Photochemical depletion of heavy CO isotopes in the Martian atmosphere

The atmosphere of Mars is enriched in heavy isotopes with respect to Earth as a result of the escape of the atmosphere to space over billions of years. Estimating this enrichment requires a rigorous understanding of all atmospheric processes that contribute to the evolution of isotopic ratios between the lower and upper atmosphere, where escape processes take place. We combine measurements of CO vertical profiles obtained by the Atmospheric Chemistry Suite on board the ExoMars Trace Gas Orbiter with the predictions of a photochemical model and find evidence of a process of photochemistry-induced fractionation that depletes the heavy isotopes of C and O in CO (δ13C = −160 ± 90‰ and δ18O = −20 ± 110‰). In the upper atmosphere, accounting for this process reduces the escape fractionation factor by ~25%, suggesting that less C has escaped from the atmosphere of Mars than previously thought. In the lower atmosphere, incorporation of this 13C-depleted CO fractionation into the surface could support the abiotic origin of recently found Martian organics. Measurements by the ExoMars Trace Gas Orbiter reveal a depletion of heavy CO isotopes in Mars’s atmosphere caused by photochemistry-induced fractionation. The observed depletion in heavy C has implications for our understanding of C escape to space and the formation of organic material on Mars’s surface.

The atmosphere of Mars is enriched in heavy isotopes with respect to Earth as a result of the escape of the atmosphere to space over billions of years. Estimating this enrichment requires a rigorous understanding of all atmospheric processes that contribute to the evolution of isotopic ratios between the lower and upper atmosphere, where escape processes take place. We combine measurements of CO vertical profiles obtained by the Atmospheric Chemistry Suite on board the ExoMars Trace Gas Orbiter with the predictions of a photochemical model and find evidence of a process of photochemistry-induced fractionation that depletes the heavy isotopes of C and O in CO (δ 13 C = −160 ± 90‰ and δ 18 O = −20 ± 110‰). In the upper atmosphere, accounting for this process reduces the escape fractionation factor by ~25%, suggesting that less C has escaped from the atmosphere of Mars than previously thought. In the lower atmosphere, incorporation of this 13 C-depleted CO fractionation into the surface could support the abiotic origin of recently found Martian organics.
Numerous lines of geomorphological and mineralogical evidence suggest that liquid water was once abundant on Mars's surface 1,2 , but it remains unclear what climatic conditions enabled this, or what drove the transition in the climate to the dry, low-pressure atmosphere we observe today. Enrichment in the heavy isotopes of several species such as N and H suggests that atmospheric escape has been an important mechanism shaping the climate and composition of the atmosphere throughout history 3,4 . Combining measured atmospheric isotope ratios with evolution models allows the estimation of the abundance of species in the atmosphere of early Mars, which demonstrates the value of a thorough understanding of the atmospheric isotope composition [5][6][7] .
Accurate estimations of the long-term evolution of the atmosphere from its isotope composition rely on two important quantities: measurements of the past and present isotopic ratios, and the net escape fractionation factor, which determines the efficiency of the heavy-isotope enrichment as the atmosphere escapes to space 8,9 . The most accurate measurements of the isotopic composition of C and O in the atmosphere of Mars were made by the Curiosity Rover, which showed an enrichment in the heavy isotopes in CO 2 with respect to Earth-like standards ( 13 C/ 12 C = 1.046 ± 0.004 VPDB and 18 O/ 16 O = 1.048 ± 0.005 VSMOW; see Methods for definition of standards), consistent with the hypothesis of substantial atmospheric loss 10,11 . On the other hand, while the fractionation factor of several escape processes of C and O has been estimated before 5,12,13 , the net escape fractionation factor, accounting for all processes from the lower atmosphere to the upper atmosphere as well, remains undetermined.
By applying this methodology to the C isotopes, the density of CO 2 in the atmosphere of early Mars has been estimated 5,14 . The escape of C at present is suggested to occur mostly in the form of hot C produced by a number of different photochemical reactions, such as the direct photodissociation of CO 2 and CO or the dissociative recombination of CO + (refs. [15][16][17]. Hu et al. 5 showed that the photodissociation of CO efficiently enriches the atmosphere in 13 C as it escapes to space, a contribution that is an important factor for explaining the current Article https://doi.org/10.1038/s41550-023-01974-2 different spectral ranges: the near-infrared (NIR) channel, which operates between 0.7 and 1.7 μm; the mid-infrared (MIR) channel operating between 2.3 and 4.2 μm and the thermal infrared channel, which samples the atmosphere between 0.7 and 17 μm. In this study, we use observations from the MIR channel, which is dedicated to solar occultation observations. For these observations, the instrument's boresight is pointed towards the Sun as it rises or sets from behind the Mars disk. By making measurements every 2.1 s, the instrument's line of sight crosses the atmosphere at different tangent altitudes above the surface, allowing the reconstruction of the atmospheric vertical profiles.
ACS observations have been previously used to measure the vertical distribution of CO and its seasonal variations in the atmosphere of Mars 23,24 . Here, we use the MIR data employed by Fedorova et al. 23 , but studying the relative abundances of 12 C 16 O, 13 C 16 O and 12 C 18 O. In particular, we use the ACS MIR measurements obtained using secondary grating position 6, which includes 14 diffraction orders (236-249) encompassing a spectral range between 3,950 and 4,193 cm −1 . We perform the retrievals in four spectral windows in diffraction orders 239 (4,012-4,025 cm −1 ), 246 (4,122-4,142 cm −1 ), 248 (4,158-4,171 cm −1 ) and 249 (4,171-4,188 cm −1 ) (see Fig. 1), to retrieve the pressure, temperature and volume mixing ratios of H 2 O and the aforementioned CO isotopes (Methods). The retrieved dataset in this study comprises all ACS MIR secondary grating position 6 observations where the four presented diffraction orders were measured, which includes 242 vertical profiles mainly measured in the second halves of Martian year (MY) 34 and MY35, covering a range of solar longitudes of L s = 166°-349° and L s = 141°-366°, respectively. This large range of observing conditions in terms of season, local time, location and dust activity allows a better characterization of any potential variability in the isotopic ratios, as well as derivation of more reliable estimates of the measurement uncertainties. Figure 2 shows a summary of all 13 C/ 12 C and 18 O/ 16 O isotopic ratios in CO measured in this dataset. We observe the 13 C/ 12 C ratio to be systematically depleted in the heavy isotopes with respect to the Earth-like standard. Weighting the measurements with their corresponding uncertainties, we derive an averaged value of 13 C/ 12 C = 0.84 ± 0.09 VPDB, where the uncertainties correspond to the s.d. of the measurements. Comparing this value with that measured in CO 2 by the Curiosity Rover, it corresponds to a depletion of 13 C/ 12 C in CO of approximately 20%. enrichment in 13 C in CO 2 measured by the Curiosity Rover. However, while CO therefore seems to be an important species for understanding the isotopic evolution of C in the atmosphere of Mars, its present-day isotopic composition has not yet been determined.
Carbon isotopic ratios can also reveal the nature of surfaceatmosphere interaction processes and mineral formation. Recently, carbon isotopic measurements by the Curiosity Rover revealed anomalously large depletions in 13 C ( 13 C/ 12 C ≈ 0.86-0.93 VPDB) on sedimentary organics potentially associated with a possible palaeosurface 18 . One of the possible explanations suggested to produce 13 C-depleted organic material is the photochemical reduction of CO 2 to formaldehyde (CH 2 O) with CO as an intermediate. In this production, the abundance of 13 C in CO is suggested to be lower than that in CO 2 as a result of the differential photolysis of 12 CO 2 and 13 CO 2 (ref. 19), transferring its isotopic signature into formaldehyde and other organics that accumulated on the surface of Gale Crater 18 . Direct measurements of the 13 C/ 12 C in CO may therefore provide constraints on whether the photoreduction of CO 2 is the mechanism responsible for the depletion of 13 C on the surface organics.
The reactions that determine the abundance of CO in the atmosphere of Mars are the same as the ones controlling the long-term stability of CO 2 : the photolysis of CO 2 to CO, and the posterior recombination of CO back into CO 2 . The stability of CO 2 in Mars's atmosphere was at first not well understood, since the photodissociation of CO 2 molecules (CO 2 + hν → CO + O) is much faster than the recombination of the products (CO + O + M → CO 2 + M). It was later realized that the stability of CO 2 is instead controlled by catalytic reactions with odd-hydrogen species (CO + OH → CO 2 + H), which can convert CO into CO 2 at a much faster rate 20,21 . This reaction network not only determines the relative abundances of CO 2 and CO in the atmosphere of Mars, but also controls the relative isotopic fractionation of C and O in these species.

Measurements of the CO isotope ratios
Here, we investigate the isotopic composition of C and O in CO using solar occultation measurements by the Atmospheric Chemistry Suite (ACS) on board the ExoMars Trace Gas Orbiter (TGO) 22 . ACS includes three spectrometers dedicated to investigating the atmosphere using a b Wavenumber (cm -1 ) Wavenumber (cm -1 ) Wavenumber (cm -1 ) Wavenumber (cm -1 )  On the other hand, the 18 O/ 16 O ratio is consistent at all altitudes with the Earth-like fractionation within the measurement uncertainties. The distribution of measured values in this case yields a weighted average isotopic ratio of 18 O/ 16 O = 0.98 ± 0.11 VSMOW, which, when compared with the isotopic ratio measured in CO 2 by the Curiosity Rover, corresponds to a depletion of the O isotopic ratio in CO of 6%, although this depletion lies within the reported uncertainties from the ACS measurements.

Modelling the CO photochemical fractionation
To understand the physical and chemical processes that give rise to the depletion of the C and O isotopic ratios measured in CO, we constructed a one-dimensional (1D) photochemical model to simulate the isotopic fractionation between CO and CO 2 . In particular, we include isotope-specific rates for the two most important reactions relating these two species: the photolysis of CO 2 into CO + O, and the recombination of CO + OH back into CO 2 . The photolysis cross-sections for 12 C 16 O 2 , 13 C 16 O 2 and 18 O 12 C 16 O are taken from the work of Schmidt et al. 19 (Extended Data Figs. 1 and 2), while the fractionation during the recombination of CO into CO 2 is modelled using data from laboratory measurements 25 (Extended Data Fig. 3). This photochemical model extends from the surface to an altitude of 200 km and is set to represent Martian mean dayside conditions (that is, solar zenith angle equal to 60° and Sun-Mars distance of 1.52 au; ref. 26). The isotopic ratios in the model are initialized following the measurements from the Curiosity Rover, and the model is then run for approximately 20 years, by which time the isotopic ratios in CO and CO 2 have converged to a steady-state solution (Extended Data Fig. 4). Figure 3 shows the derived C isotopic abundances from the photochemical model, together with the measured values from the ACS observations. The C isotopic ratios in both CO and CO 2 show a similar altitudinal trend above approximately 100 km, where they continuously decrease with increasing altitude due to diffusive separation: above the homopause, where molecular diffusion dominates, species are mixed according to their own mass-dependent scale height. As a result of this, the isotopic ratios steadily decrease with altitude, with the decrease in 13 C/ 12 C being approximately half that in 18 Fig. 4). While the isotopic ratios in CO 2 below the homopause altitude are approximately constant, those in CO are depleted in the heavy isotopes due to photochemistry-induced fractionation, mainly due to the preferential photolysis of 12  . In particular, the photochemical model suggests that below 30 km the C isotopic ratio is approximately 24% lower in CO than in CO 2 , increasing to about 9% at 100 km. The effect of the photochemistry has a smaller impact in the O isotope ratios, with the depletion of the heavy isotopes near the surface being approximately 9%. Although the ACS measurement uncertainties in the 18 O/ 16 O isotopic ratio are large compared with the magnitude of the photochemistry-induced fractionation, the lower value of the 13 C/ 12 C ratio in CO than 18 O/ 16 O predicted by the photochemical model is consistent with the ACS measurements presented in this study.

Discussion
Both the ACS measurements and the predictions from the model suggest that the photochemistry of the atmosphere of Mars produces a depletion of the heavy C and O isotopes in CO. Isotopic fractionation also occurs during the photolysis of other species such as H 2 O, whose effect in the Martian atmosphere has been extensively studied 28,29 . The photolysis-induced fractionation in H 2 O is responsible for fractionating the D/H ratio in water, molecular hydrogen and finally the escaping products, giving rise to an efficient enrichment in D/H as the atmosphere escapes to space 29-31 . Since CO contributes a substantial fraction of the C escape from Mars, this source of fractionation has to be accounted for when estimating the net escape fractionation factor. The escape of C from the atmosphere of Mars due to CO photodissociation and photoionization is most efficient at altitudes above 150 km (refs. 15,17). At these altitudes, the isotopic composition of CO is affected by both the chemistry and diffusive separation. Our model suggests that at 200 km the 13 C/ 12 C isotopic ratio in CO is ~0.75 times that of the bulk atmosphere, which combined with the fractionation factor due to escape by CO photodissociation (f ≈ 0.6; ref. 5) yields a net escape fractionation factor of ~0.45 for this reaction. This lower fractionation factor indicates that the C escape through CO photodissociation enriches the atmosphere in 13 C more efficiently than previously thought, requiring less C escape to explain the present-day enrichment of 13 C in CO 2 in the atmosphere of Mars.
On the basis of a primordial isotopic ratio of 13 C/ 12 C = 0.975 VPDB (ref. 32) and the current isotopic enrichment in CO 2 from the Curiosity Rover, Hu et al. 5 estimated that an equivalent pressure of 240 mbar of C had been lost to space throughout history, with photochemical escape being the dominant escape mechanism in the last 1.5 Gyr (~8 mbar), and sputtering of CO 2 being dominant earlier in history (~232 mbar). While the photochemistry-induced fractionation impacts only the photochemical escape fractionation factor, it can have implications for the overall estimated escape rates depending on the relative contributions from the two mechanisms. On one hand, the lower escape fractionation factor derived in this study might indicate that photochemical escape has contributed less than previously thought to the overall escape rates (that is, <8 mbar). On the other hand, it might indicate that the contribution from sputtering may have been overestimated (that is, <232 mbar) by a larger fraction than in the previous case, given the low efficiency of the sputtering process in enriching the atmosphere in 13 C (ref. 33).
The photochemistry in Mars's atmosphere therefore favours a preferential transfer of 13 C/ 12 C from CO to CO 2 , decreasing the total amount of 13 C/ 12 C in CO and transferring it to CO 2 , subsequently increasing the 13 C/ 12 C isotopic ratio of this species. However, given the orders-of-magnitude difference between the abundances of CO and CO 2 in the atmosphere of Mars, while the transfer of 13 C/ 12 C from CO to CO 2 produces a substantial effect on the overall isotopic ratio of CO, the effect on that of CO 2 is minimal. In particular, we find the near-surface isotopic ratio in CO 2 to be 0.004% higher than that in the initial atmosphere, which corresponds to an enrichment of ẟ 13 C ≈ 0.04‰. This value is much smaller than the enrichment in the heavy isotopes observed by the Curiosity Rover (ẟ 13 C = 46 ± 4‰; ref. 10), which suggests that atmospheric escape remains the most favourable process contributing to the enrichment of 13 C in the atmosphere of Mars.
The depletion of 13 C in CO observed with the ACS measurements and predicted by the photochemical model is in line with the measurements of the 13 C/ 12 C ratio in Martian organics made by the Curiosity Rover 18 . While other suggested formation mechanisms such as the photolysis of CH 4 released from the subsurface or the deposition of cosmic dust cannot be ruled out as responsible for the production of 13 C-depleted organics on Mars, our results confirm that the photolysis of CO 2 can produce strong depletions of the 13 C/ 12 C ratio in CO, an isotopic signature that can be transferred to formaldehyde or other organics that might accumulate on the surface of Mars.
The photochemical model also predicts a relative depletion of the 18 O/ 16 O isotopic ratio in CO with respect to CO 2 , although to a lesser extent than for the C isotopes. The implications of this source of fractionation for the long-term evolution of the O isotopes are nevertheless more complicated due to the much more complex O chemistry in the atmosphere of Mars. The dissociative recombination of O 2 + , responsible for most of the O escape from Mars 34 , also efficiently enriches the atmosphere in 18 O as escape occurs 13 . However, the overall escape fractionation factor will be subject to the 18 35), which will probably be affected by the isotope photochemistry. On the basis of the isotope-specific photolysis cross-sections included in this study, while the O isotopic ratios of CO 2 in the ionosphere will mostly be affected by the diffusive separation above the homopause the isotopic composition of atomic O will be affected by the preferential photolysis of  18 O in other species, which is planned but beyond the scope of the work presented here.
The results from this study highlight the important role of photochemistry in fractionating the C and O isotopic composition of the atmosphere of Mars. Future measurements of the isotopic ratios in different species may not only provide key information about the photochemical cycles in the atmosphere of Mars, but also provide accurate estimations of the fractionation between the lower and upper atmospheres, which has important implications for our understanding of the long-term evolution of Mars's climate.

Retrieval scheme
The inversion of the spectra is performed using the NEMESIS algorithm 36 , applying the same two-step methodology as in the work of Alday et al. 37 . First, the pressure and temperature profiles are determined using the CO 2 absorption features in diffraction order 239 (Fig. 1). Second, the pressure and temperature are fixed, and the abundances of the trace gases (that is, H 2 O, 12 C 16 O, 13 C 16 O and 12 C 18 O) are retrieved using all diffraction orders. All gaseous absorption in this study is modelled using precomputed line-by-line look-up tables generated using the spectroscopic parameters from the 2020 edition of the HITRAN database 38 . Apart from the retrieval of the atmospheric parameters, the instrument line shape is also fitted in each occultation using a double-Gaussian parameterization 39 . The ACS measurements (red) reveal a decrease of the 13 C/ 12 C isotope ratio in CO larger than the measured uncertainties (horizontal red lines) with respect to that in CO 2 measured by the Curiosity Rover at the surface (green). The predictions from the 1D photochemical model (dashed lines) suggest that this decrease is produced by the chemistry of the Martian atmosphere, which preferentially favours the transfer of 13 C from CO to CO 2 . The isotope ratio in CO in the upper atmosphere, responsible for a substantial fraction of the C escape to space, is substantially lower than that of the bulk CO 2 atmosphere, suggesting a lower escape fractionation factor than previously thought.
Article https://doi.org/10.1038/s41550-023-01974-2 The pressure and temperature profiles are retrieved under the assumption of an atmosphere in hydrostatic equilibrium and a known CO 2 volume mixing ratio. This approach has been widely used in the past to determine the thermal structure of the atmosphere of Mars from solar occultation observations 40 . The retrieval scheme from the NEMESIS algorithm works using the optimal estimation framework, which aims to find an optimal solution that is consistent with the observed spectra subject to a minimal departure from the a priori atmosphere. Given that the absorption features of CO 2 measured in ACS MIR secondary grating position 6 typically disappear at lower tangent altitudes than the CO ones, we use temperature profiles from simultaneous observations by ACS NIR 23 as the a priori guess in our retrievals, which ensures an accurate description of the temperature field at all altitudes. The use of accurate measurements of temperature is crucial to derive precise measurements of the CO isotopic ratios, which are highly sensitive to this parameter ( Supplementary Fig. 1).
Once the pressure and temperature profiles are determined, these are fixed, and the gaseous abundances of the CO isotopes and H 2 O are retrieved. In this case, the a priori information is taken from the OpenMARS reanalysis dataset 41 , which assimilates the temperature profiles and dust opacities measured by the Mars Climate Sounder on board the Mars Reconnaissance Orbiter into the general circulation model. The a priori isotopic composition is assumed to follow that of Earth-like conditions given by the Vienna PeeDee Belemnite ( 13 C/ 12 C = 1 VPDB = 0.0112372) and the Vienna Standard Mean Ocean Water ( 18 O/ 16 O = 1 VSMOW = 0.0020052). Given that the sensitivity of the retrieval to the CO isotopes can be different at different altitudes (that is, the 12 C 18 O absorption lines typically disappear at lower tangent heights than the 12 C 16 O), the influence of the a priori guess in the retrieval of the isotopologues might also be different. Therefore, to ensure that no biases from the a priori guess are introduced into the derivations of the isotopic ratios, the retrievals are run twice, the second time using the retrieved 12 C 16 O abundances from the previous iteration as the a priori information and scaling the abundances of the minor isotopes with the same Earth-like isotopic ratios.
In each acquisition made by ACS MIR, the observed diffraction orders encompass about 20 detector rows, which correspond to the instantaneous field of view of the instrument, with each row sampling slightly different tangent heights (Δz ≈ 150 m per row). In addition, the instrument line shape varies from row to row, which can introduce systematic uncertainties in the retrievals. To increase the confidence of the retrievals and better understand the uncertainties arising from the instrument line shape fitting, we apply the retrieval scheme to seven detector rows. The retrieved profiles from each row are then combined using a weighted average based on the retrieved uncertainties. The uncertainties are calculated considering both the error of the mean and the s.d. of the profiles 27 . We consider this method for calculating the uncertainties to provide a more accurate representation of the true uncertainty of the retrieval, in which not only the random uncertainties, but also the systematic ones, are captured.

Sensitivity of the measured isotopic ratios to the temperature field
The spectroscopic parameters tabulated in line databases such as HITRAN 38 depend on the atmospheric temperature. The dependences of these spectroscopic parameters on the temperature are different for each absorption line and gas, which might create systematic biases in the retrieval of the gaseous abundances and isotopic ratios if the temperature is not known. To determine the sensitivity of our retrievals to the temperature distribution, we perform a series of retrievals of the isotopic ratios in CO using different temperature profiles. In particular, we retrieve the abundances of 12 C 16 O, 13 C 16 O and 12 C 18 O from the spectral windows in diffraction orders 246, 248 and 249 (Fig. 1) under three assumptions: the temperature profiles are given by those derived from simultaneous measurements by ACS NIR 23 and adding a constant offset of ±5 K. These retrievals are performed for the first 20 orbits of ACS MIR measurements with secondary grating position 6 (orbits 1853-2530).
Supplementary Figure 1 summarizes the results from this experiment, showing the weighted averaged profiles of the isotopic ratios from these retrievals. These results indicate that the 13 C/ 12 C and 18 O/ 16 O ratios are highly sensitive to the temperature field: an offset of 5 K in the temperature field introduces a bias of 15-20% in the retrieved isotopic ratios. It must be noted that this kind of bias is not random but systematic, meaning that a constant offset in the assumed temperature field always leads to a bias in the isotopic ratios in the same direction. In contrast, Supplementary Fig. 1 shows that a temperature offset of 5 K only produces deviations of 1-2% in the ratio between the abundances of 13 C 16 O and 12 C 18 O. This indicates that it is indeed the retrieval of 12 C 16 O that is most affected by inaccuracies in the temperature, and the ratio between the abundances of the minor isotopes is a useful magnitude to better interpret the analysis of the isotopic ratios in this study.

Validation of the retrieved vertical profiles
Simultaneous measurements made by the NIR and MIR channels on ACS provide a unique opportunity to validate the measurements from both channels. Numerous validation exercises have been performed in previous studies, focusing on the retrievals from different spectral ranges or ACS MIR secondary grating positions 23,37,42 . In particular, Fedorova et al. 23 used all three ACS channels to validate the retrievals of the CO abundances from secondary grating position 6, data that are used in the present study. Here, we perform a similar exercise, but including comparisons of simultaneous measurements of temperature too.
The temperature retrieval is the first step of the retrieval scheme presented in this study. We use the temperature profiles from the simultaneous ACS NIR channel measurements as the a priori guess in our retrievals, given that the temperature retrievals from that channel can extend to higher altitudes than the ones derived from secondary grating position 6. To validate the retrievals of the temperature from both channels and ensure that the temperatures derived in this study are driven by the data and not the prior information, we perform a series of tests using the measurements made during the first 20 orbits. In these test retrievals, we select a prior profile given by the ACS NIR temperature profile with an offset of ±10 K.
Supplementary Figure 2 shows an example of the results for the first three measurements. These figures highlight that the retrieved temperature profiles below ~55 km from ACS MIR are driven by the measured data and are not sensitive to the assumed a priori profiles. Above this altitude, where there is no information in the spectra, the retrieved profiles tend to their respective a priori information. Using the ACS NIR profiles as the a priori guess in our retrieval scheme ensures that the retrieved profiles from ACS MIR will also be accurate above this altitude, where we can still measure the abundances of the CO isotopes.
Once the temperature profiles are retrieved, these are fixed, and the gaseous abundances of the CO isotopologues are retrieved using orders 246, 248 and 249. Supplementary Figure 3 provides a general overview of the comparison between the ACS NIR and MIR datasets, showing the differences in retrieved abundances and temperatures. The differences between the two channels are on average centred on zero, considering that different absorption bands are used and the uncertainties of the retrievals, which indicates that there are no strong systematic differences between both datasets.
Finally, once the retrieved abundance of the CO isotopes is determined, the isotopic ratios are derived. To ensure that our measurements and conclusions about the isotopic ratios are robust, we compare the retrieved isotopic ratios when using the temperature profiles from the NIR and MIR channels during the first 20 orbits (Supplementary Fig. 4). We find that using either dataset we would draw similar conclusions: the 13 C/ 12 C ratio is depleted with respect to the standard, while the 18 O/ 16 O ratio is consistent with a value of 1 VSMOW within the retrieved uncertainties. The ratio between the two Article https://doi.org/10.1038/s41550-023-01974-2 (that is, 13 C 16 O/ 12 C 18 O), not dependent on the temperature field, provides a robust confirmation that the depletion of 13 C in CO is real.
It can be observed in Supplementary Fig. 4 that the vertical profiles of the 13 C/ 12 C and 18 O/ 16 O isotopic ratios show a sudden increase below 20 km when fixing the temperature profiles from the ACS NIR channel. The magnitude of the increase is equal for both isotopic ratios, which is evidenced by the lack of such an increase in the 13 C 16 O/ 12 C 18 O ratio. This sudden increase of the isotopic ratios towards the surface could be caused by an unknown process that fractionates both isotopologues equally (that is, δ 13 C = δ 18 O). However, a more plausible explanation is that they are the result of a bias in the retrievals at these lowest altitudes in their common factor, the abundance of 12 C 16 O. In particular, the increasingly abundance of dust towards the surface negatively impacts the sensitivity of the measurements to the gaseous abundances and the temperature field. As shown in Supplementary  Fig. 1, a bias of temperature field of just a few kelvins can lead to substantial differences in the retrieved abundance of 12 C 16 O, while the retrieval of the minor isotopes is more insensitive to such deviations. Such a bias in the abundance of 12 C 16 O at the lowest altitudes can lead to equal deviations in the calculation of the 13 C/ 12 C and 18 O/ 16 O isotopic ratios, following δ 13 C = δ 18 O.
The simultaneous retrieval of the temperature field and the abundances of the CO isotopes with the ACS MIR channel mostly mitigates the bias found at the lowest altitudes ( Supplementary Fig. 4). However, the reduced sensitivity of the measurements at these altitudes due to the presence of dust can lead to a similar but smaller systematic behaviour in our retrieval scheme. Supplementary Figure 5 shows the averaged profiles of 13 C/ 12 C and 18 O/ 16 O in CO derived from the whole ACS MIR dataset used in this study. Although substantially lower than the reported uncertainties, both the C and O isotopic ratios appear to increase below 15 km towards the surface. The relationship between the two isotopic ratios at these altitudes follows δ 13 C = δ 18 O, which suggests that this feature is not real, but caused by a systematic bias in the retrievals. Since the reality of such a feature is questionable on the basis of the discussion above, we currently refrain from including the measurements at these altitudes and their potential science implications until we investigate this further with other ACS MIR measurements in other spectral ranges.

Photochemical model
The 1D photochemical model aims to solve the continuity equation given by where n i is the number density of a given species, P i and L i represent the production and loss of density due to the photochemical reactions, and the last term represents the diffusion between layers, where ɸ is the flux. The flux is calculated considering both eddy and molecular diffusion and is given by where K and D i are the eddy and molecular diffusion coefficients, H 0 is the mean scale height, H i is the gas-dependent scale height, ɑ is the thermal diffusion coefficient and T is the temperature. The diffusion term is calculated by finite differences 43 , which yields where the subscripts i and j represent the gas and layer, respectively. The coefficients in the finite-difference approximation are given by , where Δz is the altitude of each layer. The production and loss terms are calculated using the latest version of the photochemical scheme from the Mars Planetary Climate Model (PCM), which has been extensively used to model the photochemical cycles in the atmosphere of Mars [44][45][46]  It must be noted that in the version for the 18 O chemistry the total mass is not conserved, as we do not include 18 O or 18 O( 1 D) in the chemistry scheme. The only 18 O-bearing species included in the model apart from CO and CO 2 is 18 OH, whose isotopic ratio is fixed throughout the simulation to follow that measured by the Curiosity Rover in CO 2 , and whose recombination with C 16 O is assumed to produce no isotopic fractionation. Inclusion of all 18 O-bearing species would be required to model the complete chemistry of 18 O in the atmosphere of Mars. However, here we restrict our analysis to model the relative fractionation of the 18 O/ 16 O isotopic ratio between CO 2 and CO, which is not expected to be affected by this simplification, since CO molecules are solely produced by the photolysis of CO 2 , which is included in the scheme. The continuity equation is solved using a second-order Rosenbrock algorithm 47,48 . The density at the next time step is calculated following n t+1 = n t + 1.5 Δt g 1 + 0.5 Δt g 2 , where I is the identity matrix, γ = 1 + 1/√2, Δt is the time step, f(n) is the right-hand side of the differential equation (that is, the chemistry and transport terms evaluated at the density n), g 1 and g 2 are the intermideate factors allowing the calculation of the first and second-order solutions and J is the Jacobian matrix of the combined chemistry-transport system. Since the densities of a given species in a given layer are only influenced by the densities of the other gases in the same layer (that is, chemistry) and the densities of the same gas in the adjacent layers (that is, diffusion), the Jacobian matrix takes the form of a block tridiagonal matrix, which is efficiently inverted using the Thomas algorithm.
The choice of the time step is made on the basis of the maximum local error, which is estimated by the difference between the first-order and second-order solutions, following ε = n t+1 − n t+1 1 = (n t + 1.5 Δt g 1 + 0.5 Δt g 2 ) − (n t + Δt g 1 ) .
The time step is self-adjusted by Δt t+1 ≈ Δt t /√ε max , which ensures that the convergence of the system to a steady-state solution will capture the orders-of-magnitude difference in the chemical and diffusion timescales while optimizing the computational time.
The 1D photochemical model requires several inputs to represent the composition and mixing of the atmosphere of Mars. The initial temperature and density profiles are derived using an average of vertical profiles generated from the 'standard climatology' scenario of the Mars Climate Database v6. 1 (ref. 49) at different times and locations. In particular, we generate vertical profiles of temperature and gas number densities from 0 to 200 km in a grid of latitudes, local times and solar longitude L S . The latitudes expand from −80° to 80° with a step of 20°. The local time expands from 0 to 24 h with a two-hour step. The L S expands the whole year, with a step of 10°. The calculated profiles are then averaged to initialize the vertical profiles in the 1D model ( Supplementary Fig. 6). The abundances of CO 2 and CO from the Mars Climate Database are then divided into the different isotopologues by assuming a given isotopic ratio, and using where the bracketed quantities represent the abundances of the different species, and R is the isotopic ratio measured in CO 2 with the Curiosity Rover (that is, 13 30 , which were taken from other previous studies 43,50 . In particular, the eddy diffusion coefficient is given by K (z ≤ 60 km) = 10 6 cm 2 s −1 , K (z > 60 km) = 2 × 10 13 n(z) −1/2 cm 2 s −1 , as shown in Supplementary Fig. 6. The molecular diffusion coefficient is given by  Fig. 6).
The boundary conditions are similar to those used in previous studies 51 . We consider the top boundary conditions to be a zero flux for all species but for H 2 , H and O. For H 2 and H, we fix the escape velocities to be v H2 = 3.4 × 10 1 cm s −1 and v H = 3.1 × 10 3 cm s −1 , while for O we fix the escaping flux to be ɸ O = 1.2 × 10 8 cm −2 s −1 . For the lower boundary conditions, we set a zero flux for all species but H 2 O, whose density is assumed to be fixed below 50 km.
Using these inputs and the presented model setup, we run a simulation for 20 yr, by which time the O and C isotopic ratios in both CO and CO 2 have converged to a steady-state solution, as shown in Extended Data Fig. 4.

Fractionation during the photolysis of CO 2
The photolysis cross-sections of CO 2 used in the chemistry scheme of the LMD Mars PCM are composites of different laboratory measurements in different spectral ranges [52][53][54][55][56] Figure 1 shows how the calculated cross-sections for 12 C 16 O 2 compare with those tabulated in the chemistry scheme, together with a comparison between the cross-sections calculated for the different isotopologues. These results suggest that the photolysis cross-sections of 12 C 16 O 2 are larger than those of the minor isotopes (that is, 13 C 16 O 2 and 18 O 12 C 16 O) by a factor of 1.0-1.5 and 1.0-1.3, respectively. The ratio between the cross-sections of 12 C 16 O 2 and those of the minor isotopes also depends on the temperature, with that at 195 K being on average 5% larger than that at 295 K.
It is important to note that the cross-sections for the minor isotopes are only available for a limited spectral range (>150 nm), and these are lower than the cross-sections elsewhere (<150 nm), where we assume that they are equal for all isotopes. To evaluate where the cross-sections of >150 nm are most effective in the photolysis of CO 2 molecules in the atmosphere of Mars, we calculate the photolysis J values as a function of altitude (Extended Data Fig. 2). These results show that the J values are approximately constant above ~100 km, where the atmosphere is optically thin. Below this altitude, the atmosphere becomes optically thick for wavelengths lower than 150 nm, and the J values decrease with decreasing altitude as the atmosphere also becomes optically thick for wavelengths of >150 nm. Therefore, while the cross-sections of <150 nm are important above 100 km, their effect becomes negligible below 100 km, the region where the ACS observations can measure the isotopic fractionation between CO 2 and CO.
Extended Data Figure 2b shows the ratio between the photolysis rates of 12 C 16 O 2 and those of 13 C 16 O 2 and 18 O 12 C 16 O, which is higher than one at all altitudes. These results suggest that the photolysis of 12 C 16 O 2 is faster than that of the minor isotopes, meaning that 12 C 16 O will be more efficiently produced than 13 C 16 O and 12 C 18 O. The net effect of this process is to produce a depletion of the 13 C/ 12 C and 18 O/ 16 O ratios in CO with respect to those in CO 2 , and the magnitude of this effect is larger for the C isotopes than the O ones.

Fractionation during the recombination of CO + OH
The recombination of CO with the hydroxyl radical OH is an important reaction in the chemistry of the atmosphere of Mars, which is responsible for the stability of CO 2 (refs. 20,21). To model the isotopic fractionation of the C and O isotopes in this reaction, we use the data derived from laboratory measurements 25 . These measurements determined Article https://doi.org/10.1038/s41550-023-01974-2 the kinetic isotope effect at a range of pressures from 0.2 to 1.35 atm (Extended Data Fig. 3). To extrapolate the measurements to the low pressure of the Martian atmosphere, we fit this dependence using a second-order polynomial function. The fitted polynomial functions for the C and O isotopes are respectively k 13 /k 12  where the ratios 13 k/ 12 k and 18 k/ 16 k represent the ratios between the reaction rates of the minor isotopes (that is, 13 CO and C 18 O) and those of the major one (that is, 12 C 16 O) and p is the pressure in pascals. At the low pressures of the Martian atmosphere, this reaction is faster for the minor isotopes, which yields a preferential transfer of 13 C and 18 O over 12 C and 16 O from CO to CO 2 . Therefore, the net effect of this reaction is to produce a depletion of the 13 C/ 12 C and 18 O/ 16 O isotopic ratios in CO with respect to CO 2 . However, given the small magnitude of this fractionation factor with respect to that produced by the photolysis of CO 2 , this process is expected to have a very minor effect in the overall fractionation between CO 2 and CO.

Data availability
The datasets generated by the ExoMars TGO instruments analysed in this study are available in the ESA Planetary Science Archive (PSA) repository, https://archives.esac.esa.int/psa, following a six months prior access period, following the ESA Rules on Information, Data and Intellectual Property. The data products generated in this study (retrieved atmospheric parameters) can be downloaded from a Zenodo repository 57 .

Code availability
The spectral fitting and retrievals were performed using the NEMESIS radiative transfer and retrieval algorithm, which can be downloaded from a Zenodo repository 58 . The photochemical code used to model the isotopic fractionation in the atmosphere of Mars can be downloaded from a Zenodo repository 59 . Interested users are encouraged to contact the corresponding author for the usage of these tools.