Subducted carbonates, metasomatism of mantle wedges, and possible connections to diamond formation: An example from California

We investigated calcite globules and veins in two spinel-garnet peridotite xenoliths from the subSierra Nevada mantle. The studied xenoliths were entrained in a Miocene (11 Ma) volcanic plug. These carbonates are associated spatially with silicate glass inclusions, suggesting that they are primary inclusions—inclusions that formed at high temperature in the mantle and not at or close to the Earthʼs surface. The host peridotites represent samples of the lithospheric mantle wedge beneath the Mesozoic California magmatic arc, as indicated by radiogenic isotopic ratios measured on clinopyroxene separates [Sr/Sr(11 Ma) = 0.7058–0.7061, εNd (11 Ma) = –1.9 to –0.7]. Mineral chemistry of the peridotite major phases is typical of a mantle section that was depleted of melt. The δO values of olivine and orthopyroxene from the two samples are also typical of mantle rocks (δO = 6–6.5‰). In contrast, calcite veins have δO of 18–20‰ and δC of –14‰, arguing for a subducted sedimentary origin for these carbonates. Presumably, the carbonates were expelled from the downgoing slab and ß uxed into the overlying mantle wedge as CO2or CO2-H2O-rich ß uids or melts. The trace-element patterns of two analyzed calcite veins are typical of the arc signatures (e.g., depletions in high-Þ eldstrength elements) seen in calc-alkaline magmatic rocks worldwide. However, the cores of peridotite clinopyroxenes do not show that pattern, suggesting that the arc-like trace element signature was introduced via the recycled carbonate agent. A connection between mantle wedge carbonation and diamond formation in a subduction environment is proposed based on these observations. magmatism as well as the atmospheric carbon cycle over geologic timescales. It is also plausible that much of the recycled CO2 in arc volcanoes may be derived from sedimentary sections that were emplaced in the lower crust beneath arcs via shortening and not from the subducted slab; this hypothesis has yet to be tested (Stern 2002). The issue of how much CO2 is released from the slab into the upper mantle and ultimately in arc volcanoes is very much unresolved and currently suffers from little direct (observational) evidence. Rare direct observations on upper-mantle xenoliths from arc environments conÞ rm the presence of carbonic ß uids in the mantle. In this study, we present new trace-element and isotopic analyses from carbonated peridotite xenoliths from the central Sierra Nevada, California (Dodge et al. 1988). This is the Þ rst report of calcite-bearing peridotites from this locality. Crustal and mantle xenoliths from Miocene pipes in the central Sierra Nevada represent fragments of the Mesozoic-Cenozoic sub-arc lithosphere (Ducea and Saleeby 1998) of this classic Cordilleran subduction zone (Dickinson 1981). We use isotopic tracers to show that the carbonates are recycled via the subducted Farallon slab, and that they imprint an arc-like trace-element signature on peridotite clinopyroxenes, a signature that the pyroxenes did not possess originally. Carbonate ß uxing may have also been responsible for generation of sub-arc diamonds in California. SAMPLES AND TECHNIQUES The Sierra Nevada batholith is a typical Mesozoic Cordilleran magmatic arc, composed primarily of tonalitic and granodioritic plutons. Granitoid rocks are * E-mail: [email protected] DUCEA ET AL.: CARBONATE METASOMATISM IN THE MANTLE 865 known from present-day surface exposures to extend to a depth of at least 30 km in the Cretaceous crustal column (Saleeby 1990; Ducea 2001). Samples of more deeply seated rocks representing the lower crust and the upper mantle beneath the arc are entrained as xenoliths in volcanic rocks of Miocene age from the central Sierra Nevada (Dodge et al. 1988; Mukhopadhyay and Manton 1994; Ducea and Saleeby 1998). They include granulite-and eclogite-facies rocks of the arc root, as well as spinel-garnet and garnet-bearing peridotites that represent samples of the mantle wedge beneath the arc (Ducea and Saleeby 1998). Two large (5–8 cm in diameter) calcite-bearing spinel-garnet peridotites (BC115 and BC125) from the Miocene (11.1 Ma) Big Creek trachyandesitic pipe (Dodge et al. 1988; Mukhopadhyay and Manton 1994; Ducea and Saleeby 1996, 1998; Lee et al. 2000a) in the Huntington Lake area, central Sierra Nevada [37°13 ̓ N, 119° 16 W, see also Ducea and Saleeby (1996) for a location map], were selected for analysis. The samples are typical of the central Sierra Nevada peridotite suite and contain spinels that are rimmed by garnets. Calcite veins make up about 3–5% of the analyzed samples. Calcite crystals are found as globules and micro-veins that are up to 0.7 cm in diameter and 0.4 cm in width, respectively (Fig. 1). In a few cases, carbonate veins either following grain boundaries or Þ lling fractures connect patches and globules of carbonates to each other. Secondary, small euhedral olivine and spinel crystals are associated with the carbonates, typically within the carbonate. Carbonate crystals are found in spatial connection with aluminosilicate glass inclusions (Fig. 1). Major-element analyses of these glasses (see below) show that they are basaltic in composition and are distinct from the host lavas. Electron-microprobe analyses on minerals were carried out at the University of Arizona using the Cameca SX50 microprobe equipped with 5 wavelengthdispersive spectrometers (using LiF, PET, and TAP crystals). Counting times for each element were 30 seconds at an accelerating potential of 15 kV and a beam current of 10 nA. Measurements with oxide totals outside of the range 100 ± 1% were discarded from the analyzed silicates. Trace-element analyses were performed on carbonates mechanically separated from the peridotites, as well as clinopyroxene separates (about 80 mg). The following acid-leaching routine obtained a rim fraction of clinopyroxenes: unbroken clinopyroxene grains were Þ rst leached in mild, cold 2.5 M HCl for two hours. This step was performed to eliminate any additional carbonate that may reside at clinopyroxene grain boundaries; then the leachate was discarded and the grains were dried out and inspected optically. The clean grains without any grain boundary staining were used for the next step, a second leaching carried out in a hot (150 °C) mixture of concentrated HF and HNO3; after two weeks of dissolution, about 1/3 of the clinopyroxene mass was dissolved—the leachate was used for the clinopyroxene rim fractions. The undissolved clinopyroxene grains were then dried out, weighed, crushed into a powder, and dissolved in the same manner—the dissolved samples were analyzed as the “clinopyroxene core” fractions. About 10% of the dissolved core and rim samples were used for trace-element determinations in an ICP-MS instrument, the remainder being eluted in standard chromatographic columns for thermal ionization mass spectrometry (TIMS) analyses. For both fractions, the elution procedures for radiogenic isotopic analyses are as in Ducea and