Experimental Constraints on Silicate Perovskite Forming Reactions and Elastic Properties: Geophysical Implications for Chemical Heterogeneity in the Deep Mantle

Experimental Constraints on Silicate Perovskite Forming Reactions and Elastic Properties: Geophysical Implications for Chemical Heterogeneity in the Deep Mantle Dissertation zur Erlangung des Grades eines Doktors der Naturwissenschaften -Dr. rer. nat.- der Fakultät für Biologie, Chemie und Geowissenschaften der Universität Bayreuth vorgelegt von Ashima Saikia 2007 Vollständiger Abdruck der von der Fakultät für Biologie, Chemie und Geowissenschaften der Universität Bayreuth genehmigten Dissertation, zur Erlangung des akademischen Grades Doktors der Naturwissenschaften (Dr. Rer. Nat.). Prüfungssausschuß: Prof. David Rubie, Universität Bayreuth (1. Gutachter) Prof. F. Langenhorst, Universität Jena (2. Gutachter) Prof. Josef Breu, Universität Bayreuth Prof. L. Dubrovinsky, Universität Bayreuth Prof. Klaus Bitzer, Universität Bayreuth Prof. Ludwig Zöller, Universität Bayreuth Prof. S. Peiffer, Universität Bayreuth Datum der Einreichung der Dissertation: 5th October, 2007 Datum des Wissenschaftlichen Kolloquiums 6th February, 2008 2 Acknowledgements I feel immense pleasure in availing this opportunity for expressing my deepest sense of gratitude and regards to my supervisors, Dr. Dan Frost and Prof. Dave Rubie for their dynamic guidance, constructive suggestions and untiring efforts extended during my entire doctoral research work at the Bayerisches Geoinstitut, Universität Bayreuth, Germany. I am acknowledging my obligations to the Elitenetzwerk Bayern, International Graduate School program for funding my research projects and the administration of the Bayerisches Geoinstitut, for providing me necessary facilities to pursue my work. More so I am indebted for substantial help and invaluable suggestions to Dr. Tiziana Boffa Ballaran with whom I carried out my entire diamond anvil cell work, to Prof. Masaki Akaogi, University of Gakushuin, Tokyo (Japan) for allowing me to use his laboratory facility for calorimetric measurements, to Dr. Catherine McCammon for helping me out with the Mössbauer measurements and analysis, to Dr. Nobuyoshi Miyajima for assistances in the TEM studies and to Dr. Florian Heidelbach for the SEM studies and for the German translation of my thesis summary. I offer my special thanks to the entire technical and administrative stuff of the Bayerisches Geoinstitut for all the assistances. Last but not the least I thank all who have directly and indirectly sympathized with me in completing my entire dissertation. I remain Ashima Saikia Bayreuth, October 2007 3 4 Table of Contents Zusammenfassung 9 Summary 13 Chapter 1 Introduction 17 1.1 Mantle mineralogy: inferences from petrological observations, seismology and experiments 18 1.2 Motivation 23 1.3 Aims of the study 27 Chapter 2 The calcium perovskite forming reaction in the transition zone of the Earth’s mantle, implications for the mid-transition zone seismic discontinuity at 520 km depth. 31 2.1 Introduction 31 2.2 Experimental details 34 2.2.1 Synthesis experiments 34 2.2.1.1 Starting composition synthesis 34 2.2.1.2 Pressure calibrant synthesis 35 2.2.2 Multianvil experimental study 36 2.2.2.1 Multianvil technique 36 2.2.2.2 Multianvil experiments 38 2.3 Characterization and analytical techniques employed in this study 42 2.4 Results 49 2.4.1 Thermodynamic modeling 55 5 2.5 Discussion 58 2.6 Conclusions 65 Chapter 3 An equation of state study of (Fe,Al)-bearing magnesium silicate perovskite single crystals: implications for lower mantle properties. 67 3.1 Introduction 67 3.2 Experimental details 71 3.2.1 Starting materials 71 3.2.2 Multianvil synthesis experiments 72 3.2.3 Quantification of perovskite Fe3+/∑Fe ratios 75 3.2.4 Crystal compositions 78 3.3 Diamond anvil cell compression experiments 80 3.3.1 Basics of diamond anvil cell single crystal compression technique 80 3.3.2 Compression experiments 81 3.4 Equation of state results 84 3.4.1 Theoretical background 84 3.4.2 Unit cell lattice parameter variation with pressure 85 3.4.3 FE-fE plot and EoS parameters 87 3.4.4 Octahedral tilting 94 3.5 Discussion 96 3.5.1 The elasticity of the Earth’s lower mantle 96 3.5.2 The effect of pressure on perovskite substitution 102 OAlFe VIVIII 3 33 ++ 3.6 Conclusions 105 Chapter 4 A calorimetric study of the Mg3(Mg,Si)Si3O12(majorite)- Mg3Al2Si3O12(pyrope) garnet solid solution 107 4.1 Introduction 107 4.2 Starting material synthesis 109 4.2.1 Piston-cylinder synthesis experiment 110 6 4.2.2 Multianvil synthesis experiment 111 4.3 Calorimetric measurements 114 4.3.1 Basic principals 114 4.3.2 Enthalpy measurements 115 4.4 Results 119 4.5 Discussion 122 4.6 Conclusions 125 5 Conclusions 127 Appendix A 131 Appendix B 139 References 141 Erklärung 161 7 8 Zusammenfassung Im Rahmen der vorliegenden Doktorarbeit wurden drei experimentelle Untersuchungen durchgeführt, um zu verstehen, wie die Zusammensetzung des Erdmantels die Stabilität von Mineralen und ihre elastische Eigenschaften bestimmt, und wie diese wiederum die seismischen Eigenschaften des tiefen Erdmantels beeinflussen. Die Phasenbeziehungen von Calciumsilikat-Perowskit wurden in Hochdruck-Hochtemperaturexperimenten untersucht, um den Effekt seiner Bildung auf die Charakteristika der seismischen Diskontinuität bei 520 km Tiefe im Erdmantel zu bestimmen. Der Effekt von variabler Zuammensetzung auf die Kompressibilität von Magnesiumsilikat-Perowskit wurde untersucht, um die geophysikalischen Konsequenzen chemischer Heterogenität im Erdmantel zu verstehen. Kalorimetrische Messungen der Granatmischreihe Pyrop-Majorit wurden durchgeführt, um grundsätzliche thermodynamische Daten für die Modellierung der Bildungsreaktionen von Magnesiumund Calciumsilikat-Perowskit bereitzustellen. (i) Die Calciumsilikat-Perowskit bildende Reaktion in der Übergangszone des Erdmantels: Implikationen für seismische Diskontinuität in der mittleren Übergangszone bei 520 km Tiefe. Globale seismische Beobachtungen zeigen, dass das Auftreten der seismischen Diskontinuität bei 520 km Tiefe in der mittleren Mantelübergangszone ein komplexes Erscheinungsbild hat. In einigen Regionen des Erdmantels erscheint diese Diskontinuität in zwei Diskontinuitäten in leicht unterschiedlichen Tiefen aufgespalten zu sein. Es ist daher vorgeschlagen worden, dass unter den Bedingungen der mittleren Übergangszone ausser dem Phasenübergang von 9 16 The data also show that the Mg3(Mg,Si)Si3O12(majorite)-Mg3Al2Si3O12(pyrope) solid solution has strong non-ideal mixing properties. Chapter 1 Introduction Knowledge of the physical and chemical state of the deep interior of the Earth is the key to understanding its evolution and dynamics. Due to limitations in direct sampling of the deep Earth, most inferences about its interior must be based on indirect information obtained through geophysical and geochemical observations. Experimental data on mineral properties at deep mantle conditions become crucial, particularly in interpreting geophysical observations and assessing their implications for mantle geochemistry and dynamics. The mantle represents the largest geochemical reservoir of the silicate Earth, comprising about 80% of volume of the total Earth. Our main information on the composition of the mantle comes from xenoliths, pieces of the mantle brought to the surface of the Earth by volcanic activity, and peridotite massifs, sections of the mantle emplaced in the crust by tectonic movements. In addition, the geochemistry of volcanic rocks produced at mid-oceanic ridges by partial melting of the mantle aids our understanding of mantle composition and a comparison with undifferentiated meteorite samples helps to constrain the likely composition of the Earth as a whole. The constancy in composition of erupted basaltic magmas with time infers that the bulk composition of the mantle may be relatively uniform as expected for a convecting mixed reservoir. Aside from a few inclusions in diamonds that may have a deeper origin, samples from the mantle come from depths of no more than 200 km. So, inferences about deep mantle structure need to be based on geophysical observations. Geophysical observations of the mantle have identified a number of globally-observed radial discontinuities that reflect seismic waves. These discontinuities are caused by sharp changes in mantle elastic properties and density. Such 17 Chapter 1: Mantle mineralogy observations have led to the division of the mantle into an upper mantle, a transition zone and a lower mantle, each separated by seismic discontinuities at a particular depth. The discontinuities may result from mineral phase transitions in an isochemical mantle but changes in the chemical composition of the mantle with depth would also be a possible explanation. As different possibilities exist, a range of geophysical observations need to be examined to determine the most likely explanation. Except for strongly incompatible trace elements that are concentrated in the crust and siderophile and chalcophile elements that partition into the core, the mantle is the major repository for most elements in the Earth. An understanding of mantle composition and its compositional structure has, therefore, great implications for the chemical composition of the entire Earth. In addition geophysical observations contain information on the temperature of the mantle that can be extracted through comparison with experimentally determined measurements. Such information is crucial for understanding the dynamics and thermal evolution of the Earth. 1.1. Mantle mineralogy: inferences from petrological observations, seismology and experiments Xenoliths brought to the surface by kimberlite and alkali basalt magmas in addition to abyssal and massif peridotites emplaced in the crust, show that the upper part of the mantle is generally an ultramafic rock termed peridotite, which is comprised of the minerals olivine, clinopyroxene, orthopyroxene and an Al-bearing phase. At depths shallower than 70 km, this Al bearing phase is plagioclase, but it transforms to spinel and then garnet with increasing depth. The deepest xenoliths, however, generally become entrained from depths no deeper than ~200 km. Geophysical techniques, primarily seismology, then become the only method for investigating the deeper mantle. Constraints also come from observations of the moment of inertia, gravity, heat flow of the Earth and electrical conductivity measurements. These geophyiscal observations, however, require interpretation using experimentally determined mineral properties and phase equilibria determinations. The general approach adopted for deciphering the mineralogy of the mantle is to assume that the deep mantle is of a similar composition to the upper mantle as inferred 18 Chapter 1: Mantle mineralogy from peridotite samples and to experimentally determine the mineralogy of this bulk composition as a function of depth along an adiabatic thermal gradient. Based on such mineralogical models sound velocities are then calculated using available thermoelastic parameters and the results are compared with seismic reference models. By examining how well seismic velocities in the mantle match the model determinations, the assumptions made concerning the chemistry and temperature of the mantle as a function of depth can be assessed. Even though there are limitations in the existing mineral physics data that inhibit a robust interpretation of seismic data, within the uncertainties of the current data a reasonable match is found between seismic observations of S and P wave velocities with depth and models constructed for a peridotitic mantle composition (Cammarano et al., 2005). However, seismic tomography has shown that heterogeneities of thermal and chemical origin, most likely arising from the presence of subducted lithospheric slabs, exist in the mantle. In addition observations of seismic heterogeneity in the lower mantle do not seem to be well explained by thermal variations alone (Trampert et al., 2004). Some observations have also been interpreted to result from the presence of partial melting in the deep mantle (Williams and Garnero, 1996; Karato and Jung, 1998). At pressures > 3 GPa such melting is likely to only arise if the mantle is suitably enriched in volatile elements like H and C at these depths, which also raises questions as to the degree of chemical homogeneity that can be expected in the deep mantle. Chemical composition of the mantle Various approaches have been adopted to infer the average chemical composition of the mantle. The simplest approach is to assume that the composition of the mantle is the same as some pristine peridotite samples e.g., KLB1 etc. recovered at the Earth’s surface. Ringwood, (1975) proposed a chemical composition based on recombining basaltic partial melt with the refractory mantle residue left behind. At mid-oceanic ridges basalts are generated by approximately 10% partial melting of peridotite mantle. Ringwood argued that the composition of the pristine mantle, that he termed pyrolite, would be the same as a mixture of peridotite melt residue and primitive mid-oceanic ridge basalt (MORB) compositions. Rather than assuming a particular residual and basalt composition, as Ringwood did, an alternative approach is to examine melt extraction trends in peridotite 19 Chapter 1: Mantle mineralogy xenoliths and massifs and extrapolate these trends back to an unmelted precursor (Walter, 2004). Models also exist based on chondritic meteorite compositions (Allegre et al., 1995; Javoy, 1995). These models assume that the bulk Earth formed from chondrite meteorite material but that siderophile elements were extracted from this composition to the core leaving most major elements in the mantle in chondritic proportions. Because all undifferentiated meteorites have higher Si/Mg ratios than the upper mantle, such models require an additional reservoir rich in Si to form. Usually it is proposed that Si was extracted either to the core or to the lower mantle. Such models are often cited as evidence that the lower mantle may have a chemical composition different from that of the upper mantle. Seismic observations of the mantle Vital constraints can be placed on mantle mineralogy from seismological reference models (Travel time-tables of Jefferys and Bullen, 1940; 1066A, 1066B models of Gilbert and Dziewonski, 1975; PREM Preliminary Reference Earth Model of Dziewonski and Anderson, 1981; IASP 91 model of Kennet and Engdahl, 1991; AK135, Kennet et al., 1995; Cammarano et al., 2005). Seismic reference models such as the Preliminary Reference Earth Model (Dziewonski and Anderson, 1981) provide a radially symmetric velocity depth profile of the Earth based on the inversion of body wave travel time data and free oscillations of the Earth. In these models velocities in the Earth’s interior are refined to a set of polynomial functions that operate over a specified depth interval, with the assumption that mantle discontinuities occur at predetermined depths. These discontinuity depths have been determined by analyzing seismic waves that are refracted by the discontinuity or from seismic waves that are converted from S to P waves, or visa versa, at the discontinuity. Information on amplitude of these discontinuity jumps and depth interval of velocity change across the discontinuity can also be used to place much needed constraints on mantle chemical structure (Stixrude, 1997). Pyrolite mineralogy as a function of depth The variation in the proportion of minerals that would crystallize from a pyrolitic bulk composition as a function of depth is show in Fig. 1.1 for the top 1000 km of the mantle 20 Chapter 1: Mantle mineralogy (Ringwood, 1991). For purposes of discussion this diagram can be conveniently divided into two components, which occupy each side of the diagram. On the left side is olivine and the higher pressure olivine polymorphs, while the right side is composed of the nonolivine Si and Al-rich phases of the mantle. Phase transformations in olivine occur with increasing pressure over very narrow depth intervals, whereas the phase transitions in nonolivine phases are gradual and occur over broad depth intervals. At a depth of 410 km, (Mg,Fe)2SiO4 olivine transforms to the high-pressure polymorph of wadsleyite β- (Mg,Fe)2SiO4 which is generally considered to cause the 410 seismic discontinuity that is globally observed at this depth. The 410 km seismic discontinuity marks the top of the transition zone region of the mantle, so called because a transition in seismic velocity gradient occurs within this region. At around 17.5 GPa corresponding to a depth of 520 km, wadsleyite undergoes an iso-chemical phase transition to ringwoodite, which likely causes a weak seismic discontinuity observed regionally at approximately this depth. The bottom of the transition zone and top of the lower mantle occurs at 660 km depth, (approximately 24 GPa), where ringwoodite breaks down to an assemblage of (Mg,Fe)(Si,Al)O3 with the perovskite structure and (Mg,Fe)O magnesiowüstite. This transformation causes a strong globally observed seismic discontinuity at this depth. Above 3 GPa the non-olivine phases in a pyrolite composition are orthopyroxene and clinopyroxene and the Al-rich phase garnet. With increasing pressure, both orthopyroxene and clinopyroxene components start to partition into garnet. This results from the substitution of Mg, Fe and Si onto the garnet Al octahedral position to create a component with pyroxene stoichiometry called majorite {i.e., (Mg,Fe)4Si4O12}. By midtransition zone conditions pyroxenes have completely dissolved into the garnet structure with garnet having the approximate stoichiometry (Mg,Fe,Ca)3(Mg,Al,Si)2Si3O12. At midtransition zone conditions of approximately 18 GPa, CaSiO3 starts to exsolve forming the 21 Chapter 1: Mantle mineralogy Wadsle y ite (M g ,Fe) SiO 24 Upper Mantle Ma g nesium silicate Perovskite (M g ,Fe,Al)(Si,Al)O 3 Lower Mantle Transition Zone Ferropericlase (Mg,Fe)O Rin g woodite (Mg,Fe) SiO 24 Garnet (Mg,Fe,Ca) Al Si O 32312 Olivine (Mg,Fe) SiO 24 Orthopyroxene (Mg,Fe)SiO 3 Ma j oritic g arnet (Mg,Fe,Ca) (Mg,Si,Al) Si O 32312 Clinopyroxene (Mg,Ca)SiO 3 00.2 0.4 0.6 0.8 1. 0 Calcium Silicate perovskite CaSiO 3 1000 800 600 400 200 D ept h ( k m ) Figure 1.1 Pyrolite mantle mineralogy as a function of mineral volume fraction and depth variation. The small orange and pink region in upper right hand corner represents the stability field of feldspar and spinel respectively. Minerals recovered from high pressure-high temperature experiments are shown in the insets. Field of view of inset is ~200 microns. (With permission from Dr. D. J. Frost) separate phase calcium silicate perovskite. At the top of the lower mantle the remaining garnet starts to dissolve into magnesium silicate perovskite. By depths of approximately 750 km in the lower mantle, a pyrolite composition assemblage comprises magnesium silicate perovskite, magnesiowüstite and calcium silicate perovskite. This assemblage is believed to be stable throughout the bulk of the lower mantle and only at pressures corresponding to the D’’ layer on top of the core mantle boundary does magnesium silicate perovskite transform to a post perovskite polymorph with the structure of CaIrO3 (Murakami et al., 2004). 22 Chapter 1: Motivation Mantle Heterogeneity Though the major element composition of the upper mantle appears to have remained uniform over recorded geologic time, trace element and isotopic studies (Sun and McDonough, 1989) imply the presence of significant heterogeneities in the basalt source region. This may indicate the comparatively more mobile nature of incompatible trace elements in comparison to major elements or it may indicate heterogeneities being created by the presence of subducted oceanic lithosphere residing in the mantle (Christensen and Hofmaan, 1994). It is quite likely that a subducted slab would take a significant length of time to be rehomogenized by convective stirring in the mantle (Holzapfel et al., 2005). It is, therefore, quite possible that large regions of the mantle are comprised of mechanical mixtures of melt-depleted peridotite and subducted oceanic crust on a variety of length scales. This will have significant implications for the structure and composition of the mantle in addition to its potential effect on geophysical observations. 1.2 Motivation Changes in mantle mineralogy as a result of phase transformations that occur over relatively short depth intervals have been found to coincide with major seismic discontinuities in the Earth’s mantle as shown in Fig. 1.2. A significant number of experimental studies have been devoted to determining the phase relations and physical properties of mineral phases associated with major seismic discontinuities such as those at 410 km (Katsura and Ito, 1989; Irifune and Isshiki, 1998; Kiefer et al., 2001; Frost, 2003; Li et al., 1998) and 660 km (Ito and Takahashi, 1989; Shim et al., 2001; Li and Li, 2003). Of key interest is to understand how temperature and chemical variations in the mantle may affect seismically observable phenomena such as the depth, depth interval and amplitude of the discontinuities in addition to the ambient sound velocity and density of the mantle. For this experimental data are required on the influence of variable mantle chemistry on phase stabilities and elastic properties. 23 Chapter 1: Motivation 12 14 16 18 20 22 24 26 1200 1300 1400 1500 1600 1700 1800 Pressure (GPa) Te m perature (° C ) 450 500 550 600 650 700400 Depth (km) Olivine Wadsleyite Ringwoodite Perovskite + Ferropericlase [1] [2] [4] [5] [3] Figure 1.2. Pressure-temperature slopes of phase transformations in the Earth’s mantle compared with average seismic discontinuity depths (solid vertical lines) for the 410 km (in green), 520 km (in blue) and 660 km (in red) and global topography (vertical shaded regions). Double curves mark the beginning and end of divariant regions for the olivine to wadslyeite and wadsleyite to ringwoodite transformations, and the shaded area in the curves show the temperature ranges compatible with globally-observed topography of these discontinuities. Individual curves for the Mg2SiO4 ringwoodite to perovskite + ferropericalse reaction are shown from different studies [1] Irifune et al., (1998) [2] Katsura et al., (2003) [3] Fei et al., 2004 [4] Ito and Takahashi, (1989) [5] Shim et al., (2001). (With permission from Dr. D. J. Frost) The olivine to wadsleyite transformation that causes the 410 km discontinuity occurs in the MgO-FeO-SiO2 system and is therefore insensitive to chemical variations apart from the Fe/Mg ratio, although large concentrations of H2O may also have an effect (Wood 1995; Smyth and Frost, 2002; Frost and Doleĵs, 2007). The 660 km discontinuity, on the other hand, is likely to be more affected by variations in chemistry because, in addition to the MgO-FeO-SiO2 system, perovskite will also be influenced by the Al2O3 content of the mantle and the Fe2O3 content. Recent experimental studies have shown that Fe2O3 has a strong affinity for magnesium silicate perovskite, which is stabilized by Al in the structure as a result of a coupled substitution. Such a substitution mechanism can also 24 Chapter 1: Motivation affect the elastic properties of perovskite. From this analysis we can expect that the formation of magnesium silicate perovskite in the Earth’s mantle is likely to be a complex process both physically and chemically. In addition, however, there is evidence that seismic heterogeneity in the deep lower mantle may exist and that the observed variations are not well correlated with changes in temperature (Trampert et al., 2004). To understand and interpret these observations we require information on how chemistry may influence the elastic properties of major lower mantle minerals. In addition to major seismic discontinuities, minor weak seismic discontinuities are also known to exist in the Earth’s interior such as the Hales discontinuity at about 60-90 km depth, possibly caused by MgAl2O4 spinel transforming to garnet and the Lehmann discontinuity at 220 km depth that may mark a change in mantle anisotropy. Another weak discontinuity was reported by Shearer, (1990), from the mid-transition zone at 520 km depth. This has been considered to be caused by the wadsleyite to ringwoodite transition. However, recent findings of Deuss and Woodhouse, (2001) show that this discontinuity is split at some locations in the mantle into two distinct discontinuities one closer to 500 km and the other at approximately 560 km. The most likely explanation for the occurrence of a split in the 520 is that it results from the formation of calcium silicate perovskite (CaSiO3) from majoritic garnet, which also occurs in this depth range. Previously the calcium silicate perovskite forming reaction had been poorly studied in pressure-temperature space and it was hard to constrain the exact pressure and temperature range over which it occurs. Studying this reaction is further complicated by the fact that majoritic garnet from which CaSiO3 perovskite exsolves is a multi-component solid solution and as such the exsolution reaction is likely to be dependent on a significant number of variables. A number of issues have remained unsolved about perovskite-forming reactions in the Earth’s mantle, for which laboratory studies can provide substantial information. For instance, (i) The pressure, temperature and compositional effects on the formation of calcium silicate perovskite from majorite garnet have remained unaddressed. This exsolution reaction could possibly cause a seismic discontinuity but the existing data are insufficient to verify this. 25 Chapter 2: Introduction formation of calcium silicate perovskite therefore is clearly dependent on the conditions where garnet becomes saturated in CaO. Throughout the lower mantle calcium silicate perovskite is the dominant CaO bearing mineral. The formation of calcium silicate perovskite from garnet in natural systems with peridotitic and basaltic compositions occurs at approximately the same pressure inspite of differences in bulk CaO content, with basaltic compositions containing double the amount of CaO than contained in a peridotitic composition (Irifune and Ringwood, 1993; Nisihara and Takahashi, 2001). This clearly implies that the calcium solubility in garnet is dependent on some variable also other than pressure. The most likely explanation could be that the solubility of CaO in garnet is also dependent on the Al/Si ratio of garnet i.e., the proportion of the majorite component in garnet. The formation of the dense calcium silicate perovskite mineral in the Earth’s interior could cause discontinuities in the speed of sound waves as they pass through the interior, which is detectable at the Earth’s surface. Recent seismic observations of the transition zone have identified a discontinuity at the mid-transition zone depth of 520 km often designated as 520d (520 km seismic discontinuity) (Shearer, 1990; Shearer, 1996; Shearer, 2000; Deuss and Woodhouse, 2001; Gilbert et al., 2003; van der Meijde et al., 2005; Deuss et al., 2006). This discontinuity is found to be split into two discontinuities in some regions of the mantle, one at an approximately 500 km depth and another at a deeper depth of 560 km (Deuss and Woodhouse, 2001). The wadsleyite (β) to ringwoodite (γ) transition is often implicated to be the cause of the 520d (Weidner and Wang, 2000). As the exsolution of calcium silicate perovskite from majoritic garnet occurs at a similar mantle depth, it could also cause a discontinuity and thus result in a double or split 520d (Ita and Stixrude, 1992). The variability in the depth of these discontinuities could be a strong function of change in either temperature or composition between different regions of the mantle. The most likely major type of chemical variation in the silicate mantle arises due to fractionation of oceanic crust and lithosphere at the mid-oceanic ridges through partial melting. The major elements that are fractionated during partial melting are Al, Ca, Si, and Na, which become concentrated in the oceanic crust and correspondingly depleted in the lithosphere (Walter, 2003). These components are eventually recycled back to the mantle by subduction of oceanic crust. Ongoing convective stirring in the mantle may homogenize these chemically distinct domains, although this may take a significant period of time for 32 Chapter 2: Introduction homogenization to achieve local chemical equilibrium (van Keken, 2002; Holzapfel, 2005). However, if these chemically-distinct domains tend to accumulate due to density or rheological contrasts, it may led to formation of long-term heterogeneities in the mantle that are resistant to homogenization (Tackeley et al., 1993; Christensen and Hofmann, 1994; Helffrich and Wood, 2001). The major global seismic discontinuities at 410 km and 660 km depths in the Earth’s mantle correspond to phase transformation involving olivine {α-(Mg,Fe)2SiO4} to its high pressure polymorph of wadsleyite {β-(Mg,Fe)2SiO4} and ringwoodite {γ- (Mg,Fe)2SiO4} breaking down to magnesiowüstite and magnesium silicate perovskite respectively (Hellfrich and Wood, 2001). As these phases do not involve components like Ca and Al, which are significantly fractionated in the mantle, discontinuities arising due to these phase transformations tell us little about likely chemical variations in the mantle. However, a seismic discontinuity like that of 520d which may as well arise due to exsolution of calcium perovskite from majorite garnet involving components like Ca and Al, would be very sensitive to large scale mantle chemical heterogeneities; such as those that might result from the presence of significant proportions of remnant subducted oceanic crust in the mantle as the Ca and Al component are significantly fractionated at midoceanic ridges. As majorite garnet is a multi-component solid solution, the exsolution reaction of calcium perovskite from majorite garnet is perceived to be quite complex. Existing data (Irifune and Ringwood, 1993; Gasparik, 1996; Nishihara and Takahashi, 2001; Litasov and Ohtani, 2005) on this reaction are not sufficiently consistent for modeling this reaction over the range of pressure, temperature and bulk composition relevant for the mantle. Moreover, the pressure interval between experiments in existing studies is not narrow enough to accurately describe the shape of the emerging calcium perovskite stability field. The exsolution of calcium silicate perovskite for a range of mantle compositions was studied in the present work using high-pressure and high-temperature multianvil experiments. The solubility of CaSiO3 in garnet was measured as a function of garnet majorite content between 17 and 23 GPa in the temperature range 1200-1600°C in order to ascertain with high precision, how the depth interval of this reaction compares to the wadsleyite to ringwoodite transition and whether this reaction could occur over a narrow enough pressure interval to cause an observable seismic discontinuity. 33 Chapter 2: Experimental details 2.2. Experimental Details 2.2.1 Synthesis experiments 2.2.1.1 Starting composition synthesis When we consider major types of mantle rocks such as a peridotite or basalt, the major type of chemical variation in garnets of these rocks arises due to the substitution of the majorite component i.e., (Si, Mg) becoming incorporated into the octahedral Al position in the garnet structure. For this reason we choose four Ca-free garnet compositions on the (Mg,Fe)4SiO12(majorite) - (Mg,Fe)3Al2Si3O12(pyrope) join for starting materials (Table 2.1). As can be seen in the Fig. 2.1, a normal mantle peridotitic garnet will be more majoritic than a garnet from a subducted basaltic composition, which has been recycled back to the mantle. Glasses were synthesized out of reagent grade oxide mixtures of SiO2, Al2O3, Fe2O3 and MgO by fusing them at 1600°C in a 1-atmosphere furnace followed by rapid quenching in water. Glasses were analyzed for chemical composition by electron microprobe analysis using a point beam operating in wavelength dispersive mode at 15nA and 15kV. Quench recovered glasses were ground to powders and in order to reduce Fe3+ content of the glasses to Fe2+, the glass powders were reduced in a CO2/H2 gas mixing furnace, at a gas composition of 04/06 H2/CO2 (an approximate fO2 of 2 log units above iron-wüstite buffer) at 650°C for a day. Mössbauer analysis was performed on the reduced glass powders that confirmed the absence of Fe3+ in the glass powders. (Details on characterization techniques are discussed in next section). Table 2.1: Chemical composition of the starting oxide mixes for garnet glass synthesis based on 12 oxygens per formula unit. Composition SiO2 Al2O3 FeO MgO Total Si Al Fe Mg ∑Cations Peridotite 53.0 9.2 5.3 33.5 100 3.63 0.74 0.3 3.33 8 Basalt 48.3 17.4 4.8 29.5 100 3.3 1.4 0.274 3.026 8 Pyrope 43.9 24.9 4.4 26.8 100 3 2 0.249 2.751 8 Majorite 54.8 6.2 5.5 33.5 100 3.75 0.5 0.311 3.439 8 34 Chapter 2: Experimental details 0.8 0.85 0.9 0.95 1 0 0.2 0.4 0.6 0.8 1 Mg/(Mg+Fe) Extreme majorite Peridotite garnet Basaltic garnet Pyrope Mg4Si4O12 Mg3Al2Si3O12 Cubic Tetragonal Al2O3 mol% Figure 2.1: The glass starting compositions used in the present study are shown plotted on the majorite–pyrope solid solution join as a function of Mg number. In order to measure the solubility of Ca in the garnet at high pressure these glass powders were saturated in the CaSiO3 component by adding wollastonite (CaSiO3). The wollastonite phase was synthesized from an oxide mixture of calcium carbonate (CaCO3) and silica (SiO2) powder. The CaCO3 was first decarbonated for 24 hrs at 225°C in a 1atmosphere furnace. The oxide mixture of decarbonated CaCO3 and SiO2 was than placed in a 1-atmosphere furnace for 16 ½ hours at 1000°C. This furnace temperature was scheduled to heat slowly at a rate of 2°C per minute, which ensured slow release of any remaining CO2 from the sample. These samples were reground and again fused in a 1atmosphere furnace at 1300°C in successive stages over 48 hours duration in total to finally crystallise wollastonite. Powder X-ray diffraction was employed to confirm the synthesis of wollastonite. 2.2.1.2 Pressure calibrant synthesis Precise pressure determination is of utmost importance in our study. Therefore, in addition to using the normal oil pressure calibration for our multianvil experiments, we also used in-situ pressure calibration by including an olivine sample in all our experimental runs. Olivine compositions were synthesized from stoichiometric reagent grade oxide mixes of SiO2, MgO and Fe2O3. After thorough grinding of oxide mixture, pellets were made out of 35 Chapter 2: Experimental details these mixtures, that were reduced in a CO2/CO gas-mixing furnace at an fO2 of 14.1, log units at 1000-1200°C depending on the Fe concentration of the samples. After three successive steps, each of 24 hours of reduction and rehomogenization by crushing and grinding, finally homogeneous olivine crystallized. Powder x-ray diffraction on the product phases mixed with Si powder as an internal standard was used for phase identification. (Details of the characterization techniques mentioned herein are discussed in section 2.3.) 2.2.2 Multianvil experimental study 2.2.2.1 Multianvil technique Over the past 20 years use of multianvil apparatus for high-pressure, high-temperature experiments simulating mantle conditions have increased manifold and a number of publications have reviewed this technique and its applications (Kawai and Endo, 1970; Walker et al., 1990; Rubie, 1993; Rubie et al., 1993; Irifune, 2002; Frost et al., 2004; Keppler and Frost, 2005). A multianvil press works on the concept of reduction of area (A) by applying a constant force (F), thereby increasing the pressure (P) according to the relation P = F/A. Essentially, in a multianvil apparatus, a hydraulic press generates an uniaxial force which is exerted onto a set of 6 steel anvil, which is referred to as the first stage anvils. Two variations of first stage anvil design are known, a split sphere or a spilt cylinder. This set of 6 anvils creates a cubic volume that is filled with a set of eight cubes (either of tungsten carbide or sintered diamond) with truncated corners, which functions as the second stage anvils. These truncated anvils create an octahedral pressure chamber. In this pressure chamber fits in the pressure cell usually an octahedra of MgO containing the sample, which is compressed to the required pressure. A hole is drilled in the MgO octahedra for insertion of a tubular resistance heater, usually made of graphite, metal foils (inconel, platinum, rhenium) or LaCrO3. Stepped heaters, where thickness of the heater wall is increased in the central portion are also used sometimes to reduce thermal gradients across the large sample volumes. Sleeves of insulating material usually ZrO2 are placed around the heater to prevent excessive heat transport to the tungsten carbide anvils. Experimental sample is placed in the center of the pressure assembly and separated from the heater by an MgO sleeve. MgO spacers fill up the space above and below the sample capsule. The top 36 Chapter 2: Experimental details MgO spacer has a hole for insertion of thermocouple with an alumina tube for temperature measurements. Pyrophyllite gaskets are used for supporting the truncations and for pressure sealing the high-pressure chamber. The maximum pressure achievable is dependent on the force applied by the hydraulic press, the truncation edge length of the second stage anvils, the edge length of the MgO pressure cell and ultimately by the strength and hardness of the tungsten carbide. For our experimental investigations in the present work we have used different multianvil apparatus with varied designs located at the Bayerisches Geoinstitut. For our high-pressure studies we have employed different pressure assemblies i.e., different truncation edge lengths of the cubes and different octahedron edge lengths suiting our experimental pressure requirements. We have used tungsten carbide (WC) anvils as the second stage anvils from the commercial suppliers Toshiba (Japan) and Widia (Germany). WC cubes were isolated from the steel anvils by epoxy sheets and copper foils were used for contact between the pressure assembly (detail description of pressure assembly used in our study is given in section 2.2.2.2) and the first stage steel anvils. A thermocouple was inserted into the pressure assembly for temperature measurements and a copper coil protected the thermocouple in the gasket region (Fig. 2.2). A eurotherm controller converted the thermocouple e.m.f. to temperature. Figure. 2.2: Details of a multianvil apparatus and the experimental set up. On the left is a schematic view showing how the six inner anvils create a cubic space where the 8 tungsten carbide inner anvils fit in containing the pressure cell and directions of application of force. On the left a photograph showing the tungsten carbide anvil set up for an experiment. The MgO pressure medium can be seen placed inside the octahedral cavity formed by truncated tungsten carbide (WC) anvils. Out of the set of eight WC cubes, two cubes are not shown here to show the MgO octahedra inside. 37 Chapter 2: Experimental details 2.2.2.2 Multianvil experiments Multianvil experiments were performed in the pressure and temperature range of 17-23 GPa and 1200-1600°C, to constrain the CaSiO3 perovskite forming reaction as a function of composition in the mid-transition zone of the Earth’s mantle. Initial experimental considerations Initially, for our multianvil experiments we mixed the Ca-free garnet glass powder and the CaSiO3 wollastonite in a 1:1 ratio on a trail basis. The first experiment was conducted at 1400°C for duration of 24 hours; on analysis of the run products we saw clear zones in the garnets showing that the experiment had not reached equilibrium (Fig 2.3 A). This led us to increase the experimental duration to 48 hours, which did not improve the equilibrium kinetics. We therefore, increased the experimental temperature to 1600°C. The run product showed a clear lack of CaSiO3 for reaction with the garnet (Fig 2.3 B). So we finally chose a starting mixture of Ca-free garnet glass and CaSiO3 wollastonite in a 1:2 ratio (Fig 2.3 C), where we could clearly observe garnet in equilibrium with calcium perovskite. Garnet Garnet Ca-Pv Ca-Pv Garnet Ca-Pv 10 µm 10 µm (A) (B) (C) Figure. 2.3: Secondary and back-scattered electron images of different experimental run samples (A) Zonation in majorite garnet showing disequilibrium in a preliminary experiment. (B). Lack of CaSiO3 for further diffusion into majorite garnet. (C). Majorite garnet and calcium perovskite (CaPv) in equilibrium. During the experiment CaSiO3 diffuses into the garnet until it becomes saturated in this component. For the capsule material we initially tried an Al2O3 four-hole sleeve in which we placed the starting compositions in each hole. It reacted away at experimental temperatures and there was huge contamination from the capsule material into the starting materials (Fig 2.4 A). Then we tried a molybdenum (Mo) rod with spark eroded sample chambers for the 38 Chapter 2: Experimental details capsule; in this case the Fe of the starting compositions diffused into the capsule material. Finally, rhenium rod with spark-eroded sample chambers for the capsule material proved to be the right choice because it was stable at high temperatures and had no effect on the sample composition. Al 2 O 3 Al O capsule 23 reacted away Mo capsule Re capsule 100 µm 100 µm 100 µm (A) (B) (C) Figure 2.4: Different capsule materials tried out for our multianvil experiments (A) Alumina four hole sleeve which reacted away and only one sample could be recovered that too had contamination of Al2O3, (B) Mo (molybdenum) capsule made out of 1 mm diameter Mo rod with five spark eroded holes and (C) Shown here is a radial section through the high-pressure assembly the outer pressure medium and furnace with an inner four chamber Re (rhenium) capsule (white). Three sample chambers contain garnet plus Ca-perovskite assemblages (which appear lighter), while one sample chamber contains a (Mg,Fe)2SiO4 pressure calibrant sample (darker). Multianvil experiments were carried out using multianvil presses of 1000-ton and 1200-ton axial compression capacities for this study. Experiments were carried out using 10/5 and 10/4 pressure assemblies i.e., using a Cr2O3 doped MgO octahedra of 10 mm edge length in combination with tungsten carbide cubes with 5 or 4 mm truncation edge lengths. A semi-conducting LaCrO3 resistance furnace provided the electrical heating and the sample temperature was monitored using a W3%Re-W25%Re thermocouple, that was inserted axially into the furnace (Fig 2.5). The sample capsule was made out of 1mm long Re rod, that was spark eroded to produce four to five sample chambers each about 250 microns in diameter (Fig 2.4.C). The multi-chambered capsule allows us to run at least three of the garnet glass plus CaSiO3 starting materials with different majorite components (i.e., Al/Si ratios) in three chambers in a single experimental run along with a (Mg,Fe)2SiO4 powder in another chamber for pressure calibration. A Re disc of 0.025 mm thickness and an alumina disc of 0.2 mm thickness successively covered the upper surface of the sample capsule to avoid reaction with the thermocouple. Experiments were first compressed to the desired load and were subsequently heated to the required temperature for at least 24 hours. Experiments were quenched by cutting the power supply to the 39 Chapter 2: Experimental details furnace. Decompression of the experiment was usually performed over a period of 17 hours. MgO + 5 wt% Cr O octahedra 23 ZrO sleeve 2 LaCrO furnace 3 MgO sleeve Sample in a Re rod capsule Alumina cement and pyrophyllite Mo disc Thermocouple Copper coil Al O disc 23 Re foil disc 4mm Figure 2.5: A schematic diagram of a 10/4 pressure assembly used for multianvil experiments of the present study, showing an axial cross section through the assembly. (10/4 assembly = 10 mm MgO octahedron edge length and 4 mm tungsten carbide truncation edge length). During the experiments the starting glass plus CaSiO3 mixture crystallized as garnet and calcium silicate perovskite and CaSiO3 dissolved by diffusion into the garnet. We measured this solubility as a function of pressure, temperature and composition. When the majorite garnet became saturated in the Ca component, it started to exsolve Caperovskite with increasing pressure. As the garnet diffusion was found to be slow, we tried out fluxes for our experiments. We did an experiment where we chose the basaltic garnet plus CaSiO3 composition and mixed it with about 5-mol% of two different fluxes B2O3 and NaCl. The multi-chambered Re capsule allowed us to put these two fluxed compositions into two capsule chambers along with a composition without flux and a pressure calibrant in the other two chambers in the same experiment. Experiment was carried out at 19.5 GPa (oil bar pressure) at 1400°C. Analysis of the experimental run product showed that the composition with NaCl flux showed no enhancement in CaO solubility in the garnets, 40 Chapter 2: Experimental details instead it crystallized NaCl crystals. However, the one with the B2O3 flux showed significantly enhanced reaction in comparison with the same composition without any flux. This indicated that the B2O3 generated melt at the experimental temperature that promoted equilibrium. So, all the subsequent experiments were fluxed with 5-mol% of B2O3. This increased the rate of reaction considerably at 1600°C and 1400°C but reaction at 1200°C was still far too slow for equilibrium to be achieved on a feasible time scale. Reversal experiments To ensure that we achieved equilibrium in our experiments, reversal experiments were performed using Ca-bearing garnets as the starting material. For this we chose the maximum calcium-bearing garnet compositions from our forward runs. High purity oxide powders of CaO, Fe2O3, MgO, Al2O3 were mixed in the proper stoichiometric proportions and fused at 1600°C in a furnace and rapidly quenched in water to produce the desired glass phase. Glass powders were reduced under conditions similar to those used for the other glass powders (see section 2.2.1) and were characterized by Mössbauer analysis. Syntheses of the Ca-bearing garnets were carried out using a 5000-ton multianvil press, where large sample volumes could be utilized. Experiments were performed using 15 mm edge length truncated tungsten carbide cubes, which compressed a 25 mm edge length Cr doped MgO octahedra containing the sample capsule jacketed by an MgO spacer and a LaCrO3 furnace. Sample capsules of 3.5 mm in length and 2 mm in diameter were made out of Re foil, which was in contact with a Re75%W25%-W3%Re97% thermocouple for temperature measurements. Garnets were crystallized at 15 GPa and 1600°C within an experimental duration of 1.5 hours. Phase identification was carried out using powder x-ray diffraction. Table 2.2: Reversal Ca –bearing garnet glass compositions in cation proportions based on 12 oxygen per formula unit as determined from electron microprobe analysis. Composition Si Al Fe Mg Ca ∑Cations Ca-Peridotite 3.62 0.65 0.31 2.66 0.80 8 Ca-Basalt 3.37 1.06 0.29 2.32 1.04 8 Ca-Pyrope 3.27 1.34 0.27 1.99 1.20 8 Ca-Majorite 3.67 0.49 0.35 2.83 0.76 8 41 Chapter 2: Characterization techniques In the present study, the SEM technique was used on a routine basis for imaging the high-pressure, high-temperature multianvil run products that had been polished after being embedded in epoxy resin and with carbon coating to reduce charging on the surface (similar to that for EPMA analysis) at an acceleration voltage of 20 KeV and with different magnifications. Raman spectroscopy This spectroscopic method is based on the inelastic scattering of light called the Raman effect. When light is allowed to interact with a specimen, it excites the constituent molecules, which subsequently scatter the light. Most of this scattered light has a similar wavelength as the incident light while some of it is scattered with a different wavelength. This inelastically scattered light is the Raman scatter, which results from the changed molecular motions of the sample. The difference in energy between the incident light and Raman scattered light is equal to the energy of the scattering molecule and will be characteristic of a molecule and its environment in a specimen. By plotting the energy difference and the intensity of scattered light we obtain a Raman spectrum (Nasdala et al., 2004). In the present study we used a LABRAM Raman spectrometer with a He-Ne laser with the 632 nm red line excitation for phase identification of our experimental run products. This was especially useful in case of our pressure calibrants where we could easily distinguish between the (Mg,Fe)2SiO4 polymorphs using this technique. Raman spectra were collected at ambient temperature with an instrumental resolution of 2 cm-1 for the peak positions. By comparing the obtained spectra with the relevant standard Raman data available for the mineral phase from the literature we carried out the phase identification (Fig. 2.8) 48 Chapter 2: Results 600 800 1000 9000 10000 11000 12000 13000 14000 15000 Intensity Wave number cm-1 600 800 1000 1200 800 1000 1200 1400 1600 1800 Intensity Wave number cm-1 Wadsleyite Ringwoodite Figure 2.8: Raman spectra of wadsleyite and ringwoodite as crystallized by the in-situ pressure calibrants of our experiments. Phase identification was carried out by comparing these spectra with the existing Raman spectra from the literature for the concerned phase. 2.4. Results During the high pressure experiments the garnet glass and CaSiO3 compositions crystallized rapidly to garnet and perovskite respectively and CaSiO3 was dissolved into the garnet during the heating period. Experimental durations of at least 24 hours ensured equilibrium compositions at 1600°C at lower pressures. However, in some cases at higher pressures, the inner cores of garnet remained unequilbrated but equilibrium was achieved in the rims with the aid of the B2O3 flux as mentioned in the experimental section. As discussed earlier, equilibrium was crosschecked by analyzing the reversal experiments in which Ca-bearing garnets exsolved Ca perovskite at high pressure. We had equilibrium in our experiments because for a given pressure the extent of CaO solubility in the garnet phase after CaSiO3 exsolution converged to the same values as obtained in the forward runs within experimental error. At 1400°C, high Ca-bearing garnet was formed up to pressure of ~20 GPa, beyond which Ca contents in the garnet declined drastically. Time studies indicated that equilibrium was not achieved at these conditions; a possible explanation could be that at these conditions the B2O3 flux crystallized. In the case of experiments at 1200°C, even the fluxed experiments failed to reach equilibrium on a time scale of 48 hours. Longer duration requirements were not feasible. During the experiments the (Mg,Fe)2SiO4 olivine pressure calibrant crystallized to form co-existing high-pressure phases. Pressure was determined from the Mg2SiO4(forsterite)-Fe2SiO4(fayalite) phase diagram using the field of coexistence 49 Chapter 2: Results between (Mg,Fe)2SiO4 and (Mg,Fe)O magnesiowüstite plus stishovite (SiO2) (Fig. 2.9). This divariant region extends from the Fe2SiO4 ringwoodite to magnesiowüstite plus stishovite transformation at approximately 16 GPa to 23 GPa with the Fe/(Fe+Mg) ratio of both ringwoodite and magnesiowüstite (when coexisting with stishovite) decreasing with increasing pressure. This reaction has been well studied (Matsuzaka et al., 2000; Frost et al., 2001; Frost, 2003a). The Fe/(Fe+Mg) ratio of ringwoodite for example decreases by approximately 10% per GPa. As we can determine this ratio with an accuracy of approximately 1%, this gave us a precision in pressure determination of 0.1 GPa. A previous study had shown that the bulk Fe concentration of magnesiowüstite could be apparently high if ferric Fe is present. We calculated the pressure using the ringwoodite Fe concentration and used the phase relations previously determined at a similar oxygen fugacity (Frost et al., 2001; Frost, 2003a). In this way we calibrated accurately the pressure of Ca-perovksite formation relative to the pressure of phase transformations in the Mg2SiO4-Fe2SiO4 system (Frost et al., 2001; Frost, 2003b). The absolute error in pressure depends on determinations of end member phase transitions used to construct the existing Mg2SiO4-Fe2SiO4 phase diagram which are difficult to assess but could be up to 1 GPa. Our study relied more on the high precision of pressures determined relative to the Mg2SiO4-Fe2SiO4 phase diagram, rather than on accuracy in absolute pressure. When the magnesiowüstite grains were too small in our experiments to get a reliable analysis, the pressure was determined using the ringwoodite Fe/(Fe+Mg) ratio alone. In those cases where a poorly chosen Fe/(Fe+Mg) ratio of the starting olivine composition led to crystallization of a single phase ringwoodite, pressure was determined from the oil pressure of the multianvil experiment. In experiments where the pressure was measured, the determined pressure showed excellent relationship with the multianvil oil pressure. The relative uncertainty using this calibration curve in comparison to pressures determined from phase relations in Mg2SiO4-Fe2SiO4 system was approximately ± 0.5 GPa. In some of our experiments the pressure calibrants even crystallized in the stability field of perovskite plus magnesiowüstite and stishovite (Run No.S3478, S3784), see Appendix A, Table A.2 for the perovskite compositions. Experimental results are given in Table 2.4,where the relative pressure errors were calculated using the mismatch between the analysed Fe contents of ringwoodite and magnesiowüstite compared with the phase diagram of Frost et al., (2001). 50 Chapter 2: Results 0 0.2 0.4 0.6 0.8 1 16 18 20 22 24 26 Fe/(Fe+Mg) Pressure (GPa) Mg2SiO4Fe 2 SiO4 1600oC Ringwoodite Ring + Mw + Stish Pv + Mw Pv+ Mw+ Stish Mw+Stish Figure 2.9: Phase relations in the Mg2SiO4-Fe2SiO4 system between 1626 GPa at 1600 oC (Frost et al., 2001) used for calculating the pressures in our experiments. Olivine pressure calibrants during experiments crystallized into its corresponding high pressures phases of ringwoodite, magnesiowüstite and stishovite. From probe data we determined the Fe/Fe+Mg ratio of coexisting ringwoodite and magnesiowüstite after correction for ferric iron content (Frost, 2003a). By plotting the Fe/Fe+Mg values as shown marked by stars we could exactly determine the pressure in each of our experiments. Table 2.4: Experimental results for garnet compositions recovered from multianvil experiments as determined by electron microprobe analysis for different experimental runs. Results are listed as peridotite, basalt, pyrope and majorite based on the different garnet starting compositions as listed in table 2.1.Listed here are the experimental run numbers, pressures in GPa for each experiment determined as described in text and the cation proportions of the garnet compositions calculated based on 12 oxygen per formula unit. Data in Ring (ringwoodite) and MW (magnesiowüstite) columns are the Fe/Fe+Mg ratios used for calculation of pressure as described in text. (See Appendix A, Table A.3 for the probe data of all experimental runs) 51 Run no. Pressure Si Al Fe Mg Ca Total Ring MW 1600°C (in GPa) Pyrope S3550 19.2(3) 3.41(2) 1.08(4) 0.21(1) 2.25(6) 1.10(4) 8.050 0.626(3) 0.937641 S3549 19.9(2) 3.12(1) 1.52(2) 0.17(2) 2.25(5) 0.90(5) 8.038 0.542(3) 0.87(3) S3470 18.8(5) 3.32(5) 1.36(4) 0.05(2) 2.03(8) 1.22(1) 7.993 S3475 22.6(5) 3.11(6) 1.82(6) 0.207(6) 2.47(6) 0.37(3) 7.982 S3478 23(5) 3.09(2) 1.77(3) 0.28(5) 2.56(6) 0.32(3) 8.023 S3484 22.3(5) 3.08(5) 1.81(5) 0.24(7) 2.57(7) 0.30(7) 8.011 0.36(2) S3490 21.4(5) 3.10(5) 1.76(9) 0.23(1) 2.47(6) 0.44(1) 8.013 Basalt S3551 20.4(3) 3.33(6) 1.11(7) 0.1(1) 2.94(8) 0.61(4) 8.096 0.496(2) 0.88(2) S3550 19.2(3) 3.29(2) 1.36(4) 0.178(9) 1.90(6) 1.27(7) 8.023 0.626(3) 0.937641 Chapter 2: Results S3549 19.9(2) 3.38(1) 1.14(3) 0.16(2) 2.70(3) 0.65(3) 8.044 0.542(3) 0.87(3) S3548 19.5(2) 3.41(2) 1.04(3) 0.30(2) 2.56(3) 0.75(3) 8.068 0.605(5) 0.863(4) S3547 19.5(5) 3.45(2) 1.04(2) 0.21(1) 2.58(5) 0.74(4) 8.020 0.592(2) 0.844(6) S3764 21.2(3) 3.34(3) 1.26(2) 0.21(2) 2.85(6) 0.35(1) 8.021 0.426(5) 0.73(1) S3757 20.7(5) 3.36(3) 1.29(4) 0.199(8) 2.65(3) 0.47(3) 7.990 0.469(2) 0.774(6) S3783 22.3(2) 3.34(2) 1.31(2) 0.262(7) 2.81(4) 0.29(1) 8.010 0.350(4) 0.71(5) S3657 19.9(3) 3.38(2) 1.01(5) 0.44(1) 2.62(5) 0.64(2) 8.108 0.545(3) 0.88(1) S3655 19.9(5) 3.36(5) 1.25(7) 0.39(3) 2.40(6) 0.62(4) 8.018 S3655 19.9(5) 3.42(1) 1.04(3) 0.24(3) 2.77(9) 0.58(2) 8.060 S3460 18.5(5) 3.502(7) 1.06(2) 0.049(4) 2.22(6) 1.12(3) 7.970 S3470 18.8(5) 3.44(2) 1.09(1) 0.03(7) 2.37(3) 1.05(2) 8.006 S3478 23(5) 3.30(2) 1.33(3) 0.25(1) 2.89(6) 0.24(2) 8.028 S3480 22.1(5) 3.47(5) 1.16(7) 0.27(2) 2.67(4) 0.37(3) 7.950 0.380(5) S3484 22.3(5) 3.33(3) 1.34(2) 0.20(6) 2.93(3) 0.188(7) 8.000 0.36(2) S3498 21.4(5) 3.33(5) 1.28(9) 0.276(8) 2.80(6) 0.34(3) 8.031 0.28(2) Peridotite S3551 20.4(3) 3.54(4) 0.87(8) 0.26(3) 2.9(1) 0.44(8) 8.022 0.496(2) 0.88(2) S3550 19.2(3) 3.62(2) 0.70(3) 0.22(8) 2.58(3) 0.92(3) 8.030 0.626(3) 0.937641 S3549 19.9(2) 3.58(1) 0.76(2) 0.11(4) 3.04(2) 0.54(2) 8.033 0.542(3) 0.87(3) S3548 19.5(2) 3.61(1) 0.68(3) 0.23(2) 2.93(4) 0.59(3)8 8.051 0.605(5) 0.863(4) S3547 19.5(5) 3.57(2) 0.77(3) 0.167(8) 2.88(5) 0.63(2) 8.037 0.592(2) 0.844(6) S3764 21.2(3) 3.60(3) 0.80(2) 0.28(3) 3.02(2) 0.283(1) 8.000 0.426(5) 0.73(1) S3757 20.7(5) 3.65(1) 0.69(1) 0.24(5) 3.082(4) 0.322(1) 7.998 0.469(2) 0.774(6) S3784 23.5(4) 3.60(2) 0.68(9) 0.31(4) 3.4(1) 0.054(1) 8.061 0.545(3) 0.88(1) S3460 18.5(5) 3.67(4) 0.63(3) 0.10(2) 2.79(7) 0.81(5) 8.010 S3470 18.8(5) 3.664(3) 0.683(5) 0.05(2) 2.75(3) 0.84(3) 7.994 S3475 22.6(5) 3.62(2) 0.791(2) 0.28(1) 3.21(5) 0.08(2) 7.984 Majorite S3460 18.5(5) 3.78(4) 0.47(2) 0.098(9) 2.97(6) 0.65(4) 7.979 S3470 18.8(5) 3.77(1) 0.49(1) 0.034(5) 2.87(7) 0.81(3) 7.985 Reversals S3784 23.5(4) 3.15(1) 1.68(3) 0.26(2) 2.59(6 0.30(4) 8.000 0.545(3) 0.88(1) S3538 17.9(5) 3.24(5) 1.36(5) 0.05(2) 2.82(7) 0.57(6) 8.067 S3543 18.4(5) 3.14(6) 1.57(6) 0.16(3) 2.63(4) 0.55(5) 8.069 S3515 19.8(5) 3.41(2) 1.08(2) 0.22(7) 2.42(3) 1.01(4) 8.042 S3521 20.7(5) 3.55(4) 0.87(4) 0.33(2) 2.77(6) 0.48(4) 8.008 S3523 20.8(5) 3.40(2) 1.18(3) 0.23(1) 2.53(2) 0.63(2) 8.015 0.464(3) 1400°C Pyrope S3611 18.1(5) 3.17(3) 1.61(7) 0.144(7) 2.01(3) 1.08(5) 8.023 S3614 18.6(5) 3.15(3) 1.56(5) 0.197(6) 2.09(6) 1.055(5) 8.064 H2370 19.5(5) 3.10(6) 1.75(9) 0.236(5) 2.35(5) 0.58(4) 8.020 0.60(1) 0.93(1) H2375 17.9(5) 3.24(1) 1.46(4) 0.200(5) 2.00(3) 1.11(3) 8.025 H2241 19.6(5) 3.12(4) 1.76(6) 0.19(1) 2.41(5) 0.51(3) 7.996 Basalt S3611 18.1(5) 3.301(8) 1.31(2) 0.199(4) 2.12(2) 1.11(2) 8.044 S3614 18.6(5) 3.33(2) 1.17(4) 0.223(7) 2.49(2) 0.86(2) 8.080 H2375 17.9(5) 3.40(3) 1.13(7) 0.223(8) 2.22(6) 1.04(3) 8.028 Peridotite S3614 18.6(5) 3.56(1) 0.69(2) 0.263(7) 2.91(4) 0.66(4) 8.090 S3611 18.1(5) 3.60(3) 0.71(5) 0.26(5) 2.68(4) 0.78(4) 8.042 52 Chapter 2: Results The Ca contents of garnet in equilibrium with Ca-perovskite at 1600°C and at 1400°C are shown in Fig. 2.10 (A) and (B). Data for all the reversal experiments at 1600°C are shown in Fig. 2.10(A). It was observed from the experimental data at 1600°C that the solubility of the Ca in garnet in equilibrium with calcium perovskite decreases with pressure. However, the Ca solubility increases with the garnet majorite content (i.e., Al/Si ratio) at a given pressure. This implies that the CaO saturation or the initiation of CaSiO3 exsolution will occur at a higher CaO content with decreasing majorite component in majorite garnet at transition zone pressure-temperature conditions. Low Al/Si ratios of about 0.7 are representative of garnets formed in peridotitic bulk compositions, while higher values of about 1.4 are those for garnets formed in a basaltic composition. Experimental data at 1400°C also showed a similar Ca solubility trend in garnet, i.e., CaO solubility in the majorite garnet composition decreases with increasing majorite component Al/Si+Mg. However, there is a small but distinguishable decrease in the Ca solubility compared with results obtained at 1600°C. As Fe is also an important component in transition zone phases we carried out a few experiments to observe the effect of varying Fe concentration on the CaO solubility in the garnets in equilibrium with calcium perovskite. For this we mixed our fluxed starting compositions with a powder of 5-mol% Forsterite 80 composition, which is a relevant composition for the mantle depth at which this reaction takes place. We placed one ironbearing sample along with a normal starting composition in a single run using the multichambered capsule. On analysis of the experimental run products no significant change in the CaO solubility was observed compared with that of the starting samples without additional mixing of Fe. 53 Chapter 2: Results (A) Pressure(GPa) (B) 18.0 18.5 19.0 19.5 0.6 0.8 1.0 1.2 Pyrope Al = 2 Basalt Al = 1.4 Peridotite = 0.7 Garnet Ca content (p.f.u.) Pressure in GPa CaSiO3 garnet solubilty Figure 2.10. Experimental results for the Ca-content of garnet in equilibrium with Caperovskite as a function of pressure (A) at 1600°C and (B) at 1400°C. Symbols refer to starting compositions with different garnet majorite proportions named after the rock types in which the garnets would occur. Al (pfu) refers to the Al content per garnet formula unit, i.e. Al (pfu) = 2 refers to the formula (Mg,Fe)3Al2SiO12. These Al contents only reflect the starting material because the Al content is lowered during the experiments. Reversals were performed using presynthesised Ca-bearing garnets as starting materials. The curve in (A) shows the calculated Ca content of garnet for a rock of peridotite composition based on peridotite composition from McDonough and Sun, (1995). For discussion on the peridotite model see section 2.5. 54 Chapter 2: Results 2.4.1 Thermodynamic modeling The exsolution of calcium silicate perovskite from majorite garnet can be described by the following reaction Ca3Al2Si3O12+3/4(Mg,Fe)4Si4O12 = 3 CaSiO3 + (Mg,Fe)3Al2Si3O12 [1] Garnet Garnet Perovskite Garnet where, the grossular component in the garnet exsolves to form calcium silicate perovskite, which leaves the garnet increasingly aluminium rich and combined with the majorite component, produces pyrope. The experimental results were fitted to a thermodynamic model based on reaction [1]. The important variables in this fit are the volume change of the reaction and the nonideal mixing parameters for majoritic garnet. Because the data cover a range of garnetmajorite compositions, this model was used to calculate the exsolution of CaSiO3 over a range of bulk compositions relevant for the mantle. This also allowed our results to be extrapolated to lower temperatures at which slow kinetics inhibited the achievement of equilibrium in experiments. At equilibrium conditions the standard state free energy change for reaction [1] is related to the activities of the reacting components by the equation, 0 ,TP G∆ [] [ ] [ ] 4 3 0 ,12441232312323 lnln aaa Gt OSiMg Gt OSiAlCa Gt OSiAlMg TP RTKRTG −=−=∆ [2] where R is the gas constant, K is the equilibrium constant and is the activity of component i in garnet. In order to estimate activities it was assumed that Ca mixes only on the dodecahedral site in garnet and that Fe behaves identically to Mg. The activity of CaSiO Gt i a 3 is unity whereas the activities of the garnet components are described by () ( ) 3 1244 1244 XX adodec Mg oct Maj Gt OSiMg Gt OSiMg γ = [3] () ( ) 3 1 12323 12323 XX adodec Ca oct Maj Gt OSiAlCa Gt OSiAlCa −= γ [4] () ( ) 3 1 12323 12323 dodec Mg oct Maj Gt OSiAlMg Gt OSiAlMg XX a−= γ [5] where,      − =2 2nAl Xoct Maj [6] 55 Chapter 2: Results      −− =3 3nFenCa Xdodec Mg [7]       =3 nCa Xdodec Ca [8] nAl, nCa and nFe are element proportions in garnet based on 12 oxygen formula units and γ is the activity coefficient. Si and Mg are therefore assumed to be locally ordered on the octahedral site. The activity coefficients (γ) were first determined using a four-component symmetric solution model that include terms for the non-ideality of majorite mixing. However, all majorite terms, which were refined in the fitting, were very small and were therefore ignored. A ternary symmetric solution model was then used that accounted for non-ideality resulting from Mg, Ca and Fe mixing only on the dodecahedral site, i.e., ( ) WWW XX W X W XCaFeMgFeMgCa FeCa MgFeMgCa Gt OSiAlMg FeCa RT −+++= 22 12323 ln γ [9] ( ) WWW XX W X W XMgFeCaFeMgCa FeMg CaFeMgCa Gt OSiAlCa Fe Mg RT −+++= 2 2 12323 ln γ [10] where W are Margules interaction parameters. The following values were used based on published values (O’Neill and Wood, 1979; Frost, 2003b). 300= WMgFe J/mol [11] 2000= W CaFe J/mol [12] Although many studies show the pyrope-grossular solid solution to have significant asymmetric excess enthalpy, entropy and volume properties, extrapolation of the many published models to pressures in excess of 20 GPa and 1873K yields wildly disparate activity coefficients. We therefore have no option but to treat WMgCa as an adjustable parameter but the range of Ca contents covered by our data makes an asymmetric fit unconstrained and unnecessary for our purposes. Therefore, we refine symmetric terms for WMgCa but constrain them to vary over a range predefined from literature values to get, P WMgCa 3008000 += J/mol [13] where P is in GPa. 56 Chapter 2: Results Fig. 2.11, shows the RTlnK calculated from the experimental data using the activity model. The data were fited in a least squares refinement to an equation for that gave, 0 ,TP G∆ PTG TP 12560773.26140763 0 ,−+−=∆ J/mol [14] with T in K and P in GPa. As is the standard Gibbs free energy change of the pure components at pressure and temperature, this can be simply described in terms of entropy, enthalpy and volume changes by, 0 ,TP G∆ VPSTHG TP ∆+∆−∆=∆ 0 , [15] We fit the calculated values for the -RTlnK term to equation [15] .The volume change of equation [1] is determined as 12.56 cm3 mol-1 which is very reasonable given the volumes and equation of state properties of the phases involved 20000 40000 60000 80000 100000 120000 140000 17 18 19 20 21 22 23 24 Basaltic Peridotitic Pyrope Reversals 1400°C Fit 1600°C Fit 1400°C RTlnK (kJ/mol) Pressure (GPa) Figure. 2.11: A plot of RTlnK as defined in equation (2) as a function of pressure. Data from various starting compositions at 1600°C are indicated, while pyrope, basaltic and peridotitic compositions are plotted together at 1400°C. Fitted curves for described using equation (14) are shown at both temperatures. 0 ,TP G∆ 57 Chapter 2: Discussion 1200 1300 1400 1500 1600 440 480 520 560 600 Majorite - Calcium perovskite transition mid point Wadsleyite-Ringwoodite transition mid point Temperature °C Depth in Km Figure. 2.14: Comparison of the transition pressure for wadsleyite-ringwoodite and majorite-Ca-perovskite as function of mantle temperatures of 1400°C and 1600°C showing positive Clapeyron slopes as represented by the solid lines for both the reaction. The closed circles are the transition mid points for wadsleyite-ringwoodite based on work of Frost, (2003) and the open squares are the transition mid points for the majorite to Ca-perovskite from the present experimental study. A more consistent explanation is that in many regions where only a single 520 km discontinuity is observed the Ca-perovskite reaction is invisible because the Ca content of the mantle is too low i.e., the region has undergone partial melting at the surface at some point in its history and is depleted in a fertile component. A strong 560 reflection, on the other hand, would indicate the presence of fertile mantle or mantle that contains a significant component of recycled MORB crust. The impedance contrast for the Caperovskite reaction in MORB is 2.8 %. As no wadsleyite to ringwoodite transition occurs in MORB, mantle with an enriched recycled MORB component would have a weaker 510 discontinuity but a much stronger reflection at approximately 560 km, in line with some of the observations. The 520 km discontinuity becomes uplifted to approximately 500 km as it splits, which based on the Clapeyron slope of the wadsleyite to ringwoodite transition implies local mantle temperatures that are cooler by at least 100°. As these regions also display a strong Ca-perovskite transition a good explanation for the splitting would be that it occurs in areas that contain significant amounts of subducted oceanic crust. If subducted slabs descend at high angle directly into the lower mantle these regions would not be expected to 64 Chapter 2: Conclusions be nearly as globally extensive as they appear to be. However, the widespread splitting of 520 would be easily explained if subducted material is accumulating at the base of the transition zone, as a result of either density or rheological contrast with the lower mantle (Christensen, 1997; Karason and van der Hilst, 2000), creating relatively cool flatlying regions rich in oceanic crust. 2.6. Conclusions Our experimental data support the hypothesis that the wadsleyite to ringwoodite transition and the Ca-perovskite forming reactions could produce two distinct discontinuities between 500 and 560 km. The presence and amplitude of a discontinuity at approximately 560 km will be sensitive to the Ca-content of the mantle and will consequently be an indicator for the presence of fertile mantle or mantle enriched in recycled oceanic crust. This is, therefore, the first mantle discontinuity to be identified that is sensitive to the main type of large-scale chemical heterogeneity expected in the Earth’s mantle. The observed variability in 520 km seismic discontinuity splitting indicates that the mantle at this depth may be heterogeneous over quite extensive regions. On considering the geographical distribution of the split 520 km seismic discontinuity, it is observed that Deuss and Woodhouse, (2001), in their seismological study reported the presence of a distinctly split 520 km seismic discontinuity from traces across North America, the East Pacific region, Indonesia and in the areas around the North African Shield. Like wise, Gilbert et al., (2003) also reported a split 520 km seismic discontinuity in areas beneath the Colorado Plateau, the Basin and Range province and the Rocky Mountains in the North Western United States. These regions are all undoubtedly sites of modern day or past subduction. Therefore, a correlation can be drawn supporting our idea of enrichment of Ca in the mantle by subduction of oceanic lithosphere beneath continental plates from the geographical distribution of the observed split 520 km seismic discontinuity. Further studies into the variability and splitting shown by the 520 km seismic discontinuity will provide key insights into the circulation of subducted crust and the lateral distribution of chemical heterogeneity in the transition zone. 65 66 Chapter 3 The compressibility of (Fe,Al)-bearing magnesium silicate perovskite determined by single-crystal X-ray diffraction: implications for lower mantle properties. 3.1 Introduction The transformation of (Mg,Fe)2SiO4 ringwoodite to (Mg,Fe)SiO3 perovskite and (Mg,Fe)O magnesiowüstite marks the beginning of the lower mantle. This transition is the likely cause of the 670 km seismic discontinuity (Ringwood and Major, 1967). At depths below 670 km, (Mg,Fe)SiO3 perovskite becomes the dominant mantle constituent, accounting for approximately 80 volume percent of the Earth’s lower mantle. Our understanding of the structure and temperature of the lower mantle is mainly based mainly on the comparison of seismic wave velocities with model calculations for wave velocities of mineral assemblages at lower mantle conditions. The reliability of such a comparison depends on the accuracy of experimental and theoretical estimates for mineral elastic properties at the conditions of the deep mantle. A complete knowledge of the elastic properties of (Mg,Fe)SiO3 perovskite over the range of pressure, temperature and composition likely encompassed by the lower mantle is, therefore, a prerequisite for constraining the physical and chemical properties of the lower mantle. Because the lower mantle comprises the bulk of the silicate Earth, knowledge of its composition and temperature is vital to our understanding of the Earth’s composition as a whole and mantle dynamics in general. Perovskite-type compounds have ABX3 stoichiometry and are characterised by a network of BX6 corner-sharing octahedra. In the ideal perovskite structure the octahedral framework forms a cubic array and the A cations occupy the large 12-fold coordinated site. 67 Chapter 3: Intoduction This structure is cubic and has the space group mPm3 . The majority of perovskite structures, however, including (Mg,Fe)SiO3 perovskite, are distorted derivatives of the cubic structure. The most common type of distortion, as in the case of (Mg,Fe)SiO3 perovskite, arises when the size of the A-cation is too small for the 12-fold site. To accommodate such cations the octahedra tilt about the pseudocubic axes, in this way the AX bond-lengths are no longer all equal with a consequent change of the A-site coordination and reduction in symmetry from the cubic arystotype. Octahedral tilting in perovskite has been discussed in details by several authors (see for example Glazer, 1972, 1975; Megaw, 1973; Woodward 1997; Howard and Stokes, 1998). (Mg,Fe)SiO3 perovskite has an orthorhombic symmetry (space group Pbnm) due to the tilting of the BO6 framework which brings some oxygens closer to the A cations, resulting in a lowering of coordination of the A cations from [XII] to [VIII] fold (Fig. 3.1). The A site of (Mg,Fe)SiO3 perovskite is mainly occupied by divalent Mg and Fe cations and the B site by Si. Divalent Fe substitutes for Mg up to approximately 25 % into the A site at conditions compatible with the top of the Earth’s lower mantle, causing a slight expansion of the structure and a decrease in distortion (Ross and Hazen, 1989; Ross and Hazen, 1990; Hemley and Cohen, 1992). B x 6 octahedra A cation a b c Figure. 3.1: Distorted orthorhombic MgSiO3 perovskite structure, showing corner shared BX6 octahedra with dodecahedral A cation site. 68 Chapter 3: Intoduction Octahedral tilt angles can be estimated using unit cell dimensions. For orthorhombic Pbnm perovskites, with no distortion of the BX6 octahedra, three tilt angles can be defined (following the notation of Zhao et al., 1993a,b). These angles, θ, φ, and Φ represent rotations of the octahedra about the pseudo-cubic axes [110], [001] and [111] respectively and can be calculated according to the following equations: ba/cos = θ (1) ca/2cos = φ (2) bca /2cos 2 =Φ (3) Although the unit cell dimension method normally underestimates the tilt angles, as real octahedra are typically slightly distorted, it is still very useful when the atomic coordinates of orthorhombic perovskites are not known. Trivalent cations also substitute into (Mg,Fe)SiO3 perovskite although the substitution mechanism may vary depending on the cation and on pressure and temperature. There is evidence, for example, that Al enters the structure through a coupled substitution onto both A and B sites (Stebbins et al., 2001). Other workers, however, have raised the possibility that Al may replace Si on the octahedral B site with charge balance provided by the creation of an oxygen vacancy (Navrosky, 1999). Certainly at pressures >30 GPa a coupled substitution seems to occur with up to 25 % substitution of an AlAlO3 component being possible (Walter et a., 2004). Al substitution causes the octahedral distortion to increase (Hemley and Cohen, 1993). It was recently established experimentally that a significant portion of Fe in Albearing (Mg,Fe)SiO3 perovskite is in the ferric (Fe3+) state (McCammon, 1997). Support that this may also be the case in the lower mantle comes from the analyses of high ferric Fe contents in mineral inclusions trapped in diamonds that appear to have originally had the perovskite structure (Harte et al., 1999). Experimental studies have also shown that the Fe3+/∑Fe ratio in magnesium silicate perovskite is strongly correlated with Al content (Wood and Rubie, 1996; McCammon, 1997; Frost and Langenhorst, 2002; Lauterbach et al., 2000; McCammon et al., 2004). Such a strong coupling would imply that a coupled substitution takes place. Richmond and Brodholt, (1998) performed computer simulations to examine the energy associated with various trivalent cation substitution mechanisms i.e., 69 Chapter 3: Intoduction Si4+[VI] + Mg2+[VIII]↔Al3+[VI] + Fe3+[VIII] (4) 2Si4+[VI] + O2-↔2Fe3+[VI] + V0 (5) 2Si4+[VI] + O2-↔2Al3+[VI] + V0 (6) These authors reported that the coupled substitution mechanism (4), where Al substitutions onto the Si site and Fe3+ onto the Mg site, is more energetically favourable than the oxygen vacancy mechanisms (5) and (6). The incorporation of Al and the resulting stabilization of Fe3+ in the perovskite structure can potentially affect its elastic properties (Navrotsky et al., 2003). A number of observations of seismic anomalies in the lower mantle have been attributed to chemical heterogeneity (Trampert, 2004) and there are a number of possibilities as to how Al and Fe concentrations could vary in the lower mantle. In order to understanding the origin of seismic anomalies in the lower mantle a complete understanding of how cation substitutions influence the elastic properties of (Mg,Fe,)SiO3 perovskite is required. Values for the bulk modulus of end member MgSiO3 perovskite determined using static compression and dynamic techniques such as Brillouin scattering, cover a range of 250-265 GPa with K' normally fixed at 4 (Andrault et al., 2001; Daniel et al., 2001; Walter et al., 2004; Yagi et al., 2004; Walter et al., 2006). Studies of (Mg,Fe,)SiO3 perovskite indicate values in a similar range (Knittle and Jeanloz, 1987; Mao et al., 1991).Studies on the effect of Al substitution on the elastic propoerties of MgSiO3 perovskite, however, are far more inconclusive. A number of studies have shown that Al substitution causes a decrease in the bulk modulus of perovskite (Zhang and Weidner, 1999; Kubo et al., 2000; Daniel et al., 2001; Yagi et al., 2004). Some studies indicate an increase in the bulk modulus (Andrault et al. 2001; Ono et al., 2004), while several others indicate that there is no change in the bulk modulus with Al incorporation (Yagi et al., 2004; Jackson et al., 2004; Li et al., 2005). This discrepancy observed in elastic property behaviour might be attributed to different substitution mechanisms operating over different pressure or Al concentration ranges. On the other hand it may result from partial amorphisation following sample synthesis or from the use of different measurement techniques. Aside from the discrepancies in these previous studies it is likely that in the Earth the presence of Fe3+ together with Al will affect quite differently the behaviour of perovskite in comparison 70 Chapter 3: Experimental details with Fe-free samples. To date there are no studies that have specifically examined the effect of varying perovskite Fe3+ content and bulk iron concentration on the elastic properties of Al-bearing (Mg,Fe,)SiO3 perovskite. In this study single-crystal X-ray diffraction experiments have been performed at room temperature in a diamond anvil cell (DAC) to determine the equation of state parameters of well-characterized samples of Fe and Al-bearing perovskite, with different Fe3+ contents. 3.2 Experimental details 3.2.1 Starting materials Three starting compositions were employed for the synthesis of perovskite single crystals in this study (Table 3.1). The alumina contents of these compositions were fixed to a value of 5 wt %, considered to be typical for perovskite in the lower mantle, while the bulk Fe concentration was varied. Table 3.1 The glass starting compositions in wt % oxide and cation proportion used for Fe and Albearing perovskite synthesis. Cation proportions are calculated based on 3 oxygens per formula unit. No. SiO2 MgO FeO Al2O3 Total Si Mg Fe Al Total X Fe 1 54.58 36.64 3.63 5.15 100 0.929 0.935 0.052 0.103 2.019 0.05 2 52.96 35.13 6.76 5.17 100.02 0.916 0.911 0.097 0.105 2.031 0.1 3 52.8 32.07 10.09 5.1 100.06 0.925 0.843 0.147 0.1052 2.021 0.15 Starting powders of high-purity SiO2, MgO, Al2O3 and Fe2O3 were ground together in the proportions indicated in Table 3.1, then placed in a Pt crucible and fused in air at 1600°C for 15 minutes. The crucible was than rapidly quenched in water to produce a silicate glass. The glass was ground to a powder, then pelletised and reduced in a CO2-H2 gas-mixing furnace maintained at an fO2 of approximately 2 log units above the ironwüstite buffer at 650°C for 24 hours. The low temperatures prevented the glass from crystallizing. Compositional characterization of the glasses was performed using an electron microprobe. The analysis conditions were the same as listed in Chapter 2 section 71 Chapter 3: Experimental details 2.3. Mössbauer analysis on the glass powders confirmed the complete reduction of Fe2O3 to FeO. 3.2.2 Multianvil synthesis experiments The glasses were transformed to perovskite using a multi-anvil press at 25 GPa and 18002000°C. Multi-anvil experiments were performed using a pressure medium of Cr2O3 doped MgO octahedra of 8 mm edge length in combination with tungsten carbide cubes with 3 mm truncation edge lengths. The pressure assembly employed in this study is as shown in Fig. 3.2. The reduced glass powders were packed into Re or Au gold foil capsules of 2 mm length and 1 mm diameter. The heating duration of the experiments was varied from an hour to ten minutes and experiments were quenched by cutting the power supply. No thermocouple was employed in the experiments but instead a previous calibration of temperature versus power was used to provide an estimate of the temperature. It has been found that in many experiments where a thermocouple was employed, it has led to leakage of partial melt from the capsule and failure of the experiment due to damage caused by compaction of the capsule by the hard Al2O3 thermocouple sleeve. The presence of small degree partial melts is key to the growth of large single crystals. In experiment H2369, melting occurred due to the high temperatures and high Fe content, while in run H2438, H2O was added to flux melting (See Appendix B Table B.1 for run details). 72 Chapter 3: Experimental details MgO + 5 wt% Cr O octahedra 23 ZrO sleeve 2 LaCrO furnace 3 MgO sleeve Sample in a metal foil capsule MgO rod Figure 3.2: The 8/3 multianvil pressure assembly used for multianvil synthesis experiments (not to scale) as described in the text. For the crystals of the compositions with XFe = 0.10 (Crystal 1, Run no. H2369) and 0.15 (Crystal 2, Run no. H2438) we recovered single crystals of size up to a maximum of 150 microns in length on decompression of the experiment as shown in Fig. 3.3 (A) and (B). As H2O was used as flux in the synthesis of Crystal 1, FTIR (Fourier transform infra red) spectroscopic measurements were carried out on these perovskite crystals to check for the presence of water. No evidence for the incorporation of OHin the perovskite sample was detected, in accordance with previous measurements (BolfanCasanova et al., 2000). 73 Chapter 3: DAC experiments there is little evidence, would produce an oxygen excess composition due to Fe3+ or Al3+ substitution for Mg2+ or Fe2+. As can be seen in Fig. 3.8, crystals 2 and 3 plot along the charge coupled substitution mechanism join. Crystals 1 and 4, however, which have the lowest trivalent cation contents, plot between the charge coupled and the oxygen vacancy join. Oxygen vacancies are also indicated by the formulas for these perovskite crystals that contain less than 3 oxygens when normalized to 2 cations. 0.25 0.50 0.75 1.00 0.25 0.50 0.75 1.00 0.25 0.50 0.75 1.00 mol% SiO2 MgO+FeO Al2O3+Fe2O3 Oxygen excess join Charge coupled join Oxygen vacancy join Crystal 1 Crystal 2 Crystal 3 Crystal4 mol% mol% Figure. 3.8: A ternary concentration plot with the axes MgO + FeO, Al2O3+Fe2O3 and SiO2 showing the compositions of the (Fe,Al) MgSiO3crystals used in the present study in relation to different possible substitution mechanisms. 3.3 Diamond anvil cell compression experiments 3.3.1 Basics of diamond anvil cell single crystal compression technique In this study, a BGI-design diamond anvil cell (Allan et al., 1996) has been employed to perform static compression experiments. For the high-pressure experiments a single crystal is placed in a hole drilled through a hardened metal foil (gasket), which is indented 80 Chapter 3: DAC experiments between two opposed diamond anvils with flat parallel faces. This produces a pressure chamber as shown in Fig.3.9. A pressure calibrant is placed along side the sample for pressure determination within the chamber and the free volume in the pressure chamber is filled with a pressure-transmitting fluid for maintaining hydrostatic conditions. Pressure is applied to the sample by mechanically pressing the diamonds together and single crystal diffraction lines are recorded using a 4-circle diffractometer. Details of the technique can be found in Miletich et al., (2000) and Miletich, (2005). Figure. 3.9: A schematic sectional view of diamond anvil cell as is used for our compression study. 3.3.2 Compression experiments: Single crystals with well-defined habits were selected by careful observation under an optical microscope. Most crystals were too opaque to base this selection on extinction and transparency. The final selection of the crystals was performed by checking the reflection intensity and peak profiles using a Huber single crystal diffractometer. Single crystals with dimensions of approximately 120 µm x 80µm x 20 µm (Crystal 1), 130 µm x 100 µm x 50 µm (Crystal 2), and 120 µm x 90µm x 30 µm 81 Chapter 3: DAC experiments (Crystal 3) were loaded into a diamond anvil cell sample chamber drilled into a steel (T301) plate gasket pre-indented down to 90 µm depth. Diamonds with 600 µm culets were employed and the sample chamber was 300 µm in diameter. Quartz was used as an internal pressure calibrant and a 4:1 mixture of methanol:ethanol was used as the pressure transmitting medium for our study. The unit cell parameters were determined at ambient temperature at various pressures up to 9.16 GPa for Crystal 1, 7.10 GPa for Crystal 2 and 6.97 GPa for Crystal 3 on a Huber four-circle diffractometer (non-monochromatised MoKα radiation) using the 8-position centring procedure according to King and Finger (1979) and Angel et al., (2000). The maximum pressure reached during each experiment was determined by failure of the gasket and consequent loss of the perovskite single-crystals, therefore it was not possible to collect data during decompression. The centring procedure and vector least square refinement of the unit cell constants were performed by SINGLE04 software according to the protocol of Ralph and Finger, (1982) and Angel et al., (2000). The unit cell data at different pressures are tabulated in Table 3.3. Table 3.3. Unit-cell lattice parameters for (Fe, Al)- MgSiO3 single crystals. Standard deviations are in parentheses in terms of least units cited. P (GPa) a(Å) b(Å) c(Å) V((Å) Crystal 1 0.00010(1) 4.78638(19) 4.94261(16) 6.9188(4) 163.680(12) 0.295(6) 4.7844(2) 4.94052(19) 6.9161(5) 163.479(15) 1.063(6) 4.77899(18) 4.93630(15) 6.9085(4) 162.976(12) 1.982(7) 4.7732(2) 4.9310(2) 6.8995(6) 162.392(17) 2.818(9) 4.7674(2) 4.9265(3) 6.8916(6) 161.860(18) 3.991(11) 4.7595(2) 4.9198(3) 6.8800(8) 161.10(2) 4.611(11) 4.7558(2) 4.9165(2) 6.8748(6) 160.747(19) 6.501(10) 4.7441(2) 4.9066(2) 6.8574(7) 159.62(2) 7.944(10) 4.73509(18) 4.89956(16) 6.8453(5) 158.810(14) 8.785(18) 4.7306(4) 4.8953(4) 6.8376(10) 158.34(3) 9.16(2) 4.7284(4) 4.8934(3) 6.8342(9) 158.13(3) Crystal 2 0.00010(1) 4.79239(14) 4.95152(11) 6.9343(4) 164.548(11) 0.350(5) 4.79005(15) 4.94953(11) 6.9311(5) 164.327(13) 82 Chapter 3: DAC experiments 0.681(5) 4.78789(16) 4.9480(2) 6.9278(6) 164.123(16) 1.204(4) 4.78423(18) 4.94443(15) 6.9222(6) 163.747(15) 1.685(4) 4.78099(14) 4.94197(15) 6.9179(6) 163.455(14) 2.455(5) 4.77571(15) 4.93737(15) 6.9100(5) 162.933(13) 3.491(6) 4.76913(11) 4.93217(14) 6.9005(5) 162.315(12) 4.728(7) 4.76121(16) 4.92566(17) 6.8893(6) 161.568(15) 5.420(6) 4.75672(9) 4.92167(9) 6.8825(4) 161.127(9) 6.429(9) 4.75072(12) 4.91680(10) 6.8734(4) 160.552(10) 6.910(8) 4.74773(9) 4.91416(9) 6.8695(3) 160.272(9) 7.101(9) 4.74647(13) 4.91332(12) 6.8682(4) 160.173(12) Crystal 3 In air 4.79977(10) 4.97947(9) 6.98054(11) 166.837(5) 0.00010(1) 4.79978(10) 4.97961(11) 6.9799(3) 166.826(10) 0.396(4) 4.79701(11) 4.97753(13) 6.9756(4) 166.558(11) 0.513(5) 4.79639(12) 4.97699(14) 6.9745(4) 166.492(11) 0.745(5) 4.79465(14) 4.97563(16) 6.9718(5) 166.321(14) 1.176(5) 4.79181(13) 4.97311(16) 6.9671(4) 166.029(13) 1.510(5) 4.78929(17) 4.97096(18) 6.9634(6) 165.781(17) 1.943(5) 4.78665(14) 4.96852(16) 6.9591(5) 165.504(14) 2.334(5) 4.78420(13) 4.96607(14) 6.9545(4) 165.231(12) 2.756(7) 4.78122(13) 4.96370(16) 6.9502(4) 164.950(12) 3.302(7) 4.77772(13) 4.96071(13) 6.9447(4) 164.596(11) 4.285(7) 4.77151(11) 4.95556(12) 6.9346(4) 163.973(10) 4.540(8) 4.76976(12) 4.95431(12) 6.9323(4) 163.817(10) 4.930(7) 4.76740(12) 4.95202(13) 6.9283(4) 163.566(11) 5.463(7) 4.76404(11) 4.94915(13) 6.9231(4) 163.232(11) 5.969(9) 4.76110(18) 4.9465(2) 6.9188(6) 162.942(17) 6.423(8) 4.75831(12) 4.94434(13) 6.9145(4) 162.674(11) 6.729(8) 4.75623(13) 4.94286(14) 6.9115(4) 162.486(12) 6.968(8) 4.7550(2) 4.9420(3) 6.9089(8) 162.36(2) 83 Chapter 3: Equation of state results 3.4 Equation of state results 3.4.1. Theoretical background The incompressibility of a material is expressed in terms of its bulk modulus K, which is defined as K = -V(δP/δV). The bulk modulus is also a function of pressure and can be described by the pressure derivative K' = δK/δP and potentially also K'' = δ2K/δP2. Static compression measurements are performed at constant temperature and are therefore described using the isothermal bulk modulus K0. Compression in the Earth, however, is not isothermal but adiabatic and is therefore described using the adiabatic bulk modulus (KS). The relationship between the isothermal and adiabatic bulk moduli is KS = K0 (1+αγT) (7) where γ is the Gruneisen parameter and α is the thermal expansion coefficient. The relationship between volume and pressure is described using an equation of state (EoS). A number of EoS formulations exist (Angel, 2000; Duffy and Wang, 2000) but the most commonly used, particularly to describe experimental data on the compression of minerals, is the Birch–Murnaghan equation of state (Birch, 1947). This is based on the assumption that the strain energy of a solid undergoing compression can be expressed as a Taylor expansion of the Eulerian strain fE = [(V0/V)2/3-1]/2. Expansion to the fourth-order in fE gives the Birch-Murnaghan EoS: P = 3K0fE (1+2fE) 5/2{1+3/2(K' –4)fE +3/2{K0K'' + (K'- 4)(K' –3)+35/9} fE 2} (8) However, for most type of materials this EoS is truncated at the second order in strain, which requires K' to be fixed to 4 or at the third-order in strain which requires the coefficient of fE2 to be zero with an implied value of K'' = -1/K0(3K')(4K')+35/9} (Anderson, 1995). 84 Chapter 3: Equation of state results 3.4.2 Unit cell lattice parameter variations with pressure The changes in unit cell volume as a function of pressure for all three crystals are shown in Fig.3.10, where the relative volume (V/V0) is the volume normalized to the ambient pressure volume. The maximum pressure of each experiment is constrained by the point where the gasket starts to fail and diffraction lines are broadened. This occurred at slightly different pressures in each of the three experiments with crystal 1 achieving the highest pressure of just above 9 GPa. For comparison, data for pure MgSiO3 perovskite (Vanpeteghem et al., 2006) and MgSiO3 perovskite with 5-mol% Al2O3 (Zhang and Weidner, 1999) are also plotted in Fig. 3.10. 0246810 0.960 0.965 0.970 0.975 0.980 0.985 0.990 0.995 1.000 1.005 Crystal 1 XFe = 0.07 Crystal 2 XFe = 0.13 Crystal 3 XFe = 0.24 MgSiO3 MgSiO35mol% Al203 Relative volume V/V0 Pressure (GPa) Figure. 3.10:Unit cell volume variation of (Fe,Al)-MgSiO3 single crystals as a function of pressure for the 3 crystals of present study. The solid curve represents the volume variation of pure MgSiO3 of Vanpeteghem et al., (2006) and the dotted curve has been calculated using the EoS parameters reported for a MgSiO3 perovskite with 5-mol % Al2O3 by Zhang and Weidner, (1999). Although the differences between the P-V data collected in this study and those of MgSiO3 perovskite (Vanpeteghem et al., 2006) are very small, there is a slight increase in compressibility with increasing Fe content. In addition, the Fe and Al-bearing perovskites 85 Chapter 3: Equation of state results synthesised in this study are less compressible than the Fe-free Al-bearing perovskite studied by Zhang and Weidner, (1999). The axial compressibilities a/a0, b/b0, c/c0 of all the three crystals as a function of pressure are shown in Fig. 3.11. The axial compression is anisotropic with the b-axis being the least compressible. For Crystal 1 (XFe = 0.07) and Crystal 2 (XFe = 0.13) the axial compressibilities of the a and c axes are virtually identical and are very close to those of pure MgSiO3 perovskite (Vanpeteghem et al., 2006). For Crystal 3 (XFe = 0.24) the difference in compressibility between the a and c axes is larger, with a-axis being less compressible than c-axis. As a consequence, the axial ratio c/a remains constant for Crystal 1 at 1.445 and for Crystal 2 at 1.446, whereas it decreases with pressure for Crystal 3. The b/a axial ratio instead increases with pressure for all three crystals. The axial compressibilities reported for the Fe-free Al bearing perovskites by Zhang and Weidner (1999) show the a-axis to be slightly more compressible than the c-axis. This suggests that there is a different mechanism by which Al is substituted into the perovskite structure in the absence or in the presence of Fe. 86 Chapter 3: Equation of state results 01234567891011 0.984 0.986 0.988 0.990 0.992 0.994 0.996 0.998 1.000 1.002 Crystal 3 a / a 0 b / b 0 c / c 0 Relativeaxis Pressure in GPa 0246810 0.986 0.988 0.990 0.992 0.994 0.996 0.998 1.000 1.002 Crystal 1 a / a 0 b / b 0 c / c 0 Relative axis Pressure in GPa 01234567891011 0.986 0.988 0.990 0.992 0.994 0.996 0.998 1.000 1.002 Crystal 2 a / a 0 b / b 0 c / c 0 Relative axis Pressure in GPa Figure 3.11. Axial compressibilities of the (Fe,Al)-MgSiO3 perovskite single crystals. The solid curves are the 3rd-order Birch Murnaghan EoS fits to the data: (a) Crystal 1, XFe = 0.07 (b) Crystal 2, XFe = 0.13 and (c) Crystal 3, XFe = 0.24. 3.4.3 FE-fE plots and EoS parameters A normalised stress, defined as FE = P/3fE (1+2 fE)5/2 has been plotted versus the Eulerian strain fE for each crystal (Fig. 3.12) in order to have a visual assessment of the order of the Birch-Murnaghan EoS required to fit the compression data (Angel, 2000). 87 Chapter 3: Equation of state results 0.000 0.002 0.004 0.006 0.008 0.010 0.012 230 240 250 260 Normalised pressure FE GPa Eulerian strain fE Crystal 1 BM III fit BM II fit 0.000 0.002 0.004 0.006 0.008 0.010 240 255 270 285 Crystal 2 BM III fit BM II fit Normalised pressure FE GPa Eulerian strain fE 0.000 0.002 0.004 0.006 0.008 0.010 220 230 240 250 260 Crystal 3 in DAC BM III fit BM II fit Normalised pressure FE GPa Eulerian strain f Figure. 3.12 FE-fE plots based on the Birch-Murnaghan EoS for (Fe,Al)-MgSiO3 perovskite single crystals. The values of V0 used to calculate the finite strain are those measured at 1 bar. The open circles shown in the FE-fE plot of Crystal 3, are the finite strain values calculated with volume of the crystal determined in air (Equations for the weighted fits are mentioned in the text). 88 Chapter 3: Equation of state results Weighted linear fits to data points for the three crystals indicate a positive slope, which implies that K' is greater than 4. For comparison, weighted horizontal linear fits (implying K' = 4) also are shown in Fig. 3.12, although in this case we observe an increase of the chi-square. In the case of Crystal 3 the FE-fE data do not define a straight line, suggesting that a fourth-order truncation of the Birch-Murnaghan EoS might be considered. However, it has been shown (Angel, 2000) that such curvature for small strains, i.e., for data collected at low pressures, is often due to a wrong value of V0 being used in the calculation of the Eulerian strain. If the value of the volume measured with crystal 3 glued on a glass fibre (i.e., in air, Table 3.3), instead of that measured with the same crystal in the DAC is used to calculate the F-f plot, linear behaviour is obtained (Fig. 3.12, Crystal 3 open circles). Moreover, the intercept on the F axis is equal to the bulk modulus, K0, value and can hence be used to assess the goodness of the EoS fit procedure. F-f plots have been calculated also for the individual crystal axes and are shown in Fig. 3.13 indicating positive slope. Since the FE-fE plots (Fig. 3.12) suggest that K' is larger than 4 for all three crystals, the P-V data have been fitted with a 3rd order Birch–Murnaghan EoS using the Eosfit52 program (Angel, 2000). However, for comparison with data present in the literature, fitted practically uniquely with K' fixed to the value of 4, a 2nd order Birch–Murnaghan EoS has also been used. A linearised Birch-Murnaghan EoS (implemented in the Eosfit52 software) in which the cube of a unit-cell axis is used instead of the volume has been used to fit the unit cell a, b and c parameters of the three crystals. The bulk moduli so obtained are 00 31 β = K, where β 0 is the axial compressibility. The resulting equation of state parameters are reported in Table 3.4. The values obtained are in good agreement with those obtained from the linear fits of the FE-fE plots. 89 Chapter 3: Discussion 01234567891011 18.0 18.5 19.0 19.5 20.0 20.5 21.0 Crystal 1 Crystal 2 Crystal 3 MgSiO3 Distortion angle Φ(°) Pressure in GPa Figure 3.17: Distortion angle Φ for (Fe,Al)-MgSiO3 perovskites as a function of composition and pressure. The orthorhombic distortion is found to increase with increasing pressure for all compositions. Pure MgSiO3 data are also plotted for a comparison from Vanpeteghem et al., (2006). 3.5.Discussion 3.5.1 The Elasticity of the Earth’s lower mantle Assuming a pyrolite bulk mantle composition, the Earth’s lower mantle should be composed of approximately 80-volume % magnesium silicate perovskite coexisting with magnesiowüstite and calcium silicate perovskite. Comparison of seismically-inferred properties for the lower mantle with model properties based on measured bulk and shear moduli for such a mantle assemblage provides the best test for the composition and temperature of the lower mantle. The elasticity data determined from single crystal compression experiments have been combined with existing thermoelastic data in order to calculate the variation in density and bulk modulus over the depth range of 670–2571 km, which covers the entire lower mantle excluding the D’’ low velocity zone. A number of studies have reported thermoelastic properties for silicate perovskite from experiments conducted at simultaneous high pressure and temperature (Mao et al., 1991; Yeganeh–Haeri, 1991; 96 Chapter 3: Discussion Wang et al., 1994; Utsumi et .al, 1995; Funamori et al., 1996; Fiquet et al., 1998; Fiquet et al., 2000). For these calculations we have used the high temperature properties from the data set of Funamori et al., (1996) [ K0,T = 261 GPa, K'0,T = 4, (dK/dT)P =-0.028 Gpa K-1 , α0 = 1.98*10-5 K -1, α1 = 0.82*10-8 K-2 and α2 = -0.47(K)] which were determined for MgSiO3 perovskite. The room pressure volume (V0) at high temperature was calculated using thermal expansion coefficients and the equation ∫ = T TTT dTVV 298 0,0,,0 exp α (9) In view of the lack of high-temperature elastic property data for Al-bearing perovskites and the limited data for Fe-bearing perovskites, we have no option but to assume that they are the same as for the MgSiO3 end member. Mao et al., (1991) proposed that thermal expansivitites of pure and Fe-bearing perovskite were similar, but that the temperature dependence of the bulk modulus of Fe-bearing perovskite was greater, i.e., (dK/dT)P = -6.3*10-2 Gpa K-1. In order to examine this proposal we have also made calculations assuming the value of (dK/dT)P reported by Mao et al., (1991). Almost all previous studies on elastic properties of MgSiO3 perovskite have fitted the experimental data assuming a second order Birch-Murnaghan equation of state, i.e., with K' fixed at 4. Our data, however, provide strong evidence that, at higher Al and Fe concentrations a third-order Birch-Murnaghan equation of state is required with a K' higher than 4. To obtain an overview of how fixing K' to 4 and K' > 4 can effect calculated lower mantle properties; calculations have been performed using both 2nd and 3rd order fits to our data set. All the calculations were performed along the adiabatic geotherm of Brown and Shankland, (1981) fixed at 1873 K for a depth of 670 km. Densities calculated along an adiabatic temperature gradient for the three crystals are shown as a function of pressure in the mantle in Fig. 3.18. When K' is fixed at 4 and (dk/dT)P = -0.028 GPa K-1 (Funamori et al., 1996), there is a clear increase in density with increasing Fe and Al concentration at lower mantle conditions. Densities for Crystals 1 and 2 are only slightly larger than those calculated for MgSiO3 perovskite throughout most of the lower mantle. However, the slightly lower bulk moduli of Crystals 1 and 2 means that their densities become smaller than MgSiO3 perovskite close to the core mantle boundary. The difference in densities between Crystals 1 and 2 also decreases throughout the lower mantle due to the minor differences in their bulk moduli. The density of Crystal 3, which 97 Chapter 3: Discussion has the highest Fe content remains significantly larger than the other two crystals throughout the lower mantle even though it has a lower bulk modulus. When (dK/dT)P = - 0.063 GPa/K, as proposed by Mao et al., (1991), is employed for this calculation, the effect of Fe and Al on densities in the lower mantle becomes much stronger as a result of the bulk moduli being lower. The Preliminary Reference Earth Model (PREM), (Dziewonski and Anderson, 1981), density curve is quite consistent with a monomineralic perovskite lower mantle, but a perfect match would only be obtained if the lower mantle increased in Fe content towards the base. Based on similar comparisons, previous studies have argued that the lower mantle may have a higher Si/Mg ratio than the upper mantle and therefore be composed mostly of perovskite (Anderson et al., 1995). Large payoffs exist, however, between the value of (dK/dT)P used and the proportion of coexisting magnesiowüstite in the modelled assemblage (Fiquet et al., 1998). As discussed previously, the compression data reported in this chapter are actually consistent with values of K' >4 for Fe and Al bearing perovskite, and K' was also found to increase with increasing Fe content. In the lower panel of Fig. 3.18 density calculations have been performed using the results of the 3rd order Birch-Murnaghan equation of state fits. The result is a much smaller change in density in the lower mantle, compared to room pressure, as a result of Fe and Al substitution, due to the higher values of K' which make Fe-rich perovskites less compressible at lower mantle conditions. Very large changes in Fe concentration in the lower portion of the lower mantle would be required to match the changing slope of the PREM density curve when K' > 4. 98 Chapter 3: Discussion 20 40 60 80 100 120 4.2 4.4 4.6 4.8 5.0 5.2 5.4 5.6 5.8 6.0 6.2 WithK'>4 With Funamori et al. 1996 data Crystal 1 Crystal 2 Crystal 3 MgSiO3 PREM With Mao et al. 1991 data Crystal 1 Crystal 2 Crystal 3 Density (gm/cm 3 ) Pressure (GPa) 20 40 60 80 100 120 4.2 4.4 4.6 4.8 5.0 5.2 5.4 5.6 5.8 6.0 6.2 With K' = 4 With Funamori et al.1996 data Crystal 1 Crystal 2 Crystal 3 MgSiO 3 PREM With Mao et al. 1991 data Crystal 1 Crystal 2 Crystal 3 Density (gm/cm 3 ) Pressure (GPa) Figure 3.18: Density profiles calculated for the three crystals with K' = 4 and > 4 as determined from the compression data of this study. Two (dk/dT)P values were used for comparison as stated in the text. PREM data of Dziewonski and Anderson, (1981) and pure MgSiO3 density calculated based on compression data of Vanpeteghem et al., (2006), are also plotted for comparison. We have also calculated KT, the bulk modulus at pressure and temperature, for the three crystals of this study, using the equation: K T,0= K 298,0 + (dKT,0/dT)P(T-298) (10) 99 Chapter 3: Discussion where, KT,0 is the bulk modulus at temperature, K298,0 is the bulk modulus at ambient conditions and (dKT,0/dT)P is the temperature derivative of the bulk modulus, taken from either Funamori et al., (1996) or Mao et al., (1991), and the equation K T= K T,0 + (dKT,0/dP)TP (11) where (dKT,0/dP)T is the pressure derivative of the bulk modulus. An estimate for the value of KT for the lower mantle can be obtained from PREM KS (the adiabatic bulk modulus) estimates using the equation: KT=KS/(1+αγT) (12) where, α is the thermal expansivity and γ is the gruneisen parameter, values of which are tabulated for the lower mantle in Brown and Shankland, (1981). Regardless of the value of (dKT,0/dT)P employed, KT curves calculated for perovskite samples assuming K' > 4 display a much greater divergence as a function of composition than when K' is assumed to be 4 (Fig. 3.19). This is the opposite behaviour from that observed for densities, where the greatest effect of Fe content occurs for K' = 4. Very large changes in Fe content would be required, on the other hand, to influence KT in the lower mantle if K' = 4. In addition, the slope of the PREM value of KT is shallower than those calculated for the three crystals of this study if K' > 4, which means that Fe contents would have to decrease in the lower portion of the lower mantle to be consistent with the PREM slope. Again this is the opposite conclusion to that obtained based on density. KT determinations with K' > 4 are much higher at the base of the mantle compared with PREM and provide a poor fit to the observations compared to when K' = 4. The addition of magnesiowüstite to the model assemblage would reduce KT but not by more than 10 % (Fiquet et al., 1998). Under these circumstances, values of K' = 4 and (dk/dT)P = -0.028 GPa K-1 give the best agreement with the PREM value of KT for lower mantle. 100 Chapter 3: Discussion 40 60 80 100 120 200 300 400 500 600 700 800 900 1000 (dK/dT) P = -0.028GPaK -1 With K' = 4 Crystal 1 Crystal 2 Crystal 3 With K' > 4 Crystal 1 Crystal 2 Crystal 3 MgSiO 3 (K' = 4) PREM K T (P) Pressure in GPa 40 60 80 100 120 200 300 400 500 600 700 800 900 1000 (dK/dT) P = -0.063GPaK -1 With K'= 4 Crystal 1 Crystal 2 Crystal 3 With K' > 4 Crystal 1 Crystal 2 Crystal 3 MgSiO 3 K' = 4 PREM K T (P) Pressure in GPa Figure 3.19: Comparison of bulk modulus KT(P) profiles along with PREM lower mantle model (Dziewonski and Anderson, 1981) along the temperature profile of Brown and Shankland, (1981). Bulk modulus of MgSiO3 is also plotted for a comparison based on data of Vanpeteghem et al., (2006). KT profile for the three crystals calculated using (dk/dT)P of –0.028 GPa K-1 and -0.063 GPa K-1 values as reported by Funamori et al., (1996) for a MgSiO3 composition and Mao et al., (1991) for a Fe-bearing MgSiO3 composition both for a K' of 4 and K' > 4 as obtained in this study by 3rd order Birch-Murnaghan fit. 101 Chapter 3: Discussion 3.5.2 The effect of pressure on perovskite substitution. OAlFe VIVIII 3 33 ++ Experimental observations indicate (Frost et al., 2004) that even at the lowest plausible oxygen fugacity for the lower mantle, (Fe,Mg)(Al,Si)O3 perovskite contains a significant Fe3+ content (Fe3+/∑Fe > 0.5). As the upper mantle has a very low Fe3+/∑Fe ratio (<0.03) two possible scenarios for the redox state of the lower mantle can be proposed: either the lower mantle is more oxidized than the upper mantle or at lower mantle conditions Fe2+ is oxidised to Fe3+ by some agent also present in the bulk mantle. Evidence for whole mantle convection contradicts the first possibility, as it would have been impossible to maintain a low Fe3+/∑Fe ratio in the upper mantle if it were mixed through geologic time with a highly oxidised lower mantle reservoir. Although the second possibility is therefore more likely, none of the oxidising agents active in the upper mantle (e.g., CO2, SO2) are abundant enough to produce the required Fe3+ contents. In the absence of such an agent, disproportionation of FeO i.e., 3FeO = Fe+Fe2O3. (13) becomes the only mechanism capable of producing Fe3+ in sufficient abundance. Frost et al., (2004) showed experimentally that about 1 wt % of Fe metal would be required to balance the Fe3+ requirement of perovskite at 25 GPa within a pyrolitic lower mantle bulk composition. The equilibrium between (Fe,Mg)(Al,Si)O3 perovskite and Fe metal can be described by the equation: AlAlO3 + 3FeO = Fe + 2AlFeO3 (14) Perovskite Mw Metal Perovskite where, Mw is magnesiowüstite. Frost et al., (2004) measured high concentrations of the AlFeO3 component in perovskite in equilibrium with metallic Fe and concluded that this equilibrium must be shifted strongly to the right at conditions compatible with the top of 102 Chapter 3: Discussion the lower mantle. By assessing the effect of ( ), ( Fe ) and ( Fe ) substitutions on the volume of (Fe,Mg)(Al,Si)O O AlAl VIVIII 3 33 ++ Fe Si Mg AlAl VI VIII VIVIII + + ++ +↔+ 4 2 33 OSiVIVIII 3 42 ++ MgVIII VIII + +↔2 2 OAlFe VIVIII 3 33 ++ Si Mg Al VI VIII VIVIII + + ++ +↔+ 4 2 33 3 perovskite, it should be possible to determine whether disproportionation of FeO will be favoured at the higher-pressure conditions of the deep lower mantle. The volume and compression results presented in this chapter provide data that help to understand the likely volume effect. 24.4 24.6 24.8 25 25.2 25.4 25.6 0 5 10 15 20 25 30 AlAlO3 (Walter et al., 2004, 2006; Andarult et al., 2001;Kubo et al., 2000) Nishio-Hamane et al., (2005) Present study FeSiO3 (Yagi et al., 1979; Fei et al., 1996; Andarult et al., 2001) Vanpeteghem et al., (2006) molar volume (cm3mol-1) mol% AlAlO3, FeSiO3, FeAlO3 MgSiO3 AlAlO3 Fe3+AlO3 (4) (4) (4) FeSiO3 0 (8) (3) Fe3+AlO3 Figure 3.20. Effect of , , substitutions on the molar volume of magnesium silicate perovskite. OSiFe VIVIII 3 42 ++ O AlAl VIVIII 3 33 ++ OAlFe VIVIII 3 33 ++ Fig. 3.20 shows the change in molar volume of MgSiO3 perovskite as a function of possible substitution mechanisms determined using room pressure volume data from this and previous studies. The addition of and components have similar effects on the molar volume of perovskite with the former having a slightly greater effect. O AlAl VIVIII 3 33 ++ OSiFe VIVIII 3 42 ++ 103 Chapter 3: Discussion Although the perovskite samples synthesised in this study have high Fe3+/ΣFe ratios, all samples contain Fe2+ and therefore do not lie on a compositional join between MgSiO3 and . All single crystals synthesised in this study also contain 4-mol % of , which is indicated by the figure in brackets next to each data point. These data show that in comparison to and substitutions the (Fe,Mg)(Al,Si)O OAlFe VIVIII 3 33 ++ OSiVI 3 4+ FeVIII 2+ OAlVI 3 3+ OSiVI 3 4+ O AlAl VIVIII 3 33 ++ 3 OSiFe VIVIII 3 42 ++ SiFeVIII 2+ 3 perovskites have larger molar volumes. The only sample with a composition on the MgSiO3join is that synthesised by Nishio-Hamane et al., (2005) at approximately 50 GPa in a diamond anvil cell. Although the Fe OAlFe VIVIII 33 ++ OSiVIVIII 3 42 ++ 3+ content was not independently confirmed, it was assumed to be the same as that of the staring composition. The volume of this sample is smaller than the sample synthesised in our study at a similar content and a trend starts to emerge particularly when the volumes and contents of (Fe,Mg)(Al,Si)O FeVIII 3+ FeVIII 2+3 perovskite samples synthesised by Vanpeteghem et al., (2006) are also considered. Increasing the content of (Fe,Mg)(Al,Si)O O VI 3 4+ 3 perovskite samples increases their molar volumes dramatically. In other words, the substitution of into Al-free perovskite has a smaller effect on molar volume than when it substitutes into Al-bearing perovsktie. As perovskite must contain Al in the lower mantle the large effect on the molar volume should make Fe Fe 2+ substitution in perovskite unfavourable with increasing pressure. Using data in Fig. 3.20 and compression data from this study and the literature (Fei, 1996; Walter et al., 2004) the molar volume change of the equilibrium (eq. 14), at pressures of the lower mantle is calculated to be approximately –2 cm3/mol. This implies that disproportionation of FeO should be favoured with increasing pressure and therefore is likely to take place throughout the perovskite-bearing region of the lower mantle. In addition, however, the large effect on the perovskite molar volume of substitution into (Fe,Mg)(Al,Si)O OSiFe VIVIII 3 42 ++ 3 perovskite should drive Fe2+ out of perovskite with increasing pressure and into magnesiowüstite. This should cause perovskite to decrease in total Fe content with pressure but to become more Fe3+ rich. Some support of this can be found in the fact that while in this study it was not possible to synthesise pure Fe3+ bearing perovskite at 25 GPa, as attempted in the synthesis of crystal 104 Chapter 3: Conclusions 3 which contained only Fe3+ in the starting composition, this was possible in the study of Nishio-Hamane et al., (2005) performed at approximately 50 GPa. 3.6 Conclusions The substitution Al and Fe3+ into MgSiO3 perovskite occurs by a coupled substitution mechanism with the possibility that small concentrations of oxygen vacancies are present only at low trivalent cation concentrations. Static compression measurements show that the substitution of Al and Fe into MgSiO3 perovskite increases the compressibility. The axial compression is anisotropic with the b axis being the least compressible, and a and c axes having virtually identical compressibilities. Axial compressibilities of Fe-free Al-bearing perovskites (Zhang and Weidner, 1999) show the a axis to be slightly more compressible than the c axis implying that there is a different mechanism by which Al is substituted into the perovskite structure in the absence or in the presence of Fe. Fitting the data to a 2nd order Birch-Murnaghan EoS (K' = 4) gives values for the bulk modulus that decrease from 247 to 240 GPa as the Fe content increases from 0.12 to 0.22 formula units. These values are lower than the value of 253 GPa reported for MgSiO3 perovskite (Vanpeteghem et al., 2006) based on single crystal determinations. However, fF plot analyses indicate that a 3rd order BirchMurnaghan fit to the data is statistically justified. These fits indicate a larger drop in bulk modulus with Fe and Al substitution from 243 to 234 GPa and an increase in K' from 5 to 6.5. The decrease in bulk modulus likely arises from an increase in polyhedral compressibility. Combining these data with thermoelastic data from the literature indicates that perovskite densities calculated along a lower mantle geotherm are insensitive to Fe-Al substitution if the 3rd order BirchMurnaghan fit to the data is employed. The resulting values of K' from this fit which are >4 provide a poor fit to the estimated bulk modulus of the lower mantle. If on the other hand the 2nd order BirchMurnaghan fit is employed the calculated densities when compared to PREM are consistent with an increase in Fe-Al substitution in the lower portions of the lower mantle. 105 Chapter 4: Starting material synthesis MgO ceramic is inserted from one side of the assembly through the molybdenum electrode, so that the junction is located just above the sample. Synthesis experiments were performed between 17-20 GPa and 1000-1600°C (Table 4.1). The pressure calibration curves of Frost et al., (2004) were adopted for determination of pressure in our experiments. The heating duration varied from one to three hours and samples were quenched by cutting the power supply to the furnace. Cr doped MgO pressure medium ZrO sleeve 2 Re capsule LaCrO furnace 3 MgO sleeve LaCrO disc 3 MgO disc MgO rod MgO sleeve Molybdenum sleeve Thermocouple Al O 23 Copper coil Figure. 4.2: A diagrammatic representation of the box furnace assembly is shown here. The sample was contained in a Re capsule as shown surrounded by a MgO sleeve and a LaCr2O3 furnace. The molybdenum tubes serve as contacts for heating. This set up ensures low thermal gradient in the large sample volume. Table 4.1: Synthesis conditions for calorimetric samples used in this study. Piston cylinder and a multianvil press were used to synthesise them. The unit cell parameters determined by powder Xray diffraction (as discussed later in this section), for garnets are also listed here. Garnets with cubic symmetry have a=b=c and for tetragonal symmetry have a = b ≠ c where a, b, c are crystal axes so, as a and b axes are equal, only, a and c are listed here. Run no. Composition Pressure Pressure Temperature Unit cell parameters (in Å units) assembly GPa °C a-axis c-axis Z493 Py 40 18/8 19 1000 11.476 Z494 Py 20 18/8 19 1600 11.483 Z496 Py 10 18/8 18 1600 11.497 11.457 Z501 Py 30 18/8 17 1600 11.484 Z507 Majorite 18/8 19.5 1750 11.517 11.433 Z508 Py 80 18/11 17.5 1200 11.463 Z525 Py 15 18/8 18 1400 11.482 PC Pyrope 1/2inch 3 1200 11.460 112 Chapter 4: Starting material synthesis Small chips of the recovered samples were embedded in epoxy resin for quantitative chemical analysis using a JEOL JXA-8200 WD/ED electron microprobe operating in wavelength dispersive mode with a point beam at 15 nA current and 15 kV. Standards employed were andradite for Si, spinel for Al and enstatite for Mg (for details on operating conditions see Table 2.3 of Chapter 2). The determined compositions are listed in Table 4.2. Table 4.2: The cation composition of the garnet starting materials as analyzed by electron microprobe calculated based on 12 oxygen per formula unit. Abbreviation Maj. (majorite) and Pyr. (pyrope). Composition Si Mg Al Total Maj. content Pyr. content Pyrope 3.007 2.974 2.007 7.989 0.007 0.993 Pyrope 80 3.224 3.133 1.611 7.970 0.224 0.776 Pyrope 40 3.606 3.548 0.827 7.980 0.605 0.394 Pyrope 30 3.677 3.689 0.638 8.004 0.677 0.323 Pyrope 20 3.755 3.828 0.442 8.024 0.755 0.245 Pyrope 15 3.867 3.783 0.321 7.972 0.867 0.133 Pyrope 10 3.897 3.846 0.240 7.983 0.897 0.103 Majorite 4.025 3.95 0 7.974 1.025 0 The remaining portion of each sample was ground to a powder and a portion of this powder, which is presumed to have been representative of the entire sample, was characterized by X-ray powder diffraction. The samples were mixed with a small amount of Si as an internal standard (NBS standard number 640b). X-ray diffraction was performed using a Philips X’Pert Pro X–ray diffractometer operating in reflection mode; using Co Kα radiation with an wavelength of 1.78892 Å selected using a focused monochromator. The diffraction conditions were set to a step size of 0.2°, a step time of 1000s with scan speed of 0.0002°/sec with a rotating platform rotating at 1 rotation per second. The data were collected over a 2θ range of 20-120°. Cell parameter refinements were performed using the GSAS software package (see Table 4.1). Obtained Cell parameters are plotted in Fig. 4.3. No phases other than cubic or tetragonal garnet were 113 Chapter 4: Calorimetric studies detected. The cubic to tetragonal garnet transition was observed at slightly lower pyrope contents (< 13 %) than in the study of Heinemann et al., (1997). Figure 4.3: Refined cell parameters of the garnet compositions in the majorite–pyrope join after synthesis in multi-anvil and piston cylinder (filled squares). The cell parameters as reported by Heinemann et al., 1997 along the majorite–pyrope join are plotted (solid triangles) for comparison. They reported the stability field of tetragonal garnets to commence from the Maj 80 composition; however our Maj 0.867 still appears to be a cubic garnet. This discrepancy may be attributed to differences in the synthesis conditions. 4.3 Calorimetric measurements 4.3.1 Basic principals A calorimeter measures the change in heat associated with the change of state of a sample. A number of calorimetric methods have been employed in Earth sciences to provide basic thermodynamic data. Thermophysical measurements such as low temperature adiabatic calorimetry or differential scanning calorimetry (Akaogi, 1990; Navrotsky, 1997; Navrotsky, 2004) are used to determine heat capacities as a function of temperature, which are also required to determine the standard entropy of a phase. Thermochemical or reaction calorimetry, on the other hand, provides a measurement of the enthalpy of a reaction, which can be combined with other reaction enthalpies to give the heat of formation of a 114 Chapter 4: Calorimteric studies compound. A widely used type of reaction calorimetry is to measure the heat of dissolution as a sample dissolves in a solvent, either an acid or an oxide melt, to infinite dilution. There are two methods by which this can be accomplished, which deviate with respect to the initial temperature of the sample. In solution calorimetry the initial temperature of the sample is identical to that of the solvent before dissolution. In drop solution calorimetry the sample is initially at room temperature and is dropped into the high-temperature solvent. In this study drop solution calorimetry using an oxide melt was employed. This was deemed more suitable because the recovered metastable high-pressure phases could potentially breakdown during the high temperature equilibration stage (at ~700°C) of solution calorimetry. It should be noted, however, that in previous studies solution calorimenty has been extensively employed to measure solution enthalpies of high-pressure phases (Akaogi et al., 1987; Yusa et al., 1993; Akaogi and Ito, 1999). 4.3.2 Enthalpy measurements A twin calvet type microcalorimeter, based on the design described by Kleppa, (1976), installed at Gakushuin University, Tokyo, was used for our calorimetric measurements. It consisted of two sample chambers each of which was surrounded by a thermopile of PtPt10Rh thermocouples. The thermopiles were connected in opposition inside a massive kanthal block. The block was maintained at 978 K by heaters situated outside of the block (Fig. 4.4). The e.m.f. from the thermopile after amplification was recorded using an electronic integretar, which gave a graphical representation of the heat flow of the experiment (see Fig. 4.6) and a numerical value of the area under calorimetric peak after processing. 115 Chapter 4: Calorimetric studies Insulation Sample chamber Furnace Thermopile Calorimetric block Figure. 4.4 A Twin Calvet type microcaloirmeter of the type described by Kleppa, (1976) as used in this study. 2PbO.B2O3 was used as the solvent in our experiments. A single batch of solvent was made by mixing PbO and H3BO3 in a 1:1 molar proportion. The mixture was initially heated in a 1-atmosphere furnace at 1073 K for 30 minutes. The quenched glass was rehomogenised by grinding and remelted at 1323 K for one and half hours to ensure dehydration (Charlu et al., 1975). For each experiment 5 grams of 2PbO.B2O3 glass were measured into each of two Pt tubes, one for each sample chamber of the calorimeter, and melted in a furnace at 700°C for an hour. Each tube was placed into one of the calorimeter sample chambers housed in a second Pt tube jacketed by a silica glass tube inside an inconel tube. The inconel tube was inserted into the calorimeter prior to the beginning of each experiment and was allowed to reach thermal equilibrium as indicated by a stable base line from the thermopiles. Ar gas was bubbled through the solvent at a flow rate of 5 cm3/minute in order to hasten dissolution by stirring as shown in Fig. 4.5. The sample pellets of about 3 mg weight were made using a miniature pellet press and were dropped from room temperature into the 2PbO.B2O3 solvent at 978 K. Drops were alternated between the two samples chambers. Heat was either absorbed or liberated during the reaction and the thermopiles detected the change of temperature between the sample chamber and the kanthal block. The liberated 116 Chapter 4: Calorimteric studies heat gives rise to a calorimetric peak, which dies out exponentially to the original baseline with time as heat is transferred from the sample chamber to the block (Fig. 4.6). The recorded heat change is equal to the heat content of the sample plus the enthalpy of solution at the temperature of the calorimeter. The pyrope-rich samples were dissolved within an hour of dropping into the solvent; however, the majorite rich compositions required approximately 1.5 hours for complete dissolution. When the calorimetric peak dies out and gets back to the original base line the sample is considered to be totally dissolved. Pt Gas tube Inconel tube SiO glass tube 2 Pt test tube cover Pt test tube 2 PbO.B O flux 23 Sample pallet A r gas bubbling Ar gas bubbling Figure 4.5: A schematic diagram showing the set of the sample chambers used for drop solution calorimetric technique. Ar gas was bubbled through the sample chambers for hastening dissolution. 117 Chapter 4: Calorimetric studies 0 5000 10000 15000 20000 -15 -10 -5 0 5 10 15 Heat flow /mW Time / sec Base line ''Endothermic'' peak ''Exothermic'' peak Figure 4.6. A drop solution calorimetric curve of the pyrope composition showing the evolving heat of dissolution as a function of time. Each peak represents a drop solution run and are marked exothermic and endothermic peaks for distinction of heat flow values obtained from left and right sample chambers of the calorimeter (see Fig. 4.4). By integrating the area under the peaks, heat of dissolution associated with each run is determined. The heat effect is proportional to the area under the calorimetric peak but the proportionality factor must be calibrated. This calibration factor was determined for each sample chamber using Al2O3 as a known standard. Pellets of Al2O3 of approximately 3 mg were dropped into 5 gm of lead borate solvent. The integrated heat content of Al2O3 dissolution was compared with the standard ∆Hd-sol of Al2O3 of 106.73 kJ/mol to give calibration factors of 0.875 + 0.01 J/mol for the left-side and 0.896 + 0.01 J/mol for the right-side sample chambers (Table 4.3). 118 Chapter 4: Discussion Table 4.3: Calibration factor calculation using Al2O3 pellets. (Abbreviations: observed H is the observed heat, ∆H is the enthalpy of solution and S is the calibration factor). Left Side Run No. Al2O3 mass / g Al2O3 /mol Observed H / J Observed H/ KJ ∆H of Al2O3 / KJ.mol-1 S 1 0.003253333 3.1908E-05 3.0124 0.0030124 94.41003578 0.884568873 2 0.003303333 3.2398E-05 3.0242 0.0030242 93.34524406 0.874592374 3 0.00337 3.3052E-05 3.0815 0.0030815 93.23229125 0.87353407 4 0.003186667 3.1254E-05 2.9089 0.0029089 93.07354171 0.872046676 5 0.00346 3.3935E-05 3.1445 0.0031445 92.66368916 0.868206588 6 0.003286667 3.2235E-05 3.0253 0.0030253 93.8527221 0.879347157 7 0.003333333 3.2692E-05 3.1171 0.0031171 95.34678993 0.893345732 8 0.0034 3.3346E-05 3.061 0.003061 91.79488853 0.860066416 Average calibration factor 0.875713486 Right side Run No. Al2O3 mass / g Al2O3 /mol Observed H / J Observed H/ KJ ∆H of Al2O3 / KJ.mol-1 S 1 0.00325 3.1875E-05 3.0643 0.0030643 96.13510532 0.900731803 2 0.003526667 3.4588E-05 3.3203 0.0033203 95.99464319 0.899415752 3 0.003216667 3.1548E-05 3.0454 0.0030454 96.53223712 0.904452704 4 0.003353333 3.2888E-05 3.1942 0.0031942 97.12241338 0.909982323 5 0.00328 3.2169E-05 3.0046 0.0030046 93.40000628 0.875105465 6 0.00344 3.3738E-05 3.2499 0.0032499 96.32646916 0.902524774 7 0.00339 3.3248E-05 3.2332 0.0032332 97.24492779 0.911130214 8 0.003306667 3.2431E-05 3.0598 0.0030598 94.34887131 0.883995796 9 0.00325 3.1875E-05 3.0179 0.0030179 94.67941597 0.887092813 10 0.00316 3.0992E-05 2.9567 0.0029567 95.40129389 0.893856403 Average calibration factor 0.896828805 4.4 Results The measured drop solution enthalpies (∆H d-sol) of majorite-pyrope garnets are reported in Table 4.4. Due to the size of the synthesized samples a number of drops of approximately 3 mg each could be performed for each sample and the final ∆H d-sol is therefore the average of between 3-6 drops. 119 Chapter 4: Discussion Table 4.4: Drop solution enthalpies of Mg3(Mg,Si)Si3O12 (majorite)-Mg3Al2Si3O12 (pyrope) solid solution join in 2PbO.B2O3 at 978K, where Pyx is the pyrope composition determined from electromicrprobe, ∆H d-sol is the enthalpy of dissolution. Mass (g) ∆H d-sol (kJ /mol) Mass (g) ∆H d-sol (kJ /mol) Py 0.993 Py 0.776 0.00270 406.6921 0.00259 344.5220 0.00251 420.5045 0.00278 344.6312 0.00296 396.3194 0.00282 358.7211 0.00253 388.5248 0.00313 400.6793 0.00281 405.9892 Av. 403.118 ± 8.085 Av. 349.291 ± 8.166 Py 0.394 Py 0.323 0.00296 307.8711 0.00262 290.4055 0.00287 308.38 0.00255 286.598 0.00273 308.292 0.00281 292.6601 0.00261 292.6617 0.00270 303.6652 0.00273 291.9788 0.00289 294.4718 0.00273 301.6067 Av. 301.798 ± 5.796 Av. 293.472 ± 5.083 Mass (g) ∆H d-sol (kJ /mol) Mass (g) ∆H d-sol (kJ /mol) Py 0.245 Py 0.133 0.00261 214.3883 0.00299 252.4929 0.00256 230.8846 0.00268 241.8 0.00256 222.8271 0.00260 237.4546 0.00264 232.6677 0.00258 247.6614 0.00253 221.0675 Av. 224.3688 ± 5.989 Av. 245.8522 ± 7.131 Py 0.103 Maj 0.00257 281.2224 0.00257 244.5810 0.00261 265.012 0.00261 245.6691 0.00276 260.9722 0.00276 234.0641 0.00289 260.8478 0.00279 254.2783 0.00283 266.1986 0.00283 212.1408 Av. 266.251 ± 6.881 Av. 238.146 ± 12.968 The enthalpies of dissolution are plotted in Fig 4.7. The data show a negative deviation from a straight line joining the two end members. This implies that the mixing properties of the solid solution deviate positively from ideality. From the pyrope end member the 120 Chapter 4: Discussion enthalpies decrease non linearly with decreasing pyrope content, however, between a pyrope content of 0.32 and 0.24 a strong break in slope occurs and enthalpies start to increase before a final decrease occurs at the majorite end member. 020406080100 220 240 260 280 300 320 340 360 380 400 420 ∆H0 d-s KJ/mol mol % Al2O3 Mg4Si4O12 Mg3Al2Si3O12 Figure. 4.7: Enthalpies of dissolution of Mg3(Mg,Si)Si3O12-Mg3Al2Si3O12 garnets in 2PbO.B2O3 solvent at 978K. The drop dissolution enthalpy (∆H d-sol) can be written as, ∫ + ∆ = ∆Η −978 298 .dtCp Hsolsold (1) where, ∆H sol is the enthalpy of solution and the integral term accounts for the change in heat required to raise the temperature of the sample from room temperature to that of the solvent. Values for the integral term are 298 kJ/mol for pyrope (Robie et al., 1978) and 297 kJ/mol for majorite (Yusa et al., 1993) and it is assumed that this value varies linearly across the solid solution. The uncertainty in this correction is of the order of 1 kJ/mol. Values of ∆H sol are more useful for comparing with previous measurements and for calculating enthalpy changes for mineral reactions. 121 Chapter 5: Conclusions this exsolution reaction at mid-transition zone conditions, using mineral physics data, show that the impedance contrast for the initial perovskite exsolution is of a suitable magnitude to cause a discontinuity. Therefore, coupled with the wadsleyite-ringwoodite transition, two discontinuities are, possible and are consistent with the observed split in the 520 km seismic discontinuity. However, why the discontinuity should appear split in some regions of the Earth and not in others is shown not to be a function of temperature. Although temperature would tend to merge the two discontinuities together it would do so only at depths much deeper than where a single discontinuity is actually observed. What is more likely is that a single discontinuity is observed in regions where the Ca content of the mantle is low but in regions rich in Ca, such as those containing significant recycled oceanic crust, two discontinuities would be observed. Our findings allow regional seismic observations of this splitting in the 520 km discontinuity to be used as a probe for this major type of mantle heterogeneity in the mantle. This study is, therefore, the first report of a deep mantle seismic discontinuity, which can be used as a sensitive indicator of mantle chemical heterogeneity. Changes in the elastic properties of magnesium silicate perovskite have been examined as a function of the incorporation of ferric Fe and Al in the structure, using diamond anvil cell compression and single-crystal X-ray diffraction techniques. Compression experiments on Al-and Fe-bearing magnesium silicate perovskites show that Fe and Al incorporation makes the magnesium silicate perovskite structure more compressible. The Al3+ and Fe3+ substitution in magnesium perovskite will mostly occur via a coupled substitution mechanism at mantle conditions; however at low trivalent cation concentrations a small proportion of oxygen vacancies may form. A third order Birch– Murnaghan fit of the compression data yields a bulk modulus which decreases from 243 to 234 GPa with increasing Fe and Al content with a rise in K' from 5 to 6.5, which can be attributed to an increase in polyhedral compressibility. On consideration of the effects of possible substitution mechanisms on molar volumes of Al-and Fe-bearing magnesium silicate perovskite, it has been observed that in magnesium silicate perovskites existing at lower mantle conditions the substitution of the component will be energetically less favoured. Our results show that disproportionation of Fe is energetically favorable in the lower mantle conditions which would result in the enrichment of Fe OSiFe VIVIII 3 42 ++ 3+ at lower mantle conditions. 128 Chapter 5: Conclusions Slow kinetics of silicate reactions inhibits the attainment of equilibrium under feasible experimental conditions at lower temperatures. Consequently, the experimental results are generally extrapolated to lower temperatures using suitable thermodynamic models. Uncertainties in such models can be reduced considerably, if thermodynamic parameters used for such fits are independently determined as has been done in our study of the pyrope-majorite solid solution using the drop solution calorimetric technique. Enthalpies of solution along the majorite–pyrope join obtained from drop solution calorimetry show a significant deviation from ideality. The enthalpy of dissolution decreases non-linearly for compositions below 30mol% pyrope due to a symmetry change from cubic to tetragonal. The values of drop solution enthalpies on majorite–pyrope join are significantly lower than previous estimates. An estimation of the excess properties using a symmetric regular solution model gives a value of WH {interaction parameter of mixing on one cation site in garnet i.e., (Mg,Si)↔Al} of 38 kJ/mol and a solution enthalpy of the fictive cubic majorite garnet of –37 kJ/mol. This would mean a large positive enthalpy of mixing along this solid solution join. This thesis work emphasizes how a combination of complimentary experimental techniques can lead to a robust assessment of the deep interior of the Earth and contributes to a better understanding of the evolution of the Earth. Further work During the course of this thesis a number of points have become obvious where further experimental studies would benefit our understanding of the silicate perovskite forming reactions. A correlation was observed between the seismically determined geographical distribution of the split 520 km discontinuity with modern and past subduction zones as mentioned in chapter 2. This supports our idea of Ca-enrichment in the mantle via subduction of oceanic crust, which would lead to the formation of Ca-perovskite causing the second discontinuity in the 520 km region. So, further geophysical studies into the splitting and variability of the 520 km seismic discontinuity combined with our interpretation will allow us to scale lateral distribution of mantle heterogeneity on a variety of scales and will provide significant insights into the circulation of subducted oceanic lithosphere in the mantle. Furthermore as the oceanic basalt bulk compositions are rich in 129 Chapter 5: Conclusions the SiO2 component, additional phase equilibria studies involving the formation of Caperovskite from Ca2SiO4 and CaSi2O5 will lead to better constraints on the thermodynamic modeling of the Ca-perovskite forming reaction. As very little thermodynamic data exist for calcium silicate perovskite, another important aspect will be to collect high quality calorimetric data on Ca2SiO4 and CaSi2O5 to aid in the calculation of calcium perovskite thermodynamic properties from its formation reaction. Our study of the equation of state of Aland Fe-bearing magnesium silicate perovskite single crystals was performed at ambient temperature up to pressures of 10 GPa. It was observed in our study that even though a K' value greater than 4 was obtained from our results, a calculation of KT for lower mantle conditions with a K' = 4, provides a better fit to the predicted lower mantle bulk modulus. Given this discrepancy, it is necessary to extend the equation of state study to pressure-temperature range relevant for mantle conditions in order to better quantify the changes in elastic property at those conditions due to Al and Fe incorporation in the magnesium silicate structure. This will be possible, using a gas loading of the diamond anvil cell for compression experiments, which would provide hydrostatic conditions to higher pressures. In addition, a powder x-ray diffraction equation of state study could be carried out using a high intensity synchrotron X-ray source, in case of failure in producing suitable single crystals. Another important objective would be to carry out crystal structural refinement studies using single crystal X-ray diffraction, provided we can synthesize single crystals of suitable size possibly twin-free, for a better understanding of the substitution mechanism of cations in different crystal structural sites. For the calorimetric measurements on the pyrope–majorite join, it will be important to reproduce the heat of dissolution data for the majorite-rich compositions, where a strong scatter in data has been observed by our study, in order to confirm whether there is an effect of ordering of Mg and Si on the octahedral site related to synthesis conditions. A thorough assessment of the ordering of Mg and Si for the majorite-rich compositions can be carried out using a Raman spectroscopic study or a 27Al-NMR study on these majoriterich compositions. In addition, in order to employ our thermodynamic data at mantle temperatures, information on the excess entropy term will be required. This can be determined by heat capacity measurements along the majorite–pyrope join by employing differential scanning calorimetry or the recently-devised PPMS (Physical Properties Measurement System) calorimetric technique. 130 Appendix A Table A.1: Details of run numbers and phases observed. Abbreviations: Pd (peridotitic composition), Baslt (basaltic composition), Pyr (pyrope composition), Fo (forsterite composition), Pd gt (perdotitic garnet), Baslt gt. (basaltic garnet), Pyr gt (pyrope garnet), Maj gt (majoritic garnet), Ca-Pv (calcium perovskite), Al–phase (a non-stoichometric unknown phase with Al), Ring (ringwoodite), Mw (magnesiowüstite), Stish (stishovite), Oliv (olivine), Pyrx.(pyroxene), Peri (periclase),Wad (wadsleyite), Pd (peridotite), Pv (perovskite),Ca-Pd ,CaBaslt, (all Ca bearing reversal compositions), Ca-Pd gt, Ca-baslt Gt.(reversal phases with Ca) Run no. Pressure Starting compositions Phases observed (GPa) 1400°C H2375 17.9 Pd, Baslt, Pyr, Fo20 Pd gt + Ca-Pv, Baslt gt +Ca-Pv, Pyr gt + Ca-Pv, Ring + Mw + Stish. S3611 18.1 Pd, Baslt, Pyr, Fo20 Pd gt + Ca-Pv, Baslt gt +Ca-Pv, Pyr gt + Ca-Pv, Ring + Mw + Stish S3614 18.6 Pd, Baslt, Pyr, Fo20 Pd gt + Ca-Pv, Baslt gt +Ca-Pv, Pyr gt + Ca-Pv, Ring + Mw + Stish H2370 19.5 Pd, Baslt, Pyr, Fo20 Pd gt + Ca-Pv, Baslt gt +Ca-Pv, Pyr gt + Ca-Pv, Ring + Mw + Stish H2241 19.6 Pd, Baslt, Pyr, Fo20 Pd gt + Ca-Pv, Baslt gt +Ca-Pv, Pyr gt + Ca-Pv, Ring + Mw + Stish 1600°C S3460 18. Pd, Baslt, Maj, Fo90 Pd gt + Ca-Pv, Baslt gt, Maj gt + Ca-Pv, Wad S3470 18.8 Pd, Baslt, Pyr, Fo30 Pd gt + Ca-Pv, Baslt gt + Ca-Pv Pyx + Wad S3550 19.2 Pd, Baslt, Pyr, Fo30 Pd gt + Ca-Pv, Baslt gt + Ca-Pv Pyr Gt + Ca-Pv, Ring + Mw + Stish S3548 19.5 Pd, Baslt, Fo30 Pd gt + Ca-Pv, Baslt gt + Ca-Pv, Ring + Mw + Stish S3547 19.5 Pd, Baslt, Fo30 Pd gt + Ca-Pv, Baslt gt + CaPv, Ring + Mw + Stish S3549 19.9 Pd, Baslt, Pyr, Fo30 Pd gt + Ca-Pv, Baslt gt + Ca-Pv Pyr Gt, Ring + Mw + Stish S3657 19.9 Baslt, F090 + Baslt, Fo30 Baslt gt + Ca-Pv, Baslt gt + Ca-Pv + Oliv, Mw + Pyrx + Stish + Wad S3655 19.9 Baslt, F090 + Baslt Baslt gt + Ca-Pv, Baslt gt + Ca-Pv + Oliv. S3551 20.4 Pd, Baslt, Pyr, Fo30 Pd gt + Ca-Pv, Baslt gt + Ca-Pv Pyr Gt + Ca-Pv, Ring + Mw + Stish S3757 20.7 Pd, Baslt, Fo30; Fo50 Pd gt + Ca-Pv, Baslt gt + Ca-Pv, 131 Ring + Mw + Stish, Ring + Mw + stish S3480 22.1 Pd, Baslt, Pyr, Fo30 Pd gt+ Ca-Pv +Al –Phase, Baslt gt, + Ca-Pv + Al phase, Ring +Stish S3764 21.2 Pd, Baslt, Fo30 Pd gt + Ca-Pv, Baslt gt + Ca-Pv, Ring + Mw + Stish S3498 21.4 Pd, Baslt, Pyr, Fo30 Pd gt + Ca-Pv, Baslt gt + Ca-Pv Pyr gt + Al phase+ Ca-Pv, Ring + Mw + Stish S3783 22.3 Pd, Baslt, Maj, Fo70 Pd gt +Ca-Pv, Baslt gt + Ca-Pv, Ring + Mw + Stish + Pyrx. S3484 22.3 Pd, Ca-Baslt, Pyr, Fo30 Pd gt +Ca-Pv + Peri, Ca-Baslt gt + Ca-Pv, Pyr gt + Ca-Pv, Ring + Mw S3475 22.6 Pd, Baslt, Pyr, Maj Pd gt + Ca-Pv, Baslt gt + Ca-Pv, Pyr Gt + Ca-Pv, Maj gt + Ca-Pv S3478 23 Pd, Baslt, Fo30 Pd gt + Ca-Pv, Baslt gt + CaPv, + Gt + Ca-Pv, Pv + Mw + Stish S3784 23.5 Pd, Ca-Baslt, Fo70, Pd Pd gt +Ca-Pv, Ca-Baslt gt + Ca-Pv Mw + Pv, Stish + Mw +Pv + Cor Reversals (1600°C) S3538 17.95 Baslt, Ca Baslt, Fo30 Baslt gt + Pyx, Ca-Pd gt + Ca-Pv, Pyx + Wad S3543 18.4 Baslt, Ca Baslt, Fo30 Baslt gt + Pyx, Ca-Pd gt + Ca-Pv Pyx +Wad +Mw S3515 19.8 Baslt, Ca Baslt, Fo30, Fo98 Baslt gt + Ca-Pv, Ca-Baslt gt + Ca-Pv, Ring +Wad + Mw, Wad S3521 20.7 Pd, Ca Pd, Fo30 Pd gt +Ca-Pv, Ca-Pd gt + Ca-Pv Ring + Mw S3523 20.8 Pd, Ca-Pd, Ca-Baslt, Fo30 Pd gt + Ca-Pv + Al-phase +Ca-Pd gt + Ca-Pv, Ring +Stish S3784 23.5 Pd, Ca-Baslt, Fo70, Pd Pd gt +Ca-Pv, Ca-Baslt gt + Ca-Pv, Mw + Pv, Stish + Mw +Pv + Cor Table A.2: Electron microprobe data of the perovskites crystallized from the pressure calibrants in multi anvil experiments calculated based on 3 oxygens per formula unit. Fo30 (Forsterite 30), Fo70 (Forsterite 70) Run no. Pressure calibrant Pressure Si Mg Fe Total S3478 Fo30 23 GPa 0.961 0.763 0.310 2.034 S3784 Fo70 23.5 GPa 0.962 0.899 0.171 2.031 132 Table A.3: Electron microprobe data for all the experimental runs before processing of data. All the concentrations are in wt% oxide. (A) for experiments at 1600°C, (B) for experiments at 1400°C and (C) for reversal experiments. Results are listed as peridotite, basalt, pyrope and majorite based on the different garnet starting compositions as listed in table 2.1. (Abbreviation Pd gt–peridotitic garnet, Ca-Pv-calcium perovskite, Baslt gt-basaltic garnet, Maj gt –majoritic garnet, Pyr gtpyrope garnet, Fo90-forsterite 90 composition, Fo50-forsterite 50 composition, Fo30forsterite 30 composition, (Mg,Fe)O-magnesiowüstite, SiO2-stishovite, (A) Runs at 1600°C. (Run no. Composition Phases SiO2 MgO FeO CaO Al2O3 Total observed S3460 Peridotite Pd gt 52.92 24.96 1.71 11.10 7.82 98.49 Ca-Pv 59.54 10.71 0.77 16.88 3.12 91.03 Basalt Baslt gt 49.31 21.02 0.83 14.81 12.70 98.66 Majorite Maj gt 54.51 28.92 1.68 8.36 5.79 99.26 Ca -Pv 47.89 0.09 0.00 44.03 0.01 92.02 Fo 90 Olivine 42.18 56.04 2.32 0.02 0.10 100.66 S3470 Peridotite Pd gt 53.26 25.98 1.12 11.21 8.26 99.82 Ca-Pv 49.25 6.34 0.58 38.21 1.74 96.13 Basalt Baslt gt 49.17 22.74 0.55 14.09 13.23 99.77 Pyrope Pyr gt 46.97 19.24 0.82 16.16 16.30 99.48 Ca-Pv 49.68 0.75 0.03 43.66 0.58 94.71 Fo 90 Olivine 42.08 56.53 1.71 0.00 0.10 100.43 S3550 Peridotite Pd gt 50.26 23.90 3.67 11.91 8.26 98.01 Ca-Pv 50.08 0.17 0.10 47.33 0.08 97.77 Basalt Baslt gt 45.55 17.47 2.95 16.70 16.06 98.73 Ca-Pv 47.57 2.16 0.42 43.91 1.84 95.89 Pyrope Pyr gt 47.39 20.84 3.49 14.30 12.83 98.84 Ca-Pv 49.34 0.78 0.23 46.39 0.55 97.29 Fo30 olivine 32.87 17.17 51.65 0.01 0.05 101.76 (Mg,Fe)O 1.22 3.34 89.81 0.07 0.30 94.74 SiO2 98.20 0.27 2.06 0.01 0.21 100.75 S3548 Peridotite Pd gt 51.49 27.95 3.90 7.95 8.33 99.62 Ca-Pv 49.20 0.20 0.10 46.98 0.07 96.56 Basalt Baslt gt 47.90 24.00 5.07 9.83 12.45 99.25 133 Fo30 olivine 33.86 17.88 49.40 0.02 0.05 101.21 (Mg,Fe)O 3.68 8.78 85.02 0.02 0.04 97.54 SiO2 98.83 0.63 2.40 0.01 0.08 101.95 S3547 Peridotite Pd gt 51.11 27.53 2.87 8.44 9.38 99.32 Ca-Pv 50.24 0.13 0.06 47.51 0.07 98.00 Basalt Baslt gt 48.58 24.21 3.56 9.72 12.46 98.55 Ca-Pv 46.68 0.62 0.24 46.01 0.57 94.12 Fo30 olivine 33.66 18.91 49.39 0.00 0.05 102.02 (Mg,Fe)O 2.93 9.40 87.59 0.00 0.07 99.98 SiO2 96.86 0.18 1.62 0.01 0.19 98.87 S3549 Peridotite Pd gt 50.47 28.51 1.81 7.08 9.17 97.04 Ca-Pv 45.78 1.70 0.22 44.84 0.84 93.38 Basalt Baslt gt 47.46 25.27 2.70 8.55 13.60 97.58 Ca-Pv 48.52 0.36 0.10 46.44 0.10 95.52 Pyrope Pyr gt 44.08 20.65 2.82 11.52 17.84 96.91 Fo30 olivine 33.65 21.30 45.37 0.00 0.06 100.39 (Mg,Fe)O 1.08 7.40 87.90 0.02 0.17 96.57 SiO2 95.05 0.39 3.64 0.01 0.03 99.11 S3657 Basalt Basltic maj 43.46 12.43 34.93 2.89 4.52 100.36 Ca-Pv 48.27 0.35 0.30 46.38 0.33 95.62 Fe+ basalt Baslt gt 46.85 24.20 7.38 8.28 11.92 98.63 Ca-Pv 48.04 0.19 0.12 46.26 0.32 94.93 Fo30 olivine 33.67 21.25 45.76 0.00 0.06 100.75 (Mg,Fe)O 2.84 4.62 86.96 0.06 0.38 94.87 SiO2 72.70 5.19 17.46 0.01 0.08 95.44 S3655 Fe+ basallt Baslt gt 44.01 21.00 6.10 7.59 13.97 92.67 Basalt Baslt gt 48.48 26.19 4.02 7.73 12.57 98.99 Ca-Pv 47.90 1.78 0.61 45.29 0.11 97.02 Fo30 olivine 37.05 32.64 30.89 0.00 0.03 100.61 (Mg,Fe)O SiO2 99.07 0.08 0.72 0.00 0.02 99.89 S3551 Peridotite Pd gt 50.14 27.43 4.46 5.82 10.48 98.33 Ca-Pv 48.22 0.23 0.16 46.70 0.57 95.88 Basalt Baslt gt 46.41 27.00 1.73 7.95 14.18 97.27 134 Ca-Pv 39.42 0.92 0.24 43.37 3.69 87.64 Pyrope Pyr gt 45.03 21.99 3.80 9.54 18.46 98.82 Ca-Pv 50.29 0.50 0.16 47.27 0.67 98.89 Fo30 olivine 34.64 24.08 42.60 0.01 0.06 101.38 (Mg,Fe)O 5.32 5.44 82.80 0.04 0.37 93.96 SiO2 94.77 1.42 4.41 0.02 0.06 100.67 S3480 Peridotite Pd gt 53.91 28.18 4.23 4.92 8.64 99.88 Ca-pv 51.35 0.07 0.03 50.05 0.03 101.53 Basalt Baslt gt 50.82 25.90 4.62 5.35 13.73 100.42 Ca-Pv 50.85 0.09 0.09 49.40 0.05 100.48 Pyrope Pyr gt 46.51 19.56 8.61 8.77 16.70 100.15 Ca-Pv 50.21 0.18 0.34 49.42 0.43 100.58 Fo 50 Olivine 37.98 29.59 32.44 0.00 0.07 100.07 SiO2 101.54 0.05 0.54 0.02 0.03 102.17 S3498 Peridotite Pd gt 53.22 32.12 5.44 0.71 9.53 101.02 Ca-pv 51.29 0.70 0.18 45.56 0.26 97.98 Basalt Baslt gt 48.00 27.10 4.76 4.66 15.64 100.16 Ca-Pv 50.54 0.31 0.11 45.32 0.22 96.50 Pyrope Pyr gt 45.16 24.71 3.95 5.38 21.49 100.68 Ca-Pv 50.52 0.20 0.09 45.82 0.10 96.73 Fo 50 Olivine 38.72 37.48 26.17 0.00 0.08 102.46 (Mg,Fe)O 6.21 12.41 70.20 0.14 0.73 89.69 S3484 Peridotite Pd gt 52.31 33.49 3.00 0.89 9.24 98.93 Ca-pv 53.01 1.76 0.12 42.10 0.49 97.48 Basalt Baslt gt 47.63 28.78 3.44 2.38 16.66 98.88 Ca-Pv 51.29 0.23 0.09 46.93 0.09 98.63 Pyrope Pyr gt 44.03 25.42 3.84 2.04 23.32 98.64 Ca-Pv 49.31 1.60 0.33 45.24 1.43 97.91 Fo 50 Olivine 37.01 30.79 31.97 0.00 0.09 99.87 (Mg;Fe)O 0.71 21.41 75.89 0.00 0.12 98.13 S3475 Peridotite Pd gt 53.23 31.49 5.09 0.86 9.61 100.27 Ca-Pv 48.06 6.17 1.13 35.97 2.09 93.42 Basalt Baslt gt 49.03 27.26 3.28 3.03 17.03 99.62 Ca-Pv 46.97 0.43 0.15 45.41 0.72 93.68 135 Majorite Maj gt 49.32 16.34 2.41 21.32 7.36 96.75 Ca -Pv 49.06 0.96 0.20 44.84 0.58 95.64 Pyrope Pyr gt 44.96 24.00 3.58 5.04 22.35 99.94 Ca-Pv 42.84 0.52 0.26 44.77 2.38 90.77 S3478 Peridotite Pd gt 51.90 33.15 5.20 0.20 9.09 99.53 Ca-pv 48.86 0.15 0.11 46.80 0.08 96.00 Basalt Baslt gt 47.40 28.31 4.42 2.96 16.32 99.41 Ca-Pv 42.92 0.95 0.41 45.06 2.56 91.89 Pyrope Pyr gt 44.22 24.56 4.81 4.28 21.55 99.41 Ca-Pv 45.29 1.30 0.58 44.21 2.20 93.58 Fo 50 Pv 51.19 27.09 19.74 0.24 1.92 100.19 (Mg,Fe)0 0.38 20.28 78.75 0.00 0.10 99.51 SiO2 97.62 0.37 2.18 0.00 0.02 100.19 S3757 Peridotite Pd gt 52.37 29.20 4.14 4.48 8.42 98.61 Ca-pv 47.99 0.39 0.25 45.93 0.73 95.29 Basalt Baslt gt 47.64 24.98 3.36 6.26 15.54 97.78 Ca-Pv 44.64 1.13 0.49 43.13 1.16 90.55 Fo50 Olivine 34.19 26.74 39.20 0.02 0.10 100.25 (Mg,Fe)0 0.22 14.33 82.58 0.02 0.09 97.23 SiO2 79.56 7.10 12.08 0.02 0.17 98.93 Fo 30 Olivine 35.04 25.43 40.32 0.01 0.07 100.86 (Mg,Fe)0 0.94 13.63 84.07 0.01 0.08 98.73 SiO2 94.94 1.56 3.37 0.02 0.21 100.10 S3764 Peridotite Pd gt 52.70 28.41 4.82 3.67 9.60 99.19 Ca-pv 48.12 1.14 0.19 43.70 0.96 94.11 Basalt Baslt gt 48.06 27.32 3.58 4.68 15.44 99.09 Ca-Pv 49.43 0.04 0.04 46.96 0.03 96.50 Fo30 Olivine 35.64 27.84 37.14 0.00 0.09 100.70 (Mg,Fe)0 1.98 17.16 80.91 0.01 0.10 100.16 SiO2 90.05 4.11 7.14 0.00 0.25 101.56 S3784 Peridotite Pd gt 51.46 32.16 4.87 0.50 10.17 99.15 Ca-pv 49.83 0.29 0.13 46.69 0.26 97.19 Fo70 (Mg,Fe)0 6.90 29.88 64.18 0.04 0.01 101.01 Pv 55.18 34.70 11.18 0.16 0.08 101.30 136 S3783 Peridotite Pd gt 51.34 30.92 5.77 1.45 9.39 98.87 Ca-Pv 49.22 0.58 0.19 45.07 0.36 95.43 Basalt Baslt gt 50.28 15.75 2.98 23.26 4.87 97.15 Ca-Pv 48.47 0.24 0.13 44.60 0.29 93.72 Majorite Maj gt 49.83 11.87 2.27 28.60 3.73 96.29 Ca-Pv 49.04 0.16 0.08 45.60 0.11 94.99 Fo70 Olivine 37.18 32.30 31.24 0.00 0.07 100.79 (Mg,Fe)0 8.31 12.62 74.94 0.11 0.60 96.58 SiO2 91.16 1.26 3.35 0.02 0.15 95.93 (B) Runs at 1400°C Run no. Composition Phases SiO2 MgO FeO CaO Al2O3 Total observed S3611 Pyrope Pyr gt 44.92 18.91 2.48 14.50 19.34 100.14 Ca-Pv 48.94 0.44 0.12 45.83 0.45 95.78 Basalt Basalt gt 46.66 19.99 3.38 14.66 15.74 100.42 Ca-Pv 49.61 0.93 0.18 46.22 0.50 97.43 Peridotite Pd gt 51.22 26.26 4.36 9.21 8.56 99.61 CaPv 49.55 0.19 0.12 47.35 0.05 97.26 S3614 Pyrope Pyr gt 44.15 19.66 3.29 13.64 18.79 99.53 Ca-Pv 48.32 0.22 0.12 47.13 0.38 96.18 Basalt Basalt gt 47.00 23.50 3.77 11.35 14.07 99.69 Ca-Pv 48.49 0.28 0.12 47.16 0.36 96.41 Peridotite Pd gt 50.74 27.91 4.54 8.60 8.24 100.03 CaPv 49.20 0.11 0.11 47.41 0.13 96.96 H2375 Pyrope Pyr gt 45.21 18.65 3.23 14.01 17.93 99.02 Ca-Pv 50.21 0.10 0.06 46.94 0.08 97.39 Basalt Basalt gt 47.46 20.57 3.73 13.36 13.67 98.79 Ca-Pv 48.97 0.18 0.13 46.71 0.17 96.15 H2370 Pyrope Pyr 44.03 23.22 4.08 6.33 22.52 100.18 Ca-Pv 50.22 0.07 0.09 47.55 0.06 97.99 Fo 30 olivine 33.45 20.65 51.13 0.01 0.04 105.28 (Mg,Fe)O 0.75 3.24 93.06 0.10 0.31 97.46 SiO2 94.96 0.53 2.74 0.15 0.16 98.54 H2241 Pyrope Pyr gt 44.04 22.89 3.25 6.31 21.20 97.68 137 Deuss, A. and Woodhouse, J. (2001) Seismic observations of splitting of the mid-transition zone discontinuity in earth’s mantle. Science, 294: 354-357. Deuss, A., Redfern, S., Chambers, K. and Woodhouse, J. H. (2006) The nature of the 660-kilometer discontinuity in Earth's mantle from global seismic observations of PP precursors. Science, 311: 198201. Dziewonski, A. M. and Anderson, D. L. (1981) Preliminary Reference Earth Model. Physics of Earth and Planetary Interiors, 25: 297–358. Fabrichnaya, O. B., Saxena, S. K., Richet, P. and Westrum E. F. (2004) Thermodynamic data, models and phase diagrams in multicomponent oxide systems. Springer, Berlin-Heidelberg-New York, p 198. Fei, Y., Mao, H. K. B. and Mysen, O. B. (1991) Experimental determination of element partitioning and calculation of phase relations in the MgO -FeO -SiO2 system at high pressure and high temperature. Journal of Geophysical Research, 96: 2157-2169. Fei, Y., Mao, H. K., Shu, J. and Hu, J. (1992) P-V-T equation of state of magneiowüstite. Physics and Chemistry of Minerals, 18: 416-422. Fei, Y., Wang, Y. and Finger, L. W. (1996) Maximum solubility of FeO in (Mg, Fe)SiO3-perovskite as a function of temperature at 26Gpa: implication for FeO content in the lower mantle. Journal of Geophysical Research, 101(B5): 11525-11530. Fei, Y. and Bertka, C. M. (1999) Phase transitions in the Earth’s mantle and mantle mineralogy. In Fei, Y., Bertka, C. M. and Mysen, B. O., editors, Mantle petrology: Field observations and high pressure experimentation: A tribute to Francis R. Boyd, volume 6 of The Geochemical Society special Publication, p 189-207. The Geochemical Society, Houston. Fei, Y., van Orman, J., Li, J., van Westrenen, W., Sanloup, C., Minarik, W., Hirose, K., Komabayashi, T., Walter, M. and Funakoshi, K. (2004) Experimentally determined post-spinel transformation boundary in Mg2SiO4 using MgO as an internal pressure standard and its geophysical implications. Journal of Geophysical Research, 109(B02305), doi: 10.1029/2003JB002562. Fiquet, G., Andrault, D., Dewaele, A., Charpin, T., Kunz, M. and Haüsermann, D. (1998) P-V-T equation of state of MgSiO3 perovskite. Physics of Earth and Planetary Interiors, 105: 21-31. 144 Fiquet, G., Dewaele, A., Andrault, D., Kunz, M. and Le Bihan, T. (2000) Thermoelastic properties and crystal structure of MgSiO3 perovskite at lower mantle pressure and temperature conditions. Geophysical Reaserch Letters, 27(1): 21-24. Flanagan, M. P. and Shearer, P. M. (1998) Global mapping of topography on transition zone velocity discontinuities by stacking SS precursors. Journal of Geophysical Research, 103: 2673–2692. Frost, D. J., Langenhorst, F., van Aken, P. (2001) Fe-Mg partitioning between ringwoodite and magnesiowüstite and the effect of pressure, temperature and oxygen fugacity. Physics and Chemistry of Minerals, 28: 455-470. Frost, D. J. and Langenhorst, F. (2002) The effect of Al2O3 on Fe-Mg partitioning between magnesiowüstite and magnesium silicate perovskite. Earth and Planetary Science Letters, 199: 227241. Frost, D. J. (2003a) The structure and sharpness of (Mg,Fe)2SiO4 phase transformations in the transition zone. Earth and Planetary Science Letters, 216: 313-318. Frost, D. J. (2003b) Fe2+-Mg partitioning between garnet, magnesiowüstite, and (Mg,Fe)2SiO4 phases of the transition zone. American Mineralogist, 88: 387-397 Frost, D. J., Liebske, C., Langenhorst, F., McCammon, C., Trnnes, R. G. and Rubie, D. C. (2004) Experimental evidence for the existence of iron-rich metal in the Earth’s lower mantle. Nature, 428: 409-412. Frost, D. J., Poe, B. T., TrØnnes, R. G., Liebske, C., Duba, A., and Rubie, D. C. (2004) A new large volume multianvil system. Physics of Earth and Planetary Interiors, 143-144: 507-514. Frost, D. J. (2006) The stability of hydrous mantle phases. In Keppler, H. and Smyth, J. R., editors, Water in nominally anhydrous minerals, volume 62 of Reviews in Mineralogy and Geochemistry, p. 243-271, Mineralogical Society of America, Chantilly, Virginia. Frost, D. J. and Doleĵs D. (2007) Experiemntal determination of the effect of H2O on the 410 km discontinuity. Earth and Planetary Science Letters, 256(1-2): 182-195. Frost, D. J. (2007) The upper mantle and transition zone. Elements (in press). 145 Funamori, N., Yagi, T., Utsumi, W., Kondo, T., Uchida, T. and Funamori, M. (1996) Thermoelastic properties of MgSiO3 perovskite determined by in situ X-ray observations up to 30 GPa and 2000K. Jouranl of Geophysical Research, 101(B4): 8257-8269. Gasparik, T. (1990) Phase relations in the Transition Zone. Journal of Geophysical Research, 95 (B10): 15751-15769. Gasparik, T. (1996) Melting experiments on the enstatite-diopside join at 70-224 Kbar, including the melting of diopside. Contributions to Mineralogy and Petrology, 124: 139-153. Gasparik, T. (2003) Phase diagrams for geoscientists. An atlas of the Earth’s interior. Springer, Berlin Heidelberg New York, p 462. Gilbert, F. and Dziewonski, A. M. (1975) An application of normal mode theory to the retrieval of structural parameters and source mechanisms from seismic spectra. Philosophical Transactions Royal Society London, A 278: 187–269. Gilbert, H. J., Sheehan, A. F., Dueker, K. G. and Molnar, P. (2003) Receiver functions in the western United States, with implications for upper mantle structure and dynamics. Journal of Geophysical Research, 108, 10.1029/2001JB001194. Galzer, A. M. (1972) The classification of tilted octahedra in perovskites. Acta Crystallographica, B28: 3384-3392. Glazer, A. M. (1975) Simple ways of determining perovskite structures. Acta Crystallographica, A31: 756-762. Gwanmesia, G. D., Liu, J., Chen, G., Kesson, S., Rigden, S. M. and Libermann, R. C. (2000) Elasticity of the pyrope (Mg3Al2Si3O12)-majorite (MgSiO3) garnets solid solution. Physics and Chemistry of Minerals, 27: 445-452. Harte, B., Harris, Hutchinson, M. T., G. R., Watt, M. C. and Wilding, M. C. (1999) Lower mantle mineral associations in diamonds from São Luiz, Brazil. In Fei, Y., Bertka, C. M. and Mysen, B. O., editors, Mantle petrology: Field observations and high pressure experimentation: A tribute to Francis R. Boyd, volume 6 of The Geochemical Society Special Publication, p 125-153. The Geochemical Society, Houston. 146 Haselton, H. T. and Newton, R. C. (1980) Thermodynamics of pyrope-grossular garnets and their stabilities at high temperatures and high pressures. Journal of Geophysical Research, 85: 6973-6982. Hatch, D. M. and Ghose, S. (1989) Symmetry analysis of the phase transition and twinning in MgSiO3 garnet: Implications to mantle mineralogy. American Mineralogist, 74: 1221-1224. Heinemann, S., Sharp, T. G., Seifert, F. and Rubie, D. C. (1997) The cubic to tetragonal phase transition in the system majorite (Mg4Si4O12)-pyrope (Mg3Al2Si3O12), and garnet symmetry in the Earth’s transition zone. Physics and Chemistry of Minerals, 24: 206-221. Heinrich, K. F. G. and Newbury, D. E. (1991) Electron probe Quantitation. Springer, p. 412. Hemley, R. J. and Cohen, R. E. (1992) Silicate perovskite. Annual Reviews in Earth and Planetary Science, 20: 553-600. Helffrich, G. R and Wood, B.J. (1996) 410 km discontinuity sharpness and the form of the olivine -β phase diagram: resolution of apparent seismic contradictions. Geophysical Journal International, 126: F7-F12. Helffrich, G. R. and Wood, B. J. (2001) The Earth’s mantle. Nature, 412: 501-507. Hirose, K and Fei, Y. (2002) Subsolidus and melting phase relations of basaltic composition in the uppermost lower mantle. Geocheimca Cosmochimica Acta, 66: 2099-2108. Holzapfel, C., Rubie, D. C., Frost, D. J., Langenhorst, F. (2005) Fe-Mg interdiffusion in (Mg, Fe)SiO3 perovskite and lower mantle reequilibration. Science, 309: 1707-1710. Howard, C. J. and Stokes, H. T. (1998) Group theoretical analysis of octahedral tilting in perovskites, Acta Crystallographica, B54: 782-789. Irifune, T. (1987) An experimental investigation of the PyroxeneGarnet transformation in a pyrolite composition and its bearing on the constitution of the mantle. Physics of the Earth and Planetary Interiors, 45: 324-336. 147 Irifune, T. and Ringwood, A. E. (1993) Phase transformations in subducted oceanic crust and buoyancy relationships at depths of 600-800 km in the mantle. Earth and Planetary Science Letters, 117: 101-110. Irifune, T. (1994) Absence of an aluminous phase in the upper part of the Earth’s lower mantle. Nature, 370: 131-133. Irifune, T. and Isshiki, M. (1998) Iron partitioning in a pyrolite mantle and the nature of the 410-km seismic discontinuity. Nature, 349: 409-411. Irifune, T., Nishiyama, N., Kuroda, K., Inoue, T., Isshiki, M., Utsumi, W., Funakoshi, K-I., Urakawa, S., Uchida, T., Katsura, T. and Ohtaka, O. (1998) The post-spinel phase boundary in Mg2SiO4 determined by in situ x-ray diffraction. Science, 279: 1698-1701. Ita, J. and Stixrude, L. (1992) Petrology, elasticity, and composition of the mantle transition zone. Journal of Geophysical Research, 97: 6849–6866. Ito, E. and Takahashi, E. (1989) Post spinel transformations in the system Mg2SiO4-Fe2SiO4 and some geophysical implications. Journal of Geophysical Research, 94: 10637-10646. Jackson, I. and Rigden, S. M. (1996) Analysis of P-V-T data: Constraints on the thermoelastic properties of high-pressure minerals. Physics of Earth and Planetary Interiors, 96: 85-112. Jackson, J. M., Zhang, J., Shu, J., Sinogeikin, S. V. and Bass, J. D. (2005) High-pressure sound velocity and elasticity of aluminous MgSiO3 perovskite to 45 GPa: implications for lateral heterogeneity in Earth’s lower mantle. Geophysical Research Letters, 32L21305, doi: 10.1029/2005GL023522. Javoy, M. (1995) The integral enstatite chondrite model of the Earth. Geophysical Research Letters, 22(16): 2219-2222. Jeffreys, H. and Bullen, K: E. (1940) Seismological Tables. British Association Seismological Committee, London. 148 Jiang, F., Speziale, S. and Duffy, T. S. (2004) Single-crystal elasticity of grossular-and almandine-rich garnets to 11 GPa by Brillouin scattering. Journal of Geophysical Research, 109, 10.1029/2004JB003081. Karason H. and van der Hilst, R. D. (2000) Constraints on mental convection from seismic tomography, In Richards, M. A., Gordon R. G. and van der Hilst, R. D., editors, History and dynamics of global plate motion, volume 121 of Geophysical Monograph Series, p. 277-288, American Geophysical Union, Washington. Karato, S-I and Jung, H. (1998) Water, partial melting and the origin of the low velocity and high attenuation zone in the upper mantle. Earth and Planetary Science Letters, 157(3-4): 193-207. Katsura, T. and Ito, E. (1989) The system Mg2SiO4-Fe2SiO4 at high pressures and temperatures: Precise determination of stabilities of olivine, modified spinel, and spinel. Journal of Geophysical Research, 94(B11): 15663-15670. Katsura, T., Yamada, H., Shinmei, T., Kubo, A., Ono, S., Kanzaki, M., Yoneda, A., Walter, M. J., Ito, E., Urakawa, S., Funakoshi, K. and Utsumi, W. (2003) Post-spinel transition in Mg2SiO4 determined by high P-T in situ X-ray diffractometry. Physics of the Earth and Planetary Interiors, 136: 11-24. Kawai, N. and Endo, S. (1970) The generation of ultrahigh pressures by a split sphere apparatus. Review of scientific Instruments, 41: 1178-1181. Kennett, B. L. N. and Engdahl, E. R. (1991) Travel times for global earthquake location and phase identification. Geophysical Journal International, 105: 429–465. Kennett, L. N., Engdahl, E. R. and Buland, R. (1995) Constraints on the velocity structure in the Earth from travel times. Geophysical Journal International, 122: 108–124. Keppler, H. and Frost, D. C. (2005) Introduction to minerals under extreme conditions. In Miletich, R., (editor), Mineral behaviour at extreme conditions, volume 7 of European Mineralogical Union Notes in Mineralogy, p. 17, Eötvös University press, Budapest. Kiefer, B., Stixrude, L., Hafner, J. and Kresse, G. (2001) Structure and elasticity of wadsleyite at high pressures. American Mineralogist, 86: 1387-1395. 149 King, H. E. and Finger, L. W. (1979) Diffracted beam crystal centering and its application to highpressure crystallography. Journal of Applied Crystallography, 12: 374-378. Kleppa, O. J. (1976) Mineralogical applications of high–temperature reaction calorimetry. In Strens, R. G. J., editor, The physics and chemistry of Rocks and Minerals, p. 369-387, New York, Wiley Knittle, E. and Jealoz, R. (1987) Synthesis and equation of state of (Mg,Fe)SiO3 perovskite to over 100 GPa. Science, 235: 688-670. Kubo, A., Yagi, T., Ono, S. and Akaogi, M. (2000) Compressibilty of Mg0.9Al0.2Si0.9O3 perovskite. Proceedings of Japan Academy, 76B: 103-107. Kubo, T., Eiji Ohtani, E., Kondo, T., Kato, T., Toma, M., Hosoya, T., Sano, A., Takumi Kikegawa ,T. and Nagase, T. (2002) Metastable garnet in oceanic crust at the top of the lower mantle. Nature, 420: 803-806. Lauterbach, S., McCammon, C. A., van Aken, P., Langenhorst, F. and Seifert, F. (2000) Mössbauer and ELNES spectroscopy of (Mg,Fe)(Si,Al)O3 perovskite: a highly oxidised component of the lower mantle. Contributions to Mineralogy and Petrology, 138: 17-26. Libermann, R. C. and Wang, Y. (1992) Characterisation of sample environment in a uniaxial split sphere apparatus. In Syono, Y. and Manghnani, M. H., editors, High Pressure Research: Application to Earth and Planetary Sciences, p. 19–31,Terra Scientific Publishing Co., Tokyo/American Geophysical Union, Washington, DC. Li, B., Liebermann, R. C. and Weidner, D. J. (1998) Elastic Moduli of Wadsleyite (β-Mg2SiO4) to 7 Gigapascals and 873 Kelvin. Science, 281(5377): 675-677. Li, B. and Li, B. (2003) Compressional and shear wave velocities of ringwoodite {gamma}-Mg2SiO4 to 12 GPa. American Mineralogist, 88: 1312-1317. Li, L., Brodholt, J. P., Stackhouse, S., Weidner, D. J., Alfredsson, M. and Price, G. D. (2005) Elasticity of (Mg,Fe)(Si,Al)O3perovskite at high pressure. Earth and Planetary Science Letters, 240: 529-536. 150 Li, L., Weidner, D.J., Brodholt, J., Alfè, D., Price, G. D, Caracas, R. and Wentzcovitch, R. (2006) Elasticity of CaSiO3 perovskite at high pressure and high temperature. Physics of Earth and Planetary Interiors, 155: 249–259. Litasov, K. D. and Ohtani, E. (2005) Phase relations in hydrous MORB at 18-26 GPa: Implications for heterogeneity of the lower mantle. Physics of Earth and Planetary Interiors, 150: 239-263. Mao, H. K., Hemley, R. J., Fei, Y., Shu, J. F., Chen, L. C., Jephcoat, A. P. and Wu, Y. (1991) Effect of pressure, temperature and composition on lattice parameters and density of (Fe,Mg)SiO3 peovskites to 30 Gpa. Journal of Geophysical Research, 96(B5): 8069-8079. Matsuzaka, K., Akaogi, M., Suzuki, T. and Suda, T. (2000). Mg-Fe partitioning between silicate spinel and magnesiowüstite at high pressure: experimental determination and calculation of phase relations in the system Mg2SiO4-Fe2SiO4. Physics and Chemistry of Minerals, 27: 310-319. McCammon, C. (1997) Perovskite as a possible sink for ferric iron in the lower mantle. Nature, 387: 694-696. McCammon, C. (1998) The crystal chemistry of ferric iron in Fe0.05Mg0.95SiO3 perovskite as determined by Mössbauer spectroscopy in the temperature range 80-293K. Physics and Chemistry of Minerals, 25: 292-300. McCammon, C. (2004) Mössbauer spectroscopy: applications, in Spectroscopic methods. In Beran, A. and Libowitzky, E., editors, Spectroscopic Methods in Mineralogy, volume 6 of European Mineralogical Union Notes in Mineralogy, p. 369-398, Eötvös University Press, Budapest. McCammon, C. A., Lauterbach, S., Seifert, F., Langenhorst, F., and van Aken, P. A. (2004) Iron oxidation state in lower mantle assemblages I Empirical relations derived from high–pressure experiments. Earth and Planetary Science Letters, 222: 435-449. McDonough, W. F. and Sun, S. S. (1995) The composition of the Earth. Chemical Geology, 20: 223253. Megaw, H. D. (1973) Crystal structure: a working approach, Philadelpheia, Saunders. 151 Meng, Y., Fei, Y., Weidner, D. J., Gwanmesia, G. D. and Hu, J. (1994) Hydrostatic compression of γMg2SiO4 to mantle pressures and 700 K: Thermal equation of state and related thermoelastic properties. Physics and Chemistry of Minerals, 21: 407-412. Miletich, R., Allan, D. R. and Kuhs, W. F. (2000) High-pressure single-crystal techniques. In Hazen, R. M. and Downs, R. T., editors, High-Temperature and High–Pressure Crystal Chemistry, volume 41 of Reviews in Mineralogy and Geochemistry, p. 445-519, Mineralogical Society of America, Chantilly, Virginia. Miletich, R., (2005) Diffraction techniques: Shedding light on structural changes at extreme conditions. In Miletich, R., editor, Mineral Behaviour at Extreme conditions, in volume 7 of European Mineralogical Union Notes in Mineralogy, p. 281-331, Eötvos University Press, Budapest. Murakami, M., Hirose, K., Kawamura, K., Sata, N., Ohisi, Y. (2004) Post perovskitephase transition in MgSiO3, Science, 304(5672): 855-858. Nasdala, L., Smith, D. C., Kaindal, R. and Zieman, M. A. (2004) Raman Spectroscopy: Analytical Perspectives in Mineralogical Research. In Beran, A. and Libowitzky, E., editors, Spectroscopic Methods in Mineralogy, volume 6 of European Mineralogical Union Notes in Mineralogy, p. 281-243, Eötvös University Press, Budapest. Navrotsky, A. (1997) Progress and New directions in High temperature calorimetry revisited. Physics and Chemistry of Minerals, 24: 22-241. Navrotsky, A. (1999) A lesson from ceramics, Science, 284: 1788-1789. Navrotsky, A., Schoenitz, M., Kojitani, H., Xu, H., Zhang, J., Weidner, D. J., Jeanloz and R. (2003) Aluminium in magnesium silicate perovskite: formation, structure, and energetics of magnesium–rich defect solid solutions. Journal of Geophysical Research, 108(B7), 2330, doi: 10.1029/2002JB002055. Navrotsky, A. (2004) Thermochemistry, energetic modeling and systematics. In Gramaccioli, C, editor, Energetic modeling in minerals, volume 4 of European Mineralogical Union Notes in Mineralogy, p. 5-31, Eötvös University Press, Budapest. 152 Newton, R. C., Charlu, T. V. and Kleppa, O. J. (1977) Thermochemistry of high pressure garnets and clinopyroxenes in the system CaO-MgO-Al2O3-SiO2. Geochimica et Cosmochimica Acta, 41(3): 369377. Nishihara, Y. and Takahashi, E. (2001) Phase relation and physical properties of an Al-depleted komatiite to 23 GPa. Earth and Planetary Science Letters, 190: 65-77. Nishio-Hammane, D., Nagai, T., Fujino, K., Seto, Y., Takafuji, N. (2005) Fe3+ and Al solubilities in MgSiO3 perovskite: implications of the Fe3+AlO3 substitution in MgSiO3 perovskite at the lower mantle conditions. Geophysical Research Letters, 32:L16306, doi: 10.1029/2005GL023529. Ono, S., Kikegawa, T. and Iizuka, T. (2004) The equation of state of orthorhombic perovskite in a pyrolitic mantle upto 80 GPa: Implications for chemical composition of the lower mantle, Physics of the Earth and Planetary Interiors, 145(1-4): 9-17. O’Neill, H. St. C. and Wood, B. J. (1979) An experimental study of Fe-Mg partitioning between garnet and olivine and its calibration as a geothermometer. Contributions to Mineralogy and Petrology, 70: 59–70. Ralph, R. L. and Finger, L. W. (1982) A computer program for refinement of crystal orientation on matrix and lattice constants from diffractometer data with lattice symmetry constraints. Journal of Applied Crystallography, 15: 537-539. Reed, S. (1996) Electron Microprobe Analysis and Scanning Electron Microscopy in Geology. Cambridge University Press, Cambridge, p. 187. Richmond, N. C. and Brodholt, J. P. (1998) Calculated role of aluminium in the incorporation of ferric iron into magnesium silicate perovskite. American Mineralogist, 83(9-10): 947-951. Ringwood, A. E. (1967) The pyroxene-garnet transformation in the Earth’s mantle. Earth and Planetary Science Letters, 2: 255-263. Ringwood, A. E. and Major, A. (1967) Some high-pressure transformations of geophysical significance. Earth and Planetary Science Letters, 2, 106-110. Ringwood, A. E. (1975) Composition and petrology of the Earth’s mantle. McGraw-Hill, New York, p. 618. 153