Hot and Dry Deep Crustal Xenoliths from Tibet

See allHide authors and affiliations

Science  31 Mar 2000:
Vol. 287, Issue 5462, pp. 2463-2466
DOI: 10.1126/science.287.5462.2463


Anhydrous metasedimentary and mafic xenoliths entrained in 3-million-year-old shoshonitic lavas of the central Tibetan Plateau record a thermal gradient reaching about 800° to 1000°C at a depth of 30 to 50 kilometers; just before extraction, these same xenoliths were heated as much as 200°C. Although these rocks show that the central Tibetan crust is hot enough to cause even dehydration melting of mica, the absence of hydrous minerals, and the match of our calculated P-wave speeds and Poisson's ratios with seismological observations, argue against the presence of widespread crustal melting.

The processes that formed the Tibetan Plateau—the highest and largest topographic feature on Earth, and the archetypal signature of continent-continent collision—remain unclear. Models of wholesale underthrusting of India (1), lower crustal flow (2), intracontinental subduction (3, 4), or distributed shortening followed by convective thinning of thickened mantle lithosphere (5) make predictions about the thickening, heating, and melting histories of the plateau, but we have little direct knowledge of, for example, the thickness and thermal regime of Tibetan crust during much of the Cenozoic. The ages of crustal extension (6), crustal shortening (7), rapid upper crustal cooling (8), and potassic volcanism (9) have all been used to infer when the Tibetan Plateau attained its present thickness of 60 to 75 km, but few studies quantify crustal thickness or thermal gradient. Moreover, the presence of potassic volcanism has been causally linked to convective thinning of overthickened mantle lithosphere—and this seemed a plausible explanation when volcanism appeared to be only 13 million years (My) old and younger. However, recent reports (10) reveal that this style of volcanism has occurred in Tibet for 45 My, which calls into question the role of convective thinning, which should be an event with a time scale of about 5 My (11). Here we report a detailed analysis of xenoliths from the Tibetan Plateau, which reveal that the lower crust includes anhydrous metasedimentary granulite-facies rocks equilibrated at temperatures of 800° to 1100°C at depths of 30 to 50 km.

Alkalic volcanic rocks have erupted on the Tibetan Plateau since at least 45 million years ago (Ma) (Fig. 1). The most widespread and common are shoshonitic to ultrapotassic trachybasalts to trachytes (9, 10, 12–14). Xenoliths and xenocrysts were recently discovered in these volcanic rocks at eight localities (12, 14, 15). We examined inclusions from the north-central part of the Qiangtang terrane (16) in an area typified by north-south Pliocene-Quaternary grabens (17). We observed basaltic flows overlying relatively fresh, porphyritic, trachyandesite to trachyte flows and rhyolites; broad-beam electron-microprobe analyses show that the trachytic flows are moderately shoshonitic (K2O/Na2O = 1.5 to 2.0). Phenocrysts in the shoshonites include plagioclase (An28–42), sanidine (Or56–59), partly resorbed quartz, pargasite, apatite, and abundant partly resorbed biotite. The groundmass is dominated by microcrystalline sanidine, phlogopite, ilmenite, titaniferous magnetite, apatite, and glass (18). The more evolved flows locally contain millimeter- to centimeter-sized xenoliths and xenocrysts.

Figure 1

Volcanic rocks and xenolith localities on the Tibetan plateau (3, 9, 10, 12–14). Xenoliths: asterisk, this study; boxed asterisk (12, 14). (Inset) Ages of volcanic rocks dated by K/Ar and Ar/Ar methods; fields with ages in multiple ranges are shown schematically.

The xenoliths comprise mafic rocks and siliciclastic metasedimentary rocks (19). The mafic rocks include biotite-bearing clinopyroxenite and amphibolite in addition to clinopyroxene-orthopyroxene-plagioclase granulites. The metasedimentary samples are dominated by garnet + orthopyroxene + plagioclase + potassium feldspar + quartz ± biotite ± sillimanite ± hercynite ± monazite ± zircon; the most aluminous sample contains cordierite + sillimanite + orthopyroxene + quartz. These mineralogies and mineral compositions clearly demonstrate that the metasedimentary protoliths were pelites and graywackes (20). The xenocrysts are mainly sanidine, garnet, biotite, plagioclase, orthopyroxene, and quartz—similar to the phenocryst assemblage in the volcanic rocks except for the addition of garnet.

The mineral textures and compositions provide unambiguous evidence for multistage recrystallization of the xenoliths. The primary recrystallization is characterized by the coexistence of almandine garnet + orthopyroxene + biotite + plagioclase + potassium feldspar ± quartz ± sillimanite ± osumilite in metasedimentary rocks and orthopyroxene + clinopyroxene + plagioclase ± biotite in mafic rocks (21). The mafic rocks are unfoliated to weakly foliated and preserve cumulus textures, whereas the metasedimentary rocks exhibit compositional layering, a weak foliation, and granoblastic, equigranular textures typical of granulite-facies rocks.

Many of the xenoliths reveal that higher temperatures succeeded formation of the primary mineral assemblage. This secondary overprinting is defined texturally by the breakdown of phlogopite to orthopyroxene + spinel (22) symplectites that are an order of magnitude finer grained (50 to 150 μm) and compositionally different from those in the primary assemblage. The habits and shapes of these reaction products are typical of granulite-facies symplectites developed at H2O-undersaturated conditions (23).

Later, when the xenoliths were entrained in the magma that carried them to the surface, silicate liquid intruded into fractures in the xenoliths, forming pockets of quenched melt and a variety of textural and chemical changes, including dehydration melting of biotite to micrometer-scale orthopyroxene + spinel, growth of clinopyroxene + potassium feldspar by reaction of melt with plagioclase and quartz, growth of new orthopyroxene rims from the melt, formation of monazite rims on spinel, reaction of garnet with the melt to form andesine + biotite + magnetite, formation of coronas tens of micrometers thick of submicrometer-scale plagioclase + spinel + orthopyroxene symplectites along garnet-plagioclase grain boundaries, in situ melting of plagioclase, and dissolution of feldspars and quartz. Xenocrysts within the host volcanic rocks display similar features, including growth of secondary rims on orthopyroxene and plagioclase and resorption of phlogopite and quartz.

The mineral compositions and zoning in the Qiangtang xenoliths are compatible with a scenario in which a regional thermal gradient was perturbed more than once by magmatic heat. Recrystallization pressures and temperatures were assessed mostly by partitioning of Al, Fe, and Mg between orthopyroxene and garnet (Fig. 2) (19, 24). Mineral cores in the metasedimentary xenoliths show Fe-Mg apparent temperatures (T Mg) that are colder than Al apparent temperatures (T Al), implying cooling toT Mg from an initial heating toT Al (Fig. 2). Core-to-rim increases in Mg/Fe in garnet and decreases in Mg/Fe in orthopyroxene indicate subsequent heating to 950° to 1300°C (25), withT Mg =T Al. The preservation of heating-induced zoning at these temperatures requires that the heating lasted <1 My (26). The coincidence of rimT Mg and T Alrequires rapid cooling of the sort expected during volcanic eruption. Orthopyroxene-clinopyroxene pairs in the mafic xenoliths have homogeneous cores showing T Mg of 850° to 1015°C, and zoned rims, where two pyroxenes are adjacent, showing heating of 35° to 150°C above the coreT Mg. Pressures calculated for the metasedimentary xenoliths range from 0.8 to 1.3 GPa, implying extraction depths of about 30 to 50 km.

Figure 2

Tibetan xenolith thermobarometry (39) and calculated P-wave velocities (34). Pale and medium gray bands show inferred “normal” and magmatically elevated thermal gradients, respectively.

Two separate sanidine xenocrysts dated by the40Ar/39Ar method yielded ages of 2.5 ± 0.2 Ma and 3.2 ± 0.1 Ma (27). These are unambiguously eruption ages of the volcanic rocks because sanidines do not quantitatively retain Ar at the high temperatures demonstrated for these xenoliths. We determined208Pb/232Th ages from 200-μm-diameter monazite inclusions in a 1-cm garnet (12G15A) with an ion microprobe (19, 28). The garnet contains cracks that intersect all the monazite inclusions, and the cracks exhibit alteration associated with silicate melt intrusion. The monazites exhibit U zoning and sector and oscillatory zoning visible with back-scattered electron imaging. Two monazites with dentate grain boundaries show a range in 208Pb/232Th spot ages from about 14.2 to 4.2 Ma. A subhedral, equant grain, part of a polymineralic inclusion intruded by melt, gave five spot ages of about 3.4 Ma, and spot ages from a second, texturally similar, monazite grain range from 16.2 to 4.5 Ma. We interpret these textures and ages to indicate Pb loss at about 3.4 Ma. Calculations show that 200-μm monazite grains will lose all their Pb in much less than 1 My if subjected to temperatures >900°C (29), confirming that the highest temperatures experienced by the Qiangtang xenoliths were transient.

We interpret the history of the Qiangtang xenolith suite to be (i) deposition of the metasedimentary rocks on Earth's surface; (ii) burial, muscovite-dehydration melting, and extraction of leucogranitic liquids (30); (iii) attainment of the characteristic granulite-facies metamorphism at 800° to 900°C, biotite-dehydration melting and fluorination of remaining biotite; (iv) additional reaction, including the breakdown of F-biotite and dissolution of feldspar and quartz at 1300°C caused by injection of magma into the lower crust; followed shortly by (v) decompression and quenching during eruption about 3 Ma. That these conclusions pertain to a broad area of the plateau and may reach as far back in time as 15 Ma is suggested by almandine + spinel + sillimanite xenoliths in 15-Ma volcanic rocks (12) about 300 km northeast of our locality (Fig. 1).

The xenolith temperatures indicate that a thermal gradient of 17°C per kilometer has prevailed since at least 3 Ma. Seismic data suggest that the present temperature of the uppermost mantle beneath Qiangtang is 840° to 1170°C (31). Satellite magnetic data suggest that temperatures > 550°C prevail across the Tibetan Plateau at depths of 15 ± 5 km (Fig. 2), and this places the H2O-saturated granite solidus at a depth of 17 ± 5 km across the plateau, such that melt may be present below this depth across the entire plateau (32). The slower-than-normal seismic wave speeds and somewhat high Poisson's ratio of the crust imply the presence of partial melt (33). The xenolith suite shows, however, that although the Qiangtang lower crust is at temperatures above the solidus of H2O-saturated continental crustal rocks—indeed, above the dehydration-melting solidus of mica (Fig. 2)—and has been since 3 Ma, the nearly anhydrous character of the xenoliths (the OH in biotite having been replaced by F, for example) means that the crust need not be molten in spite of the extreme temperatures. Moreover, the seismic wave speeds and Poisson's ratios we calculate for the xenoliths, considering only their crystalline phases and excluding melt (Fig. 2) (19), straddle those from seismological models of Qiangtang terrane (33–35).

Turner et al. (13) interpreted Tibetan potassic lavas to have been generated by partial melting of enriched mantle, in part because of high K2O and light rare-earth-element concentrations, radiogenic Sr and Pb isotopic ratios, and relatively unradiogenic Nd. However, if the lower crust of Tibet is partly metasedimentary, instead of wholly gabbroic, and is hot, mantle-generated melts that traverse the lower crust will have considerably different interaction with the lower crust than is commonly considered. Indications of in situ melting of biotite, feldspar, and quartz in the metasedimentary xenoliths, and of resorption of biotite, feldspar, and quartz in the volcanic rocks hosting the xenoliths, reveal that partial melting of the xenoliths or other metasedimentary rocks contributed some silicate liquid to the mantle-derived magma. The widespread occurrence of fine-grained, undigested xenocrysts suggests that the unusual chemical patterns of some Tibetan lavas may be a mixture of lower Tibetan crustal fragments with a mantle-derived melt. Moreover, Patiño Douce and McCarthy (36) showed that dehydration melting of mica—the melting documented in the Qiangtang xenoliths—produces (ultra)potassic melts and garnet-cordierite-orthopyroxene–rich residues. Such rocks would have previously undergone H2O-saturated melting to yield trondhjemitic liquids and mica-rich residues; this inferred early melting event could have occurred either during introduction to the deep crust if the protolith was low-grade sediments such as the Songpan-Ganze flysch (37) or possibly long before if the protolith metasedimentary rocks were already dry granulites. Surface geologic exposures have been used to infer that upper crustal lithologies are widespread beneath central and northern Tibet (37); if so, the currently popular model for the Tibetan ultrapotassic-shoshonitic volcanism, melting of enriched lithospheric mantle due to convective thinning of its lower parts, should be reconsidered.

As an alternative to convective thinning of the mantle lithosphere, Deng (3) suggested that potassic volcanism on the plateau is linked to intracontinental subduction. This idea received support from Meyer et al. (4), who proposed that the distribution and age of plateau volcanism is directly linked to successive episodes of upper-middle crustal shortening accommodated by intracontinental subduction of lower crust and mantle, and from Kapp et al. (37), who proposed that the Songpan-Ganze flysch was thrust beneath Qiangtang in the early Mesozoic, and later catalyzed Cenozoic volcanism. The pattern of volcanism on the plateau (Fig. 1) (38) suggests a meld of both models: from 30 to 45 Ma volcanism straddled the Jinsha suture in eastern Tibet, between 29 and 6 Ma volcanic rocks erupted in a small area southwest of the Kunlun fault, and subsequent eruptions have occurred in the same area as well as farther west. If the pattern of volcanism can be used as a guide, the major changes in the evolution of the plateau distal to the Himalayan front occurred at 30 and 5 Ma.


View Abstract

Stay Connected to Science

Navigate This Article