Cretaceous mantle of the Congo craton: Evidence frommineral and fluidinclusions in Kasai alluvial diamonds
Charles W. Kosman a,⁎, Maya G. Kopylova a, Richard A. Stern b, James W. Hagadorn c, James F. Hurlbut ca Department of Earth, Ocean and Atmospheric Sciences, University of British Columbia, 2020 - 2207 Main Mall, Vancouver, BC, V6T1Z4, Canadab Department of Earth and Atmospheric Science, University of Alberta, 1-26 Earth Sciences Building, Edmonton, AB T6G2E3, Canadac Department of Earth Sciences, Denver Museum of Nature & Science, 2001 Colorado Blvd, Denver, CO 80205, USA
a b s t r a c ta r t i c l e i n f o
Article history:Received 15 January 2016Accepted 7 July 2016Available online xxxx
Alluvial diamonds from the Kasai River, Democratic Republic of the Congo (DRC) are sourced from Cretaceouskimberlites of the Lucapa graben in Angola. Analysis of 40 inclusion-bearing diamonds provides new insightsinto the characteristics and evolution of ancient lithospheric mantle of the Congo craton. Silicate inclusions per-mitted us to classify diamonds as peridotitic, containing Fo91–95 and En92–94, (23 diamonds, 70% of the suite), andeclogitic, containing Cr-poor pyrope and omphacite with 11–27% jadeite (6 diamonds, 18% of the suite). Fluid in-clusion compositions of fibrous diamonds are moderately to highly silicic, matching compositions of diamond-forming fluids from other DRC diamonds. Regional homogeneity of Congo fibrous diamond fluid inclusion com-positions suggests spatially extensive homogenization of Cretaceous diamond forming fluids within the Congolithospheric mantle. In situ cathodoluminescence, secondary ionmass spectrometry and Fourier transform infra-red spectroscopy reveal large heterogeneities in N, N aggregation into B-centers (NB), and δ13C, indicating thatdiamonds grew episodically from fluids of distinct sources. Peridotitic diamonds contain up to 2962 ppm N,show 0–88% NB, and have δ13C isotopic compositions from −12.5‰ to −1.9‰ with a mode near mantle-likevalues. Eclogitic diamonds contain 14–1432 ppm N, NB spanning 29%–68%, and wider and lighter δ13C isotopiccompositions of−17.8‰ to−3.4‰. Fibrous diamonds on average containmore N (up to 2976 ppm) and are re-stricted in δ13C from−4.1‰ to−9.4‰. Clinopyroxene-garnet thermobarometry suggests diamond formation at1350–1375 °C at 5.8 to 6.3 GPa,whereas N aggregation thermometry yields diamond residence temperatures be-tween 1000 and 1280 °C, if the assumed mantle residence time is 0.9–3.3 Ga. Integrated geothermobaromteryindicates heat fluxes of 41–44 mW/m2 during diamond formation and a lithosphere-asthenosphere boundary(LAB) at 190–210 km. The hotter-than-average cratonic mantle may be attributable to contemporaneous riftingof the southern Atlantic, multiple post-Archean reactivations of the craton, and/or proximal Cretaceous plumes.
Keywords:Congo CratonAlluvial diamondsMineral and fluid inclusionsCretaceous mantleCarbon isotopesNitrogen content and aggregation state
1. Introduction
Old, cold, and thick roots of lithosphere with the potential to growand store diamonds extend beneath Archean cratons. This subcontinen-tal lithospheric mantle (SCLM) records the growth and modification ofEarth's continents spanning the Archean to present. SCLM mineralsand fluids trapped within diamonds permit us to trace ancient mantleprocesses at 150 km and deeper. Because they are shielded from a con-stantly changingmantle environment, these inclusions provide insightsinto themantle compositions and the thermal states of early continents.
Ancient, cratonic roots are neither homogenous nor static. Plume ac-tivity, rifting, and accretion are capable of modifying the parageneses,structure, and thermal state of SCLM, sometimes to the demise of thecratonic root (e.g. Helmstaedt andGurney, 1995). Thus, characterization
of SCLM within and between cratons, viewed in the context of eachcraton's geodynamic history, helps us to better understand evolutionand changes in mantle dynamics.
The SCLM beneath the Kaapvaal, Slave, Superior, and Siberiancratons have been characterized in substantial detail utilizing diamondinclusions. These studies revealed the mantle lithologies, the thermalstate and the thickness of the lithosphere at the time of kimberliteemplacement and have been fundamental in constructing models ofSCLM evolution. Despite the Congo craton's size confirmed by a high-resolution shear-wave tomographic model of Africa (Begg et al.,2009), few studies have investigated the underlying lithospheric man-tle, including only one systematic study of Congo diamond inclusionsthus far (Mveumba Ntanda et al., 1982).
To fill this knowledge gap, we studied inclusions in diamonds of theKasai block of the Congo craton to constrain for the first time thegeotherm at the time of kimberlite emplacement, the lithosphere-asthenosphere-boundary, the diamond host rocks, and the diamond-
j ourna l homepage: www.e lsev ie r .com/ locate / l i thos
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
forming fluids of part of the Congo craton. These data, obtainable onlyfrom diamonds and their inclusions, complement analogous data forother cratons globally and contribute to understanding the general pat-tern of evolution of the continental mantle.
2. Regional geology
The Congo craton is one of four cratons that comprise the Africancontinent. The craton is composed of several Archean blocks, adjacentProterozoic mobile belts and areas covered by Proterozoic and Phaner-ozoic sediments (Fig. 1). The exposed Kasai block in the southeastern
part of the craton is a heterogeneous Paleoarchean (3.49–3.33Ga) gran-ulite complex (De Waele et al., 2008). Zircons from heavy mineral con-centrates northward of the Kasai river yield laser ablation inductivelycoupled plasma mass spectrometry (LA ICP-MS) U-Pb ages as old as3.6 Ga (Batumike et al., 2009a). The Kasai block is thought to havemostly accreted to the Congo craton by 2.8 Ga (Begg et al., 2009; DeWaele et al., 2008). The 1.8 Ga Karagwe-Ankole Belt (KAB) and KibaraBelt (KIB) (e.g., Tack et al., 2010) on the easternmargin of the craton re-sulted from collisionwith the Tanzanian craton (Begg et al., 2009). Near1.37 Ga, a failed intracratonic rift is recorded within the sedimentaryand volcanic sequences of the Kibara belt (Tack et al., 2010). The
5°S
10°S
15°E10°E 20°E 25°E 30°E
15°S
25°S
20°S
0 500 km
Kibar
aB
elt (
KIB
)
DamaraBelt
Kasai Block
AngolaBlock Lufilian Belt
West C
ongolianB
elt
Congo Basin
Mbuji-Mayi
Gilson 1
Archean crustArchean crust and Paleoproterozoic supracrustalsreworked in Paleoproterozoic (Eburnian Orogeny)
Poorly known province of Paleo-Mesoproterozoiccrust reworked during Pan-African Orogeny
Fig. 1. Geological and geographic positions of sampling locations for studied Kasai diamonds. (A) Geology and tectonic features of central Africa and the Congo craton with locations ofknown kimberlite fields. The geologic map on the larger panel is modified from Begg et al. (2009) with select legend entries preserved. The Karagwe-Ankole Belt (KAB) is outside theboundaries of the inset to the NE. The dotted line indicates Lucapa corridor. Dark purple areas indicatemajor kimberlite fields from Boyd and Danchin (1980); Jelsma et al. (2009); Jelsmaet al. (2013); Campeny et al. (2014) and deWit and Jelsma (2015). TheGilson 1 oil well within the Congo Basin (Lucazeau et al., 2015) is the site of the quoted heatflowmeasurement. Theblack square outline marks the position of Inset B. (B) Inset from (A), showing the relief and rivers with locations of kimberlites and alluvial diamondmines. Light purple areas representhigh altitude, and light green represents low altitude. National and provincial boundaries are in black. White circles represent kimberlites after aforementioned sources and white starsrepresent exploited alluvial diamond deposits in 2013 (after Chambel et al., 2013). Areas of alluvial diamond concentration (in blue) afterDietrich (2000). Area in red represents the regionfrom which the study samples were collected.
2 C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
Neoproterozoic Lufilian Arc obscures the southerly continuity of the KIB(Begg et al., 2009, Fig. 1A). The center of the craton is extensively cov-ered in Neoproterozoic and later basinal deposits (Begg et al., 2009). Aroot of high seismic velocity is as thick as 400 km beneath the centerof the craton and is interpreted as the lithosphere (Begg et al., 2009).
The distribution of Angolan kimberlites and other alkaline magmasis controlled by tectonic structures. Most Angolan kimberlites are con-centrated within the Lucapa corridor or “graben”, a 50–90 km wide,northeast-southwest trending lineament of tectonic weakness episodi-cally active since the Paleoproterozoic (Jelsma et al., 2009, Fig. 1A). TheLucapa corridor is along strike with a 400-km long set of discontinuousfaults in southwest and central Angola (deBoorder, 1982). In addition tokimberlites, the graben also hosts carbonatites and alkaline magmas.Magmas classified as carbonatites appear to be more concentrated insouthwestern Angola off cratonic margins, whereas kimberlites tendto cluster along the northeastern part of the corridor (e.g., Campenyet al., 2014; Jelsma et al., 2009). The distribution of western SouthAfrican, Namibian, and Lucapa graben kimberlites, carbonatites, and al-kaline magmas was proposed to be controlled by asthenospheric meltmigration along pre-existing fracture zones reactivated during southernAtlantic rifting 135 Ma ago (Jelsma et al., 2009). Alluvial diamond de-posit distribution is controlled by the predominately northwardflowingrivers, and thus alluvial deposits are typically to the north of kimberlitefields (Fig. 1B). The deposits are Mid-Cretaceous gravels of the Calondaformation and recent alluvials eroding and redepositing the Calondaformation (Pereira et al., 2003).
Kimberlites of northeast Angola have been dated to the Early Creta-ceous (Robles-Cruz et al., 2012a and references therein); most recently,Robles-Cruz et al. (2012a) reported a U-Pb SHRIMPweighted mean ageof zircons from Catoca of 117.9 ± 0.7 Ma (MSWD 1.3), contemporane-ous with the age of rifting of the southern Atlantic Ocean. Kimberlitesfrom central Angola along the Lucapa corridor are Triassic in age(216–252 Ma; Jelsma et al., 2013) and the Chicuatite kimberlite insouthwestern Angola is 372 ± 8 Ma (Egorov et al., 2007). The CongoCraton is thought to have been above the present day Tristan daCunha mantle plume and associated large low-shear wave velocityprovince of the deep mantle in the Cretaceous, leading some to specu-late about a genetic link between central African kimberlite magmatismand continental breakup with the plume activity (Torsvik et al., 2010).
3. Materials and methods
3.1. Samples
Diamonds in this study include 138 samples from the DenverMuseum of Nature and Science micromount mineral collection, whichwas assembled in the 1960s by Paul Seel and later donated to themuseum. Specimens used in this research were collected from theKasai River of the Katanga Province, Democratic Republic of the Congo(DRC) and selected for the presence of inclusions. The primary sourcesof the diamonds are inferred to be the Early Cretaceous kimberlites ofnortheastern Angola (Fig. 1B), because the present regime of dominant-ly northward flowing rivers in NE Angola had been established by acomplex history of subsidence of the Congo basin beginning inNeoproterozoic time (Linol et al., 2015). The diamonds of this studyare inferred to have been liberated and transported from kimberlitesof northeastern Angola rather than Mbuji-Mayi kimberlites, which are300 km northeast of the Lunda Norte province of northeastern Angolaand at a lower altitude within the Congo basin (Fig. 1B). Diamonds innortheastern Angola are hosted within primary kimberlites, the MidCretaceous gravels of the Calonda formation, and recent alluvial-elluvial gravels deposits consisting primarily of Calonda and Kalahariformation sediments (Pereira et al., 2003 and references therein).Samples from this study were collected from the latter secondary andtertiary collectors.
3.2. Analytical methods
3.2.1. Electron probe microanalysis (EPMA)Twenty-nine non-fibrous diamondswith relatively largemineral in-
clusionswere polished using a diamond-impregnated steel scaife to ex-pose individual mineral inclusions. The inclusions were re-polishedusing diamond pastes to eliminate a possible addition of Fe from thescaife and mounted in an acrylic disk using epoxy or a combination ofaluminum foil and conductive mounting adhesive. Quantitative chemi-cal analyses were undertaken on a CAMECA SX-50 electron microprobe(University of British Columbia, Department of Earth, Ocean and Atmo-spheric Sciences; Appendix B of the supplementary material). Allelements were analyzed with a beam current of 20 nA, an accelerationvoltage of 15 kV, a peak count time of 20 s and two 10 s backgrounds.Detection limits for most oxides were below 0.08 wt.%, and below0.04 wt.% for SiO2, Al2O3, MgO, and CaO. The exceptions are detectionlimits of 0.11 wt.% for Cr2O3, MnO, and NiO in select phases.
Six of the 14 fibrous and coated diamonds with high inclusion den-sity were chosen for EPMA analysis of fluid inclusions. These diamondswere abraded to half their thickness using the diamond-impregnatedsteel scaife in order to sample inclusions from the entirety of thediamonds' growth histories from core to rim. Polished diamonds werecleaned and their exposed inclusions were dissolved by submergingthem in 68–70% HNO3 at 100 °C for 24 h, followed by 48–50% HCl at100 °C for 24 h.
Unexposedfluidmicro-inclusions (b1 μm) situated immediately be-neath the polished surface were identified in electron backscattermodeon the microprobe. The inclusions appear as bright sub-rounded spotswithout any visible surface expression. Inclusions were analyzed witha 1 μm wide beam at a current of 20 nA and an acceleration voltage of15 keV. Only analyses with totals above 4wt.% were selected for furtherwork. Low totals are expected because the inclusion volumes are signif-icantly smaller than the beamwidth (Kopylova et al., 2010; Navon et al.,1988). Weiss et al. (2008) demonstrate that EPMA analysis of fluidmicro-inclusions is accurate to better than ±15%for major elementanalysis.
3.2.2. Fourier transform infrared spectroscopy (FTIR)Infrared spectra were collected on a Nicolet 6700 Fourier transform
infrared spectrometer (CGL-GRS Swiss Canadian Gemlab, Vancouver,Canada). Absorbance spectra were measured at maximum light trans-mission. Background spectra were collected for 120 s prior to the anal-ysis of each sample and were subtracted from each measuredabsorbance spectra. Count time for each spectrumwas 40 s at a spectralresolution of 0.5 cm−1. One to three analyses were performed persample.
Nitrogen content and nitrogen aggregation state of the diamondswere calculated by deconvoluting peaks corresponding to A- and B-centers using spreadsheet 3i-4 provided by John Chapman of Rio TintoDiamonds, Ltd. The spreadsheet uses a least squares approach tofit base-line corrected, thickness-normalized spectra to modeled A, B, C andD nitrogen peak spectra in accordance with the methodology ofMendelssohn andMilledge (1995). Baseline correctionswere performedvisually with coordinates at 404 cm−1, 1523 cm−1, and 4000 cm−1
wavenumbers and sample thickness was normalized based on peakheight at 2160 cm−1 wavenumber. IaA and IaB nitrogen content werecalculated using absorption coefficients of 16.5 ppm IaA cm−1 and75 ppm IaB cm−1 at 1280 cm−1 wavenumber, the former in accordancewith Boyd et al. (1994).
3.2.3. Secondary ion mass spectrometry (SIMS)Mount preparation and secondary ion mass spectrometry (SIMS)
were carried out at the Canadian Centre for Isotopic Microanalysis(CCIM) at the University of Alberta. Sectioned and polished diamondswere cast into a single 25-mm diameter epoxy mount, along with asample of diamond reference material. The mount was coated with
3C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
10 nm of Au before scanning electron microscopy (SEM) using a ZeissEVO MA15 instrument, operating at 15 kV and 3 nA beam current.Cathodoluminescence SEM imageswere obtainedusing a parabolicmir-ror coupled to a high-sensitivity, broadband photomultiplier detector.The mount was subsequently coated with an additional 20 nm Au filmprior to SIMS analysis.
Isotopes of carbon (13C/12C) were determined using the CamecaIMS-1280 multi-collector ion microprobe at CCIM. Analytical tech-niques follow the methods and reference materials detailed by Sternet al. (2014) and summarized by Palot et al. (2014). The primarybeam conditions used a 20 keV acceleration voltage on 133Cs+ ions fo-cused to a diameter of ~15 μmand beamcurrents of ~2.3 nA. The prima-ry beamwas rastered across a 20 x 20 μm area for 30 s prior to analysis.Negative secondary ionswere extracted through 10 kV to the groundedtransfer section. Automated tuning of the secondary ions in the transfersection preceded each analysis. Secondary ion collection conditions forC-isotopes included an entrance slit width of 100 μm, field aperture of5 x 5 mm, a field aperture-to-sample magnification of 100×, and afully-open energy slit. Both 12C− and 13C−were analyzed simultaneous-ly in Faraday cups (L’2 using 1010 Ω amplifier, and FC2 with 1011 Ω) atmass resolutions of 1975 and 2900, respectively. Faraday cup baselineswere determined once at the start of the session. Mean count rates for12C− and 13C− were typically 1.2 x 109 and 1.3 x 107 counts/s, respec-tively, determined over a 75 s counting interval, with total analysistime of 210 s. The analytical sequence interspersed measurements ofunknowns with the natural diamond S0270, havingδ13CVPDB = −8.88 ± 0.10 ‰ in a 4:1 ratio. Instrumental mass fraction-ation (IMF) for 13C−/12C− was−23.1‰, determined from all the repli-cate analyses (N=60) of S0270 diamond. The standard deviation of the13C−/12C− values was 0.05‰ after correction for systematic within-session IMF drift of +0.5‰ over 18 h. Uncertainties of individualδ13CVPDB analyses propagate within-spot (~±0.05‰, 1σ), between-spot (±0.15‰, 1σ), and between-session errors (±0.01‰, 1σ); togeth-er, errors average ±0.32‰ (2σ). A larger-than-normal (typically±0.05‰, 1σ) between-spot blanket error was applied to account forpossible biases resulting from the analysis of diamonds over a wide ra-dius (i.e., ±8 mm, compared to ±5 mm normally) away from themount center.
Nitrogen abundances were determined immediately followingdetermination of C isotopes from the identical spot locations, typically5 spots per diamond (Appendix C of the supplementary material). Theprimary beam was rastered for 60 s prior to analysis. Secondary ioncollection conditions included an entrance slit width of 40 μm, fieldaperture of 3 x 3 mm, and energy slit width of 40 eV transmittinglow-energy ions. The molecular ions [12C14N−]/[12C12C−] wereanalyzed simultaneously using a Faraday cup–Faraday cup (L’2–FC2)combination where 12C14N− N 1 × 105 counts/s, and Faraday cup–electron multiplier combination (L’2–EM) where b1 × 105 counts/s.Mass resolution was 7000 and 2100 for 12C14N− and 12C12C−, respec-tively. Electron multiplier counts were corrected for background anddead time (40 ns). Total analysis time was 240 s, including pre-analysis raster, secondary ion centering, and peak counting time of50 s. The sensitivity factor for N in diamondwas determined by analysisof diamond reference material S0280 on a separate mount having N =1670 atomic ppm (±5‰ absolute, 95% confidence; Stern et al., 2014).
4. Results
4.1. Physical characteristics
The morphology of the diamonds, together with the degree of re-sorption and color, were observed with a binocular microscope(Table 1, Appendix A of the supplemental material). Most non-fibrousdiamonds are of octahedral morphology, followed by resorbed dodeca-hedra, polycrystalline aggregates, cuboidmorphologies,macles, and un-identifiable morphologies. The samples are colorless, brown, yellow,
pink, green, and gray. Fourteen diamonds were classified as fibrous orfibrous coated diamonds based on their low degree of transparency,cube or twinned cube morphologies, and uneven surface appearance.Surface features of the diamonds are diverse. Many octahedral dia-monds showed no surface relief, while others had strongly steppedfaces and striation styles. Only a few diamonds showed lightly frostedsurfaces. Signs of alluvial origin including mechanical wear or radiationdamage spots are sparse. The wear is expressed as a pattern of shallowcracks (Afanasiev et al., 2000) on a few dodecahedral diamonds; abra-sive rounding of edges or mechanical polishing typical of alluvial dia-monds abraded in a marine beach environment for several hundredmillion years (Afanasiev et al., 2000) are absent.
The observedmorphologies result frommultiple growth and resorp-tion events as evidenced from cathodoluminescence (CL) photographs.They reveal that all diamonds contain multiple layers with distinct CLcharacteristics (Fig. 3). The interpreted growth zone boundaries rangefrom rounded to angular. The constituent layers of rounded growthzones commonly appear truncated by more exterior growth zones,interpreted as diamond resorption between growth events. Four ofthe 40 diamonds imaged in CL display more complex structures thatsuggest more than one core is present. CL brightness varies widelyacross samples, with the entire observed spectrum of brightness
Table 1Resorption, colors, and habit of 138Kasai River diamonds (resorption classification followsthe scheme of McCallum et al., 1994).
Resorption class # Color # Habit #
1 33 Colorless 49 Octahedron 492 4 Brown 37 Dodecahedron 353 11 Yellow 17 Polycrystalline aggregate 134 16 Pink 9 Cube or cubo-octahedron 125 37 Green 9 Macle 106 18 Gray 3 Unidentified 5Unidentified 5 Fibrous or fibrous coated ⁎14
⁎ Fibrous diamonds are not categorized in color or resorption.
Fig. 2. Representative photographs of studied Kasai River diamonds. Distance betweenblack lines at the bottom of each photograph indicate 1-mm scale (A) Peridotiticdiamond 11626 bearing olivine and orthopyroxene inclusions. (B) Peridotitic diamond11691 bearing exclusively olivine mineral inclusions. (C) Eclogitic diamond 11712bearing both garnet and clinopyroxene inclusions. (D) Fibrous coated diamond 10284.
4 C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
occurringwithin single samples. Thefibrous diamonds andfibrous coatshave significantly dimmer CL emission than non-fibrous growth zones.
Transparent mineral inclusionswere observed in a subset of the col-lection. The inclusions ranged in size from 15 μm to 200 μm along theirmaximum dimension, averaging 40–50 μm, and spanned equant, pro-late, and oblate shapes. Some inclusions displayed octahedral morphol-ogies imposed by the host diamond, including stepped faces andtrigons, while others were anhedral. All inclusions appeared colorlessunder the binocular microscope.
4.2. Compositions of mineral inclusions
EPMA analysis of 106 mineral inclusions in 29 diamonds revealedfour main phases: olivine, orthopyroxene, low-Cr garnet, andomphacitic clinopyroxene (Table 2 and 3; Appendix B of thesupplementary material). Most of the diamonds (n = 21) host onlyone mineral phase, with olivine-only diamonds comprising approxi-mately half the sample set (n = 15). Four diamonds host onlyomphacite, one diamond hosts only orthopyroxene, and one diamondcontains only low-Cr garnet. Two-phase assemblages comprise theremainder of the sample set (n = 8). Olivine + orthopyroxeneassemblages constitute the most abundant two-phase assemblagetype and the second most abundant assemblage overall with a total ofseven samples. A single diamond hosts both low-Cr garnet andomphacite. None of the analyzed inclusion grains are zoned.
Olivine is themost abundantmineral inclusion of the studied sampleset. A total of 64 olivine inclusions from 22 diamonds were identified
through energy-dispersive X-ray spectroscopy (EDS), though only 51inclusions in 20 diamonds were large enough to yield wavelength-dispersive X-ray spectroscopy (WDS) analyseswith sufficiently high to-tals. Mg# (100 * molar Mg/(Mg+Fe)) of the olivine inclusions rangesfrom 90.6 to 94.6 with a mean of 93.1. This mean Mg# lies closer tothemeanMg# of the harzburgitic (93.2) than the lherzolitic (92.0) oliv-ine diamond inclusions of the worldwide dataset (Stachel and Harris,2008, Fig. 4). One diamond (11691) shows large heterogeneity in oliv-ine Mg#, with a single inclusion towards the rim of the diamond mea-sured at 90.6 in contrast to six other olivines whose Mg# averages93.1 (Appendix B of the supplementary material). All other olivine-bearing diamonds show less than 0.5 difference in Mg#, often lessthan 0.3, between olivine inclusions.
Orthopyroxene is the second most abundant inclusion phase, with15 inclusions analyzed from eight diamonds. The Al2O3 and Cr2O3 con-tent are strongly correlated with one another (r = 0.90, Fig. 5A). TheAl2O3 content of thirteen of the inclusions is less than 1.5 wt.% suggest-ing their equilibration with garnet (Boyd et al., 1997)., while two inclu-sions froma single diamondmay be equilibratedwith spinel.Mg#of theorthopyroxene inclusions ranges from 91.6 to 94.0 with a mean of 93.7(Fig. 5B), slightly higher than the average Mg# of the olivine.Orthopyroxene analyses broadly occupy the sameMg# – CaO space de-fined by other orthopyroxene diamond inclusions from cratons world-wide (Fig. 5B) and cluster near the overlap of harzburgite andlherzolitic parageneses (Stachel and Harris, 2008). One Ca-richorthopyroxene (CaO= 0.93 wt.%) lies outside the defined fields.
Clinopyroxene is the third most abundant inclusion phase with 14inclusions from five diamonds. Mg# of the clinopyroxene ranges from63.9 to 77.1, with an average of 71.6. Calculated jadeite componentranges from 10.7 to 27.3 mol% (mean 18.9 mol%). Na is present in con-centrations ranging from0.21 to 0.47 andAl from0.22 to 0.63 (both on a4 cation unit basis). Clinopyroxenewith lower concentrations of Na andAl host them in nearly 1:1 cation ratios,whereas grainswith higher con-tents of both tend to contain excess Al cations over Na cations (Fig. 6).This excess is attributable to the presence of Tschermak clinopyroxenecomponents (Stachel and Harris, 2008). The K2O content is up to0.58 wt.% (mean= 0.27wt.%). The lowMg#, significant jadeite compo-nent, low Cr (maximum 0.21 wt.% Cr2O3) and elevated Al (minimum5.06 wt.% Al2O3) of clinopyroxenes assign them to an eclogitic paragen-esis. The MgO content displays the most heterogeneity between
Fig. 3. Representative cathodoluminescence (CL) images of six polished diamonds of the Kasai River. Points correspond to where δ13C and N ppm were analyzed via SIMS. δ13Cmeasurements for each spot are indicated in regular font, N concentration measurements are indicated in italicized text. Sample MB2-5 (D) is a fibrous cuboid diamond, all otherimages are of non-fibrous diamonds.
Table 2Summary of Kasai River alluvial diamond inclusion phases and parageneses.
Inclusions species # diamonds # inclusions Bulk composition
⁎ Addition symbols indicate the respective number of each phase within diamondsbearing multiple phases.
5C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
diamonds, with a range of 10.5 wt.% due to two highly magnesianclinopyroxenes belonging to a single diamond (sample 11664). Al-though they overlap with websteritic compositions in terms of theirNa and Al cation concentrations, their Mg #s (77.0, 77.1) lie closer tothe worldwide mean of eclogitic clinopyroxenes (76.8) than that ofwebsteritic clinopyroxenes (83.6; Stachel and Harris, 2008). Al2O3 isthe next most heterogeneous oxide with a range of 9.6 wt.%, spanningAl2O3 content from 5.1 wt.% in sample 11664 to 14.6 wt.% in sample11712. For diamonds that contain more than one clinopyroxene inclu-sion, maximum heterogeneity occurs between the five analyzedclinopyroxenes of sample 11635, with a range of 1.11 wt.% for MgOand 1.07 wt.% for Al2O3. Range of oxide contents is less than 1 wt.% forall other analyzed oxides.
Garnet is the least abundant inclusion type with 13 inclusions fromtwo diamonds. Only seven were large enough to yield sufficiently highWDS oxide totals. Pyrope garnet Mg# is restricted to 49.7–53.1. TheCr2O3 content of garnet is below detection limit in all but two analyses.CaO content ranges from 8.94 to 10.39 wt.%. TiO2 content is below de-tection limit in most instances, though averages to 0.68 wt.% betweentwo inclusions in a single diamond (11712). These criteria classify allobserved garnets as G3 eclogitic in the scheme of Grütter et al. (2004).Na2O ranges from 0.23 to 0.48 wt.%, averaging 0.33 wt.% across all ana-lyzed garnet inclusions. FeO (range 2.43 wt.%), CaO (range 1.46 wt.%),and Al2O3 (range 0.93 wt.%) contents show the most variation acrossthe suite. However, garnets belonging to the samediamond are less het-erogeneous, with a maximum range of 0.37 wt.% FeO for two analyzedgarnets of sample 11712 and a maximum range of 0.80 wt.% Al2O3 forfive analyzed garnets of sample 11601.
Studied Kasai alluvial diamondswith inclusions are divided into twoparageneses. The 23 diamonds bearing inclusions of olivine,orthopyroxene, or both are assigned to peridotitic parageneses. Six dia-monds bear clinopyroxene, garnet, or both. LowMg# and Cr-poor com-positions of garnet unambiguously assign them to eclogitic parageneses(Table 2 and 3).
4.3. Compositions of fluid inclusions
A total of 48 micro-inclusions across six fibrous/fibrous coated dia-monds were analyzed via EPMA. The average fluid inclusion composi-tion of each diamond is listed in Table 4. Samples 10274, 11671, MB2-1, and MB2-3 are entirely fibrous and dense with fluid inclusionsthroughout; sample MB2-6 is dense with fluid inclusions except forone clear growth band near its edge; and sample MB2-4 hosts fluid
inclusions only in thin fibrous coats mantling an octahedral growthzone. Normalization of individual analyses to K+Na, Ca+Fe+Mg,and Si+Al cations indicates that most analyzed inclusions are silicic(Fig. 7).
Each diamond is relatively restricted in the composition of micro-inclusions. The clear exception to this is diamond 11671, with one anal-ysis of anomalously low Si+Al content and another with anomalouslylow K+Na. These analyses may represent accidental entrapment ofsolid or mixed solid and fluid material, rather than pure daughter min-erals crystallized from the trapped fluid, or a bias introduced by anom-alously large daughter minerals (Kopylova et al., 2010). No systematicchemical trends with CL-identified growth zones are observed for anyof the six diamonds.
4.4. N analysis using different techniques: comparison between FTIRand SIMS
Thirty-one octahedral diamonds were measured for nitrogen con-centration both through FTIR and SIMS spot analyses within growthzones identified with CL. The analyses provide an opportunity to com-pare the two methods of N concentration analysis, the former measur-ing only optically active N, the latter measuring all N. Twenty-eightdiamonds show agreement between the N concentration range deter-mined on SIMS and bulk N concentration determined through FTIRspectra deconvolution, while for 2 diamonds (11655 and 11738) the Nanalyses do not match. In both instances, the bulk calculated N concen-tration exceeds the maximum N concentration determined from SIMSspot analyses. Both diamonds have very low N concentration deter-mined by both methods. Sample 11655 ranges from 1.0 to 10.3 N ppmin spot analyses and bulk N concentration was calculated as40 N ppm. Sample 11738 similarly ranges from 0.8 ppm to 46.4 ppm,whereas bulk N concentration is calculated as 55 ppm.
The distributions of SIMS spot analyses within samples 11655 and11738 sufficiently cover their respective growth zones identified in CL.Thus, havingmissed a growth zone of higher N concentration exceedingthe bulk calculated N concentration seems unlikely. Rather, FTIR spectradeconvolution is an inherently less precise technique for low-N concen-tration diamonds. The magnitude of change in modeled peak heightcorresponding to A-aggregated and B-aggregated nitrogen is smallerat lower nitrogen contents than for higher nitrogen contents, e.g., thedifference in peak height between 10 ppm NA and 20 ppm NA is lessthan that between 130 ppm NA and 140 ppm NA (or NB).
Table 3Select electron microprobe analyses of silicate mineral inclusions.
- Not measured.⁎ All oxides and total reported in wt.%.§ Ol = Olivine, OPX = Orthopyroxene, CPX = clinopyroxene, Gt = Garnet.† Jd = Jadeite.‡ Below detection limit.
6 C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
The FTIR spectra of sample (11699) went off scale as a result of largesample thickness and high actual N concentration so its peaks could notbe deconvoluted. SIMS data indicates the diamond ranges from10.8 ppm N to 1314.9 ppm N, with the majority of growth zones (andtherefore diamond volume) in excess of 760 ppm N. Therefore, the Nconcentration estimates of this sample based on FTIR and SIMS are inagreement.
Despite the uncertainties, the agreement between FTIR spectradeconvolution and direct SIMS measurement of N concentration is suf-ficiently robust, considering that the former relies on a visual estimationof baseline and spectra fit. For further constraints we used N concentra-tion values provided by FTIR measurements, as they directly relate to
corresponding values of N aggregation and calculated temperatureestimates.
4.5. Nitrogen characteristics and nitrogen geothermometry
Infrared spectra were obtained for 32 octahedrally grown diamonds.Calculated N concentration varies from 25 ppm to 600 ppm, with an av-erage of 230 ppm. N concentration appears bimodal, with a large modenear 100–149 N ppm and a smaller mode near 350–399 ppm. The de-gree towhich total nitrogen is aggregated into B-centers (denoted here-after as NB) ranges from 0 to 88%. NB proportion appears Gaussian, witha mode at 60–70%. Diamonds of eclogitic parageneses, on average, con-tain both more nitrogen and greater proportion of NB than diamonds ofperidotitic or unknown parageneses (Fig. 6). All but one analysis yield-ing 2962 N ppm contain less than 1350 N ppm. Most fibrous diamondN ppm content ranges from 957 to 2160 ppm.
The proportion of N aggregated as B-centers is a function of nitrogencontent, mantle storage time, and mantle storage temperature (Tayloret al., 1990). The kinetics of transformation is strongly dependent onmantle storage temperature, but weakly dependent on mantle storagetime. This provides a geothermometer surprisingly consistent with con-ventional coexistingmineral inclusion geothermobarometers but a poorgeochronometer. If the diamonds are known to be at least 200 Ma,
0
5
10
15
0
2
4
6
8
10
12
0
10
20
30
96 989492908886
A
(n = 173)
B
C
Mg/(Mg+Fe)*100
Worldwide harzburgitic olivine inclusions
(n = 49)
Worldwide lherzoliticolivine inclusions
Mveumba Ntanda et al., 1982 (n = 2)
This study (n = 49)Kasai olivine inclusions
Fig. 4. (A) Mg# of olivine inclusions in alluvial diamonds of the Kasai River (blue), andMbuji-Mayi kimberlites (green) of Mveumba Ntanda et al. (1982). Dashed line marksmean Mg# of olivines from peridotite (garnet lherzolite) xenoliths from theSomacuanza kimberlite after Boyd and Danchin (1980). (B) and (C) Mg# of worldwideharzburgitic and lherzolitic olivine diamond inclusions, respectively after Stachel andHarris (2008).
90 91 92 93 94 95 96 970.0
0.2
0.4
0.6
0.8
1.0
CaO
Mg#
0.5 1.0 1.5 2.00.0
0.2
0.4
0.6
0.8
1.0
Al2O3
Cr 2
O3
A
B
GarnetPeridotite
SpinelPeridotite
Fig. 5. (A) Cr2O3 vs. Al2O3 of orthopyroxene inclusions from alluvial diamonds of the KasaiRiver (red circles). Orange square indicates the mean orthopyroxene composition fromperidotite xenoliths of the Somacuanza kimberlite (Boyd and Danchin, 1980), the singleavailable xenolith orthopyroxene analysis reported in the literature for Congo craton.Dashed line indicates cutoff of Boyd et al. (1997) between spinel and garnet peridotiteorthopyroxene compositions. (B) Mg# vs. CaO of orthopyroxenes from alluvial diamondsof the Kasai River. Symbology as in (A). Green and blue areas outline compositional fieldsof lherzolitic and N95% of harzburgitic orthopyroxenes, respectively, within worldwidedataset of Stachel andHarris (2008). Averaged analytical error bars (upper right)were cal-culated using electron microprobe XMasPlus software.
7C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
calculated temperatures vary little with different assumed mantleresidence time (Navon, 1999; Stachel and Harris, 2008).
In order to use the degree of N aggregation to infer mantle temper-atures, the mantle residence time of the diamonds should beconstrained based on the geological history of the craton. Because theKasai crust formed 3.33–3.6 Ga and the underlying mantle is expectedto form concurrently with the crust (Pearson and Wittig, 2008), theKasai SCLM should also have similar ages. However, worldwide agesof peridotitic diamonds as determined by Re-Os chronology extendonly as far back as 3.3 Ga (Shirey and Richardson, 2011). In the absence
of definitive Re-Os diamond ages older than 3.3 Ga, the maximum resi-dence time of peridotitic Kasai diamonds are hypothesized to be 3.2 Ga,given that minimum eruption ages of NW Angolan kimberlites are113 ± 1 Ma, i.e., Early Cretaceous (Robles-Cruz et al., 2012a, and refer-ences therein).
Mantle residence time for eclogitic diamondsmay be different, sincecratonic eclogites are thought to be the product of high pressure meta-morphism of a basaltic parent, i.e., subducted MORB (Jacob, 2004). Inparticular, Nikitina et al. (2014) advocate a subduction origin for Kasaieclogite xenoliths based on petrography, major and trace elements,and Proterozoic U-Pb ages of the eclogite xenoliths contemporaneouswith subduction around the Congo craton. Diamonds with eclogitic in-clusions therefore should have mantle residence times only as old asthe difference in age between the most recent subduction eventsresponsible for introducing eclogitic paragenesis to the SCLM andkimberlite eruption ages. Viable ages of Congo eclogites are constrainedby discordant 2.97–1.24 Ga ages of zircon in eclogitic xenoliths ofthe Catoca kimberlite (Nikitina et al., 2014), the Kibara orogeny (1.8–1.4 Ga) and the convergence of the Congo-Tanzania-Bangweulu cratonswith the Kalahari cratons at 1.0 Ga (Begg et al., 2009). The minimummantle residence time of the Kasai diamonds is assessed based on therespective time for eclogitic diamonds. Global Re-Os dating of eclogiticsulfide inclusions in diamonds indicate a near absence of eclogitic dia-monds younger than 900 Ma (Shirey and Richardson, 2011). The mini-mum mantle residence time is estimated to be on the order of 900 Mabased on the global Re-Os dating and on the 1.0 Ga last accretionaryevent to have subducted MORB beneath the southern Congo Craton/Kasai block (Begg et al., 2009).
Figure 8 displays mantle storage temperature isotherms as a func-tion of N concentration and N aggregation for maximum andminimumresidence times of 3.2 Ga and 900 Ma. Comparison of the two assumedmantle residence times indicates temperature discrepancies of less than50 °C for any given diamond. The studied Kasai diamonds resided attime-averaged temperatures between 1000 to 1280 °C given a resi-dence time of 0.9–3.2 Ga.
0.2 0.4 0.6 0.8Al cations
0.0
0.2
0.4
0.6
0.8
Na
catio
ns
Diamond Inclusions
Mvemba Ntanda
This work
Eclogite xenoliths
Kyanite eclogite xenoliths
Feldspathic eclogites
Xenoliths
Fig. 6. Sodium vs aluminum cations for eclogitic (omphacitic) clinopyroxenes. Closed redcircles indicate clinopyroxene inclusions from alluvial diamonds of the Kasai River. Solidline indicates a 1:1 ratio. Open circles indicate omphacitic clinopyroxene compositionsfrom diamond inclusions of Mbuji Mayi (Mveumba Ntanda et al., 1982). Eclogiticxenolith data from Camutue after Boyd and Danchin (1980), Mbuji-Mayi after El Fadiliand Demaiffe (1999), Catoca after Robles-Cruz et al. (2012b), and Catoca and Cat-115after Nikitina et al. (2014). Kyanite bearing eclogitic xenolith data from Chicundo afterBoyd and Danchin (1980) and Mbuji-Mayi after El Fadili and Demaiffe (1999).Feldspathic eclogitic xenolith data are from Vale do Queve after Boyd and Danchin(1980). Purple area outlines compositional field of worldwide eclogitic clinopyroxenediamond inclusions after Stachel and Harris (2008). Analytical error bars (upper right)were calculated using electron microprobe XMasPlus software.
Table 4Averaged composition of fluid inclusions in the fibrous cuboid Kasai River diamonds.
Sample number 11671 MB2-1 10274 MB2-3 MB2-4 MB2-6
Average of 11 7 6 15 5 4
Average Std. dev. Average Std. dev Average Std. dev. Average Std. dev. Average Std. dev. Average Std. dev.
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
4.6. Composition of carbon isotopes and their relationship to nitrogencharacteristics
Carbon isotopic compositions (δ13CVPDB) of 40 diamonds (23 perido-titic, 5 eclogitic, 4 unknown non-fibrous, 8 fibrous) (Fig. 9, Table 5,Appendix C of the supplementary material) ranged from −27.3‰ to−1.9‰. Carbon isotopic compositions follow previously establishedcorrelations to diamond paragenesis (Cartigny, 2005). Eclogiticdiamonds span a wider range of isotopic compositions (−17.8‰ to−3.4‰, mean −8.3‰), than peridotites (−12.5‰ to −1.9‰, mean
−3.9‰). Non-fibrous diamonds of unknownparagenesismay comprisesome of both, given their range from −27.3‰ to −3.1‰ (mean−12.4‰). Fibrous diamonds are restricted to −9.3‰ to−4.1‰ δ13C.
Non-fibrous diamonds with peridotitic inclusions and diamonds offibrous growth habits (parageneses unknown) display relatively re-stricted δ13C and N ranges (Fig. 7). All but three spot analyses of perido-titic octahedral diamonds are between−6‰ to 0‰ δ13C. The exceptionsare diamond core analysis of −12.5‰ δ13C (MB3-4) and two analysesfrom the core of sample 11738 of−8.5‰ and−8.0‰ δ13C. Fibrous dia-mond spot analyses have lower δ13C but with a similar overall range
0
20
40
60
80
0 20 40 60 80
0
20
40
60
80
Si + Al
Ca + Fe + MgK + Na
11671
MB2-1
10274
MB2-3
MB2-4
MB2-6
Fig. 7. EPMA analyses of inclusions in fibrous diamonds normalized to 100% Si+Al, Ca+Fe+Mg, and K+Na cation components. Gray outlined areas represent global observedcompositional fields for silicic, saline, and carbonatitic fluids from Smith et al. (2012) and references therein. Light blue field outlines observed compositions of Zaire cubic diamondsfluid inclusion compositions after Navon et al. (1988) and Kopylova et al. (2010).
1
10
100
1000
10000
0 10 20 30 40 50 60 70 80 90 100
1050
˚C– 3.2 Ga
1100 ˚C – 3.2 Ga
1150 ˚C – 3.2 Ga
1200 ˚C – 3.2 Ga
1250 ˚C – 3.2 Ga
1100 ˚C – 900 Ma
1150 ˚C – 900 Ma
1200 ˚C – 900 Ma
1250 ˚C – 900 Ma
Tot
al N
(pp
m)
%NB
Eclogitic
Peridotitic
Unknown
Fig. 8. Temperature isotherms as a function of total nitrogen concentration (N ppm) and degree to which nitrogen is aggregated as B-centers (%NB). Isotherms calculated based on theequations of Taylor et al. (1990) and Leahy and Taylor (1997). Solid isotherms calculated assuming a mantle residence time of 3.2 Ga, dotted lines assume a mantle residence time of900 Ma. Analytical error bars (lower right) represent ±10% error after Stachel et al. (2002).
9C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
from−9.3‰ to−4.1‰. Fibrous diamonds have on average elevated ni-trogen content compared to peridotitic diamonds.
Non-fibrous eclogitic diamonds are less restricted in their δ13Cisotopic composition. Three of five eclogitic diamonds (11635, 11664,11712) are isotopically indistinguishable from the peridotitic diamonds(range −5.4‰ to −3.2‰). The remaining two eclogitic diamonds(11601, 11731) are isotopically light, with one diamond ranging from−17.8‰ to −14.3‰ and the other from −13.7‰ to−13.0‰.
Diamonds of unknown paragenesis display the most variation intheir δ13C composition. One of the four diamonds (11634) overlapswith the peridotitic and two eclogitic diamonds (range −4.1‰ to−3.1‰). A second diamond is more depleted (11690, −10.9‰ to−9.0‰) and a third sample is uniformly more depleted than all others(11660,−23‰ to−22.1‰). Another diamond (sample 11733c, Fig. 2c)of unknown paragenesis is the isotopically lightest sample and displaysthe most internal isotopic variation. The unresorbed core displays δ13Cof −27.2‰ and −27.3‰; the surrounding resorbed growth zone isisotopically heavier with δ13C of −5.8‰, −5.3‰, −6.1‰ (fromcore to rim); followed by a similarly isotopically heavy, slightlyresorbed growth zone (δ13C = −6.0‰) and an unresorbed coat(δ13C = −5.7‰).
The diamonds showvarying degrees of internal isotopic heterogene-ity. All but two diamonds are restricted to a range of b5‰ δ13C. A peri-dotitic diamond (MB3-4) spans isotopic values of −12.5‰ to −3.6‰,with the isotopically light analysis concentrated in the core and theheavier values averaging−3.9‰. Diamond11733 is the singlemost iso-topically heterogeneous diamond known to date worldwide.
4.7. Cathodoluminescence
The sample suite as a whole and individual samples are heteroge-neous in cathodoluminescent emission. Spot analyses in regions ofdim cathodoluminescence in octahedral zones generally yield low Nconcentration and those within regions of bright cathodoluminescenceyield high N concentration. There are many exceptions to this pattern,however, particularly when the growth zone has fine oscillations in CLcharacter. Large variations in N concentration may be present withinsingle diamonds whose CL characteristics appear relatively uniform.Spot analyses taken in growth zones with a mottled CL appearance(e.g. Fig. 8f, spot 6) generally have N concentration less than 100 ppm,often much less, though unmottled growth layers from the same dia-monds often rival these values. Fibrous diamonds are many times dim-mer in CL than their octahedrally grown counterparts despite their high
average N concentration. Additionally, the fibrous diamonds are signif-icantly more uniform in CL character. No systematic variation of δ13C asa function of CL emission is observed.
5. Discussion
5.1. Sources of diamond carbon and their variation
Non-fibrous Kasai diamonds match the carbon isotope signatures ofother diamonds worldwide (Cartigny, 2005; Stachel et al., 2009). Kasaiperidotitic and eclogitic diamonds have a prominent mode in C isotopecomposition close to −5‰ (Fig. 9). Kasai eclogitic diamonds have δ13Cspanning down to −18.0‰ (sample 11731). This follows the globalpattern, in which diamonds belonging to eclogitic parageneses are lessrestricted in carbon isotopic composition than their peridotitic counter-parts and feature more depleted isotopic compositions (Cartigny, 2005;Stachel et al., 2009). The lighter carbon of eclogitic diamonds is thoughtto broadly reflect subducted organic carbon-rich sediment-bearingsources (Sobolev and Sobolev, 1980).
Select non-fibrous diamonds not assigned a paragenesis by mineralinclusions can be classified based on δ13C. Fromworldwide correlationsbetween diamond paragenesis and δ13C, two non-fibrous diamonds(11600 and 11733) of unknown paragenesiswith growth zones bearingδ13C values−27‰ and−23‰may reasonably be classified as eclogitic.
-30 -20 -10 0 10
500
1000
1500
2000
2500
3000
3500
N ppm
δ13C (‰)
Peridotitic (n = 23)
Non-fibrous
Eclogitic (n = 5)
Unknown (n = 4)
Fibrous (n = 8)
Fig. 9. Carbon isotope composition (δ13C) vs N concentration (ppm) of alluvial diamondsof the Kasai River. Number in legend indicates the number of diamonds of eachparagenesis.
Table 5Carbon isotope (δ13C) and N concentration statistics of select Kasai diamonds.
- Not measured.⁎ δ13C‰ is reported relative to VPBD.
10 C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
A third diamond of unknown paragenesis averaging −10.1‰ δ13C(11690) and a fourth averaging−3.8‰ δ13C (11634)may not unambig-uously be assigned a paragenesis due to the overlap between peridotiticand eclogitic δ13C.
Carbon isotopic signatures of Kasai fibrous coats and fibrous dia-monds overlap with those from all other localities worldwide(Cartigny, 2005; Klein-BenDavid et al., 2010). Their δ13C is close to thatof mantle carbon but shifted towards lighter δ13C values and overlapswith the modeled mode of δ13C of diamond precipitated from carbonatemelt with mantle-like δ13C (Smart et al., 2011). Rapid precipitation fromcarbonatitic melts with C isotope compositions near that of the mantlemight explain the observed uniformity in δ13C and high N concentration(Cartigny, 2005; Navon, 1999) of fibrous diamonds worldwide.
Two Kasai diamonds (11733 andMB3-4) have large contrasts in δ13C(22‰ and 8.9‰, respectively) across growth zones observed within CLimages. Prior investigations of single-diamond δ13C isotopic heterogene-ity have yielded isotopic variation between growth zones typically on theorder of 4‰; larger heterogeneities have seldom been observed(Wiggers de Vries et al., 2013a and references therein). A number of pro-cesses have been invoked to explain carbon isotope variation within in-dividual diamonds, including Rayleigh fractionation (Cartigny et al.,2001). The δ13C composition of diamonds precipitated by fluids undergo-ing Rayleigh fractionation is controlled by thefluid redox state. Growth ofdiamonds from reduced carbon species (i.e., methane) produces 13C-enriched diamond with progressively 13C-depleted compositions.Growth of diamonds from oxidized species (i.e., CO2 or CO3
2−) producesthe reverse, i.e., 13C-depleted compositions with progressively 13C-enriched compositions (Deines, 1980), essentially tapering off at +7‰δ13C relative to the isotopic composition of the initial media. Anotherproposed process is the CO2 escape model of Cartigny et al. (2001) inwhich escaping 13C-enriched CO2 leaves behind an isotopically depletedmedia for diamond growth and hence progressively depleted in 13C.
The two Kasai diamonds with large δ13C heterogeneities have 13C-enriched rims relative to their cores not explicable by these processes.Smart et al. (2011)modeled the frequency and shift of δ13C of diamondsprecipitated at 1100 °C in Rayleigh fractionation processes from oxi-dized fluids/melts. If the model of Smart et al. (2011) begins withfluid/melt at−27‰ δ13C and−12.5‰ δ13C (the C isotope compositionsof the cores of diamonds 11733 and MB3-4) one would expect essen-tially no diamond heavier than−20‰ and−5.5‰, respectively, to pre-cipitate. This precludes Rayleigh fractionation as a viablemechanism forthe abrupt and large changes in isotopic compositions of the adjacentzones observed within some of the Kasai alluvial diamonds. Instead,we attribute formation of diamonds with strongly dissimilar δ13C, Nconcentration and CL between growth zones to precipitation of dia-mond in separate episodes from distinct sources. A similar processwas proposed by Wiggers de Vries et al. (2013a, 2013b) for Yakutiandiamonds with comparable characteristics.
Though carbon isotope compositions vary for the CL-identifiedgrowth zones in sample 11691, precipitation of this diamond inseparate episodes cannot explain its anomalously low-Mg# olivine.The growth zone towhich this inclusion belongs shows only slightly dif-ferent 13C and N concentration characteristics compared with othergrowth zones in the same sample (−4.0‰ δ13C, 7 N ppm, comparedto −2.8‰ δ13C, 185 N ppm average in the core and −3.5‰ δ13C,249 N ppm on the very outer growth zone). Other peridotitic diamondswith greater or equal internal δ13C and N variation (e.g., 11730, 11747,MB3-1) show less variation in inclusion Mg#. Sample 11691 hostedthe most olivine inclusions of any diamond, many toward the rim ofthe specimen, further complicating this simple explanation for theanomalously low Mg#.
5.2. Composition of diamond-forming fluids
Fluid inclusions trapped within diamonds provide the best con-straints on the composition of diamond-forming fluids below the
Congo craton. The micro-inclusions analyzed across the six diamondsrange from 31.3% to 79.7% Si + Al cations when renormalized tothe end members Si + Al (silicic), Mg + Fe + Ca (carbonatitic), andK + Na (saline) and are thus classified as silicic (Fig.7, Table 4, Navon,1999).
Few prior studies have investigated compositions of the crystallizedfluids trapped within fibrous diamonds from the Congo craton. Navonet al. (1988) and Kopylova et al. (2010) used similar EPMA methodsto investigate the compositions of fluid inclusions from DRC fibrous di-amonds from the kimberlites of Mbuji-Mayi. The compositional overlapbetween Kasai fluid inclusions and those from the DRC, and the absenceof carbonatitic and saline fluid compositions jointly from the Kasai andDRC diamonds suggest homogeneity in the diamond fluid compositionsdespite their distance in space and time. The 117.9 ± 0.7 Ma Catocakimberlite (Robles-Cruz et al., 2012a) is ~300 km from the 70 MaMbuji-Mayi kimberlites (Schärer et al., 1997; Figure 1B). The age ofthe diamond-formingfluids is broadly similar to the age of thehost kim-berlites, predating them by ~5 Ma, as implied by the poorly aggregatedN in fibrous diamonds globally (Navon, 1999).
The compositional uniformity in the fluid captured by fibrous dia-monds from the sizable block of the Congo craton is interesting, al-though one cannot assess whether this is the norm or the exceptionfor other cratons. In the absence of systematic EPMA fluid inclusionstudy of spatially distributed kimberlites in themantle, the Slave cratonprovides the only comparable example. Fibrous diamonds from aDiavikkimberlite (Klein-BenDavid et al., 2007), Panda kimberlite (Ekati mine;Tomlinson et al., 2006), and Fox kimberlite (Ekati mine; Weiss et al.,2015) emplaced between 70Ma and 45Ma (Heaman et al., 2003) with-in 40 km of one another contain compositionally inhomogeneous fluidinclusions, spanning all three defined end-member compositions.Thus, the central Slave mantle in the Paleogene produced diamondsfrom widely varying fluids, unlike the Cretaceous Congo mantle.
A possible explanation for the homogeneity or heterogeneity of thediamond-forming fluids may be found within the framework of theWeiss et al. (2015) model, which postulates that the geochemistry offluid inclusionswithin diamonds is a function of their source and subse-quent interaction with SCLM lithologies. Saline (Na + K rich) fluidsoriginating from dehydration reactions within subducting slabs are pa-rental to both carbonatitic and silicic fluid inclusions; saline composi-tions evolve towards carbonatitic compositions upon interaction withperidotitic lithologies, whereas interaction with eclogitic lithologiesproduce silicic compositions (Weiss et al., 2015).
Within this model Kasai fibrous diamonds and fibrous coats andthose previously studied from Mbuji-Mayi (Kopylova et al., 2010;Navon et al., 1988) record interaction of a saline fluid with eclogitic li-thologies underlying the Congo Craton. The absence of fluid composi-tions other than silicic in the DRC and Kasai diamonds thus far suggestthat eclogites comprise a volumetrically significant part of rocks parentto diamond in the Congo SCLM.Weiss et al. (2015) hypothesize that thesource of fluids to Slave craton fibrous diamonds is related to low-anglesubduction of the Farallon plate. Perhaps it is the relatively recent se-quence of events of subduction, fibrous diamond formation, and kim-berlite eruption in the Slave Craton that is responsible for its fluidinclusion diversity. Subducted fluids beneath the Kasai block mayhave had longer time to interact with SCLM lithologies (i.e., between1.0 Ga Proterozoic subduction to Early Cretaceous kimberlite emplace-ment) and therefore transform their original composition to silicic,whereas primitive fluid inclusions survive within the Slave suite.
TheWeiss et al. (2015)model could additionally be thebasis for con-clusions on the evolution of Kasai diamondparent rocks over time. Non-fibrous, older octahedral diamonds with aggregated N from Kasai are67% peridotitic and 19% eclogitic (based on mineral inclusion assem-blages and inferences from δ13C), whereas younger fibrous diamondsare fully eclogitic. This trend indicates that eclogitic parageneses mayhave became more significant diamond host rocks in the Congo SCLMthrough time.
11C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
5.3. Thermal state of the Congo Craton in the Cretaceous
Temperature estimates based on theN aggregation state of diamond,together with traditional mineral thermobarometry for diamonds andmantle xenoliths, constrain the thermal state and the minimal litho-sphere depth of the Congo craton in the Cretaceous.
The studied Kasai River diamond suite contained only one diamond(11712) amenable to thermobarometry, a specimen in which garnetcoexists with omphacite. Temperatures of equilibration for the garnet(two grains, one an average of two analyses and the other a singleanalysis) and clinopyroxene (one analysis) at 5 GPa using thegeothermometer of Nakamura (2009) are 1303 °C and 1318 °C,i.e., within the 25 °C method uncertainty from one another. A more ac-curate thermobarometric constraint on diamond 11712 can bemade bydrawing univariant PT lines that correspond to the coexisting averagegarnet-clinopyroxene pair using the Nakamura (2009) thermometer(thick solid line on Fig. 9) and the Beyer et al. (2015) barometer (dot-dashed line on Fig. 9). The intersection of the two univariant lines indi-cates a pressure-temperature coordinate of 5.8 GPa (191 km depth),1350 °C, along the 44 mW/m2 geotherm of Pollack and Chapman(1977). This PT point fits well the field of single clinopyroxenemacrocrysts temperatures and pressures from nearby Cretaceous kim-berlite Cucumbi 79 (Robles-Cruz et al., 2012b; green field in Fig. 9).The single PT point yielded by traditional thermobarometry for dia-mond 11712 exceeds the N aggregation temperature estimates(1000–1280 °C, Fig. 8 and Fig. 10). This result could be ascribed to thehigher temperatures of freezing-in of the mineral equilibrium for non-touching inclusions (Stachel and Harris, 2008 and references therein)and an incomplete consistency between two independent methods of
thermometry, each with its own freezing-in limits and temperaturescales.
The minimum temperatures of the diamond residence as impliedby the N aggregation place the lower bound on the thermal state ofthe Congo craton. At this temperature of 1000 °C the 41 mW/m2
geotherm enters the diamond stability field. The highest heat flow isconstrained by the garnet-clinopyroxene thermobarometry and is44 mW/m2. This feasible range of 41–44 mW/m2 is hotter than37–42 mW/m2 recorded by diamond inclusions for Archean cratonicinteriors worldwide (Stachel and Harris, 2008), but compare verywell with thermobarometry for mantle xenoliths and macrocrysts inkimberlites of the Congo craton. Univariant PT lines for eclogites fromthe Catoca and Cat-115 pipes of NE Angola (Nikitina et al., 2014) enterthe diamond stability field at 1000–1260 °C (dashed lines on Fig. 9).Garnet lherzolites and phlogopite garnet wehrlites of the Catoca,Tchiuzo, Cucumbi-79, and Alto Cuila-63 kimberlites plot at820–1250 °C and 3.6–5.8 GPa if the thermobarometer of Nimis andTaylor (2000) is employed (Nikitina et al., 2014; Robles-Cruz et al.,2012a, 2012b). These values constrain the geotherm to between 40and 45 mW/m2.
Several tectonothermal events of the Kasai block may serveto explain the high calculated geotherms. Firstly, the higher heat flowmay relate to reworking of Archean crust in Paleoproterozoic,Mesoproterozoic, and Neoproterozoic events as indicated by zirconages (Batumike et al., 2009b). In particular, Neoproterozoic ages for zir-con correspond well to ages for magmatic activity and formation of theKatangan Basin (880 Ma–760 Ma), the peak of the Lufilian orogeny(570 Ma), and the closure of the Katangan Basin (540 Ma) on thesouth and south-east margins of the Kasai block.
4445
4241
40
38
800 1000 1200 1400
T (°C)
P (G
Pa)
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
7.5
145
125
105
165
185
205
225
245
Minimum Naggregation temperature
Maximum Naggregation temperature
Minimum LAB
Maximum LAB
Fig. 10. Pressure-temperature diagram of Kasai diamond inclusions with northeast Angola mantle xenolith data. Thin solid gray lines are geotherms after Pollack and Chapman (1977).Thin horizontal dashed line is the graphite-diamond transition after Kennedy and Kennedy (1976). Solid near-vertical line is a univariant PT line calculated using the thermometer ofNakamura (2009) for the garnet-clinopyroxene pair of Kasai diamond 11712. Dashed line is univariant PT line calculated from the barometer of Beyer et al. (2015). Vertical N aggregationtemperature lines represent minimum (1000 °C) and maximum (1280 °C) calculated mantle residence temperatures. Red circles and orange squares are garnet lherzolite and garnetphlogopite wehrlite temperatures, respectively, after Robles-Cruz et al. (2012b), calculated using the thermometer of Nimis and Taylor (2000). Green field represents a majority(N90%) of single clinopyroxene macrocryst temperatures and pressures from kimberlite Cucumbi 79 after Robles-Cruz et al. (2012b), using Nimis and Taylor (2000). Blue near-verticallines represent univariant PT lines of eclogite xenolith mineral data (mineral core analyses only) of Nikitina et al. (2014) calculated using Nakamura (2009). The two purple lines labeled“Minimum LAB” and “Maximum LAB” correspond to the depth of intersection of the Nakamura (2009) and Beyer et al. (2015) univariant PT lines and the depth of intersection of the41 mW/m2 geotherm with the maximum calculated N aggregation temperature, respectively.
12 C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
Alternatively, the high geotherms may be directly attributable torifting of the southern Atlantic. Contemporaneous eruption of kimber-lites on the São Francisco craton (Chaves et al., 2008), in South Africa(Jelsma et al., 2009), and Catoca (Robles-Cruz et al., 2012a) each alongnortheast-southwest trending lineaments, is consistent with the timingof southern Atlantic rifting. Rifting may increase the heat flow inadjacent cratonic SCLM; for example, Miller et al. (2012) report heatingand destruction of the Southern Superior SCLM attributable to thethermal disturbance of theMidcontinent Rift at 1.1 Ga, which increasedthe heat flow from 39–41 mW/m2 pre-rift to 41–42 mW/m2 in the Ju-rassic. Similarly, Smit et al. (2014) report a geotherm of 42 mW/m2
constrained from 1.1 Ga diamonds near the Midcontinent Rift, versus39 mW/m2 from diamonds of Jurassic kimberlites. South AfricanSCLM xenoliths record a regional protracted heating event (~100 °C,about+2mW/m2) associatedwith Gondwana supercontinent breakupstarting at ~200 Ma propagating from east to west (Bell et al., 2003).
Lastly, elevated geothermsmay be a function of the proximity to theTristan plume head during the time of Congo kimberlite eruptions(e.g., Torsvik et al., 2010). O'Connor and Duncan (1990) and Gibsonet al. (1999) infer the Tristan da Cunha plume to have been underneathsoutheastern Brazil at the time of onset of continental breakup based onthe location of the Parana-Etendeka continental flood basalts. The largeplume (~1500-km diameter) may have been expressed across parts ofthe Congo craton SCLM.
Present day heat flow measurements of Southern DRC for compari-son with our determined paleogeotherms are available, though scarce.Sebagenzi et al. (1993) report the single existing heat flow measure-ment from a borehole at Mbuji-Mayi to be 44 mW/m2, with muchhigher geotherms in the Kibara belt (53.0 ± 30 mW/m2 to 69.0 ±26 mW/m2 across three boreholes). A more recent heat flow estimateof 40 ± 5 mW/m2 based on data from the Gilson 1 (Fig. 1A) andM'bandaka oil wells by Lucazeau et al. (2015) is in good agreementwith the former. These estimates are in good agreementwith the calcu-lated 41–44mW/m2 in this study. This continuity is expected given thatno major tectonic events have affected the Congo Craton since theCretaceous.
The constrained thermal regime and the restriction of diamonds tothe lithosphere (Boyd and Gurney, 1986) enable assessment of the lith-osphere thickness. The 41mW/m2model geotherm intersects themax-imum N aggregation temperatures at 6.3 GPa (208 km). If the highergeotherm of 44 mW/m2 is assumed, the greatest depth where the dia-monds occur in the Congo mantle is 5.8 GPa (191 km) (Fig. 10). The190–210 km lithosphere matches the typical cratonic lithosphere of~200 km as defined petrologically (e.g., Eaton et al., 2009; Kopylovaet al., 2016 and references therein). The lithosphere beneath the Kasaiblock may be thinner than the lithosphere in the center of the Congocraton as suggested by geophysical surveys (Begg et al., 2009). The po-sition of the petrological lithosphere-asthenosphere boundary belowthis craton is shallower than that mapped by seismic methods, similarto the relationships between petrological and geophysical lithosphereobserved globally (Eaton et al., 2009). Specifically, Hansen et al.(2009) report an LAB potentially as deep as 285 kmbeneath the Bogoin,Central African Republic permanent seismic station (BCGA), thoughthey cast doubt on this estimate. Global seismic tomographymodels in-dicate high velocity domains underlying the Congo craton, suggestinglithospheric roots as deep at 300–400 km (Begg et al., 2009).
5.4. Peridotitic and eclogitic origin of diamonds
None of the 23 peridotitic diamonds bear clinopyroxene orgarnet and therefore may not be unambiguously assigned to lherzoliticor harzburgitic parageneses. Nevertheless, high Mg# of olivineinclusions in the suite (Fig. 4) and absence of peridotitic garnet andclinopyroxene suggest extremely depleted peridotitic mantle, likelyharzburgitic. Analyzed clinopyroxenes are low inMg# and bear a signif-icant jadeite component, classifying them as eclogitic. Na2O and K2O
concentration in garnet and clinopyroxene, respectively, are elevated(Na2O in garnet = 0.21 to 0.48 wt.%, average 0.30 wt.%; K2O inclinopyroxene = 0 to 0.58 wt.%, average 0.27 wt.%), typical for dia-mondiferous parageneses (Stachel and Harris, 2008). On the basis of di-amond inclusion parageneses, at least 70% of the suite is peridotitic(n= 23) and 18% is eclogitic (n= 6; remainder n = 4 unknown). Car-bon isotopic data supports these assignments of paragenesis, wherein12.5% of the non-fibrous samples can be unambiguously assigned toeclogitic parageneses (δ13C b -13‰, n = 4).
Prior studies of DRC and Angolan diamond inclusions and xenolithshave also revealed the presence of both peridotitic and eclogitic para-geneses within Congo craton SCLM (e.g., Boyd and Danchin, 1980;Mveumba Ntanda et al., 1982; Nikitina et al., 2014; Robles-Cruz et al.,2012b). The origin of eclogitic parageneses within the suite is likely at-tributable to subduction during cratonic amalgamation on the southernmargin of the Congo craton in Proterozoic time (Nikitina et al., 2014).Peridotites of the Congo craton have never been studied in detail, butin general peridotitic SCLM is the residuum of extreme degrees of lowpressure polybaric partial melting, possibly the Archean equivalent ofmodern day oceanic spreading centers imbricated in Archean time(Pearson and Wittig, 2008).
The proportion of eclogitic to peridotitic diamonds does not reflectthe true abundance of SCLM parageneses at depth. The observed 12–18% of eclogitic stones among the studied Kasai diamonds is higherthan b4 vol.% of eclogites at depth (McLean et al., 2007; Russell et al.,2001; Schulze, 1989) due to preferential formation of diamonds ineclogites (Gurney et al., 2005 and references therein) and because theLucapa kimberlites were emplaced off the thickest part of the Congocraton root (Begg et al., 2009). Kimberlites sampling cratonic marginscontain significant proportions of eclogitic diamonds attributable topost-Archean subduction events (Jacob, 2004; Shirey et al., 2003).
6. Conclusions
Kasai alluvial diamonds provide insights into the evolution of kim-berlites and of the greater Congo craton. These diamonds exhibit chem-ical and mineralogical characteristics suggesting that they experiencedepisodic growth spanning a variety of residence times from isotopicallyheterogeneous sources. Kasai diamonds sourced carbon from bothsubducted organic material and the mantle. Olivine, orthopyroxene,clinopyroxene, and garnet inclusions in these diamonds indicate thatthey originated from garnet-facies harzburgitic (70%) and eclogitic(18%) sources. Silicic fluid inclusions in fibrous diamonds require spa-tially extensive homogenization of the diamond-forming fluid withinthe Congomantle. Clinopyroxene-garnet thermometry suggests that di-amonds formed at 1350–1375 °C, and may have experienced residencetemperatures between 1000 and 1280 °C, over their 0.9–3.3 Ga mantleresidence time. Observed heat fluxes are slightly higher than those ex-pected for cratonic heat flows. These elevated heat flows might reflectcontemporaneous development of adjacent rifting, thermal disturbanceby plume and multiple post-Archean reactivations of the craton.
Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.lithos.2016.07.004.
Acknowledgments
The Denver Museum of Nature and Science is thanked for makingthe diamond collections available for research. Branko Deljanin isthanked for allowed use of his infrared spectrometer. John Chapman isacknowledged for providing nitrogen deconvolution spreadsheets.Wewould also like to thankMati Raudsepp and Edith Czech for their as-sistance in collecting microprobe analyses. Susan Bucknam, LarryHavens and RickWicker are thanked for their assistance with catalogu-ing and photographing diamonds used in this research. This researchwas supported by an NSERC discovery grant no. RGPIN-2014-04629 toMGK.
13C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
Batumike, J.M., Griffin,W.L., O’Reilly, S.Y., 2009a. Lithospheric mantle structure and the di-amond potential of kimberlites in southern DR Congo. Lithos 112, 166–176.
Batumike, J.M., Griffin, W.L., O’Reilly, S.Y., Belousova, E.A., Pawlitschek, M., 2009b. Crustalevolution in the central Congo-Kasai Craton, Luebo, DR Congo: Insights from zirconU–Pb ages, Hf-isotope and trace-element data. Precambrian Research 170 (1),107–115.
Bell, D.R., Schmitz, M.D., Janney, P.E., 2003. Mesozoic thermal evolution of the southernAfrican mantle lithosphere. Lithos 71 (2), 273–287.
Beyer, C., Frost, D.J., Miyajima, N., 2015. Experimental calibration of a garnet–clinopyroxene geobarometer for mantle eclogites. Contributions to Mineralogy andPetrology 169 (2), 1–21.
Boyd, F.R., Danchin, R.V., 1980. Lherzolites, eclogites and megacrysts from some kimber-lites of Angola. American Journal of Science 280 (2), 528–549.
Boyd, F.R., Gurney, J.J., 1986. Diamonds and the African lithosphere. Science 232 (4749),472–477.
Boyd, S.R., Kiflawi, I., Woods, G.S., 1994. The relationship between infrared absorption andthe A defect concentration in diamond. Philosophical Magazine B 69 (6), 1149–1153.
Boyd, F.R., Pokhilenko, N.P., Pearson, D.G., Mertzman, S.A., Sobolev, N.V., Finger, L.W.,1997. Composition of the Siberian cratonicmantle: evidence from Udachnaya perido-tite xenoliths. Contributions to Mineralogy and Petrology 128 (2–3), 228–246.
Campeny, M., Mangas, J., Melgarejo, J.C., Bambi, A., Alfonso, P., Gernon, T., Manuel, J., 2014.The Catanda extrusive carbonatites (Kwanza Sul, Angola): An example of explosivecarbonatitic volcanism. Bulletin of Volcanology 76 (4), 1–15.
Cartigny, P., 2005. Stable isotopes and the origin of diamond. Elements 1 (2), 79–84.Cartigny, P., Harris, J.W., Javoy, M., 2001. Diamond genesis, mantle fractionations and
mantle nitrogen content: a study of δ13C–N concentrations in diamonds. Earth andPlanetary Science Letters 185 (1), 85–98.
Chambel, L., Caetano, L., Reis, M., 2013. One century of Angolan diamonds. (Retrievedfrom) https://xmbl.files.wordpress.com/2013/10/one-century-of-angolan-diamonds-hd2.pdf (on December 10th, 2015).
Chaves, M.L.S.C., Benitez, L., Brandão, P.R.G., Girodo, A.C., 2008. Kimberlito Canastra-1 (SãoRoque de Minas, MG): geologia, mineralogia e reservas diamantíferas. Revista Escolade Minas 61 (3), 57–364.
de Boorder, H., 1982. Deep-reaching fracture zones in the crystalline basement surround-ing the West Congo System and their control of mineralization in Angola and Gabon.Geoexploration 20 (3), 259–273.
de Waele, B., Johnson, S.P., Pisarevsky, S.A., 2008. Palaeoproterozoic to Neoproterozoicgrowth and evolution of the eastern Congo Craton: its role in the Rodinia puzzle.Precambrian Research 160 (1), 127–141.
deWit, M.C., Jelsma, H.A., 2015. A Review of the Kimberlites of the Democratic Republic ofCongo. In: deWit, M.F., Gauillocheau, deWit, M.C.J. (Eds.), Geology and Resource Po-tential of the Congo Basin. Springer-Verlag, Berlin, pp. 361–369.
Deines, P., 1980. The carbon isotopic composition of diamonds: relationship to diamondshape, color, occurrence and vapor composition. Geochimica et Cosmochimica Acta44 (7), 943–961.
Dietrich, C., 2000. Inventory of formal diamondmining in Angola. Angola'sWar Economy:The Role of Oil and Diamonds, p. 151.
Egorov, K.N., Roman’ko, E.F., Podvysotsky, V.T., Sablukov, S.M., Garanin, V.K., D’yakonov,D.B., 2007. New data on kimberlite magmatism in southwestern Angola. Russian Ge-ology and Geophysics 48 (4), 323–336.
El Fadili, S., Demaiffe, D., 1999. Petrology of eclogite and granulite nodules from the MbujiMayi kimberlites (Kasai, Congo): the significance of kyanite-omphacite intergrowths.In: Gurney, J.J., Gurney, J.L., Pascoe, M.D., Richardson, S.H. (Eds.), The J.B. Dawson Vol-ume, Proceedings of the VIIth International Kimberlite Conference. Red Roof Design,Cape Town, pp. 205–213.
Gibson, S.A., Thompson, R.N., Leonardos, O.H., Dickin, A.P., Mitchell, J.G., 1999. The sec-tored extent of plume-lithosphere interactions during continental flood-basalt gene-sis: geochemical evidence from Cretaceous magmatism in southern Brazil.Contributions to Mineralogy and Petrology 137 (1–2), 147–169.
Grütter, H.S., Gurney, J.J., Menzies, A.H., Winter, F., 2004. An updated classification schemefor mantle-derived garnet, for use by diamond explorers. Lithos 77 (1), 841–857.
Gurney, J.J., Helmstaedt, H.H., Le Roex, A.P., Nowicki, T.E., Richardson, S.H., Westerlund,K.J., 2005. Diamonds: crustal distribution and formation processes in time andspace and an integrated deposit model. Economic Geology 100, 143–177.
Hansen, S.E., Nyblade, A.A., Julia, J., 2009. Estimates of crustal and lithospheric thickness insub-Saharan Africa from S-wave receiver functions. South African Journal of Geology112 (3–4), 229–240.
Heaman, et al., 2003. ages of kimberlite magmatism in NA.Helmstaedt, H.H., Gurney, J.J., 1995. Geotectonic controls of primary diamond de-
posits: implications for area selection. Journal of Geochemical Exploration 53(1), 125–144.
Jacob, D.E., 2004. Nature and origin of eclogite xenoliths from kimberlites. Lithos 77 (1),295–316.
Jelsma, H., Barnett, W., Richards, S., Lister, G., 2009. Tectonic setting of kimberlites. Lithos112, 155–165.
Jelsma, H., Krishnan, U., Perritt, S., Preston, R., Winter, F., Lemotlo, L., van der Linde, G.,Armstrong, R., Phillips, D., Joy, S., Costa, J., 2013. Kimberlites from Central Angola: acase study of exploration findings. Proceedings of 10th International KimberliteConference 1. Springer, India, pp. 173–190.
Kennedy, C.S., Kennedy, G.C., 1976. The equilibrium boundary between graphite anddiamond. Journal of Geophysical Research 81, 2467–2470.
Klein-BenDavid, O., Izraeli, E.S., Hauri, E., Navon, O., 2007. Fluid inclusions in diamondsfrom the Diavik mine, Canada and the evolution of diamond-forming fluids.Geochimica et Cosmochimica Acta 71 (3), 723–744.
Klein-BenDavid, O., Pearson, D.G., Nowell, G.M., Ottley, C., McNeill, J.C., Cartigny, P., 2010.Mixed fluid sources involved in diamond growth constrained by Sr–Nd–Pb–C–N iso-topes and trace elements. Earth and Planetary Science Letters 289 (1), 123–133.
Kopylova, M.G., Navon, O., Dubrovinsky, L., Khachatryan, G., 2010. Carbonatitic mineralo-gy of natural diamond-forming fluids. Earth and Planetary Science Letters 291 (1),126–137 (The influence of the Glennie domain deep structure on the diamonds inSaskatchewan kimberlites. Geologiya i Geofizika, 38, 481–491).
Kopylova, M.G., Beausoleil, Y., Goncharov, A., Burgess, J., Strand, P., 2016. Spatial distribu-tion of eclogite in the Slave cratonic mantle: The role of subduction. Tectonophysics672, 87–103.
Leahy, K., Taylor, W.R., 1997. The influence of the Glennie domain deep structure on thediamonds in Saskatchewan kimberlites. Geologiya i Geofizika 38 (2), 451–460.
Linol, B., deWit, M.J., Guillocheau, F., Robin, C., Dauteuil, O., 2015. Multiphase phanerozoicsubsidence and uplift history recorded in the Congo basin: A complex successorbasin. In: de Wit, M.F., Gauillocheau, de Wit, M.C.J. (Eds.), Geology and ResourcePotential of the Congo Basin. Springer-Verlag, Berlin, pp. 213–227.
Lucazeau, F., Armitage, J., Kabongo, É.K., 2015. Thermal Regime and Evolution of theCongo Basin as an Intracratonic Basin. In: de Wit, M., Gauillocheau, F., de Wit, M.C.J.(Eds.), Geology and Resource Potential of the Congo Basin. Springer-Verlag, Berlin,pp. 229–244.
McCallum,M.E., Huntley, P.M., Falk, R.W., Otter, M.L., 1994. Morphological, resorption andetch feature trends of diamonds from kimberlite populations within the Colorado–Wyoming state line district, USA. In: Meyer, H.O.A., Leonardos, O.H. (Eds.), Diamonds:Characterization, Genesis and Exploration. CPRM Special Publication 1/B, pp. 78–97.
McLean, H., Banas, A., Creighton, S., Whiteford, S., Luth, R.W., Stachel, T., 2007. Garnetxenocrysts from the Diavik mine, NWT, Canada: Composition, color, and paragenesis.The Canadian Mineralogist 45 (5), 1131–1145.
Mendelssohn, M.J., Milledge, H.J., 1995. Geologically significant information from routineanalysis of the mid-infrared spectra of diamonds. International Geology Review 37(2), 95–110.
Miller, C.E., Kopylova, M.G., Ryder, J., 2012. Vanished diamondiferous cratonic root be-neath the Southern Superior province: evidence from diamond inclusions in theWawa metaconglomerate. Contributions to Mineralogy and Petrology 164 (4),697–714.
Mveumba Ntanda, F., Moreau, J., Meyer, H.O.A., 1982. Particularites des inclusionscristallines primaires des diamants du Kasai, Zaire. Canadian Mineralogist 20,217–230.
Nakamura, D., 2009. A new formulation of garnet–clinopyroxene geothermometer basedon accumulation and statistical analysis of a large experimental data set. Journal ofMetamorphic Geology 27 (7), 495–508.
Navon, O., 1999. Diamond formation in the Earth's mantle. In: Gurney, J.J., Gurney,J.L., Pascoe, M.D., Richardson, S.H. (Eds.), The J.B. Dawson Volume, Proceedingsof the VIIth International Kimberlite Conference. Cape Town, Red Roof Design,pp. 584–604.
Nikitina, L.P., Korolev, N.M., Zinchenko, V.N., Felix, J.T., 2014. Eclogites from the uppermantle beneath the Kasai Craton (Western Africa): Petrography, whole-rock geo-chemistry and U Pb zircon age. Precambrian Research 249, 13–32.
Nimis, P., Taylor, W.R., 2000. Single clinopyroxene thermobarometry for garnet peri-dotites. Part I. Calibration and testing of a Cr-in-Cpx barometer and an enstatite-in-Cpx thermometer. Contributions to Mineralogy and Petrology 139 (5),541–554.
O’Connor, J.M., Duncan, R.A., 1990. Evolution of the Walvis Ridge-Rio Grande risehot spot system: implications for African and South American plate motions overplumes.
Palot, M., Pearson, D.G., Stern, R.A., Stachel, T., Harris, J.W., 2014. Isotopic constraints onthe nature and circulation of deep mantle C–H–O–N fluids: Carbon and nitrogensystematics within ultra-deep diamonds from Kankan (Guinea). Geochimica etCosmochimica Acta 139, 26–46.
Pearson, D.G., Wittig, N., 2008. Formation of Archean continental lithosphere and its dia-monds: the root of the problem. Journal of the Geological Society 165 (5), 895–914.
Pereira, E., Rodrigues, J., Reis, B., 2003. Synopsis of Lunda geology, NE Angola: implicationsfor diamond exploration. Comunicações do Instituto Geológico e Mineiro 90,189–212.
Pollack, H.N., Chapman, D.S., 1977. On the regional variation of heat flow, geotherms, andlithospheric thickness. Tectonophysics 38, 279–296.
Robles-Cruz, S.E., Escayola, M., Jackson, S., Galí, S., Pervov, V., Watangua, M., Gonçalves, A.,Melgarejo, J.C., 2012a. U–Pb SHRIMP geochronology of zircon from the Catocakimberlite, Angola: implications for diamond exploration. Chemical Geology 310,137–147.
Robles-Cruz, S.E., Melgarejo, J.C., Galí, S., Escayola, M., 2012b. Major-and Trace-ElementCompositions of Indicator Minerals that Occur as Macro- and Megacrysts, and of Xe-noliths, from Kimberlites in Northeastern Angola. Minerals 2 (4), 318–337.
Russell, J.K., Dipple, G.M., Kopylova, M.G., 2001. Heat production and heat flow in themantle lithosphere, Slave craton, Canada. Physics of the Earth and Planetary Interiors123 (1), 27–44.
14 C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
Schärer, U., Corfu, F., Demaiffe, D., 1997. U–Pb and Lu–Hf isotopes in baddeleyite and zir-con megacrysts from the Mbuji-Mayi kimberlite: constraints on the subcontinentalmantle. Chemical Geology 143 (1), 1–16.
Schulze, D.J., 1989. Constraints on the abundance of eclogite in the upper mantle. Journalof Geophysical Research - Solid Earth 94 (B4), 4205–4212.
Sebagenzi, M.N., Vasseur, G., Louis, P., 1993. First heat flow density determinations fromsoutheastern Zaire (central Africa). Journal of African Earth Sciences (and the MiddleEast) 16 (4), 413–423.
Shirey, S.B., Richardson, S.H., 2011. Start of the Wilson cycle at 3 Ga shown by diamondsfrom subcontinental mantle. Science 333 (6041), 434–436.
Shirey, S.B., Harris, J.W., Richardson, S.H., Fouch, M., James, D.E., Cartigny, P., Deines, P.,Viljoen, F., 2003. Regional patterns in the paragenesis and age of inclusions in dia-mond, diamond composition, and the lithospheric seismic structure of SouthernAfrica. Lithos 71 (2), 243–258.
Smart, K.A., Chacko, T., Stachel, T., Muehlenbachs, K., Stern, R.A., Heaman, L.M., 2011. Di-amond growth from oxidized carbon sources beneath the Northern Slave Craton,Canada: a δ13C–N study of eclogite-hosted diamonds from the Jericho kimberlite.Geochimica et Cosmochimica Acta 75 (20), 6027–6047.
Smit, K.V., Stachel, T., Stern, R.A., 2014. Diamonds in the Attawapiskat area of the Superiorcraton (Canada): evidence for a major diamond-forming event younger than 1.1 Ga.Contributions to Mineralogy and Petrology 167 (1), 1–16.
Smith, E.M., Kopylova, M.G., Nowell, G.M., Pearson, D.G., Ryder, J., 2012. Archean mantlefluids preserved in fibrous diamonds from Wawa, Superior craton. Geology 40 (12),1071–1074.
Sobolev, V.S., Sobolev, N.V., 1980. New proof on very deep subsidence of eclogitized crust-al rocks. Doklady Akademii Nauk SSSR 250 (3), 683–685.
Stachel, T., Harris, J.W., 2008. The origin of cratonic diamonds—constraints from mineralinclusions. Ore Geology Reviews 34 (1), 5–32.
Stachel, T., Harris, J.W., Aulbach, S., Deines, P., 2002. Kankan diamonds (Guinea) III; δ13Cand nitrogen characteristics of deep diamonds. Contributions to Mineralogy andPetrology 142, 465–475.
Stachel, T., Harris, J.W., Muehlenbachs, K., 2009. Sources of carbon in inclusion bearing di-amonds. Lithos 112, 625–637.
Stern, R.A., Palot, M., Howell, D., Stachel, T., Pearson, D.G., Cartigny, P., Oh, A., 2014.Methods and reference materials for SIMS diamond C- and N-isotope analysis.Canadian Centre for Isotopic Microanalysis, Research Report 14-01. University ofAlberta, Education and Research Archive (Retrieved from https://era.library.ualberta.ca/files/cc08hf71g, December 10th, 2015).
Tack, L., Wingate, M.T.D., De Waele, B., Meert, J., Belousova, E., Griffin, B., Tahon, A.,Fernandez-Alonso, M., 2010. The 1375 Ma “Kibaran event” in Central Africa: Promi-nent emplacement of bimodal magmatism under extensional regime. PrecambrianResearch 180 (1), 63–84.
Taylor, W.R., Jaques, A.L., Ridd, M., 1990. Nitrogen-defect aggregation characteristics ofsome Australasian diamonds: time-temperature constraints on the source regionsof pipe and alluvial diamonds. American Mineralogist 75, 1290–1310.
Torsvik, T.H., Burke, K., Steinberger, B., Webb, S.J., Ashwal, L.D., 2010. Diamonds sampledby plumes from the core–mantle boundary. Nature 466 (7304), 352–355.
Weiss, Y., Griffin, W.L., Elhlou, S., Navon, O., 2008. Comparison between LA-ICP-MS andEPMA analysis of trace elements in diamonds. Chemical Geology 252 (3), 158–168.
Weiss, Y., McNeill, J., Pearson, D.G., Nowell, G.M., Ottley, C.J., 2015. Highly saline fluidsfrom a subducting slab as the source for fluid-rich diamonds. Nature 524 (7565),339–342.
Wiggers de Vries, D.F., Bulanova, G.P., De Corte, K., Pearson, D.G., Craven, J.A., Davies, G.R.,2013a. Micron-scale coupled carbon isotope and nitrogen abundance variations in di-amonds: Evidence for episodic diamond formation beneath the Siberian Craton.Geochimica et Cosmochimica Acta 100, 176–199.
Wiggers de Vries, D.F., Pearson, D.G., Bulanova, G.P., Smelov, A.P., Pavlushin, A.D., Davies,G.R., 2013b. Re–Os dating of sulphide inclusions zonally distributed in single Yakutiandiamonds: evidence for multiple episodes of Proterozoic formation and protractedtimescales of diamond growth. Geochimica et Cosmochimica Acta 120, 363–394.
15C.W. Kosman et al. / Lithos xxx (2016) xxx–xxx
Please cite this article as: Kosman, C.W., et al., Cretaceousmantle of the Congo craton: Evidence frommineral and fluid inclusions in Kasai alluvialdiamonds, Lithos (2016), http://dx.doi.org/10.1016/j.lithos.2016.07.004
