Journal of Petrology Advance Access originally published online on March 4, 2008
Journal of Petrology 2008 49(4):841-855; doi:10.1093/petrology/egn007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Seismic Properties of Anita Bay Dunite: an Exploratory Study of the Influence of Water
1Research School of Earth Sciences, Australian National University, Canberra, Act 0200, Australia
2Institute for Study of the Earth's Interior, Okayama University, Misasa, Tottori, 682-0193, Japan
RECEIVED DECEMBER 8, 2006; ACCEPTED FEBRUARY 11, 2008
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONGEOPHYSICAL IMPLICATIONSSUMMARYREFERENCES |
|---|
As a pilot study of the role of water in the attenuation of seismic waves in the Earth's upper mantle, we have performed a series of seismic-frequency torsional forced-oscillation experiments on a natural (Anita Bay) dunite containing accessory hydrous phases, at high temperatures to 1300°C and confining pressure (Pc) of 200 MPa, within a gas-medium high-pressure apparatus. Both oven-dried and pre-fired specimens wrapped in Ni–Fe foil within the (poorly) vented assembly were recovered essentially dry after 50–100 h of annealing at 1300°C followed by slow staged cooling. The results for those specimens indicate broadly similar absorption-band viscoelastic behaviour, but with systematic differences in the frequency dependence of strain-energy dissipation Q–1, attributed to differences in the small volume fraction of silicate melt and its spatial distribution. In contrast, it has been demonstrated that a new assembly involving a welded Pt capsule retains aqueous fluid during prolonged exposure to high temperatures—allowing the first high-temperature torsional forced-oscillation measurements under high aqueous fluid pore pressure Pf. At temperatures >1000°C, a marked reduction in shear modulus, without concomitant increase in Q–1, is attributed to the widespread wetting of grain boundaries resulting from grain-scale hydrofracturing and the maintenance of conditions of low differential pressure Pd = Pc – Pf . Staged cooling from 1000°C is accompanied by decreasing Pf and progressive restoration of significantly positive differential pressure resulting in a microstructural regime in which the fluid on grain boundaries is increasingly restricted to arrays of pores. The more pronounced viscoelastic behaviour observed within this regime for the Pt-encapsulated specimen compared with the essentially dry specimens may reflect both water-enhanced solid-state relaxation and the direct influence of the fluid phase. The scenario of overpressurized fluids and hydrofracturing in the Pt-encapsulated dunite specimen may have some relevance to the high Q–1 and low-velocity zones observed in subduction-zone environments. The outcomes of this exploratory study indicate that the presence of water can have a significant effect on the seismic wave attenuation in the upper mantle and provide the foundation for more detailed studies on the role of water.
KEY WORDS: seismic wave attenuation; water; dunite; hydrous mineral; shear modulus; viscoelasticity; olivine; grain-scale hydrofracturing
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONGEOPHYSICAL IMPLICATIONSSUMMARYREFERENCES |
|---|
Seismological observations have revealed that the propagation of seismic waves in the Earth's mantle is significantly affected by viscoelastic processes, causing attenuation and frequency dependence (dispersion) of wave speeds (e.g. Kanamori & Anderson, 1977
; Anderson & Hart, 1978). It is widely accepted that such viscoelastic processes are associated with the motion of crystal defects such as grain boundaries and dislocations in rocks at high temperatures and/or the stress-induced redistribution of fluids including melt (e.g. Nowick & Berry, 1972; Minster & Anderson, 1981; Karato & Spetzler, 1990; Jackson, 2007). However, further understanding of these processes has been limited by lack of direct measurements of viscoelasticity for mantle materials such as olivine at high temperature and seismic frequency.
In recent years the application of seismic-frequency torsional forced-oscillation methods to fine-grained polycrystalline olivine has become feasible and revealed that elastic behaviour breaks down at sufficiently high temperatures, even in melt-free material, resulting in stronger temperature sensitivity of wave velocity and associated attenuation (Tan et al., 1997; Gribb & Cooper, 1998; Jackson et al., 2002). These viscoelastic effects vary systematically with grain size and have been attributed to elastically and/or diffusionally accommodated grain-boundary sliding (Cooper, 2002; Jackson et al., 2002; Faul et al., 2004).
A pronounced effect of water on viscoelastic processes has been suggested (Karato & Jung, 1998; Karato, 2003) based on the analogy with its effect on creep, in which a trace amount of water significantly weakens olivine-rich rocks (Carter & Avé Lallemant, 1970; Chopra & Paterson, 1984; Karato et al., 1986; Mei & Kohlstedt, 2000). Water weakening of olivine aggregates in creep is currently interpreted in terms of increased concentrations of point defects, such as Si and/or Mg/Fe vacancies, resulting in enhanced rates of ionic diffusion and dislocation climb (e.g. Mei & Kohlstedt, 2000; Kohlstedt, 2006; Walker et al., 2007). However, the influence of water on the micro-strain viscoelasticity has not yet been investigated experimentally.
This exploratory study serves to test two hypotheses: (1) that water can be retained in the specimen at high temperatures (up to 1150°C) during the long duration (50–100 h) forced-oscillation experiments; (2) that the presence of water has an effect on the seismic wave attenuation in mantle materials. Here we report the first seismic-frequency measurements of the influence of water on seismic wave dispersion and attenuation in polycrystalline olivine. For these exploratory experiments we chose a naturally occurring material (see Jackson et al., 1992)—in this case a relatively fine-grained, hydrated but unserpentinitized, dominantly olivine rock from Anita Bay (New Zealand), previously used in the studies that first demonstrated the effect of water on creep behavior of olivine (Chopra & Paterson, 1981, 1984). We measured both pre-fired and initially untreated specimens (either wrapped in Ni–Fe foil or Pt-encapsulated) to isolate the effect of water.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONGEOPHYSICAL IMPLICATIONSSUMMARYREFERENCES |
|---|
Specimen preparation
Starting material
The natural dunite specimen from Anita Bay, New Zealand is a 30 cm diameter boulder that consists primarily of olivine (>90 vol. % of composition Fo93, Chopra & Paterson, 1981
), orthopyroxene and chromite but also contains minor amounts of hydrous phases, in the form of chlorite, tremolite and thin sheets of talc. Cylindrical samples cored from the same boulder were used in the studies by Chopra & Paterson (1981, 1984) and Faul & Jackson (2007). It is compositionally banded with a strong mineral lineation but foliation is weak and no crystallographic preferred orientation is indicated, consistent with previous studies of Anita Bay material (Turner, 1942; Wood, 1972). A trimodal grain-size distribution is evident with an overall average grain size of 100 µm. The fine-grained fraction (20–30% of thin-section area) has a grain size around 20 µm and consists mainly of lineation-parallel domains of orthopyroxene mixed with olivine. The fine-grained material is crack free and has straight grain boundaries. The medium-grained fraction (50–60%) consists almost exclusively of 
100 µm sized olivine grains with no undulose extinction and minor intragranular fracturing. Olivine–olivine grain boundaries in these regions are distinctly serrated (Fig. l). Large porphyroblasts of olivine and orthopyroxene (up to 8 mm in size and elongated parallel to the lineation) form 10–20% of the sample. They display undulose extinction and at least two generations of fractures.
|
Cylindrical specimens were cored parallel to the lineation and precision ground to 11·5 mm in diameter and 30 mm in length. All four samples were cored in close proximity to each other. The density of a representative oven-dried specimen was determined by the Archimedean method involving immersion in ethanol to be 3·2986(2) x 103 kg/m3. This is 99·7% of the theoretical density of olivine (Fo93) (Chopra & Paterson, 1984
), in accordance with a dunite composition.
Heat treatment prior to mechanical testing
In preparation for mechanical testing the rock specimens were either simply oven-dried (at 110°C; runs 1066 and 1093) or fired at 1200°C for 20 h (runs 1055 and 1086) in a controlled oxygen fugacity furnace using a gas mixture comprising equal partial pressures of CO and CO2, yielding log10fO2(Pa) –6 (Muan & Osborn, 1965). Based on the weight loss on ignition, the initial volatile content of the as-received rock was estimated to be 0·20 wt %, which is within the range of previous estimates (0·18 ± 0·02 wt %) (Chopra & Paterson, 1981).
Infrared spectroscopy for [H2O] determination
The water content [H2O] was determined by unpolarized IR spectroscopy on doubly polished disks nominally of 0·5 mm thickness cut from the cylindrical dunite specimens (Tan et al., 2001). For determination of the average concentration of water in the bulk specimens, an aperture of at least 4 mm diameter was employed, whereas for spot analyses, for example within olivine porphyroblasts, an aperture of 70 µm diameter was selected. The concentration of molecular water, contained mainly in fluid inclusions, and responsible for the broad absorption band at wavenumber 
of 3000–3700 cm–1 (Fig. 2), was estimated from the absorbance measured at 3400 cm–1 with an absorption coefficient of 8·1 cm–1 (mol/l)–1 (Rossman, 1988).
|
Hydroxyl chemically bound within the olivine structure is responsible for sharp IR absorption peaks within the wavenumber ranges 3280–3380 and 3450–3510 cm–1, superimposed upon the broad-band absorption attributed to fluid inclusions (Fig. 2d). To facilitate comparison with the water concentrations reported by Zhao et al. (2004
) for similar conditions, their procedure was adopted for the estimation of [H2O], notwithstanding emerging indications that it may systematically overestimate water concentrations estimated from unpolarized spectra (Kovács et al., 2008; Sambridge et al., 2008). First, we calculated the integrated absorbance associated with such peaks normalized to 1 cm thickness and appropriately averaged over similar spectra for multiple randomly oriented crystals (Fig. 2d). The molar absorption coefficient I() of Paterson (1982) with an ‘orientation factor’ of 0·5 was used to calculate the concentration of water, which was then multiplied by 3·5 following the calibration of Bell et al. (2003). The spectrum of the as-received specimen can be resolved into two components, a sharp peak at 3690 cm–1, reflecting the presence of hydrous phases, and a broader band around 3200–3600 cm–1. The sharp peak was eliminated and the broad-band absorption greatly diminished by firing under controlled atmosphere at 1200°C (Fig. 2a), consistent with breakdown of the hydrous phases and expulsion of most of the molecular water (Table 1).
|
Mechanical testing
The experimental conditions and specimen characteristics are presented in Table 1. The pre-fired specimens were wrapped in Ni70Fe30 foil and the oven-dried specimens were either wrapped in Ni70Fe30 foil or sealed within a welded Pt capsule for mechanical testing within an outer mild-steel jacket with torsional forced-oscillation methods (Jackson & Paterson, 1993
; Jackson, 2000). At the outset of each experiment, the assembly was first annealed (for 45–100 h) at the highest temperature [1300°C for pre-fired and oven-dried Ni-Fe wrapped specimens and 1150°C for Pt-sealed specimen] and 200 MPa confining pressure. During this interval, forced-oscillation tests were repeatedly conducted to monitor time dependence in the mechanical behaviour associated with any microstructural evolution. After stabilization of the mechanical behaviour, mechanical data were collected during slow staged-cooling (1–2°C/min) from the annealing temperature to room temperature. The IR spectroscopy for the recovered oven-dried Ni–Fe wrapped specimen (1066) indicated that it retained no more water than the pre-fired specimen (1055; Table 1)—in contrast to previous experiments of shorter duration with similar assemblies (Chopra & Paterson, 1981). The alternative of a welded stainless steel capsule lined internally with sleeves of talc and Ni used in compressive tests by Mei & Kohlstedt (2000) was rejected as impractical for quantitative torsional oscillation tests because of the difficulty of correcting the results for the unknown and potentially time-dependent complex modulus of the dehydrating talc sleeve. Instead, we conducted further measurements with the specimen enclosed within a welded Pt capsule of 280 µm cylindrical wall thickness and 100 µm thick ends, to achieve H2O-saturated conditions.
The data obtained from the forced-oscillation tests were processed to obtain the shear modulus G and strain-energy dissipation Q–1 vs period To for the interval 1–1000 s (Jackson et al., 2004). In a significant refinement of established procedures, the calibration factors for the displacement transducers were corrected for drift during the long intervals (typically 5–9 h) of pre-programmed data acquisition. The torsional compliance of the specimen assembly was compared with that for a foil-free reference assembly containing an alumina control specimen. Combined G and Q–1 data for relatively high temperatures (800–1300°C) for each specimen were parameterized in an internally consistent manner with a phenomenological model based on a Burgers-type creep function (Faul & Jackson, 2005; Jackson, 2005) briefly described in Table 2 (first footnote). Owing to the low strain amplitudes (10–5) of torsional oscillation, the microstructures remained unchanged by mechanical testing.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONGEOPHYSICAL IMPLICATIONSSUMMARYREFERENCES |
|---|
Overview of shear modulus and dissipation data
The forced-oscillation data (both shear modulus G and dissipation Q–1 at high temperatures, T>1000°C for pre-fired and oven-dried, T>800°C for Pt-encapsulated specimen) show strong viscoelastic behaviour and are generally well fitted by the Burgers model as illustrated in Figs 3–6
. The parameters of the Burgers model for each of the four specimens are listed in Table 2. Monotonic variation of Q–1 with oscillation period for each specimen is observed, which is similar to that obtained in previous studies for ‘melt-free’ olivine aggregates (e.g. Jackson et al., 2002), in spite of the presence in each specimen of a small silicate melt fraction, and in the Pt-encapsulated specimen, of 2 vol. % of an aqueous fluid phase (see Microstructures section below). At lower temperatures, the specimen behaves nearly elastically, with relatively low levels of dissipation (< 0·01) and no measurable dispersion. The usual mildly monotonic increase of shear modulus with decreasing temperature is disrupted at low temperatures (< 600°C, not plotted) by thermal micro-cracking attributed to the relatively large grain-size and multi-phase character of the dunite material under conditions of relatively low confining pressure (200 MPa).
|
|
|
|
Although the Pt-encapsulated specimen was mechanically tested at a substantially lower maximum temperature (1150°C) than for the other specimens (1300°C), the magnitudes of the dissipation Q–1 for these different conditions are broadly comparable, indicating that the presence of H2O has considerable influence on the mechanical behavior.
Comparison of G and Q–1 data for the Ni–Fe-wrapped oven-dried (1066) and pre-fired (1055 and 1086) specimens
As shown in Fig. 3, the pre-fired (1055) and oven-dried (1066) specimens display broadly similar levels of dissipation, but with a dependence on oscillation period To that is markedly stronger for the pre-fired specimen (1055) (
logQ–1/logTo 
= 0·43 compared with = 0·27 for the oven-dried specimen, Table 2). The systematically lower values of Q–1 at short periods for specimen 1055 are associated with a narrower spread of the values of the modulus at short period, which is reflected in a value (2·7 x 10–4 K–1) of 
lnJU/T, the temperature sensitivity of the unrelaxed compliance (Table 2, first footnote), that is not much greater than the anharmonic value for Fo90 olivine (2·1 x 10–4 K–1, Anderson & Isaak, 1995). This indicates that significant dissipation does not extend to relaxation times much shorter than 1 s even for temperatures as high as 1300°C. An experiment on another pre-fired specimen (run 1086) of Anita Bay dunite (Fig. 4), confirms the strong frequency dependence of Q–1 for the pre-fired material but with consistently somewhat lower Q–1 and significantly higher G for 1086 presumably reflecting the heterogeneity in the Anita Bay dunite boulder from which the specimens were cored.
Comparison of G and Q–1 data for the oven-dried specimens 1066 (Ni–Fe-wrapped) and 1093 (Pt-encapsulated)
For the Pt-encapsulated specimen (1093), a satisfactory fit to the Burgers model required exclusion of data pertaining to the highest temperatures and longest periods (e.g. 214–1000 s at 1075°C, 22–1000 s at 1120°C and 1–1000 s at 1150°C; Fig. 5). The excluded Q–1 data are in fact broadly compatible with the fit to the lower temperature data but this is not valid for G. Thus, there is a significant reduction in modulus at relatively long periods and temperatures >1000°C without any associated increase in Q–1. The possibility that the modulus deficit reflects the high pore pressure of H2O will be addressed in the following section.
Q–1 is consistently higher for the Pt-sealed specimen 1093 than for the Ni–Fe-wrapped specimen 1066 by 0·5 log unit, but the frequency dependences are broadly consistent (with 0·26; Fig. 6). The offset is essentially independent of temperature. The shear modulus is substantially lower for 1093 than for 1066 in the regime where the extended Burgers fit applies. The systematically lower modulus and greater dissipation for the former specimen suggest that the presence of water might play an important role in the mechanical behaviour.
Characteristics and microstructures of recovered specimens
Microstructures
Representative microstructures for three types of recovered specimen are shown in Figs 1, 7 and 8. The melt fractions for the oven-dried and pre-fired specimens (Table 1) were estimated via light microscopy and scanning electron microscopy (SEM). Recovered specimens re-equilibrated and quenched from 1300°C at 300 MPa, expected to preserve the maximum melt fraction relevant to experimental conditions, show generally very small melt fractions averaging <0·1%. Melt is distributed heterogeneously and is concentrated in areas where amphibole or chlorite was present or in adjacent olivine-rich regions of 100 µm grain size. Those hydrous phases have decomposed into olivine + orthopyroxene + melt ± clinopyroxene ± spinel. Melt fractions of up to 10 vol. % occur locally in those areas. Large gradients in melt fraction, particularly in pre-fired specimens, mean that melt fraction can drop to zero within as little as 50 µm distance from those former hydrous phases. On the sample scale, large areas with no melt present exist. In different parts of each sample, different melting paths are taken, in effective isolation, for different hydrous impurities. Consequently, the composition of glass formed by melt solidification is highly variable: for example, compositions in the pre-fired specimen (1055) show ranges of 53–58 wt % for SiO2, 14–21 wt % Al2O3 and are very potassic, with 3 wt % K2O (Table 3). In the specimens that have been stage-cooled for attenuation measurements, the melts formed at high temperatures would have had the opportunity to partially crystallize.
|
|
|
For the oven-dried Ni–Fe-wrapped specimen (1066) subject to progressive dehydration during the course of the mechanical testing, the initial melting presumably occurred in the presence of an interconnected aqueous pore fluid with Pf
Pc (see Discussion) explaining the more uniform glass composition (especially SiO2, Al2O3, CaO, Na2O and K2O; Table 3), than for 1055. The much higher melt fraction in the recovered material than for 1055 presumably reflects a higher proportion of the non-olivine phases in that core.
For the specimen (6579) quenched from 1150°C under Pt-encapsulated conditions, the melt retained (0·1 vol. %) is relatively homogeneously distributed and also reasonably uniform in chemical composition except for MgO and FeO, but is distinctively higher in SiO2 and lower in Na2O than the other melts. The melt pockets contain euhedral–subhedral orthopyroxene and subordinate clinopyroxene along with residual silicate melt. The melt composition (Table 3) is lower in Na2O but otherwise closely comparable with the nano-crystalline plagioclase previously reported in melt-bearing olivines recovered from staged cooling (Faul et al., 2004).
In contrast, the Pt-encapsulated specimen 1093 recovered after staged cooling contains no detectable melt even at sites where the microstructure and mineral assemblage point to the prior existence of hydrous-phase grains. However, the average 2 vol. % porosity is distributed heterogeneously throughout the specimen with fractions locally up to 30% over regions of 300 µm2 at prior hydrous-grain sites. Because 6579, quenched from 1150°C, showed a small amount of melt (0·1 vol. %), it seems reasonable to assume that 1093 similarly contained a small amount of melt at the highest temperature. Apparently this melt had almost completely crystallized during staged cooling in the presence of 2 vol. % of an aqueous phase. Many olivine–olivine grain boundaries contain micropore arrays, as well as larger lens-shaped pores with apparent dihedral angles of 30–50°. Olivine and orthopyroxene grains bordering the largest pore structures at grain junctions are conspicuously faceted. The pore volume has been partially redistributed into the olivine-dominant 100 µm grain-size regions, where the majority of triple junctions are occupied (pore volumes of 0·5–1·5%) indicating considerable connectivity of the pore fluids (Fig. 8). Original olivine grains have undergone grain growth in parts of these regions, and grain boundaries have become curved to straight (Fig. 1). The multitude of pore arrays and lenses, and the presence of many open grain boundaries are consistent with high and distributed fluid pressure and pervasive hydrofracturing.
Grain shape change and grain growth in the olivine-dominant regions were also clear in the oven-dried and Ni–Fe-wrapped specimen 1066 (Fig. 1), whereas the grain-boundary shapes in pre-fired specimen 1055 are indistinguishable from those of the starting material. It thus seems reasonable to speculate that grain-boundary mobility is positively correlated with the bulk OH content.
Water in the recovered specimens
The bulk molecular H2O contents for the recovered specimens inferred from the IR absorption, as described above, are 34–38 wt ppm for pre-fired (1055 and 1086) and oven-dried Ni–Fe-wrapped specimens (1066) and 156–187 wt ppm for Pt-encapsulated specimens 6579 and 1093 (Table 1). The spot analyses of 1093 with 70 µm aperture reveal significant variability in [H2O] within and amongst the (randomly oriented) olivine porphyroblasts (Fig. 2d). [H2O] is systematically somewhat higher in matrix regions, implying that H2O is preferentially accommodated in grain boundaries and in any small amount of glass (Fig. 2c).
Within the olivine porphyroblasts of 1093, broad-band absorption diagnostic of molecular water mainly in fluid inclusions is similarly dominant, at a level of 20–80% of the absorbance for the bulk specimen (Fig. 2c). The superimposed sharp absorption peaks at 3353 and 3325 cm–1 are attributed to hydroxyl associated with the trivalent cations Fe3+ and Cr3+, respectively, whereas the other prominent peak (at 3478 cm–1) may reflect hydroxyl related to Si vacancies and/or Ti4+-related defects (Berry et al., 2005, 2007; Grant et al., 2007; Walker et al., 2007). Such sharp peaks (Fig. 2d) provide evidence of structurally bound hydroxyl at concentrations of 2–3 wt ppm H2O. Evidence of newly crystallized hydrous phases, in the form of a sharp absorption peak at 3690 cm–1, was also found in both bulk and matrix analyses (Fig. 2c). The weight loss on ignition of the recovered specimen (1093) was 0·28 wt % (Table 1), indicating that most of the volatiles initially contained in the specimen were retained in the Pt-capsule following sustained high-T mechanical testing.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONGEOPHYSICAL IMPLICATIONSSUMMARYREFERENCES |
|---|
Solid-state viscoelastic relaxation in dry dunite
The results of the present study that are most closely comparable with previous seismic-frequency data are for the oven-dried Ni–Fe-wrapped specimen 1066, recovered essentially dry. A useful comparison can be made with the Andrade–Gaussian pseudo-period model for Q–1 for a suite of synthetic melt-bearing olivine polycrystals (Jackson et al., 2004
, table 4). For comparison with the data from the present study for specimen 1066, this model has been evaluated at 100 µm grain size and at the representative melt fractions of 0·01 and 0·1 vol. %. A good match to the data for 1066 is obtained for 
= 0·01% (Fig. 9a). However, for = 0·1%, the estimated melt fraction for specimen 1066 (Table 1), the model deviates systematically from the data at short periods where a subtle Q–1 plateau at temperatures of 1150–1250°C predicted by the model is not strongly expressed in the data (Fig. 9b). Thus there is no compelling evidence for a melt-related dissipation peak despite the fact that the olivine grain edges are perhaps sufficiently commonly rounded at grain-edge triple junctions (Fig. 7c) to facilitate elastically accommodated grain-boundary sliding as envisaged by Faul et al. (2004). However, the milder dependence of Q–1 upon oscillation period for the oven-dried Ni–Fe-wrapped specimen 1066 might in fact mask the presence of a dissipation peak of low amplitude associated with a silicate melt fraction somewhat greater and more uniformly distributed than for the pre-fired specimens.
|
Despite the microstructural complexities of Anita Bay dunite, the Andrade–Gaussian pseudo-period model of simple synthetic melt-bearing olivine (Jackson et al., 2004
) provides a reasonable approximation to the data for the natural dunite specimen 1066. This demonstrates that Anita Bay dunite can be usefully compared with simpler polycrystalline mantle rocks and that Anita Bay dunite can be used as a first-order proxy for mantle material in examination of the influence of water on seismic wave attenuation.
H2O in the Pt-encapsulated specimen 1093
The results obtained for the Pt-encapsulated specimen (1093) display consistently lower shear modulus and higher dissipation than the foil-wrapped specimen 1066. To interpret this contrast in seismic properties, we first estimate the pore fluid pressure using the equation of state for pure H2O given by Holland & Powell (1991, Fig. 2a). The volume and mass mf fractions of the H2O pore fluid are related by
|
|
r and f are the densities of the rock matrix and fluid, respectively.
For mf = 0·0028 (for 1093, Table 1) and r = 3·3 g/cm3 and pore fluid pressure Pf of 200 MPa (equal to confining pressure Pc) is predicted to decrease linearly with temperature from 3·1% at 1150°C to 2·0% at 750°C—compared with our estimate of 2% porosity in the recovered specimen. At the highest temperatures of runs 1093 and 6579, the condition Pf = Pc is thus expected to be realized. Any tendency for Pf to exceed Pc will be countered by swelling of the Pt capsule and enclosing steel jacket to increase the volume available for the pore fluid. On cooling, restoration of a significant differential pressure Pd = Pc – Pf (sometimes called effective pressure) is expected for temperatures <750°C.
Considering the low initial porosity in the as-received dunite, the 2 vol. % porosity must have been created by dehydration accompanied by the development of a small silicate melt fraction, leading to high pore-fluid pressure and pervasive grain-boundary hydrofracturing. This microstructural evidence, along with the IR evidence of molecular water in fluid inclusions within the olivine porphyroblasts and the substantial weight loss of 0·28% on oven drying of the recovered specimen (Table 1), suggests that the specimen remained H2O-saturated throughout the experiment (see Grant et al., 2007). In these circumstances, the available H2O would have been partitioned among a substantial volume fraction (2%) of an aqueous fluid phase, a subordinate amount (0·1% at the maximum temperature) of an immiscible silicate melt phase (e.g. Hirschmann et al., 2005) subject to essentially complete progressive crystallization during staged cooling, and hydroxyl chemically bound in the crystal structure of the nominally anhydrous minerals.
Using the recent results of Zhao et al. (2004), we estimate the equilibrium solubility in the olivine of specimen 1093 to be 26 wt ppm H2O at Pc = Pf = 200 MPa and 1150°C, for the relatively oxidizing conditions expected to prevail within the Pt capsule (Grant et al., 2007). However, Mackwell & Kohlstedt (1990) reported re-equilibration of dissolved water on timescales of hours in millimetre-sized crystals during cooling to 1000°C, suggesting that equilibrium in this study would have been maintained to lower temperatures of perhaps 800°C during staged cooling involving 25°C steps and 6–10 h at each stage. This scenario is consistent with more recent data concerning H diffusivity reviewed by Ingrin & Blanchard (2006). At 800°C, a substantially lower solubility of about 8 wt ppm H2O is expected (Zhao et al., 2004). This estimate is to be compared with the concentrations of 2–3 wt ppm H2O for the olivine porphyroblasts in the Pt-encapsulated specimen from the sharp IR absorption peaks. Significant variability amongst the various porphyroblasts (Fig. 2d) probably reflects a combination of two factors: variation of absorbance with lattice orientation, and diffusion profiles associated with incomplete re-equilibration to lower [H2O] during staged cooling (see Peslier & Luhr, 2006). The observed concentration of structurally bound hydroxyl in the olivine porphyroblasts of the Pt-encapsulated specimen, along with the presence of fluid inclusions, is therefore broadly consistent with expectations based on equilibrium partitioning with the pore fluid during staged cooling.
Water-enhanced solid-state viscoelastic relaxation
The Pt-encapsulated specimen 1093 has a similarly low value of [0·25(1)] to 1066 (recovered essentially dry following in situ dehydration) but generally much higher Q–1 and lower G (Fig. 6). A modulus deficit, essentially independent of oscillation period, is attributed to the direct influence of the fluid phase as discussed in the next section. However, the water-saturated conditions maintained in specimen 1093 are also expected to promote solid-state viscoelastic relaxation—through a combination of grain-boundary and intragranular mechanisms. Increased grain-boundary mobility has already been inferred microstructurally. The effective grain-boundary viscosity 
should be markedly reduced by the presence of the aqueous fluid phase, so that the characteristic timescale (is proportional to ) for relaxation associated with grain-boundary sliding might be significantly less than for essentially anhydrous conditions (e.g. for specimen 1066). The mild period dependence of Q–1 would then be explained as for 1066 by the superposition upon the dissipation background of a broad peak, having its impact at lower temperatures than for 1066 because of the influence of water in reducing the grain-boundary viscosity. The elevated water fugacity may also enhance intragranular viscoelastic relaxation by increasing the mobility of dislocations as suggested by Karato (2003).
The direct influence of the fluid phase
Both wetted grain boundaries and their tendency to evolve into arrays of isolated grain-boundary pores are evident in the Pt-encapsulated (1093) specimen (Fig. 8). However, the presence of pores at most olivine triple junctions indicates the preservation of high fluid connectivity. Presumably the hydrofracturing and high Pf created and maintained a high degree of boundary wetting at the highest temperatures. During subsequent staged cooling, Pf eventually became significantly less than Pc and the arrays of isolated grain-boundary pores began to form. In specimen 1093, the small silicate melt fraction evidently crystallized very effectively, and presumably early, during the staged cooling from 1150°C. In this scenario, it is the 2 vol. % aqueous fluid phase, present throughout staged cooling but with progressively lower Pf, that is relevant to the observed viscoelastic relaxation.
It is widely recognized that elastic moduli and wave speeds will be influenced by the presence of a bulk pore fluid to a degree determined mainly by the properties of the porous or cracked rock matrix and the saturating fluid, notably the volume fraction of pore fluid, the aspect ratio of the fluid-filled ‘inclusions’ (defined as minimum/maximum dimension), and the bulk modulus of the fluid. The effective elastic properties are also predicted to vary with the timescale or frequency of the measurement. For the saturated isolated regime, probed at the MHz frequencies of ultrasonic techniques, experimentally determined elastic properties can be related to those of the dense matrix via the ‘soft-fluid’ saturation theory of O’Connell & Budiansky (1974). Ultrasonic methods have been used by Kono et al. (2007) to measure the reduction of compressional wave speed (VP) resulting from the in situ dehydration of hydrous minerals in altered ultramafic rocks of low initial porosity encapsulated, like those of the present study, within welded metal containers. Observed reductions in VP , reaching only 15% for = 0·12 at 1 GPa and 1000°C and interpreted as representative of the saturated isolated regime, suggest an average aspect ratio of 0·15 indicative of sub-equant fluid inclusions rather than extensively wetted grain boundaries with aspect ratios <0·01.
For conditions representative of the present study of specimen 1093 ( = 0·02 and water saturation with Pc = Pf = 200 MPa at 1150°C), and the average aspect ratio of 0·15 inferred from the preceding analysis, a shear modulus reduction of 5% would be predicted within the saturated isolated regime. A further 2% shear modulus relaxation would be expected under these conditions from local fluid flow [O’Connell & Budiansky, 1977, equation (25)] yielding an essentially frequency-independent modulus deficit of perhaps 7% for the saturated isobaric regime presumably probed by the present low-frequency forced-oscillation measurements.
At the highest temperatures of the present study, the pressurized fluid phase may exert additional influence by compromising the mechanical coupling between torsion rods and specimen that depends upon maintenance of significant differential pressure Pd (Jackson et al., 1984). Low Pd provides a plausible explanation for the anomalous reduction of modulus without associated dissipation evident at progressively longer periods for temperatures above 1000°C.
The direct influence of the fluid phase on the measured modulus is expected to diminish during staged cooling on account of the previously mentioned microstructural changes. As grain boundaries that were at least transiently wetted at Pf Pc evolve to form the arrays of isolated grain-boundary pores and the connected grain-edge tubules at lower T and Pf, the average aspect ratio of the fluid inclusions will increase substantially, resulting in a progressively smaller modulus deficit.
In summary, these first forced-oscillation data provide compelling evidence of markedly lower shear moduli and higher attenuation under water-saturated conditions. These effects are attributable in part to the presence of a bulk pore fluid and partly to the enhancement of solid-state relaxation processes by the high water fugacity. However, the heterogeneity and microstructural complexity of the Anita Bay dunite tested in this exploratory study preclude a definitive quantitative interpretation. It is intended that a further study on the role of water on seismic wave attenuation will use simpler synthetic polycrystalline olivine specimens. Such a study will provide more conclusive answers concerning the effect of water on seismic wave attenuation in the upper mantle.
|
GEOPHYSICAL IMPLICATIONS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONGEOPHYSICAL IMPLICATIONSSUMMARYREFERENCES |
|---|
As discussed in the previous section, the homogeneous distribution of fluids along grain boundaries, presumably resulting from hydrofracturing, is consistent with our semi-quantitative estimate of Pf
Pc at high temperature. Broadly analogous conditions may be realized, at least transiently, during dehydration of subducting oceanic lithosphere, where pore pressures comparable with the prevailing lithostatic pressure might induce hydrofracturing, thereby facilitating fluid transport into the overlying mantle wedge triggering partial melting (Davies, 1999). The development and persistence of such conditions of high pore-fluid pressure will depend strongly on the grain-scale permeability, which is a function of the instantaneous wetting behaviour and the temporal evolution of the fluid–solid microstructure (e.g. Faul, 1997). Increasing pressure with depth in the upper mantle increases the silica content of the aqueous fluid resulting in lower dihedral angles, connectivity of the fluid phase distributed in grain-edge tubules, and hence higher permeability (e.g. Mibe et al., 1998). The presence of a few per cent of a fluid phase thus accommodated mainly in grain-edge tubules would have limited impact upon the seismic wave speeds. However, the complete wetting of grain boundaries, inferred for the highest temperatures of the present study and recently reported by Yoshino et al. (2007) for pressures corresponding to the deepest part of the upper mantle, results in much larger reductions in wave speeds that are likely to be seismologically observable. Thus the reduced wave speeds and pronounced dissipation observed in the fluid-pressurized specimens of the present study and that of Kono et al. (2007) may be relevant particularly to the deepest zones of low velocity and high attenuation in mantle-wedge environments (e.g. Zhao et al., 1992; Takanami et al., 2000; Nakajima & Hasegawa, 2003). However, the very different timescales for grain-scale hydrofracturing, textural re-equilibration and fluid migration in the laboratory and subduction-zone settings, along with the markedly different fluid compositions, preclude any immediate quantitative seismological application of these experimental data. |
SUMMARY |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONGEOPHYSICAL IMPLICATIONSSUMMARYREFERENCES |
|---|
An exploratory study of the role of water in upper-mantle seismic-wave attenuation and dispersion has been undertaken on a natural dunite, of 100 µm average olivine grain size, containing
0·3 wt % H2O mainly in accessory hydrous silicate minerals. Cylindrical samples, either oven-dried or thoroughly dehydrated by pre-firing at 1200°C under controlled atmosphere, and either wrapped in Ni–Fe foil or sealed within a welded Pt capsule, have been tested in torsional forced oscillation at seismic periods of 1–1000 s and low strain amplitudes, under conditions of simultaneous high pressure (200 MPa) and temperature (to 1300°C).
Both oven-dried and pre-fired specimens, wrapped in Ni–Fe foil in the standard vented experimental assembly, were recovered after staged cooling following annealing for tens of hours at 1300°C and 200 MPa, containing no more than 40 wt ppm bulk molecular H2O. On the other hand, Pt encapsulation of dunite allows retention of essentially the entire inventory of water, including 150 wt ppm bulk molecular water trapped as fluid inclusions in the recovered specimens, for the duration of similar experiments at peak temperatures of 1150°C. In situ dehydration of the hydrous minerals liberates water at a pore pressure Pf locally and transiently greater than the confining pressure, resulting in pervasive hydrofracturing and redistribution of the dominant aqueous fluid and coexisting minor silicate melt phases. During staged cooling from 1150°C the 2 vol. % porosity thus created is occupied mainly by the aqueous pore fluid. The decrease of Pf with decreasing temperature leads eventually to the restoration of a significantly positive differential pressure Pd = Pc – Pf and a consequent transition from grain-boundary wetting for Pd 0 towards arrays of isolated grain-boundary pores for Pd > 0.
Each specimen displays intense viscoelastic relaxation at sufficiently high temperatures; both pronounced dissipation Q–1 and associated inverse dependence of the shear modulus upon oscillation period are measured. The behaviour is of the ‘high-temperature background’ type, whereby Q–1 varies monotonically with period and temperature without any well-resolved dissipation peak. For each of the pre-fired specimens, the absence of a melt-related dissipation peak is attributed to the highly localized distribution of a small melt fraction (
0·05 vol. %); the bulk of the specimen is simply melt-free. For the Ni–Fe-wrapped oven-dried specimen (1066), the somewhat more abundant melt (0·1 vol. %) is more uniformly distributed on account of the hydrofracturing associated with transiently high Pf early in the experiment. A poorly resolved broad melt-related peak of low amplitude superimposed on the background dissipation may explain the milder period dependence of Q–1 for this specimen.
Where 2 vol. % of an aqueous fluid phase is retained (in the Pt-encapsulated specimen) the lowest shear moduli and highest levels of attenuation were measured. The mild period dependence of Q–1, like that for 1066, may be attributable to the superposition on the background of a poorly resolved dissipation peak associated with grain-boundary sliding. The higher dissipation and lower modulus for a given temperature (Fig. 6) than for specimen 1066 may reflect both water-enhanced solid-state relaxation (grain-boundary and intragranular mechanisms) and the direct influence of the fluid phase. The anomalously low moduli measured at the highest temperatures are attributed to conditions of low differential pressure resulting in extensive wetting of grain boundaries by aqueous fluid and compromised mechanical coupling between the specimen and torsion rods. It has been shown in this exploratory study that water may have a significant effect on seismic wave attenuation in mantle material, thereby giving incentive for more detailed studies.
|
ACKNOWLEDGEMENTS |
|---|
The expert assistance of Harri Kokkonen and Craig Saint in the preparation of specimens and experimental assemblies, and in microstructural characterization, is gratefully acknowledged. We thank David Green for his consistent support of our work on seismic wave attenuation in upper-mantle materials and for his valuable guidance on petrological matters. Detailed reviews by S. Karato and anonymous reviewers of earlier versions of the manuscript are greatly appreciated.
|
FOOTNOTES |
|---|
Present address: Japan Manned Space Systems Corporation, MY Bldg., 1-1-26, Kawaguchi, Tsuchiura, Ibaraki 300-0033, Japan. 
Present address: Department of Earth Sciences, Boston University, Boston, MA 02215, USA.
Present address: Faculty of Geosciences, Department of Earth Sciences, Utrecht University, PO Box 80021, 3508 TA Utrecht, The Netherlands.
*Corresponding author. Telephone: +61 2 6125 2498. Fax: +61 2 6125 8253. E-mail: Ian.Jackson{at}anu.edu.au
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONGEOPHYSICAL IMPLICATIONSSUMMARYREFERENCES |
|---|
Anderson DL, Hart RS. Q of the Earth. Journal of Geophysical Research (1978) 83:5869–5882.[CrossRef]
Anderson DL, Isaak DG. Elastic constants of mantle minerals at high temperature. In: Mineral Physics and Crystallography: A Handbook of Physical Constants, AGU Reference Shelf, Vol. 2—Ahrens TJ, ed. (1995) Washington, DC: American Geophysical Union. 64–97.
Bell DR, Rossman GR, Maldener J, Endisch D, Rauch F. Hydroxide in olivine: A quantitative determination of the absolute amount and calibration of the IR spectrum. Journal of Geophysical Research (2003) 108. doi:10.1029/2001JB000679.
Berry AJ, Hermann J, O’Neill H. StC, Foran GJ. Fingerprinting the water site in mantle olivine. Geology (2005) 33:869–872.
Berry AJ, O’Neill H. StC, Hermann J, Scott DR. The infrared signature of water associated with trivalent cations in olivine. Earth and Planetary Science Letters (2007) 261:134–142.[CrossRef][Web of Science]
Carter NL, Avé Lallemant HG. High-temperature flow of dunite and peridotite. Geological Society of America Bulletin (1970) 81:2181–2202.
Chopra PN, Paterson MS. The experimental deformation of dunite. Tectonophysics (1981) 78:453–473.[CrossRef][Web of Science]
Chopra PN, Paterson MS. The role of water in the deformation of dunite. Journal of Geophysical Research (1984) 89:7861–7876.[CrossRef]
Cooper RF. Seismic wave attenuation: Energy dissipation in viscoelastic crystalline solids. In. Plastic Deformation in Minerals and Rocks. Mineralogical Society of America, Reviews of Mineralogy and Geochemistry—Karato S, Wenk H, eds. (2002) 51:253–290.
Davies JH. The role of hydraulic fractures and intermediate depth earthquakes in generating subduction-zone magmatism. Nature (1999) 398:142–145.[CrossRef]
Faul UH. Permeability of partially molten upper mantle rocks from experiments and percolation theory. Journal of Geophysical Research (1997) 102:10299–10311.[CrossRef]
Faul UH, Jackson I. The seismological signature of temperature and grain size variations in the upper mantle. Earth and Planetary Science Letters (2005) 234:119–134.[CrossRef][Web of Science]
Faul UH, Jackson I. Diffusion creep of dry, melt-free olivine. Journal of Geophysical Research (2007) 112. B04204, doi:10.1029/2006JB004586.
Faul UH, Fitz Gerald JD, Jackson I. Shear-wave attenuation and dispersion in melt-bearing olivine polycrystals II. Microstructural interpretation and seismological implications. Journal of Geophysical Research (2004) 109. B06202, doi:10.1029/2003JB002407.
Grant K, Brooker RA, Kohn SC, Wood BJ. The effect of oxygen fugacity on hydroxyl concentrations and speciation in olivine: Implications for water solubility in the upper mantle. Earth and Planetary Science Letters (2007) 261:217–229.[CrossRef][Web of Science]
Gribb TT, Cooper RF. Low-frequency shear attenuation in polycrystalline olivine: grain-boundary diffusion and the physical significance of the Andrade model for viscoelastic rheology. Journal of Geophysical Research (1998) 103:27267–27279.[CrossRef]
Hirschmann MM, Aubaud C, Withers AC. Storage capacity of H2O in nominally anhydrous minerals in the upper mantle. Earth and Planetary Science Letters (2005) 236:167–181.[CrossRef][Web of Science]
Holland TJB, Powell B. A Compensated-Redlich–Kwong (CORK) equation for volumes and fugacities of CO2 and H2O in the range 1 bar to 50 kbar and 100–1600°C. Contributions to Mineralogy and Petrology (1991) 160:170–182.
Ingrin J, Blanchard M. Diffusion of hydrogen in minerals. In. Water in Nominally Anhydrous Minerals. Mineralogical Society of America, Reviews of Mineralogy and Geochemistry—Keppler H, Smyth JR, eds. (2006) 62:291–320.
Jackson I. Laboratory measurement of seismic wave dispersion and attenuation: recent progress. In. Earth's Deep Interior: Mineral Physics and Tomography from the Atomic to the Global Scale. Geophysical Monograph, American Geophysical Union—Karato S, Forte AM, Liebermann RC, et al, eds. (2000) 117:265–289.
Jackson I. Laboratory measurement of seismic wave dispersion and attenuation at high pressure and temperature. In: Advances in High-pressure Technology for Geophysical Applications—Chen J, Wang Y, Duffy TS, et al, eds. (2005) Amsterdam: Elsevier. 95–115.
Jackson I. Properties of rocks and minerals—physical origins of anelasticity and attenuation in rock. In: Treatise on Geophysics, Volume 2—Schubert G, ed. (2007) Oxford: Elsevier. 493–525.
Jackson I, Paterson MS. A high-pressure, high-temperature apparatus for studies of seismic wave dispersion and attenuation. Pure and Applied Geophysics (1993) 141:445–456.[CrossRef][Web of Science]
Jackson I, Paterson MS, Niesler H, Waterford RM. Rock anelasticity measurements at high pressure, low strain amplitude and seismic frequency. Geophysical Research Letters (1984) 11:1235–1238.[CrossRef][Web of Science]
Jackson I, Paterson MS, Fitz Gerald JD. Seismic wave attenuation in Åheim dunite: an experimental study. Geophysical Journal International (1992) 108:517–534.[CrossRef][Web of Science]
Jackson I, Fitz Gerald JD, Faul UH, Tan BH. Grain-size-sensitive seismic wave attenuation in polycrystalline olivine. Journal of Geophysical Research (2002) 107. doi:10.1029/2001JB001225.
Jackson I, Faul UH, Fitz Gerald JD, Tan BH. Shear-wave attenuation and dispersion in melt-bearing olivine polycrystals I. Specimen fabrication and mechanical testing. Journal of Geophysical Research (2004) 109. B06201, doi:10.1029/2003JB002409.
Kanamori H, Anderson DL. Importance of physical dispersion in surface wave and free oscillation problems: Review. Review of Geophysics (1977) 15:105–112.
Karato S. Mapping water content in the upper mantle, In. The Subduction Factory. Geophysical Monograph, American Geophysical Union—Eiler J, ed. (2003) 138:135–152.
Karato S, Jung H. Water, partial melting and the origin of the seismic low-velocity and high-attenuation zone in the upper mantle. Earth and Planetary Science Letters (1998) 153:193–207.
Karato S, Spetzler HA. Defect microdynamics in minerals and solid-state mechanism of seismic wave attenuation and velocity dispersion in the mantle. Reviews of Geophysics (1990) 28:399–421.[CrossRef][Web of Science]
Karato SI, Paterson MS, Fitz Gerald JD. Rheology of synthetic olivine aggregates: Influence of grain size and water. Journal of Geophysical Research (1986) 91:8151–8176.
Kohlstedt DL. The role of water in high-temperature rock deformation. In. Water in Nominally Anhydrous Minerals. Mineralogical Society of America, Reviews of Mineralogy and Geochemistry—Keppler H, Smyth JR, eds. (2006) 62:377–396.
Kono Y, Ishikawa M, Arima M. Effect of H2O released by dehydration of serpentine and chlorite on compressional wave velocities of peridotites at 1 GPa and up to 1000°C. Physics of the Earth and Planetary Interiors (2007) 161:215–223.[CrossRef][Web of Science]
Kovács I, Hermann J, O’Neill H. StC, Fitz Gerald JD, Sambridge M, Horvath G. Quantitative absorbance spectroscopy with unpolarized light, Part II: Experimental evaluation and development of a protocol for quantitative analysis of mineral IR spectra. American Mineralogist (2008) (in press).
Mackwell SJ, Kohlstedt DL. Diffusion of hydrogen in olivine—implications for water in the mantle. Journal of Geophysical Research (1990) 95:5079–5088.
Mei S, Kohlstedt DL. Influence of water on plastic deformation of olivine aggregates 1. Diffusion creep regime. Journal of Geophysical Research (2000) 105:21457–21469.[CrossRef]
Mibe K, Fujii T, Yasuda A. Connectivity of aqueous fluid in the Earth's upper mantle. Geophysical Research Letters (1998) 25:1233–1236.[CrossRef][Web of Science]
Minster JB, Anderson DL. A model of dislocation controlled rheology for the mantle. Philosophical Transactions of the Royal Society of London, Series A (1981) 299:319–356.[CrossRef]
Muan A, Osborn EF. Phase Equilibria among Oxides in Steelmaking (1965) Reading, MA: Addison Wesley.
Nakajima J, Hasegawa A. Estimation of thermal structure in the mantle wedge of northeastern Japan from seismic attenuation data. Geophysical Research Letters (2003) 30. 1760, doi:10.1029/2003GL017185.
Nowick AS, Berry BS. Anelastic Relaxation in Crystalline Solids (1972) New York: Academic Press.
O’Connell RJ, Budiansky B. Seismic velocities in dry and saturated cracked solids. Journal of Geophysical Research (1974) 79:5412–5426.
O’Connell RJ, Budiansky B. Viscoelastic properties of fluid-saturated cracked solids. Journal of Geophysical Research (1977) 82:5719–5735.
Paterson MS. The determination of hydroxyl by infrared absorption in quartz, silicate glasses and similar minerals. Bulletin of Mineralogy (1982) 105:20–29.
Peslier AH, Luhr JF. Hydrogen loss from olivines in mantle xenoliths from Simcoe (USA) and Mexico: Mafic alkalic magma ascent rates and water budget of the sub-continental mantle. Earth and Planetary Science Letters (2006) 242:302–319.[CrossRef][Web of Science]
Rossman GR. Vibrational spectroscopy of hydrous components. In: Spectroscopic Methods in Mineralogy and Geology—Hawthorne FC, ed. (1988) Washington, DC: Mineralogical Society of America. 183–206.
Sambridge M, Fitz Gerald JD, Kovács I, O’Neill H. StC, Hermann J. Quantitative IR spectroscopy with unpolarized light, Part I: Physical and mathematical development. American Mineralogist (2008) (in press).
Takanami T, Sacks IS, Hasegawa A. Attenuation structure beneath the volcanic front in northeastern Japan from broad-band seismograms. Physics of the Earth and Planetary Interiors (2000) 121:339–357.[CrossRef][Web of Science]
Tan BH, Jackson I, Fitz Gerald JD. Shear wave dispersion and attenuation in fine-grained synthetic olivine aggregates: preliminary results. Geophysical Research Letters (1997) 24:1055–1058.[CrossRef][Web of Science]
Tan BH, Jackson I, Fitz Gerald JD. High-temperature viscoelasticity of fine-grained polycrystalline olivine. Physics and Chemistry of Minerals (2001) 28:641–664.[CrossRef][Web of Science]
Turner FJ. Preferred orientation of olivine crystals in peridotites, with special reference to New Zealand examples. Transactions of the Royal Society of New Zealand (1942) 68:570–98.
Walker AM, Hermann J, Berry AJ, O’Neill H. StC. Three water sites in upper-mantle olivine and the role of titanium in the water-weakening mechanism. Journal of Geophysical Research (2007) 112. B05211, doi:10.1029/2006JB004620.
Wood BL. Metamorphosed ultramafites and associated formations near Milford Sound, New Zealand. New Zealand Journal of Geology and Geophysics (1972) 15:88–128.[Web of Science]
Yoshino T, Nishihara Y, Karato S. Complete wetting of olivine grain boundaries by a hydrous melt near the mantle transition zone. Earth and Planetary Science Letters (2007) 256:466–472.[CrossRef][Web of Science]
Zhao D, Hasegawa A, Horiuchi S. Tomographic imaging of P and S wave velocity structure beneath northeastern Japan. Journal of Geophysical Research (1992) 97:19909–19928.[CrossRef]
Zhao YH, Ginsberg SB, Kohlstedt DL. Solubility of hydrogen in olivine: dependence on temperature and iron content. Contributions to Mineralogy and Petrology (2004) 147:155–161.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. J. Angel, J. M. Jackson, H. J. Reichmann, and S. Speziale Elasticity measurements on minerals: a review European Journal of Mineralogy, June 1, 2009; 21(3): 525 - 550. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||