Origin and evolution of mafic volcanics of Sumatra (Indonesia) : their mantle sources, and the roles of subducted oceanic sediments and crustal contamination

The Quaternary volcanoes along the Sunda arc of Indonesia are a well known example of subduction-related volcanism. Relatively primitive rocks in the arc are rare, so that in the last two decades much attention has been dedicated to the study of the isotopic systematics of the mafic volcanics. These suggest that sediments - or fluids derived from the sediments - subducted along the Sunda Trench have some effect on the chemistry of the arc volcanics only in some sectors of the arc, whereas the isotopic signature of other sectors mainly reflects the composition of their mantle source. However, very little is known about the composition of the mantle beneath the arc, and good quality, representative analyses of oceanic sediments from the northeastern Indian Ocean are scarce. Moreover, the importance of crustal contamination has not been sufficiently tested, at least partly due to the lack of exposure of arc basement in the best studied areas. Good-quality, modern analyses of volcanic rocks from the Sunda arc mainly come from the relatively easily-accessible areas, but little is known about the least accessible volcanoes in the largest island in the arc, Sumatra, despite the fact that the occurrence of both granitic rocks and very primitive basalts was reported some 70 years ago. In the first part of this thesis, after a brief review of the studies on the volcanism in the Sunda-Banda arc, an attempt is made to define the chemical and isotopic composition of \typical\" oceanic sediments and continental crust for use in the modelling of contamination processes along the Sunda Trench. \"Basement\" granitoids and arc-related granitoids from several localities in Sumatra were investigated and described and compared with the better known granitoids of peninsular southeast Asia. Two groups of ganitoids have been identified. Arc-related granitoids generally older than the Quaternary arc volcanics are found along the arc follow the calcalkaline trend of arc rocks and have Sr Nd and Pb isotope systematics similar to those of Indian Ocean basalts (including basalts from south Sumatra). East of the Semangko fault granitoids and pyroclastic rocks are mainly of \"S-type\" and have considerably high initial 87Sr/86Sr values similar to those found in the Central Granitoid Province of Southeast Asia. Therefore the Semangko fault and the Quaternary arc may represent an important tectonic boundary defining the southeastern margin of the SIBUMASU terrane. No systematic variations in geochemical and isotopic composition from north to south Sumatra have been observed in the two groups of granitoids. This suggests that there are no major variations in the isotopic composition of the source of arc volcanism and of the continental crust from north to south Sumatra. Oceanic basalts from the Northeastern Indian Ocean were studied to test the possibility of the existence beneath the Sunda arc of mantle sources with OIB composition and \"enriched\" (EM HIMU) isotopic signatures. Basalts with an isotopically \"enriched\" component are described for the first time in three new locations in the Northeastern Indian Ocean: the Cocos Plateau the Investigator Ridge and site DSDP 22211. Only the occurrence of this type of basalts along the Investigator Ridge is consistent with a \"fixed hot spot\" model of evolution of the Northeastern Indian Ocean. Pre-\"Gondwana rifting\" basalts in the Argo Abyssal Plain seem to have isotope systematics similar to Atlantic and Pacific MORB. For the sediments the large (but qualitatively poor) existing database was integrated with new data to provide a more precise picture of variations of chemical and isotopic composition of the sedimentary cover in time and distance from and along the arc. The variability in major and trace element composition observed in oceanic sediments can be modelled as the result of variable degrees of dilution of a detrital component by organogenic calcite and silica. Pb isotopes can distinguish between pre- and postMiocene sediments and the Sr Nd and Pb isotope systematics in old sediments are consistent with their derivation from volcanogenic material with isotope systematics similar to those of Indian Ocean basalts and material from pre-Mesozoic continental crust similar to that found in Sumatra. On the other hand post-Miocene sediments seem to be enriched in a U-rich component. Using the new data the distribution of volcanism along Sumatra and the west Sunda arc and the effects of crustal and sediment contamination are briefly reviewed and discussed. These data suggest that neither contamination of the mantle source by bulk sediment nor bulk assimilation of crustal material by an uprising melt can satisfactorily account for the Sr Nd and Pb isotope systematics observed in most arc rocks in the west Sunda arc. Fluids released from the subducted sediments do not seem to be able to reproduce the range and spatial distribution of 87Sr/86Sr - 143Nd/144Nd values in the west Sunda volcanic rocks. Pb isotopic values and their variations along the arc also suggest that subducted sediments did not have important effects on the Pb systematics of the west Sunda volcanic rocks. On the contrary assimilation of partial melts of crustal material by uprising magmas can account for the Sr Nd and Pb isotope systematics in the Sumatran and Jawanese arc rocks and is consistent with the spatial distribution of Sr Nd and Pb isotope values observed from north Sumatra to Lombok. This model requires that Indian Ocean OIB-like mantle as well as MORB mantle be present in the mantle wedge and also requires the existence of pre-Mesozoic continental crust in Sumatra and west Jawa but its absence in the Bali sector from east Jawa to Lombok. Relatively low 3He/4He and high B/Be and a 18 0 values usually interpreted as evidence for metasomatism of the mantle wedge by slab-derived material are rare in the west Sunda volcanics and are best explained by processes occurring at shallow levels (degassing of the magma percolation of meteoric fluids assimilation of crustal material) in the magma chamber or at near-surface levels. 230Th/232Th values in Holocene rocks are indicative of an OIB source and there is no evidence for 230Th - 238U disequilibrium and metasomatism of the mantle source by slab-derived fluids. In the second part of this thesis all three localities in Sumatra where olivine-phyric basalts had been previously reported (Sukadana Bukit Telor and Bukit Mapas) have been studied together with other alleged occurrences of olivine-phyric basalts elsewhere in Indonesia. Whole rock major trace and Sr Nd and Pb isotope analyses together with detailed mineralogical studies were carried out to determine the composition of the source and the evolution of these isotopically and compositionally rather unusual (in an arc environment) rocks and their similarity to OIB and MORB basalts from the Indian Ocean and Australia is evaluated The basaltic andesites of Bukit Mapas are olivine-phyric but closely resemble the other Sumatran arc volcanics in their mineral and whole-rock composition and Sr Nd and Pb isotope systematics. These rocks are characterised by the coexistence of Al-spinel and Cr-spinel inclusions in olivine phenocrysts probably as the result of mixing at relatively shallow depth between a \"normal\" basaltic melt and a high-Al melt derived from the melting of crustal material. Two types of olivine-bearing basalts (Sukadana basalts) were identified in south Sumatra. The high-Ti basalts have TiO 2 > 2.3% are slightly alkaline and have HFSE and MREE and HREE concentrations similar to those of OIB but are relatively depleted in LILE and LREE. Isotopically they resemble basalts from the Ninetyeast Ridge and some Indian Ocean MORB and there is no evidence for their derivation from isotopically strongly enriched hot spot magmas. Despite their proximity to the arc there is no evidence for metasomatism of their source by subducted sediments and the depletion in LREE and L1LE suggests that their source might have suffered a previous episode of depletion: maybe the extraction of arc melts as suggested by the fact that these rocks are spatially and temporally associated with arc rocks and that arc-related granitoids in Sumatra derived from an isotopically similar source. Low-Ti basalts have TiO2 < 1.6% are tholeiitic and have variable major and trace element composition with the most \"depleted\" samples showing trace element composition similar to E-MORB but with negative HFSE peaks and depleted HREE. The low-Ti basalts may derive from a source similar to that of the high-Ti basalts through a combination of two processes: higher degrees of partial melting responsible for the lower content in incompatible elements and crustal contamination which accounts for the Sr Nd and Pb isotopic variations and for the correlation between isotope ratios and LILE/HFSE values. The composition of some spinel inclusions in the low-Ti basalts suggests that relatively large degrees of crustal contamination may produce melts which may have major and trace elements and Sr Nd and Pb isotopic ratios very similar to those of arc basalts. Basalts from Buldt Telor in central Sumatra are rich in mantle xenoliths (lherzolite) and resemble the high-Ti basalts of Sukadana but are overall more primitive and more alkaline and probably represent the composition of the unmetasomatised mantle wedge at some distance from the volcanic arc. Isotopically they resemble basalts of the Ninetyeast Ridge and of some Indian Ocean MORB. In general the basalts from Sukadana and Bukit Telor have trace element and isotopic characteristics similar to those of tholeiitic to mildly alkaline basalts from Kerguelen the Ninetyeast Ridge and from the eastern Australian volcanic province. Similar magmatism may also occur within the southeast part of the Eurasian plate (Karimunjawa Islands and Quaternary volcanics in central and western Kalimantan)."