Cu Isotopes in Marine Black Shales Record the Great Oxidation Event

1 Cu isotopes in marine black shales record the Great Oxidation Event

3 Ernest Chi Fru1*, Nathalie Pérez Rodríguez1, Camille A. Partin2, Stefan Lalonde3, Per

4 Andersson4, Dominik Weiss5, Abderrazak El Abani6, Ilia Rodushkin7, Kurt Konhauser8.

6 1Stockholm University, Department of Geological Sciences and Bolin Centre for Climate

7 Research, SE-10691, Stockholm, Sweden

8 2University of Saskatchewan, Department of Geological Sciences, 114 Science Place,

9 Saskatoon, SK S7N 5E2, Canada.

10 3CNRS–UMR 6538 Laboratoire Domaines Océaniques, Institut Universitaire Européen de la

11 Mer, Université de Bretagne Occidentale, Place Nicolas Copernic, Plouzané 29280 France.

12 4Swedish Museum of Natural History, Department of Geosciences, Box 50 007

13 SE-104 05 Stockholm, Sweden.

14 5Imperial College, Department of Earth Science and Engineering, Royal School of Mines,

15 London SW7 2BP, UK.

16 6Université de Poitiers UMR 7285-CNRS, Institut de Chimie des Milieux et Matériaux de

17 Poitiers - 5, rue Albert Turpin (Bât B35) 86073 Poitiers cedex.

18 7ALS Laboratory Group. ALS Scandinavia AB. Aurorum 10. S-977 75 Luleå, Sweden.

19 8Department of Earth and Atmospheric Sciences, University of Alberta, Edmonton, Alberta

20 T6G 2EG, Canada.

22 *Corresponding author: [email protected]

1 25 Abstract

26 The oxygenation of the atmosphere ~2.45–2.32 billion years ago (Ga) is one of the most

27 significant geological events to have affected Earth’s redox history. Our understanding of the

28 timing and processes surrounding this key transition is largely dependent on the development

29 of redox-sensitive elemental proxies, many of which remain unexplored. Here we report a

65 30 shift from negative to positive copper isotopic composition (δ CuERM-AE633) in organic carbon–

31 rich shales spanning the Great Oxidation Event (GOE), after which they maintain positive

32 values comparable to modern shales. The data cover the period 2.66 to 2.08 Ga, and suggest

33 that prior to 2.3 Ga, preferential removal of 65Cu by Fe(III)(oxyhydr)oxides left seawater and

34 marine biomass depleted in 65Cu but enriched in 63Cu. As banded iron formation (BIF)

35 deposition waned, biomass sampled a dissolved Cu reservoir that was progressively less

36 fractionated relative to the continental pool. Consequently, δ65Cu values ≤0 disappear from

37 the shale record after ~2.3 Ga, when many proxies converge on permanent atmospheric

38 oxygenation. This shift towards heavy δ65Cu values is traceable through Phanerozoic shales to

39 modern marine settings, where marine dissolved and sedimentary δ65Cu are universally

56 40 positive, coinciding with a shift to negative sedimentary δ Fe values and increased marine

41 sulfate after the GOE. The evolving δ65Cu composition is suggested to provide a more

42 straightforward explanation for ferruginous versus sulfidic conditions than iron isotope

43 compositions during the GOE transition. The permanent shift in sedimentary Cu isotope

44 compositions provides new insights into Precambrian marine cycling of this critical

45 micronutrient.

2 48 Significance statement

49 Redox-sensitive transition metals and their isotopes provide some of the best lines of evidence

50 for reconstructing early Earth’s oxygenation history, especially related to permanent atmo-

51 spheric oxygenation after ~2.4 Ga. We show a shift from dominantly negative to permanently

52 positive copper isotope compositions in black shales spanning the height of banded iron for-

53 mation deposition, the Great Oxidation Event, and the shift from ferruginous to more oxic and

54 euxinic seawater conditions with enhanced oceanic sulfate, ~2.66-2.08 Ga. This shift is re-

55 flected in black shale iron isotope compositions, but copper isotopes appear to provide a better

56 sensitivity towards distinguishing iron- vs. sulfide-rich marine conditions in the sedimentary

57 record. Our results provide new insights into copper bioavailability coupled to Earth’s oxy-

58 genation history.

3 61 /body

62 Introduction

63 Shortly after the Archean-Paleoproterozoic boundary, the Great Oxidation Event (GOE)

64 ~2.45-2.32 billion years ago (Ga) marks the first permanent accumulation of oxygen in Earth-

65 ’s atmosphere (1-4). Evidence for the GOE comes from a convergence of proxies, including

66 the evolution of redox–sensitive element concentrations and their isotopic compositions in

67 marine sedimentary rocks (see refs 2 & 4 for reviews). Accompanying the GOE was an in-

68 crease in the oxidative weathering of a relatively untapped continental sulfide reservoir (1-12)

69 and the onset of near–continental margin and intracratonic basin sulfide-rich (euxinic) water

70 column conditions caused by bacterial sulfate reduction of an increasing oceanic sulfate pool

71 (4, 6-12). These events are reflected in the transition from Neoarchean oceans characterized

72 by abundant iron(III)(oxyhydr)oxide precipitation and the deposition of banded iron forma-

73 tions (BIF), to relative BIF scarcity and increased global marine pyrite burial for much of the

74 Paleoproterozoic (6-17).

75 Light iron (Fe) isotope enrichments in sedimentary pyrite older than 2.4 Ga records

76 oceans strongly affected by BIF deposition and dissimilatory Fe(III) reduction, while general-

77 ly near-zero to positive Fe isotope compositions between 2.3-1.8 Ga reflect near-complete

78 draw-down of the Fe reservoir by enhanced oxidation and/or marine pyrite burial (12, 18).

79 The lack of substantial δ56Fe variations in black shale pyrite after 1.5 Ga coincides with

80 oceans becoming even less Fe-rich due to waning submarine hydrothermal activity (12, 13,

81 16, 17). To provide a new and complementary perspective on water column redox evolution

82 surrounding the GOE, we turned to sedimentary copper (Cu) isotope compositions, known to

83 respond to redox shifts tuned to Fe and sulfur (S) cycling. This relationship predicts that the

84 preferential sequestration of 65Cu by Fe(III)(oxyhydr)oxides (e.g., 19, 20) precipitated from

85 ferruginous Archean oceans may have enriched seawater in residual 63Cu. Conversely, oxida-

4 86 tive continental weathering of sulfides, such as chalcocite (Cu2S) and chalcopyrite (CuFeS2),

87 should have increased the supply of dissolved Cu(II) and delivered more 65Cu–rich runoff to

88 the oceans (21-31) after the GOE. Oxidative dissolution of Cu(I) sulfide minerals to dissolved

65 89 Cu(II) entails Cu isotope fractionation between the reservoirs by up to 1.2-3.1‰ (∆ Cu(Cu(II)–

90 Cu(I) sulfide)) (30). Moreover, Cu isotope fractionation between surface/groundwater, secondary

91 Cu minerals, and primary Cu sulfide sources, all enrich the aqueous phase in 65Cu, yielding

92 fractionations of up to 12‰ between dissolved Cu and its source (30). Riverine supply of wa-

93 ters enriched in 65Cu, as the result of continental sulfide oxidation (1-17), should have influ-

94 enced the highly productive post-GOE continental margins and intracratonic basins. We can

95 thus hypothesize that the GOE should be accompanied by a transition to seawater that is en-

96 riched in 65Cu relative to continental sources, similar to the modern ocean (32).

97 We further hypothesize that before the GOE, massive Archean Fe(III)(oxyhydr)oxide

98 precipitation in the form of BIF preferentially buried 65Cu, enriching residual seawater in 63Cu.

99 This 63Cu, in turn, would have become incorporated into planktonic biomass that scavenged

100 Cu from seawater. After the GOE, increased oxidative weathering of a previously unweath-

101 ered continental sulfide pool (7) should have progressively delivered isotopically heavy river-

102 ine Cu(II) to the oceans as iron oxide deposition waned in the face of increasingly common

103 oxic or euxinic conditions. Collectively, this sequence of events predicts that sedimentary

104 δ65Cu values should have become increasingly positive across the GOE. To test this hypothe-

105 sis, we selected a suite of well-characterized, low metamorphic grade black shale samples de-

106 posited before, during, and after the GOE, spanning ~2.7-2.1 Ga. by other studies. The sample

107 suite spans the Transvaal Supergroup in South Africa and the Francevillian Series in Gabon,

108 both of which have been widely used for reconstructing key intervals in Earth’s oxygenation,

109 and their continental margin setting and depositional redox conditions have been previously

56 13 110 established (1, 9, 33, 34, 35). Our coeval δ Fe (redox sensitive proxy), δ Corg (biological ac-

5 111 tivity proxy) and δ65Cu records, spanning Late Archean iron-rich to Paleoproterozoic sul-

112 fide-rich marginal ocean conditions, provide new insight into the relative importance of fer-

113 ruginous vs. sulfidic conditions across the GOE as well as the Precambrian cycling of Cu, a

114 critical micronutrient.

116 Results and Discussion

117 Temporal changes in sedimentary Cu dynamics on the evolving Earth: Changes in the

118 sources and mechanisms of Cu delivery to Archean and Paleoproterozoic oceans would have

119 affected both biological activity and Cu isotopic composition. To interpret the δ65Cu data, we

120 first reconstruct temporal Cu sedimentary concentrations and the relative importance of iron

121 and sulfide sinks from the BIF and black shale archives (Fig. 1). The shale and BIF records of

122 authigenic Cu enrichment (Figs. 1A and 1B), normalized to detrital input using Ti (e.g., refs 7,

123 26, 36), are remarkably static, showing little significant change on either side of the GOE.

124 Authigenic Cu enrichment in BIF is ~1000 times higher relative to black shale (Fig. 1C). Af-

125 ter the GOE, BIF deposition decreased by approximately three orders of magnitude in terms

126 of tonnage (Fig. 1D), likely associated with declining hydrothermal activity and increasing

127 marine sulfate content resulting from enhanced oxidative continental sulfide weathering (11,

128 12, 37-40). Consequently, during peak BIF deposition, BIF likely exerted a stronger influence

129 on the concentration and isotope composition of dissolved Cu than black shale deposits.

130 Moreover, Cu/Ti molar ratios in black shale generally fall within an order of magnitude of the

131 upper continental crust (Fig. 1C), implying that black shales are not a strong Cu sink and

132 therefore not expected to significantly influence the Cu isotope budget to the same degree as

133 iron oxide-rich sediments. This fundamental shift in Cu exit channels likely contributed to the

134 Cu isotope evolution of seawater and pelagic sediments across the GOE (see below).

6 136 Temporal changes in sedimentary Cu isotopic compositions: Predominantly negative

137 δ65Cu values in ferruginous late Archean oceans increase from –0.55 to +0.74‰ by ~2.3 Ga,

138 after which δ65Cu values ≤0 disappear from our shale record (Fig 2A; Table 1; Fig. S1). This

139 irreversible upward stratigraphic change, which points to secular modification of seawater

140 chemistry from the Late Archean to the Paleoproterozoic, is significant at the 95% confidence

65 13 141 interval (ANOVA, p<0.05). δ Cu values appear to plot inversely to δ Corg with respect to

142 time (Fig. 2B), suggesting co-variance in Cu and C cycling in ancient oceans.

143 The preferential adsorption of dissolved 65Cu by Fe(III)(oxyhydr)oxides yields a

65 144 ∆ Cusolution-Fe(III)(oxyhydr)oxides fractionation of up to ~1‰ (19, 20). During late Archean peak BIF

145 deposition at ~2.48 Ga, ~4.5×1012 mol Fe/yr may have been removed by iron oxide sinks,

146 thereby also removing ~1×108 moles of Cu from the ocean annually (41). Residual

147 bioavailable Cu in the water column would have been left enriched in the lighter isotope 63Cu.

148 This appears reflected in the Neoarchean data, where black shales reach the lowest δ65Cu

149 values in our dataset. With the dramatic reduction of BIF deposition by a factor of 1000

150 following the GOE (Fig. 1D), a shift to heavy dissolved marine 65Cu enrichment appears

151 inevitable. Shale δ65Cu values (following seawater; see below) may readily be explained by

152 Raleigh distillation across important shifts in the proportion of marine Cu exiting via iron

153 oxide sinks (Fig. 3). Similar isotopic distillation of Cu occurs today in supergene weathering

154 environments (30). Indeed, starting from seawater with a modern Cu isotopic composition of

155 ~0.5‰, δ65Cu values as low as those observed between 2.66 to 2.44 Ga cannot be achieved

156 during adsorption to iron oxides in the absence of open-system behavior (Fig. 3)

157 Moving into the GOE, increasing sedimentary δ65Cu relative to the Neoarchean may

158 also be partially explained by an increasing supply of isotopically heavy Cu via continental

159 sulfide weathering (22-31) that became more important after 2.4–2.3 Ga, when the

160 atmosphere was permanently oxygenated. Abiotic and biotic oxidation of Cu minerals by

7 65 65 161 molecular oxygen produces a solution enriched in Cu, with a ∆ Cusolution-source mineral value of up

162 to ~3‰ (21-23, 27, 30). Cu isotope fractionation during precipitation of secondary Cu

163 minerals further enriches the aqueous phase in 65Cu, leading to a cumulative fractionation of

164 up to 12‰ between dissolved Cu and its source (30). While shifts in both source Cu isotopic

165 composition and the importance of iron oxide sinks may have influenced the sedimentary

166 δ65Cu record, the Cu enrichment record in both shales and BIF are relatively static (Fig. 1A,

167 1B), which indicates a more important role for the latter.

168 The adsorption and uptake of dissolved Cu by bacteria appears to favour the light

65 169 isotope at acidic pH. However, a negligible isotopic effect (i.e., ∆ Cusolution-phytoplankton≈0) is

170 consistently observed in the pH range of modern oceans (20, 32, 42). Although the reasons for

171 this are unclear, our data imply that the light Neoarchean δ65Cu values are best explained by

172 inheritance from a seawater component that was isotopically enriched in 63Cu due to ~200

173 million years of intense BIF deposition that preferentially removed 65Cu (Fig. 4A). This

174 isotopically light seawater Cu was probably passively adsorbed by biomass and preserved

175 alongside sedimentary organic matter since active biological Cu assimilation was likely

176 limited until after the GOE (43). After the GOE, the reduction in BIF deposition, coupled to

177 increased acid rock drainage on land related to the oxidation of sulfide minerals (7), changed

178 the Cu isotopic composition of seawater to permanently positive values.

13 179 In our sample suite, the increasing depletion of δ Corg with time is in agreement with

180 reports for marine sedimentary rocks during this period (4). For instance, the anomalously

13 181 light δ Corg values of the FD Formation in the Francevillian Series have previously been

182 linked to massive oxidation of organic matter and methanotrophy (9, 44, 45). The appearance

65 183 of this unique δ Corg at a time linked to an expanding sulfur reservoir (Fig. S3) associated

184 with microbial sulfate reduction (9, 46), suggests the existence of a potential anaerobic

185 methane oxidation pathway coupled to sulfate reduction. However, the absence of a

8 186 corresponding shift in the δ65Cu signal from that seen at the height of the GOE, ~2.3 Ga, when

65 187 δ Corg was much higher, suggests that biological processes responsible for generating the

65 65 188 distribution of δ Corg through time leaves no visible imprint on the δ Cu signal. Together

189 with the proposition that the changing Fe(III)(oxyhydr)oxides reservoir exerted a major

190 influence on Cu bioavailability, the data assert a similar view in the modern ocean where

191 biological activity appears to play a negligible part in the δ65Cu composition of marine waters

192 (32). Not much is known on the microbial composition of the early oceans, whether similar to

193 now or not, but a recent study suggests many cyanobacteria carry primitive Precambrian

194 genes that may have been active in autotrophic CO2 fixation (47)

196 Paired Cu and Fe isotope records of marine redox evolution: In the Transvaal Supergroup,

197 a narrow heavy range in δ56Fe (+0.21 to +0.29 ‰) is recorded by ~2.66-2.5 Ga in shales (Fig.

198 2C; Fig. S2). This decreases to negative values by ca. 2.45 Ga, and evolves to slightly positive

199 values with greater variability (–0.22 to +0.32‰) for Timeball Hill shales deposited during

200 the GOE interval (Fig. 2C; Fig. S2). In the Francevillian Series, more oxic conditions

201 dominated ~2.15-2.12 Ga (9), during which bulk δ56Fe remains slightly heavy (+0.23‰ to

202 +0.49‰) in the FB1 and FB2 Formations (Fig. 2C; Fig. S2). Soon after, at ~2.084 Ga, δ56Fe

203 values display broader variability during euxinic conditions (9) in the FD Formation of the

204 Francevillian series, attaining values as low as -0.55‰ (Fig. 2C; Fig. S2).

205 The δ56Fe values are used in an attempt to underline the sensitivity of δ65Cu systematics

206 as a potential GOE transition proxy. This is because δ56Fe values in pyrites are argued to

207 characterize three distinct stages across the GOE transition and beyond (12), where the δ56Fe

208 Archean signal controlled by mass deposition of BIF, contains variable and generally negative

209 values. An intermediate early Proterozoic stage gives way to an interval with higher and

210 stable values linked to continental weathering (12). However, the trajectory of bulk δ56Fe

9 211 evolution from positive to near-zero values between 2.66 and 2.45 Ga, gradually increasing

212 from 2.45 to 2.15 Ga, and returning to more negative and variable values at 2.08 Ga, is

213 complex, suggesting an inverse relationship between bulk δ56Fe and pyrite-specific δ56Fe (12).

214 This in part is likely due to the local nature of the bulk Fe sink but also to the fact that

215 different environmental processes can generate similar Fe isotopic compositions (12, 18, 42,

216 48). Independent constraints on water column and early diagenetic redox conditions that are

217 available from our sample suite (1, 9, 33, 34, 35, 49) significantly narrow the processes likely

218 responsible for the δ56Fe compositions. The positive values preserved under ferruginous

219 conditions ca. 2.66 Ga most likely reflect precipitation of a small portion of the dissolved Fe

220 pool under weakly oxidizing conditions (12). Near-zero and variable values ~2.45–2.15 Ga

221 likely represent a more important proportion of the local dissolved Fe pool being deposited as

222 iron oxides or sulfides with increasing environmental oxygenation (12, 49). And a return to

223 bulk negative values under euxinic conditions at ca. 2.08 Ga (9) is best explained by

224 enrichment of light isotopes into iron sulfides formed in the water column or during early

225 diagenesis (48).

226 Taken together, the evolution of the sedimentary δ65Cu composition in the Late Archean

227 versus post-GOE Paleoproterozoic marine sedimentary deposits is more straightforward in

228 interpretation, since it agrees with the δ56Fe shifts of the pure pyrite proxy (12) than observed

229 for bulk (Figs 2A & C). The transition to positive δ65Cu values in the Timeball Hill Formation

230 and the Francevillian FD Formation, coincidental with expanding euxinia generated through

231 bacterial sulfate reduction (1, 9), is also recorded by a positive δ56Fe transition in sedimentary

232 pyrites (12), validating bulk δ65Cu as a potential GOE transition proxy. Moreover, the higher

233 consistency of the δ65Cu signal compared to bulk δ56Fe further highlights its value in

234 identifying the GOE transition in sedimentary deposits, supporting the idea that the δ56Fe of

10 235 pyrites captures the protracted nature of the GOE, culminating in the muting of negative black

236 shale δ65Cu at 2.3 Ga.

238 Implications for Cu cycling at the Archean-Proterozoic boundary: The complete

239 disappearance of δ65Cu values ≤0 after 2.3 Ga suggests that Cu isotopes captured a uni-

240 directional shift in the oxygenation of the atmosphere-ocean system across the GOE. Protein

241 structure phylogenomics suggest the global emergence of large-scale biological Cu utilization

242 between ~2.4-2.2 Ga—i.e., during the GOE (43). Evidence from biological assimilation by

243 modern ocean microbial populations (32, 50, 51) suggests that microbial Cu assimilation does

244 not fractionate Cu isotopes, and thus should not influence the Cu isotope composition of

245 seawater nor be expected to drive Cu isotopic evolution from Archean to Paleoproterozoic

246 seawater. Increased riverine Cu and sulfate supply as the result of sulfide mineral weathering

247 across the GOE was accompanied by a decrease in Fe(III)(oxyhydr)oxide precipitation after

248 the GOE, as evident from the dramatic reduction in BIF deposition (Fig. 1D) and increased

249 precipitation of sulfide minerals (12, 15-17) after the GOE. Indeed, our data indicate an

250 increase in bulk δ34S values beginning ~2.15 Ga (Table 1 & Fig. S2), coincident with the

251 general expansion of the marine sulfate budget after the GOE (1-4, 6, 8, 11). The isotopically

252 heavy Cu associated with an increasing continental supply would have been actively

253 incorporated into biological enzymes and/or adsorbed to phytoplankton cell surfaces (20, 32).

254 As mentioned above, modern ocean and experimental studies suggest a net zero isotopic shift

255 during the assimilation of dissolved δ65Cu (~0.5-0.9‰) by phytoplankton biomass. Instead,

256 the isotopically heavy δ65Cu values in post-GOE shales are most parsimoniously explained by

257 continental weathering processes (21-30) delivering dissolved riverine Cu with a δ65Cu

258 signature of up to ~0.7‰ (30, 52-54) to the oceans.

11 259 The simple explanation for a shift from light Late Archean to permanently heavy

260 Proterozoic bulk δ65Cu may be complicated by episodes in the Neoarchean record, 2.66-2.5

261 Ga, when mass balance between BIF Cu and Shale Cu may have approached equilibrium,

262 suggested by coeval Cu/Ti signals (Fig. 1C). This may corrupt the simplicity of the δ65Cu

263 model because it is possible that the very light δ65Cu values produced during these episodes

264 may be reflecting the formation of Cu sulfides when the black shale and BIF Cu sinks are not

265 markedly different. Reduction in the Fe(III)(oxyhydr)oxide sink across the GOE would

266 therefore make black shale sulfides a major Cu sink, swinging the authigenic δ65Cu signal to

267 heavier values difficulty to distinguish from a new continental source carrying a similar

268 signal. However, the only evidence of when considerable change in the sulfur reservoir may

269 have impacted Cu content coincides with an expanding sedimentary sulfur reservoir and a

270 concomitant reduction in the Fe(III)(oxyhydr)oxides reservoir, after 2.4 Ga (Fig. S3). The

271 persistently low bulk sulfur signal before this time is taken to mean that Cu concentrations

272 were regulated mainly by a fluctuating but broadly larger Fe(III)(oxyhydr)oxide reservoir, not

273 sulfide.

274 The fact that contemporary dissolved seawater δ65Cu is isotopically heavier by 1.5‰

275 relative to riverine inputs is further linked to fractionation against 65Cu during adsorption to

276 authigenic ferromanganese marine precipitates (52-54). This view seems to contradict the fact

277 that experimental and riverine Fe(III)(oxyhydr)oxides preferentially remove 65Cu. However,

278 the isotopically-light Cu in modern ferromanganese deposits appears associated with Mn(IV)

279 instead of Fe(III) (52-54). Since birnessite has been shown to preferentially remove the light

280 Cu isotope (53, 54), this may further explain why after the GOE sedimentary marine δ65Cu

281 values became permanently positive, since Mn(IV) oxides are only precipitated in the

282 presence of free oxygen (55). In the Archean oceans, Mn(IV) precipitation was generally

283 inhibited, but became significant across the GOE, exemplified by the deposition of Earth’s

12 284 largest Mn(IV) oxide deposits ~2.4-2.2 Ga (55). Similarly, our sample set records an increase

285 in Mn content from the Archean (Fig. S4) and our predominantly positive δ65Cu

286 Paleoproterozoic values, though lower (Table 1), fall within the modern marine sedimentary

287 range (32).

288 Similar and mostly positive δ65Cu values, of up to 0.49‰, characterize Cambrian

289 marine shales genetically linked to the oxidative weathering of Precambrian basement rocks

290 (56). The Cambrian shales accumulated under euxinic marine conditions characterized by

291 intense bacterial sulfate reduction in a lagoonal transgressive environment (56, 57). These

292 sulfidic depositional conditions are similar to the situation under which the Francevillian FD

293 and Timeball Hill formations were deposited (1, 9). If direct microbial production and

294 consumption of organic matter has a negligible effect on marine δ65Cu dissolved values (e.g.,

295 32), then it is reasonable to assume that early microbial diagenetic processing of sedimentary

296 organic matter would have simply recycled and preserved the Cu isotopic signatures

297 originating from organic matter in the marine water column. Although the generation of

298 secondary minerals enriched in 63Cu through fluid migration in reworked clastic sediments has

299 been noted (58), such metasomatic processes are not known to have affected our samples (1,

300 9, 33, 34). Moreover, metasomatism is often local and, therefore, cannot explain the secularly

301 consistent sedimentary δ65Cu pattern from ~2.66-2.3 Ga or the reproducibility of positive

302 sedimentary δ65Cu values after 2.3 Ga.

304 Conclusions

305 Across the GOE boundary, we observe a consistent trend from negative δ65Cu and positive

306 δ56Fe values during ferruginous conditions, towards heavy δ65Cu and light δ56Fe signatures in

307 post-GOE sulfidic sedimentary successions. We interpret the transition in marine δ65Cu

308 values, from negative to positive across the GOE, to reflect a progressive increase in sulfide

13 309 availability after the GOE, itself related to increased oxidative continental weathering, and in

310 turn, to waning BIF deposition. Our dataset reveals a disappearance of negative δ65Cu values

311 from the black shale record after ~2.3 Ga, when the GOE was at its height, and supports a

312 change from low marine sulfate conditions during the Archean to sulfate-rich oceans after the

313 GOE. The gradual change in our δ65Cu values continues for ~150 million years after the

314 disappearance of mass independent sulfur isotope fractionation at ~2.45 Ga (3, 4), a key

315 marker of the GOE (59), re-affirming that the GOE was a protracted event that took place

316 between 2.45-2.3 Ga (7).

318 Methods

319 Bulk Cu and Fe isotope measurements were performed at ALS Scandinavia on a multicollect-

320 or inductively coupled plasma mass spectrometer (MC-ICP-MS) at ALS Scandinavia. See

321 supplementary information for details of sample preparation. A Neptune Plus was used at me-

322 dium resolution mode. A standard-sample bracketing technique was used during the measure-

323 ment. The reported error is 2σ based on long time reproducibility. Prior to the analysis of cop-

324 per stable isotopes, the Cu fraction was diluted to 0.5 Cu mg l-1 (where possible) followed by

325 addition of 2 mg l-1 Zn for mass bias correction. The Fe fraction was diluted to 2 mg l-1. A plot

326 of δ57/56Fe vs. δ56/54Fe reveals strictly mass-dependent fractionation, confirming interference-

327 free measurement of all three isotopes. Online mass bias correction was achieved by spiking

328 the samples with nickel (internal standard) to 4 mg l-1 (60). Values are reported in per mil (‰)

329 according to equation 1 and 2. A detailed description of the operating conditions and measure-

330 ment, as well as reproducibility and accuracy data is reported in 61.

65 65 63 65 63 332 δ Cu = [( Cu/ Cu)sample/( Cu/ Cu)ERM-AE633 - 1] × 1000 (1)

56 56 54 56 54 333 δ Fe = [( Fe/ Fe)sample/( Fe/ Fe)IRMM-014 - 1] × 1000 (2)

14 334

335 Bulk stable isotopes of C, S and N were analysed at the Stable Isotope Laboratory (SIL) of the

336 Department of Geological Sciences, Stockholm University as described in supplementary

337 information.

15 340 Acknowledgements

341 We will like to thank Hans Schöberg for help with setting up methods for Cu isotope analysis

342 and Karin Wallner for help in the clean lab at the Swedish Museum of Natural History.

343 Thanks to François Gauthier Lafaye for his logistic support. This work was initiated by a

344 Marie Curie fellowship (Grant No: PIEF.GA-2010-276475) and partly by funds from the

345 European Research Council (Grant No: 336092) and the Bolin Centre for Climate Change.

347 Author contribution

348 E.C.F. Designed the project; N.P.R. and E.C.F. conducted research. N.P.R. and I.R.

349 performed isotopic analysis. S.L., C.P. and K.K. compiled Cu data through time. A.E.A.

350 provided rock samples. P.A. and D.W. contributed to Cu isotopic analysis methods. N.P.R.,

351 P.K., S.L., C.P., K.K. and E.C.F. analysed data. E.C.F. wrote the paper. All authors

352 contributed in editing the manuscript. K.K. would like to thank the Natural Sciences and

353 Engineering Research Council of Canada (NSERC) for their financial support.

16 365 References

366 1. Bekker A, et al. (2004) Dating the rise of atmospheric oxygen. Nature 42:117–120.

367 2. Canfield DE (2005) The early history of atmospheric oxygen: Hommage to Robert M.

368 Garrels. Annual Rev Earth Plan Sci 33:1–36.

369 3. Guo Q, et al. (2009) Reconstructing Earth's surface oxidation across the Archean–Pro-

370 terozoic transition. Geology 37:399–402.

371 4. Lyons TW, Reinhard, CT, Planavsky NJ (2014) The rise of oxygen in Earth’s early ocean

372 and atmosphere. Nature 506:307–315.

373 5. Farquhar J, Zerkle AL, Bekker A (2011) Geological constraints on the origin of oxy-

374 genic photosynthesis. Photosynth Res 107:11–36.

375 6. Canfield DE (1998) A new model for Proterozoic ocean chemistry. Nature 396:450–

376 453.

377 7. Konhauser KO, et al. (2011) Aerobic bacterial pyrite oxidation and acid rock drainage dur-

378 ing the Great Oxidation Event. Nature 478:369–373.

379 8. Planavsky NJ, Bekker A, Hofmann A, Owens JD, Lyons TW (2012) Sulfur record of

380 rising and falling marine oxygen and sulfate levels during the Lomagundi event. Proc

381 Natl Acad Sci U.S.A. 109:18300–18305.

382 9. Canfield DE, et al. (2013) Oxygen dynamics in the aftermath of the Great Oxidation

383 of Earth’s atmosphere. Proc Natl Acad Sci U.S.A. 110:16736–16741.

384 10. Reinhard CT, et al. (2013) Proterozoic ocean redox and biogeochemical stasis. Proc

385 Natl Acad Sci U.S.A. 110:5357–5362.

17 386 11. Scott C, et al. (2014) Pyrite multiple-sulfur isotope evidence for rapid expansion and

387 contraction of the early Paleoproterozoic seawater sulfate reservoir. Earth Plan Sci

388 Lett 389:95–104.

389 12. Rouxel OJ, Bekker A, Edwards KJ (2005) Iron isotope constraints on the Archean and

390 Paleoproterozoic ocean redox state. Science 307:1088–1091.

391 13. Poulton SW, Canfield DE (2011) Ferruginous conditions: A dominant feature of the

392 ocean through Earth's history. Elements 7:107–112.

393 14. Planavsky NJ, McGoldrick P, Scott CT, Li C, Reinhard CT, Kelly AE, Chu X, Bekker

394 A, Love GD, Lyons TW (2011) Widespread iron-rich conditions in the mid-Proterozo-

395 ic ocean. Nature 477:448-451.

396 15. Frei RG, Stolper C, Canfield DE (2013) Fluctuations in late Neoproterozoic atmo-

397 spheric oxidation — Cr isotope chemostratigraphy and iron speciation of the late Edi-

398 acaran lower Arroyo del Soldado Group (Uruguay). Gondwana Res 23:797-811.

399 16. Bekker A et al. (2010) Iron formation: The sedimentary product of a complex

400 interplay among mantle, tectonic, oceanic, and biosphere processes. Econ Geol

401 105:467-508.

402 17. Planavsky N., Rouxel OJ, Bekker A., Hoffmann A, Little CTS, Lyons TW. (2012).

403 Iron isotope composition of some Archean and Proterozoic iron formations. Geochim

404 Cosmochim Acta 80:158–169 .

405 18. Johnson CM, Beard BL, Roden EE (2008) The iron isotope fingerprints of redox and

406 biogeochemical cycling in modern and ancient Earth. Annual Rev Earth Plan Sci

407 36:457-493.

18 408 19. Balistrieri LS, Borrok DM, Wanty RB, Ridley WI (2008) Fractionation of Cu and Zn

409 isotopes during adsorption onto amorphous Fe(III) oxyhydroxide: Experimental mix-

410 ing of acid rock drainage and ambient river water. Geochim Cosmochim Acta 72:311-

411 328.

412 20. Pokrovsky OS, Viers J, Emnova EE, Kompantseva EI, Freydier R (2008) Copper iso-

413 tope fractionation during its interaction with soil and aquatic microorganisms and met-

414 al oxy(hydr)oxides: Possible structural control. Geochim Cosmochim Acta 72:1742–

415 1757.

416 21. Ehrlich S, Butler I, Halicz L, Rickard D, Oldroyd A, Mathews A (2004) Experimental

417 study of copper isotope fractionation between aqueous Cu(II) and covellite, CuS.

418 Chem Geol 209:259–269.

419 22. Fernandez A, Borrok DM (2009) Fractionation of Cu, Fe, and Zn isotopes during the

420 oxidative weathering of sulfide-rich rocks. Chem Geol 263:1-12.

421 23. Rodríguez NP, Khoshkhoo M, Sandström A, Rodushkin I, Alakangas L, Öhlander B

422 (2015) Isotopic signature of Cu and Fe during bioleaching and electrochemical leach-

423 ing of chalcopyrite concentrate. Inter J Min Processing 134:58-65.

424 24. Mathur R, Schlitt WJ (2010) Identification of the dominant Cu ore minerals providing

425 soluble copper at Cañariaco Peru through Cu isotope analyses of batch leach experi-

426 ments. Hydrometallurgy 101:15–19.

427 25. Fantle MS, DePaolo DJ (2004) Iron isotopic fractionation during continental weather-

428 ing. Earth Plan Sci Lett 228:547–562.

429 26. Mathur R, Jin L, Prush V, Paul J, Ebersole C, Fornadel A, Williams JZ, Brantley S

19 430 (2004) Cu isotopes and concentrations during weathering of black shale of the Marcel-

431 lus Formation, Huntingdon County, Pennsylvania (USA). Chem Geol 304-305: 175–

432 184.

433 27. Kimball BE, Mathur R, Dohnalkova C, Wall J, Runkel RL, Brantley SL (2009) Cop-

434 per isotope fractionation in acid mine drainage. Geochim Cosmochim Acta 73:1247–

435 1263.

436 28. Rodríguez NP, Engström E, Khoshkhoo M, Rodushkin I, Nason P, Alakangas L, Öh-

437 lander B (2013) Copper and iron isotope fractionation in mine tailings at the Laver and

438 Kristineberg mines, northern Sweden. Appl Geochem 32: 204–215.

439 29. Wall A.J, Mathur R, Post JE, Heaney PJ (2011) Cu isotope fractionation during

440 bornite dissolution: An in situ X‐ ray diffraction analysis. Ore Geol Rev 42:62–70.

441 30. Sherman DM (2013) Equilibrium isotopic fractionation of copper during oxidation/re-

442 duction, aqueous complexation and ore-forming processes: Predictions from hybrid

443 density functional theory. Geochim Cosmochim Acta 118:85–97.

444 31. Liu S-A, Teng F-Z. Li S, Wei G-J, Ma J-L, Li D (2014) Copper and iron isotope frac-

445 tionation during weathering and pedogenesis: Insights from saprolite profiles.

446 Geochim Cosmo Acta 146, 59-75.

447 32. Takano S, Tanimizu M, Hirata T, Sohrin Y (2014) Isotopic constraints on biogeo-

448 chemical cycling of copper in the ocean. Nat Comm 5:5663 doi:10.1038/ncomms6663.

449 33. Kendall B, Reinhard CT, Lyons TW, Kaufman AJ, Poulton SW, Anbar AD, 2010.

450 Pervasive oxygenation along late Archaean ocean margins. Nat Geosci, 3:647-652.

451 doi: 10.1038/ngeo942

20 452 34. Ngombi-Pemba L, El Albani A, Meunier A, Grauby O, Gauthier-Lafaye F (2014)

453 From detrital heritage to diagenetic transformations, the message of clay minerals con-

454 tained within shales of the Palaeoproterozoic Francevillian basin (Gabon) Precambri-

455 an Res 255:63–76.

456 35. Chi Fru E, Arvestål E, Callac N, El Albani A, Kilias S, Argyraki A, Jakobsson M,

457 2015. Arsenic stress after the Proterozoic glaciations. Nat. Sci. Rep. 5:17789, DOI:

458 10.1038/srep17789.

459 36. Demory F, Oberhänsli H, Nowaczyk NR, Gottschalk M, Wirth R, Naumann R (2005)

460 Detrital input and early diagenesis in sediments from Lake Baikal revealed by rock

461 magnetism. Global Plan Change, 461:145-166.

462 37. Parnell J, Mark DF, Frei R, Fallic A.E, Ellam RM (2014) 40Ar/39Ar of exceptional con-

463 centration of metals by weathering of Precambrian rocks at the Precambrian-Cambrian

464 boundary. Pre Res 246:54–63.

465 38. Gallagher M, Turner EC, Kamber BS (2015) In situ trace metal analysis of

466 Neoarchean-Ordovician shallow-marine microbial-carbonate-hosted pyrites. Geobio-

467 logy 13:316–339.

468 39. Large RS, Halpin JA, Danyushevsky LV, Maslennikov VV, Bull SW, Long JA,

469 Gregory DD, Lounejeva E, Lyons TW, Slack PJ, McGoldrick PJ (2014) Trace element

470 content of sedimentary pyrite as a new proxy for deep-time ocean-atmosphere. Earth

471 Plan Sci Lett 389:209–220.

472 40. Saito AM, Sigman DM, Morel FMM (2003) The bioinorganic chemistry of the ancient

473 ocean: the co-evolution of cyanobacterial metal requirements and biogeochemical

474 cycles at the Archean-Proterozoic boundary? Inorg Chimica Acta 356:308–318.

21 475 41. Konhauser KO, Hamade T, Raiswell R, Morris RC, Ferris FG, Southam G, Canfield

476 DE (2002) Could bacteria have formed the Precambrian banded iron formations. Geo-

477 logy 30:1079-1082.

478 42. Mathur R, Ruiz J, Titley S, Liermann L, Buss H, Brantley S (2005) Cu isotopic frac-

479 tionation in the supergene environment with and without bacteria. Geochim

480 Cosmochim Acta 69:5233–5246.

481 43. Dupont CL, Butcher A, Valas RE, Bourne PE, Caetano-Anolles G (2010) The history

482 of biological metal utilization inferred through phylogenomic analysis of protein struc-

483 tures. Proc Natl Acad Sci U.S.A. 107:10567–10572.

484 44. Weber F, Gauthier-Lafaye F (2012) No proof from carbon isotopes in the

485 Francevillian (Gabon) and Onega (Fennoscandian shield) basins of a global oxidation

486 event at 1980-2090 Ma following the Great Oxidation Event (GOE). Comptes rendus

487 - Géosci 345:28–35.

488 45. Kump et al. (2011) Isotopic Evidence for Massive Oxidation of Organic Matter Fol-

489 lowing the Great Oxidation Event. Science 334:1694-1696.

490 46. El Albani et al., (2010) Large colonial organisms with coordinated growth in oxy-

491 genated environments 2.1 Gyr ago. Nature 466:100–104.

492 47. Shih PM, Occhialini A, Cameron JC, Andralojc PJ, Parry MAJ, Kerfeld CA (2016)

493 Biochemical characterization of predicted Precambrian RuBisCO. Nat Comm 7:10382

494 doi:10.1038/ncomms10382.

495 48. Guilbaud R, Butler IB, Ellam RM (2011) Abiotic pyrite formation produces a large Fe

496 isotope fractionation. Science 332:1548–1551.

22 497 49. Bankole O, El Albani A, Rouxel O, Ngombi-Pemba L, Bekker A, Gauthier Lafaye F

498 (2015) Origin and significance of Paleoproterozoic redbeds in the FA formation,

499 Francevillian Basin, Gabon: A geochemical and isotopic constraints. Goldschmidt

500 2015. 194.

501 50. Navarrete JU, Borrok DM, Viveros M, Ellzey JT (2011) Copper isotope fractionation

502 during surface adsorption and intracellular incorporation by bacteria. Geochim

503 Cosmochim Acta 75:784-799

504 51. Morel FM, Price NM (2003) The biogeochemical cycles of trace metals in the oceans.

505 Science 300:944-7.

506 52. Little SH, Vance D, Walker-Brown C, Landing WM (2014) The oceanic mass balance

507 of copper and zinc isotopes, investigated by analysis of their inputs, and outputs to fer-

508 romanganese oxide sediments. Geochim Cosmochim Acta 125:673–693.

509 53. Little SH, Sherman DM, Vance D, Hein JR (2014) Molecular controls on Cu and Zn

510 isotopic fractionation in Fe-Mn Crusts. Earth Plan Sci Letts 396:213-222.

511 54. Manceau A, Nagy KL (2015) Comment on “Molecular controls on Cu and Zn isotopic

512 fractionation in Fe-Mn crusts” by Little et al. Earth Plan Sci Letts 441:310-312.

513 55. Roy S (2006) Sedimentary manganese metallogenesis in response to the evolution of

514 the Earth system. Earth-Sci Rev 77: 273–305.

515 56. Asael D, Mathews A, Bar-Mathews A, Halicz L (2007) Copper isotope fractionation

516 in sedimentary copper mineralization (Timna Valley, Israel). Chem Geol 243:238–

517 254.

518 57. Shlomovitch N, Bar-Matthews M, Matthews A (1999) Sedimentary and epigenetic

23 519 copper mineral assemblages in the Cambrian Timna Formation, southern Israel. Israel

520 J. Earth Sci 48:195–208.

521 58. Mason TFD, et al. (2005). Zn and Cu isotopic variability in the Alexandrinka vol-

522 canic-hosted massive sulphide (VHMS) ore deposit, Urals, Russia. Chem Geol

523 221:170–187.

524 59. Farquhar J, et al. (2000) Atmospheric influence of Earth’s earliest sulfur cycle. Sci-

525 ence 289, 756–758.

526 60. Malinovski D, Stenberg A, Rodushkin I, Andren H, Ingri J, Öhlander B, Baxter D

527 (2003) Performance of high resolution MC-ICP-MS for Fe isotope ratio measurements

528 in sedimentary geological materials J Anal At Spectrom 18:687-695.

529 61. Rodushkin I, Pallavicini N, Engström E, Sörlin E, Öhlander B, Ingri J, Baxter D

530 (2015) Assessment of natural variability of B, Cd, Cu, Fe, Pb, Sr, Tl and Zn

531 concentrations and isotopic compositions in leaves, needles and mushrooms using

532 single sample digestion and two-column matrix separation. J Anal At Spectrom, In

533 Press. DOI: 10.1039/C5JA00274E.

539 Figure Caption

24 540

541 Fig. 1. Copper distribution in banded iron formations and black shales throughout Earth

542 history (see Table S1 & S2 for details). A) and B) molar Cu/Ti ratios in shales and

543 banded iron formations, respectively. C) their enrichments compared (black shale in

544 green, banded iron formation in red), relative to the average continental crust (white

545 line). D) banded iron formation tonnage over time, reproduced from reference 16.

65 13 56 65 547 Fig. 2. Box and whisker isotopic plots (δ C, δ Corg, δ Fe) across the GOE interval. A) δ Cu

548 values in ~2.7 to 2.0 Ga black shales, showing a transition to heavier values with the on-

13 56 549 set of the GOE. B) and C), corresponding δ Corg and δ Fe values, respectively.

551 Fig. 3. Rayleigh isotope distillation of δ65Cu between seawater and iron oxide sinks. Black

552 shale δ65Cu values are assumed to reflect contemporaneous seawater composition (32).

553 An experimentally–determined 65Cu value of 0.75‰ between seawater and Fe(III)

554 (oxyhydr)oxides (19, 20) was used to calculate the isotopic composition of Cu

555 sequestered into iron oxides and residual isotopic composition of seawater, assuming an

556 initial δ65Cu of seawater of 0.5‰ in the absence of BIF deposition, similar to modern

557 (32).

559 Fig. 4. Conceptual model for Late Archean and Early Proterozoic marine cycling and isotope

560 composition of Cu. A) During the late Archean, the marine Cu cycle lacked oxidative

561 weathering inputs, and was strongly influenced by iron oxide deposition B) During the

25 562 early Proterozoic, isotopically heavy Cu was supplied by continental weathering of

563 sulfides, while iron oxide sinks waned, both promoting 63Cu depletion from seawater

564 and pelagic sediments. Fe-OB stands for iron-oxidizing bacteria.