Patterns in the hydrogen and trace element compositions of mantle olivines

 The concentrations of hydrogen and the other trace elements in olivines from mantle xenoliths have been determined by secondary ion mass spectrometry (SIMS) for clarifying the incorporation mechanism and the behavior of the hydrogen. The hydrogen contents in olivines from mantle xenoliths range from 10 to 60 ppm wt. H₂O and the concentration range is consistent with the previous infrared (IR) spectroscopic data. IR spectra of the olivine crystals show no effects of the weathering or secondary alteration. The hydrogen is distributed homogeneously among olivine grains in each mantle xenolith. However, the hydrogen contents of the olivine crystals are less than those for the olivine phenocrysts crystallized from the host magma. Olivine inclusions in diamonds also show similar hydrogen contents to the xenolithic olivines. Thus the hydrogen content of xenolithic olivines does not attain equilibrium with water in the host magma during the transportation from the Earth's mantle to the surface, and is taken as a reflection of the hydrogen condition in the mantle. Correlations of hydrogen with trivalent cation contents in garnet peridotitic olivines indicate the incorporation of hydrogen into mantle olivines by a coupled substitution mechanism, with the hydrogen present in the form of hydroxyl in oxygen positions adjacent to the M site vacancies. The hydrogen content of xenolithic olivines increases with pressure but decreases with increasing temperature, suggesting importance of olivine as a water reservoir at low temperature regions such as in subducting slabs. Received August 15, 1995/Revised, accepted November 19, 1996     

Hydroxyl in mantle olivine xenocrysts from the Udachnaya kimberlite pipe

The incorporation of hydrogen in mantle olivine xenocrysts from the Udachnaya kimberlite pipe was investigated by Fourier-transform infrared spectroscopy and secondary ion mass spectrometry (SIMS). IR spectra were collected in the OH stretching region on oriented single crystals using a conventional IR source at ambient conditions and in situ at temperatures down to −180°C as well as with IR synchrotron radiation. The IR spectra of the samples are complex containing more than 20 strongly polarized OH bands in the range 3,730–3,330 cm⁻¹. Bands at high energies (3,730–3,670 cm⁻¹) were assigned to inclusions of serpentine, talc and the 10 Å phase. All other bands are believed to be intrinsic to olivine. The corresponding point defects are (a) associated with vacant Si sites (3,607 cm⁻¹ E || a, 3,597 E || a, 3,571 cm⁻¹ E || c, 3,567 E || c, and 3,556 E || b), and (b) with vacant M1 sites (most of the bands polarized parallel to a). From the pleochroic behavior and position of the OH bands associated with the vacant M1 sites, we propose two types of hydrogen—one bonded to O1 and another to O2, so that both OH vectors are strongly aligned parallel to a. The O2–H groups may be responsible for the OH bands at higher wavenumbers than those for the O1–H groups. The multiplicity of the corresponding OH bands in the spectra can be explained by different chemical environments and by slightly different distortions of the M1 sites in these high-pressure olivines. Four samples were investigated by SIMS. The calculated integral molar absorption coefficient using the IR and SIMS results of 37,500±5,000 L mol H₂O cm⁻² is within the uncertainties slightly higher than the value determined by Bell et al. (J Geophys Res 108(B2):2105–2113, 2003) (28,450±1,830 L mol H₂O cm⁻²). The reason for the difference is the different distributions of the absorption intensity of the spectra of both studies (mean wavenumber 3,548 vs. 3,570 cm⁻¹). Olivine samples with a mean wavenumber of about 3,548 cm⁻¹ should be quantified with the absorption coefficient as determined in this study; those containing more bands at higher wavenumber (mean wavenumber 3,570 cm⁻¹) should be quantified using the value determined by Bell et al. (J Geophys Res 108(B2):2105–2113, 2003).

Hydroxyl in olivines from mantle xenoliths in kimberlites of the Siberian platform

From a total of 335 olivine crystal grains, crystallographically orientated platelets and, where possible, parallelepipeds were prepared, chemically analysed by electron microprobe, examined under the polarisation microscope, and studied by polarised FTIR microscope-absorption-spectrometry in the _OH vibrational range, 3,000–3,800 cm^–1. The 335 crystal grains were extracted from 174 different specimens of Yakutian upper mantle material, including 97 xenoliths that represent all the rock types occurring in all the kimberlites of the Siberian shield. The other specimens were mega- and macrocrysts or inclusions in diamonds and garnets. Analysis of the polarised _OH-spectra allowed distinction between hydroxyl in non-intrinsic separate inclusions, NSI, and in isolated local or condensed extended defects, intrinsic to the olivines, ILD or CED, respectively. As the two latter types cannot be distinguished by vibrational spectroscopy, and as they are presumably interconnected by intracrystalline condensation reactions of the type n [ILD][CED]_n, we propose to symbolise them as [ILD/CED]. Of the total of 70 polarised _OH-bands that were found in the whole set, 17 are caused by NSI, 53 by [ILD/CED]. Total mean integrated _OH-band intensities, (̄_int)_total with ̄_int=(_||a+_||b+_||c)_int/3, were determined from the spectra. They yielded the contents of structurally unallocated water, using the recent calibration of the IR-method (Bell et al. 2003). The range is 0<wt. ppm (H₂O)_total<419. Olivines included in diamonds were found to be free of hydroxyl (detection limit of the single crystal IR-spectrometry, ca. 1 wt. ppm water). The total water contents of the different types of olivines increase in the sequence groundmass crystals < megacrysts < macrocrysts. NSI are: (1) Serpentine plus talc with _OH in the range 3,704–3,657 cm^–1, either polarised along a of the olivine matrix (Pbnm setting) or unpolarised. Approximately 232 olivines out of the 335 contain such NSI. Serpentine and talc occur mostly together, in rare cases one of them alone and if so, mostly talc. (2) Mg-edenite or Mg-pargasite occur rarely and with _OH at 3,709–3,711 cm^–1. NIS types (1) and (2) are presumably formed by metasomatic alterations of the host olivines, the orientated ones probably in the mantle, the unorientated ones during later stages. (3) The spectra of 23 olivine crystals, displayed specifically a _OH-band, polarised c>a>b, at 3,327–3,328 cm^–1, an energy typical of _OH in hydrous wadsleyite. We assume this phase to be present as NIS in the respective olivines, possibly as relic phase. (4) Weak bands between 3,175 and 3,260 cm^–1 polarised along c, are tentatively assigned to molecular water NSI with relatively strong hydrogen bonds to the matrix. We did not find larger clusters of molecular water, i.e. liquid-like water with its characteristic broad band centred at ca. 3,400 cm^–1. We did also not find any humite minerals as an NSI. Of the 53 _OH-bands intrinsic to olivine, the 29 most abundant and strong ones were subject to further analysis in terms of OH^–-bearing structural defects [ILD/CED]. Nearly all these bands are strongly polarised along a. Two bands at 3,672 and 3,535 cm^–1 are assigned to boron-related defects, [ILD/CED]_B. Five bands at 3,573, 3,563, 3,541, 3,524 and 3,512 cm^–1 are intensity-correlated and are assigned to Si-depleted titan-clinohumite-like defects, [ILD/CED]_(thl). The other, so far unidentified _OH of [ILD/CED] are suggested to originate from OH^– in different types of (Mg, Fe)-depleted defects recently predicted and discovered by TEM. These are called [ILD/CED]_(KWK). Eight mostly strong bands of them occur at energies higher than 3,573 cm^–1, [ILD/CED]_(KWK)-H, 13 strong to medium strong bands occur below 3,500 cm^–1, [ILD/CED]_(KWK)-L. Such intrinsic defects may occur alone, [ILD/CED]_(thl) and [ILD/CED]_(KWK)-H, or in different combinations with each other, [ILD/CED]_(KWK)-H+[ILD/CED]_(thl), [ILD/CED]_(KWK)-H+[ILD/CED]_(KWK)-L and [ILD/CED]_(KWK)-H+[ILD/CED]_(thl)+[ILD/CED]_(KWK)-L. Though there are indications that the occurrences of such types and combinations of the intrinsic OH^–-bearing defects in the olivines are related to the types and genetic peculiarities of their host rocks, straightforward and simple correlations do not exist. The reasons for this and also for the great number of varieties of intrinsic [ILD/CED] are discussed.Editorial responsibility: J. Hoefs

---

Velocity structure of the continental upper mantle: evidence from southern Africa

The velocity model for southern Africa of Qiu et al. [Qiu, X., Priestley, K., McKenzie, D., 1996. Average lithospheric structure of southern Africa. Geophys. J. Int. 127, 563–587] is revised so as to satisfy both the regional seismic waveform data and the fundamental mode Rayleigh wave phase velocity data for the region. The revised S-wave model is similar to the original model of Qiu et al. except that the high velocity, upper mantle lid extends to 160 km depth in the revised model rather than to 120 km in the original model. Sensitivity tests of the regional seismic data show that the minimum velocity in the S-wave low velocity zone can be as high as 4.45 km s⁻¹ compared to 4.32 km s⁻¹ in the Qiu et al. model. The vertical S-wave travel time for the revised south African model is compared with the vertical S-wave travel times for the global tomographic models S12WM13 and S16B30, and they are found to be similar.     

Mineralogy of the upper mantle: A review of the minerals in mantle xenoliths from kimberlite

The importance of ultramafic and eclogitic xenoliths in kimberlite as representing the rocks and minerals of the upper mantle has been widely perceived during the last decade. Studies of the petrology and mineral chemistry of these mantle fragments as well as of inclusions in diamond, have led to significant progress in our understanding of the mineralogy and chemistry of the upper mantle. For example, it is now known that textural differences in the ultramafic xenoliths (lherzolite, harzburgite, pyroxenite and websterite) are partially reflected in chemical differences. Thus xenoliths that display a ‘fluidal’ texture, indicative of intense deformation are less depleted in Ca, Al, Na, Fe and Ti than those xenoliths in which granular textures are predominant. It is believed this relative depletion may indicate the sheared (fluidal texture) xenoliths are representative of primary, undifferentiated mantle. This material on partial melting would produce ‘basaltic-type’ material, and leave a residuum whose chemistry and mineralogy is reflected by the granular xenoliths.Also present in kimberlite are large single phase xenoliths that may be either one single crystal (xenocryst, megacryst) or an aggregate of several crystals of the same mineral (discrete xenolith, or discrete nodule). These large single phase samples consist of similar minerals to those occurring in the ultramafic xenoliths but chemically they are distinct in being generally more Fe-rich. The relation between these xenocrysts to their counterparts in the ultramafic xenoliths is unknown. Also unknown, at the present time, is the exact relation between diamond and kimberlite. Evidence obtained from study of the mineral inclusions in diamond suggests that diamond forms in at least two chemically distinct environments in the mantle; one eclogitic, the other, ultramafic. Interestingly, this suggestion is true for diamonds from worldwide localities.The mineral-chemical results of studies on xenoliths and inclusions in diamond have been convincingly interpreted in the light of experimental studies. It is now possible based on several different geothermometers and barometers to determine relatively reasonable physical conditions for the final genesis of many of these mantle rocks. For the most part the final equilibration temperatures range between 1000 and 1400°C and pressure in the region 100–200 km. These conditions are consistent with an upper mantle origin. Future studies will undoubtedly attempt to more concisely, and accurately, define these conditions, as well as understand better the chemical and spatial relationship of the rock-types in the mantle.     

The mineral chemistry of new titanates from the jagersfontein kimberlite,South Africa: Implications for metasomatism in the upper mantle

New members of the crichtonite mineral series are described in which K, Ba, Ca and REE are in significant concentrations (5 wt% oxides) filling the A formula position in AM₂₁O₃₈. These phases are chromium (16 wt% Cr₂O₃) titanates (58 wt% TiO₂) enriched in ZrO₂ (5 wt%) and constitute a mineral repository for refractory and large ion lithophile elements in the upper mantle. The mineral senes coexists with Mg-Cr-ilmenite, Nb-Cr-rutile, and Ca-Cr (NbZr) armalcolite that have equally unusual chemistries. Kimberlitic crichtonites are depleted in the intermediate lanthanides but highly enriched in LREE and HREE with chondrite normalized abundances of 10³ to 10⁵. Crichtonite, armalcolite, and Nb-Cr-rutile occupy a compositional range in TiO₂ contents bridging the gap between ilmenite and rutile, two minerals having a widespread distribution in kimberlites and mantle-derived nodule suites.In common with other associations, and based on similarities in mineral chemistry, it is concluded that these minerals formed at P = 20–30 kb, 900–1100°C by reaction of peridotite with metasomatizing fluids. Kimberlitic crichtonite may be expressed as spinel + Cr-ferropseudobrookite, and armalcolite is equivalent to Cr-geikielite + rutile in the system (FeMg)-TiO₂-Cr₂O₃. This system contains a number of Cr-Ti compounds not found as minerals but it is proposed that the ubiquitous occurrence of ilmenite intergrowths in kimberlitic rutile results from decomposition of high pressure αPbO₂-type crystallographic shear structures. The new minerals have exotic chemistries and the high K-affinities broaden the scope for the origin of alkalic rocks, the generation of highly potassic magmas in the upper mantle, and suggest that alkali metasomatism may be pervasive.     

Phlogopite in mantle xenoliths and kimberlite from the Grib pipe,Arkhangelsk province,Russia:Evidence for multi-stage mantle metasomatism and origin of phlogopite in kimberlite

We present petrography and mineral chemistry for both phlogopite,from mantle-derived xenoliths(garnet peridotite,eclogite and clinopyroxene-phlogopite rocks)and for megacryst,macrocryst and groundmass flakes from the Grib kimberlite in the Arkhangelsk diamond province of Russia to provide new insights into multi-stage metasomatism in the subcratonic lithospheric mantle(SCLM)and the origin of phlogopite in kimberlite.Based on the analysed xenoliths,phlogopite is characterized by several generations.The first generation(Phil)occurs as coarse,discrete grains within garnet peridotite and eclogite xenoliths and as a rock-forming mineral within clinopyroxene-phlogopite xenoliths.The second phlogopite generation(Phl2)occurs as rims and outer zones that surround the Phil grains and as fine flakes within kimberlite-related veinlets filled with carbonate,serpentine,chlorite and spinel.In garnet peridotite xenoliths,phlogopite occurs as overgrowths surrounding garnet porphyroblasts,within which phlogopite is associated with Cr-spinel and minor carbonate.In eclogite xenoliths,phlogopite occasionally associates with carbonate bearing veinlet networks.Phlogopite,from the kimberlite,occurs as megacrysts,macrocrysts,microcrysts and fine flakes in the groundmass and matrix of kimberlitic pyroclasts.Most phlogopite grains within the kimberlite are characterised by signs of deformation and form partly fragmented grains,which indicates that they are the disintegrated fragments of previously larger grains.Phil,within the garnet peridotite and clinopyroxene-phlogopite xenoliths,is characterised by low Ti and Cr contents(TiO_21 wt.%,Cr_2 O_31 wt.% and Mg# = 100 × Mg/(Mg+ Fe)92)typical of primary peridotite phlogopite in mantle peridotite xenoliths from global kimberlite occurrences.They formed during SCLM metasomatism that led to a transformation from garnet peridotite to clinopyroxene-phlogopite rocks and the crystallisation of phlogopite and high-Cr clinopyroxene megacrysts before the generation of host-kimberlite magmas.One of the possible processes to generate low-Ti-Cr phlogopite is via the replacement of garnet during its interaction with a metasomatic agent enriched in K and H_2O.Rb-Sr isotopic data indicates that the metasomatic agent had a contribution of more radiogenic source than the host-kimberlite magma.Compared with peridotite xenoliths,eclogite xenoliths feature low-Ti phlogopites that are depleted in Cr_2O_3 despite a wider range of TiO_2 concentrations.The presence of phlogopite in eclogite xenoliths indicates that metasomatic processes affected peridotite as well as eclogite within the SCLM beneath the Grib kimberlite.Phl2 has high Ti and Cr concentrations(TiO_22 wt.%,Cr_2O_31 wt.% and Mg# = 100× Mg/(Mg + Fe)92)and compositionally overlaps with phlogopite from polymict brecc:ia xenoliths that occur in global kimberlite formations.These phlogopites are the product of kimberlitic magma and mantle rock interaction at mantle depths where Phl2 overgrew Phil grains or crystallized directly from stalled batches of kimberlitic magmas.Megacrysts,most macrocrysts and microcrysts are disintegrated phlogopite fragments from metasomatised peridotite and eclogite xenoliths.Fine phlogopite flakes within kimberlite groundmass represent mixing of high-Ti-Cr phlogopite antecrysts and high-Ti and low-Cr kimberlitic phlogopite with high Al and Ba contents that may have formed individual grains or overgrown antecrysts.Based on the results of this study,we propose a schematic model of SCLM metasomatism involving phlogopite crystallization,megacryst formation,and genesis of kimberlite magmas as recorded by the Grib pipe.     

Post-emplacement serpentinization and related hydrothermal metamorphism in a kimberlite from Venetia, South Africa

The common serpentine–diopside matrix assemblage in volcaniclastic kimberlite (VK) at the Venetia Mine, South Africa is ascribed to a secondary origin, because of post‐emplacement serpentinization and associated hydrothermal metamorphism. Volcaniclastic deposits with 20–30% porosity infill kimberlite pipes in the waning stages of kimberlite eruptions. Olivine macrocrysts are typically rimmed by talc and are pseudomorphed by lizardite, with minor magnetite. The fine matrix consists of mixtures of lizardite, chlorite, smectite, brucite, calcite, titanite and andradite, an assemblage which either pseudomorphed microcrysts or in‐filled voids. Locally we recognize microcryst pseudomorphs rich in sub‐microscopic mixtures of lizardite with smectite, and other microcryst pseudomorphs and void‐filling matrix rich in chlorite and lizardite. Interstitial lizardite and associated phyllosilicates (brucite, smectite and chlorite) crystallized progressively from meteoric or hydrothermally derived pore waters, and Si⁴⁺ and Mg²⁺ released into the fluid phase during serpentinization of olivine macrocrysts. Radial‐fibrous fringes of diopside microlites around crystals display void‐filling textures because of unrestricted growth into pore spaces. Secondary diopside is attributed to Si⁴⁺, Mg²⁺ and Ca²⁺ cations released into the fluid phase by interaction with olivine, calcite and plagioclase in siliceous xenoliths. The paucity of primary, fine‐grained groundmass phases resistant to alteration, for example, perovskite and spinel, precludes an origin for the intergrain matrix as altered interstitial ash, glass or a late‐stage kimberlite melt. Isovolumetric replacement of olivine results in a volume increase of 60% so that pore spaces in the original deposit can be easily filled up with serpentine. The source of Al³⁺ to form chlorite and smectite is attributed to alteration of plagioclase in xenoliths which comprise 20–30 vol.% of the deposit. Titanite, hydro‐andradite and second‐generation diopside precipitate as hydrothermal minerals from calcium‐bearing serpentinizing fluids in replacement reactions and as void‐filling minerals. Consideration of mineral equilibria in the CaO‐MgO‐SiO₂‐H₂O‐CO₂ system constrains the common matrix assemblage of lizardite and diopside in X_CO2)–T space. At 300 bar, the assemblage is stable only at temperatures below 370 °C and X_CO2 < 0.01. This upper limit on temperature is well below the plausible solidus of ultrabasic magmas. Furthermore, the requirement of trace CO₂ in the fluid phase implies a post‐emplacement external source rather than ‘autometamorphism’ from kimberlite‐derived fluids, because of high P_CO2 commonly inferred for kimberlite magmas.     

Density structure of the cratonic mantle in southern Africa: 1. Implications for dynamic topography

The origin of high topography in southern Africa is enigmatic. By comparing topography in different cratons, we demonstrate that in southern Africa both the Archean and Proterozoic blocks have surface elevation 500–700 m higher than in any other craton worldwide, except for the Tanzanian Craton. An unusually high topography may be caused by a low density (high depletion) of the cratonic lithospheric mantle and/or by the dynamic support of the mantle with origin below the depth of isostatic compensation (assumed here to be at the lithosphere base). We use free-board constraints to examine the relative contributions of the both factors to surface topography in the cratons of southern Africa. Our analysis takes advantage of the SASE seismic experiment which provided high resolution regional models of the crustal thickness.We calculate the model of density structure of the lithospheric mantle in southern Africa and show that it has an overall agreement with xenolith-based data for lithospheric terranes of different ages. Density of lithospheric mantle has significant short-wavelength variations in all tectonic blocks of southern Africa and has typical SPT values of ca. 3.37–3.41 g/cm³ in the Cape Fold and Namaqua–Natal fold belts, ca. 3.34–3.35 g/cm³ in the Proterozoic Okwa block and the Bushveld Intrusion Complex, ca. 3.34–3.37 g/cm³ in the Limpopo Belt, and ca. 3.32–3.33 g/cm³ in the Kaapvaal and southern Zimbabwe cratons.The results indicate that 0.5–1.0 km of surface topography, with the most likely value of ca. 0.5 km, cannot be explained by the lithosphere structure within the petrologically permitted range of mantle densities and requires the dynamic (or static) contribution from the sublithospheric mantle. Given a low amplitude of regional free air gravity anomalies (ca. + 20 mGal on average), we propose that mantle residual (dynamic) topography may be associated with the low-density region below the depth of isostatic compensation. A possible candidate is the low velocity layer between the lithospheric base and the mantle transition zone, where a temperature anomaly of 100–200 °C in a ca. 100–150 km thick layer may explain the observed reduction in Vs velocity and may produce ca. 0.5–1.0 km to the regional topographic uplift.     

Titanochondrodite and titanoclinohumite derived from the upper mantle in the Buell Park kimberlite,Arizona, USA

Titanoclinohumite, titanochondrodite, and associated Na-bearing tremolite occurring as crystal fragments in the Buell Park kimberlite are not likely crystallization products of a kimberlite magma. They more likely formed as phases in hydration assemblages of peridotite at temperatures below 700 ° C and pressures below 18 kbar. The crystals were dispersed in kimberlite as rock fragments were comminuted during transport to the surface.     

Petrology and geochemistry of diamondiferous Mesoproterozoic kimberlites from Wajrakarur kimberlite field,Eastern Dharwar craton,southern India: genesis and constraints on mantle source regions

The petrology and geochemistry of some new occurrences of Mesoproterozoic diamondiferous hypabyssal-facies kimberlites from the Chigicherla, Wajrakarur-Lattavaram and Kalyandurg clusters of the Wajrakarur kimberlite field (WKF), Eastern Dharwar craton (EDC), southern India, are reported. The kimberlites contain two generations of olivine, and multiple groundmass phases including phlogopite, spinel, calcite, dolomite, apatite, perovskite, apatite and rare titanite, and xenocrysts of eclogitic garnet and picro-ilmenite. Since many of the silicate minerals in these kimberlites have been subjected to carbonisation and alteration, the compositions of the groundmass oxide minerals play a crucial role in their characterisation and in understanding melt compositions. While there is no evidence for significant crustal contamination in these kimberlites, some limited effects of ilmenite entrainment are evident in samples from the Kalyandurg cluster. Geochemical studies reveal that the WKF kimberlites are less differentiated and more primitive than those from the Narayanpet kimberlite field (NKF), Eastern Dharwar craton. Highly fractionated (La/Yb = 108–145) chondrite-normalised distribution patterns with La abundances of 500–1,000 × chondrite and low heavy rare earth elements (HREE) abundances of 5–10 × chondrite are characteristic of these rocks. Metasomatism by percolating melts from the convecting mantle, rather than by subduction-related processes, is inferred to have occurred in their source regions based on incompatible element signatures. While the majority of the Eastern Dharwar craton kimberlites are similar to the Group I kimberlites of southern Africa in terms of petrology, geochemistry and Sr–Nd isotope systematics, others show the geochemical traits of Group II kimberlites or an overlap between Group I and II kimberlites. Rare earth element (REE)-based semi-quantitative forward modelling of batch melting of southern African Group I and II kimberlite source compositions involving a metasomatised garnet lherzolite and very low degrees of partial melting demonstrate that (1) WKF and NKF kimberlites display a relatively far greater range in the degree of melting than those from the on-craton occurrences from southern Africa and are similar to that of world-wide melilitites, (2) different degrees of partial melting of a common source cannot account for the genesis of all the EDC kimberlites, (3) multiple and highly heterogeneous kimberlite sources involve in the sub-continental lithospheric mantle (SCLM) in the Eastern Dharwar craton and (4) WKF and NKF kimberlites generation is a resultant of complex interplay between the heterogeneous sources and their different degrees of partial melting. These observations are consistent with the recent results obtained from inversion modelling of REE concentrations from EDC kimberlites in that both the forward as wells as inverse melting models necessitate a dominantly lithospheric, and not asthenospheric, mantle source regions. The invading metasomatic (enriching) melts percolating from the convecting (asthenosphere) mantle impart an OIB-like isotopic signature to the final melt products.     

Density structure of the cratonic mantle in Southern Africa: 2. Correlations with kimberlite distribution,seismic velocities,and Moho sharpness

We present a new regional model for the depth-averaged density structure of the cratonic lithospheric mantle in southern Africa constrained on a 30′ × 30′ grid and discuss it in relation to regional seismic models for the crust and upper mantle, geochemical data on kimberlite-hosted mantle xenoliths, and data on kimberlite ages and distribution. Our calculations of mantle density are based on free-board constraints, account for mantle contribution to surface topography of ca. 0.5–1.0 km, and have uncertainty ranging from ca. 0.01 g/cm³ for the Archean terrains to ca. 0.03 g/cm³ for the adjacent fold belts. We demonstrate that in southern Africa, the lithospheric mantle has a general trend in mantle density increase from Archean to younger lithospheric terranes. Density of the Kaapvaal mantle is typically cratonic, with a subtle difference between the eastern, more depleted, (3.31–3.33 g/cm³) and the western (3.32–3.34 g/cm³) blocks. The Witwatersrand basin and the Bushveld Intrusion Complex appear as distinct blocks with an increased mantle density (3.34–3.35 g/cm³) with values typical of Proterozoic rather than Archean mantle. We attribute a significantly increased mantle density in these tectonic units and beneath the Archean Limpopo belt (3.34–3.37 g/cm³) to melt-metasomatism with an addition of a basaltic component. The Proterozoic Kheis, Okwa, and Namaqua–Natal belts and the Western Cape Fold Belt with the late Proterozoic basement have an overall fertile mantle (ca. 3.37 g/cm³) with local (100–300 km across) low-density (down to 3.34 g/cm³) and high-density (up to 3.41 g/cm³) anomalies. High (3.40–3.42 g/cm³) mantle densities beneath the Eastern Cape Fold belt require the presence of a significant amount of eclogite in the mantle, such as associated with subducted oceanic slabs.We find a strong correlation between the calculated density of the lithospheric mantle, the crustal structure, the spatial pattern of kimberlites, and their emplacement ages. (1) Blocks with the lowest values of mantle density (ca. 3.30 g/cm³) are not sampled by kimberlites and may represent the “pristine” Archean mantle. (2) Young (< 90 Ma) Group I kimberlites sample mantle with higher density (3.35 ± 0.03 g/cm³) than the older Group II kimberlites (3.33 ± 0.01 g/cm³), but the results may be biased by incomplete information on kimberlite ages. (3) Diamondiferous kimberlites are characteristic of regions with a low-density cratonic mantle (3.32–3.35 g/cm³), while non-diamondiferous kimberlites sample mantle with a broad range of density values. (4) Kimberlite-rich regions have a strong seismic velocity contrast at the Moho, thin crust (35–40 km) and low-density (3.32–3.33 g/cm³) mantle, while kimberlite-poor regions have a transitional Moho, thick crust (40–50 km), and denser mantle (3.34–3.36 g/cm³). We explain this pattern by a lithosphere-scale (presumably, pre-kimberlite) magmatic event in kimberlite-poor regions, which affected the Moho sharpness and the crustal thickness through magmatic underplating and modified the composition and rheology of the lithospheric mantle to make it unfavorable for consequent kimberlite eruptions. (5) Density anomalies in the lithospheric mantle show inverse correlation with seismic Vp, Vs velocities at 100–150 km depth. However, this correlation is weaker than reported in experimental studies and indicates that density-velocity relationship in the cratonic mantle is strongly non-unique.     

Spinel lherzolite xenoliths from the Premier kimberlite (Kaapvaal craton, South Africa): Nature and evolution of the shallow upper mantle beneath the Bushveld complex

Coarse-grained, granular spinel lherzolites xenoliths from the Premier kimberlite show evidence of melt extraction and metasomatic enrichment, documenting a complex history for the shallow mantle beneath the Bushveld complex. Compositions of orthopyroxene, clinopyroxene and spinel indicate equilibration within the spinel–peridotite facies of the upper mantle, at depths from 80 to 100 km and temperatures from 720 to 850 °C. Bulk compositions have lower Mg-number [atomic 100 Mg/(Mg + Fe*)] than previously studied spinel peridotites from Premier, and have higher Mg/Si than low-temperature coarse grained garnet lherzolites, suggesting shallower melting conditions or metasomatic enrichment. Clinopyroxene in one sample is highly LREE-depleted indicating very minor modification of a residue of 20% melt extraction, whereas the calculated REE pattern for a melt in equilibrium with a mildly LREE-depleted sample is similar to MORB or late Archean basalt, possibly related to the Bushveld Complex. Bulk and mineral compositions suggest minimal refertilization by silicate melts in four out of six samples, but REE patterns indicate introduction of a LIL-enriched component that may be related to kimberlite.     

Origin of melt pockets in mantle xenoliths from southern Patagonia,Argentina

Peridotite mantle xenoliths collected north of Gobernador Gregores, Patagonia, affected by cryptic and modal metasomatism bear melt pockets of unusually large size. Melt pockets consist of second generation olivine (ol2), clinopyroxene (cpx2) and spinel (sp2) ± relict amphibole (amph) immersed in a yellowish vesicular glass matrix. Amphibole breakdown was responsible for melt pocket generation as suggested by textural evidence and proved by consistent mass-balance calculations: amph → cpx2 + ol2 + sp2 + melt. Composition of calculated amphibole in amphibole-free melt pockets is very similar to that measured in amphibole-bearing melt pockets from the same xenolith, i.e. amphibole was consumed in the melt pocket generation process. In melt pockets devoid of relict amphibole, mass-balance calculations show remarkable differences between the calculated amphibole and the measured amphibole compositions in melt pockets from the same xenolith. The participation of minor proportions of a consumed reactant phase could be a reasonable explanation. In some samples the calculated phase proportion of glass is in excess compared to modal estimations based on backscattered electron images, probably because a portion of the generated melt was able to migrate out of the melt pockets. Compositional inhomogeneity of cpx2 and variable Ti Kd in cpx2 vs. glass in the same melt pocket reflect fast nucleation and growth and disequilibrium crystallisation, respectively. This and the difference between forsterite content in calculated equilibrium olivine and second generation olivine, suggest that mineral equilibrium was inhibited by rapid quenching of melt pockets.     

Equilibration and reaction in Archaean quartz-sapphirine granulite xenoliths from the Lace kimberlite pipe, South Africa

Ultrahigh-temperature quartz-sapphirine granulite xenoliths in the post-Karoo Lace kimberlite, South Africa, comprise mainly quartz, sapphirine, garnet and sillimanite, with rarer orthopyroxene, antiperthite, corundum and zinc-bearing spinel; constant accessories are rutile, graphite and sulphides. Comparison with assemblages in the experimentally determined FMAS and KFMASH grids indicates initial equilibration at >1040 °C and 9–11  kbar. Corona assemblages involving garnet, sillimanite and minor cordierite developed on a near-isobaric cooling P–T  path as both temperature and, to a lesser extent, pressures decreased. Garnet-orthopyroxene Fe-Mg exchange thermometers record temperatures of only 830–916 °C. These estimates do not indicate the peak metamorphic conditions but instead reflect the importance of post-peak Fe-Mg exchange during cooling. Correction of mineral Fe-Mg compositions for this exhange using a convergence approach of Fitzsimons & Harley (1994 ) leads to retrieved P–T  estimates from garnet-orthopyroxene thermobarometry ( c . 1000 °C and 10.5±0.7  kbar) that are consistent with the petrogenetic grid constraints. U-Pb dating of a single zircon grain gives an age of 2590±83  Ma, interpreted as the age of the metamorphic event. Protolith major and trace element chemistries of the xenoliths differ from sapphirine-quartzites typical of the Napier Complex (Antarctica) but are comparable to less siliceous, high Cr and Ni, sapphirine granulites reported from several ultrahigh temperature granulite terranes.     

华北克拉通古老岩石圈地幔的多次地质事件:来自金伯利岩中橄榄岩捕虏体的启示

本文提供的两件蒙阴岩区金伯利岩中的蛇纹石化石榴石橄榄岩捕虏体,整体发育剪切-变形结构,其中的辉石有三种类型,代表了三次地质事件,他们是:(1)石榴石中的自形单斜辉石包裹体Py;(2)粗粒不规则形状的斜方辉石Py1;(3)具反应边及定向排列的斜方辉石Py2。 Py具有高Na2O和Al2O3,及低Mg#和CaO的特征,暗示所赋存的橄榄岩未遭受过明显的熔融作用。推测Py为早期阶段地幔"岩浆海"结晶时被石榴石包裹的矿物。在手标本及薄片中普遍见到Py2切过Py1,表明Py1形成早于Py2。Py1的 Cr(669×10-6~9503×10-6), Ni (1941×10-6~4750×10-6)含量和Mg#(0.91~0.94)比值较高, 而Py2中的Cr (725×10-6~1926×10-6) , Ni (902×10-6~2989×10-6) 和Mg# (0.88~0.90)值较低,说明Py1是早期经部分熔融的橄榄岩耐熔残余中的顽火辉石残留。相反,Py2可能是软流圈来源的熔体与耐熔橄榄岩反应的结果。剪切/变形 以及交代事件则发生于上述反应之后或者与之同时。依据主元素特征,较早的Py1的耐熔程度反而高于Py2,本文称之为地幔组成的"逆向演化"。 看来,这种逆向成分演化不仅发生在中新生代,而且也发生于古老地幔,甚至是贯穿于整个地幔演化的历史时期。与已发表的有关地幔形成年龄的资料对比,Py、Py1和 Py2的年龄估计分别是>3.8Ga, 2.5Ga/1.4~1.3Ga和0.9~0.7Ga 。另外1件碳酸盐化橄榄岩捕虏体,采自复县金伯利岩,具有明显的剪切-变形结构,最终形成时间可能与Py2接近。     

Associated chemical and carbon isotopic composition variations in diamonds from Finsch and Premier kimberlite,South Africa

The carbon isotopic composition of 66 inclusion-containing diamonds from the Premier kimberlite, South Africa, 93 inclusion-containing diamonds and four diamonds of two diamond-bearing peridotite xenoliths from the Finsch kimberlite, South Africa was measured. The data suggest a relationship between the carbon isotopic composition of the diamonds and the chemical composition of the associated silicates. For both kimberlites similar trends are noted for diamonds containing peridotite-suite inclusions (P-type) and for diamonds containing eclogite-suite inclusions (E-type): Higher δ¹³C P-type diamonds tend to have inclusions lower in SiO₂ (ol), Al₂O₃ (opx, gt), Cr₂O₃, MgO, $\text{Mg}\text{(Mg + Fe)}$ (ol, opx, gt) and higher in FeO (ol, opx, gt) and CaO (gt). Higher δ¹³C E-type diamonds tend to have inclusions lower in SiO₂, Al₂O₃ (gt, cpx), MgO, $\text{Mg}\text{(Mg + Fe)}$ (gt), Na₂O, K₂O, TiO₂ (cpx) and higher in CaO, $\text{Ca}\text{(Ca + Mg)}$ (gt, cpx).Consideration of a number of different models that have been proposed for the genesis of kimberlites, their xenoliths and diamonds shows that they are all consistent with the conclusion that in the mantle, regions exist that are characterized by different mean carbon isotopic compositions.