The Ca isotope compositions of mare basalts offer a novel insight into the heterogeneous nature of the lunar mantle. We present new high-precision Ca isotope data for a suite of low-Ti and high-Ti mare basalts obtained using our collision cell MC-ICP-MS/MS instrument, Proteus. Mare basalts were found to have a Ca isotope composition resembling terrestrial basalts (${\delta }^{44/40}$Ca_{SRM 915a} =0.78–0.89‰) even though they are derived from a differentiated, refractory cumulate mantle source. Modelling of Ca isotope fractionation during crystallisation of a lunar magma ocean (LMO) indicates that the dominantly harzburgitic cumulates of the lunar interior should be isotopically heavier than Earth's mantle (${\delta }^{44/40}$Ca_{SRM 915a} =1.1–1.2‰ versus 0.93‰, respectively). These are balanced by an isotopically light lunar anorthosite crust, consistent with data for lunar anorthosite and feldspathic breccia meteorites.

We investigate the major element and Ca isotope composition of partial melts of various cumulate reservoirs by combining pMELTS models with equilibrium isotope fractionation mass balance calculations. The principal finding is that harzburgite cumulates alone are too refractory a source to produce low-Ti magmas. Partial melts of harzburgite cumulates have too low CaO contents, too high Al₂O₃/CaO and too high ${\delta }^{44/40}$Ca_{SRM 915a} to resemble low-Ti magmas. From Ca isotope constraints, we find that the addition of 10–15% of late-stage, clinopyroxenite cumulates crystallising at 95% LMO solidification is required to produce a suitable source that can generate low-Ti basalt compositions. Despite the addition of such late-stage cumulates pMELTS finds that these hybrid sources are undersaturated in clinopyroxene and are thus consistent with experimental constraints that the mantle sources of the lunar magmas are clinopyroxene-free. High-Ti basalts have slightly lower ${\delta }^{44/40}$Ca_{SRM 915a} (0.80–0.86‰) than low-Ti magmas (0.85–0.89‰) and clearly elevated TiO₂/CaO. No suitable hybrid source involving ilmenite-bearing cumulates (IBC) was found that could reproduce melts with appropriate ${\delta }^{44/40}$Ca and major element systematics. Instead, we suggest that metasomatism of low-Ti mantle sources by IBC melts is the most plausible way to generate high-Ti magma sources and the rich diversity in TiO₂ contents of lunar basalts and pyroclastic glasses.

海马玄武岩的 Ca 同位素组成提供了对月球地幔异质性质的新见解。我们展示了使用我们的碰撞池 MC-ICP-MS/MS 仪器 Proteus 获得的一套低钛和高钛玄武岩的新高精度 Ca 同位素数据。海玄武岩被发现具有类似于陆地玄武岩的 Ca 同位素组成（${\delta }^{44/40}$Ca _{SRM 915a}  =0.78–0.89‰)，即使它们来自分化的、难熔的累积地幔源。对月球岩浆海洋 (LMO) 结晶过程中 Ca 同位素分馏的建模表明，月球内部主要的哈茨伯里岩堆积体在同位素上应该比地球地幔重。${\delta }^{44/40}$Ca _SRM  915a 分别为 1.1–1.2‰ 和 0.93‰）。这些由同位素轻的月球斜长岩地壳平衡，与月球斜长岩和长石角砾岩陨石的数据一致。

我们通过将 pMELTS 模型与平衡同位素分馏质量平衡计算相结合，研究了各种累积储层的部分熔体的主要元素和 Ca 同位素组成。主要的发现是，单独的菱镁矿堆积物过于难熔，无法产生低钛岩浆。方辉石部分熔体的 CaO 含量太低，Al ₂ O ₃ /CaO 太高，太高${\delta }^{44/40}$Ca _{SRM 915a}类似于低钛岩浆。根据 Ca 同位素约束，我们发现需要添加 10-15% 的后期单斜辉石，在 95% 的 LMO 凝固时结晶，以产生可以生成低钛玄武岩成分的合适来源。尽管添加了这样的后期累积，pMELTS 发现这些混合源在单斜辉石中是不饱和的，因此与月球岩浆的地幔源不含单斜辉石的实验限制一致。高钛玄武岩略低${\delta }^{44/40}$Ca _{SRM 915a} (0.80–0.86‰) 比低钛岩浆 (0.85–0.89‰) 和明显升高的 TiO ₂ /CaO。没有发现合适的混合源涉及含钛铁矿的累积物 (IBC)，可以用适当的方法复制熔体${\delta }^{44/40}$Ca 和主要元素系统学。相反，我们认为 IBC 熔体对低钛地幔源的交代是产生高钛岩浆源和月球玄武岩和火山碎屑玻璃中丰富多样的 TiO ₂含量的最合理方式。