The generation of sodic granite magmas, western Palmer Land, Antarctic Peninsula

Subduction-related Mesozoic to Cainozoic granites s.l. in western Palmer Land, Antarctic Peninsula, have similar chemical compositions to Archean tonalite-trondhjemite-granodiorite (TTG) suites, Phanerozoic slab-melts (adakites), and to experimental partial melts of basaltic material in equilibrium with amphibole ± pyroxene ± garnet. They are predominantly sodic, metaluminous and most have Al₂O₃ > 15 wt% and Y < 18 ppm. All are light rare earth element (LREE)-enriched (2 < La/Yb_n <30) and most have small Eu anomalies. They have a wide range of initial ɛNd_(t) (−6.8 to +4.5) and ɛSr_(t) (+293.4 to −3.7), but most Pb isotope compositions deviate by < 0.3% from their mean. The Pb isotope data indicate a crustal component to all the granites, which Sr and Nd isotope variations suggest is pre-Triassic–Triassic. The ²⁰⁷Pb/²⁰⁴Pb_(t) range from 15.602 to 15.666 and appear to preclude a significant Proterozoic, or older, crustal component. The granites have chemical and isotopic compositions that suggest they are not partial melts of subducted oceanic lithosphere, as has been suggested for some Archean and Phanerozoic TTG magmas. We conclude that they were produced by mixing between basaltic-andesitic arc magmas, partial melts of juvenile basaltic lower crust and pre-Triassic crust. The low H(heavy)REE+Y content of some of the granites requires that garnet was a residual phase in the crust during partial melting, indicating a crustal thickness of >36 km. Between Triassic and Tertiary times the initial ɛNd_(t) of the magmatism increased and ɛSr_(t) decreased, suggesting that new continental crust was produced during this period. Underplating by mafic magma was an important crustal growth mechanism in the arc: the generation of abnormally thick crust, and its subse quent fusion, is considered to be a consequence of ca. ≥ 180 Ma of subduction and associated magmatism in the region. An implication of the model is that dense garnet-amphibolite and eclogite residues from partial melting of the lower crust will accumulate. In theory, the setting was appropriate for such residues to detach from the base of the crust and to sink into the convecting mantle. Such a process would leave the rest of the crust enriched in large ion lithophile elements/LREE, but depleted in HREE+Y. Received: 2 October 1995 / Accepted: 5 January 1997