Technical Comments

Comment on “Abiotic Pyrite Formation Produces a Large Fe Isotope Fractionation”

See allHide authors and affiliations

Science  03 Feb 2012:
Vol. 335, Issue 6068, pp. 538
DOI: 10.1126/science.1211804


Guilbaud et al. (Reports, 24 June 2011, p. 1548) suggest that the geologic record of Fe isotope fractionation can be explained by abiological precipitation of pyrite. We argue that a detailed understanding of the depositional setting, mineralogy, and geologic history of Precambrian sedimentary rocks indicates that the Fe isotope record dominantly reflects biological fractionations and Fe redox processes.

Interpreting the origin of isotopic variations preserved in the rock record requires consideration of geologic, petrographic, and geochemical contexts. Guilbaud et al. (1) described a kinetic Fe isotope fractionation factor for abiological production of pyrite from aqueous and solid FeS, which potentially provides important insight into Fe isotope compositions of sulfides in the rock record. These authors suggest that negative δ56Fe values measured in pyrite from the geologic record could have been produced by this process, implying that Fe isotopes cannot be used to trace ancient biologically or abiologically mediated redox processes. We find such an interpretation to be faulty for three reasons: (i) The authors ignore evidence for biological and abiological redox processes contained in the diversity of Precambrian samples studied to date; (ii) the authors do not fully discuss the limited amount of low-δ56Fe pyrite produced in their model; and (iii) the isotopic fractionations favored by the authors reflect extreme conditions produced in the laboratory and are unlikely to be representative of natural processes.

We first address the Fe isotope record for Neoarchean and Paleoproterozoic marine sedimentary rocks because it bears on the Fe redox processes that Guilbaud et al. (1) dismiss. Fig. 1A shows Fe isotope compositions of Fe oxides and Fe carbonates in banded iron formations (BIFs), Fe-rich shales, and Fe-rich and Fe-poor Ca-Mg carbonates. In all of these samples, pyrite Fe comprises <20% of the total Fe budget, and for most samples, especially BIFs, pyrite is insignificant or absent. The model proposed by (1) cannot, therefore, explain low-δ56Fe values in such samples. Many Fe-poor, Ca-Mg carbonates have very negative δ56Fe values that have been interpreted to reflect the Fe isotope composition of a marine photic zone with low amounts of aqueous ferrous Fe [Fe(II)aq] after extensive Fe oxidation and precipitation of Fe oxide/hydroxides (2). The Fe-rich BIFs that were deposited coevally in the same basins have average δ56Fe values that provide sufficient mass balance to the Fe-poor Ca-Mg carbonate inventory, providing an example of an Fe redox couple (Fig. 1A). Indeed, as discussed in the references cited in Fig. 1, combining Fe isotope analyses with other geochemical evidence, notably carbon isotopes, indicates that the redox processes of microbial dissimilatory iron reduction (DIR) and Fe oxidation are likely means by which a >3 per mil (‰) range in δ56Fe values was produced.

Fig. 1

Fe isotope and chemical compositions of shales, BIFs, and carbonates of Neoarchean and Paleoproterozoic age. δ56Fe values are defined relative to the average of igneous rocks. (A) δ56Fe versus total Fe contents (FeT) for various types of sedimentary rocks for which pyrite Fe comprises <20% of the total Fe budget. Magnetite and hematite BIF samples, as well as many of the siderite BIF samples (1012), were collected from monomineralic layers, and thus their FeT values assume stoichiometries of Fe3O4 (magnetite), Fe2O3 (hematite), and FeCO3 (siderite) [72.4 weight percent (wt %) Fe, 70.0 wt % Fe, and 48.3 wt % Fe, respectively]. Total Fe values of all other samples (2, 1315) were measured from bulk rock digestions. All samples are from the Hamersley and Transvaal basins. (B) δ56Fe versus the percent of Fe that is pyrite Fe for shales that have low carbonate contents (<2 wt % Ccarb). Samples that contain low-δ56Fe pyrite formed by the model proposed by (1) would necessarily have low total Fe contents and a high percentage of Fe as pyrite (blue squares in the upper left region of plot) for such a signal to be detected in the bulk sample. All shale samples were analyzed as bulk rock digestions (2, 13, 14). It is important to note that the relations in Fig. 1B are cast in terms of percentage of Fe in pyrite, rather than the degree of pyritization, the latter of which references pyrite abundance to “reactive” (HCl-extractable) Fe, which may be problematic because of alterations to the reactive Fe pools during even small extents of metamorphism. (C) δ56Fe versus the ratio of FeT to total Al (AlT) for low-Ccarb shales [the samples from (B) for which AlT data are available] separated by the percentage of Fe that is pyrite Fe and by total Fe contents. These samples are shales that contain siliciclastic Fe, which will dilute any negative δ56Fe values contained in pyrite or other minerals (indicated schematically by the arrow). The FeT/AlT of average Archean shale is that of Taylor and McLennan (3).

We next turn to pyrite-bearing shales to test the relationship of δ56Fe and pyrite abundance predicted by the Guilbaud et al. (1) model. The model involves precipitation of ~10% of marine Fe(II)aq to form FeS, followed by reaction of FeS to pyrite; the maximum decrease in δ56Fe values for pyrite occurs when small amounts (<20%) of FeS are converted to pyrite, equivalent to ~2% pyritization of the initial marine Fe(II)aq inventory [~98% removal of Fe(II)aq]. Such signals are unlikely to be found in bulk shales, in which much of the record of negative δ56Fe values in Precambrian rocks lies, because a negative-δ56Fe signal for pyrite would generally be diluted by Fe-bearing silicates (δ56Fesilicates ~ 0‰). The low-δ56Fe signal proposed by (1) would be most likely detected in rocks that have low total Fe contents and large proportions of Fe as pyrite and is possibly recorded in less than 10% of the samples plotted in Fig. 1B that have such a combination. In contrast, the numerous Fe-rich shales that have low proportions of their Fe inventory in pyrite and negative δ56Fe values (Fig. 1B) cannot be explained by this model. Additionally, virtually all of the measured low-δ56Fe shales of Neoarchean and Paleoproterozoic age are enriched in Fe [FeT/AlT greater than 0.5, the value of average Archean shale (3)], and there is no correlation between Fe enrichment, δ56Fe value, and the proportion of pyrite in the Fe inventories of these samples (Fig. 1C). Focusing on samples that have >20% of their Fe inventory in pyrite, FeT/AlT– δ56Fe variations suggest the presence of pyrite that has a very low δ56Fe value (Fig. 1C). The fact that the majority of these rocks have FeT/AlT ratios equal to or greater than that of average Archean shale suggests, however, that the low-δ56Fe pyrite component is unlikely to reflect a process that occurred by the ~98% Fe(II)aq removal that is required by the model of Guilbaud et al. (1) and more likely reflects Fe addition. DIR has been recognized as an efficient “pump” for explaining net addition of low-δ56Fe Fe to deep sections of marine basins [e.g., (4)].

Next, we turn to the experimental results obtained by (1). Guilbaud et al. chose an initial kinetic isotope fractionation for precipitation of FeS from Fe(II)aq that was measured for FeS precipitated over time scales of minutes to hours (5, 6), and the very large kinetic fractionation between FeS and pyrite they measured was obtained over periods of hours. Although many experimental studies indicate that pyrite can be formed rapidly, studies of modern marine environments indicate rates of pyrite formation that are orders-of-magnitude slower than those of laboratory experiments (7). Kinetically induced stable isotope fractionations can depend upon rates [e.g., (8)], and yet this critical point is not discussed by (1). Thus, we contend that the fractionation factors used by (1) represent laboratory conditions that may not be reflective of natural conditions, especially considering that 65 to 77% isotopic re-equilibration occurs between FeS and Fe(II)aq in 2 to 4 days at 25°C (9). Indeed, the lack of correlation in the experiments between percent pyritization and the measured Fe isotope fractionation factor [figure 1 in (1)] suggests that these data may largely represent experimentally induced kinetic effects.

We conclude that the Fe isotope compositions of Neoarchean and Paleoproterozoic marine sedimentary rocks are the result of numerous processes, including abiological and biological Fe redox processes. It is possible that some low-Fe samples preserve the signal proposed by (1), although their model is not a likely explanation for many low-δ56Fe samples from the rock record, and their model does not address the temporal changes in the Fe isotope compositions of Precambrian sedimentary rocks.

References and Notes

  1. Acknowledgments: This work was supported by the NASA Astrobiology Institute.
View Abstract

Navigate This Article