Primitive lunar mantle materials at the Chandrayaan-3 landing site

Notably, the comparison of measured abundances of Na and K between the Chandrayaan-3 and previous landing sites reveal substantially lower values of these elements in the soils at the Chandrayaan-3 landing site, as shown in Fig. 3a, b (Table 1). Compared to Apollo 16 soils in the lunar highlands, the maximum concentrations of Na and K measured at the Chandrayaan-3 landing site are less by approximately 40% and 80%, respectively. This difference increases to about 50% for Na and 90% for K when compared to Apollo 12 and 14 soils, likely attributable to the presence of KREEP-rich rock components in the source regions^20,21. Contrary to Na and K, the highland soils analyzed at the Chandrayaan-3 landing site exhibit an enrichment in S relative to the highland soils from Apollo 16 and Luna 20 (Fig. 3c). The average S content of the Chandrayaan-3 landing site is 1200 ppm, with the overall range being approximately 300-500 ppm higher than that of the soils from Apollo 16 and Luna 20. Furthermore, a comparison with other Apollo sites that have highland soils, such as Apollo 15 and 17, reveals that they also have lower S values, ranging from 310-830 ppm, as shown in Fig. 3c. The Na-Al₂O₃, K-Al₂O₃, and S-Al₂O₃ trends in Fig. 3d-f (Table 2) suggest that the Apollo, Luna 20, and Chang’e-5 soils originated from distinct igneous sources compared to those at the Chandrayaan-3 landing site. It is important to investigate the origin of these asymmetries in abundances of Na, K, and S and infer the probable sources that led to the enrichment-depletion of these elements at the Chandrayaan-3 landing site.

To understand the Na, K, and S abundances observed at the Chandrayaan-3 landing site, we consider contribution from both indigenous and exogenous sources²² (Fig. 4). The landing site’s average K₂O content is 460 ± 70 ppm¹⁴, which is substantially lower than the <3500 ppm K₂O levels found in the low-K KREEP region²³. Further, the mean value of Na₂O abundance is 2640 ± 560 ppm at the landing site¹⁴, which is comparatively lower than the average Na₂O content in the Apollo 16 (4700 ppm) and Luna 20 (4000 ppm) highland landing sites. However, the S content (mean abundance: 1200 ± 200 ppm) in the landing site is higher relative to the highland soils from Apollo 16 (290-900 ppm) and Luna 20 (800 ppm) landing sites. For the highland soils, the lower and upper limits of the quantity of Type I CC meteorites are constrained as 0.7-1.7% based on the mean concentrations of iridium (Ir) and gold (Au)²⁴. The quantity of meteoritic material can be less than 1.7% based on the abundance of Ir in the average soil metal. The Type I CC meteorites contain approximately 0.5 wt% of Na, 0.05 wt% of K, and 6 wt% of S^{25,26,27,28,29}. Considering the constrained values of the quantity of Type I CC meteorites, it would contribute approximately 35-85 ppm Na, 3.5-8.5 ppm K, and 420-1020 ppm S to the southern high-latitude highland soils at the Chandrayaan-3 landing site^{25,26,27,28,29}. The Na and K contributions fall below the measurement uncertainties of these elements at the landing site¹⁴, suggesting Type I CC meteorites are an insignificant source for Na and K. Although the S contribution from Type I CC meteorites exceeds the total S content found in some lunar highland soils, such as those from the Apollo 16 and Luna 20 landing sites^25,30, it remains inadequate to explain the sulfur levels at the Chandrayaan-3 landing site. The S content measured along the Pragyan traverse path ranges between 900-1400 ppm (mean abundance: 1200 ppm), which is approximately 200-400 ppm higher than what would be contributed by Type I CC meteorites. Given that anorthosites contribute only 40 ppm S²⁵, the exceptionally lower Na and K values, coupled with higher S content, cannot be comprehended by contributions from KREEP and/or Type I CC meteorites. Consequently, such a scenario would have implications on the concentration of KREEP during the SPA basin impact.

The terrain’s nature, composition, and geological setting where the SPA basin impacted depend on the extent of LMO crystallization and the timing of the LMO overturn. The earliest lunar crust likely formed between 4.45 and 4.4 Ga³¹, with LMO crystallization persisting for another 150–200 million years³¹. This timeline aligns with the ages of lunar meteorite Kalahari 009 (4.369 ± 0.007 Ga)³², FAN 60025 (4.360 ± 0.003 Ga)³³, and KREEP-rich samples (4.22-4.27 Ga)³¹. Models of LMO crystallization³⁴ suggest that urKREEP liquid residue⁶, the final 0.5% of the LMO, formed after 99.5% LMO had crystallized³⁴. The SPA basin, which formed around 4.3 Ga³⁵, predates both the LMO overturn³⁶ and urKREEP crystallization^6,35. Owing to the immense scale of the SPA basin, it excavated substantial material from the lower crust/upper mantle^37,38,39 (Fig. 4). However, urKREEP was not a predominant component of the lower crust/upper mantle at the time and place of the SPA basin impact, possibly explaining the low Na and K abundances at the Chandrayaan-3 landing site. In this case, it is likely that the higher S content be related to the SPA basin ejecta from the lunar interior, such as the lower crust/upper mantle.

On a global scale, S concentrates in the residual silicate melt sandwiched between the anorthosite crust and the cumulate mantle (Supplementary Fig. 1a). Studies suggest that this late stage residual melt becomes progressively enriched in troilite (FeS), with its concentration potentially surpassing the S concentration at sulfide saturation (SCSS) during the end of the LMO crystallization⁴⁰. S begins to exsolve as FeS after about 85–90% of the magma ocean has crystallized^40,41. Modeling results demonstrate that FeS sulfide liquid saturation occurs when ~96-98% of the LMO has crystallized⁴¹. Ultimately, all S is eventually exsolved as FeS at the end of the LMO crystallization process. The segregated sulfide is then transported to greater depths during the overturn phase^40,41. The SPA basin has excavated FeS from the lunar interior, including the lower crust/upper mantle of the Moon³⁷, as shown in Supplementary Fig. 1b. The suggested timing of KREEP formation implies that S in the lunar interior was excavated possibly when the sulfide saturation was being reached, while the KREEP layer still in the process of formation. Therefore, the SPA basin ejecta would have exhibited elevated S values without corresponding elevated KREEP signatures. The subsequent impacts on the SPA basin ejecta likely led to the mixing of sulfur-rich and sulfur-poor materials in the vicinity (Supplementary Fig. 1c). The close proximity of the Chandrayaan-3 landing site to the SPA basin rim increases the likelihood of mixing of the sulfur-rich SPA basin ejecta and the adjacent units. However, the amount of S contributed by SPA basin ejecta to the Chandrayaan-3 landing site is still unclear.

We now focus on assessing the S contribution from SPA basin ejecta, composed of lower crust and upper mantle materials, to the soils at the Chandrayaan-3 landing site. Assuming ferroan anorthosite (FAN), pyroxene, and olivine as corresponding endmember for upper crust, lower crust⁴², and upper mantle^43,44, the triangular plot (Supplementary Fig. 2) suggests a mixing of nearly 8% of lower crust and 14% of upper mantle materials at this site¹⁴. Apollo 17 norite 78236, a pristine lower crust sample⁴⁵, contains 200 ppm S¹⁵. At the Chandrayaan-3 landing site, the concentrations of anorthosite and Type I CC meteorites are 77%¹⁴ and 0.7-1.7%²⁴, respectively. The remaining 21.3-22.3% is distributed between the lower crust (8%) and upper mantle (14%) materials (Table 3). Typically, 8% lower crustal materials would add only 16 ppm of S to the soils at the Chandrayaan-3 landing site, which is insufficient to explain the observed excess of 200-400 ppm S. To first order, this suggests that the dominant component of S was likely contributed from the upper mantle. Since S was excavated from the upper mantle in the form of FeS, the mean Fe content at the Chandrayaan-3 landing site (4.9 ± 0.25 wt%)¹⁴ is correspondingly higher than the mean Fe content (4 wt%) measured in Apollo 16 highland soils.

We have strengthened our S abundance calculations by estimating the concentration of nickel (Ni) at the Chandrayaan-3 landing site using the same set of mixing model components applied to S (Supplementary Table 1). Type I CC meteorites, which contain 1 wt% Ni, would contribute 178.5 ppm Ni to lunar soils for a 1.7% contribution of Type I CC material. Further, the Ni content in the upper crust (FAN), lower crust (pyroxene), and upper mantle (olivine) is approximately 30 ppm, 25 ppm, and 400 ppm, respectively. The APXS instrument measured a Ni concentration of 300 ± 50 ppm at the landing site. Supplementary Table 1 shows the Ni contributions from individual mixing model components, yielding a mean estimated Ni abundance of approximately 260 ppm. This value closely aligns with the APXS measurement, suggesting that the mixing model used in the study is robust.

To ascertain the S content of the upper mantle, the ideal samples would be those in which the S concentration of the lunar mantle has been directly quantified. However, such direct samples are lacking in the existing lunar collections. Consequently, our focus should shift to mantle-derived magmatic samples for this investigation, though these likely underwent rapid S degassing while passing through the lunar crust and erupting onto the surface⁴⁶. Therefore, it is reasonable to estimate S abundance based on undegassed mantle-derived materials. The olivine-hosted melt-inclusions found in the undegassed Apollo 17 magma (e.g., 74235) exhibit a S abundance of 950 ppm and contain sulfide globules⁴⁷. However, these melts were not saturated with sulfide in their mantle source regions⁴⁷. The experiments on S solubility in lunar melts have revealed that they have a great ability to dissolve S and to achieve sulfide saturation in their mantle source regions they may require S concentration higher than 2000 ppm^17,48. However, the SCSS value varies substantially with factors such as pressure, temperature, and compositional parameters like FeO, SiO₂, and TiO₂^49,50,51. Therefore, it is crucial to constrain the SCSS value before using it in our work. To address this, we refer to a mean SCSS value of 3923 ppm⁴¹, determined from sulfur solubility experiments conducted across different LMO crystallization stages. Modeling studies indicate that SCSS initially decreases as the percentage of LMO crystallization increases (up to 96–98%) but does not exceed 4000 ppm for pure FeS liquid in different LMO crystallization scenarios⁴¹. Therefore, a SCSS range of 2000–4000 ppm is a reasonable estimate for our work, capturing the potential roles of sulfides, the timing of SPA basin formation, and KREEP concentration across different LMO crystallization stages. Consequently, 14% upper mantle material would contribute approximately 280-565 ppm of S to the soils at the Chandrayaan-3 landing site (Table 3). At a mean SCSS range of 3000 ppm, 14% upper mantle material would contribute around 420 ppm of S. This value of S concentration is in good agreement with the excess 200–400 ppm of S measured in the southern high-latitude highland soils by Chandrayaan-3 mission (Fig. 5). Therefore, APXS measurements of excess S likely provides a potential proof of primitive mantle materials in the Moon’s southern high-latitude region, particularly at the highland landing site of the Chandrayaan-3 mission.

The LMO^19,52 crystallization model constrains that: (1) after about 85-90% of LMO crystallization, S begins to exsolve as FeS⁴¹; (2) FeS sulfide saturates the lunar mantle between 96% and 98% of LMO crystallization⁴¹; (3) the SPA basin forms around 4.3 Ga³⁵, a period preceding the crystallization of the KREEP layer, and (4) KREEP is the remaining 0.5% liquid residue after 99.5% LMO crystallization⁶, which forms during 4.22-4.27 Ga³¹. The APXS data show that the SPA basin formation excavated lower crust/upper mantle materials¹⁴, which are low in Na and K concentrations due to the lack of adequately concentrated KREEP-rich materials in the source regions, but high in S, likely sourced from the primitive mantle. The APXS measured abundances of Na and K at the Chandrayaan-3 landing site align with the LMO crystallization model, and are sufficiently robust to support the LMO hypothesis for the Moon’s formation.

Finally, to understand whether the excess 200-400 ppm S measured at the Chandrayaan-3 landing site could be sourced by surface condensation of atmospheric S-carrying components, we discuss the loss mechanism of volatiles and examine the surface temperature of the landing site (see Supplementary Discussions 1 and 2). Condensation of S compound like H₂S is believed to be intense inside the permanently shadow regions (PSRs) due to the presence of a number of ultra-cold trap regions with temperatures as low as 20-30 K⁵³ (Supplementary Discussion 2). A cold-trapping mechanism likely trapped the outgassed H₂S from the mare regions⁵⁴, thereby, making it the second most abundant volatile present at the poles after H₂O⁵⁵ (Supplementary Discussion 2). Previous studies have shown that buried SO₂ ice can remain stable only in areas where mean surface temperatures stay below 110 K⁵⁶, a condition found exclusively in the Moon’s polar areas. At the Chandrayaan-3 landing site around 70° S, the modeled diurnal maximum surface temperature observed at local noon ranges between 250 and 300 K^57,58,59. Such a temperature is much higher than the volatility temperature of S⁶⁰; therefore, the cold-trapping mechanism of surface enrichment of S for geologically long periods is unlikely at the landing site. A different possibility (Supplementary Discussion 1), which hypothesizes ³²S implantation into highland soils, would elevate S abundance while decreasing ³⁴S values⁶¹. However, the APXS experiment aboard the Chandrayaan-3 mission did not perform isotope-level measurements; therefore, this scenario cannot be examined. As a result, we suggest that the enrichment of S by 200-400 ppm at the Chandrayaan-3 landing site may be contributed from the primitive mantle.