3.1 Ca Isotope Difference Between Diogenites and Eucrites and Implications for Their Petrological Relationships

The difference in Ca isotopes between eucrite and diogenite meteorites is unlikely to be caused by terrestrial weathering because fall and find meteorites in a same group/type do not show systematic Ca isotope differences (Amsellem et al., 2017; S. Huang & Jacobsen, 2017; Klaver et al., 2021; Magna et al., 2015; Valdes et al., 2014; Wu et al., 2020). Rather, the variation in Ca contents between eucrites and diogenites results from their different mineralogy. Most of the diogenites are monomineralic orthopyroxene cumulates with low CaO contents (<2.5 wt.%), while eucrites are more Ca-rich basaltic rocks containing orthopyroxene, pigeonite, high-Ca pyroxenes, and plagioclase phenocrysts (Mittlefehldt, 2015). The δ^44/40Ca variation among eucrites and diogenites and their correlation with Ca content and Mg# is likely caused by magmatic processes (Figure 1). Given the preference of orthopyroxene for heavy Ca isotopes (Feng et al., 2014; S. Huang et al., 2010), the high δ^44/40Ca of the diogenites can result from equilibrium crystallization of orthopyroxene that is Ca-poor and Mg-rich.

To test the petrological relationship between eucrites and diogenites, that is, whether they are co-genetic, we model the Ca isotope and major element implications (Mg# and Ca contents) of this process using a simple orthopyroxene fractional crystallization model (Figure 1), based on the experimentally determined liquid line of descent of Ashcroft and Wood (2015). The bulk δ^44/40Ca of the parental Vestan magma ocean in this model is assumed to be ordinary chondrites (OCs), with δ^44/40Ca = 0.92 ± 0.11‰ (S. Huang & Jacobsen, 2017; Valdes et al., 2014). The details of the parameters and methods are given in the Supporting Information. As shown in Figure 1, this model adequately explains the difference in δ^44/40Ca between the cumulate diogenites and derivative eucritic melt as an equilibrium signature. As orthopyroxene has a low Ca content, its crystallization has little leverage on the Ca budget of the melt. Hence, δ^44/40Ca of the melt only decreases by 0.01‰ even after removal of 40% orthopyroxene, which is consistent with the general homogeneity in δ^44/40Ca between eucrites with highly variable CaO contents (7.1–12.8 wt.%). This Ca isotope model also has significance on the petrogenesis of eucrites and diogenites. Formation of diogenites mostly result from the crystallization of OPX in the eucritic melt, and eucrites and diogenites can be co-genetic (Mandler & Elkins-Tanton, 2013; Righter & Drake, 1997; Ruzicka et al., 1997; Warren, 1997).

However, our Ca isotope model cannot reproduce the Ca isotope composition of the two diogenites, NWA 7831 and Dhofar 700, as their δ^44/40Ca is resolvably lower than the modeled composition of orthopyroxene cumulates. This inconsistency could reflect the two diogenites represent mixtures between orthopyroxene and the residual melt. Alternatively, some diogenites may crystallize from the different melts, due to their diversity of heavy REEs (Barrat & Yamaguchi, 2014; Barrat et al., 2008). However, note that, this argument about REEs is based on an assumption that HED parent body has chondritic REE composition, which might not be true, because a REE-depleted reservoir has not been found in HED sample collections, considering eucrites are largely rich (seven to ten times) in REEs relative to chondrites (Consolmagno et al., 2015).

3.2 Genetic Relationships Between HEDs and Mesosiderites

The anomalous eucrite, Pasamonte (δ^44/40Ca = 0.85 ± 0.05‰) has the same Ca isotope composition as other eucrites, which is consistent with their similar mass-independent ⁵⁴Cr/⁵²Cr (Trinquier et al., 2007). Note that Ibitira, another anomalous eucrite (Wilkening & Anders, 1975), also has the same mass-independent ⁵⁰Ti/⁴⁷Ti as normal eucrites (Trinquier et al., 2009). The anomalous eucrites, that is, Pasamonte and Ibitira, should have similar metal isotope compositions as normal eucrites (Trinquier et al., 2009). However, their different mass-independent ¹⁷O/¹⁶O (Mittlefehldt, 2015; Scott et al., 2009) could be evidence that these anomalous eucrites come from the Vestoids, that is, smaller asteroids around Vesta (Burbine et al., 2001), and experienced some secondary alteration (e.g., fluid-assisted metasomatic processes) that changed their O isotope compositions (Shisseh et al., 2023) but not their metal isotope compositions.

Similar δ^44/40Ca values between howardites and eucrites attest to howardites being physical mixtures of eucrites and diogenites (Mittlefehldt, 2015). The mean Ca contents and δ^44/40Ca of eucrites and diogenites are ∼10 wt.% and ∼2 wt.%, and ∼0.83‰ and ∼1.10‰, respectively, so the product of mixing (mass for mass) would be dominated by the isotopic signature of eucrites. For mesosiderites, the metal phase is Ca-free, so the Ca isotope composition of the two mesosiderites are representative of their silicate portions. Both Youxi (0.87 ± 0.004‰) and NWA 1182 (0.79 ± 0.01‰) have δ^44/40Ca values falling into the range of that of eucrites (0.83 ± 0.04‰, 2SD), supporting the idea that the silicate proportions of mesosiderites could have similar precursors as Vesta (Haba et al., 2019; Mittlefehldt, 2015). This is also consistent with similar mass-independent ⁵⁴Cr/⁵²Cr and ⁵⁰Ti/⁴⁷Ti between mesosiderites and HEDs, considering that Cr and Ti are also lithophile elements at Vesta's conditions (Trinquier et al., 2007, 2009).

3.3 Calcium Stable Isotope Composition of Bulk Vesta and Ca Isotopic Heterogeneities in the Inner Solar System Bodies

Since Ca is non-siderophile and does not exist in the core (Rubin, 2011), the Ca isotope composition of bulk silicate Vesta should be mostly representative of that of bulk Vesta. Since no mantle rocks of Vesta are in our meteorite collection, eucrites and diogenites are the only direct samples from the silicate proportions of Vesta. Diogenites consisting mostly of orthopyroxene and minor olivine are cumulates and do not represent the average composition of the Vestan mantle (Lunning et al., 2015). Also, the CaO contents in diogenites (0.8 wt.%–2.4 wt.%) are relative low, relatively to that of bulk silicate Vesta, for example, ∼2.5 wt.%, estimated by cosmochemical models (Ruzicka et al., 1997; Wänke & Dreibus, 1980). Hence, the diogenites with isotopically heavy Ca nay be difficult to link the Ca isotope composition of bulk Vesta. Alternatively, the Ca-rich basaltic eucrites (with 7 wt.%–13 wt.% CaO) represent a most important Ca reservoir of Vesta.

In Figure 2, we compared the basalts from different planets in the inner solar system, including for example, normal mid-ocean ridge basalts (N-MORB) from Earth (Eriksen & Jacobsen, 2022; H. Zhu et al., 2018), lunar basalts (Klaver et al., 2021; Valdes et al., 2014; Wu et al., 2020), shergottites from Mars (Magna et al., 2015), and eucrites from Vesta. The δ^44/40Ca value for N-MORB (0.84 ± 0.12‰; 2SD, N = 28), lunar basalts (0.86 ± 0.13‰; 2SD, N = 22), eucrites (0.83 ± 0.04‰; 2SD, N = 8; excluding Pasamonte) and shergottites (0.95 ± 0.22‰; 2SD, N = 14) have nearly identical Ca isotope compositions. Note that the δ^44/40Ca data for Apollo lunar basalts measured by collision-cell-based MC-ICP-MS using the double spike technique are more homogeneous: 0.85 ± 0.06‰ (2SD, N = 13) (Klaver et al., 2021). However, basalts generally do not represent the Ca isotope composition of the bulk planetary body. For example, the bulk Earth δ^44/40Ca value (∼0.94‰) is around ∼0.1‰ higher than the average N-MORB, which is caused by low-degree partial melting with clinopyroxene retaining a significant fraction of Ca in the residue (Eriksen & Jacobsen, 2022; Kang et al., 2017; Klaver et al., 2021; Soderman et al., 2022). Similarly, mare basalts are low-degree partial melts of a highly heterogeneous lunar mantle, but their δ^44/40Ca is consistent with a bulk (silicate) Moon that is isotopically heavier than lunar basalts and similar to bulk (silicate) Earth (e.g., Klaver et al., 2021; Wu et al., 2020). Also considering the relatively large Ca isotope variation and its 2SD uncertainty (e.g., >0.10‰) of basalts of a same planet/asteroid, we cannot directly compare their basalts for Ca isotope comparison between different planets.

The distinct petrogenesis of eucrites means that, unlike terrestrial MORB or lunar basalts, eucrites can be used to gauge the Ca isotope composition of bulk silicate Vesta. The major element composition of eucrites is consistent with fractional crystallization of orthopyroxene from a high-degree parental melt (Shearer et al., 1997). Subsequent evolution of this melt through fractional crystallization of orthopyroxene gave rise to the eucrites and diogenites. Calcium-poor, orthopyroxene-rich diogenites are in Ca isotope equilibrium with a eucritic melt (Figure 1). Our modeling shows that fractional crystallization of isotopically heavy orthopyroxene does not effectively alter δ^44/40Ca of the residual melt (Figure 1). The Ca content of orthopyroxene is so low that, even after 50% orthopyroxene removal, δ^44/40Ca of the residual melt has decreased by <0.03‰. From this, it follows that δ^44/40Ca of the eucrites, which represent the residual melt after 25%–50% orthopyroxene crystallization, is similar to their parental melt well withing the analytical resolution. As this parental melt is contains >99% of Ca on Vesta, with the remaining <1% hosted in deep olivine cumulates and the core, we argue that δ^44/40Ca of bulk Vesta is best estimated by the normal eucrites, that is, 0.83 ± 0.04‰ (2SD, N = 8; excluding Pasamonte).

A mass-balance calculation for the Ca isotope reservoirs in Vesta will be helpful to validate this hypothesis. Vesta has a thick crust (Consolmagno et al., 2015; Russell et al., 2012) that can be up to 80 km as thick as the Vestan mantle (Clenet et al., 2014), so the Vestan crust has a higher volume than the mantle by 7 times: V_crust = π × (160³−80³); V_mantle = π × 80³. Although the eucritc Vestan crust is intruded by diogenites, like the howadite, its Ca isotope composition should also be eucretic. Considering the much larger volume of the Vestan crust, although the Ca isotope composition of the olivine-rich mantle is unknown, we can predict the Ca isotope composition of bulk silicate Vesta can be represented by eucrites.

Previous studies have given the δ^44/40Ca values of bulk Earth and Mars, 0.94 ± 0.05‰ (Kang et al., 2017) and 1.04 ± 0.09‰ (Magna et al., 2015), respectively (Figure 2). As for Moon, lack of the mantle derived rocks limits the accurate estimation of its bulk δ^44/40Ca value, so we do not discuss it in this study. Although the uncertainty for the δ^44/40Ca value for Mars is relatively large, it can be seen Vesta has clearly different δ^44/40Ca value from Earth and Mars. Hence, the three inner solar system bodies, that is, Earth, Mars, and Vesta may have different mass-dependent Ca isotope compositions. The nucleosynthetic anomaly of ⁴⁸Ca isotope (expressed as ⁴⁸Ca/⁴⁴Ca ratios) is also a robust fingerprint to test the kinships between solar system materials (Dauphas et al., 2014; Moynier et al., 2010; Schiller et al., 2018, 2015; K. Zhu et al., 2023). In fact, Earth, Mars, and Vesta have different mass-independent ⁴⁸Ca/⁴⁴Ca signatures (Dauphas et al., 2014; Schiller et al., 2018), consistent with their mass-dependent Ca isotope difference. However, δ^44/40Ca and the mass-independent ⁴⁸Ca/⁴⁴Ca compositions of Earth, Mars, and Vesta do not correlate with each other, possibly because the different origins of the two Ca isotope fractionation mechanisms, that is, mass-dependent isotope fractionation and nucleosynthesis.

Volatile depletion is a key stage during planetary differentiation (Allègre et al., 2001), which could fractionate isotopes (e.g., Paniello, Day, & Moynier, 2012), even for some more refractory elements, for example, Si (Pringle et al., 2014), Ca and Ti (Zhang et al., 2014). Compared to Earth and Mars, Vesta has lower Rb/Sr and K/U, suggesting it is more depleted in volatile elements (Davis, 2006), and this volatile depletion of Vesta has been traced by Zn (Paniello et al., 2012), K (Tian et al., 2019) and Cr stable isotope measurements (K. Zhu et al., 2019). Kinetic isotope fractionation during element evaporation would enrich the heavy isotopes in the residue, which is not consistent with the isotopically lighter Ca of Vesta relative to those of Earth and Mars, so the Ca isotope difference is not likely caused by Ca evaporation during volatile loss. Since no other planetary processes are likely to fractionate Ca isotopes, their isotope difference can reflect their different precursor materials, and furthermore suggest Ca isotopic heterogeneities in the inner solar system. Earth, Moon, and Vesta accreted at different regions in the inner solar system and thus their Ca isotope difference suggests that their precursor materials experienced different condensation histories.

3.4 Ca Isotopes Test the Chondrule-Rich Accretion Model for Terrestrial Planets

Bulk chondrites and chondritic components are potential candidates for the planetary precursors (Allègre et al., 1995; Johansen et al., 2021). Ordinary chondrites (OCs) and enstatite chondrites (ECs) have a non-carbonaceous (NC) isotope anomaly signature, and thus constitute inner solar system material (Dauphas et al., 2014; Schiller et al., 2018; Steele et al., 2011; Trinquier et al., 2007, 2009). It is consistent that the inner-solar system bodies, including Earth, Mars, and Vesta, have similar δ^44/40Ca values as the NC chondrites, including OCs (OCs; 0.92 ± 0.11‰; 2SD, N = 7), and ECs (0.96 ± 0.08‰; 2SD, N = 6), instead of those of carbonaceous chondrites (CCs) with δ^44/40Ca values ranging from ∼0.2‰ to ∼1.0‰ (Amsellem et al., 2017; S. Huang & Jacobsen, 2017; Schiller et al., 2018; Valdes et al., 2014).

Amsellem et al. (2017) first found that chondrules in CV chondrites, with average δ^44/40Ca values of 1.10 ± 0.10‰, mimic those of the BSE with δ^44/40Ca = 1.05 ± 0.04 (S. Huang et al., 2010), and bulk silicate Mars, δ^44/40Ca = 1.04 ± 0.09‰ (Magna et al., 2015). Also considering chondrules in carbonaceous chondrites including CV, Ornans-type (CO) and Renazzo-type (CR) groups have similar volatile element (e.g., Mn, K, Ga, and Zn) contents and Si/Mg ratios as terrestrial primitive mantle, the authors argue that chondrules may contribute most (>90%) of the precursor materials to accretion of Earth and maybe other terrestrial planets (Amsellem et al., 2017). However, the mass-dependent Ca isotope composition of BSE has been revised to δ^44/40Ca = 0.94 ± 0.05‰, through investigating the fertile spinel and garnet peridotites that experienced little or no melting and metasomatism (Kang et al., 2017). This updated δ^44/40Ca data of BSE only overlaps the edge of the average δ^44/40Ca value CV chondrules (1.11 ± 0.14‰; 2SE, N = 12; Amsellem et al., 2017; Bermingham et al., 2018). In fact, this comparison is not valid, because the nucleosynthetic anomalies of multiple metal isotopes, for example, Ca (Dauphas et al., 2014), Ni (Steele et al., 2012), Cr (Trinquier et al., 2007), Ti (Trinquier et al., 2009), Mo (Budde et al., 2019), suggests that CV chondrites accreted in the outer solar system.

OCs have closer nucleosynthetic isotope compositions to the inner solar system bodies (Budde et al., 2019; Dauphas et al., 2014; Steele et al., 2012; Trinquier et al., 2007, 2009), so comparison of Ca isotopes between OC chondrules (Schiller et al., 2018) and Earth, Mars and Vesta should be more reasonable to test the chondrule-rich model. The δ^44/40Ca values for OC chondrules vary from ∼0.05‰ to ∼0.72‰ (Figure 2) that are obviously lower than those of Earth, Vesta, and Mars. Hence, from mass-dependent Ca isotope perspective, it does not support the pebble accretion model, which predicts that chondrules may contribute to the accretion of terrestrial planets (Johansen et al., 2021), although some OC chondrules have overlapping mass-independent ⁴⁸Ca/⁴⁴Ca compositions as Vesta (Schiller et al., 2018). As for Earth, ECs have similar mass-independent isotope compositions as Earth for multiple elements (Dauphas, 2017), and are considered as the possible precursor materials for Earth. Therefore, the δ^44/40Ca data for EC chondrules will be useful to discuss the chondrule-Earth model (e.g., K. Zhu et al., 2020).