Petrogenetic models for convergent margins should be consistent with the global systematics of convergent margin volcanic compositions. A newly developed tool for compiling and screening data from the GEOROC database was used to generate a global dataset of whole rock chemical analyses from arc front stratovolcano samples. Data from 227 volcanoes within 31 volcanic arc segments were first averaged by volcano and then by arc to explore global systematics. Three different methods of data normalization produce consistent results that persist across a wide range of Mg# [Mg#=Mg/(Mg+Fe)]. Remarkably coherent systematics are present among major and trace element concentrations and ratios, with the exception of three arcs influenced by mantle plumes and Peru/N. Chile, which is built on exceptionally thick crust. Chemical parameters also correlate with the thickness of the overlying arc crust. In addition to previously established correlations of Na6.0 with Ca6.0 and crustal thickness, correlations are observed among major elements, trace elements, and trace element ratios (e.g. La/Yb, Dy/Yb, Zr/Sm, Zr/Ti). Positive correlations include “fluid mobile,” “high field strength,” and “large ion lithophile” element groups, with concentrations that vary by a factor of five in all groups. Incompatible element enrichments also correlate well with crustal thickness, with the greatest enrichment found at arcs with the thickest crust. Intra-crustal processes, however, do not reproduce the global variations. High pressure fractionation produces intermediate magmas enriched in aluminum, but such magmas are rare. Furthermore, differences among magma compositions at various volcanic arcs persist from primitive to evolved compositions, which is inconsistent with the possibility that global variations are produced by crystal fractionation at any pressure. Linear relationships among elements appear to be consistent with mixing between depleted primary magma and an enriched contaminant, but the required composition of the theoretical enriched end-member is not similar to compositions expected in the deep crust or to any known rock composition. The large-scale chemical variations among volcanic arcs are therefore likely to be generated by processes in the subducting slab or mantle wedge, rather than the crust. While crustal processes are important in the differentiation of convergent margin magmas, they do not account for the systematics presented here. Models that attribute the chemical variability of arc magmas to slab or wedge processes are also constrained to be consistent with the global chemical systematics, and are discussed in Turner and Langmuir (2015).
Understanding magmatic processes such as crystallization and melting recorded in natural samples requires calibration of mineral–melt trace element partition coefficients (D) and their dependence on temperature, pressure, oxygen fugacity (fO2) and chemical composition. However, few experimental studies have focused on measuring trace element partition coefficients for a large number of trace elements, in the various minerals present in basaltic rocks, and under diverse conditions, particularly of variable fO2. Twenty-seven 0.1 MPa experiments provide partition coefficients for major elements and Sc, Ti, V, Mn, Co, Ni, Ga, Sr, Y, Nb, Ba, Ce, Nd, Eu, Gd, and Yb for the mineral phases olivine, plagioclase, orthopyroxene and clinopyroxene. The experimental conditions range from 1150 to 1190 °C with oxygen fugacities from QFM to NNO+2. Run products were analyzed by laser-ablation ICP-MS. The new partition coefficients, combined with previously published data, can be used to model crystallization processes at low pressure. Partitioning of multivalent cations V, Fe and Eu varies as a function of the redox conditions, consistent with previous work, and can be used to constrain oxidation states of magmatic source regions. The V/Yb ratio is shown to be a useful proxy for oxidation state. The V/Yb ratio varies during mantle melting as a function of oxidation state of the mantle source, and it is not modified during fractional crystallization of olivine ± plag ± cpx. V/Yb increases from MORB, BABB to arc lavas, suggesting a progressive increase of fO2 from QFM to NNO+2. Apparent fO2 of arc lavas, however, is quite variable. These results demonstrate that sub-arc mantle displays a larger range of redox conditions toward a more oxidized mantle than the MORB mantle.
A comparison of geochemical and Sr–Nd–Pb isotopic compositions for Deccan Continental Flood Basalts (CFBs) and Central Indian Ridge (CIR) Basalts is presented: these data permit assessment of possible parental linkages between the two regions, and comparison of their respective magmatic evolutionary trends in relation to rift-related tectonic events during Gondwana break-up. The present study reveals that Mid-Ocean Ridge Basalt (MORB) from the northern CIR and basalts of Deccan CFB are geochemically dissimilar because of: (1) the Deccan CFB basalts typically show a greater iron-enrichment as compared to the northern CIR MORB, (2) a multi-element spiderdiagram reveals that the Deccan CFBs reveal a more fractionated slope (Ba/YbN>1), as compared to relatively flat northern CIR MORB (Ba/YbN<1), (3) there is greater REE fractionation for Deccan CFB than for the northern CIR MORB (i.e., La/YbN∼2.3 and 1 respectively) and (4) substantial variation of compatible–incompatible trace elements and their ratios among the two basalt groups suggests that partial melting is a dominant process for northern CIR MORB, while fractional crystallization was a more important control to the geochemical variation for Deccan CFB. Further, incompatible trace element ratios (Nb/U and Nb/Pb) and radiogenic isotopic data (Sr–Pb–Nd) indicate that the northern CIR MORBs are similar to depleted mantle [and/or normal (N)-MORB], and often lie on a mixing line between depleted mantle and upper continental crust. By contrast, Deccan CFB compositions lie between the lower continental crust and Ocean island basalt. Accordingly, we conclude that the basaltic suites of the northern CIR MORB and Deccan CFB do not share common parentage, and are therefore genetically unrelated to each other. Instead, we infer that the northern CIR MORB were derived from a depleted mantle source contaminated by upper continental crust, probably during the break up of Gondwanaland; the Deccan CFB are more similar to Ocean island basalt (Reunion-like) composition, and perhaps contaminated by lower continental crust during their evolution.
The Tarim continental flood basalts (CFBs) provide important clues about the genesis and magmatic evolution of the Early Permian Tarim Large Igneous Province (Tarim LIP) in northwestern China. Here we present results of LA–MC–ICPMS Lu–Hf isotope analysis on Early Permian (ca. 290Ma) zircons extracted from the Tarim CFBs in the Keping area, northwest of the Tarim Basin. Zircons from two sub-groups of Keping basalts (Groups 1a and 1b) have similar Lu–Hf isotopic compositions and exhibit a relatively large range of 176Hf/177Hf ratios between 0.282422 and 0.282568. Their negative εHf(t) values (−6.8–−1.4) are generally lower than the whole-rock εHf(t) values of their host basalts (−2.8–2.1), and are distinct from other known intrusive rocks (−0.3–7.1) in the Tarim LIP and their hosted zircons (4.9–8.8). Systematic studies of Hf isotopic data from Tarim and its adjacent regions reveal that these zircons are probably xenocrysts, sourced from coeval igneous rocks in the South Tianshan Orogen (e.g., the Lower Permian Xiaotikanlike Formation volcanic and pyroclastic rock suite). This, together with the presence of Precambrian zircons in Keping basalts, clearly indicates crustal contamination during their eruptions and provides hints about the potential contaminant sources. Geochemical modeling further suggests that the earlier erupted Group 1b basalts experienced more contamination, predominantly by some high Th–U–Pb rock components, most likely from the South Tianshan Orogen. The later erupted Group 1a basalts in the Keping area have been less contaminated with mainly the Tarim Precambrian rocks. Another group of the Tarim CFBs in the Northern Tarim Uplift (Group 2) appears to have undergone negligible crustal contamination but possesses evidence for variable source compositions. The modeling also indicates that the uncontaminated parental magmas of various Tarim LIP rocks (from the picrites and basalts to ultramafic–mafic and syenitic intrusive rocks) exhibit a wide range of εNd(t) values (ca. −5–5), reflecting source isotopic heterogeneity, which may be a consequence of plume–lithosphere interaction during the generation of the Tarim LIP.
Detailed major element, trace element and isotopic study of the FAMOUS and North Famous segments within the geochemical gradient south of the Azores platform provides new constraints on controls on chemical variations at the segment scale and the origin of plume geochemical gradients. A comprehensive investigation of 110 samples along the entire length of the FAMOUS segment, coupled with a recent extensive melt inclusion study by Laubier et al. (2012), shows large trace element diversity within a single segment and substantial isotopic variability that largely correlates with trace element variations. Substantial variations are also present along the short (18km) North Famous segment despite the presence of an axial volcanic ridge. These results confirm multiple supply of magmas along the length of these segments, the lack of a centrally supplied magma chamber, and the ability of melting processes to deliver highly diverse melts over short distances and times. With the exception of one group of high Al2O3, low SiO2 magmas (HiAl–LoSi) largely recovered in the original small FAMOUS area, the data can be simply explained by a two-component mixing model coupled with melting variations. The HiAl–LoSi magmas reflect assimilation and mixing in the crust, an interpretation supported by the diverse melt inclusions in these lavas. Since the mantle heterogeneity reflects two-component mixing, the end members can be constrained. Surprisingly, source mixing between the Azores plume and depleted mantle cannot produce the observations. This is evident regionally from the fact that nearly all basalts have highly incompatible trace element ratios (e.g., Th/La, Nb/La) as high or higher than the most plume-influenced MORB near the Azores hotspot, despite being over 300km farther south and much less enriched isotopically. To account for the elevated highly incompatible trace element ratios, a metasomatic component formed by adding deep, low-degree melts of Azores plume material to a depleted mantle is required. The regional gradient south of the Azores then requires different processes along its length. Close to the Azores, plume material mixes with depleted mantle. The pure plume influence is spatially restricted, and enrichment farther to the south is caused by shallow mantle metasomatized by low-degree melts from deep plume flow. North Famous lavas are spatially closer to the Azores and yet are more depleted in trace elements and isotopes than FAMOUS lavas, suggesting delivery of the enriched component to individual segments is influenced by additional factors such as segment size and offset. The extent to which these processes operate in other regions of plume–ridge interaction remains to be investigated.
Permian continental flood basalts (CFBs), as a main component for Tarim Large Igneous Province (TLIP), are widely distributed in the western and central parts of the Tarim Basin. Here we present a systematic study of platinum-group elements (PGEs) combined with conventional major, trace and Sr–Nd isotopic geochemistry to characterize the Tarim CFBs from the Yingan section in the Keping area, west of the Tarim Basin. The basaltic units with a total thickness of ca. 400m in the section can be divided into two basaltic sequences: the lower Kupukuziman and the upper Kaipaizileike sequences. Both of the sequences display extremely low PGE concentrations, with Os, Ir, Ru, Rh, Pt and Pd contents of 0.014–0.106, 0.007–0.072, 0.035–0.253, 0.011–0.078, 0.043–0.149 and 0.026–0.124ppb, respectively. The Keping basalts exhibit a marked increase in Cu/Pd ratios (>105) albeit with a narrow range of lower Pd/Ir ratios (<50), different from the PGE-undepleted basalts of the Siberian Traps, Emeishan Large Igneous Province (ELIP), East Greenland CFBs and Deccan Traps. Furthermore, the Tarim CFBs in Keping and elsewhere in the basin have very low Cu/Zr ratios (<0.5 commonly), indicating that their parent magmas were significantly depleted in chalcophile elements. The Tarim CFB magmas are estimated to have formed by <5% partial melting, and were probably S-saturated during partial melting, leaving residual sulfide in the mantle source. The differences in Th/Yb and (Th/Nb)N ratios between the Kupukuziman and Kaipaizileike sequences suggest that the Keping basalts incorporated variable degrees of assimilated crustal components from the Precambrian basement in the Tarim Basin. Modeling indicates that the earlier erupted Kupukuziman lavas witnessed relatively larger extent of assimilation than the later Kaipaizileike lavas. Nevertheless, the crustal assimilation did not trigger sulfur saturation and sulfide segregation for the basaltic magma in the crust. Furthermore, a magma mixing process between the relatively primitive and more evolved residual magmas of the same lineage during magma chamber replenishment is suggested according to the PGEs and Sr–Nd isotope variations in the Kaipaizileike sequence. This process might have also induced limited sulfur saturation and sulfide segregation in the late stage of the magmatic evolution. A comparison of the chalcophile element characteristics between the basalts in the Tarim Basin and the mafic–ultramafic rocks in the Eastern Tianshan and Beishan Rift areas indicates that the degree of partial melting plays an important role on the PGE geochemistry and potential PGE–(Ni–Cu) ore deposits in the TLIP.