Landing site description and Yutu rover operations

The location of CE-3 from a global to a ground view is shown in Fig. 1. CE-3 landed on the rim of a young crater (∼27–80 Myr old6), initially informally named Purple Palace4 and now formally named Zi Wei (Fig. 1c). The diameter of the crater is ∼450 m, which would have excavated ∼40–50 m beneath the surface. The ejecta of the impact should cover the entire CE-3 landing site and the region explored by the Yutu rover, evidenced by the blocky surface seen by the landing camera (Fig. 1d) and big boulders encountered by Yutu during its traverse (Fig. 1d,e). The Panoramic Camera imaged two types of rocks; one is a mainly light-toned and coarse-grained rock and the other is a relatively darker, fine-grained rock (Supplementary Figs 1 and 2, Supplementary Note 1). During 32 days of surface operations, Yutu travelled 114 m in this region and made four sets of in situ and stand-off measurements at four locations (Fig. 1d)3,4,6. The APXS and VNIS aboard the Yutu rover acquired compositional and spectral measurements at four locations (CE3-0005, -0006, -0007 and -0008), as shown in Fig. 1d. A detailed description of the instruments, measurements and data processing are given in the Methods section and refs 12, 13, 14, 15, 16.

Figure 1: Location of the Chang'e-3 landing site. (a) Chang'e-1 CCD image with boundaries of typical mare basalt units7. (b) Chang'e-2 CCD image and (c) LROC NAC image (LROC NAC M1142582775R). (d) The traverse map of the Yutu rover and the locations of APXS and VNIS measurements. (e) Panoramic view of the ‘Zi Wei’ crater by the Panoramic Camera on the Yutu rover at the CE3-0008 site. Full size image

Chemical compositions and normative mineralogies from APXS

APXS spectra show peaks of Mg, Al, Si, K, Ca, Ti, Cr, Fe, Ni, Sr, Zr and Y from the CE-3 soils (Fig. 2a). We used the peak-area ratio of measured samples and the calibration target (Supplementary Table 1) to derive the chemical compositions of three measured soils (CE3-0006_2, -0006_3 and -0008, Table 1). In general, the major element concentrations of the three soils at the CE-3 landing site are similar to each other and represent a distinctive composition (Table 1, Fig. 2b–d). They are characterized by low SiO 2 (∼41.2 wt.%), very high FeO (∼22.8 wt.%), high CaO (∼12.1 wt.%), intermediate TiO 2 (∼5.0 wt.%) and modest Al 2 O 3 (∼9.7 wt.%).

Figure 2: X-ray spectrum and chemical compositions of Chang'e-3 soils from APXS. (a) APXS spectrum CE3-0006_2 overlain on the calibration spectrum. Comparison of Chang’e-3 site surface soil samples with Apollo and Luna samples2,17 in (b) FeO versus TiO 2 , (c) FeO versus MgO and (d) FeO versus CaO. Full size image

Table 1: Compositional data in weight percent and results of CIPW norm of Chang'e-3 soils from Yutu APXS after calibration. Full size table

When compared with Apollo and Luna soils and basaltic rocks (Fig. 2b–d)2,17, the TiO 2 versus FeO relation of the CE-3 soils bears some similarity with Apollo 12 ilmenite basalts, but CE-3 soils have higher FeO and TiO 2 (Fig. 2b). The CE-3 soils have MgO concentrations in the range 6.3–11.0 wt.% (Table 1) with derived Mg# (=Mg/(Mg+Fe) × 100)<50 (Fig. 2c) at the low end, but a higher CaO compared with other mare samples (Fig. 2d), and deviating from the KREEP—feldspathic highlands—mare compositional triangle based on the returned lunar samples. These compositional features suggest that the CE-3 soils differ significantly from other known lunar basaltic materials.

On the basis of chemical composition (Table 1), we calculated the abundances of normative minerals of CE-3 soils using a CIPW (Cross, Iddings, Pirsson and Washington) norm. The major CIPW norm results are summarized in Table 1 and the detailed results are shown in Supplementary Tables 2,3. For CE3-0006_2 and CE3-0006_3 (sampling sites ∼10 cm apart), we calculated the mean value as ‘Mean_0006.’ For the norms summarized in Table 1, we combined the high-Ca pyroxene components as diopside (Di) and the Ca-poor components as hypersthene (Hy). Given the analytical uncertainties associated with the APXS data (Supplementary Table 4), the main difference is in MgO, which is significantly higher in CE3-0008 (11 wt.%) compared with CE3-0006 (6.7 wt.%). The difference in MgO translates to a difference in the relative abundance of olivine and pyroxene and in the ratio of Di to Hy (Table 1). The higher MgO concentration of the CE3-0008 soil results in a higher Mg# (46) compared with CE3-0006 (34) and a higher Fo (=Mg/(Mg+Fe) × 100) in olivine, that is, ∼Fo 51 for the CE3-0008 soil and ∼Fo 40 for the CE3-0006 soil. The CE3-0006 soil is also richer in the high-Fe endmember for both Di and Hy (Table 1) as a result of the difference in MgO. Considering analytical uncertainties for Al, Si, Ca, Fe and Ti, normative abundances of plagioclase and ilmenite are the same in -0006 and -0008, within analytical uncertainties.

In Table 1, the ‘Means_all’ column shows the average chemical composition and normative mineralogy summary of the CE-3 landing site soils. The soils have a high percentage of normative pyroxene (∼42 wt.%), with most being high-Ca pyroxene, Di (29 wt.%), that is, about two times the Hy (13 wt.%). The normative feldspar content (27 wt.%) is within the range of many lunar basaltic samples. The normative olivine content (20 wt.%, corresponding to 17 vol.%) of CE-3 is at the high end of the range for known lunar basalts (for example, Apollo 12 olivine basalt has ∼20 vol.% olivine2). In the CE-3 soils, olivine is Fe-rich with relatively low average Fo content (∼43). The normative ilmenite contents of the three CE-3 soils are similar, averaging ∼9 wt.%. The average Mg# of the soils is ∼38, indicating the exceptionally ferroan character of source rocks that make up the local surface soils.

Mineral chemistry and mineral modes based on VNIR spectra

The visible-NIR (near-infrared) spectra (Fig. 3a,b) of four VNIS observations show characteristic 1 and 2 μm absorption features owing to the electron transfer of Fe2+ in the M1 and M2 sites of lunar mafic silicates18. The spectra (Fig. 3a) have obvious absorption features and relatively flat profiles, indicating a low degree of space weathering, consistent with the fact that the CE-3 landing site sits on a relatively young Eratosthenian lava flow and the fresh ejecta of the young and fresh Zi Wei crater.

Figure 3: Visible-NIR spectral properties and mineral chemistry of Chang'e-3 soils from VNIS. (a) Combined VNIS spectra (450–2,400 nm) from sites 0005, 0006, 0007 and 0008. The inset image is from site CE3-0006 of the VNIS (450–950 nm) image mode at 750 nm. The dashed circle indicates the region measured by the VNIS-point spectral mode (900–2,400 nm). (b) VNIS spectra after continuum removal. (c) Pyroxene VNIS peak positions of the CE-3 soils overlain on experimental results from Adams27 and Cloutis and Gaffey28. (d) Fo values of olivine in four CE-3 soils derived from VNIS spectra, overlain on calibration lines (Sunshine and Pieters20). Full size image

To estimate the average composition of minerals contributing to the spectra, we apply the modified Gaussian model (MGM)19,20 to deconvolve the spectral bands. We find that the spectra from CE3-0005 and CE3-0008 sites have wide and strong 1 μm absorption bands but shallow 2 μm band depths (Fig. 3a,b), thus they should have a higher 1–2μm band area ratio (BAR), which implies the presence of a significant amount of olivine in the soils of these two sites15,21,22,23. The absorption components of all four continuum-removed spectra (Fig. 3b) are calculated using MGM, as mixtures of three endmembers, high-Ca pyroxene (HCP), low-Ca pyroxene (LCP) and olivine (Supplementary Note 5, Table 2). This combination is the most complicated for this type of spectral deconvolution24,25,26. The results of the MGM deconvolution are shown in Table 2.

Table 2: MGM results of the four VNIS spectra. Full size table

Extensive laboratory studies of terrestrial and synthetic pyroxenes provide the basis to correlate the 1- and 2-μm band positions with their chemical compositions27,28,29,30. We plot the central positions of deconvolved 1 and 2 μm bands from HCP and LCP components based on the data of Adams27 and Cloutis and Gaffey28 (Fig. 3c). Here we define the HCP as wollastonite (Wo)>30 and LCP as Wo<30, keeping with previous work by Sunshine et al.31 and Klima et al.30. By comparison, the compositional features of LCP of the four soils are similar (Fig. 3c) and very Fe-rich, relative to orthopyroxene examined by Adams27 and Cloutis and Gaffey28. However, the HCP compositions of CE-3 soils occur in two groups; CE3-0006 and CE3-0007 are slightly richer in Ca and in Fe than CE3-0005 and CE3-0008, consistent with APXS data (Table 1). The pyroxene chemistry of the CE-3 soils derived from VNIS data thus supports their general Fe-rich character, with CE3-0006 and CE3-0007 having even higher Fe contents, consistent with APXS results.

The volume percentage ratio of HCP and LCP (HCP/LCP) can be estimated using the band-strength ratios of 1 and 2 μm bands from MGM deconvolution of VNIS spectra19,24,29,30,32. The HCP/LCP vol.% ratios for four CE-3 soils were calculated using both the 1- and 2-μm band-strength ratios. The results for each soil using two ratios are consistent (Table 2), indicating an equivalent compositional effect on both 1 and 2 μm bands. Overall, the HCP/LCP ratios in four CE-3 soils are similar, with HCP about two times LCP in abundance.

The MGM-derived band positions of olivine shift as a function of Fo content, thus they can be used to estimate olivine chemistry20,25. We plot central positions of two deconvolved M1 component bands of CE-3 olivine (Fig. 3d) with trend lines determined on terrestrial samples by Sunshine and Pieters20. The central positions of the two olivine M1 bands occur at 870–884 nm and 1234–1261, nm for the four CE-3 soils, suggesting they are Fe-rich (30<Fo<55, Fig. 3d). Specifically, spectra indicate that CE3-0005 and CE3-0008 soils have higher olivine Fo values than the other two soils.

The precise location of the M2 band (∼1050, nm) of olivine in the VNIS spectrum is difficult to determine via MGM deconvolution20,33 (Supplementary Note 5). However, we plotted the central positions of olivine M2 bands of four CE-3 soils derived from MGM deconvolution in Fig. 3d, which also shows a trend along the trend line determined by Sunshine and Pieters20. Therefore, olivine chemistry of the CE-3 soils derived from VNIS data supports their general Fe-rich character, consistent with normative analysis of the APXS results (Fo ∼43 on average, Table 1). The band-strength ratios of the HCP 1-μm band to the olivine M1 band near 1.25 μm could also be used to estimate their volume percentage24. The four CE-3 soils can be divided into two groups (Table 2): the CE3-0005 and -0008 soils are richer in olivine on the basis of VNIS analysis (HCP/OL=2.0 and 2.3) than CE3-0006 and CE3-0007 (HCP/OL=3.0 and 3.3, respectively).

Refinement of mineral mode by correlated APXS and VNIS

A key result from both the APXS and VNIS data is the inferred abundance of olivine. The APXS data indicate relatively low SiO 2 and high FeO+MgO, resulting in a significant proportion of olivine in the norm (10 vol.% in CE3-0006 and 30 vol.% in CE3-0008, Table 1). Our MGM analysis of the VNIS spectra also reflects high olivine contents (the HCP/OL ratio is 3.0 for CE3-0006 and 2.3 for CE3-0008, Table 2). A high olivine content coupled with intermediate to high TiO 2 makes the CE-3 soil and the basalt from which it derives unique among the known lunar samples, similar to a basalt type that has been inferred from orbital data34, but not until the CE-3 mission verified by in situ or sample analysis.

Second, olivine chemistry derived from the norm analysis based on the APXS composition (Table 1) and from the central positions of the olivine M1 bands in VNIS spectra (Fig. 3d) both support the Fe-rich character of olivine in CE-3 soils. Fe-rich olivine was predicted on the basis of remote sensing of this area21, thus the Fe-rich olivine found by CE-3 indicates a relatively evolved magma from late-stage volcanic activity in the Imbrium basin35.

We also find similarity in pyroxene features inferred from normative analyses of the APXS composition (Table 1) and deconvolved VNIS spectra (Table 2). For example, both sets of analyses suggest CE-3 soils are rich in HCP. A good match was found between HCP/LCP ratios in CE3-0006 soil derived from VNIS (2.0, Table 2) and the Di/Hy components derived from APXS analysis (2.0, Table 1). The CIPW norm analysis based on APXS composition indicates a much higher Di/Hy ratio for CE3-0008 (8.7, Table 1) than the HCP/LCP derived from VNIS (2.3, Table 2). This large difference results in part from the effect of the normative pyroxene components in which the Hy component has no Ca. In reality the LCP pyroxene (in this case, pigeonite) does contain Ca. The effect is greatest in CE3-0008 because it contains so much olivine that there is little normative Hy and abundant Di. Moreover, the VNIS and APXS target areas were within a short distance of each other (<1 m), so we do not expect such a large variation in this less (space-) weathered basaltic regolith.

To provide a better comparison between APXS data and VNIS data, we use typical compositions of lunar mare minerals (that is, olivine, augite, pigeonite and plagioclase) as endmembers in a mixing-model calculation instead of the normative calculation results (Supplementary Table 7). The results of this mixing analysis are shown in Table 3.

Table 3: Mineralogy of Chang'e-3 soils derived from APXS data using mixture modelling of the chemical composition. Full size table

From the mixing analysis, the refined mineral mode of CE3-0008 yields an Aug/Pig ratio of 2.4 (Table 3), which matches well with the VNIS HCP/LCP ratio (2.3, Table 2) based on VNIS spectral deconvolution. For CE3-0006, the Aug/Pig value in the refined mineral mode (2.0) and the HCP/LCP value in the VNIS-derived mode (2.0) are essentially the same. Considering these results, the APXS and VNIS mineral modal data are consistent.

The CE3-0008 soil may have a greater abundance of material contributed from deeper levels of the nearby Zi Wei crater (Fig. 1e), with a composition similar to the nearby light-toned rocks (Supplementary Figs 2 and 3) such as the ‘Outer Fence’ boulder. From an image-based mineral modal estimation of Outer Fence, we infer ∼20 vol.% in plagioclase phenocrysts (Supplementary Fig. 3, Supplementary Table 5, Supplementary Note 2), which is in general agreement with our estimations (∼33 vol.%) of nearby CE3-0008 soil. Moreover, the regolith reflectance differences observed by the Lunar Reconnaissance Orbiter Camera Narrow Angle Cameras (LROC NAC) are rather limited around the landing site (Supplementary Fig. 4, Supplementary Note 3). The Al 2 O 3 content of CE-3 landing site based on experience of lunar samples is most likely in the range of 7–10 wt.% (Supplementary Fig. 5, Supplementary Note 4). The reflectance of the nearby rocks at the CE-3 landing site may result from texture-related human-eye brightness exaggeration (that is, some workers36 initially interpreted the rock as an aluminous basalt with a plagioclase content possibly exceeding 40 vol.% (see Supplementary Notes 2–4 for additional discussion).

Correspondence between the landing site and remote sensing