Pb Isotope Signature of a Low-μ ( 238 U/ 204 Pb) Lunar Mantle Component

The chemical and isotopic characteristics of terrestrial basalts are constrained within the concept of mantle chemical geodynamics that explains the existing variety of basaltic rocks within a framework of several end-member reservoirs in Earth’s mantle. In contrast, there is no comparable fully developed model explaining the isotopic composition of lunar basaltic rocks, in part owing to the lack of well-constrained age–isotope relationships in different groups of basalts identified on the Moon. Notably, the absence of agreement upon ages includes basalts from a unique group of meteorites collectively known as ‘YAMM’ (basalts Yamato-793169: Y-793169, Asuka-881757: A-881757, Miller Range 05035: MIL 05035 and regolith breccia Meteorite Hill 01210: MET 01210), which appear to show chemical signatures different from all other known lunar basaltic rocks. We present high-precision Pb–Pb ages and initial Pb isotopic ratios for two samples from this group, MIL 05035 and A-881757. These meteorites have Pb isotope ratios different from those of the other lunar basalts, suggesting they are derived from a distinct and depleted mantle source, with a 238 U/ 204 Pb ratio ( μ value) lower than any other mantle source. Their depletion in rare earth elements, in conjunction with recalculated initial Nd and Sr isotopic ratios from published data and using our new age, appear to support this conclusion. The chemical and Sr-Nd-Pb isotopic characteristics of this low-μ source appear to be the opposite of those of the KREEP reservoir and many, if not all, features described in other lunar basalts (such as low-and high-Ti mare basalts) can be explained by a binary mixing of material derived from low-μ and KREEP-like reservoirs. This mixing might be the result of a slow, convection-like mantle overturn.


INTRODUCTION
Our current understanding of mantle evolution on the Moon is framed within the concept of differentiation of a global lunar magma ocean (LMO; Warren & Wasson, 1979;Warren, 1985).In this hypothesis, crystallization of LMO resulted in the formation of a stratified lunar mantle (e.g.Elardo et al., 2011), with geochemically distinct domains, which are the sources of the different chemical types of lunar basalts (Neal & Taylor, 1992;Shearer & Papike, 1999).This hypothesis implies that the earlier Mg-rich olivine and Mg-rich pyroxene cumulates formed from LMO could be the mantle sources of low-titanium types of basalts (Taylor & Jakes, 1974;Longhi, 1992;Neal & Taylor, 1992;Giguere et al., 2000;Joy et al., 2008).The late stages of the crystallization sequence would be marked by formation of ilmenite-rich cumulates and an anorthosite crust, leaving a residual liquid enriched in incompatible elements like potassium (K), rare earth elements (REE), phosphorus (P) and uranium (U).This incompatible elementenriched liquid eventually crystallized to form a shallow mantle layer named urKREEP, which has been proposed as the source of enriched KREEP basalts (e.g. Warren & Wasson, 1979).The high-Ti basalts would have been derived from a source with a contribution from the late ilmenite-rich cumulates (Taylor & Jakes, 1974;Longhi, 1992;Neal & Taylor, 1992;Snyder et al., 1992;Beard et al., 1998).Gravitational instability in the LMO cumulate pile might have developed as a result of the crystallization sequence, resulting in a mantle overturn possibly just after the final LMO solidification (4372 ± 35 Ma), which would have produced some degree of heterogeneity within the lunar mantle (Hess & Parmentier, 1995;Elkins-Tanton et al., 2011;Borg et al., 2020;Sio et al., 2020).This concept seems to be supported by chemical and isotopic data obtained from different types of low-Ti and high-Ti basalts (collectively known as mare basalts), in which varying quantities of material derived from urKREEP were added either to their sources or assimilated during their ascent (e.g.Neal & Taylor, 1992;Snyder et al., 1994;Sprung et al., 2013;Hallis et al., 2014).Reinforcing this previous interpretation, the most recently obtained Pb isotope data indicate a systematic increase in KREEP proportion from older to younger low-Ti basalts (Snape et al., 2019;Merle et al., 2020).However, the potential mantle components contributing to the chemical characteristics of the diverse groups of lunar basalts remain unidentified.
In contrast to Apollo samples, a small set of low-Ti lunar basalts from a group of meteorites collectively known as 'YAMM' (basalts Y-793169, A-881757 and MIL 05035 and regolith breccia MET 01210) shows chemical signatures compatible with a depleted mantle (Misawa et al., 1993;Torigoye-Kita et al., 1995;Nyquist et al., 2007;Terada et al., 2007).They have lower REE contents, a depletion in light REE compared to middle and heavy REE, no negative Eu anomalies and apparently (due to poorly constrained age) low initial 87 Sr/ 86 Sr and high initial 143 Nd/ 144 Nd, suggesting a KREEP-free depleted mantle source (Joy et al., 2008;Liu et al., 2009;Arai et al., 2010;Elardo et al., 2014;Srivastava et al., 2022).Previous Pb-isotopic studies of meteorites from this group indicated a low-μ ( 238 U/ 204 Pb ratio) of their source (μ Basalt Source : μ BS ), in the range of ∼10-22 (Misawa et al., 1993;Torigoye-Kita et al., 1995).To date, the basalts of the 'YAMM' meteorites (hereafter 'YAM' basalts) have not been investigated using the recently developed SIMS (Secondary Ion Mass Spectrometry) Pb approach, which largely eliminates the inf luence of terrestrial contamination that can bias U-Pb ages, Pb initial isotope ratios and 238 U/ 204 Pb (Snape et al., 2016;Merle et al., 2020;Connelly et al., 2022).In addition, this approach presents the advantage of obtaining accurate and precise ages as well as initial Pb compositions of the samples (Snape et al., 2016;Merle et al., 2020).Therefore, it is unknown whether the 'YAM' basalts follow the Pb isotope trend defined by studied Apollo basalts, suggesting that they can represent the most depleted KREEP-free end-member of mixing, or fall off this trend.In the latter case, this might indicate an entirely different reservoir on the Moon unrelated to the sources of other lunar basalts (e.g.Arai et al., 2010).Furthermore, the YAM basalts were thought to have crystallized ∼3800-3900 Ma (Misawa et al., 1993;Torigoye-Kita et al., 1995;Nyquist et al., 2007;Fernandes et al., 2009;Zhang et al., 2010) but there was no agreement upon their crystallization age.This precluded the recalculation of robust and precise initial Sr and Nd isotope ratios using previously published data in order to establish Sr-Nd-Pb systematics for the YAM basalts suitable to compare with other lunar basalts.To tackle this issue, we have determined Pb-Pb ages and initial Pb isotopic compositions of A-881757 and MIL 05035 using SIMS.
Previous investigations of the YAM basalts (Yanai, 1991;Koeberl et al., 1993;Warren & Kallemeyn, 1993;Joy et al., 2008;Fernandes et al., 2009;Liu et al., 2009;Arai et al., 2010) showed strong mineralogical and chemical similarities between these samples, suggesting that A-881757 and MIL 05035 might even originate from the same lava flow (Arai et al., 2010).The reported presence of Caphosphates, K-feldspars and Zr-rich minerals (e.g.Misawa et al., 1993;Terada et al., 2007;Zhang et al., 2010) clearly render them as suitable for initial Pb isotope composition and Pb-Pb isochron age determination by SIMS (Snape et al., 2016;Merle et al., 2020), allowing a re-evaluation of their μ BS values, free of terrestrial contamination of the mantle source of the YAM basalts.With our new Pb-Pb age, we recalculated the initial Sr and Nd isotopic ratios that allowed us to (1) establish Sr-Nd-Pb systematics for the YAM basalts and (2) compare these isotopic ratios with those from other lunar basalts to provide a proper characterization of the mantle sources of the lunar basaltic magmatism.
All 27 analyses from A-881757 define an isochron (Fig. 2) with an age of 3864.8 ± 3.7 Ma (95% confidence, MSWD = 0.72, P = 0.84) that is indistinguishable within uncertainties to the age of MIL 05035.Three analyses of potassium feldspar yielded identical 204 Pb/ 206 Pb and 207 Pb/ 206 Pb ratios within uncertainties and define weighted average values for 204 Pb/ 206 Pb and 207 Pb/ 206 Pb ratio of 0.0354 ± 0.0025 (2σ, MWSD = 0.22, P = 0.80) and 1.2111 ± 0.014 (2σ, MSWD = 1.2, P = 0.29), respectively.As these ratios are slightly lower than those of MIL 05035, these initial Pb compositions could be underestimated.If the feldspars contain small quantities of U, the presence of additional radiogenic Pb formed in situ from decay of this U will produce an apparent initial Pb composition with lower 204 Pb/ 206 Pb and 207 Pb/ 206 Pb ratio.To test for this possibility and determine a more accurate initial ratio, we used a hybrid SIMS-TIMS method ('Method B' of Connelly et al., 2022).This method involves correcting the 207 Pb/ 206 Pb, 204 Pb/ 206 Pb and 238 U/ 206 Pb ratios measured by Thermal Ionization Mass Spectrometry (TIMS) in mineral or bulk-rock fractions for terrestrial contamination and using projection of the TIMS data onto the SIMS isochron.These terrestrial Pb-free ratios are used in the U decay equations to calculate the initial Pb isotope ratios (see details in supplementary material).For A-881757, we used previously published TIMS data obtained from three plagioclase fractions (Misawa et al., 1993) to calculate the initial 207 Pb/ 206 Pb and 204 Pb/ 206 Pb ratios (see details in supplementary material).The resulting initial Pb isotopic compositions for these three plagioclase analyses are indistinguishable within uncertainties and define a mean value for 204 Pb/ 206 Pb and 207 Pb/ 206 Pb of 0.0404 ± 0.0103 and 1.344 ± 0.244 (2σ).This composition is similar to the initial Pb isotope ratios measured in MIL 05035 within uncertainties (Table 1).

Revisiting the age of the YAM basalts
The new Pb-Pb ages presented here are in the range of previously published dates (Misawa et al., 1993;Torigoye-Kita et al., 1995;Nyquist et al., 2007;Zhang et al., 2010), but are determined with an order of magnitude of improvement in precision and robustness (see Merle et al., 2020 for a review of the robustness of the previously published dates).These ages also appear to be similar to the age of KREEP basalts, although the Pb-Pb age of the latter has a large uncertainty (77 Ma; Snape et al., 2019).A recent study suggested that the YAM basalts may be formed during a relatively shallow melting event related to impact-induced decompression linked to the Late Heavy Bombardment (Srivastava et al., 2022).However, this interpretation seems in contradiction with a partially melted deep-mantle source for these basalts (e.g.Joy et al., 2008).Furthermore, our new Pb-Pb dates yield a weighted average age of 3864 ± 3 Ma (2σ) for YAM basalts.This is younger than the Imbrium basin, the youngest on the Moon, for which there is a robust age of 3922 ± 12 Ma (Liu et al., 2012;Nemchin et al., 2021).

Chemical characteristics of the mantle source of the YAM basalts
Initial Pb isotope compositions determined for MIL 05035 and A-881757 allow calculation of μ values in their mantle sources (μ BS ) during a two-stage model (see methods).For MIL 05035 the observed range of μ BS can be expressed as a mean and 2 sigma standard deviation of 93 ± 9. A similar estimate for A-881757 of 60 ± 54 is significantly less precise, largely due to the less precise estimate of the initial Pb composition for this sample that itself results from use of relatively imprecise plagioclase U-Pb data (Misawa et al., 1993) used in correcting for terrestrial contamination and in situ ingrowth of radiogenic Pb.This result indicates that the μ BS for A-881757 is unlikely to exceed ∼120-150 and, similar to that of MIL 05035, is significantly lower than previously estimated μ BS for Apollo basalts (μ BS = 360-650 for both low-Ti and high-Ti basalts, Snape et al., 2019).Both A-881757 and MIL 05035 also appear to have higher estimated μ BS than previous estimates made for YAM basalts (i.e. 10 ± 3 in A-881757 and 22 ± 4 in Y-793169, Misawa et al., 1993 andTorigoye-Kita et al., 1995) but these earlier lower estimates are likely explained by their not fully accounting for terrestrial contamination.On the 204 Pb/ 206 Pb vs 207 Pb/ 206 Pb diagram (Fig. 3a), A-881757 and MIL 05035 plot distinctly away from both low-Ti and high-Ti basalts and KREEP basalts.Significantly higher initial 204 Pb/ 206 Pb of A-881757 and MIL 05035 supports a relatively unradiogenic mantle source for YAM basalts (Fig. 3a).The lower μ BS in YAM basalts, coupled with their lower contents of incompatible trace elements, a distinctive depletion in LREE relative to the MREE and HREE and a f lat MREE-HREE pattern compared to the other mare basalts are also consistent with their derivation from a depleted mantle source.

The low-μ source as a distinct mantle chemical end-member
In the 204 Pb/ 206 Pb versus 207 Pb/ 206 Pb plot (Fig. 3a) and also in the 87 Sr/ 86 Sr versus 143 Nd/ 144 Nd diagram (Fig. 3b), the YAM basalts plot distinct from the both KREEP and other lunar basalts, regardless of the age of the samples, supporting the uniqueness of the isotopic signature of YAM basaltic samples.The lack of a negative Eu-anomaly, unlike the other lunar basalts, suggests that the source of the YAM basalts must have formed early in the LMO sequence, before plagioclase saturation (Joy et al., 2008;Arai et al., 2010;Elardo et al., 2014).
In contrast, the urKREEP mantle component formed at the end of the LMO crystallization ∼4370 Ma (Borg et al., 2019).This component is clearly identified based on isotope ratios and characterized by a strong negative Eu anomaly, strong enrichment in LREE, high μ and Rb/Sr and low Sm/Nd (Warren & Wasson, 1979;Shearer et al., 2006).These chemical signatures are the opposite of those of the source of the YAM basalts.As the YAM basalts might be derived from a urKREEP-free source (Srivastava et al., 2022), this suggests that KREEP and YAM basalts sources may represent two extreme chemical end-member reservoirs in the lunar mantle.
The true KREEP basalts as defined by the chemical characteristics of the Apollo 15 basalts (e.g.sample 15 386) are closely associated with the Procellarum KREEP Terrane (PKT, Jolliff et al. 2000).The lack of a KREEP-like contribution in the chemical characteristics of the YAM basalts suggests that the source of the latter is located outside of PKT (Srivastava et al., 2022) and that the KREEP and low-μ mantle components could be geographically distinct.Proposed locations for the origin of the YAMM meteorites include the Schickard crater (Arai et al., 2010) or larger basins like the Mare Crisium, Fecundidatis, Australe or Humboldtianum (Nyquist et al., 2007;Joy et al., 2010).The fact that the typical KREEP basalts from the Apollo 15 landing site (Apollo 15, sample 15386) and YAM basalts erupted at a similar time, close to 3865 Ma, without noticeable chemical evidence of interaction (Nyquist et al., 2007;Joy et al., 2008;Arai et al., 2010;Srivastava et al., 2022) also argues for isolation of their sources within the mantle.All these lines of evidence suggest that the YAM basalts are derived from a unique mantle component formed relatively early in the LMO crystallization sequence.

Possible genetic link between the low-μ mantle component and the main chemical groups of mare basalts
Mixing between a urKREEP source and a less enriched component has been previously suggested to explain the chemical and isotopic characteristics of both low-Ti and high-Ti basalts (e.g.Hughes et al., 1988;Jerde et al., 1994;Rankenburg et al., 2007;Wang et al., 2012;Valencia et al., 2019;Merle et al., 2020).It is possible that the source of YAM basalts may represent this less enriched component for at least some mare basalts, considering that all available data from the lunar basalts plot between the theoretical evolution curves of the urKREEP source and low-μ mantle source in the Pb-Pb and Sr-Nd plots (Fig. 3).
A binary mixing of urKREEP that has a μ value potentially exceeding 2000, with a depleted mantle source component (the source of the YAM basalts) that has a μ value of ∼90 (see other parameters in supplementary material), can explain the Pb isotope compositions of most lunar basalts with a few to ∼10-15% contribution of KREEP (Fig. 3a).This also appears to be supported by 87 Sr/ 86 Sr and 143 Nd/ 144 Nd systematics (Fig. 3b).These estimates agree with previous studies suggesting potential KREEP mixing Fig. 207 Pb/ 206 Pb vs. 204 Pb/ 206 Pb plots for MIL 05035 and A-881757.All data points are represented as error crosses.Top panel: plot representing data from MIL 05035.Black crosses are data used for the isochron and light grey crosses are data rejected from the regression.Middle panel: data from MIL 05035 used for the isochron calculations.Bottom panel: data from A-881757.No data were rejected for the construction of the isochron.For colour code corresponding to the analysed phases refers to the digital version of this manuscript (red = phosphate; black = sulphide; blue = K-rich feldspar or K-rich feldspathic glass; green = K-rich feldspar+phosphate intricate mixture).The thick blue crosses are data used for calculation of initial 207 Pb/ 206 Pb and 204 Pb/ 206 Pb ratios.Also shown, the composition of average terrestrial Pb using the values from Stacey & Kramers (1975).Decay constants used for age calculation are according to Steiger & Jäger (1977).
of up to 10% for the low-Ti basalts and up to 15% for the high-Ti basalts (Jerde et al., 1994;Snyder et al., 1994).The NWA 773 clan present the highest contribution of KREEP-like component with ∼20% (Fig. 3a), lower than the 30% contribution previously estimated (Jolliff et al., 2003).This model also seems to confirm the involvement of a KREEP-like component in the chemical characteristics of the Chang'e 5 samples, but our estimated contribution (∼15%) is higher than previously calculated (1-1.5%,Zong et al., 2022).

Identification of the process leading to mantle component mixing
An apparent increase in the proportion of KREEP as basalts become progressively younger has been documented by Pb isotope ratios for the low-Ti and high-Ti basalts (Snape et al., 2019;Merle et al., 2020;Fig. 3a).This feature may point out to the mechanism(s) responsible for the mixing.The addition of KREEP to lunar basalts has been proposed in a number of previous studies and explained in a number of ways: (i) Assimilation of KREEP-rich materials by basaltic magma during its ascent or at lunar surface (e.g.Neal & Taylor, 1992;Jolliff et al., 2003;Wang et al., 2012).However, an age-related increase of KREEP proportion (Fig. 3a) is difficult to explain by melt assimilation of material either during the ascent or on the lunar surface.The potential contaminants of basaltic magmas are the urKREEP mantle layer, which likely formed during the last stage of LMO crystallization (∼4350 Ma; Borg et al., 2020), and surface material enriched in KREEP components formed at the latest by Imbrium impact (at ∼3920 Ma), which probably excavated KREEPrich rocks and distributed these materials over a large area on the near side of the Moon.These materials could also have been excavated and distributed over the lunar surface before that by earlier basin-forming events.
(ii) Incompatible element-enriched liquids may become trapped during basalts source formation (e.g.Snyder et al., 1994) (iii) Different degrees of admixing of KREEP into basalt sources during rapid mantle overturn might occur (Snyder et al., 1992).
The correlation between basalt age and proportion of mixed enriched material (Fig. 3; Merle et al., 2020) argues against these latter two mechanisms.In the first mechanism, these potential KREEP-rich contaminants should have existed from the onset of the basaltic magmatism on the Moon, and the proportion of assimilated KREEP-rich materials should be randomly distributed in different basalts irrespective of the time of their formation.Equally, the presence of trapped melt in the source(s) of basalts should exist from the time of LMO crystallization.It would potentially support heating and even melting where a large proportion of this trapped melt is located.Trapped melt would also be favourable for long-lasting sustained melting through time and possibly lead to a progressively more radiogenic signature in basaltic melts.However, it cannot account for the absence of older basalts (>3900 Ma) with a high proportion of KREEP mixed into the melt.A similar argument can be made against different levels of admixing of KREEP into basalt sources during a rapid mantle overturn in the immediate aftermath of LMO crystallization.An alternative explanation of the observed isotopic relationships is proposed here, which invokes slow (rather than instantaneous) mantle overturn, akin to terrestrial mantle convection.This type of slow overturn could introduce KREEP materials to the parts of the mantle beneath the urKREEP layer, gradually increasing enrichment of the deeper lunar mantle in time.Carlo simulation) was used to determine the Pb evolution curve of the mantle source of the YAM basalts and for the KREEP source, we used a μ source of 3800 ± 500 (Snape et al., 2019).(b) Sr-Nd initial ratios of lunar basalts.Measured ratios were corrected for in situ decay using ages from Snape et al. (2019) for Apollo rocks, age compilation of Merle et al. (2020) and this work for meteorites.Decay constant values from Steiger & Jäger, 1977 ( 238 U, 235 U and 87 Rb) and Lugmair & Marti, 1978 ( 147 Sm).Details of calculation given in supplementary material.To construct the time-related evolution curves of lunar mantle components, we used 87 Rb/ 86 Sr = 0.0121 ± 0.0029 and 147 Sm/ 144 Nd = 0.3257 ± 0.0407 for the low-μ mantle source and for the KREEP-like mantle source, 87 Rb/ 86 Sr source = 0.1486 ± 0.0271 and 147 Sm/ 144 Nd source = 0.1847 ± 0.0403 (see supplementary materials for details).Nd data from Lugmair et al., 1975;Nyquist et al., 1979Nyquist et al., , 1981;;Unruh et al., 1984;Misawa et al., 1993;Snyder et al., 1994Snyder et al., , 1997aSnyder et al., , 1997bSnyder et al., , 1998;;Rankenburg et al., 2007;Borg et al., 2009;Haloda et al., 2009;Sprung et al., 2013;McLeod et al., 2014;Borg et al., 2019;Tian et al., 2021. Sr data from Compston et al., 1970;Papanastassiou & Wasserburg, 1970;Papanastassiou et al., 1970;Cliff et al., 1971;Compston et al., 1971;Murthy et al., 1971;Papanastassiou & Wasserburg, 1971;Chappell et al., 1972;Murthy et al., 1972;De Laeter et al., 1973;Evensen et al., 1973;Papanastassiou & Wasserburg, 1973;Birck et al., 1975;Papanastassiou & Wasserburg, 1975;Nyquist et al., 1975;Nyquist, 1977;Nyquist et al., 1976Nyquist et al., , 1977Nyquist et al., , 1979Nyquist et al., , 1981;;Murthy & Coscio Jr., 1976, 1977;Misawa et al., 1993;Snyder et al., 1994Snyder et al., , 1997aSnyder et al., , 1998;;Borg et al., 2004;Rankenburg et al., 2007;Borg et al., 2004Borg et al., , 2009;;Elardo et al., 2014;Tian et al., 2021.

CONCLUSIONS
The new precise ages of 3862.1 ± 4.5 and 3864.8 ± 3.7 Ma determined for two meteorite samples representing the unusual YAM basalts make the time of their formation indistinguishable within uncertainties from that of KREEP basalts.Nevertheless, YAM and KREEP basalts show contrasting isotope characteristics indicative of significant difference in their sources, with one being significantly depleted and other one enriched in incompatible elements.All other basalts from the Moon show intermediate initial isotope compositions for multiple systems (Nd, Sr and Pb) suggesting a possible origin by mixing between material derived from low-μ and KREEP reservoirs for many, if not all, of these basalts.We suggest that this mantle component mixing resulted from a slow mantle overturn process rather similar to convection.

Table 1 :
Summary of new U-Pb data for MIL 05035 and A-881757 This age difference and the absence of radioisotope age estimate for the ∼3850-Ma-old Orientale basin(Stöff ler et al., 2006)makes it purely hypothetical to link formation of YAM basalts and impact basin-forming events.