Decompression Melting of Peridotite Xenoliths

Peridotite xenoliths commonly exhibit melt veining, variably attributed to mantle metasomatism, host magma infiltration, or decompression melting. In order to test these origins and constrain the P–T–t path of xenolith ascent, we study spinel-lherzolite xenoliths and host scoria from a cinder cone north of the Grand Canyon (Sullivan Ranch). Micro-x-ray computed tomography (μCT) scans of xenoliths show that melt veins are ubiquitous at this locality. They make up ~ 10% of bulk xenoliths, are highly interconnected at the mm-scale and include bubbles. Bubbles are surrounded by silica-rich residual glass, although veins are largely crystalline, composed of secondary plagioclase, clinopyroxene (cpx), olivine and Cr-spinel (sp). The reconstructed bulk melt vein is basaltic (50–52% SiO2), and mineral compositions within the melt vein (high-Fo olivine, high-Ti spinel, high-Cr and lower-La/Sm cpx) are distinct from primary minerals away from melt veins. The reconstructed incongruent melting reactions for two different xenoliths (63cpx + 72opx + 13sp ➔ 100liq + 48ol, on average) are similar to those from equilibrium peridotite melting experiments at 1 GPa, even though minerals are not in equilibrium across the 3–10 cm-scale xenoliths. The timescale for xenolith melting was derived from Fe-Mg zonation profiles in primary olivine (Fo89–90) adjacent to melt veins (with Fo91–92 rims) that indicate diffusion over 1–3 weeks. This timescale is similar to that derived for olivine grains at the xenolith exterior that were diffusively equilibrating with the host magma for ~ 1 week. The similarity of timescales from the interior melt veins to the exterior olivine indicates that in situ melting occurred during entrainment and transport. Ascent rates from the entrainment depth (~ 40 km) to the surface are 0.05–0.15 m/s, similar to other primitive cinder cone eruptions and consistent with Stokes settling velocities. Xenocrysts in the host scoria yield diffusion timescales of hours to days, consistent with xenolith disaggregation late in the ascent history. Evidence for limited interaction with the host magma comes from trace element zonation in melt veins that approach the host composition in the outer 1 cm of the xenolith. The interaction does not appear to be driven by diffusion, as elements with very different diffusivities (e.g. Cl vs. Zr) show similar lengthscales of zonation. Instead, host melt infiltration could have occurred due to a combination of capillary forces and inflation of the xenolith driven by the volume change of melting and bubble growth in the melt veins. Given the P–T–t path of these lherzolite xenoliths, which should be similar to those erupted from other alkali basalt cinder cones from around the world, decompression melting is inevitable. The extent of decompression melting will be a function of the host magma P–T path, the xenolith water content, and the proportion of touching cpx-opx-sp. Such melting consumes cpx and may lead to systematic misclassification of xenoliths as refractory harzburgites that were originally lherzolites. Geochemical studies of xenoliths should take care to avoid the outer cm (or more) that may have been contaminated by the carrier magma, and secondary minerals formed from infiltration, reaction, melting and/or crystallization.