Journal of Petrology Advance Access originally published online on January 7, 2005
Journal of Petrology 2005 46(4):749-781; doi:10.1093/petrology/egh096
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Phase Relations and Stability of Magnetoplumbite- and Crichtonite-Series Phases under Upper-Mantle P–T Conditions: an Experimental Study to 15 GPa with Implications for LILE Metasomatism in the Lithospheric Mantle
1 INSTITUT FÜR MINERALOGIE UND PETROGRAPHIE, UNIVERSITÄT INNSBRUCK, INNRAIN 52, A-6020 INNSBRUCK, AUSTRIA
2 DEPARTMENT OF MECHANICAL AND MATERIALS ENGINEERING, FLORIDA INTERNATIONAL UNIVERSITY, 10555 WEST FLAGLER STREET, MIAMI, FL 33175, USA
3 BAYERISCHES GEOINSTITUT, UNIVERSITÄT BAYREUTH, UNIVERSITÄTSSTRASSE 30, D-95447 BAYREUTH, GERMANY
RECEIVED FEBRUARY 26, 2004; ACCEPTED NOVEMBER 9, 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL AND ANALYTICAL...PREVIOUS EXPERIMENTAL WORKRESULTSDISCUSSIONSUMMARY OF CONCLUSIONSREFERENCES |
|---|
High-pressure–high-temperature experiments were performed in the range 7–15 GPa and 1300–1600°C to investigate the stability and phase relations of the K- and Ba-dominant members of the crichtonite and magnetoplumbite series of phases in simplified bulk compositions in the systems TiO2–ZrO2–Cr2O3–Fe2O3–BaO–K2O and TiO2–Cr2O3–Fe2O3–BaO–K2O. Both series of phases occur as inclusions in diamond and/or as constituents of metasomatized peridotite mantle xenoliths sampled by kimberlites or alkaline lamprophyres. They can accommodate large ion lithophile elements (LILE) and high field strength elements (HFSE) on a wt % level and, hence, can critically influence the LILE and HFSE budget of a metasomatized peridotite even if present only in trace amounts. The Ba and K end-members of the crichtonite series, lindsleyite and mathiasite, are stable to 11 GPa and 1500–1600°C. Between 11 and 12 GPa, lindsleyite breaks down to form two Ba–Cr-titanates of unknown structure that persist to at least 13 GPa. The high-pressure breakdown product of mathiasite is a K–Cr-titanate with an idealized formula KM7O12, where M = Ti, Cr, Mg, Fe. This phase possesses space group P63/m with a = 9·175(2) Å, c = 2·879(1) Å, V = 209·9(1) Å3. Towards high temperatures, lindsleyite persists to 1600°C, whereas mathiasite breaks down between 1500 and 1600°C to form a number of complex Ti–Cr-oxides. Ba and K end-members of the magnetoplumbite series, hawthorneite and yimengite, are stable in runs at 7, 10 and 15 GPa between 1300 and 1400°C coexisting with a number of Ti–Cr-oxides. Molar mixtures (1:1) of lindsleyite–mathiasite and hawthorneite–yimengite were studied at 7–10 GPa and 1300–1400°C, and 9–15 GPa and 1150–1400°C, respectively. In the system lindsleyite–mathiasite, one homogeneous Ba–K phase is stable, which shows a systematic increase in the K/(K + Ba) ratio with increasing pressure. In the system hawthorneite–yimengite, two coexisting Ba–K phases appear, which are Ba rich and Ba poor, respectively. The data obtained from this study suggest that Ba- and K-dominant members of the crichtonite and magnetoplumbite series of phases are potentially stable not only throughout the entire subcontinental lithosphere but also under conditions of an average present-day mantle adiabat in the underlying asthenosphere to a depth of up to 450 km. At still higher pressures, both K and Ba may remain stored in alkali titanates that would also be eminently suitable for the transport of other ions with large ionic radii.
KEY WORDS: crichtonite; magnetoplumbite; high-P–T experiments; phase relations; upper mantle
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...PREVIOUS EXPERIMENTAL WORKRESULTSDISCUSSIONSUMMARY OF CONCLUSIONSREFERENCES |
|---|
The study of peridotite xenoliths sampled by magmas originating from upper-mantle depths has shown that metasomatism is a widespread phenomenon in the subcontinental mantle lithosphere (e.g. Kempton, 1987
; Menzies et al., 1987, and references therein) and is likely to be a prerequisite for the generation of deep-seated incompatible element enriched rocks such as kimberlites, lamproites and alkali basalts (Fraser et al., 1985; Tainton & McKenzie, 1994). More recently, metasomatism involving Ti phases has also been documented in oceanic upper-mantle peridotites sampled by hotspot magmas (Grégoire et al., 2000a, 2000b). Metasomatism is often characterized by the addition of water and/or incompatible high field strength elements (HFSE) and large ion lithophile elements (LILE), which may lead to the crystallization of both silicate and oxide phases not previously present in the peridotitic upper mantle (see Harte, 1987). The most commonly observed metasomatic silicates are phlogopite and amphibole, which are able to store high concentrations of elements with large ionic radii, such as Ba, Sr, Rb, Cs and Pb, in highly coordinated lattice positions. In many cases, (Fe)Ti-oxides are part of the metasomatic assemblages. The most common metasomatic oxides are rutile and/or ilmenite, which are characteristic of Fe–Ti metasomatism typically found in mantle xenoliths entrained in alkali basalts, but also as part of kimberlitic megacryst suites (e.g. Mitchell, 1977; Menzies et al., 1987; Schulze et al., 1995). Significant amounts of rutile and ilmenite may also be present in metasomatized mantle rocks sampled by kimberlites and lamproites (Dawson & Smith, 1977; Harte et al., 1987) and both phases are present as inclusions in diamond (e.g. Sobolev & Yefimova, 2000, and references therein). In rare cases, armalcolite, which is a member of the pseudobrookite FeTi2O5–MgTi2O5 series, may join the metasomatic oxide assemblage in both subcontinental (e.g. Jones et al., 1982; Haggerty, 1983; Jaques et al., 1990; Varlamov et al., 1996) and suboceanic (Grégoire et al., 2000b) lithospheric settings. A typical feature of these metasomatic (Fe)Ti-oxides is variable and, in part, extremely high concentrations of Cr and HFSE such as Zr, Hf, Nb and Ta, reaching wt % levels (e.g. Mitchell, 1977; Tollo & Haggerty, 1987; Wang et al., 1999).
A third group of metasomatic oxides that has been reported so far from subcontinental mantle settings only is alkali-bearing titanates of the crichtonite and magnetoplumbite series. These phases are characterized by a layer structure and a highly complex composition (Townes et al., 1967; Grey et al., 1976; Obradors et al., 1985). K–Ba-rich members of both series are present as part of a suite of Ti-oxides involving Cr–Nb-rich rutile, ilmenite, Cr–Ti-spinel and, more rarely, armalcolite (Fe,Mg)Ti2O5 in xenoliths and/or as xenocrysts sampled by kimberlites or lamprophyres (Jones et al., 1982; Haggerty et al., 1983, 1989; Zhou et al., 1984; Erlank et al., 1987; Nixon & Condliffe, 1989; Kostrovitsky & Garanin, 1991; Varlamov et al., 1996; Kiviets et al., 1998; Wang et al., 1999; Bulanova et al., 2004). The occurrence of Cr–Ti oxides with compositions indicative of crichtonite and magnetoplumbite phases as inclusions in diamond has been reported by Sobolev et al. (1988, 1997), Leost et al. (2003) and Bulanova et al. (2004), and testifies to the high-pressure stability of these phases. Textures of oxide phase intergrowths (Haggerty et al., 1983; Nixon & Condliffe, 1989; Haggerty, 1991; Velde, 2000) indicate that phases of the crichtonite and magnetoplumbite series have formed through metasomatic alteration of pre-existing Ti–Cr oxide-bearing assemblages by LILE–HFSE–rare earth element (REE)-rich fluids or melts in subcontinental lithospheric settings. The very high concentrations of LILE, HFSE and light REE (LREE) (up to 103–104 times chondritic abundance) in addition to potassium (Haggerty, 1983, 1987; Jones & Ekambaram, 1985; Konzett et al., 2000; Grégoire et al., 2002) found in crichtonite and magnetoplumbite phases, and a pressure stability extending into the stability field of diamond (Foley et al., 1994), make these phases potentially important reservoirs for a large number of incompatible trace elements in the source regions of deep-seated alkaline magmas, such as kimberlites and lamproites (Foley, 1992a, 1992b; Mitchell, 1995), even if present only in minute amounts. Ti-oxides other than crichtonite and magnetoplumbite series phases that are reported to contain significant amounts of alkali elements include (in very rare cases) armalcolite (see Jaques et al., 1990), priderite K2M2+Ti7O16–K2M3+2Ti6O16 (e.g. Mitchell & Lewis, 1983; Mitchell, 1995) and loparite [(Na,K)0·5REE0·5TiO3]–tausonite (SrTiO3)–perovskite (CaTiO3) (LTP) solid solutions. Both priderite and LTP solid solutions have been described as inclusions in diamond (Jaques et al., 1989; Kopylova et al., 1997). It should be noted that apart from their potential as hosts of incompatible (trace) elements in the mantle, phases with crichtonite structure have received attention because of their suitability as storage media for high-level radioactive waste, especially radionuclides with large ionic radii (Krauskopf, 1988).
This study was undertaken to constrain the P–T stability field of K–Ba-dominant members of the crichtonite and magnetoplumbite series and to identify and characterize their breakdown products. This makes it possible (1) to assess whether crichtonite and magnetoplumbite phases are restricted to the source regions of lithospheric magmas, such as lamproites or alkaline lamprophyres (Fraser et al., 1985; Mitchell & Bergman, 1991; Foley, 1992a, 1992b) or may be part of deeper asthenospheric enriched sources for group I kimberlites (Mitchell, 1995), and (2) to assess the potential of these phases as carriers of incompatible trace elements (LILE) as an alternative to, or outside the stability fields of, more common silicate hosts such as micas or amphiboles.
Classification of crichtonite and magnetoplumbite phases
Phases of the crichtonite and magnetoplumbite structural series are characterized by the general formulae AM21O38 and AM12O19, respectively (Adelsköld, 1938; Rouse & Peacor, 1968). The A- and M-sites are occupied by large and small cations, respectively, and the classification is based on the dominant A-site cation. In the case of the crichtonite series this may be Sr (crichtonite), Ca (loveringite), Na (landauite), U + REE (davidite), Pb (senaite), Ba (lindsleyite), or K (mathiasite) (Haggerty, 1983, and references therein). In the magnetoplumbite series, the Pb-, K-, and Ba-dominant members are magnetoplumbite, yimengite and hawthorneite or haggertyite (Zhou et al., 1984; Grey et al., 1987, 1998; Haggerty et al., 1989). In both series, the M-site may be occupied by Ti, Cr, Fe, Mg, Zr, Nb, V or Zn, with Ti usually accounting for >50% of the M formula position (Haggerty, 1987). The crichtonite structure (rhombohedral R-3) is based on a nine-layer close-packed anion lattice in which the A- and M-cations occupy 12-coordinated and 8- or 6-coordinated positions, respectively (Grey et al., 1976; Gatehouse et al., 1983). Magnetoplumbite (hexagonal P63/mmc) is isostructural with barium ferrite BaFe12O19 (Townes et al., 1967; Obradors et al., 1985), and is composed of alternating spinel-like slabs (M12O16) with the M-cations in 4- or 6-coordination, and a perovskite-like layer (AMO3) with the A-cation in 12-fold coordination (Grey et al., 1987).
Natural crichtonite- and magnetoplumbite phases from upper-mantle settings usually show an excess of total cations when normalized to 38 or 19 oxygens, along with a considerable variation in their A- and M-site cation sums (see below), indicating the ability of these structures to accommodate a significant amount of vacancies. Some titanate analyses reported in the literature as crichtonite-group phases combine the high TiO2 contents (>65 wt %) and very low Cr2O3 contents (
1–2 wt %) typical of pseudobrookite with high concentrations (>1–2 wt %) of LILE and alkali cations such as Ba, Sr, K and Na, the latter typical of crichtonite (e.g. Wang et al., 1999). On the other hand, titanate analyses are reported that show the range in TiO2 and Cr2O3 typical for crichtonite combined with extremely low concentrations of A-site cations (e.g. Varlamov et al., 1996). In the absence of crystal structure data, however, an assignment of phases with these ‘hybrid’ compositions to a specific group of titanate phases should be made with caution.
|
EXPERIMENTAL AND ANALYTICAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...PREVIOUS EXPERIMENTAL WORKRESULTSDISCUSSIONSUMMARY OF CONCLUSIONSREFERENCES |
|---|
Composition and preparation of starting materials
Six starting compositions were used in the present study, representing simplified end-member compositions of lindsleyite (LI), mathiasite (MA), hawthorneite (HA), yimengite (YI) and 50:50 lindsleyite–mathiasite (LIMA) and hawthorneite–yimengite (HAYI) solid solutions with all Fe as Fe3+ (Fetot = Fe3+), following Foley et al. (1994)
. The structural formulae of the starting materials are as follows:
- mathiasite K(Ti13Cr4FeZrMg2)O38;
- lindsleyite Ba(Ti12Cr4Fe2ZrMg2)O38;
- yimengite K(Ti3Cr5Fe2Mg2)O19;
- hawthorneite Ba(Ti3Cr4Fe2Mg3)O19;
- LIMA (Ba0·5K0·5)(Ti13Cr3·5FeZrMg2·5)O38;
- YIHA (Ba0·5K0·5)(Ti3Cr4·5Fe2Mg2·5)O19.
- lindsleyite Ba(Ti12Cr4Fe2ZrMg2)O38;
|
Experimental and analytical conditions
The starting materials (Table 1) were synthesized from high-purity (
99·99%) oxides following the procedure described by Foley et al. (1994). Experiments (Table 2) were performed with 600 t and 500 t Walker-type multi-anvil presses at the Geophysical Laboratory (GL) of the Carnegie Institution of Washington and the Bavarian Research Institute of Experimental Geochemistry and Geophysics, University of Bayreuth (BGI), respectively. Assembly sizes and materials are as follows: at the GL: P 10 GPa: 18/11 (18 mm edge-length of octahedron, 11 mm truncation of WC cubes) assemblies with octahedra cast from an MgO-based ceramic (Walker, 1991) and a graphite furnace; P > 10 GPa: 10/5 assemblages with precast MgO octahedra combined with pyrophyllite gaskets and a rhenium furnace (see Bertka & Fei, 1997, fig. 2a and b); at the BGI: P 10 GPa: 18/11 assemblies with precast MgO octahedra combined with pyrophyllite gaskets and a graphite furnace; P > 10 GPa: 10/5 assemblages with precast MgO octahedra combined with pyrophyllite gaskets and a LaCrO3 furnace. At the BGI, all experiments were conducted with precast MgO octahedra in combination with pyrophyllite gaskets. In all experiments, temperatures were measured with W3%Re–W25%Re thermocouples and both pressure and temperature were computer-controlled during the entire duration of the runs. Detailed descriptions of the GL and BGI experimental and calibration procedures have been given by Rubie et al. (1993) and Bertka & Fei (1997).
|
|
Starting materials were pre-dried at 150°C overnight, placed in 1·18 or 2·00 mm outer diameter Pt100 capsules and welded shut immediately. Initial experiments with this capsule material resulted in an almost complete loss of Fe to the capsule walls. Therefore, an additional inner graphite capsule (high-purity graphite ring with compressed graphite powder as bottom and lid) was used in subsequent experiments (see Table 2 and Fig. 1) to minimize Fe loss. Two experimental charges were placed in individual capsules and one capsule (10/5 assemblies) or in some cases two capsules (18/11 assemblies) were placed in a single assembly with the combined height of both capsules not exceeding 3·5 mm (Fig. 1). Attempts to place four capsules with 1·0 mm outer diameter in an 18/11 assembly were abandoned after two successful runs (JKW84, JKW43) because of problems in handling assembly parts. After the runs, recovered Pt capsules were embedded in epoxy resin and ground to expose the centre of the charges for microprobe analysis, Raman spectrometry and extraction of crystals for structure analysis. In runs at >10 GPa, the inner graphite capsule partially transformed to diamond.
|
Phase compositions were analysed by electron microprobe using a JEOL superprobe at the GL and a Cameca SX100 at the University of Vienna. Analytical conditions were 50 nA and 15 kV with a residence time t of 20–30 s on peak and t/2 s on backgrounds of the X-ray lines. Ba and Ti were analysed using the L
and K lines, respectively, on a LiFH crystal. Raw counts were corrected online with the ZAF correction procedure. In the absence of information on the Fe2+/Fe3+ ratios, Fetot is quoted as Fe3+ as used in the starting materials.
Phases were identified based on their chemical composition and, in part, on single-crystal structure analysis. Identification of crichtonite and magnetoplumbite phases was straightforward because of their distinctive stoichiometry and similarity to both natural counterparts and experimental run products that were identified by powder diffraction (Foley et al., 1994). An identification of the phases coexisting with crichtonite and magnetoplumbite, however, often turned out to be impossible because of (1) the small grain size and modal amounts prohibiting single crystal analysis and (2) a lack of data on high-P–T phase relations and compositions of oxides in the system TiO2–Cr2O3–MgO–FeO. Even in simple systems such as TiO2–Cr2O3 or TiO2–ZrO2–Cr2O3–Fe2O3, phase relations are complex even at low pressures, with a multitude of distinctive stoichiometries and the possibility to form polysomatic series (see Flörke & Lee, 1970; Grey et al., 1973). Thus, it has to be kept in mind that some of the Ti–Cr–Fe–Mg oxides listed as distinctive phases in the present study may not be phases with distinctive stoichiometry and structure, but may actually represent members of a single solid solution series with variable composition. Therefore, the number of distinctive structure types of phases coexisting with crichtonite and magnetoplumbite may actually be lower than the number of unidentified phases labelled U1–U9 described below and listed in Tables 4–7.
|
|
|
|
Single-crystal X-ray diffraction (XRD) data were collected on a Bruker Smart CCD diffractometer equipped with graphite-monochromatized MoKa radiation (
= 0·71069 Å). A hemisphere of three-dimensional XRD data (0° < 2
< 54°) was collected with frame widths of 0·3° in 
and 30 s counting time per frame. The XRD data were analysed to locate peaks for the determination of the unit-cell parameters. The structure of a K–Cr-titanate crystal from JKW86 was solved and refined with SHELX97. Unpolarized Raman spectra were recorded with a DilorXY spectrometer equipped with confocal optics and a Peltier-cooled CCD detector. A microscope with 100x objective was used to focus a He–Ne laser beam with 632 nm excitation wavelength onto the polished surface of experimental charges. Accumulation times from 10 to 180 s were used with a laser power of 4–8 mW to avoid beam damage. Initial attempts to record the Raman spectra of crichtonite and magnetoplumbite phases with a 514 nm Ar+ laser resulted in an instantaneous thermal destruction of these phases.
|
PREVIOUS EXPERIMENTAL WORK |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...PREVIOUS EXPERIMENTAL WORKRESULTSDISCUSSIONSUMMARY OF CONCLUSIONSREFERENCES |
|---|
Both crichtonite and magnetoplumbite phases are stable at 1 atm (e.g. Chase & Wolten, 1965
; Kohn & Eckart, 1965; Peterson & Grey, 1995) and, in the case of magnetoplumbite phases, represent industrially important ferrimagnetic materials such as barium ferrite BaFe12O19 (Obradors et al., 1985). By comparison, only two studies have dealt with the high-pressure stability of crichtonite and magnetoplumbite phases. Podpora & Lindsley (1984) synthesized lindsleyite and mathiasite with complex natural compositions [as given by Haggerty et al. (1983)] at 2·0 GPa and 1300°C, and 2·2 GPa and 900°C. Foley et al. (1994) demonstrated the stability of end-member lindsleyite, mathiasite, yimengite, hawthorneite and solid solutions of lindsleyite–mathiasite (LIMA) and hawthorneite–yimengite (HAYI) in the system TiO2–ZrO2–Cr2O3–Fe2O3–MgO–BaO–K2O in a P–T range 3–5 to 5·0 GPa and 1150–1350°C. |
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...PREVIOUS EXPERIMENTAL WORKRESULTSDISCUSSIONSUMMARY OF CONCLUSIONSREFERENCES |
|---|
Textures and chemical homogeneity of the phases
At P
10 GPa, crichtonite and magnetoplumbite phases usually form subhedral to rounded or lath-shaped grains with a size not exceeding 
50 µm (Fig. 2) that show strong orientation contrast in back-scattered electron (BSE) imaging (e.g. Fig. 2g). Textural equilibration is often indicated by smooth grain boundaries with regular triple grain junctions (e.g. Fig. 2a) giving a polygonal granoblastic or granuloblastic array. At pressures >10 GPa, the average grain size decreases to <20 µm and the textures become less equilibrated, with more xenomorphic grain shapes and irregular grain boundaries irrespective of run durations. All experiments except JKW78 are considered subsolidus. In JKW78 at 8·8 GPa and 1600°C, thin interstitial films along grain boundaries and irregular patches with K-rich composition were taken as an indication of the presence of small amounts of melt (Fig. 2b).
|
Within runs, the phase distribution is usually homogeneous. Apart from magnetoplumbite and crichtonite phases, a number of additional phases appeared in all experiments (see below). Within the stability fields of magnetoplumbite and crichtonite phases, their presence can be explained by shifts in the bulk composition as a result of: (1) partial reduction of Fe3+ in the starting material to Fe2+; (2) loss of Fe to the Pt capsule and/or of alkalis to the graphite liner. The latter effect was observed despite careful drying of the starting materials and the graphite liners, and may also occur in alkali-rich systems at high P and T (see Konzett & Ulmer, 1999
). These phases commonly form rounded to subhedral but in part euhedral grains (Fig. 2) interstitial between magnetoplumbite and crichtonite phases, and their modal amount may range from 20 to 30% to only a few grains. In some cases, they show a more zonal distribution with enrichment along the graphite–charge interface, indicating T or fO2 gradients in the capsule. Because experiments of the present study are synthesis experiments and because of the notorious sluggishness of oxide equilibration, the metastable nature of some of the phases coexisting with crichtonite and magnetoplumbite cannot be ruled out.
The compositional variation of phases within runs is generally larger than that encountered in high-pressure experiments involving silicate systems, because of the sluggishness of equilibration of the oxide phases and the absence of water. The strongest within-sample variation is observed for Cr, Ti and Fe, with up to 4% relative variation with respect to the averaged values. With the exception of rutile and TiO2-II (the -PbO2-structured TiO2 modification), individual grains rarely show significant compositional zoning. Rutile or TiO2-II may show irregular cores with variable ZrO2 concentrations. Wherever possible, these cores were avoided during analysis. Sometimes, however, the small grain size of rutile made a partial beam overlap with these cores inevitable, which contributed to the scatter in the data (see below). It should be noted that no relation was found between the degree of textural equilibration and the homogeneity of phases. In the K2O–BaO-bearing bulk systems, irregular variations in K2O and BaO contents of solid-solution phases may be present. Traces of unreacted starting material are frequently present as small remnants of ZrO2 in the cores of ZrTiO4 or Zr phase grains. In a few cases (e.g. JKW88), small irregular remnants of unreacted MgO and/or BaTiO3 were found in runs at P > 10 GPa.
Phase relations of mathiasite, lindsleyite and LIMA solid solution
In bulk composition I, mathiasite is stable at pressures between 7 and 11 GPa and temperatures between 1300° and 1500°C (Fig. 3a and Table 2). Towards high pressures, mathiasite breaks down continuously to form a K–Cr-titanate that appears at 8·8 GPa and 1500°C and coexists with mathiasite to a pressure of 11 GPa (see Konzett & Yang, 2001). In runs at 12 and 13 GPa, mathiasite is no longer stable and is replaced by the K–Cr-titanate as the only carrier of potassium. The temperature stability of mathiasite extends to 1500°C at 8·8 GPa. At this pressure, mathiasite is replaced at 1600°C by K–Cr-titanate and small (<5%) amounts of a K-rich interstitial phase that is interpreted as quenched melt (Fig. 2b). Phases coexisting with mathiasite are rutile–TiO2-II, ZrTiO4, a Zr–Cr-titanate with a fluorite-derived structure (Konzett & Yang, 2001) and a number of unidentified Ti–Cr oxides.
|
In bulk composition II, lindsleyite is stable between 7 and 11 GPa in the temperature range 1300–1600°C (Fig. 3b and Table 2). High-pressure breakdown products of lindsleyite are two unidentified Ba–Cr-titanates (Table 4) that first appear at 11 GPa and 1400°C coexisting with lindsleyite and completely replace lindsleyite at 12 GPa and 1400°C. Unlike mathiasite, lindsleyite is still stable in runs at 1600°C. Phases coexisting with lindsleyite are rutile–TiO2-II and ZrTiO4 in addition to two unidentified Cr-titanates and the Zr–Cr-titanate also observed in bulk composition I.
Lindsleyite–mathiasite solid solutions were synthesized in the P–T range 7–10 GPa and 1300–1400°C from bulk composition V, to compare with the solid solution behaviour of the Ba–K-dominant members of the magnetoplumbite series (Foley et al., 1994). Phases coexisting with LIMA are rutile–TiO2-II and ZrTiO4 in all runs (Fig. 2d). At 9 and 10 GPa, additional phases are present in modal amounts <2%. Only a few grains were identified even after microprobe element mapping of the entire charge. One of these phases present at 9 GPa is K rich and strongly resembles yimengite in terms of both oxide wt % and cation numbers (see Tables 5 and 6). Two other Cr–Fe–Mg-titanate phases, U4 and U5 (Table 5), remain unidentified. Because of the very small modal amounts of these phases, it is possible that they might be present in other runs using bulk composition V but were overlooked as only a single section plane through the experimental charges was examined. One run at 13 GPa and 1300°C produced very small (<5 µm) and inhomogeneous grains of a K–Ba phase interstitial between abundant rutile and ZrTiO4. Despite the inhomogeneity and the likeliness of beam overlap with adjacent phases, analyses of the K–Ba phase consistently yielded only 43–46 wt % TiO2, which is significantly lower than the TiO2 contents of LIMA solid solutions (see below). This was taken as an indication that LIMA solid solution is no longer stable at 13 GPa and 1300°C.
Phase relations of yimengite, hawthorneite and YIHA solid solutions
Yimengite and hawthorneite were synthesized from bulk compositions III and IV at 7, 10 and 15 GPa at temperatures between 1300 and 1400°C (Fig. 2f and g). In all runs (see Table 2), one or two additional phases were found coexisting with yimengite and hawthorneite with modal abundances of <3 to 20 vol. % as determined by image analysis (Tables 6 and 7).
Attempts to synthesize 50:50 yimengite–hawthorneite solid solutions using bulk composition VI failed in the pressure range 9–15 GPa. Instead, two Ba–K phases were found in all runs (see Fig. 2e), which is consistent with results obtained by Foley et al. (1994). Those workers also reported two coexisting Ba–K phases in this system at 4·3 and 5·0 GPa and 1150–1350°C, and took this finding as an indication of the presence of a solvus in the system yimengite–hawthorneite. In all runs of this study using bulk composition VI, one or two phases appear in addition to the two yimengite–hawthorneite solid solutions (Fig. 2e and Table 7).
Mineral chemistry
Crichtonite phases
When normalized to 38 oxygens and assuming all Fe as Fe2O3 (Fetot = Fe2O3), almost all crichtonite analyses obtained from the experiments of the present study show a deviation from the ideal AM21O38 stoichiometry in terms of total cation as well as A-site and M-site cation sums. This variation is most pronounced for lindsleyite with 
cations, A-cations and M-cations in the range 21·55–22·29, 0·84–1·28 and 20·39–21·40 (n = 51), respectively. The least variation was observed for experimental LIMA analyses, which cluster tightly around the ideal stoichiometric values (Fig. 4). Except for lindsleyite analyses, no correlation can be observed between pressure and the degree of non-stoichiometry. Lindsleyite shows a significant A-site cation surplus of 1·21 ± 0·04 and 1·18 ± 0·06 atoms per formula unit (a.p.f.u.) along with an M-site deficiency of 20·58 ± 0·03 and 20·43 ± 0·03 a.p.f.u. at 10 GPa and 1600°C and 11 GPa and 1600°C compared with lower-temperature runs that show A-site deficiency and M-site surplus (see Tables 3 and 4). The non-stoichiometry of crichtonite analyses is consistent with data for natural crichtonites from upper-mantle settings (Jones et al., 1982; Haggerty, 1983, 1987; Haggerty et al., 1983; Zhou et al., 1984; Varlamov et al., 1996; Wang et al., 1999; Konzett et al., 2000; Grégoire et al., 2002) that show an even stronger variation in stoichiometric coefficients with values for cations, A-cations and M-cations in the range 21·68–22·82, 0·40–1·50 and 20·67–21·84 (n = 32), respectively (see Fig. 4). It should be kept in mind, however, that an assignment of phases to the crichtonite series is usually based on the similarity of electron microprobe analyses to those of crichtonite phases and not on structure analysis. This includes data with anomalously low A-cations in the range 0·40–0·49 reported by Varlamov et al. (1996).
|
The chemical variability of crichtonite phases synthesized in Pt capsules with graphite liners is most pronounced in the Ti and Cr contents, which always show the highest standard deviations for averaged concentrations for individual runs (Tables 3–5). Because there is no evidence for significant amounts of vacancies, at least not in runs at
1500°C, Ti 
Cr exchange requires a coupled substitution involving mono- or divalent cations. General trends in the covariation of Cr with Ti, Mg and Fe (Fig. 5a–c) are similar for crichtonite phases, albeit with some variation in the degree of correlation, and are consistent with a substitution 2 Cr Ti + Mg, combined with Mg Fe2+ and possibly Cr Fe3+. Although Fe is present in the starting material as Fe3+ only, the positive correlation between Fe and Mg observed in runs using bulk compositions II and V would be consistent with some Fe2+ in the run products as a result of Fe reduction in the presence of graphite–diamond. The extent of Cr Fe3+ exchange is difficult to estimate because the negative Cr–Fe correlation would be consistent with both 2Cr Ti + (Mg, Fe2+) and Cr Fe3+.
|
Crichtonites synthesized in run JKW43 without an inner graphite liner show an almost complete lack of Fe combined with much higher Mg and Cr contents compared with crichtonites from JKW68 synthesized at identical P and T conditions but in the presence of graphite–diamond (Fig. 5). The negative correlation of both Fe–Cr and Fe–Mg (Fig. 5b and c) for crichtonite compositions in runs JKW43 and JKW68 could be explained by a substitution of the type 2Cr + Mg
Ti + 2Fe2+. The presence of Fe2+ required for this exchange would be consistent with fO2 permitting Fe loss to the Pt capsule through the formation of a Pt–Fe alloy. A comparison of the Fe–Mg data (apart from direct inspection of Fe concentrations) further shows that run B02-11 must have suffered some Fe loss in spite of the presence of a graphite liner.
Apart from the Ti–Cr variation as discussed above, analyses of the LIMA solid solution are also characterized by a systematic increase in K/(Ba + K) ratios with pressure (Fig. 5d), which requires a coupled substitution involving a monovalent cation. The positive correlation between K and Ti observed for LIMA analyses (Fig. 5e) would be consistent with an exchange of the type K + Ti Cr + (Ba + Mg). Although there is some scatter in the data, negative correlations between K and Cr and a positive correlation between Mg and Ba (Fig. 5f) would support this assumption.
The Raman spectra of mathiasite, lindsleyite and LIMA solid solution are characterized by broad bands in the wavenumber region <800 cm–1 (Fig. 6a). The most prominent feature is a broad and asymmetric band in the region 657–695 cm–1 that may be split in the case of lindsleyite into two adjacent bands at 662 and 702 cm–1. Towards lower wavenumbers, bands at 543–560, 432–437, 316–324 and 201–205 cm–1 are present in all crichtonite phases (Fig. 6a). As a result of anisotropy, additional bands at 354 cm–1 in mathiasite are present only as shoulders on the 316–324 cm–1 bands in lindsleyite and LIMA solid solution. Similarly, bands at 243 and 248 cm–1 present in the LIMA and mathiasite spectra appear as shoulders only in the lindsleyite spectrum.
|
K–Cr-titanate
This phase is a breakdown product of mathiasite in runs with bulk composition I, and is the only carrier of potassium under subsolidus conditions apart from mathiasite in this system. Compared with mathiasite, the K–Cr-titanate contains much less TiO2 (31–37 wt %) and more Cr2O3 (30–37 wt %), and twice as much K2O (6·5–7·0 wt %).
The XRD study on a K–Cr-titanate single crystal from JKW86 (bulk VII) shows that it possesses an unreported structure with the unit cell parameters a = 9·175(2) Å, c = 2·879(1) Å, V = 209·9(1) Å3, and space group P63/m. The crystal structure of this K–Cr-titanate phase contains three cationic sites M1, M2 and M3, which are occupied by (Ti4+ + Zr4+ + Cr3+ + Fe3+), Mg2+ and K+, respectively. Its main feature is double chains of M1 octahedra, which consist of two adjacent single chains that share octahedral edges running parallel to the c-axis. The double chains link corners with each other to form a framework with two types of tunnels, one occupied by M2 cations and the other by M3 (Fig. 7). A more comprehensive description of the structure will be reported elsewhere.
|
On a 12 oxygen basis suggested by the structure analysis, K–Cr-titanate analyses are characterized by a significant variation in
R2+ + R3+ and R4+, ranging from 4·29 to 5·29 and from 2·28 to 2·99 a.p.f.u., respectively (Fig. 8a, Table 3). Total cation sums, too, show a variation in the range 8·08–8·37 a.p.f.u. At constant temperature, both total cations and R2+ + R3+ tend to increase, whereas R4+ decreases. Cation sums for JKW90 (11 GPa and 1600°C) deviate from the trend, with the possibility that high- and low-temperature runs at 1600 and 1400°C follow two separate trends (Fig. 8a). The large cation position, on the other hand, does not show any significant variation with pressure or temperature, and ranges between 0·78 ± 0·02 and 0·81 ± 0·01 a.p.f.u. (Table 3).
|
As was the case for crichtonite phases, major variations in K–Cr-titanate compositions are observed in Cr and Ti with an increase in Cr with increasing pressure (Fig. 8b). Possible exchange mechanisms to account for the observed compositional variation of K–Cr-titanate are more difficult to evaluate compared with crichtonite phases because of the variability of cation sums with the possibility of vacancies being involved. Nevertheless, element covariations (e.g. Fig. 8b–d) would be consistent with exchange mechanisms of the form 2 R3+
R4+ + R2+. Again, some reduction of Fe3+ is likely to have occurred in the presence of graphite–diamond. A negative correlation of Fe with Ti (Fig. 8d) and positive correlations with Mg and Cr would be consistent with the presence of both Fe2+ and Fe3+ involved in 2 Cr Ti + (Mg + Fe2+) and 2 (Cr + Fe3+) Ti + Mg. Clearly, in the absence of Mössbauer data on Fe2+/Fe3+ ratios, these considerations must remain speculative.
Rutile–TiO2-II
The most common phase accompaying crichtonite and magnetoplumbite phases is rutile or its high-pressure equivalent TiO2-II possessing -PbO2 structure (Linde & DeCarli, 1969). It always contains significant amounts of Cr, Zr and Fe, with Cr being the major impurity. The maximum observed concentrations of minor elements in rutile are 14·6 wt % Cr2O3, 7·3 wt % ZrO2 and 2·3 wt % Fe2O3 (Tables 3–5). The rutile–TiO2-II transition, which passes through 6·5 GPa and 1000°C and 11·0 GPa and 1500°C (Withers et al., 2003, and references therein), does not have an obvious influence on the minor element contents of TiO2 as there are no discontinuities in element correlations for TiO2.
In the absence of pentavalent cations, the substitution of trivalent cations such as Cr or Fe3+ requires a shift in rutile composition towards CrO1·5 and FeO1·5, thus producing a non-stoichiometric rutile with a composition TiO2–x (Dirstine & Rosa, 1979). Whereas Zr occupies Ti lattice positions, the incorporation of Cr is thought to take place on unoccupied octahedral voids (= interstitial) of the rutile structure, introducing local lattice defects with corundum-type structure according to 2 Ti4+ + O2– 2 (Cr3+, Fe3+) + vO (Flörke & Lee, 1970); charge compensation is ensured by the introduction of oxygen vacancies (Dirstine & Rosa, 1979). In spite of its chemical simplicity, the system Ti–Cr–O is highly complex in terms of crystallographic structures and discrete compounds encountered (Flörke & Lee, 1970; Gibb & Anderson, 1972). According to Gibb & Anderson (1972), the rutile structure sensu stricto is stable with 0–5 mol % CrO1·5, whereby Cr substitution leads to the introduction of localized corundum-type structure defects. Between 5 and 15 mol % CrO1·5, anion sites are removed by crystallographic shear to achieve charge compensation, and for CrO1·5 concentrations in the range >15 to <50 mol %, a series of discrete phases with a general formula Cr2Tin–2O2n–1 with n = 6–12 becomes stable (Fig. 9). At >50 mol % CrO1·5, the system is saturated with respect to Cr2O3 (Flörke & Lee, 1970; Gibb & Anderson, 1972). The introduction of ZrO2 and Fe2O3 leads to additional complexities because for suitable Cr:Fe and Ti:Zr ratios, the system Cr2O3–Fe2O3–TiO2–ZrO2 contains a series of phases with a general formula (Cr,Fe)2p(Ti,Zr)p+2qO5p+4q or pM3O5·qM2O4 with p = 1–4 and q = 0–5, which can be represented as polysomatic intergrowths [i.e. intergrowths of distinct structure modules in various porportions; see Thompson (1978)] of V3O5 and -PbO2 structure types (Grey et al., 1973) (Fig. 9). It should be noted that the crystallographic characterization of TiO2–Cr2O3 solid solutions is based on 1 atm experiments and that the distinction of compositional fields with characteristic structures (i.e. rutile sensu stricto, crystallographic shear) is not necessarily valid for high pressures.
|
In the rutiles of the present study, combined TiO2 + ZrO2 is always
82 wt % and negatively correlated with Cr close to a 1:1 exchange line. (Cr, Fe3+)O1·5 does not exceed 15 mol %, thus falling into the fields of rutile sensu stricto and TiO2 with crystallographic shear (Gibb & Anderson, 1972). When normalized to 1·0 cation, the maximum oxygen deficiency required to maintain charge balance is 0·09 or (Ti, Cr, Fe3+)O1·91 (see Tables 3–5).
Raman spectra of rutiles were recorded from runs B01-6 and JKW43 (bulk I) at 7 and 8 GPa. All spectra show four bands at 258–261, 429–430, 614 and 693 cm–1 (Fig. 6c). By comparison, a spectrum of pure synthetic rutile only shows three strong bands at 240, 446 and 610 cm–1. The shift of band positions in Cr-rich rutile compared with pure rutile and the appearance of the additional band at 693 cm–1 is thought to be due to the presence of Cr in the TiO2 lattice. It should be noted, however, that the 693 cm–1 band does not correspond to any of the Cr–O vibrational modes known from Cr2O3 (see Brown et al., 1968; Mougin et al., 2001). A band shift as a result of a change in interatomic distances resulting from the high pressure of formation seems unlikely because the shift of the 446 cm–1 band (e.g. Ti–O stretching) towards lower wavenumbers is opposite to the shift of this band as observed in in situ high-pressure Raman studies (Mammone & Sharma, 1979; Liu & Mernagh, 1992). Raman spectra of TiO2-II containing between 5 and 10 wt % Cr2O3 were recorded for runs JKW88 (bulk II) and JKW76 (bulk I). Eight bands were identified at 178, 285, 315, 337, 354, 426, 531 and 585 cm–1 (Fig. 6c). In contrast to rutile, the band positions of TiO2-II seem to be virtually unaffected by the presence of substantial Cr and Zr, and agree very well with the positions reported by El Goresy et al. (2001) at 175, 285, 315, 340, 358, 428, 532 and 575 cm–1 for natural TiO2-II from the Ries crater in Germany.
ZrO2–TiO2 solid solution
Most runs (Table 2) contain grains of a ZrO2–TiO2 solid solution, which, however, are usually too small to be analysed. In cases where the size permitted a reliable analysis, compositions with 30–35 wt % TiO2 and 65–68 ZrO2 were obtained (Tables 3–5). Normalized to two oxygens, this would be equivalent to Ti0·40–0·44Zr0·60–0·55O2. According to Feighery et al. (1999), however, an orthorhombic compound ZrTiO4 is stable instead of tetragonal (Zr,Ti)O2 in the system ZrO2–TiO2 between 40 and 55 mol % ZrO2. In the present study, ZrTiO4 always shows a slight Zr excess compared with the ideal 1:1 Zr/Ti ratio.
Zr phase
With 48–58 wt % TiO2, 21–34 wt % ZrO2 and minor amounts of Cr, Fe and Mg (Tables 3 and 4), this phase is the most Zr rich apart from the ZrO2–TiO2 solid solution encountered in this study, and appears in high-temperature runs at 1600°C in bulk compositions I and II. It was one of the phases selected for structure analysis. An averaged composition derived from run JKW78 was used as starting composition for a synthesis experiment at 8·8 GPa and 1600°C. Structure analysis of a single crystal extracted from JKW81 showed that the Zr phase is isostructural with fluorite. A more complete discussion of the Zr phase structure and possible implications for fluorite-type structures at high pressures and temperatures will be presented elsewhere.
Magnetoplumbite phases
Like crichtonites, all magnetoplumbite phases in this study show a deviation from the ideal AM12O19 stoichiometry in accordance with data from natural magnetoplumbites from upper-mantle settings. Yimengite and hawthorneite synthesized at 7, 10 and 15 GPa show a range in A-site and M-site occupancies of 0·93–1·28 and 11·73–12·24 a.p.f.u., respectively. Total cation sums are in the range 12·68–13·33 (Fig. 4d–f). TiO2 and Cr2O3 contents of 19–33 wt % and 33–44 wt %, respectively, are closely comparable with the compositions of natural yimengites and hawthorneites (Zhou et al., 1984; Haggerty et al., 1989). It is interesting to note that yimengite and hawthorneite are the only carriers of large cations in all runs using bulk compositions III and IV except in run JKW43 using bulk IV. In this run, the shift in bulk composition associated with the almost complete loss of Fe from the experimental charge resulted in the stability of an additional Ba–Cr–Ti phase (Ba phase 3, Table 6). By comparison, such partitioning of the large cation into a second phase could not be observed in JKW43 using bulk composition III at identical P and T conditions.
In all runs using bulk composition VI, the 50:50 hawthorneite–yimengite solid solution represented by this bulk composition is unstable. Instead, two coexisting Ba–K phases (HAYI and YIHA in Table 2), are present together with small amounts of unidentified phases U7–U9 (see Table 2). This is unlike the situation encountered in the lindsleyite–mathiasite system (see above) and consistent with results obtained by Foley et al. (1994) at somewhat lower pressures using the same bulk composition. In this study, the coexisting Ba–K phases are considered to be magnetoplumbite phases following the suggestion by Foley et al. (1994), and were renormalized to 19 oxygens accordingly.
Cation sums of both hawthorneite-rich and yimengite-rich solid solutions (HAYIss and YIHAss) deviate from the ideal values. This deviation is especially pronounced for HAYIss, which shows total cation sums of 11·46–12·65 and M-site cation sums of 10·10–11·67 as compared with the ideal values of 13·00 and 12·00 (Fig. 10). Between 5 and 13 GPa, both hawthorneite-rich and yimengite-rich solid solutions (HAYIss and YIHAss) show a systematic decrease in K a.p.f.u. and K/(K + Ba) along with a concomitant increase in Ba (Fig. 11c, e and f). In the case of HAYIss, K decreases from 0·36 to 0·06 a.p.f.u. while Ba increases from 0·68 to 1·30 a.p.f.u., resulting in a decrease in K/(K + Ba) from 0·35 to 0·04. In coexisting YIHAss, K decreases from 1·38 to 0·48 a.p.f.u., accompanied by an increase in Ba from 0·02 to 0·58 a.p.f.u., resulting in a decrease in K/(K + Ba) from 0·99 to 0·45. Run B02-10 shows a slight reversal of this trend with increasing Ba and K a.p.f.u., especially for HAYIss. The significance of this reversal, however, would have to be tested by additional runs. It might also be an analytical artefact as a result of the small grain size of HAYIss in this run and the possibility of beam overlap with adjacent K-rich YIHAss grains and/or analysis of undetected K-rich inclusions in HAYIss. Consistent with the results by Foley et al. (1994), HAYIss always has lower Mg and Mg/(Mg + Fe3+) compared with YIHAss (Fig. 11b).
|
|
An interesting feature of the compositional variation of the Ba–K phases in bulk composition VI is the strong increase in the Ti/Cr ratio of HAYIss above 10 GPa with an increase in Ti from 2·98 Ti a.p.f.u. at 10 GPa to 4·89 and 5·58 Ti a.p.f.u. at 11 and 13 GPa (Fig. 11a and b). A possible explanation would be Fe loss despite the use of a graphite liner because the abrupt increase in Ti/Cr of HAYIss is accompanied by a strong decrease in Fe3+ a.p.f.u. from 1·8–1·9 at P
10 GPa to 0·7–0·9 at 11 and 13 GPa. On the other hand, run B02-10 also produced HAYIss with high Ti/Cr compared with runs at P 10 GPa, but in this case combined with Fe3+ a.p.f.u. values almost identical to those encountered at low pressures. At present, the significance of this change in Ti/Cr cannot be explained satisfactorily and would require structure analysis of HAYIss.
Raman spectra of magnetoplumbite phases are presented in Fig. 6b. Like crichtonite phases, both yimengite and hawthorneite show relatively simple spectra with broad bands in the wavenumber region <800 cm–1. The most prominent feature of the yimengite spectrum is a broad double band at 629 and 695 cm–1 with additional small bands at 545, 471 and 285 cm–1. By comparison, only one strong band at 680 cm–1 is present in the hawthorneite spectrum, with an additional strong band at 516 cm–1 and small bands at 459, 352 and 285 cm–1. The spectrum of the Ba-rich phase (HAYI) in the system yimengite–hawthorneite is similar to that of hawthorneite, with the strongest band at 678 cm–1. The spectrum of the K-rich phase (YIHA), on the other hand, is different from either the yimengite or hawthorneite spectrum but very similar to that of the K–Cr-titanate encountered in bulk composition I. In view of the overall compositional similarity between YIHA and the K–Cr-titanate, the similarity in Raman spectra of the two phases raises the question of whether YIHA and HAYI do represent magnetoplumbite-structured phases coexisting along a solvus as suggested by Foley et al. (1994). An answer to this question, however, will have to await structure analysis of both YIHA and HAYI.
Unidentified phases
A number of unidentified Ti–Cr–(Ba) phases are present together with crichtonite and magnetoplumbite phases for which no data on their structures are available and a distinction is solely based on electron microprobe analysis. As noted above, the distinction is therefore somewhat arbitrary and only preliminary in the absence of data on the crystal structure of these phases. Nevertheless, a formula calculation was attempted for some of the phases to give at least an educated guess on possible stoichiometries. No attempt has been made to calculate stoichiometries based on a fixed number of cations, especially when phases contain significant Fe with an unknown Fe2+/Fe3+ ratio, because this would place tight and unjustified constraints on cation ratios and result in grossly misleading stoichiometries.
Unidentified Ba-bearing phases (Ba phases 1–3). Two Ba-bearing Cr–Ti-rich phases appear as breakdown products of lindsleyite at pressures between 11 and 13 GPa. Both phases contain 50–52 wt % TiO2 and 15–20 wt % Cr2O3 and both are significantly richer in Ba than lindsleyite, with 20·6–22·2 wt % and 15·3–17·5 wt % BaO in the Ba-rich and Ba-poor phase, respectively (Table 4). Both Ba-bearing phases show a systematic difference in composition, with the Ba-rich phase always being richer in Ba and Ti and poorer in Zr, Cr, Fe and Mg compared with the Ba-poor phase (Table 4). Whether these two phases have similar structures and represent compositions on opposite limbs of a solvus is not known at present and would require more experiments to higher pressures to outline the shape of a potential solvus. Assuming that both phases contain one structural position that is exclusively occupied by Ba, and that this position can be fully occupied (i.e. vacancies on the Ba position are not an integral part of the structures), then the compositions of the Ba-rich and Ba-poor phase would be consistent with BaM8O15 and a BaM11O20 stoichiometry, respectively.
Ba phase 3 was found only in run JKW43 using bulk composition IV. Its appearance is thought to be due to Fe loss from bulk composition IV as a result of the lack of a graphite liner. Assuming Cr a.p.f.u. = n x 2 and a separate lattice position for Ba that shows a degree of occupation similar to that of magnetoplumbites and crichtonites, then the smallest reasonable oxygen basis would be 20, resulting in a formula Ba0·88Mg2·68(Cr,Fe3+)4·38Ti4·94O20 with cations = 12·00. Normalization to 19 oxygens (i.e. assuming a solvus generated by unmixing on the M-site) would result in a total cation sum and M-site cations much lower than found in any high-pressure magnetoplumbite reported or synthesized in this study.
Unidentified Ti–Cr–Fe–Mg phases. A number of Ti–Cr–Fe–Mg phases appear in addition to the crichtonite and magnetoplumbite phases described above. Modal amounts of these phases range from 20% to only one grain found in an individual run. These phases, labelled U1–U9 in Table 2, have a wide range in TiO2 and Cr2O3 contents from 6 to 75 wt % TiO2 and from 15 to 60 wt % Cr2O3. Averaged or representative compositions are listed in Tables 4–7. Some of the phases have compositions similar to natural Ti–Cr phases such as U4 and U8, or similar to phases known from low-pressure experiments such as U2.
Phase U4 is comparable in composition with a Cr–Ti oxide described by Stachel et al. (1998) as an inclusion in diamond that was tentatively identified as an unusually Cr-rich ilmenite. Both compositions of U4 and that reported by Stachel et al. (1998) are listed in Table 5. It should be noted that a wide range in Ti–Cr–Fe–Mg oxide compositions (e.g. phases U6 and U7) can be normalized to two cations and three oxygens assuming a ilmenite stoichiometry and a variation in composition as a result of an exchange 2 (Cr + Fe3+) Ti + (Mg + Fe2+). In the absence of information on Cr2O3 and Fe2O3 solubility limits in ilmenite at high P and T, and in view of the complexity of the Cr–Ti–Fe–Mg system even at low pressures, no attempt has been made to normalize other Ti–Cr–Fe–Mg phases to an ilmenite stoichiometry.
Phase U9 was found together with U8 in run B01-7 as euhedral to subhedral crystals. It contains 6·5 wt % TiO2 and 60 wt % Cr2O3, which would be similar in composition to a Ti-rich Cr-spinel. Normalization to four oxygens and three cations yields stoichiometric coefficients that can be expressed in terms of known spinel end-members Mg2TiO4 (16 mol %), FeFe2O4 (5 mol %), FeCr2O4 (37 mol %) and MgCr2O4 (41 mol %) with a negligible excess of Cr2O3 of 0·3 mol %.
Phase U2 may be a member of the Ti2–nCr2O2n–1 series of Andersson phases (Flörke & Lee, 1970). Combined TiO2 + ZrO2 contents near 78 wt % indicate a compound Ti7Cr2O17 ideally containing 78·6 wt % TiO2 and 21·4 wt % Cr2O3. In fact, normalization of phase U2 to 17 oxygens yields total cations = 9·04–9·18, Ti + Zr = 6·93–7·09 and Cr + Fe + Mg (Fetot = Fe2O3) = 1·85–2·22.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...PREVIOUS EXPERIMENTAL WORKRESULTSDISCUSSIONSUMMARY OF CONCLUSIONSREFERENCES |
|---|
What factors govern the stability of crichtonite and magnetoplumbite phases in the upper mantle?
The formation of crichtonite and magnetoplumbite series phases is commonly attributed to metasomatic alteration of Cr-spinel in depleted Cr-rich harzburgites by incompatible (trace) element enriched fluids or melts (e.g. Nixon & Condliffe, 1989
). An inclusion of Cr-rich and Al-depleted spinel coexisting with Cr-rich pyrope and enstatite in a Sr-magnetoplumbite-containing diamond described by Sobolev et al. (1997) also indicates a harzburgitic source for magnetoplumbite. The fact that many crichtonites and magnetoplumbites seem to have grown on pre-existing Cr-spinel (e.g. Haggerty, 1983, 1987; Nixon & Condliffe, 1989; Velde, 2000; Bulanova et al., 2004) is presumably related to the similarity of spinel- and crichtonite–magnetoplumbite structures that favours nucleation of crichtonite or magnetoplumbite on a spinel substrate. The source of the Ti required for titanate formation is more equivocal. Menzies et al. (1987) assumed that Ti is not introduced by metasomatism but derived locally from the pre-metasomatic whole-rock Ti budget. This can be a reasonable assumption for a primitive mantle containing 0·2 wt % TiO2 (McDonough & Sun, 1995), but not for an average cratonic peridotite, for which McDonough (1994) and Rudnick et al. (1998) reported only 0·08 wt % TiO2. The presence in crichtonite and/or magnetoplumbite assemblages of very Ti-rich Cr-spinels with up to 7 wt % TiO2 (Nixon & Condliffe, 1989; see also Haggerty, 1983; Bulanova et al., 2004) is atypical of cratonic (harzburgitic) spinels (see McDonough & Rudnick, 1998) and also suggests metasomatic addition of Ti prior to or contemporaneous with the incompatible trace element infiltration.
The presence of lindsleyite–mathiasite in spinel-free fertile metasomatized lherzolites (Konzett et al., 2000; Grégoire et al., 2002) shows that the presence of Cr-spinel is not an indispensible prerequisite for crichtonite phase formation in the upper mantle. Instead, silicate sources such as clinopyroxene must be invoked as the Cr source. The unusually low Cr2O3 contents of 7·9–10·9 wt % reported by Konzett et al. (2000) for LIMA may thus reflect the absence of Cr-spinel.
Apart from the availability of LILE and a local source of Cr, other factors that govern the formation of crichtonite and magnetoplumbite phases in peridotitic upper mantle remain unclear. This is especially true for magnetoplumbites because only very few occurrences are reported that represent either isolated diamond inclusions or macrocrysts from kimberite concentrates and, hence, do not allow the study of the original environment of formation. In the case of crichtonite phases, the situation is better because crichtonites are (amongst other occurrences, see above) known from a suite of metasomatized peridotites that can be placed in a genetic context (see Erlank et al., 1987).
The apparent rarity of crichtonite phases as compared with the widespread presence of phlogopite and/or amphibole in metasomatized peridotites indicates that in most cases hydrous silicates preferentially incorporate LILE introduced by metasomatism. This is possible because incompatible (trace) element metasomatism is almost always accompanied by the introduction of water (e.g. Menzies & Hawkesworth, 1987, and references therein), whereby a hydrous fluid may act as a complexing agent for trace element transport. The distribution of crichtonite phases in metasomatized peridotites from the Kaapvaal Craton (Erlank et al., 1987, and references therein) indicates that it is the hydrogen/LILE ratio that is critical for crichtonite stability. It is only in the most advanced states of metasomatism that LIMA phases become stable in addition to phlogopite and/or amphibole. To the authors' best knowledge, no occurrence of LIMA has been described where these phases are present in the absence of hydrous silicates. Therefore, it is only in cases where the ability of hydrous phases (whose modal amount is controlled by the amount of hydrogen available) to accommodate LILE is exhausted because of a high LILE/H ratio that crichtonite may become stable, provided sufficient Cr and Ti is available.
The availability of Ti is controlled by the ability of coexisting phases to incorporate this element. By far the most common metasomatic Ti phases are rutile and ilmenite, both of which may also accommodate Cr on a wt % level (e.g. Tollo & Haggerty, 1987) and are stable instead of crichtonite–magnetoplumbite if LILE are locked up in coexisting hydrous silicates. At high pressures, the ability of silicate phases to compete for Ti has to be taken into account, especially in fertile alkali-enriched bulk compositions. Peridotites sampled by kimberlites often contain garnets with high TiO2 contents reaching wt % levels in addition to significant Na (e.g. Peltonen et al., 1998; Sobolev et al., 1998; Stachel et al., 1998). The ability of garnet to incorporate large amounts of Ti and Na through exchange of the type M+ + Ti4+ M2+ + M3+ is confirmed by high-pressure experiments (e.g. Konzett, 1997; Zhang et al., 2003). This means that in garnet-bearing systems, the availability of Ti for titanate formation decreases with increasing pressure as a result of partitioning into garnet.
A comparison of the evolution of the A-site cation occupancies in crichtonites and magnetoplumbites synthesized in the present study shows a different behaviour for the two series of phases. Crichtonites show a systematic increase in K/Ba with pressure whereas magnetoplumbites show decreasing K/Ba on both limbs of the solvus (if one accepts the idea of a solvus based on experimental evidence and in the absence of exsolution textures known from natural samples) (Figs 5 and 11).In this respect, the K–Ba systematics in crichtonites are similar to the K–Na systematics in K–Na-bearing silicates from peridotites such as amphiboles, micas (e.g. Konzett et al., 1997; Konzett & Fei, 2000) and clinopyroxene (the last only at pressures >5 GPa; see Harlow, 1997). This increasing preference of silicates for K with increasing pressure is attributed to the much higher compressibility of the potassium ion and its ability to fit more tightly in the large and highly coordinated lattice positions into which both K and Na enter (e.g. Konzett et al., 1997). Because no data are available on the variation in K–Ba systematics of silicates with pressure, it is not clear whether decreasing K/Ba in magnetoplumbite represents an exception to a more general preference of K over Na and other cations with large ionic radii.
Results obtained from bulk composition VI indicate that the system yimengite–hawthorneite is much more complex than the system mathiasite–lindsleyite. The compositional evolution of coexisting Ba–K phases with significant changes between 10 and 11 GPa and the similarity of Raman spectra of YIHA and the K–Cr-titanate at least hint at the possibility that phase compositions and structures in the system yimengite–hawthorneite are not simply controlled by a solvus. Further investigations will, however, be required to gain a more comprehensive understanding of this system.
The potential role of oxide phases as LILE carriers
Amongst the multitude of natural LILE-bearing titanates primarily known from kimberlites and lamproites (see Haggerty, 1987; Mitchell & Bergman, 1991; Mitchell, 1995) only members of the crichtonite and magnetoplumbite series of phases in addition to priderite are known to occur as presumably syngenetic inclusions in diamonds (Sobolev et al., 1988, 1997; Jaques et al., 1989; Kopylova et al., 1997; Bulanova et al., 2004). Synthesis experiments by Foley et al. (1994) at 5 GPa and 1350°C confirmed the high-pressure stability of these phases. The presence in a diamond of a Cr–Sr-rich loparite containing 4·4 wt % K2O reported by Kopylova et al. (1997) from the River Ranch kimberlite, Limpopo belt, SE Africa, however, was attributed by those workers to growth of this phase outside the stability field of diamond.
Experiments in this study indicate that Ba- and K-rich members of the crichtonite and magnetoplumbite series of phases have P–T stability fields that extend far into the diamond stability field to a depth of at least 300 km in the case of crichtonites and to at least 400–450 km in the case of magnetoplumbites. Breakdown of crichtonite phases between 11 and 12 GPa leads to the formation of new K- and Ba-titanates that may act as LILE carriers to even greater depths. The temperatures at which magnetoplumbite and crichtonite phases were synthesized in the present study are as high as 1400–1600°C, which suggests that these phases would be stable not only throughout the entire continental (cratonic) lithosphere but also along an average present-day mantle adiabat in the underlying asthenosphere (Fig. 12). P–T estimates of 6–7 GPa at 1300–1500°C reported by Leost et al. (2003) for a diamond inclusion suite including a Cr-titanate (crichtonite according to those workers) + Ti-rich phlogopite would thus be consistent with the results of this study. Once crichtonite and/or magnetoplumbite phases have formed by metasomatic alteration of a sufficiently Cr- and Ti-rich source, they may act as an important source of LILE and HFSE during partial melting of the lithosphere or asthenosphere even if present only in trace amounts. K–Ba-titanates of the crichtonite and magnetoplumbite series of phases can thus be expected in the source regions of lamproites and both group I and group II (orangeites sensu Mitchell, 1995) kimberlites (see Foley, 1992a; Mitchell, 1995) (Fig. 12), although one should not take this to imply that these phases are commonly present.
|
Most recently, Bulanova et al. (2004)
reported 124 ppm Rb, 3·7 ppm Cs and 2543 ppm Sr in syngenetic yimengite inclusions in diamond, underlining the importance of this phase as an LILE carrier. Likewise, extremely high LILE and HFSE concentrations reaching wt % levels for SrO, BaO and ZrO2 and >1000 ppm for REE are known from LIMA phases (Jones & Ekambaram, 1982; Grégoire et al., 2002). Unfortunately, no data are available on the near-solidus behaviour of crichtonites and magnetoplumbites in a metasomatized peridotite. Therefore, it is not known (1) whether these phases can coexist with a partial melt and control its LILE budget and (2) whether they would survive thermal erosion of the lithosphere and become entrained in convecting asthenospheric mantle.
The role of pseudobrookite (armalcolite) [(Mg,Fe)Ti2O5] as an LILE carrier in the upper mantle remains unclear. The pseudobrookite structure does not possess large interstitial cation sites suitable for LILE (see Yang & Hazen, 1998, and references therein). On the other hand, up to 0·30 wt % K2O and 0·77 wt % BaO have been reported for Cr-rich armalcolites by Haggerty (1983). Jaques et al. (1990) gave even higher K2O concentrations of up to 2·10 wt % for armalcolites from diamond-bearing peridotites sampled by the Argyle lamproite. Experiments by Friel et al. (1977) have shown that pseudobrookite (Mg0·5Fe0·5)Ti2O5 breaks down at P 1·5 GPa and 1200°C to form rutile + ilmenite. Hence, pseudobrookite would act as an LILE carrier in shallow upper mantle only. This is consistent with an occurrence of armalcolite coexisting with feldspar reported by Grégoire et al. (2000b) for which P–T conditions of 1·3 GPa and 1150–1200°C have been estimated.
|
SUMMARY OF CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...PREVIOUS EXPERIMENTAL WORKRESULTSDISCUSSIONSUMMARY OF CONCLUSIONSREFERENCES |
|---|
(1) High-P–T experiments have been conducted in the range 7–15 GPa and 1150–1600°C to determine the P–T stability and breakdown products of K- and Ba-dominant members of the crichtonite (AM21O38) and magnetoplumbite (AM12O19) series of phases.
(2) Lindsleyite (LIN) Ba(Ti12Cr4Fe2ZrMg2)O38 and mathiasite (MA) K(Ti13Cr4FeZrMg2)O38 are stable to 11 GPa and 1500–1600°C. Between 11 and 12 GPa, lindsleyite breaks down to form two Ba–Cr-titanates of unknown structure that persist to at least 13 GPa. The high-P breakdown product of mathiasite is a K–Cr-titanate with an idealized formula KM7O12, where M = Ti, Cr, Mg, Fe. This phase possesses space group P63/m with a = 9·175(2) Å, c = 2·879(1) Å, V = 209·9(1) Å3.
(3) Hawthorneite (HA) Ba(Ti3Cr4Fe2Mg3)O19 and yimengite (YI) K(Ti3Cr5Fe2Mg2)O19, are stable in runs at 7, 10 and 15 GPa between 1300 and 1400°C coexisting with a number of Ti–Cr-oxides.
(4) Molar mixtures (1:1) of LIMA and HAYI were studied at 7–10 GPa and 1300–1400°C, and 9–15 GPa and 1150–1400°C, respectively. In the system LIMA, one homogeneous Ba–K phase is stable, which shows a systematic increase in the K/(K + Ba) ratio with increasing P. In the system HAYI, two coexisting Ba–K phases appear, which are Ba rich and Ba poor, respectively. This is consistent with results obtained by Foley et al. (1994).
(5) Results of this study suggest that crichtonite and magnetoplumbite phases are potentially stable throughout the entire subcontinental lithosphere and the underlying asthenosphere to a depth of up to 450 km, and may thus be present in the source regions of both group I and group II kimberlites and lamproites, where these phases would act as major LILE and HFSE carriers. At still higher pressures, both K and Ba may remain stored in alkali-titanates.
(6) By comparison with natural LIMA phase assemblages, we conclude that only in the most advanced stages of upper-mantle metasomatism would LIMA phases become stable in addition to phlogopite and/or amphibole. It is only in cases where the ability of hydrous phases to accommodate LILE is exhausted as a result of a high LILE/H ratio that LIMA phases may form, provided sufficient Cr and Ti is available. The same is probably true for HAYI phases. Because of increased partitioning of Ti into garnet via coupled substitutions involving Na, the availability of Ti for oxide formation in general decreases with increasing P.
(7) An important aspect of the high-P–T stability of LIMA and HAYI phases not covered by this study is their near-solidus behaviour. In particular, it is not known whether these phases possess a supersolidus stability in peridotitic systems and, hence, can control the LILE and HFSE budget of low-percentage partial melts by solid–melt partitioning or whether they break down at the solidus without the ability to buffer the melt composition.
|
ACKNOWLEDGEMENTS |
|---|
We gratefully acknowledge the help of Christos Haddidiacos, Geophysical Laboratory, and Theodoros Ntaflos, University of Vienna, during microprobe analysis. We are also indebted to Erwin Mayer and Christoph Salzmann, Department of Inorganic Chemistry, University of Innsbruck, for providing access to the Raman spectrometer and for assistance during recording of the spectra. Reviews by Michel Grégoire, Ben Harte and Nick Sobolev helped to improve the quality of the manuscript. Financial support for this study was provided by the Carnegie Institution of Washington and by the European High-Pressure Facility EU ‘Access to Research Infrastructures’ Programme during a visit of the first author to the Bavarian Research Institute of Experimental Geochemistry and Geophysics, University of Bayreuth.
* Corresponding author. Telephone: +43-(0)512-507-5506. Fax: +43-(0)512-507-2926. E-mail: juergen.konzett{at}uibk.ac.at
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL AND ANALYTICAL...PREVIOUS EXPERIMENTAL WORKRESULTSDISCUSSIONSUMMARY OF CONCLUSIONSREFERENCES |
|---|
Adelsköld, V. (1938). X-ray studies on magnetoplumbite, PbO·6Fe2O3 and other substances resembling ‘beta-alumina’, Na2O·11Al2O3. Arkiv før Kemi Mineralogioch Geologi, Series A-12 29, 1–9.
Bertka, C. M. & Fei, Y. (1997). Mineralogy of the Martian interior up to core–mantle boundary pressures. Journal of Geophysical Research 102, 5251–5264.[CrossRef]
Brown, D. A., Cunningham, D. & Glass, W. K. (1968). The infrared and Raman spectra of chromium (III) oxide. Spectrochimica Acta 24A, 965–968.
Bulanova, G. P., Muchemwa, E., Pearson, G., Griffin, B. J., Kelly, S., Klemme, S. & Smith, C. B. (2004). Syngenetic inclusions of yimengite in diamond from Sese kimberlite (Zimbabwe)—evidence for metasomatic conditions of growth. Lithos 77, 181–192.[CrossRef][Web of Science]
Chase, A. B. & Wolten, G. M. (1965). Aluminum and gallium analogs of magnetoplumbite. Journal of the American Ceramic Society 48, 276–277.[CrossRef][Web of Science]
Dawson, J. B. & Smith, J. W. (1977). The MARID (mica–amphibole–rutile–ilmenite–diopside) suite of xenoliths in kimberlite. Geochimica et Cosmochimica Acta 41, 309–323.[CrossRef][Web of Science]
Dirstine, R. T. & Rosa, C. J. (1979). Defect structure and related thermodynamic properties of nonstoichiometric rutile (TiO2–x) and Nb2O5 doped rutile. Part I: The defect structure of TiO2–x (rutile) and partial molar properties for oxygen solution at 1273 K. Zeitschrift für Metallkunde 70, 322–329.
El Goresy, A., Chen, M., Gillet, P., Dubrovinsky, L., Graup, G. & Ahuja, R. (2001). A natural shock-induced dense polymorph of rutile with -PbO2 structure in the suevite from the Ries crater in Germany. Earth and Planetary Science Letters 192, 485–495.[CrossRef][Web of Science]
Erlank, A. J., Waters, F. G., Hawkesworth, C. J., Haggerty, S. E., Rickard, R. S. & Menzies, M. (1987). Evidence for mantle metasomatism in peridotite nodules from Kimberley pipes, South Africa. In: Menzies, M. & Hawkesworth, C. J. (eds) Mantle Metasomatism. London: Academic Press, pp. 221–312.
Feighery, A. J., Irvine, J. T. S., Flagg, D. P. & Kaiser, A. (1999). Phase relations at 1500°C in the ternary system ZrO2–Y2O3–TiO2. Journal of Solid State Chemistry 143, 273–276.[CrossRef]
Flörke, O. W. & Lee, C. W. (1970). Andersson Phasen, dichteste Packung und Wadsley Defekte im System Ti–Cr–O. Journal of Solid State Chemistry 1, 445–453.
Foley, S. F. (1992a). Petrological characterization of the source components of potassic magmas: geochemical and experimental constraints. Lithos 28, 187–204.[CrossRef][Web of Science]
Foley, S. F. (1992b). Vein-plus-wall-rock melting mechanisms in the lithosphere and the origin of potassic alkaline magmas. Lithos 28, 435–453.[CrossRef][Web of Science]
Foley, S. F., Höfer, H. & Brey, G. (1994). High-pressure synthesis of priderite and members of the lindsleyite–mathiasite and hawthorneite–yimengite series. Contributions to Mineralogy and Petrology 117, 164–174.[CrossRef][Web of Science]
Fraser, K. J., Hawkesworth, C. J., Erlank, A. J., Mitchell, R. H. & Scott-Smith, B. H. (1985). Sr, Nd, and Pb isotope and minor element geochemistry of lamproites and kimberlites. Earth and Planetary Science Letters 76, 57–70.[CrossRef][Web of Science]
Friel, J. J., Harker, R. I. & Ulmer, G. C. (1977). Armalcolite stability as a function of pressure and oxygen fugacity. Geochimica et Cosmochimica Acta 41, 403–410.[CrossRef][Web of Science]
Gatehouse, B. M., Grey, I. E. & Smyth, J. R. (1983). Structure refinement of mathiasite (K0·62Na0·14Ba0·14Sr0·10)1·0[Ti12·90Cr3·10Mg1·53Fe2·15Zr0·67Ca0·29(V,Nb,Al)0·36]21·0O38. Acta Crystallographica, Section C 39, 421–422.[CrossRef]
Gibb, R. M. & Anderson, J. S. (1972). The system TiO2–Cr2O3: electron microscopy of solid solutions and crystallographic shear structures. Journal of Solid State Chemistry 4, 379–390.[CrossRef]
Grégoire, M., Moine, B. N., O'Reilly, S., Cottin, J. Y. & Giret, A. (2000a). Trace element residence and partitioning in mantle xenoliths metasomatized by highly alkaline, silicate- and carbonate-rich melts (Kerguelen Islands, Indian Ocean). Journal of Petrology 41, 477–509.
Grégoire, M., Lorand, J. P., O'Reilly, S. Y. & Cottin, J. Y. (2000b). Armalcolite-bearing, Ti-rich metasomatic assemblages in harzburgitic xenoliths from the Kerguelen Islands: implications for the oceanic mantle budget of high-field strength elements. Geochimica et Cosmochimica Acta 64, 673–694.[CrossRef][Web of Science]
Grégoire, M., Bell, D. R. & LeRoex, A. P. (2002). Trace element geochemistry of phlogopite-rich mafic mantle xenoliths; their classification and their relationship to phlogopite-bearing peridotites and kimberlites. Contributions to Mineralogy and Petrology 142, 603–625.[Web of Science]
Grey, I. E., Reid, A. F. & Allpress, J. G. (1973). Compounds in the system Cr2O3–Fe2O3–TiO2–ZrO2 based on intergrowth of the -PbO2 and V3O5 structural types. Journal of Solid State Chemistry 8, 86–99.
Grey, I. E., Lloyd, D. J. & White, J. S. (1976). The structure of crichtonite and its relationship to senaite. American Mineralogist 61, 1203–1212.[Abstract]
Grey, I. E., Madsen, I. C. & Haggerty, S. E. (1987). Structure of a new upper-mantle, magnetoplumbite-type phase, Ba[Ti3Cr4Fe4Mg]O19. American Mineralogist 72, 633–636.[Abstract]
Grey, I. E., Velde, D. & Criddle, A. J. (1998). Haggertyite, a new magnetoplumbite-type mineral from the Prairie Creek (Arkansas) lamproite. American Mineralogist 83, 1323–1329.[Abstract]
Haggerty, S. E. (1983). The mineral chemistry of new titanates from the Jagersfontein kimberlite, South Africa, implications for matasomatism in the upper mantle. Geochimica et Cosmochimica Acta 47, 1833–1854.[CrossRef][Web of Science]
Haggerty, S. E. (1987). Metasomatic mineral titanates in upper mantle xenoliths. In: Nixon, P. H. (ed.) Mantle Xenoliths. New York: John Wiley, pp. 671–690.
Haggerty, S. E. (1991). Oxide textures—a mini atlas. In: Lindsley, D. H. (ed.) Oxide Minerals: Petrologic and Magnetic Significance. Mineralogical Society of America, Reviews in Mineralogy 25, 130–218.
Haggerty, S. E., Smyth, J. R., Erlank, A. J., Rickard, R. S. & Danchin, R. V. (1983). Lindsleyite (Ba) and mathiasite (K): two new chromium-titanates in the crichtonite series from the upper mantle. American Mineralogist 68, 494–505.[Abstract]
Haggerty, S. E., Grey, I. E., Madsen, I. C., Criddle, A. J., Stanley, C. J. & Erlank, A. J. (1989). Hawthorneite, Ba[Ti3Cr4Fe4Mg]O19: a new metasomatic magnetoplumbite-type mineral from the upper mantle. American Mineralogist 74, 668–675.[Abstract]
Harlow, G. E. (1997). K in clinopyroxene at high pressure and temperature: an experimental study. American Mineralogist 82, 259–269.[Abstract]
Harte, B. (1987). Metasomatic events recorded in mantle xenoliths: an overview. In: Nixon, P. H. (ed.) Mantle Xenoliths. New York: John Wiley, pp. 625–640.
Harte, B., Winterburn, P. A. & Gurney, J. J. (1987). Metasomatic and enrichment phenomena in garnet peridotite facies mantle xenoliths from the Matsoku Kimberlite pipe, Lesotho. In: Menzies, M. A. & Hawkesworth, C. J. (eds) Mantle Metasomatism. London: Academic Press, pp. 145–219.
Jaques, A. L., Hall, A. E., Sheraton, J. W., Smith, C. B., Sun, S., Drew, R. M., Foudoulis, C. & Ellingsen, K. (1989). Composition of crystalline inclusions and C-isotopic composition of Argyle and Ellendale diamonds. In: Jaques, A. L., Ferguson, J. & Green, D. H. (eds) Kimberlites and Related Rocks 2: Their Mantle/Crust Setting. Geological Society of Australia Special Publication 14, 966–989.
Jaques, A. L., O'Neill, H. St. C., Smith, C. B., Moon, J. & Chappell, B. W. (1990). Diamondiferous peridotite xenoliths from the Argyle (AK1) lamproite pipe, Western Australia. Contributions to Mineralogy and Petrology 104, 255–276.[CrossRef][Web of Science]
Jones, A. P. & Ekambaram, V. (1985). New INAA analysis of a mantle derived titanate mineral of the crichtonite series, with particular reference to the rare earth elements. American Mineralogist 70, 414–418.[Abstract]
Jones, A. P., Smith, J. V. & Dawson, J. B. (1982). Mantle metasomatism in 14 veined peridotites from Bultfontein mine, South Africa. Journal of Geology 90, 435–453.[Web of Science]
Kempton, P. D. (1987). Mineralogical and geochemical evidence for differing styles of metasomatism in spinel lherzolite xenoliths: enriched mantle source regions of basalts? In: Menzies, M. A. & Hawkesworth, C. J. (eds) Mantle Metasomatism. London: Academic Press, pp. 45–89.
Kiviets, G., Phillips, D., Shee, S. R., Vercoe, S. C., Barton, E. S., Smith, C. B. & Fourie, L. F. (1998). 40Ar/39Ar dating of yimengite from the Turkey Well kimberlite, Australia: the oldest and the rarest. In: Extended Abstracts of the 7th International Kimberlite Conference. Cape Town: University of Cape Town, pp. 432–433.
Kohn, J. A. & Eckart, D. W. (1965). Mixed-layer polytypes related to magnetoplumbite. American Mineralogist 50, 1371–1380.[Web of Science]
Konzett, J. (1997). Phase relations and chemistry of Ti-rich K-richterite-bearing mantle assemblages: an experimental study to 8·0 GPa in a Ti–KNCMASH system. Contributions to Mineralogy and Petrology 128, 385–404.[CrossRef][Web of Science]
Konzett, J. & Fei, Y. (2000). Transport and storage of potassium in the Earth's upper mantle and transition zone: an experimental study to 23 GPa in simplified and natural bulk compositions. Journal of Petrology 41, 583–603.
Konzett, J. & Ulmer, P. (1999). The stability of hydrous potassic phases in lherzolitic mantle—an experimental study to 9·5 GPa in simplified and natural bulk compositions. Journal of Petrology 40, 629–652.[CrossRef][Web of Science]
Konzett, J. & Yang, H. (2001). Phase relations and crystal chemistry of complex titanates under upper mantle P–T conditions. Journal of Conference Abstracts 6(1), 579.
Konzett, J., Sweeney, R. J., Thompson, A. B. & Ulmer, P. (1997). Potassium amphibole stability in the upper mantle: an experimental study in a peralkaline KNCMASH system to 8·5 GPa. Journal of Petrology 38, 537–568.[CrossRef][Web of Science]
Konzett, J., Armstrong, R. A. & Günther, D. (2000). Modal metasomatism in the Kaapvaal craton lithosphere: constraints on timing and genesis from U–Pb zircon dating of metasomatized peridotites and MARID-type xenoliths. Contributions to Mineralogy and Petrology 139, 704–719.[CrossRef]
Kopylova, M. G., Rickard, R. S., Kleyenstueber, A., Taylor, W. R., Gurney, J. J. & Daniels, L. R. M. (1997). First occurrence of strontian K–Cr-loparite and Cr-chevkinite in diamonds. Russian Geology and Geophysics 38, 405–420.
Kostrovitsky, S. I. & Garanin, V. K. (1991). Chrome titanate inclusions of unusual composition in pyropes from lamprophyres and kimberlites. In: Extended Abstracts of the 5th International Kimberlite Conference, pp. 525–526.
Krauskopf, K. B. (1988). Geology of high-level nuclear waste disposal. Annual Review of Earth and Planetary Sciences 16, 173–200.[CrossRef][Web of Science]
Leost, I., Stachel, T., Brey, G. P., Harris, J. W. & Ryabchikov, I. D. (2003). Diamond formation and source carbonation: mineral associations in diamonds from Namibia. Contributions to Mineralogy and Petrology 145, 15–24.[Web of Science]
Linde, K. & DeCarli, P. S. (1969). Polymorphic behaviour of titania under dynamic loading. Journal of Chemical Physics 50, 319–325.[CrossRef]
Liu, L.-G. & Mernagh, T. P. (1992). Phase transitions and Raman spectra of anatase and rutile at high pressures and room temperature. European Journal of Mineralogy 4, 45–52.
Mammone, J. F. & Sharma, S. K. (1979). Raman study of TiO2 under high pressures at room temperature. Carnegie Institution of Washington Yearbook 78, 636–640.
McDonough, W. F. (1994). Chemical and isotopic systematics of continental lithospheric mantle. In: Meyer, H. O. A. & Leonardos, O. H. (eds) Proceedings of the 5th International Kimberlite Conference. Brasilia: Companhia de Pesquisa de Recursos Minerais, pp. 478–485.
McDonough, W. F. & Rudnick, R. L. (1998). Mineralogy and composition of the upper mantle. In: Hemley, R. J. (ed.) Ultrahigh-Pressure Mineralogy: Physics and Chemistry of the Earth's Deep Interior. Mineralogical Society of America, Reviews in Mineralogy 37, 139–164.
McDonough, W. F. & Sun, S.-s. (1995). The composition of the Earth. Chemical Geology 120, 223–253.[CrossRef][Web of Science]
Menzies, M. A. & Hawkesworth, C. J. (eds) (1987). Mantle Metasomatism. London: Academic Press, 472 pp.
Menzies, M. A., Rogers, N., Tindle, A. & Hawkesworth, C. J. (1987). Metasomatic and enrichment processes in lithospheric peridotites, an effect of asthenosphere–lithosphere interaction. In: Menzies, M. A. & Hawkesworth, C. J. (eds) Mantle Metasomatism. London: Academic Press, pp. 313–359.
Mitchell, R. H. (1977). Geochemistry of magnesian ilmenites from kimberlites in South Africa and Lesotho. Lithos 10, 29–37.[CrossRef][Web of Science]
Mitchell, R. H. (1991). Kimberlites and lamproites: primary sources of diamonds. Geoscience Canada 18, 1–16.[Web of Science]
Mitchell, R. H. (1995). Kimberlites, Orangeites and Related Rocks. New York: Plenum, 410 pp.
Mitchell, R. H. & Bergman, S. C. (1991). The Petrology of Lamproites. New York: Plenum, 447 pp.
Mitchell, R. H. & Lewis, R. D. (1983). Priderite-bearing xenoliths from the Prairie Creek mica peridotite, Arkansas. Canadian Mineralogist 21, 59–64.
Mougin, J., Le Bihan, T. & Lucazeau, G. (2001). High-pressure study of Cr2O3 obtained by high-temperature oxidation by X-ray diffraction and Raman spectroscopy. Journal of Physics and Chemistry of Solids 62, 553–563.[CrossRef]
Nixon, P. H. & Condliffe, E. (1989). Yimengite of K–Ti metasomatic origin in kimberlitic rocks from Venezuela. Mineralogical Magazine 53, 305–309.
Obradors, X., Collomb, A., Pernet, M., Samaras, D. & Joubert, J. C. (1985). X-ray analysis of the structural and dynamic properties of BaFe12O19, hexagonal ferrite at room temperatures. Journal of Solid State Chemistry 56, 171–181.[CrossRef]
Peltonen, P., Huhma, H., Tyni, M. & Shimizu, N. (1998). Garnet peridotite xenoliths from kimberlites of Finland: nature of the continental mantle at an Archaean Craton–Proterozoic mobile belt transition. In: Gurney, J. J., Gurney, J. L., Pascoe, M. & Richardson, S. H. (eds) Proceedings of the 7th International Kimberlite Conference. Cape Town: Red Roof Design, pp. 664–676.
Peterson, R. C. & Grey, I. E. (1995). Preparation and structure refinement of synthetic Ti3+-containing lindsleyite, BaMn3Ti18O39. Canadian Mineralogist 33, 1083–1089.
Podpora, C. & Lindsley, D. H. (1984). Lindsleyite and mathiasite: synthesis of chromium-titanates in the crichtonite (A1M21O38) series. EOS Transactions, American Geophysical Union 65, 293.
Rouse, R. C. & Peacor, D. R. (1968). The relationship between senaite, magnetoplumbite and davidite. American Mineralogist 53, 869–879.[Web of Science]
Rubie, D. C., Karato, S., Yan, H. & O'Neill, H. St. C. (1993). Low differential stress and controlled chemical environment in multi-anvil high pressure devices. Physics and Chemistry of Minerals 20, 315–322.[Web of Science]
Rudnick, R. L., McDonough, W. F. & O'Connell, R. J. (1998). Thermal structure, thickness and composition of continental lithosphere. Chemical Geology 145, 399–415.
Schulze, D. J., Anderson, P. F. N., Hearn, B. C. & Hetman-Casey, M. (1995). Origin and significance of ilmenite megacrysts and macrocrysts from kimberlite. International Geology Review 37, 780–812.
Sobolev, N. V. & Yefimova, E. S. (2000). Composition and petrogenesis of Ti-oxides associated with diamonds. International Geology Review 42, 758–767.[Web of Science]
Sobolev, N. V., Yefimova, E. S., Kaminskiy, F. V., Lavrientiev, Y. G. & Usova, L. V. (1988). Titanate of complex composition and phlogopite in the diamond stability field. In: Composition and Processes of Deep-seated Zones of Continental Lithosphere. Novosibirsk: Nauka, pp. 185–186.
Sobolev, N. V., Kaminsky, F. V., Griffin, W. L., Yefimova, E. S., Win, T. T., Ryan, C. G. & Botkunov, A. I. (1997). Mineral inclusions in diamonds from the Sputnik kimberlite pipe, Yakutia. Lithos 39, 135–157.[CrossRef][Web of Science]
Sobolev, N. V., Yefimova, E. S. & Koptil, V. I. (1998). Mineral inclusions in diamonds in the northeast of the Yakutian diamondiferous province. In: Gurney, J. J., Gurney, J. L., Pascoe, M. & Richardson, S. H. (eds) Proceedings of the 7th International Kimberlite Conference. Cape Town: Red Roof Design, pp. 816–822.
Stachel, T., Harris, J. W. & Brey, G. P. (1998). Rare and unusual mineral inclusions in diamonds from Mwadui, Tanzania. Contributions to Mineralogy and Petrology 132, 34–47.[CrossRef][Web of Science]
Tainton, K. M. & McKenzie, D. (1994). The generation of kimberlites, lamproites, and their source rocks. Journal of Petrology 35, 787–817.
Thompson, J. B., Jr (1978). Biopyriboles and polysomatic series. American Mineralogist 63, 239–249.[Abstract]
Tollo, R. P. & Haggerty, S. E. (1987). Nb–Cr-rutile in the Orapa Kimberlite, Botswana. Canadian Mineralogist 25, 251–264.[Web of Science]
Townes, W. D., Fang, J. H. & Perrotta, A. J. (1967). The crystal structure and refinement of ferrimagnetic barium ferrite of BaFe12O19. Zeitschrift für Kristallographie 125, 437–449.[Web of Science]
Varlamov, D. A., Garanin, V. K. & Kostrovitskiy, S. I. (1996). Exotic high-titanium minerals as inclusions in garnets from lower crustal and mantle xenoliths. Transactions of the Russian Academy of Science, Earth Science Section 345A, 352–355.
Velde, D. (2000). Mineralogy of mafic xenoliths and their reaction zones in the olivine lamproite from Prairie Creek Arkansas and the paragenesis of haggertyite, Ba[Fe6Ti5Mg]O19. American Mineralogist 85, 420–429.
Walker, D. (1991). Lubrication, gasketing, and precision in multianvil experiments. American Mineralogist 76, 1092–1100.[Abstract]
Wang, L., Essene, E. J. & Zhang, Y. (1999). Mineral inclusions in pyrope crystals from Garnet Ridge, Arizona, USA: implications for processes in the upper mantle. Contributions to Mineralogy and Petrology 135, 164–178.[CrossRef][Web of Science]
Waters, F. G. & Erlank, A. J. (1988). Assessment of the vertical extent and distribution of mantle metasomatism below Kimberley, South Africa. Journal of Petrology, Special Lithosphere Issue, 185–204.
Withers, A. C., Essene, E. J. & Zhang, Y. (2003). Rutile/TiO2-II phase equilibria. Contributions to Mineralogy and Petrology 145, 199–204.[Web of Science]
Yang, H. & Hazen, R. M. (1998). Crystal chemistry of cation order–disorder in pseudobrookite-type MgTi2O5. Journal of Solid State Chemistry 138, 238–244.[CrossRef]
Zhang, R. Y., Zhai, S. M., Fei, Y. M. & Liou, J. G. (2003). Titanium solubility in coexisting garnet and clinopyroxene at very high pressure: the significance of exsolved rutile in garnet. Earth and Planetary Science Letters 216, 591–601.[CrossRef][Web of Science]
Zhou, J., Yang, G. & Zhang, J. (1984). Mathiasite in kimberlite of China. Acta Mineralogica Sinica 9, 193–200.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||