V43A-1415 1340h
High Ferric Iron Content of Diopsides From Cape Verdian Lavas and Dykes; Implications for the Oxidation State of the Lower Mantle.
The oxidation state of bulk rock basalts is not necessarily reflecting the oxidation state of the source regions because pre- and post-eruptive redox modifying processes are likely to have influenced bulk rock chemistry. To circumvent this problem, the ferric iron content of clinopyroxenes separated from 10 Cape Verdian rocks - six lavas and four dykes - were analyzed with $^{57}$Fe moessbauer spectroscopy at 295 K. In addition, four of the pyroxenes were analyzed at 15 K. A fresh well-crystallized dyke rock was selected to represent the oxidation state of the magma. The distribution coefficient Fe$^{3+}$/Fe$^{2+}$ between rock and clinopyroxene was found to be 1, and is thought to be a good approximation. The spectra reveal that 38 - 45% of the iron atoms in the clinopyroxenes are ferric; a ferric iron percentage considered large for mantle rock. The Fe$^{3+}$/Fe$_{tot}$ ratio calculated from micro probe analyses, all overestimate the true Fe$^{3+}$/Fe$_{tot}$ of the clinopyroxenes by up to 20%. The bulk host rock basalts exhibit very large variations in redox state as expressed in a Fe$^{3+}$/Fe$_{tot}$ ratio variations between 0.3 and 0.9, but this is not accompanied by a similar redox variation in the clinopyroxenes; an indication of the stability of the Fe$^{3+}$/Fe$_{tot}$ ratio after its incorporation in the crystal lattice. Thus, the oxidized character of the clinopyroxenes possibly reflects a high-level oxidized state of the lower mantle, which is believed to be the source region of the Cape Verde Mantle Plume. This is in concert with experimental studies in recent years of high pressure Al-rich perovskite having a possible Fe$^{3+}$/Fe$_{tot}$ as large as 0.7, i.e that lower mantle could be a considerable sink for Fe$^{3+}$. The result suggests that lower mantle has been subject to addition of ferric iron since segregation of the core.
V43A-1416 1340h
Mantle Redox Conditions in the North Atlantic Igneous Province
The North Atlantic igneous province (NAIP) has long been viewed as a region of anomalous mantle upwelling related to plume activity, continental rifting, and a heterogeneous mantle source. Prior to continental rifting in the Tertiary, the northern portion of the region was the site of closure of the Iapetus ocean basin. This tectonic event may have contributed to heterogeneities within the upper mantle and altered its oxidation state relative to the ambient mantle. Vanadium has been shown to be a useful indicator of redox conditions due to its multiple valence states (e.g. [1-2]). In mantle minerals, vanadium becomes increasingly incompatible under more oxidizing conditions [3]. Because both scandium and vanadium are moderately incompatible during melting, the Sc/V ratio of primitive basalts can be used to investigate the oxidation state of the mantle [1-3]. We have examined the Sc/V ratios of primitive lavas from the mid-Atlantic ridge (MAR), Iceland, and the East Greenland margin to determine if there are spatial or temporal variations in the oxidation state of the NAIP mantle. The Sc/V ratios for MAR basalts are 0.13-0.20 (GEOROC chemical database); while Icelandic basalts range from 0.10-0.25 with an average of 0.16 (1 $\sigma$=0.05). The entire range of Sc/V ratios of the Paleogene East Greenland basalts is 0.07-0.17 with an average of 0.10 (1 $\sigma$= 0.05). The Sc/V ratios of Icelandic basalts are similar to MAR basalts, but the East Greenland lavas are distinctly lower than both the MAR and Iceland. The Sc/V ratio also can vary as a function of mean pressure of melting (i.e. spinel versus garnet lherzolite). To test the relative importance of melting systematics, source composition, and oxygen fugacity on the Sc/V systematics for NAIP basalts, we incorporated the oxygen-fugacity-dependent V mineral-melt partitioning data of [3] into the polybaric decompression melting model REEBOX [4]. The best-fit model parameters for the majority of the Iceland and MAR basalts constrain the oxygen fugacity of the mantle for the modern ridge system to be one log unit below the Ni-NiO buffer. To model the entire range of East Greenland lavas requires that the Paleogene mantle source was $\sim$0.5-0.8 log units more oxidized than the Iceland source. These differences may be attributed to a change in the composition of the Iceland plume or reflect the involvement of metasomatized upper mantle associated with Iapetus subduction in the formation of the East Greenland basalts. [1] Canil, D., 1997, Nature 389, 842-845; [2] Canil, D. 1999, Geochem. Cosmochim. Acta 63, 557-572; [3] Canil and Fedortchouk, 2000, J. Geophys. Res. 105, 26003-26016; [4] Fram and Lesher, 1993, Nature 363, 712-715
V43A-1417 1340h
Ferrous/Ferric Ratios in 1984 Mauna Loa Lavas: A Contribution to Understanding the Oxidation State of Hawaiian Magmas
The oxygen fugacity of basaltic magma is a fundamental intensive variable that controls the iron redox state of the melt, and has a strong influence on the sequence and composition of minerals that crystallize from a cooling magma, and therefore on the composition of a fractionated melt. Additionally, the oxygen fugacity of basaltic magma is thought to reflect the oxygen fugacity of the mantle source or, at the very least, to place an upper limit on that of the source. In Hawaiian magmas, the oxygen fugacity is widely accepted as being close to the FMQ (fayalite-magnetite-quartz) buffer. This assumption is based largely on analyses of lavas from Kilauea volcano. There are very few published measurements on lavas from the other Hawaiian volcanoes. Ferric/ferrous ratios have been used to estimate the oxygen fugacity of lavas erupted in 1984 on Mauna Loa Volcano, Hawaii. Rapidly quenched lavas erupted close to vents are less oxidized than rapidly quenched lavas scooped from lava flows away from the vents, or samples that have cooled slowly. These results demonstrate that sampling is of critical importance in determining the oxidation state of a lava. The oxidation state of the vent lavas are below, or at, MW (magnetite-wüstite), and are significantly lower than that previously reported for Hawaiian lavas (about FMQ). Similarly, rapidly quenched lavas from the ongoing Kilauea eruption and Loihi seamount, all have oxygen fugacities that are close to MW and on the low side of the range previously reported for Hawaiian lavas. From this we conclude that the initial oxygen fugacity of parental Hawaiian magmas is close to MW, not FMQ, and that previous estimates of the oxidation state of Hawaiian lavas may be suspect, having undergone subaerial or near-surface oxidation during eruption. The implications are that the plume source of these magmas is also at or below MW, but not as reduced as the mantle source of mid-ocean ridge basalts. Additionally, Mauna Loa lavas appear to be slightly more reduced than Kilauea or Loihi lavas, perhaps indicating heterogeneous oxidation within the Hawaiian plume.
V43A-1418 1340h
{\it f}O$_{2}$ Determination by Sulfur K$\alpha$ Peak Shift and Olivine-Spinel Oxygen Barometry: Implications for Mantle Wedge Heterogeneities in the Cascadia Subduction Zone
Oxidation states of primitive, subaerially-erupted basalts, in the Oregon segment of the Cascade volcanic arc, have been determined by S K$\alpha$ peak shift measurements and olivine-spinel oxygen barometry. Olivine-hosted melt inclusions, with sulfur concentrations ranging from 0.080 to 0.503 wt.% S, have oxidation states varying from -0.8 to +1.8 log units, relative to the QFM oxygen buffer. High-K basalts have sulfur concentrations from 0.27 to 0.503 wt.% S with corresponding oxygen fugacities of +1.2 to +1.8 log units ($\Delta$QFM). Calc-alkaline basalts and OIB-like basalts have a greater range in log {\it f}O$_{2}$ (-0.8 to +1.0 $\Delta$QFM), significantly lower than the high-K basalts. Olivine-spinel oxygen barometry corresponds reasonably well with {\it f}O$_{2}$ determinations by S K$\alpha$ peak shift measurements. Preliminary analysis of olivine-spinel pairs from another high-K forearc lava suggests oxidation states may reach values greater than +3 log units ($\Delta$QFM). Additionally, high oxidation states have been calculated for several olivine-spinel pairs from OIB-like lavas. Coupled with increased sulfur and chlorine concentrations in these lavas, this may suggest the influence of a more oxidized source region, and/or metasomatic agent, for the generation of these magmas. Although previous studies have indicated that the oxidation states of basalts may not be truly representative of the mantle oxidation state, the relative difference between oxygen fugacities of various basalts may correlate to heterogeneities in the source region. This is expressed, in terms of {\it f}O$_{2}$, as a dichotomy between the groups of lavas. The high {\it f}O$_{2}$ suggested for the Oregon Cascade forearc lavas may result from metasomatic enrichment of a depleted mantle source. This is similar to recent modeling presented for the forearc region of the Mexican Volcanic Arc, and supported by the presence of relatively oxidized harzburgite xenoliths from the southern Washington Cascade backarc, believed to result from metasomatic enrichment of previously depleted peridotite.
V43A-1419 1340h
Platinum-group elements and oxidation condition of Paleozoic sub-continental lithospheric mantle in southernmost South America: Xenolith study in the Pali-Aike Region, Chile
Mantle xenoliths are abundant in Quarternary alkaline basalts in the Pali-Aike area of southernmost South America. The area is in the Andean back-arc region and most samples were derived from the depth of $>$60 km. There are two types of mantle xenoliths: cumulates of partial melt and mantle residues. The former are pyroxenites (orthopyroxenite, websterite) and contains significant contents of globular sulfides ($>$0.1 %), Mg-ilmenite ($>$ 1 %), and Phl ($>$ 2 %). The bulk rocks are low in PGE, 2.20-15.5 ppb, and show positive-sloped mantle-normalized patterns, similar to host basalts. This confirms the incompatible nature of Pt group PGE. Ol websterite, which shows low P and T, $\sim$ 15 kb and $780\deg$C, has the highest Mg\# (0.74) and lowest Cr\#sp ($\sim$0.186) and has fO2 similar to FMQ buffer, representing a relatively shallow cumulate of partial melt. The mantle residues are Grt lherzolite, Grt-Spl lherzolite and Grt-Spl harzburgite. Grt lherzolite and Grt-Spl lherzolite have high CaO (2.43-3.33wt %) and Al2O3(3.14- 4.18 wt %), whereas Grt-Spl harzburgites are low in CaO (0.99-1.21wt %). Sulfides are rare, and occur as inclusions in Ol and Opx, and as films along boundaries of silicate and oxide minerals. Some harzburgites are modally metasomatized by partial melt, forming Phl and Prg amphibole. The melt itself was solidified into ilmenite- and sulfide-rich pyroxenite veinlets. The calculation of fO2 using the Sp-Ol-Opx equilibria shows that the fertile garnet-bearing peridotites are reduced, $\Delta$fO2 ranging from FMQ-0.50 to -0.20 with the Cr\#sp from 0.29 to 0.30, similar to those in oceanic peridotites. Depleted harzburgites have slightly elevated, but comparable fO2 (FMQ-0.36 -FMQ+0.39) and Cr\#sp (0.28-0.33). The mantle residues contain total PGE ranging from 6.92 to 22.1 ppb, slightly lower than the primitive mantle value, but show flat normalized patterns. Metasomatized harzburgites contain comparable Cu and total PGE contents as anhydrous peridotites. The data suggest that the metasomatism by partial melt originated from the underlying asthenosphere was not accompanied by significant change in PGE and redox state.
V43A-1420 1340h
Interaction Between Magma Fluids and Lithosphere Rocks Under Crest Zone of MAR: Mineralogical and Petrophysical Consequences
Using numerical and physical experiments dynamics of mass-change in the lithosphere under the zones joining rift valleys of MAR and transform faults was modeled. `Black smokers', methane gas flows, and bubbly carbon deposits, as products of hydrocarbon condensation, present in these zones. Numerical experiments were completed using flow-reactor scheme of PC Selector Win for gas flows of compositions: C (0.1-4), O (0-2), H (0.5-4), Cl (0.05-0.5), F (0.01-1), S (0.01-0.1), and N (0.02-0.1). Weight fraction of gas mixture in rocks was 1.5-0.01%, P from 45-10 kbar to 30-100 bar, T=$1200-400\deg$C. The fluid-rock interaction time was t=1-100 steps. Density change for new-formed rock in the lithosphere profile was estimated by virtual mineral composition recounting for each time step. Verification of physicochemical model was carried out by comparison of changed rocks and numerically obtained condensates, as well as minerals and solid, gas and liquid carbon phases, obtained experimentally using the equipment to study catalytic conversion of synthesis-gas flow (H$_{2}$=65%, CO=34.8%, N$_{2}$=0.2% vol.). It was shown that above the boiling boundary of basic liquids a field of convective mass transfer should form in the lithosphere. This field includes a number of zones of initial rock change with regions of solid phase depleting and condensing. The ranges of rock composition change due to `reduced' and oxidized' gas mixtures were studied. The density change for ultra-basic rock in the lithosphere is related to spatial and time change of oxygen potential, which current values at the beginning of the interaction process are buffering by rocks, and then - by the values at the system input. In the case when reduced gas mixtures exist, an oxidation roll is forming in the flow, for oxidized mixtures - a reduction roll. At the fluid output at the sea bottom their composition is the most oxidized. When fluids and initial rocks of the lithosphere interact, changed rock mixtures of anomalously high density appear (30-60% higher as compared to the initial value). These mixtures appear under the temperatures t$_{55-100}$=$750-650\deg$C (reduced) and $675-500\deg$C (oxidized). For complex mineral columns the densities of rocks appearing when interacting with the reduced fluids are higher than those due to oxidized (not less than 3-4%). Rock densities for t$_{0}$, when interacting with the reduced fluids for the temperature interval $1100-900\deg$C, are 2-4% higher. For the temperature interval $900-750\deg$C the density inversion for changed rocks was found, the density being decreased to 5-8% for t$_{5-35}$ and increased to 2-4% for t $>$ t$_{50}$ relative to t$_{0}$. Relative densities of the rocks due to interaction with the reduced fluids for the above-mentioned temperature range are 2-4% less than the initial ones. The least densities were found for the temperature range $900-750\deg$C. Above the boundary of the magma fluid source within the depleting field the density changes are orthogonal for both reduced and oxidized fluids. Here the reduced fluid influence causes first the density decrease, and then its anomalous increase, with the transition though zero for the time range t$_{5-55}$. For the oxidized fluids the density decreases to about 4-5% relative to zero from high to low temperatures. This work has been completed with the financial support of the RFBR (Grant N 04-05-64107), the Integration Project of the Presidium of SB RAS (Grant N 6.1.1).
V43A-1421 1340h
High-T Non-Stoichiometry of Titanomagnetite - An Experimental Re-Examination
Despite its usually modest modal abundance, titanomagnetite (Tmt, magnetite-ulvoespinel solid solution) is an important mineral in igneous rocks, because it is a major carrier of rock magnetism and yields valuable information about temperature and redox conditions during the magmatic stage. Like other spinels, Tmt ((Fe,Ti)$_{3-d}$O$_{4}$) can depart from stoichiometry at high T, which influences both its thermobarometric and magnetic properties. The current formulations of the Fe-Ti oxide geothermobarometer do not take into account the Tmt non-stoichiometry and there is only a limited set of data concerning its influence on the Curie temperature. Previous studies on high-T Tmt non-stoichiometry (e.g. Hauptman, 1974) disagree regarding the vacancy concentration (v.c.) and its dependence on the ulvoespinel content (X$_{Usp}$). Our aim is to quantify the maximum v.c. as a function of T and X$_{Usp}$ with different methods. To obtain Tmt with max. v.c., we have synthesized assemblages of Tmt with ilmenite-hematite$_{ss}$ (Ilm$_{ss}$) in the system Fe-Ti-O at $1100\deg$C -$1300\deg$C, 1 bar, and $\Delta$FMQ = -4 to +5. Fragments of synthesis products were sealed in evacuated silica glass ampoules and annealed at $950\deg$C. Due to vacancy relaxation in Tmt, annealing produces Ilm$_{ss}$ exsolution lamellae and rims, i.e. a net increase in the Ilm$_{ss}$ proportion (Lattard, 1995). For comparison, we have also synthesized Tmt with wuestite (Wue, Fe$_{1-x}$O) (min. v.c.). V.c.'s have been estimated from modal abundances and phase compositions (EMP) of annealed vs. high-T synthesis samples and from Fe$^{3+}$/Fe$_{tot}$determinations using Fe L$_{2,3}$-edge electron energy-loss spectroscopy (EELS), combined with EMP analyses. We have also considered lattice parameters (a$_{0}$ decreases with increasing v.c. at constant X$_{Usp}$, e.g. Senderov et al., 1993). Our current results suggest that Tmt(+Ilm$_{ss}$) is close to stoichiometry at X$_{Usp}$$<$0.7 at all investigated temperatures. At X$_{Usp}$$>$0.7 v.c. increases with increasing X$_{Usp}$ and T. At X$_{Usp}$=0.81 and $1300\deg$C, the maximum v.c. amounts to 1.3\pm$0.5 cat$%$, corresponding to $\delta$x10$^{2}$=3.8\pm$1.6 (EELS). These values are in agreement with Senderov et al. (1993) and are also supported by annealing results and lattice parameters. First results for Tmt(+Wue) indicate significant concentrations of cation interstitials; e.g. 0.6\pm$0.8 cat% ($\delta$x10$^{2}$=-1.9\pm$2.5) for X$_{Usp}$=0.80 at $1300\deg$C (EELS). Our results are essentially in agreement with Senderov et al. (1993) concerning both the max. v.c. and its dependence on X$_{Usp}$ but disagree with Aggarwal & Dieckmann (2002), especially regarding X$_{Usp}$ dependence. This might be due to the fact that both our results and those of Senderov et al. (1993) were obtained on quenched samples, while Aggarwal & Dieckmann (2002) have used in-situ thermogravimetry. We cannot exclude that upon quenching vacancy-rich Tmt adjust to oxygen-poorer, Fe-richer Tmt. In this case, Tmt should exsolve very fine lamellae of Ilm$_{ss}$. Careful examinations with SEM, EMP and TEM did not reveal any evidence for such quenching phenomena. To improve quantification of Tmt non-stoichiometry we will intensify EELS examinations, which have emerged as being most reliable. We will also carry out EMP oxygen analyses on samples in the system Fe-Ti-O and on Mg- and Al-bearing Tmt (relevant for natural compositions). Aggarwal S., Dieckmann R. (2002) PCM 29, 695-706. Hauptmann Z. (1974) GJRAS 38, 19-47. Lattard D. (1995) AM 80, 968-981. Senderov E., Dogan A.U., Navrotsky A. (1993) AM 78, 565-573.
V43A-1422 1340h
Pressure-Induced Valence Change and Hydrogen Disorder in Fe(OH)$_{2}$
Vibrational and M\"{o}ssbauer spectroscopy reveal that pressure causes the hydrogen lattice of brucite-structured Fe(OH)$_{2}$ to become disordered above 10 GPa, leading to self-oxidation of Fe$^{2+}$----$^{P}$----$>$Fe$^{3+}$$+{\it e}$^{-}$. This result documents one mechanism by which water (hydrogen) dissolved in the Earth's mantle can induce changes in valence and chemical bonding at depth, regardless of buffering by other mineral phases. Infrared absorption (IR) and Raman (R) spectroscopy at 7-21 GPa and 293 K show that the A$_{2u}$ (IR) and A$_{1g}$ (R) OH-stretching modes decrease with pressure (-1.3 $\pm$ 0.1 cm$^{-1}$/GPa and -4.9 $\pm$ 0.2 cm$^{-1}$/GPa, respectively) in a manner expected for hydrogen-bonded systems. Both modes show non-linear broadening, with a $\sim$ 4-fold increase in width over the experimental pressure range. This is interpreted as a signature of the hydrogen-ion positions becoming disordered, as previously proposed for the isostructural Mg(OH)$_{2}$, Ca(OH)$_{2}$ and Co(OH)$_{2}$, and attributed to the effects of H - H Coulombic repulsion caused by compression along the {\it c}- crystallographic axis. In the case of Fe(OH)$_{2}$, M\"{o}ssbauer spectroscopy reveals the internal (self-) oxidation of the Fe site at starting at similar pressures as the hydrogen disordering. Displacement (and disordering) of H alters the Coulomb potential at the iron site, thus furnishing the driving force for delocalization of the weakly bonded 3d electron of Fe. Once the electron is lost from Fe, the site remains oxidized on decompression apparently because of the affinity of iron for the 3+ oxidation state. Thus, the hydrogen disordering is observed to be reversible upon decompression, whereas the self-oxidation is not. Such changes in valence can strongly affect the phase stability, minor- and trace-element partitioning, and transport properties of ''water''-bearing minerals in the Earth's mantle.