Coupled Sr-Nd-Pb isotope and trace element ratios in olivine-hosted melt inclusions from subduction-related magmatism in central Italy

Abstract

The complex intercontinental collision zone of peninsular Italy is characterised by diverse potassium-rich magmatic products that reflect large temporal and spatial variations in subducted material in the mantle source. Understanding the chemical geodynamics of Italy—a post-collisional subduction setting marked by large sediment input—is key to deciphering both the extent and mechanisms of subduction recycling worldwide. However, primary melt systematics and thereby source characteristics are obscured in bulk lavas by magma mixing/mingling and crustal assimilation at shallow levels. Hence the need for isotope, major and trace element data on melt inclusions (MIs) trapped in high-forsterite olivine, which represent partial melts that have by-passed such secondary modifications, and thus more fully record the geochemical heterogeneity of the mantle source. Previously the analysis of coupled Sr-Nd-Pb isotope systems in small (ø < 300 µm) MIs was hindered by analytical limitations. Recent advances in TIMS technology, i.e. the use of 10^13 Ω resistors in the feedback loop of Faraday cup amplifiers (Koornneef et al., 2015), now allow isotope analysis of exceedingly low abundances of Sr (2 ng), Nd (30 pg) and Pb (100 pg). In contrast to in situ techniques, chemical separation of the elements of interest prior to TIMS analysis eliminates matrix effects and interfering isotopes, and thus leads to more accurate and precise data. Although this approach requires dissolution of the MI and host olivine, the latter contains insignificant amounts of Sr, Nd and Pb (De Hoog et al., 2010), meaning its contribution is negligible. Nonetheless, wet chemistry procedures require miniaturised, ultra-low blank techniques to minimise blank contributions (Koornneef et al., 2015). Total procedural blanks for these techniques are typically <20 pg Sr, <1 pg Nd and <10 pg Pb, but can be corrected for using the elemental abundances of Sr, Nd and Pb determined through isotope dilution by means of single (Sr and Nd) and double spike techniques (Pb; Klaver et al., 2016). In addition, the residual matrix, collected as eluate during column chemistry, is analysed for trace element ratios by ICP-MS (e.g. Koornneef et al., 2017). Repeat analyses of small, representative aliquots of international rock standards AGV-1 (4–8 ng Sr; 200–400 pg Nd; 250–500 pg Pb; n = 6) and JB-2 (17 ng Sr; 600 pg Nd; 500 pg Pb; n = 3), corrected for column yields for each element, yield trace element ratios (e.g. La/Sm, Cs/Ba, Sm/Gd) typically reproducible to within 5–15% 2RSD. Sr-Nd-Pb isotope data for AGV-1 (87Sr/86Sr = 0.70400 ± 2; 143Nd/144Nd = 0.51278 ± 8; 206Pb/204Pb = 18.93 ± 2; 207Pb/204Pb = 15.65 ± 2; 208Pb/204Pb = 38.55 ± 6; 2SD; n = 8) and JB-2 (87Sr/86Sr = 0.703685 ± 8; 143Nd/144Nd = 0.51307 ± 8; 206Pb/204Pb = 18.340 ± 5; 207Pb/204Pb = 15.56 ± 1; 208Pb/204Pb = 38.27 ± 4; 2SD; n = 3) are in excellent agreement with the GeoReM preferred values (Jochum et al., 2016). The new combined methods are applied to ~20 homogenised high-potassium (HKS) to melilitite MIs hosted by primitive (Fo92–90) olivines from three key Quaternary volcanic centres (Vulsini, Sabatini and Alban Hills) in the Roman Magmatic Province, central Italy. Most MIs were previously analysed for major and trace elements by EPMA and LA-ICP-MS. Systematic covariations are recorded between proxies for sediment metasomatism such as K2O, U/Th, U/Nb, Cs/Rb, Be and 87Sr/86Sr within the MIs. Whereas 143Nd/144Nd show little variation, marked unradiogenic Pb isotope compositions far exceed those reported for regional volcanic rocks. The primitive melt compositions indicate the involvement of isotopically distinct and trace element–enriched mantle domains below central Italy. We infer that the covariations reflect melt extraction from a small-scale heterogeneous mantle source that was modified by sediment melts derived from the subducted Adriatic slab. These findings will ultimately be used to improve our understanding of subduction recycling and deep mantle element fluxes.