Print this page


  • 2019
    • Brewer, Aaron - Ph.D. Dissertation
      Magnesium isotope fractionation associated with biotic and abiotic weathering -and- Developing a scalable method for rare earth element extraction from non-traditional feedstocks using engineered Escherichia coli 2019, Brewer,Aaron,Aaron Brewer Magnesium isotope fractionation associated with biotic and abiotic weathering -andDeveloping a scalable method for rare earth element extraction from non-traditional feedstocks using engineered Escherichia coli Aaron W Brewer A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy University of Washington 2019 Reading Committee: Fang-Zhen Teng Chair Yongqin Jiao Drew Gorman-Lewis Program Authorized to Offer Degree: Earth and Space Sciences Copyright 2019 Aaron W Brewer ii University of Washington ABSTRACT Magnesium isotope fractionation associated with biotic and abiotic weathering -and- Rare earth element extraction from non-traditional feedstocks using engineered Escherichia coli Aaron W Brewer Chairman of the Supervisory Committee: Fang-Zhen Teng Ph D Department of Earth and Space Sciences This dissertation is divided into two sections with the first discussing magnesium Mg isotope behavior during biotic and abiotic rock weathering and the second describing the application of engineered microbes for the selective recovery of rare earth elements REEs from non-traditional feedstocks Magnesium isotopes exhibit observable mass dependent fractionation during a variety of rock weathering processes permitting them to serve as a tracer of chemical and biological weathering First Bacillus subtilis endospore-mediated forsterite dissolution experiments demonstrating the effects of cell surface reactivity on Mg isotope fractionation are discussed The endospore surfaces preferentially adsorbed 24 Mg from the forsterite dissolution products with calculated adsorbed Mg isotope compositions reaching 26Mg approx -0 51 compared to the aqueous phase at approx -0 37 These results demonstrate the effect of cell surface reactivity on Mg isotope fractionation in isolation separate from other biological processes such as metabolism and organic acid production Second the Mg isotope compositions of granite and granodiorite iii weathering profiles were examined to investigate Mg isotope behavior during the weathering of felsic rocks Mg isotope fractionation in these profiles was primarily controlled by primary biotite weathering and secondary illite formation and weathering Finally North Cascade Volcanic Arc basalts and andesites were analyzed for Mg isotopes to assess the extent and mechanism of crustal contribution to this magmatic system The 26Mg of these samples vary from within the range of ocean island basalts the lightest being -0 33 0 07 to heavier compositions as heavy as -0 15 0 06 The heavy Mg isotope compositions are best explained by the addition of crustal materials during assimilation and fractional crystallization The results show that Mg isotopes may be a valuable tracer of crustal input into a magma supplementing more traditional methods Next we investigate biosorption as a potential means of recovering REEs from nontraditional feedstocks We examined how REE adsorption by engineered E coli is controlled by various geochemical factors relevant to natural geofluids including total dissolved solids TDS temperature pH and the presence of specific competing metals REE biosorption is largely unaffected by high TDS concentrations although high concentrations of some metals e g U and Al and decreasing pH below 5-6 were found to limit REE recovery REE extraction efficiency and selectivity increase with temperature up to 70 C which is best explained by the thermodynamic properties of metal complexation on the bacterial surfaces Together these data demonstrate the potential utility of biosorption for selective REE recovery from geothermal fluids however the cells alone are generally not suitable for industrial-scale extraction operations In the second portion of this section we immobilize the engineered cells by encapsulating them in polyethylene glycol diacrylate PEGDA microparticles for use in fixed-bed columns We demonstrate that optimal REE recovery 2 6 mg Nd g dry sorbent occurs at an influent flow rate of 1 ml min pH of 6 and maximum REE concentration of 21 mM The microparticles exhibit minimal loss in performance iv over 9 adsorption desorption cycles Furthermore they have a strong preference for heavy REEs particularly Eu Sm Yb and Lu which may permit the separation of individual rare earth metals The results of this study represent a major step towards making biosorption a viable industrial scale REE extraction technology v To my parents Laurie and Terry Brewer and my sister Katherine Brewer vi ACKNOWLEDGEMENTS This work is ultimately the result of the combined efforts of many people who provided their scientific expertise personal and professional guidance and moral support I would first like to express my gratitude to my two primary supervisors Dr Fang-Zhen Teng and Dr Yongqin Jiao They are both leaders in their fields who contributed significantly to the conception and development of these projects throughout the duration of my Ph D Without them and their continued efforts I would never have reached this point I would also like to thank my committee members Dr Drew Gorman-Lewis and Dr Bruce Nelson Both have provided invaluable advice for my research efforts through the years In particular their feedback has greatly improved this dissertation Dr Dan Park although not technically one of my supervisors certainly deserves recognition for unofficially filling that role with such aplomb I would also like to acknowledge Dr Rory Barnes my Graduate Student Representative for being an excellent mentor and teacher and for his willingness to help Undergraduate assistants often go without recognition proportional to their efforts and I have truly been fortunate to have three undergraduates who went far above and beyond the expectations of their position If their graduate student put out the call for some help in the lab Khadijah Homolka Jiarui Zhou and Florence Yuen could always be counted on to be there probably with a snack to share and a sarcastic comment about my failing work ethic Finally some would see it as a sign of true megalomania to ignore the contributions of the people who raised him so I feel obliged to acknowledge my parents Laurie and Terry Brewer and my sister Katherine Brewer Who can adequately describe the role of one s family in the progression of their life so I will leave it at that and say a final thanks to everyone named and unnamed who has made this work possible vii Table of Contents List of Figures x List of Tables xii Chapter 1: Introduction 1 Chapter 2: Magnesium isotope fractionation during microbially enhanced forsterite dissolution 18 Abstract 18 1 Introduction 19 2 Materials and Methods 22 2 1 Endospore preparation 22 2 2 Forsterite preparation 23 2 3 Direct and indirect dissolution experiments 23 2 4 Chemical analyses 28 2 5 Magnesium isotope analyses 28 2 6 Saturation state modeling 29 3 Results 30 4 Discussion 32 4 1 Mg isotope fractionation during forsterite dissolution 32 4 2 Effects of secondary mineral precipitation 35 4 3 Mg isotope fractionation during endospore surface adsorption 36 4 4 Implications 39 5 Conclusions 41 6 References 42 Chapter 3: Magnesium isotope fractionation during granite weathering 47 Abstract 47 1 Introduction 48 2 Samples 50 3 Methods 55 4 Results 56 5 Discussion 57 5 1 Lateral transport 58 5 2 Primary and secondary mineral behavior 59 5 3 Comparison with other weathering profiles 62 5 4 Implications 64 6 Conclusions 66 7 References 67 Chapter 4: Magnesium Isotopes as a Tracer of Crustal Materials in Volcanic Arc Magmas in the Northern Cascade Arc 72 Abstract 72 1 Introduction 73 2 Samples 75 3 Methods 77 4 Results 78 5 Discussion 80 5 1 Mg isotope variations in the North Cascade Volcanic Arc 80 5 2 Hypotheses for Mg isotope variations 81 5 3 Modeling crustal input 84 viii 6 Summary 89 7 References 90 Chapter 5: Recovery of rare earth elements from geothermal fluids through bacterial cell surface adsorption 94 Abstract 94 1 Introduction 95 2 Methods 97 2 1 Bacterial strains and growth conditions 97 2 2 Blue Mountain geofluid REE biosorption 97 2 3 REE biosorption with the Great Salt Lake brine 98 2 4 Temperature dependence 99 2 5 Metal competition experiments 99 2 6 ICP-MS analysis 99 2 7 Thermodynamic analysis 100 3 Results and Discussion 100 3 1 REE recovery from the Blue Mountain geofluid 100 3 2 Effects of high TDS 102 3 3 Effects of competing metals 104 3 4 Effects of pH 105 3 5 Effects of temperature 107 3 6 Thermodynamic analysis of temperature-dependent biosorption 109 3 7 Implications for REE extraction 111 4 Supporting Information 112 5 References 117 Chapter 6: Selective recovery of rare earth elements from non-traditional feedstocks using bacteria immobilized in polymer microparticles 122 Abstract 122 1 Introduction 123 2 Methods 125 2 1 Bacterial strains and growth conditions 125 2 2 Microparticle synthesis 126 2 3 Batch sorption experiments 126 2 4 Breakthrough columns 127 2 5 ICP-MS analysis 128 3 Results and Discussion 128 3 1 Microparticle synthesis and characterization 128 3 2 REE adsorption in a fixed-bed column 130 3 3 Column Reusability 133 3 4 Selective REE recovery 134 3 5 Implications for REE extraction 136 4 Supporting Information 137 5 References 138 Chapter 7: Summary and Future Work 142 ix List of Figures Figure 2-1: Si M Mg M and Mg Si mol mol changes with time for four forsterite dissolution assays 25 Figure 2-2: Variation of 26Mg with a time b Si M and c Mg M for four forsterite dissolution assays 31 Figure 2-3: 26Mg of four forsterite dissolution assays as a function of % Mg adsorbed 33 Figure 2-4: a Mg Si mol mol and b 26Mg as a function of hematite saturation index in the four forsterite dissolution assays 36 Figure 2-5: Isotope composition of the Mg adsorbed onto endospore surfaces compared to aqueous composition as a function of time 37 Figure 3-1: Mineral composition as a function of depth for four granite and granodiorite weathering profiles 51 Figure 3-2: Normalized MgO content as a function of depth for four granite and granodiorite weathering profiles 54 Figure 3-3: wt% MgO correlation with chemical index of weathering for four granite and granodiorite weathering profiles 55 Figure 3-4: 26Mg as a function of depth for four granite and granodiorite weathering profiles 57 Figure 3-5: 26Mg variation with the illite fraction of Mg-bearing minerals for four granite and granodiorite weathering profiles 61 Figure 3-6: Rayleigh distillation models of 26Mg variation with wt% MgO for a range of weathering profiles 62 Figure 3-7: 26Mg as a function of chemical index of weathering for four granite and granodiorite weathering profiles 63 Figure 4-1: Magnesium isotope composition of volcanic arc samples 74 Figure 4-2: a Ba and Nb contents in volcanic arc samples b Sr isotope composition and Sr contents in volcanic arc samples 75 Figure 4-3: Variation of 26Mg with a wt% MgO and b Sr isotope composition in the North Cascade Volcanic Arc samples 81 Figure 4-4: Variation of 26Mg with a Sm Yb and b Dy Yb in the North Cascade Volcanic Arc samples 82 Figure 4-5: Variation of 26Mg with a Ba Th b Th Yb and c Pb Ce in the North Cascade Volcanic Arc samples 83 Figure 4-6: Assimilation and fractional crystallization models and bulk mixing models of 26Mg variation with a wt% MgO and b Sr isotope composition in the North Cascade Volcanic Arc samples 85 x Figure 5-1: a Composition of the Blue Mountain Geofluid before biosorption and of the adsorbed metals and b Fold purity increase of each metal following biosorption 102 Figure 5-2: REE biosorption as a function of total dissolved solids 103 Figure 5-3: Effect of competing metals at various concentrations on REE biosorption 104 Figure 5-4: REE and major element biosorption as a function of pH 106 Figure 5-5: a REE biosorption b non-REE extraction efficiency as a function of temperature in the Great Salt Lake brine c Metal extraction efficiency in simple buffer solution as a function of temperature d Terbium extraction efficiency at several temperature conditions 108 Figure 5-6: a Van t Hoff plots of lanthanum-acetate and project lanthanum-bacteria complexation b Metal-bacteria stability constants as a function of temperature 110 Figure 5-7: Schematic showing process flow for a potential industrial-scale biosorption system for REE extraction from geofluids 111 Supplementary Figure 5-1: Correlation between metal-bacteria logK and metal-acetate logK for several metals 115 Supplementary Figure 5-2: Live dead staining of LBT-displayed E coli cells following heating to 70 C 116 Supplementary Figure 5-3: Terbium recovery from the Great Salt Lake brine up to 100 C 116 Figure 6-1: Nd a adsorption capacity b adsorption kinetics and c desorption kinetics for the hybrid microparticles and cells alone 129 Figure 6-2: Nd breakthrough behavior in a fixed-bed column packed with microparticles with and without cells 130 Figure 6-3: Nd breakthrough as a function of influent a flow rate b Nd concentration and c pH 131 Figure 6-4: Nd a adsorption capacity and b breakthrough behavior over 9 adsorption desorption cycles 134 Figure 6-5: a Multi-REE breakthough and b REE selectivity in a fixed-bed column with the microparticle adsorbent 135 Supplementary Figure 6-1: Tb adsorption capacity as a function of a cell density and b PEGDA content for the hybrid microparticles 137 Supplementary Figure 6-2: Comparison of Nd breakthrough results using colorimetric methods Arsenazo and ICP-MS 138 xi List of Tables Table 2-1: Magnesium and silicon concentrations and Mg isotope compositions of dissolution assay solutions and forsterite powder 26 Table 2-2: Mg isotope composition of reference materials 30 Table 3-1: Mineralogical compositions of granite and granodiorite weathering profile samples 52 Table 3-2: Magnesium isotope and chemical composition of granite and granodiorite weathering profile samples and standards 53 Table 4-1: Magnesium isotope and trace element compositions of North Cascade Volcanic Arc samples and standards 79 Table 4-2: North Cascade Volcanic Arc modeling parameters and end-member compositions 84 Table 5-1: Chemical composition of representative geofluids 101 Supplementary Table 5-1: Chemical composition of the Blue Mountain geothermal fluid before and after biosorption 113 Supplementary Table 5-2: Terbium sodium and magnesium extracted from the Great Salt Lake solution through biosorption 114 Supplementary Table 5-3: Terbium adsorption by LBT-displayed E coli in MES buffer pH 6 114 xii CHAPTER 1: INTRODUCTION The research discussed here covers two distinct sections which will be examined in sequence The first section using magnesium isotopes to trace weathered rock and elucidate biotic and abiotic weathering processes is primarily the result of research performed at the University of Washington The second section applying microbes to selectively adsorb rare earth elements from various feedstock solutions is the culmination of work conducted at Lawrence Livermore National Laboratory as part of the Livermore Graduate Scholar Program Part 1: Magnesium isotope fractionation associated with biotic and abiotic weathering and weathered materials Chemical weathering is a major determinant of the state of the atmosphere hydrosphere biosphere and lithosphere Silicate weathering in combination with carbonate precipitation removes carbon dioxide from the atmosphere exerting a major influence on the atmospheric CO2 budget Urey 1952 The chemical erosion and disintegration of igneous rocks provides the components for the formation of clays and other sedimentary rocks and affects river and seawater composition Edmond et al 1979 These aqueous weathering products are also essential for life supplying many of the elements necessary for cell growth and metabolism Finally the subduction of weathered material can have an impact on the chemical composition of the middle and lower crust as well as upwelling magmas Hofmann 2003 Planck 2014 Chemical weathering is therefore involved directly or indirectly in a wide range of geological processes 1 Magnesium Mg is a major element that is abundant in the hydrosphere biosphere and lithosphere Magnesium has three stable isotopes 24Mg 25 Mg and 26 Mg and many geochemical processes cause measurable mass dependent Mg isotope fractionation due to the 8% mass difference between 24 Mg and 26 Mg Teng 2017 and references therein Although Mg isotope analyses have now been applied to the study of many of these processes providing progress towards a systematic knowledge of Mg isotope behavior some gaps remain in our understanding of this valuable geochemical tracer Minimal Mg isotope fractionation is observed at high temperatures so the compositions of uncontaminated mid-ocean ridge basalts and of the mantle are essentially homogeneous 26Mg -0 25 0 07 and -0 25 0 04 respectively Teng et al 2010a Magnesium leaching from igneous rocks into the hydrosphere during chemical weathering produces isotopically heavy residual rock as high as 1 81 and a correspondingly light fluid phase Tipper et al 2008 2010 Li et al 2010 Teng et al 2010b Liu et al 2014 Wimpenny et al 2014 Although this preferential leaching of 24 Mg may be generally representative of most chemical weathering systems there are many additional processes that play variable roles in determining Mg isotope behavior in natural weathering environments A natural weathering system will involve primary mineral dissolution and alteration secondary mineral formation surface adsorption desorption biological processes and potentially horizontal deposition of additional material The final Mg isotope composition of a weathered sample will be the result of some combination of these processes and may be informative about the weathering history of the sample For example Mg isotopes have been used to identify loess deposition in a weathering profile Teng et al 2010b and can distinguish between adsorbed and structural Mg in secondary clay minerals Wimpenny et al 2014 Mg isotopes may similarly be 2 used as a biosignature in the future once the fractionation effects of various biological processes have been more systematically examined The utility of Mg isotopes as an indicator of weathering will be more fully realized by assessing the distinct effects of individual weathering processes the cumulative results of those processes in natural weathering systems and the fate of weathered material in the environment Magnesium isotope behavior associated with biological processes has largely remained unexplored Several studies have investigated Mg isotope behavior in plants Black et al 2006 2008 Bolou-Bi et al 2010 2012 Uhlig et al 2017 fungi Fahad et al 2016 and in biogenic carbonates Higgins and Schrag 2010 Immenhauser et al 2010 Pokrovsky et al 2011 Pokharel et al 2017 assessed Mg isotope fractionation during uptake into microbes but otherwise Mg interactions with microbial life have not been tested The biosphere may play a particularly large and complex role in Mg isotope fractionation during chemical weathering Microbes may influence Mg isotope behavior by increasing the rates of mineral dissolution and Mg release into the hydrosphere through direct surface adhesion organic acid production and or the removal of aqueous dissolution products via uptake and adsorption Lee and Fein 2000 Wightman and Fein 2004 Song et al 2006 Uptake and adsorption may also affect the Mg isotope composition of the fluid phase in addition to affecting weathering rates These effects may be compared to Mg isotope behavior during purely chemical dissolution of specific minerals which has been investigated for brucite Li et al 2014 and forsterite Wimpenny et al 2010 Chapter 2 assesses the effects of cell surface reactivity on Mg isotope fractionation by studying forsterite dissolution in the presence of Bacillus subtilis endospores at various experimental conditions Because endospores are metabolically dormant and do not produce 3 significant organic acids they affect the dissolution process only through cell surface reactivity Olsen and Rimstidt 2008 Song et al 2006 Lee and Fein 2000 Vandiviere et al 1994 Harrold 2014 Their sorption behavior is also generally representative of vegetative cell surfaces Fein et al 2005 Gorman-Lewis et al 2006 Harrold and Gorman-Lewis 2013 and they are therefore ideal for isolating the Mg isotope fractionation associated with cell surface reactivity in weathering systems The endospore surfaces may affect Mg isotopes by attaching to the mineral surfaces to increase dissolution rate and or by interacting with the aqueous dissolution products Dissolution assays with and without dialysis tubing preventing direct interaction between the mineral powder and the endospores permit those processes to be differentiated In combination the two assays show that the endospores do not appear to have an effect on Mg isotopes during Mg leaching from the forsterite over the analyzed time period 1043 days The compositions of the total dissolution products for the biotic and abiotic control assays are indistinguishable if all other processes are excluded However the endospores do cause an increase in the Mg isotope composition of the aqueous phase by preferentially adsorbing dissolved 24 Mg The original forsterite dissolution assays were performed by Zoe Harrold working under the supervision of Drew Gorman-Lewis at the University of Washington Fang-Zhen Teng contributed to the conception of the Mg isotope component of this work The results of this study are in the final stages of preparation to be submitted to Geobiology The laboratory dissolution experiments are informative because they permit an individual chemical process to be studied in isolation however natural weathering systems are the combined result of many distinct processes For example soil formation Immenhauser et al 2011 Ma et al 2015 and clay mineral precipitation and surface chemistry Huang et al 2012 Ma et al 2015 Opfergelt et al 2014 Tipper et al 2012 Wimpenny et al 2014 may have a 4 significant impact on the final composition of a weathering system in addition to mineral leaching and dissolution The Mg isotope composition of several natural weathering profiles have been assessed to date focusing mainly on mafic lithologies such as basalt Huang et al 2012 Liu et al 2014 Pogge von Strandmann et al 2012 and diabase Teng et al 2010b although one shale profile has also been analyzed Ma et al 2015 However the Mg isotope composition of granite and granodiorite weathering profiles has not previously been the subject of systematic study This gap in our understanding is significant because granites are the most abundant rock-type in the upper crust while granodiorite is taken as representative of the bulk chemical composition of the upper crust Rudnick and Gao 2003 Wedepohl 1995 Magnesium isotope analyses of granite and granodiorite weathering profiles may provide important insights into weathering on the continents and into the Mg cycle in these environments Chapter 3 investigates Mg isotope fractionation during granite and granodiorite weathering specifically with regard to the dissolution and alteration of primary Mg-bearing minerals and the formation and behavior of secondary clay minerals Mg isotope composition in these profiles was primarily the result of primary biotite weathering and the formation and weathering of secondary clay minerals particularly illite As biotite weathers some 24 Mg is preferentially lost to the hydrosphere producing increased Mg isotope compositions from the bedrock until the upper 2 m of the profiles However significant Mg is preserved in the rock through incorporation into secondary illite minimizing the fractionation effect At depths 12 repeat measurements of the standards during an analytical session 2SD for the average of each assays is two times the standard deviation of the sample population 26 In the biotic indirect or dialysis assays any changes in dissolution behavior associated with the endospores can be attributed to the adsorption of dissolved Mg onto the endospore surfaces Harrold and Gorman-Lewis 2013 decreasing the chemical activity of Mg in the bulk solution Harrold 2014 While forsterite dissolution is expected to always be far from equilibrium at these temperature and pressure conditions Olsen and Rimstidt 2008 White and Brantley 1995 it is possible that the preferential leaching of Mg from the crystal lattice could result in a Si-rich layer that can polymerize at the experimental pH of 7 5 Pokrovsky and Schott 2000a 2000b Magnesium dissolution products may be required to diffuse through this layer and the rate of diffusion would be dependent upon the Mg activity of the bulk solution Pokrovsky and Schott 2000a 2000b Therefore Mg adsorption onto the endospores would increase the rate of forsterite dissolution An increased adsorption rate was initially observed however the rates had returned to the abiotic level -11 40 0 09 molg-1s-1 2SD prior to the period for which samples were tested for Mg isotope 10-43 days Harrold 2014 Magnesium is also highly depleted relative to Si in the dissolution products over the entire incubation period due to Mg adsorption onto the endospore surfaces with an average Mg:Si ratio of approximately 0 60 0 28 2SD Harrold 2014 Fig 1 In the biotic homogenous or free assays the endospores can adhere directly to the mineral surfaces in addition to interacting with the aqueous dissolution products Endospore adhesion onto the mineral surfaces occurs via the complexation of anionic oxygen moieties on the endospore surface with Mg on the mineral surface Harrold 2014 This complexation shifts the electron density from the oxygen in the Si-O-Mg forsterite structure to the Mg-ligand bond making the breaking of the bridging O bond more thermodynamically favorable Olsen and Rimstidt 2008 Again the dissolution rate did initially increase but returned to approx abiotic 27 levels -10 97 0 37 2SD prior to the period for which Mg isotopes were analyzed Harrold 2014 Magnesium is also depleted relative to Si in the dissolution products over the entire incubation period but less depleted than in the dialysis assays with an average Mg:Si ratio of approximately 1 14 0 12 2SD again due to Mg adsorption onto the endospore surfaces Harrold 2014 Fig 1 It is possible that the direct forsterite-endospore interaction in the homogenous endospore assays may also release excess Si into solution compared to the abiotic assays Harrold 2014 2 4 Chemical Analyses In Harrold 2014 magnesium and iron concentrations were measured on an inductively coupled plasma optical emission spectrometer ICP-OES Silicon concentrations were measured using the molybdenum blue colorimetric method which quantifies the concentrations of aqueous monomeric and dimeric silicic acid Cation and silica concentrations generated during the biotic and abiotic forsterite dissolution assays are characterized and discussed in detail in Harrold 2014 2 5 Magnesium Isotope Analyses Magnesium isotope analyses were performed at the Isotope Laboratory in the Department of Earth and Space Sciences at the University of Washington All procedures are similar to those discussed in previous papers Li et al 2010 Teng et al 2007 2010a Teng and Yang 2014 Yang et al 2009 Filtered samples were transferred to Savillex screw-top Teflon beakers and dried To ensure complete dissolution the dried samples were subjected to a 6:1 HF-HNO3 acid digestion followed by a 3:1 HCl-HNO3 acid digestion The samples were kept in each acid at approx 28 125 C for several days To achieve cation separation the samples were passed twice through a column containing 1 8 ml Bio-Rad 200-400 mesh AG50W-X8 cation exchange resin in a 1N HNO3 media and were eluted using 1N HNO3 The collected aliquot contains 99% of the Mg in the sample to prevent isotope fractionation within the column and limits the concentrations of other elements to less than 5% of the concentration of Mg Magnesium isotope compositions were analyzed using the sample-standard bracketing method on a Nu Plasma II MC-ICP-MS Mg isotope data is expressed in standard -notation: %& 1 1000 % &% where x refers to masses 25 or 26 The reference materials Seawater n 8 and San Carlos Olivine n 10 were analyzed at least once during each analytical session each time yielding a 26Mg value of -0 83 0 07 and -0 25 0 07 Table 2 These values agree with previously published data Foster et al 2010 Hu et al 2016 Ling et al 2011 Teng et al 2015 2 6 Saturation State Modeling Saturation indices were modeled using the PHREEQC geochemical modeling software Ksp at 298 K and 1 atm and relevant equilibrium phases were obtained from the phreeqc dat database Solution conditions were 25 mM NaCl ionic strength a pH of 7 5 and a temperature of 298 K 29 Sample 26Mg 2SD 25Mg 2SD San Carlos Olivine Duplicate Replicate Replicate Replicate Duplicate Replicate Replicate Replicate Replicate Seawater Replicate Duplicate Duplicate Duplicate Replicate Replicate Replicate -0 29 -0 23 -0 25 -0 22 -0 28 -0 27 -0 26 -0 26 -0 27 -0 30 -0 85 -0 88 -0 83 -0 88 -0 88 -0 84 -0 88 -0 88 0 06 0 09 0 09 0 09 0 09 0 05 0 07 0 07 0 07 0 05 0 06 0 06 0 09 0 05 0 07 0 07 0 07 0 05 -0 09 -0 11 -0 13 -0 08 -0 13 -0 15 -0 12 -0 16 -0 14 -0 16 -0 43 -0 44 -0 41 -0 49 -0 46 -0 41 -0 41 -0 46 0 07 0 11 0 11 0 11 0 11 0 08 0 06 0 06 0 06 0 07 0 07 0 07 0 11 0 08 0 06 0 06 0 06 0 07 Table 2: Mg isotope composition of reference materials 2SD Two standard deviation of the population of n n 12 repeat measurements of the standards during an analytical session 3 Results The Mg isotope composition of the samples overall ranged from 26Mg -0 39 0 06 to -0 08 0 07 The Mg isotope composition of each of the four assays remained approximately constant within experimental error over the duration of the incubation period 1043 days The homogenous abiotic control assay ranged from 26Mg -0 39 0 06 to -0 28 0 06 with an average Mg isotope composition of 26Mg -0 35 0 08 while the abiotic dialysis assay ranged from 26Mg -0 35 0 05 to -0 26 0 09 with an average Mg isotope composition of 26Mg -0 31 0 09 Fig 2 There is no statistically significant 30 difference in Mg isotope composition between the abiotic free and dialysis assays two-tailed ttest p 0 5360 and therefore the two assays will be discussed together The biotic homogenous assay ranged from 26Mg -0 30 0 05 to -0 21 0 06 with an average Mg isotope composition of 26Mg -0 26 0 06 and the biotic dialysis assay ranged from 26Mg -0 17 0 06 to -0 08 0 07 with an average Mg isotope composition of 26Mg -0 11 0 07 Fig 2 These two biotic assays are statistically distinct from each other p 0 0012 and from the abiotic assays p 0 0455 and 0 0005 for the free and dialysis assays respectively All sample data is reported in Table 1 and all reference material data is reported in Table 2 Figure 2: Mg isotope composition a over the duration of the incubation period b as a function of aqueous Si concentration M and c as a function of aqueous Mg concentration M Error bars represent the 2SD for each sample The triangles represent the dialysis assays and the circles represent the homogenous assays while the red and blue symbols represent the biotic and abiotic assays respectively The solid line represents the Mg isotope composition of the pristine forsterite Mg and Si concentration data are from Harrold 2014 Data are reported in Table 1 31 4 Discussion In this section we first describe the Mg isotope fractionation associated with forsterite dissolution We then examine the fractionation of aqueous Mg during secondary mineral precipitation and adsorption onto the endospore surfaces Finally we discuss the implications of these findings for our understanding of chemical weathering and the future of Mg isotope research 4 1 Mg isotope fractionation during forsterite dissolution Forsterite dissolution resulted in the preferential release of 24 Mg into solution for both biotic and abiotic assays Because additional processes such as adsorption and or secondary mineral precipitation are involved in producing the measured aqueous Mg isotope composition the effects of these processes must be excluded prior to determining the composition of the dissolution products Congruent forsterite dissolution will produce aqueous Mg:Si molar ratios reflecting the original mineral composition 1 8:1 and any change in that Mg:Si ratio is likely due to additional processes The percent of dissolved Mg that was removed from solution due to adsorption and or precipitation can be estimated by comparing the measured aqueous Mg concentration to the Mg concentration calculated based on measured aqueous Si concentration and assuming a Mg:Si ratio of 1 8 % 1 %& 100 1 8 %& It is assumed that no Si is lost due to precipitation or adsorption A few abiotic samples exhibit minimal 15% aqueous Mg removal can be estimated and compared to these measured values The Mg isotope composition across the four assays varies approx linearly with Mg:Si mol mol and % Mg removed Fig 3 We can therefore perform a linear regression and calculate the Mg isotope composition of the dissolution products in the biotic and abiotic assays at 0% aqueous Mg removed Fig 4 The calculated aqueous Mg isotope composition of the dissolution products in the biotic assays is 0 41 comparable to the pristine abiotic assays average 26Mg -0 37 0 02 and the measured isotope composition of samples at biotite microcline quartz with clay concentrations increasing with degree of weathering Dethier and Bove 2011 Soils contain minor aeolian contributions as well as laterally transported upslope material Dethier and Bove 2011 Magnolia Road 0 2 Depth m Cave Creek South 0 2 4 4 6 6 8 0 5 10 15 Lee Hill Road 0 8 0 5 10 0 Kaolinite Illite Biotite Smectite 1 Depth m 1 15 Hurricane Hill 2 3 2 4 5 3 6 4 0 5 10 15 20 Weight % 25 30 7 0 5 10 15 20 25 30 Weight % Figure 1: Variation of mineral composition wt% with depth for each of the four profiles Note the differences in both x and y-axis scales between the panels Data are from Dethier and Bove 2011 and are reported in Table 1 The mineralogy of the samples follows the expected trends in regard to Mg-bearing minerals with primary biotite generally decreasing with decreasing depth being replaced by clay minerals primarily illite but also kaolinite and smectite in some cases In the Magnolia Road profile biotite concentrations decrease 3 89 wt% from the bedrock to the near surface regolith with a corresponding 4 60 wt% increase in smectite 10 60 wt% increase in illite and 1 01 wt% increase in kaolinite Table 1 Figure 1 The Lee Hill Road profile is slightly more complex with a 2 29 wt% decrease in biotite from the bedrock to the upper portion of the saprolite but then a 10 14 wt% increase from the saprolite to the regolith Figure 1 There is a sharp increase in kaolinite 2 92 wt% smectite 3 34 wt% and illite 2 34 wt% across that saprolite regolith 51 Sample Biotite Illite Smectite Kaolinite Chlorite Calcite Rock Type Hurricane Hill LC-03-31a LC-03-31b LC-03-31c LC-03-31d LC-03-31f 20 4 14 6 15 7 19 1 0 10 0 0 0 7 7 3 3 3 7 9 0 0 0 0 0 0 0 0 0 0 0 2 0 0 1 7 0 0 10 7 0 9 0 9 1 7 0 7 0 2 0 00 0 00 0 00 0 00 0 00 Bedrock Oxidized Bedrock Oxidized Bedrock Oxidized Bedrock Regolith hydrothermal Magnolia Road LC-03-32a LC-03-32b LC-03-32c LC-03-32d LC-03-32e 12 71 11 32 8 68 9 51 8 82 0 54 4 02 3 65 12 30 11 14 0 00 0 00 0 00 2 73 4 60 1 48 1 55 1 68 2 70 2 49 0 29 0 02 0 00 0 07 0 09 Bedrock Oxidized Bedrock Oxidized Bedrock Saprolite Regolith grus Cave Creek South LC-04-02 TC-05-S-01a1 TC-05-S-01b1 TC-05-S-01c2 TC-05-S-01d1 TC-05-S-01e1 4 71 1 52 2 87 2 79 3 95 3 94 0 06 10 47 1 82 4 52 8 43 7 35 0 00 3 81 3 56 3 07 4 98 4 73 1 29 2 56 2 94 3 28 5 46 5 54 0 00 0 00 0 00 0 00 0 00 0 00 Bedrock Saprolite Saprolite Saprolite Regolith soil Regolith soil Lee Hill Road TC-05-S-02a TC-05-S-02b TC-05-S-02c TC-05-S-02d 19 49 16 18 17 20 27 34 0 00 5 54 3 93 6 27 0 00 0 00 0 00 3 34 0 48 1 55 1 57 4 49 0 20 0 10 0 05 0 05 Bedrock Saprolite Saprolite Regolith soil 16 31 0 94 0 00 0 72 0 29 11 77 0 00 0 00 0 68 0 00 Bedrock Bedrock Bedrock Bedrock Additional Samples DC-04-110 DC-04-116 DC-04-117 DC-05-01 Table 1: Mineralogical compositions of weathering profile samples Quantitative XRD data for Magnolia Road Cave Creek South Lee Hill Road and additional samples from Dethier and Bove 2011 All data are expressed in weight percent 52 Sample 26Mg 2SD 25Mg 2SD Depth m Rock Type MgO wt% Mg Norm Hurricane Hill LC-03-31a LC-03-31b LC-03-31c LC-03-31d LC-03-31f -0 07 -0 15 0 07 -0 15 0 55 0 06 0 06 0 06 0 06 0 06 -0 05 -0 08 0 07 -0 07 0 30 0 07 0 07 0 07 0 07 0 07 6 50 3 50 2 50 1 50 0 50 Bedrock Oxidized Bedrock Oxidized Bedrock Oxidized Bedrock Regolith hydrothermal 2 44 2 05 2 00 2 39 0 30 1 00 1 03 0 77 0 97 0 17 Magnolia Road LC-03-32a LC-03-32b LC-03-32c LC-03-32d LC-03-32e -0 17 -0 12 -0 10 0 00 -0 12 0 06 0 06 0 06 0 06 0 06 -0 10 -0 06 -0 08 0 04 -0 06 0 07 0 07 0 07 0 07 0 07 8 00 7 00 6 00 1 50 0 80 Bedrock Oxidized Bedrock Oxidized Bedrock Saprolite Regolith grus 1 61 2 09 1 33 2 33 0 58 1 00 1 17 1 00 0 91 0 36 Cave Creek South LC-04-02 TC-05-S-01a1 TC-05-S-01b1 TC-05-S-01c2 TC-05-S-01d1 TC-05-S-01e1 0 07 0 24 0 10 0 21 0 14 0 09 0 06 0 06 0 06 0 06 0 06 0 06 0 02 0 15 0 03 0 09 0 07 0 07 0 07 0 07 0 07 0 07 0 07 0 07 7 00 1 65 0 85 0 60 0 40 0 15 Bedrock Saprolite Saprolite Saprolite Regolith soil Regolith soil 0 39 0 34 0 36 0 43 0 52 0 77 1 00 0 74 0 88 0 98 0 90 0 67 Lee Hill Road TC-05-S-02a TC-05-S-02b TC-05-S-02c TC-05-S-02d -0 17 -0 04 -0 12 -0 14 0 06 0 06 0 06 0 06 -0 07 0 01 -0 06 -0 07 0 07 0 07 0 07 0 07 3 30 2 20 0 80 0 30 Bedrock Saprolite Saprolite Regolith soil 2 79 3 14 3 17 3 73 1 00 1 07 1 11 1 19 Additional Samples DC-04-110 DC-04-116 DC-04-117 DC-05-01 -0 14 -0 26 -0 18 -0 15 0 06 0 06 0 06 0 06 -0 10 -0 10 -0 09 -0 08 0 07 0 07 0 07 0 07 0 00 0 00 0 00 0 00 Bedrock Bedrock Bedrock Bedrock 2 47 1 86 3 69 1 87 Reference Materials Basalt BR Replicate Replicate Replicate Seawater Duplicate -0 33 -0 35 -0 36 -0 36 -0 85 -0 87 0 06 0 06 0 06 0 06 0 06 0 06 -0 20 -0 17 -0 18 -0 22 -0 42 -0 46 0 07 0 07 0 07 0 07 0 07 0 07 Table 2: Magnesium isotope and chemical compositions of weathering profile samples and reference materials Elemental data from Dethier and Bove 2011 2SD Two standard deviation of the population of n n 22 repeat measurements of the standards during an analytical session Mg Norm MgO TiO2 sample MgO TiO2 bedrock 53 Magnolia Road 0 Cave Creek South Lee Hill Road boundary as well Figure 1 The Hurricane Hill 1 Cave Creek South profile exhibits Depth m 2 3 a 3 19 wt% decrease in biotite 4 5 from the bedrock to the saprolite 6 7 but then an increase up through 8 9 0 0 5 1 1 5 0 0 5 1 1 5 0 MgO norm 0 5 1 1 5 0 0 5 1 Figure 2: Variation of normalized MgO concentration with depth for three weathering profiles Magnolia Road Cave Creek South and Lee Hill Road and one hydrothermal weathering profile Hurricane Hill MgO Norm MgO TiO2 sample MgO TiO2 bedrock Data are from Dethier and Bove 2011 and are reported in Table 2 1 5 the saprolite kaolinite and increases regolith consistently from 1 29 wt% to 5 54 wt% but the trends in illite and smectite are more erratic increasing and decreasing across the profile Figure 1 Finally the Hurricane Hill profile contains 20 4 wt% biotite in the unaltered bedrock which decreases to only 0 1 wt% in the upper regolith with corresponding increases in kaolinite 0 2 to 10 7 wt% and illite 0 to 7 9 wt% Figure 1 The Mg concentration of these four profiles also demonstrates the effect of weathering Figure 2 The Magnolia Road Cave Creek South and Hurricane Hill profiles exhibit a decrease in Mg concentration as expressed in terms of normalized MgO MgO Norm MgO TiO2 sample MgO TiO2 bedrock between the bedrock and the upper 2 meters of the profile At Magnolia Road Cave Creek South and Hurricane Hill this decrease is approximately 9% 26% and 23% respectively then in the upper 2 meters Mg is depleted again dropping a further 55% 7% and 60% respectively Figure 2 The Lee Hill Road profile does not follow this trend and instead demonstrates a 19% increase in Mg from the bedrock to the near surface regolith likely due to the deposition of less weathered upslope material Figure 2 The profiles generally exhibit a decrease in MgO concentration with increasing chemical index of weathering 54 4 CIW Al2O3 Al2O3 CaO Na2O 100 as is given that Mg-bearing minerals are progressively lost during weathering Figure 3 MgO wt% expected Magnolia Road Cave Creek South Lee Hill Road Hurricane Hill Additional Bedrock Samples 3 5 3 Methods 3 2 5 2 1 5 1 0 5 0 65 70 All Mg isotope analyses were performed at the Isotope Laboratory at the University of Washington and follow the procedures reported in previous studies 75 CIW 80 85 90 Figure 3: Variation of Mg concentration wt% MgO with chemical index of weathering CIW where CIW Al2O3 Al2O3 CaO Na2O 100 Data are reported in Table 2 Li et al 2010 Teng et al 2007 2010a Teng and Yang 2014 Yang et al 2009 Powdered samples in Savillex screw-top beakers were dissolved in two stages to ensure complete dissolution First concentrated HF and HNO3 were added in a 3:1 ratio and the samples were heated for at least one week at approximately 125 C after which the sample was completely dried Second concentrated HCl and HNO3 were added in a 3:1 ratio heated for several days again at approximately 125 C and finally dried To achieve Mg purification the samples were passed through columns containing Bio-Rad 200-400 mesh AG50W-X8 cation exchange resin and were eluted using 1N HNO3 This procedure ensures that the collected fraction contains 99% of the total sample Mg to prevent fractionation within the column and that the matrix elements were limited to less than 5% of the concentration of Mg to prevent matrix effects Teng and Yang 2014 Magnesium isotope compositions were analyzed using the sample-standard bracketing method on a Nu Plasma II MC-ICP-MS Mg isotope data is expressed in standard -notation: %& 1 1000 % &% 55 where x refers to masses 25 or 26 The reference materials Seawater n 2 and Basalt BR n 4 were analyzed at least once during each analytical session each time yielding a 26Mg value of 0 83 0 07 and -0 41 0 09 Table 2 These values agree with previously published data Foster et al 2010 Ling et al 2011 Teng et al 2015 The mineralogy of the Hurricane Hill samples was analyzed at the Space Science and Astrobiology Division at NASA s Ames Research Center Bulk samples were crushed and passed through a 250 m sieve spiked with 10% -Al2O3 as an internal standard and micronized in ethanol using a McCrone Mill The micronized samples were dried under a heat lamp saturated with Vertrel and vortexed with plastic beads to promote random orientation The samples were then sieved 250 m and side-loaded into an XRD mount All powdered samples were analyzed on a Rigaku Smartlab diffractometer using Cu k-alpha radiation and a scintillation detector with 1 divergence and receiving slits Samples were scanned from 5 to 65 at 0 02 2 steps with a 2 second per step count time Quantitative mineralogy was calculated using the full pattern peak-fitting program RockJock 11 Eberl 2009 4 Results Magnesium isotope compositions for weathering profile samples and reference materials are reported in Table 2 along with major element data for the rock samples including those from Dethier and Bove 2011 The mineralogy of the Hurricane Hill samples together with others from Dethier and Bove 2011 is reported in Table 1 The Mg isotope composition of the non-hydrothermally altered samples generally increases slightly from the bedrock to the upper 2 meters of the profile and then above 2 m decreases again back to near bedrock values At Magnolia Road 26Mg increases from -0 17 0 06 in the bedrock to 0 00 0 06 in the saprolite at 1 5 meters depth and then decreases 56 to -0 12 0 06 in the overlying grus Cave Creek South exhibits an increase in 26Mg from 0 07 0 06 in the bedrock to 0 24 0 06 in the saprolite at 1 65 meters and then decreases to 0 09 0 06 in the uppermost soil sample at 0 15 meters At Lee Hill Road the trend continues moving from a 26Mg value of -0 17 0 06 in the bedrock to -0 04 0 06 in the saprolite at 2 20 meters and then decreasing to -0 14 0 06 in the soil at 0 30 meters The Hurricane Hill hydrothermal sample shows a major increase in one near surface sample 0 5 m from a bedrock value of -0 07 0 06 to 0 55 0 06 at the surface Figure 4 However the profiles generally exhibit a small increase in Mg isotope composition with decreasing depth from the bedrock until the upper 2 meters in which a decrease is observed in the non-hydrothermal profiles Figure 4 While the bedrock compositions of these samples vary widely from -0 17 to 0 07 the effects of weathering within each profile are apparent and reveal a consistent trend of Mg isotope fractionation Magnolia Road Cave Creek South Lee Hill Road Hurricane Hill Figure 4: 26Mg variations with depth for the three weathering profiles Magnolia Road Cave Creek South and Lee Hill Road and one hydrothermal weathering profile Hurricane Hill Error bars represent two times the standard deviation 0 06 Data are reported in Table 2 5 Discussion In this section we discuss three possible mechanisms for Mg isotope fractionation during granite weathering: lateral deposition primary mineral alteration and dissolution and the 57 formation alteration and dissolution of secondary minerals We then compare these granite and granodiorite weathering profiles with other weathering profiles that have been analyzed for Mg isotopes to date Finally we discuss the implications of constraining the effects of granite weathering on Mg isotopes for Mg cycling on the continents the historical shift of continental crust composition and the reconstruction of paleo-weathering environments 5 1 Lateral Transport Aeolian deposition in the upper section of a weathering profile can alter its Mg isotope composition e g Liu et al 2014 Loess for example is known to demonstrate a variety of Mg isotopic compositions depending on its source material and transport processes Huang et al 2013 Li et al 2010 Wimpenny et al 2014b Because carbonate minerals tend to be isotopically light in regard to Mg loess containing a significant carbonate fraction typically has a low Mg isotope composition while more silicate-rich loess tends to be heavier Huang et al 2013 Li et al 2010 Liu et al 2014 Wimpenny et al 2014b To produce the observed nearsurface decrease in 26Mg the deposited material would almost certainly need to contain a significant carbonate fraction However the calcite concentration in these profiles never exceeds 0 3 wt% and no other carbonate minerals are observed so it is unlikely that carbonate could play a role in controlling Mg isotope composition Table 1 Therefore while aeolian sediments may be present in the near-surface portion of the profiles they are unlikely to be responsible for the observed shift towards lighter 26Mg values given the low levels of carbonate present throughout the profiles The lateral transport of less weathered silicate material from upslope may also play a role in the decrease in Mg isotope composition in the near surface samples Liu et al 2014 The Cave Creek South and Lee Hill Road locations exhibit an increase in biotite content in this 58 portion of the sections most likely indicating the deposition of less-weathered sediment Figure 1 The addition of relatively pristine material could result in a decrease in 26Mg given that rock gradually loses light Mg to the hydrosphere during weathering However no systematic increase in Mg concentration is observed which one would expect if such deposition had occurred Figure 2 With the current data it is not possible to quantify the extent of these additions given that the nature of the source materials and the fraction of the total Mg that they contributed are unknown but given the absence of increased Mg concentrations the effect on Mg isotopes is likely minimal 5 2 Primary and secondary mineral behavior Previous weathering profile studies focusing on mafic lithologies have found that 26 Mg 24Mg ratios generally increase with increased weathering intensity Huang et al 2012 Liu et al 2014 Teng et al 2010b Pogge von Strandmann et al 2012 a trend which is also observed in these more felsic profiles Magnesium in the unweathered granite and granodiorite bedrock in these profiles is primarily contained in biotite and the loss of biotite in the rock may affect its Mg isotope composition Ryu et al 2011 In these profiles however very little Mg is removed from the rock until the upper two meters of the profile Figure 2 The Mg remains within secondary minerals primarily illite with minor contributions from oxides and other clays This transition from primary biotite to secondary minerals results in limited Mg isotope fractionation because nearly all of the Mg is preserved in the rock retaining the original isotope composition The small decrease in Mg concentration at depths below two meters is likely responsible for the small increase in Mg isotope composition in those sections as heavy Mg is preferentially incorporated into secondary clay minerals while light Mg is lost to the hydrosphere The weathering of primary minerals largely biotite therefore has a small but 59 observable impact on Mg isotopes in these granite and granodiorite weathering environments imparting a relatively heavy isotopic signature in the residual rock at depths greater than 1 5-2 m The Mg isotope composition of these weathering profiles is mainly controlled by the formation and loss of secondary clay minerals A decrease in the Mg isotope ratios is observed in the upper two meters of each non-hydrothermal profile primarily due to changes in secondary mineral abundance particularly the illite content Illite can be produced in situ during weathering or hydrothermal activity as an alteration product of feldspars and micas and can subsequently be lost due to further weathering Meunier and Velde 2004 The Mg in illite like in biotite is primarily structural while Mg in clays such as kaolinite is primarily adsorbed onto surface and interlayer sites Drever 1988 Structural sites in clays and similar minerals show a systematic preference for the heavy isotope of Mg while Mg adsorbed onto clay surfaces is typically isotopically light Opfergelt et al 2012 2014 Wimpenny et al 2014a As the illite fraction of total major Mg-bearing minerals biotite illite chlorite and smectite increases so does the Mg isotope composition Figure 5 However the upper two meters of the non-hydrothermal profiles show a decrease in Mg isotope composition and the Magnolia Road and Cave Creek South locations also show a general decrease in illite Figures 1 and 4 As illite is weathered in the profile Mg likely shifts from being primarily present in structural sites to being adsorbed onto surface sites in minerals such as kaolinite resulting in heavy Mg being lost to the hydrosphere and leaving the solid phase enriched in light Mg The strong positive correlation between 26Mg and the illite fraction of total major Mg-bearing minerals biotite illite chlorite and smectite is likely primarily due to a combination of in situ illite production below 2 m and subsequent loss in the near surface samples Figure 5 60 Magnolia Road Cave Creek South Lee Hill Road Hurricane Hill 26Mg 0 6 0 4 0 2 0 -0 2 26Mg 0 6 0 4 0 2 0 -0 2 0 0 2 0 4 0 6 0 8 1 0 0 2 0 4 0 6 0 8 1 Illite Total Mg-bearing Minerals Figure 5: 26Mg variation with illite fraction of Mg-bearing minerals expressed in terms of weight percent The Mg-bearing minerals considered here are illite biotite smectite and chlorite illite fraction wt% illite wt% illite wt% biotite wt% smectite wt% chlorite The Mg contribution from other minerals is negligible Error bars represent two times the standard deviation for 26Mg 0 06 Data are reported in Tables 1 and 2 The observed Mg isotope fractionation associated with Mg loss to the hydrosphere can be modeled using Rayleigh fractionation as also observed in the weathering profiles of other rock types Teng 2017 Note that Rayleigh fractionation models provide only a limited approximation of isotope behavior in natural weathering systems The models assume a constant fractionation factor between the solid and liquid phases during rock leaching However a natural weathering system is comprised of many complex chemical interactions and some of these processes such as adsorption onto secondary minerals are not accounted for in the Rayleigh models The equation governing Rayleigh fractionation is 26Mgresidual rock 26Mgbedrock 1000 f 1 1 1000 where f fraction of bedrock Mg remaining in the weathered sample Mgresidual 26Mg 24Mg residual rock Mgbedrock rock 26 The apparent fractionation factor values Mg 24Mg fluid vary from 1 00001 to 1 00040 for these samples Figure 6 similar to the values observed during the weathering of mafic rocks 1 00005 1 0004 Liu et al 2014 The Hurricane Hill hydrothermal profile samples reflect a higher 61 Literature Data Figure 6: Panel A shows Mg concentration wt% MgO correlated with 26Mg for all measured samples Magnolia Road red squares Cave Creek South blue squares Lee Hill Road green squares and Hurricane Hill purple squares profiles are included as well as additional bedrock samples black squares from the area Dashed curves represent Rayleigh distillation models of Mg loss and Mg isotope fractionation for three different values of the fractionation factor top: 1 00040 middle: 1 00015 bottom: 1 00001 The orange triangle represents the pristine bedrock used in the Rayleigh distillation models MgO wt% 3 73 and 26Mg -0 20 Panel B shows the experimental colored squares and modeling data dashed lines from Panel A as well as weathering profile data from other regions and lithologies grey circles Literature data from Huang et al 2012 Liu et al 2014 Pogge von Strandmann et al 2012 and Teng et al 2010b The dotted lines represent Rayleigh distillation models of Mg removal from pristine bedrock and its effect on Mg isotope composition For the upper line 1 00040 and for the bottom line 1 00001 The blue triangle represents the pristine bedrock used in the Rayleigh distillation models MgO wt% 16 0 and 26Mg -0 25 Data are reported in Table 2 fractionation factor falling near 1 00040 while the Lee Hill Road samples exhibit minimal fractionation with Mg loss and therefore have a low near 1 00001 Figure 6 This range of values is similar to that found in other weathering profiles Figure 6b Despite the complications associated with the clay mineralogies in these profiles they still display behavior broadly typical of Rayleigh fractionation 5 3 Comparison with other weathering profiles Compared to the other weathering profiles analyzed for Mg isotopes primarily with mafic lithologies these granite and granodiorite profiles exhibit a small degree of fractionation 62 however this difference is unsurprising given the mineralogical and weathering conditions present in each profile Profiles developed on diabase and basalt bedrock exhibit more extreme fractionation -0 24 to 1 81 in one basalt profile for example Figure 6b Liu et al 2014 The samples in those profiles have experienced much more extreme Mg depletion often greater than 99% while the non-hydrothermal granite and granodiorite profiles have only lost up to 64% Mg Table 2 Huang et al 2012 Liu et al 2014 Teng et al 2010b The hydrothermal profile lost 83% Mg which may explain its more dramatic Mg isotope fractionation than the other profiles Table 2 Given that one of the dominant controls on Mg isotope fractionation in weathering environments is the preferential loss of residue enriched in 26 24 Mg to the hydrosphere leaving a rock Mg it makes sense that more extreme Mg depletion will produce more extreme Mg isotope fractionation The explanation for the more extreme weathering observed in these other profiles is likely primarily the result of mineralogical differences The abundant Mg-bearing minerals in the mafic profiles primarily pyroxene are more readily weathered than biotite in these felsic profiles Figure 7: 26Mg correlation with Chemical Index of Weathering where CIW Al2O3 Al2O 3 CaO Na2O 100 Cation concentrations are expressed in terms of weight percent Magnolia Road red squares Cave Creek South blue squares Lee Hill Road green squares and Hurricane Hill purple squares profiles are included as well as additional bedrock samples black squares from the area The grey symbols represent literature data from other weathering profiles The dashed black line expresses the average upper continental crust Mg isotope composition 26Mg -0 22 Li et al 2010 Data are reported in Table 2 so Mg is more readily lost to the hydrosphere Ehlers and Blatt 1982 The Mg isotope fractionation in the granite and granodiorite profiles is less dramatic than that previously observed in mafic profiles however Mg isotopes remain informative regarding weathering in these lithologies 63 5 4 Implications Granite and granodiorite weathering processes play a major role in Mg cycling in many continental surface environments and Mg isotopes could help elucidate these processes In these environments the sources of Mg in groundwater soils and the biosphere may be limited to the weathering of felsic rocks and aeolian deposition To understand Mg isotope behavior in these regions it is therefore important to have a clear picture of these weathering processes and how they affect the isotopic composition of the Mg that is released into near-surface environments In particular the ability to differentiate the effects of primary mineral alteration and dissolution secondary mineral formation alteration and dissolution and lateral transport would permit significant insights into surface processes The variety of trends observed in these four profiles is evident in the correlation between 26Mg and the chemical index of weathering CIW a measure of weathering and alteration intensity demonstrating the complexity of Mg isotope behavior in these systems Figure 7 The Cave Creek South and Hurricane Hill samples show a strong positive correlation while the Magnolia Road and Lee Hill Road profiles show no variation in 26Mg with CIW Figure 7 Magnesium isotopes are clearly sensitive to the specific mechanisms of chemical weathering sensitive enough to exhibit different behaviors even within profiles from the same locale and a complete understanding of this Mg isotope behavior could provide valuable insight into these complex weathering systems Magnesium isotope behavior during continental weathering may also help explain the role of preferential Mg weathering in the compositional shift of the continental crust from basalt to andesite During weathering MgO Na2O and CaO are leached from the crustal rock into the hydrosphere while other components such as SiO2 Fe2O3 and TiO2 are more likely to be retained in the rock Garrels and McKenzie 1971 Na2O remains in solution or is precipitated as 64 an evaporite and CaO is precipitated as carbonate but MgO is primarily fixed in oceanic crust and ultimately subducted Holland 1984 Rudnick 1995 MgO is thus recycled from the crust back into the mantle which may have contributed to the transition in the crust from more mafic lithologies to the modern andesitic composition Rudnick 1995 The present study demonstrates that with the exception of the near surface samples the weathering of granite and granodiorite results in a progressive increase in 26Mg in the residual rock as MgO is lost to the hydrosphere which is in agreement with previous studies of the effects of weathering on other lithologies Huang et al 2012 Liu et al 2014 Ma et al 2015 Pogge von Strandmann et al 2012 Teng et al 2010b The Mg isotope composition of the continental crust varies significantly -0 52 to 0 92 Li et al 2010 If sedimentary and formerly sedimentary material comprises a significant fraction of the crust the crust overall should evolve towards a heavy Mg isotope composition which can be tested by systematic studies of crustal compositions through time Magnesium isotope fractionation during the weathering of felsic lithologies particularly the fractionation effects observed during the formation of secondary minerals may be applied as an indicator of past climate and lithological conditions The type and composition of clays and other secondary minerals are controlled by the lithology of the rock being weathered and environmental factors such as pH temperature and water availability so these variables indirectly determine the Mg isotope composition of a profile For example the transition from biotite with primarily structural Mg to kaolinite with primarily exchangeable Mg is associated with a change in Mg isotope composition Drever 1988 Opfergelt et al 2012 2014 Wimpenny et al 2014a In pristine weathering profiles the clay mineralogy itself could be used a tracer of the weathering environment but in paleosol profiles that have been heavily altered by metamorphism the clay composition may not be preserved However the Mg isotope 65 composition of a rock is not significantly affected by metamorphism Geske et al 2012 Li et al 2011 Wang et al 2015 so this isotopic indicator of climate may remain While this study focuses Mg isotope behavior during the transitions between biotite illite and kaolinite the relationships between other primary and secondary minerals during weathering could be used in a similar way With a thorough understanding of the relationship between Mg isotopes and clay mineralogy Mg isotopes in preserved weathering profiles may be used to reconstruct paleoweathering environments Huang et al 2016 6 Conclusions Based on the Mg isotope analysis of four weathering profiles from granite and granodiorite lithologies from Boulder Creek Colorado USA the following conclusions are drawn: 1 The Mg isotope composition of three non-hydrothermal weathering profiles varies from -0 17 0 06 to 0 07 0 06 in the bedrock to -0 14 0 06 to 0 09 0 06 near the surface and -0 07 0 06 to 0 55 0 06 in one hydrothermal section 2 Magnesium isotope fractionation is primarily controlled by the illite content in the rocks increasing with illite production at depths greater than 2 m then decreasing in the near surface samples most likely from illite weathering with a small contribution from the lateral transport of less weathered material 3 Magnesium isotope fractionation can be described by Rayleigh distillation with values from 1 00001 to 1 00040 4 With additional study magnesium isotopes may be used as paleoclimate indicators in preserved felsic weathering profiles even altered profiles because they are sensitive to clay mineralogy 66 Acknowledgements We would like to thank Shui-Jiong Wang Yan Hu and Lie-Meng Chen for their assistance throughout this study Khadijah Karrington Homolka for her work in the clean lab Keith Morrison and Tom Bristow for their help with the XRD analyses and Kang-Jun Huang and Diana Carlsson for their careful editing and insightful comments This work was supported by NSF grant EAR-1340160 References Brenot A Cloquet C Vigier N Carignan J France-Lanord C 2008 Magnesium isotope systematics of the lithologically varied Moselle river basin France Geochim Cosmochim Acta 72 5070-5089 Dethier D P Bove D J 2011 Mineralogical and geochemical changes from alteration of granitic rocks Boulder Creek Catchment Colorado Vadose Zone J 10 858-866 Drever J I 1988 The Geochemistry of Natural Waters Prentice Hall New Jersey Eberl D D 2009 User s guide to rockjock: A program for determining quantitative mineralogy from powder X-ray diffraction data U S Geological Survey Open-file report 03-78 Edmond J M Measures C McDuff R E Chan L H Collier R Grant B Gordon L I Corliss J B 1979 Ridge crest hydrothermal activity and the balances of the major and minor elements in the ocean: The Galapagos data Earth Planet Sci Lett 46 1-18 Ehlers E G Blatt H 1982 Petrology: Igneous Sedimentary and Metamorphic W H Freeman and Company California Foster G L Pogge von Strandmann P A E Rae J W B 2010 Boron and magnesium isotopic composition of seawater Geochem Geophys Geosys 11 1-10 Garrels R M MacKenzie F T 1971 Evolution of sedimentary rocks Norton New York Geske A Zorlu J Richter D K Buhl D Niedermayr A Immenhauser A 2012 Impact of diagenesis and low grade metamorphosis on isotope 26Mg 13C 18O and 87Sr 88Sr and elemental Ca Mg Mn Fe and Sr signatures of Triassic sabkha dolomites Chem Geol 332333 45-64 Higgins J A Schrag D P 2010 Constraining magnesium cycling in marine sediments using magnesium isotopes Geochim Cosmochim Acta 74 5039-5053 Hofmann A W 2003 Sampling mantle heterogeneity through oceanic basalts: Isotopes and 67 trace elements in: Carlson R W Ed The Mantle and Core Treatise on Geochemistry Elsevier-Pergamon Oxford pp 61-101 Holland H D 1984 The chemical evolution of the atmosphere and oceans Princeton Univ Press Princeton NJ Huang K -J Teng F -Z Wei G -J Ma J- L Bao Z -Y 2012 Adsorption- and desorptioncontrolled magnesium isotope fractionation during extreme weathering of basalt in Hainan Island China Earth Planet Sci Lett 359-360 73-83 Huang K -J Teng F -Z Elsenouy A Li W -Y Bao Z -Y 2013 Magnesium isotopic variations in loess: Origins and implications Earth Planet Sci Lett 374 60-70 Huang K -J Teng F -Z Shen B Xiao S Lang X Ma H -R Fu Y Peng Y 2016 Episode of intense chemical weathering during the termination of the 365 Ma Marinoan glaciation PNAS 113 520 14904-14909 Immenhauser A Buhl D Richter D Niedermayer A Reichelmann D Dietzel M Schulte U 2010 Magnesium-isotope fractionation during low-Mg calcite precipitation in a limestone cave Field study and experiments Geochim Cosmochim Acta 74 4346-4364 Kelley S A Chapin C E 2004 Denudation history and internal structure of the Front Range and Wet Mountains Colorado based on apatite-fission-track thermochronology New Mexico Bureau of Geology & Mineral Resources Bulletin 160 41-77 Leonard E M 1989 Climatic change in the Colorado Rocky Mountains: Estimates based on modern climate at late Pleistocene equilibrium lines Arctic and Alpine Research 21 245-255 Li W -Y Teng F -Z Ke S Rudnick R L Gao S Wu F -Y Chappell B W 2010 Heterogeneous magnesium isotopic composition of the upper continental crust Geochim Cosmochim Acta 74 6867 6884 Li W -Y Teng F -Z Xiao Y Huang J 2011 High-temperature inter-mineral magnesium isotope fractionation in eclogite from the Dabie orogeny China Earth Planet Sci Lett 304 224-230 Ling M -X Sedaghatpour F Teng F -Z Hays P D Strauss J Sun W 2011 Homogeneous magnesium isotopic composition of seawater: an excellent geostandard for Mg isotope analysis Rapid Commun Mass Spectrom 25 2828-2836 Liu S A Teng F -Z He Y Ke S Li S 2010 Investigation of magnesium isotope fractionation during granite differentiation: Implication for Mg isotopic composition of the continental crust Earth Planet Sci Lett 297 646-654 Liu X -M Teng F -Z Rudnick R R McDonough W F Cummings M L 2014 Massive magnesium depletion and isotope fractionation in weathered basalts Geochim Cosmochim Acta 135 336 349 68 Ma L Teng F -Z Jin L Ke S Yang W Gu H -O Brantley S L 2015 Magnesium isotope fractionation during shale weathering in the Shale Hills Critical Zone Observatory: Accumulation of light Mg isotopes in soils by clay mineral transformation Chem Geol 397 3750 Meunier A Velde B 2004 Illite: Origins Evolution and Metamorphism Springer-Verlag Berlin Heidelberg New York Opfergelt S Georg R B Delvaux B Cabidoche Y -M Burton K W Halliday A N 2012 Mechanisms of magnesium isotope fractionation in volcanic soil weathering sequences Guadeloupe Earth Planet Sci Lett 341-344 176-185 Opfergelt S Burton K W Georg R B West A J Guicharnaud R A Sigfusson B Siebert C Gislason S R Halliday A N 2014 Magnesium retention on the soil exchange complex controlling Mg isotope variations in soils soil solutions and vegetation in volcanic soils Iceland Geochim Cosmochim Acta 125 110-130 Plank T 2014 The chemical composition of subducting sediments in: Rudnick R L Ed The Crust Treatise on Geochemistry Elsevier Oxford pp 607-629 Pogge von Strandmann P A E Opfergelt S Lai Y -J Sigfusson B Gislason S R Burton K W 2012 Lithium magnesium and silicon isotope behavior accompanying weathering in a basaltic soil and pore water profile in Iceland Earth Planet Sci Lett 339-340 11-23 Pokrovsky B G Mavromatis V Pokrovsky O S 2011 Co-variation of Mg and C isotopes in late Precambrian carbonates of the Siberian Platform: A new tool for tracing the change in weathering regime Chem Geol 290 67-74 Rudnick R L Gao S 2003 Composition of the Continental Crust in: Carlson R W Ed The Crust Treatise on Geochemistry Elsevier-Pergamon Oxford pp 1-64 Ryu J -S Jacobson A D Holmden C Lundstrom C Zhang Z 2011 The major ion 44 40Ca 44 42Ca and 26 24Mg geochemistry of granite weathering at pH 1 and T 25 C: power-law processes and the relative reactivity of minerals Geochim Cosmochim Acta 75 6004 6026 Shen B Jacobsen B Lee C -T A Yin Q -Z Morton D M 2009 The Mg isotopic systematics of granitoids in continental arcs and implications for the role of chemical weathering in crust formation Proc Natl Acad Sci 106 20652-20657 Teng F -Z Wadhwa M Helz R T 2007 Investigation of magnesium isotope fractionation during basalt differentiation: Implications for a chondritic composition of the terrestrial mantle Earth Planet Sci Lett 261 84 92 Teng F -Z Li W -Y Ke S Marty B Dauphas N Huang S Wu F Y Pourmand A 2010a Magnesium isotopic composition of the Earth and chondrites Geochim Cosmochim Acta 74 4150 4166 69 Teng F -Z Li W -Y Rudnick R L Gardner L R 2010b Contrasting lithium and magnesium isotope fractionation during continental weathering Earth Planet Sci Lett 300 6371 Teng F -Z Yang W 2014 Comparison of factors affecting the accuracy of high-precision magnesium isotope analysis by multi-collector inductively coupled plasma mass spectrometry Rapid Commun Mass Spectrom 28 19-24 Teng F -Z Li W -Y Ke S Yang W Liu S -A Sedaghatpour S Wang S -J Huang K J Hu Y Ling M -X Xiao Y Liu X -M Li X -W Gu H -O Sio C K Wallace D A Su B -X Zhao L Chamberlin J Harrington M Brewer A 2015 Magnesium isotopic compositions of international geological reference materials Geostand Geoanal Res 29 329339 Teng F -Z 2017 Magnesium Isotope Geochemistry Rev Mineral Geochem vol 82 219-287 Thompson R S 1991 Pliocene environments and climates in the western United States Quat Sci Rev 10 115-132 Tipper E T Galy A Gaillardet J Bickle M J Elder- field H Carder E A 2006 The magnesium isotope budget of the modern ocean: Constraints from riverine magnesium isotope ratios Earth Planet Sci Lett 250 241 253 Tipper E T Galy A Bickle M J 2008 Calcium and magnesium isotope systematics in rivers draining the Himalaya-Tibetan-Plateau region: Lithological or fractionation control Geochim Cosmochim Acta 72 1057-1075 Tipper E T Calmels D Gaillardet J Louvat P Capmas F Dubacq B 2012 Positive correlation between Li and Mg isotope ratios in the river waters of the Mackenzie Basin challenges the interpretation of apparent isotopic fractionation during weathering Earth Planet Sci Lett 333-334 35-45 Urey H C 1952 The Planets: Their Origin and Development Yale University Press New Haven CT 245 pp Wang S -J Teng F -Z Li S -G Hong J -A 2015 Magnesium isotopic systematics of mafic rocks during continental subduction Geochim Cosmochim Acta 143 34-48 Wedepohl K H 1995 The composition of the continental crust Geochim Cosmochim Acta 59 1217-1232 Wimpenny J Gislason S R James R H Gannoun A Pogge von Strandmann P A E Burton K W 2010 The behavior of Li and Mg isotopes during primary phase dissolution and secondary mineral formation in basalt Geochim Cosmochim Acta 74 5259-5279 Wimpenny J Colla C A Yin Q -Z Rustad J R Casey W H 2014a Investigating the behavior of Mg isotopes during the formation of clay minerals Geochim Cosmochim Acta 128 178-194 70 Wimpenny J Yin Q -Z Tollstrup D Xie L -W Sun J 2014b Using Mg isotope ratios to trace Cenozoic weathering changes: A case study from the Chinese Loess Plateau Chem Geol 376 31-43 Yang W Teng F -Z Zhang H -F 2009 Chondritic magnesium isotopic composition of the terrestrial mantle: A case study of peridotite xenoliths from the North China craton Earth Planet Sci Lett 288 475 482 71 CHAPTER 4: Magnesium Isotopes as a Tracer of Crustal Materials in Volcanic Arc Magmas in the Northern Cascade Arc This chapter is published as: Brewer A B Teng F -Z Mullen E 2018 Magnesium isotopes as a tracer of crustal materials in volcanic arc magmas in the Northern Cascade Arc Frontiers in Earth Sciences 6:21 Abstract Fifteen North Cascade Arc basalts and andesites were analyzed for Mg isotopes to investigate the extent and manner of crustal contributions to this magmatic system The 26Mg of these samples vary from within the range of ocean island basalts the lightest being -0 33 0 07 to heavier compositions as heavy as -0 15 0 06 The observed range in chemical and isotopic composition is similar to that of other volcanic arcs that have been assessed to date in the circum-pacific subduction zones and in the Caribbean The heavy Mg isotope compositions are best explained by assimilation and fractional crystallization within the deep continental crust with a possible minor contribution from the addition of subducting slab-derived fluids to the primitive magma The bulk mixing of sediment into the primitive magma or mantle source and the partial melting of garnet-rich peridotite are unlikely to have produced the observed range of Mg isotope compositions The results show that Mg isotopes may be a useful tracer of crustal input into a magma supplementing traditional methods such as radiogenic isotopic and trace element data particularly in cases in which a high fraction of crustal material has been added 72 1 Introduction Volcanic arcs involve significant crust-mantle interactions particularly through assimilation and fractional crystallization as magma rises to the surface and or the addition of subducted sediment altered oceanic crust and slab-derived fluids to the mantle source or the primitive magma Kelemen et al 2007 Plank 2014 This varied crustal material can be a major determinant of the final composition of these igneous rocks Quantifying the crustal contribution to these magmas and the mechanism s by which these materials were added informs our understanding of the petrogenetic history of volcanic arcs motivating the development of chemical tracers of these processes The unique systematics of Mg isotopes in mantle and crustal materials offer a new approach for tracing crustal additions to arc magmas With a few exceptions most significant Mg isotope fractionation occurs at the low temperatures of Earth s surface while little fractionation occurs at high temperatures and pressures Teng 2017 and references therein The Mg isotope compositions of uncontaminated mid-ocean ridge basalts and of the mantle are 26Mg -0 25 0 07 and -0 25 0 04 respectively Teng et al 2010a Large deviations from the well-constrained mantle Mg isotope composition in an unweathered mafic rock may be indicative of crustal input to the magma Surface material displays a wide range of 26Mg values from -5 57 to 1 81 reflecting the variety of fractionation mechanisms possible at low temperatures Teng 2017 During chemical weathering for example light isotopes are preferentially removed from a rock which in combination with variations in watershed lithology produces isotopically light river Pogge von Strandmann et al 2008 Tipper et al 2008 Teng et al 2010b Huang et al 2012 and ocean water -0 83 0 07 Foster et al 2010 Ling et al 2011 The residual silicate rock and sediment is typically isotopically heavy due to these same 73 processes as heavy as 1 81 Tipper et al 2008 2010 Li et al 2010 Teng et al 2010b Liu et al 2014 This chemical weathering and other associated processes such as carbonate precipitation have produced an extremely heterogeneous upper crust in regard to Mg isotopes and the subduction of this material has also resulted in isotopically heterogeneous lower crust and mantle wedge material Li et al 2010 Teng et al 2013 Yang et al 2016 Wang et al 2017 Recent investigations into the Mg isotope systematics of volcanic arc systems in the circumPacific subduction zones and in the Caribbean have found samples with a wide range of Mg isotope compositions -0 35 0 05 to 0 06 0 04 which has been attributed to variable slab-derived fluid additions Figure 1 Teng et al 2016 Li et al 2017 To explore the effects of crustal contamination on Mg isotopes in volcanic arcs we measured 15 samples from the Figure 1: Mg isotope composition of volcanic arc samples Lesser Antilles data from Teng et al 2016 Kamchatka Philippines Costa Rica and Lau data from Li et al 2017 North Cascade data from the present study The black line and green bar represent the mantle composition based on peridotite xenoliths 26Mg -0 25 0 04 from Teng et al 2010a North Cascade data are reported in Table 1 Error bars represent the 2SD for each sample North Cascade Volcanic Arc where both uncontaminated primary magma and contaminated evolved magmas have been previously observed These samples are geochemically well characterized and thought to record a range of crustal content due the addition of subducted material and or crustal assimilation during magma transport Figure 2 Mullen and Weis 2013 2015 Mullen and McCallum 2014 74 Mullen et al 2017 Our results show that the Mg isotope compositions of these arc magmas vary from within the established values for uncontaminated mantle-derived mafic magmas to relatively heavy compositions likely due to crustal contamination Figure 1 The use of Mg isotopes as a tracer of crustal material in magma is not limited to specific sites but could be applicable to a variety of volcanic arc settings Teng et al 2016 Li et al 2017 1800 Mt Baker Bridge River Cones Glacier Peak Chilliwack Batholith Primitive Magma ent 1600 ucte d Co mpo n 1200 1000 800 Sub d Ba ppm 1400 a Kamchatka Philippines Costa Rica Lau Primitive Mantle Depleted Mantle 600 400 elting Partial M 200 0 0 5 10 15 20 25 30 35 40 Nb ppm 0 7039 b 0 7038 Subducted Component 87 Sr 86Sr 0 7037 0 7036 0 7035 0 7034 0 7033 0 7032 0 7031 C AF 0 703 0 7029 0 200 400 600 800 1000 1200 Sr ppm 1400 1600 1800 2000 Figure 2: a Ba and Nb contents of the North Cascade samples as well as literature data from other volcanic arcs where available The first labeled gray arrow approximates the effects of the addition of a subducted component bulk sediment prior to 10% partial melting of the depleted mantle and the second approximates the effects of partial melting of the primitive mantle for the North Cascade based on Mullen and Weis 2013 Primitive mantle composition is from Sun and McDonough 1989 Depleted mantle composition is from Salters and Stracke 2004 b Sr isotope composition and Sr content of the North Cascade samples as well as literature data from other volcanic arcs where available The first labeled gray arrow approximates the effects of the addition of a subducted component bulk sediment to a primitive magma and the second approximates the effects of assimilation and fractionalization of a gabbro assimilant for the North Cascade based on Mullen and Weis 2013 The primitive magma composition is based on Mullen and Weis 2013 The colored circles represent those samples identified as most primitive from Mt Baker and the Bridge River Cones while the colored triangles represent samples that exhibit crustal contamination North Cascade data are reported in Table 1 Literature data are represented by gray symbols and are from Li et al 2017 2 Samples The Cascade Arc located on the western margin of North America is the result of the subduction of the Juan de Fuca plate beneath the North American plate Most of the associated magmatic rocks contain geochemical evidence for the addition of subducted material derived from both oceanic crust and sediment to their mantle sources Mullen et al 2017 In addition some magmas assimilated continental crust during transit to the surface Mullen et al 2017 We choose 15 samples from the North Cascade Volcanic Arc for Mg isotope analysis in an effort to 75 cover samples exhibiting a wide range of known crustal content from negligible to considerable based on trace element compositions and Sr-Nd-Hf-Pb isotope data The analyzed samples represent a range of locations and rock types within the northern segment of the Cascade Arc including the Mt Baker volcanic field the Bridge River Cones Glacier Peak and the Chilliwack batholith Tepper 1996 Mullen and Weis 2013 2015 Mullen and McCallum 2014 Mullen et al 2017 To ensure that no weathering has occurred all samples were collected from the interior of lava flows were confirmed to have low LOI contents 99% of the Mg in the sample to prevent isotope fractionation within the column and limits the concentrations of the other elements to less than 5% of the concentration of Mg Teng et al 2007 The purified Mg samples were then analyzed on a Nu Plasma MC-ICP-MS using the standard-sample bracketing method The Mg concentrations of the sample and standard solutions were matched to within 5% to prevent mass bias caused by a concentration mismatch Teng and Yang 2014 The solutions contained approx 300 ppb Mg in 3% nitric acid The samples were introduced using the wet plasma method and 24 Mg 25 Mg and 26 Mg were analyzed simultaneously using three Faraday cups H5 Ax and L4 Results are presented in delta notation: 77 %& 1 1000 % &% where x refers to mass 25 or 26 The precision of the measured 26Mg 24Mg ratio for one sample solution at the 2SD level based on repeat standard analyses during a single analytical session is 0 07 comparable to previous Mg isotope studies Li et al 2010 Teng et al 2010a Ling et al 2011 Teng and Yang 2014 The reference materials San Carlos Olivine n 4 and Seawater n 3 were each analyzed at least once during an analytical session each time yielding a 26Mg value within the established 2SD of -0 25 and -0 83 respectively These values agree with previously published data Teng et al 2015 Hu et al 2016a 4 Results Magnesium isotopic compositions MgO concentrations relevant trace element data and Sr isotope compositions for the North Cascade Arc are reported in Table 1 The Mt Baker samples range from 26Mg -0 32 to -0 15 0 06 the Bridge River Cones samples range from 26Mg -0 33 to -0 22 0 07 the Mg isotope composition of the Chilliwack Batholith and Glacier Peak samples is 26Mg -0 23 and -0 22 0 07 respectively Weak correlations between 26Mg and MgO content 87Sr 86Sr Th Yb and Pb Ce may be evident in these samples but given the number of processes that can affect the chemical and isotopic composition of arc magmas clear correlations are not expected Figures 3 and 5 No correlation is observed between Mg isotope composition and Ba Th Dy Yb or Sm Yb Figures 4 and 5 The increase in 26Mg from the MORB-like composition is small however it is resolvable with the current precision 0 07 78 Sample North Cascade Arc Mt Baker Tarn Plateau Sulphur Creek Lake Shannon Park Butte Cathedral Crag Coleman Pinnacle 101B Coleman Pinnacle 105 Coleman Pinnacle 86 Table Mountain Bridge River Cones Tuber Hill East Dot Tuber Hill East Cap Tuber Hill East Plateau Nichols Valley Chilliwack Batholith Mount Sefrit Glacier Peak Dishpan Gap Standards Seawater Replicate Replicate SC Olivine Duplicate Duplicate Replicate 26Mg 2SD 25Mg 2SD MgO wt% Sm Yb Dy Yb Th Yb Pb Ce Ba Th Ba ppm Nb ppm Sr ppm 87 Sr 86Sr -0 28 -0 23 -0 29 -0 32 -0 15 -0 15 0 07 0 06 0 07 0 06 0 07 0 06 -0 19 -0 13 -0 19 -0 18 -0 11 -0 09 0 06 0 04 0 06 0 04 0 06 0 04 7 76 5 42 6 44 8 38 3 96 1 36 2 199 1 878 1 586 1 470 2 944 1 929 2 078 2 069 1 918 1 939 2 219 1 688 1 295 0 446 0 410 0 303 1 163 3 326 0 122 0 084 0 100 0 088 0 097 0 192 155 8 206 7 195 5 331 7 229 0 120 0 335 279 215 199 426 894 3 57 7 40 5 13 2 56 4 95 10 16 860 563 486 502 1194 603 0 703109 0 703240 0 703213 0 703156 0 703513 0 703383 -0 17 -0 20 -0 17 0 06 0 07 0 07 -0 10 -0 08 -0 12 0 04 0 06 0 06 2 84 2 59 3 08 3 949 3 110 1 963 2 178 1 991 1 817 3 010 2 749 1 846 0 185 0 187 0 169 158 2 148 5 132 1 938 894 585 9 77 9 65 7 79 1801 1394 673 0 703718 0 703686 0 703475 -0 22 -0 30 -0 33 -0 25 0 06 0 07 0 07 0 07 -0 08 -0 16 -0 16 -0 14 0 04 0 06 0 06 0 06 3 44 4 60 3 10 4 44 2 272 2 928 2 752 3 043 1 685 2 432 2 314 2 361 1 568 0 808 0 625 0 816 0 173 0 043 0 077 0 037 257 9 143 1 212 6 140 0 655 289 440 287 7 0 34 0 32 8 36 4 618 571 514 567 0 703495 0 703042 0 703186 0 703052 -0 23 0 07 -0 09 0 06 6 80 2 550 2 015 1 329 0 164 161 3 300 3 19 799 0 703441 -0 22 0 07 -0 14 0 06 4 47 2 129 1 817 1 378 0 139 112 7 374 3 87 631 0 703566 -0 81 -0 88 -0 83 -0 23 -0 25 -0 31 -0 24 0 07 0 07 0 07 0 07 0 07 0 06 0 06 -0 43 -0 54 -0 50 -0 05 -0 20 -0 15 -0 11 0 06 0 06 0 06 0 06 0 06 0 04 0 04 Table 1: Magnesium isotope and trace element compositions of samples and standards North Cascade major and trace element data from Mullen et al 2017 Mullen and McCallum 2013 2014 Mullen and Weis 2015 2SD Two standard deviation of the population of n 15 n 22 repeat measurements of the standards during an analytical session 79 5 Discussion 5 1 Mg isotope variations in the North Cascade Volcanic Arc The alkali basalts from the Bridge River Cones are essentially free of crustal contamination based on trace element and Sr-Nd-Hf-Pb isotope data a conclusion which is supported by Mg isotope systematics Figure 2 Mullen and Weis 2013 The samples with 26Mg between -0 33 and -0 25 0 07 fall within the range of Mg isotope compositions exhibited by MORBs 26Mg -0 31 to -0 19 and OIBs 26Mg -0 35 to -0 18 Teng et al 2010a They do not reach the heavier compositions found elsewhere in the North Cascade for example at Mt Baker Figure 3 Thus the Mg isotope composition of these basalts supports the conclusion that the Bridge River Cones likely reflect the melting of a primitive mantle source with little to no contribution from subducted material or assimilated continental crust Figure 3 Mullen and Weis 2013 The calc-alkaline arc basalts from Mt Baker Glacier Peak and the Chilliwack batholith and the andesites from Mt Baker and the Bridge River Cones do show the influence of subducted material and or assimilation and fractional crystallization of the continental crust Mullen and McCallum 2014 While all of these samples are thought to exhibit some crustal contamination we consider the Tarn Plateau and Park Butte samples to be the most primitive given their high MgO contents 7 wt% high Mg 0 6 and relatively high compatible trace element content e g Cr 200 ppm Mullen and Weis 2013 2015 Mullen and McCallum 2014 The Mg isotope compositions of these samples are also quite light -0 28 0 07 and 0 32 0 07 respectively similar to the primitive alkali basalts from the Bridge River Cones which are known to contain little crustal contamination Figure 3 The remainder of the calcalkaline basalts and andesites display variable crustal involvement from subducted material 80 and or assimilated continental crust Tepper 1996 Mullen and Weis 2013 2015 Mullen and McCallum 2014 The relatively heavy Mg isotope compositions of these samples likely reflects the addition of isotopically heavy crustal material such as subducted sediment 26Mg -3 65 to 0 52 altered oceanic crust 26Mg -2 76 to 0 21 and deep continental crust 26Mg -0 76 to 0 19 Figure 3 Huang 2013 Teng et al 2013 Hu et al 2017 Yang et al 2016 The following sections will examine the possible causes of the elevated Mg isotope compositions of these arc samples a b Figure 3: a Variation of 26Mg with wt% MgO b Variation of 26Mg with 87Sr 86Sr The colored circles represent those samples identified as most primitive from Mt Baker and the Bridge River Cones while the colored triangles represent samples that exhibit crustal contamination The black line and green bar represent the mantle composition based on peridotite xenoliths 26Mg -0 25 0 04 from Teng et al 2010a North Cascade data are reported in Table 1 Error bars represent the 2SD for each sample 5 2 Hypotheses for Mg Isotope Variations The observed increase in Mg isotope composition may be the result of one of three processes or a combination thereof: the primary melt was derived from partial melting of garnetrich peridotite isotopically heavy subducted material was added to the primitive magmas or the mantle source or the primitive magmas underwent assimilation and fractional crystallization during transport within the continental crust 81 Garnet has a light Mg isotope composition compared to coexisting silicates Li et al 2011 2016 Wang et al 2012 2014 Huang et al 2013 The difference in Mg coordination between garnet 8-fold and most silicate minerals 6-fold causes 24 Mg to be preferentially incorporated into garnet leaving the surrounding minerals enriched in 26 Mg Li et al 2011 2016 Wang et al 2012 2014 Huang et al 2013 Both equilibrium and disequilibrium intermineral fractionation can occur between garnet and coexisting silicates Li et al 2016 Therefore partial melting that leaves a garnet-rich a residue could produce a melt with a relatively heavy Mg isotope composition However the trace element data do not support this explanation for the origin of the isotopically heavy magmas Because of b the affinity of the HREE for garnet partial melts leaving a garnet-rich residue would also display relatively high Sm Yb and Dy Yb ratios Lassiter and DePaolo 1997 Therefore 26Mg should be Figure 4: Variation of 26Mg with a Sm Yb and b Dy Yb ratios The colored circles represent those samples identified as most primitive from Mt Baker and the Bridge River Cones while the colored triangles represent samples that exhibit crustal contamination North Cascade data are reported in Table 1 Literature data are represented by gray symbols and are from Li et al 2017 Error bars represent the 2SD for each North Cascade sample positively correlated with Sm Yb and Dy Yb which is not observed in the North Cascade or any other volcanic arc assessed to date Figure 4 Furthermore previous work has ruled out residual garnet for Mt Baker and Glacier Peak based on trace element modeling Mullen and Weis 2013 2015 Residual garnet associated with the Bridge River Cones is likely however those samples do not exhibit heavy Mg isotope compositions Mullen and Weis 2013 We therefore conclude 82 that partial melting in the presence of garnet is unlikely to have produced the observed Mg isotope compositions The addition of a crustal component to the magmas is more consistent with the Mg isotope data Because crustal material such as subducted a sediment generally has lower MgO contents than the mantle source and primitive Average magmas McDonough and Sun 1995 Rudnick and Gao b 2003 Plank 2014 crustal input into a magma is generally associated with a decrease in MgO Fractional crystallization can produce similar changes in chemical composition however isotopes with the exception of fractional crystallization does not fractionate Mg c processes associated with garnet Teng 2017 and references therein A heavy Mg isotope composition in these samples is also associated with increased 87 Sr 86Sr which is expected given that crustal materials have elevated 87 Sr 86Sr compared to most mantle-derived melts Kelemen et al 2007 Plank 2014 Slab-derived fluid additions to the mantle wedge are typically associated with increases in Pb Ce and Ba Th ratios while Figure 5: Variation of 26Mg with Ba Th a Th Yb b and Pb Ce c The black triangles represent the average MORB composition MORB 26Mg value -0 25 0 07 is from Teng et al 2010a Th Yb 0 1113 Pb Ce 0 0384 and Ba Th 72 2772 are from Gale et al 2013 The colored circles represent those samples identified as most primitive from Mt Baker and the Bridge River Cones while the colored triangles represent samples that exhibit crustal contamination Literature data are represented by gray symbols and are from Li et al 2017 North Cascade data are reported in Table 1 Error bars represent the 2SD for each North Cascade sample 83 subducted sediment melts are typically associated with increased Th Yb ratios Weak correlations between 26Mg and Pb Ce and Th Yb may be present in these samples but no correlation with Ba Th is observed Figure 3 Taken together the North Cascade Volcanic Arc samples exhibit a range of Mg isotope compositions that is best explained by crustal additions rather than garnet involvement 5 3 Modeling Crustal Input To investigate the origin of crustal contributions to the North Cascade Arc an AFC model DePaolo 1981 combining primitive magma and lower continental crust and two different two-component bulk-mixing models primitive magma subducted sediment and Sample End-Member Compositions Depleted Mantle Primitive Magma Lower Crust Subducted Sediment 26Mg 1 -0 25 -0 30 -0 052 0 203 MgO wt% 4 38 2 8 38 6 575 2 536 Sr ppm 4 9 8 502 4015 3236 87 Sr 86Sr depleted mantle subducted sediment were constructed for the North Cascade Arc Figure 4 0 70260 0 70311 0 704005 0 714906 Modeling Parameters DSr 3 27 DMgO 3 07 r 0 57 6 The models are designed to be representative of the calc-alkaline arc basalts from Mt Baker Glacier Peak and the Chilliwack batholith and 1 Teng et al 2010a 2 Yang et al 2016 3 Hu et al 2017 4 Salters and Stracke 2004 5 Mullen and Weis 2013 6 Plank 2014 7 Teng et al 2016 Table 2: Modeling parameters and endmember compositions the andesites from Mt Baker and the Bridge River Cones but not the alkali basalts from the Bridge River Cones since they are known to be petrogenetically distinct from the other samples All end-member compositions and other model parameters are listed in Table 2 The primitive magma composition was selected based on the samples judged to be the most primitive in the Mt Baker dataset Tarn Plateau and Park Butte The depleted mantle composition is from Salters and Stracke 2004 with the exception of the Mg isotope composition -0 25 which is the canonical mantle value described in Teng et al 2010a and 84 a b Figure 6: AFC assimilation-fractional crystallization and bulk mixing models for 26Mg vs wt% MgO a and 87Sr 86Sr b for the North Cascade samples and available literature data The colored circles represent those samples identified as most primitive from Mt Baker and the Bridge River Cones while the colored triangles represent samples that exhibit crustal contamination The dashed and dotted line represents bulk mixing between depleted mantle and subducted sediment the dashed line represents bulk mixing between primitive magma and subducted sediment the solid line represents assimilation and fractional crystallization of lower continental crust by the primitive magma Table 2 contains the end-member compositions modeling parameters and their sources The black circles along the modeled curves represent 10% bulk mixing increments and 10% crystallization increments in the AFC model The black triangle represents the initial primitive magma end-member Error bars represent the 2SD for each North Cascade sample North Cascade data are reported in Table 1 Literature data grey symbols are from Li et al 2017 and Teng et al 2016 85 elsewhere Huang et al 2011 Hu et al 2016b Wang et al 2016 The lower continental crust composition is based on the North Cascade lower crust end-member from Mullen and Weis 2013 The Mg isotope composition of the lower crust in this region is unknown so a reasonable composition -0 05 was selected based on the range exhibited by lower crustal material in Yang et al 2016 Finally the subducted sediment end-member is based on the subducted sediment at the Cascade from Plank 2014 However the Mg isotope composition of the subducted sediment component affecting the mantle source or primitive magma is unknown so again a reasonable value 0 20 was selected based on Hu et al 2017 For the AFC models the ratio of assimilation to crystallization was fixed at 0 5 and the bulk partition coefficients for MgO and Sr were estimated and fixed at 3 0 and 3 2 respectively after the volcanic arc modeling of Teng et al 2016 Minor changes in the bulk partition coefficients have little impact on the model Although some of the relevant modeling parameters can only be estimated these models do demonstrate that AFC is a feasible explanation for the observed data while bulk mixing alone is not The North Cascade Arc two-component bulk mixing models are unable to reproduce the observed trends in the data particularly the Sr isotope compositions To approximate the MgO content and Mg isotope composition of some North Cascade samples with the primitive magma mixing model the sediment must contribute more than 50% of the final magma which is unreasonably high Figure 6a The Sr and Mg isotope primitive magma mixing model predicts very little change in Mg isotope composition across the whole range of Sr isotope compositions observed in the samples which does not agree with the observed data Figure 5b The twocomponent mantle source mixing model requires an even higher fraction of crustal material more than 80% to produce the observed Mg isotope compositions given the high MgO content 86 of the mantle and again no change in Mg isotope composition is predicted for the observed range in Sr isotopes Figure 5 Therefore bulk sediment additions to the primitive magma or mantle source are unlikely to be the cause of the elevated Mg isotope compositions found in the North Cascade Arc samples Assimilation and fractional crystallization in the deep continental crust on the other hand can match the observed trends without an excessively high crustal contribution The observed major element and isotopic trends can be reproduced with between 100% and 60% liquid remaining in the system Figure 6 Unlike the bulk mixing models AFC can approximate the observed increase in Mg isotope composition within the range of Sr isotope compositions exhibited by the samples Figure 6 Given the elevated Pb Ce and Ba Th ratios and the apparent correlation between 26Mg and Pb Ce in these samples fluid additions may also have made a minor contribution to the final Mg isotope compositions However quantifying the slab-derived fluid effect on Mg isotopes is not currently possible given our limited understanding of Mg isotope behavior during slab dehydration Assimilation and fractional crystallization is the favored explanation for the increase in Mg isotope composition and while minor effects from the addition of slab-derived fluids to the mantle source and primitive may have occurred those processes are unlikely to the primary cause of the heavy Mg isotope compositions observed in the North Cascade Volcanic Arc The Mg isotope compositions exhibited by the North Cascade Volcanic Arc samples are similar to those from previously analyzed volcanic arcs Figure 1 Li et al 2017 analyzed arc samples from Kamchatka 26Mg -0 35 0 05 to -0 24 0 08 the Philippines -0 19 0 05 to 0 06 0 04 Costa Rica -0 32 0 01 to -0 27 0 05 and Lau -0 26 0 02 to -0 16 0 05 while Teng et al 2016 analyzed samples from the Lesser Antilles -0 24 0 07 to - 87 0 10 0 07 The Cascade samples -0 33 0 07 to -0 15 0 06 have similar Mg isotope compositions to the samples from all of those locations except the Philippines which have unusually heavy compositions Li et al 2017 The present dataset fills in a gap in the Mg isotope data for arc volcanics from the circum-Pacific subduction zones representing the margin along the northwestern coast of North America The similarity to these other arc samples including those from the Lesser Antilles in the Caribbean demonstrates that the processes affecting Mg isotopes are likely consistent from margin to margin Much of the combined data for the volcanic arcs worldwide can be approximated by the same assimilation and fractional crystallization model used with the North Cascade samples Figure 6 AFC processes are therefore the most likely explanation for the range in Mg isotope data observed in volcanic arcs Small differences in the Mg isotope compositions of the primitive magma subducted sediment and assimilated lower crust between and within different arcs likely do impact the final magma composition producing the observed variety between the arcs Although crustal input to the North Cascade magmas produced only small just beyond two-sigma analytical uncertainty variations from normal mantle values this likely required a significant crustal addition Nonetheless improvements in analytical precision may make possible the use of Mg isotopes as a valuable tracer of crustal recycling that is complementary to established methods such as Sr-Nd-Hf-Pb isotopic and trace element data The crustal materials that may be involved in arc volcanism including subducted sediment and sediment melt altered oceanic crust slab-derived fluids and assimilated continental crust have extremely varied Mg isotope compositions This variability along with the lack of fractionation during most high temperature processes may permit Mg isotopes to be a valuable tool in tracing arc volcanic processes With a thorough understanding of the composition of these different reservoirs Mg 88 isotope systematics combined with existing geochemical indicators would elucidate the nature of crustal input in a given arc Furthermore Mg isotopes may provide a solution for detecting crustal input in settings where commonly-used radiogenic isotopes are ambiguous due to lack of isotopic contrast between the crust and primary magmas e g Mullen et al 2017 These findings also represent a significant step in our understanding of the Mg cycle by demonstrating that silicate material from the crust can alter the isotopic composition of a magma 6 Summary The main conclusions from this study are: 1 The Mg isotope composition of samples from the North Cascade Arc range from -0 33 0 07 to -0 15 0 06 2 The alkali basalts from the Bridge River Cones reflect the partial melting of the mantle source with minimal crustal contamination while the calc-alkaline basalts and andesites from Mt Baker Glacier Peak Chilliwack Batholith and the Bridge River Cones do have Mg isotope compositions indicative of variable crustal contamination 3 The samples with high 26Mg are best explained by the addition of isotopically heavy deep continental crust to primitive magmas through assimilation and fractional crystallization with a possible minor contribution from slab-derived fluids 4 With further study and analytical improvements Mg isotopes will become a useful tool in understanding the generation and evolution of magmatic rocks Acknowledgements We would like to thank Shui-Jiong Wang for his extensive help throughout this study Florence Yuen Khadijah Karrington Homolka and Jiarui Zhou for their work in the clean lab and Kwan-Nang Pang and I Stewart McCallum for their insightful comments P Adam and J 89 Tepper are acknowledged for providing samples from the Bridge River Cones and the Chilliwack batholith respectively We would also like to thank Julia Ribeiro for her careful editing and Thomas Zambardi and two anonymous reviewers for their constructive comments on a previous version of this manuscript This work was supported by an NSF grant EAR17407706 References DePaolo DJ 1981 Trace element and isotopic effects of combined wallrock assimilation and fractional crystallization Earth Planet Sc Lett 53:189-202 doi:10 1016 0012-821X 81 90153-9 Foster GL Pogge von Strandmann PAE Rae JWB 2010 Boron and magnesium isotope composition of seawater Geochem Geophy Geosy 11 8 doi:10 1029 2010GC003201 Gale A Dalton CA Langmuir CH Su Y Schilling JG 2013 The mean composition of ocean ridge basalts Geochem Geophy Geosy 14 3 :1525-2027 doi:10 1029 2012GC004334 Hu Y Harrington MD Sun Y Yang Z Konter J Teng FZ 2016a Magnesium isotopic homogeneity of San Carlos olivine: a potential standard for Mg isotopic analysis by multicollector inductively coupled plasma mass spectrometry Rapid Commun Mass Sp 30 19 :21232132 doi:10 1002 rcm 7700 Hu Y Teng FZ Zhang HF Xiao Y Su BX 2016b Metasomatism-induced mantle magnesium isotopic heterogeneity: Evidence from pyroxenites Geochim Cosmochim Ac 185:88-111 doi:10 1016 j gca 2015 11 001 Hu Y Teng FZ Plank T Huang KJ 2017 Magnesium isotopic composition of subducting marine sediments Chem Geol 466:15-31 doi:10 1016 j chemgeo 2017 06 010 Huang F Chen LJ Wu ZQ Wang W 2013 First-principles calculations of equilibrium Mg isotope fractionations between garnet clinopyroxene orthopyroxene and olivine: Implications for Mg isotope thermometry Earth Planet Sc Lett 367:61-70 doi:10 1016 j epsl 2013 02 025 Huang F Zhang ZF Lundstrom CC Zhi XC 2011 Iron and magnesium isotopic compositions of peridotite xenoliths from Eastern China Geochim Cosmochim Ac 75:3318-3334 doi:10 1016 j gca 2011 03 036 Huang KJ 2013 The behavior of magnesium isotopes during low-temperature water-rock interactions PhD Dissertation China University of Geosciences Wuhan China Huang KJ Teng FZ Wei GJ Ma JL Bao ZY 2012 Adsorption- and desorption-controlled magnesium isotope fractionation during extreme weathering of basalt in Hainan Island China Earth Planet Sc Lett 359:73-83 doi:10 1016 j epsl 2012 10 007 90 Kelemen PB Hanghoj K Greene AR 2007 One view of the geochemistry of subductionrelated magmatic arcs with an emphasis on primitive andesite and lower crust In: Holland HD Turekian KK ed Treatise on Geochemistry 2nd Edition vol The Crust Elsevier Oxford pp 170 doi:10 1016 B0-08-0435751 03035-8 Lassiter JC Depaolo DJ 1997 Plume Lithosphere interaction in the generation of continental and oceanic flood basalts: Chemical and isotopic constraints In: Mahoney JJ Coffin MF ed Large Igneous Provinces: Continental Oceanic and Planetary Flood Volcanism American Geophysical Union Washington D C doi:10 1029 GM100p0335 Li SG Yang W Ke S Meng X Tian H Xu L et al 2017 Deep carbon cycles constrained by a large-scale mantle Mg isotope anomaly in eastern China Nat Sci Review 4 1 : 111-120 doi:10 1093 nsr nww070 Li WY Teng FZ Ke S Rudnick RL Gao S Wu FY et al 2010 Heterogeneous magnesium isotopic composition of the upper continental crust Geochim Cosmochim Ac 74 23 :6867-6884 doi:10 1016 j gca 2010 08 030 Li WY Teng FZ Xiao YL Gu HO Zha XP Huang J 2016 Empirical calibration of the clinopyroxene-garnet magnesium isotope geothermometer and implications Contrib Mineral Petrol 171:61 doi:10 1007 s00410-016-1269-1 Li WY Teng FZ Xiao YL Huang JA 2011 High-temperature inter-mineral magnesium isotope fractionation in eclogite from the Dabie orogen China Earth Planet Sc Lett 304 12 :224-230 doi:10 1016 j epsl 2011 01 035 Ling MX Sedaghatpour F Teng FZ Hays PD Strauss J Sun WD 2011 Homogeneous magnesium isotopic composition of seawater: an excellent geostandard for Mg isotope analysis Rapid Commun Mass Sp 25 19 :2828-2836 doi:10 1002 rcm 5172 Liu XM Teng FZ Rudnick RL McDonough WF Cummings ML 2014 Massive magnesium depletion and isotope fractionation in weathered basalts Geochim Cosmochim Ac 135:336-349 doi:10 1016 j gca 2014 03 028 McDonough WF Sun SS 1995 The Composition of the Earth Chem Geol 120 3-4 :223-253 doi:10 1016 0009-2541 94 00140-4 Mullen EK McCallum IS 2013 Coexisting pseudobrookite ilmenite and titanomagnetite in hornblende andesite of the Coleman Pinnacle flow Mount Baker Washington: Evidence for a highly oxidized arc magma Am Mineral 98 2-3 :417-425 doi:10 2138 am 2013 4185 Mullen EK McCallum IS 2014 Origin of Basalts in a Hot Subduction Setting: Petrological and Geochemical Insights from Mt Baker Northern Cascade Arc J Petrol 55 2 :241-281 doi:10 1093 petrology egt064 Mullen EK Weis D 2013 Sr-Nd-Nf-Pb isotope and trace element evidence for the origin of alkali basalts in the Garabaldi Belt northern Cascade arc Geochem Geophy Geosy 14:31263155 doi:10 1002 ggge 20191 91 Mullen EK Weis D 2015 Evidence for trench-parallel mantle flow in the northern Cascade Arc from basalt geochemistry Earth Planet Sc Lett 414:100-107 doi:10 1016 j epsl 2015 01 010 Mullen EK Weis D Marsh NB Martindale M 2017 Primitive arc magma diversity: New geochemical insights in the Cascade Arc Chem Geol 448:43-70 doi:10 1016 j chemgeo 2016 11 006 Plank T 2014 The chemical composition of subducting sediments In: Holland HD Turekian KK ed Treatise on Geochemistry 2nd Edition vol The Crust Elsevier Oxford pp 607-629 doi:10 1016 B978-0-08-095975-7 00319-3 Pogge von Strandmann PAE Burton KW James RH van Calsteren P Gislason SR Sigfusson B 2008 The influence of weathering processes on riverine magnesium isotopes in a basaltic terrain Earth Planet Sc Lett 276 1-2 :187-197 doi:10 1016 j epsl 2008 09 020 Rudnick RL Gao S 2003 Composition of the Continental Crust In: Rudnick RL Holland HD Turekian KK ed Treatise on Geochemistry vol The Crust Elsevier Oxford pp 1-64 doi:10 1016 B0-08-043751-6 03016-4 Salters VJM Stracke A 2004 Composition of the depleted mantle Geochem Geophy Geosy 5 5 doi:10 1029 2003GC000597 Sun W McDonough WF 1989 Chemical and isotopic systematics of oceanic basalts: Implications for mantle composition and processes Geological Society London Special Publications vol 42 313-345 doi:10 1144 GSL SP 1989 042 01 19 Teng FZ 2017 Magnesium Isotope Geochemistry Reviews in Mineralogy and Geochemistry vol 82 219-287 doi:10 2138 rmg 2017 82 7 Teng FZ Li WY Ke S Yang W Liu SA Sedaghatpour F et al 2015 Magnesium isotopic composition of international geostandards Geochem Geophy Geosy 39:329-339 doi: 10 1111 j 1751-908X 2014 00326 x Teng FZ Hu Y Chauvel C 2016 Magnesium isotope geochemistry in arc volcanism P Natl Acad Sci USA 113 26 :7082-7087 doi:10 1073 pnas 1518456113 Teng FZ Li WY Ke S Marty B Dauphas N Huang SC et al 2010a Magnesium isotopic composition of the Earth and chondrites Geochim Cosmochim Ac 74 14 :4150-4166 doi:10 1016 j gca 2010 04 019 Teng FZ Li WY Rudnick RL Gardner LR 2010b Contrasting lithium and magnesium isotope fractionation during continental weathering Earth Planet Sc Lett 300 1-2 :63-71 doi:10 1016 j epsl 2010 09 036 Teng FZ Wadhwa M Helz RT 2007 Investigation of magnesium isotope fractionation during basalt differentiation: Implications for a chondritic composition of the terrestrial mantle Earth Planet Sc Lett 261 1-2 :84-92 doi:10 1016 j epsl 2007 06 004 92 Teng FZ Yang W 2014 Comparison of factors affecting the accuracy of high-precision magnesium isotope analysis by multi-collector inductively coupled plasma mass spectrometry Rapid Commun Mass Sp 28 1 :19-24 doi:10 1002 rcm 6752 Teng FZ Yang W Rudnick RL Hu Y 2013 Heterogeneous magnesium isotopic composition of the lower continental crust: A xenolith perspective Geochem Geophy Geosy 14 9 :3844-3856 doi:10 1002 ggge 20238 Tepper JH 1996 Petrology of mafic plutons associated with calc-alkaline granitoids Chilliwack Batholith north Cascade Washington J Petrol 37 6 :1409-1436 doi:10 1093 petrology 37 6 1409 Tipper ET Galy A Bickle MJ 2008 Calcium and magnesium isotope systematics in rivers draining the Himalaya-Tibetan-Plateau region: Lithological or fractionation control Geochim Cosmochim Ac 72 4 :1057-1075 doi:10 1016 j gca 2007 11 029 Wang SJ Teng FZ Li SG Hong JA 2014 Magnesium isotopic systematics of mafic rocks during continental subduction Geochim Cosmochim Ac 143:34-48 doi:10 1016 j gca 2014 03 029 Wang SJ Teng FZ Li SG Zhang LF Du JX He YS et al 2017 Tracing subduction zone fluid-rock interactions using trace element and Mg-Sr-Nd isotopes Lithos 290-291:94-103 doi:10 1016 j lithos 2017 08 004 Wang SJ Teng FZ Scott JM 2016 Tracing the origin of continental HIMU-like intraplate volcanism using magnesium isotope systematics Geochim Cosmochim Ac 185:78-87 doi:10 1016 j gca 2016 01 007 Wang SJ Teng FZ Williams HM Li SG 2012 Magnesium isotopic variations in cratonic eclogites: Origins and implications Earth Planet Sc Lett 359:219-226 doi:10 1016 j epsl 2012 10 016 Yang W Teng FZ Li WY Liu SA Ke S Liu YS et al 2016 Magnesium isotopic composition of the deep continental crust Am Mineral 101 1-2 :243-252 doi:10 2138 am-20165275 Yang W Teng FZ Zhang HF 2009 Chondritic magnesium isotopic composition of the terrestrial mantle: A case study of peridotite xenoliths from the North China craton Earth Planet Sc Lett 288 3-4 :475-482 doi:10 1016 j epsl 2009 10 009 93 CHAPTER 5: Recovery of rare earth elements from geothermal fluids through bacterial cell surface adsorption This chapter is in revision for publication as: Brewer A B Chang E Park D M Kou T Li Y Lammers L N Jiao Y 2018 Recovery of rare earth elements from geothermal fluids through bacterial cell surface adsorption Environmental Science & Technology Abstract Rare earth elements REEs are in increasing demand in the modern economy and yet meaningful REE production is limited to only a few locations worldwide which motivates the development of novel strategies to enable cost-effective REE recovery from non-traditional feedstocks We investigate biosorption as a potential means of recovering REEs from geothermal fluids a low-grade but abundant REE source We have previously engineered E coli to express lanthanide binding tags LBTs on the cell surface and the resulting strain showed an increase in both REE adsorption capacity and selectivity Here we examined how REE adsorption by the engineered E coli is affected by various geochemical factors relevant to geofluids including total dissolved solids TDS temperature pH and the presence of specific competing metals REE biosorption is robust to high TDS concentrations with high extraction efficiency and selectivity observed in geofluids containing REE concentrations as low as 100 ppb and TDS as high as 165 000 ppm Among several metals tested U Al and Pb were found to be the most competitive causing significant reductions 25% in REE biosorption when present at concentrations 3 to 11-fold higher than the REEs Optimal REE biosorption occurs between pH 5-6 with significant loss in sorption capacity 65% as pH decreases from 6 to 2 REE 94 extraction efficiency and selectivity increase as a function of temperature up to 70 C which can be explained by the thermodynamic properties of metal complexation on the bacterial surface Together these data demonstrate the potential utility of biosorption for selective REE recovery from geothermal fluids by defining the optimal and boundary conditions for this extraction technology Introduction Rare earth elements REEs are becoming increasingly significant to the international economy with the emergence and development of new technologies particularly in the area of clean energy Common applications of REEs include automotive and industrial catalysts permanent magnets and electronics However the supply of these metals is uncertain and potentially at risk globally As of 2011 more than 95% of REE production came from China even though there are significant reserves worldwide 1 Given the limited sources of REE production and the increasing demand for these metals around the world it is crucial to explore new REE feedstocks and to develop new and improved methods of REE extraction 1-5 Biosorption has gained interest in recent years as a potential clean sustainable method for REE recovery The high binding affinity of native cell surfaces for REEs relative to most nonREEs permits selective extraction of these valuable metals from low-grade feedstocks 4 6-8 Microbes are relatively inexpensive to produce in large quantities and native biomass has a naturally high capacity for REE adsorption 4 9-13 Cell surfaces can withstand multiple cycles of adsorption and desorption enabling reuse of biomass 14 Furthermore biosorption processes are not expected to contribute hazardous chemical wastes to a REE extraction scheme unlike some conventional methods such as solvent extraction 15 16 Microbial surface adsorption could present 95 a clean and effective means of REE recovery from a variety of feedstocks including those that are low-grade and traditionally unexploited We have previously bioengineered Escherichia coli to express lanthanide binding tags LBTs on the cell surface to enhance their natural adsorptive properties improving both extraction efficiency and selectivity for REEs 17 LBT display increased REE adsorption capacity from 13 to 28 mg Tb g dry cell weight in a simple buffer solution and also improved REE binding selectivity 2 to 10-fold over most non-REE metals present in mine tailing leachates 17 The native cell surface functional groups primarily carboxylate and phosphate 18-21 do continue to play a major role in REE recovery with the LBT-displayed cells In this study we investigate the REE adsorption performance of the engineered LBT-displayed strain of E coli under various geochemical conditions characteristic of geofluids Geothermal fluids are abundant low-grade feedstocks that are currently being investigated as a potential source of REEs 4 6 22 23 Total REE concentrations reported in geofluids range from sub-ppb to low-ppm levels 24 Fluids with REE concentrations at ppb levels or higher are nearly always acidic pH 9 adsorption desorption cycles with minimal loss in REE recovery performance Furthermore the microparticles exhibit a strong preference for heavy REEs particularly Eu Sm Yb and Lu compared to the light 122 REEs which may permit the separation of individual REEs with an appropriate method design The results of this study represent a major step towards transitioning engineered microbes from promising REE adsorbents to a practical industrial scale REE extraction technology Introduction Rare earth elements REEs are critical components of many green energy technologies such as wind turbines and consumer products including mobile phones laptops and appliances 1 However greater than 85% of the global REE supply is obtained from China leaving the global market vulnerable to supply restrictions 2 Given the lack of economically exploitable REE ore deposits in many locations various abundant non-traditional REE resources e g geothermal fluids 3 coal byproducts 4 5 and electronic wastes6 are currently being explored as potential REE feedstocks The adverse chemical characteristics of many of these feedstocks relative to high-grade ores including high concentrations of non-REE metals and low REE contents present a significant challenge for industrial-scale exploitation using traditional extraction technologies The development of new REE extraction methodologies that can effectively exploit these non-traditional feedstocks is thus important for diversifying the REE supply chain and reducing dependency on foreign imports Adsorption-based solid-phase extraction systems offer promise for extracting REEs from low-grade feedstocks Many adsorbents have been tested 7 including bioengineered microbes 8-10 organic wastes such as seafood11 and vegetable byproducts 12 various hydrogels 13 zero-valent iron particles 14 and mineral powders 15 Each of these materials has advantages and disadvantages for REE extraction For example organic wastes are inexpensive abundant and non-hazardous however they may not have the high selectivity for REEs that an engineered hydrogel will exhibit 7 Zero-valent iron particles on the other hand are highly adsorbent but degrade at some environmentally-relevant conditions 14 An ideal adsorbent would exhibit both a 123 high capacity and a high selectivity for REEs and be inexpensive readily available mechanically and chemically stable and environmentally friendly Biosorption offers a sustainable method for REE extraction and purification that is well suited for non-traditional REE feedstocks Native microbial surfaces typically have a high binding affinity for REEs compared to most non-REE metals enabling REEs to be selectively concentrated on the cell surface relative to the aqueous phase 16-19 Microbes are relatively inexpensive to produce in large quantities exhibit fast adsorption and desorption kinetics and can be reused over multiple REE recovery cycles 8 20-24 Furthermore unlike solvent extraction which is the current industrial standard for REE separation biosorption processes would not contribute to the production of hazardous chemical wastes 25 26 Recent work from our group and others has revealed the potential utility of genetic engineering to enhance the already high REE binding capacity and selectivity of native microbial surfaces Genetic alteration of teichoic acids a major cell wall component of Bacillus subtilis 27 and heterologous expression and display of lanthanide binding peptides on the surface of Caulobacter crescentus8 and E coli9 has enhanced microbial REE adsorption capacity and selectivity Although the adsorptive characteristics of these microbes are appropriate for selective REE extraction there are several drawbacks to their use in industrial-scale applications in their native form including a tendency to obstruct fluid flow relatively low cell stability and the risk of cells escaping into the feedstock 28 Immobilizing the cells often by imbedding them in a permeable matrix such as a polymer circumvents these issues limiting obstruction to flow stabilizing the cells and preventing their release into the fluid phase The ideal carrier material would be biocompatible highly permeable and mechanically and chemically stable To date several natural and synthetic polymers including polyacrylamide 29 silica 30 polyvinyl alcohol 31 and more commonly 124 alginate 31-34 have been used to immobilize cells for metal adsorption applications in matrix-type capsules or microparticles With cells uniformly distributed throughout the material these microparticles exhibit both the advantageous chemical characteristics of the biological material and the advantageous physical characteristics of the polymer Polyethylene glycol diacrylate PEGDA is an attractive polymer matrix for cell immobilization as it is highly customizable capable of being modified to display a range of physical and chemical characteristics and is nonadsorptive unlike other common polymers such as alginate and silica gel 35-38 For biomining purposes where the purity of the adsorbed metals is of paramount importance it is crucial to minimize any non-specific adsorption by the carrier material Finally adsorbed REEs must be recovered through targeted desorption with an eluent such as citrate or EDTA These eluents would degrade some polymer materials such as alginate which is ionically crosslinked with divalent metals but PEGDA remains stable even after long exposure to these chemicals With these considerations in mind we have synthesized PEGDA microparticles impregnated with a high concentration of LBT-displayed E coli cells to create an adaptable adsorbent suited for selective REE recovery in a continuous flow column Methods Bacterial strains and growth conditions The E coli strain with an lpp-ompA-dLBT expression plasmid was grown in LB media containing 50 g ml ampicillin Induction of LBT expression occurred at mid-exponential phase using 0 002% arabinose for 3 hr at 30 C Park et al 2016 and 2017 include a complete description of plasmid construction and LBT expression 8 9 Cells were harvested washed once in 0 9% NaCl and normalized by OD600 in preparation for microparticle synthesis 125 Microparticle synthesis Three components are required for microparticle synthesis 1 polymer precursor: 1% g g TPO-L photoiniator 2 4 6-Trimethylbenzoylphenyl phosphinic acid ethyl ester Rahn AG and 99% polyethylene glycol diacrylate Mn 575 Sigma Aldrich 2 cell suspension: 1x1011 cells ml OD600 40 LBT-displayed E coli in 10 mM MES pH 6 and 3 oil phase: 1% g g Triton X-100 surfactant Sigma Aldrich and 99% 10cSt PDMS oil polydimethylsiloxane Clearco Products 10 mM MES pH 6 was used as the cell suspension solution for all microparticle production batches except for one large batch used to pack a 315 ml column for which 0 9% NaCl was used The polymer precursor and cell suspension were first mixed at a 1:3 ratio 25% PEGDA by volume to produce the final aqueous phase The oil phase and aqueous phases were then mixed to a ratio of 7:1 vol vol in 40 ml silanized borosilicate vials The vials were vigorously shaken for 15 s by hand to produce an emulsion and were immediately exposed to UV 4 W cm2 at 365 nm for 120 s to polymerize the droplets comprised of PEGDA and LBT-displayed cells The PDMS oil was removed from the polymerized microparticles by vacuum filtration through a 20 m nylon mesh filter The microparticles were rinsed 10 times in 0 9% NaCl solution to remove any residual oil before being stored in 10 mM PIPES buffer pH 7 or 0 9% NaCl for the 315 ml column batch at 4 C until use Batch sorption experiments Microparticles were collected on a 20 m nylon mesh filter and any residual liquid was removed The microparticles were then distributed into 5 ml Eppendorf tubes at a controlled wetweight To determine the dry weight the particles were dried for 72 hours at 65 C and reweighed For the batch adsorption capacity experiment 4 5 ml of feedstock with varying Nd concentrations in 10 mM MES buffer pH 6 was added to 0 2 g wet weight of the 126 microparticles For the batch adsorption kinetics experiment 5 ml of 500 M Nd in 10 mM MES buffer pH 6 was added to 0 2 g wet weight of microparticles The tubes were mixed in a rotator for the duration of the experiments 30 min for the adsorption capacity tests and 1-120 min for the adsorption kinetics tests The Nd concentrations remaining in the solutions were tested colorimetrically with Arsenazo and the amount of Nd adsorbed was calculated based on mass balance Breakthrough columns Empty glass columns Biorad were packed with the cell PEGDA microparticles Column dimensions used were 15 x 0 5 cm 20 x 0 5 cm and 100 x 1 cm The microparticle suspension was initially added gravimetrically then packed down by adding slight pressure to the top of the column Under pressure the sorbent bed compressed and more microparticles were added This procedure was repeated until the entire column volume was filled The microparticle sorbent was conditioned with 10 mM MES buffer pH 6 at 5x the bed volume prior to REE adsorption The REE feedstock was then pumped through the column at a measured rate of 1 ml min unless otherwise specified For most tests the feedstock concentration was kept at 500 M Nd in 10 mM MES buffer pH 6 though exceptions are described and discussed below The effluent was collected in 1 ml aliquots during the adsorption step Following adsorption columns were desorbed with 1 5x fixed bed volume of 10 mM citrate and again conditioned with 10 mM MES buffer pH 6 prior to reuse The microparticles were removed from the columns that would no longer be used and dried down to assess dry adsorbent weight REE concentrations of the samples were assessed by ICP-MS and or colorimetrically with Arsenazo Non-REE concentrations were measured by ICP-MS 127 ICP-MS analysis Nd breakthrough column sample analyses were performed using a Thermo XSeriesII ICPMS run in standard mode at UC Santa Cruz The sample introduction system was an ESI PFAST nebulizer pumped at 120 l min Selectivity column sample analyses were conducted at Idaho National Laboratory on an Agilent 7900 ICP-MS run in either hydrogen or helium gas modes Results and Discussion Microparticle synthesis and characterization To encapsulate a high density of LBT-displayed E coli in PEGDA microparticles we employed a bulk emulsion and polymerization technique Microparticles were produced and tested over a range of cell densities and PEGDA concentrations in order to determine the optimal microparticle formulation As expected REE adsorption scaled in proportion to cell density between 1 5x1010 and 9 6x1010 cells ml Supplementary Figure 1a suggesting that cell surface availability was not limited by increased cell density within the tested range With a cell density above 4 3x1010 cells ml a higher PEGDA content 15 vol% was required to maintain the physical integrity of the produced microparticles data not shown This increase in PEGDA content had a minimal negative effect on REE sorption per unit wet weight between 10 and 25 vol% Supplementary Figure 1b As such all cell PEGDA microparticles described in this work were formulated with 10x1010 cells ml in 25 vol% PEGDA The microparticles were assessed in batch to determine their baseline REE adsorption behavior Nd was used for these experiments given its abundance in many REE feedstocks and its high criticality The calculated adsorption capacity was 2 9 mg Nd g dry particles This capacity represents a significant decrease 88 7% from the LBT-displayed cells alone which 128 have an adsorption capacity of 25 7 mg Nd g dry cells 9 This decrease is expected given that PEGDA which is not adsorptive contributes 86 6% of the total dry weight of the microparticles as discussed above The observed REE adsorption capacity is comparable to published values for some other adsorbents which range from 9 adsorption desorption cycles Figure 4: a Nd adsorption capacity and b Nd breakthrough behavior of the microparticles in a fixed-bed column after 1 to 9 adsorption desorption cycles The feedstock solution was 500 M Nd in 10 mM MES buffer pH 6 for each test 35 ml 10 mM citrate was used for desorption and then the column was washed with 100 ml 10 mM MES buffer pH 6 prior to the next adsorption cycle The column dimensions were 15 cm x 0 5 cm and flow rate was 1 ml min Selective REE Recovery A continuous flow column that is capable of achieving a degree of separation between the individual REEs would be advantageous because it would allow the user to have some control over the final product composition rather than just generating a total REE mixture The LBTdisplayed E coli cells have previously been shown to exhibit some selectivity between different REEs 9 and a sufficiently large column may be able to exploit this selectivity for REE separation A 100 x 1 cm column was packed with the microparticles to test REE selectivity with a synthetic 134 multi-REE influent The feedstock contained equal concentrations 500 M of Y and all of the lanthanides except Pm in 10 mM MES buffer pH 6 The sorbent generally exhibited a preference for the heavy REEs over the light REEs with a particular affinity for Sm Eu Yb and Lu Figure 5 Light REEs that were initially adsorbed onto the column prior to reaching adsorption capacity were ultimately desorbed and replaced by the heavy REEs as evidenced by the spike in La and Ce concentration in the effluent reaching concentrations above that of the initial feedstock as heavy REE adsorption increased Figure 5a This observed selectivity for heavy REEs would permit us to achieve a degree of the separation between the REEs in a breakthrough column and may be further exploited with a chromatography column system design to separate individual REEs or REE pairs Figure 5: a Breakthrough curves for all REEs except Pm and Sc from a 100 cm x 1 cm fixed bed column b Factor of concentration for each REE The initial feedstock had approx equal concentrations of all REEs in a 10 mM MES buffer pH 6 solution matrix The flow rate through the column was 1 ml min 135 Implications for REE extraction The LBT-displayed cells and a number of other REE adsorbents have shown great promise in terms of their REE adsorption capacity and selectivity however many of these adsorbents have not yet been incorporated into a practical industrial-scale REE recovery technology The requirements for an industrial-scale REE adsorbent go beyond simply its adsorptive characteristics In addition to having adequate REE adsorption capacity and kinetics the sorbent must be reusable stable in the presence of harsh REE feedstocks and eluents capable of being applied in continuous flow systems and easily and sustainably produced Based on these criteria the cell PEGDA microparticles are suited for REE extraction processes beyond the bench scale unlike many REE sorbents including LBT-displayed cells alone Immobilizing LBT-displayed cells in PEGDA microparticles creates a REE adsorbent that has the advantageous adsorptive characteristics of the cell surfaces and the advantageous physical characteristics of the polymer While the microparticles do have a lower adsorption capacity than the cells immobilization has successfully enabled the cells to be used in continuous flow column extraction systems The microparticles are also highly stable capable of being recycled and reused through many adsorption desorption cycles Furthermore the microparticles are relatively simple to synthesize in large quantities allowing the technology to be scaled to the requirements of the specific application This microparticle synthesis method could be easily scaled to rapidly produce large volumes of sorbent for industrial-scale REE recovery operations for example by incorporating an in-line emulsifier and UV source to continuously mix and polymerize the microparticles The microparticles are an effective green REE adsorbent that is readily scalable broadly reusable and can be applied in continuous flow systems 136 Acknowledgements We would like to thank Kevin Paulsen for his assistance silanizing the borosilicate vials AB acknowledges funding from the Livermore Graduate Scholar Program from Lawrence Livermore National Laboratory This work was performed under the auspices of the U S Department of Energy by Lawrence Livermore National Laboratory under Contract DEAC5207NA27344 LLNL-TH-763869 Supporting Information Figure S1 Tb adsorption capacity as a function of cell density and PEGDA content for the cell PEGDA microparticles Figure S2 Comparison of Nd breakthrough data for a single column based on colorimetric assays Arsenazo and ICP-MS analyses Figure S1: a M Tb adsorbed as a function of cell density in the cell PEGDA microparticles PEGDA was set at 10 vol% for all microparticles except for the highest tested point 10x1010 cells ml for which 15 vol% PEGDA was required to maintain particle integrity b M Tb adsorbed as a function of PEGDA fraction vol% in the cell PEGDA microparticles Cell density was set at 1 5x1010 cells ml In all sorption assays 0 5 g of microparticles was exposed to 4 ml of 300 M Tb in 10 mM MES buffer pH 6 in a 5 ml Eppendorf tube for 30 minutes except for the highest cell density 10x1010 cells ml which required 600 M Tb due to its high adsorption capacity Error bars represent the SD of the assays in triplicate 137 Arsenazo vs ICP-MS Figure S2: Comparison of a breakthrough arsenazo - 2 6 mg g curve with Nd concentrations measured ICP-MS - 2 6 mg g colorimetrically with Arsenazo black line and symbols and with an ICP-MS red line and symbols References 1 DOE Critical Materials Strategy In 2011 2 Zhou B Li Z Chen C Global potential of rare earth resources and rare earth demand from clean technologies Minerals 2017 7 1-14 3 Wood S A Behavior of rare earth elements in geothermal systems: A new exploration exploitation tool In DOE Ed Geothermal Reservoir Technology Research: 2001 4 Taggart R K Hower J C Dwyer G S Hsu-Kim H Trends in the Rare Earth Element Content of U S -Based Coal Combustion Fly Ashes Environ Sci Technol 2016 50 11 5919-5926 5 Qi H Hu R Zhang Q REE Geochemistry of the Cretaceous lignite from Wulantuga Germanium Deposit Inner Mongolia Northeastern China International Journal of Coal Geology 2007 71 2 329-344 6 Lixandru A Venkatesan P J nsson C Poenaru I Hall B Yang Y Walton A G th K Gau R Gutfleisch O Identification and recovery of rare-earth permanent magnets from waste electrical and electronic equipment Waste Management 2017 68 482489 7 Anastopoulos I Bhatnagar A Lima E C Adsorption of rare earth metals: A review of recent literature Journal of Molecular Liquids 2016 221 954-962 138 8 Park D M Reed D W Yung M C Eslamimanesh A Lencka M M Anderko A Fujita Y Riman R E Navrotsky A Jiao Y Q Bioadsorption of Rare Earth Elements through Cell Surface Display of Lanthanide Binding Tags Environ Sci Technol 2016 50 5 2735-2742 9 Park D M Brewer A Reed D W Lammers L N Jiao Y Recovery of Rare Earth Elements from Low-Grade Feedstock Leachates Using Engineered Bacteria Environ Sci Technol 2017 51 22 13471-13480 10 Jin H Y Park D M Gupta M Brewer A W Ho L Singer S L Bourcier W L Woods S Reed D W Lammers L N Sutherland J W Jiao Y Q Techno-economic Assessment for Integrating Biosorption into Rare Earth Recovery Process Acs Sustain Chem Eng 2017 5 11 10148-10155 11 Takahashi Y Kondo K Miyaji A Watanabe Y Fan Q Honma T Tanaka K Recovery and Separation of Rare Earth Elements Using Salmon Milt PLOS ONE 2014 9 12 e114858 12 Varshini J S C Nilanjana D Screening of biowaste materials for the sorption of cerium III from aqueous environment Research Journal of Pharmaceutical Biological and Chemical Sciences 2014 5 402-408 13 Bertagnolli C Grishin A Vincent T Guibal E Recovering Heavy Metal Ions from Complex Solutions Using Polyethylenimine Derivatives Encapsulated in Alginate Matrix Industrial & Engineering Chemistry Research 2016 55 8 2461-2470 14 Crane R A Sapsford D J Sorption and fractionation of rare earth element ions onto nanoscale zerovalent iron particles Chem Eng J 2018 345 126-137 15 Gok C Neodymium and samarium recovery by magnetic nano-hydroxyapatite Journal of Radioanalytical and Nuclear Chemistry 2014 301 3 641-651 16 Dodson J R Parker H L Garcia A M Hicken A Asemave K Farmer T J He H Clark J H Hunt A J Bio-derived materials as a green route for precious & critical metal recovery and re-use Green Chem 2015 17 4 1951-1965 17 Jacinto J Henriques B Duarte A C Vale C Pereira E Removal and recovery of Critical Rare Elements from contaminated waters by living Gracilaria gracilis J Hazard Mater 2018 344 531-538 18 Lo Y C Cheng C L Han Y L Chen B Y Chang J S Recovery of high-value metals from geothermal sites by biosorption and bioaccumulation Bioresource Technol 2014 160 182-190 19 Kucuker M A Wieczorek N Kuchta K Copty N K Biosorption of neodymium on Chlorella vulgaris in aqueous solution obtained from hard disk drive magnets PLoS One 2017 12 4 e0175255 20 Jiang M Y Ohnuki T Tanaka K Kozai N Kamiishi E Utsunomiya S Postadsorption process of Yb phosphate nano-particle formation by Saccharomyces cerevisiae Geochim Cosmochim Ac 2012 93 30-46 139 21 Moriwaki H Yamamoto H Interactions of microorganisms with rare earth ions and their utilization for separation and environmental technology Appl Microbiol Biot 2013 97 1 1-8 22 Ozaki T Gillow J B Kimura T Ohnuki T Yoshida Z Francis A J Sorption behavior of europium III and curium III on the cell surfaces of microorganisms Radiochim Acta 2004 92 9-11 741-748 23 Texier A C Andres Y Le Cloirec P Selective biosorption of lanthanide La Eu Yb ions by Pseudomonas aeruginosa Environ Sci Technol 1999 33 3 489-495 24 Tsuruta T Accumulation of rare earth elements in various microorganisms J Rare Earth 2007 25 5 526-532 25 Nash K L Jensen M P Analytical-scale separations of the lanthanides: A review of techniques and fundamentals Sep Sci Technol 2001 36 5-6 1257-1282 26 Xie F Zhang T A Dreisinger D Doyle F A critical review on solvent extraction of rare earths from aqueous solutions Miner Eng 2014 56 10-28 27 Moriwaki H Koide R Yoshikawa R Warabino Y Yamamoto H Adsorption of rare earth ions onto the cell walls of wild-type and lipoteichoic acid-defective strains of Bacillus subtilis Appl Microbiol Biot 2013 97 8 3721-3728 28 Zhu Y Chapter 14 - Immobilized Cell Fermentation for Production of Chemicals and Fuels In Bioprocessing for Value-Added Products from Renewable Resources Yang S -T Ed Elsevier: Amsterdam 2007 pp 373-396 29 Texier A C Andres Y Faur-Brasquet C Le Cloirec P Fixed-bed study for lanthanide La Eu Yb ions removal from aqueous solutions by immobilized Pseudomonas aeruginosa: experimental data and modelization Chemosphere 2002 47 3 333-42 30 Cabuk A Akar T Tunali S Tabak O Biosorption characteristics of Bacillus sp ATS-2 immobilized in silica gel for removal of Pb II J Hazard Mater 2006 136 2 31723 31 Tsekova K T D Ganeva S Removal of heavy metals from industrial wastewater by free and immobilized cells of Aspergillus niger International Biodeterioration & Biodegradation 2010 64 447-451 32 Xu J Song X C Zhang Q Pan H Liang Y Fan X W Li Y Z Characterization of metal removal of immobilized Bacillus strain CR-7 biomass from aqueous solutions J Hazard Mater 2011 187 1-3 450-8 33 Preetha B V T Batch and continuous biosorption of chromium VI by Rhizopus arrhizus Separation and Purification Technology 2007 57 126-133 34 Kuhn S P P R M Adsorption of mixed metals and cadmium by calcium-alginate immobilized Zoogloea ramigera Appl Microbiol Biot 1989 31 5-6 613-618 35 Michelini E Roda A Staying alive: new perspectives on cell immobilization for biosensing purposes Analytical and Bioanalytical Chemistry 2012 402 5 1785-1797 36 Lee K G Park T J Soo S Y Wang K W Kim B I I Park J H Lee C -S Kim D H Lee S J Synthesis and utilization of E coli-encapsulated PEG-based 140 microdroplet using a microfluidic chip for biological application Biotechnology and Bioengineering 2010 107 4 747-751 37 Son H A Choi S K Jeong E S Kim B Kim H T Sung W M Kim J W Microbial Activation of Bacillus subtilis-Immobilized Microgel Particles for Enhanced Oil Recovery Langmuir 2016 32 35 8909-8915 38 Akselrod G M Timp W Mirsaidov U Zhao Q Li C Timp R Timp K Matsudaira P Timp G Laser-Guided Assembly of Heterotypic Three-Dimensional Living Cell Microarrays Biophysical Journal 2006 91 9 3465-3473 39 Chu K H Hashim M A Copper biosorption on immobilized seaweed biomass: Column breakthrough characteristics Journal of Environmental Sciences 2007 19 8 928932 141 CHAPTER 7: Summary and Future Work The first portion of this dissertation investigates Mg isotope fractionation associated with biological and chemical weathering and weathered materials I first discuss a laboratory study of the effects of cell surface reactivity on Mg isotopes in a forsterite weathering system Next I characterize the Mg isotope fractionation associated granite and granodiorite weathering in natural environments Finally I conclude with the use of Mg isotopes to trace the fate of upper crustal materials in volcanic arc magmas The described results further our understanding of Mg isotope behavior during biological and chemical weathering and the effects of upper crustal material on the Mg isotope composition of volcanic arc rocks The main conclusions from these chapters are discussed below: 1 Chapter 2 discusses Mg isotope fractionation during forsterite dissolution in the presence of Bacillus subtilis endospores Because the endospores do not have an active metabolism or release significant organic acids their surface reactivity is the only way that they affect the weathering system The endospores therefore are ideal for demonstrating the isolated effects of cell surface reactivity on Mg isotopes during weathering As the forsterite dissolves 24 Mg is preferentially leached from the mineral into the aqueous phase and the presence of endospores has little impact on this process despite increasing initial dissolution rate However the endospore surfaces preferentially adsorb 24 Mg from the aqueous phase causing the Mg isotope composition of the liquid phase to increase in proportion to the fraction of aqueous Mg adsorbed With further study of the other ways in which microbes can fractionate Mg isotopes this geochemical tracer may function as a biosignature in weathered rock 142 2 Chapter 3 describes the Mg isotope composition of four granite and granodiorite weathering profiles from Boulder Creek Colorado USA As weathering progresses the primary biotite is lost but most Mg is retained in the solid phase as secondary illite A small increase in Mg isotope composition is observed at 2 m depth during this process as some 24Mg is preferentially lost to the hydrosphere However as weathering continues the secondary illite is lost with Mg retained in the profile adsorbed onto other clay minerals like kaolinite Exchangeable Mg is typically isotopically light which is reflected in a decrease in Mg isotope composition in this upper portion of the weathering profiles These results demonstrate the effects of chemical weathering on the Mg isotope composition of felsic lithologies which informs Mg cycling in many continental surface environments 3 Chapter 4 presents Mg isotope data for a suite of well-characterized samples from the North Cascade Volcanic Arc that are known to exhibit variable input from upper crustal materials Mullen and Weis 2013 2015 Mullen and McCallum 2014 Mullen et al 2017 The Mg isotope results in conjunction with trace element data demonstrate that the addition of isotopically heavy upper crustal material to upwelling magmas during assimilation and fractional crystallization caused a small but measurable increase in Mg isotope composition in some locations A small contribution to Mg isotope composition due to bulk mixing of subducted materials cannot be ruled out With further study Mg isotope systematics may supplement traditional geochemical tracers to monitor the addition of crustal materials to primitive magmas in a variety of geological settings Magnesium isotope behavior has become increasingly well characterized for a wide range of geological processes but many avenues remain for further investigation In the context of weathering specifically I believe that important advances could be made in our mechanistic 143 understanding of specific weathering processes Natural weathering systems are typically the result of a combination of several distinct processes that can be difficult to differentiate or quantify Primary mineral dissolution secondary mineral formation surface adsorption and uptake into plants for example could all play a role in determining the ultimate composition of a weathered rock Laboratory studies at controlled conditions can isolate the Mg isotope fractionation associated with a single process in isolation Armed with the results of a more comprehensive series of these laboratory studies a researcher could interpret Mg isotope data from natural weathering environments much more rigorously than is currently possible Part 2 The second portion of this dissertation describes the application of engineered E coli for the selective extraction of rare earth elements REEs from non-traditional feedstocks First I investigate the possibility of using these microbes to recover REEs specifically from natural geofluids The work assesses the chemical characteristics of these feedstocks such as pH and temperature in terms of their effect on REE biosorption Next I encapsulate the engineered microbes in polymer microcapsules to create a scalable hybrid adsorbent that may be practical and economically-feasible for industrial REE extraction and purification operations 1 Natural geofluids have several characteristics that present obstacles for REE extraction including low REE contents high total dissolved solids TDS high concentrations of competitive metals low pH and high temperatures Chapter 5 discusses the effects of these variables on REE adsorption by engineered E coli The beneficial adsorptive characteristics of the cells high capacity and high selectivity are retained at a wide range of chemical conditions relevant to geofluids The results obtained during this study will inform the use of biosorption for practical REE recovery from geothermal fluids 144 2 Chapter 6 discusses the encapsulation of engineered E coli in polyethylene glycol diacrylate PEGDA microparticles for use in continuous flow REE extraction systems The synthesis procedure is described along with the fundamental physical and chemical characteristics of the microparticles When packed in a fixed bed column the hybrid adsorbent is an effective REE extractant at a variety of feedstock concentrations and pH conditions and with several influent flow rates The optimal conditions for REE recovery are assessed as is the possibility of the separation of individual REEs in a column extraction system using this adsorbent The work described in this chapter represents a significant step towards industrial scale application of biosorption for REE extraction and purification The importance of REEs to the modern economy in combination with their potential supply vulnerability makes the design of novel REE extraction technologies from novel feedstocks a matter of immediate practical concern Ample progress has been made in the initial development and characterization of many promising REE-specific adsorbents for solid phase extraction systems However these materials have generally not progressed to a technology readiness level that approaches industrial scale application I believe that advancements in polymer encapsulation particularly in the area of microfluidics may permit a variety of adsorbents to make the jump from proof of concept to actual application Once these technologies have been adequately developed a wide range of applications become possible from remediation to green mining to analytical chromatography just to name a few References Mullen EK McCallum IS 2014 Origin of Basalts in a Hot Subduction Setting: Petrological and Geochemical Insights from Mt Baker Northern Cascade Arc J Petrol 55 2 :241-281 145 Mullen EK Weis D 2013 Sr-Nd-Nf-Pb isotope and trace element evidence for the origin of alkali basalts in the Garabaldi Belt northern Cascade arc Geochem Geophy Geosy 14:31263155 Mullen EK Weis D 2015 Evidence for trench-parallel mantle flow in the northern Cascade Arc from basalt geochemistry Earth Planet Sc Lett 414:100-107 Mullen EK Weis D Marsh NB Martindale M 2017 Primitive arc magma diversity: New geochemical insights in the Cascade Arc Chem Geol 448:43-70 146
    • Lilien, David - Ph.D. Dissertation
      Understanding Antarctic ice-stream flow using ice-flow models and geophysical observations
      Appendix 1 2019, Lilien,David,David Lilien
    • Zheng, Ling - Ph.D. Dissertation
      Testing the theory of radiation belt electron loss by hiss and electromagnetic ion cyclotron waves 2019, Zheng,Ling,Ling Zheng c Copyright 2019 Ling Zheng Testing the theory of radiation belt electron loss by hiss and electromagnetic ion cyclotron waves Ling Zheng A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy University of Washington 2019 Reading Committee: Michael P McCarthy Chair Robert Holzworth Robert Winglee Program Authorized to Offer Degree: Earth and Space Sciences University of Washington Abstract Testing the theory of radiation belt electron loss by hiss and electromagnetic ion cyclotron waves Ling Zheng Chair of the Supervisory Committee: Professor Michael P McCarthy Department of Earth and Space Sciences Hiss chorus and electromagnetic ion cyclotron waves EMIC wave are three major wave modes that are widely investigated and included in the radiation belt electron models to explain electron precipitation The quasi-linear theories of electron loss through pitch angle diffusion by hiss and EMIC waves were proposed in 1970s Since then the testing of the theories is still going on though some progresses had been made Comparison of theoretical predictions to electrons distribution at loss cone is one effective way to evaluate the theories The main obstruction of loss cone testing was from the lack of measurements of the electron loss cone distribution with enough pitch angle and energy resolution and simultaneous wave activities at the heart of radiation belt This thesis is devoted to testing the hiss and EMIC waves diffusion theories from the perspective of the electron loss cone distribution by utilizing the previously unnoticed overlap of UARS and CRRES missions in 1991 The conclusions are as following: 1 Two cases showing the consistency between quasi-linear theory of hiss diffusion and observed loss cone distribution are found 2 In 25 out of 38 cases hiss wave power is far insufficient in precipitating the large amount of electrons observed Loss mechanisms other than hiss chorus and EMIC waves are needed to account for the discrepancy 3 Three EMIC wave events were investigated In the first case the isotropic distribution sign of strong diffusion is caused by process other than EMIC wave diffusion demonstrating that simultaneous presence of EMIC wave and electron precipitation does not guarantee any connection between the two Large discrepancy between quasi-linear theory of EMIC wave diffusion and observed electron loss cone distribution is found from the second case The strong depletion of electrons by EMIC waves predicted by the current theory was not found In the third case the resonant energy goes beyond the instrument limit of electron detectors thus no conclusion is drawn TABLE OF CONTENTS Page List of Figures iii Chapter 1: Introduction 1 1 Radiation belt and relativistic electrons 1 2 Previous research on hiss wave as a loss mechanism 1 3 Previous research on electromagnetic ion cyclotron wave as 1 4 Motivation science question and outline of this thesis a loss mechanism 1 1 3 6 10 Chapter 2: Data set 2 1 UARS satellite 2 2 CRRES satellite 2 3 UARS and CRRES magnetic conjunction 13 13 15 17 Chapter 3: Theoretic framework of quasi-linear diffusion 3 1 Wave-particle interaction as a diffusion process 3 2 Formula of pitch angle diffusion coefficient 21 21 25 Chapter 4: Comparison of hiss wave diffusion theory to electron loss cone distribution 4 1 Electron precipitation events consistent with hiss diffusion 4 2 Electron precipitation events inconsistent with hiss diffusion 4 3 Conclusion for hiss diffusion theory testing 31 33 36 41 Chapter 5: 5 1 5 2 5 3 5 4 Comparison of EMIC wave diffusion to precipitating and trons both at low Earth orbit and magnetic equator Case studies Concluding remarks from case studies Comparison to findings from others Conclusions of testing EMIC wave diffusion theory i trapped elec 42 42 56 56 60 Chapter 6: Conclusion and future work 61 Bibliography 62 Appendix A: A 1 Derivation of N for hiss diffusion coefficient A 2 Derivation of N Y for EMIC diffusion coefficient A 3 Source term estimation using observations inside loss cone A 4 Some notes on diffusion coefficients code implementation and 68 68 69 69 70 ii testing LIST OF FIGURES Figure Number 1 1 Page schematic view of Earth s magnetosphere formed by interaction of Earth s magnetic field and solar wind The ball at the center is Earth with arrow curves the magnetic field lines The green shaded area are inner and outer radiation belts Credit from spaceweathercenter org 2 2 1 Orientation of UARS HEPS electron telescopes 14 2 2 schematic view of a magnetic conjunction between UARS at 585km and CRRES at about 33 000km magnetic equator 18 2 3 Temporal distribution of 38 conjunction cases in Oct 1991 19 2 4 Spatial distribution of 38 conjunction cases in Oct 1991 20 4 1 Survey plot of wave power spectrum from CRRES PWE Plasma Wave Experiment on Oct 8th 1991 Orbit 1059 The plot is provided by Physics Department University of Iowa Th broad band hiss is typically from tens of Hz to 2 kHz The red line is electron cyclotron frequency 32 Hiss wave power spectra at 21:15:00-21:15:54 Oct 8 1991 when UARS and CRRES are at magnetic conjunction 34 Comparison of hiss diffusion curve to measured electron flux at loss cone he vertical magenta dash line plots edge of bounce loss cone a 100 keV electron flux vs hiss diffusion the red stars are measured electron flux from three electron telescopes inside loss cone the horizontal color bars are field of view of each telescope the thick blue curve is electron flux distribution predicted by hiss diffusion the black stars are integral of blue curve over field of view of each telescope the dash blue curve is obtained by reducing source term by a factor of 2 b similar to panel a for 500 keV electrons c similar to panel a for 1MeV electrons 35 4 2 4 3 iii 4 4 4 5 4 6 4 7 Comparison of modified hiss diffusion curve to measured electron flux at loss cone The vertical magenta dash line plots edge of bounce loss cone whereas the horizontal green dash line is detector sensitivity a 100 keV electron flux vs hiss diffusion the red stars are measured electron flux from three electron telescopes inside loss cone the horizontal color bars are field of view of each telescope the thick blue curve is electron flux distribution predicted by hiss diffusion with diffusion coefficient increased by a factor of 10 000 the black stars are integral of blue curve over field of view of each telescope b similar to panel a for 500 keV electrons with diffusion coefficient boosted up by a factor of 2000 c similar to panel a for 1MeV electrons with diffusion coefficient boosted up by a factor of 2000 37 Comparison of hiss diffusion curve to measured electron flux at loss cone The vertical magenta dash line plots edge of bounce loss cone a 100 keV electron flux vs hiss diffusion the red stars are measured electron flux from three electron telescopes inside loss cone the horizontal color bars are field of view of each telescope the thick blue curve is electron flux distribution predicted by hiss diffusion b similar to panel a for 500 keV electrons c similar to panel a for 1MeV electrons 38 Electron flux pitch angle profile with sharp edge of loss cone and flat distribution inside loss cone The vertical magenta dash line plots edge of bounce loss cone The horizontal green dash line is detector sensitivity 39 Comparison of hiss diffusion curve to measured electron flux at loss cone The the vertical magenta dash line plots edge of bounce loss cone The horizontal green dash line is detector sensitivity a 100 keV electron flux vs hiss diffusion the red stars are measured electron flux from three electron telescopes inside loss cone the horizontal color bars are field of view of each telescope the thick blue curve is electron flux distribution predicted by hiss diffusion the dash blue line is hiss curve with diffusion coefficient decreased by a factor of 10 b similar to panel a for 500 keV electrons c similar to panel a for 1MeV electrons 40 iv 5 1 Three EMIC wave events from CRRES with case 1 on October 1st 1991 and case 2 and 3 at two consecutive orbits on Oct 4th 1991 The right column panels plot wave power spectrum on frequency domain versus universal time L shell and magnetic local time The black double lines on UT axis marks the time slots when UARS is at footprint of CRRES aka conjunction The red double line marks the reference time Two white dash lines from top to bottom within each panel are local hydrogen and helium gyro-frequencies The left column panels plot wave power spectrum at the time slots marked the double black and red lines in the right column The black curve is wave power spectrum at the conjunction time while red curve is wave power at reference time The blue vertical dash lines show cutoff frequencies bounding the peaks 43 5 2 Trapped and precipitating electron flux observed by UARS at three UARS CRRES conjunctions The left middle and right columns correspond to case 1 2 and 3 respectively From top to bottom a 0 5 MeV electron differential flux with red for electron telescope right outside loss cone blue green and cyan for three telescopes well inside loss cone the horizontal magenta dash lines show detector sensitivity those below sensitivity line are due to over-correction of correction algorithm b and c 1MeV and 2MeV electron differential flux with color code the same as 0 5 MeV electrons The events pointed by the two black arrows are re-plotted in Figure 5 3 and 5 4 d difference of L shell and magnetic local time of UARS magenta and CRRES blue e L profile of 1MeV electron flux at time window of one minute right before and after conjunction red for electron telescope right outside loss cone blue green and cyan for three telescopes well inside loss cone The vertical two black dash lines bound the conjunction window 45 Comparison of theoretical loss cone distribution to measurements from UARS around 21:55 UT Oct 4 1991 The diffusion coefficients are computed assuming parallel propagating EMIC waves Row a and b are for 2MeV and 1MeV electrons respectively The red thick curves are prediction given by EMIC wave diffusion theory Red dash lines are EMIC wave curve with diffusion coefficients decreased by a factor of 16 for 1MeV electrons and 13 for 2 MeV electrons The black dots are electron differential flux measured by four telescopes on board UARS with one right outside loss cone three inside loss cone The horizontal error bars are field of view of telescopes The pink vertical dash lines mark the edge of bounce loss cone The detector sensitivity is marked by horizontal blue dash lines The pitch angles of UARS telescopes at low Earth orbit are converted into equatorial values in degree 49 Similar to Figure 5 3 assuming oblique propagating EMIC waves 50 5 3 5 4 v 5 5 Precipitating protons from UARS at 550km The upper and bottom panels plot proton flux at pitch angle 22 and 14 respectively Both are well inside bounce loss cone The UARS CRRES conjunction is at 20:54:30 to 20:55:20 well inside the proton precipitation window The lower energy cutoff is 74 keV Notice that the energy is in eV and flux is in cm 2 s 1 sr 1 eV 1 5 6 Two choices of cutoff frequencies The black curve is wave power spectrum at case 2 conjunction around 21:55 UT Oct 4 1991 The green curve is the wave power spectrum used as background and measured around 20:21 UT Oct 4 1991 The magenta vertical dash line marks lower end cutoff frequency of 0 45 Hz blue and red lines mark two choices of cutoff frequencies of 1 665 Hz and 2 1 Hz at higher end 5 7 Comparison of trapped electron from MEA on board CRRES to EMIC wave diffusion a differential flux of 1090 keV electrons at pitch angles from 5 to 90 b thick blue curve is 1090 keV electron differential flux at pitch angle of 5 measured by MEA The thin solid red line is prediction from EMIC wave diffusion with bounce-drift averaged model The dash magenta line is prediction from EMIC wave diffusion with bounce-averaged model c EMIC wave spectrogram versus universal time L shell and magnetic local time The two white meandering dash lines from top to bottom mark local hydrogen and helium gyro-frequencies 5 8 hydrogen band EMIC wave amplitude square versus electron flux at pitch angle of 5 measured by MEA on board CRRES The black and red dots are for 509 keV and 1090 keV respectively 5 9 snapshot of Figure 3 of Miyoshi et al 2008 38 a Emission profile of Hb and be count rates of energetic ions and electrons observed by POES-17 on September 5 2005 with UT MLAT the McIlwain L-value and magnetic local time MLT of the satellite footprint during the period when the satellite was crossing over Athabasca The vertical blue line indicates the time when the satellite footprint crossed the stable isolated proton aurora MLAT and L-value are calculated by the IGRF model 5 10 Snapshot of Figure 8 of Zhang XJ et al 2016 56 a Observed electron pitch angle distributions from Van Allen Probes at L 5 77 b The evolution of electron distribution after interaction with EMIC and hiss c only EMIC and d only hiss waves for 40 min Initial distributions are shown in dotted lines observations or model results at t 40 min are shown in solid lines vi 51 52 53 55 58 59 ACKNOWLEDGMENTS I appreciate UW Department of Earth and Space Sciences and my advisor Prof Michael P McCarthy for supporting me with RA TA and fellowships through my entire life pursuing this doctorate degree of space physics I also thank Michael for great support and close advising for so many years and Profs Robert Holzworth Robert Winglee and Erika Harnett for being in my committee to make sure that I am on the right track I thank Profs Michael McCarthy Robert Holzworth and Robert Winglee sitting in my reading committee to prove this thesis Thanks Prof Robert Holzworth for admitting me to this department I thank Rudy Frahm Pesnell William Gaines Edward from UARS PEM group for help with electron data and IDL code for loading the data Rudy Frahm provided details about correction algorithm which is useful for interpretation of electron spectrum I would also like to thank Larry Granroth from University of Iowa for offering the CRRES plasma wave CDF files generated by French Aerospace Lab ONERA The wave plots from website hosted by Physics department of University of Iowa were used along with ONERA cdf files Since CRRES plasma wave data was not publicly archived this work is not possible without the help of people from University of Iowa Dr Nigel Meredith from British Antartic Survey kindly provided me raw EMIC wave data and good feedback on data analysis The EMIC wave investigation would not be possible without Dr Meredith s help I thank emeritus Professor Brian Fraser from University of Newcastle for providing the detailed list and plots of EMIC wave events The communication with my friend Yong Fu Wang from School of Earth and Space Science Peking University is helpful to better understand the quasi-linear theory of plasma wave diffusion and navigating through MATLAB code The discussion with office mates Hao Zheng Paul Sturmer and Todd Anderson were also beneficial vii DEDICATION to the readers viii 1 Chapter 1 INTRODUCTION This chapter will start with the basics of radiation belt the source and loss processes that control the flux of relativistic electrons inside the outer part of radiation belt Hiss and electromagnetic ion cyclotron waves EMIC wave were proposed in 1970s to be two important electromagnetic waves in radiation belt that precipitate radiation belt electrons into Earth s atmosphere via a process named pitch angle diffusion A review on quasi-linear theories of hiss and EMIC waves diffusion as well as a summary of observations will be presented in details This chapter will conclude with the motivation science questions and the outline of following chapters 1 1 Radiation belt and relativistic electrons The Earth s magnetic field is produced by the liquid at core and its effects extend to outer space It interacts with the continuous flow of magnetized charged particles from the Sun named solar wind forming Earth s magnetosphere which is the outermost boundary of Earth s territory The magnetosphere is compressed in the day side and stretched at nightside See Figure1 1 Within magnetosphere the radiation belt is located about 1 1 to 8 Earth radii from center of the Earth It is also called Van Allen belt after Dr James Van Allen who discovered it in 1958 53 Radiation belt is further divided into two parts by the slot region of very few energetic electrons at 2 to 3 Earth radii The inner belt is relatively stable Outer radiation belt is very dynamic and is famous for relativistic electrons that are a threat to the nearby satellites The energy of relativistic electrons concerned here is above 100 000 electron volts 100 keV The flux of outer radiation belt electrons is highly dynamic and can fluctuate several 2 Figure 1 1: schematic view of Earth s magnetosphere formed by interaction of Earth s magnetic field and solar wind The ball at the center is Earth with arrow curves the magnetic field lines The green shaded area are inner and outer radiation belts Credit from spaceweathercenter org orders of magnitude in time scales of days hours and even shorter The local electron flux is determined by the relative strength of source and loss Solar wind and Earth s atmosphere provide seed electrons which can be further energized by a variety of processes As electrons are transported radially toward the Earth through Earth s approximately dipole magnetic field and energized through adiabatic processes The energetic electrons can also be built up locally by very low frequency and ultra low frequency waves When the solar wind is very strong and push the dayside boundary of magnetosphere into geosynchronous orbit radiation belt electrons are lost to interplanetary space by crossing the magnetopause Radiation 3 belt electrons are also being precipitated into Earth s north and south polar atmosphere by ELF VLF electromagnetic or electrostatic waves via a process called pitch angle diffusion The electron precipitation can be detected by satellites at low Earth orbit of several hundred kilometers above the Earth s surface The bremsstrahlung X-rays associated with precipitating electrons can be detected by stratospheric balloon experiments ELF VLF from 20 Hz to 30 kHz electromagnetic waves of both natural and anthropogenic origins were proposed as mechanisms that dump radiation belt electrons into Earth s atmosphere via pitch angle diffusion not long after the discovery of radiation belt Pitch angle is the angle between electron velocity and local Earth s magnetic field At the magnetic equator electrons with near zero or 180 degree pitch angle can move along the magnetic field line freely traveling from radiation belt toward the thick atmosphere and get lost Electrons with equatorial pitch angle close to 90 degree are trapped in Earth s magnetic field The small range of pitch angle near either zero or 180 degree forms a region called loss cone The electrons within loss cone will get lost quickly whereas those outside loss cone will stay trapped The Lorentz force of the waves either accelerate or decelerate the electrons increase or decrease the pitch angle of electrons randomly Depending on the distribution function of electrons and properties of the waves the pitch angle and energy diffusion will go in certain direction If the trapped electron flux is larger than flux of electrons inside loss cone which is always true in outer radiation belt the waves will move electrons from large pitch angle to small pitch angle to smooth out the gradient Some very low frequency electron cyclotron waves 24 and ultra low frequency ion cyclotron waves e g EMIC waves are able to reduce the pitch angle of trapped electrons and thus precipitate them into Earth s atmosphere 1 2 1 2 1 Previous research on hiss wave as a loss mechanism Basics of hiss wave and the theory of hiss as a loss mechanism Hiss wave is broad band incoherent electron cyclotron waves that is observed mostly inside plasmasphere which largely overlaps with radiation belt in space and is filled with dense cold 4 electrons of several electron volt during both geomagnetically active and quiet time The frequency of hiss can go up to 2 kHz and reach to as low as 20 Hz with most wave power at several hundred Hz 32 Hiss was first proposed by Lyons et al in 1970s 32 31 as the main mechanism inside plasmasphere that precipitate electrons via pure pitch angle diffusion forming the slot region that separates the inner and outer radiation belt More details about the theoretic framework of hiss diffusion is in Chapter 3 1 2 2 Evidence hunting The hiss diffusion theory has been backed by general agreement with profile of trapped electrons Relying on simple models of hiss wave and cold plasma density Lyons 32 computed pitch angle profile and decay rate of trapped electrons Agreement with observations from Ogo-5 and OV3-3 satellites was found in slot regions The hiss was assumed by Lyons to be uniform over entire plasmasphere Utilizing electrons and hiss waves measured by CRRES spacecraft Meredith in 2006 reported the decay time of trapped electrons at pitch angle of 90 to be consistent with pitch angle scattering by hiss traveling at small and moderate wave normal angles the angle between local Earth s magnetic field and wave vector 35 Similarly to Lyons work 32 this comparison is on average and over long term days weeks or months though a more realistic global model of hiss was adopted More recently Thorne in 2013 48 reported that the slow decay of an unusual new ring of relativistic electrons in outer radiation belt can be well explained by global plasmaspheric hiss scattering The comparison of hiss wave to trapped electrons has limitations The hiss diffusion is weak most of the time due to its small amplitude of several picotesla The trapped electron flux is several orders of magnitude of higher than precipitating flux A small bite on trapped electrons by hiss could be not noticeable Back to early 1990s Imhof 18 found that time variation of electron flux inside loss cone resembles the variation of hiss wave amplitude while electron flux just outside loss cone doesn t correlate with hiss wave variation at all This is similar to a recent discovery by Kasahara in early 2018 about interaction of electrons with chorus wave which like hiss is also whistler mode electron cyclotron wave and considered as 5 important mechanism for electron precipitation 23 Kasahara concluded that chorus waves strongly modulate electron flux inside loss cone while showing no clear effect on electrons outside loss cone In order to see the effects of hiss scattering a long time observation is needed At longer time frame the hiss scattering is easier to be obscured by other processes An incomplete list of processes of concerned includes radial diffusion local acceleration electron loss through magnetosphere boundary crossing ring current effect and wave modes other than hiss The electrons inside the loss cone are very sensitive to wave scattering which makes one-to-one correspondence testing possible Testing the hiss theory with electrons inside loss cone also eliminates possibilities of loss processes other than precipitation Some attempt had been made on this since not long after Lyons 32 suggested hiss as loss mechanism responsible for radiation belt electron precipitation In early 1980s Imhof et al 16 investigated 17 conjunctions between P78-1 at 600km and ISEE-1 at 2 to 3 earth radii in the magnetosphere Hiss wave measurements from ISEE-1 along with cold plasma density inferred from upper hybrid frequency were compared to simultaneous measurements of quasi-trapped 65 115 pitch angle at low Earth orbit electron spectra 68 -1120 keV at P78-1 In general the wave frequency electron energy and cold plasma density were found mutually consistent with precipitation caused by first order cyclotron resonance with parallel propagating hiss at the equator As mentioned earlier Imhof et al in early 1990s 18 reported good correlation between loss cone electron flux and hiss wave observed by CRRES spacecraft at the vicinity of magnetic equator The best correlation between waves and electrons is seen for wave near 256Hz and electrons near 50keV Li et al in 2014 26 compared the ratio of integral electron flux between two electron telescopes approximately 0 and 90 pitch angles on board POES satellite to that predicted by hiss diffusion under steady state assumption when POES is at magnetic footprint of Van Allen probes The POES measurement agrees with estimation from hiss diffusion within a factor of 10 Breneman et al in 2015 8 investigated two cases when hiss waves were observed by Van Allen probes and simultaneously measurements of X-rays associated with precipitating electrons were seen by high altitude balloon 6 arrays BARREL at the magnetic footprint The temporal structure of X-rays resembles hiss variation Hardman et al in 2015 15 used ground-based VLF receiver network to investigate electron precipitation during a magnetic storm The weak electron precipitation that s missed by POES but captured by their network is claimed to be consistent with hiss diffusion In summary the long term averaged trapped electrons profile is found to be consistent with hiss diffusion theory Some attempt of looking into loss cone had been made 1 3 1 3 1 Previous research on electromagnetic ion cyclotron wave as a loss mechanism Basics of EMIC wave and the theory of EMIC wave as a loss mechanism Electromagnetic Ion Cyclotron Wave EMIC is at Pc1 and 2 bands of ultra lower frequency with a typical frequency range of 0 1-5Hz EMIC wave consists of three distinct bands: hydrogen band with frequency lower than proton gyro-frequency and above He gyro-frequency helium band with frequency in between He gyro-frequency and O gyro-frequency and oxygen band which is below O gyro-frequency but higher than Alfvenic ULF waves at Pc3- Pc5 EMIC waves arise when ion temperature perpendicular to local magnetic field is larger than parallel temperature The temperature anisotropy is caused by substorm injections ring current intensification during magnetic storms and solar wind impulse in dayside magnetosphere Thorne and Kennel in 1971 47 first introduced EMIC waves as loss mechanism for precipitating relativistic electrons into atmosphere One year later Lyons and Thorne further addressed the EMIC wave theory with more quantitative framework 30 The features of electron precipitation associated with EMIC waves are summarized as following First EMIC waves are generated by temperature anisotropy of tens of keV ring current ions with the perpendicular temperature much higher than parallel temperature Thus it is often to see that relativistic electron precipitation is accompanied by almost simultaneous ion precipitation Second due to the large amplitude of EMIC waves with typical value of 7 1-10nT the pitch angle diffusion could reach strong diffusion limit such that an isotropic distribution of electrons across loss cone is expected Also due to strong diffusion the trapped electron flux will decrease in a fast pace at the absence of a strong source to replenish the electrons Summers calculations in 2003 suggested that during magnetic storms EMIC waves can remove the electrons completely in several hours even though waves reside at only one percent of magnetic local time 46 Third the minimum resonant energy of electrons is 1MeV suggested by Thorne and Kennel in 1971 due to the ultra low frequency nature of EMIC waves Electrons with energy lower than 1 MeV was believed to not interact with EMIC waves Later theories show the minimum resonant energy can go down to several hundred keV 36 and even at 100 keV under nonlinear resonance 39 Fourth EMIC waves can only scatter electrons at small to medium pitch angles leaving electrons around 90 intact due to zero overlap between cyclotron resonance and Landau resonance for relativistic electrons Last Thorne and Kennel predicted that relativistic electrons precipitation REPs will be mostly in the dusk sector from evening to midnight in magnetic local time and confined in L 3 5-5 at the vicinity of plasmapause This is straightforward inference from ring current ion drifting path and high cold plasma density of plasmasphere The cloud of ring current ions injected from midnight plasmasheet drift westward into dusk side and then go to morning side The high cold plasma density inside plasmasphere greatly lowers the resonant energy from tens of MeV to several hundred keV 1 3 2 Evidence hunting Strong electron precipitation accompanied by simultaneous proton precipitation is believed to be an important though not unique signature of EMIC wave scattering Thorne and Andreoli in 1980 49 investigated 313 relativistic electron precipitation REP events from an analysis of over 14 months of data from S3-3 satellite at low altitude The majority 306 of the events occur in a narrow latitudinal zone at nightside and were embedded within a broader region of strong ion precipitation Thorne and Andreoli attributed 302 of the events to be caused by electrostatic ion cyclotron waves EIC waves Only four events were believed to 8 be caused by EMIC waves An example of the four events is given in Figure 3 of Thorne and Andreoli s paper It shows isotropic electron distribution which is the sign of strong diffusion accompanied by simultaneous broader strong ion precipitation Simultaneous EMIC wave observation was not available As will be mentioned later in this thesis the event reported by Thorne and Andreoli is at a boundary that looks like what other authors identified as trapping boundary even though it is at relatively low L of 4 35 Isotropic electron distribution at trapping boundary is often seen regardless of the presence of EMIC waves Imhof et al in 1986 20 selected 40 relativistic electron events that occurred at plasmapause crossing The event selection criteria excludes events near trapping boundary The occurrence rate of these relativistic electron events at plasmapause according to Imhof is one percent Nine out of the 40 electron enhancement events come with ion precipitation at the same time and location The electron enhancement preferentially occurs at relativistic energies of from several hundred keV to greater than 4 MeV which is consistent with the energy selective nature of EMIC wave scattering However Imhof pointed out that the electron flux pitch angle distribution is far from isotropic The loss cone electron flux is at least one order of magnitude lower than electrons at 90 Measurement of bremstrahlung X-rays associated with relativistic electron precipitation by balloon experiments and electron spectra measurements by low Earth orbit both show preference of relativistic electron precipitation over dusk side mostly from evening to midnight quadrant of magnetic local time 37 9 This agrees with what the theory suggested in 1970s by Thorne Kennel and Lyons 47 30 However satellite observations imply that the peak location of EMIC waves is in dayside mostly from late morning to early afternoon 4 57 Smith et al in 2016 suggest that current sheet scattering rather than EMIC wave should be studied further as a potentially important loss mechanism to account for the large amount of relativistic electron precipitation at nightside 45 Simultaneous observations of EMIC wave activities in magnetosphere and electron precipitation provide good insight into the validity of the theory Some authors claimed to find the evidence of electron precipitation caused by EMIC waves whereas others reported 9 absence of electron precipitation during the presence of strong EMIC waves Miyoshi et al in 2008 38 reported a greater than 800 keV electron precipitation accompanied by simultaneous 30-80 keV proton precipitation observed by POES At the same time EMIC wave was registered by ground-based magnetometer at nearby L shell Rodger et al in 2015 42 reported greater than 140-230 keV electron precipitation event observed by POES during a strong helium band EMIC wave event observed by Van Allen probes Contrarily also from POES observation Usanova et al in 2014 50 reported absence of precipitation during a strong EMIC event registered by both Van Allen probes and ground-based magnetometers Engebretson et al in 2015 12 investigated a strong up to 25nT p-p global scale of EMIC wave event lasting for 8 hours in UT and covering over 12 h in MLT being observed by both Van Allen probes and GOES satellites and multiple ground magnetometers in Antarctica near dawn Finland at local noon Russia in the afternoon and Canada from duskside to midnight During such strong and spatially large EMIC wave event ten passes of POES satellites near footprint of Van Allen probes detected strong 30-80 keV subauroral proton precipitation indicating strong interaction of protons with EMIC waves At the same time no enhancement of electron precipitation was observed Both Usanova and Engebretson concluded no evidence of significant electron loss by EMIC waves was found which contradicts with what Summers in 2003 predicts from quasi-linear theory Case studies on associating Balloon X-rays with EMIC wave events in magnetosphere were carried out as well Li et al in 2014 27 found the estimated amount of X-rays from electron pitch angle scattering by EMIC waves observed by GOES is only a factor of 2 7 higher than X-ray flux observed by BARREL balloon arrays The observations from Van Allen probes and ground stations along with some assumptions about electron spectrum distribution and electron to X-ray models were used to overcome the one L shell displacement between GOES and BARREL Zhang et al 55 studied another conjunction case between Van Allen probes and BARREL Since EMIC wave event is highly localized both in space and time Zhang concluded that the X-ray event is likely associated with EMIC wave observed by Van Allen probes 10 The selective pitch angle scattering by EMIC waves at low and medium pitch angle produces so called pancake electron distribution with electrons around 90 degree intact and electrons at lower pitch depleted In both Usanova 50 and Engebretson 12 s work it is found that pitch angle distribution of ultra-relativistic electrons at 2 3-5 6 MeV is more pancake-like when EMIC waves are present The quasi-linear diffusion coefficients of EMIC wave diffusion show that higher energy electrons are confined at narrower pitch angle range centered at 90 degree This is consistent with observation as well Zhang XJ et al in 2016 56 simulated time evolution of electron pitch angle distribution during a long-lasting EMIC wave event The simulation agrees with the measurements of electrons at energy greater than 1 8 MeV and medium pitch angles but fails to reproduce the electron profile at energy lower than 1 8 MeV and small pitch angles In summary observations show the pancake distribution of electrons at ultra-relativistic energy can be explained by the EMIC wave diffusion There exists seemingly conflicting evidences of electron precipitation associated with EMIC waves 38 42 50 12 Some authors reported no evidence supporting the strong depletion of electrons by EMIC wave as predicted by the current quasi-linear theory 1 4 Motivation science question and outline of this thesis The identification of loss mechanisms will help construct a good physics-based radiation belt electron model Hiss and EMIC waves are two of the three major wave modes that had been suggested to precipitate electrons from radiation belt into Earth s atmosphere The testing and validation of the theories have not been finished despite of some progresses made since 1970s More work needs to be done to verify which part of the theories agree with observations and which part need to be modified to fit the observations This thesis is to test the theories by comparing the predicted and measured electron distribution at loss cone quantitatively evaluate the discrepancy between current quasi-linear theories of hiss and EMIC wave diffusion and observations peek into the legitimate regime of the theories and give some guidance to the modifications of the theories 11 Though there exist increasing evidences consistent with electron loss by hiss wave diffusion a quantitative comparison of loss cone electron distribution to simultaneous hiss wave events on one-to-one correspondence is lacking The few one-to-one correspondence study carried out using POES electron measurements and balloon X-rays mentioned in previous sections is limited by the instrument capacity POES satellites only have two electron telescopes pointing at 0 and 90 which means it s impossible to construct electron pitch angle distribution inside loss cone which is vital for inferring diffusion by particular wave mode The energy resolution of electron detectors onboard POES is coarse 30keV 300keV 1M eV too POES also has electron-proton cross contamination problem Hardman et al in 2015 15 reported weak precipitation seen by ground-based VLF receivers but invisible to POES probably due to its small geometric factor Though both Van Allen probes and BARREL provide high quality wave and X-ray data the electron distribution inside loss cone can not be inferred from X-rays The one-to-one correspondence analysis on loss cone distribution will be carried out in this thesis About 38 cases of hiss diffusion will be analyzed This relatively large number of cases will give more insight into the legitimate regime of hiss theory and the role of hiss waves in depleting radiation belt electrons than from the one or two cases published by other authors As mentioned in previous section the strong diffusion predicted by the theory 46 suggests EMIC wave could be very important loss mechanism during magnetic storm On the other hand seemingly contradictory evidences of electron precipitation associated with EMIC waves have been reported from different authors The strong capacity of electron removal predicted by theory is also being challenged by the lack of observed significant loss of electrons The observed pancake distribution of electrons at ultra-relativistic energy gives some credit to the theory This thesis will help to have a clearer picture of EMIC wave diffusion and the legitimacy of current theory from the perspective of electron loss cone distribution One of the biggest obstacles of testing either hiss or EMIC theory is lack of electron measurements at loss cone with enough energy and pitch angle resolution and simultaneous 12 wave measurements at high altitude magnetic equator New space missions are needed if we are not able to unearth qualified data set from old space missions Fortunately during the last ten days of the prematurely dead CRRES mission in early October 1991 there are about 50 conjunctions between CRRES sitting at the vicinity of magnetic equator and UARS orbiting low Earth orbit at the magnetic footprint of CRRES The study carried out in this thesis relies on this piece of data set The details about data instrumentation and overview of UARS and CRRES missions will be presented in Chapter 2 Chapter 3 is devoted to the theoretic framework of quasi-linear diffusion theory and the customary formulas derived from quasi-linear theory for testing hiss and EMIC wave diffusion The measured wave power is fed to a formula for computing diffusion coefficients Full cold plasma dispersion relationship of EMIC waves is incorporated in the EMIC wave diffusion coefficient formula Comparison of hiss diffusion to electron loss cone distribution will be presented in Chapter 4 A similar comparison for EMIC waves is in Chapter 5 Chapter 6 summarizes the entire thesis Appendix lists more details about diffusion coefficient formulas code implementation and testing and misc 13 Chapter 2 DATA SET 2 1 2 1 1 UARS satellite Mission overview The UARS Upper Atmosphere Research Satellite was lauched in September 12th 1991 by Space Shuttle Discovery into a nearly circular orbit at nominal altitude of 585km with inclination of 57 The orbit processed 5 per day relative to the Sun Like most low Earth orbit satellites the orbital period is about one hour and half It was the first observatory of NASA s Mission to Planet Earth carrying out first systematic and comprehensive study of Earth s stratosphere and mesosphere This mission was designed to survive for three years Six out of ten instruments functioned for more than 14 years The official decommission was on December 14th 2005 More details about the mission can be found in Carl Reber s introductory GRL paper 40 and NASA s UARS main science page 2 1 2 Electron and proton instrument Though the mission focused on atmospheric science the Particle Environment Monitor PEM on board UARS provided quality measurements of radiation belt electrons and protons with high energy and pitch angle resolutions At most of the time when UARS is at the footprint of outer radiation belt there are three electron telescopes well inside bounce loss cone and one right outside loss cone For testing wave-particle theory with loss cone distribution UARS PEM provides so far the best data set compared to existing or past missions such as POES SAMPEX and some cubsat missions PEM consists of four instruments: the atmospheric X-ray imaging AXIS vector mag- 14 netometer VMAG medium particle spectrometer MEPS and high energy particle spectrometer HEPS HEPS electron and proton data is used for the research on radiation belt electron in this thesis The satellite is three-axis stabilized There are four HEPS telescopes mounted at zenith each with a 15 half cone angle facing upward hemisphere There are two telescopes at nadir looking at the Earth see 2 1 HEPS electron detectors measure electrons from 30 keV to 5 MeV at 32 energy bins The energy bin size is 15 keV at low energy and 100 keV at high energy The relatively large geometric factor of 0 54 cm2 sr makes it possible to see low precipitating flux For more details about PEM and HEPS see J D Winningham s instrument paper 54 Level 2 HEPS electron data in binary format with time resolution of Figure 2 1: Orientation of UARS HEPS electron telescopes 4 086 s used in this thesis can be located at website hosted by NASA Goddard flight center https: mirador gsfc nasa gov with searching keyword of UARS HEPSA The 15 IDL code modified from the code provided by Dr W Dean Pesnell for reading the binary data is attached in the appendix The PEM data can also be plotted and downloaded from Southwest Research Institute Virtual Sun Earth Observatory http: vseo space swri edu with customary data format The first set of HEPS data became available on Oct 1st 1991 2 2 2 2 1 CRRES satellite Mission overview The CRRES Combined Release and Radiation Effects Satellite was launched on July 25 1990 from Cape Canaveral Air Force Station by Atlas I The mission was planned for three years but communications lost on early morning of October 12 1991 The initial orbit was 350 by 33 584km with an inclination of 18 1 and orbital period of 613 4 minutes The orbit is similar to the double Van Allen probes launched in August 2012 The spacecraft traversed the heart of outer equatorial radiation belt slowly at its apogee For more details about CRRES mission please read M H Johnson s paper 22 and other online resources 2 2 2 Electron instrument The CRRES MEA Magnetic Electron Spectrometer utilizes the principal of momentum analysis in a solenoid magnetic field to measure electrons spectrum The electrons are bent by uniform magnetic field by 180 before hitting the detector array It measures electrons at center energy from 153 to 1582 keV with 18 energy bins The angular resolution at different energy bins varies from 2 82 to 16 48 In the data set used in this thesis there are 19 pitch angle bins with equal size of 5 from 0 to 90 The time resolution is 25 s The MEA electron spectrum is used to for comparison to the EMIC wave scattering at equatorial radiation belt The CDF format MEA electron data is accessible at ftp: spdf gsfc nasa gov pub data crres particle_mea mea_h0_cdaweb 1991 For more information about MEA read the instrument paper by Vampola 52 16 2 2 3 Wave instrument The CRRES PWE Plasma Wave Experiments provides measurements of electric fields from 5 6 Hz to 400kHz and magnetic fields from 5 6 Hz to 10kHz with a dynamic range of at least 100 dB The module consists of three sensors: 1 a 100m tip-to-tip electric dipole antenna 2 a search coil magnetometer 3 a 94m sphere-to-sphere double probe electric antenna The instrument paper by Roger Anderson from University of Iowa provides more details about PWE 5 Like the data from most of instruments of CRRES mission PWE data is not archived Our research relies on two data sources: 1 the CDF files that seem to be cooperative product by French aerospace lab ONERA and PWE PI Roger Anderson 2 gif plots of waves activities from University of Iowa The CDF files provide cold plasma density electron gyro-frequency electric field and magnetic field wave power above 100Hz The pixels of gif plots are converted into electric wave power according to the color bars The gif-generated electric field wave power extends to frequency below 100Hz The CDF files and gif plots can be obtained from Physics Department of University of Iowa 2 2 4 Magnetometer The fluxgate magnetometer on board CRRES was to measure both Earth s magnetic field and ultra low frequency waves such as EMIC waves at Pc 1and 2 bands and Pc3-5 Alfvenic ULF waves It also provides the pitch angle for particle instruments The magnetometer s dynamic range goes from a fractional to 48 000 nanotesla The resolution is 0 07 nT at high gain and 3 3 nT at low gain mode For ULF wave measurement the noise level at high sensitivity is about 2 10 3 nT 2 Hz 1 The EMIC wave spectrogram processed from magnetometer data used in this thesis is provided by Dr Nigel Meredith in British Antarctic Survey who might obtained the data from emeritus Prof Brian Fraser from University of Newcastle A list of identified EMIC wave events compiled by Brian Fraser is obtained by the author of this thesis from Dr Nigel Meredith too 17 2 3 UARS and CRRES magnetic conjunction The UARS overlapped with CRRES in the first eleven days of October 1991 before CRRES was lost in early October 12 There are more than 50 conjunctions during this period About 38 conjunction cases were analyzed The rest 12 cases were not analyzed due to the lack of either hiss wave data or cold plasma density data EMIC waves were observed at three conjunctions A conjunction occurs when the two spacecraft s sitting approximately at the same magnetic field tube at the same time as shown by the schematic view of Figure 2 2 The spatial closeness is defined in terms of McIlain L-shell and magnetic local time It is arbitrarily decided that UARS and CRRES are at conjunction when they are within 0 2L and 1 hour magnetic local time from each other within one exception of 0 3L and 1 2 hour This criteria is not largely different from that has been used by other scientists and tighter than 0 5L or even 1L by some authors 27 A conjunction event typically lasts for 50 to 60 seconds All the conjunctions occur when UARS is at northern hemisphere and CRRES is near the magnetic equator The temporal and spatial distribution of all 38 conjunctions are given by 2 3 and 2 4 18 Figure 2 2: schematic view of a magnetic conjunction between UARS at 585km and CRRES at about 33 000km magnetic equator 19 0 20 40 Dst nT 60 80 100 120 140 Dst conjunction events 160 180 10 01 10 02 10 03 10 04 10 05 10 06 10 07 10 08 10 09 10 10 UT Figure 2 3: Temporal distribution of 38 conjunction cases in Oct 1991 10 11 20 7 6 5 6 5 5 L 5 4 5 4 3 5 3 2 5 2 11 12 13 14 15 16 17 magnetic local time hr Figure 2 4: Spatial distribution of 38 conjunction cases in Oct 1991 18 19 21 Chapter 3 THEORETIC FRAMEWORK OF QUASI-LINEAR DIFFUSION The formulas needed for comparison of hiss and EMIC wave theory to the observed electron distribution at loss cone are derived from quasi-linear theory in this chapter The solution of electron distribution function at steady state given by Kennel and Petschek in 1960s will be reviewed and followed by a brief proof for extension to non-steady state New diffusion coefficients formulas for both hiss and EMIC wave are derived from quasilinear theory at a cold plasma Compared to the conventional formulas used in Lyons et al 1972 32 the assumptions of Gaussian wave frequency is dropped while the Gaussian wave normal angle distribution is preserved 3 1 Wave-particle interaction as a diffusion process The theories from 1960s state the wave-particle interaction is a diffusion process described by diffusion equation The Lorentz force of wave field scatter particles reducing the gradient of particle distribution on pitch angle or energy space The magnitude of diffusion coefficient is determined by the wave properties If wave frequency is relatively low and particle energy is high energy diffusion is negligible and only pure pitch angle diffusion is considered This assumption is valid for hiss and EMIC wave scattering 3 1 1 Existing solution of loss cone distribution function at steady state Kennel and Petschek in 1966 24 first proposed that the existence of loss cone makes a sharp gradient of electron distribution at the vicinity of loss cone which generates whistler mode electron cyclotron waves The waves smooth out the gradient by pushing trapped electrons into loss cone The wave growth rate depends on both the magnitude of gradient and number 22 of electrons at resonance The wave growth rate decreases as gradient decreases That is called electrons being dumped into atmosphere by self-generated waves At steady state by assuming a source of particles balancing the loss and a reasonable wave damping rate Kennel and Petschek s diffusion equation for electrons inside loss cone at equatorial magnetosphere is written as 1 0 0 f D 0 0 f TE 3 1 where f is phase space density as function of equatorial pitch angle 0 at constant energy The right hand side term is loss to the atmosphere TE is quarter bounce period It s assumed that there is no source inside loss cone and diffusion coefficient D is constant due to small value of 0 The small size of loss cone at the equator implies sin 0 0 Equation 3 1 can be transformed into modified Bessel equation with variable manipulation As will be shown later the solution inside loss cone is modified Bessel function first kind The diffusion outside loss cone is written as 1 sin 0 0 f D sin 0 s 0 v 0 3 2 where D is diffusion coefficient s 0 v is source term It s assumed there is no additional loss outside loss cone By requiring finite electron flux at the center of loss cone assuming continuity of f and its derivative at the edge of loss cone and f 0 2 0 an analytical solution inside loss cone is obtained S v S v f 0 v h 0 D D D TE I0 0 D TE c I1 c D TE 3 3 where c is size of loss cone D diffusion coefficient inside loss cone S v is integration of source term over pitch angle I0 and I1 are zeroth and first order modified Bessel function of first kind The solution outside loss cone is given by sin 0 S v f 0 v h c ln D sin c 3 4 To obtain 3 4 it is assumed that D 0 D cos 0 which is a good approximation at small and medium pitch angles 23 3 1 2 Extension of steady state solution to non-steady state Kennel and Petschek s theory can not explain the very low flux of electrons at the slot region since the electron flux is too low to be able to self generate waves Lyons Thorne and Kennel in 1972 32 suggested whistler mode hiss wave filled with the entire plasmasphere can scatter radiation belt electrons to very low flux Since hiss wave is generated by a different group of electrons rather than the electrons being scattered it is called parasitic diffusion Research shows chorus wave and lightning contribute as source of hiss wave too In the same year Lyons and Thorne 30 proposed EMIC wave is able to parasitically diffuse radiation belt electrons at fast pace due to it large amplitude One thing we need to keep mind is that diffusion means gradient is being reduced and the distribution function is relaxed That is true even for parasitic diffusion Attaining a steady state is not guaranteed in the parasitic diffusion by hiss and EMIC waves The electrons loss might not be replenished especially in the slot region The balance between wave damping and growth is not guaranteed either A non-steady state solution of loss cone distribution is needed In this section I will prove that solution of loss cone distribution can be obtained by slightly tweaking the steady-state solution The non-steady state diffusion equations are given by 3 5 and 3 6 Roberts in 1969 41 found that any arbitrary pitch angle distribution that might result from an injection event will soon decay to the lowest mode which is very similar to the steady state In other words at non-stead state the shape of distribution function f over pitch angle space is approximately constant The electron flux will see an exponential decay with a characteristic time determined by the magnitude of diffusion coefficient which depends on the strength of waves That makes it possible to use separable variable method to solve equations 3 5 and 3 6 by assuming f 0 t F t g 0 Equation 3 5 become 3 7 where is called electron life time f 1 t 0 0 f f D 0 0 c 0 TE 3 5 24 1 f t sin 0 0 f D sin 0 0 c 0 dF 1 d dtF t g sin 0 d 0 dg D sin 0 d 0 1 1 TE 3 6 3 7 Right hand side of 3 7 is essentially a zeroth-order modified Bessel equation and rewritten as 3 8 The solution is zeroth-order modified Bessel function of first kind equation 3 9 The pitch angle function h 0 in the solution is different from 3 3 only by small modification from As the theory says 30 at small pitch angle is approximately reciprocal of diffusion coefficient The diffusion coefficient of a typical hiss wave is less than 10 3 s 1 At the strong diffusion D than 1 TE 20 2TE 49 Under any situation D is at least two orders of magnitude smaller Therefore h 0 in equation 3 9 become identical as that in 3 3 u2 p 0 where u 0 DTE and 1 0 TE 1 TE dg d2 g u u2 g 0 2 du du 1 1 TE D 1 TE 3 8 20 2TE 1 TE p 0 X v I0 0 D TE p f 0 v X v h 0 c 0 I1 c D T 0 E D T 3 9 E where X v is a constant at given energy and determined by boundary conditions at the edge of loss cone Equations 3 3 and 3 9 show that the pitch angle shape of loss cone distribution can be completely determined by diffusion coefficients from waves whereas the knowledge about electrons outside loss cone is needed to determine the scale of the loss cone population In a steady state when source equals loss the source term S v in 3 3 can be estimated by by integrating the observed electron flux from the three UARS telescopes well inside loss cone see appendix for more details As will be shown in Chapter 4 steady state is a good assumption for the hiss waves which diffuse electrons slowly and have small fluctuation for relatively long time Equation 3 3 along with source estimation is used for hiss wave investigation The steady state might not be a good assumption for EMIC waves which 25 cause strong diffusion and last for short time equation 3 9 is used instead and the scale factor X v is determined by the electron flux right outside loss cone assuming continuity of distribution function across loss cone and the pitch angle shape preservation at non-steady state 3 2 Formula of pitch angle diffusion coefficient Diffusion coefficient along with proper boundary conditions are essential to solve the diffusion equations The bounce-averaged pitch angle diffusion coefficient formulas for hiss and EMIC waves developed by Lyons in 1970s 32 29 from cold plasma quasi-linear diffusion theory have been widely used for several decades The wave frequency distribution was assumed to be a Gaussian distribution with lower and upper frequency cutoff The real wave power spectrum is rarely to be Gaussian One Gaussian fit could deviate the real wave distribution largely Multiple Gaussian fit was used to fit wave power spectrum by some authors 34 It has to been done manually for each spectrum and good fit is not always guaranteed To avoid this cumbersome work in this thesis new formulas of diffusion coefficients for both hiss and EMIC waves are derived The measured wave power spectra is fed to the algorithm without any pre-processing of wave power with curve fitting The rest of this section will start with the relativistic quasi-linear diffusion theory from Lyons Thorne and Kennel s work in 1970s 28 32 29 and walk through the process of deriving new formulas A brief description of MATLAB code implementation and testing is in the appendix Contact the author for MATLAB code In resonant diffusion and low frequency limit the relativistic quasi-linear diffusion equation is reduced to a simple form 28 given by f 1 X f Dn sin t sin n 3 10 for n 6 0 e2 Dn lim V V 2 2 mPk Z k dk 0 kk 2 n k 2 kk n m Pk k 3 11 26 where V is volume of plasma is relativistic factor P is parallel momentum of particles n is harmonic number m is electron rest mass is electron gyro-frequency k and are wave vector and frequency the n k is function of wave electric field given by n k vk Ek R Jn 1 Ek L Jn 1 Ek k Jn v 2 3 12 where Ek k is electric field component parallel to background magnetic field the perpendicular electric field is expressed in terms of right hand Ek R and left hand Ek L components vk and v are particle parallel and perpendicular velocities the argument of first kind Bessel function Jn is k P m is the particle gyro-frequency Ek R Ek L and Ek k can be expressed in terms of wave magnetic power Bk 2 in k space by using cold plasma relation Since the measured wave power is usually in frequency space Bk 2 is further converted to wave power B 2 as function of frequency assuming see Lyons papers 28 29 Bk 2 V B 2 G N 3 13 where N is the normalization factor V is plasma volume G is wave normal angle distribution Though the in G implies the wave frequency dependency of wave normal angle distribution in this thesis wave normal angle distribution is assumed to be the same over all wave frequencies by following Lyons method In the rest part of the thesis the notation will be dropped from wave normal angle term The formulas of N for both hiss and EMIC wave are given in the appendix 3 2 1 Hiss wave bounce-averaged pitch angle diffusion coefficient For whistler wave with moderate wave normal angle and frequency between electron and proton gyro-frequencies with a greatly simplified dispersion relationship 3 14 used by Lyons and other authors 28 29 35 The electric field n k is a related to Bk 2 by 3 15 2 2 pe c2 k 2 2 e cos 3 14 27 2 is plasma frequency k and are magnitude of wave vector and angular frequency where pe e is electron gyro-frequency is wave normal angle c is speed of light in vaccum Bk 2 n k 8c2 2 k kk 2 1 cos Jn 1 1 cos Jn 1 2 3 15 where c is vacuum speed of light is wave normal angle The argument of firs kind Bessel function Jv is k P m e Bk 2 is the wave power on k space By plugging 3 15 3 13 and A 6 into 3 11 transferring from k space to and X tangent of wave normal angle space and evaluating all terms at resonance condition of kk n m Pk 3 11 is rewritten as n e2 2e Dn 2 I 8 Pk2 pe Z Xmax dX Xmin X 2 1 B g X n X 1 X 2 4 n e2 2e In 2 I 8 Pk2 pe 3 16 where I is defined in appendix A 6 n X 1 cos Jn 1 1 cos Jn 1 2 Xmin and Xmax are determined by resonance condition dispersion relationship and the frequency range of hiss waves the integral is replaced by In for compactness wave frequency is evaluated via dispersion relationship and resonance condition at each value of X B 2 is measured wave power spectrum g X is wave normal angle distribution The diffusion coefficient at loss cone is insensitive to wave normal angle models 14 A Gaussian normal angle distribution centered at zero degree with width of 80 is adopted The cyclotron resonance diffusion coefficient at given local pitch angle can be obtained by summing 3 16 over harmonics as following D X e2 2eq h2 n In 2 I 8 P 2 cos2 pe n 3 17 where in this thesis the highest harmonic computed is n 5 the contribution from harmonics higher than 5 is negligible as demonstrated by Lyons and other authors 32 local electron gyro-frequency e is related to equatorial gyro-frequency eq by latitude dependent term h for Earth s dipole field h is given by 1 1 3 sin2 2 h cos6 3 18 28 According to Lyons work 32 the bounce-averaged diffusion coefficient can be evaluated over magnetic latitude and is related to local pitch angle diffusion coefficient by Z m cos 1 D 2 cos7 d D 0 T 0 0 cos 0 3 19 where 0 is equatorial pitch angle T 0 1 30 0 56 sin 0 m is minimum of particle bouncing mirroring latitude and latitude where waves are reflected back to equator For hiss wave the reflecting latitude is believed to be around 45 The diffusion coefficient at loss cone is insensitive to change of m from mirroring latitude to 45 Finally the bounce-averaged pure pitch angle diffusion coefficient is written as Z m X e2 2eq cos7 h2 1 D 0 d n In 1 2 cos2 T 8 P 2 pe 0 0 1 h sin2 0 2 I 0 3 2 2 3 20 EMIC wave bounce-averaged pitch angle diffusion coefficient The full dispersion relationship of EMIC wave see Albert 2003 3 is given by 2 pe c2 k 2 1 2 2 e 2 3 21 by normalizing wave frequency over proton gyro-frequency it is rewritten as 2 M 2 pe Y 2 X Y 2 2 kk c 3 22 where M is electron-to-proton mass ratio X is tangent of wave normal angle Y p is wave frequency over proton gyro-frequency For one value of X the resonant Y has three values with one in each of the three ion bands Hydrogen Helium and Oxygen By using similar notations used by Albert 3 is given by 1 1 1 1 X Y AX 2 1 X 2 Sign L A2 X 4 B 2 1 X 2 2 3 23 2 R L where A 12 P1 12 L1 R1 B 12 L1 R1 R L and P are Stix s cold plasma wave 2 coefficients normalized by pe 2e and given by R 1 X i i M i 1 i Y Zi 3 24 29 L 1 X i i M i 1 i Y Zi 3 25 X 1 2 P 1 M i Zi i M Y 2 i 3 26 where i mi mp is ion-to-proton mass ratio i ni ne is ion-to-electron density ratio Zi is charge number The summation is over ion species with electron excluded The relative proportion of ion species affect the bands of EMIC waves through R L and P terms A 70%H 20%He and 10%O is adopted for magnetic storm time ion proportion as followed by convention 3 The electric field n k in 3 12 is related to magnetic field power via cold plasma relation and given by Bk 2 n k 2 2 3 27 1 An 1 Jn 1 An 1 Jn 1 2 4 3 28 n k 2 where n k 2 is given by n k 2 The coefficients in front of Bessel function Jv are functions of X and Y through dispersion relationship and given by 2 X An 1 s 2 pe c2 kk2 2 D X2 2 S where D 2 S Y 2 pe 2 c kk2 2 pe c2 kk2 2 2 pe 2 c kk2 2 i Y i i Y 2 1 3 29 P c2 kk2 1 X 2 2 M pe D 2 S i i Y 2 i i Y 2 1 3 30 P By plugging 3 27 into 3 11 and using similar manipulation from hiss diffusion coefficient derivation the local pitch angle diffusion coefficient from EMIC wave scattering is given by me e2 2eq h2 X 2 D Dn n In 4 c2 P 3 cos3 n6 0 n6 0 X 3 31 30 where Z Xmax In Xmin B 2 Y g X n k 2 XdX N Y 1 X 2 3 32 kk nmPe e k The normalization factor N Y is given in the appendix Xmin and Xmax determined by resonance condition dispersion relationship and frequency range of EMIC waves The bounceaveraged diffusion coefficient of EMIC wave pitch angle scattering is finally given by Z max X me e2 3eq h3 1 7 d cos D 0 n2 In 3 33 2 P 3 T cos2 M 2 pe 1 sin2 0 h n 0 0 min where min is usually zero from the equator max depends on both resonance condition and wave latitude distribution In this thesis the most conservative choice of 8 is adopted for max according to Fraser and Nguyen s findings that all EMIC wave polarization are seen within 8 of magnetic equator whereas linear polarization dominates over 20 30 latitude 13 A more relaxed latitude model gives larger diffusion coefficient 31 Chapter 4 COMPARISON OF HISS WAVE DIFFUSION THEORY TO ELECTRON LOSS CONE DISTRIBUTION About 38 conjunction cases are analyzed for testing hiss diffusion theory The RMS hiss amplitude varies from a few picotesla to about 120 picotesla as will be shown in later sections Steady state is assumed if hiss wave has been present and fluctuates less than one percent in a time window that is at least twice longer than a typical conjunction time of 50-60 seconds The hiss wave is qualitatively identified by the broad band waves spreading from tens to 1-2 kHz above 10 9 nT 2 Hz 1 when CRRES is traversing outer radiation belts see Figure 4 1 Since the wave normal angle information is missing it is unclear if there is some magnetosonic wave mixed in hiss at the low end frequency of tens of hertz Hiss wave has been known to fill the entire plasmasphere 32 Here it is assumed that hiss wave exist over all latitude Three wave normal angle models are used The first model assumes hiss wave normal angle is a Gaussian distribution 0 center 80 width The second model assumes center of the Gaussian increases with latitude linearly The wave normal angle becomes approximately 90 at latitude of 45 The third model also assumes center of Gaussian increasing with latitude but more slowly The wave normal angle approaches 90 at mirroring latitude The parallel model has been used widely 32 36 The second model is based on some observations that hiss becomes very oblique and bounce back at about latitude of 45 14 The third model is tried since hiss can be observed at aurora altitude 33 32 Figure 4 1: Survey plot of wave power spectrum from CRRES PWE Plasma Wave Ex- periment on Oct 8th 1991 Orbit 1059 The plot is provided by Physics Department University of Iowa Th broad band hiss is typically from tens of Hz to 2 kHz The red line is electron cyclotron frequency 33 4 1 Electron precipitation events consistent with hiss diffusion Figure 4 1 is a survey plot of electric field component of plasma wave from a few Hz to several kHz The broad band whistler mode hiss is labeled At universal time 21:15:00 - 21:15:54 UARS is at footprint of CRRES The wave power spectrum at the conjunction window is plotted in Figure 4 2 The total RMS wave amplitude is about 120 pT The wave power along with cold plasma density L shell and electron energies are fed to the hiss diffusion coefficient formula in Chapter 3 to compute quasi-linear diffusion coefficients using three different latitude models It turns out that diffusion at loss cone is insensitive to latitude model The diffusion coefficients at loss cone from parallel model are 0 0022s 1 at 100 keV 8 8 10 4 s 1 at 500 keV and 2 4 10 4 s 1 at 1 MeV A comparison of hiss diffusion curve to measured electron distribution at loss cone is given by Figure 4 3 The hiss curve is given by 3 3 in Chapter 3 The three similar panels a b and c are for 100 keV 500keV and 1MeV electrons respectively The thick blue curve is electron flux predicted by hiss diffusion theory given by equation 3 3 Under steady state the source equals loss The source term is estimated by adding the amount of electrons seen by three UARS telescopes inside loss cone The red stars are measured electron flux with horizontal error bar indicates the field of view of each telescope The pitch angle at low Earth orbit to transformed to equatorial value using equation of conservation of first adiabatic variant which relates electron pitch angles at different points along dipole magnetic field line The black stars are obtained by integrating the blue curve over field of view of each telescope for direct comparison to red stars The dash blue curve is just shift of thick blue curve by reducing strength of source term by a factor of 2 to account for the uncertainty of source estimation At 100 keV and 500 keV hiss diffusion curve agrees well with measured electron distribution in terms of shape The agreement at 1 MeV is worse but it does not exclude the possibility that hiss along with other loss mechanisms precipitate electrons This case from Figure 4 3 is the best among all 38 cases in terms of shape match between measurements and theory for electrons at 100 and 500 keV No chorus or EMIC wave was 34 4 1 2 x 10 1 wave power nT2 Hz 1 0 8 0 6 0 4 0 2 0 100 200 300 400 500 600 700 800 900 1000 frequency Hz Figure 4 2: Hiss wave power spectra at 21:15:00-21:15:54 Oct 8 1991 when UARS and CRRES are at magnetic conjunction 35 08 Oct 1991 21:15:00 Luars 4 26 Lcrres 4 38 mltuars 16 84 mltcrres 16 7 dL 0 12 dmlt 0 14 3 b 2 a 10 electron flux cm 2 s 1 sr 1 keV 1 electron flux cm 2 s 1 sr 1 keV 1 10 2 10 1 10 1 10 0 10 1 10 0 10 0 1 2 3 4 5 6 equatorial pitch angle degree 0 1 2 3 4 5 6 equatorial pitch angle degree c electron flux cm 2 s 1 sr 1 keV 1 1 10 0 10 1 10 2 10 0 1 2 3 4 5 6 equatorial pitch angle degree Figure 4 3: Comparison of hiss diffusion curve to measured electron flux at loss cone he vertical magenta dash line plots edge of bounce loss cone a 100 keV electron flux vs hiss diffusion the red stars are measured electron flux from three electron telescopes inside loss cone the horizontal color bars are field of view of each telescope the thick blue curve is electron flux distribution predicted by hiss diffusion the black stars are integral of blue curve over field of view of each telescope the dash blue curve is obtained by reducing source term by a factor of 2 b similar to panel a for 500 keV electrons c similar to panel a for 1MeV electrons 36 observed at this conjunction which reduces the possibility of other wave modes Another conjunction between UARS and CRRES occurs at universal time 11:38:01 - 11:38:34 the same day at slot region The diffusion coefficient at 100 keV given by theory is zero An empty loss cone is observed At 500 keV and 1MeV the diffusion coefficients are at small values 10 6 10 7 s 1 The trapped electron flux is lower than detector sensitivity There is no electrons available to be scattered This case is also considered as being consistent with hiss diffusion theory 4 2 Electron precipitation events inconsistent with hiss diffusion Most of the time hiss diffusion seems to be far from adequate to account for the amount of precipitation observed Figure 4 4 shows the observed electron flux at three energy bins versus modified hiss diffusion curve at universal time 00:14:50-00:15:50 Oct 5 1991 The wave amplitude is about 13 pT The hiss diffusion coefficients are magnified by a factor of 10 000 at 100 keV and 2000 at 500 keV and 1MeV to match the measurements The hiss diffusion is negligible compared to the large amount of precipitating electrons observed There is no chorus or EMIC wave present at this conjunction Among 38 cases there are 25 cases resembling this one with two cases coinciding with EMIC waves and 4 coinciding with chorus waves Figure 4 5 gives another type of electron precipitation that can not be explained by hiss diffusion The flat isotropic distribution across loss cone is either from strong diffusion or non-diffusive mechanism The hiss diffusion is too small to cause strong diffusion as indicated by the steep diffusion curve Sometimes the electron distribution is step-like with a flux drop at the edge of loss cone and flat inside loss cone One example is given by Figure 4 6 This electron profile does not look like from a diffusion process Some non-diffusive process like atmospheric backscattering is a possible mechanism There is only one case that hiss diffusion theory seems to overestimate the amount of precipitating electrons given as shown in panel a of Figure 4 7 It is at universal time 37 05 Oct 1991 00:14:50 Luars 4 72 Lcrres 4 46 mltuars 17 07 mltcrres 16 3 dL 0 26 dmlt 0 77 2 10 b electron flux cm s 1 sr 1 keV 1 D x 10000 2 10 1 10 D x 2000 0 10 2 electron flux cm 2 s 1 sr 1 keV 1 a 0 2 10 10 0 1 2 3 4 5 equatorial pitch angle degree 0 1 2 3 4 5 equatorial pitch angle degree electron flux cm 2 s 1 sr 1 keV 1 c D x 2000 0 10 2 10 0 1 2 3 4 5 equatorial pitch angle degree Figure 4 4: Comparison of modified hiss diffusion curve to measured electron flux at loss cone The vertical magenta dash line plots edge of bounce loss cone whereas the horizontal green dash line is detector sensitivity a 100 keV electron flux vs hiss diffusion the red stars are measured electron flux from three electron telescopes inside loss cone the horizontal color bars are field of view of each telescope the thick blue curve is electron flux distribution predicted by hiss diffusion with diffusion coefficient increased by a factor of 10 000 the black stars are integral of blue curve over field of view of each telescope b similar to panel a for 500 keV electrons with diffusion coefficient boosted up by a factor of 2000 c similar to panel a for 1MeV electrons with diffusion coefficient boosted up by a factor of 2000 38 01 Oct 1991 19:08:11 Luars 5 78 Lcrres 6 1 mltuars 14 69 mltcrres 13 33 dL 0 32 dmlt 1 36 3 10 b electron flux cm 2 s 1 sr 1 keV 1 electron flux cm 2 s 1 sr 1 keV 1 a 2 10 1 10 2 10 0 1 2 3 4 0 equatorial pitch angle degree 1 2 3 4 equatorial pitch angle degree electron flux cm 2 s 1 sr 1 keV 1 c 1 10 2 10 0 1 2 3 4 equatorial pitch angle degree Figure 4 5: Comparison of hiss diffusion curve to measured electron flux at loss cone The vertical magenta dash line plots edge of bounce loss cone a 100 keV electron flux vs hiss diffusion the red stars are measured electron flux from three electron telescopes inside loss cone the horizontal color bars are field of view of each telescope the thick blue curve is electron flux distribution predicted by hiss diffusion b similar to panel a for 500 keV electrons c similar to panel a for 1MeV electrons 39 07 Oct 1991 17:46:55 Luars 4 91 Lcrres 4 91 mltuars 11 63 mltcrres 11 93 dL 0 dmlt 0 31 b a 3 electron flux cm 2 s 1 sr 1 keV 1 electron flux cm 2 s 1 sr 1 keV 1 10 2 10 1 10 1 10 0 10 1 10 2 10 0 10 0 1 2 3 4 5 equatorial pitch angle degree 0 1 2 3 4 5 equatorial pitch angle degree electron flux cm 2 s 1 sr 1 keV 1 c 0 10 1 10 2 10 3 10 0 1 2 3 4 5 equatorial pitch angle degree Figure 4 6: Electron flux pitch angle profile with sharp edge of loss cone and flat distribution inside loss cone The vertical magenta dash line plots edge of bounce loss cone The horizontal green dash line is detector sensitivity 40 02 Oct 1991 19:11:08 Luars 6 14 Lcrres 6 15 mltuars 14 46 mltcrres 15 2 dL 0 02 dmlt 0 74 0 a 3 electron flux cm s 1 sr 1 keV 1 2 10 D 10 2 electron flux cm 2 s 1 sr 1 keV 1 b 10 10 1 10 1 10 2 10 3 10 0 1 2 3 equatorial pitch angle degree 0 1 2 3 equatorial pitch angle degree electron flux cm 2 s 1 sr 1 keV 1 c 2 10 3 10 0 1 2 3 equatorial pitch angle degree Figure 4 7: Comparison of hiss diffusion curve to measured electron flux at loss cone The the vertical magenta dash line plots edge of bounce loss cone The horizontal green dash line is detector sensitivity a 100 keV electron flux vs hiss diffusion the red stars are measured electron flux from three electron telescopes inside loss cone the horizontal color bars are field of view of each telescope the thick blue curve is electron flux distribution predicted by hiss diffusion the dash blue line is hiss curve with diffusion coefficient decreased by a factor of 10 b similar to panel a for 500 keV electrons c similar to panel a for 1MeV electrons 41 19:11:08 - 19:12:20 Oct 2 1991 The amplitude of hiss is about 62pT The diffusion coefficient at 100 keV has to go down by a factor of 10 to fit the observation As mentioned earlier the choice of wave normal angle model does not make big difference There is no knob to be tuned in the model to fit the measured electron flux The UARS electron correction algorithm at this event is less reliable at 100 keV due to proton over correction It is unclear if the discrepancy is caused by assumptions in the diffusion model or electron data correction algorithm 4 3 Conclusion for hiss diffusion theory testing As mentioned in Chapter 2 there was a magnetic storm with minimum Dst -149 nT in the first ten days of October 1991 The 38 magnetic conjunctions between UARS and CRRES spread out over all phases of the magnetic storm Hiss has been proposed to be the main loss mechanism during quiet time 31 48 Thus the storm time cases might not be ideal for testing the theory of electron precipitation by hiss diffusion There are two cases showing good consistency between hiss theory and measured electron loss cone distribution In only one case hiss predicts more electrons than observed at 100 keV due to some unknown factor either from theory or data correction algorithm Most of the time hiss diffusion is far from sufficient to account for the large amount of precipitation observed which suggests existence of other unknown mechanisms There is no case that the presence of hiss causes no electron precipitation Therefore it is concluded that the hiss diffusion theory seems be consistent with the electron loss cone distribution when no other loss mechanism is present 42 Chapter 5 COMPARISON OF EMIC WAVE DIFFUSION TO PRECIPITATING AND TRAPPED ELECTRONS BOTH AT LOW EARTH ORBIT AND MAGNETIC EQUATOR As mentioned in the first chapter quasi-linear theory of EMIC wave scattering predicts strong diffusion at loss cone fast depletion of trapped electrons pancake-shape distribution of trapped electrons resulting from selective scattering at low and medium pitch angles and relativistic electron precipitation embedded in broader precipitation of tens of keV ions In the following sections the measured loss cone distribution will be compared to what the quasi-linear theory predicts The trapped electron profile and selective pitch angle scattering will be examined as well At the end the findings will be compared to those from other authors to have a better understanding of current quasi-linear theory of EMIC diffusion and the role of EMIC waves in radiation belt electron loss 5 1 5 1 1 Case studies Testing the theory with electrons at loss cone Among about 38 magnetic conjunctions between UARS and CRRES during the first ten days of October 1991 only three coincide with EMIC wave activities at the vicinity of magnetic equator Right column of Figure 5 1 show three groups EMIC waves labeled as case 1 2 and 3 observed by CRRES satellite at north of magnetic equator with latitude within about 8 All three EMIC wave events are characterized as peaking at hydrogen band and being left hand polarized most of the time according to comprehensive list of EMIC waves by Emeritus Prof Brian Fraser from University of Newcastle The EMIC waves are spotted with a combination of manual checking on the wave power near ion cyclotron frequencies 43 frequency Hz Case 1 0 10 2 10 5 4 1 0 5 2 0 1 0 05 0 L MLT 10 8 frequency Hz 0 10 2 10 2 Case 3 6 01 Oct 1991 0 10 frequency Hz 18:30 19:00 5 76 12 93 6 04 13 25 19:30 6 24 13 53 0 2 10 4 1 0 5 2 0 1 0 05 L MLT 8 10 0 10 frequency Hz 20:30 21:00 21:30 22:00 6 41 13 83 6 42 14 08 6 37 14 32 6 26 14 57 0 01 100 50 6 10 5 4 1 0 5 2 0 1 0 05 04 Oct 1991 18:45 19:00 19:15 19:30 19:45 L MLT 5 83 12 89 5 97 13 04 6 09 13 19 6 19 13 32 6 27 13 46 0 Figure 5 1: Three EMIC wave events from CRRES with case 1 on October 1st 1991 and case 2 and 3 at two consecutive orbits on Oct 4th 1991 The right column panels plot wave power spectrum on frequency domain versus universal time L shell and magnetic local time The black double lines on UT axis marks the time slots when UARS is at footprint of CRRES aka conjunction The red double line marks the reference time Two white dash lines from top to bottom within each panel are local hydrogen and helium gyro-frequencies The left column panels plot wave power spectrum at the time slots marked the double black and red lines in the right column The black curve is wave power spectrum at the conjunction time while red curve is wave power at reference time The blue vertical dash lines show cutoff frequencies bounding the peaks 100 50 10 5 0 10 0 01 6 04 Oct 1991 frequency Hz Case 2 2 0 10 frequency Hz 100 50 2 1 wave power intensity nT Hz 10 8 0 01 wave power intensity nT2 Hz 1 2 10 44 wave power above the instrument noise level of about 10 3 nT 2 Hz and polarization analysis carried by Prof Brian Fraser The black curves in the left column of Figure 5 1 are wave power spectrum on frequency domain at the conjunction time intervals marked by the double black lines in the right column The red curves are the background wave power spectrum at reference when there is no EMIC wave marked by the red double lines in the right column The EMIC wave power spectrum is obtained by subtracting red curve from black curve The two blue vertical dash lines are cutoff frequencies picked manually It is assumed that wave power is zero outside the frequency range bounded by cutoff frequencies Then the wave power spectrum along with L shell cold plasma density and electron energy are fed to the diffusion coefficients formula given in Chapter 3 to compute diffusion coefficients It is assumed that EMIC waves are confined within 8 latitude of magnetic equator The diffusion coefficient will be larger with larger latitude range Only diffusion at hydrogen band is computed Two Gaussian wave normal angle models are used: parallel propagating and oblique at 30 The Gaussian width for both models is 15 The minimum resonant energy is computed as well to have independent check on the diffusion coefficient formula The minimum energy is determined by using equations 3 4 and 5 in Denton et al 2015 11 Denton s formulas are based on first order resonance condition and EMIC wave dispersion relationship The electron flux from UARS at conjunctions are plotted in the top three rows of Figure 5 2 The left middle and three columns correspond to case 1 2 and 3 respectively The red dots are electron flux observed right outside loss cone Blue green and cyan are for electrons inside loss cone The closeness of conjunctions in terms of McIlain L shell and magnetic local time is plotted in Row d The bottom row plots L shell profile of both trapped red and precipitating blue green and cyan 1MeV electron flux at a time window of one minute right before and after conjunctions The center location of conjunctions is marked by the vertical dash lines For the first conjunction at universal time 19:08:11 to 19:08:40 on Oct 1st 1991 isotropic distribution across loss cone is observed The peak EMIC wave power is about 3 28 nT 2 Hz 1 45 a 2 MeV 1 MeV 0 5 MeV L cm 2 s 1 sr 1 keV 1cm 2 s 1 sr 1 keV 1cm 2 s 1 sr 1 keV 1 MLT hour case 1 case 2 case 3 0 10 1 10 2 10 b 3 100 10 1 10 2 10 c 3 100 10 1 10 2 10 3 10 d 1 MeV cm 2 s 1 sr 1 keV 1 LUARS LCRRES 0 1 e MLTUARS MLTCRRES 1 19:08:15 19:08:24 19:08:32 UT 01 Oct 1991 20:54:31 20:54:48 20:55:06 UT 04 Oct 1991 20:55:2319:16:36 19:16:53 19:17:11 19:17:28 UT 04 Oct 1991 0 10 1 10 2 10 3 10 5 3 5 51 5 73 L 5 95 5 26 5 98 6 69 L 7 41 5 5 62 L 6 25 6 88 Figure 5 2: Trapped and precipitating electron flux observed by UARS at three UARS CRRES conjunctions The left middle and right columns correspond to case 1 2 and 3 respectively From top to bottom a 0 5 MeV electron differential flux with red for electron telescope right outside loss cone blue green and cyan for three telescopes well inside loss cone the horizontal magenta dash lines show detector sensitivity those below sensitivity line are due to over-correction of correction algorithm b and c 1MeV and 2MeV electron differential flux with color code the same as 0 5 MeV electrons The events pointed by the two black arrows are re-plotted in Figure 5 3 and 5 4 d difference of L shell and magnetic local time of UARS magenta and CRRES blue e L profile of 1MeV electron flux at time window of one minute right before and after conjunction red for electron telescope right outside loss cone blue green and cyan for three telescopes well inside loss cone The vertical two black dash lines bound the conjunction window 46 at 0 925 Hz The cold plasma density is 7 23 cm 3 The cutoff frequencies are 0 75 Hz at lower end and 1 2 Hz at higher end The local proton gyro-frequency is 2 23 Hz The minimum resonant energy given by Denton s formula at 1 2 Hz is about 8 35 MeV The diffusion coefficients are all zeros at 0 5 1 and 2 MeV The isotropic distribution looks like to be the signature of strong diffusion from EMIC wave as predicted by the theory However the zero diffusion coefficients and high minimum resonant energy excludes this possibility Furthermore the isotropic distribution is spotted twice again without presence of EMIC waves at UT 20:44:51 and 22:22:16 when UARS traverses the same L and approximately same magnetic local time Since early 1970s it has been known that isotropic distribution is very commonly found at trapping boundary of radiation belts where the particle gyroradius is comparable to the curvature of Earth s magnetic field 16 By looking at wider time window of one minute right before and after conjunction bottom row of Figure 5 2 it is clear that case 1 conjunction occurs right at some boundary while case 2 and 3 are not Trapping boundary is a spatial structure that can be seen quite often by multiple passes of low Earth orbit satellites regardless of presence of EMIC waves More details about properties of trapping boundaries can be found in Imhof s investigation on UARS data in 1997 17 Therefore it is likely that the isotropic distribution here is because UARS is at trapping boundary That EMIC wave and strong precipitation appear at the same time doesn t mean the two should have connection This case demonstrates the importance of quantitative evaluation in linking EMIC wave to electron precipitation The second conjunction occurs around 20:55 Oct 4th 1991 The electron precipitation is embedded in a broad ion precipitation Figure 5 5 show such enhancement of tens of keV proton precipitation at a time window that well encapsulates the time frame of electron precipitation at 20:54:30 - 20:55:20 The flux of precipitating protons at 74 -80 keV is greater than 103 cm 2 s 1 sr 1 keV 1 Though measurement of trapped protons is not available from POES observed EMIC wave- driven proton precipitation at about same location see Engebretson et al in 2015 12 this amount of precipitating proton flux is comparable to typical flux of trapped protons This is a signature of EMIC wave scattering protons The 47 EMIC wave is more intense than that in the first case peaking 69 68 nT 2 Hz 1 at 1 005 Hz The cold plasma density is 11 36 cm 3 The cutoff frequencies are 0 45 at lower end and 2 1 Hz at higher end The local proton gyro-frequency is 2 13 Hz The minimum resonant energy is 139 keV The loss cone diffusion coefficients from parallel propagating wave model are 0 0107 s 1 at 0 5 MeV 0 0151 s 1 at 1 MeV and 0 0298 s 1 at 2MeV Oblique wave model gives 0 0070 s 1 at 0 5 MeV 0 0199 s 1 at 1 MeV and 0 0176 s 1 at 2 MeV Either model suggests diffusion hits the strong diffusion limit of 0 0066 s 1 at 0 5 MeV 0 0073 s 1 at 1 MeV and 0 0075 s 1 at 2 MeV Thus an isotropic distribution across loss cone is expected Contrarily the measured trapped flux is one to two order of magnitude higher than flux inside loss cone as shown in the middle panels of Row a - c in Figure 5 2 The closest point between trapped and precipitating flux occurs at 20:54:52 and 20:55:05 UT pointed by the arrows The electron flux at these two time stamps is re-plotted in Figure 5 3 and 5 4 over pitch angle space for comparison to the theoretical EMIC wave diffusion curves As mentioned in Chapter 3 the theoretical curves are forced to line up with electron flux right outside loss cone Even at the nearest point the discrepancy between observation and theory is still very large The diffusion coefficients given by parallel model have to go down by a factor of about 16 for 1MeV electrons and 13 for 2 MeV electrons to fit the observation The diffusion coefficients given by oblique model need to be decreased by a factor of 21 at 1 MeV and 9 at 2 MeV It seems that the discrepancy shrinks from 1 MeV to 2 MeV In case 2 since the higher end cutoff frequency is close to the proton gyro-frequency small change of the cutoff frequency will change the minimum resonant energy by noticeable amount If the cutoff is chosen to be at 1 665 Hz as shown in Figure 5 6 the minimum resonant energy jumps to 1 75 MeV The diffusion coefficients at 0 5 and 1 MeV go to zero Then the 1 MeV electron precipitation at 21:55:05 UT can not be explained by EMIC wave diffusion But the diffusion coefficient of 2 MeV is 0 0154 s 1 which still goes beyond strong diffusion limit The large discrepancy between the theory and observation still exists The third conjunction case occurs around 19:17 Oct 4th 1991 The peak EMIC wave power is about 20 93 nT 2 Hz 1 at 1 55 Hz The cold plasma density is 9 73 cm 3 The cutoff 48 frequencies are 1 4 Hz at lower end and 1 65 Hz and higher end The local proton gyrofrequency is 2 77 Hz The diffusion coefficients from both parallel and oblique models are zero at 0 5 1 and 2 MeV The minimum resonant energy is 6 3 MeV The non-empty loss cone might be filled by other processes if the theory gives the correct amount of diffusion 5 1 2 Testing the theory with profile of trapped electrons The second EMIC wave event lasts for about two hours with a 12min gap between 21:42 and 21:54 The CRRES spacecraft is moving very slow and mostly azimuthally It stays at the same L shell of about 6 4 and spans only 0 65 hr in magnetic local time from 20:23 to 21:42 UT which suggests the electron flux variation observed by CRRES is mostly temporal variation It is good opportunity to test what has been suggested by Summers 46 that EMIC wave residing in just 1 percent of magnetic local time is able to empty the radiation belt at a time scale of a couple of hours The time evolution of electron flux from CRRES MEA is compared to EMIC wave diffusion curve In short time period the characteristic decay time of EMIC waves at small pitch angle is reciprocal of bounce averaged diffusion coefficient 30 In a time period much longer than drift period the magnetic local time distribution of EMIC waves need to be considered to compute bounce-drift averaged diffusion It is assumed that EMIC wave is confined within 0 65 hr of magnetic local time as from the observation The EMIC wave diffusion curves for both bounce and bounce-drift averaged diffusion are plotted in the middle row of Figure 5 7 The thin solid and dash red lines are bounce-drift averaged and bounce-averaged diffusion respectively The thick wiggly blue curve is electron flux at pitch angle of 5 measured by MEA electron telescope on board CRRES at 1090 keV Obviously the measured electron profile matches neither bounce-drift averaged nor bounce-averaged diffusion At the vicinity of wave onset the 1090 keV electron flux drops by about a factor of 2 at all pitch angles as shown in the top row of Figure 5 7 There seems to be a connection between electron flux drop and EMIC wave However the flux drop seems to begin at 20:22 one minute before the wave onset and occurs at all pitch angles EMIC wave diffusion theory 49 0 EMIC predicted EMIC predicted diffusion coef decreased by 13 measured 2 MeV electron flux at 04 Oct 1991 20:54:52 10 1 10 cm 2 s 1 sr 1 keV 1 a edge of bounce loss cone 2 10 3 100 10 EMIC predicted EMIC predicted diffusion coef decreased by 16 measured 1 MeV electron flux at 04 Oct 1991 20:55:05 1 10 sr 1 keV 1 b cm 2 s 1 detector sensitivity 2 10 3 10 0 0 5 1 1 5 2 2 5 3 equatorial pitch angle degree Figure 5 3: Comparison of theoretical loss cone distribution to measurements from UARS around 21:55 UT Oct 4 1991 The diffusion coefficients are computed assuming parallel propagating EMIC waves Row a and b are for 2MeV and 1MeV electrons respectively The red thick curves are prediction given by EMIC wave diffusion theory Red dash lines are EMIC wave curve with diffusion coefficients decreased by a factor of 16 for 1MeV electrons and 13 for 2 MeV electrons The black dots are electron differential flux measured by four telescopes on board UARS with one right outside loss cone three inside loss cone The horizontal error bars are field of view of telescopes The pink vertical dash lines mark the edge of bounce loss cone The detector sensitivity is marked by horizontal blue dash lines The pitch angles of UARS telescopes at low Earth orbit are converted into equatorial values in degree 50 0 10 1 10 cm 2 s 1 sr 1 keV 1 EMIC predicted EMIC predicted diffusion coef decreased by 9 measured 2 MeV electron flux at 04 Oct 1991 20:54:52 edge of bounce loss cone 2 10 3 100 10 1 10 sr 1 keV 1 EMIC predicted EMIC predicted diffusion coef decreased by 21 measured 1 MeV electron flux at 04 Oct 1991 20:55:05 cm 2 s 1 detector sensitivity 2 10 3 10 0 0 5 1 1 5 2 2 5 equatorial pitch angle degree Figure 5 4: Similar to Figure 5 3 assuming oblique propagating EMIC waves 3 51 101 100 2 0 105 1 5 105 10-1 1 0 105 10-2 20:52 1991-10-04 20:54 20:56 20:58 4 5 105 4 0 105 3 5 105 5 3 0 10 2 5 105 101 100 2 0 105 1 5 105 10-1 1 0 105 10-2 20:52 1991-10-04 20:54 20:56 I cnts cm 2-ster-s-e eV eV cnts cm 2-ster-s-eV eV eV 4 5 105 4 0 105 5 3 5 10 3 0 105 2 5 105 20:58 Figure 5 5: Precipitating protons from UARS at 550km The upper and bottom panels plot proton flux at pitch angle 22 and 14 respectively Both are well inside bounce loss cone The UARS CRRES conjunction is at 20:54:30 to 20:55:20 well inside the proton precipitation window The lower energy cutoff is 74 keV Notice that the energy is in eV and flux is in cm 2 s 1 sr 1 eV 1 52 2 10 wave spectrum at reference wave spectrum at conjunction 1 10 2 1 Hz 0 10 nT2 Hz 1 665 Hz 1 10 2 10 0 45 Hz 3 10 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 EMIC wave frequency Hz Figure 5 6: Two choices of cutoff frequencies The black curve is wave power spectrum at case 2 conjunction around 21:55 UT Oct 4 1991 The green curve is the wave power spectrum used as background and measured around 20:21 UT Oct 4 1991 The magenta vertical dash line marks lower end cutoff frequency of 0 45 Hz blue and red lines mark two choices of cutoff frequencies of 1 665 Hz and 2 1 Hz at higher end 5 53 2 cm 2 s 1 sr 1 keV 1 10 pa 5o pa 10o pa 20o pa 30o pa 40o pa 50o pa 60o pa 70o pa 80o pa 90o 1 10 0 10 1 10 2 10 cm 2 s 1 sr 1 keV 1 1 10 measured 1090 keV e flux pa 5o 0 10 EMIC predicted 1090 keV e flux bd model EMIC predicted 1090 keV e flux b model 100 50 frequency Hz 4 10 5 3 2 1 0 5 1 0 1 0 05 0 04 Oct 1991 20:24 L MLT 6 40 13 78 20:38 20:52 21:07 21:21 21:36 21:50 22:04 22:19 6 42 13 90 6 43 14 02 6 42 14 13 6 40 14 25 6 36 14 37 6 30 14 49 6 23 14 62 6 14 14 75 Figure 5 7: Comparison of trapped electron from MEA on board CRRES to EMIC wave diffusion a differential flux of 1090 keV electrons at pitch angles from 5 to 90 b thick blue curve is 1090 keV electron differential flux at pitch angle of 5 measured by MEA The thin solid red line is prediction from EMIC wave diffusion with bounce-drift averaged model The dash magenta line is prediction from EMIC wave diffusion with bounce-averaged model c EMIC wave spectrogram versus universal time L shell and magnetic local time The two white meandering dash lines from top to bottom mark local hydrogen and helium gyro-frequencies 0 01 wave intensity nT2 Hz 1 5 54 suggests the diffusion is effective only at low and medium pitch Furthermore multiple such flux pulsation were observed in several hours right before the wave onset It is unlikely that EMIC wave would cause such flux drop A correlation study between square of EMIC wave amplitude and electron flux at small pitch angles from UT 20:15 to 22:20 is carried out to see if there is any hidden relationship between EMIC wave and electrons at the edge of loss cone Figure 5 8 The diffusion coefficient is proportional to wave amplitude square The characteristic decay time is reciprocal of diffusion coefficient for electrons at small pitch angles As shown in Figure 5 8 the correlation between trapped electron flux outside loss cone and EMIC wave amplitude is very weak The Pearson correlation coefficients are -0 08 with p-val of 0 2 at 509 keV and -0 12 with p-val of 0 04 at 1090 keV Another interesting signature of EMIC wave diffusion is the selective scattering at low and medium pitch angles Though the electrons near 90 are being peeled off by other processes starting at 20:38 UT which make it more difficult to see selective scattering there are time intervals without obscuring during the 2-hour long wave activities No selective scattering is observed Before preceding to conclusion one needs to consider if other processes could mask the EMIC wave scattering from being distinguished There exist whistler mode electron waves mostly hiss at the same time EMIC wave is observed The magnitude of diffusion coefficient by hiss scattering is at about 10 5 s 1 three orders of magnitude smaller than EMIC wave diffusion The electron depletion by EMIC will dominate the loss process according to what the theory predicts The 1 MeV electron flux at all pitch angles start recovering at 21:00 UT and goes back to about the same level before 20:22 UT and stay relatively constant If the strong scattering suggested by EMIC wave diffusion theory exists it would require a very strong local sources to replenish the MeV electrons in a relatively short time 55 2 CRRES MEA e flux cm 2 s 1 sr 1 keV 1 10 1 10 0 10 0 5 10 15 20 25 30 35 2 Hydrogen band EMIC wave amplitude square nT Figure 5 8: hydrogen band EMIC wave amplitude square versus electron flux at pitch angle of 5 measured by MEA on board CRRES The black and red dots are for 509 keV and 1090 keV respectively 56 5 2 Concluding remarks from case studies To summarize the electron loss cone distribution deviates largely from what theory predicts The discrepancy is at least between a factor 9 to 21 depending on the electron energy and wave normal angle models It seems that the discrepancy at 2 MeV is smaller than at 1 MeV Neither selective pitch angle scattering nor significant depletion of trapped electrons is observed at energy up to 2 MeV 5 3 Comparison to findings from others This section will dive deep into the findings by other authors since 1970s and incorporate findings in this thesis to have a better understanding of EMIC wave diffusion First of all let us look back to history to see if evidence of strong diffusion isotropic distribution at loss cone was reported Secondly I will relate new findings in this thesis to the seemingly contradictory evidences reported by Rodger and Usanova and Engerbretson The end of this section will be a discussion of EMIC wave diffusion at ultra-relativistic energy of greater than 2 MeV In earlier days when simultaneous wave measurements in magnetosphere was lacking electron precipitation with simultaneous ion precipitation observed by low Earth orbit satellites were investigated for searching evidence of relativistic electron precipitation by EMIC wave scattering At the same year Thorne and Kennel 47 first introduced EMIC waves as loss mechanism in 1971 Vampola reported three relativistic events that seemed to be associated with EMIC wave scattering 51 All three events are just inside plasmapause between late evening and midnight The isotropy increases toward higher energy There was no information about wave and protons in Vampola s paper Later Thorne and Andreoli in 1980 49 attributed four isotropic electron precipitation events to EMIC wave scattering Figure 3 of Thorne and Andreoli s paper gives an example of such event Though simultaneous wave measurement was not available the broader simultaneous ion precipitation and selective precipitation at higher energy greater than 850 keV look very promising The 57 event given by Thorne and Andreoli occurs at a boundary at local time 20 18 hour and L shell of 4 35 This boundary looks very similar to what later study identified as trapping boundary 17 Isotropic distribution is quite often to be spotted at trapping boundary By avoiding the trapping boundary events Imhof in 1986 identified nine energy selective electron precipitation accompanied by ion precipitation near plasmapause at evening local time The distribution is far from isotropic with electrons near 0 often one order of magnitude lower than electrons at 90 To my best knowledge no isotropic events were reported by other authors after Thorne and Andreoli in 1980 As simultaneous wave measurements either from the ground or space become available one-to-one correspondence analysis had been reported Miyoshi et al in 2008 associated an electron precipitation event observed by POES to simultaneous EMIC waves recorded by ground magnetometer Figure 5 9 is a snapshot of Miyoshi s main findings The hump of greater than 3 MeV flux on the bottom panel is believed to be evidence of EMIC wave -induced electron precipitation It is unclear how much of the variation is spatial effort given that POES moves fast at low Earth orbit and both trapped and precipitating electrons at lower energy also see a big hump Furthermore the trapped 3 MeV electron flux is not available for comparison to precipitating electrons The wave amplitude in space was estimated from the ground-based observation The diffusion coefficient was then estimated with assumptions on the wave properties and cold plasma condition Miyoshi s conclusion is EMIC waves caused the observed electron and proton precipitation Rodger et al in 2015 42 reported a burst of protons and electrons precipitation observed by POES when Van Allen probe at conjunction observes strong helium band wave at L greater than 5 The precipitating electron flux is estimated to be 1 25 104 cm 2 s 1 sr 1 at energy above 140-230 keV The POES trapped electron flux is not reported From eyeball estimation of trapped electron flux from Van Allen probes in Figure 4 of Rodger s paper the trapped electron flux near the edge of loss cone is at least one order of magnitude higher than estimated precipitating flux and even higher at higher pitch angles The diffusion coefficient is not given by Rodger It is unknown how close is the observation to the theory As 58 Figure 5 9: snapshot of Figure 3 of Miyoshi et al 2008 38 a Emission profile of Hb and be count rates of energetic ions and electrons observed by POES-17 on September 5 2005 with UT MLAT the McIlwain L-value and magnetic local time MLT of the satellite footprint during the period when the satellite was crossing over Athabasca The vertical blue line indicates the time when the satellite footprint crossed the stable isolated proton aurora MLAT and L-value are calculated by the IGRF model 59 Figure 5 10: Snapshot of Figure 8 of Zhang XJ et al 2016 56 a Observed electron pitch angle distributions from Van Allen Probes at L 5 77 b The evolution of electron distribution after interaction with EMIC and hiss c only EMIC and d only hiss waves for 40 min Initial distributions are shown in dotted lines observations or model results at t 40 min are shown in solid lines demonstrated in case 1 study in previous sections simultaneous observations of EMIC waves and electron precipitation doesn t necessarily imply EMIC wave causes the precipitation Contrarily to Rodger s findings as mentioned in Chapter 1 Usanova and Engebretson reported the absence of electron precipitation at the presence of strong and long lasting EMIC waves The amount of electrons is too low to be detected by POES The findings in this thesis agree with Usanova and Engebreson s findings There is no evidence of significant loss of electrons caused by EMIC waves 60 Before drawing conclusion for this chapter it is worthwhile to investigate greater than 2 MeV electrons Though EMIC seems to be ineffective at scattering electrons with energy smaller than 2 MeV it is still possible that EMIC wave is able to precipitate electrons at ultrarelativistic energies Both Usanova and Engebretson reported the pitch angle distribution of several MeV electrons is more pancake-like at the presence of EMIC wave Zhang XJ et al in 2016 56 simulated EMIC wave diffusion for a pancake case by taking advantage of identical orbit and time elapse between the two Van Allen probes Figure 5 10 is snapshot of Figure 8 in Zhang XJ s paper The Van Allen probe B and A went through the same point in time lapse of 40 min The observation from probe B is used as initial condition for the simulation The simulation agrees with the observations at energy greater than 1 8 MeV and medium pitch angles It does not reproduce the electron profile at lower energy These findings agree with one of the findings in this thesis that the discrepancy between measurement and theory inside loss cone decreases at higher energy This could give a guidance to the direction of theory modifications 5 4 Conclusions of testing EMIC wave diffusion theory In summary three EMIC wave events have been analyzed The first case shows isotropic distribution which results from process other than EMIC wave scattering It demonstrates that simultaneous presence of electron precipitation and EMIC waves doesn t guarantee any connection between the two A large discrepancy between observation and what EMIC wave diffusion theory predicts is found from the second case If EMIC wave were able to precipitate electrons from several hundred keV to about 2 MeV the magnitude of diffusion should be significantly smaller than what current theory suggests In the third case the resonant energy is much higher than the upper limit of electron detector no conclusion is drawn from that It is likely that EMIC wave is more effective at higher energies 61 Chapter 6 CONCLUSION AND FUTURE WORK Hiss EMIC and chorus waves have been the three major waves included in radiation belt electron models and simulations by many space scientists Two of the three waves hiss and EMIC waves are tested against the electron loss cone distribution in this thesis The loss cone electron pitch angle distribution that can be completely explained by hiss diffusion theory is found Hiss wave is also found to be far insufficient to account for the electron precipitation among 25 out of 38 cases Only 4 out of the 25 cases see the presence of either chorus or EMIC waves which suggests loss mechanisms other than the three major waves modes are present The strong diffusion of electrons predicted by EMIC wave theory has not been found Current quasi-linear theory of EMIC wave diffusion significantly overestimates the electron precipitation by EMIC waves at energy up to at least 2 MeV More observations at ultra-relativistic energies are needed to quantitatively evaluate the role of EMIC wave in radiation belt electron loss 62 BIBLIOGRAPHY 1 J M Albert Quasi-linear pitch angle diffusion coefficients: Retaining high harmonics Journal of Geophysical Research: Space Physics 99 A12 :23741 23745 1994 2 J M Albert Analysis of quasi-linear diffusion coefficients Journal of Geophysical Research: Space Physics 104 A2 :2429 2441 1999 3 J M Albert Evaluation of quasi-linear diffusion coefficients for emic waves in a multispecies plasma Journal of Geophysical Research: Space Physics 108 A6 2003 4 B J Anderson R E Erlandson and L J Zanetti A statistical study of pc 12 magnetic pulsations in the equatorial magnetosphere: 1 equatorial occurrence distributions Journal of Geophysical Research: Space Physics 97 A3 :3075 3088 1992 5 ROGER R ANDERSON DONALD A GURNETT and DANIEL L ODEM Crres plasma wave experiment Journal of Spacecraft and Rockets 29 4 :570 573 Jul 1992 6 Hendry A T Experimental evidence and properties of EMIC wave driven electron precipitation Thesis Doctor of Philosophy University of Otago 2018 7 L W Blum A Halford R Millan J W Bonnell J Goldstein M Usanova M Engebretson M Ohnsted G Reeves H Singer M Clilverd and X Li Observations of coincident emic wave activity and duskside energetic electron precipitation on 1819 january 2013 Geophysical Research Letters 42 14 :5727 5735 2015 8 A W Breneman A Halford R Millan M McCarthy J Fennell J Sample L Woodger G Hospodarsky J R Wygant C A Cattell J Goldstein D Malaspina and C A Kletzing Global-scale coherence modulation of radiation-belt electron loss from plasmaspheric hiss Nature 523:193 EP Jun 2015 9 Max D Comess David M Smith Richard S Selesnick Robyn M Millan and John G Sample Duskside relativistic electron precipitation as measured by sampex: A statistical survey Journal of Geophysical Research: Space Physics 118 8 :5050 5058 2013 10 G Davidson and M Walt Loss cone distributions of radiation belt electrons Journal of Geophysical Research 82 1 :48 54 1977 63 11 R E Denton V K Jordanova and J Bortnik Resonance of relativistic electrons with eletromagnetic ion cyclotron waves Geophys Res Lett 2015 12 M J Engebretson J L Posch J R Wygant C A Kletzing M R Lessard C -L Huang H E Spence C W Smith H J Singer Y Omura R B Horne G D Reeves D N Baker M Gkioulidou K Oksavik I R Mann T Raita and K Shiokawa Van allen probes noaa goes and ground observations of an intense emic wave event extending over 12 h in magnetic local time Journal of Geophysical Research: Space Physics 120 7 :5465 5488 2015 13 B J Fraser and T S Nguyen Is the plasmapause a preferred source region of electromagnetic ion cyclotron waves in the magnetosphere Journal of Atmospheric and Solar-Terrestrial Physics 63 11 :1225 1247 2001 The Plasmasphere Revisited: A Tribute to Donald Carpenter 14 Yuzhu Gao Fuliang Xiao Qi Yan Chang Yang Si Liu Yihua He and Qinghua Zhou Influence of wave normal angles on hiss-electron interaction in earth s slot region Journal of Geophysical Research: Space Physics 120 11 :9385 9400 2015 2015JA021786 15 Rachael Hardman Mark A Clilverd Craig J Rodger James B Brundell Roger Duthie Robert H Holzworth Ian R Mann David K Milling and Eva Macusova A case study of electron precipitation fluxes due to plasmaspheric hiss Journal of Geophysical Research: Space Physics 120 8 :6736 6748 2015 2015JA021429 16 W L Imhof R R Anderson J B Reagan and E E Gaines Coordinated measurements of slot region electron precipitation by plasmaspheric wave bands Journal of Geophysical Research: Space Physics 87 A6 :4418 4426 1982 17 W L Imhof D L Chenette E E Gaines and J D Winningham Characteristics of electrons at the trapping boundary of the radiation belt Journal of Geophysical Research: Space Physics 102 A1 :95 104 1997 18 W L Imhof R M Robinson H L Collin J R Wygant and R R Anderson Simultaneous equatorial measurements of waves and precipitating electrons in the outer radiation belt Geophysical Research Letters 19 24 :2437 2440 1992 19 W L Imhof R M Robinson H L Collin J R Wygant and R R Anderson Simultaneous measurements of waves and precipitating electrons near the equator in the outer radiation belt Journal of Geophysical Research: Space Physics 99 A2 :2415 2427 1994 64 20 W L Imhof H D Voss J B Reagan D W Datlowe E E Gaines J Mobilia and D S Evans Relativistic electron and energetic ion precipitation spikes near the plasmapause Journal of Geophysical Research: Space Physics 91 A3 :3077 3088 1986 21 Juro Ishida Susumu Kokubun and Robert L McPherron Substorm effects on spectral structures of pc 1 waves at synchronous orbit Journal of Geophysical Research: Space Physics 92 A1 :143 158 1987 22 M H Johnson and John Kierein Combined release and radiation effects satellite crres : Spacecraft and mission Journal of Spacecraft and Rockets 29 4 :556 563 Jul 1992 23 S Kasahara Y Miyoshi S Yokota T Mitani Y Kasahara S Matsuda A Kumamoto A Matsuoka Y Kazama H U Frey V Angelopoulos S Kurita K Keika K Seki and I Shinohara Pulsating aurora from electron scattering by chorus waves Nature 554:337 EP Feb 2018 24 C F Kennel and H E Petschek Limit on stably trapped particle fluxes Journal of Geophysical Research 71 1 :1 28 1966 25 G V Khazanov and K V Gamayunov Effect of emic wave normal angle distribution on relativistic electron scattering in outer rb NASA Marshall Space Flight Center Huntsville AL United States 2007 26 W Li B Ni R M Thorne J Bortnik Y Nishimura J C Green C A Kletzing W S Kurth G B Hospodarsky H E Spence G D Reeves J B Blake J F Fennell S G Claudepierre and X Gu Quantifying hiss-driven energetic electron precipitation: A detailed conjunction event analysis Geophysical Research Letters 41 4 :1085 1092 2014 2013GL059132 27 Zan Li Robyn M Millan Mary K Hudson Leslie A Woodger David M Smith Yue Chen Reiner Friedel Juan V Rodriguez Mark J Engebretson Jerry Goldstein Joseph F Fennell and Harlan E Spence Investigation of emic wave scattering as the cause for the barrel 17 january 2013 relativistic electron precipitation event: A quantitative comparison of simulation with observations Geophysical Research Letters 41 24 :8722 8729 2014 28 L R Lyons R M Thorne and C F Kennel Electron pitch-angle diffusion driven by oblique whistler-mode turbulence Journal of Plasma Physics 6 3 :589606 1971 29 Lawrence R Lyons Pitch angle and energy diffusion coefficients from resonant interactions with ioncyclotron and whistler waves Journal of Plasma Physics 12 3 :417432 1974 65 30 Lawrence R Lyons and Richard Mansergh Thorne Parasitic pitch angle diffusion of radiation belt particles by ion cyclotron waves Journal of Geophysical Research 77 28 :5608 5616 1972 31 Lawrence R Lyons and Richard Mansergh Thorne Equilibrium structure of radiation belt electrons Journal of Geophysical Research 78 13 :2142 2149 1973 32 Lawrence R Lyons Richard Mansergh Thorne and Charles F Kennel Pitch-angle diffusion of radiation belt electrons within the plasmasphere Journal of Geophysical Research 77 19 :3455 3474 1972 33 Spasojevic M Statistics of auroral hiss and relationship to auroral boundaries and upward current regions Journal of Geophysical Research: Space Physics 2016 34 Nigel P Meredith Richard B Horne Sarah A Glauert and Rodger R Anderson Slot region electron loss timescales due to plasmaspheric hiss and lightning-generated whistlers Journal of Geophysical Research: Space Physics 2007 35 Nigel P Meredith Richard B Horne Sarah A Glauert Richard M Thorne Danny Summers Jay M Albert and Roger R Anderson Energetic outer zone electron loss timescales during low geomagnetic activity Journal of Geophysical Research: Space Physics 111 A5 :n a n a 2006 A05212 36 Nigel P Meredith Richard M Thorne Richard B Horne Danny Summers Brian J Fraser and Roger R Anderson Statistical analysis of relativistic electron energies for cyclotron resonance with emic waves observed on crres Journal of Geophysical Research: Space Physics 108 A6 2003 37 R M Millan R P Lin D M Smith K R Lorentzen and M P McCarthy X-ray observations of mev electron precipitation with a balloon-borne germanium spectrometer Geophysical Research Letters 29 24 :47 1 47 4 2002 38 Y Miyoshi K Sakaguchi K Shiokawa D Evans J Albert M Connors and V Jordanova Precipitation of radiation belt electrons by emic waves observed from ground and space Geophysical Research Letters 35 23 2008 39 Yoshiharu Omura and Qinghua Zhao Relativistic electron microbursts due to nonlinear pitch angle scattering by emic triggered emissions Journal of Geophysical Research: Space Physics 118 8 :5008 5020 2013 40 Carl A Reber The upper atmosphere research satellite uars Geophysical Research Letters 20 12 :1215 1218 1993 66 41 Charles S Roberts Pitch-angle diffusion of electrons in the magnetosphere Reviews of Geophysics 7 1-2 :305 337 1969 42 Craig J Rodger Aaron T Hendry Mark A Clilverd Craig A Kletzing James B Brundell and Geoffrey D Reeves High-resolution in situ observations of electron precipitation-causing emic waves Geophysical Research Letters 42 22 :9633 9641 2015 43 A A Saikin J -C Zhang R C Allen C W Smith L M Kistler H E Spence R B Torbert C A Kletzing and V K Jordanova The occurrence and wave properties of h - he - and o -band emic waves observed by the van allen probes Journal of Geophysical Research: Space Physics 120 9 :7477 7492 2015 44 H J SINGER W P SULLIVAN PETER ANDERSON F MOZER P HARVEY J WYGANT and WILLIAM MCNEIL Fluxgate magnetometer instrument on the crres Journal of Spacecraft and Rockets 29 4 :599 601 Jul 1992 45 David M Smith Eric P Casavant Max D Comess Xinqing Liang Gregory S Bowers Richard S Selesnick Lasse B N Clausen Robyn M Millan and John G Sample The causes of the hardest electron precipitation events seen with sampex Journal of Geophysical Research: Space Physics 121 9 :8600 8613 2016 46 Danny Summers and Richard M Thorne Relativistic electron pitch-angle scattering by electromagnetic ion cyclotron waves during geomagnetic storms Journal of Geophysical Research: Space Physics 108 A4 2003 47 R M Thorne and C F Kennel Relativistic electron precipitation during magnetic storm main phase Journal of Geophysical Research 76 19 :4446 4453 1971 48 R M Thorne W Li B Ni Q Ma J Bortnik D N Baker H E Spence G D Reeves M G Henderson C A Kletzing W S Kurth G B Hospodarsky D Turner and V Angelopoulos Evolution and slow decay of an unusual narrow ring of relativistic electrons near l 3 2 following the september 2012 magnetic storm Geophysical Research Letters 40 14 :3507 3511 2013 49 Andreoli L J Thorne R M Exploration of the Polar Upper Atmosphere NATO Advanced Study Institutes Series Series C Mathematical and Physical Sciences Springer Dordrecht 1980 50 M E Usanova A Drozdov K Orlova I R Mann Y Shprits M T Robertson D L Turner D K Milling A Kale D N Baker S A Thaller G D Reeves H E Spence C Kletzing and J Wygant Effect of emic waves on relativistic and ultrarelativistic electron populations: Ground-based and van allen probes observations Geophysical Research Letters 41 5 :1375 1381 2014 67 51 A L Vampola Electron pitch angle scattering in the outer zone during magnetically disturbed times Journal of Geophysical Research 76 19 :4685 4688 1971 52 A L VAMPOLA J V OSBORN and B M JOHNSON Crres magnetic electron spectrometer afgl-701-5a mea Journal of Spacecraft and Rockets 29 4 :592 595 Jul 1992 53 JAMES A VAN ALLEN and LOUIS A FRANK Radiation around the earth to a radial distance of 107 400 km Nature 183:430 EP Feb 1959 54 J D Winningham J R Sharber R A Frahm J L Burch N Eaker R K Black V A Blevins J P Andrews J Rudzki M J Sablik D L Chenette D W Datlowe E E Gaines W I Imhof R W Nightingale J B Reagan R M Robinson T L Schumaker E G Shelley R R Vondrak H D Voss P F Bythrow B J Anderson T A Potemra L J Zanetti D B Holland M H Rees D Lummerzheim G C Reid R G Roble C R Clauer and P M Banks The uars particle environment monitor Journal of Geophysical Research: Atmospheres 98 D6 :10649 10666 1993 55 Jichun Zhang Alexa J Halford Anthony A Saikin Chia-Lin Huang Harlan E Spence Brian A Larsen Geoffrey D Reeves Robyn M Millan Charles W Smith Roy B Torbert William S Kurth Craig A Kletzing J Bernard Blake Joseph F Fennel and Daniel N Baker Emic waves and associated relativistic electron precipitation on 2526 january 2013 Journal of Geophysical Research: Space Physics 121 11 :11 086 11 100 2016 56 X -J Zhang W Li Q Ma R M Thorne V Angelopoulos J Bortnik L Chen C A Kletzing W S Kurth G B Hospodarsky D N Baker G D Reeves H E Spence J B Blake and J F Fennell Direct evidence for emic wave scattering of relativistic electrons in space Journal of Geophysical Research: Space Physics 121 7 :6620 6631 2016 57 X -J Zhang W Li R M Thorne V Angelopoulos J Bortnik C A Kletzing W S Kurth and G B Hospodarsky Statistical distribution of emic wave spectra: Observations from van allen probes Geophysical Research Letters 43 24 :12 348 12 355 2016 68 Appendix A A 1 Derivation of N for hiss diffusion coefficient In a volume V the total magnetic energy can be expressed in terms of either on frequency or wave vector k domain Z 1 B d 2 3 2 V 0 Z 1 Bk d k 2 2 2 3 Z Z dkk Bk 2 k dk A 1 0 By plugging 3 13 and transforming from k kk to X tan domain the relationship is rewritten as Z 0 2 B 2 d 2 2 R Z 2 B 0 0 k g X k J k kX dX d N A 2 Thus 2 N 2 2 Z g X k J 0 kk k dX X A 3 Jacobian is given by kk kk k J k X Xk k A 4 k X With the help of simplified cold plasma dispersion relation for whistler wave c2 k2 2 2 pe e cos The four terms in the determinant collapses to two terms and Jacobian is rewritten as J kk k 3 pe 3 1 1 2 X 1 X 2 4 X c 2k e e A 5 Finally the normalization factor is given by 1 1 2 pe 3 N 2 2 32 c e Z Xmax Xmin 1 3 Xg X 1 X 2 4 dX 1 2 pe 3 I 2 2 32 c e A 6 69 A 2 Derivation of N Y for EMIC diffusion coefficient The Jacobian from EMIC wave dispersion relation is 2 2 2 pe M XY 2 2M Y pe k kk k k J 2 2 2 X 2 e c c kk M Y A 7 Thus N Y is given by 2 pe M 2Y 2 N Y 2 2 e c2 A 3 Z Xmax Xmin 2 kk X 2M Y pe dXg X 2 c2 kk M Y A 8 Source term estimation using observations inside loss cone At low Earth orbit UARS location each of the three UARS electron telescopes inside loss cone measures differential electron flux Ji i 1 2 3 the unit of Ji is s 1 cm 2 sr 1 keV 1 Assume d sin d d is the solid angle element where is the angle from local magnetic field B and is azimuthal angle around B The total integral electron flux inside loss cone is then determined by Jtotal XZ i i Ji d XZ i Ji sin d d 2 i XZ i Ji sin d A 9 i where flux azimuthal symmetry is assumed the integral range i for each telescope is a ring-shape patch that is created by spinning telescope around local magnetic field and determined by the field of view of the telescope the angle between center of the telescope and local magnetic field B and adjusted for overlaps between two adjacent telescopes and the boundary of loss cone at UARS The unit of Jtotal is s 1 cm 2 keV 1 Jtotal is also given by Z Jtotal Jd A 10 losscone where the integral is over entire loss cone and the differential flux J is related to distribution function by J P 2 fP P P2 fv 0 v m3 A 11 70 where P and m are momentum and mass of electrons fv 0 v is the loss cone solution given by 3 3 and copied here with subscript v omitted S v S v f 0 v h 0 D D D TE I0 0 D TE c I1 c D TE A 12 Source term S v is determined by equating the theoretical flux from distribution function and measurement from UARS A 4 Some notes on diffusion coefficients code implementation and testing The diffusion coefficients formulas for both hiss and EMIC waves in Chapter 3 are implemented in MATLAB Contact me if you are interested in using or testing my code In Lyons et al 1972 32 Gaussian models for hiss wave frequency and normal angle distribution were adopted I first implemented Lyons 72 diffusion coefficient formulas and reproduced Figure 4 of Lyons et al 1972 32 This figure shows how cyclotron and Landau resonance diffusion coefficients varying with equatorial pitch angle with arbitrary unit In order to have a sense of the magnitude of diffusion coefficients the outputs of my code were compared to Figure 1 of Albert 1994 1 Albert s plot is based on the same parameters used by Lyons et al 1972 except with wave amplitude specified My code reproduces Albert s plot The code outputs were also compared to results from several other authors e g Gao et al 2015 14 The agreements were found Once the Gaussian diffusion coefficient code was tested I then implemented the more general version of the diffusion coefficient formula 3 20 and tested against the Gaussian code The two sets of code give exactly the same diffusion coefficients with identical Gaussian wave inputs EMIC wave code has the same structure as hiss wave code except the implementation of dispersion relationship The dispersion relationship of EMIC wave is far more complicated than hiss wave which makes the code more complicated and computationally expensive The dispersion relationship part was tested heavily against results from Albert 2003 3 Figure 1 and 2 in Albert 2003 3 were reproduced and similar plots were obtained when the parameters were tweaked The diffusion coefficients were also compared to Summers 2003 46 My 71 diffusion coefficients are one order of magnitude lower than Summers That is because the bounce averaged diffusion coefficient is one order of magnitude lower than local equatorial values as mentioned in Summers 2003 72 VITA As the first college student in my direct family since my great great-grand father and the first graduate student from my extended family since early 20th century I thank everyone in both my immediate and extended family for great support for so many years I also thank the help from friends I made in elementary school middle school high school college and now both in China and in United States
    • Zheng, Hao - Ph.D. Dissertation
      An Investigation of Lightning-Generated Whistler Waves in the Inner Magnetosphere 2019, Zheng,Hao,Hao Zheng Copyright 2019 Hao Zheng An Investigation of Lightning-Generated Whistler Waves in the Inner Magnetosphere Hao Zheng A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy University of Washington 2019 Reading Committee: Robert Holzworth Chair Michael McCarthy Abram Jacobson Program Authorized to Offer Degree: Earth and Space Sciences University of Washington Abstract An Investigation of Lightning-Generated Whistler Waves in the Inner Magnetosphere Hao Zheng Chair of the Supervisory Committee: Professor Robert Holzworth Earth and Space Sciences In this dissertation we use a new and unique dataset to investigate lightning-generated whistler waves in the Earth s inner magnetosphere Global lightning data and high-resolution waveform data in the inner magnetosphere are combined to study lightning-generated whistler waves including source propagation and potential scattering effects Lightning can produce strong broadband radio waves especially in the very low frequency VLF band from 300 Hz to 30 kHz A fraction of wave energy leaks into the ionosphere during reflections in the Earth-ionosphere waveguide and becomes lightning-generated whistler waves Lightning-generated whistler waves and energetic electrons are important in the inner magnetosphere dynamics The recent launch of the near-equatorial Van Allen Probes provides a great opportunity to observe and study lightning-generated whistler waves During the conjunction work between World Wide Lightning Location Network WWLLN and Van Allen Probes we successfully predict the occurrence of lightning-generated whistler waves near the geomagnetic equator at low L-shells L 3 with a rate of 80% About 22 6% of whistler waves observed by Van Allen Probes correspond to possible source lightning in the WWLLN data which agrees with the detection efficiency of WWLLN About 40 1% additional whistler waves observed by Van Allen Probes may be related with WWLLN lightning if the source region is extended from 2000 km to the global area The far-field radiated energy of lightning may not be the dominant factor for the appearance of lightning-generated whistler waves if it is larger than 100 J By using the highresolution waveform data from Van Allen Probes we can study the possible pitch angle scattering process in both ducted whistler wave and non-ducted whistler wave events Diffusion coefficients calculated from the high-resolution waveform data show that strong lightninggenerated whistler waves may be important for scattering electrons with energies around 100 keV Ducted whistler waves may be more effective in pitch angle scattering than non-ducted whistler waves due to smaller wave normal angles The work in this dissertation shows the possibility to simulate the electron precipitation caused by lighting-generated whistler waves in individual cases TABLE OF CONTENTS List of Figures iv List of Tables vii Chapter 1 Introduction 1 1 1 The Earth s Magnetosphere 1 1 2 The Radiation Belts 2 1 3 The Significance of Radiation Belts Research 3 1 4 Particle Motions 4 1 5 Pitch Angle Scattering 6 1 6 Whistler-Mode Waves in the Inner Magnetosphere 6 1 7 Lightning and Lightning-Generated Whistler Waves 7 1 8 Electron Precipitation Caused by Lighting-Generated Whistler Waves 9 1 9 Summary and Thesis Structure 10 Chapter 2 A Statistical Study of Whistler Waves Observed by Van Allen Probes RBSP and Lightning Detected by WWLLN 21 2 1 Introduction 21 2 2 RBSP and WWLLN 25 2 2 1 Prediction 27 2 2 2 Dechirping 28 2 2 3 Propagation Model 30 2 2 4 One-to-One Coincidence Between WWLLN and RBSP 31 i 2 3 Statistical Results 32 2 4 Discussions 35 2 5 Conclusions 40 Chapter 3 Interactions Between Energetic Electrons and Strong Lightning-Generated Whistler Waves Observed at High L-Shells 50 3 1 Introduction 50 3 2 Database 53 3 2 1 Van Allen Probes 53 3 2 2 WWLLN 54 3 3 Observations 54 3 3 1 Oblique Whistler Event 54 3 3 2 Ducted Whistler Event 59 3 4 Discussions 60 3 5 Conclusions 62 Chapter 4 Summary and Future Work 75 4 1 Conclusions 75 4 2 Future Work Suggestions 76 Appendix A WWLLN Service Unit Test 88 A 1 New Design 88 A 2 Computer Selection 88 A 3 Design and Layout 89 A 4 Construction 90 ii A 5 Test Procedures 90 A 6 Software Setup 91 iii LIST OF FIGURES Figure 1 1 Schematic of the Earth s magnetosphere with principal particle regions Rice Space Institute Server adapted from T W Hill 12 Figure 1 2 Various types of satellite anomalies caused by high levels of charged particles NASA Credit: JHUAPL 13 Figure 1 3 Three basic components of particle motion in the inner magnetosphere NASA 14 Figure 1 4 The relation of mirror height to the equatorial pitch angle and the concept of a losscone angle Figure 2 8 in Bortnik 2004 15 Figure 1 5 Normal first-order cyclotron resonance between electromagnetic circularly polarized waves and charged particles Figure 8 in Tsurutani and Lakhina 1997 16 Figure 1 6 Illustration of a lightning discharge radiating ELF and VLF waves that propagate away from the source in the Earth-ionosphere waveguide and leak into the magnetosphere Figure 1 2 in Bortnik 2004 17 Figure 1 7 Illustration of the lightning-induced precipitation process Stanford VLF Group website 18 Figure 1 8 Observations of sferics top whistlers middle and electron flux bursts bottom in LEP events Figure 1 in Voss et al 1984 19 Figure 1 9 Observations of drift loss cone event top Drift loss cone electrons observed on board SAMPEX and UARS are plotted in spectrogram format on their orbital tracks middle Electron spectrum at UARS bottom electron precipitation over different L-shells Plate 5 in Blake et al 2001 20 Figure 2 1 Prediction and real results for a RBSP-A and b RBSP-B on 17 July 2013 The red lines show the real result and blue lines show the prediction result 42 Figure 2 2 Spectrogram and waveform of Bu component on RBSP-A from 18:27:20 5 UT to 18:27:21 5 UT on 17 July 2013 a Original spectrogram b Dechirped spectrogram c Original waveform d Dechirped waveform e Timing of WWLLN lightning strokes near iv the satellite footpoints The vertical dashed lines represent the WWLLN lightning time observed in first half of the second 43 Figure 2 3 Histogram of corrected dechirped peak time minus WWLLN stroke time in every 2s dechirping window after correction for two propagation terms a Including all WWLLN strokes b Including WWLLN strokes within 10 000 km c Including WWLLN strokes within 2000 km 44 Figure 2 4 a c L shell value distribution of data downloaded by both RBSP-A and B in seconds: a July 2013 b August and September 2013 and c March and April 2014 d Footpoint location of RBSP satellites when whistlers are observed with a one-to-one coincident WWLLN source lightning e Location of WWLLN lightning strokes which are one-to-one coincident with whistlers observed on RBSP satellites Figures 2 4d and 2 4e are all plotted in geomagnetic coordinate system 45 Figure 2 5 Distribution of one-to-one coincident lightning with energy vs arc distance from lightning to satellite footpoint a Scatter plot b Histogram of arc distance from lightning to satellite footpoint c Histogram of lightning energy 46 Figure 2 6 Scatter plot of lightning energy from WWLLN and Poynting flux calculated from EMFISIS at RBSP for one-to-one coincident events Both linear fitting and power law fitting results are shown here as solid and dashed curves fitting formula in text 47 Figure 2 7 A nose whistler example observed by RBSP-A from 07:15:38 5 UT to 07:15:40 5 UT on 16 July 2014 a EFW spectrogram b EMSIFIS spectrogram c Timing of WWLLN lightning strokes near the satellite footpoints The vertical dashed line also represents the WWLLN lightning time 48 Figure 3 1 Overview of wave measurements from RBSP-A on 20141004 from 0900 to 0910 ac and from 0908 to 0909 d-h a Frequency-time spectrogram of magnetic field spectral density b Electric field spectral density c Waveform of Bu component UVW is the satellite spinning coordinate system with W axis as the spin axis d Same as a e Same as c f Whistler wave magnetic spectral density only wave spectra with intensities at least 5 times greater than the background median value are shown g Poynting vector angle h Wave normal angle The horizontal lines in Figure 3 1a and 3 1d indicate 0 1 fce where fce represents the local electron gyrofrequency 64 v Figure 3 2 Wave power spectral density as a function of frequency a and wave normal angle b for oblique whistler event on 20141004 Red blue and black curves indicate the spectrum of all waves hiss wave and lightning-generated whistler waves 65 Figure 3 3 Bounce-averaged pitch angle a and momentum b diffusion coefficients of nonducted whistlers as a function of pitch angle and energy for oblique whistler event on 20141004 66 Figure 3 4 WFR survey mode measurements of two Van Allen Probes at same L range a c Magnetic field spectral density b d Electric field spectral density 67 Figure 3 5 WWLLN lightning map when two Van Allen Probes pass the same L shells The blue and magenta dots represent the footpoints of two Van Allen Probes 68 Figure 3 6 Energetic electron pitch angle distribution for 102 keV black and 132 keV red from Van Allen Probe A solid and B dash when they passed L 3 40 69 Figure 3 7 a L difference b MLT difference c-d L-shell values and MLT of RBSP-A solid and Metop-B dash e Lightning whistlers on RBSP-A f Electron flux from 0 telescope on Metop-B for 40 keV red 130 keV green and 287 keV blue 70 Figure 3 8 Overview of wave measurements from RBSP-A on 20160625 from 0844 to 0845 a Frequency-time spectrogram of magnetic field spectral density b Electric field spectral density c Waveform of Bu d Whistler wave magnetic spectral density e Poynting vector angle f Wave normal angle The horizontal lines in Figure 3 8a indicate 0 5fce and 0 1fce 71 Figure 3 9 Same format as Figure 3 2 for ducted whistler event on 20160625 72 Figure 3 10 Same format as Figure 3 3 for ducted whistler event on 20160625 73 Figure 3 11 Pitch angle distribution of RBSP-A in 2 minutes solid: before dash: after Different colors indicate different energy channels 74 Figure A 1 WWLLN Service Unit Test v2 design 93 Figure A 2 WWLLN Service Unit Test v2 schematic 94 Figure A 3 Schematic for Service Unit box holes 95 Figure A 4 Schematic for Service Unit mounting holes 96 Figure A 5 Overview of WWLLN Service Unit Test 97 vi LIST OF TABLES Table 2 1 Statistical results of whistler waves observed by RBSP and WWLLN lightning 49 vii ACKNOWLEDGEMENTS At first I would like to thank my advisor Robert Holzworth for his amazing guidance and support over the past 6 years I am deeply grateful that he provided countless advices with his board insights in science and in life His help is extremely important for me an international student who is far away from home to dive into the mystery space I will always enjoy our discussions about science career and life I also want to thank my committee members Michael McCarthy Abram Jacobson John Sahr and Mike Wallace for their support over the years Special thanks to Michael McCarthy for his wealth of knowledge and enthusiastic support in both theoretical research and hardware engineering In addition I would like to thank Qianli Ma and Jinxing Li for providing the assistant of simulation work in Chapter 3 and for their contributions in discussing the results My friends in JHN 266 made my graduate life easier than fighting alone I will miss the wonderful journey at UW with Michael Hutchins Brian Burkholder Todd Anderson Paul Sturmer and of course Leo who I know for more than 10 years Finally I would like to thank my family for their long-time support My parents and parentsin-law may never understand my research but they just share their endless love without any hesitate My wife Tianshi Skye has shown incredible patience and understanding for the past 6 years Thank you Tianshi I love you viii DEDICATION To Tianshi ix 1 Chapter 1 INTRODUCTION Magnetosphere is the region around a planet dominated by the planet s magnetic field The shape of the Earth s magnetosphere is the direct result of being blasted by solar wind The Earth s inner magnetosphere is the region with nearly dipolar magnetic field which usually points to the region within 8 RE radius of Earth to the center of the Earth at night side There are two important parts inside the inner magnetosphere which are separated by plasma energies not the locations First is the plasmasphere which has dense and cold plasma The other part is called radiation belts where the energetic particles are located This is a brief introduction of the Earth s inner magnetosphere In the following sections we will specifically explain the magnetosphere the radiation belts basic motions of charged particles important plasma waves and the wave-particle interactions in the inner magnetosphere 1 1 THE EARTH S MAGNETOSPHERE The solar wind plasma cannot easily penetrate the Earth s magnetic field but is mostly deflected around it As explained in Baumjohann and Treumann 1997 the interplanetary magnetic field lines cannot penetrate the Earth s field lines The boundary separating the two different regions is called magnetopause And the cavity generated by the Earth s field has been named magnetosphere At the dayside solar wind compresses the field to about 10 RE while the nightside magnetic field is stretched out into a long magnetotail to 60 RE The structure of the Earth s magnetosphere and some related regions are shown in Figure 1 1 The plasmasphere is a region consisting of dense 10 cm-3 but low energy 1 3 Abel and Thorne 1998a There are three main types of whistler-mode waves in the inner magnetosphere: hiss wave chorus wave and lightning-generated whistler wave Hiss wave which is also called plasmaspheric hiss is frequently observed at a high-density region in the plasmasphere The emissions of hiss are incoherent with a wide frequency band between 50 7 Hz to 1 kHz Li et al 2013 The key feature of hiss waves is the structureless and mixed phase at similar frequency band There are three main generation sources of hiss waves in the inner magnetosphere: local generation by electron injections chorus propagated from the region out of the plasmapause and mixed lightning-generated whistlers Hiss waves may cause the loss of energetic electrons with energies from tens of keV to several MeV in the plasmasphere Ni et al 2013 Unlike hiss waves chorus waves are coherent emissions between 0 1fce and fce Chorus waves are usually observed outside of the plasmapause with two types of structure rising tone and falling tone The rising tones usually has a stronger wave amplitude than falling tones Li et al 2011 Lightning-generated whistler waves whistler waves or whistlers are also coherent emissions The frequency of terrestrial whistlers is 1 30 kHz and in the plasmasphere the lightning-generated whistlers have an upper cutoff at local electron plasma frequency and gyrofrequency The electron whistlers usually have a structure of falling tone which means the high frequency components travel faster than the low frequency components In this work we will focus on lightning-generated whistler waves and their contribution to the pitch angle scattering of energetic electrons in the inner magnetosphere In the recent decades it is shown that some other types of waves may also be important for energetic electron precipitation in the inner magnetosphere like man-made VLF transmissions and electromagnetic ion cyclotron EMIC waves e g Meredith et al 2003a Ma et al 2017 1 7 LIGHTNING AND LIGHTNING-GENERATED WHISTLER WAVES Lightning can produce strong broadband radio waves called sferics which propagate in the Earth-ionosphere waveguide and are detected thousands of kilometers away from their source The strongest waves are in the ELF VLF bands from 300 Hz to 30 kHz During propagation in the Earth-ionosphere waveguide the ELF VLF waves reflect many times between the surface of the 8 Earth and the ionosphere Since the ionosphere is not a perfect conductor a fraction of wave energy leaks into the ionosphere and propagates upward to the top of ionosphere and further to the inner magnetosphere Figure 1 6 illustrates the process with black arrows showing different paths in the Earth-ionosphere waveguide and red arrows showing the upward propagation of lightninggenerated whistler waves Note that Figure 1 6 is just a cartoon plot The direction of lightninggenerated whistler waves won t be vertical upward in the ionosphere until multiple refractions due to the Snell s law It is also shown in this figure that there are multiple entry points for lightninggenerated whistler waves to the ionosphere at reflections Lightning-generated whistler is a right-handed circularly polarized wave The dispersion relation of electromagnetic wave propagating in the cold magnetized plasma can be explained by Appleton-Hartree equation: n2 1 1 2 2 2 1 2 2 4 4 2 1 2 2 2 1 2 1 5 where n is the refractive index is the electron cyclotron frequency is the electron plasma frequency and is the wave frequency For lightning-generated whistler waves in the inner magnetosphere the approximation of 1 is applied on 1 5 and we can get 2 n 1 1 1 2 1 6 Based on 1 6 we can also calculate the group velocity of lightning-generated whistler waves when 2 1 7 9 From the group velocity formula we can find that when frequency is low the high frequency component has a higher group velocity which agrees with the falling tone structure 1 8 ELECTRON PRECIPITATION CAUSED BY LIGHTING-GENERATED WHISTLER WAVES Figure 1 7 illustrates the basic physics processes how lightning-generated whistler waves can cause the precipitation of energetic electrons 1 A lightning discharge produce broadband radiation especially in VLF frequency range 2 A fraction of energy from VLF waves leaks into the inner magnetosphere as whistler waves which propagate along or in some angles with magnetic field to the magnetic equator 3 Whistler waves can effectively interact with energetic electrons in the radiation belts when the resonance conditions meet which leads to the pitch angle scattering of electrons 4 The new mirror point of electrons is too low so the electrons collide with atmospheric particles and precipitate into the atmosphere This phenomenon which is named Lightninginduced Electron Precipitation LEP was firstly discovered using in situ rocket measurements e g Rycroft 1973 Goldburg et al 1985 LEP is a known troposphere-to-magnetosphere coupling mechanism The first satellite measurements of LEP events was reported by Voss et al 1984 As shown in Figure 1 8 multiple electron flux bursts are observed at several hundred kilometers altitude with almost simultaneous existence of whistler waves In the LEP events the particles precipitate during the initial bounce path after the wave-particle interactions so we say that the pitch angle of particles drops into the bounce loss cone Besides the bounce loss cone there is another type of loss cone called drift loss cone Figure 1 9 shows a drift loss cone precipitation event caused by lightning The drift loss cone is different with the bounce loss cone in that the particle doesn t precipitate within one bounce period after the wave-particle interactions 10 but precipitate within one drift period around the Earth The reason for that is the South Atlantic Anomaly SAA The particle in the L-shell range of SAA may be temporarily trapped if the pitch angle is larger than the local bounce loss cone angle but may still precipitate during its eastward drift when its pitch angle is smaller than the bounce loss cone angle at SAA 1 9 SUMMARY AND THESIS STRUCTURE The lightning-generated whistler waves and energetic particles are significant in the inner magnetosphere dynamics It is important to investigate the lightning-generated whistler waves including the lightning source the propagation from troposphere to magnetosphere the waveparticle interactions in the inner magnetosphere and the electron precipitations as a result In previous work there are several limitations of available data so it is hard to statistically study the correlations between global lightning and whistler waves in the inner magnetosphere The limitations usually come from two parts data of lightning and data of whistler waves Lightning data is usually limited by the resolution or coverage For example lightning data may only be available at some local regions or globally but with only a low resolution In this thesis we will use the lightning data from World Wide Lightning Location Network WWLLN which is a longrange network capable of locating global lightning strokes in space and time WWLLN is capable to locate lightning to within a few kilometers and ten microseconds and also provide a far-field energy from the detected strokes in the 6 18 kHz band The other limitation comes from the satellite measurements of whistler waves in the inner magnetosphere In previous research the lightning-generated whistler waves are usually studied at hundreds of kilometer altitude like DEMETER satellite CRRES may be the only satellite specifically designed for radiation belts research in the past There are a lot of studies on long time scale plasma waves based on CRRES 11 data But the data may not be accurate for lightning-generated whistler waves due to low time resolution Lightning-generated whistler waves usually have a time scale within second so it is required to have a high sample rate in the data In this thesis we will use the most recent data from Van Allen Probes RBSP which are designed to understand the dynamics in the Earth s radiation belts The burst mode data obtained with a high sampling rate from Van Allen Probes can be a powerful tool to study detailed changes in the radiation belts By using the data from WWLLN and Van Allen Probes we will study the propagation of lightning-generated whistler waves and the related pitch angle scattering processes The results of this thesis are based on previous studies but working on totally new data sets and they can show a new view or path to understand the dynamics of lightning-generate whistler waves in the inner magnetosphere In Chapter One we introduce background information previous understandings and the reason of our research In Chapter Two we will show the statistical results of one-to-one coincidence between lightning and lightning-generated whistler waves observed in the inner magnetosphere In Chapter Three we will focus on two events to study pitch angle scattering processes related with two types of lightning-generated whistler waves In Chapter Four we will summarize the results in this thesis and also provide some suggestions for future work 12 Figure 1 1 Schematic of the Earth s magnetosphere with principal particle regions Rice Space Institute Server adapted from T W Hill 13 Figure 1 2 Various types of satellite anomalies caused by high levels of charged particles NASA Credit: JHUAPL 14 Figure 1 3 Three basic components of particle motion in the inner magnetosphere NASA 15 Figure 1 4 The relation of mirror height to the equatorial pitch angle and the concept of a loss-cone angle Figure 2 8 in Bortnik 2004 16 Figure 1 5 Normal first-order cyclotron resonance between electromagnetic circularly polarized waves and charged particles Figure 8 in Tsurutani and Lakhina 1997 17 Figure 1 6 Illustration of a lightning discharge radiating ELF and VLF waves that propagate away from the source in the Earth-ionosphere waveguide and leak into the magnetosphere Figure 1 2 in Bortnik 2004 18 Figure 1 7 Illustration of the lightning-induced precipitation process Stanford VLF Group website 19 Figure 1 8 Observations of sferics top whistlers middle and electron flux bursts bottom in LEP events Figure 1 in Voss et al 1984 20 Figure 1 9 Observations of drift loss cone event top Drift loss cone electrons observed on board SAMPEX and UARS are plotted in spectrogram format on their orbital tracks middle Electron spectrum at UARS bottom electron precipitation over different L-shells Plate 5 in Blake et al 2001 21 Chapter 2 A STATISTICAL STUDY OF WHISTLER WAVES OBSERVED BY VAN ALLEN PROBES RBSP AND LIGHTNING DETECTED BY WWLLN This chapter is published as: Zheng H R H Holzworth J B Brundell A R Jacobson J R Wygant G B Hospodarsky F S Mozer and J Bonnell 2016 A statistical study of whistler waves observed by Van Allen Probes RBSP and lightning detected by WWLLN J Geophys Res Space Physics 121 2067 2079 doi:10 1002 2015JA022010 2 1 INTRODUCTION Previous observations of energetic electron flux indicate that the Earth s radiation belts are distributed in two distinct zones separated by a region of depleted flux called the slot e g Horne et al 2003 Figure 2 The structure of the inner zone L 2 tends to be relatively stable in comparison to the outer zone L 3 which is highly dynamic Millan and Thorne 2007 Over the past few decades it has been shown that this difference is primarily related to the source and loss mechanisms that control radiation belt electrons It is believed that in the inner magnetosphere pitch angle scattering including both Coulomb collisions and resonant scattering by whistlermode waves controls the loss of energetic electrons Abel and Thorne 1998a 1998b Coulomb collisions with atmospheric constituents are the dominant loss process for energetic electron E 100 keV inside L 1 3 Walt and MacDonald 1964 Abel and Thorne 1998a Above L 1 3 the long-term energetic electron population is largely controlled by whistler mode waves including plasmaspheric hiss lightning-generated whistlers and man-made transmitter signals The calculations in Abel and Thorne 1998a suggest that all three types of whistler mode waves 22 may play important roles in the loss of energetic electrons but different types of whistler mode waves may interact with different electron energies and be dominant at different L shells Specifically the lightning-generated whistlers which are the concern of this work are suggested to become important at L 2 0 provide the dominant scattering process at L 2 4 and still make a contribution at L 3 2 Abel and Thorne 1998a Recent studies have also shown that whistler mode chorus wave can provide local acceleration of energetic electrons by efficient energy diffusion in the outer radiation belts e g Summers et al 2002 Meredith et al 2003b Li et al 2014 It has also been shown that intense whistler mode chorus emission may cause microburst electron precipitation into the atmosphere Thorne et al 2005 Previous studies have shown that whistlers play an important role in the dynamics of the radiation belts and are partly responsible for the loss of the energetic electrons e g Dungey 1963 Voss et al 1984 1998 Abel and Thorne 1998a 1998b Lauben et al 2001 Rodger et al 2004 Millan and Thorne 2007 Meredith et al 2009 Pitch angle scattering of energetic Van Allen belt electrons by whistlers can result in the precipitation of these electrons into the atmosphere Dungey 1963 Lightning-induced electron precipitation LEP from the Earth s radiation belts caused by whistler wave-particle interaction is a known troposphere-to-magnetosphere coupling mechanism The first satellite measurements of LEP events were obtained as a result of the SEEP experiment on S81-1 satellite Voss et al 1984 By using the same data Voss et al 1998 found that a single LEP burst 10 3 erg s 1 cm 2 in the slot region is estimated to deplete 0 001% of the particles in the region Whistlers can be important for pitch angle diffusion of 100 250 keV electrons in the 2 L 3 range Voss et al 1998 Rodger et al 2004 provide evidence for the relative significance of the electron losses driven by whistler-induced electron precipitation and that caused by VLF transmitters Magnetospherically reflected whistlers generated by lightning 23 are also considered to be a source of plasmaspheric hiss Sonwalkar and Inan 1989 By using ray-tracing simulations it was shown that whistlers produced by a single lightning flash but entering the magnetosphere at different points can form a continuous hiss-like signal at a fixed point Draganov et al 1992 The analysis of DE-1 and IMAGE data showed that the geographic distribution of hiss over the 500 Hz to 3 kHz frequency range is similar to the geographic distribution of lightning strokes Green et al 2005 Similar results are shown in the analysis of CRRES data for the plasmaspheric hiss at higher frequencies f 2 kHz Meredith et al 2006 Plasmaspheric electron densities are an important and fundamental parameter in the dynamics of Earth s radiation belts The propagation of whistlers in the plasmasphere is strongly connected with cold electron density Whistler data from ground-based stations were used to identify the large-scale electron-density irregularities in the plasmasphere Park and Carpenter 1970 and also to present a statistical study of equatorial plasmaspheric electron density and associated flux tube electron content Park et al 1978 Understanding the link between lightning activities and whistler observations on satellites may assist in estimating of plasmaspheric electron densities and testing of magnetospheric models of electron density and energy distribution Liemohn and Scarf 1964 The broadband radio waves produced by lightning discharge called sferics can propagate in the Earth-ionosphere waveguide and be detected thousands of kilometers away from the source A portion of the sferics can penetrate into the ionosphere coupling with the whistler mode in the very low frequency VLF band and travel upward obliquely within tens of degrees along the geomagnetic field line Helliwell 1965 The source and the destination of lightning-generated whistlers have been studied for years Early in situ rocket-borne measurements demonstrated the one-to-one connection between whistlers observed in the upper atmosphere ionosphere and 24 individual lightning strokes from specific thunderstorms within 1000 2000 km e g Holzworth et al 1985 1999 Kelley et al 1990 Li et al 1991 SCATHA satellite VLF data indicated that whistlers are rarely detected near the magnetic equator across the L shells of L 5 5 9 0 Koons 1985 The apparent scarcity of whistlers near geostationary altitude as covered by the SCATHA satellite suggests that there may be few if any propagation paths from the Earth-ionosphere waveguide to the outer regions of magnetosphere But it also may indicate that VLF spectral data are not always investigated with high enough time resolution to easily detect lightning whistlers Gurnett and Inan 1988 examined the data from the DE spacecraft and discovered several lightning whistlers up to 15 kHz or even 25 kHz at L 4 Gurnett and Inan 1988 Figures 21 23 Holzworth et al 1999 used a global three-dimensional two-fluid code to investigate the propagation of whistlers at 0 5 1 and 2 0 kHz into the high-latitude magnetosphere The results show that with southward interplanetary magnetic field the energy of whistlers which start from magnetic latitude above 70 can propagate to near magnetopause or high-altitude magnetosphere In recent decades localized and global ground lightning detection networks have been gradually improving which has proven critical in linking whistlers to their source lightning strokes Santol k et al 2009 analyzed three lightning strokes detected by METEORAGE sferics received by Nan ay station and the corresponding whistlers observed by the DEMETER satellite at 707 km altitude The electric field data and optical flashes measurements on the C NOFS satellite were used with simultaneous global lightning location information from WWLLN in Holzworth et al 2011 to show that whistlers have abundant access to the ionosphere even close to the magnetic equator Both papers showed that the one-to-one coincidence between whistler waves observed at LEO and individual lightning strokes and the penetration into the topside ionosphere occurs at nearly vertical wave vector angles due to the gradient of electron density Fiser et al 2010 used 25 two reference samples to automatically detect the fractional hop whistlers on the DEMETER satellite A local lightning detection network in Europe was used to match the lightning and whistler data They found that the amplitudes of whistlers decrease monotonically with horizontal distance up to 1000 km from the source lightning and the amplitude of whistlers are stronger at nighttime than during the daytime In this paper we will explore the connection between lightning sferics and whistlers by using data from the Van Allen Probes formerly known as the Radiation Belt Storm Probes RBSP and World Wide Lightning Location Network WWLLN The highresolution waveform data obtained near the geomagnetic equator provide a wider area coverage in the inner magnetosphere than LEO satellites like DEMETER Global lightning data including the location time and energy for every individual stroke are simultaneously collected in the conjunction periods One-to-one coincidences between whistlers observed by RBSP and lightning strokes detected by WWLLN are explored in this work 2 2 RBSP AND WWLLN The RBSP satellites were launched in August 2012 into a near-equatorial orbit with 10 inclination apogee altitude of 30 050 31 250 km 5 8 RE from the center of the Earth and perigee altitude of 500 675 km Stratton et al 2013 These constraints place the satellites in orbits that cut through both the inner and outer radiation belts After launch the satellites were officially renamed to the Van Allen Probes The fundamental purpose of the RBSP mission is to provide a better understanding of the processes that drive changes within the Earth s radiation belts The Electric Field and Waves EFW instrument Wygant et al 2013 on the RBSP can provide both 3-D electric field and 3-D magnetic field waveform data The burst mode data we used in this work have a sampling rate of 16 4 ksamples s 1 Another instrument called RBSP-EMFISIS The Electric and Magnetic Field Instrument Suite and Integrated Science Kletzing et al 2013 can 26 also provide 3-D electric and magnetic field waveform data with a sampling rate of 35 ksamples s 1 in 6 s blocks The burst mode of EMFISIS can be trigged automatically or manually WWLLN is a global very low frequency lightning-location system using the time-of-grouparrival TOGA technique Dowden et al 2002 It can detect both cloud-to-ground CG and intracloud IC lightnings but the type of lightning is not distinguished in the data At present WWLLN includes over 70 participating stations This network improves in accuracy and detection efficiency with increased number of stations The number of lightning strokes located increased from 10 6 million to 28 1 million 165% when the number of WWLLN stations increased from 11 in 2003 to 30 in 2007 Rodger et al 2009 As of 2011 the network had an estimated detection efficiency of about 11% for CG lightning over the continental United States and the number can increase to 30% for higher peak current lightning Hutchins et al 2012 Knowledge of individual stroke locations with high temporal and spatial accuracy is very helpful for studying VLF energy radiation and the global electric circuit e g Hutchins et al 2013 2014 This paper presents data collected during two periods of conjunction work from July to September 2013 and from March to April 2014 during which we were able to collect specific burst mode RBSP data The conjunction work of 2013 between WWLLN and RBSP began on 1 July 2013 and ended on 15 September 2013 The following study period started on 15 March 2014 and ended on 30 April 2014 For each day of the two time periods a prediction was computed in advance to determine the best times to collect the broadband wave data Burst mode sampling was conducted during the predicted time for each satellite In addition to the statistical study conducted in this work we also provide some examples from other burst mode sampling intervals when high L shell whistlers were seen 27 2 2 1 Prediction Due to the limit of data storage onboard and in agreement with the RBSP instrument team the burst mode for this lightning study at 16 4 ksamples s on RBSP-EFW was limited to 10 min per day per satellite in 2014 only RBSP-B was available for the data collection Additionally due to data download limitations everyday we had to select 3 or 4 min out of the 10 min stored broadband data which were then downloaded In order to predict the best 10 min d to collect high sampling rate burst mode data on RBSP satellites we traced the satellite magnetic footpoints over a lightning occurrence map The footpoints data were calculated at an altitude of 100 km using the T89c magnetic field model developed by Tsyganenko 1989 and is provided by the SSC 4D Orbit Viewer from NASA http: sscweb gsfc nasa gov tipsod This lightning occurrence map is composed of all global WWLLN lightning data from the same date in the last 3 years 2010 2012 That is we used 3 weeks of data 1week from each of the last 3 years centered at the day of the year for which we want a prediction This map therefore had global lightning data from 21 days We then used the predicted satellite ephemeris with 1 min time resolution to identify the magnetic field footpoints in both Northern and Southern Hemispheres From previous research Santol k et al 2009 Fiser et al 2010 Holzworth et al 2011 we know that the entry point of whistlers into the ionosphere can be thousands of kilometers away from the source lightning In our work a source area of 20 20 box was used in the prediction instead of a circle with 1000 km radius in order to get a faster calculation in the program Although the area of 20 20 box is not the same at different latitudes the predicted peaks will not change significantly because the results are dominated by the variance of lightning occurrence rate at different latitudes In our prediction for every 1 min footpoint location we summed all lightning strokes within the 20 20 box centered at the footpoint This gave us a 1 min resolution prediction of possible strokes at 28 the magnetic footpoint for the whole day Figures 2 1a and 2 1b show the prediction results for RBSP-A and RBSP-B on 17 July 2013 as an example The blue lines represent the prediction result using WWLLN data from 2010 to 2012 and the red lines represent the real lightning number in the same footpoint box using actual WWLLN data on 17 July 2013 The time difference between prediction peak and actual peak is within 5 min for both satellites If this example is true in general it would suggest that if we set a time period with 5 min centered at the predicted peak time there should be lightning strokes detected around the footpoints within the predicted 10 min The conjunction work between WWLLN and RBSP started in July 2013 after the prediction test for the whole month of June was finished For RBSP-A and RBSP-B 93 3% and 96 7% of daily time shifts between actual peak and prediction peak are within 5 min Even on days when the time difference between the actual and predicted peaks is larger than 5 min there are still many lightning strokes detected during the predicted 10 min period After the RBSP burst mode sampling on the satellites was finished we checked the actual WWLLN lightning stroke numbers detected within the 20 20 box for every 1 min footpoint location and selected the best 3 4 min to download everyday 2 2 2 Dechirping Figure 2 2 shows a 1s spectrogram and waveform plot of the RBSP-A observation on 17 July 2013 during 18:27:20 5 18:27:21 5 UT period Figure 2 2a shows the power spectral density PSD of Bu component UVW coordinate system is the spacecraft coordinate system where the W axis is the spin axis and is orthogonal to U and V axis In Figure 2 2a at least five intense whistler events are detected and identified by numbered oblique arrows There are also some weak events shown as unnumbered vertical arrows in Figure 2 2a The original waveform data of Bu is shown in Figure 2 2c Figure 2 2e shows the lightning strokes detected by WWLLN in the 20 20 box centered 29 at the footpoint The lightning strokes detected in the first half second are shown as dashed vertical lines in Figures 2 2a 2 2d From the original data Figures 2 2a and 2 2c it is difficult to determine whether there is an actual one-to-one coincidence between the lightning detected by WWLLN and whistlers observed by RBSP-A or if they are actually unrelated events that happened to be observed simultaneously In the ionosphere at low latitude and midlatitudes the time delay of a whistler approximately varies as where f is the wave frequency and D is the dispersion constant Jacobson et al 2011 developed an automated algorithm called dechirping for recognizing and selecting the signatures of electron whistlers observed on C NOFS satellite Here we employ this method to identify the best dispersion constant between 0 and 400 in each analysis window detailed definition described in Jacobson et al 2011 section 4 1 and equation 1 In the final step we shift different frequency parts of waveform back to the arrival time by using the best fit to for each time interval Figures 2 2b and 2 2d are the dechirped results for the data in Figures 2 2a and 2 2c In Figure 2 2b the whistlers detected in Figure 2 2a have been corrected for the time delay and are no longer dispersed Instead the whistler energy is now sharpened in the vertical red bars between 1 kHz and 8 kHz In order to reduce the contribution of noise below the waveguide cutoff and alias effects around the Nyquist frequency low 1 kHz and high 8 kHz cutoff frequencies were applied before the dechirping process After the dechirping process the vertical dashed lines WWLLN lightning time and vertical red bars dechirped whistler time are located quite close in time The dechirping method enables us to identify even low-intensity whistlers from background noise The two weak signals in Figure 2 2a vertical arrows become much more prominent in Figure 2 2b also vertical arrows The dechirping process also makes it possible to identify overlapping whistlers Using the Number 3 whistler in Figure 2 2a as an example we see it contains two to three whistler traces observed at almost the same time In Figure 30 2 2d the number 3 event clearly resolves as several independent dechirped whistler packets which are shown in the dechirped waveform Considering the leftward migration of signal energy in the dechirping a 50% overlap between the time windows was applied We only compared the lightning strokes and dechirped peaks in the first half second because the dechirped peak near the end of the observation window may not be accurate since only the high-frequency packet of whistler waveform underwent the dechirping process In Figure 2 2 all six lightning strokes in first 0 5 s detected in the 20 20 box centered at the footpoints correspond to whistler signals observed by RBSP-A The correspondence between Number 5 whistler and WWLLN lightning stroke is also good and can be identified in the next time window There is still a small timedifference between the dechirped waveform peak and WWLLN lightning time because we have not yet accounted for the propagation time 2 2 3 Propagation Model WWLLN provides lightning stroke location with better than 15 s temporal accuracy and 10 km spatial accuracy Jacobson et al 2006 We assume that the VLF sferics generated from lightning strokes propagate in the Earth-ionosphere waveguide to the footpoint of satellites with a speed slightly less than c Dowden et al 2002 where c is the speed of light in a vacuum Near the footpoint they couple with the plasma and propagates obliquely along the geomagnetic field to the RBSP satellites according to the oblique whistler dispersion relation This type of propagation process has been used extensively in previous work e g Holzworth et al 1999 Jacobson et al 2011 It is based on the quasi-longitudinal approximation to the Appleton-Hartree dispersion relation Helliwell 1965 To compare the WWLLN lightning stroke time with the dechirped peak time from RBSP we subtract two terms from the dechirped peak time The first is the speed of light propagation time 31 along the geomagnetic field line from the footpoint to the satellites Here we used IGRF-11 model in the Geopack DLM to trace the geomagnetic field and calculated the length of the geomagnetic field line The second term is the propagation time in the Earth-ionosphere waveguide from WWLLN lightning stroke to the footpoint of RBSP satellites After subtracting the two terms on the dechirped peak time the whistler is effectively transported back to the possible candidate of source lightning We call it the corrected dechirped peak time in this paper 2 2 4 One-to-One Coincidence Between WWLLN and RBSP The goal of this paper is to find possible one-to-one coincidences between lightning and whistlers by comparing the WWLLN lightning time and corrected dechirped peak time from RBSP For all the continuous burst mode data we divided Bu component into several analysis windows Each window includes 16 384 samples 1 s and 50% overlapping 8196 samples Every 1 s window was then extended to a 2 s window by adding another 16 384 empty data just before the Bu data in order to leave enough space for the leftward migration of the signal during the dechirping process Every new 2 s window was fed into the dechirping process and all dechirped peaks between 1 s and 1 5 s were saved for later analysis except for the first and last window the dechirped peaks of first and last window were searched from 0 s to 1 5 s and from 1 s to 2 s respectively Once there was at least one dechirped peak found in the analysis window we loaded the WWLLN lightning data detected in the same 2 s analysis window as the source candidates for the dechirped peaks At the same time the footpoint with shorter arc distance spherical distance between one of the footpoint to lightning location to each lightning stroke was selected from two hemispheres and used to correct the dechirped peak The time differences between source lightning candidates and corrected dechirped peaks in the same analysis window were calculated and the histogram results are shown in Figure 2 3 We have already shown that the time difference between 32 WWLLN lightning and dechirped peak is usually a few milliseconds in Figure 2 2 so only time differences between 100 ms and 100 ms are plotted here Figure 2 3a shows a histogram of the time difference between corrected dechirped peak and WWLLN lightning tcorrected tWWLLN in 1ms bin for all stroke locations Figures 2 3b and 2 3c are similar but only include WWLLN stroke locations within 10 000 km and 2000 km away from the nearest footpoint In Figures 2 3a 2 3c a clear peak of match numbers is located between -30 ms and 0 There are several possible reasons for the time difference The first reason is the uncertainty of the dispersion constant Take a whistler observed on 5 August 2013 as an example In the 236th analysis window with a dispersion constant of 220 5 s1 2 the dechirped peak was located at 20:00:58 748679 This peak was located in the 1 5 s to 2 s of 236th window so it was not saved here but would be saved in the 237th window While in the 237th analysis window with a dispersion constant of 208 0 s1 2 the same peak was located at 20:00:58 761069 instead 12 39 ms difference The second reason is the uncertainty of two subtracted propagation terms Waveguide and field line arc lengths are calculated from models which also generated errors Finally the errors may be introduced by the propagation model since it may not work perfectly for all latitudes L shells The maximum absolute value of time difference above the noise level between corrected dechirped peak and its WWLLN source was found to be around 30 ms in Figure 2 3 By using this number an automatic one-to-one coincidence search was applied to the whole data set For every dechirped peak we loaded global WWLLN lightning data corrected the dechirped peak time and then found the one-to-one coincidence with time difference from -30 ms to 0 2 3 STATISTICAL RESULTS During the conjunction work in 2013 we downloaded 192 min over 65 separate days from RBSPA and 192 min over 66 days from RBSP-B In 2014 we downloaded 221 min over 39 days only 33 from RBSP-B In total we collected 605 min of data across 170 distinct days for RBSP-A and RBSP-B In July 2013 no special selection criteria were used so the burst mode period was biased toward the low-latitude region where lightning is most prevalent For the highly eccentric RBSP orbit low satellite altitude always corresponds to low L shell Figure 2 in Rodger et al 2004 showed that the lightning activity may reach to the high L shell region in the summer of the Northern Hemisphere From August 2013 to September 2013 the prediction was changed to focus on the footpoints region with geographic latitude larger than 40 which is poleward of the regions with highest lightning activities Figure 2 4 shows the L shell coverage of RBSP data used in our work As mentioned before the burst mode sampling period is only requested when it includes the peak number of strokes in the daily prediction For July 2013 the best prediction time occurred when the satellites were conjugate to low-latitude regions where lightning is most prevalent Figure 2 4a Thus during the first period July 2013 whistler observations were only made at low L shells Beginning in August 2013 our focus shifted to the region with geographic latitude larger than 40 In August and September 2013 the L shell values of satellite footpoints were therefore significantly higher than in July 2013 Figure 2 4b Most of the data in this period were sampled between L 2 and 3 There are 780 s 13 min of data sampled at L 5 region In March and April 2014 the L shell value coverage returned to L 2 5 due to the seasonal change of global lightning Figure 2 4c Figure 2 4d shows the global coverage of the satellites footpoints when whistlers were observed by RBSP satellites All of the dechirped peaks are observed inside of L 3 The data mainly cover four regions with three small portions located at 100 130 170 220 and 230 250 and a large region located at 310 40 These are approximately corresponding to Europe Africa Asian Oceania and America respectively In Figure 2 4d the longitude coverage is limited in 34 several parts because lightning has a higher occurrence rate on continents than over oceans Figure 2 4e shows the distribution of lightning strokes that are one-to-one coincident with whistlers observed by RBSP satellites Figure 2 4d refers to the L value of the RBSP satellites during whistlers observation while Figure 2 4e refers to the L value of the lightning source location In Figure 2 4e the matched lightning strokes show full coverage of all geomagnetic longitudes and have peak numbers around the same regions of dechirped peaks in Figure 2 4d There are some lightning strokes at high L shell larger than L 3 in Figure 2 4e which are the sources of whistlers observed by RBSP inside of L 3 in the magnetosphere This means the distance between lightning and satellite footpoint may be larger than 1000 2000 km which is consistent with previous research e g Holzworth et al 1999 Table 2 1 shows the statistical results of both RBSP-A and RBSP-B The total of 605 min data are divided into three periods: 151 min in July 2013 233 min in August and September 2013 and 221 min in March and April 2014 Whistlers are observed in 485 min 80 2% of data with 140 min 92 7% 150 min 64 4% and 195 min 88 2% in the three periods respectively A total number of 20 986 whistlers are observed by RBSP satellites The 8308 39 6% of them are oneto-one coincident with WWLLN lightning strokes Specifically 2526 5938 42 5% 730 2454 29 7% and 5052 12 594 40 1% of the observed RBSP whistlers are one-to-one coincident with WWLLN lightning strokes in the three periods In the 485 min 206 470 lightning strokes are detected by WWLLN Four percent of them are found to be one-to-one coincident with whistlers observed on RBSP If we narrow the lightning locations to the area within 2000 km from the footpoints there are 38 777 lightning strokes detected in the 485 min and 15 3% of them are oneto-one coincident with whistlers observed on RBSP 35 2 4 DISCUSSIONS As explained in the Section 2 1 previous studies found that whistlers play a significant role in the dynamics of the radiation belts But the connection between source lightning strokes and subsequent whistlers in the magnetosphere has been difficult to study due to lack of simultaneous high time resolution waveform and global lightning observations In this paper we present a new data set that identifies the source lightning locations for specific whistlers in the inner magnetosphere with the help of WWLLN and RBSP Unlike Fiser et al 2010 the lightning data used in our work are not simply narrowed to the region near the footpoints of satellites Additionally we do not use reference whistlers to automatically detect whistlers since the dispersion factor may change at different regions The final difference between two studies is the propagation of whistlers At low altitude the main propagation process happens in the Earthionosphere waveguide between footpoint and source lightnings While in the inner magnetosphere the whistlers travel a long path along or within some angles of field lines till they are observed by satellites We did not compare the amplitude of whistlers with horizontal distance or the amplitude of whistlers between day and night like what Fiser et al 2010 did in this paper The propagation model used in this work is quite simple and does not include the various propagation mechanisms of whistlers in the magnetosphere Whether whistlers are ducted or non-ducted remains an outstanding question Although lightning is considered the only source for whistlers the observed one-to-one coincident rate between whistlers and lightning is much less than 100% This rate is limited by lightning detection efficiency the strength of whistlers and how well we understand the propagation of whistlers In this work we used the WWLLN lightning location data from 2010 2012 to forecast the lightning activity along the daily trajectory of RBSP footpoints Then we recorded the 10 min of 36 RBSP burst mode data for the time period with peak lightning stroke counts in the 20 20 grid centered at the footpoint After onboard recording we were only able to download 3 4 min of the daily 10 min recording During the entire 605 min of downloaded data whistlers are observed 80 2% of the time we predicted suggesting that satellites should have a high probability of observing whistlers if their footpoints are located within a few thousand kilometers of an active thunderstorm The occurrence of these high lightning activity areas can be predicted by using archival WWLLN data We can use the results of this method to predict the occurrence of lightning-generated whistler waves in the inner magnetosphere and its related phenomena In the 485 min of data 20 986 dechirped peaks are observed by RBSP satellites 39 6% of those peaks are one-to-one coincident with a WWLLN lightning stroke It is obvious that WWLLN does not detect the source lightning for every whistler observed by RBSP satellites It is found that WWLLN only has a 20 40% detection efficiency for strong CG lightning peak current larger than 55 kA Rodger et al 2009 Abarca et al 2010 so the possible source lightning may be missed by WWLLN A test was also undertaken to evaluate the probability that we might match whistlers with some noncorresponding lightning due to the high occurrence rate of global lightning During the 485 min 20 random times for every second were set as the dechirped peaks and fed into the same automatic one-to-one coincidence search with WWLLN lightning The random match rate is around 16 7% After subtracting the random match rate we find that at least about 22 9% of whistlers correspond to a source lightning stroke detected by WWLLN This number is comparable to the detection efficiency of WWLLN for strong lightning About 4 0% of WWLLN lightning strokes correspond to the whistlers observed by the RBSP satellites Figure 2 5a shows the scatter plot of all the coincident WWLLN lightning strokes with energy versus arc distance in the waveguide to the nearest satellite footpoint Figures 2 5b and 2 5c 37 are the histograms of energy and arc distance shown in Figure 2 5a In Figure 2 5b we see that the number of coincident lightning strokes decreases from 4000 with the increase of arc distance below 7000 km Above 7000 km the number of coincident lightning strokes shows another small peak of 200 This phenomenon is also found in the one-to-one coincidence between WWLLN and C NOFS data see Jacobson et al 2011 Figure 16 They explained it as an expected behavior of the spherical-shell effect In Figure 2 5c the energy of coincident lightning strokes shows a peak of 1 10 kJ If we only consider the range from 50 J to 500 kJ the average and median energy of the coincident lightning strokes are about 7 5 kJ and 2 6 kJ respectively while the average and median energy of all detected lightning strokes are 6 5 kJ and 2 4 kJ respectively Hutchins et al 2012 showed that the median energy of WWLLN lightning strokes is around 2 kJ after April 2011 see their Figure 8 which is consistent with our results In Figure 2 5a the majority of the scatter points above 100 J may suggest that any lightning stroke with energy 100 J may generate a whistler strong enough to be detected by the RBSP-EFW Figure 2 5b shows that the lightning strokes detected near the footpoints within 7000 km have a larger potential to generate a whistler propagated to the inner magnetosphere In Table 2 1 the numbers of lightning strokes detected by WWLLN within 2000 km of footpoints both Northern and Southern Hemispheres are shown as 12 699 6103 and 20 005 respectively The coincident rate increased from about 4 0% to 15 3% if the source region was narrowed to within 2000 km of footpoints If the source regions of whistlers are extended beyond 2000 km from the footpoints the coincident whistlers can increase from 5932 to 8308 40 1% Based on the one-to-one coincident match between lightning-generated whistlers and lightning we compared the energy of lighting from WWLLN data and Poynting flux calculated from RBSP data The measurement along spin axis w on EFW is not capable to calculate the 38 Poynting flux so we use the EMFISIS data at the same period when it is available The calculation of Poynting flux density followed the work in Santol k et al 2010 Figure 2 6 showed the scatter plot for each lightning-whistler pair when EMFISIS data is available The Poynting flux is integrated between 1 8 kHz with an FFT bin size of 68 36 Hz 35000 Hz 512 In Figure 2 6 it is shown that there is not a strong linear correlation between lightning energy and whistler Poynting flux Most of the points are located at 103 to 104 J which is near the average energy of WWLLN lightning as we explained before We applied a linear fitting solid on these data and the formula is y 6829 41 20 11 x y is the lighting energy and x is the Poynting flux A power law fitting is also shown in the figure as dashed curve and the formula is lg y 3 36 0 13 lg x It looks like the power law fitting works better than the linear fitting here From the fitting results we can find that the propagation of energy from lightning to whistlers in the inner magnetosphere is still a complicated problem The main reasons are first there can be multiple entry points for lightning energy to leak into the ionosphere during propagation in the waveguide secondly the channel size of each path for whistlers in the magnetosphere is also unknown Sometimes there is a large flux tube duct which can reach to 0 5 RE we have an example in Chapter 3 but it is not necessary for non-ducted whistlers However this work showed a possibility to predict the appearance of lightning-generated whistlers at satellite locations and we may be able to provide a Poynting flux estimate based on the fitting equation We note that in August and September of 2013 the occurrence rate of one-to-one coincidence 29 7% is much lower than in July 2013 42 5% or in March and April 2014 40 1% One reason may be that fewer lightning strokes are detected at high latitude in August and September Although the arc distance from the possible source lightning to the footpoint varies from 0 to 18 000 km the match rate of one-to-one coincidence between lightning and whistlers is still 39 dominated by lightning near the footpoint especially within 2000 km Figures 2 4a 2 4c show that the L shell coverage of the RBSP data in August and September 2013 is larger than in other two periods The occurrence rate of lightning at high latitudes is lower than found at low latitude and midlatitude Hutchins et al 2012 and the number of lightning strokes detected at the footpoints in August and September 2013 is lower than for the other two periods Table 2 1 Another reason may be the failure of finding one-to-one coincidence between lightning and whistlers at high L shells In Figure 2 4d no dechirped peaks are observed above L 3 during 7 min of data sampling between L 3 and 3 5 and 13 min of data sampling above L 5 in Figure 2 4b This could be due to two different situations: 1 no whistlers are observed by RBSP satellites at high L shells 2 whistlers are observed by RBSP but do not pass the dechirping process After carefully checking the data above L 3 it was found that no whistlers were observed when L is larger than 5 Although this is consistent with the work of Koons 1985 we cannot conclusively state that lightning has no impact at L 5 There are three possible reasons whistlers may not be observed above L 5 First the magnetic field around the geomagnetic equator is small at high L shells so the whistlers would have a low upper cutoff frequency and may be masked by lowfrequency noise or even be reflected before they arrived at the satellites Second the amplitude of whistlers may be not strong enough due to the attenuation Burkholder et al 2013 showed that within the ionosphere there is an approximately 3 order of magnitude loss of energy from the footpoint to C NOFS satellites Third the propagation of whistlers at high L shell may not follow the same magnetic field line so the lightning-generated whistler waves near the footpoint may not reach the geomagnetic equator of the same L shell The propagation of magnetospherically reflected whistlers is studied by ray trace method e g Bortnik et al 2002 2003 By reviewing recent EMFISIS data collected during the Northern Hemisphere summer in 2015 we identified 40 several lightning-generated whistler waves with upper cutoff frequencies of 2 kHz observed near the RBSP apogee L 5 797 which may correspond to WWLLN located lightning It is also found that in the 7 min data located between L 3 and 3 5 there were whistlers observed by RBSP but which failed to pass the dechirping process This may be due to the fact that the dispersion is stronger and the whistler waves may no longer be fully displayed in the 1 s window and the dispersion constant may also exceed the number we set At last nose whistlers may be observed when L shell is large Figure 2 7 shows an example of a nose whistler observed by RBSP-A at L 3 88 on 16 July 2014 Lightning near the footpoint is detected at 07:15:38 64 Figure 2 7c and the dashed line in Figure 2 7 About 0 9 s later both EFW and EMFISIS captured the nose whistler with a nose frequency of about 4 5 kHz and an upper cutoff frequency of about 8 kHz As shown in Figures 2 7a and 2 7b EMFISIS provides more information at high frequency 8 12 kHz The dechirping method used in our work has a band-pass filter of 1 8 kHz which is no longer appropriate for nose whistlers because the nose frequency is usually lower than 6 kHz or even 4 kHz in the outer magnetosphere Helliwell 1965 Figures 4 19 With a better dechirping process in the future we can probably get more information about the contribution of lightning to the high L shell whistlers in the magnetosphere 2 5 CONCLUSIONS A study of simultaneous observations of global lightning and whistlers was conducted from July to September 2013 and from March to April 2014 Global lightning data of the past 3 years from the WWLLN were used to forecast lightning conditions along the trajectory of RBSP s magnetic field footpoints Ten minutes with the highest lightning probability around the footpoints were selected for burst mode recording on the satellites Data were downloaded for short time periods during 170 days leading to a total of 605 min of high-resolution waveform which were statistically 41 analyzed in this paper By using this method lightning-generated whistlers near the magnetic equator at low L shell regions can be successfully predicted with a rate of 80 2% This new data set should prove valuable for the future study of whistler-related phenomena About 22 6% of the whistlers observed by the satellites correspond to possible source lightning in the actual WWLLN data which closely matched the time and location This rate also agrees with the detection efficiency of WWLLN The source regions of whistlers are extended 2000 km from the footpoints in this study About 40 1% more whistlers observed by the RBSP satellites are found to correspond with WWLLN lightning Lightning strokes with energy larger than 100 J all have the potential to generate a whistler and propagate to the inner magnetosphere We show that whistlers strongly correspond to WWLLN lightning at low L shell L 3 regions The correspondence between high L shell whistlers and lightning still exist but need further study Acknowledgments The authors wish to acknowledge partial support for this research from CRDF grant RUG1-7084-PA-13 from NOAA grant NA10OAR4320148 and from NSF grant 1443011 The authors wish to thank the World Wide Lightning Location Network http: wwlln net a collaboration among over 50 universities and institutions for providing the lightning location data used in this paper The authors wish to acknowledge the Space Sciences Laboratory at University of Berkeley and the University of Minnesota for providing EFW data data link: http: www space umn edu rbspefw-data PI: John Wygant and the University of Iowa for providing the EMFISIS data data link: http: emfisis physics uiowa edu data index PI: Craig Kletzing both of which are instruments on board the Van Allen Probes 42 Figure 2 1 Prediction and real results for a RBSP-A and b RBSP-B on 17 July 2013 The red lines show the real result and blue lines show the prediction result 43 Figure 2 2 Spectrogram and waveform of Bu component on RBSP-A from 18:27:20 5 UT to 18:27:21 5 UT on 17 July 2013 a Original spectrogram b Dechirped spectrogram c Original waveform d Dechirped waveform e Timing of WWLLN lightning strokes near the satellite footpoints The vertical dashed lines represent the WWLLN lightning time observed in first half of the second 44 Figure 2 3 Histogram of corrected dechirped peak time minus WWLLN stroke time in every 2s dechirping window after correction for two propagation terms a Including all WWLLN strokes b Including WWLLN strokes within 10 000 km c Including WWLLN strokes within 2000 km 45 Figure 2 4 a c L shell value distribution of data downloaded by both RBSP-A and B in seconds: a July 2013 b August and September 2013 and c March and April 2014 d Footpoint location of RBSP satellites when whistlers are observed with a one-to-one coincident WWLLN source lightning e Location of WWLLN lightning strokes which are one-to-one coincident with whistlers observed on RBSP satellites Figures 2 4d and 2 4e are all plotted in geomagnetic coordinate system 46 Energy 106 105 104 103 102 101 100 0 6 12 18 12 18 Arc distance 1000 km Stroke number 10000 1000 100 10 0 6 Arc distance 1000 km Energy 106 105 104 103 102 101 100 100 101 102 103 Stroke number 104 105 106 Figure 2 5 Distribution of one-to-one coincident lightning with energy vs arc distance from lightning to satellite footpoint a Scatter plot b Histogram of arc distance from lightning to satellite footpoint c Histogram of lightning energy 47 Figure 2 6 Scatter plot of lightning energy from WWLLN and Poynting flux calculated from EMFISIS at RBSP for one-to-one coincident events Both linear fitting and power law fitting results are shown here as solid and dashed curves fitting formula in text 48 Figure 2 7 A nose whistler example observed by RBSP-A from 07:15:38 5 UT to 07:15:40 5 UT on 16 July 2014 a EFW spectrogram b EMSIFIS spectrogram c Timing of WWLLN lightning strokes near the satellite footpoints The vertical dashed line also represents the WWLLN lightning time 49 Table 2 1 Statistical results of whistler waves observed by RBSP and WWLLN lightning Number of sampling minute Number of sampling minute with dechirped peaks Percentage Number of dechirped peaks Number of coincident dechirped peaks Percentage Number of global WWLLN lightning strokes Number of coincident WWLLN lightning strokes Percentage Number of WWLLN lightning strokes within 2000 km Number of coincident WWLLN lightning strokes within 2000 km Percentage July 2013 151 Aug Sep 2013 233 Mar Apr 2014 221 Total 605 140 150 195 485 92 7% 5938 64 4% 2454 88 2% 12 594 80 2% 20 986 2526 730 5052 8308 42 5% 29 7% 40 1% 39 6% 54 675 62 520 89 275 206 470 2526 730 5052 8308 4 6% 1 2% 5 7% 4 0% 12 699 6103 20 005 38 777 2004 452 3476 5932 15 8% 7 4% 17 4% 15 3% 50 Chapter 3 INTERACTIONS BETWEEN ENERGETIC ELECTRONS AND STRONG LIGHTNING-GENERATED WHISTLER WAVES OBSERVED AT HIGH LSHELLS 3 1 INTRODUCTION In the past few decades it has been shown that there are three main loss mechanisms of radiation belt electrons: losses across the magnetopause precipitation into the upper atmosphere and outward radial diffusion The losses across the magnetopause and outward radial diffusion have been used to explain the rapid depletion of relativistic electrons in the outer radiation belt down to L 5 e g Bortnik et al 2006a Turner et al 2012 Electron precipitation into the upper atmosphere is mainly controlled by pitch angle scattering process including Coulomb collisions and resonant wave-particle interactions between different types of plasma waves and energetic electrons e g Abel and Thorne 1998a 1998b Pitch angle scattering outside the plasmasphere is mainly due to whistler-mode chorus waves Horne and Thorne 2003 Thorne et al 2005 Inside the plasmasphere the loss of electrons is mainly due to Coulomb collisions lightning-generated whistlers plasmaspheric hiss and manmade VLF transmitter signals Based on the calculations in Abel and Thorne 1998a three types of VLF waves may interact with energetic electrons of different energies and are dominant at different L-shells By using the average wave intensity lightning whistlers are suggested to be important drivers between 2 0 L 3 2 but much weaker than plasmaspheric hiss at L 2 4 Bortnik et al 2003 qualitatively estimated the energy distribution and lifetimes of magetospherically reflected MR whistlers in the plasmasphere and indicated that these MR whistlers may play a more important role than assumed in previous work at slot region Rodger et 51 al 2003 suggested that long-term loss driven by whistlers is more significant than all other loss processes for energetic electrons in the range of 50 150 keV at L 2 2 4 Based on the average wave densities from CRRES measurements Meredith et al 2007 estimated the loss timescales due to plasmaspheric hiss and lightning-generated whistlers in and above the slot region The results showed that the loss of electrons with energies larger than 500 keV in the region 2 5 L 4 0 is dominated by plasmaspheric hiss with small wave normal angles At lower energies ducted whistlers are more effective as a scattering process than plasmaspheric hiss but still with a much longer loss timescale By adding the observations from SAMPEX satellites and comparing the new results with their previous work Meredith et al 2009 found that plasmaspheric hiss alone can t account for the observed loss timescales at 2 0 L 2 4 The observed loss required the combined effects of plasmaspheric hiss and ducted lightning-generated whistlers Although the power of whistlers used in their calculations is two orders of magnitude less than hiss it is shown that whistlers can still play an important role in electron loss at the inner slot region In the past decade CRRES data were used in different models to estimate the scattering loss due to plasmaspheric hiss and lightning-generated whistlers e g Meredith et al 2007 Glauert et al 2014 It is believed that lightning-generated whistlers only have an effect on 500 keV electrons due to the frequency range 2 0 f 5 0 kHz and have weaker effects than plasmaspheric hiss due to the lower average wave amplitude one or two orders of magnitude less Lightning-induced electron precipitation LEP from the Earth s radiation belts is a known troposphere-to-magnetosphere coupling mechanism caused by wave-particle interactions Voss et al 1984 showed the first LEP event from the observations of SEEP experiment on S81-1 satellite It was shown that a single LEP burst 10-3 erg s-1 cm-2 is estimated to deplete 0 001% of the particles in the slot region Voss et al 1998 The LEP events proved that lightning whistlers 52 may be important for pitch angle scattering of 100 250 keV electrons in the slot region Simultaneous observations of LEP events and lightning-generated whistlers have been observed on the DEMETER satellite with apogee 700 km altitude Inan et al 2007 The LEP events are observed within 4 near the magnetic equator Zheng et al 2016 Due to the limited data sampling rate or the limited space coverage on satellites long-timeaveraged wave emissions over some specific frequency ranges were used to estimate the effect of lightning-generated whistlers However in our recent work we found that the wave amplitude of lightning-generated whistlers in specific events can be much stronger than the average wave intensities used in previous work In this paper we would like to show you two strong lightninggenerated whistlers events observed at L 3 near the geomagnetic equator with peak amplitude of waveform perturbation larger than 0 1 nT We will present direct evidence of strong lightninggenerated whistlers scattering energetic electrons in the outer radiation belt 3 2 3 2 1 DATABASE Van Allen Probes The Van Allen Probes formerly known as the Radiation Belt Storm Probes RBSPs were launched in August 2012 into a near-equatorial orbit with 10 inclination The two probes traverse roughly the same orbit with 1 1 RE 5 8 RE from the center of the Earth which cut through both the inner and outer radiation belts Stratton et al 2013 The Electric and Magnetic Field Instrument Suite and Integrated Science EMFISIS wave instrument on board provides 3-D measurements of wave electric and magnetic fields and DC magnetic fields Kletzing et al 2013 The six-channel Waveform Receiver WFR on EMFISIS has a continuous coverage of wave power spectral density in the survey mode from 10 Hz to 12 kHz and a variable coverage of full waveform data in the burst mode with a sampling rate up to 35 ksamples s The Magnetic Electron 54 Ion Spectrometer MagEIS provides electron distributions with energies from 30 keV to 4 MeV at a time cadence of 11 s Blake et al 2013 3 2 2 WWLLN WWLLN detects lightning by monitoring the Time-Of-Group-Arrival TOGA of VLF lightning sferics at five or more stations simultaneously located globally Dowden et al 2002 This network has improved in accuracy and detection efficiency with the addition of many new stations since 2004 In 2011 the network had an estimated detection efficiency of about 11% for Cloud-Ground strokes and can be 50% for stokes with higher peak current Hutchins et al 2012 Zheng et al 2016 previously showed that the conjunction between Van Allen Probes and WWLLN is helpful to find the source lightning of whistlers observed around magnetic equator especially at low Lshells 3 3 3 3 1 OBSERVATIONS Oblique Whistler Event Figure 3 1 shows an overview of lightning-generated whistler measurements from RBSP-A on 20141004 from 0900 to 0910UT In this event we had rare continuous burst mode data from EMFISIS for about 20 minutes only the last 10 minutes data are shown here During this event Van Allen Probes were close to the magnetic equator MLAT magnetic latitude 5 therefore we directly used the local measurements to represent the magnetic equatorial parameters The L value of RBSP-A varied from 3 69 to 3 08 in 20 minutes and MLT was around 8 0 As illustrated in Figure 3 1a and 3 1b it is still difficult to identify the dispersed whistler pattern in magnetic electric wave power spectra with such a condensed plot Every individual lightninggenerated whistler wave appeared as spikes in the frequency range between 0 4kHz There are 55 also two narrow band of waves at 1kHz and 300 Hz which may be the plasmaspheric hiss waves Unlike the previous studies in the introduction the wave amplitudes of whistlers in this event are very strong The magnetic perturbation was even larger than 0 2 nT at 090528UT and 090812UT with multiple peaks higher than 0 1 nT Figure 3 1c In figure 3 1d 3 1h the 1minute plot from 0908 to 0909UT could provide more details about this event The spectrogram in Figure 3 1d is very similar to the first MR non-ducted whistler spectrogram from OGO 1 which usually contains a series of echo train with decreasing nose frequency Smith et al 1968 In this event each group of whistler waves and three possible echoes became more dispersed after multiple reflections It also agrees with the ray tracing simulation results at L 3 3 5 in Bortnik et al 2003 In order to select the frequency spectrum of whistler waves and get rid of background only PSD higher than 5 times of its median value at each frequency bin are kept in Figure 3 1f From the Poynting vector angle shown in Figure 3 1g it is clearly seen that whistler echoes passed the satellite point from different directions The wave normal angle calculated by SVD method Santol k et al 2003 of non-ducted whistler waves was already very large when the first wave arrived at satellite and became even closer to 90 after several more reflections Figure 3 1h As shown in Figure 3 1 strong non-ducted whistler waves were observed by Van Allen Probes and detailed necessary wave parameters were provided to model the scattering effects Here we used the EMFISIS measurements from 0908 to 0909 UT in Figure 3 1f as the average wave frequency spectrum The electron density and background magnetic field amplitude were obtained from in situ measurements producing a high ratio of electron plasma frequency to gyrofrequency Since the wave normal angle of non-ducted whistlers increased rapidly after first reflection Bortnik et al 2006b Meredith et al 2007 we assumed that the wave normal angle 56 remained highly oblique at high L-shells The input wave parameters we obtained from RBSP-A are shown in Figure 3 2 Specifically the red blue and black curves in figure 3 2a indicate the averaged power spectral density of all waves Figure 3 1d hiss wave median value of Figure 3 1d and only whistlers Figure 3 1f In this event the total wave spectrum can also be divided into three frequency ranges: mostly hiss waves below 500 Hz a mixture of hiss waves and lightning-generated whistler waves from 500 Hz to 1 1 kHz and mostly lightning-generated whistler waves from 1 1 kHz to 4 kHz Unlike previous work which only calculated the contribution of lightning-generated whistlers from 2 kHz the event shows that we can extend the lower cutoff frequency to 500 Hz The wave spectral densities shown here are slightly different from average results at quiet times Figure 2 in Meredith et al 2007 with stronger whistler intensity and weaker hiss intensity in this event The wave normal angle spectrum of whistlers Figure 3 1h is also shown in Figure 3 2b The majority of the waves have large wave normal angles around 65 and we applied a Gaussian fitting for the wave normal angle distribution in the following model To quantify the scattering effects of this strong non-ducted whistler event we calculated the bounce-averaged diffusion coefficients by using Full Diffusion Code Ni et al 2008 2011 Shprits & Ni 2009 at different pitch angles and energies Cyclotron resonances with harmonic numbers ranging from -10 to 10 including Landau resonance were included in the calculations The bounce-averaged pitch angle diffusion coefficients are shown as a function of pitch angle PA or and electron kinetic energy Ek in Figure 3 3 The dominant scattering effect of lightning-generated whistlers is pitch angle scattering with a much weaker effect of momentum diffusion There are some interesting features related with this event of strong non-ducted whistler scattering There are two parts of large diffusion coefficients in Figure 3 3a In the near 57 perpendicular direction PA 70 90 the electrons of 10 to 100 keV may be efficiently scattered due to Landau resonance For the resonant electron energies from 30 keV to 100 keV the diffusion coefficients remain large from near loss cone to intermediate pitch angles like 30 There is also a rise to high energy around several hundred keV of around pitch angle of 20 The maximum value of can reach to 10-5 s-1 which is almost the same level as hiss waves in previous work Ni et al 2013 To examine the evidence of strong non-ducted whistler scattering measurements of RBSP-B were also checked since the two Van Allen Probes were in a very similar orbit In this event two probes passed the same L shell at similar MLT with RBSP-B trailing by 86 minutes Different geomagnetic indices on 04 October 2014 were also checked The minimum Dst was -8 nT the minimum SYM-H was -15 nT and Kp index was 1 Under such a quiet geomagnetic condition we assumed that the structure of the plasmasphere didn t change too much within 86 minutes Therefore the difference of measurements on RBSP-B and RBSP-A could be a temporal effect more than a spatial effect The survey mode data of EMFISIS measurements from RBSP-A and RBSP-B when passing the same L-shells were shown in Figure 3 4 Although the resolution of the survey mode data is not as high as burst mode data in Figure 3 1 we can still find multiple spikes as the indication of non-ducted whistlers on both satellites It is shown that the non-ducted whistlers still exist at the L-shells after 86 minutes when RBSP-B arrived Meanwhile the lightning data from WWLLN were also checked during these two periods In Figure 3 5a and 3 5b the location of thunderstorms near the satellite footpoints doesn t change significantly since the thunderstorms normally last a time scale of hours The lightning source of the strong non-ducted whistlers may originate from one of the thunderstorms in the North Atlantic Ocean based on the longitude of satellite footpoints and the ray tracing results at L 3 5 in Bortnik et al 2003 Due 58 to the rotation of the Earth in 86 minutes the footpoints of RBSP-B were westward of the footpoints of RBSP-A when passing the same L-shell From previous research Holzworth et al 2011 Zheng et al 2016 it was shown that the entry point of the whistler waves into the ionosphere can be many thousands of kilometers away from the source lightings The highest amplitudes of whistlers may be observed up to distances of 500 km to 1000 km Fiser et al 2010 Jacobson et al 2011 Therefore the whistlers generated from one thunderstorm may have the ability to illuminate a longitude range of 20 to 50 depends on the latitude In Figure 3 5 it is shown that the footpoints of two Van Allen Probes are located at two sides of an intense line of thunderstorms in the North Atlantic both within 10 of these storms After a short time of rotation it is still possible that the non-ducted whistlers observed on both Van Allen Probes were generated from the lightning strokes of the same thunderstorm area which propagated in the Earthionosphere waveguide and illuminated the same area in the inner magnetosphere The pitch angle distribution from both satellites at L 3 40 are shown in Figure 3 6 The variation in 86 minutes is not large as expected But the pitch angle distribution of 100 keV electrons still shows a weak diffusion process which agrees with diffusion coefficients The electrons around 80 have a small peak which corresponds with the Landau resonance at a perpendicular direction which is also known as a top-hat distribution Meanwhile the electron phase space density PSD slightly decreases at intermediate pitch angles and then increases at lower pitch angles near the loss cone We also examined the lightning-induced electron precipitation LEP by using measurements of POES MetOp satellites Evans and Green 2004 MetOp-B is close to the region of whistlers for less than 1 minute Figure 3 7 presents the data of RBSP-A and MetOp-B during the conjunction period Here we require the L-shell value difference between two satellites to be less than 0 1 which result in a 40 second conjunction window 59 During this period the MLT difference is no larger than 0 65 which is also reasonable The medium energy proton and electron detector MEPED on POES can provide integral electron flux in four energy bands 40 keV 130 keV 287 keV and 612 keV In this 40-second window there are five precipitation peaks observed at 07:05 07:09 07:17 07:23 and 07:35 07:39 The time resolution of MEPED data is 2 seconds so sometimes it is hard to make a one-to-one match between lightning whistlers and flux peaks Unlike the LEP events at low-L shell the peak flux is only four times of background here This is probably due to the sensitivity of POES satellites and the low trapped flux at around 100 keV near the outer radiation belt boundary In some of the peaks the flux of 287 keV is higher than 130 keV that s caused by the geometry factors at different energy channels The original count rate data not shown are checked to have reasonable numbers at different energy channels 3 3 2 Ducted Whistler Event Figure 3 8 shows an overview of ducted whistler measurements from RBSP-A on 20160625 from 0844 to 0845UT In this event RBSP-A was also close to the magnetic equator around L 3 96 The maximum magnetic perturbation of the whistlers was larger than 0 4 nT at 08:44:28 5UT The frequency range of the ducted whistlers was much higher than non-ducted whistlers due to the higher plasma density not shown here The electron densities which inferred from the upper hybrid resonance frequency fuh on EMFISIS Kurth et al 2015 showed a density increase from 08:42 to 08:59 UT indicating a duct here From previous studies the ducted whistlers usually propagate along the magnetic field lines and can be received also reflected at the conjugate point of the other hemisphere Helliwell 1965 The Poynting vector angle showed that the initial whistlers propagated from northern hemisphere to the satellite Two-hop whistler echoes were also observed coming from southern hemisphere As a ducted whistler event the wave normal angle in 60 this event stayed at a small number after one reflection 10 of lightning strokes from WWLLN dataset were found within 10 away from the northern magnetic footpoint of the satellite Each of them was about 1 second before the arrival of whistlers on RBSP-A Here we also used the wave parameters provided on RBSP-A to model the scattering effects of ducted whistlers The average wave frequency spectrum electron density and background magnetic field amplitude obtained from RBSP-A is shown in Figure 3 9a The two peaks in the wave frequency spectrum correspond to different frequency ranges of the initial whistlers and whistler echoes We assumed that the wave normal angle didn t change during reflections in the duct The bounce-averaged diffusion coefficients are shown in Figure 3 10 The results are similar to plasmaspheric hiss but the scattering energy is much lower due to its higher frequency Ma et al 2015 The strong ducted whistlers may cause the energetic electron loss mostly at several to tens of keV with an extension to a hundred keV In this event the other Van Allen Probe was far away and it didn t pass the same L shell at a similar MLT in a short time period We also didn t find any LEO satellites passing through the same L-shells in 1 minute Figure 3 11 shows the electron distribution variation in 2 minutes on RBSP-A The L-shell changed from 3 91 to 3 97 The electron loss at large pitch angles occurred at energies lower than 143 keV At some energies lower than 108 keV the electron PSD increased at low pitch angles near loss cone This agrees with our diffusion coefficient results but it can also be a spatial effect since RBSP-A moved 400 km 0 06 RE 3 4 DISCUSSIONS In this paper we present two strong events of lightning-generated whistlers from Van Allen Probes near the geomagnetic equator of outer radiation belt By using full waveform data from EMFISIS we conducted a detailed study of lightning whistler events around geomagnetic equator Lightning 61 whistlers at L 3 can be divided into non-ducted and ducted types by checking the wave normal angles During the non-ducted whistler event wave and particle measurements from two probes show a trace of pitch angle scattering In this event the whistlers generated from the same thunderstorm may be observed by both probes passing the same L shell within 86 minutes of separation The pitch angle distribution around one hundred keV on both probes showed a possible top-hat near 90 and an increase near the loss cone region which is a sign of the diffusion process Only the wave components with strong power spectral density which is at least 5 times higher than background were selected as whistler waves in our study Possible hiss waves were removed by this method due to its continuous existence at similar strength The wave spectral intensity of whistlers in this event is at least one order higher than the average intensity calculated in previous research Using the wave properties of non-ducted whistlers observed on RBSP-A we simulated the whistler wave scattering effects by modeling the evolution of the electron distribution Bounce-averaged diffusion coefficients were calculated first based on the averaged wave frequency spectrum electron density background magnetic field and L-shell location Lowaltitude MetOp satellite which has passed similar L-shells and MLT the same time provides the chance for the conjunction observations Several electron precipitation events were observed related with lightning-generated whistlers The ratio of peak flux over background is not as large as LEP events at low L-shells e g L 2 3 in Voss et al 1984 This is probably caused by the sensitivity of telescope design and low trapped electron flux near the outer radiation belt boundary Another ducted whistler event is also studied in this paper A duct with increased electron density at a region of 0 4 RE is observed on RBSP-A Multiple lightning-generated whistlers are detected with quasi-parallel propagation in a 1-minute window The source lightning of some whistlers can be found in the WWLLN lightning map near the satellite footpoint In this event the 62 wave spectral intensity of whistlers is at least 20 times higher than the average intensity found before The bounce-averaged diffusion coefficients showed similar results as hiss waves but working mostly in a lower energy range Two probes comparison could be made in this event But the variation of electron PSD on RBSP-A in 2 minutes agreed with the simulations Although the simulation results showed that the strong ducted whistlers could deplete the electrons below 100 keV in about 1 2 hours we would also like to point out that the duration of electron duct is a complicated question and it may not exist long enough and distributed at all MLT 3 5 CONCLUSIONS We conducted a detailed study of two lightning-generated whistler wave events at L 3 region Our results are as follows: 1 The wave intensity of lightning-generated whistlers can be much higher than the averaged results found in previous work Specifically the non-ducted whistler event in our work is at least one order higher and the ducted whistlers is more than 20 times higher 2 The strong lightning-generated whistlers can be important for scattering electrons with energies near the 100 keV level 3 Lightning-generated whistler waves may be underestimated for pitch-angle scattering effects at L 3 regions Acknowledgments The authors wish to acknowledge support from NSF grant 1443011 The authors wish to thank the World Wide Lightning Location Network http: wwlln net a collaboration among over 50 universities and institutions for providing the lightning location data used in this paper The authors wish to acknowledge University of Iowa for providing the EMFISIS data data link: http: emfisis physics uiowa edu data index and LANL for providing MagEIS 63 data data link: https: www rbsp-ect lanl gov data_pub both of which are instruments on board the Van Allen Probes 64 Figure 3 1 Overview of wave measurements from RBSP-A on 20141004 from 0900 to 0910 a-c and from 0908 to 0909 d-h a Frequency-time spectrogram of magnetic field spectral density b Electric field spectral density c Waveform of Bu component UVW is the satellite spinning coordinate system with W axis as the spin axis d Same as a e Same as c f Whistler wave magnetic spectral density only wave spectra with intensities at least 5 times greater than the background median value are shown g Poynting vector angle h Wave normal angle The horizontal lines in Figure 3 1a and 3 1d indicate 0 1 fce where fce represents the local electron gyrofrequency 65 Figure 3 2 Wave power spectral density as a function of frequency a and wave normal angle b for oblique whistler event on 20141004 Red blue and black curves indicate the spectrum of all waves hiss wave and lightning-generated whistler waves 66 Figure 3 3 Bounce-averaged pitch angle a and momentum b diffusion coefficients of nonducted whistlers as a function of pitch angle and energy for oblique whistler event on 20141004 67 Figure 3 4 WFR survey mode measurements of two Van Allen Probes at same L range a c Magnetic field spectral density b d Electric field spectral density 68 Figure 3 5 WWLLN lightning map when two Van Allen Probes pass the same L shells The blue and magenta dots represent the footpoints of two Van Allen Probes 69 Figure 3 6 Energetic electron pitch angle distribution for 102 keV black and 132 keV red from Van Allen Probe A solid and B dash when they passed L 3 40 70 Figure 3 7 a L difference b MLT difference c-d L-shell values and MLT of RBSP-A solid and Metop-B dash e Lightning whistlers on RBSP-A f Electron flux from 0 telescope on Metop-B for 40 keV red 130 keV green and 287 keV blue 71 Figure 3 8 Overview of wave measurements from RBSP-A on 20160625 from 0844 to 0845 a Frequency-time spectrogram of magnetic field spectral density b Electric field spectral density c Waveform of Bu d Whistler wave magnetic spectral density e Poynting vector angle f Wave normal angle The horizontal lines in Figure 3 8a indicate 0 5fce and 0 1fce 72 Figure 3 9 Same format as Figure 3 2 for ducted whistler event on 20160625 73 Figure 3 10 Same format as Figure 3 3 for ducted whistler event on 20160625 74 Figure 3 11 Pitch angle distribution of RBSP-A in 2 minutes solid: before dash: after Different colors indicate different energy channels 75 Chapter 4 SUMMARY AND FUTURE WORK In the following chapter I briefly summarize the key findings in this dissertation and provide suggestions for future work 4 1 CONCLUSIONS In this dissertation we used a new and unique dataset to study lightning-generated whistler waves in the Earth s inner magnetosphere Global lightning data and high-resolution waveform data in the radiation belts are connected together to study the lightning-generated whistler waves including the source propagation wave-particle interactions and electron precipitation The conjunction work between WWLLN and Van Allen Probes is conducted from July to September 2013 and from March to April 2014 which provided the simultaneous observations of global lightning and whistler waves used in this thesis Global lightning data of the past 3 years from WWLLN are used to forecast the lightning occurrence along the magnetic footpoints trajectories of the Van Allen Probes In each day a ten-minute window is selected for data collecting in burst mode at the satellites By using this method lightning-generated whistler waves near the geomagnetic equator at low L-shells were successfully predicted with a rate of 80% About 22 6% of the whistlers observed by the Van Allen Probes corresponds to possible source lightning in the WWLLN data both in time and position This rate agrees with the detection efficiency of WWLLN About 40 1% more whistlers observed by Van Allen Probes may correspond with WWLLN lightning if the source region is extended from 2000 km to the global area The energy of lightning may not be the dominant factor to decide the appearance or absence of the lightning-generated whistler waves in the inner magnetosphere if the far-field radiated 76 energy of lightning is larger than 100 J The new data set based on our method can be valuable for the future study of whistler-related phenomena In previous research the pitch angle scattering of energetic electron by lightning-generated whistler waves is estimated by using average wave strength and hypothetical wave normal angle distribution The results may not be accurate for specific cases In this thesis we show detailed analyses for two types of whistler events using new data from Van Allen Probes The detailed waveform and wave normal angle data of lightning whistlers only are used for calculating the diffusion coefficients The two events in this thesis both showed stronger wave intensities than the averaged results in the previous work Specifically the ducted whistler event with small wave normal angles is larger than the oblique whistler event with large wave normal angles The diffusion coefficients show that the strong lightning-generated whistler waves may be important for scattering electrons with energies around 100 keV Ducted whistler waves may be more effective in pitch angle scattering processes than non-ducted whistler waves due to the small wave normal angles But from another perspective the propagation of ducted whistler requires an electron density gradient and it may not exist long enough The work in this thesis shows the possibility to simulate the electron precipitation caused by lightning-generated whistler waves in individual cases 4 2 FUTURE WORK SUGGESTIONS The work in this thesis shows several important results with the new data from global lightning and high-time resolution waveform data in the radiation belts But there are still several studies can be done to extend the understanding in the future Ray tracing method is very useful to simulate the propagation of lightning-generated whistler waves from ionosphere to the inner magnetosphere But currently there is not a complete model to 77 explain the propagation from lightning to the leak points at the ionosphere The most common hypothesis used for this process is that any point within 2000 km or more from the lightning can be a possible entry point for lightning-generated whistlers In the future study some simulation works can be done to generate a whistler wave strength map based on the distance to the lightning location After that we can combine this map with the ray tracing simulations to get a 3-D whistler wave intensity map generated from each individual lighting stroke Background plasma density from existing models or real measurements can be used as input for ray propagation model There are some preliminary results at 700 - 800 km e g Jacobson et al 2016 but it is necessary to extend the model to a high altitude to the inner magnetosphere In this thesis we only show two whistler events at high L-shells from a list with several hundred whistler events As mentioned in Chapter Two the dechirping method no longer works well for the whistler detection at high L-shells We try to use two methods to identify the whistler waves at high L-shells The first thing we tried is the dechirping method with a high frequency cutoff around 0 3 or 0 4 fce The second thing we used is the cross-matching method in the previous work We manually generated several reference whistler waves with different dispersion constants as candidates and try to match them with wave spectrograms to see if any one fits better than a set threshold But neither method can solve the problem perfectly In the future work we may build a convolution neural network to identify the whistler wave patterns in the spectrograms There is already a neural network trained for DEMETER satellite to identify the whistler phenomena at low altitude The parameters of this neural network are not open and it only works for well-structed 0 whistler waves at low altitude In the future work a new neural network model can be trained if there are enough whistler training samples at high L-shells 78 After we have a better solution on automatic whistler wave detection at high L-shells we can statistically analyze all whistler wave events to get a better understanding on the whistler wave intensity distribution It is also possible to get more electron precipitation cases with the conjunction work of low altitude satellites like new CubeSats with high resolution electron flux data and the Van Allen Probes 79 References Abarca S F K L Corbosiero and T J Galarneau 2010 An evaluation of the Worldwide Lightning Location Network WWLLN using the National Lightning Detection Network NLDN as ground truth J Geophys Res 115 D18206 doi:10 1029 2009JD013411 Abel B and R M Thorne 1998a Electron scattering loss in Earth s inner magnetosphere: 1 Dominant physical processes J Geophys Res 103 A2 2385 2396 doi:10 1029 97JA02919 Abel B and R M Thorne 1998b Electron scattering loss in Earth s inner magnetosphere: 2 Sensitivity to model parameters J Geophys Res 103 A2 2397 2407 doi:10 1029 97JA02920 Baumjohann W and R A Treumann 1997 Basic Space Plasma Physics World Scientific Press Blake J B Inan U S Walt M Bell T F Bortnik J Chenette D L and Christian H J 2001 Lightning-induced energetic electron flux enhancements in the drift loss cone J Geophys Res 106 A12 29733 29744 doi:10 1029 2001JA000067 Blake J B et al 2013 The Magnetic Electron Ion Spectrometer MagEIS Instruments Aboard the Radiation Belt Storm Probes RBSP Spacecraft Space Sci Rev 179 1-4 doi:10 1007 S11214-013-9991-8 Bortnik J U S Inan and T F Bell 2002 L dependence of energetic electron precipitation driven by magnetospherically reflecting whistler waves J Geophys Res 107 A8 1150 doi:10 1029 2001JA000303 Bortnik J U S Inan and T F Bell 2003 Frequency-time spectra of magnetospherically reflecting whistlers in the plasmasphere J Geophys Res 108 A1 1030 doi:10 1029 2002JA009387 Bortnik J 2004 Precipitation of radiation belt electrons by lightning-generated magnetospherically reflecting whistler waves Stanford University Press Bortnik J Thorne R M O Brien T P Green J C Strangeway R J Shprits Y Y and Baker D N 2006a Observation of two distinct rapid loss mechanisms during the 20 November 2003 radiation belt dropout event J Geophys Res 111 A12216 doi:10 1029 2006JA011802 80 Bortnik J Inan U S and Bell T F 2006b Temporal signatures of radiation belt electron precipitation induced by lightning-generated MR whistler waves: 1 Methodology J Geophys Res 111 A02204 doi:10 1029 2005JA011182 Burkholder B S M L Hutchins M P McCarthy R F Pfaff and R H Holzworth 2013 Attenuation of lightning-produced sferics in the Earth-ionosphere waveguide and lowlatitude ionosphere J Geophys Res Space Physics 118 3692 3699 doi:10 1002 jgra 50351 Carpenter D L 1963 Whistler evidence of a knee in the magnetospheric ionization density profile J Geophys Res 68 6 1675-1682 doi:10 1029 JZ068i006p01675 Dowden R L 1971 VLF discrete emissions deduced from Helliwell s theory J Geophys Res 76 13 3046 3054 doi:10 1029 JA076i013p03046 Dowden R L J B Brundell and C J Rodger 2002 VLF lightning location by time of group arrival TOGA at multiple sites J Atmos Sol Terr Phys 64 7 817 830 doi:10 1016 S1364-6826 02 00085-8 Draganov A B U S Inan V S Sonwalkar and T F Bell 1992 Magnetospherically reflected whistlers as a source of plasmaspheric hiss Geophys Res Lett 19 3 233 236 doi:10 1029 91GL03167 Dungey J 1963 Loss of Van Allen electrons due to whistlers Planet Space Sci 11 6 591 595 doi:10 1016 0032-0633 63 90166-1 Evans D S and M S Greer 2004 Polar Orbiting Environmental Satellite Space Environment Monitor-2: Instrument descriptions and archive data documentation NOAA Tech Mem 93 version 1 4 Space Weather Predict Cent Boulder Colo Fiser J J Chum G Diendorfer M Parrot and O Santolik 2010 Whistler intensities above thunderstorms Ann Geophys 28 1 37 46 doi:10 5194 angeo-28-37-2010 Glauert S A Horne R B and Meredith N P 2014 Three-dimensional electron radiation belt simulations using the BAS Radiation Belt Model with new diffusion models for chorus plasmaspheric hiss and lightning-generated whistlers J Geophys Res Space Physics 119 268 289 doi:10 1002 2013JA019281 Goldberg R A J R Barcus L C Hale and S A Curtis 1986 Direct observation of magnetospheric electron precipitation stimulated by lightning J Atmos Terr Phys 48 3 293 299 81 Green J L S Boardsen L Garcia W W L Taylor S F Fung and B W Reinisch 2005 On the origin of whistler mode radiation in the plasmasphere J Geophys Res 110 A03201 doi:10 1029 2004JA010495 Green J L S Boardsen L Garcia W W Taylor S F Fung and B W Reinisch 2006 Reply to comment on On the origin of whistler mode radiation in the plasmasphere by Green et al J Geophys Res 111 A09211 doi:10 1029 2006JA011622 Gurnett D A and U S Inan 1988 Plasma wave observations with the dynamics explorer 1 spacecraft Rev Geophys 26 2 285 316 doi:10 1029 RG026i002p00285 Helliwell R A 1965 Whistlers and Related Ionospheric Phenomena Stanford Univ Press Stanford Calif Holzworth R H M C Kelley C L Siefring L C Hale and J D Mitchell 1985 Electrical measurements in the atmosphere and the ionosphere over an active thunderstorm: 2 Direct current electric fields and conductivity J Geophys Res 90 A10 9824 9830 doi:10 1029 JA090iA10p09824 Holzworth R H R M Winglee B H Barnum Y Li and M C Kelley 1999 Lightning whistler waves in the high-latitude magnetosphere J Geophys Res 104 A8 17 369 17 378 doi:10 1029 1999JA900160 Holzworth R H M P McCarthy R F Pfaff A R Jacobson W L Willcockson and D E Rowland 2011 Lightning-generated whistler waves observed by probes on the Communication Navigation Outage Forecast System satellite at low latitudes J Geophys Res 116 A06306 doi:10 1029 2010JA016198 Horne R B N P Meredith R M Thorne D Heynderickx R H A Iles and R R Anderson 2003 Evolution of energetic electron pitch angle distributions during storm time electron acceleration to megaelectronvolt energies J Geophys Res 108 A1 1016 doi:10 1029 2001JA009165 Hutchins M L R H Holzworth J B Brundell and C J Rodger 2012 Relative detection efficiency of the World Wide Lightning Location Network Radio Sci 47 RS6005 doi:10 1029 2012RS005049 Hutchins M L R H Holzworth K S Virts J M Wallace and S Heckman 2013 Radiated VLF energy differences of land and oceanic lightning Geophys Res Lett 40 2390 2394 doi:10 1002 grl 50406 82 Hutchins M L R H Holzworth and J B Brundell 2014 Diurnal variation of the global electric circuit from clustered thunderstorms J Geophys Res Space Physics 119 620 629 doi:10 1002 2013JA019593 Inan U S Piddyachiy D Peter W B Sauvaud J A and Parrot M 2007 DEMETER satellite observations of lightning induced electron precipitation Geophys Res Lett 34 L07103 doi:10 1029 2006GL029238 Jacobson A R R Holzworth J Harlin R Dowden and E Lay 2006 Performance assessment of the World Wide Lightning Location Network WWLLN using the Los Alamos Sferic Array LASA as ground truth J Atmos Oceanic Technol 23 8 1082 1092 doi:10 1175 JTECH1902 1 Jacobson A R R H Holzworth R F Pfaff and M P McCarthy 2011 Study of oblique whistlers in the low-latitude ionosphere jointly with the C NOFS satellite and the WorldWide Lightning Location Network Ann Geophys 29 5 851 863 doi:10 5194 angeo-29851-2011 Jacobson A R Holzworth R H Pfaff R and Heelis R 2016 Automated identification of discrete lightning-generated multiple-dispersed whistler waves in C NOFS-VEFI very low frequency observations Radio Sci 51 1547 1569 doi:10 1002 2016RS005989 Kelley M C Siefring C L Pfaff R F Kintner P M Larsen M Green R Holzworth R H Hale L C Mitchell J D and Le Vine D 1985 Electrical measurements in the atmosphere and the ionosphere over an active thunderstorm: 1 Campaign overview and initial ionospheric results J Geophys Res 90 A10 9815 9823 doi:10 1029 JA090iA10p09815 Kelley M C J G Ding and R H Holzworth 1990 Intense ionospheric electric and magnetic field pulses generated by lightning Geophys Res Lett 17 12 2221 2224 doi:10 1029 GL017i012p02221 Kletzing C et al 2013 The Electric and Magnetic Field Instrument Suite and Integrated Science EMFISIS on RBSP Space Sci Rev 179 1 4 127 181 doi:10 1007 s11214013-9993-6 Koons H C 1985 Whistlers and whistler-stimulated emissions in the outer magnetosphere J Geophys Res 90 A9 8547 8551 doi:10 1029 JA090iA09p08547 83 Kurth W S S De Pascuale J B Faden C A Kletzing G B Hospodarsky S Thaller and J R Wygant 2015 Electron densities inferred from plasma wave spectra obtained by the Waves instrument on Van Allen Probes J Geophys Res Space Physics 120 904 914 doi:10 1002 2014JA020857 Lauben D S U S Inan and T F Bell 2001 Precipitation of radiation belt electrons induced by obliquely propagating lightning-generated whistlers J Geophys Res 106 A12 29 745 29 770 doi:10 1029 1999JA000155 Li W Bortnik J Thorne R M and Angelopoulos V 2011 Global distribution of wave amplitudes and wave normal angles of chorus waves using THEMIS wave observations J Geophys Res 116 A12205 doi:10 1029 2011JA017035 Li W et al 2013 An unusual enhancement of low-frequency plasmaspheric hiss in the outer plasmasphere associated with substorm-injected electrons Geophys Res Lett 40 37983803 doi:10 1002 grl 50787 Li W et al 2014 Radiation belt electron acceleration by chorus waves during the 17 March 2013 storm J Geophys Res Space Physics 119 4681 4693 doi:10 1002 2014JA019945 Li Y Q R H Holzworth H Hu M McCarthy R D Massey P M Kintner J V Rodrigues U S Inan and W C Armstrong 1991 Anomalous optical events detected by rocketborne sensor in the WIPP campaign J Geophys Res 96 A2 1315 1326 doi:10 1029 90JA01727 Liemohn H B and F L Scarf 1964 Whistler determination of electron energy and density distributions in the magnetosphere J Geophys Res 69 5 883 904 doi:10 1029 JZ069i005p00883 Ma Q Li W Thorne R M Ni B Kletzing C A Kurth W S Hospodarsky G B Reeves G D Henderson M G Spence H E Baker D N Blake J B Fennell J F Claudepierre S G and Angelopoulos V 2015 Modeling inward diffusion and slow decay of energetic electrons in the Earth s outer radiation belt Geophys Res Lett 42: 987 995 doi: 10 1002 2014GL062977 Ma Q Mourenas D Li W Artemyev A and Thorne R M 2017 VLF waves from ground based transmitters observed by the Van Allen Probes: Statistical model and effects on plasmaspheric electrons Geophys Res Lett 44 6483 6491 doi:10 1002 2017GL073885 84 Meredith N P Thorne R M Horne R B Summers D Fraser B J and Anderson R R 2003a Statistical analysis of relativistic electron energies for cyclotron resonance with EMIC waves observed on CRRES J Geophys Res 108 1250 doi:10 1029 2002JA009700 A6 Meredith N P R B Horne R M Thorne and R R Anderson 2003b Favored regions for chorus-driven electron acceleration to relativistic energies in the Earth s outer radiation belt Geophys Res Lett 30 16 1871 doi:10 1029 2003GL017698 Meredith N P R B Horne M A Clilverd D Horsfall R M Thorne and R R Anderson 2006 Origins of plasmaspheric hiss J Geophys Res 111 A09217 doi:10 1029 2006JA011707 Meredith N P Horne R B Glauert S A and Anderson R R 2007 Slot region electron loss timescales due to plasmaspheric hiss and lightning-generated whistlers J Geophys Res 112 A08214 doi:10 1029 2007JA012413 Meredith N P R B Horne S A Glauert D N Baker S G Kanekal and J M Albert 2009 Relativistic electron loss timescales in the slot region J Geophys Res 114 A03222 doi:10 1029 2008JA013889 Meredith N P Horne R B Isles J D and Green J C 2016 Extreme energetic electron fluxes in low Earth orbit: Analysis of POES E 30 E 100 and E 300 keV electrons Space Weather 14 136 150 doi:10 1002 2015SW001348 Millan R and R Thorne 2007 Review of radiation belt relativistic electron losses J Atmos Sol Terr Phys 69 3 362 377 doi:10 1016 j jastp 2006 06 019 Ni B R M Thorne Y Y Shprits and J Bortnik 2008 Resonant scattering of plasma sheet electrons by whistler-mode chorus: Contribution to diffuse auroral precipitation Geophys Res Lett 35 L11106 doi:10 1029 2008GL034032 Ni B Thorne R M Meredith N P Shprits Y Y and Horne R B 2011 Diffuse auroral scattering by whistler mode chorus waves: Dependence on wave normal angle distribution J Geophys Res 116 A10207 doi:10 1029 2011JA016517 Ni B Bortnik J Thorne R M Ma Q and Chen L 2013 Resonant scattering and resultant pitch angle evolution of relativistic electrons by plasmaspheric hiss J Geophys Res Space Physics 118 7740 7751 doi:10 1002 2013JA019260 85 Park C G and D L Carpenter 1970 Whistler evidence of large-scale electron-density irregularities in the plasmasphere J Geophys Res 75 19 3825 3836 doi:10 1029 JA075i019p03825 Park C G D L Carpenter and D B Wiggin 1978 Electron density in the plasmasphere: Whistler data on solar cycle annual and diurnal variations J Geophys Res 83 A7 3137 3144 doi:10 1029 JA083iA07p03137 Rodger C J Clilverd M A and McCormick R J 2003 Significance of lightning-generated whistlers to inner radiation belt electron lifetimes J Geophys Res 108 1462 doi:10 1029 2003JA009906 A12 Rodger C J R J McCormick and M A Clilverd 2004 Testing the importance of precipitation loss mechanisms in the inner radiation belt Geophys Res Lett 31 L10803 doi:10 1029 2004GL019501 Rodger C J J B Brundell R H Holzworth and E H Lay 2009 Growing detection efficiency of the World Wide Lightning Location Network AIP Conf Proc 1118 1 15 20 doi:10 1063 1 3137706 Rycroft M J 1973 Enhanced energetic electron intensities at 100 km altitude and a whistler propagating through the plasmasphere Planet Space Sci 21 241 251 doi:10 1016 00320633 73 90009-3 Santol k O Parrot M and Lefeuvre F 2003 Singular value decomposition methods for wave propagation analysis Radio Sci 38 1010 doi:10 1029 2000RS002523 1 Santol k O M Parrot U Inan D Bure ov D Gurnett and Chum 2009 Propagation of unducted whistlers from their source lightning: A case study J Geophys Res 114 A03212 doi:10 1029 2008JA013776 Santol k O Pickett J S Gurnett D A Menietti J D Tsurutani B T and Verkhoglyadova O 2010 Survey of Poynting flux of whistler mode chorus in the outer zone J Geophys Res 115 A00F13 doi:10 1029 2009JA014925 Shprits Y Y and B Ni 2009 Dependence of the quasi-linear scattering rates on the wave normal distribution of chorus waves J Geophys Res 114 A11205 doi:10 1029 2009JA014223 86 Smith R L and Angerami J J 1968 Magnetospheric properties deduced from OGO 1 observations of ducted and nonducted whistlers J Geophys Res 73 1 1 20 doi:10 1029 JA073i001p00001 Sonwalkar V S and U S Inan 1989 Lightning as an embryonic source of VLF hiss J Geophys Res 94 A6 6986 6994 doi:10 1029 JA094iA06p06986 Stratton J R Harvey and G Heyler 2013 Mission overview for the Radiation Belt Storm Probes mission Space Sci Rev 179 1 4 29 57 doi:10 1007 s11214-012-9933-x Summers D C Ma N P Meredith R B Horne R M Thorne D Heynderickx and R R Anderson 2002 Model of the energization of outer-zone electrons by whistler-mode chorus during the October 9 1990 geomagnetic storm Geophys Res Lett 29 24 2174 doi:10 1029 2002GL016039 Thorne R M T P O Brien Y Y Shprits D Summers and R B Horne 2005 Timescale for MeV electron microburst loss during geomagnetic storms J Geophys Res 110 A09202 doi:10 1029 2004JA010882 Thorne R M R B Horne and N P Meredith 2006 Comment on Onthe origin of whistler mode radiation in the plasmasphere by Green et al J Geophys Res 111 A09210 doi:10 1029 2005JA011477 Tsurutani B T and Lakhina G S 1997 Some basic concepts of wave-particle interactions in collisionless plasmas Rev Geophys 35 4 491 501 doi:10 1029 97RG02200 Tsyganenko N 1989 A magnetospheric magnetic field model with a warped tail current sheet Planet Space Sci 37 1 5 20 doi:10 1016 0032-0633 89 90066-4 Turner D L Y Y Shprits M Hartinger and V Angelopoulos 2012 Explaining sudden losses of relativistic electrons during geomagnetic storms Nature Phys 8 doi:10 1038 NPHYS2185 Van Allen J A and L A Frank 1959 Radiation around the earth to a radial distance of 107 400 km Nature 183 430-434 doi:10 1038 183430a0 Voss H et al 1984 Lightning-induced electron precipitation Nature 312 740 742 doi:10 1038 312740a0 Voss H D M Walt W L Imhof J Mobilia and U S Inan 1998 Satellite observations of lightning-induced electron precipitation J Geophys Res 103 A6 11 725 11 744 doi:10 1029 97JA02878 87 Walt M and W M MacDonald 1964 The influence of the Earth s atmosphere on geomagnetically trapped particles Rev Geophys 2 4 543 577 doi:10 1029 RG002i004p00543 Walt M 1994 Introduction to geomagnetically trapped radiation Cambridge University Press Wygant J et al 2013 The electric field and waves instruments on the radiation belt storm probes mission Space Sci Rev 179 1 4 183 220 doi:10 1007 s11214-013-0013-7 Zheng H Holzworth R H Brundell J B Jacobson A R Wygant J R Hospodarsky G B Mozer F S and Bonnell J 2016 A statistical study of whistler waves observed by Van Allen Probes RBSP and lightning detected by WWLLN J Geophys Res Space Physics 121 2067 2079 doi:10 1002 2015JA022010 88 Appendix A WWLLN SERVICE UNIT TEST A 1 NEW DESIGN The WWLLN service unit was redesigned due to the audio problems in the Gumstix operating system with new version of kernels The stereo signals may randomly swap the left and right channels which is a big problem for the analysis of VLF data There are some other changes in the design of service unit In the new design the Gumstix is removed from the service unit and we separate the computer and service unit again like we did in SUv3 By doing that the new design is more compatible with different computers and we can easily switch to another one if the current choice is discontinued or there is a major system failure which can t be fixed at our end easily All the connectors related to the Gumstix are all removed including the HDMI Ethernet and USB ports We can no longer remotely control the preamp power supply but we revised the design of yellow LEDs so it can directly monitor the currents feeding into the preamp The new design is also compatible with both two types of GPS units using in SUv3 and SUv4 The two voltage regulators used in SUv4 are all replaced by DC DC convertors to guarantee enough power for the circuits The components of new design are all located on one side now instead of two sides in the previous design A 2 COMPUTER SELECTION The Raspberry Pi 3 Model B was chosen to be the current computer instead of a regular desktop used with SUv3 There are several advantages of that 1 it is light and small 2 it is not expensive 3 there is a large community using it so the support is much better than Gumstix There are some basic requirements which should be satisfied for the future upgrades like Internet access Ethernet connection preferred Linux Operating System USB connections and video outputs There are 89 some other soft requirements like sound card RAM size MicroSD card options cost and service support The new version of Raspberry Pi was chosen for its popularity and support The operating system of Raspberry Pi called Raspbian is a Debian-based computer operating system The large open source community can provide an environment with flexibility security and robust But since the Raspberry Pi doesn t come with a sound card we need to find a USB sound card which can analyze 48 kHz stereo signals The USB sound card we chose here is Behringer U-Control UCA222 It is powered by USB port and have stereo input choice with RCA connectors A 3 DESIGN AND LAYOUT The board design is shown in Figure A 1 with component name and values The connections between service unit and Raspberry Pi is not shown in this figure This a minor problem which can be fixed in the future design It is designed to be compatible with two types of GPS units But the output voltage of two types of GPS units are different The voltage at pin 3 of socket 1 is 3 3V but the voltage at pin 3 of socket 2 is 5V The voltage difference will not destroy the circuit but may generate different brightness of LED A1 when it is triggered So in the future design it may be better to put one resistor for each TXD output before the conjunction point instead of using only one R6 The layout is shown in Figure A 2 There are four connectors in the front side left of Figure A 2 including GPS serial jack GPS unit connector stereo jack and preamp jack There is only one power jack at the back side right of Figure A 2 We kept the two-way GPS serial communication between computer and service unit by using the FT232 board The new version of FT232 now uses micro-USB jack instead of mini-USB jack The connector position of two GPS units are aligned in the horizontal directions but the heights are different so before we drill the 90 service unit box we still need to know which type of GPS unit will be put inside of it The stereo jack is directly connected to the USB sound card through a stereo-to-RCA cable and the USB sound card is connected to Raspberry Pi A 4 CONSTRUCTION The construction work for the new service unit also includes two parts: board and box At first we assemble the board with all the necessary components socket and jumpers Then we do several tests see Section A 5 to make sure the board is working correctly After that we start the construction of box In Figure C 3 and C 4 we have the schematics for box holes and mounting holes on each side After the drill work we mount the service unit board into box plug in all the cables as shown in Figure C 5 and do the final test in the Raspberry Pi system A 5 TEST PROCEDURES In the initial board test we don t mount the GPS unit or plug in cables When the input voltage slowly increases from 0 to 12V the current should reach about 45 mA Test the voltage for GPS and preamp The SMT360 socket should have two voltages 5V on pin 1 and 3 3V on pin 2 The NLC-SKII-CP2_V2 0 socket should only have one voltage 5V on pin2 All other pins in the GPS part should be zero We also need to test the voltage for preamp The voltage between pin 5 and pin 1 should be 15V and the voltage between pin 4 and pin 1 should be -15V Turn off the board and mount the GPS unit At 12 V input voltage the board should draw about 73 mA With all the cables connected including the GPS and preamp antenna the board should draw about 117 mA at 12 V input voltage Both yellow LEDs should be on Since the GPS unit is not configured so far we will talk about the red LEDs later 91 A 6 SOFTWARE SETUP The Raspberry Pi system usually boots from a MicroSD card so we need to build a operating system in a MicroSD card first Here are the steps to build the system: 1 Download the official Raspbian image from the official website The current version is Raspbian Stretch 2 Find right tool and write the image to the MicroSD card 3 Mount the MicroSD card to any Linux machine then copy the directory of firstRunFiles to home pi firstRunFiles under the rootfs directory in the MicroSD card The firstRunFiles directory includes the WWLLN software and all necessary configuration files 4 Unmount the MicroSD card safely and plug it into the Raspberry Pi 5 Connect the full system as shown in Figure C 5 boot the Raspberry Pi 6 Run install sh and setup sh under the firstRunFiles directory to install required packages create accounts and finish the configurations 7 After reboot make other necessary configurations in the GUI of raspi_config like allow ssh access don t use pi as default log in account etc 8 Configure the GPS unit Run the sendTSIP_ py script and follow the instructions There is another way to configure the GPS We can directly connect the GPS serial to a Windows PC and configure it by Trimble software After the GPS configuration both two red LEDs should flash once per second 9 Check the GPS status by running readTSIP_ py script 10 Check VLF data We need to check the vlf png file under home sferix public_html directory This file should provide the latest spectrogram of VLF The sferics log file 92 under home sferix sferics can provide more detailed information about the sferics results 11 Change the default webpage in the web browser to the vlf png file so the host can easily visit it The SSH tunnel settings are still under test so the network configurations are not listed here The Raspberry Pi will run with DHCP at this moment It is easy to change it to a static IP if the host has one At first open the file at etc dhcpcd conf with vi or nano Then scroll all the way to the bottom of the file and add one of the following snippets Here is an example: interface eth0 static ip_address 192 168 0 10 24 static routers 192 168 0 1 static domain_name_servers 192 168 0 1 Here interface defines which network interface you are setting the configuration for ip_address is the IP address that you want to set your device to the 24 is the subnet mask which indicates 255 255 255 0 in this example routers is the IP address of your gateway probably the IP address or your router and domain_name_servers is the IP address of your DNS You can add multiple DNS here separated with a single space After you type all the required information save the file and reboot You can check the IP information by the ifconfig command or any website which can help to identify your IP 93 Figure A 1 WWLLN Service Unit Test v2 design 94 Figure A 2 WWLLN Service Unit Test v2 schematic 95 Figure A 3 Schematic for Service Unit box holes 96 Figure A 4 Schematic for Service Unit mounting holes 97 Figure A 5 Overview of WWLLN Service Unit Test VITA Hao Zheng left his hometown to join Peking University for a B S in Space Physics After that he spent 3 more years to get a M S in Space Physics at Peking University His master thesis studied the Bursty Bulk Flows and Dipolarization Fronts in the magnetotail In 2012 he took a 12hour flight to join University of Washington studying lightning-generated whistler waves
  • 2018
    • Bapst, Jonathan - Ph.D. Dissertation
      Mars’ Water Cycle Seen Through an Ice Lens 2018, Bapst,Jonathan,Jonathan Bapst Copyright 2018 Jonathan Bapst Mars Water Cycle Seen Through an Ice Lens Jonathan Bapst A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy University of Washington 2018 Reading Committee: Joshua L Bandfield Chair Shane Byrne Michelle Koutnik Program Authorized to Offer Degree: Earth and Space Sciences University of Washington Abstract Mars Water Cycle Seen Through an Ice Lens Jonathan N Bapst Chair of the Supervisory Committee: Dr Joshua L Bandfield Department of Earth and Space Sciences The water cycle of Mars has been intensely studied thanks to data from orbiting spacecraft and landers on the surface Water is of major interest due to its vital connection to life on Earth The global cycle is driven by the planet s major source of water vapor the north polar residual water-ice cap and the amount of this vapor is very sensitive to polar insolation Climate can be expressed in the behavior of water ice present at the surface both seasonally and perennially Here I enhance our knowledge of Mars current and recent climate by investigating the occurrence and physical structure of water ice at the surface with a combination of data analysis and numerical modeling In chapter 2 I analyze surface temperatures and albedo from broadband thermal-infrared and visible observations in order to characterize the seasonal cycle of water frost I identify extensive deposition of water ice in the northern hemisphere during autumn whereas the southern hemisphere shows little evidence for water ice deposition I argue this is a result of the configuration of major water sources at present i e the primary source of water vapor being at the north pole In chapter 3 I analyze the stability of icy outliers of the northern polar cap of Mars with both high-resolution imagery and numerical models These outliers are mounds of water ice 1050 km in diameter up to 2 km thick and are not contiguous with the residual cap the mostequatorward being at 70 N They are potentially unstable so secular changes in mound extent are explored using multi-year high-resolution images I estimate the current annual mass balance using a coupled 1-D thermal and atmospheric model Both lines of study support the outliers being close to equilibrium suggesting at present surface deposits may equilibrate at the same pace as the changing orbit In chapter 4 I derive maps of thermophysical properties of the north polar residual cap itself with thermal models and binned temperature data I investigate depth-density relationships in the polar water ice deposit and find that porous ice overlies a denser ice substrate which could indicate recent accumulation We find a noticeable difference between the interior and edge of the residual cap supporting recent accumulation over the interior and ablation at the edges The properties I derive will provide insight into the recent behavior of the martian north polar residual cap which remains largely unknown TABLE OF CONTENTS List of Figures iv List of Tables x Chapter 1 Introduction 1 Chapter 2 Hemispheric Asymmetry in Martian Seasonal Surface Water Ice from MGS TES 5 2 1 Introduction 5 2 1 1 Background 5 2 1 2 Lander Observations of Frost 7 2 2 Methods 10 2 2 1 General Methodology 10 2 2 2 TES Instrument Description 12 2 2 3 TES Bolometer Data 13 2 2 4 Dataset Constraints and Binning Methodology 14 2 3 Results 15 2 3 1 Overview of TES Bolometric Temperature and Albedo 15 2 3 2 Atmospheric Effects on Albedo and Temperature 19 2 3 3 Seasonal Water Frost Northern Hemisphere 22 2 3 4 Seasonal Water Frost Southern Hemisphere 26 2 4 Discussion 27 2 4 1 Comparison of Northern and Southern Hemisphere Observations 27 2 4 2 Comparison of TES and Lander Observations 30 i 2 4 3 Implications of Widespread Seasonal Water Frost 31 2 5 Conclusions 35 Chapter 3 On the Icy Edge at Louth and Korolev Craters 37 3 1 Introduction 37 3 2 Methods 40 3 2 1 High-resolution Image Analysis 41 3 2 2 Thermal Modeling 42 3 2 3 Atmospheric Model 47 3 3 Results 54 3 3 1 Multiyear HiRISE Image Analysis 54 3 3 2 Modeling Results 55 3 4 Discussion 61 3 4 1 Potential for Observing Change at Louth 61 3 4 2 Louth and Korolev Comparison 62 3 4 3 Seasonally-varying Albedo 63 3 4 4 Near-infrared Observations at Louth 66 3 5 Conclusions 67 Chapter 4 Thermophysical Properties of the North Polar Residual Ice Cap of Mars 70 4 1 Introduction 70 4 2 Methods 74 4 2 1 Observational Data 74 4 2 2 Model and Fitting Procedures 76 ii 4 3 Results 81 4 3 1 Homogeneous Model 81 4 3 2 Depth-varying Models 82 4 3 3 Best-fit Comparison 85 4 3 4 Properties of Residual Ice 85 4 4 Discussion and Conclusions 87 4 4 1 Comparison to Previous Work 87 4 4 2 Spatial Heterogeneity of the North Residual Cap 88 4 4 3 Ice Conductivity Models and Temperature Dependency 89 4 4 4 Conclusions 91 Chapter 5 Conclusions 93 5 1 Astrobiological Relevance 94 Bibliography 97 Supplementary Material 111 iii LIST OF FIGURES Figure 1 1 Mars obliquity history for the past 10 Myr from Laskar et al 2004 2 Figure 2 1 Images from Viking Lander 2 showing unfrosted left and frosted right surfaces in Utopia Planitia 48 N The frosted image is from LS 279 winter solstice and is interpreted as seasonal water frost Svitek and Murray 1990 The images are capturing roughly the same scene with the same rock 1 m circled in both images for reference 8 Figure 2 2 Example of seasonal water frost at 42 109 N 312 031 E High Resolution Imaging Science Experiment HiRISE images left ESP_032192_2225_COLOR 2 PM and middle ESP_030847_2225_COLOR 2 PM and corresponding Thermal Emission Imaging System THEMIS I07941019 5 PM smoothed brightness temperatures right Temperatures are consistent with the presence of water frost and are too warm for the presence of CO2 ice 9 Figure 2 3 Estimate of zonally-averaged frost-free surface albedo from TES These values were subtracted from albedo data to create change in albedo maps e g Fig 2 4 For details on calculating these values see section 2 4 2 12 Figure 2 4 Zonally-averaged 2 PM TES color-shaded albedo and temperatures three contours at 170 200 and 230 K Dominant features include seasonal polar caps dust storms and natural variability in the martian surface albedo e g more reflective northern hemisphere Note the large gap in data in MY25 due to a global dust event Color bar limits do not reflect the full range of the data 16 Figure 2 5 Zonally -averaged 2 PM TES color-shaded albedo change relative to summer surface see Section 2 1 Note the asymmetrical behavior in mid-to-high latitude albedo between the northern and southern hemispheres in late autumn Also note the general lack of interannual variations with the MY 25 dust storm being most obvious 17 Figure 2 6 The behavior of zonally-averaged TES albedo and AM PM temperature for 40 50 60 and 70 N in MY 26 Data are re-binned in 3 LS increments for clarity Dashed gray line indicates zero change in albedo As seen in Figures 2 4 and 2 5 the onset of albedo iv increase is earlier at higher latitudes and is consistent with water ice temperatures too high for CO2 50 N and 70 N can be compared to the VL2 and PHX sites respectively 18 Figure 2 7 Scatter plots showing trends in MY 26 zonally-averaged TES albedo change and temperature Dashed lines include interpretations of surface ices from this study Colors indicate latitude The largest albedo changes occur at an abrupt temperature characteristic of CO2 ice in equilibrium with the atmosphere d i e turbulent includes eddy diffusion i e forced convection and free convection Eddy diffusivity increases with altitude until z h then remains constant Free convection becomes less effective with height because density inversions only occur near the surface where sufficient water vapor is available The surface accumulation ablation model z0 40 high-TI unit 750 J m-2 K-1 s-1 2 in contrast to the surrounding lower-TI regolith Paige et al 1994 also show that the presence of thin mm coatings of dust are consistent with IRTM observations Putzig et al 2007 prescribe albedo in their fitting approach retrieve a TI value 74 for each temperature measurement and average the results However near-surface layering Bandfield and Feldman 2008 Putzig et al 2014 results in a seasonally-dependent apparent TI that is difficult to compare to that derived from temperature measurements acquired over a significant portion of the martian year e g from Paige et al 1994 and this work Regardless their averaged TI is consistent with the NRC as a high-TI unit In section 4 4 1 the results of our work will be compared to these earlier studies where applicable 4 2 4 2 1 METHODS Observational Data Data used in this study were acquired by the Thermal Emission Spectrometer TES aboard Mars Global Surveyor MGS MGS was operational over four Mars Years MY specifically MY24-28 see Piqueux et al 2015a for a description of the martian calendar The primary science orbit of MGS was inclined approximately 93 which resulted in high data density near the martian poles Fig 4 1 useful for a study of polar surface properties Unfortunately it also results in a region of low-to-zero data density poleward of 87 N This region of the pole was only observed sporadically via cross-track spacecraft rolls and off-nadir pointing of TES Because data are sparse here we ignore this region for this work 75 Figure 4 1 Number of observations per 10 by 10 km bin with latitude labeled Note the increase in density towards the pole and the logarithmic scale TES consists of a thermal-infrared spectrometer as well as co-aligned visible and thermal-infrared bolometers A single TES footprint is approximately 3 by 3 km but is lengthened by a factor of 2-3 in the along-track direction due to smearing Christensen et al 2001 Titus et al 2001 Putzig et al 2005 We assume brightness temperatures derived from the thermal infrared bolometer represent kinetic surface temperatures similar to previous studies using TES data to interpret thermophysical properties Putzig et al 2005 Putzig et al 2007 Bandfield and Feldman 2008 Data include observations poleward of 70 N for all Mars years and are restricted to emission angles 20 Additionally data with bad nadir opacity ratings are omitted Christensen et al 2001 Opacity ratings are sparse over the NRC due to relatively low daytime peak surface temperatures e g 90 We find it is especially important to omit spring data LS 0-90 as during this time the seasonal layer of CO2 ice is sublimating at the surface Piqueux et al 2015b A 1-D thermal model can only represent the surface as either CO2 covered or not whereas in reality the defrosting process is heterogeneous at scales smaller than TES can observe Searls et al 2010 During spring TES measurements may contain a mix of temperatures from CO2-covered and CO2-free surfaces potentially resulting in erroneous fits between our model and the observations We explored fitting with other seasonal windows starting at different times i e LS 0-180 Differences in results using a seasonal window starting between LS 90-150 were small with our preferred choice of LS 110 maximizing data beyond the defrosting season The choice of end season also made little difference as long as the window includes the onset of seasonal CO2 ice LS 180-200 Based on these tests we select a nominal window of LS 110270 for the derivations presented here 4 3 RESULTS Model fits of albedo and near-surface thermal properties are presented in this section Because our model is designed to retrieve the thermal properties of porous ice our results over regolith-covered surfaces are not valid and should be ignored for clarity bins with derived surface TI
    • Batbaatar, Jigjidsurengiin - Ph.D. Dissertation
      Quaternary glaciation in Central Asia 2018, Batbaatar,Jigjidsurengiin,Jigjidsurengiin Batbaatar Copyright 2018 Jigjidsurengiin Batbaatar Quaternary glaciation in Central Asia Jigjidsurengiin Batbaatar A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy University of Washington 2018 Reading Committee: Alan Gillespie Chair Summer Rupper David Montgomery Program Authorized to Offer Degree: Department of Earth and Space Sciences University of Washington Abstract Quaternary glaciations in Central Asia Jigjidsurengiin Batbaatar Chair of the Supervisory Committee: Emeritus Research Professor Alan Gillespie Department of Earth and Space Sciences The mountain ranges in Central Asia were heavily glaciated during the Quaternary Period The relative magnitudes of paleoglaciers varied spatially and temporally The chronology of glacial standstills was established from 37 sites spread over nine ranges in Central Asia using cosmic-ray exposure luminescence and radiocarbon dating techniques and the magnitudes of glacial standstills were estimated from equilibrium-line altitudes ELA of the paleoglaciers The study sites were chosen in regions where the ELAs predicted from numerical modeling to exhibit different sensitivities to changes in air temperature and precipitation The dating and the ELA depression from modern values ELA reveal that some paleoglaciers in humid regions advanced to their local maxima before the global Last Glacial Maximum LGM a period characterized by abrupt sea-level decrease and existence of large ice sheets on the continents The glacier ELAs in these humid regions were modeled to be more sensitive to air temperatures than to precipitation Most of the paleoglaciers advanced to their maximum during the global LGM suggesting a strong response to minima in insolation and air temperature at the time In cold and arid regions most of the paleoglaciers advanced to their maximum tens of thousands of years before the global LGM and their extents during the global LGM were limited to their cirques In one particular case a paleoglacier in the Mongolian Gobi advanced to its maximum during early Holocene without evidence of earlier glaciation This spatially and temporally asynchronous glaciations in cold and arid regions were probably driven more sensitively to changes in precipitation than in air temperatures a consistent response per the numerical sensitivity models The results highlight the importance of precipitation in controlling the advances of paleoglaciers in a continental setting TABLE OF CONTENTS List of Figures ii List of Tables v Chapter 1 Introduction 1 1 1 Motivation 1 1 2 Methods and approach 12 1 3 Organization of the dissertation 14 1 4 References in Chapter 1 16 Chapter 2 Outrburst floods of the Maly Yenisei 20 2 1 Abstract for Part I 21 2 2 Introduction to Part I 22 2 3 Timing of the floods 54 2 4 Discussion in Part I 72 2 5 Summary 76 2 6 References in Chapter 2: Part I 79 2 7 Abstract for Part II 88 2 8 Introduction to Part II 89 2 9 Methods in Part II 93 2 10 Results 107 2 11 Discussion in Part II 144 2 12 Conclusions to Part II 149 2 13 Supplementary material to Chapter 2 152 2 14 References in Chapter 2: Part II 162 Chapter 3 Asynchronous glaciations in arid continental climate 172 3 1 Abstract 173 3 2 Introduction to Chapter 3 174 3 3 Methods 180 3 4 Results 208 3 5 Discussion 240 3 6 Summary and Conclusions 250 3 7 References in Chapter 3 252 Chapter 4 Spatial pattern of glaciations across the climate-transect of Central Asia 259 4 1 Abstract 260 4 2 Introduction to Chapter 4 262 4 3 Methods 267 4 4 Results 276 4 5 Discussion 310 4 6 Summary and conclusions 315 4 7 References in Chapter 4 341 Chapter 5 Summary and conclusions 350 5 1 Summary and conclusions 350 5 2 Discussion 355 5 3 References in Chapter 5 359 i LIST OF FIGURES Figure 1 1 Glacial asynchrony around the globe 3 Figure 1 2 Geography of the mountain ranges in Central Asia 4 Figure 1 3 Areas with available glacial geomorphological maps in Central Asia 6 Figure 1 4 Paleo-ELA estimation at the Hoit Aguy mountain southern Siberia 8 Figure 1 5 Annual precipitation and summer temperature at the modern ELAs 9 Figure 1 6 Annual precipitation vs melt fraction of total ablation 11 Figure 1 7 Sites with new radiometric ages reported in this dissertation 13 Figure 2 1 Peak discharge and volume of water from major Quaternary floods 23 Figure 2 2 Map of Ob and Yenisei river basins 24 Figure 2 3 Major tributaries to the Yenisei river and their local names 29 Figure 2 4 Basin map of the upper Yenisei river 32 Figure 2 5 Giant current ripples in the Kyzyl basin 36 Figure 2 6 Maps of the lower Tengis river and the Maly Yenisei 43 Figure 2 7 Topographic profile along the Maly Yenisei and across Darhad basin 45 Figure 2 8 Paleolake extents in Darhad basin 51 Figure 2 9 Stratigraphic relationship of sediment outcrops in Darhad basin 58 Figure 2 10 Normalized distribution density of 10Be ages from literature 65 Figure 2 11 Inundation map of a flood from the Darhad paleolake 1710 m level 74 Figure 2 12 Shorelines and highstands of Darhad paleolakes 77 Figure 2 13 Darhad basin map with lake levels 92 Figure 2 14 10Be sampling sites in Sayan Hoit Aguy Darhad and Hangai 102 Figure 2 15 Stratigraphic column of the DBC1 core 110 Figure 2 16 10Be sampling site on the end moraine of the Tengis glacier 128 Figure 2 17 East Sayan sampling sites 130 Figure 2 18 10Be sampling site at the M nh Saridag mountain 133 Figure 2 19 10Be sampling site at the pass between Darhad and H vsg l basins 134 Figure 2 20 Hoit Aguy moraine map 136 ii Figure 2 21 10Be sampling site near the Gyalgar mountain area 138 Figure 2 22 Normalized distribution density of the new 97 10Be ages 145 Figure 2 23 Two alternative models for subsidence of the Darhad basin floor 148 Figure 2 24 Shorelines and highstands of Darhad paleolakes 150 Figure 2 25 Potential increase in the apparent exposure ages due to erosion 154 Figure 2 26 Precipitation gradients in the vicinity of Darhad basin 159 Figure 2 27 Summer temperature and the ELA in the vicinity of Darhad basin 160 Figure 3 1 Geographic location of the Hangai and Gobi-Altai ranges 176 Figure 3 2 Modern annual precipitation and summer seasonality 178 Figure 3 3 Modern annual precipitation and melt fractions at the study sites 179 Figure 3 4 Moraines in the M nh M snii valley Gichginii range 209 Figure 3 5 Gichginii plateau and glacial extent in the M nh M snii valley 210 Figure 3 6 Surface features of the G1 ridge in the M nh M snii cirque 212 Figure 3 7 Surface features of the upper part of the G2 moraine 213 Figure 3 8 Surface features of the upper part of the G3 moraine 214 Figure 3 9 Surface features of the lower part of the G4 moraine 215 Figure 3 10 Camel plots for the 10Be ages from Gichginii range 216 Figure 3 11 Sample locations in Sutai range 218 Figure 3 12 Camel plots for the 10Be ages from Sutai range 221 Figure 3 13 Non-glacial samples at Sutai 223 Figure 3 14 Ih Bogd range and the 10Be sample locations 225 Figure 3 15 Camel plots for the 10Be ages from Ih Bogd range 226 Figure 3 16 Otgontenger site and the 10Be sample locations 228 Figure 3 17 Camel plots for the 10Be ages from Otgontenger Hangai range 229 Figure 3 18 Bumbat site and the 10Be sample locations 230 Figure 3 19 Camel plots for the 10Be ages from Bumbat valley Hangai range 231 Figure 3 20 Field photos of some of the sampled boulders 232 Figure 3 21 Annual precipitation and summer air temperature at the ELA 234 Figure 3 22 Sublimation- and melt-dominated regimes for glacier ablation 236 Figure 3 23 Timing of local glaciation and the ELA lowering 239 iii Figure 3 24 Ephemeral running water on the South Col of Chomolungma 246 Figure 3 25 Schematic diagram relating ELA depression to change in precipitation 248 Figure 4 1 Study areas in continental Central Asia 263 Figure 4 2 Summer air temperature and melt fraction vs annual precipitation 265 Figure 4 3 Kax Kurta valley Altai range and the 10Be sample locations 278 Figure 4 4 Study sites in Diehanjelegou Alashanje and Daxigou East Tian Shan 282 Figure 4 5 Study sites in Muzart and Tailan West Tian Shan 284 Figure 4 6 Study sites in Barskoon and Suek West Tian Shan 287 Figure 4 7 Study site in Gulbel West Tian Shan 289 Figure 4 8 Study site in Choktal West Tian Shan 291 Figure 4 9 Study site in Altyn Tagh Qilian Shan 293 Figure 4 10 Study site in Dumda Dunde ice cap Qilian Shan 295 Figure 4 11 Study site in Gangshiqia Qilian Shan 297 Figure 4 12 Study sites near Askaiqin lake and Quanshui glacier West Kunlun 299 Figure 4 13 Study site in Karakax West Kunlun 302 Figure 4 14 Map of modern ELA in Central Asia 303 Figure 4 15 Map of MIS 2 maximum ELAs in Central Asia 304 Figure 4 16 Latitudinal gradients of modern and MIS 2 ELAs 306 Figure 4 17 Melt fraction vs the timing of the local LGM standstills 309 Figure 4 18 Glacier advance and retreat in response to changing climate 314 Figure 5 1 Normalized ELA depressions in Central Asia through time 356 iv LIST OF TABLES Table 2 1 Selected large outburst floods from glacier-dammed lakes 26 Table 2 2 Geographic names used in Chapte 2 27 Table 2 3 Names of geographic features and their local variants 29 Table 2 4 Highstands and volumes of the paleolake Darhad 52 Table 2 5 14C samples collected in Darhad basin 59 Table 2 6 Published and recalculated 10Be ages in and around Darhad basin 66 Table 2 7 Sedimentation breaks in the DBC1 core 111 Table 2 8 Diatom species in the lake sediments from the DBC1 core 116 Table 2 9 Radioactivity data for the luminescence samples 118 Table 2 10 Luminescence ages calculated by various models 119 Table 2 11 Luminescence ages from Darhad basin 120 Table 2 12 14C samples and data 122 Table 2 13 10Be data for new samples reported in Chapter 2 125 Table 2 14 10Be ages for samples from East Sayan Hangai Hoit Aguy Sarig 126 Table 2 15 ELA for modern glaciers and modern climate 142 Table 2 16 10Be exposure ages calculated using various scaling schemes 155 Table 2 17 Increase in calculated exposure ages after correcting for erosion 156 Table 3 1 10Be exposure ages from the Gobi-Altai and Hangai ranges 182 Table 3 2 10Be data used for exposure-age calculation 184 Table 3 3 Outlier evaluation by normalized deviation method 191 Table 3 4 Outlier evaluation by Chauvenet s criteria 193 Table 3 5 Outlier evaluation by Peirce s criteria 195 Table 3 6 Summary of outlier evaluations 197 Table 3 7 Equilibrium-line altitude estimations 202 Table 3 8 Temperature and relative humidity measurements at Sutai 205 Table 3 9 Summary of modern climate at the study sites 206 Table 3 10 Changes in paleoclimate parameters compared to modern values 207 v Table 3 11 Average age and total uncertainty for the 10Be ages 243 Table 4 1 Data for new cosmogenic nuclide analysis in Chapter 4 316 Table 4 2 New 10Be and 26Al exposure ages reported in Chapter 4 319 Table 4 3 Summary of outlier evaluations 321 Table 4 4 Mean 10Be ages for groups with two samples 325 Table 4 5 Modern and paleo-ELA estimates for glaciers considered in Chapter 4 327 Table 4 6 Modern and paleo-ELAs for dated moraines and modern climate 336 vi : 37 vii - - - - viii ACKNOWLEDGEMENTS This dissertation was a culmination of support by many individuals to whom I am grateful I am indebted to my advisor Alan R Gillespie for his excellent mentorship He showed me how to be a better scientist a better colleague and a better human Thank you Alan for your unbelievable patience The scientific works of my committee members were formative to my own work I would like to thank James Feathers David Montgomery Gerard Roe Summer Rupper and John Stone for their ample inspiration and guidance Specific connections and opportunities built a strong foundation for my research Amgalan Bayasgalan took me on a trip to Darhad basin where I met Alan for the first time and D Sukhbaatar of Damdin Da Foundation provided logistical support in the collection of the early samples reported in my dissertation I am grateful for the opportunity to contribute to great research by Matthew Smith and Rivka Amit I was not alone in carrying the intellectual and physical weight of my research many people supported me with mentorship expertise equipment and finances As colleagues and mentors B Charlotte Schreiber Ari Matmon David Fink Amit Mushkin Michele Koppes Yehouda Enzel ZhongPing Lai and Toshiyuki Fujioka were crucial to the success of my research Ron Sletten Howard Conway Steve Warren and Al Rasmussen lent their expertise and equipment for the weather measurements at Sutai mountain Galhuu Bat-Ochir and Byambaa helped me in the field Jody Bourgeois provided financial support for that trip through the department s graduate student awards I am grateful to Carrie Garrison-Laney and Brian Sherrod who analyzed the diatoms in my samples ix Without the support of the incredible staff at the Earth and Space Sciences department of UW my life would have been much harder Robert Winglee Bruce Nelson Harvey Greenberg Eunice Yang Kathy Gabriel No ll Bernard-Kingsley Monick Keo Ed Mulligan and Dave McDougall all provided unwavering support My fellow graduate students shared much-needed friendship and company during coffee breaks: Moon Young Choi Zo Harrold Matthew Smith Max Needle Karl Lang Mike Turzewski and everybody else My parents Tsetseg Damba and Jigjidsuren Gombojav prodded me into a world of science and supported me all the way With my love adoration and respect thank you My brothers Bodisuren and Bodibaatar have been good caretakers at home during my adventures abroad My boys Tsogt-Ochir and Munh-Orgil were pure joy in my life except when they were asking for iPad passwords Last but certainly not the least I cannot thank enough my wife Bayarkhuu Bat-Ochir who sacrificed her career and followed me along this journey x DEDICATION To my wife Bayaraa xi 1 Chapter 1 INTRODUCTION 1 1 MOTIVATION The distinctive mountain landscapes carved by past glaciers and adorned by modern glaciers near high peaks fascinate every observer of nature The almost-synchronous demise of modern glaciers around the globe however has become the self-evident pulse on the current warming climate attributed to humans This is because the glaciers directly respond to local climate conditions and by aggregating the spatial and temporal patterns of glacier advances and retreats this response can be used to infer regional changes in climate Roe et al 2017 Furthermore glacial deposits left from past glacial advances and retreats can be used to reconstruct paleoclimate variables in places where no other proxies are available In general an accurate reconstruction of paleoclimate from glacial deposits requires the following three steps: 1 Establish the chronology of glacial advances and or standstills using radiometric dating 2 Establish and quantify a consistent metric for the state of a glacier 3 Establish the relationship between glacier state and climate parameters Each of these steps has been addressed in the literature which I will summarize in the following paragraphs and which will highlight the problems that need improvement This dissertation is focused on glacial evidence in Central Asia because a large number of glacial deposits were left under various climatic or environmental conditions The methods and approach used in this dissertation are summarized in Section 1 2 and the main results of the Chapters 2 4 are summarized in Section 1 3 of this Chapter 2 The high mountain ranges of Central Asia were formed around 50 million years ago when the Eurasian and Indian continental plates collided Molnar and Tapponier 1975 and the cold peaks of these mountains have been modified by glaciers during the glacial cycles of the Pleistocene epoch e g Koppes and Montgomery 2009 Despite the vast expanse of glacial landscapes and deposits in Central Asia the continent was not accessible to a broad scientific community from World War II to about 1990 due to the political divide between the countries in eastern and western hemispheres The political thaw between the two camps around 1990 s brought increased collaboration between researchers which allowed for example a comparison of glaciations in the North and South Americas Europe with those in Central Asia One of the first such global observations by Gillespie and Molnar 1995 Figure 1 1 noted two general types of asynchronism between the glacial advances in different climate regions: 1 Asynchrony in the timing of maximum glacial extents and 2 Asynchrony in the amplitude of glacial extents during the same climate stade In other words when the most continental ice sheets were gaining their maximum extents during the global Last Glacial Maximum global LGM around 21 ka many mountain glaciers were not at their maximum extents Instead many mountain glaciers extended to their maximum lengths tens of thousands of years before the global LGM from which Gillespie and Molnar 1995 noted the importance of precipitation in regulating the glacier growth However the constraints on the ages of glacial advances were poor at that time and the highresolution temporal comparison of different glacier advances was not possible Since then the development and widespread use of cosmogenic nuclides have allowed directly dating the boulders on moraines which precisely mark the past positions of glaciers A compilation of cosmic-ray exposure ages previously published in literature is now available in a global database 3 http: expage github io and comprises an invaluable tool for establishing temporal comparison of glacial advances Figure 1 1 Glacial asynchrony around the globe from Gillespie and Molnar 1995 Schematic summary of the timing and relative extent of the of the late Pleistocene alpine glacial advances surveyed in Gillespie and Molnar 1995 The authors noted that the ages were poorly constrained and this presentation was speculative Locations of the study sites are shown by crosses and the exact coordinates are available in the original article Although there are more than 2000 individual cosmogenic nuclide ages are available for approximately 500 glacial deposits in Central Asia most of the ages are from the PamirKarakorum-Himalaya regions in which the precipitation is derived mostly from Indian monsoon Figure 1 2 In this dissertation the study sites are chosen in the mountain ranges located in more continental climates such as the Sayan Hangai Gobi-Altai Tian Shan Qilian Shan and Kunlun ranges to sample from the less represented regions Chapter 2 covers the East Sayan ranges and Hangai ranges establishing the glacial chronology and the chronology of related giant glacial 4 outburst floods down the upper Yenisei River in the northernmost parts of Central Asia Chapter 3 focuses in the Gobi-Altai and Hangai ranges documenting an extreme example of asynchronous glaciations within a short distance Chapter 4 expands the glacial chronology to include the Tian Shan Qilian Shan and Kunlun ranges and evaluated the glacial history in the context of the glacial chronology in other regions Figure 1 2 Geography of the mountain ranges in Central Asia Dashed lines indicate the modern and global LGM limits of the East Asian monsoon Shi 2002 Solid lines indicate the general direction of major air flows Benn and Owen 1998 5 The growth of a glacier is dictated by its mass balance the difference between the annual accumulation and ablation It is a straightforward metric of glacier growth: Precipitation fallen as snow accumulates in high altitudes and if it is greater than the mass lost to sublimation and seasonal melt then the glacierized ice flows downhill to a lower altitude until it is warm enough to melt the remaining ice Relationships between the glacier mass balance and climate is addressed in the next paragraph and here I discuss only the measuring of the glacier mass balance In a geometric sense mass of a glacier can be estimated at various altitudes by multiplying the glacier volume with measured or assumed density of ice The volume can be measured in three quantities: length width and thickness Among these three the length is the most straightforward measure of glacier positions past and present Indeed the longest records of historical measurements of glaciers were of their lengths Roe et al 2017 Glacier width varies from the accumulation zone to the ablation zone but is relatively easy to estimate for modern glaciers from aerial and satellite images For example Global Land Ice Measurements from Space project GLIMS uses satellite images to outline the glacier ice extents over the globe to compile a worldwide inventory of glaciers Randolph Glacier Inventory: RGI 2017 For paleoglaciers however it is not so easy to estimate the paleo-width because the glaciers are not there and no automatic detection-algorithm has been developed to delineate the paleoglacier extents Therefore the mapping of paleoglacier extents require extensive field investigation and photo interpretative mapping and these are laborintensive efforts As far as I know there is no worldwide map of paleoglacial landscapes and deposits A few regional-scale maps exist for Central Asia Figure 1 3 covering NE Tibet Heyman et al 2008 Central Tibet Mor n et al 2011 SE Tibet Fu et al 2012 the Tian Shan Stroeven et al 2013 and the Altai and Western Sayan Blomdin et al 2014 in addition to individual maps provided with the dating studies of glacial deposits 6 Figure 1 3 Areas with available glacial geomorphological maps in Central Asia from Blomdin et al 2014 The maps are for a Altai and Western Sayan Mountains Blomdin et al 2014 b Tian Shan Stroeven et al 2013 c Tangula Shan Mor n et al 2011 d Bayan Har Shan Heyman et al 2008 and e Shaluli Shan Fu et al 2012 Thickness is the hardest of the three quantities to estimate For a modern glacier its bed topography can be estimated using a numerical model based on the empirical relationship between the bed slope and ice thickness e g Frey et al 2004 Kraaijenbrink et al 2017 For paleoglaciers Benn and Hulton 2010 developed a simple numerical model to reconstruct a surface topography of a glacier based on the basal stress caused by the weight of the ice These labor-intensive mapping techniques of glacier volumes however do not directly relate to the amount of accumulation and ablation of a glacier On the other hand equilibrium-line altitude ELA of a glacier an altitude of a zone on a glacier-surface in which the annual accumulation and ablation equals provides a direct measure of glacial advance and retreat in response to the climate 7 forcing acting on the glacier Among the many ways of estimating the ELA the Toe-Headwall Altitude Ratio THAR and Accumulation-Area Ratio AAR methods are accepted to be the most reliable e g Porter 1975 Meierding 1982 Porter 2001 The AAR method assumes that the accumulation area of the glacier covers a certain portion of the total area of the glacier 65% for most temperate debris-free glaciers Porter 2001 and it requires an accurate estimation of glacier outline and thickness of the paleoglacier which is hard to acquire for the reasons mentioned above The THAR method assumes a simple valley geometry for the glacier and estimates the ELA at the threshold value of the altitude difference between the toe and the headwall of a glacier Eq 1 1 % % 1 1 where At is toe altitude Ah is headwall altitude THAR is a threshold value The value of THAR varies from glacier to glacier but the recent compilation studies have shown that the median altitude of the glaciers were appropriate estimations of THAR ELAs e g Heyman 2014 Nuimura et al 2015 Maximum Elevation of Lateral Moraine MELM method assumes that the deposition of till occurs only below the ELA where ablation exceeds accumulation Meierding 1975 Porter 2001 and is a useful estimation of ELA where only lateral moraines exist and no end moraine is found However due to small chance of depositing lateral moraines on steep slopes or subsequent erosion the MELM considered to be the least reliable of the methods mentioned above Meierding 1982 Porter 2001 Batbaatar and Gillespie 2012 used these three techniques to estimate the ELAs for well-defined mountain glaciers in the Eastern Sayan range and found that the disagreement between the ELAs estimated with the three methods was 100 m Figure 1 4 which was no more than the technique precision given by Meierding 1982 and Gillespie 1991 All ELAs for modern and paleoglaciers presented in the Chapter 2 4 were estimated using the THAR and MELM methods and the justification for the use of the methods is provided in each instance 8 Figure 1 4 Paleo-ELA estimation at the Hoit Aguy mountain southern Siberia from Batbaatar and Gillespie 2012 The ELA values m asl were estimated using THAR AAR and MELM methods The location is near N 51 5 4 and E 98 7 The final step in the reconstruction of paleoclimate from glacial deposits depends on how reliable we understand the relationship between the ELA and the climate In general there are two models that establishes this relationship: 1 Empirical models based on the summer temperature and the annual precipitation at the ELA e g Ohmura et al 1992 and 2 Physically based numerical models that account for all the atmospheric variables and calculate the ablation for comparison to the accumulation e g Kayastha et al 1999 M lg and Hardy 2004 Rupper and Roe 2008 9 An empirical curve relating the summer mean air temperature and annual precipitation at the ELA was first established by Ohmura et al 1992 by extrapolating the above-mentioned variables for 70 glaciers worldwide Sakai et al 2015 Figure 1 5 used a high-resolution gridded climate dataset to establish a similar empirical curve estimated for the modern glaciers in Central Asia 4000 3500 y 370 7e0 3623x R 0 84 Annual precipitation mm 3000 2500 2000 1500 1000 500 0 -8 -6 -4 -2 0 2 4 6 8 Summer JJA air temperature C Figure 1 5 Annual precipitation and summer temperature at the modern ELAs for Central Asian glaciers The black dots are the actual data from Sakai et al 2015 and the red circles are calculated values from the equation derived from fitted exponential curve 10 From this empirical curve two different regime of glaciers emerge: 1 In regions with summer temperature 0 C the glaciers are sensitive to both temperature and precipitation and the annual precipitation is always 150 mm 2 In regions with summer temperature 2 C the glaciers are most sensitive to precipitation and the annual precipitation ranges from 40 250 mm This empirical relationship between summer air temperature and annual precipitation should also hold true for the paleoglaciers and if one of the variables is known from an independent proxy then the other could be estimated using the paleo-ELA This is very appealing for its simplicity and establishes an important framework for reconstructing paleoclimate from glacial records Physically based numerical models calculate the ablation at various altitudes of a glacier using important atmospheric variables driving the glacier growth such as air temperature incoming and outgoing short and longwave radiations and accounts for wind speed and cloud cover The accumulation will exceed ablation at higher altitudes due to colder environment than in the lower altitudes The decreased ablation with lowering altitude then must equal to accumulation at an altitude where the ELA is defined Rupper and Roe 2008 developed such a model to predict the ELA sensitivities of Central Asian glaciers to climate variables and highlighted two modes of glacier ablation Figure 1 6 : 1 In regions with 500 mm annual precipitation melt is responsible for most of the ablation and the ELA is more sensitive to air temperature than to precipitation 2 In regions with
    • Burgener, Landon - Ph.D. Dissertation
      A Window into Terrestrial Paleoclimate: Soil Carbonate Formation Processes and Climate Proxy Applications
      Appendix 1 2018, Burgener,Landon,Landon Burgener
    • Danner, Mariah - M.S. Thesis
      Cratering Characteristics of the Europa Kinetic Ice Penetrator 2018, Danner,Mariah,Mariah Danner Copyright 2018 Mariah L Danner i Cratering Characteristics of the Europa Kinetic Ice Penetrator Mariah L Danner A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science University of Washington 2018 Reading Committee: Robert Winglee Chair Erika Harnett Michael McCarthy Program Authorized to Offer Degree: Earth and Space Sciences University of Washington ii Abstract Cratering Characteristics of the Europa Kinetic Ice Penetrator Mariah L Danner Chair of the Supervisory Committee: Professor Robert Winglee Earth and Space Sciences This thesis further develops the Europa Kinetic Ice Penetrator EKIP landing technique for airless bodies as well as characterizes the effect EKIP would have on Europa s surface Damage to the extremophile Planococcus Halocryophilus OR1 PHOR1 during a laboratory hypervelocity impact test was studied the effect of rapid application of pressure to microbes frozen in ice Significant die-off occurred however PHOR1 microbes survived a 2 2km s impact Field testing the second-stage deployment as well as to characterize crater morphology of the EKIP system was conducted With low impact velocities penetrators consistently had deeper narrower craters than natural impactors rocks and showed less radial and sub-impactor compression This and future crater data into harder substrates will create a cratering hardness curve for this design impactor into airless bodies This curve used with the eventual in situ craters can be used to constrain the hardness and other physical properties of the surface of icybodies iii TABLE OF CONTENTS List of Figures vii List of Tables xi Chapter 1 Introduction 1 1 1 Background 1 1 2 Previous Work 3 1 2 1 The Grave-Digger Asteroid Penetrator 3 1 2 2 Ice Penetrator Laboratory Shots 5 Chapter 2 Microbial Impact 7 2 1 Experimental Set-Up 8 2 1 1 AERB Ram Gun Set-Up 8 2 1 2 Microbe- Planococcus halocryophilus OR1 9 2 1 3 Ice Preparation 11 2 1 4 Sampling and Testing Error Bookmark not defined 2 2 2 2 1 2 3 Results 13 Microbial Survivability 14 Future Microbial Work 15 Chapter 3 EKIP Field Testing 17 3 1 Penetrator Design 18 3 1 1 Nozzle Design 20 3 1 2 Straight Bore 21 iv 3 1 3 3 2 Finless 24 Penetrator Drop 25 3 2 1 Control Impactors 28 3 2 2 Engineered Impactors 30 3 3 Penetrator Results 31 3 3 1 Crater Morphology 32 3 3 2 Individual impactors 33 3 3 3 Field Test Results and Summary 48 Chapter 4 Second Stage 49 4 1 Second Stage Design 49 4 1 1 Birdie Design 49 4 1 2 Construction 51 4 2 Birdies 52 4 2 1 Birdie 1- Tinkerbell 52 4 2 2 Birdie 2 Flora 54 4 2 3 Birdie 3 Fauna 56 4 2 4 Birdie 4 Merriweather 59 4 3 Results and Future Work 60 Chapter 5 Proposed Research 60 5 1 Alaskan Glacier Test 60 5 2 Alaskan Sea Ice Test 61 5 3 Birdie Development 61 v 5 4 Biological Capture and Detection 61 Bibliography 62 Appendix A 64 vi LIST OF FIGURES Figure 1 1 Artist rendition of Europa Clipper image credit: NASA JPL-Caltech 2 Figure 1 2 Artist rendition of the Europa Kinetic Ice Penetrator departing from the Europa Clipper After separation from the Clipper the lander separates into two stages As the first stage impacts the surface d it creates a plume that interacts with and decelerates the second stage e and f 3 Figure 1 3 Artistic rendition of asteroid penetrator system Hollow bore up the center of the impactor allows for rock flow into a sample chimney Image courtesy of Chad Truitt 4 Figure 1 4 Rock core created from rock penetrator 5 Figure 1 5 Modified ice projectiles Left: Spike projectile used for initial testing Right: Cylindrical projectile with interior bevel 6 Figure 2 1 Ram gun diagram courtesy of www aa washington edu research ramaccel facility 8 Figure 2 2 Ram gun set-up 9 Figure 2 3 Planococcus halocryophilus OR1 Nadia C S Mykytczuk et al 2012 10 Figure 2 4 Ice block diagram for bio-shot 12 Figure 2 5 LIVE DEAD staining of the frozen control sample 14 Figure 2 6 LIVE DEAD staining of the impacted sample 14 Figure 3 1 Photo of the four finned penetrators From the left: Stumpy Long John Firefly and Slim Jim 19 Figure 3 2 Basic schematic of Stumpy Though not to scale the interior nozzle design is visible 21 Figure 3 3 Schematic of Slim Jim 22 Figure 3 4 Schematic of Long John 23 Figure 3 5 The Taku Glacier with the Taku Towers in the distance The drop site is directly below the Towers onto the center of the glacier The hole pictured is the mass balance hole shown in Figure 3 16 26 vii Figure 3 6 Camp 10 the base of operations for the drop test operated by the Juneau Ice Research Program 26 Figure 3 7 Ice balance pit dug out by JIRP Density measurement holes along left side of image measurement probe for scale Ice lenses indicated by red arrows beginning of firn layer indicated by yellow arrow 27 Figure 3 8 The helicopter piloted by Coastal Helicopters above Camp 10 Projectiles were slid out of tubes from open door Three separate flights were needed for all drops to occur 28 Figure 3 9 Typical crater morphology of an impact crater Similar to what is expected from the control rocks dropped during the test Figure modified from Osinski 2013 29 Figure 3 10 Diagram of penetrator crater 31 Figure 3 11 Projectile drop map Key: Circles penetrators diamonds rocks Blue Black Density measurements Pink Red no density measurements A Slim Jim B Stumpy C Long John D In Steel E Fire Fly F Outer-Steel 1 2 3 Rocks 32 Figure 3 12 A: Full depth of Stumpy s crater measured to the end of the projectile measured to 1 84m This image was taken after crater infill was removed B: 11 34kg rock crater apparent depth measured 0 5m true depth 1 1m 34 Figure 3 13 Close-up of Stumpy before surrounding snow and ice was removed Visible ice lens outside the penetrator indicated by finger is not present inside the debris snow within the penetrator This indicated the ice lens was pulverized upon impact 35 Figure 3 14 Stumpy s crater prior to removal of crater infill Colored squares indicate where density measurements were taken: Orange Control 1 Blue Control 2 Red Sample 1 Green Sample 2 36 Figure 3 15 A: Full depth of Slim Jim s crater measured to end of penetrator 1 5m Colored squares correspond to density samples: Orange control pink sample 1 sample 2 not shown B: Corresponding rock impact measured to actual apparent depth of 3m width of 4m 38 Figure 3 16 A: Top view of Slim Jim s crater Very little ejecta created clear marking of where fins impacted B: Corresponding rock impact crater from above Wide shallow crater created with minor ejecta forming a crater rim 38 viii Figure 3 17 A: Core created from Slim Jim impact Core was removed from penetrator during removal of penetrator from crater The very front of the core was still attached to the firn layer impacted by the penetrator and was unable to be removed 39 Figure 3 18 Crater created by Long John Penetrated to a depth of 1 2m at end of penetrator 40 Figure 3 19 A: Possible compression lines created by a surface feature on Long John B: Compressional lines horizontally parallel to the 9 98kg rock 41 Figure 3 20 Ice core created by Long John 42 Figure 3 21 Impact crater of Inner Steel finless design A: Full crater with colored squared indicating density samples: Blue control red steel aluminum transition pink nose bulb B: Close up of the nose bulb 43 Figure 3 22 Core created by Inner Steel finless design 44 Figure 3 23 Out steel finless landing position Due to a flat spin the penetrator impacted on its side 45 Figure 3 24 Impact of Firefly A: Depth of crater to end of penetrator only 3cm B: Crater with penetrator removed 46 Figure 3 25 The core created by Firefly more similar to a cookie cutter 46 Figure 3 26 Plot of penetrator mass to crater depth for the finned penetrators Though the more massive penetrators did impact deeper two similar in mass penetrators with differing diameters were different This could be due to more air resistance on the larger diameter penetrator 47 Figure 3 27 Plot of impactor to snow density Chute refers to the path the penetrator rock traveled through the snow The sub-impactor density for the rocks was much higher than that of the penetrators 47 Figure 4 1 Birdie design without chute fabric A: Design mid-way open B: Arms stowed 50 Figure 4 2 Birdie design with chute fabric Penetrator shown attached to front electronics to the top A: Design mid-way open B: Arms stowed 51 Figure 4 3 Tinkerbell chute A: Shown being stowed by hand B: Chute open and upside down 53 ix Figure 4 4 Tinkerbell impact A: Untouched landing position B: Chute when removed from penetrator 54 Figure 4 5 Flora undisturbed impact orientation Due to minimal air resistance the birdie fell straight down and impacted perpendicular to the glacier surface 55 Figure 4 6 Close-up of Flora s Impact A: Over compression of the springs caused the springs to bend Otherwise no damage occurred to the skeleton B: Snow imprint from Flora s skeleton Based off small indentations one arm bounced once before spreading to a larger angle than designed 56 Figure 4 7 Full configuration of Fauna and Long John 57 Figure 4 8 Fauna landing position A: Viewed from the south B: Viewed from the north 58 Figure 4 9 Trajectory path of birdies through wind bands Colored bands show wind bands corresponding color arrow shows wind direction and magnitude Dotted line shows trajectory without wind interaction Dashed line shows trajectory of birdie steering into oncoming wind 59 Figure 4 10 Merriweather impact viewed from east A and south B 59 x LIST OF TABLES Table 2 1 DAPI Staining Results 15 Table 3 2 Penetrator Attribute Table 19 Table 3 3 Crater Density Table Stumpy 36 Table 3 4 Crater Density Table Slim Jim 39 Table 3 5 Crater Density Table Long John 42 Table 3 6 Crater Density Table Inner Steel Finless 44 Table 4 7 Birdie Attribute Table 52 xi 1 Chapter 1 INTRODUCTION 1 1 BACKGROUND The Galilean moons of Jupiter have recently been the focus of interplanetary science missions with continued interest in the icy moon Europa as an important astrobiology target Of particular interest to its subsurface ocean which may sustain life Research into the chemistry and geology of both the icy shell and oceanic subsurface continue see Appendix A Previous missions such as Rosetta s Philae lander and its unsuccessful anchoring to cometary ice exemplify our lack of in-depth knowledge into the dynamics of extra-planetary ice The next mission the Europa Clipper is being designed to clip or side-swipe Europa as the satellite navigates through the Jovian system s complex orbits and radiation bands Due to congressional mandate a lander will launch from the Europa Clipper to explore Europa s surface Two main lander designs are underway a traditional lander designed by NASA s Jet Propulsion Laboratory JPL and the Europa Kinetic Ice Penetrator EKIP led by the University of Washington in collaboration with NASA Ames Winglee et al 2018 The EKIP concept is designed around the process theory of cryo-braking a technique to slow the lander down in a near-zero atmosphere environment using a two stage system Cryobraking uses the transfer of kinetic energy momentum from a kinetic impact to form a channelized debris plume to an impact-lander As shown in Figure 1 2 the Europa Kinetic Ice Penetrator will separate from the Europa Clipper departing the satellite with a small Delta V relative to the spacecraft but with a speed of roughly 4 km s relative to Europa s orbital speed With the velocity imparted from the Clipper the lander will approach the Europan surface separating into two distinct stages Once separated the first stage will continue towards the 2 surface while the second stage deploys a mechanical chute The penetrator will impact the ice creating the channelized ice plume necessary for cryo-braking As the second stage interacts with the ice plume the second stage will decelerate to a velocity the onboard instrumentation can withstand The momentum transfer will be enough to slow but not stop the second stage allowing for a low-velocity impact of the second stage Due to the braking being caused by relative velocity differences of the two stages instead of fuel contamination of Europa s surface is less likely This thesis will focus on research done on the effects of the EKIP landing system would have on Europa s surface namely the craters caused by the impactor system to develop an empirical theory for engineered impacts into extra-planetary ice It will be possible to apply this theory to better understand the dynamic continuum of planetary ice in a way that is difficult using traditional methods The effects of impacts on possible microbial life will also be addressed Figure 1 1 Artist rendition of Europa Clipper image credit: NASA JPL-Caltech 3 Figure 1 2 Artist rendition of the Europa Kinetic Ice Penetrator departing from the Europa Clipper After separation from the Clipper the lander separates into two stages As the first stage impacts the surface d it creates a plume that interacts with and decelerates the second stage e and f 1 2 1 2 1 PREVIOUS WORK The Grave-Digger Asteroid Penetrator The Europa Kinetic Ice Penetrator derived from a previously funded project the GraveDiggers or the asteroid penetrators also being developed at the University of Washington This system of asteroid sampling utilizes impactors contained on a satellite which once launched will impact a given asteroid As shown in Figure 1 3 the penetrator body consists of a hardened nose cone with a hollow bore for which the impacted substrate flows Once the sample 4 fills an interior sample return canister a tether is able to retrieve the sample from the asteroid Preliminary testing of the Grave-Digger system created a rock core shown in Figure 1 4 When the penetrator impacted sandstone at a velocity nearing Mach 2 the sandstone flowed into the sample bore but stagnated within the penetrator body Truitt 2016 Redesign of the interior sample return canister including a shock absorbing spring enabled rock sample to flow into the canister and eject upon impact Winglee et al 2017 The ejected sample retained the stratigraphy of the impacted rock along with the macro-biology in the overlaying soil Nose cone and experimental designs from the Grave-Digger system have been modified for the EKIP landing system utilizing the penetrator to break up the ice to eject a transient atmosphere that can be used to provide passive braking for a following payload Figure 1 3 Artistic rendition of asteroid penetrator system Hollow bore up the center of the impactor allows for rock flow into a sample chimney Image courtesy of Chad Truitt 5 Figure 1 4 Rock core created from rock penetrator 1 2 2 Ice Penetrator Laboratory Shots The Europa Kinetic Ice Penetrator presented new obstacles that needed to be overcome before large scale testing could begin Proof of concept for the EKIP landing system creating an ice plume with an engineered projectile impacting at hyper velocity needed more rigorous analysis the channelization of impact ejecta into a plume has not previously been studied Initial testing with miniaturized penetrator designs was conducted in the RAM Accelerator at the University of Washington Aeronautics and Astronautics Engineering Department This design compares to the European Space Agency s ESA penetrator program designed to withstand an impact up to 300 m s Collinson 2008 The University of Washington penetrator is designed to supersede this impact velocity In order to optimize the characteristics of the ejecta modifications were made to the Asteroid penetrator nozzles Initially numerical models of miniaturized penetrator 38 mm diameter impacts were created These designs are then shot at high velocities into ice blocks Impacts velocities reached between 1-2 kilometers per second with the target blocks varying in size between 35 to 100 gallons of fresh ice High speed cameras determined velocities and ejecta characteristics such as plume morphology Figure 1 5 shows two projectile designs tested The 6 spike projectile design is most similar to a rocket nose cone or bullet the hypothesis for this design being a deeper penetration into the ice target The projectile is made of 6061 Aluminum with a magnetic strip attached to record velocity during testing The second design the cylinder projectile is also made of 6061 Aluminum Unlike the spike projectile the cylinder projectile is hollow with an interior bevel to create a more efficient cutting edge upon impact The cylinder projectile design is both lighter and was determined to create a more channelized ejecta plume than the spike projectile though penetration depth was less Robinson 2017 The plume created by the cylindrical projectile travels roughly 1 3 the velocity of the impactor an empirical relationship established by laboratory testing Robinson 2017 Figure 1 5 Modified ice projectiles Left: Spike projectile used for initial testing Right: Cylindrical projectile with interior bevel 7 Chapter 2 MICROBIAL IMPACT Possible microbial life or an environment able to support microbial life is a major aspect of the interest in Europa s ocean and ice crust Ice thicker than a few feet s ability to block radiation Paranicas et al 2002 protects any microbial life able to survive in the subsurface ocean from the extreme radiation emitted from Jupiter as well at the cosmic radiation from the sun With the continued interest in possible microbial life in Europa s subsurface the logistics behind landing on the moon s surface are substantial Not only must the process of landing be considered but the effect the landing system has on the surrounding ice must be considered as well To study possible microbes using the EKIP landing system it must be ensured the landing technique will not destroy evidence of microbes in the impacted ice With an initial impact velocity of the penetrator of 4 km s it is imperative to determine the effect the impact shockwaves and subsequent cratering have on the microbes Related research studies have been conducted in regards to the panspermia theory that extremophilic microbes shielded within an impactor i e meteoroid not only survive hyper velocity impacts but spread to the surrounding substrate following impact Mastrapa et al 2001 The microbes in Mastrapa et al are well protected shielded within the impactor similar to how shuttles protect astronauts from heat during launch Instead of protection from heat the microbes are protected from extreme acceleration and rapid changes in acceleration which could damage cell membranes In comparison this study focused upon microbes contained within a substrate impacted by an outside projectile The microbes in our study are much more vulnerable to changes in acceleration or pressure waves that could damage cell membranes The objective of this study is to map where the most damage to the microbes occurs if the microbe survives dies or is completely destroyed within the impact crater 8 2 1 EXPERIMENTAL SET-UP The set-up used for this experiment has been used for numerous previous experiments regarding ice impact shots utilizing inter-departmental coordination and resources 2 1 1 AERB Ram Gun Set-Up The Aeronautic Engineering Research Building AERB at the University of Washington contains a Ram Accelerator gun run by Dr Carl Knowlen The ram gun is capable of accelerating a projectile more than 8 km s Hertzberg 1988 utilizing similar combustion techniques to that of a jet-engine In the experiments described below the gun was used in the light gas gun mode where only helium was used and no combustible were present to avoid heating of the ice by any exhaust gases Some contamination of the ice does occur from byproducts that have coated the gun barrel surface from previous combustion experiments Figure 2 1 Ram gun diagram courtesy of www aa washington edu research ramaccel facility For this experiment the full length of the 40-meter-long barrel was evacuated to near vacuum pressure The entry diaphragm is made out of 2mm thick plastic film known to rupture 9 at pressures able to accelerate the projectile up to 2 km s The ice block was held horizontally a half meter from the exit diaphragm care was taken to ensure centering of the ice block with the end of the barrel Figure 2 2 Ram gun set-up 2 1 2 Microbe- Planococcus halocryophilus OR1 Microbe choice was key for this experiment The conditions on Europa are more extreme than any on Earth temperatures ranging from -187 15C to -141 15C on the surface no atmosphere and saline waters Spencer 1988 An extremophile is an organism evolved to survive in extreme 10 environments such as the extreme heat of hydrothermal vents or extreme cold beneath a polar ice sheet With assistance from the Fishery Sciences department we chose the extremophilic microbe Planococcus halocryophilus OR1 PHOR1 for this research Figure 2 3 Planococcus halocryophilus OR1 Nadia C S Mykytczuk et al 2012 Other extremophiles are able to live in either extremely acidic basic anoxic or irradiated environments PHOR1 is an extremophile suited to saline freezing conditions therefore uniquely suited for this experiment The microbe is able to reproduce at -16C and survive down to -20C while being able to hibernate if frozen to -60C with minimal die off Mykytczuk 2013 PHOR1 is highly researched previously and is known as an extremely robust and reliable easily recognizable microbe to work with Importantly PHOR1 has previously been tested with LIVE DEAD staining the process planned to determine survivability in this experiment Though it is obvious that this specific microbe would not survive the conditions on Europa it is assumed that if an extremophile is able to evolve to survive the extreme conditions 11 on Earth theoretically an extremophile on Europa would be able to match and survive in Europa s environments as well For this experiment it is not imperative for all the microbes to survive only that some survive the impact The objective is to prove the impact does not destroy all evidence of life The microbes grown by the Fisheries department at the University of Washington were grown in a protein solution to provide the microbes with food to reduce die off and then diluted into two gallons of water The two-gallon solution was then frozen to -20C over a couple of hours Though flash freezing with liquid nitrogen would have reduced die off the process to flash freeze two gallons was impractical for the experiment The microbes were frozen for less than 24 hours prior to the experiment being conducted 2 1 3 Ice Preparation The ice used in the experiment was frozen in the Quaternary Research Center s QRC D freezer the same used in the freezing of ice targets for the original EKIP ice test shots The freezer fluctuates between -35C to -20C through daily defrost cycles During the defrost cycles the ice stays cold enough to not actually melt the purpose of the defrost cycle being to remove any ice on the condenser in the freezer The ice block was frozen into a 100-gallon water barrel a meter in diameter The ice was frozen in layers roughly 10 cm thick roughly 20 gallons of fresh water at a time Water chilled to the point of slush forming in buckets was poured into the main barrel to limit crystalline orientation occurring as much as possible Fresh water was frozen until 30cm from the top of the bucket 35ppm saline water with a composition of half table salt NaCl and half Epsom salt MgSO4 made up the upper half of the ice bucket The saline ice formed into a ring with a 30cm high 30cm diameter cylinder in the middle That cylinder was then filled with a separately 12 frozen sample of microbe filled ice The entire ice block was then frozen for roughly 20 hours before the experiment took place Figure 2 4 Ice block diagram for bio-shot Though ice layers may cause boundary effects during the impact i e shock wave reflection over layer boundaries the freezing process of a singular mass of ice of this size roughly 800lbs of ice would be impractical for time requirements for this experiment Other boundary effects are possible due to the joint of the microbial ice from the saline ice causing a weak juncture where the two intersect An ejecta shield attached to the exit diaphragm of the Ram gun collected samples This ejecta shield was sterilized using isopropyl alcohol immediately before the experiment took place Larger debris from the crater collected into a half-meter wide splash bucket also sterilized using isopropyl alcohol before testing Factory sterilized 3cm sample tubes collected samples from inside the crater All samples were stored immediately in a cooler to reduce melting 13 The dye LIVE DEAD from thermofisher stained the microbes to determine survivability This dye stains living cells fluorescent green while dead cells and cell remnants stain fluorescent red This process has the benefit of allowing dead but not destroyed cells to still be distinguishable versus stains that only allow visualization of living cells With help from the Astrobiology lab within the Earth and Space Sciences department microscopic analysis of a nonfrozen control sample frozen un-impacted sample and impacted sample occurred As a precaution DAPI testing was also conducted on both a control and impacted sample of the microbe by the Oceanography and Fisheries departments on campus This was to ensure a lab bias wasn t an issue as well as to avoid a staining bias by the microbe DAPI testing is a more robust stain in regards to longevity of sample life however the stain only bonds to nondamaged cells meaning the test would only be beneficial if microbes survived the impact Though steps were taken to reduce contamination the nature of the testing site adds an inherent possibility contamination from the lab or even the barrel of the gun Though contamination was a mild concern in this experiment the chosen microbe PHOR1 is highly recognizable under microscope The microbe also cannot survive in fresh water which means contamination of PHOR1 in the fresh ice sample or water frozen to make the target isn t plausible As long as PHOR1 is found in impacted samples contamination by other microbes is less of an issue 2 2 RESULTS The experiment was conducted November 2016 in the of the AERB A 60-gram cylindrical aluminum projectile was used to impact the ice at 2 2 km second High speed camera and GoPro 14 footage was taken of the experiment Roughly 29 gallons 0 11 cubic meters of ice was ejected or removed from the 100-gallon ice block 2 2 1 Microbial Survivability After both LIVE DEAD and DAPI staining it was determined that the microbe Planococcus halocryophilus OR1 survived the 2 2 km second impact Using LIVE DEAD Baclite staining it is visible that small microbes resembling PHOR1 survived the impact though large amounts of damage and die off occurred Minor contamination also is visible in both the control and impacted samples Due to the survival of the planted microbe the occurrence of contamination is of little concern for this experiment Figure 2 5 LIVE DEAD staining of the frozen control sample Figure 2 6 LIVE DEAD staining of the impacted sample 15 DAPI staining conducted by the Oceanography and Fisheries department showed similar die off but confirmed the existence of living PHOR1 cells within the samples from both within the crater and the ejecta caught in the ejecta shield Similar minimal amounts of contamination assumed to be a strain of E coli was also noted using the DAPI staining Table 2 1 DAPI Staining Results Key: C refers to the control sample collected prior to testing E refers to ejecta collected after testing RT refers to temperature as in room temperature No growth apparent implies the microbe is dormant or inactive 2 3 FUTURE MICROBIAL WORK The notable result from the experiment is the microbe survived the impact The lack of complete destruction of the cellular material let alone the survival of Planococcus halocryophilus OR1 from a 2 2 km second impact means the Europa Kinetic Ice Penetrator s impact landing system may be a viable option for the possible research of microbial life on Europa Further testing including more robust freezing must be completed A fully saline ice block with higher densities of microbial ice frozen to higher hardness would remove some boundary 16 effects from the experiment Removing layering to an extent would also be optimal Higher care will also be taken to reduce contamination in further testing Beside changes in freezing techniques the velocity of impacts will be gradually increased in further testing Increasing velocities up to the predicted 4 km second impact velocity of the EKIP lander will be necessary to fully rule out possible microbial destruction from the proposed impact landing system During these tests characterization of zones within the crater will be analyzed to determine which zones contain the most microbial damage 17 Chapter 3 EKIP FIELD TESTING The initial work on the Europa Kinetic Ice Penetrator impact landing system design has focused heavily on miniaturized versions of the first stage impactor the penetrator for laboratory tests Testing of the full system including separation of the second stage payload section as well as electronic survivability needed to be conducted as part of my research Previous miniaturized designs of the Europa Kinetic Ice Penetrator s impactor system have been tested in the laboratory These tests though at high velocities similar to those expected for the mission boundary effects caused by size restrictions cause issues with crater morphology and shock wave propagation through the ice Field testing on Easton Glacier of Mt Baker WA was planned but later canceled due to the target area being determined too small of an area for safe operation A field test was then planned for Juneau Alaska to ensure a large enough target area Existing designs were modified into true-to-scale projectiles to impact a glacier increasing the boundary limits to effectively infinite proportions i e depth horizontal path The second stage of the EKIP mission the payload section needs to have a compact design that can fit with the space restrictions of the spacecraft and at the same time have a large crosssectional area to attain a sufficiently large braking surface to allow efficient deceleration of the payload section within the generated transient atmosphere These requirements lead to the conclusion that the payload section must have a deployable surface Different designs were considered It was concluded that a shuttlecock design provides the greatest aerodynamic stability and structural integrity for the proposed operating conditions With this being the initial design testing of separation balance structural integrity sizing and material strength must be tested for further designs Further discussion of the second stage is included in Chapter 4 18 This field test though intended to be a proof of concept for the EKIP landing system is also being used as the first of several empirical tests to catalogue the morphology of the EKIP lander in ices of differing hardness such as snow sea ice blue glacial ice and basalt When the depth and shape of the crater in each hardness substrate is combined into an empirical curve the results of an impact on an actual icy planetary body can be fit into the curve This data will give insight into the ice dynamics of the impacted surface in a way remote sensing is unable to Furthermore these tests will be used to ground truth the biology laboratory experiments to ensure the compressional effects seen in the miniaturized shots are still accounted for in true to scale tests 3 1 PENETRATOR DESIGN The first stage of the EKIP landing system the penetrator determines the ability of the second stage to slow down Through the creation and channelization of ejecta from the impact of the penetrator an artificial atmosphere can be created on Europa enabling the use of a chute-centric deceleration technique The channelized ejecta or the ejecta plume must be both dense and fast enough to impart a considerable amount of momentum to the second stage To ensure this miniaturized impactors were shot into ice at high velocities The basic shape of the penetrator was determined but due to mass and size constraints the actual design of the penetrator needed to be further designed at larger scale in field testing An overview of each of the six penetrators designed for field testing are contained in the table below For in depth descriptions of each penetrator see following sections 19 Table 3 2 Penetrator Attribute Table Penetrator Design Mass kg Material Length cm Bore Diameter cm Stumpy Nozzle 12 A2 tool steel 35 15 max 7 6 min Slim Jim Straight Bore 7 3 Steel Aluminum 51 7 6 Long John Straight Bore 9 9 Steel Aluminum 39 10 1 Firefly Straight Bore 3 4 Aluminum 22 7 6 Inner Steel Finless 6 4 38 10 1 Outer Steel Finless 4 1 33 7 6 Steel Aluminum Steel Aluminum Figure 3 1 Photo of the four finned penetrators From the left: Stumpy Long John Firefly and Slim Jim 20 3 1 1 Nozzle Design The nozzle design of the penetrator mimics that of an actual rocket nozzle Because the ejecta needs to be kept at high velocity the design of a nozzle is ideal Rocket nozzles using a converging-diverging nozzle design are able to accelerate propellant converting thermal energy to directed motion Utilizing the converging-diverging nozzle a penetrator should be able to accelerate the ejecta created from impact in a similar fashion During laboratory testing the ejecta was accelerated to roughly 1 3 the velocity of the impactor Robinson 2016 However during field testing an inert penetrator the terminal velocity is subsonic meaning the nozzle cannot accelerate the ejecta However this design will be used in future supersonic impacts so the nozzle design must also be used during the first field test For field testing several different designs were considered The penetrator nicknamed Stumpy was designed with the intent to mimic a nozzle The intent of this penetrator was to maximize mass while minimizing impact surface A2 tool steel was chosen for its high density to increase mass as well as its resistance to fracturing Because the penetrator would only be used for a single impact the durability was of less importance than maximizing the mass However the penetrator had to withstand the impact without significant deformation or run the risk of blocking ejecta flow from misshaping With a total mass of 11 8 kg 20 cm length and 15 cm width Stumpy is the second smallest but most massive of the penetrators The exterior beveled edge used as a cutting edge coincides with a straight cut converging angle into a 7 6cm diameter straight bore The nozzle then continues to a 10cm long straight diverging cut extending out to 14 5 cm diameter at the outlet of the penetrator Four steel fins were welded to the exterior of the penetrator to ensure stable flight Even with the attached fins the minimal distance between the center of pressure CP to the center of mass CM increased the chances 21 for unstable flight or tumbling For this reason the fins were extended back past the end of the penetrator to extend the distance between the CP and CM Figure 3 2 Basic schematic of Stumpy Though not to scale the interior nozzle design is visible 3 1 2 Straight Bore The second design type the straight bore is similar to designs for the nosecone of the asteroid penetrator As implied the bore hole of the penetrator in this design is straight unlike that of the nozzle design In this design the leading edge often has an exterior or interior exterior bevel to create a cutting edge for initial impact In comparison to the nozzle design the straight bore design intends for the ejecta to eject en masse instead of breaking apart and possibly decelerating within the penetrator 22 The first straight bore design nicknamed Slim Jim is a dual material design between steel and aluminum The steel at the leading edge is beveled to decrease impact surface with a joint continuing into the aluminum The steel impacting edge has durability and mass while the aluminum addition increases the over al length With a diameter of 7 6 cm and length of 51 cm this design was to maximize stability while falling With such a long body the center of pressure in Slim Jim is much more conducive to stable flight than Stumpy Small fins were welded to the aluminum body to further increase stability However due to the excessively long straight bore the chances of flow stagnation before ejecta can exit the penetrator increase heavily as seen in previous testing with the rock penetrator Truitt 2016 Figure 3 3 Schematic of Slim Jim 23 The second straight bore design nicknames Long John is another dual material design of aluminum and steel With the same material orientation as Slim Jim as well as the same bevel on the leading steel edge Long John is extremely similar in design to Slim Jim However the name is a misnomer with Long John actually being shorter than Slim Jim At 39 cm length and 10 1 cm diameter this design has a larger impact surface but is less likely to cause flow stagnation of the ejecta The wider bore may cause flow deceleration in relation to a smaller diameter bore Figure 3 4 Schematic of Long John The final straight bore design with the nickname of Firefly is the smallest designed penetrator This design is purely made of aluminum with an interior exterior bevel on the front edge Totaling just under 18cm in length and 3kg in mass Firefly is dwarfed in size by the other designs With minimal mass Firefly will have a small terminal velocity in comparison to the 24 other more massive penetrators To increase the mass from the pure aluminum design iron weights were added to the nose edge of the penetrator With the minimal length stagnation of the ejecta is unlikely though lacking mass and speed may minimize the amount of ejecta created upon impact 3 1 3 Finless The final penetrator design the Finless design was the final to be created These penetrators were included as a fail-safe for the drop test created as an attempt to quickly increase the number of penetrators able to be dropped This was to hopefully increase the probability of a successful test drop as well as increasing the chances of finding the penetrators after they ve impacted This design is another set of dual material impactors without a straight bore and without the attachment of fins This lack of fins reduces air resistance skin friction during flight meaning the same mass penetrator has the ability to achieve higher velocities However the lack of fins also destabilizes the flight of the penetrator increasing the chance of tumbling during flight To reduce tumbling the penetrators were heavily weighted to the front edge of the penetrators using material weighting The first penetrator of this design Inner Steel is designed exactly as the name implies An interior bevel on a 7 6cm diameter steel pipe was cut with the steel pipe then being inserted into a snuggly fitting aluminum pipe With the increasing diameter from 7 6 cm to 10 1 cm a primitive compression zone and divergence was created With the total length of 38 cm the primitive nozzle created decreased the chance of flow stagnation of the ejecta The total mass of the penetrator 6 4 kg makes this penetrator the median mass of the penetrators As mentioned above the lack of fins on this design may lead to tumbling To minimize this effect the penetrator was heavily weighted with towards the leading edge This was done using the conjoining of steel and aluminum With a 25 large ratio of steel to aluminum the majority of the mass lies in the leading end However due to the steel portion being extremely long the mass is spread up into the penetrator instead of being focused near the leading edge This could cause an additional lack of stability The second finless design Outer Steel is yet again designed as the name implies Going from a beveled 12 7 cm bore steel tube to a 10 6 cm aluminum tube all ejecta entering the penetrator is compressed This primitive nozzle design may cause an increase in ejecta velocity though at the minimal extent could lessen flow stagnation in the ejecta flow With a total length of 33 cm and the steel extending only 10 cm the majority of the mass is weighted at the leading edge of the penetrator This weighting of the material will hopefully counteract the destabilization caused by the lack of fins Outer Steel being 4 1 kg is the 2nd smallest penetrator designed 3 2 PENETRATOR DROP Field testing for the Europa Kinetic Ice Penetrator s drop test was conducted from 17 July 2017 to 23 July 2017 on the Juneau Icefield Alaska The actual drop test occurred on 20 July 2017 on the Taku Glacier through coordination with the Juneau Ice Research Program 26 Figure 3 5 The Taku Glacier with the Taku Towers in the distance The drop site is directly below the Towers onto the center of the glacier The hole pictured is the mass balance hole shown in Figure 3 7 Figure 3 6 Camp 10 the base of operations for the drop test operated by the Juneau Ice Research Program 27 The Taku Glacier a 386mi2 Alaskan tidewater glacier sits at an elevation of 4700 feet above sea level at the site of the drop The glacier was covered in an annual snow layer roughly 2m thick The annual snow had an average density of 0 42g cm3 interspersed with two to three ice lenses indicated on the figure below The local geology consists of feldspathic granite interspersed with quartzite dikes Cobbles from a nearby nunatak were collected for additional drop test projectiles The Taku Towers granitic peaks standing at 1597 m were used as the minimum allowable cloud ceiling for the drop to occur A helicopter provided by Coastal Helicopters was used for the drop Figure 3 7 Ice balance pit dug out by JIRP Density measurement holes along left side of image measurement probe for scale Ice lenses indicated by red arrows beginning of firn layer indicated by yellow arrow 28 Figure 3 8 The helicopter piloted by Coastal Helicopters above Camp 10 Projectiles were slid out of tubes from open door Three separate flights were needed for all drops to occur 3 2 1 Control Impactors While the overall goal was to test the specifically designed penetrators and document the differing effects of each design it was also necessary to compare the craters created by the penetrators with those of natural impactors i e rocks Though impact cratering is heavily studied the majority of the studies focus on impacts at high velocities This test will focus on low velocity impacts Though a rough comparison similarities between the rock craters and high velocity impact craters known to impact theory can be shown With the same comparison the low velocity impact craters can be extrapolated to predict crater morphology of high velocity penetrator craters Future high velocity field testing will further compare and hopefully validate this relation 29 In order to accurately compare craters from the penetrators cobble sized rocks of similar masses to the penetrators were dropped These rocks chosen from a scree hill on a nunatak are the same composition as the surrounding geology Through classic impact theory the crater s typically have depth to diameter ratios between 1:5 and 1:7 Osinski 2013 regardless of speed or size of impactor This characteristic ratio creates a shallow wide crater with specific ejecta patterns shown in the figure below The craters created during the field test lack several key components from Osinski s diagram shown below The ejecta ad allochthonous crater-fill deposits are not present due to the low impact velocity The impactor in hypervelocity impacts is usually destroyed while the rocks dropped were notably still present The broken up substrate shown with the rock craters could be analogous to the impact breccias referred to in the figure The rock craters fall under the category of simple craters which are roughly bowl shaped and are controlled through compressional forcing Because of the low velocity impact and the effect of gravity the rock craters create more of a pit than a traditional crater shape Figure 3 9 Typical crater morphology of an impact crater Similar to what is expected from the control rocks dropped during the test Figure modified from Osinski 2013 30 The control rocks were 7 26kg 9 98kg and 11 34kg chosen for their roughly spherical shape in order to imitate natural impactors Though impactors this size i e meteoroids would burn up in Earth s atmosphere the lack of atmosphere on Europa would enable smaller impactors to reach the surface The rock crater created by the 11 34kg rock measured to an apparent depth of 0 5m a true depth of 1 1m and a diameter of 1m Thus the rock crater s depth to diameter ratio stands at 1:2 Though a smaller ratio than normally seen with natural impactors this can be explained by the low velocity of the impact 3 2 2 Engineered Impactors Based upon laboratory experiments the engineered impactors the penetrators have a distinctive crater morphology completely different from that of a natural impactor When impacted at mid to high velocities into blocks of ice the penetrators have a typical depth to diameter ratio of 3:1 to 5:1 depending on the velocity of impact This reversal of the depth versus diameter ratio in comparison to natural impactors is most likely caused by the excavation design of the penetrators With the large impact surface on natural impactors the majority of the force of impact is imparted both horizontally and vertically The momentum separated into the horizontal direction increases the substrate fracturing in the horizontal plane while reducing the depth of the crater While ejecta from natural impactors typically reenters the crater or is deposited on the rim of the crater ejecta from the penetrators is shown to evacuate the crater almost entirely with side walls of the crater typically caving inwards 31 Figure 3 10 Diagram of penetrator crater The penetrators also have a distinctive plume of ejecta in the laboratory shots a characteristic unlikely to occur in the field test due to low velocities and soft wet snow as a substrate 3 3 PENETRATOR RESULTS Wind conditions for the drop averaged 10knots from the South at the helicopter 10knot katabatic winds from the North at ground level Cloud base was roughly 4000 feet above the glacier Projectiles were dropped roughly 2500 feet about the glacier Due to the crossing wind patterns 32 during the drop the projectiles scattered along a roughly 1 km cross section of the glacier shown in the map below Figure 3 11 Projectile drop map Key: Circles penetrators diamonds rocks Blue Black Density measurements Pink Red no density measurements A Slim Jim B Stumpy C Long John D In Steel E Fire Fly F Outer-Steel 1 2 3 Rocks 3 3 1 Crater Morphology The cratering between the cobbles and the penetrators was highly different The cobbles showed the characteristic impact crater hypothesized thought impact theory: the craters were relatively shallow and wide However the rock craters were deeper than typical natural impact craters possibly due to the extremely soft substrate and low velocities 33 The overall shape of the craters created by the penetrators had an average depth to diameter ratio of 19:1 to 20:1 This extreme ratio difference between not only natural impactors but laboratory shots could be explained yet again by the low density and highly compactible substrate Unlike the ice in the laboratory shots which fractured upon impact the snow in the field test acted more like a cookie cutter going through clay during impact Density measurements were taken at multiple crater sites including a rock and four penetrators A map of craters with density measurements taken is shown below Detailed density measurements for each crater can be found in section 3 3 2 3 3 2 Individual impactors The impactor section will focus on each individual crater created by each penetrator Craters created by the rocks will be used for comparison purposes only Stumpy The crater created by Stumpy shown in Figures 3 23 through 3 25 was the second deepest crater created With a depth measured to end of penetrator of 1 84m and diameter of 0 12m the crater has a depth to diameter ratio of 15:1 The reversal of the ratio between the penetrator and the rock crater agrees with results from laboratory shots 34 Figure 3 12 A: Full depth of Stumpy s crater measured to the end of the projectile measured to 1 84m This image was taken after crater infill was removed B: 11 34kg rock crater apparent depth measured 0 5m true depth 1 1m No plume was created by the impact of this penetrator However through examination of the crater the nozzle design began the process needed to create a plume The nozzle penetrator is designed to compress and then expand material passing through the nozzle This process pulverizes material and though difficult to see in snow the destruction of ice lenses impacted by the penetrator as shown in figure 3 13 clearly show that material going through the nozzle is not merely being slid past like a cookie cutter Further evidence of this process can be seen in the density measurements taken As shown in figure 3 14 the crater infill leftover from the penetrator passing through the ice shows evidence of being broken up The density of the surrounding annual snow layer averages 0 45 g cm3 throughout this area of the glacier The crater infill has a density of 0 41 g cm3 while the density of the snow directly below the penetrator was measured to be 35 0 61g cm3 These measurements along with all the density measurements were taken roughly 48 hours after impact the impact This means refreezing will equalize densities to a minor extent This means that the lower density of the infill may have been initially lower directly following impact Regardless the lower density infill and lack of evidence of wall collapse could mean the pulverized snow ice in the infill would have been evacuated into a plume had the penetrator impacted at a higher velocity The higher density snow seen below the penetrator is mostly likely caused by compression waves as the penetrator lost momentum and flow stagnated within it Figure 3 13 Close-up of Stumpy before surrounding snow and ice was removed Visible ice lens outside the penetrator indicated by finger is not present inside the debris snow within the penetrator This indicated the ice lens was pulverized upon impact 36 Figure 3 14 Stumpy s crater prior to removal of crater infill Colored squares indicate where density measurements were taken: Orange Control 1 Blue Control 2 Red Sample 1 Green Sample 2 Table 3 3 Crater Density Table Stumpy Sample Control 1 Control 2 Sample 1 Tunnel Sample 2 nose Depth cm 75 75 75 210 Mass g 43 40 44 90 41 10 60 50 Volume cm3 99 00 99 00 99 00 99 00 Density g cm3 0 44 0 45 0 42 0 61 37 Slim Jim The long welded two material penetrator Slim Jim was the first penetrator to be excavated Though the penetrator impacted to a smaller depth than Stumpy if measured to the end of the penetrator Slim Jim impacted the largest depth when measured to the nose of the penetrator Slim Jim s nose impacted the firn layer making this the only penetrator to penetrate all the way through the seasonal snow layer Blue dye used to mark the impact zone colored the crater infill This penetrator crater had a depth to diameter ratio of 10:1 The smaller ratio of Slim Jim than Stumpy may be accounted for by the mass difference and subsequent velocity difference between the two penetrators The crater shape from above of Slim Jim versus the rock crater shown in figure 3 27 differ substantially The penetrator crater appears as if the penetrator cut straight through the snow upon impact similar to that of a cookie cutter Clear impressions of where the fins impacted is visible A small amount of debris can be seen along one side of the penetrator crater possibly due to the penetrator impacting at a small angle before straightening Another possibility could be that the penetrator had a slight spin upon impact causing debris to be created However due to the asymmetric occurrence of the debris it is most likely that the penetrator came in at a small angle The impact created an ice core of compressed layers from the annual snow shown in figure 3 17 The layering could easily be seen with the naked eye Upon the accidental drop of the core the core easily broke apart into individual layers believed to be snow events of the annual snow compressed into ice Density measurements of the crater were completed with crater infill being slightly less dense than the control snow around the crater A sample taken from below the nose of the penetrator showed a higher density 0 13 g cm3 higher than the control This is believed to be due to the compression caused by the ice core building beneath the penetrator forcing snow 38 beneath the penetrator to become denser Unfortunately density measurements could not be taken of the ice core Figure 3 15 A: Full depth of Slim Jim s crater measured to end of penetrator 1 5m Colored squares correspond to density samples: Orange control pink sample 1 sample 2 not shown B: Corresponding rock impact measured to actual apparent depth of 3m width of 4m Figure 3 16 A: Top view of Slim Jim s crater Very little ejecta created clear marking of where fins impacted B: Corresponding rock impact crater from above Wide shallow crater created with minor ejecta forming a crater rim 39 Figure 3 17 A: Core created from Slim Jim impact Core was removed from penetrator during removal of penetrator from crater The very front of the core was still attached to the firn layer impacted by the penetrator and was unable to be removed Table 3 4 Crater Density Table Slim Jim Sample Control Sample 1 Sample 2 Depth 30cm 100cm 230cm Mass 44 50 42 00 57 50 Volume 99 00 99 00 99 00 Density 0 45 0 42 0 58 Long John The crater created by Long John shown in Figure 3 18 was excavated the second day The penetrator impacted to a depth of 1 3m to the end of the penetrator The crater had an average measured width of 13 9cm giving the crater a depth to diameter ratio of 9 4:1 Of the penetrators that came in perpendicular this one penetrated the least This may be due to the similar mass to Slim Jim yet larger drag due to surface features That same surface feature likely caused 40 compression halfway through the penetrator as the projectile traveled through the snow As shown in part A of figure 3 19 slight deformation can be seen leading away from the surface indent in Long John s steel portion In comparison to compression features caused by the rock impact as shown in part B of figure 3 19 Long John s surface feature caused minimal deformation Figure 3 18 Crater created by Long John Penetrated to a depth of 1 2m at end of penetrator 41 B Figure 3 19 A: Possible compression lines created by a surface feature on Long John B: Compressional lines horizontally parallel to the 9 98kg rock As with Slim Jim Long John also created a snow ice core The core shown in Figure 3 20 corresponded only to the length of the penetrator unlike Slim Jim s which extended behind the end of the penetrator This core exhibited no layering leading to the conclusion that flow did not stagnate until the penetrator stopped moving In comparison to Slim Jim s core which showed evidence of flow stagnation halfway through impact Long John s core shows that the wider bore throat along with a shorter throat lessens the chance of flow stagnation Surface features on the core itself indicate either the snow spiraled while flowing through the penetrator or more likely the penetrator was still spiraling at depth The core is noticeably denser than the surrounding snow including that of the chute This is most likely due to the compression the snow underwent while flowing and subsequent refreezing Based upon malleability testing conducted upon removal the core created by Long John was noticeably easier to indent than the core created by Slim Jim indicating the density of the core created by Slim Jim was denser than that of the core created by 42 Long John This could mean the snow in Slim Jim s core was much more compressed than that of Long John s even though the impact velocities should have been similar This would indicate the larger bore penetrator allowed less stagnation than the smaller bore penetrator Figure 3 20 Ice core created by Long John Table 3 5 Crater Density Table Long John Sample Control 1 Control 2 Core 1 Core 2 Sample 1 chute Sample 2 nose Depth 50cm 50cm N A N A 50cm 215cm Mass 49 00 48 00 421 00 354 50 47 50 53 50 Volume 99 00 99 00 628 32 477 52 99 00 99 00 Density 0 49 0 48 0 67 0 74 0 48 0 54 Inner Steel Finless The Inner Steel finless design penetrator impacted at an angle curving slightly once it penetrated into the snow Because of the non-perpendicular impact the penetrator achieved very little depth The depth to diameter ratio for this impact is slightly skewed due to the impact angle with the actual diameter being difficult to determine at the entryway of the crater Using the 43 diameter near the end of the penetrator the depth to diameter ratio stands at roughly 4:1 An interesting feature of this impact is the nose-bulb of ice created at the nose of the penetrator shown in Figure 3 21 This feature was likely created as a result of flow stagnation The flow most likely stagnated instantly upon impact because of the low impact velocity as well as the impact angle imparting much of the force downwards while the penetrator went sideways The remaining force in the penetrator combined with the larger compressional face created a larger compressional wave a bow shock of sorts at the lead of the penetrator When the penetrator ceased movement the compressional wave compressed the surrounding snow into a spherical ice lens The density of the nose bulb was 2 g cm3 higher than the control and slightly higher density than the core and chute Figure 3 21 Impact crater of Inner Steel finless design A: Full crater with colored squared indicating density samples: Blue control red steel aluminum transition pink nose bulb B: Close up of the nose bulb The Inner Steel finless design also created a snow core upon impact It s doubtful that flow actually occurred with this impact with the resulting core most likely being the result of a cookie 44 cutter effect An interesting effect with the core as shown in figure 3 22 is the change in diameter corresponding to the expanded inner diameter of the second half of the penetrator Because the snow had to first go through the smaller diameter opening the snow had to expand into the second larger half This is most likely a pressure effect while refreezing due to the lack of other evidence of flow Figure 3 22 Core created by Inner Steel finless design Table 3 6 Crater Density Table Inner Steel Finless Sample Control Core Nose Bulb Steel Al transition Depth 15 cm 50cm 100cm 75cm Mass 36 00 37 00 49 60 48 00 Volume 99 00 84 82 99 00 99 00 Density 0 36 0 44 0 50 0 48 45 Outer Steel Finless Due to the lack of fins and miss-weighting of the Outer Steel finless design the penetrator tumbled during deployment and continued in a flat spin throughout flight This lead to extremely minimal velocity upon impact as well as the penetrator landing flatly on its side No true crater was formed by this impact Figure 3 23 Out steel finless landing position Due to a flat spin the penetrator impacted on its side Firefly The smallest penetrator Firefly impacted at very low velocity The minimal mass of the penetrator led to a depth of only 10cm resulting in a depth to diameter ratio of roughly 2:1 The penetrator lost a fin during flight or impact and was not retrieved A second fin broke upon impact as seen in Figure 3 24 46 The impact created a cookie cutter-like snow core The core was extremely unconsolidated less so than the surrounding snow Because of this it may be possible that the material in the core may have ejected had the penetrator impacted at higher velocity However it is also possible that the material would have stagnated and been buried with the penetrator Figure 3 24 Impact of Firefly A: Depth of crater to end of penetrator only 3cm B: Crater with penetrator removed Figure 3 25 The core created by Firefly more similar to a cookie cutter 47 Crater Depth Penetrator Mass to Crater Depth 2 1 8 1 6 1 4 1 2 1 0 8 0 6 0 4 0 2 0 3 4 5 6 7 8 9 Penetrator Mass 10 11 12 13 Figure 3 26 Plot of penetrator mass to crater depth for the finned penetrators The more massive penetrators impacted deeper However a definitive relationship cannot be implied due to multiple untracked variables Crater Depth Penetrator Mass to Crater Depth 2 1 8 1 6 1 4 1 2 1 0 8 0 6 0 4 0 2 0 3 4 5 6 7 8 9 10 11 12 13 Penetrator Mass Figure 3 27 Plot of impactor to snow density Chute refers to the path the penetrator rock traveled through the snow The sub-impactor density for the rocks was much higher than that of the penetrators 48 3 3 3 Field Test Results and Summary The field test of the EKIP system had multiple objectives with focus on both science and engineering goals The engineering goal to test the effectiveness of different penetrator designs was not truly completed with this field test Due to low velocity impacts as well as the low hardness of the substrate impacted a plume was not created Future field tests at high velocities will be needed to fully test this component The science objective to use crater morphology of penetrator craters to create an empirical curve was moderately successful With a low hardness the impact into snow will be the lower end member on a hardness constraining curve However a high velocity impact into snow would yield more accurate data for future studies The characteristic morphology of the penetrator craters into snow together with further data points into harder substrates with future field tests will allow a reasonably accurate extrapolation to apply to an actual impact on Europa 49 Chapter 4 SECOND STAGE The EKIP mission s goal is to successfully land a scientific payload onto the surface of Europa: That scientific payload necessitates the engineering of a second stage robust enough in design to survive the descent through the plume and the impact while protecting the instruments inside The electronics payload is yet to be determined but may include motion sense microscopes seismometers or magnetometers Due to the mass restrictions accorded to space travel the second stage must be both robust in strength and light in mass The second stage must withstand high velocity debris as it interacts with the plume This field test was intended to determine whether the chosen design which was compact and light was able to deploy from the first stage and to determine if the flight was stable Future testing would be necessary to refine designs to increase mechanical strength 4 1 4 1 1 SECOND STAGE DESIGN Birdie Design The penetrator is projected to impact the surface at roughly 4 km s Winglee 2018 with the second stage hopefully slowing to 1km s upon impact To achieve a momentum-transfer of 3 km s or more from the ejecta to the second stage a large surface area is needed to interact with the ejecta The entire second stage needs to remain as light as possible to decrease the momentum needed to slow it down The combination of these two requirements a large surface area with minimized mass led to two competing designs for the second stage 50 The first design resembled an umbrella attached to the end of a rocket The arms made of aluminum were extremely robust however the arms were minimal in length to conserve an already considerable mass measuring 1 3ft2 in surface area To balance the mass contained at the rear of the lander the nose of the lander was weighted with lead The overall mass of the lander sans electronic payload weighed 17lbs The large mass combined with the minimal surface area caused concern The second and working design was adapted to mimic the flight dynamics of a badminton shuttlecock or birdies This design allows for a large lightweight and stable lander Due to the shape of the fins the birdie requires little to no mass to fly stably After construction the average birdie weighed 2lbs with a surface area of 18ft2 Springs connecting the two metal bulkheads exert a constant pull to open the birdie To stow the birdie the springs must be fully stretched which may fatigue the springs over long duration stowage Figure 4 1 Birdie design without chute fabric A: Design mid-way open B: Arms stowed 51 Figure 4 2 Birdie design with chute fabric Penetrator shown attached to front electronics to the top A: Design mid-way open B: Arms stowed 4 1 2 Construction To reduce mass while ensuring the survivability of the lander material selection was put under much consideration For an actual lander meant to land on an icy moon such as Europa the skeleton would need to be a strong lightweight metal such as titanium The chute for the actual lander would need to be highly shock resistant and prevent tearing leading to the use of a Kevlar weave fabric However due to the cost of such materials the prototypes created and tested by our lab were modified To keep mass down the skeleton of the birdie was a mix of stainless steel and aluminum The arms were made of extruded carbon fiber with the chute being made of P1 parachute material Though the P1 material is not tear resistant enough to survive impacts of ejecta faster than 1 km s it would be suitable for the low velocity test 52 The skeleton was machined in house while the chute was sewn by a volunteer parachute rigger Two prototype miniaturized chutes were sewn one later becoming an actual prototype used for testing Tinkerbell Following the miniaturized test chutes three full scale chutes were sewn at varying lengths An electronics package was developed for two birdies with the intention to control separation and deployment timing and altitude The electronics consisted of a timer that initiates the burn of a blast charge or pack of black powder meant to separate the stages The timer is started by the removal of a pull pin switch to ensure no ignition could occur in the helicopter 4 2 BIRDIES Individual reports on each birdie tested are outlined below A consolidated table of characteristics for each tested birdie is below Table 4 7 Birdie Attribute Table Birdie Color Mass kg Surface Area m2 Stowed Length cm Electronics Penetrator Pairing Spring Force Tinkerbell Dark Green 0 13 0 44 61 No Firefly 50 2 springs Flora Purple 0 23 1 9 107 No Slim Jim 90 3 springs Fauna Bright Green 0 18 1 5 91 Yes Off Long John 90 3 springs Merriweather Orange 0 16 0 7 76 Yes On Stumpy 60 2 springs 4 2 1 Birdie 1- Tinkerbell Tinkerbell the smallest of the birdies to be tested measured 76cm stowed and had a total surface area of 0 44 m2 The skeleton was made of stainless steel and aluminum with a zeroporosity 0-P nylon rip-stop parachute material used for the chute This chute was sewn as a 53 prototype design to determine the process for sewing the three larger chutes It was later decided to use Tinkerbell during the actual test for an additional data point Figure 4 3 Tinkerbell chute A: Shown being stowed by hand B: Chute open and upside down Tinkerbell was attached to the penetrator Firefly through a friction fit The lander combination was dropped from roughly 760 meters above the surface of the glacier As seen in Figure 4 4A the chute failed to separate from the penetrator during the drop This is believed to be due to the minimal mass which didn t allow the lander combination to reach adequate velocity to create enough pressure to overcome the friction fit When removed from the penetrator as seen in Figure 4 4B Tinkerbell s chute was slightly damaged The carbon fiber rods ending the skeleton s arms poked through the ribbing material edging the parachute material There was no further damage to the chute The skeleton was undamaged by the impact 54 Figure 4 4 Tinkerbell impact A: Untouched landing position B: Chute when removed from penetrator 4 2 2 Birdie 2 Flora Flora s skeleton is made of aluminum stainless steel and pultruded carbon fiber tubes The carbon fiber tubes were filled with an extruded carbon fiber rod to act as a stabilizing core The chute was sewn using two-porosity 2P nylon rip-stop parachute material The higher porosity material in comparison to Tinkerbell s chute did not affect any material properties and was used purely for price efficiency The birdie had a total mass of 0 23kg Flora was paired with the penetrator Slim Jim The system was the second combined system to be dropped from the helicopter directly following the drop of Tinkerbell Firefly The drop from the helicopter went without issue until the separation of the birdie from the penetrator 55 at which time the chute tore from the skeleton This was due to thin nylon thread being used to attach the chute to the body in an attempt to lessen the possibility of binding the hinges of the skeleton The birdie skeleton fell straight down after separation from the chute with visual reference being kept from the helicopter until impact The birdie skeleton was not affected by cross winds as the following birdies were This is most likely due to the lack of available surface area for wind to push against once the chute was stripped Upon impact as shown in Figure 4 6 the birdie skeleton was undamaged Flora is estimated to have reached to highest velocity of the birdies due to the less of air resistance due to a lack of chute Figure 4 5 Flora undisturbed impact orientation Due to minimal air resistance the birdie fell straight down and impacted perpendicular to the glacier surface 56 Figure 4 6 Close-up of Flora s Impact A: Over compression of the springs caused the springs to bend Otherwise no damage occurred to the skeleton B: Snow imprint from Flora s skeleton Based off small indentations one arm bounced once before spreading to a larger angle than designed 4 2 3 Birdie 3 Fauna The construction of Fauna was identical to that of Flora The birdie had a total mass of 0 18 kg This lander combination was the only set to test an electronic separation The electronics primarily a timer ignited a black powder blast charge 12 seconds after being dropped from the helicopter The blast charge removed a plastic cap from the opening of the penetrator allowing air pressure to separate the birdie This configuration allowed for the birdie to remain stowed for a longer amount of time than the other birdies To avoid the chute stripping from the skeleton as happened with Flora 3mm parachute cord was used to tie the chute to the skeletons of both Fauna and Merriweather With the additional ties the chute remained attached during flight Due to the design of the birdie to point into the path of most resistance the birdie flew sideways when interacting with the wind during 57 flight A distinctive scallop shaped flight path tracing the wind patterns occurred when the birdie entered the two separate wind bands as illustrated in Figure 4 9 Figure 4 7 Full configuration of Fauna and Long John 58 Figure 4 8 Fauna landing position A: Viewed from the south B: Viewed from the north 59 Figure 4 9 Trajectory path of birdies through wind bands Colored bands show wind bands corresponding color arrow shows wind direction and magnitude Dotted line shows trajectory without wind interaction Dashed line shows trajectory of birdie steering into oncoming wind 4 2 4 Birdie 4 Merriweather The construction was identical to that of Flora The birdie had a total mass of 0 16kg This lander configuration was originally set up to run an electronic separation payload identical to that of Fauna however a friction separation was decided upon flight preparation to avoid multiple deployment failures if the electronics failed Due to the mass differences between the highly massive penetrator and relatively light birdie separation occurred almost instantaneously upon release from the helicopter Merriweather also contained only two springs in comparison to the other large birdies containing three These two factors led to Merriweather only partially deploying during flight Even with partial deployment Merriweather displayed the same scalloping descent while encountering the wind bands though to a less extreme degree than Fauna Figure 4 10 Merriweather impact viewed from east A and south B 60 4 3 RESULTS AND FUTURE WORK Overall each birdie both failed and succeeded in different ways Though Flora s chute stripped it is now known how the birdies would react without air resistance during flight Fauna separated when the timer was set though it was decided an altitude based deployment system would allow for better chute interaction with possible plumes Merriweather exemplified the need for strong resistance to ejecta in this case represented by air resistance The most pressing result of this field test of the second stage was determining the flight stability of the second stage after deployment as well as the means of deployment The design avoided tumbling during deployment and flight but failed to open fully in face of high velocity A chute deployment with higher spring-force may improve this problem For the deployment technique the timer worked suitably well though the blast charge must be replaced with another release mechanism Future designs of the birdies will focus on more robust skeletons as well as deployment electronics For birdies truly intended for interaction with ice plumes Kevlar chutes would be needed instead of nylon The carbon fiber skeletons though light proved to brittle to sustain impacts with high velocity ejecta A stronger alternative would be needed Chapter 5 PROPOSED RESEARCH 5 1 ALASKAN GLACIER TEST A second future trip to the Juneau Icefield is in planning for August 2018 This test will occur on the Gilkey Glacier in an attempt to impact ice instead of seasonal snowpack This will also be 61 done with rocket propulsion instead of static drops from a helicopter with the intent to reach higher velocities Sugar motors are being created which in contrast to the regular rocket motors used contain no ammonium perchlorate This chemical is toxic and will be avoided to not allowed contamination of the glacier 5 2 ALASKAN SEA ICE TEST A further impact test is planned for March 2018 to Barrow AK to impact sea ice at high velocities This test is intended to create ice plumes in a more controllable situation than on the Juneau Icefield Additionally sea ice is less dense than blue glacier ice giving an additional data point of substrate relative hardness 5 3 BIRDIE DEVELOPMENT Additional to the improvements mentioned in section 4 3 further development of the second stage is being conducted An energy absorption material Truitt 2016 modified and contained within the second stage to protect electronics at high velocity impacts is currently being developed The design will be tested with small scale impacts similar to the initial ice penetrator laboratory testing 5 4 BIOLOGICAL CAPTURE AND DETECTION Further testing of the biology shots will be continued with a focus on microbial damage within impact zones Caution will be taken to minimize contamination during these experiments An additional test at larger scale will be conducted in Barrow AK during the planned sea-ice field test Trapped algae within the sea ice will be analyzed optically to determine the effect of a supersonic impact in the surrounding ice 62 BIBLIOGRAPHY Collinson G Planetary Penetrators - The Vanguard for the Future Exploration of the Solar System Journal of the British Interplanetary Society 61 2008 : 198-202 Web Gowen R A Penetrators for in Situ Subsurface Investigations of Europa Elsevier 48 4 2011 : 725-42 Advances in Space Research Pergamon 19 June 2010 Web 27 Dec 2017 Hertzberg A A P Bruckner and D W Bogdanoff Ram Accelerator - A New Chemical Method for Accelerating Projectilesto Ultrahigh Velocities AIAA Journal 26 2 1988 : 195-203 Web Mastrapa R M E H Glanzberg J N Head H J Melosh and W L Nicholson Survival of Bacteria Exposed to Extreme Acceleration: Implications for Panspermia Earth and Planetary Science Letters Elsevier 14 June 2001 Web 28 Dec 2017 Mykytczuk Nadia Bacterial Growth at 15 C Molecular Insights from the Permafrost Bacterium Planococcus Halocryophilus Or1 Medscape Log In The ISME Journal 07 Feb 2013 Web 28 Dec 2017 Osinski Gordon R and Elisabetta Pierazzo Impact Cratering: Processes and Products Hoboken NJ: John Wiley & Sons 2013 Print Paranicas C J M Ratliff B H Mauk C Cohen and R E Johnson The Ion Environment near Europa and Its Role in Surface Energetics Geophysical Research Letters 29 5 2002 : n pag Web Robinson Tessa Robert M Winglee and Carl Knowlen Europa Kinetic Ice Penetrator System for Hyper Velocity Instrument Deposition Thesis University of Washington 2017 N d N p : n p n d Print 63 Spencer J R L K Tamppari T Z Martin and L D Travis Emperatures on Europa from Galileo Photopolarimeter-Radiometer: Nighttime Thermal Anomalies Science284 5419 1988 : 1514-516 Web Truitt Chad Planetary Penetrators for Sample Return Missions Thesis University of Washington 2016 N p : n p n d Print Winglee R M T Robinson M Danner and J Koch Cryo-braking Using Penetrators for Enhanced Capabilities for the Potential Landing of Payloads on Icy Solar System Objects Acta Astronautica 144 2018 : 136-46 Web Winglee R m C Truitt and R Shibata High Velocity Penetrators Used a Potential Means for Attaining Core Sample for Airless Solar System Objects Acta Astronautica 137 2017 : 274-86 Web Wunnemann K and B A Ivanov Numerical Modelling of the Impact Crater Depth diameter Dependence in an Acoustically Fluidized Target Planetary and Space Science Pergamon 04 Oct 2003 Web 28 Dec 2017 64 APPENDIX A Past and Pending Missions to near Europa or Icy Bodies: Mission Launch Year Decommissioned Pioneer 10 & 11 1973 -- Galileo 1989 -- Galileo Europa Mission GEM 1997 -- Rosetta 2004 2014 Impact Philae Juno 2011 -- Europa Clipper Orbiter & Lander 2020 s -- Denotes future pending mission VITA A short bio of the author is required for a Ph D dissertation at the University of Washington The vita section does not go into the Table of Contents The formatting style follows the text of the dissertation
    • Hart, Chloe - Ph.D. Dissertation
      Thermodynamics of Acidophiles: Energetics of microbial growth, response to substrate availability, and interactions with heavy metals 2018, Hart,Chloe,Chloe Hart c Copyright 2018 Chloe E Hart Thermodynamics of Acidophiles: Energetics of microbial growth response to substrate availability and interactions with heavy metals Chloe E Hart A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy University of Washington 2018 Reading Committee: Drew Gorman-Lewis Ph D Chair Roger Buick Ph D John Baross Ph D Program Authorized to Offer Degree: Department of Earth and Space Sciences University of Washington Abstract Thermodynamics of Acidophiles: Energetics of microbial growth response to substrate availability and interactions with heavy metals Chloe E Hart Chair of the Supervisory Committee: Drew Gorman-Lewis Ph D Department of Earth and Space Sciences Chemical and physical properties play a crucial role in the type and efficiency of microbial activity that can be supported within an ecosystem In return the presence of microbial activity affects the surrounding environment driving biogeochemical cycles altering chemical fluxes via metabolism and surface interactions and affecting the lithosphere through mineral precipitation or dilution Fluctuations in aqueous geochemistry however can alter growth efficiency and overall energetic demands of microorganisms This dissertation uses a thermodynamic approach to explore energy requirements of acidophilic microorganisms in response to chemical changes and how their cell surfaces interact with heavy metals Chapter 2 explores the effects of energy source availability on microbial energetics with the sulfuroxidizing Archaea Acidianus ambivalens A ambivalens in three decreasing concentrations of dissolved oxygen from aerobic to microaerobic The results show that growth proceeds most efficiently under low levels of oxygen while high levels of oxygen require the most energy likely due to higher maintenance energy demands as a result of oxidative stress Chapter 3 studies the energetic response to environmental redox conditions by quantifying bioenergetics of A ambivalens during anaerobic growth with H2 and sulfur A ambivalens growth was not affected by environmental oxidation state and shows similar growth efficiencies and energy budgets between anaerobic and microaerobic conditions However microaerobic growth required less overall Gibbs energy suggesting a slight preference for growth on sulfur and O2 in low-oxygen environments Chapter 4 characterizes the thermodynamics of growth for two mesophilic bacteria Acidithiobacillus ferrooxidans and Acidithiobacillus thiooxidans Energetics were determined during aerobic growth with Fe2 and oxygen for A ferrooxidans and with sulfur and oxygen for both A ferrooxidans and A thiooxidans A ferrooxidans grew most efficiently with sulfur and O2 indicating a preference over growth with Fe2 Energetics for all three sulfur oxidizers A ambivalens A ferrooxidans and A thiooxidans are compared and show a significant correlation between Gibbs energy consumed enthalpies of growth and biomass yield suggesting growth energetics are impacted more by chemistry changes via catabolism than biochemical differences between species In Chapter 5 we measure cadmium adsorption to the thermoacidophile Sulfolobus acidocaldarius cell surfaces and provide the first thermodynamic description of surface adsorption of Cd by S acidocaldarius These results will not only help us understand interactions between natural environments and microbial activity but will also help evaluate habitability of environments and energy available for life elsewhere TABLE OF CONTENTS Page List of Figures iv List of Tables vi Chapter 1: Introduction 1 1 Organization of Dissertation 1 2 Motivation 1 1 1 Chapter 2: Energetics of Acidianus ambivalens in response to oxygen availability 2 1 Introduction 2 2 Materials and Methods 2 2 1 Maintenance culture procedures 2 2 2 Growth experiments 2 2 3 Calorimetric procedures 2 2 4 Chemical measurements 2 2 5 Overall growth modeling 2 2 6 Gibbs energy consumption 2 3 Results 2 3 1 Calorimetric results 2 3 2 Culture growth results 2 3 3 Thermodynamic components of growth 2 3 4 Response to carbon and oxygen availability 2 4 Discussion 2 4 1 Energetic costs for life 2 4 2 Energetic demands in aerobic and microaerobic conditions 2 4 3 Thermodynamics of microbial growth 2 5 Conclusion i 5 5 8 8 9 9 10 10 12 13 13 16 17 21 24 24 26 27 28 Chapter 3: 3 1 3 2 3 3 3 4 3 5 Energetics of A ambivalens during anaerobic fects of oxidation state on growth Introduction Methods 3 2 1 Culture maintenance 3 2 2 Cell growth 3 2 3 Calorimetric procedures 3 2 4 Chemical measurements 3 2 5 Overall growth modeling 3 2 6 Gibbs energy consumption Results 3 3 1 Calorimetry results 3 3 2 Thermodynamic Description of Growth Discussion 3 4 1 Anaerobic and Aerobic Growth Energetics 3 4 2 Power Consumption and Energy Budgets 3 4 3 Energetic Costs for A ambivalens Conclusion Chapter 4: Energetics of Acidithiobacillus spp during 4 1 Introduction 4 2 Methods 4 2 1 Culture maintenance 4 2 2 Calorimetric experiments 4 2 3 Chemical measurements 4 2 4 Overall growth modeling 4 2 5 Gibbs energy consumption 4 3 Results 4 3 1 A ferrooxidans Fe II oxidation 4 3 2 A ferrooxidans S oxidation 4 3 3 A thiooxidans S oxidation 4 4 Discussion 4 4 1 Iron oxidation by A ferrooxidans ii sulfur reduction and sulfur and ef 30 30 32 32 32 32 33 33 35 36 36 38 41 41 45 47 48 iron II oxidation 49 49 51 51 51 52 53 54 56 56 60 63 67 67 67 71 72 74 Thermodynamic characterization of cadmium adsorption onto cells of the thermoacidophile Sulfolobus acidocaldarius Introduction Materials and Methods 5 2 1 Cell Growth 5 2 2 Adsorption Experiments 5 2 3 Desorption Experiments 5 2 4 Calorimetric Experiments 5 2 5 Derivation of model parameters Results and Discussion 5 3 1 Cd Adsorption 5 3 2 Surface Complexation Modeling 5 3 3 Enthalpies of Cd Adsorption 5 3 4 Temperature Dependence Conclusions 75 75 77 77 77 78 78 79 82 82 82 85 87 90 Chapter 6: Conclusions 6 0 1 Summary of work 91 91 Bibliography 94 4 5 4 4 2 Substrate utilization A ferrooxidans 4 4 3 Open and closed system growth 4 4 4 Sulfur oxidation energetics Conclusion Chapter 5: 5 1 5 2 5 3 5 4 Appendix A: Supplementary Materials A 1 Chapter 2 A 2 Chapter 4 A 3 Chapter 5 112 112 116 119 Appendix B: Curriculum vitae 123 iii LIST OF FIGURES Figure Number Page 2 1 Example heat curves for each of the three growth conditions 14 2 2 Typical heat flow curve for A ambivalens and growth curve 15 2 3 Measured enthalpies of growth and calculated Gibbs energy consumed 19 2 4 Calculated Gibbs energy consumption and heat produced during growth compared to standard state Gibbs energy and enthalpy of the overall growth reaction 20 2 5 Gibbs energy consumed compared to initial dissolved oxygen availability 22 2 6 Gibbs energy consumed compared to initial dissolved CO2 availability 23 3 1 Calorimetric heat flow curves for all Acidianus ambivalens replicates during microbial growth with H2 S 37 Gibbs energy consumed and enthalpies of growth for A ambivalens growth with H2 S compared to standard state 40 Energetics of aerobic microaerobic and anaerobic growth normalized to moles of electrons transferred 44 Biomass produced and power consumed during microbial growth for A ambivalens under anaerobic microaerobic and aerobic conditions 46 Example heat flow curve and growth curve of Acidithiobacillus ferrooxidans during iron oxidation 57 4 2 Energetics of Acidithiobacillus ferrooxidans during iron II oxidation 59 4 3 Energetics of Acidithiobacillus ferrooxidans during growth on sulfur 62 4 4 Energetics of growth for Acidithiobacillus thiooxidans during sulfur oxidation 66 4 5 Combined energetics data from Acidithiobacillus ferrooxidans and Acidithiobacillus thiooxidans normalized per electron mole 69 4 6 Total cells produced and power consumed for all growth results 70 4 7 Combined sulfur oxidation energetics from mesophilic Bacteria Acidithiobacillus ferrooxidans and Acidithiobacillus thiooxidans and thermoacidophilic Archaea Acidianus ambivalens 73 3 2 3 3 3 4 4 1 iv 5 1 5 2 5 3 Cd adsorption by Sulfolobus acidocaldarius as a function of pH with surface complexation model results Raw data from isothermal calorimetry and corrected heat with surface complexation modeling fits Cd partitioning onto S acidocaldarius at 25 C and 75 C A 1 Summary of heat flow curves produced by Acidianus ambivalens during sulfur oxidation under recommended growth conditions A 2 Summary of heat flow curves produced by Acidianus ambivalens during sulfur oxidation under oxygen-limited growth conditions A 3 Summary of heat flow curves produced by Acidianus ambivalens during sulfur oxidation under oxygen- and carbon-limited growth conditions A 4 Summary of heat flows produced by Acidithiobacillus ferrooxidans during iron oxidation A 5 Calorimetric heat flow for replicates of Acidithiobacillus ferrooxidans during sulfur oxidation A 6 Heat flow recorded by calorimetry for Acidithiobacillus thiooxidans during sulfur oxidation v 83 88 89 113 114 115 116 117 118 LIST OF TABLES Table Number 2 1 Page Summary of data collected calculated for each replicate of the three tested oxygen regimes for Acidianus ambivalens 18 Acidianus ambivalens growth with H2 S Average growth data and energetic results for Acidianus ambivalens under aerobic and microaerobic conditions with S O2 42 4 1 4 2 4 3 Acidithiobacillus ferrooxidans growth on Fe II O2 Acidithiobacillus ferrooxidans growth on S O2 Acidithiobacillus thiooxidans growth on S O2 58 61 64 5 1 5 2 Sulfolobus acidocaldarius Model Parameters 79 84 3 1 3 2 vi 39 ACKNOWLEDGMENTS I would like to thank all the faculty and staff in the Department of Earth and Space Sciences and the Astrobiology program Without your dedication and support this degree would not have been possible The faculty I had the pleasure of interacting with helped me look at my research from fresh perspectives and expand my love for science I would also like to thank my brilliant committee members Drew Gorman-Lewis Roger Buick John Leigh and John Baross for your stimulating lectures helpful guidance and feedback along the way I especially would like to thank my advisor Drew Gorman-Lewis Thank you for taking me on as a graduate student and allowing me to pursue my research interests Through the chaos whirlwind of worry and broken electrodes I greatly appreciated the calm and balance you brought to every situation I am beyond grateful for the teaching outreach mentoring and research opportunities you ve provided or encouraged and helping me grow as a scientist Additionally I have to thank my friends and family To my friends in ESS and Astrobiology you became a fantastic support system both in your academic and intellectual support and reminding me to leave the lab every once in a while I am also tremendously grateful to my family back home I appreciate your support and love through this entire journey Thank you for believing in me and for all the laughter we share every time we are together To my parents thank you for all your inspiration To both you have taught me to find humor and happiness in good times and in bad though I am still need reminding not to sweat the small stuff To my mother Tammy you have taught me tenacity independence and strength To my father Joe I can t thank you enough for fueling my first spark of interest in microbiology and environmental science Along the way you have also taught me vii how to work hard play hard and tell all the worst jokes And to Zachary Thank you for your never-ending support over the years I don t know what I would do without your friendship and love Thank you for putting a smile on my face when I need it most and pushing me to be my best self You are the best cheerleader even 2400 miles away I look forward to our future adventures together viii DEDICATION To my wonderful parents Joe & Tammy for their love support and endless laughter ix 1 Chapter 1 INTRODUCTION 1 1 Organization of Dissertation The introduction chapter provides a brief overview of the motivation for this dissertation work Chapters 2-5 have been written as stand-alone papers Chapter 2 Energetics of Acidianus ambivalens in response to oxygen availability was submitted to the journal Geobiology in October 2017 declined with encouragement to resubmit and will be resubmitted in June 2018 Chapter 3 Energetics of A ambivalens during anaerobic sulfur reduction and effects of oxidation state on growth will be submitted to the journal Geomicrobiology Chapter 4 Acidithiobacillus spp growth energetics during iron II and sulfur oxidation will be submitted to Geochimica et Cosmochimica Acta and Chapter 5 Surface complexation of cadmium by thermoacidophile Sulfolobus acidocaldarius will be submitted to the journal Chemical Geology A compiled bibliography for all chapters is located at the end of the dissertation 1 2 Motivation In geothermal and acidic environments geochemistry plays a vital role in determining the type and abundance of microbial activity Extreme chemolithoautotrophs grow and thrive in these harsh conditions through energy-yielding reduction-oxidation reactions with inorganic substrates i e sulfur iron hydrogen All life requires energy to drive reactions such as growth and cell maintenance therefore fluctuations in the geochemistry can alter microbial activity by affecting the availability of energy sources nutrients oxidation state of the environment or introducing harmful contaminants As microorganisms are ubiquitous on Earth and are key players in biogeochemical cycles it is important to understand the energy 2 requirements of microbial growth and the factors that can alter overall energetics and growth efficiency Previous energetic research has focused primarily on the amount of Gibbs energy required and or growth efficiency for heterotrophs and methanogens Therefore there is a gap in our knowledge concerning the energetics of extreme chemolithoautotrophs and their response to environmental factors Quantifying the energy necessary for microbial growth in extreme environments helps establish the chemical and physical limits to life Chemolithoautotrophic extremophiles are considered the most relevant analogs for life elsewhere in the Solar System One strategy to evaluate habitability of other planets and moons is to Follow the energy Following the energy is driven by the notion that all life requires energy thus the objective is to search for environments with chemical disequilibrium that could sustain life Information from extraterrestrial environments can be acquired through remote sensing in situ measurements with rovers landers or in the future sample return missions This can help us determine the physical and chemical conditions e g temperature oxidation state energy sources and could allow us to estimate the amount of energy available and type of metabolisms that could be supported there The gap in our knowledge concerning energetics of thermophiles and acidophiles and their astrobiological significance make these species ideal subjects for energetics research This dissertation takes a geochemical and thermodynamic approach to understanding microbial energetics of acidophilic chemolithoautotrophs and their surface interactions with heavy metals In the following chapters I characterize microbial growth for different acidophilic Archaea and Bacteria to determine how growth varies based on chemical properties of the environment catabolic reactions used to drive growth species-specific traits that affect maintenance energy requirements and how microbial surfaces interact with metals found in acidic waters In Chapter 2 I quantified growth energetics in terms of Gibbs energy consumed G and enthalpy produced H for the thermoacidophile Acidianus ambivalens A ambivalens during aerobic sulfur oxidation Microbial growth modeled using macrochemical equations 3 to describe overall metabolism required the most Gibbs energy under high oxygen concentrations The large energetic demand was accompanied by low growth efficiencies in terms of biomass yield Energetic demand decreased and growth efficiency increased as oxygen became more limited These findings suggest high oxygen concentrations increase maintenance energy requirements and as a result overall energetic demand for A ambivalens growth through oxidative stresses In Chapter 3 energetics of A ambivalens was characterized under anaerobic conditions the oxidation of hydrogen with sulfur serving as the energy-yielding catabolic reaction Overall metabolic efficiency during growth under anaerobic conditions was compared to growth by sulfur oxidation in aerobic and microaerobic conditions Chapter 2 to investigate the effects of environmental redox state on energetics Despite the lower energy requirements to synthesize biomolecules in anaerobic environments anaerobic growth metabolizing H2 S yielded similar efficiencies as microaerobic growth metabolizing S O2 However normalized to moles of electrons transferred microaerobic growth consumed less Gibbs energy than anaerobic and fully aerobic growth suggesting a slight preference for growth with O2 and sulfur in low-oxygen environments Chapter 4 characterized the thermodynamics of growth of two mesophilic bacteria Acidithiobacillus ferrooxidans and Acidithiobacillus thiooxidans Energetics were determined during aerobic growth with Fe2 and oxygen for A ferrooxidans and with sulfur and oxygen for both A ferrooxidans and A thiooxidans over a range of growth efficiencies A ferrooxidans required less Gibbs energy during growth and produced high biomass yields when grown with sulfur compared to growth with Fe2 indicating a preference for the higher energy-yielding sulfur oxidation reaction Microbial energetics for all three sulfur oxidizers A ambivalens A ferrooxidans and A thiooxidans were compared resulting in the potential to predict growth energetics for other sulfur-oxidizing microorganisms In Chapter 5 we studied the interactions between cell surfaces of the thermoacidophile Sulfolobus acidocaldarius and cadmium a highly toxic heavy metal Surface adsorption of Cd was determined as a function of pH and for different biomass concentrations demonstrat- 4 ing Cd readily adsorbed to the cells Using surface complexation modeling and calorimetry I provided the first thermodynamic description of surface adsorption of Cd by S acidocaldarius By characterizing microbial growth for a variety of organisms and metabolic strategies under different growth conditions I hope to understand how and to what extent different environmental factors can affect overall energetics This research will help us further understand the complex relationship between life and energy and help identify suitable habitable environments in the solar system and beyond 5 Chapter 2 ENERGETICS OF ACIDIANUS AMBIVALENS IN RESPONSE TO OXYGEN AVAILABILITY This manuscript will be submitted to the journal Geobiology Co-authored by Chloe E Hart Drew Gorman-Lewis 2 1 Introduction Microorganisms are ubiquitous on Earth and key players in biogeochemical cycles Even in the harshest environments we find microorganisms making a living on chemical disequilibrium e g 84 78 155 In geothermal regions such as Yellowstone National Park YNP we find life built to withstand elevated temperatures 70 C low pH 3 toxic gases i e H2 S and harsh metal loid s 124 Chemotrophic microorganisms dominate the hottest regions in YNP above the temperature limit for photosynthetic life 73 C 87 169 2 Their metabolic activity is fundamental for cycling essential elements such as sulfur and iron 24 93 Many case studies have been conducted in YNP to identify the inorganic energy sources and determine how much Gibbs energy G is available to support chemolithotrophic life e g 113 148 109 79 140 Inorganic sulfur compounds are common in acidic geothermal regions 176 16 15 and thus play an essential role in microbial metabolisms 24 102 One of the highest energy yielding reactions available in acidic hot springs is aerobic sulfur oxidation which produces sulfuric acid 140 Oxygen availability can be naturally low in hot spring environments due to the decrease in solubility at elevated temperatures and chemical data has shown dissolved oxygen can vary both spatially within a geothermal setting e g 176 123 35 75 and within a hot spring environment mm- m scale 23 Changes in 6 oxygen availability consequently affects the amount of potential Gibbs energy available for microorganisms requiring oxygen as a terminal electron acceptor during metabolism Because of this available Gibbs energy variability laboratory-based studies are necessary to better understand microbial energetic demands and to what extent environmental conditions may affect them After temperature and pH e g 47 energy source availability is a major factor on both the abundance and distribution of life in an ecosystem 154 3 33 Energy source limitation can impact microbial growth beyond the amount of biomass produced For example mesophilic heterotrophs have shown an increase in growth efficiency in terms of cells produced per amount of substrate consumed as a response to carbon limitations 133 39 52 Fluctuations in growth efficiency impacts the environment by altering the extent of Gibbs energy consumption and chemical fluxes influenced by metabolism While sulfur metabolisms have been widely investigated in many organisms e g 24 102 we lack a thermodynamic understanding of energy usage under extreme conditions in response to fluctuations of energy sources Toward this end we investigated aerobic sulfur oxidation by Acidianus ambivalens A ambivalens under various oxygen levels and characterized Gibbs energy consumption G and enthalpy of growth H A ambivalens is an ideal microorganism to investigate aerobic sulfur oxidation A ambivalens is a crenarchaeon that grows optimally between 75-85 C in acidic pH 2 5 environments 180 181 51 A ambivalens is metabolically versatile and can grow anaerobically by H2 oxidation with sulfur H2 oxidation with Fe3 and sulfur oxidation with Fe3 or under aerobic conditions with oxygen as described by the reaction S 1 5 O2 H2 O 2 H SO4 2 180 181 51 91 4 A ambivalens can fix inorganic carbon through the 3-hydroxypropionate 4-hydroxybutyrate pathway 22 though yeast extract can also be used to stimulate growth 181 Multiple investigators have studied molecular and genetic aspects of metabolism in A ambivalens and the species is regarded as a well-studied crenarchaeon 180 181 51 77 97 However we lack comprehensive studies of its energetics under varying energy availability 7 Researchers have taken various approaches to quantify microbial energy usage both in laboratory settings and in natural environments In a laboratory setting G consumed during microbial growth can be measured typically based on the catabolic reaction of the organism in question to quantify energetic budgets for specific microbes e g 147 4 Computationally energetic demands can also be calculated in terms of G necessary for biosynthesis e g 5 7 108 6 While several investigations focus on energetic demands of microbial growth in terms of G it does not completely describe the energetics of growth As shown in Equation 2 1 Gibbs energy is dissipated as enthalpy H and entropy S T is absolute temperature G H T S 2 1 Regardless of metabolic pathway a portion of Gibbs energy is dissipated as heat H 0 or heat could be adsorbed H 0 Liu et al 2001 Consequently to thoroughly characterize energy usage of microbes it is necessary to quantify G consumption during growth in addition to total enthalpy via calorimetry Some research has included calorimetry as a tool to measure total microbial activity in terms of heat produced e g 158 18 175 25 111 131 but few studies have fully characterized energetics of individual species in terms of both G and H Calorimetry is the ideal means to measure the enthalpy of growth at constant temperature and pressure The focus of this work was to determine how oxygen availability affects Gibbs energy consumption by A ambivalens and how that Gibbs energy is dissipated We merged two approaches to quantify microbial growth energetics The first approach largely used in biotechnology fields models both catabolic and anabolic reactions of metabolism to create an overall growth reaction 162 164 68 20 The production of biomass is treated as a basic chemical reaction where the metabolic reactants progress irreversibly to the metabolic products and new cell material The second approach proposed by Smith and Shock 147 quantified Gibbs energy consumption through the change in catabolic chemical species in the medium Both catabolic and biosynthetic reactions influence the change in chemical 8 composition of the medium Therefore it is assumed anabolic reactions were inherently taken into account with this approach even though they were not explicitly used in the calculations We used our combined model of overall microbial growth in addition to chemical composition data to determine Gibbs energy consumption of cultures Calorimetric measurements of culture growth allowed us to determine the total enthalpy of growth By working with a pure culture in controlled growth conditions we can isolate the energetic effects due primarily to the availability of oxygen The results show that G consumption and H produced per C-mole biomass are significantly reduced as oxygen availability decreases and accompany an increase in biomass yield This characterization provides a fundamental energetic description of A ambivalens growth that is essential for understanding how geochemical factors impact energy usage 2 2 2 2 1 Materials and Methods Maintenance culture procedures Maintenance cultures of A ambivalens ATCC 49204 were grown according to growth conditions described by Zillig et al 181 using Brocks basal salt medium 26 supplemented with additional KH2 PO4 0 02 M final concentration absence of yeast extract and medium adjusted to pH 2 8 with H2 SO4 prior to autoclaving Sulfur powder was sterilized through three sequential autoclave cycles at 100 C for 30 minutes Maintenance cultures were prepared aseptically in 100 mL septa vials with 50 mL growth medium 2 g L sterile sulfur powder and 2% v v cell mixture inoculation Once sealed with butyl rubber stoppers and crimped with aluminum seals vials were injected with 100% CO2 gas for a final concentration of 10% CO2 in the headspace and incubated at 76 C for 6-7 days 9 2 2 2 Growth experiments Three distinct growth conditions were investigated designated by the initial amount of oxygen provided In the recommended growth RG experiments culture procedures were based on published literature and were identical to the maintenance culture procedures described above The aim was for initial oxygen concentrations in the medium to be near saturation 200 M at 76 C Oxygen-limited OL experiments were prepared following the maintenance culture protocol but the headspace was purged with 100% CO2 gas to remove excess O2 in the headspace yet provide ample inorganic carbon for biomass production The design aimed to provide an initial dissolved oxygen concentration of 50% saturation 100 M Lastly the design for oxygen- and CO2 -limited OCL experiments was constructed to provide an initial dissolved oxygen concentration of 25% saturation 50 M The design was also aimed to test if growth efficiency in OL experiments were affected by the increased availability of inorganic carbon The OCL experiments were prepared following the maintenance culture protocol but the experiment vials were brought into an anaerobic chamber with a 99% N2 and 1% H2 atmosphere to seal the vials The final headspace was N2 injected with CO2 for a final concentration of 10% CO2 These steps were taken to further minimize O2 in the headspace and lose some additional O2 from the media during vacuuming and gas exchange in the airlock chamber Control vials were also prepared in this manner and brought up to growth temperature 76 C to verify there was minimal hydrogen gas contamination as A ambivalens can use H2 as an electron donor to reduce sulfur 2 2 3 Calorimetric procedures Enthalpies of growth H were measured with a TA Instruments TAM III Nanocalorimeter that measures heat flow between a reaction cell and a reference cell 80 165 The calorimeter heat flow response was calibrated by an electrical heating procedure verified by measuring the heat of protonation of trishydroxymethylaminomethane 61 All experiment cultures were grown in 4 mL Hastelloy calorimetric cells containing 2 5 10 mL total culture and 2 g L sterile sulfur Both reaction and reference Hastelloy cells were filled with identical media and headspace constituents The reaction cell contained an aliquot of A ambivalens to provide an initial cell density of 105 cells mL and the reference cell was kept sterile and void of cells The heat flow of the samples was monitored over time and integration of the heat flow signal after baseline correction resulted in the total heat produced during growth 2 2 4 Chemical measurements At the beginning and end of growth chemical data were collected to measure the change in components of the growth medium Sulfate concentrations were determined through Hach R TNTplus spectrophotometric sulfate test vials pH was measured with a combination microelectrode Microelectrodes Inc calibrated daily with four NIST standards and dissolved oxygen was measured with a Unisense Clark-type dissolved oxygen microsensor Both pH and dissolved oxygen measurements were taken at 76 C due to their sensitivity to temperature Slides for cell counts were prepared using a polycarbonate 0 2 m filter membrane with SYBR Green I dye following the procedure of Lunau et al 100 and examined with appropriate filters on a Zeiss Axiostar Plus microscope 2 2 5 Overall growth modeling Equations reported by Heijnen and Kleerebezem 68 Equations 2 2-2 7 below were applied to model the overall energetics using simplified macrochemical equations The equations are based on biosynthesis of one carbon-mole C-mol of generic biomass represented by the formula CH1 8 O0 5 N0 2 34 63 94 162 68 The biomass formula is comprised of the four major elements of cellular material and has a molar mass of 24 6 g C-mol Autotrophic organisms like A ambivalens synthesize biomass during the anabolic reaction represented 11 by Equation 2 2 a e donor b N source c H d H2 O e CO2 2 2 f oxidized e donor CH1 4 O0 4 N0 2 0 The Gibbs energy necessary to drive anabolism is generated through redox reactions during catabolism A generic catabolic reaction is shown in Equation 2 3 g e donor h e acceptor i oxidized e donor 2 3 j reduced e acceptor 0 Metabolic coefficients a through j were determined by solving a series of linear equations to satisfy mass charge and degree of reduction balances Supplementary material Reactants consumed during growth result in negative coefficients while products result in positive coefficients For aerobic growth on sulfur the derived anabolic reaction to produce 1 C-mol of generic biomass for A ambivalens becomes Equation 2 4: 0 7S 1CO2 0 2N H4 1 3H2 O 0 7SO42 1 6H 1CH1 8 O0 5 N0 2 0 2 4 The catabolic reaction for sulfur oxidation to sulfuric acid results in the following reaction displayed in Equation 2 5 1S 1 5O2 1H2 O 2H 1SO42 0 2 5 The catabolic multiplicative factor fcat Equation 2 6 represents the number of times the catabolic reaction is performed in relation to the anabolic reaction to produce 1 C-mol of biomass The catabolic factor is calculated from experimental data using the amount of biomass yield YX D defined as C-mol of biomass produced per mole electron donor consumed and the anabolic coefficient of the electron donor YDan or coefficient g in Equation 2 2 Biomass produced in C-mol was calculated by converting the number of cells to moles of carbon using the value 2 5 fmol-C cell the average carbon content per cell for A ambivalens performing sulfur reduction with hydrogen reported by Amenabar et al 4 via 13 C assimilation fcat 1 YDan YX D 2 6 12 The overall growth reaction coefficients are determined by combining fcat Equation 2 6 anabolic coefficients Equation 2 2 and the catabolic coefficients Equation 2 3 as shown in Equation 2 7 Overallgrowthreaction Anabolism fcat Catabolism 2 7 The microbial growth model established by Heijnen and Kleerebezem 68 was applied to growth of A ambivalens Catabolic factors were calculated for each experiment by using the biomass yield and stoichiometric coefficient of sulfur in the anabolic reaction Equations 2 6 and 2 4 to produce overall growth reactions Equation 2 7 for each experimental replicate The overall growth reactions were used for further thermodynamic calculations 2 2 6 Gibbs energy consumption Gibbs energy consumed during growth G was calculated for each experiment based on the activities of all the chemical species in the overall growth reaction and the stoichiometric coefficients of the overall growth reaction determined by the biomass yield in each replicate see above The standard state Gibbs energies of formation G f for the chemical species in the growth reaction were calculated at the experimental temperature with the revised Helgeson-Kirkham-Flowers HKF equations of state 72 71 using SUPCRT92 139 83 For biomass G f -67 kJ mol was used for all growth experiments 69 Standard state Gibbs energies of reactions G r of the overall growth reaction determined for each experiment were calculated using Equation 2 8 where G f products and G f reactants are the standard state Gibbs energies of formation for the products and reactants of the growth reaction respectively G r G f products G f reactants 2 8 Gibbs energy available for growth Gr is defined by Equation 2 9 where G r is modified with the concentration-dependent reaction quotient Q Gr G r 2 3026 R T logQ 2 9 13 The reaction quotient Q is calculated with the activities of the products and reactants involved in the overall growth equation and is shown in Equation 2 10 where aYi ogr denotes thermodynamic activity of the ith chemical species and Y ogr represents the stoichiometric reaction coefficient of the ith species in the overall growth reaction Equation 2 7 Q aYi ogr 2 10 Activities of the aqueous species were calculated using PHREEQC 119 and sulfur was assumed to have an activity of 1 Activity of biomass in solution was converted from molality C-mol kg solvent assuming an activity coefficient of 1 Overall Gibbs energy consumed G during growth for each experiment was calculated with Equation 2 11 as the difference between initial Gr available calculated from the initial chemical composition of the medium solution and biomass and final Gr available calculated from the final chemical composition of the medium solution and biomass G Gr initial Gr f inal 2 11 Microbial growth was ultimately evaluated in terms of G consumed and H produced both in kJ C-mol biomass 2 3 2 3 1 Results Calorimetric results A ambivalens produced clear heat flow signals during growth in all three oxygen conditions tested see Supplementary Material A typical heat flow signal over time for each oxygen regime is depicted in Figure 2 1 The heat flow signal paralleled the growth phase of the culture with lag exponential and stationary phase corresponding with the initial baseline exponential increase and peak heat flow signal and return to baseline respectively as shown in the example in Figure 2 2 RG produced the largest heat flow signals with a lag phase of approximately 67 hours before spending 53 hours in exponential phase The peak heat flows reached 33-48 W after approximately 113 hours of growth and returned to baseline after 14 50 Recommended growth O2 limited 45 40 O2 and CO2 limited Heat flow W 35 30 25 20 15 10 5 0 0 50 100 150 Time hr Figure 2 1: Example heat curves produced by the calorimeter during microbial growth for each of the three experimental conditions in terms of watts of heat over time in hours Total heat produced during growth in joules was calculated by integrating the area under each heat flow curve Watts over time in seconds Recommended growth conditions produced the largest heat signals red dotted line followed by oxygen-limited blue dashed line and the lowest heat signals were produced in the oxygen- and carbon-limited replicates green solid line Heat flow W 15 A 40 20 0 0 20 40 60 80 100 120 140 160 100 120 140 160 Time hr 10 Cells cells mL 4 6 B 3 2 1 0 0 20 40 60 80 Time hr Figure 2 2: A Typical heat curve for A ambivalens displaying heat W over growth time hours recorded by the calorimeter B Growth curve from external cultures monitored in tandem to calorimeter culture example depicting concentration of cells cells mL over the course of growth in hours Heat flow increases during exponential growth of the culture and returns to baseline as the culture reaches stationary phase Different symbols in growth curve represent cell counts from different replicates 16 119 hours OL growth conditions produced the second largest average heat signal The lag phase was shorter than RG with an approximate duration of 30 hours followed by 40 hours of exponential growth Peak heat flows ranged from 4 3 to 15 5 W and occurred after about 52 hours of growth before returning to baseline after 70 hours OCL experiments were the lowest heat signals but were similar to OL experiments in lag phase and total growth duration Lag phase occurred for 30 hours followed by 40 hours of exponential growth After approximately 59 hours peak heat flow occurred and reached 2-4 5 W Heat flow returned to baseline after approximately 72 hours 2 3 2 Culture growth results Data from each replicate of the experimental conditions are summarized in Table 2 1 Cultures grown under RG conditions consumed an average of 10 mol of sulfur during growth These conditions resulted in the most total biomass production approximately 2-fold greater than either oxygen-limited conditions The average biomass was calculated to be 1 18 10 8 C-mol and the biomass yield ranged from 0 0006 to 0 0017 C-mol S-mol OL growth conditions consumed much less sulfur with an average of 1 5 mol consumed during growth The average total biomass production was 6 93 10 9 C-mol and yields were higher than RG with a range of 0 0023 to 0 0132 C-mol S-mol Similar sulfur consumption and total biomass values were produced in OCL experiments as OL conditions The average sulfur consumption was 1 8 mol in OCL experiments and average total biomass produced was 5 61 10 9 C-mol Biomass yields ranged from 0 0014 to 0 0099 C-mol S-mol Cell morphology remained consistent between the different growth experiments there was no apparent expression of appendages or other changes to the cell surface across all replicates Cell diameters of the coccus cells were statistically similar across the different oxygen treatments P 0 1 for all t-tests The mean for RG N 69 cells were 1 06 0 26 m OL N 52 cell diameters were 1 01 0 19 m and OCL N 52 cell diameter mean was 1 04 0 18 m 17 2 3 3 Thermodynamic components of growth Modeling overall growth to determine biomass yield G consumed and H produced allows one to interpret energy efficiency and usage by microbes in a thermodynamic framework Biomass yield is a good indication of how closely the catabolic reaction is coupled to the anabolic reaction 133 68 A relatively higher biomass yield indicates more catabolic energy is conserved as new biomass as opposed to other cellular processes or non-growth reactions G and H both normalized to the amount of C-mol produced can determine how much Gibbs energy is necessary for growth and the extent of Gibbs energy dissipation as enthalpy 64 147 By measuring both we can see more detailed changes in energetics across the tested oxygen regimes Total heat produced during growth was determined for each replicate by integrating the heat flow curve obtained via calorimetry - Hof growth was calculated by taking the quotient of total heat and biomass produced for a final normalized value in kJ C-mol The total heat produced during growth decreased across the tested oxygen regimes RG with the highest amount of dissolved oxygen available 190 M released the largest amount of enthalpy during growth ranging from 3 59 105 to 7 02 105 kJ C-mol OL conditions 110 M O2 produced less heat with - H values between 5 89 104 to 1 62 105 kJ C-mol OCL conditions 56 M O2 produced the lowest amount of heat with - H values ranging from 3 02 104 to 5 77 104 kJ C-mol Gibbs energy consumed - G followed the same trend as - H across the different oxygen conditions - G calculations ranged from 1 104 to 3 04 104 for RG growth conditions 5 94 102 to 2 66 103 for OL conditions and 2 25 102 to 1 65 103 for OCL conditions These results are portrayed in Figure 2 3 as a function of biomass yield for each individual experiment Both - H and - G for RG conditions were very large but resulted in the lowest biomass yield The two oxygen limited conditions generally followed the same negatively correlated trend between - H - G and biomass yield but resulted in greater biomass yields than RG experiments 18 Table 2 1: Summary of data collected calculated for each replicate of the three tested oxygen regimes for Acidianus ambivalens Biomass C-nmol S moles Recommended growth 14 8 4 16 Avg Initial CO2 mM G kJ C-mol H kJ C-mol Energetic yield fmol J 0 0017 182 0 9 1 00 104 3 65 105 100 0 0015 182 0 9 1 21 104 3 62 105 83 4 5 71 10 8 0 0013 192 0 9 1 42 10 6 9 8 0 0006 192 0 9 3 04 104 7 02 105 33 4 5 3 59 10 69 11 10 3 0 0011 198 0 9 1 44 10 10 11 0 0 0009 198 0 9 2 16 104 5 35 105 46 4 5 67 12 5 03 10 10 1 0 0012 191 0 9 1 71 10 1 6 0 0039 123 13 7 1 70 103 6 43 104 2 4 4 71 10 1493 1153 7 99 10 592 6 8 1 4 0 0049 123 13 7 6 79 10 8 1 1 0 0 0081 107 13 7 8 72 102 1 62 105 3 5 464 377 4 0 1 8 0 0023 107 13 7 2 16 10 6 5 2 3 0 0029 100 13 7 2 66 103 1 44 105 2 4 1 55 10 9 9 0 8 0 0132 100 13 7 5 94 10 6 9 1 5 0 0059 110 13 7 1 44 103 1 11 105 964 0 0025 28 0 9 2 63 102 3 02 104 3930 Oxygen- and carbon-limited 6 2 2 5 Avg Initial O2 M 14 Oxygen-limited 6 4 Avg 10 2 Yield c-mol S 5 89 10 1709 2 5 1 8 0 0014 28 0 9 2 25 102 3 17 104 4564 3 6 0 8 0 0048 63 0 9 1 65 103 3 49 104 605 8 0 3 0 0 0027 63 0 9 1 60 103 5 77 104 625 6 2 2 2 0 0028 78 0 9 2 83 102 5 59 104 513 7 2 0 7 0 0099 78 0 9 5 15 102 5 51 104 1021 5 6 1 8 0 0040 56 0 9 7 57 102 4 43 104 1876 - H kJ C-mol 19 106 Recommended growth conditions O2 limited 105 - G kJ C-mol O2 and CO2 limited 104 103 102 0 0 005 0 01 0 015 Biomass yield C-mol S-mol Figure 2 3: Measured enthalpies of growth - H and calculated Gibbs energy consumed G for the replicates in the three experimental growth conditions on a log scale in relation to the corresponding biomass yield Filled symbols represent measured enthalpy values upper half and corresponding open symbols show Gibbs energy consumed lower half Recommended growth red circles O2 -limited blue squares O2 and CO2 -limited green triangles Each symbol represents one replicate in the designated growth condition and error bars are included for all measurements 20 105 10 - H r - G or - H kJ C-mol 9 - G r closed markers - H open markers - G 8 7 Recommended growth conditions 6 5 4 3 O2 limited 2 O2 and CO2 limited 1 0 0 0 005 0 01 0 015 Biomass yield C-mol S-mol Figure 2 4: Black curves represent standard enthalpy - H r solid line and Gibbs energy - G r dashed line for the overall growth reaction for A ambivalens as it relates to biomass yield Standard state values for the overall growth reaction were calculated for 1 molal of pure reactants and products at 76 The standard state curves act as theoretical results for how microbial growth will behave at different biomass yields Plotted on top of the standard state curves are actual growth data each symbol represents one replicate in the designated growth condition The filled symbols represent measured enthalpies of growth - H determined via calorimetry and open symbols represent Gibbs energy consumed G calculated from chemical data for all three growth conditions recommend growth: red circles O2 -limited: blue squares O2 - and CO2 -limited: green triangles The measured enthalpies of growth fall along or close to the theoretical enthalpy of growth solid line All - G values were greater than zero but much smaller than the theoretical Gibbs energy of growth dashed line Error bars are included for all measurements 21 When standard state values of G r and H r of the modeled overall growth reaction for A ambivalens were compared H r was always greater than G r across the biomass yield values with the difference between the lines representing entropic thermal energy T S r 150 164 as shown in Figure 2 4 Therefore enthalpy released during growth must be large to compensate for an unfavorable change in T S r that would otherwise hinder microbial growth The behavior of the standard state values predicts the primary mode of Gibbs energy dissipation will be through the release of heat and growth will be highly exothermic This prediction held true for all experiments - H G and measured - H fell along or near H r modeled with the exception of some OCL replicates Calculated - G however was considerably less than G r modeled for all three growth scenarios Though Gibbs energy was primarily dissipated as enthalpy - H - G the results suggest that a substantial portion of Gibbs energy is dissipated as T S In all experiments - G was smaller than the amount of enthalpy produced - H - G following standard state predictions however the standard state values did not accurately predict T S or - G 2 3 4 Response to carbon and oxygen availability To evaluate the energetics of growth in response to oxygen and inorganic carbon availability both initial dissolved oxygen activity and initial CO2 activity were plotted against - G Figure 2 5 and 2 6 respectively Two sample t-tests showed oxygen values for each growth condition were statistically different from each other P 0 05 for all three tests ANOVA and post analysis Tukey tests revealed RG - G values were statistically different from OL and OCL data 0 05 confidence level but OL and OCL were statistically indistinguishable ANOVA and Tukey tests were conducted for initial CO2 and showed OL contained significantly more inorganic carbon while RG and OCL were the same There were significant differences in initial activity of CO2 between OL and OCL experiments but insignificant differences in - G during growth This suggests the differences in Gibbs energy consumption and enthalpy across the three tested conditions were directly related to the initial dissolved oxygen activity in the medium while initial CO2 activity had no measurable effect 22 104 4 3 5 - G kJ C-mol 3 Recommended growth conditions 2 5 2 1 5 1 O2 and CO2 limited 0 5 O2 limited 0 0 0 05 0 1 0 15 Initial dissolved oxygen activity mmol 0 2 Figure 2 5: Gibbs energy consumed per C-mol biomass produced for the tested oxygen treatments All three oxygen treatments had significantly different beginning dissolved oxygen concentrations P 0 The primary mode of Gibbs energy dissipation is another approach to characterize how microbial life grows from a thermodynamic perspective Microbial growth of A ambivalens by sulfur oxidation is an excellent example of enthalpy-driven growth H
    • Hodson, Keith - Ph.D. Dissertation
      Structural controls and diagenetic conditions associated with fluid migration on the Moab Fault, UT
      Appendix 1
      Appendix 2 2018, Hodson,Keith,Keith Hodson
    • Hu, Yan - Ph.D. Dissertation
      Tracing subduction zone processes with magnesium isotopes 2018, Hu,Yan,Yan Hu Tracing subduction zone processes with magnesium isotopes Yan Hu A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy University of Washington 2018 Reading Committee: Fang-Zhen Teng Chair Bruce K Nelson Ronald S Sletten Program Authorized to Offer Degree: Earth and Space Sciences Copyright 2018 Yan Hu University of Washington Abstract Tracing subduction zone processes with magnesium isotopes Yan Hu Chair of the Supervisory Committee: Professor Fang-Zhen Teng Department of Earth and Space Sciences Subduction and recycling of oceanic plates change the chemical composition of mantle and affect its physical properties thereby modulating Earth s dynamics Stable Mg isotopes 26Mg can trace this recycling process as crustal materials are highly fractionated compared to the average mantle composition This dissertation focuses on Mg isotope fractionation during subduction-related processes and the consequent mantle heterogeneity The dissertation first compares inconsistent 26Mg values of San Carlos peridotitic olivines that are published by several laboratories We analyzed mineral grains from two San Carlos peridotites with disparate lithologies and determined that all mineral phases have indistinguishable 26Mg values to within 0 07 With analytical precision and accuracy being confirmed the significance of anomalous 26Mg values can be understood in the context of mantle heterogeneity The next paper investigates the Mg isotopic heterogeneity in mantle pyroxenites that have formed by multi-stage interactions between peridotites and melts with diverse origins Pyroxenites formed by reaction with melts derived from subducted oceanic crust and carbonate sediments are shown to have variable 26Mg values 1 51 to 0 10 In contrast pyroxenites that are formed by reaction with silicate melts from deep mantle has 26Mg values similar to common mantle peridotites Therefore subducted oceanic slab can indeed produce local Mg isotopic heterogeneities The third paper aims to characterize the Mg isotopic compositions of subducting sediments which account for a substantial Mg input to global subduction zones We analyzed 77 bulk sediments that were recovered from drill cores worldwide these cores contain diverse lithologies with a wide compositional variation The sediments display a large variation in 26Mg ranging from 1 34 carbonate-rich sediments to 0 46 clay-rich sediments indicating that sediment recycling is a feasible process to alter the Mg isotopic composition of local mantle The fourth paper assesses the Mg isotopic distribution in a typical mantle wedge overlying an active subduction zone where fluids released from subducting slab acting as a transfer medium for elements between slab and mantle wedge A suite of sub-arc peridotites from Avacha Russia are chosen to represent western Pacific mantle wedge The isotopic compositions of these peridotites are similar to normal mantle peridotites suggesting that fluids produced by slab dehydration at relatively shallow depth pressure of 2-3 GPa are too depleted in Mg to leave a measurable fingerprint on the Mg-rich mantle wedge Therefore large-scale dehydration of isotopically distinct minerals at higher pressures such as serpentine are likely to be required to produce mantle domains with atypical 26Mg Collectively this dissertation confirms the existence of locally heterogeneous mantle domains caused by subducted slabs highlights the potential of recycling sediments for introducing heterogeneous 26Mg to the mantle and emphasizes the importance of subduction zone thermal structure which controls the dehydration path of a subducting slab in generating atypical 26Mg in mantle wedge The results from this dissertation contribute to the overall development of using Mg isotopic system to trace crustal recycling and associated mantle heterogeneities Table of Contents List of Figures iv List of Tables vi Chapter 1 Introduction 10 Chapter 2 Magnesium isotopic homogeneity of San Carlos olivine: a potential standard for Mg isotopic analysis by multi-collector inductively coupled plasma mass spectrometry 17 Abstract 17 1 Introduction 18 2 Experimental 22 2 1 Samples 22 2 2 Analytical methods 24 3 Results and discussion 26 3 1 Data accuracy and precision 26 3 2 Magnesium isotopic composition of mineral separates from San Carlos peridotites 27 3 3 Comparisons with previous data 27 3 4 Evaluating the influence of resin cleaning on Mg isotopic analysis 34 3 5 Previous inter-laboratory discrepancies on San Carlos olivine: sample heterogeneity vs analytical artifacts 39 4 Conclusions 41 Chapter 3 Metasomatism-induced mantle magnesium isotopic heterogeneity: Evidence from pyroxenites 43 Abstract 43 1 Introduction 44 2 Sample petrology 46 3 Methods for Mg isotopic analysis 50 4 Results 55 4 1 Whole-rock major and trace element compositions 55 4 2 Mineral chemistry 57 4 3 Temperature and pressure estimates 62 4 4 Magnesium isotopic compositions 65 i 4 4 1 Whole-rocks 65 4 4 2 Mineral separates 69 5 Discussion 70 5 1 Garnet-bearing lherzolites and pyroxenites 72 5 1 1 Origin of garnet-bearing xenoliths 72 5 1 2 Whole-rock Mg isotopic variation 73 5 1 3 Inter-mineral Mg isotope fractionation 74 5 1 4 Mechanism of disequilibrium garnet olivine pyroxene fractionation 79 5 2 Garnet-free websterites and orthopyroxenite 81 5 3 Garnet-free clinopyroxenites 82 5 4 Implications on mantle 26Mg heterogeneity 86 6 Conclusions 89 7 Appendix A Supplementary data 90 Chapter 4 Magnesium isotopic composition of subducting marine sediments 91 Abstract 91 1 Introduction 92 2 Samples 95 3 Analytical methods 99 4 Results 101 4 1 Detrital sediments 109 4 1 1 Turbidites 109 4 1 2 Terrigenous clays 111 4 2 Hydrogenetic and hydrothermal clays 112 4 3 Biogenic sediments 113 5 Discussion 114 5 1 Mechanisms of Mg isotope fractionation in marine sediments 114 5 1 1 Mineralogical control on marine sediment 26Mg values 116 5 1 2 Detrital sediments: source signature and sorting effects 118 5 1 2 1 Source signature: provenance heterogeneity 120 5 1 2 2 Source signature: intensity of chemical weathering 124 5 1 2 3 Sediment transport: grain-size sorting 127 ii 5 1 3 Non-detrital sediments 128 5 1 3 1 Hydrogenetic clays and Mn nodule 128 5 1 3 2 Hydrothermal clays 132 5 2 Sedimentary Mg input to global subduction zones 134 5 3 Implications for arc magmatism and mantle Mg isotopic heterogeneity 139 6 Conclusions 140 7 Appendix A Supplementary data 141 Chapter 5 Magnesium cycling at subduction zones constrained by Mg isotopic composition of sub-arc mantle beneath southern Kamchatka 143 Abstract 143 1 Introduction 144 2 Geologic setting 147 3 Samples and evidence for slab - mantle interaction 148 4 Analytical methods 153 5 Results 154 6 Discussion 158 6 1 Magnesium isotopic compositions of slab-derived fluids 164 6 2 Magnesium isotopic compositions in mantle wedge peridotites 165 6 3 Behavior of Mg in the mantle wedge-slab system 168 6 4 Comparison with arc lavas: Implications for Mg cycling in subduction zones and beyond 171 7 Conclusions 176 Chapter 6 Summary and future work 178 References 182 iii List of Figures Figure 2-1 Compilations of 26Mg data for a mantle olivine from peridotite xenoliths and b San Carlos olivine 20 Figure 2-2 The two San Carlos peridotites analyzed in this study 23 Figure 2-3 Magnesium three-isotope plot relative to DSM-3 standard 28 Figure 2-4 a Mg isotopic compositions of mineral separates from the San Carlos SC peridotites b Inter-mineral Mg isotope fractionation of Cpx-Ol and Opx-Ol pairs from the San Carlos peridotites 30 Figure 2-5 Compilation of inter-mineral Mg isotope fractionation of a Cpx-Ol and b Opx-Ol pairs in peridotite xenoliths 33 Figure 2-6 Magnesium isotopic compositions of a lherzolitic San Carlos olivine and b Hawaiian seawater processed through column chemistry using different types of resins 35 Figure 3-1 Petrological and petrographic characteristics of Hannuoba garnet-bearing xenoliths 48 Figure 3-2 Major A C and trace D element variations of Hannuoba HNB mantle xenoliths studied here in relation to their MgO wt % 56 Figure 3-3 Trace element features for Hannuoba mantle xenoliths 58 Figure 3-4 Mineral chemical composition plots for Hannuoba mantle xenoliths 60 Figure 3-5 Wo En Fs classification quadrilateral of pyroxenes from Hannuoba mantle xenoliths 61 Figure 3-6 Mg correlations between co-existing A Cpx Ol and B Opx Ol 63 Figure 3-7 Compilation of 26Mg for A mantle-related rocks and B mineral separates 71 Figure 3-8 26Mg of the Hannuoba garnet-bearing xenoliths in relation to their MgO wt % 75 Figure 3-9 Inter-mineral Mg isotope fractionation 77 Figure 3-10 Ti Eu vs La Yb N plot for distinguishing between carbonatite metasomatism and silicate metasomatism Coltorti et al 1999 84 Figure 3-11 Trace element and Mg isotopic compositions of the two Hannuoba clinopyroxenites suggest involvement of carbonatites in their genesis 85 Figure 4-1 Geographic map showing the sample locations of the ten drill sites analyzed in this study yellow square and the other two sites analyzed in previous work white square Teng et al 2016 96 Figure 4-2 Major element variability of subducting sediments investigated in this study 97 Figure 4-3 Magnesium isotopic compositions of subducting marine sediments 108 Figure 4-4 Positive correlations of 26Mg with MgO in sediments from A ODP 701 South Sandwich and B DSDP 211 Nicobar Fan indicate a mixing trend between Mg-depleted siliceous oozes and Mg-rich detrital clays 115 Figure 4-5 Mineralogical control on Mg isotopic composition of marine sediments 117 Figure 4-6 Positive correlations between 26Mg values and Cs Rb contents in carbonaterich sediments 119 Figure 4-7 26Mg variation in subducting sediments with relation to their contrasting provenance compositions 122 iv Figure 4-8 The effect of source rock heterogeneity on Mg isotopic composition of subducting sediments 123 Figure 4-9 26Mg variation in detrital sediments during progressive chemical weathering 126 Figure 4-10 26Mg variation related to grain-size sorting during sediment transport 129 Figure 4-11 26Mg variation in Tonga hydrogenetic sediments reflects the preferential uptake of heavy Mg isotopes during authigenic mineral formation 131 Figure 4-12 26Mg variation related to hydrothermal sediments 133 Figure 4-13 Variations of Mg mass fluxes and estimated average 26Mg values for sedimentary sections subducting at major subduction zones 136 Figure 5-1 Geological map of the Kamchatka peninsula Russia 149 Figure 5-2 Geochemical evidence for subduction-related metasomatism recorded in the Avacha peridotite xenoliths 152 Figure 5-3 26Mg in bulk Avacha peridotite xenoliths and their constituent minerals plotted against modal and chemical parameters 162 Figure 5-4 Inter-mineral fractionation between pyroxene Px and olivine Ol in Avacha peridotites 163 Figure 5-5 Lack of correlation between 26Mg in Avacha peridotites with common indices for slab fluid-peridotite interaction 166 Figure 5-6 Illustration of Mg isotope evolution of mantle peridotites during open-system fluid-peridotite interaction at different fluid rock ratios and different 26Mgrock-fluid values 172 Figure 5-7 Compilations of published arc lava 26Mg data with their trace element characteristics 173 Figure 5-8 The average 26Mg value for each arc as a function of slab temperature and slab depth 175 v List of Tables Table 2-1 Magnesium isotopic compositions of reference materials 29 Table 2-2 Magnesium isotopic compositions of mineral separates from San Carlos peridotites 31 Table 2-3 Magnesium isotopic compositions of Hawaiian seawater and lherzolitic San Carlos olivine processed through resins cleaned in different ways 36 Table 2-4 Resin codes and associated Mg blank values 38 Table 3-1 Magnesium isotopic compositions of standards analyzed during the course of this study and comparison with literature data 53 Table 3-2 Reproducibility check of Mg isotopic measurements relative to standard DSM3 54 Table 3-3 Temperature C and pressure GPa estimates for Hannuoba garnet-bearing xenoliths 64 Table 3-4 Magnesium isotopic compositions relative to standard DSM3 of mineral separates and whole rock powders from Hannuoba mantle xenoliths 66 Table 3-5 Inter-mineral Mg isotope fractionations in Hannuoba garnet-bearing xenoliths 67 Table 3-6 Magnesium isotopic compositions of mineral separates and whole rock powders from Hannuoba garnet-free pyroxenite xenoliths basalts and a Qinling garnet peridotite xenolith 68 Table 4-1 Magnesium isotopic compositions of marine sediments from various DSDP and ODP sites outboard of subduction zones 102 Table 4-2 Calculated Mg fluxes and bulk 26Mg values of global subducting sedimentary columns 135 Table 5-1 Mg isotopic compositions of standards analyzed during the course of this study 155 Table 5-2 Mg isotopic compositions of bulk Avacha harzburgite xenoliths 156 Table 5-3 Table 5-3 Mg isotopic compositions of mineral separates 159 Table 5-4 Table 5-4 Inter-mineral Mg isotope fractionation 161 vi Acknowledgments On the 6-hour flight to visit the flowing lava in Hawaii I started the first sentence in my PhD dissertation I think it is all started from there my passion for geology and my love for rocks I became involved in Mg isotope research as a visiting scholar in my PhD advisor s Prof Fang-Zhen Teng s world-renowned laboratory in University of Arkansas Fayetteville when I was pursuing my master s degree and changed my major from Gemology to Geochemistry Back in that time I know nothing about a clean laboratory nor mass spectrometers Therefore I would like to express my deepest gratitude to my advisor for teaching me analytical chemistry from scratch for providing a rich environment to carry my PhD research for trusting me in the laboratory and on the instruments for encouraging me to interact with people from diverse disciplines for caring about not only my academic progress but also my daily life My PhD work will not be possible without the help from my committees I would like to thank Prof Ronald Sletten for his knowledge on surface processes and aquatic chemistry for providing opportunities of fantastic field trips for teaching me do not feel sorry for myself and do not be defeated by myself I am grateful to Prof Bruce Nelson for his expertise on both instrumental analysis and isotope geochemistry for his thoughtful advice on my research and constant support during various geologic society meetings as well as making me realize having small hands are something that I could put on my resume when applying jobs in mass spectrometry I thank Prof John Stone for sharing his interests in rocks minerals meteorites and stars His passion for teaching always motivates me to be a good teaching assistant I would also like to recognize Prof Bo Zhang my Graduate 7 School Representative for his constructive input and for raising thoughtful questions at my General and Final Exams My PhD work has also been benefited greatly from interactions with all my coauthors I would like to specially thank Prof Terry Plank at Columbia University for her support and collaborative research on global subducting sediments as well as her mentoring during the CIDER workshop at University of Berkeley in 2017 summer I am also grateful to Prof Ionov Dimitri from Universit de Montpellier France for providing precious sub-arc peridotite samples for my research as well as sharing with me his professional knowledge on mantle xenoliths I also thank my friend Prof Xiao-Ming Liu from University of North Carolina at Chapel Hill for teaching me how to tune the Nu Plasma MC-ICPMS and to take criticism seriously but not personally I would like to express my sincere thanks to our laboratory manager Dr Scott Kuehner for maintaining the instrument and clean laboratory with extreme care patience and enthusiasm for his technical academic and life support I like his saying Let s see what we can break whenever I ask him for help on the instrument Profs Antony Irvine and Bernard Evans are thanked for fruitful and fun discussions on planetary mantle and subduction-zone processes Prof Stewart McCallum is thanked for teaching me how to read his stories from rock thin sections and for his help when I was the teaching assistant for Igneous Petrology In addition I am indebted to all the people working in the main office as well as No ll Bernard-Kingsley Meghan Oxley Dave McDougall and Nathan Briley for helping me with various daily questions and problems I am also grateful to my academic brothers and sisters and my fellow graduate students for their support and help along this beautiful journey 8 My additional thanks are due to the various funding sources supported my PhD endeavors including the National Science Foundation Graduate Student Research Grant from Geological Society of America 2014 and 2015 Robert and Nadine Bassett Fellowship 2014 Kenneth C Robbins Fellowship 2015 George Edward Goodspeed Geology Scholarship 2015 Peter Misch Fellowship 2016 Harry Wheeler Scholarship 2017 David A Johnston Memorial Fellowship 2018 and Inquisitive Graduate Student Support Fund 2014 from the Department of Earth and Space Sciences at the University of Washington These awards not only provided financial support for the completion of my PhD research but also built confidence in my scholarly abilities Last but not the least I would like to express my greatest appreciation to my family in China for their consistent support love encouragement and inspiration over all these years This will not be an end but rather a brand-new start and I am feeling excited 9 Chapter 1 Introduction Subduction zones are the primary sites where oceanic plates dive into the mantle and forming juvenile continental crust Accompanying these processes are active cycling of elements and volatiles between Earth s surface and interior The efficiency of this cycling process varies for different chemical species which is related to the nature of the species and the thermal structure of a given subduction zone As a result certain species are preferentially transferred from the subducting slab to the overlying mantle wedge which are subsequently incorporated to arc magmas and return back to surface Materials that are retained in the downgoing plate subduct to the deeper mantle and contribute to the development of long-term chemical heterogeneities in the mantle Therefore subduction zone processes play a central role in the dynamic evolution of the Earth Isotope systematics allow tracing subduction-related modification in the mantle since oceanic slab inherits a crustal- or seawater-like isotope signature during seawater hydrothermal alteration and sedimentation before entering the trench Studies of radiogenic radioactive and cosmogenic isotopes such as Sr Nd Pb Hf Os 238U 235U and 10 Be on oceanic basalts and mantle peridotites suggest that the mantle is compositionally heterogeneous on all spatial and temporal scales with subducted materials being one of the essential contributors to enriched mantle domains e g Tera et al 1986 Zindler and Hart 1986 Morris and Tera 1989 O Reilly and Griffin 2013 Hofmann 2014 Andersen et al 2015 Stable isotopes such as oxygen isotopes are time-independent sensitive tracers for recycled slab materials because they fractionate significantly during low-temperature interactions with Earth s surface whereas fractionations occurring at mantle temperatures are largely negligible e g Woodhead et al 1987 1993 Eiler 2001 Despite the ever- 10 increasing evidence of subduction recycling and mantle heterogeneity fundamental questions and controversies remain For example where do the enriched crust-like signature of arc magmas originate from How do elements transfer from downgoing plate to arc magmas And could subducted materials be responsible for chemical heterogeneities seen in intra-plate continental alkaline basalts as well as ocean island basalts Rapid growth in studies of non-traditional stable isotopes e g Li Mg Fe and K have been facilitated by analytical advances in Multi-Collector Inductively Coupled Plasma Mass Spectrometry MC-ICPMS over the past decade These advances allow subpermil isotopic variations to be resolved and provide a new technique for studying subduction zone processes The magnesium isotopic system is proving to be an effective tracer of such processes due to the sharp contrast in the values and ranges of 26Mg ratios 26Mg 26Mg 24Mg sample 26Mg 24Mg DSM3 1 1000 between mantle 26Mg 0 25 0 07 Teng et al 2010a and surface materials 5 57 to 1 81 Wombacher et al 2011 Liu et al 2014 and negligible fractionation during metamorphic dehydration Teng et al 2013 Li et al 2014 Wang et al 2014b Magnesium is an abundant 2 cation in both silicate rocks and the hydrosphere constituting 15% of the Earth by mass It has three naturally occurring stable isotopes 24Mg 78 99% 25 Mg 10 00% and 26 Mg 11 01% Rosman and Taylor 1998 The large relative mass differences 4 to 8% and its high solubility in fluids lead to significant massdependent Mg isotope fractionation during low-temperature surface processes For example continental weathering preferentially leaches light Mg isotopes from rock resulting in light Mg isotopic composition in rivers average 26Mg 1 09 Tipper et 11 al 2006a and oceans average 26Mg 0 83 0 09 Foster et al 2010 Ling et al 2011 and a complementary isotopically heavy weathering residue e g Teng et al 2010b Liu et al 2014 On the other hand during carbonate precipitation the weak Mg-O bond in carbonate minerals cause preferential incorporation of light Mg isotopes from aqueous solutions Saenger and Wang 2014 Therefore carbonates have a characteristic light 26Mg signatures typically between 5 and 1 see review in Teng 2017 which makes the Mg isotopes an effective tracer for subducted carbonate By contrast similarly strong Mg-O bond strength shared by common mantle minerals i e olivine clinopyroxene and orthopyroxene limits the extent of equilibrium isotope fractionation during mantle partial melting and fractional crystallization resulting in a uniform 26Mg mantle on a global scale 0 25 0 07 Teng et al 2007 2010a Exceptions to this are spinel and garnet which have a 4-fold and 8-fold coordination for Mg respectively Therefore spinel is enriched in heavy Mg isotopes 0 to 0 65 Young et al 2009 Liu et al 2011 Xiao et al 2013 whereas garnet is enriched in the light isotopes 2 15 to 0 37 Wang et al 2012 2015c Hu et al 2016b An et al 2017 The application of Mg isotopes as a tracer for crust-mantle interaction requires both clearly defined values for surface and mantle Mg reservoirs as well as the knowledge of processes and mechanisms governing these distributions Magnesium is the most abundant cation in the mantle MgO 36 77 wt % Palme and O Neill 2014 and over 99% of the Mg inventory resides in the mantle Mass-balance calculation suggests difficulty in perturbation of the homogeneous Mg isotopic distribution in the mantle however subduction recycling does appear to alter these values For example continental and arc magmas have display large 26Mg variation ranging from 0 61 to 0 06 Yang et al 12 2012 Wang et al 2016 Li et al 2017 Su et al 2017 Sun et al 2017 However oceanic basalts particularly ocean island basalts sampled on a global scale do not show similarly large variation but are rather homogeneous 0 26 0 07 n 110 Teng et al 2010a This paradox raises the question that whether the anomalous 26Mg values in continental and arc magmas reflect genuine mantle signature or are a crust-modified signature during magma ascent If the former is the case then what processes are uniquely involved in basalts formed in continental and arc settings from those produced in oceanic realm Basalts are derived from partial melting of the mantle and thus they are indirect mantle samples Studies on mantle xenoliths offer a more direct view of mantle composition because they are fragments of solid mantle that are incidentally entrained in magmas and brought to the surface during volcanic eruptions This dissertation explores the extent distribution and mechanism of mantle Mg isotopic heterogeneity recorded in mantle xenoliths and their linkage with subduction-related processes The first chapter examines the nature of large inter-laboratory difference in 26Mg measured on San Carlos peridotites while the next three chapters focus on Mg isotopic compositions of mantle pyroxenite xenoliths modern subducting marine sediments and sub-arc mantle peridotite xenoliths This dissertation is presented as four independent papers that have either already been published or will be submitted for publication They are described briefly below Chapter 2 examines Mg isotopic heterogeneity of San Carlos peridotite xenoliths which display large inconsistent 26Mg values in olivine samples measured from different laboratories Two batches of olivine grains from a lherzolite and four batches of olivine clinopyroxene and orthopyroxene grains from a harzburgite have been analyzed The six different batches of olivine grains yield homogeneous 26Mg to within 0 03 2SD and 13 are isotopically indistinguishable to co-existing pyroxene grains Therefore previously documented large inter-laboratory discrepancies are likely not accurate and our study further supports that mantle peridotites in general have a homogeneous Mg isotopic composition This work was published as an article entitled Magnesium isotopic homogeneity of San Carlos olivine: a potential standard for Mg isotopic analysis by multi-collector inductively coupled plasma mass spectrometry in Rapid Communications in Mass Spectrometry With our analytical accuracy being confirmed by Chapter 2 the measured anomalous 26Mg values in the following chapters are attributed to genuine mantle heterogeneity Chapter 3 examines the existence of small-scale heterogeneity in local mantle domains caused by melt-rock interaction i e metasomatism To explore this possibility this study presents the first systematic Mg isotopic analyses on pyroxenite xenoliths which are physical evidence for melt infiltration into the lithospheric mantle Twenty-five pyroxenite with diverse petrological occurrences and origins were measured which revealed a large range in whole-rock 26Mg values with large inter-mineral disequilibrium isotope fractionations that are explained to reflect the various origins of the metasomatic melts These heterogeneous 26Mg values may be linked to the large variations seen in continental basalts as they often contain contributions from lithospheric mantle to various externs The relatively cold lithospheric mantle and lack of convection also facilitate preservations of mantle heterogeneities This work was published as an article entitled Metasomatism-induced mantle magnesium isotopic heterogeneity: Evidence from pyroxenites in Geochimica et Cosmochimica Acta A natural question following the findings in Chapter 3 is the origin of metasomatic 14 melts in the mantle Chapter 4 presents the first systematic investigation on Mg isotopic variation in subducting marine sediments which have been considered as an essential contributor to isotopic heterogeneity of the mantle A total of 77 bulk sediments collected from 10 drill sites worldwide yield a wide 26Mg range from 1 34 to 0 46 the variation of which appears to be linked to sediment lithology In addition the flux-weighted 26Mg of 0 336 is estimated for Global Subducting Sediments GLOSS-II which is isotopically lighter than the mantle average Therefore this study demonstrates that subducting marine sediments are a source of heterogeneous 26Mg in the mantle This work was published as an invited research article entitled Magnesium isotopic composition of subducting marine sediments in Chemical Geology The highly variable Mg isotopic input from subducting slab suggests that components with atypical Mg isotopic compositions may be present in the mantle wedge as fractionated crustal materials are recycled into the mantle at subduction zones Chapter 5 presents the first systematic study on the Mg isotopic composition of a typical suprasubduction zone mantle domain by analyzing 23 large and fresh harzburgite xenoliths from the Avacha volcano in Kamchatka Russia Both sub-arc mantle xenoliths and arc lavas are well studied for this region e g Kersting and Arculus 1995 Kepezhinskas et al 1996 1997 Kepezhinskas and Defant 1996 Yogodzinski et al 2001 Widom et al 2003 Ionov 2010 Ishimaru and Arai 2008a b c 2011 Kayzar et al 2014 B nard et al 2017 This study finds that melting residues in the mantle wedge in the western Pacific where dehydration of slab crust at pressures of 2-3 GPa serves as the major water supplier for melting and metasomatism likely have a mantle-like 26Mg due to the low Mg concentrations in fluids released at such depths Large-scale dehydration of isotopically 15 distinct phases at higher pressures may be required to produce isotopically diverse arc lavas such as those seen in lavas from Lesser Antilles and Philippines Teng et al 2016 Li et al 2017 In addition since Mg rich carbonate magnesite and K-rich silicate such as phlogophie and phengite do not typically dehydrate at sub-arc depth they may preserve their distinct 26Mg to deeper mantle which may be tapped later by ultra potassic magmas e g Liu et al 2015 Sun et al 2017 This work will be submitted soon as an article entitled Magnesium cycling at subduction zones constrained by Mg isotopic composition of sub-arc mantle beneath southern Kamchatka The perspectives for this work and future suggested research are presented in Chapter 6 16 Chapter 2 Magnesium isotopic homogeneity of San Carlos olivine: a potential standard for Mg isotopic analysis by multi-collector inductively coupled plasma mass spectrometry This chapter is published as: Hu Y Teng Harrington M D Sun Y Konter J and Teng F -Z 2016 Magnesium isotopic homogeneity of San Carlos olivine: a potential standard for Mg isotopic analysis by multi collector inductively coupled plasma mass spectrometry Rapid Commun Mass Spectrom 30 2123-2132 Abstract RATIONALE: Previous analyses on San Carlos olivine from Arizona USA have shown inter-laboratory 26Mg differences of up to 0 67 while mantle olivine samples worldwide are homogeneous at a current analytical uncertainty of 0 1 The differing measurements on San Carlos olivine may be attributable to analytical artifacts or sample heterogeneity The latter must be ruled out before using it as a standard for Mg isotopic analysis METHODS: To examine sample homogeneity two different batches of San Carlos olivine from a lherzolite and four batches from a harzburgite have been analyzed together with coexisting harzburgitic pyroxene In addition the effect of acid purity on resin performance and the reusability of AG50W-X8 resin for Mg separation have been evaluated by processing another batch of lherzolitic San Carlos olivine and Hawaiian seawater through both new and used resins cleaned with different acids 17 RESULTS: Six different batches of olivine grains from two San Carlos peridotite xenoliths show homogeneous 26Mg values to within 0 03 and all the mineral phases in the harzburgite are in Mg isotope equilibrium Furthermore there is no resolvable 26Mg shift in either lherzolitic San Carlos olivine or Hawaiian seawater by using either new or used resins that were cleaned with single-distilled or double-distilled acids CONCLUSIONS: The new data are consistent with the narrow 26Mg range of mantle olivine worldwide while they stand in contrast to the wide range measured on the same San Carlos olivine powder in different laboratories Therefore previous inter-laboratory discrepancies reflect analytical artifacts instead of sample heterogeneity and San Carlos olivine is a suitable standard for Mg isotopic analysis 1 Introduction The application of multi-collector inductively coupled plasma mass spectrometry MC-ICP-MS in magnesium Mg isotope geochemistry has made it possible to achieve a routine precision of better than 0 1 compared with a precision of 1-2 by thermal ionization or secondary ionization mass spectrometry This tenfold improvement in analytical capability now allows small mass-dependent Mg isotope fractionation in terrestrial samples especially in high-temperature samples to be resolved Studies on peridotite xenoliths and oceanic basalts have shown that the mantle which holds 99% of Earth s total Mg inventory has a homogeneous Mg isotopic composition 26Mg 0 25 0 07 2SD Teng et al 2010a A compilation of olivine data from peridotite xenoliths worldwide also suggests a narrow 26Mg range 0 25 0 12 2SD despite differences in age Phanerozoic to Archean and degree of melt extraction and metasomatic 18 refertilization of the host peridotites Fig 2-1a Handler et al 2009 Yang et al 2009 Liu et al 2011 Pogge von Strandmann et al 2011 Xiao et al 2013 Wang et al 2015 a b 2016 However olivine grains from San Carlos peridotites Arizona USA seem to span an unexpectedly large range from 0 73 to 0 06 Fig 2-1b Pearson et al 2006 Teng et al 2007 Wiechert and Halliday 2007 Handler et al 2009 Huang et al 2009 Wimpenny et al 2010 Pogge von Strandmann et al 2011 Sio et al 2013 Wang et al 2015 a b 2016 More strikingly analyses on aliquots of the same large batch of San Carlos olivine powder by multiple laboratories have still yielded a wide range from 0 55 to 0 19 Fig 2-1b Young et al 2009 Chakrabarti and Jacobsen 2010 Liu et al 2010 Yang et al 2012 Bouvier et al 2013 As mass dependent isotope fractionation produced during chemical purification and instrumental analysis follows the same behavior as occurs in nature it is still unclear whether these discrepancies are the result of analytical artifacts or sample heterogeneity Understanding the source of these inconsistent results is important as they indicate either that the mantle is not universally homogeneous or that there is inter-laboratory bias for Mg isotopic analysis The peridotite xenoliths from San Carlos as in most other studied locations represent a range of mantle processes and sources If sample heterogeneity is the cause previous suggestion of the homogeneous chondritic Mg isotopic composition of the mantle Teng et al 2010a might need to be re-examined On the contrary if the differing measurements are caused by analytical issues results from different laboratories cannot be compared directly More importantly the 0 36 interlaboratory difference on aliquots of the same San Carlos olivine powder 0 55 to 0 19 Fig 2-1b Young et al 2009 Chakrabarti and Jacobsen 2010 Liu et al 2010 Yang et al 2012 Bouvier et al 2013 19 Figure 2-1 Compilations of 26Mg data for a mantle olivine from peridotite xenoliths and b San Carlos olivine Data sources: Handler et al 2009 Yang et al 2009 Liu et al 2011 Pogge von Strandmann PvS et al 2011 Xiao et al 2013 and Wang et al 2016 Handler et al 2009 Pogge von Strandmann PvS et al 2011 Wang et al 2015 a b 2016 Pearson et al 2006 Teng et al 2007 Wiechert and Halliday W&H 2007 Huang et al 2009 Wimpenny et al 2010 Sio et al 2013 Young et al 2009 Chakrabarti and Jacobsen C&J 2010 Liu et al 2010 Yang et al 2012 Bouvier et al 2013 and Lundstrom unpublished The gray bar in each panel represents two standard deviations 2SD of the average mantle 26Mg 0 25 0 07 Teng et al 2010 which is shown as the black vertical line in the center of the gray bar Error bars in this and the following figures represent 2SD UW Isotope Laboratory at the University of Washington and UA Isotope Laboratory at the University of Arkansas both of which use the same analytical method as in this study 20 leaves the argument for a non-chondritic Earth by some laboratories questionable as their measured differences between peridotites and chondrites are only 0 1 to 0 3 Wiechert and Halliday 2007 Young et al 2009 These differences have not been found in other studies that reported data from simultaneously analyzed peridotites and chondrites Teng et al 2007 Handler et al 2009 Yang et al 2009 Teng et al 2010a Pogge von Strandmann 2011 Chakrabarti and Jacobsen 2010 Bouvier et al 2013 further suggesting that previous argument for a non-chondritic Earth may be based on an analytical artifact Investigating the Mg isotopic homogeneity of San Carlos olivine is also important with respect to its use as a standard for data quality control and inter-laboratory comparison Olivine has several advantages as a Mg isotopic standard if its Mg isotopic composition can be demonstrated to be homogeneous First olivine is a common mineral in a wide diversity of lithologies on Earth Moon and in some meteorites e g pallasite and chassigny Second olivine can be used to monitor laboratory-induced Mg isotope fractionation through the whole procedure starting with sample dissolution through chemical purification and instrumental analysis Olivine in hence better in this respect than solution standards such as Cambridge-1 that do not need sample preparation as solid materials Third only a small quantity of olivine is needed as a standard due to its high Mg content which also prevents Mg isotopic resetting by late-stage modifications Finally fresh clear and large olivine grains are easily accessible from peridotite localities worldwide Olivine from mantle xenoliths is also more representative of mantle Mg isotopic composition than some altered peridotite whole rock standards e g USGS dunite standard DTS-1 21 Here we further examine the Mg isotopic homogeneity of San Carlos olivine and its use as a standard for Mg isotopic analysis Independent analyses on four different batches of olivine clinopyroxene and orthopyroxene grains from a San Carlos harzburgite yield consistent results suggesting homogeneity within and among all phases The other two batches of olivine from a San Carlos lherzolite have identical 26Mg values to those from the harzburgite suggesting inter-xenolith Mg isotopic homogeneity The results from this study together with the narrow 26Mg range of olivine grains from global peridotite xenoliths argue for analytical issues in reported inconsistent measurements on San Carlos olivine San Carlos olivine can therefore be used as a reliable standard for establishing analytical accuracy and long-term reproducibility for Mg isotopic measurement 2 Experimental 2 1 Samples The peridotite xenoliths used in this study were collected from the Peridot Mesa 33 33 N 110 47 W a well-known mesa-capping lava flow from one of the vents in the San Carlos volcanic field southwestern United States There abundant ultramafic xenoliths with gem-quality coarse olivine grains were carried to the surface by Late Tertiary to Quaternary alkaline volcanism Both lherzolite and harzburgite are present in the locality 45 to 90% olivine interlayered with pyroxene-rich rocks in some cases Frey and Prinz 1978 Galer and O Nions 1989 One lherzolite and one harzburgite xenolith were selected for Mg isotopic analyses based on their freshness large mineral grains and characteristic mineral assemblages of each peridotite type Fig 2-2 Both peridotites contain Mg rich olivine and Cr-rich diopside and therefore are classified as Group I xenoliths after Frey and Prinz 1978 The peridotite xenoliths were interpreted as 22 Figure 2-2 The two San Carlos peridotites analyzed in this study a Hand specimen of the harzburgite b Photomicrograph of the lherzolite under cross-polarized light Ol olivine Cpx clinopyroxene Opx orthopyroxene Spl spinel 23 melting residues but have undergone multi-stage metasomatic modifications which might be related to the pyroxene-rich rocks Frey and Prinz 1978 Galer and O Nions 1989 Olivine separates were hand-picked from the two coarse-grained peridotites Fig 2-2 Coexisting clinopyroxene and orthopyroxene were also picked from the harzburgite to evaluate inter-mineral Mg isotope equilibrium Previous major element analyses on constituent minerals suggested a complete chemical equilibrium between coexisting minerals and the equilibrium temperature calculated for San Carlos peridotites is around 1022 C Galer and O Nions 1989 Moreover as a major element Mg has a restricted range of concentrations in olivine Mg 87-91 clinopyroxene Mg 88-93 and orthopyroxene Mg 88-92 where Mg represents the molar ratio of 100 Mg Mg Fe2 Galer and O Nions 1989 2 2 Analytical methods Magnesium separation and isotopic analyses were performed at the Isotope Laboratory of the University of Washington UW Seattle WA USA The procedures follow those detailed in previous studies conducted at the Isotope Laboratory of the University of Arkansas UA Fayetteville AR USA Teng et al 2007 Yang et al 2009 Li et al 2010 Teng et al 2010a Teng and Yang 2014 and are summarized here Mineral separates from the two San Carlos peridotites were selected for purity and clarity under a binocular microscope In order to examine intra-xenolith Mg isotopic homogeneity and inter-mineral isotope equilibrium four batches of mineral grains for each mineral phase were hand-picked from the San Carlos harzburgite Meanwhile two batches of olivine grains from the San Carlos lherzolite were picked to examine inter-xenolith Mg 24 isotopic homogeneity Only a small number of mineral fragments were selected in each mineral batch to minimize the homogenization effect from mixing of large quantities of samples The mineral separates were first cleaned in an ultrasonic bath using Milli-Q water 18 2 M cm resistivity Millipore Billerica MA USA then powdered using an agate mortar and a pestle The mineral powders were subsequently dissolved and fluxed in screwtop Teflon beakers Savillex Eden Prairie MN USA in a series of concentrated acids: HF-HNO3 HCl-HNO3 and HNO3 The sample solutions were evaporated to dryness between each step and were finally dissolved in 1 N HNO3 in preparation for column chemistry Magnesium separation and purification were achieved via cation-exchange chromatography with 200 400 mesh AG50W-X8 pre-cleaned resin Bio-Rad Hercules CA USA following the procedure first outlined by Teng et al 2007 In brief the resin was first conditioned in 1 N HNO3 Matrix elements were then removed with 16 mL 1 N HNO3 The Mg fraction was subsequently collected with 19 mL of 1 N HNO3 Two rounds of column chemistry were performed using this same elution sequence to ensure a pure Mg solution The magnesium isotopic ratios were measured on a Nu Plasma Nu Instruments Wrexham UK high-resolution multi-collector inductively coupled plasma mass spectrometer using the sample-standard bracketing method with an individual beaker for each standard replicate Pure Mg solutions in 3% double-distilled HNO3 300 ppb were introduced into the source region using a quartz Cinnabar spray chamber Glass Expansion Pocasset MA USA and a MicroMist glass concentric nebulizer 100 L min Glass Expansion The intensities of sample solutions were diluted to within 95% of the intensities of the bracketing standard solutions The three Mg isotopes 24Mg 25 25 Mg and 26 Mg were well separated in low-resolution mode with a typical 24Mg signal of 3 5-4 5 V The magnesium isotopic data are reported relative to the DSM-3 standard which is a pure Mg solution made of 10 g of pure Mg metal Dead Sea Magnesium Ltd Israel dissolved in 1 L of 0 3 N HNO3 3 Results and discussion The magnesium isotopic data are reported in Table 2-1 for the reference materials Table 2-2 for the mineral separates from San Carlos peridotites and Table 2-3 for the results of the lherzolitic San Carlos olivine and Hawaiian seawater processed through different types of resins The resin codes used in Table 2-3 and Fig 2-6 are explained in Table 2-4 All the samples analyzed during the course of this study fall on a single mass-dependent fractionation line with a slope of 0 513 R2 0 994 Fig 2-3 Hereafter we only discuss 26Mg values 3 1 Data accuracy and precision An USGS peridotite standard PCC-1 was processed and analyzed to assess data accuracy which yielded a 26Mg value of 0 22 in agreement with the recommended value of 0 229 0 055 Teng et al 2015 Furthermore Hawaiian seawater which has been used as a reliable in-house standard in the laboratory for over 5 years was analyzed repeatedly to further monitor data accuracy and long-term reproducibility Table 2-1 Three duplicates of the Hawaiian seawater yielded 26Mg values between 0 83 and 0 84 with an average of 0 83 0 04 consistent with the recommended value of 0 843 0 057 Teng et al 2015 26 3 2 Magnesium isotopic composition of mineral separates from San Carlos peridotites All the mineral separates from the two San Carlos peridotites analyzed in this study have a homogeneous Mg isotopic composition with limited variation Fig 2-4a The harzburgitic and lherzolitic olivine grains have similar 26Mg ranges 0 27 to 0 23 with an average of 0 24 0 03 These values are consistent with previous measurements on different olivine grains from the same San Carlos lherzolite hand specimen 0 27 to 0 23 Wang et al 2016 Wang et al 2015 a b The two types of pyroxene also yield identical 26Mgvalues from 0 25 to 0 21 average 0 23 0 04 for clinopyroxene and from 0 26 to 0 20 average 0 22 0 05 for orthopyroxene The lack of measurable fractionation between them 26MgCpx-Ol 0 01 0 05 26MgOpx-Ol 0 02 0 06 indicates an inter-mineral Mg isotope equilibrium as olivine and pyroxene have a similar Mg-bonding environment and hence similar Mg O bond strength Fig 2-4b Liu et al 2011 Shauble 2011 3 3 Comparisons with previous data A compilation of 26Mg data for peridotitic olivine worldwide suggests an overall homogeneity at current analytical uncertainties Fig 2-1a regardless of the different melt extraction and modification histories undergone by their host peridotites Handler et al 2009 Furthermore the compiled fractionation factors between clinopyroxene and olivine and between orthopyroxene and olivine reveal a broad Mg isotope equilibrium among the major mantle minerals Fig 2-5 These observations suggest a general Mg isotopic 27 Figure 2-3 Magnesium three-isotope plot relative to DSM-3 standard SC San Carlos HI Hawaiian Ol olivine Cpx clinopyroxene Opx orthopyroxene PCC-1 is an USGS peridotite standard 28 Table 2-1 Magnesium isotopic compositions of reference materials 26Mg 2SD 25Mg 2SD PCC-1 USGS peridotite 0 22 0 07 0 12 0 05 Recommended value 0 229 0 055 -0 101 0 01 Hawaiian seawater 0 84 0 07 0 42 0 05 chemistry duplicate 0 83 0 07 0 44 0 06 chemistry duplicate 0 83 0 07 0 42 0 05 Hawaiian seawater average 0 83 0 04 0 42 0 03 Recommended value 0 843 0 057 0 433 0 040 Reference Materials 2SD two times the standard deviation of multiple analyses of bracketing standards during an analytical session Chemistry duplicate indicates analyses of different aliquots of the dissolved sample that were processed individually through chemical separation Recommended values for PCC-1 and Hawaiian seawater are from Teng et al 2015 29 Figure 2-4 a Mg isotopic compositions of mineral separates from the San Carlos SC peridotites b Inter-mineral Mg isotope fractionation of Cpx-Ol and Opx-Ol pairs from the San Carlos peridotites The x-axis in a represents 26Mg values while in b it represents 26Mg between pyroxene Px and olivine Ol Data from Wiechert and Halliday W&H 2007 Young et al 2009 and Chakrabarti and Jacobsen C&J 2010 which also reported 26Mg data for Cpx-Ol and Opx-Ol pairs are plotted for comparison In addition published data for San Carlos olivine analyzed in UW and UA laboratories by other authors have also been plotted for comparison including Wang et al 2015a b 2016 Sio et al 2013 Liu et al 2010 and Yang et al 2012 The gray bar in a represents 2SD of the average mantle 26Mg 0 25 0 07 Teng et al 2010a which is shown as the black vertical line The orange and blue bars in b represent the calculated ranges of equilibrium inter-mineral fractionation of Opx-Ol and Cpx-Ol pairs respectively at temperatures between 800 and 1200 C The theoretical fractionation factors used in the calculation are from Schauble 2011 The vertical orange and blue lines in the center of the bars represent the fractionation factor at 1000 C in each case which is the average equilibrium temperature for San Carlos peridotites Galer and O Nions 1989 30 Table 2-2 Magnesium isotopic compositions of mineral separates from San Carlos peridotites 26Mg 2SD 25Mg 2SD 0 25 0 27 0 27 0 23 0 26 0 24 0 23 0 23 0 23 0 24 0 07 0 07 0 07 0 07 0 04 0 07 0 07 0 05 0 07 0 03 0 13 0 14 0 13 0 11 0 13 0 12 0 10 0 11 0 10 0 12 0 05 0 05 0 05 0 06 0 03 0 05 0 06 0 04 0 05 0 03 Cpx-1 Cpx-2 Cpx-3 Cpx-4 Harzburgitic Cpx average 0 24 0 21 0 22 0 25 0 23 0 07 0 07 0 07 0 07 0 04 0 12 0 12 0 12 0 13 0 12 0 06 0 05 0 06 0 06 0 01 Opx-1 Opx-2 Opx-3 repeat Opx-3 average Opx-4 Harzburgitic Opx average 0 22 0 26 0 19 0 21 0 20 0 21 0 22 0 07 0 07 0 07 0 07 0 05 0 07 0 05 0 10 0 12 0 09 0 09 0 09 0 11 0 10 0 06 0 06 0 06 0 05 0 04 0 06 0 03 Ol-1 San Carlos lherzolite repeat repeat average Ol-2 Lherzolitic Ol average 0 25 0 23 0 27 0 25 0 23 0 24 0 07 0 07 0 07 0 04 0 07 0 04 0 13 0 12 0 15 0 13 0 10 0 12 0 05 0 05 0 06 0 03 0 05 0 03 Minerals San Carlos harzburgite Ol-1 Ol-2 repeat chemistry duplicate Ol-2 average Ol-3 chemistry duplicate Ol-3 average Ol-4 Harzburgitic Ol average Ol olivine Cpx clinopyroxene Opx orthopyroxene Repeat indicates instrumental analyses of the same purified sample solution in different analytical session 31 homogeneity in mantle olivine and are consistent with the narrow 26Mg range of global peridotite xenoliths Teng et al 2010a Therefore the Mg isotopic composition of mantle olivine seems well buffered by its high Mg concentration while any isotopic heterogeneity is expected to be erased quickly due to the fast diffusion rate of Mg at mantle temperature e g 1 10 18 m2 s at 980 C for Fo85-93 Chakraborty 1997 Measurements on San Carlos olivine however have shown noticeable interlaboratory variation in 26Mg values ranging from 0 73 to 0 06 Fig 2-1b Pearson et al 2006 Teng et al 2007 Wiechert and Halliday 2007 Handler et al 2009 Huang et al 2009 Wimpenny et al 2010 To eliminate the potential influence from sample heterogeneity Chakrabarti and Jacobsen 2010 prepared a homogenized San Carlos olivine sample by powdering 50g of hand-picked olivine grains This sample was measured at four laboratories on three different types of instruments: Thermo Neptune Thermo Fisher Scientific Waltham MA USA GVI IsoProbe-P GV Instruments Manchester UK and Nu Plasma Nu Instruments The reported 26Mg values are similar 0 19 to 0 32 Young et al 2009 Liu et al 2010 Yang et al 2012 Bouvier et al 2013 except for a much lower 26Mg value of 0 55 obtained by Chakrabarti and Jacobsen 2010 Despite the 0 67 total variation in published 26Mg data for the San Carlos olivine Teng et al 2007 Wiechert and Halliday 2007 the results from most studies Huang et al 2009 Young et al 2009 Liu et al 2010 Wimpenny et al 2010 Pogge von Strandmann 2011 Yang et al 2012 Bouvier et al 2013 Sio et al 2013 Wang et al 2015 a b 2016 including this one are similar and fall within the mantle range For example olivine grains separated from five different San Carlos peridotite specimens by five different authors using the same analytical method as used in this study gave a narrow 32 Figure 2-5 Compilation of inter-mineral Mg isotope fractionation of a Cpx-Ol and b Opx-Ol pairs in peridotite xenoliths Data sources: Handler et al 2009 Yang et al 2009 Liu et al 2011 Pogge von Strandmann PvS et al 2011 Xiao et al 2013 Wang et al 2016 Wiechert and Halliday W&H 2007 and Chakrabarti and Jacobsen C&J 2010 The gray bars in a and b represent the calculated ranges of equilibrium intermineral fractionation of Cpx-Ol and Opx-Ol pairs respectively at temperatures between 800 and 1200 C The theoretical fractionation factors used in the calculation are from Schauble 2011 The dark gray vertical lines in the center of the bars represent the fractionation factor at 1000 C in each case 33 26Mg range of 0 32 to 0 23 Fig 2-4a Liu et al 2010 Yang et al 2012 Sio et al 2013 Wang et al 2015 a b 2016 although these measurements were performed in two laboratories UA and UW Given the limited range of results from this study and from mantle olivine in general the wide range of 26Mg values obtained from the same San Carlos olivine powder measured in different laboratories therefore suggests that analytical issues are responsible for the previous inconsistent measurements of San Carlos olivine 3 4 Evaluating the influence of resin cleaning on Mg isotopic analysis Both Teng and Yang 2014 and Chang et al 2003 have noted that the cleaning procedure for ion-exchange resin can affect the accuracy of Mg isotopic measurements as resin impurities may cause instability in instrumental fractionation The acids used in the routine resin cleaning procedure in the Isotope Laboratory of University of Washington are prepared by sub-boiling distillation from single-distilled HCl trace metal grade matrix element concentration 1 ppb Thermo Fisher Scientific using a Savillex DST-1000 acid purification unit We evaluated the influence of acid purity on resin performance by comparing the analytical results from resins cleaned with single-distilled HCl to those cleaned with double-distilled HCl In addition as ion-exchange resin may show deterioration in performance due to exposure to oxidizing HNO3 during column chemistry we also investigated the reusability of Bio-Rad AG50W-X8 resin for Mg isotopic analysis Separate batches of lherzolitic San Carlos olivine and Hawaiian seawater were processed through different types of resins either new or used resins cleaned with either double-distilled or single-distilled HCl The lherzolitic San Carlos olivine yielded 26Mg 34 Figure 2-6 Magnesium isotopic compositions of a lherzolitic San Carlos olivine and b Hawaiian seawater processed through column chemistry using different types of resins The resin codes are described in Table 2-4 The red horizontal line in each panel represent the average 26Mg value of all the measured data 35 Table 2-3 Magnesium isotopic compositions of Hawaiian seawater and lherzolitic San Carlos olivine processed through resins cleaned in different ways 26Mg 2SD 25Mg 2SD 0 84a 0 83a 0 83a 0 88 0 80 0 83 0 07 0 07 0 07 0 07 0 07 0 05 0 42 0 44 0 42 0 45 0 41 0 43 0 05 0 06 0 05 0 06 0 06 0 04 SW-NR2 0 84 0 07 0 43 0 06 SW-NR3 0 80 0 07 0 42 0 06 SW-UR1 0 82 0 07 0 40 0 06 SW-UR2 repeat repeat average 0 82 0 87 0 87 0 85 0 06 0 07 0 07 0 04 0 41 0 44 0 43 0 42 0 05 0 06 0 06 0 03 SW-UR3 repeat average 0 83 0 83 0 83 0 06 0 07 0 05 0 44 0 42 0 43 0 05 0 06 0 04 SW-UR4 repeat average Hawaiian Seawater average n 12 0 85 0 83 0 84 0 84 0 07 0 07 0 05 0 05 0 44 0 43 0 43 0 43 0 06 0 06 0 04 0 03 Lherzolitic San Carlos Ol SC-NR1 repeat repeat digestion replicate digestion duplicate repeat repeat chemistry duplicate 0 25b 0 23b 0 27b 0 23b 0 26 0 25 0 26 0 26 0 07 0 07 0 07 0 07 0 06 0 07 0 07 0 07 0 13 0 12 0 15 0 10 0 14 0 14 0 12 0 14 0 05 0 05 0 06 0 05 0 05 0 06 0 06 0 06 Samples Hawaiian Seawater SW-NR1 chemistry duplicate chemistry duplicate chemistry duplicate chemistry duplicate average 36 average 0 25 0 03 0 14 0 03 SC-NR2 repeat chemistry duplicate average 0 22 0 23 0 23 0 23 0 07 0 07 0 07 0 04 0 10 0 10 0 10 0 10 0 06 0 06 0 06 0 04 SC-NR3 repeat chemistry duplicate average 0 22 0 21 0 23 0 22 0 06 0 07 0 07 0 04 0 12 0 12 0 14 0 12 0 05 0 06 0 06 0 03 SC-UR1 repeat chemistry duplicate average 0 25 0 27 0 23 0 25 0 06 0 07 0 07 0 04 0 12 0 13 0 12 0 12 0 05 0 06 0 06 0 03 SC-UR2 repeat repeat chemistry duplicate average 0 21 0 22 0 24 0 23 0 22 0 06 0 07 0 07 0 07 0 03 0 12 0 12 0 12 0 11 0 12 0 05 0 06 0 06 0 06 0 03 SC-UR3 repeat chemistry duplicate average 0 23 0 25 0 23 0 24 0 06 0 07 0 07 0 04 0 11 0 12 0 12 0 11 0 05 0 06 0 06 0 03 SC-UR4 repeat repeat chemistry duplicate average Lherzolitic San Carlos Ol average n 24 0 22 0 24 0 21 0 24 0 23 0 24 0 06 0 07 0 07 0 07 0 03 0 03 0 11 0 13 0 11 0 12 0 12 0 12 0 05 0 06 0 06 0 06 0 03 0 02 a: data from Table 2-1 b: data from Table 2-2 Digestion replicate indicates full chemical replicates starting with sample dissolution through chemical separation and instrumental analysis 37 Table 2-4 Resin codes and associated Mg blank values Resin type Mg blank ng Cleaning procedure New resin NR NR1 3 6 New resin 3 passes of double-distilled 6 N HCl 1 pass of double-distilled 1 N HNO3 NR2 3 4 New resin 3 passes of single-distilled 6 N HCl 2 passes of double-distilled 1 N HNO3 NR3 3 7 New resin 3 passes of single-distilled 6 N HCl 1 pass of double-distilled 1 N HNO3 UR1 3 7 Used resin 3 passes of double-distilled 6 N HCl 2 passes of double-distilled 1 N HNO3 UR2 5 8 Used resin 3 passes of double-distilled 6 N HCl 1 pass of double-distilled 1 N HNO3 UR3 4 2 Used resin 3 passes of single-distilled 6 N HCl 2 passes of double-distilled 1 N HNO3 UR4 3 8 Used resin 3 passes of single-distilled 6 N HCl 1 pass of double-distilled 1 N HNO3 Used resin RR 38 values from 0 25 to 0 22 similar to the values of the other two batches of lherzolitic San Carlos olivine while the Hawaiian seawater gave 26Mg values from 0 85 to 0 80 with an average of 0 84 0 05 n 15 comparable with the recommended value of 0 843 0 057 Teng et al 2015 Fig 2-6 In addition purified Mg solutions collected from all types of resins had similar levels of Mg blanks typically 5 ng that are negligible compared with the amounts of loaded Mg 10 g Table 2-4 These results suggest that the Bio-Rad AG50W-X8 resin can be re-used for Mg separation and that single-distilled HCl can be used directly for resin cleaning without further purification Therefore previous differing measurements on San Carlos olivine must have been caused by other factors during chemical separation or instrumental analysis 3 5 Previous inter-laboratory discrepancies on San Carlos olivine: sample heterogeneity vs analytical artifacts Among all the studies where San Carlos olivine was analyzed Wiechert and Halliday 2007 and Chakrabarti and Jacobsen 2010 reported 26Mg values significantly different from those of other studies Figs 2-1b and 2-4a However the clinopyroxene olivine and orthopyroxene olivine pairs were in isotope equilibrium in both studies Fig 2-4b In comparison although the 26Mg data for San Carlos olivine and orthopyroxene reported by Young et al 2009 fall within the mantle range their clinopyroxene separates yielded much higher 26Mg values Fig 2-4a These clinopyroxene separates had 0 40 higher values than coexisting olivine grains resulting in the only disequilibrium clinopyroxene olivine pairs reported for San Carlos peridotites so far Fig 2-4b Young et al 2009 ruled out the presence of spinel inclusions as the cause of these 39 isotopically heavy clinopyroxene grains as this would require an unrealistically high 50% contribution of Mg from spinel impurities However the calculated whole-rock 26Mg values agreed better with the measured values when using the clinopyroxene 26Mg values calculated from 26MgCpx-Ol than when using the measured clinopyroxene 26Mg values Therefore these abnormally high 26Mg values of clinopyroxene might be erroneous as was pointed out by the authors On the other hand the low 26Mg value of 0 58 for San Carlos olivine reported by Pearson et al 2006 may be related to analytical issues of the laser ablation LA -MC-ICP-MS method such as the matrix effect associated with laser ablation Norman et al 2006a The over 4 Mg isotopic variation in olivine grains from different peridotite xenoliths reported in their study has not been found in other LA-MCICP-MS studies Norman et al 2006 a b Xie et al 2011 or solution aspiration MC-ICPMS studies Handler et al 2009 Yang et al 2009 Liu et al 2011 Pogge von Strandmann et al 2011 Xiao et al 2013 Wang et al 2016 As a result previous inter-laboratory disparities are more likely to reflect systematic analytical artifacts in different laboratories rather than Mg isotopic heterogeneity within the San Carlos olivine A variety of factors can result in analytical artifacts that are several times greater than the typically quoted precision 0 1 for 26Mg values Among them the levels of matrix removal and Mg recovery are of primary importance The presence of matrix elements e g Al Fe Ti Mn Ca Na K Ni even in trace amount for certain elements has been shown to cause severe drift in Mg isotopic measurement Galy et al 2001 Huang et al 2009 Wombacher et al 2009 Teng et al 2010a An et al 2014 Teng and Yang 2014 For example a Ca Mg ratio of 0 05 can cause a nearly 0 60 shift in 26Mg values Galy et al 2001 Wombacher et al 2009 due to spectral interference of doubly 40 charged 48 Ca on 24 Mg Teng and Yang 2014 This may explain the anomalously high 26Mg values obtained by Young et al 2009 as clinopyroxene hosts a considerable amount of CaO 20 wt % compared with olivine and orthopyroxene 99% purity and triple cleaned with Milli-Q 50 water in an ultra-sonic bath 10 min per wash The mineral and whole-rock samples were attacked sequentially with Optima-grade concentrated HF HNO3 HNO3 HCl and HNO3 following the method detailed in Yang et al 2009 An ultrasonic bath was used to accelerate dissolution by disrupting sample size and hence increasing the solid-acid interface All samples were finally taken up in 1 N HNO3 and loaded onto pre-cleaned BioRad cation-exchange resin AG50W-X8 200-400 mesh Matrix elements were eluted away in the first 16 mL of 1 N HNO3 and Mg was collected in the following 19 mL of the same acid This purification procedure was repeated twice to ensure maximum elimination of the matrix The total procedural blank was 30% skeletal opals Biosiliceous oozes in this study have SiO2 varying mostly between 60 wt% and 80 wt% 98 except an FeO-rich 28 wt% SiO2-poor 38 wt% radiolarian ooze from DSDP 291 Philippines The Mg content in these siliceous oozes may reflect impurities of detrital clays or volcanic ashes as their MgO Al2O3 ratios generally fall on the dilution trends of turbidites and terrigenous clays A few samples have slightly higher MgO Al2O3 ratios and higher FeO content Fig 4-2C and D suggesting uptake of Mg by biogenic silica and formation of authigenic aluminosilicate minerals e g Donnelly and Merrill 1977 Cole and Shaw 1983 In comparison calcareous oozes are mainly calcium carbonate in nature with limited Mg present as a substitute for Ca Therefore their Mg isotopic compositions could also be variably affected by more Mg-rich clay impurities 3 Analytical methods Chemical separation and isotopic analysis were performed at the Isotope Laboratories of both University of Washington Seattle and University of Arkansas Fayetteville following previously described techniques Teng et al 2007 2010a Li et al 2010 Sample powders containing approximately 50 g Mg were dissolved in clean Savillex screw-top beakers in a mixture of Optima-grade concentrated HF-HNO3-HCl at 120 C Up to three stages of cation-exchange chromatographic columns were used in this study to separate Mg from the matrix elements First the majority of Ca was removed using Bio-Rad Poly-prep 10 mL columns containing 1 mL highly cross-linked Bio-Rad AG50W-X 12 resin in 12 N HCl Ling et al 2013a This step was repeated for samples with high Ca Mg ratios 50 The solutions collected from the Ca removal column were dried down and converted to 1 N HNO3 media then passed through borosilicate glass 99 columns packed with Bio-Rad AG50W-X 8 resin This step was repeated for all samples to ensure complete removal of major matrix elements e g Na K Al and Fe Finally Mnrich samples Mn Mg 0 1 were passed through an additional Mn removal column using Bio-Rad AG50W-X8 resin in 0 5 N HCl-95% acetone Schiller et al 2010 After this step sample solutions were evaporated to dryness and converted to nitrate form for instrumental analysis Magnesium isotope ratios were measured on three different instruments including a Nu Plasma HR MC-ICP-MS at the University of Arkansas Fayetteville and both a Nu Plasma HR MC-ICP-MS and a Nu Plasma II MC-ICP-MS at the University of Washington Seattle Purified samples were diluted to 300 to 500 g kg Mg concentration in 3% HNO3 Optima-grade and introduced into the plasma source under wet plasma conditions using a MicroMist low-flow nebulizer in conjunction with a low-volume Cinnabar spray chamber Magnesium isotopic data are measured using standard-sample bracketing method Teng and Yang 2014 and are reported as relative deviations from the DSM-3 standard Galy et al 2003 in notation according to the following formula: xMg 24Mg sample Mg x -1 1000 24 Mg Mg x DSM3 where X 25 or 26 The associated error is 2SD which represents two times the standard deviation of multiple analyses of bracketing standards during an analytical session Teng et al 2015 To evaluate the accuracy of the full chemical procedure two shale standards SCo1 and SGR-1 from the United States Geological Survey USGS and a basalt standard JB1 from the Geological Society of Japan GSJ were processed along with sediment samples 100 The silty marine shale SCo-1 and the carbonate-rich shale from the Green River Formation SGR-1 yield average 26Mg values of 0 86 0 06 n 2 and 0 98 0 06 n 5 respectively while the basalt JB-1 gives an average 26Mg value of 0 25 0 03 n 5 These values are consistent with the recommended values of 0 89 0 08 1 00 0 08 and 0 28 0 10 respectively Table A1 Teng et al 2015 Teng 2017 The accuracy of column chemistry and the external reproducibility are further examined by processing and analyzing two in-house standards San Carlos olivine and Hawaiian seawater with each batch of column chemistry and instrumental analysis The lherzolitic San Carlos olivine yields an average 26Mg value of 0 25 0 04 n 28 comparable to literature average of 0 24 0 03 Hu et al 2016b the Hawaiian seawater yields an average 26Mg value of 0 84 0 05 n 20 identical to the recommended value of 0 84 0 06 Teng et al 2015 The results from repeat duplicate and replicate analyses agree within 0 05 for both 26Mg and 25Mg Table 4-A1 and Table 4-1 even though some of the samples were processed in two different laboratories and measured on three different instruments Therefore the overall analytical uncertainty over the course of this study is compatible with the long-term external precision of this laboratory i e better than 0 07 for 26Mg and 0 05 for 25Mg Teng et al 2015 4 Results Magnesium isotopic compositions of marine sediments are reported in Table 4-1 and are plotted in Fig 4-3 The linear regression between 26Mg and 25Mg values of all samples and standards analyzed in this study yields a slope of 0 515 R2 0 997 Hence only 26Mg values are presented and discussed hereafter 101 Table 4-1 Magnesium isotopic compositions of marine sediments from various DSDP and ODP sites outboard of subduction zones Sample Unit Depth Litho Age 26Mg 2 SD 25Mg 2 SD Description m Ma DSDP Site 174 Cascadia 174A-1-1-62-67 1 0 62 TB 0 00 0 20 0 06 0 12 0 06 silty clay 174A-2-6-47-53 1 45 47 TB 0 20 0 33 0 07 0 15 0 05 clay Repeat 0 37 0 07 0 19 0 03 Duplicate 0 39 0 07 0 21 0 05 repeat 0 39 0 07 0 21 0 05 average 0 38 0 04 0 20 0 02 174A-8-3-93-98 1 98 43 TB 0 44 0 40 0 06 0 21 0 06 clay repeat 0 41 0 07 0 22 0 03 duplicate 0 43 0 07 0 21 0 05 repeat 0 41 0 07 0 20 0 03 average 0 41 0 03 0 21 0 02 174A-8-6-57-62 1 102 57 TB 0 46 0 49 0 06 0 24 0 06 sandy 174A-19-2-107-113 1 201 57 TB 0 90 0 50 0 06 0 23 0 06 sand duplicate 0 49 0 07 0 24 0 05 average 0 49 0 05 0 24 0 04 174A-19-4-32-36 1 203 82 TB 0 91 0 41 0 06 0 19 0 06 silty 174A-27-2-125-130 1 277 75 TB 1 24 0 49 0 06 0 22 0 06 silty clay duplicate 0 50 0 07 0 26 0 05 average 0 49 0 05 0 24 0 04 174A-33-2-106-111 2 344 06 TB 1 54 0 40 0 06 0 20 0 06 silty clay 174A-33-2-132-137 2 344 32 TB 1 54 0 57 0 06 0 28 0 06 sand duplicate 0 58 0 07 0 29 0 05 average 0 57 0 05 0 29 0 04 174A-37-2-123-127 2 505 73 TB 2 15 0 44 0 06 0 20 0 06 clay duplicate 0 40 0 07 0 21 0 05 repeat 0 44 0 07 0 22 0 03 average 0 43 0 04 0 22 0 02 174A-37-2-134-137 2 505 84 TB 2 15 0 53 0 06 0 23 0 06 sand duplicate 0 54 0 07 0 25 0 05 average 0 54 0 05 0 24 0 04 174A-39-2-22-24 2 751 72 TB 7 94 0 21 0 06 0 11 0 06 clayst 174A-39-2-145-150 2 752 95 TB 8 00 0 28 0 07 0 13 0 07 sandst repeat 0 30 0 06 0 15 0 06 repeat 0 30 0 06 0 15 0 04 duplicate 0 33 0 06 0 17 0 07 102 repeat 0 29 0 06 0 12 0 06 repeat 0 31 0 06 0 16 0 04 average 0 30 0 02 0 15 0 02 178-1-88-94 DSDP Site 178 - Alaska repeat 0 20 0 07 0 10 0 07 dark green-green silty 0 06 clay replicate 0 22 0 08 0 11 0 05 average 0 21 0 04 0 11 0 03 178-12-2-112-117 1 TB 98 62 TB 0 00 0 20 0 06 0 11 replicate 0 25 0 08 0 14 0 07 dark green-green silty 0 05 clay average 0 25 0 05 0 14 0 04 178-24-3-87-92 1 0 88 1 207 87 TB 0 94 0 25 0 07 0 14 2 00 0 18 0 07 0 08 0 06 green silty clay replicate 0 19 0 08 0 07 0 05 average 0 19 0 05 0 08 0 04 178-29-5-129-134 1 256 29 TB 2 23 0 12 0 06 0 08 0 07 green silty clay replicate 0 13 0 09 0 06 0 07 average 0 12 0 05 0 07 0 05 0 07 ash 178-39-5-109-110 2 398 59 VC 0 14 0 09 0 08 178-47-1-127-130 2 506 77 TB 3 30 0 19 0 07 0 06 0 07 turb top clay repeat 0 15 0 07 0 07 0 06 repeat 0 18 0 07 0 08 0 05 duplicate 0 19 0 06 0 09 0 04 replicate 0 19 0 09 0 09 0 07 average 0 18 0 03 0 08 0 02 178-47-1-141-145 2 506 91 TB 3 30 0 18 0 07 0 09 0 06 Turb bottom sand duplicate 0 18 0 07 0 09 0 05 replicate 0 19 0 09 0 09 0 07 average 0 18 0 04 0 09 0 03 178-49-1-66-71 2 591 66 TB 4 09 0 06 0 09 0 03 0 06 turb top clay repeat 0 06 0 07 0 02 0 07 average 0 06 0 06 0 03 0 04 178-49-1-85-90 2 591 85 TB 4 10 0 16 0 06 0 09 0 07 turb bottom sand repeat 0 17 0 07 0 08 0 05 repeat 0 15 0 07 0 08 0 06 duplicate 0 16 0 07 0 08 0 06 average 0 16 0 03 0 08 0 03 178-52-2-127-131 2 688 77 TB 5 00 0 24 0 06 0 09 0 06 diatom clay repeat 0 23 0 07 0 09 0 06 repeat 0 23 0 06 0 10 0 04 duplicate 0 21 0 06 0 11 0 04 103 replicate 0 20 0 09 0 09 0 07 average 0 22 0 03 0 10 0 02 0 06 0 09 0 02 0 06 brown clay 178-54-4-109-113 3 747 59 HT 178-57-1-136-140 3 769 36 HT 45 00 0 44 0 09 0 23 0 06 brown clay repeat 0 47 0 06 0 22 0 06 duplicate 0 45 0 06 0 23 0 04 replicate 0 46 0 09 0 23 0 07 average 0 46 0 04 0 23 0 03 211-1-2-36-41 DSDP Site 211 - Nicobar 1 duplicate 1 86 TB average 211-1-3-126-127 1 4 26 VC duplicate average 0 50 0 48 0 10 0 27 0 07 ash-rich diatom ooze 0 48 0 05 0 25 0 05 0 48 0 05 0 26 0 04 0 41 0 37 0 10 0 20 0 07 ash layer 0 34 0 05 0 17 0 05 0 35 0 05 0 18 0 04 211-2-4-11-16 1 13 61 TB 2 00 0 30 0 05 0 14 0 05 clay-rich diatom ooze 211-5-2-36-40 2 134 86 TB 3 28 0 06 0 05 0 02 0 06 rad-rich clay repeat 0 04 0 07 0 03 0 04 repeat 0 03 0 06 0 02 0 05 average 0 05 0 03 0 02 0 03 211-6-4-50-55 2 185 59 TB 3 82 0 43 0 10 0 22 0 07 silty clay duplicate 0 42 0 05 0 22 0 05 average 0 42 0 05 0 22 0 04 211-6-4-60-63 2 185 69 TB 3 82 0 36 0 10 0 18 0 07 silty sand repeat 0 37 0 05 0 18 0 05 average 0 37 0 05 0 18 0 04 211-7-1-131-136 3 229 31 TB 4 28 0 91 0 07 0 45 0 07 silty sand repeat 0 90 0 06 0 44 0 06 duplicate 0 91 0 05 0 47 0 05 average 0 90 0 03 0 45 0 03 211-9-1-109-115 3 295 59 TB 4 98 0 32 0 06 0 17 0 06 silty clay 701A-1H-4-145-150 1A 5 9 ODP Site 701 - South Sandwich 701B-1H-5-145-150 1A 77 4 TC 0 20 0 20 0 07 0 11 0 06 diatom ooze ash TC 3 88 0 12 0 07 0 08 0 06 diatom ooze ash 701B-9H-5-46-50 1B TC 4 77 0 38 0 07 0 20 0 06 diatom ooze ash 701C-27X-2-140-150 2A 246 7 TC 11 40 0 09 0 06 mud diatom 701C-31X-2-140-150 2A 284 7 TC 18 80 0 03 0 07 0 03 0 06 silic clay 701C-42X-1-140-150 2A 387 7 TC 32 00 0 18 0 07 0 11 0 06 clay duplicate 0 18 0 07 0 10 0 06 average 0 18 0 05 0 11 0 04 150 104 0 07 0 05 321-1-53-58 DSDP Site 321 - Peru 321-3-6-24-30 1 0 53 TC 0 5 0 32 0 06 0 16 0 06 siliceous detrital clay 1 19 08 TC 1 9 0 30 0 06 0 14 0 06 siliceous detrital clay 321-4-2-34-40 1 22 34 TC 2 2 0 26 0 06 0 14 0 06 siliceous detrital clay 321-5-4-27-33 2 34 77 TC 4 7 0 24 0 06 0 13 0 06 volcanic glass clay 321-5-4-145-150 2 35 95 VC 5 1 0 38 0 06 0 20 0 07 ash duplicate 0 35 0 07 0 18 0 06 average 0 36 0 05 0 19 0 04 0 22 0 07 0 09 0 07 volcanic-glass rich clay repeat 0 22 0 07 0 10 0 06 average 0 22 0 05 0 10 0 05 321-6-4-33-39 321-7-1-93-97 2 3 44 33 TC 49 93 HD 8 1 10 1 0 12 0 07 0 03 0 07 brown zeolite clay duplicate 0 11 0 07 0 06 0 05 average 0 11 0 05 0 05 0 04 495-3-1-25-31 1 19 25 TC DSDP Site 495 - Central America 495-4-5-20-26 1 34 14 TC 0 5 0 27 0 07 0 14 0 06 hemipelagic mud 1 1 0 13 0 07 0 08 0 06 biogenic silty mud 495-13-3-31-37 1 117 31 HD 4 4 0 00 0 06 diatomaceous mud 495-16-7-30-35 1 150 78 HD 8 5 0 04 0 06 0 02 0 06 siliceous mud 2 8 0 20 0 10 0 10 0 07 rad diatom clay grey 0 21 0 07 0 11 0 05 0 21 0 06 0 11 0 04 0 21 0 07 0 11 0 06 rad nanno ooze grey repeat 0 23 0 10 0 10 0 07 repeat 0 22 0 07 0 11 0 05 average 0 22 0 05 0 11 0 04 844B-4H-4-56-57 IB 28 6 ODP Site 844B - Central America repeat TC average 844B-6H-6-102-104 844B-21X-6-100-102 IB 51 03 TC IIB 193 5 BC 7 7 0 07 0 02 14 3 1 34 0 05 0 70 0 05 nanno ooze light grey repeat 1 33 0 07 0 68 0 05 average 1 34 0 04 0 69 0 03 844B-30X-6-145-150 IIB 280 6 BC 17 0 0 53 0 10 0 25 0 07 nanno ooze light grey repeat 0 54 0 07 0 28 0 05 repeat 0 56 0 05 0 28 0 05 average 0 55 0 04 0 27 0 03 0 16 0 07 0 08 0 07 silty yellow clay duplicate 0 14 0 07 0 05 0 06 average 0 15 0 05 0 06 0 04 0 03 DSDP Site 291 - Philippines 291-1-1-130-135 1 1 3 TC 0 4 291-2-1-133-139 1 61 33 TC 20 4 0 01 0 07 brown silty clay 291-3-1-126-132 3 80 26 HD 26 8 0 04 0 07 0 02 0 07 yellow rad ooze 105 0 07 291-4-1-88-94 4 98 88 HD 33 0 0 12 0 07 0 07 291-4-2-132-137 4 100 82 HD 33 6 0 00 0 07 0 00 0 07 brown nanno-rad silty 0 07 clay brown rad zeolite clay repeat 0 01 0 07 0 01 0 05 average 0 01 0 05 0 01 0 04 0 07 0 07 0 04 0 07 brown Fe-zeolite clay repeat 0 05 0 06 0 03 0 04 duplicate 0 07 0 06 0 03 0 04 average 0 06 0 04 0 03 0 03 291-4-4-62-67 4 103 12 HD 34 4 DSDP Site 294 5 - Ryukyu 294-1-3-56-61 0 23 0 07 0 09 0 07 silt-rich clay duplicate 0 22 0 07 0 12 0 06 average 0 23 0 05 0 11 0 04 0 07 0 07 0 05 0 07 brown clay repeat 0 06 0 07 0 04 0 05 repeat 0 06 0 07 0 05 0 06 duplicate 0 07 0 07 0 04 0 05 average 0 07 0 04 0 04 0 03 294-4-1-110-115 294-4-4-45-50 1 1 2 3 56 94 1 TC HD 97 95 HT 2 0 53 5 55 0 0 58 0 06 0 30 0 04 brown-black clay repeat 0 54 0 07 0 28 0 05 average 0 57 0 04 0 29 0 03 294-6-1-118-123 2 106 68 HT 55 0 0 15 0 07 0 06 0 07 pebbly black Fe clay duplicate 0 18 0 05 0 09 0 05 average 0 17 0 04 0 08 0 04 0 10 0 07 0 06 0 07 dark brown clay repeat 0 09 0 07 0 04 0 06 average 0 10 0 05 0 05 0 05 295-2-1-34-39 295-2-6-84-89 2 120 34 HD 2 128 34 HT 51 6 55 0 0 48 0 06 0 25 0 06 black-brown clay duplicate 0 47 0 07 0 25 0 05 repeat 0 49 0 07 0 24 0 03 average 0 48 0 04 0 25 0 02 295-3-2-30-35 2 140 8 HT 55 0 0 50 0 06 0 24 0 06 black clay 1 1 05 1 05 0 18 0 06 0 11 0 07 yellow brown clay DSDP Site 595 6 - Tonga 596-1-1-105-108 TC repeat 0 14 0 07 0 06 0 06 duplicate 0 14 0 10 0 05 0 07 repeat 0 18 0 05 0 10 0 05 average 0 17 0 03 0 08 0 03 596-1-cc 1 5 50 TC 18 48 0 12 0 07 0 07 106 0 03 medium brown clay duplicate 0 12 0 07 0 05 0 05 average 0 12 0 05 0 06 0 03 595A-2-3-57-60 HD 21 12 0 01 0 07 0 01 0 07 brown black clay repeat 0 02 0 06 0 00 0 04 duplicate 0 01 0 05 0 01 0 05 average 0 01 0 00 0 03 595A-2-cc 1 2 6 37 0 03 12 20 HD 47 22 0 18 0 06 0 09 0 04 dark brwn Mn nodule repeat 0 15 0 07 0 09 0 06 average 0 17 0 05 0 09 0 03 596-2-6-70-72 2 13 70 HD 54 55 0 02 596-3-1-115-117 2 16 25 HD 67 0 09 0 01 0 06 very dark brown clay 0 04 0 09 0 02 0 06 dark brown clay repeat 0 05 0 07 0 02 0 06 average 0 05 0 06 0 02 0 04 596-3-5-100-102 2 22 10 HD 69 36 0 11 0 09 0 06 0 06 dark brown clay 596-5-2-114-116 3 36 94 HD 75 34 0 07 0 06 0 05 0 06 dark brown clay chert duplicate 0 06 0 06 0 02 0 04 average 0 06 0 04 0 03 0 03 38 00 HD 75 77 0 22 0 09 0 11 repeat 0 21 0 07 0 10 0 06 yellow brown 0 06 porcellanite repeat 0 21 0 07 0 11 0 06 duplicate 0 22 0 07 0 11 0 05 average 0 21 0 04 0 11 0 03 57 60 HD 83 67 0 09 0 09 0 05 0 06 yellow chert repeat 0 11 0 06 0 06 0 06 average 0 10 0 05 0 05 0 04 595A-5-cc 595A-8-5-100-103 3 3 Note: repeat repeat instrumental analysis on the same Mg-cut solution Duplicate means repeat column chemistry and instrumental analysis Replicate means re-dissolution column chemistry and instrumental analysis denotes samples analyzed on a MC-ICPMS at the University of Washington Seattle denotes samples analyzed on a MC-ICPMS at the University of Arkansas Fayetteville Samples without the star specification were analyzed on a Nu II MC-ICPMS at the University of Washington Seattle Lithological code: TB turbidite TC terrigenous clay HD hydrogenetic HT hydrothermal BC biogenic carbonate 107 Figure 4-3 Magnesium isotopic compositions of subducting marine sediments A 26Mg variation as a function of drill site latitude Symbols with the same colors represent samples from the same drill sites B Histogram of 26Mg in subducting sediments shows a similar distribution frequency to that of C upper continental rocks and sediments Data sources: marine sediments: this study and Teng et al 2016 for Antilles sediments upper crustal rocks and sediments: Shen et al 2009 Li et al 2010 2014 Liu et al 2010 Telus et al 2012 Huang et al 2013 Ling et al 2013b Liu et al 2014 Wimpenny et al 2014b and Wang et al 2015a 2015b 26Mg ranges of seawater Ling et al 2011 and mantle Teng et al 2010a are plotted as references 108 The 26Mg values of the 77 studied bulk sediments span a wide range comparing to global seawater and the mantle varying from 1 34 to 0 46 Fig 4-3A and B Among the ten drill sites located between 60 N and 60 S those close to the equator display the greatest variability as would be expected from the higher levels of biogenic inputs In general terrigenous clays and hydrogenetic clays occupy the high- 26Mg end while calcareous oozes reside at the low- 26Mg end Turbidites and hydrothermal clays have a wide spread of 26Mg with values both higher and lower than the mantle range Overall the 26Mg variation covers a similar spectrum to that reported for rocks and sediments on continents Fig 4-3B and C There is no uniform down-core variation with respect to either MgO or 26Mg except that the terrigenous clays from DSDP 321 Peru display a monotonic increase in 26Mg with both depth and age Detailed results for the subducting marine sediments are reported below as three broad lithological groups: 1 detrital sediments turbidites terrigenous clays and associated volcanic ashes 2 hydrogenetic and hydrothermal clays and 3 biogenic siliceous and calcareous oozes 4 1 Detrital sediments 4 1 1 Turbidites Turbidites are represented by sand-silt-clay sequences that are deposited relatively rapidly and typically close to land The three turbidite sections from Cascadia DSDP 174 Alaska DSDP 178 and Nicobar Fan DSDP 211 have 26Mg values comparable to but less variable than the turbidites-dominated mudrocks from the British Caledonides 26Mg 0 24 0 43 2SD Wang et al 2015b and suspended load bedload sediments from 109 rivers that drain dominantly silicate catchments 26Mg 0 29 0 59 2SD Pogge von Strandmann et al 2008 Wimpenny et al 2011 Furthermore the three sections show minor yet distinguishable Mg isotopic characteristics Fig 4-3A DSDP sites 174 and 178 are both located within the Northeast Pacific Turbidite Province Horn et al 1970 and sediments from these sites contain primarily of detrital mica with minor amounts of montmorillonite amphibole chlorite and dolomite Zemmels and Cook 1973 Site 174 located in the Astoria Submarine Fan contains abundant landderived detritus Based on their Nd-Hf-Pb isotopes the majority of the turbidites are sourced from Precambrian terranes that are over 1000 km away inland with only the top layer of sediments being supplied by the Columbia River Basalt Prytulak et al 2006 The shallowest sample of the Cascadia turbidites has a 26Mg value 0 20 that is comparable to the Columbia River Basalts 0 33 to 0 15 Liu et al 2014 but is distinguishably heavier than the deeper samples This shift in sediment 26Mg is consistent with the change in sediment provenance The rest of the section can be further divided into two units separated by a seismic discontinuity while they share a similar range in Hf-Nd-Pb isotopes Prytulak et al 2006 Similarly the 26Mg values range from 0 49 to 0 38 for Unit 1 and from 0 57 to 0 40 for Unit 2 In addition the claystone and sandstone at the bottom of the section have higher 26Mg values 0 21 and 0 30 than the unconsolidated sediments Site 178 located in the Gulf of Alaska is characterized by highly uniform petrologic and climatologic conditions on the continent and a strong nepheloid layer Ewing et al 1970 Sediments there are rich in mica and chlorite and contain appreciable ice-rafted erratics Zemmels and Cook 1973 reflecting the dominance of physical 110 weathering and low intensity of chemical weathering at high latitude regions Griffin et al 1968 They display a more restricted 26Mg range of 0 25 to 0 12 apart from a slightly heavier clay 26Mg 0 06 These values are similar to those of fresh oceanic basalts 0 26 0 07 Teng et al 2010a and glacial flour from the Canadian shield 0 165 to 0 106 Tipper et al 2012 Site 211 is located in the eastern equatorial Indian Ocean The turbidites supplied by the Ganges-Brahmaputra River contain abundant illite with lesser amounts of smectite chlorite and kaolinite in the 2 m fractions Mica with small amounts of amphibole are the dominant Mg-bearing minerals in the coarser fractions Venkatarathnam 1974 With the exception of an isotopically light carbonate-rich silty sand 26Mg 0 90 the three other silty clay sand samples have similar 26Mg values 0 42 to 0 32 The silicic volcanic ash sample derived from the Indonesian Island arc contains abundant smectite and has a similar 26Mg value of 0 35 4 1 2 Terrigenous clays This group consists mainly of land-derived detritus transported by eolian processes Consequently sediments from this group are dominated by fine-grained clay minerals Sediments from DSDP site 321 and ODP site 701 are representative samples of this group Site 321 is located on the eastern edge of the Nazca Plate where the Peru-Chile trench effectively traps windborne volcanic and terrigenous sediments from the Andes Quilty et al 1976 Clay sediments from the top two units have a fairly uniform chemical composition with high contents of volcanic glass and ash The Mg-bearing minerals presented include kaolinite chlorite and mica illite as well as montmorillonite that is 111 closely associated with volcanic glass Zemmels and Cook 1976 These clays and a single ash sample have similar 26Mg values 0 36 to 0 22 that fall within the range of typical igneous rocks Li et al 2010 Liu et al 2010 Teng et al 2010a The three terrigenous clays from central America DSDP site 495 and ODP site 844 have 26Mg values 0 27 to 0 13 that mainly reflect Mg hosted in the volcanic glass and smectite Latouche and Maillet 1982 In contrast to the largely mantle-like 26Mg values of clays from the Peru and Central America trenches those from ODP site 701 which is located on the western flank of the Mid-Atlantic Ridge have generally higher 26Mg values 0 03 to 0 18 Similarly high 26Mg values have been reported for terrigenous clays from DSDP site 543 North Antilles Teng et al 2016 In addition to those clay-dominated sites sediment sections cored at DSDP sites 291 294 295 and 595 596 which consist primarily of hydrogenetic clays see next section also contain a thin layer of terrigenous clays on top with 26Mg values varying from 0 23 to 0 03 4 2 Hydrogenetic and hydrothermal clays Hydrogenetic sediments are collected mainly from DSDP sites 291 294 295 and 595 596 which are cored on the outer swell of the Philippine trench in the northeastern West Philippine Basin and in the south central west Pacific respectively Hydrogenetic clays from sites 291 and 294 295 display a significantly negative Ce anomaly that is indicative of seawater derivation Piper 1974 Their Mg is mainly hosted in montmorillonite with minor mica and clinoptilolite Cook et al 1975 In comparison the extremely low sedimentation rate at sites 595 596 leads to the deposition of metalliferous clays rich in hydrogenetic amorphous oxides up to 90% Fe-Mn micronodules and fish 112 teeth The 14 hydrogenous clays from various sites have relatively uninform 26Mg values with an average of 0 02 0 14 except for a single Mn nodule from site 595 that has a slightly lower 26Mg value of 0 17 Sediment sections from sites 294 295 Ryukyu also contain a lower unit that is enriched in hydrothermal clays This unit is composed of iron oxides goethite and hematite zeolite and manganese micronodules in addition to the clay fraction that is dominated by montmorillonite Cook et al 1975 These FeO-rich up to 42 wt% hydrothermal clays generally have low 26Mg values 0 57 to 0 48 except for a pebbly black clay which has the lowest FeO concentration 17 wt% and a 26Mg value 0 17 similar to detrital clays In comparison the two basal brown clays from Alaskan turbidites are isotopically heavy with 26Mg values of 0 06 and 0 46 These two Alaskan clays are dominated by montmorillonite and kaolinite in composition and have much lower FeO contents 8 9 wt% Zemmels and Cook 1973 4 3 Biogenic sediments Biogenic sediments in this study mainly include radiolarian diatom oozes and calcareous oozes which are commonly abundant at sites in the equatorial region For example ODP site 844B is located within the eastward-flowing North Equatorial Countercurrent and close to the Costa Rica Dome The two nanno oozes from the lower unit are characterized by low 26Mg values 0 55 and 1 34 similar to those of the siliceous chalk oozes from DSDP site 144 Teng et al 2016 The five diatomaceous oozes from ODP site 701 South Sandwich and DSDP site 211 Nicobar Fan and a single radiolarian nanno ooze from ODP site 844B have been mixed with varying proportions of 113 volcanic ash clay and mud These samples thus display a positive correlation between 26Mg and MgO contents and have 26Mg values 0 48 to 0 12 that are comparable to the detrital clays Fig 4-4 By contrast an FeO-rich 28 wt% yellow radiolarian ooze from DSDP site 291 yields a 26Mg value 0 04 that is similar to those of the hydrogenetic clays The zeolitic radiolarian-bearing chert and porcellanite from DSDP site 595 Tonga have 26Mg values of 0 10 and 0 21 respectively slightly higher than those of the hydrogenetic sediments from the same site 5 Discussion The Mg isotopic composition of subducting marine sediments investigated here ranges from 1 34 to 0 46 the variation of which mirrors that of rocks and sediments from the upper continental crust Fig 4-3 This similarity is consistent with the dominance of land-derived detrital components in the subducting marine sediments In this section we first explore the mechanisms of Mg isotope fractionation in different types of marine sediments Then we calculate the sedimentary Mg input to global subduction zones Finally we discuss the impact of crustal recycling on arc magmatism and mantle Mg isotopic heterogeneity 5 1 Mechanisms of Mg isotope fractionation in marine sediments The overall distribution of Mg and its isotopes in marine sediments are primarily controlled by 1 the relative proportions of different phases with distinct Mg isotopic signatures and 2 the processes by which Mg is transported into the ocean and incorporated into the sediments In detail different mechanisms are responsible for Mg 114 Figure 4-4 Positive correlations of 26Mg with MgO in sediments from A ODP 701 South Sandwich and B DSDP 211 Nicobar Fan indicate a mixing trend between Mgdepleted siliceous oozes and Mg-rich detrital clays The yellow bar in this and following figures represent the mantle range 26Mg 0 25 0 07 Teng et al 2010a 115 isotope fractionation in different types of samples The isotopically light calcareous oozes are consistent with the preferential uptake of light Mg isotopes from seawater during precipitation of carbonates Meanwhile the 26Mg values of calcareous and siliceous oozes may be biased by the presence of various proportions of detrital components Detrital sediments i e turbidites and terrigenous clays display a mineralogy and size dependence on 26Mg Fractionation in hydrogenetic clays may be related to clay authigenesis while variation in hydrothermal clays is presumably associated with exchange with seawater hydrothermal fluids These factors are considered in the following sections in turn 5 1 1 Mineralogical control on marine sediment 26Mg values The subducting marine sediments studied here to a first order can be viewed as mixtures between silicates and carbonates which differ significantly in their Mg isotopic compositions Teng 2017 Carbonates are enriched in light Mg isotopes Saenger and Wang 2014 whereas clay minerals and soils formed as weathering residue are generally more enriched in heavy Mg isotopes compared to the unweathered parent rocks Teng et al 2010b Huang et al 2012 Liu et al 2014 Nonetheless both silicate- and carbonaterich sediment groups display large variations in 26Mg see Saenger and Wang 2014 Teng 2017 for recent reviews Sediment mineralogy carbonate vs clay can be estimated by Rb Sr and CaO TiO2 ratios as Rb and Ti are mainly retained in detrital phases whereas Ca and Sr are much more enriched in carbonates Fig 4-5A The broad inverse relationship between 26Mg value and CaO TiO2 ratio in both marine and continental sediments is consistent with the 116 Figure 4-5 Mineralogical control on Mg isotopic composition of marine sediments A Negative correlation between Rb Sr and CaO TiO2 ratios suggests that clay and carbonate are the two major mineral constituents of these sediments B Negative relationship between 26Mg values and CaO TiO2 ratios indicates an increasingly light Mg isotopic composition of the sediments with increasing amounts of carbonate present in the sediments The open diamond symbols in Fig 4-5 B represent literature data of various upper crust rocks and sediments see Fig 4-3 caption for source information 117 mineralogy as a first order control with the calcareous oozes being the low- 26Mg endmember 0 55 and 1 34 among the measured sediments Fig 4-5B Their bulk 26Mg values are similar to those of the carbonate-rich CaO 20 wt% sediments from DSDP site 144 reported by Teng et al 2016 whereas are considerably higher than the leached fractions of calcareous sediments from the Atlantic 1 78 1 29 n 13 and Pacific oceans 2 43 1 63 n 9 Rose-Koga and Albar de 2010 This shift in 26Mg toward higher values may reflect the presence of clay impurities as supported by the positive correlations between their 26Mg values and Cs and Rb contents which have a strong affinity for clay minerals Fig 4-6 Sawhiney 1972 Clay minerals are the dominant constituent of lithogenous component in marine sediments and they display a wide range in 26Mg both within and between different clay mineral groups e g 2 variability Wimpenny et al 2014a This variation in 26Mg is proportional to the Mg in structural versus exchangeable sites with the former favoring heavy Mg isotopes and the latter preferentially taking up light Mg isotopes from fluids Wimpenny et al 2014a In addition to various clay minerals turbidites could preserve significant amount of primary mafic silicates which generally have a mantle-like 26Mg value except for isotopically light garnets Li et al 2011 Wang et al 2012 Therefore mixing of different types of silicate phases and Mg isotopic heterogeneity in clay minerals can lead to large Mg isotopic variation in silicate-rich sediments 5 1 2 Detrital sediments: source signature and sorting effects Magnesium in detrital sediments is mainly derived from the continents and their 26Mg values are affected principally by source composition weathering and sorting 118 Figure 4-6 Positive correlations between 26Mg values and Cs Rb contents in carbonaterich sediments which indicate the influence of clay impurities on the bulk 26Mg values of these sediments 119 history 5 1 2 1 Source signature: provenance heterogeneity The provenance of the studied sediments is diverse as indicated by their wide variation in Nd-Pb-Hf-Li isotopes Plank and Langmuir 1998 Bouman et al 2004 Chan et al 2006 Prytulak et al 2006 Carpentier et al 2008 2009 Vervoort et al 2011 Tang et al 2014 Since different types of protoliths are likely to carry a distinguishable Mg isotopic fingerprint 26Mg variation in detrital sediments may partly reflect their source heterogeneity For example while most granites have a mantle-like 26Mg Li et al 2010 Liu et al 2010 those incorporated weathered crustal components have been found to be more enriched in heavy Mg isotopes Shen et al 2009 Li et al 2010 On the contrary some continental granites and basalts are isotopically lighter relative to the mantle as light as 0 47 which may reflect metasomatism of their mantle sources by low- 26Mg carbonates Yang et al 2012 Huang et al 2015 Liu et al 2015 Ke et al 2016 Tian et al 2016 Wang et al 2016 Metasomatism by carbonate-rich fluids has also been proposed to account for some low- 26Mg granulites Yang et al 2016 As for sedimentary protoliths some siliciclastic rocks e g mudrocks and shales can be isotopically light if containing appreciable amounts of carbonates Wang et al 2015b Teng 2017 Otherwise they are similar to the mantle e g greywacke and tillite Li et al 2010 or more often shifting toward high- 26Mg end Li et al 2010 Ma et al 2015 The negative La Sm vs Nd i correlation in these sediments indicates a mixed source between mature evolved upper crustal materials and more juvenile crustal materials recent mantle derivatives and seawater where a higher La Sm ratio and a more 120 negative Hf i value indicate derivation from a continental source Fig 4-7A The terrigenous clays from Peru and Central America and turbidites from Alaska have Nd i values close to zero and relatively low La Sm ratios These sediments received significant detrital contributions from volcanic arcs or juvenile terranes which are expected to have 26Mg values similar to the oceanic basalts Their relatively homogeneous and mantle-like 26Mg values are consistent with this expectation Fig 4-7B On the other hand samples with lower Nd i values and higher La Sm ratios which indicate an increased input from more mature continental sources display a large dispersion in 26Mg Fig 4-7B consistent with the prolonged intra-crustal processes e g melting weathering and sedimentation underwent by these sources Sediments from Cascadia DSDP 174 Nicobar Fan DSDP 211 and Antilles margins DSDP 144 and 543 have the most enriched radiogenic Nd-Hf-Pb isotopes Among them the Cascadia turbidites have a distinguishably light Mg isotopic composition Their Pb isotopes suggest predominant derivation from a distal Proterozoic source from North America mixed with a proximal primitive source like the Columbia River Basalts Prytulak et al 2006 The 26Mg values in Cascadia turbidites are positively correlated with Nd i values Fig 4-7B and negatively correlated with 207 Pb 204Pb and La Sm ratios Fig 4-8A and B These correlations suggest that the basaltic source has a mantle-like 26Mg signature while the more evolved source has a low 26Mg signature The latter can be attributed to the presence of isotopically light carbonate particularly Mg-rich dolomite Zemmels and Cook 1973 and garnet Prytulak et al 2006 in these sediments Sediments from the Antilles margins on the contrary are characterized by heavy Mg isotopic compositions Teng et al 2016 These sediments have Nd-Hf-Pb-Li isotopes 121 Figure 4-7 26Mg variation in subducting sediments with relation to their contrasting provenance compositions A The negative correlation between La Sm and Nd i in detrital sediments indicates their source compositions ranging from old continental crust to juvenile crust B 26Mg variation in detrital sediments as a function Nd i Neodymium isotopic data are from Carpentier et al 2008 and Vervoort et al 2011 122 Figure 4-8 The effect of source rock heterogeneity on Mg isotopic composition of subducting sediments A and B Correlations between 26Mg 207Pb 204Pb and La Sm in Cascadia turbidites suggest isotopic mixing between an evolved continental source with a low 26Mg signature and a primitive source with mantle-like 26Mg Pb isotopic data are from Prytulak et al 2006 C and D Correlations between 26Mg K2O and TiO2 Al2O3 in South Sandwich terrigenous sediments indicate source mixing between a weathered high- 26Mg continental source and a juvenile volcanic source with mantle-like 26Mg 123 that indicate derivation from a highly weathered and hence high- 26Mg low- 7Li old continental source on the South American craton Westbrook et al 1984 White et al 1985 In comparison sediments from ODP site 701 South Sandwich have intermediate Nd i values and La Sm ratios Their 26Mg values are positively correlated with MgO and K2O concentrations Fig 4-4A and Fig 4-8C and negatively correlated with Nd i and TiO2 Al2O3 Fig 4-7B and Fig 4-8D These correlations suggest a binary mixing between volcanic ash and K-rich terrigenous clay likely illite which is the dominant type of clay minerals in marine sediments from the Atlantic Ocean Riley and Chester 1971 Specifically the three ash mud-bearing siliceous oozes have less negative Nd i values and mantle-like 26Mg values 0 38 to 0 12 consistent with their volcanogenic origin whereas the three terrigenous clay samples have lower Nd i values and higher 26Mg values 0 03 to 0 18 The highest 26Mg value is identical to an illite clay sample measured by Wimpenny et al 2014a 5 1 2 2 Source signature: intensity of chemical weathering Chemical weathering exerts a strong control on the geochemistry and mineralogy of siliciclastic sediments e g Nesbitt and Young 1982 McLennan et al 1993 Magnesium isotopes are also significantly fractionated during this process with preferential loss of light isotopes to fluids and accumulation of heavy ones in the weathered residue see Teng 2017 for a recent review Notably the highest 26Mg values reported so far up to 1 8 were found in bauxites that formed via extreme weathering of basalts with great Mg depletion MgO 0 12 to 0 25 wt% Liu et al 2014 The degree of chemical weathering underwent by sediment source rocks can be 124 quantified by a number of chemical indices The chemical index of alteration CIA Nesbitt and Young 1982 and chemical index of weathering CIW Harnois 1988 have been used widely to monitor the decomposition of feldspar which is the most abundant rock-forming mineral group in the upper continental crust These two indices are calculated in a similar way except that CIW excludes K2O in the equation due to its inconsistent geochemical behaviors during weathering being leached or accumulated Unweathered rocks typically have CIA values around 50 or less if they contain abundant amphibole CIA 10 30 or clinopyroxene CIA 0 20 On the contrary CIA values for secondary minerals are much higher approximately 70 to 85 for smectites and illites and close to 100 for kaolin-group minerals Nesbitt and Young 1982 The marine sediments investigated here have CIA values ranging between 40 and 64 indicating that their sources were subjected to low to moderate degrees of chemical weathering compared to shales average CIA 70 75 Taylor and McLennan 1985 Both the calculated CIA and CIW values form a broad positive correlation with 26Mg in detrital sediments consistent with the general trend of preferential retention of heavy Mg isotopes during chemical weathering Fig 4-9A and B Scatter in the data may be caused by compositional differences in sediment protoliths Mg isotopic heterogeneity in clay minerals and different degrees of sorting underwent by the sediments see next section The rough negative correlation between 7Li and 26Mg particularly in the carbonate-rich sediments from DSDP 144 South Antilles Fig 4-9C lends further support to the role of chemical weathering which leads to preferential leaching of heavy Li isotopes thus Li and Mg isotopes fractionate in the opposite direction see Tomascak et al 2016 Penniston-Dorland et al 2017 for recent reviews In addition to 7Li Li concentration in sediments is also an effective weathering index due to its strong 125 Figure 4-9 26Mg variation in detrital sediments during progressive chemical weathering Chemical Index of Alteration CIA values in A were calculated using the formula CIA 100 molar Al2O3 Al2O3 CaO Na2O K2O after Nesbitt and Young 1982 Chemical Index of Weathering CIW values in B were calculated using a similar formula without K2O Harnois 1988 CaO in both CIA and CIW takes into account only CaO content in the silicate fraction of the sediments The open diamond symbols represent literature data of various upper crust rocks and sediments see Fig 4-3 caption for source information Li isotopic data in C are from Bouman et al 2004 Chan et al 2006 and Tang et al 2014 When no direct measurements are available 7Li values were taken as those of sediments with similar lithology from the same or nearby sites 126 affinity to clay minerals Qiu et al 2009 The trend of increasing 26Mg value with increasing Li content in Alaskan turbidites thus reflects a weathering effect Fig 4-9D Intensity of chemical weathering can also be evaluated by using element pairs with different lability to weathering such as MgO Na2O and Al2O3 K2O ratios in which cases Mg and Al are less mobile compared to Na and K respectively e g Schneider et al 1997 These weathering-induced 26Mg variation patterns have been observed in detrital sediments from ODP 701 South Sandwich and DSDP 543 North Antilles Fig 4-9E and F 5 1 2 3 Sediment transport: grain-size sorting Grain-size sorting Grain-size sorting along with sediment transport leads to systematic fractionation of sediment mineralogy and chemical composition between finegrained and associated coarse-grained fractions Magnesium isotopes are likely to be fractionated during this process as a result of changes in the relative proportions of mineral phases carrying distinct Mg isotopic signatures For example some loess sediments have their 26Mg values vary inversely with their grain sizes Huang et al 2013 This trend is consistent with the observation that sedimentary rocks dominated by fine-grained clay minerals such as shale and pelite comprise the high- 26Mg endmember on continents e g Li et al 2010 Ma et al 2015 as opposed to the coarser loess sediments that can host appreciable carbonates and are therefore depleted in heavy Mg isotopes Huang et al 2013 Wimpenny et al 2014b The TiO2 Zr ratio is a sensitive index for the degree of sediment sorting since the coarse-grained fraction of the sediments tends to concentrate heavy minerals e g zircon hence Zr together with quartz whereas TiO2 tends to follow Al2O3 127 and therefore is preferentially retained in the fine-grained weathering products Garcia et al 1994 TiO2 Zr ratios correlate positively with sediment 26Mg values in detrital sediments particularly in turbidites Fig 4-10A and B consistent with the trend of concentration of fine-grained isotopically heavy clay minerals with increasing degree of sediment sorting The effect of zircon sorting has also been well documented to cause the negative deviation of Hf in zircon-bearing detrital sediments from the Terrestrial Array Patchett et al 1984 Carpentier et al 2009 Vervoort et al 2011 Along with zircon detrital carbonate and other heavy minerals such as garnet are preferentially concentrated in the coarser sandier fractions This may explain the positive correlation between 26Mg and Hf i values in the Cascadia turbidites with the sand-rich fractions containing more abundant low- 26Mg carbonate garnet and low- Hf i old zircon Fig 4-10C 5 1 3 Non-detrital sediments 5 1 3 1 Hydrogenetic clays and Mn nodule Hydrogenetic sediments are enriched in authigenic components formed by direct precipitation from seawater or submarine alteration of pre-existing materials The hydrogenetic sediments analyzed in this study commonly display higher MgO Al2O3 ratios than detrital sediments Fig 4-2C suggesting in-situ formation of magnesian clays with uptake of Mg from seawater They have a relatively homogenous and high 26Mg value of 0 02 0 14 n 14 This is consistent with results from modeling of sediment pore water which suggest that the formation of authigenic clays preferentially remove heavy Mg from the fluid phase into their structure with a fractionation factor of 0 to 1 25 Higgins and Schrag 2010 Likewise brucite synthesis experiments found that the 128 Figure 4-10 26Mg variation related to grain-size sorting during sediment transport A TiO2 Zr ratio is a sensitive index for sediment sorting and measures the proportion of finegrained clays relative to coarse-grained quartz and zircon B The effect of sorting is particularly prominent in turbidites C As a result of sorting low- Hf i zircon and low 26Mg carbonate garnet were concentrated in the coarser sand fractions leading to the positive correlation between 26Mg and Hf i in Cascadia turbidites Hafnium isotopic data for these sediments are from Vervoort et al 2011 129 precipitated brucite is isotopically heavier than the solution Wimpenny et al 2014a The hydrogenetic sediments from Tonga DSDP sites 595 596 also have elevated Li concentrations and Li isotopic compositions relative to the upper continental crust which were attributed to adsorption of isotopically heavy seawater-derived Li Chan et al 2006 Therefore Mg and Li isotopes fractionate by different mechanisms during the formation of hydrogenetic sediments with heavy Mg isotopes being incorporated into clay structure while heavy Li isotopes being adsorbed to clays and hydroxyoxides resulting in a positive correlation between 26Mg and 7Li in Tonga sediments Fig 4-11 as opposed to the general negative correlation in detrital sediments produced by chemical weathering Fig 4-9C In addition the porcellanite and chert samples from Tonga have 26Mg values similar to the hydrogenetic clays from the same site but are isotopically heavier than siliceous oozes from other drill sites Their elevated 26Mg values may as well reflect formation of isotopically heavy authigenic clays during diagenetic recrystallization of amorphous opal Hydrogenetic sediments also differ significantly from detrital sediments in that they have higher Fe and Mn concentrations due to the abundance of ferro-manganese oxides Fig 4-2D The single Mn nodule sample analyzed here belongs to the hydrogenetic vernadite class due to its high Co concentration 3120 mg kg Chan et al 2006 in contrast with the diagenetic Mn nodules that are dominated by the todorokite class with no or low Co content Halbach et al 1981 Uspenskaya et al 1987 The 26Mg value of this sample 0 17 falls at the high- 26Mg end of the wide range reported for global Mn nodules 2 94 to 0 69 indicating insignificant early diagenesis which drives 26Mg in Mn 130 Figure 4-11 26Mg variation in Tonga hydrogenetic sediments reflects the preferential uptake of heavy Mg isotopes during authigenic mineral formation 7Li data are from Chan et al 2006 131 nodules to lighter values Rose-Koga and Albar de 2010 5 1 3 2 Hydrothermal clays Hydrothermal clays typically occur in the basal layer of sediment columns immediately above the basement rocks They consist primarily of authigenic Fe-Mn precipitates from spreading ridge hydrothermal fluids admixed with rock detritus Fig 412A They display large variation in 26Mg value which shows a negative correlation with FeO Al2O3 ratio Fig 4-12B In detail the two brown clay samples from DSDP site 178 Alaska have a chemical composition similar to that of the average terrigenous matter TM Their Mg budgets are dominated by montmorillonite with minor mica and chlorite Zemmels and Cook 1973 and they have high 26Mg values 0 06 and 0 46 that are commonly seen in clay minerals By contrast the four hydrothermal black brown clay samples from DSDP sites 294 295 Ryukyu have much higher Fe and Mn concentrations and lower Al concentrations that plotted close to the composition of the East Pacific Rise metalliferous deposit EPR These hydrothermal clays contain abundant hematite and lesser amount of montmorillonite Cook et al 1975 The relatively low 26Mg values of 0 48 to 0 57 in these hydrothermal clays might reflect Mg exchange between clays and 26Mg-depleted seawater or formation of secondary Mg-bearing minerals that have a preference for light Mg isotopes at hydrothermal temperatures Pogge von Strandmann et al 2008 The single pebbly black hydrothermal clay has a much higher Al content than the other three hydrothermal clays from Ryukyu 16 wt% vs 3 6 wt% Its higher 26Mg value of 0 17 may reflect dominance of terrigenous detritus Therefore a mixed contribution of Mg from the continental detritus and seawater hydrothermal fluids may 132 Figure 4-12 26Mg variation related to hydrothermal sediments TM and EPR represent the average compositions of terrigenous materials and of East Pacific Rise deposits respectively from B strom et al 1976 133 produce the Mg isotopic variation seen in these hydrothermal sediments 5 2 Sedimentary Mg input to global subduction zones To evaluate the variation in Mg flux and isotopic signature of sediment sections delivered to modern subduction zones on a global basis we calculated the mass-weighted average 26Mg value and total Mg flux for all sediment columns that have been included in the calculation of GLOSS-II which account for 86% of the global flux Plank 2014 The comprehensive datasets of subducting marine sediments over worldwide trench systems provided in Plank 2014 serve as a basis for estimating the global subducting sedimentary Mg input Similar to the elemental flux calculation in Plank and Langmuir 1998 the sedimentary Mg flux per unit arc length F is calculated using the following equation: F C 1- w t R where C w t and R represent depth-integrated concentration water content thickness density and convergence rate respectively The average 26Mg value for each trench considered is calculated based on averaged data for individual lithologic units see Appendix B for calculation details Calculation results are summarized in Table 4-2 and Fig 4-13 For more than half of the trenches no Mg isotopic analysis is available and their 26Mg values are estimated from the correlation with CaO TiO2 ratios Fig 4-B1 or taken as the averages of compositionally similar sediment endmembers Despite the potential uncertainties in these estimates they provide a preliminary quantification of the 26Mg values for a variety of sedimentary columns that are being delivered to the trenches The calculated sedimentary Mg fluxes differ by two orders of magnitude among 134 Table 4-2 Calculated Mg fluxes and bulk 26Mg values of global subducting sedimentary columns Trench N Zealand Kermadec Tonga Vanuatu E Sunda Java Sumatra Andaman Makran Philippines Ryuku Nankai Marianas Izu-Bonin Japan Kurile Kamchatka Aleutians Alaska Cascadia Mexico CentAm Colombia Peru Chile-35 Chile-40 Chile-45 Sandwich N Antilles S Antilles GLOSS-II Subd rate cm yr 2 7 6 5 16 8 1 7 6 5 4 7 3 3 5 6 7 7 1 3 4 4 1 8 2 7 8 7 5 6 5 3 3 5 5 2 7 5 9 7 6 7 2 7 5 7 5 6 2 1 8 1 8 Thickness m 1600 200 70 650 500 300 1400 3500 4200 120 160 350 476 408 285 345 435 350 780 900 170 425 290 125 1000 1200 1500 200 235 1750 Density g cc 2 00 1 40 1 31 1 60 1 71 1 65 1 95 2 05 2 05 1 31 1 40 2 20 1 86 1 74 1 58 1 64 1 70 1 64 1 80 1 88 1 37 1 62 1 44 1 47 1 88 1 88 1 88 1 49 1 66 1 88 Water % content 30 00 58 00 62 57 38 35 34 89 40 30 24 49 20 17 20 17 62 57 58 00 20 00 23 00 31 57 42 68 38 80 35 15 41 54 32 40 30 00 59 01 48 69 53 04 50 00 30 00 30 00 30 00 50 00 37 30 30 00 Trench km Length 600 1400 1350 1800 1000 2010 1000 1500 950 1550 1350 800 1400 1050 800 1650 550 1900 800 1300 1450 1450 1050 1500 495 550 275 800 400 400 Sediment % mass 2 7% 0 8% 0 5% 6 9% 2 9% 2 9% 7 1% 19 0% 16 8% 0 5% 0 7% 1 1% 2 8% 1 5% 1 2% 3 3% 1 5% 2 8% 3 0% 4 0% 0 5% 2 7% 0 9% 0 8% 3 5% 4 8% 3 0% 0 5% 0 1% 1 2% 135 Average Mg wt% 3 16 1 14 1 62 1 67 1 48 1 64 1 50 1 46 2 11 1 86 1 83 1 34 2 16 0 98 1 13 0 93 0 80 1 63 1 96 1 53 1 73 0 92 0 95 1 11 1 83 1 84 1 88 1 35 1 42 1 15 1 658 Mg mass % 5 1% 0 5% 0 5% 6 9% 2 6% 2 8% 6 4% 16 7% 21 4% 0 5% 0 7% 0 9% 3 7% 0 9% 0 8% 1 8% 0 7% 2 8% 3 5% 3 7% 0 6% 1 5% 0 5% 0 5% 3 8% 5 3% 3 4% 0 4% 0 1% 0 8% Mg Flux per g yr cm arc length 1910 87 89 867 577 316 1450 2504 5057 74 122 248 590 196 238 251 286 329 985 634 86 227 109 77 1738 2176 2780 125 63 477 Total Mg kg yr Flux 1 15E 08 1 22E 07 1 21E 07 1 56E 08 5 77E 07 6 35E 07 1 45E 08 3 76E 08 4 80E 08 1 14E 07 1 65E 07 1 98E 07 8 26E 07 2 05E 07 1 90E 07 4 15E 07 1 57E 07 6 24E 07 7 88E 07 8 24E 07 1 24E 07 3 30E 07 1 15E 07 1 16E 07 8 60E 07 1 20E 08 7 64E 07 1 00E 07 2 51E 06 1 91E 07 2 25E 09 Average 26 Mg 0 28 0 24 0 02 0 29 0 09 0 08 0 32 0 36 0 51 0 01 0 14 0 44 0 21 0 18 0 16 0 13 0 16 0 26 0 15 0 42 0 24 0 61 0 43 0 50 0 32 0 34 0 38 0 00 0 06 0 07 -0 336 Figure 4-13 Variations of Mg mass fluxes and estimated average 26Mg values for sedimentary sections subducting at major subduction zones Sediment sections displayed in the lower panel of the figure are after Plank 2014 The blue horizontal line represents the average 26Mg value for GLOSS-II this study and the red horizontal line represents the mantle average from Teng et al 2010a Note that 26Mg axis is in reverse order and higher blue bars are more negative and therefore isotopically lighter all values are negative except for one 136 the arcs investigated here Fig 4-13 ranging from 2 5 107 kg yr at the North Antilles margin to 4 8 109 kg yr at the Makran trench The variation is mainly a function of sediment type and section thickness Generally Mg fluxes are the highest in trenches fed by thick turbidites For example turbidites subducting at Andaman and Makran account for almost 40% of the Mg budget of GLOSS-II In addition trenches that contain a thick volcaniclastic section such as New Zealand and Vanuatu trenches could contribute significant amount of Mg 12% due to high Mg concentration in volcaniclastic sediments On the other hand biogenic-dominated trenches such as South Sandwich Peru and Colombia are usually at the low-Mg flux end 0 5% due to low Mg concentration in carbonate and opaline silica and their commonly thin sediment thickness Convergence rates are generally similar among different trench systems hence they play a secondary role in determining the sediment flux except for a few cases where they are considerably different For example although the sediment section at Tonga is only 70 m thick its fast subduction rate of 16 cm year results in a similar Mg flux 1 2 108 kg yr to that of the South Antilles 1 9 108 kg yr where the sediment column is 1750 km thick but subducts at a much slower rate of 1 8 cm year Table 4-2 Similar to the large range in regional Mg fluxes estimated average sedimentary 26Mg values for different trenches also display a significant arc-to-arc variation from 0 61 to 0 02 reflecting different dominant lithology among global sediments Fig 4-13 Sediment sections rich in clay-size detritus derived from mature and highly weathered continental sources acquire an accompanying heavy Mg isotopic signature e g Antilles and South Sandwich in comparison with less-differentiated volcaniclasticdominated or glacially transported sediments that preserve mantle-like 26Mg signatures 137 e g Peru and Alaska Grain-size sorting also fractionates Mg isotopes within turbidites Those sequences enriched in coarse-size detritus could possibly host higher amounts of low- 26Mg carbonate and or garnet leading to a lighter Mg isotopic signature Enrichment in hydrogenetic phases leads to heavier Mg isotopic compositions in some sediment sections e g Philippines and Tonga as a result of preferential uptake of heavy Mg isotopes during authigenic clay formation On the contrary a few sites near the equatorial regions such as Central America Peru and Colombia contain a high fraction of biogenic carbonate which drives their bulk 26Mg to variably lighter values The global variation in sedimentary 26Mg input to active subduction zones falls within the range found in continental sediments while the flux-weighted 26Mg average for GLOSS-II 0 336 is slightly lower than both the average upper continental crust 0 22 Li et al 2010 and normal upper mantle 0 25 0 07 Teng et al 2010a The relatively lower 26Mg of GLOSS-II is consistent with the fact that GLOSS-II as well as its Mg content is heavily weighted by a few isotopically light turbidites sequences and volcaniclastic sections Fig 4-13 Table 4-2 The thick turbidite sequences subducting at Sumatra Andaman Makran Cascadia and Chile margins together account for 60% of the total Mg flux of the GLOSS-II and they have variably low 26Mg averages 0 51 to 0 32 Table 4-2 Among them the lowest average 26Mg value occur at the thickest Makran section 4200 m Plank 2014 that contributes to 20% of the Mg flux of the GLOSS-II Likewise the thick MgO-rich volcaniclastics sections at New Zealand-Hikurangi trough and Vanuatu trench result in high average MgO values in the incoming sediment columns e g up to 5 24 wt% in the former Plank 2014 Meanwhile the thick carbonate units above the volcaniclastics layers drive the bulk 26Mg values 138 0 30 to slightly lower than the mantle average Furthermore local enrichment of carbonate in some sediment sections that close to the equator such as sediments from Central America and Colombia-Ecuador trenches may also contribute to the overall lower 26Mg average for the GLOSSII In addition marine sediments generally underwent lower degree of chemical weathering and hence are less enriched in heavy Mg isotopes compared with widespread highly weathered shales on continents 5 3 Implications for arc magmatism and mantle Mg isotopic heterogeneity Subducted sediments were considered to be one of the key components in producing the enriched mantle endmembers e g Hofmann 2014 Non-mantle 26Mg values found in some mantle derivatives have also been tentatively ascribed to contributions from subducted slabs Yang et al 2012 Wang et al 2012 2015a 2016 Xiao et al 2013 Huang et al 2015 Liu et al 2015 Hu et al 2016a Tian et al 2016 Teng et al 2016 Li et al 2017 Our study provides the direct evidence that Mg isotopic compositions of subducting marine sediments are indeed highly heterogeneous Fig 4-13 More importantly trenches with high sedimentary Mg fluxes have average 26Mg values different from the mantle suggesting they may exert a significant impact on mantle Mg isotopic heterogeneity Furthermore recent studies have revealed that altered oceanic crust and abyssal peridotites also have heterogeneous Mg isotopic compositions Huang 2013 Liu et al 2017 Since metamorphic dehydration does not fractionate Mg isotopes Li et al 2011 2014 Teng et al 2013 Wang et al 2014a 2015b 2015c the subducting slab as a whole can carry significant amount of distinct Mg isotopes into the mantle Different slab components however are expected to release Mg at different depths 139 and by different mechanisms hence they may contribute differently to arc magmatism and mantle Mg isotopic heterogeneity Marine sediments may undergo partial melting during subduction because the primary hydrous phase in sediments breaks down above the fluidsaturated solidus Johnson and Plank 1999 However dehydration is more likely to occur than melting for basaltic crust at the thermal structure of most modern subduction zones and these fluids further promote sediment melting In addition the slab-mantle interface is a sedimentary-mafic ultramafic mixing zone Bebout 2007 6 Conclusions Marine sediments from ten drill sites in front of the world s major subduction zones display large variation in their Mg isotopic compositions which mainly reflects differences in sediment mineralogy provenance composition and sedimentary processes The following conclusions can be drawn: 1 The bulk marine sediments have highly heterogeneous 26Mg values 1 34 to 0 46 the variation of which largely mirrors that found in rocks and sediments on continents 2 Sediment mineralogy plays an important role on the Mg isotopic distribution among different sediment types Carbonate-rich sediments favor light Mg isotopes while clayrich sediments are generally more enriched in heavy Mg isotopes 3 Chemical weathering and sediment textural mineralogical sorting can potentially alter the relative proportions of fine-grained clay to coarse-grained sand and therefore drive the 26Mg of clay-rich marine sediments to heavier values while the presence of detrital carbonate and or garnet in the coarser fraction drives turbidites to variably 140 lower 26Mg values 4 Hydrogenetic clays have a relatively uniform and on average heavy Mg isotopic composition consistent with preferential uptake of heavy isotopes by authigenic clays In comparison hydrothermal clays have highly variable 26Mg values which may result from variable mixing between detrital phases and hydrothermal fluids or seawater 5 The Mg mass flux and isotopic composition of subducting sediment sections vary substantially from arc to arc Recycling of these isotopically heterogeneous marine sediments together with altered oceanic crust and abyssal peridotites into the mantle may lead to local mantle domains with distinct Mg isotopic compositions and generate heterogeneous volcanic rocks 7 Appendix A Supplementary data Supplementary data to this article can be found online at http: dx doi org 10 1016 j chemgeo 2017 06 010 Acknowledgements We would like to thank the Deep Sea Drilling Program and Ocean Drilling Program for providing the core samples for this study Brian Atwater Charlotte Schreiber Ronald Sletten Scott Kuehner and Melissa Harrington are thanked for stimulating discussions Constructive comments of Ming Tang and an anonymous reviewer and efficient editorial handling by Michael E B ttcher are gratefully acknowledged This work was financially supported by the National Science Foundation EAR-1340160 to F Z Teng National Natural Science Foundation of China 41602343 to K -J Huang 2015 141 graduate student research grant from the Geological Society of America to Y Hu and George Edward Goodspeed Geology Scholarship from the Earth and Space Sciences Department of the University of Washington to Y Hu 142 Chapter 5 Magnesium cycling at subduction zones constrained by Mg isotopic composition of sub-arc mantle beneath southern Kamchatka This chapter is in preparation for publication as: Hu Y Teng F -Z and Dmitri A Ionov in prep Magnesium cycling at subduction zones constrained by Mg isotopic composition of sub-arc mantle beneath southern Kamchatka Abstract Subduction of the oceanic slab may add to mantle components with stable isotopic signatures fractionated at low-temperatures The highly variable Mg isotopic compositions of the subducted oceanic slab input and arc lava output imply that components with atypical Mg isotopic compositions are present in the mantle wedge Magnesium isotopic data on samples from the sub-arc mantle are still limited however To characterize the Mg isotopic composition of a typical supra-subduction zone mantle domain 23 large and fresh spinel harzburgite xenoliths from the Avacha volcano in Kamchatka Russia were analyzed These harzburgites were formed by 30% melt extraction at fluid fluxing conditions followed by subduction-derived small-scale fluid metasomatism They are similar in modal and chemical composition to forearc harzburgites and harzburgite xenoliths from the western Pacific The 26Mg values of the Avacha xenoliths display limited variation from 0 30 to 0 21 overlapping the mantle average at 0 25 0 07 The overall mantlelike and uniform 26Mg range of the Avacha sub-arc peridotites may be due to their high MgO contents 44 wt% and low fluid rock mass ratios during flux melting and metasomatism Clinopyroxene a rare late-stage mineral 0 24 0 10 2SD n 5 143 appears to be in Mg isotopic equilibrium with olivine 0 26 0 04 2SD n 17 and orthopyroxene 0 23 0 04 2SD n 17 Collectively this study finds that melting residues in the mantle wedge in the western Pacific where dehydration of slab crust serves as the major water supplier for melting and metasomatism likely have a mantle-like 26Mg Fluids released from oceanic crust at pressures of 2 to 3 GPa are well buffered during their migration in the mantle wedge due to their low Mg concentrations Large-scale dehydration of isotopically distinct materials at higher pressures may be required to produce isotopically diverse arc lavas 1 Introduction Slab-mantle interaction at subduction zones is central to global element cycling arc magmatism and seismic activities which ultimately contribute to the formation of the continental crust and physicochemical evolution of the mantle The Mg isotope system has recently been suggested as a potential tracer of crustal recycling due to its solubility in fluids the lack of measurable Mg isotope fractionation during metamorphic dehydration e g Teng et al 2013 Li et al 2014 Wang et al 2014 and the large isotopic contrast between typical mantle rocks and surface materials Teng 2017 and references therein Magnesium is overwhelmingly enriched in the mantle relative to the crust and has shown a uniform isotope ratio 26Mg in the majority of peridotites and oceanic basalts Teng et al 2010a By contrast low-temperature continental chemical weathering and seafloor alteration fractionate Mg isotopes significantly leading to a highly heterogeneous subducting slab that contains marine sediments altered oceanic crust and altered abyssal peridotite Wimpenny et al 2012 Beinlich et al 2014 Teng et al 2016 Hu et al 2017 144 Liu et al 2017 Huang et al 2018 Therefore contributions of recycled crustal materials to sub-arc mantle could potentially be seen in 26Mg values of arc lavas that deviate from the normal mantle range Studies of arc lavas from Lesser Antilles and Philippines have found 26Mg values higher than the normal mantle which are attributed to hydration and modification of the mantle wedge by fluids released from subducting slabs in particularly abyssal peridotites Teng et al 2016 Li et al 2017 The highly variable Mg isotopic compositions of the subducting slab input and arc lava output suggest that slab-derived Mg isotopic signatures could be detected in the mantle wedge which serves as the primary sink for various slab-released fluids melts and is the main source of arc magmas Compared to arc lavas that commonly undergo magmatic differentiation and crustal assimilation during their ascent through the overlying crust arc peridotite xenoliths are fragments of mantle lithosphere above the subducted slab accidentally entrained in arc magmas and brought to the surface by explosive eruptions They are thus direct samples of materials processed in the mantle wedge that may provide insights into the melting and slab-mantle interaction processes that are independent of but complementary to observations from arc lavas However Mg isotopic data on arc peridotite xenoliths remain sparse due to the scarcity of representative samples making the sub-arc mantle one of the most important but least documented mantle reservoirs The Kamchatka peninsula in far eastern Russia Fig 5-1 hosts one of the few arcs where mantle xenoliths are available The most common xenoliths near the arc front are refractory spinel harzburgites interpreted as residues of flux melting in sub-arc mantle variably affected by slab-related fluids and or melts during or after their incorporation in the mantle lithosphere Halama et al 2009 Ionov 2010 2011 B nard and Ionov 2012 145 Hopp and Ionov 2011 Ionov et al 2011 2013 B nard et al 2013 2016 2017 Much work on Kamchatka xenoliths including those inland from the arc examined microstructural deformation volatile and fluid-mobile elements and Sr-Os-Pb-4He 3He isotopes along with the formation of late-stage minerals glass inclusions and a variety of pyroxene-rich veins Soustelle et al 2010 Kepezhinskas et al 1995 1996 1997 2002 Kepezhinskas and Defant 1996 Arai et al 2003 Widom et al 2003 Saha et al 2005 Ishimaru et al 2006 2007 2009 Bryant et al 2007 Arai and Ishimaru 2008 Ishimaru and Arai 2008a b c 2011 Partial melting of sub-arc mantle beneath Kamchatka generated a wide variety of lavas that display elemental and isotopic enrichments H-O-Sr-Nd-PbHf and U series Hochstaedter et al 1994 Kersting and Arculus 1995 Kepezhinskas et al 1996 1997 Turner et al 1998 Pineau et al 1999 Dorendorf et al 2000 Churikova et al 2001 2007 Ishikawa et al 2001 Yogodzinski et al 2001 Dosseto et al 2003 Duggen et al 2007 Portnyagin et al 2007 Auer et al 2009 Dosseto and Turner 2014 Kayzar et al 2014 To further constrain the Mg isotope systematics in subduction zones we analyze a suite of 23 spinel harzburgite xenoliths from the Avacha volcano in southern Kamchatka arc These samples are uncommonly large fresh and texturally equilibrated Importantly they do not show evidence for late-stage interaction with host magmas or extensive recrystallization reported for other xenolith suites from Avacha e g Arai et al 2003 Ishimaru et al 2007 Ishimaru and Arai 2008 Therefore their composition reflects the nature of original sub-arc lithospheric mantle The Mg isotopic composition of sub-arc mantle peridotites are examined in conjunction with published arc lava 26Mg data to explore the relationship between slab dehydration and 26Mg variation in the lavas Our 146 results demonstrate that the mantle wedge beneath the southern Kamchatka arc which represents a typical subduction zone in the western Pacific has largely a uniform and mantle-like Mg isotopic composition despite interaction with slab-derived fluids Subduction of oceanic crust with its overlying sediments has long been considered an essential mechanism of elemental cycling from Earth s surface to its interior thereby producing chemical and isotopic heterogeneity in mantle rocks and mantle-derived magmas e g Armstrong 1968 Hofmann 2014 and references therein Recently this process has also been called upon to explain local Mg isotopic variations in the mantle e g Yang et al 2012 Xiao et al 2013 Hu et al 2016a Li et al 2017 The majority of mantle peridotites and oceanic basalts have similar Mg isotopic compositions 26Mg 0 25 0 07 2SD Teng et al 2010a suggesting homogeneous distribution of Mg isotopes in the mantle Nevertheless some mantle xenoliths e g wehrlite pyroxenite and eclogite and mantle-derived lava possess 26Mg values that lie outside the well-defined mantle range and could reflect a contribution from recycled crustal Mg Wang et al 2012 2015a 2016 Yang et al 2012 Xiao et al 2013 Huang et al 2015 Liu et al 2015 Hu et al 2016a Tian et al 2016 Teng et al 2016 Li et al 2017 2 Geologic setting The Kamchatka arc lies in the northern part of the c a 2000 km-long KamchatkaKurile convergent margin formed by subduction of the Mesozoic Pacific plate beneath the Eurasian plate in northwestern Pacific at a rate of 8 cm yr Gorbatov et al 1997 Fig 51 Orthogonal intersection with the Aleutian transform fault-trench system where the Pacific plate subducts beneath the North American plate sets the northern limit of the 147 Kamchatka arc and divides it into a volcanically inactive north segment and active central and south segments Hochstaedter et al 1994 Kepezhinskas et al 1995 At the north segment westward subduction of the young 25 Ma and hot oceanic lithosphere formed by short-lived spreading in the Komandorsky basin Bogdanov 1988 had produced arc volcanism in the Sredinny Range SR until 2 Ma Firsov 1987 Closure of the Vetlovka basin and subsequent accretion of the Kronotsky terrane led to formation of the Eastern Volcanic Front EVF at the southern segment and the 200 km-wide Central Kamchatka depression CKD between the SR and EVF in the central segment The Avacha Avachinsky stratovolcano 53 15 N 158 52 E is located in the middle part of the EVF Fig 5-1 It is situated 120 km above the slab surface where the 70 km-thick Pacific slab is subducting at 55 Gorbatov et al 1997 A layer of low Pvelocity has been observed at 70 to 130 km and was interpreted to represent the asthenospheric mantle wedge Nizkous et al 2006 The Avacha volcano formed in late Pleistocene producing volcanic ash-fall or trephra and pyroclastic flow deposits of calcalkaline affinity 3 Samples and evidence for slab - mantle interaction A unique feature of partial melting in arc settings is the high fluid flux supplied by down-going slab which facilitates melt extraction leaving behind highly refractory peridotite residues Bizimis et al 2000 The xenoliths analyzed in this study are mediumto coarse-grained spinel harzburgites Ionov 2010 B nard et al 2017 They are primarily composed of olivine 67-82 5% and orthopyroxene Opx 18-30% as well as accessory spinel late-stage interstitial clinopyroxene Cpx 1-3% formed mainly by exsolution from 148 Figure 5-1 Geological map of the Kamchatka peninsula Russia The boxed area of Kamchatka is enlarged on the left to show the locations of major Quaternary volcanoes Modified from Kersting and Arculus 1995 and Ionov 2010 EVF Eastern Volcanic Front SR Sredinny Range CKD Central Kamchatka Depression 149 high-temperature Opx on cooling and amphibole 1% These highly refractory 44 wt % MgO peridotites containing no primary Cpx were inferred to have formed by high degree 28-35% melt extraction at 45 to 65 km depth Ionov 2010 Compared to the local Moho depth at 37 km Levin et al 2002 these harzburgites represent the shallow lithospheric mantle beneath Avacha affected by fluid metasomatism Ionov 2010 veining Halama 2009 B nard et al 2013 2016 and high-temperature creep-related deformation Soustelle et al 2010 A distinctive characteristic of the Avacha peridotite suite is the paradoxical enrichment in silica hosted by opx despite their highly refractory nature Fig 5-2a which requires an enrichment process in addition to partial melting Herzberg 2004 It was initially suggested based on studies of xenoliths containing fine-grained Opx that the Avacha peridotites were enriched in Opx after the melting shortly before their capture by magma due to reaction of primary olivine with subduction-related fluids e g Ishimaru et al 2007 By contrast Ionov 2010 reported silica enrichments in peridotites from this study that contain no fine-grained Opx and argued that the enrichments took place during the melting because of the input of slab-derived components and effects of fluids on modal proportions in the residues Recently modeling of mantle melting further suggested that the Avacha peridotites could be residues after extraction of picrites and high Ca-boninites B nard and Ionov 2012 B nard et al 2016 from a hybrid source that formed by interaction of refractory mantle with slab-derived components before flux melting B nard et al 2017 The subduction of old and hence cold Pacific plate 100 Ma beneath southern Kamchatka implies that slab dehydration beneath the EVF produces hydrous fluids rather 150 than melts which is supported by thermal modeling for the Kamchatka arc Manea and Manea 2007 This is consistent with the fact that nearly all Avacha harzburgites contain accessory amphibole and show U-shaped trace element patterns with widespread but lowdegree enrichments in fluid-mobile elements Rb Ba U Pb Sr and B relative to fluidimmobile melt-mobile elements e g MREE Nb Y Ionov 2010 similar to serpentinized forearc peridotites Savov et al 2005 Fig 5-2 Locally at high fluid rock ratios the reaction of such fluids with refractory mantle produced small-volume boninitic melts that formed rare thin veins B nard and Ionov 2013 as well as melt inclusions in spinel B nard et al 2016 in some xenoliths To sum up the Avacha xenoliths show ubiquitous but weak effects of percolation of slab-derived fluids on bulk-rock compositions in addition some xenoliths contain veins formed by boninitic melts The Sr-Nd-Pb isotopic signatures of the southern Kamchatkan basalts and xenoliths are similar to those in Pacific MORB suggesting the dominant role of fluids from the altered oceanic crust with limited sediment contribution 1% in the volatile input to the mantle wedge Kersting and Arculus 1995 Kepezhinskas et al 1997 Churikova et al 2001 Saha et al 2005 Duggen et al 2007 This is consistent with the radiogenic Os isotopes of peridotite xenoliths Widom et al 2003 Saha et al 2005 and high- 18O signatures from lava-hosted phenocrysts Dorendorf et al 2000 In addition studies of volatiles in melt inclusions from lava-hosted olivine and B concentration and isotopic signatures of the lavas suggest a possible role of serpentinite dehydration Ishikawa et al 2001 Churikova et al 2007 Portnyagin et al 2007 Serpentinite dehydration is independently supported also by the observation of a double seismic zone in Kamchatka Gobatov et al 1997 which is commonly attributed to breakdown of antigorite Peacock 151 Figure 5-2 Geochemical evidence for subduction-related metasomatism recorded in the Avacha peridotite xenoliths Major and trace element concentrations are from Ionov 2010 Ce Pb and U Th abundances are normalized to primitive mantle of McDonough and Sun 1995 The compositions of serpentinized forearc peridotites Savov et al 2005 and N-MORB Sun and McDonough 1989 are plotted for comparison The field of slab melts trapped in the Kamchatka mantle are based on analyses of glass veins and melt inclusions in Kamchatka mantle xenoliths Kepezhinskas et al 1996 Kepezhinskas and Defant 1996 152 et al 2001 4 Analytical methods Magnesium isotopic analysis was performed at the Isotope Laboratory of the University of Washington Seattle Protocols for sample digestion chemical separation and instrumental analysis are outlined in Teng et al 2007 2010 and are summarized here The whole-rock powders analyzed here are the same as those used in Ionov 2010 and B nard et al 2017 Olivine and pyroxene fragments were hand-picked under a binocular Whole-rock powders and pre-cleaned mineral fragments were dissolved in a mixture of Optima HF-HNO3-HCl with no insoluble residue present in the solution After complete dissolution sample solutions were evaporated to dryness and then fluxed with 1 N HNO3 for column chromatography Magnesium was quantitatively extracted and purified by two passes of the pre-cleaned Bio-Rad cation exchange resin AG50W-X8 resin 200-400 mesh in 1 N HNO3 Magnesium isotopic ratios were then measured on a Nu Plasma II multicollector inductively coupled plasma mass spectrometer MC-ICPMS using samplestandard bracketing method Dried Mg cuts were diluted to 300 ppb Mg concentration in 3% Optima HNO3 shortly before analysis The sample solutions were introduced under wet plasma conditions and were analyzed in low-resolution mode Sample intensity was adjusted to within 10% of that of the bracketing standard Magnesium isotopic data are expressed as per mil deviation relative to the DSM3 standard: x Mg xMg 24Mg x sample 24 Mg Mg -1 1000 DSM3 where x 25 or 26 The associated error of each data is 2SD which represents two times 153 the standard deviation of multiple analyses of bracketing standards during an analytical session The accuracy of the method was checked by processing and measuring three international rock standards and two in-house standards San Carlos olivine and Hawaiian seawater along with the unknown samples Table 5-1 The two peridotite standards PCC-1 and DTS-1 yield average 26Mg of 0 25 0 03 n 4 and 0 31 0 05 n 2 and the basalt standard JB-1 has a similar average 26Mg of 0 28 0 03 n 3 The average 26Mg for 15 measurements on San Carlos olivine and 7 analyses on Hawaiian seawater are 0 25 0 03 and 0 85 0 03 respectively The results of all these standards agree well with recommended values listed in Table 5-1 Ling et al 2011 Teng et al 2015 Hu et al 2016a 5 Results Magnesium isotopic compositions for bulk peridotites are reported in Table 5-2 along with mineral modes MgO and Al2O3 contents and estimated equilibrium temperatures reported in Ionov 2010 and B nard et al 2017 The Mg isotopic compositions of mineral separates are presented in Table 5-3 Table 5-4 reports the degree of inter-mineral Mg isotopic fractionation calculated as 26MgX-Y 26MgX 26MgY where X and Y refer to two different mineral phases All measurements during the course of this study fall on a single mass-dependent fractionation line with a slope of 0 51 R2 0 99 not shown and therefore only 26Mg values are discussed hereafter 154 Table 5-1 Mg isotopic compositions of standards analyzed during the course of this study Standard 26Mg 2SD 25Mg 2SD Reference PCC-1 0 25 0 05 0 14 0 04 Harzburgite USGS 0 25 0 05 0 11 0 06 0 26 0 07 0 14 0 06 0 26 0 07 0 14 0 05 Average 0 25 0 03 0 14 0 02 Recommended 0 23 0 06 0 10 0 01 Teng 2017 DTS-1 Dunite USGS Average Recommended 0 31 0 31 0 31 0 30 0 07 0 06 0 05 0 01 0 17 0 15 0 16 0 13 0 06 0 05 0 04 0 01 Teng 2017 JB-1 Basalt GSJ 0 28 0 25 0 29 0 28 0 28 0 06 0 07 0 06 0 03 0 1 0 15 0 13 0 16 0 15 0 15 0 04 0 04 0 04 Teng 2017 0 02 0 04 0 26 0 23 0 26 0 24 0 26 0 26 0 23 0 22 0 24 0 25 0 26 0 25 0 26 0 22 0 25 0 25 0 24 0 05 0 05 0 05 0 07 0 07 0 06 0 08 0 06 0 07 0 06 0 07 0 07 0 07 0 06 0 06 0 03 0 03 0 13 0 12 0 15 0 11 0 13 0 13 0 12 0 12 0 13 0 13 0 13 0 14 0 15 0 12 0 13 0 13 0 12 0 05 0 06 0 04 0 05 0 06 0 04 0 05 0 05 0 03 0 04 0 04 0 04 0 05 0 04 0 04 0 02 0 02 Hu et al 2016a 0 86 0 83 0 83 0 86 0 87 0 85 0 85 0 85 0 84 0 07 0 05 0 07 0 05 0 06 0 07 0 07 0 03 0 06 0 45 0 43 0 42 0 42 0 43 0 45 0 44 0 43 0 43 0 06 0 04 0 04 0 05 0 05 0 05 0 06 0 03 0 04 Teng et al 2015 Average Recommended San Carlos olivine Average Recommended Hawaiian seawater Average Recommended 155 Table 5-2 Mg isotopic compositions of bulk Avacha harzburgite xenoliths Sample Spl harzburgite Av-1 re-dissolution average Av-2 Av-3 Av-4 duplicate re-dissolution average Av-5 Av-6 duplicate re-dissolution average Av-7 Av-8 re-dissolution average Av-9 duplicate average Av-10 duplicate average Av-11 duplicate average Av-12 Al2O3 wt % T Ca-Opx T Ol-Spl C C 0 91 0 50 903 950 45 37 45 74 43 73 0 91 0 91 0 91 0 55 0 41 0 75 906 946 920 969 950 792 44 56 45 97 0 91 0 91 0 63 0 48 900 993 988 995 46 02 44 50 0 91 0 91 0 42 0 78 916 958 907 976 46 19 0 91 0 50 923 964 45 00 0 91 0 62 949 989 46 21 0 91 0 44 930 914 46 05 0 91 0 62 917 978 26Mg 2SD 25Mg 2SD MgO wt % Mg 0 29 0 29 0 29 0 30 0 30 0 28 0 30 0 30 0 29 0 26 0 27 0 27 0 25 0 26 0 27 0 30 0 29 0 29 0 27 0 27 0 27 0 28 0 21 0 25 0 30 0 30 0 30 0 24 0 07 0 05 0 04 0 07 0 07 0 07 0 07 0 05 0 04 0 06 0 06 0 07 0 05 0 04 0 08 0 07 0 05 0 04 0 05 0 07 0 04 0 05 0 07 0 04 0 05 0 07 0 04 0 05 0 13 0 15 0 14 0 17 0 14 0 13 0 14 0 16 0 14 0 13 0 15 0 14 0 14 0 14 0 13 0 15 0 14 0 14 0 14 0 13 0 13 0 15 0 10 0 12 0 15 0 16 0 16 0 12 0 04 0 05 0 03 0 04 0 04 0 03 0 04 0 05 0 02 0 04 0 04 0 05 0 05 0 03 0 05 0 05 0 05 0 04 0 05 0 04 0 03 0 05 0 04 0 03 0 05 0 04 0 03 0 05 45 67 156 duplicate average Av-13 duplicate average Av-14 duplicate average Av-15 duplicate average Av-16 duplicate average Av-17 duplicate average Av-33 Av-40 Av-41 duplicate average Av-42 duplicate average Av-G1 duplicate average Av-G2 duplicate average 0 25 0 24 0 29 0 24 0 27 0 27 0 28 0 27 0 27 0 22 0 25 0 29 0 29 0 29 0 27 0 23 0 25 0 27 0 30 0 28 0 27 0 28 0 31 0 28 0 29 0 20 0 22 0 21 0 23 0 22 0 23 0 07 0 04 0 05 0 07 0 04 0 05 0 07 0 04 0 05 0 07 0 04 0 05 0 07 0 04 0 05 0 07 0 04 0 07 0 07 0 06 0 05 0 04 0 07 0 05 0 04 0 07 0 07 0 05 0 06 0 07 0 05 0 12 0 12 0 16 0 11 0 13 0 14 0 12 0 13 0 12 0 11 0 12 0 14 0 13 0 13 0 13 0 12 0 12 0 15 0 15 0 14 0 13 0 14 0 16 0 14 0 15 0 11 0 11 0 11 0 12 0 11 0 12 0 04 0 03 0 05 0 04 0 03 0 05 0 04 0 03 0 05 0 04 0 03 0 05 0 04 0 03 0 05 0 04 0 03 0 04 0 04 0 05 0 04 0 03 0 04 0 04 0 03 0 04 0 05 0 03 0 04 0 05 0 03 157 45 90 0 91 0 53 954 997 44 47 0 91 0 85 915 972 44 82 0 91 0 63 960 972 46 20 0 92 0 52 989 993 46 54 0 91 0 50 895 915 44 69 45 93 46 00 0 91 0 91 0 91 0 66 0 52 0 62 45 42 0 91 0 59 44 44 0 90 0 71 44 86 0 91 0 71 Overall both the bulk peridotites and their constituent minerals display narrow variations in 26Mg which are not correlated with their modal or major element compositions Fig 5-3 The 26Mg values of the 23 Avacha peridotites vary from 0 30 to 0 21 with an average of 0 27 0 05 2SD Similarly olivine separates from the Avacha harzburgites have uniform 26Mg ranging between 0 30 and 0 23 Opx and Cpx have slightly higher and more heterogeneous 26Mg values of 0 27 to 0 19 and 0 32 to 0 20 respectively The average 26Mg are 0 26 0 04 2SD n 15 for olivine 0 23 0 04 2SD n 15 for Opx and 0 24 0 10 2SD n 5 for Cpx The 15 olivine Opx pairs yield 26MgOpx-Ol from 0 01 to 0 08 with an average of 0 04 0 04 while the five olivine Cpx pairs yield 26MgCpx-Ol from 0 06 to 0 08 with an average of 0 02 0 11 Fig 5-4 6 Discussion The Avacha tuff-hosted xenoliths analyzed in this study were rapidly carried to surface by volatile-rich andesitic magmas and cooled down immediately after the explosive eruption They have not undergone observable post-eruption alteration as indicated by petrographic data as well as mass gain on ignition proportional to iron contents and hence lack of hydrous phases apart from rare accessory amphibole of mantle origin Ionov 2010 Therefore these xenoliths record genuine sub-arc signatures and provide an opportunity to examine the nature and processes of Mg isotope distribution in an active subduction zone setting Detailed petrographic and geochemical investigation on the Avacha peridotite suite suggested that their modal and chemical compositions mainly reflect melt extraction and moderate subduction-related enrichments Ionov 2010 Since 158 Table 5-3 Mg isotopic compositions of mineral separates Olivine Ol Samples Av-1 duplicate Fine grain duplicate average Av-2 duplicate average Av-3 duplicate average Av-4 duplicate average Av-7 duplicate average Av-8 duplicate duplicate average Av-9 Fine grain duplicate duplicate average Av-10 26Mg 0 27 0 23 0 23 0 26 0 25 0 26 2SD 0 07 0 06 0 06 0 06 0 03 0 07 25Mg 0 13 0 10 0 12 0 14 0 12 0 14 2SD 0 06 0 05 0 52 0 04 0 03 0 06 0 26 0 06 0 13 0 04 0 28 0 32 0 30 0 25 0 07 0 06 0 04 0 06 0 15 0 17 0 16 0 14 0 06 0 04 0 04 0 05 0 28 0 26 0 29 0 28 0 24 0 24 0 26 0 07 0 06 0 06 0 04 0 06 0 07 0 06 0 14 0 12 0 15 0 14 0 11 0 12 0 12 0 06 0 05 0 04 0 03 0 05 0 06 0 04 0 25 0 22 0 04 0 07 0 12 0 11 0 03 0 06 Orthopyroxene Opx 26Mg 0 18 0 18 0 18 0 23 0 19 0 21 0 23 0 22 0 24 0 24 0 24 0 22 0 24 0 22 0 24 0 21 0 23 0 25 0 28 2SD 0 07 0 06 0 05 0 07 0 03 0 05 0 07 0 04 0 05 0 07 0 04 0 05 0 07 0 04 0 05 0 07 0 04 0 05 0 07 25Mg 0 10 0 09 0 09 0 11 0 10 0 10 0 12 0 11 0 12 0 12 0 12 0 11 0 13 0 12 0 12 0 11 0 12 0 14 0 15 2SD 0 05 0 04 0 04 0 06 0 02 0 04 0 06 0 03 0 04 0 06 0 03 0 04 0 06 0 03 0 04 0 06 0 03 0 04 0 06 0 26 0 17 0 18 0 20 0 21 0 19 0 22 0 04 0 06 0 05 0 06 0 07 0 03 0 05 0 14 0 09 0 10 0 11 0 10 0 10 0 11 0 03 0 04 0 04 0 04 0 06 0 02 0 04 159 Clinopyroxene Cpx 26Mg 0 24 0 24 2SD 0 05 0 07 25Mg 0 11 0 12 2SD 0 06 0 04 0 24 0 04 0 11 0 03 0 32 0 33 0 32 0 05 0 06 0 04 0 16 0 15 0 16 0 06 0 04 0 03 0 21 0 24 0 05 0 07 0 11 0 12 0 06 0 04 0 22 0 04 0 12 0 03 duplicate duplicate average Av-11 duplicate average Av-12 duplicate duplicate Fine grain duplicate average Av-13 duplicate average Av-14 duplicate average Av-15 duplicate average Av-16 duplicate average Av-17 duplicate average 0 25 0 24 0 23 0 29 0 29 0 29 0 22 0 26 0 06 0 06 0 04 0 07 0 06 0 05 0 07 0 06 0 13 0 12 0 12 0 16 0 14 0 14 0 12 0 14 0 05 0 04 0 03 0 06 0 05 0 04 0 06 0 04 0 22 0 07 0 10 0 06 0 22 0 27 0 25 0 27 0 25 0 23 0 04 0 05 0 07 0 04 0 05 0 07 0 11 0 15 0 13 0 14 0 14 0 11 0 03 0 04 0 06 0 03 0 04 0 06 0 29 0 29 0 27 0 27 0 27 0 27 0 25 0 07 0 06 0 03 0 07 0 06 0 04 0 07 0 14 0 14 0 13 0 14 0 15 0 14 0 13 0 06 0 05 0 03 0 06 0 04 0 04 0 05 0 28 0 28 0 28 0 27 0 07 0 06 0 04 0 06 0 15 0 14 0 14 0 13 0 05 0 04 0 03 0 05 0 27 0 25 0 26 0 07 0 06 0 05 0 13 0 14 0 14 0 06 0 05 0 04 0 22 0 25 0 24 0 25 0 23 0 24 0 23 0 23 0 23 0 22 0 26 0 23 0 21 0 22 0 21 0 21 0 22 0 21 0 05 0 06 0 03 0 05 0 07 0 04 0 05 0 07 0 04 0 05 0 07 0 04 0 05 0 07 0 04 0 05 0 07 0 04 0 11 0 13 0 12 0 13 0 11 0 12 0 10 0 13 0 11 0 11 0 13 0 12 0 10 0 11 0 10 0 10 0 12 0 11 0 04 0 04 0 02 0 04 0 06 0 03 0 04 0 06 0 03 0 04 0 06 0 03 0 04 0 06 0 03 0 04 0 06 0 03 160 0 22 0 22 0 24 0 07 0 07 0 06 0 12 0 11 0 12 0 04 0 03 0 04 0 23 0 04 0 11 0 02 0 21 0 19 0 20 0 07 0 06 0 04 0 12 0 09 0 11 0 04 0 04 0 03 Table 5-4 Inter-mineral Mg isotope fractionation Samples Av-1 Av-2 Av-3 Av-4 Av-7 Av-8 Av-9 Av-10 Av-11 Av-12 Av-13 Av-14 Av-15 Av-16 Av-17 26MgOpx-Ol 0 06 0 04 0 02 0 08 0 01 0 02 0 06 0 01 0 02 0 03 0 03 0 03 0 05 0 06 0 05 2SD 0 04 0 08 0 07 0 06 0 07 0 05 0 05 0 05 0 06 0 04 0 06 0 08 0 06 0 07 0 06 161 26MgCpx-Ol 0 01 2SD 0 05 0 06 0 07 0 03 0 05 0 04 0 05 0 08 0 06 Figure 5-3 26Mg in bulk Avacha peridotite xenoliths and their constituent minerals plotted against modal and chemical parameters The grey band and horizontal solid line in each panel represent normal mantle range and average from Teng et al 2010 Olivine mode whole-rock WR Al2O3 content and Mg of olivine Ol orthopyroxene Opx and clinopyroxene Cpx are from Ionov 2010 162 Figure 5-4 Inter-mineral fractionation between pyroxene Px and olivine Ol in Avacha peridotites The solid blue and black lines correspond to theoretical equilibrium fractionation lines for Cpx-Ol and Opx-Ol pairs respectively at temperature of 950 C using the parameters provided in Schauble 2011 The blue dash line represents Cpx-Ol equilibrium fractionation line based on measurements on natural samples Liu et al 2011 The average temperature for Avacha peridotites are from Ionov 2010 163 partial melting does not produce measurable Mg isotope fractionation Teng et al 2007 2010a the discussion below mainly focuses on the effects of fluid-rock interaction on the Mg isotope systematics in subduction zones 6 1 Magnesium isotopic compositions of slab-derived fluids The Mg isotopic compositions of the major components of the subducting slabs marine sediments altered oceanic crust and mantle overlap broadly and can be highly varied with 26Mg values both higher and lower than the normal mantle Systematic investigation of Mg isotope systematics in global marine sediments revealed an 26Mg range from 1 34 to 0 52 with the highly weathered continental detritus-dominated silicate sediments defining the high- 26Mg end-member and carbonate-dominated sediments as the low- 26Mg end-member Teng et al 2016 Hu et al 2017 Analyses on altered Pacific crust drilled outboard of the Mariana trench yielded highly variable but mostly heavier-than-mantle 26Mg values 2 76 to 0 21 that are attributed to prolonged low-temperature seawater alteration leading to a gradual build-up of isotopically heavy secondary saponite Huang et al 2018 Likewise abyssal peridotites that underwent extensive hydrothermal alteration and marine weathering are enriched in heavy Mg isotopes 26Mg of 0 25 to 0 10 due to the preferential loss of light Mg into seawater while serpentinization process itself is largely isochemical and does not fractionate Mg isotopes Wimpenny et al 2012 Beinlich et al 2014 Liu et al 2017 Consistent with these slab inputs exhumed metamorphic slab equivalents also display very broad Mg isotopic variations For example cratonic eclogite xenoliths that are fragments of ancient subducted oceanic crust exhibited a wide range in 26Mg 1 60 to 164 0 17 Wang et al 2012 2015a Similarly ultramafic rocks from the Catalina Schist that are considered to represent fragments of slab-mantle interface have 26Mg values ranging from 0 50 to 0 01 with the variation attributed to different modal compositions developed during fluid-peridotite reactions Li et al 2016 Furthermore Mg isotopic exchange between relatively 26 Mg-rich silicates and much lighter fluids or melts equilibrated with carbonates may further differentiate the Mg isotopic compositions of silicate minerals Wang et al 2014a Therefore slab-derived fluids are expected to have variable Mg isotopic compositions 6 2 Magnesium isotopic compositions in mantle wedge peridotites Despite various lines of evidence for fluid metasomatism Fig 5-2 in the Avacha xenolith suite on one hand and highly heterogeneous Mg isotopic compositions inferred for dehydration fluids from subducted slabs on the other hand the peridotite xenoliths analyzed in this study revealed strikingly uniform Mg isotopic compositions that are not correlated with common indices of fluid metasomatism such as Ba La and Pb Ce Fig 55 The 26Mg values in all the 23 Avacha peridotites and their constituent minerals fall within the range for normal upper mantle defined by the Mg isotopic composition of global peridotite xenoliths and oceanic basalts Fig 5-3 In addition the bulk 26Mg values of the Avacha peridotites are identical to those of arc lavas reported for Kamchatka Li et al 2017 Similarly the same Avacha xenoliths display mantle-like 7Li isotopic signatures Ionov and Seitz 2008 which is remarkable considering that lithium is highly fluidmobile and thus a sensitive tracer for slab-derived fluids e g Zack et al 2003 Furthermore the lack of Mg isotope differences between the Avacha peridotites and normal 165 Figure 5-5 Lack of correlation between 26Mg in Avacha peridotites with common indices for slab fluid-peridotite interaction PM stands for primitive mantle and its trace element concentrations are from Sun and McDonough 1989 Its 26Mg is assumed to be identical to the normal mantle average from Teng et al 2010a 166 mantle is consistent with mantle-like Fe isotope signatures in these rocks Weyer and Ionov 2007 This could be explained by the high mass ratios of both Mg and Fe in the mantle peridotites relative to metasomatic melts or fluids Pogge von Strandmann et al 2011 obtained slightly different 26Mg values for six Avacha samples analyzed in this study and for 7Li data by Ionov and Seitz 2008 They found a negative correlation between 7Li and 26Mg which was attributed to diffusiondriven interaction with slab-derived fluids In addition their results for samples Av-1 -4 and -6 26Mg 0 16 to 0 10 Pogge von Strandmann et al 2011 are higher than those we obtained Since their duplicate analyses on the same samples may differ by 0 07 yet the total variation in 26Mg is within the claimed analytical reproducibility the apparent 7Li vs 26Mg correlation and higher 26Mg values might be accidental The mantle-like Mg isotopic composition of the Avacha sub-arc peridotite xenoliths is on the other hand comparable to published Mg isotopic data on some other rocks of likely mantle wedge origin For example the Horoman massif peridotites from Japan are interpreted as samples of ancient mantle wedge above the Hidaka subduction zone Despite various lines of geochemical evidence and the formation of secondary Opx that suggests infiltration by slab-derived fluids melts e g Yoshikawa and Nakamura 2000 the Horoman peridotites yielded mantle-like 26Mg values 0 27 to 0 19 Lai et al 2015 Su et al 2015 reported a 26Mg range from 0 28 to 0 14 for variably metasomatized Tibetan ophiolitic peridotites and attributed the slightly heavier 26Mg values in some of their samples to hydrothermal alteration of the oceanic crust Su et al 2015 It appears that not only the Avacha xenoliths but also other sub-arc mantle wedge suites show no distinctive Mg isotopic signatures relative to normal upper mantle caused 167 by subduction-related fluid infiltration The uniform Mg isotopic compositions of the bulk Avacha peridotites are further supported by the equilibrium isotope fractionation between coexisting olivine and Opx 26MgOpx-Ol 0 01 to 0 08 over the entire temperature range in the Avacha samples in this study Fig 5-4A Unlike Opx however Cpx appears to slightly deviate from the equilibrium fractionation line toward lighter values Fig 5-4A Furthermore the Mg isotope fractionation between Cpx and olivine 26MgCpx-Ol display a positive correlation with 26MgCpx implying that the variation of inter-mineral fractionation is mainly caused by the variation in 26MgCpx Fig 5-4B These observations suggest slight isotope disequilibrium between olivine and Cpx which is consistent with the late-stage origin of Cpx Ionov 2010 Furthermore the larger 0 2-0 3 mm Cpx grains in the Avacha peridotites that were handpicked for the analyses appear to be mainly metasomatic formed by reaction with percolating fluids by contrast to smaller Cpx grains exsolved from coarse residual Opx on cooling 6 3 Behavior of Mg in the mantle wedge-slab system The data on the Avacha sub-arc mantle peridotites in this study suggest that interaction of the arc mantle with fluids expelled from dehydrated Pacific slab does not affect its Mg isotope composition within the analytical uncertainty Fig 5-5 This can be explained by the following reasons Firstly Mg behaves conservatively during metamorphic dehydration at relatively shallow depths therefore only limited amounts of Mg partition to the fluids MgO 0 5 wt% Manning 2004 and references therein This is supported by the consistent Mg isotopic ratios measured in a series of genetically related 168 prograde metamorphic rocks ranging from greenschist to amphibolite to eclogite in a continental subduction zone Wang et al 2014b Similar conclusion has been reached for metamorphic dehydration of metapelites after as much as 75% loss of the initial volatiles Li et al 2014 These studies imply that the majority of slab-hosted Mg will retain in silicates and be carried to the deeper mantle whereas the fluids released beneath the arc volcanic regions must have relatively low Mg contents Secondly although Mg can be transported by fluids it is a highly compatible element with a strong affinity for Mg-silicates relative to fluids Drock fluid of 10 at subMg arc depth Ayer et al 1997 Consequently it would be preferentially incorporated into mantle minerals along fluid pathways Given that the depth to the slab surface beneath Avacha is approximately 100-120 km Gorbatov et al 1997 whereas the peridotite xenoliths originate from the mantle lithosphere above 65 km depth Ionov 2010 ambient mantle peridotites may effectively sequester fluid-carried Mg during its ascent in a process analogous to chromatography e g Ionov et al 2002 Therefore probably only limited amounts of slab fluids could make their way to the shallow part of the mantle wedge from where the xenoliths were entrained by andesitic magmas This is consistent with the low abundance of hydrous amphibole 1 % and relatively low La Yb N 1 in the bulk peridotite xenoliths Ionov 2010 Thirdly the Mg isotopic compositions of slab-expelled fluids are expected to be strongly and continuously modified during their migration in the mantle The equilibrium Mg isotope fractionation between olivine and Opx in the Avacha harzburgites indicates that Mg isotopes reached exchange equilibrium within and between individual samples This is further supported by the lack of correlation between 26MgOpx-Ol with chemical 169 compositions of Opx and olivine Fig 5-3 the lack of Mg concentration zoning in minerals and the identical Mg isotopic compositions obtained for different batches of dissolution for the same mineral phase from a single xenolith sample Table 5-3 Therefore diffusive Mg isotopic equilibration between metasomatic fluids and ambient peridotite is probably a relatively recent and rapid process in the sub-arc mantle This inference is also supported by the links of amphibole and other metasomatic phases disseminated in the peridotites with quenched veins B nard and Ionov 2013 The Mg isotopic composition of percolating fluids and fluid-rich melts can thus be effectively buffered by the mantle wedge as a result of the high Mg content and much larger mass of the mantle wedge compared to the relatively limited amounts of fluids The lack of a clear fluid-imparted 26Mg signature in the Avacha harzburgites does not rule out the possibility that fluid metasomatism may alter the mantle wedge locally where the fluid flux is unusually high This has been demonstrated by significant Mg isotope fractionation in exhumed metamorphic rocks from subduction channels or fore-arc m lange where fragments of different lithologies interact with each other Pogge von Strandmann et al 2015 Chen et al 2016 Wang et al 2017 Figure 5-6 illustrates the expected 26Mg variations in mantle rocks that exchange Mg isotopes with an isotopically enriched fluid phase at various fluid rock ratios in an open-system It shows that fluids derived from oceanic crust are too depleted in MgO to shift the Mg isotopic composition of the mantle because such a shift requires high fluid rock ratios and large 26Mgrock-fluid In comparison fluids derived from serpentinized peridotites are more effective metasomatic agents for Mg In addition the efficiency of metasomatism could be further enhanced if the fluids travel through the mantle by channelized flow in which case they 170 will react less with surrounding peridotites and therefore preserve their 26Mg signatures for longer travel distances during fluid ascent 6 4 Comparison with arc lavas: Implications for Mg cycling in subduction zones and beyond Compositions of arc lavas provide additional observational constraints on the role of slab input in sub-arc mantle sources Although Mg isotopic data on arc lavas remain scarce they have revealed significant among-arc variations These lavas display dominant contributions from slab-derived fluids except that those from Lesser Antilles also exhibit a strong sediment signal Fig 5-7A and B Arc lavas from the Kamchatka and Costa Rica volcanic fronts display mantle-like Mg isotopic compositions whereas those from Lesser Antilles and Philippines are characterized by variably higher 26Mg values which are attributed to dehydration fluids released from altered abyssal peridotites Teng et al 2016 Li et al 2017 There are no apparent correlations between the Mg isotopic compositions of these lavas and their Ba Th and Dy Yb ratios Fig 5-7C and D indicating that neither the amount of fluid additions to their mantle sources nor fractional crystallization of garnet from the magmas could be responsible for the variably 26 Mg-enriched arc lava compositions Thermal structure of a given subduction zone has a profound influence on the dehydration path of a subducting plate and the nature and flux of slab-mantle mass transfer agent fluid vs melt which ultimately affect the compositions of arc magmas The average 26Mg values of these arcs show co-variation trends with their subduction parameters e g 171 Figure 5-6 Illustration of Mg isotope evolution of mantle peridotites during open-system fluid-peridotite interaction at different fluid rock ratios and different 26Mgrock-fluid values The latter is indicated on each of the solid lines in the figures The mantle is assumed to have an average MgO content identical to that of Avacha harzburgites 45 4 wt% with a 26Mg of normal mantle average 0 25 Teng et al 2010a The MgO concentration ranges of fluids derived from oceanic basalt and peridotite are from Kessel et al 2005 and Dvir et al 2011 respectively These fluids are assumed to have a 26Mg of 1 1 Wang et al 2015b 172 Figure 5-7 Compilations of published arc lava 26Mg data with their trace element characteristics A and B Trace element and isotopic characteristics for the published arc lava data in literature which indicate a dominant role of slab-derived fluids in Philippines Kamchatka and Costa Rica whereas both slab-fluids and sediments contribute to genesis of Lesser Antilles arc lavas C and D Magnesium isotopic compositions of published arc lava data plotted against indices for fluid addition and garnet crystallization Data are from Teng et al 2016 and Li et al 2017 173 increasingly higher 26Mg values with an increase in slab surface temperatures and slab depths Fig 5-8 These correlations suggest that the Mg isotopic signatures of the volcanic front lavas may be controlled by specific dehydration reactions that occur at specific P-T conditions in the subduction zone For lavas that are derived at relatively shallow depths the primary water suppliers of the oceanic crust are amphibole chlorite lawsonite zoisite and paragonite Schmidt and Poli 2014 These minerals either do not contain much Mg e g lawsonite zoisite or likely have Mg isotopic compositions similar to those of mantle olivine and pyroxene due to similar Mg-O bond strength e g amphibole By contrast for magmas produced by deeper partial melting the higher slab temperatures facilitate water liberation from the serpentinized peridotite layer of the oceanic slab or those from above the slab surface Compared to oceanic crust serpentinites have much higher water contents and Mg concentrations More importantly the relevant phyllosilicates in serpentinized peridotites generally have Mg isotopic compositions that are different from those in the normal mantle For example an extremely low 26Mg value has been reported for brucite 1 82 Wimpenny et al 2014 whereas variably heavy Mg isotopic compositions have been measured in talc and antigorite up to 0 30 Beinlich et al 2014 as well as in altered serpentinites e g Wimpenny et al 2012 Liu et al 2017 At typical melting depths beneath the arcs K-bearing hydrous phases e g phlogopite and phengite and magnesite which commonly have heavy up to 0 59 Li et al 2011 Liu et al 2011 and light e g 1 Beinlich et al 2014 Mg isotopic compositions respectively are still stable and hence capable of carrying their distinctive Mg isotopic signatures further down into the deep mantle They may later be involved in the formation of isotopically light continental basalts and potassium-rich rocks such as 174 Figure 5-8 The average 26Mg value for each arc as a function of slab temperature and slab depth Slab parameters are from Syracuse et al 2010 Magnesium isotopic data are from Teng et al 2016 and Li et al 2017 175 those in eastern China Yang et al 2012 Huang et al 2015 Tian et al 2016 Li et al 2017 Su et al 2017 Sun et al 2017 Alternatively such melts formed at greater depths could be consumed in the subcontinental lithospheric mantle to generate a diverse range of metasomatic rocks such as wehrlite and pyroxenite that were found to have variable bulk Mg isotopic compositions and disequilibrium inter-mineral Mg isotope fractionations e g Yang et al 2009 Xiao et al 2013 Hu et al 2016b 7 Conclusions The heterogeneous Mg isotopic compositions in different components of downgoing slabs and non-mantle Mg isotopic ratios found in a few studied arc lavas implicate that the mantle wedge may bear atypical Mg isotopic signatures Our Mg isotopic data however show that there is no significant difference between the normal mantle and the metasomatized sub-arc lithospheric mantle sampled by the Avacha xenoliths from the Kamchatka arc Therefore we conclude that a typical mantle wedge where slab input is dominated by addition of fluids from the subducting crust probably has an overwhelmingly mantle-like Mg isotopic composition This could be due to a combination of chromatographic removal of Mg from slab-derived fluids and diffusive Mg isotopic reequilibration between migrating fluids and ambient mantle rocks Arc lavas that display abnormal Mg isotopic compositions may have formed in subduction zones with deeper and hotter slab surface that allows dehydration of serpentinized peridotites to occur Acknowledgements This work was financially supported by the National Science Foundation EAR-1340160 to F -Z Teng and Harry Wheeler Scholarship to Y Hu D 176 Ionov acknowledges support from the Chinese Academy of Sciences CAS President s International Fellowship Initiative PIFI for 2017-18 and from the PNP-INSU program in France Scott Kuehner Bruce Nelson and Bernard Evans are thanked for constructive comments and discussions 177 Chapter 6 Summary and future work This dissertation investigates Mg isotope systematics in crustal and mantle materials I first examine the analytical results of other laboratories compared to ours Next the recycling of crustal materials into the mantle is considered which is the primary mechanism for transporting fractionated low-temperature stable isotopic signatures to the mantle Specifically I approach this topic by studying the Mg isotopic composition of lithological heterogeneities i e pyroxenites in the peridotite-dominated mantle exploring the source materials for metasomatic melts that may lead to anomalous mantle Mg isotopic compositions and investigating processes that can influence the Mg isotope redistribution The results from this dissertation contribute to understanding of the Mg isotope behaviors in subduction zone processes and lay the foundation for using Mg isotopes to trace crust-mantle recycling The main conclusions drawn from this dissertation along with future directions are discussed below: 1 The first topic Chapter 2 investigates intra-mineral and inter-mineral Mg isotopic heterogeneity of two peridotite xenoliths from San Carlos Arizona Previous measurements of a single or a few olivine samples showed large differences between values obtained by different laboratories in the measured 26Mg e g Wiechert and Halliday 2007 Chakrabarti and Jacobsen 2010 We measured two xenoliths where we separated the mineral grains of olivine clinopyroxene and orthopyroxene and determined that all the minerals have a homogeneous Mg isotope composition These results indicate that previously documented large inter-laboratory discrepancies are likely not accurate We find that the San Carlos peridotites have a normal peridotitelike Mg isotopic composition and may be useful as a standard for comparing 178 consistency to other laboratories 2 Chapter 3 focus on Mg isotope variation in mantle pyroxenites which represent the local parts of the mantle that have been significantly modified high melt rock ratio by reacting with melts generated from the mantle Our study finds that reaction of mantle peridotites with metasomatic melts of diverse origins particularly those linked to subducted oceanic crust and carbonates can produce local mantle domains with anomalous Mg isotopic compositions this is opposed to the largely homogeneous Mg isotopic compositions shared by upper mantle peridotites that are not strongly modified These isotopically-altered mantle domains may subsequently serve as the source region of isotopically distinct continental basalts Since the anomalously light continental basalts were explained to reflect contributions from carbonatitic melts supplied by subducted oceanic slab Yang et al 2012 Huang et al 2015 future examination on Mg isotopic variation in carbonated peridotites mantle carbonatites and Mg isotope fractionation during silicate-carbonate immiscibility are required to further our understanding of carbonate-peridotite interaction in the mantle which are still rarely studied 3 Chapter 4 presents a systematic investigation on Mg isotopic variations in subducting sediments recovered from drill cores conducted by Deep Sea Drilling Project and Ocean Drilling Program in front of 10 active subduction zones These sediments display diverse lithologies and serve as a substantial Mg input to global subduction zones albeit their Mg isotope systematics are poorly constrained The 77 bulk sediments measured yield a wide variation in 26Mg 1 34 to 0 46 and together with our study on altered oceanic crust Huang et al in revision they support that subduction 179 of oceanic slab is an important process introducing the large isotope fractionations produced at the Earth s surface into the deep mantle generating local mantle heterogeneities In this respect future studies on oceanic island basalts with endmember isotopic compositions are needed in particular for basalts with a HIMU signature enriched mantle with elevated enriched mantle type II with high 87 238 U 204Pb and with EM II signature Sr 86Sr which that are believed to have incorporated subducted slab components in their mantle sources 4 Chapter 5 presents a Mg isotope investigation on a suite of supra-subduction zone peridotites from Avacha Russia which were chosen due to their genuine sub-arc nature and their representative of western Pacific mantle wedge Results from this study along with published arc lava data Teng et al 2016 Li et al 2017 suggest that fluids released from oceanic crust at pressures of 2 3 GPa are well buffered during their migration in the mantle wedge due to their low Mg concentrations Large-scale dehydration of isotopically distinct materials at higher pressures may be required to produce isotopically diverse arc lavas Therefore to further constrain the role of subduction of oceanic slab in arc magma genesis future studies on arc lavas derived from subduction zones with contrasting thermal structures are needed which permits inter-arc comparison on how dehydration of different hydrous phases at various pressure-temperature conditions affect the Mg isotopic compositions of arc magmas 5 In addition to work presented in this dissertation and the proposed future directions coupling Mg isotope systematics with other non-traditional stable isotopes such as potassium K and boron B which behave differently from Mg during geological processes will give a more complete picture of the working mechanism for subduction 180 zone recycling in general Both K and B are highly enriched in the crust relative to the mantle In addition the heavier isotopes of both the elements are preferentially enriched in seawater These heavy isotopic signatures in the seawater could be transferred to the oceanic slab during low-temperature interactions with seawater which then may be subducted to the mantle and release fluids or melts that inherited these heavy isotopic signatures to the mantle In this way K and B isotopes may provide an archive of recent slab addition whereas Mg isotopes preserve a fingerprint of comparatively longer-term modification event and mantle heterogeneity 181 References Adam J and Green T 2001 Experimentally determined partition coefficients for minor and trace elements in peridotite minerals and carbonatitic melt and their relevance to natural carbonatites Eur J Mineral 13 815 827 All gre C J and Turcotte D L 1986 Implications of a two-component marble-cake mantle Nature 323 123 127 An Y -J Wu F Xiang Y -X Nan X -Y Yu X Yang J -H Yu H -M Xie L -W and Huang F 2014 High-precision Mg isotope analyses of low-Mg rocks by MC-ICPMS Chem Geol 390 9 21 An Y Huang J -X Griffin W L Liu C and Huang F 2017 Isotopic composition of Mg and Fe in garnet peridotites from the Kaapvaal and Siberian cratons Geochim Cosmochim Acta 200 167 185 Anders E and Grevesse N 1989 Abundances of the elements: meteoritic and solar Geochim Cosmochim Acta 53 197 214 Anderson D L 2005 Large igneous provinces delamination and fertile mantle Elements 1 271 275 Arai S Ishimaru S and Okrugin V M 2003 Metasomatized harzburgite xenoliths from Avacha volcano as fragments of mantle wedge of the Kamchatka arc: implication for the metasomatic agent Island Arc 12 233 246 Arai S and Ishimaru S 2008 Insights into petrological characteristics of the lithosphere of mantle wedge beneath arcs through peridotite xenoliths: a review J Petrol 49 665 695 182 Armstrong R 1968 A model for Sr and Pb isotope evolution in a dynamic earth Rev Geophys 6 175 199 Auer S Bindeman I Wallace P Ponomareva V and Portnyagin M 2009 The origin of hydrous high- 18O voluminous volcanism: diverse oxygen isotope values and high magmatic water contents within the volcanic record of Klyuchevskoy volcano Kamchatka Russia Contrib Mineral Petrol 157 209 230 Ayers J C Dittmer S K and Layne G D 1997 Partitioning of elements between peridotite and H2O at 2 0 3 0 GPa and 900 1100 C and application to models of subduction zone processes Earth Planet Sci Lett 150 381 398 Basu A R Junwen W Wankang H Guanghong X and Tatsumoto M 1991 Major element REE and Pb Nd and Sr isotopic geochemistry of Cenozoic volcanic rocks of eastern China: implications for their origin from suboceanic-type mantle reservoirs Earth Planet Sci Lett 105 149 169 Bebout G E 2007 Metamorphic chemical geodynamics of subduction zones Earth Planet Sci Lett 260 373 393 Beinlich A Mavromatis V Austrheim H and Oelkers E H 2014 Inter-mineral Mg isotope fractionation during hydrothermal ultramafic rock alteration implications for the global Mg-cycle Earth Planet Sci Lett 392 166 176 B nard A and Ionov D A 2012 A new petrogenetic model for low Ca boninites: Evidence from veined sub arc xenoliths on melt mantle interaction and melt fractionation Geochem Geophys doi:10 1029 2012GC004145 183 Geosyst 13 Q0AF05 B nard A and Ionov D A 2013 Melt- and fluid-rock interaction in supra-subduction lithospheric mantle: evidence from andesite-hosted veined peridotitic xenoliths J Petrol 54 2339 2378 B nard A Nebel O Ionov D A Arculus R J Shimizu N and M trich N 2016 Primary silica-rich picrite and high-Ca boninite melt inclusions in pyroxenite veins from the Kamchatka sub-arc mantle J Petrol 57 1955 1982 B nard A Arculus R J Nebel O Ionov D A and McAlpine S R B 2017 Silicaenriched mantle sources of subalkaline picrite-boninite-andesite island arc magmas Geochim Cosmochim Acta 199 287 303 Bigeleisen J 1965 Chemistry of isotopes Science 147 463 471 Bizimis M Salters V J M and Bonatti E 2000 Trace and REE content of clinopyroxenes from supra-subduction zone peridotites Implications for melting and enrichment processes in island arcs Chem Geol 165 67 85 Bodinier J -L and Godard M 2014 Orogenic ophiolitic and abyssal peridotites In The Mantle and Core Treatise on Geochemistry ed R W Carlson ElsevierPergamon Oxford pp 103 167 Bogdanov N A 1988 Geology of the Komandorsky deep basin J Phys Earth 36 65 71 Bolou-Bi E B Vigier N Brenot A and Poszwa A 2009 Magnesium isotope compositions of natural reference materials Geostand Geoanal Res 33 95 109 Bostr m K Joensuu O Vald s S Charm W and Glaccum R 1976 Geochemistry and origin of East Pacific sediments doi:10 2973 dsdp proc 34 149 1976 184 sampled during DSDP Leg 34 Bouman C Elliott T and Vroon P Z 2004 Lithium inputs to subduction zones Chem Geol 212 59 79 Bourdon B Tipper E T Fitoussi C and Stracke A 2010 Chondritic Mg isotope composition of the Earth Geochim Cosmochim Acta 74 5069 5083 Bouvier A Wadhwa M Simon S B and Grossman L 2013 Magnesium isotopic fractionation in chondrules from the Murchison and Murray CM2 carbonaceous chondrites Meteorit Planet Sci 48 339 353 Breeding C M Ague J J and Brocker M 2004 Fluid-metasedimentary interactions in subduction zone m lange: Implications for the chemical composition of arc magmas Geology 32 1041 1044 Brenot A Cloquet C Vigier N Carignan J and France-Lanord C F 2008 Magnesium isotope systematics of the lithologically varied Moselle river basin France Geochim Cosmochim Acta 72 5070 5089 Brey G P and Kohler T 1990 Geothermobarometry in fourphase lherzolites II: new thermobarometers and practical assessment of existing thermobarometers J Petrol 31 1353 1378 Brooks C James D E and Hart S R 1976 Ancient lithosphere: its role in young continental volcanism Science 193 1086 1094 Bryant J A Yogodzinski G M and Churikova T G 2007 Melt mantle interactions beneath the Kamchatka arc: Evidence from ultramafic xenoliths from Shiveluch volcano Geochem Geophys Geosyst 8 Q04007 doi:10 1029 2006GC001443 185 Carpentier M Chauvel C and Mattielli N 2008 Pb-Nd isotopic constraints on sedimentary input into the Lesser Antilles arc system Earth Planet Sci Lett 272 199 211 Carpentier M Chauvel C Maury R C and Mattielli N 2009 The zircon effect as recorded by the chemical and Hf isotopic compositions of Lesser Antilles forearc sediments Earth Planet Sci Lett 287 86 99 Chan L -H Leeman W P and Plank T 2006 Lithium isotopic composition of marine sediments Geochem Geophys Geosys 7 doi: 10 1029 2005GC001202 Chang V T -C Makishima A Belshaw N S O Nions R K 2003 Purification of Mg from low-Mg biogenic carbonates for isotope ratio determination using multiple collector ICPMS J Anal At Spectrom 18 296 301 Chakrabarti R and Jacobsen S B 2010 The isotopic composition of magnesium in the inner Solar System Earth Planet Sci Lett 293 349 358 Chakraborty S 1997 Rates and mechanisms of Fe Mg interdiffusion in olivine at 980 1300 C J Geophys Res 102 12317 12331 Chakraborty S 2010 Diffusion coefficients in olivine wadsleyite and ringwoodite Rev Mineral Geochem 72 603 639 Chen S O Reilly S Y Zhou X Griffin W L Zhang G Sun M Feng J and Zhang M 2001 Thermal and petrological structure of the lithosphere beneath Hannuoba Sino-Korean Craton China: evidence from xenoliths Lithos 56 267 301 Chen Y -X Schertl H P Zheng Y -F Huang F Zhou K and Gong Y -Z 2016 Mg-O isotopes trace the origin of Mg-rich fluids in the deeply subdcuted continental crust of Western Alps Earth Planet Sci Lett 456 157 167 186 Cherniak D J and Dimanov A 2010 Diffusion in pyroxene mica and amphibole Rev Mineral Geochem 72 641 690 Choi S H Mukasa S B Zhou X H Xian X H and Andronikov A V 2008 Mantle dynamics beneath East Asia constrained by Sr Nd Pb and Hf isotopic systematics of ultramafic xenoliths and their host basalts from Hannuoba North China Chem Geol 248 40 61 Churikova T Dorendorf F and W rner G 2001 Sources and fluids in the mantle wedge below Kamchatka evidence from across-arc geochemical variation J Petrol 42 1567 1593 Churikova T W rner G Mironov N and Kronz A 2007 Volatile S Cl and F and fluid mobile trace element compositions in melt inclusions: implications for variable fluid sources across the Kamchatka arc Contrib Mineral Petrol 154 217 239 Cole F G and Shaw H F 1983 The nature and origin of authigenic smectites in some Recent marine sediments Clay Minerals 18 239 252 Coltorti M Bonadiman C Hinton R W Siena F and Upton B G J 1999 Carbonatite metasomatism of the oceanic upper mantle: evidence from clinopyroxenes and glasses in ultramafic xenoliths of Grande Comore Indian Ocean J Petrol 40 133 165 Coltorti M 2008 Metasomatism in oceanic & continental lithospheric mantle Geol Soc Spec Pub No 293 eds M Coltorti and M Gregoire Cook H E Zemmels I and Matti J C 1975 X-ray mineralogy data far western Pacific Leg 31 Deep Sea Drilling Project doi: 10 2973 dsdp proc 31 app 1975 187 Donnelly T W and Merrill L 1977 The scavenging of magnesium and other chemical species by biogenic opal in deep-sea sediments Chem Geol 19 167 186 Downes H 2007 Origin and significance of spinel and garnet pyroxenites in the shallow lithospheric mantle: ultramafic massifs in orogenic belts in Western Europe and NW Africa Lithos 99 1 24 Dorendorf F Wiechert U and W rner G 2000 Hydrated sub-arc mantle: a source for the Kluchevskoy volcano Kamchatka Russia Earth Planet Sci Lett 175 69 86 Dosseto A Bourdon B Joron J L and Dupr B 2003 U-Th-Pa-Ra study of the Kamchatka arc: New constraints on the genesis of arc lavas Geochim Cosmochim Acta 67 2857 2877 Dosseto A and Turner S 2014 Reappraisal of uranium-series isotope data in Kamchatka lavas: implications for continental arc magma genesis Geol Soc London Spec Publ 385 103 116 Duggen S Portnyagin M Baker J Ulfbeck D Hoernle K Garbe-Sch nberg D and Grassineau N 2007 Drastic shift in lava geochemistry in the volcanic-front to rear-arc region of the Southern Kamchatkan subduction zone: evidence for the transition from slab surface dehydration to sediment melting Geochim Cosmochim Acta 71 452 480 Dvir O Pettke T Fumagalli P and Kessel R 2011 Fluids in the peridotite water system up to 6 GPa and 800 C: new experimental constrains on dehydration reactions Contrib Mineral Petrol 161 829 844 188 Ewing M and Connary S D 1970 Nepheloid layer in the North Pacific In Hayes J D Ed 1970 Geological Investigations of the North Pacific Geol Soc Am Mem 126 Fan Q C and Hooper P R 1991 The Cenozoic basaltic rocks of eastern China: petrology and chemical composition J Petrol 32 765 810 Fan W -M Zhang H -F Baker J Jarvis K E Mason P R D and Menzies M A 2000 On and off the North China Craton: where is the Archaean keel J Petrol 41 933 950 Fan Q C Zhang H F Sui J L Zhai M G Sun Q and Li N 2005 Magma underplating and Hannuoba present crust mantle transitional zone composition: xenolith petrological and geochemical evidence Sci China Ser D 48 1089 1105 in Chinese with English abstract Fan Q C Du X X Sui J L and Zhao Y W 2010 Genesis of carbonatite from Hannuoba and Yangyuan Acta Petrol Sin 26 3189 3194 in Chinese with English abstract Firsov L V 1987 Geochronology of magmatic rocks from the southwestern part of the Koryak upland Olyutor Depression In: Nikolaeva I V ed Regional Geochronology of Siberia and Soviet Far East Novosibirsk: Nauka pp 7 32 in Russian Flynn L P Mouginis-Mark J and Korton K A 1994 Distribution of thermal areas on an active lava flow field: landsat observations of Kilauea Hawaii July 1991 Bull Volcanol 56 284 296 189 Foley S F 2008 Ancient lithosphere: its role in young continental volcanism Nat Geosci 1 503 510 Foster G L Pogge von Strandmann P A E and Rae J W B 2010 Boron and magnesium isotopic composition of seawater Geochem Geophys Geosyst 11 Q08015 http: dx doi org 10 1029 2010GC003201 Frey F A and Prinz M 1978 Ultramafic inclusions from San Carlos Arizona: petrologic and geochemical data bearing on their petrogenesis Earth Planet Sci Lett 38 129 176 Frey F A 1980 The origin of pyroxenites and garnet pyroxenites from Salt Lake Crater Oahu Hawaii: trace element evidence Am J Sci 280-A 427 449 Galer S J G and O Nions R K 1989 Chemical and isotopic studies of ultramafic inclusions from the San Carlos Volcanic Field Arizona: A bearing on their petrogenesis J Petrol 30 1033 1064 Galy A Belshaw N S Halicz L and O Nions P K 2001 High-precision measurement of magnesium by multiple-collector inductively coupled plasma mass spectrometry Inter J Mass Spectrom 208 89 98 Galy A Bar-Matthews M Halicz L O Nions P K 2002 Mg isotopic composition of carbonate: insight from speleothem formation Earth Planet Sci Lett 201 105 115 Galy A Yoffe O Janney P E Williams R W Cloquet C Alard O Halicz L Wadhwa M Hutcheon I D Ramon E and Carignan J 2003 Magnesium isotope heterogeneity of the isotopic standard SRM980 and new reference materials for magnesium-isotope-ratio measurements J Anal At Spectrom 18 1352 1356 190 Ganguly J 2010 Cation diffusion kinetics in aluminosilicate garnets and geological applications Rev Mineral Geochem 72 559 601 Gao Z -Y and Zhang L -C 1994 A study on the genesis of the mantle-derived inclusions and the host rock in Dashan volcanic cone J NW Univ 24 61 65 in Chinese with English abstract Garcia D Fonteilles M and Moutte J 1994 Sedimentary fractionations between Al Ti and Zr and the genesis of strongly peraluminous granites J Geol 102 411 422 Garrido C J and Bodinier J -L 1999 Diversity of mafic rocks in the Ronda peridotite: evidence for pervasive melt rock reaction during heating of subcontinental lithosphere by upwelling asthenosphere J Petrol 40 729 754 Gorbatov A Kostoglodov V Su rez G and Gordeev E 1997 Seismicity and structure of the Kamchatka subduction zone J Geophys Res 102 17883 17898 Grant T Harlov D and Rhede D 2016 Experimental formation of pyroxenite veins by reactions between olivine and Si Al Ca Na and Cl-rich fluids at 800 C and 800 MPa: implications for fluid metasomatism in the mantle wedge Am Mineral 101 808 818 Griffin J J Windom H and Goldberg E D 1968 The distribution of clay minerals in the World Ocean Deep-Sea Res 15 433 459 Halama R Savov I P Rudnick R L and McDonough W F 2009 Insights into Li and Li isotope cycling and sub-arc metasomatism from veined mantle xenoliths Kamchatka Contrib Mineral Petrol 158 197 222 191 Halbach P Scherhag C Hebisch U and Marchig V 1981 Geochemical and mineralogical control of different genetic types of deep-sea nodules from the Pacific Ocean Miner Deposita 16 59 64 Handler M R Baker J A Schiller M Bennett V C and Yaxley G M 2009 Magnesium stable isotope composition of Earth s upper mantle Earth Planet Sci Lett 282 306 313 Harley S L and Green D H 1982 Garnet orthopyroxene barometry for granulites and peridotites Nature 300 697 701 Harnois L 1988 The CIW index: A new chemical index of weathering Sediment Geol 55 319 322 Herzberg C 2004 Geodynamic information in peridotite petrology J Petrol 45 2507 2530 Higgins J A and Schrag D P 2010 Constraining magnesium cycling in marine sediments using magnesium isotopes Geochim Cosmochim Acta 74 5039 5053 Hippler D Buhl D Witbaard R Richter D K and Immenhauser A 2009 Towards a better understanding of magnesium-isotope ratios from marine skeletal carbonates Geochim Cosmochim Acta 73 6134 6146 Hochstaedter A G Kepezhinskas P K Defant M J Drummond M S and Bellon H 1994 On the tectonic significance of arc volcanism in northern Kamchatka J Geol 102 639 654 Hofmann A W 2014 Sampling mantle heterogeneity through oceanic basalts: isotope and trace elements Vol 3 Treatise on Geochemistry second ed eds H Holland and K Turekian Elsevier Oxford pp 67 101 192 Hopp J and Ionov D A 2011 Tracing partial melting and subduction-related metasomatism in the Kamchatkan mantle wedge using noble gas compositions Earth Planet Sci Lett 302 121 131 Horn D R Horn B M Delach M N 1970 Sedimentary provinces of the North Pacific In Hayes J D Ed Geological Investigations of the North Pacific Geol Soc Am Mem 126 1 22 Hu Y Teng F -Z Zhang H -F Xiao Y and Su B -X 2016a Metasomatism-induced mantle magnesium isotopic heterogeneity: Evidence from pyroxenites Geochim Cosmochim Acta 185 88 111 Hu Y Harrington M D Sun Y Yang Z Konter J and Teng F -Z 2016b Magnesium isotopic homogeneity of San Carlos olivine: a potential standard for Mg isotopic analysis by multi-collector inductively coupled plasma mass spectrometry Rapid commun Mass Spectrom 30 2123 2132 Hu Y Teng F -Z Plank T and Huang K -J 2017 Magnesium isotopic composition of subducting marine sediments Chem Geol 466 15 31 Huang F Glessner J Ianno A Lundstrom C C and Zhang Z -F 2009 Magnesium isotopic composition of igneous rock standards measured by MC-ICP-MS Chem Geol 268 15 23 Huang F Zhang Z Lundstrom C C and Zhi X 2011 Iron and magnesium isotopic compositions of peridotite xenoliths from Eastern China Geochim Cosmochim Acta 75 3318 3334 193 Huang K -J Teng F -Z Wei G -J Ma J -L and Bao Z -Y 2012 Adsorption- and desorption-controlled magnesium isotope fractionation during extreme weathering of basalt in Hainan Island China Earth Planet Sci Lett 359 360 73 83 Huang F Chen L Wu Z and Wang W 2013 First-principles calculations of equilibrium Mg isotope fractionations between garnet clinopyroxene orthopyroxene and olivine: implications for Mg isotope thermometry Earth Planet Sci Lett 367 61 70 Huang K -J 2013a The behavior of magnesium isotopes during low-temperature waterrock interactions PhD dissertation China University of Geosciences Wuhan China Huang K -J Teng F -Z Elsenouy A Li W -Y and Bao Z -Y 2013b Magnesium isotopic variations in loess: Origins and implications Earth Planet Sci Lett 374 60 70 Huang K -J Teng F -Z Plank T Staudigel H Hu Y and Bao Z -Y 2018 Magnesium isotopic composition of the altered oceanic crust: Implications for the magnesium geochemical cycle Geochim Cosmochim Acta 238 357 373 Huang J Li S -G Xiao Y L Ke S Li W -Y and Tian Y 2015 Origin of low 26Mg Cenozoic basalts from South China Block and their geodynamic implications Geochim Cosmochim Acta 164 298 317 Ionov D A Chanefo I and Bodinier J -L 2005 Origin of Fe-rich lherzolites and wehrlites from Tok SE Siberia by reactive melt percolation in refractory mantle peridotites Contrib Mineral Petrol 50 335 353 194 Ionov D A and Seitz H M 2008 Lithium abundances and isotopic compositions in mantle xenoliths from subduction and intra-plate settings: mantle sources vs eruption histories Earth Planet Sci Lett 266 316 331 Ionov D A 2010 Petrology of mantle wedge lithosphere: new data on supra-subduction zone peridotite xenoliths from the andesitic Avacha volcano Kamchatka J Petrol 51 327 361 Ionov D A B nard A and Plechov P Y 2011 Melt evolution in subarc mantle: evidence from heating experiments on spinel-hosted melt inclusions in peridotite xenoliths from the andesitic Avacha volcano Kamchatka Russia Contrib Mineral Petrol 162 1159 1174 Ionov D A Bodinier J -L Mukasa S B and Zanetti A 2002 Mechanisms and sources of mantle metasomatism: major and trace element compositions of peridotite xenoliths from Spitsbergen in the context of numerical modeling J Petrol 43 2219-2259 Irving A J 1980 Petrology and geochemistry of composite ultramafic xenoliths in alkali basalts and implications for magmatic processes within the mantle Am J Sci 280 389 426 Ishikawa T Tera F and Nakazawa T 2001 Boron isotope and trace element systematics of the three volcanic zones in the Kamchatka arc Geochim Cosmochim Acta 65 4523 4537 Ishimaru S Arai S Ishida Y Shirasaka M and Okrugin V M 2006 Melting and multistage metasomatism in the mantle wedge beneath a frontal arc inferred from highly 195 depleted peridotite xenoliths from the Avacha volcano southern Kamchatka J Petrol 48 395 433 Ishimaru S and Arai S 2008a Nickel enrichment in mantle olivine beneath a volcanic front Contrib Mineral Petrol 156 119 131 Ishimaru S and Arai S 2008b Calcic amphiboles in peridotite xenoliths from Avacha volcano Kamchatka and their implications for metasomatic conditions in the mantle wedge Geol Soc London Spec Publ 293 35 55 Ishimaru S and Arai S 2008c Highly silicic glasses in peridotite xenoliths from Avacha volcano Kamchatka arc implications for melting and metasomatism within the sub-arc mantle Lithos 107 93 106 Ishimaru S and Arai S 2011 Peculiar Mg Ca Si metasomatism along a shear zone within the mantle wedge: inference from fine-grained xenoliths from Avacha volcano Kamchatka Contrib Mineral Petrol 161 703 720 Jacob D 2004 Nature and origin of eclogite xenoliths from kimberlites Lithos 77 295 316 Johnson M C and Plank T 1999 Dehydration and melting experiments constrain the fate of subducted sediments Geochem Geophys Geosyst 1 doi: 10 1029 1999GC000014 Kayzar T M Nelson B K Bachmann O Bauer A M and Izbekov P E 2014 Deciphering petrogenic processes using Pb isotope ratios from time-series samples at Bezymianny and Klyuchevskoy volcanoes Central Kamchatka Depression Contrib Mineral Petrol 168 1067 196 Ke S Teng F -Z Li S -G Gao T Liu S -A He Y and Mo X 2016 Mg Sr and O isotope geochemistry of syenites from northwest Xinjiang China: Tracing carbonate recycling during Tethyan oceanic subduction Chem Geol 437 109 119 Kelemen P B Hart S R and Bernstein S 1998 Silica enrichment in the continental upper mantle via melt rock reaction Earth Planet Sci Lett 164 387 406 Kersting A B and Arculus R J 1995 Pb isotope composition of Klyuchevskoy volcano Kamchatka and North Pacific sediments: implications for magma genesis and crustal recycling in the Kamchatkan arc Earth Planet Sci Lett 136 133 148 Kepezhinskas P Defant M J and Drummond M S 1995 Na metasomatism in the island arc mantle by slab melt-peridotite interaction: evidence from mantle xenoliths in the north Kamchatka arc J Petrol 36 1505 1527 Kepezhinskas P Defant M J and Drummond M S 1996 Progressive enrichment of island arc mantle by melt-peridotite interaction inferred from Kamchatka xenoliths Geochim Cosmochim Acta 60 1217 1229 Kepezhinskas P and Defant M J 1996 Contrasting styles of mantle metasomatism above subduction zones: constraints from ultramafic xenoliths in Kamchatka Geophys Monogr AGU 96 307 313 Kepezhinskas P McDermott F Defant M J Hochstaedter A Drummond M S Hawkesworth C J Koloskov A Maury R C and Bellon H 1997 Trace element and Sr-Nd-Pb isotopic constraints on a three-component model of Kamchatka arc petrogenesis Geochim Cosmochim Acta 61 577 600 197 Kepezhinskas P Defant M J and Widom E 2002 Abundance and distribution of PGE and Au in the island-arc mantle: implications for sub-arc metasomatism Lithos 60 113 128 Keshav S and Sen G 2004 The depth of magma fractionation in the oceanic mantle: insights from garnet-bearing xenoliths from Oahu Hawaii Geophys Res Lett 31 L04611 Keshav S Sen G and Presnall D C 2007 Garnet-bearing xenoliths from Salt Lake Crater Oahu Hawaii: high-pressure fractional crystallization in the oceanic mantle J Petrol 48 1681 1724 Kessel R Ulmer P Pettke T Schmidt M W and Thompson A B 2005 The waterbasalt system at 4 to 6 GPa: phase relations and second critical endpoint in a K-free eclogite at 700 to 1400 C Earth Planet Sci Lett 237 873 892 Knoche R Sweeney R J and Luth R W 1999 Carbonation and decarbonation of eclogites: the role of garnet Contrib Mineral Petrol 135 332 339 Korenaga J and Kelemen P B 2000 Major element heterogeneity in the mantle source of the North Atlantic igneous province Earth Planet Sci Lett 184 251 268 Lai Y -J Pogge von Strandmann P A E Dohmen R Takazawa E and Elliott T 2015 The influence of melt infiltration on the Li and Mg isotopic composition of the Horoman Peridotite Massif Geochim Cosmochim Acta 164 318 332 Latouche C and Maillet N 1982 X-ray mineralogy studies Leg 67 sites 495 500 : the middle America Trench doi:10 2973 dsdp proc 67 120 1982 198 transect off Guatemala Le Bas M J Lemaitre R W Streckeisen A and Zanettin B 1986 A chemical classification of volcanic-rocks based on the total alkali silica diagram J Petrol 27 745 750 Lee C T and Rudnick R L 1999 Compositionally stratified cratonic lithosphere: petrology and geochemistry of peridotite xenoliths from the Labait volcano Tanzania In Proceedings of the 7th International Kimberlite Conference l eds J J Gurney J L Gurney and M D Pascoe Red Roof Design Cape Town pp 503 521 Levin V Park J Brandon M Lees J Peyton V Gordeev E and Ozerov A 2002 Crust and upper mantle of Kamchatka from teleseismic receiver functions Tectonophysics 358 233 265 Li S -G Yang W Ke S Meng X Tian H Xu L He Y Huang J Wang X -C Xia Q Sun W Yang X Ren Z -Y Wei H and Liu Y 2017 Deep carbon cycles constrained by a large-scale mantle Mg isotope anomaly in eastern China Natl Sci Rev 4 111 120 Li W -Y Teng F -Z Ke S Rudnick R L Gao S Wu F -Y and Chappell B W 2010 Heterogeneous magnesium isotopic composition of the upper continental crust Geochim Cosmochim Acta 74 6867 6884 Li W -Y Teng F -Z Xiao Y -L and Huang J 2011 High-temperature inter-mineral magnesium isotope fractionation in eclogite from the Dabie orogen China Earth Planet Sci Lett 304 224 230 199 Li W -Y Teng F -Z Wing B A and Xiao Y 2014 Limited magnesium isotope fractionation during metamorphic dehydration in metapelites from the Onawa contact aureole Maine Geochem Geophys Geosyst 15 408 415 Li W -Y Xiao Y -L and Teng F -Z 2016 Metasomatism-induced magnesium isotope fractionation in ultramafic rocks: evidence from the Franciscan Complex California American Geophysical Union Fall Meeting San Francisco Lindsley D H and Dixon S A 1976 Diopside-enstatite equilibria at 850 to 1400 C 5 to 35 kb Am J Sci 276 1285 1301 Ling M -X Sedaghatpour F Teng F -Z Hays P D Strauss J and Sun W 2011 Homogeneous magnesium isotopic composition of seawater: an excellent geostandard for Mg isotope analysis Rapid Commun Mass Spectrom 25 2828 2836 Ling M -X Liu Y -L Williams I S Teng F -Z Yang X -Y Ding X Wei G -J Xie L H Deng W -F and Sun W -D 2013a Formation of the world s largest REE deposit through protracted fluxing of carbonatite by subduction-derived fluids Sci Rep 3 http: dx doi org 10 1038 srep01776 Ling M -X Li Y Ding X Teng F -Z Yang X -Y Fan W -M Xu Y -G and Sun W D 2013b Destruction of the North China Craton induced by ridge subductions J Geol 121 197 213 Liu C Q Masuda A and Xie G H 1994 Major- and trace-element compositions of Cenozoic basalts in eastern China: petrogenesis and mantle source Chem Geol 114 19 42 200 Liu D Zhao Z -D Zhu D -C Niu Y -L Widom E Teng F -Z DePaolo D -J Ke S Xu J -F Wang Q and Mo X -X 2015 Identifying mantle carbonatite metasomatism through Os Sr Mg isotopes in Tibetan ultrapotassic rocks Earth Planet Sci Lett 430 458 469 Liu P -P Teng F -Z Dick H J B Zhou M -F and Chung S -L 2017 Magnesium isotopic composition of the oceanic mantle and oceanic Mg cycling Geochim Cosmochim Acta 206 151 165 Liu Y S Gao S Liu X M Chen X M Zhang W L and Wang X C 2003 Thermodynamic evolution of lithosphere of North China Craton: records from lower crust and upper mantle xenoliths from Hannuoba Chin Sci Bull 48 2371 2377 Liu Y S Yuan H L Gao S Hu Z C Wang X Liu X M and Lin W L 2004 Zircon U Pb ages of olivine pyroxenite xenolith from Hannuoba: links between the 97 158 Ma basaltic underplating and granulite-facies metamorphism Chin Sci Bull 49 1055 1062 Liu Y S Gao S Lee C T A Hu S H Liu X M and Yuan H L 2005 Melt peridotite interactions: links between garnet pyroxenite and high-Mg signature of continental crust Earth Planet Sci Lett 234 39 57 Liu S -A Teng F -Z He Y Ke S and Li S -G 2010a Investigation of magnesium isotope fractionation during granite differentiation: implication for Mg isotopic composition of the continental crust Earth Planet Sci Lett 297 646 654 201 Liu X -M Teng F -Z Rudnick R L McDonough W F Cummings M L 2014 Massive magnesium depletion and isotope fractionation in weathered basalts Geochim Cosmochim Acta 135 336 349 Liu Y S Gao S Gao C G Hu Z C and Ling W L 2010b Garnet-rich granulite xenoliths from the Hannuoba basalts north China: petrogenesis and implications for the Mesozoic crust mantle interaction J Earth Sci 21 669 691 Liu Y S Gao S Hu Z Gao C Zong K and Wang D 2010c Continental and oceanic crust recycling-induced melt peridotite interactions in the Trans-North China Orogen: U Pb dating Hf isotopes and trace elements in zircons from mantle xenoliths J Petrol 51 537 571 Liu S -A Teng F -Z Yang W and Wu F -Y 2011 High-temperature inter-mineral magnesium isotope fractionation in mantle xenoliths from the North China craton Earth Planet Sci Lett 308 131 140 Ma L Teng F -Z Jin L -X Ke S Yang W Gu H -O and Brantley S L 2015 Magnesium isotope fractionation during shale weathering in the Shale Hills Critical Zone Observatory: Accumulation of light Mg isotopes in soils by clay mineral transformation Chem Geol 397 37 50 Macris C A Young E D and Manning C E 2013 Experimental determination of equilibrium magnesium isotope fractionation between spinel forsterite and magnesite from 600 to 800 C Geochim Cosmochim Acta 118 18 32 Manea V C and Manea M 2007 Thermal models beneath Kamchatka and the Pacific plate rejuvenation from a mantle plume impact In: Eichelberger J Izbekov P 202 Ruppert N Gordeev E Lees J eds Volcanism and tectonics of the Kamchatka peninsula and adjacent arcs Geophys Monogr 172 77 90 Manning C E 2004 The chemistry of subduction-zone fluids Earth Planet Sci Lett 223 1 16 Mavromatis V Meister P and Oelkers E H 2014 Using stable Mg isotopes to distinguish dolomite formation mechanisms: A case study from the Peru Margin Chem Geol 385 84 91 McDonough W F and Sun S -S 1995 The composition of the Earth Chem Geol 120 223 253 McDonough W F 1990 Constraints on the composition of the continental lithospheric mantle Earth Planet Sci Lett 101 1 18 McLennan S M Hemming S McDaniel D K and Hanson G N 1993 Geochemical approaches to sedimentation provenance and tectonics Geol Soc Lond Spec Publ 284 21 40 Melson W G Vallier T L Wright T L Byerly G and Nelen J 1976 Chemical diversity of abyssal volcanic glass erupted along the Pacific Atlantic and Indian Ocean seafloor spreading centres In The Geophysics of the Pacific Ocean Basin and its Margin Am Geophys Union Washington DC pp 351 368 Nesbitt H W and Young G M 1982 Early Proterozoic climates and plate motions inferred from major element chemistry of lutites Nature 299 715 717 Nickel K G and Green D H 1985 Empirical geothermobarometry for garnet peridotites and implications for the nature of the lithosphere kimberlites and diamonds Earth Planet Sci Lett 73 158 170 203 Nimis P and Taylor W R 2000 Single clinopyroxene thermobarometry for garnet peridotites Part I Calibration and testing of a Cr-in-Cpx barometer and an enstatitein-Cpx thermometer Contrib Mineral Petrol 139 541 554 Nizkous I V Sanina I A Kissling E and Gontovaya L I 2006 Velocity properties of the lithosphere in the ocean-continent transition zone in the Kamchatka region from seismic tomography data Izvestiya Phys Solid Earth 42 286 296 Norman M D McCulloch M T O Neill H S and Yaxley G M 2006 Magnesium isotopic analysis of olivine by laser-ablation multi-collector ICP-MS: composition dependent matrix effects and a comparison of the Earth and Moon J Anal At Spectrom 21 50 54 Norman M D Yaxley G M Bennett V C and Brandon A D 2006 Magnesium isotopic composition of olivine from the Earth Mars Moon and pallasite parent body Geophys Res Lett 33 L15202 DOI: 10 1029 2006GL026446 O Reilly S Y and Griffin W L 2013 Mantle Metasomatism In: Metasomatism and the Chemical Transformation of Rock Lecture Notes in Earth System Sciences Springer Berlin Heidelberg Patchett J White W M Feldmann H Kielinczuk S and Hofmann A W 1984 Hafnium rare earth element fractionation in the sedimentary system and crustal recycling into the Earth s mantle Earth Planet Sci Lett 69 365 378 Peacock S M 2001 Are the lower planes of double seismic zones caused by serpentine dehydration in subducting oceanic mantle Geology 29 299 302 204 Pearson N J Griffin W L Alard O and O Reilly S Y 2006 The isotopic composition of magnesium in mantle olivine: Records of depletion and metasomatism Chem Geol 226 115 133 Penniston-Dorland S Liu X -M and Rudnick R L 2017 Lithium Isotope Geochemistry Rev Mineral Geochem 82 165 206 Pineau F Semet M P Grassineau N Okrugin V M and Javoy M 1999 The genesis of the stable isotope O H record in arc magmas: the Kamtchatka s case Chem Geol 153 93 124 Piper D 1974 Rare Earth Elements in the sedimentary cycle: A summary Chem Geol 14 285 304 Plank T and Langmuir C H 1998 The chemical composition of subducting sediment and its consequences for the crust and mantle Chem Geol 145 325 394 Plank T 2014 The chemical composition of subducting sediments In The Crust ed R L Rudnick Vol 4 Treatise on Geochemistry second ed eds H Holland and K Turekian Elsevier Oxford pp 607 629 Pogge von Strandmann P A E 2008 Precise magnesium isotope measurements in core top planktic and benthic foraminifera Geochem Geophys Geosyst 9 Q12015 http: dx doi org 10 1029 2008GC002209 Pogge von Strandmann P A E Elliott T Marschall H R Coath C Lai Y J Jeffcoate A B and Ionov D A 2011 Variations of Li and Mg isotope ratios in bulk chondrites and mantle xenoliths Geochim Cosmochim Acta 75 5247 5268 205 Pogge von Strandmann P A E Dohmen R Marschall H R Schumacher J C and Elliott T 2015 Extreme magnesium isotope fractionation at outcrop scale records the mechanism and rate at which reaction fronts advance J Petrol 56 33 58 Portnyagin M Hoernle K Plechov P Mironov N and Khubunaya S 2007 Constraints on mantle melting and composition and nature of slab components in volcanic arcs from volatiles H2O S Cl F and trace elements in melt inclusions from the Kamchatka Arc Earth Planet Sci Lett 255 53 69 Prytulak J Vervoort J D Plank T Yu C 2006 Astoria Fan sediments DSDP site 174 Cascadia Basin: Hf Nd Pb constraints on provenance and outburst flooding Chem Geol 233 276 292 Qian S P Ren Z R Zhang L Hong L B and Liu J Q 2015 Chemical and Pb isotope composition of olivine-hosted melt inclusions from the Hannuoba basalts North China Craton: implications for petrogenesis and mantle source Chem Geol 401 111 125 Qiu L Rudnick R L McDonough W F and Merriman R J 2009 Li and 7Li in mudrocks from the British Caledonides: metamorphism and source influences Geochim Cosmochim Acta 73 7325 7340 Quilty P G Sachs H M Benson W E Vallier T L and Blechschmidt G 1976 Sedimentologic history Leg 34 Deep Sea Drilling Project doi: 10 2973 dsdp proc 34 166 1976 Rapp R P Shimizu N Norman M D and Applegate G S 1999 Reaction between slabderived melts and peridotite in the mantle wedge: experimental constraints at 3 8 GPa Chem Geol 160 335 356 206 Richter F M Dauphas N and Teng F -Z 2009 Non-traditional fractionation of nontraditional isotopes: evaporation chemical diffusion and Soret diffusion Chem Geol 258 92 103 Riley J P and Chester R 1971 Introduction to marine chemistry Academic Press London and New York Rudnick R L McDonough F and Chappell B W 1993 Carbonatite metasomatism in the northern Tanzanian mantle: petrographic and geochemical characteristics Earth Planet Sci Lett 114 463 475 Rudnick R L Gao S Ling W L Liu Y S and McDonough W F 2004 Petrology and geochemistry of spinel peridotite xenoliths from Hannuoba and Qixia North China Craton Lithos 77 609 637 Rudnick R L and Gao S 2014 Composition of the continental crust In The Crust ed R L Rudnick Vol 4 Treatise on Geochemistry second ed eds H Holland and K Turekian Elsevier Oxford pp 1 51 Rose-Koga E F and Albar de F 2010 A data brief on magnesium isotope compositions of marine calcareous sediments and ferromanganese nodules Geochem Geophys Geosyst 11 doi: 10 1029 2009GC002899 Saenger C and Wang Z -R 2014 Magnesium isotope fractionation in biogenic and abiogenic carbonates: implications for paleoenvironmental proxies Quatern Sci Rev 90 1 21 Savov I P Ryan J G D Antonio M Kelley K and Mattie P 2005 Geochemistry of serpentinized peridotites from the Mariana Forearc Conical Seamount ODP leg 207 125: implications for the elemental recycling at subduction zones Geochem Geophys Geosyst 6 Q04J15 doi:10 1029 2004GC000777 Sawhiney B 1972 Selective sorption and fixation of cations by clay minerals: a review Clays Clay Miner 20 93 100 Schauble E A 2011 First-principles estimates of equilibrium magnesium isotope fractionation in silicate oxide carbonate and hexaaquamagnesium 2 crystals Geochim Cosmochim Acta 75 844 869 Schiller M Baker J A and Bizzarro M 2010 26Al 26Mg dating of asteroidal magmatism in the young Solar System Geochim Cosmochim Acta 74 4844 4864 Schilling J G Zajac M Evans R Johnston T White W Devine J D and Kingsley R 1983 Petrologic and geochemical variations along the Mid-Atlantic Ridge from 27 N to 73 N Am J Sci 283 510 586 Schmidt M W and Poli S 2014 Devolatilization during subduction In: Treatise in Geochemistry vol 4 669 701 Schneider R R Price B M ller P J Kroon D and Alexander I 1997 Monsoon related variations in Zaire Congo sediment load and influence of fluvial silicate supply on marine productivity in the east equatorial Atlantic during the last 200 000 years Paleoceanography 12 463 481 Shen B Jacobsen B Lee C -T A Yin Q -Z and Morton D M 2009 The Mg isotopic systematics of granitoids in continental arcs and implications for the role of chemical weathering in crust formation Proc Natl Acad Sci USA 106 20652 20657 208 Sio C K I Dauphas Teng N F -Z Chaussidon M Helz R T and Roskosz M 2013 Discerning crystal growth from diffusion profiles in zoned olivine by in-situ MgFe isotopic analyses Geochim Cosmochim Acta 123 302 321 Sobolev N V Lavrent ev Y G Pokhilenko N P and Usova L V 1973 Chrome-rich garnets from the kimberlites of Yakutia and their parageneses Contrib Mineral Petrol 40 39 52 Sobolev A V Hofmann A W Sobolev S V and Nikogosian I K 2005 An olivine-free mantle source of Hawaiian shield basalts Nature 434 590 597 Sobolev A V Hofmann A W Kuzmin D V Yaxley G M Arndt N T Chung S L Danyushevsky L V Elliott T Frey F A and Garcia M O 2007 The amount of recycled crust in sources of mantle-derived melts Science 316 412 417 Song Y and Frey F A 1989 Geochemistry of peridotite xenoliths in basalt from Hannuoba eastern China: implications for subcontinental mantle heterogeneity Geochim Cosmochim Acta 53 97 113 Spera F J 1984 Carbon dioxide in petrogenesis III: role of volatiles in the ascent of alkaline magma with special reference to xenolith-bearing mafic lavas Contrib Mineral Petrol 88 217 232 Stracke A Zindler A Salters V J M McKenzie D Blichert- Toft J Albar de F and Gr nvold K 2003 Theistareykir revisited Geochem Geophys Geosyst 4 http: dx doi org 10 1029 2001GC000201 Su B -X Zhang H -F Hu Y Santosh M Tang Y -J and Xiao Y 2012a The genesis of mantle-derived sapphirine Am Mineral 97 856 863 209 Su B -X Zhang H -F Ying J -F Tang Y -J Hu Y and Santosh M 2012b Metasomatized lithospheric mantle beneath the Western Qinling Central China: insight into carbonatite melts in the mantle J Geol 120 671 681 Su B -X Chung S -L Zarrinkoub M H Pang K -N Chen L Ji W -Q Brewer A Ying J F and Khatib M M 2014 Composition and structure of the lithospheric mantle beneath NE Iran: constraints from mantle xenoliths Lithos 202 203 267 282 Su B -X Teng F -Z Hu Y Shi R -D Zhou M -F Zhu B Liu F Gong X -H Huang Q S Xiao Y Chen C and He Y -S 2015 Iron and magnesium isotope fractionation in oceanic lithosphere and sub-arc mantle: Perspectives from ophiolites Earth Planet Sci Lett 430 523 532 Su B -X Hu Y Teng F -Z and Xiao Y 2017 Magnesium isotope constraints on subduction contribution to Mesozoic and Cenozoic East Asian continental basalts Chem Geol 466 116 122 Sun S -s and McDonough W F 1989 Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and processes Geol Soc London Spec Publ 42 313 345 Sun Y Teng F -Z Ying J -F Su B -X Hu Y Fan Q -C and Zhou X -H 2017 Magnesium isotopic evidence for ancient subducted oceanic crust in LOMU-like potassium-rich volcanic rocks J Geophys Res 122 7562 7572 Syracuse E M van Keken P E and Abers G A 2010 The global range of subduction zone thermal models Phys Earth Planet Inter 183 73 90 Tang M Rudnick R L and Chauvel C 2014 Sedimentary input to the source of Lesser Antilles lavas: A Li perspective Geochim Cosmochim Acta 144 43 58 210 Tang Y -J Zhang H -F Nakamura E Moriguti T Kobayashi K and Ying J -F 2007 Lithium isotopic systematics of peridotite xenoliths from Hannuoba North China Craton: implications for melt rock interaction in the considerably thinned lithospheric mantle Geochim Cosmochim Acta 71 4327 4341 Tang Y -J Zhang H -F Ying J -F Zhang J and Liu X -M 2008 Refertilization of ancient lithospheric mantle beneath the central North China Craton: evidence from petrology and geochemistry of peridotite xenoliths Lithos 101 435 452 Tang Y -J Zhang H -F Nakamura E and Ying J -F 2011 Multistage melt fluidperidotite interactions in the refertilized lithospheric mantle beneath the North China Craton: constraints from the Li Sr Nd isotopic disequilibrium between minerals of peridotite xenoliths Contrib Mineral Petrol 161 845 861 Tatsumoto M Basu A R Huang W K Wang J W and Xie G H 1992 Sr Nd and Pb isotopes of ultramafic xenoliths in volcanic-rocks of Eastern China: enriched components EMI and EMII in subcontinental lithosphere Earth Planet Sci Lett 113 107 128 Taylor S R and McLennan S M 1985 The Continental Crust: Its Composition and Evolution Blackwell Oxford Telus M Dauphas N Moynier F Tissot F L H Teng F -Z Nabelek P I Craddock P R and Groat L A 2012 Iron zinc magnesium and uranium isotopic fractionation during continental crust differentiation: The tale from migmatites granitoids and pegmatites Geochim Cosmochim Acta 97 247 265 211 Teng F -Z and Yang W 2014 Comparison of factors affecting the accuracy of highprecision magnesium isotope analysis by multi-collector inductively coupled plasma mass spectrometry Rapid Commun Mass Spectrom 28 19 24 Teng F -Z Wadhwa M and Helz R T 2007 Investigation of magnesium isotope fractionation during basalt differentiation: implications for a chondritic composition of the terrestrial mantle Earth Planet Sci Lett 261 84 92 Teng F -Z Li W -Y Ke S Marty B Dauphas N Huang S C Wu F -Y and Pourmand A 2010a Magnesium isotopic composition of the Earth and chondrites Geochim Cosmochim Acta 74 4150 4166 Teng F -Z Li W -Y Rudnick R L and Gardner L R 2010b Contrasting lithium and magnesium isotope fractionation during continental weathering Earth Planet Sci Lett 300 63 71 Teng F -Z Yang W Rudnick R L and Hu Y 2013 Heterogeneous magnesium isotopic composition of the lower continental crust: A xenolith perspective Geochem Geophys Geosys 5 doi: 10 1002 ggge 20238 Teng F -Z Li W -Y Ke S Yang W Liu S -A Sedaghatpour F Wang S -J Huang K -J Hu Y Ling M -X Xiao Y Liu X - M Li X -W Gu H -O Sio C K Wallace D A Su B -X Zhao L Chamberlin J Harrington M and Brewer A 2015 Magnesium isotopic compositions of international geostandards Geostandard Geoanal Res 39 329 339 Teng F -Z Hu Y and Chauvel C 2016 Magnesium isotope geochemistry in arc volcanism Proc Natl Acad Sci U S A 113 7082 7087 Teng F -Z 2017 Magnesium Isotope Geochemistry Rev Mineral Geochem 82 219 287 212 Tian H -C Yang W Li S -G Ke S and Chu Z -Y 2016 Origin of low 26Mg basalts with EM-I component: Evidence for interaction between enriched lithosphere and carbonated asthenosphere Geochim Cosmochim Acta 188 93 105 Tipper E T Galy A and Bickle M J 2006a Riverine evidence for a fractionated reservoir of Ca and Mg on the continents: implications for the oceanic Ca cycle Earth Planet Sci Lett 247 267 279 Tipper E T Galy A Gaillardet J Bickle M J Elderfield H and Carder E A 2006b The magnesium isotope budget of the modern ocean: constraints from riverine magnesium isotope ratios Earth Planet Sci Lett 250 241 253 Tipper E T Lemarchand E Hindshaw R S Reynolds B C and Bourdon B 2012 Seasonal sensitivity of weathering processes: Hints from magnesium isotopes in a glacial stream Chem Geol 312 313 80 92 Tomascak P B Magna T Dohmen R 2016 Advances in Lithium Isotope Geochemistry Springer International Publishing Turner S McDermott F Hawkesworth C and Kepezhinskas P 1998 A U-series study of lavas from Kamchatka and the Aleutians: constraints on source composition and melting processes Contrib Mineral Petrol 133 217 234 Urey H C 1947 The thermodynamic properties of isotopic substances J Chem Soc London 562 581 Uspenskaya T Y Gorshkov A I Sivtsov A V 1987 Mineralogy and internal structure of Fe-Mn nodules from the Clarion Clipperton fracture zone Int Geol Rev 29 363 371 213 Vervoort J D Plank T Prytulak J 2011 The Hf Nd isotopic composition of marine sediments Geochim Cosmochim Acta 75 5903 5926 Venkatarathnam K 1974 Mineralogical data from sites 211 212 213 214 and 215 of the Deep-Sea Drilling Project Leg 22 and origin of noncarbonate sediments in the equatorial Indian Ocean doi:10 2973 dsdp proc 22 123 1974 von Huene R and Scholl D W 1991 Observations at convergent margins concerning sediment subduction subduction erosion and the growth of continental crust Rev Geophys 29 279 316 Walter M J 2003 Melt extraction and compositional variability in mantle lithosphere In Treatise on Geochemistry-The mantle and core ed R W Carlson Carnegie Institute of Washington Washington DC pp 363 394 Wang F -Z Jin L -Y and Xu Y -R 1987 The study of ultramafic inclusions in Cenozoic basalt in Shandong Linqu of Shandong Earth Sci 12 249 256 in Chinese with English abstract Wang C Jin Z -M Gao S Zhang J -F and Zheng S 2010 Eclogite-melt peridotite reaction: experimental constraints on the destruction mechanism of the North China Craton Sci China Earth Sci 53 797 809 Wang S -J Teng F -Z Williams H M and Li S -G 2012 Magnesium isotopic variations in cratonic eclogites: origins and implications Earth Planet Sci Lett 359 360 219 226 Wang S -J Teng F -Z and Li S -G 2014a Tracing carbonate-silicate interaction during subduction using magnesium and oxygen isotopes Nat Commun 5 http: dx doi org 10 1038 ncomms6328 214 Wang S -J Teng F -Z Li S -G and Hong J -A 2014b Magnesium isotopic systematics of mafic rocks during continental subduction Geochim Cosmochim Acta 143 34 48 Wang S -J Teng F -Z Rudnick R L and Li S -G 2015a Magnesium isotope evidence for a recycled origin of cratonic eclogites Geology 43 1071 1074 Wang S -J Teng F -Z Rudnick R L and Li S -G 2015b The behavior of magnesium isotopes in low-grade metamorphosed mudrocks Geochim Cosmochim Acta 165 435 448 Wang S -J Teng F -Z and Bea F 2015c Magnesium isotopic systematics of metapelite in the deep crust and implications for granite petrogenesis Geochem Perspect Lett 1 75 83 Wang S -J Teng F -Z and Scott J 2016 Tracing the origin of continental HIMU-like intraplate volcanism using magnesium isotope systematics Geochim Cosmochim Acta 185 78 87 Wang S -J Teng F -Z Li S -G Zhang L -F Du J -X He Y -S and Niu Y 2017 Tracing subduction zone fluid-rock interactions using trace element and Mg-Sr-Nd isotopes Lithos 290-291 94 103 Westbrook G K Mascle A and Biju-Duval B 1984 Geophysics and the structure of Lesser Antilles Forearc In: Biju-Duval B Moore J C J Eds Initial Reports of the Deep Sea Drilling Program 78A U S Government Printing Office Washington pp 23 28 Wells P R A 1977 Pyroxene thermometry in simple and complex systems Contrib Mineral Petrol 62 129 139 215 Weyer S and Ionov D A 2007 Partial melting and melt percolation in the mantle: the message from Fe isotopes Earth Planet Sci Lett 259 119 133 White W M Dupr B and Vidal P 1985 Isotope and trace element geochemistry of sediments from the Barbados Ridge-Demerara Plain region Atlantic Ocean Geochim Cosmochim Acta 49 1875 1886 Widom E Kepezhinskas P and Defant M 2003 The nature of metasomatism in the subarc mantle wedge: evidence from Re Os isotopes in Kamchatka peridotite xenoliths Chem Geol 196 283 306 Wiechert U and Halliday A N 2007 Non-chondritic magnesium and the origins of the inner terrestrial planets Earth Planet Sci Lett 256 360 371 Wilshire H G and Shervais J W 1975 Al-augite and Cr-diopside ultramafic xenoliths in basaltic rocks from western United States Phys Chem Earth 9 257 272 Wimpenny J G slason S R James R H Gannoun A Pogge von Strandmann P A E and Burton K W 2010 The behaviour of Li and Mg isotopes during primary phase dissolution and secondary mineral formation in basalt Geochim Cosmochim Acta 74 5259 5279 Wimpenny J Burton K W James R H Gannoun A Mokadem F and G slason S R 2011 The behaviour of magnesium and its isotopes during glacial weathering in an ancient shield terrain in West Greenland Earth Planet Sci Lett 304 260 269 Wimpenny J Harvey J and Yin Q 2012 The effects of serpentinization on Mg isotopes in Mid-Atlantic ridge peridotite American Geophysical Union Fall Meeting San Francisco 216 Wimpenny J Colla C A Yin Q -Z Rustad J R and Casey W H 2014a Investigating the behaviour of Mg isotopes during the formation of clay minerals Geochim Cosmochim Acta 128 178 194 Wimpenny J Yin Q -Z Tollstrup D Xie L -W and Sun J 2014b Using Mg isotope ratios to trace Cenozoic weathering changes: A case study from the Chinese Loess Plateau Chem Geol 376 31 43 Wombacher F Eisenhauer A Heuser A and Weyer S 2009 Separation of Mg Ca and Fe from geological reference materials for stable isotope ratio analyses by MC-ICPMS and double-spike TIMS J Anal Atom Spectrom 24 627 636 Wombacher F Eisenhauer A B hm F Gussone N Regenberg M Dullo W -C and R ggeberg A 2011 Magnesium stable isotope fractionation in marine biogenic calcite and aragonite Geochim Cosmochim Acta 75 5797 5818 Woodhead J D Greenwood P Harmon R S and Stoffers P 1993 Oxygen isotope evidence for recycled crust in the source of EM-type ocean island basalts Nature 362 809 813 Woodhead J D Harmon R S and Fraser D G 1987 O S Sr and Pb isotope variations in volcanic rocks from the northern Mariana islands implications for crustal recycling in intra-oceanic arcs Earth Planet Sci Letters 83 39 52 Xiao Y and Zhang H -F 2011 Effects of melt percolation on platinum group elements and Re-Os systematics of peridotites from the Tan-Lu fault zone eastern North China Craton J Geol Soc Lond 168 1201 1214 217 Xiao Y Teng F -Z Zhang H -F and Yang W 2013 Large magnesium isotope fractionation in peridotite xenoliths from eastern North China craton: product of melt rock interaction Geochim Cosmochim Acta 115 241 261 Xie L -W Yin Q -Z Yang J -H Wu F -Y and Yang Y -H 2011 High precision analysis of Mg isotopic composition in olivine by laser ablation MC-ICP-MS J Anal At Spectrom 26 1773 1780 Xu X -S O Reilly S Y Griffin W L Zhou X -M and Huang X - L 1998 The nature of the Cenozoic lithosphere at N shan Eastern China In Mantle Dynamics and Plate Interactions in East Asia eds M F J Flower S L Chung C H Lo and T Y Lee American Geophysical Union pp 167 196 Xu Y G 2002 Evidence for crustal components in the mantle and constrains on crustal recycling mechanism: pyroxenite xenoliths from Hannuoba North China Chem Geol 182 301 322 Yan J Chen J -F Xie Z and Zhou T -X 2003 Mantle derived xenoliths in the late Cretaceous basalts in eastern Shandong: new constraints on the timing of lithospheric thinning in east China Chin Sci Bull 48 1570 1574 in Chinese with English abstract Yang W Teng F -Z and Zhang H -F 2009 Chondritic magnesium isotopic composition of the terrestrial mantle: a case study of peridotite xenoliths from the North China craton Earth Planet Sci Lett 288 475 482 Yang W Teng F -Z Zhang H -F and Li S -G 2012 Magnesium isotopic systematics of continental basalts from the North China craton: implications for tracing subducted carbonate in the mantle Chem Geol 328 185 194 218 Yang W Teng F -Z Li W -Y Liu S -A Ke S Liu Y -S Zhang H -F and Gao S 2016 Magnesium isotopic composition of the deep continental crust Am Mineral 101 243 252 Yaxley G M and Green D H 1998 Reactions between eclogite and peridotite: mantle refertilisation by subduction of oceanic crust Schweiz Mineral Petrogr Mitt 78 243 255 Yaxley G M Crawford A J and Green D H 1991 Evidence for carbonatite metasomatism in spinel peridotite xenoliths from western Victoria Australia Earth Planet Sci Lett 107 305 317 Ying J -F Zhou X -H and Zhang H -F 2004 Geochemical and isotopic investigation of the Laiwu Zibo carbonatites from western Shandong Province China and implications for their petrogenesis and enriched mantle source Lithos 75 413 426 Ying J -F Zhang H -F Tang Y -J Su B -X and Zhou X -H 2013 Diverse crustal components in pyroxenite xenoliths from Junan Sulu orogenic belt: implications for lithospheric modification invoked by continental subduction Chem Geol 356 181 192 Yogodzinski G M Lees J M Churikova T G Dorendorf F W erner G and Volynets O N 2001 Geochemical evidence for the melting of subducting oceanic lithosphere at plate edges Nature 409 500 504 Yoshikawa M and Nakamura E 2000 Geochemical evolution of the Horoman peridotite complex: implications for melt extraction metasomatism and compositional layering in the mantle J Geophys Res 105 2879 2901 219 Young E D and Galy A 2004 The isotope geochemistry and cosmochemistry of magnesium In Geochemistry of Nontraditional Stable Isotopes Rev Mineral Geochem eds C M Hohnson B L Beard and F Albarede pp 197 230 Young E D Tonui E Manning C E Schauble E and Macris C A 2009 Spinel-olivine magnesium isotope thermometry in the mantle and implications for the Mg isotopic composition of Earth Earth Planet Sci Lett 288 524 533 Young E D Manning C E Schauble E A Shahar A Macris C A Lazar C and Jordan M 2015 High-temperature equilibrium isotope fractionation of non-traditional stable isotopes: experiments theory and applications Chem Geol 395 176 195 Yu C -M Zheng J -P and Griffin W L 2006 Petrography and geochemistry of peridotite xenoliths from Hannuoba and significance for lithospheric mantle evolution J China Univ Geosci 17 25 33 in Chinese with English abstract Yu S -Y Xu Y -G Ma J -L Zheng Y -F Kuang Y -S Hong L -B Ge W -C and Tong L -X 2010 Remnants of oceanic lower crust in the subcontinental lithospheric mantle: trace element and Sr Nd O isotope evidence from aluminous garnet pyroxenite xenoliths from Jiaohe Northeast China Earth Planet Sci Lett 297 413 422 Zack T Tomascak P B Rudnick R L Dalp C and McDonough W F 2003 Extremely light Li in orogenic eclogites: the role of isotope fractionation during dehydration in subducted oceanic crust Earth Planet Sci Lett 208 279 290 Zemmels I and Cook H E 1973 X-Ray mineralogy of sediments from the Northeast Pacific and Gulf of Alaska Leg 18 Deep Sea Drilling Project doi:10 2973 dsdp proc 18 app4 1973 220 Zemmels I and Cook H E 1976 X-Ray mineralogy from the Nazca Plate Leg 34 Deep Sea Drilling Project doi:10 2973 dsdp proc 34 151 1976 Zhang H -F Nakamura E Kobayashi K Zhang J Ying J -F Tang Y -J and Niu L -F 2007a Transformation of subcontinental lithospheric mantle through deformation-enhanced peridotite-melt reaction: evidence from a highly fertile mantle xenolith from the North China craton Int Geol Rev 49 658 679 Zhang H -F Ying J -F Shimoda G Kita N T Morishita Y Shao J -A and Tang Y -J 2007b Importance of melt circulation and crust mantle interaction in the lithospheric evolution beneath the North China Craton: evidence from Mesozoic basalt-borne clinopyroxene xenocrysts and pyroxenite xenoliths Lithos 96 67 89 Zhang S -H Zhao Y Song B Hu J -M Liu S -W Yang Y -H Chen F -K Liu X -M and Liu J 2009 Contrasting Late Carboniferous and Late Permian-Middle Triassic intrusive suites from the northern margin of the North China craton: geochronology petrogenesis and tectonic implications Geol Soc Am Bull 121 181 200 Zhang H -F Deloule E Tang Y -J and Ying J -F 2010 Melt rock interaction in remains of refertilized Archean lithospheric mantle in Jiaodong Peninsula North China Craton: Li isotopic evidence Contrib Mineral Petrol 160 261 277 Zhang J -F Wang C and Wang Y F 2012 Experimental constraints on the destruction mechanism of the North China Craton Lithos 149 91 99 Zhang H -F 2009 Peridotite melt interaction: a key point for the destruction of cratonic lithospheric mantle Chin Sci Bull 54 3417 3437 221 Zhao X -M Zhang H -F Zhu X -K Tang S and Tang Y 2010 Iron isotope variations in spinel peridotite xenoliths from North China Craton: implications for mantle metasomatism Contrib Mineral Petrol 160 1 14 Zheng J P O Reilly S Y Griffin W L Lu F -X and Zhang M 1998 Nature and evolution of Cenozoic lithospheric mantle beneath Shandong peninsula SinoKorean Craton eastern China Int Geol Rev 40 471 499 Zheng J P Griffin W L Qi L O Reilly S Y Sun M Zheng S Pearson N Gao J F Yu C M Su Y P Tang H Y Liu Q S and Wu X L 2009 Age and composition of granulite and pyroxenite xenoliths in Hannuoba basalts reflect Paleogene underplating beneath the North China Craton Chem Geol 264 266 280 Zhi X -C Song Y Frey F A Feng J -L and Zhai M -Z 1990 Geochemistry of Hannuoba basalts eastern China: constraints on the origin of continental alkali and tholeiitic basalt Chem Geol 88 1 33 Zhu B Q 1998 Theory and Applications of Isotope Systematics in Geosciences: Evolution of Continental Crust and Mantle in China in Chinese Science Press Beijing Zindler A and Hart S 1986 Chemical geodynamics Annu Rev Earth Planet Sci 14 494 571 222
    • Kehrl, Laura - Ph.D. Dissertation
      Understanding ice-sheet dynamics using geophysical observations and numerical ice-flow models 2018, Kehrl,Laura,Laura Kehrl Copyright 2018 Laura Kehrl Understanding ice-sheet dynamics using geophysical observations and numerical ice-flow models Laura Kehrl A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy University of Washington 2018 Reading Committee: Ian R Joughin Chair Knut Christianson Benjamin E Smith Program Authorized to Offer Degree: Earth and Space Sciences University of Washington Abstract Understanding ice-sheet dynamics using geophysical observations and numerical ice-flow models Laura Kehrl Chair of the Supervisory Committee: Dr Ian R Joughin Applied Physics Laboratory Mass loss from the world s ice sheets is one of the largest sources of uncertainty in sealevel rise projections for the 21st century One way to improve sea-level rise projections is to better understand the processes driving past ice-sheet mass loss This dissertation investigates past changes in ice flow for two regions: 1 Helheim and Kangerlussuaq Glaciers two fast-flowing tidewater glaciers in Southeast Greenland and 2 the Allan Hills Blue Ice Area a slow-flowing blue ice area in East Antarctica For both regions I constrain changes in ice-sheet dynamics using geophysical observations and interpret those changes using numerical ice-flow models At Helheim and Kangerlussuaq I examine seasonal and interannual variations in surface velocity elevation and terminus position from 2001 to 2016 I show that glacier dynamics depend on the extent of floating ice near the terminus Helheim s grounded terminus calved small nontabular icebergs while Kangerlussuaq s floating ice tongue calved large tabular icebergs Furthermore terminus-driven seasonal speedups and dynamic thinning were generally larger at Helheim than at Kangerlussuaq likely due to its grounded rather than floating ice tongue To interpret the observed changes at Helheim and Kangerlussuaq I use inverse methods to investigate changes in basal conditions under the two glaciers The basal shear stress under Helheim and Kangerlussuaq decreased or remained relatively constant during terminus-driven speedup events suggesting that changes in the stress balance were generally supported outside of the region of fast flow Finally I use the inferred basal shear stresses to help constrain the form of the basal sliding law At the Allan Hills Blue Ice Area I combine ice-penetrating radar data an ice-flow model and age constraints to determine a potential site to drill a million-year-old ice core I also show that thickness anomalies in the englacial stratigraphy suggest that glacier velocity was 30% of present-day values during the last glaciation While the dynamics of the Allan Hills Blue Ice Area are likely unimportant for sea-level rise projections an ice core from the region could provide insight into the past stability of the Ross Sea Sector and West Antarctic Ice Sheet TABLE OF CONTENTS List of Figures v List of Tables viii Chapter 1 Introduction 1 1 1 Motivation 1 1 2 Research approach and tools 2 1 3 Organization of the dissertation 2 Chapter 2 Seasonal and interannual variabilities in terminus position glacier velocity and surface elevation at Helheim and Kangerlussuaq Glaciers from 2008 to 2016 4 2 1 Abstract 4 2 2 Introduction 5 2 3 Methods 7 2 3 1 Glacier velocity 7 2 3 2 Terminus position 8 2 3 3 Iceberg-calving behavior 9 2 3 4 Glacier surface elevation 10 2 3 5 Bed elevation and flotation condition 10 2 3 6 Surface-elevation change rate 12 2 3 7 Glacier surface runoff ice-m lange rigidity and sea-ice fraction 14 2 4 2 4 1 Results 15 i Helheim Glacier 15 2 4 2 2 5 Kangerlussuaq Glacier 17 Discussion 19 2 5 1 Long-term behavior from 2008 to 2016 19 2 5 2 Seasonal variations in glacier velocity and surface elevation 21 2 5 3 Seasonal variability in terminus position and iceberg-calving behavior 25 2 6 Conclusions 29 2 7 Acknowledgments 30 Chapter 3 Basal conditions for Helheim and Kangerlussuaq Glaciers Southeast Greenland from 2001 to 2016 43 3 1 Abstract 43 3 2 Introduction 44 3 3 Methods 47 3 3 1 Data 47 3 3 2 Modeling 48 3 3 3 Driving stress 51 3 4 Results 52 3 4 1 Data-model misfit and regularization 52 3 4 2 Spatial patterns in the inferred basal shear stress 53 3 4 3 Temporal variations in the inferred basal shear stress 54 3 4 4 Relationship between the basal velocity and basal shear stress 57 3 5 ii Discussion 57 3 5 1 Spatial variations in the basal shear stress 57 3 5 2 Interannual variations in the basal shear stress 59 3 5 3 Seasonal variations in the basal shear stress 59 3 5 4 Basal sliding law 61 3 6 Conclusions 64 3 7 Data availability 65 3 8 Acknowledgments 65 Chapter 4 Evaluating the duration and continuity of potential climate records from the Allan Hills Blue Ice Area East Antarctica 76 4 1 Abstract 76 4 2 Introduction 76 4 3 Methods 78 4 3 1 Ice-penetrating radar 78 4 3 2 Age constraints 79 4 3 3 Ice-flow model 79 4 4 Results 83 4 5 Discussion 84 4 6 Conclusions 86 4 7 Acknowledgments 87 Chapter 5 Summary and future work 92 5 1 Dynamics of Helheim and Kangerlussuaq Glaciers Southeast Greenland 92 5 2 Dynamics of the Allan Hills Blue Ice Area East Antarctica 93 References 95 Appendix A 107 iii A 1 Assumption of negligible basal melt 107 A 2 Surface-elevation change rate error estimates 108 Appendix B 113 Appendix C 128 Uncertainty estimates for radar-detected layer ages 128 iv LIST OF FIGURES Figure 2 1 Overview map and cross section for Helheim 32 Figure 2 2 Overview map and cross section for Kangerlussuaq 33 Figure 2 3 Observational record for Helheim 34 Figure 2 4 Terminus position and ice-m lange conditions for Helheim 36 Figure 2 5 Range of observed surface velocities and elevations at Helheim 37 Figure 2 6 Observational record for Kangerlussuaq 38 Figure 2 7 Terminus position and ice-m lange conditions at Kangerlussuaq 39 Figure 2 8 Linear trends in surface velocity and elevation at Kangerlussuaq 40 Figure 2 9 Range of observed surface velocities and elevations at Kangerlussuaq 41 Figure 2 10 Terminus position and dynamic surface-elevation change rate 42 Figure 3 1 Overview map of Helheim and Kangerlussuaq 66 Figure 3 2 Data-model misfit for Kangerlussuaq 67 Figure 3 3 Basal shear stress for Kangerlussuaq 68 Figure 3 4 Basal sliding ratio at Kangerlussuaq and Helheim 70 Figure 3 5 Basal shear stress at Kangerlussuaq from 2001 to 2016 71 Figure 3 6 Basal shear stress at Kangerlussuaq from 2011 to 2015 72 Figure 3 7 Basal shear stress at Helheim from 2004 to 2015 73 Figure 3 8 Basal shear stress at Helheim from 2011 to 2015 74 v Figure 3 9 Basal shear stress vs basal velocity 75 Figure 4 1 Overview map of the Allan Hills Blue Ice Area East Antarctica 88 Figure 4 2 Measured and modeled depth-age scale for Radar Track 1 89 Figure 4 3 Model sensitivity to different parameters 90 Figure 4 4 Dated radiostratigraphy and model-derived age structure along Radar Track 1 91 Figure A 1 Tabular and non-tabular iceberg calving 109 Figure A 2 Comparison of the DEM and flux-gate methods 110 Figure A 3 Flux-gate analysis for Helheim 111 Figure A 4 Flux-gate analysis for Kangerlussuaq 112 Figure B 1 Depth-averaged temperature at Helheim and Kangerlussuaq 113 Figure B 2 Data-model misfit for Helheim 114 Figure B 3 Basal shear stress for Helheim 115 Figure B 4 Basal shear stress at Kangerlussuaq for FS-CT 117 Figure B 5 Basal shear stress at Kangerlussuaq for FS-MT 118 Figure B 6 Basal shear stress at Kangerlussuaq for SSA-CT 119 Figure B 7 Basal shear stress at Kangerlussuaq for SSA-MT 120 Figure B 8 Basal shear stress at Helheim for FS-CT 121 Figure B 9 Basal shear stress at Helheim for FS-MT 122 Figure B 10 Basal shear stress at Helheim for SSA-CT 123 vi Figure B 11 Basal shear stress at Helheim for SSA-MT 124 Figure C 1 Average age difference for radar-detected layers 129 Figure C 2 Height of 500-ka and 1-Ma ice above the bed 129 vii LIST OF TABLES Table B 1 Surface-elevation and -velocity data sources for KG Surface-elevation measurements are from ASTER Howat et al 2007 SPIRIT Korona et al 2009 Worldview WV and TanDEM-X TDM Kehrl et al 2017 Surface-velocity measurements are derived from optical feature tracking of ASTER and Landsat images HOWAT Howat 2017 and from speckle tracking of TerraSAR-X and TanDEM-X radar images TSX Joughin et al 2010a 2016 125 Table B 2 Surface-elevation and -velocity data sources for HG See Table B 1 for a description of abbreviations 126 Table B 3 Pearson correlation coefficients and p-values for the relationship between the basal shear stress and basal velocity at KG for the lower glacier upper glacier and combined region of fast flow 127 Table B 4 Pearson correlation coefficients and p-values for the relationship between the basal shear stress and basal velocity at HG for the lower glacier upper glacier and combined region of fast flow 127 viii ACKNOWLEDGEMENTS First and foremost I want to thank my supervisor Ian Joughin for his stimulating discussions and unwavering support as I decided on a career path that is right for me He has been an excellent role model and I will miss our discussions about science and about life I also want to thank my committee members Ben Smith Knut Christianson Michelle Koutnik and Cecilia Bitz for their support over the years I appreciate that their doors have always been open and that they have always been willing to lend a hand when needed The glaciology community at the University of Washington has been like a family over the last five years Twit Conway while not on my committee served as the mentor for my chapter on the Allan Hills Blue Ice Area He pushed me to ask difficult questions and supported me when I didn t know the answers Nick Holschuh and Daniel Shapero also provided inspiration for the ideas in this dissertation My officemates David Lilien David Shean Max Stevens Taryn Black and Kristin Poinar always managed to put a smile on my face Bob Hawley and Huw Horgan while not at the University of Washington helped me survive my undergraduate and master s degrees to reach this point They have been wonderful mentors and friends over the years and I appreciate their continued support and wittiness Lastly I want to thank my family and friends for reminding me that there is life outside of my dissertation My grandparents Gagi and Grandpa Bud have always supported me no matter what I wanted to do My brother Ryan has spent countless hours on the phone with me Most importantly I want to thank my husband Mike and cuddle monster Zoey for their unconditional love and support ix DEDICATION To my grandparents Gagi and Grandpa Bud who supported me every step of the way x 1 Chapter 1 Introduction 1 1 Motivation Over the next century sea-level rise will likely displace people from their homes and increase the incidence of storm-related flooding IPCC 2014a To mitigate these effects it is important that we accurately predict future sea-level rise One of the greatest sources of uncertainty in sea-level rise projections is the contribution from the world s ice sheets The ice sheets currently account for one-fifth of total sea-level rise but their contributions are expected to rise over the coming century IPCC 2014b The ice sheets lose mass through changes in ice discharge surface mass balance and basal mass balance Variations in ice discharge are driven by changes in ice-sheet dynamics In Antarctica satellite measurements numerical models and in situ measurements show increased ice discharge from the ice streams draining the continent likely caused by a loss of ice shelf buttressing due to enhanced basal melt Pritchard et al 2012 In Greenland many tidewater glaciers have retreated sped up and thinned in the last two decades also likely as a result of changes in the ocean Rignot and Kanagaratnam 2006 Moon and Joughin 2008 Moon et al 2012 Yet many of the processes driving these rapid changes remain poorly understood IPCC 2014b making it difficult to predict ice-sheet mass loss over the next century One way to improve projections of ice-sheet mass loss is to better understand the processes driving recent changes in the ice sheets Satellites started collecting measurements over the ice sheets in the 1970s and the temporal and spatial resolution of these measurements has only improved since then These measurements provide a window into the recent past and an opportunity to understand changes in ice-sheet dynamics on seasonal interannual and decadal timescales These measurements provide the foundation for two of the chapters in this dissertation 2 Another way to improve predictions is to better understand the link between the climate and ice-sheet dynamics in the distant past By analyzing the behavior of the ice sheets in the past we may develop a better understanding of what might happen in the future For example marine and geologic evidence suggests that the West Antarctic Ice Sheet may have collapsed during the previous interglacial Marine Isotope Stage 5e when the sea level was 6 to 9 m higher than it is today Alley et al 2015 Dutton et al 2015 No direct measurements of climate and ice flow exist for this period or other past periods and so we instead turn to geophysical measurements and ice cores for these analyses These ideas provide the foundation for my remaining chapter 1 2 Research approach and tools In this dissertation I combine geophysical observations and numerical ice-flow models to understand recent and past changes in ice-sheet dynamics The geophysical observations constrain the changes in ice flow and thickness and the numerical models assist the interpretation of those changes in terms of physical processes For changes in ice-sheet dynamics that occurred during the satellite era I rely on satellite measurements For changes in ice-sheet dynamics that occurred long before the satellite era I turn to ice-penetrating radar to map the englacial stratigraphy and then interpret the stratigraphy in terms of the ice-flow dynamics that must have caused it 1 3 Organization of the dissertation This thesis is organized into five chapters beginning with this introduction and ending with a summary and suggestions for future work The middle three chapters are scientific papers Chapter 2 is a manuscript entitled Seasonal and interannual variabilities in terminus position glacier velocity and surface elevation at Helheim and Kangerlussuaq Glaciers from 2008 to 2016 which was published in Journal of Geophysical Research: Earth Surface Kehrl et 3 al 2017 This paper developed a detailed timeseries of terminus position glacier surface velocity surface elevation and iceberg-calving behavior to investigate geometric and environmental controls on the seasonal and interannual evolution of Helheim and Kangerlussuaq Glaciers two of Greenland s largest tidewater glaciers The supplement for this paper can be found in Appendix A Chapter 3 is a manuscript entitled Basal conditions for Helheim and Kangerlussuaq Glaciers Southeast Greenland from 2001 to 2016 which will be submitted to The Cryosphere in October 2018 This manuscript used the timeseries developed in Chapter 2 combined with measurements from other studies Howat et al 2005 2007 Howat 2017 to run more than 20 basal inversions to characterize seasonal and interannual variations in basal conditions at Helheim and Kangerlussuaq The supplement for this paper can be found in Appendix B Chapter 4 is a manuscript entitled Evaluating the duration and continuity of potential climate records from the Allan Hills Blue Ice Area East Antarctica which was published in Geophysical Research Letters Kehrl et al 2018 This paper combined an ice-flow model icepenetrating radar data and existing ages constraints to determine a potential site for a continuous million-year-old ice core in the Allan Hills The supplement for this paper can be found in Appendix C 4 Chapter 2 Seasonal and interannual variabilities in terminus position glacier velocity and surface elevation at Helheim and Kangerlussuaq Glaciers from 2008 to 2016 Chapter 2 in full is a reprint of Seasonal and interannual variabilities in terminus position glacier velocity and surface elevation at Helheim and Kangerlussuaq Glaciers from 2008 to 2016 authored by L M Kehrl I Joughin D E Shean D Floricioiu and L Krieger as it appears in Journal of Geophysical Research: Earth Surface 2017 The supplement for this paper can be found in Appendix A The dissertation author was the primary investigator and author of this paper 2 1 Abstract The dynamic response of Greenland tidewater glaciers to oceanic and atmospheric change has varied both spatially and temporally While some of this variability is likely related to regional climate signals glacier geometry also appears to be important In this study we investigated the environmental and geometric controls on the seasonal and interannual evolution of Helheim and Kangerlussuaq Glaciers Southeast Greenland from 2008 to 2016 by combining year-round satellite measurements of terminus position glacier velocity and surface elevation While Helheim remained relatively stable with a lightly grounded terminus over this time period Kangerlussuaq continued to lose mass as its grounding line retreated into deeper water By summer 2011 Kangerlussuaq s grounding line had retreated into shallower water and the glacier had a 5km-long floating ice tongue We also observed seasonal variations in surface velocity and elevation at both glaciers At Helheim seasonal speedups and dynamic thinning occurred in the late summer when the terminus was most retreated At Kangerlussuaq we observed summer speedups due to surface-melt-induced basal lubrication and winter speedups due to ice-shelf retreat We suggest that Helheim and Kangerlussuaq behaved differently on a seasonal timescale 5 due to differences in the spatial extent of floating ice near their termini which affected icebergcalving behavior Given that seasonal speedups and dynamic thinning can alter this spatial extent these variations may be important for understanding the long-term evolution of these and other Greenland tidewater glaciers 2 2 Introduction The contribution of the Greenland Ice Sheet to sea-level rise more than quadrupled from 1991 2001 to 2002 2011 Shepherd et al 2012 as a result of enhanced surface melt and increased ice discharge from tidewater glaciers Enderlin et al 2014 Van Den Broeke et al 2016 While this widespread increase in ice discharge has been attributed likely correctly to ocean warming Holland et al 2008 Hanna et al 2009 Murray et al 2010 Straneo et al 2010 there has been significant spatial variability in the dynamic response of individual glaciers Moon et al 2012 Tidewater glaciers in the same fjord which are likely subject to similar oceanic and atmospheric conditions have behaved differently Rignot et al 2016 Motyka et al 2017 indicating that individual glacier characteristics such as glacier geometry likely play an important role in modulating a glacier s dynamic response Enderlin et al 2013 Amundson 2016 Felikson et al 2017 The dynamics of tidewater glaciers are sensitive to changes near the terminus Nick et al 2009 Many Greenland tidewater glaciers have a grounded or nearly grounded terminus and consequently the terminus position often closely corresponds with the grounding-line position location where the ice transitions from grounded to floating Several processes have been proposed to link terminus retreat to oceanic and atmospheric changes including enhanced submarine melt Holland et al 2008 Motyka et al 2011 ice-m lange weakening Joughin et al 2008a Amundson et al 2010 and increased hydrofracture of water-filled crevasses Benn et 6 al 2007 Once retreat is initiated the glacier s geometry will affect its dynamic response Retreat into deeper water can promote speedup thinning and further retreat whereas retreat into shallower water can cause the glacier to slow and to perhaps stabilize Meier and Post 1987 Schoof 2007 Furthermore lateral constrictions in glacier width can help stabilize the terminus at a particular location Gudmundsson et al 2012 As a result of these dynamic feedbacks it can be difficult to attribute a change in glacier dynamics to a specific oceanic or atmospheric forcing To address this difficulty several studies have turned to seasonal records of terminus position Howat et al 2010 Seale et al 2011 Schild and Hamilton 2013 and glacier velocity Moon et al 2014 2015 By comparing these records to different climatic variables over many years these studies have attempted to correlate seasonal changes in terminus position and or glacier velocity to oceanic and atmospheric changes However often there is no clear relationship between glacier dynamics and environmental change: the timing of the seasonal onset of retreat and speedup varies spatially from glacier to glacier and temporally from year to year at individual glaciers Schild and Hamilton 2013 Moon et al 2014 Seasonal and multiyear variations in glacier geometry ice thickness surface slope and grounding-line position have been put forward as a possible mechanism to help explain this temporal variability at individual glaciers e g Schild and Hamilton 2013 but until recently we have lacked the necessary seasonal surface-elevation records to explore this hypothesis further In this study we combined year-round records of terminus position glacier velocity and surface elevation from 2008 to 2016 to investigate seasonal and multiyear changes in glacier geometry at Helheim and Kangerlussuaq Glaciers Figure 2 1 and Figure 2 2 the two largest tidewater glaciers in Southeast Greenland Helheim and Kangerlussuaq collectively drain 8% of the Greenland Ice Sheet area Nick et al 2013 Both glaciers rapidly retreated accelerated and 7 thinned in the early 2000s Rignot et al 2004 Howat et al 2005 2007 Stearns and Hamilton 2007 The resulting dynamic mass loss from these two glaciers alone accounted for roughly 30% of the 2000-2012 dynamic mass loss from the entire Greenland Ice Sheet Enderlin et al 2014 After their retreats ended in 2006 both glaciers slowed but Helheim stopped thinning while Kangerlussuaq continued thinning Howat et al 2011 Bevan et al 2012 The glaciers have also behaved differently on a seasonal timescale terminus positions typically varied seasonally at Kangerlussuaq but showed little seasonality at Helheim Joughin et al 2008b Schild and Hamilton 2013 By comparing the varying evolution of Helheim and Kangerlussuaq over seasonal and multiyear timescales we develop new insights into how differences in glacier geometry affect a glacier s dynamic response to oceanic and atmospheric changes In particular we focus our analysis on changes in the spatial extent of floating ice near the terminus which can affect iceberg-calving behavior Benn et al 2007 2 3 Methods We combined several different data sets to develop seasonal records of glacier velocity terminus position iceberg-calving behavior surface elevation and surface-elevation change rates at Helheim and Kangerlussuaq from 2008 to 2016 To interpret these observations we also considered bed elevations ice-m lange conditions sea-ice fraction SIF and modeled glacier surface runoff in our analysis 2 3 1 Glacier velocity To measure glacier velocity we applied speckle tracking techniques Joughin 2002 Joughin et al 2010 to pairs of synthetic aperture radar SAR images from the German Aerospace Center s DLR TerraSAR-X TSX mission Krieger et al 2013 This satellite started acquiring 8 data over Helheim and Kangerlussuaq in September 2008 A second virtually identical satellite TDX was launched in June 2010 to complete the tandem configuration of the TanDEM-X TDM mission The repeat period of each satellite allows velocity estimates to be determined over intervals as short as 11 days Due to missed acquisitions some velocity estimates were computed using 22- or 33-day intervals We did not attempt estimates for intervals longer than 33 days leaving some gaps in our record All velocity estimates were smoothed with a moving average filter to a spatial resolution of 300 m Conversion from the radar line-of-sight to the horizontal across-track direction under a surface-parallel flow assumption can yield slope-dependent errors of up to 3% Joughin et al 2010a For observations collected along the same satellite track this is a systematic error common to all estimates so the error does not apply to changes in velocity Of our 264 velocity products 237 90% were collected along the same track for each glacier 2 3 2 Terminus position We measured terminus position by digitizing the location where the calving front intersects the longitudinal profiles shown in Figure 2 1a and Figure 2 2a We chose this method over the box method Moon and Joughin 2008 because it allowed us to include satellite images with high cloud cover and Landsat 7 images with gaps due to the scan line corrector SLC failure which improved the temporal resolution of our records Terminus positions were digitized in all available TSX TDX radar images and Worldview-1 2 3 WV and Landsat 7 8 panchromatic scenes We report terminus position relative to the 2008-2016 average with a negative positive value indicating a more retreated advanced terminus position than average To assess uncertainties in the measured terminus positions we compared digitized terminus positions from satellite images that were acquired on the same day when no large calving events occurred between the acquisitions Consequently the uncertainty estimates account for 9 errors due to manual digitization and errors introduced by including satellite images with different spatial resolution acquisition geometry and spectral characteristics The 22 pairs of coincident digitized terminus positions differed by 2-42 m with a root-mean-squared RMS difference of 24 m 2 3 3 Iceberg-calving behavior We used the available satellite images to assess the calved iceberg type Two types of icebergs have been observed previously at Helheim and Kangerlussuaq: tabular and non-tabular Joughin et al 2008b Tabular icebergs have a large longitudinal width-to-height ratio and do not capsize when they calve These icebergs can be distinguished in the satellite images by the presence of crevasses on their surfaces Figure A 1a Non-tabular icebergs capsize when they calve due to their smaller width-to-height ratio and greater buoyancy-driven torque Figure A 1b Amundson et al 2010 James et al 2014 We compared subsequent satellite images to determine the type of new icebergs in the fjord after each iceberg-calving episode We refer to periods of iceberg calving as episodes rather than as events because we cannot determine if the icebergs calved during a single event or during a series of events that occurred over several days Iceberg-calving episodes were divided into three types: 1 tabular 2 non-tabular and 3 mixed both tabular and non-tabular We considered an episode to be tabular if new tabular icebergs appeared in the fjord that could account for the observed retreat between subsequent satellites images If new tabular icebergs appeared but could not account for all of the observed retreat then the episode was considered mixed All other episodes were considered nontabular This approach may incorrectly characterize some tabular or mixed iceberg-calving episodes as non-tabular if the tabular icebergs broke up and overturned between subsequent 10 satellite images However the observed tabular icebergs typically remained intact for several weeks suggesting that we observed a substantial fraction of tabular icebergs before they broke up 2 3 4 Glacier surface elevation We combined point surface-elevation measurements from NASA s Airborne Topographic Mapper ATM Krabill et al 2014 with digital elevation models DEM from WV GeoEye-1 SPIRIT Korona et al 2009 and TanDEM-X TDM DEMs from WV and GeoEye-1 were created following the stereo-photogrammetry techniques outlined in Shean et al 2016 TDM DEMs were processed from bistatic SAR acquisitions using the Integrated TanDEM-X Processor ITP Rossi et al 2012 Large absolute elevation offsets in the resulting TDM DEMs are caused by baseline-dependent interferometric SAR InSAR height ambiguities and were corrected by adjusting the absolute phase offset during InSAR processing To reduce georeferencing errors all DEMs were co-registered to ground control points GCPs over exposed rock We adjusted the DEMs using a rigid-body translation that minimized the elevation difference between the DEMs and the GCPs Shean et al 2016 All available ICESat-1 data Zwally et al 2003 LVIS data Blair and Hofton 2010 and ATM data over exposed bedrock surfaces were included as GCPs After co-registration the uncertainty of each DEM was estimated using the normalized median absolute deviation NMAD H hle and H hle 2009 of all GCP-DEM differences Shean et al 2016 The average NMAD for the WV and TDM DEMs was 1 18 m and 1 77 m respectively 2 3 5 Bed elevation and flotation condition While several gridded bed-elevation products exist for Helheim and Kangerlussuaq e g Bamber and Griggs 2013 Morlighem et al 2014 there are large discrepancies among these 11 products due to different interpolation methods Consequently to avoid errors introduced by interpolation we used bed-elevation point measurements from the Helheim and Kangerlussuaq 2006-2014 Composite V3 products from the Center for Remote Sensing of Ice Sheets CReSIS We also included the CReSIS radar transect from 21 May 2001 in our analysis for Helheim Some of the radar thicknesses in the Composite V3 products were collected over floating ice To determine if a radar thickness was collected over floating ice we used the ATM surfaceelevation measurements that were collected concurrently with the radar-thickness measurements If the measured surface elevation was less than or equal to the flotation height for a given radar thickness assuming an ice density of 917 kg m-3 and seawater density of 1025 kg m-3 then we considered the ice at that location to be floating and removed that measurement from our analysis of bed elevations Almost all radar thicknesses acquired over the lower 5 km of Kangerlussuaq were collected over floating ice Figure 2 2c After removing these points the RMS for the bedelevation crossover differences in regions below sea level improved from 113 m to 45 m for Helheim and from 90 m to 44 m for Kangerlussuaq Given that crossover differences were
    • Koehler, Matthew - Ph.D. Dissertation
      The co-evolution of life and the nitrogen cycle on the early Earth 2018, Koehler,Matthew,Matthew Koehler Copyright 2018 Matthew C Koehler The co-evolution of life and the nitrogen cycle on the early Earth Matthew C Koehler A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy University of Washington 2018 Reading Committee: Roger Buick Ph D Chair David Catling Ph D Eric Steig Ph D Program Authorized to Offer Degree: Earth and Space Sciences & Astrobiology University of Washington Abstract The co-evolution of life and the nitrogen cycle on the early Earth Matthew C Koehler Chair of the Supervisory Committee: Roger Buick Ph D Earth and Spaces Sciences & Astrobiology Nitrogen is an essential element for all life as we know it Its abundance and speciation in the atmosphere and ocean has been dynamic through Earth s history The dynamic nature of nitrogen cycling is invariably linked to the dynamic nature of oxygenation on the early Earth and the evolution of these two biogeochemical cycles influenced biological evolution and diversification Exploration of the mode tempo and scale of such changes will add to our mechanistic understanding of these components of the inhabited Earth-system and should provide insight into how these cycles might operate on an alien planet that is different from our own Here I explore four critical points in Earth s redox and biological evolution: Chapter 2 The dominantly anoxic world of the Mesoarchean when aerobic nitrogen metabolisms did not exist and bioavailable nitrogen could only be obtained through biological nitrogen fixation or remineralized ammonium Biological nitrogen fixation by the Mo-nitrogenase dominated Chapter 3 Incipient oxygenation 300 Myr before the great oxidation event We start to characterize the mode tempo and scale of these transient oxygenation events and provide the oldest evidence of aerobic nitrogen cycling 2 66 Gyr ago Chapter 4 The state of nitrogen cycling during the Mesoproterozoic We find further evidence that the Mesoproterozoic was characterized by an oceanic nitrate minimum and that this nitrate scarcity could have prevented the radiation of early eukaryotes until the end of the Proterozoic Chapter 5 Nitrogen cycling across the Hirnantian glaciation and end Ordovician mass extinction event We find that oceans during this period were characterized by widespread ocean anoxia that led to a scarcity in nitrate much like in the Mesoproterozoic Nitrate scarcity was temporarily alleviated in near-shore settings during glaciation Much like in the Pacific Ocean during the Last Glacial Maximum the carbonate compensation depth at multiple sites seems to have deepened substantially during peak Hirnantian glaciation These results highlight the relationship between ecology evolution nitrogen speciation abundance and surface oxygen TABLE OF CONTENTS List of Figures vi List of Tables xiii Chapter 1: Introduction 17 Chapter 2: Nitrogen isotope evidence for anoxic deep marine environments from the Mesoarchean Mosquito Creek Formation Australia 22 2 1 Abstract 22 2 2 Introduction 22 2 3 Geologic Setting 23 2 4 Methods 28 2 5 Results 30 2 6 Discussion 32 2 6 1 Isotopic fidelity 32 2 6 2 Redox and paleoecology 34 2 6 3 Marine or non-marine 36 2 6 4 Implications for microbial evolution 40 2 7 Acknowledgements 40 Chapter 3: Transient surface ocean oxygenation recorded in the 2 66 Ga Jeerinah Formation Australia 41 3 1 Abstract 41 3 2 Introduction 42 i 3 3 Geologic Setting 44 3 4 Results 45 3 4 1 AIDP-2 48 3 4 2 AIDP-3 48 3 5 Discussion 51 3 5 1 Proxy Alteration 51 3 5 2 Interpretation of nitrogen isotopic data 53 3 5 3 Interpretation of selenium concentrations and isotopic ratios 54 3 5 4 Transient surface ocean oxygenation in the Neoarchean: mode and tempo 55 3 6 Conclusion 57 3 7 Methods 58 3 8 Acknowledgements 58 Chapter 4: Spatial and temporal trends in Precambrian nitrogen cycling: a Mesoproterozoic offshore nitrate minimum 60 4 1 Abstract 60 4 2 Introduction 61 4 3 Precambrian nitrogen cycling 64 4 4 Location and geologic setting 68 4 4 1 Bangemall Supergroup 68 4 4 2 Roper Group 71 4 5 Analytical methods 73 4 5 1 Sample preparation for bulk rock analyses 73 4 5 2 Kerogen extraction for organic nitrogen 73 ii 4 5 3 Isotopic analyses 74 4 6 Results 75 4 6 1 Bangemall basin 75 4 6 2 Roper basin 76 4 7 Discussion 77 4 7 1 Diagenesis and metamorphism 77 4 7 1 1 Oxic diagenesis: effects on preserved 15Nbulk 77 4 7 1 2 Anoxic diagenesis: effects on preserved 15Nbulk 81 4 7 1 3 Diagenetic effects on the 13C of organic matter 82 4 7 1 4 Metamorphism 83 4 7 2 Carbon cycling in the Bangemall and Roper Basins 85 4 7 3 Nitrogen cycling in the Bangemall and Roper Basins 86 4 7 4 15Nbulk variability in the Bangemall and Roper basins 91 4 7 5 Comparing the Bangemall and Roper basins to the Belt basin 96 4 7 6 Mesoproterozoic nitrate minimum 97 4 7 7 Implications for life 100 4 8 Conclusion 102 4 9 Acknowledgements 103 Chapter 5: Nitrogen and carbon biogeochemistry across the Ordovician-Silurian boundary at Dob s Linn Scotland 105 5 1 Abstract 105 5 2 Introduction 106 5 3 Geologic setting 109 iii 5 4 Methods 110 5 4 1 Sample preparation: bulk and kerogen isolates 110 5 4 2 Analytical Methods 111 5 4 3 Percent carbonate 112 5 5 Results 113 5 5 1 Carbonate 113 5 5 2 Nitrogen and carbon 113 5 6 Discussion 115 5 6 1 Potential for post-depositional alteration 115 5 6 1 1 Diagenesis Error Bookmark not defined 5 6 1 2 Metamorphism 120 5 6 1 3 Diffusion advection effects on 15Nbulk 120 5 6 2 Interpretation of nitrogen isotopes 124 5 6 3 Carbon cycling 129 5 7 Conclusion 136 5 8 Acknowledgements 137 Chapter 6: Conclusion 138 Bibliography 141 Appendix 161 7 1 Chapter 2 161 7 2 Chapter 3 168 7 2 1 Methods 168 iv 7 2 2 Identifying authigenic Se enrichments 170 7 2 3 Correlating the cores 171 7 2 4 Metamorphism 172 7 2 5 Figures 174 7 3 Chapter 4 182 7 4 Chapter 5 188 v LIST OF FIGURES Figure 2 1 Location map modeled after 21 showing A a simplified map of the Pilbara Craton B a simplified map of the Mosquito Creek Basin highlighted in A 25 Figure 2 2 ABDP 5 stratigraphy from 50m to 150m depth The section analyzed in this study is indicated by the black and white striped bar Modified from 1 26 Figure 2 3 Field photo of rippled sandy C silty D and shaley E turbidite facies of the Mosquito Creek Formation modified from 21 Red arrow indicates stratigraphic up The samples studied here predominately came from D and E facies The coin for scale is 25 mm in diameter 27 Figure 2 4 Simplified stratigraphic section with 13Corg 15Nbulk 15Nkerogen TOC% Grey circles are from bulk measurements and red squares represent kerogen measurements 31 Figure 2 5 Cross-plots of carbon and nitrogen isotope abundance measurements A 15Nbulk vs 13Corg B 15Nkerogen vs 13Corg C 15Nkerogen vs 15Nbulk D 13Corg vs TOC E 15Nbulk vs TN F 15Nbulk vs C Nbulk G 15Nkerogen vs C Nkerogen H 13Corg vs TOC for sandstone samples TOC Total organic carbon TN Total nitrogen Note all isotope measurements are expressed in parts per thousand and C N ratios are atomic ratios 33 Figure 2 6 Schematic of the relationship between 13Corg values and oxidant availability Scenarios 1 and 2 are described in section 2 6 2 38 Figure 2 7 Cross-plots of 13Corg vs 15Nbulk for the A Mesoarchean and B Neoarchean Grey circles are marine data orange X s are non-marine data blue circles are from this study Horizontal dashed line separates 15Nbulk values consistent with nitrogen fixation by the Mo-nitrogenase below from 15Nbulk values consistent with some degree of aerobic nitrogen cycling above Shaded colors represent the following: Green - 13Corg values consistent with carbon fixation using the Calvin Cycle Blue - 13Corg values consistent with carbon fixation using the Wood-Ljungdahl pathway Purple - 13Corg values consistent with methanotrophic carbon cycling Only least-altered data from the Archean were included and were taken from a compilation in 38 with added data from 17 45 While there are vi more 15N values from the Archean not shown here the fidelity of those measurements based on potential alteration and or large analytical uncertainly see supplementary materials of 41 have been excluded 39 Figure 3 1 Crossplots of A 15Nkerogen values and C N ratios B Se enrichment factor EF and Se ppm C Se ppm and TOC % and D Se ppm and total sulfur TS % For A Note the linear covariance in AIDP-3 stage I compared to the rest of the core Samples from AIDP-2 come from all stages For B the procedure for calculating EF is described in the SI Appendix Note the particularly strong enrichments in AIDP-2 Stage II Samples from AIDP-3 come from all stages For C and D Note the weaker correlations in AIDP-2 compared to AIDP-3 The variation and so the increase in selenium abundance in AIDP-2 cannot be explained through variations in TOC and TS Conversely much of the variation in selenium abundance in AIDP-3 can be explained by variations in TOC and TS This suggests the increase in Se during stage II of AIDP-2 is likely caused by an increased flux of Se to the basin 47 Figure 3 2 Chemostratigraphy of AIDP-2 showing 15Nbulk grey dots 15Nkerogen red dots C N ratios 82 78Se and Se concentrations Stages are delineated by horizontal dotted lines Shaded area for 82 78Se represents crustal values Down-core depth is given in meters 49 Figure 3 3 Chemostratigraphy of AIDP-3 showing 15Nbulk grey dots 15Nkerogen red dots C N ratios 82 78Se and Se concentrations Stages are delineated by horizontal dotted lines Lithological symbols follow the Figure 3 2 legend Note the different 15N and C N scales between AIDP-2 and AIDP-3 required to accommodate Stage I of AIDP-3 Shaded area for 82 78Se represents crustal values Down-core depth is given in meters 50 Figure 4 1 Average nitrogen isotopic composition of bulk marine sedimentary rocks from offshore environments Data are compiled from the literature see 41 for references and from this study Each point represents a time-point average the black solid line marks the running mean over three points Where no basinal gradient is present all data were used Samples from hydrothermal cherts and amphibolite facies were excluded 67 Figure 4 2 Locations of the Bangemall Supergroup and Roper Group boxes and the approximate locations of sample collection dots 69 vii Figure 4 3 Stratigraphic 15Nbulk plots of Bangemall transects A Fords Creek B Irregully C Wandarry Depositional facies in each panel are as in panel D with deep subtidal above the top dot-dash line shallow subtidal in the middle and peritidal below the bottom dot-dash line Panel D is a normalized compilation of all the Bangemall transects where the positions of each point are relative to the thickness of the shallow subtidal facies The red line indicates the running mean over three points 78 Figure 4 4 15Nkerogen vs TOC Nkerogen total organic carbon kerogen-bound nitrogen for Bangemall yellow diamonds and Roper black circle samples 84 Figure 4 5 15Nbulk plotted against 13Corg for the Bangemall A and Roper B basins The plot also includes those samples from the Bangemall basin that do not belong to one of the three main transects shown in Figure 4 3A-C Error bars are 1 92 Figure 4 6 Histogram of Proterozoic offshore 15Nbulk data A Neoproterozoic era 1 00 548 Ga with data from Ader et al 125 and Kikumoto et al 189 B Mesoproterozoic era 1 6-1 0 Ga with data from St eken 24 Luo et al 171 and this study C Paleoproterozoic era 2 5-1 6 Ga with data from Busigny et al 113 Kump et al 115 and Godfrey et al 72 Subsets of the datasets listed above were taken to only include offshore environments: from Ader et al 125 the Camil Carmelo and Copacel sections from Brazil and all data from Svalbard and North Canada From Kikumoto et al 189 all data from the Doushantuo Formation From St eken 24 data from the Newland Formation in Deep Creek From the Bangemall and Roper of this study deep and basinal data respectively From Luo et al 171 all data From Busigny et al 113 data from the Brockman Iron Formation From Kump et al 115 data from above 180 meters in core depth from 0m to 180m From Godfrey et al 72 data from cores MGS-7 and MGS-8 The inset in panel A shows modern marine sediment data from Tesdal et al 179 The modern data show no correlation with water depth and are therefore not separated by facies Paleoproterozoic data from the Aravalli Group 74 190 were excluded because they are of higher metamorphic grade and their depositional environment is uncertain 93 Figure 4 7 Schematic of the proposed Mesoproterozoic nitrogen cycle Shown are fluxes F both in and out of a shallow water nitrate reservoir as we propose for the Bangemall and Roper basins Blue flux arrows represent fluxes that when varied are most likely able to viii change the isotopic composition of a relatively small nitrate reservoir All other fluxes are unlikely to change the isotopic composition of the nitrate reservoir Flux labels are as follows: Ffixation N2 fixation flux Fnitri nitrification of organic matter to nitrate Fassim assimilation of nitrate into biomass Fden wc water-column denitrification Fden sed sedimentary denitrification Fupwell upwelling of ammonium from anoxic waters 95 Figure 5 1 Cross-plots of nitrogen and organic carbon isotopes abundances and ratios from Dob s Linn shales bentonites and kerogen isolates Grey dots are sample measurements that follow the axes labels 116 Figure 5 2 Nitrogen and organic carbon isotope chemostratigraphy through the Dob s Linn section Symbols for bulk shale bentonite and kerogen measurements are as in Figure 5 1 117 Figure 5 3 Nitrogen and organic carbon chemostratigraphy through the Dob s Linn section Symbols for bulk shale bentonite and kerogen measurements are as in Figure 5 1 118 Figure 5 4 P-value relationships comparing 15N populations between a depositional environments within each period and b periods for each depositional environment Green boxes represent a statistically significant rejection of the null hypothesis that the population data from the compared parameters are equal Red boxes represent a failure to reject the null hypothesis Orange boxes represent a limited number of locations samples such that any rejection or acceptance of the null hypothesis should be questioned All statistical tests were two-tailed T-tests with an 0 05 133 Figure 5 5 Comparison of outer shelf and basinal 15N values between the eras of the Proterozoic and data from the O-S boundary The colored bars in the O-S panel are from outer shelf-basinal depositional environments The rest of the shallower data are included in black and white for further comparison Notice the similarities between the Mesoproterozoic and O-S nitrogen data Figure modified from Koehler et al 2017 and the included excluded Proterozoic data are as in figure 6 of that reference More recent Proterozoic nitrogen data from Canfield et al 2018 and Zerkle et al 2017 are also excluded due to their shallower depositional environments 134 ix Figure 0 1 Cross-plots of 13Corg vs 15Nbulk comparing sections with known oxygenation events to the data from this study AIDP labels refer to data from the Jeerinah Formation Koehler et al in review McRae labels refer to data from the Mt McRae Shale Garvin et al 2009 161 Figure 0 2 Cross-plots of 13Corg vs 15Nbulk comparing sub-anoxic marine section to the data from this study AIDP and McRae labels are as in Error Reference source not found Soanesville and Witwatersrand data are from St eken et al 2015 162 Figure 0 3 Cross-plots of 13Corg vs 15Nbulk comparing non-marine data from the Mesoarchean and Neoarchean to the data in this study Hardey and Bellary data are from St eken et al 2017 Witwatersrand including the Booysens data are from St eken et al 2015 Booysens points that are distinguished from the rest of the Witwatersrand data were deposited in brackish water marine environment with a noticeable fluvial contribution Lalla Rookh data come from St eken and Buick 2018 163 Figure 0 4 Chemostratigraphy of the Carawine Dolomite from AIDP-2 showing 15Nbulk 13C and C N ratios Down-core depth is given in meters The 15Nbulk values are similar to the underlying stage III in AIDP-2 174 Figure 0 5 Drill cores AIDP-2 and AIDP-3 correlated using the impact spherule layer and nitrogen isotope Stage II as described in the main text The dashed line for the lower Stage II boundary as it approaches AIDP-3 indicates that the precise stratigraphic extent of elevated 15N values that defines Stage II is unknown for AIDP-3 This is because Stage I in AIDP-3 hosts altered 15N values 175 Figure 0 6 Same as Figure 3 2 in text but shows selenium EF normalized to aluminum instead of selenium ppm This coupled with Figure 3 1C and Figure 3 1D demonstrates that the increase in Se during Stage II of AIDP-2 is likely not due to lithological changes For comparison the Se EF enrichment in the McRae shale is over 3 times greater than the enrichment found here Grey vertical bars represent crustal values 176 Figure 0 7 Same as Figure 3 3 in text but shows selenium EF normalized to aluminum instead of selenium ppm Note the different x-axis scale from Figure 0 6 Grey vertical bars represent crustal values 177 x Figure 0 8 Thin section images from the basalt underlying the Jeerinah Fm at 252 72m stratigraphic depth containing characteristics of multiple metasomatic episodes Panel a is groundmass comprising K-feldspar spherulites Kfs Panels b-e are of amygdales b and c are of the same large amygdale filled with chert Cht and sericite Ser with calcite Cal veins and blocky quartz Qtz d and e are of different amygdales with abundant chlorite Chl chlorite overgrowths and larger calcite crystals f is a large chert vein with sericite intraveins surrounded by blocky quartz crystals and with abundant chlorite in some regions g and h are zoomed-in images of f Scale bars are 1mm for panels a and f 200 m for panels b c d and g and 100 m for panels e and h 179 Figure 0 9 Schematic representation of the mechanisms that result in elevated 15N values as described in the text Note that all pathways denoted by the red and blue arrows have a net isotopic preference for the light isotope 14N That preference is expressed in instances of partial conversion red arrows but not in instances of complete conversion blue arrows The starting 15N value for the initial NH4 in each case is 0 consistent with nitrogen fixation by the Mo-nitrogenase Note that the only mechanism that incorporates a spatial component is mechanism i For a full schematic of the nitrogen cycle including fractionation factors and oxidation state for each reaction species see Figure 4 of 38 180 Figure 0 10 Least-altered 15N values from 3 4 Ga to 2 0 Ga Circles are marine data points X s are non-marine data points Black points are lower- sub greenschist samples grey points are greenschist samples and red points are from this study Blue shaded area covers 15N values consistent with nitrogen fixation by the Mo-nitrogenase Data taken from a compilation in 38 with added data from 17 45 282 While there are more 15N values from the Archean not shown here the fidelity of the measurements based on potential alteration and or large analytical uncertainly see supplementary materials of 41 have been excluded 181 Figure 0 11 Cross-plot between 13Ccarb and 18Ocarb for carbonates detected during the Hirnantian at Dob s Linn 188 xi Figure 0 12 Total nitrogen and organic carbon abundances through the Dob s Linn section Symbols are as in Figure 5 1 of the main text 189 Figure 0 13 Relationship between bulk Corg N ratios and total organic carbon abundances for shales and bentonites a through the sampled Dob s Linn section and b zoomed in on shales from the Upper Hartfell Shale Bentonites in b are from all parts of the section Symbols are as in Figure 5 1 of the main text 190 Figure 0 14 Histogram of 15N values across the O-S boundary Notice how data from shallower depositional environments are on-average more positive regardless of latitude 191 xii LIST OF TABLES Table 3 1 Stratigraphic stages of AIDP-2 and AIDP-3 with associated nitrogen and carbon measurements Stratigraphic stage boundaries were determined using the linear regression model described above C N ratios are given as atomic ratios 46 Table 4 2 List of nitrogen cycle steps that induce large isotope fractionation effects and their respective fractionation factors Fractionations are expressed as 15Nsubstrate 15Nproduct References: 1 Zhang et al 37 2 Casciotti 58 3 Frey et al 120 4 Brunner et al 121 5 Fulton et al 122 6 Fuchsman et al 102 7 Altabet & Francois 31 8 Kessler et al 123 9 Sigman et al 124 65 Table 4 3 15Nbulk values for the Bangemall basin transects and whole basin the Roper basin and the Belt basin Belt data are from St eken 24 Parenthetical facies labels are specific to the Roper basin Uncertainties are expressed as 1 n number of samples 79 Table 4 4 13Corg in the Bangemall Roper and Belt basins Parenthetical facies labels are specific to the Roper basin Data from the Jixian basin are taken from Guo et al 158 Uncertainties are expressed as 1 n number of samples 80 Table 5 5 Table showing average 15N values in age and depositional bins Standard deviations are 1 Data are from this study Luo et al 266 Melchin et al 239 and Laporte et al 199 Parenthetical numbers represent number of units analyzed in each bin 132 Table 0 6 Bulk measurements from the Mesoarchean Mosquito Creek Formation Chapter 2 166 Table 0 7 Kerogen measurements from the Mesoarchean Mosquito Creek Formation Chapter 2 167 Table 0 8 Kerogen data Included are data from both the Roper and the Bangemall groups Chapter 4 182 Table 0 9 New data from the Bangemall Group Chapter 4 183 xiii Table 0 10 New data from the Roper Group Facies abbreviations: t d shoreline tidedominated shoreline sand d shelf sand-dominated shelf s d shelf storm-dominated shelf c s platform coastal sand platform Chapter 4 186 xiv ACKNOWLEDGEMENTS I am extremely fortunate to have had Roger as an advisor His creativity and near encyclopedic knowledge of early Earth geology and biogeochemistry are admirable but equally admirable is his professionalism and character He is incredibly reasonable easy to work with easy get along with and his door has always been open Not to mention his sense of humor is right up my alley Being a great scientist means little if you re not a great person Roger is most definitely both If I become a scientist when I grow up he is the kind I want to be My time at UW is sprinkled with inspirational people who have helped me develop academically and personally Particularly I have admired the passion Jody B and Charlotte S AKA the Empress of Evaporites have demonstrated in their works and I have learned so much from them They have both been exceedingly kind to me and are both forces of nature David C has also been kind to me and his sense of humor is also something I always appreciate I am very glad he is on my committee Eric S and Sharon D have also been exceptional committee members and I am grateful for their comments and guidance Eavesdropping on Eric super passionately talking about isotopes in the isolab always put a smile on my face Fangzhen T is a great teacher and has always been a pleasure to talk to He has shown real interest and enthusiasm in my academic pursuits and he too has a fantastic sense of humor Lab members Mike K Jana M Eric G and Owen L have been a great support system I am certain they are going to make up the next generation of super-scientists if that is what they want Me It might be nice to open up a tavern am I rambling None of my work or any other light-stable isotope work in UW ESS would be possible without Andy Schauer He is the best As an engineer as a scientist and as an occasionally grumpy but still humorous person Sorry I still haven t paid for stuff xv DEDICATION To the poor souls who read this entire thing And to my friends and family who know better I love you all xvi Chapter 1: INTRODUCTION The six most important bioessential elements for life as we know it are carbon hydrogen oxygen nitrogen phosphorus and sulfur Five of these elements are in the top ten most abundant elements in the universe Planetary bodies both inside and outside of the Solar System likely harbor these elements in different forms potentially in varying phases and redox states on their surfaces where life is more likely to emerge and proliferate It makes sense then to use Earth s biochemistry as an analog for other life in the universe This assumes that extraterrestrial life would utilize similar nutrients and energy sources as our own If we believe that these assumptions are our best anchor for the search for signs of life on other planetary bodies then the natural prerequisite to this search is to examine how life and these elements co-evolved through Earth s history Understanding how the abundance and speciation of these elements have changed in Earth s atmosphere and oceans through time provides a mechanistic foundation for exploring biogeochemical evolution on other worlds that may be different from ours This is the astrobiological foundation for the following dissertation that focuses on the co-evolution of life and the nitrogen cycle on early Earth The earliest signs of life on Earth come from equivocal trace fossils and organic matter from 4 1 to 3 8 Ga The abundance and speciation of nitrogen in the atmosphere and ocean during this time is relatively unknown Before the evolution of biological nitrogen fixation1 fluxes of nitrogen from the atmosphere to the ocean were dominated by NOx species produced from atmospheric N2 by lightning and bolide impact plumes Importantly these abiotic fixation fluxes would have been at least an order of magnitude less than modern biological nitrogen fixation 1 The biologically-mediated conversion of N2 gas to bioavailable nitrogen 17 St eken E E Kipp M A Koehler M C Buick R 2016 Biological utilization of this NO3 NO2- could have supported a proportionally smaller ancient biosphere before the evolution of biological nitrogen fixation There is firm isotopic evidence that biological nitrogen fixation by the Mo-nitrogenase evolved before 3 2 Ga St eken E E Buick R Guy B M Koehler M C 2015 Koehler M C Buick R Barley M E in review but ubiquitous alteration of older sedimentary rocks inhibits extending this isotopic record further back in time Still with this finding we know that the biosphere had a first-order control on the nitrogen cycle since the Mesoarchean Until the evolution of denitrification2 the nitrogen cycle would be missing a significant return flux to the atmosphere and so fixed nitrogen could have possibly accumulate in the oceans sediments and crust A growing body of evidence points to substantially lower atmospheric pressure in the Neoarchean mainly due to a lower partial pressure of nitrogen A modelling study suggests that a relatively high biomass burial rate in the Archean could have resulted in a drawdown of atmospheric N2 to less than half of modern levels St eken E E Kipp M A Koehler M C Schwieterman E W Johnson B Buick R 2016 The first strong evidence for denitrification is from the Neoarchean over 250 Myr before the Great Oxidation Event3 GOE Koehler M C Buick R Kipp M A St eken E E Zaloumis J In Press Interestingly the available evidence suggests there was at least a 500 Myr gap between the evolution of nitrogen fixation and denitrification These early signs of denitrification before the GOE are associated with transient whiffs of oxygen in the surface ocean and atmosphere Enhanced oxidative weathering due to increases in atmospheric oxygen may have caused a rebound in atmospheric N2 by shortening the residence time of crustal nitrogen After the 2 The biologically-mediated reduction of oxidized nitrogen species nitrate nitrite to N2 gas The Great Oxidation Event is characterized by the increase of atmospheric oxygen over 10-5 present atmospheric levels It is dated to 2 33 Ga 3 18 GOE nitrogen isotopes indicate a globally significant increase in oceanic nitrate and aerobic nitrogen cycling almost reminiscent of the modern ocean This period is often described as an oxygen overshoot Whether or not the oxygen overshoot was a period when surface oxygen approached modern levels it represents a temporary maximum in ocean oxygenation and nitrate levels Towards the end of the Paleoproterozoic and through the Mesoproterozoic the nitrogen isotope record indicates a decrease in oceanic nitrate particularly in offshore environments Koehler M C St eken E E Kipp M A Buick R 2017 This is coincident with the appearance of eukaryotic microfossils in the geologic record Eukaryotic microfossils primarily rely on nitrate as a nitrogen source because they are unable to fix atmospheric N2 and do not compete well against prokaryotic organisms for NH4 The onshore-offshore gradient in oceanic nitrate correlates with an onshore-offshore decrease in the richness and abundance of eukaryotic microfossils It thus appears that low Mesoproterozoic nitrate levels likely caused by a decrease in surface oxygen levels resulted in the restriction of eukaryotes to nearshore environments potentially delaying their proliferation and rise to ecological dominance until ocean and atmospheric oxygen levels increased in the Neoproterozoic Paleozoic While surface oxygenation was originally thought to have risen sharply in the Neoproterozoic new ocean redox data suggests that deep-sea oxygenation was probably protracted through the Paleozoic Indeed nitrogen isotopes across the Ordovician-Silurian boundary are reminiscent of the Mesoproterozoic suggesting low oceanic nitrate levels and a scarcity of fixed nitrogen particularly in offshore environments Koehler M C St eken E E Prave A R in prep It is possible that full ocean oxygenation and a rise to near-modern nitrate levels did not occur until the rise of land plants in the Devonian Even so ocean anoxic events associated with mass 19 extinctions through the Phanerozoic periodically decreased ocean nitrate to levels similar to those in the Mesoproterozoic As alluded to above the four chapters in this dissertation examine the co-evolution of nitrogen cycling surface oxygen and life at critical times on the ancient Earth The first explores the Mesoarchean redox and nitrogen cycle landscape at relatively high temporal resolution This was apparently before the advent of aerobic nitrogen cycling The second chapter looks for the oldest signs of aerobic nitrogen cycling in the Neoarchean and explores the mode tempo and scope of transient oxygenation events before the GOE The third chapter deals with oceanic nitrate availability in the Mesoproterozoic and how it related to eukaryotic evolution The last chapter focuses on nitrogen cycling carbon cycling and the redox landscape in the Paleozoic specifically across the Ordovician-Silurian boundary This period hosts the first major glaciation of the Phanerozoic and one of the five big mass extinction events By enhancing our understanding of how nitrogen oxygen and life evolved on the early Earth the chapters in this dissertation add to the astrobiological exploration of how biogeochemical cycles may evolve on strange new worlds and how this may affect habitability and the evolution of extraterrestrial life My role in the projects discussed in this dissertation are as follows: Roger Buick and I devised chapter 2 I conducted all analyses Roger Buick and I interpreted the results and wrote the manuscript Samples were provided by Mark Barley Chapter 3 was devised by Roger Buick All nitrogen and carbon analyses were performed by me Selenium analyses were performed by Michael Kipp Eva St eken Jon Zaloumis and myself Roger Buick and I performed the petrographic work Myself Roger Buick Michael Kipp and Eva St eken interpreted the results and wrote the manuscript Samples were provided by Roger Buick and the Agouron Institute Drilling Project Chapter 4 was devised by myself Roger Buick and Eva St eken Analyses were 20 performed by primarily by myself with help from Eva St eken and Michael Kipp All authors contributed to data interpretation and writing the manuscript Samples were provided by Roger Buick and Andy Knoll Chapter 5 was devised by myself and Eva St eken I collected the samples with help from Tony Prave and Eva St eken I carried out all the analyses and largely interpreted the result with help from Eva St eken I wrote the manuscript 21 Chapter 2: NITROGEN ISOTOPE EVIDENCE FOR ANOXIC DEEP MARINE ENVIRONMENTS FROM THE MESOARCHEAN MOSQUITO CREEK FORMATION AUSTRALIA This manuscript has been submitted to the journal Precambrian Research Co-authored by Matthew C Koehler Roger Buick and Mark E Barley 2 1 Abstract Recent evidence suggests that oxygenated environments in the Mesoarchean were limited to the shallowest marine and fluvio-lacustrine settings It is becoming increasingly clear that during the Neoarchean oxidizing conditions spread to the photic zone of deeper shelf and basinal depositional environments Here we present nitrogen and carbon isotope ratios from the Mosquito Creek Formation of the Nullagine Group 2 9 Ga to further explore the Mesoarchean redox landscape The 15N and 13Corg values are invariant and suggest an ecosystem dominated by nitrogen fixers anaerobic nitrogen cycling and CO2 fixation by the Calvin Cycle These data i support the apparent Mesoarchean trends of decreasing oxidant availability and methane cycling from onshore to offshore depositional environments ii provide further evidence that the Mosquito Creek Formation was deposited in a deep marine setting and iii contain 15N values that highlight the persistence of nitrogen fixation by Mo-nitrogenase and the dearth of aerobic nitrogen metabolisms in the Mesoarchean 2 2 Introduction It has been argued based on sulfur isotope evidence that Earth s atmosphere was persistently oxygenated throughout the Archean era 1 Now this is distinctly a minority view 22 with abundant evidence contradicting it from samples collected at high stratigraphic resolution and analyzed for quadruple sulfur isotopes 2 nitrogen isotopes 3 4 selenium isotopes 5 and trace metal abundances and isotopes 6 11 These indicate instead that the lower atmosphere during the Neoarchean was generally anoxic except for a few transient whiffs of oxygen e g 6 However few relevant studies have been done to determine if this pattern extends back into the Mesoarchean The emerging view is that free O2 in the Mesoarchean was limited to localized production in the shallowest marine 12 15 and perhaps fluvial-lacustrine-terrestrial environments 16 17 Apart from a single study 18 of Mo and Cr isotopes which have complex fractionation pathways and thus lesser redox sensitivity distal-marine rocks have not been examined using other redox proxies to further explore the Mesoarchean redox landscape Here we analyze nitrogen and carbon isotope ratios in rocks of the Mosquito Creek Formation from the same ABDP 5 drill-core that was studied to formulate the provocative oxic Archean atmosphere hypothesis 1 Our data are consistent with i anoxic waters in distal-marine settings ii the apparent ecological irrelevance of nitrification-denitrification metabolisms before the start of the Neoarchean and iii a facies-dependent trend of more depleted 13C values in fluvial-lacustrineproximal marine environments compared to more distal marine settings in the Mesoarchean 2 3 Geologic Setting The samples analyzed here came from the Nullagine Group which is located in the southeastern corner of the Pilbara Craton in northwestern Australia Figure 2 1 This unit has been subdivided into the underlying Coondamar Formation and the upper Mosquito Creek Formation with our samples all coming from the latter The samples are predominantly kerogenous shales and siltstones from the finer-grained D and E portions of graded turbidites deposited distally in 23 subaqueous fans formed beneath subaerial fan-deltas Figure 2 2 Figure 2 3 19 20 The tectonic setting of the basin in which these turbidities were deposited is controversial It was initially interpreted as a fore-arc basin facing an open ocean to the south 19 based on facies and provenance analysis Alternatively deposition of the Mosquito Creek basin along a passive continental margin was proposed based on a geochemical discriminant factor analysis 21 also in a setting with an open connection to the ocean but in this case to the west By contrast Nijman and colleagues 20 advocated deposition in an intracontinental intermontane basin that may or may not have been marine based on the fining-upward style of sedimentation and overall underfilling of the basin Our data may help to partially resolve this controversy Our samples 47 in total were obtained at roughly meter-scale resolution from the drill-core ABDP 5 which was collared at Lat 21 41 53 7 S Long 120 37 14 5 E in the Eastern Creek goldmining area 55 km to the northeast of the town of Nullagine 1 These rocks have been metamorphosed from sub- to lower-greenschist facies 1 22 and have been strongly deformed multiple times resulting in tight folding thrust faulting and a variably developed schistosity 21 Their depositional age is constrained between a maximum detrital zircon U Pb date of 2971 15 Ma from near the base of the 5 km thick Mosquito Creek Formation 21 and a minimum Pb-Pb model age from galena in cross-cutting lode gold mineralization of 2905 9 Ma Thorpe et al 1992 24 Figure 2 1 Location map modeled after 21 showing A a simplified map of the Pilbara Craton B a simplified map of the Mosquito Creek Basin highlighted in A 25 Figure 2 2 ABDP 5 stratigraphy from 50m to 150m depth The section analyzed in this study is indicated by the black and white striped bar Modified from 1 26 Figure 2 3 Field photo of rippled sandy C silty D and shaley E turbidite facies of the Mosquito Creek Formation modified from 21 Red arrow indicates stratigraphic up The samples studied here predominately came from D and E facies The coin for scale is 25 mm in diameter 27 2 4 Methods Samples for bulk nitrogen and carbon analyses were prepared using standard methods that have been utilized successfully in previous studies 23 24 All glassware and the silica sand used for cleaning were combusted in a muffle furnace at 500oC overnight to remove organic contaminants All chemicals used were reagent-grade Rock samples from ABDP 5 were cut into 3 x 3 x 2 cm chips and all surfaces were shaved with a diamond-tipped rock saw to minimize modern contamination Chips were then manually crushed between a large mortar and steel plate each wrapped in clean aluminum foil to reduce contamination The aluminum foil was changed between each sample The resulting smaller 1 cm3 chips were then transferred to small clean glass beakers Each sample was then sonicated with ethanol for 2 minutes decanted and then with 2N HCl for 20 seconds to further remove modern contaminants After decanting the HCl samples were rinsed 3 times with 18M milli-Q water and left covered in a 60 C ventilated drying oven until completely dry Samples were then pulverized using an aluminum oxide puck mill that was cleaned with ethanol 18M milli-Q water and silica-sand between each sample The resulting sample powders were stored in glass scintillation vials Powder aliquots of about 0 5 grams were weighed into glass centrifuge tubes and were acidified in three iterations with 6N HCl For each iteration the acid was left to digest the sample covered in a 60oC ventilated oven overnight 24 hours After decanting the acid from the last iteration the samples were washed three times with 18M milli-Q water to remove the HCl The water from the last rinse was then decanted and covered samples were left to dry in a 60oC ventilated oven for two days After drying samples were weighed again to determine sample carbonate concentrations gravimetrically 28 Kerogen was extracted from a subset of samples again using standard methods 23 24 adapted from 25 3-4 grams of powder were weighed in 250ml Teflon bottles and acidified with 100ml of 50% v v hydrofluoric acid HF in a warm shaking water bath 55oC overnight to dissolve the silicate matrix Because fluoride precipitates can form during this process the samples were then reacted with 100ml of saturated boric acid in 50% v v HF to remove them Samples were then washed thoroughly in three iterations with 200ml of 18M milli-Q water Kerogen was then transferred from the Teflon bottles to glass scintillation vials with 10ml of 18M milliQ water and subsequently frozen Samples were then freeze-dried for two days Total organic carbon TOC 13Corg total nitrogen TN 15Nbulk and 15Nkerogen values were measured using a Costech ECS 4010 Elemental Analyzer coupled to a Thermo Finnigan MAT253 continuous-flow isotope-ratio mass spectrometer Samples were bracketed by three in-house standards calibrated to reference materials USGS40 and USGS41 26 to calibrate sample analyses and monitor accuracy Each run also contained two analyses of a fourth rock-standard UW McRae to monitor long-term precision Average analytical accuracy and precision for 15N measurements were 0 15 0 11 and 0 15 respectively Average analytical accuracy and precision for 13C measurements were 0 1 0 06 and 0 08 respectively Nitrogen and carbon isotope values are presented in standard delta notation relative to air and VPDB respectively 29 2 5 Results The most notable aspect of the Mosquito Creek nitrogen and carbon data is how invariant it is through the section Figure 2 4 15Nbulk values average -1 8 0 29 1 where the standard deviation among the samples is less than the average standard deviation of sample replicates 0 33 1 13Corg values average -32 3 0 41 1 with close to half of the variability 0 16 coming from the four sandstone samples This is greater than the average standard deviation among sample replicates 0 09 1 Total nitrogen abundances of 0 009 0 002 % total organic carbon levels of 0 34 0 1 % total inorganic carbon values of 24 3 6 % and Corg Nbulk atomic ratios of 42 8 9 all have standard deviations 1 that are below 30% of the average 15Nkerogen values are slightly more depleted and variable than 15Nbulk values of the same sample set averaging -2 5 0 64 1 The average standard deviation for sample replicates is 0 22 1 In contrast 13Ckerogen values are almost identical to the corresponding bulk-rock 13Corg ratios Figure 2 4 Corg Nkerogen atomic ratios average 360 46 5 and the fraction of total nitrogen in a bulk sample that is bound into kerogen is 0 11 0 016 30 Figure 2 4 Simplified stratigraphic section with 13Corg 15Nbulk 15Nkerogen TOC% Grey circles are from bulk measurements and red squares represent kerogen measurements 31 2 6 2 6 1 Discussion Isotopic fidelity Because the metamorphic grade of the Mosquito Creek Formation is no higher than lower greenschist facies post-depositional thermal alteration of the primary nitrogen and carbon isotopic signal has likely been less than 2 28 30 Indications that these isotopic systems have been significantly thermally altered such as a positive correlation between 15N values and C N ratios or a large disparity between 15Nkerogen and 15Nbulk values are not present in our dataset Figure 2 4 Figure 2 5 Post-depositional diagenetic alteration can lead to an increase of the primary 15N signal by up to 4 under oxic conditions through preferential deamination of 14N from organic matter followed by nitrification of this isotopically light NH4 and loss from the system 31 34 However the absence of bioturbation in the Mesoarchean would have likely left sediments anoxic regardless of overlying water redox conditions so this diagenetic process would have been inoperative Diagenetic alteration under anoxic conditions can potentially lead to a decrease in the primary 15N value of organic matter if there is significant in situ growth of 15N-depleted nitrogen-fixing bacteria among sedimentary organic matter with positive 15N values or if there is preferential deamination of 15 N-rich compounds such as proteins 32 35 36 However the consistently low 15N values throughout the section suggest that in situ growth of N-fixing bacteria or deamination of 15N-rich compounds during diagenesis was unlikely to have affected the primary isotopic composition by more than a couple of per mil Thus the original 15N of the sedimentary organic matter was probably close to what it is now 32 Figure 2 5 Cross-plots of carbon and nitrogen isotope abundance measurements A 15Nbulk vs 13Corg B 15Nkerogen vs 13Corg C 15Nkerogen vs 15Nbulk D 13Corg vs TOC E 15Nbulk vs TN F 15Nbulk vs C Nbulk G 15Nkerogen vs C Nkerogen H 13Corg vs TOC for sandstone samples TOC Total organic carbon TN Total nitrogen Note all isotope measurements are expressed in parts per thousand and C N ratios are atomic ratios 33 2 6 2 Redox and paleoecology The nitrogen isotopic values of almost all samples both bulk and kerogen measurements are between -3 to -1 suggesting an ecosystem dominated by nitrogen fixation using the molybdenum nitrogenase 37 with no signs of any aerobic nitrogen cycling utilizing NO3- Partial nitrification of ammonium followed by quantitative denitrification or quantitative nitrification followed by partial denitrification would preferentially remove 14 N from the system as N2 and increase the residual 15N values preserved in the rock record 28 38 These scenarios require free O2 and are inconsistent with our data Redox data from the literature suggests that the Mesoarchean was an era where only the shallowest marine and fluvial-lacustrine environments had appreciable oxidants 12 16 for metabolisms such as non-quantitative sulfate reduction 12 and perhaps anaerobic oxidation of methane 17 Together with studies of hydrothermally influenced banded iron formations 39 and distal marine shale mudstones 40 41 our data from the deep-water Mosquito Creek Fm further support this paleogeographic redox trend showing that oxidants were limited or absent in offshore settings Such a trend could have resulted from i lower concentrations of oxygen sinks such as organic matter Fe II and reduced sulfur species onshore e g 12 and or ii greater local oxygen production due to greater nutrient availability e g 15 in shallow marine and terrestrial environments The Mosquito Creek carbon isotope data average -32 can be explained by an ecosystem dominated by CO2 fixation using the Calvin Cycle but do not preclude a subordinate role for chemosynthetic organisms Lacking distinct evidence for dissolved O2 and oxygenic photosynthesis it is more likely that ecological dominance of anoxygenic photosynthesizers are responsible for the measured 13Corg values The 13Corg of the four sandstone samples are enriched compared to the shale samples by an average of 1 This enrichment is coupled to lower TOC 34 concentrations Within the sandstone samples there is a negative correlation between TOC and 13Corg which could be explained by the progressive dilution of heavier carbon through the addition of lighter carbon This could arise when enriched detrital carbon was progressively diluted with the more depleted carbon source responsible for the 13Corg of the shale samples It could also reflect differences in diagenesis associated with the lithologic change from shale to sandstone with light carbon being preferentially removed by migrating fluid phases in sands with higher porosity and permeability Regardless this small difference of 1 does not affect our overall interpretation of the carbon data Interestingly both the Mesoarchean 17 and Neoarchean 42 are characterized by environmental gradients in 13Corg values with fluvial-lacustrine settings being the most depleted consistent with a larger role of methanogens methanotrophs Though both the Mesoarchean and Neoarchean are characterized by this facies-dependent trend in local carbon-fixing ecologies the carbon isotope record suggests more vigorous methanogenesis methanotrophy in all Neoarchean environments even in offshore settings So while the golden age of methanogenesis methanotrophy is attributed to the Neoarchean 42 when evidence of oxygenation extends to deeper marine environments our data support the idea that methane cycling in the Mesoarchean seems to have been restricted to lakes rivers and shallow marine settings 17 Non-marine carbon isotope measurements from the Mesoarchean range from 28 to 47 while marine measurements range from 27 to 33 These ranges suggest a more diverse carbon fixation ecology that included methanogens in non-marine environments The tight grouping of marine 13C values from this and previous studies indicate more homogenous and less diverse carbon fixation strategies in offshore environments 35 The relationship between oxidant availability methanogenesis and methanotrophy is complex It is likely that increasingly depleted 13C values correspond to greater oxidant availability and a coincident rising role of methanotrophs from the Mesoarchean to the Neoarchean as combined carbon isotope and redox data would suggest This scenario would require greater rates of biogenic methane production than oxidant consumption such that anaerobic and or aerobic methane oxidation rates would increase with oxidant supply Figure 2 6 scenario 1 Importantly this would not require any change in methane production across the Meso Neoarchean boundary However once the oxidant flux overwhelms the biogenic methane flux and their respiratory consumption starts competing with methanogenesis for reduced substrates the inverse relationship would be true In such a case progressively lighter 13C values would result from a decreasing oxidant flux less competition for organic carbon and H2 between methanogens and organisms that use anaerobic respiration and so perhaps the proliferation of methanogens Figure 2 6 scenario 2 Therefore a more quantitative understanding of the relationship between oxidants and methane during the Precambrian is critical when interpreting local carbon isotope records 42 44 2 6 3 Marine or non-marine These data help to resolve the controversy over the depositional setting of the Mosquito Creek basin Though they cannot inform about its tectonic style they can illuminate whether it was marine 19 21 or potentially non-marine as in the intermontane basin model 20 Comparison of the 15N and 13C values of the Mosquito Creek rocks with near-contemporary units of known marine or non-marine origin Figure 2 7 allows environmental discrimination In carbon-nitrogen isotope space the Mosquito Creek data most closely resemble the marine Witwatersrand 2 9 Ga and Soanesville 3 19 Ga samples Figure 2 7A and Figure 0 2 36 However they differ markedly from most Mesoarchean and Neoarchean non-marine samples and Neoarchean marine samples Figure 2 7B Error Reference source not found Figure 0 2 and Figure 0 3 with only the Lalla Rookh 3 0 Ga data plotting anywhere nearby However the Lalla 37 Figure 2 6 Schematic of the relationship between 13Corg values and oxidant availability Scenarios 1 and 2 are described in section 2 6 2 38 Figure 2 7 Cross-plots of 13Corg vs 15Nbulk for the A Mesoarchean and B Neoarchean Grey circles are marine data orange X s are non-marine data blue circles are from this study Horizontal dashed line separates 15Nbulk values consistent with nitrogen fixation by the Mo-nitrogenase below from 15Nbulk values consistent with some degree of aerobic nitrogen cycling above Shaded colors represent the following: Green - 13Corg values consistent with carbon fixation using the Calvin Cycle Blue - 13Corg values consistent with carbon fixation using the Wood-Ljungdahl pathway Purple - 13Corg values consistent with methanotrophic carbon cycling Only least-altered data from the Archean were included and were taken from a compilation in 38 with added data from 17 45 While there are more 15N values from the Archean not shown here the fidelity of those measurements based on potential alteration and or large analytical uncertainly see supplementary materials of 41 have been excluded 39 Rookh samples have distinctly lighter 13Corg values with a greater spread than the tightly clustered Mosquito Creek data suggesting a more variable environment and microbiota than the homogeneous setting implied by the low Mosquito Creek variance Thus this strongly supports the idea that the Mosquito Creek basin was marine 19 21 rather than deposited in an intracontinental intermontane basin 20 2 6 4 Implications for microbial evolution Data from the Archean suggest that nitrogen fixation by the molybdenum nitrogenase had evolved by 3 2 Ga 41 was widespread by 2 9 Ga 41 this study and that nitrification and denitrification did not become significant metabolisms in the oceans until 2 65 Ga 4 Given the evidence suggesting some in situ oxygen production in the shallowest marine and fluvial-lacustrine environments as far back as 3 Ga 12 17 it is possible that the nitrification and denitrification evolved before evidence for their ecological significance first appeared in the Neoarchean However high-resolution exploration of older sections have yielded no strong evidence for aerobic nitrogen cycling 41 this study suggesting that nitrification and denitrification only became ecologically relevant with the later expansion of oxygenated surface waters 2 7 Acknowledgements This study was funded by NASA grant NNX16AI37G and NSF FESD grant 338810 to R B and U W ESS Departmental Awards to M C K We thank the U W Isolab and Andy Schauer for technical support and we are particularly indebted to Sarah Barley for facilitating this research and for reviewing the manuscript 40 Chapter 3: TRANSIENT SURFACE OCEAN OXYGENATION RECORDED IN THE 2 66 GA JEERINAH FORMATION AUSTRALIA This manuscript has been submitted to the journal Proceedings of the National Academy of Sciences Co-authored by Matthew C Koehler Roger Buick Michael A Kipp Eva E St eken and Jonathan Zaloumis 3 1 Abstract Many paleoredox proxies indicate low-level and dynamic incipient oxygenation of Earth s surface environments during the Neoarchean 2 8-2 5 Ga prior to the Great Oxidation Event GOE at 2 4 Ga The mode tempo and scale of these redox changes are poorly understood because data from various locations and ages suggest both protracted and transient oxygenation Here we present bulk-rock and kerogen-bound nitrogen isotope ratios as well as bulk-rock selenium abundances and isotope ratios from drill-cores sampled at high stratigraphic resolution through the Jeerinah Formation 2 66 Ga Fortescue Group Western Australia to test for changes in the redox state of the surface environment We find that both shallow and deep depositional facies in the Jeerinah Fm display episodes of positive primary 15N values ranging from 4 to 6 recording aerobic nitrogen cycling that requires free O2 in the upper water column Moderate selenium enrichments up to 5 4 ppm in the nearer-shore core may indicate coincident oxidative weathering of sulfide minerals on land though not to the extent seen in the younger Mt McRae Shale that records a well-documented whiff of atmospheric oxygen at 2 5 Ga Unlike the Mt McRae Shale Jeerinah selenium isotopes do not show a significant excursion concurrent with the positive 15N values Our data are thus most parsimoniously interpreted as evidence for transient 41 surface ocean oxygenation lasting less than 50 Myr extending over hundreds of kilometers and occuring well before the GOE The nitrogen isotope data clearly record nitrification and denitrification providing the oldest firm evidence for these microbial metabolisms 3 2 Introduction Despite widespread agreement about general trends 46 our understanding of the mode and tempo of Earth s oxygenation still suffers from considerable uncertainties Among the more salient features of this record is a proposed whiff of oxygen in surface environments at 2 5 Ga well before the onset of the Great Oxidation Event GOE at 2 3-2 4 Ga 47 48 - which has been identified using combined trace metal and nitrogen isotope datasets collected at high stratigraphic resolution 3 5 6 Exploration of other Meso- and Neoarchean records has revealed dynamic evolution of surface ocean redox conditions and oxidative continental weathering leading up to the permanent establishment of oxidizing conditions at Earth s surface 8 10 12 16 44 49 55 To further elucidate the dynamic redox landscape in the Neoarchean we conducted a coupled survey of nitrogen and selenium isotopes and abundances at high stratigraphic resolution in two drill cores spanning a near shore-offshore transect across the 2 66 Ga Jeerinah Formation in Western Australia These complementary proxies allow us to assess the relative magnitude and spatial extent of redox fluctuations Nitrogen is essential for life as we know it and its biogeochemical behavior is highly redoxsensitive Since at least 3 2 Ga it is evident that the biosphere has exerted the primary control on Earth s surface nitrogen fluxes regulating nitrogen availability and speciation in the ocean and atmosphere 28 41 Metabolisms that drive nitrogen speciation in the ocean transition of nitrogen among different redox states can use a variety of electron donors for reduction such as organic 42 matter Fe2 and sulfide but for oxidation are primarily reliant on dissolved O2 as an electron acceptor due to the high Eh of nitrate 38 Ammonium oxidation to nitrate can occur down to low nM oxygen concentrations 56 57 which is a fraction of a percent of modern well-oxygenated ocean waters typically of 300 M dissolved O2 Thus a significant pool of nitrate in an ancient ocean would suggest that at least minor amounts of oxygen were freely available to allow for nitrification: the biologically mediated conversion of ammonium to nitrite nitrate Modern microbial metabolisms are known to have a diverse range of effects on the relative abundances of the nitrogen isotopes 14N and 15N summarized in Table 1 of ref 38 58 59 For example N2 fixation using Mo-nitrogenase causes an insignificant isotopic fractionation during the conversion of N2 to organic-bound nitrogen but biological denitrification of NO3- to N2 strongly prefers 14 N producing isotopically light N2 gas and an isotopically heavy residual dissolved NO3- pool Hence nitrogen isotopes in ancient sedimentary rocks have allowed paleoecological interpretations of nitrogen metabolisms in ancient ecosystems These interpretations are only semi-quantitative due to overlapping fractionation factors for different metabolic pathways and because of the potential for isotopic resetting by post-depositional alteration Still the presence of nitrate in concentrations high enough for significant bio-assimilation can be distinguished isotopically from anaerobic settings that are dominated by nitrogen fixation followed by proximal re-assimilation of liberated ammonium 3 4 23 24 e g 28 38 41 We also studied the behavior of selenium Se a redox-sensitive element whose oxyanions are stable at similar redox potentials to nitrogen oxyanions to determine whether changes in marine nitrogen cycling were concurrent with changes in weathering dynamics on land Unlike nitrate which is primarily produced during biomass remineralization and nitrification in the water column Se oxyanions are mainly sourced from oxidative weathering on land 60 43 Additionally if Se fluxes are sufficiently large and oxygenated waters sufficiently widespread Se isotopes can become fractionated during incomplete reduction of Se oxyanions 61 The nitrogen and selenium proxies combined can thus inform us about the loci of oxygenation 3 3 Geologic Setting As part of the Agouron Institute Drilling Project in 2012 two ultra-clean diamond drill cores AIDP-2 and AIDP-3 were sampled across the boundary between the Hamersley and Fortescue Groups in Western Australia These holes were collared at 21 16 51 S 120 50 02 E AIDP-2 and 21 46 32 S 117 34 11 E AIDP-3 across a basinal depth gradient with AIDP-2 intersecting shallower near-shore facies and AIDP-3 deeper offshore facies AIDP-2 consists of 158 meters of the Carawine Dolomite that for this study was sampled at low stratigraphic resolution SI Appendix Figure 0 4 and 128 meters of predominantly black shales from the Jeerinah Formation that were sampled at one-meter resolution AIDP-3 is 138 meters deep capped by the basal Marra Mamba Iron Formation but with samples taken exclusively from the underlying Jeerinah Formation at one-meter resolution The shallower depositional environment of AIDP-2 is indicated by the stromatolitic carbonates of Carawine Dolomite 62 and by the shoreface setting of the basal Woodiana Member of the Jeerinah Formation which consists of occasionally stromatolitic orthoquartzites 63 In contrast AIDP-3 transects only a basinal banded iron formation and sulfidic kerogenous shales The two holes can be stratigraphically correlated using a horizon of meteorite impact spherules 64 which occurs in a 20 meter thick megabreccia towards the bottom of the Carawine Dolomite in AIDP-2 and in a 1 centimeter thick graded spherule layer in the upper Jeerinah Formation in AIDP-3 SI Appendix Figure 0 5 Rocks from both AIDP-2 and AIDP-3 are generally thought to be within the prehnite-pumpellyite metamorphic 44 facies 65 but an alternative upper greenschist facies interpretation exists 66 However this latter interpretation is inconsistent with the pelitic mineralogy of the AIDP shales as key index minerals of upper greenschist conditions e g biotite chloritoid are absent Also it depends upon a chlorite geothermometer 67 that i unjustifiably assumes all iron in the chlorite is Fe2 which can lead to temperature overestimations 68 and ii yields infeasibly broad temperature ranges 140oC for samples in close stratigraphic proximity Hence we favor the prehnite-pumpellyite metamorphic facies assignment addressed further in the SI Appendix 3 4 Results Se abundances are tightly scattered around an average of 1 2 1 1 ppm throughout both cores but a moderate excursion up to 5 4 ppm occurs around 363 m in AIDP-2 After normalizing the data to Al TOC and total sulfur TS the Se excursion in AIDP-2 still appears statistically significant Figure 3 1B C D and SI Appendix Figure 0 6 and Figure 0 7 Multiple regression confirms this inference showing that less than half of the observed Se enrichment can be explained by lithological variability R2 0 49 Figure 3 1B C D 82 78Se shows no systematic trend across either of the two cores average 0 3 0 3 In contrast 15N displays a large range from -2 to 14 with systematic shifts detected by our linear regression model This identified three distinct stages I to III from oldest to youngest in each core Table 3 1 The only exception to the model boundary placement is the boundary between stage I and II in AIDP-3 where the boundary position was handpicked based on marked changes in TOC % TN % and C N ratios atomic Importantly this boundary still falls within the range of stratigraphic positions where the model predicts a significant change in 15Nbulk values 45 Table 3 1 Stratigraphic stages of AIDP-2 and AIDP-3 with associated nitrogen and carbon measurements Stratigraphic stage boundaries were determined using the linear regression model described above C N ratios are given as atomic ratios AIDP-2 Stage III II I Strat meters 344 18 - 278 75 377 89 - 344 99 428 16 - 380 37 15Nbulk 0 38 0 61 4 33 0 53 2 76 0 71 13Corg -43 2 2 5 -43 0 1 9 -42 0 1 5 TN % 0 08 0 03 0 12 0 02 0 04 0 01 TOC % 4 8 2 3 7 7 2 4 2 8 0 8 C N 81 39 74 17 95 24 TIC % 11 4 4 4 9 5 2 5 16 3 4 9 TN % 0 06 0 02 0 06 0 01 0 02 0 02 TOC % 5 8 2 4 4 7 0 7 2 4 2 1 C N 109 32 85 14 156 63 TIC % 12 1 8 6 15 0 3 0 14 7 2 9 AIDP-3 Stage III II I Strat meters 127 83 - 66 03 155 75 - 129 65 204 37 - 156 89 15Nbulk 2 39 0 86 6 07 0 36 10 2 2 14 13Corg -41 7 1 5 -41 7 1 0 -44 2 0 9 46 Figure 3 1 Crossplots of A 15Nkerogen values and C N ratios B Se enrichment factor EF and Se ppm C Se ppm and TOC % and D Se ppm and total sulfur TS % For A Note the linear covariance in AIDP-3 stage I compared to the rest of the core Samples from AIDP-2 come from all stages For B the procedure for calculating EF is described in the SI Appendix Note the particularly strong enrichments in AIDP-2 Stage II Samples from AIDP3 come from all stages For C and D Note the weaker correlations in AIDP-2 compared to AIDP-3 The variation and so the increase in selenium abundance in AIDP-2 cannot be explained through variations in TOC and TS Conversely much of the variation in selenium abundance in AIDP-3 can be explained by variations in TOC and TS This suggests the increase in Se during stage II of AIDP-2 is likely caused by an increased flux of Se to the basin 47 3 4 1 AIDP-2 Stage I in AIDP-2 is characterized by the most variable 15Nbulk values in the core averaging around 2 8 In this stage TOC % and TN % are at their lowest From stage I to stage II 15Nbulk suddenly increase by an average of 1 5 to the highest values recorded in the core 5 43 and become less variable This rise in 15Nbulk values is associated with increases in both TOC % and TN % by about a factor of 3 From stage II to stage III 15Nbulk values decrease by 4 on average to a minimum of -0 71 TOC % and TN % also decrease across this transition but not to levels as low as in stage I 15Nkerogen values are systematically lower than 15Nbulk measurements throughout the core by an average of 1 6 0 9 and so show the same general trend as 15Nbulk values across all three stages Figure 3 2 3 4 2 AIDP-3 Stage I in AIDP-3 has the most positive and variable 15Nbulk values and C N atomic ratios in either core averaging 11 0 and 174 respectively Notably in this same stage TOC % and TN % averages are the lowest in either core Across the boundary between stages I and II 15Nbulk values decrease by 5 on average Figure 3 3 TOC % and TN % values both increase by over a factor of 3 and C N atomic ratios decrease to an average similar to all stages apart from stage I of AIDP-3 The transition from stage II to stage III is defined by another decrease in 15Nbulk of about 4 on average similar to the stage II-III transition in AIDP-2 TOC % TN % and C N ratios atomic do not change significantly from stage II to stage III Like AIDP-2 in AIDP-3 15Nkerogen values are systematically lower than 15Nbulk Whereas the average difference between the two measurements is consistent throughout AIDP-2 in AIDP3 it varies in a stepwise fashion throughout the core the smallest differences are in stage III and 48 Figure 3 2 Chemostratigraphy of AIDP-2 showing 15Nbulk grey dots 15Nkerogen red dots C N ratios 82 78Se and Se concentrations Stages are delineated by horizontal dotted lines Shaded area for 82 78Se represents crustal values Down-core depth is given in meters 49 Figure 3 3 Chemostratigraphy of AIDP-3 showing 15Nbulk grey dots 15Nkerogen red dots C N ratios 82 78Se and Se concentrations Stages are delineated by horizontal dotted lines Lithological symbols follow the Figure 3 2 legend Note the different 15N and C N scales between AIDP-2 and AIDP-3 required to accommodate Stage I of AIDP-3 Shaded area for 82 78Se represents crustal values Down-core depth is given in meters 50 the largest are in stage I A noteworthy relationship in stage I is that 15Nkerogen values correlate strongly with atomic C Nkerogen and C Nbulk ratios Figure 3 1A This is not the case in stages II and III and all of AIDP-2 3 5 3 5 1 Discussion Proxy Alteration Post-depositional processes such as diagenesis and metamorphism have the potential to alter the primary nitrogen and Se isotopic signals preserved in sedimentary rocks For example Se isotope fractionation could potentially occur during diagenetic oxyanion reduction However this would require the presence of a stable Se oxyanion reservoir in the overlying water column which would in turn imply oxic conditions Such diagenetic Se reactions would therefore strengthen environmental redox signals During metamorphism Se is fairly immobile 69 and so its isotopic ratios are unlikely to be altered Nitrogen isotopes on the other hand can undergo more significant alteration during diagenesis and metamorphism 29 The 15N of sinking and sediment biomass can increase during early diagenesis under oxic conditions through the preferential remineralization oxidation and loss of 14 N common in modern marine settings but is less susceptible to alteration under the mainly anoxic bottom waters sediments 31 33 34 characteristic of the Neoarchean It is possible that for samples with initially positive 15Nbulk values early diagenesis under anoxic conditions can decrease 15Nbulk values by the addition of isotopically light 15N of -2 to 1 biomass via the in-situ growth of nitrogen-fixing bacteria samples that have a primary 15Nbulk signal around 0 would not be significantly affected by this mechanism 32 35 36 Primary 15N values may also decrease by the preferential deamination of 15N-enriched organic compounds such as proteins 32 Even if more relevant in the Neoarchean 51 neither of these mechanisms can fully account for the shifts in 15Nbulk values of up to 4 between stages in both cores nor would they affect our interpretations regarding episodes of enriched 15N values Anoxic diagenesis may lead to small isotopic differences between the kerogen and silicate nitrogen fractions and even smaller isotopic changes in bulk rock measurements but metamorphism metasomatism can lead to greater isotopic disparity and changes that are not always predictable e g 70 71 Below greenschist facies such thermal effects can lead to changes in the primary nitrogen isotopic signal of bulk rocks by 1-2 29 As previously mentioned the Carawine Dolomite and Jeerinah Formation were likely metamorphosed to prehnite-pumpellyite grade 65 and so may have been subject to these small changes to the primary 15N values As metamorphic grade and metasomatic alteration increases so does the difference in 15N between bulk rock and kerogen fractions which may reach up to 13 72 To test if metamorphism metasomatism has affected our samples we considered the relationship between 15N values and C N ratios as well as the difference in 15N values between the two nitrogen fractions kerogen and silicate If thermal processes have significantly affected our samples increasing 15N values should correspond to increasing C N ratios as nitrogen is more mobile than carbon and 14N is more mobile than 15N under such conditions The only samples that show this relationship come from stage 1 in AIDP-3 Figure 3 1A This is also the stage where the differences in 15N values between the two nitrogen fractions are the greatest and most variable Because of this stage 1 of AIDP-3 evidently displays a secondary isotopic signal due to thermal alteration This most likely had a metasomatic component as the kerogen samples are not persistently more altered higher C N and 15N with stratigraphic depth but show local variability that could have resulted from focused fluid-rock interactions This is consistent with previous 52 observations that underlying basalt units locally experienced metasomatic alteration 73 Indeed the basalt underlying the Jeerinah Fm in AIDP-3 has metamorphic metasomatic textures including zones of K-feldspar spherulites veins and amygdales filled with chlorite calcite silica and patches of sericitization all indicating multiple periods of metasomatic alteration SI Appendix Figure 0 8a-h Because all samples from AIDP-2 and from stages II and III of AIDP-3 lack petrographic and geochemical signs of significant thermal alteration the 15N values from these stages are interpreted to be close to primary compositions 3 5 2 Interpretation of nitrogen isotopic data 15N values between -2 and 1 as seen in AIDP-2 Stage III are generally interpreted to reflect marine nitrogen cycling dominated by nitrogen fixation using Mo-nitrogenase 37 Elevated 15N values 1 as in all other stages can be interpreted in one of three ways e g 23 24 SI Appendix Figure 0 9 : i Partial bio-assimilation from a dissolved NH4 pool that would preferentially remove 14 N locally creating isotopically light biomass and an isotopically heavy residual NH4 pool If this heavy residual pool were transported to another location in the basin it could be fully assimilated and biomass there would record isotopically heavy 15N values 74 However because AIDP-2 shallow and AIDP-3 deep represent a broad cross-basinal transect and no corresponding isotopically light values 2 see 24 28 38 for detailed discussion i It has been hypothesized that in an anaerobic nitrogen cycle partial assimilation of an NH4 pool by organisms can leave the residual NH4 pool enriched in 15N as life preferentially assimilates 14N into biomass 74 If this enriched pool of NH4 is subsequently transported to a different location and assimilated by organisms two isotopic facies result: the first location preserves light 15Nbulk and the second preserves heavy 15Nbulk ii Partial nitrification of NH4 can create a nitrate pool that is depleted in 15N and a residual NH4 pool that is enriched in 15N because nitrification preferentially selects for the lighter 87 isotope Assimilation of the NH4 pool will result in heavy 15Nbulk so long as the light nitrate is removed from the system either through subsequent complete denitrification or relocation Partial nitrification is rare in modern water columns occurring where seasonally fluctuating oxygen concentrations occur in transiently stratified waters 177 178 However it is possible that low oxygen concentrations in the Mesoproterozoic enabled more widespread partial nitrification coupled with the removal of the light nitrate pool by complete denitrification Subsequent uptake of the leftover heavy NH4 and further remineralization of organic matter to NH4 could result in a range of positive 15Nbulk values 24 177 iii Partial denitrification of a nitrate pool in the water column will leave the residual nitrate enriched in 15 N as the biologically governed steps of denitrification like nitrification preferentially use 14N over 15N So N2 g the most common end-product of denitrification would be isotopically depleted in 15N and organisms could assimilate the remaining heavy nitrate pool This is the mechanism producing heavy 15Nbulk in the modern ocean e g 76 179 Mechanisms i and ii should result in two distinct isotopic reservoirs one that is relatively depleted below -2 and one that is relatively enriched above 1 There are no samples or sample sets from this study that are lighter than the 15Nbulk values that would result from N2 fixation by Mo-nitrogenase -2 to 1 37 We cannot rule out that a much lighter facies exists somewhere within or adjacent to the Bangemall and Roper basins but was not sampled Nevertheless explanations i and ii are unlikely for several reasons First the Black Sea our best modern analog for the Mesoproterozoic ocean has a large NH4 reservoir in the anoxic bottom water but underlying sediments do not record evidence of partial NH4 88 assimilation values are close to 0 reflecting nitrogen limitation and N2 fixation in the photic zone 122 The same is true for the modern anoxic Cariaco basin 180 This argues against an isotopically light nitrogen reservoir resulting from partial NH4 assimilation in the similarly chemically stratified Mesoproterozoic ocean option i Hydrodynamically the Black Sea and Cariaco basin are probably more stagnant than open marine settings like the Roper and Bangemall basins and so it is conceivable that these Mesoproterozoic settings experienced a relatively higher upwelling flux of NH4 from deeper waters into the photic zone In theory this scenario could favor non-quantitative NH4 assimilation However in the modern ocean NO3- assimilation usually goes to completion with minimal net isotopic fractionations because nitrogen is the proximally limiting nutrient e g 34 179 181 There is no a priori reason to expect that this would have been different if the nitrogen compound was NH4 instead of NO3- This is especially true if low Mo levels reduced N2 fixation rates in the Mesoproterozoic ocean and exacerbated nitrogen limitation 98 Furthermore upwelling would have been most pronounced along the continental margin rather than far offshore and so the hypothetical isotopically light reservoir should be preserved in our sample set which is not the case Regarding option ii in the Black Sea nitrification quickly goes to completion under suboxic conditions at the chemocline 102 The same is true in bacterial cultures and oxygen minimum zones along open marine margins 182 184 suggesting that partial nitrification may also have been rare in the past as long as surface waters were at least mildly oxidizing Indeed in modern environments partial nitrification requires transient seasonal changes in the environment such as fluctuating sea ice cover and is not known to operate over long geologic timescales This leaves option iii partial denitrification at the chemocline cf 124 or in sediment pore waters 123 as the most plausible mechanism responsible for heavier 15Nbulk values in the shallower facies of the Bangemall and Roper basins We thus 89 interpret the large variations of 15Nbulk in the Bangemall and Roper basins as reflecting differential mixing between components derived from N2 fixation alone responsible for the most depleted values between -2 to 1 and from an aerobic nitrogen cycle coupled with varying degrees of partial denitrification of a nitrate pool followed by nitrate assimilation yielding 15Nbulk values above 1 As some peritidal samples have anomalously light 15Nbulk values it may be that N2 fixation temporarily dominated in microbial mats that were transiently cut off from the marine nitrate pool at low tide which would be consistent with microfossil evidence from the Bangemall basin peritidal facies where Palaeopleurocapsa morphologically similar to a modern genus of nitrogen-fixing cyanobacteria Section 4 4 1 is found 94 All shallow subtidal samples have 15Nbulk values above 1 barring one outlier suggesting that some nitrate was always available in the water column and that there was a permanent chemocline where partial denitrification was occurring Deep subtidal samples have lower 15Nbulk values than shallow Table 4 3 and contain samples within the -2 to 1 range Figure 4 5 Figure 4 6 Almost all deep-water values for both basins are lighter than 3 i e lighter than most modern and recent marine sediments 4 e g 185 187 This suggests that offshore sites had less nitrate available compared to shallower facies and the modern deep ocean Thus our data is consistent with at least some nitrate in all depositional environments in both the Bangemall and Roper basins but relatively more in near-shore facies than offshore When we discuss nitrate availability we refer to availability in the zones of highest biological production as this zone will result in the dominating isotopic signal preserved in sediments This is probably the photic zone and it is unlikely that nitrate could have existed much deeper in the water column if bottom waters during the Mesoproterozoic were predominantly anoxic 90 4 7 4 15Nbulk variability in the Bangemall and Roper basins The variability of 15Nbulk values in the Bangemall and Roper shallow water facies can be better understood by drawing an analogy to the variability of 34S values proposed for the Proterozoic 188 In this conceptual model the rate of change in the nitrate isotopic composition in the shallow water environments is controlled by the size of the nitrate reservoir and the size and isotopic composition of fluxes that both add and remove isotopically distinct nitrogen The relationship is as follows with fluxes labeled in Figure 4 7: d NO3- dt Fupwell upwell Fnitri nitri Ffixation fixation Fassim assim Fden sed den sed Fden wc den wc MNO3- d NO3- dt is the isotopic rate of change of the shallow water nitrate reservoir Fupwell is the flux of nitrogen being upwelled from deep water environments upwell is equivalent to 15NNH4 - 15NNO3 which is the isotopic difference of the ammonium being upwelled and the existing shallow water nitrate reservoir Fnitri nitri is the isotopic flux term of organic nitrogen that is nitrified within the shallow water environment nitri is 0 because nitrification likely goes to completion and will preserve the isotopic composition of shallow water nitrate from which the organic matter was likely derived Ffixation fixation is the isotopic flux contributed by N2 fixation in shallow water environments This flux could contribute to changes in the isotopic composition of shallow water nitrate as fixation equals 15Nfixation - 15NNO3- 91 Figure 4 5 15Nbulk plotted against 13Corg for the Bangemall A and Roper B basins The plot also includes those samples from the Bangemall basin that do not belong to one of the three main transects shown in Figure 4 3A-C Error bars are 1 92 Figure 4 6 Histogram of Proterozoic offshore 15Nbulk data A Neoproterozoic era 1 00 548 Ga with data from Ader et al 125 and Kikumoto et al 189 B Mesoproterozoic era 1 6-1 0 Ga with data from St eken 24 Luo et al 171 and this study C Paleoproterozoic era 2 5-1 6 Ga with data from Busigny et al 113 Kump et al 115 and Godfrey et al 72 Subsets of the datasets listed above were taken to only include offshore environments: from Ader et al 125 the Camil Carmelo and Copacel sections from Brazil and all data from Svalbard and North Canada From Kikumoto et al 189 all data from the Doushantuo Formation From St eken 24 data from the Newland Formation in Deep Creek From the Bangemall and Roper of this study deep and basinal data respectively From Luo et al 171 all data From Busigny et al 113 data from the Brockman Iron Formation From Kump et al 115 data from above 180 meters in core depth from 0m to 180m From Godfrey et al 72 data from cores MGS-7 and MGS-8 The inset in panel A shows modern marine sediment data from Tesdal et al 179 The modern data show no correlation with water depth and are therefore not separated by facies Paleoproterozoic data from the Aravalli Group 74 190 were excluded because they are of higher metamorphic grade and their depositional environment is uncertain 93 It may be the case however that Ffixation is 0 because N2 fixation was probably negligible onshore where nitrogen was more available Fassim assim is the isotopic flux of nitrogen being assimilated by shallow water organisms This is likely also 0 as there is no evidence for non-quantitative assimilation negative 15Nbulk values Fden sed den sed is the isotopic flux associated with sedimentary denitrification den sed is also close to 0 as sedimentary denitrification does not impart large isotopic fractionations on the nitrate in the overlying water column 123 191 Fden wc den wc is the isotopic flux from canonical water column denitrification den wc is around 20 MNO3- is the total onshore nitrate reservoir in grams or moles Removing the 0 terms we are left with: d NO3- dt Fupwell upwell Fden wc den wc MNO3- The average isotopic rate of change is 14 7 4 7 and 0 3 per 100 meters within the shallow water facies of the Bangemall Roper and Belt basins respectively The sedimentation rates between the basins were likely different but if we assume that they were somewhat comparable then these isotopic rates of change may reflect primary changes in shallow water nitrate For the Belt basin d NO3- dt is an order of magnitude less than in the Bangemall and Roper basins In the context of our model this would have to result from the Belt basin either having a greater nitrate reservoir in shallow waters MNO3- or smaller fluxes Fupwell upwell and Fden wc den wc It is unlikely that the Belt basin had a greater nitrate reservoir compared to the Bangemall and Roper basins because it was likely more restricted from the open ocean see following section As a result of this restriction upwelling and water column denitrification fluxes were likely smaller in magnitude in the Belt basin than those in the Bangemall and Roper basins 94 Figure 4 7 Schematic of the proposed Mesoproterozoic nitrogen cycle Shown are fluxes F both in and out of a shallow water nitrate reservoir as we propose for the Bangemall and Roper basins Blue flux arrows represent fluxes that when varied are most likely able to change the isotopic composition of a relatively small nitrate reservoir All other fluxes are unlikely to change the isotopic composition of the nitrate reservoir Flux labels are as follows: Ffixation N2 fixation flux Fnitri nitrification of organic matter to nitrate Fassim assimilation of nitrate into biomass Fden wc water-column denitrification Fden sed sedimentary denitrification Fupwell upwelling of ammonium from anoxic waters 95 In this conceptual model then the Bangemall and Roper basins having a greater connection to the open ocean had larger upwelling and denitrification fluxes that when altered resulted in changes in the isotopic composition of the shallow nitrate reservoir This variability is reflected in the variability of 15Nbulk values from the shallow water depositional environments in each basin Variability in the deep basin can be attributed to a small nitrogen reservoir where incursions of nitrate into the deep ocean could not fully be isotopically buffered by existing NH4 If the fixed nitrogen reservoir had been larger in the Mesoproterozoic ocean then it would have been less susceptible to isotopic change The degree of variability in our sample sets may thus be a reflection of a small and isotopically variable nitrogen supply in comparison to the Paleo- and Neoproterozoic settings that are more uniform Section 4 7 6 4 7 5 Comparing the Bangemall and Roper basins to the Belt basin The Bangemall and Roper basins show facies-dependent trends in nitrogen isotopes similar in direction to but smaller in magnitude than the Mesoproterozoic Belt Supergroup Table 4 3 24 suggesting that such a pattern may have been common in the Mesoproterozoic Era The Belt basin however has a deep depositional environment where 80% n 21 of the samples fall within the range of Nif N2 fixation compared to 36% and 8% in the Bangemall and Roper respectively This difference could be an artifact of sampling if relatively deeper facies were sampled in the Belt and not captured in the Bangemall transects and Roper samples However the Belt basin could instead have been more restricted and consequently more strongly stratified than the Bangemall and Roper basins which may have led more rapid depletion of the dissolved nitrate reservoir This hypothesis is supported by the geometry of the Belt basin which is thought to have been formed by intracontinental rifting 192 and may only have had limited exchange with the 96 open ocean for some of its history e g 193 195 The unusually steep gradient in organic carbon isotopes from -32 2 to -22 9 which is not seen in any other Mesoproterozoic basins 158 this study further supports significant water mass stratification in the Belt basin Hence the subtler carbon and nitrogen isotopic gradients in the Roper and Bangemall basins may be more representative of global marine conditions Importantly none of the three basins show nitrogen isotope values within the range of vanadium or iron-based nitrogenases -6 37 suggesting that Mo was available in sufficient quantities for the dominance of the molybdenum nitrogenase despite being present at low concentrations in the Mesoproterozoic ocean 98 4 7 6 Mesoproterozoic nitrate minimum While available data are consistent with a basinal gradient in 15Nbulk and thus in nitrogen speciation in the Mesoproterozoic Section 4 7 3 this was not the case in the earlier and later Precambrian In the mid-Archean Witwatersrand Supergroup 2 87-2 96 Ga sediments deposited near estuaries 1 2 1 0 are on average slightly heavier than marine sediments further offshore -1 6 0 8 41 but almost all of these values fall within the range of biological N2 fixation The subtle gradient is thus more likely a result of varying Fe2 availability to diazotrophic microbes 173 rather than a gradient in nitrate abundance Nitrate was likely scarce in all parts of the mid-Archean ocean including shallow waters 41 consistent with very low levels of atmospheric oxygen at this time 46 52 196 Surface water nitrate levels may have increased in the late Archean with the onset of low levels of oxidative weathering and enhanced oxygenation of the surface ocean 10 44 50 51 53 55 In the Ghaap Group in South Africa 2 67-2 52 Ga bulk 15Nbulk values have a mean of 4 6 2 0 and show no systematic variation between different facies which include shallow-water microbialites and deeper-water siliciclastic 97 sediments 4 These results were interpreted as evidence of aerobic nitrogen cycling 4 implying that nitrate had become a significant ion in the surface ocean at this time Across the ArcheanProterozoic boundary in the Hamersley Group in Western Australia 2 50-2 46 Ga combined data sets from Garvin et al 3 and Busigny et al 113 capture offshore marine facies from the outer shelf and the shelf edge respectively and in both settings values are mostly above 4 especially after the whiff of oxygen at 2 5 Ga 3 5 6 Although data from contemporaneous shallow marine sediments are not available these fairly heavy values in offshore sediments are distinct from the comparatively light values found in the Mesoproterozoic 24 this study and provide strong evidence for a significant reservoir of nitrate throughout the surface ocean at the end of the Archean and extending into the early Paleoproterozoic From the late Paleoproterozoic Godfrey et al 72 analyzed drill-core samples along a basinal profile in the Animikie Group 1 871 84 Ga and reported a subtle gradient of 1 0-1 4 from onshore to offshore the latter being slightly lighter but nearly all their values 98% were above 3 irrespective of environment Hence nitrate was probably relatively abundant in the surface ocean across all environments in the Animikie basin This may also have been the case in most of the Neoproterozoic where bulk 15N values are mostly above 2 and show no systematic basinal gradient 125 From the late Neoproterozoic onwards nitrate depletion is only reported during temporary anoxic events 197 204 throughout most of the Phanerozoic the nitrogen cycle seems to have been predominantly aerobic with little spatial variance 187 Hence the Mesoproterozoic basins analyzed in this study appear to be anomalous in displaying subtle but significant facies-dependent variation in nitrogen isotopes and by inference nitrogen speciation We cannot rule out the possibility that this pattern is biased by latitudinal or oceanographic effects because all of the Mesoproterozoic sites were originally located at low latitudes 30 98 205 206 and in epicontinental seas which could have enhanced stagnation and stratification of the water column The Paleoproterozoic Animikie basin on the other hand formed at a higher latitude 60 207 possibly under colder temperatures which would have favored downward mixing of oxidants produced in the upper ocean Samples from other localities and better constraints on paleolatitude for other Precambrian basins would be needed to test this possibility The relatively light 15Nbulk values in offshore sediments from the Mesoproterozoic are unique and may have been a global characteristic of this time period Figure 4 6 Figure 4 1 shows a compilation of bulk nitrogen isotopic compositions from offshore marine environments highlighting the decline between 1 7 Ga and 1 2 Ga or possibly later This interval post-dates the proposed mid-Paleoproterozoic oxygen overshoot 2 3-2 0 Ga 116 118 208 209 and has recently been identified as a time when atmospheric pO2 may have dropped back to as little as 0 1% or as great as 4% 144 210 of present atmospheric levels until a second potentially protracted rise to nearer modern amounts across the Neoproterozoic Paleozoic possibly beginning at 800 Ma 97 211 Statistical analysis of global Fe-speciation data indicates that while subsurface anoxia was widespread throughout the Proterozoic Eon euxinia was disproportionately common in Mesoproterozoic oceans 100 consistent with lower atmospheric oxygen levels Marine sulfate concentrations are also thought to have declined after 1 7 Ga from 10 mM to less than 1 8 mM followed by a return to 2-3 mM after 1 3 Ga 117 119 188 212 Given that the redox potential of nitrate is intermediate between that of sulfate and oxygen 213 it is plausible that the abundance of nitrate in the surface ocean also declined in the mid-Proterozoic This would have encouraged microbial N2 fixation leading to relatively low 15Nbulk values in offshore marine sediments Figure 4 6 Nitrification of ammonium to nitrate requires oxygen and so the production of nitrate was perhaps favored in shallow waters where O2 was actively being produced 99 In contrast nitrification may have been suppressed further offshore where O2 production was lower 4 7 7 Implications for life Both the Bangemall and the Roper basins have fossil assemblages that are consistent with an onshore-offshore trend of decreasing organismic diversity and abundance seawards 93 94 St eken 24 suggested there could be a linkage between the basinal nitrate gradient observed in the Belt basin and the fossil distributions in the Bangemall and Roper basins if they also had a nitrate gradient Although our data are inconsistent with complete nitrate depletion offshore in the Roper and Bangemall basins unlike in the Belt basin several features of our results indicate that nitrate concentrations were probably significantly lower than in the Paleoproterozoic Neoproterozoic and modern ocean First 15N values were below 3 in offshore Mesoproterozoic sediments compared to 4 to 5 in the Cenozoic and modern 179 187 Figure 4 6 which likely reflects a mixture of two biological inputs from nitrate-assimilating organisms 15N 0 and diazotrophs 15N 0 The latter would not have been ecologically significant if nitrate had been abundant because N2 fixation is energetically costly Second the lightest 15Nbulk data from the peritidal facies suggest rapid nitrate depletion and domination by N2 fixers during temporary restriction from the marine nitrate reservoir This is consistent with microfossil evidence from the Bangemall basin peritidal facies where Palaeopleurocapsa which resembles a modern genus of nitrogen-fixing cyanobacteria Section 4 4 1 is found 94 Further support comes from biomarker evidence from the late Mesoproterozoic Taoudeni basin 1 1 Ga which indicates that even some shallow waters during the Mesoproterozoic could have also been deficient in oxidized nitrogen species 214 Lastly the variability in 15N throughout the basins 100 is best explained by a small nitrate reservoir whose relative size and isotopic composition were easily perturbed Similar variability is seen in sulfur isotopes from the Mesoproterozoic which is interpreted as an artifact of a small sulfate reservoir 188 215 It is likely that the magnitude of nitrogen speciation trends varied between different basins nevertheless all the currently available nitrogen isotope data point towards generally low Mesoproterozoic nitrate concentrations in the surface ocean with a minimum in offshore waters If so then nitrogen availability may have contributed to the ecological distribution of marine organisms As photosynthetic eukaryotes are apparently outcompeted by prokaryotes in nitrate-limited environments 103 107 216 217 it is likely that the open ocean was dominated by prokaryotic organisms with eukaryotes perhaps only inhabiting the most oxygenated part of the water column Nearer shore a more diverse ecosystem including abundant eukaryotes may have developed in relatively nitrate-rich waters The very shallowest peritidal settings may again have excluded eukaryotes not because of anoxia but due to periodic nitrate depletion during intervals of restricted water exchange at low tides This ecological gradient may have also had evolutionary consequences in that eukaryote diversification was not possible in offshore and onshore nitratepoor settings but was instead confined to near-shore waters that were relatively nitrate-rich It is also possible that eukaryotic life was inhibited directly by episodic upwelling of anoxic and sometimes sulfidic waters 218 these two mechanisms of eukaryotic inhibition are not mutually exclusive and likely both occurred If so then perhaps eukaryotes underwent a major evolutionary radiation and rise to ecological dominance only after a Neoproterozoic oxygen increase resulting in globally prevalent nitrification and deeper or less widespread anoxic water masses Thus our data support the hypothesis of Anbar & Knoll 92 that nitrogen availability may have been a key constraint on the evolution of eukaryotes 101 4 8 Conclusion Nitrogen isotope data from the Bangemall and Roper basins considered in concert with the Belt basin 24 are consistent with the idea that distinct facies-dependent nitrogen regimes largely aerobic near-shore and partially to fully anaerobic offshore were a common feature in the early Mesoproterozoic Peak enrichment in 15Nbulk occurs in shallow and peritidal depositional environments and cannot solely be explained by post-depositional alteration There is no apparent systematic cross-basin bias of oxic versus anoxic diagenesis or metamorphism so it is likely that these heavy values reflect the primary isotopic composition of biomass forming in the water column The most plausible explanation for positive 15Nbulk values in the shallower waters is that a pool of dissolved nitrate was partially denitrified and the residual isotopically heavy nitrate was subsequently assimilated into biomass as in the modern ocean e g 76 Instances of light 15Nbulk in peritidal environments probably represent transient periods of isolation from the marine nitrate supply at low tide leaving fixation as the primary source of nitrogen Light 15Nbulk values in deep water samples are consistent with a predominance of N2 fixation by the Mo-Fe nitrogenase slightly heavier samples likely record mixing with biomass from nitrate assimilators during intervals when nitrate was more available in the surface ocean as in parts of the modern redoxstratified Cariaco basin 219 Such mixing implies that nitrate concentrations were low because any isotopic signal from N2 fixation would be erased without a nitrate deficit in the water column 122 It is not clear whether the proposed spatial and temporal trends in nitrogen cycling indicate reduced concentrations of trace metals oxygen or both in deep water environments during the Mesoproterozoic where aerobic nitrogen cycling seems to have been limited Our results are 102 consistent with metal-nitrogen co-limitation controlled by the extent of euxinic conditions 92 98 220 but probably only to a degree that limited nitrification and denitrification offshore while N2fixation by the Mo-Fe nitrogenase was able to persist Our data are also consistent with a Mesoproterozoic oxygen decline 116 117 119 209 as nitrogen isotopic ratios are lower in the Mesoproterozoic than in the Paleo- and Neoproterozoic suggesting relatively lower nitrate concentrations A basinal gradient of dissolved oxygen concentrations higher near-shore to lower offshore could potentially also produce the nitrogen isotopic trend seen in our datasets by limiting the extent of nitrification Thus the theoretical prediction of two spatially separated states of the nitrogen cycle in the mid-Proterozoic 129 is now expressed in three different Mesoproterozoic basins though how much this pattern directly depended on the location and extent of underlying euxinic waters is still unclear Regardless such patterns would probably have restricted eukaryote biomass and evolution to near-shore settings as hypothesized by Anbar and Knoll 92 though the key limiting factor may not have been nitrogen fixation rates but nitrogen speciation 4 9 Acknowledgements We thank NSF EAR FESD grant 1338810 RB NASA grant NNX16AI37G RB the Agouron Institute RB the NASA Astrobiology Institute s Virtual Planetary Laboratory RB the NSF Graduate Research Fellowship Program MAK and the Department of Earth and Space Sciences University of Washington Goodspeed Geology Fellowship MCK Misch Fellowship MCK and the Kenneth C Robbins Field Study Fellowship MCK for funding Andrew Schauer Virginia Littell and the University of Washington Isolab for technical support and Simon Poulton for three additional Roper Group samples We also thank Chris Reinhard two 103 anonymous reviewers and the associate editor James Farquhar for their comments that significantly improved this paper 104 Chapter 5: NITROGEN AND CARBON BIOGEOCHEMISTRY ACROSS THE ORDOVICIAN-SILURIAN BOUNDARY AT DOB S LINN SCOTLAND This manuscript will be submitted to the journal Paleogeography Paleoclimatology Paleoecology Co-authored by Matthew C Koehler Eva E St eken and Tony Prave 5 1 Abstract Marine ecology co-evolved with dramatic environmental change across the Ordovician- Silurian boundary In particular apparent trends in nitrogen cycling both latitudinal and with respect to depositional environment suggest extreme spatial-temporal heterogeneity in the availability of dissolved-nitrogen species associated with global climate change and mass extinction Missing from these data-sets are analyses from the paleo-mid-latitudes and facies that represent the most distal continental margin Here we present nitrogen and carbon isotope abundance data from the O-S Global Boundary Stratotype Section and Point at Dob s Linn Scotland Nitrogen isotope and abundance data from bulk rock shale and bentonite samples as well as kerogen extractions indicate that unlike in tropical shallow-water sections the sub-tropical locale at Dob s Linn shows no changes in nitrogen cycling associated with the Hirnantian glaciation and suggest persistent nitrogen limitation Diminishing oxygen minimum zones cause by increased ocean ventilation towards the end of the Ordovician apparently increased bioavailable nitrogen concentrations in shallow water tropical settings alleviating nitrogen limitation in these environments If perturbations to the nitrogen cycle in part caused the mass extinction and changes in climate then these results highlight the importance of a climate-denitrification positive feedback in regulating the global environment Carbonate data from this and other sections suggest a 105 deepening of the carbonate compensation depth during the Hirnantian which is also consistent with increased ocean ventilation This indicates Pacific-style responses of the CCD to glacial interglacial periods were operational across the O-S boundary and that the expansion of deep-sea abiotic carbonate deposition and preservation could have in-part mediated changes to surface CO2 during these extreme changes in climate 5 2 Introduction The Ordovician-Silurian O-S boundary is characterized by dramatic changes in Earth s climate surface environments and biosphere It is generally thought that through the Ordovician atmospheric pCO2 levels and surface temperatures steadily decreased from greenhouse conditions 221 culminating in the Hirnantian glaciation This climatic trend likely caused the progressive ventilation of the oceans due to stronger latitudinal temperature gradients 222 causing deepocean anoxia to diminish a process that has been used in part to explain the Great Ordovician Biodiversification Event GOBE 223 The initiation of the Hirnantian glacial period likely occurred when pCO2 levels fell below a threshold level 224 225 eventually resulting in the expansion of a Gondwanan ice sheet to volumes that rivaled or surpassed those of the Pleistocene Last Glacial Maximum 226 The beginning and end of the glaciation were punctuated by two mass extinction pulses that in combination eradicated 85 3% of all marine invertebrate animal species 227 The close associations between global climate change and mass extinction pulses have led to the common view that the former caused the latter but there are more recent data linking widespread ocean anoxia not necessarily correlated with global climate change to these extinctions 228 As with all documented mass extinction events astronomic e g 229 230 geologic e g 231 233 and biologic drivers e g 234 have been proposed as mechanisms for 106 global environmental change and extinctions Further insights into these cause-effect relationships can be gained from biogeochemical cycles The carbon isotope record from the Ordovician indicates periods of enhanced organic carbon burial caused by increases in marine primary productivity e g 235 leading to increased CO2 sequestration from the atmosphere The start of the largest magnitude carbon isotope excursion is documented from the early Hirnantian stage with a peak in the late Hirnantian consistent with rapid CO2 drawdown and global cooling Sulfur isotopes from carbonateassociated sulfate and pyrite covary through the Ordovician and recent interpretations of these records suggest more vigorous ocean ventilation by the middle Ordovician 222 Similar sulfurredox records through the Hirnantian were initially interpreted as representing increased anoxia at the start of glaciation 236 but reevaluation of the single-reservoir model and independent proxies suggest that these records may indicate enhanced ocean ventilation and O2 at the start of the Hirnantian 222 It has been proposed that changes in the biogeochemical nitrogen cycle specifically changes in the global rate of denitrification6 in part contributed to the onset and termination of glaciation 237 as well as the abundance and distribution of prokaryotic and eukaryotic organisms in the global ocean 238 240 Nitrous oxide N2O is a by-product of denitrification and a potent greenhouse gas so decreases in the global denitrification rate caused by glacial eustatic sea-level fall or diminishing oxygen minimum zones could act as a positive cooling feedback by decreasing the N2O flux to the atmosphere Also because denitrification removes bioavailable nitrogen NO3and NO2- from the ocean eukaryotic organisms that rely more heavily on the presence of fixed nitrogen in the form of nitrate may become advantaged if global rates of denitrification decreased 6 Denitrification is the biologically mediated reduction of oxidized nitrogen species NO3- NO2- to N2 gas through multiple intermediate nitrogen oxide products 107 If such a feedback between climate redox and global rates of denitrification existed then nitrogen cycling is inherently linked to the carbon cycle and the mass extinctions pulses It is worth noting that recent high-resolution uranium isotope data from across the O-S boundary suggests a decoupling between global ocean redox and eustatic sea-level climate at the end-Hirnantian as the last mass extinction pulse is associated with an ocean anoxic event that begins during a sea-level high stand and lasted through a sea-level low stand 228 Importantly this does not preclude a relationship between anoxia the nitrogen cycle and mass extinctions but instead suggests climate and sea-level may at times be decoupled from these systems Nitrogen isotopic data from sections in present-day China 237 Canada 239 and USA 199 record nitrogen limitation in the paleotropics 10o N 15o N and 5o S respectively spurred by widespread ocean anoxia and high rates of denitrification before near the end and after the Hirnantian During the Hirnantian positive 15N excursions indicate temporary alleviation from nitrogen limitation in shallow tropical sections Shallow-water sections from the high-paleolatitudes 60o S in the present-day Czech Republic break from this pattern showing positive 15N values from the Katian through the Hirnantian suggesting persistent nitrate availability 239 The emerging spatial-temporal trends in nitrogen cycling from the available data suggest redox changes the extent of oxygen minimum zones across the O-S boundary affected nitrogen bioavailability in tropical shallow-water settings Here we add nitrogen and carbon isotope and abundance data from the O-S Global Boundary Stratotype Section and Point at Dob s Linn Scotland This is the first nitrogen isotope dataset from mid-latitudes across the O-S boundary and is one of the deepest sections recording 15N values and transient increases in deepsea carbonate abundances during the Hirnantian 108 5 3 Geologic setting Sample were collected from the Upper Hartfell Shale and the conformably overlying Lower Birkhill Shale formations of the Moffat Shale Group located in the Central Belt of the Scottish Southern Uplands terrain The 11 meter section spans from the late Katian Rawtheyan to the Rhuddanian and was sampled at 30 cm resolution The Upper Hartfell shale is a massive organic-lean 0 02% - 0 15% total organic carbon TOC grey shale with subordinate layers of organic-rich 1% - 2% TOC graptolitic shale referred to as the anceps bands 241 These bands are reminiscent of the Lower Birkhill Shale which is a laminated organic-rich 1 2% - 2 0% TOC graptolitic black shale with abundant disseminated and laminated pyrite These formations at Dob s Linn have only been subjected to prehnite-pumpellyite facies metamorphism 242 The sampled section contains over 20 bentonite beds through both formations ranging from 1 cm to 12 cm in thickness Seven of these beds were sampled for nitrogen and carbon analyses Geochemical similarities between these bentonite horizons and bentonite beds deposited in eastern Baltica suggest a common volcanic source that transported ash more than 1200 km from modern-day Scandinavia which was located on the eastern margin of the Iapetus paleo-ocean 243 This correlation has qualitatively led to the hypothesis that perhaps the Dob s Linn sediments were deposited near the middle of the Iapetus Ocean before its closing 243 Classic interpretations however suggest deposition on the eastern continental margin of Laurentia at 30oS north of the Iapetus suture and on the more distal section of the continental slope 244 245 This is perhaps more consistent with the micro-turbidite sedimentary structures in the Dob s Linn sediments 241 244 Regardless deposition at Dob s Linn represents the most distal-marine sedimentary package across the O-S to be analyzed for nitrogen isotopes 109 5 4 5 4 1 Methods Sample preparation: bulk and kerogen isolates Bulk samples and kerogen extractions were prepared using standard methodology e g 23 24 Outcrop samples were cut into 3 x 3 x 2 cm chips and weathering rinds were shaved off with a diamond tip saw to remove modern contaminants The chips were then crushed with a steel pestle on a steel plate which were covered with clean aluminum foil between each sample The resulting smaller chips and coarse powder were transferred to clean glass beakers that had previously been heated at 500oC overnight to remove contamination Each sample was then sonicated with ethanol for 2 minutes followed by 2N HCl for 20 seconds to further remove modern contaminants The acid was rinsed from the sample three times with 18M milli-Q water and samples were covered and left in a ventilated oven at 60oC until dry Samples were then finely powdered using an aluminum oxide puck mill that was thoroughly cleaned with ethanol 18M milli-Q water and pre-combusted silica sand between each sample Sample powders were stored in clean glass scintillation vials Before analysis bulk shale and bentonite samples were decarbonated using the methodology outlined in Section 5 4 3 Methods for kerogen extractions were adapted from 25 Depending on organic carbon abundance estimates 1-4 grams of unacidified powder were weighed into 250ml Teflon bottles 100ml of 50% v v hydrofluoric acid HF were added to sample bottles to react overnight in a shaking water bath at 55oC The acid was then decanted after centrifugation and next any resulting fluoride precipitates were eliminated by treating the samples with 100ml of saturated boric acid in 50% v v HF overnight in a shaking water bath at 55oC The acid was decanted and samples were rinsed 3 times with 200ml of 18M milli-Q water Kerogen was transferred from the Teflon bottles to clean glass scintillation vials frozen and then freeze-dried for two days 110 5 4 2 Analytical Methods 15Nbulk 15Nkerogen total nitrogen TN total organic carbon TOC and 13Corg data were measured using a Costech ECS 4010 Elemental Analyzer coupled via a Conflo III to a Thermo Finnigan MAT253 continuous-flow isotope-ratio mass spectrometer Groups containing a maximum of 6 samples were bracketed by three in-house standards calibrated to reference materials USGS40 and USGS41 26 that were used to calibrate both abundance and isotope analyses as well as to monitor accuracy within individual runs A fourth in-house rock standard was used to monitor long-term precision Average analytical accuracy for 15Nbulk kerogen and 13Corg through all runs was 0 16 0 16 and 0 08 0 06 respectively and average analytical precision was 0 11 and 0 02 respectively The average standard deviation of sample replicates was 0 11 for all 15N measurements and 0 15 for all 13Corg measurements 13Ccarb and 18Ocarb analyses were performed on unacidified sample powders using a Kiel III Carbonate Device coupled to a Delta V Plus Thermo Finnigan Isotope Ratio Mass Spectrometer The Kiel parameters were set to a phosphoric acid temperature of 80oC 10 drop acidification for each sample and a reaction time of 10 minutes in order to digest more recalcitrant carbonates such as siderite Internal calcite standards C1 and C2 that had been calibrated to international standards NBS19 LSVEC and NBS18 were used to calibrate 13Ccarb and 18Ocarb values on the VPDB scale A third quality control standard CQS2 was used to track long term precision In total 15 standards were run with 20 samples from Dob s Linn The average analytical accuracy based on all standards was 0 02 0 02 and 0 04 0 03 for 13Ccarb and 18Ocarb respectively The average analytical precision based on the in-house standard CQS2 was 0 02 and 0 05 for 13Ccarb and 18Ocarb respectively The average standard deviation of sample 111 replicates was 0 1 for all 13Ccarb measurements and 0 2 for all 18Ocarb measurements Because all the standards are calcite calcite was the assumed mineralogy when correcting the raw 18Ocarb data for fractionation during phosphoric acid digestion Fractionation factors during phosphoric acid digestion of carbonate vary depending on mineralogy siderite aragonite dolomite etc 246 If dolomite siderite or aragonite were the primary carbonate the 18Ocarb data presented here would only shift by approximately 1 and their relative values would remain the same 246 5 4 3 Percent carbonate The percent carbonate of our samples was initially calculated gravimetrically e g 23 24 Sample powders were acidified with 6 M HCl at 60oC in three iterations over three days They were then thoroughly rinsed three times with 18 Milli-Q deionized water to remove the acid The weight of the samples before and after the acidification was used to estimate percent carbonate The resulting values were 10% greater than values measured by the Kiel device prompting suspicion that the phosphoric acid in the Kiel was not properly digesting recalcitrant carbonate in the Dob s Linn samples Because recalcitrant carbonates such as siderite are notoriously hard to digest in phosphoric acid special attention was given to conditions that will yield complete digestion e g 247 To test if the Kiel III Carbonate Device was completely digesting all carbonates in our samples we used a methodology similar to the open vessel digestion of Fernandez et al 2016 which was shown to more rapidly digest siderite Approximately 50 mg of two Dob s Linn samples were weighed into separate Pyrex reaction tubes Room-temperature phosphoric acid was added to the reaction tubes but did not interact with the sample at the bottom of the tube because of its high viscosity 112 Reaction tubes were then attached to a vacuum line to remove the headspace and subsequently flame-sealed They were then vortexed to mix the acid with the sample and placed in an oven overnight at 100oC The tubes were intermittently vortexed during the hot digestion period to ensure the acid was completely mixing with the sample After the digestion period the tubes were cracked on a vacuum line and gasses were cryogenically cleaned to isolate the CO2 produced by the digestion The amount of CO2 produced by the digestions was then measured manometrically 5 5 5 5 1 Results Carbonate The two Dob s Linn samples that were reacted with phosphoric acid overnight at 100oC and measured manometrically are within 1% of the percent carbonate values produced by Kiel analysis suggesting that the Kiel carbonate values are reliable This means the gravimetricallydetermined carbonate abundances were inaccurate overestimating by more than 10% likely due to the dissolution of non-carbonate species by the hot concentrated HCl As a result all carbonate values isotopes and abundances discussed below refer to the Kiel-measured values 5 5 2 Nitrogen and carbon The lithological change between the Upper Hartfell and Lower Birkhill is driven by changes in the abundance of organic matter TOC % is on average an order of magnitude greater in the Lower Birkhill 1 6 0 3 than in the Upper Hartfell 0 11 0 11 Because the C N atomic ratios of organic matter are similar - and the abundance of silicate-bound nitrogen is invariant - 113 through both formations the fraction of nitrogen bound in organic matter Fkerogen 7 rather than in silicates is an order of magnitude greater in the Lower Birkhill 26 2 3 6% than in the Upper Hartfell 2 8 2 6% In both formations silicate-bound nitrogen is the dominant nitrogen fraction This is especially true for the bentonite samples where TOC % values are comparable to Upper Hartfell shale but TN % values are approximately triple those from shale samples Figure 5 3 and Figure 0 12 Because the silicate-bound nitrogen fraction is large and silicate-bound nitrogen abundances are invariant in the Upper Hartfell and bentonite samples it is no surprise that bulk C N atomic ratios are strongly correlated to TOC % in these samples Figure 0 13 This correlation is much weaker in the Lower Birkhill because the variability in organic matter C N atomic ratios in this more organic rich formation contributes more significantly to bulk C N atomic values Carbonate % in bulk shale samples are all below 1% the limit of detection for these runs except during the middle Hirnantian where values peak at 23% Figure 5 3 15Nbulk measurements through both the Upper Hartfell Shale and Lower Birkhill Shale are relatively invariant and range from 0 7 to -0 1 Figure 5 1a and Figure 5 2 15Nbulk measurements from the bentonite horizons are equally invariant and on average slightly more negative ranging from -0 47 to -1 25 15Nkerogen measurements through both formations are more varied ranging from -6 0 to 1 3 with seven out of ten samples containing 15Nkerogen values that are 1 more depleted than 15Nbulk values of the same sample Figure 5 1e and Figure 5 2 13Corg measurements from the shales agree with those of previous studies e g 248 249 with values up to -29 3 during the middle Hirnantian followed by a steady decrease to more depleted values averaging -32 8 starting at the lithological boundary to the Birkhill Shale Figure 7 1 % % % % 114 5 2 One 13Corg measurement in the Rawtheyan is enriched to a similar magnitude -29 1 also corroborating a prior study 248 Bentonite samples are all enriched in 13Corg compared to adjacent shale samples ranging from -26 to -30 Figure 5 1c Differences between kerogen and bulk shale 13Corg measurements of the same sample are all less than 0 5 13Ccarb and 18Ocarb values for these carbonates average -4 9 0 4 and -5 3 0 3 respectively Figure 0 11 5 1 5 1 1 Discussion Potential for post-depositional alteration Post-depositional alteration of nitrogen and carbon isotopes can occur during diagenesis and metamorphism To interpret our data as primary biological and environmental signals these effects need to be addressed 5 5 2 1 Diagenesis Depending on the redox state of the water column and marine sediments diagenesis can have varying effects on both carbon and nitrogen isotope ratios In modern marine sediments residual organic nitrogen is often enriched in 15N as a result of preferential 14N loss during oxic diagenesis e g 31 33 34 The magnitude of this enrichment depends on the sediment redox state and sedimentation rates with greater enrichments 3 - 4 occurring in oxic sediments with low sedimentation rates These enrichments approach 0 in anoxic sediments with high sedimentation rates 157 Depletion of 15N in residual organic nitrogen after diagenesis has been observed in a few modern marine environments e g 35 36 and during laboratory experiments 32 The mechanism for depletion is thought to be the addition of 15N-depleted biomass through the in-situ growth of N2-fixing bacteria that have an isotopic composition of -2 to 1 Of course this 115 Figure 5 1 Cross-plots of nitrogen and organic carbon isotopes abundances and ratios from Dob s Linn shales bentonites and kerogen isolates Grey dots are sample measurements that follow the axes labels 116 Figure 5 2 Nitrogen and organic carbon isotope chemostratigraphy through the Dob s Linn section Symbols for bulk shale bentonite and kerogen measurements are as in Figure 5 1 117 Figure 5 3 Nitrogen and organic carbon chemostratigraphy through the Dob s Linn section Symbols for bulk shale bentonite and kerogen measurements are as in Figure 5 1 118 would only cause a depletion of 15N in residual organic nitrogen if the organic matter was originally enriched in 15N i e biomass dominated by N2-fixers is unlikely to be isotopically altered by the addition of biomass from more N2-fixers Iron speciation data from Dob s Linn tell an inconclusive story regarding bottom-water redox conditions through this section because they straddle the empirically-defined oxic-anoxic boundary 236 It is possible that during the Rawtheyan sediments were deposited under more oxic conditions than during the Hirnantian and Rhuddanian However systematic trends in diagenetic alteration of original 15N values are difficult to evaluate due to apparent nitrogen migration and 15N homogenization through this section see section 5 5 2 3 If our rough calculations of original Upper Hartfell 15N values are correct section 5 5 2 3 then the lack of 15N variability through the section suggests that preferential large scale diagenetic changes to any part of the studied section are unlikely Carbonates preserved in the middle Hirnantian sediments at Dob s Linn have relatively light 13Ccarb and 18Ocarb values suggesting that some of the carbonate formed diagenetically from respired biomass Using average Hirnantian 13Ccarb and 13Corg values from other coeval sections 0 to 5 and -28 respectively 239 and the average 13Ccarb value from Dob s Linn mass balance predicts 20% to 30% of the carbon in the Dob s Linn carbonates comes from respired organic matter This means that the carbonate may have formed during early diagenesis rather than syndepositionally There is no correlation between carbonate abundance or 13Ccarb and 13Corg values suggesting that the 13Corg excursion is unlikely to be a result of carbon isotopic equilibration between organic matter and carbonate which only occurs at higher grades above mid-greenschist facies e g 165 167 119 5 5 2 2 Metamorphism The effects of sub-greenschist metamorphism on 15N values is thought to be 2 28 29 Classic signs of metamorphic alteration include increasing 15N values correlated to increasing C N ratios and a divergence between 15Nsilicate and 15Nkerogen values of a given sample This is because nitrogen is more mobile than carbon and 14 N more mobile than 15 N under such conditions There is no correlation between 15N and C N ratios in the Dob s Linn data Figure 5 1d and Figure 5 1f but two samples display 15Nsilicate-kerogen value of greater than 4 In rocks of low metamorphic grade that have not undergone fluid alteration and migration of ammonium 15Nbulk values are usually a good approximation for the original 15N values of deposited organic matter 29 However the fact that all bentonite horizons are characterized by large TN % values very low C N ratios and 15Nbulk values closely resembling the 15Nkerogen from the Lower Birkhill Figure 5 1b suggests that NH4 may have been mobile after sediment deposition and may have homogenized silicate-bound 15N through the section see below 5 5 2 3 Diffusion advection effects on 15Nbulk It is possible that 15Nbulk values through the section particularly through the more organic- lean Upper Hartfell do not strictly reflect the stratigraphically equivalent 15N of original organic matter During the rapid deposition of bentonite horizons TOC is expected to be low because the sedimentation rate greatly exceeds biomass burial If silicate-bound nitrogen is sourced from stratigraphically equivalent organic matter then TN in these bentonite horizons should be correspondingly low TOC in the Dob s Linn bentonite horizons throughout the section 0 05% 0 02% are indeed low comparable to the Upper Hartfell but TN values 0 16% 0 016 are triple and double those seen in Upper Hartfell and Lower Birkhill shales respectively Figure 5 1b and S2 This suggests that the bentonite beds were enriched in ammonium after deposition which 120 is most parsimoniously explained by NH4 diffusion through these sediments during illitization The bentonite layers have higher TN values than the shale samples from the section because the bentonites have proportionally more illite Given the consistent silicate-bound nitrogen abundances through both formations 0 053% 0 006 as illustrated by the linear relationship between the TOC % and C Nbulk of samples from the Upper Hartfell and bentonite horizons shown in Figure 0 13 and the proportional increase observed in illite-rich bentonite horizons it seems likely that the silicate-bound nitrogen fraction throughout the section was derived from fluids with similar NH4 concentrations and 15N values A plausible source for this NH4 may have been the organic-rich Birkhill Shale where NH4 may have been mobilized during diagenetic degradation of organic matter as is commonly observed in modern organic-rich sediments e g 250 251 In principle NH4 can also be sourced from the overlying water column in areas where deep waters are dominated by euxinia 252 This is because sulfate reduction cannot be thermodynamically coupled to NH4 oxidation and therefore leads to accumulation of dissolved NH4 However iron speciation data from Dob s Linn record possible hints of euxinia only in the early Silurian but not during the Ordovician i e long after the deposition of the first ash beds Hammarlund et al 2012 It is therefore unlikely that the NH4 in the bentonites was directly absorbed from the water column during deposition Post-depositional NH4 migration as described above is more likely Importantly the isotopic composition of the bentonites deviates by less than 1 from the surrounding sediments including the Birkhill Shale suggesting that NH4 mobility and assimilation into secondary illite does not impart a significant isotopic fractionation 121 We can estimate the percent of nitrogen in Upper Hartfell samples that is authigenic vs derived from the Lower Birkhill using the following equation if we assume the Upper Hartfell had a similar C N % ratio to the Lower Birkhill before the addition of allogenic nitrogen: % & & - - - 5 1 - where TNauth hartfell is the nitrogen percent of the bulk sample that is estimated to be indigenous in an Upper Hartfell sample TOChartfell is the total organic carbon percent of an Upper Hartfell sample and TOCbirkhill and TNbirkhill are the average organic carbon percent and average total nitrogen percent of Lower Birkhill shale samples respectively Knowing the total nitrogen percent of the bulk sample TNbulk hartfell we can then calculate the fractions of authigenic and allogenic nitrogen in Upper Hartfell samples: & 1 & 5 2 Using these calculations on all Upper Hartfell samples we see that 93% 3 of the nitrogen is likely allogenic derived from then Lower Birkhill We can explore estimates for the original nitrogen isotopic composition of both these fractions using the following equations: 12 N 12 N& & 456 7 89: 456 7 - - : - - 5 3 456 7 89: - - 456 7 - - - : 5 4 where 15Nbirkhill and 15Nhartfell are the nitrogen isotopic compositions of the allogenic derived from the Lower Birkhill and authigenic fractions respectively 15Nbulk hartfell is the bulk nitrogen isotopic value of an Upper Hartfell sample 15Noriginal hartfell is an estimate of the authigenic nitrogen isotope composition and 15Nestimated birkhill is an estimate of the nitrogen isotopic composition of mobile NH4 derived from the Lower Birkhill 122 If we assume 15Noriginal hartfell is best approximated by the 15Nkerogen values from the Upper Hartfell ranging from -6 0 to 1 3 then using equation 5 3 would yield 15Nbirkhill values ranging from 0 35 to 1 0 which is exactly the range of measured Lower Birkhill 15Nsilicate values However 15Nbirkhill is so insensitive to 15Noriginal hartfell that when using average 15Nbulk hartfell FN birkhill and FN hartfell from the Upper Hartfell 15Noriginal hartfell values ranging from -10 to 35 would yield 15Nbirkhill values within the range of all 15N values measured from the Lower Birkhill from -2 to 1 Because of this it is less informative to guess at authigenic Upper Hartfell 15N values in order to estimate the 15N value of NH4 exported from the Lower Birkhill to the Upper Hartfell Perhaps the best approximation of these fractional 15N values is to calculate 15Nhartfell equation 5 4 by using the average 15Nsilicate of the Lower Birkhill as 15Nestimated birkhill This is because the 15Nsilicate of the Lower Birkhill most likely represents the mobile NH4 in pore fluids that was incorporated into the Lower Birkhill silicates and that eventually diffused down into the Upper Hartfell Doing this yields 15Nhartfell values between -2 5 and 1 0 Importantly this range encompasses almost all 15N values measured from the Upper Hartfell bulk and kerogen If instead average 15Nbulk and 15Nkerogen from the Lower Birkhill are used as 15Nestimated birkhill the estimated 15Nhartfell ranges widen to 2 6 to 12 8 and 11 7 to 45 5 respectively Most of these values are unfeasibly high compared to the rest of the Paleozoic and so the average 15Nsilicate of the Lower Birkhill appears to be the best estimate of mobile NH4 and so also the allogenic nitrogen component in the Upper Hartfell 123 5 5 3 Interpretation of nitrogen isotopes It is clear from mass-balance calculations that most of the nitrogen within Upper Hartfell samples are derived from the organic-rich Lower Birkhill where NH4 was mobilized in pore fluids and permeated the rest of the section Using the range of 15Nsilicate values measured from the Lower Birkhill to estimate the 15N values of permeating NH4 yields calculated 15N values of indigenous Upper Hartfell nitrogen ranging from -2 5 to 1 0 Most 15Nkerogen values from the Upper Hartfell another estimation of indigenous 15N values fall within this range Using this range of values as characteristic of the Upper Hartfell it seems almost all nitrogen isotope values both bulk and kerogen measurements from the Dob s Linn section fall within the window attributed to nitrogen fixation by the Mo-nitrogenase -2 to 1 This precludes the presence of a large nitrate pool because coupled nitrification and partial water column denitrification produce isotopically light N2 and isotopically heavy residual nitrogen e g NH4 NO3- so long as these metabolisms do not fully convert the bioavailable nitrogen to N2 The isotopically heavy residual is reflected in biomass as 15N values greater than 1 e g the modern ocean nitrate average is 5 76 So the measured values between -2 to 1 at Dob s Linn suggests that bioavailable nitrogen may have been persistently limited in this offshore environment across the O-S boundary This limitation was likely due to i rapid conversion of NH4 to NO3- in the photic zone and ii rapid rates of denitrification that quantitatively converted all NO3- to N2 depleting bioavailable nitrogen and giving nitrogen fixers an ecological advantage 237 The same nitrogen isotope trend is observed across the O-S boundary at Vinini Creek Nevada USA which was also deposited beyond the continental shelf 199 The addition of the Dob s Linn nitrogen data highlights the spatial-temporal trends in nitrogen cycling presented in 124 Luo et al 2016 Though the scarcity of sections measured for nitrogen isotopes across the O-S boundary prohibits confident reconstructions of the global ocean at that time it appears that within each age Katian Hirnantian and Rhuddanian there are statistically significant differences in 15N values between sections at different latitudes and depositional environments Table 5 5 Figure 5 4 This perhaps is not surprising because latitudinal trends in marine nitrogen cycling are observed in the modern ocean Higher oxygen concentrations in high-latitude waters on the modern Earth inhibits water column denitrification a reaction requiring suboxic-anoxic waters resulting in higher nitrate concentrations 31 These latitudinal trends are primarily driven by physical oceanographic processes that should have operated 425 My ago colder polar and subpolar waters have higher oxygen solubilities and stronger high-latitude winds drive more vigorous vertical water column mixing 253 It should then be expected that higher-latitude waters in the Ordovician ocean were more nitrate-rich To continue with the analogy of global oceanic nitrogen cycling trends between the modern and Ordovician the apparent diametric response in nitrogen isotopes must be addressed In the modern ocean there is a negative correlation between nitrate concentration and the 15N of suspended and sedimentary particulate nitrogen PN 31 The increased oxygen concentration near the poles makes it so water column denitrification is inhibited resulting in less nitrate 15 N enrichment as there is little preferential loss of the light isotope through the partial conversion of nitrate to 15N-depleted N2 As waters move towards the equator oxygen minimum zones form where O2 demand outpaces O2 supply It is in these zones where nitrate is partially converted into 15 N-depleted N2 depleting the nitrate while raising its 15N value These oxygen minimum zones make up a relatively small portion of the modern ocean and there are few 125 instances of quantitative denitrification that allow for nitrogen fixation to dominate Because of this water column denitrification and nitrogen fixation are often decoupled in the modern ocean 31 253 What results are areas with the lowest nitrate concentrations in the modern ocean tend to have the largest 15N values During the Ordovician however the pO2 in the atmosphere and dissolved oxygen concentrations in the ocean were likely lower than modern 100 254 Suboxic and anoxic waters made up a larger portion of the ocean and as a result water column denitrification was likely more vigorous in oxygen minimum zones leading to more areas of quantitative denitrification 15N values in this instance likely had a positive rather than negative correlation with nitrate concentrations because zones of intense water column denitrification would have been coupled to zones of nitrogen limitation and nitrogen fixation Outside these areas of quantitative denitrification which likely occurred in upwelling zones open ocean PN 15N values depended on local areas of partial water column denitrification and more restricted shallow-marine basin PN 15N values depended on the N:P ratios of advected marine waters 253 The N:P ratio of advected seawater controlled which nutrient was limiting and so whether nitrogen fixation could dominate High-latitude sections across the O-S boundary showing consistent 15N values averaging 3 suggest either i persistent local partial water column denitrification or ii high oxygen concentrations in polar subpolar waters inhibited water column denitrification as observed in the modern ocean and the enriched 15N values resulted from spatially-removed zones of partial water column denitrification The sub-tropical and equatorial records from Dob s Linn and Vinini Creek represent persistent oxygen minimum zones across the O-S boundary locally depleting fixed nitrogen allowing for the local ecological dominance of nitrogen fixers 199 This is consistent 126 with recent evidence that suggests ocean anoxia through this period was focused in these more offshore environments 228 But what specifically drove the temporal changes in 15N values from equatorial shallowmarine environments across the O-S boundary To address this we can look toward how nitrogen cycling in the modern ocean responded to climate change over the last 300 kyrs A first order control on the extent of global denitrification is high-latitude O2 supply to the global thermocline which is again controlled by oxygen solubility and wind-controlled rates of thermocline circulation 253 During glacial periods colder more oxygenated waters entered a more rapidly circulating thermocline driven by stronger winds The increased O2 supply from these waters towards the equator diminished oxygen minimum zones and global rates of denitrification What resulted were waters with relatively higher average N:P ratios that when supplied to equatorial shallow-marine basins would i limit nitrogen fixation as primary productivity was instead limited by phosphorus ii favor a significant steady-state nitrate pool and iii yield more positive 15N values through either local partial water column denitrification or a higher average 15N value of the advected nitrogen This response has been observed in the eastern Mediterranean where 15N records indicate more nitrogen replete conditions during glacial periods and nitrogen limitation with significant contribution from nitrogen fixers in interglacial periods 253 The eastern Mediterranean is ultimately supplied by water from the Atlantic Ocean In coeval 15N records from deeper non-restricted Atlantic Ocean sections 15N values were higher due to enhanced - but non-quantitative local water column denitrification during interglacial periods suggesting lower N:P ratio waters from the Atlantic were being fed into the eastern Mediterranean causing nitrogen limitation and spurring nitrogen fixation 253 The opposite relationship is recorded in glacial periods 127 It is important to note two mechanisms that superimpose spatial heterogeneity on nitrogen cycling isotopes in the global ocean: i local primary productivity oxidant demand and ii proximity to zones of denitrification and fixation Locations in the ocean removed from zones of denitrification and fixation do not necessarily record any 15N fluctuations during glacialinterglacial cycles over the last 300 kyrs and would not necessarily fluctuate across the Hirnantian glaciation In summary the positive nitrogen isotope excursions in shallow equatorial records indicate that ocean waters in general were more nitrogen rich in the Hirnantian than during the Katian and Rhuddanian due to the climatically-driven reduction in oxygen minimum zones and rates of denitrification as described above Introduction of waters with relatively higher N:P ratios into more restricted shallow-marine basins created a nitrogen excess that is reflected as positive 15N excursions in these sections Either fixed nitrogen 15N values were on average globally higher due to increased partial water column denitrification during the Hirnantian or shallow-water marine environments hosted local partial water column denitrification leading to the more positive Hirnantian 15N values Either way these data support climatically-driven changes to the global oceanic nitrogen cycle If these changes do in fact represent major perturbations to global rates of denitrification transiently low denitrification rates and accompanying low N2O fluxes a potent greenhouse gas during the Hirnantian could have acted as a positive feedback for global cooling 237 It is worth noting that depositional environment trends in 15N values within these ages are reminiscent of basinal gradients detected at other times in Earth s history that are characterized by large ocean redox heterogeneity such as in the Mesoproterozoic 23 24 Statistical consideration of iron speciation data from mid-Proterozoic to Paleozoic outer shelf basinal samples suggests no 128 difference in the proportion of anoxic bottom waters through this time and also similarities between the Mesoproterozoic and Paleozoic in the proportion of anoxic bottom waters that were sulfidic 100 Because redox state - the extent and vigor of oxygen minimum zones - seems to exert a primary control on spatial-temporal trends in nitrogen cycling it may not be surprising that the two periods have closely related 15N values Figure 5 5 Relatively low 15N values in the Mesoproterozoic when compared to the Neo- and Paleoproterozoic have previously been interpreted as reflecting a temporal minimum in oceanic nitrate concentrations It is possible that periods in the Paleozoic like the O-S boundary are similarly characterized by low oceanic nitrate even when nitrate concentrations should be at a maximum during the Hirnantian glaciation 5 5 4 Carbon cycling The trend in deep-sea carbonate deposition at Dob s Linn mirrors trends observed in the Pacific Ocean during glacial-interglacial cycles over the past 300 kyrs As originally recognized in the Pacific ocean 255 it is evident that the marine carbonate compensation depth8 CCD responds to changes in global climate e g 256 258 Such responses have been extensively documented through the Cenozoic particularly during the Quaternary where proxies from deep sea sediment cores record changes in carbonate preservation 259 260 CO32- concentrations 261 and carbonate dissolution rates 262 263 that correlate to climate proxies The timing of these correlations is well known: carbonate accumulation in sediments is at a maximum during late glacial periods and early deglaciation 262 264 8 The carbonate compensation depth is the depth in the water column where carbonate dissolution balances carbonate production such that no carbonate accumulates in sediments below this depth 129 There has been contention however regarding the mechanisms that drive these observed changes in deep-water carbonate accumulation e g 265 is the dominating control changes to CaCO3 biogenic sources or changes to carbonate preservation i e carbonate dissolution sinks Recently evidence has been piling up in support of preservation control on deep-sea CaCO3 accumulation 265 CaCO3 preservation is dominantly controlled by the saturation state of CO32 where the degree of over under-saturation controls the degree of preservation dissolution During glacial periods over the past 300 kyrs increased CO32- concentrations in mid to deep waters slowed carbonate dissolution resulting in more carbonate accumulation in deep-sea sediments 265 268 Increases in CO32- were driven by increases in the ratio of alkalinity to the dissolved inorganic carbon pool Specifically lower sea-levels during glacial periods i reduced the area of continental shelves and associated neritic carbonate deposition and ii induced weathering of emergent marine carbonates both of which raised CO32- concentrations and favored carbonate preservation at greater depths Carbonate compensation also had a part in controlling carbonate preservation and accumulation: The transfer of CO2 from the surface ocean to the atmosphere during glacial atmospheric pCO2 lows increases CO32- concentrations favoring carbonate accumulation whereas the opposite is true during interglacial atmospheric pCO2 highs A secondary influence on carbonate preservation is CO2 accumulation in sediment pore waters and ocean bottom waters through the remineralization of organic matter Deep waters that contain higher concentrations of CO2 whether from higher rates of aerobic respiration and or longer deep-water residence times are more corrosive and result in a shallower CCD Faster thermocline circulation and venting of deep water CO2 to the atmosphere during glacial times would increase CO32- concentrations lower the CCD and favor carbonate accumulation in deepsea sediments However because of this dependence on primary productivity as well as the vigor 130 and path of ocean circulation it is important to note the relationship between the CCD and climate change is spatially heterogeneous e g the CCD in the Pacific Ocean responded differently to the Last Glacial Maximum than the CCD in the Atlantic Ocean e g 269 In the equatorial and NE Pacific data over the last 45 ka show the CCD fluctuated a maximum of 1800 m with maximum deepening associated with glacial periods and maximum shallowing associated with interglacial periods 259 Carbonate data from Dob s Linn suggest that these mechanisms may have been operating across the Hirnantian The data presented here show no carbonate accumulation in Katian and Rhuddanian sediments but substantially increased carbonate accumulation up to 23% during the Hirnantian glaciation Importantly none of the carbonate preserved in these sediments are biogenic the disseminated carbonate indicates early authigenic formation from sediment pore waters The fact that planktonic foraminifera and coccolithophores had not evolved and proliferated by the early Paleozoic and that the 13Ccarb values are fairly negative further support the authigenic nature of the Hirnantian carbonates Because the 13Ccarb values indicate that only a fraction of the carbonate carbon was derived from respiration and the majority from seawater DIC local processes such as anaerobic oxidation of organic matter that promote carbonate precipitation likely had limited influence This means that during the Hirnantian deep waters above Dob s Linn sediments were near or over-saturated with respect to the solid carbonate phase such that only minor increases in the saturation state in pore waters caused by anaerobic respiration would yield carbonate accumulation Carbonate preservation in this section is also consistent with enhance thermocline circulation as suggested by nitrogen isotopes as ventilation of deep water CO2 would lessen the corrosiveness of bottom waters 131 Table 5 5 Table showing average 15N values in age and depositional bins Standard deviations are 1 Data are from this study Luo et al 270 Melchin et al 239 and Laporte et al 199 Parenthetical numbers represent number of units analyzed in each bin Shallow Tropics Deep Tropics Deep Subtropics Shallow Polar Katian 1 1 0 5 3 0 2 0 4 2 -0 4 2 4 1 3 0 0 2 2 Hirnantian 2 0 1 0 3 0 6 0 3 2 0 1 0 8 1 3 1 0 4 2 132 Rhuddanian 1 1 0 5 2 0 1 0 3 1 -0 1 0 6 1 N A Figure 5 4 P-value relationships comparing 15N populations between a depositional environments within each period and b periods for each depositional environment Green boxes represent a statistically significant rejection of the null hypothesis that the population data from the compared parameters are equal Red boxes represent a failure to reject the null hypothesis Orange boxes represent a limited number of locations samples such that any rejection or acceptance of the null hypothesis should be questioned All statistical tests were twotailed T-tests with an 0 05 133 Figure 5 5 Comparison of outer shelf and basinal 15N values between the eras of the Proterozoic and data from the O-S boundary The colored bars in the O-S panel are from outer shelf-basinal depositional environments The rest of the shallower data are included in black and white for further comparison Notice the similarities between the Mesoproterozoic and O-S nitrogen data Figure modified from Koehler et al 2017 and the included excluded Proterozoic data are as in figure 6 of that reference More recent Proterozoic nitrogen data from Canfield et al 2018 and Zerkle et al 2017 are also excluded due to their shallower depositional environments 134 Lowering of the CCD below Dob s Linn sediments deposited in a continental slope abyssal plain depositional environment only hints at the potential magnitude of changes to the global carbonate cycle across the Hirnantian glaciation Carbonate records from other sections can qualitatively add to this story but only limited interpretations can be made because there are few sections and their paleogeography s were separated Both deep and shallow South China sections the Wangjiawan and Nanbazi respectively show a similar pattern of shale deposition during the Katian and Rhuddanian and carbonaceous shale argillaceous limestone deposition during the Hirnantian 271 272 Deposition at the basinal to continental rise Vinini Creek USA section 273 is complicated by significant sediment contributions from platform carbonates and silicates Changes in carbonate abundance at Vinini Creek often in the form of skeletal fragments and peloids has been attributed to the relative productivity of a carbonate factory on a nearby subtidal platform There the Katian D ornatus zone contains sediments that have up to 90% carbonate by weight 199 but the subsequent Katian zones leading up to the Hirnantian P pacificus and D mirus are notably more carbonate poor dominated by organic-rich shales with subordinate lime mudstones Hirnantian sediments at Vinini Creek comprise lime mudstones and thickening limestone beds 199 Cessation and resumption of carbonate production has been qualitatively linked to changes in sea level but it is possible that platform carbonate productivity and export to the Vinini Creek basin was influenced by a fluctuating CCD The carbonate poor P pacificus and D mirus zones would thus represent a period of CCD shallowing where corrosive bottom waters prohibited allochthonous carbonate accumulation The differences between deepwater carbonate deposition at Vinini Creek and deposition at Dob s Linn and the two South China sections could reflect paleo-CCD spatial heterogeneity between the western Panthalassic Iapetus and paleo-Tethys Oceans or instead the unusual periplatform depositional setting at Vinini Creek 135 Other sections across the O-S seem to have never been below the CCD e g 239 This finding supports the importance of deep-sea authigenic carbonate in the global carbon cycle 274 275 particularly before the evolution of planktonic foraminifera and coccolithophores and extends the relationship observed between deep-sea carbonate cycling and climate in the Pacific Ocean during the Quaternary back to the Ordovician 5 6 Conclusion In agreement with Luo et al 2016 we find that the Katian is characterized by 15N values consistent with nitrogen limitation at Dob s Linn and in all other studied sections save for the two polar sections 239 This is suggestive of widespread anoxia and high rates of denitrification The CCD in the paleo-Tethys and Iapetus oceans may have been fairly shallow at this time potentially above the inner-shelf Nanbazi sediments and seemingly above the Wangjiawan 276 and Dob s Linn sediments An increase in tropical 15N values Table 5 5 and deep-sea carbonate deposition during the Hirnantian supports diminishing oxygen minimum zones and global rates of denitrification as well as more vigorous thermocline circulation Suppressed denitrification and N2O production could have acted as a positive feedback for climate cooling Distal slope and rise environments at Dob s Linn and Vinini Creek were persistent oxygen minimum zones and 15N values suggest they remained nitrogen limited due to high denitrification rates suggesting that areas of deeper-water anoxia persisted 228 Importantly this relationship between deep-sea carbonate deposition and climate change is an O-S example of the Quaternary Pacific-style climatically induced CCD oscillations The nitrogen cycle and carbonate deposition landscape returned to Katian averages with the termination of the Hirnantian glaciation 136 5 7 Acknowledgements This study was funded by University of Washington Department of Earth and Space Sciences Harry Wheeler Scholarship and Jody Bourgeois Graduate Student Support Fund granted to MCK We thank the U W Isolab and Andy Schauer for technical support and Roger Buick for helpful discussions and comments We also thank Sebastian Fischer and K rt praus for help in the field 137 Chapter 6: CONCLUSION Nitrogen isotopes from the geologic record are a window into how the abundance and speciation of nitrogen oxygenation and life co-evolved through time This work contributes to our understanding of how biogeochemical cycles evolved on Earth and informs our understanding of how these cycles may evolve on other worlds that are fundamentally different e g no plate tectonics never accumulate surface oxygen etc The interactions between life and dynamic changes in these Earth systems helps provide a rough framework for how these relationships could occur on other inhabited worlds and how they might be expressed in the atmosphere for detection This dissertation focuses on key periods in Earth s redox and biosphere evolution to explore the timing and conditions that led to environmental and metabolic diversity A summary of the chapters follows: Chapter 2: Recent evidence suggests that oxygenated environments in the Mesoarchean were limited to the shallowest marine and fluvio-lacustrine settings It is becoming increasingly clear that during the Neoarchean oxidizing conditions spread to the photic zone of deeper shelf and basinal depositional environments Here we present nitrogen and carbon isotope ratios from the Mosquito Creek Formation of the Nullagine Group 2 9 Ga to further explore the Mesoarchean redox landscape The 15N and 13Corg values are invariant and suggest an ecosystem dominated by nitrogen fixers anaerobic nitrogen cycling and CO2 fixation by the Calvin Cycle These data i support the apparent Mesoarchean trends of decreasing oxidant availability and methane cycling from onshore to offshore depositional environments ii provide further evidence that the Mosquito Creek Formation was deposited in a deep marine setting and iii contain 15N values that highlight the persistence of nitrogen fixation by Mo-nitrogenase and the dearth of aerobic nitrogen metabolisms in the Mesoarchean Chapter 3: Many paleoredox proxies indicate low-level and dynamic incipient oxygenation of Earth s surface environments during the Neoarchean 2 8-2 5 Ga prior to the Great Oxidation Event GOE at 2 4 Ga The mode tempo and scale of these redox changes are poorly understood because data from various locations and ages suggest both protracted and transient oxygenation Here we present bulk-rock and kerogen-bound nitrogen isotope ratios as well as bulk-rock selenium abundances and isotope ratios from drill-cores sampled at high stratigraphic resolution through the Jeerinah Formation 2 66 Ga Fortescue Group Western Australia to test for changes 138 in the redox state of the surface environment We find that both shallow and deep depositional facies in the Jeerinah Fm display episodes of positive primary 15N values ranging from 4 to 6 recording aerobic nitrogen cycling that requires free O2 in the upper water column Moderate selenium enrichments up to 5 4 ppm in the nearer-shore core may indicate coincident oxidative weathering of sulfide minerals on land though not to the extent seen in the younger Mt McRae Shale that records a well-documented whiff of atmospheric oxygen at 2 5 Ga Unlike the Mt McRae Shale Jeerinah selenium isotopes do not show a significant excursion concurrent with the positive 15N values Our data are thus most parsimoniously interpreted as evidence for transient surface ocean oxygenation lasting less than 50 Myr extending over hundreds of kilometers and occuring well before the GOE The nitrogen isotope data clearly record nitrification and denitrification providing the oldest firm evidence for these microbial metabolisms Chapter 4: Fixed nitrogen is an essential nutrient for eukaryotes As N2 fixation and assimilation of nitrate are catalyzed by metalloenzymes it has been hypothesized that in Mesoproterozoic oceans nitrate was limited in offshore environments by low trace metal concentrations and high rates of denitrification in anoxic and episodically euxinic deep water masses restricting eukaryotes to near-shore environments and limiting their evolutionary innovation To date this hypothesis has only been tested in the Belt Supergroup 1 4 Ga with results that support an onshore-offshore nitrate gradient as a potential control on eukaryote ecology Here we present bulk nitrogen and organic carbon isotopic data from non-isochronous cross-basinal facies across the Bangemall 1 5 Ga and the Roper 1 4-1 5 Ga basins to better understand the extent and variability of onshore-offshore nitrogen isotope gradients in the Mesoproterozoic Both basins show an average 1-2 enrichment in 15Nbulk from deep to shallow facies with a maximum range from -1 offshore to 7 5 onshore Unlike the Belt basin the Bangemall and Roper basins show some offshore 15Nbulk values that are enriched beyond the isotopic range induced by biological N2 fixation alone This suggests a mixture of aerobic and anaerobic metabolisms offshore In shallow waters where 15Nbulk enrichment peaks an aerobic nitrogen cycle was evidently operating vigorously Even though isotopic signatures of aerobic nitrogen cycling are seen in all parts of the Bangemall and Roper basins our data are consistent with a lateral gradient in nitrate availability within the photic zone with higher concentrations in near-shore environments than offshore The variability in 15Nbulk values in each depositional environment and the consistently low 15N values from Mesoproterozoic units compared to the 139 Paleoproterozoic and Neoproterozoic suggest that nitrate concentrations in the global ocean were likely low This trend is now seen in all three Mesoproterozoic basins so far examined and contrasts with the Paleoproterozoic and Neoproterozoic where nearly all 15Nbulk data plot above the N2 fixation window Thus we propose that the Mesoproterozoic ocean was characterized by a nitrate minimum between the Paleo- and Neoproterozoic with the lowest concentrations in offshore environments This inference is consistent with a Mesoproterozoic O2 decline following a temporary Paleoproterozoic O2 peak and it further supports the idea that nitrate limitation offshore may have contributed to the restriction of photosynthetic eukaryotes to near-shore environments delaying their rise to ecological dominance until the Neoproterozoic Era Chapter 5: Marine ecology co-evolved with dramatic environmental change across the Ordovician-Silurian boundary In particular apparent trends in nitrogen cycling both latitudinal and with respect to depositional environment suggest extreme spatial-temporal heterogeneity in the availability of dissolved-nitrogen species associated with global climate change and mass extinction Missing from these trends are analyses from the paleo-mid-latitudes and facies that represent the most distal continental margin Here we present nitrogen and carbon isotope abundance data from the O-S Global Boundary Stratotype Section and Point at Dob s Linn Scotland Nitrogen isotope and abundance data from bulk rock shale and bentonite samples as well as kerogen extractions indicate that unlike in tropical shallow-water sections the sub-tropical locale at Dob s Linn shows no changes in nitrogen cycling associated with the Hirnantian glaciation which suggests persistent nitrogen limitation Diminishing anoxia caused by increased ocean ventilation towards the end of the Ordovician was apparently biased towards shallow water tropical settings If perturbations to the nitrogen cycle in part caused the mass extinction and changes in climate then these results highlight the importance of the tropical inner-shelf marine biosphere in regulating the global environment Carbonate data from this and other sections suggest a deepening of the carbonate compensation depth during the Hirnantian also consistent with increased ocean ventilation This indicates Pacific-style responses of the CCD to glacial interglacial periods were operational across the O-S boundary and that an expansion of deep-sea carbonate deposition preservation could have constituted a large reactive reservoir for surface CO2 during these extreme changes in climate 140 BIBLIOGRAPHY 1 Ohmoto H Watanabe Y Ikemi H Poulson SR Taylor BE 2006 Sulphur isotope evidence for an oxic Archaean atmosphere Nature 442 7105 :908 911 2 Kaufman AJ et al 2007 Late Archean biospheric oxygenation and atmospheric evolution Science 317 5846 :1900 1903 3 Garvin J Buick R Anbar AD Arnold GL Kaufman AJ 2009 Isotopic Evidence for an Aerobic Nitrogen Cycle in the Latest Archean Science 323 5917 :1045 1048 4 Godfrey LV Falkowski PG 2009 The cycling and redox state of nitrogen in the Archaean ocean Nat Geosci 2 10 :725 729 5 St eken EE Buick R Anbar AD 2015 Selenium isotopes support free O-2 in the latest Archean Geology 43 3 :259 262 6 Anbar AD et al 2007 A Whiff of Oxygen Before the Great Oxidation Event Science 317 5846 :1903 1906 7 Duan Y et al 2010 Molybdenum isotope evidence for mild environmental oxygenation before the Great Oxidation Event Geochim Cosmochim Acta 74 23 :6655 6668 8 Kendall B Creaser RA Reinhard CT Lyons TW Anbar AD 2015 Transient episodes of mild environmental oxygenation and oxidative continental weathering during the late Archean Sci Adv 1 10 :e1500777 9 Kendall B Brennecka GA Weyer S Anbar AD 2013 Uranium isotope fractionation suggests oxidative uranium mobilization at 2 50 Ga Chem Geol 362:105 114 10 Kendall B et al 2010 Pervasive oxygenation along late Archaean ocean margins Nat Geosci 3 9 :647 652 11 Raiswell R et al 2011 Formation of syngenetic and early diagenetic iron minerals in the late Archean Mt McRae Shale Hamersley Basin Australia: New insights on the patterns controls and paleoenvironmental implications of authigenic mineral formation Geochim Cosmochim Acta 75 4 :1072 1087 12 Eickmann B et al 2018 Isotopic evidence for oxygenated Mesoarchaean shallow oceans Nat Geosci 11:133 138 13 Homann M Heubeck C Airo A Tice MM 2015 Morphological adaptations of 3 22 Ga-old tufted microbial mats to Archean coastal habitats Moodies Group Barberton Greenstone Belt South Africa Precambrian Res 266:47 64 14 Planavsky NJ et al 2014 Evidence for oxygenic photosynthesis half a billion years before the Great Oxidation Event Nat Geosci 7 4 :283 286 141 15 Riding R Fralick P Liang L 2014 Identification of an Archean marine oxygen oasis Precambrian Res 251:232 237 16 Crowe SA et al 2013 Atmospheric oxygenation three billion years ago Nature 501 7468 :535- 17 St eken EE Buick R 2018 Environmental control on microbial diversification and methane production in the Mesoarchean Precambrian Res 304:64 72 18 Wille M et al 2013 Mo-Cr isotope evidence for a reducing Archean atmosphere in 3 46-2 76 Ga black shales from the Pilbara Western Australia Chem Geol 340:68 76 19 Eriksson K Krapez B Fralick P 1994 Sedimentology of Archean Greenstone Belts - Signatures of Tectonic Evolution Earth-Sci Rev 37 1 2 :1 88 20 Nijman W Clevis Q de Vries ST 2010 The waning stage of a greenstone belt: The Mesoarchaean Mosquito Creek Basin of the East Pilbara Western Australia Precambrian Res 180 3 4 :251 271 21 Bagas L Bierlein FP Bodorkos S Nelson DR 2008 Tectonic setting evolution and orogenic gold potential of the late Mesoarchaean Mosquito Creek Basin North Pilbara Craton Western Australia Precambrian Res 160 3 4 :227 244 22 Kisch HJ Nijman W 2010 Metamorphic grade and gradient from white K-micas of Na-mica bearing sedimentary rocks in the Mosquito Creek Basin East Pilbara Craton Western Australia Precambrian Res 176 1 4 :11 26 23 Koehler MC Stueeken EE Kipp MA Buick R Knoll AH 2017 Spatial and temporal trends in Precambrian nitrogen cycling: A Mesoproterozoic offshore nitrate minimum Geochim Cosmochim Acta 198:315 337 24 St eken EE 2013 A test of the nitrogen-limitation hypothesis for retarded eukaryote radiation: Nitrogen isotopes across a Mesoproterozoic basinal profile Geochim Cosmochim Acta 120:121 139 25 Robl TL Davis BH 1993 Comparison of the HF-HCl and HF-BF3 maceration techniques and the chemistry of resultant organic concentrates Org Geochem 20 2 :249 255 26 Qi H Coplen TB Geilmann H Brand WA B hlke JK 2003 Two new organic reference materials for 13C and 15N measurements and a new value for the 13C of NBS 22 oil Rapid Commun Mass Spectrom 17 22 :2483 2487 27 Qi H Coplen TB Geilmann H Brand WA B hlke JK 2003 Two new organic reference materials for 13C and 15N measurements and a new value for the 13C of NBS 22 oil Rapid Commun Mass Spectrom 17 22 :2483 2487 28 Ader M et al 2016 Interpretation of the nitrogen isotopic composition of Precambrian sedimentary rocks: Assumptions and perspectives Chem Geol 429:93 110 29 St eken EE Zaloumis J Meixnerova J Buick R 2017 Differential metamorphic effects on nitrogen isotopes in kerogen extracts and bulk rocks Geochim Cosmochim Acta 217:80 94 142 30 Thomazo C Papineau D 2013 Biogeochemical Cycling of Nitrogen on the Early Earth Elements 9 5 :345 351 31 Altabet MA Francois R 1994 Sedimentary nitrogen isotopic ratio as a recorder for surface ocean nitrate utilization Glob Biogeochem Cycles 8 1 :103 116 32 Lehmann MF Bernasconi SM Barbieri A McKenzie JA 2002 Preservation of organic matter and alteration of its carbon and nitrogen isotope composition during simulated and in situ early sedimentary diagenesis Geochim Cosmochim Acta 66 20 :3573 3584 33 M bius J 2013 Isotope fractionation during nitrogen remineralization ammonification : Implications for nitrogen isotope biogeochemistry Geochim Cosmochim Acta 105:422 432 34 Thunell RC Sigman Daniel M Muller-Karger Frank Astor Yrene Varela Ramon 2004 Nitrogen isotope dynamics of the Cariaco Basin Venezuela Glob Biogeochem Cycles 18 3 doi:10 1029 2003GB002185 35 Fry B et al 1991 Stable Isotope Studies of the Carbon Nitrogen and Sulfur Cycles in the BlackSea and the Cariaco Trench Deep-Sea Res Part -Oceanogr Res Pap 38:S1003 S1019 36 Libes S Deuser W 1988 The Isotope Geochemistry of Particulate Nitrogen in the Peru Upwelling Area and the Gulf of Maine Deep-Sea Res Part -Oceanogr Res Pap 35 4 :517 533 37 Zhang X Sigman DM Morel FMM Kraepiel AML 2014 Nitrogen isotope fractionation by alternative nitrogenases and past ocean anoxia Proc Natl Acad Sci U S A 111 13 :4782 4787 38 St eken EE Kipp MA Koehler MC Buick R 2016 The evolution of Earth s biogeochemical nitrogen cycle Earth-Sci Rev 160:220 239 39 Teutsong T et al 2017 Petrography and geochemistry of the Mesoarchean Bikoula banded iron formation in the Ntem complex Congo craton Southern Cameroon: Implications for its origin Ore Geol Rev 80:267 288 40 Guy BM et al 2012 A multiple sulfur and organic carbon isotope record from nonconglomeratic sedimentary rocks of the Mesoarchean Witwatersrand Supergroup South Africa Precambrian Res 216:208 231 41 St eken EE Buick R Guy BM Koehler MC 2015 Isotopic evidence for biological nitrogen fixation by molybdenum-nitrogenase from 3 2 Gyr Nature 520 7549 :666-U178 42 Flannery DT Allwood AC Van Kranendonk MJ 2016 Lacustrine facies dependence of highly 13Cdepleted organic matter during the global age of methanotrophy Precambrian Res 285:216 241 43 Izon G et al 2015 Multiple oscillations in Neoarchaean atmospheric chemistry Earth Planet Sci Lett 431:264 273 44 Zerkle AL Claire MW Domagal-Goldman SD Farquhar J Poulton SW 2012 A bistable organicrich atmosphere on the Neoarchaean Earth Nat Geosci 5 5 :359 363 143 45 St eken EE et al 2017 Environmental niches and metabolic diversity in Neoarchean lakes Geobiology 15 6 :767 783 46 Lyons TW Reinhard CT Planavsky NJ 2014 The rise of oxygen in Earth s early ocean and atmosphere Nature 506 7488 :307 315 47 Bekker A et al 2004 Dating the rise of atmospheric oxygen Nature 427 6970 :117 120 48 Gumsley AP et al 2017 Timing and tempo of the Great Oxidation Event Proc Natl Acad Sci U S A 114 8 :1811 1816 49 Eigenbrode JL Freeman KH 2006 Late Archean rise of aerobic microbial ecosystems Proc Natl Acad Sci 103 43 :15759 15764 50 Gregory DD et al 2015 The chemical conditions of the late Archean Hamersley basin inferred from whole rock and pyrite geochemistry with Delta S-33 and delta S-34 isotope analyses Geochim Cosmochim Acta 149:223 250 51 Kurzweil F Wille M Schoenberg R Taubald H Van Kranendonk MJ 2015 Continuously increasing 98Mo values in Neoarchean black shales and iron formations from the Hamersley Basin Geochim Cosmochim Acta 164:523 542 52 Kurzweil F et al 2013 Atmospheric sulfur rearrangement 2 7 billion years ago: Evidence for oxygenic photosynthesis Earth Planet Sci Lett 366:17 26 53 Siebert C Kramers JD Meisel T Morel P Nagler TF 2005 PGE Re-Os and Mo isotope systematics in Archean and early Proterozoic sedimentary systems as proxies for redox conditions of the early Earth Geochim Cosmochim Acta 69 7 :1787 1801 54 St eken EE Catling DC Buick R 2012 Contributions to late Archaean sulphur cycling by life on land Nat Geosci 5 10 :722 725 55 Wille M et al 2007 Evidence for a gradual rise of oxygen between 2 6 and 2 5Ga from Mo isotopes and Re-PGE signatures in shales Geochim Cosmochim Acta 71 10 :2417 2435 56 Bristow LA et al 2016 Ammonium and nitrite oxidation at nanomolar oxygen concentrations in oxygen minimum zone waters Proc Natl Acad Sci U S A 113 38 :10601 10606 57 Bristow LA et al 2017 N-2 production rates limited by nitrite availability in the Bay of Bengal oxygen minimum zone Nat Geosci 10 1 :24 29 58 Casciotti KL 2009 Inverse kinetic isotope fractionation during bacterial nitrite oxidation Geochim Cosmochim Acta 73 7 :2061 2076 59 Robinson D 2001 15N as an integrator of the nitrogen cycle Trends Ecol Evol 16 3 :153 162 60 St eken EE et al 2015 The evolution of the global selenium cycle: Secular trends in Se isotopes and abundances Geochim Cosmochim Acta 162:109 125 144 61 Johnson TM Herbel MJ Bullen TD Zawislanski PT 1999 Selenium isotope ratios as indicators of selenium sources and oxyanion reduction Geochim Cosmochim Acta 63 18 :2775 2783 62 Murphy MA Sumner DY 2008 Tube structures of probable microbial origin in the Neoarchean Carawine Dolomite Hamersley Basin Western Australia Geobiology 6 1 :83 93 63 Nishizawa M et al 2010 Grain-scale iron isotopic distribution of pyrite from Precambrian shallow marine carbonate revealed by a femtosecond laser ablation multicollector ICP-MS technique: Possible proxy for the redox state of ancient seawater Geochim Cosmochim Acta 74 9 :2760 2778 64 Rasmussen B Blake TS Fletcher IR 2005 U-Pb zircon age constraints on the Hamersley spherule beds: Evidence for a single 2 63 Ga Jeerinah-Carawine impact ejecta layer Geology 33 9 :725 728 65 French KL et al 2015 Reappraisal of hydrocarbon biomarkers in Archean rocks Proc Natl Acad Sci U S A 112 19 :5915 5920 66 Peters CA Piazolo S Webb GE Dutkiewicz A George SC 2016 In search of early life: Carbonate veins in Archean metamorphic rocks as potential hosts of biomarkers Earth Planet Sci Lett 453:44 55 67 Lanari P Wagner T Vidal O 2014 A thermodynamic model for di-trioctahedral chlorite from experimental and natural data in the system MgO FeO Al 2 O 3 SiO 2 H 2 O: applications to P T sections and geothermometry Contrib Mineral Petrol 167 2 :968 68 Vidal O Lanari P Munoz M Bourdelle F De Andrade V 2016 Deciphering temperature pressure and oxygen-activity conditions of chlorite formation Clay Miner 51 4 :615 633 69 Large RR et al 2014 Trace element content of sedimentary pyrite as a new proxy for deep-time ocean-atmosphere evolution Earth Planet Sci Lett 389:209 220 70 Haendel D Muhle K Nitzsche H Stiehl G Wand U 1986 Isotopic Variations of the Fixed Nitrogen in Metamorphic Rocks Geochim Cosmochim Acta 50 5 :749 758 71 Jia Y 2006 Nitrogen isotope fractionations during progressive metamorphism: A case study from the Paleozoic Cooma metasedimentary complex southeastern Australia Geochim Cosmochim Acta 70 20 :5201 5214 72 Godfrey LV Poulton SW Bebout GE Fralick PW 2013 Stability of the nitrogen cycle during development of sulfidic water in the redox-stratified late Paleoproterozoic Ocean Geology 41 6 :655 658 73 Teitler Y et al 2015 Ubiquitous occurrence of basaltic-derived paleosols in the Late Archean Fortescue Group Western Australia Precambrian Res 267:1 27 74 Papineau D et al 2009 High primary productivity and nitrogen cycling after the Paleoproterozoic phosphogenic event in the Aravalli Supergroup India Precambrian Res 171 1 4 :37 56 145 75 Thomazo C Ader M Philippot P Extreme 15N-enrichments in 2 72-Gyr-old sediments: evidence for a turning point in the nitrogen cycle Geobiology 9 2 :107 120 76 Sigman DM Karsh KL Casciotti KL 2009 Nitrogen Isotopes in the Ocean Encyclopedia of Ocean Sciences Second Edition ed Steele JH Academic Press Oxford pp 40 54 77 Rouxel O Ludden J Carignan J Marin L Fouquet Y 2002 Natural variations of Se isotopic composition determined by hydride generation multiple collector inductively coupled plasma mass spectrometry Geochim Cosmochim Acta 66 18 :3191 3199 78 Olson SL Kump LR Kasting JF 2013 Quantifying the areal extent and dissolved oxygen concentrations of Archean oxygen oases Chem Geol 362:35 43 79 Lalonde SV Konhauser KO 2015 Benthic perspective on Earth s oldest evidence for oxygenic photosynthesis Proc Natl Acad Sci 112 4 :995 1000 80 Merchant SS Helmann JD 2012 Elemental Economy: Microbial Strategies for Optimizing Growth in the Face of Nutrient Limitation Advances in Microbial Physiology Vol 60 ed Poole RK Academic Press Ltd-Elsevier Science Ltd London pp 91 210 81 Huertas MJ L pez-Maury L Giner-Lamia J S nchez-Riego AM Florencio FJ 2014 Metals in Cyanobacteria: Analysis of the Copper Nickel Cobalt and Arsenic Homeostasis Mechanisms Life 4 4 :865 886 82 Pavlov AA Kasting JF Eigenbrode JL Freeman KH 2001 Organic haze in Earth s early atmosphere: Source of low-C-13 Late Archean kerogens Geology 29 11 :1003 1006 83 Arney G et al 2016 The Pale Orange Dot: The Spectrum and Habitability of Hazy Archean Earth Astrobiology 16 11 :873 899 84 Ono S et al 2003 New insights into Archean sulfur cycle from mass-independent sulfur isotope records from the Hamersley Basin Australia Earth Planet Sci Lett 213 1 2 :15 30 85 Scott CT et al 2011 Late Archean euxinic conditions before the rise of atmospheric oxygen Geology 39 2 :119 122 86 Nelson DR Trendall AF Altermann W 1999 Chronological correlations between the Pilbara and Kaapvaal cratons Precambrian Res 97 3 4 :165 189 87 Arndt N Nelson D Compston W Trendall A Thorne A 1991 The Age of the Fortescue Group Hamersley Basin Western-Australia from Ion Microprobe Zircon U-Pb Results Aust J Earth Sci 38 3 :261 281 88 van Kessel MAHJ et al 2015 Complete nitrification by a single microorganism Nature 528 7583 :555- 89 Daims H et al 2015 Complete nitrification by Nitrospira bacteria Nature 528 7583 :504- 146 90 St eken EE Foriel J Nelson BK Buick R Catling DC 2013 Selenium isotope analysis of organicrich shales: advances in sample preparation and isobaric interference correction J Anal At Spectrom 28 11 :1734 1749 91 WAIS Divide Project Members et al 2013 Onset of deglacial warming in West Antarctica driven by local orbital forcing Nature 500 7463 :440 444 92 Anbar AD Knoll AH 2002 Proterozoic ocean chemistry and evolution: A bioinorganic bridge Science 297 5584 :1137 1142 93 Javaux EJ Knoll AH Walter MR 2001 Morphological and ecological complexity in early eukaryotic ecosystems Nature 412 6842 :66 69 94 Buick R Knoll AH 1999 Acritarchs and microfossils from the Mesoproterozoic Bangemall Group northwestern Australia J Paleontol 73 5 :744 764 95 Arnold GL Anbar AD Barling J Lyons TW 2004 Molybdenum isotope evidence for widespread anoxia in mid-proterozoic oceans Science 304 5667 :87 90 96 Canfield DE 1998 A new model for Proterozoic ocean chemistry Nature 396 6710 :450 453 97 Planavsky NJ et al 2014 Low Mid-Proterozoic atmospheric oxygen levels and the delayed rise of animals Science 346 6209 :635 638 98 Reinhard CT et al 2013 Proterozoic ocean redox and biogeochemical stasis Proc Natl Acad Sci U S A 110 14 :5357 5362 99 Shen Y Knoll AH Walter MR 2003 Evidence for low sulphate and anoxia in a mid-Proterozoic marine basin Nature 423 6940 :632 635 100 Sperling EA et al 2015 Statistical analysis of iron geochemical data suggests limited late Proterozoic oxygenation Nature 523 7561 :451 454 101 Godfrey LV Glass JB 2011 Chapter twenty-two - The Geochemical Record of the Ancient Nitrogen Cycle Nitrogen Isotopes and Metal Cofactors Methods in Enzymology Research on Nitrification and Related Processes Part A ed Klotz MG Academic Press pp 483 506 102 Fuchsman CA Murray JW Konovalov SK 2008 Concentration and natural stable isotope profiles of nitrogen species in the Black Sea Mar Chem 111 1 2 :90 105 103 Lindell D Post A 1995 Ultraphytoplankton Succession Is Triggered by Deep Winter Mixing in the Gulf-of-Aqaba eilat Red-Sea Limnol Oceanogr 40 6 :1130 1141 104 Latasa M Bidigare RR 1998 A comparison of phytoplankton populations of the Arabian Sea during the Spring Intermonsoon and Southwest Monsoon of 1995 as described by HPLC-analyzed pigments Deep-Sea Res Part Ii-Top Stud Oceanogr 45 10 11 :2133 2170 147 105 Karl DM Bidigare RR Letelier RM 2001 Long-term changes in plankton community structure and productivity in the North Pacific Subtropical Gyre: The domain shift hypothesis Deep-Sea Res Part Ii-Top Stud Oceanogr 48 8 9 :1449 1470 106 Bouman HA et al 2011 Water-column stratification governs the community structure of subtropical marine picophytoplankton Environ Microbiol Rep 3 4 :473 482 107 Fawcett SE Lomas M Casey JR Ward BB Sigman DM 2011 Assimilation of upwelled nitrate by small eukaryotes in the Sargasso Sea Nat Geosci 4 10 :717 722 108 Poulin FJ Franks PJS 2010 Size-structured planktonic ecosystems: constraints controls and assembly instructions J Plankton Res 32 8 :1121 1130 109 Farquhar J Zerkle AL Bekker A 2011 Geological constraints on the origin of oxygenic photosynthesis Photosynth Res 107 1 :11 36 110 Czaja AD et al 2010 Iron and carbon isotope evidence for ecosystem and environmental diversity in the similar to 2 7 to 2 5 Ga Hamersley Province Western Australia Earth Planet Sci Lett 292 1 2 :170 180 111 Reinhard CT Raiswell R Scott C Anbar AD Lyons TW 2009 A Late Archean Sulfidic Sea Stimulated by Early Oxidative Weathering of the Continents Science 326 5953 :713 716 112 Buick R 2007 Did the Proterozoic Canfield Ocean cause a laughing gas greenhouse Geobiology 5 2 :97 100 113 Busigny V Lebeau O Ader M Krapez B Bekker A 2013 Nitrogen cycle in the Late Archean ferruginous ocean Chem Geol 362:115 130 114 Yamaguchi KE 2002 Geochemistry of Archean-Paleoproterozoic black shales: the early evolution of the atmosphere oceans and biosphere PhD Diss Pa State Univ:485 pp 115 Kump LR et al 2011 Isotopic Evidence for Massive Oxidation of Organic Matter Following the Great Oxidation Event Science 334 6063 :1694 1696 116 Bekker A Holland HD 2012 Oxygen overshoot and recovery during the early Paleoproterozoic Earth Planet Sci Lett 317:295 304 117 Planavsky NJ Bekker A Hofmann A Owens JD Lyons TW 2012 Sulfur record of rising and falling marine oxygen and sulfate levels during the Lomagundi event Proc Natl Acad Sci U S A 109 45 :18300 18305 118 Hardisty DS et al 2014 An iodine record of Paleoproterozoic surface ocean oxygenation Geology 42 7 :619 622 119 Scott C et al 2014 Pyrite multiple-sulfur isotope evidence for rapid expansion and contraction of the early Paleoproterozoic seawater sulfate reservoir Earth Planet Sci Lett 389:95 104 148 120 Frey C Hietanen S Juergens K Labrenz M Voss M 2014 N and O Isotope Fractionation in Nitrate during Chemolithoautotrophic Denitrification by Sulfurimonas gotlandica Environ Sci Technol 48 22 :13229 13237 121 Brunner B et al 2013 Nitrogen isotope effects induced by anammox bacteria Proc Natl Acad Sci 110 47 :18994 18999 122 Fulton JM Arthur MA Freeman KH 2012 Black Sea nitrogen cycling and the preservation of phytoplankton delta N-15 signals during the Holocene Glob Biogeochem Cycles 26:GB2030 123 Kessler AJ et al 2014 The isotope effect of denitrification in permeable sediments Geochim Cosmochim Acta 133:156 167 124 Sigman DM et al 2009 The dual isotopes of deep nitrate as a constraint on the cycle and budget of oceanic fixed nitrogen Deep-Sea Res Part -Oceanogr Res Pap 56 9 :1419 1439 125 Ader M et al 2014 Ocean redox structure across the Late Neoproterozoic Oxygenation Event: A nitrogen isotope perspective Earth Planet Sci Lett 396:1 13 126 Saito MA Sigman DM Morel FMM 2003 The bioinorganic chemistry of the ancient ocean: the co-evolution of cyanobacterial metal requirements and biogeochemical cycles at the ArcheanProterozoic boundary Inorganica Chim Acta 356:308 318 127 Zerkle AL House CH Cox RP Canfield DE 2006 Metal limitation of cyanobacterial N-2 fixation and implications for the Precambrian nitrogen cycle Geobiology 4 4 :285 297 128 Glass JB Axler RP Chandra S Goldman CR 2012 Molybdenum limitation of microbial nitrogen assimilation in aquatic ecosystems and pure cultures Front Microbiol 3:331 129 Boyle RA et al 2013 Nitrogen cycle feedbacks as a control on euxinia in the mid-Proterozoic ocean Nat Commun 4:1533 130 Canfield DE Teske A 1996 Late Proterozoic rise in atmospheric oxygen concentration inferred from phylogenetic and sulphur-isotope studies Nature 382 6587 :127 132 131 Thomson D Rainbird RH Planavsky N Lyons TW Bekker A 2015 Chemostratigraphy of the Shaler Supergroup Victoria Island NW Canada: A record of ocean composition prior to the Cryogenian glaciations Precambrian Res 263:232 245 132 Martin DM Thorne AM 2004 Tectonic setting and basin evolution of the Bangemall Supergroup in the northwestern Capricorn Orogen Precambrian Res 128 3 :385 409 133 Martin DM Sircombe KN Thorne AM Cawood PA Nemchin AA 2008 Provenance history of the Bangemall Supergroup and implications for the Mesoproterozoic paleogeography of the West Australian Craton Precambrian Res 166 1 :93 110 134 Zi J-W et al 2015 In situ U Pb geochronology of xenotime and monazite from the Abra polymetallic deposit in the Capricorn Orogen Australia: Dating hydrothermal mineralization and fluid flow in a long-lived crustal structure Precambrian Res 260:91 112 149 135 Buick R Des Marais DJ Knoll AH 1995 Stable isotopic compositions of carbonates from the Mesoproterozoic Bangemall group northwestern Australia Chem Geol 123 1 :153 171 136 Martin D 2004 Peperite in the Backdoor Formation and its significance to the age and tectonic evolution of the Bangemall Supergroup pp 53 59 137 Wingate M 2002 Age and palaeomagnetism of dolerite sills intruded into the Bangemall Supergroup on the Edmund 1:250 000 map sheet Western Australia Geological Survey of Western Australia Available at: http: research-repository uwa edu au en publications age-andpalaeomagnetism-of-dolerite-sills-intruded-into-the-bangemall-supergroup-on-the-edmund1250-000-map-sheet-western-australia cb47aeb7-4016-4d3e-a479-4838d657db91 html Accessed June 13 2018 138 Page RW Jackson MJ Krassay AA 2000 Constraining sequence stratigraphy in north Australian basins: SHRIMP U Pb zircon geochronology between Mt Isa and McArthur River Aust J Earth Sci 47 3 :431 459 139 Kendall B Creaser RA Gordon GW Anbar AD 2009 Re-Os and Mo isotope systematics of black shales from the Middle Proterozoic Velkerri and Wollogorang Formations McArthur Basin northern Australia Geochim Cosmochim Acta 73 9 :2534 2558 140 Kralik M 1982 Rb Sb age determinations on Precambrian carbonate rocks of the Carpentarian McArthur basin Northern Territories Australia Precambrian Res 18 1 :157 170 141 Jackson MJ Raiswell R 1991 Sedimentology and carbon-sulphur geochemistry of the Velkerri Formation a mid-Proterozoic potential oil source in northern Australia Precambrian Res 54 1 :81 108 142 Abbott ST Sweet IP 2000 Tectonic control on third-order sequences in a siliciclastic ramp-style basin: an example from the Roper Superbasin Mesoproterozoic northern Australia Aust J Earth Sci 47 3 :637 657 143 Johnston DT et al 2008 Sulfur isotope biogeochemistry of the Proterozoic McArthur Basin Geochim Cosmochim Acta 72 17 :4278 4290 144 Cox GM et al 2016 Basin redox and primary productivity within the Mesoproterozoic Roper Seaway Chem Geol 440:101 114 145 Jackson J- Sweet IP Powell TG 1988 STUDIES ON PETROLEUM GEOLOGY AND GEOCHEMISTRY MIDDLE PROTEROZOIC McARTHUR BASIN NORTHERN AUSTRALIA I: PETROLEUM POTENTIAL APPEA J 28 1 :283 302 146 Peat CJ Muir MD Plumb KA McKirdy DM Norvick MS 1978 Proterozoic microfossils from the Roper Group Northern Territory Australia 17 147 Javaux EJ Knoll AH Walter MR 2004 TEM evidence for eukaryotic diversity in mid-Proterozoic oceans Geobiology 2 3 :121 132 150 148 Summons RE Powell TG Boreham CJ 1988 Petroleum geology and geochemistry of the Middle Proterozoic McArthur Basin Northern Australia: III Composition of extractable hydrocarbons Geochim Cosmochim Acta 52 7 :1747 1763 149 Dutkiewicz A Volk H Ridley J George S 2003 Biomarkers brines and oil in the Mesoproterozoic Roper Superbasin Australia Geology 31 11 :981 150 Dutkiewicz A Volk H Ridley J George SC 2004 Geochemistry of oil in fluid inclusions in a middle Proterozoic igneous intrusion: implications for the source of hydrocarbons in crystalline rocks Org Geochem 35 8 :937 957 151 Volk H George SC Dutkiewicz A Ridley J 2005 Characterisation of fluid inclusion oil in a MidProterozoic sandstone and dolerite Roper Superbasin Australia Chem Geol 223 1 :109 135 152 Siljestr m S et al 2013 Analysis of single oil-bearing fluid inclusions in mid-Proterozoic sandstones Roper Group Australia Geochim Cosmochim Acta 122:448 463 153 Flannery EN George SC 2014 Assessing the syngeneity and indigeneity of hydrocar