Carbonates in the Martian meteorite Allan Hills 84001 formed at 18 ± 4 °C in a near-surface aqueous environment

Abstract

Despite evidence for liquid water at the surface of Mars during the Noachian epoch, the temperature of early aqueous environments has been impossible to establish, raising questions of whether the surface of Mars was ever warmer than today. We address this problem by determining the precipitation temperature of secondary carbonate minerals preserved in the oldest known sample of Mars’ crust—the approximately 4.1 billion-year-old meteorite Allan Hills 84001 (ALH84001). The formation environment of these carbonates, which are constrained to be slightly younger than the crystallization age of the rock (i.e., 3.9 to 4.0 billion years), has been poorly understood, hindering insight into the hydrologic and carbon cycles of earliest Mars. Using “clumped” isotope thermometry we find that the carbonates in ALH84001 precipitated at a temperature of approximately 18 °C, with water and carbon dioxide derived from the ancient Martian atmosphere. Furthermore, covarying carbonate carbon and oxygen isotope ratios are constrained to have formed at constant, low temperatures, pointing to deposition from a gradually evaporating, subsurface water body—likely a shallow aquifer (meters to tens of meters below the surface). Despite the mild temperatures, the apparently ephemeral nature of water in this environment leaves open the question of its habitability.

The extreme difficulty in achieving mild surface temperatures in early Mars climate models (1) is in disagreement with widespread geomorphological evidence for surface water runoff during the Noachian epoch (2, 3). A robust determination of the temperature at or near the surface of Noachian Mars would provide insight into this apparent paradox, but has been challenging to establish. Carbonate minerals can record and preserve information regarding the temperature and chemistry of their formation environment through aspects of their oxygen and carbon isotopic composition (δ¹⁸O and δ¹³C, respectively). These minerals have been observed in a range of Martian geological materials, including dust, bedrock outcrops, and several meteorites (4–9). Collectively, these observations highlight a role for carbonate formation in the global Martian carbon cycle and the evolution of its atmosphere. The oldest known carbonate from Mars, or any other planetary body, occurs as a minor constituent (approximately 1% by weight) in the meteorite Allan Hills 84001 (ALH84001), which has a crystallization age of approximately 4.1 billion years (10). The carbonate in ALH84001, geochronologically constrained to be slightly younger in age [between 3.9 and 4.0 billion years old; (11)], provides a unique window into the hydrologic cycle, carbon cycle, and climate of Noachian Mars.

Texturally, the carbonates in ALH84001 occur as (i) chemically and isotopically zoned ovoid concretions, veins, and void fillings and (ii) regions of massive carbonate, variably intergrown with unweathered feldspathic glass and orthopyroxene, the latter of which makes up the bulk of the meteorite (12–14). The concretions display wide systematic variation in chemical and isotopic composition. They have Ca-Fe-rich cores with (O isotope ratios reported relative to Standard Mean Ocean Water) values as low as approximately 5‰ and (C isotope ratios reported relative to the Pee Dee Belemnite standard) values of 25–35‰ (15–18). Their rims are Mg-rich, sometimes nearly pure MgCO₃, and have and values as high as approximately 30‰ and 65‰, respectively. The veins and void fillings display covarying chemical and isotopic compositions similar to the concretions (12). The massive carbonates are usually internally homogeneous, though chemical and isotopic variability has been observed between isolated occurrences. Several studies have reported an ankeritic (Ca-rich) and isotopically light ( of 0–10‰) composition for the massive carbonates (e.g., ref. 16).

The carbonates in ALH84001 have been studied by a variety of geochemical and petrographic methods, but no agreement exists concerning the processes and environment of their precipitation. Hypotheses for their formation span a range of temperatures between subfreezing and more than 700 °C, a range of depths between the surface and several kilometers, and a range of processes including impact melting of preexisting carbonates, hydrothermal alteration of crustal rocks, precipitation from a rapidly evaporating water body, and biological activity (13–16, 19–28). Textural observations indicate that after carbonate growth, ALH84001 experienced brief, localized, shock-related heating, but natural remanent magnetization and argon thermochronometry place upper limits of approximately 40 °C on the short-duration (approximately 10 min) and approximately 20–30 °C on the long-duration (billions of years) thermal history of this rock (29, 30).

Previous methods for estimating the temperature of carbonate formation in ALH84001, though informative, have placed only loose constraints (e.g., refs. 29 and 30) or have depended on assumptions about the chemical and isotopic composition of the carbonate’s parent fluid (e.g., refs. 16 and 22). Multiply substituted, or “clumped,” isotope thermometry avoids this problem because it is based on homogeneous thermodynamic equilibrium within the carbonate mineral (31, 32). The abundance of rare isotopologues, measured as mass 47 CO₂ (mostly ¹³C¹⁸O¹⁶O) evolved from acid digestion of carbonate minerals, is higher than predicted from a random distribution of the heavy isotopes. This excess is denoted Δ₄₇ and varies systematically with carbonate precipitation temperature (32, 33). With temperature known, measurements of the δ¹⁸O and δ¹³C of the carbonates provide direct constraints on the isotopic compositions of their parent fluid (water and CO₂).

Results

We performed stepped phosphoric acid digestions of three aliquots of ALH84001 and measured the bulk and clumped isotopic composition of the evolved CO₂ (see Methods and SI Text). The amount of CO₂ evolved from the digestion steps was about one-tenth to one-third the amount typically required for clumped isotope analysis. Additionally, the acid digestion duration was up to 36 times longer than typically applied, increasing the potential for contamination. This combination of factors necessitated special extraction apparatus, purification steps, and analytical protocols (SI Text, Table S1, and Figs. S1–S3). The δ¹⁸O and δ¹³C values of the carbonate extracted in the three digestion steps (Table 1 and Table S2) were reproducible among the three different meteorite aliquots and fall within the range of previous measurements (15–18, 21, 22). Combining these δ¹⁸O and δ¹³C measurements with prior constraints on the relationship between stable isotope and major element composition in ALH84001 carbonates (15–18), we estimated the chemical composition of carbonate sampled by the digestion steps (Fig. 1 and SI Text). The sequential extraction technique yielded increasing Mg content with reaction time, a relationship expected from the slower dissolution of the Mg-rich outer concretion layers (34).

Table 1.

Isotopic composition of ALH84001 carbonates extracted by stepped acid digestion

	0–1 h					1–4 h					4–12 h
	N	δ18O‰	δ13C‰	Δ47‰	T °C	N	δ18O‰	δ13C‰	Δ47‰	T °C	N	δ18O‰	δ13C‰	Δ47‰	T °C
Aliquot A (320 mg)	—	13.05	35.52	—	—	—	16.68	40.87	—	—	—	18.96	40.82	—	—
Aliquot B (1,625 mg)	5	13.38 (±0.28)	34.60 (±0.07)	0.642 (±0.039)	27	5	17.32 (±0.28)	41.05 (±0.07)	0.638 (±0.039)	28	1	17.99 (±0.28)	40.94 (±0.07)	0.742 (±0.098)	3
Aliquot C (1,550 mg)	8	14.11 (±0.16)	35.14 (±0.02)	0.693 (±0.027)	14	9	16.27 (±0.16)	40.67 (±0.02)	0.643 (±0.027)	26	7	18.17 (±0.16)	42.39 (±0.02)	0.711 (±0.028)	10

δ¹⁸O and δ¹³C are reported relative to the SMOW and PDB standards, respectively. N is the number of data acquisitions suitable for clumped isotope analysis (typical clumped isotope measurements comprise eight acquisitions), reflecting the gas yield from the digestion steps. Error estimates are based on identical treatment of carbonate standards of known bulk and clumped isotopic composition, which was only done for aliquots B and C (SI Text)

Fig. 1.

Major element composition of the ALH84001 carbonates digested in the three steps, estimated from their δ¹⁸O and δ¹³C values and correlations between major element and isotopic composition (see text).

The first aliquot of the meteorite (approximately 320 mg) yielded too little gas for reliable Δ₄₇ measurements, but has δ¹⁸O and δ¹³C values that are consistent with subsequent extractions of two larger aliquots (1,625 and 1,550 mg, respectively). The Δ₄₇ measurements of these two larger aliquots (Table 1) yield six temperatures between and , all within uncertainty of an error-weighted average temperature of 18 ± 4 °C.

Discussion

Establishing the Robustness of the Temperature Determination.

We addressed potential sources of uncertainty (outside of analytical uncertainty) in interpreting the Δ₄₇ data as a precipitation temperature for the carbonates in ALH84001. The crystal orientation distribution of the carbonate concretions (SI Text and Fig. S4) suggests that the carbonates have not been subject to recrystallization since their radial growth from aqueous solution. In the absence of recrystallization, resetting of the carbonates’ isotopic composition to yield the observed, strongly covarying relationship between δ¹⁸O and δ¹³C is unlikely. Together, these observations indicate that the measured Δ₄₇ was acquired during initial carbonate precipitation.

Precipitation at high temperature and subsequent reordering of carbonate clumped isotopes upon cooling is unlikely because the temperature recorded under such a scenario (i.e., blocking temperature) would be approximately 200–300 °C (35, 36). Mixing between two physical or isotopic end members with bulk isotopic composition at either end of the observed δ¹³C and δ¹⁸O range in ALH84001 carbonate may lead to spuriously low apparent temperatures, but is also unlikely, as indicated both by previous observations of the distribution of ALH84001 carbonate chemical and isotopic composition (15–18) and by limits on the time scale for isotopic equilibrium (SI Text and Fig. S5).

Several additional hypotheses for interpreting the Δ₄₇ data can be rejected on the basis of the relationships observed among the δ¹⁸O, δ¹³C, and Δ₄₇ measured in the carbonate (Fig. 2). Carbonate precipitation over a range of temperatures (e.g., ref. 22) does not fit the observed δ¹⁸O and δ¹³C array (Fig. 2A). Because the stepped acid digestion could not perfectly separate the concretion cores from their rims, the range of we observed was narrow (approximately 13–19‰) relative to the full range observed in microprobe studies (approximately 5–30‰). As a result, the uncertainty in a linear fit to the reliable δ¹⁸O-Δ₄₇ data (aliquots B and C) is large at the ends of the full δ¹⁸O range (Fig. 2B, dashed curves). Thus, in a formal hypothesis testing sense, the 95% confidence bounds marginally include an explanation of the data by kinetic isotope effects (KIEs) of the sort observed in some cave and cryogenic carbonates (37–40). These KIEs, which are thought to arise in part from rapid, disequilibrium CO₂ degassing, have been observed to lead to spurious apparent values of Δ₄₇ that are lower than the true values by as much as 0.12‰. Applied to our measured Δ₄₇, this would translate into an error-weighted average temperature of carbonate formation in ALH84001 of -7 °C (instead of approximately 18 °C if the Δ₄₇ we measure is correct). Temperatures below 0 °C can be rejected because they would cause the water from which the carbonates formed to freeze and are thus inconsistent with the increase in δ¹⁸O from the concretion cores to their rims. The remaining temperature range between 0 and 18 °C is possible, but the large uncertainty in the fit and the better correspondence to a hypothesis of constant temperature (Fig. 2B) lead us to conclude that KIEs of the type discussed above are unlikely in the case of ALH84001 carbonates. Altogether the data are best explained by carbonate precipitation at a constant temperature of approximately 18 °C. This result rules out a high-temperature origin of ALH84001 carbonates (e.g., refs. 13, 14, and 19) and imposes strict constraints on the remaining low-temperature hypotheses.

Fig. 2.

Comparison of Δ₄₇, δ¹⁸O, and δ¹³C values (see SI Text) with models for the isotopic variation in ALH84001 carbonates. (A) Relationship between δ¹³C and δ¹⁸O. The solid black line and the broken black curves are a weighted total least squares linear fit to the data and a 95% confidence bound on the fit, respectively. The diagonal gray line shows the equilibrium relationship between δ¹³C and δ¹⁸O over the range of temperatures required to explain the range of observed in microprobe studies (approximately 5–30‰). A hypothesis of evolving temperature can be rejected on the basis of the misfit between these lines. (B) Relationship between Δ₄₇ and δ¹⁸O. The solid black line and the broken black curves are a weighted total least squares linear fit to the data and a 95% confidence bound on the fit, respectively. The diagonal gray line shows the equilibrium relationship between Δ₄₇ and δ¹⁸O over the range of temperatures required to explain the observed range (approximately 5–30‰). The horizontal gray line shows the isotopic evolution due to evaporation at a constant temperature. The temperature of 21.2 °C is the average temperature of the two leftmost data points. The light and dark orange wedges are model predictions for the relationship between Δ₄₇ and δ¹⁸O that is expected from rapid degassing and associated kinetic isotope effects (38). The confidence bounds on the fit to the data are most consistent with constant temperature, less consistent with an evolving temperature (which can be ruled out by the relationship between δ¹³C and δ¹⁸O) and only marginally include kinetic isotope effects due to rapid degassing (see text).

Formation Environment of ALH84001 Carbonates.

Previously only loose constraints on the temperature of carbonate growth in ALH84001 were available, and past models for their formation required one or more assumptions about temperature, the open/closed nature of the system, the ratio of CO₂ to H₂O, or the isotopic composition of the parent fluid. Here, by independently constraining temperature, we are able to use the isotopic variation in the carbonates to develop a model for their formation environment and mechanism.

From the Δ₄₇ constraints, we find that the striking core to rim coevolution of (approximately 5 → 30‰) and (approximately 25 → 65‰), observed in previous ion-microprobe studies (15–18) is best explained by gradual evaporation of a shallow subsurface aqueous solution at a constant, low temperature. The evaporative water loss drives carbonate precipitation and CO₂ degassing under equilibrium conditions. The observed isotopic variation in the carbonates suggests a system in poor short-term communication with the atmosphere—otherwise the δ¹⁸O and δ¹³C would be buffered by exchange with the atmospheric CO₂ reservoir. This rules out, for example, formation of the ALH84001 carbonates in a long-lasting, open lake. Instead we suggest that distillative loss of the isotopically light vapor phase (gaseous CO₂ and water vapor) from a semiisolated environment (Fig. 3A) gradually enriched the residual water and dissolved inorganic carbon in the heavy isotopes, resulting in the coupled increase in δ¹⁸O and δ¹³C values from the concretion cores to their rims. A system open to the atmosphere, where evaporation and carbonate precipitation occurred too rapidly to allow equilibrium with the gas phase, may also be consistent with the coevolution of δ¹⁸O and δ¹³C values, but is not supported by the observed δ¹⁸O-Δ₄₇ relationship (Fig. 2B). Finally, although the observed δ¹³C increase can also be explained by organic matter synthesis or oxidation, the existence of extraterrestrial organic matter in ALH84001 remains controversial, due mainly to the difficulty in ruling out contamination by terrestrial organic matter (27). Furthermore, the covariation in carbon and oxygen isotope ratios suggests a common process affected both reservoirs.

Fig. 3.

Model for carbonate formation in ALH84001. (A) Physical model of a shallow subsurface aquifer. The depth of carbonate formation is constrained by a combination of the cosmic ray exposure pattern of ALH84001 and Martian meteorite ejection models (48, 49). (B) Reduction of the physical model to a geochemical model of carbonate precipitation and CO₂ degassing driven by gradual evaporation of water in a confined volume, coupled to loss of the vapor phase.

Support for the hypothesis of carbonate formation from an evaporating subsurface aqueous reservoir comes from observed covariation between δ¹⁸O and δ¹³C in terrestrial carbonates from cold or arid regions. These analogs are imperfect because even in cold, dry, and relatively organic-poor environments, organic matter fixation or respiration contributes to the preserved isotopic signal to variable degree. Even so, the isotopic composition of Mg-bearing carbonates formed on ultramafic mine tailings in northwestern Canada (latitude ≥58 °N) form arrays strikingly similar to those observed in ALH84001 carbonates in δ¹⁸O-δ¹³C space (41), as do endostromatolites in an impact structure in the Canadian high Arctic (42), soil carbonates from the Mojave and Atacama deserts (43, 44), and caliche formed on basalts in young volcanic fields in Arizona (45). All of these studies discuss hypotheses that explain the observed isotopic trends with evaporation and CO₂ degassing.

To further examine this hypothesis we developed a model of carbonate precipitation (Fig. 3B) in which carbonate mineral saturation and charge balance are maintained, as is chemical and isotopic equilibrium among the aqueous solution, the instantaneous precipitate, and the gas phase in the regolith during gradual evaporation (Methods and SI Text). The range of δ¹⁸O values in the concretions requires distillative loss of 42–92% of the water, depending on whether the transport of water toward the surface is rate limited by vapor-phase diffusion. The observed covariation in δ¹⁸O and δ¹³C can be reproduced (Fig. 4) if approximately 87–98% of the noncarbonate carbon loss from solution is transported to the surface. The remaining 2–13% causes a pore pCO₂ increase of 5–15%, consistent with the modest pressure gradient that would be required to cause distillative CO₂ loss from the subsurface in the scenario considered here. These results are relatively insensitive to the choice of initial pCO₂, the nonprecipitating ion concentration, and the particular carbonate mineralogy (SI Text and Fig. S6).

Fig. 4.

Comparison of observed and modeled isotopic composition. Gray crosses are combinations of δ¹⁸O and δ¹³C microprobe data correlated by Mg content, including uncertainty in the correlation (SI Text). The black curves show a weighted total least squares linear fit through these data (solid) and 95% confidence bounds on the fit (broken). The black triangles are measurements from this study. The gray rectangle brackets the highest values of achievable by water–rock oxygen isotope exchange, given the uncertainty on temperature and on the measured δ¹⁸O of the ALH84001 silicates (SI Text). The light orange field shows the coevolution of δ¹⁸O and δ¹³C during 42% evaporation and diffusion-limited water loss from a subsurface reservoir, accompanied by carbonate precipitation and CO₂ degassing and diffusion, at a temperature of 17.5 ± 5 °C (SI Text). The darker orange field is for 92% evaporative water loss and accompanying carbonate precipitation if transport to the surface is not rate-limited by diffusion. The gray line with a negative slope shows the δ¹⁸O and δ¹³C coevolution if the water slowly freezes instead of evaporating. The initial pCO₂ in the calculations was 600 millibar and the nonprecipitating solute concentrations were 0 mol liter^-1. The results are relatively insensitive to this choice (SI Text and Fig. S6).

Although the fluid in an environment such as the one we describe could, in principle, come from depth (magmatic or hydrothermal fluid), there is evidence that the parent fluid of the ALH84001 carbonates came from the surface (atmosphere–hydrosphere). The carbonates have an oxygen isotopic composition that is anomalous with respect to Martian igneous rocks [Δ¹⁷O > 0.3; (46)]. Similarly, sulfides in this sample have a “mass-independent” sulfur isotopic composition [Δ³³S ≠ 0; (47)]. Both signatures were likely generated by atmospheric reactions and support a surface source of the fluid. Furthermore, water in oxygen isotope equilibrium with the Ca-rich concretion cores (determined by microprobe studies) would, at the temperatures we measured, have δ¹⁸O more than 10‰ lower than equilibrium with coexisting silicate minerals (Fig. S7) and CO₂ in equilibrium with the concretion cores is tens of permil too high in δ¹³C to have come directly from the Martian mantle (though the δ¹³C of the Martian mantle is poorly constrained, SI Text).

At the measured growth temperature, the Mg-rich carbonates in the concretion rims have δ¹⁸O 10–15‰ heavier than equilibrium with the silicates (Fig. S7). This, together with the persistence of the mass-anomalous oxygen isotope composition, implies that the combination of time and temperature was insufficient to allow appreciable oxygen isotope exchange between water and rock, supporting a low-temperature, transient aqueous environment. This is consistent also with the absence of alteration minerals (e.g., hydrous phyllosilicates) in ALH84001 other than the carbonates and the associated sulfides and magnetite (13, 14).

Conclusions

In sum, the evidence points to a short-lived, low-temperature, aqueous environment, initially containing surface-derived CO₂ and H₂O but subsequently characterized by poor communication with the atmosphere, such that isotopic buffering did not occur. Distillative loss of CO₂ and H₂O can explain the observed δ¹⁸O and δ¹³C trends. This combination of characteristics is consistent with partial to complete evaporation of a subsurface aqueous reservoir, coupled to carbonate precipitation and CO₂ degassing under equilibrium conditions. We imagine that an ephemeral regolith aquifer at a depth of a few to tens of meters would be a plausible setting for this process (Fig. 3). The existence of a near-surface aqueous environment during the Noachian epoch, recorded by the carbonates in ALH84001, is consistent with, though need not require, a warmer early Mars—a hydrothermal or impact-related source for the required heat is equally possible. Though the mild temperatures point to an environment that might be considered habitable, the presence of water was also ephemeral, suggesting a time frame probably too short for life to have evolved de novo.

Methods

Three aliquots of ALH84001 were subjected to stepped phosphoric acid digestion. Because of the slower dissolution kinetics of magnesian carbonates, the first step in each of the digestions (0–1 h) sampled the faster-reacting, Ca-rich carbonates, whereas the second and third steps (1–4 and 4–12 h, respectively) sampled more Mg-rich compositions (34). Finer separation (e.g., by lower reaction temperature or shorter reaction intervals) was not attempted due to the relatively large amount of CO₂ required for clumped isotope analysis, the susceptibility of small samples to contamination, and the precious nature of the sample. The bulk and clumped isotopic composition of evolved CO₂ was measured on a Thermo Finnigan MAT 253 isotope ratio gas source mass spectrometer. The small sample size necessitated measurement using a microvolume. Measurements were standardized to carbonates of known bulk and clumped isotopic composition, treated identically to the meteorite aliquots and measured using a microvolume. See SI Text for full analytical methods.

With knowledge of the precipitation temperature, we constructed an isotopic evolution model and examined hypotheses for carbonate precipitation that could also quantitatively explain the observed covariation of carbon and oxygen isotope ratios in the carbonate. Prescribing values for the model-free parameters—the initial CO₂ pressure (pCO₂) and the concentration of nonprecipitating ions—we solve for the pH, solute concentrations, and pCO₂ as the water evaporates. The oxygen isotopic evolution depends on the cumulative fraction of water lost to evaporation and vapor-phase diffusion. The carbon removed by carbonate precipitation is equal to the loss of the precipitating cation, which is solved by the model. The remaining carbon lost from solution is divided between transport to the surface and an increase in pore pCO₂. The carbon isotopic evolution, therefore, depends on the relative proportions of these removal processes (carbonate precipitation, diffusive transport, and pore overpressurization). See SI Text for a full description of the model and parametric sensitivity analysis results.