Journal of Petrology Advance Access originally published online on November 30, 2006
Journal of Petrology 2007 48(2):231-252; doi:10.1093/petrology/egl067
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Ferric Iron in CaTiO3 Perovskite as an Oxygen Barometer for Kimberlite Magmas II: Applications
School of Earth and Ocean Sciences, University of Victoria, 3800 Finnerty Road, Victoria, BC, V8W 3P6, Canada
RECEIVED OCTOBER 27, 2005; ACCEPTED OCTOBER 16, 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSAMPLESRESULTSDISCUSSIONSUPPLEMENTARY DATAAPPENDIX: DETAILED DESCRIPTION...REFERENCES |
|---|
We apply an oxygen barometer based on the Fe content of CaTiO3 perovskite to estimate the oxygen fugacity (fO2) during the crystallization and emplacement of kimberlites in different eruptive phases of a single pipe, or between different pipes, clusters or provinces. Mineral chemical data for perovskite were compiled from the literature and obtained in our detailed study of perovskites from 11 kimberlites at Somerset Island and Lac de Gras, Canada. Perovskite compositions in kimberlites record a range in fO2 of many orders of magnitude from NNO–5 to NNO+6 [where log fO2 is given relative to the nickel–nickel oxide (NNO) buffer]. The range of fO2 recorded by different parageneses of perovskite within a single pipe can vary up to three orders of magnitude with trends toward both oxidation and reduction during crystallization. Kimberlites record some of the greatest ranges, and the highest known fO2 conditions for any terrestrial magma. This is attributed to the presence of deep and oxidized source regions and the variable interplay of ferric–ferrous vs carbon–fluid equilibria during ascent of kimberlite magmas. Three kimberlite pipes from the Lac de Gras field show that higher fO2 values correlate with higher proportions of more resorbed diamonds, suggesting that this variable has a measurable effect on the physical properties of diamonds in a pipe.
KEY WORDS: kimberlites; oxygen fugacity; perovskite; diamond; redox; mantle
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSAMPLESRESULTSDISCUSSIONSUPPLEMENTARY DATAAPPENDIX: DETAILED DESCRIPTION...REFERENCES |
|---|
Kimberlites have a special place in petrology as being one of the few primary host rocks for diamond, and the most deep-seated igneous rock known. The oxygen fugacity (fO2) is an intensive variable that varies more than any other in igneous rocks (Carmichael, 1991
). If the fO2 of magmas reflects that of their source region (Carmichael, 1991), that of kimberlites affords a view into the mantle source for magmas (i.e. asthenosphere) at depths in the Earth greater than 200 km. Furthermore, fO2 controls the phase equilibria during the differentiation of magmas (Osborn, 1959; Carmichael & Ghiorso, 1986), and the speciation and stability of phases with redox-sensitive elements (C, Fe, S, etc.), such that conditions of high fO2 may promote the corrosion of diamonds carried by kimberlite.
Olivine and chromite are early liquidus phases during the crystallization of kimberlite magma (Mitchell, 1986). Fedortchouk & Canil (2004) examined chromite in fresh olivine phenocrysts in kimberlites from Lac de Gras, Canada, and showed that they formed between 1030°C and 1170°C at a maximum fO2 of NNO–2·2 (where NNO is the nickel–nickel oxide buffer). It is possible, however, that the fO2 of kimberlites and many magmas changes during their ascent, crystallization and emplacement from deep in the mantle to the surface (Ballhaus, 1993; Ballhaus & Frost, 1994). To characterize the fO2 for the entire evolution of kimberlites from source to emplacement requires an approach to determine their fO2 along the complete liquid line of descent.
Perovskite (CaTiO3) is an accessory phase whose appearance in igneous rocks is favoured under conditions of low silica activity (Verhoogen, 1962; Carmichael & Nicholls, 1967). For this reason, perovskite is common in most kimberlites, generally crystallizing after chromite, olivine and monticellite (Mitchell, 1973). Estimates of kimberlite fO2 based on heterogeneous equilibria involving perovskite were first made by Carmichael & Nicholls (1967), who noted results more oxidized than for most magmas (
NNO+1). Their calculation assumed equilibrium with diopside, now known typically to be a result of contamination of kimberlite magma (Mitchell, 1986).
The Fe content of CaTiO3 perovskite in kimberlite has been experimentally calibrated by Bellis & Canil (2006) as an oxygen barometer using an empirical relationship
|
| (1) |
, and Nb and Fe as cations per three oxygens). For the conditions of their experiments, equation (1) shows no dependence on temperature (T) or the bulk composition of the kimberlite starting material, and has an estimated uncertainty of ±1 log fO2 unit. Because perovskite can crystallize in several different generations in kimberlites (Chakmouradian & Mitchell, 2000), this oxybarometer can potentially unravel, in rich detail, the fO2 recorded during kimberlite formation and emplacement.
In this article, we first study the fO2 recorded by CaTiO3 perovskite from several kimberlites in the literature. We then use published data on a single northern Alberta pipe (Eccles et al., 2004), and new mineral chemical data we have gathered in a detailed study of different perovskite parageneses from 11 well-characterized kimberlite pipes, to determine the variation in fO2 between eruptive phases of a single pipe, locally between a cluster of different pipes, or within and between kimberlite provinces on a continental scale. Finally, we compare the redox state of kimberlites with other mantle-derived magmas and comment on the oxygen fugacity of these rocks.
|
SAMPLES |
|---|
|
TOPABSTRACTINTRODUCTIONSAMPLESRESULTSDISCUSSIONSUPPLEMENTARY DATAAPPENDIX: DETAILED DESCRIPTION...REFERENCES |
|---|
Literature compilation
Kimberlite occurrences are associated with Archean or Proterozoic cratons throughout the world (Haggerty, 1994
). To first examine the fO2 recorded by perovskite in kimberlites distributed worldwide, we compiled kimberlite whole-rock and perovskite analyses from the literature. The parageneses of the perovskites and detailed information on kimberlite type are in most cases not reported in these studies, although we show later that these become important in interpreting the fO2 of kimberlites during emplacement.
In one case, published data on both zoned and unzoned perovskites from lapilli in the crater facies of the Phoenix kimberlite (Eccles et al., 2004) make it possible to examine the variation of fO2 within a single pipe. The Phoenix kimberlite intrudes Proterozoic basement rocks of the Talston magmatic arc domain (1·78–1·98 Ga) that are overlain by 500 m of Phanerozoic sedimentary rocks of the Western Canada Sedimentary Basin of Alberta, Canada. The northern Alberta kimberlite province is dominated by pyroclastic kimberlites (PK) followed by resedimented volcaniclastic kimberlite (RVK) (Eccles et al., 2004). The Phoenix pipe has an emplacement age of 70 Ma based on Rb–Sr phlogopite and U–Pb perovskite ages (Aravanis, 1999). More recently, Eccles et al. (2003) determined emplacement ages of 77·6±1·1 Ma based on U–Pb analysis of perovskite. The different ages are interpreted to reflect separate eruption events in the same pipe, as several different types of lapilli containing perovskite are found in the PK rocks.
Detailed perovskite parageneses
To examine the fO2 of kimberlites on a continental scale, as well as within and between different pipes, we studied in detail the perovskite parageneses from 11 kimberlites from North America (Table 1). Back-scattered electron (BSE) images of samples were acquired with a Philips XL30 scanning electron microscope (SEM). Electron microprobe (EMP) analyses were carried out with a CAMECA SX50 electron microprobe. Major and minor elements were determined at 15·0 kV acceleration voltage and a beam current of 20·1 nA with a 1 µm beam. Analytical conditions were 20 s counting time on peaks for all major elements except for Fe, Ce and La (60 s), Sr (120 s) and Nb (160 s). Natural and synthetic standards were used for calibration. Standards were albite (Na), olivine (Mg, Si), orthoclase (Al, K), apatite (P), diopside (Ca), rutile (Ti), synthetic rhodonite (Mn) and fayalite (Fe). Drake standard glasses (Drake & Weill, 1972) were used as standards for Ce, La, Sr and Nb in glass and perovskite phases. The ‘PAP’ 
(
Z) method (Pouchou & Pichoir, 1985) was used for data reduction. Compositions of each perovskite are given in Table 2. Whole-rock analyses of major and selected trace elements in the kimberlites (Table 3) were obtained by X-ray fluorescence and LECO induction furnace at McGill University, using methods and analytical details as given by Fedortchouk & Canil (2004).
|
|
|
Somerset Island
On Somerset Island, and the nearby Brodeur peninsula in the Canadian Arctic, kimberlites intrude middle Proterozoic to early Paleozoic sediments overlying crystalline Precambrian basement (Innuitian Province) in the Boothia uplift (Stewart, 1987
). The region is believed to overlie an Archean mantle root (Irvine et al., 1999). The Somerset Island kimberlites are mostly brecciated diatremes and hypabyssal root zones with rare magmatic kimberlite (Mitchell & Meyer, 1980) such as at the Nikos 3 pipe (Schmidberger & Francis, 1999). The Somerset Island kimberlites vary in age from 88 to 105 Ma based on U–Pb dating of perovskite (Heaman, 1989; Smith et al., 1989). Perovskite from six Somerset Island kimberlites occurs as discrete, unzoned crystals or as grains mantled by Fe–Ti oxides (Fig. 1). Detailed descriptions of the kimberlites and their perovskite parageneses are given in the Appendix.
|
Lac de Gras
Kimberlites from the Lac de Gras area intrude a granite–greenstone terrain in the central Archean Slave Province (Bleeker & Davis, 1999
) of northwestern Canada. About 40% of the Lac de Gras kimberlites are of Eocene age (45·2–53·3 Ma), and occur in four distinct episodes (Creaser et al., 2003): (1) the Mark Array (47·8±0·3 Ma); (2) the Panda Array (53·2±0·3 Ma); (3) the A154 Array (55·3±0·3 Ma); (4) the Cobra Array (59·0±0·7 Ma). The Eocene pipes appear to be the only kimberlites of economic significance at Lac de Gras (Heaman et al., 2004). Our samples are from a variety of kimberlite types or facies (resedimented volcaniclastic, tuffisitic and hypabyssal ‘magmatic’).
Perovskites from the Lac de Gras kimberlites are varied and complex, and Chakhmouradian & Mitchell (2001) distinguished three types based on composition, some of which are shown in Fig. 2. Type I perovksites are rare earth element (REE)–Nb–Al poor with mantles having high Sr. Type II perovskites have elevated Al, Fe, Nb and light REE (LREE), are found as discrete crystals or as rims on ilmenite, and represent the overwhelming majority of grains. Type III perovskites are notably enriched in Na, Sr, Nb and LREE, and are found as rims on Type I and II perovskites. The evolutionary trend from Types I to III perovskite is one of increasing Nb (Chakhmouradian & Mitchell, 2001). Type III perovskites are believed to represent the re-equilibration of Type I and II grains at near-solidus conditions (Chakhmouradian & Mitchell, 2001), and for this reason we did not apply perovskite oxybarometry to Type III parageneses. Detailed descriptions of each pipe and its perovskite paragenesis are given in the Appendix.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSAMPLESRESULTSDISCUSSIONSUPPLEMENTARY DATAAPPENDIX: DETAILED DESCRIPTION...REFERENCES |
|---|
Data from the literature
The Fe2O3 content of perovskites from natural kimberlites compiled from the literature varies from 1 to 2 wt%, which corresponds to fO2 conditions of NNO–5 to NNO+1 (Fig. 3a). Nearly the entire range of fO2 embodied in the literature database is recorded by perovskite within different lapilli of the Phoenix pipe alone. Perovskite cores in the crater facies of the Phoenix pipe display an average Fe2O3 content consistently lower than the rims, corresponding to differences in relative fO2 of NNO–2 to NNO in the cores, compared with NNO+1 to NNO+2 in the rims (Fig. 3b and c).
|
NNO) calculated using the perovskite oxygen barometer of Bellis & Canil (2006Somerset Island
Within the Somerset Island cluster, perovskites from six pipes display a range of Fe2O3 content from 0·8 to 2·6 wt% (Table 2), corresponding to a relative fO2 of NNO–4 to NNO+1 (Fig. 4). Within the Nikos 1 pipe, perovskites with Fe–Ti oxide rims record a relative fO2 of NNO–2·0, whereas resorbed grains record slightly more reduced conditions of NNO–3 (Table 2, Fig. 5). In the Zulu pipe, perovskite cores record a relative fO2 of NNO+0·5 to NNO+1· 8, whereas rims have a relative fO2 of NNO–1· 9 to NNO+0·6. Unzoned perovskites record a relative fO2 of NNO–0·9 to NNO+1· 1 (Table 2, Fig. 5).
|
|
Lac de Gras
Within the Lac de Gras cluster, perovskites from five pipes have a wide range of Fe2O3 contents from 1 to almost 14 wt% that are strongly correlated with Nb content (0·3–13 wt% Nb2O5). Type I, II and III perovskites are present only in the Grizzly pipe, the latter type by definition as an alteration on Type I and II grains. The pervoskites in this evolutionary trend record a progressive increase in fO2 (Fig. 5). The remaining Lac de Gras pipes studied contain only Type II grains with a range of Fe2O3 contents from 2 to 5 wt%, representing a range relative fO2 of NNO–4 to NNO+8 (Fig. 6). Different phases (volcaniclastic and hypabyssal) of the Aaron kimberlite record the same fO2 (Fig. 6)
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSAMPLESRESULTSDISCUSSIONSUPPLEMENTARY DATAAPPENDIX: DETAILED DESCRIPTION...REFERENCES |
|---|
The range of fO2 recorded by kimberlite perovskites
The Fe content of perovskite in kimberlites records a range in fO2 of many orders of magnitude (NNO–5 to NNO+6), the largest extreme observed for this variable in any magma type (Fig. 7). Because our fO2 estimates are empirical, based simply on the Fe content of perovskite alone, it might appear that the extreme range of fO2 we observed simply reflects varying Fe contents of kimberlite magmas from which they crystallized. The experiments of Bellis & Canil (2006
) show that fO2, and not the bulk Fe content of the system, is the determining factor for the variation in the Fe2O3 content of perovskites. This assertion is supported by the lack of correlation between the Fe content of perovskite (and thus calculated fO2) and that of the bulk kimberlite (Fig. 8). Furthermore, kimberlites have a restricted range in bulk Fe content (8 wt% FeO*; Fig. 9), such that the range in Fe content of perovskites, and in calculated fO2 in our study, cannot be attributed to varying bulk Fe content of different kimberlites.
|
|
|
We note a systematic positive correlation of the fO2 recorded by perovskites with the Mg/Fe of the bulk-rock for kimberlites having magmatic textures (Fig. 10). Resedimented volcaniclastic rocks lie off this trend because their bulk-rock chemistry does not represent that of a liquid in any way. The bulk Mg and Fe content of the magmatic kimberlites represents that of their Mg-silicates (mainly olivine and monticellite). Because the latter phases prefer Fe2+ over Fe3+, more oxidized melts precipitating and accumulating these minerals will have more of their total Fe as Fe3+, resulting in a higher bulk Mg/Fe.
|
The fO2 estimated from perovskite can be compared with that recorded by olivine–chromite oxybarometry for two kimberlites from this study. Fedortchouk et al. (2005
) estimated a relative fO2 of NNO–2·9±0·2 for chromite included in olivine phenocrysts from the Panda pipe, within uncertainty of our fO2 estimates using perovskite from this pipe (Fig. 6). In the Grizzly pipe, olivine–chromite pairs record from NNO–3 to NNO–2·5, within uncertainty of the one Type I perovskite analysed, but lower than Type II parageneses from the same pipe (Fig. 5).
We consider the consistency in fO2 recorded by olivine + chromite and perovskite, and the variation of fO2 with bulk Mg/Fe of the whole-rocks, as evidence that perovskite oxybarometry gives reasonable results. The calibration for perovskite oxybarometry has a large uncertainty of ±1 log fO2 unit (Bellis & Canil, 2006), but it has the potential advantage that it can be applied to discrete stages in the evolution of kimberlite magma that are recognizable by the different compositional types of perovskite in kimberlite pipes (Figs 1 and 2).
Variability in fO2 within a pipe
Kimberlites experience multiple stages of crystallization under changing conditions during ascent, which might be encoded in the zonation observed in natural perovskites (Chakhmouradian & Mitchell, 2000). The wide range of fO2 recorded by perovskites from the Phoenix kimberlite (Figs 3c, d and 5) is attributed to the fact that these perovskites crystallized from different emplacement episodes or magma batches. The samples from the Phoenix pipe are derived from several different lapilli in the crater facies environment, which are considered to have resulted from mixing of multiple eruptions (Eccles et al., 2004). Mixing of different phenocrysts is common in kimberlite (Mitchell, 1986) and the perovskites have recorded the fO2 from varied and separate eruptions in the development of this pipe. The cores of the Phoenix pipe perovskites record consistently more reduced conditions relative to the rims (Fig. 3), suggesting that this kimberlite magma underwent oxidation during differentiation and emplacement.
Within the Nikos 1 pipe from Somerset Island, different fO2 estimates were recorded by texturally distinct perovskites within a single phase of intrusion. For example, in the hypabyssal sample JP1-101D, two separate generations of perovskite were identified. Perovskite cores with rims of Fe–Ti oxides are most abundant, but isolated resorbed grains also occur. As crystallization proceeded, incompatible Nb should have increased in the melt and concentrated in the perovskite. The low-Nb perovskites with rims of Fe–Ti oxides are thereby interpreted as ‘early’ whereas the resorbed, unzoned perovskites with high Nb are ‘late’ during magma crystallization. The continuous variation in relative fO2 from NNO–2 in early, Fe–Ti oxide-mantled perovskites, to NNO–3 in late-stage, highly resorbed, unzoned perovskites suggests that a decrease in fO2 occurred during emplacement.
Zulu is a tuffisitic kimberlite, and discrete zoned and unzoned perovskites from this pipe also define a trend of reduction during emplacement. Perovskite cores record relative fO2 conditions from NNO+0·5 to NNO+1· 9, whereas rims record NNO–2 to NNO+0·5 (Fig. 5). Unzoned perovskites from this pipe record fO2 conditions intermediate between these extremes (from NNO–1·0 to NNO+1· 0). In the Grizzly pipe, perovskites show an evolutionary trend from reduced Type I varieties (NNO–3·4) to more oxidized Type II grains (NNO–1 to NNO+1) (Fig. 5).
The evolutionary trend of perovskites from our detailed study shows that both oxidation and reduction can ensue during crystallization. There is no consistent relationship between the absolute value or trend in fO2 and the texture present in the host kimberlite, suggesting that the mode of eruption has no role in the fO2 recorded by the perovskite.
Controls on the fO2 of kimberlite magmas
It is well known that more oxidized magmas tend to be more hydrous (Carmichael, 1991), and this also appears to be true for kimberlites, if their H2O contents (6–15 wt%) are deemed to be primary. A summary of recent experiments on the effect of H2O on the Fe3+/Fe2+ of silicate liquids shows little or no effect in basic melts, although none of the melts so far investigated is as ultrabasic as kimberlite (Botcharnikov et al., 2005). Given the current body of experimental data it appears that high H2O contents are not the reason for the high fO2 recorded by perovskite in some of kimberlite magmas.
At constant fO2 and composition, the Fe3+/Fe2+ of silicate liquids increases with decreasing T (Kennedy, 1948). The Fe3+/Fe2+ of liquids also increases with the concentration and the identity of the alkali ion (K>Na>Li) at constant T and fO2 (Paul & Douglas, 1965; Fudali, 1965). Because the experimental calibration using Fe in perovskite as an oxygen barometer is performed at temperatures above those recorded or inferred for kimberlite magmas (Mitchell, 1986; Fedortchouk & Canil, 2004) it is possible that the Fe-rich perovskites observed in some kimberlites are due to the large amounts of Fe3+ favoured in potassic kimberlitic liquids at low temperatures (<1000°C). Our estimates of fO2 based on the Fe content of perovskite, however, are not unlike those of Carmichael & Nicholls (1967), who used heterogeneous equilibria between perovskite and coexisting phases to infer similarly high fO2 values (NNO+1 to NNO+2·5) of kimberlite and madupite liquids. Furthermore, there is a consistent correlation of high fO2 and water contents with high K contents of the magmas (Lange et al., 1993), and the results we observe, as well as those for some shoshonite lavas erupted at Stromboli (Cortes et al., 2006), are the extreme upper ends of this general trend. Given the evidence for highly oxidized magmas in the potassic rock clan, we consider the very high fO2 recorded by perovskites not to be an artifact of the oxybarometer.
Several different datasets constrain the oxidation state of mantle lithosphere or mantle-derived magmas. The Fe3+/Fe2+ of mid-ocean ridge basalt (MORB) glasses record a relative fO2 of NNO–1 (Bezos & Humler, 2005), whereas the lavas of Kilauea range from NNO–1·5 to NNO–0·5 (Carmichael & Ghiorso, 1986; Carmichael, 1991). In western Mexico, basalts and andesites encompass a range in fO2 from NNO–0·5 to NNO+2·5, but lamprophyres, minettes and related alkaline lavas reach extremes in fO2 of NNO+3 to NNO+5 (Carmichael et al., 1996). Thermobarometric estimates of relative fO2 recorded by samples of mantle lithosphere vary from NNO–6 to NNO+1 (Ballhaus, 1993; Woodland & Koch, 2003; McCammon & Kopylova, 2004) a range more limited than that of mantle-derived magmas erupted at the surface (Fig. 7).
A number of different processes can alter the fO2 of melts during their emplacement, including the partial melting process (Holloway, 1998), crystallization (Carmichael & Nicholls, 1967) decompression (Carmichael & Ghiorso, 1986) and volatile degassing (Sato, 1978). Large amounts of crystallization of principally Fe2+-bearing minerals may serve to increase the Fe3+/Fe2+ and fO2 of magmas (Carmichael Nicholls, 1967). Olivine and monticellite crystallize before perovskite in kimberlite magma (Mitchell, 1986), and preferentially partition Fe2+, causing an increase in the Fe3+/
Fe ratio of the residual liquid. Experimental data on the crystallization of kimberlite (Bellis & Canil, 2006) suggest that up to 30% crystallization of olivine and monticellite in kimberlite magma has occurred by the onset of perovskite saturation. Perovskites in some kimberlites, therefore, could record a high Fe3+/Fe ratio present in the melt simply because of their later appearance on the liquidus after considerable crystallization of Fe3+-free olivine or monticellite. Water loss by degassing or decompression during ascent can cause copious amounts of crystallization (Blundy & Cashman, 2001). Kimberlite magmas have substantial water concentrations that affect their crystallization temperatures (Eggler & Wendlandt, 1978; Edgar et al., 1988) and may undergo profound olivine and monticellite crystallization by H2O loss, leaving more Fe3+ to partition into perovskite upon saturation in that mineral.
Calculations to test this hypothesis by simply treating Fe3+ as an incompatible trace element do not create a sufficiently oxidized residual liquid to explain the levels of Fe2O3 (and thus high fO2) recorded in kimberlite perovskites. Perovskites from kimberlites with a relative fO2 of NNO+2 or higher contain more than 2 wt% Fe2O3. The DFe2O3 Pv/liq is as high as 0·15 at NNO+2 (Bellis & Canil, 2006), requiring perovskites with greater than 2 wt% Fe2O3 to have crystallized from liquids with greater than 13 wt% Fe2O3. Assuming Fe3+ is as incompatible as Al in either monticellite or olivine (i.e. DFe2O3 ol, mont/liq 0·01), and Rayleigh fractional crystallization, a parent kimberlite magma with 8 wt% FeO* and an Fe3+/Fe value of 0·25 requires more than 85% fractional crystallization to attain this level of Fe2O3 in the liquid. More reasonable amounts of crystallization (50%) are attained for parent liquids that have an Fe3+/Fe ratio of >0·5. Thus, early crystallization of olivine and monticellite raises the Fe3+/Fe in derivative liquids, but in the case of kimberlites, the parent liquids need to have already been very oxidized (Fe3+/Fe>0·5) to explain the levels of Fe2O3 found in their perovskites by this mechanism.
Whether volatile loss or dissociation from magmas can affect their redox state by reactions such as
|
| (2) |
; Mueller, 1971; Sato, 1978; Carmichael & Ghiorso, 1986). This mechanism, when explored quantitatively by calculation, is an inefficient process to oxidize silicate liquids, and can proceed only if there remains a large volatile component relative to Fe, to exceed the buffering effect of the Fe3+/Fe2+ ratio in most terrestrial magmas (Candela, 1986; Ballhaus & Frost, 1993). Such a condition might be realized in kimberlites, however, given their high H2O and CO2 contents. It remains uncertain if the H2O contents of most kimberlites are primary, for the H2O content of the whole-rock may simply reflect the introduction of fluid to serpentinize olivine and hydrate other anhydrous minerals during emplacement. We restrict our analysis to the fresh kimberlites from this study for which both perovskite analyses and H2O and CO2 contents are known (Table 3). One way to view these data is that the higher amounts of H2O preserved in the whole-rock (now as hydrous minerals) suggest that minimal H2O was lost by degassing (loss) of H2 and thereby reduced according to equation (2).
Primitive kimberlite magma with up to 15 wt% CO2 can be produced by less than 1% melting of a CO2-bearing peridotite (Gudfinnson & Presnall, 2005). Analyses of kimberlites that are believed to approximate primary kimberlite magma (i.e. Jericho, Canada; Wesselton, South Africa) contain between 10 and 17 wt% ‘primary’ CO2 (Price et al., 2000). During emplacement, kimberlites could elevate their CO2 and H2O contents even more by crystallization of volatile-free minerals.
The CO2 contents of whole-rock kimberlites from this study strongly correlate with both CaO and Sr, and are considered to be primary CO2 bound in carbonate globules that segregated as liquids from the (silicate) magma during emplacement (Dawson & Hawthorne, 1973; Clarke & Mitchell, 1975; Clement, 1975). When these segregations are extracted from the conjugate (silicate) kimberlite magma, the latter component, which eventually crystallizes perovskite, could undergo CO2 loss to reduce the silicate magma:
|
| (3) |
).
All the kimberlites from this study have a pressure of origin of at least 5 or 6 GPa, as confirmed by the presence of diamonds and the P–T arrays of the mantle xenoliths they entrain (Kjaarsgaard, 1992; Boyd & Canil, 1997; MacKenzie & Canil, 1999; Schmidberger & Francis, 1999). Experimental work shows that perovskite is stable in the crystallization interval of kimberlite to at least 3 GPa (Edgar et al., 1988). If crystallization ensues during ascent from this pressure, there is ample opportunity for perovskite grains from successive magma batches to have mixed while forming a pipe. Such grains may record similar or varying fO2 depending on the conditions the magma batches experienced during decompression and ascent, explaining the heterogeneity in the fO2 recorded by different grains or generations of perovskite within the same kimberlite pipe (Figs 4 and 6). Thus, perovskite fO2 is mainly controlled by the redox state of the magma somewhere between its deep source in the mantle and its eruption at the surface; we have no evidence where along this path the perovskite crystallized. Nonetheless, the redox state of a magma along this path can be estimated by examining the interplay of two competing internal buffers: carbon–fluid and ferric–ferrous equilibria (Ballhaus, 1993).
The effect of the ascent path on buffering equilibria in the mantle has been explored quantitatively by Ballhaus & Frost (1994). Adiabatically rising magma in equilibrium with a peridotite mantle buffered by ferric–ferrous equilibria will oxidize 0·3–0·6 log fO2 units/GPa during decompression, depending on whether garnet or spinel is stable. The oxidation state of the magma upon arrival at the surface would be dictated by the initial fO2 at its source, its Fe3+ content, and the distribution of Fe3+ amongst mantle phases (Fig. 11). In contrast, a mantle source buffered by the presence of carbon (CCO-carbon–carbon dioxide) would undergo reduction upon decompression, as a result of the opposite pressure dependence of carbon–fluid equilibria (Ballhaus & Frost, 1994). In this way, the oxidation state of a magma, and whether it oxidized or reduced on ascent, will depend on its depth of origin, degree of melting and the depth at which it separates from a carbon-bearing source region (Ballhaus, 1993).
|
In a general way these model calculations can explain some of the wide variations and trends observed in fO2 of the kimberlites in our dataset. For example, the reduced Lac de Gras and Somerset Island pipes could represent magmas buffered by CCO during rise from a source region at or near the fayalite–magnetite– quartz (FMQ) buffer (path A in Fig. 11). If such magmas were not buffered by CCO, but instead by ferric–ferrous equilibria, their sources would have to be impoverished in ferric iron and significantly reduced (e.g. FMQ – 4; path B in Fig. 11). Conversely, the oxidized extreme of kimberlites, recorded exclusively by the Lac de Gras pipes, requires a mantle source rich in ferric iron at or several log units above FMQ (path C in Fig. 10). The source of these oxidized kimberlites could also be at the normal mantle value of
FMQ , but would then require a pressure of origin of at least 10 GPa.
That the fO2 recorded by magmas is a reflection of their source region has long been postulated for many other mantle-derived magmas (Carmichael & Ghiorso, 1986; Carmichael, 1991). If so, the exceptionally high fO2 recorded by the kimberlites from this study would require a source region more oxidized than currently observed in mantle rocks (Carmichael, 1991). Similar high oxidation states are also encompassed by the potassic minette lavas of Mascota, western Mexico. To explain these high fO2 conditions, Carmichael et al. (1996) suggested partial melting of phlogopite (as veins) in the mantle source of the minettes. Mantle phlogopite has both a high Fe3+/Fe2+ and high Mg-number, and during partial melting as veins in lherzolite would produce low volumes of oxidized magma with a high Fe3+/Fe2+ and Mg-number. As partial melting of a phlogopite-veined lherzolite proceeds, liquids produced will have increasing Mg-number and Fe3+/Fe2+ as the proportion of phlogopite increases in the partial melt (Carmichael et al., 1996). The case for the Mascota minettes is supported by major and trace element data for that rock series that lie along trends projecting to phlogopite. No such trends are evident for fresh Lac de Gras samples or any other kimberlite rock series, but may well be obscured by the mixing of different and varied xenocrystic components in kimberlites (compared with the glassy Mascota series lavas). Given this limitation, we cannot quantitatively confirm whether phlogopite in the source region produces the high fO2 values recorded by perovskites in kimberlites.
Either crystallization of early Fe2+-bearing phases or an oxidized mantle source for kimberlite-rich phlogopite veins can serve to explain the extremely high fO2 recorded by kimberlite perovskites. A mechanism involving buffering by carbon–fluid equilibria in C-rich melts such as kimberlite, which separated from their mantle source, or a very reduced mantle source low in Fe3+, can account for their range to low fO2. Detailed examination of perovskite generations in individual pipes leads us to believe that all these processes can be operative during emplacement to exert a control on the resultant fO2 recorded by perovskite. Given this potential to either oxidize or reduce on ascent, we might predict that some kimberlite magmas are more ‘diamond-friendly’ than others.
Kimberlite fO2 and diamonds
The stability and dissolution of carbon-bearing minerals in kimberlite magmas are directly influenced by the fO2 of the magma during its ascent according to reaction (3). With increasing fO2, diamond in the kimberlite magma should be oxidized to CO2. If a kimberlite magma resides at too high an fO2 for too long during its ascent, it is possible that diamonds may be resorbed and the kimberlite would decrease in grade and/or quality.
To evaluate this possibility we compare the fO2 estimates determined in this study with diamond grade and quality assessed from thousands of stones in selected kimberlites from the Lac de Gras area (Fedortchouk et al., 2005). Although diamond grade is probably a result of the quantity of diamond sampled in the mantle by ascending kimberlite magma, estimates of kimberlite fO2 may have some utility in predicting the quality or properties of these diamonds. Figure 12 shows that the kimberlite pipe with the highest fO2 (i.e. Misery) has the highest proportion of highly resorbed diamonds (67%), whereas pipes with lower estimated fO2 (i.e. Panda and Grizzly) have lower proportions of highly resorbed diamonds (14% at Panda and 50% at Grizzly) (Fedortchouk et al., 2005). The kimberlite pipe with the lowest estimated fO2 (i.e. Panda), has higher proportions of lesser-resorbed diamonds (54%), whereas pipes with higher estimated fO2 (i.e. Grizzly and Misery), have lower proportions of lesser-resorbed diamonds (20% at Grizzly and 10% at Misery). These results suggest that fO2 affects the properties and textures of diamonds preserved in the magma, as predicted according to reaction (3). In this way, perovskite oxybarometers as applied in this study may have the potential to aid in predicting the diamond quality in a given pipe.
|
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONSAMPLESRESULTSDISCUSSIONSUPPLEMENTARY DATAAPPENDIX: DETAILED DESCRIPTION...REFERENCES |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
APPENDIX: DETAILED DESCRIPTION OF KIMBERLITES AND PEROVSKITE TYPES |
|---|
|
TOPABSTRACTINTRODUCTIONSAMPLESRESULTSDISCUSSIONSUPPLEMENTARY DATAAPPENDIX: DETAILED DESCRIPTION...REFERENCES |
|---|
Somerset Island
Batty Complex
The Batty Complex is a dark green hypabyssal kimberlite. Perovskites are 20–100 µm, subhedral to euhedral and unzoned (Fig. 1a).
Elwin Bay
The Elwin Bay pipe is transitional between hypabyssal and tuffistic kimberlite, containing altered olivines and pelletal lapilli. Perovkites are 50–100 µm, euhedral and unzoned (Fig. 1b).
Zulu
Zulu is a brown, highly altered tuffistic kimberlite with altered olivine and pelletal lapilli. There are at least two generations of perovskite: one zoned with 30–35 µm, subhedral to euhedral, cores mantled by 10–15 µm rims (Fig. 1c), and a second as 20–50 µm, subhedral to euhedral, unzoned perovskites.
K4
The K4 pipe is a dark green hypabyssal kimberlite. Perovskites are 30–80 µm, subhedral to euhedral and resorbed (Fig. 1d); some display mottled compositional zoning.
Nikos 1
JP1 is a tuffistic kimberlite breccia, containing 20–100 µm, subhedral perovskites with thin rims (< 1 µm) of Fe–Ti oxide. Perovskites also occur as mantles on Fe–Ti oxides that mantle perovskite. The inverse texture also occurs, with Fe–Ti oxides mantling Pv, all of which are mantled by Fe–Ti oxides. JP1-101D is a dark green hypabyssal kimberlite, containing two separate generations of perovskite: 50 µm, anhedral, perovskite cores with thick rims of Fe–Ti oxide are most abundant (Fig. 1e), with less common 20–40 µm, subhedral to euhedral, resorbed perovskites (Fig. 1f).
Nikos 3
JP3 is a dark green hypabyssal kimberlite, containing 50 µm discrete, euhedral perovskites that are unzoned, some of which are resorbed. A second generation of perovskites occur as 30 µm mantles on anhedral Fe–Ti–Mg cores.
Lac de Gras
Torrie
The Torrie pipe is composed of kimberlite tuff and tuff breccia rich in organic material. Perovskites occur as discrete, compositionally zoned grains (Fig. 2a).
Misery
The Misery pipe is mostly volcaniclastic kimberlite associated with dykes of macrocrystic hypabyssal kimberlite, and containing two generations of perovskite similar in habit: one generation with relatively low Nb, 10–20 µm, subhedral and unzoned, and a second generation with more elevated Nb (Fig. 2b).
Grizzly
The Grizzly pipe is a dark green, hypabyssal kimberlite, containing three generations of perovskite. A generation of subhedral to euhedral cores with relatively low Nb content (Type I), are overgrown with Type III rims (high Nb) (Fig. 2c). Compositionally zoned perovskites are 10–30 µm, subhedral to euhedral, and are typically low in Nb (Type II) (Fig. 2d). Unzoned perovskites elevated in Nb (Type III) were also recognized.
Panda
The Panda pipe is mainly a resedimented volcaniclastic kimberlite, containing 10–30 µm, unzoned, euhedral, perovskites (Fig. 2e).
Aaron
The Aaron pipe is composed of grey, hypabyssal kimberlite, with little or no alteration (sample AN2) as well as friable, green pelletal volcaniclastic kimberlite, with serpentine alteration and abundant wall-rock xenoliths (samples AN4 and AN5). Perovskites from the hypabyssal kimberlite are 50–100 µm, angular and in some instances resorbed. Perovskites from the volcaniclastic kimberlite are 20–100 µm, euhedral and have compositional zoning (Fig. 2f).
|
ACKNOWLEDGEMENTS |
|---|
We thank J. Carlson and BHP Billiton for access to the Lac de Gras drill cores and permission to publish. M. Raudsepp assisted with the EMP work at UBC, R. Eccles provided unpublished data from the Phoenix kimberlite, and J. Pell generously donated the Somerset Island samples. We are especially grateful to C. Ballhaus, B. R. Frost, P. Jugo and C. McCammon for their thorough reviews. This research was supported by a NSERC Discovery Grant to D.C.
*Corresponding author. E-mail: dcanil{at}uvic.ca
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSAMPLESRESULTSDISCUSSIONSUPPLEMENTARY DATAAPPENDIX: DETAILED DESCRIPTION...REFERENCES |
|---|
Apter DB, Harper FJ, Wyatt BA, Scott Smith BH. (1984) The geology of the Mayeng kimberlite sill complex, South Africa. Proceedings of the 3rd International Kimberlite Conference(ElsevierIn Kornprobst J (Ed.). , Amsterdam) pp. 43–57.
Aravanis T. (1999) Legend property assessment report, Birch Mountains area, Alberta. Alberta Energy and Utilities Board, Alberta Geological Survey, Assessment File. 2003 pp. 23.
Ballhaus C. (1993) Redox states of lithospheric and asthenospheric upper mantle. Contributions to Mineralogy and Petrology 114:331–348.[CrossRef][ISI]
Ballhaus C and Frost BR. (1993) Comment on ‘Constraints on the origin of the oxidation state of mantle overlying subduction zones: An example from Simcoe, Washington, USA’ by A. D. Brandon and D. S. Draper. Geochimica et Cosmochimica Acta 62:329–331.
Ballhaus C and Frost BR. (1994) The generation of oxidized CO2-bearing basaltic melts from reduced CH4-bearing upper mantle sources. Geochimica et Cosmochimica Acta 58:4931–4940.[CrossRef][ISI]
Bellis AJ and Canil D. (2006) Ferric iron in CaTiO3 perovskite as an oxygen barometer for kimberlitic magmas I: experimental calibration. Journal of Petrology (in press).
Berg GW and Carlson JA. (1998) The Leslie kimberlite pipe of Lac de Gras, Northwest Territories, Canada: evidence of near surface hypabyssal emplacement. In Gurney J (Ed.). Extended Abstracts, 7th International Kimberlite Conference(University of Cape Town, Cape Town) pp. 81–83.
Bezos A and Humler E. (2005) The Fe3+/Fe ratios of MORB glasses and their implications for mantle melting. Geochimica et Cosmochimica Acta 69:711–725.[CrossRef][ISI]
Bleeker W and Davis WJ. (1999) The 1991–1996 NATMAP Slave Province Project: Introduction. Canadian Journal of Earth Science 36:1033–1042.[CrossRef]
Blundy J and Cashman K. (2001) Ascent-driven crystallisation of dacite magmas at Mount St Helens, 1980–1986. Contributions to Mineralogy and Petrology 140:631–650.[ISI]
Botcharnikov RE, Koepke J, Holtz F, McCammon C, Wilke M. (2005) The effect of water activity on the oxidation and structural state of Fe in a ferro-basaltic melt. Geochimica et Cosmochimica Acta 69:5071–5085.[CrossRef][ISI]
Boyd FR and Canil D. (1997) Peridotite xenoliths from the Slave craton, NWT. Lunar and Planetary Science Contribution 921:34.
Candela PA. (1986) The evolution of aqeous vapor from silicate melts: effect on oxygen fugacity. Geochimica et Cosmochimica Acta 50:1205–1211.[CrossRef][ISI]
Carmichael ISE. (1991) The redox states of basic and silicic magmas: a reflection of their source regions? Contributions to Mineralogy and Petrology 106:129–141.[CrossRef][ISI]
Carmichael ISE and Ghiorso MS. (1986) Oxidation–reduction relations in basic magma: a case for homogeneous equilibria. Earth and Planetary Science Letters 78:200–210.[CrossRef][ISI]
Carmichael ISE and Nicholls J. (1967) Iron–titanium oxides and oxygen fugacities in volcanic rocks. Journal of Geophysical Research 72:465–487.
Carmichael ISE, Lange RA, Luhr JF. (1996) Quaternary minettes and associated volcanic rocks of Mascota, western Mexico: a consequence of plate extension above a subduction modifies mantle wedge. Contributions to Mineralogy and Petrology 124:302–333.[CrossRef][ISI]
Chakhmouradian AR and Mitchell RH. (2000) Occurrence, alteration patterns and compositional variation of perovskite in kimberlites. Canadian Mineralogist 38:975–994.
Chakhmouradian AR and Mitchell RH. (2001) Three compositional varieties of perovskite from kimberlites of the Lac de Gras field (Northwest Territories, Canada). Mineralogical Magazine 65:1133–148.
Christie DM, Carmichael ISE, Langmuir CH. (1986) Oxidation states of mid-ocean ridge basalt glasses. Earth and Planetary Science Letters 79:397–411.[CrossRef][ISI]
Clarke DB and Mitchell RH. (1975) Mineralogy and petrology of the kimberlite from Somerset Island, N.W.T, Canada. Physics and Chemistry of the Earth 9:123–136.[CrossRef]
Clement CR. (1975) The emplacement of some diatreme facies kimberlites. Physics and Chemistry of the Earth 9:51–59.[CrossRef]
Cortes JA, Wilson M, Clondiffe E, Francalacci L. (2006) The occurrence of forsterite and highly oxidizing conditions in basaltic lavas from Stromboli volcano, Italy. Journal of Petrology 47:1345–1373.
Creaser RA, Grutter H, Carlson J, Crawford B. (2003) Macrocrystal phlogopite Rb–Sr dates for the Ekati property kimberlites, Slave Province, Canada: evidence for multiple intrusive episodes in the Paleocene and Eocene. Lithos 76:399–414.[CrossRef]
Dawson JB. (1967) Geochemistry and origin of kimberlite. In Wyllie PJ (Ed.). Ultramafic and Related Rocks(John Wiley, New York) pp. 241–251.
Dawson JB. (1972) Kimberlites and their relationship to the upper mantle. Royal Society of London Philosophical Transactions, Series A 271:297–311.