Genesis of Diamond: a Mantle Saga

AmericanMineralogist, Volume 70, pages344-355, 1985

Genesisof diamond:a mantle saga

HsNnv O. A. MeYEn Department of Geosciences Purdue Uniuersity West Lafayette, lndiana 47907

Abstract A model for the genesisof natural diamond is presentedbased on the physical,chemical and mineralogicalproperties and featuresof diamond. Optical studiessuggest that individual diamonds have had complex growth histories in which growth and dissolution may have occurred.Growth was not always continuousnor did diamonds grow in necessarilysimilar chemicalenvironments. Evidence for this is provided by variation in the nitrogen and trace elementcontents in diamondsas well as information from studiesof the mineralsincluded in diamond. Isotopic data suggestthat diamonds formed from carbon whose sourcesvaried isotopically.The possibilityexists that somediamonds may be productsof recycledsubducted carbon, whereasothers have formed from primordial material either through magmatic or metasomaticprocesses. It is also likely that most diamonds formed in the Archaean or Proterozoic.The cognate host rocks for diamond in the mantle were severalbut can be broadly grouped into eclogitic and ultramafic (peridotitic); however, in mineralogic and chemical detail these rocks are quite diverse.Although diamond is commonly found in kimberlite and in lamproite at the earth's surface,these two rocks are not geneticallyrelated to diamond formation. Instead they are the transporting vehiclesin which diamond ascended rapidly from mantledepths to the crust.

Introduction initial evidencefor this latter idea was the discovery of Although diamond has beena sourceof fascination,the diamond in an eclogitexenolith from kimberlite (Bonney, origin of this mineral has for centuriesperplexed man. 1899).Du Toit (1906),Wagner (1914)and Sutton (1928) Greek philosophers and medieval alchemists ziscribed modified this idea and suggestedthat the eclogiteand peri- many mystical properties to diamond. When taken as a dotite xenoliths were cognate with the kimberlite. Dia- powder,voluntarily or involuntarily, it could, among other mond was thus genetically related to the early crys- things, cure diseases,poison ones enemiesor make the tallization of kimberlitemagma. honest strong and agile. An unusual belief, especially held The above two hypotheseshave lasted until the present in Greece and India was that diamond could procreate and have developed with some modifications into the itself-a boon to the owner of a diamond mine. phenocrystversus xenocryst schools(Dawson, 1980).For In 1772 Lavoisier demonstrated that diamond, like example,Gurney et al. (1979)and Harte et al. (1980)main- carbon, would burn in air. However, it was only later in tain that diamonds are geneticallyrelated to early crys- 1797 that Smithson Tennant proved that diamond consist- tallization products of kimberlite within the upper mantle ed of carbon. This led several gentlemen scientists of the and are thus phenocrysts.In contrast, Meyer and Tsai nineteenth century to suggest that diamond was formed (1976a),Robinson (1978),and Meyer (1982 a,b) have through the action of heat and pressure on plant remains arguedthat diamondsare accidentalinclusions in kimber- (Des Cloizeaux, 1855; Goppert, 1862). lite and thus are xenocrysts;the associationof diamond The discovery of diamonds in a volcanic rock (kimber- and kimberlite being one of passengerand transporting lite) at Kimberley, South Africa in 1871 led to more scien- vehicle. tific, and less philosophical studies. This did not, however, Most scientistsfamiliar with diamond concedethat dia- deter various authors from presenting opposing view- mond has grown stably within the upper mantle (Kennedy points, as summarized by Williams (1932). For example, and Nordlie, 1968;Meyer and Boyd, 1972;Orlov, L973; Lewis (1887) considered that diamond formed in the crust Sobolev,1974; Robinson, 1978). Omitted for purposesof as the kimberlite host rock solidifled-the carbon being this discussionare the polycrystalline aggregatesof dia- derived from coal and other carbonaceous material. mond (carbonado,framesite, boart) which have received In contrast others maintained that diamonds had orig- little scientificstudy (Trueb and DeWys,1969; 1971; Trueb inally formed in ultrabasic rocks at depths, and were subse- and Barrett, 1972; Gurney and Boyd, 1982) and whose quently released as the rocks fractured upon incorporation origin is even more uncertain than the single crystal dia- into the kimberlite melt (Harger, 1905; Holmes, 1936).The mond consideredhere. 0003-{04xl85/0304-{344$02.00 344 MEYER: GENES/S OF DIAMOND: A MANTLE SAGA 345

Suggestionsas to the sourceof carbon from which diamond forms have been diverse and range from coal and plant remains as favored in the 1800's,to carbon dioxide and methanetoday. However,whether or not the carbon is primitive or from recycledcrustal material is a necessary question.Current studieson carbon isotopes(Deines, 1980; 1982;Milledge et al., 1983),as well as on nitrogen (Becker, 1982)and rare gases(Ozima and Zashu, 1983)bear on this question. An important factor in understandingthe formation of natural diamond is afforded by detailed examination of minerals included in diamond. Thesestudies, mostly crys- tallographic and mineralogical,have beenreviewed by So- bolev (1974),Meyer and Tsai (1976a),Harris and Gurney (1979)and Meyer (1982a).Isotopic studies of these small inclusions in diamond are now possibleand future work should provide significantresults concerning diamond and the evolution of the upper mantle. This paper suggestsa model for the genesisof diamond, and its subsequentpassage to the earth's surface.Current interest in the evolution of the upper mantle and magma generationconsiders diamond to be an unreactivechemical probe from the depths. One aim of the discussionand model presentedherein is to place the genesisof diamond within the correct context of mantle processes.It is also hoped that the discussionwill remove various misunder- standingsthat are prevalent with respectto diamond and its relationshipto kimberlite and other rocks. A subsidiary aim of this paper is to bring to the attention of mineral- ogists the large amount of important information contributed by physiciststo diamond research,and equally to exposephysicists to geologicalprocesses attendent on diamond formation and the subsequenthistory of diamond. Although the host rocks for diamond at the earth's surface are kimberlite and lamproite, it is believedthat these Fig. l. (a) A transparent,clear and well shapedoctahedron of rocks are not geneticallyrelated to diamond. Accordingly, diamond.Octahedral edge I mm. (b) A polishedand etchedsec- it is not the aim of this paper to dwell on the nature of the tion throughan octahedraldiamond (2.4 mm on edge),showing chemical and mineralogical differenc€sbetween various the internalstratigraphy of diamond.Various growth layers are kimberlites,and betweenkimberlites and lamproites.The eitherType I or TypeII diamond(Figure 3.1; Harrison and Tol- ansky,1964). interestedreader is referredto Dawson (1980)for kimberlites, Mitchell (1984)for lamproites, and the proceedings volumesof the threeInternational Kimberlite Conferences. cuboid surfaces.More detailed descriptionsof this phe- Physical features of diamond nomenon are to be found in Suzuki and Lang (1976)and A considerableamount of detailedstudy into the physics Lang(1979).The stratigraphyof diamond can also be illus- of diamond has been undertaken over the past 35 years trated by cathodoluminescenceon polishedsurfaces ofdia- (Berman,1965; Field, 1979).Much of this researchhas mond (Moore, 1979) and by X-ray topography (Lang, significanceto mineralogy and bears on the formation of 1979).The growth stratigraphyis observedbecause various diamond. layers consist of either Type I or Type II diamond, and Figure la is a photographofa typical clear and colorless these two types have different chemical and physical diamond without any visible flaws. This clarity, shown by properties(Table 1). many diamonds,suggests to the observercrystallization in The presenceof Type I and II diamond was first demon- a singleuninterrupted event. This is not the case.In Figure strated by Robertson et al. (1934)based on differencesin lb is shown a polished and etched surfaceof a diamond UV and IR absorption.Lonsdale (1942) showed that Type displayinga seriesof geometricallayers-referred to as the I diamond producesextra X-ray diflraction reflections,or stratigraphy of diamond (Harrison and Tolansky, 1964; spikes.These spikes were interpreted to be due to platelets Seal, 1965). These patterns were interpreted by Frank within the diamond structure (Frank, 1956).Kaiser and (1966)as being due to periodic growth on octahedraland Bond (1959),and later Lightowlersand Dean (1964)proved t46 MEYER: GENES/SOF DIAMOND: A MANTLE SAGA the presenceof nitrogen in Type I diamondsand showeda Table 1. Someproperties of diamond correlation betweennitrogen content and optical absorp- i ficat ion tion at 7.8 pm (1282 cm- r). In contrast Type II diamond CI ass Type la - Xosr comnon, appror. 981 of natural dianonds. Contains containsvery little nitrogen and no platelets. nitrogen up to 2500 ppn by uc. ss aggregares and platelets

The aboveevidence suggests that the stratigraphyof dia- Type lb - Rare in naturc but mosc synthecic diamondE are of lhis Lype mond recordsperiodic growth in chemicallydifferent envi- Nitrogen < 20 pr by pt in drsperscd Eubstitutional fom. Type lla - Very ra.e. NitroAen. 20 pF by ur, Often the very ronments,at least with respectto the amount of nitrogen large gen diamonds are rhis rype. present.Alternatively, the rate of crystallization,or length Type 1lb - Ext.eoely rare 3nd generally blc in color. Semi_conducrins of residencetime after partial growth may also contribute (B .20 pr by wt.) uosc pure Eype of diadond. to differencesin nitrogencontent. Space croup: Fd3tu - Oil At high pressureit is possibleto causemigration of the unit 3.56683 + o.oO00I to 3.56125. O.oOOOrl substitutional nitrogen in diamond so that the nitrogen cell: Densi ry: aggregates(including platelets)found in natural diamonds Type I - 1.51537 + 0.00005 go cn-3- Type 1I - 3,51506 + 0,00005 go cn-r are produced(Evans and Qi, 1982).Based on the resultsof this study Evans and Qi suggestthat Type Ia diamond Ultrsviolet and lnfrared - Type I SrronS absorpLion <140 nn and blteen 6 to llxlo3 nm must haveexisted for between200 million and 2000million Type II - Trsnspare.r to 225 nm a.d berueen 6 to llxlo3 nm years.This range in time is admittedly large but is due to insuffrcientknowledge of the activation energyof migration Type ra - various (e.e. colortess, pale yerrov, brom) of certain aggregatesin diamond at high pressureand tem- rype lb - various (e.s. yerlou, dark brom) perature. Nevertheless,the data indicate that some dia- type lIa _ Colorless, broen - lllue monds have had gestationperiods in excessof the age of TYpe tIb the kimberlite eruption that transportedthem to the crust. Therftal ConductrviEy Type la - 6OO-1OOOm-I K-l (at 293"K) In summary, although individual diamonds are grossly Type lI - ZOOO-ztOOm-1 K-l (at 293'K) similar in physical properties detailed examination shows Resrs t Lvr ly subtle differencesresulting in four distinct types of dia- Tvpe I - >1016 ohm n mond (Table 1). Growth of diamonds is crystallogra- Iype lla - ca.lu'" ohn n Typ€ IIb - tol - to5 ' phically discontinuous and reflects possible variation in "n. chemistryof the growth environment.Experimental aggre- gation of nitrogen to form platelets and other nitrogen aggegatesin diamond suggestsvery long residencetimes techniques (e.g. nuclear activation and ultra sensitive for diamond within the upper mantle prior to reachingthe gamma ray spectroscopy) have enabled trace elements to earth'ssurface. be studied in a single diamond, and the results should provide significant information regarding the chemical differences between the various types ofdiamond. Chemical features of diamond Various diamonds have diflerent trace element contents The presence of nitrogen as a major impurity in dia- (Sellschop, 1979). This variation in trace element content mond has been noted, as also has the occurrence of boron suggests that different diamonds have not always formed ( < 20 ppm) which accounts for the semi-conducting under identical chemical conditions even though the dia- properties of Type IIb diamond (Table 1). As a chemical monds may have been obtained from the same kimberlite sink for various elements, diamond is fairly poor with pipe. If color is due to variation in trace element content, regard to concentration levels, although 58 elements have been recognized at the trace impurity level (Sellschop, 1979).Most of the results reported by Sellschop were ob- Table 2. Elementspresent in diamond (maximumppm by weight) tained using instrumental neutron activation analysis, but Element Elenent Anount EleE€nt early studies used emission spectrographic techniques (Chesley, 1942;Raal, 1957). Elements occurring in amounts H I r00 s90 2 I IO cl c NT 80 greater than I ppm are listed in Table 2 (Sellschop, 1979), N 5500 K48 Cu 40 and most of these elements are typically present in silicate o r700 ca 195 Ag 30 and sulfide magmas. Na 34 Ti8 Ba 58 In a significant contribution, Fesq et al. (1975)suggested Mg 370 cr t000 t4 that in certain instances the trace elements in diamond AI I00 ln5 Ce t1 si EO Fe t40 Hg 5 were contained in sub-microscopic inclusions that repre- sented quenched, or temperature re-equilibrated, melt from Elenent6 present in diarcnd F, Sc, V, As, Rb, Sr, Sb, C3, Ih, Eu, which diamond had crystallized; composition of the melt

then the wide range of colors in different diamonds from al. (1980)obtained a mean value for 61sN of +L5h, the same pipe can also be used, in part, to support the whereasBecker (1982) reported a mean value of *29oln tor above arguement (Robinson, 1978). five diamonds,with a rangeof 0 to * Sc/n.Both thesemean valuesat presentresult in diamond being isotopicallylight- Isotopic studies er in nitrogen than mantle rocks (Becker and Clayton, Geochemical information bearing on the genesis of dia- 1977)whose range is +6 to +20o/',.Studies of rare gasesin mond is available from studies on the isotopes of carbon, diamond (Ar, He, Ne) have been undertakenby Takaoka nitrogen and rare gases. and Onma (1978),Oama and Zashu (1983),and Ozima et al. (1983).The latter authors reported 3HelnHevalues Carbon for three diamondsthat werein excessof the primordial value On the basis of a few measurements, the d13C value for for meteorites(1.42 + 2 x l0-a; Reynoldset al., 1978),and diamond was once considered to be generally in the range close to the solar value (-{ x 10-a; Black, 1972).A -4 -9o/oo to (Craig, 1953; Wickman, 1956).More recently second group of diamonds have much lower 3He/aHe several authors (Galimov et al., 1978; Gurkina et al., 1979: ratios. Ozima and coworkers propose that the diamonds Sobolev et al., 1979;Deines, 1980and Milledge et al., 19g3) with high 3He74Heratios trapped primitive helium soon have extended the range from *3 to -34%oo.Nevertheless, after the formation of the earth. whereasthose with lower the majority of diamonds studied (e.g.,Deines, 19g0) have valuesmay have formed later and included more evolved 613Cvalues in the range -4 to -8oln. helium from the mantle. In a significant study, Milledge et al. (1983) have exam- Most of the variations in isotopic contents of diamond ined the d13C content of 18 Type II diamonds and ob- can be explainedby consideringthat the carbon has passed tained a range of values between 0 and -32y@, although through a subductioncycle (Frank, 1966).In contrast,very - 404r/36Ar only two values were below l5y@. It was concluded that high ratios obtained by Takaoka and Ozima Type II diamond is isotopically lighter than Type Ia dia- (1978)are interpreted as indicating that the carbon from mond which presumably comprises the majority of dia- which diamond formed has never reachedthe earth's sur- monds. as noted above. face; a similar comment is pertinent for the high 3He/4He The existenceof a range of dr3C values for diamond has valuesof Ozima et al. (1983).Melton and Giardini (1980) a bearing on the source of carbon and the chemical reac- obtaineda 4o{rf361ltratio of 190for a diamondfrom the tions that produce diamond. Recent models (Deines, 1980; Prairie Creek diatreme,Arkansas, well below that of the Mitchell, 1975) as to the role of a vapor phase in diamond atmospherewhich is taken as 296. Interestingly,this dia- formation have proven ambiguous, although the presence treme has recentlybeen reclassified as a lamproitic kimber- of a vapor could help explain the range of d13C. Deines lite (Scott Smith and Skinner, 1984),whereas Mitchell and (1980) concluded that this range was most likely inherited Lewis (1983)regard part of the diatremeto be a madupite. from the source carbon, suggesting isotopic heterogeneity The data of Melton and Giardini (1980)represent the first in the mantle. Mitchell (1975) in contrast has pointed out published results of rare gas contentsin diamond from a that variation in the isotopic composition of diamond may non-kimberlitic primary source(see also Roedder,1984, p. be the result of re-equilibration (isotopic) of the solid phase 508-511). with changing carbon-bearing gas compositions. Geochemicaland isotopic studies show that diamond Discussed later is the fact that minerals included in dia- contains many elementsin trace amounts, but generally mond can be assigned to either an ultramafic or eclogitic those occurringin the largestconcentrations are similar to suite. A careful study by Sobolev et al. (1979) involved those presentin silicateand sulfide magmas.This suggests examination of d13C values of diamonds in terms of wheth- that diamond grew in a similar environment to most er they contained ultramafic or eclogitic suite inclusions. silicate-bearingrocks. Isotopic data, particularly the d13C For diamonds with ultramafic inclusions the majority had values,show that the carbon from which diamond formed a 613C value of -5%oo and a restricted range between -2 was isotopically variable. Furthermore, the variation in and -9'/oo. Conversely, diamonds with eclogitic suite in- d13Cbetween Type I and Type II diamond can be inter- clusions had 613C values ranging from 0 to -34%o. As preted as diamond having formed in more than one envi- pointed out by Milledge et al. (1983), this latter range is ronment, and one possibility is that somediamonds repre- similar to that of Type II diamond, whereas the peak at sent recycled crustal carbon. It is important that future -5Y* for the ultramafic suite diamonds corresponds to isotopic studiesbe carried out on well-documentedspeci- that of Type I. A study of the type of diamond in which mens,including geographicsource and type of diamond. eclogitic inclusions occur has yet to be undertaken. Mineral inclusions Nitrogen and rare gases Various studiesof mineralsincluded in natural diamond Attempts to determine the origin of diamond and the are documented by Sobolev (1974), Meyer and Tsai geochemical nature of the upper mantle have also involved (1976a),Harris and Gurney (1979)and more recently by studies of the isotopes of nitrogen as well as those of rare Meyer (1982a).The significanceof mineral inclusionswith gases,notably helium. Data on nitrogen isotopes are sparse regard to diamond genesis and the upper mantle and consequently any conclusions are speculative. Wand et characterizationwas appreciatedearly (e.g.,Bauer, 1896), 348 MEYER: GENESISOF DIAMOND: A MANTLE SAGA but apart from Sutton (1921)and Williams (1932)little Table 3. Minerals occurringas inclusionsin diamond work was done until the 1950's(G0belin, 1952; Mitchell and Giardini, 1953;Futergendler, 1956, 1958, 1960). These Eclogitic suite were mostly optical and X-ray diflraction studies,and it Ultrenefic Suite Ol i vine Omphaci ie was not until 1967that the first electronmicroprobe analy- EnEtatite Py rope-alnandine ses were obtained (Meyer, 1968; Meyer and Boyd, 1969, Diops ide Kyani te Cr -pyrope coesite (Qurtz) 1972).Since then others have contributed significantly to PhI ogopi te Rut i Ie Cr-sp ine I Ruby the study of inclusions(see Meyer, I982a for references). Mg-i 1reni te I lneni te A number of resultsfrom thesestudies bear on the gen- Zircoo Chrooite Sul f ides su1 fide s esis of diamond. For example,most inclusionscan be as- Dianond Dianond signed to one of two mineral suites-an ultramafic suite and an eclogitic suite (Table 3). Membersof the two suites EP IGENETIC are mutually exclusive;that is, mineralsof one suite do not Serpentinc, coethite, craphite, Kaolinite Henatite, Sellaite, X€notine coexist in the same diamond with minerals of the other suite.The discoveryof this phenomenon(Meyer and Boyd, IJNCERTAIN PAMGENESIS proof L972;Prinzet al., 1975)was the first that diamonds Huscovite, Bioti!e, Senidine, Uagnetite, Anphibole grew in more than one geochemicalenvironment. It should be noted that inclusionscan also be subdivided into protogenetic,syngenetic, and epigenetic.For purposes chemistriesthat differ in detail from similar minerals that ofthis discussiononly proto- and syngeneticinclusions are occur in kimberlite. This is not to say that one cannot find considered.Many syngeneticinclusions, particularly oli- in kimberlite rare examplesof mineralsthat are chemically vine, display a cubo-octahedralmorphology that is im- equivalent to those occurring as inclusions (Gurney and posed by the diamond host. In spite of diligent searchby Switzer,1973) but theseminerals are most probably xeno- scientiststhrough severaltens ofthousands ofdiamonds no crystsunrelated to kimberlite crystallization. macroscopic or microscopic fluid or gaseousinclusions havebeen found (Roedder,1982; 1984). Kimberlite and lamproite The majority of mineral inclusionsare small (- 100 pm) Kimberlites are widely distributed throughout the conti- and monominerallic,although bi- and polyminerallic in- nental regionsof the world (Bardet,1974, 1977; Wilson, clusionsdo occur. Multiphase inclusionsare important be- 1982).Most are not diamond-bearing,and comparedto the causethe chemicaland physical information they provide diamondiferousones have been little studied. Diamondi- enablesestimation of the pressureand temperatureof equi- ferouskimberlites appear to be confined mainly to the in- libration of the inclusions,and by inferencethat of the host teriors of stable cratons. such as Southern Africa and Si- during diamond genesis.Estimated conditions of equili- beria, although somedo occur closeto continentalmargins bration for co-existingultramafic suite inclusions in dia- in Liberiaand SierraLeone (Bardet,1974; Haggerty, 1982). mond are between 900 and 1300'C and 45 to 65 kbar Dawson (1980)has provided a review of kimberlites and (Prinz et al., 1975; Meyer and Tsai, 1976b; Hervig et al., their xenoliths,and recentlyClement et al. (1984)have pro- 1980;Boyd and Finnerty, 1980).These values lie within the posed a redefinition and classificationof kimberlite (see region of diamond stability (Kennedyand Kennedy, 1976) alsoSkinner and Clement,1979). and are similar to those obtainedfor diamond-bearingand Kimberlite occursin diatremes,dikes and rarely as sills; diamond-freegarnet lherzolite xenoliths(Shee et al., 1982). multiple intrusions within a single diatreme are common. It is not possibleto determineunequivocally pressures of Neighboring diatremes,or pipes, may contain similar or equilibration for eclogitesuite inclusions,but temperatures very different suitesof xenoliths,or none at all. Somekim- lie within the rangecalculated for the ultramaficsuite. berlites contain almost entirely xenoliths of eclogite, for A major unsolvedproblem at presentis the direct deter- example the Roberts Victor kimberlite, whereas others, mination of the age of diamond from the diamond itself. such as the Kimberley pipes, are rich in lherzolitic xeno- Currently, it is necessaryto obtain the age of syngenetic liths. Multiple intrusionsin a singlepipe often contain their mineral inclusions and to assumethis approximatesthe own suite of xenoliths,or xenocrystsand more importantly diamond age. Kramers (1979) examined batches of diamond. Type I and II diamonds can occur in the same inclusion-bearing diamonds from kimberlites in South pipe, and diamondswith ultramafic inclusionscoexist with Africa, and using lead isotopic techniquesdemonstrated ones that contain eclogitic suite minerals.In spite of the that in general the inclusions,and by inferencethe dia- economicimportance of diamond, it is a trace mineral in monds, were much older than the kimberlite eruption. In kimberlite and rangesonly up to I part in 2.5 million; a the caseof the Finsch and Kimberley kimberlites,the age quoted figure is I in 20 million (e.g.,Kennedy and Nordlie, of the diamondsis greaterthan 2 b.y., whereasthe kimber- 1968). lites erupted about 90 m.y. ago. More recent work by Confusion has arisen due to the discoveryof diamond- Ozimaet al. (1983)suggests that somediamonds may have bearingrocks in north-westernAustralia that unfortunately agescomparable to the ageof the earth. were referredto initially as kimberlite (Jacqueset a1.,1982; Most ultramafic and ecloeitic suite inclusions have McCulloch et al., 1983).The so-called"kimberlites" in that MEYER: GENES/SOF DIAMOND: A MANTLE SAGA 349 region are tuffaceouslamproites (i.e.,Ellendale, Argyle)---- 1980; Anderson, 1982; Allegre, 1982)and conduit forma- someof which are rich in diamonds(Jacques et al., 1984). tion (Mercier, 1979\.McGetchin and Ullrich (1973)calcu- Lamproitesare ultrapotassicrocks that may be plutonic, lated a speedof ascentfor kimberlite of approximately70 hypabyssal,or volcanic. In the latter instanceboth lavas, km/hr on the basis of xenolith size and density.A similar tuffs and vent breccias may be present. Generally the value was computedby Mercier (1979)from olivine anneal- various rocks occur in groups,or fields,but worldwide they ing data. Both theserates of ascentare consistentwith the are relatively restricted(e.g., Leucite Hills, Wyoming; West cooling rates calculated by McCallister et al. (1979) for Kimberley, N.W. Australia).Mineralogically leucite(or sa- exsolution phenomenain pyroxenesfrom kimberlite. The nidine) and Ti-rich minerals are almost ubiquitous. Ti- evidenceat present thus favors a rapid ascent,possibly amphibole,Ti-phlogopite, priderite, etc. are usually present hours, for kimberlite (and diamond) from depths in the and dependingupon the abundanceand type of minerals upper mantle. Although geochemicalprocesses in a diapir severalrock names are possible(e.g., orendite, wolgidite, may have contributed to the formation of kimberlite, the wyomingite,fitzroyite, etc.-Hughes, 1972, p. 321; Mitchell, route through the upper mantle was probably by crack 1984). propagation(D. H. Eggler,pers. comm., 1984). The significanceof the diamond-bearinglamproites is It can be argued that the speedof ascenthas servedto that kimberlite is now not the only primary crustal source preservediamonds that would havebeen absorbed into the of diamond.It has beenreported that somenon-kimberlitic kimberlite magmawith slow ascent.Other factorsaflecting rocks in the Soviet Union contain diamonds (Kaminsky the stability of diamond would be temperature,particularly and Gevorkin,1976;Kaminsky, 1980;Dawson, 1980). In the length of residencetime at high temperaturesand pres- several instancesre-evaluation of what were once con- sures outside the diamond stability field, and oxygen fu- sidered"odd" kimberliteshas shownthem to be lamproitic. gacity. Preliminary experiments(Meyer, unpub.) on the The diamond-bearingPrairie Creek diatreme in Arkansas effect of temperatureandfo, on the oxidation rate of dia- is one such example,and is now referredto as lamproite mond at room pressurehave provided interesting data. At (Scott Smith and Skinner, 1982,1984), and most likely the an fo, equivalentto 10-12 atm. and a temperatureof metakimberlites(Bardet, 1974)of the Ivory Coast (Knopf, 1000'C, roughly between the FMQ and MW (fayalite- 1970)and Gabon are lamproites. magnetite-quartz, magnetite-wustite) buflers, diamond The question of whether or not diamonds from lam- would disappearin about 2l days.In contrast under the proiteshave differentinclusions than thosefrom kimberlite same1[o, but at 800"C diamond would remain for about has been answeredin part from evidenceat Prairie Creek, 110 years. At present the /or-temperature conditions for Arkansas.Diamonds from the Prairie Creek "lamproite" generationof kimberlitic, or lamproitic, magmasat depth contain similar inclusions to those in diamonds from are unknown. Arculus et al. (1982),Haggerty and Tomp- kimberlite-namely, olivine, enstatite,diopside, Cr-pyrope, kins (1983)and Eggler (1983)suggest conditions in the pyrope-almandine,chromite and sulfides (Newton et al., regionof the FMQ-MW bufers. 1977; Pantaleo et al., 1979).It is anticipated that similar Mantle rocks and diamond mineral inclusions will be present in diamonds from the lamproitesin N.W. Australia. A large number of eclogite xenoliths that contain diamond are known (Hatton and Gurney, 1979)but only a Discussion small number of diamondiferousultramafic xenoliths have beenfound. In contrast, the majority of diamondsstudied Relation of diamonil to kimberlite and lamproite contain ultramaflc suite inclusions (Boyd and Finnerty, The presenceof diamonds in both kimberlite and lam- 1e80). proite, the presenceof similar inclusionsin diamondsfrom Diamonds that contain eclogitic suite inclusions, and kimberlite and lamproite, the chemicaldifferences between diamonds that occur in eclogite are obviously easily as- theseinclusions and cognateminerals in the host kimber- signed to an eclogite source rock. Sobolev et al. (1972) lite and lamproite,plus the largedisparity betweenthe ages documentedthis relationship by examining inclusions in of diamond and kimberlite intrusions lead to the con- diamond that itself was contained in an eclogite.The inclusion that diamond is not geneticallyrelated to either clusionsand the constituentminerals of the eclogite(clin- kimberlite or lamproite.The relationshipof thesetwo rock opyroxeneand garnet)were chemically equivalent, suggest- typesto diamond is that they are the transportingmedium ing that the diamond is a constituent of the eclogite and by which diamond ascendsto the earth'ssurface. Diamond thus formed contemporaneouslywith the eclogitehost. is best describedas a xenocryst in kimberlite and lam- In contrast,it is lesseasy to determinethe sourcerock in proite. which diamonds with ultramafic suite inclusions formed. One reasonfor this is that few rocks are known that have Transport of iliamond mineral phasesof comparablechemistry to the inclusions, The processof kimberlite ascentthrough the mantle is Furthermore,the ultramafic xenolithsthat do contain dia- unknown although severalauthors have suggestedvarious mond are chemically and mineralogically varied and in- models which include zone refining (Harris and Middle- clude dunites (garnet * olivine-Pokhilenko et al., 1977), most, 1969),diapirism (Green and Guegen, 1974; Wyllie, harzburgites (garnet + olivine + orthopyroxene- 350 MEYER: GENES/SOF DIAMOND: A MANTLE SAGA

Sobolev, 1974; McCallum and Eggler, 1976), and lherzol- within the same general range (900 to 1300"C; 45 to 65 ites (garnet + olivine + clinopyroxene + kbar) as those obtained for xenoliths and megacrysts,in- orthopyroxene-Dawson and Smith, 1975; Shee et al., cluding thosewhich contain diamond.Thus it appearsthat 1982). Diamond-bearing dunites consist of olivine and diamonds with inclusionshave equilibrated in the mantle garnet whose chemistries are somewhat similar to those of at depths between140 and 200 km; within the bounds of the ultramafic suite inclusions from Siberian diamonds the asthenosphereand roughly coincidingwith the thermal (Pokhilenko et al., 1977). Thus some Siberian diamonds maximum suggestedby Anderson (1980)to occur in the may have been derived from the disaggregation of dunitic temperatureprofile of the upper mantle. rocks, but such dunites have not been recognized elsewhere Cry stallization of diamond in the world. In contrast Shee et al. (1982), with specific reference to the Finsch kimberlite, South Africa, report no The manner in which diamond grows is a subject of similarity in chemistry between the constituent minerals of controversy.It has beensuggested that diamond forms in a a diamond-bearing garnet lherzolite and inclusions in loose solid state i.e., metamorphic reaction (Meyer and Boyd, diamonds from the same kimberlite. 1969;Boyd and Finnerty,1980), or is the productof some Any model of diamond genesis for the ultramafic suite form of metasomatism(Shee et al., 1982)or is of igneous diamonds also must explain the chemical variation, on a origin and crystallizedfrom a magma (Meyer and Boyd, worldwide basis, of some of the mineral inclusion types. 1972;Harte et al., 1980;Meyer, I982a,b). For example, clinopyroxene inclusions show a range in The evidenceis conflicting. The small size of the in- Cal(Ca * Mg) values from sub-calcic(<40 Cal(Ca + Mg)) clusions plus the fact that most are monominerallic led to normal diopsides, while some are Cr-rich (Sobolev et al., Meyer and Boyd (1972\to suggestthat diamonds crystal- 1975).It is unlikely that the sub-calcic diopside inclusions lized from a melt. It is possiblethat the chemicaldissimi- would have formed in the same environment as the Cr-rich larity (particularly with regard to the ultramafic suite) of ones, and thus it seems probable that diamond forms in inclusionsin diamondsand mineralsfrom the host xeno- more than one rock type of ultramafic character. liths is a result of the inclusionshaving formed early on the The aforementioned differences in chemistry and miner- liquidus and having been removed from further chemical alogy of the ultramafic xenoliths and inclusions, as well as reactionswith the liquid by the armoring effect of the dia- the variety of xenolith types (dunite, harzburgite, lherzolite, mond. Similar views have been expressedby Fesq et al. etc.) that contain diamond are also paralleled by the miner- (1975)and Robinson(1978). alogical variety of eclogite xenoliths. Diamonds witb coes- Although Gurney et al. (1979)and Harte et al. (1980) ite inclusions (Milledge, 1961; Sobolev et al., 1976) and consider diamond to be an igneous phase they relate its coesiteeclogite (Smyth and Hatton, 1977)most likely have genesisdirectly to the kimberlite magma. In view of the formed under conditions chemically distinct from those of disparity betweenkimberlite agesand the age of diamond, ruby-bearing diamond (Meyer and Gubelin, 1981) or cor- growth from a kimberlite magmais untenable. undum eclogite. The presenceof alternatingstratigraphy of Type I and II Further evidence in support of diamond having grown in diamond suggestsdiscontinuous growth and minor chemi- several rock types within the broad classification ofeclogite cal variation in the environmentin which growth occurred. and ultramafic rock is the variation of dr3C values. noted Clues as to whetherthe growth processwas either igneous earlier, for eclogitic and ultramaflc suite diamonds. The or metamorphicmay be found through a detailedstudy of view that diamond is a constituent of several mantle rocks the relationship betweenmineral inclusions and the host has also been expressedby Sobolev (1974). diamond strateigraphy. On the basis of various geothermometersand barometers,particularly those of O'Neil and Wood (1979),Boyd growth Pressures and temperatures of diamond and Finnerty (1980)cautiously suggest that the majority of The proto- and syngenetic inclusions in diamond are all diamonds with ultramafic suite inclusionshave formed in phasesthat are stable at high pressuresand temperatures, subsolidusevents. The main reason for this conclusionis and are compatible with formation in the stability field for the fact that most inclusions in diamonds have equili- diamond. Estimates of equilibration conditions for dia- brationtemperatures below 1150'C. monds are based on the chemistries of the inclusions and Metasomatic processesin the mantle are currently utilize the various geothermometers and barometers that widely discussed(Bailey, 1984; Boettcher et al., 1979; are also used in determining pressure and temperature re- Wyllie, 1980)although recentlyWalker (1983)has cau- gimes for xenoliths (e.g., Boyd, 1973; MacGregor, 1974; tioned, quite correctly, against solving all geochemical Lindsley and Dixon, 1975; Wells, 1977; O'Neil and Wood, problemsof the mantle by invoking the "deusex machina" 1979). Ideally for thermobarometric calculations three of metasomatism.Metasomatising agents relevant to dia- coexisting and touching inclusions should be present in the mond formation are variouscombinations of C, H and O. diamond. This is seldom the case and accordingly some Experimental studies of COr-HrO-peridotite systems assumptions have to be made. In spite of probable errors in (Eggler, 1977; 1978;Wyllie, 1977, 1978;Eggler and Wend- the use of these geothermobarometers, the equilibration landt, 1979;Ellis and Wyllie, 1980)have provided signifi- temperatures and pressures obtained for diamonds are cant data and modelsfor the genesisand eruption of kim- MEYER: GENESISOF DIAMOND: A MANTLE SAGA 351

berlite magma (e.g.,Wyllie, 1980).However, the formation speculativemodel of Dickey et al. (1983)provides a ready of kimberlite with or without CO, has little to do with the source of carbon. In this model elementalcarbon in the much earlier genesisof diamond.Furthermore, if most dia- lower mantle is in the form of a metallic liquid. Subsequent mondsare Archaeanor Proterozoic,it could be questioned perturbations causeupward migration of the carbon re- as to whetherpresent thermal modelsof the mantle (Sleep, sulting in the formation of mantle plumes.Oxidation of the 1979;Anderson, 1980) are valid for discussionsof the gen- carbon produces CO, which causespartial melting and esisof diamond. carbonation of the mantle silicate rocks. Theoretically,if Carbon in a melt is unlikely to be in elementalform elementalcarbon ascendedfrom the lower mantle it should (Olaffson and Eggler, 1983)but rather as some dissolved pass through the hexagonaldiamond phase-lonsdaleite speciessuch as CO, or CHo. Whether CO, or CHo is (Bundy and Kasper, 1967; Frondel and Marvin, 1967) presentin the melt dependson the redox conditions.Stud- beforeinverting to cubic diamond at lower pressures.How- ies of ilmenite (Haggerty and Tompkins, 1983; Arculus et ever, the absenceof lower mantle phasesas mineral in- al., 1982)and megacrysts(Eggler, 1983) suggest /., values clusionssuggests it is unlikely that the carbon of Dickey et in the mantle to range from fayalite-magnetite-quaftzto al. (1983)directly forms diamond. magnetite-wustite.These conditions possibly do not reflect Recycledcrustal carbon in the form of carbonatein the those of the early mantle which may have been more re- mantle may be presentwithin regions of subduction.It is duced (Haggerty and Tompkins, 1982). Sobolev et al. stableto appropriatedepths (Wyllie, L978,1980) and under (1981)have reportedmetallic Fe inclusionsin Siberiandia- the correct conditions diamond should be produced.The monds. In the event of a more reduced mantle, perhaps influence of such carbon is unknown, but generation of equivalent to iron-wustite bufer, the predominant dis- some diamondsfrom recycledcarbon (say eclogitic types), solvedspecies in a magma would be CHo (Woerman and and other diamond from primitive carbon (mostly ultrama- Rosenhauer,1982; Eggler and Baker, 1982;Eggler et al., fic types)would accountfor various chemicaland isotopic 1980;Ryabchikov et al., 1981).The occurrenceof CHn in features.If this is the caseand if eclogitic diamonds are the mantle and its role in diamond formation has beenthe also very old (i.e. Proterozoicor older) then a corollary is subjectof modellingby Deines(1980). that subductionmust have beenoperative during the early It has been hypothesized(Gold, 1979; Gold and Soter, historyof the earth. 1979)that the deep mantle is being degassedof methane. Melton and coworkers(1973, 1974)reported the presence Summary of CO, and CHn among gasesthat were evolved from Various partial melting or metasomaticevents in which diamondsthey crushedbetween steel anvils in a massspec- dissolved CHo or CO, were present, resulted in crys- trometer.They found indications that other organic com- tallization of diamond at depths of approximately 180 km plexesof C and H were present,but their results require in the upper mantle. Isotopic data, especially that of independentsubstantiation. As noted earlier, no macro- carbon, are compatiblewith the possibility that diamonds scopic fluid or gaseousinclusions have been observedin iormed from carbon whosesources were isotopically differ- diamond, although CO, has been recorded in fluid in- ent. Variation in isotopecontent could havebeen produced clusionsin olivines from xenoliths in kimberlite (Roedder, by inhomogeneityin primitive mantle material,the pres- 1963,1982,1984). ence of recycled crustal carbon, or both. The growth of In summarythe evidencegenerally favors the formation diamond was not unique to a singlerock but took placein of diamond either directly from an igneousmelt or from any mantle material in which chemical interactions pro- some type of metasomatisingfluid that has pervaded duced elemental carbon. Consequentlyseveral types of various types of mantle rock, perhapsat different times.It rock broadly grouped under eclogiteand ultramafic (peri- is not clear whether the carbon was derived from CO, or dotite) are the original cognatehosts of diamond.Evidence CHn but the presenceof CHn in diamond,metallic iron as suggeststhat this crystallization of diamond occurred in + inclusions,and Cr2 in olivinesincluded in diamond indi- the Archaean or Proterozoic.Subsequent mobilization of catereduced conditions which would favor CH.. mantle diapirs and associatedphysical and thermal pertur- bation of the proximal mantle rocks resultedin the incor- Source of mantle carbon poration of diamond and xenoliths. Dependingupon the The source of the carbon that forms diamond is an locationofthe diapirismrelative to the lithosphere,various enigma.Variations in d13Cvalues for diamondsmay reflect magmatic end products could have been produced.Kim- either an isotopically inhomogeneousmantle or the pres- berlite or lamproite magmas,formed as a consequenceof enceof recycledcrustal material, or both. A similar com- diapiric processes,conveyed diamond rapidly to the crust. ment is valid for the 3He/4He and 404r/364'r values,al- Rapid ascentwas possibly through crack propagation in though more data are obviouslyneeded in order to remove the mantle, and through a thick cool lithosphere whose ambiguities. Meteorites, particularly the carbonaceous thermal regimeresulted in little lossof diamond. variety,have a rangeof 613Cvalues that is similarto that for diamond (Robertand Epstein,1982). Acknowledgments In addition to continual degassingof the mantle in CHn I thankDrs. F. R. Boyd,D. Eggler,T. Evans,I. D. MacGregor, and CO2 as suggestedby Gold (1979),the intriguing and R H. Mitchell, M. Moore, J. P. F. Sellschop and P. J. Wyllie for 352 MEYER: GENESISOF DIAMOND: A MANTLE SAGA their constructivereviews of the manuscript.Many of the ideas Chesley,F. G. (1942)Investigation of the minor elementsin dia- expressedherein were formulated during the award from NSF of mond. AmericanMineralogist, 27, 20-36. grant EAR76-22698;however, I hastento add that errors in inter- Clement,C. R., Skinner,E. M. W. and Scott-Smith,B. H. (1984) pretation are mine alone. Kimberlite redefined.Journal of Geology,92, 22T228. Craig, H. (1953)The geochemistryof stable carbon isotopes.Ge- Note aildeilin press: ochemicaet CosmochimicaActa, 3, 53-92. After the presentmanuscript was written and acceptedfor pub- Dawson, J. B. (1980)Kimberlites and Their Xenoliths. Springer- lication, Richardsonet al. (Science,310,198-202,1984) have pre- Verlag,New York. sentedisotopic evidencethat chrome-pyrop€garnet inclusions(ul- Dawson,J. B. and Smith, J. V. (1975)Occurrence of diamond in a tramafic suite inclusions)in diamondshave model agesof 3.2 to mica-garnetlherzolite xenoliths from kimberlite. Nature, 254, 3.5 b.y., whereasthe host South Africa kimberlites were intruded 580-581. roughly90 m.y.B.P. Deines, P. (1980)The carbon isotopic composition of diamond: relationship to diamond shape,color, occurrenceand vapor composition.Geochimica et CosmochimicaActa, 44,943-961- Deines,P. (1982)The relationshipbetween inclusion composition Reference and carbon isotopic composition of host diamond. (abstr') Allegre,C. J. (1982)Diapirism and magma genesis:an instability Terra Cognita,2,2O2. analysis.(abstr.) Terra Cognita,2, 239. DesCloizeaux,A. (1855)Note sur le diamant noir. Annalles des Anderson,D. L. (1982)The role of kimberlite in mantle evolution. Mines Memoires,8, 30,t-306. (abstr.)Terra Cognita,2, 239. Dickey, J. S., Bassett,W. A., Bird, J. M. and Weathers,M' S. Anderson, O. L. (1980) The temperature profile of the upper (1983)Liquid carbonin the lower mantle.Geology,ll,2l9-22o. mantle.Journal of GeophysicalResearch, 85, 7003-7010. Du Toit, A. L. (1906)Geological survey of the easternportion of Arculus, R. J., Dawson, J. B., Mitchell, R. H. and Gust, D. A. Griqualand West. Eleventh Report Geological Commission (1982)The intrinsic oxygen fugacities(/Or's) of megacrystil- CapeColony, 89-176. menitesfrom southernAfrica kimberlites,Types A and B spinel Eggler,D. H. (1977)The principle of the zone of invariant vapor peridotitesfrom San Carlos, Arizona. (abstr.)Terra Cognita, 2, composition: An example in the system 228. CaO-MgO-SiO2lOz-H2O and implications for the mantle Bailey,D. K. (1984)Kimberlite: "The Mantle Sample"formed by solidus. Yearbook of the CarnegieInstitution of Washington, 't6,428435. ultra metasomatism.In Kornprobst, Ed., Kimberlites I: Kim- berlite and RelatedRocks, p. 323-334.Elsevier, New York. Eggler, D. H. (19?8)The effect of CO2 upon partial melting of Bardet, M. G. (19'14)Geologie du diamant. vol. II. Memoires du peridotite in the system NarO{aO-Al2O3-MgO-SiO2{O2 BureauRecherches Geologie et Mineralogie.No.83, Paris. to 35 kb, with an analysisof melting in a peridotite-HzO{.O2 Bardet,M. G. (1977)Geologie du diamant. vol. III. Memoiresdu system.American Journal of Science,278,301343. BureauRecherches Geologie et Mineralogie.No.83, Paris. Eggler,D. H. (1983)Upper mantle oxidation state: evidencefrom Bauer,M. (1896)Precious Stones.2 vols. Reprint oftranslation by olivine-orthopyroxene-ilmeniteassemblages. Geophysical Re' L. J. Spencer,1904. Dover Publications,New York, 1968. searchLetters, 10, 365-368. Becker, R. H. (1982)Nitrogen isotopic ratios of individual dia- Eggler, D. H. and Baker, D. R. (1982)Reduced volatiles in the mond samples. (abstr.) Fifth International Conference Ge- systernC-o-H: implicationsto mantle melting,fluid formation, ochronology, Cosmochronology, Isotope Geology. Nikko, and diamond genesis.In S. Akimoto and M. Monghnani, Eds., Japan. High PressureResearch in Geophysics,p.237-250. Center for Becker, R. H. and Clayton, R. H. (1977) Nitrogen isotopes in AcademicPublications, Tokyo. rocks. (abstr.) Transactions of the American Geophysical Eggler, D. H. and Wendlandt, R. F. (1979)Experimental studies Union, 58,536. on the relationship between kimberlite magmas and partial Berman, R. (1965) Physical Properties of Diamond. Clarendon melting of peridotite.In F. R. Boyd and H. O. A. Meyer, Eds., Press,Oxford. Kimberlites, Diatremes and Diamonds: Their Geology, Pet- Black, D. C. (1972)On the origins of trapped heliurn,neon, and rology and Geochemistry,p. 330-338.American Geophysical argon isotopic variations in meteorites-I. Gas-rich meteorites, Union, Washington,D.C. lunar soil and breccia.Geochemica et CosmochimicaActa, 36, Eggler,D. H., Baker, D. R. and Wendlandt, R. F. (1980)/O, of 347,375. the assemblagegraphite-€nstatite-forsterite-magnesite: experi- Boettcher,A. L., O'Neil, J. R., Windom, K. E., Stewart,D. C. and ment and application to mantle/O2 and diamond formation. Wilshire,H. G (1979)Metasomatism and the genesisof kimber- (abstr.)Geological Society of America Program with Abstracts, lites and alkali basalts.In F. R. Boyd and H. O. A. Meyer, Eds., 12,420. The Mantle Sample:Inclusions in Kimberlites and Other Vol- Ellis, D. E. and Wyllie, P. J. (1980)Phase relations in the system canics,p. 173 182.American GeophysicalUnion, Washington, MgO-SiOr-HrO-CO2 at pressuresup to 100 kbar and pet- DC. rologicalimplications. American Mineralogist, 65, 5'!0-556. Bonney,T. G. (1899)The parent rock of the diamond in South Evans, T. and Qi, Z. (19E2\The kinetics of the aggregationof Africa. Proceedingsofthe Royal Society,65,223-236. nitrogen atomsin diamond.Proceedings ofthe Royal Societyof Boyd, F. R. (1973)A pyroxenegeotherm. Geochimica et Cosmo- London,A381, 159-178. chimica Acta. 37. 2533-2546. Fesq, H. W., Bibby, D. M., Erasrnus,C. S., Kable, E. J. D. and Boyd, F. R. and Finnerty, A. A. (1980)Conditions of origin of Sellschop,J. P. F. (1975)A comparativetrace elementstudy of natural diamonds of peridotite affrnity.Journal of Geophysical diamondsfrom Premier,Finsch and Jagersfonteinmines, South Research.85. 6911-6918. Africa.Physics and Chemistryofthe Earth,9,817-836. Bundy, F. P. and Kasper,J. S. (1967)Hexagonal diamond-a new Field, J. E. (1979\The Propertiesof Diamond. Academic Press, form of carbon.Journal of ChemicalPhysics, 46,3437-346. New York. MEYER:GENESTSOF DIAMOND: A MANTLE SAGA 353

Frank, F. C. (1956)On the X-ray diffraction spikesof diamond. Hatton, C. J. and Gurney,J. I. (1979)A diamond-graphiteeclogite Proceedingsof the Royal Societyof London, A37,168-174. from the Roberts Victor mine. In F. R. Boyd and H. O. A. Frank, F. C. (1966)Defects in diamond.Proceedings International Meyer, Eds.,The Mantle Sample: Inclusionsin Kimberlites and Industrial Diamond Conference,p. 119-135. Industrial Dia- Other Volcanics, p. 29-36. American Geophysical Union, mond Information Bureau,London. Washington,D.C. Frondel, C. and Marvin, U. B. (1967)Lonsdaleite, a hexagonal Hervig R. L., Smith, J. V., Steele,I. M., Gurney, J. J., Meyer, H. polymorph of diamond.Nature, 214,587-589. O. A. and Harris, J. W. (1980)Diamonds: Minor elementsin Futergendler,S. I. (1956)The study of inclusionsin diamondsby silicate inclusions: Pressure-ternperatureimplications. Journal the method of x-ray structural analysis.(in Russian)Zapiski of Geophysical Research,85, 69194929. VsesoyznyiMineralogia Obshchestva, 85, 568-656. Holmes, A. (1936)A contribution to the petrology of kimberlite Futergendler,S. L (1958)X-ray study of solid inclusionsin dia- and its inclusions.Transactions of the Geological Society of mond (in Russian)Soviet Physics-{rysrallography, 3, 49U97. South Africa, 39,379427. Futergendler,S. L (1960)X-ray study of solid inclusionsin the Hughes,C, J. (1982)Igreous Petrology.Elsevier, New York. Ural and Yakutia diamonds.Materials Vsesoiuznogo Nauchno- Jacques,A. L., Gregory,G. P., Lewis,J. D. and Ferguson,J. (1982) Issledovatel'skogoGeologicheskogo Instituta, no. zt0, 3_82 (in The ultrapotassicrocks of the West Kimberley region, Western Russian). Australia, and a new class of diamondiferouskimberlitc. (Ex- Galimov, E. M., Kaminsky, E. V. and Ivanovskaya,I. N. (1928) tendedabstr.) Terra Cognit4 2,251152. Study of carbon isotop€ compositionsof diamonds from the Jacques,A. L., Lewis, J. D., Smith, C. B., Gregory, G. P., Fergus- Urals, Timan, Sayany,Ukraine and other regions.(in Russian) on, I., Chappell, B. W. and McCulloch, M. T. (1984) The Geokhimiy43,3fi-349. diamond-bearing ultrapotassic (lamproitiQ rocks of the West Gold, T. J. (1979)Terrestrial sourcesof carbon and earthquake Kimberley region,Western Australia. In Kornprobst, Ed., Kim- outgassing.Journal of PetroleumGeology, l, 1-19. berlite I: Kimberlites and related rocks, 223-254.Elsevier, New Gold, T. J. and Soter, S. (1979) Brontides: Natural explosive York. noises.Science, 2M, 37l-37 5. Kaiser, W. and Bond, W. L. (1959)Nitrogen, a major impurity in Goppert,H. R. (1862)Uber Einschlusein diamant.Breslau. cornmontype I diamond.Physics Review, f 15,857-863. Green,H. W. and Guegen, Y . (1974)Origin of kimberlite pipesby Kaminsky, F. V. (1980)Authentic and valid diamond occurrences diapiric upwelling in the upper mantle. Nature physical Sci- in alkali-basaltoidsand ultramafic (non-kimberlitic)bodies. (in enes,249,617-620. Russian)Zapiski VsesoyrrmyiMineralogia Obshchestva,109, Giibelin, E. (1952) Inclusions in diamonds. Journal of Gem- 488-493. mology, 3, 174-187. Kaminsky, F. V. and Gevorkin, P. G. (1976)Non-kimberlitic pri Gurkina, G. A., Ivanovskaja,I. N., Kaminski, F. V. and Galimov, mary diamonds. (in Russian) Izvestia An. Armenia SSR, 2, E. M. (1979)Geokhimiy4 1897-1905. 32-q. Gurney, J. J. and Boyd, F. R. (19E2)Mineral intergrowths with Kennedy, C. S. and Kennedy, G. C. (1976) The equilibrium polycrystalline diamonds from the Orapa Mine, Botswana. boundary betweengraphite and diamond. Journal Geophysical Yearbook of the CarnegieInstitution of Washington,8l,267- Research,81, 24'l l-247 4. 273. Kennedy,G. C. and Nordlie, B. E. (1968)The genesisof diamond Gurney, J. J. and Switzer,G. S. (1973)The discoveryof garnets deposits.Economic Geology, 63, 495-503. closely related to diamonds in the Finsch pipe, South Africa. Knopf, D. (1970)Les kimberliteset les rochesapparentees de Cote Contributionsto Mineralogyand Petrology,39, 103-116. d'Ivoire. SODEMI, Abidjan. Gurney, J. J., Harris, J. W. and Pickard, R. S. (1979)Silicate and Kramers, J. D. (1979)Pb, U, Sr, K, and Rb in inclusion-bearing oxide inclusionsin diamonds from the Finsch kimberlite pip€. diamonds and mantle-derivedxenoliths from southern Africa. In F. R. Boyd and H. O. A. Meyer,Eds., Kimberlites, Diatremes Earth and PlanetarySciences Letters,42, 58-70. and Diamonds: Their Geology,Petrology and Geochemistry,p. Lang, A. R. (1979) Internal structure. In J. E. Field, Ed., The 1-15.American Geophysical Union, Washington,D.C. Propertiesof Diamond, p.425469. AcademicPress, New York. Haggerty, S. E. (1982)Kimberlites in WesternLiberia: An over- Lewis, C. (1887)On a diamondiferousperiodotite and thc genesis view of the geologicalsetting in a plate tectonic framework. of the diamond.Geological Magazine, 4,22-24. Journal of GeophysicalResearch, 87, l08f f-10826. Lightowlers,E. C. and Dean, P. J. (1964)Measurement of nitrogen Haggerty,S. E. and Tompkins, L. A. (1983)R€dox stateof Earth's concentrationsin diamond by photon activation analysisand uppermantle frorn kimberlitic ilmenites.Nature, 3O3,295-ZW. optical absorption. Diamond Research1964, p. 21-25. Sup- Harger, H. S. (1905)The diamond pipes and fissuresof South plementto Industrial Diamond Review,London. Africa. Transactions of the GeologicalSociety of South Africa, Lindsley, D. and Dixon, S. H. (1975) Coexisting diopside and 8. 110-134. enstatiteat 20 kbar and 900-1200"C.American Journal of Sci- Harris, J. (1979) W. and Gurney,l. J. Inclusionsin diamonds.In J. ence,276,1285-1301. E. Field, Ed., The Propertiesof Diamond, p. 555-594.Academic (1942) Press,New York. Lonsdale,K. Extra reflexionsfrom the two types of diamond. Proceedingsofthe Royal Society,4179,315-320. Harris, P. G. and Middlemost, E. A. K. (1969)The evolution of kimberlites.Lithos, 3, 77-88. MacGregor, l. D. (1974) The system MgO-A.lrOr-SiOr: Solu- Harrison, E. R. and Tolansky,S. (1964)Growh history of a natu- bility of AlrO, in enstatitefor spineland garnetperidotite corn- ral octahedraldiamond. ProceedingsRoyal Societyof London, positions.American Mineralogist, 59, I 10-l 19. 41279,490496. McCallister,R. H., Meyer, H. O. A. and Aragon, R. (1979)Partial Harte, 8., Gurney, J. J. and Harris, J. W. (1980)The formation of thermal history of two exsolvedclinopyroxenes from the Thaba peridotitesuite inclusions in diamonds.Contributions to Miner- Putsoa kimberlite pipe, Lesotho. In F. R. Boyd and H. O. A. alogy and Petrology,72, l8l-190. Meyer, Eds.,The Mantle Sample:lnclusions in Kimberlitesand 354 MEYER: GENESISOF DIAMOND: A MANTLE SAGA

Other Vofcanics, p. 244-248. American Geophysical Union, Moore, M. (1979)Optical studiesof diamond and their surfaces:A Washington,D.C. reviewof the late ProfessorTolansky's work. In J. E. Field, Ed., McCallum, M. E. and Eggler,D. H. (1976)Diamonds in an upper The Propertiesof Diamond, p.243-277. AcademicPress, New mantle periodotite nodule from kimberlite in SouthernWyom- York. ing. Science,192, 253-256. Newton, M. G., Melton, C. E. and Giardini, A. A. (1977)Mineral McCulloch, M. T., Jacques,A. L, Nelson, D. R. and Lewis, J. D. inclusionsin an Arkansasdiamond. American Mineralogist,62, (1983)Nd and Sr isotopesin kimberlites and lamproites from 583-586. WesternAustralia: an enrichedmantle origin. Nature, 3O2,4W- Olaffson,M. and Eggler,D. H. (1983)Phase relations of amphibo- 403. le, amphibole-carbonate,and phlogopite-carbonateperidotite: McGetchin, T. R. and Ullrich, G. W. (1973)Xenoliths in maars petrologic constraintson the asthenosphere.Earth and Plane- and diatremeswith inferencesfor the Moon. Mars and Venus. tary ScienceLetters, 64, 305-315. Journalof GeophysicalResearch, 78, 1833-1853. O'Neil, H.St. C. and Wood, B. J. (1979)An experimentalstudy of Mercier, J.-C. C. (1979)Peridotite xenoliths and the dynamicsof Fe-Mg partitioning betweengarnet and olivine and its calibra- kimberlite intrusion. In F. R. Boyd and H. O. A. Meyer, Eds., tion as a geothermometer.Contributions to Mineralogy and The Mantle Sample:Inclusions in Kimberlites and Other Vol- Petrology,70,59-70. canics,p. 197-212.American GeophysicalUnion, Washington, Orlov, Yu. L. (1973)The mineralogy of the diamond. Izdatel'stva D.C. Nauka. (Transl.,J. Wiley and Sons,New York, 1977). Melton, C. E. and Giardini, A. A. (1974)The composition and Ozima, M. and Zashu, S. (1983)Primitive helium in diamonds. significanceof gas releasedfrom natural diamondsfrom Africa Science,219, 1067-1068. and Brazil. AmericanMineralogist, 59, 77 5-782. Ozima, M., Zashu, S. and Nitoh, o. (1983)tHeloHe ratio, noble Melton, C. E. and Giardini, A. A. (1980)The isotopecomposition gas abundanceand K-Ar dating of diamonds-An attempt to of argon includedin an Arkansasdiamond and its significance. searchfor the recordsof early terrestrialhisrory. Geochimicaet GeophysicalResearch Letters, 7, 461464. CosmochimicaActa, 47, 2217-2224. Melton, C. E., Salotti, C. A. and Giardini, A. A. (1974)The obser- Pantaleo,N. S., Newton, M. G., Gogineni, S. V., Melton, C. E. vation of nitrogen,water, carbon dioxide,methane and argon as and Giardini, A. A. (1979)Mineral inclusionsin four Arkansas impurities in natural diamond. American Mineralogist, 57, diamonds: their nature and significance. American Mineral- 1518-1523. ogist,64,1059-1062. Meyer, H. O. A. (1968)Inclusions in diamond. Yearbook of the Pokhilenko, N. P., Sobolev,N. V. and Lavrentyev,Yu. G. (1977) CarnegieInstitution of Washington,66, 44450. Xenoliths of diamantiferousultrarnafic rocks from Yakutian Meyer,H. O. A. (1982a)Mineral inclusionsin natural diamond. kirnberlies.(abstr.) Second International Kimberlite Conference. International GemologicalSymposium Proceedings, p. 441465. AmericanGeophysical Union, Washington,D.C. GemologicalInstitute of America,Los Angeles. Prinz, M., Manson, D. V., Hlava, P. F. and Keil, K. (1975)In- Meyer, H. O. A. (1982b)The genesisof diamond. (abstr.)Terra clusions in diamonds. Garnet lherzolite and eclogite assem- Cognita,2,199. blages.Physics and Chemistryofthe Earth,9,79'l-815. Meyer, H. O. A. and Boyd, F. R. (1969)Inclusions in Diamond. Raal, F. A. (1957)A spectrographicstudy of the minor element Yearbook of the CarnegieInstitution of Washington,68, 315- contentof diamond.American Mineralogist, 42,35+361. 324. Reynolds,J. H., Frick, U., Neil, J. M. and Phinney,D. L. (1978) Meyer, H. O. A. and Boyd, F. R. (1972)Composition and origin of Rare gas-rich separatesfrom carbonaceouschondrites. Ge- crystallineinclusions in natural diamonds.Geochimica et Cos- ochimicaet CosmochimicaActa, 42, 1775-1797. mochimicaActa. 36. 1255-1273. Robert, F. and Epstein,S. (1982)The concentrationand isotopic Meyer,H. O- A. and Giibelin,E. (1981)Ruby in diamond.Gems compositionof hydrogen,carbon and nitrogen in carbonaceous and Gemology,l7, 153-156. meteorites.Geochimica et CosmochimicaActa,46, 81-95. Meyer, H. O. A. and Tsai, H.-M. (1976a)Mineral inclusionsin Robertson,R., Fox, J. J. and Martin, A. E. (1934)Two types of natural diamond: their nature and significance.Mineral Science diamond. Philosophical Transactions of the Royal Society, and Engineering, 8, 242-261. A232,463-538. Meyer, H. O. A. and Tsai, H.-M. (1976b)Mineral inclusionsin Robinson, D. N. (1978)The characteristicsof natural diamond diamond: Temperatureand pressureof equilibration, Science, and their interpretation.Minerals Scienceand Engineering,10, 191,849-851. 55-'t2. Milledge,H. J. (1961)Coesite as an inclusionin C. E. C. synthetic Roedder,E. (1963)Studies of fluid inclusionsII: freezingdata and diamonds.Nature, 190, 1181. their interpretation.Economic Geology, 58, 167 -211. Milledge, H. J., Mendelssohn,M. J., Seal,M. Rouse,J. E., Swart, Roedder,E. (1982)Fluid inclusions in gemstones:Valuable de- P. K. and Pillinger, C. T. (1983)Carbon isotopic variations in fects. International Gemological Symposium Proceedings,p. spectraltype II diamonds.Nature, 303,791-792. 479-502.Gemological Institute of Americ4 Los Angeles. Mitchell, R. H. (1975)Theoretical aspects of gaseousand isotopic Roedder,E. (1984)Fluid Inclusions.Reviews in Mineralogy, Vol. equilibria in the systemC-H-O-S with application to kimber- 12.Mineralogical Society of America. lite. Physicsand Chemistryof the Earth, 9, 903-915. Ryabchikov,I. E., Green,D. H., Wall, V. J. and Brey,G. P. (1981) Mitchell, R. H. (1985)A review of the mineralogy of lamproites. The oxidation state of carbon in the reduced-velocityzone. TransactionsGeological Society ofSouth Africa, in press. GeochemistryInternational, 18, 221-232. Mitchell, R. H. and Lewis,R. D. (1983)Priderite-bearing xenoliths Seal, M. (1965) Structure in diamonds as revealedby etching. from the Prairie Creek mica peridotite, Arkansas. Canadian AmericanMineralogist, 50, 105-123. Mineralogist,21, 59-64. Sellschop,J. P. F. (1979)Nuclear probes in physical and geo- Mitchell, R. S. and Giardini, A. A. (1953) Oriented olivine in- chemicalstudies of natural diamonds.In J. E. Field, Ed., The clusionsin diamond.American Mineralogist, 38, 136-138. Propertiesof Diamond, p. 107-163.Academic Press, New York. MEYER: GENESISOF DIAMOND: A MANTLE SAGA 355

Scott Smith, B. H. and Skinner,E. M. W. (1984)A new look at Suzuki, S. and Lang, A. R. (1976)Internal structuresof natural Prairie Creek, Arkansas. In Kornprobst, Ed., Kimberlites I: diamonds revealing mixed growth habit. Diamond Research Kimberlitesand relatedrocks, p. 255154. Elsevier,New York. 1976, p. 39 47. Supplementto Industrial Diamond Review, Shee, S. R., Gurney, J. J. and Robinson, D. N. (1982) Two London. diamond-bearingperidotie xenoliths from the Finsch kimber- Takaoka, N. and Ozima, M. (1978) Rare gas isotopic compo- lite, South Africa. Contributions to Mineralogy and Petrology, sitionsin diamond.Nature, 271,4546. 81,79-87. Trueb, L. F. and Barrett, C. S. (1972) Microstructural investi- Skinner,E. M. W. and Clement,C. R. (1979)Mineralogical classi- gation of ballas diamond. American Mineralogist, 57, 1664- fication of southern African kimberlites.In F. R. Boyd and H. 1680- O. A. Meyer, Eds., Kimberlites, Diatremes and Diamonds: Trueb,L. F. and De Wys,E. C. (1969)Carbonado:Natural poly- Their Geology, Petrology and Geochemistry, p. 129-139. crystallinediamond. Science, 165, 799-802. AmericanGeophysical Union, Washington,D.C. Trueb, L. F. and De Wys, E. C. (1971)Carbon from Ubangi-a Sleep,N. H. (1979)Thermal history and degassingof the earth: microstructuralstudy. American Mineralogist, 56, 1252-1268. Somesimple calculations. Journal of Geology,87,671-686. Wagner, P. A. (1914)The Diamond Fields of Southern Africa. Smyth,J. R. and Hatton, C. J. (19'17)A coesite-sanidinegrospydite Reprinted l97l,C. Struik (fty) Ltd, CapeTown. from the Roberts Victor kimberlite. Earth and Planetary Sci- Walker, D. (1983)New developrnentsin magmaticprocesses. Re- encesLetters, 34, 28+290. viewsof Geophysicsand SpacePhysics, 21,1372-1384. Sobolev,N. V. (1974)Deep-seated inclusions in kimberlites and Wand, U., Nitzsche,H.-M., Muhle, K. and Wetzel,K. (1980)Ni- the problem of the composition of the upper mantle. (Trans- trogen isotope composition in natural diarnond-First results. lation by American Geophysical Union, Washington, D.C., Chemieder Erde, 39,85-87. r977). Wells, P. R. A. (1977)Pyroxene thermometry in simple and com- Sobolev,N. V., Ellimova, E. S. and Pospelova,L. N. (1981)Native plex systems.Contributions in Mineralogy and Petrology, 62, iron in Yakutian diamonds, its paragenesis.(in Russian)Ge- 129-139. ology and Geophysics,Doklady Akademii Nauk SSR,Siberian Wickman, F. E. (1956)The cycle of carbon and stable carbon Division. 12.25-29. isotopes.Geochimica et CosmochimicaActa,9, 13G153. Sobolev,N. V., Efimova, E. S., Koptil, V. I., Lavrent'yev,Yu. G., Williams, A. F. (1932)The Genesisof Diamond, 2 vols. London: and Sobolev,V. S. (1976)Inclusions of coesite,garnet and om- E. Benn. phacite in Yakutian diamond: the first discovery of coesite Wilson, A. N. (1982)Diamonds from birth to eternity. Gemologi- paragenesis.(in Russian)Doklady Akademii Nauk SSSR,230, cal Institute of America,Los Angeles. 1442-1446. Woerman,E. and Rosenhauer,M. (1982)Oxygen fugacity and the Sobolev,N. V., Galimov, E. M., Ivanovskaya,I. N. and Yefimova, fluid C-H-O phasein the earth'smantle. (abstr.) Terra Cognita, E. S. (1979)The carbon isotopecompositions of diamondscon- 2,227. taining crystallographicinclusions. (in Russian)Doklady Aka- Wyflie, P. J. (1977)Peridotite-Co2-HrO and carbonatiticliquids demiiNauk SSSR,249,1217-1220. in the upper asthenosphere.Nature, 266,4547. Sobolev,V. S., Sobolev,N. V. and Lavrent'yev,Yu. G. (1972) Wyllie, P. J. (1978) Mantle fluid compositions buffered in Inclusions in diamond extractedfrom a diamondiferouseclo- peridotite-CO2-HrO by carbonates,amphibole, and phlogo- gite.(in Russian)Doklady AkademiaNauk SSSR,2O7, l&-167. pite. Journal of Geology,86, 687-713. Sobolev,V. S., Sobolev,N. V. and Laurent'yev,Yu. G. (1975) Wyllie, P. J. (1980)The origin of kimberlite. Journal of Geophysi- Chrome-rich clinopyroxenesfrom the kimberlites of Yakutia. cal Research,85, 6902-6910. NeuesJahrbuch fiir Mineralogie,Abhandlungen, 123, 213-218. Sutton, J. R. (1921)Inclusions in diamonds from South Africa. MineralogicalMagazine, 9, 2O8-21O. Sutton, J. R. (1928)Diamond: A descriptivetreatise. T. Murby, Manuscript receioed, February 9, 1984; London. acceptedfor publication, Nouember 15, 1984.