Comment on “Reanalysis of the Viking results suggests perchlorate and organics at midlatitudes on Mars” by Rafael Navarro-González et al.

[1] The reanalysis of the Viking 1976 data by Navarro-González et al. [2010, 2011]asserts that the finding of chloromethane in one Martian soil sample at landing site 1 (VL-1) and dichloromethane in both samples collected at site 2 (VL-2), originally attributed to terrestrial contamination [Biemann et al., 1976, 1977], indicates instead the presence of perchlorate and organic carbon. This conclusion is based on experiments with Atacama Desert soil in the presence of added 1 w% magnesium perchlorate and in the absence of added perchlorate. With added perchlorate chloromethane and dichloromethane are simultaneously and exclusively produced at temperatures of ≥350°C. This is not consistent with the Viking results. Kinetic modeling by Navarro-González et al. [2010] assumes that these two chlorohydrocarbons are produced by reaction of chlorine derived from the thermal decomposition of the perchlorate with methane even though the latter is not detected when the soil sample is heated without perchlorate. The absence of the equally likely production of chlorinated aromatic molecules does not support the conclusions presented. [2] In a recent article, Navarro-González et al. [2010] (referred to hereinafter as “the authors”; all figure numbers unless stated otherwise refer to that paper) note that NASA's Phoenix mission, which touched down in the Martian Arctic (68.3°N, 127°W) 25 May 2008, discovered perchlorate (ClO4−) as a component of the collected soil samples [Hecht et al., 2009]. Based on experiments with perchlorate-doped terrestrial soils and hypothetic modeling, the authors came to the remarkable conclusion that the soil samples analyzed by the Viking Molecular Analysis Experiment (commonly called the “Viking GCMS” (Gas Chromatograph - Mass Spectrometer)) in the summer of 1976 [Biemann et al., 1976, 1977] missed the organic material present at the Viking Landing 1 (VL-1) site at Chryse Planitia (22.7°N, 48.2°W) and Viking Landing 2 (VL-2) site at Utopia Planitia (48.3°N, 226.0°W). [3] These conclusions are based on the following assumptions and assertions: (1) perchlorate must also exist at the two Viking sites; (2) that the organic matter actually present in the soil samples analyzed by Viking was oxidized and/or chlorinated, thus preventing its detection and the resulting products, CH3Cl and CH2Cl2, were detected by the Viking GCMS, but at the time were interpreted to be terrestrial contaminants [Biemann et al., 1976, 1977]; and, therefore, assert that (3) Viking actually detected the presence of ≤0.1% perchlorate and 1.5–6.5 ppm organic carbon at the VL-1 site and ≤0.1% perchlorate and 0.7–2.6 ppm organic carbon at the VL-2 site. [4] There are several arguments against the points in paragraph 3: First, Phoenix found 0.4–0.6% perchlorate, 0.01–0.04% chloride (i.e., between 10 and 60 times more perchlorate than chloride), and a soil pH of 7.7 ± 0.5 [Hecht et al., 2009]. However, at the Viking sites the soil is assumed to be highly acidic [Navarro-González et al., 2006], as indicated by their use of ferric sulfate in sulfuric acid [Navarro-González et al., 2006, Figure 5], to achieve complete oxidation of stearic acid, as a model of the destruction of organic material upon heating to 750°C. This notion was, in part, based on the detection of jarosite by the Mars Rover Opportunity [Klingelhöfer et al., 2004]. Second, in their introduction the authors devote a considerable part to the abundance of chlorine at various sites on Mars where it has been measured. The Viking X-ray fluorescence based Cl data average 0.73% at Chryse and 0.44% at Utopia [Clark et al., 1982], not 0.1% and 0.9%, respectively, as the authors incorrectly state. While the total chlorine concentrations for the Arctic Phoenix site and the two Viking sites appear comparable, there is no reason to assume that their chloride/perchlorate ratio would be similar at the Viking sites. Thus, the extrapolation from the Arctic Phoenix site to the two Viking sites (on opposite sides of the planet) is questionable. There is, of course no reason to expect a correlation between chloride and perchlorate in the Martian soil, because it has been suggested that on Mars perchlorate is produced in the atmosphere by photochemical reactions, electrostatic discharge, and gas-solid reactions involving oxidation of volcanic HCl by O3 and its photochemical decomposition products like H2O2, HO· etc. [Catling et al., 2010], which also could give rise to surface variability in the concentration of perchlorate. The same processes would result in the oxidative destruction of the organic material present in the regolith when it is suspended in the atmosphere during the periodic sandstorms, which keep the dust particles exposed to radiation and to these oxidants for significant periods of time. These oxidation processes have been suggested previously [Biemann et al., 1976, 1977; Biemann, 2007] as an explanation of the absence of organic matter at the surface of Mars. [5] Second, all the experiments and calculations of the authors show that CH3Cl and CH2Cl2are produced simultaneously in a ∼20:1 ratio. However, the former was detected only when one of the two VL-1 samples were analyzed, and the latter only with the two VL-2 samples. The authors also completely ignore the fact that the Viking data clearly demonstrated the detection of Freon-E, and for VL-2 also methyl-fluorosiloxane and other cleaning solvents, such as acetone, benzene, toluene and xylene (see Table 1). This was attributed to slight contamination of the VL-1 GCMS and more severe contamination of the VL-2 instrument by the solvents used to clean the spacecraft and instruments. If the detection of the various terrestrial contaminants shown inTable 1seems somewhat erratic, it should be kept in mind that the cleaning of such complex instruments is not a simple matter. Each component had to be cleaned separately, then assembled to sub-systems and finally to the complete instrument. For example, the atmospheric inlet system was built and cleaned separate from the soil loader and pyrolyzer sub-assembly [Rushneck et al., 1978]. Also the analysis of the atmosphere involved only the inlet volume (which was, of course, never heated during the entire mission), with a tube extending above the lander's deck, and connected directly to the mass spectrometer. In contrast, the soil analyses used a much more complex acquisition and processing system. Furthermore, the two GCMS instruments were cleaned and assembled at separate times. In addition, the sample acquisition system, including the sampler arm, distribution system, grinding and sieving, to the oven carousel, were each cleaned separately. Acetone, dichloromethane, benzene, methanol, etc. were commonly used cleaning agents and silylating agents were used in the preparation of GC columns. Furthermore, for operational reasons, soil samples could not be immediately placed into the oven and analyzed. For example, VL-1 sample 1 was collected and processed on sol 8 (29 July 1976) but kept in the hopper until sol 17 (7 August) before delivery to the oven and then heated to 200°C [Biemann et al., 1976], providing ample time to adsorb any methanol, HCl and/or CH3Cl, and other contaminants that may have been present. [6] Finally, the experiments carried out by the authors in support of their assertions (1) did not use a set-up corresponding to the Viking GCMS design, operation and protocol, (2) do not agree with the Viking CH3Cl and CH2Cl2 data, or (3) are not carried out in a meaningful way. [7] With respect to point 1 above, the authors use an assembly of commercially available instruments plumbed together, in contrast to the completely integrated, miniaturized Viking GCMS and its sample processing and heating system [Biemann, 2007; Biemann et al., 1977]. A significantly important difference is the authors' commercial pyrolyzer, in which the sample is placed into a quartz capillary, closed at both ends by quartz wool plugs, over which the carrier gas (He) flows. This is in considerable contrast to the integrated, flow-through ovens of the Viking design. The authors' total system leads to a sensitivity considerably lower than that of the Viking GCMS: even after correction [Navarro-González et al., 2011] the lowest signal for CH2Cl2 (at 350°C) detected (see corrected Figure 4 of Navarro-González et al. [2011]) is about 10 ppb, while in the Viking Lander 2, sample 2, 0.04–0.08 ppb of CH2Cl2 was detected (see Table 1) using a sample of ≤100 mg in the 67 μL oven [Biemann et al., 1977; Rushneck et al., 1978]. [8] Regarding point 2 in paragraph 6, the Viking data (see Table 1) show the detection of CH3Cl when sample 1 (but none from sample 2) collected at the VL-1 site was heated to 200°C, while Figure 4 ofNavarro-González et al. [2010] demonstrates that heating of the model Atacama soil with added perchlorate produces CH3Cl only at ≥350°C. Similarly, CH2Cl2reached its maximum when either VL-2 sample 1 or 2 were heated to 350°C, while the simulation experiment (Figure 4) demonstrates that at that temperature no CH2Cl2 is produced. [9] Finally, regarding point 3 in paragraph 6, it should be noted that CH3Cl is a gas at room temperature and not used in the cleaning procedure for the Viking instruments. Its detection, already in pre-flight analyses [Biemann et al., 1976], was attributed to a reaction between adsorbed residual methanol and HCl, a process well known to organic chemists and so stated in the paper [Laniewski et al., 1998] cited by the authors. Nevertheless, they subjected a solution of 0.4% HCl in methanol to their experimental set-up, heating the solution in the quartz tube from 30°C to 200°C, 350°C, 500°C, 750°C and 1000°C in separate experiments. It was found that CH3Cl is indeed produced at temperatures of ≥350°C and primarily at 500°C, but then with the concomitant production of CH2Cl2 and benzene. It is not clear how there could have been any methanol left after the quartz tube had been flushed with He for 3 min., then heated to 200°C (or higher) and held there for 30 s. [10] The reason why the authors in these experiments detected only those compounds they were looking for (CH3Cl, CH2Cl2, C6H6 and C6H5-CH3) is due to the mode of operation of the mass spectrometer and its data system: selective ion monitoring of only the m/z values 50, 84, 78 and 91, as the legend to Figure 2 states. This mode is commonly used only to detect or quantify specific components in complex mixtures, but not to survey all the unknown products that may be present. Regardless whether this mode is accomplished by “hopping” from one of the selected m/z values to the next, or by automatically picking the selected m/z signal out of a full scan, the result is the same (except that the former provides higher sensitivity than the latter and is thus preferred for quantitative analyses): only those molecules whose mass spectrum exhibits a significant signal at the m/z being monitored will be detected. One can also plot the full mass spectrum of the component which produces that m/z value, without interference from anything else present in the mixture. For very simple mixtures of only two, three or four chromatographically well resolved components it is certainly not necessary to reconstruct the gas chromatogram by selective ion monitoring of up to four m/z values. [11] The conclusion that there actually was so much perchlorate (≤0.1%) and so much organic matter (0.7–6.5 ppm as organic carbon) in the soil samples analyzed by Viking is based on the assumption that it is methane, and only methane, that reacts with the chlorine produced by the thermal decomposition of magnesium perchlorate. The same rate expression database used by the authors (http://kinetics.nist.gov) for their kinetic modeling predicts that benzene and toluene detected by VL-2 (seeTable 1), or produced upon heating either the VL- 1 or VL-2 samples, if they did indeed contain organic carbon, would both be chlorinated along with methane, but neither C6H5Cl nor C7H7Cl were detected by Viking. It is to be noted that the authors' Figure 4 does not show methane to be produced in the experiment, while benzene and toluene are the major (and only) organic products. [12] Aside from these kinetic considerations, it is well known to organic chemists, that aromatic hydrocarbons are easily halogenated. For example, aromatic chlorohydrocarbons, in addition to aliphatic ones, are produced upon heating of environmental samples [Laniewski et al., 1998]. Even more relevant is the work of Steininger and Goesmann [2010], also inspired by the Phoenix data, showing that upon heating of either phenylalanine or benzoic acid in the presence of magnesium perchlorate a variety of chlorinated aromatics were observed. The authors' mode of operation of the mass spectrometer (and data processing) (see section 4) precluded the detection of chlorobenzene and chlorotoluene by not also monitoring m/z 112,114, 126 and 128. Furthermore, the reaction of chlorine with methane in the gas phase and on solids proceeds all the way to CCl4 [Treger and Rozanov, 1989] and one should therefore expect that at least CHCl3 (most abundant fragment at m/z 83) would also have been detected. [13] There are still other major errors and inconsistencies that bring into question the overall conclusions of the authors about the level of organic material supposedly detected by Viking and in their Atacama samples. The values for benzene and toluene produced upon heating the Atacama soil reported in Figure 4 of the original paper by far exceeded the organic carbon present in the sample. This was recently corrected [Navarro-González et al., 2011] by stating that the volume of the CO2 used to calibrate the instrument had been 10 cm3, not 40 cm3 as originally assumed. This brought the values for O2, NO, N2O, C6H6, C6H5CH3, CH3Cl and CH2Cl2 down by a factor of 4. In support of this correction the authors state that [Navarro-González et al., 2011, paragraph 2] “the revised values … agree stoichiometrically with the estimated concentration of soil organic carbon determined independently….”: 32 +/− 6 ppm organic carbon heated to 750°C produced 36.5 +/− 14.5 ppm benzene and 6.2 +/− 2.1 ppm toluene, equivalent to 39.4 +/− 15.6 ppm of organic carbon. Clearly, rather than supporting the correction by a factor of 4, it disproves it, because it is unreasonable to claim that the organic carbon in a soil sample can be exclusively and quantitatively converted into just two aromatic hydrocarbons. Furthermore, in a paper by the same group [Navarro-González et al., 2006], it is stated that Atacama soils from the Yungay area containing 20–40 ppm organic carbon produces only 1–4 ppm benzene (the only compound reported) upon heating the soil to 750°C. Such values are much more reasonable. [14] Not only is it difficult to understand how the organic carbon in the Atacama soil is converted upon heating exclusively into benzene and toluene, but also difficult to rationalize why other, related molecules are not formed. Considering that the soil used exhibited a C/N ratio of 8.2 one would expect also the formation of some nitrogen containing molecules. For example, when ∼100 mg of Antarctic soil #638 containing ∼300 ppm of organic carbon [Cameron, 1971] was heated to 500°C in the oven of a laboratory version of the Viking GCMS a complex reconstructed gas chromatogram was obtained, in which 26 aliphatic and aromatic compounds could be identified on the basis of their mass spectra, including 5 nitrogen containing ones, ranging from acetonitrile to toluonitrile [Biemann et al., 1977]. That this soil did contain almost ten times as much organic carbon than the Atacama sample should not significantly change the complexity of the products, just the abundance of the components. [15] The original error in the abundances is also rationalized by the additional description of the analytical procedure [Navarro-González et al., 2011]: all the components analyzed for, their amounts being shown in Figure 4 of the original paper and in the correction [Navarro-González et al., 2011], were estimated from a calibration curve of carbon dioxide “assuming that all chemical species would have a similar response to CO2by GC-MS” [Navarro-González et al., 2011, paragraph 1] It is now explicitly stated that “no calibration curves were done for each individual component” [Navarro-González et al., 2011, paragraph 1]. From sections 2.3, 2.4 and 3.1 of the original paper one must assume that selective ion monitoring of m/z 50, 84, 78, and 91 was used for quantitation of CH3Cl, CH2Cl2, C6H6 and C6H5CH3. Without proper calibration for each of these compounds, reliable results cannot be obtained, because it is well known that the response (peak height) in a mass spectrum is dependent on the ionization cross section of the molecule, its extent of fragmentation, the mass discrimination of the MS, and other more minor factors. Considering the ramifications of the authors' conclusions, which could influence the design of future missions to Mars, it would have been prudent to acquire the appropriate calibration data by injecting a known amount of a known four-component mixture into the GCMS. [16] Finally, the kinetic model used by the authors to predict the supposed amounts of organic carbon in the Viking samples fails to predict the relative amounts of dichloromethane and chloromethane released in the Atacama Desert sample they analyzed in the presence of magnesium perchlorate and shown in Figure 4. Their corrected Figure 4 [Navarro-González et al., 2011] shows that the ratio of dichloromethane to chloromethane is ∼0.1 at 500°C whereas in their model results shown in Figure 6, this ratio ranges from ∼5 to 104. Their model predicts that the dichloromethane would always be released in greater excess relative to chloromethane, which is not consistent with the Atacama results. The model also does not include the steps of further chlorination to tri- and tetrachloromethane, as well as any catalytic effects of the surface of the soil particles [Treger and Rozanov, 1989]. Even if, as the authors claim, their model is only a first order approximation, its failure to accurately predict the release of dichloromethane and chloromethane in a natural terrestrial sample raises a serious question about the validity of their model derived Viking organic carbon estimates from the amounts of these species detected in some of the Viking analyses. [17] The assertion by Navarro-González et al. [2010, paragraph 33] that “the Viking Landers did detect the presence of organics in the soil” of Mars, based on the detection of CH3Cl with one of the two samples of the VL-1 site and of CH2Cl2with both of the VL-2 samples is not supported by the experimental data. In order to validate their assertions the experimental date were searched specifically only for these two molecules but neglecting any other chlorinated aliphatic or aromatic hydrocarbons. Similarly, the kinetic modeling included only rate expressions (in the gas phase) for the conversion of CH4 to CH3Cl and CH2Cl2 but not the continued chlorination to CHCl3 and CCl4or for the chlorination of benzene or other hydrocarbons. Equally questionable is the suggestion that in situ life detection missions to Mars be initiated immediately. In fact, what is urgently needed are well designed experiments to analyze the surface of the planet for the presence of specific organic molecules, in order to ascertain what types of organic material may be present, if any. We know that meteoritic infall is disseminating such organic material continuously although radiolysis may compromise its long-term survival at the surface [Kminek and Bada, 2006]. If the meteoritic material cannot be found, it must be that destructive processes prevail, which at the same time would destroy any prebiotic chemistry long before it reaches the point of the “origin of life” chemistry. [18] We would like to thank an anonymous reviewer for bringing the paper by Treger and Rozanov [1989] to our attention. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.