Pierre Schiano

Osmium-strontium-neodymium-lead isotopic covariations in mid-ocean ridge basalt glasses and the heterogeneity of the upper mantle

Abstract Osmium, strontium, neodymium, and lead isotopic data have been obtained for 30 hand picked samples of basaltic glass from the Pacific, Atlantic and Indian mid-oceanic ridges. Large variations in Os isotopic ratios exist in the glasses, from abyssal peridotite-like values to radiogenic compositions similar to oceanic island basalts (187Os/186Os and 187Os/188Os ratios range from 1.06 to 1.36 and from 0.128 to 0.163, respectively). Os isotopic and elemental data suggest the existence of mixing correlations. This relationship might be ascribed to secondary contamination processes; however, such a hypothesis cannot account for the negative correlation observed between Os and Nd isotopes and the existence of complementary covariations between Os and Sr Pb isotopes. In this case, Os Sr Nd Pb isotopic variations are unrelated to late post-eruption or shallow level contamination. These relationships provide strong evidence that the Os isotopic composition of the samples are derived from the mantle and thus implies a global chemical heterogeneity of the oceanic upper mantle. The results are consistent with the presence of recycled oceanic crust in the mantle sources of mid-ocean ridge basalts, and indicate that the unique composition of the upper mantle below the Indian ocean results from its contamination by a mantle component characterized by radiogenic Os and particularly unradiogenic Nd and Pb isotopic compositions.

Oxygen-isotope evidence for recycled crust in the sources of mid-ocean-ridge basalts

Mid-ocean-ridge basalts (MORBs) are the most abundant terrestrial magmas and are believed to form by partial melting of a globally extensive reservoir of ultramafic rocks in the upper mantle. MORBs vary in their abundances of incompatible elements (that is, those that partition into silicate liquids during partial melting) and in the isotopic ratios of several radiogenic isotope systems. These variations define a spectrum between ‘depleted’ and ‘enriched’ compositions, characterized by respectively low and high abundances of incompatible elements. Compositional variations in the sources of MORBs could reflect recycling of subducted crustal materials into the source reservoir, or any of a number of processes of intramantle differentiation. Variations in 18O/16O (principally sensitive to the interaction of rocks with the Earths hydrosphere) offer a test of these alternatives. Here we show that 18O/16O ratios of MORBs are correlated with aspects of their incompatible-element chemistry. These correlations are consistent with control of the oxygen-isotope and incompatible-element geochemistry of MORBs by a component of recycled crust that is variably distributed throughout their upper mantle sources.

Osmium isotope disequilibrium between mantle minerals in a spinel-lherzolite

Rhenium (Re)–osmium (Os) isotope and elemental data have been obtained for coexisting silicates and sulphide from a spinel-lherzolite from Kilbourne Hole, New Mexico. These results confirm that sulphide dominates the Os budget in mantle rocks, but a significant proportion of the Re is located in the silicates. The silicates and interstitial sulphide yield indistinguishable Os isotope compositions suggesting these phases were in equilibrium at the time of xenolith eruption. However, sulphide inclusions in silicate minerals preserve significantly less radiogenic Os isotope compositions, suggesting that they have been shielded from reaction or diffusion by their silicate hosts. Diffusion rates in silicates and sulphides are likely to be rapid at mantle temperatures. However, for the sulphide inclusions diffusion through the silicate host is likely to be impaired by the high partition coefficient for Os between sulphide and silicate phases, in the range of 104 to 106. Consequently, sulphide inclusions trapped in silicates may be unable to equilibrate either with other minerals or with an intergranular melt, and in principle may preserve significantly older Re–Os age information than coexisting phases.

Cogenetic silica-rich and carbonate-rich melts trapped in mantle minerals in Kerguelen ultramafic xenoliths: Implications for metasomatism in the oceanic upper mantle

In an attempt to characterize metasomatic agents for the oceanic upper mantle, we have undertaken a study of melt and fluid inclusions trapped in metasomatized peridotite nodules (anhydrous spinel lherzolites and harzburgites) from the Kerguelen Islands (southern Indian Ocean). These xenoliths contain three types of genetically related inclusions hosted by olivine, clinopyroxene and orthopyroxene. These are silicate melt inclusions, carbonate-rich inclusions and CO2 fluid inclusions. These inclusions are secondary in nature and form trails along fracture planes in the sheared peridotites. Heating experiments conducted on silicate melt inclusions give an estimation of the entrapment temperatures ( ≈ 1250°C) and indicate that there is no genetic relationship between the inclusions and their host minerals. The chemical composition of the silicate melt inclusions is characterized by normative quartz and feldspar components, with SiO2 ≈ 60wt%, Al2O3 ≈ 20wt%, Na2O and K2O each ≈ 4–5wt%, FeO and MgO 1000 ppm, H2O ⩾ 1.2%, and oversaturation of the melt with CO2. The trace element signature is characterized by LREE enrichment, negative HFSE (Ti and Zr) anomalies and a TiZr value of 17. The trapped melt has crystallized the following minerals: K-rich amphibole (kaersutite), diopside, rutile, ilmenite and carbonate (magnesite). Carbonate-rich inclusions, interpreted as trapped carbonate melt, have crystallized calcite. The carbonate-rich inclusions are often physically connected with the silicate melt inclusions, indicating the former existence of a homogeneous melt which later unmixed into two separate melts. Cogenetic relationships between CO2 inclusions and both carbonate melt inclusions and silicate melt inclusions yield a minimum trapping pressure for all types of inclusions of 12.5 kbar at 1250°C, corresponding to upper mantle depths. Based on their daughter mineral types, their chemical composition and high volatile element contents, the silicate-carbonate melt inclusions trapped in the ultrabasic xenoliths of the Kerguelen Islands are interpreted as small amounts of a metasomatic melt phase. These melt inclusions cannot result from melting of the anhydrous peridotite assemblages in which they have been trapped. They must represent an exotic, migrating metasomatic melt phase in the oceanic lithosphere below the Kerguelen Islands.

Assessment of the Zr/Hf fractionation in oceanic basalts and continental materials during petrogenetic processes

In order to evaluate the widely accepted assumption that Zr/Hf ratios are uniform and chondritic (i.e. equal to 36.6) in terrestrial rocks [Jochum et al., Geochim. Cosmochim. Acta 50 (1986) 1173–1183], precise Zr/Hf measurements on oceanic basalts, continental materials and chondrites have been obtained by isotope dilution technique using thermal ionisation mass spectrometry and magnetic sector-multiple collector ICP–MS. The results indicate that Zr/Hf ratio may substantially fractionate during petrogenetic processes. A well-defined negative correlation observed between Sc concentrations and Zr/Hf ratios indicates that during fractional crystallisation, the latter are controlled by the precipitation of clinopyroxene. Although clinopyroxene is the major phenocryst phase, minor mineral phases such as sphene and amphibole must be taken into account to explain the fractionation of highly evolved alkaline suites. On the other hand, the comparison between mid-ocean ridge basalts (MORB) and oceanic island basalts (OIB) suggests that DZr<DSm<DHf<DEu during partial melting. Finally, after filtering the data for such fractionation effects, we observe that MORB and continental material (with the exception of granites) display relatively uniform chondritic Zr/Hf ratios, ranging from 35.41 to 38.37 and from 36.28 to 38.71, respectively. This result implies that the extraction of the continental crust from the initially primitive mantle did not result in large Zr/Hf variations. By contrast, OIB are characterised by distinctively higher Zr/Hf ratios, ranging from 36.86 to 43.93, and this may strongly constrain the addition of plume-related material to the chemical budget of the continental crust. Moreover, residual garnet influence will result in the generation of OIB with lower Zr/Hf ratios and recycling of continental and oceanic crustal materials in the sources of OIB will not strongly modify their Zr/Hf ratios. These observations thus give additional support to our explanation that the variation of Zr/Hf ratios in OIB mainly reflect the melting process.

Primitive mantle magmas recorded as silicate melt inclusions in igneous minerals

Abstract This paper reviews work on primitive silicate melt inclusions in basalt phenocrysts and evaluates its significance. Primary melt inclusions are small quantities of silicate melt included in minerals during coarsening, growth or recrystallisation of the crystal structure. Because they contain liquids formed in thermodynamic equilibrium with their host minerals, primary melt inclusions produced at different stages of evolution of the melts will record the liquid line of descent of magmatic systems. Moreover, assuming they become closed to the surrounding systems since the time of entrapment, primary melt inclusions in early-formed crystals may isolate pristine samples of high-pressure mantle-derived melts that were trapped prior to mixing at shallower levels. They can thus be used to obtain information on the processes that create magmas and the nature of their mantle source regions. The first part of the review focuses on the general characteristics of silicate melt inclusions. Several aspects of the evolution of silicate melt inclusions after formation are investigated, in particular (1) the effect of elastic deformation of the host phase on the evolution of pressure inside an inclusion; (2) the changes that may occur in the inclusions during cooling, including the separation of immiscible fluid phases and crystallisation of new minerals; and (3) the efficiency of the melt inclusion isolation from the influence of the external system. The second part documents inferences derived from the composition of primitive melt inclusions in basalt phenocrysts, on mantle source compositions, melting and melts transportation. In order to discuss these results, two examples are detailed. First, the trace element compositions of primitive melt inclusions in early-formed phenocrysts in mid-ocean-ridge basalts are consistent with those expected for small melt fractions of peridotite, progressively depleted by fractional melting with a small threshold porosity and followed by high-level mixing. Secondly, the compositional changes in primitive melt inclusions in basalts from Mount Etna volcano (Eastern Sicily, Italy) reflects a progressive transition from a predominantly mantle-plume source to an island-arc magmatic source.

Transfer of sulfur in subduction settings: an example from Batan Island (Luzon volcanic arc, Philippines)

Sulfur abundances have been determined in silicic and basaltic melt inclusions in olivines from harzburgitic xenoliths and a basaltic lava sample, all from Batan Island. In mantle xenoliths, olivines (Fo80–91) are present as neoblasts or in finely recrystallized patches. The most magnesian olivines (Fo89.7–91, CaO 85, CaO = 0.25 wt%) and their melt inclusions (CaO/Al2O3 from 0.8 to 1.15) have recorded early stages of crystallization. The sulfur concentrations for these calc-alkaline basaltic melts are estimated between 1720 and 3200 ppm, with a mean value at 2550 ppm (1σ=390) and S/Cl ratio at nearly 1. This is in agreement with the idea that arc basaltic melts may contain high concentrations of sulfur (S > 2000 ppm), at 1200°C. However, the heterogeneous distribution of S and its partitioning between silicate melts, H2O-rich vapor and S-bearing solid phases as illustrated by the Batan mantle xenoliths would result in highly variable sulfur concentrations in island arc basaltic magmas, mostly controlled by fO2 and fS2.

The distribution and behaviour of rhenium and osmium amongst mantle minerals and the age of the lithospheric mantle beneath Tanzania

Rhenium–osmium (Re–Os) isotope and elemental abundances have been obtained for primary mantle minerals, metasomatic phases, and a range of mantle rock types from xenoliths in recent volcanics in northern Tanzania. Re and Os abundances for sulphide and coexisting silicates in garnet lherzolites from Lashaine confirm that sulphide dominates the Os budget, but also show that Re is almost exclusively sited in silicate phases. Silicate minerals from two different samples yield 187Re–188Os ages of 15.4±6.1 and 31.4±6.3 Myr, respectively. Comparison with 232Th–208Pb (267.1±4.4 Myr) 147Sm–143Nd (164±18 Myr) and 87Rb–87Sr (in equilibrium at the present-day) ages for the same silicate minerals suggests differential closure between these isotope systems, and a closure temperature of ≥670°C for the Re–Os system. Remarkably, sulphide inclusions were not affected by diffusional equilibration between the silicates, and preserve significantly older age information. Model calculations suggest that sulphide–silicate equilibration ceased some 200–300 Ma, and the Os isotope composition of the sulphide (187Os/188Os=0.10432±0.00013) suggests a minimum age of 3.4 Gyr. Most xenoliths possess Os isotope compositions that are less radiogenic than the present-day chondritic mantle indicating that they experienced Re-loss some time ago. Samples showing evidence for modal metasomatism have high Re concentrations and Re/Os ratios, but their relatively unradiogenic Os isotope compositions suggests that the metasomatism occurred recently, consistent with data for metasomatic vein minerals. In contrast, some dunites possess both high Re/Os ratios and radiogenic Os isotope compositions. These samples differ from those affected by modal metasomatism in having low Re and exceptionally low Os concentrations. These results provide quantitative constraints on the distribution of Re and Os amongst mantle minerals, highlight the potential of Re–Os isotope dating of sulphide inclusions for establishing the early history of mantle mineral assemblages, and demonstrate that mantle processes themselves (metasomatism and dunite formation) can significantly modify the Os isotope chemistry of mantle rocks.

Large-scale silicate liquid immiscibility during differentiation of tholeiitic basalt to granite and the origin of the Daly gap

The dearth of intermediate magmatic compositions at the Earths surface, referred to as the Daly gap, remains a major issue in igneous petrology. The initially favored explanation invoking silicate liquid immiscibility during evolution of basalt to rhyolite has lost support because of the absence of any firm geological evidence for separation of Fe- and Si-rich liquids in igneous rocks. This work presents a record of large-scale magmatic differentiation due to immiscibility in the tholeiitic Sept Iles intrusion (Canada), one of the largest layered plutonic bodies on Earth. Gabbroic cumulate rocks from the Critical Zone of this intrusion show a bimodal distribution in density and P2O5 content, despite identical major element chemistry of the principal magmatic phases. Immiscibility is supported by the presence of contrasting Fe-rich and Si-rich melt inclusions trapped in cumulus apatite. Phase diagrams and well-documented occurrences of small-scale immiscibility confirm that liquid-liquid unmixing and the separation of Fe-rich and Si-rich liquids may contribute significantly to the Daly gap along the tholeiitic liquid line of descent.

On the preservation of mantle information in ultramafic nodules, glass inclusions within minerals versus interstitial glasses

This study questions the assumption that silicate melts preserved as glass inclusions in minerals and as interstitial films or pockets in mantle xenoliths have identical chemical compositions to one another and are both suitable for inferring deep mantle melt compositions. Theoretical models of the elastic behavior of melt inclusions indicate that only limited decompression of the inclusion takes place during ascent of the host xenolith, whereas the pressure of an interstitial melt follows the external pressure. Consequences of such behavior are considered using simple model systems, such as the alkali-bearing system forsterite–nepheline–SiO2. With decreasing pressure, phase boundaries shift to silica normative compositions. Consequently, reequilibration of small-degree melts of peridotite, which are characterized by olivine–nepheline normative compositions at moderate pressure, yield quartz normative compositions at a lower pressure. Also, melt inclusions are simpler systems, i.e. the trapped melts are in contact with a single mineral phase, in contrast with interstitial glasses. We show here that experimental heating of the inclusions restores the original melt composition of melt inclusions prior to cooling, which is not possible for interstitial glasses. Data obtained for rehomogenized glass inclusions and interstitial glasses associated in the same xenoliths from intraplate ocean islands and subduction zones confirm the model predictions. In intraplate nodules, the highly silicic, alkali-rich melts preserved as glass inclusions inside minerals are olivine–nepheline normative and represent high-pressure (1 GPa) near-solidus melts in equilibrium with a peridotitic assemblage. In contrast, the silica-normative composition of the interstitial glasses records their last pressure of equilibration, i.e., shallow-level conditions. In subduction zone settings, exsolution of the oversaturated H2O-rich volatile phase and decompression cause conflicting effects on the composition of the melts trapped in xenoliths. The composition of glass inclusions is characterized by higher levels of volatile elements, mainly H2O, and by a higher quartz norm than interstitial glasses. This is consistent with the hypothesis that the glass inclusions represent quenched melts from H2O-saturated peridotite or amphibolite/eclogite systems, whereas the composition of the interstitial glasses indicates reequilibration at low pressure, probably induced by the H2O-rich volatile loss