Crystallization of Megacrysts From Protokimberlite Fluids
We report petrological, geochemical and isotopic data for low-Cr and high-Cr suite megacrysts from the Jericho kimberlite (N. Canada). The megacrysts comprise garnet, clinopyroxene, orthopyroxene and olivine; ilmenite is present only in the low-Cr suite. The megacrysts occur mostly as polycrystalline variably- recrystallized intergrowths. Our observations suggest that the Jericho megacrysts cannot be magmatic, i.e. they are not crystallized from a large body of magma that allows for crystal fractionation. The megacrysts are proposed to form metasomatically from a melt or a fluid. Megacrysts do not have magmatic textures or extended MgO compositional trends typical of magmatic fractionation. Radiogenic Hf compositions and fractionated REE patterns of Cr-rich megacrysts as compared to low-Cr megacrysts elsewhere imply a lower fluid/wall rock ratio than that characteristic of magmatic processes. The megacrysts consist of at least two unequilibrated parageneses and formed as a result of highly localized metasomatic recrystallization of previously existing larger mantle minerals. Rb-Sr, Sm-Nd and Lu-Hf isochron ages for the megacrysts are similar to the eruption age of the kimberlite and represent blocking of the isotope systems at the time of eruption. Trace element patterns of melts buffered by megacrystal clinopyroxene and garnet were computed using recent experimental partition coefficients. Melts buffered by megacrystal clinopyroxene are similar in REE composition to that of the host Jericho kimberlite, whereas melts equilibrated with garnet are less fractionated and resemble primitive Group I kimberlites from South Africa. Previous modeling of melts equilibrated with megacrysts failed to produce REE patterns equivalent to kimberlitic melts and so concluded that the patterns are closer to those of the alkali basalt. Our modeling yields a very different result because 1) we based the modeling on high-Cr megacrysts that have steeper REE patterns, and 2) we used partition coefficients for carbonatitic high-pressure, high-temperature melts. Our data suggests that the fluid that led to metasomatic recrystallization of mantle wall-rock minerals and formation of megacrysts was kimberlitic in nature. The megacrysts are equilibrated with a kimberlite melt with respect to trace elements and have Hf and Nd isotopic compositions similar to the host kimberlite. The range in initial Sr isotope compositions of the host Jericho kimberlite can be modelled by addition of 5-10% of local crust such as limestone or granite, into a magma that had an initial unradiogenic Sr isotope composition similar to that of the megacrysts.
The Diamond Potential of the Tuwawi Kimberlite (Baffin Island, Nunavut).
Baffin Island, underlain by Archean crust of the Rae craton with Paleoproterozoic reworking, is known to contain several kimberlites of possibly Cretaceous age. The most recent findings of kimberlite are located at the northwestern end of Baffin Island on the Brodeur Peninsula. The Tuwawi kimberlite, one in the cluster of 3 kimberlites, has an inverted cone shape. We studied drill core samples of kimberlite and mantle xenoliths from the Tuwawi kimberlite to constrain its diamond potential. Hypabyssal and volcaniclastic kimberlite types have been identified among available kimberlite core. Hypabyssal kimberlite is the predominant type in Tuwawi. The kimberlite consists of olivine macrocrysts set in a carbonate-serpentine groundmass with olivine microphenocrysts, phlogopite and spinel. Volcaniclastic kimberlite is characterized by the presence of 1) irregularly-shaped juvenile lapilli; 2) two semi-intermixed dark cryptocrystalline matrix materials; 3) olivine grains with a restricted size distribution and angular shapes. These features suggest mild sorting of the kimberlite, a possible incorporation of mud to the matrix, an epiclastic origin and formation in the crater facies. Peridotites and a garnet clinopyroxenite are found as xenoliths in the Tuwawi kimberlite. Peridotites include garnet lherzolite, garnet, spinel, and garnet-spinel harzburgites, and dunite. Both coarse and deformed (porphyroclastic and mosaic-porphyroclastic) textures are present within the peridotite xenoliths, and Cr- diopside from deformed xenoliths shows higher TiO2 (0.16 wt%) content than in coarse peridotites. Pyrope (Mg70-82) is present in all but one sample, whereas spinel occurs only in coarse peridotites and shows strong heterogeneity. It is controlled by random intra-grain compositional changes in FeO (from 13 to 16 wt%), MgO, Al2O3 and Cr2O3 (from 43 to 57 wt% ). Olivine and orthopyroxene in all xenoliths are very magnesian (Fo85-87 and En86-89), slightly more so in coarse peridotites. Pressures and temperatures of mineral equilibria for the xenoliths were estimated using various two- pyroxenes, garnet-pyroxene, olivine-garnet, and olivine-spinel geothermobarometers. Porphyroclastic garnet lherzolite and garnet clinopyroxenite were formed at 1100-1140°C and 54-56 kb. Deformed peridotites are equilibrated at higher temperatures and pressures than coarse peridotites. Garnet peridotites and pyroxenites show higher temperatures than spinel peridotites. These patterns match the commonly observed mantle lithological columns below cratons. In comparison to temperature and pressure data from kimberlites of Somerset Island, xenoliths from Tuwawi plot farther into the diamond stability field and at a lower geothermal gradient (∼42 mW/m2). The majority of mantle xenoliths from Cretaceous Somerset Island plot in the graphite stability field along a geotherm of ∼44 mW/m2. Our study identified several factors that give a positive outlook on the diamond potential of the Tuwawi kimberlite. These factors include 1) a preservation of the crater facies kimberlite, and 2) kimberlite sampling of the deep diamondiferous mantle. The diamond potential is reduced by the estimated 42 mW/m2 geothermal gradient that is hotter than the desired low geotherm for Archean cratons.
Re-Os Isotopes in Kimberlite Oxides
Over the past two decades, the Rhenium-Osmium (Re-Os) decay scheme has been increasingly used to evaluate the timing of crustal growth and mantle depletion events in the continental lithosphere. This method has utilized mantle peridotite xenoliths entrained in kimberlite and related rocks, but in many cases, such xenoliths are highly altered or absent, negating their use for geochronology. We have investigated the use of oxide minerals present in kimberlite as monitors of mantle Re-Os characteristics in the absence of peridotite xenoliths. Our Re-Os data, obtained from oxide minerals in kimberlites and lamprophyres from North America and South Africa, over a range of emplacement ages (0.4- 1.18 Ga), suggests that meaningful Re Depletion Ages (TRD) can be determined in many cases, based upon the agreement between these Re-Os TRD ages and other independent age estimates. Major element data suggests a link between Os content and chromium-rich spinel abundance. Previous research investigating platinum group metal abundances (PGMs) in mantle xenoliths has also supported this link, and has shown that Os-rich alloys can be formed around the edges of chromite grains in sulphur under- saturated conditions. Simple mass balance calculations suggest that <30 Os-rich alloy grains of <2 μm diameter could account for the entire Os budget of the oxide mineral separates analyzed, although Os alloys were not observed during microprobe image analysis. Conversely, plots of Fe# vs Ti# and Cr2O3 vs TiO2 suggest that the spinel samples are primary magmatic phases and not xenocrystic, indicating that PGMs form Os alloy inclusions in kimberlitic spinel, or partition directly into the crystal structure. Calculated initial Os isotopic compositions for oxides at the time of eruption also have potential to constrain the origin and subsequent magma history. Some samples have depleted signatures indicating that the mantle source for the magma was highly depleted close to the emplacement time. Others have supra-chondritic initial Os isotopic compositions indicating either source enrichment prior to kimberlite/lamprophyre genesis, or interaction with highly radiogenic fluids close to the time of magma formation
Experimental Study of Surface Dissolution Features on Kimberlite Indicator Minerals
During the ascent to the Earth's surface kimberlite magmas entrain mantle minerals - chromites, ilmenites, garnets and the most desirable - diamonds. Kimberlite magma partially dissolves these minerals during the ascent, producing different types of surface features on the minerals. Experiments showed that surface features on diamonds can be used to constrain composition of magmatic fluid. However, examining mantle minerals with more complex chemical compositions, such as chromites and ilmenites, could provide more detailed information about the composition and evolution of fluid system in the magmas, as determination of the depth of their entrainment is possible. This study experimentally investigates dissolution of chromites and ilmenites in melts with C-O-H fluid. The surface features produced at these conditions are then compared to the surface features on minerals recovered from kimberlites. The experiments were done in a piston-cylinder apparatus at 1350 - 1400°C and 1 GPa. Rounded natural mineral grains were placed in a synthetic mixture of Ca-Mg-Si-C-H-O composition with 0, 5, 13, 15, and 31 wt% H2O and 0, 5, and 27 wt% CO2. The experimental results investigated using Field-Emission Scanning Electron Microscope showed that angular step-like dissolution surfaces, which are common for natural kimberlitic chromites, develop only in the presence of H2O-rich fluid phase. The reaction of chromite with H2O dissolved in the melt and with dry melt caused smoothing of chromite surfaces. Chromite dissolution in CO2-rich melts produced rounded and polyhedral relief features. Both the smooth and polyhedral types of features are not typical for natural kimberlite-hosted chromite grains. Ilmenite underwent rapid dissolution at our experimental conditions. In H2O-rich fluid ilmenite produced "pyramidal" type of surface features previously described as the most common for natural kimberlitic ilmenites. The experimental results were compared to the natural minerals recovered from Misery and Grizzly kimberlites, EKATI Mine property, N.W.T., Canada. Misery chromites and ilmenites have features similar to those experimentally produced in the presence of H2O-rich fluid. This agrees well with fluid composition constrained from the surface features of Misery diamonds. Grizzly chromites have surface features which were not observed in our experiments and need further studies. Our study shows that surface features on xenocrystal minerals can provide information about the behaviour of volatiles in mantle- derived magmas. Preliminary results suggest that kimberlites containing chromites which exhibit angular step- like surfaces have significant H2O fluid during emplacement, and may be associated with good quality diamonds.
An Experimental Study of Diamond Dissolution in Cl--H2O Systems: Implications for Mechanisms of Diamond Oxidation and Kimberlitic Fluids
A diverse array of textures occurs on natural diamond surfaces, representing both growth and dissolution events. Dissolution features produced by fluids depend strongly on fluid composition. Similarities between diamond surfaces produced in H2O-fluid bearing experiments and surfaces of kimberlite-hosted diamonds suggest a water-rich kimberlitic fluid composition. However, many diamonds' surfaces show markedly different appearance, even within a single kimberlite, indicating that the compositional history of diamond dissolving fluids is complex. The occurrences of chloride-rich fluid inclusions in coated and cloudy diamonds, as well as chloride-rich mineralogy of the unaltered Udachnaya-East kimberlite (Yakutia) indicate that chloride may be an important constituent in diamond-fluid interactions in kimberlite or in the mantle. We experimentally investigated diamond dissolution at 1 GPa and 1350°C in the systems H2O-NaCl and H2O-KCl to determine the effect of chloride salts on the character of diamond resorption. We then compared the results to the surfaces of natural kimberlite-hosted diamonds. Our results clearly show that diamond surface features are dependent on the chloride salt content of aqueous fluid. Addition of chloride salts reduced edge and vertex dissolution. Octahedral faces underwent more intensive development of triangular etch pits of smaller size than in water fluid. A distinctive feature observed only in chloride-bearing experiments is the development of deep etch channels, which are reported on some natural diamonds. These surface features may be used as a basis for comparison to natural diamonds to evaluate the importance of alkali chlorides in kimberlitic or mantle fluids. Additionally, our results provide new insights into the mechanism of diamond oxidation.
S/Se In Sulfide Inclusion In Diamond
Sulfides are among the most common minerals found as inclusions in diamonds. Being protected from any alteration after diamond formation, they likely represent the most pristine sulfide sample of mantle rocks. Their chemical composition in major and minor elements (mainly Ni, Cu and Cr), as determined using Electron Probe Micro Analyse (EPMA), is commonly used to determine the rock type in which the diamond formed. Here we propose to apply the same technique to the trace element abundance determination. We performed selenium (Se) on sulfide inclusion in diamonds. The S/Se value could help understanding whether the diamond formed in an eclogitic or peridotitic environment and may also constrain on the magmatic differentiation of diamonds host rock as well as provide a potential surface (hydrothermal) signature in diamond inclusions. A trace element measurement scheme has been developed by EPMA at the CAMPARIS centre (Paris). Se-abundance was obtained using a 30 kV accelerating voltage and 100nA probe current. Total counting time was 800s for peak (1.1 Å ) and 400s for background on both side of peak. Analyses were duplicated by μPIXE using the LPS nuclear microprobe facility (SIS2M CEA Saclay, France). Maps from 30x30 μm2 to 70x70 μm2 were obtained by scanning a 4x4 μm2 proton beam of 3MeV, 600 pA, (0.4 to 2 μC). The two techniques show good agreement and we conclude that EPMA is well suited for accurate and precise Se measurements. We analysed five samples; two monosulfide solid solution (MSS) (Ni > 22 wt%) typical of the peridotitic paragenesis (P-type), and three Ni-poor sulfides (Ni < 7 wt%) typical of the eclogitic paragenesis (E-type). In P-type sulfides, Se-content (260 ppm) is significantly higher than previously reported in sulfides from mantle-derived lherzolites (40-160 ppm), pyroxenites (25-45 ppm) or harzburgite. The value of S/Se in MSS is low (∼1400) compared to those of the primitive mantle reservoir (3,300; McDounough et al., 1995 Chemical Geology,120, 223-253.) and CI chondrites (∼2,540; Dreibus et al., 1995, EPSL, 255, 9-21.). This S/Se values suggests a more compatible behaviour for Se; it is typical of highly refractory mantle rocks having experienced a high degree of partial melting prior to diamond formation. In E-type sulfides, Se-content ranges from 69 to 120 ppm, a range consistent with one reported for E-type sulfides extracted from Yakutian diamonds (16-106 ppm, Bulanova et al. (1996) Contrib. Mineral. Petrol., 124, 111-125). The S/Se values range from 3.2*103 to 5.6*103. These values belong to the magmatic sulfide field but do not reveal any evidence for a surface S signal (hydrothermal sulfides, for example, have S/Se ratios up to 104, Luguet et al. (2004) Chemical Geology, 208, 175-194). This preliminary study shows that selenium content can discriminate between sulfide parageneses associated with diamonds. As a supplement to major element analysis, it also constrains the degree of partial melting of the diamond's host-rocks.
Removal of Brown Colour From Diamonds During Storage in the Lithospheric Mantle
Brown colour in natural diamonds is produced by plastic deformation during residence in the mantle. Dislocation movement generates vacancies, which aggregate into clusters of about 30-60 vacancies. The resulting electronic configuration at each cluster leads to the broad, featureless absorption pattern associated with common brown colour. Less commonly, other brownish colours can be attributed to hydrogen or isolated nitrogen atoms, H4 centres, or possibly oxygen. The common brown colour can be removed by high-pressure-high-temperature (HPHT) treatment. This process involves pressures and temperatures of 5-9 GPa and 1800-2700 ° C, respectively. Treatment may take several minutes or hours. It has been suggested that brown diamonds stored in the lithospheric mantle should lose their colour by analogy to HPHT treatment. If so, brown colour must be the product of late deformation (close to kimberlite eruption) or the brown diamonds must be stored above the diamond stability field (in a cooler part of the lithosphere). Is it reasonable to expect brown colour to be destroyed in the lithospheric mantle? An objective analysis of this question must consider temperature. Higher temperatures result in faster colour removal. HPHT treatment occurs at 1800-2700 ° C, whereas inclusion thermometry places most lithospheric diamonds in the range of 900-1400 ° C. Destruction of the brown colour centre involves breaking up vacancy clusters. The activation energy required to do this can be estimated as the energy of an isolated monovacancy, minus the energy per vacancy of the cluster, plus the vacancy migration energy. Data from recent literature produces a value of 7.7 eV. The Arrhenius equation can be modified to show how reaction time varies with temperature. The activation energy can be used in conjunction with experimental HPHT data to extrapolate reaction times from HPHT temperatures to lithospheric mantle temperatures. For a certain reduction in brown colour produced by HPHT (T1, t1), the equation below shows the time required (t2) to produce the same reduction at a different temperature (T2): ln(t2) = ln(t1) + (Ea/k)(1/T2 - 1/T1) where Ea is activation energy, k is the Boltzmann constant, and temperatures are in kelvin. A relatively potent treatment of 1 hour at 2500 ° C is comparable to about 109 years at 1150 ° C. Most diamonds would lose any brown colour in this scenario. Modest colour reduction is detectable after 5 hours at 1800 ° C. This is comparable to about 109 years at 950 ° C, or about 105 years at 1150 ° C. The time required to destroy brown colour in the lithospheric mantle is significant at the scale of geological time. Brown diamonds should easily maintain their colour during cooler mantle storage at or below 1000 ° C. Warmer temperatures toward the base of the lithosphere may be able to reduce or eliminate brown colour within a reasonable geological time frame. The survival of brown colour in the lithospheric mantle does not require the colour to be formed late in the storage history nor does it require metastable storage in the graphite stability field.
Unusual Cathodoluminescence in Diamonds: Evidence for Metamorphism or a Source Characteristic
Cathodoluminescence (CL) is a useful means of diamond "fingerprinting". CL-active cratonic macrodiamonds usually cathodoluminesce blue or yellow, and always exhibit prominent wide CL emittance peaks at 430-450 nm and 480-490 nm. Exceptions to this norm are diamond suites recently discovered in the Archean rocks metamorphosed in the greenschist facies. These macrodiamonds cathodoluminesce red, orange and yellow, and invariably exhibit the most prominent CL peak at 520 nm. The diamond suites with the unusual CL are derived from two different locations within the Michipicoten Greenstone Belt (Southern Superior craton), near the town of Wawa (Ontario). One suite is extracted from the 2.68-2.74 Ga polymict volcanic breccias and lamprophyres and the other suite - from the 2.68 Ga sedimentary conglomerates grading into overlying sandstones of the Dore assemblage. The diamondiferous conglomerates are found in an area 8 km south of the breccias and 12 km northeast of Wawa. CL emittance of macrodiamonds (> 0.5 mm) extracted from the breccias consists of a broad band at 520 nm, a sharp peak at 575.5 nm, and several lines at 550-670 nm. The conglomerate macrodiamonds mostly show a dominant peak at 520 nm, whereas corresponding microdiamonds exhibit two peaks at about 576 and 600 nm. None of the diamonds show a maximum peak at 420 nm. Polycrystalline stones from conglomerates show distinct CL spectra and colours for all intergrown crystals in the same diamond. The relative abundances of the CL colors of the conglomerate diamonds are orange-red (46%), yellow (28%), orange-green (10%), green (6%), and non-uniform colors (10%). These colours are more diverse than mostly orange CL colours in the breccia diamonds; this results from a larger variety of positions and intensity of CL peaks in the conglomerate diamonds. We propose two models for explaining the presence of the 520 nm CL peak in the breccia and conglomerate diamonds in Wawa. The first model suggests metamorphism as the main factor influencing the CL colors of the suites. Diamonds in the volcaniclastic breccias and sedimentary conglomerates may have come from different deep sources, but acquired similar cathodoluminescence due to a metamorphic overprint. Metamorphic fluids have been shown to have a potential to percolate through diamond fractures and affect diamond inclusions. Furthermore, diamonds found in the Kokchetav metamorphic massif are reported to have green CL with an emission at 514-537 nm. The "metamorphic" model is supported by the contrast in the diamond indicator minerals recovered from the volcaniclastic breccias and sedimentary conglomerates. Only the latter contain kimberlite indicator minerals from a proximal source, such as diopside and garnet with preserved kelyphitic rims. The second model suggests the presence of the 520 nm CL peak controlling the green-red CL visible colors is an internal characteristic of the two Wawa diamond suites related to their origin from the same deep source. Currently, we are studying the N content and aggregation state of the conglomerate diamonds using the Fourier transform infrared technique to compare these data with the corresponding values for the breccia diamonds. Further work is needed to determine if either model can explain all observed properties of the Wawa diamond suites.
Mapping a Mantle Xenolith Using Micro X-ray Diffraction
With the advent of Micro X-ray diffraction (μXRD), mineral mapping is now possible at the thin section scale, using crystal structural rather than chemical properties. To test the effectiveness of this technique, a Brüker D8 Discover μXRD was used to create mineral maps from two garnet lherzolite xenoliths from the Bultfontein kimberlite recovered on the Boshof road dumps (Kimberley, South Africa). This technique generated a 2D representation of the major mineralogy (olivine + garnet + clinopyroxene + orthopyroxene + phlogopite); it also enabled estimation of the modal proportions of each mineral. For one polished section, two sets of μXRD data were collected along grids having spacing of 0.5 mm (1148 points) and 1.5 mm (516 points), respectively. The 0.5 mm spacing coincided with the X-ray beam diameter, leaving no gaps in coverage. Data were also collected using a grid of 1.5 mm spacing to examine the effectiveness of a low- resolution map, which would be advantageous to decrease the time required for data collection and processing. To test the potential for automation of the method, data were collected for a second polished thin section, using the high resolution grid (0.5 mm spacing, 3312 data points). General Area Diffraction Detection System (GADDS) images were collected and processed at each step. Debye rings were integrated along small "slices" of 2θ to yield the total number of counts in that area. The 2θ ranges used to identify each phase were chosen such that they coincided with regions where there was no overlap with peaks from any of the other four major phases. To compensate for variable orientation of the grains in the thin sections, several d-spacings (2θ ranges) were integrated for each mineral, and then added together to create five "overall" mineral maps. To create final maps, minerals were determined to be 'present' in cells where the number of counts was above a threshold level in the "overall" map for that mineral. The maps provided fairly accurate estimates of the modal mineralogy (when compared to point counts), with the added benefit of recording the spatial distribution. Some areas of the thin sections were not assigned to any mineral, due to the orientation of the mineral not satisfying Bragg's law, or the mineral being highly altered in that area of the thin section. Further development of this technique will focus on refining the quality of the final maps. This technique could potentially be used to provide information about preferred orientation of minerals in mantle xenoliths and other rocks, or to distinguish minerals which exist in different hydrated states.