P–T Phase Relations of Silicic, Alkaline, Aluminous Mantle-Xenolith Glasses Under Anhydrous and C–O–H Fluid-saturated Conditions

JOURNAL OF PETROLOGY VOLUME 38 NUMBER 9 PAGES 1187–1224 1997 P–T Phase Relations of Silicic, Alkaline, Aluminous Mantle-Xenolith Glasses Under Anhydrous and C–O–H Fluid-saturated Conditions DAVID S. DRAPER∗ AND TREVOR H. GREEN GEMOC: ARC NATIONAL KEY CENTRE FOR THE GEOCHEMICAL EVOLUTION AND METALLOGENY OF CONTINENTS, SCHOOL OF EARTH SCIENCES, MACQUARIE UNIVERSITY, N.S.W. 2109, AUSTRALIA RECEIVED OCTOBER 14, 1996 REVISED TYPESCRIPT ACCEPTED APRIL 14, 1997 High-pressure liquidus experiments on three silicic, aluminous, alkaline melts, modelled on glasses found in many mantle xenoliths, show that part of this compositional range is saturated with harzburgite (or possibly lherzolite) under anhydrous conditions. Under C–O–H fluid-saturated conditions with XH2O=0·5, phlogopite mica is present along with anhydrous phases similar to those found under dry conditions. Phlogopite is the sole liquidus phase when XH2O=1·0. At XH2O=0·5 and 3·0 GPa, garnet, kyanite and carbonate minerals appear as near-liquidus phases and the shape of the liquidus surface is reminiscent of that of the carbonated peridotite solidus. Saturation of these liquids with harzburgite, and comparisons with calculated melt silica activities, suggests that these liquids would face no chemical or thermal obstacles to circulating amongst and coexisting with harzburgitic mantle. Also, there is textural evidence that these melts may be mobile. Accordingly, these kinds of liquids could act as cryptic metasomatic agents. If mantle at ~45–90 km depth is pre-enriched in low-melting-temperature components, and probably volatiles, via the ascent and percolation of alkaline, mafic liquids (along geotherms that cross inflections in the solidus of CO2-bearing peridotite), then subsequent low-degree partial melts could yield the liquids that are ultimately trapped as xenolith glasses. INTRODUCTION xenoliths The compositions of silicate melts potentially in equilibrium with the Earth’s mantle have long been of interest to petrologists and geochemists. Much useful information about such melts has come from the study of melt inclusions in phenocrysts in volcanic rocks and in the minerals of ultramafic mantle xenoliths. In addition, many mantle xenoliths contain silicate glasses as a discrete phase, rather than (or in addition to) inclusions (see Appendix A for references). In this paper, we will refer to discrete-phase xenolith glasses simply as xenolith glasses and those found as inclusions in minerals as inclusion glasses. Xenolith glasses have been reported from mantle xenoliths hosted by alkaline, mafic magmas both from intra-plate and from subduction-related tectonic settings. The glasses have a wide range of compositions and textural relations: they are found as intergranular blebs, irregular patches, veins, and thin films that appear to wet all grain faces. Most glasses account for, at most, only a few volume percent of their host xenolith, and some much less than 1 vol. %. Xenolith glasses have a very wide range of major element compositions, falling into every compositional field except picrobasalt in the alkali vs silica classification of LeMaitre (1989). Figure 1 summarizes the published compositions of 361 xenolith glasses (data sources given in Appendix A). Very many are alkaline (Na2O+K2O up to 17 wt %). The mg-numbers (defined as 100×Mg/ (Mg+Fe)[molar], all Fe as FeO) range from 25 to 90, ∗Corresponding author. Present address: Department of Geosciences, University of Texas at Dallas, P.O. Box 830688, Richardson, TX 75083-0688 USA. e-mail:  Oxford University Press 1997 KEY WORDS: experimental petrology; metasomatism; phase equilibria; JOURNAL OF PETROLOGY VOLUME 38 with a distribution maximum around 60; MgO contents show a very wide range, from <1 to >12 wt %. There is a rough negative correlation of TiO2 contents (not plotted) with SiO2, with the more silicic glasses containing <1% TiO2 and the more mafic glasses containing 3–5 wt % TiO2. Al2O3 contents do not correlate with SiO2, mg-number, or molar Na/K, and range from 12 to 25 wt %, with a few as high as 30–32 wt %. Like TiO2, contents of FeO, MgO and CaO (not plotted) have a crude negative correlation with SiO2, so that the majority of the most silica-rich glasses (>63 wt % SiO2) have <3 wt % FeO+MgO and <3 wt % CaO. Most xenolith glasses have more Na than K, and molar Na/K ratios are largest in glasses with 50–60% SiO2. However, 33 of the 361 published glass compositions plotted in Fig. 1 have Na/K ratios of <1; the lowest value reported is 0·16. Most reported xenolith glasses were found in spinelfacies peridotites; one exception is the study of Hunter & Taylor (1982), who reported glasses in garnet kimberlite from Pennsylvania, USA. It should be noted also that there appear to be no major-element compositional features that are peculiar to a particular tectonic setting. Many xenolith glasses can be satisfactorily explained by mechanisms including infiltration of the host lava (e.g. Kuo & Essene, 1986; Garcia & Presti, 1987), melting of typical mantle peridotite [under either anhydrous conditions (e.g. Maaløe & Printzlau, 1979) or via volatile flux (e.g. Jones et al., 1983; Griffin et al., 1984)], or melting of amphibole- or phlogopite-bearing peridotite assemblages (e.g. Ellis, 1976; Amundsen, 1987; Francis, 1991). Most of the glasses that these workers explained via such processes fall in the comparatively silica-poor range of the compositions plotted in Fig. 1 (i.e. less than ~55 wt % SiO2). More recently, Chazot et al. (1996) and Yaxley et al. (1997) proposed models invoking breakdown of amphibole or phlogopite to give rise to silica-undersaturated liquids that then react with mantle minerals, principally orthopyroxene, and crystallize olivine and/ or clinopyroxene, leaving more silica-rich residual liquids. The majority of the compositions reported in these two papers also have SiO2 contents not exceeding 53–55 wt %, and many are even less silica rich. A reasonably large class of glasses appears, at first glance, to remain unexplained by processes like those outlined above. We denote these as extreme-composition glasses. They are rich in SiO2 (> ~60 wt %), Al2O3 (18–20 wt %) and the alkalis (4–10 wt % Na2O, 4–8 wt % K2O), and poor in MgO (2–4 wt %), FeO∗ (all Fe as Fe2+, 2–4 wt %) and CaO (Ζ5 wt %). In contrast, most of the comparatively less silica-rich liquids noted above are richer in MgO, FeO∗ and (especially) CaO, and poorer in K2O. The seeming inability to explain extremecomposition glasses by the kinds of mechanisms listed above led to suggestions by Edgar et al. (1989), Draper (1992), Schiano et al. (1992), Schiano & Clocchiatti (1994) NUMBER 9 SEPTEMBER 1997 (the latter two studies focusing mostly on inclusion glasses) and Zinngrebe & Foley (1995) that such melts could represent a type of metasomatic agent circulating in the upper mantle. Chazot et al. (1996) and Yaxley et al. (1997) suggested that glass compositions as (...truncated)