Iron isotope biogeochemistry of Neoproterozoic marine shales

Abstract

Iron isotopes have been widely applied to investigate the redox evolution of Earth’s surface environments. However, it is still unclear whether iron cycling in the water column or during diagenesis represents the major control on the iron isotope composition of sediments and sedimentary rocks. Interpretation of isotopic data in terms of oceanic redox conditions is only possible if water column processes dominate the isotopic composition, whereas redox interpretations are less straightforward if diagenetic iron cycling controls the isotopic composition. In the latter scenario, iron isotope data is more directly related to microbial processes such as dissimilatory iron reduction. Here we present bulk rock iron isotope data from late Proterozoic marine shales from Svalbard, northwestern Canada, and Siberia, to better understand the controls on iron isotope fractionation in late Proterozoic marine environments. Bulk shales span a $\delta$⁵⁶Fe range from −0.45$‰$ to +1.04$‰$. Although $\delta$⁵⁶Fe values show significant variation within individual stratigraphic units, their mean value is closer to that of bulk crust and hydrothermal iron in samples post-dating the ca. 717–660 Ma Sturtian glaciation compared to older samples. After correcting for the highly reactive iron content in our samples based on iron speciation data, more than 90% of the calculated $\delta$⁵⁶Fe compositions of highly reactive iron falls in the range from ca. −0.8$‰$ to +3$‰$. An isotope mass-balance model indicates that diagenetic iron cycling can only change the isotopic composition of highly reactive iron by $<$1$‰$, suggesting that water column processes, namely the degree of oxidation of the ferrous seawater iron reservoir, control the isotopic composition of highly reactive iron. Considering a long-term decrease in the isotopic composition of the iron source to the dissolved seawater Fe(II) reservoir to be unlikely, we offer two possible explanations for the Neoproterozoic $\delta$⁵⁶Fe trend. First, a decreasing supply of Fe(II) to the ferrous seawater iron reservoir could have caused the reservoir to decrease in size, allowing a higher degree of partial oxidation, irrespective of increasing environmental oxygen levels. Alternatively, increasing oxygen levels would have led to a higher proportion of Fe(II) being oxidized, without decreasing the initial size of the ferrous seawater iron pool. We consider the latter explanation as the most likely. According to this hypothesis, the $\delta$⁵⁶Fe record reflects the redox evolution of Earth’s surface environments. $\delta$⁵⁶Fe values in pre-Sturtian samples significantly heavier than bulk crust and hydrothermal iron imply partial oxidation of a ferrous seawater iron reservoir. In contrast, mean $\delta$⁵⁶Fe values closer to that of hydrothermal iron in post-Sturtian shales reflects oxidation of a larger proportion of the ferrous seawater iron reservoir, and by inference, higher environmental oxygen levels. Nevertheless, significant iron isotopic variation in post-Sturtian shales suggest redox heterogeneity and possibly a dominantly anoxic deep ocean, consistent with results from recent studies using iron speciation and redox sensitive trace metals. However, the interpretation of generally increasing environmental oxygen levels after the Sturtian glaciation highlights the need to better understand the sensitivity of different redox proxies to incremental changes in oxygen levels to enable us to reconcile results from different paleoredox proxies.

Introduction

The distribution of iron isotopes in sediments and sedimentary rocks is a powerful measure of the biogeochemical cycle of Fe in the modern and ancient ocean. By far the largest fractionation of Fe isotopes is associated with redox transformation between ferrous and ferric iron (e.g., Schauble et al., 2001, Johnson et al., 2002, Johnson et al., 2008b, Welch et al., 2003, Anbar et al., 2005), and hence Fe isotope ratios have been applied to track the evolution of the seawater iron reservoir in response to increasing atmospheric oxygen levels throughout Earth’s history (e.g., Rouxel et al., 2005, Planavsky et al., 2012, Fan et al., 2014, Tahata et al., 2015, Zhang et al., 2015). In spite of the dominant role of redox processes in iron isotope fractionation, the interpretation of Fe isotope data from ancient sediments is not straightforward because of different isotopic compositions of iron sources to the ocean and the complex cycling of iron in the water column and during early diagenesis. Nevertheless, distinct features in the iron isotope record appear to broadly correlate with the evolution of atmospheric O₂. The Neoarchean to early Paleoproterozoic (ca. 2.8–2.3 Ga) record comprises highly variable, but typically negative, sedimentary pyrite $\delta$⁵⁶Fe values ($\sim$−3.5$‰$ to +1.2$‰$) (Rouxel et al., 2005, Archer and Vance, 2006). Younger Proterozoic (2.3–0.54 Ga) pyrites also span a broad range from $\sim$−2.0$‰$ to +4.0$‰$ but are generally less negative (Tahata et al., 2015), whereas Phanerozoic ($<$0.54 Ga) and modern sedimentary pyrites are also less negative and span a narrower range from $\sim$−1.6$‰$ to +1.2$‰$ (e.g., Rouxel et al., 2005, Severmann et al., 2006, Severmann et al., 2008, Fehr et al., 2008, Fehr et al., 2010). The Fe isotope record of banded iron formation (BIF) and other Fe oxyhydroxide-rich rocks also displays a systematic secular trend. Archean and early Paleoproterozoic iron formation show a large variation in bulk rock $\delta$⁵⁶Fe values from −2.4$‰$ to +1.8$‰$ with most samples displaying heavy isotope ratios (see compilations in Planavsky et al., 2012, Busigny et al., 2014). Younger Fe oxyhydroxide-rich rocks span a narrower range from −0.9$‰$ to +1.2$‰$ with most samples falling between −0.5$‰$ and +0.5$‰$ (Planavsky et al., 2012, Busigny et al., 2014).

Two broad and contrasting approaches have been implemented in interpreting the record of iron isotopes in the Precambrian. One approach stresses the importance of ferrous iron oxidation across the chemocline in an anoxic Precambrian ocean as the controlling mechanism for the Fe isotopic composition of sedimentary rocks and minerals (Rouxel et al., 2005, Planavsky et al., 2012, Busigny et al., 2014). Whereas quantitative oxidation masks isotope fractionations and produces ferric oxyhydroxides with Fe isotope ratios comparable to the coeval seawater ferrous iron reservoir, non-quantitative oxidation leads to preservation of a significant positive isotope effect upon precipitation of Fe-oxyhydroxides with concomitant depletion of the ferrous iron reservoir (Dauphas et al., 2004, Rouxel et al., 2005, Planavsky et al., 2012, Busigny et al., 2014, Mendes et al., 2017). By extension, lower oxygen concentration in the Neoarchean and early Paleoproterozoic led to incomplete oxidation of the ferrous seawater iron inventory, producing isotopically heavy Fe-oxyhydroxides (precipitated as BIF with heavy $\delta$⁵⁶Fe) and leaving behind a depleted ferrous reservoir from which pyrites formed (Rouxel et al., 2005, Planavsky et al., 2012). In contrast, higher oxygen levels later in Earth’s history allowed for more complete oxidation of ferrous iron, resulting in a narrower range in iron isotope ratios and isotopically heavier pyrites. However, this hypothesis has been challenged because the inferred large negative seawater isotope shift requires that a significant proportion of the dissolved iron inventory precipitated as Fe-oxyhydroxides (50–90%; Yamaguchi and Ohmoto, 2006, Johnson et al., 2008a).

A second approach to interpreting secular trends in the sedimentary iron isotope record emphasizes the importance of diagenetic and microbial processes such that the iron-isotopic composition of BIF and pyrite predominantly reflects dissimilatory iron reduction, which produces large amounts of isotopically light ferrous iron (Johnson et al., 2008a, Johnson et al., 2008b, Heimann et al., 2010, Craddock and Dauphas, 2011, Li et al., 2015). In this case, the secular trend in Fe isotopes in the Precambrian is thought to reflect the evolution of heterotrophic organisms, whereas the link to Earth’s evolving redox state is ambiguous. This interpretation easily accounts for strongly negative values seen in some samples; however, while diagenetic processes can explain iron isotope partitioning among different mineral phases, shifts in the bulk rock composition require physical separation of distinct iron isotope pools, which is a process that is redox dependent.

A diverse set of redox proxies suggest that atmospheric oxygen levels increased in the Neoproterozoic (e.g., Fike et al., 2006, Canfield et al., 2007, Sahoo et al., 2012, Och and Shields-Zhou, 2012, Ader et al., 2014, Sperling et al., 2015, Kunzmann et al., 2017). However, the degree and exact timing of oxygenation are not well understood (Sahoo et al., 2012, Sahoo et al., 2016, Kunzmann et al., 2017, Miller et al., in press). Although iron isotope studies may provide valuable insights, only a small number of iron isotope datasets on the Neoproterozoic have been published (Halverson et al., 2011, Fan et al., 2014, Zhang et al., 2015, Tahata et al., 2015, Cox et al., 2016a). Iron isotope ratios in Sturtian iron formation deposited during post-glacial transgression show a systematic trend of up-section increasing $\delta$⁵⁶Fe values, which has been interpreted as oxidation of a successively decreasing proportion of the ferrous iron reservoir during deepening of the depositional environment (Halverson et al., 2011, Cox et al., 2016a). Sedimentary pyrites in Cryogenian (ca. 717–635 Ma) black shales from South China show a systematic decrease from positive values (ca. 0$‰$ to +1$‰$) to values comparable to hydrothermal iron sources (ca. −0.5$‰$ to +0$‰$), which Zhang et al. (2015) interpreted to reflect near-quantitative oxidation of the ferrous iron reservoir and, by inference, increasing environmental oxygen levels. Tahata et al. (2015) reported Fe isotope data from pyrites in carbonates, shales, sandstones, and diamictites from a Neoproterozoic succession in Svalbard and suggested that heavy $\delta$⁵⁶Fe values before and after the Sturtian glaciation attest to an anoxic-ferruginous deep ocean. The bulk rock Fe isotopic composition of carbonates, phosphorites, and diagenetic chert from the Ediacaran (ca. 635–541 Ma) Doushantuo Formation in South China were also interpreted to reflect a ferruginous deep ocean (Fan et al., 2014). Although these interpretations are broadly consistent with redox constraints from other proxies (e.g., Canfield et al., 2008, Johnston et al., 2010, Johnston et al., 2013, Li et al., 2010, Sperling et al., 2013, Sperling et al., 2015, Kunzmann et al., 2015), limited data hinders the reconciliation of the iron isotope record with proposed models of oxygenation.

The reconstruction of a detailed secular Fe isotope trend is further complicated because previous studies focused on different rock types and either presented bulk rock or mineral-specific isotope data. Data from specific minerals (like pyrite) yield more pronounced isotopic variations because the biogeochemical signals are undiluted by detrital sedimentary components. However, specific minerals may not fully capture seawater signals. For example, pyrite formed in non-euxinic environments does not quantitatively capture dissolved ferrous iron (Berner, 1984), making it difficult to extract information about changing seawater chemistry. In contrast, although bulk rock analyses include detrital iron sources, this method has the advantage of capturing both diagenetic and water column processes. Furthermore, recrystallization during burial should not affect the bulk rock composition, even though it may redistribute isotopes between individual mineral phases.

Here we report the bulk rock Fe isotope composition of 124 marine shales from Svalbard, northwestern Canada, and Siberia deposited between 1050 and 570 Ma. Our first aim is to evaluate the relative importance of isotopic fractionation in the water column versus fractionation during diagenesis on the bulk shale Fe isotope composition. To address this question, we created an isotope mass-balance model, which suggests that water column processes, i.e. the quantitative or non-quantitative oxidation of a ferrous iron reservoir, are more important than fractionation during diagenesis in setting bulk rock compositions. Based on conclusions drawn from modeling, we qualitatively interpret our new record of the secular variation of Fe isotopes in Neoproterozoic shales (Fig. 1) in the context of the relative proportion of the oxidized Fe(II) pool. Favoring a control by environmental oxygen levels, the Fe isotope record qualitatively suggests an increase in oxygen levels after the 717–660 Ma Sturtian glaciation. Nevertheless, significant iron isotopic variation in post-Sturtian units suggest redox heterogeneity and a deep ocean that possibly remained dominantly anoxic, at least through the mid-Ediacaran.