Unravelling lunar mantle source processes via the Ti isotope composition of lunar basalts
Affiliations | Corresponding Author | Cite as | Funding information
Published by the European Association of Geochemistry
under Creative Commons License CC BY-NC-ND 4.0
Keywords: titanium isotopes, stable isotopes, Moon, Mare basalt
- Share this article
Article views:1,991Cumulative count of HTML views and PDF downloads.
- Download Citation
- Rights & Permissions
top
Abstract
Figures and Tables
 Figure 1 Measured Ti isotope compositions of lunar samples. All uncertainties unless stated otherwise are 95 % confidence interval (c.i.) of at least 6 measurements of the same aliquot. See main text and Supplementary Information for details. bce = breccia, ilm = ilmenite; QNB = quartz-normative basalt, olv = olivine, pgt = pigeonite. |  Table 1 Summary of reference materials and measured samples. For reference materials, n gives the number of sequences in which the material has been measured at least 6 times. Abbreviations are the same as in Figure 1. Two aliquots of OL-Ti were ran through the chemical separation process and are given as OL-Ti (chemistry). |  Figure 2 Plots of (a) δ49Ti vs. MgO/TiO2, (b) δ49Ti vs. Ta/Hf and (c) δ49Ti vs. U that can be used to discriminate between processes leading to variations in δ49Ti. Symbols are the same as in Figure 1, black arrows indicate the direction in which a process would influence the values along y- and x-axes. Partial melting assumes the presence of ilmenite in the source, fractional crystallisation always implies fractional crystallisation of ilmenite. AFC implies the assimilation of an IBC component during ilmenite-free fractional crystallisation of a low-Ti magma. (b) The estimated Ta/Hf value of urKREEP is ~0.11 (Warren and Taylor, 2014). (c) Contamination by urKREEP (coloured arrow), would imply higher U contents (not observed). urKREEP is based on our first model (see Supplementary Information). |
| Figure 1 | Table 1 | Figure 2 |
top
Introduction
It is widely accepted that the Moon formed as the result of an impact between one or more planetesimals and the proto-Earth (e.g., Asphaug, 2014
Asphaug, E. (2014) Impact Origin of the Moon? Annual Review of Earth and Planetary Sciences 42, 551–578.
). The last phases to solidify in the impact-induced Lunar Magma Ocean (LMO) were K-, REE-, and P-rich residual components (urKREEP) and complementarily an ilmenite-bearing cumulate (IBC; Warren and Wasson, 1979Warren, P.H., Wasson, J.T. (1979) The origin of KREEP. Reviews of Geophysics 17, 73.
; Snyder et al., 1992Snyder, G.A., Taylor, L.A., Neal, C.R. (1992) A chemical model for generating the sources of mare basalts: Combined equilibrium and fractional crystallization of the lunar magmasphere. Geochimica et Cosmochimica Acta 56, 3809–3823.
). Subsequent magmatic processes concluded with the eruption of lunar mare basalts (e.g., Gross and Joy, 2016Gross, J., Joy, K.H. (2016) Evolution, Lunar: From Magma Ocean to Crust Formation. In: Cudnik, B. (ed.) Encyclopedia of Lunar Science. Springer International Publishing, Cham, 1–20.
). Experimental studies and the Hf and Nd isotope composition of mare basalts suggest that low-Ti mare basalts likely result from partial melting of lunar mafic cumulates, whereas high-Ti mare basalts are thought to reflect IBC involvement (e.g., Longhi, 1992Longhi, J. (1992) Experimental petrology and petrogenesis of mare volcanics. Geochimica et Cosmochimica Acta 56, 2235–2251.
; Sprung et al., 2013Sprung, P., Kleine, T., Scherer, E.E. (2013) Isotopic evidence for chondritic Lu/Hf and Sm/Nd of the Moon. Earth and Planetary Science Letters 380, 77–87.
). However, whether high-Ti basalts result from IBC assimilation by mafic low-Ti magmas (e.g., Münker, 2010Münker, C. (2010) A high field strength element perspective on early lunar differentiation. Geochimica et Cosmochimica Acta 74, 7340–7361.
), or from the partial melting of a hybridised lunar mantle source (Snyder et al., 1992Snyder, G.A., Taylor, L.A., Neal, C.R. (1992) A chemical model for generating the sources of mare basalts: Combined equilibrium and fractional crystallization of the lunar magmasphere. Geochimica et Cosmochimica Acta 56, 3809–3823.
), remains ambiguous. Titanium isotope variations in mare basalts may be used to discriminate between the two scenarios: refractory, lithophile, and fluid-immobile Ti is predominantly tetravalent albeit lunar samples may contain significant amounts of Ti3+ (Simon and Sutton, 2017Simon, S.B., Sutton, S.R. (2017) Valence of Ti, V, and Cr in Apollo 14 aluminous basalts 14053 and 14072. Meteoritics & Planetary Science 52, 2051–2066.
; Leitzke et al., 2018Leitzke, F.P., Fonseca, R.O.C., Göttlicher, J., Steininger, R., Jahn, S., Prescher, C., Lagos, M. (2018) Ti K-edge XANES study on the coordination number and oxidation state of Titanium in pyroxene, olivine, armalcolite, ilmenite, and silicate glass during mare basalt petrogenesis. Contributions to Mineralogy and Petrology 173, 103.
). The principal Ti-bearing phase in the lunar mantle is ilmenite whose VI fold coordinated crystal site preferentially incorporates lighter over heavier Ti isotopes (Schauble, 2004Schauble, E.A. (2004) Applying Stable Isotope Fractionation Theory to New Systems. Reviews in Mineralogy and Geochemistry 55, 65–111.
). The Ti isotope composition of terrestrial samples (given as δ(49Ti/47Ti)OL-Ti, relative to the Origins Lab reference material, henceforth δ49Ti; Millet and Dauphas, 2014Millet, M.-A., Dauphas, N. (2014) Ultra-precise titanium stable isotope measurements by double-spike high resolution MC-ICP-MS. Journal of Analytical Atomic Spectrometry 29, 1444.
) ranges between -0.046 and +1.8 ‰. Observed covariations between δ49Ti and SiO2, TiO2, and FeO contents of terrestrial volcanic rocks suggest that stable Ti isotope variation is mainly driven by the onset of fractional crystallisation of Fe-Ti oxides during magmatic differentiation (Millet et al., 2016Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
; Greber et al., 2017aGreber, N.D., Dauphas, N., Bekker, A., Ptáček, M.P., Bindeman, I.N., Hofmann, A. (2017a) Titanium isotopic evidence for felsic crust and plate tectonics 3.5 billion years ago. Science 357, 1271–1274.
,bGreber, N.D., Dauphas, N., Puchtel, I.S., Hofmann, B.A., Arndt, N.T. (2017b) Titanium stable isotopic variations in chondrites, achondrites and lunar rocks. Geochimica et Cosmochimica Acta 213, 534–552.
; Deng et al., 2018aDeng, Z., Moynier, F., Sossi, P.A., Chaussidon, M. (2018a) Bridging the depleted MORB mantle and the continental crust using titanium isotopes. Geochemical Perspectives Letters 9, 11–15.
, 2019Deng, Z., Chaussidon, M., Savage, P., Robert, F., Pik, R., Moynier, F. (2019) Titanium isotopes as a tracer for the plume or island arc affinity of felsic rocks. Proceedings of the National Academy of Sciences 201809164.
; Mandl et al., 2018Mandl, M.B., Fehr, M.A., Schönbächler, M. (2018) Titanium stable isotope fractionation on the Moon: Evidence for inter- mineral isotopic fractionation. Goldschmidt Abstracts 2018 1666.
). Consequently, the Ti isotope composition of lunar magmas might be a sensitive indicator for the crystallisation of the IBC, and its assimilation or partial melting: Millet et al. (2016)Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
reported δ49Ti variations in three low-Ti mare basalts with a pooled δ49Ti of -0.008 ± 0.019 ‰ and five high-Ti mare basalts with δ49Ti between +0.011 and +0.033 ‰. The presence of more fractionated δ49Ti in mare basalts, although suggested, has not yet been reported (Millet et al., 2016Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
). Indeed, for the urKREEP end member, as represented by the lunar meteorite Sayh al Uhaymir (SaU) 169, a tentative δ49Ti of +0.330 ± 0.034 ‰ was obtained (Greber et al., 2017bGreber, N.D., Dauphas, N., Puchtel, I.S., Hofmann, B.A., Arndt, N.T. (2017b) Titanium stable isotopic variations in chondrites, achondrites and lunar rocks. Geochimica et Cosmochimica Acta 213, 534–552.
).Here we report the δ49Ti of 24 representative lunar samples in order to investigate their mantle sources in the context of three major end members: the ambient lunar mantle (low-Ti), the late stage cumulates (IBC, high-Ti), and the residual KREEP-rich component. Notably, the processes that affect δ49Ti in lunar mantle cumulates and corresponding melts also fractionate high field strength element ratios (HFSE; Münker, 2010
Münker, C. (2010) A high field strength element perspective on early lunar differentiation. Geochimica et Cosmochimica Acta 74, 7340–7361.
). High precision HFSE, W, U, and Th data, obtained for the same samples (Thiemens et al., 2019Thiemens, M.M., Sprung, P., Fonseca, R.O.C., Leitzke, F.P., Münker, C. (2019) Early Moon formation inferred from hafnium–tungsten systematics. Nature Geoscience 12, 696–700.
), can help to constrain δ49TiurKREEP, δ49TiIBC and to identify the processes leading to δ49Ti variations in lunar samples.top
Results
Titanium isotope measurements were performed using the Thermo Neptune Plus MC-ICPMS at the University of Cologne with an intermediate precision better than ±0.023 ‰ (2 x standard deviation, henceforth s.d.) for spiked reference materials BCR-2, JB-2, OL-Ti, and Col-Ti. Total blank contribution was always less than 10 ng total Ti and is negligible compared to at least 30 µg of processed sample Ti (20 µg for 68115). More detailed information on the analytical protocol are given in the Supplementary Information. Low-Ti mare basalts show small, resolvable variations in δ49Ti between -0.030 and +0.055 ‰ with an average of +0.010 ± 0.015 ‰ (2 s.d.; n = 12, Fig. 1, Table 1). Most high-Ti samples range from +0.009 to +0.047 in δ49Ti (average of +0.026 ± 0.036 ‰, 2 s.d.; n = 7), sample 75035 has a comparatively high δ49Ti value of +0.115 ‰. The high-Ti average (including 75035) yields a δ49Ti of +0.037 ± 0.071 ‰. KREEP-rich lithologies have δ49Ti values between +0.117 and +0.296 ‰.
Figure 1 Measured Ti isotope compositions of lunar samples. All uncertainties unless stated otherwise are 95 % confidence interval (c.i.) of at least 6 measurements of the same aliquot. See main text and Supplementary Information for details. bce = breccia, ilm = ilmenite; QNB = quartz-normative basalt, olv = olivine, pgt = pigeonite.
Table 1 Summary of reference materials and measured samples. For reference materials, n gives the number of sequences in which the material has been measured at least 6 times. Abbreviations are the same as in Figure 1. Two aliquots of OL-Ti were run through the chemical separation process and are given as OL-Ti (chemistry).
| δ49Ti | 2 s.d. | n | 95 % c.i. | |||
| JB-2 | -0.044 | 0.023 | 7 | 0.011 | ||
| BCR-2 | -0.025 | 0.012 | 4 | 0.010 | ||
| OL-Ti mean | 0.000 | 0.001 | 8 | 0.000 | ||
| OL-Ti (chemistry) | -0.001 | 0.030 | 2 | 0.137 | ||
| Col-Ti mean | 0.206 | 0.021 | 8 | 0.009 | ||
| Low-Ti rocks | ||||||
| 12022 | Apollo 12 | Low-Ti ilm basalt | 0.029 | 0.022 | 6 | 0.011 |
| 12051 | " | Low-Ti ilm basalt | 0.030 | 0.022 | 8 | 0.009 |
| 12063 | " | Low-Ti ilm basalt | 0.055 | 0.022 | 6 | 0.012 |
| 12054 | " | Low-Ti ilm basalt | 0.028 | 0.026 | 6 | 0.014 |
| 12004 | " | Low-Ti olv basalt | 0.006 | 0.010 | 6 | 0.005 |
| 12053 | " | Low-Ti pgt basalt | -0.013 | 0.026 | 6 | 0.014 |
| 15495 | Apollo 15 | Low-Ti ONB | 0.011 | 0.017 | 6 | 0.009 |
| 15555 | " | Low-Ti ONB | -0.008 | 0.026 | 6 | 0.014 |
| 15556 | " | Low-Ti ONB | 0.007 | 0.030 | 6 | 0.016 |
| 15065 | " | Low-Ti QNB | -0.030 | 0.014 | 6 | 0.007 |
| 15058 | " | Low-Ti QNB | -0.010 | 0.034 | 6 | 0.018 |
| 15545 | " | Low-Ti QNB | 0.011 | 0.032 | 6 | 0.017 |
| Low-Ti mean ± 2 s.d. | 0.010 | 0.047 | 12 | |||
| High-Ti rocks | ||||||
| 10017 | Apollo 11 | High-Ti ilm basalt | 0.009 | 0.024 | 6 | 0.012 |
| 10020 | " | High-Ti ilm basalt | 0.011 | 0.022 | 6 | 0.011 |
| 10057 | " | High-Ti ilm basalt | 0.009 | 0.024 | 6 | 0.012 |
| 74255 | Apollo 17 | High-Ti ilm basalt | 0.043 | 0.010 | 6 | 0.005 |
| 74275 | " | High-Ti ilm basalt | 0.045 | 0.010 | 6 | 0.005 |
| 75035 | " | High-Ti ilm basalt | 0.115 | 0.025 | 6 | 0.013 |
| 79135 | " | High-Ti bce | 0.019 | 0.017 | 6 | 0.009 |
| 79035 | " | High-Ti bce | 0.047 | 0.018 | 6 | 0.010 |
| High-Ti mean 1 ± 2 s.d. | 0.026 | 0.036 | 7 | |||
| High-Ti mean 2 ± 2 s.d. | 0.037 | 0.071 | 8 | |||
| KREEP-rich rocks | ||||||
| 14305 | Apollo 14 | KREEP-rich bce | 0.296 | 0.030 | 6 | 0.016 |
| 14310 | " | KREEP basalt | 0.263 | 0.035 | 6 | 0.018 |
| 72275 | Apollo 17 | KREEP-rich bce | 0.185 | 0.026 | 6 | 0.014 |
| 68115 | Apollo 16 | KREEP-rich bce | 0.117 | 0.027 | 6 | 0.014 |
top
Discussion
Estimating the δ49Ti of urKREEP and the IBC. The co-genetic relationship between urKREEP and IBC permits studying the coupled δ49Ti evolution of both reservoirs. Complementarily to the low δ49TiIBC, the residual LMO is expected to have positive δ49Ti (Millet et al., 2016
Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
; Greber et al., 2017bGreber, N.D., Dauphas, N., Puchtel, I.S., Hofmann, B.A., Arndt, N.T. (2017b) Titanium stable isotopic variations in chondrites, achondrites and lunar rocks. Geochimica et Cosmochimica Acta 213, 534–552.
). Furthermore, U became enriched in urKREEP, whereas Ti, Hf and Zr are more compatible in ilmenite and are extracted from the LMO during IBC formation. The observed positive co-variation of δ49Ti and U concentration in KREEP-rich samples may therefore indicate variable portions of the urKREEP-component (Fig. 2b). Our KREEP-rich sample with highest δ49Ti and highest U concentration is identical within uncertainty to the previous estimate for δ49TiurKREEP (Greber et al., 2017bGreber, N.D., Dauphas, N., Puchtel, I.S., Hofmann, B.A., Arndt, N.T. (2017b) Titanium stable isotopic variations in chondrites, achondrites and lunar rocks. Geochimica et Cosmochimica Acta 213, 534–552.
). Two KREEP-rich samples along with SaU169 have U/Hf, U/Zr and U/Ti in the range of the urKREEP estimate (calculated using data by Warren and Taylor, 2014Warren, P.H., Taylor, G.J. (2014) The Moon. Treatise on Geochemistry (Second Edition) 2, 213–250.
), which results in a conservative estimate of δ49TiurKREEP of +0.296 ± 0.067 ‰ (see Supplementary Information). To model δ49TiurKREEP in a different approach, we constrained the evolution of Ti, Hf and Ta concentrations in the remaining liquid during fractional crystallisation for various LMO solidification models (Snyder et al., 1992Snyder, G.A., Taylor, L.A., Neal, C.R. (1992) A chemical model for generating the sources of mare basalts: Combined equilibrium and fractional crystallization of the lunar magmasphere. Geochimica et Cosmochimica Acta 56, 3809–3823.
; Lin et al., 2017Lin, Y., Tronche, E.J., Steenstra, E.S., van Westrenen, W. (2017) Experimental constraints on the solidification of a nominally dry lunar magma ocean. Earth and Planetary Science Letters 471, 104–116.
; Charlier et al., 2018Charlier, B., Grove, T.L., Namur, O., Holtz, F. (2018) Crystallization of the lunar magma ocean and the primordial mantle-crust differentiation of the Moon. Geochimica et Cosmochimica Acta 234, 50–69.
; Rapp and Draper, 2018Rapp, J.F., Draper, D.S. (2018) Fractional crystallization of the lunar magma ocean: Updating the dominant paradigm. Meteoritics & Planetary Science 53, 1432–1455.
) starting with LMO element abundances by Münker (2010)Münker, C. (2010) A high field strength element perspective on early lunar differentiation. Geochimica et Cosmochimica Acta 74, 7340–7361.
and calculated δ49Timelt using a Rayleigh distillation model (see Supplementary Information). The intersect of the respective Ti, Hf and Ta concentrations with urKREEP concentration estimates by Warren and Taylor (2014)Warren, P.H., Taylor, G.J. (2014) The Moon. Treatise on Geochemistry (Second Edition) 2, 213–250.
together with the modelled δ49Timelt-evolution line yield a range for δ49TiurKREEP: all estimates using the above mentioned models and LMO concentrations of Ta (δ49TiurKREEP = +0.308 ± 0.064 ‰), Hf (δ49TiurKREEP = +0.290 ± 0.040 ‰) and Ti (δ49TiurKREEP = +0.238 ± 0.085 ‰) are identical to our KREEP-rich sample with highest δ49Ti (+0.296 ± 0.016 ‰), and previous estimates for δ49TiurKREEP (Greber et al., 2017bGreber, N.D., Dauphas, N., Puchtel, I.S., Hofmann, B.A., Arndt, N.T. (2017b) Titanium stable isotopic variations in chondrites, achondrites and lunar rocks. Geochimica et Cosmochimica Acta 213, 534–552.
). Our models predict a final δ49TiIBC between -0.006 and -0.012 ‰, consistent with estimates by Millet et al. (2016)Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
.Identifying magmatic processes in lunar basalt sources. The average δ49Ti values obtained for low- and high-Ti mare basalts (+0.010 ± 0.047 ‰ and +0.037 ± 0.071 ‰) are consistent with findings by Millet et al. (2016)
Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
and are indistinguishable from their Bulk Silicate Earth estimate (BSE: +0.005 ± 0.005 ‰, 95 % c.i.; see Fig. 1). The δ49Ti value of our most primitive sample, the Apollo 12 olivine basalt 12004 (possibly tapping bulk lunar mantle) is identical with the δ49TiBSE estimate by Millet et al. (2016)Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
and the chondritic δ49Ti value proposed by Greber et al. (2017b)Greber, N.D., Dauphas, N., Puchtel, I.S., Hofmann, B.A., Arndt, N.T. (2017b) Titanium stable isotopic variations in chondrites, achondrites and lunar rocks. Geochimica et Cosmochimica Acta 213, 534–552.
but lower than the chondritic δ49Ti value suggested by Deng et al. (2018b)Deng, Z., Moynier, F., van Zuilen, K., Sossi, P.A., Pringle, E.A., Chaussidon, M. (2018b) Lack of resolvable titanium stable isotopic variations in bulk chondrites. Geochimica et Cosmochimica Acta 239, 409–419.
. Possible explanations for the observed intra-group δ49Ti variations in mare basalts include fractional crystallisation of ilmenite during petrogenesis, assimilation of an IBC- or KREEP-component during fractional crystallisation, partial melting an IBC or mantle source heterogeneity. To distinguish between these processes we first consider the results of experimental studies that simulated the petrogenesis of lunar basalts (see Supplementary Information; Longhi, 1992Longhi, J. (1992) Experimental petrology and petrogenesis of mare volcanics. Geochimica et Cosmochimica Acta 56, 2235–2251.
and references therein): For instance, experimentally synthesised low-Ti-like samples showed that ilmenite has been one of the first solidus phases (Longhi, 1992Longhi, J. (1992) Experimental petrology and petrogenesis of mare volcanics. Geochimica et Cosmochimica Acta 56, 2235–2251.
and references therein), which would preferentially incorporate light Ti isotopes, increasing the δ49Timelt during fractional crystallisation of ilmenite (Millet et al., 2016Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
). High-Ti basalts from Apollo 17 were shown to originate from greater depths than Apollo 12 and 15 low-Ti mare basalts were successfully modelled by partial melting of an ilmenite-rich source (Longhi, 1992Longhi, J. (1992) Experimental petrology and petrogenesis of mare volcanics. Geochimica et Cosmochimica Acta 56, 2235–2251.
and references therein). This can explain the observed elevated δ49Tihigh-Ti values (See Fig. 2a; Millet et al., 2016Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
. In contrast, the full assimilation of an IBC-component during fractional crystallisation of a low-Ti magma would decrease the δ49Ti value of the basaltic melt (Fig. 2a). Thus, the small intra-group variations in δ49Ti values of low- and high-Ti basalts can best be explained by fractional crystallisation of ilmenite and melting of ilmenite-bearing sources, respectively.In addition to experimental petrological studies, geochemical tools can help distinguish between the discussed processes. For example, the compatibility of HFSEs in ilmenite is variable (Münker, 2010
Münker, C. (2010) A high field strength element perspective on early lunar differentiation. Geochimica et Cosmochimica Acta 74, 7340–7361.
): tantalum is the most compatible HFSE in ilmenite, whereas Hf is the most incompatible (e.g., DTa > 1 > DHf; Leitzke et al., 2016Leitzke, F.P., Fonseca, R.O.C., Michely, L.T., Sprung, P., Münker, C., Heuser, A., Blanchard, H. (2016) The effect of titanium on the partitioning behavior of high-field strength elements between silicates, oxides and lunar basaltic melts with applications to the origin of mare basalts. Chemical Geology 440, 219–238.
; or DTa > DHf, van Kan Parker et al., 2011van Kan Parker, M., Mason, P.R.D., van Westrenen, W. (2011) Trace element partitioning between ilmenite, armalcolite and anhydrous silicate melt: Implications for the formation of lunar high-Ti mare basalts. Geochimica et Cosmochimica Acta 75, 4179–4193.
). Thus, fractional crystallisation of ilmenite would increase the δ49Timelt (Millet et al., 2016Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
) but decrease the Ta/Hfmelt (Münker, 2010Münker, C. (2010) A high field strength element perspective on early lunar differentiation. Geochimica et Cosmochimica Acta 74, 7340–7361.
). In contrast, partial melting of an IBC component would increase the Ta/Hfmelt and the δ49Timelt value (DTa ~ 2 x DHf for ilmenite, e.g., Leitzke et al., 2016Leitzke, F.P., Fonseca, R.O.C., Michely, L.T., Sprung, P., Münker, C., Heuser, A., Blanchard, H. (2016) The effect of titanium on the partitioning behavior of high-field strength elements between silicates, oxides and lunar basaltic melts with applications to the origin of mare basalts. Chemical Geology 440, 219–238.
). Apollo 12 low-Ti ilmenite basalts show lower Ta/Hf and positive δ49Ti relative to the most primitive lunar basalt analysed in this study, (12004, see Figs. 1, 2b), indicating the early fractional crystallisation of ilmenite, which is consistent with experimental results (Longhi, 1992Longhi, J. (1992) Experimental petrology and petrogenesis of mare volcanics. Geochimica et Cosmochimica Acta 56, 2235–2251.
and references therein). Relatively low δ49Ti values in Apollo 15 low-Ti basalts may indicate the assimilation of an IBC material (Longhi, 1992Longhi, J. (1992) Experimental petrology and petrogenesis of mare volcanics. Geochimica et Cosmochimica Acta 56, 2235–2251.
and references therein). However, assimilation of an IBC component seems inconsistent with relatively low Ta/Hf and comparatively low TiO2 contents of ~1.3 %. The assimilation of a KREEP-component would cause a substantial increase in the δ49Ti and U concentration of the melt while barely fractionating Ta/Hf. Such trends are not observed indicating that KREEP components played an insignificant role during the petrogenesis of low- and high-Ti basalts (Fig. 2c). Apollo 17 high-Ti samples are rich in TiO2, exhibit relatively low MgO/TiO2 but high δ49Ti and higher Ta/Hf relative to our most primitive sample. In agreement with the experimental evidence, this geochemical pattern strongly indicates partial melting of an IBC component during the petrogenesis of Apollo 17 high-Ti basalts (Longhi, 1992Longhi, J. (1992) Experimental petrology and petrogenesis of mare volcanics. Geochimica et Cosmochimica Acta 56, 2235–2251.
and references therein). These processes are further modelled and discussed in the Supplementary Information.Figure 2 Plots of (a) δ49Ti vs. MgO/TiO2, (b) δ49Ti vs. Ta/Hf and (c) δ49Ti vs. U that can be used to discriminate between processes leading to variations in δ49Ti. Symbols are the same as in Figure 1, black arrows indicate the direction in which a process would influence the values along y- and x-axes. Partial melting assumes the presence of ilmenite in the source, fractional crystallisation always implies fractional crystallisation of ilmenite. AFC implies the assimilation of an IBC component during ilmenite-free fractional crystallisation of a low-Ti magma. (b) The estimated Ta/Hf value of urKREEP is ~0.11 (Warren and Taylor, 2014
Warren, P.H., Taylor, G.J. (2014) The Moon. Treatise on Geochemistry (Second Edition) 2, 213–250.
). (c) Contamination by urKREEP (coloured arrow), would imply higher U contents (not observed). urKREEP is based on our first model (see Supplementary Information).Comparison with Fe isotope systematics. Previous studies observed a bimodal distribution of δ57Fe values for lunar low-Ti and high-Ti basalts: higher δ57Fe values in high-Ti basalts were attributed to the presence of ilmenite during partial melting (Sossi and Moynier, 2017
Sossi, P.A., Moynier, F. (2017) Chemical and isotopic kinship of iron in the Earth and Moon deduced from the lunar Mg-Suite. Earth and Planetary Science Letters 471, 125–135.
; Sossi and O’Neill, 2017Sossi, P.A., O’Neill, H.St.C. (2017) The effect of bonding environment on iron isotope fractionation between minerals at high temperature. Geochimica et Cosmochimica Acta 196, 121–143.
), which is consistent with experimental studies (Longhi, 1992Longhi, J. (1992) Experimental petrology and petrogenesis of mare volcanics. Geochimica et Cosmochimica Acta 56, 2235–2251.
and references therein) as well as our Ta/Hf and δ49Ti data. However, intra-group variation in δ57Fe of low-Ti basalts was not observed (Weyer et al., 2005Weyer, S., Anbar, A., Brey, G., Munker, C., Mezger, K., Woodland, A. (2005) Iron isotope fractionation during planetary differentiation. Earth and Planetary Science Letters 240, 251–264.
; Sossi and Moynier, 2017Sossi, P.A., Moynier, F. (2017) Chemical and isotopic kinship of iron in the Earth and Moon deduced from the lunar Mg-Suite. Earth and Planetary Science Letters 471, 125–135.
; Poitrasson et al., 2019Poitrasson, F., Zambardi, T., Magna, T., Neal, C.R. (2019) A reassessment of the iron isotope composition of the Moon and its implications for the accretion and differentiation of terrestrial planets. Geochimica et Cosmochimica Acta 267, 257–274.
) as the δ57Fe of low-Ti basalts is mainly controlled by fractional crystallisation of olivine and pyroxene, which would preferentially incorporate light Fe isotopes, similar to ilmenite (Δ57FeIlm-Ol = +0.01 ‰; Sossi and O’Neill, 2017Sossi, P.A., O’Neill, H.St.C. (2017) The effect of bonding environment on iron isotope fractionation between minerals at high temperature. Geochimica et Cosmochimica Acta 196, 121–143.
). Thus, δ49Ti appear more appropriate to discriminate better between the petrogenetic processes culminating in lunar basalts. Additional Cr, V and Hf isotope systematics are discussed in the Supplementary Information.top
Conclusion
New Ti isotope data for a representative set of lunar samples show that there is a clear offset between KREEP-rich samples and mare basalts, in agreement with previous work (Millet et al., 2016
Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
; Greber et al., 2017bGreber, N.D., Dauphas, N., Puchtel, I.S., Hofmann, B.A., Arndt, N.T. (2017b) Titanium stable isotopic variations in chondrites, achondrites and lunar rocks. Geochimica et Cosmochimica Acta 213, 534–552.
). Our data reveal intra-group variation amongst olivine- and quartz-normative and (low-Ti) ilmenite mare basalts with the latter recording higher δ49Ti than the former. High-Ti mare basalts have overall higher δ49Ti. The δ49Ti of our lunar sample suite, coupled with HFSE data, suggests that the fractional crystallisation of ilmenite and partial melting of an IBC-component are the principal processes affecting the δ49Ti of Apollo 12 low-Ti mare basalts, and of Apollo 17 high-Ti mare basalts, respectively. This conclusion is consistent with results of previously published experimental studies (Longhi, 1992Longhi, J. (1992) Experimental petrology and petrogenesis of mare volcanics. Geochimica et Cosmochimica Acta 56, 2235–2251.
and references therein). This is in excellent agreement with previous experimental data and the heavy δ57Fe of high-Ti mare basalts (Longhi, 1992Longhi, J. (1992) Experimental petrology and petrogenesis of mare volcanics. Geochimica et Cosmochimica Acta 56, 2235–2251.
; Weyer et al., 2005Weyer, S., Anbar, A., Brey, G., Munker, C., Mezger, K., Woodland, A. (2005) Iron isotope fractionation during planetary differentiation. Earth and Planetary Science Letters 240, 251–264.
; Sossi and Moynier, 2017Sossi, P.A., Moynier, F. (2017) Chemical and isotopic kinship of iron in the Earth and Moon deduced from the lunar Mg-Suite. Earth and Planetary Science Letters 471, 125–135.
; Charlier et al., 2018Charlier, B., Grove, T.L., Namur, O., Holtz, F. (2018) Crystallization of the lunar magma ocean and the primordial mantle-crust differentiation of the Moon. Geochimica et Cosmochimica Acta 234, 50–69.
). Based on coupled HFSE and δ49Ti data, the petrogenesis of high-Ti mare basalts by assimilation of IBC-component by low-Ti magma is unlikely.top
Acknowledgements
We thank CAPTEM for providing lunar samples and Horst Marschall for handling the manuscript. The authors would like to thank Marc-Alban Millet and Nicolas Greber for sharing the OL-Ti reference material, Wim van Westrenen and one anonymous reviewer for constructive comments that greatly improved the manuscript. ROCF is grateful for funding of a Heisenberg Professorship by the Deutsche Forschungsgemeinschaft (DFG grants FO 698/6-1 and FO 698/11-1). CM acknowledges funding by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No. 669666). SK was partially funded through a UoC Advanced Post Doc grant within the Excellence Initiative to PS and acknowledges the UoC Graduate School of Geosciences for providing a fellowship (GSGS-2019X-07).
Editor: Horst R. Marschall
top
References
Asphaug, E. (2014) Impact Origin of the Moon? Annual Review of Earth and Planetary Sciences 42, 551–578.
Show in contextIt is widely accepted that the Moon formed as the result of an impact between one or more planetesimals and the proto-Earth (e.g., Asphaug, 2014).
View in article
Charlier, B., Grove, T.L., Namur, O., Holtz, F. (2018) Crystallization of the lunar magma ocean and the primordial mantle-crust differentiation of the Moon. Geochimica et Cosmochimica Acta 234, 50–69.
Show in context To model δ49TiurKREEP in a different approach, we constrained the evolution of Ti, Hf and Ta concentrations in the remaining liquid during fractional crystallisation for various LMO solidification models (Snyder et al., 1992; Lin et al., 2017; Charlier et al., 2018; Rapp and Draper, 2018) starting with LMO element abundances by Münker (2010) and calculated δ49Timelt using a Rayleigh distillation model (see Supplementary Information).
View in article
This is in excellent agreement with previous experimental data and the heavy δ57Fe of high-Ti mare basalts (Longhi, 1992; Weyer et al., 2005; Sossi and Moynier, 2017; Charlier et al., 2018).
View in article
Deng, Z., Chaussidon, M., Savage, P., Robert, F., Pik, R., Moynier, F. (2019) Titanium isotopes as a tracer for the plume or island arc affinity of felsic rocks. Proceedings of the National Academy of Sciences 201809164.
Show in context Observed covariations between δ49Ti and SiO2, TiO2, and FeO contents of terrestrial volcanic rocks suggest that stable Ti isotope variation is mainly driven by the onset of fractional crystallisation of Fe-Ti oxides during magmatic differentiation (Millet et al., 2016; Greber et al., 2017a,b; Deng et al., 2018a, 2019; Mandl et al., 2018).
View in article
Deng, Z., Moynier, F., Sossi, P.A., Chaussidon, M. (2018a) Bridging the depleted MORB mantle and the continental crust using titanium isotopes. Geochemical Perspectives Letters 9, 11–15.
Show in contextObserved covariations between δ49Ti and SiO2, TiO2, and FeO contents of terrestrial volcanic rocks suggest that stable Ti isotope variation is mainly driven by the onset of fractional crystallisation of Fe-Ti oxides during magmatic differentiation (Millet et al., 2016; Greber et al., 2017a,b; Deng et al., 2018a, 2019; Mandl et al., 2018).
View in article
Deng, Z., Moynier, F., van Zuilen, K., Sossi, P.A., Pringle, E.A., Chaussidon, M. (2018b) Lack of resolvable titanium stable isotopic variations in bulk chondrites. Geochimica et Cosmochimica Acta 239, 409–419.
Show in context The δ49Ti value of our most primitive sample, the Apollo 12 olivine basalt 12004 (possibly tapping bulk lunar mantle) is identical with the δ49TiBSE estimate by Millet et al. (2016) and the chondritic δ49Ti value proposed by Greber et al. (2017b) but lower than the chondritic δ49Ti value suggested by Deng et al. (2018b).
View in article
Greber, N.D., Dauphas, N., Bekker, A., Ptáček, M.P., Bindeman, I.N., Hofmann, A. (2017a) Titanium isotopic evidence for felsic crust and plate tectonics 3.5 billion years ago. Science 357, 1271–1274.
Show in contextObserved covariations between δ49Ti and SiO2, TiO2, and FeO contents of terrestrial volcanic rocks suggest that stable Ti isotope variation is mainly driven by the onset of fractional crystallisation of Fe-Ti oxides during magmatic differentiation (Millet et al., 2016; Greber et al., 2017a,b; Deng et al., 2018a, 2019; Mandl et al., 2018).
View in article
Greber, N.D., Dauphas, N., Puchtel, I.S., Hofmann, B.A., Arndt, N.T. (2017b) Titanium stable isotopic variations in chondrites, achondrites and lunar rocks. Geochimica et Cosmochimica Acta 213, 534–552.
Show in contextObserved covariations between δ49Ti and SiO2, TiO2, and FeO contents of terrestrial volcanic rocks suggest that stable Ti isotope variation is mainly driven by the onset of fractional crystallisation of Fe-Ti oxides during magmatic differentiation (Millet et al., 2016; Greber et al., 2017a,b; Deng et al., 2018a, 2019; Mandl et al., 2018).
View in article
Indeed, for the urKREEP end member, as represented by the lunar meteorite Sayh al Uhaymir (SaU) 169, a tentative δ49Ti of +0.330 ± 0.034 ‰ was obtained (Greber et al., 2017b).
View in article
Complementarily to the low δ49TiIBC, the residual LMO is expected to have positive δ49Ti (Millet et al., 2016; Greber et al., 2017b).
View in article
Our KREEP-rich sample with highest δ49Ti and highest U concentration is identical within uncertainty to the previous estimate for δ49TiurKREEP (Greber et al., 2017b).
View in article
The intersect of the respective Ti, Hf and Ta concentrations with urKREEP concentration estimates by Warren and Taylor (2014) together with the modelled δ49Timelt-evolution line yield a range for δ49TiurKREEP: all estimates using the above mentioned models and LMO concentrations of Ta (δ49TiurKREEP = +0.308 ± 0.064 ‰), Hf (δ49TiurKREEP = +0.290 ± 0.040 ‰) and Ti (δ49TiurKREEP = +0.238 ± 0.085 ‰) are identical to our KREEP-rich sample with highest δ49Ti (+0.296 ± 0.016 ‰), and previous estimates for δ49TiurKREEP (Greber et al., 2017b).
View in article
The δ49Ti value of our most primitive sample, the Apollo 12 olivine basalt 12004 (possibly tapping bulk lunar mantle) is identical with the δ49TiBSE estimate by Millet et al. (2016) and the chondritic δ49Ti value proposed by Greber et al. (2017b) but lower than the chondritic δ49Ti value suggested by Deng et al. (2018b).
View in article
New Ti isotope data for a representative set of lunar samples show that there is a clear offset between KREEP-rich samples and mare basalts, in agreement with previous work (Millet et al., 2016; Greber et al., 2017b).
View in article
Gross, J., Joy, K.H. (2016) Evolution, Lunar: From Magma Ocean to Crust Formation. In: Cudnik, B. (ed.) Encyclopedia of Lunar Science. Springer International Publishing, Cham, 1–20.
Show in context Subsequent magmatic processes concluded with the eruption of lunar mare basalts (e.g., Gross and Joy, 2016).
View in article
Leitzke, F.P., Fonseca, R.O.C., Michely, L.T., Sprung, P., Münker, C., Heuser, A., Blanchard, H. (2016) The effect of titanium on the partitioning behavior of high-field strength elements between silicates, oxides and lunar basaltic melts with applications to the origin of mare basalts. Chemical Geology 440, 219–238.
Show in context For example, the compatibility of HFSEs in ilmenite is variable (Münker, 2010): tantalum is the most compatible HFSE in ilmenite, whereas Hf is the most incompatible (e.g., DTa > 1 > DHf; Leitzke et al., 2016; or DTa > DHf, van Kan Parker et al., 2011).
View in article
In contrast, partial melting of an IBC component would increase the Ta/Hfmelt and the δ49Timelt value (DTa ~ 2 x DHf for ilmenite, e.g., Leitzke et al., 2016).
View in article
Leitzke, F.P., Fonseca, R.O.C., Göttlicher, J., Steininger, R., Jahn, S., Prescher, C., Lagos, M. (2018) Ti K-edge XANES study on the coordination number and oxidation state of Titanium in pyroxene, olivine, armalcolite, ilmenite, and silicate glass during mare basalt petrogenesis. Contributions to Mineralogy and Petrology 173, 103.
Show in context Titanium isotope variations in mare basalts may be used to discriminate between the two scenarios: refractory, lithophile, and fluid-immobile Ti is predominantly tetravalent albeit lunar samples may contain significant amounts of Ti3+ (Simon and Sutton, 2017; Leitzke et al., 2018).
View in article
Lin, Y., Tronche, E.J., Steenstra, E.S., van Westrenen, W. (2017) Experimental constraints on the solidification of a nominally dry lunar magma ocean. Earth and Planetary Science Letters 471, 104–116.
Show in contextTo model δ49TiurKREEP in a different approach, we constrained the evolution of Ti, Hf and Ta concentrations in the remaining liquid during fractional crystallisation for various LMO solidification models (Snyder et al., 1992; Lin et al., 2017; Charlier et al., 2018; Rapp and Draper, 2018) starting with LMO element abundances by Münker (2010) and calculated δ49Timelt using a Rayleigh distillation model (see Supplementary Information).
View in article
Longhi, J. (1992) Experimental petrology and petrogenesis of mare volcanics. Geochimica et Cosmochimica Acta 56, 2235–2251.
Show in context Experimental studies and the Hf and Nd isotope composition of mare basalts suggest that low-Ti mare basalts likely result from partial melting of lunar mafic cumulates, whereas high-Ti mare basalts are thought to reflect IBC involvement (e.g., Longhi, 1992; Sprung et al., 2013).
View in article
To distinguish between these processes we first consider the results of experimental studies that simulated the petrogenesis of lunar basalts (see Supplementary Information; Longhi, 1992 and references therein):
View in article
For instance, experimentally synthesised low-Ti-like samples showed that ilmenite has been one of the first solidus phases (Longhi, 1992 and references therein), which would preferentially incorporate light Ti isotopes, increasing the δ49Timelt during fractional crystallisation of ilmenite (Millet et al., 2016).
View in article
High-Ti basalts from Apollo 17 were shown to originate from greater depths than Apollo 12 and 15 low-Ti mare basalts were successfully modelled by partial melting of an ilmenite-rich source (Longhi, 1992 and references therein).
View in article
Apollo 12 low-Ti ilmenite basalts show lower Ta/Hf and positive δ49Ti relative to the most primitive lunar basalt analysed in this study, (12004, see Figs. 1, 2b), indicating the early fractional crystallisation of ilmenite, which is consistent with experimental results (Longhi, 1992 and references therein).
View in article
Relatively low δ49Ti values in Apollo 15 low-Ti basalts may indicate the assimilation of an IBC material (Longhi, 1992 and references therein).
View in article
In agreement with the experimental evidence, this geochemical pattern strongly indicates partial melting of an IBC component during the petrogenesis of Apollo 17 high-Ti basalts (Longhi, 1992 and references therein).
View in article
Previous studies observed a bimodal distribution of δ57Fe values for lunar low-Ti and high-Ti basalts: higher δ57Fe values in high-Ti basalts were attributed to the presence of ilmenite during partial melting (Sossi and Moynier, 2017; Sossi and O’Neill, 2017), which is consistent with experimental studies (Longhi, 1992 and references therein) as well as our Ta/Hf and δ49Ti data.
View in article
This conclusion is consistent with results of previously published experimental studies (Longhi, 1992 and references therein).
View in article
This is in excellent agreement with previous experimental data and the heavy δ57Fe of high-Ti mare basalts (Longhi, 1992; Weyer et al., 2005; Sossi and Moynier, 2017; Charlier et al., 2018).
View in article
Mandl, M.B., Fehr, M.A., Schönbächler, M. (2018) Titanium stable isotope fractionation on the Moon: Evidence for inter- mineral isotopic fractionation. Goldschmidt Abstracts 2018 1666.
Show in contextObserved covariations between δ49Ti and SiO2, TiO2, and FeO contents of terrestrial volcanic rocks suggest that stable Ti isotope variation is mainly driven by the onset of fractional crystallisation of Fe-Ti oxides during magmatic differentiation (Millet et al., 2016; Greber et al., 2017a,b; Deng et al., 2018a, 2019; Mandl et al., 2018).
View in article
Millet, M.-A., Dauphas, N. (2014) Ultra-precise titanium stable isotope measurements by double-spike high resolution MC-ICP-MS. Journal of Analytical Atomic Spectrometry 29, 1444.
Show in context The Ti isotope composition of terrestrial samples (given as δ(49Ti/47Ti)OL-Ti, relative to the Origins Lab reference material, henceforth δ49Ti; Millet and Dauphas, 2014) ranges between -0.046 and +1.8 ‰.
View in article
Millet, M.-A., Dauphas, N., Greber, N.D., Burton, K.W., Dale, C.W., Debret, B., Macpherson, C.G., Nowell, G.M., Williams, H.M. (2016) Titanium stable isotope investigation of magmatic processes on the Earth and Moon. Earth and Planetary Science Letters 449, 197–205.
Show in context Observed covariations between δ49Ti and SiO2, TiO2, and FeO contents of terrestrial volcanic rocks suggest that stable Ti isotope variation is mainly driven by the onset of fractional crystallisation of Fe-Ti oxides during magmatic differentiation (Millet et al., 2016; Greber et al., 2017a,b; Deng et al., 2018a, 2019; Mandl et al., 2018).
View in article
Consequently, the Ti isotope composition of lunar magmas might be a sensitive indicator for the crystallisation of the IBC, and its assimilation or partial melting: Millet et al. (2016) reported δ49Ti variations in three low-Ti mare basalts with a pooled δ49Ti of -0.008 ± 0.019 ‰ and five high-Ti mare basalts with δ49Ti between +0.011 and +0.033 ‰.
View in article
The presence of more fractionated δ49Ti in mare basalts, although suggested, has not yet been reported (Millet et al., 2016).
View in article
Complementarily to the low δ49TiIBC, the residual LMO is expected to have positive δ49Ti (Millet et al., 2016; Greber et al., 2017b).
View in article
Our models predict a final δ49TiIBC between -0.006 and -0.012 ‰, consistent with estimates by Millet et al. (2016).
View in article
The average δ49Ti values obtained for low- and high-Ti mare basalts (+0.010 ± 0.047 ‰ and +0.037 ± 0.071 ‰) are consistent with findings by Millet et al. (2016) and are indistinguishable from their Bulk Silicate Earth estimate (BSE: +0.005 ± 0.005 ‰, 95 % c.i.; see Fig. 1).
View in article
The δ49Ti value of our most primitive sample, the Apollo 12 olivine basalt 12004 (possibly tapping bulk lunar mantle) is identical with the δ49TiBSE estimate by Millet et al. (2016) and the chondritic δ49Ti value proposed by Greber et al. (2017b) but lower than the chondritic δ49Ti value suggested by Deng et al. (2018b).
View in article
For instance, experimentally synthesised low-Ti-like samples showed that ilmenite has been one of the first solidus phases (Longhi, 1992 and references therein), which would preferentially incorporate light Ti isotopes, increasing the δ49Timelt during fractional crystallisation of ilmenite (Millet et al., 2016).
View in article
This can explain the observed elevated δ49Tihigh-Ti values (See Fig. 2a; Millet et al., 2016).
View in article
Thus, fractional crystallisation of ilmenite would increase the δ49Timelt (Millet et al., 2016) but decrease the Ta/Hfmelt (Münker, 2010).
View in article
New Ti isotope data for a representative set of lunar samples show that there is a clear offset between KREEP-rich samples and mare basalts, in agreement with previous work (Millet et al., 2016; Greber et al., 2017b).
View in article
Münker, C. (2010) A high field strength element perspective on early lunar differentiation. Geochimica et Cosmochimica Acta 74, 7340–7361.
Show in contextHowever, whether high-Ti basalts result from IBC assimilation by mafic low-Ti magmas (e.g., Münker, 2010), or from the partial melting of a hybridised lunar mantle source (Snyder et al., 1992), remains ambiguous.
View in article
Notably, the processes that affect δ49Ti in lunar mantle cumulates and corresponding melts also fractionate high field strength element ratios (HFSE; Münker, 2010).
View in article
To model δ49TiurKREEP in a different approach, we constrained the evolution of Ti, Hf and Ta concentrations in the remaining liquid during fractional crystallisation for various LMO solidification models (Snyder et al., 1992; Lin et al., 2017; Charlier et al., 2018; Rapp and Draper, 2018) starting with LMO element abundances by Münker (2010) and calculated δ49Timelt using a Rayleigh distillation model (see Supplementary Information).
View in article
For example, the compatibility of HFSEs in ilmenite is variable (Münker, 2010): tantalum is the most compatible HFSE in ilmenite, whereas Hf is the most incompatible (e.g., DTa > 1 > DHf; Leitzke et al., 2016; or DTa > DHf, van Kan Parker et al., 2011).
View in article
Thus, fractional crystallisation of ilmenite would increase the δ49Timelt (Millet et al., 2016) but decrease the Ta/Hfmelt (Münker, 2010).
View in article
Poitrasson, F., Zambardi, T., Magna, T., Neal, C.R. (2019) A reassessment of the iron isotope composition of the Moon and its implications for the accretion and differentiation of terrestrial planets. Geochimica et Cosmochimica Acta 267, 257–274.
Show in context However, intra-group variation in δ57Fe of low-Ti basalts was not observed (Weyer et al., 2005; Sossi and Moynier, 2017; Poitrasson et al., 2019) as the δ57Fe of low-Ti basalts is mainly controlled by fractional crystallisation of olivine and pyroxene, which would preferentially incorporate light Fe isotopes, similar to ilmenite (Δ57FeIlm-Ol = +0.01 ‰; Sossi and O’Neill, 2017).
View in article
Rapp, J.F., Draper, D.S. (2018) Fractional crystallization of the lunar magma ocean: Updating the dominant paradigm. Meteoritics & Planetary Science 53, 1432–1455.
Show in contextTo model δ49TiurKREEP in a different approach, we constrained the evolution of Ti, Hf and Ta concentrations in the remaining liquid during fractional crystallisation for various LMO solidification models (Snyder et al., 1992; Lin et al., 2017; Charlier et al., 2018; Rapp and Draper, 2018) starting with LMO element abundances by Münker (2010) and calculated δ49Timelt using a Rayleigh distillation model (see Supplementary Information).
View in article
Schauble, E.A. (2004) Applying Stable Isotope Fractionation Theory to New Systems. Reviews in Mineralogy and Geochemistry 55, 65–111.
Show in contextThe principal Ti-bearing phase in the lunar mantle is ilmenite whose VI fold coordinated crystal site preferentially incorporates lighter over heavier Ti isotopes (Schauble, 2004).
View in article
Simon, S.B., Sutton, S.R. (2017) Valence of Ti, V, and Cr in Apollo 14 aluminous basalts 14053 and 14072. Meteoritics & Planetary Science 52, 2051–2066.
Show in context Titanium isotope variations in mare basalts may be used to discriminate between the two scenarios: refractory, lithophile, and fluid-immobile Ti is predominantly tetravalent albeit lunar samples may contain significant amounts of Ti3+ (Simon and Sutton, 2017; Leitzke et al., 2018).
View in article
Snyder, G.A., Taylor, L.A., Neal, C.R. (1992) A chemical model for generating the sources of mare basalts: Combined equilibrium and fractional crystallization of the lunar magmasphere. Geochimica et Cosmochimica Acta 56, 3809–3823.
Show in context The last phases to solidify in the impact-induced Lunar Magma Ocean (LMO) were K-, REE-, and P-rich residual components (urKREEP) and complementarily an ilmenite-bearing cumulate (IBC; Warren and Wasson, 1979; Snyder et al., 1992).
View in article
However, whether high-Ti basalts result from IBC assimilation by mafic low-Ti magmas (e.g., Münker, 2010), or from the partial melting of a hybridised lunar mantle source (Snyder et al., 1992), remains ambiguous.
View in article
To model δ49TiurKREEP in a different approach, we constrained the evolution of Ti, Hf and Ta concentrations in the remaining liquid during fractional crystallisation for various LMO solidification models (Snyder et al., 1992; Lin et al., 2017; Charlier et al., 2018; Rapp and Draper, 2018) starting with LMO element abundances by Münker (2010) and calculated δ49Timelt using a Rayleigh distillation model (see Supplementary Information).
View in article
Sossi, P.A., Moynier, F. (2017) Chemical and isotopic kinship of iron in the Earth and Moon deduced from the lunar Mg-Suite. Earth and Planetary Science Letters 471, 125–135.
Show in context Previous studies observed a bimodal distribution of δ57Fe values for lunar low-Ti and high-Ti basalts: higher δ57Fe values in high-Ti basalts were attributed to the presence of ilmenite during partial melting (Sossi and Moynier, 2017; Sossi and O’Neill, 2017), which is consistent with experimental studies (Longhi, 1992 and references therein) as well as our Ta/Hf and δ49Ti data.
View in article
However, intra-group variation in δ57Fe of low-Ti basalts was not observed (Weyer et al., 2005; Sossi and Moynier, 2017; Poitrasson et al., 2019) as the δ57Fe of low-Ti basalts is mainly controlled by fractional crystallisation of olivine and pyroxene, which would preferentially incorporate light Fe isotopes, similar to ilmenite (Δ57FeIlm-Ol = +0.01 ‰; Sossi and O’Neill, 2017).
View in article
This is in excellent agreement with previous experimental data and the heavy δ57Fe of high-Ti mare basalts (Longhi, 1992; Weyer et al., 2005; Sossi and Moynier, 2017; Charlier et al., 2018).
View in article
Sossi, P.A., O’Neill, H.St.C. (2017) The effect of bonding environment on iron isotope fractionation between minerals at high temperature. Geochimica et Cosmochimica Acta 196, 121–143.
Show in context Previous studies observed a bimodal distribution of δ57Fe values for lunar low-Ti and high-Ti basalts: higher δ57Fe values in high-Ti basalts were attributed to the presence of ilmenite during partial melting (Sossi and Moynier, 2017; Sossi and O’Neill, 2017), which is consistent with experimental studies (Longhi, 1992 and references therein) as well as our Ta/Hf and δ49Ti data.
View in article
However, intra-group variation in δ57Fe of low-Ti basalts was not observed (Weyer et al., 2005; Sossi and Moynier, 2017; Poitrasson et al., 2019) as the δ57Fe of low-Ti basalts is mainly controlled by fractional crystallisation of olivine and pyroxene, which would preferentially incorporate light Fe isotopes, similar to ilmenite (Δ57FeIlm-Ol = +0.01 ‰; Sossi and O’Neill, 2017).
View in article
Sprung, P., Kleine, T., Scherer, E.E. (2013) Isotopic evidence for chondritic Lu/Hf and Sm/Nd of the Moon. Earth and Planetary Science Letters 380, 77–87.
Show in contextExperimental studies and the Hf and Nd isotope composition of mare basalts suggest that low-Ti mare basalts likely result from partial melting of lunar mafic cumulates, whereas high-Ti mare basalts are thought to reflect IBC involvement (e.g., Longhi, 1992; Sprung et al., 2013).
View in article
Thiemens, M.M., Sprung, P., Fonseca, R.O.C., Leitzke, F.P., Münker, C. (2019) Early Moon formation inferred from hafnium–tungsten systematics. Nature Geoscience 12, 696–700.
Show in context High precision HFSE, W, U, and Th data, obtained for the same samples (Thiemens et al., 2019), can help to constrain δ49TiurKREEP, δ49TiIBC and to identify the processes leading to δ49Ti variations in lunar samples.
View in article
van Kan Parker, M., Mason, P.R.D., van Westrenen, W. (2011) Trace element partitioning between ilmenite, armalcolite and anhydrous silicate melt: Implications for the formation of lunar high-Ti mare basalts. Geochimica et Cosmochimica Acta 75, 4179–4193.
Show in contextFor example, the compatibility of HFSEs in ilmenite is variable (Münker, 2010): tantalum is the most compatible HFSE in ilmenite, whereas Hf is the most incompatible (e.g., DTa > 1 > DHf; Leitzke et al., 2016; or DTa > DHf, van Kan Parker et al., 2011).
View in article
Warren, P.H., Taylor, G.J. (2014) The Moon. Treatise on Geochemistry (Second Edition) 2, 213–250.
Show in context Two KREEP-rich samples along with SaU169 have U/Hf, U/Zr and U/Ti in the range of the urKREEP estimate (calculated using data by Warren and Taylor, 2014), which results in a conservative estimate of δ49TiurKREEP of +0.296 ± 0.067 ‰ (see Supplementary Information).
View in article
The intersect of the respective Ti, Hf and Ta concentrations with urKREEP concentration estimates by Warren and Taylor (2014) together with the modelled δ49Timelt-evolution line yield a range for δ49TiurKREEP: all estimates using the above mentioned models and LMO concentrations of Ta (δ49TiurKREEP = +0.308 ± 0.064 ‰), Hf (δ49TiurKREEP = +0.290 ± 0.040 ‰) and Ti (δ49TiurKREEP = +0.238 ± 0.085 ‰) are identical to our KREEP-rich sample with highest δ49Ti (+0.296 ± 0.016 ‰), and previous estimates for δ49TiurKREEP (Greber et al., 2017b).
View in article
Warren, P.H., Wasson, J.T. (1979) The origin of KREEP. Reviews of Geophysics 17, 73.
Show in context The last phases to solidify in the impact-induced Lunar Magma Ocean (LMO) were K-, REE-, and P-rich residual components (urKREEP) and complementarily an ilmenite-bearing cumulate (IBC; Warren and Wasson, 1979; Snyder et al., 1992).
View in article
Weyer, S., Anbar, A., Brey, G., Munker, C., Mezger, K., Woodland, A. (2005) Iron isotope fractionation during planetary differentiation. Earth and Planetary Science Letters 240, 251–264.
Show in context However, intra-group variation in δ57Fe of low-Ti basalts was not observed (Weyer et al., 2005; Sossi and Moynier, 2017; Poitrasson et al., 2019) as the δ57Fe of low-Ti basalts is mainly controlled by fractional crystallisation of olivine and pyroxene, which would preferentially incorporate light Fe isotopes, similar to ilmenite (Δ57FeIlm-Ol = +0.01 ‰; Sossi and O’Neill, 2017).
View in article
This is in excellent agreement with previous experimental data and the heavy δ57Fe of high-Ti mare basalts (Longhi, 1992; Weyer et al., 2005; Sossi and Moynier, 2017; Charlier et al., 2018).
View in article
top
Supplementary Information
The Supplementary Information includes:
- Sample Description
- Methods
- Results
- Discussion
- Tables S-1 to S-7
- Figures S-1 to S-9
- Supplementary Information References
Download the Supplementary Information (PDF).
Figures and Tables
Figure 1 Measured Ti isotope compositions of lunar samples. All uncertainties unless stated otherwise are 95 % confidence interval (c.i.) of at least 6 measurements of the same aliquot. See main text and Supplementary Information for details. bce = breccia, ilm = ilmenite; QNB = quartz-normative basalt, olv = olivine, pgt = pigeonite.
Table 1 Summary of reference materials and measured samples. For reference materials, n gives the number of sequences in which the material has been measured at least 6 times. Abbreviations are the same as in Figure 1. Two aliquots of OL-Ti were run through the chemical separation process and are given as OL-Ti (chemistry).
| δ49Ti | 2 s.d. | n | 95 % c.i. | |||
| JB-2 | -0.044 | 0.023 | 7 | 0.011 | ||
| BCR-2 | -0.025 | 0.012 | 4 | 0.010 | ||
| OL-Ti mean | 0.000 | 0.001 | 8 | 0.000 | ||
| OL-Ti (chemistry) | -0.001 | 0.030 | 2 | 0.137 | ||
| Col-Ti mean | 0.206 | 0.021 | 8 | 0.009 | ||
| Low-Ti rocks | ||||||
| 12022 | Apollo 12 | Low-Ti ilm basalt | 0.029 | 0.022 | 6 | 0.011 |
| 12051 | " | Low-Ti ilm basalt | 0.030 | 0.022 | 8 | 0.009 |
| 12063 | " | Low-Ti ilm basalt | 0.055 | 0.022 | 6 | 0.012 |
| 12054 | " | Low-Ti ilm basalt | 0.028 | 0.026 | 6 | 0.014 |
| 12007 | " | Low-Ti olv basalt | 0.006 | 0.010 | 6 | 0.005 |
| 12053 | " | Low-Ti pgt basalt | -0.013 | 0.026 | 6 | 0.014 |
| 15495 | Apollo 15 | Low-Ti ONB | 0.011 | 0.017 | 6 | 0.009 |
| 15555 | " | Low-Ti ONB | -0.008 | 0.026 | 6 | 0.014 |
| 15556 | " | Low-Ti ONB | 0.007 | 0.030 | 6 | 0.016 |
| 15065 | " | Low-Ti QNB | -0.030 | 0.014 | 6 | 0.007 |
| 15058 | " | Low-Ti QNB | -0.010 | 0.034 | 6 | 0.018 |
| 15545 | " | Low-Ti QNB | 0.011 | 0.032 | 6 | 0.017 |
| Low-Ti mean ± 2 s.d. | 0.010 | 0.047 | 12 | |||
| High-Ti rocks | ||||||
| 10017 | Apollo 11 | High-Ti ilm basalt | 0.009 | 0.024 | 6 | 0.012 |
| 10020 | " | High-Ti ilm basalt | 0.011 | 0.022 | 6 | 0.011 |
| 10057 | " | High-Ti ilm basalt | 0.009 | 0.024 | 6 | 0.012 |
| 74255 | Apollo 17 | High-Ti ilm basalt | 0.043 | 0.010 | 6 | 0.005 |
| 74275 | " | High-Ti ilm basalt | 0.045 | 0.010 | 6 | 0.005 |
| 75035 | " | High-Ti ilm basalt | 0.115 | 0.025 | 6 | 0.013 |
| 79135 | " | High-Ti bce | 0.019 | 0.017 | 6 | 0.009 |
| 79035 | " | High-Ti bce | 0.047 | 0.018 | 6 | 0.010 |
| High-Ti mean 1 ± 2 s.d. | 0.026 | 0.036 | 7 | |||
| High-Ti mean 2 ± 2 s.d. | 0.037 | 0.071 | 8 | |||
| KREEP-rich rocks | ||||||
| 14305 | Apollo 14 | KREEP-rich bce | 0.296 | 0.030 | 6 | 0.016 |
| 14310 | " | KREEP basalt | 0.263 | 0.035 | 6 | 0.018 |
| 72275 | Apollo 17 | KREEP-rich bce | 0.185 | 0.026 | 6 | 0.014 |
| 68115 | Apollo 16 | KREEP-rich bce | 0.117 | 0.027 | 6 | 0.014 |
Figure 2 Plots of (a) δ49Ti vs. MgO/TiO2, (b) δ49Ti vs. Ta/Hf and (c) δ49Ti vs. U that can be used to discriminate between processes leading to variations in δ49Ti. Symbols are the same as in Figure 1, black arrows indicate the direction in which a process would influence the values along y- and x-axes. Partial melting assumes the presence of ilmenite in the source, fractional crystallisation always implies fractional crystallisation of ilmenite. AFC implies the assimilation of an IBC component during ilmenite-free fractional crystallisation of a low-Ti magma. (b) The estimated Ta/Hf value of urKREEP is ~0.11 (Warren and Taylor, 2014
Warren, P.H., Taylor, G.J. (2014) The Moon. Treatise on Geochemistry (Second Edition) 2, 213–250.
). (c) Contamination by urKREEP (coloured arrow), would imply higher U contents (not observed). urKREEP is based on our first model (see Supplementary Information).