Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Chromium isotopes, iron speciation, and the evolution of Earth's surface chemistry through time Bauer, Kohen Witt 2019

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2019_september_bauer_kohen.pdf [ 67.13MB ]
JSON: 24-1.0379441.json
JSON-LD: 24-1.0379441-ld.json
RDF/XML (Pretty): 24-1.0379441-rdf.xml
RDF/JSON: 24-1.0379441-rdf.json
Turtle: 24-1.0379441-turtle.txt
N-Triples: 24-1.0379441-rdf-ntriples.txt
Original Record: 24-1.0379441-source.json
Full Text

Full Text

Chromium isotopes, iron speciation, andthe evolution of Earth’s surface chemistrythrough timebyKohen Witt BauerMSc, University of Toronto, 2013BSc, Queens University, 2011A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinThe Faculty of Graduate and Postdoctoral Studies(Geological Sciences)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)June 2019c© Kohen Witt Bauer , 2019The following individuals certify that they have read, and recommend to the Faculty of Graduateand Postdoctoral Studies for acceptance, the dissertation entitled:Chromium isotopes, iron speciation and the evolution of Earth’s surface chemistry through timeSubmitted by Kohen W. Bauer in partial fulfillment of the requirements for the degree of Doctorof Philosophy in Geological SciencesExamining Committee:Dr. Sean A. Crowe, Microbiology and Immunology & Earth, Ocean and Atmospheric SciencesSupervisorDr. Roger Francois, Earth, Ocean and Atmospheric SciencesSupervisorDr. Mark JellinekSupervisory Committee MemberDr. Gregory DippleUniversity ExaminerDr. Simon DonnerUniversity ExaminerDr. Michael TenzerChairAdditional Supervisory Committee Members:Dr. Mark Jellinek, Earth, Ocean and Atmospheric SciencesSupervisory Committee MemberDr. Stephen Calvert, Earth, Ocean and Atmospheric SciencesSupervisory Committee MemberiiAbstractThe oxygen concentration of the ocean atmosphere system regulates the nature, activity anddiversity of life on Earth. Atmospheric and ocean oxygenation is tightly coupled to the globalbiogeochemical cycles of C, N, P, S and Fe, as well as climate. Reconstructing the historyof oxygen on planet Earth, therefore, is a key component to understanding the evolution oflife. Our emergent picture of the evolution of Earth’s surface redox state with its links tothe evolution of life and climate relies heavily on interpretations of geochemical informationpreserved in the rock record. The Cr isotope and Fe-speciation proxies are two widely appliedtools used to diagnose redox conditions in both modern and ancient depositional environments.Many aspects of the precise mechanisms that lend the use of these two transition metals aspaleoredox proxies, however, remain unclear, confounding accurate reconstructions of paleo-oxygen concentrations that rely on Cr isotope and Fe-speciation data. In this work I studied Crisotope and Fe speciation proxy systematics to develop more nuanced frameworks for how thesetwo paleoredox proxies may be employed to reconstruct depositional redox states in both modernand past environments. I determined the Cr isotope and Fe mineral composition of modernmarine hydrothermal sediments, revealing Cr isotope fractionations that imply deposition from anoxygenated deep ocean. I determined Cr isotope fractionations associated with the reduction ofCr(VI) in modern ferruginous sediments, revealing that the magnitude of Cr isotope fractionationin such environments is linked to the speciation of Fe and the oxygen penetration depth of thesediments. I determined Fe-speciation and trace metal abundances of sediments deposited duringoceanic anoxic event 1a (OAE1a), revealing that during this interval the oceans were anoxicand Fe-rich (ferruginous) for more than 1 million years. Lastly, I determined the Fe-speciationof suspended and sedimented material from two modern ferruginous lakes, revealing that themineral magnetite forms authigenically in the ferruginous water columns. This new knowledgeof Cr and Fe proxy systematics will allow for more refined interpretations of paleo oxygenconcentrations based on Cr isotope and Fe-speciation signals captured in the rock record throughtime.iiiLay SummaryToday, oxygen concentrations in Earth’s atmosphere are high enough to support complex animallife – including perhaps most obviously, humans. Atmospheric oxygen concentrations high enoughto sustain animal life, however, are a relatively new development in Earth’s history, and thereis much evidence to show that oxygen concentrations at Earth’s surface have fluctuated greatlythrough time. Reconstructing the evolutionary history of oxygen, therefore, is a key componentto understanding the evolution of life. Earth’s rock record serves as an archive of the geologicpast, and preserved within are geochemical clues as to what past oxygen concentrations in thebiosphere were like. In this work I studied the geochemistry of Cr and Fe in modern and ancientrocks and sediments, to create new knowledge on the implementation of Cr and Fe as paleoredoxproxies that will allow for more nuanced reconstructions of the complex history of oxygen inEarth’s surface environments.ivPrefaceThis work was made possible through the contributions and dedication of many collaborators. Dr.Sean Crowe and Roger Francois, as the research advisors were involved in all aspects of this workincluding experimental design, data analysis and interpretation and writing. Sections of this workare partly and wholly published, in press, or in review. Copyright licenses were obtained and arelisted below.• Chapter 1: Kohen W. Bauer wrote the main text with editorial support from Ashley Davidson,Roger Francois and Sean A. Crowe. A review manuscript is in preparation.• Chapter 2: Kohen W. Bauer wrote the main text with editorial support from Sean A. Croweand Roger Francois. Kohen W. Bauer, Sean A. Crowe and Roger Francois designed theresearch. Simon Poulton collected samples. Kohen W. Bauer, Dan Asael and Devon Coleperformed the laboratory work. Kohen W. Bauer, Noah Planavsky, Steve Calvert and Sean A.Crowe analyzed and interpreted the data. Editorial support was received from the entire listof authors. The manuscript is in review.• Chapter 3: Kohen W. Bauer wrote the main text with editorial support from Sean A. Crowe.Sean A. Crowe and Roger Francois designed the research. Kohen W. Bauer and Sean A.Crowe collected samples. Kohen W. Bauer, Devon Cole and Bleuenn Gueguen performedthe laboratory work. Kohen W. Bauer, Noah Planavsky, Jens Kallmeyer and Sean A. Croweanalyzed and interpreted the data. Editorial support was received from the entire list ofauthors. The reference for the published manuscript can be found as follows:Bauer, Kohen W., et al. ”Chromium isotope fractionation in ferruginous sediments.”Geochimica et Cosmochimica Acta 223 (2018): 198-215.• Chapter 4: Kohen W. Bauer wrote the main text with editorial support from Sean A. Croweand Roger Francois. Kohen W. Bauer designed the research. Elisabetta Erba and CinziaBottini collected samples. Kohen W. Bauer performed the laboratory work. Kohen W. Bauer,Sergei Katsev, Mark Jellinek and Sean A. Crowe analyzed and interpreted the data. Editorialsupport was received from the entire list of authors. The manuscript is in review.• Chapter 5: Kohen W. Bauer wrote the main text with editorial support from Sean A.Crowe and Roger Francois. Sean A. Crowe and Kohen W. Bauer designed the research.Kohen W. Bauer, Sean A. Crowe, Rachel Simister, Aurele Vuillemin, Andre Friese, JensvKallmeyer and Celine Michiels collected samples. Kohen W. Bauer, Matthijs Smit, and JamesByrne performed the laboratory work. Kohen W. Bauer and Sean A. Crowe analyzed andinterpreted the data. Editorial support was received from the entire list of authors. Themanuscript is in review.• Chapter 6: Kohen W. Bauer wrote the main text.Throughout this dissertation the word ’we’ refers to Kohen W. Bauer unless otherwise stated.None of the work encompassing this dissertation required consultation with the UBC ResearchEthics Board.viTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiiAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Oxygenation history of the Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Chromium biogeochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2.1 Chromium speciation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.2.2 Chromium redox reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.3 The Cr isotope paleoredox proxy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.3.1 Oxidative weathering of crustally bound Cr(III) . . . . . . . . . . . . . . . . . 91.3.2 Riverine delivery of Cr to the oceans . . . . . . . . . . . . . . . . . . . . . . . 101.3.3 Hydrothermally sourced Cr(III) . . . . . . . . . . . . . . . . . . . . . . . . . . 121.3.4 Chromium in seawater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131.3.5 Reduction of seawater Cr(VI) and burial of Cr(III) in marine sediments . . . 151.4 The Fe-speciation paleoredox proxy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161.4.1 Fe geochemistry in Earth’s surface environments . . . . . . . . . . . . . . . . 161.5 Fe-speciation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181.6 Dissertation overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20vii2 Chromium isotopes in marine hydrothermal sediments . . . . . . . . . . . . . . . . . . 232.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262.2.1 Chemical extraction techniques . . . . . . . . . . . . . . . . . . . . . . . . . . 272.2.2 Analytical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292.2.3 XRD analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302.2.4 Cr purification and isotope ratio determination . . . . . . . . . . . . . . . . . 302.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322.3.1 Mineralogical analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322.3.2 Site 598 elemental concentrations and Cr mass balance . . . . . . . . . . . . 332.3.3 Mn-oxide concentrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392.3.4 Chromium isotopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422.4.1 Cr in the carbonate phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422.4.2 Cr isotope composition of oxygenated deep-sea sediments and their detritalinputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432.4.3 Cr(III) oxidation in serpentinizing environments . . . . . . . . . . . . . . . . 462.4.4 Reduction of Cr(VI) in open vs. closed systems . . . . . . . . . . . . . . . . . 472.4.5 Diagenetic oxidation of Cr(III) . . . . . . . . . . . . . . . . . . . . . . . . . . . 502.4.6 Isotopically light Cr in hydrothermal sediments . . . . . . . . . . . . . . . . 522.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543 Chromium isotope fractionation in ferruginous sediments . . . . . . . . . . . . . . . . 553.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603.2.1 Sample location and experimental framework . . . . . . . . . . . . . . . . . . 603.2.2 Whole core incubation experiment . . . . . . . . . . . . . . . . . . . . . . . . 603.2.3 Laboratory sediment slurry experiment . . . . . . . . . . . . . . . . . . . . . 613.2.4 Sediment leaching experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . 623.2.5 Chromium isotope methodology . . . . . . . . . . . . . . . . . . . . . . . . . 623.2.6 Modelling Cr isotope fractionation . . . . . . . . . . . . . . . . . . . . . . . . 643.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 693.3.1 Whole core incubations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 693.3.2 Sediment leach experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 733.3.3 Slurry experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 733.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 773.4.1 Modelling Cr(VI) reduction and isotope fractionation . . . . . . . . . . . . . 773.4.2 Comparison of Cr isotope fractionations obtained in slurries and core incu-bations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81viii3.4.3 Implications for the chromium isotope record . . . . . . . . . . . . . . . . . . 853.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 904 Ferruginous oceans during OAE1a and the collapse of the seawater sulphate reservoir 914.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 924.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 934.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 934.3.1 Ferruginous conditions during OAE1a . . . . . . . . . . . . . . . . . . . . . . 934.3.2 Modelling Aptian seawater sulphate concentrations . . . . . . . . . . . . . . 974.3.3 Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1024.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035 Magnetite biomineralization in ferruginous waters and early Earth evolution . . . . . 1045.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1055.2 Fe-speciation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1085.3 Magnetite formation, Fe reduction, and Fe recycling . . . . . . . . . . . . . . . . . . 1125.4 Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1146 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1176.1 Summary of the dissertation and key research findings . . . . . . . . . . . . . . . . 1176.1.1 Cr isotopes recorded in marine hydrothermal sediments . . . . . . . . . . . 1176.1.2 Cr isotope fractionation in modern and ancient ferruginous environments . 1196.1.3 Dynamics in Earth surface oxidant budgets during OAE1a . . . . . . . . . . 1206.1.4 Magnetite formation in modern and ancient ferruginous environments . . . 121Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123AppendicesA Chapter 3: supplementary material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142A.1 Cr isotope model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142A.1.1 Oxygen model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145B Chapter 4: supplemental material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147B.1 Geological setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147B.1.1 Cismon drill core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147B.1.2 DSDP site 463 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147B.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147B.2.1 Fe-speciation analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147B.3 Evaluating Fe-speciation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153ixB.3.1 Dilution of authigenic Fe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153B.4 Possible post depositional shale alteration . . . . . . . . . . . . . . . . . . . . . . . . 154B.4.1 Oxidation of reactive Fe(II) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154B.5 Idealized 1D water column reaction transport model . . . . . . . . . . . . . . . . . . 155B.6 Diagenetic model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159B.7 Evaporite deposition model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161B.8 Marine sulphur budgets and sulphate drawdown . . . . . . . . . . . . . . . . . . . . 162C Chapter 5: supplemental material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167C.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167C.1.1 Sample collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167C.1.2 Confirming the selectivity of oxalate extractable Fe . . . . . . . . . . . . . . . 169C.1.3 XRD analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171C.1.4 Fe flux calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171C.1.5 Saturation state calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171C.1.6 SEM microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172C.1.7 Raman microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175xList of Tables1.1 Cr river, groundwater and estuary fluxes . . . . . . . . . . . . . . . . . . . . . . . . . 112.1 Summary of extraction techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282.2 Mineralogical analyses of the sediments from site 598 . . . . . . . . . . . . . . . . . . 332.3 Element concentrations and Cr isotope composition determined in the bulk (L6NHClleach) site 598 sediments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352.4 Element concentrations determined via sequential extractions in site 598 sediments 382.5 Mn-oxide:Fe molar ratios in the upper 20 cm of the three samples sites . . . . . . . 402.6 Compilation of modern oxic pelagic sediment Cr isotope values . . . . . . . . . . . 442.7 Compilation of fraction factors during reduction of Cr(VI) adapted from [1] . . . . 493.1 Whole core incubation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 713.2 Rayleigh model parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 713.3 Cr isotope composition of the 0.5 M HCl extractable pool in upper 0.5 cm ofsediment of two duplicate cores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 733.4 Slurry experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 743.5 1D diffusion model parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 783.6 Compilation of experimental isotope fractionations for this study . . . . . . . . . . . 823.7 Compilation of isotope fractionations during reduction of Cr(VI) adapted from [1] . 833.8 Fe speciation data for the upper 0.5 cm of sediment in Lake Matano . . . . . . . . . 835.1 Fe fluxes in the Malili lakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109A.1 A range of boundary conditions used in the oxygen model . . . . . . . . . . . . . . . 146B.1 Description of revised Fe-speciation leach procedure . . . . . . . . . . . . . . . . . . 148B.2 Cismon Core Fe-speciation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149B.3 DSDP Site 463 Fe-speciation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150B.4 Cismon Core Fe-speciation. Canonical extraction . . . . . . . . . . . . . . . . . . . . 151B.5 DSDP site 463 Fe-speciation. Canonical extraction . . . . . . . . . . . . . . . . . . . . 152B.6 Reactions in sediment diagenetic model . . . . . . . . . . . . . . . . . . . . . . . . . . 159B.7 Parameters of diagenetic model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160B.8 Diagenetic model results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160xiB.9 Evaportation model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161B.10 Box model S-fluxes and isotopic compositions . . . . . . . . . . . . . . . . . . . . . . 163B.11 Range of coupled δ34SCAS - δ34Spyr data for OAE1a . . . . . . . . . . . . . . . . . . . 164C.1 Description of Fe-speciation extractions . . . . . . . . . . . . . . . . . . . . . . . . . . 168C.2 Water column Fe-speciation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169C.3 Results of qXRD phase analysis of LT deep sediments (wt.%) . . . . . . . . . . . . . 170xiiList of Figures1.1 The modern chromium biogeochemical cycle . . . . . . . . . . . . . . . . . . . . . . . 41.2 The reactivity of Fe species targeted by the Fe-speciation sequential extraction . . . 172.1 Leach procedure schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272.2 Two end member-mixing diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342.3 Mass balance of Cr in the site 598 sediments . . . . . . . . . . . . . . . . . . . . . . . 362.4 Element concentrations at site 598 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372.5 Cr:Fe composition of vent particles and sediments deposited at site 598 . . . . . . . 392.6 Cr-isotope composition of sediments deposited at site 598 . . . . . . . . . . . . . . . 412.7 Cr cycling in an anoxic vs. oxic deep ocean . . . . . . . . . . . . . . . . . . . . . . . . 533.1 A schematic showing how the depth of the non-reactive zone (NRZ) controlseffective isotope fractionation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663.2 Cr(VI) Concentration (µmol l– 1) vs. Time (hours) during whole-core incubationexperiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 703.3 δ53Cr values (h) for whole-core incubation experiments . . . . . . . . . . . . . . . . 723.4 Cr concentration (µmol l– 1) as a function of time (hours) during slurry experiments 743.5 0.5 M HCl extractable Fe(II) concentrations (µmol l– 1) through time (hours) duringthe slurry experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 753.6 Variation of δ53Cr values (h) of the slurry experiments through time (hours) . . . . 763.7 Modelling of porewater δ53Cr(VI) values through time using different appliedintrinsic fractionation factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803.8 Sensitivity analysis for our steady state model . . . . . . . . . . . . . . . . . . . . . . 863.9 Oxygen penetration depth model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 884.1 Fe-speciation and carbon isotope records for Cismon and DSDP Site 463 . . . . . . . 944.2 Fe-speciation and Fe/Al records of the Cismon (left panels) and DSDP Site 463(right panels) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 954.3 Models of marine sulphur cycling during OAE1a . . . . . . . . . . . . . . . . . . . . 984.4 Evolution of the Cretaceous seawater sulphate reservoir . . . . . . . . . . . . . . . . 1005.1 Chemical and physical properties of LM and LT . . . . . . . . . . . . . . . . . . . . . 107xiii5.2 Malili lake Fe-speciation and magnetite fluxes . . . . . . . . . . . . . . . . . . . . . . 1085.3 Authigenic magnetite morphologies in the Malili lakes . . . . . . . . . . . . . . . . . 111A.1 Measured pore water oxygen profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . 142A.2 Hypothetical rate distribution determined from Cr(VI) and Fe(II) concentrationprofiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143A.3 Cr model output example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145B.1 Results of OM leach test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153B.2 STot/FeHR vs. FeHR/FeTot for Cismon samples deposited during OAE1a . . . . . . . 154B.3 FeHR/FeTot vs. CIA for the Cismon core . . . . . . . . . . . . . . . . . . . . . . . . . . 155B.4 1D water column reaction transport sensitivity analysis results . . . . . . . . . . . . 157B.5 1D water column reaction transport isotope results . . . . . . . . . . . . . . . . . . . 158B.6 Results of gypsum supersaturation model . . . . . . . . . . . . . . . . . . . . . . . . . 162B.7 Area specific modern marine pyrite burial rates statistics . . . . . . . . . . . . . . . . 164B.8 Box model sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166C.1 XRD spectra of LT deep sediment (200 m) . . . . . . . . . . . . . . . . . . . . . . . . . 170C.2 SEM-EDS statistical results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172C.3 Detailed photomicrographs of framboidal magnetite surfaces . . . . . . . . . . . . . 173C.4 Framboidal forms of magnetite in LM sediment . . . . . . . . . . . . . . . . . . . . . 174C.5 Authigenic magnetite in LT sediment . . . . . . . . . . . . . . . . . . . . . . . . . . . 174C.6 Detrital magnetite morphologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175xivAcknowledgementsI consider myself extremely lucky to have found something I am deeply passionate about. WhenI reflect back over my studies, which have now been ongoing for more than a decade, I arriveat a memory about what first got me interested in Earth science. Dr. John Hanes instructed thefirst geology course I took during my undergrad at Queens University. It was John’s incrediblyenigmatic, dedicated and weird teaching style that inspired me to want to learn more, his passionfor Earth science was contagious. Over my scientific career I have been fortunate to meet manypassionate individuals; through collaborations, at scientific meetings, during fieldwork, in thehallway at the department, in the classroom and beyond, it is these encounters that remind methat science is the people involved. For the most part, it is this community of people that motivatesme to continue to want to do science. This dissertation is the result of over four years of researchthat has benefited from the contributions of many.Foremost, I thank my thesis advisors Sean Crowe and Roger Francois for their mentorship,inspiration, tolerance and support. I thank them both for being exceptional role models. Sean andRoger were both very supportive of my ideas, and they allowed me the freedom to explore manyscientific avenues. Through this trust in me, I was able to build my own scientific confidence.I thank Sean for giving me the opportunity to travel the world in support of my work. Theopportunities to travel widely to participate in fieldwork, attend scientific meetings, and fulfillsome of my own hobbies, were some of the most enjoyable aspects of my thesis. I also thank themboth for challenging me. During particularly tricky parts of my thesis work, they allowed mestruggle just the right amount, in order for me to develop scientific character. It is largely dueto Sean and Roger’s excellent mentorship that I still have the aspiration to continue forward inacademia.I would like to thank my committee members for not only shaping me scientifically, butas a person as well. Thank you Stephen Calvert and Mark Jellinek for being two of the mostoutstandingly helpful and inspirational people I have had the pleasure of knowing. I thank Stevefor always taking the time to chat with me. The many scientific discussions we shared over emailor in his office meant a lot to me. His deep wisdom and humble attitude were great motivators. IfI ever make it as deep into a scientific career as Steve has, I only hope to remain as cool. I thankMark for his mentorship. Mark is an intense person in all the most positive ways. This intensityshows in his scientific style, but also in his approach to life. It is something that resonated withme very deeply. When I think about how I want to develop my own scientific philosophies, how Ixvwill want to engage with my own future students, or how I plan to carry myself professionally,much of this will be modelled after Mark’s outstanding example. I would also like to thank themembers of the examining committee who were present at my oral defence; Greg Dipple, SimonDonner and Michael Tenzer.I am very thankful for my wonderful lab mates. Sean and Roger did an amazing job atcurating and incredible team, and these people made it a joy to come to work everyday. Thankyou Maureen Soon, Kate Thompson, Ashley Davidson, Julia Huggins, Jenifer Spence, RachelSimister, Niko Finke, Celine Michiels, Arne Sturm and Genna Patton; you will always remain notonly colleagues but also great friends.I am fortunate to have collaborated with many inspirational people during the course of thisthesis. Without the insight and expertise of these individuals, much of this work would not havebeen possible. I thank Noah Planavsky for his unending support in all of my Cr isotope work, thecornerstone of my thesis. I am truly appreciative (and proud) of our collaborations and I lookforward to many more years of scientific teamwork and friendship. Many thanks to membersof Noah’s group; Devon Cole, Dan Asael, Bleuenn Gueguen and Terry Isson, for all their help.Thank you to the GFZ geomicrobiology team. Jens Kallmeyer, Andre Friese, Aurele Vuilleminand Axel Kitte. I am so pleased to have met such fun, intelligent and inspirational people so earlyin my PhD. I thank Jim Russell, Henrik Vogel, Luis Ordonez and Daniel Ariztegui for all thethoughtful discussions about the Malili lakes. Thank you to James Byrne and Andreas Kapplerfor their insights on Fe biogeochemistry. I thank Sergei Katsev for his insight on geochemicalmodelling. Sergei was always dedicated to chatting on Skype or via email, and he provided anexpert understanding of many geochemical problems discussed during our collaborations. Finally,thank you to Simon Poulton, Elisabetta Erba and Cinzia Bottini for generous sharing of theirsample collections.I would like to thank my all my friends. I am so fortunate to have met and shared my lifewith so many incredible people. I am a big proponent of work life balance, and without sucha strong community of friends to draw inspiration from; my scientific creativity would havebeen nonexistent. Each in their own way, they made this thesis bearable. Firstly my band mates,Matt, Max and David. I am thankful to have experienced so much with these gentlemen. Theykept me sane, and together gave me an outlet to go completely insane. We have accomplisheda lot together and I am very proud of us. Thank you Cole Nowicki. Cole is someone I respectvery deeply and his friendship has been monumental to me from the very first day I arrived inVancouver. Thank you Michelle Pezel and Antisocial. Michelle is the most generous, selfless andhardworking individual I have ever met and she shines as an example on how to work hard andenjoy life at the same time. In many ways Michelle, and the Antisocial Skateboard Shop defineVancouver for me. Thank you Amanda Field for the immense support over the last year, duringthe most intense period of thesis writing, you are special.My deepest thanks go to my family. I have always had the full encouragement and support ofxvimy family to pursue my passions. Mom, Dad, Brennyn, Garrett, Dean, James, Josee, thank you foralways pushing me to be my best, and for always being there for me when I wasn’t.xviiDedicationI dedicate this thesis to the Toronto Raptors.xviiiChapter 1Introduction1.1 Oxygenation history of the EarthThe oxygen concentration in Earth’s ocean-atmosphere system is tightly coupled to elementalcycles and is linked to the evolution of life and the advent of oxygenic photosynthesis – arguablythe most profound evolutionary development in Earth history. The O2 concentration of theocean-atmosphere system, however, has changed greatly over geologic time and this change inoxygen concentrations has consequences for the evolution of Earth’s biosphere [2]. On the modernearth with 21% atmospheric oxygen, aerobic respiration is the most energetically favourablemetabolism [3]. For the majority of Earth’s history, however, oxygen was likely present onlyat trace concentrations (0.01% present atmospheric level, PAL) [4–7]. The oxygenation of theatmosphere is coupled to biological evolution, as the atmospheric O2 pool represents a balancebetween oxygen produced via oxygenic photosynthesis and consumed during respiration andreaction with reduced chemical species [8–10]. On geologic timescales the redox state of Earth’ssurface environments influences the exchange of carbon between oxidised and reduced reservoirsand ultimately the amount of organic carbon buried in the sedimentary reservoir – liberatingfree O2 [10–12]. Discovery of both biological and geochemical evidence for oxygen in the ocean-atmosphere system, therefore, is a key component in refining knowledge on of the evolution oflife and Earth’s surface chemistry.Accumulating evidence indicates two important events in the evolutionary history of oxygen[2, 6]. The first is an initial accumulation of O2 in the atmosphere around 2.4 Ga, which isknown as the great oxidation event (GOE) [13–15]. The GOE is believed to signify the first majoraccumulation of free O2 in the atmosphere, supported by the disappearance of easily oxidiseddetrital minerals (pyrite, uraninite) from the rock record, the abrupt disappearance of a non1mass-dependant sulphur isotope fractionation signal, and multitude other geochemical proxies[16, 17, 17]. All of these proxies taken in tandem indicate a fundamental accumulation of lowlevels of O2 in the atmosphere at this time. The second step toward modern atmospheric O2concentrations occurred sometime between 0.8 – 0.5 Ga (the so-called Neoproterozoic OxidationEvent, NOE, e.g. [18, 19], which ushered in an ocean-atmosphere system with PAL. Whileimprecise, the timing of the NOE is also concomitant with the diversification and radiationof complex life [20, 21]. The ∼2.0 billion years, however, between the GOE and the NOE is atime interval referred to as the “boring billion”. The Mesoprotoerozoic, the geologic era that isbookended by the first major rise in atmospheric oxygen at the GOE, and then the explosion ofmulti-cellular life at the NOE - was likely the time interval that accommodated the evolutionof eukaryotes [22] and as such is anything but “boring”. Conflicting lines of evidence forhighly variable O2 concentrations in the ocean-atmosphere system, however, also characterize theMesoproetozoic. Some suggest O2 concentrations were as high as 4% PAL - high enough to supportcomplex animal respiration [23, 24]. Contrasting this is a growing body of evidence suggestingthat atmospheric O2 remained at consistently low concentrations (< 0.1% PAL) [4, 7, 25]. Giventhat complex animals have minimum oxygen requirements [26], the increased diversity of thebiosphere at the end Precambrian has been hypothesized to be linked to a rise in environmentalO2 concentrations [7, 27]. A contrasting view; is that the dramatic rise of animals to ecologicaldominance at that time, simply reflects genetic innovation independent of environmental controls[28]. Major questions still exist, therefore, as to how the evolution of life and oxygen concentrationsmight be related.More recent stable isotope evidence suggests whiffs of O2 potentially linked to oxygenicphotosynthesis as far back as ∼3.0 Ga, much before both the GOE and NOE [29–31]. Importantly,these lines of evidence suggest the possibility that oxygenic photosynthesis may have evolvedduring the Archean Eon, yet oxygen did not accumulate in the atmosphere due to strong fluxesof reductants at Earth’s surface prior to the Great Oxidation Event (GOE)[14]. It is clear that theocean-atmosphere oxygenation state over Earth’s 4.5 billion year history has been nothing short ofdynamic. Even in the last 540 millions years of the current Phanerozoic Eon, atmospheric [O2]has changed by as much as 20% [32, 33]. Overall, new knowledge on how the oxygenation state2of Earth has evolved throughout geologic time will inform us about the evolution of life, thebiogeochemical cycling of key nutrients, mass extinctions, and climate change.Our emerging picture of the protracted oxygenation of Earth’s surface through time reliesheavily on our interpretation of geochemical signals preserved in the rock record. There has beenmuch work done to develop geochemical tools that inform on oxygenation states, in essence oxygenproxies, and to determine the mechanistic underpinnings that make these particular proxies useful.To this end, some of the most widely used redox proxies in both modern and ancient sedimentarysystems are based on the concentration and isotopic composition of transition metals such as Cr, V,U, Mo and Fe preserved in sediments and sedimentary rocks [34–36]. The utility of these metalsin serving as paleoredox proxies is due to the straightforward observation that their solubility inthe natural environment is linked to their redox state and thus to local redox conditions, and thisnormally results in strong sedimentary enrichments under oxygen poor conditions [36]. Whilethis first order observation is certainly true for both Cr and Fe, certain aspects regarding thesystematics of how these two metals are cycled under varying environmental conditions remainunknown. In this thesis, therefore, I aim to place better constraints on the biogeochemical cyclingof Cr and Fe, two transition metals whose geochemistry in the natural environment is very muchinterconnected. This will allow development of more robust frameworks for the use of these twometals as paleoredox proxies and enable more nuanced application to help unravel the complexhistory of oxygen in Earth’s surface environments.1.2 Chromium biogeochemistryChromium (Cr) isotope ratios are emerging as a sensitive and robust proxy for the evolving redoxstate of the ocean-atmosphere system over geologic time-scales [7, 30, 31, 37–40]. First-orderassessments of order of magnitude changes in the oxygen concentration of the atmosphere andoceans, indeed, can be recorded by large amplitude (several per mil) shifts in the Cr isotopiccomposition of marine chemical sediments, notably banded iron formations (BIF), ironstones,shales and sedimentary carbonates [30, 31, 39–42]. So far, Cr isotope data aligns with broad trendsin the stepwise oxygenation of the Earth based on other proxies [4, 6, 7, 30], however, in depth3application of the proxy and elucidation of oceanic paleoredox states during key intervals in Earthhistory, still requires a more detailed framework to define the biogeochemical processes governingisotope fractionation and to ultimately know how different δ53Cr signals are preserved in therock record. While Cr isotopes appear to hold great promise for the reconstruction of paleoredoxconditions, more nuanced applications would require a much more granular view of global Crcycling and its impact on the distribution of Cr isotopes in the relevant sedimentary pools thatserve as Cr repositories and archives of the past. To develop a framework for how Cr geochemistryis linked to ocean-atmosphere oxygen content, I must first introduce current knowledge of themodern Cr biogeochemical cycle. The current, and relatively high-level picture of the modernglobal Cr cycle is illustrated schematically in (Fig. 1.1).Cr2O3             Cr(VI)aqMnO2Cr(VI)aq , δ53Cr = -0.12 – 5.00‰Continental runoffCr(VI)aq, δ53Cr  = 0.13 – 1.24‰Cr(VI)aq                  Cr(OH)3 Fe(II), H2SCr2O3           Cr(III)aqH+Cr(OH)3                   Cr(III)aq δ53Cr BSE 1234Reactions1.  Cr(III) oxidation2.  Acid Cr(III) dissolution3.  Hydrolysis 4.  Cr(VI) reductionHydrothermal ventsChemical sedimentsSoilsContinental weatheringSeafloor weatheringδ53Cr = -0.57 – 5.00‰1Figure 1.1: The modern chromium biogeochemical cycle, modified from [30]. We display the Cr isotope compo-sition of the two main Cr fluxes to the ocean (Continental runoff, δ53Cr = -0.12 – +5.00h, [43–45], andHydrothermal vents, δ53Cr = -0.12 ± 0.1h, the igneous silicate Earth composition (ISE), [46]), as well as thefour main reactions (numbered) that control Cr isotope fractionation. The δ53Cr composition of the marinesediment reservoir ranges from -0.57h to 5.00h. Grey arrows indicate the cycling of Cr species that haveundergone a redox reaction, whereas black arrows indicate the cycling of Cr species that have not.41.2.1 Chromium speciationChromium exists in oxidation states from -2 to +6, and in the natural environment it is presentmainly as Cr(III) and Cr(VI) species. These two redox states have remarkably different geochemicalbehaviours, dictating both their distribution and mobility in the environment. The reduced state,Cr(III), is typically found in octahedral coordination and is the most stable state, forming a widerange of solid Cr-species. Solid Cr(III) species are incorporated in a variety of rock formingminerals that are abundant in Earth’s crust, and thus Cr concentrations in Earth’s crust exhibit arange of 100 - 300 µg g– 1 [47, 48]. Based on thermodynamic considerations, in seawater, dissolvedCr(III) is largely insoluble and should only be prevalent under strongly reducing conditions [49].The dissolution of Cr(III) from crustal rocks, and its subsequent delivery to the ocean, is onlypromoted under acidic conditions (pH < 4 ) [49]. Cr(III) is also highly particle reactive and formsstrongly hydrolyzed species in solution based on natural pH and redox potential. The particlereactive nature of Cr(III) species is also thought to be responsible for it’s low concentration inmost natural waters as it sorbs strongly to mineral surfaces and organic matter at pH > 4 andsubsequently scavenged and removed from solution [47, 50, 51]. When considering relevant Cr(III)species under a given set of conditions, anionic hydroxo species, mixed ligand complexes andpolynuclear complexes are usually ignored due to their limited presence at typical environmentalpH; Cr(III) species are thus usually reported as Cr(OH)2+, which dominates at pH 5, and Cr(OH)3,which dominates at pH 8.In natural waters, oxidized Cr, Cr(VI), takes a tetrahedral coordination and forms a variety ofmobile oxospecies (HxCrO2 – x4 ) depending on the pH of the environment and Cr(VI) concentrations[50, 52]. Under similar environmental conditions, most Cr(VI) solid species are soluble [51].Cr(VI) is known to be toxic and carcinogenic and it is widely used in industrial activities,leading to contamination of surface and ground waters [53]. At circumneutral pH, Cr(VI) ismostly present as the highly soluble oxyanion chromate (CrO2 –4 ) and is stable in oxygenatedenvironments. For concentrations < 10 µmol l– 1, typical of natural waters, Cr(VI) will be foundonly as HCrO–4 or CrO2 –4 (dependent on pH); chromic acid and hydrogen dichromate (H2CrO4 andHCr2O7 – , respectively) are found only at low pH and dichromate (Cr2O72 – ) species only at high5Cr concentrations [52]. In contrast to Cr(III) species, in solution at natural pH, Cr(VI) oxyanionsare weakly sorbing and are thus highly mobile [47, 51, 54]. The distinct redox geochemistry of thedifferent Cr species in the natural environment is ultimately linked to fractionation of its isotopesand its emergence as a paleoredox proxy.1.2.2 Chromium redox reactionsRedox reactions determine the mobility, speciation and prevalence of Cr in the natural environment.Some relevant redox reactions which I explore below, include: oxidation by molecular oxygen (O2),hydrogen peroxide (H2O2), and manganese (oxyhydr)oxides, and reduction by Fe(II), hydrogensulphide (H2S), and H2O2. Thermodynamics imply the presence of Cr(III) to be relatively limitedunder oxic conditions. Observations of Cr(III) concentrations in nature that are much higher thanexpected based on thermodynamics alone, demonstrates the large role of kinetics in determiningthe state of Cr found in the environment [51]. Cr(III) has a large crystal field stabilization, thehighest of all transitional metals, a product of having its 3d electrons in high-spin octahedralcoordination, which gives rise to its strong kinetic stability [55, 56]. Given thermodynamicequilibrium calculations, and in the absence of Cr(III)’s strong kinetic stability, one would expectto find a large ratio of Cr(VI)/Cr(III) of 10 20 at a pH of 8.1 and an oxidizing pE of 12.5 [55]. Givenratios of Cr(VI)/Cr(III) that have been measured in natural environments are much smaller (arange of < 1 – 70), it is likely that in-situ reduction of Cr(VI) to Cr(III) must occur in order toachieve such high concentrations of Cr(III); kinetics help to maintain these concentrations, butalone are not capable of producing Cr(III) in such large quantities [55, 57].Two common reductants of Cr(VI) in aqueous environments are Fe(II) and H2S. Previousworkers determined reduction by Fe(II) to have an overall rate constant given by the equation;log(k) = 11.93+ 0.95 · pH–(4260.1/T)–1.06 · I0.5(= 0.2) (1.1)at pH 5 - 8.7, temperature between 5 - 40◦C, and I = 0.01 – 2 M [56]. Cr(VI) reduction kineticswere also determined by [58] for Fe(II) bearing minerals. The authors found the reaction kineticsto decrease as a function of Fe-mineralogy as follows: goethite ∼ ferrihydrite >> montmorillite6> Fe-clays [59]. It has also been hypothesized that photo-reduction of Fe(III) occurring in oceansurface waters may provide a significant a source of Fe(II), which is then able to reduce Cr(VI) toCr(III) [60]. At high pH, oxygen is more efficient at oxidizing Fe(II) to Fe(III), and thus Fe(II) isnot a significant reductant of Cr(VI) under such conditions [61].Under anoxic conditions at pH < 5.5, H2S is the dominant reductant of Cr(VI) [62]. Hydrogensulphide reduces Cr(VI) with an overall rate equation given by;log(k) = 16.19–1.06 · pH–(2300.9/T)(= 0.07) (1.2)This reaction has a first order dependence on pH, no significant effect by ionic strength and a firstorder response to the concentration of H2S. Given both Fe(II) and H2S concentrations are limitedunder oxidizing conditions, it has been suggested that microbial reduction of Cr(VI) may be thedominant pathway controlling Cr(VI) reduction in oxygenated waters [62].The rate of oxidation of Cr(III) by molecular O2 is sluggish, and this allows for the build up ofCr(III) in natural systems, despite the fact that Cr(VI) is the thermodynamically favoured species.In a previous laboratory experiment that explored the reaction kinetics of Cr(III) with O2, authorsdemonstrated that over a two week period less than 2% of initial Cr(III) was oxidized over a pHrange of 5.9 - 9.9, and at room temperatures [61]. The reaction rate can be increased with elevatedtemperatures, however, it is still significantly slower than other kinetic oxidation reactions, suchthat Cr(III) will be oxidized by other chemical species before being oxidized by dissolved oxygen[61].Hydrogen peroxide is can act as both an oxidant and reductant of Cr. The oxidation of Cr(III)by H2O2 can be described by the following equation, valid at pH 5 - 7, I = 0 - 1M, temperaturebetween 5 - 40◦C;log(k) = −4.60+ 0.87 · pH–(5.13/T)(= 0.06) (1.3)This rate equation provides an estimated half life in surface waters of about 24 days, compared tooxidation by O2, estimated at about 500 days [57]. In highly acidic environments at pH < 5, H2O2becomes a reductant of Cr(VI). This reaction, however, has only been explored in relatively few7natural environments, for example within atmospheric aerosols and hot springs [57, 63].Cr(III) oxidation by manganese (oxyhydr)oxides phases is the primary mechanism of Cr(VI)production in natural environments. The kinetic studies of this general reaction differ widely dueto the varying mineralogy of both Mn and Cr bearing phases, and the pH of the environmentwhere the reaction takes place [64]. Previous workers demonstrate that this reaction is oftenbiphasic, and initially Cr(III) is oxidized very rapidly (within the first 30 minutes of reaction time),after which its oxidation slows dramatically [61, 64–66]. This change in reaction rate has beensuggested to be a result of Cr(III) precipitates that form on the surface of the Mn (oxyhdry)oxidesduring the initial reaction phase, and this oxide coating inhibits continued reaction [47, 67].Importantly, Mn (oxyhdry)oxides act as an electron acceptor in a direct reaction to oxidize Cr(III)at rates orders of magnitude faster than molecular O2, and thus play a key role in cycling of Crspecies and attendant isotope fractionations that occur in the natural environment.1.3 The Cr isotope paleoredox proxyThe dramatically different geochemical behaviour between oxidized and reduced Cr species hasled to the development of Cr as a paleoredox proxy, as strong fractionation of Cr isotopes ismediated by both equilibrium and kinetic redox reactions [31, 68–71]. Chromium has four naturallyoccurring isotopes, 50Cr (4.35%), 52Cr (83.79%), 53Cr (9.50%) and 54Cr (2.36%). Equilibrium isotopetheory predicts that the heavy isotope 53Cr will preferentially partition into Cr-species withstronger bonds (tetrahedral vs. octahedral co-ordination), and theoretical calculations predictup to 6 - 7h equilibrium isotope fractionation between Cr(III) and Cr(VI) species at standardstate conditions [70, 71]. Large equilibrium fractionations between Cr redox species are predictedand have been observed in laboratory experiments, however, given that in natural settings, onlyvery limited amounts of Cr(VI) have been observed to sorb on Cr(III) (oxyhydr)oxide surfaces,the conditions and timescales of these equilibrium reactions are only relevant in select few earthsurface environments [71].81.3.1 Oxidative weathering of crustally bound Cr(III)Cr resides in the continental crust bound in its relatively low solubility trivalent redox state Cr(III)and hosted in silicate and oxide minerals [47, 72]. Cr(III) is liberated through oxidative weatheringas Cr(VI), which forms through solid-state catalysis by reaction with Mn-oxides [65]. Duringoxidation, Cr(III) loses electrons and the Cr(VI) produced is enriched in the 53Cr isotope. Thisoccurs through a likely combination of Cr(III) oxidation to Cr(VI), and partial back reduction ofCr(VI) (or reactive Cr(IV) and Cr(V) intermediates) [30, 73]. For example, when crustal boundCr(III) is oxidized to Cr(VI), this leaves Cr(III) in the reactant (weathering profile) isotopicallylight. If isotopic equilibration between Cr(VI) and Cr(III) in aqueous solutions is not reached, andno further redox reactions take place, then Cr(VI) will retain the isotopic composition impartedduring the oxidative weathering process [30, 31]. As mentioned above, the oxidation of Cr(III) inthe presence of molecular O2 alone is sluggish and thus for Cr(III) oxidation to be relevant onthe timescales of environmental transport, the reaction requires the presence of manganese oxidephases as a catalyst. Manganese oxide production, in turn, requires molecular oxygen. Thus byextension, the oxidative half of the Cr biogeochemical cycle and the preservation of fractionated Crisotopes in marine sediments may provide a record of oxidative Cr cycling and thus the presenceof O2 in Earth’s surface environment.Relatively few studies have examined the δ53Cr composition of minerals contained in thevast diversity of crustal rocks. Chromium is a highly compatible element in magmas and as aresult mafic and ultramafic rocks are the major reservoirs in Earth’s crust [74, 75]. The δ53Crcomposition of high temperature crustal rocks and minerals is mostly homogeneous (δ53Cr = -0.124± 0.1h, [46, 74]) and this narrow range of crustal δ53Cr composition sets the comparative baselinefor determining whether the δ53Cr composition of other environmental samples is fractionatedoutside of this inventory range, known as the igneous silicate earth composition (ISE). More recentmineral specific δ53Cr measurements conducted on Cr-rich chromites, however, show that theseminerals tend to have δ53Cr compositions that fall within the ISE, but with a heavier mean valueof -0.079 ± 0.129h [74]. Other studies also demonstrate small but measurable heterogeneity in theδ53Cr composition of mantle-derived minerals. For example, chromite minerals have consistently9heavier δ53Cr compositions relative to chromite free igneous rocks (peridotites) [75]. Ultimately,weathering of continentally derived Cr(III) contained in the crustal minerals, is the principlesource of Cr delivered to the oceans via runoff (reaction 1, Fig. 1.1).1.3.2 Riverine delivery of Cr to the oceansThrough riverine transport, oxidative weathering sustains the largest Cr flux to the oceans today[25, 44, 76, 77]. Rock weathering and anthropogenic inputs of Cr to rivers and are estimatedat 5 x 10 8 mol yr– 1 and 2 x 10 9 mol yr– 1, respectively [78]. Cr may be in particulate or ionicdissolved form, which based on thermodynamics calculations, would comprise primarily of Cr(VI)in oxidized water at a pH range 5 - 9, though particle transport makes up a large portion of thetotal Cr load in large rivers [77, 79, 80]. For example, Cr speciation of the Columbia River andestuary shows that Cr(VI) makes up > 90% of the total dissolved Cr, with similar enrichments inboth the estuary and the mouth of the river [81]. Dissolved Cr(III) is present, but only makes up3% of the total Cr, likely due to it’s slow oxidation kinetics [65]. Particulate Cr is high especiallyin the estuary due to flocculation processes and removal of Cr(III) with increased salinity [81].Cr concentrations measured in the within San Francisco Bay display comparable trends to thoseobserved in the Columbia River, but large variations throughout the extended region are observed[68]. The authors assign these differences primarily to physical mixing, resulting from localizedinputs of Cr(III) and in-situ reduction of Cr(VI) in certain areas. The bed lithology underlyingrivers also plays a key role in determining the speciation and concentration of Cr in a givencatchment, and typically Cr concentrations are higher in rivers that drain ultramafic catchmentsrelative to rivers that drain more felsic lithologies [77]. Globally, there is a large range in the fluxof total dissolved Cr, ranging from 2.2 x 10 2 mol yr– 1 to 2.5 x 10 8 mol yr– 1, the highest fluxescoming from anthropogenically-contaminated waters (McClain and Maher, 2016). The total Crconcentration in rivers also ranges over a few orders of magnitude, from 5 – 1923 nM [77]. Acompilation of river, groundwater and estuarine Cr fluxes can be seen in Table 1.1.A fraction of Cr may also be delivered to the oceans as Cr(III) in rivers, possibly stabilized asCr(III)-organic complexes [85, 86]. Cr(III) is also highly particle reactive and exhibits typicalcationic behaviour as its affinity for adsorption increases with pH - this is in contrast to the Cr(VI)10Table 1.1: Cr River, Groundwater and Estuary Flux Estimates. †World area and discharge from [? ]. ‡Worldaverage Cr(VI) concentration. ?Global Cr(VI) flux = sum of Cr(VI) flux/fraction of world discharge. Globalarea normalized Cr(VI) flux = Global Cr(VI) flux/world area. κWorld average CrTot concentration. φGlobalCrTot flux = sum of CrTot flux / fraction of world discharge. ψGlobal area normalized CrTot flux = GlobalCrTot flux / world area. θ[82]. χ[81]. σ[83]. ρ[84]. ζ[77]. υ[43] World Rivers Ground Water Estuary Area (103 km2) 79500†	   Discharge (km3 yr-1) 37288†   Cr(VI) (nmol l-1) 149.8‡	 3.8 – 1000θ  1.00 – 2.69χ  Cr(VI) flux (106 mol yr-1) 1716	   Cr(VI) flux (mol km-2 yr-1) 21.59◊	   Cr(TOT) (nmol l-1) 175.16κ  5.0 – 1000ρ  3.0 – 14.0σ  Cr(TOT) flux (106 mol yr-1) 3506φ    Cr(TOT) flux (mol km-2 yr-1) 44.10ψ    Citation for concentration ξ ν   Table 4. Cr River, Groundwater and Estuary Flux Estimates. †World area and discharge from Dai and Trenberth (2003). ‡World average Cr(VI) concentration. Global Cr(VI) flux = sum of Cr(VI) flux/fraction of world discharge. ◊Global area normalized Cr(VI) flux = Global Cr(VI) flux/world area. κWorld average Cr(TOT) concentration. φGlobal Cr(TOT) flux = sum of Cr(TOT) flux / fraction of world discharge. ψGlobal area normalized Cr(TOT) flux = Global Cr(TOT) flux / world area. θAllard (1995). χCranston and Murray (1980). σCampbell and Yeats (1984). ρIzbicki et al., (2008). ξMcClain and Maher (2016). ν Raddatz et al., (2011)   	anion that is generally soluble at circumneutral pH [49]. It has also been demonstrated thatorganic matter (OM) may play a large role in complexing Cr(III). Previous workers demonstratethat Cr(III) ions are quickly adsorbed and precipitated in the absence of soluble OM [87, 88].Thus in most natural waters above pH 5.5, Cr(III) may be present in higher concentrations thanwhat is expected based on thermodynamics due to stabilization via OM [85]. Generally, Cr(III) isotherwise only soluble and liberated to solution at acidic pH.The δ53Cr composition of rivers reflects the products of oxidative Cr(III) weathering on thecontinents, imparting the variable but consistently heavy δ53Cr compositions observed in seawater[1, 41] (Fig. 1.1). Quantitative information on the magnitude of the Cr flux from different rivers iscontinually improving, however, limited data exists on the isotopic composition of river waterglobally. Cr isotope data measured on select river water samples from around the world reveal thatthe Cr isotope composition of the global river flux is quite heterogeneous. Cr isotope compositionsdetermined in river water and suspended particles from the Connecticut River, USA, display arange in δ53Crriver water between -0.17 and +0.92h, whereas the suspended particles exhibit aδ53Cr composition between -0.11 and +0.13h [45]. This data is consistent with the framework that11river water is expected to deliver a heavy δ53Cr flux to the ocean, yet considerable variability maybe introduced by runoff seasonality and catchment lithology (see section above). Likewise, theδ53Cr composition of the Uruguay River exhibits a muted range of heavy δ53Cr compositions witha mean of only 0.08h [44], and in contrast, the Glenariff River waters display a much wider rangeof δ53Crcompositions between -0.17 and +1.68h [44]. Again, authors suggest the considerablevariability in the δ53Cr composition of these particular rivers can be attributed to local catchmentlithology and the prevailing redox conditions that may fluctuate over each river’s course. Overall,current data reveals that the δ53Cr composition of the global river flux is overwhelmingly heavy,with a range that is consistent with and overlaps the range of current measurements of the δ53Crcomposition of seawater (see seawater section below). Heterogeneity in the river flux δ53Cr,however, does exist, and this may modulate the isotope composition of ocean water especially inclose proximity to river mouths.1.3.3 Hydrothermally sourced Cr(III)The other principle source of Cr to the oceans is hydrothermal vents. The hydrothermal contri-bution of Cr to the marine Cr inventory, however, remains poorly constrained and is currentlyestimated to represent < 1% of the Cr flux supplied from rivers [25]. While hydrothermal ventfluids are rich in dissolved Cr(III) sourced through the acid dissolution of basaltic seafloor, muchof this Cr(III) is likely scavenged by Fe (oxy)hydroxide particles that form when Fe(II) ladenvent fluids mix and react with ambient oxygenated seawater [89–94]. Deposition of these Fe(oxy)hydroxides may thus capture the entirety of vent derived Cr for deposition in proximalhydrothermal sediments [95]. Given the often high abundance of ferrous Fe, a potent reduc-tant of Cr(VI), in vent fluids, hydrothermal sediments may also capture a fraction of seawaterCr(VI), making vent systems a net sink for seawater Cr. Recent data, however, also shows thathydrothermal Fe particles can be transported great distances from their vent sources [96]. Whilethe fate of Cr in these particles remains entirely unknown at this time, long distance transport ofhydrothermal Fe and associated elements may have implications for the distribution of Cr in theoceans. Nevertheless, it is clear that an appreciable fraction of Cr from vent fluids finds its wayto the seafloor in association with hydrothermal Fe oxyhydroxide sediments, either from direct12incorporation of Cr(III) into Fe (oxy)hydroxides, reduction of seawater Cr(VI) via vent Fe(II)aq, ora combination of both processes.Acid dissolution of Cr(III) is thought to be associated with minimal isotope fractionation[30, 97] and given that acid dissolution of basalt liberates Cr to vent fluids, the efflux of Cr(III)from hydrothermal vent waters is expected to carry an isotopic signature no different fromthe igneous silicate earth (ISE), -0.124 ± 0.1h [46]. Furthermore, precipitation of authigenicCr(III) phases or sorption of Cr(III) onto oxides is assumed to operate free of significant isotopefractionation [37, 38]. If hydrothermal Cr(III) escaped removal by (oxy)hydroxide particles in thevicinity of vents, it would likely carry an igneous reservoir Cr isotope composition—significantlylighter than the riverine influx of Cr(VI) [30, 46]. Recent laboratory experiments however, indicatethat organic acids are capable of dissolving and complexing significant amounts of Cr(III) fromCr(III) (oxyhydr)oxides, a redox independent process resulting in isotope fractionations of up to1.5h [86]. Although the δ53Cr composition of organic ligand bound Cr(III) may play a role insetting the δ53Cr composition of seawater, the relative importance of these non-redox processesfor Cr(III) cycling in natural environments is currently unknown.1.3.4 Chromium in seawaterChromium exists at trace concentrations in the ocean, and the concentration of dissolved speciestypically ranges between 1 - 10 nM [1, 77, 81, 98–101]. In oxygenated seawater Cr(VI) should bethe dominant species, and often accounts for upwards of 95% of the total dissolved Cr, whichagrees with thermodynamic calculations [102]. However, the measured ratio of Cr(VI)/Cr(III)ranges from 0.04 to 70, with significant Cr(III) pools not uncommon [81, 98, 99, 101]. High Cr(III)concentrations may arise due to the kinetic stability of Cr(III), but local inputs could also contributeif in close proximity to a river mouth (polluted or naturally high Cr) or a hydrothermal ventsource [101]. Data recording Cr speciation in seawater is rather sparse given the number of yearsit has been of interest. Previous workers measured total dissolved Cr profiles in the equatorialPacific, with some corresponding Cr(III) data points [103]. Typically, total Cr increased from thesurface to 250 m, and below 1000 m either maintained a relatively homogenous concentration orgradually increased with depth. In the equatorial Pacific, total Cr concentrations ranged from13about 7 - 10 nM with Cr(III) dominating at 60 - 99% of the total Cr [103]. Previous workers alsomeasured two distinct Cr profiles in the northern Pacific, one in fully oxygenated north-easternPacific water and another in the seasonally anoxic fjord, Saanich Inlet in coastal British Columbia[79, 81]. The authors observe that in oxygenated waters, Cr(VI) is the dominant species presentwith concentrations between 1.5 - 3 nM. However, below the oxycline of Saanich Inlet, Cr(VI)concentrations decrease substantially to < 0.25 nM, while Cr(III) becomes the dominant speciespresent reaching concentrations of about 1 nM. This implies Cr(VI) reduction under anoxicconditions combined with Cr(III) sorption onto sinking particles delivered from the oxic layer. Inthe Atlantic Ocean, [98] measured Cr speciation in the Sargasso Sea, finding seasonal differencesin speciation, yet a relatively constant Cr inventory. Cr(VI) was typically dominant and total Crconcentrations ranged from 2.5 - 4.5 nM. Cr(III) concentrations were strongly positively correlatedwith biomass and Cr concentrations measured were between about 0.1 - 2.5 nM. Two seawater[Cr] profiles were determined by [41] in the oxygenated waters off the coast of South Hampton,and a third profile was obtained in the Argentine Basin. The authors observe no systematic shiftsin the total Cr concentration with depth in either location, with roughly constant [Cr]total of ∼2nM in the South Hampton waters and ∼6 nM determined in the Argentine Basin waters.Coupled seawater [Cr] and δ53Cr measurements imply that the global oceans are heterogeneouswith respect to Cr concentration and isotopic composition. For example, Cr concentrationsand δ53Cr compositions were determined in the Atlantic, Pacific and Arctic Ocean waters [1].Authors find, similar to [41], the δ53Cr composition of seawater is heavy, falling within a rangeof ∼0.5 – 1.5h. This implies that there is global heterogeneity in δ53Crcomposition of seawater.This heterogeneity in the δ53Crseawater composition may arise from supply of fresh waters fromrivers and or local water column reduction of Cr(VI). The authors also demonstrate that mostseawater δ53Cr values can be described by Rayleigh fractionation law, likely resulting from Cr(VI)reduction to Cr(III) in oxygen minimum zones, operating with a isotope fractionation factor ofe = 0.80h. Sediments deposited on the seafloor are thought to record the ambient seawaterδ53Cr composition, (e.g. [30, 31, 41, 104, 105]), however few marine sediment samples recordcontemporaneous modern seawater δ53Cr values, making it difficult to surmise under whatconditions the sedimentary reservoir faithfully captures the Cr isotope composition of seawater14and whether this is altered depending on lithology or subsequent diagenesis.1.3.5 Reduction of seawater Cr(VI) and burial of Cr(III) in marine sedimentsReduction of Cr(VI) to Cr(III) is accompanied by large and characteristic isotope fractionations(ranging from e = 0.5 – 6.0h) and occurs via a range of both abiotic and microbial pathways[37, 43, 105–108]. Given Cr(VI) is the dominant form of Cr in natural waters, its reduction plays acrucial role in Cr cycling in modern oceans, lakes and rivers as the Cr(III) product is removed tothe sedimentary Cr pool [51, 56]. Cr(VI) removal from seawater results from reduction processeswhich enrich the reaction products in the light isotope (52Cr) relative to the reactant. As reductionprogresses the reactant becomes increasingly enriched in the heavy isotope (53Cr) compared tothe Cr(III) product. As mentioned, Cr(VI) reduction is associated with a wide range of electrondonors with variable reaction kinetics, producing different magnitudes of Cr isotope fractionations.However, all reaction processes identified so far result in the remaining Cr(VI)reactant being furtherenriched in the heavy isotope relative to the Cr(III) product.In sediments underlying oxygenated seawater, sedimentary Cr fluxes are dominated bylithogenic (detrital) Cr(III), with a mean δ53Cr composition of -0.1 ± 0.1h [104]. An importantconsideration when measuring δ53Cr in marine sediments, therefore, is the contribution of detritalCr introduced into the environment in question. Currently, δ53Cr corrections for detrital Cr areoften applied using the average Cr:Ti composition of the continental crust (0.02) and assuming thatdetrital Cr pool has a δ53Cr composition of the ISE [4, 7, 35, 104]. In marine sediments where thedetrital flux from the continents is particularly high, detrital Cr will dilute the overall authigenicδ53Cr signal, assuming these phases have a δ53Cr composition of the ISE. More informationon the riverine composition with respect to δ53Cr, and a much broader data set of the δ53Crcomposition of mineral archives of Cr, are both required to narrow down processes of continentalrock weathering, oxidation of Cr(III) and subsequent delivery of Cr(VI) to the ocean.In marine sediments underlying anoxic seawater, quantitative reduction of Cr(VI) may resultin the δ53Cr signature of deposited Cr(III) capturing the ambient seawater composition, likely dueto the abundance of potential electron donors in oxygen free settings (H2S, Fe(II), organic matter)[104, 105]. The reasoning here is that these marine settings may act as essentially closed systems15and Cr(VI) reduction from seawater may be complete, capturing the contemporaneous seawaterδ53Cr value. In contrast, in regions of the ocean with low-oxygen bottom waters where theunderlying sediments have shallow oxygen penetration depths, the sediment δ53Cr compositionhas the potential to be fractionated from the contemporaneous seawater composition [105]. Thisis because O2 inhibits reduction of CrO42 – and fractionation will be limited by diffusion at thesediment water interface. Therefore, it is likely that the δ53Cr composition of modern marinesediments is highly dependant on the geochemical conditions of the diagenetic environment, likeoxygen concentrations, a central theme of my thesis research.1.4 The Fe-speciation paleoredox proxy1.4.1 Fe geochemistry in Earth’s surface environmentsTwo positions over from Cr on the periodic table, is Fe – arguably the most informative metalfor the assessment of local redox conditions in modern and ancient depositional environments.This is largely because Fe is a ubiquitous component of sedimentary environments for it is one ofthe most abundant elements in the Earth’s crust [16]. Similar to Cr, iron’s utility as a paleoredoxproxy stems principally from its susceptibility to undergo redox transformations under surface(or near-surface) conditions [16, 109].Iron cycling in a given sedimentary environment is governed by its supply and speciation[16], which are highly dependent on both pH and redox potential [109]. Reduced ferrous Fe(II)is soluble in anoxic waters at circumneutral and alkaline pH [109]. In anoxic waters, however,aqueous Fe(II) is often titrated out of the system in the presence of sulphide, due to reactionsbetween Fe and S and the low solubility of Fe-S species [110]. In sulphide limited anoxic systems,Fe(II) may also form other reduced Fe-minerals such as carbonates (siderite, FeCO3), phosphates(vivianite, Fe3(PO4)2 · 8H2O) or mixed valence oxides (magnetite, Fe3O4). In contrast to Fe(II),ferric Fe(III) is highly insoluble at circumneutral and alkaline pH. In freshwaters at neutral pH,the dominant dissolved inorganic Fe species is Fe(OH)30 with a very minor contribution fromthe aqueous Fe3+ ion [109]. Similar to Cr(III), however, the solubility of Fe(III) in both fresh andseawater can be enhanced by complexation with dissolved organic ligands [111–114]. Due to its16low solubility, Fe(III) is mostly transported as particulate or colloidal material in most aqueousenvironments, making it one of the few major elements that is transported in the solid phase[115, 116].Solid Fe phases are found with a broad range of chemical structures with varying degrees ofcrystallinity, which have implications for their reactivity in the natural environment (Fig. 1.2).Mineral FormulaFerrihydrite Fe3+O3 - 0.5H20Lepidocrocite Fe3+O(OH)Goethite Fe3+O(OH)Haematite Fe3+2O3Magnetite Fe3+2Fe2+O4Siderite Fe2+CO3Silicates (K,H3O)(Al,Mg,Fe2+)2(Si,Al)4O10Ferrihydrite < Lepidocrocite < Goethite < Hematite < Magnetite < SilicatesIncreasing resistance to sulphidation / reductive dissolutionDecreasing reactivity toward microbial metal reductionFigure 1.2: The reactivity of Fe species targeted by the Fe-speciation sequential extractionFe contained in primary rock forming minerals exists in both reduced and oxidized states.Common sedimentary Fe(III) species comprise disordered ferrihydrite ((Fe3+)2O3 · 1/2 H2O), whichages to ordered ferrihydrite, and given enough time, to other more crystalline forms of Fe suchas goethite (FeO(OH)), lepidocrocite (γ-FeO(OH)) or hematite (Fe2O3), as well as Fe-containingsilicates [117–119]. These mineral transformations are often catalyzed by microbial metabolismsand can occur on the timescales of transport within a given system [16, 120]. Microbiologytherefore, plays a critical role in the biogeochemical cycling of Fe as the majority of Fe(III) reductionoccurs through an energy yielding respiratory metabolism that couples organic carbon oxidation17to Fe(III) reduction [120–123]. As such, Fe(III) reduction and Fe mineral transformations alsoplay a quantitatively important role in global carbon degradation in both aquatic and terrestrialenvironments [124, 125]. A plethora of laboratory experiments furthermore, demonstrate that theextent of microbial Fe(III) reduction is directly linked to the crystallinity of the electron acceptingFe-phase, in such a way that higher degrees of crystallinity typically yield lower rates of microbialFe reduction and Fe(II) production [126].1.5 Fe-speciationThe biogeochemical features of Fe cycling at Earth’s surface have led to widespread use of Fe-speciation analyses as paleoredox proxies [127]. The Fe-speciation paleoredox proxy, in simpleterms, uses the distribution of Fe between different phases (oxides and carbonates in particular)that are susceptible to abiotic and biologic reduction on short, diagenetic time scales (< 1000 years)[110, 127–129]. Enrichment of these minerals, termed “reactive (FeHR)”, relative to the total ironpool including non biologically reactive phases (FeTot) – typically Fe silicates, occurs in sedimentsdeposited beneath anoxic waters that are either ferruginous (Fe2+ bearing) or sulphidic (H2Sbearing, also known as euxinic) [129]. The Fe-speciation proxy, and specifically the thresholdsused to delineate anoxic and Fe-rich vs. sulphide-rich depositional conditions, is calibrated basedon observations in modern marine environments. For example, in a wide range of modern oxicsediments the FeHR pool comprises up to 38% of total sedimentary Fe (FeHR/FeTot = 0.38) withan average value for FeHR/FeTot = 0.26 ± 0.08, defining the modern detrital baseline [118, 130].Enrichments in FeHR that are in excess of this detrital background ratio (> 0.38) indicate a sourceof reactive Fe that reflects processes that lead to its enrichment under anoxic conditions [131]. Inanoxic basins, a set of processes known as the benthic Fe shuttle, together constitutes the mainmechanism driving FeHR enrichment. The benthic Fe shuttle works to enrich FeHR phases bytransporting them from shelf sediments, either as precursor Fe (oxyhydr)oxides or as dissolvedFe(II), to the deeper water column where they are ultimately scavenged and buried.The Fe-speciation proxy is designed, furthermore, to distinguish between these two redoxstates (ferruginous vs. sulphidic) based on the extent to which the FeHR pool has reacted with18H2S produced during sulphate reduction, and has been converted to the mineral pyrite (FeS2).Sedimentary pyrite formation, in turn, requires three key ingredients; organic carbon, a supplyof reactive Fe, and sulphate. In typical sulphate-rich, anoxic marine sediments, organic carbonis degraded through sulphate reduction – driving pyrite formation in the presence of reactiveFe. If the system lacks a source of reactive Fe – the sediments are Fe limited, pyrite formationceases, and the sulphide produced during sulphate reduction may accumulate in those porewaters.Under extreme cases, the absence of reactive Fe phases to buffer H2S, may result in sulphideextending into the water column and the development of euxinic conditions. In contrast, a lack ofpyritization of the FeHR pool in sediments that also contain appreciable organic matter impliesthat pyrite formation was sulphate limited [119, 124, 127]. Under extreme cases, the absence ofsulphate reduction to drive Fe pyrite formation, may result in dissolved Fe(II) extending intothe water column and the development of ferruginous conditions. The Fe-speciation paleoredoxproxy, therefore, may diagnose sediment deposition under euxinic or ferruginous redox statesthrough quantification of the Fe-bearing mineral pools involved in the biogeochemical cycling ofFe and S in a given depositional system.The Fe-speciation paleoredox proxy has been applied widely to ancient sediments and euxinicmodern analog settings, for example the sulphidic Black Sea, the seasonally sulphidic Saanich Inletand the stratified Cariaco Basin [34, 105, 131–134]. Detailed studies of the Fe-cycle in ferruginousenvironments, however, have only started to emerge, as ferruginous water bodies are rare on themodern Earth. There are no modern examples of ferruginous conditions in marine settings today,and thus much of our framework for Fe-cycling under ferruginous conditions has come throughstudy of modern lacustrine environments [135–142].Modern ferruginous environments provide natural laboratories to examine processes extensibleto the Precambrian oceans [137, 143], however, a precise framework for how the Fe cycle operatesunder such conditions is lacking. The Malili Lakes system, Sulawesi Island, Indonesia, comprisesof a series of interconnected lakes, together encompassing the largest ferruginous basins on theplanet and thus the ferruginous basins of the Malili Lakes provide a an key analog environmentto investigate processes in the Fe biogeochemical cycle that were likely widespread during thePrecambrian eons. Ultramafic rocks of ophiolitic origin dominate the catchment basin surrounding19the lakes, and weathering of these lithologies has led to development of exceptionally Fe-richlateritic soils [137]. Heavy tropical rains deliver strong fluxes of Fe-(oxyhydr)oxides from these soilsto the lakes, which exert an overwhelming influence on the lakes’ biogeochemistry [137, 138, 142].The two most prominent lakes in the system, Lake Matano and Lake Towuti, are physicallystratified and characterized by persistently anoxic Fe(II)-rich (∼140 and ∼10 µM respectively)and virtually sulphate free deep waters [138, 143]. Comprehensive Fe-speciation surveys of riverparticulates and sediments reveal complexities in Fe-cycling dynamics during source to sinktransport [119, 144] that are currently poorly understood. Therefore although extremely rareon the modern Earth, ferruginous water bodies represent understudied natural laboratories forexploring the ecology and biogeochemistry of Fe-rich waters extensible to the ferruginous oceansthat existed during much of Earth’s history. The Malili lakes provide a necessary analog system inwhich to conduct studies that improve the framework of the Fe-speciation proxy; a second themeof my thesis research.1.6 Dissertation overviewTo create new knowledge on the Cr-isotope and Fe-speciation paleoredox proxies and apply themto reconstructing paleo oxygen conditions during key intervals in Earth history:(i) I conducted experiments to determine Fe-speciation and Cr isotope fractionations insediments from modern depositional environments with varying geochemistry (oxic, anoxic andferruginous).(ii) I developed more detailed conceptual frameworks that enable nuanced interpretations ofseawater redox conditions that rely on Cr-isotope and Fe-speciation signals preserved in the rockrecord.(iii) I constructed computer models that incorporate Cr and Fe proxy systematics to supportreconstructions of the evolution of Earth’s surface chemistry through time.Chapter 2: Cr isotope fractionation in marine hydrothermal sedimentsThis chapter explores the Cr concentration and isotopic composition of hydrothermal sedimentsdeposited on the eastern flank of the South East Pacific Rise (SEPR). In this chapter I test20the hypothesis that the δ53Cr composition of marine hydrothermal sediments is linked to theoxygenation content of deep ocean water. I elucidate links between Cr, Fe and Mn cycling inhydrothermal environments and reveal that the δ53Cr composition of the SEPR sediments isanomalously light compared to the δ53Cr composition determined in all other oxygenated marinesediments previously. I propose in this chapter that light δ53Cr compositions preserved in marinehydrothermal sediments may be a diagnostic signal for the evolution of oxygenated conditions inthe deep ocean, which may be traced through time.Chapter 3: Cr isotope fractionation in ferruginous sedimentsThis chapter places constraints on the magnitude of Cr isotope fractionation during Cr(VI)reduction in ferruginous sediments. In this chapter I test the hypothesis that diagenetic Cr(VI)reduction in ferruginous sediments imparts a characteristic isotope fractionation linked to theavailability of different Fe-species and the depth of oxygen penetration into the sediment pile.Detailed experiments and models developed in Chapter 3 reveal that Cr(VI) reduction has adistinguishing isotope fractionation in ferruginous sediments linked to the depth of oxygenpenetration in the sediment pile and diagenetic reactions with different Fe-species. In this chapterI propose that this has implications for interpreting the δ53Cr record contained in ancient Fe-richsediments, which may hold clues as to the emergence of oxygenated seafloor environments.Chapter 4: Ferruginous conditions during OAE1aIn Chapter 4, I test the hypothesis that the redox state of seawater during oceanic anoxic event1a (OAE1a) was ferruginous. Chapter 4 elucidates that the prevailing redox state of the globaloceans during OAE1a, was indeed ferruginous. Through geochemical analyses of sedimentsdeposited during OAE1a and geochemical modelling, I propose that the development of anoxicand Fe-rich conditions at this time is linked to a catastrophic depletion of the seawater sulphatereservoir. I further propose that extremely rapid dynamics in Earth surface oxidant budgets asobserved during OAE1a, may be a more common feature of the Phanerozoic than previouslyimagined.Chapter 5: Fe-cycling and magnetite formation in a modern ferruginous environmentThis chapter investigates Fe cycling in the ferruginous water columns and sediments of LakeMatano and Lake Towuti, Indonesia. I test the hypothesis that magnetite, a multivalent Fe-mineral,21forms in the water columns of these two ferruginous lakes. Through Fe-speciation and microscopicwork I show for the first time that magnetite forms pelagically in this environment. I propose thatpelagic magnetite generation may be a key component of iron formation (IF) genesis and mayplay a previously overlooked role in the coupling of the Fe and C biogeochemical cycles operatingunder ferruginous conditions. I further propose that the preservation of water column magnetitein ferruginous sediments may also serve as an important biosignature.Chapter 6: ConclusionsThis chapter summarizes the findings of my work on the Cr isotope and Fe-speciationpaleoroedox proxies. This chapter also addresses knowledge gaps and future challenges presentin the study of transition metals and their continued development as paleoredox proxies.22Chapter 2Chromium isotopes in marinehydrothermal sedimentsHydrothermal chromium (Cr) cycling contributes to marine Cr inventories and their Cr isotopiccomposition, yet Cr isotope effects associated with such processes remain poorly documented.Here we determine the distribution, isotopic composition, and diagenetic mobility of Cr inhydrothermal sediments from the flank of the South East Pacific Rise (SEPR, DSDP-site 598). Wefind that Cr is primarily associated with the metalliferous iron (oxyhydr)oxide component of thesediment (0.4 – 3.6 ppm), whereas Cr concentrations are much lower in the dominant carbonatephase (< 0.03 ± 0.2 ppm). The Cr:Fe ratio of the metalliferous component, however, decreaseswith increasing depth below the sediment water interface, with an apparent loss of > 80% Crfrom the sediment relative to Fe. We propose this loss is tied to oxidation of authigenic Cr(III) toCr(VI) followed by diagenetic remobilization and efflux from the sediment pile. The bulk δ53Crcomposition of the SEPR sediments is isotopically light (-0.24 to -0.57h), and we argue this resultsfrom the partial reduction of oxic seawater-bearing Cr(VI) by reduced hydrothermal vent fluidsenriched in Fe(II)aq. Diagenetic oxidation, furthermore, of the reactive Cr pool by Mn-oxides andloss of Cr(VI) from the sediment may further deplete the sediment in 53Cr during diagenesis. Theδ53Cr composition of the detrital Cr fraction of the sediment (average δ53Cr composition = -0.05 ±0.04h) falls within the igneous silicate earth (ISE) range, revealing that detrital Cr has a marginalinfluence on the bulk sediment δ53Cr composition at this location. Together our results showthat light δ53Cr compositions in hydrothermal sediments are imparted through a combinationof processes previously overlooked in the marine Cr biogeochemical cycle, but the overall δ53Crcomposition of such sediments may provide a rich source of information on paleoredox conditions.232.1 IntroductionChromium (Cr) isotope ratios are emerging as proxy that can track the evolving redox stateof the ocean-atmosphere system over geologic time-scales [4, 7, 30, 39, 104, 145]. The simplestapplication of the Cr isotope system uses large amplitude shifts in the Cr isotopic composition ofsedimentary rocks, notably banded iron formations (BIF) and ironstones, together with shales andlimestones to record the onset of Cr redox cycling, which is indicative of the presence of oxygenin Earth’s surface environment [4, 7, 30, 31, 40]. However, to provide more insight into the Crisotope system and pursue more nuanced applications of the proxy, a more refined view of globalCr cycling is required. Specifically, improving knowledge of the distribution of Cr isotopes in awider range of sedimentary reservoirs that serve as Cr repositories and archives of past Earthsurface chemistry is essential for the development of a global isotope mass balance. Furthermore,exploration of the modern Cr cycle also provides a means to test many of the assumptions builtinto the current interpretation of the Cr isotope record.Our emerging picture of the global Cr cycle is illustrated schematically in (Fig. 1.1, Chapter 1).Chromium resides in the continental crust bound in its relatively low solubility trivalent redoxstate Cr(III), and is hosted in silicate and oxide minerals [47, 50]. In the modern Earth system,with O2 comprising 21% of the atmosphere, this Cr is liberated to runoff predominantly throughoxidative weathering at Earth’s surface, which produces Cr(VI) from Cr(III) through heterogenoussolid-state catalysis by Mn-oxides [51, 61]. It has also been suggested that Cr(III) may also beoxidized by H2O2 in the local absence of O2 in serpentinizing systems [63], however available O2is ultimately required to generate this peroxide and other oxidants of Cr at the outset. Riverinetransport of Cr(VI) produced during oxidative weathering sustains the largest Cr flux to themodern oceans [25, 76, 77]. Once in the oceans, this Cr(VI) is likely removed from seawater viareduction by an as yet unknown diversity of electron donors, to Cr(III) which, in turn, is scavengedby settling particles and ultimately deposited in sea-floor sediments [37, 104–106, 108]. A fractionof Cr may also be delivered to the oceans via rivers as Cr(III), partly stabilized as Cr(III)-organiccomplexes, or dissolved in low pH systems [47, 51, 146].The contribution of hydrothermal systems to the marine Cr inventory remains poorly con-24strained, and may represent a net source or sink of Cr in the global marine Cr mass balance.Previous work suggests Cr sourced from hydrothermal vents represents 0.02% of the Cr riverineflux, or 12 x 106 mol yr – 1, implying that the contribution of Cr to seawater from hydrothermal flu-ids is likely very small [25]. While vent fluids emit Cr(III) derived from acid dissolution of basalticocean crust, much of this Cr(III) is scavenged by Fe (oxyhydr)oxide particles formed during themixing of Fe(II)-laden vent fluids with ambient oxygenated seawater [90, 91, 94, 100]. Deposition ofthese Fe (oxyhydr)oxides may thus quantitatively capture the vent flux of Cr, which is then largelydeposited in proximal hydrothermal sediments [95]. However, given the typically strong vent fluxof ferrous Fe – a potent reductant of Cr(VI) – precipitation of hydrothermal Fe (oxyhydr)oxidesmay also capture a fraction of seawater Cr(VI), rendering vent systems a net sink for seawater Cr[95]. A fraction of hydrothermal Fe particles can also be transported great distances from theirvent sources [94, 96, 147]. While the fate of Cr in these particles remains unknown, long distancetransport of hydrothermal Fe and associated elements may distribute hydrothermal Cr over longdistances in the oceans. Cr(III) associated with Fe (oxyhydr)oxides generally has a low solubility(e.g., [50]) and, therefore, deposition of Cr in hydrothermal sediments may lead to effective andquantitative burial of hydrothermal Cr with such phases. It has also been suggested, however, thatthere is extensive Cr remobilization in surficial continental-margin Mn-oxide bearing sediments[148], which implies the potential for Cr remobilization from Mn-oxide bearing hydrothermalsediments. Additional constraints on the fate of Cr deposited in hydrothermal sediments are thusneeded to better constrain the role of vent systems in the marine Cr cycle.Chromium isotopes are fractionated during redox reactions. During oxidative weathering, theresulting Cr(VI) is enriched in the 53Cr (heavy) isotope through a possible combination of Cr(III)oxidation to Cr(VI), and partial re-reduction of Cr(VI) [30, 73]. Despite current uncertainty aroundthe precise mechanism for this fractionation, heavy δ53Cr(VI) is exported from the weatheringenvironment in runoff, leaving Cr(III) in the residual soil isotopically light (53Cr depleted) [30,31, 44, 149]. Oxidative weathering thus sets the heavy δ53Cr composition observed in some rivers(e.g. [31, 44, 45]), and contributes to the heavy isotopic composition of seawater (modern valueslie between 0.5 and 1.5h, [1, 41]). On the other hand, dissolution of Cr(III), unless mediatedby strongly complexing organic acids (see [86]), is unlikely to be associated with large isotope25fractionation [30, 97]). Given that acid dissolution of basalt provides Cr to vent fluids, the efflux ofCr(III) from hydrothermal vent waters is expected to carry an isotopic signature no different fromthe igneous silicate earth (ISE), -0.12 ± 0.10h [46]. As a result, Cr(III) escaping removal via Fe(oxyhydr)oxide particle scavenging in the vicinity of vents, would likely contribute a Cr isotopeflux indistinguishable from ISE to the marine Cr inventory, with potential to partly offset heavysignals imparted through the riverine influx of Cr(VI) [30, 46]. Hydrothermal Cr(III) precipitates,however, might also be expected to carry a lighter but variable Cr isotopic composition, ifdiagenetic reactions remobilized Cr as result of partial oxidation or, if hydrothermal particlescaptured seawater Cr(VI) through partial reduction.To assess Cr deposition and cycling in hydrothermal settings we have determined the abun-dances and distribution of Cr, Mn and Fe, together with Cr isotope ratios, from the western flankof the South East Pacific Rise (SEPR), one of the fastest spreading centres on the ridge crest systemwith rates of up to 14.5 cm yr – 1 [150]. Three sediment cores from the SEPR at DSDP sites 598, 599and 600 were examined, with a focus on the more distal site 598, which is currently 1130 km westof the ridge crest. These sites have collected particulate hydrothermal materials derived from thecrest of the SEPR spreading center and transported by westward-flowing deep-ocean circulation.Sequential extraction techniques were employed to target Cr in different mineral phases of thesediment, and we used the molar ratios of Cr, Fe and Mn to evaluate post-depositional alteration.We then assessed the implications of this redistribution for marine Cr budgets, ocean isotope massbalance, and the use of Cr isotopes as a proxy for the oxygenation state of the oceans.2.2 MethodsThe three sediment cores used in this study were collected during DSDP Leg 92 in 1984. Theywere previously used by [150], to investigate the bulk sediment major element geochemistryand the history of hydrothermal sedimentation on the south-eastern flank of the SEPR. Previousworkers [151], conducted a study of the association of P with Fe (oxyhydr)oxides and other phasesin the upper 5 m at site 598. This site represents the most distal location from the ridge crest andvia seafloor spreading is now located 1130 km from the active vent. Using splits from the same26sample suite [151] we constructed a profile from the core surface to 5.5 m depth, corresponding toan age of 5.5 Myr at 5.5 m depth [151]. To investigate vent particle alteration processes, Mn-oxideand Fe concentrations were determined in 5 samples across the top 20 cm of cores from the othersites (599 and 600), which are closer to the ridge crest, allowing us to track chemical changes inthe Mn:Fe ratio of surface sediments in relation to the proximity to the ridge crest. The sedimentsat all three sites are reported to be compositionally simple, mainly comprising two phases, namelyforaminferal and nannofossil carbonates (37 - 95%) and metalliferous hydrothermal precipitates(3 - 40% of bulk sediment) [152]. Importantly, the sediments contain only small amounts oflithogenous detrital material as the sample locations are well removed from continental inputsources [152].2.2.1 Chemical extraction techniquesA selective extraction scheme was used to quantify Cr in the different operationally defined phasesof the hydrothermal sediment. Figure 2.1 is a schematic illustration of these procedures.6 N HCl  (FeTOT)Na-dithionite  (FeGOE)Hydroxylamine HCl  (FeHFO)FeTOT - (FeGOE + FeHFO)  (Fesil)6 N HCl  (L6NHCl)NH2OH-HCl  (MnOx)0.5 N HCl  (LHFO)Na-dithionite  (LGOE)a) b)Bulk Sediment Leach 1Bulk Sediment Leach 2Sequential Extraction 1Bulk Sediment Leach 1Sequential Extraction 16 N HCl  (LSil)Na-Acetate (FeCarb)Acetic Acid (LCarb)Figure 2.1: Leach procedure schematic. a) Sediment extraction techniques applied in this study. The chemicalextraction techniques are denoted beside each operationally defined sediment phase with the leachdesignation abbreviation in brackets (refer to Table 2.1). b) Sediment extraction techniques applied in [151]on identical sediment sections. We note that in [151] metal contents of the detrital silicate fraction of thesediment (FeSil) was calculated as the difference between FeTOT and (FeGOE + FeHFO) fractions.27We employed four different leaches in sequence, adapted from the procedures of [129], and twoseparate bulk digests. Table 2.1 contains a detailed breakdown of our leaching procedures andhow they relate to the nomenclature employed in [151].Table 2.1: Summary of extraction techniques. †[152]‡[151]?[153][129]κ[154]Fraction designation Sediment fraction Extractant Sequential Digests LCarb Carbonate 10% acetic acid, 4 h†  LHFO Poorly crystalline hydrous ferric oxides (FeHFO in ‡) 0.5 N HCl, 1 h1 LGOE Ferric (oxyhydr)oxides (FeGOE in ‡) 0.35 M acetic acid/0.2 M Na-citrate Na-dithionite, 2 h◊  LSil Silicates (FeSil in ‡) Near boiling 6 N HCl: 24 h◊ Bulk Sediment Digests L6NHCl FeTOT Near boiling 6 N HCl: 24 h◊ MnOx Mn Oxides 0.1 M NH2OH-HCl, 2 hk  †(Barrett,	et	al.	1987)	‡(Poulton	and	Canfield	2006)1	,(.)  ◊(Poulton and Canfield 2005) k(Neaman, et al. 2004)				Barrett,	T.	J.,	P.	N.	Taylor,	and	J.	Lugowski		 1987	 Metalliferous	sediments	from	DSDP	Leg	92:	The	East	Pacific	Rise	transect.	Geochimica	Et	Cosmochimica	Acta	51(9):2241-2253.	Neaman,	A.,	et	al.		 2004	 Improved	methods	for	selective	dissolution	of	manganese	oxides	from	soils	and	rocks.	European	Journal	of	Soil	Science	55(1):47-54.	Poulton,	S.	W.,	and	D.	E.	Canfield		 2005	 Development	of	a	sequential	extraction	procedure	for	iron:	implications	for	iron	partitioning	in	continentally	derived	particulates.	Chemical	Geology	214(3-4):209-221.	—		 2006	 Co-diagenesis	of	iron	and	phosphorus	in	hydrothermal	sediments	from	the	southern	East	Pacific	Rise:	Implications	for	the	evaluation	of	paleoseawater	phosphate	concentrations.	Geochimica	Et	Cosmochimica	Acta	70(23):5883-5898.	Thamdrup,	B.,	H.	Fossing,	and	B.	B.	Jorgensen		 1994	 Manganese,	iron	and	sulfur	cycling	in	a	coastal	marine	sediment,	Aarhus	Bay,	Denmark.	Geochimica	Et	Cosmochimica	Acta	58(23):5115-5129.		The first leach in sequence (LCarb) targeted the carbonate fraction. Powdered samples of 200 mgwere weighed into centrifuge tubes and leached with 10% acetic acid [152]. The acid additionwas in excess of that needed to dissolve 200 mg of pure CaCO3. Samples were placed on ashaker for 4 hours and subsequently centrifuged. An aliquot of supernatant was taken anddiluted for Cr concentration analyses via inductively coupled plasma mass spectrometry (ICP-MS)and inductively coupled plasma optical emission spectrometry (ICP-OES). A subset of sedimentsamples were analysed for total inorganic carbon via coulometry before and after application ofthe first leach to ensure all CaCO3 was successfully dissolved. Following re oval of carbonate, theresidual sediment was subjected to the second leach in sequence, which targets poorly crystallineFe-(oxyhydr)oxide phases (lepidocrocite, ferrihydrite) [153] using trace metal grade 0.5 N HCl[153] (LHFO, which is broadly analogous to the FeHFO fraction in [151]). The supernatant wasdecanted, filtered, and diluted for measurement via ICP-MS.Following the LHFO extraction, the remaining sediment contained more crystalline metal(oxyhydr)oxides as well as refractory detrital components. In order to differentiate between themore crystalline authigenic metal (oxyhydr)oxides (e.g., goethite and hematite) and refractorydetrital minerals, we first subjected the residual sediment to 0.35 M acetic acid/0.2 M Na-citratebuffered Na-dithionite leach for 2 hours (LGOE, which targets minerals such as goethite and28hematite, and is referred to as the FeGOE fraction in [151]. The supernatant was decanted followingcentrifugation, filtered, and diluted for measurement via ICP-OES. We did not directly determinethe Cr concentration of this sediment phase analytically due to perceived matrix effects of theLGOE extract in the quadrupole ICP-MS, and thus estimated the Cr concentration of this leachusing a mass balance approach (see Section 3.2 below).Following the LGOE extraction a small volume of greyish residual sediment remained, compris-ing the refractory component of the sediment (not extractable in the LCarb, LHFO or LGOE fractions).We leached the residual sediment from LGOE with near boiling 6 N HCl for 24 h (LSil, analogousto the FeSil fraction in [151]). Again, samples were centrifuged and an aliquot of the supernatantwas removed and diluted for analysis via ICP-MS and ICP-OES. We also performed a separatebulk digest with 6 N HCl on 200 mg sediment by adding 10 ml of acid in sealed centrifuge tubes(L6NHCl). Again, samples were centrifuged and an aliquot of the supernatant was removed anddiluted for analysis via ICP-MS and ICP-OES.The reactive Mn and Fe (oxyhydr)oxide content in sediments from the upper 10 cm at sites 599and 600, and from 5.5 m at site 598, were determined following the methods of [154, 155]. Samplesof 7.5 mg were weighed into centrifuge tubes and leached for 2 h with 0.1 M hydroxylaminehydrochloride (LMnOx). The amount of NH2OH – HCl added was in a 1:2000 solid – solutionratio. Samples were centrifuged and an aliquot of supernatant was taken for analysis by ICP-OES.Reaction time was limited to 2 h to restrict dissolution of poorly crystalline iron-oxides [154]. Ironconcentrations in these extracts accounted for only 0.01% of the bulk Fe in the sediments.2.2.2 Analytical methodsChromium concentrations in our leachates were determined by ICP-MS, while major elements Al,Fe and Mn were determined by ICP-OES. The error on [Fe] measurements was ± 8% based onthe difference between triplicate measurements, and the detection limit, calculated as three timesthe standard deviation of the blank (n = 5) was 8 ppb in solution and roughly 35 ppm in the solidphase based on our dilutions. The error on our ICP-OES [Mn] measurements was ± 5% based onthe difference between triplicate measurements, and the detection limit, calculated as three timesthe standard deviation of the blank (n = 5) was 6 ppb in solution and roughly 30 ppm in the solid29phase based on our dilutions. The error on [Cr] measurements is ± < 1% based on the differencebetween triplicate measurements, and the detection limit, calculated as three times the standarddeviation of the blank (n = 5), translates to 0.028 ppm in the solid phase.2.2.3 XRD analysisWe used quantitative X-ray diffraction (qXRD) methods on 6 bulk sediment samples to determinethe sediment mineralogy. The samples were smear mounted with ethanol on non-diffractingsilica plates. Continuous-scan X-ray powder-diffraction data were collected over a range 3-80◦2θwith CoKα radiation on a Bruker D8 Focus Bragg-Brentano diffractometer equipped with an Femonochromator foil, 0.6 mm (0.3◦) divergence slit, incident- and diffracted-beam Soller slits and aLynxEye detector. The long fine-focus Co X-ray tube was operated at 35 kV and 40 mA, usinga take-off angle of 6◦. We analyzed the X-ray diffractograms using the International Centre forDiffraction Database PDF-4 and Search-Match software by Bruker. X-ray powder-diffraction dataof the samples were refined with Rietveld program Topas 4.2 (Bruker AXS).2.2.4 Cr purification and isotope ratio determinationChromium isotope ratios were determined on a Multi-Collector Inductively-Coupled Plasma MassSpectrometer (MC-ICP-MS) at the Yale Metal Geochemistry Center using a double-spike correctionfor isotope fractionation during column chemistry and instrumental mass bias. The spike wasadded to acid splits after the digestion procedure (bulk sediment 6 N HCl digestion, L6NHClleachate as well as the LSil 6 N HCl digestion performed sequentially). Although isotope effectsdue to leaching in 6 N HCl have not been tested directly, we do not expect any measurable isotopefractionation during acid digestions of Cr, even at higher acid molarity (see [25, 30]). We note thatprevious studies on Fe isotope ratios in rocks and sediments have also shown that there appearsto be no isotope effect during proton-promoted acid leaching of Fe oxides, which is in contrast toligand promoted dissolution [156]. We used three separate column separations [46, 105] to ensurecomplete removal of Fe, Ti, and V during chemical purification. Initially, a split of the 6 N HClsample solution was spiked, evaporated to dryness, and the residues brought up in 1 N HCl. Theamounts of 50Cr-54Cr double-spike (50Cr-52Cr = 462.917, 53Cr/52Cr = 0.580, 54Cr/52Cr = 354.450,30calibrated in the Department of Geology, University of Illinois at Urbana-Champagne and addedas Cr(III)), were adjusted so that the spike/sample ratio (i.e., (54Cr)spk/(52Cr)smp) was 0.5. Prior tothe first column step (AG1-X8 anionic resin, 100-200 mesh), Cr(III) was oxidized to Cr(VI) usingpotassium peroxidisulfate (heating the samples for 2 h at 110◦C). The AG1-X8 resin was cleanedwith mQ water, 3 N HNO3 and 6 N HCl. The matrix was eluted with 24 ml of 0.2 N HCl and 4ml 2 N HCl, with Cr subsequently reduced with 5% H2O2 and eluted with 5 ml of 2 N HNO3.The second step removes traces of Fe that may remain after the first elution. Microcolumns werefilled with 0.2 ml AG1-X8 resin. The columns were cleaned using mQ water and 3 N HNO3, andsamples were loaded and collected with 1.2 ml of 6 N HCl. Traces of Ti are often left after the firsttwo column steps. Therefore, as the last step, a cation resin AG50W-X8 (200-400 mesh) was usedto ensure complete Ti and Cr separation. The resin was cleaned with mQ water, 3 N HNO3 and 6N HCl, followed by sample loading in 3 ml of 0.5 N HNO3 and by matrix elution with 1 ml of 0.5N HNO3, 2 ml of 0.5 N HF and 6 ml of 1 N HCl. Cr was collected after loading with 4 ml of 0.5 NHNO3 and after elution with 5 ml of 1.8 N HCl.Chromium isotopes were measured on a NeptunePlus (Thermo-Finnigan) MC-ICP-MS using amodified version of the analytical protocol of [46]. Samples were run in high-resolution mode toresolve polyatomic interferences such as 40Ar12C+, 40Ar14N+ and 40Ar16O+. Although our chemicalprocedure results in nearly complete removal of Fe, Ti and V, these elements were monitored bymeasuring 56Fe, 49Ti and 51V, and samples were corrected for potential interferences of 54Fe on54Cr, as well as 50Ti and 50V on 50Cr. Samples at a concentration of ∼100 ng g– 1 were introducedinto the plasma with a PFA µFlow nebulizer (∼50 µL min– 1) coupled with an Apex IR (ElementalScientific) without additional N2 gas or membrane desolvation. With a standard sample cone andX skimmer cone under high-resolution mode, the sensitivity obtained was ∼310 – 10 A 52Cr on100 ng g– 1 solution. All ion beams were measured on faraday detectors. The double-spike datareduction model is based on the iterative method described by [157]. A spiked Cr isotope standardNIST SRM 979 was measured bracketing every three natural samples to ensure machine stability.Chromium isotope ratios are reported relative to bracketing standards using conventional deltanotation; (δ53Cr = [(53Cr/52Cr)sample/(53Cr/52Cr)NIST – 979 – 1] x 1000h).External precision is reported as two sigma (2σ) uncertainty, calculated based on duplicate31analysis of geological reference materials (GRMs) processed through ion-exchange chromatog-raphy columns along with samples (BHVO-2 and Nod-A-1 were systematically processed with70 samples). The δ53Cr value for BHVO-2 is -0.11 ± 0.07h (n = 3), which is similar to publishedvalues in the literature for BHVO-1 (geostandard collected at the same location as BHVO-2) [46],and Nod-A-1 yielded a δ53Cr value of 0.08 ± 0.05h (n = 4). Sample duplicates, including columnprocedure duplicates, digested duplicates, and replicate measurements on the MC-ICP-MS re-vealed a 2σ uncertainty similar to that determined for GRMs, i.e. 0.09h and an average duplicateoffset of 0.043h. Measurement precision is calculated via replicate measurements of the isotopicstandard NIST SRM 979 during each analytical session (two standards bracket every sample), and2σ values are better than 0.04h. In addition, a two-standard deviation of the mean (2se) wassystematically calculated using the 50 cycles of measurement obtained for each sample duringMC-ICP-MS analysis, and was generally 0.04 - 0.06h on the delta values. This precision andaccuracy are comparable to long term means at the Yale Metal Geochemistry Center.2.3 Results2.3.1 Mineralogical analysesQuantification of the relative mineral proportions via qXRD on 6 sub-samples (Table 2.2) showsthat the sediment samples examined here consist almost entirely of calcite and goethite, with verysmall contributions from crystalline lithogenic minerals (< 12% ; Table 2.2), which are inverselyproportional to calcite concentration.32Table 2.2: Mineralogical analyses of the sediments from site 598.Sample Depth (cm)  2.5 42.5 75.6 246.5 398 499.5 % Albite  1 2 4 8  2 % Calcite 99 85 78 56 97 83 % Goethite  11 14 33 2 15 % Hercynite  1  2 1  % Kaolinite   2 1   % Quartz low   2    Total 100 100 100 100 100 100 	The detrital fraction is dominated by the feldspar albite, a felsic mineral, in addition to about 1%hercynite, a spinel phase (Fe2+Al23+O4) into which Cr can substitute for Al (Fe2+Cr23+O4) (Table2.2).2.3.2 Site 598 elemental concentrations and Cr mass balanceWithin our sequential leach scheme we first targeted the carbonate phase (LCarb, Table 2.1).Sediments from site 598 have CaCO3, contents ranging from 42.8 to 93.3 ± 0.1 wt% [158]. Asnoted previously, this fraction is dominated by coccoliths (comprising roughly 85% ), with a small(up to 4% ) fraction of planktonic foraminifera [150, 152]. Carbonate concentration in the sedimentare inversely related to metalliferous phases and this leads to mutual dilution of both authigenicand detrital metals in carbonate rich intervals. Triplicate measurements of Cr concentrations inthis phase were below detection limit (0.03 ppm) in all of the samples from site 598. The low Crconcentrations in the carbonates at site 598 are consistent with the linear relationship between Crand Fe in the L6NHCl bulk sediment leachate (Table 2.1), calculated for samples deeper than 15033cm (Fig. 2.2) where the average molar Cr:Fe ratio remains nearly constant (Fig. 2.2).0 50,000 100,000 150,000[Fe] (ppm)[Cr] (ppm)Figure 2.2: Two end member-mixing diagram. Displayed are the Cr and Fe concentrations in the L6NHClleach for samples below 250 cm, where the Cr:Fe ratio remains constant. The intercept of 0.04 ± 0.3 ppmprovides the [Cr] expected in the carbonate phase. The dotted lines represent a 90% confidence belt on theregression.This linear trend indicates two end-member mixing between Cr deficient carbonate and the higherCr contents of the metalliferous Fe-rich phases. The intercept (0.041 ± 0.2 ppm) provides anindependent estimate of the Cr concentration in the CaCO3, which is near the detection limit ofour analytical method (0.03 ppm).The bulk sediment (6 N HCl extractable phases, L6NHCl, Table 2.3) Fe concentration determinedin this study is within ± 5% of the FeTOT values reported by [151].34Table 2.3: Element concentrations and Cr isotope composition determined in the bulk (L6NHCl leach) site598 sediments. We note that the detrital Cr concentration was determined as the ratio of CrSil to total Cr inthe Cr6NHCl leach; thus % Cr detrital = (CrSil / Cr6NHCl) * 100%. ND “Not Determined”.Bulk Sediment Leach   Depth (cm)  Element concentrations in L6NHCl (6 N HCl) Calculated Detrital fraction Calculated δ53Cr Authigenic (‰) Calculated 2se (‰) Cr (ppm) Fe (wt%) Al (wt%) δ53Cr (‰) 2se (‰) Cr (%) δ53Cr (‰) 2se (‰ 0.5 1.1 1.3 ND -0.35 0.05 ND ND ND 2.5 1.2 1.2 ND -0.37 0.05 ND ND ND 4.5 1.3 0.9 0.1 -0.32 0.05 53 -0.7 0.2 8.5 1.0 1.1 0.2 -0.33 0.05 42 -0.1 0.2 21 1.5 2.5 0.3 -0.3 0.05 48 -0.4 0.3 26.5 1.6 2.3 0.3 -0.29 0.05 61 ND ND 34.5 2.0 3.7 0.5 -0.28 0.05 51 -0.4 0.3 42.5 3.1 4.7 0.6 -0.37 0.04 50 -0.6 0.2 52.5 3.2 5.5 0.7 -0.28 0.04 ND ND ND 68.5 3.6 6.5 0.8 -0.24 0.05 ND ND ND 76.5 3.4 7.2 0.9 -0.3 0.05 59 -0.6 0.2 84.5 3.2 6.6 0.8 -0.34 0.05 63 -0.7 0.2 101 2.6 4.8 0.6 -0.36 0.04 ND ND ND 110.5 2.0 3.7 0.4 -0.34 0.05 58 -0.7 0.2 121.5 1.9 3.4 0.4 -0.35 0.11 81 -0.7 0.4 138.5 1.2 2.3 0.3 -0.27 0.06 54 -0.4 0.3 150.5 1.4 2.8 0.3 -0.27 0.06 52 -0.9 0.3 166.5 1.3 2.9 0.3 -0.28 0.05 46 -0.4 0.3 186.5 1.3 3.4 0.4 ND ND 51 ND ND 196.5 1.4 3.8 0.4 -0.37 0.06 51 -1.4 0.2 206.5 1.8 7.1 0.7 ND ND 45 ND ND 216.5 2.3 8.9 0.8 -0.38 0.05 43 -0.6 0.2 236.5 2.7 ND ND -0.5 0.04 ND ND ND 246.5 2.6 15.0 1.4 -0.37 0.04 41 -0.6 0.2 266.5 2.3 11.2 1.0 -0.45 0.04 37 -0.6 0.2 277.5 2.1 9.1 0.8 -0.57 0.04 41 -0.8 0.2 286.5 1.5 6.4 0.5 -0.47 0.04 41 -0.7 0.2 306.5 0.9 2.8 0.2 ND ND 51 ND ND 326.5 0.8 2.6 0.2 NA NA 56 ND ND 347.5 1.4 5.7 0.5 -0.44 0.05 46 -0.6 0.2 398.5 0.7 2.3 0.2 ND ND 56 ND ND 424.5 0.6 2.4 0.2 ND ND 45 ND ND 499.5 1.6 7.1 0.5 -0.56 0.05 39 -0.8 0.2 549.5 0.4 1.2 0.1 ND ND 39 ND ND 	Chromium concentrations in the bulk sediment L6NHCl leach range from 0.4 – 3.6 ppm, and arestrongly correlated with Fe [151] – the highest Cr concentrations are associated with Fe-richsamples, whereas carbonate-rich samples are low in Cr. The mass balance of Cr in the individualmetalliferous sediment phases of the site 598 sediments is plotted in Fig. 2.3.354.58.521.026.534.542.576.584.5101.0110.5121.5138.5150.5166.5186.5196.5206.5216.5236.5246.5266.5277.5286.5306.5347.5398.5424.5499.5549.5Depth (cm)0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0Cr (ppm)L6NHClLHFOLGOELSil0 50 100CrPhase/Cr  L6NHCla bFigure 2.3: Mass balance of Cr in the site 598 sediments. a) Concentrations of Cr in the LHFO, LGOE and LSilsequential leaches. We also plot the bulk Cr concentration of the of the sediment (L6NHCl leach). b) Relativeabundance of Cr in each sediment fraction normalized to the bulk Cr in the L6NHCl leach. We note thatdepth intervals where [Cr] was not determined in all leach fractions (see Tables 2.4 and 2.3) have beenomitted from the profiles.There is a minor fraction of the Cr contained in the LHFO fraction (Fig. 2.4b), dominantly comprisedof highly reactive hydrous ferric oxide minerals such as ferrihydrite [151], and this is mostlyrestricted to the upper 150 cm (Fig. 3.3).360.0 2.0 4.0Mn-Oxide (wt%)0100200300400500600Depth (cm)0.0 2.0 4.0MnOxide:Fe6NHClMnOxide (wt%)MnOxide:Fe Molar Ratio0.0 2.5 5.0Cr6NHCl (ppm)0.0 10.0 20.0Fe6NHCl (wt%)L6NHCl Cr (ppm)L6NHCl Fe (wt%)0.0 1.5 3.0CrSil (ppm)0.0 0.5 1.0FeSil (wt%)LSil Cr (ppm)LSil Fe (wt%)0.0 1.5 3.0CrGOE (ppm)0.0 10.0 20.0FeGOE (wt%)LGOE Cr (ppm)LGOE Fe (wt%)0.0 0.5 1.0CrHFO (ppm)0.0 0.4 0.8FeHFO (wt%)Cr Fe40 70 100CaCO3 (wt%)0100200300400500600Depth (cm)CaCO30.0 25.0 50.0Cr:FeHFO (x104)Cr/FeHFO0.0 3.5 7.0Cr:FeGOE (x104)Cr/Fe GOE0.0 0.5 1.0Cr:FeSil (x104)Crdet/Fedet0.0 6.0 12.0Cr:Fe6NHCl (x104)Cr/FeTOTa b c d ef g h i jFigure 2.4: Element concentrations at site 598. a) Carbonate concentration [151]. b) Cr (blue data points) andFe (orange data points) in the LHFO fraction. c) Cr and Fe in the LGOE fraction. d) Cr and Fe in the LSilfraction. e) Cr and Fe in the bulk L6NHCl fraction. f) Mn-Oxide (purple) and Mn-Oxide:Fe6NHCl (red datapoints). g) Cr:Fe ratio of the LHFO fraction. h) Cr:Fe ratio of the LGOE fraction. i) Cr:Fe ratio of the LSilfraction. j) Cr:Fe ratio of the L6NHCl fraction.The Cr concentration of the LSil fraction ranges from 0.3 – 2.5 ppm (Fig 2.4d), indicating that therefractory Cr pool, accounts for 37 – 82% of the total Cr (L6NHCl, Table 2.3). As discussed in section2.1, we were unable to analytically determine the Cr concentration of the LGOE leachate, and thuswe determine this via mass balance; LCrGOE = LCr6NHCl - (LCrHFO + LCrSil; Table 2.4). The majorityof Cr in the site 598 sediments is thus mainly contained in the LGOE (Na-dithionite extractable,Table 2.4) sediment fraction, comprising crystalline Fe (oxyhydr)oxide mineral phases such asgoethite, and the refractory LSil fractions (Fig. 2.4c-d and Table 2.4).The average Cr:Fe ratio of the sediments is 4.8 x 10 – 5, with a maximum value of 10.2 x 10 – 5and minimum value of 1.9 x 10 – 5. In the upper 150 cm of the sediment pile there is a clear and37Table 2.4: Element concentrations determined via sequential extractions in site 598 sediments. We note thatCr concentration of the LGOE fraction was determined as the difference between total Cr (Cr6NHCl, Table2.3) and the sum of the sequential phases; thus CrGOE = Cr6NHCl - (CrHFO + CrSil). ND “Not Determined”,BLD “Below Detection”.Sequential Extractions  Element concentrations in LHFO (0.5 N HCl) fraction Element concentrations in LGOE (Na-Dith) fraction Element concentrations in LSil (6 N HCl) Depth (cm) Cr (ppm) Fe (wt%) Al (wt%) Cr (ppm) Fe (wt%) Al (wt%) Cr (ppm) Fe (wt%) Al (wt%) δ53Cr (‰) 2se (‰) 0.5 0.3 ND ND ND ND ND ND ND ND ND ND 2.5 ND ND ND ND ND ND ND ND ND ND ND 4.5 0.7 0.2 0.0 0 1.0 0.1 0.6 0.1 0.1 -0.01 0.02 8.5 0.3 0.2 0.0 0.2 1.1 0.1 0.5 0.1 0.1 -0.28 0.11 21 0.3 0.3 0.1 0.3 2.4 0.1 1.0 0.2 0.1 -0.07 0.03 26.5 0.20 0.3 0.1 0.2 2.7 0.1 1.2 0.2 0.1 ND ND 34.5 0.16 0.3 0.1 0.5 3.6 0.1 1.3 0.3 0.1 -0.10 0.02 42.5 0.19 0.4 0.2 0.8 5.6 0.2 2.1 0.5 0.2 -0.06 0.03 52.5 ND ND ND ND ND ND ND ND ND ND ND 68.5 ND 0.7 0.2 ND 7.3 0.3 ND 0.5 0.3 0.04 0.06 76.5 0.19 0.4 0.3 0.7 7.3 0.3 2.5 0.7 0.3 -0.04 0.03 84.5 0.09 0.4 0.2 0.7 6.6 0.3 2.4 0.7 0.3 -0.06 0.03 101 0.3 0.4 0.2 0.1 5.4 0.2 2.2 0.3 0.2 -0.05 0.03 110.5 0.09 0.2 0.1 0.8 4.1 0.1 1.1 0.3 0.1 -0.06 0.03 121.5 0.08 0.2 0.1 0.3 4.0 0.1 1.5 0.3 0.1 -0.22 0.05 138.5 0.10 0.2 0.1 0.4 2.3 0.1 0.7 0.2 0.1 -0.08 0.03 150.5 0.06 0.2 0.1 0.5 3.0 0.1 0.8 0.2 0.1 0.15 0.03 166.5 0.05 0.1 0.1 0.6 2.9 0.1 0.7 0.2 0.1 -0.08 0.03 186.5 0.05 0.2 0.1 0.5 3.6 0.1 0.7 0.2 0.1 ND ND 196.5 0.03 0.2 0.1 0.6 4.6 0.1 0.8 0.2 0.1 0.32 0.05 206.5 0.05 0.3 0.2 0.6 7.0 0.2 1.1 0.4 0.2 ND ND 216.5 0.08 0.3 0.2 0.8 9.7 0.2 1.4 0.5 0.2 -0.04 0.05 236.5 0.07 0.5 0.4 0.6 12.0 0.3 2.0 0.7 0.3 -0.01 0.03 246.5 0.10 0.4 0.4 0.7 13.3 0.3 1.8 0.9 0.3 -0.03 0.03 266.5 0.05 0.4 0.3 0.9 11.7 0.2 1.3 0.7 0.2 -0.07 0.03 277.5 0.07 0.3 0.2 1.0 9.7 0.2 1.1 0.5 0.2 -0.08 0.03 286.5 0.04 0.2 0.1 0.7 6.4 0.1 0.7 0.3 0.1 -0.07 0.03 306.5 0.03 0.1 0.1 0.4 3.2 0.1 0.4 0.1 0.1 ND ND 326.5 BLD 0.1 0.1 ND  2.3 0.0 0.4 0.1 0.0 ND ND 347.5 0.05 0.2 0.1 0.7 6.1 0.1 0.7 0.3 0.1 -0.11 0.03 398.5 0.11 0.1 0.0 0.3 2.6 0.0 0.3 0.1 0.0 ND ND 424.5 0.07 0.1 0.0 0.2 2.5 0.0 0.3 0.1 0.0 ND ND 499.5 0.03 0.3 0.1 0.9 8.9 0.1 0.7 0.3 0.1 -0.05 0.05 549.5 0.03 0.1 0.0 0.1 1.5 0.0 0.3 0.0 0.0 ND ND Average          -0.04 0.04 	38systematic decrease in the ratio of total Cr:Fe (L6NHCl, Fig. 2.4j), below which the ratio remainsrelatively constant (Fig. 2.4g-h). These more deeply buried intervals have a Cr:Fe ratio (2.0 x10 – 5), which is 5.5 times lower than those at the core top (10.2 x 10 – 5), and 20 times lower thanfresh hydrothermal vent plume particles (40 x 10 – 5, [91], Fig. 2.5).0 500 1,000 1,500Distance (km) Molar Ratio (x105 )Vent Particle Composition  (Feely et al., 1996)Surface Sediment CompositionBelow 250cm CompositionFigure 2.5: Cr:Fe composition of vent particles and sediments deposited at site 598. Displayed is a comparison ofthe Cr:Fe molar ratios of suspended vent particles measured by [91], compared to the Cr:Fe composition ofhydrothermal sediments at site 598 (Cr and Fe concentrations from the L6NHCl leach).Within the upper 150 cm, we also observe a decrease in the amount of LHFO extractable Cr, and anincrease in LGOE extractable Cr (Fig. 2.3) and Fig. 2.4h).2.3.3 Mn-oxide concentrationsTo directly test the capacity of the SEPR sediments to oxidize Cr, we also determined reactiveMn-oxide concentrations in the sediment. The Mn (oxyhdr)oxide concentrations in the sediment atsite 598 follow similar trends to those of bulk Fe6NHCl (Fig. 2.4g), with a maximum concentration39of 3.9 wt% at the sediment water interface. Extractable Mn in the upper 20 cm of sediment at sites599 and 600 also shows that the total Mn:Fe (oxyhydr)oxide ratio also increases with distancefrom the ridge crest (Table 2.5).Table 2.5: Mn-oxide:Fe molar ratios in the upper 20 cm of the three samples sites.Site Distance from ridge crest (km) Mn:Fe Molar Ratio of Upper 20 cm 600 340 0.28 599 640 0.24 598 1130 0.39 	2.3.4 Chromium isotopesChromium extracted from bulk SEPR site 598 sediments with a 6 N HCl (L6NHCl) leach (Table 2.3,Fig. 2.6) has δ53Cr values that are significantly lower (-0.24 to -0.57h, 2 tailed t-test p-value = 4 x10 – 6 at the 95% confidence level) than the ISE (-0.12 ± 0.10h) and oxic abyssal ocean sediments(0.23 to -0.14h; [46, 104]).40a b-1.00 -0.80 -0.60 -0.40 -0.20 0.00 0.20δ53Cr (‰)0100200300400500Depth (cm)δ53CrBulk L6NHClδ53CrDetrital LSilAuthigenic Component-0.60 -0.40 -0.20δ53Cr (‰) Cr:Fe Molar Ratio Sediments Below 150 cmFigure 2.6: Cr-isotope composition of sediments deposited at site 598. a) Cr isotope composition of the site 598sediments. The grey shaded region represents the ISE range (-0.12 ± 0.10h). The large errors on the δ53Crcomposition of the estimated authigenic component result from error propagation in our mass balancecalculation; δ53Crauthigenic = (δ53Crmeasured – (δ53Crdetrital * X)) / (1-X). b) Relationship between the Cr:Feratio and δ53Cr composition of the bulk L6NHCl leach for sediments deposited below 150 cm. grey shadedregion represents a 95% confidence belt on the linear regression with an r2 value of 0.80.The bulk L6NHCl leach contains both the authigenic reactive Cr phases (LHFO and LGOE) as wellas the refractory Cr (LSil). There is a general decrease in δ53Cr with increasing sediment depthand decreasing Cr:Fe ratios (Fig. 2.6). The lightest δ53Cr values come from samples below 150 cm.We also determined the Cr isotope composition of the LSil sediment fraction, comprising only therefractory sediment phases, of which all δ53Crvalues fall within error of the ISE with an average41δ53Crcomposition of -0.05 ± 0.04h (Table 2.4, Fig. 2.6). The LSil leach was performed after themore reactive and likely authigenic phases of Cr had been extracted (LHFO and LGOE), and thuswe consider Cr contained in the LSil fraction to represent detrital Cr. The isotopic composition ofthe detrital LSil Cr in site 598 sediments is not significantly isotopically fractionated relative to theISE composition (2 tailed t-test p-value = 0.26 at the 95% confidence level).To estimate the isotopic composition of the reactive and likely authigenic component of thesite 598 sediments (the sum of LHFO and LGOE extractable Cr), we correct for detrital Cr with nearcrustal isotopic composition using a mass balance calculation;δ53Crauthigenic = (δ53Crmeasured–(δ53Crdetrital · X))/(1− X) (2.1)where δ53Crauthigenic is the Cr isotopic composition of authigenic Cr, δ53Crmeasured is the bulk Crisotope composition of our L6NHCl leach (Table 2.3), δ53Crdetrital is the Cr isotope composition ofthe detrital fraction of the sediment, which we take from our LSil isotope values. X is the fractionof detrital Cr in the L6NHCl leach. We use the Cr concentration of our LSil leachates from oursequential extractions as best estimates of the detrital Cr contribution (Table 2.4). The results ofour mass balance calculation are shown in (Fig. 2.6) and we note that the large uncertainties onthe δ53Cr composition of the authigenic component result from error propagation in our massbalance calculation. Our calculation demonstrates that once corrected for isotopic composition ofdetrital Cr, the δ53Crauthigenic values of the SEPR sediments are even lighter relative to the ISE thanthe measured L6NHCl bulk isotope compositions.2.4 Discussion2.4.1 Cr in the carbonate phasePrevious workers attribute the variability in metal concentrations in the sediment at site 598to changes in the accumulation rate of CaCO3 rather than to fluctuations in the depositionrate of the metalliferous hydrothermal component [151]. Biogenic carbonate, however, is thedominant sedimentary component at site 598 and so we first considered it might also be the42main carrier phase of Cr. We first considered the possibility that any calcite bound Cr releasedduring the carbonate leach was re-adsorbed onto the remaining Fe-phases. Cr adsorption to Fe(oxyhydr)oxides is maximal in a pH range of 5 - 6 and less than 30% is adsorbed below pH 4[159]. Although re-adsorption of calcite bound Cr to Fe (oxyhydr)oxides during our carbonateleach cannot explicitly be ruled out, our 10% acetic acid leach was at pH 4, hence we do notexpect significant adsorption. Modern planktonic foraminifera, furthermore, typically exhibitCr concentrations between 0.05 and 0.3 ppm [160], which is consistent with biogenic carbonateshaving lower Cr concentrations than limestones [40]. Experimental studies show that Cr(VI)incorporation into calcite (as CrO2 –4 ) increases with increasing Cr(VI) concentration in solution[161]. We applied the partition coefficient (Kd*) from this experimental work [161], where;Kd∗ = [Cr(VI)]solid/[Cr(VI)]solution (2.2)When [Cr]solution drops below a minimum of 10 mmol l– 1, Kd* reaches a maximum value of 95. Ifthis value is extended to Cr(VI) concentrations of 1.95 ± 0.15 nmol l– 1 observed by [76] in surfacewaters above the East Pacific Rise, we calculate an expected [Cr]solid of 0.2 ppm in calcite, alsobelow our detection limit. Together, our observations imply that a negligible fraction of the Cr inSEPR sediments is associated with carbonate phases.2.4.2 Cr isotope composition of oxygenated deep-sea sediments and their detritalinputsThe δ53Cr composition of the majority of previously measured marine sediments deposited fromoxygenated seawater fall within the ISE range (-0.12 ± 0.10h, [46, 104]), while ferromanganesenodules accumulating beneath oxygenated seawater are isotopically light (Table 2.6).43Table 2.6: Compilation of modern oxic pelagic sediment Cr isotope values. †[104]‡[46]?[162] δ53Crbulk (‰) δ53Crleach (‰) 2se (‰) Reference  -0.06 NA 0.04 †   0.06 NA 0.04  -0.06 NA 0.04  -0.06 NA 0.04  -0.01 0.02 0.04  -0.05 NA 0.04  -0.07 NA 0.04  -0.08 NA 0.04  -0.08 NA 0.04  -0.13 NA 0.04  -0.14 NA 0.04  -0.13 NA 0.04  -0.08 NA 0.04  -0.06 NA 0.04  -0.02 NA 0.04  -0.03 0.01 0.04  -0.03 0.01 0.04  -0.07 NA 0.04  0 0.1 0.04  0.23 NA 0.04  0.02 NA 0.04  -0.01 NA 0.04  -0.1 -0.21 0.04  -0.11 NA 0.04  -0.1 NA 0.04  -0.016 NA 0.048 ‡   -0.053 NA 0.048  -0.007 NA 0.048  -0.54 NA 0.03    -0.48 NA 0.04  -0.85 NA 0.04  -0.41 NA 0  -0.37 NA 0.03  -0.47 NA 0.04  -0.38 NA 0.02  -0.34 NA 0.02  -0.15 NA 0.04  -0.34 NA 0.03  -0.32 NA 0.05 Average -0.15 -0.014 0.05  	This yields an average δ53Crvalue of -0.15 ± 0.05h for oxic marine sediments. Leaching ex-periments conducted to isolate the authigenic Cr pool from these sediments also yielded δ53Crvalues that were not statistically different from the bulk sediment (-0.01 ± 0.05h, [104]). Currentdata for oxic deep-sea sediments, therefore, suggest heavily fractionated authigenic Cr poolscomprise a very minor fraction of the total sediment Cr, and thus the δ53Crcomposition of thesesediments reflects a detrital input with composition identical to the ISE [46, 104]. The sediments44at site 598 were also deposited under oxygenated deep-sea conditions, and yet in contrast to otheroxygenated sediments, are overwhelmingly isotopically light. Given that most Cr in oxic modernmarine sediments has been attributed to detrital inputs, we first considered the possibility that thelight δ53Cr compositions at site 598 originated from detrital Cr.The eolian dust flux at site 598 has been estimated [163] and comprises only 0.8 – 4.0 wt% ofthe bulk sediment in the upper 6 m of the sediment column. Our qXRD measurements verifythat the mineralogy of the sediment is mainly biogenic calcite and authigenic goethite, with traceamounts of albite (0 - 8% ), which likely represents the major mineralogical component of theeolian dust (Table 2.2). Albite (NaAlSi3O8) is not likely to contain Cr in its crystal lattice, but wehave detected the detrital spinel mineral hercynite (Fe2+Al2O4), which can contain Cr(III). Sincewe do find an appreciable amount of the total Cr in the site 598 sediment is contained in theLSil fraction (Table 2.3), we speculate this could comprise a refractory Cr component. We thusconsider the possibility that some or all Cr in our bulk L6NHCl leachates is detrital rather thanhydrothermal and/or authigenic, and that the isotopically light bulk Cr isotope composition of theSEPR sediments might have been inherited from eolian dust. Eolian dust might acquire light Cras the result of oxidative weathering of the continents, which leads to the removal of isotopicallyheavy Cr(VI) via runoff [1, 30, 31, 44, 45, 164], while the residual soil becomes isotopically light[30, 42, 149]. This isotopically light continental material may thus contribute to eolian dust fluxesto marine sediments, providing a possible explanation for the light δ53Cr values of the SEPRsediments, if they comprise mostly detrital Cr, which we evaluate below.Aluminium is typically employed as a tracer for detrital input in many types of marinesediment. Although Al is not enriched in hydrothermal particulates [147], the positive enrichmentrelationship between Fe, Al and Cr, may indicate a similar source for these metals with mutualdilution by carbonate components controlling metal concentrations as a function of depth. Anauthigenic origin for Al is a possibility, however, given that a considerable fraction (∼70% ) ofthe total Al in the sediment as reported in [150], is contained in the more reactive LHFO and LGOEsediment fractions. Thus Cr:Al ratios of our leach fractions cannot be used to determine thedetrital Cr content of the sediment. We further suggest that the relatively large component ofauthigenic Al in the SEPR sediments may have resulted from the formation of authigenic clay45minerals [165, 166], a process needing further study at the SEPR.To test the hypothesis that highly weathered detrital material imparted the light δ53Cr compo-sition in the SEPR sediments, we determined the δ53Cr composition of the detrital fraction, LSil,directly. As noted above, the detrital component of other marine sediments measured to datecarries a δ53Cr indistinguishable from ISE values, and like these sediments, the LSil fraction at site598 also carries δ53Cr values that are indistinguishable from the ISE (Fig. 2.6). This shows thateven though refractory Cr phases of likely detrital origin comprise an appreciable proportion ofthe hydrothermal sediment Cr, their isotopic composition is not responsible for imparting theisotopically light δ53Cr composition of the bulk sediment. Our Cr:Fe ratios, furthermore, pointto diagenetic remobilization of an appreciable reactive, and thus likely authigenic, Cr fractionwith little Cr redistribution occurring in the detrital LSil fraction (Fig. 2.4i). Taking these twoobservations together implies that the light SEPR δ53Cr is authigenic in origin, and if LSil capturesdetrital Cr, then eolian dust in this part of the Pacific ocean carries the δ53Cr of ISE, which hasremained relatively constant over the last 5.7 million years and is consistent with previous studiesof Cr in marine sediments [46, 104].2.4.3 Cr(III) oxidation in serpentinizing environmentsAs the delivery of detrital Cr to the SEPR sediments (Section 2.4.1 above) does not control theirisotopic composition, we next considered whether the light δ53Cr recorded in the hydrothermalsediments at site 598 could have been inherited directly from the hydrothermal fluids. Theδ53Cr composition of hydrothermal fluids is likely set by alteration reactions (serpentinization)occurring within the hydrothermal vent, and so we consider whether these reactions couldtheoretically impart light δ53Cr. Two previous studies that investigated the δ53Cr compositionof serpentinized ocean crust found it to be isotopically heavy (up to +1.22h) relative to theigneous silicate earth (ISE) [74, 167]. The authors interpreted these findings to be the result of acomplex series of Cr redox reactions, whereby seawater Cr(VI) entrained into the hydrothermalvent environment experiences partial reduction in the immediate vicinity of the serpentinizationsite, leading to isotopically heavy residual δ53Cr(IV) in the hydrothermal fluids and shiftingthe δ53Cr composition of any serpentinite captured Cr(III) to heavier δ53Cr values. Hydrogen46peroxide can be present in serpentinizing systems [63], with the potential to oxidize Cr(III) andthereby contribute heavy Cr(IV) to hydrothermal fluids, leaving residual serpentinites isotopicallylight [63]. Such a mechanism, however, is inconsistent with all current data, which insteadsuggest that the net result of hydrothermal alteration processes is to produce serpentinites andhydrothermal fluids with heavy δ53Cr compositions [74, 167]. We consider it unlikely, therefore,that serpentinization reactions generate hydrothermal fluids with a light δ53Cr composition andthus these processes likely did not significantly contribute to setting the δ53Cr composition of thehydrothermal sediments deposited at site 598.2.4.4 Reduction of Cr(VI) in open vs. closed systemsHydrothermal fluids provide a flux of Cr to seawater [25], though the degree to which seawaterCr may be reduced and scavenged by hydrothermal fluids remains uncertain. This means thathydrothermal systems may in fact be a net sink rather than a source of Cr to seawater. To assessthe extent to which Cr(III) in the SEPR sediments might originate from hydrothermal vs. seawatersources we consider concentrations of Cr in vent fluids, seawater and fresh plume particles. Cr(III)can be enriched in plume fluids by up to 10 times compared to seawater [100]. Adopting aconservative end member Cr of 5 times ambient seawater concentration (similar to measured ventfluid concentrations [100, 101]), would give a vent fluid Cr content of up to 50 nM Cr, prior todilution with seawater. Typical Fe concentrations in SEPR hydrothermal fluids are on the order of10 mM [168]. Using these values, the molar Cr:Fe ratio in SEPR hydrothermal fluids is at most 5x 10 – 6, which is two orders of magnitude lower than the Cr:Fe ratio of on axis plume particles[91] (Fig. 2.5) and three orders of magnitude lower than the SEPR sediments at Site 598. Thissuggests that most Cr in hydrothermal plume particles is not derived from vent fluids but rathercomes from seawater in the immediate vicinity of the vent, consistent with studies conductedon other hydrothermal systems [95, 147]. Other estimates indeed suggest that 8 - 10% of all Crsupplied to the ocean by rivers is removed via hydrothermal plume scavenging and deposited inridge crest sediments [94]. This suggests that scavenging of seawater Cr is a non-negligible part ofthe global marine Cr budget, and the resulting redox reactions involved may impart importantisotopic effects, as we explore below.47Partial reduction of seawater Cr(VI) provides one likely explanation for the light δ53Cr valueswe observed in the SEPR sediments. Partial reduction of seawater Cr(VI) upon mixing withFe(II) (and H2S) laden vent fluids, and subsequent co-precipitation of Fe (oxyhydr)oxides has thepotential to impart large Cr isotope fractionations, as observed in previous laboratory experiments[106, 107]. While the extent to which evolving vent plumes mix with seawater, and hence how openthe plume is to Cr exchange with seawater, is difficult to estimate, we consider two end-memberscenarios in which Cr isotope fractionation at the SEPR results from fully open or fully closedsystem behaviour. A fully open, well-mixed system would most likely be achieved during reactionbetween Cr(VI) and Fe(II) in the water column. This would occur when the rate of mixing betweenseawater and the hydrothermal plume is so fast that the concentration and δ53Cr composition ofCr(VI) in the latter remains the same as the former (i.e. a lack of reservoir effects). In this scenario,[Cr(VI)] and δ53Cr in the plume water is constant in time and space, and the δ53Cr of particlesproduced during seawater Cr(VI) reduction can be described by:δ53Cr(I I I)product = δ53Cr(VI)seawater–e (2.3)where the Cr isotopic composition of seawater and the isotope fractionation factor (e) do notchange with distance from the vent.To estimate the possible range of δ53Cr(III)product values, we have compiled relevant isotopefractionation factors (e) from the literature, which span a suite of appropriate reductants for Cr(VI)in hydrothermal systems (Table 2.7).48Table 2.7: Compilation of fraction factors during reduction of Cr(VI) adapted from[1].†[1]‡[169]?[107][106]κ[37]φ[108]ψ[170]θ[73]Reference Fractionation factor (‰) Reductant †	 0.8 Organic matter in ocean surface water ‡	 1.8 Dissolved and solid phase Fe(II) 	 3.91; 2.67; 2.65; 2.11 Goethite; siderite; green rust; FeS ◊	 3.60; 1.50 Ferrous Fe; ‘Green rust’ κ  3.51 Magnetite φ  4.20; 3.11 Low-pH Fe(II); organic reductants ψ  4.11; 1.75 Shewanella oneidensis bacteria θ  3.54; 5.0 Acidic H2O2; pH-neutral H2O2 (kinetic) 	Under open system behaviour, the smallest fractionation factor (e = 0.8h), proposed by [1] forCr(VI) reduction in ocean surface waters, would only yield δ53Cr(III)product values as light as theδ53Cr observed in the SEPR sediments if the initial δ53Crseawater is only mildly heavy to beginwith (δ53Cr = 0.13h). Given, however, that the modern average δ53Crseawater is 0.66h, and morespecifically, average Pacific ocean water is also 0.66h [1, 171], then Cr(VI) reduction in the SEPRsystem would need to be accompanied by an isotope fractionation > 0.8h to account for thelight δ53Cr values in the SEPR sediments, if partial Cr(VI) reduction was exclusively responsible.Larger isotope fractionation factors, however, are expected. Dissolved ferrous iron is a potentand likely reductant of Cr(VI) in the vicinity of hydrothermal vents, as are mixed valence Fe(oxyhydr)oxides that form as direct products of Fe(II) oxidation (Table 2.7). Reduction of Cr(VI) byferrous iron carries a fractionation in the range of e = 3.6 – 4.2h [106, 108]. These larger isotopefractionation factors produce δ53Cr(III) values lighter than our data over the entire range of knownδ53Crseawater compositions, including the highest values measured to date (δ53Crseawater = 1.24h,[1]. In contrast, particulate Fe(II)-bearing phases carry more muted isotope fractionation factorsin the range of e = 2.11 – 2.65h (Table 2.7). Thus reduction via such particles could account forthe observed range of Cr isotope fractionation at the SEPR if seawater Cr(VI) had an isotopic49composition between δCr53seawater = 0.62 and 2.41h, consistent with current measurements ofδCr53seawater. Therefore, based on isotope data alone, our data can be described by partial reductionof Cr(VI) in a hydrothermal system that is entirely open, using isotope fractionations that fall in acontinuum between 0.8h and 2.65h – magnitudes consistent with isotope fractionation factorsderived from laboratory experiments with a variety of ferrous Fe reductants (Table 2.7). Variabilityin the isotope fractionation factor, δ53Crcomposition of local seawater, and the extent of fluidmixing may also explain some of the dynamics observed in the δ53Crvalues down core at theSEPR.In contrast to a fully open system where partial Cr(VI) reduction occurs in a vigorously mixedwater column, a fully closed system would be approached if Cr(VI) reduction occurred withinthe sediment pile at site 598. In this scenario, increasing burial and continued Cr(VI) reductionwithin the sediment would drive the authigenic δ53Cr of Cr(III) formed deeper in the sedimenttowards the initial δ53Cr composition of seawater at the sediment water interface. This processwould also tend to cause Cr:Fe ratios of the sediment to increase with depth. Neither relationshipis observed (Fig. 2.4g-h), and thus Cr(VI) reduction within the sediment pile is unlikely to be thedominant process driving Cr isotope fractionation at site 598. In fact, we observe the oppositerelationship between sediment δ53Cr composition and burial depth, finding that the lightest δ53Crvalues measured come from samples with the lowest Cr:Fe ratios. We also observe a relativelycoherent relationship in which δ53Cr decreases with increasing burial depth and decreasing Cr:Feratios. We also recognize, however, that our data may fall somewhere on a continuum betweenfully open and fully closed system behaviour, and thus record a combination of water columnand diagenetic Cr redox processes, the latter of which we explore below.2.4.5 Diagenetic oxidation of Cr(III)Oxidation of Cr(III) to Cr(VI) by manganese oxides has been shown under laboratory conditions[61, 65, 172] and in natural soils [72], and can be described by the following reaction [65, 66]:Cr3+ + 3 MnOOH⇒ HCrO−4 + 3Mn2+ + 2OH− (2.4)50which produces soluble Cr(VI) oxyanions. In sediments, this Cr(VI) would be mobile and couldultimately escape by diffusion to overlying seawater. Previous workers determined the distributionof porewater [Cr] in the upper 20 cm of multiple continental margin sites off the coast of California,suggesting Cr loss in Mn-oxide rich surface sediments, and demonstrating low retention of Cr insediments under oxic conditions in the presence of Mn-oxides [148]. We observe manganese oxidesthroughout the sediment profile and these are present in excess of that necessary to oxidize alloxide bound Cr, based on the stoichiometry noted above. The diagenetic interaction of Mn-oxideswith hydrothermal Cr(III) to induce oxidative remobilization as Cr(VI) is therefore a plausible andlikely explanation for the apparent redistribution of Cr. Based on our current understanding of Crisotope fractionation, and by analogy to oxidative Cr mobilization in weathering environments[30, 31, 42, 149], oxidatively remobilized Cr(VI) from SEPR sediments is expected to be isotopicallyheavier than the residual Cr(III), either due to fractionation imparted directly during oxidation(e.g., [73]) or due to partial reduction back to Cr(III), which favours the light Cr isotope [30, 37].Chromium oxidation and remobilization from mineral phases hosting Cr(III) is dependenton the solubility of the host minerals [51]. There is a progressive change in the fraction oftotal Fe represented in LHFO and LGOE at site 598 [151], and this is interpreted as a progressivetransformation of poorly stable Fe-minerals (ferrihydrite) to more stable phases. This is largelycomplete by a depth of 150 cm at site 598. We observe that the Cr:Fe molar ratio decreasessteadily through the top 150 cm of sediment (Fig. 2.4j), which coincides with the zone offerrihydrite transformation to goethite (Fig. 3.3b). Ferrihydrite transformation to goethite takesplace via dissolution-reprecipitation reactions, which presumably release Cr from ferrihydrite andredistribute it to porewaters where it can be sequestered by less reactive Fe phases or becomemobile through diffusive transport. Assuming a constant Cr:Fe ratio during deposition suggestsprogressive Cr loss relative to Fe with increasing depth in the upper 150 cm of sediment (Fig. 2.4j).The downcore changes suggest that Cr in ferrihydrite is redistributed into goethite by a depth of150 cm, and that Cr loss and remobilization from ferrihydrite is complete by this depth, with lessthan 5% of the total Cr remaining in LHFO phases (Fig. 3.3, Table 2.4). Cr(III) oxidation driven byrelease of Cr from LHFO and subsequent reaction with Mn-oxides provides a likely explanationfor the observed distribution of Cr within the upper 150 cm of the sediment and its apparent loss51from the sediment. Such a scenario is also consistent with the downcore trends in δ53Cr which arestrongly correlated to Cr:Fe (Fig. 2.6b), implying a genetic link between these two geochemicalfeatures. δ53Cr values indeed become progressively lighter as Cr is lost relative to Fe, and this isconsistent with the directionality of fractionation expected during oxidative Cr remobilization.Thus, partial reduction of seawater Cr(VI) in the hydrothermal plume and diagenetic Cr(III)oxidation can together account for the light δ53Cr values and Cr(VI) loss observed in the sedimentat the SEPR.Remobilization of Cr below 150 cm appears limited, indicating that sediments older than2.5 Myr are poorly reactive towards Cr dissolution and consequently Cr(III) oxidation andredistribution. In these deeper sediments, Cr is dominantly hosted by goethite and residual poorlyreactive detrital phases, although in other marine sediments reactive Fe phases such as ferrihydriteoften convert to hematite as the stable end member [173]. At the SEPR, we expect this goethitephase to be the primary carrier of both the Cr isotope signal derived from seawater and anysubsequent early diagenetic reactions, for ultimate preservation in the geologic record. Cr isotopesignals imparted through partial reduction of seawater Cr(VI) and subsequently overprintedthrough Cr remobilization and isotope fractionation in hydrothermal sediments may thus bepartly controlled by Fe (oxyhydr)oxide mineral ageing.2.4.6 Isotopically light Cr in hydrothermal sedimentsThe light δ53Cr values of the hydrothermal sediments at site 598 are unique compared to mostoxygenated marine sediments studied to date [41, 104, 105, 162]. Based on our results from DSDPsite 598, we argue that partial reduction of seawater Cr(VI) and oxidative diagenesis combine toleave hydrothermal sediments isotopically light (depleted relative to the ISE). Importantly, bothof these mechanisms require oxygenated seawater in the vicinity of the hydrothermal system,opening the possibility that isotopically light Cr values measured in hydrothermal sediments maybe diagnostic of oxygen-bearing deep ocean water. In an anoxic deep ocean, Cr(III) sourced fromhydrothermal vents would enter seawater and precipitate in hydrothermal sediments withoutundergoing redox cycling. The resulting hydrothermal precipitates would carry an ISE δ53Crcomposition (Fig. 2.7a).52 Cr(OH)3 Fe(II)Reactions1.  Hydrolysis Anoxic SedimentsAnoxic Deep OceanAnoxic Seawater, Dissolved Fe(II), H2Sδ53Crsediments ≈ -0.121‰ (BSE)Cr2O3           Cr(III)aqCr(OH)3               Cr(III)aq Seafloor weatheringCr2O3      1no Cr(VI)reductionwithin plumeFe(II)Fe(II)Fe(II)Cr(VI)aqCr(VI)aq         Cr(OH)3 Fe(II)Reactions1.  Cr(III) oxidation2.  Cr(VI) reduction3.  Hydrolysis Oxygenated Sedimentsδ53Crseawater ≈ 1.0‰Oxygenated Deep OceanOxygenated Seawaterδ53Crsediments < -0.2‰Cr2O3        Cr(III)aqCr(OH)3               Cr(III)aq Seafloor weatheringCr2O3        Cr(VI)aqMnO21Cr(VI)aq2 3Fe(II), H2S: ε = 1.57‰Partial seawater Cr(VI)reduction within plumeFe(II)Fe(II)Fe(II)a bFigure 2.7: Cr cycling in an anoxic vs. oxic deep ocean. Depicted are schematics detailing the expected δ53Crcomposition of hydrothermal sediments deposited in both an anoxic and oxic deep ocean. Blue arrowsindicate Cr that has undergone redox cycling whereas black arrows indicate no redox state change. a) In adominantly anoxic ocean, no Cr(VI) reduction takes place in the plume vicinity. This deposits unfractionatedhydrothermally sourced Cr(III) in the sediment and preserves the initial ISE δ53Cr composition liberatedfrom vent efflux. b) In an oxygenated ocean, the heavy Cr(VI) reservoir may be partially reduced duringinteraction with electron donor rich hydrothermal plumes. Depending on the magnitude of the fractionationfactor associated with this reaction (see Table 2.7), this is expected to precipitate depleted δ53Cr in thesediment, making light δ53Cr values diagnostic of oxygenated deep ocean waters that have interacted withhydrothermal plumes. This light δ53Cr signal is only further reinforced during diagenesis with manganeseoxides in the sediment pile.In strong contrast to the anoxic deep ocean scenario, the modern oxygenated deep oceans containisotopically heavy Cr(VI), as the result of oxidative weathering on land [1, 30, 31, 44, 45, 171].Cr(VI) readily reacts with reductants (H2S, Fe(II)) that accumulate in the ocean under anoxicconditions, and thus Cr(VI) accumulation in the deep ocean only occurs when it is oxygenated.The isotopically light signals we find in modern hydrothermal sediments develop due tothe local interaction of hydrothermal reductants with Cr(VI) in oxic seawater, with a possible orlikely partial overprint through oxidative Cr(III) diagenesis (Fig. 2.7b). These observations setup a framework for using negative δ53Cr isotope anomalies in iron oxide dominated deep-seahydrothermal sediments, as a proxy for deep-water oxygenation. Light δ53Cr compositions haveonly been observed as the result of oxidative remobilization of Cr(III) in paleosols and modernsoils, and in strongly oxidizing ferromanganese nodules (e.g., [30, 149, 162, 174]), and here throughCr redox cycling in hydrothermal plume fallout sediments. Examination of existing δ53Cr records53reveals occurrences of isotopically light δ53Cr values in Neoproterozoic hydrothermal sediments[175], implying deposition from oxygenated deep ocean waters at this time. We propose thatthis framework can be expanded through further measurements of δ53Cr in older hydrothermalsediments. In principle this should enable reconstruction of deep ocean oxygen dynamics throughgeologic history.2.5 ConclusionsSediments deposited at the SEPR are enriched in hydrothermal Fe and Mn (oxyhydr)oxides. Inthe vicinity of hydrothermal vents, seawater-sourced Cr(VI) is likely reduced by Fe(II) laden ventfluids and the Cr(III) produced is scavenged by freshly-formed Fe precipitates and depositedin the sediments. This partial reduction of Cr(VI) from oxygenated seawater imparts a lightδ53Cr fingerprint in the plume particles, which are deposited with δ53Cr values lighter than theISE range (-0.12 ± 0.10h). Sedimentary Mn oxides may subsequently catalyze the oxidation ofsediment Cr(III), inducing the release of as much as 80% of Cr in the form of soluble Cr(VI) tothe overlying water column. This remobilization may also carry an isotopic fractionation, withpotential to enhance the overwhelmingly light δ53Cr imparted during partial reduction. Overall,these processes combine to set a variable, but distinctly light, range in δ53Cr of between -0.24 and-0.57h. Iron diagenesis transforms less stable Fe phases (e.g., ferrihydrite) to more stable goethiteby a depth of 150 cm (2.5 Myr), slowing subsequent diagenetic Fe alteration and likely locking inthe δ53Cr composition below this transition. Together, these processes leave a diagnostic record ofisotopically light δ53Cr that signals deposition from Cr(VI) bearing oxygenated deep seawater – asignal that can potentially be tracked through time, thus providing impetus to further exploreancient hydrothermal sediments as archives of δ53Cr that may demarcate deep ocean oxygenationin the geologic record.54Chapter 3Chromium isotope fractionation inferruginous sedimentsFerrous Fe is a potent reductant of Cr(VI), and while a number of laboratory studies havecharacterized Cr isotope fractionation associated with Cr(VI) reduction by ferrous iron, theexpression of this fractionation in real-world ferrous Fe-rich environments remains unconstrained.Here we determine the isotope fractionation associated with Cr(VI) reduction in modern ferrousFe-rich sediments obtained from the previously well studied Lake Matano, Indonesia. Whole coreincubations demonstrate that reduction of Cr(VI) within ferruginous sediments provides a sink forCr(VI) leading to Cr(VI) concentration gradients and diffusive Cr(VI) fluxes across the sedimentwater interface. As reduction proceeded, Cr(VI) remaining in the overlying lake water becameprogressively enriched in the heavy isotope (53Cr), increasing δ53Cr by 2.0 ± 0.1h at the end ofthe incubation. Rayleigh distillation modelling of the evolution of Cr isotope ratios and Cr(VI)concentrations in the overlying water yields an effective isotope fractionation of eeff = 1.1 ± 0.2h(53Cr/52Cr), whereas more detailed diagenetic modelling implies an intrinsic isotope fractionationof eint = 1.80 ± 0.04h. Parallel slurry experiments performed using anoxic ferruginous sedimentyield an intrinsic isotope fractionation of eint = 2.2 ± 0.1h. These modelled isotope fractionationsare corroborated by direct measurement of the δ53Cr composition on the upper 0.5 cm of LakeMatano sediment, revealing an isotopic offset from the lake water of eeff = 1.1 – 1.5h. The data andmodels reveal that effective isotope fractionations depend on the depth at which Cr(VI) reductiontakes place below the sediment water interface—the deeper the oxic non-reactive zone, the smallerthe effective fractionation relative to the intrinsic fractionation. Based on the geochemistry of thesediment we suggest the electron donors responsible for reduction are a combination of dissolved55Fe(II) and 0.5 M HCl extractable (solid phase) Fe(II). Our results are in line with the range ofintrinsic fractionation factors observed for such phases in previous laboratory studies. We suggestthat intrinsic isotope fractionations of around 1.8h, may be broadly characteristic of ferruginousenvironments, but we note that the partitioning of ferrous Fe between dissolved and solid phasesmay modulate this value. These results indicate that seawater δ53Cr is only captured with high-fidelity by ferruginous sediments when oxygen penetration, and therefore the upper boundary ofthe zone of Cr(VI) reduction, extends to more than 10 cm below the sediment-water-interface, ascan be the case in sediments deposited below oligotrophic waters. In more productive regions,with thinner oxic zones, ferruginous sediments would record δ53Cr as much as 1.8h lower thanseawater δ53Cr. This implies that a range of sediment δ53Cr compositions, that include that of theigneous silicate earth (ISE), are possible even when seawater is isotopically heavier than the ISE.3.1 IntroductionThe chromium isotope composition of marine chemical sediments broadly tracks the redoxevolution of the ocean-atmosphere system (e.g., [7, 30, 31, 40]. Notably, large shifts in the Crisotope composition of banded iron formation (BIFs), ironstones, shales, and carbonates resultfrom dynamics in Cr biogeochemical cycling that ultimately lead to changes in the isotopiccomposition of Cr in seawater [1, 4, 7, 31, 39, 41]. It has been proposed that the Cr isotope proxycaptures the protracted oxygenation of the Earth’s surface through the Precambrian Eons (e.g.[4, 6]. More nuanced applications of the proxy, however, require more complete knowledge ofthe processes governing Cr isotope fractionation, the Cr isotope composition of seawater, andultimately how this isotope composition is preserved in the rock record.The current model of Cr biogeochemical cycling and its attendant Cr isotope fractionation isillustrated schematically in Fig. 1.1, Chapter 1. Cr(III) hosted in crustal rocks is oxidized, largelythrough solid-state catalysis by Mn-oxides [65, 67], during weathering of continental rocks underthe modern well oxygenated atmosphere. This oxidative weathering liberates Cr(VI) to riversand groundwater [48, 50, 61, 65, 67, 72] and the subsequent riverine transport sustains the largestCr flux to the oceans today [25, 76, 77]. Cr(III) is highly particle reactive and exhibits typical56cationic sorption behaviour with greater affinity for particles at higher pH. This is in contrastto the Cr(VI) oxyanion that is generally soluble at circumneutral pH [49], and sorbs to particlesmore readily at low pH. It has also been shown that organic matter (OM) plays a large role incomplexing Cr(III) [85] and Cr(III) ions readily adsorb and precipitate in the absence of solubleOM. In most natural waters above pH 5.5, Cr(III) is thus present in concentrations higher thanexpected based on equilibrium solubility due to stabilization via OM [85]. Likewise, organic acidspromote dissolution of Cr(III) hydroxides under alkaline conditions, potentially mobilizing Cr(III)[176]. Speciation in the presence of such organic acids is dominated by these ligand complexesrather than hydrolyzed Cr(III) species at pH > 5 [177]. Cr(III) is otherwise only appreciablysoluble and mobile under acidic pH [97].The majority of chromium in seawater is in the form of CrO2 –4 ions which are the directproduct of continental weathering and are most likely removed from seawater via reduction toCr(III) and subsequent association to particles and sedimentation [105]. Reduction can occur via aplethora of electron donors (organic compounds, reduced sulphur species, reduced Fe species,anaerobic and aerobic microbes), of which ferrous Fe species are particularly effective, includingFe(II)aq and Fe(II)–bearing minerals [106]. Cr(VI) reduction via Fe(II)aq is very rapid (half-liferanging from minutes to hours) and obeys pseudo-first order reaction kinetics at pH < 10 [56, 178].Reduction of Cr(VI) with solid phase Fe(II) is generally slower, occurring on half-life timescales ofhours to days [37].Reduction of Cr(VI) is generally accompanied by large and characteristic Cr isotope frac-tionations (∼+1.0 to +4.5h), where isotope fractionation (e) herein refers to the difference inisotopic composition of the reactant and instantaneously formed product [37, 43, 106–108, 179].Laboratory experiments demonstrate that Cr(VI) reduction preferentially incorporates the lightisotopes into the product Cr(III) [37, 46]. Isotopic fractionations during Cr(VI) reduction fordifferent Fe species range from ∼1.5h for Fe(II)aq + Fe(II)/Fe(III)greenrust (a multi-valence Fe(oxyhydr)oxide, [106, 107] to ∼3.5h for Fe(II)/Fe(III)magnetite [37], demonstrating that the reactionmechanism plays a major role in determining the magnitude of fractionation expected. Kineticisotope fractionation occurs during the reduction of Cr(VI) likely due to the rearrangement ofCr-O bonds and the resulting change in Cr bonding from tetrahedral to octahedral coordination57[71]. Determining Cr isotope fractionation during reduction of Cr(VI) is crucial for interpretingthe Cr-isotope signal in the paleorecord. While the isotope fractionations accompanying Cr(VI)reduction are well established in closed-system laboratory settings, relatively few studies haveextended such experiments into direct tests of fractionation in natural environments [180, 181],and this lack of data from natural environments limits our ability to make robust interpretationsof the Cr isotope record.Ferrous Fe is abundant in a wide range of modern marine sediments [148], and appears tohave also been particularly important during Earth’s early history. Ferruginous ocean conditions(anoxic and Fe-rich) persisted through much of Earth’s history [182] and thus Cr(VI) reductionvia Fe, likely played a major role in setting the authigenic δ53Cr recorded in BIFs, ironstones,and ferruginous shales. After ∼800 Ma, Cr(VI) was likely supplied to the Fe-rich oceans viaoxidative weathering in O2 rich surface environments. It has been shown that Cr(III) may alsobe oxidized to Cr(VI) by H2O2 in the local absence of O2 in serpentinizing systems [63], howeverO2 is ultimately required to generate this peroxide and other oxidants of Cr at the outset. Thebatch experiments of [106] indicate that if Cr(VI) reduction in solution is near quantitative, theδ53Cr of the initial solution is inherited by the Cr in the Fe(III)-(oxyhydr)oxide products. This islikely to occur as long as Fe(II) is in excess, as might be expected for Archean eon Fe(II)aq richwater-columns. Likewise, it has been suggested that in sediments deposited under an anoxicwater column, adsorbed authigenic Cr can be buried with little to no fractionation relative to theseawater from which it was deposited. A possible modern example of this is the anoxic CariacoBasin, whose sediments have had a constant δ53Cr composition for the last 14.5 kyr, similar to thatof contemporary δ53Cr of Atlantic deep waters [104, 105]. Thus δ53Cr values measured in anoxicsediments—sediments deposited from overlying water that is oxygen free—may faithfully recordthe contemporaneous δ53Cr seawater composition [104, 105]. In contrast, it has been shown thatfractionation between seawater Cr and authigenic Cr can be appreciable in sediments overlain byoxygenated seawater [104]. The δ53Cr values recorded in such sediments are variable [104], andthis suggests Cr isotopes are fractionated during early diagenesis, a process which has receivedlittle attention but likely plays a critical role setting the δ53Cr values recorded in marine sediments.The contrasting Cr isotope data obtained from these two modern marine environments highlights58the possible importance of transport mechanisms in setting authigenic δ53Cr values. In anoxicwater columns, transport across the chemocline is much slower than the timescales of Cr(VI)reduction reactions and thus Cr(VI) reduction is nearly quantitative with very small fractionationsfrom local seawater [105]. In contrast, transport into the sediment pile is dominated by diffusion,and thus a key factor controlling Cr isotope fractionation is the length scale of the diffusional pathto the zone of Cr(VI) reduction [105, 183–185].We sampled sediments from ferruginous Lake Matano, Indonesia, to directly test for Cr isotopefractionation during reduction in modern Fe-rich sediments. Iron speciation in sediments can beassessed through selective sequential extractions [118], and through Fe speciation analyses it ispossible to discern between depositional conditions that are oxic, anoxic and ferruginous (dissolvedFe(II) present), and anoxic and sulphidic (dissolved H2S present). Ferruginous sediments areoften enriched in reactive Fe relative to total Fe, and have ratios FeReactive/FeTotal > 0.4. Eventhough the sampling site at Lake Matano is overlain by mildly oxygenated water, the underlyingsediments are greatly enriched in reactive Fe and their FeReactive/FeTotal ratios are >> 0.4, whichoperationally define them as ferruginous. We performed whole core incubations and sedimentslurry experiments and we modelled our results using three different approaches, Rayleighfractionation, a parameterized diffusion model, and a diagenetic reactive-transport model. Weview these sediments from Lake Matano as analogous to the Fe-rich sulphide poor clasticsediments (e.g. shales) common in the Proterozoic and early Paleozoic during the emergenceand early diversification of animals (e.g., [23, 182]. We provide new insights into the magnitudeof fractionation for Cr(VI) reduction in ferruginous settings, to refine interpretation of the δ53Crisotope record during that period of Earth’s history. In addition, this work provides foundationthat can be used to interpret Cr isotope distributions in diverse environmental settings and therock record.593.2 Methods3.2.1 Sample location and experimental frameworkLake Matano is located in Central Sulawesi Island, Indonesia. The lake has a very small surface areato depth ratio (Lake Matano is the 8th deepest lake in the world at > 590 m). This combined withthe absence of strong seasonal temperature fluctuations maintain a persistent pycnocline at 100 mdepth, separating an oxic surface mixed layer from anoxic bottom waters. Sulfate concentrationsare low (< 20 µM) in the surface mixed layer [143] and the ultramafic catchment basin suppliesabundant Fe oxyhydroxides to the lake supporting the deposition of Fe-rich sediments [143].Chromium concentrations are naturally high in the lake (180 nM) due to oxidative weathering ofCr-rich ultramafic bedrocks [137].3.2.2 Whole core incubation experimentIn spring 2015 we obtained 6 gravity cores from a site located at the northwestern end of the lake,at 60 m water depth. The water is oxygenated to the sediment-water interface at this locationand Cr(VI) was thus the dominant species present. The oxygen penetration depth within thesediment at this location is roughly 2 mm (Appendix A). We collected cores (core barrel size; 8 cmin diameter, 57 cm tall) with a well-preserved sediment water interface. Four of the most pristinecores were selected for whole-core incubation experiments. We sub-cored the larger gravity coresinto smaller cores to perform the incubations (core barrel size; 5.5 cm in diameter, 42 cm tall),while maintaining an undisturbed sediment water interface, capturing an average volume of 520ml of bottom water overlying an average of 20 cm of sediment. Our incubations were performedon site, as to minimize core disturbance during transport. The 4 cores (herein referred to as A, B,C, D) were set up in a core stand and magnetic stir bars were installed 10 cm above the sedimentwater interface in each tube to mix the overlying water without disturbing the sediment waterinterface, in order to homogenize the overlying water. Stirring in this fashion has been shown toreproduce turbulent flow velocities at the sediment water interface similar to in-situ conditions[186]. The cores were pre-incubated for 3 hours to allow return to steady-state conditions postcore recovery. The overlying lake waters preserved during coring, were then spiked with the60aim of augmenting natural concentrations by A = 0.5, B = 5, C = 2.5, and D = 1 µmol l– 1 Cr(VI)respectively, in addition to the background Cr(VI) concentration of the overlying lake water (180nM, [137]. Chromium was added using a 10 mM stock solution of potassium dichromate preparedin 18 mQ deionized water. Over the course of 70 hours, 20 ml aliquots were taken from theoverlying waters of the whole cores at regular intervals and Cr(VI) concentrations were measuredspectrophotometrically using the method of [187]. Spectrophotometric measurements were usedon site to monitor the progress of Cr(VI) reduction and drawdown in the overlying water and todetermine the amount of double-spike to add to the samples in order to obtain robust isotopemeasurements. We also isolated 50 ml aliquots of the overlying water immediately after spikingeach core with Cr(VI). We measured Cr(VI) in these 50 ml separates at each time point to trackreduction taking place directly in the water column versus that occurring within the sediment ofthe whole core incubations. Complementing our [Cr(VI)] measurements, aliquots of 20 ml werealso sampled at the same time points and acidified with 0.5 M HCl for future Cr isotope analysis.3.2.3 Laboratory sediment slurry experimentTo complement our whole core incubations and study Cr(VI) reduction without diffusive limitation,we also performed an anoxic slurry experiment using lake Matano sediment that was sub-sampledin the upper 5 cm from the two cores not used in our whole core incubations, and stored inanoxic bags under N2 until the slurry incubations were performed. Bicarbonate buffered mQwater (pH 7) was bubbled with a N2/CO2 gas mixture (80/20, 99.99%) to purge oxygen. Slurrieswere prepared in a glove bag under an N2 atmosphere by adding 8 g of sediment to 200 ml ofanoxic water in 250 ml serum bottles, which were then sealed with a rubber stopper. Using an N2flushed syringe, Cr(VI) was spiked into each bottle, to an initial slurry concentration of 5 µmoll– 1, from a stock of 1 mmol l– 1 K2Cr2O7. Samples (1 ml) were then taken periodically using N2flushed syringes and [Cr(VI)] was measured spectrophotometrically [187]. [Fe(II)] and [FeTot] ofthe 0.5 M HCl extractable sediment pool were also measured spectrophotometrically [188] as wasthe pH at each time point. Aliquots of 10 ml of slurry were filtered and preserved in 0.5 M HClfor Cr isotope analysis at each time point.613.2.4 Sediment leaching experimentWe also performed a 0.5 M HCl leach in order to measure Fe speciation in the lake sediment [129].We sectioned two separate gravity cores at a resolution of 0.5 cm; these cores were not the onesused in the Cr doping experiments. We digested roughly 500 mg of sediment from each depthinterval in 1.5 ml of 0.5 M HCl for 1 hour in order to release reactive solid phase Fe(II) and Fe(III).We conducted Fe concentration measurements spectrophotometrically using the ferrozine methodon site. Aliquots were preserved for δ53Cr measurement on these 0.5 M HCl sediment extracts.Porewater Fe-speciation measurements were also obtained from a gravity core sampled from thelake in 2010. This core was also recovered at a water depth of 60 m, hence the overlying water wasoxygenated, similar to our site sampled in 2015. The core sampled for porewater was sectioned atthe same resolution and Fe concentrations were measured spectrophotometrically on site. Theporewater extraction performed in 2010 was conducted in a N2 flushed glove bag using rhizons.3.2.5 Chromium isotope methodologyWe followed a procedure very similar to [104] and modified after [105] and [46], in order tomeasure the Cr isotope composition of our samples. Chromium isotopes were measured on aMultiple Collector-Inductively Coupled Plasma-Mass Spectrometer (MC-ICP-MS, ThermoFisherScientific Neptune Plus) using a double-spike correction for isotope fractionation during columnchemistry and instrumental mass bias at the Yale Metal Geochemistry Center. We used threesteps [105] to ensure complete removal of Fe, Ti, and V during chemical purification. Initially,a split of the 0.5 M HCl sample solution was spiked after digestion, evaporated to dryness,and the residues brought up in 1 M HCl. The amounts of 50Cr – 54Cr double-spike (50Cr/52Cr =462.917, 53Cr/52Cr = 0.580, 54Cr/52Cr = 354.450), were adjusted so that the spike/sample ratio(i.e., (54Cr)spk/(52Cr)smp) was 0.5. Prior to the first column step (AG1-X8 anionic resin, 100-200mesh), Cr(III) was oxidized to Cr(VI) using potassium peroxidisulfate and heating the samplesfor 2 hours at 110◦C. The AG1-X8 resin was cleaned with mQ water, 3 M HNO3 and 6 M HCl.The matrix was eluted with 24 ml of 0.2 M HCl and 4 ml 2 M HCl, with Cr subsequently reducedwith a mixture of 2 M HNO3 and 0.5% H2O2 for 30 min and eluted with 5 ml of this mixture. The62second step removes traces of Fe that may remain after the first elution. We used microcolumnsfilled with 0.2 ml AG1-X8 resin, 100-200 mesh. The columns were cleaned using mQ water and 3M HNO3, and samples were loaded and collected with 0.8 ml of 6 M HCl. Traces of Ti can be leftafter the first two columns steps. Therefore, as the last step we used a cationic resin AG50W-X8(200-400 mesh) to ensure complete Ti and Cr separation. The resin was cleaned with mQ water, 3M HNO3 and 6 M HCl, followed by sample loading in 3 ml of 0.5 M HNO3. The Cr was collectedwith 1 ml of 0.5 M HNO3 and the matrix was eluted with 2 ml of 0.5 M HF and 6 ml of 1 M HCl.Remaining Cr on the columns was collected with 5 ml of 1.8 M HCl.Chromium isotopes were measured using a modified analytical protocol of [46], and Crconcentrations were determined by isotope dilution. Samples were run in high-resolution mode toresolve polyatomic interferences such as 40Ar12C+, 40Ar14N+ and 40Ar16O+. Although our chemicalprocedure should ensure complete removal of Fe, Ti and V, we monitored these elements bymeasuring 56Fe, 49Ti and 51V, and corrected our samples for potential interferences of 54Fe on54Cr, and 50Ti and 50V on 50Cr. The double-spike data reduction model is based on the iterativemethod described by [157]. We measured a spiked Cr isotope standard NIST SRM 979 before andafter each natural sample to ensure machine stability.External precision is reported as two sigma (2σ) uncertainty, calculated based on sampleconcentration matched duplicate analysis of geological reference materials (GRMs) processedthrough ion-exchange chromatography columns along with samples (BHVO-2 and Nod-A-1were systematically processed with 18 samples). Both geostandards returned values similar tomeasurements published in the literature (BHVO-2 δ53Cr = -0.11 ± 0.07h (n = 3), Nod-A-1 yieldedδ53Cr = 0.08 ± 0.05h (n = 4)). Measurement precision is calculated via replicate measurements ofthe isotopic standard NIST SRM 979 during each analytical session (two standards bracket everysample), and 2 values are better than 0.04h. In addition, a two-standard deviation of the mean(2se) was systematically calculated using the 50 cycles of measurement obtained for each sampleduring MC-ICP-MS analysis, and was generally 0.04 - 0.06h on the delta values. The measured53Cr/52Cr ratios are reported in δ notation, where δ53Cr values for reasons of simplification aremultiplied by 1000 and reported as per mil (h) deviations relative to that of a known Cr isotope63standard (NIST 979).δ53Cr = [(53Cr52Cr sample)(53Cr52Cr NIST979)− 1] (3.1)3.2.6 Modelling Cr isotope fractionationTo examine Cr(VI) reduction and evolution of δ53Cr(VI) during the whole core incubations andslurries, we apply the Rayleigh Fractionation model to the Cr pool remaining in the overlying orinterstitial water during the course of the experiments. As reduction of Cr(VI) progresses in boththe whole core incubations and slurries, it is expected that the remaining Cr(VI) reactant becomesincreasingly enriched in the heavy isotope (53Cr) relative to the Cr(III) product. The proportions ofthe heavy and light isotopes of Cr in the reactant and the instantaneous product can be describedby a fractionation factor α;α =53Cr52Cr reactant53Cr52Cr product(3.2)We can also express the magnitude of isotopic fractionation more conveniently by defining e;e = (α− 1) · 1000h (3.3)Assuming a closed well-mixed system and constant fractionation factor (α) equation (4) relates theratios of the heavy and light isotope in the reactant to the progress of the chemical reaction;ln(RtR0) = (α− 1) · ln(ft) (3.4)Where Rt and R0 are the ratios of 53Cr/52Cr at time t, and initially (before any reaction) respectively,ft is the fraction of reactant remaining at time t, and α is the fractionation factor. In naturalsedimentary systems it has been abundantly demonstrated that Rayleigh models can underestimatetrue, intrinsic fractionation factors as a result of open system behaviour, where reaction productscan be physically separated from the sampled reactant pool by permeable materials that restrictreactant transport to the location where the reaction takes place [183–185, 189–191]. Overall these64reactive-transport phenomena lead to reservoir effects that result in a measured or expressedfractionation factor that is smaller than the intrinsic fractionation factor.While the slurries experiments are expected to yield estimates of intrinsic fractionation factor,whole core incubations will likely yield smaller effective fractionation factors reflecting theexperimental conditions. To obtain another estimate of the intrinsic fractionation factor we thusutilize a second idealized 1D diffusion model [184, 185, 192], whereby removal of Cr(VI) froman overlying water is driven by diffusion into underlying sediments, where reduction occurs inthe pores, assuming a spatially uniform first order Cr(VI) reaction rate throughout the reactivezone. This model is useful as it allows for an estimated correction for the reservoir effects thatdevelop during the removal of Cr within the sediment porewater. Similarly to [185], we assumethat reductants are absent in the presence of oxygen, and thus the oxic surface sediment layercreates a “non-reactive” zone (NRZ) that displaces Cr(VI) reduction deeper into the sediment.This zone increases the distance across which Cr(VI) must diffuse between overlying water andthe reaction sites in the sediment [184, 185]. Figure 3.1 is a schematic and idealized representationof how the depth of the NRZ controls the effective fractionation.65Cr(VI)SW Cr(VI)PW Cr(III)SEDk1kLk1NRZAs L     0 [CrVI]PW    [CrVI]SW f     1 δ53CrSED     δ53CrSW – (εint)     As L     ∞ [CrVI]PW     0 f     0 δ53CrSED     δ53CrSW     δ53CrSED = δ53CrSW – f (εint)             Figure 3.1: A schematic showing how the depth of the non-reactive zone (NRZ) controls effective isotope fractionation.In this idealized schematic, seawater chromium (Cr(VI)SW) diffuses through a NRZ of a given thickness(length scale, L) to reach porewater in the anoxic zone (Cr(VI)PW). k1 is the diffusion rate constant of Cr(VI)through the NRZ and k is the reduction rate constant of Cr(VI) in anoxic pore waters, producing sedimentCr(III)SED. The fraction of Cr(VI) diffusing out of the anoxic zone is represented by f = k1 [Cr(VI)]PW / k1[Cr(VI)]SW, where [Cr(VI)]PW and [Cr(VI)]SW are the Cr(VI) concentration in anoxic pore water and oxicseawater or lake water. The effective isotope fractionation of Cr in such a system is controlled by the ratiobetween the rate of Cr(VI) reduction in the anoxic zone and the rate of diffusion of Cr(VI) through the NRZ.If the rate of reduction increases for a given rate of diffusional supply, f ⇒ 0 and δ53CrSED ⇒ δ53CrSW.Since the rate of supply decreases with increasing thickness of the NRZ (L), for a given rate of reduction,δ53CrSED ⇒ δ53CrSW as L ⇒ ∞. On the other hand, as L ⇒ 0, the rate of Cr(VI) supply to anoxic porewaters increases and for a given rate of reduction, [Cr(VI)]PW increases, f⇒ 1 and δ53CrSED ⇒ δ53CrSW –eint.However, it is important to note that end members in this idealized system are not physicallypossible. For example, one end member in this idealized system assumes that all reaction of Cr(VI)occurs in direct contact with the overlying water, and hence does not consider concentration orisotopic gradients that in reality would develop due to diffusion into some small layer of sedimentin which the reaction takes place. At steady state, the net transport of Cr(VI) is equal to the sum ofCr(VI) diffusion from bottom water into pore waters (which depends on Cr(VI) concentration in66pore waters at the top of the anoxic zone) and Cr(VI) diffusing from pore waters to bottom water:k1 · [Cr(VI)]SW = (k1+ k) · [Cr(VI)]PW (3.5)Where [Cr(VI)]SW and [Cr(VI)]PW are the Cr(VI) concentration in bottom water and in pore waterrespectively, at the top of the sediment anoxic zone, k1 is diffusion rate constant of Cr(VI) throughthe NRZ, and k is the reduction rate constant of Cr(VI) in pore waters. If Cr isotope fractionationduring diffusion through the NRZ is negligible:δ53CrSED = δ53CrSW– f · (eint) (3.6)Where δ53CrSED is the isotopic composition of the sediment pool, eint is the intrinsic isotopefractionation during Cr reduction, and f is the proportion of Cr diffusing from pore water tobottom water:f = (k1 · [Cr(VI)]PW)/(k1 · [Cr(VI)]SW) = [Cr(VI)]PW/[Cr(VI)]SW (3.7)As the NRZ depth approaches zero, the [Cr(VI)]PW approaches [Cr(VI)]SW and the proportion ofCr diffusing from pore water to bottom water approaches unity, then we obtain the scenario whereall Cr reduction occurs at the sediment water interface in contact with bottom water, intrinsicisotope fractionation approaches full expression and sediment is isotopically lighter than bottomwater, which represents an idealized case:δ53CrSED = δ53CrSW–(eint) (3.8)As the NRZ depth approaches infinity, the [Cr(VI)]PW approaches zero (because the rate of Cr(VI)supply to anoxic pore water becomes vanishingly smaller than its rate of reduction) and theproportion of Cr diffusing from porewater to bottom water approaches zero. In this case, f = 0and:δ53CrSED = δ53CrSW (3.9)67Between these two extremes, the sediment Cr isotopic composition is lower than seawater buthigher than expected if intrinsic isotope fractionation is fully expressed.To estimate the intrinsic isotope fractionation (αint) from experiments where the effective alpha(αeff) was determined from Rayleigh models of measurements in the overlying water column, weemployed the model of [185]. We assumed that 52Cr and 53Cr have identical diffusivities, then αintcan be calculated as follows [184, 185, 192];αint =(α2e f f · λ2)((L + λ− (αe f f · L)2) (3.10)Where αeff is the effective isotopic fractionation factor, calculated from the Rayleigh distillationmodel, and L is the length of the nonreactive zone (cm), which we take as the depth of O2penetration into these sediments. The measured O2 penetration depth at the site in Lake Matanois 0.2 cm (Fig. A.1). λ is the length scale (cm) of penetration of Cr(VI) into the reaction zone (thee-folding distance for concentration decrease) and is defined as follows;λ =√(Dk) (3.11)where D is the diffusion coefficient of Cr(VI) (cm2 s– 1) and k is the first-order reaction rate constant(s– 1).We also constructed our own reactive transport model to explicitly describe the intrinsic isotopefractionation factor and the reservoir effects that develop. The model only considers diffusion –no advection, and parameterizes Cr(VI) reaction rates using a Gaussian distribution. This modeldiffers from the [184] approach, as it does not assume a spatially uniform reaction rate within thereactive zone. Rather, this approach is based on the assumption that Cr(VI) reduction rates arelimited by the availability of electron donors within the NRZ and that the rate maximum mustoccur below the depth of oxygen penetration, where potential reductants of Cr(VI) accumulate.Below the depth at which the rate maximum occurs, Cr(VI) reduction rates decline again, limitedby the availability of Cr(VI) itself as its supply becomes exhausted in response to reduction. Acomplete description of this model can be found in the Supplementary Material (Appendix A).Furthermore, we complement our diagenetic Cr modelling with an oxygen penetration model68adapted from [193], which allows us to link δ53Cr composition to O2 penetration depths (AppendixA). Overall, our δ53Cr results from both the whole core incubations and slurry experiments weremodelled using these different approaches, ultimately yielding a comprehensive description of Crisotope fractionation during sediment diagenesis.3.3 Results3.3.1 Whole core incubationsOver the time course of our whole-core incubations [Cr(VI)] decreased an average of 75% in allcores (Fig. 3.2, Table 3.1).690 10 20 30 40 50 60 70Time (Hours)-1.2-1.1-1.0-0.9-0.8-0.7-0.6-0.5-0.4-0.3-0.2log[Cr] (µmol l-1) 0 10 20 30 40 50 60 70Time (Hours)-[Cr] (µmol l-1) 0 10 20 30 40 50 60 70Time (Hours)-0.6-0.5-0.4-0.3-0.2-[Cr] (µmol l-1)0 10 20 30 40 50 60 70Time (Hours)-0.5-0.4-0.3-0.2-[Cr] (µmol l-1)A BC DFigure 3.2: Cr(VI) Concentration (µmol l– 1) vs. Time (hours) during whole-core incubation experiments. Displayedis the log[Cr(VI)] as a function of time for each incubation (tubes A, B, C and D, labelled in the upper rightcorner of each plot). Initial [Cr(VI)] (µmol l– 1) were as follows A = 0.5, B = 5.0 C = 1.0 and D = 2.4. Blackdata points were measured spectrophotometrically on-site, grey data points were measured via isotopedilution. Linear regression is fit to the isotope dilution concentration measurements and the dotted linesrepresent a 95% confidence belt.The difference between the spectrophotometric [Cr(VI)] measurements performed on-site andthe ID-ICP-MS measurements, is greater than the precision of the spectrophotometric methoddetermined by repeated measurements of standards (Table 3.1), which likely reflect additional errorin natural samples. However, we have only used the much more precise ID-ICP-MS measurementsin all of our modelling. We have only used spectrophotometric measurements used site to monitorthe progress of Cr(VI) reduction and drawdown in the overlying water and to determine, theamount of double-spike to add to the samples in order to obtain robust isotope measurements70Table 3.1: Whole core incubation. Spectrophotometric [Cr(VI)] data has subscript “s”, NA stands for “notanalyzed”. Spectrophotometric precision is 0.1 µmol l– 1, isotope dilution precision is 0.02 µmol l– 1[Cr(VI)] (µmol l-1) δ53/52Cr (‰) Time (hours) A As B Bs C Cs D Ds A 2se B 2se C 2se D 2se 0 0.52 0.4 3.49 2.4 0.97 1.0 2.44 2.2 0.16 0.03 0.12 0.03 0.09 0.03 0.02 0.04 5.5 NA 0.4 NA 4.3 NA 1.0 NA 2.1 NA NA NA NA NA NA NA NA 12.5 0.37 0.3 3.09 3.5 0.69 1.0 1.56 1.8 0.48 0.03 0.67 0.03 0.35 0.04 0.44 0.04 20.5 NA 0.2 NA 2.3 NA 0.6 NA 1.1 NA NA NA NA NA NA NA NA 29.5 0.20 0.2 2.59 1.8 0.48 0.5 1.00 0.7 1.04 0.08 1.37 0.08 0.77 0.03 0.91 0.03 35.5 NA 0.1 NA 1.7 NA 0.5 NA 0.7 NA NA NA NA NA NA NA NA 44.5 NA 0.1 1.68 1.7 0.33 0.4 0.65 0.6 NA 0.04 1.44 0.04 1.18 0.03 1.32 0.03 54.5 0.16 0.1 1.44 1.2 0.33 0.3 0.52 0.4 1.54 0.04 1.66 0.04 1.38 0.04 1.68 0.03 69.5 0.07 0.1 1.13 1.0 0.26 0.3 0.35 0.4 1.84 0.04 1.78 0.04 1.84 0.05 2.03 0.04 	(Table 3.1). Our high [Cr(VI)] treatment (Tube B, 5.0 µmol l– 1), was sampled too quickly afterthe initial spike, not allowing for equilibration, as observed by the low [Cr(VI)] at time point 1compared to time point 2. To assess the relationship between [Cr(VI)] and the rate of reduction, weexamine the decrease in Cr(VI) concentrations observed in the overlying water through time. Thechange in [Cr(VI)] can be described by a pseudo first-order rate constant, k, calculated from theslope of a linear regression of ln(Ct/C0) vs. time. The pseudo first-order rate constants describingCr(VI) reduction in the four different cores fall in a narrow range from 5.65 – 7.46 x 10 – 6 s– 1 witha mean value of 6.3 ± 0.7 x 10 – 6 s– 1, indicating reduction rates are independent of the [Cr(VI)] inthe overlying water, over the concentration range from 0 - 5 µmol l– 1 (Table 3.2).Table 3.2: Rayleigh model parametersTube Label C0 (umol l-1) ε  Value (‰) 2 se (‰) Alpha First Order Rate Constant (cm s-1) A 0.5 0.9 0.1 1.0009 5.64 x 10-6 B 3.5 1.3 0.3 1.0013 6.13 x 10-6 C 1.0 1.3 0.1 1.0013 5.77 x 10-6 D 2.4 1.0 0.03 1.0010 7.45 x 10-6 	We also tested for Cr(VI) reduction in the overlying water. To do this we isolated aliquots of lakewater at the beginning of the incubation and measured [Cr(VI)] over the time course. No Cr(VI)reduction was observed over 40 hours in the aliquots of lake water that were separated from eachcore, demonstrating that the reduction observed in the whole-cores occurs within the sediment71with little to no internal Cr reduction in the overlying water.Each of the whole-core incubations shows that Cr reduction is concomitant with isotopefractionation. All incubations display increasing δ53Cr with decreasing [Cr(VI)] that is consistentbetween the cores. The overlying waters therefore become progressively enriched in the heavyisotope (53Cr) as reduction proceeds (Fig. 3.3, Table 3.1).0 10 20 30 40 50 60 70Time (Hours)δ53 Cr (per mil)Tube ATube BTube CTube D-2.0-1.5-1.0-0.50.0ln(C/C0)*ln(R/R0)Group: 1Group: 2Group: 3Group: 4ABFigure 3.3: δ53Cr values (h) for whole-core incubation experiments. A) Variation of δ53Cr values over time.B) Rayleigh fractionation model for all data points, where R0 and C0 are the initial 53/52Cr ratio andconcentration and R and C are the ratio and concentration at time t. The slope of the line is (α-1) · 1000.Effective isotope fractionation factors were estimated using the Rayleigh distillation model, whichassumes closed system behaviour and relates the δ53Cr evolution to the change in [Cr(VI)] for eachcore (Table 3.2). Rayleigh model fractionation factors were obtained through a linear regression of72ln(R/R0) vs. ln(C/C0), where R and C are the 53Cr/52Cr ratio and the [Cr(VI)] at time t respectively,and R0 and C0 are the initial 53Cr/52Cr ratio and [Cr(VI)]. This regression yields a slope of m= (α-1)·1000h with an error equal to the least squares regression. The mean Rayleigh isotopefractionation from the regressions obtained on the 4 separate cores with Cr(VI) concentrations of0.5, 5.0, 2.5 and 1.0 was eeff = 1.1 ± 0.2h.3.3.2 Sediment leach experimentsThe 0.5 M HCl extractable sediment pool of the upper 0.5 cm at the site is also rich in Cr andFe. The 0.5 M HCl extractable Cr concentrations in the upper 0.5 cm obtained from duplicatecores range from 0.46 - 0.64 µmol cm– 3 and concentrations of solid phase reactive Fe(II) andFe(III) are 15.28 and 0.55 µmol cm– 3 respectively. Porewater Fe(II) concentrations measured in2010 were 1.26 µmol cm– 3. The authigenic Cr isotope composition of these sediments is heavy,having a range of δ53Cr compositions from 1.49h to 1.99h ± 0.03h (Table 3.3), which are clearlyfractionated compared to the first reported δ53Cr measurements of the lake water, which suggesta δ53Cr compositional range for the overlying water of 2.20 to 3.30 ± 0.15h, and thus yield apossible isotope separation of ∆53Crlakewater – sediment = +0.21h to +1.81h ± 0.15h.Table 3.3: Cr isotope composition of the 0.5 M HCl extractable pool in upper 0.5 cm of sediment of twoduplicate cores[Cr] (ug g-1) δ53Cr (‰) Isotope Separation Factor ∆ (‰) (δ53Crlakewater - δ53Crsediments) 95.4 , 132.2 1.49 , 1.99 ± 0.03 0.21 , 1.81 ± 0.15 	3.3.3 Slurry experimentsAs a direct experimental test of the Cr(VI) reduction rate and effective fractionation we observedin our open system whole core incubations, we also measured Cr(VI) reduction and isotopefractionation in a closed system slurry. The results of our slurry experiment can be seen in Table3.4, and over the course of 100 hours [Cr(VI)] decreased by an average of 98% in all bottles (Fig.3.4).73Table 3.4: Slurry experiment results [Cr(VI)] (µmol l-1) δ53/52Cr (‰) Fe(II) (µmol l-1) Fe(III) (µmol l-1) Time (hours) B1 First Order Rate Constant (cm s-1) B2 First Order Rate Constant (cm s-1) B1 B2 B1 B2 B1 B2 0.0 4.4 8.67 x 10-6 4.4 1.04 x 10-5 0.1 ± 0.03 0.1 ± 0.03 570 480 6060 6150 0.9 3.3  3.5  0.2 ± 0.03 0.3 ± 0.03 330 350 3700 3350 1.8 2.8  2.9  0.4 ± 0.03 0.4 ± 0.04 280 280 3540 3240 6.8 2.7  2.8  0.8 ± 0.04 0.8 ± 0.03 70 100 3070 3200 23.6 1.9  1.8  1.8 ± 0.03 2.0 ± 0.03 180 230 3580 3790 28.6 1.7  1.6  2.1 ± 0.04 2.4 ± 0.03 190 160 3680 3680 48.6 0.9  0.8  3.7 ± 0.03 4.1 ± 0.03 270 190 3780 3810 95.6 0.2  0.1  7.0 ± 0.07 8.9 ± 0.07 240 240 3760 3690 	0 10 20 30 40 50 60 70 80 90 100Time (Hours)[Cr] (µmol l-1)III0 10 20 30 40 50 60 70 80 90 100Time (Hours)-1.5-1.0- [Cr] (µmol l-1)ABFigure 3.4: Cr concentration (µmol l– 1) as a function of time (hours) during slurry experiments. A) [Cr] for BottlesI and II (duplicate) measured during the slurry experiment. B) log[Cr] vs. Time for the slurry experimentfit with a linear regression. Dashed lines represent a confidence belt of 95%.74Initial reduction of Cr(VI) was rapid, possibly due to heterogeneity in the slurries before thereactants are well mixed, a phenomenon also observed by [108]. The pH was stable at an averagevalue of 6.4 during the entire experiment. We also measured [Fe(II)] and [Fe(III)] of the reactivesolid phase Fe fraction of the sediment from the slurries at each time point and the results areplotted in (Fig. 3.5).0 10 20 30 40 50 60 70 80 90 100Time (Hours)0100200300400500600[Fe(II)] (µmol l-1)B1 Fe(II)B2 Fe(II)0 10 20 30 40 50 60 70 80 90 100Time (Hours)3,0003,5004,0004,5005,0005,5006,0006,500[Fe(III)] (υmol l-1)B1 Fe(III)B2 Fe(III)ABFigure 3.5: 0.5 M HCl extractable Fe(II) concentrations (µmol l– 1) through time (hours) during the slurry experiment.0.5 M HCl extractable Fe(II) concentrations (µmol l– 1) through time (hours) during the slurry experiment.We attribute the initial drop in [Fe(II)] to Fe(II) sorption or possibly to traces of residual oxygen inthe serum bottles which may have induced oxidation of Fe(II) at the beginning of the time course.Similar to our whole core incubation experiments, the slurries display preferential removal ofthe lighter over the heavier Cr isotopes with time. The Rayleigh model can also be applied to ourslurry data (Fig. 3.6).750 10 20 30 40 50 60 70 80 90 100Time (Hours)δ53 Cr (per mil)B1B2-4.0-3.5-3.0-2.5-2.0-1.5-1.0-0.50.0ln(C/C0)0.0000.0010.0020.0030.0040.0050.0060.0070.0080.009ln(R/R 0)ABFigure 3.6: Variation of δ53Cr values (h) of the slurry experiments through time (hours). A) Variation of δ53Crvalues as a function of time for bottles I and II (duplicates), note that the error bars are smaller than thepoints themselves. B) Rayleigh model for the slurry data. The slope of the regression is (α-1), where the efor bottles I and II are 2.2 ± 0.11 and 2.3 ± 0.09h respectively.76Applying the Rayleigh model to both slurry bottles yields isotope fractionations of 2.2 ± 0.1hand 2.3 ± 0.1h for B1 and B2 respectively. Comparing the isotope fractionation induced in ourwhole-core incubations (eeff = 1.1 ± 0.2h) to our slurries confirms that the apparent isotopefractionation observed in the whole core incubations was smaller than the intrinsic isotopefractionation determined from the slurries.3.4 Discussion3.4.1 Modelling Cr(VI) reduction and isotope fractionationWe compare our isotope fractionations obtained from each experiment to develop an accurateand robust description of Cr isotope fractionation in ferruginous sediments. The effective isotopefractionation (eeff = 1.1 ± 0.2h) estimated through a Rayleigh model applied to the δ53Cr(VI)composition of the overlying waters from whole-core incubations likely underestimates theintrinsic isotope fractionation (eint) imparted during Cr(VI) reaction within the sediments dueto a diffusive boundary layer effect [185]. Since Cr(VI) transport is limited by diffusion into thesediment, the isotopic composition of Cr(VI) becomes progressively disconnected from the originalsource pool of Cr(VI) in the overlying water with increasing depth in the sediment [185]. Overthe diffusive path 52Cr(VI) is preferentially removed by reduction and so the pool of porewaterCr(VI) remaining is driven towards heavier δ53Cr values. δ53Cr values increase as a function ofdepth below the sediment-water interface due to progressive reduction and removal of porewaterCr(VI). In the reducing zone, the instantaneous δ53Cr(III) also becomes progressively enriched inthe heavy isotope (as the reactant pool contains less 52Cr). The δ53Cr(III) values are offset fromthe pore waters by a constant δ53Cr value (i.e. the intrinsic isotope fractionation, eint), which isimparted by the reaction. We estimate eint using equations (10) and (11) and substituting ourRayleigh approximation for eeff (see Methods section). Table 3.5, shows a complete list of theparameters we employ in this model for variables D, k (s– 1), eeff (h), and (cm).77Table 3.5: 1D diffusion model parametersTube k (s -1) D (cm2 / s) εeff (‰) λ (cm) L (cm) εint (‰) A 9.54 x 10-6 3.23 x 10-6 1.1 0.75 0.2 2.8 B 9.54 x 10-6 3.23 x 10-6 0.9 0.72 0.2 2.3 C 9.54 x 10-6 3.23 x 10-6 1.1 0.74 0.2 2.9 D 9.54 x 10-6 3.23 x 10-6 1.0 0.65 0.2 2.7 	We utilize a literature value for Cr(VI) diffusion (3.23 x 10 – 6 cm2 s– 1, [194]. Without tighterconstraints on the rate of Cr(VI) reduction in the sediment itself, we use an approach similarto [185], and assume that the reduction rate per unit volume pore water in the sediments isthe same as that determined for our well-mixed slurries, yielding an average value k = 9.54 x10 – 6 s– 1. We assume this rate constant to be a minimum value for k and performed a sensitivityanalysis on eint as it relates to k, by varying this rate constant. Increasing the rate constant by anorder of magnitude from 9.54 x 10 – 6 s– 1 to 9.54 x 10 – 5 s– 1 in equation (10) results in a relativedifference of 17% in the overall intrinsic isotope fractionation resulting in a decrease of eint from2.80 to 2.31h. We note, that the rate calculated from our slurry experiments is 44% larger thanthe pseudo first order rate calculated from our Rayleigh model of the whole core incubations, aphenomenon also highlighted by [189] when using Rayleigh models to evaluate field isotope data.We used O2 penetration depth as a proxy for the depth of the NRZ (L) in our whole cores, whichwe parameterized as 0.2 cm. We directly determined oxygen penetration depth in Lake Matanosediments on cores recovered near our sampling site (Fig A.1), and detailed discussion on theinfluence of the depth of the NRZ on the model results is in section 4.3 below. Although we cannotexplicitly rule out that the NRZ depth of 2 mm may have changed slightly during core recovery,this effect has been shown to be minor (an oxygen penetration depth difference of up to ∼30%),when the oxygen penetration depth is shallow [195]. We do not expect appreciable porewaterexpulsion, decompression or temperature change to have varied greatly during core recoveryand thus after our pre-incubation of the cores, we expect a return to steady-state conditionsand an oxygen penetration depth very close to in-situ conditions (2 mm). As an example ofthis parameterized model calculation, we take the experimental boundary conditions of core D78(Table 3.2), apply them in equations (10) and (11), and observe that both the porewater Cr(VI)and reduced Cr(III) become progressively enriched in 53Cr with increased depth (λ), offset by theintrinsic isotope fractionation (eint) of 2.7h, for tube D (Table 3.5). If we apply this calculation toeach of the four incubations individually, we calculate a mean parameterized eint = 2.7 ± 0.2h.To explore the intrinsic fractionation in more detail we set up a reactive-transport model.This model differs from the other modelling approaches in that we set Cr(VI) reduction ratesbelow the NRZ to follow a Gaussian distribution as a function of depth. The model is setup as apseudo steady-state for each individual time point, using the measured [Cr], and δ53Cr(VI) fromour whole core incubation (Tube D), as boundary conditions. To estimate time-scales to reacha pseudo steady-state following the Cr(VI) doping, we calculated the time for Cr(VI) to reachthe reactive zone (2 mm) via diffusion. These calculations indicate this condition was satisfiedwithin 20 minutes of Cr(VI) injection (Appendix A). We run our model for each time point usingthe set of boundary conditions obtained from Tube D (Table 3.1) in order to describe changes inδ53Cr(VI) over the time course. We note that we have measured δ53Cr values at 6 time points,however, we interpolate the δ53Cr data in order to have a full set of boundary conditions from timepoints 2 – 8. The complete details of our model can be seen in Appendix A. As mentioned, Cr(VI)transport into the zone of reduction, occurs via diffusion through a specified depth of non-reactivezone (NRZ). We assume O2 inhibits Cr(VI) reduction within this zone, similar to selenium (Se)reduction modelled by [185]. Cr(VI) reduction only begins below the depth of O2 penetration intothe sediment, measured as 0.2 cm in lake Matano (Fig A.1). As mentioned, below the non-reactivezone (NRZ), we set Cr(VI) reduction rates to follow a Gaussian distribution. Without data onporewater Cr(VI) concentrations this is a somewhat arbitrary choice, but is justified based on ourknowledge of the redox structure in coastal sediment porewaters [148], which generally implyelectron donor limitation above the rate maximum and Cr(VI) limitation below the rate maximum.We modelled the integrated isotopic composition of Cr(VI)reactant and Cr(III)product for eachtime point at pseudo steady-state, producing a range of modelled δ53Cr(VI) and δ53Cr(III) valuesfor the entire time-course (2-8). We considered a range of isotope fractionation factors (eint = 1.0015– 1.003) and tracked how the modelled δ53Cr(III) and δ53Cr(VI) would change over the course ofthe incubations when these fractionations were imposed. We then compare these model outputs to79the measured isotope δ53Cr(VI) in the the overlying water (Table 3.1) to find the fractionation factorthat best describes the time-course data. We found that an intrinsic fractionation factor of αint =1.0018 (eint = 1.80 ± 0.04h) leads to a model that best reproduces the measured δ53Cr(VI) dataof the overlying water (Fig. 3.7) and is lower than the eint of 2.7h from the more parameterizedmodel.0 10 20 30 40 50 60 70Time (Hours)δ53 Cr (per mil)Measured DataAlpha 1.0015Alpha 1.0018Alpha 1.0020Alpha 1.0025Figure 3.7: Modelling of porewater δ53Cr(VI) values through time using different applied intrinsic fractionationfactors. We apply a range of fractionation factors (i.e., α = 1.002; 1.001; 1.0018; 1.0015; 1.0025) and model theδ53Cr(VI) in overlying lake water. Measured δ53Cr(VI) values in overlying lake water from the whole-coreincubation experiments are shown as black dots. A best fit to measured data is obtained utilizing an α =1.0018 or eint = 1.80h. The error on this best fit is expected to be <0.1h.Our various approaches to quantifying the fractionation factors that accompany Cr(VI) reductionin our whole core incubations are all founded on solid theory, but open system behaviour andtransport phenomena are variably considered in these approaches and these phenomena play animportant role in shaping how intrinsic fractionation is expressed at the scale of our observations.80Our reactive transport model contains an explicit and realistic parameterization of Cr(VI) reductionand transport rates, and thus likely provides the closest approximation to the intrinsic isotopefractionation (eint = 1.80 ± 0.04h) for Cr(VI) reduction within the sediment.Our three modelling approaches yield different estimates for the intrinsic isotope fractionationduring Cr(VI) reduction in the ferruginous sediment of Lake Matano. In these approachestransport phenomena and the simulation of kinetic isotope fractionation are variably considered.We did not however, explicitly employ a time dependant reaction transport model to quantitativelyassess reaction mechanisms. These models do exist and have been previously applied to the Crisotope system [196, 197]. Hence, treatment of our data in such models in the future may providevaluable further insight into Cr cycling and isotope fractionation in ferruginous environments.3.4.2 Comparison of Cr isotope fractionations obtained in slurries and coreincubationsOur whole core incubations and slurry experiments offer another window into the developmentof reservoir effects during Cr(VI) reduction in sediments underlying mildly oxygenated watercolumns. In Lake Matano and hence in our modelling approaches discussed above, we assumea completely open system at steady-state conditions, a constant isotope fractionation and anunchanging Cr(VI) reservoir. This produces a constant isotopic offset between the Cr(VI) pool andthe extractable authigenic Cr(III) pool (a scenario where the instantaneous Cr(III) product is equalto the accumulated Cr(III) product). Direct measurement of the authigenic δ53Cr composition of theupper 0.5 cm of sediment (Table 3.3) confirms that the isotopic composition of the sediment is offsetfrom the overlying lake water, as expected in diffusion regulated systems. This defines an isotopeseparation of ∆53Crlakewater – sediment = 0.21h to 1.81h ± 0.15h, based on first measurements ofthe lake water δ53Crlakewater = 2.20h to 3.30h ± 0.15h. This large range in the isotope separationvalues arises due to a 1.10h variability in the reported δ53Crlakewater composition, however wenote that this range captures values predicted based on the fractionation in our experiments. Inaddition, Table 3.6 summarizes our experimental results, and the isotope fractionations obtainedvia different experimental and modelling approaches.81Table 3.6: Compilation of experimental isotope fractionations for this study  Model Approach Isotope Fractionation ε (‰) 2se (‰) Whole Core Incubation Rayleigh 1.1  0.2 1D Diffusion 2.7  0.2 Reactive Transport 1.80 0.04 Slurry Experiment Rayleigh (B1) 2.2 0.1 Rayleigh (B2) 2.3 0.1 Direct Measurement ∆ = δ53Crlakewater – δ53Crproduct 0.21 – 1.81 0.15 	Our slurry experiments yield an intrinsic isotope fractionation (eint = 2.2h) that is greater than theRayleigh model (eeff = 1.1 ± 0.2h), but below that of the parameterized model (eint = 2.7 ± 0.2h)[184]. The intrinsic isotope fractionation that emerges from a different description of transport(eint = 1.80 ± 0.04h) that explicitly accounts for reservoir effects is smaller than the fractionationfactor from slurry incubations. Since this model explicitly describes Cr(VI) reduction rates andreservoir effects, the difference between the eint of the whole-core incubations and that of theslurries is likely real rather than an artefact of the calculation. We suggest that this difference islikely due to the relative role of different Fe species (Fe(II)aq and Fe(II)solidphase) under whole-coreversus slurry conditions, which we explore in detail below.Table 3.7, includes a compilation of existing e values for Cr(VI) reduction via a variety ofdifferent electron donors, and it has been shown that mixed valence Fe-(oxyhydr)oxides have aparticularly important role in Cr(VI) reduction [198].82Table 3.7: Compilation of isotope fractionations during reduction of Cr(VI) adapted from [1]. †[179] ‡[107][180] [106] κ[37] φ[199] ψ[108] θ[170] χ[73]Reference Isotope Fractionation (‰) Reductant †	 3.03; 2.17; 3.14; 3.01 G. sulfurreducens, Shewanella sp. (NR), P. stutzeri DCP-Ps1, D. vulgaris ‡	 3.91; 2.67; 2.65; 2.11 Goethite, siderite, green rust, FeS 	 3.07; 2.38 Incubated contaminated sediment ◊	 3.60; 1.50 Ferrous Fe; ‘Green rust’ κ  3.51 Magnetite φ  1.95; 0.38 Pseudomonas: aerobic; denitrifying ψ  4.20; 3.11 Low-pH Fe(II); organic reductants θ  4.11; 1.75 Shewanella oneidensis bacteria χ  3.54; 5.0 Acidic H2O2; pH-neutral H2O2 (kinetic) 	For comparison, [106] found that the isotope fractionation for reduction of Cr(VI) by a mixtureof dissolved Fe(II) and Fe(II) + Fe(II)greenrust was e = 1.5h, which is similar to our whole coreresults. In the batch experiments of [106], Fe(II)aq concentrations were high and hence Fe(II)aq waslikely the primary electron donor, leading to initial Cr(VI) reduction to Cr(III)-bearing green rustwhich has been shown to impart large isotope fractionations (e = 3.6h, Table 3.7). In the sameexperiments, reduction by Fe(II)-bearing green rust was shown to have a more muted isotopefractionation (e = 1.5h, Table 3.7), as adsorption and reduction of Cr(VI) in green rust interlayersimparts a minor isotope fractionation [106]. Table 3.8, displays a tabulation of both solid phaseand porewater Fe speciation data, measured on the upper 0.5 cm of Lake Matano sediment.Table 3.8: Fe speciation data for the upper 0.5 cm of sediment in Lake Matano 0.5 M HCl Extractable Fe in Upper 0.5 cm of sediment Fe(II) (µmol cm-3) 15.28 Fe(III) (µmol cm-3) 0.55  Porewater Fe in Upper 0.5 cm Fe(II) (µmol cm-3) 1.26 	83We observe that solid phase (0.5 M HCl extractable) Fe(II) concentrations are more than tentimes higher than [Fe(II)aq] measured in the porewaters. Based on the ferruginous nature of thesediments and this Fe-speciation data, we hypothesize our isotope fractionation of e = 1.80h islikely due to a combination of reduction via both Fe(II)aq and Fe(II)solidphase, and we hypothesizethat Cr(VI) reduction below the zone of oxygen penetration (2 mm), is likely governed by thehigher concentrations of reactive solid phase Fe(II), more so than Fe(II)aq. We note our modelledisotope fractionation (eint = 1.80h) is marginally larger than (e = 1.5h) from [106], thus weconduct a simple mass balance calculation in order to estimate the possible contribution (X) ofaqueous Fe(II) vs. solid phase Fe(II) species which may contribute to the overall fractionationfactor of 1.80h;(X · eFe(I I)) + ((1− X) · eFe(I I)solidphase) = 1.8h (3.12)Where X is the proportion of Cr(VI) reduced by a particular reductant (here: Fe(II), and eFe(II) andeFe(II)solidphase are the isotope fractionations for Fe(II)aq and Fe(II)solidphase (h), respectively). Weset eFe(II) = 3.6h and eFe(II)greenrust = 1.5h (e.g. [106]), and calculate X = 14%. This implies that 86%of Cr(VI) reduction likely occurs via Fe(II)solidphase in the upper 1 cm of Lake Matano, with a minorcontribution from Fe(II)aq. We note that we cannot explicitly rule out the impact of microbialCr(VI) reduction on the overall fractionation factor but acknowledge the almost certain role ofmicrobial metabolisms in Fe-redox reactions. Briefly for comparison [170] measured a ∼70%reduction in chromate concentration over a 15 – 20 day time interval in batch experiments withShewanella oneidensis using lactate as the organic substrate, and found an isotope fractionation (eint= 1.75h) similar to Fe(II)greenrust (Table 3.7). Previous workers also performed batch experiments,instead using Desulfovibrio vulgaris and pyruvate as the organic substrate and found a ∼90%reduction in chromate concentration over a 29 day time interval [179]. The rates of microbialCr(VI) reduction in both of the aforementioned experiments are an order of magnitude slowerthan our measured rates at similar Cr(VI) concentrations, implying the likely minor role of directmicrobial Cr(VI) reduction in ferruginous sediments.We also apply our mass balance calculation to our slurry data (again using equation 12), in84order to make the same assessment of the possible contributions of Fe(II)aq and Fe(II)solidphase,to the overall isotope fractionation. We use the same values; eFe(II) = 3.6h and eFe(II)solidphase =1.5h and calculate X = 35%. This suggests that reduction via Fe(II)aq may be more than twiceas important for Cr(VI) reduction in our slurry experiments compared to the intact sediments.Our reasoning for this is that in the slurry experiments, Fe(II) may be liberated from the intactsediment to the water added upon homogenization during vigorous shaking, which increases theFe(II)aq that is available to react with Cr(VI) and thus also the intrinsic isotope fractionation. Thiscan be observed in Figures 3.4 and 3.5, where the initial strong decrease of Cr(VI) in the slurryexperiments might have been due to a combined reduction by dissolved Fe(II)aq liberated fromporewater, and Fe(II)-bearing minerals. By quickly using up the Fe(II)aq reservoir the reductionrate of Cr(VI) then decreased after a couple of hours. These results indicate that the relativeconcentrations of mixed-valence Fe species can play an important role in setting the magnitude ofisotope fractionations expected in ferruginous systems. An isotope fractionation of eint = 1.80his characteristic for intact ferruginous sediments with relatively low porewater Fe(II) and highabundances of solid phase mixed-valence Fe minerals. Notably, we demonstrate that Cr(VI)reduction via a combination of Fe(II)aq and Fe(II)solidphase or similar phases will generally leadto sediment δ53Cr(III) signals that are offset from the overlying water from which it deposited.The magnitude of this isotope fractionation plays an important role in interpreting δ53Cr records,which we explore below.3.4.3 Implications for the chromium isotope recordOur data reveals that Cr(VI) reduction during early diagenesis is important in setting the δ53Crof ferruginous sediments. Diffusion of Cr(VI) and the development of a δ53Cr gradient betweenoverlying water and the zone of reduction, depends on the thickness of the NRZ. The thickness ofthe NRZ is directly dependant on the depth of oxygen penetration within the sediments, whichis linked to the O2 concentration of overlying water, sedimentation rates, and organic matterconcentration and reactivity. Assuming that eint = 1.80h is the characteristic intrinsic isotopefractionation for ferruginous sediments, we explore the effects of non-reaction zone depth onauthigenic depth integrated δ53Cr(III) value (i.e., the δ53Cr buried), which is ultimately important85for interpreting the sedimentary δ53Cr record. To explore how the δ53Cr composition behavesunder different environmental conditions, we performed a sensitivity analysis using our model.We applied a range of initial δ53Cr seawater values (0.0 - 2.0h), reflecting possible paleo-seawatercompositions [1, 41, 171], and track resulting depth integrated δ53Cr(III) composition as a functionof NRZ thickness (0.01 – 100 mm), resulting in a range of buried δ53Cr(III) from -1.79 to +1.95h(Fig. 3.8).0.01 0.1 1 10 100NRZ Depth (mm)-2.0-1.5-1.0-δ53 Cr(III) (‰)5 mmFigure 3.8: Sensitivity analysis for our steady state model. Circles are sediment δ53Cr(III) when seawaterδ53Cr(VI) is 0.0h, squares are sediment δ53Cr(III) when seawater δ53Cr(VI) is 0.3h, triangles are sedimentδ53Cr(III) when seawater δ53Cr(VI) is 1.0h, and crosses are sediment δ53Cr(III) when seawater δ53Cr(VI) is2.0h. The grey shaded area represents the ISE δ53Cr compositional range. Note the logarithmic x-axis.We also tested the model with boundary conditions similar to Lake Matano ([Cr(VI)] = 120 nmoll– 1, δ53Crlakewater = 3.0h, NRZ = 2 mm, eint = 1.80h) and find that the δ53Cr(III)sediment value ourmodel produces is 1.95h, which is within error of the measured sediment δ53Cr values (Table863.3). We observe that the modelled δ53Cr(III) is sensitive to the δ53Cr value of the overlyingwater, suggesting that authigenic sedimentary δ53Cr values are likely to be in part controlledby δ53Crseawater despite relatively large intrinsic isotope fractionation during Cr(VI) reduction insediments. The model also clearly illustrates that authigenic δ53Cr values from sediments overlainby oxygenated seawater best capture the seawater δ53Cr when the NRZ is thick, but are moresensitive to environmental conditions, like oxygen concentrations at the sediment-water interface,with thinner NRZ. We note, that in instances where the NRZ may be very thick, diffusion betweenthe overlying water and the reduction sites will be inhibited and supply of Cr to the reductionsites may be restricted. Thus, the amount of authigenic Cr formed will be small and possibly verydifficult to distinguish from detrital Cr.To examine the sensitivity of sediment δ53Cr values to O2 concentrations in overlying seawater,we used a 1D diagenetic model, which tracks O2 concentrations below the sediment water interface(the details of the model can be found in the Appendix A) [193, 195, 200–202]. We perform asensitivity test on this model by varying sedimentation rate, and the O2 concentration in theoverlying seawater (Fig. 3.9).870.0 0.5 1.0 1.5[O2] (µmol l-1) (mm)[O2] (umol l-1) Sedimentation =0.05 cm yr-1[O2] (umol l-1) Sedimentation =0.5 cm yr-1[O2] (umol l-1) Sedimentation =5 cm yr-1Figure 3.9: Oxygen penetration depth model. We model the oxygen penetration depth of marine sediment as afunction of sedimentation rate and organic matter reactivity, using a steady-state equation from [193]. Theinitial O2 at the sediment water interface is chosen to be 1.25 µmol l– 1 (1% PAL) to simulate hypothesizedlow oxygen conditions during the Neoproterozoic.To validate links between δ53Cr and bottom water O2, we compare diagenetic model outputsto data from Lake Matano. We obtain a NRZ depth comparable to Lake Matano when bottomwater [O2] is 5 – 50 µmol and the sedimentation rate is 0.5 cm yr– 1. These values are similar toLake Matano’s sedimentation rate of ∼ 0.1 cm yr– 1 and water column O2 concentrations [203].Notably, millimeter scale NRZ depths are permissible at low O2 concentrations (∼ 1% PAL) evenwhen organic matter reactivity is low, which we parameterize after [204], indicating that strongdiagenetic Cr isotope fractionation is possible even in low productivity environments.Our diagenetic models can be used to gain several insights into the sedimentary Cr isotoperecord. Foremost, our Cr model can be used to gauge the conditions that will lead to sedimentaryrocks with an igneous silicate earth (ISE) δ53Cr composition (-0.12 ± 0.1h, [46]). In an ocean-atmosphere system with O2 concentration below the threshold for manganese oxide production,Cr would be cycled in surface environments as reduced Cr(III), leading to sedimentary δ53Crcompositions that largely fall within the ISE range, indicating a lack of redox Cr cycling on the88continents [4, 7, 30]. Our model indicates, however, that seawater δ53Cr as high as the intrinsicisotope fractionation (in this case 1.8h) could still produce a δ53Cr(III) of 0h (Fig. 3.8). Therefore,if authigenic Cr concentration in the geological archive is high and measureable, it is possible thatthe thickness of the NRZ was small, which would have resulted in significant isotope fractionation.Our modelling shows that it is possible to obtain a NRZ of a couple millimeters (2.05 mm),under low oxygen conditions (1% PAL, 1.25 µmol l– 1), if the reactivity of organic matter andsedimentation rate is sufficiently low. On the other hand, under the same diagenetic conditions,our modelling also predicts a NRZ of just under 5 mm, when bottom water oxygen concentrationsare 100% PAL (Fig. 3.8). A NRZ of this depth still produces δ53Cr compositions in the sedimentthat fall within the ISE range when the seawater δ53Cr composition is 0.3h. Notably, the currentsedimentary Cr isotope record suggests seawater was typically characterized by δ53Cr heavierthan 0.3h [4], during periods in which the ocean-atmosphere system is independently constrainedto be fully oxygenated. This means that even small heavy Cr isotopes enrichments in sedimentmay indicate appreciable ocean-atmosphere oxygen. We acknowledge that there are limitations inthe application of our modelling approach to the reconstruction of past environments and thatadditional factors not constrained in our model may play a role in governing expressed Cr isotopefractionation such as, variable seawater Cr concentrations and dynamics in the physical, chemical,and biological properties of marine sediments through time. Despite these caveats however, ourwork demonstrates how NRZ depth exerts control on the Cr isotope cycle and the δ53Cr preservedin marine sediments, notably that δ53Cr values may be offset from seawater by as much as themagnitude of the intrinsic isotope fractionation.Lastly, and more speculatively, this work suggests it could be possible to use the Cr isotopevalues of marine sediments to estimate a sediment oxygen penetration depth. The range ofintegrated isotope fractionation factors imparted during early diagenetic Cr sequestration inferruginous sediments is appreciable (> 0.5h) relative to analytical error (< 0.1h), and may thusbe used to estimate paleomarine oxygen levels when the δ53Cr value of coeval seawater can beindependently constrained. This approach is appealing since bottom water oxygen concentrations,interesting in their own right, can also be linked to minimum atmospheric oxygen concentrations.893.5 ConclusionsWhole core incubations performed on ferruginous sediments demonstrate that Cr(VI) reductioninduces Cr(VI) concentration gradients that drive diffusive Cr fluxes across the sediment waterinterface. As reduction proceeded, the remaining lake water Cr(VI) became progressively enrichedin the heavy isotope (53Cr). Closed system Rayleigh distillation modelling yields effective isotopefractionations that fall within a narrow range eeff = 1.1 ± 0.2h, independent of initial Cr(VI)concentrations. Direct measurement of the δ53Cr composition of the upper 0.5 cm of Lake Matanoreveals an eeff = 1.1 – 1.5h. More detailed diagenetic modelling that accounts for open systembehaviour predicts an intrinsic isotope fractionation eint = 1.80 ± 0.04h, implying that appreciablereservoir effects develop during Cr(VI) reduction in sediments. Based on the magnitude of thisfractionation factor and previous experimental work, we suggest the Cr(VI) reductant in LakeMatano sediments is a combination of dissolved and solid phase Fe(II). Our results show that theisotopic composition of authigenic Cr in sediments can be offset from the δ53Cr values in overlyingwater when the zone of Cr(VI) reduction is sufficiently close to the sediment water interfaceresulting in an intrinsic fractionation that is expressed with minimal reservoir effects. Becauseof this reservoir effect, δ53Cr(III) similar to the ISE cannot reliably be interpreted as indicatingthe absence of atmospheric oxygen without additional constraints on the δ53Cr composition ofseawater. When the zone of Cr(VI) reduction moves deeper into the sediment, authigenic δ53Crvalues more faithfully record seawater δ53Cr. Our work further shows how the Cr isotope values ofmarine sediments from the geologic record may be used to estimate sediment oxygen penetrationdepths, which are linked to the oxygenation of the ocean-atmosphere system.90Chapter 4Ferruginous oceans during OAE1a andthe collapse of the seawater sulphatereservoirEarth surface oxidant budgets are assumed to have existed in a relatively stable state since thePrecambrian Eons. Oceanic anoxic events (OAEs) are intervals of widespread to global oceananoxia and they are linked to dynamics in the global biogeochemical cycles, major climaticanomalies, and biological crises. These recurrent perturbations to the Earth system over thelast 200 Myr thus challenge the paradigm of relative stasis in Phanerozoic Earth surface oxidantpools. Little is known, however, of the precise redox chemistry of seawater during OAEs andthis confounds models that aim to quantitatively link ocean anoxia to broader-scale oxidantbudgets in the Earth system. Here we show that during OAE1a, 120 Ma ago, the oceans wereanoxic and Fe-rich (ferruginous) for more than 1 million years. Like the ferruginous oceans ofthe Precambrian eons, development of ferruginous conditions at this time requires low seawatersulphate concentrations, which we show dropped to < 50 µM or more than a thousand timeslower than modern. This collapse in the seawater sulphate pool over just a few hundred thousandyears reveals previously unrecognized instability in Phanerozoic Earth surface oxidant budgetswith potential to dramatically alter global biogeochemical cycles, marine biology and climate onremarkable time-scales.914.1 IntroductionSeawater chemistry is generally thought to have evolved to its current well-oxygenated, sulphate-rich state between 540 and 420 million years ago (Ma) [12, 205]. Throughout much of thepreceding 3.5 billion years the oceans were largely anoxic, dominantly iron-rich (ferruginous),and punctuated by intervals of widespread euxinia (hydrogen sulphide rich) [7, 182]. Theseconditions waned in the early Phanerozoic, and thus, for much of the last 500 Ma, marine andglobal biogeochemical cycles were thought to have been relatively stable, operating much as theydo today [206? ]. Large-scale oceanic anoxic conditions, however, re-emerged periodically in thePhanerozoic [207] and were particularly prevalent during warm periods such as the Cretaceous.The oceans developed euxinia during a number of these oceanic anoxic events (OAEs)[208–210] when pelagic microbial respiration was channelled through sulphate reduction producinghydrogen sulphide that accumulated in poorly ventilated water masses. Emerging evidence,however, also suggests that ferruginous conditions may have re-emerged at least transientlyduring some of the OAEs (OAE2, OAE3 and the end-Permian OAE)[210, 211] implying large-scaledynamics in ocean chemistry and the sulphur cycle, since development of ferruginous conditionshinges on the balance between Fe and S delivery and removal from the oceans [182].Sulphur isotope signals during the Cretaceous period reveal that seawater sulphate was drawndown to low millimolar concentrations over millions of years before OAE1a in response to episodicdeposition of basinal-scale evaporites [212, 213]. Although the precise timing of pre-OAE1aevaporite deposition remains poorly constrained [214], ensuing seawater sulphate concentrations(< 3 mM) were much lower than the modern ocean (28 mM) and this low seawater sulphatewould have strongly influenced global biogeochemical cycling. For example, seawater sulphateconcentrations are an important control on marine methane (CH4) cycling, with super-millimolarsulphate concentrations attenuating the release of CH4 generated in modern marine sedimentsto the atmosphere through microbial anaerobic methane oxidation [215, 216]. Sub-millimolarsulphate concentrations, in contrast, can lead to large-scale oceanic CH4 efflux with correspondingimplications for climate [217]. While S-isotope data imply low sulphate concentrations, existingmodels lack the resolution needed to place robust upper limits on possible marine sulphate92concentrations that would facilitate reconstruction of the climate system and the effects of OAEdevelopment on the biosphere.4.2 MethodsTo explore marine sulphate concentrations in the early Cretaceous, we studied a suite of sedi-mentary rocks from the eastern margin of the Paleo-Tethys Ocean (Cismon drill core) and themiddle of the Paleo-Panthalassic Ocean (Deep Sea Drilling Project (DSDP) Site 463) that capturethe Aptian OAE1a (∼120 Ma). This interval is delineated by the deposition of organic matter(OM)-rich black shales and lasted for more than a million years [218, 219]. We analysed rocksfrom OAE1a using selective sequential extractions to partition Fe speciation between operationallydefined phases (Supplementary Information), and distinguish between Fe burial as pyrite andother forms of reactive Fe (i.e. non-lithogenic). These sequential extractions target Fe-phasesconsidered highly reactive (FeHR = sum of all non-silicate Fe) towards sulphide (pyritization) andbiological and abiological Fe(III) reduction under anoxic conditions [110, 128, 129]. Fe-speciationanalyses thus enable us to discriminate between sediments deposited under anoxic ferruginousversus euxinic water column states, based on the degree of pyritization of the FeHR pool [129, 182].Notably, preservation of FeHR in sediments that also contain appreciable organic matter impliesthat pyritization was sulphate limited [124].4.3 Results and Discussion4.3.1 Ferruginous conditions during OAE1aOAE1a is accompanied by a C-isotope excursion and deposition of OM-rich sediment, which areevident between ∼18 - 24 m in the Cismon core and ∼615 - 625 m in DSDP Site 463 (Fig. 4.1).93Nannoconid DeclineNannoconid DeclineNannoconid CrisisNannoconid CrisisFigure 4.1: Fe-speciation and carbon isotope records for Cismon and DSDP Site 463. The grey shaded region(OAE1a) represents ∼1.1 Mya [219], C-isotope stages C2–C7 [219–221]. Panels (a-e) are Cismon data andpanels (f-g) are DSDP Site 463 data. (a) Cismon carbon isotope data from [222]. Rhenium concentrationdata from [223]. (b) Carbonate C after [224], and organic matter C data after [223]. (c) FePyritizable; Sum ofall pyritizable FeHR pools (FeCarb, FeOM, FeOx). (d) FeSil; silicate Fe (e) FePyr; Pyrite Fe. Green diamondsrepresent pyrite concentration data from [212]. (f) DSDP Site 463 carbon isotope data from [223]. Rheniumconcentration data from [223]. (g) Carbonate C data from [222], organic C data from [222, 225]. (h)FePyritizable; Sum of all pyritizable FeHR pools (FeCarb, FeOM, FeOx) (i) FeSil; silicate Fe. (j) FePyr; Pyrite Fe.Lower dashed lines represents the start of the nannoconid decline and the upper dashed lines representsthe start of the nannoconid crisis [218]. Integrated stratigraphy of the Cismon and DSDP Site 463 after[218, 222]. 94Our Fe-speciation analyses reveal enrichments of pyritizable Fe (FeHR) across these same intervalsin both Cismon and DSDP Site 463 rocks (Fig. 4.1), relative to rocks stratigraphically above andbelow. Ratios of FeHR/FeTot > 0.38 imply sediment deposition beneath anoxic waters if ratios ofFeTot/Al are also > 0.5 and Corg contents are > 0.5 wt% [129, 226]. FeHR/FeTot values recordedin the Cismon core during OAE1a are consistently above 0.38 and indeed have FeTot/Al > 0.5along with Corg > 0.5 wt%, characteristic of deposition below an anoxic water column (Fig. 4.2).Fe Pyr + Fe Ox(new) / Total FeHR new NO OMFe’Pyr/FeHR Fe’Pyr/FeHRFigure 4.2: Fe-speciation and Fe/Al records of the Cismon (left panels) and DSDP Site 463 (right panels). Thevertical and horizontal dotted lines refer to the oxic-anoxic threshold (FeHR/FeTot = 0.38) and ferruginous-euxinic threshold (FePyr/FeHR = 0.70) respectively. The open circles represent FePyr’/FeHR, the maximumpossible amount of pyrite present in the samples assuming the very unlikely scenario where the entireFeOx pool is a result of pyrite oxidation. The solid vertical line in the top panels refer to the Fe/Al ratioof 0.5. The solid vertical line in the bottom panels refer to the average rhenium concentration of oxicpelagic sediments, with the purple shading representing a 3 sigma uncertainty on this value (0.4 pbb[227]). Litho-, bio- and magneto-stratigraphy is the same as for Figure 4.1. The images are SEM (greyscale)and petrographic (color) micrographs of pristine framboidal pyrites preserved in the Cismon sedimentsdeposited during OAE1a.FeHR/FeTot ratios in rocks that bound OAE1a have FeHR/FeTot < 0.38) (Fig. 4.2). FeHR/FeTot inrocks deposited at DSDP Site 463 also capture intervals with values > 0.38 and Fe/Al > 0.5,similarly implying deposition under anoxic conditions. Some FeHR/FeTot values of sediments95deposited during OAE1a at site 463, however, are below the threshold (> 0.38) used to diagnoseanoxic ocean conditions and are thus ambiguous to depositional redox state based on Fe-speciationalone. We note however, that Fe-speciation analyses cannot diagnose sediment deposition underoxic conditions as the mass balance and supply of FeHR precludes it from being enriched in alldepositional environments simultaneously. Ratios below 0.38 can thus also result from depositionbeneath an anoxic water column and instead likely signify a restricted supply of FeHR to thispelagic site (Fig. 4.2). In strong contrast, epicontinental sediments like those deposited at theCismon site likely served as the main oceanic sink of FeHR. FePyr/FeHR ratios are a direct measureof the degree of pyritization of the highly reactive Fe pool and ratios > 0.7 signal depositionunder a euxinic water column [110, 128, 228]. Values below 0.7 indicate non-pyritized FeHR and,by definition, reveal insufficient sulphide supply to pyritize the available FeHR thus precludingaccumulation of free sulphide and the development of euxinia. Values below 0.7 in sedimentsdeposited under anoxic conditions signal sulphate depletion and a ferruginous water column. Allrocks deposited at the Cismon and DSDP 463 sites during the OAE1a interval have FePyr/FeHR<< 0.7 indicating deposition under strictly non-euxinic ferruginous conditions.As an additional test of depositional redox states we examined the distribution of rhenium(Re) - a trace element that is enriched under both ferruginous and euxinic anoxic conditions,though to a lesser extent under euxinic conditions [34]. Sediments from both the Cismon andDSDP sites are highly enriched in rhenium throughout the OAE1a interval relative to average oxicpelagic clay [227], confirming deposition under anoxic conditions at both sites (Fig. 4.2). Moreover,deposition under euxinic conditions is generally accompanied by strong molybdenum enrichment,and yet molybdenum is almost entirely absent from the sediments deposited during OAE1a,while many other redox sensitive trace elements (Cr, Re, V) are strongly enriched (SupplementaryInformation, Table B.1 and [229]). Together our Fe-speciation data coupled to Re and Mo datafrom both sites strongly suggest anoxic ferruginous conditions and imply that such conditionsspanned from the Tethys to the Pacific Ocean during OAE1a. Development of these pervasiveferruginous conditions, furthermore, implies that delivery of Fe to the oceans outpaced that ofS with corresponding implications for the Aptian marine sulphur cycle and seawater sulphateconcentrations.96Prior work, however, suggests that sedimentary rocks can be subject to sample storageartefacts that develop due to exposure to oxygen in the atmosphere and subsequent oxidationof sedimentary pyrite [230, 231]. We thus took care to work with well-preserved rocks, butnevertheless also evaluated the extent to which post depositional oxidation of FePyr in our samplescould have led to false diagnosis of either anoxic or ferruginous conditions (see SupplementaryInformation section 4). Oxidation converts pyrite and siderite to Fe-(oxyhydr)oxides and thuscauses redistribution of Fe from FePyr and FeCarb to FeOx. Since these three pools are summed inthe FeHR pool, oxidation would not have impacted FeHR/FeTot or Fe/Al ratios, and thus not ledto false diagnosis of anoxia based on Fe speciation. Diagnosis of ferruginous conditions is basedon FePyr/FeHR <0.7, and since pyrite oxidation would decrease this ratio it could lead to falsediagnosis. We therefore tested the extent to which oxidation might have altered FePyr/FeHR bysumming FeOx, the product of oxidation, and FePyr to come up with a maximum possible pre-oxidation ratio, FePyr’/FeHR. The small size of the FeOx pool relative to FePyr indicates negligiblepost depositional oxidation and we thus find that although it is possible that samples from site463 have experienced greater degree of pyrite oxidation, FePyr’/FeHR is mostly below the 0.7threshold, and indicates deposition under ferruginous conditions (Fig. 4.2). Likewise, if total S(pyrite S and any S pool resulting from oxidation) is used as a proxy for pre-oxidation pyriteFePyr”, FePyr”/FeHR also remains below the 0.7 (Fig. S2). Furthermore, the preservation of bothpristine framboidal pyrites (Fig. 4.2 and Fig. B.2) and abundant FeCarb (Table B.4 and B.5) isentirely inconsistent with pervasive post depositional sample oxidation. Collectively, therefore,our Fe speciation and trace element data unambiguously document deposition of both DSDP andCismon sediments under ferruginous conditions during the Aptian. Development of ferruginousconditions at this time would have required very low seawater sulphate concentrations.4.3.2 Modelling Aptian seawater sulphate concentrationsTo reconstruct Aptian seawater sulphate concentrations, we setup a 1D water-column reactivetransport model (see Supplementary Information) to simulate a stratified Cretaceous ocean. Pyriteburial fluxes calculated by combining sedimentation rates with sediment pyrite concentrations[219] tether our model outputs to the geologic record and place upper limits on the flux of sulphate97that can be converted to pyrite through microbial sulphate reduction and reaction of the sulphideproduced with FeHR, either directly in the water column or in bottom sediments. With microbialsulphate reduction rates parameterized based on modern marine ecosystems and analogies tolow-sulphate ferruginous environments [143, 232] our water column modelling yields an averageestimate for seawater sulphate of less than 5 µM (Fig. 4.2).A B CFigure 4.3: Models of marine sulphur cycling during OAE1a. (a) Depth integrated sulphate reduction rates asa function of initial sulphate concentration. The solid green line represents the average pyrite depositionalflux whereas the pink shaded region delineates the entire range of pyrite depositional fluxes observed atthe Cismon site. (b) Modelled rates of microbial sulphate reduction in a stratified ocean water column withdifferent surface seawater sulphate concentrations. (c) Resulting sulphate concentration profiles.Similar results are obtained modelling sulphate reduction rates in underlying sediments (TableB.6). We note that the low sulphate reduction rates in our models are also consistent with the largeS-isotope fractionations observed in Cismon pyrites, given that large S-isotope fractionations tendto develop at low sulphate reduction rates [233]. The large S-isotope fractionations recorded inthe Cismon pyrites, however, may also require a contribution from S disproportionation (δ34Spyrite-47h [212], Fig. B.5) which is known to be active in modern Fe-rich sediments [234] . Imposinghigher sulphate concentrations in our models with realistic rates of sulphate reduction andtransport across oceanic pycnoclines or the sediment-water interface yield pyrite burial fluxesmuch higher than those recorded in Cismon rocks. Our modelling thus constrains Aptian seawatersulphate concentrations to more than a thousand times lower than modern seawater sulphateconcentrations.Previous studies suggested that seawater sulphate concentrations [235] were drawn down to98low mM levels preceding OAE1a as the result of basin-scale evaporite (BSE) deposition associatedwith the opening of the South Atlantic [212, 213]. The exact timing of BSE deposition, however,is poorly constrained and its onset may have occurred as early as the Barramanian (129 – 124Ma), up to 5 Myr before the initiation of OAE1a [214]. Our results nonetheless, require furtherreduction of the seawater sulphate pool from low mM to low µM concentrations in a time framecommensurate with the onset of ferruginous conditions prior to the initiation of OAE1a, asdelineated by the carbon isotope record. Evaporite mineral deposition effectively draws sulphatedown from modern day concentrations of 28 mM to 1 mM, but when sulphate concentrations dropbelow 1 mM, seawater saturation with respect to sulphate-minerals (gypsum) during evaporationrequires unrealistically high Ca2+ concentrations (see Supplementary Information). Seawatersulphate drawdown to low mM concentrations therefore requires a second phase of sulphatesequestration and an alternative burial mechanism. Another important sink for seawater sulphateis microbial sulphate reduction and pyrite deposition and burial. Expansion of this sink over afew 100 kyrs preceding OAE1a provides a plausible mechanism to lower seawater sulphate to µMconcentrations and drive development of ferruginous conditions, which we explore below.Sulphur mass balance modelling reveals that an increase in global pyrite deposition ratesfrom 0.66 to a maximum of 1.31 Gmol yr– 1, assuming modern S-input fluxes (see SupplementaryInformation), would cause seawater sulphate concentrations to drop to low µM levels (Fig. 4.4).994 μMFerruginous OAE1a20253035404550556065Model Time (Myr)01,0002,0003,000[SO 42-] (μM)20253035404550556065Model Time (Myr) Flux (Gmol yr-1) EvaporiteNon Fractionated-SPyrite20253035404550556065Model Age (Ma) (‰)SInputABCFigure 4.4: Evolution of the Cretaceous seawater sulphate reservoir. (a) The grey line is our model result for theevolution of seawater sulphate concentrations. The green shading represents the timing of the nannoconniddecline [218], grey shading represents the relative timing of OAE1a based on C-isotope stratigraphy [220],while the dotted vertical line represents the timing of the large negative C-isotope excursion that commonlydemarcates the start of the OAE1a event. (b) Modelled S input and burial fluxes. (c) Isotopic compositionof Aptian seawater sulphate. The grey line represents our model results for the evolution of δ34Ssulphate.The green circles are δ34SCAS from [221] and the blue circles are 34Ssulphate from [236]. The vertical dottedline represents the timing of the negative C-isotope excursion [221].We note that pyrite deposition rates are higher under anoxic water columns [132] and thus theincrease in pyrite burial could be achieved by expanding the global extent of water column100anoxia. For example, if water column anoxia expanded from 0.1%, its extent in the modernocean [237], to 4%, as estimated for OAEs [238], the increase from 0.66 to 1.31 Gmol yr– 1 couldbe achieved with area specific pyrite deposition rates of 0.1 mol S m– 2 yr– 1 in regions of oceananoxia, which is similar to rates of pyrite burial in sediments underlying modern OMZs [239, 240].This pyrite burial interval would also drive an increase in the δ34S composition of seawatersulphate due to enhanced burial of isotopically light S in pyrite. Such an increase in δ34S is indeedrecorded in carbonate associated sulphate (CAS) providing geological constraints for the timingand magnitude of the pyrite burial event (Fig 4.4). We also note that the expansion of ocean anoxiaassociated with the pyrite burial event, as constrained by the CAS record, is contemporaneous withmarine biological crises prior to OAE1a [218] that are a possible result of ocean deoxygenation[241] (Fig. 4.4). S budgets (see Supplementary Information) during the OAE1a interval itself,however, imply an additional sink for marine sulphate that balances enhanced S inputs throughhydrothermalism [221]. Mass balances based on the isotopic composition of CAS deposited duringOAE1a as a proxy for δ34S of seawater sulphate further suggests that this additional sink carriesnear seawater δ34S values. The magnitude of this additional sink is between, 5.3 x 10 11 and 2.2 x10 12 mol yr– 1, depending on the S concentration in hydrothermal fluids at this time [242].In the absence of gypsum deposition at seawater sulphate concentrations less than 1 mM, wesuggest that sulphur was likely removed as a combination of biomass derived organic sulphur,CAS itself, or barite associated with organic matter, all of which are known to operate as sulphursinks in low sulphate modern and ancient environments [243, 244]. While biomass associatedorganic S has been shown to be a major pathway for S burial in lacustrine environments [245, 246],it is often overlooked as an important component of the marine S cycle. Marine organisms arecharacterized by C:S molar ratios of 50:1 [247, 248], and thus the burial flux of organic S is likelyon the order of 5 wt% that of organic C. Given the total S and organic matter contents of theCismon sediments [223], and assuming all non-pyrite S is buried as biomass associated organic S,we calculate an average organic matter (OM) C:S molar ratio of ∼45 during OAE1a, revealing OMburied during OAE1a has a similar C:S composition to that of modern, supporting the idea thatduring intervals of increased OM burial biomass associated organic S may become an importantsink of S from the ocean. Marine organisms, furthermore, assimilate S with a δ34S composition101nearly identical to seawater [248–250]. Given pre-OAE1a primary production similar to modern(48 Gt C yr– 1)[237] and OM that is ∼5 wt% S, the dramatic (up to a 35-fold [223]) increase inOM burial during OAE1a would be more than sufficient (4.5 Gmol yr– 1) to remove most of the Sentering the ocean at this time (3.3 Gmol yr– 1). Combined, our geochemical data documentingferruginous conditions during OAE1a, δ34S records of seawater sulphate, and our models, thusimply that biomass associated organic S likely represents an important but previously overlookedsink for sulphur in the low sulphate Aptian oceans. More importantly, our data and modelsreveal sulphate drawdown to low µM concentrations is possible on 100 kyr timescales with nonanomalous pyrite burial rates similar to modern oxygen deficient marine environments like thePeru Margin [240], and Chilean OMZ [239] (Fig. 4.4b and Fig. B.7). Development of ferruginousconditions during the Aptian can thus be attributed to widespread anoxia and ensuing pelagicsulphate reduction immediately prior to OAE1a against the backdrop of low Cretaceous seawatersulphate concentrations.4.3.3 ImplicationsAt 28 mM, seawater sulphate is an oxidant pool twice the size of modern atmospheric O2. Adecline to low µM seawater sulphate concentrations thus indicates a collapse of global oxidantpools during OAE1a with implications for marine ecology, biogeochemical cycling, and climate.Water column anoxia, for example, may have extended at least transiently into the photic zoneduring OAE1a with potential to influence photosynthetic ecology. Biomarkers indicative ofphototrophic Clorobi have indeed been recovered in sediments deposited during OAE1a [225, 251].While phototrophic Clorobi are best known as sulphide oxidizing phototrophs [131], they arealso known to grow on ferrous iron (photoferrotrophy) and were likely key primary producers inthe ferruginous oceans of the Precambrian [137]. Preservation of biomarkers from phototrophicClorobi in Aptian ferruginous sediments may thus signal the return of photoferrotrophy to thePhanerozoic oceans. At the same time, low-sulphate, ferruginous ocean conditions would haveboth channelled organic matter degradation from sulphate reduction to methanogenesis andprecluded the consumption of this methane through anaerobic oxidation with sulphate. Earthsystem modelling indeed shows that seawater sulphate concentrations below 1 mM promote102massive marine methane fluxes to the atmosphere with attendant greenhouse warming [217].4.4 ConclusionsDevelopment of ferruginous conditions during OAE1a thus reveals large-scale dynamics inEarth’s biogeochemical cycles over intervals of just a few hundred thousand years (Fig. 4.4).The development of ferruginous ocean conditions during multiple OAEs [208, 210] may thussignify a general instability in Earth surface redox budgets and the recurrent reorganization ofmajor oxidant pools at Earth’s surface, like seawater sulphate, during the Phanerozoic Eon. Themechanisms driving this reorganization remain uncertain, but could be addressed through betterconstraints on global sulphur budgets and the drivers of ocean de-oxygenation, as well as thedevelopment of Earth system models that resolve such large-scale biogeochemical dynamics overrelatively short time scales.103Chapter 5Magnetite biomineralization inferruginous waters and early EarthevolutionBurial of vast quantities of magnetite (Fe2+Fe3+2 O4) in Precambrian sedimentary rocks knownas iron formations (IFs) played a key role in the protracted oxidation of Earth’s surface duringthe Precambrian eons. Formation of this magnetite is commonly attributed to diagenetic andor metamorphic reactions, though laboratory experiments also show magnetite formation as aproduct of low temperature microbial metabolisms. Uncertainty in the precise provenance ofmagnetite in IFs confounds models that aim to mechanistically connect the processes that droveIF deposition to Earth surface redox budgets. Here we show magnetite formation occurs directlyin the water column of two modern ferruginous lakes. We find that biologically reactive Fe(III)phases are quantitatively reduced in the anoxic waters of both lakes, leading to the formationof primary authigenic magnetite. The formation of this magnetite and its links to microbial Fereduction thus control sediment redox budgets and imply primary magnetite formation may havebeen a principle mode of Fe delivery to IFs. Combined, the role of magnetite burial in settingEarth’s surface redox budgets and the locus of magnetite formation in the water column, suggestthat seawater chemistry and biological reactions would have played a key role in regulating thepace of Earth’s surface oxidation throughout the Precambrian eons. We also find that primarymagnetite takes conspicuous framboidal forms, which given the link to Fe reduction, may providea biological signature on early Earth and Mars.1045.1 IntroductionBiogeochemical cycling of iron (Fe) and carbon (C) plays a key role in setting Earth’s surfaceredox budgets and climate. Burial of reduced forms of C and Fe in marine sediments as organicmatter (OM) and ferrous Fe (Fe(II)), in particular, increase Earth’s surface oxidation state andrepresent quantitatively important net sources of oxygen to the modern atmosphere [8, 15, 252].Oxidation of crustal Fe(II) to form Fe(III)-bearing minerals, by contrast, is a sink for oxygen [5, 14].The magnitudes of these sources and sinks are influenced by dynamics in coupled C and Fecycling, which induce secular variation in Earth’s surface chemistry and climate [14, 15, 253]and over geological timescales this can lead to fundamentally different ocean-atmosphere redoxstates and climate systems [14, 253]. During much of the Precambrian eons, for example, theocean-atmosphere system was nearly oxygen-free and rich in reduced chemical species (Fe(II),H2 and CH4) [254, 255]. Intense deposition of Fe minerals at this time formed sedimentary rocksknown as iron formations (IFs) as the result of coupled C and Fe cycling in the oceans, andthese enormous deposits contain vast quantities of both oxidized and reduced Fe [117, 256]. IFmineralogy, notably, is dominated by magnetite (Fe2+Fe3+2 O4) [117] and burial of magnetite in IFshas been implicated in the redox evolution of the early biosphere [5, 14, 257, 258]. For example,the oxidation of crustal Fe(II) to form magnetite in IFs represents a net oxidant sink for theocean-atmosphere system, while the ultimate production of hydrogen as a corresponding sinkfor reduced equivalents can lead to planetary oxidation through H2-loss to space [253]. Factorscontrolling magnetite formation and burial in IFs and associated links to C cycling, therefore,likely played a key role in the protracted oxidation of Earth’s surface throughout the Precambrianeons.The precise pathways that drove magnetite formation and deposition in IFs remain poorlyconstrained and thus models that aim to mechanistically link the formation of IFs to the evolutionEarth’s surface redox budgets remain equivocal. The origins of magnetite observed in IFs arecontentious, but magnetite is often given an early diagenetic and or metamorphic provenance[257, 258]. Diagenetic and metamorphic models for magnetite in IFs suggest primary Fe depositionas Fe(III)-(oxyhydr)oxides (ferrihydrite, goethite, green rust) and these precursor phases were105transformed to magnetite during later-stage diagenetic and thermochemical Fe-reduction [121,259, 260]. Diagenetic and thermochemical origins for magnetite in IFs imply a decoupling ofIF mineralogy from the primary water column processes that cause IF deposition, and thusmagnetite would need not directly reflect the physicochemical and biological conditions ofthe ocean-atmosphere system at the time of IF deposition. Direct precipitation of primarymagnetite from the Precambrian oceans may also have been possible, as magnetite formation hasbeen documented in the laboratory as the product of biological processes including anoxygenicFe(II) photosynthesis [261] and Fe(III) respiration [121, 122], that were likely widespread underferruginous ocean conditions [142, 261–263]. The formation of primary magnetite would directlyimplicate these processes as mechanisms that control the deposition of IFs, set their mineralogy,oxidation-state, and chemical composition, and thus influence Earth surface redox budgets. Italso would imply that magnetite records primary information on the chemistry and ecology ofancient seawater. Primary water column magnetite formation and its role in IF deposition isseldom considered, however, in part because low temperature magnetite precipitation in naturalenvironments is infrequently observed today [260].Modern ferruginous environments provide natural laboratories to examine processes extensibleto the Fe-rich Precambrian oceans [137, 143]. To investigate Fe-cycling and mineral formationunder ferruginous conditions we conducted experiments in, and collected samples from LakesMatano and Towuti (herein referred to as LM and LT), on Sulawesi Island, Indonesia. LM andLT are part of the interconnected Malili Lakes system. The catchment basin surrounding thelakes is dominated by ultramafic rocks of ophiolitic origins and weathering of these rocks hasled to development of exceptionally Fe-rich lateritic soils [137, 140]. Heavy tropical rains deliverstrong fluxes of Fe (oxyhydr)oxides from these soils to the lakes, which exert an overwhelminginfluence on the lakes’ biogeochemistry [137, 142]. Both LM and LT are physically stratified andcharacterized by persistently anoxic Fe(II)-rich (∼140 and ∼10 µM respectively) and virtuallysulphate free deep waters [143] (Fig. 5.1).1060.0 5.0 10.0[Fe] (uM)050100150200Depth (m)0.0 75.0 150.0[Fe] (uM)050100150200Depth (m)0 125 250[O2] (uM)85 90 95 100Light Transmission (%)7.0 7.5 8.0 8.5 9.0pH-5 0 5 10 15Saturation Indicesa b c dLake TowutiLake MatanoMgFMgGSidGRMgFMgGSidGRFigure 5.1: Chemical and physical properties of LM (bottom panels) and LT (top panels). The blue hexagonsrepresent the sediment trap deployment depths in LT and LM, 110 and 160 m and 90 and 130 m respectively.a) Dissolved O2 and Fe(II). b) Light transmission c) pH. d) Mineral saturation indices. Closed black circlesrepresent the SI for magnetite calculated using Fe(III) concentrations assuming ferrihydrite saturation(MgF). Open black circles represent the SI for magnetite calculated using Fe(III) concentrations assuminggoethite saturation (MgG). We note both lakes’ deep waters are oversaturated with respect to magnetite.Closed brown circles represent the SI for siderite (Sid). Closed green circles represent the SI for carbonategreen rust (GR).107We combined geochemical and mineralogical analyses to parse Fe cycling and pelagic Fe mineralformation in both LM and LT, revealing that water column Fe reduction leads directly to theformation of primary authigenic magnetite. This magnetite comprises a major component of theprimary Fe mineral assemblage exported to the underlying sediments.5.2 Fe-speciationCanonically reactive Fe (FeHR; see Supplementary Information), which is thought to representFe-species reactive towards anaerobic microbial respiration and chemical reduction [128], is animportant component (> 90%) of the total Fe pool in suspended and deposited material fromboth LM and LT (Fig. 5.1 and Table 5.1).Fe - Non ReactiveFe - Highly ReactiveFe(II)HRFe(III)HRFeAcaFe(II)HClFe(III)HClFeDithFeOxaFeMagneticFeNon MagneticFeNon Magnetic - Oxalate ExtractableFe - Non ReactiveFe - Highly ReactiveFe(II)HRFe(III)HRFeAcaFe(II)HClFe(III)HClFeDithFeOxaFeMagneticFeNon MagneticFeNon Magnetic - Oxalate ExtractableFe - Non ReactiveFe - Highly ReactiveFe(II)HRFe(III)HRFeAcaFe(II)HClFe(III)HClFeDithFeOxaFeMagneticFeNon MagneticFeNon Magnetic - Oxalate ExtractableTowuti 110 mTowuti 160 mMatano 90 mFe - Non ReactiveFe - Highly ReactiveFe(II)HRFe(III)HRFeAcaFe(II)HClFe(III)HClFeDithFeOxaFeMagneticFeNon MagneticFeNon Magnetic - Oxalate ExtractableMatano 130 mabcdFe HRFe  NRFeNRFeHRFe(II)Fe(III)FeAcaFe(II)HClFe(III)HClFeDithFeOxaFeNonMagneticFeMagneticFeNMOEFe(II)Fe(III)FeAcaFe(II)HClFe(III)HClFeDithFeOxaFe NonMagneticFe MagneticFeNMOE0 20 40 60 80 100 0 20 40 60 80 100Fe Phase (% Fe Tot ) 110 m05101520253035Magnetite Flux (mmol m-2 yr-1)160 m(0-0.5 cm)(35-40 cm)130 mFe Phase (% Fe Tot ) 90 m{{Fe HRFe HRMatanoTowutife90 m 130 mFigure 5.2: Malili lake Fe-speciation and magnetite fluxes. a-d) Water column Fe-speciation sunburst plots. Eachoperationally defined Fe-phase is normalized to the total Fe content of the sample. FeMagnetic, FeNonMagneticand FeNMOE, refers to our sediment mass balance oxalate leaching tests (see Supplementary Information).e) Photographs of filtered sediment trap material from LM. We note the striking color difference of materialsedimenting above (shallow sediment trap, 90 m) and below (deep sediment trap, 130 m) the chemocline. f)Magnetite fluxes in the sediment traps and sediments in LM and LT. LT core top samples come from the 0 –0.5 cm sediment depth interval, whereas core bottom samples from the 35 – 40 cm sediment depth interval.108Table 5.1: Fe fluxes in the Malili lakesSediment Trap Samples Magnetite Flux (mmol m-2 yr-1) FeAca Flux (mmol m-2 yr-1) Fe(III)HCl (mmol m-2 yr-1) Total FeHR(II) Flux (mmol m-2 yr-1) Total FeHR(III) Flux (mmol m-2 yr-1) Magnetite Production (mmol m-2 yr-1) Fe(III) Reduction (mmol m-2 yr-1) Matano, 90 m 11 13 19 32 73 8 52 Matano, 130 m 19 15 16 84 107 Towuti, 110 m 13 28 73 26 145 7 30 Towuti, 160 m 20 47 101 56 245 Sediment Samples        Matano Core Top (0.25 cm) ND ND ND ND ND ND ND Matano Core Bottom (20 cm) ND ND ND ND ND Towuti Core Top (0.25 cm) 29 52 0 84 89 0 31 Towuti Core Bottom (40 cm) 27 57 0 114 51 	Analyses of water column particles collected from sediment traps deployed above and beloweach lake’s chemocline, as well as their bottom sediments, inform on the locus of Fe mineraltransformations that take place both during transport through the water column and during earlysediment diagenesis. Fe mineral concentrations in material recovered from sediment traps andbottom sediments were determined using operationally defined sequential extractions designedto speciate Fe [110, 129, 228]. The speciation of Fe in sedimenting material from LM and LT issimilar, with both shallow and deep sediment traps revealing that most of the Fe in these particlesis in the reactive, FeHR, form (Fig. 5.2 and Table C.1, see Supplementary Information). TheFeHR pools are dominated by FeAca (operationally defined as siderite), Fe(III)HCl (operationallydefined as ferrihydrite) and FeDith (operationally defined as goethite) phases. In LM, these threepools together account for 83% and 70% of the total particulate Fe in the shallow and deeptraps respectively, whereas in LT they account for 79% and 77% (Fig 5.2, Table C.1). These poolscomprise Fe-minerals that are thermodynamically favourable substrates for microbial Fe-reductionunder anaerobic standard-state conditions [109]. The strong flux of FeHR thus provides an amplepotential source of Fe(II) that can accumulate under anoxic conditions.The Fe(III)HCl pool, traditionally thought to represent the form most easily accessed formicrobial respiration Lovley1986, delivered to the lakes anoxic waters is entirely reduced directlywithin the water column and is thus quantitatively converted to dissolved Fe(II) and Fe(II) bearingmineral phases (Table 5.1, Fig. 5.2). Comparing the speciation of Fe captured in the upper sedimenttraps and buried in the uppermost sediments reveals that reduction of a considerable fractionof the Fe(III)HCl pool takes place in the water column (Table 5.1, Fig. 5.2). In LT consumption109of the Fe(III)HCl pool occurs entirely between the bottom sediment trap and the sediment waterinterface, rendering the underlying sediments devoid of Fe(III)HCl. Reduction of Fe(III) in thewater column is also commensurate with increases in the proportion of Fe(II) bearing phases,magnetite and FeAca, which in the deep traps increases by over 27% in LM and 44% in LT, relativeto the upper sediment traps (Fig. 5.2, Table 5.1 and C.1). Due to Fe-reduction and the formationof Fe(II)-bearing phases – like magnetite – directly in the water column, the deep sedimentsof LM and LT receive negligible Fe(III)HCl. The conclusions drawn from these results, and inparticular that Fe(III)HCl is converted to magnetite (Fig. 5.2), however, hinges on the selectivity ofthe extractions and their ability to the target specific mineral phases. Notably, the oxalate leach isknown to dissolve non-magnetite phases in some environments [231] and so we explore oxalateselectivity below.The oxalate extraction is highly selective for magnetite in sediment trap material and depositedsediments from LM and LT, confirming that our Fe-speciation results accurately quantify magnetite.We directly tested the selectivity of the oxalate extraction by first leaching water column particlesand sediments from LM and LT with dithionite to dissolve Fe (oxyhydr)oxides (see SupplementaryInformation), leaving magnetite and residual unreactive Fe-silicate minerals [129]. After leachingthese particulates with dithionite, we physically separated magnetite grains from the residualmaterial using neodymium magnets (FeMagnetic, Fig. 5.2), and analyses of the separated grainsconfirmed their identity as magnetite (see section below and Supplementary Information). Wealso leached the residual non-magnetic material (FeNonMagnetic, Fig. 5.2) with oxalate, recovering< 0.1% (< 0.8% from deposited sediments) of the total Fe (FeNonMagneticOxalateExtractable, Fig. 5.2),or less than 1% of the total oxalate extractable Fe. This demonstrates that more than 99% of theFe extracted with oxalate can be attributed to magnetite and less than 1% to the non-selectiveextraction of other phases. We thus conclude that, while operationally defined, the oxalateextraction applied here is highly selective for magnetite and yields precise and quantitativeinformation on its abundance in sediment trap material and deposited sediments.We also confirmed the presence of magnetite in the water columns and sediments of LMand LT using microscopic techniques (see Supplementary Information) and these analyses revealdifferent forms of magnetite that can be ascribed to authigenic and detrital origins. For example,110in the deep sediment traps we observe magnetite framboids, which are clearly authigenic, andirregularly shaped partly dissolved grains of likely detrital provenance (Fig. 5.3 and Fig. C.6).5 uma10 umb1.5 umd1 um111110e2 umfc4 um1 umg h0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0Energy (KeV)01,0002,0003,0004,0005,000IntensityiFeFeMgSiAlO Fe:O = 2.80 400 800 1,200 1,600Wave Number (cm-1)IntensityFigure 5.3: Authigenic magnetite morphologies in the Malili lakes. a) Water column magnetite framboidcaptured in the LM deep sediment trap. Orange arrow indicates EDS spot location and corresponds toorange spectra in (g). b) Water column magnetite framboid captured in the LT deep sediment trap. c) Aclose up image of the framboid surface from (a). d) Magnetite framboid preserved in the LM sediment.Green arrow indicates EDS spot location and corresponds to green spectra in (g). e) A close up imageof the framboid surface from (b), displaying immature magnetite octahedra. f) A close up image of theframboid surface from (a), displaying well-formed euhedral magnetite octahedra with identifiable crystalfaces. g) SEM-EDS spectra of the framboids from (a) and (d). The orange and green arrows demarcate theEDS spot locations and corresponding spectral curves with the same colors from (a) and (d) respectively.Both framboids have Fe:O stoichiometry diagnostic of magnetite (see Supplementary Information). h)Raman spectrum of sediment framboids from LM. Solid spectra correspond to 4 different framboids. Greytriangles on the x-axis correspond to prominent spectral bands for magnetite 44. We observe diagnosticmagnetite peaks at wave numbers ∼306, ∼450 - 490, ∼538 and ∼668 (cm– 1), with second order scatteringbetween 1200 - 1400 (cm– 1). i) A close up image of the framboid surface from (c), displaying nannoscaleoctahedral magnetite crystals.Framboids are one of magnetite’s most enigmatic crystal forms and these have only been sporadi-111cally observed in ancient sedimentary rocks and meteorites [264–266]. Magnetite framboids alsoformed in response to microbial Fe reduction in laboratory experiments [267]. These framboidalforms were not observed in material recovered from upper sediment traps, but are a ubiquitouscomponent (∼5%) of the magnetite populations in LM and LT sediments (Fig. C.2 and Fig. C.4),implying they form under anoxic conditions as the likely result of microbial Fe reduction (Fig. 5.3and Fig C.4,C.5,C.6). Detailed observations of framboid surfaces reveal that they are comprisedof aggregates of nano-scale euhedral, octahedral crystals – a common crystal habit of magnetite(Fig. 5.3c, e, f). Micro chemical analyses, furthermore, reveal that these crystals have Fe:Ostoichiometries diagnostic of magnetite (Fig. 5.3g and see Supplementary Information). Ramanmicrospectroscopy yielded reflections at wavenumbers (310, 500 and 700 (cm– 1)) diagnostic formagnetite providing definitive mineralogical identification (Fig. 5.3h and see SupplementaryInformation). Although the mechanisms causing framboid formation remain largely unknown,pyrite of similar morphology is known to form at redox interfaces in other stratified water columns[268, 269]. Our observations thus imply that magnetite formation, including framboids, is theresult of Fe(III)HCl reduction, primarily in the water columns of LM and LT.5.3 Magnetite formation, Fe reduction, and Fe recyclingComparisons of magnetite fluxes recorded in the water columns and sediments reveals that mostauthigenic magnetite in LM and LT forms in the water column. By combining sedimentation ratesin both LM and LT with the Fe speciation of suspended and deposited material, we determinedwater column and sediment magnetite fluxes (Fig. 5.2, Table 5.1 and Table C.2). Magnetite fluxesrecorded in the lower sediment traps are 79% and 55% greater than in the upper sediment trapsfrom LM and LT, respectively (Fig. 5.2 and Table 5.1). This demonstrates magnetite formation inthe water column, at depths between the two sediment traps. In LT magnetite fluxes recorded inthe upper sediments are also greater than in lower sediment traps (Fig 5.2, Table 5.1), and thisindicates further magnetite formation, either in the deep waters or the very uppermost sediments.Throughout the upper 30 cm of the LT sediments, however, the flux of magnetite is roughlyconstant (Fig. 5.2e) revealing a lack of net magnetite formation or dissolution and implying112that the primary source of authigenic magnetite is the water column. In LM, we were unable tocalculate sediment magnetite fluxes, as the lake’s steep bathymetry effectively focuses detrital Fe(6 N HCl extractable FeSil) to the deeper sediments, diluting authigenic Fe-mineral phases (seeSupplementary Information). Our results thus reveal that water column magnetite formation isthe principal source of magnetite to the underlying sediments in LT.Magnetite formation in LM and LT is associated with Fe reduction in the water column anduppermost sediments. Increases in the fluxes of magnetite below the chemoclines of LM andLT are commensurate with the conversion of Fe(III)HR to Fe(II)HR, which indicates Fe reductionwithin this depth interval. Between the upper and lower sediment traps, magnetite formationaccounts for 16% and 24% of this Fe reduction, respectively (Table 5.1). In LT, there is continuedconversion of Fe(III)HR to Fe(II)HR in the anoxic waters between the lower sediment trap and uppermost sediments and magnetite formation accounts for 31% of this (Table 5.1). Pelagic magnetitetherefore represents a substantial sink for Fe(II) resulting from the reduction of Fe(III)HR speciesdirectly in the water columns of LM and LT.The Fe(III)HCl delivered to the upper sediment trap of LT is entirely converted to magnetiteand Fe(II)Aca in the water column and this implies that an appreciable fraction of the Fe(III)HCldelivered to the lake is preserved as magnetite in the sediments. The flux of Fe(III)HCl deliveredto the upper sediment trap is 73 mmol m2 yr– 1, and this compares with magnetite and Fe(II)Acafluxes of 29 and 52 mmol m2 yr– 1 in the upper most sediments, respectively (Table 5.1). Theabsence of Fe(III)HCl in these uppermost sediments demonstrates that it is quantitatively reducedin the water column and uppermost sediments. FeDith fluxes are the same in the upper sedimenttrap and uppermost sediment revealing a lack of conversion of Fe(III)HCl to Fe(III)Dith. Massbalance thus requires that the consumption of Fe(III)HCl in the water column is entirely tied to theproduction of magnetite (36%) and FeAca (64%). Given that 66% of Fe in magnetite is Fe(III), thisindicates that 23% of the Fe(III) delivered to the upper sediment trap is ultimately preserved inmagnetite.The accumulation of dissolved Fe(II) in the water columns of LM and LT implies Fe recyclingwith potential to support microbial metabolisms. In LM and LT we compared rates of diffusiveFe(II) supply to the fluxes of Fe(III)HCl delivered to the upper sediment traps. In LT the diffusional113flux of Fe(II) is 130 mmol m2 yr– 1 (see Supplementary Information) which, when divided by thedelivery flux of Fe(III)HCl implies recycling of not more than 2 times. In LM the diffusional flux ofFe(II) is 55 mmol m2 yr– 1 (see Supplementary Information) which, when divided by the deliveryflux of Fe(III)HCl implies recycling of not more than 3 times. This limited recycling reveals thatFe dependant microbial metabolisms are restricted by the both the delivery flux of Fe(III)HCl andits conversion to magnetite and FeAca, which are ultimately removed through sedimentation andburial.5.4 ImplicationsOur findings from LM and LT imply that pelagic magnetite formation can be a important modeof Fe delivery to ferruginous sediments and by extension suggest that magnetite derived fromthe Precambrian oceanic water column may have been a primary contributor to IF deposition.Magnetite formation in lakes LM and LT is linked to water column reduction of Fe(III)HCl as thelikely source of Fe(II) in magnetite, and a similar process could thus be envisioned for magnetiteformation in the Precambrian oceans. Magnetite formation is undoubtedly influenced by waterchemistry, and in LM and LT it forms at circumneutral pH, at < 1 to 10s of µM Fe(II), andat mM concentrations of dissolved inorganic carbon (DIC) (Fig. 5.1 and [137]). Notably, theseFe(II) concentrations are far lower than those thought required to induce magnetite formationin laboratory experiments [121, 122, 267]. Conditions similar to LM and LT, however, couldbe expected to support primary magnetite formation in the Precambrian oceans and, indeed,such conditions are in line with current reconstructions of the Archean and Proterozoic seawaterchemistry [262]. Notably, in LM and LT these conditions also support the formation and depositionof an appreciable, though poorly defined FeAca phase, which may, in addition to magnetite, alsohave been an important contributor to IFs.Magnetite formed in the water column of Precambrian oceans would have directly recordedprimary information on the chemistry and ecology of ancient seawater. Given that magnetiteformation in response to Fe reduction depends on concentrations of DIC [270, 271], the presenceof primary magnetite, or its metamorphic products, in IFs would place constraints on Precambrian114ocean pCO2 and pH. For example, the formation of primary water column magnetite impliesthat magnetite in IFs may have formed in thermodynamic equilibrium with the atmosphere [270],placing constraints on Precambrian partial pressures of both CO2 and H2. Although this model isintriguing, magnetite may also form diagenetically; far from equilibrium with the ocean [271]. Theprecise conditions that promote or inhibit primary magnetite formation thus need to be workedout through further research.Our observation of pelagic magnetite formation, furthermore, also supports the idea that theFe isotopic composition of IF magnetite captures signatures of water column Fe drawdown inresponse to Fe oxidation. Such water column Fe drawdown leaves residual Fe(II) isotopicallylight, and minerals that sequester Fe(II) in such a gradient record this light signal [257, 272, 273].Isotopically light magnetites were deposited in IFs from the Meso- to Neoarchean and have beenvariably interpreted to reflect either water column Fe(II) drawdown [135, 273, 274] or sedimentarydiagenetic Fe(III) reduction [257]. Our findings from LM and LT imply both that Fe reduction ismostly pelagic and that magnetites form from Fe reduction in the water column, suggesting thatthe former pathway likely contributed to the Fe isotope composition of IFs. This could be furthertested through Fe isotope studies in LM and LT.The role of magnetite burial in setting Earth’s surface redox budgets combined with ourobservation of water column magnetite formation in LM and LT implicates coupled C and Fecycling in ferruginous Precambrian water columns in the protracted oxidation of Earth’s surface.The extent to which C and Fe cycling in the water column lead to magnetite, siderite, or Fe(III)(oxyhydr)oxide export to underlying sediments sets the redox state of these sediments. Notably,the export of magnetite sequesters Fe(III) leaving excess reductant to ultimately fuel H2 production[14, 253] through biological and photochemical reactions. Loss of this hydrogen to space leadsto net oxidation of the Earth [253]. In LT, 23% of the Fe(III)HCl is converted to magnetite inassociation with microbial Fe(III) reduction. With global fluxes of up to 45 Tmol yr– 1 reactive Feto the Precambrian oceans (Thompson et al., 2019), a similar fraction of conversion to magnetite,which is consistent with typical BIF magnetite contents [275], would lead to hydrogen productionof 5.2 Tmol yr– 1, equivalent to nearly twice the modern oxidant production through organiccarbon burial in marine sediments [14]. Factors controlling magnetite formation, like water115column Fe(II) concentrations, pH, DIC concentrations, and likely rates of microbial metabolismwould thus influence H2 production rates and ultimately planetary oxidation rates.In LM and LT, primary magnetites often form in conspicuous framboidal morphologies(Fig. 5.3), which can thus be used to identify such primary magnetites in the rock record or inextraterrestrial materials. Framboids of similar size and morphology of iron’s most oxidized form,hematite (Fe2O3), have been observed in multiple Precambrian IFs [276–278], and indeed thesehematite framboids have variably been interpreted to have formed through post depositionaloxidation of precursor pyrite framboids of biogenic origins [278]. Such oxidation of pyrite tomagnetite, however, is accompanied by a significant molar volume reduction of 60%, renderingpreservation of the framboid microstructure unlikely [268]. Leading models of framboid formationrely on the ferromagnetic properties of the precursor Fe-phases that control framboid aggregation[268], and thus it is unlikely that the framboids observed in IFs were originally depositedas non-magnetic hematite. Given the near absence of sulfate in the ferruginous Malili lakesand Precambrian oceans, we alternatively suggest these framboidal hematite grains were likelydeposited as magnetite, and subsequently oxidized to hematite during subsequent diagenesisand or metamorphism. We also note that recrystallizing of biogenic nano-magnetites to largermagnetite grains with textures common in IFs, has been observed in laboratory experiments, evenunder low-grade pressure and temperature regimes [279]. Thus formation of magnetite framboidsin LM and LT, in association with microbial Fe reduction, and their formation as the result ofmicrobial Fe reduction in previous lab experiments [267], further implies that magnetite framboidsprovide a signature for anaerobic microbial respiration and suggest that ancient sedimentary andmeteorite magnetite framboids may be relicts of ancient microbial Fe metabolisms. Ability tovisually identify framboids and, as we show here, determine their mineralogy through Ramanmicrospectroscopy, holds considerable promise for detecting these biosignatures from ancientferruginous oceans on Earth, or even remotely on Mars and other terrestrial planets.116Chapter 6Conclusions6.1 Summary of the dissertation and key research findingsThe research reported in this dissertation has explored the geochemical behaviour of the transitionmetals Cr and Fe under a wide range of environmental conditions. This investigation has led to thedevelopment of more nuanced conceptual frameworks for how the distribution and associations ofthese metals may be used to gain further insight into the redox state of both modern and ancientlacustrine and oceanic environments. A main focus of this work has been to provide furthercalibration of the Cr isotope and Fe-speciation paleoredox proxies in modern environments, with aspecific emphasis on ferruginous depositional conditions – the inferred redox state of ocean watersover much of Earth’s history. New knowledge of Cr and Fe proxy systematics will allow for moredetailed reconstructions of depositional redox states that rely on Cr isotope and Fe-speciationsignals captured in the rock record through time.6.1.1 Cr isotopes recorded in marine hydrothermal sedimentsHydrothermally-sourced enriched metalliferous sediments deposited at the South East PacificRise are enriched in hydrothermal Fe and Mn (oxyhydr)oxides. In the vicinity of hydrothermalvents, seawater Cr(VI) is inferred to undergo reduction by Fe(II) delivered to the sea floor by ventfluids, and the Cr(III) produced is scavenged by freshly-formed Fe precipitates and depositedin the local sediments. This implies that hydrothermal systems are likely a net sink, rather thana net source of Cr to the global oceans. Partial reduction of Cr(VI) from oxygenated seawater,furthermore, imparts a light δ53Cr fingerprint to the resulting plume particles, which have δ53Crvalues lighter than the ISE range (-0.12 ± 0.10h). The abundant sedimentary Mn-oxides maysubsequently catalyze the oxidation of deposited Cr(III), inducing the release of as much as 80% of117Cr in the form of soluble Cr(VI) back to the overlying water column. This remobilization may alsocarry an isotopic fractionation, with potential to enhance the overwhelmingly light δ53Cr impartedduring partial reduction. Overall, these processes combine to set a variable, but distinctly light Crisotopic signal to hydrothermal sediments and together, these processes imply deposition fromCr(VI) bearing oxygenated deep seawater – a signal that can potentially be tracked through time.This work provides impetus to further explore ancient hydrothermal sediments as archivesof δ53Cr that may demarcate deep ocean oxygenation in the geologic record. To date, light δ53Crcompositions have only been observed as the result of oxidative remobilization of Cr(III) inmodern soils, paleosols, modern soils, and in strongly oxidizing ferromanganese nodules (e.g.,[30, 149, 162, 174]). In this thesis I show that a light δ53Cr signal characterizes hydrothermal plumefallout sediments, thereby establishing that the δ53Cr composition of modern marine hydrothermalsediments is linked to the oxygenation state of the deep ocean. I propose therefore that thisframework can be expanded through further Cr isotope determinations of older hydrothermal sed-iments. In principle this should enable reconstruction of deep ocean oxygen dynamics over a rangeof geological timescales. For example, existing δ53Cr records of Neoproterozoic hydrothermal sed-iments reveal occurrences of isotopically light δ53Cr [175], implying deposition from oxygenateddeep ocean waters at that time. Likewise, rare earth element (REE) anomalies determined inPaleoproterozoic hydrothermal deposits likely record whiffs of deep ocean oxygenation [280].Studies that couple Cr isotope and REE systematics of hydrothermal sediments may thus provideinvaluable tools for investigating the timing of deep ocean oxygenation.It is still unclear whether the oxidation of Cr(III) to Cr(VI) in the environment causes Crisotope fractionation. It is possible that the oxidative dissolution of Cr(III) bearing minerals duringreaction with Mn oxides, may favour release of the light isotope (52Cr) to solution. The preciseisotope effects, however, associated with Cr(III) oxidation during reaction with Mn oxides, arepoorly constrained and highly dependent on specific physicochemical conditions. I speculatein this work that diagenetic oxidation of Cr(III) by Mn-oxides in hydrothermal sediments mayremobilize an appreciable amount of authigenic Cr from well-oxygenated deep sea sediments.Clearly, additional work on the mechanisms controlling Cr(III) oxidation via Mn-oxides acrossa broad range of conditions needs to be conducted to resolve these issues. I propose that this118can be addressed through laboratory experiments that track the isotopic composition of both theCr(VI) and residual Cr(III) as a function of time, during kinetic reactions with Mn-oxides. Theseexperiments would benefit from a systematic exploration of these processes over a wide rangeof oxidant species (different Mn-oxide minerals) and reductant species (different Cr(III) bearingminerals).6.1.2 Cr isotope fractionation in modern and ancient ferruginous environmentsThe fractionation of Cr isotopes during redox reactions can involve a plethora of electron donors,and the magnitude of isotopic fractionation associated with these reactions can be large andvariable [31]. To place constraints on isotope effects associated with the reduction of Cr(VI) duringreaction with different Fe(II) species, I conducted experiments using sediments from Lake Matano,Indonesia, a modern ferruginous envrionment. Whole core incubations performed on ferruginoussediments demonstrate that Cr(VI) reduction induces Cr(VI) concentration gradients that drivediffusive Cr fluxes across the sediment water interface. As reduction of Cr(VI) progressed, theremaining lake water Cr(VI) became increasingly enriched in the heavy isotope (53Cr). Detaileddiagenetic modelling that accounts for open system behaviour predicts an intrinsic isotopefractionation eint = 1.80 ± 0.04h, implying that appreciable reservoir effects develop duringCr(VI) reduction in sediments. Based on the magnitude of this fractionation factor and previousexperimental work, I suggest the Cr(VI) reductant in Lake Matano sediments is a combinationof dissolved and solid phase Fe(II). My results show that the isotopic composition of authigenicCr in sediments can be offset from the δ53Cr values in overlying water when the zone of Cr(VI)reduction is sufficiently close to the sediment water interface resulting in an intrinsic fractionationthat is expressed with minimal reservoir effects. This work demonstrates how the Cr isotopevalues of marine sediments from the geologic record may be used to estimate sediment oxygenpenetration depths, which are linked to the oxygenation of the ocean-atmosphere system.Previous workers have attempted to elucidate links between the evolution of eukaryotic lifeand the oxygenation state of the mid-Proterozoic oceans using Cr mass balance models informedby sedimentary Cr enrichments [25]. These quantitative models, however, lack parameterizationof the isotopic composition of the different Cr input and burial fluxes to and from the ocean119respectively, and in addition, they fail to include an approach for capturing authigenic Cr burialrates in different regions of the ocean. My work provides a new framework for evaluating themagnitude of sedimentary Cr enrichment and isotopic composition, which is directly linked tothe oxygenation state of the sediment water interface. I propose therefore, that use of my model;in combination with measurements of the Cr concentration and isotopic composition of sedimentsin the geologic record will facilitate more detailed reconstructions of the oxygenation state ofseawater in the past. This mass balance approach would be even more powerful if the redox stateof depositional environment is independently constrained by the use of the other proxies (forexample Fe-speciation). In follow-on work I will couple my diagenetic model to an ocean Crmass balance model that includes the isotopic composition of the relevant input fluxes. I will usethese models to interpret δ53Cr signals preserved in marine sediments that accumulated duringintervals when the redox state of seawater fluctuated dramatically (for example during oceanicanoxic events (OAEs), such as during the Paleocene-Eocene Thermal Maximum, and duringmodern expansions of oxygen minimum zones). This approach will help to place constraints onthe drivers and relative timing of large-scale Earth system perturbations that resulted in changesto the redox state of the sediment water interface in different regions of the oceans.6.1.3 Dynamics in Earth surface oxidant budgets during OAE1aDevelopment of ferruginous conditions during OAE1a reveals large-scale dynamics in Earth’sbiogeochemical cycles over intervals of just a few hundred thousand years. I propose thatthe development of ferruginous ocean conditions during multiple OAEs [208, 210] may thussignify a general instability in Earth surface redox budgets and the recurrent reorganization ofmajor oxidant pools at Earth’s surface, such as seawater sulphate, during the Phanerozoic Eon.Indeed, there is emerging evidence that dynamics in Earth surface oxidant budgets develop overtimescales that are much more rapid than previously thought [281], and this general instability ofEarth’s redox state may have been especially pronounced during last 540 Myr of Earth history.For example, sulphur isotope signals recorded in barites deposited across the Paleocene-EoceneThermal Maximum (PETM), reveal significant ocean de-oxygenation and development of sulphidicocean waters on timescales relevant to humankind (< 10 kyr)[281]. To expand our emerging120view of how the Earth system responds during intervals of biogeochemical upheaval, we requirehigher-resolution records than those currently available. I therefore propose, that future workin this realm must include a focus on conducting studies at a fine-scale sample resolution. Forexample, sedimentary profiles that document geochemical change on timescales < 10 kyrs, areurgently needed. To fully constrain dynamics in the seawater sulphate reservoir specifically,furthermore, determinations of δ34Spyrite must be combined with δ34S data that are a proxy for theδ34S composition of seawater, such as sedimentary barite. Further examination of high-resolution∆34Sseawater – pyrite will place better constraints on global sulphur budgets and the drivers of oceande-oxygenation.In this work I developed a marine sulphur mass balance model to explore oceanic sulphurbudgets during OAE1a. This exploration into S biogeochemistry during OAE1a led me tohypothesize that biomass associated organic S was a major pathway for S burial during OAEs.To further constrain the potentially important yet previously overlooked role of organic S burialduring intervals of ocean anoxia, I propose that future studies should aim to produce high-resolution measurements of the S content and isotopic composition of organic matter buried inmarine sediments throughout the Phanerozoic. This will yield will critical data needed to refinemarine S-budgets during some of the longest or largest perturbations to the Earth system to haveoccurred in the last 540 Myr.6.1.4 Magnetite formation in modern and ancient ferruginous environmentsBy combining Fe-speciation experiments, imaging techniques and conceptual models, I demon-strate that water column formation of magnetite is an important feature of the Fe-cycle operatingunder ferruginous conditions. This implies that magnetite contained in Iron Formations (IFs)may not only record primary information on the geochemistry of Precambrian seawater but mayalso represent a previously unrecognized shuttle of both Fe(II) and poorly biologically reactiveFe(III) to the sediment water interface under ferruginous conditions. The direct precipitation ofreduced Fe-minerals in the water column has important implications for coupled cycling of Feand C, the evolution of Earth’s redox budget, and the protracted oxidation of Earth’s surface envi-ronments. These findings improve the current framework for IF genesis and must be considered121in interpretations of the mineralogy and isotopic composition of IFs and Earth’s early Fe-cycle.One exciting finding of this work includes the discovery of framboidal magnetite – thefirst occurrence of this enigmatic crystal form documented in a modern ferruginous setting.Examples of framboidal magnetite have also been sporadically identified in extraterrestrialsamples [265, 266]. Identification of framboidal magnetites whose genesis may be linked toFe-reduction, therefore, may represent an important biosignature in the search for extraterrestriallife on other Fe-rich planetary bodies (for example Mars). I propose that future studies shouldaim to place constraints on the precise environmental conditions that favor microbial Fe-reductionleading to the precipitation of framboidal magnetite. Framboidal magnetite furthermore, is easilyidentified via Raman microspectroscopy. Significant progress in understanding the compositionand physical processes associated with the surfaces of planetary bodies has been made in recentdecades, and this is due in large part to the increasing sophistication of orbiters, landers androvers that can now be equipped with advanced analytical instrumentation [282]. It is believedthat the optimization of Raman microspectrometers for rover based in-situ analysis of planetarymaterials, will be completed within the next decade [282]. I propose therefore, that Ramanspectroscopy provides an exciting and accurate approach for identifying biogenic Fe-minerals,and this instrumentation should be at the forefront of future discussions involving the potentialcapabilities of rover based space missions with the goal of fingerprinting signs of extraterrestriallife.122Bibliography[1] K. Scheiderich, M. Amini, C. Holmden, and R. Francois. Global variability of chromium isotopesin seawater demonstrated by pacific, atlantic, and arctic ocean samples. Earth and Planetary ScienceLetters, 423:87–97, 2015.[2] L. R. Kump. The rise of atmospheric oxygen. Nature, 451(7176):277–278, 2008.[3] P. N. Froelich, G. P. Klinkhammer, M. L. Bender, et al. Early oxidation of organic matter in pelagicsediments of the eastern equatorial atlantic: suboxic diagenesis. Geochimica Et Cosmochimica Acta,43(7):1075–1090, 1979.[4] D. B. Cole, C. T. Reinhard, X. L. Wang, et al. A shale-hosted cr isotope record of low atmosphericoxygen during the proterozoic. Geology, 44(7):555–558, 2016.[5] Heinrich D Holland. The chemical evolution of the atmosphere and oceans. Princeton University Press,1984.[6] T. W. Lyons, C. T. Reinhard, and N. J. Planavsky. The rise of oxygen in earth’s early ocean andatmosphere. Nature, 506(7488):307–315, 2014.[7] Noah J. Planavsky, Christopher T. Reinhard, Xiangli Wang, et al. Low mid-proterozoic atmosphericoxygen levels and the delayed rise of animals. Science, 346(6209):635–638, 2014.[8] R. A. Berner. The long-term carbon cycle, fossil fuels and atmospheric composition. Nature,426(6964):323–326, 2003.[9] L. R. Kump. Alternative modeling approaches to the geochemical cycles of carbon, sulfur, andstrontium isotopes. American Journal of Science, 289(4), 1989.[10] L. R. Kump and R. M. Garrels. Modeling atmospheric o2 in the global sedimentary redox cycle.American Journal of Science, 286(5):337–360, 1986.[11] L. R. Kump and M. A. Arthur. Interpreting carbon-isotope excursions: carbonates and organic matter.Chemical Geology, 161(1-3), 1999.[12] L. R. Kump, S. L. Brantley, and M. A. Arthur. Chemical, weathering, atmospheric co2, and climate.Annual Review of Earth and Planetary Sciences, 28:611–667, 2000.[13] J. Farquhar, H. M. Bao, and M. Thiemens. Atmospheric influence of earth’s earliest sulfur cycle.Science, 289(5480):756–758, 2000.[14] H. D. Holland. Volcanic gases, black smokers, and the great oxidation event. Geochimica Et Cos-mochimica Acta, 66(21):3811–3826, 2002.[15] H. D. Holland. The oxygenation of the atmosphere and oceans. Philosophical Transactions of the RoyalSociety B-Biological Sciences, 361(1470):903–915, 2006.[16] Donald E. Canfield, Erik Kristensen, and Bo Thamdrup. Aquatic geomicrobiology. Advances in marinebiology, 48:1–599, 2005.123[17] J. Farquhar and B. A. Wing. Multiple sulfur isotopes and the evolution of the atmosphere. Earth andPlanetary Science Letters, 213(1-2):1–13, 2003.[18] L. M. Och and G. A. Shields-Zhou. The neoproterozoic oxygenation event: Environmental perturba-tions and biogeochemical cycling. Earth-Science Reviews, 110(1-4):26–57, 2012.[19] G. A. Shields and B. J. W. Mills. Tectonic controls on the long-term carbon isotope mass balance.Proceedings of the National Academy of Sciences of the United States of America, 114(17):4318–4323, 2017.[20] D. H. Erwin, M. Laflamme, S. M. Tweedt, et al. The cambrian conundrum: Early divergence and laterecological success in the early history of animals. Science, 334(6059):1091–1097, 2011.[21] D. H. Rothman, J. M. Hayes, and R. E. Summons. Dynamics of the neoproterozoic carbon cycle.Proceedings of the National Academy of Sciences of the United States of America, 100(14):8124–8129, 2003.[22] B. Rasmussen, I. R. Fletcher, J. J. Brocks, and M. R. Kilburn. Reassessing the first appearance ofeukaryotes and cyanobacteria. Nature, 455(7216):1101–U9, 2008.[23] Erik A. Sperling, Charles J. Wolock, Alex S. Morgan, et al. Statistical analysis of iron geochemicaldata suggests limited late proterozoic oxygenation. Nature, 523(7561):451–454, 2015.[24] S. C. Zhang, X. M. Wang, H. J. Wang, et al. Sufficient oxygen for animal respiration 1,400 million yearsago. Proceedings of the National Academy of Sciences of the United States of America, 113(7):1731–1736,2016.[25] Christopher T. Reinhard, Noah J. Planavsky, Leslie J. Robbins, et al. Proterozoic ocean redox andbiogeochemical stasis. Proceedings of the National Academy of Sciences of the United States of America,110(14):5357–5362, 2013.[26] Daniel B. Mills, Lewis M. Ward, CarriAyne Jones, et al. Oxygen requirements of the earliest animals.Proceedings of the National Academy of Sciences of the United States of America, 111(11):4168–4172, 2014.[27] Erik A. Sperling, Galen P. Halverson, Andrew H. Knoll, Francis A. Macdonald, and David T. Johnston.A basin redox transect at the dawn of animal life. Earth and Planetary Science Letters, 371:143–155,2013.[28] N. J. Butterfield. Oxygen, animals and oceanic ventilation: an alternative view. Geobiology, 7(1):1–7,2009.[29] A. D. Anbar, Y. Duan, T. W. Lyons, et al. A whiff of oxygen before the great oxidation event? Science,317(5846):1903–1906, 2007.[30] Sean A. Crowe, Lasse N. Dossing, Nicolas J. Beukes, et al. Atmospheric oxygenation three billionyears ago. Nature, 501(7468):535–+, 2013.[31] Robert Frei, Claudio Gaucher, Simon W Poulton, and Don E Canfield. Fluctuations in precambrianatmospheric oxygenation recorded by chromium isotopes. Nature, 461(7261):250, 2009.[32] N. M. Bergman, T. M. Lenton, and A. J. Watson. Copse: A new model of biogeochemical cycling overphanerozoic time. American Journal of Science, 304(5):397–437, 2004.[33] R. A. Berner and D. E. Canfield. A new model for atmospheric oxygen over phanerozoic time.American Journal of Science, 289(4):333–361, 1989.[34] S. E. Calvert and T. F. Pedersen. Geochemistry of recent oxic and anoxic marine sediments: implica-tions for the geological record. Marine Geology, 113(1-2):67–88, 1993.124[35] D. B. Cole, S. Zhang, and N. J. Planavsky. A new estimate of detrital redox-sensitive metal concentra-tions and variability in fluxes to marine sediments. Geochimica Et Cosmochimica Acta, 215:337–353,2017.[36] Nicolas Tribovillard, Thomas J. Algeo, Timothy Lyons, and Armelle Riboulleau. Trace metals aspaleoredox and paleoproductivity proxies: An update. Chemical Geology, 232(1-2):12–32, 2006.[37] A. S. Ellis, T. M. Johnson, and T. D. Bullen. Chromium isotopes and the fate of hexavalent chromiumin the environment. Science, 295(5562):2060–2062, 2002.[38] A. S. Ellis, T. M. Johnson, and T. D. Bullen. Using chromium stable isotope ratios to quantify cr(vi)reduction: Lack of sorption effects. Environmental Science & Technology, 38(13):3604–3607, 2004.[39] R. Frei, C. Gaucher, L. N. Dossing, and A. N. Sial. Chromium isotopes in carbonates - a tracer forclimate change and for reconstructing the redox state of ancient seawater. Earth and Planetary ScienceLetters, 312(1-2):114–125, 2011.[40] G. J. Gilleaudeau, R. Frei, A. J. Kaufman, et al. Oxygenation of the mid-proterozoic atmosphere: cluesfrom chromium isotopes in carbonates. Geochemical Perspectives Letters, 2(2):178–+, 2016.[41] P. Bonnand, R. H. James, I. J. Parkinson, D. P. Connelly, and I. J. Fairchild. The chromium isotopiccomposition of seawater and marine carbonates. Earth and Planetary Science Letters, 382:10–20, 2013.[42] Robert Frei and Ali Polat. Chromium isotope fractionation during oxidative weathering-implicationsfrom the study of a paleoproterozoic (ca. 1.9 ga) paleosol, schreiber beach, ontario, canada. PrecambrianResearch, 224:434–453, 2013.[43] Amanda L. Raddatz, Thomas M. Johnson, and Travis L. McLing. Cr stable isotopes in snake riverplain aquifer groundwater: Evidence for natural reduction of dissolved cr(vi). Environmental Science& Technology, 45(2):502–507, 2011.[44] J. D’Arcy, M. G. Babechuk, L. N. Dossing, C. Gaucher, and R. Frei. Processes controlling thechromium isotopic composition of river water: Constraints from basaltic river catchments. GeochimicaEt Cosmochimica Acta, 186:296–315, 2016.[45] W. H. Wu, X. L. Wang, C. T. Reinhard, and N. J. Planaysky. Chromium isotope systematics in theconnecticut river. Chemical Geology, 456:98–111, 2017.[46] R. Schoenberg, S. Zink, M. Staubwasser, and F. von Blanckenburg. The stable cr isotope inventoryof solid earth reservoirs determined by double spike mc-icp-ms. Chemical Geology, 249(3-4):294–306,2008.[47] S. E. Fendorf. Surface reactions of chromium in soils and waters. Geoderma, 67(1-2):55–71, 1995.[48] B. R. James, J. C. Petura, R. J. Vitale, and G. R. Mussoline. Hexavalent chromium extraction fromsoils: a comparison of five methods. Environmental Science & Technology, 29(9):2377–2381, 1995.[49] F. C. Richard and A. C. M. Bourg. Aqueous geochemistry of chromium: a review. Water Research,25(7):807–816, 1991.[50] C. Oze, S. Fendorf, D. K. Bird, and R. G. Coleman. Chromium geochemistry in serpentinizedultramafic rocks and serpentine soils from the franciscan complex of california. American Journal ofScience, 304(1):67–101, 2004.[51] D. Rai, L. E. Eary, and J. M. Zachara. Environmental chemistry of chromium. Science of the TotalEnvironment, 86(1-2):15–23, 1989.125[52] R. Rakhunde, L. Deshpande, and H. D. Juneja. Chemical speciation of chromium in water: A review.Critical Reviews in Environmental Science and Technology, 42(7):776–810, 2012.[53] C. D. Palmer and P. R. Wittbrodt. Processes affecting the remediation of chromium-contaminatedsites. Environmental Health Perspectives, 92:25–40, 1991.[54] D. Rai, B. M. Sass, and D. A. Moore. Chromium (iii) hydrolysis constants and solubility of chromium(iii) hydroxide. Inorganic Chemistry, 26(3):345–349, 1987.[55] M Pettine, Frank J Millero, and T La Noce. Chromium (iii) interactions in seawater through itsoxidation kinetics. Marine Chemistry, 34(1-2):29–46, 1991.[56] M. Pettine, L. D’Ottone, L. Campanella, F. J. Millero, and R. Passino. The reduction of chromium (vi)by iron (ii) in aqueous solutions. Geochimica Et Cosmochimica Acta, 62(9):1509–1519, 1998.[57] M Pettine and Frank J Millero. Chromium speciation in seawater: The probable role of hydrogenperoxide. Limnology and Oceanography, 35(3):730–736, 1990.[58] I. J. Buerge and S. J. Hug. Kinetics and ph dependence of chromium(vi) reduction by iron(ii).Environmental Science & Technology, 31(5):1426–1432, 1997.[59] Ignaz J Buerge and Stephan J Hug. Influence of mineral surfaces on chromium (vi) reduction by iron(ii). Environmental Science & Technology, 33(23):4285–4291, 1999.[60] H. Zhang and R. J. Bartlett. Light-induced oxidation of aqueous chromium(iii) in the presence ofiron(iii). Environmental Science & Technology, 33(4):588–594, 1999.[61] D. C. Schroeder and G. F. Lee. Potential transformations of chromium in natural waters. Water Airand Soil Pollution, 4(3-4):355–365, 1975.[62] S. Fendorf, B. W. Wielinga, and C. M. Hansel. Chromium transformations in natural environments:The role of biological and abiological. processes in chromium(vi) reduction. International GeologyReview, 42(8):691–701, 2000.[63] C. Oze, N. H. Sleep, R. G. Coleman, and S. Fendorf. Anoxic oxidation of chromium. Geology,44(7):543–546, 2016.[64] Gautier Landrot, Matthew Ginder-Vogel, Kenneth Livi, Jeffrey P Fitts, and Donald L Sparks.Chromium (iii) oxidation by three poorly-crystalline manganese (iv) oxides. 1. chromium (iii)-oxidizing capacity. Environmental science & technology, 46(21):11594–11600, 2012.[65] L. E. Eary and D. Rai. Kinetics of chromium(iii) oxidation to chromium(vi) by reaction withmanganese dioxide. Environmental Science & Technology, 21(12):1187–1193, 1987.[66] A. Manceau and L. Charlet. X-ray absorption spectroscopic study of the sorption of cr(iii) at theoxide-water interface: I. molecular mechanism of cr(iii) oxidation on mn oxides. Journal of Colloid andInterface Science, 148(2):425–442, 1992.[67] S. E. Fendorf and R. J. Zasoski. Chromium (iii) oxidation by. delta.-manganese oxide (mno2). 1.characterization. Environmental Science & Technology, 26(1):79–85, 1992.[68] K. E. Abusaba and A. R. Flegal. Chromium in san francisco bay: superposition of geochemicalprocesses causes complex spatial distributions of redox species. Marine Chemistry, 49(2-3):189–199,1995.[69] Y. L. Li, K. O. Konhauser, and M. G. Zhai. The formation of magnetite in the early archean oceans.Earth and Planetary Science Letters, 466:103–114, 2017.126[70] E. Schauble, G. R. Rossman, and H. P. Taylor. Theoretical estimates of equilibrium chromium-isotopefractionations. Chemical Geology, 205(1-2):99–114, 2004.[71] Xiangli Wang, Thomas M. Johnson, and Andre S. Ellis. Equilibrium isotopic fractionation and isotopicexchange kinetics between cr(iii) and cr(vi). Geochimica Et Cosmochimica Acta, 153:72–90, 2015.[72] Christopher Oze, Dennis K. Bird, and Scott Fendorf. Genesis of hexavalent chromium from naturalsources in soil and groundwater. Proceedings of the National Academy of Sciences of the United States ofAmerica, 104(16):6544–6549, 2007.[73] Sonja Zink, Ronny Schoenberg, and Michael Staubwasser. Isotopic fractionation and reaction kineticsbetween cr(iii) and cr(vi) in aqueous media. Geochimica Et Cosmochimica Acta, 74(20):5729–5745, 2010.[74] Juraj Farkas, Vladislav Chrastny, Martin Novak, et al. Chromium isotope variations (delta cr-53/52)in mantle-derived sources and their weathering products: Implications for environmental studies andthe evolution of delta cr-53/52 in the earth’s mantle over geologic time. Geochimica Et CosmochimicaActa, 123:74–92, 2013.[75] J. Shen, J. Liu, L. P. Qin, et al. Chromium isotope signature during continental crust subductionrecorded in metamorphic rocks. Geochemistry Geophysics Geosystems, 16(11):3840–3854, 2015.[76] C. Jeandel and J. F. Minster. Isotope dilution measurement of inorganic chromium(iii) and totalchromium in seawater. Marine Chemistry, 14(4):347–364, 1984.[77] C. N. McClain and K. Maher. Chromium fluxes and speciation in ultramafic catchments and globalrivers. Chemical Geology, 426:135–157, 2016.[78] Jerome O Nriagu and Evert Nieboer. Chromium in the natural and human environments, volume 20.John Wiley & Sons, 1988.[79] RE Cranston and JW Murray. The determination of chromium species in natural waters. AnalyticaChimica Acta, 99(2):275–282, 1978.[80] RE Mesmer and CF Baes. The hydrolysis of cations. Ed. W i 1 ey, EUA, 1976.[81] R. E. Cranston and J. W. Murray. Chromium species in the columbia river and estuary. Limnology andOceanography, 25(6):1104–1112, 1980.[82] David Blowes. Tracking hexavalent cr in groundwater. Science, 295(5562):2024–2025, 2002.[83] John A Campbell and Philip A Yeats. Dissolved chromium in the st. lawrence estuary. Estuarine,Coastal and Shelf Science, 19(5):513–522, 1984.[84] John A. Izbicki, James W. Ball, Thomas D. Bullen, and Stephen J. Sutley. Chromium, chromiumisotopes and selected trace elements, western mojave desert, usa. Applied Geochemistry, 23(5):1325–1352,2008.[85] B. R. James and R. J. Bartlett. Behavior of chromium in soils. vi. interactions between oxidation-reduction and organic complexation 1. Journal of Environmental Quality, 12(2):173–176, 1983.[86] E. M. Saad, X. L. Wang, N. J. Planavsky, C. T. Reinhard, and Y. Z. Tang. Redox-independent chromiumisotope fractionation induced by ligand-promoted dissolution. Nature Communications, 8, 2017.[87] R. J. Bartlett and J. M. Kimble. Behavior of chromium in soils: I. trivalent forms. Journal ofEnvironmental Quality, 5(4):379–383, 1976.[88] R. J. Bartlett and J. M. Kimble. Behavior of chromium in soils: Ii. hexavalent forms. Journal ofEnvironmental Quality, 5(4):383–386, 1976.127[89] J. Dymond and W. Eklund. A microprobe study of metalliferous sediment components. Earth andPlanetary Science Letters, 40(2):243–251, 1978.[90] J. Dymond and S. Roth. Plume dispersed hydrothermal particles: A time-series record of settling fluxfrom the endeavour ridge using moored sensors. Geochimica Et Cosmochimica Acta, 52(10):2525–2536,1988.[91] R. A. Feely, E. T. Baker, K. Marumo, et al. Hydrothermal plume particles and dissolved phosphate overthe superfast-spreading southern east pacific rise. Geochimica Et Cosmochimica Acta, 60(13):2297–2323,1996.[92] R. A. Feely, T. L. Geiselman, E. T. Baker, G. J. Massoth, and S. R. Hammond. Distribution andcomposition of hydrothermal plume particles from the ashes vent field at axial volcano, juan de fucaridge. Journal of Geophysical Research-Solid Earth and Planets, 95(B8):12855–12873, 1990.[93] R. A. Feely, J. F. Gendron, E. T. Baker, and G. T. Lebon. Hydrothermal plumes along the eastpacific rise, 8 40 to 11 50 n: Particle distribution and composition. Earth and Planetary Science Letters,128(1-2):19–36, 1994.[94] M. D. Rudnicki and H. Elderfield. A chemical model of the buoyant and neutrally buoyant plumeabove the tag vent field, 26 degrees n, mid-atlantic ridge. Geochimica Et Cosmochimica Acta, 57(13):2939–2957, 1993.[95] R. P. Trocine and J. H. Trefry. Distribution and chemistry of suspended particles from an activehydrothermal vent site on the mid-atlantic ridge at 26n. Earth and Planetary Science Letters, 88(1-2):1–15,1988.[96] Joseph A. Resing, Peter N. Sedwick, Christopher R. German, et al. Basin-scale transport of hydrother-mal dissolved metals across the south pacific ocean. Nature, 523(7559):200–U140, 2015.[97] Kurt O. Konhauser, Stefan V. Lalonde, Noah J. Planavsky, et al. Aerobic bacterial pyrite oxidation andacid rock drainage during the great oxidation event. Nature, 478(7369):369–+, 2011.[98] Douglas P. Connelly, Peter J. Statham, and Anthony H. Knap. Seasonal changes in speciation ofdissolved chromium in the surface sargasso sea. Deep-Sea Research Part I-Oceanographic Research Papers,53(12):1975–1988, 2006.[99] S. Emerson, R. E. Cranston, and P. S. Liss. Redox species in a reducing fjord: equilibrium and kineticconsiderations. Deep-Sea Research Part a-Oceanographic Research Papers, 26(8):859–878, 1979.[100] S. Sander and A. Koschinsky. Onboard-ship redox speciation of chromium in diffuse hydrothermalfluids from the north fiji basin. Marine Chemistry, 71(1-2):83–102, 2000.[101] S. Sander, A. Koschinsky, and P. Halbach. Redox speciation of chromium in the oceanic water columnof the lesser antilles and offshore otago peninsula, new zealand. Marine and Freshwater Research,54(6):745–754, 2003.[102] Henry Elderfield. Chromium speciation in sea water. Earth and Planetary Science Letters, 9(1):10–16,1970.[103] D Grimaud and G Michard. Concentration du chrome dans deux profils de l’oce´an pacifique. MarineChemistry, 2(3):229–237, 1974.[104] B. Gueguen, C. T. Reinhard, T. J. Algeo, et al. The chromium isotope composition of reducing andoxic marine sediments. Geochimica Et Cosmochimica Acta, 184:1–19, 2016.128[105] Christopher T. Reinhard, Noah J. Planavsky, Xiangli Wang, et al. The isotopic composition ofauthigenic chromium in anoxic marine sediments: A case study from the cariaco basin. Earth andPlanetary Science Letters, 407:9–18, 2014.[106] L. N. Dossing, K. Dideriksen, S. L. S. Stipp, and R. Frei. Reduction of hexavalent chromium byferrous iron: A process of chromium isotope fractionation and its relevance to natural environments.Chemical Geology, 285(1-4):157–166, 2011.[107] Anirban Basu and Thomas M. Johnson. Determination of hexavalent chromium reduction using crstable isotopes: Isotopic fractionation factors for permeable reactive barrier materials. EnvironmentalScience & Technology, 46(10):5353–5360, 2012.[108] Jacquelyn W. Kitchen, Thomas M. Johnson, Thomas D. Bullen, Jianming Zhu, and Amanda Raddatz.Chromium isotope fractionation factors for reduction of cr(vi) by aqueous fe(ii) and organic molecules.Geochimica Et Cosmochimica Acta, 89:190–201, 2012.[109] W. Stumm and J. J. Morgan. Aquatic chemistry; an introduction emphasizing chemical equilibria innatural waters. Current Contents/Agriculture Biology & Environmental Sciences, (41):18–18, 1988.[110] T. W. Lyons and S. Severmann. A critical look at iron paleoredox proxies: New insights from moderneuxinic marine basins. Geochimica Et Cosmochimica Acta, 70(23):5698–5722, 2006.[111] H. Boukhalfa and A. L. Crumbliss. Chemical aspects of siderophore mediated iron transport. Biometals,15(4):325–339, 2002.[112] F. J. Millero, S. Sotolongo, and M. Izaguirre. The oxidation kinetics of fe (ii) in seawater. GeochimicaEt Cosmochimica Acta, 51(4):793–801, 1987.[113] F. J. Millero, W. S. Yao, and J. Aicher. The speciation of fe (ii) and fe (iii) in natural waters. MarineChemistry, 50(1-4):21–39, 1995.[114] T. Nagai, A. Imai, K. Matsushige, K. Yokoi, and T. Fukushima. Dissolved iron and its speciation in ashallow eutrophic lake and its inflowing rivers. Water Research, 41(4):775–784, 2007.[115] J. M. Martin and M. Meybeck. Elemental mass-balance of material carried by major world rivers.Marine Chemistry, 7(3):173–206, 1979.[116] J. M. Martin and H. L. Windom. Present and future roles of ocean margins in regulating marinebiogeochemical cycles of trace elements. Ocean Margin Processes in Global Change, pages 45–67, 1991.[117] C. Klein. Some precambrian banded iron-formations (bifs) from around the world: Their age, geologicsetting, mineralogy, metamorphism, geochemistry, and origin. American Mineralogist, 90(10):1473–1499,2005.[118] R. Raiswell and D. E. Canfield. Sources of iron for pyrite formation in marine sediments. AmericanJournal of Science, 298(3):219–245, 1998.[119] R. Raiswell and D. E. Canfield. The iron biogeochemical cycle past and present. GeochemicalPerspectives, 1(1):1–220, 2012.[120] B. Thamdrup. Bacterial manganese and iron reduction in aquatic sediments. Advances in MicrobialEcology, Vol 16, 16:41–84, 2000.[121] D. R. Lovley. Magnetite formation during microbial dissimilatory iron reduction. Iron Biominerals,pages 151–166, 1991.[122] D. R. Lovley, J. F. Stolz, G. L. Nord, and E. J. P. Phillips. Anaerobic production of magnetite by adissimilatory iron-reducing microorganism. Nature, 330(6145):252–254, 1987.129[123] Derek R Lovley. Dissimilatory fe (iii) and mn (iv) reduction. Microbiological reviews, 55(2):259–287,1991.[124] D. E. Canfield, B. Thamdrup, and J. W. Hansen. The anaerobic degradation of organic matter indanish coastal sediments: iron reduction, manganese reduction, and sulfate reduction. Geochimica EtCosmochimica Acta, 57(16):3867–3883, 1993.[125] Bei Wu, Wulf Amelung, Ying Xing, Roland Bol, and Anne E Berns. Iron cycling and isotopefractionation in terrestrial ecosystems. Earth-Science Reviews, 2018.[126] E. E. Roden and J. M. Zachara. Microbial reduction of crystalline iron(iii) oxides: Influence of oxidesurface area and potential for cell growth. Environmental Science & Technology, 30(5):1618–1628, 1996.[127] R. Raiswell, D. S. Hardisty, T. W. Lyons, et al. The iron paleoredox proxies: A guide to the pitfalls,problems and proper practice. American Journal of Science, 318(5):491–526, 2018.[128] D. E. Canfield, R. Raiswell, and S. Bottrell. The reactivity of sedimentary iron minerals toward sulfide.American Journal of Science, 292(9):659–683, 1992.[129] S. W. Poulton and D. E. Canfield. Development of a sequential extraction procedure for iron:implications for iron partitioning in continentally derived particulates. Chemical Geology, 214(3-4):209–221, 2005.[130] Christopher T Reinhard, Timothy W Lyons, Olivier Rouxel, et al. Iron speciation and isotope perspectiveson Palaeoproterozoic water column chemistry, pages 1483–1492. Frontiers, 2013.[131] D. E. Canfield and A. Teske. Late proterozoic rise in atmospheric oxygen concentration inferred fromphylogenetic and sulphur-isotope studies. Nature, 382(6587):127–132, 1996.[132] S. E. Calvert and R. E. Karlin. Relationships between sulfur, organic-carbon, and iron in the modernsediments of the black-sea. Geochimica Et Cosmochimica Acta, 55(9):2483–2490, 1991.[133] T. W. Lyons, J. P. Werne, D. J. Hollander, and R. W. Murray. Contrasting sulfur geochemistry and fe/aland mo/al ratios across the last oxic-to-anoxic transition in the cariaco basin, venezuela. ChemicalGeology, 195(1-4):131–157, 2003.[134] C. T. Scott, A. Bekker, C. T. Reinhard, et al. Late archean euxinic conditions before the rise ofatmospheric oxygen. Geology, 39(2):119–122, 2011.[135] Vincent Busigny, Noah J Planavsky, Didier Je´ze´quel, et al. Iron isotopes in an archean ocean analogue.Geochimica et Cosmochimica Acta, 133:443–462, 2014.[136] S. A. Crowe, D. E. Canfield, D. A. Fowle, et al. Biogeochemistry and microbial ecology of a modern,ferruginous chemocline. Geochimica Et Cosmochimica Acta, 74(12):A196–A196, 2010.[137] S. A. Crowe, C. Jones, S. Katsev, et al. Photoferrotrophs thrive in an archean ocean analogue.Proceedings of the National Academy of Sciences of the United States of America, 105(41):15938–15943, 2008.[138] S. A. Crowe, A. H. O’Neill, S. Katsev, et al. The biogeochemistry of tropical lakes: A case study fromlake matano, indonesia. Limnology and Oceanography, 53(1):319–331, 2008.[139] Nicholas Lambrecht, Chad Wittkop, Sergei Katsev, Mojtaba Fakhraee, and Elizabeth D Swanner.Geochemical characterization of two ferruginous meromictic lakes in the upper midwest, usa. Journalof Geophysical Research: Biogeosciences, 2018.[140] J. M. Russell, S. Bijaksana, H. Vogel, et al. The towuti drilling project: paleoenvironments, biologicalevolution, and geomicrobiology of a tropical pacific lake. Scientific Drilling, 21:29–40, 2016.130[141] A. Vuillemin, A. Friese, M. Alawi, et al. Geomicrobiological features of ferruginous sediments fromlake towuti, indonesia. Frontiers in Microbiology, 7, 2016.[142] A. Zegeye, S. Bonneville, L. G. Benning, et al. Green rust formation controls nutrient availability in aferruginous water column. Geology, 40(7):599–602, 2012.[143] S. A. Crowe, J. A. Maresca, C. Jones, et al. Deep-water anoxygenic photosythesis in a ferruginouschemocline. Geobiology, 12(4):322–339, 2014.[144] D. E. Canfield. The geochemistry of river particulates from the continental usa: Major elements.Geochimica Et Cosmochimica Acta, 61(16):3349–3365, 1997.[145] C. Holmden, A. D. Jacobson, B. B. Sageman, and M. T. Hurtgen. Response of the cr isotope proxy tocretaceous ocean anoxic event 2 in a pelagic carbonate succession from the western interior seaway.Geochimica Et Cosmochimica Acta, 186:277–295, 2016.[146] A. M. Yusof, C. H. Chia, and A. K. H. Wood. Speciation of cr(iii) and cr(vi) in surface waters witha chelex-100 resin column and their quantitative determination using inductively coupled plasmamass spectrometry and instrumental neutron activation analysis. Journal of Radioanalytical and NuclearChemistry, 273(3):533–538, 2007.[147] C. R. German, J. Hergt, M. R. Palmer, and J. M. Edmond. Geochemistry of a hydrothermal sedimentcore from the obs vent-field, 21 degrees n east pacific rise. Chemical Geology, 155(1-2):65–75, 1999.[148] T. J. Shaw, J. M. Gieskes, and R. A. Jahnke. Early diagenesis in differing depositional environments:The response of transition metals in pore water. Geochimica Et Cosmochimica Acta, 54(5):1233–1246,1990.[149] M. G. Babechuk, I. C. Kleinhanns, and R. Schoenberg. Chromium geochemistry of the ca. 1.85 ga flinflon paleosol. Geobiology, 15(1):30–50, 2017.[150] M. W. Lyle. Major element composition of leg 92 sediments ( east pacific rise). Initial Reports of theDeep Sea Drilling Project, 92:355–370, 1986.[151] S. W. Poulton and D. E. Canfield. Co-diagenesis of iron and phosphorus in hydrothermal sedimentsfrom the southern east pacific rise: Implications for the evaluation of paleoseawater phosphateconcentrations. Geochimica Et Cosmochimica Acta, 70(23):5883–5898, 2006.[152] T. J. Barrett, P. N. Taylor, and J. Lugowski. Metalliferous sediments from dsdp leg 92: The east pacificrise transect. Geochimica Et Cosmochimica Acta, 51(9):2241–2253, 1987.[153] B. Thamdrup, H. Fossing, and B. B. Jorgensen. Manganese, iron and sulfur cycling in a coastal marinesediment, aarhus bay, denmark. Geochimica Et Cosmochimica Acta, 58(23):5115–5129, 1994.[154] A. Neaman, B. Waller, F. Mouele, F. Trolard, and G. Bourrie. Improved methods for selectivedissolution of manganese oxides from soils and rocks. European Journal of Soil Science, 55(1):47–54,2004.[155] T. T. Chao. Selective dissolution of manganese oxides from soils and sediments with acidifiedhydroxylamine hydrochloride. Soil Science Society of America Proceedings, 36(5):764–&, 1972.[156] Jan G Wiederhold, Stephan M Kraemer, Nadya Teutsch, et al. Iron isotope fractionation duringproton-promoted, ligand-controlled, and reductive dissolution of goethite. Environmental science &technology, 40(12):3787–3793, 2006.[157] C. Siebert, T. F. Nagler, and J. D. Kramers. Determination of molybdenum isotope fractionation bydouble-spike multicollector inductively coupled plasma mass spectrometry. Geochemistry GeophysicsGeosystems, 2:art. no.–2000GC000124, 2001.131[158] S. W. Poulton and D. E. Canfield. Co-diagenesis of iron and phosphorus in hydrothermal sedimentsfrom the southern east pacific rise: Implications for the evaluation of paleoseawater phosphateconcentrations. Geochimica Et Cosmochimica Acta, 70(23):5883–5898, 2006.[159] James A Davis and James O Leckie. Surface ionization and complexation at the oxide/water interface.3. adsorption of anions. Journal of Colloid and Interface Science, 74(1):32–43, 1980.[160] X. L. Wang, N. J. Planavsky, P. M. Hull, et al. Chromium isotopic composition of core-top planktonicforaminifera. Geobiology, 15(1):51–64, 2017.[161] Yuanzhi Tang, Evert J. Elzinga, Young Jae Lee, and Richard J. Reeder. Coprecipitation of chromatewith calcite: Batch experiments and x-ray absorption spectroscopy. Geochimica Et Cosmochimica Acta,71(6):1480–1493, 2007.[162] Wei Wei, Robert Frei, Tian-Yu Chen, et al. Marine ferromanganese oxide: A potentially importantsink of light chromium isotopes? Chemical Geology, 2018.[163] David K Rea and Maurice K Bloomstine. Neogene history of the south pacific tradewinds: evidence forhemispherical asymmetry of atmospheric circulation. Palaeogeography, palaeoclimatology, palaeoecology,55(1):55–64, 1986.[164] Cora Paulukat, Lasse N. Dossing, Sisir K. Mondal, Andrea R. Voegelin, and Robert Frei. Oxidativerelease of chromium from archean ultramafic rocks, its transport and environmental impact - a crisotope perspective on the sukinda valley ore district (orissa, india). Applied Geochemistry, 59:125–138,2015.[165] G. R. Heath and J. Dymond. Genesis and transformation of metalliferous sediments from the eastpacific rise, bauer deep, and central basin, northwest nazca plate. Geological Society of America Bulletin,88(5):723–733, 1977.[166] P. Michalopoulos and R. C. Aller. Early diagenesis of biogenic silica in the amazon delta: Alteration,authigenic clay formation, and storage. Geochimica Et Cosmochimica Acta, 68(5):1061–1085, 2004.[167] X. L. Wang, N. J. Planavsky, C. T. Reinhard, et al. Chromium isotope fractionation during subduction-related metamorphism, black shale weathering, and hydrothermal alteration. Chemical Geology,423:19–33, 2016.[168] J. L. Charlou, Y. Fouquet, J. P. Donval, et al. Mineral and gas chemistry of hydrothermal fluids on anultrafast spreading ridge: East pacific rise, 17 degrees to 19 degrees s (naudur cruise, 1993) phaseseparation processes controlled by volcanic and tectonic activity. Journal of Geophysical Research-SolidEarth, 101(B7):15899–15919, 1996.[169] Kohen W Bauer, Bleuenn Gueguen, Devon B Cole, et al. Chromium isotope fractionation in ferruginoussediments. Geochimica et Cosmochimica Acta, 223:198–215, 2018.[170] E. R. Sikora, T. M. Johnson, and T. D. Bullen. Microbial mass-dependent fractionation of chromiumisotopes. Geochimica Et Cosmochimica Acta, 72(15):3631–3641, 2008.[171] Cora Paulukat, Geoffrey J Gilleaudeau, Pavel Chernyavskiy, and Robert Frei. The cr-isotope signatureof surface seawater—a global perspective. Chemical Geology, 444:101–109, 2016.[172] E. Nakayama, T. Kuwamoto, S. Tsurubo, and T. Fujinaga. Chemical speciation of chromium in seawater: Part 2. effects of manganese oxides and reducible organic materials on the redox processes ofchromium. Analytica Chimica Acta, 130(2):401–404, 1981.[173] U. Schwertmann and E. Murad. Effect of ph on the formation of goethite and hematite fromferrihydrite. Clays and Clay Minerals, 31(4):277–284, 1983.132[174] Alfons Berger and Robert Frei. The fate of chromium during tropical weathering: A laterite profilefrom central madagascar. Geoderma, 213:521–532, 2014.[175] Alcides N. Sial, Marcel S. Campos, Claudio Gaucher, et al. Algoma-type neoproterozoic bifs andrelated marbles in the serido belt (ne brazil): Ree, c, o, cr and sr isotope evidence. Journal of SouthAmerican Earth Sciences, 61:33–52, 2015.[176] O. W. Duckworth, M. M. Akafia, M. Y. Andrews, and J. R. Bargar. Siderophore-promoted dissolutionof chromium from hydroxide minerals. Environmental Science-Processes & Impacts, 16(6):1348–1359,2014.[177] J. P. Gustafsson, I. Persson, A. G. Oromieh, et al. Chromium(iii) complexation to natural organicmatter: Mechanisms and modeling. Environmental Science & Technology, 48(3):1753–1761, 2014.[178] D. L. Sedlak and P. G. Chan. Reduction of hexavalent chromium by ferrous iron. Geochimica EtCosmochimica Acta, 61(11):2185–2192, 1997.[179] Srabanti Basu, Monikankana Dasgupta, and Bhaswati Chakraborty. Removal of chromium (vi) bybacillus subtilis isolated from east calcutta wetlands, west bengal, india. International Journal ofBioscience, Biochemistry and Bioinformatics, 4(1):7, 2014.[180] Emily C. Berna, Thomas M. Johnson, Richard S. Makdisi, and Anirban Basui. Cr stable isotopes asindicators of cr(vi) reduction in groundwater: A detailed time-series study of a point-source plume.Environmental Science & Technology, 44(3):1043–1048, 2010.[181] Christoph Wanner, Urs Eggenberger, Daniel Kurz, Sonja Zink, and Urs Ma¨der. A chromate-contaminated site in southern switzerland–part 1: Site characterization and the use of cr isotopes todelineate fate and transport. Applied geochemistry, 27(3):644–654, 2012.[182] Simon W. Poulton and Donald E. Canfield. Ferruginous conditions: A dominant feature of the oceanthrough earth’s history. Elements, 7(2):107–112, 2011.[183] A. Mariotti, J. C. Germon, P. Hubert, et al. Experimental determination of nitrogen kinetic isotopefractionation: some principles; illustration for the denitrification and nitrification processes. Plant andSoil, 62(3):413–430, 1981.[184] J. A. Brandes and A. H. Devol. Isotopic fractionation of oxygen and nitrogen in coastal marinesediments. Geochimica Et Cosmochimica Acta, 61(9):1793–1801, 1997.[185] Scott K. Clark and Thomas M. Johnson. Effective isotopic fractionation factors for solute removal byreactive sediments: A laboratory microcosm and slurry study. Environmental Science & Technology,42(21):7850–7855, 2008.[186] H. Rasmussen and B. B. Jorgensen. Microelectrode studies of seasonal oxygen uptake in a coastalsediment: Role of molecular diffusion. Marine Ecology Progress Series, 81(3):289–303, 1992.[187] Z. Marczenko. Spectrophotometric determination of trace elements. Crc Critical Reviews in AnalyticalChemistry, 11(3):195–260, 1981.[188] L. L. Stookey. Ferrozine—a new spectrophotometric reagent for iron. Analytical Chemistry, 42(7):779–&,1970.[189] Y. Abe and D. Hunkeler. Does the rayleigh equation apply to evaluate field isotope data in contami-nant hydrogeology? Environmental Science & Technology, 40(5):1588–1596, 2006.[190] B. M. Van Breukelen and H. Prommer. Beyond the rayleigh equation: Reactive transport modeling ofisotope fractionation effects to improve quantification of biodegradation. Environmental Science &Technology, 42(7):2457–2463, 2008.133[191] C. T. Green, J. K. Bohlke, B. A. Bekins, and S. P. Phillips. Mixing effects on apparent reaction ratesand isotope fractionation during denitrification in a heterogeneous aquifer. Water Resources Research,46, 2010.[192] T. M. Johnson and D. J. Depaolo. Interpretation of isotopic data in groundwater-rock systems: Modeldevelopment and application to sr isotope data from yucca mountain. Water Resources Research,30(5):1571–1587, 1994.[193] Ellery D Ingall and Philippe Van Cappellen. Relation between sedimentation rate and burial oforganic phosphorus and organic carbon in marine sediments. Geochimica et Cosmochimica Acta,54(2):373–386, 1990.[194] Bernard P Boudreau. Diagenetic models and their implementation, volume 606. Springer Berlin, 1997.[195] R. N. Glud, J. K. Gundersen, B. B. Jorgensen, N. P. Revsbech, and H. D. Schulz. Diffusive and totaloxygen uptake of deep-sea sediments in the eastern south atlantic ocean: in situ and laboratorymeasurements. Deep-Sea Research Part I-Oceanographic Research Papers, 41(11-12):1767–1788, 1994.[196] C. Wanner and E. L. Sonnenthal. Assessing the control on the effective kinetic cr isotope fractionationfactor: A reactive transport modeling approach. Chemical Geology, 337:88–98, 2013.[197] C. Wanner, J. L. Druhan, R. T. Amos, P. Alt-Epping, and C. I. Steefel. Benchmarking the simulation ofcr isotope fractionation. Computational Geosciences, 19(3):497–521, 2015.[198] E. J. Tomaszewski, S. Lee, J. Rudolph, H. F. Xu, and M. Ginder-Vogel. The reactivity of fe(ii)associated with goethite formed during short redox cycles toward cr(vi) reduction under oxicconditions. Chemical Geology, 464:101–109, 2017.[199] R. Y. Han, L. P. Qin, S. T. Brown, J. N. Christensen, and H. R. Beller. Differential isotopic fractionationduring cr(vi) reduction by an aquifer-derived bacterium under aerobic versus denitrifying conditions.Applied and Environmental Microbiology, 78(7):2462–2464, 2012.[200] Robert A Berner. Early diagenesis: a theoretical approach. Princeton University Press, 1980.[201] W. J. Cai and C. E. Reimers. Benthic oxygen flux, bottom water oxygen concentration and core toporganic carbon content in the deep northeast pacific ocean. Deep-Sea Research Part I-OceanographicResearch Papers, 42(10):1681–1699, 1995.[202] W. J. Cai and F. L. Sayles. Oxygen penetration depths and fluxes in marine sediments. MarineChemistry, 52(2):123–131, 1996.[203] S. A. Crowe, S. Katsev, K. Leslie, et al. The methane cycle in ferruginous lake matano. Geobiology,9(1):61–78, 2011.[204] P. J. Muller and A. Mangini. Organic carbon decomposition rates in sediments of the pacificmanganese nodule belt dated by 230th and 231pa. Earth and Planetary Science Letters, 51(1):94–114,1980.[205] Daniel A. Stolper and C. Brenhin Keller. A record of deep-ocean dissolved o2 from the oxidationstate of iron in submarine basalts. Nature, 553(7688):323–327, 2018.[206] I. Halevy, S. E. Peters, and W. W. Fischer. Sulfate burial constraints on the phanerozoic sulfur cycle.Science, 337(6092):331–334, 2012.[207] H. C. Jenkyns. Geochemistry of oceanic anoxic events. Geochemistry Geophysics Geosystems, 11:30,2010.134[208] M. O. Clarkson, R. A. Wood, S. W. Poulton, et al. Dynamic anoxic ferruginous conditions during theend-permian mass extinction and recovery. Nature Communications, 7, 2016.[209] M. M. M. Kuypers, R. D. Pancost, I. A. Nijenhuis, and J. S. S. Damste. Enhanced productivity ledto increased organic carbon burial in the euxinic north atlantic basin during the late cenomanianoceanic anoxic event. Paleoceanography, 17(4), 2002.[210] S. W. Poulton, S. Henkel, C. Marz, et al. A continental-weathering control on orbitally drivenredox-nutrient cycling during cretaceous oceanic anoxic event 2. Geology, 43(11):963–966, 2015.[211] C. Marz, S. W. Poulton, B. Beckmann, et al. Redox sensitivity of p cycling during marine black shaleformation: Dynamics of sulfidic and anoxic, non-sulfidic bottom waters. Geochimica Et CosmochimicaActa, 72(15):3703–3717, 2008.[212] M. L. Gomes, M. T. Hurtgen, and B. B. Sageman. Biogeochemical sulfur cycling during cretaceousoceanic anoxic events: A comparison of oae1a and oae2. Paleoceanography, 31(2):233–251, 2016.[213] U. G. Wortmann and B. M. Chernyavsky. Effect of evaporite deposition on early cretaceous carbonand sulphur cycling. Nature, 446(7136):654–656, 2007.[214] L. R. Tedeschi, H. C. Jenkyns, S. A. Robinson, et al. New age constraints on aptian evaporites andcarbonates from the south atlantic: Implications for oceanic anoxic event 1a. Geology, 45(6):543–546,2017.[215] K. Knittel and A. Boetius. Anaerobic oxidation of methane: Progress with an unknown process.Annual Review of Microbiology, 63:311–334, 2009.[216] W. S. Reeburgh. Oceanic methane biogeochemistry. Chemical Reviews, 107(2):486–513, 2007.[217] S. L. Olson, C. T. Reinhard, and T. W. Lyons. Limited role for methane in the mid-proterozoicgreenhouse. Proceedings of the National Academy of Sciences of the United States of America, 113(41):11447–11452, 2016.[218] Elisabetta Erba, Cinzia Bottini, Helmut J. Weissert, and Christina E. Keller. Calcareous nannoplanktonresponse to surface-water acidification around oceanic anoxic event 1a. Science, 329(5990):428–432,2010.[219] A. Malinverno, E. Erba, and T. D. Herbert. Orbital tuning as an inverse problem: Chronology of theearly aptian oceanic anoxic event 1a (selli level) in the cismon apticore. Paleoceanography, 25, 2010.[220] A. P. Menegatti, H. Weissert, R. S. Brown, et al. High-resolution delta c-13 stratigraphy through theearly aptian ”livello selli” of the alpine tethys. Paleoceanography, 13(5), 1998.[221] J. V. Mills, M. L. Gomes, B. Kristall, et al. Massive volcanism, evaporite deposition, and the chemicalevolution of the early cretaceous ocean. Geology, 45(5):475–478, 2017.[222] C Bottini, E Erba, D Tiraboschi, et al. Climate variability and ocean fertility during the aptian stage.Climate of the Past, 11(3):383–402, 2015.[223] C. Bottini, A. S. Cohen, E. Erba, H. C. Jenkyns, and A. L. Coe. Osmium-isotope evidence for volcanism,weathering, and ocean mixing during the early aptian oae 1a. Geology, 40(7):583–586, 2012.[224] Yong-Xiang Li, Timothy J Bralower, Isabel P Montan˜ez, et al. Toward an orbital chronology for theearly aptian oceanic anoxic event (oae1a, 120 ma). Earth and Planetary Science Letters, 271(1):88–100,2008.135[225] Yvonne van Breugel, Stefan Schouten, Harilaos Tsikos, et al. Synchronous negative carbon isotopeshifts in marine and terrestrial biomarkers at the onset of the early aptian oceanic anoxic event 1a:Evidence for the release of c-13-depleted carbon into the atmosphere. Paleoceanography, 22(1), 2007.[226] M. O. Clarkson, S. W. Poulton, R. Guilbaud, and R. A. Wood. Assessing the utility of fe/al andfe-speciation to record water column redox conditions in carbonate-rich sediments. Chemical Geology,382:111–122, 2014.[227] D. Colodner, J. Sachs, G. Ravizza, et al. The geochemical cycle of rhenium: a reconnaissance. Earthand Planetary Science Letters, 117(1-2):205–221, 1993.[228] S. W. Poulton, P. W. Fralick, and D. E. Canfield. The transition to a sulphidic ocean similar to 1.84billion years ago. Nature, 431(7005):173–177, 2004.[229] K. B. Follmi, M. Bole, N. Jammet, et al. Bridging the faraoni and selli oceanic anoxic events: latehauterivian to early aptian dysaerobic to anaerobic phases in the tethys. Climate of the Past, 8(1):171–189, 2012.[230] P. Kraal, C. P. Slomp, A. Forster, M. M. M. Kuypers, and A. Sluijs. Pyrite oxidation during samplestorage determines phosphorus fractionation in carbonate-poor anoxic sediments. Geochimica EtCosmochimica Acta, 73(11):3277–3290, 2009.[231] S. P. Slotznick, J. M. Eiler, and W. W. Fischer. The effects of metamorphism on iron mineralogy andthe iron speciation redox proxy. Geochimica Et Cosmochimica Acta, 224:96–115, 2018.[232] D. E. Canfield, F. J. Stewart, B. Thamdrup, et al. A cryptic sulfur cycle in oxygen-minimum-zonewaters off the chilean coast. Science, 330(6009):1375–1378, 2010.[233] K. S. Habicht and D. E. Canfield. Sulfur isotope fractionation during bacterial sulfate reduction inorganic-rich sediments. Geochimica Et Cosmochimica Acta, 61(24):5351–5361, 1997.[234] M. Kunzmann, T. H. Bui, P. W. Crockford, et al. Bacterial sulfur disproportionation constrains timingof neoproterozoic oxygenation. Geology, 45(3):207–210, 2017.[235] M. N. Timofeeff, T. K. Lowenstein, M. A. da Silva, and N. B. Harris. Secular variation in themajor-ion chemistry of seawater: Evidence from fluid inclusions in cretaceous halites. Geochimica EtCosmochimica Acta, 70(8), 2006.[236] A. Kampschulte and H. Strauss. The sulfur isotopic evolution of phanerozoic seawater based on theanalysis of structurally substituted sulfate in carbonates. Chemical Geology, 204(3-4):255–286, 2004.[237] John H Martin, George A Knauer, David M Karl, and William W Broenkow. Vertex: carbon cycling inthe northeast pacific. Deep Sea Research Part A. Oceanographic Research Papers, 34(2):267–285, 1987.[238] Jeremy D. Owens, Benjamin C. Gill, Hugh C. Jenkyns, et al. Sulfur isotopes track the global extentand dynamics of euxinia during cretaceous oceanic anoxic event 2. Proceedings of the National Academyof Sciences of the United States of America, 110(46):18407–18412, 2013.[239] Philipp Bo¨ning. Trace element signatures of Peruvian and Chilean upwelling sediments. Thesis, 2005.[240] F. Scholz, S. Severmann, J. McManus, et al. On the isotope composition of reactive iron in marinesediments: Redox shuttle versus early diagenesis. Chemical Geology, 389:48–59, 2014.[241] Wei-Jun Cai, Xinping Hu, Wei-Jen Huang, et al. Acidification of subsurface coastal waters enhancedby eutrophication. Nature geoscience, 4(11):766, 2011.136[242] L. R. Kump and W. E. Seyfried. Hydrothermal fe fluxes during the precambrian: Effect of low oceanicsulfate concentrations and low hydrostatic pressure on the composition of black smokers. Earth andPlanetary Science Letters, 235(3-4):654–662, 2005.[243] T. J. Horner, H. V. Pryer, S. G. Nielsen, et al. Pelagic barite precipitation at micromolar ambient sulfate.Nature Communications, 8, 2017.[244] G. Paris, J. F. Adkins, A. L. Sessions, S. M. Webb, and W. W. Fischer. Neoarchean carbonate-associatedsulfate records positive delta s-33 anomalies. Science, 346(6210):739–741, 2014.[245] M. J. Mitchell, D. H. Landers, D. F. Brodowski, G. B. Lawrence, and M. B. David. Organic andinorganic sulfur constituents of the sediments in 3 new-york lakes - effect of site, sediment depth andseason. Water Air and Soil Pollution, 21(1-4):231–245, 1984.[246] N. R. Urban. Retention of sulfur in lake-sediments. Environmental Chemistry of Lakes and Reservoirs,237:323–369, 1994.[247] K. M. Fagerbakke, M. Heldal, and S. Norland. Content of carbon, nitrogen, oxygen, sulfur andphosphorus in native aquatic and cultured bacteria. Aquatic Microbial Ecology, 10(1):15–27, 1996.[248] I. R. Kaplan, K. O. Emery, and S. C. Rittenberg. The distribution and isotopic abundance of sulphurin recent marine sediments off southern california. Geochimica Et Cosmochimica Acta, 27(APR):297–&,1963.[249] I. R. Kaplan and S. C. Rittenberg. Microbiological fractionation of sulphur isotopes. Journal of GeneralMicrobiology, 34(2):195–&, 1964.[250] A. C. Utne-Palm, A. G. V. Salvanes, B. Currie, et al. Trophic structure and community stability in anoverfished ecosystem. Science, 329(5989):333–336, 2010.[251] J. S. S. Damste, S. G. Wakeham, M. E. L. Kohnen, J. M. Hayes, and J. W. Deleeuw. A 6,000–yearsedimentary molecular record of chemocline excursions in the black sea. Nature, 362(6423):827–829,1993.[252] M. W. Claire, D. C. Catling, and K. J. Zahnle. Biogeochemical modelling of the rise in atmosphericoxygen. Geobiology, 4(4):239–269, 2006.[253] James F. Kasting. What caused the rise of atmospheric o-2? Chemical Geology, 362:13–25, 2013.[254] D. C. Catling, K. J. Zahnle, and C. P. McKay. Biogenic methane, hydrogen escape, and the irreversibleoxidation of early earth. Science, 293(5531):839–843, 2001.[255] Aubrey L. Zerkle, Markw Claire, Shawn D. Domagal-Goldman, James Farquhar, and Simon W.Poulton. A bistable organic-rich atmosphere on the neoarchaean earth. Nature Geoscience, 5(5):359–363, 2012.[256] N. J. Beukes and C. Klein. Geochemistry and sedimentology of a facies transition—from microbandedto granular iron-formation—in the early proterozoic transvaal supergroup, south africa. PrecambrianResearch, 47(1-2):99–139, 1990.[257] C. M. Johnson, B. L. Beard, C. Klein, N. J. Beukes, and E. E. Roden. Iron isotopes constrain biologic andabiologic processes in banded iron formation genesis. Geochimica Et Cosmochimica Acta, 72(1):151–169,2008.[258] J. C. G. Walker. Suboxic diagenesis in banded iron formations. Nature, 309(5966):340–342, 1984.137[259] M. Halama, E. D. Swanner, K. O. Konhauser, and A. Kappler. Evaluation of siderite and magnetiteformation in bifs by pressure-temperature experiments of fe(iii) minerals and microbial biomass.Earth and Planetary Science Letters, 450:243–253, 2016.[260] R. Karlin, M. Lyle, and G. R. Heath. Authigenic magnetite formation in suboxic marine sediments.Nature, 326(6112):490–493, 1987.[261] K. O. Konhauser, D. K. Newman, and A. Kappler. The potential significance of microbial fe(iii)reduction during deposition of precambrian banded iron formations. Geobiology, 3(3):167–177, 2005.[262] I. Halevy, M. Alesker, E. M. Schuster, R. Popovitz-Biro, and Y. Feldman. A key role for green rust inthe precambrian oceans and the genesis of iron formations. Nature Geoscience, 10(2):135–+, 2017.[263] N. J. Tosca, S. Guggenheim, and P. K. Pufahl. An authigenic origin for precambrian greenalite:Implications for iron formation and the chemistry of ancient seawater. Geological Society of AmericaBulletin, 128(3-4):511–530, 2016.[264] A. C. Itambi, T. von Dobeneck, M. J. Dekkers, and T. Frederichs. Magnetic mineral inventory ofequatorial atlantic ocean marine sediments off senegal-glacial and interglacial contrast. GeophysicalJournal International, 183(1):163–177, 2010.[265] Y. Kimura, T. Sato, N. Nakamura, et al. Vortex magnetic structure in framboidal magnetite revealsexistence of water droplets in an ancient asteroid. Nature Communications, 4, 2013.[266] D. Suk, D. R. Peacor, and R. Vandervoo. Replacement of pyrite framboids by magnetite in limestoneand implications for palaeomagnetism. Nature, 345(6276):611–613, 1990.[267] Edward J O’Loughlin, Christopher A Gorski, and Michelle M Scherer. Effects of phosphate onsecondary mineral formation during the bioreduction of akaganeite (-feooh): Green rust versusframboidal magnetite. Current Inorganic Chemistry, 5(3):214–224, 2015.[268] R. T. Wilkin and H. L. Barnes. Formation processes of framboidal pyrite. Geochimica Et CosmochimicaActa, 61(2):323–339, 1997.[269] R. T. Wilkin, H. L. Barnes, and S. L. Brantley. The size distribution of framboidal pyrite in modernsediments: An indicator of redox conditions. Geochimica Et Cosmochimica Acta, 60(20):3897–3912, 1996.[270] M. T. Rosing, D. K. Bird, N. H. Sleep, and C. J. Bjerrum. No climate paradox under the faint earlysun. Nature, 464(7289):744–U117, 2010.[271] C. T. Reinhard and N. J. Planavsky. Mineralogical constraints on precambrian p(co2). Nature,474(7349):E1–E2, 2011.[272] B. L. Beard, C. M. Johnson, K. L. Von Damm, and R. L. Poulson. Iron isotope constraints on fe cyclingand mass balance in oxygenated earth oceans. Geology, 31(7):629–632, 2003.[273] Olivier J Rouxel, Andrey Bekker, and Katrina J Edwards. Iron isotope constraints on the archean andpaleoproterozoic ocean redox state. Science, 307(5712):1088–1091, 2005.[274] N. Planavsky, O. Rouxel, A. Bekker, et al. Iron-oxidizing microbial ecosystems thrived in latepaleoproterozoic redox-stratified oceans. Earth and Planetary Science Letters, 286(1-2):230–242, 2009.[275] C. Klein and N. J. Beukes. Geochemistry and sedimentology of a facies transition from limestoneto iron-formation deposition in the early proterozoic transvaal supergroup, south africa. EconomicGeology, 84(7):1733–1774, 1989.[276] Jung Ho Ahn and Peter R Buseck. Hematite nanospheres of possible colloidal origin from aprecambrian banded iron formation. Science, 250(4977):111–113, 1990.138[277] DE Ayres. Genesis of iron-bearing minerals in banded iron formation mesobands in the dales gorgemember, hamersley group, western australia. Economic Geology, 67(8):1214–1233, 1972.[278] MS Lougheed and JJ Mancuso. Hematite framboids in the negaunee iron formation, michigan;evidence for their biogenic origin. Economic geology, 68(2):202–209, 1973.[279] Yi-Liang Li, Kurt O Konhauser, Andreas Kappler, and Xi-Luo Hao. Experimental low-grade alterationof biogenic magnetite indicates microbial involvement in generation of banded iron formations. Earthand Planetary Science Letters, 361:229–237, 2013.[280] J. F. Slack, T. Grenne, A. Bekker, O. J. Rouxel, and P. A. Lindberg. Suboxic deep seawater in the latepaleoproterozoic: Evidence from hematitic chert and iron formation related to seafloor-hydrothermalsulfide deposits, central arizona, usa. Earth and Planetary Science Letters, 255(1-2):243–256, 2007.[281] Weiqi Yao, Adina Paytan, and Ulrich G Wortmann. Large-scale ocean deoxygenation during thepaleocene-eocene thermal maximum. Science, 361(6404):804–806, 2018.[282] I. B. Hutchinson, R. Ingley, H. G. M. Edwards, et al. Raman spectroscopy on mars: identification ofgeological and bio-geological signatures in martian analogues using miniaturized raman spectrom-eters. Philosophical Transactions of the Royal Society a-Mathematical Physical and Engineering Sciences,372(2030), 2014.[283] B. B. Jorgensen. A theoretical model of the stable sulfur isotope distribution in marine sediments.Geochimica Et Cosmochimica Acta, 43(3):363–374, 1979.[284] DAVID J Toth and ABRAHAM Lerman. Organic matter reactivity and sedimentation rates in theocean. American Journal of Science, 277(4):465–485, 1977.[285] E. Erba and R. L. Larson. The cismon apticore (southern alps, italy): A ”reference section” for thelower cretaceous at low latitudes. Rivista Italiana Di Paleontologia E Stratigrafia, 104(2):181–191, 1998.[286] P. H. Roth. Mid-cretaceous nannoplankton from the central pacific implications for palaeoceanography.Initial Reports of the Deep Sea Drilling Project, 62:471–489, 1981.[287] A. Ando, T. Kakegawa, R. Takashima, and T. Saito. New perspective on aptian carbon isotopestratigraphy: Data from delta c-13 records of terrestrial organic matter. Geology, 30(3):227–230, 2002.[288] E. Erba. Calcareous nannofossils and mesozoic oceanic anoxic events. Marine Micropaleontology,52(1-4):85–106, 2004.[289] A. Tessier, P. G. C. Campbell, and M. Bisson. Sequential extraction procedure for the speciation ofparticulate trace metals. Analytical Chemistry, 51(7):844–851, 1979.[290] M. A. Huertadiaz and J. W. Morse. A quantitative method for determination of trace metal concentra-tions in sedimentary pyrite. Marine Chemistry, 29(2-3):119–144, 1990.[291] YH Li and JE Schoonmaker. Chemical composition and mineralogy of marine sediments, volume 7. na,2003.[292] H. W. Nesbitt and G. M. Young. Early proterozoic climates and plate motions inferred from majorelement chemistry of lutites. Nature, 299(5885):715–717, 1982.[293] Sean A. Crowe, Guillaume Paris, Sergei Katsev, et al. Sulfate was a trace constituent of archeanseawater. Science, 346(6210):735–739, 2014.[294] I. H. Tarpgaard, H. Roy, and B. B. Jorgensen. Concurrent low- and high-affinity sulfate reductionkinetics in marine sediment. Geochimica Et Cosmochimica Acta, 75(11):2997–3010, 2011.139[295] Marc Lliro´s, Tamara Garcı´a–Armisen, Franc¸ois Darchambeau, et al. Pelagic photoferrotrophy andiron cycling in a modern ferruginous basin. Scientific reports, 5:13803, 2015.[296] Valeria Luciani, Miriam Cobianchi, and Claudia Lupi. Regional record of a global oceanic anoxicevent: Oae1a on the apulia platform margin, gargano promontory, southern italy. Cretaceous Research,27(6):754–772, 2006.[297] L. N. Neretin, II Volkov, M. E. Bottcher, and V. A. Grinenko. A sulfur budget for the black sea anoxiczone. Deep-Sea Research Part I-Oceanographic Research Papers, 48(12):2569–2593, 2001.[298] D. Kadko. Upwelling and primary production during the us geotraces east pacific zonal transect.Global Biogeochemical Cycles, 31(2):218–232, 2017.[299] A. Gnanadesikan, D. Bianchi, and M. A. Pradal. Critical role for mesoscale eddy diffusion insupplying oxygen to hypoxic ocean waters. Geophysical Research Letters, 40(19):5194–5198, 2013.[300] W. B. Whitman, D. C. Coleman, and W. J. Wiebe. Prokaryotes: The unseen majority. Proceedings of theNational Academy of Sciences of the United States of America, 95(12):6578–6583, 1998.[301] S. Louca and S. A. Crowe. Microscale reservoir effects on microbial sulfur isotope fractionation.Geochimica Et Cosmochimica Acta, 203:117–139, 2017.[302] Boswell A Wing and Itay Halevy. Intracellular metabolite levels shape sulfur isotope fractionationduring microbial sulfate respiration. Proceedings of the National Academy of Sciences, 111(51):18116–18125, 2014.[303] D. E. Canfield, B. Thamdrup, and S. Fleischer. Isotope fractionation and sulfur metabolism by pureand enrichment cultures of elemental sulfur-disproportionating bacteria. Limnology and Oceanography,43(2):253–264, 1998.[304] C. M. Hansel, C. J. Lentini, Y. Z. Tang, et al. Dominance of sulfur-fueled iron oxide reduction inlow-sulfate freshwater sediments. Isme Journal, 9(11):2400–2412, 2015.[305] Sergei Katsev and Sean A. Crowe. Organic carbon burial efficiencies in sediments: The power law ofmineralization revisited. Geology, 43(7):607–610, 2015.[306] C. L. Blattler, H. C. Jenkyns, L. M. Reynard, and G. M. Henderson. Significant increases in globalweathering during oceanic anoxic events 1a and 2 indicated by calcium isotopes. Earth and PlanetaryScience Letters, 309(1-2):77–88, 2011.[307] D. D. Adams, M. T. Hurtgen, and B. B. Sageman. Volcanic triggering of a biogeochemical cascadeduring oceanic anoxic event 2. Nature Geoscience, 3(3):201–204, 2010.[308] K. Burke and A. M. C. Sengor. Ten metre global sea-level change associated with south atlantic aptiansalt deposition. Marine Geology, 83(1-4):309–312, 1988.[309] U. G. Wortmann and A. Paytan. Rapid variability of seawater chemistry over the past 130 millionyears. Science, 337(6092):334–336, 2012.[310] T. J. Bralower and H. R. Thierstein. Low productivity and slow deep-water circulation in mid-cretaceous oceans. Geology, 12(10):614–618, 1984.[311] James M. Russell, Hendrik Vogel, Bronwen L. Konecky, et al. Glacial forcing of central indonesianhydroclimate since 60,000 y bp. Proceedings of the National Academy of Sciences of the United States ofAmerica, 111(14):5100–5105, 2014.[312] E. Viollier, P. W. Inglett, K. Hunter, A. N. Roychoudhury, and P. Van Cappellen. The ferrozine methodrevisited: Fe(ii)/fe(iii) determination in natural waters. Applied Geochemistry, 15(6):785–790, 2000.140[313] S. Katsev, S. A. Crowe, A. Mucci, et al. Mixing and its effects on biogeochemistry in the persistentlystratified, deep, tropical lake matano, indonesia. Limnology and Oceanography, 55(2):763–776, 2010.141Appendix AChapter 3: supplementary materialA.1 Cr isotope modelTo estimate the time-scale over which our whole core incubations reached steady state, we estimated thetime for Cr(VI) to reach the reactive zone and penetrate into it via diffusion. To estimate the diffusionaltransit time t, we utilized the Einstein-Smoluchowski relationship;t = L2/(2 · D) (A.1)Where L is the depth to the reactive zone (2 mm) as determined in Lake Matano and D is the diffusioncoefficient adjusted for the salinity, temperature and porosity of our site in Lake Matano (2.8 x 10 – 5 cm2 s– 1).0 50 100 150 200 250[O2] (µmol l-1)-0.3-0.2- (mm)Figure A.1: Measured pore water oxygen profiles. A depth of 0 mm marks the sediment water interface.This calculation yields an estimated diffusion time of 23 minutes, suggesting our whole core incubationsreached steady-state on timescales shorter than our sampling interval, as well as the time-scale over whichsignificant boundary conditions would change.To model Cr-isotope fractionation in an open system, we setup a diagenetic reactive-transport model inwhich the profiles of the individual Cr-isotopes were described independently. The steady-state modelwas calculated for various time points, based on the changing Cr(VI) concentration at the sediment waterinterface. The model was adapted from the approach of [283] and [143]. The model predicts Cr(VI) andCr(III) distributions under steady-state conditions by describing changes in Cr(VI) concentrations with142depth as a function of rates of diffusional transport and reduction of Cr(VI) to Cr(III) as:Dzd2[Cr(VI)]dx2− RCr = 0 (A.2)where Dz is the site specific sediment diffusion coefficient (2.8 x 10 – 5 m2 d– 1) for Cr(VI) [194], x is thedepth in the sediment (m), and RCr is the Cr(VI) reduction rate (µmol m– 3 d– 1). In the absence of datadescribing Cr(VI) reduction rates, we set these rates of Cr(VI) reduction (RCr) in the model with an explicitlyspecified depth distribution function to generate modelled Cr(VI) profiles. Cr(VI) reduction rates were setfor a normal distribution:RCr = Rmax · 1σ√2pie−12 ((x−µ)σ )2(A.3)where σ is the variance (0.0015 m), x-µ is depth relative to the centre of the reactive zone, and Rmax is aconstant that determines the maximum rate of Cr(VI) reduction. We set Rmax at each time point in orderto reproduce the integrated rates obtained in our incubations. This depth distribution is based on theassumption that rates of Cr(VI) reduction are limited by electron donor concentrations as depths above themaximum Cr(VI) reduction and limited by [Cr(VI)] below this same depth. Furthermore, Figure A.2 below,displays two idealized concentration profiles for Cr(VI) and Fe(II), which when applied in a rate law similarto the one determined experimentally by [58], result in a Gaussian distribution of Cr(VI) reduction rates.0 10 20 30 40 50 60 70 80Concentration (nmol l-1)0.0000.0010.0020.0030.0040.0050.006Depth (m)[Cr(VI)][Fe(II)]0.00 0.05 0.10 0.15 0.20 0.25Rate (µmol l-1 m-3 d-1)0.0000.0010.0020.0030.0040.0050.006Depth (m)Figure A.2: Hypothetical rate distribution determined from Cr(VI) and Fe(II) concentration profiles. Hypotheticalrate distribution determined from Cr(VI) and Fe(II) concentration profilesGradients in Cr(VI) concentration were computed by integrating the Cr(VI) reduction rates, and concen-trations were then computed through integration of the gradients. Specific rates of 53Cr(VI) and 52Cr(VI)143reduction and concentrations of 53Cr(VI) and 52Cr(VI) are related by a fractionation factor (α):R53Cr = (R52Cr · [53Cr])/(α · [52Cr]) (A.4)At fractionations of between 1.0 and 5.0h, the rate of 52Cr reduction (R52Cr) 0.8982 · RCr to within onepercent, so R52Cr was calculated as 0.8982 · RCr. The factor of 0.8982 represents the abundance of thelight isotopes of Cr. For our calculations, an initial approximation of R53Cr was thus taken as 0.1018 · RCr.The differential equations determining steady-state values of [53Cr] and R53Cr, were solved numerically.The model was highly sensitive to the initial R53Cr, and model convergence was only possible within anarrow range of initial R53Cr values - this sensitivity to gradient boundary conditions is common in suchmodels [283]. The model requires a specified value of αint, however because this is the unknown parameterto be fit to the measured data, and we used a trial and error approach until we achieved convergence.The values for [53Cr] and R53Cr were obtained iteratively until the model converged on an αint within ±<0.00001 of the specified αint value. For example, if the specified α was 1.0018, convergence was deemedacceptable at a modelled α of between 1.01801 and 1.01799. In other words, modelled fractionations wereaccurate to within ± 0.01h δ53Cr. The model was run with α values of 1.0010, 1.0015, 1.0018, 1.0020 and1.0025, and using the measured Tube D [Cr(VI)] and δ53Cr composition at each time point, as boundaryconditions (Table 3.1, Chapter 3). The fit of the modelled δ53Cr(VI) values to the measured δ53Cr(VI) valueswas assessed visually, and of the αint values tested, a fractionation factor αint of 1.0018 ± 0.0004, appearedto fit the data best (Fig. 3.7, Chapter 3).The δ53Cr(III) produced was calculated by subtracting the intrinsic fractionation ((α -1) · 1000h) ata given depth from the δ53Cr(VI) at the same depth. The sedimentary δ53Cr values, averaged over thesediment column, for the Cr(III) exported to the sediments were calculated by integrating δ53Cr(VI) · RCrover depth. Figure A.3 below shows a complete model output example.1440 1,000 2,000 3,000 4,000 5,000Cr Reduction Rates (µM / m3 d)0.0000.0010.0020.0030.0040.005Depth (m)Time Point 2Time Point 3Time Point 4Time Point 5Time Point 6Time Point 7Time Point 80 500 1,000 1,500 2,000 2,500[Cr] (µM / m3)0.0 0.5 1.0 1.5 2.0 2.5δ53Cr(VI) (per mil)0.0000.0010.0020.0030.0040.005Depth (m)-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0δ53Cr(III) (per mil)A BC DFigure A.3: Cr model output example. Displayed is the model output for our reactive transport model forwhole core incubation with boundary conditions similar to the time points of Tube D. The model runsare steady-state simulations with boundary conditions set to approximate the Cr(VI) and δ53Cr(VI) of theoverlying water at 8 time points. A) Shows the Cr(VI) reduction rates as a function of depth for each timepoint. B) Shows the change in [Cr(VI)] as it diffuses into the sediment and is reduced. C) Displays theevolution of the porewater δ53Cr(VI) pool. D) Displays the evolution of the product δ53Cr(III) pool.A.1.1 Oxygen modelTo calculate changes in concentration of dissolved oxygen below the sediment water interface, we utilize amodel for organic carbon diagenesis developed by [193].DO2d2O2dx2−ω dO2dx− 1.3FKcGc = 0 (A.5)where x is the depth below the water-sediment interface; ω sedimentation rate; F is bulk dry density (1.7g cm– 3) divided by the porosity (90%); GC is the concentration of solid organic carbon (in moles C perunit dry sediment): KC is the first order rate constant of organic carbon decomposition and CO2 is theconcentration of dissolved O2. We parameterize KC based on the following [204, 284];KC = 1.8ω1.47 (A.6)We note that this parameterization of the reactivity of organic matter does not include the quality of organicmatter being delivered to the sediment water interface, however both [284] and [200], found this relationshipto be acceptable based on modelling of interstitial nutrient depth profiles. We then solve equation (A.5)using the approach presented in [194], and a number of permutations of the different boundary conditionsto calculate CO2 as a function of depth (x) in the sediment (Fig. 3.9, Chpater 3). Table A.1 details the rangeof boundary conditions employed in the model for CO2, ω, and KC.145Table A.1: A range of boundary conditions used in the oxygen modelBottom Water O2 (µmol l-1) Sedimentation Rate (cm yr-1) KC (cm yr-1) Porosity (%) Diffusion (cm2 s-1) 250.00 5.00 19.1800 90.00 0.0000126 150.00 0.50 0.6500   50.00 0.05 0.0220   5.00 0.005 0.0007   2.50     1.25     	146Appendix BChapter 4: supplemental materialB.1 Geological settingB.1.1 Cismon drill coreWe worked with samples from the Cismon APTICORE, a reference section for the Barremian-Aptian intervalat low latitudes [220, 223, 285]. The site was located on the continental margin of the Mesozoic Tethys, onthe eastward deepening slope between the Trento Plateau and the Belluno Basin [223]. Lithologically theOAE1a level is characterized by marlstone alternating with black shales and discrete radiolarian beds [285].We targeted our Fe-speciation analyses on intervals also studied in [223].B.1.2 DSDP site 463DSDP Site 463 was drilled at a water depth of 2525 m in the ancient structural high of the westernMid-Pacific Mountains (21◦01’N, 174◦40.07’E) [223]. During the Early Cretaceous, Site 463 was locatedat a paleo-latitude of ∼20◦S with a paleo-depth of ∼1 km [286]. The OAE1a occurs at Site 463 between∼626 - 615 mbsf, corresponding to ∼12 m of tuffaceous limestone [287] containing a number of discreteorganic-rich horizons with TOC up to 7.5 (wt%) [225, 288].B.2 MethodsB.2.1 Fe-speciation analysisRock samples from both Cismon drill core and DSDP Site 463 were crushed using an agate mill to avoidmetal contamination, and crushed sediment samples were then further milled to fine powders using anagate hand mortar and pestle. Sample masses of 500 mg of sediment were weighed into 50 ml centrifugetubes, and subjected to the Fe-speciation sequential extraction scheme based on [129] and performed asindicated in Table B.1.147Table B.1: Description of revised Fe-speciation leach procedure. †[153] ‡[289] ?[129] [290]Fraction Designation SEDIMENT FRACTION EXTRACTANT Sequential Digests Fecarb Reactive Fe in carbonate and poorly crystalline phases 0.5 M HCl, 1 h †	 FeOM Reactive Fe in organic matter 0.02 M HNO3 / 30% H2O2 adjusted to pH 2 ‡	 FeGoe Reactive Fe in goethite and hematite 0.35 M acetic acid / 0.2 M Na-citrate Na-dithionite, 2 h 	 FeMagnetite Reactive Fe in magnetite 0.2 M ammonium oxalate / 0.17 M oxalic acid solution pH 2, 6 h 	 FeSilicate Unreactive Fe in silicates Near boiling 6 N HCl, 24 h 	 FePyr Reactive Fe in pyrite Concentrated HNO3, 2 h ◊	 	Our “highly reactive, FeHR” pool is defined as carbonate-associated Fe (FeCarb, 0.5 N HCl extractable Fe),ferric (oxyhydr)oxides including magnetite (FeOxides, sum of dithionite and oxalate extractable Fe, FeGoeand FeMagnetite, Table B.1), organic matter associated Fe (FeOM) and pyrite (FePyr) (FeHR = FeCarb + FeOxides+ FeOM + FePyr). The FeTot pool is defined as the sum of all FeHR pools and Fe contained in silicate minerals(FeSil). We note that organic matter associated Fe (FeOM) is not included in the calibrated extractionscheme of [129]. We have included this pool to gain more information on authigenic Fe deposited duringOAE1a and have included this in our calculation of FeHR to emphasize reactive Fe pools that would pyritizein the presence of H2S (Table C.3 and Table B.3).148Table B.2: Cismon Core Fe-speciation. Revised extraction scheme.Stratigraphic Interval (m) Fecarb (wt%) FeOM (wt%) FeOxide (wt%) FePyr (wt%) FeTot (wt%) FeHR/FeTot FePyr/FeHR Fe/Al 11.9 0.00 0.02 0.13 0.01 0.65 0.25 0.05 0.57 13.2 0.02 0.02 0.23 0.02 1.28 0.22 0.06 0.53 15.8 0.03 0.02 0.18 0.01 1.09 0.23 0.04 0.61 16.8 0.02 0.02 0.24 0.02 1.27 0.24 0.07 0.63 17.4 0.04 0.02 0.12 0.00 0.69 0.26 0.01 0.73 18.7 0.04 0.03 0.11 0.02 0.49 0.40 0.10 0.75 18.9 0.05 0.02 0.19 0.03 0.87 0.32 0.09 0.74 19.3 0.05 0.42 1.62 0.57 3.80 0.70 0.21 1.24 19.5 0.04 0.12 0.64 0.07 2.28 0.38 0.09 0.58 19.9 0.05 0.53 1.36 0.40 3.38 0.69 0.17 1.13 20.3 0.07 1.55 0.31 0.08 3.48 0.58 0.04 0.51 21.6 0.00 0.06 0.63 0.07 1.81 0.42 0.10 0.63 21.9 0.03 0.03 0.58 0.10 1.76 0.42 0.13 0.67 22.4 0.02 0.03 1.00 0.18 1.82 0.67 0.14 1.03 22.5 0.05 0.62 1.05 0.53 2.94 0.76 0.24 1.32 23.1 0.06 1.20 1.01 0.21 3.39 0.73 0.09 1.44 23.1 0.05 0.28 0.65 0.03 2.75 0.37 0.03 0.60 23.2 0.04 0.03 0.52 0.12 1.70 0.41 0.17 0.68 23.3 0.05 0.96 0.33 0.01 2.48 0.54 0.01 0.68 23.4 0.04 1.17 0.59 0.06 2.97 0.63 0.03 0.82 23.5 0.00 1.69 0.42 0.04 3.46 0.62 0.02 0.77 23.5 0.05 0.39 0.75 0.29 2.06 0.72 0.20 1.00 23.5 0.04 0.05 0.80 0.13 1.48 0.70 0.13 1.01 23.7 0.03 0.02 0.15 0.01 1.34 0.16 0.03 0.47 24.2 0.03 0.03 0.13 0.00 0.71 0.26 0.01 0.68 25.1 0.02 0.03 0.11 0.00 0.76 0.22 0.02 0.65 26.2 0.03 0.05 0.09 0.00 0.63 0.29 0.00 0.74 27.6 0.07 0.06 0.05 0.00 0.36 0.53 0.01 0.82 29.4 0.04 0.04 0.24 0.03 2.16 0.16 0.08 0.50 31.3 0.03 0.04 0.10 0.00 0.47 0.36 0.00 0.80 32.4 0.00 0.02 0.14 0.01 1.57 0.11 0.04 0.55 	149Table B.3: DSDP Site 463 Fe-speciation. BLD stands for “below detection”. NA stands for “not analyzed”.Revised extraction scheme.Stratigraphic Interval (m) Fecarb (wt%) FeOM (wt%) FeOxide (wt%) FePyr (wt%) FeTot (wt%) FeHR/FeTot FePyr/FeHR Fe/Al 547.08 BLD NA 0.35 0.01 1.84 0.19 0.03 0.37 549.29 BLD 0.01 0.14 0.00 0.45 0.34 0.01 0.49 557.01 BLD 0.02 0.12 0.00 0.29 0.46 0.00 0.62 557.28 BLD 0.04 0.06 0.00 0.23 0.39 0.01 0.55 559.04 BLD 0.05 0.07 0.00 0.36 0.30 0.01 0.49 559.41 BLD 0.02 0.17 0.00 0.59 0.32 0.00 0.52 559.69 BLD 0.02 0.07 0.00 0.39 0.21 0.01 0.42 567.57 BLD 0.03 0.10 0.00 0.44 0.27 0.01 0.42 576.55 BLD 0.03 0.10 0.00 0.34 0.37 0.02 0.54 587.025 BLD 0.05 0.15 0.00 0.45 0.40 0.00 0.76 595.05 0.04 0.12 0.22 0.00 0.48 0.59 0.00 0.85 607.105 0.01 0.15 0.23 0.02 0.63 0.52 0.04 0.59 607.58 0.36 0.07 3.92 0.03 6.09 0.67 0.01 0.94 614.755 0.05 0.08 0.20 0.00 2.68 0.12 0.01 0.3 616.23 0.06 0.78 0.23 0.01 3.96 0.22 0.01 0.31 616.88 0.34 NA 2.15 0.13 4.60 0.53 0.05 0.41 617.885 0.03 0.12 0.11 0.01 0.75 0.30 0.03 0.38 618.81 0.03 1.67 0.27 0.03 1.18 0.70 0.01 0.84 622.59 0.05 0.02 0.22 0.01 2.60 0.12 0.05 0.28 622.71 0.24 0.04 0.28 0.02 2.02 0.26 0.04 0.31 622.84 0.35 0.07 0.80 0.03 3.03 0.36 0.02 0.37 623.19 0.04 0.01 0.18 0.01 2.73 0.09 0.04 0.28 623.45 0.17 0.13 0.66 0.03 4.02 0.23 0.03 0.38 623.62 0.04 0.00 0.12 0.00 1.21 0.13 0.01 0.34 623.77 0.07 0.01 0.12 0.00 1.12 0.17 0.01 0.33 623.985 0.03 0.02 0.28 0.01 2.16 0.15 0.02 0.34 624.565 0.05 0.01 0.12 0.00 1.13 0.16 0.02 0.31 624.905 0.07 0.02 0.10 0.00 1.29 0.14 0.01 0.28 625.085 0.12 0.44 2.38 0.01 4.23 0.62 0.00 0.85 625.32 0.14 0.72 2.99 0.03 5.09 0.65 0.01 0.94 625.49 0.01 0.02 0.39 0.01 1.39 0.31 0.02 0.5 625.7 0.05 0.11 0.13 0.01 0.63 0.37 0.03 0.45 626.49 0.00 0.02 0.74 0.00 2.81 0.27 0.01 1.97 627.175 0.03 0.05 0.12 0.00 0.41 0.41 0.01 0.51 627.575 0.01 0.02 0.17 0.00 1.02 0.19 0.01 0.56 628.335 0.03 0.07 0.12 0.00 0.51 0.36 0.01 0.51 628.585 0.01 0.02 0.33 0.00 1.88 0.19 0.01 0.36 635.4 0.02 0.02 0.45 0.00 1.41 0.34 0.00 0.41 643.535 0.02 0.04 0.34 0.01 1.16 0.33 0.01 0.5 	150We note however, our detection of an anoxic FeHR/FeTot signal above a threshold of 0.38 during OAE1ais not dependant on the inclusion of this leach. Indeed, for DSDP Site 463 sediments, FeHR values withand without the OM fraction are not significantly different at the 95% confidence level using a two-tailedstudent t-test. Furthermore, Fe/Al ratios of the OM leach average 4.4 during OAE1a, clearly indicating anauthigenic source of Fe in this leach fraction as Fe/Al ratios are well above the detrital composition of theupper continental crust of 0.62 [291].We also have performed the Fe-speciation extraction using the canonical protocol [129] and arrive atthe same conclusion of anoxic ferruginous conditions during OAE1a (Table B.4 and B.5).Table B.4: Cismon Core Fe-speciation. Canonical extraction. BLD stands for “below detection”. NA standsfor “not analyzed”.Stratigraphic Interval (m) FeCarb (wt%) FeOxides (wt%) FeMag (wt%) FeTot (wt%) FeOM (wt%) FeHR/FeTot Mo (ppm) Cr (ppm) V (ppm) 11.9 0.10 0.05 0.02 0.48 0.01 0.27 BLD 34.0 18.7 13.2 0.09 0.06 0.04 0.96 0.12 0.24 NA   15.8 0.12 0.06 0.05 0.79 0.06 0.27 BLD 92.7 34.5 16.8 0.11 0.08 0.06 0.91 0.10 0.27 BLD 122.9 48.9 17.4 NA NA NA NA NA NA NA   18.7 NA NA NA NA NA NA NA   18.9 0.10 0.07 0.05 0.54 0.09 0.36 BLD 48.7 18.0 19.3 0.28 0.13 0.09 0.89 1.20 0.66 53.9 216.2 211.3 19.5 0.14 0.10 0.09 1.07 0.59 0.46 12.8 214.6 116.7 19.9 0.27 0.14 0.10 0.99 1.08 0.62 18.6 321.1 311.7 20.3 0.18 0.13 0.08 0.86 0.81 0.58 18.6 325.4 144.1 21.6 0.10 0.11 0.08 1.07 0.50 0.42 BLD 185.3 102.4 21.9 0.09 0.08 0.08 0.97 0.33 0.37 BLD 136.3 58.4 22.4 0.20 0.11 0.06 0.55 0.91 0.70 NA    22.5 NA NA NA NA NA NA NA    23.1 NA NA NA NA NA NA 13.9 149.0 193.1 23.1 0.15 0.11 0.09 1.39 0.23 0.29 BLD 231.4 76.0 23.2 0.13 0.11 0.07 1.03 0.23 0.35 BLD 165.6 63.8 23.3 0.16 0.14 0.07 1.20 0.62 0.45 BLD 121.0 169.7 23.4 0.19 0.12 0.07 1.27 0.76 0.47 BLD 104.0 97.4 23.5 0.18 0.12 0.07 1.26 0.82 0.49 10.6 113.3 96.5 23.5 0.18 0.10 0.09 0.58 0.67 0.64 BLD 69.9 72.7 23.5 0.20 0.12 0.04 0.43 0.72 0.71 BLD 57.4 48.1 23.7 0.13 0.13 0.03 0.58 0.06 0.37 BLD 57.5 23.9 24.2 0.10 0.07 0.03 0.52 0.06 0.33 BLD 62.5 26.2 25.1 0.11 0.07 0.02 0.40 0.07 0.41 BLD 40.5 14.6 26.2 0.12 0.05 0.01 0.25 0.03 0.48 BLD 26.0 7.5 27.6 NA NA NA NA NA NA NA   29.4 NA NA NA NA NA NA 9.0 22.3 5.5 31.3 NA NA NA NA NA NA NA   32.4 0.11 0.13 0.05 1.17 0.12 0.25 BLD 115.3 47.0 	151Table B.5: DSDP site 463 Fe-speciation. Canonical extraction. BLD stands for “below detection”. NA standsfor “not analyzed”. FeOM and FePyr from previous extraction.Stratigraphic Interval (m) FeCarb (wt%) FeOxides (wt%) FeMag (wt%) FePyr (wt%) FeOM (wt%) FeSil (wt%) FeHR/FeTot 547.08 0.00 0.28 0.10 0.01 NA 1.66 0.19 549.29 0.02 0.12 0.02 0.00 0.01 0.44 0.29 557.01 NA NA NA NA NA NA NA 557.28 0.04 0.07 0.01 0.00 0.02 0.17 0.45 559.04 0.07 0.06 0.02 0.00 0.02 0.23 0.42 559.41 NA NA NA NA NA NA NA 559.69 0.04 0.05 0.02 0.00 0.01 0.30 0.29 567.57 NA NA NA NA NA NA NA 576.55 0.03 0.08 0.03 0.00 0.02 0.26 0.39 587.025 NA NA NA NA NA NA NA 595.05 0.21 0.14 0.06 0.00 0.06 0.22 0.68 607.105 NA NA NA NA NA NA NA 607.58 0.65 0.53 0.55 0.03 0.03 2.63 0.41 614.755 0.03 0.15 0.04 0.00 0.04 1.43 0.15 616.23 0.15 0.20 0.03 0.01 0.39 1.30 0.37 616.88 1.03 0.52 0.66 0.13 NA 2.61 0.47 617.885 0.06 0.09 0.03 0.01 0.06 0.82 0.23 618.81 0.11 0.15 0.03 0.03 0.84 1.10 0.51 622.59 0.02 0.18 0.05 0.01 0.01 2.47 0.10 622.71 0.18 0.37 0.10 0.02 0.02 2.15 0.24 622.84 0.34 0.47 0.27 0.03 0.03 2.32 0.33 623.19 0.01 0.16 0.05 0.01 0.00 2.50 0.09 623.45 NA NA NA NA NA NA NA 623.62 0.02 0.12 0.03 0.00 0.00 1.00 0.15 623.77 0.07 0.13 0.03 0.00 0.01 0.98 0.19 623.985 0.02 0.21 0.09 0.01 0.01 2.02 0.14 624.565 0.03 0.12 0.02 0.00 0.01 1.17 0.14 624.905 0.07 0.13 0.03 0.00 0.01 0.83 0.22 625.085 0.74 0.38 0.60 0.01 0.22 2.21 0.47 625.32 0.86 0.42 0.68 0.03 0.36 2.34 0.50 625.49 0.12 0.16 0.18 0.01 0.01 1.05 0.31 625.7 0.18 0.13 0.02 0.01 0.05 0.63 0.39 626.49 0.78 0.27 0.78 0.00 0.01 1.17 0.61 627.175 0.14 0.10 0.06 0.00 0.03 0.43 0.44 627.575 0.08 0.15 0.22 0.00 0.01 0.89 0.34 628.335 0.11 0.06 0.06 0.00 0.03 0.36 0.43 628.585 0.08 0.15 0.18 0.00 0.01 1.09 0.28 635.4 0.19 0.18 0.21 0.00 0.01 1.07 0.36 643.535 0.16 0.14 0.12 0.01 0.02 0.76 0.37 	 152At the end of this extraction we also tested the importance of including the OM leach. Following thecanonical Fe-speciation extraction, we subjected the residual sediment to the OM leach, observing that theFeHR pool is underestimated by 30% if this leach is not included in the extraction scheme (Fig. B.1, TableB.1, Table B.4).0 10,000 20,000 30,000 40,000[Fe] (ug g-1) Depth (m)Total Fe - Determined via FusionTotal Fe - no OMTotal Fe - including OM0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0FeHR/FeTot0. HR0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7FeOx (wt%) +FeOx)0.0 0.2 0.4 0.6 0.8 1.0FeHR/FeTot S/Fe HRFigure B.1: Results of OM leach test. After performing the canonical Fe-speciation protocol [129] (blue data)we observe that failure to include the OM leach (red data) under estimates the total Fe pool by 30% inorganic rich intervals.Fe concentrations of our leachates were measured by both flame atomic adsorption spectroscopy(Flame AAS) and inductively coupled plasma optical emission spectroscopy (ICP-OES). For our flame AASmeasurements precision on triplicate measurements was 1.2% and our limit of detection in solution was80 µg l– 1, or roughly 35 µg g– 1 Fe based on our dilutions. For our ICP-OES measurements precision ontriplicate measurements for Fe was 2.2% and our limit of detection in solution was 6 µg l– 1, or roughly30 µg g– 1 based on our dilutions. For Al analysis via ICP-OES, precision on triplicate measurements was1.2% and our limit of detection in solution was 6 µg l– 1, or roughly 33 µg g– 1 based on our dilutions. Ourextractions dissolved > 92% of the Fe from the PACS-2 international reference standard.B.3 Evaluating Fe-speciationB.3.1 Dilution of authigenic FeTo directly test if our Fe-speciation values were influenced by high sedimentation rates, we measured theFe/Al ratios of our sequential extractions (Fig. 4.2 Main Text, Table C.3, B.3), as high sedimentation rateswill cause a dilution of the FeHR pool [118, 182]. We tested whether samples with FeHR/FeTot values > 0.38in OAE1a sediments recorded in both Cismon and DSDP Site 463 also have Fe/Al > 0.5, and indeed allsamples that are above this anoxic threshold also pass this Fe/Al cut-off. To further test if the FeHR in oursamples is authigenic, we also compared Fe/Al values recorded during OAE1a to the composition of thecontinental crust, as a proxy for the detrital Fe/Al ratios. This yields a more stringent cut-off of Fe/Al =0.62 [291]. Sediments from Cismon have Fe/Al compositions consistently near or above the detrital baselineof 0.62 [291] during OAE1a, whereas samples that bound the event are near or below this threshold (Fig.1534.1 Main Text). At DSDP Site 463, sediments with FeHR/FeTot values above 0.38, also have Fe/Al valuesabove 0.62 (Fig. 4.2, Chapter 3).B.4 Possible post depositional shale alterationB.4.1 Oxidation of reactive Fe(II)Although we took care to work with well-preserved rocks, we assessed the extent to which our samplesmay have been affected by oxidative weathering of authigenic pyrite. Prior to analyses the majority ofshale samples were olive greenish grey with no evidence of Fe oxide staining, suggesting negligible postdepositional oxidation. We nevertheless considered the possibility of oxidation in more detail. It should benoted that pyrite and FeCarb minerals, such as siderite, are the dominant FeHR minerals that contain Fe(II)and may undergo oxidation. In both cases, however, oxidation will not impact FeHR/FeTot or Fe/Al ratios,and hence any interpretation of anoxia remains robust as the oxidation of FeHR(II) would transfer this Fe tothe FeOx pool, and hence this would not influence diagnosis of anoxia based on Fe speciation. Much of theFeHR is preserved as siderite (Table B.4 and B.5), which itself is susceptible to oxidation as well as pyrite.This observation also argues strongly against pyrite oxidation and implies that even the FeOx is a primarycomponent of the FeHR pool.Post depositional oxidation of pyrite, however, could decrease FePyr/FeHR with potential to obscurediagnosis of deposition under euxinic water column conditions. Pyrite preserved in the Cismon and DSDPsite 463 sediments is pristine and framboidal, which is entirely inconsistent with pervasive post depositionaloxidation (Fig. 4.2, Chapter 3). Pyrite oxidation would effectively transfer Fe from the pyrite pool (FePyr) tothe oxide pool (FeOx) and redistribute S into non-pyrite pools. In the unlikely scenario that all of the FeOxpool was the result of pyrite oxidation, summing FePyr and FeOx as a proxy for total FePyr yields typicalFePyr/FeHR of < 0.7, which is still below the degree of pyritization that develops under euxinic depositionalconditions (Fig. 4.2, Chapter 3). Even if total S (including organic-S) is used in place of FePyr, however, thesediments still have far less S relative to FeHR than expected following deposition under euxinic conditions(Fig. B.2).Figure B.2: STot/FeHR vs. FeHR/FeTot for Cismon samples deposited during OAE1a. Substituting total S for pyriteS does not change our interpretation of ferruginous conditions during OAE1a.154Only a fraction of the highly reactive Fe, furthermore, exists in the oxide pool, and much of the highlyreactive Fe is present as siderite (Table B.4 and B.5), which is itself highly reactive towards oxygen. Sideritepreservation thus implies negligible post depositional oxidation.As an additional test for secondary alteration through weathering, we plotted Fe-speciation as a functionof the chemical index of alteration CIA [292];CIA = ([Al])/([Al] + [Ca] + [Na]) · 100% (B.1)where element concentrations of Al, Na and Ca are taken from the FeSil leachate (see Table B.1). Forthe Cismon samples we observe no relationship between CIA and FeHR/FeTot (Fig. B.3), indicating ourFe-speciation results are not linked to the degree of chemical alteration of our samples.0 10 20 30 40 50 60 70 80 90 100CIA (%) Totf=0.00101562X+0.372951 r2 = 0.0156839Figure B.3: FeHR/FeTot vs. CIA for the Cismon core. FeHR/FeTot vs. CIA for the Cismon core.B.5 Idealized 1D water column reaction transport modelThis model was developed and described in detail in [293] and an abbreviated description is providedhere. The model explores how sulphate levels during OAE1a influenced rates of pyrite deposition andthe fractionation of sulphur isotopes in a stratified Cretaceous ocean water column. The model for theocean water column predicts sulphate distributions under steady-state conditions by describing changes insulphate with depth as a function of vertical transport and bacterial reduction of sulphate to sulphide as:Kzd2[SO2−4 ]dx2− RSR = 0 (B.2)where Kz is the vertical mixing coefficient, x is the depth in the water column, and RSR is the sulphatereduction rate. Sulphate reduction rates were calculated with a Michaelis-Menten description of sulphate155reduction kinetics:RSR =Vmax[SO2−4 ]Km + [SO2−4 ](B.3)where Km is the half-saturation constant, which was taken as 5 µM for low sulphate marine environments[294]. Higher values for Km generally lead to residual sulphate in deep waters and are thus incompatiblewith the Fe speciation data and ferruginous depositional conditions (Fig 4.3, Chapter 3). Vmax is themaximum rate of sulphate reduction when sulphate supply is unlimited and thus corresponds to thescenario when sulphate reduction is limited by organic matter availability. Vmax can therefore be estimatedbased on models of organic matter degradation rates in modern marine systems. As a starting point wethus parameterized Vmax , according to carbon degradation rates in the modern ocean [237]. At these ratesfor Vmax (1 - 10 mmol m– 2 yr– 1), however, sulphate reduction rates lead to pyrite deposition rates muchhigher than those recorded in the Cismon sediments. We recognized, though, that sulphate reduction inferruginous environments, is only 1 - 10% of total respiration [295]. Scaling modern marine Vmax by afactor of 10 to 100 then yields pyrite burial rates equivalent to those recorded in Cismon sediments (average315 ± 300 µmol m– 2 yr– 1, complete range 15 – 1200 µmol m– 2 yr– 1). Model outputs were thus tightlytethered to the geological record through pyrite burial fluxes and Fe speciation. We note, that incompletereduction of sulphate, for example as a result of inhibition by alternative electron acceptors such as nitrate,or cryptic sulphur cycling with sulphide re-oxidation coupled to nitrate reduction, would be incompatiblewith the burial of significant amounts of organic matter. Both processes would induce sulphate reductionin sediments and drive further pyritization of FeHR, implying development of ferruginous conditions alsorequires low water column nitrate. We also considered an end member scenario where minimal sulphatereduction occurs in the water column and sulphate reduction occurs in the sediment (see Section 6 below).To gauge the sensitivity of our model to the value of Kz, we performed a sensitivity analysis over arange of values for this parameter. OAE1a is characterized by decreased latitudinal temperature gradientsand sluggish ocean circulation [296], implying a slower than modern vertical mixing coefficient that isconsistent with a more strongly stratified ocean. We thus varied Kz from 0.01 – 0.25 m2 d– 1, of which thelowest value is a factor of 10 smaller than the Kz in surface waters of the modern day Black Sea [297] andthe highest value is a factor of 10 smaller than modern ocean upwelling zones [298]. The results of oursensitivity analysis can be seen in (Fig. B.4), showing that the smallest value of Kz that yields the largestestimate of seawater [SO2 –4 ] is 0.1 m2 d– 1.156Figure B.4: 1D water column reaction transport sensitivity analysis results. The contours represent the parameterspace required to achieve quantitative water column reduction of the initial seawater sulphate concentration(y-axis) while also reproducing the pyrite burial rates observed at the Cismon site.We note that this is very similar to the Kz (0.13 m2 d– 1) required to balance oxygen consumption in modernhypoxic ocean waters (O2 < 10 µM, [299]).The volume specific rates of sulphate reduction rendered by our model at low µM sulphate concen-trations are also similar to those in modern ferruginous environments like Lake Matano, Indonesia andKabuno Bay in East Africa, illustrating that such rates are well within the physiological capacity of modernsulphate reducing bacteria. To compare our modelled RSR to modern cell specific sulphate reduction rates(csSRR), we assume a coastal shelf water column cell density of 5 x 10 – 5 cells ml– 1 [300]. We furtherassume that sulphate reducers comprise about 10% of the total microbial community in the upper reachesof ferruginous water columns, similar to observed abundances in modern low sulphate ferruginous watercolumns [293]. Based on our range of modelled RSR we calculate csSRR values of 1.2 x 10 – 8 to 2.4 x 10 – 9 lcell– 1 d– 1.We also used our model to test the capacity of sulphate reduction to generate the pyrite S isotopiccompositions observed in the Cismon rocks. Sulphate concentrations as a function of depth in themodeled water column were computed by integrating the sulphate reduction rates, and concentrationswere then computed through integration of the gradients. Specific rates of 34SO2 –4 and32SO2 –4 reductionand concentrations of 34SO2 –4 and32SO2 –4 are related by a fractionation factor (α):α =R32SR[34SO2−4 ]R34SR[32SO2−4 ](B.4)157The fractionation factor can also be more easily written as epsilon (e);e = (α− 1) · 1000h (B.5)At fractionations of between 10 and 80h, the rate of 32SO2 –4 reduction (R32SR) 0.96RSR to within one percent,so R32SR was calculated as 0.96RSR. For our calculations, an initial approximation of R34SR was thus taken as0.04RSR. The differential equations required to determine steady-state values of [34SO2 –4 ] and R34SR, weresolved numerically. The model was highly sensitive to the initial R34SR, and model convergence was onlypossible within a narrow range of initial R34SR values. The values for [34SO2 –4 ] and R34SR were obtainediteratively until the model converged on an α within ± <0.0001 of the specified value. For example, if thespecified α were 1.020, convergence was deemed acceptable at a modelled α of between 1.0199 and 1.0201.In other words, modelled fractionations were accurate to within ± 0.1h δ34S.To explore how our modelled RSR affects the S-isotope composition of sulphide exported as pyrite insulphate limited water columns, we imposed a fractionation factor of e = 70h, which is the maximumequilibrium fractionation between sulphate and sulphide during sulphate reduction by bacteria [17]. Weused this large fractionation factor based on the observation that the isotopic offset between seawater andpyrite during OAE1a is ∼65h (∆34SSeawater – pyrite = ∼65h, [212]). Taking initial δ34Sseawater = 18h [212],as indicated from analyses of carbonate associated sulphate that bracket OAE 1a sediments, our depthintegrated δ34Spyrite reproduces the measured Cismon δ34Spyrite = -47h [212], with an imposed S-isotopefractionation factor of e = 70h and a seawater sulphate concentration of 50 µM (Fig. B.5).Figure B.5: 1D water column reaction transport isotope results. We plot the isotopic offset between seawatersulphate (δ34Sseawater = 18h) and sulphide formed in the water column using an imposed fractionationfactor of 70h, as a function of seawater sulphate concentration. The green diamond distribution plot at leftillustrates sedimentary ∆34Ssulphate – sulphide calculated from bulk Cismon pyrites [212] and using 18h as aconservative value for δ34S of seawater sulphate. The dashed lines delineate values between the 25 th and75 th percentiles, whereas the solid lines delineate the 5th and 95th percentiles, encompassing 90% of theCismon pyrite data.158We note, modelling work presented in [301] and [302] demonstrates that it is possible to expressthis degree is S-isotope fractionation between the sulphate and sulphide pool within the range of ourmodelled csSRR (∼2 x 10 – 9 l cell– 1 d– 1) as discussed above. The results of our isotope modelling, however,imply a slightly higher upper limit on seawater sulphate concentration than that permitted by the pyriteburial fluxes, and thus the strong fractionations apparent from the Cismon pyrites may reflect sulphurdisproportionation, which is expected under ferruginous conditions [234, 303, 304].B.6 Diagenetic modelSimulations of sulphate reduction in anoxic sediment were performed with a simplified one-dimensionalreaction-transport model(Fakhraee, et al. 2017). Vertical distributions of dissolved chemical species withinthe sediment were obtained by solving the diagenetic equation at steady state:0 = Diφd2Cidx2+∑jRij (B.6)Here x is depth below the sediment surface, Ci is the concentration of species i (sulphate or sulphide), Di isthe molecular diffusion coefficient, and Rij are rates of reactions (Tables B.6 and B.7). Advection due toburial was neglected for solutes because its time scale is much longer than that of diffusion. For simplicity,the porewater concentration of dissolved Fe2+ was imposed in the model to increase into the sediment inproportion to the square root of depth x, in resemblance to the typical profiles in modern low-sulphateenvironments. As the rate of sulphate reduction is linked to the rate of carbon mineralization RC (TableB.6), the latter was calculated as a power law of the organic carbon age t at depth x, as detailed in [305].Table B.6: Reactions in sediment diagenetic model.	Reaction Rate expression Sulphide Precipitation Rpyr = kpyr[H2S][Fe2+] Sulphate Reduction !!" = 0.5!![!"!!!]!! + [!"!!!] 159Table B.7: Parameters of diagenetic model.	Parameter Symbol Value Unit Depth of model domain L 10-50 cm OC content at interface Corg 3 % OC age at interface tinit 1-10 yr Sulphide precipitation rate constant kpyr 0.1 (µM yr)-1 Monod constant for SO42- reduction Km 5-200 µM Max. Fe2+ concentration at L [Fe2+]max 5-50 µM Burial Velocity ν 0.15 mm/yr Porosity ϕ 0.85 - The boundary conditions for solutes were imposed as fixed-concentrations at the sediment-water interface(assuming zero for hydrogen sulphide) and zero-flux at the bottom of the model domain. The porewaterdistributions were obtained by solving the boundary-value problem in MATLAB. The depth-integrated rateof pyrite precipitation was then calculated by integrating the rate of sulphide precipitation Rpyr over depth,while verifying that the precipitation rate at the bottom of the domain was negligible. The diagenetic modelresults can be seen in Table B.8.Table B.8: Diagenetic model results. Shaded model runs represent pyrite burial fluxes consistent with theCismon core.[SO42-] at the Sediment Water Interface (µM) Diagenetic Pyrite Burial (µmol m-2 yr-1) Km (µM) 100 7178 5 50 6033 5 25 3700 5 10 1600 5 5 818 5 2 331 5 1 166 5 	160B.7 Evaporite deposition modelTo investigate the sensitivity of sulphate draw down during evaporite mineral formation to seawatersulphate concentrations, we performed a forward modelling exercise using the PHREEQC 3.0 computermodel. Ion activity coefficients were calculated using Pitzer’s ion-interaction model, which accountsfor the non-ideality of aqueous solutions. The simulations were performed using an estimate for thechemical composition of Aptian seawater [235], as input for the initial solution composition. We model thissolution in equilibrium with pCO2 = 500 ppm at 25◦C and impose no charge balance. The initial chemicalcomposition of our seawater solution is summarized in Table B.9.Table B.9: Chemical composition of Aptian seawater used for evaporation modelling [235] and miner-als included in the equilibrium assemblage. SO2 –4 and Ca2+ concentrations modelled over a range ofcompositions.Ion Composition (mg/L) Chloride (Cl-) 20029 Sodium (Na+) 9564 Sulphate (SO42-) 20 - 400 Magnesium (Mg2+) 1021 Calcium (Ca2+) 400 - 40,000 Potassium (K+) 430 Bicarbonate(HCO3-) 141 Strontium (Sr2+) 8 Bromide (Br-) 68 CO2 (gas) 500 ppm   Mineral  Formula anhydrite CaSO4 bischofite  MgCl2 · 6 H2O bloedite  Na2Mg(SO4)2 · 4 H2O calcite CaCO3 carnallite  KMgCl3 · 6 H2O epsomite  MgSO4 · 7 H2O gypsum  CaSO4 · 2 H2O halite  NaCl hexahydrate  MgSO4 · 6 H2O kieserite  MgSO4 · H2O polyhalite  K2MgCa2(SO4)4 · 2 H2O sylvite  KCl 	To simulate sulphate mineral formation, we performed an evaporation experiment by varying the initialseawater composition with respect to SO2 –4 and Ca2+ over a range of concentrations (Table B.9). In a set ofsequential steps, PHREEQC was constrained to reduce the original water mass of the seawater by 10% in thefirst evaporation step, and then reduce the resulting new water mass by 10% in the second evaporation step,and so on while allowing mineral precipitation if saturation was reached. After 9 evaporation steps, 90% ofthe original water mass was depleted without replenishment. During evaporation, PHREEQC calculatesseveral brine parameters including the mineral saturation indices (SI), ionic evolution, and number ofmoles of precipitated mineral salts. Specifically, we tracked over what range of initial seawater SO2 –4 andCa2+ concentrations; supersaturation of the mineral gypsum was possible, as a function of the degree ofevaporation. The modelling does not take into account kinetic factors such as those caused by day-nighttemperature changes and the time to reach equilibrium, which affect the crystallization path. Furthermore,the modelling does not provide information on the fate of trace elements in the mineral salts. The results ofour modelling can be seen in Figure B.6.161Ca2+ Concentrations Required to Saturate Gypsum (SI>0)1 2 3 4 5 6 7 8 9 10Seawater [SO42-] (mM)4045505560657075808590Degree of Evaporation (%)20406080100120140160180200Seawater [Ca2+ ](mM)Figure B.6: Results of gypsum supersaturation model.Our simulation shows that over the range of initial seawater SO2 –4 and Ca2+ concentrations tested, lowseawater SO2 –4 concentrations are a major inhibitor to gypsum precipitation. Below initial SO2 –4 concentra-tions of 3-5 mM, gypsum precipitation requires anomalously high Ca2+ concentrations, and or extremelyhigh degrees of seawater evaporation. Given previous estimates for Aptian seawater [Ca2+] [235, 306], it isunlikely that continued evaporite deposition was capable of drawing seawater SO2 –4 concentrations downbelow 1 mM during OAE1a.B.8 Marine sulphur budgets and sulphate drawdownGiven that evaporation alone cannot draw sulphate concentrations down below 1 mM (section 7 above),we tested the capacity of sulphate reduction to draw down the sulphate reservoir to the low µM levelsimplied by our reaction transport and diagenetic models (section 5 and 6 above). To track the mass andisotopic composition of marine sulphate we employed a sulphur box model similar to models constructedby previous workers [221, 307] using the following equations;d(Ms)dt= Fw + Fh(Fpyr + Fevap + FOM + FCAS) (B.7)δ34Ssulphatedt=((Fwδw + Fhδh)δ34Ssulphate · (Fw + Fh)Fpyr · ∆34S)Ms(B.8)where MS is the mass of sulphate in the ocean; Fh, Fw, are the hydrothermal, and weathering input fluxesof S, respectively; Fpyr , FOM, FCAS and Fevap are the burial fluxes of pyrite, organic sulphur, carbonateassociated sulphate and evaporites, respectively. δ34Ssulphate is the S isotope composition of seawatersulphate; δw and δh are the S-isotope composition of the weathering and hydrothermal inputs respectively,162∆34S is the average isotope fractionation factor associated with pyrite deposition. The initial steady state forthe S cycle was determined assuming that the magnitude and isotopic composition of the input fluxes (bothhydrothermal and weathering) during the Early Cretaceous were comparable to the modern (Table B.10).Table B.10: Box model S-fluxes and isotopic compositions. Model results are plotted in (Fig. 4.4) of theMain Text. †[221] ‡[213] ?[212] Flux Symbol Value (Gmol yr-1) Isotopic Composition Relative Timing (Myr) Reference Pre-OAE1a Steady State Hydrothermal Fh 0.53 3.2‰ NA †	 Weathering Fw 1.03 5.2‰ NA †	 Evaporite Burial Fevap 0.52 ∆34S = 0‰ NA This Study Pyrite Burial Fpyr 0.66 ∆34S  = -26‰ NA This Study CAS FCAS 0.03 ∆34S  = 0‰ NA This Study Organic Sulphur FOM 0.35 ∆34S  = 0‰ NA This Study Maximum During Model Hydrothermal Fh 2.30 3.2‰ 0.5 †	 Evaporite Burial Fevap 1.12 ∆34S  = 0‰ 4.0 ‡	 Pyrite Burial Fpyr 1.31 ∆34S  = -62‰ 0.3 	 Organic Sulphur FOM 1.89 ∆34S  = 0‰ 1.0 This Study Minimum During Model Run Evaporite Burial Fevap 0 ∆34S  = 0‰ 1.2 This Study Pyrite Burial Fpyr 0.53 ∆34S = -26‰ 0.5 This Study 	Furthermore, the model assumes an ocean volume of 1.38 x 10 18 m3 [308] and an initial sulphate concentra-tion of 2.5 mM (as opposed to modern levels of 28 mM), in accordance with Cretaceous estimates based onthe chemical composition of fluid inclusions encased in halite and previous modelling work [213, 235, 309].We set the weathering flux of sulphate to the oceans, Fw = 1.035 x 10 12 mol yr– 1 and the hydrothermalflux, Fh = 5.29 x 10 11 mol yr– 1 [221]. Initial pyrite burials rates were set at, Fpyr = 6.58 x 10 11 mol yr– 1.We applied some additional complexity in our model by assuming that, Fevap = 5.26 x 10 11 mol yr– 1,is not the only non-redox dependant S-burial flux at steady-state, and also included the S-burial fluxescarbonate associated sulphate (FCAS) and biomass associated organic sulphur, FOM, as additional non-redoxdependant burial fluxes. We estimate the steady-state CAS burial flux using the sulphate concentration ofCAS in the Cismon core [212] and an estimate for global carbonate burial [19], Fcarb = 2.99 x 10 10 mol yr– 1.We estimate the steady-state biomass associated organic S-burial flux using a previously published estimatefor modern global primary productivity in ocean surface waters (48 Gt C yr– 1) [237]. Applying a typicalmolar C:S ratio of 50:1 for marine biomass [247], we calculate a maximum generation of biomass associatedorganic S of 7.0 x 10 13 mol yr– 1. Assuming only a small fraction (0.5%) of this biomass S becomes buried inmarine sediments globally, FOM = 3.5 x 10 11 mol yr– 1. The modelled S fluxes are tabulated in Table B.10.We used our model to examine the timescales on which the expansion of ocean anoxia and ensuingsulphate reduction is able to draw seawater sulphate down to low µM sulphate predicted from our Fespeciation and pyrite burial flux data for the Aptian oceans, while also reconciling existing isotope data[212, 221, 236]. From steady-state, we force our model by increasing the burial of evaporites for ∼5 Myrprior to OAE1a [213] while also applying a modest increase in the pyrite burials rates (Fig. 4.4, Chapter4). Pyrite burial rates during this pre-OAE1a phase are consistent with an increase in the extent of oceananoxia from 0.1 to 2.5%. Once seawater sulphate concentrations drop below 1 mM, the evaporite burial fluxceases (see section 6 above), and we require further seawater sulphate drawdown during a brief interval(∼300 kyr) of euxinia prior to the onset of OAE1a and the negative C-isotope excursion (Fig. 4.4, Chapter4). Peak pyrite burial rates during this brief interval are consistent with an expansion of ocean anoxia to4% with ensuring pyrite burial rates comparable to those observed in the modern day Peru Margin and163Chilean oxygen minimum zone (OMZ) (∼0.1 mol m– 2 yr– 1) [239, 240] (Fig. B.7).0.20 0.25 0.30 0.35 0.40 0.45Pyrite Burial Rates (mol m-2 yr-1)0100200300400500Frequency0.0 0.5 1.0 1.5 2.0 2.5Pyrite Burial Rates (mol m-2 yr-1)10,0005,00005,00010,00015,000FrequencyCariaco BasinBlack Sea0.0 0.1 0.2 0.3 0.4 0.5Pyrite Burial Rates (mol m-2 yr-1)5,00005,00010,000FrequencyPeru MarginChilean OMZFigure B.7: Area specific modern marine pyrite burial rates statistics. 20,000 bootstrapped resampling of themean pyrite burial rates for the Peru Margin [240], Chilean OMZ [239], Cariaco Basin [133] and the BlackSea [110, 132]. The bottom panel includes data from all 4 environments.An expansion of ocean anoxia and pyrite burial during this pre-OAE1a interval is also consistent withexisting S-isotope records (Fig. 4.4 and Table B.11) and the onset of biological crises in the Aptian oceans[218].Table B.11: Range of coupled δ34SCAS - δ34Spyr data for OAE1a. †[212]Section ∆34S (‰) Reference Cismon   Mean 62.3 †	 STD ± 4.9 †	 Permanente   Mean 36.0 †	 STD ± 12.3 †	 	164Previous workers observe a coupling between the Aptian 87Sr/86Sr and δ34Ssulphate records, hypothe-sized to result from strong hydrothermal fluxes at the initiation of OAE1a and thus at the onset of OAE1awe apply the hydrothermal forcing described in [221]. Models of hydrothermal fluid chemistry, predictcomplete removal of H2S from the hydrothermal fluid when seawater is sulphate free versus 28 mM[242]. Given the much lower than modern Cretaceous seawater sulphate concentrations predicted by ourmodels, we cannot rule out the possibility that the S/Sr composition of hydrothermal vent fluids was muchlower than the modern, and the resulting hydrothermal efflux of S during OAE1a was muted relative tomodern. The burial of S, therefore, would vary in proportion to the size of the incoming hydrothermalflux. Regardless of the hydrothermal vent fluid composition, in order for seawater sulphate concentrationto remain low during the ferruginous conditions of OAE1a, we must balance hydrothermal S-input byintroducing an additional sink for sulphate; biomass associated organic S burial (∆34S = 0h, [249, 250]),during which the sulphate reservoir stabilizes below 50 µM for the entirety of the ferruginous interval,with a minimum recorded value of ∼4 µM. Our modelled increase in the FOM burial flux is consistentwith the up to ∼35-fold increase in the burial of organic matter during OAE1a, and is also consistent withOM concentrations in the Cismon sediments as well as other sedimentary sections that contain OAE1a[223]. Our model, therefore, shows that it is possible to draw seawater sulphate concentrations down andmaintain low µM concentrations of sulphate during OAE1a, due to an initial sulphate sequestration eventwith pyrite burial rates observed in modern day oxygen minimum zones [239, 240] and an increase inorganic S-burial during OAE1a itself. A model sensitivity analyses can bee seen in (Fig. B.8).16520253035404550556065Model Time (Myr)1,0001,5002,0002,5003,000Sulphate (uM)101214161820222426δ34 Ssulphate (‰)20253035404550556065Model Time (Myr) (Gmol yr-1) Evaporite_1Organic-S_1Pyrite_1S-input20253035404550556065Model Time (Myr)1,5002,0002,5003,0003,5004,0004,5005,000Sulphate (uM)101214161820222426δ34 Ssulphate (‰)20253035404550556065Model Time (Myr) (Gmol yr-1)20253035404550556065Model Time (Myr)05001,0001,5002,0002,5003,000Sulphate (uM)121416182022δ34 Ssulphate (‰)20253035404550556065Model Time (Myr) (Gmol yr-1)20253035404550556065Model Time (Myr)05001,0001,5002,0002,5003,000Sulphate (uM)101520253035δ34 Ssulphate (‰)20253035404550556065Model Time (Myr) (Gmol yr-1)a bc de fg hFigure B.8: Box model sensitivity analysis. Grey lines represent modelled seawater sulphate concentrations.Blue lines represent modelled δ34Ssulphate. Green data points are δ34SCAS from (Mills, et al. 2017). (a) 4.2 xFh for 0.5 Myr followed by 1.5 x Fh for 5 Myr, 2.35 x Fevap for 4 Myr followed by no evaporite burial for1 Myr, 4.5 x FOM for 1 Myr. (b) Modelled results prescribed in (a). (c) 4.2 x Fh for 0.5 Myr followed by1.5 x Fh for 5 Myr, 2.35 x Fevap for 4 Myr followed by no evaporite burial for 1 Myr. (d) Modelled resultsprescribed in (c). (e) 4.2 x Fh for 0.5 Myr followed by 1.5 x Fh for 5 Myr, 4.5 x FOM for 1 Myr, 1.4 x Fpyr for3 Myr followed by 2 x Fpyr for 0.3 Myr, FOM for 1 Myr. (f) Modelled results prescribed in (c). (g) 4.2 x Fhfor 0.5 Myr followed by 1.5 x Fh for 5 Myr, Increase to 2 x Fpyr over 10 Myr. (h) Modelled results prescribedin (g).Together our models yield a range of estimates for seawater sulphate concentration (between 5 and 50µM). At a seawater sulphate concentration of ∼4 µM and peak S-fluxes in our box model, we calculatethe residence time of sulphate to be comparable to modern ocean mixing time (∼3000 years) and thusheterogeneity in the ocean with respect to sulphate concentration would be expected, especially giventhe more sluggish than modern ocean circulation and deep water renewal rates that characterized theCretaceous [310].166Appendix CChapter 5: supplemental materialC.1 MethodsC.1.1 Sample collectionSediment trap sampling was performed in May and June 2015. Sediment traps consisting of four corebarrels (8 cm in diameter, 57 cm tall) were deployed at the specified depths in each lake, and their timeof installation recorded. Sedimentation rates in LM and LT are 80 and 19 cm ky– 1 respectively [137, 311].During recovery, half of the sediment traps at each depth (2/4) were filtered on board using a peristalticpump onto glass fibre filters (0.2 µM) and preserved in a 15 ml falcon tube containing 5 ml 0.5 N HClimmediately. The material in the remaining two traps was quantitatively vacuumed pumped along withthe overlying water into 12 ml exetainers without leaving headspace.Short coring as well as water column geochemical sampling was performed in both lakes in years 2014and 2015. Water temperature, oxygen concentration, light fluorescence reemitted by chlorophyll a and lighttransmission profiles were collected on site using a submersible conductivity-temperature-depth probe(CTD; Sea-Bird, SBE-19; Sea-Bird Electronics, Bellevue, WA, USA). All water samples were collected with 5L Go-Flow (Niskin; General Oceanics, Miami, FL, USA) bottles attached in series to a stainless steel cableand a hand-operated winch. The bottles were placed at depth to an accuracy of ± 1 m with the help of acommercial fish finder (Furuno, FCV 585; Furuno Electric Co., Nishinomiya, Japan). We obtained pH in thefield via portable pH meter by homogenizing 2 ml of sediment in 2 ml of deionized water and measuringthe supernatant after 2 min. Several sediment short cores (< 0.5 m) were retrieved from both lakes using agravity corer. Sediment sampling took place at a water depth of 200 m in Lake Towuti, and a water depthof 200 m in Lake Matano. The sites in both lakes are overlain by anoxic and Fe(II)-rich (ferruginous) water.Short cores were sectioned in a N2 flushed glove bag at a resolution of 0.5, 1, and 2 cm resolution forthe upper 1, 1–10, and below 10 cm, respectively. A sub sample of 0.5 g of sediment from each interval wasimmediately extracted in 1 ml 0.5 N HCl, and Fe-speciation of the easily extractable phases was measuredspectrophotmetrically on site using the ferrozine assay [153, 312]. The residual sediment from each sectionwas preserved in N2 flushed falcon tubes and sealed in N2 flushed aluminium foil bags.Short core sediment Fe-speciation measurements were performed on anaerobically preserved sedimentsamples following the method of [129] at the University of British Columbia. Sample masses of 100 - 200 mgof sediment were weighed into 15 ml centrifuge tubes, and the sequential extraction scheme was followedprecisely as indicated in [129], only substituting 0.5 M HCl in place of the hydroxylamine hydrochlorideleach [153] (Table C.1).167Table C.1: Description of Fe-speciation extractions. †[129] ‡[153]Operationally defined Fe – mineral phases Extractant Symbol Siderite 1 M Na-acetate pH 4.5, 24 h †	 FeAca Lepidocrocite, Ferrihydrite 0.5 M HCl, 1 h ‡	 FeHCl Goethite, Hematite 0.35 M acetic acid/0.2 M Na-citrate Na-dithionite, 2 h †	 FeDith Magnetite 0.2 M ammonium oxalate/0.17 M oxalic acid, 6 h †	 FeOxa Silicate Fe Near boiling 6 M HCl, 24 h †	 Fe6N 	Fe-speciation was conducted on filtered sediment trap material by applying each extraction directly to thefilter paper contained in a 15 ml centrifuge tube (Table C.2).168Table C.2: Water column Fe-speciation results. Towuti 2015 110 m Towuti 2015 160 m Phase % of FeTot % of FeTot NR 10 6 HR 90 94 Fe(II) 14 18 Fe(III) 77 77 FeAca 7 5 FeHCl(II) 4 11 FeHCl(III) 39 32 FeDith 33 41 FeOxa 2 2 FeOxa(III) 5 4    FeNon Magnetic 94 74 FeMagnetic 6 25 FeNon Magnetic Oxalate Extractable 0.5 1    Total Fe (μg) 436 740  Matano 90 m Matano 130 m Phase % of FeTot % of FeTot NR 7 6 HR 93 94 Fe(II) 29 41 Fe(III) 64 53 FeAca 24 23 FeHCl(II) 1 15 FeHCl(III) 17 8 FeDith 41 39 FeOxa 3 3 FeOxa(III) 6 6    FeNonMagnetic 99 96 FeMagnetic 0 4 FeNonMagnetic Oxalate Extractable 1 1    Total Fe (μg) 525 948 	All Fe concentration measurements were performed using a Flame Atomic Absorption Spectrophotometer(Flame AAS). Precision on triplicate measurements was 1.2% and our limit of detection in solution was 0.08µg ml– 1. Our extractions dissolved > 92% of the Fe from the PACS-2 international reference standard.C.1.2 Confirming the selectivity of oxalate extractable FeTo test and confirm the selectivity of the oxalate extraction for magnetite, we first measured the magneticsusceptibility (MS) of sediment samples from the deep basin of LM and LT, before and after the oxalateextraction using a handheld magnetic susceptibility meter (Exploranium, Kappameter KT-9). After removal169of crystalline Fe-phases (hematite, goethite) via dithionite treatment, but prior to the oxalate leach, sedimentsamples (∼1 g) have an initial MS of (0.07), relative to a background of (-0.01). After the oxalate extractionis applied the sediments do not retain any magnetic field and return to baseline (-0.01).As a further and more detailed of the selectivity of the oxalate extraction for magnetite, we leached LTsediments (1 g) deposited from ferruginous bottom waters at a depth of 200 m using a dithionite leachto dissolve crystalline Fe-(oxyhydr)oxides) while leaving magnetite and residual unreactive Fe-silicateminerals untouched [129]. The amount of dissolved Fe measured in our dithionite extractions is withinerror of the abundance of goethite as confirmed independently via quantitative XRD, verifying that thedithionite leach is highly selective for goethite (Table C.3).Table C.3: Results of qXRD phase analysis of LT deep sediments (wt.%)Mineral Ideal Formula Towuti deep sediment 0 - 0.5 cm  Fe  Fe (Dithionite leached) Towuti deep sediment 35 - 40 cm Fe Fe (Dithionite leached) Quartz SiO2 11   11   Serpentine Mg3Si2O5(OH)2 14   18   Actinolite Ca2(Mg,Fe2+)5Si8O22(OH)2 3   3   Albite (Na,Ca)(Al,Si)4O8 3   3   Clinochlore (Mg,Fe2+)5Al(Si3Al)O10(OH)8 15   16   Illite K0.65Al2.0Al0.65Si3.35O10(OH)2    2   Goethite α-FeO(OH) 9 6 5.1 7 4 4.4 Nontronite (Na,Ca)0.3(Fe23+)2(Si,Al)4O10(OH)2·nH2O 45 18  40 16  	After leaching the bulk sediment with dithionite, we separated magnetite grains from the residual sedimentusing neodymium magnets and confirmed via XRD that this magnetically purified phase is almostexclusively magnetite (Fig. C.1a, see section below).10 20 30 40 50 60 70 802 θ (degrees)Intensity10 20 30 40 50 60 70 802 θ (degrees)MMMMMMMNLLN LA BFigure C.1: XRD spectra of LT deep sediment (200 m). a) Magnetically extracted grains from LT core topsediment. b) Residual sediment after removal of magnetic mineral grains. M corresponds to magnetitepeaks, L corresponds to lizardite peaks and N corresponds to nontronite peaks.After magnetite separation, magnetite peaks are no longer detectable in the residual sediment via XRD (Fig.170C.1a). We then leached the residual magnetite-free residual sediment with oxalate, recovering only ∼0.8%of the total Fe in the sediment. In sum our tests confirm that in the Malili lakes system (see Main Text forwater column results), the oxalate extraction is highly selective for magnetite and other Fe-mineral phasesare not readily leached by this technique.C.1.3 XRD analysisWe analyzed LT sediment samples via quantitative X-ray diffraction (qXRD), to determine the mineralogyof the sediment (Table C.3). The samples were smear mounted with ethanol on non-diffracting silica plates.Continuous-scan X-ray powder-diffraction data were collected over a range 3-80◦2θ with CoKα radiationon a Bruker D8 Focus Bragg-Brentano diffractometer equipped with an Fe monochromator foil, 0.6 mm(0.3◦) divergence slit, incident- and diffracted-beam Soller slits and a LynxEye detector. The long fine-focusCo X-ray tube was operated at 35 kV and 40 mA, using a take-off angle of 6◦. We analyzed the X-raydiffractograms using the International Centre for Diffraction Database PDF-4 and Search-Match softwareby Bruker. X-ray powder-diffraction data of the samples were refined with Rietveld program Topas 4.2(Bruker AXS).C.1.4 Fe flux calculationsWater column and sediment Fe fluxes in LM and LT were determined by combining each lake’s sedimen-tation rate with the Fe-speciation results. In the water columns, area specific Fe fluxes were determinedby dividing the concentration of Fe captured by the sediment trap in each operationally defined mineralphase (mmol), by the area of the sediment trap (0.005 m2) and the deployment time (0.017 and 0.008 yearsin LM and LT respectively) to yield Fe fluxes in units of mmol m– 2 yr– 1. In the short core sediments areaspecific Fe fluxes were determined by multiplying volume specific Fe concentrations (mmol m– 3) by thesedimentation rates (0.0008 and 0.00019 m yr– 1 in LM and LT respectively) to yield sediment Fe fluxes ofmmol m– 2 yr– 1. Deep water diffusive Fe(II) gradients in each lake were determined by multiplying Fe(II)concentration gradients with diffusivity coefficients. Bottom water concentrations of Fe(II) in LM and LTare 0.140 and 0.010 mmol l-1 respectively (Fig. 5.1 Main Text), and these dissolved pools are quantitativelyoxidized at the chemocline (Fig. 5.1 Main Text), driving upward diffusive fluxes of Fe(II). We calculate Fe(II)gradients as 1.5 and 0.6 mM m– 4 in LM and LT respectively (Fig. 5.1 Main Text). Upward diffusional Fe(II)fluxes were estimated by multiplying the Fe(II) gradients with each lake’s eddy diffusivity coefficient (0.1and 0.6 m2 d– 1 in LM and LT respectively, [143, 313]), yielding Fe(II) fluxes of 54.8 and 131.4 mmol m2 yr– 1,respectively.C.1.5 Saturation state calculationsSaturation indices were calculated for the relevant mineral species as;SI = −log IAPKsp(C.1)where IAP is the ion activity product for the relevant mineral phase and Ksp is its corresponding solubilityproduct. Saturation indices greater than 0 indicate supersaturation, whereas those less than 0 indicateundersaturation with respect to a given mineral phase (Fig. 5.1, Main Text). Solubility products for allmineral phases were taken from the PHREEQC database. Water column dissolved inorganic carbon (DIC)concentrations were calculated based on charge balance using the concentration of all major ions in solution.Water column dissolved Fe(III) concentrations in LM and LT were calculated assuming both goethite andferrihydrite saturation.171C.1.6 SEM microscopyWe prepared magnetic separates from both the sediment traps and sediments of LM and LT for SEMimaging. Anaerobically preserved water column and sediment samples were mounted on 12.5 mm SEMstubs and coated with 5 nm of Ir. We also treated a sub-set of anaerobically preserved water column andsediment samples with dithionite, to remove crystalline Fe-(oxyhydr)oxide mineral phases [129]. We thencarefully and as quantitatively as possible, separated magnetic grains using neodymium magnets. Themagnetic separates were then mounted onto 12.5 mm SEM stubs and coated them with 5 nm of Ir to ensureconductivity. All SEM images were obtained using a Helios FIB-SEM using magnetically purified sedimenttrap and sediment material.To verify the SEM-EDS analyses accurately differentiate magnetite from other Fe-oxide phases, weprepared two end-member Fe-mineral standards; magnetite and goethite. On each standard we collectedover 10 distinct EDS spots and compiled their Fe:O stoichiometries (wt%). We then performed bootstrapresampling of the mean Fe:O compositions for these standard minerals (Fig. C.2).Figure C.2: SEM-EDS statistical results. Plotted are histograms of the bootstrap resampled mean values ofthe Fe:O composition for the mineral standards and water column and sediment magnetite framboids.At the 95% confidence interval, goethite and magnetite are easily distinguishable based on our EDS results(p value = 0.0135). We also obtained EDS spectra on framboids collected in the water column and sedimentsof LM and LT (n = 17) and performed bootstrapped resampling of their mean Fe:O values (Fig. C.2). Atthe 95% confidence interval, the Malili lake framboid composition and the magnetite mineral standardare not statistically different based on our EDS results (p value = 0.786). Consistent with the dearth ofsulphur in the Maili lakes system, furthermore, we did not detect sulphur in any of our EDS analyses,further supporting a water column origin and ruling out diagenetic pyrite oxidation as a mechanism formagnetite framboid formation [266].To our knowledge, the Malili lakes are the only known modern ferruginous environment to host a wellpreserved framboidal magnetite population. In the anoxic water column, magnetite framboids recoveredfrom the sediment traps are smaller (< 10 µm in diameter, Fig. 5.3 Main Text) and their surface crystallitesare more anhedral relative to their sedimentary counterparts (Fig. 5.3 Main Text and Fig. C.3).172Close up framboids Matano2 um3 um2 um2 um2 um2 uma b cd e fFigure C.3: Detailed photomicrographs of framboidal magnetite surfaces. a) Nannoscale magnetite octahedra,LM sediment. b) Nannoscale magnetite octahedra, LM sediment. c) Nannoscale magnetite octahedra,LM sediment. d) Nannoscale magnetite octahedra, LM anoxic sediment trap. e) Nannoscale magnetiteoctahedra, LM sediment. f) Triangular forms of magentite, LM anoxic sediment trap.The water column framboids surfaces show triangular crystals, indicative of ocatahedron, which may be inearly growth stages. The sedimentary population of magnetite in both lakes display a wide range of size(20 – 50 mum in diameter) (Fig C.3, Fig. C.4 and Fig. C.5).17320 um10 um10 um30 um20 um10 um10 um6 um30 um50 umMatano Sedimenta b cd e fg h iFigure C.4: Framboidal forms of magnetite in LM sediment.Towuti Sediment2 um5 um10 um5 um10 um4 uma b cd e fFigure C.5: Authigenic magnetite in LT sediment. a) pristine octahedral magnetite crystals. b) Framboidalcluster. c) Framboid with triangular and octahedral surface crystals. d) Small framboid nucleus. e)Framboid. f) Clusters of nannoscale octahedral crystals embedded in an unknown matrix.174The framboids have pristine euhedral octahedral crystals with well-defined faces at miller indices of 111,110. The framboidal morphology of magnetite stands in contrast to detrital morphologies observed in thesediments (Fig. C.6).Detrital Sediment grains20 um10 um10 um50 um5 um10 uma b cd e fFigure C.6: Detrital magnetite morphologies. a) Octahedral crystal with pervasive dissolution pits. b) Largerounded subhedral crystals with pervasive dissolution pits, grooves and surface etchings. c) Large roundedeuhedral octaherdral crystal with surface etchings. d) Large euhedral octaherdral crystal with surfaceetchings. e) Cracked and broken euhedral crystal with pervasive surface dissolution pits. f) Large roundedeuhedral octaherdral crystal with surface etchings.C.1.7 Raman microscopyMagnetite grains were magnetically separated from LM sediment and samples were mounted in epoxyand polished using micro-diamond. The polished samples were then subjected to Raman spectroscopicanalysis using a Horiba Ltd. XploRa Plus µ-Raman spectrometer at the Department of Earth, Ocean andAtmospheric Sciences, University of British Columbia (UBC), Vancouver B.C., Canada. Analyses were doneusing a green laser (λ = 532 nm), which provided a power at the sample surface of 2.5 mW. Analyses weredone using a 100x objective focused down to a spot ∼1 µm in diameter. Spectral slit width was set at 100µm and the confocal hole was kept at 300 µm. Data were collected during three cycles of 30 s to optimizethe full width at half maximum of resolved Raman bands, while minimizing possible effects of heating oroxidation.175


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items