Introduction

Noble gases in lunar soils are mainly sourced from solar wind implantation, in situ decay of radionuclides (notably radiogenic 4Her and 40Arr from the decay of U and Th and 40K, respectively), and spallation of the target nuclides irradiated by cosmic rays (for instance, cosmogenic 3Hec, 21Nec, and 38Arc). They provide a critical record of the evolution of the sun and the complex physical processes that have occurred on the lunar surface1,2,3,4,5. The results of previous noble gas analyses of samples from the Apollo and Luna missions showed obvious variations, which were closely related to lunar soil origin, the frequency and intensity of impact events, and the interaction of the solar wind with the lunar surface1,6,7.

China’s Chang’E-5 mission successfully returned 1731 g of lunar soil from the Northern Oceanus Procellarum of the Moon on December 17, 2020. The landing site is located at a relatively high latitude which has not been previously reached and far from the Apollo and Luna landing sites. The site was selected because of the presence of young basaltic rocks as deduced by impact crater counting. The age of the rocks was later confirmed by the precise Pb‒Pb dating, with a crystallization age of 1963 ± 57 Ma reported by Che et al.8 and 2030 ± 4 Ma reported by Li et al.9. The returned soil is among the youngest maria on the moon found so far8,9,10. In this study, we report on the concentrations and isotopic compositions of He, Ne, and Ar of the Chang’E-5 lunar soil (hereafter, CE-5 soil) to better understand the related physical processes that occurred on the new landing site over the past 2 Ga years.

Results and discussion

The Chang’E-5 sample studied in this work was a fine-grained soil sample scooped from the lunar surface (Sample number: CE5C0100YJFM001). The grains included mineral fragments, basaltic clasts, agglutinates, breccias, and glasses (Fig. 1), and the grain size was concentrated in the range of 1–200 μm11. We conducted He, Ne, and Ar analyses on the bulk soil, different grain sizes fractions, and different types of minerals (ilmenite, plagioclase, and pyroxene-olivine) and glass beads. In addition, single grains or groups of a few mineral grains were selected for Micro-CT scanning. In that process, the surface area and volume of each grain were measured in order to examine the correlation of noble gas concentration with the surface area and volume of the grains.

Fig. 1: Images of the bulk sample under the microscope.
figure 1

Ilm ilmenite: black, opaque, euhedral, or subhedral crystals; Pl plagioclase: colorless to white, transparent, euhedral, or subhedral crystals; Aug/Ol pyroxene-olivine: yellow to light brown, transparent, euhedral, or subhedral crystals; Glass beads: round beads with variable colors and uniform transparency.

Full size image

Elemental abundance ratios

The concentrations of all three noble gases (He, Ne, and Ar) in the soil showed a significant negative correlation with the grain size (Supplementary Data 1). The smaller the grain size, the larger the specific surface area. Thus, the concentrations of noble gases in CE-5 soil have an obvious positive correlation with the surface area. As previously reported, the implantation depth of the solar wind ions is only several tens of nanometers, which renders the solar wind gas concentration closely dependent on the grain surface area12,13,14. Therefore, the strong surface-correlation of the noble gas concentrations in the CE-5 soils suggests that the noble gases are sourced predominantly from the implantation of solar wind. The most recent measurement of unfractionated elemental ratios and isotopic compositions of solar wind He, Ne, and Ar were carried out by Heber et al. with targets from the Genesis mission (Table 1)15. However, it is inevitable that the solar wind noble gases trapped in lunar soils become fractionated. The extent of fractionation is not only closely related to the secondary processes (e.g., micrometeorite impacts, soil transportation and burial, and the irradiation of cosmic rays) that occurred on the lunar soils but also to the different retentivity of noble gases in different minerals in the soil.

Table 1 Elemental and isotopic composition of He, Ne, and Ar of solar wind collected by the Genesis mission15.
Full size table

Signer et al. reported that the specific retention efficiency of different minerals for He and Ne is quite different (retention efficiency: ilmenite > olivine-pyroxene > plagioclase), while all minerals retain solar wind Ar similarly well16. This was now also observed in the CE-5 soil we studied, as the 36Ar concentrations of all three minerals were essentially the same, differing by only ~7%. Conversely, the 4He and 20Ne concentrations of different minerals exhibited obvious differences. Expectedly, the ilmenite has the highest 4He/36Ar and 20Ne/36Ar values, demonstrating its best He and Ne retentivity. In contrast, the plagioclase with the lowest 4He/36Ar and 20Ne/36Ar ratios revealed large losses of He and Ne. However, the very similar 4He/36Ar values in both the plagioclase and pyroxene-olivine groups demonstrated that the latter group also had poor He retentivity. In comparison, the higher 20Ne/36Ar value of pyroxene-olivine indicated that it could better retain Ne than plagioclase (Fig. 2). The glass beads group has even lower concentrations of three noble gases than all three mineral types, possibly due to part of the glass beads having been formed from solar wind bearing particles by impact events that resulted in severe losses of these gases.

Fig. 2: Elemental ratios of different minerals and soil with different grain sizes versus their 36Ar concentrations.
figure 2

a 4He/36Ar ratios plotted versus 36Ar concentrations. b 20Ne/36Ar ratios plotted versus 36Ar concentrations.The uncertainties (1σ) of the data are smaller than symbol sizes.

Full size image

The elemental ratios of 4He/36Ar and 20Ne/36Ar could be used to distinguish the origin of the lunar soil, as material from mare terrains is significantly richer in He and Ne than that from highland regions, mainly because the mare soils are richer in ilmenite than highland soils7,16. Signer et al. summarized the 4He/36Ar and 20Ne/36Ar values of the soils from highland and mare samples. The 4He/36Ar and 20Ne/36Ar values of the soils from the highland are seldom higher than 150 and 3, respectively, whereas those of soils from mare are significantly higher, exceeding 200 and 4, respectively16. For the CE-5 soils analyzed in this study, the mean 4He/36Ar and 20Ne/36Ar ratios are 661.1 and 7.1, respectively (Fig. 3 and Supplementary Data 1), clearly showing that the soil has typical mare terrain characteristics. In addition, the 4He/36Ar and 20Ne/36Ar values obtained in this study were higher than those of most mare soils from the Apollo and Luna missions. It is reported that the Ti-rich samples appear to have better He and Ne retentivity; hence, they tend to have higher 4He/36Ar and 20Ne/36Ar ratios relative to Ti-poor samples7,17,18,19,20,21. Taking the TiO2 content into consideration, we compared the 4He/36Ar and 20Ne/36Ar ratios of the CE-5 soil with those of the soil samples retrieved by Apollo and Luna missions. To avoid the grain size effect, the data we used for the comparison were generally from samples with approximately the same grain size. Even considering the effect of the TiO2 content of the local basalt, both the 4He/36Ar and 20Ne/36Ar ratios of CE-5 lunar soil show relatively high values (Fig. 4), indicating that the solar wind in CE-5 soil was less fractionated than that in most other lunar soils. The high 4He/36Ar and 20Ne/36Ar values of CE-5 soil indicate that the soil suffered only relatively weak secondary processes. Such processes could be caused by a variety of factors, including micrometeorite impacts, cosmic rays irradiation, soil burial and compaction, mixing with hot ejecta from craters, and so on. They could lead to surface erosion of the soil grains, which in turn result in the partial loss of solar wind gas from the soil. It is natural to recognize that the lighter noble gas (e.g., He) is more sensitive to such secondary processes than the heavier noble gas (e.g., Ar). Therefore, the weaker these secondary processes are, the higher the 4He/36Ar and 20Ne/36Ar values of the soil are likely to be. Thus, the high 4He/36Ar and 20Ne/36Ar values of CE-5 soil may indicate that the secondary processes it endured were not strong.

Fig. 3: Elemental abundance ratios of the different lunar soils.
figure 3

a 4He/36Ar ratios plotted versus 4He/20Ne ratios. b 20Ne/36Ar ratios plotted versus 4He/20Ne ratios.Unfractionated solar wind 4He/36Ar and 20Ne/36Ar ratios were shown in Table 1. It could be seen that 4He/36Ar ratio and 20Ne/36Ar ratios in mare soils are less fractionated than those in highland soils. Furthermore, CE-5 soils seem to preserve more light noble gas compared with most mare soils returned by other missions. Analytical uncertainties of 4He/36Ar ratio and 20Ne/36Ar ratio of CE-5 soil are at 1σ. The elemental abundance ratios of Apollo and Luna soils are from refs. 7,16,17,18,19,27,50,51,52,53,54,55,56.

Full size image
Fig. 4: Relationship between TiO2 contents and 4He/36Ar and 20Ne/36Ar ratios of the lunar soils.
figure 4

The number in the brackets indicates the soil grain size(the unit of the grain size is μm). The data of the Apollo and the Luna soils are from ref. 17,50,51,52,53,54,55,56.

Full size image

Isotopic compositions

The measured values of (3He/4He)m of the bulk sample ranged from (3.91 ± 0.06) × 10−4 to (4.18 ± 0.06) × 10−4 [mean value: (4.02 ± 0.11) × 10−4], (20Ne/22Ne)m from 12.57 (±0.16) to 13.08 (±0.16) [mean value:12.88 (±0.22)], and (21Ne/22Ne)m from 0.038(±0.001) to 0.042 (±0.001)[mean value: 0.040 ± (0.002)] (Supplementary Data 1). The 20Ne/22Ne and 21Ne/22Ne values prove that the noble gas in CE-5 soil is solar wind dominated as they all plot in the area between the fractionated solar wind and solar wind in the neon three-isotope plot (Supplementary Fig. 2). The isotopic composition differences of the bulk samples may be a result of inhomogeneity due to the very small sample size for each analysis. The measured values do not accurately represent the trapped noble gas isotopic composition because the samples also contain a small fraction of in situ radiogenic (4Her) and cosmogenic nuclides (3Hec,20Nec, 21Nec, and 22Nec) that should be properly subtracted. The grain size dependence of the (3He/4He)m and (21Ne/22Ne)m ratios allows for calculating the concentrations of the volume correlated 3Hec and 21Nec17,19,22. For example, one could plot the (21Ne/22Ne)m versus 1/21Nem of different grain size fractions to calculate the (21Ne/22Ne)tr from the ordinate intercept of the line and 21Nec from the slope of the least square fit. Unfortunately, however, our data did not show the correlation between the grain size and isotopic composition as described in the literature, mostly because we could not prepare well-defined homogeneous grain size fractions due to the very small sample size available (Supplementary Fig. 3). Therefore, we attempted to calculate the cosmic-ray exposure (CRE) age to evaluate the contribution of the cosmogenic nuclides 3Hec, 20Nec, and 22Nec. In principle, 3Hec,21Nec, and 38Arc all could be used to calculate the CRE age. However, the 3Hec does not work because we couldn’t correct the measured 3He for the dominant solar wind 3Hesw.The 38Arc is also difficult to calculate as the measured (38Ar/36Ar)m value (Supplementary Data 1) is too close to the solar wind (38Ar/36Ar)sw value (Table 1), which means that the amount of 38Arc is too small compared with the solar wind 38Arsw. In contrast, the (21Ne/22Ne)m values are different enough from the solar wind value to reliably calculate 21Nec. Assuming (21Ne/22Ne)tr = 0.03125 (±0.00101) and (21Ne/22Ne)c = 0.91 (±0.08) reported by Signer et al.16, we calculated the 21Nec exposure age by adopting the production rates reported by Leya et al.23 for 2π irradiation geometry. Since these are related to the soil depth, here we adopted the maximum production rate assuming the depth of the sample during irradiation near the lunar surface was 30 g/cm2. The exposure ages of bulk samples calculated in this way ranged from 363 (±45) to 566 (±86) Ma, with an average age of 432 (±116) Ma (Supplementary Data 2). As discussed above, such variable 21Nec exposure ages of the bulk soil also result from the inhomogeneity of the samples due to the extremely small sample size we used. What should be noted here is that, as discussed above, considerable Ne loss occurred due to diffusion; therefore, the 21Ne exposure ages should be regarded as a lower limit to the true exposure time to cosmic rays radiation. Based on the 21Nec values, the 3Hec concentration of the soils was calculated according to the production rate given by Reedy24. However, a large fraction of 3Hec may have been lost via diffusion on the lunar surface25,26. We were unable to clarify this loss as it is closely related to the exposure history of the soil, and the physical processes that occurred on the soil like the flux and intensity of the micrometeorite impacts. Somewhat arbitrarily assuming a 50% loss of 3Hec by diffusion, the retained 3Hec amount would account for 4.8‒7.7% of the amount of measured 3He in CE-5 soil. As for the radiogenic 4Her, we calculated its concentration through the U and Th contents reported by Li et al.11 and the rock age of ~2 Ga determined by Pb‒Pb dating8,9.Also adopting a 50% loss of 4Her, the retained 4Her fraction would not exceed 1% of the measured 4He. With these corrections, the trapped (3He/4He)tr ratios of the bulk samples range between(3.73 ± 0.06) × 10−4 and (3.89 ± 0.06) × 10−4, which seem similar to values reported for Apollo 11 and Apollo 15 soils17,27,28. The isotopic compositions of He and Ne of different minerals also exhibited significant variations. The (3He/4He)m value in the pyroxene-olivine group was relatively high [(4.66 ± 0.08) × 10−4)], whereas, in the ilmenite and plagioclase groups, the (3He/4He)m values were relatively low [(4.13 ± 0.06) × 10−4) and (4.12 ± 0.06) × 10−4), respectively]. The glass beads group had the lowest (3He/4He)m value, indicating that severe fractionation occurred during the mineral remelting process to form the beads. We consider that the (3He/4He)m value of the pyroxene-olivine group may have been significantly higher than that in the ilmenite group due to a relatively large contribution of the cosmogenic nuclide 3Hec. Assuming that the ilmenite and olivine-pyroxene groups had the same exposure age, we adopted a mean exposure time of bulk soils with 432 Ma (Supplementary Data 2) to estimate the concentrations of the 3Hec in different mineral groups according to their different production rates (Supplementary Data 3).The calculated concentration of 3Hec in the pyroxene-olivine group was only slightly higher than that in the ilmenite group because-unlike for 21Nec and 38Arc-elemental production rates of 3Hec only weakly depend on the chemical composition of a mineral. Therefore, the slight difference in 3Hec concentration between these two mineral groups would be insufficient to result in a significant difference in their (3He/4He)m values, as long as the concentrations of trapped He in both samples would be comparable. However, as discussed above, the concentration of solar wind 3Hesw trapped in ilmenite was significantly higher than that in pyroxene-olivine, leading to a lower proportion of 3Hec to 3Hesw in ilmenite relative to that of pyroxene-olivine. In other words, the contribution of 3Hec in pyroxene-olivine must be obviously higher than that in ilmenite. This must be the dominant reason for the higher (3He/4He)m value of the pyroxene-olivine group. After the 3Hec subtraction (also assuming a 50% loss of diffusion), we obtained (3He/4He)tr values of (4.07 ± 0.06) × 10−4 for the ilmenite group and (4.17 ± 0.07) × 10−4 for the pyroxene-olivine group, showing that the (3He/4He)tr values of the two groups were more similar than the (3He/4He)m values. To obtain the (20Ne/22Ne)tr value, we subtracted 22Nec according to the (21Ne/22Ne)c = 0.91 (±0.08)16, the corrected (20Ne/22Ne)tr values of the bulk samples ranged from 12.71 (±0.16) to 13.17 (±0.16). The ilmenite group, which exhibited the best Ne retentivity, showed a higher (20Ne/22Ne)tr value of 13.64 (±0.17) relative to other mineral groups (Supplementary Data 4). The relatively low value in the glass beads group, as in the case of (3He/4He)tr, indicates a pronounced isotopic mass fractionation effect due to the significant gas loss during the melting process.

The trapped (40Ar/36Ar)tr value is a semi-quantitative measure of the solar wind antiquity of lunar soils20,29,30,31,32,33. As the solar wind is essentially free of 40Ar, the trapped 40Ar in the soil is basically entirely derived from the lunar exosphere by ionization and acceleration by the electromagnetic field induced by the solar wind and final implantation into the soil, while the trapped 36Ar derives from the solar wind. The 40Ar in the exosphere, which is mainly sourced from the radioactive decay of the 40K in the lunar interior, has decreased over time30,31,32. Therefore, the smaller (40Ar/36Ar)tr values are, the younger the solar wind and the soils preserved. Eugster et al. reported that the (40Ar/36Ar)tr values decrease from about 13 some 3.7 Ga ago, to about 0.5 for samples irradiated during the last several 10 Ma ago31. For CE-5 soil, the most reliable (40Ar/36Ar)tr may come from the potassium-poor pyroxene-olivine group because the contribution of radiogenic 40Ar* produced by the 40K decay process is very small. The pyroxene-olivine group had a (40Ar/36Ar)m value of 0.69 (±0.01). The single pyroxene-olivine grain returned a (40Ar/36Ar)m value of 0.56 (±0.01), which must be very close to the trapped (40Ar/36Ar)tr value of the CE-5 soil, as the contribution of the radiogenic 40Ar was negligible. Therefore, the CE-5 soil was likely irradiated relatively late in lunar history, perhaps as late as 10 Ma ago, but certainly later than 1 Ga ago. The relatively higher (40Ar/36Ar)m value of the pyroxene-olivine group compared to that of the pyroxene-olivine single grain may have resulted from K-bearing impurities in the pyroxene-olivine group (Supplementary Data 5). The glass beads and plagioclase mineral groups showed relatively higher (40Ar/36Ar)m values relative to the pyroxene-olivine group, which must be due to the radiogenic 40Arr contribution of the samples. Higher (40Ar/36Ar)m values were also found in the two single (several) plagioclase grain(s) groups. Regarding the difference in the 40Ar/36Ar values of these two groups, we believe it may result from the inhomogeneity in K between the different grains. The average (40Ar/36Ar)m value of the analyzed CE-5 bulk soil was 0.87 (±0.09) (Supplementary Data 1). After subtraction of radiogenic 40Arr calculated through the K contents reported by Li et al.11 and the rock age of ~2 Ga determined by Pb‒Pb dating8,9, we corrected the (40Ar/36Ar)tr of the bulk sample with the value of 0.65 (±0.05) that was expectedly consistent with the (40Ar/36Ar)tr value of the pyroxene-olivine group, which is free of the contribution of radiogenic 40Ar* (Fig. 5).

Fig. 5: (40Ar/36Ar)m ratios plotted versus the 36Ar concentrations in CE-5 soil.
figure 5

a (40Ar/36Ar)m ratios in different minerals and soil with different grain sizes. Different minerals have similar 36Ar concentrations, and the grain size negatively correlates with the 36Ar concentration. b (40Ar/36Ar)m ratios in groups with single (several) grain(s). Note that each group has two symbols representing the grain weight, which is related to the 36Ar concentration and is calculated by the two end-densities of the mineral, respectively. The 40Ar/36Ar ratios of the single (several) pyroxene/olivine grain(s) are close to the corrected (40Ar/36Ar)tr ratios of the bulk soil, as they have negligible radiogenic 40Ar because of the extremely low K content. Analytical uncertainties of the 40Ar/36Ar ratio of the bulk sample are at 1σ. The uncertainties of the 36Ar concentration of the single (several) pyroxene/olivine grain(s) mainly resulted from the uncertainty of the mineral density and uncertainty of grain volume determination (see details in the “Methods” section).

Full size image

The (3He/4He)tr ratio inversely correlates with the (40Ar/36Ar)tr value (Fig. 6). As discussed above, the lower the (40Ar/36Ar)tr value, the more recent the solar wind captured by the soil. Based on this, it has been argued that the 3He/4He of the solar wind may have increased over time34,35. This apparent secular change in (3He/4He)sw was explained by the mixing of the 3He produced by the incomplete burning of H in the outer convective zone of the sun, which is the solar wind source36,37. However, Heber et al. measured (3He/4He)tr using the high-resolution closed system stepwise etching method, which could effectively extract the solar wind He and Ne through the precise control of the erosion depth of the soils38,39,40,41, and found that not only the apparent solar wind (3He/4He)sw increases with the antiquity of the solar wind, but the (20Ne/22Ne)sw appears to show a similar increase41. The magnitude of the increase in 20Ne/22Ne was ~5%/Ga, which is difficult to explain. Therefore, it is suggested that the apparent increase in (3He/4He)tr of the soils with young antiquity age may be caused by secondary processes of the soils on the lunar surface2,39,40,41. As discussed above, the secondary processes could result in the partial removal of the solar wind component that resides at the outmost layer of the grains. It was shown that the trapped solar noble gases become isotopically heavier in the deeper grain layers41. Thus, the stronger the secondary processes the soil endured, the more severe the isotopic fractionation of the trapped solar wind noble gas was. We compared the (3He/4He)tr and (20Ne/22Ne)tr values in CE-5 soils with those in other lunar soils (Figs. 6 and 7). Both the (3He/4He)tr and (20Ne/22Ne)tr values of the CE-5 soil are less fractionated than in other lunar soils, i.e. the CE-5 values plot closer to the unfractionated solar wind values. Especially the ilmenite, shows a (20Ne/22Ne)tr ratio very close to that of the unfractionated solar wind, probably the least fractionated value reported for a lunar soil so far. This remarkable finding indicated that the secondary processes the CE-5 soils have undergone were weak. This observation is in agreement with the relatively good retention of solar wind 4He and 20Ne, indicated by the relatively high 4He/36Ar and 20Ne/36Ar, which also indicated that the trapped noble gas record in the CE-5 soil is less disturbed by secondary processes than that of other lunar soils. As discussed above, multiple reasons could lead to such secondary processes. Further work is required to better understand the fractionation processes in detail, but the CE-5 soil seems to be indeed a particularly valid extension to the lunar sample suite.

Fig. 6: Correlation of the (3He/4He)tr ratios versus (40Ar/36Ar)tr ratios of the lunar samples.
figure 6

The latter is an antiquity indicator of the solar wind, as introduced above. The (3He/4He)tr ratios of the CE-5 bulk soil and ilmenite both fit the correlation. The uncertainties of (3He/4He)tr ratios of CE-5 samples were estimated on the assumption that the cosmogenic 3He and radiogenic 4He with different loss proportions (the symbol represents 50% loss of 3Hec). The data of Apollo soils are from previous reports16,17,19,22,28,31,53,54,57.

Full size image
Fig. 7: Comparison of the isotopic composition of Ne in the lunar soil.
figure 7

The (20Ne/22Ne)tr ratios of the CE-5 bulk soil seem less fractionated than those of the Apollo soils, particularly for the (20Ne/22Ne)tr ratio in ilmenite, which is very close to that of the modern solar wind, indicating that the secondary process endured by the CE-5 soils was not strong. The relatively higher 21Ne/22Ne ratios of the CE-5 soil than those of other lunar soils indicate that the sample contains a certain proportion of cosmogenic 21Ne, which may result from the longer cosmic-ray irradiation time than those of the Apollo samples. The uncertainties of (20Ne/22Ne)tr ratios in CE-5 samples are at 1σ. The data of Apollo soils are from previous reports16,17,19,22,28,31,53,54,57.

Full size image

Integrated exposure time

The integrated exposure time refers to the time that the soil is directly exposed to the solar wind on the uppermost lunar surface16. We used the concentration of 36Ar, which is better retained than 4He and 20Ne, to estimate the integrated exposure time of the soil. Geiss et al. measured the solar wind flux using a metallic foil at the lunar surface and obtained a 36Ar flux of the solar wind of 600 atoms cm−2 s−142,43, based on which the effective acquisition rate of the outermost regolith layer could be calculated as ~1.1 × 10−10 cm3 · STP cm−2 yr−116. Therefore, if the surface area and the 36Ar amount could be measured for each mineral grain, it would be possible to estimate the integrated mean residence time of the grain in the uppermost layer of the regolith. Signer et al. calculated the integrated exposed times of the Apollo minerals with 150‒200 μm grain size, ~2000 (soil 60051) to 13,000 years (soil 15021)16. We conducted Micro-CT scanning for each grain of four groups (two groups with single grain and the other two groups with several grains) to determine the surface area of each group, after which the grains of each group were totally melted to determine the 36Ar amount. The results showed that the 36Ar amounts of all four groups are linearly correlated with their surface areas, and the 36Ar amount per grain surface area in all four groups was similar (Fig. 8, Supplementary Data 6). These findings indicate that the grains of each group may have endured the same exposure-burial history. According to the implanted rate of the 36Ar mentioned above, we calculated the integrated exposure times of the four groups to be between 4778 (±57)and 6168 (±189) years on the assumption that no 36Ar was lost. Compared with the Apollo soils, it appeared that the integrated exposure time of CE-5 soil was in the middle range. However, considering that the basalt of the Chang’E-5 landing site is relatively young (~2 Ga) compared to that of the Apollo landing site (3.0‒4.5 Ga), and assuming that the CE-5 soil is derived from the local basalt, the integrated exposure time of the minerals of CE-5 must be close to the upper limit of those of the Apollo soils if the turnover frequency of the lunar sample is constant during the lunar history. Signer et al. suggested that the relatively short integrated exposure time of the sample may indicate that the soil was admixed by grains sputtered from the impact of nearby rocks, as the grains formed by the impact of the rocks may not have been previously exposed to the solar wind16. From this point, the relatively long integrated exposure time of the CE-5 minerals may indicate that the soil of the CE-5 landing site was not admixed by sputtered materials from nearby rocks seriously.

Fig. 8: 36Ar amount of the grain(s) plotted versus its(their) surface area.
figure 8

S/V represents the ratio of surface area to volume of the grain(s). The 36Ar amount correlates well with the surface area but shows no relationship with the volume.

Full size image

Methods

Sample preparation

Sample preparation work was carried out in an ultra-clean room to avoid contamination of the surroundings. Firstly, three aliquots sample with very small amounts (sample weight: ~0.1 mg of each aliquot) were taken from the sample bottle, then transferred to three small copper cups (wall thickness: 0.15 mm, inner diameter: 4 mm, height: 5 mm) (Supplementary Fig. 1). A sample in each copper cup was completely melted in one step by the laser for He, Ne, and Ar analyses.

Secondly, another two aliquots (sample weight: ~1 mg for each aliquot) were taken from the vials for mineral separation and grain size sorting, respectively. According to their different colors and shapes, three mineral groups with >80% of target minerals (ilmenite, pyroxene-olivine, and plagioclase) and one group with 100% glass beads were identified. Three additional groups of different grain sizes were selected (då 50 μm, 50 μmå då 20 μm, and d <20 μm) according to the length of the longest side of the grain from another aliquot of the CE-5 bulk sample. Both the mineral separation and grain size selection were carried out under a binocular microscope using hand-picking method. Except for the glass beads group, each group was randomly divided into two portions. One portion was cast in epoxy resin as standard 25-mm mounts for mineral identification and grain size analysis. The other portion was transferred to a small copper cup (Supplementary Fig. 1), which was used to weigh the sample mass by a precise balance and then also as a crucible placed into the laser chamber for the noble gas extraction. No statistical work on mineralogy was conducted on glass beads due to the extremely small amount.

Lastly, four additional groups of minerals with single or several grains were selected from the CE-5 bulk soil, two groups being plagioclase grains and two olivine-pyroxene grains. One plagioclase group consisted of a single large grain (grain size: ~200 μm), while the other represented several small grains (grain sizes: 50–100 μm). Similarly, one of the olivine-pyroxene groups consisted of one single grain (grain size: ~200 μm) and the other group included several smaller grains (grain sizes: 50–100 μm). All grains were adhered to glue tape for Micro-CT scanning to measure the surface area and volume. After scanning, the mineral grains were removed from the tape and transferred to the copper cup for gas extraction by laser melting method.

Mineral and grain size characterization

The quantitative mineral analysis and particle size statistics of each group in this study were determined using a Zeiss Sigma 300 scanning electron microscope, equipped with a Bruker X Flash 6130 energy dispersive spectrometer with an advanced mineral identification and characterization system. It operated at 15 kV, and a resolution of 0.11‒0.93 μm/pixel. The maximum length calculation method was used to determine the particle size, and the particle size distribution were calculated by the weight of the particles in different size range as a percentage of the total weight of all particles. For the estimate of the bulk soil composition, modal abundances of the different mineral and glass phases were converted to mass abundance using the following densities:2.73 g/cm3 for anorthite, 3.40 g/cm3 for clinopyroxene, 3.7 g/cm3 for olivine, 4.39 g/cm3 for fayalite, 3.24 g/cm3 for glass, 4.72 g/cm3 for ilmenite, 4.78 g/cm3 for ulvöspinel, 4.75 g/cm3 for troilite, 2.62 g/cm3 for silica, 3.15 g/cm3 for apatite, 5.75 g/cm3 for baddeleyite, 4.79 g/cm3 for chromite, 7.6 g/cm3 for iron, 4.65 g/cm3 for zircon, and 2.52 g/cm3 for K-bearing minerals. On this basis, accurate identification of minerals was carried out using electron probe data. The chemical composition of different lunar minerals and glass was determined by wavelength dispersive spectroscopy using a JXA-8100 electron microprobe operating at 20 kV, 10 nA, with a 1–2 µm beam diameter.

Measurement of the surface area and volume of single grain

This study used an Xradia 610 Versa micro-CT scanner to obtain the surface area and volume of the mineral (plagioclase, pyroxene, and olivine) grains. The voltage of the X-ray source was 120 kV and the 20 X detector was applied to obtain a voxel size as fine as 0.25 μm. Such a small voxel size guarantees that even the smallest particle is represented by at least thousands of voxels to depict the particle morphology. This study used a machine-learning algorithm “ilastik” (the official website ilastik.org provides further details) to conduct mineral segmentation from raw CT images that considers both voxel intensity and mineral edge detection at the same time44.

The volume of each particle was then obtained by counting the voxel number of a particle with known unit voxel volume. 3D erode (peels off a layer of voxels from the 3D volume) and dilate (adds a layer of voxels to the 3D volume) functions in Fiji were utilized to estimate the surface area of the particles45. The surface area of each particle was determined by dividing the total volume of eroded and dilated voxels by the thickness of the surface layer. We consider the maximum uncertainty in volume calculation to occur when the whole surface layer was mistakenly classified, therefore, the uncertainty value was calculated by dividing the eroded or dilated voxels over the total volume. The number of voxels in the eroded surface layer is typically slightly less than the dilated counterpart, and the difference between the number of eroded and dilated voxels divided by their average was considered as the uncertainty in particle surface area calculation. The image intensity is correlated to the mineral composition and density46,47. Minerals with higher density and higher average atomic number are brighter in CT images. Sometimes, the boundary between different minerals is not sharp and there could be a gradual transition between different minerals, and even a single mineral grain could show variation in image brightness as its composition may change gradually. Through image segmentation, we could determine the number of mineral types in a single grain, to check whether the particle was pure or not. The plagioclase grains are obviously dark due to relatively low density and low average atomic number. However, it is more difficult to clearly distinguish between pyroxene and olivine due to their similar density. Even under the binocular, these two minerals show the same yellow color. Therefore, we put pyroxene and olivine into one group for discussion.

He, Ne, and Ar analyses

Sample weighing

To minimize sample loss, for weighing we transferred the sample directly to the copper cup in which the sample was later melted. We first weighed the empty crucible on a high-precision balance (resolution:0.001 mg), then added the sample and weighed the crucible with the sample. Each sample was weighed four times to ensure the accuracy of the weights. For single grains or assemblages of a few grains, this method would not have been precise enough, therefore, the sample masses were calculated by the volume of the grain determined by Micro-CT, and the respective mineral density. The uncertainty of the weight determination of these samples was mainly governed by the uncertainty of the mineral density. For the pyroxene-olivine minerals, it was difficult to separate the pyroxene and olivine under the microscope; hence, we used the densities of 3.40 g/cm3 for pyroxene and 4.07 g/cm3 for olivine to determine the mass uncertainty. For plagioclase, we used a density between 2.56 g/cm3 and 2.69 g/cm3.

Gas extraction

A CO2 laser (λ = 10.6 μm) was used to extract the gas by total-fusion. A small copper cup containing the weighed sample was placed into a copper holder in the laser chamber. After evacuating the chamber to high vacuum, it was baked at 100 °C for 72 h to fully remove the adsorbed atmospheric gas. Finally, the sample was melted with a 2-mm-diameter laser beam and ~8 W of power. After each extraction, the power was increased to 10 W to check that the sample gas had been thoroughly extracted upon melting.

Gas analysis

The extracted noble gases were purified using three SAES ST-101 getters, two getters at 450 °C and another one at room temperature. He, Ne, and Ar were analyzed using a Helix SFT (Thermo Fisher Scientific) split flight tube noble gas mass spectrometer in static mode. After purification, the heavier noble gases (Ar, Kr, and Xe) were trapped by a charcoal finger immersed in liquid nitrogen, leaving the pure He and Ne to be let into the mass spectrometer for analysis. The two He isotopes were simultaneously measured (4He on the Faraday cup and 3He on the multiplier). The three Ne isotopes were detected in peak-jumping mode (20Ne and 22Ne on the Faraday cup and 21Ne on the multiplier). A charcoal finger at 77 K near the ion source was used to freeze the Ar and CO2 during measurements. Even so, the 40Ar+ and CO2+ were measured during each analysis for correction of interferences of the 40Ar++ and CO2++ to the 20Ne and 22Ne. However, the 40Ar+ signal was seldom beyond 1% of the 20Ne signal; thus, the contribution of 40Ar++ to 20Ne+ was negligible. The CO2+ signal was not more than 10% of that of 22Ne+. Considering the correction factor of the CO2++/CO2+ (1.3 ± 0.3%)48, the contribution of the CO2++ to 22Ne+ was ~0.1%, thus it could also be neglected. After the He and Ne analysis, the charcoal finger was heated to release the heavier noble gases for the Ar analysis. The three Ar isotopes were also measured in peak-jumping mode (40Ar and 36Ar on the Faraday cup and 38Ar on the multiplier).

Data correction

The standard for the correction of the measured 3He/4He value of lunar soil was an artificial gas mixture with a 3He/4He value of ~5.00E−04, more than two orders of magnitude higher than the atmospheric value of ~1.40E-06.To ascertain its accurate He isotopic composition, the standard gas was distributed to several associated laboratories at ETH (Zurich, Switzerland), INGV (Palermo, Italy), and IGGCAS (Beijing, China) for comparison. The 3He/4He ratios of the three laboratories closely agree with each other (Supplementary Data 7) and we adopt the average value of 4.95(±0.07) × 10−4. The Ne and Ar isotopic compositions were corrected for mass discrimination with an air standard. The He and Ne concentrations of the samples were calculated by comparing the corresponding signal intensity with those of known amounts of the air standard. Ar concentrations of the samples were calibrated with the 40Ar-39Ar dating sanidine standard YBCs. Splits of this standard, with a radiogenic 40Ar* abundance of 1.16(±0.01) × 10−5cc·STP/g(calculated by its K content and crystallization age proposed by Wang et al.49), were precisely weighed and then melted to release 40Ar*.

The blanks were measured according to the same procedure used for the sample analysis. For the laser method, the mean blank levels were 1.01(±0.05) × 10−10 cc. STP in 4He, 1.98(±0.08) × 10−12 cc. STP in 20Ne, and 3.59(±0.09) × 10−11 cc. STP in 40Ar with the laser turned off. All blanks were measured before individual sample analysis and subtracted for sample correction.