Iron Isotope Biosignatures

See allHide authors and affiliations

Science  17 Sep 1999:
Vol. 285, Issue 5435, pp. 1889-1892
DOI: 10.1126/science.285.5435.1889


The 56Fe/54Fe of Fe-bearing phases precipitated in sedimentary environments varies by 2.5 per mil (δ56Fe values of +0.9 to −1.6 per mil). In contrast, the56Fe/54Fe of Fe-bearing phases in igneous rocks from Earth and the moon does not vary measurably (δ56Fe = 0.0 ± 0.3 per mil). Experiments with dissimilatory Fe-reducing bacteria of the genus Shewanella algae grown on a ferrihydrite substrate indicate that the δ56Fe of ferrous Fe in solution is isotopically lighter than the ferrihydrite substrate by 1.3 per mil. Therefore, the range in δ56Fe values of sedimentary rocks may reflect biogenic fractionation, and the isotopic composition of Fe may be used to trace the distribution of microorganisms in modern and ancient Earth.

Fractionation of light stable isotopes such as C, O, N, and S is controlled by inorganic processes related to temperature changes and phase transitions, and by biological processes (1). This dual control can make it difficult to interpret the origin of isotopic differences in rocks. For example, excursions in δ13C values in deep-sea sediments can be interpreted as a function of changes in the productivity of the oceans or the partial pressure of CO2 of the atmosphere (2). In contrast, intermediate-mass elements such as Fe may not be fractionated substantially by inorganic processes because the relative mass difference between Fe isotopes is less than that of C, O, N, or S isotopes. However, biological processes may produce measurable Fe-isotopic fractionation because the metabolic processing of Fe involves a number of steps, such as transport across membranes and uptake by enzymes (3), that may fractionate isotopes.

Few studies have documented biological fractionation of transition metal elements (4) because of the difficulty of measuring precisely the isotopic ratios of transition metals (5). Thermal ionization mass spectrometry (TIMS) can produce high-precision isotope ratio measurements of these metals and, in the case of Fe, is not subject to large interferences by Ar-containing species (for example, 40Ar16O and40Ar14N), as is inductively coupled plasma mass spectrometry. However, TIMS produces large mass-dependent isotope fractionations during the course of a measurement, which must be corrected before the natural isotopic composition of a sample can be determined. Previous attempts to correct instrumental, mass-dependent isotopic fractionation of Fe used an empirical approach that produced data with a 1σ precision of only 2 to 3 per mil for56Fe/54Fe (6), an uncertainty that exceeds the range in nature. Here we used a mixed double spike to correct for instrumental mass bias (7–9). With the use of this technique, it is possible to make Fe isotope ratio measurements that are precise to ±0.2 to 0.3 per mil (1σ) for56Fe/54Fe. We report Fe-isotopic ratios in conventional per mil notation:Embedded Image Embedded Imagewhere (56Fe/54Fe)E-Mis the average 56Fe/54Fe measured for 15 terrestrial igneous rocks, ranging in composition from peridotite to rhyolite, and five high-Ti lunar basalts. The average56Fe/54Fe measured for the Earth-moon system is 15.7028 (7). Terrestrial and lunar rocks comprise an isotopically homogenous igneous iron reservoir that is thought to represent the bulk isotopic composition of Earth and the moon.

Two sets of experiments were performed to determine the magnitude of Fe-isotopic fractionation that might be produced by microorganisms. Experiment 1, run in duplicate at the University of Wisconsin–Milwaukee (U.W.-Milwaukee), used S. algae(strains BCM 8 and BrY) grown on a ferrihydrite substrate in an LM growth medium (10). After inoculation of the ferrihydrite + growth medium solution with the cells, the bacteria were allowed to reduce Fe for 8 hours. One run was contained in a 10,000 molecular weight dialysis bag (ferrihydrite + cells + growth medium) suspended in a 100-ml flask of LM growth medium, which allowed the ferrihydrite and cells to be removed from the growth medium. The other run was done in a 100-ml flask, followed by separation of the ferrihydrite and cells from the growth medium + hydrolyzed Fe(II), by filtration. Ferrous iron in the growth medium solution was precipitated by adding ultrapure ammonia, to increase the pH to 9 to 10 immediately after the solution was exposed to the atmosphere, and then allowed to sit for 3 to 5 days. Abiological control experiments were run in parallel; addition of ammonia did not precipitate any iron, confirming that no substantial amount of ferrous Fe was generated. The ammonia precipitation procedure effectively removes the Fe(II) from the growth medium as a ferric oxyhydroxide, which was centrifuged and washed three times in doubly distilled H2O. The precipitate was dissolved in 6 M HCl, after which followed chemical processing and isotopic analysis with the methods of (7, 8).

Experiment 2, performed at the Jet Propulsion Laboratory (JPL), usedS. algae BrY grown on ferrihydrite in an LB growth medium (11). Three runs were made, harvesting Fe 13, 15, and 23 days after inoculation, which produced Fe(II) contents of 5.5, 11.1, and 35.6 parts per million (ppm) Fe, respectively (12). Each solution was sterilized with a 0.2-μm filter. Reacted ferrihydrite from runs 2 and 3 was saved for Fe isotope analysis. Splits (50 ml) of the Fe(II) solutions were evaporated to dryness and the organic material combusted in quartz crucibles in a muffle furnace at 700°C for 8 hours. The remaining solids were dissolved in 6 M HCl and processed for Fe isotope analysis as in (7, 8). A parallel set of 50-ml aliquots from each run was processed, for comparison with the samples that were combusted, where Fe was harvested as an oxyhydroxide using 10 ml of 30% H2O2 in ammonia to bring the pH to 9 to 10. The precipitated Fe oxyhydroxide was treated in the same manner as in the U.W.-Milwaukee experiments. The JPL experiments also included an abiological control, which was harvested by the H2O2 + ammonia precipitation technique, as well as by the evaporation and combustion technique.

The high levels of Fe(II) that are produced in the experiments that contain bacteria, relative to those in the abiological controls, are interpreted to reflect biologically produced Fe(II) because the54Fe/56Fe of the Fe(II) is significantly less than the 56Fe/54Fe ratio of the starting ferrihydrite substrate (Fig. 1 and Table 1). These data show that the Fe-reducing bacteria S. algae preferentially reduce 54Fe relative to 56Fe, as reflected in a 1.3 per mil negative shift in δ56Fe values. The U.W.-Milwaukee experiments (contained in dialysis bags) do not display as large an iron isotope fractionation as the experiments that were run in flasks—the differences in the isotopic compositions are smaller than the analytical error of the measurements—but the results are consistent (Table 1). Although Fe(II) concentrations were not measured in the U.W.-Milwaukee experiments, clumping of the ferrihydrite in the dialysis bag experiments effectively reduced the amount of substrate surface area available to the bacterium. The armoring of the ferrihydrite by clumping probably drove the reaction further to completion, toward the original isotopic composition of the starting ferrihydrite, which would have produced higher δ56Fe(II) values in the experiments that were contained in dialysis bags, compared with the flask experiments.

Figure 1

(A) δ56Fe versus the inverse of iron concentration (in parts per million) of the Fe produced from Fe-reducing bacteria, starting ferrihydrite, and abiological control of the experiments conducted at JPL. The Fe-isotopic composition of the abiological control can be explained by mixing a low δ56Fe value that is inferred from the growth media and the high δ56Fe value of the ferrihydrite (see text for discussion). The low δ56Fe of the iron reduced by bacteria is best explained by mass-dependent fractionation caused by bacteria preferentially reducing54Fe relative to 56Fe. The curves are closed system Rayleigh distillation curves with αferrihydrite-Fe(II) values of 1.0013 (solid curve) and 1.0026 (dashed curve). Tick marks for the curve with an α value of 1.0013 represent the proportion of ferrihydrite consumed and from right to left are 0.003, 0.004, 0.005, 0.01, 0.02, 0.03, 0.04, 0.05, 0.1, 0.2, 0.4, 0.6 and 0.8. FH, ferrihydrite; FSC, ferrihydrite starting composition; AC, abiological control; H1, Fe(II) from harvest 1; H2, Fe(II) from harvest 2; H3, Fe(II) from harvest 3. (B) δ56Fe versus the proportion of ferrihydrite consumed by Fe-reducing bacteria. The Fe isotope composition of the ferrous Fe produced by bacteria and the reacted ferrihydrite may be consistent with a closed system Rayleigh distillation process that has a fractionation factor of 1.3 per mil for the 56Fe/54Fe. FH2, ferrihydrite harvest 2; FH3, ferrihydrite harvest 3.

Table 1

Fe isotope compositions of sedimentary Fe and bacteria-mediated Fe products. Pacific Ocean nodules collected at the following: sample 1, 16°N, 125°W; sample 2, 57°S, 95°W; sample 3, 10°N, 136°W; sample 4, 35°N, 160°W. Atlantic nodule sample 5 was collected from the Blake Plateau. Analytical methods are reported in (7, 8). Isotopic ratios are the averages of “n” duplicate isotopic analyses. Errors in the δ56Fe values reflect internal statistics (n = 1) or external error based on duplicates (n ≥ 2).

View this table:

The JPL experiments were scaled so that the isotopic composition of the abiological control could be analyzed. This control has an anomalously low δ56Fe compared with the Fe(II) from the bacterial experiment. Moreover, there are significant differences between the Fe-isotopic compositions determined by the precipitation technique and the evaporation and combustion technique, for harvest 1 as well as for the abiological control (Fig. 1 and Table 1). The low δ56Fe of the abiological control can be interpreted in two ways. One explanation is that partial dissolution of ferrihydrite occurs in the complex organic medium, and this produces significant Fe isotope fractionation. If the low δ56Fe of the abiological control is a result of isotopic fractionation caused by dissolution of ferrihydrite in the growth medium, the fractionation factor (Fe(II)-ferrihydrite) would be about −2.6 per mil for56Fe/54Fe, as calculated from the Fe contents of the abiological control (0.34% of the total ferrihydrite). An alternative explanation is that the low δ56Fe value of the abiological control represents a mixture between Fe of normal isotopic composition (no isotopic fractionation during partial dissolution of ferrihydrite) and the Fe blank of the growth medium with an anomalously low δ56Fe. The JPL growth medium has an Fe blank of 0.32 ppm, the bulk of which comes from the tryptone and yeast extract. We have been unable to measure precisely the Fe isotopic composition of the growth medium because of the low Fe contents and the processing complexities associated with the organic matrix. If the low δ56Fe of the abiological control reflects a mixture of growth medium blank and normal Fe (partially dissolved ferrihydrite), then mass balance calculations show that the growth medium blank must have a δ56Fe of −8 per mil (Fig. 1A). Such a low δ56Fe could be produced through multicyclic processing of Fe by yeast during preparation of the commercial yeast extract.

The Fe-isotopic compositions of the Fe(II) produced by Fe-reducing bacteria grown on ferrihydrite are consistent only with biologically produced Fe isotope fractionation, despite the fact that the abiological control produced low δ56Fe values. The Fe(II) contents of the biological experiments (5.5 to 35.6 ppm Fe) are too large to have been affected by the Fe blank of the growth medium or the abiological control, even if the abiological control reflects only isotopic fractionation produced during the dissolution of ferrihydrite (Fig. 1). For example, if there was no biological fractionation of Fe isotopes and the bacteria served only as a catalyst for the ferrihydrite dissolution that was accompanied by Fe isotope fractionation, then the products harvested from the three experiments should lie along a Rayleigh distillation curve that defines a fractionation factor (Fe(II)-ferrihydrite) of −2.6 per mil (Fig. 1). Alternatively, if the Fe(II) in the experiments reflects a mixture of biologically produced Fe that was not isotopically fractionated and a “blank” of low δ56Fe (abiological control), then the isotopic compositions of the biological experiments should plot along a mixing line defined by the δ56Fe of the stock ferrihydrite and the abiological control (Fig. 1). Neither of these explanations is consistent with the relation observed between δ56Fe and 1/Fe, and it seems most likely that Fe(II) produced in the bacteria-containing experiments undergoes biologically produced Fe-isotopic fractionation.

The differences between the Fe-isotopic compositions of the Fe(II) harvested by the precipitation technique, and of that harvested by the combustion technique for the abiological control and harvest 1 (which contained the lowest Fe contents) probably reflect differing proportions of biologically produced Fe and Fe blank. The evaporation and combustion method includes Fe(II) that has been produced by bacteria and any Fe blank in the growth media (which may have a low δ56Fe, as indicated by the abiological control). In contrast, the Fe precipitation method most likely contains only Fe that has been processed by bacteria if the blank Fe of the growth media is complexed by an organic ligand.

The Fe-isotopic fractionation produced by S. algae BrY grown on a ferrihydrite substrate may follow a Rayleigh distillation law (Fig. 1B), where the fractionation factor (ferrihydrite-Fe(II)) is 1.3 per mil for 56Fe/54Fe. If confirmed by longer experimental runs, extreme isotopic fractionations may be produced in the remaining ferrihydrite substrate. For example, when 80% of the substrate is consumed, the remaining fraction would have a δ56Fe of +1.8 per mil and the ferrous iron produced would have a δ56Fe of −0.8 per mil. Low57Fe/56Fe values have been reported for Fe that has been processed by Fe-reducing bacteria (13).

We have measured anomalous Fe-isotopic compositions in iron-bearing minerals from sedimentary environments, including Fe-Mn nodules from the Pacific and Atlantic Oceans and individual layers from banded iron formations (Fig. 2 and Table 1). Modern sediments (Fe-Mn nodules) and ancient sedimentary rocks (Precambrian banded iron formations) show 2 to 3 per mil Fe-isotopic variations that can be explained by bacterially produced fractionation, given the constancy of Fe-isotopic compositions measured in igneous rocks from Earth and the moon (7).

Figure 2

δ56Fe of iron deposited in sedimentary environments. The shaded band is the baseline for the isotopic composition of inorganic Fe, as determined from analysis of a variety of igneous rocks from Earth and the moon (7). The measured range of Fe isotope compositions in natural samples is 2.5 per mil, which is an order of magnitude larger than the analytical uncertainty. Variations in the measured Fe isotope composition of these natural samples are best explained by biological processes (see text for discussion).

Two banded Fe formations have been analyzed, a Proterozoic sample from the Empire Mine in Michigan and an Archean sample from the Sudan Mine in Minnesota. δ56Fe values of the dark-colored Fe-rich layers from the Archean and Proterozoic samples are close to zero. In contrast, the light-colored, Fe-poor layers have positive δ56Fe values: The δ56Fe values of the red layers of the Archean sample overlap that of the bulk earth, whereas the green layer from the Proterozoic sample has a positive δ56Fe value of +0.9 per mil.

Our experimental results predict that the light-colored, Fe-poor layers of banded iron formations should have positive δ56Fe values if Fe-reducing bacteria were involved in Fe mobilization. Because mobile (reduced) Fe should have negative δ56Fe values (Fig. 1), we infer that loss of reduced Fe during the development of banded iron formations should produce Fe-poor residues with high δ56Fe values. It has been proposed that Fe(III) in sediments may be reduced by bacteria, dissolved, and transported from its place of deposition during the genesis of banded iron formations (14), which would leave behind a substrate with a positive δ56Fe. In contrast, if dark, Fe-poor layers result from inorganic Fe precipitation, then no isotopic anomalies would be expected. It is unclear if the lack of large Fe-isotopic anomalies in the Archean formation indicates a lack of biologic activity or if the Fe-isotopic composition of this sample was homogenized by supergene enrichment, as suggested by the specular hematite veins that cross cut layers in the analyzed sample. High57Fe/56Fe in aquifer sediment leachates and low57Fe/56Fe for groundwater from the same site (4) are thought to reflect mass balance in a biologically mediated system.

  • * To whom correspondence should be addressed. E-mail: beardb{at}


View Abstract

Stay Connected to Science

Navigate This Article