V13F-01 INVITED
Timescales for Reequilibration of Major Elements in Olivine-Hosted Melt Inclusions
Geochemical evidence demonstrates that near-fractional partial melts form over a range of upper mantle conditions from peridotites with variable bulk compositions. The importance of determining the initial compositions of melt inclusions lies in the ability of the major elements to record information about the conditions at which melt generation takes place, and the involvement of exotic lithologies in the melt generation process. However, before the major-element compositions of melt inclusions can be interpreted reliably, the effects of post-entrapment processes must be quantitatively understood. Diffusive reequilibration of olivine-hosted melt inclusions during cooling, magma mixing, or decompression can produce irreversible compositional changes in response to changes to the melt outside of the host olivine. Factors that influence the timescales for reequilibration of major elements in olivine-hosted melt inclusions are the sizes of the inclusion and the host olivine, which control the diffusion lengthscale, and the olivine-melt partition coefficient and diffusivity, which control the flux through the olivine grain. Among the major elements, Fe, Mg and Ca have the greatest potentials for undergoing diffusive reequilibration. The concentration of H2O in olivine-hosted melt inclusions must also be evaluated with care. Based on experimentally determined diffusivities and partition coefficients, reequilibration of Fe and Mg occurs on a timescale of years, while Ca reequilibrates over a few decades. Diffusion of H in olivine is rapid enough to allow reequilibration in a matter of hours. Important questions the must be answered in order to fully assess the reliability of the melt inclusion compositions are whether the history of Fe/Mg transport through the olivine can be uniquely determined, whether there are factors that rate limit transport of H, and whether gradients in slower diffusing elements can be used to constrain re-equilibration timescales.
V13F-02 INVITED
Detecting and Correcting Melt Inclusion Modification
Post entrapment diffusive modification of melt inclusions may mute or erase primary signatures. Corrections for post-entrapment crystallization (PEC) and Fe-loss are routinely applied and, because recent experimental studies suggest rapid diffusion of trace components into and out of olivine-hosted inclusions, the ability to discriminate between primary and secondary signatures is now even more critical. Two tools may assist in this endeavor. XANES measurements of Fe3+/ΣFe ratios in undegassed ol-hosted basaltic melt inclusions from global arcs are 16-36% (n=16), significantly higher than the 7-10% commonly assumed, and higher than in MORB or BABB lavas (Kelley and Cottrell, this mtg). The Fe3+/ΣFe ratios indicate melt-host equilibrium, with significantly less PEC or Fe-loss than would have been otherwise assumed. We conclude that Fe2+ diffusion has been minimal; therefore the residence time of these primitive inclusions in an evolved magma must have been short. Fe3+/ΣFe correlates positively with water concentration, but not with CO2 and S concentrations or Mg#. The oxidized nature of arc lavas and melt inclusions may therefore indicate an oxidized source rather than late-stage degassing or fractionation. Trace element concentrations evolve with time if an inclusion is out of equilibrium with its host. The numerical model of Cottrell et al., 2002, makes specific predictions about how suites of melt inclusions evolve, creating a tool to detect post-entrapment modification. Recent laboratory measurements of REE diffusion in olivine greatly diverge (at 1300°C, 1015 vs 1019m2/s). If REE diffusivity is extremely fast, melt inclusion HREE diversity shouldn't survive more than a few years in a magma chamber; but if slow, HREE variance could be preserved for >104 yrs. Model analysis of published suites of ol-hosted inclusions indicates that either REE diffusion is quite slow, or the residence time of melt inclusions at high temperature is very short. Loss of variance in suites of pl-hosted inclusions is consistent with long (>103 yrs) residence times. Suites of ol- and pl-hosted inclusions from the same magmatic system can therefore bracket residence times if diffusivities are known, or put reasonable bounds on diffusion rates.
V13F-03
Primitive Melt Inclusions from Multiple Samples from the FAMOUS Zone: Insights into the Mantle Melting Column and the Fractionation Processes
On mid-ocean ridges, the influential work by Sobolev and Shimizu (Nature, 1993) and Sobolev (Petrology, 1996) has inferred fractional melting during polybaric upwelling by showing that olivine-hosted inclusions were formed over a range of pressures. However melt inclusion studies have often concerned single MORB samples and may be seen as anecdotal in the sense that they are neither repeated nor globally verified. Recent modeling and experimental results also suggest the importance of post-entrapment processes for major and trace elements. This study presents major and trace element data in 300 olivine-hosted melt inclusions from 11 samples from the FAMOUS segment on the Mid-Atlantic Ridge. Published data from Shimizu (Phys. Earth Planet. Int., 1998) and Kamenetsky (EPSL, 1996; spinel-hosted inclusions) are also reported. In parallel, major and trace element measurements were performed in 150 glasses of the segment in order to have consistent datasets. Melt inclusions, trapped in olivine phenocrysts Mg#85-92, display complex trends in major element plots and can be divided into three groups. Group 1, the largest, is characterized by high MgO (9.4-13.4 wt.%), intermediate SiO2 and Al2O3 contents. Group 2 displays distinctively high Al2O3 (up to 18.4 wt.%), low SiO2 (as low as 46.5 wt.%) and high MgO (10.5-12.8 wt.%) contents, along with low CaO and variable TiO2, K2O and incompatible element concentrations. Group 3 consists of the melt inclusions trapped in less primitive olivines (Mg#<88.5) and displays higher SiO2, CaO and trace element contents. In the lava population, two groups can be distinguished. A small subset, that shares many features with the group 2 melt inclusions, displays high MgO and Al2O3 and low SiO2 and incompatible element contents. This type of lava - high-Al, low-Si and high-Mg - has been previously reported for various mid-ocean ridges (e.g., le Roux et al., Contrib. Min. Petrol., 2002; Eason and Sinton, EPSL, 2008). The second group plots along liquid lines of descent at low pressure starting from the compositions of the group 1 melt inclusions. Modeling of continuous polybaric melting and crystallization shows that the different inclusion groups are derived from melts formed at various pressures in the melting column (~12-6 kbar). After segregation from the mantle, the three batches of melts are fractionated at distinct pressures. The group 2 melt inclusions are consistent with the highest pressure of melt formation and a major role of cpx+olivine fractionation at high pressure (8 kbar), whereas group 3 results indicate the lowest pressure of extraction and entrapment (1kbar). An important observation is that high-Al, low-Si lavas contain melt inclusions from both the low-Si, high-Al group 2 and normal compositions (groups 1 and 3). These lavas can be reproduced by mixing between these two populations of inclusions, followed by some extent of differentiation. Therefore, this study shows that lavas represent averages of melts differentiated from the melt inclusions, and that the major element variability among inclusions can be explained by the combined effects of polybaric melting and crystallization at variable pressure. Trace element compositions of group 1 and 2 melt inclusions show large variations; incompatible element ratios (Ba/La, Rb/Nb, etc) suggest local source heterogeneity. Further modeling will be carried out in order to distinguish between the effects of partial melting and source composition.
V13F-04
Relative timing of crustal contamination in short-lived small-volume volcanic centers
Melt inclusion compositional variability in conjunction with inclusion host and whole rock geochemistry can provide insight into both magma source variability and the timing of contamination and (or) melt aggregation relative to crystallization. Paricutin Volcano (Mexico) is often cited as a classic example of assimilation and fractional crystallization processes with increases in whole rock SiO2 (wt%), δ18O, and 87Sr/86Sr, observed over the lifespan (1943-52) of Paricutin. Olivine and opx-hosted melt inclusions are utilized to identify discrete batches of magma and to examine the relative timing of crustal contamination during the evolution of the volcanic system. A high-SiO2 (58-65 wt%) and low-SiO2 (50-60 wt%) inclusion group are found in each lava in this study. Incompatible element ratios are similar between high- and low-SiO2 groups within a sample, and the two populations are most easily explained by late-stage fractional crystallization. High Ba/Nb in Feb 1943, distinct from later 1943-46 inclusions, supports prior evidence of an initially distinct magma batch. Melt inclusions record a rapid change in composition during 1947, characterized in part by an increase in Ba/Nb and K2O/TiO2, rather than a gradual evolution over the volcano's lifespan. Importantly, there is no compositional overlap between melt inclusions in early and late erupted lavas, indicating that the inclusions do not record concurrent assimilation and fractional crystallization characteristic of an AFC model. Comparison of xenolith glass and bulk crustal xenolith compositions, will allow us to test bulk versus partial crustal assimilation models. Melt inclusion geochemistry suggests that 1) early lavas record multiple primitive magma compositions, 2) contamination in late magmas occurred early, prior to significant crystallization, and 3) contamination was either a very rapid process, or the change in composition records a transition to erupting a more evolved and contaminated batch of magma, distinct from that erupted from 1943-46. These new data allow us to re-evaluate AFC processes in small basaltic systems as well as hypothesize about magma plumbing beneath the volcanic edifice.
V13F-05 INVITED
Experimental and Natural Evidence for Rapid Water Exchange Between Melt Inclusions in Olivine and Host Magma
We have carried an experimental study aimed at evaluating the ability of olivine to isolate chemically melt inclusions from the host magma after their entrapment. We demonstrate that nearly 'dry' (<0.5 wt% of water) melt inclusions from Galapagos Plateau basalt gain up to 2.5 wt% of water if they are placed for 2 days in a water-bearing melt at 200 MPa and 1140 °C. Amount of structurally bound water in olivine crystals also increased, maintaining equilibrium with hydrous matrix melt (D olivine-melt ~ 0.002). Despite the complete re-equilibration of the system with respect to water, no or only partial re-equilibration was achieved with respect to major, volatile (S, Cl) and incompatible trace elements between matrix melt, olivines and melt inclusions. The apparent diffusion coefficient of water in olivine is found to be more or equal to 5E-12 m2/s at our experimental conditions that is as fast as proton-vacancy diffusion and at least 3-4 orders of magnitude faster than for other major (e.g., Fe and Mg) and trace elements at dry or hydrous conditions. These results indicate that inclusions in olivine can rapidly and selectively exchange water with matrix melt, probably, through combination of proton diffusion and molecular water transport along dislocations in olivine. The rapid re-equilibration of melt inclusions with matrix melt or atmosphere can explain the decoupling of water and incompatible trace elements (e.g, water vs. potassium) reported for suites of primitive inclusions from oceanic settings (43 °N MAR, FAMOUS, Galapagos) and island arcs (Kamchatka, Central America, Cyprus). Rare cases of well preservation of initial water content in suites of co- genetic inclusions imply very short residence time (a few hours) of the olivine phenocrysts in magma with contrasting water content during fractionation and transport to the surface and rapid quenching upon eruption.
V13F-06
Revisiting the Compositions and Volatile Contents of Olivine-Hosted Melt Inclusions From the Mount Shasta Region
Recent controversy over the origin of high-Mg andesites from the Mount Shasta region has prompted us to re-examine olivine-hosted melt inclusions from this area. We analyzed olivine-hosted (Fo88-94) melt inclusions from tephra of the S17 cinder cone (N flank Whaleback), north of Mount Shasta, using FTIR and electron microprobe. The melt inclusion compositions (uncorrected for post-entrapment crystallization) range from 54-63 wt% SiO2 and 4.0-6.7 wt% MgO, and they can be divided into high (>12 wt%) and low (<10 wt%) CaO groups with distinct P2O5/K2O ratios. High-CaO inclusions have Fo91-94 host olivines, whereas low-CaO inclusions have Fo87-88 host olivines. S, Cl, and F contents are 700-2300 ppm, 1300-2700 ppm and 0-600 ppm, respectively. H2O contents (uncorrected) vary from 0.5-3.8 wt% and agree with the range of H2O contents determined by Anderson (1974 and unpublished data) using the electron probe H2O-by- difference technique (0-6.4 wt%). CO2 contents range from below detection (~25 ppm) up to 800 ppm. S and Cl contents are generally higher for high-CaO inclusions than for low-CaO inclusions, whereas F, H2O, and CO2 contents are similar for both groups. All volatiles show trends consistent with degassing during crystallization and must be taken as minimum amounts dissolved in the melt. Comparison of melt inclusions with whole rock compositions (Baker et al., 1994) suggest significant Fe-loss from the glasses (~2-4 wt%). Total extents of post-entrapment crystallization were estimated by the addition of olivine to the melt composition (with 8 wt% FeOT) until equilibrium between the melt and host olivine was achieved (13-48 wt% crystallization). Corrected SiO2 and MgO contents of the high-CaO inclusions (Fo91-95 olivine) are 49-52 wt% and 17-19 wt%, respectively. The minimum initial H2O content for inclusions in high-Fo olivines is 2.5-4.7 wt% based on our new data and Anderson's (1974 and unpublished) data. Our data provide strong evidence for the existence of volatile-rich, high-Mg basaltic andesite melts beneath the Mount Shasta region.
V13F-07
Sulfur Isotope Variation in Basaltic Melt Inclusions from Krakatau Revealed by a Newly Developed Secondary Ion Mass Spectrometry Technique for Silicate Glasses
Sulfur is a ubiquitous element with variable valance states (S2-, S0, S4+, S6+) allowing for its participation in a wide variety of chemical and biogeochemical processes. However, its potential as an isotopic tracer in magmatic processes has not been fully developed and is crucial to understanding of sulfur recycling in subduction zones and between Earth's major reservoirs, mantle, lithosphere and coupled hydrosphere-atmosphere. Previous studies of silicate glasses and melt inclusions have been hampered by lack of an in situ isotopic measurement technique with spatial resolution of 10 to 100 microns. We have developed a new secondary ion mass spectrometry (SIMS) analytical technique for measurement of 34S/32S ratios in silicate glasses utilizing the IMS 1280 at Woods Hole Oceanographic Institution. A beam of 133Cs+ ions with 13 keV energy and current of 1-2 nA is focused onto a 10 micron spot and rastered over 30 × 30 microns. A Normal Incidence Electron Gun was used to compensate excess charge. The rastered beam is then centered to the optical axis of the machine, and a mechanical aperture is placed on the image plane to limit the area of analysis to the central 15 × 15 microns. The energy slit width was adjusted to 50 eV. A mass resolving power of 5500 was sufficient for eliminating mass interferences. A suite of synthetic and natural glasses with δ34SVCDT values spanning from - 5.6‰ to 18.5‰ with SiO2 from 44-72 weight % were measured. Magnitude of the instrumental mass fractionation (α) for 34S/32S ratios is 0.991 and is constant for all the glasses measured despite their compositions. Precision of individual measurements of 34S/32S ratios is 0.4 ‰, or better. Preliminary δ34S measurements of olivine-hosted basaltic melt inclusions in pre- 1883 basaltic scoria from Krakatau volcano Indonesia vary from -5.6 to 7.9‰ with sulfur concentrations from 490 to 2170 ppm, respectively. Host olivines are Fo77-80 and inclusions generally need minor to no post-entrapment corrections using KDFe-MgOl-liquid = 0.30. Most inclusions are basaltic though a few range up to basaltic andesite. Sulfur X-ray wavelength scans of melt inclusions indicates 60% of dissolved sulfur present as SO4. Dissolved H2O (by FTIR) ranges from 2.7 to 4.1 weight %, and CO2 concentrations are currently being determined. Dissolved Cl in melt inclusions ranges from 600 to 1100 ppm. Possible correlations of δ34S values with dissolved volatiles (by SIMS, and FTIR), and trace element concentrations are being evaluated.
V13F-08 INVITED
How robust are melt inclusions in olivine for mantle study?
Melt inclusions in olivine are commonly used to address composition of mantle sources and processes of mantle melting. Here I will assess basic assumption of such studies: ability of melt inclusions to preserve chemical characteristics of parental melts formed in the mantle. Using published and new data I will demonstrate that olivines generally trap their melt inclusions during shallow crystallization in the crust by fast growth in the thermal boundary layers of mixing magma batches. Whether such trapped melts represent parental melts or they were modified by reaction on route could be recognized by chemical evidence of shallow equilibrium with plagioclase or pyroxene. Some inclusions reveal post-trapping modification of heaviest REE and H2O by volume diffusion through host olivine as predicted by the high diffusion coefficients for these elements in olivine (e.g . Spandler et al, Nature 2007 and Portnyagin et al, EPSL 2008). However, the contents of the middle and light REE and those of all highly incompatible trace elements are fully preserved. I conclude that (1) time scales in the magmatic system are short enough to preserve the abundances of highly and moderately incompatible elements inherited from primary melts, and (2) there is no evidence that wall-rock interaction has contributed significantly to the incompatible-element composition of Hawaiian melt inclusions.