Contrasting Garnet Lherzolite Xenolith Suites from the Letšeng Kimberlite Pipes: Inferences for the Northern Lesotho Geotherm

Abstract

The Main and Satellite Pipe kimberlites at Letseng-la-Terae, Lesotho, both contain abundant crustal and mantle xenoliths. In garnet lherzolite xenoliths that are the subject of this paper, textures range from coarse to highly deformed in samples from both pipes, but the most extreme deformation is seen only in Main Pipe samples. Development of reaction coronas around the garnets in all samples indicates retrogression to spinel peridotite facies conditions prior to entrainment in the host kimberlites. Bulk rock compositions of most samples are depleted relative to pristine mantle compositions. Most Main Pipe samples equilibrated at 950–1,000 °C and pressures of 40–45 Kbar, with others extending the temperature range to 820 °C and 1,350 °C. In the temperature range 950–1,000 °C, there is no overlap with Satellite Pipe xenoliths which equilibrated at <800 and >1,305 °C. High-T samples from both pipes lie on the high-T side of the Cretaceous Kalahari geotherm of Rudnick and Nyblade (1999), indicating a perturbed geotherm, similar to that deduced from Thaba Putsoa xenoliths but unlike the non-perturbed geotherm inferred from mantle xenoliths in other Cretaceous Lesotho kimberlites. Hence, xenolith suites from individual kimberlite intrusions each have distinctive thermal characteristics, and there is no consistently perturbed Cretaceous geothermal gradient beneath Lesotho.

Introduction

Kimberlite pipes and dykes intrude the Karoo (Carboniferous–Triassic) sandstones and basalts in the Maluti Mountains of northern Lesotho, to the east of the major Cretaceous kimberlite province of South Africa (Nixon 1973). Zircon from the Mothae pipe has been dated at 87.1 Ma (Davis 1977), and perovskite from other pipes (Kao, Liqhobong, Lipelaneng, Mothae, Ngopetseu) has yielded ages ranging from 88.8 ± 1.4 to 94.7 ± 1.1 Ma (1 sigma errors—Hoese 2009).

The Main and Satellite kimberlite pipes at Letšeng-la-Terae (29°00′S 28°43′E) (for brevity hereafter referred to as Letšeng) contain abundant crustal and mantle xenoliths and afford the opportunity to compare two peridotite xenolith suites which can reasonably be assumed to have been transported up the same or closely adjacent upper mantle conduit(s). The pipes, which are only ~350 m apart (Fig. 1), crop out at ~3,100 m above sea level, forming the world’s highest diamond mine. Features of the bulk diamond production are the overall low diamond grade but the occasional occurrence of exceptionally large, high-quality (type IIA) diamonds. Harris et al. (1979) and Bowen et al. (2009) report significant differences in diamond morphology and colour, but not of size–frequency distribution, between the diamond populations from the two pipes. An objective of this study is to compare the Letseng xenolith suite with that from Thaba Putsoa, Lesotho, from which a perturbed geothermal gradient has been inferred with implications for major frictional heating at the base of the Kalahari craton (Boyd 1973a).

Fig. 1

The Letšeng kimberlite pipes, showing the distribution of kimberlite types in the Main Pipe at the 55 m level (Lock 1980) and the Satellite Pipe at the 70 m level (Palmer et al. 2008)

General Geology

The two pipes (Fig. 1) are separated by basalt wall rock that forms abundant xenoliths in both pipes. In the Lesotho province, kimberlites intruding the Drakensberg basalts have a maximum stratigraphic age of ca 180 Ma (Duncan and Marsh 2006), but the Letseng pipes have not been dated radiometrically. Other xenoliths are of Karoo sediments; gneisses and granulites that have their analogues in the basement of southern Africa; and upper mantle peridotites (s.l).

The Main Pipe (surface area 17.2 ha) comprises four petrographically distinct types of kimberlite, the contacts between which are gradational over a few metres (Lock, 1980). The kimberlites are broadly volcanoclastic (terminology of Scott Smith et al. 2008), but with distinctive characteristics, and the pipe is mainly infilled by autolithic kimberlite. This is penetrated by a smaller body of highly garnetiferous kimberlite which is zoned, with increasing abundance and/or size of peridotite xenoliths and garnet megacrysts towards its centre. It is cut by three dykes, which are interpreted as the hypabyssal facies of the garnetiferous kimberlite. Penetrating the garnetiferous kimberlite are two smaller bodies of highly micaceous kimberlite, the younger of which contains fresh olivines. The intrusive relationships between the different kimberlite types suggest a younging towards the centre of the diatreme.

The Satellite Pipe (surface area 5.2 ha) mainly contains kimberlite that is similar to the Main Pipe autolithic kimberlite, though Palmer et al. (2008) recognise five texturally different varieties and, in the south-west part of the pipe, reworked kimberlite that surrounds a major basalt raft. Sedimentary xenoliths are more abundant than in the Main Pipe, whereas peridotite xenoliths are much rarer. An early dyke of micaceous kimberlite is truncated by the pipe. The emplacement of the two pipes is assumed to be pene-contemporaneous, but there are no dates on either intrusion.

Ultramafic Xenoliths

The xenoliths reported here were collected in situ underground from the Main Pipe and from the surface diggings on the Satellite Pipe. However, kimberlite from both pipes is processed at the plant and the xenoliths piled in a common dump; so, when collected from the dump, the source of xenoliths is unclear. In the case of samples studied by Simon et al. (2003), neither texture nor location (?Main or ?Satellite Pipe) is reported so they are here referred to as being from “Pipe not known”.

The most common ultramafic xenoliths are garnet or chromite lherzolites and harzburgites, but rarer types are lherzolites containing both garnet and Cr-spinel, garnet dunites, garnet websterites, garnet pyroxenites, Al-spinel lherzolites (some of which also contain garnet) and barren harzburgites (i.e. containing low-Ca–Al enstatites—Hervig et al. 1980). Primary phlogopite is present in a few lherzolites, and secondary phlogopite, generally adjacent to garnet, is present in samples that also show incipient serpentinisation. Most rock types are found in both the Main and Satellite pipes, but garnet lherzolites are approximately three times more abundant in the Main Pipe. Barren harzburgites, in which the orthopyroxenes contain very low amounts of CaO and Al₂O₃ (Hervig et al. 1980), have been recorded only from the Main Pipe.

Petrography

Garnet lherzolites are the focus of this contribution. Measured modes and maximum grain sizes in coarse specimens are as follows: olivine 54–78 modal  %, up to 6 mm; enstatite 11–41 %, 8 mm; garnet 3–14 %, 5 mm (including the reaction coronas); and green diopside 0.3–10.3 %, 4 mm. Diopside is often spatially close to garnet. A small number of samples also contain rare chrome-rich spinel or Al-rich spinel (up to 0.3 mm). Grain sizes in deformed rocks are less than in coarse samples.

Due to varying deformation and recrystallisation, the peridotites exhibit a range of textures from coarse texture (not recrystallised), through increasingly recrystallised porphyroclastic and mosaic-structured to laminated and disrupted (LAD) textures (terminology of Harte 1977). In the LAD rocks, garnet and its surrounding reaction coronas are disrupted and strung out into fine-grained chains, interspersed, with laminae of fine-grained olivine and pyroxene neoblasts. Rocks showing recrystallisation are collectively referred to hereafter as “deformed”. A range of textures are seen in samples from both pipes (Table 1); however, the Main Pipe suite is dominated by coarse-textured rocks and the most extreme LAD deformation has not been recognised in Satellite Pipe xenoliths.

Table 1 Textures and equilibration conditions of Letšeng garnet lherzolites

Whole Rock Chemistry

Due to small xenolith size in some cases, not all samples listed in Table 1 were analysed. Compositions of the analysed samples are given in Table 2, with comparable analyses of other pertinent garnet lherzolites and a reference analysis for primitive mantle. With one high-Fe exception, the Letšeng rocks, whether coarse or deformed, are depleted in TiO₂, Al₂O₃, FeO and CaO relative to the primitive mantle composition. With the exception of the high-Fe sample, the Mg# values (a depletion indicator) of the Letseng lherzolites have a narrow range (91.0–93.2) with a slight tendency for lower values in deformed samples (Fig. 2). Coarse “sterile” (i.e. depleted) and sheared “fertile” (i.e. undepleted) garnet lherzolites from Thaba Putsoa show greater differences. Cr/(Cr + Al) ratios (another indicator of depletion) show a range from 0.149 to 0.367, with an overlap between coarse and deformed samples; however, there is no overall correlation between the two depletion parameters. Also, in the Letseng samples, there is no correlation between degree of depletion and deformation, unlike Thaba Putsoa where high deformation is linked to high “fertility”. The Fe-rich sample 3 is also relatively rich in Ti and Ca, and although differing in containing little Na₂O, in both its Mg# and Cr/(Cr + Al) ratio, it strongly resembles Thaba Putsoa fertile lherzolite which has been interpreted as having been metasomatised during the kimberlite event. The bulk compositions of most Letšeng lherzolites are similar to many other Kalahari craton lherzolite suites in Cretaceous kimberlites and also show little compositional difference, in terms of SiO₂, FeO, Al₂0₃ and CaO, to off-craton peridotites (Janney et al. 2010). However, they are more refractory than garnet lherzolites in the Proterozoic Premier kimberlite (Table 2), the undepleted composition of which is attributed to sampling before the early Mesozoic major melting/depletion event that produced the Mesozoic Karoo volcanics (Danchin 1979).

Table 2 Whole-rock analyses of Letseng lherzolites^a (numbers from Table 1) and other reference samples

Fig. 2

Bulk rock depletion parameters for Letšeng garnet lherzolites and other mantle compositions. Data from Table 2

Phase Chemistry

Primary Phases

Analyses of the phases in twenty-four garnet lherzolites were carried out by WDS on electron microprobes at the universities of Edinburgh and Cambridge. At Edinburgh, where most analyses were performed, the probe was a Cameca SX50, and standards used were the following: for Si and Ca—wollastonite; Ti—rutile; Al—corundum; Fe, Mn—metals; Mg—periclase; Na—jadeite; K—orthoclase; Na and K were analysed early in the routine to avoid migration/volatility effects. Counting times were 30 s on peaks and 15 s on backgrounds. Analyses were performed with a spot beam of ~2 μm at 20 kV and a probe current of 20 nA. Data were reduced using the PAP routine procedure (Pouchou and Pichoir 1991). Our data are supplemented by data on seven additional lherzolites from Boyd (1973b), Hervig et al. (1980 ) and Simon et al. (2003) (Table 1).

Olivines (Table 3) lie mainly in the compositional range Mg#90–94 with two exceptions, one from each pipe, that are more Fe-rich (Mg# < 90). There are no significant differences in the olivine compositions in coarse and deformed xenoliths, and there is a reasonable correlation between their Mg# and the Mg# of the coexisting pyroxenes and garnet.

Table 3 Compositions of olivines in Letšeng garnet lherzolites

Orthopyroxenes (Table 4) show a narrow range of Mg# (91.6–93.7) with the exception of the Fe-rich samples (Mg# 89.8 and 90.7). There is a range in CaO (0.20–1.59 wt%) and, with the exception of the Fe-rich specimens, there is a tendency towards slightly higher Ca and Fe but lower Mg in enstatites from deformed xenoliths in both pipes. All enstatites contain appreciable Al₂O₃ (0.69–1.29 wt %), and those with Al₂O₃ > 1.2 wt % are consistently from deformed samples and coexist with subcalcic diopsides (Ca/(Ca + Mg) < 0.37).

Table 4 Compositions of orthopyroxenes in Letšeng garnet lherzolites

Clinopyroxenes (Table 5) range from ureyitic and chrome diopsides to diopsides and subcalcic diopsides (terminology of Stephens and Dawson 1977), with a wide range in Cr₂O₃ (0.75–3.66 wt %). The wide range of Ca/(Ca + Mg) (0.338–0.487) indicates a range of equilibrium temperatures (Davis and Boyd 1966). Subcalcic diopsides (Ca/[Ca + Mg] < 0.40) are present in deformed rocks from both pipes. Intermediate values (0.40–0.46) are common in the clinopyroxenes in Main Pipe samples, but not in those from the Satellite Pipe which have a bimodal grouping of Ca/(Ca + Mg). In the Main Pipe samples, there is no overall correlation between the Ca/(Ca + Mg) ratio and texture, but in coarse samples from both pipes, clinopyroxenes have relatively high Ca/(Ca + Mg) ratios and those from highly deformed samples show relatively low values, with implications for equilibration temperatures.

Table 5 Compositions of clinopyroxenes in Letšeng garnet lherzolites

Garnet compositions (Table 6) show a considerable range in both Mg# (range 78.0–86.5) and Cr content (Cr₂O₃ = 2.77–11.1 wt %); Cr₂O₃ varies directly with CaO and inversely with Al₂O₃. There are no systematic chemical differences between garnets from coarse and deformed rocks in the Main Pipe samples, but those in coarse samples from the Satellite Pipe are more iron-rich than in the deformed rocks and are most similar to the single Fe-rich coarse sample from the Main Pipe. Most, including those co-existing with relatively Fe-rich olivines and pyroxenes, are chrome pyropes (Group 9 of Dawson and Stephens 1975), but one high Ca–Cr garnet in sample 8 is a knorringitic uvarovite pyrope. None is compositionally similar to low-Ca, high-Cr inclusions in diamond.

Table 6 Compositions of garnets in Letšeng garnet lherzolites

Spinel A small number of the garnet lherzolites contain magnesian chromite in addition to garnet. Compositionally, in having higher Cr but lower Mg, they are unlike the spinels occurring in both granuloblastic spinel lherzolites and spinel–garnet lherzolites from Letseng (Lock 1980).

Neoblast compositions. Compared with primary grains, both olivine and enstatite neoblasts show no changes in Mg# but consistent, small increases in CaO, for example 0.08 wt % CaO in an olivine neoblast compared with 0.04 wt % in primary olivine. Although this Ca increase might appear small, it is relevant to the whole rock Ca budget due to the modal dominance of olivine in the neoblast population. In the case of deformed clinopyroxenes, the neoblasts contain less CaO, resulting in a slight decrease in the Ca/(Ca + Mg) ratio. Overall lower Ca in the clinopyroxene neoblasts coincides with a concomitant increase in the olivine and enstatite neoblasts. Based on the enstatite solubility in diopside criterion, this can be interpreted as reflecting slightly increased temperatures during recrystallisation.

Conditions of Equilibration of the Letseng Garnet Lherzolites: Inferences for the Lesotho Geotherm

Garnet lherzolites permit calculation of their temperatures and pressures of equilibration, as determined in experimental systems. We have used the two-pyroxene thermometer of Taylor (1998) combined with the Al-in-enstatite in equilibrium with garnet (Nickel and Green 1985) for pressure estimates, as recommended by Nimis and Grütter (2010) in a rigorous review of the various methods of PT calculation. The same methods have also been used to recalculate, for comparison, the data of Boyd (1973a) and Nixon and Boyd (1973), used in their much-discussed model for a perturbed Cretaceous geothermal gradient below northern Lesotho.

The equilibration PTs of the primary assemblages in the Letšeng garnet lherzolites are given in Table 1 and plotted on Fig. 3. Also shown is the Kalahari craton geotherm of Rudnick and Nyblade (1999), which is based on both surface heat-flow and xenolith equilibrium data calculated by the methods of Brey and Köhler (1990); Brey et al. (1990) which, for a given pressure (depth), in most cases give similar temperatures to those obtained by the method used in this paper. Also shown is the graphite–diamond transition of Bundy (1980), preferred to that of Kennedy and Kennedy (1976) because a diamond-bearing garnet lherzolite xenolith from the Mothae pipe, 7 km north-west of Letseng (PT conditions 43.0 kb and 1035 °C—Dawson and Smith 1975), plots closer to the diamond stability field of Bundy (1980) although still plotting within the graphite stability field.

Fig. 3

PT equilibration of Letseng garnet lherzolites (data from Table 1). Other data sources: Kalahari geotherm—Rudnick and Nyblade (1999); graphite–diamond transition—Bundy (1980); Matsoku diamondiferous lherzolite—Dawson and Smith (1975). The solid and broken lines in red are the trends for Main Pipe coarse and deformed xenoliths, respectively

From Fig. 3, several facts are apparent:

• 1. Xenoliths with coarse textures have apparently equilibrated under relatively low PT conditions, though low-temperature estimates (<900 °C) should be treated with caution (Nimis and Grutter 2010).

• 2. The highest PT conditions are found for deformed xenoliths from both pipes; in this respect, they are similar to high-T, sheared “fertile” xenoliths in the Thaba Putsoa kimberlite (Boyd 1973a). However, a major difference, as noted earlier, is that the Letšeng deformed xenoliths are depleted.

• 3. The Main Pipe xenoliths (with one exception) have equilibrated along a relatively continuous linear trend, whereas the PT conditions for Satellite Pipe xenoliths are bimodal—one group being low PT, the other at much higher PT.

• 4. In the Main Pipe xenolith suite, at around 1,000 °C and 40 kb, there is overlap between coarse and deformed xenoliths (including those with the most extreme LAD textures), in this respect being like garnet lherzolite xenoliths from the Kimberley area (Dawson et al. 1975). The most highly deformed xenoliths (i.e. with LAD textures) did not equilibrate at the highest temperatures (see data in Table 1), so there is no overall correlation between the temperature of equilibration and degree of deformation as found at Thaba Putsoa.

• 5. Many Main Pipe xenoliths, coarse and deformed, derive from an intermediate PT regime apparently not sampled by the Satellite Pipe kimberlite. (The “Pipe not known” samples of Simon et al. (2003) also fall within this PT area).

• 6. For a given temperature, the Satellite Pipe xenoliths have equilibrated at lower pressures than the Main Pipe xenoliths, thus showing substantial differences in the mantle PT regimes sampled by the two pipes.

• 7. Compared with the Kalahari craton geotherm, the Main Pipe xenolith linear array is parallel to, but cooler than, the geotherm down to a depth of ~45 kb. The lowest PT xenoliths from the Satellite Pipe plot close to the geotherm, but they are exceptions, and as noted above (a), their temperature estimates should be treated with caution. However, at P > 45 kb, xenoliths from both pipes plot on the high-T side of the geotherm. This indicates a perturbed geotherm as proposed for northern Lesotho by Boyd (1973a).

• 8. The Letšeng PT array ranges from low- to high-PT conditions, in this respect resembling the lherzolites from Thaba Putsoa, but unlike a relatively narrow PT range found in xenolith suites in some other nearby Lesotho kimberlites—Matsoku (Gurney et al. 1975) and Pipe 200 (Carswell et al. 1979)—and from the Kimberley area (Dawson et al. 1975; Boyd and Nixon 1978).

• 9. Most Main Pipe and all Satellite Pipe samples lie within the graphite stability field.

Conclusions

• 1. Most Letšeng samples derive from depleted mantle, in this respect being like most lherzolites in other Cretaceous kimberlites. Only a very few samples are more Fe-rich.

• 2. The garnet lherzolite suites in the two pipes apparently derive from different sections of the upper mantle. This conclusion is in accord with observations that the two pipes have different diamond populations.

• 3. Some, but not all, deformed xenoliths from both pipes have equilibrated on the high-temperature side of the Kalahari craton geotherm, indicating they were thermally perturbed, but to different extents.

• 4. Although the thermal perturbation is like that found in deformed rocks from Thaba Putsoa, other xenolith suites from nearby kimberlite pipes do not have this high-temperature overprint. Hence, there is not a consistently perturbed geothermal gradient Lesotho-wide in the upper mantle in the mid-Cretaceous. The reason for these localised thermal differences is debatable, but a possible explanation is differing thermal input into the mantle aureoles surrounding either individual ascending kimberlites or around coarsely crystalline mantle intrusions that have been proposed as the source of the mantle megacryst suites found in most kimberlites (Gurney et al. 1975)

.

Barry Dawson died suddenly at his home in Edinburgh in late January 2013. He will be especially missed by all of his family, colleagues and diverse friends including those of the kimberlite carbonatite and upper mantle community. Barry was known and loved for his wit, charm and conviviality together with his extensive knowledge of kimberlite geology, poetry, ornithology, and African history. He was a raconteur and a great after-dinner speaker, with a penchant for singing rather rude and amusing songs. Barry was widely regarded as the patriarch of the “kimberlite family”. He was well-respected by university faculty, students and industrial geoscientists. Typically, he never held a grudge against persons who held opposing views. Many of us will really miss staying at the “Braid Farm Road Hotel” enjoying the hospitality and G&Ts so generously provided by our host.

Barry was born in Leeds (UK) and attended Leeds University from which he graduated with a B.Sc. in Geology in 1957. His interest in all things African was developed as graduate student with the Research Institute of African Geology. His work there took him to Basutoland (now Lesotho) to study kimberlite intrusions and culminated in his Ph.D (1960). Parts of his thesis appeared in the seminal paper Basutoland Kimberlites (Geol. Soc. America Bull., 73, 1962). This work was the first modern study of kimberlites and was notable in that Barry introduced the term “fluidization” to kimberlite geology; there are still arguments as to the role of this process in diatreme emplacement! Subsequent, to his graduation Barry, and his late wife Christine, embarked again for Africa to join the Tanganyika Geology Survey. It was here he discovered the amazing sodium carbonate volcanism of Oldoinyo Lengai and earned himself a permanent place in petrological history with the publication of his paper in Nature (1962) on these extraordinary rocks. At the time of his death, some 50 years later, Barry was still studying Oldoinyo Lengai and other Tanzanian volcanoes in the Gregory Rift.

In his academic career, Barry was initially employed at the University of St. Andrews (1964–1978), followed by appointment as Sorby Professor of Geology University of Sheffield, and ultimately at the University of Edinburgh 1988–1997. He retired from active teaching in 1997 and continued an active program of research in his two favourite topics; upper mantle-derived xenoliths, and Recent volcanic rocks of northern Tanzania. In recognition of his research he was awarded the Collins medal of the Mineralogical Society of Great Britain in 2012 and the Clough Medal of the Geological Society of Edinburgh (1999). He was a Fellow of the Royal Society of Edinburgh and the German Academy of Scientists.

Barry’s involvement with kimberlites continued throughout his life. In particular, he was one of the convenors of the 1st International Kimberlite Conference held in Cape Town (1973) with field excursions led by him and Peter Nixon to Lesotho. This conference resulted in world-wide attention being given to kimberlites and upper mantle-derived xenoliths, and to Barry as “the authority” on all things “kimberlitic”. Subsequently, Barry was a convenor of the 2nd Kimberlite Conference held in Santa Fe (1978); an event at which major advances in kimberlite geology were reported; these resulting from studies of the material collected in Lesotho and South Africa. As a consequence of the success of these conferences Barry was instrumental in the formation of the International Kimberlite Conference Advisory Committee and was Chairman of this committee from 1986–1998.

During the last years of his life Barry undertook several expeditions to the Oldoinyo Lengai area, even making the difficult trek to the top of the volcano Kerimasi. As late as December 2012 he was back in Lesotho doing field work. One thing he regretted was that he had not yet climbed all of the Scottish Munros—something he was working on still. His passing is a great loss to us all. To some of us involved in kimberlite and carbonatite research he can never be replaced as he was a mentor and guide throughout our careers. Perhaps Barry was one of the last of the great “gentlemen scientists”?

Roger H. Mitchell

Emeritus Professor of Geology

Lakehead University, Thunder Bay, Ontario

Canada