Potassium isotope composition of Mars reveals a mechanism of planetary volatile retention

Significance

Using spacecraft data and elemental abundances derived from martian meteorites, earlier studies set a paradigm of a volatile- and water-rich Mars relative to Earth. Nevertheless, inherent difficulty in determining the volatile budget of bulk silicate Mars (BSM) makes it challenging to directly compare the extents of volatile depletions among differentiated bodies in the Solar System. This study provides an alternative for evaluating the nature of volatiles on Mars using potassium (K) isotopes. The K isotopic composition of BSM and the strong correlation between δ⁴¹K and planet mass reveals that the sizes of planetary bodies fundamentally control their ability to retain volatiles. This could further shed light on the habitability of planets and assist with constraining unknown parent body sizes.

Abstract

The abundances of water and highly to moderately volatile elements in planets are considered critical to mantle convection, surface evolution processes, and habitability. From the first flyby space probes to the more recent “Perseverance” and “Tianwen-1” missions, “follow the water,” and, more broadly, “volatiles,” has been one of the key themes of martian exploration. Ratios of volatiles relative to refractory elements (e.g., K/Th, Rb/Sr) are consistent with a higher volatile content for Mars than for Earth, despite the contrasting present-day surface conditions of those bodies. This study presents K isotope data from a spectrum of martian lithologies as an isotopic tracer for comparing the inventories of highly and moderately volatile elements and compounds of planetary bodies. Here, we show that meteorites from Mars have systematically heavier K isotopic compositions than the bulk silicate Earth, implying a greater loss of K from Mars than from Earth. The average “bulk silicate” δ⁴¹K values of Earth, Moon, Mars, and the asteroid 4-Vesta correlate with surface gravity, the Mn/Na “volatility” ratio, and most notably, bulk planet H₂O abundance. These relationships indicate that planetary volatile abundances result from variable volatile loss during accretionary growth in which larger mass bodies preferentially retain volatile elements over lower mass objects. There is likely a threshold on the size requirements of rocky (exo)planets to retain enough H₂O to enable habitability and plate tectonics, with mass exceeding that of Mars.

Examining the presence, distribution, and abundance of volatile elements and compounds, including water, on Mars has been a central theme of space exploration for the past 50 y. The majority of all past, ongoing, and future Mars missions involve the direct or indirect study of volatile element inventories, including the recent “Perseverance” and “Tianwen-1” missions (1, 2). The direct study of volatiles in martian meteorites, along with remote sensing efforts, have significantly broadened understanding of the volatile inventory of Mars and spurred the development of competing bulk chemistry models for Mars. These models can broadly be divided into three groups: 1) those based on cosmochemical implications of elemental ratios (3–5), 2) those attempting to reproduce the martian O isotope composition by mixing different proportions of chondritic materials (6, 7) and equating these to volatile abundances, and 3) those combining spacecraft data and meteorite chemistry to estimate the composition of bulk silicate Mars (8). These models all adopted the ratios of volatile element K to the refractory elements U and Th as a proxy of volatile depletion because these elements are all highly incompatible and lithophile during igneous processes, that is, such ratios are not strongly affected by partial melting followed by melt fractionation that leads to the formation of basaltic rocks and their derivatives that constitute the present-day martian crust. Furthermore, the concentrations of K, Th, and U of this martian crust can be measured remotely from orbit using gamma-ray spectrometry (GRS). All previous models for the composition of bulk silicate Mars, as well as GRS data of exposed martian surface materials, have shown that Mars has elevated K/Th as well as higher contents of a greater suite of moderately volatile elements relative to Earth (Fig. 1), together implying a volatile-rich early Mars (8–10). A caveat with these models is the inherent difficulty in determining the K/Th of bulk silicate Mars from surface data as well as the marked inconsistency between meteorite analyses and GRS data of martian surface regions (9).

Fig. 1.

Potassium to thorium ratios versus the corresponding K concentrations of martian meteorites (basaltic, olivine-phyric, and lherzolitic shergottites and other categories), the martian surface detected by the Mars Odyssey Gamma Ray Spectrometer, and terrestrial mid-ocean ridge basalts (MORB) and ocean island basalts (OIB). To avoid potential terrestrial contamination effects, only meteorite falls and Antarctic finds are plotted. Data are from compilations for martian meteorites (47, 57–63), Mars Odyssey GRS data from ref. 64, MORB (65), and for OIB from GEOROC. The bulk silicate Earth has a K abundance of 240 ppm and a K/Th of 2,900 (66). Note the systematically high K/Th for martian surface materials measured by GRS compared with martian meteorites. Terrestrial rocks, especially MORB, exhibit a wide K/Th range overlapping with martian meteorite samples.

An alternative means of examining the volatile history of Mars is by measuring the isotopic ratios of moderately volatile elements (MVE) in martian meteorites. Of the MVEs, which are defined as having 50% equilibrium condensation temperatures (50%T_c) of less than 1,335 K at a total pressure of 10⁻⁴ bar for a solar system gas composition (11), K is one of the most abundant (50% T_c = 1,006 K(11)). The isotopic ratios of K in igneous rocks from planetary bodies are insensitive to igneous processes [e.g., melting and fractional crystallization (12)] and secondary effects such as impact-induced vaporization (13) and eruptive degassing (14) and thus are a strong proxy for volatile depletion in planetary interiors. Here, the K isotope compositions of 20 martian meteorites are reported. These meteorites have previously been established to originate from Mars on the basis of triple-oxygen isotope systematics, trapped noble gas inventories, and the generally young crystallization ages (<1.34 Ga as with the majority of martian meteorites) (15–18). The 20 examined meteorites cover a range of rock types (basaltic, olivine-phyric, lherzolitic, and picritic shergottites, nakhlites [clinopyroxene-rich cumulates], a chassignite [cumulate dunite], and a basaltic crustal breccia [NWA 7034], SI Appendix, Table S1) and geochemical signatures (incompatible element-enriched, -intermediate, and -depleted shergottites). With the exception of the basaltic breccia NWA 7034, only observed meteorite falls and Antarctic meteorite finds were considered in order to avoid uncertainties related to terrestrial contamination and alteration, which commonly affect hot desert finds (19).

Potassium Isotopic Composition of Bulk Silicate Mars

Potassium isotopic compositions of martian samples, reported as δ⁴¹K = ([(⁴¹K/³⁹K)_sample/(⁴¹K/³⁹K)_standard] − 1) × 1,000), range from −0.59 to −0.08‰ (Fig. 2A). Where different subfragments of meteorite samples could be analyzed (SI Appendix, Table S1), all but two samples yielded analytically indistinguishable K isotopic compositions (SI Appendix). The δ⁴¹K values of the two aliquots of Nakhla (a witnessed fall) differ by 0.34‰. This feature likely reflects postcrystallization aqueous modifications on Mars as previously suggested for nakhlites (20). The lower K/Al and K/Nb as well as elevated Ba/La in the isotopically light Nakhla aliquot suggest that the K isotopic composition of this aliquot stems from an alteration of primary igneous minerals. Potassium isotope fractionation during aqueous alteration on the martian surface is not yet understood. Nonetheless, terrestrial analogs show that regardless of weathering environment and secondary minerals formed, the weathered product is an isotopically light endmember in the global K cycle (21, 22). These results hint at an enrichment of K and other volatiles in the martian crust and that the K cycle on Mars is likely to be complex at the surface.

Fig. 2.

(A) Potassium isotopic compositions of martian meteorites. Data are compiled from this study and refs. 19 and 23. The dashed line and light gray bar represent the δ⁴¹K_NIST of bulk silicate Earth (BSE, –0.43 ± 0.17‰ defined by ref. 24); the darker gray bar represents the BSE value (–0.48 ± 0.03‰) defined by ref. 56. No significant differences are observed among mean values of each martian lithology. Samples with different subsamples from different sources analyzed in this study and compiled from the literature are labeled (e.g., Shergotty, Zagami, Nakhla, Tissint, EETA 79001A, and MIL 03346). Color scheme for shergottites: green denotes depleted, orange intermediate, and red enriched shergottites. (B) Histogram and “summed Gaussian” plot of the K isotopic compositions of Mars, the Earth, Moon, and the asteroid 4-Vesta. Bins for the histogram are in 0.05‰ increments. (C) Potassium isotopic compositions of the four parent bodies. The shaded areas and dashed lines denote the δ⁴¹K_NIST of the bulk silicate planets. The Earth is represented by a set of mid-ocean ridge basalts, oceanic basalts, and back-arc basin basalts, which provide an estimate for bulk silicate Earth (24). The K isotopic composition of the average upper continental crust (UCC) is also presented (67). Lunar mare basalts are plotted as representative of the bulk silicate Moon to eliminate the effects of secondary K isotope distributions on the Moon (25, 68). The δ⁴¹K_NIST of the asteroid 4-Vesta is represented by a group of eucrites and howardites reported in ref. 19. Data for Mars are compiled from refs. 19 and 23 and this study. Only meteorite falls and Antarctic finds are considered in calculating the K isotopic compositions of bulk silicate values for planets.

Excluding the isotopically light Nakhla aliquot and the hot desert meteorite NWA 7034, martian meteorites measured in this study and refs. 19 and 23 define a mean δ⁴¹K of −0.28 ± 0.18‰ (2 SDs, n = 30). Potassium tends to be quantitively removed into the melt fraction during the partial melting of mantle rocks owing to its incompatibility in the residual mantle mineralogy; accordingly, the K isotopic composition of the melt is expected to be unfractionated with respect to its mantle source (12). No correlation is found between bulk Mg# [100 × Mg/(Mg + Fe) in atomic ratios] and their corresponding δ⁴¹K. Moreover, there is no resolvable difference between the mean δ⁴¹K values of shergottites (including their petrological subgroups), nakhlites, Chassigny, and NWA 7034, indicating only a limited K isotopic fractionation in Mars. We hence use the mean δ⁴¹K of −0.28 ± 0.18‰ as a direct proxy for the K isotopic composition of bulk silicate Mars. This newly determined δ⁴¹K value indicates that bulk silicate Mars is ∼0.2‰ heavier than bulk silicate Earth [−0.43 ± 0.17‰ (24)] but isotopically lighter than the bulk silicate Moon and asteroid 4-Vesta (the probable parent body of howardite, eucrite, and diogenite meteorites) by ∼0.2‰ [−0.07 ± 0.09‰ (25)] and ∼0.65‰ [+0.36 ± 0.16‰ (19)], respectively. These differences are significant on the basis of summed Gaussian distributions of δ⁴¹K values (Fig. 2B) and statistically significant as determined by a two-sample Student’s t test (P_{t test} << 0.001 between any two group means).

Nebular Processes in Shaping the K Isotope Systematics of Planetary Bodies

A key issue relating to the differences in K isotope compositions among various planetary bodies is whether these reflect mass-dependent variations in “nebular” processes, mass-independent variations among different nucleosynthetic sources, or planetary-scale volatile depletion processes. Nucleosynthetic variations of K are considered unlikely, as known nucleosynthetic isotope anomalies to date are limited to elements having 50% T_c above 1,400 K (26), except for the highly volatile noble gases. Moreover, evidence against nucleosynthetic variability as the cause of K isotope variations is found when considering δ⁴¹K versus the mass-independent μ⁴⁸Ca (27). Variable μ⁴⁸Ca among inner Solar System bodies has been attributed to a rapid change in the composition of disk materials associated with early mass accretion to the proto-Sun rather than spatial variations in nucleosynthetic components (28). Crucial to this assessment is that the μ⁴⁸Ca values of the Earth and Moon are indistinguishable (SI Appendix, Fig. S1). The decoupling of μ⁴⁸Ca and δ⁴¹K of the Earth and Moon along with the slight negative correlation between μ⁴⁸Ca and δ⁴¹K in the inner disk bodies suggest that neither nucleosynthetic anomalies nor ⁴¹K production from ⁴¹Ca decay can be the main cause of K isotope variations.

Another potential cause of K isotopic fractionation is nebular-scale partial evaporation or incomplete condensation of volatiles in planetary feedstocks before planet formation (29). In this scenario, variations in δ⁴¹K would occur because of different extents of volatile depletion in precursor materials (29). The different accretionary time scales for planetary bodies may have enabled initial admixing of early-accreted volatile depleted components with later-stage volatile-rich materials, especially in larger planetary bodies. It is, however, difficult to reconcile the isotope effect generated during gas–dust interactions (condensation and evaporation) with the tight correlation observed between δ⁴¹K and parent body sizes (Fig. 3A and SI Appendix, Fig. S2). This is because δ⁴¹K heterogeneity in building blocks would be anticipated to produce δ⁴¹K variations among bulk planets that contradict the observation, that is, the distribution of volatile-rich planetary building blocks in the outer Solar System and “drier” feedstocks in the inner Solar System would be expected to have led to more scattered K isotopic compositions that might correlate with the heliocentric distance of planetary bodies. Additional evidence precluding a nebular volatile depletion is the decoupling of K/U and K/Th from K isotopic compositions. Different accretion timescales for planetary bodies may have enabled initial admixing of volatile-depleted components with later-accreted volatile-rich feedstocks, especially to large parent bodies such as the Earth. If nebular volatile depletion and accretion durations were the only contributing factors accounting for K isotope variations among bulk planets, then K/U and K/Th should also scale with planetary body size. However, intermediate mass objects in the Solar System, such as Mercury (K/U of ∼12,800) (30) and Mars (K/U of ∼15,000) (31), have K/Th and K/U values broadly similar to, or slightly higher than, those of larger mass objects such as Earth (K/U of ∼13,800) (32) and Venus (K/U of ∼10,000) (31) based on surface measurements. Taken together, it is unlikely that the K isotope variations in the inner Solar System are a consequence of nebular processes but rather reflect the loss of volatile elements primarily during the formation and accretion of the planets.

Fig. 3.

(A) Average δ⁴¹K of Mars, Earth, Moon, and Vesta versus their corresponding surface gravity. Potassium isotope data are compiled from refs. 19, 23 to 25, and 68 and this study. (B) Average δ⁴¹K of enstatite chondrites and the four differentiated terrestrial planetary bodies versus their Mn/Na. The K isotope data are compiled from refs. 9, 23 to 25, 68, and 69. The bulk Earth Mn/Na is from ref. 35 and takes into account the metal silicate partitioning behavior of Mn during core formation in Earth. The Mn/Na data of enstatite chondrite, Mars, Moon, and 4-Vesta are from ref. 34. Error bars on data points represent two SDs (2SD) on the full dataset. The goodness of the linear regression fit is shown as R². The black solid lines represent the 95% confidence error envelopes (shaded area). (C) Average δ⁴¹K of four parent bodies versus their corresponding water content estimates. The water contents are from ref. 39. Note a negative correlation between water contents and their corresponding δ⁴¹K among four planetary bodies.

Volatile Loss during Accretionary Growth of Planetary Bodies

Day and Moynier (33) suggested that the volatile element inventories of planetary bodies in the Solar System are related to physical characteristics, including surface gravity, although the relationship was not well defined. A striking feature of the new average δ⁴¹K data for bulk silicate Mars is that it lies on a negatively correlated trend (R² = 0.995) with the three other known bulk silicate planetary compositions in the inner Solar System as a function of surface gravity and other related parameters (Fig. 3A and SI Appendix, Fig. S2). The mean δ⁴¹K values for the respective parent bodies in the inner Solar System also correlate (R² = 0.870) with their Mn/Na (Fig. 3B), an independent chemical index for late-stage volatilization during accretionary growth (34–36), as a stronger loss of Na relative to Mn was favored under increasingly oxidizing conditions relative to the solar nebula (34). Additionally, Rb behaves geochemically similar to K while being somewhat more volatile both under solar nebular conditions and during planetary evaporation (11, 37). Although the currently available data are limited for stable Rb isotope systematics (38), the mean δ⁸⁷Rb of Earth, the Moon, and 4-Vesta also negatively correlate with surface gravity (SI Appendix, Fig. S3). Most importantly, the negative correlation between planetary sizes and stable isotopic compositions also applies to the highly volatile element H (Fig. 3C) in which bulk planet water contents correlate with their K isotopic compositions (39). This suggests that the same underlying cause is responsible for the depletion of K and water among planetary bodies, which is consistent with a recent study showing that enstatite meteorites contain sufficient quantities of H to account for Earth’s water content, obviating the need for a late delivery of water through carbonaceous chondrites or comets (40). These relationships confirm that smaller differentiated planetary bodies, with lower gravity, underwent greater volatile depletion and stronger K isotope fractionation than larger bodies relative to their precursor materials. More importantly, the relationships suggest that K depletion resulted from planetary processes affecting the bodies themselves rather than from nebular processes acting on their precursors.

Vapor loss during the accretion of small, low-gravity bodies and concomitant chemical fractionation has been proposed as an important contributor to the elemental and isotopic budgets of planetary bodies (41–44). Variability among volatile-depleted accreting materials could have led to variations in the K isotope compositions of the planetary bodies relative to the initial δ⁴¹K of the Solar System. A “sum” effect (i.e., integrated over an object’s accretionary growth) of fractionation owing to vapor loss from planetary bodies has been proposed to account for Mg isotopic fractionation among planetary bodies on the basis of simulations of collisional accretion and a vapor-melt fractionation model derived from thermodynamic properties (Fig. 4A) (43). In this model, the Mg isotopic composition of the Earth requires such large vapor mass losses that it demands the Earth accreted from relatively volatile-rich planetary feedstocks once it grew larger than approximately Mars size to account for the abundances of moderately volatile elements such as K and Na. Such a model may account for the similarity of the K isotopic compositions of Earth and enstatite chondrites (presumable building blocks of the Earth). Nonetheless, it is unlikely to cause the tight correlation between bulk planetary δ⁴¹K and their current masses from integration over all accretionary collisions as observed in Fig. 3A. We hypothesize that such correlation instead implies that the ultimate K inventory of a bulk planet is dominated by the final accreted mass, regardless of its accretion and differentiation history, unlike more refractory elements such as Li and Mg that require more extreme temperatures for significant vaporization and are hence less likely to be dominated by a late-stage vaporization event (Fig. 4B) (43, 45). This hypothesis is further supported by the fact that the Moon exhibits higher δ⁴¹K than that of the Earth. The Moon is considered to have formed later in the Solar System’s history by a single-stage giant collision between a Mars-sized body and the proto-Earth that homogenized the stable isotopes of refractory elements (46). That the Moon does not exhibit the same K isotope composition as the Earth is a major impediment to a time-integrated volatile depletion mechanism for K, indicating that the volatile depletion of planetary feedstocks alone cannot explain the variations between δ⁴¹K and surface gravity.

Fig. 4.

Schematic diagram of mechanisms of volatile loss during accretionary growth that can potentially generate the K isotope compositions observed in different parent bodies. Two scenarios are shown here: (A) Time-integrated effect of vapor loss and isotope fractionation over a planetary body’s growth history. Two mechanisms are involved in vapor loss during accretion: impact-induced vaporization and vapor losses above impact-generated magma ponds (43). (B) Chemical fractionations dominated by a late-stage evaporation event. Large bodies are able to retain volatiles more efficiently once they reached sufficient sizes (“critical size”). In contrast, the gravity of small bodies is insufficient to prevent volatile losses, which results in preferential loss of vapor relatively enriched in light K isotopes, leaving the residual portion enriched in heavy K isotopes. Such K isotopic fractionation effects are shown in SI Appendix, Fig. S4, with different equilibrium fractionation factors under different temperatures. The new K isotope data presented here are in line with the conclusion that the ultimate K (and possibly MVEs with similar volatility) budget of rocky planetary bodies in our Solar System is dominated by vapor loss at a late stage in their accretion.

A “Dry” Versus “Wet” Mars

The K isotopic compositions among inner Solar System planetary bodies and their strong dependence on planetary sizes are consistent with isotopic fractionation during the depletion of MVE from these bodies. Small, differentiated bodies such as 4-Vesta and the Moon, irrespective of their formation processes, had insufficient gravity to prevent volatile losses, leading to the preferential escape of isotopically light K and resultant heavy silicate portions (Fig. 4B). In contrast, Earth, and by analogy Venus, reached sufficient size to retain volatiles that accreted to them more efficiently once they exceeded a certain size (Fig. 4B). The gas giants (e.g., Jupiter and Saturn) fall at the end of the spectrum, quantitatively retaining all volatile elements and compounds as evidenced by their atmospheric compositions resembling that of the Sun. An intriguing question arises regarding the mass threshold required to retain K (and other elements and compounds with similar volatility), partially or fully, in rocky planets. Previous numerical models placed the threshold at approximately the size of Mars, above which vapor loss ceased (43). Our new K isotope data are consistent with planets larger than Mars enabling enhanced retention of volatiles and having highly variable K/Th and K/U. As such, variability in elemental ratios may demonstrate the variability in the volatile abundances of planetary feedstocks and later crustal differentiation processes.

The new K isotope results for Mars, in the broader context of planetary volatile depletion, are seemingly at odds with the longstanding paradigm based on K/Th and K/U that Mars is volatile rich. The new K isotope constraint indicates that the volatile abundances of Mars were primarily set during accretionary growth. The K to refractory element ratios, however, were likely modified from the original ratios acquired during accretion (SI Appendix). Supporting evidence comes from GRS surveys (9) of the martian surface, which consistently point to a volatile-rich crustal environment relative to igneous martian meteorites. Mars likely underwent an early-stage differentiation followed by a long-term isolation of mantle reservoirs as indicated by the greater variance in time-integrated Rb/Sr and Sm/Nd of martian meteorites compared with terrestrial igneous rocks (47, 48). The apparent high K/Th and K/U of Mars are therefore likely to be a consequence of sampling of a highly differentiated volatile-rich surface environment, a conclusion that is also consistent with heterogeneous volatile reservoirs previously reported for Mars from the D/H of martian meteorites (49). This differs from Earth, where some of the volatile elements are subducted and recycled into the mantle through active plate tectonic processes. On the basis of K isotopes, the volatile budget of bulk Mars is low, and the likelihood for the habitability of Mars is limited, as it lies roughly at the threshold in which a rocky planet is large enough to retain significant budgets of life-sustaining volatiles. The strong correlation between K isotopes and planetary gravity could therefore provide key insights into constraining parent body sizes (e.g., angrite and ureilite parent bodies) and for determining the “Goldilocks” zone for habitability of exoplanets on the basis of their masses in addition to distance from their host star.

Materials and Methods

As the current sample collection contains no martian meteorites directly excavated from the mantle, the K isotopic composition of bulk silicate Mars can only be inferred from the partial melting products of their mantle source regions. To a first order, martian lithologies can be divided into shergottites (basaltic, lherzolitic, olivine-phyric, and picritic on the basis of their mineralogy and texture; or depleted, intermediate, and enriched on the basis of incompatible trace elements contents), nakhlites (clinopyroxenites), chassignites (dunites), an orthopyroxenite (ALH 84001), a basaltic regolith breccia (NWA 7034/7533 and other pairings), and the >2 Ga augite basalts (NWA 7635/8159). Widespread shock metamorphism is recognized in most martian meteorites to various degrees (e.g., plagioclase is commonly transformed into maskelynite in basaltic shergottites; in more extreme shock events, high-pressure polymorphs can be present as well, including ringwoodite and silicate perovskite). The depleted and enriched shergottites likely crystallized at different oxygen fugacity, with enriched shergottites crystallizing under more oxidized conditions (along with high La/Yb, high initial ⁸⁷Sr/⁸⁶Sr, and negative ε¹⁴³Nd), depleted shergottites under more reduced conditions (along with low La/Yb, low initial ⁸⁷Sr/⁸⁶Sr, and positive ε¹⁴³Nd), and intermediate shergottites being a mixture of enriched and depleted endmembers. This suggests the existence of either heterogeneous martian mantle reservoirs (49) or an ancient martian crustal component and a primitive depleted mantle source (50). The nakhlite–chassignite suite exhibits both the same crystallization and ejection ages and is thought to have a comagmatic origin (51). A total of 20 martian meteorites (23 subsamples) were investigated in this study, covering a wide range of geochemical and petrologic types. Most samples from the suite have previously been analyzed for stable Li, Mg, Ca, and V isotope compositions (20, 52–54). High-precision data from previous studies were also included (19, 23).

The high-precision K isotope analytical procedure used here was adopted from ref. 55. About 10 to 150 mg of homogeneously pulverized samples was dissolved in a concentrated HF-HNO₃ mixture (∼3:1 vol/vol) either in Teflon beakers or Parr high-pressure digestion vessels. The samples were refluxed with 6 M HCl after being completely dried down on a hot plate. The fully dissolved samples were then desiccated and redissolved in 0.7 M HNO₃. Ultrapure water (Milli-Q, 18.2 MΩ cm) and double-distilled HCl and HNO₃ were used in sample dissolution and ion-exchange chromatography to reduce blank levels. The samples were centrifuged before loading onto columns for chemical purifications.

A three-step procedure was deployed to separate K from other matrix elements, including one “big column” (inside diameter [ID] = 1.5 cm, Bio-Rad Econo-Pac, resin volume of 19 mL in water medium) and two repeated “small column” steps (ID = 0.5 cm, Bio-Rad Glass Econo-Column, resin volume of 2.4 mL in 4N HCl). Bio-Rad AG50-X8 (100 to 200 mesh) cation-exchange resin was precleaned and conditioned in 0.7 M and 0.5 M HNO₃ for the “big” and “small” columns, respectively. Detailed protocols can be found in table 1 of ref. 55. All elemental concentration measurements, including major and trace elements, were performed using a Thermo Scientific iCAP Q Quadrupole inductively coupled plasma mass spectrometer (ICP-MS) with a typical uncertainty of 5% (2 relative standard deviations [RSD]). A K recovery rate of ∼100% is required to avoid artificial isotope fractionations during ion-exchange chromatography. Hence, fractions collected before and after K-collecting volumes were monitored to ensure >99% K yields for all samples measured in the course of this study. Furthermore, the proportion of matrix elements was constrained to within 2% of the total K inventories to eliminate drift introduced by matrix effects during measurements (55). The total procedure blank was 0.26 ± 0.15 µg, with a distinct K isotopic composition of –1.31 ± 0.05‰ (55). In all cases, contamination introduced by blanks was negligible.

Potassium isotope analyses were performed with a Neptune Plus MC-ICP-MS (Thermo Fisher Scientific). To enhance the instrumental sensitivity while minimizing the generation of hydrides and oxides, an APEX Omega desolvating nebulizer (Elemental Scientific) was used as the sample introduction system. The measurements were conducted on the low-mass “shoulder” of the peak in the instrument’s high-resolution mode (m/Δm > 10,000) to resolve the peaks of ⁴⁰Ar¹H⁺ and ⁴¹K⁺. A conventional standard sample standard bracketing technique was used, and concentrations between samples and standards were matched to within 3% to avoid concentration effects on K isotope measurements. National Institute of Standards and Technology Standard Reference Materials (NIST SRM) 3141a was used as the K standard solution. All results are expressed in the delta notation, where

$${\mathrm{\delta }}^{41}\text{K}\left(‰\right)=\left(\left[{\left({\hspace{0.17em}}^{41}\text{K}{/}^{39}\text{K}\right)}_{\text{sample}}/{\left({\hspace{0.17em}}^{41}\text{K}{/}^{39}\text{K}\right)}_{\text{NISTSRM}3141\text{a}}–1\right]\times 1000\right).$$

Each sample was measured approximately 10 times in an ∼300 ppb solution. The internal (i.e., within-run) reproducibility is reported as two SEs (2SE, typically ∼0.05‰) along with the weighted mean value for each sample. The long-term reproducibility (∼0.11‰, two SDs, 2SD) of the technique has been evaluated by routinely measuring the same geo-reference material over ∼20 mo in different analytical sessions (55). The same geo-reference material, BHVO-2, was also measured here (along with the martian samples) in each analytical session to monitor instrumental drift among different sessions and to undertake interlaboratory comparisons (SI Appendix, Table S1). Within analytical uncertainties, the K isotopic values for BHVO-2 reported here are indistinguishable from those found in other studies (19, 55, 56).

Data Availability

All study data are included in the article and/or SI Appendix.