sign in sign up
<<
Home , Phlogopite

American Minerulogist, Volume 69, pages 237-249, 1984

Kimberlitic chlorites from Sierra Leone, West Africa: unusual chemistriesand structural polytypes

LINoe A' TolrPrtNsl

Geology Department, University of Massachusetts Amherst. Massachusetts 01003

S. W. BelleY

Department of Geology and Geophysics (Jniversity of Wisconsin, Madison, Wisconsin 53706

nNo STePHBNE' HeccpnrY

Geology Department, (Iniversity of Massachusetts Amherst. Massachusetts 01003

Abstract

Chloritesoccur as ovoid discretenodules (l-4 cm), as anhedralmacrocrysts (0'20 mm) and euhedral crystals (0.05 mm) in , and as cleavagereplacement products of high Fe (reversely pleochroic), low Ti + cr phlogopites.The nodules are severely deformed and four textural groups are recognized (fibrous, platy, tectonized, multiple )that consist of intergrown chlorites and vermiculites of variable compositions. Major chemicaltypes in the nodulesare classified as highFe (25 wt.%), intermediateFe (15 wt.%),and low Fe (8wt.%) varietieswith a subgroupconsisting of low Fe and high K(0.8- 2.3 wt.VoKzO). The chlorites have high SiO2(46.6 wt.Vomax) and MgO (28 wt.% max), and low Al2O3 g-12 wt.%) contents. High Fe3* and later alteration result in vacancies (corroborated by electron densities)in the interlayer sheet. The chlorites and vermiculites are dominantly of the rare la structural type, intergrown with more typical IIb chlorite in somespecimens. Two-layer "s" and "t" stackingsequences identified in Ia chloritesand vermiculites suggestalteration from parent lM andlMl phlogopites,respectively. Chloriti- zation of peridotitic phlogopite nodules and macrocrysts took place at relatively low P-T conditions concurrent with, or followed by crystallization of primary magmaticchlorite in the kimberlite groundmass.

Introduction the upper mantle, and phlogopite (Prinz et al.' 1975; Sobolev,1977), (Giardini etal.,1974)' and type ID Chlorite is a common constituent of many rock types chlorite (Mitchell and Giardini, 1977)havebeen reported. and a characteristic component of low grade greenschist Nixon and Kreston (1973) describe green kimberlitic faciesmetamorphism. It is, however,stable over a range chlorite and phlogopite cored by chlorite in the Khabes of temperature, pressure, and volatile conditions (Albee, dike, Lesotho. A green pleochroic chlorite occurs along 1962).Primary igneous chlorite occurs in some diabases the outer three meters of a kimberlite from the De Beers (Lehmann, 1965)and Fawcett (1965)reports chlorite in Mine (Fergusonet al., 1973).These may be late-stage residualliquids from basaltsin Mull. Quenchedchlorite is alteration products of phlogopite associated with, or noted by Schreyer and Yoder (1964)in experiments at 5 resulting from fluidization and kimberlite emplacement and l0 kbar at temperaturesabove the cordierite solidus (Kreston, 1973).McGetchin et al. (1973)have character- in the system Mg-cordierite-H2o. inclusion ized three pleochroic types of chlorites resulting from studies support the notion of hydrous phyllosilicates in variations in Cr2O3content from the Moses Rock kimber- lite andthe CaneValley carbonatitedikes, Utah. Diopsid- pyroxenes spinel lherzolite are considered to be I Presentaddress: B. P. ,75 Rua Martin Ferreira, ic from via reactions with a Boto Fogo, Rio de Janeiro, Brazil. Reprint requeststo S. E. the precursors to these chlorites Haggerty. volatile-rich fluid phase during kimberlite-carbonatiteas- m03-004r84/0304-0237$02.00 237 238 TOMPKINS ET AL.: KIMBERLITIC CHLORITES

cent. Chlorite replacement of phlogopite along cleavage igneous is used for minOral melt derivatives whereas replace- planes is also described from the Mir pipe in Siberia ment (or secondary)implies a solid state reaction derived from (Frantsesson, 1970)) and this appears to be directly deuteric processesor post-consolidationactivity. associatedwith carbonatitic activity in the pipe. Chlorite was first recognized as a constituent of the Optical techniques Koidu kimberlite complex in Sierra Leone by Grantham Chlorite nodules were examined and classifiedin hand speci- and Allen (1960),but few details are provided. Systematic men and under a binocular microscope. Thick polished mounts samplingof the complex (Tompkins, 1983;Tompkins and were found to be more useful than polished-thin-sectionsin Haggerty, l983a,b) now shows that chlorite is abundant revealing subtleties in cross fiber relations. Transmitted and and is present in three modes: as large (up to 4 cm) reflectedlight microscopy was used for the study of . discrete megacrystic nodules, as a primary crystalline Chemistry component of the kimberlite, and as a replacementprod_ uct of phlogopite. Chlorites most similar to these are Chemical analyses of assemblagesand intergrowths reported from only one other kimberlite locality, the were determinedusing an ETEC automatedelectron microprobe Mbujimayi complex inZaire (Fieremansand Ottenburgs, at the University of Massachusetts.Accelerating vOltagewas 15 kV, with a 1979). The analytical data, however, are not directly beamdiameter of 5 pm and a specimencurrent of 0.03 nA. correction factors are those applicable because large concentrations of quartz are of Bence and Albee (1968)and Albee and Ray (1970). Synthetic (s) and narural (n) presentin bulk chemicalanalyses. The presentstudy was standardsused are as followed: pyrope garnet, p-130 (n) for Fe, undertakento characterizethe Koidu chlorites, composi- Al; Mn-ilmenite (n) for Ti, Mn; Foq, p-130 (n) for Mg; kyanite tionally and structurally, in an attempt to define more (n) for Si; chromite, Cr-cats (s) for Cr; jadeite, Jd-35 (s) for Na, accurately their genesisand petrological affinities, and to Ca; sanidine(n) for K. establish whether chlorite is a viable mineral host for Iron was determined by microbeam analysis but ferrous iron volatiles in the upper mantle. was determined on discrete nodules by wet chemical analyses following the Wilson cold digestion procedure as outlined by Geological setting Maxwell (1968). Triplicate analyses were compared with two standards,BCR-I and Knippa-l. These standards The Cretaceous-Jurassickimberlite field in the Leo are routinely used in XRF analysesat the University of Massachusetts;the uplift of eastern Sierra Leone is comprised of three USGS analysesare given in Flanagan(1976). Deviations of the kimberlite pipes, a ring-dike structure, and multiple sets standardsfrom the mean were no greater than +0.10 wt.Vo.The of en-echelonkimberlite dikes intruded into Archean age wet chemical ferrous iron determinations of chlorite nodules granitic rocks (Hall, 1970).This district forms part of a represent bulk compositions. Ferrous and ferric iron contents larger kimberlite province that includes other bodies in were calculatedfrom weighted averagesbased on chlorite modal Liberia, Guinea, Mali, and the Ivory Coast (Bardet, analysesin those nodules having more than one compositional 1974). The pipes at Koidu are multiple intrusions and rype. severalfacies of kimberlite are recognized(Tompkins and X-ray crystallography Haggerty, l983a,b). Kimberlites at pipe I are typically coarse grained and brecciated, whereas those at pipe 2 Four chlorite nodules were studied by X-ray powder and are finer grained and have carbonatitic affinities basedon single crystal techniques. Samples CHK-I and CHK-6 were major and minor element bulk chemistries (Tompkins, selected as representative of the fibrous chlorites and CHK-2 1983). The dikes are massive and distinctly basaltic in and CHK-13 of the platy and tectonized chlorites. Gandolfi patterns were taken appearance.Chlorite nodules have thus far been recog- of individual fibers and plates as well as Debye-Scherrer patterns using crushed material in nized only at Pipe 2. Primary groundmasschlorite, and capillary tubes with FeKo radiation. From three to six individual fibers or chlorite replacement of phlogopite is best developed in plates from each nodule were also studied by precession and the dikes and is less prevalent in the diatreme facies. All Weissenberg methods to determine the precise stacking se- kimberlite facies are diamond-bearing and the mantle- quence of layers. The 00/ intensities for crystals from samples derived nodule suite is composedofeclogites and discrete CHK-I and CHK-2 were measuredon a single crystal ditrrac- megacrysts of Mg-ilmenite, sub-calcic clinopyroxene, tometer, corrected for Lp and absorption factors, and used to and pyrope-rich garnet. calculate one-dimensionalelectron density maps.

Analytical procedures Chlorite occurrenceand mineral associations Terminology Discrete chlorite nodules Noduleis usedin the sensegenerally applied to polymineralic xenolithshosted in kimberlite,and megacrysts are large (usually The discretechlorite nodules(Fig. l) are similar to >2 cm) discretenodules of a singlemineral. Because of the othermembers of the megacrystsuite (pyroxene, garnet, geneticimplications implied by phenocrystsand xenocrysts these ilmenite)at Koidu with respectto size(l-4 cm)and their termsare only used in specificinstances and the \ermmacrocryst oval and abraidedappearance. In hand specimenthe is employedwhere there are uncertaintiesin origin. primary nodulesare dark forest greenin color with veinletsof TOMPKINS ET AL.: KIMBERLITIC CHLORITES 239

averaging:0.05 mm in cross-section.They occur either as disseminated crystals in the carbonatitic-kimberlite groundmassor as aggregateplatelets (Fig. 3b) in carbon- ate rich pockets of less carbonated kimberlite. In some cases the platelets are optically zoned (Fig. 3a). These chlorites as well as the others are strongly pleochroic in deep blue-green with abnormal blue interference colors and a small negative2V(-151. Occasionaldecomposition of macrocrystic chlorite results in the formation of admix- tures of goethite + hematite * complex intergrowths of finely disseminatedcalcite and silicate minerals (Fig. 3c). Expansionof the sheetsilicate and swelling and distortion Fig. 1. A representativecollection of the four discrete of the chlorite takes place. chloritenodule groups identified at Koidu. Texturaltypes from left to rightare: back row, all groupF; frontrow, groupsT, P, C; Replacement chlorite center,group F. Reversely (R) pleochroic phlogopites (a > y, where a : tan, and 7 : colorless) form cores to normally (N) pleochroic phlogopites (a 1 ^1,where a = colorless and 7 mustard yellow serpentine. Their habit is variable and : tan in color). The early phlogopites(R) are preferential- four major groups are recognized. Group F (fibrous) ly replaced by chlorite along cleavage planes (Fig. 3d). chlorites are cohesive and consist of nodules with a The phlogopites are consistently >0'1 mm in longest filiform to bladed habit. Group P (platy) are characterized dimension,and phlogopites(R) similar to thesehave been by chlorites that form books or plates typical ofphyllosili- described from a number of kimberlites in South Africa cate minerals. Group T (tectonizefi are characterizedby (Farmerand Boettcher, 1981). single distinct parting planes. Textures are platy to fi- brous and many nodules are bent or multiply contorted. Mineral chemistrY Group C (multiple cleavage) are composed of nodules Representativeaverages of electron microbeam analy- with two distinct parting directions. Examples of a selec- ses of the Koidu chlorites are listed in Tables 1-3. tion of thesegroups are illustrated in Figure 2a-f. Anasto- Previously published kimberlitic chlorite compositions mosing networks of crosscutting fibers (Fig. 2a-d) are are also presentedand have been recalculatedbased on a present in some nodules and these crossfibers form theoretical 28 oxygen formula cell. In most of the Koidu distinct subparallel groups. Typically, one set of cross- chlorites studied, two, three, or in some cases four fibers is observed but two sets may also be present (Fig. distinct compositions were determined. Some of the 2d). Compositionally, these fine cross-fibers are usually componentsare evident in the photomicrographsin Fig- similar to the host phase. In Figure 2d, however, hard- ure 2 but others are not illustrated and are on a much finer ness, reflectivity, and chemical differences are evident scale. and the harder and darker cross-fibers are selectively replacedby serpentine(highly reflective phase).In Figure 2e-f the flame-like structures are two compositionally Major elements distinct chlorites intergrown along the same cleavage The unusual chemistry of the chlorites is distinguished planes. Kink banding perpendicular to the cleavagedis- by their low Al2O3 (8-12 wt.%), high SiOz Q8.97-41.65 rupts this parallism giving the appearanceof cross-cutting wt.Vo), and MgO (19.0-28.0 wt.Vo) values coupled with fibers. These textures are interpreted to result from high Fe (up to 29.5 wt.Voas FeO). deformation and rehealing along brittle planes Nodules (CHK-1,2, and 3) Iyith two compositional and are very similar to those described by Boyd and types in groups F, D, and T (Table l) have hosts with low Nixon (1978)for lherzolitic phlogopites from kimberlites FeOt 1:7 wt.%),high SiOr1:4lwt.Vo), andAl2O3 averag- in South Africa. ing 11.5 wt.Vo. This compositional type is referred to as low iron chlorite. By contrast, the high iron chlorites Macrocrysts and groundmass chlorite average21.5 wt.% FeOr, are lower in SiO2(31'7 wt.%o), When calcite is the major matrix mineral, groundmass and lower in Al2O3 $.6 wt.%). The average MgO con- chlorite is found with only minor modal amounts of Ti, tents in both the high and low iron chlorites is 26 wt.%io. Cr-rich phlogopite. In those kimberlites where matrix Nodule CHK-7 (group F) has three compositionaltypes serpentineis predominant, however, groundmassTi, Cr- (Table 2), a high iron chlorite host (the inverse of CHK-1' rich phlogopite is abundant and chlorite is sparse or 2, and 3), a low Fe chlorite (8.7 wl% FeOr), and a absent. Large (0.25-0.5 cm) chlorites are anhedral and chlorite of intermediate FeOr 03.1 wt.%) content. Alu- xenocrystic in appearancebut finer grained groundmass minum and SiOz vary sympathetically and MgO ranges chlorites are presentas perfect euhedralplatelets (Fig. 3a) from 24-28 wt.Vo. 2N TOMPKINS ET AL.: KIMBERLITIC CHLORITES

Fig. 2. Reflected light photomicrographs of discrete nodules in polished sections illustrating the variety of observed textural forms. a) CHK-I. Anastomosingnetwork of cross-cuttingfibers (dark phase).Both the host chlorite and cross-fiberchlorite are low Fe chlorites (analysisA, Table 1). b) A coarser cross-fiberfrom CHK-1. c) CHK-7. The host phase is the lighter and softer phase. This phaseis high in Fe (analysisC, Table 2). Cross-fibersare low in Fe (analysesA and B, Table 2), and are easily distinguishedin this nodule by its greater hardnessand lower reflectivity. The highly reflective phaseis serpentinereplacing the low Fe cross-fibers. d) CHK-7. This photomicrographis similar to C except that the cross-fibersare completely replaced by serpentine.e) CHK-I. Well formed kink bands run approximately N-S perpendicular to the cleavagegiving the appearenceof cross-cutting fibers. The lighter flameJike phaseis the high Fe chlorite (analysisB, Table 1), the darker flame-likephase is the low Fe chlorite (analysisA, Table l). f) CHK-I. Sameas e, exceptthat fine cross-cuttingfibers are visible. Scalebar = 0.2 mm.

Nodule CHK-13 (group T) has four compositional 38, 48, and 9) are given in Table 2, representinggroups components(Table 2) of which three are approximately in F, T, and C. These nodules have not been characterized equal proportions. Here again,low and high iron chlorites in the same detail but the major compositional host are present, but in addition there are two intermediate phases are high, low, or intermediate iron chlorites, iron chlorites having 10 and 15 wt.To FeOy. similar in major element compositions to those described Compositionsof five additionalnodules (CHK-4. lB. above. The replacementchlorites (Fig. 3d) in phlogopite TOMPKINS ET AL.: KIMBERLITIC CHLORITES 241

Table 1. Compositionsof discretechlorite nodules.

CHK-2 s to4 33.51 4L.44 35.98 31.09 4L.2I 4r.29 4L.65 47 0.17 0.15 0.17 0.14 0.18 0. 16 0. 16 0.20 o.2L o.20 o.22 o.22 o.22 ;i;E. 10,89 9.16 r0.46 8.33 11.71 10,02 8.18 11,83 11.56 11.46 11.89 LL.72 1r. 81 0,L2 0.09 0.11 0.07 0.08 0.08 0. 11 o.r2 o.r2 o.l2 0.16 0.16 0.16. 4.oi- ;;?;:' 10.81+ t L.72+ 0.39+ Fe6 t.oJ zo. Jt 2.74 19.01 7.84 2.88 18.83 8.00 8.00 3, 33 6.66 6.97 3.16 Mgo 25.84 2L.9L 24.86 26,86 2s.88 27,AO 26,73 26.30 26.83 26.2r 25.78 26.OO MnO o.o2 0.04 0.03 0,09 0.07 0,08 0,08 0.03 0.20 0.04 0.01 0.00 0.01 CaO 0.15 0.22 o,L7 o.24 0.18 o.2r o.25 0.09 0.10 0.11 0.00 0.09 0.05 NarO 0.01 0.03 0.02 0,02 0.0r- o.02 0,00 0.00 0,o2 0.00 0.00 0.02 0.01 KzO 0.01 0.00 0.01 0.08 0.47 0.28 0.06 0.26 0.79 o.25 0.15 r.45 0. 80 Totel 86.02 91.48 88.49 85.3s 87,86 87. 80 86.56 88.47 88.59 88.93 85,95 87,83 87,83

catlons to 28 Oxygens' 7. 700 s1 7.810 6.772 7,355 b.4)4 t.t>4 6.927 6.468 7.662 7.704 7 ,435 7.797 7.765 T1 0.025 0.023 o,o24 0.023 0.026 0.023 0.025 0.028 0.029 0.028 0.032 0.032 0.031 2.623 2.589 2. 580 AI 2.448 2.L82 2.3t9 2.O7? 2.542 2.273 2.OO7 2.592 2.543 2.498 0.018 0.015 0.016 0.011 0.012 0.013 0.0r8 0.018 0.018 0.018 0.023 0.024 o.023 fil:r 1.530 1.698 0,889 0. 568 fe r.249 4.4s5 0.431 3.363 L.227 o.464 3.277 L.243 L.248 0.514 r.043 1,093 0.490 183 Me 7.345 6.599 6.969 8.468 7.2L8 7.567 8.62L 7.408 7,3r4 7.396 7.3r3 7,203 7. Mn 0.003 0.007 0.005 0.017 0.011 0.013 0,014 0.004 0.031 0.007 0.001 0.000 0,001 Ce 0.030 0.o47 0.034 0.055 0,036 0.043 0,056 0,018 0.020 o.o22 0.000 0,018 0,010 Na 0.005 0.012 0.007 0.008 0.007 0.007 0.000 0,000 0.007 0.000 0.000 0,007 0,003 r 0.002 0.000 0.002 0.o22 0.112 0.068 0.016 0.061 0.188 0.059 0.036 0.347 0, 189 lotal 18.935 20.113 r8.692 20.498 18.985 19.096 20.502 19.034 79.tO2 18.865 18.858 19,078 L8.778

#76 77 344 48

27525 50 50 10882 50 50 0.300 Tet, Al 0.190 L.228 0.645 L.546 0.246 1.073 r.532 0.338 0.296 0.565 o.203 0.23s 2.354 2.280 Oct. AL 2.25E 0.954 L.674 0.531 2.316 r.200 o.475 2.254 2,247 1.933 2,420

(x) eeuple nuber and textural type. * Based on the ulit cell asamlng theoretlcal O(oH) content + Besed on the dlffereace obtalned by vet chsical analysls for Feo T NuEber of aElysea averaged together I Estluted rcda1 percent ueed ln bulk atraly8is coEputatlon as well as the groundmasschlorites (Fig, 3a-b, Table 3) samples;a few of the latter are enriched in Mg and are are high Fe compositional types. slightly removed from the arcuate trend displayed by the Consideringthe bulk compositions of the entire group, remaining groups. the three chemical chlorite tlpes may now be defined as low (7-10 wt.VoFeOr), intermediate (12-14 wt.% FeOr), Minor elements and high (17-20 wt.%) iron varieties. The Koidu chlorites have minor concentrationsof MnO In order to maintain charge balance, a large amount of (0.09wl% max), Cr2O3(0.51wt.Vo max), and TiO2(0.14- ferric iron is required for the higher Fe chlorite varieties 0.43wt.%). There is a tight cluster of the latter two oxides in the absenceof major concentrations of other trivalent around0.1 wt.% CrzOr and 0.2 wt.% TiOz Gie' 5). The cations. Even though the low Fe varieties contain higher few anomalously high contents are associated with octahedralAl. considerableFe3* or octahedralvacancies groundmasschlorites and one ofthe chlorite nodules.The are necessaryassuming a formula fit to theoretical chlo- minor element concentration of the Koidu chlorites are rite. The low Fe varieties, however, yield a reasonable markedly different from those of the Utah specimens, cation total to a theoretical formula of 22 which are high chromium chlorites (0.23-3'90 wt.% oxygens and these phases may indeed be vermiculite or Cr2O3).The chlorite from the Mir pipe, however is similar interstratified vermiculite/phlogopite(i.e., CHK- I 3, anal- to the Koidu sampleshaving O.26wt.Vo Cr2O3,and 0.32 ysis D) or vermiculite/chlorite. Becauseof probable mul- wt.VoTiO2. ticomponent complexities in these phasesthe theoretical 28 oxygens is used. This is a reasonableassumption for K, and Ca most chlorites (Foster, 1962)but it is possible that some Na, oxygen has replaced (OH); this increasesthe cation total Sodium contents are minimal in all kimberlitic chlorites but decreases tetrahedral Al which in many cases is with a maximum of 0.07 wt.VoNa2Ofor those at Koidu to already fairly low. Variations in Al, Fe2*, and Mg are 0.03 wt.% Na2O for the Utah and Siberia samples. shown in Figure 4 and are compared to chlorites from Potassium,however, is present in concentrationsof up to other kimberlites. A well-definedtrend is apparentfor the 2.31 wL% KzO, and CaO contentsup to 0.25 wt-Vowere Koidu and Mir chlorites and for some of the Utah detected for Koidu. The modal proportions of the high 242 TOMPKINS ET AL.: KIMBERLITIC CHLORITES

'Iable 2. Compositionsof discrete chlorite nodules.

x cnK_7(F) cnK-13(T) cHK-4(c) CHK-IB(F) cHK-3B(r) CUK-48(c) CHK-9(F) A.EI-BULKAgCDBI'LK sro^ 41.00 37.73 2s,b4 1l-tt za-sz trlgr arJs tglzo 3-,83 34.52 3s.11 32.s3 32.12 t9.L2 rroi 0.19 o.r9 0.14 0.16 0,22 o.2o O.2o 0,21 o.2r 0.15 0.20 0.35 0.20 0.21 At^6. L2.02 10.91 9.12 9.18 B.bo 9,95 11.5G 11.16 L0.27 10.17 10.96 9'42 9'32 10.19 crioi O.I5 0,13 O,o8 O.IO. O.O5 0,09 0.08 0,09 O,O8r 0.11 0.12 0.0s 0.I3 0.07 ;;?;: -:--: -:::: -::-- 14.6s- 6.ss' 14.8e+ 7.33+ ts.st+ 1t.66' 3.s6 ' re6 8.67 13.0G 22.31 5.t7 29.55 t5.44 7.76 10.03 10.38 3.28 4.9r 2.90 3,43 4.03 MgO 27.54 27.95 23.79 24,96 19.66 27,sL 25.54 24.66 23.42 21.41 29.29 28.07 27 'O9 25.L3 MnO 0.01 o.0r 0.05 o.o5 o,ol 0.01 o.0o 0.01 o,o1 0,08 0.09 0.05 0.06 0.03 cao 0.07 o.o9 0.14 0.!2 o,I3 o.o9 o.o5 0.15 0.I2 0,17 0.33 0.13 0.I4 0.03 Nd^O 0,00 0.02 o.o1 0.04 o.0o o,oo o.o3 0'02 0.06 0'16 0'06 0.10 0.03 0.00 g.Sf K_6 0,Or O.O2 O.O1 0,Ol O.O1 0.03 0.09 2.3r O,8I 0.01 o.25 -!.02 0.16 i3l"r €=ffi !d5t Efti s-6;Ti frn iffi 6:68 ffi 3ffi eo.e5 8f75- se.r6 6e76 E t6 catlons to 28 oxygens *

sl 7.548 7.L43 5.470 6,412 5.347 6.892 7.8L2 7.s54 7.048 6.534 5.688 6,308 6.2L7 7,7O5 T1 0,027 0.027 0.024 0.024 0.037 0.030 0.028 0.030 0.031 0.022 0,029 0,051 0,029 0.031 AI 2.508 2.434 2.089 2.170 2.222 2.3L4 2,574 2.535 2.381 2.262 2.46L 2. ls3 2,126 2,366 0.022 0.019 0.014 0.016 0.010 0.014 0.012 0.014 0.013 0.016 0,018 0.008 0.021 0.011 2.2rL 1.028 2.rr5 1.051 2.272 2.573 0.587 1.335 2.058 4.072 0.917 5.418 2.548 7,226 1.616 1.708 0.518 0.782 0.471 0,555 0.664 r.E 1,557 7.888 7.740 7,451 6.425 8.091 7.r92 7.083 6,857 7.?IL 8.317 8.112 7,815 ?,318 Mn o.oor o,ool o,oro 0,008 0.002 0.001 0.000 0.001 0.001 0.012 0,015 0.009 0.009 0.005 Ca 0.013 o.or.8 0.033 0.025 0.030 0.019 0.012 0'031 0.025 0.034 0.068 0.027 0.029 0.006 Na o.ooo o.ooo 0.008 o.oo2 0.016 0.000 0.000 0.012 0.007 0.022 0,060 0.022 0.037 0.012 0.002 0.062 0.005 0.002 0,040 K 0.002 o.oos 0.003 0.004 0.003 0.007 -9,9? .9.197. =949: Td]66e IOEEJ ltJr3 19-03 n:461 19Jm 20,510 1t.9i6 18.878 L9.443 t9,1L2 rs:Td 1t.51 T6.438 rEZIS

#5415 10585 ) l5

7201070 3492334

Tet, Al 0.452 0.857 r.530 L.528 r.653 r.108 0.188 0.445 0.952 r.466 L.3L2 r,692 1.783 0.294 Oct. AI 2.155 t,577 0,559 0.642 0.569 r.206 2.345 2.089 r,429 0.795 1,149 0,451 0.343 2.072

(X) silple textur6I type. * Based on the unlt cell aseunlng theoretlcal O(OH) content. + Based on the diffrrence obtalned by wer chehlcal analy€ls for FeO, # NuEber of atralyses averaged together. Z Eetlmted nodal pexcst used in butk analysis conputation

K2Ochlorites are2Vo (CHK-3, groupT, analysisC, Table two subtle groups are present. One group consists ofthe l), 50% (CHK-6, group F, analysisB, Table l), and34Vo high Fe chlorites (i.e., softer lighter phase, Fig. 2c and (CHK-I3, groupT, analysisD, Table2). Theseare all low light flame-like phase Fie. Ze-f) plotting to the left of the Fe chlorites. Both K and Ca are consideredto be incom- clinochlore-serpentinejoin. When ferric iron is consid- patible with the chlorite structure (Pauling, 1930;Foster, ered, thesechlorite compositionsmove towards the "Tri- 1962).It is likely that the K (+Ca) is present eirher in an Di" octahedral region parallel to the join clinochlore- interstratified phlogopite-(chlorite and/or vermiculite) "Di". The secondgroup is definedby the low Fe and high phase or in incompletely replaced phlogopite crystals. Si (+K2O) chlorites(i.e., crossfibers,Fig. 2c and the dark phasesin Fig. 2e-f). Thesecompositions plot towards Combined elemental relations (vermiculite) with ferric iron having less influence. Com- In order to consider all elements on a single two- positions are displaced to the right of these data after dimensionalplot, the compositional space shown in Fig- considerationof Fe3+ in the analyses. ure 6 is useful and is adoptedfrom Robinson et al. (1982). Naturally occuring chlorites with octahedral vacancies All kimberlitic chlorite analyses are plotted including were summarizedby Eggleton and Bailey (1967)and are those from the Cane Valley diatreme, the Moses Rock plotted as a field in Figure 6. Natural dioctahedral chlo- dikes, and the Mir pipe. Ideal trioctahedral ("Tri") rites show a range in Si-R3* with Al being the dominant clinochlore is shown and with increasing substitution of trivalent cation. "Di-Tri" octahedral chlorites show an Al ideal chlorite compositions plot along the clinochlore- enrichment of Si relative to ideal compositions and again amesitejoin (Fig. 6). Ideal chlorite compositions with Al the trivalent cation is predominantly Al. No naturally substitution less than that of ideal clinochlore plot to- occuring "Tri-Di" octahedral chlorites have yet been wards serpentine and away from clinochlore. Data for found (Bailey, 1980).In chlorites,the lower the Si and the kimberlitic chlorites from Utah and Siberia lie along this higher the tetrahedral Al values, the stronger will be the trend. Displacementsfrom this join and towards - resulting interlayer bonding (Foster, 1962).Ideally, octa- spinel (ol-sp) indicate that ferric iron is present. Similar- hedral R3+ cations should equal tetrahedral Al3+ cations, ly, chlorites plotting to the right ofthisjoin have variable but this rarely happens in nature. For the high Fe degrees of octahedral vacancies. The Koidu chlorites chlorites, tetrahedral Al is far in excess of octahedral Al show an enrichment of Si relative to ideal clinochlore and (Tables l-3). In order to maintain charge balance some TOMPKINS ET AL,: KIMBERLITIC CHLORITES u3

Table 3. Compositionsof serpentineand other chlorites. Chlorite structural types

S.rp.ntlnr GroundMaa Chlorllaa X-ray study shows that the layer-interlayer assem- cBI(-r g&.1 lo-8G-17-pZA-B blages are dominantly of the Ia type in all four samples 810. 42.O9 40.90 32.78 33.52 32.94 32,94 31.5r Bailey and Brown 0.07 0,06 .0.19 0,23 0,37 0.19 0,43 examined. This terminology is from lr-6- 0.50 0.52 9.66 9,35 9.59 9,28 g,71 (1962), of 0,00 0.08 0,40 0.04 0.05 0,5l in which the symbol I meansthat the directions rrloi J ::-:: staggerof the octahedral sheetsin the 2:1 layer and the 116 16.II r7.27 18.82 17.89 17.85 uto 38.86 40,87 25,83 26,33 27.03 26,97 21,72 interlayer are the same, and d means that interlayer MnO 0.04 0.02 0.08 0.0t 0.04 0.07 0.05 C.0 0.18 0.15 0,12 0.12 0.19 0.L2 cations project normal to the layers onto tetrahedral N.^O 0.01 0,00 0.00 0.00 0.00 0.00 0.04 x-6 0.03 0,02 0.36 0.01 0.16 0.20 0.04 cationsin the 2:1 layers above and below (Shirozuand Mo -qg. -9.4 ,NA,NANA NA _U- lotel 85.73 85.64 84,25 87.30 89,lr 87,79 E6.38 Bailey, 1965). The close approach of interlayer and tetrahedralcations in the a position leadsto cation-+ation oxY8.n.r Crtlonr to 7 cltlon! to 28 orygme" repulsion and consequentinstability relative to the alter- sr 2.007 1,9s3 6,620 6.785 6.598 6,672 6.s23 Tt 0.003 0,002 0.028 0.035 0.056 0.029 0.057 native b position. This accounts for the fact that Ia A1 0.034 0,029 2.300 2.231 2,265 2.2L7 2,L24 the chlorite C.e. 0.001 0,000 0.013 0,054 0.007 0.010 0,083 chlorites were the least abundant among 303 F.;. specimens included in a survey of the abundances of B.- 0.154 0.123 3.050 2.925 3,154 3.032 3.090 M! 2,754 2.911 r.081 7.948 8,074 8.146 8.369 structural types in nature by Bailey and Brown (1962).On !h 0.001 0.001 0.013 0.011 0.007 0.013 0.009 ce 0.005 0,009 0.034 0,025 0,026 0.041 0,025 the other hand, the Ia structural type is the only one Nd 0.001 0.000 0.000 0.000 0,002 0,000 0.015 K 0.002 0.001 0.093 0.003 0.040 0.053 0.010 found to date in vermiculite, where enhancedstability is Nr q.99,t 0.000 NA NA NA NA NA Torel 4.974 5.029 20.242 20.030 20.230 20.2t3 20.315 gained primarily by having relatively few interlayer cat- 05r 3t317 ions and secondarilyby cation order patterns that achieve Tet. Al 1,380 I.2L4 L,442 r.328 !,477 local chargebalance (Shirozu and Bailey, 1966).Interlay- oct. Al 0,920 1.019 0.854 0.889 0.647 er vacancieswould enhancethe stability of the Ia struc- Publtsh€d data on kiDberlltlc chlorlte6 tural type in chlorite also, and it is noteworthy that Hayes 55710 20 (1970) tentatively concluded that Ia chlorites found in s10. 32,40 31.90 30.60 35.50 31.60 30.48 34.30 r1o; o. 03 o. oo 0,03 0.08 0.03 0,32 0.00 sedimentaryrocks are transitional to vermiculite. This is A1^6- 1r.40 12.I0 15,90 7,07 14.10 4.42 8.70 r-znJ I on r.05 0.23 2.40 0.26 0,00 the case in the Koidu chlorites also, where significant P.-n- 7,3L ;;3-3 2.eo 2.4o 4,40 7,70 4.10 6.81 9.20 numbersof interlayer vacanciesin all of the sampleshave r,tto 35.10 35.00 33.40 34,30 33,90 25.L8 40.30 Hdo 0.02 0.00 0.02 0.I0 0,04 0.06 0,00 been inferred from the chemical data of the preceding Cao 0.00 0,00 0.00 0.00 0.01 3.70 0.50 in Naro 0.00 0.00 0.01 0.03 0.01 0.00 0.00 section and interlayer vacancies have been confirmed Kro 0.02 0.00 0.01 0.36 0.00 0.30 0,00 two of the samples (CHK-I, and CHK-2) based on NIO NA NA NA NA NA 0.13 NA rotal 8:6f E3l3o 95.43 86.31 86.19 82.97 93.00 electrondensity measurements(Fig. 7). Cetlons to 2E oxygenEi Figure 7 was constructed using 28 00/ reflections for si 6.287 6.203 5.946 7.108 6.078 6.445 6.307 Tl 0,005 0.000 0.005 0.012 0.005 0.051 0,000 each crystal after least-squaresrefinement of z-coordi- A1 2.607 2.773 3,64! 1.623 3,265 2.099 1.886 C.* 0.593 0.599 0.152 0.035 0.365 0.043 0.000 nates, cation multiplicities, and isotropic B values. The F"i. r.164 = 0ois very Fe- 0.471 0.390 0.714 r.254 0,660 r,204 1.415 electron density in the interlayer sheet at zlc MS 10.153 10.144 9.672 9,9s5 9.7L9 7.936 1t .045 low (about 13e- in CHK-I and22 e- in CHK-2) relative !h 0.003 0,000 0.004 0.016 0.007 0.010 0.000 Ca 0.000 0.000 0.000 0.000 0.002 0,639 0.098 to that of the octahedralcation plane in the 2 : I layer at zlc Ne 0,000 0.000 0,005 0.012 0.005 0,000 0.000 K 0.002 0.000 0.002 0.089 0.000 0.081 0.000 = 180" (about 44 e- and 46 e-). The densities for the Nl NA NA NA NA NA O,O22 : ToteL 20.III 20.109 20.r5r 20.104 20.106 19.894 20.757 tetrahedral cation plane at zlc 110.9"suggest substitu- Tet. At 1.713 1,797 2.054 0,892 !.922 1.555 1.693 tion of 0.1 atoms of tetrahedral Fe3* per formula unit in oct. Al 0.894 0,976 r.587 0.731 1.343 0,544 0,193 CHK-I and 0.2 atomsin CHK-2. * Besed oo the unit cetl essMlng theoretlcal O(OH) content. The low interlayer cation electron density and the # NsDber of eElyses everaSed togethet, 3r5r7 froo carbonallte-kloberllte dtkes, Cene VaIIey dlatreEe, Mccelchin g! broad peaks in the map of CHK-I suggest that this aI, (1973) 4,6 fron kiDb€rltte, Moses Rock dl.ke (6aDp1e 1416A)' uccetchln g 91, (1973) specimenapproaches true vermiculite, whereasspecimen 10 froo Mlr ptpe' siberler FtenlseE€on' (1970), table 24. This has been confirmed 20 Dtemnd iocluBlon, Hltchell and Glerdlnl (1977). CHK-2 is slightly more chloritic. by heating tists. Piecesof specimensCHK-I, CHK-2, CHK-6, and CHK-13 were ground under acetone, pre- Fe3* must proxy for octahedralAl or some substitution of pared as oriented slides, and heated at 6fi)"C in a muffie (OH)- for 02- may take place. There is probably an furnace for one hour. X-ray powder diffractograms ob- excessof Fe3+beyond the l:1 replacementof Al3*, and tained in a dry air environment showed that the l4A Fe3* will, therefore, replace octahedral divalent cations spacings of the untreated samples had collapsed com- in a ratio of 2:3.In the low Fe chlorites,octahedral Al is pletely to l0A in specimensCHK-I and CHK-6 and about greater than tetrahedral Al and the replacement of R2+ 6}VohaAcollapsedto l0A in specimensCHK-2 and CHK- cations in the 2:3 ratio is dominatedby Al3*; ferric iron is 13. The complete collapse of specimensCHK-I and the major substituting ion in the high Fe chlorites. In CHK-6 relative to the two other Koidu specimensis in either case, octahedral vacanciesare required. accord with the differencesobserved in their 00/ intensi- 244 TOMPKINS ET AL.: KIMBERLITIC CHLORITES

Fig. 3. a) Euhedral groundmasschlorite from section KD-80-5. Microprobe analysesof this grain are given in Table 4. Transmitted light, air objective. b) Stacked chlorite platelets in groundmasskimberlite from section KD-80-5. Transmitted light, air objective. c) Chlorite (center) alteration to goethite and hematite (highly reflective phase),from section KD-80-13. Microprobe analysesare not given. Transmitted and reflected light, air objective. d) Early reversely pleochroic (R) phlogopite rimmed by late stagekimberlitic phlogopite with normal pleochroism (N), from section KD-80-7, area A, grain 3. Dark areas in the core phlogopite are chlorite. Microprobe traverse analysesacross this grain are given in Tompkins and Haggerty (1983b).Transmitted light, air objective. Scale bars (a-b : 0.05mm;(c-d) : 0.2mm.

ties. Vermiculite samplesCHK-I and CHK-6 have ex- Shirozuand Bailey (1966)refined the structureof a2- tremely strong l4A intensities,which are a consequence layer "s" type vermiculite from Llano County, Texas, of the small amount of scattering power present in the and consider the "s" sequenceto be the most stable interlayers. The 144 peaks in the other two samplesare structure for vermiculite. De la Calle et al. (1976)and De intensifiedalso, but not to the samedegree. For example, la Calle et al. (1978)determined experimentally that the the l4Al7A inftnsity ratio for the untreated CHK-I "s" structure occurs only as a result of vermiculitization sampleis 37.0and for the CHK-2 sampleis 1.8. of phlogopite-lM.The transformation is reversible in the Although stacking of adjacent l4A structural units is laboratory. They designate the "s" structure as the V1 often semi-randomin chlorite. this is not the case in the type and illustrate identification of different vermiculite Koidu specimens. The fibrous vermiculite specimens structures by means of the diagnostic 02/ reflections. CHK-I and CHK-6 were found to be the same,for the They point out that the observed difuseness of these fibers sampled,and to have the fibers oriented parallel to reflections in V1 indicates a semi-orderedstructure hav- crystallographic L All fibers have a 2-layer structure of ing mistakesin the regularity of the +bl3 and -bl3 shifts, the type labeled"s" by Mathiesonand Walker (1954),in relative to the ideal regular "s" model, such that there is which thereis an L type shift within each2:llayer (-al3 equal probability of a given layer having a +bl3 or -bl3 parallel to the symmetry plane) and adjacent layers are shift. also shifted by +bl3 and -bl3 relative to one another. No Specimen CHK-2 is platy. One of the four platelets second phase was found by either powder or single studied by precessionphotography showed a superposi- crystal patterns of the CHK-I and CHK-6 samplesfor the tion of Ia and IIb type patterns. The other platelets fibers studied. showed only the Ia single crystal pattern, but abundant TOMPKINS ET AL.: KIMBERLITIC CHLORITES 245

KIMBERLITIC CHLORITES

"NCFM" 5,4 sp A KIMBERLITIC Fig. 6. A ternarydiagram summarizing all dataon kimberlitic presented paper.The compositionalplotting Fe2+ CHLORITES AI chlorites in this schemeis adoptedfrom Robinsonet al. (1982)in which Fig. 4. All chloriteanalyses are plotted as a functionof Fe2*- "NCFM" : Fe2*+ Mg+ Mn + Ca+ 2Na+ 2K - Ti,and"A" Mg-Al. Opencircles = low Fe chlorites;Hatched circle : low : Al + Fe3* + Cr + 2Ti - Na - K. Areasoutlined represent Fe/phlogopiteanalysis CHK-l3-D, Table2; Closedcircles : low natural occurring Di-Tri and Di-octahedralchlorites as Fe samplesnot fully characterized;Open apex up triangles: summarizedby Eggletonand Bailey (1967).Pyx = pyroxene, highFe chlorites;Closed apex up triangles= samplesnot fully Serp = serpentine;Phlog : phlogopite;Ol = olivine; Sp = characterized;Open apex down triangles= groundmass spinel.Symbols are the sameas in Figure4. Seetext for further chlorites;Open square : Mir pipe,Siberia (Frantsesson, 1970); explanation. Openhexagons : CaneValley diatreme and Moses Rock dikes, Utah (McGetchinet al., 1973);Open diamond : chlorite inclusionin diamond(Mitchell and Giardini, 1977). phlogopite-2Ml and vermiculite "t" : Vrr are rare in nature. De la Calle et al. (1976) and De la Calle et al. (1978)conclude from the modulated streaks along the 02/ extraneous streaks from other incoherently oriented row line that V1 also is a semi-orderedstructure in which phase(s) also are present. Powder patterns of the bulk there is an equal probability of a +bl3 or a -bl3 shift of a sample confirmed the presence of Ia and IIb materials. given layer. Instead of describing it as an average struc- The Ia material in this specimen has a 2-layer sequence ture ofthe regular"q" and "r" sequences,therefore, V1 labeled "t" by Mathieson and Walker (1954).They found is better describedas semi-orderedwith a layer sequence this sequencein a vermiculite from Kenya, which they given by +120' !b13, -120o !b13, +120" !b13, etc. It is believed to be the averageresult ofthe equal probability of adoption of the 2-layer sequences"q" and "r". In 2{ ocr both of these structures adjacent layers are related by INfERLI 5OL HcO lox 2rEr "r?l 120"rotations, but differ in the senseof the bl3 shifts. De 7z rrurealaven>l- Y22'{ LAYER la Calle et al. (1976) and De la Calle et al. 0978) determinedexperimentally that vermiculite "t" the struc- cHK-l"s" ture (V1 in their terminology) transforms only to phlogo- pite-zMr, which also incorporates 120"layer rotations in its structure and is considered the parent material. Both

KIMBERLITICCHLORITES V X oq ae F= o

F ol od o3 o5 Cr2O3(WT %) 100 120 140 i60 180 Fig. 5. Chemical variation plot from individual Koidu chlorite 40 60 80 z/c Pl analyses(apex up triangles).Only the groundmasschlorites have been distinguishedas shown by apex down triangles. Square : Fig. 7. One dimensional electron density projections for Mir pipe, Siberia (Frantsesson,1970). chloritesamples CHK-l and CHK-2. 246 TOMPKINS ET AL.: KIMBERLITIC CHLORITES not known in specimenCHK-2 whether or not the Ia speciesin peralkaline conditions (Dickenson and Hess, stackingsequence is restricted to the collapsiblevermicu- l98l). Potassiumwill act as an efficient charge balancing litic layers and the IID sequenceto the chloritic layers, speciesand tetrahedral Al3+ expands the framework to but this is a possibility. allow the substitution of Fe3+in the tetrahedral site. The Platesfrom specimenCHK-I3 readily break into fibers early core phlogopites(R) are reversely pleochroic, most upon cutting. Six fibers were found to give superposition probably due to the effects oftetrahedral Fe; they repre- of Ia and IID chlorite patterns on precessionphotographs. sent crystallization from a depletedpotassium-rich source This overlap prevented determination of the precise in which tetrahedral Fe3* played an important role in the stacking sequenceof layers in each phase. The IIb absenceof sufficientAl3* . Late stagereactions of dissem- chlorite is the most abundant structural type and is inated early phlogopite nodules with the kimberlite liquid believedby Bailey and Brown (1962)to be the most stable has permitted the nucleation and growth of Ti and Cr chlorite. It is the only chlorite to be synthesizedto date, enrichedphlogopites. and is the only chlorite structural type found in the Forbes and Flower (1974) have demonstrated experi- greenschistmetamorphic facies, except in clearly retro- mentally that Ti-rich phlogopites are more stable under gradereactions. Thus it is consideredthe highertempera- mantle conditions than Ti-free phlogopites.If Ti acts as a ture form of chlorite, somewhat analogous to the lM stabilizer to the phlogopite structure then the preferential polytype in the phlogopite system and the ZMl polytype replacement of the reversely pleochroic phlogopites by in the muscovitesystem. It is not known whetherbr not chlorite may result from Ti depletion and Fe enrichment. the Ia stacking sequenceis restricted to the collapsible The other possibility is that chlorite replacement oc- layersin specimenCHK-13. curred during a "chlorite event" as the rising kimberlite magma passed through the chlorite stability field. Late Discussion stageprecipitation of Ti-rich groundmassphlogopite unal- Experimental studies on the alteration of phlogopite to tered by, and in conjunction with chlorite (Fig. 3) further vermiculite,or vice-versa,(De la Calleet al.,1976;DeLa emphasizesthe stabilizing effect of Ti on phlogopite and Calle et al., 1978)show without exception that 2-layer the probable primary occurrence of groundmasschlorite. type V1(="s") structuresform from lMphlogopites,and Water-CO2 activity appears to be the driving mecha- that type Vn (:"t") structures result from 2M1 phlogo- nism for chloritization in the kimberlite. Vermiculite is pites. Neither of these 2-layer structures has been recog- present only in weathered kimberlite (Fairbain and Rob- nized previously in chlorite and may be restricted to the ertson; 1966)under hydrous and post emplacementcondi- vermiculitephases. Soboleva et al. (1982)studied phlogo- tions. When CO2 activity is low, phlogopite (R+N) is pites from the Yakutian kimberlites and determined that more stable and chlorite is present in minor amounts. all specimensthey examined are lM polytypes. These Some core phlogopites may even remain totally unaffect- data coupled with the selective replacement of early ed by chloritization. As phlogopite becomes unstable macrocrystic phlogopites by chlorite at Koidu suggests under conditions of increasing carbonitization, the reac- that the precursor phase to the discrete chlorite nodules tion to chlorite requires the formation of brucite and the was probably phlogopite and that the ultimate end prod- releaseof interlayer K. Magnesiumions must, therefore, uct would be vermiculite. This sequenceof kimberlite pre be in abundance and readily available for linkage with and post emplacement alteration (phlogopite-chlorite- hydroxyl molecules.In the carbonatitickimberlites, oli- vermiculite) is consistent with that proposed by Kreston vine is pseudomorphedby calcite, and chlorite is stable (1973). relative to phlogopite (Tompkins and Haggerty, 1983b). The phlogopite (R)-chlorite macrocrysts lend strong The carbonitization of olivine is, hence, a probable support in favor of a similar phlogopite composition for source to produce the required flux of Mg ions required the discretenodules; some xenocryst-like chlorites in the for brucite formation. The Xs.olXco, ratio or changesin groundmassmay representdisaggregated nodules. Identi- P-T conditions would be required to decreasesufficient- cal Cr2O3and TiO2 contents betweenthe chlorite nodules ly, subsequentto carbonitization, to permit the formation and core phlogopites(R) are observed(Tompkins, 1983). of brucite and the ultimate demise of phlogopite. This is Such low Cr2O3,TiO2, and Al2O3values represent growth necessarybecause positively charged brucite layers re- in a highly depleted environment. Deformation kink place the interlayer K ions. Highly mobile K ions can be bands in the nodules (Fig. 2) and chemical homogeneity more readily transported in volatile rich fluids and may of the core (R) phlogopites (Tompkins and Haggerty, precipitate to form other late-stage K-bearing phases. l9E3a) is compatible with crystal growth in a mantle Erlank and Rickard (1977)and Erlank et al. (1982)consid- sourceregion prior to incorporation in the kimberlite host er K-metasomatismto be widespreadin the upper mantle (Boyd and Nixon, 1978;Farmer and Boettcher, l98l). and although we have no direct evidencethat K-mobiliza- Experimental studies on the oxidation state of Fe in tion is related to the processeswe invoke, the removal of potassium rich melts show that Fe is dominantly in the K is nonethelessrequired for the transformation of phlog- trivalent state, and that Fe3+ behaves as a tetrahedral opite to chlorite. It is possible that the high K-chlorites TOMPKINS ET AL.: KIMBERLITIC CHLORITES 247 contain a vestige of earlier phlogopite and in this sense habits; (2) coarse anhedral macrocrysts and small euhe- represents incipient transformation to chlorite and/or dral platelets in kimberlite groundmass;and (3) as prefer- vermiculite. Primary, euhedralgroundmass chlorite is Fe- ential replacement products of reversely pleochroic enrichedand theseare comparablein most respectsto the phlogopite. The chlorites are compositionally distinctive, most Fe-rich chlorite nodules. high in SiO2 and MgO and low in Al2O3contents. Three The formation of phlogopite was at lower P-T condi- dominant compositional types are designatedas high (25 tions than the diamond/graphitetransition in the depleted wt.%FeO), intermediate(15 wL%) andlow (8 wt.%) FeO portion of the mantle (Farmer and Boettcher, varieties, with a minor type that is low in iron but hieh in 1981).Phloeopite intergrowths with diamond (Prinz et al., K2O. Compositions,comparisons with ideal chlorite, and 1975)are typically very high in TiO2 90.8 wt.%) consis- electron density measurements,suggest extensive vacan- tent with growth from an evolved liquid. rraeruo suite cies in the interlayer sheet. Fibrous specimens are Ia rocks (which are most similar to the core phlogopite (R) in structural types, ubiquitous in vermiculite but among the composition) are considered to have equilibrated under least abundant of the chlorites found in nature. There is relatively shallow depths resulting from early plutonic evidencefor nearly complete transformation to vermicu- kimberlite pulses (Dawson and Smith, 1977). In the lite and for phlogopite as a precursor. Platy chlorites are system MgO-SiOaCOrHzO the subsolidusformation of intergrown la andllb structural types, the latter being of brucite and carbonate from forsterite and vapor occurs relatively high temperatureorigin, the most stable among with decreasingtemperature and/or increasing pressure chlorites, and the most abundant structural type in na- and H2O/CO2values (Ellis and Wyllie, 1980).Based on ture. the stability of K-richterite (Kushiro and Erlank, 1970), Euhedral kimberlite groundmasschlorites are unequiv- the unnro suite nodules are considered to have crystal- ocally of primary magmaticorigin and high TiO2 contents lized at depths of 75-100 km. At these depths the chlorite are consistent with crystallization from a residual melt. dehydrationreaction curve (Jenkins,1981) intersects the Carbonatiteassociations or high calcite contents, and the Precambrianshield geotherm and is also coincident with alteration of phlogopite to Ia chlorite implies chloritiza- the depth where CO2 ceases to be a dominant solidus tion under high Xco2 activity. This condition prevailed in component (Wyllie, 1980), implying that high fluxes of a volumetrically minor component of Pipe 2 and in many CO2 vapor are present. of the kimberlite dikes, suggestingthat COz volatilization Experimental and theoretical studies of the Mg end- was impaired, possibly becausethese facies were passive member clinochlore (Jenkins, l98l; Obata and Thomp- intrusions. By contrast, the dominant diatreme facies son, l98l) show that chlorite (type IIb) can be stable at were explosive and enriched in H2O, and this resulted in low temperatures (<9fi)'C) up to :75 km in the upper serpentine + phlogopite. Although eclogites are the only mantle and an alternative possibility is that some of the Iithic nodules identified at Koidu that can be directly Koidu chlorites (types Ia, and lalllb) are not pseudo- linked to a fertile asthenosphere,it is also possible that morphs after phlogopite. Chlorite in diamond inclusions metasomatizedIithosphere was sampled in the form of (type Ib; Mitchell and Giardini, 1977) are similar to our phlogopite nodules that are now completely transformed low Fe chlorite compositional type in FeO, SiO2, and to chlorite. Al2O3contents only,In the two inclusionsreported, one is The chlorites we describe may not be of mantle origin; exceptionally enriched in CaO (10.7 wt.%) and probably additional chemical and crystallographic data are re- representsan overlap with coexisting clinopyroxene, the quired on these and chlorites in other kimberlites. Experi- other is a high magnesianvariety (40.3 wt.%) with almost mental data appropriate to these compositions are also twice the concentration of MgO than those from Koidu. needed to determine the conditions required for Ia and Although, the most stable chlorite polytype is type IIb it IIb type chlorite formation, and to addressthe problem of is interesting to note that the diamond inclusion is not a chlorite stability in the upper mantle. IIb but type ID. Whether type la chlorites are stable at mantle P and 7"s under appropriate conditions is uncer- tain. Given these uncertainties it is still possible that the Acknowledgments Koidu chlorite nodules are indeed primary upper mantle This researchwas partially supportedby NationalScience phases.The experimental data are not entirely appropri- Foundationgrants EAR 78-02539, EAR83-08297 (toS.E.H.), and ate to the compositions we discuss and data are clearly EAR-8106124(to S.W.B.).Access to, andlogistical support at required to evaluate the high pressure stability of low the Koidu kimberlite complex was made possibleby B.P. Al2O3,high Mg-Fe chlorites. Mineralsand SierraLeone SelectionTrust. Managementand geologicalstaff are thankedfor their help in obtainingcritical Conclusions samplesin this andthe overall project. Peter Robinson has had a continuedinterest in the chloriteproblem and is thankedfor Chlorite is present in three modes of occurrence in the fruitful discussions.D. Leonardwas responsible for maintaining Koidu kimberlite complex: (l) discrete nodules with the electronmicrobeam facility and D. Volz aidedin the wet fibrous, platy, tectonized, or multiply cleaved textural chemistry.Critical reviewsand constructivecomments were 248 TOMPKINS ET AL.: KIMBERLITIC CHLORITES provided by D. Hewitt and S. Bohlen. To all we express our Lesotho Kimberlites, p.2O7-213.Lesotho National Develop- gratitude. ment Corporation, Maseru. Fieremans,M. and Ottenburgs, R. (1979)Kimberlite inclusions References and chlorite nodules from the kimberlite-breccia of Mbuji- Mayi (Eastern Kasai) Zaire. SocieteBelge de Geologie.Bulle- Albee, A. L. (1962)Relationships between the mineral associa- tin/ BelgischeVereniging Voor Geologie, SE,205-224. tion, chemical composition and physical properties of the Flanagan, F. J., Ed. (1976)Descriptions and analysesof eight chlorite series. American Mineralogist, 47, 851-870. new USGS rock standards. Geological Survey professional Albee, A. L. and Ray, L. (1970)Correction factors for electron paper E40, 192pages. probe microanalysis of silicates, oxides, carbonates, phos- Forbes, W. C. and Flower, M. F. J. (1974)Phase relations of phates, and sulfates. Analytical Chemistry, 42, 1408-1414. titan phlogopite KzM&TiAlzSi6O2o(OH)4a refractory phasein Bailey, S. W. (19E0)Summary of recommendationsof erpee the upper mantle? Earth and Planetary Science Letters.22, nomenclaturecommittee on clay minerals. American Mineral- 60-66. ogist,65,l-7, Foster, M. D. (1962) Interpretation of the composition and a Bailey, S. W. and Brown, B. E. (1962)Chlorite polytypism:I. classificationof the chlorites. Geological Survey professional Regular and semi-random one-layer structures. American paper 14 A. Mineralogist,47, 819-850. Frantsesson,E. V. (1970)The petrology of kimberlites. Translat- Bardet, M. G. (1974) Geologie du diamant. Memoires du ed by D.A. Brown, Department of Geology Publication No. B.R.G.M..n'83. tone I and II. 150, Australian National University, Canberra. Bence, A. E. and Albee, A. L. (1968)Empirical correction Giardini,A. A., Hurst, V. J., Melton, C. E. and Stormer,Jr., J. factors for the electron microanalysis of silicates and oxides. C. (1974)Biotite as a primary inclusion in diamond: Its nature Journal of Geology, 76, 3E2-403. and signiflcance.American Mineralogist, 59, 783-789. Boyd, F. R. and Nixon, P. H. (1978)Ultramafic nodulesfrom the Grantham, D. R. and Allen, J. B. (1960) Kimberlites in Sierra Kimberley pipes, South Africa. Geochimicaet Cosmochimica Leone. OverseasGeology and Mineral Resources.The Quar- Acta,42, 1367-1382. terly Bulletin of the OverseasGeological Survey, London. 8, Dawson, J. B. and Smith, J. V. (1977)The uenro (- 5:25. amphibole-rutile-ilmenite-diopside) suite of in kim- Hall, P. K. (1970)The diamond fields of Sierra Leone. Geologi- berlite. Geochimica et Cosmochimica Acta, 41, 309-323. cal Survey of Sierra Leone. De la Calle,C., Dubernat,J., Suquet,H.,Pezerzt, H., Gaultier, Hayes, J. B. (1970)Polytypism ofchlorite in sedimentaryrocks. J., and Mamy, J. (1976) Crystal structure of two-layer Mg- Clays and Clay minerals, 18, 285-306. vermiculitesand Na, Ca-vermiculites.In S. W. Bailey, Ed., Jenkins, D. M. (1981)Experimental phase relations of hydrous Proceedingsof the International Clay Conference,p.201-28. modelled in the system H2O-CaO-MgO-AI2O3- Applied Publishing Ltd., Illinois. SiO2.Contributions to Mineralogy and Petrology, 77, 166-176. De la Calle,C., Suquet,H., Dubernat,J. and Pezerat,H. (1978) Kreston, P. (1973)Ditrerential thermal analysisof kimberlites. In Mode d'empilement des feuillets dans les vermiculites hydra- P.H. Nixon, Ed. Lesotho kimberlites, p. 269-279. Lesotho teesa'deux couches'.Clay Minerals, 13,275-297. National Development Corporation, Maseru. Dickenson,M. P. and Hess, P. C. (1981)Redox equilibriaand Kushiro, I. and Erlank, A. J. (1970)Stability of potassicrichter- the structural role of iron in alumino-silicatemelts. Contribu- ite. Annual report of the Director of the GeophysicalLabora- tions to Mineralogy and Petrology, 78,352-157. tory, CarnegieInstitution of WashingtonYear Book, 68,231- Eggleton, R. A. and Bailey, S. W. (1967)Structural aspects of 233. dioctahedralchlorite. American Mineralogist, 52, 673-689. Lehmann, E. (1965)Non-metasomatic chlorite in igneousrocks. Ellis, D. E. and Wyllie, P. J. (19E0)Phase relations and their Geological Magazine, lO2, 24-35. petrological implications in the system MgO-SiO2-H2O-CO, Mathieson, A. McL. and Walker, G. F. (1954)Crystal structure at pressuresup to 100kb. American Mineralogist, 65, 540-556. of magnesium-vermiculite.American Mineralogist, 39, 231- Erlank, A. J., Alsopp, H. J., Hawkesworth,C. J., and Menzies, 255. M.A. (1982)Chemical and isotopic characterizationof upper Maxwell, J. A. (l%8) Rock and Mineral Analysis. Chemical mantle metasomatismin peridotite nodules from Bultfontein. Analysis,27, 419. Terra Cognita, 2, 261-263. McGetchin, T. R., Nikhanj, Y. S. and Chodos,A. A. (1973) Erlank, A. J. and Rickard, R. S. (1977) Potassium richterite Carbonatite-kimberlite relations in the Cane Valley Diatreme, bearing peridotites from kimberlite and the evidence they San Juan County, Utah. Journal ofGeophysical Research,TE, provide for upper mantle .Abs. Vol., Second lE54-1869. International Kimberlite Conference,Santa Fe, New Mexico. Mitchell, R. S. and Giardini, A. A. (1977)Some mineral inclu- Fairbairn, P. E. and Robertson, R. H. S. (1966) Stagesin the sions from African and Brazilian : their nature and tropical weathering of kimberlite. Clay Minerals, 6, 351-370. significance. American Mineralogist, 62, 756-7 62. Farmer, G. L. and Boettcher, A. L. (1981)Petrological and Nixon, P. H. and Kreston, P. (1973)Butha-Buthe dyke swarm crystal-chemical significance of some deep-seated phlogo- and associatedkimberlite blows. In P.H. Nixon, Ed. Lesotho pites. American Mineralogist, 66, 1154-1163. Kimberlites, p.197-2O6. Lesotho National Development Cor- Fawcett, J. J. (1965)Alteration products ofolivine and pyroxene poration, Maseru. in lavas from the Isle of Mull. Mineralogical Magazine, Obata,M. and Thompson,A. B. (19E1)Amphibole and chlorite 35, 55-68. in mafic and ultramafic rocks in the lower crust and upper Ferguson,J., Danchin,R. V., and Nixon, P. H. (1973)Fenitiza- mantle, A theoretical approach. Contributions to Mineralogy tion associatedwith kimberlite magmas.In P. H. Nixon, Ed., and Petrology, 77, 74-81. TOMPKINS ET AL.: KIMBERLITIC CHLORITES 249

Pauling, L. (1930)The structure of the chlorites. Proceedingsof Soboleva,S. V., Khar'kiv, A. D., Zinchuk, N. N. and Kotel'ni- the National Academy of Science, 16, 578-582. kov, D. D. (1982) Characteristics of phlogopite of mantle Prinz, M., Manson, D. V., Hlava, P. F. and Keil, K. (1975) origin. International Geology Review, 24, 35. Inclusions in diamonds: garnet lherzolite and eclogite assem- Tompkins, L. A. (1983)The Koidu Kimberlite Complex, Sierra blages.In L.H. Ahrenset al., Eds., Physicsand Chernistryof Leone, West Africa. M.S. thesis, University of Massachu- the Earth. 9.797-816. setts. Robinson,P., Spear,F. S., Schumacher,J. C., Laird, J., Klein, Tompkins, L. A. and Haggerty, S. E. (1983a)The Koidu C., Evans,B. W. and Doolan, B. L. (1982)Phase relations of kimberlite complex, Sierra Leone: Geologicalsetting, petrolo- metamorphicamphiboles: natural occurenceand theory. In D. gy, and mineral chemistry. Proceedingsof the third Intema- R. Veblen and P. H. Ribbe, Eds., Amphiboles: Petrology and tional Kimberlite Conference, Developments in Petrology, experimentalphase relations ,98, p. l-228. Reviews in Miner- Volume I, ElsevierPub. Co., in press. alogy, Mineralogical Society of America, Washington, D.C. Tompkins, L. A. and Haggerty, S. E. (1983b)The Koidu Schreyer,W. and Yoder, H. S., Jr. (1964)The system Mg- kimberlite complex, Sierra Leone: Geologicalsetting, petrolo- cordierite-H2O and related rocks. Neues Jahrbuchfiir Minera- gy, and mineral chemistry: appendix. IIIeme conf. int. kimber- logie, Monatshefte,101, 271-342. lites, "Documents," J. Kornprobsted., Ann. Univ. Clermont Shirozu, H. and Bailey, S. W. (1965)Chlorite polytipism III: Fd., in press. Crystal structure of an orthohexagonaliron chlorite. American Wyllie, P. J. (1980)The origin of kimberlite. Journal of Geophys- Mineralogist, 50, 868-885. ical Research.85. 6902-6910. Shirozu,H. and Bailey, S. W. (l%6) Crystal structureof a two- layer Mg-vermiculite. American Mineralogist, 51, 1124-1143. Sobolev,N. V. (1977)Deep-seated inclusions in kimberlitesand the problem of the composition of the upper mantle. American Manuscript received,February 21, 1983; GeophysicalUnion, Washington,D.C. acceptedfor publication, October3, 1983.