| Journal of Petrology | Pages |
© 1999 Oxford University Press |
|
|
Kimberlite-derived xenoliths provide constraints on the composition, structure and thermal state of the upper mantle underlying many of the world's major cratons. Before the 1990s the lack of kimberlite occurrences from the Slave craton (Northwest Canada) made the lithosphere underlying this craton terra incognita. The first kimberlite was found in the Lac de Gras area of the Slave craton in 1991, and the following `diamond rush' has discovered more than 150 kimberlite pipes (Pell, 1997). Xenoliths hosted by these kimberlite bodies offer the first opportunity to study the petrology of the upper mantle for this part of the Earth. To date, there have been only preliminary petrological studies of mantle xenoliths of the Slave craton (Boyd & Canil, 1997; Kopylova et al., 1997; MacKenzie & Canil, 1998; Pearson et al., 1998). The purpose of this paper, therefore, is to provide the first detailed description of the mantle rocks underlying the Slave craton, to constrain their conditions of origin and to compare our results with parallel results from other cratons. The Slave Structural Province of the Northwest Territories is one of several Archaean nuclei of the North American Craton. These nuclei, including the Superior Province, the Slave Province, and the Nain Province, were welded together in Palaeoproterozoic time (2·5-1·6 Ga) (Percival, 1996). The Slave craton comprises dominantly late Archaean (2·7-2·6 Ga) supracrustal and plutonic rocks (Padgham & Fyson, 1992), within which are blocks of older (4·0-2·8 Ga) gneiss and younger sedimentary rocks (Percival, 1996). The Earth's oldest known rocks, the Acasta gneisses (4·02 Ga), occur in the western part of the Slave craton (Bowring & Housh, 1995). Kimberlite has episodically intruded the Slave lithosphere from the Cambrian to Tertiary (Pell, 1997). The distribution of kimberlite pipes in the central Slave Province (Fig. 1) is similar to that observed in other kimberlite fields; most pipes are distributed along a main trend (NNW), whereas subordinate clusters of pipes are arranged in a direction orthogonal to the main trend (NNE and ENE) (Kjarsgaard, 1996). Most of the kimberlite bodies in the Slave Province do not crop out at surface, but are concealed by lakes or glacial till and are fairly small (Pell, 1997); they are interpreted as variably eroded, carrot-shaped diatremes fashioned after `miniature' versions of the classic South African pipes (Kjarsgaard, 1996). The majority of the Slave kimberlite occurrences are in the Lac de Gras area, where crater, diatreme, and hypabyssal facies kimberlite have all been identified. Diamonds have been found in 35 pipes located within the Slave craton, and at least seven are known to have economic quantities of diamonds (Pell, 1997). Figure This study focuses on xenoliths derived from the Jericho kimberlite pipe (65°59'55"N, 111°28'45"W). The Jericho pipe is located 400 km NE of Yellowknife near the northern end of Contwoyto Lake, intrudes Archaean granitoids (Bowie, 1994) and supracrustal rocks of the Slave craton, is dated at 172 ± 2 Ma (Rb-Sr and U-Pb geochronology; Heaman et al., 1997), and is significantly diamondiferous (1·18 ct/t, Lytton Minerals Ltd Press Release, 1997). The Jericho kimberlite is a multiphase intrusion consisting of a precursor dyke and at least two pipes (Cookenboo, 1998). Mineralogically, the Jericho kimberlite is a typical non-micaceous kimberlite lacking groundmass tetraferriphlogopite (Mitchell, 1995). Chemically, on the basis of concentrations of TiO2, K2O, SiO2 and Pb (Smith et al., 1985), the Jericho kimberlite is classified as Group Ia (Kopylova et al., 1998b), and is similar to most of the other Slave kimberlites (Pell, 1997). Our study is based on a large and comprehensive suite of ~200 mantle xenoliths. They are normally <15 cm in diameter, with rare specimens being as large as 30 cm. Megacrysts (commonly <5 cm in size) are abundant. Our collection, excluding eclogites, has the following proportions of rock types: 48% coarse peridotite (18% spinel-garnet and 30% spinel free), 31% porphyroclastic peridotite, 16% megacrystalline pyroxenite, and 5%ilmenite-garnet wehrlite and clinopyroxenite. Eclogite xenoliths make up ~25% of the mantle-derived xenoliths in the kimberlite and have been described by
Kopylova et al., (1998c). Xenoliths of coarse peridotite are composed of olivine (55-90%) and lesser amounts of clinopyroxene (0-10%), orthopyroxene (0·5-10%), garnet (0·5-5%), and spinel (0-1%). The predominant rock types are lherzolite and harzburgite; wehrlite is scarce. Coarse peridotite occurs in spinel-garnet and garnet facies. The coarse texture derives from early crystallization of subhedral, large olivine (3-10 mm) and orthopyroxene (0·8-5 mm) and late crystallization of anhedral clinopyroxene (1-7 mm), garnet and spinel (Fig. 2a). Wehrlite has a poikilitic texture, where large clinopyroxene oikocrysts include smaller grains of olivine. Rarely orthopyroxene shows exsolution lamellae of clinopyroxene ± garnet or spinel. Light lilac-coloured garnet, 2-3 mm across, is subhedral or anhedral and forms along grain boundaries. In some instances, samples contain phlogopite and amphibole which appear to be in textural equilibrium with primary mantle minerals, suggesting a primary-metasomatic origin (Winterburn et al., 1990). Figure The majority of the specimens show equant or tabular coarse texture (Harte, 1977). Anhedral spinel found within interstices between olivine or as irregular inclusions within pyroxene and garnet indicates that the deformation textures can be classified as `primary' (Mercier & Nicolas, 1975). Only one peridotite was found that showed a possible `secondary' granoblastic texture as implied by the occurrence of subhedral spinel enclosed in olivine. Secondary alteration includes the development of phlogopite replacing clinopyroxene and garnet, talc after orthopyroxene, and serpentine (green fibrous chrysotile) after olivine and orthopyroxene. Secondary carbonate is commonly found on the margins of pseudomorphs of olivine and orthopyroxene. Porphyroclastic peridotite can be subdivided on the basis of deformation textures into porphyroclastic non-fluidal (Fig. 2b) and fluidal peridotite. Among the latter group, peridotite is further classified into non-disrupted (Fig. 2c) and disrupted varieties (Fig. 2c) (Harte, 1977). Both static and dynamic recrystallization were active in the formation of porphyroclastic peridotites. Static recrystallization during annealing created euhedral, tabular olivine neoblasts (e.g. Bouiller & Nicolas, 1975; Gueguen, 1977; Mercier, 1985), whereas dynamic recrystallization resulted in the development of equant polygonal neoblasts near rims of large porphyroclasts. Fluid trapped in olivine porphyroclasts assisted grain boundary migration and recrystallization as shown by the tabular habits of neoblasts [Drury & Van Roermund, (1989) and references therein]. Porphyroclastic non-fluidal peridotite Porphyroclastic non-fluidal peridotite comprises olivine porphyroclasts (40-90%), olivine neoblasts (3-40%), orthopyroxene (5-12%), clinopyroxene (0-10%), and garnet (2-10%). Samples show a moderate to strong preferred orientation of olivine and pyroxenes. Olivine porphyroclasts are generally 1·5-7 mm in diameter, show undulatory extinction and kink bands, and can contain numerous fluid inclusions (Fig. 3a). Olivine neoblasts (0·05-0·5 mm) exhibit both euhedral, tabular (Fig. 3b), and equant polygonal habits near rims of large porphyroclasts. Orthopyroxene (up to 4 mm long) forms large porphyroclasts, either strained or not, with peripheral recrystallization into tiny neoblasts. Emerald-green clinopyroxene ( <= 4 mm) is subhedral or has been recrystallized into neoblasts. Garnet (0·5-3 mm) is colourless or light blue in thin sections, and is euhedral or elongate in a direction consistent with the olivine preferred orientation. Rare xenoliths host coexisting grains of garnet characterized by different colours, indicating significant intergranular compositional heterogeneity. Figure Porphyroclastic fluidal non-disrupted peridotite The transition to the more deformed fluidal-porphyroclastic texture is marked by recrystallization of pyroxenes into fine neoblasts, i.e. by `tails' of tinypyroxene neoblasts stretching out from larger pyroxene porphyroclasts (Fig. 2c). Intact rounded garnet grains define the non-disrupted nature of these samples. The rocks with strong preferred orientation are made up of olivine porphyroclasts (20-45%), olivine neoblasts (35-70%), clinopyroxene (1-5%), orthopyroxene (1-15%), and garnet (1-3%). Olivine porphyroclasts are strongly deformed as evidenced by extreme undulatory extinction, narrowly spaced (100) sub-boundaries, low-angle dislocation walls, and transformation of sharp sub-boundaries into grain boundaries. Olivine neoblasts are 0·5 mm in size, equant or tabular, and free of dislocations and fluid inclusions. Orthopyroxene forms both larger strained and deformationally twinned orthopyroxene porphyroclasts and fine neoblasts drawn out from porphyroclasts into streaks. Clinopyroxene occurs in bands of neoblasts that surround or are disrupted by incompletely recrystallized porphyroclasts. Clinopyroxene neoblasts often make up `envelopes' and `stress shadows' around rigid garnet. Clinopyroxene grains show more deformation than orthopyroxene, and clinopyroxene porphyroclasts are less abundant than orthopyroxene porphyroclasts, indicating prompt recrystallization of the former. Under the conditions realized in the Jericho mantle, clinopyroxene was apparently the second most weak and plastic mineral after olivine. Late clinopyroxene, as tiny grains, develops in interstices between olivine neoblasts and especially along the margins of orthopyroxene porphyroclasts (Fig. 3c). Porphyroclastic fluidal disrupted peridotite Among the Jericho mantle xenoliths, these samples are the most deformed and contain layers, stringers, and lenses of disaggregated garnet. In contrast to other porphyroclastic peridotites, disrupted specimens have distinct mineral chemistry and lower temperatures of equilibrium. The peridotite exhibits a strong preferred orientation accentuated by mineral banding. Equant olivine neoblasts (0·03-0·05 mm) form up to 70% of the rock, and orthopyroxene (10-15%) occurs in thin, alternating, wavy bands of neoblasts with rare porphyroclasts. Clinopyroxene (1-5%) and garnet (0-3%) neoblasts formmonomineralic layers. Garnet neoblasts are extensively replaced by phlogopite. Megacrystalline pyroxenite xenoliths include websterite, clinopyroxenite, and orthopyroxenite. Modal proportions of minerals in the pyroxenite vary significantly: orthopyroxene (30-65%), clinopyroxene (0-65%), olivine(0-10%), garnet (1-25%), spinel (0-1%). This group of rocks is characterized by extremely large grain size (5-30 mm), irregular, curvilinear grain outlines (Fig. 3d), and the presence of exsolution lamellae and symplectite intergrowths. Such textures are metastable under elevated P-T conditions (Field & Haggerty, 1994; Passchier & Trouw, 1996). Pyroxenite has a magmatic allotriomorphic texture defined by anhedral pyroxenes or by subhedral orthopyroxene and anhedral clinopyroxene. Some pyroxenite shows porphyroclastic textures. Clinopyroxene is commonly coarser than orthopyroxene, although clinopyroxene may also be present as small exsolution lamellae in orthopyroxene cores. Olivine forms anhedral small grains along pyroxene margins and irregular crystals poikilitically enclosed by clinopyroxene. Late garnet occurs as small round inclusions in clinopyroxene, in vermicular anhedral grains, or as an exsolved phase in orthopyroxene. Rare accessory spinel is enclosed in garnet and clinopyroxene. Other accessory phases include ilmenite, pentlandite and pyrrhotite. A magmatic origin for megacrystalline pyroxenite is inferred from their magmatic texture and the presence of unequilibrated mineral intergrowths metastable under mantle P-T conditions. Wehrlite and clinopyroxenite with abundant garnet (40-50%) and ilmenite (3-10%) are treated as a separate group of rocks. The proportion of olivine to clinopyroxene varies considerably (from 4:1 to 1:4) and the rocks span a compositional range of ilmenite-garnet clinopyroxenite to ilmenite-garnet wehrlite. These specimens are characterized by zoned orange garnet, mosaic-porphyroclastic texture (Fig. 3e), and the presence of megacrystalline ilmenite and/or garnet. Olivine occurs as neoblasts and as symplectite intergrowths with garnet. Clinopyroxene ranges in habit from euhedral rhombic grains (0·5-1·5 mm; Fig. 3f) to anhedral. It also forms strained porphyroclasts and rare lenticles of neoblasts. Garnet is typically coarser grained (1-4 mm), euhedral or polygonally shaped, and is zoned in terms of mineral inclusions. Cores and rims are free of inclusions, whereas the interior zones between the core and the rim are crammed with small inclusions of ilmenite (Fig. 3h), euhedral bluish grey-brown spinel, and an anisotropic mineral (Px? Ol?) ± phlogopite. Not all three zones are necessarily present in every garnet grain. Within the interior zone, ilmenite and silicate grains coarsen towards the garnet rims, indicating inward growth and recrystallization of the assemblage. Clinopyroxene and garnet are commonly partially melted or corroded. Ilmenite forms irregular-shaped crystals filling interstices [sideronitic texture, as defined by
Moorhouse, (1959)] and appears to have crystallized last of all phases (Fig. 3f). Rare grains of amphibole occur in euhedral rhombic crystals. The petrography of the ilmenite-garnet wehrlite and clinopyroxenite suite suggests a two-stage origin. Coarse garnet and clinopyroxene are interpreted to represent the initial crystallization event. Subsequent partial melting and corrosion is evidenced by the `spongy', cloudy, turbid appearance of some clinopyroxene grains, and replacement of garnet by a fine-grained aggregate of ilmenite + pyroxene + garnet relicts ± olivine, spinel, and phlogopite. This melting event (or metasomatic reaction?) transformed up to 80-100% of former garnet grains into thick rims of ilmenite-silicate intergrowths. The intergrowths are perhaps analogous to kelyphitic rims normally found on garnets, but are developed to an extreme degree. Alternatively, the intergrowths could be an example of ilmenite-silicate symplectites common to kimberlites (Haggerty, 1987). Late in situ recrystallization created euhedral pyroxene, olivine-garnet symplectites, and euhedral garnet rims mantling the interior zones of the ilmenite-silicate intergrowths within garnet. The megacryst suite includes forsterite, diopside, ilmenite, pyrope, enstatite, phlogopite and spinel. Modal proportions of megacrystal minerals vary between individual phases of the Jericho kimberlite. The megacrysts are normally 1-3 cm long, but diopside can reach lengths of 20-30 cm. Monomineral and monocrystal megacrysts are rare; most of them occur as polycrystalline intergrowths and/or contain smaller grains of minerals from the megacrystal assemblage. At Jericho, there is a complete textural transition from isolated megacrysts of orthopyroxene, clinopyroxene and garnet to megacrystal intergrowths of two pyroxenes and garnet to pyroxenitic nodules. Megacrystal olivine is very often deformed, and shows sub-boundaries, deformational twinning, and veins of polygonal and tabular neoblasts. Ilmenite often occurs as coarse-grained, allotriomorphic intergrowths withpyrope and diopside; megacrysts of pyrope and diopside are zoned in Ti and Cr. Based on mineral chemistry, petrography, and the presence of ilmenite, the Jericho megacrysts belong to the common `Cr-poor' group (Schulze, 1987). Electron-microprobe analysis of mineral compositions was performed with a fully automated CAMECA SX-50 microprobe (Department of Earth and Ocean Sciences, University of British Columbia), operating in wavelength-dispersion mode. Silicates were analysed at an accelerating voltage of 15 mV and a 20 mA beam current. On-peak counting times were 10 s for major and 20 s for minor elements. Precision (2[sgr], relative %) and minimum detection limits (absolute wt %) calculated from counting statistics and replicate analysis, respectively, are as follows: SiO2 (0·7, 0·03); Na2O (3, 0·02), MgO (1, 0·05); Al2O3 (3, 0·08); CaO (1·5, 0·02); TiO2 (22, 0·03); Cr2O3 (9, 0·04); MnO (27, 0·03); FeO (3, 0·04), NiO (20, 0·03). Raw data were treated with the `PAP' [phi]([rho]Z) on-line correction program. Individual phases in a sample were analysed as 10-15 points in cores and rims over 4-5 grains; phases used for thermobarometry were analysed at points of their mutual contact. Mineral compositions were averaged for homogeneous phases. Samples with strongly zoned minerals were studied by scanning electron microscopy (SEM) and zoning profiles for the mineral grains were made. Table 1 lists select mineral analyses for each rock type; an entire set of mineral compositions for the 37 studied specimens used for thermobarometry is available via ftp (http://www.oup.co.uk/petroj/hdb/Volume_40/Issue_01/dataset/egc006_dat.html and http://perseus.geology.ubc.ca/research/data). Table 1. Average compositions of minerals from representative samples of xenoliths
Garnet was analysed for trace elements (Table 2) with the Guelph scanning proton microprobe (Department of Physics, University of Guelph) and data were processed via the Guelph PIXE software package. The energy of the proton beam was 3 MeV, I was 7·5 nA, and spot size was 10 µm. Counts were collected over 4·5 min, which produced minimum detection limits of 5, 3 and 4 ppm on Ni, Ga and Zr, respectively. Table 2. Average trace element composition (ppm) of peridotitic garnet and estimated Ni-in-garnet temperatures
Olivine compositions range in mg-number from 88 to 93. The most magnesian olivine (Fo91-93) occurs in spinel-garnet coarse peridotite (Fig. 4). The range of olivine compositions from other peridotite xenoliths is Fo90·5-92 for garnet coarse peridotite, and Fo89-93 for porphyroclastic peridotites. Olivine found as megacrysts, in megacrystalline pyroxenite or in ilmenite-garnet wehrlite-clinopyroxenite is substantially less magnesian (Fo88-90·5). Concentrations of Ni in olivine from coarse and porphyroclastic peridotites have modal values of 3000-3100 ppm (0·39 wt % NiO). Ni concentrations in spinel-garnet coarse peridotite are generally higher than those in spinel-free coarse peridotite and porphyroclastic peridotite. The lowest Ni content (1300-1600 ppm) for olivine derives from samples of ilmenite-garnet wehrlite and clinopyroxenite; olivine in megacrystalline pyroxenites contains 1800-1900 ppm Ni. Figure Olivine from coarse peridotite has Cr, Ca and Ti contents below detection. In contrast, olivine in porphyroclastic peridotite exhibits elevated levels of Cr (0·04-0·11 wt% Cr2O3) and Ca (0·04-0·06 wt % CaO). In general, olivine porphyroclasts are unzoned and are identical in composition to neoblasts. However, rare samples (e.g. samples 22-7 and 21-3) contain porphyroclasts and neoblasts of differing composition where the porphyroclastic olivine has lower abundances of Cr (below minimum detection limit (MDL) vs 0·07 wt % Cr2O3), Ca (below MDL vs 0·04 wt % CaO), and higher NiO (0·42 vs 0·33 wt %) and mg-numbers (91·0 vs 90·7). Olivine in ilmenite-garnet wehrlite and clinopyroxenite can be reliably distinguished by the elevated Ti content (0·05 wt % TiO2). Orthopyroxene compositions vary little within and between different xenolith types (mg-number = 90-94). The modal value of mg-number decreases from spinel-garnet peridotite (mg-number = 92·5-93), to garnet-only peridotite (mg-number = 92-92·5), to porphyroclastic peridotite (mg-number = 91·5-92·5), and to pyroxenite (mg-number = 90-91·5). In all cases, orthopyroxene is more Mg rich than the coexisting olivine, suggesting that the parageneses are equilibrated (Gurney et al., 1979). Orthopyroxene compositions from different rock types are distinct in terms of their Cr2O3 and CaO contents (Fig. 5). Orthopyroxene in porphyroclastic peridotite contains the highest concentrations of Cr2O3 and CaO (Cr2O3 > 0·29 wt %, CaO > 0·47 wt %). Only orthopyroxene from a uniquely shallow spinel-garnet peridotite (22-1) has Ca and Cr contents more typical of high-temperature nodules. Enstatite in spinel-garnet coarse peridotite is depleted in Ca compared with orthopyroxene in garnet-only coarse peridotite (Fig. 5). Enstatite from samples of porphyroclastic peridotite or from megacrysts also tends to have higher Na2O (0·11-0·26 wt %) than does enstatite from low-temperature peridotite (0-0·14 wt %). TiO2 and Al2O3 range from 0 to 0·15 wt % and from 0·4 to 0·8 wt %, respectively. Figure Orthopyroxene grains are homogeneous and similar in composition regardless of habits and textural origin. However, several samples contain orthopyroxene that shows significant and diverse compositional zoning in Al, Cr, Ca, and Fe. Clinopyroxene in peridotite, pyroxenite and as megacrysts has high Cr content (0·6-2·2 wt % Cr2O3) and, as such, is classified as Cr-diopside. Diopside with up to 0·3% TiO2 is present in ilmenite-garnet wehrlite-clinopyroxenite. Progressively more Fe-rich, Cr-diopside occurs in coarse spinel-garnet peridotite, coarse garnet peridotite, porphyroclastic peridotite, and pyroxenite and megacrysts. Cr-diopside from coarse spinel-garnet peridotite shows a strong negative correlation between Cr2O3 and both CaO and mg-number (Fig. 6). In contrast, Cr-diopside from porphyroclastic peridotites and coarse spinel-garnet peridotite exhibits no correlation between Cr2O3 and mg-number. Cr-diopside from pyroxenites and megacrysts is compositionally identical (Fig. 6). Late clinopyroxene, present in porphyroclastic peridotite, is considerably lower in Al2O3 (0·5-0·8%), Cr2O3 (0·6-1·1%) and Na2O (0·6-0·8%), higher in TiO2 (0·3-0·6%), and demonstrates highly variable Ca/Mg ratio. Figure Whereas olivine and orthopyroxene show little or no compositional zoning, clinopyroxene commonly exhibits patchy compositional zoning near rims of grains (outer 50-100 µm) or associated with `spongy' areas in grains. This zoning always involves a decrease in Al2O3 (0·5-1 wt %) and Na2O (0·5-1·5 wt %) rimwards. Superimposed on this pattern are other chemical trends: (1) Ca enrichment with Cr depletion; (2) Ca enrichment with little or no Cr enrichment; (3) Fe-Ti enrichment. The overall chemical variations are: 2-3 wt % CaO, 0·4-0·8 wt % Cr2O3, and 0·2-0·5 wt % FeO and TiO2. Some of the Al-Na, and Cr-poor clinopyroxene growing in outermost rims of primary clinopyroxene may represent late interstitial clinopyroxene that is not equilibrated with the primary assemblage. Garnet in Jericho peridotites contains high MgO, low CaO and moderate Cr2O3, and is classified as pyrope. All pyrope compositions plot within the `G9' field of lherzolitic, Ca-saturated garnet compositions established by
Dawson & Stephens, (1975). However, some of the garnet-bearing samples are, in fact, harzburgite. Garnet in porphyroclastic peridotite is more Ti rich, Mg rich, and Cr rich (Fig. 7) on average than garnet in coarse peridotite: 7·70 wt % Cr2O3 ± 2·02 wt % vs 4·17 ± 1·26 wt %, respectively. Plotted as Cr2O3 vs CaO, pyrope compositions define two trends: a common `lherzolitic' trend (Sobolev et al., 1973) parallel to the G9-G10 boundary, and a more exotic trend showing less Cr2O3 enrichment with increasing Ca. The latter trend extends well into the compositional field of wehrlitic garnet (Sobolev et al., 1973) and comprises pyrope compositions from spinel-bearing peridotite. Ilmenite-garnet wehrlite-clinopyroxenite rocks contain titaniferous pyrope (1-3 wt % TiO2), which is more enriched in Ti than all previously described garnet in kimberlite and kimberlite-hosted xenoliths (Dawson & Stephens, 1975; Mitchell, 1986; Solovjeva et al., 1994). Figure Trace element concentrations were measured in garnet from coarse, porphyroclastic, and ilmenite-bearing peridotite (Table 2). Coarse peridotite contains 11-48 ppm Ni, 0-12 ppm Ga, 5-26 ppm Y, and 2-57 ppm Zr. Porphyroclastic peridotite demonstrates similar levels of Ga (7-10 ppm) and Y (11-17 ppm), but elevated levels of Ni (39-92 ppm) and Zr (22-72 ppm). Garnet in ilmenite-garnet wehrlite is considerably richer in minor elements containing 25-88 ppm Ni, 7-36 ppm Ga, 8-22 ppm Y, and 18-145 ppm Zr. More than half of the Jericho peridotitic garnets show Y concentrations exceeding 10 ppm and one-third of peridotitic garnets demonstrate high Zr (>30 ppm); such elevated levels of Y and Zr are usually taken as evidence of relatively `high-temperature' and `low-temperature' metasomatic enrichment, respectively (Griffin & Ryan, 1995; Griffin et al., 1996). The common linkage between enrichment in Y and Zr and temperature, though, is not observed at Jericho. Here, the cryptic metasomatism was not associated with any particular range of temperature, as no correlation is noted between Y and Zr concentration and Ni content. The metasomatic introduction of incompatible elements resulted in high variance of Ni, Y, and Zr from grain to grain far exceeding the normal variance seen in homogeneous equilibrated rocks. Garnet from some samples also shows major element heterogeneity between grains and within grains. The within-grain zoning is commonly patchy in peridotitic garnet (Fig. 3g) and areas in rims are usually depleted in CaO (<1·2 wt %) and Cr2O3 (<2·5 wt %). Ti-pyrope in ilmenite-garnet wehrlite and clinopyroxenite exhibits extreme variation in TiO2 (from 0·34 to 3 wt % in one sample) and corresponding variations in CaO, Al2O3, and SiO2 apparently linked to garnet habits and degree of melting. Rims of euhedral garnet show oscillatory zoning in Ti and Cr (Fig. 3h). Spinel (chromite with 0-0·4 wt % TiO2, 33-60% Cr2O3, and 9-16% MgO) occurs only in coarse peridotites and demonstrates common negative correlation of Mg and Cr. The low-Cr character places the Jericho xenolithic chromite far below the diamond inclusion field (Gurney & Zweistra, 1995) in the Cr-Mg space. As discussed above, some xenoliths contain mantle minerals that show within-grain or between-grain compositional variations. Compositions of cores of mineral grains demonstrate the lowest within-grain variances, whereas rims of mineral grains can be heterogeneous and overgrown by secondary clinopyroxene. We have therefore restricted our use of thermobarometric calculations to mineral grains that showed homogeneous core compositions. We believe that these estimates will yield the closest approximation to an ambient palaeo-thermal regime. Based on these restrictions we obtained 37 P-T estimates for Jericho xenoliths (Table 3). Pressures and temperatures obtained from rims of zoned minerals clearly must reflect dynamic conditions caused by perturbations of geothermal gradients resulting from tectonic, magma generation, or emplacement events. We have investigated the impact of using rim rather than core mineral compositions in our thermobarometric calculations. For coarse peridotites which tend to show better textural equilibrium, individual P-T points shifted on average by 80°C and 4 kbar, but remained within the same P-T array. If rim compositions of minerals in porphyroclastic peridotite were used, P-T points shifted by 10-100°C and 6 kbar, became more scattered towards lower P, and sometimes moved off the actual P-T array. Table 3. Pressure and termperature estimates
Equilibrium P-T estimates are computed for orthopyroxene-bearing samples (peridotites and megacrystalline pyroxenite) based on compositions of coexisting garnet-clinopyroxene-orthopyroxene. We have applied several of the geothermobarometric solutions recommended for kimberlite-derived peridotitic assemblages. Temperature estimates were made with the geothermometer of
Finnerty & Boyd, (1987) (FB), the geothermometer of
O'Neill & Wood, (1979) (OW), and the two-pyroxene geothermometer of
Brey & Köhler, (1990) (BK). Pressure was estimated with the geobarometer of
MacGregor, (1974) (MG) and the Al-in-Opx geobarometer of
Brey & Köhler, (1990) (BK). The above methods were coupled in three different combinations of thermometers and barometers (Table 3): the FB-MG solution (Fig. 8a), the OW-MG solution (Fig. 8b), and the BK solution (Fig. 8c). Figure Among the three solutions, the FB-MG method was shown to be in best agreement with the diamond-graphite constraint, placing diamond-bearing xenoliths in the diamond stability field and graphite-bearing xenoliths in the graphite stability field (Pearson et al., 1994). One concern is that the FB thermometer is based on Ca partitioning between clinopyroxene and orthopyroxene, and some pyroxene from Jericho xenoliths shows zoning in Ca. As an independent check for two-pyroxene temperatures, we also calculated temperatures based on Fe-Mg exchange between olivine and garnet (OW-MG solution). The OW thermometer reportedly agrees with the FB thermometer to within ±50°C (Ryan et al., 1996). The BK thermobarometer is recommended by the fact that it is calibrated on recent phase equilibria experiments for natural lherzolite compositions at 10-60 kbar and 900-1400°C (Brey et al., 1990). Furthermore, the BK method lacks the problems allegedly associated with the
MacGregor, (1974) and the
Finnerty & Boyd, (1987) techniques (Carswell & Gibb, 1987; Carswell, 1991). The Jericho xenoliths offer several features that can be used to evaluate the accuracy of calculated pressures and temperatures. These features include: (1) the presence of spinel-garnet peridotite, which should record P-T conditions consistent with the experimentally constrained spinel-garnet transition curve, and (2) the highly diamondiferous character of the Jericho pipe, which implies that the kimberlite sampled the `diamond window' in the mantle. The `diamond window' was originally defined as a P-T interval within the diamond stability field of the mantle that hosts peridotite with depleted (in Y, Zr, and Ti) mineral chemistry (Griffin & Ryan, 1995). We could use the `diamond window' as a petrological constraint, as this depletion in normally seen only in low-temperature, coarse peridotite. In other words, in diamondiferous kimberlite we should expect a relatively large interval of P and T where low-temperature coarse peridotite in the diamond stability field does not coexist with high-temperature porphyroclastic peridotite. All three thermobarometric solutions place spinel-garnet peridotite at P-T conditions close to the independently calculated spinel-garnet transition curve (Fig. 8). In natural samples this transition is not univariant and occurs over a range of pressures and temperatures (O'Neill, 1981; Webb & Wood, 1986). The computed spinel-garnet transition curve lies in the lower part of the spinel-garnet peridotite horizon for the FB-MG and the OW-MG solutions. The lowest two P-T estimates for spinel-garnet peridotite in the FB-MG solutions are probably underestimated, as they plot outside the garnet stability field. In contrast, P-T estimates for spinel-garnet peridotite derived from the BK model are centred on the spinel-garnet transition curve, and thus best satisfy this criterion. Both the FB-MG and the OW-MG solutions meet the second condition, namely, that the samples define a relatively wide (large [Delta]P) `diamond window' (Fig. 8). The BK results suggest a narrow window (7 kbar, Fig. 8c) because porphyroclastic peridotite occupies the same depth interval as coarse garnet peridotite. The mean position of the P-T array does not change appreciably as a result of the different solution techniques. Only the presence and position of the inflection point is significantly affected by the choice of geothermobarometers. The FB-MG and the OW-MG solutions show high-temperature inflections at 190 km (Fig. 8a) and 180 km (Fig. 8b), respectively. The BK-generated P-T array (Fig. 8c) shows no apparent inflection, but does show more scatter among the higher temperature points. Instead, it suggests the coexistence of low-temperature coarse peridotite and pyroxenite with high-temperature porphyroclastic peridotite at depths of 160 km and greater. Major-element thermobarometry was supplemented by an independent geothermometric method based on partitioning of Ni between chrome-pyrope and olivine [Griffin et al., 1989; Canil, 1994; Ryan et al., 1996 (RG thermometer)]. Temperatures were calculated for Jericho xenoliths (Table 2) using measured Ni concentrations in olivine, which are commonly higher than the fixed value of 2900 ppm used by
Ryan et al., (1996). Relative to the formulation of
Canil, (1994), the RG thermometer reproduces the OW olivine-garnet and the FB two-pyroxene temperatures better, because the RG thermometer was empirically calibrated against the FB thermometer (Griffin et al., 1989) and recalibrated (Ryan et al., 1996) against the OW thermometer. Based on thermobarometric results, the peridotite samples can be assigned to a low-temperature and a high-temperature suite. All samples which are consistent with a steady-state conductive geotherm are considered as part of the low-temperature suite. Peridotite xenoliths that fall off the geotherm by shifting to higher temperatures belong to the high-temperature class. AtJericho, low-temperature peridotite is dominantly coarse; only two out of 37 samples have disrupted porphyroclastic texture. High-temperature peridotite samples always show non-disrupted porphyroclastic textures. Ni thermometry also supports a lower-temperature origin of coarse peridotite compared with porphyroclastic Jericho peridotite. Coarse peridotite yielded temperatures (Ryan et al., 1996) of 580-990°C, with spinel-bearing specimens having the lowest temperatures. Temperatures of 1000-1230°C were recorded for porphyroclastic peridotite, and 990-1180°C for ilmenite-garnet wehrlite and clinopyroxenite. The thermobarometry data of Fig. 9 can be used to compare the thermal state of the Slave mantle with cratonic mantle from elsewhere. The central Slave mantle as probed by the Jericho, Grizzly and other Lac de Gras kimberlites is colder than the mantle sampled by other kimberlites of the North American plate (Montana, Colorado-Wyoming, Kirkland Lake, and Kentucky kimberlites). This conclusion holds true regardless of the thermobarometric method applied. Furthermore, theJericho and Lac de Gras kimberlites located in the deep interior of the North American craton probed deeper and cooler lithosphere than that sampled by kimberlites located on the periphery of the North American craton. This supports the
Finnerty & Boyd, (1987) hypothesis that the lithosphere-asthenosphere boundary shallows toward the Arctic (Somerset Island) in the north and New Mexico in the south because of their hotter thermal state. Nevertheless, some of the central Slave kimberlites may be consistent with a slightly hotter upper mantle than found below Jericho, as suggested by data from the Torrie pipe (MacKenzie & Canil, 1998). Within a global context, the thermal state of the north-central Slave mantle represents the cool end of mantle cratonic regimes
(Nyblade & Pollack, 1993), and is definitely cooler than cratonic upper mantle from beneath Kaapvaal andSiberian cratons (Fig. 9). Figure The mantle beneath the northern part of the central Slave craton shows many features common to cratonic mantle elsewhere. Peridotite is the dominant rock type within the mantle underlying the Slave craton. Similar to mantle sampled by most Type I kimberlites, there are two distinct suites of low-temperature and high-temperature peridotite. Jericho low-temperature garnet peridotite has Mg-enriched mineral compositions, typical of peridotite underlying other cratons. The similarity in olivine composition between the Slave and other cratons indicates their comparable degree of depletion. The modal mg-number of olivine in low-temperature Jericho peridotite (92-93) is identical to that reported from other Slave peridotite (Boyd & Canil, 1997; MacKenzie & Canil, 1998) and parallels the average range for Kaapvaal and Udachnaya (Boyd et al., 1997). Low-temperature peridotite is generally coarse textured, and only rarely displays disrupted textures. At Jericho, xenoliths of high-temperature peridotite show Fe-, Ca- and Ti-enriched mineral compositions, and always have porphyroclastic textures. Both these features are shared by most other world-wide cratonic peridotite suites (Boyd, 1987; Harte & Hawksworth, 1989). In addition, garnet within high-temperature peridotite is always non-disrupted. Disrupted garnet is present only within the low-temperature suite, where it is attributed to higher stress regimes associated with cooler mantle. This parallels what has been described from the Kaapvaal craton (Boyd & Nixon, 1978). The compositional range found in peridotitic garnet at Jericho suggests that the peridotitic mantle beneath the central Slave shares one more feature with portions of the Kaapvaal cratonic mantle. Specifically, there is a pronounced lack of G10 (harzburgitic) garnet compositions found in Jericho harzburgite (Fig. 7), in Jericho kimberlite concentrate (Kopylova et al., 1998a), and in the Lac de Gras harzburgite (Pearson et al., 1998). In the Kaapvaal craton, the absence of G10 garnet compositions was interpreted to result from equilibration of harzburgitic assemblages with a lherzolite-dominated mantle (Boyd & Nixon, 1978; Skinner, 1989). In such harzburgite, garnet is saturated in Ca despite the local absence of clinopyroxene in rocks. One last feature that is shared by peridotite of the Slave and Siberian cratons is the presence of a late-stage Na-, Al- and Cr-depleted clinopyroxene. The clinopyroxene occurs as a fine-grained coating along orthopyroxene grain boundaries in peridotite xenoliths recovered from the Jericho pipe as well as from the Grizzly pipe (Boyd & Canil, 1997). Similar clinopyroxene has been described in Udachnaya peridotite and was ascribed to secondary crystallization during eruption (Boyd et al., 1997). Some substantive differences between peridotitic xenoliths from Jericho and peridotite described from other cratons suggest corresponding differences in the character of the underlying peridotitic upper mantles. Jericho peridotitic xenoliths show three unique chemical features, including: (1) an anomalously high proportion of chemically unequilibrated samples; (2) a distinct Cr enrichment in mineral chemistry of high-temperature peridotite, relative to low-temperature samples; (3) unique trends in garnet and clinopyroxene compositions within spinel-bearing peridotite that derive from equilibration with spinel. Almost half of the analysed samples of high- and low-temperature peridotite are chemically unequilibrated; minerals show abundant between- and within-grain chemical variation. The chemical variation is irregular, in that individual grains show patchy zoning, generally restricted to the rims, and mineral chemistry does not always correlate with grain shape and origin (e.g. porphyroclast vs neoblast). The most heterogeneous minerals are clinopyroxene and garnet; the least heterogeneous is olivine. Olivine, orthopyroxene and garnet in high-temperature (spinel-free) peridotite are Cr rich compared with low-temperature peridotite, whereas clinopyroxene is not. However, it is mainly within high-temperature peridotite that clinopyroxene shows compositional zoning involving Cr enrichment. These aspects of Cr enrichment of mineral assemblages are not described for other high-temperature peridotite xenoliths. There is, however, a distant similarity to patterns seen in sheared low-temperature nodules from the Kimberley pipes. In two of the Kimberley nodules, orthopyroxene porphyroclasts were found to be lower in Cr, Ti, Al and Ca than orthopyroxene neoblasts (Boyd, 1975). Similar effects have been sought, but not found, in sheared lherzolite from Thaba Putsoa, Lesotho, and Frank Smith. Garnet from the Kimberley deformed rocks tends to be more Cr rich than garnet from coarse peridotite, but this may reflect the presence of chromite (Boyd & Nixon, 1978). Cr enrichment in minerals from Jericho high-temperature peridotite may result from one of two processes. First, the high Cr content could be a primary feature reflecting the absence of spinel in these deep-seated rocks. Alternatively, the high Cr content may derive from secondary processes linked to mantle deformational events. The latter explanation is suggested by rims of clinopyroxene porphyroclasts and neoblasts that show Cr enrichment, and by coexisting (albeit rare) porphyroblasts of olivine that have retained their early Cr- and Ca-depleted core compositions. Equilibration with spinel affected the composition of clinopyroxene and garnet in Jericho spinel-garnet peridotite. Clinopyroxene compositions from spinel-garnet peridotite show an unusual negative correlation in Mg and Cr (Fig. 6). The trend in clinopyroxene compositions is reminiscent of the compositional variations commonly found in mantle, Ti-poor spinel and also characteristic of the Jericho chromite. Typically, the Cr contents of these spinel compositions are positively correlated with Fe (Haggerty, 1991), implying a negative correlation with Mg. If the partitioning of Cr and Mg into silicates is controlled by the presence of spinel with a constant Cr/Mg ratio, we might expect similar negative Mg-Cr correlations in contemporaneous silicate phases. Pyrope garnet that has also equilibrated with spinel defines a strong and unique compositional trend that is less enriched in Cr relative to the common `lherzolitic' trend (Fig. 7). This trend in garnet composition is found in garnet from spinel-garnet peridotite and from heavy mineral concentrates at Jericho, and appears to be controlled by the substitution of uvarovite for pyrope (Ca2Cr3 Mg2Al3). Smith & Boyd, (1992) have described a similar trend for garnet compositions from spinel-garnet peridotite from South Africa. Furthermore, similar ranges in garnet composition have been reproduced experimentally (Fig. 10) for natural compositions of lherzolite (Brey et al., 1990; Brey, 1991). The experiments were performed at mantle conditions (i.e. 900-1200°C and 3-6 GPa) and used spinel-bearing starting materials. Based on these observations we suggest that this trend in garnet compositions results from equilibration with spinel. An outstanding problem is that this `spinel-garnet equilibrium' trend is apparently rarely found in mantle peridotite and kimberlite concentrates, despite the common occurrence of spinel in coarse low-temperature peridotite (Boyd, 1987). This suggests that coexisting spinel and garnet in mantle peridotite are not always at equilibrium. Figure The mantle underlying the north-central Slave craton comprises, in part, an unusually high proportion of magmatic-textured, non-peridotitic rocks that appear to be related in origin to megacrysts. The magmatic suite includes unique high-temperature megacrystalline pyroxenite and ilmenite-bearing rocks. Pyroxenite rocks in cratonic mantle are found in the Kaapvaal craton in Matsoku (Harte et al., 1975) and Monastery kimberlites (Gurney et al., 1991), and in North America in Colorado-Wyoming kimberlite (Eggler & MacCallum, 1974), but are most widespread in Siberia. There, they have been described in the Mir, Obnazhennaya and Udachnaya pipes (Solovjeva et al., 1987; Spetsius & Serenko, 1990), and in the last they represent up to 6% of the mantle xenoliths (Solovjeva et al., 1994). In all of these occurrences, pyroxenite xenoliths have metamorphic granoblastic textures ranging from megacrystalline to fine grained (Solovjeva et al., 1987), and relatively low equilibrium temperatures (600-1100°C). Past workers have suggested that mantle-derived pyroxenite originated as relatively young, magmatic intrusions (Harte & Hawksworth, 1989; Harte et al., 1987) or perhaps as igneous cumulates (Harte et al., 1975). Subsequently, these igneous mantle rocks have been recrystallized, as evidenced by widespread exsolution of all pyroxenes, and sometimes deformed. In general, these rocks become increasingly granoblastic and fine grained as the extent of recrystallization increases (Solovjeva et al., 1987, , 1994). Pyroxenite from Jericho differs from other occurrences of cratonic pyroxenite in two aspects. First, the suite is dominated by magmatic textures rather than textures of metamorphic recrystallization or subsolidus re-equilibration. Second, the suite records significantly higher equilibrium temperatures, indicating a formation within the deep thermally disturbed mantle. This combination of characteristics, quenching of high-temperature formation conditions and original magmatic textures, places substantial constraints on the thermal history of the pyroxenite. It implies that the samples were removed from ambient mantle conditions shortly after their formation or the surrounding mantle conditions changed drastically. At Jericho, the suite of megacrystalline pyroxenite seems to be related genetically to the megacryst (low-Cr) suite. The mineral chemistry of the megacrysts is identical to that of megacrystalline pyroxenites (e.g. Fig. 6). Furthermore, there is a complete textural transition from megacrystalline pyroxenite to megacrystic intergrowths of garnet-pyroxene to isolated megacrysts. Pyroxenite and megacrysts occupy the same depth interval in the Slave mantle, as demonstrated by clinopyroxene-garnet thermometry (Kopylova et al., 1998c) and two-pyroxene-garnet thermobarometry (Table 3). The transitional nature of megacrystalline pyroxenites is a common feature found also in kimberlite-hosted xenoliths from other cratons. In part, this association has been masked by classification problems involving small megacrystalline pyroxenite xenoliths (Solovjeva et al., 1994). The problem arises in knowing whether to treat each sample as a representative of a megacryst suite or as a xenolith. For example, small pyroxenitic nodules comprising several grains of pyroxene are sometimes ascribed to a megacryst suite (Mitchell, 1986; Schulze, 1987) whereas megacrysts with extensive exsolution features have been described as polymineral xenoliths (Kirkley et al., 1984). The Jericho megacrystalline pyroxenite could represent pegmatites derived from precursor mafic magmas. Crystallization within these earlier magmas, which are intrinsically related to kimberlite magmas, has been suggested as a source for Cr-poor megacrysts (Schulze, 1987). The ilmenite-garnet wehrlite and clinopyroxenite suite of xenoliths described at Jericho belong to a rare variety of cratonic rocks containing ilmenite. Other descriptions of their counterparts in kimberlite-derived xenolith suites include ilmenite peridotite and pyroxenite xenoliths from Siberia (Ponomarenko, 1977; Rodionov et al., 1988; Solovjeva et al., 1994), Fe-Ti pyroxenite veins in peridotite xenoliths from South Africa (Harte et al., 1987), and other South African occurrences of exotic ilmenite-garnet-bearing pyroxenite xenoliths (Boyd et al., 1984; Clarke & McKay, 1990). The Jericho ilmenite-bearing xenoliths share a number of traits with these occurrences. For example, these rocks tend to have: (1) highly variable mineral modes (Tomkins & Haggerty, 1984; Clarke & MacKay, 1990; Solovjeva et al., 1994), (2) sideronitic texture (Boyd et al., 1984; Rodionov et al., 1988; Clarke & MacKay, 1990; Solovjeva et al., 1994), (3) mosaic-textured silicate minerals in contact with ilmenite (Harte et al., 1987; Rodionov et al., 1988; Solovjeva et al., 1994), (4) megacrystalline grains of ilmenite and garnet (Rodionov et al., 1988; Solovjeva et al., 1994), and (5) abundant compositional zoning in minerals (Boyd et al., 1984; Rodionov et al., 1988). In general, past workers have argued that these rocks represent crystallization products from a magma that is related both to the megacryst suite and to later kimberlite magmas (Boyd et al., 1984; Rodionov et al., 1988; Clarke & MacKay, 1990; Solovjeva et al., 1994). We suggest that the Jericho ilmenite-garnet wehrlites and clinopyroxenites originated by partial melting and magmatic and/or metasomatic recrystallization of pre-existing garnet-clinopyroxene segregations. The parental magmas were also a source for the ubiquitous ilmenite (± garnet and clinopyroxene) megacrysts found in the kimberlite, and were notably enriched in incompatible elements as evidenced by extremely high Zr and Ga contents in garnet. On the basis of our observations at Jericho, we suggest that the two pyroxenitic suites (megacrystalline and ilmenite bearing) may be linked in origin to megacrystal magmas. These pyroxenitic xenoliths volumetrically form up to 13% of our xenolith collection. Even considering the over-representation of exotic rock types in the collection, such a proportion of pyroxenitic magmatic rocks is unusually high for a cratonic mantle. We think that this phenomenon might not be a characteristic of the north-central Slave mantle, but is rather related to temporal and spatial vagaries of the processes that created the Jericho kimberlite magma. P-T arrays from kimberlite-hosted mantle xenoliths offer insight into the thermal structure of the mantle and commonly are used to constrain cratonic geotherms. Common practice is to match the observed P-T array against model steady-state, conductive geotherms established for average models for continental lithosphere and various values of model surface heat flow (Q0) (Pollack & Chapman, 1977). Alternatively, other workers have used forward modelling methods to simulate a geotherm for an assumed crust-mantle configuration (e.g. Bussod & Williams, 1991; Cull et al., 1991). The geometries of these conductive geotherms, whether based on average models of the Earth or specific models for a region, are highly sensitive to the architecture and composition of the crust and mantle (e.g. Chapman, 1986; Rudnick et al., 1998), and may be completely invalid if the assumed structure is inappropriate and can lead to serious misinterpretations (Rudnick et al., 1998). Therefore, we have taken a slightly different approach in our interpretation of the P-T array. Our approach is to adopt and fit a relatively simple, one-dimensional, analytical model for steady-state conductive heat transfer to the P-T data array to constrain values of surface heat flow at the time of kimberlite emplacement (Russell & Kopylova, 1998). This method is relatively independent of assumptions about crustal structure and, therefore, is especially well suited for regions with sparse geophysical data or for deducing palaeo-geotherms. The model steady-state conductive geotherm assumes an exponential decrease in heat-producing elements over a critical depth D and has the form (Lachenbruch, 1968; Crowley, 1987)
At this point we invert the P-T array data (Fig. 8) to obtain model estimates of surface heat flow (Q0) and D given geologically defined values of surface heat production (A0) and thermal conductivity (K). We assigned a constant value of 2·5 to K, based on reported values of K for upper (2·37 W/m per K) and lower (2·64 W/m per K) crust from cool cratonic regimes (Chapman, 1986). The bedrock geology for an area of 14 300 km2 surrounding the Jericho kimberlite was used to calculate an average value of surface radiogenic heat production (A0) of 2·16 µW/m3 using the surface area proportions of rock types and the bulk chemical compositions reported by
Davis, (1991) (Table 4). Table 4. Parameters used to calculate a weighted average value for surface heat generation based on the bedrock geology surrounding the Jericho kimberlite pipe
Our analysis utilizes only P-T data that are representative of the stable, low-temperature portion of the geotherm; we have not included P-T points from high-temperature peridotite in our inversion (Fig. 11). The results for high-temperature peridotite fall off the ambient geotherm and define a thermal disturbance that is seen in many Type I kimberlite geotherms and that is commonly ascribed to convection processes associated with the magma-bearing zone at the lithosphere to asthenosphere transition (e.g. Boyd & Gurney, 1986; Boyd, 1987; Harte & Hawkesworth, 1989). Figure We fitted the model equation to P-T data sets resulting from both the FB-MG and the BK solutions. Optimal values of the model parameters Q0 and D have been derived by minimization of the [chi]2 merit function (Press et al., 1986):
where Tobs is the observed temperature value, Sj is the uncertainty in Tobs because of analytical variance and T* is the model temperature. The model geotherm derived from the FB-MG data array has parameters D = 19·6 km and Q0 = 52·3 mW/m2. The optimal solution reproduces 95% of the P-T data to within ±40°C. Also shown (Fig. 11) is the model geotherm fitted to the BK array of P-T estimates. It should be noted that although the latter P-T estimates are offset to lower values of P and T, the corresponding model fit parameters are nearly identical in value (D = 19·5 km; Q0 = 52·9). Thus, regardless of which of these thermobarometric methods is preferred, the deduced thermal state of the Slave lithosphere is the same. We tested the sensitivity of our results to variations in K; a 10% change in the value of K (e.g. 2·75 W/m per K) causes a 5% variation in the fit parameters (e.g. <1 km in D and <3 mW/m2 in Q0). The estimated model values of surface heat flow Q0 = 52·3-52·9 mW/m2 show an excellent agreement with two heat flow measurements available for the Slave craton: 54·4 ± 0·4 mW/m2 (Beck & Sass, 1966) and 53·3-54·4 mW/m2
(Lewis & Wang, 1992). These estimates suggest that the Slave craton has unusually high surface heat flux for a Precambrian terrane. However, this distinction does not extend deeper into the upper mantle. Our model predicts a thermal gradient in the mantle of 4·3°C/km, which implies a mantle heat flow of 12·9 mW/m2 for an assumed K for the mantle of 3·0 W/m per K. This assumption is based on latest estimates of thermal diffusivity of the upper mantle [(7-8) * 10-7 m2/s; Katsura, 1995); heat capacity (1·24 J/kg per K; Stacey, 1992) and density (Turcotte & Schubert, 1982). The mantle heat flow of the north-central Slave craton is, therefore, identical to other estimates of mantle heat flow for the Canadian shield (12 mW/m2, Mareschal et al., 1997; and 14 mW/m2, R. Hyndman & T. Lewis, in preparation) and for other Precambian terranes (10-15 mW/m2, Rudnick et al., 1998). The relatively cool upper mantle below the Slave craton is, however, not at odds with the observed high surface heat flow. Heat production within the radiogenically enriched Slave crust more than adequately compensates the surface heat flux for the slightly cooler deep mantle underlying the craton. (1) The following types of mantle xenoliths from the Middle Jurassic Jericho kimberlite in the Slave craton of Northern Canada are described: (a) coarse peridotite (48%; spinel-garnet and spinel free); (b) porphyroclastic peridotite (31%; non-fluidal, fluidal non-disrupted and fluidal disrupted); (c) megacrystalline pyroxenite (16%); (d) ilmenite-garnet wehrlite and clinopyroxenite (5%).
(2) The varieties of the upper-mantle xenoliths, and their proportions, petrography and mineralogy, mainly resemble those from kimberlites of other cratons. Based on major and minor element thermobaromety, the peridotite samples are assigned to a low-temperature and a high-temperature suite. The low-temperature depleted peridotite is coarse or disrupted porphyroclastic; high-temperature fertile peridotite samples always show non-disrupted porphyroclastic textures.
(3) Some substantive differences between peridotitic xenoliths from Jericho and peridotite described from other cratons suggest corresponding differences in the character of the underlying peridotitic upper mantles. Jericho peridotitic xenoliths show three unique chemical features: (a) an anomalously high proportion of chemically unequilibrated samples; (b) a distinct Cr enrichment in mineral chemistry of high-temperature peridotite, relative to low-temperature samples; (c) unique trends in garnet and clinopyroxene compositions within spinel-bearing peridotite that reflect equilibration with spinel.
(4) The mantle below Jericho comprises, in part, an unusually high proportion of magmatic-textured, non-peridotitic rocks that appear to be related in origin to megacrysts. The magmatic suite includes unique high-temperature megacrystalline pyroxenite and ilmenite-bearing rocks. In contrast to other occurrences of cratonic pyroxenites, the Jericho pyroxenites have magmatic rather than metamorphic granoblastic textures, and higher equilibrium temperatures, indicating formation within the deep thermally disturbed mantle.
(5) The cool upper mantle below the central Slave coexists with a highly radiogenic, hot crust. The central Slave mantle is colder than the mantle beneath the rest of the North American craton and the mantle beneath Kaapvaal and Siberian cratons. A new approach was used to invert the Jericho thermobarometric data to obtain model estimates of surface heat flow given geologically defined values of surface heat production and thermal conductivity. The estimated model values of surface heat flow of the north-central Slave craton are Q0 = 52-53 mW/m2; whereas the mantle heat flow is 13 mW/m2. These values show excellent agreement with both measured values of surface heat flow for the Slave craton and theoretically predicted values of mantle heat flow beneath cratons. This research was made possible by an NSERC-Industry Collaborative Research and Development grant (94-0972) sponsored by Canamera Geological Ltd. We particularly wish to acknowledge support received from J. Dupuis and B. Dynes of Canamera Geological Ltd, and thank Lytton Minerals Ltd for permission to study and publish details on samples from Jericho. Critical helpful comments from J. Boyd, D. Francis, and W. Griffin on an earlier version of the manuscript are greatly appreciated.INTRODUCTION
GEOLOGICAL SETTING
XENOLITH PETROGRAPHY
Coarse peridotite
Porphyroclastic peridotite
Megacrystalline pyroxenite
Ilmenite-garnet wehrlite and clinopyroxenite
Megacrysts
MINERAL CHEMISTRY
Analytical methods
Olivine
Orthopyroxene
Clinopyroxene
Garnet and spinel
PRESSURE-TEMPERATURE CALCULATIONS
Method
Results
DISCUSSION
Commonality of the Slave peridotitic mantle
Unique features of the Slave peridotitic mantle
Unique aspects of the Jericho pyroxenitic mantle
`Cool' upper mantle and `hot' crust
CONCLUSIONS
ACKNOWLEDGEMENTS
REFERENCES
