Potentially biogenic carbon preserved in a 4.1 billion-year-old zircon

From an initial population of over 10,000 Jack Hills zircons (6), we examined 656 grains with ages over 3.8 Ga for the presence of graphitic inclusions. The zircons were mounted in epoxy and polished to expose their interiors. The search protocol included an initial screening for opaque inclusions using transmitted light microscopy. Seventy-nine candidates thus identified were then targeted for Raman spectroscopy from which we documented two zircons containing partially disordered graphite (Fig. 1, Inset) beneath their polished surfaces (RSES 81-10.14 in a cracked region; RSES 61-18.8 in a crack-free region). We did not consider RSES 81-10.14 further due to the potential for contamination via ingress on cracks.

A concordant U-Pb age of 4.10 ± 0.01 Ga was obtained on a polished internal surface of zircon RSES 61-18.8 (6). Its low U content (∼100 ppm; Supporting Information) minimizes the potential for radiation damage and is a contributing cause for its 99% U-Pb concordancy (6). A roughly 30 × 60 × 20-µm sliver containing two carbonaceous phases was milled using a Ga⁺ focused ion beam (FIB) and attached to a tungsten needle via a platinum weld for synchrotron transmission X-ray microscopy (19) at beam line 6-2c of the Stanford Synchrotron Radiation Lightsource (SSRL). The 40-nm spatial resolution of this imaging method revealed no through-going cracks or defects associated with these inclusions that exhibit the graphite crystal habit (Fig. 1; also Supporting Information). Due to their isolation within a crack-free region of an isotopically undisturbed zircon, conditions shown to preserve primary inclusion assemblages (20), we interpret these graphitic inclusions to have been incorporated during crystallization of this igneous zircon.

The zircon sliver was remounted in indium, coated with gold, and analyzed for δ¹³C on a CAMECA ims1270 secondary ion mass spectrometer (SIMS) using a 5-nA Cs⁺ primary beam (21) with static multicollection following a 2-min cleaning via rastering of the primary beam. Note that the two inclusions had not previously been exposed to epoxy or any other source of laboratory carbon contamination and were revealed only by drilling with the primary Cs⁺ ion beam. During analysis of the smaller inclusion, we found that count rates increased to a maximum value as the inclusion was breached followed by a tailing to background (Supporting Information). This precludes the possibility that our data were affected by surface contamination.

SIMS measurement of C isotopes in carbonaceous materials is known to be little affected by instrumental mass fractionation. For example, a previous study found less than 2‰ variation in an intercomparison of hydrocarbon standards that varied in H/C by a factor of 7 (21). Thus, we used epoxy, with δ¹³C of −26.8‰ (21), as a standard material for instrumental mass fractionation corrections and Escherichia coli with δ¹³C of −24‰ (21) as a secondary standard. The two inclusions yielded similar results in both carbon abundance and δ¹³C. Maximum signals of ∼10⁵ counts per second (cps) of ¹²C₂⁺ for the inclusions was at least an order of magnitude higher than for the C background in the indium mounting material (∼10⁴ cps) and much higher than a clean area in the zircon (∼10³ cps). The δ¹³C values of the two inclusions are indistinguishable within error, and thus we combined the results to obtain an average of −24 ± 5‰. This is notably light with respect to terrestrial inorganic carbon and most meteoritic carbon, but consistent with terrestrial biogenic carbon (Fig. 2). Following carbon isotopic analysis, we measured rare earth element (REE) and Ti abundances using the ims1270 with a 15-nA O⁻ primary beam. Two spots yielding measurable Ti correspond to an average apparent crystallization temperature of ∼660 °C, similar to the average Hadean Jack Hills value (9), with REE consistent with crystallization under reduced conditions (using the method of ref. 22) and a continental trace element signature (following the discriminant diagrams of ref. 23).

Given the isolation of these inclusions provided by their entrapment in a crack-free and isotopically undisturbed zircon crystal, as well as the vanishingly slow self-diffusion of carbon in graphite under crustal conditions (24), we interpret them to be petrologically primary and isotopically closed systems. Because these inclusions are at least as old as their 4.10-Ga host, the observation of isotopically light carbon raises the possibility that biologic processes were operating during the Hadean. To be sure, not all light carbon isotopic signals are biogenic.

Abiotic processes that could produce light δ¹³C during the Hadean include Fischer–Tropsch mechanisms (25) and carbon isotopic fractionation by diffusion (26), incorporation of meteoritic (ranging from +68‰ to −60‰ δ¹³C) materials, mid-ocean ridge basalt degassing (27), and high-temperature disproportionation of siderite (28). However, in the absence of experimental evidence for particulate carbon formation via a Fischer–Tropsch process (ref. 25; cf. ref. 27) or diffusive or mineral disproportionation mechanisms that could selectively lead to light carbon condensates that were ultimately incorporated in a low-temperature (i.e., ∼650 °C) granitoid zircon, a biogenic origin seems at least as plausible. The quantity of extraterrestrial carbon necessary to dominate the carbon budget of a felsic magma protolith would be significant and profoundly influence the chemistry of that magma away from its granitoid character unless separated from the host meteorite. A possible mechanism to effect this on ancient Earth might be through sedimentary processing and concentration of meteoritic carbon in pelagic sediments. Even if meteoritic carbon could be concentrated in this fashion, our observed isotopic signal is uncharacteristic of the most likely candidates, carbonaceous chondrites, which contain on average 3.5 wt% C (29). Although δ¹³C of various components ranges between +68‰ and −34‰ (30), only 10% of bulk carbon measurements have δ¹³C less than −20‰ (31), and thus this class of meteorites would be a fortuitous, although possible, source for both inclusions. Our view that isotopically light C in ancient terrestrial materials may be of biogenic origin is consistent with interpretations of light δ¹³C in mantle diamonds (e.g., ref. 32). Although diamonds as low as approximately −10‰ are able to be accounted for by isotopic fractionation of abiogenic precursors, more negative values down to −23‰ do not appear to be explicable by such processes and strongly suggest an origin from subducted organic matter (33). Although we cannot rule out an abiogenic source, the eventual acquisition of a large database of Hadean carbon isotope measurements will permit the plausibility of hypothetical abiogenic origins to be more closely scrutinized.

By contrast, biogenic organic matter is incorporated into the vast majority of preserved clastic sediments, can reach 10% by weight in Phanerozoic sediments (34), and has averaged δ¹³C approximately −25‰ over the past 3.5 Ga (13). Given this historical trend and the several lines of evidence indicative of sediment involvement in the Jack Hills zircon magmas (8, 10–12), the simplest interpretation of our data are that the carbonaceous inclusions represent graphitized organic carbon present during melting of a pelitic protolith at 4.10 Ga. This interpretation is entirely consistent with the observed low crystallization temperature, continental affinity, and reduced environment (35) of zircon formation.

The isolated, primary nature of the graphite in sample RSES 61-18.8, along with the lack of diamond features in its Raman spectrum, differs from the abundant graphite previously reported (16, 17) in Jack Hills zircons, which was found often intersecting cracks in the host zircon, and all of which were reported to contain diamond (17). As noted, the diamonds were later found to be contamination (18). The rare occurrence of primary graphite in this study (in <0.2% of >3.8-Ga zircons) also contrasts with these earlier results (graphite in 4% of zircons of all ages), further suggesting that much or all of the earlier-purported graphite identified was contamination.

There are considerable limitations of basing any inference regarding early Earth on a single zircon containing primary carbonaceous inclusions. Instead, we see this contribution as demonstrating the feasibility of perhaps the only approach that could lead to establishing a Hadean carbon isotope record. In this regard, we emphasize that because >3.8-Ga grains make up only ∼5% of the Jack Hills zircon population (6), RSES 61-18.8 represents an abundance of carbon-bearing Jack Hills zircons of only about 1-in-10,000.

This study extends the terrestrial carbon isotope record ∼300 My beyond the previously oldest-measured samples from southwest Greenland. Our interpretation that the light C isotope signature in primary graphitic inclusions could reflect biologic processes is consistent with an estimate from molecular divergence in prokaryote phylogenetic relationships that a terrestrial biosphere had emerged by 4.1 Ga (36). Confirming such a connection would represent a potentially transformational scientific advance. However, given the low occurrence of carbon-bearing Hadean zircons, establishing a Hadean carbon cycle and its possible bearing on the origin of life will require enormous and sustained efforts.

Zircons were mounted in epoxy and polished to expose their interiors, typically involving loss of approximately one-third of the zircon during polishing. Polishing was accomplished with 1,200-grit silicon carbide paper and 1-µm diamond paste. The search protocol included an initial screening for opaque inclusions using transmitted light microscopy and a 40× objective lens.

Raman spectroscopy by Anatoliy Kudryavtsev, Anke Watenpuhl, Olivier Beyssac, and Daniel Wilkinson, using confocal Raman systems in the laboratories of William Schopf, Andrew Steele, Olivier Beyssac, and Bruce Dunn, made possible our search for carbonaceous inclusions. FIB milling by Noah Bodzin, X-ray imaging by Wendy Mao and Crystal Shi [who are supported by National Science Foundation (NSF)–Division of Earth Sciences (EAR) Grant 1055454], and sample mounting by Christopher Snead were also essential. We thank Elizabeth Boehnke for X-ray image reconstruction and for Movie S2. We thank the three reviewers, whose comments greatly improved our manuscript. This project was funded by a Simons Collaboration on the Origin of Life Postdoctoral Fellowship (to E.A.B.) and NSF-EAR Grant 0948724 (to T.M.H.). The UCLA ion microprobe facility is partially supported by a grant from the Instrumentation and Facilities Program of EAR, NSF. Data reported in this manuscript are presented in full in Supporting Information. Portions of this research were carried out at the SSRL, a Directorate of SLAC National Accelerator Laboratory and an Office of Science User Facility operated for the US Department of Energy, Office of Science by Stanford University.