Episodic diamond growth beneath the Kaapvaal Craton at Jwaneng Mine, Botswana

Important implications for the interior workings of the Earth can be drawn by studying diamonds and their inclusions. To better understand the timing and number of diamond forming events beneath the NW margin of the Kaapvaal Craton, a comprehensive reassessment of Jwaneng’s diamond populations has been undertaken. We report new inclusion abundance data from the visual examination of ~130,000 diamonds that validate the predominance of an eclogitic diamond suite (up to 88%) with on average 5% inclusion-bearing diamonds (with inclusions >10 μm in size). From this population, polished plates from 79 diamonds of eclogitic and peridotitic paragenesis have been studied with cathodoluminescence (CL) imaging and infrared spectroscopy (FTIR) traverses. The majority (80%) record major changes in N concentration and aggregation states, as well as sharp boundaries in the CL images of individual plates that are interpreted to demarcate discrete diamond growth events. In addition, bulk FTIR data have been acquired for 373 unpolished diamonds. Silicate inclusions sampled from distinct growth zones define 2 compositional groups of omphacites and pyrope-almandines associated with different N contents in their diamond hosts. These findings reinforce previous observations that at Jwaneng at least seven individual diamond forming events can be identified – 3 peridotitic and 4 eclogitic. The results demonstrate that detailed examination of diamond plates by CL imaging and FTIR traverses is necessary to unveil the complex history recorded in diamonds.


Introduction
Diamonds are considered 'ancient messengers' from Earth's interior.Inclusions trapped during distinct diamond forming events provide insight into the temporal tectono-magmatic evolution of the subcontinental lithospheric mantle (SCLM) and the deep carbon cycle.However, the absolute number of diamond-forming events recorded in the mantle beneath an individual diamond mine and the scale of these events, remains unknown.Hence, it is vital to establish the timing of diamond forming events: e.g., by analysing individual sulphide inclusions for Re-Os isochron ages (Pearson et al. 1998;Wiggers de Vries et al. 2013).Initial attempts have also been made to date individual silicate diamond inclusions using the Sm-Nd isochron method e.g., at the Orapa kimberlite cluster in Botswana and at Finsch, Kimberley and Venetia diamond mines, South Africa (Koornneef et al. 2017;Smith et al. 1991;Timmerman et al. 2017).These data indicate multiple temporally-distinct diamond forming events at the same diamond mine.Similar conclusions have been reached in studies of the 240 Ma Jwaneng kimberlites on the NW rim of the Archaean Kaapvaal Craton (Kinny et al. 1989).Based on C and N stable isotope ratios, N concentration and N aggregation state data obtained by Fourier-transform infrared Editorial handling: T. Stachel Electronic supplementary material The online version of this article (https://doi.org/10.1007/s00710-018-0582-y)contains supplementary material, which is available to authorized users.
(FTIR) spectroscopy and major element compositions of inclusions, multiple eclogitic and peridotitic populations have been inferred: the age of eclogitic diamond formation at Jwaneng was previously determined using a 2-point Sm-Nd isochron on composite silicate inclusions and yielded an age of 1540 ±20 Ma with indication of a second population formed at around 1 Ga (Richardson et al. 1999).Similarly, sulphide inclusions define 2 generations of eclogitic diamonds, one with a Proterozoic (1.5 Ga) age and another suite of Archean (2.9 Ga) age (Richardson et al. 2004;Thomassot et al. 2009).There are no isochron ages available for the peridotitic diamond suite at Jwaneng and only their relative age is constrained based on inclusion chemistry and C-isotope and FTIR data from the diamond hosts (Deines et al. 1997;Stachel et al. 2004).Those FTIR and C-N-isotope data for peridotitic diamonds were acquired on the same fragments but there are no references which part of the diamond (i.e., core or rim) these pieces represent (Cartigny et al. 1998;Deines et al. 1997;Thomassot et al. 2009).
The present study reports a detailed assessment of the diamond populations at Jwaneng to further constrain the number of diamond-forming events beneath an individual mine.New inclusion abundance data from the visual examination of ~130,000 diamonds refine proportions of different inclusion suites.Of these, bulk FTIR data were acquired for 373 whole diamonds, and polished plates from 79 diamonds of eclogitic and peridotitic paragenesis were studied through cathodoluminescence (CL) imaging, with FTIR traverses conducted and major element inclusion data interpreted based on recognised growth zones.

Samples and methods
Over a 4-year period (2013 to 2017) run of mine production was examined from Jwaneng Mine (north, centre and south pipes) with diamonds ranging in size from 0.1 to 0.8 carat (Diamond Trading Company (DTC) sieve classes +07, +09, +11, and 3 grainers) at the Diamond Trading Company in Gaborone, Botswana (DTCB).In total, over 130,000 diamonds were characterised visually for their inclusion content using a binocular microscope.Inclusions were defined as recognisable minerals >10 μm in size.The main distinction criterion is based on differences in colour between, e.g., eclogitic pyrope-almandine garnet (orange), omphacitic clinopyroxene (cpx; pale green) and kyanite (blue) inclusions and peridotitic pyrope garnet (purple), olivine or orthopyroxene (yellow to colourless) and chromite (brown to black) inclusions.Sulphide inclusions were almost ubiquitously associated with a characteristic rosette fracture system around the inclusion.If not otherwise reported, the examined diamonds are of gem and near-gem quality.

Relative proportions of inclusion suites
On average, 5% of the diamonds at Jwaneng contain inclusions (Table 1).Sulphide inclusions are the most abundant (~6 3%, Fig. 1).Diamonds containing multiple sulphides are more abundant but the sulphides are smaller in size compared to diamonds containing single sulphide inclusions.A visual distinction between eclogitic and peridotitic sulphide inclusions was only possible when silicate inclusions were also present in the diamond, and diamonds with combined eclogitic silicate and sulphide inclusions account for ~5% of the inclusion-bearing diamonds.Eclogitic inclusions dominate the silicate population, averaging 69% (Table 1).Six percent of inclusions were unidentified silicates, 6% contained P-type silicate inclusions (Fig. 1), <0.1% had combined peridotitic silicate/ sulphide inclusions.A detailed overview of the 15 individual production parcels examined for relative inclusion proportions can be found in the Online Resources (ESM 1).

Diamond morphology
Polished plates were mainly prepared from octahedral to dodecahedral diamonds, but irregular shapes and macles were also present.Surface features show resorption, stepped faces and trigons.The majority of these diamonds were colourless but some diamonds with different shades of yellow and brown, 2 green coated (from radiation staining) and one pink diamond were also present.Further details are presented in the Online Resources (ESM 2).

CL imaging
Cathodoluminescence imaging of the central plates (Fig. 2) reveals growth histories of the individual diamonds ranging from a single well-defined growth zone (Fig. 2b) to samples having multiple (2 to 4+) growth zones that are recognized by sharp, successive boundaries (Fig. 2e-i).A growth zone may be compromised of multiple similar (genetically related) growth layers.In contrast, resorption horizons show irregular geometry (Fig. 2d core) and can be difficult to distinguish from superimposition of different growth directions (Fig. 2f intermediate zone), especially when successive precipitation of growth layers was involved in diamond formation (Fig. 2f, k).A resorption horizon can also represent a surface for a new growth zone (Fig. 2l intermediate zone).

Modelled mantle residence temperatures
The N aggregation state of a diamond in the mantle varies strongly with residence temperature (and to a lesser extent with time) and therefore can be used to obtain insight into a diamond's thermal history.If the approximate formation time of a diamond (or a distinct growth layer) is known, a time averaged residence temperature can be calculated (Leahy and Taylor 1997).
Temperatures modelled for the FTIR data of plates and bulk diamonds in Fig. 3 were calculated for diamond genesis at 1, 1.5 and 2.9 Ga and kimberlite eruption at 240 Ma, based on ages reported in the literature (Richardson et al. 1999(Richardson et al. , 2004)).Assuming diamond formation at 1540 Ma, mantle residence temperatures of bulk diamonds and the plates vary from ~1010 to >1250 °C.Some plates (Fig. 3) have a relatively homogeneous temperature record (JW020, JW083, JW407), while others seem to reflect a gradual shift (e.g., JW024, JW044, JW102) that in some cases might not be beyond the uncertainty in the data (e.g.JW044).In contrast, other examples imply distinct changes of up to several tens of degrees between individual growth zones that cut across isotherms (e.g., JW033, JW060, JW114, JW147, JW175, JW194).Overall, the FTIR data of the plates (Fig. 3) show a clustering of diamond cores at higher N aggregation (70 to 90 %B; ~1200 to 1250 °C for all 3 calculated ages) and to a lesser extent of lower aggregation states (10 to 30 %B; ~1050 to 1150 °C), where also the majority of rims is located.Data from intermediate growth zones span the whole range.It should be noted that if the analysed volume is not homogeneous in N concentration (which is certain to be the case for most bulk diamonds and some points from traverses on plates), there will be a systematic and predictable overestimate of the mantle residence temperatures (Kohn et al. 2016) that may have some effect on the distribution of the FTIR data in Fig. 3.

General remarks
Inclusion compositional data are summarized in Table 2 for 30 omphacite and 15 pyrope-almandine inclusions liberated from 16 diamonds.All omphacite inclusions (Fig. 4a) are characterised by moderate molar Mg# (71 to 87) and low molar Cr# (<5).Pyrope-almandine inclusions (Fig. 4b) have low Cr 2 O 3 (0.5 wt%) with CaO between 4 and 14 wt%.These compositions fall in the eclogite (G4 and G5) garnet fields as defined by Grütter et al. (2004) and are comparable with previous studies of Jwaneng inclusions (Richardson et al. 1999;Richardson et al. 2004;Stachel et al. 2004).The compositional variation between inclusions, liberated from the same diamond, is between zero to 2.3 wt% in CaO and < 0.3 wt% in Cr 2 O 3 for garnet and between 0.3 to 5.6 in Mg# and < 0.7 in Cr# for cpx.

Inclusion abundance
The average proportion of inclusion-bearing diamonds for 130,000 Jwaneng diamonds, comprising DTC sieve sizes +7, +9, +11, and 3 grainers (equivalent to a combined range from 0.1 to 0.8 carats), is 5%.Previous studies of inclusion- Inclusions were defined as recognisable minerals (>10 μm) for counting purposes; in total ~130,000 diamonds were examined.The inclusion abundance varies between diamond size fractions that ranged from 0.1 to 0.8 carat (sieve class +7 to 3 grainers) and averaged 5% at Jwaneng.Note that ~99% of diamonds containing both silicate and sulphide inclusions are of eclogitic paragenesis.Hence, up to 88% of all inclusion-bearing diamonds at Jwaneng are inferred to be eclogitic bearing diamonds at Jwaneng report 0.7% from 127,000 diamonds of DTC sieve size +05 (Stachel and Harris 2008).A possible explanation for the different abundance data can be differences in sampling criteria (e.g., different inclusion size cut off) and the overall time of the campaign (sampling of different kimberlite facies and varying input of the 3 pipes).
Based on the observation of associated silicate and oxide inclusions, over 99% of sulphide inclusions are thought to be of eclogitic paragenesis and hence, at least 88% of all inclusionbearing diamonds at Jwaneng are eclogitic.Only 6% of inclusions are unambiguously assigned to a peridotitic assemblage while the paragenesis of the remaining 6% of inclusions could not be identified (Fig. 1).
A general predominance of the eclogitic inclusion suite for Jwaneng (Table 1) was also noted in previous studies (Gurney et al. 1995).Stachel et al. (2004) and references therein) link the eclogitic predominance to Proterozoic magmatic and tectonic events that triggered eclogitic diamond formation, potentially resorbing parts of any pre-existing peridotitic diamond suite.

General remarks
Studies of diamond populations worldwide almost always reveal variations in colour and morphology (Gurney et al. 2004).Deines et al. (1997) suggested a rough classification of Jwaneng diamonds based upon their colour, morphology and paragenesis.They concluded E-type diamonds were more often colourless, partially resorbed and of irregular shape, compared to the P-type diamonds that were more commonly octahedral and green coated or brown and deformed.No coherent relationships were reported between N characteristics and diamond paragenesis.The same lack of relationship is observed in the present dataset.While bulk FTIR analyses in were plotted.Isotherms for assumed mantle residence times (calculated after Leahy and Taylor (1997)) of 2.66, 1.26 and 0.76 billion years are based on diamond formation at 2.9, 1.5 and 1 Ga and kimberlite eruption at 240 Ma.Both datasets (plates and bulk diamond analyses) reported here cover most of the range of N content and the entire range of N aggregation previously documented R Episodic diamond growth beneath the Kaapvaal Craton at Jwaneng Mine, Botswana this study and the literature provide broad insights into the N characteristics of the prevalent diamond populations (Online Resources ESM 6), this approach averages the significant N concentration variation found in diamonds that record multiple growth events and systematically overestimates the N aggregation.
Further information on growth relationships between Jwaneng diamonds and their eclogitic inclusions can be found in a complementary study of Davies et al. (2018).The diamonds of that study were also sampled during the campaign at DTC. et al. (1997) state that peridotitic diamonds (n = 20) tend to have lower N contents compared to the eclogitic suite but are similar in N aggregation state and postulate 3 different Episodic diamond growth beneath the Kaapvaal Craton at Jwaneng Mine, Botswana

Conclusion
The data from this study support five main conclusions.(i) Counting statistics for 130,000 diamonds from Jwaneng emphasise the predominance of an overwhelming (>88%) eclogitic diamond suite with individual production parcels containing up to 10% inclusion-bearing diamonds (5% on average).(ii) Over 80% of the eclogitic and peridotitic diamond plates record changing FTIR characteristics with 2, 3 or 4 individual growth zones potentially spanning extended timescales during the evolution of the Kaapvaal Craton, while a minority of inclusion-bearing diamonds were likely formed in a single event.This observation has not been quantified for inclusion-free diamonds.(iii) FTIR data acquired for bulk diamonds integrate N defects from multiple growth zones and consequently do not record the total internal variability, especially when major changes between individual growth zones are present.(iv) Two compositional subgroups of eclogitic garnet and cpx inclusions with different characteristics in N content of their diamond hosts were identified and imply different diamond forming conditions (i.e. a combination of different protoliths, formation depth, fluid composition and most probably age).(v) Detailed examination of diamond plates by CL imaging and FTIR identifies at least 7 individual diamond-forming events recorded in peridotitic (3) and eclogitic (4) diamonds.The scale and timing between these individual events is yet to be fully resolved, as is their distribution with depth.

Fig. 1
Fig.1Average relative proportions of inclusions in the run of mine production of the Jwaneng mine, Botswana.The inclusions are subdivided into eclogitic (E), peridotitic (P) and unidentified silicates; single and multiple sulphides; and eclogitic sulphides and silicates.Inclusions were defined as recognisable minerals (>10 μm) for counting purposes; in total ~130,000 diamonds were examined.The

Table 1
Relative inclusion proportions in diamonds at Jwaneng

Table 2
Averaged major element composition (EPMA results quoted in wt%) for inclusions in diamonds from Jwaneng *Zone = approximate location of inclusion in core, intermediate (int) and rim zones based on visual observation only **n = number of individual analyses per grain