Jan Spacek, Clay A. Abraham, and Steven A. Benner

A preprint of a manuscript submitted to Astrobiology.

If you understand the reasoning presented in this paper, you will understand why I’m spending every waking moment working on IMPRESS and ALF projects to seek life on Mars before we send humans there, and ideally before Tianwen-3 brings the samples back.

Abstract

The 1976 Viking Carbon Assimilation–Pyrolytic Release (CAPyR) experiment operationally tested whether carbon supplied as ¹⁴CO₂ and ¹⁴CO would, during a 120-hour incubation, enter Martian soil in a form where radiolabel was not released when the incubated soil was flushed at 120 °C, but was released upon pyrolysis at 635 °C, in a form captured in an organic vapor trap (Chromosorb-P/CuO). Radioactive material released from the soil at 635 °C, but not trapped, was measured as Peak 1. Chromosorb-trapped material was subsequently oxidized by CuO, and its radioactivity was measured as Peak 2. According to the design, net Peak 2 would be interpreted as representing reduced organic material synthesized from ¹⁴CO₂ and ¹⁴CO, a synthesis that presumptively required Martian biology. Run at two sites on Mars, Chryse Planitia (with results labeled C1 through C6) and Utopia Planitia (with results labeled U1 through U3), seven of nine usable incubations gave significant net Peak 2 signals, including one incubation done in the dark. In C1, the signal corresponded to ~10 pmol of Peak 2 carbon if derived from ¹⁴CO or ~30 pmol of Peak 2 carbon if derived from ¹⁴CO₂. This corresponds to ca. 10³ Escherichia coli-equivalents of carbon presumptively fixed by Martian soil biology. Subsequent advances in our understanding of Martian soil chemistry, in particular, the recognition that perchlorate may have been present in the Viking soils, and that perchlorate would oxidize fixed organic carbon at 635 °C, may require reconsideration of how the CAPyR data are interpreted. Perchlorate oxidation at 635 °C of fixed ¹⁴C-labeled organic carbon to ¹⁴CO₂/¹⁴CO may have shifted, or “repartitioned”, ¹⁴C-label away from Peak 2 (the organic carbon signal) and into Peak 1. This perchlorate-specific repartitioning was not considered in previous CAPyR analyses. Further, Hubbard (1979) sought abiotic explanations of some CAPyR results by seeking positive Peak 2 signals with several mineral substrates under Viking-relevant conditions. Although these experiments did not reproduce the complete set of Viking data, they raise the possibility that Peak 2 radiolabel might not be biologically specific. These data are examined in detail, confirming the conclusion that no demonstrated abiotic mechanism has yet reproduced all of the CAPyR observations under Mars-relevant conditions. Thus, while Viking CAPyR did not establish extant life, the observations remain consistent with a biological carbon-assimilation system whose relevant activity survives the 90 °C pretreatment and is strongly impaired by pretreatment at 175 °C.

Keywords: Mars; Viking; carbon assimilation; pyrolytic release; perchlorate; biosignatures.

1. Introduction

Independent of any interpretations, the 1976 Viking Carbon Assimilation–Pyrolytic Release (CAPyR) experiment was operationally designed to detect transfer of radioactive carbon, supplied as ¹⁴CO₂ and ¹⁴CO gases, into organic material that remained within a soil sample after excess radiolabeled gases were removed with a 120 °C flush. Operationally, the transferred radiolabel was subsequently released from the soil by pyrolysis at 635 °C in a form that would be trapped by an organic vapor trap (Chromosorb-P/CuO) operating at 120 °C. The trapped fraction was then to be oxidized to ¹⁴CO₂ by the CuO, after which it emerged from the trap into a radiation detector, and quantitated in what came to be called “Peak 2” (Hubbard et al., 1970; Horowitz et al., 1972; Brown et al., 1978).

This workflow was extensively tested with terrestrial soils before flight. In more than 80 analyses of sterilized soils, none met the investigators’ criteria for a statistically significant response from the trapped Peak 2 organic fraction (Hubbard et al., 1970). Significant trapped radioactivity had been found under illuminated or dark conditions only in soils containing metabolically active organisms, including Antarctic dry-valley soils (Hubbard et al., 1970).

Under this design, on Mars, the radioactivity in Peak 2 was to be assigned as arising from nonvolatile organic material synthesized biologically from the supplied ¹⁴CO₂ and ¹⁴CO, carbon “fixed” during a 5-day incubation (Hubbard et al., 1970; Horowitz et al., 1977). The fixation was not limited to photosynthetic (illumination was provided in most of the experiments). Any biological process that incorporated the supplied inorganic carbon-14 into material retained during the 120 °C flush, released at 635 °C, and trapped in the Chromosorb, could contribute to the Peak 2 signal (Horowitz et al., 1972).

On Mars, many of the CAPyR experiments produced small, but statistically significant, net Peak 2 signals that were intended to “represent organic matter synthesized from ¹⁴CO or ¹⁴CO₂ during the incubation” (Horowitz et al., 1977) (Table 1). Although this was originally interpreted as a sign of Martian biology (Jastrow, 1977), the CAPyR investigators themselves ultimately judged a biological origin unlikely, in part because the response appeared to be more thermally stable than expected for hypothetical Martian biology (Horowitz et al., 1977).

Klein’s contemporaneous comparison of the Viking biology experiments overall similarly favored nonbiological chemistry, but stated that the explanation for the apparent CAPyR organic synthesis remained obscure (Klein, 1978). More recent reassessments of the Viking biology experiments have emphasized that perchlorate, discovered long after the Viking experiments were complete (Hecht et al., 2009), may change interpretation of certain Viking measurements that were originally suggested to show an absence of life (Navarro-González et al. 2010, McKay et al., 2025), and possibly to indicate that the Martian soils were ‘self-sterilizing’ (Tsapin et al. 2000, Crawford et al. 2003). The discussion surrounding perchlorate has, however, focused primarily on three other Viking experiments: the Label Release (LR) experiment, the Gas Exchange (GEx) experiment, and the gas chromatograph-mass spectrometer (GC-MS) experiment. This paper complements the set and reanalyzes data from the CAPyR experiment.

This reanalysis distinguishes three levels of inference: (i) what Viking directly measured, (ii) what molecular process the CAPyR design intended that measurement to represent, and (iii) what biological or abiotic mechanism could have generated the measured fraction.

The reanalysis is necessary at this time because preflight and post-Viking experiments show that retention of (bi)carbonate label in alkali soils both on Earth and Mars, oxidation of fixed organics by perchlorate and other oxidants during pyrolysis, the possibility that small volatile chlorinated products from perchlorate oxidation observed by the Viking GC-MS may have arisen from fixed organics but not trapped at the operating temperature of the Chromosorb-P column, and abiotic mineral chemistry might alter the relation between the precursor chemistry and the measured CAPyR Peaks 1 and 2.

2. Experimental Logic and Viking Results

2.1 CAPyR measurement principle and analytical workflow

A standard CAPyR experiment incubated approximately 0.25 cm³ of Martian surface material for 120 hours in a chamber to which had been added, in the chamber’s headspace, a 92:8 (volume:volume) mixture of ¹⁴CO₂ and ¹⁴CO (Horowitz et al., 1977; Brown et al., 1978) (Fig. 1). Standard early runs began at ~9.8 mbar total pressure. Run temperatures across the mission spanned approximately 8 to 26 °C (Horowitz et al., 1977). Most runs were illuminated with a xenon lamp behind a Schott WG335 filter that suppressed wavelengths below approximately 320 nm; dark runs were also performed (Horowitz et al., 1977; Brown et al., 1978).

After incubation, the soil was heated to 120 °C to vent the radiolabeled atmosphere (Horowitz et al., 1977). The soil was then heated at approximately 625–635 °C to fragment, or pyrolyze, any organic material that had been fixed. Evolved gases were swept through a Chromosorb-P/CuO organic-vapor trap maintained near 120 °C. Radiolabel in ¹⁴CO₂, ¹⁴CO, and other volatile ¹⁴C-labeled products that were not trapped then moved into a radiation counter, which registered a signal that came to be known as Peak 1 (Horowitz et al., 1977). It was recognized that Peak 1 radioactivity could have come from organic material fixed in the soil but then oxidized to ¹⁴CO₂ and/or ¹⁴CO with oxygen atoms derived from the soil. Indeed, this was observed in preflight tests of the CAPyR architecture on Earth.

The material retained on the organic vapor trap was later released at 640 °C. Here, any ¹⁴C-labeled organic fragments were oxidized by CuO embedded in the Chromosorb-P to give ¹⁴CO₂. This material passed to the radiation counter, and was counted as Peak 2.

To manage any ¹⁴CO₂/¹⁴CO that might enter into the trapped pathway, Horowitz et al. empirically estimated the background term as Peak 2(0) = 28.8 + 2.84 × 10⁻⁵ × (Peak 1), with a standard deviation of 27 dpm (disintegrations per minute). Net Peak 2 radiolabel was calculated to be the measured Peak 2 minus Peak 2(0) (see Fig. 1 in Horowitz et al., 1977).

The experiment added no organic nutrient mixtures. Water was introduced to selected samples as vapor rather than as bulk liquid. Thus, other than a slight increase in overall pressure, maintaining the temperature at 8 – 26 °C, removal of light < 320 nm, and an increase in the CO/CO₂ ratio, the samples saw little change in their natural environment.

Preflight tests demonstrated picomole-scale analytical sensitivity to retained radiocarbon under the tested conditions, an amount of carbon equivalent to that present in ~10² E. coli cells (Horowitz et al., 1972).

The design interpretation required that the pyrolysis-trapping-oxidation sequence deliver to net Peak 2 no radioactivity other than radioactivity from carbon that had traveled from ¹⁴CO₂/¹⁴CO into reduced organics and back to ¹⁴CO₂ by pyrolysis into fragmented organics, followed by their trapping and subsequent oxidation by CuO. Accordingly, careful preflight control experiments were carried out to constrain any contribution of unfixed ¹⁴CO₂/¹⁴CO to Peak 2 (Horowitz et al., 1972; Hubbard et al. 1971; Horowitz et al. 1977). Further, the design interpretation viewed the reduction of ¹⁴CO₂/¹⁴CO to give reduced organic carbon as possible only by biological catalysis.

Figure 1. Viking CAPyR workflow redrawn for this reanalysis from the Viking CAPyR descriptions (Hubbard et al., 1970; Horowitz et al., 1977; Brown et al., 1978). Rounded boxes indicate analytical operations. Peak 1 and Peak 2 are operational fractions indicating the amount of ¹⁴C counted in a beta particle detector, without explicit molecular identification. The experimental design intended, however, for Peak 2 radioactivity to largely represent ¹⁴C-labeled organic fragments retained in the Chromosorb-P organic vapor trap, where those fragments were the pyrolysis products of organic molecules synthesized by reduction of the supplied ¹⁴CO₂ and/or ¹⁴CO.

2.2 Preflight recognition of chemical ambiguities

Before flight, CAPyR developers identified chemistry that might deliver both false-negative and false-positive results (Hubbard et al., 1970). In nitrate-rich Atacama soil, oxidation of reduced ¹⁴C-labeled organic material by nitrate repartitioned radiolabel into Peak 1, which represented ¹⁴CO₂ and/or ¹⁴CO not removed in the 120 °C flush, and away from the trapped-organic fraction counted as Peak 2 attributed to biological carbon fixation. This showed that soil oxidants might decrease recovery of fixed ¹⁴C-containing organics during pyrolysis (Hubbard et al., 1970). Indeed, nitrates were considered (and ruled out) as an oxidant that destroyed organics during heating up to 500 °C, causing the lack of organics detected by GCMS (Biemann et al., 1976).

The Atacama test soil contained approximately 13 wt% nitrate, orders of magnitude above nitrate measured by Curiosity. In this respect, Atacama soil is not representative of Martian soil. However, these preflight data showed the possibility that a soil oxidant might diminish the amount of Peak 2 radioactivity that might otherwise arise from pyrolysis of biological organics (Hubbard et al., 1970; Stern et al., 2015). These data became relevant after the Phoenix discovery of perchlorate in Martian soils (Hecht et al. 2009). Perchlorate oxidizes most organic materials at high temperatures, including at 635 °C, primarily to CO₂, but also to chlorinated organics (Navarro-González et al. 2010), including those observed by the Viking GC-MS on Mars (Biemann et al., 1976).

Different preflight experiments showed that sterilized alkaline soil could retain large amounts of ¹⁴CO₂ and return most of that label upon pyrolysis to the ¹⁴CO₂ Peak 1 fraction while delivering very little (0.02% in the Atacama soil test) into the Peak 2 trapped-organic fraction (Hubbard et al., 1970). This established that Peak 1 could contain a substantial abiotic mineral-carbon contribution. Simple (bi)carbonate inorganic carryover into Peak 2 was small in the tested preflight matrices (Hubbard et al., 1970) and accounted for by the Peak 2(0) calculation (Horowitz et al., 1977) discussed above.

However, the two possibilities that Peak 1 radiolabel might arise by

(i) perchlorate-oxidation of ¹⁴C-labeled organic molecules (presumptively generated biologically) or

(ii) ¹⁴C-labeled (bi)carbonates stable in the 120 °C flush but releasing ¹⁴CO₂ at temperatures > 600 °C (non-biology)

make it impossible without further analysis to assign Peak 1 radioactivity as the consequence of biology or non-biology. In retrospect, had an acidic gas (e.g., HCl) been included in the 120 °C flush, these two could have been cleanly distinguished, as (bi)carbonates readily release CO₂ with acid, while organics do not. Inclusion of a volatile acid in the flush is recommended should any future mission seek to reproduce the CAPyR data.

The team also considered Kolbe-type carboxylation, where CO₂ reacts without biological catalysis with electron-rich aromatic compounds to form radiolabeled benzoic acids, as a possible abiotic route that brought the radiolabel from ¹⁴CO₂ into organic material. Experiments showed, however, that alkaline soil containing phenol or humic acid produced negligible amounts of trapped radiolabel that later generated Peak 2 radioactivity (18 and 3 counts per minute with 1% phenol and 10% humic acid, respectively). Hubbard et al. noted that the pyrolysis would decarboxylate those benzoic acids, delivering any ¹⁴C fixed by this path into Peak 1 ¹⁴CO₂ (Hubbard et al., 1970).

Last, preflight work had seen abiological synthesis of ¹⁴C-labeled material that would generate Peak 2 label from CO and water, driven by high-energy (<300 nm) UV light (Hubbard et al., 1971; Horowitz et al., 1977). This motivated the use of a lamp (as opposed to Martian sunlight) with a >320-nm filter as the source of illumination when seeking photosynthetic carbon fixation (Hubbard et al., 1971; Horowitz et al., 1977). Hubbard later suggested that this spectral safeguard did not eliminate all abiotic sources of Peak 2 radiolabel (Hubbard, 1979). This suggestion is analyzed in Sections 3.2–3.5.

2.3 Run-level results and diagnostic treatments

Ten CAPyR experiments were attempted. Cycles with the prefix “C” were conducted by Viking 1 at Chryse Planitia. Cycles with the prefix “U” were conducted by Viking 2 at Utopia Planitia. Seven (C1 through C6 and U1) had statistically significant positive net Peak 2 under the published correction model, whereas U2 and U3 did not. The data from U4 were unusable because of an apparent valve S-11 failure venting Peak 2 material before it could be counted (Horowitz et al., 1977), leaving nine usable cycles summarized in Table 1. Note that the U4 experiment Peak 1 and Peak 2 were affected differentially by the leaking valve because of the data collection timing. Peak 1 signal was collected over a period of <20 minutes; Peak 2 was collected for >12 hours. Whether the valve was partially faulty before the U4 cycle, and thus affected the Peak 2 data obtained in cycles U1–U3, is not discussed in the literature.

The experiments were repeated under different conditions (Horowitz et al., 1977). The U1 Peak 2 radioactivity was substantial even though the incubation had been done without light. This suggests that a mechanism requiring illumination during the incubation cannot explain every positive Viking run (Horowitz et al., 1977).

Darkness may not, however, exclude abiotic mineral chemistry in general, as Hubbard (1979) reported some CAPyR-positive mineral responses from experiments done in the dark. As discussed below, the formation of ¹⁴C-labeled organic material from ¹⁴CO₂ and/or ¹⁴CO without light, or another source of energy, is thermodynamically puzzling.

Table 1. Run-level conditions and principal CAPyR results for the nine usable Mars experiments. Values are from Horowitz et al. (1977). U4 is omitted because of an apparent valve failure.

C1 produced the largest net Peak 2, 842 ± 29 dpm (Horowitz et al., 1977). Horowitz et al. estimated that this corresponded to approximately 10 pmol of carbon fixed if derived entirely from CO, or approximately 30 pmol if derived entirely from CO₂ (Horowitz et al., 1977). The difference arises because the ¹⁴CO₂ and ¹⁴CO had different specific activities. If interpreted biologically, that amount corresponds to approximately 10³ Escherichia coli-equivalents of carbon using an order-of-magnitude cellular carbon content. This is an illustrative mass-scale comparison (Phillips and Milo, 2009), and cannot be used to infer the initial population density without information about metabolic rates.

More information is available from variations on the CAPyR experiment. Thus, the C2 sample, from the same dig as the C1 sample, was stored for 19 sols and then heated to 175 °C for 3 h. Its net Peak 2 was approximately 88% lower than fresh C1, while its Peak 1 was essentially unchanged (Horowitz et al., 1977). This would indicate that the Peak 2 signal was heat-labile, while the Peak 1 signal was not, notwithstanding the presence of perchlorate.

C4, C5, and C6 were aliquots from the same sample acquisition and delivered similar net Peak 2 data, despite testing fresh, after 69 sols of storage plus the C5 water/thermal treatment, and after 139 sols of storage, respectively (Horowitz et al., 1977). These experiments differed in several variable including radiolabeled-gas injections, storage, water handling, and heating histories. However, the run histories directly constrain interpretations of thermal and storage sensitivity (Horowitz et al., 1977), and the differences in Peak 1 radioactivities are noteworthy. The variability may be at least partially explained by different amounts of labeled gas supplied in the experiments. C4, C5, and C6 received one, three, and six radioactive-gas injections, respectively (Horowitz et al., 1977; Brown et al., 1978).

3. Reanalysis in Light of Post-Viking Chemical and Mineralogical Information

3.1 High-temperature oxidants can reduce Peak 2

The Mars Phoenix lander measured approximately 0.4–0.6 wt% perchlorate in its Martian soil samples (Hecht et al., 2009). While perchlorate is unreactive towards organics at temperatures below ~200 °C, most organic materials are oxidized by perchlorate at temperatures over ~400 °C to generate CO₂ and chlorinated products (Navarro-González et al., 2010), including methyl chloride (CH₃Cl) observed by the Viking GC-MS (Biemann et al. 1976; Glavin et al., 2013; Guzman et al., 2018).

Perchlorate was not directly measured at either Viking landing site, so its abundance in the CAPyR soil samples remains unknown. Controlled experiments show that perchlorate-driven combustion of organic matter to CO₂/CO and substrate-dependent degradation/chlorination deliver a range of chlorinated organic compounds (Royle et al., 2018; Millan et al., 2020). However, the ability of Chromosorb-P to trap small volatile chlorinated organics (e.g. CH₃Cl) at 120 °C remains unknown. It is likely that different organics generate different ratios of CO₂, small chlorinated organics (which may not be trapped by Chromosorb at 120 °C), and large chlorinated organics (which are likely trapped by Chromosorb at 120 °C) (Sephton et al., 2014).

These experiments directly establish high-temperature oxidation and chlorination of fixed reduced carbon that may have caused an underestimation of the amount of biologically fixed carbon. However, no post-Viking experiments used the complete Viking CAPyR Chromosorb-P/CuO analytical sequence. Under the CAPyR workflow, perchlorate chemistry could have repartitioned some pre-pyrolysis reduced ¹⁴C towards the ¹⁴CO₂/¹⁴CO and volatile chlorinated organics entering Peak 1 and away from Peak 2. The magnitude of that repartitioning remains unknown until measurements are made in a reconstructed CAPyR system, including a detailed analysis of how different organic molecules give different ratios of CO₂, very volatile organic products that are not trapped on Chromosorb, and less volatile organic fragments that are trapped.

The preflight nitrate experiment establishes the same direction of effect, proving the principle (Hubbard et al., 1970). With nitrate as well as perchlorate, net Peak 2 would underestimate reduced labeled carbon present immediately before pyrolysis. Again, this was considered (for nitrate) by those analyzing the GC-MS data (Biemann et al., 1976).

High-temperature oxidation does not, by itself, explain how a reduced ¹⁴C-labeled nonvolatile precursor would have formed before pyrolysis under the original organic-fixation interpretation of Peak 2. Hubbard (1979) and Quinn and Pacheco (2013) independently proposed abiotic routes to CAPyR nominally life-positive results. As discussed in Sections 3.2–3.5, these proposals are thermodynamically problematic, and (in any case) do not reproduce all of the relevant Viking signal classes and treatment responses.

The essentially unchanged Peak 1 in C1 versus C2 is also an important constraint. If much of the C1-to-C2 Peak 2 decrease was caused by the 175 °C treatment, most of the Peak 1 cannot be plausibly derived from that same heat-labile precursor. Peak 1 is therefore best treated as not being predominantly derived from the same heat-labile precursor fraction. Conversely, Peak 2 cannot be assumed to contain all reduced carbon that may have existed before pyrolysis because perchlorates and nitrates can divert pre-existing reduced carbon toward the Peak 1 pathway during heating (Navarro-González et al., 2010; Royle et al., 2018).

Reanalysis of Viking GC-MS data in light of post-Viking data is useful, with some caution. Chlorobenzene detected in the Viking data, but originally excluded as presumed instrument background, could have had a perchlorate chlorine source and a Martian carbon source (Guzman et al., 2018). The result illustrates that high-temperature products can contain Mars-derived components without uniquely identifying the specific structure of the pre-pyrolysis carbon precursor (Guzman et al., 2018). It also removes the former simplicity of treating failure by GC-MS to observe parent organics during Viking heating as evidence that reduced organic carbon had been absent before analysis (Navarro-González et al., 2010; Guzman et al., 2018).

3.2 Laboratory abiotic chemistry reproduced only a subset of CAPyR results

Hubbard’s 1979 experiments with Martian soil simulants remain an important effort to explore the possibility that the operational CAPyR outcomes were not biology-specific. These experiments used a Test Standards Module with mechanical components and operational sequences similar to the flight CAPyR system. This included a WG335 filter that removed wavelengths from a standard lamp below approximately 320 nm (Hubbard, 1979).

Peak 2 radiolabel was observed with hematite, maghemite, magnetite, a Mars-analog mixture, and Fe-exchanged montmorillonites (Hubbard, 1979). Some responses were interpreted as photocatalytic, while dry magnetite also produced a dark response. This suggested that neither the Viking spectral cutoff nor darkness alone excludes abiotic mineral chemistry as a potential source of Peak 2 signal (Hubbard, 1979).

This notwithstanding, any suggestion that ¹⁴CO₂ can generate ¹⁴C-labeled organic molecules by mineral catalysis in the absence of light, or with light having wavelengths only greater than 320 nm, seems thermodynamically paradoxical. A suggestion that ¹⁴CO can generate ¹⁴C-labeled organics is only marginally less so. The reaction CO + H₂O → organic H₂CO₂ is energetically uphill by ~30 kJ/mol at one bar of CO. The reduction of CO₂ + H₂O → organic H₂CO + O₂ is much more unfavorable thermodynamically. The source of reducing equivalents in Hubbard’s dark chemistry remains unresolved. The abiotic pathway will remain paradoxical until or unless it satisfies redox and thermodynamic constraints. This is examined further below.

If this thermodynamic paradox is overlooked, the Hubbard experiments suggest (as Claim A) that selected mineral systems might generate abiotic CAPyR-positive material resulting in Peak 2 radiolabel (Hubbard, 1979). However, they did not establish a Claim B: that the abiotic chemistry explains the complete set of CAPyR signals in the Viking regolith (Hubbard, 1979). Accordingly, Hubbard concluded that duplication of the whole of the Viking CAPyR data had not been achieved (Hubbard, 1979).

Specifically, several candidate soil substrates produced Peak 1 values much larger than Viking, perhaps involving generation of ¹⁴C-labeled carbonates and bicarbonates. Further, responses to water and thermal treatment also differed among substrate systems (Hubbard, 1979). Hubbard (1979) recognized another issue: Moisture approximately doubled the TSM response in some Hubbard experiments, whereas the Viking experiments showed no reproducible moisture effect. Horowitz et al. (1977) ultimately favored sample heterogeneity rather than water inhibition as the explanation for the differing Utopia responses to added moisture.

Thus, it remains open whether an abiotic process can reproduce the Viking CAPyR data as a whole.

3.3 Mineralogical applicability and the partially circular analog argument

Earth-based laboratory work to analyze the Viking data must begin with a choice of material to simulate Martian soil. Hubbard identified model-soil selection as the aspect of the simulations in which he had the least confidence. The then available Viking X-ray fluorescence data constrained elemental abundances, but did not assign mineral phases (Hubbard, 1979).

Recognizing this ambiguity, the original ‘Mars analog soil’ used in those experiments contained 51% nontronite and 26% bentonite, making it 77% clay of some type (Hubbard, 1979). Later Fe²⁺– and Fe³⁺-exchanged montmorillonites were considered particularly promising because they produced positive Peak 2 with Peak 1 values closer to the Viking range (Hubbard, 1979).

The similarity of the analytical output being explained, the ¹⁴C-label in Peak 1 in particular, was used in part to support the geological realism of the substrate chosen to explain that output (Hubbard, 1979). Analytical agreement can be evidence for a chemical model only after the substrate itself has independent geological support. Otherwise, one substrate is selected from many simply because it generates data to be fitted.

Modern mineralogical measurements provide independent constraints that were unavailable in 1979. Thermal infrared analysis of Martian atmospheric dust indicates a framework silicate component, probably feldspar, as a major component, with lesser olivine, pyroxene, amorphous material, hematite, and magnetite. This differs from a composition that is dominated by clays (Hamilton et al., 2005).

Rocknest data provided the first in situ X-ray diffraction measurement on Mars to identify mineral phases. Direct X-ray diffraction of the Rocknest aeolian soil identified plagioclase, olivine, augite, pigeonite, minor iron-bearing phases, and a substantial amorphous component, with a crystalline assemblage similar to basaltic materials (Bish et al., 2013).

No corresponding measurements constrain the minerals in the Viking Chryse or Utopia samples. Thus, Rocknest observations do not eliminate the possibility of Viking soils containing smectites. However, they remove independent support for treating a 77%-clay or artificially Fe-exchanged montmorillonite substrate as a default representation of Viking surface fines (Hamilton et al., 2005; Bish et al., 2013). The mineralogy weakens the claim that the abiotic chemistry explains the complete set of CAPyR signals in the Viking regolith (Section 3.2).

3.4 Thermal response and storage behavior may favor a heat-labile source over a Hubbard 1979 model

Leaving aside the thermodynamic challenges of processes where mineral catalysts generate organic material from CO₂ and/or CO without an energy source, we can compare biological and abiological models for their ability to understand the variants of the Viking CAPyR experiments. The C1/C2 comparison provides the strongest thermal attenuation in the Viking data that might be interpreted as the consequence of biology. C1 gave 842 ± 29 dpm net Peak 2, whereas C2 gave 105 ± 29.5 dpm net Peak 2 after 19 sols of storage plus 175 °C heating for 3 h, a ~88% decrease (Horowitz et al., 1977).

Because storage and heat treatment were varied together, the entire decrease cannot be assigned to the 175 °C heating. Thus, C4, C5, and C6, aliquots from one Sandy Flats 3 acquisition, gave similar net Peak 2 after 0, 69, and 139 sols of storage, despite additional treatment differences and differences in the number of radiolabeled-gas injections (discussed in Section 2.3).

The comparison also constrains attempts to explain CAPyR and the Viking Labeled Release (LR) response (the oxidation of ¹⁴C-labeled organics to give ¹⁴CO₂). Klein’s cross-experiment summary reported that the Peak 2 signal was retained after prolonged storage, while the LR response disappeared. The samples in that comparison were stored at spacecraft temperatures of approximately 20 °C (Klein, 1978).

Brown et al. (1978) noted that the biology instrument used a common Soil Distribution Assembly that retained sufficient material from each acquisition for subsequent redistribution to the GEx, CAPyR, and LR experiments. However, publications do not establish identical microscopic humidity or headspace histories for every stored aliquot. The comparison is therefore best described as a broadly common dry spacecraft-temperature storage environment.

Under that shared regime assumption, the difference between the LR and CAPyR argues against the same labile storage-sensitive agents being responsible for both LR and CAPyR positive responses. The observations summarized by Klein (1978) argue for the presence of different agents with different resistances to storage conditions. The carbon-fixing agent persisted at low-temperature storage for over 139 sols and partially survived ~170 °C for 3 h treatment. The nutrient-oxidizing agent is inactivated by 82-sol storage, and partially inactivated by 50 °C for 3 h treatment and fully inactivated by 160 °C for 3 h treatment.

Hubbard’s mineral systems also did not reproduce the Viking CAPyR thermal attenuation (Hubbard, 1979). Candidate substrates were pre-dried at 160 °C before testing. Nevertheless, substantial abiotic activity remained (Hubbard, 1979). In the Test Standards Module, a simulated sterilization treatment of 170 °C for 3 h left approximately 73% of the corresponding light-dry Peak 2 response of the Mars analog, and sterilized maghemite did not show a Viking-like reduction in Peak 2 (Hubbard, 1979).

Hubbard therefore cautioned that the sterilization effect was not statistically resolved in his limited analog series, and that the tested mineral systems did not exhibit the strong attenuation observed between Viking C1 and C2 (Hubbard, 1979). For the CAPyR data, Klein had already noted that a successful nonbiological catalyst would need to remain active after heating at approximately 90 °C yet be strongly impaired by 175 °C (Klein, 1978).

The Viking thermal response therefore preferentially supports an unknown heat-labile source of Peak 2 over the specific mineral-catalyzed abiotic mechanisms suggested by Hubbard. This is consistent with biology, especially recognizing that some Terran spores can be recovered in viable form after dry heat exposures to 115-170 °C (Kempf et al., 2008) and brief heating up to 200 °C (Schubert & Beaudet 2011). Thus these data do not exclude a biological source. However, they also do not exclude any yet unknown abiotic carbon-fixing species with different thermal susceptibility.

C5 defines the lower-temperature side of that behavior. The sample was briefly brought to 120 °C for approximately 1–2 min after water addition and venting, then held near 90 °C for about 112 min before the remainder of the 5-day incubation near 17 °C; its net Peak 2 remained statistically similar to fresh C4 (Horowitz et al., 1977).

The C5 pretreatment was not necessarily sterilizing even by terrestrial standards. Vacuum-dried wild-type Bacillus subtilis spores retained 55% viability after 5 min at 120 °C and 14% after 30 min, while 90% remained viable after 60 min at 90 °C (Setlow and Setlow, 1995). Thus, survival during the brief 120 °C excursion followed by approximately 112 min near 90 °C in C5 is compatible with known terrestrial biology.

This does not imply that Martian life would resemble bacterial spores. It establishes only that a robust survival of the C5 pretreatment is not incompatible with all known terrestrial biology. The thermal stability of hypothetical Martian molecular biology is empirically unconstrained. The relevant biological requirement is survival or persistence of the carbon-assimilation machinery followed by activity during the cooler incubation.

3.5 Other proposed abiotic routes remain unvalidated

The preflight Kolbe-type tests directly examined abiotic carboxylation of electron-rich aromatic organics by ¹⁴CO₂ (Hubbard et al., 1970). Alkaline soil containing phenol or humic acid delivered little trapped label into Peak 2. This is consistent with thermal decarboxylation returning the incorporated carbon to ¹⁴CO₂, essentially the reverse of the carboxylation reaction (Hubbard et al., 1970). This specific abiotic carboxylation route was not considered to threaten a Peak 2 false positive signal (Hubbard et al., 1970).

Quinn and Pacheco later proposed a different process for abiological organic synthesis. In a 2013 conference abstract, they reported the formation of trace CH₃Cl and CH₂Cl₂ during thermal analysis of carbonate/perchlorate/iron-oxide mixtures under some helium conditions. They suggested that supplied CO₂ might first form labeled carbonate and then generate chloromethanes during heating (Quinn and Pacheco, 2013).

This suggestion presents thermodynamic challenges. The reduction of CO₃^2- (or HCO₃^–) with anything short of a metal hydride to a carbon at a -2 oxidation state (as in CH₃Cl) is strongly disfavored thermodynamically. It would be chemically surprising if carbon in CH₃Cl were to be shown by isotopic labeling to be formed from carbonate.

Further, experimentally established perchlorate chemistry is already known to generate chlorinated products from pre-existing reduced organic carbon (Navarro-González et al., 2010; Keppler et al., 2014; Millan et al., 2020). Perchlorate combustion of trace contaminants in Quinn and Pacheco samples may provide an alternative carbon source for the observed traces of CH₃Cl and CH₂Cl₂. Accordingly, Quinn and Pacheco (2013) suggested that additional work was necessary to establish the observation and mechanism.

3.6 Synthesis: Biology is consistent with observations, abiotic pathway is missing

 Despite its thermodynamic challenges, some mineral chemistry may yield positive Peak 2 signals, more likely from ¹⁴CO than ¹⁴CO₂. Generally analogous processes are known at higher pressures of CO (Evans et al., 2006). While reducing equivalents do not easily come from iron at its +3 oxidation state, magnetite has iron at a lower oxidation state that might conceivably provide reducing equivalents. However, such chemistry is not known to us.

Thus, no demonstrated abiotic mechanism has yet reproduced the combination of Viking Peak 1/Peak 2 behaviors, light/dark results, survival of the C5 treatment, and strong C2 attenuation. Because the Viking soil samples themselves were heterogeneous, a successful model need not reproduce every run identically. An actionable model should generate both positive Peak 2 and also reproduce the principal classes of observations and the diagnostic perturbation responses.

However, a biological interpretation remains consistent with the CAPyR data. Under the original interpretation, a Martian biological system would need to incorporate supplied ¹⁴CO₂ and/or ¹⁴CO into a retained reduced-carbon precursor during the 5-day incubation at 8 to 26 °C (Horowitz et al., 1977). Its relevant carbon-assimilation activity would need to persist through the C5 120 °C/90 °C pretreatment yet be strongly impaired by the C2 175 °C treatment (Horowitz et al., 1977).

Such thermal response resembles some life as we know it (Kempf et al., 2008; Setlow & Setlow, 1995; Schubert and Beaudet, 2011). Likewise, the significant U1 result, obtained in the dark, is compatible with nonphotosynthetic carbon fixation, including atmospheric trace-gas-supported chemolithoautotrophy (Ji et al., 2017).  

The CAPyR data neither identify a metabolic pathway nor establish living cells. They show only that a Martian biological catalyst with this experimentally constrained phenotype is consistent with the Viking observations. An abiotic catalyst showing carbon fixation both under UV illumination and in the dark, together with thermal instability at 175 °C remains unidentified.

4. Future Earth-Based Testing

This reanalysis motivates experiments seeking to match the manifold of CAPyR results here on Earth, without an expensive flight to Mars. Experiments with hematite, maghemite, magnetite, original clay-rich Mars analog, and Fe-exchanged montmorillonites should be replicated as historical positive controls because they establish a range of possible abiotic CAPyR-positive responses (Hubbard, 1979). Their thermodynamic paradoxes must then be resolved.

In parallel, substrates should be selected using modern models of Mars mineralogy, including basalt-derived plagioclase/pyroxene/olivine assemblages, amorphous Fe-bearing material, measured minor iron oxides, and systematically varied clay abundances (Hamilton et al., 2005; Bish et al., 2013).

The reconstruction should then reproduce the Viking CAPyR workflow, including gas composition, illumination spectrum, incubation, the 120 °C flush, high-temperature sample pyrolysis, Chromosorb-P/CuO trap, and trap oxidation (Horowitz et al., 1977; Brown et al., 1978). Each candidate should be subjected to light/dark controls, storage-only controls, a C5-like thermal treatment, and a 170–175 °C three-hour treatment, with Mars-relevant perchlorate added as a separate variable. (Bi)carbonate and Kolbe-type routes can be tested in the same system, and any proposed chloromethane pathway should require isotope-proven carbon provenance and direct measurement of retention of this and other very volatile chlorinated organics by the CAPyR trap. A candidate mechanism should be judged against both Peak 1/Peak 2 generation and attenuation as a treatment response, noting what fraction of organics oxidized by perchlorate is moved from Peak 2. Producing Peak 2 alone is insufficient.

5. Conclusions

The CAPyR reanalysis can be reduced to six technical points.

1. CAPyR was operationally designed and preflight-tested to detect the transfer of ¹⁴C-carbon from supplied ¹⁴CO₂ and ¹⁴CO into a retained radiolabeled fraction that, under the intended molecular interpretation, represented newly (biologically) synthesized organic material (Hubbard et al., 1970; Horowitz et al., 1972; Brown et al., 1978).

2. Operationally, Viking repeatedly measured the life-positive Peak 2 signals that were sought by the experiment. Seven of nine usable runs had statistically significant positive net Peak 2 signals, including the dark U1 run. The response was heterogeneous among samples and was diminished by heat treatment (Horowitz et al., 1977).

3. Preflight experiments with nitrate-containing soils directly showed that an oxidizing agent during pyrolysis can move reduced ¹⁴C away from the Peak 2 fraction captured by the organic vapor trap into ¹⁴CO₂ in Peak 1 (Hubbard et al., 1970), an under-counting of fixed carbon. Modern experiments independently show this process with perchlorate (Navarro-González et al., 2010; Royle et al., 2018; Millan et al., 2020). Under the Viking CAPyR workflow, analogous chemistry could therefore cause Peak 2 to underestimate reduced carbon present before pyrolysis, although this repartitioning has not yet been measured with the complete CAPyR workflow.

4. This oxidizing chemistry based on perchlorate cannot possibly provide a low-temperature route that would generate a reduced nonvolatile precursor under any organic-fixation interpretation. Abiotic reduction routes have been proposed, most importantly Hubbard’s mineral chemistry and the Quinn–Pacheco carbonate-to-organics hypothesis (Hubbard, 1979; Quinn and Pacheco, 2013). However, they are largely silent on the origin of reducing equivalents and the thermodynamic paradoxes that they present. Thus, no mechanism has yet reproduced the principal Viking observations as a whole, and those that may have done so remain paradoxical from the perspective of chemical thermodynamics.

5. Modern Mars chemistry and mineralogy weaken two premises that once favored a nonbiological interpretation of the CAPyR results. In the presence of perchlorate, high-temperature analysis can destroy or transform pre-existing organics. Thus, any failure, by the GC-MS or CAPyR, to recover parent organics or their fragments does not establish that none were present before heating (Navarro-González et al., 2010; Guzman et al., 2018). At the same time, the clay systems used to seek non-biological results for the CAPyR data were selected under a mineralogical model not independently supported as a default composition of Martian aeolian fines. Further, even these systems did not reproduce the thermal attenuation and other details of the CAPyR observables.

6. If the C1 net Peak 2 carbon arose from biological assimilation, its magnitude corresponds to approximately 10 pmol if derived from CO or 30 pmol if derived from CO₂, on the order of 10³ E. coli-equivalents of carbon in mass (Horowitz et al., 1977; Phillips and Milo, 2009). The biological interpretation requires a Martian carbon-assimilation system whose relevant activity operates during the 8–26 °C incubation, survives the C5 pretreatment, including heating briefly at 120 °C and subsequently at 90 °C, and is strongly impaired at 175 °C. Horowitz et al. (1977) considered survival above 90 °C too robust for life expected on Mars. Newer data show, that thermal resistance profile is not too far from life as we know it (Setlow & Setlow, 1995; Kempf et al., 2008; Schubert and Beaudet, 2011). The Viking data do not conclusively establish such a microbial system on Mars, but they have not shown it to be inconsistent with the observations.

This should motivate renewed (Jastrow, 1977) consideration of the view that the CAPyR results remain consistent with biology.

Taken together, the LR (Levin & Straat, 2016) and CAPyR results show that Martian soil samples both oxidized introduced organic nutrients and reduced introduced CO₂ and/or CO into nonvolatile organics, with the two activities associated with agents having markedly different thermal stabilities. This combination of observations is consistent with a microbial community containing distinct metabolic activities. No demonstrated abiotic agent or combination of agents has yet reproduced both response classes together with their low thermal stabilities. An abiotic explanation accounting for both LR and CAPyR observations should provide a simultaneous accumulation of a low-temperature oxidant and a low-temperature reductant in upper Martian soil.

These considerations should redirect Mars exploration priorities back toward the direct search for extant life on Mars (Stoker et al., 2021; Benner, 2026). This priority is also emphasized for the Martian subsurface by Nisson et al. (2026). Distributed, low-cost mission architectures such as IMPRESS provide one practical route for returning modern in situ life-detection experiments to multiple Martian surface and shallow-subsurface sites (Spacek et al., 2026).

References

Benner SA. Meet the Neighbors: Life on Mars and How to Find It. Allen Lane: London, UK; 2026.

Biemann K, Oro J, Toulmin P, III, et al. Search for organic and volatile inorganic compounds in two surface samples from the Chryse Planitia region of Mars. Science 1976, 194(4260): 72–76.

Bish DL, Blake DF, Vaniman DT, et al. X-ray diffraction results from Mars Science Laboratory: Mineralogy of Rocknest at Gale Crater. Science 2013;341(6153):1238932.

Brown FS, Adelson HE, Chapman MC, et al. The biology instrument for the Viking Mars mission. Rev Sci Instrum 1978;49(2):139–182.

Crawford RL, Paszczynski A, Allenbach L. Potassium ferrate [Fe (VI)] does not mediate self-sterilization of a surrogate Mars soil. BMC Microbiology 2003; 3(1): 4.

Evans WJ, Lipp MJ, Yoo CS, Cynn H, Herberg JL, Maxwell RS, Nicol MF. Pressure-induced polymerization of carbon monoxide: Disproportionation and synthesis of an energetic lactonic polymer. Chemistry of Materials 2006; 18(10):2520-2531.

Glavin DP, Freissinet C, Miller KE, et al. Evidence for perchlorates and the origin of chlorinated hydrocarbons detected by SAM at the Rocknest aeolian deposit in Gale Crater. J Geophys Res Planets 2013;118(10):1955–1973.

Guzman M, McKay CP, Quinn RC, et al. Identification of chlorobenzene in the Viking gas chromatograph-mass spectrometer data sets: Reanalysis of Viking mission data consistent with aromatic organic compounds on Mars. J Geophys Res Planets 2018;123(7):1674–1683.

Hamilton VE, McSween HY Jr, Hapke B. Mineralogy of Martian atmospheric dust inferred from thermal infrared spectra of aerosols. J Geophys Res Planets 2005;110(E12):E12006.

Hecht MH, Kounaves SP, Quinn RC, et al. Detection of perchlorate and the soluble chemistry of Martian soil at the Phoenix lander site. Science 2009;325(5936):64–67.

Horowitz NH, Hubbard JS, Hobby GL. The carbon assimilation experiment: The Viking Mars lander. Icarus 1972;16(1):147–152.

Horowitz NH, Hobby GL, Hubbard JS. Viking on Mars: The carbon assimilation experiments. J Geophys Res 1977;82(28):4659–4662.

Hubbard JS, Hobby GL, Horowitz NH, et al. Measurement of ¹⁴CO₂ assimilation in soils: An experiment for the biological exploration of Mars. Appl Microbiol 1970;19(1):32–38.

Hubbard JS, Hardy JP, Horowitz NH. Photocatalytic production of organic compounds from CO and H₂O in a simulated Martian atmosphere. Proc Natl Acad Sci U S A 1971;68(3):574–578.

Hubbard JS. Laboratory simulations of the pyrolytic release experiments: An interim report. J Mol Evol 1979;14(1–3):211–221.

Jastrow R (1977) Until the Sun Dies. New York, W. W. Norton & Company

Ji M, Greening C, Vanwonterghem I, et al. Atmospheric trace gases support primary production in Antarctic desert surface soil. Nature 2017;552(7685):400–403.

Kempf MJ, Schubert WW, Beaudet RA. Determination of lethality rate constants and D-values for Bacillus atrophaeus (ATCC 9372) spores exposed to dry heat from 115°C to 170 °C. Astrobiology 2008;8(6):1169–1182.

Keppler F, Harper DB, Greule M, et al. Chloromethane release from carbonaceous meteorite affords new insight into Mars lander findings. Sci Rep 2014;4:7010.

Klein HP. The Viking biological experiments on Mars. Icarus 1978;34(3):666–674.

Levin, GV, Straat, PA. The case for extant life on Mars and its possible detection by the Viking labeled release experiment. Astrobiology 2016, 16(10), 798-810.

McKay CP, Quinn RC, Stoker CR. The Viking biology experiments on Mars revisited. Icarus 2025;431:116466.

Millan M, Szopa C, Buch A, et al. Influence of calcium perchlorate on organics under SAM-like pyrolysis conditions: Constraints on the nature of Martian organics. J Geophys Res Planets 2020;125(5):e2019JE006359.

Navarro-González R, Vargas E, de la Rosa J, et al. Reanalysis of the Viking results suggests perchlorate and organics at midlatitudes on Mars. J Geophys Res 2010;115(E12):E12010.

Nisson DM, Mickol RL, Robinson A, Lloyd KG. Life detection in the Martian subsurface: Advancements in microbial metabolic research under Mars laboratory and field-analog conditions since Viking. Astrobiology 2026;26(Suppl 1):53S–71S. doi:10.1177/15311074261464014.

Phillips R, Milo R. A feeling for the numbers in biology. Proc Natl Acad Sci U S A 2009;106(51):21465–21471.

Quinn RC, Pacheco DJ. Production of chlorinated hydrocarbons during the thermal decomposition of metal carbonates and perchlorate salts. In: 44th Lunar and Planetary Science Conference; March 18–22, 2013; The Woodlands, TX. Lunar and Planetary Institute: Houston, TX; 2013; Abstract 2664.

Royle SH, Oberlin E, Watson JS, et al. Perchlorate-driven combustion of organic matter during pyrolysis–gas chromatography–mass spectrometry: Implications for organic matter detection on Earth and Mars. J Geophys Res Planets 2018;123(7):1901–1909.

Schubert WW, Beaudet RA. Determination of lethality rate constants and D-values for heat-resistant Bacillus spores ATCC 29669 exposed to dry heat from 125°C to 200°C. Astrobiology 2011;11(3):213–223.

Sephton MA, Lewis JM, Watson JS, Montgomery W, Garnier C Perchlorate‐induced combustion of organic matter with variable molecular weights: Implications for Mars missions. Geophysical Res Lett 2014; 41(21): 7453-7460.

Setlow B, Setlow P. Small, acid-soluble proteins bound to DNA protect Bacillus subtilis spores from killing by dry heat. Applied and Environmental Microbiology. 1995;61(7):2787–2790.

Spacek J, Alpern H, Dineen T, et al. IMPRESS: A Planetary Penetrator Network for Astrobiology, Prospecting, and Exploration of Mars. Astrobiology 2026. doi:10.1177/15311074261471218.

Stern JC, Sutter B, Freissinet C, et al. Evidence for indigenous nitrogen in sedimentary and aeolian deposits from the Curiosity rover investigations at Gale crater, Mars. Proc Natl Acad Sci U S A 2015;112(14):4245–4250.

Stoker C, Blank JG, Boston P, et al. We should search for extant life on Mars in this decade. Bull AAS 2021;53(4):33. doi:10.3847/25c2cfeb.36ef5e33

Tsapin AI, Goldfeld MG, Nealson KH, Kemner KM, Moskovitz B Self-sterilizing properties of martian soil: possible nature and implications. SAE Technical Paper. 2000 2000-01-2043, SAE International, Warrendale, PA.