Author's personal copy Mineralogy of the lower mantle: A review of 'super-deep' mineral inclusions...

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Mineralogy of the lower mantle: A review of ‘super-deep’ mineral inclusionsin diamond

Felix KaminskyKM Diamond Exploration Ltd, West Vancouver, British Columbia, Canada

a b s t r a c ta r t i c l e i n f o

Article history:Received 3 November 2010Accepted 19 October 2011Available online 15 November 2011

Keywords:Lower mantleD″ layerDiamondFerropericlaseStishoviteTAPP

Starting from the late 1980s, several groups of lower-mantle mineral inclusions in diamond have been found.Three associations were established among them: juvenile ultramafic, analogous to eclogitic, and carbonati-tic. The juvenile ultramafic association strongly predominates, and it is composed of ferropericlase, MgSi-,CaSi- and CaTi-perovskites, stishovite, tetragonal almandine-pyrope phase (TAPP), and some others. The as-sociation analogous to the upper-mantle eclogitic association, formed from subducting lithosphere, com-prises: majorite, CaSi-perovskite bearing compositional Eu anomalies, phase ‘Egg’ with a tetragonalstructure, and stishovite. The carbonatitic association is represented by various carbonates, halides, and asso-ciated minerals. Some mineral associations (wüstite+periclase and native iron+iron carbides) are, possibly,related to the D″ layer at the core/mantle boundary. The mineralogical composition of the lower mantle isnow understood to be more complex than had been suggested in theoretic and experimental works. The pro-portion of ferropericlase in the lower mantle is higher than it was suggested before, and its composition ismore iron-rich (mg=0.36–0.90) as compared to experimental and theoretical data. Free silica (stishovite)is always present in lower-mantle associations, and a separate aluminous phase (TAPP) has been identifiedin several areas. These discrepancies suggest that the composition of the lower mantle differs to that of theupper-mantle, and experiments based solely on ‘pyrolitic’ compositions are not, therefore, applicable to thelower mantle. These data indicate a probability of an alternative to the CI-chondrite model of the Earth'sformation, for example, an enstatite-chondrite model.

© 2011 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1282. History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

2.1. General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1282.2. South Africa and South Australia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1282.3. Brazil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1292.4. Guinea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1292.5. Canada . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1292.6. Other suggested localities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

3. Minerals and mineral associations in the lower mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1303.1. General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1303.2. Juvenile ultramafic mineral association . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

3.2.1. General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1303.2.2. Ferropericlase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1303.2.3. MgSi-perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1313.2.4. CaSi-perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1333.2.5. CaTi-perovskite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1343.2.6. Stishovite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1343.2.7. Tetragonal almandine-pyrope phase (TAPP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

Earth-Science Reviews 110 (2012) 127–147

E-mail address: [email protected].

0012-8252/$ – see front matter © 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.earscirev.2011.10.005

Contents lists available at SciVerse ScienceDirect

Earth-Science Reviews

j ourna l homepage: www.e lsev ie r .com/ locate /earsc i rev


3.2.8. Phase with a composition of olivine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1363.2.9. Spinel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1363.2.10. Manganoilmenite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1363.2.11. Titanite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1373.2.12. Native nickel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1373.2.13. Native iron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1373.2.14. Magnetite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1373.2.15. Unidentified Si–Mg phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1383.2.16. Majorite garnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1383.2.17. Moissanite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138

3.3. Analog of the eclogitic association . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1383.3.1. General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1383.3.2. Phase ‘Egg’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1393.3.3. Stishovite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1393.3.4. Additional indications of the ‘eclogitic’ association in the lower mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139

3.4. Carbonatitic association . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1393.4.1. General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1393.4.2. Calcite and dolomite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1403.4.3. Nyerereite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1403.4.4. Nahcolite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1413.4.5. Halides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1413.4.6. Other minerals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

4. Minerals suggested from deeper levels of the Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1414.1. Wüstite and periclase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1414.2. Iron carbides and native iron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

5. Composition of the lower mantle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1436. Discussion and conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145Appendix A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

1. Introduction

It has long been assumed that our view of the structure and compo-sition of the lowermantle (over a half of the Earth's total volume) reliedsolely upon indirect methods, such as geophysical and/orexperimental data, direct observation being out of thequestion. Howev-er, in the last two decades, a series of lower mantle minerals has beendiscovered, encapsulated and preserved in diamond crystals formed inthe lower mantle and delivered to near-surface regions by kimberliteeruptions. While these most frequently appear as single minerals theysometimes form paragenese with other phases. Some of these wereformed before their being entrapped by diamonds while the remainderwere probably encapsulated by the diamonds formed in lower-mantleprovenance melts and fluids. Irrespective of these considerations, theyprobably represent material from the deepest levels of the Earth avail-able for direct investigation. Compositionally, such lower-mantle min-erals resemble those either predicted theoretically or synthesizedexperimentally under ultra-high pressure conditions (20–25 GPa andhigher), corresponding to depths of 500–600 km and greater.

However, the observed diamond-bearing phases are not identical tothe experimentally-produced phases, and are characterized by specificchemical features and, sometimes, distinct crystallographic structures.

At the present time being, the number of lower-mantle material‘samples’ available for study is very limited, less than, for example, thenumber of Lunar samples retrieved by the Apollo and Luna missions.This does not diminish the scientific value of lower-mantle mineralsamples, however, as they allow for the preliminary reconstruction ofreal compositions and conditions of formation in the deepest zones ofthe Earth providing critical information relating to early stages of theEarth's formation. Accordingly, it is appropriate to summarize the exist-ing relevant data as a basis for formulatingmodels and drawing prelim-inary conclusions. With these goals in view, the author has made use ofmaterial obtained primarily from Brazil and Venezuela, in conjunctionwith the existing experimental data and published materials fromother regions, where similar material has been found.

2. History

2.1. General

Experiments performed in the 1980s demonstrated the presence a se-ries of phase transitions in the Earth's interiorwithin the pressure interval20–25 GPa. The most important of these was: (Mg,Fe)2SiO4→(Mg,Fe)O+(Mg,Fe)SiO3 (wadsleyite, ringwoodite→ferropericlase+ ‘MgSi-perov-skite’)which occurs at c. 24 GPa andmarks a transition from the upper tothe lower mantle corresponding to a depth of 650–670 km (Ringwoodand Irifune, 1988). According to this reaction, the lower-mantle indicatorassociation is: ferropericlase+a (Mg,Fe)SiO3 phase with the structure ofperovskite (‘MgSi-perovskite’). Ferropericlase, without this associationhas been identified as an inclusion in diamond from a number of areas;however, these finds cannot be considered as a fully reliable indicator ofa lower-mantle paragenesis.

2.2. South Africa and South Australia

The first reported occurrence of ferropericlase+(Mg,Fe)SiO3 phase(described as enstatite)was identified in diamond from theKoffifonteinkimberlite pipe, South Africa (Scott Smith et al., 1984). Subsequently,the compositional data were confirmed by new analyses of ferroperi-clase and “enstatite” (the primary perovskite structure of which hadnot yet been established) in diamond from the same pipe (Mooreet al., 1986; Scott Smith et al., 1984). Two ferropericlase grains werealso found in association with‘enstatite’ from a diamond in the Orroroodyke, South Australia. However, these were extracted from different di-amond crystals and cannot be reliably considered as paragenetic. How-ever, lower-mantle minerals (associations of ferropericlase, MgSi-perovskite, and ‘olivine’) were also recently identified in diamondfrom a kimberlitic dyke (K7) and ancient placer deposits from thesame area (Tappert et al., 2009) suggesting that the Orroroo area ofSouth Australia is the second, reliable source of lower-mantle material(Fig. 1).

128 F. Kaminsky / Earth-Science Reviews 110 (2012) 127–147


2.3. Brazil

In the early 1990s, numerous grains of ferropericlase, MgSi-perovskite, CaSi-perovskite, and other minerals were identified inthe São Luiz placer deposit (Mato Grosso State, Brazil) and subjectedto detailed petrographic study (Wilding et al., 1991; Harte and Harris,1994; Harris et al., 1997; Harte et al., 1999; Hutchison et al., 2001).These studies led a systematic study of lower-mantle mineral associ-ations from the Juina area where the São Luiz placer deposit is located(Fig. 1).

Lower-mantle assemblages were not only found in this deposit,but also in diamonds from within the same area from the rivers:Mutum, Vermelho, Chicoria (Kaminsky et al., 2001), and Soriso(Hayman et al., 2005). Lower-mantle diamonds have been also re-cently been discovered in primary kimberlitic sources in the vicinityof these rivers. These kimberlites, referred to as the Pandrea pipes,host diamonds that also contain lower-mantle mineral inclusions(Kaminsky et al., 2009a, 2010) as do diamonds from an earlierknown pipe (Collier-4) from the same area (Kaminsky et al., 2009a;Bulanova et al., 2010) and diamond from the Machado River placerdeposit, located approximately 200 km west of the Juina area(Fig. 1) (Bulanova et al., 2008). The Juina area is thus one of themajor sources of geochemical data relating directly to lower-mantlematerial.

2.4. Guinea

Lower-mantle inclusions in diamond have also been identified inthe Kankan placer deposit in Guinea (Fig. 1), where they predominateover upper-mantle associations (Stachel et al., 2000, 2002). The lowermantle assemblages mainly comprise grains of ferropericlase, in somecases coexisting with MgSi- and CaSi-perovskites and SiO2 (invertedfrom stishovite).

2.5. Canada

The fifth area, where lower-mantle minerals have been identifiedwithin diamond, is the Slave kimberlitic province in the NorthwestTerritories of Canada. Most of these superdeep inclusions are foundin kimberlite pipe DO27, where they comprise approximately 11% of

all inclusions, the remainder being represented by ‘common’ upper-mantle minerals (Davies et al., 2004). In addition to DO27, diamondfrom other pipes in the area (DO18 and Ranch Lake) has yielded fer-ropericlase and SiO2 as inclusions in diamond, CaSi-perovskite beingincluded in diamond in samples from pipe A-21 (Davies et al.,2004). In one of the major productive kimberlite pipes (Panda) ferro-periclase has been observed in diamond, in association with SiO2,CaSi-perovskite, ‘olivine’, and spinel, comprising c. 5% of all diamondinclusions in this pipe (Tappert et al., 2005a). Another economic dia-mondiferous pipe, A-154S, contains numerous ferropericlase inclu-sions (Donnelly et al., 2007; van Rythoven and Schulze, 2009).

2.6. Other suggested localities

In addition to the five major areas described above, about a dozenoccurrences of single minerals from different areas which cannot beascribed with certainty to a lower-mantle provenance, althoughtheir resemblance to the examples described above should not be dis-counted. The first of such occurrences are ferropericlase inclusions indiamond, unassociated with other and have been reported from theGuaniamo placer deposit in Venezuela (Kaminsky et al., 2000), lam-proites in Arkansas (Newton et al., 1977) and kimberlites fromSloan (Otter and Gurney, 1989) in USA, South African pipes such asthe Monastery, South Africa (Moore et al., 1986), Letseng-la-Terai, Le-sotho (McDade and Harris, 1999), River Ranch, Zimbabwe (Kopylovaet al., 1997) and Mwadui, Tanzania (Stachel et al., 1999). In addition,ferropericlase has been observed as single inclusions in placer de-posits in north-eastern Yakutia (Sobolev et al., 1999) and microdia-monds from the Udachnaya pipe (Zedgenizov et al., 1998, 2001) inRussia. In addition to ferropericlase, the Monastery pipe diamondhas inclusions of majoritic garnet (Moore and Gumey, 1985), and inthose of the Jagersfontein pipe, in South Africa, high-Si majorite gar-net (XSi up to 3.5) along with a phase with the composition of olivine(Chinn et al., 1998; Tappert et al., 2005b), all of which could be con-sidered as potential lower-mantle phase constituents.

In addition to the diamond-bearing phases, megacrysts of badde-leyite from the Mbuji Mayi kimberlites have been suggested as havingoriginated in the lower-mantle, presumably having transformed froma cubic ZrO2 lower mantle structure to their present upper mantlemonoclinic lattice (Kerschhofer et al., 2000; Schärer et al., 2011).

Fig. 1. Finds of lower-mantle material on the Earth.

129F. Kaminsky / Earth-Science Reviews 110 (2012) 127–147


Microprobe analyses identified majorite garnet in association withhigh-Al (4.57 wt.% Al2O3) MgSi-perovskite, CaSi-perovskite, picroil-menite, diamond, and other minerals from xenoliths and garnetmegacrysts in alnoitic pipes and sills in the Malaita Island (OntongJava Plateau, Southwest Pacific). These led Collerson et al. (2000) toinfer a lower-mantle origin for these phases although subsequent an-alytical mineralogical studies raised some doubts concerning this in-terpretation (Neal et al., 2001).

Interestingly, mineral phases of deep-seated origin have been ob-served in ophiolitic ultramafic sequences in the Swiss Alps and Tibet.For example, three previously unknown structures of FeTiO3 wereidentified in peridotites from the Arami Massif, Switzerland, indicat-ing that the originally-exsolved phase was a high-pressure perovskitepolymorph of ilmenite, implying a minimum depth of origin of300 km (Dobrzhinetskaya et al., 1996). Moreover, in chromititesfrom the Luobusa ophiolite, southern Tibet, high-pressure nitridesand oxides were identified in association with diamond, moissanite,boron carbide, phase TiO2-II, native iron, and other minerals. This as-sociation indicates their formation at the depths of greater than300–400 km (Dobrzhinetskaya et al., 2009). It seems possible, there-fore, that some ultramafic rocks may have lower-mantle origins.

3. Minerals and mineral associations in the lower mantle

3.1. General

Geological data have shown that the mineral composition of thelower mantle is more complex than previously thought, as shown inTable 1, listing lower-mantle minerals in diamond from Brazil,Guinea, and Canada identified to date. Only minerals confirmed aslower-mantle phases are included in the table: ferropericlase andMgSi-perovskite (as lower-mantle indicator mineral phases) and min-erals associated with these in a single diamond grain, frequently as in-tergrowths (‘touching association’) (Table 2). Some of the listedminerals (e.g., ‘olivine’ and, possibly, titanite) have, in fact, anotherstructure. Three mineral associations can be distinguished for thelower-mantleminerals: juvenile, ultramafic, analogous (but not similar)to the ultramafic upper-mantle association; analogous to the eclogiticassociation of the upper mantle; and carbonatitic association.

The juvenile association almost certainly predominates. Its composi-tion, as shown below, does not resemble that of upper-mantle modelssuch as ‘pyrolite’. Along with the identified lower-mantle minerals,some minerals of possibly deeper provenance have been identifiedamong inclusions in diamond. These are the wüstite+periclase as-sociation and iron carbides and nitrocarbides that, it may be conjec-tured, formed within the D″ layer at the core/mantle boundary.

3.2. Juvenile ultramafic mineral association

3.2.1. GeneralMajor rock-forming minerals of the upper-mantle juvenile ultra-

mafic association are olivine, orthopyroxene, clinopyroxene, andpyrope-almandine garnet, which at the transition zone are replacedby majorite garnet, orthorhombic wadsleyite [β-(Mg,Fe)2SiO4], andringwoodite with a spinel-type structure [γ-(Mg,Fe)2SiO4]. Accordingto experimental data, isochemical transformations in both olivine-wadsleyite and wadsleyite-ringwoodite take place at pressures of12–16 GPa and 18–22 GPa respectively (Fig. 2). At pressures of c.24 GPa and higher, wadsleyite and ringwoodite decompose to formferropericlase and a phase (Mg,Fe)SiO3 that has a perovskite-typestructure (‘MgSi-perovskite’) (e.g., Ringwood and Irifune, 1988;Chudinovskikh and Boehler, 2001). This mineral association is prima-ry indicator of lower-mantle conditions (Fig. 2). Under the samepressure–temperature conditions, majorite transforms into MgSi-perovskite with a possible admixture of stishovite and an Al-phase(e.g. Irifune et al., 1996). Under the same conditions, phase CaSiO3

with a perovskite-type structure (‘CaSi-perovskite’) and other min-erals listed in Table 1 are stable as well.

The relationship of each mineral to the juvenile ultramafic associ-ation is defined by its coexistence (frequently touching intergrowth)with ferropericlase and MgSi-perovskite. To date, such definite at-tachment to the juvenile ultramafic association is determined forCaSi- and CaTi-perovskite, stishovite, TAPP (tetragonal almandine-pyrope phase), phase with the composition of olivine, spinel andchrome spinel, titanite, picroilmenite and manganoilmenite, nativeiron and native nickel. The occurrences of these minerals are pre-sented in Table 2, from which it is clear that they all are closely relat-ed to one another, forming a single association.

3.2.2. FerropericlaseFerropericlase (Mg,Fe)O (fPer) is the most common mineral of the

juvenile ultramafic association. It forms small (b1 mm) grains of anoctahedral, cubic, or cubic-octahedral habit which can be single inclu-sions in diamond or form intergrowths with MgSi-perovskite, CaSi-perovskite, stishovite (SiO2), TAPP, phase with the composition of ol-ivine, chrome spinel, ilmenite, and native nickel. Ferropericlase grainscomprise, in some areas, up to 50–56% of all identified lower-mantleinclusions. This fact contradicts experimental data, according towhich the ferropericlase share in the lower mantle should compriseonly 16–20%; the rest should belong to MgSi- and CaSi-perovskite(e.g., Irifune, 1994; Fei and Bertka, 1999; Wood, 2000) (Fig. 2). Sucha high content of ferropericlase in the lower mantle may partly beexplained by its higher stability compared to the perovskite phases;however, the main reason for the discrepancy between experimentaldata and real observations is likely caused by the composition of theupper mantle different from that of a ‘pyrolite’ model.

The second discrepancy between experimental data and observedfacts is the composition of ferropericlase, and primarily its high rangein terms of magnesian index,mg=0.36–0.90. According to the exper-iments with natural systems under pressures at 25–60 GPa, the mag-nesian index of ferropericlase in the lower-mantle material withmg=0.80–0.95 should be 0.73–0.88 (e.g. Wood, 2000; Lee et al.,2004). Theoretically, for the most likely lower-mantle magnesianindex, mg=0.89–0.92, the magnesian index of ferropericlase shouldbe localized within a much narrower range, at c. 0.90 (Kesson andFitz Gerald, 1991). However, in the natural environment ferroperi-clase may locally be much more iron-rich, up to mg=0.36 (Fig. 3).

There are regional differences in the magnesian index of ferroperi-clase values fromdifferent regions. For example, ferropericlase fromCan-ada, South Australia and South Africa has mg=0.80–0.90, that is closeto the theoretical values. Ferropericlase from Guinea, along with similarmg values, also has lowermg values of 0.75.Moreover, iron-rich ferroper-iclase from Brazil (reaching the composition of magnesiowüstite) com-prises almost a half of all grains in this region (c. 46.5%). It has beensuggested that such great deviations may relate to the hypothetical de-composition of MgSi-perovskite, to form iron-rich ferropericlase andSiO2 (Fei et al., 1996a, 1996b; Harte et al., 1999); however, this sugges-tion only replaces one discrepancywith another: how, in the lowerman-tle withmg=0.80–0.95, can exist a highly iron-rich MgSi-perovskite.

The ferric iron proportion (Fe3+/ΣFe) in ferropericlase is low; only0–7% of the total iron content (McCammon et al., 1997).

In order to solve both discrepancies (high ferropericlase concen-tration in the lower mantle and its ferruginous composition), Liu(2002) proposed a model of decarbonation of ferromagnesite, havingvariable iron content, with the formation of ferropericlase with a var-iable magnesian index, carbon (in the form of diamond), and oxygen.Not to mention the complexity of this model, the idea that ferromag-nesite occurs in the lower mantle and that suggestion concerning itsvariability in iron-content remain enigmatic. Hence, variations inboth the magnesian index of ferropericlase and its high content inthe lower mantle (compared to experimental and theoretical data)remain unexplained.



Minor elements in ferropericlase are Cr, Ni, Mn, and Na, whichcomprise up to 1–2.5 wt.% each (Table 3). Ni has a close positive cor-relation with mg, while Mn has a negative correlation with mg. Na ispositively related to Cr. Trace elements in ferropericlase have anirregular distribution. Their concentrations vary widely; only enrich-ment in Li, Ta, Th, and U remains constant. REE tend to be enrichedin HREE, but not in all samples (Fig. 4).

3.2.3. MgSi-perovskiteMgSi-perovskite (Mg,Fe)SiO3 (MgSiPrv) associates in diamond

with ferropericlase, CaSi-perovskite, TAPP, phase with the

composition of olivine, chrome spinel, native nickel, and sulfide.Owing to its instability, following extraction from diamond, thisphase underwent retrograde transformation to a pyroxene-typestructure. The retrograde phase inherits an initial perovskite chemicalcomposition, and importantly a high Al content. In contrast to ensta-tite from the upper-mantle, in which Al2O3 usually comprises 0.3–0.9 wt.% and never exceeds 1 wt.% of the mineral, MgSiPrv commonlycontains 1–2.7 wt.% Al2O3, and rarely this value can be in the range of8.3–12.6 wt.%, in a group of grains from São Luiz in Brazil (Table 4).This corresponds to experimental data, according to which, the gar-net–perovskite transition, occurring under 24–26.5 GPa pressure, in

Table 1List of lower-mantle minerals included in diamond.

Brazil Guinea Canada

Rio Sao Luiz Rio Mutum,Vermelho,Chicoria

Rio Soriso Pandreapipes

Collier-4 pipe Machadoarea

Kankanplacerdeposit

DO27, DO18,A21, Pandaet al. pipes

Wilding et al.,1991; Harteet al., 1999

Kaminskyet al., 2001

Hayman et al.,2005; Kaminskyet al., 2009b

Kaminskyet al., 2009a

Kaminsky et al.,2009a; Bulanovaet al., 2010

Bulanovaet al., 2008

Stachelet al., 2000

Davies et al.,2004; Tappertet al., 2005a

Ultramafic associationfPer fPer fPer fPer fPer fPer fPer fPerMgSiPrv MgSiPrv MgSiPrv MgSiPrv MgSiPrv MgSiPrv MgSiPrvCaSiPrv CaSiPrv CaSiPrv CaMgSiPrv CaSiPrv CaSiPrv CaSiPrv CaSiPrv

CaTiPrv CaTiPrv CaTiSiPrvSti Sti Sti ? Sti ? Sti StiTAPP TAPP TAPP TAPP

‘Ol’ ‘Ol’ ‘Ol’ ‘Ol’Spl CrSpl SplMn-Ilm Mn-IlmPilm PilmTit Tit

Ni0 Ni0 Ni0

Fe0 Fe0 Fe0 Fe0

Mag MagSi–Mg phase

Maj Maj Maj MajSulph Sulph Sulph

Mois‘Eclogitic’ association

EggSti

Carbonatitic associationCalDol CaMg-carb Dol

SdNyerNahHlSylHyCotSiO2

TiO2-IIWo-IICuspMtcCa-GrtApAnhIlm‘Ol’SplPhlMagSulph

Minerals possibly from the D″ layerWuPerCohChalHaxFe0



Table2

Coex

isting

lower-m

antlemineralsin

diam

ond.



the MgSiO3–Al2O3 system, has an increase of Al2O3 with an increasingpressure (Irifune et al., 1996; Wood, 2000). If so, the Al2O3 content inMgSi-perovskite may prove to be a barometer for lower-mantle min-eral associations; and, in this case, the São Luiz association is the dee-pest among those observed to date.

For MgSi-perovskite, a low Ni content (or its complete absence) ischaracteristic; Ni is concentrated in ferropericlase associated withMgSiPrv (Kesson and Fitz Gerald, 1991; Stachel et al., 2000). Themost significant minor element in MgSiPrv, in addition to Al, is Cr(Table 4). Water may have a significant role in the composition ofMgSiPrv. From IR data, its solubility in MgSiPrv is 1000–1800 ppm,under a pressure of 25–26 GPa (Litasov et al., 2003). If this is correct,it could imply that the lower mantle contains 3.42°×1021 kg of water,or more than two and a half times that of the of the world's oceans. (Itshould be noted, however, that to date there has been no confirma-tion for this model.)

The value of mg in MgSiPrv, in contrast to ferropericlase, is ratherconstant. In correspondence with a Fe distribution coefficient for thenatural pair fPer–MgSiPrv, Fe content is greater than in the mineralfPer: mgMgSiPrv is usually 0.88–0.95 (Kesson and Fitz Gerald, 1991;Harte et al., 1999). These data do not correspond to the experimentaldata, however, according to which the magnesian index forMgSiPrv issuggested to be lower, at 0.80–0.88 (Fei et al., 1996a, 1996b; Lee et al.,

2004). The absence of less magnesian MgSi-perovskite in nature is,possibly, related to the instability of iron-rich MgSiPrv which decom-poses to iron-rich ferropericlase and SiO2 (Fei et al., 1996a, 1996b;Harte et al., 1999).

There are signs of regional differences in values of mg for MgSi-perovskite from different regions: in Guinea and Canada MgSiPrv ismore magnesian (mg=0.93–0.95) than in Brazil, where less magne-sian samples are present, up to mg=0.86 (Table 4). Fig. 5 representsdata on the partition of ferrous iron Fe2+ in MgSi-perovskite and fer-ropericlase. One can see that the Fe2+-partition is variable, in differ-ent regions: the Brazilian samples are more ferruginous than theGuinean ones, mainly because of the more iron-rich ferropericlase.It is not discounted also that these regional differences relate to vari-ations in Al content in MgSi-perovskite from different regions, be-cause Al content causes an increase of the proportion of ferric ironin MgSi-perovskite (McCammon, 1997). Subsequent experimentalwork has demonstrated that while an increase of Al inMgSiPrv resultsin an increase in the Fe3+ solubility within this mineral, it has nomeasurable effect on the Fe2+ partitioning between MgSiPrv andfPer (Frost and Langenhorst, 2002).

Trace element concentrations in MgSi-perovskite are low. The REEdistribution shows a depletion in LREE (Fig. 6).

3.2.4. CaSi-perovskiteCaSi-perovskite CaSiO3 (CaSiPrv) is stable within the pressure

range from almost 100 GPa to 14–16 GPa (Shim et al., 2000), i.e.,

Fig. 2. Experimental data on mineral composition and phase transformations inthe mantle within a pressure range of 3–35 GPa. After Stixrude and Lithgow-Bertelloni(2007); Kennedy and Kennedy (1976) [diamond/graphite]; Zhang et al. (1996) [coesite/stishovite]; Helffrich and Wood (2001) [mantle geotherm].

Fig. 3. Magnesian index of ferropericlase inclusions in diamond. Only data on provenlower-mantle inclusions are used.

Notes to Table 2Notes: Br-SL — Brazil, placer São Luiz (Wilding et al., 1991; Harte et al., 1999; Kaminsky et al., 2001).Br-Ver — Brazil, placer Vermelho (Kaminsky et al., 2001).Br-Ch — Brazil, placer Chicoria (Kaminsky et al., 2001).Br-Sor — Brazil, placer Soriso (Hayman et al., 2005).Br-Pan — Brazil, Pandrea pipes (Kaminsky et al., 2009a).Бp-К4 — Brazil, Collier-4 pipe (Bulanova et al., 2010).Br-Ma — Brazil, placers Machado (Bulanova et al., 2008).Guin — Guinea, placer Kankan (Stachel et al., 2000).DO27 — Canada, DO27 pipe (Davies et al., 2004).DO18 — Canada, DO18 pipe (Davies et al., 2004).Koffi — ЮAP, pipe Koffifontein (Scott Smith et al., 1984; Moore et al., 1986).K7 — South Australia, K7 dyke (Tappert et al., 2009).Panda — Canada, Panda pipe (Tappert et al., 2005).Ranch — Canada, Ranch Lake pipe (Davies et al., 2004).Spr — South Australia, placer Springfield (Tappert et al., 2009).The use of bold font indicates mineral intergrowth (‘touching association’).



from the lower parts of the lower mantle up to the transition zone.Following its release from diamond, CaSiPrv acquires a retrogradestructure, that of larnite or walstromite (Gasparik et al., 1994;Joswig et al., 1999). CaSiPrv forms small, colorless or milky-whitegrains with a characteristic, for inclusions in diamond, ‘negative’shape. It associates with ferropericlase and MgSi-perovskite, as wellas with CaTi-perovskite, ‘olivine’, titanite, and native Ni (Table 2).

During the course of the retrograde phase transitions, CaSiPrvmaintains an unchanged chemical composition. The composition isremarkably clean, only with minor admixtures of Al2O3, FeO, SrO,and K2O (Table 5).

The REE distribution in CaSiPrv, in contrast toMgSiPrv, has a signif-icant (to one order) enrichment of LREE (e.g., the ratio of La/Yb=3.6–37.7; Fig. 7). Such distribution may be explained by a multiple in-crease of the distribution coefficient for LREE comparatively to HREEwith the increasing pressure (Wang et al., 2000). In such a case, theenrichment of the LREE may be used as a qualitative geobarometer.

Some CaSiPrv grains have a positive Eu-anomaly that reflects thecomposition of subducted lithospheric slabs. Such grains, most likely,belong to the ‘eclogitic’ association (see below).

3.2.5. CaTi-perovskitePerovskite with the ‘common’ (for the lithosphere) CaSiO3 compo-

sition (CaTiPrv) has been identified in single diamond grains from theJuina placers (Kaminsky et al., 2001; Hayman et al., 2005). CaTiPrv oc-curs in association with ilmenite, a Si–Mg phase, and majorite garnet.It has admixtures of SiO2 (1.05–2.06 wt.%), Al2O3 (0.64–1.48 wt.%),FeO (0–0.17 wt.%) and Na2O (0–0.33 wt.%). According to experimen-tal data, CaTiO3–CaSiO3 solid solution is stable within the lower man-tle conditions (Kubo et al., 1997).

3.2.6. StishoviteSiO2 is not characteristic for the ultramafic association of the

upper mantle; it occurs there only as part of an eclogitic association.

Fig. 4. Chondrite-normalized trace element patterns in ferropericlase. Chondrite composition here and in the following spider-diagrams (Figs. 6–8, 10–12) derives fromMcDonough andSun (1995).

Table 4Minor elements in MgSi-perovskite (wt.%).

Oxides Brazil Guinea Canada

Rio São Luiz Rio Mutum. Vermelho and Chicoria Rio Soriso Kankan DO27 pipe

Wilding, 1990; Harte et al., 1999 Kaminsky et al., 2001 Hayman et al., 2005 Stachel et al., 2000 Davies et al., 2004

Al2O3 1.23–2.66/8.3–12.6 2.16 1.60–3.1 0.55–1.68 0.69–2.03Cr2O3 0.1–1.19 0.17 0.20–0.22 0.17–0.29 0.23–0.36NiO 0.01–0.03 b0.06 0.01–0.03mg 0.86–0.95 0.87 0.91–0.94 0.93–0.95 0.93–0.95

Table 3Minor elements in ferropericlase (wt.%).


São Luiz Rio Mutum. Vermelho.Chicoria

Rio Soriso Pandrea pipes Collier-4 pipe Kankan DO27. DO18. A21.Ranch Lake pipes

Panda pipe

Wilding, 1990;Harte et al., 1999

Kaminsky et al.,2001

Hayman et al.,2005

Kaminsky et al., 2009a Kaminsky et al.,2009a

Stachel et al.,2000

Davies et al.,2004

Tappert et al.,2005a

Cr2O3 0.19–2.56 0.11–2.26 0–1.30 0.10–0.65 0.09–0.27 0.12–1.04 0.41–2.56 0.69–0.75NiO 0.10–1.49 0.02–1.37 0.11–1.46 0.59–1.34 1.02–1.48 0.51–1.46 0.33–1.55 1.34–2.14MnO 0.07–0.81 0.19–0.94 0.16–1.46 0.15–0.42 0.14–0.18 0.16–0.26 0.13–0.34 0.12–0.24Na2O 0.02–1.25 0.04–1.35 0–1.20 0.01–0.24 0–0.03 0.07–0.79 0.04–1.11 0.07–0.19mg 0.36–0.87 0.49–0.83 0.45–0.89 0.75–0.81 0.76–0.86 0.75–0.92 0.84–0.90 0.80–0.87



Neither does the model composition of the lower mantle host free sil-ica (Fig. 2). Among lower-mantle inclusions in diamond from all stud-ied regions, SiO2 is, however, present; it occurs in association withcharacteristic lower-mantle minerals, such as ferropericlase andCaSi-perovskite. CaSiPrv is stable at pressures only greater than14–16 GPa, i.e., only within the stishovite stability field, whichmeans that SiO2 in these inclusions should be stishovite. The in-situstishovite structure has been established only for the 'eclogitic' asso-ciation (see below); there have been no studies of the SiO2 structurein a juvenile ultramafic association, to date. One may expect that inthe deepest levels of the lower mantle, SiO2 has an orthorhombicstructure (e.g., Ono et al., 2002).

Because SiO2 is in disequilibriumwith other phases of the ultramaficassociation, and because SiO2 permanently associates with ferroperi-clase, amodel of theMgSi-perovskite dissolutionwith the resultant for-mation of fPer and Sti was proposed: (Mg,Fe)SiO3→(Mg,Fe)O+SiO2.MgSi-perovskite→ ferropericlase+stishovite. Such transformation isexperimentally proven, but only for Fe-rich systems (Stachel et al.,2000 and references therein). Possibly, in the natural, lower-mantle en-vironment, this reaction can explain the presence only of high-Mg,MgSi-perovskite (with mg=0.88–0.96) and Fe-rich ferropericlase (asa dissolution product of primary Fe-rich MgSi-perovskite).

The composition of natural stishovite is close to pure SiO2; somegrains contain insignificant (less than 0.09 wt.%) admixtures of Al,Ti, Cr, Fe, Mn, Ni, Mg, and Ca.

3.2.7. Tetragonal almandine-pyrope phase (TAPP)Small, cubo-octahedral or elongate-tabular, apple-green grains

with a stoichiometric composition similar to pyrope-almandine gar-net were identified in diamond from the São Luiz placer deposit inthe Juina area, Brazil, in association with ferropericlase and MgSi-perovskite (Harte and Harris, 1994). Based on the composition, theylack evidence for majoritic solid solution and are very low-Ca (0.03–0.54 wt.% CaO), as compared to garnet from an upper-mantle assem-blage (either peridotitic or eclogitic). The subsequent crystallographicanalysis of this mineral demonstrated its tetragonal structure, withthe space group I-42d and a=6.526 Å and c=18.182 Å (Harris etal., 1997). The mineral is now referred to as TAPP. It is low in Fe(FeO=4.60–10.06 wt.%; fe=0.09–0.39), with a predominance of fer-ric iron (Fe3+/Fetot=66–74%; McCammon et al., 1997). Consideringan average Fe3+/Fetot ratio to be 70%, the ferrous iron index Fe2+#in TAPP is only 3–7%. The formula of the mineral may be presentedas (Mg,Fe3+)(Al,Cr,Mn)2(Mg,Fe2+)2Si3O12, where the first cationgroup has four-fold symmetry with tetrahedral coordination, andthe second and the third groups have two-fold symmetry with octa-hedral coordination.

Cr is the major admixture in TAPP (1.34–3.00 wt.% Cr2O3), al-though it is very low compared to upper-mantle, peridotitic, garnet(typically 3–10 wt.% Cr2O3). In addition, TAPP contains minor

Fig. 5. Fe2+ partition in coexisting MgSi-perovskite and ferropericlase in diamondsfrom Guinea, Brazil, and RSA (after McCammon et al., 2004; Hayman et al., 2005).The Fe3+ contents are measured from Mössbauer spectra. The line indicates the bestfit for all Fe2+ data (after McCammon et al., 2004).

Fig. 6. Chondrite-normalized trace element patterns in MgSi-perovskite.

Table 5Minor elements in CaSi-perovskite (wt.%).


Rio São Luiz RioVermelhoand Chicoria

RioSoriso

Kankan DO-27and A21pipes

Pandapipe

Wilding et al.,1991; Harteet al., 1999


Haymanet al.,2005

Stachelet al.,2000

Davieset al.,2004

Tappertet al.,2005a

Al2O3 0.05–0.11 0.03–0.11 0–0.26 0.01–0.66 0–0.02 0.09SrO n.a. 0.01–0.06 n.a. 0.06–0.85 n.a. 0.10K2O 0–0.02 0–0.07 0–0.14 0.01–0.73 0–0.07 0.10



admixtures of NiO (0.01–0.07%), MnO (0.05–0.96%), Na2O (0.02–1.13%) and K2O (0–0.02%) (Table 6).

TAPP has yet to have been made in any laboratory experiment forultramafic systems under high pressure conditions. It is the only alu-minous phase to occur in a juvenile ultramafic association. Prior tothe acknowledgment of the presence of an aluminous phase in thelower mantle, all Al was thought to reside in MgSi-perovskite(Irifune, 1994; Kesson et al., 1995).

Some researchers, in confirming the structural features of TAPPand accepting this as a new, non-garnet mineral phase, suggest thatits origin relates to a retrograde process, occurring outside of the sta-bility field of the lower mantle, during the ascent of magmatic mate-rial to the surface (Finger and Conrad, 2000; Brenker et al., 2002). Analternative opinion, however, on the deep-seated origin of TAPP as aphase replacing majorite in the lower mantle (Harte, 2010) seemsmore likely because of a close association (sometimes ‘touching’) ofTAPP with ferropericlase and MgSi-perovskite (see Table 2).

The concentration of trace elements in TAPP (except Nb, Zr, andHf) is below chondritic; the REE distribution, however, is similar tochondrite (Fig. 8).

3.2.8. Phase with a composition of olivineA mineral phase with a composition identical to olivine (Mg,

Fe)2SiO4 (‘Ol’) was identified in diamond from placer deposits of theJuina area in Brazil, where it forms colorless grains, associated with fer-ropericlase, MgSi- and CaSi-perovskites, and TAPP (Wilding et al., 1991;Kaminsky et al., 2001; Hayman et al., 2005), as well as in diamond fromKankan, Guinea where it associates with TAPP (Stachel et al., 2000).These grains have wide variations in Mg (mg=0.88–0.97), includinglow-Mg varieties (mg=0.88–0.89) with low concentrations of Ni,which are not characteristic for usual, upper-mantle olivine (Table 7).

Such low-Mg ‘olivine’ usually associates with ferropericlase whichconcentrates Ni. The formation of this phase is, most likely, related toa retrograde transformation of some lower-mantle phase, or with a ret-rograde reaction fPer+MgSiPrv→Ol. The latter is less probable because‘Ol’ in diamond frequently associates not only with fPer but withMgSiPrv as well (sometimes in a 'touching' association).

More Fe-rich ‘olivine’ (mg=0.79) was identified as a singleinclusion lacking any association with other lower-mantle mineralsin diamond from the South African pipe, Jagersfontein (Chinn et al.,1998). It is likely that this ‘olivine’ belongs to the lower mantle orthe transition zone because in other diamond crystals from thesame pipe are identified majorite garnet inclusions.

3.2.9. SpinelChromian spinel (Mg,Fe)(Cr,Al,Fe)2O4 (CrSpl), in association with

ferropericlase and MgSi-perovskite, was identified in the Chicoriaplacer and within the Pandrea pipes from Juina, Brazil (Kaminsky etal., 2001, 2009a). In comparison to chrome spinel inclusions inupper-mantle diamond, the identified grains are less chromian (35–55 wt.% Cr2O3), while those present in diamond from the placer de-posits have high concentrations of iron (33–34 wt.% FeO) and titani-um (10–11 wt.% TiO2) (Fig. 9).

The magnesian index (mg) of the identified lower-mantle, chromespinel varies in a range from mg=0.35–0.36 in Juina placers tomg=0.42–0.64 in the Pandrea pipes; in both cases this value islower than for that present within lithospheric diamond(mg=0.51–0.80). MREE and HREE values in chrome spinel are closeto chondritic; the distribution is characterized by LREE-depletion(Fig. 10).

Besides chrome spinel, alumospinel has been identified in diamondfrom several kimberlite pipe occurrences: Panda, Canada and Collier-4, Brazil. These have mg=0.76–0.82 (Tappert et al., 2005a; Bulanovaet al., 2010).

3.2.10. ManganoilmeniteManganoilmenite (Mn-Ilm) was known previously only as a late-

or postmagmatic, metasomatic phase in kimberlites, carbonatites,agpaitic and ultramafic pegmatites, and some other rocks. In diamondfrom the Juina area, manganoilmenite was identified, for the firsttime, as a primary-magmatic phase with a homogeneous internalstructure. It associates with ferropericlase, perovskite, majorite gar-net, and the Si–Mg phase (Kaminsky et al., 2001, 2009a; Kaminskyand Belousova, 2009). In one of the diamond crystals from the

Fig. 7. Chondrite-normalized trace element patterns in CaSi-perovskite.

Table 6Minor elements in TAPP (wt.%).

Oxides Brazil Guinea

Rio São Luiz Rio Vermelho Rio Soriso Kankan

Harte et al.,1999

Kaminsky et al.,2001

Hayman et al.,2005

Stachel et al.,2000

TiO2 0.01–0.06 (4.23) 7.55 4.71 1.01Cr2O3 1.34–3.00 1.74 2.74 0.02NiO 0.01–0.07 b0.06 0.03mg 0.83–0.92 0.81 0.87 0.61



Pandrea-7 pipe, manganoilmenite associates with picroilmenite(magnesian ilmenite). In addition to the Juina area, manganoilmenitein diamond is known from the Guaniamo area, Venezuela (Sobolev etal., 1998; Kaminsky et al., 2000).

Manganoilmenite has an almost stoichiometric composition ofilmenite with the exception of unusually high concentration of man-ganese (MnO=0.42–11.46 wt.%) and an elevated admixture of vana-dium (V2O3=0.13–0.43 wt.%) (Table 8). An admixture of zinc doesnot exceed 0.5 wt.% ZnO. Trace element patterns in manganoilmeniteand picroilmenite are demonstrated in Fig. 11. They are similar toeach other. Both manganoilmenite and picroilmenite have high con-centrations of V, Zr, Nb, Ta, Th, and U. Concentrations of Ni, Cr, Co,Ta, and Nb in manganoilmenite are lower than in picroilmenite.

According to calculations with the use of generalized gradient ap-proximation, ilmenite may be stable only in the uppermost levels ofthe lower mantle under the 20–26 Gpa pressure (Yu et al., 2011).

3.2.11. TitaniteTitanite, CaTiSiO5 (Ttn), was identified in placers of Rio Chicoria

(Kaminsky et al., 2001) and Rio Soriso (Hayman et al., 2005) in theJuina area, Brazil, in association with CaSi-perovskite. Structural anal-ysis of this mineral was not performed; it is possible that it is a poly-morph modification of CaTiSiO5. The only chemical analysis of thismineral has an admixture of alumina (3.16 wt.% Al2O3) and iron(0.06 wt.% FeO). Deficit of TiO2 and low totals (93.70%) suggest theexistence of other, non-analyzed elements.

3.2.12. Native nickelNative nickel (Ni0) was identified in diamond from Rio São Luiz

(Wilding et al., 1991) and Rio Chicoria from the Juina area, Brazil, in

association with CaSi-perovskite (Kaminsky et al., 2001). It is alsofound in pipe DO27, Canada, occurring in association with ferroperi-clase and MgSi-perovskite (Davies et al., 2004). Its peculiar chemicalfeature is a noticeable admixture of potassium (in the samplefrom the Chicoria, 0.21 wt.% K2O). Other admixtures are Fe (1.01–1.29 wt.% FeO), Ti (0.02–0.07 wt.% TiO2), Cr (0.01 wt.% Cr2O3), Mn(0.07 wt.% MnO), and Na (0.02 wt.% Na2O).

3.2.13. Native ironNative iron (Fe0) was identified in diamond from Rio Soriso in as-

sociation with low-Ni pyrrhotite (Hayman et al., 2005), from thePandrea-1 pipe in association with ferropericlase and chrome spinel(Kaminsky et al., 2009a), both from the Juina area, Brazil, and fromthe placer Kankan in Guinea (Stachel et al., 2000). The mineral has anoticeable admixture of Cr (1.13–2.37 wt.% Cr2O3), as well as Ni(0.28–0.40 wt.% NiO), Mg (0.16–0.94 wt.% MgO), Mn (0.17–0.47 wt.% MnO), Ti (0.02–0.12 wt.% TiO2) and Al (0.09–0.27 wt.%Al2O3).

3.2.14. MagnetiteMagnetite (Mag) was identified in diamond from the São Luiz and

Soriso placer deposits in Brazil, without any lower-mantle mineral

Fig. 8. Chondrite-normalized trace element patterns in TAPP.

Table 7Minor elements in ‘olivine’ (wt.%).

Oxides Brazil Canada SouthAustralia

Rio Vermelhoand Chicoria

Rio Soriso DO27 pipe Panda pipe Dyke K7


Haymanet al., 2005

Davieset al., 2004

Tappertet al., 2005a

Tappertet al., 2009

TiO2 b0.03 0.05–0.07 0 0.003–0.01 0.02Al2O3 0.03 0.12–0.80 0.03 0.02–0.05 0Cr2O3 0.15–0.24 – 0 0.01–0.03 0.07NiO 0.32–0.38 – 0.08 0.19–0.22 0.47MnO 0.05–0.14 0.12–0.15 0.09 0.06–0.09 0.11mg 0.88 0.89–0.95 0.97 0.94–0.95 0.91

Fig. 9. Fe and Ti (cations) vs. Cr/(Cr+Al)at in chrome spinel.



association (except SiO2). However, a great number of the grains(more than twenty) and some features of their compositions (for ex-ample, 2.10–3.03 wt.% Cr2O3) provide a basis to suggest their lower-mantle origin. Other chemical admixtures in magnetite are Si (0.17–0.28 wt.% SiO2), Ti (2.97–3.61 wt.% TiO2), Al (3.44–6.92 wt.% Al2O3),Mg (2.04–3.58 wt.% MgO), and Na (0–0.11 wt.% Na2O).

3.2.15. Unidentified Si–Mg phaseAn unidentifiedmineral, with the composition of Si–Mg in an approx-

imate atomic ratio of 1.37 was identified in diamond from the Chicoriaplacer in Brazil; it occurs in association with perovskite, manganoilme-nite, andmajorite (i.e., a lower-mantle origin is not obvious). The compo-sition of the mineral consists only of Si (61.95 wt.% SiO2) and Mg(30.33 wt.% MgO), with admixtures of Al (0.10 wt.% Al2O3) and Fe(0.45 wt.% FeO). Considering the low analytical totals (92.71 wt.%) forthis material, one may suggest the presence of volatiles in this phase.

3.2.16. Majorite garnetMajorite garnet (majorite), Mg3(Mg,Fe,Al,Si)2Si3O12 (Maj), in

which an excess of Si in octahedral coordination exists, is the majormineral of the transition zone (e.g., Harte, 2010; Fig. 2). In majoritegrains from the Monastery and the Jagersfontein pipes, South Africa,the silica excess is as high as XSi=3.2–3.6 (Moore and Gumey,1985; Chinn et al., 1998; Tappert et al., 2005b). Some experimentsdemonstrated that majorite is stable under pressures of up to28 GPa (Irifune and Ringwood, 1993); and the association of majorite,in diamond from the Juina placers in Brazil, with perovskite and il-menite (Wilding et al., 1991; Kaminsky et al., 2001) allows also to as-sume its presence in the uppermost horizons of the lower mantle.

The chemical composition of majorite varies within a range ofmg=0.50–0.87; mostly within a range of mg=0.60–0.70 (Table 9).The main admixture is Na (0.27–1.12; in one of the grains 5.89 wt.%Na2O). This corresponds to the experimental data about synthesis of

majorite with up to 12 mol. % of Na in garnet solid solution (Bobrovet al., 2008a; Dymshits et al., 2010). Sodium majorite Na2MgSi5O12,like TAPP, has tetragonal structure where excess Si begins to occupythe octahedral site that is compensated by the entry of Na in the Xsite of the garnet structure (Bobrov et al., 2008b; Bindi et al., 2011).Na-majorite may be an important concentrator of sodium in theupper part of the lower mantle and transition zone.

Concentrations of microelements in majorite are usually one-twomagnitudes higher than in chondrite, except Ni, Ba, and Pb whichform negative anomalies (Fig. 12). The patterns of REE in majoritefrom different regions are also distinct (Fig. 12). Majorite from Guineaand Brazil have almost linear, chondritic-like distribution (with asmall predominance of LREE in majorite from Guinea), while majoritefrom South Africa (pipes Monastery and Jagersfontein) is significantlydepleted with respect to LREE; the most contrasting distribution forthe samples is present in those from the Jagersfontein pipe.Manymajor-ite grains have a positive or a negative Eu anomaly, most likely indicat-ing their ‘eclogitic’ genesis with formation of these grains likely to haveoccurred within subducted lithosphere. The other majorite grains, thatlack Eu-anomalies, may belong to the juvenile ultramafic association.

3.2.17. MoissaniteMoissanite SiC (Mois) was identified only in diamond from the São

Luiz placer deposit (Wilding et al., 1991) and does not associate withother minerals. It may, therefore, only be suggested to have formed inthe lower mantle.

3.3. Analog of the eclogitic association

3.3.1. GeneralTectonic reconstructions based on geophysical data (e.g., Helffrich

and Wood, 2001) provide a model for the super-deep (perhaps downto the core/mantle boundary) subduction of lithospheric plates. During

Fig. 10. Chondrite-normalized trace element patterns in chrome spinel.

Table 8Chemical characteristics of manganoilmenite.

Characteristic Juina, Brazil Carbonatites fromJacupiranga, Brazil

Other igneousrocks

Metamorphic mafic and ultramaficrocks, W. Australia

Inclusions in diamond Kimberlites

MnO, wt.% 0.42–11.46 0.63–2.49 2.29–7.87 1.44–15.15 1.15–7.38MgO 0.0.81 0–0.24 15.69–23.46 0.0.46 0.01–0.49TiO2 50.13–56.15 55.49–57.79 55.83–59.24 46.23–52.02 42.80–52.66V2O3 0.13 0.21–0.43 – 0.06 0.01–0.43NiO 0–0.12 0–0.02 4.56–6.80 – –

MnTiO3, mol% 0.9–23.5 1.3–5.0 4.2–14.6 3.2–40.4 3.0–16.4MgTiO3 0–2.0 0–0.6 52.6–75.6 0–6.0 0.1–2.3



the course of subduction, of a plate with a MORB-like composition,under pressures of greater than 10 GPa, the eclogitic garnet–pyroxeneassociation is replaced by an association ofmajorite+stishovite.Withinthe slabs, boundaries of perovskite-forming reactions displace to higherpressures (comparing to the juvenile lowermantle) as a consequence oftheir negative P–T slopes (e.g., Saxena, 2010). This association remainsstable to some 25–28 GPa pressure, when majorite transforms intothe association: CaSi-perovskite+MgSi-perovskite+stishovite. Insome experiments with a basaltic starting-composition, an unidentifiedaluminous phasewas observed, andmajorite remained stable in associ-ationwithMgSi-perovskite (Irifune and Ringwood, 1993; Irifune, 1994;Hirose et al., 1999; Akaogi, 2007). Under the same conditions, in a hy-drous system, a newhydrous aluminum silicate AlSiO3OH (initially sug-gested as Al5Si5O17OH) was produced (Egglton et al., 1978): AlOOH+SiO2→AlSiO3OH (diaspor+stishovite→phase ‘Egg’). This mineralwas termed phase ‘Egg’ (Schmidt et al., 1998; Ono, 1999; Sano et al.,2004); its stability was estimated at up to 40 GPa (Vanpeteghem et al.,2003). For this reason, phase ‘Egg’, in associationwith stishovite, is con-sidered as an analog of the eclogitic association in the lower mantle.

A natural association of phase ‘Egg’+stishovite was established indiamond from Rio Soriso, Juina area, Brazil (Wirth et al., 2007). Bothminerals were identified with the aid of Raman spectroscopy andelectron diffractometry carried out in situ, owing to which the miner-al structures remained unchanged.

3.3.2. Phase ‘Egg’Phase ‘Egg’ AlSiO3OH forms an aggregate of euhedral, cubic and

cubo-octahedral grains, 20–30 to 200–300 nm in size (Fig. 13). TheSi/Al ratio in phase “Egg” is approximately 1:1 (46.74–53.84% and45.63–52.60 respectively), and a minor admixture of Fe (0.29–1.52%) is present. The presence of OH in phase ‘Egg’ was identifiedwith an aid of pre-peak at about 528 eV in the electron energy-lossspectrum, and, for the Raman spectra, in the OH-stretching region be-tween 1800 cm−1 and 3000 cm−1. The presence of OH in the

composition of the mineral allows this to be considered as one ofthe principal water reservoirs in the deep Earth.

Phase ‘Egg’ has predominantly a tetragonal crystal structure(ao=0.7435 nm and co=0.706 nm); only in one sample has it beenfound to display a monoclinic structure. In this respect, it differs tothat of phase ‘Egg’ studied before (Schmidt et al., 1998), and so forthis reason this is not strictly phase ‘Egg’ but a polymorph of this min-eral with relics of phase ‘Egg’ present in some grains.

In some of the Raman-spectra maxima, in the areas of 2150 cm−1

and 2500 cm−1, features were detected, which are characteristic for aphase δ-AlOOH. A coexistence of this phase with phase ‘Egg’ points tothe most likely pressure interval of their joint origin at 23–27 GPa, i.e.in the uppermost horizons of the lower mantle (Sano et al., 2004).

3.3.3. StishoviteStishovite with a tetragonal structure forms subordinate spheroi-

dal grains, c. 20 nm in size. Chemically it is almost pure silica with aminor amount of Al, which is characteristic for stishovite formedunder an elevated pressure (Ono, 1999; Sano et al., 2004).

Numerous pores of irregular shape are present within this stisho-vite (Fig. 13A). These contained a fluid which was released to the vac-uum of the FIB during TEM specimen preparation. Quench products ofthe fluid contain minor amounts of K, Ca, Ba, P, S, Cl and F, that werealso detected along the walls of the pores. This porosity, possibly, re-flects the initial stage of the degassing of phase ‘Egg’ at the pressureconditions near its stability boundary.

3.3.4. Additional indications of the ‘eclogitic’ association in the lowermantleIn addition to the observed phase ‘Egg’+stishovite association,

some CaSi-perovskite grains and numerous majorite grains havebeen shown to contain either positive or negative Eu-anomaly, intheir REE distribution curves (Figs. 7 and 12). These anomalies mayrelate to the composition of subducted lithosphere, and such anoma-lous grains, most likely, belong to the ‘eclogitic’ association.

3.4. Carbonatitic association

3.4.1. GeneralIn upper-mantle diamond, single finds of carbonates (e.g., Sobolev

et al., 1997), sometimes in association with sylvite, apatite, and mica(Klein-BenDavid et al., 2006; Logvinova et al., 2008) have beenreported. Sylvite, probably, has been identified previously in two oc-tahedral diamond crystals from Shandong Province, China, where itwas described as a “high-potassium and high-chlorine inclusion(K=48–58 wt.%; Cl=37–43 wt.%)” in association with “silicate

Fig. 11. Chondrite-normalized trace element patterns in manganoilmenite.

Table 9Minor elements in majorite (wt.%).

Oxides Brazil

Rio São Luiz Corigo Chicoria Collier-4 pipe

Harte et al., 1999 Kaminsky et al., 2001 Bulanova et al., 2010

Al2O3 12.71–20.59 17.22–18.88 21.10–22.22Cr2O3 0.03–0.27 (2.96) 0.07–0.12 0.03–0.07Na2O 0.27–1.12 (5.89) 0.61–0.88 0.40–0.98mg 0.57–0.87 0.60–0.63 0.50–0.60



mineral inclusions”, that were suggested to represent as a product of theliquid phase trapped during the primary stage of diamond growth (Chenet al., 1992). A carbonatitic association of minerals (e.g., carbonates andhalides) was identified recently in diamond from Rio Soriso, Juina area,Brazil (Brenker et al., 2007; Wirth et al., 2009; Kaminsky et al., 2009b)(Table 1). Carbonatitic minerals associate (sometimes closely, forming‘touching’ associations) with typical lower-mantle minerals (such asCaSi-perovskite and ferropericlase; Brenker et al., 2007; Kaminsky et al.,2009b), and so may be attributed to the lower mantle, as well.

The carbonatitic association is now believed to represent lower-mantle material, which earlier was identified as chloride–carbonate,gas–fluid inclusions in diamond from different regions (e.g., Izraeliet al., 2001). Those inclusions are a product of the two evolutionaltrends among initial magmatic melts, silicate–carbonate melt andcarbonate–chloride melt (Navon et al., 2003). The existence of thetwo immiscible melts which are similar to the melt and brine inclu-sions in natural diamond has been proven experimentally (Safonovet al., 2007). The mineral association described below belongs to thecarbonate–chloride evolutional line of the primary, lower-mantle liq-uid. It seems likely that, in the future, products of the second, silicate–carbonate line may also be found.

A peculiar feature of the carbonatitic association is the presence ofminerals enriched with volatiles (nahcolite, halides, cuspidine, anhy-drite, and phlogopite). In combination with porosity found in some

minerals (e.g., coesite) this feature evidences the great role of volatilecomponents during the course of formation of the lower-mantle car-bonatitic association.

3.4.2. Calcite and dolomiteCalcite and dolomite were found not only in Rio Soriso diamond,

but (as single grains) in the Machado area, Brazil, as well (Bulanovaet al., 2010). Calcite is characterized by an admixture of Sr(0.74 wt.% SrO) and Ba (0.0nwt.% SrO).

3.4.3. NyerereiteNyerereite Na2Ca(CO3)2 (Nyer) is a rare mineral, unstable at sur-

face conditions, which was identified before only in the recent natro-carbonatitic lavas of Oldoinyo Lengai volcano, Tanzania (Dawson,1962), within chloride–carbonate nodules from the Siberian pipe,Udachnaya (Kamenetsky et al., 2006), as well as being present in aninclusion within perovskite from the Guli carbonatite complex inNorthern Siberia (Kogarko et al., 1991). In a diamond from Rio Soriso,nyerereite associates with calcite, apatite, wollastonite-II, phase withthe composition of olivine, spinel, and wüstite (Fig. 14) The chemicalcomposition of nyerereite in diamond is very similar to that fromother localities, with only a slightly lower calcium content and higherconcentration of potassium (Table 10; Fig. 15).

Fig. 12. Chondrite-normalized trace element patterns in majorite.

Fig. 13. Electron-microscope images of an aggregate of phase ‘Egg’+stishovite with an extensively porous structure (A) and a single, cubic crystal of phase ‘Egg’ (B). After Wirth etal. (2007).



3.4.4. NahcoliteNahcolite NaHCO3 (Nah), in natrocarbonatites, is one of the inter-

mediate products of nyerereite hydratation. In Juina diamond, it wasidentified from a characteristic Raman spectrum.

3.4.5. HalidesA series of halideswas identified in diamondwithin a large inclusion

of SiO2. Halides are represented by halite, NaCl (Hl), sylvite, KCl (Syl),hydrophilite, CaCl2 (Hy), and cotunnite, PbCl2 (Cot). Silica groundmasshas presently a retrograde, coesite structure; originally it was, mostlikely, stishovite. Anhydrite and plattnerite associate with the halides.

3.4.6. Other mineralsPhase TiO2-IIwas identified in a close (‘touching’) associationwith ha-

lite. It is a rutile polymorph with an orthorhombic structure α-PbO2,which is characteristic for high pressure conditions (Withers et al.,2003). The mineral has admixtures of FeO (0.21–0.44 at.%) and Cr2O3

(0.09–0.13 at.%), as well as micro-admixtures Al and V (not quantified).Wollastonite-II CaSiO3 (Wo-II) forms intergrowths with calcite, cus-

pidine and monticellite; in another case with nyerereite and Ca-garnet. Its composition is (in at.%): Ca=49.7; Si=50.1; Fe=0.2. Withthe ratio of Ca/Si close to 1, its structure appeared to be not perovskitic,as one could expect, but triclinic, corresponding to wollastonite-II.

Cuspidine Ca4Si2O7F2 (Cusp) forms close intergrowths with monti-cellite and wollastonite-II. Its chemical composition (in at.%) is:Si=25.8–26.0; Ca=49.0–49.4; F=24.6–25.2; with the ratio Ca/Si c.2:1. Unit cell dimensions are: a0=0.755 nm; b0=1.043 nm;c0=1.06 nm. Judging by mineral interrelations, cuspidine replaceswollastonite-II.

Monticellite CaMgSiO4 (Mtc) has a chemical composition (in at.%):Si=29.8; Ca=33.6; Mg=23.9; Fe=11.4 (mg=0.83), with minoradmixtures of Al (0.7), Ti (0.3), and Mn (0.3). Ca-garnet (Ca-Grt)with an admixture of Zr, according to the cell parametera0=1.2365, belongs to the andradite-kimzeyite-schorlomite series.

Apatite Ca5(PO4)3(F,Cl) (Ap) always associates with calcite andnyerereite. It contains admixtures of La, Ce, and Nd.

Among other minerals, in the carbonatitic association, anhydrite,ilmenite, the mineral phase with the composition of olivine, spinel(magnesioferrite), phlogopite, magnetite, and Fe–Cu sulfides (with anadmixture of Ni) are present.

4. Minerals suggested from deeper levels of the Earth

Two peculiar mineral associations were found in a diamond con-taining the carbonatitic association: wüstite+periclase and iron car-bides+native iron.

4.1. Wüstite and periclase

In one of the diamond crystals from the Juina area, Brazil, an associ-ation: wüstite+periclase was identified. Earlier iron-rich periclase andmagnesiowüstite were known from other areas: in a diamond fromthe Monastery pipe, South Africa, a grain with mg=0.12 was found(Moore et al., 1986), and in a diamond from theGuaniamo area, Venezu-ela, almost pure wüstite with mg=0.002 was found (Kaminsky et al.,2000). However, these were only single grains. In the Juina area, almostpure wüstite (mg=0.02–0.03; unit cell parameter a0=0.4192 nm)closely (‘touching’) associates with high-Mg ferropericlase (Kaminskyet al., 2009b; Fig. 16). An explanation to the association of such phases,which are polar different in composition, can be found in the results ofexperiments by Dubrovinsky et al. (2001) who heated ferropericlasewith mg=0.80 to temperatures in excess of 1000 K at pressures ofover 80 GPa, simulating the stability of the solid solution at physical con-ditions relevant to the Earth's lowermantle. After heating at 85 GPa theyobserved, in the quenched sample, two cubic phases with lattice param-eters of 4.224(4) Å and 4.323(3) Å.

The first corresponds to periclase with mg=0.89; and the secondcorresponds to pure wüstite. The results demonstrated that ferroperi-clase under pressures in excess of 85 GPa and high temperatures maydissociate into Mg-rich and Fe-rich oxide components. These experi-mental results can further be explained in the data of Badro et al.(2003), who showed that at pressures exceeding 70 GPa (or at depthsin excess of 2000 km) the partition coefficient of ferrous iron betweenferropericlase and perovskite increases from approximately zero (ap-proximately equal partitioning of iron between the two compoundswith the initialmg=0.83) to approximately 10–14, i.e., all iron concen-trates into the perovskite forming highly iron-rich wüstite. The concen-tration of iron in ferropericlase continues in association with a post-perovskite phase as well, at pressures of up to 154 GPa (Sakai et al.,2010). In this case, the origin of Mg-poor wüstite may be attributed tothe lowermostmantle, i.e., to the D″ layer at the core–mantle boundary.

4.2. Iron carbides and native iron

In the same diamond from the Juina area, containing the wüstite+periclase inclusion, another interesting association, forming an aggre-gate, was identified: iron carbides+native iron+magnetite+graphite(Kaminsky andWirth, 2011). Native iron, cohenite, andmagnetite havebeen found as inclusions in diamond before; however, in this case notonly cohenite (Fe,Ni,Co)3C, but other carbides, such as chalypite Fe2O(or Fe7C3) and haxonite (Fe,Ni,Co)23C6 were identified. The latter twowere known before only frommeteorites. The Fe/C ratio in carbides var-ies from 1.65 to 3.98, grouping at around 2.33 (chalypite), 3.0 (cohe-nite), and 3.83 (haxonite) (Fig. 17). Most grains are chalypite. Theycontain high concentrations of nitrogen (7.3–9.1 at.%) and are, in fact,

Fig. 14. Electron microscope image of polymineralic inclusion in diamond comprising:nyerereite, phase with the composition of olivine, spinel, apatite, and phlogopite (?).After Kaminsky et al. (2009b).

Table 10Chemical composition of nyerereite (wt.%), after Kaminsky et al. (2009b).

Oxide #1659 #1734

CaO 20.6 21.8Na2O 18.3 23.1K2O 13.8 10.0SrO 0 0.15P2O5 3.4 1.12Total 56.1 56.17



nitrocarbide. In this case, the experimental formula is (Fe,Ni,Cr)7(C0.73–0.81N0.27–0.19)3.

In addition to carbides and carbonitrides, another interestingphase was identified, in the same inclusion, with a much higher nitro-gen concentration. The diffraction patterns of that phase are differentfrom both Fe3C and Fe2C: (001) 0.447 nm; (100) 0.402 nm; and (101)0.299 nm. As such, they could be indexed based on the diffractionpatterns of a synthetic alloy ∈−Fe3(N0.80 C0.20)1.395 which has a Fe/(C+N) ratio of 2.15, i.e., this is similar to the Fe2C composition butwith a C/(C+N) ratio of only 0.20: (001) 0.4406 nm; (100)0.4135 nm; and (101) 0.3015 nm (Leineweber et al., 2001). The C/Nratio in this synthetic alloy can vary because both carbon and nitro-gen are localized in the interstitial positions of the lattice(Leineweber et al., 2001). This compound, having a trigonal crystalstructure, is carbonitride and is of great importance to metallurgy.

Nitrocarbides and carbonitrides have not previously been foundunder natural terrestrial conditions. However, nitrides (e.g., osborniteTiN and BN) are known in chromitites from the Luobasa ophiolites inTibet, where they associate with diamond (Dobrzhinetskaya et al.,2009).

The system Fe–N–C under high pressure conditions is not studied todate. Hence, in order to establish the P–T conditions for the iron carbide(carbonitride, nitrocarbide)+native iron+graphite+diamond associ-ation, one must turn to experimental data on the Fe–C system. In thissystem, cohenite Fe3C, is stable at pressures of less than 5–6 GPa. Athigher pressures Fe7C3, becomes the stable species, and at 50 GPa it re-places cohenite from the liquidus and subsolidus associations (Lordet al., 2009), until cohenite fully disappears at c. 120 GPa, correspondingto depths of c. 3400 km (i.e., within the outer core; Lord et al., 2009)(Fig. 18). Accordingly, one could expect that, under outer-core tolower-mantle conditions, Fe7C3 (exclusively or predominately) shouldcrystallize from an iron–carbon liquid.

It is interesting that, in the Fe–C system, under such extreme P–Tconditions (starting from c. 3700 km depth, in the outer core) dia-mond crystallizes first and can associate, in the subsolidus, withFe7C3 and with cohenite Fe3C, if the carbon concentration is highenough (Fig. 18). These experimental data confirm that an associationof diamond+chalypite+graphite may crystallize from an iron–car-bon liquid (locally enriched in carbon) either in the region of the up-permost stratified layer within the outer core or at the D″ layer, alongwith the association wüstite+periclase.

Fig. 15. Chemical composition of nyerereite. After Kaminsky et al. (2009b).

Fig. 16. Magnesian index of coexisting wüstite and periclase. Fig. 17. Histogram of Fe/C ratios for carbides.



The presence of native iron and graphite, in the studied iron car-bide associations, indicates that the oxygen fugacity for this associa-tion was lower than the iron–wüstite oxygen buffer (c. −9.5 to−10.5 log fO2; Fedorov et al., 2002), and possibly as low as c.−13.95 log fO2 (Solovova et al., 2002) for the association: nativeiron+cohenite in basalt from Disko Island, Greenland. According tothe model of the Earth's core growth by Galimov (2005), such condi-tions prevailed during only the first 100 million years of the Earth'sdevelopment.

5. Composition of the lower mantle

In Table 11 and Fig. 19 we present frequencies of lower-mantleminerals included in diamond in three major regions, Brazil, Canada,and Guinea, which have more or less representative numbers (tensand low hundreds) of lower-mantle mineral inclusions. The propor-tions of minerals vary in different areas, which may be explained ei-ther as regional differences or not full representativity of the samplesets (particularly for Canadian kimberlitic pipes, where only twentygrains of lower-mantle minerals have to date been identified). De-spite these differences, the frequencies of the lower mantle mineralsfor all areas are similar to each other, deviating from the average byonly 10–15%; this value may be considered as an error value forthe total average. It is significant to note that these figures differ

drastically from the composition of the lower mantle as suggestedfrom experimental data (e.g., Fei and Bertka, 1999). In that themost common mineral is ferropericlase, which comprises 48.0–63.3%(average 55.4%) of all minerals in the lower mantle.

This is in contrast to an abundance of c. 18% as suggested by exper-imental data, i.e., approximately three times higher. In contrast to fer-ropericlase, MgSi-perovskite comprises, in all studied regions, only5.0–10.2% (average 7.5%), i.e., approximately ten times lower thanhas been suggested as an average composition in the lower mantle(c. 77%). CaSi-perovskite, according to geological data, is more thantwice as common compared to experimental data (10.0–14.3% withan average of 12.0% against c. 5%). The most important feature ofthe real composition of the lower mantle is a permanent presence,in all regions and areas, of free silica (as stishovite), in the lower man-tle. Stishovite frequency, among the lower mantle minerals, is 2.1–15.0% (average 8.4%). The other minerals (CaTi-perovskite, TAPP, aphase with the composition of that of olivine, spinel, ilmenites, tita-nite, native nickel and iron, magnetite, and sulfides), have frequencies0.1–4.3% each.

6. Discussion and conclusions

Among lower-mantle minerals identified as inclusions in ‘super-deep’ diamond, can be distinguished three mineral associations:

Fig. 18. Phase diagram of the Fe–C system (after Lord et al., 2009).

Table 11Frequency of lower-mantle minerals included in diamond.

Mineral Brazil, Juina area Guinea Canada, Slave province Totalaver.

Rio Sao Luiz Rio Mutumet al.

Rio Soriso Pandrea pipes Aver. Kankan placer DO27 and oth. Panda pipe Aver.

Grains % Grains % Grains % Grains % Grains % Grains % Grains %

fPer 23 60.5 28 45.2 97 55.0 7 33.4 48.0 31 63.3 6 60.0 5 50.0 55.0 55.6MgSiPrv 4 10.5 1 1.6 14 8.0 2 9.5 7.4 5 10.2 1 10.0 5.0 7.5CaSiPrv 5 13.2 4 6.5 32 18.2 2 9.5 11.8 7 14.3 1 10.0 1 10.0 10.0 12.0CaTiPrv 1 1.6 4 2.3 1.0 0.3Sti 1 2.6 3 4.8 2 1.1 2.1 4 8.2 2 20.0 1 10.0 15.0 8.4TAPP 5 13.2 1 1.6 3 1.7 4.1 1.4Ol' 3 4.8 9 5.1 2.5 2 20.0 10.0 4.2Spl 5 8.1 5 23.8 7.9 1 10.0 5.0 4.3Ilm 14 22.6 2 9.5 7.9 2.6Tit 1 1.6 2 1.1 1.4 0.5Ni0 1 1.6 0.4 0.1Fe0 1 0.7 1 4.8 1.4 1 2.0 1.1Mag 9 5.1 1.3 0.4Sulph 3 1.7 2 9.5 2.8 1 2.0 1.6Total 38 100.0 62 100.0 176 100.0 21 100.0 100.0 49 100.0 10 100.0 10 100.0 100.0 100.0Reference Harte

et al., 1999Kaminskyet al., 2001

Haymanet al., 2005

Kaminskyet al., 2009a

n=297 Stachelet al., 2000

Davieset al., 2004

Tappertet al., 2005a

n=20 n=366



juvenile ultramafic, ‘eclogitic’-type, and carbonatitic. The juvenile(ultramafic) obviously predominates among them. The ‘eclogitic’-type association is related only to subducted slabs which likely com-prise an insignificant volume of the deep Earth, and a carbonatiticassociation, most likely, is associated with lower-mantle, carbon-richpartial melts caused by carbonate-induced suppression of the liquidusin the vicinity of thermochemical upwelling (Collerson et al., 2010). Inthis case, the mineral and chemical composition of the lower mantlecan be approximated to the composition of the juvenile, ‘ultramafic’association. However, it is important to point out that the mineral

frequency of the lower mantle minerals cannot readily be used as amodel for the overall mineral composition of the lower mantle.

If we accept the frequencies of minerals from Table 11, the chem-ical composition of the lower mantle is (in wt.%): 42.6 MgO, 23.0%FeO, 21.4% SiO2, 5.9% CaO, 2% TiO2 and Cr2O3, 1.6% Al2O3, 0.7% NiO,and 0.4% K2O. This seems very unlikely. We suggest, therefore, thatthis unrealistic composition for the lower mantle is an artifact of thehigh proportion of ferropericlase among inclusions in diamond. Fer-ropericlase is, to date, the most common mineral identified amongthe other lower-mantle phases. For example, while studying diamond

Fig. 19. Frequency of lower-mantle minerals included in diamond.



crystals from the Juina area, we initially identified only ferropericlaseand ilmenite inclusions; and only after a subsequent special, detailed,instrumental study we were able to find several dozen other lower-mantle inclusions (Kaminsky et al., 2001). The major discrepanciesbetween observed geological data and theoretical/experimental re-sults are as follows:

1. The proportion of ferropericlase in the lowermantle, even consideringits artificial overrating, is higher than it has previously been suggested.

2. The composition of ferropericlase in the lower mantle is more iron-rich than was suggested earlier. The observed magnesian index offerropericlase is within a range ofmg=0.36–0.90, while the theoret-ical and experimental magnesian index was estimated at 0.73–0.88.

3. Free silica (stishovite) is always present in lower-mantle associa-tions, where it usually associates with ferropericlase (sometimeswith native iron). Silica inclusions were identified in all sets oflower-mantle minerals observed in diamond from all regions andareas: 2.6% from Rio São Luiz, 1.1% from Rio Soriso; 4.8% fromother placer deposits in the Juina area; 8.2% from the Kankanarea, Guinea, and 10–20% from Canadian kimberlites (Harte et al.,1999; Kaminsky et al., 2001; Hayman et al., 2005; Stachel et al.,2000; Davies et al., 2004; Tappert et al., 2005a). It is not accountedfor in experimental models of lower mantle compositions.

4. Separate aluminous phase (TAPP) was identified in several areas,in association with ferropericlase and MgSi-perovskite.These discrepancies suggest that the composition of the lowermantle differs significantly to that of the upper-mantle, and exper-iments, based, solely, on ‘pyrolitic’ compositions are not generallyapplicable for the lower mantle.

Based on the presented materials, two fundamental conclusionscan be made.

1. The Earth's mantle has different compositions in its upper andlower parts and may be considered as heterogeneous. This idea isconsistent to that proposed earlier on the basis of both geochemi-cal and isotopic studies, and seismic data (e.g., Anderson, 2002;Cammarano and Romanowicz, 2007).

2. These differences and the presence of free silica in the lower man-tle indicate the likelihood of an alternative model, to that of CI-chondrite to account for the Earth's current structure; for example,an enstatite-chondrite model (e.g., Javoy et al., 2010).

These suggestions should, therefore, be considered in future geo-physical interpretations and in the design of all models for the evolu-tion of the Earth. They require confirmation, however, that can beaddressed through further data collection and compilation of the iso-tope characteristics of the lower-mantle material.

Acknowledgements

Manypeople have contributed to the subject of this paper, and Iwouldparticularly like to thank: P. Andreazza, E. Belousova,W.Griffin, S. O'Reilly,R. Wirth, and O. Zakharchenko for past collaborations. G. Bulanova, B.Harte, and T. Stachel helped considerably by providing spreadsheets foranalytical data. L. Kogarko and Yu. Litvin made useful comments to thetext. This paper is based upon the presentation of the 2010 VernadskyReading, and I thank the Vernadsky Institute of Geochemistry andAnalyt-ical Chemistry, Russian Academy of Sciences, and particularly Professor E.Galimov, for the encouragement to publish the substance of this paper.

Appendix A

AbbreviationsGPa gigapascal; 1 GPa=10 kilobarsAnh anhydrite CaSO4

Ap apatite Ca5(PO4)3(F,Cl)

Cal calcite CaCO3

CaSiPrv CaSi-perovskite CaSiO3

CaTiPrv perovskite CaTiO3

Chal chalypite Fe2C (Fe7C3)Coh cohenite Fe3CCot cotunnite PbCl2CrSpl chrome spinel (chromian spinel) (Mg,Fe)(Cr,Al,Fe)2O4

Cusp cuspidine Ca4Si2O7F2Dol dolomite CaMg(CO3)2Egg phase “Egg” AlSiO3OHf=Fe/(Fe+Mg)mol iron indexFe0 native ironFe2+#=100 Fe2+/(Fe2++Mg)mol ferrous iron indexfPer ferropericlase (Mg,Fe)OGrt garnet (Mg,Fe,Mn)3Al2(SiO4)3Hax haxonite (Fe,Ni,Co)23C6Hl halite NaClHREE heavy rare earth elements (Er to Lu)Hy hydophilite CaCl2LREE light rare earth elements (La to Nd)mg=Mg/(Mg+Fe)mol magnesian indexMgSiPrv MgSi-perovskite (Mg,Fe)SiO3

Mn-Ilm manganoilmenite (Fe,Mn)TiO3

Mag magnetite Fe3O4

Maj majorite Mg3(Mg,Fe,Al,Si)2Si3O12

Mois moissanite SiCMtc monticellite CaMgSiO4

Nah nahcolite NaHCO3

Ni0 native nickelNyer nyerereite Na2Ca(CO3)2Ol phase of olivine composition (Mg,Fe)2SiO4

Per periclase MgOPhl phlogopite KMg3(Si3AlO10)(F,OH)2Pilm picroilmenite (magnesian ilmenite) (Fe,Mg)TiO3

REE rare earth elementsSd siderite FeCO3

Spl spinel MgAl2O4

Syl sylvite KClSti stishovite SiO2

Sulph sulfideTAPP tetragonal phase of almandine-pyrope composition

(Mg,Fe3+)(Al,Cr,Mn)2(Mg,Fe2+)2Si3O12

Ttn titanite CaTiSiO5

Wo wollastonite CaSiO3

Wu wüstite FeOXSi cation proportion of silica in garnet based on 12 oxygens

References

Akaogi, M., 2007. Phase transitions of minerals in the transition zone and upper partof the lower mantle. In: Ohtani, E. (Ed.), Advances in High-pressure Mineralogy.Geological Society of America Special Paper 421, pp. 1–13.

Anderson, D.L., 2002. The case for irreversible chemical stratification of the mantle. In-ternational Geology Review 44 (2), 97–116.

Badro, J., Fiquet, G., Guyot, F., Rueff, J.-P., Struzhkin, V.V., Vanko, G., Monaco, G., 2003.Iron partitioning in Earth's mantle; toward a deep lower mantle discontinuity.Science 300 (5620), 789–791.

Bindi, L., Dymshits, A.M., Bobrov, A.V., Litasov, K.D., Shatsky, A.F., Ohtani, E., Litvin,Yu.A., 2011. Crystal chemistry of sodium in the Earth's interior: the structure ofNa2MgSi5O12 synthesized at 17.5 GPa and 1700 °C. American Mineralogist 96(2–3), 447–450.

Bobrov, A.V., Litvin, Yu.A., Bindi, L., Dymshits, A.M., 2008a. Phase relations and forma-tion of sodium-rich majorite garnet in the system Mg3Al2Si3O12–Na2MgSi5O12 at7.0 and 8.5 GPa. Contributions to Mineralogy and Petrology 156 (2), 243–257.

Bobrov, A.V., Kojitani, H., Akaogi, M., Litvin, Yu.A., 2008b. Phase relations on thediopside-hedenbergite-jadeite join up to 24 GPa and stability of Na-bearing major-ite garnet. Geochimica et Cosmochimica Acta 72, 2392–2408.

Brenker, F.E., Stachel, T., Harris, J.W., 2002. Exhumation of lower mantle inclusions indiamond: ATEM investigation of retrograde phase transitions, reactions and exso-lution. Earth and Planetary Science Letters 198 (1–2), 1–9.



Brenker, F.E., Vollmer, C., Vincze, L., Vekemans, B., Szymanski, A., Janssens, K., Szaloki, I.,Nasdala, L., Kaminsky, F., 2007. Carbonates from the lower part of transition zoneor even the lower mantle. Earth and Planetary Science Letters 260 (1–2), 1–9.

Bulanova, G.P., Smith, C.B., Kohn, S.C., Walter, M.J., Gobbo, L., Kearns, S., 2008. MachadoRiver, Brazil—a newly recognised ultradeep diamond occurrence. 9th InternationalKimberlite Conference Extended Abstract No. 9IKC-A-00233.

Bulanova, G.P., Walter, M.J., Smith, C.B., Kohn, S.C., Armstrong, L.S., Blundy, J., Gobbo, L.,2010. Mineral inclusions in sublithospheric diamonds from Collier 4 kimberlitepipe, Juina, Brazil: subducted protoliths, carbonated melts and primary kimberlitemagmatism. Contributions to Mineralogy and Petrology 159 (4), 489–510.

Cammarano, F., Romanowicz, B., 2007. Insights into the nature of the transition zonefrom physically constrained inversion of long-period seismic data. PNAS 104(22), 9139–9144.

Chen, F., Guo, J., Wang, S., Liu, C., Chen, J., Wang, Z., Lei, P., 1992. High-K and high-Cl in-clusions in diamond and mantle metasomatism. Acta Mineralogica Sinica 12 (3),193–198 (in Chinese, with English Abstr.).

Chinn, J.L., Milledge, H.J., Gurney, J.J., 1998. Diamonds and inclusions from the Jagersfon-tein kimberlite. Seventh International Kimberlite Conference Extended Abstracts,Cape Town, pp. 156–157.

Chudinovskikh, L., Boehler, R., 2001. High-pressure polymorphs of olivine and the 660-km seismic discontinuity. Nature 411 (6837), 574–577.

Collerson, K.D., Hapugoda, S., Kamber, B.S.,Williams, Q., 2000. Rocks from themantle tran-sition zone: majorite-bearing xenoliths from Malaita, Southwest Pacific. Science 288(5469), 1215–1223.

Collerson, K.D., Williams, Q., Ewart, A.E., Murphy, D.T., 2010. Origin of HIMU and EM-1domains sampled by ocean island basalts, kimberlites and carbonatites: the role ofCO2-fluxed lower mantle melting in thermochemical upwellings. Physics of theEarth and Planetary Interiors 181 (1), 112–131.

Davies, R.M., Griffin, W.L., O'Reilly, S.Y., Doyle, B.J., 2004. Mineral inclusions and geo-chemical characteristics of microdiamonds from the DO27, A154, A21, A418,DO18, DD17 and Ranch Lake kimberlites at Lac de Gras, Slave Craton, Canada. Lithos77 (1–4), 39–55.

Dawson, J.B., 1962. Sodium carbonate lavas from Oldoinyo Lengai, Tanganyika. Nature195 (4846), 1075–1076.

Dobrzhinetskaya, L., Green II, H.W., Wang, S., 1996. Alpe Arami: a peridotite massiffrom depths of more than 300 kilometers. Science 271 (5257), 1841–1845.

Dobrzhinetskaya, L.F., Wirth, R., Yang, J., Hutcheon, I.D., Weber, P.K., Green II, H.W.,2009. High-pressure highly reduced nitrides and oxides from chromitite of aTibetan ophiolite. Proceedings of the National Academy of Sciences 106 (46),19233–19238.

Donnelly, C.L., Stachel, T., Creighton, S., Muehlenbachs, K., Whiteford, S., 2007.Diamonds and their mineral inclusions from the A154 South pipe, Diavik DiamondMine, Northwest territories, Canada. Lithos 98 (1–4), 160–176.

Dubrovinsky, L.S., Dubrovinskaia, N.A., Annersten, H., Halenius, E., Harryson, H., 2001.Stability of (Mg0.5Fe0.5)O and (Mg0.8 Fe0.2)O magnesiowüstites in the lower man-tle. European Journal of Mineralogy 13 (5), 857–861.

Dymshits, A.M., Bobrov, A.V., Litasov, K.D., Shatsky, A.F., Ohtani, E., Litvin, Yu.A., 2010.Experimental study of the pyroxene-garnet phase transition in the Na2MgSi5O12

system at pressures of 13–20 GPa: first syntheses of sodium majorite. DokladyEarth Sciences 434 (1), 1263–1266.

Egglton, R.A., Boland, J.N., Ringwood, A.E., 1978. High pressure synthesis of a new alu-minum silicate: Al5Si5O17(OH). Geochemical Journal 12, 191–194.

Fedorov, I.I., Chepurov, A.A., Dereppe, J.M., 2002. Redox conditions of metal-carbonmelts and natural diamond genesis. Geochemical Journal 36, 247–253.

Fei, Y., Wang, Y., Finger, L.W., 1996a. Maximum solubility of FeO in (Mg,Fe)SiO3 perov-skite as a function of temperature at 26 GPa: Implication for FeO content in thelower mantle. Journal of Geophysical Research 101 (B5), 11525–11530.

Fei, Y., Bertka, C.M., 1999. Phase transitions in the Earth's mantle and mantle mineral-ogy. In: Fei, Y., Bertka, C.M., Mysen, B.O. (Eds.), Geochemical Society Special Publi-cation No. 6, pp. 189–207.

Fei, Y., Wang, Y., Finger, L.W., 1996b. Maximum solubility of FeO in (Mg,Fe)SiO3 perov-skite as a function of temperature at 26 GPa: implication for FeO content in thelower mantle. Journal of Geophysical Research 101 (B5), 11525–11530.

Finger, L.W., Conrad, P.G., 2000. The crystal structure of “Tetragonal Almandine-PyropePhase” (TAPP): a reexamination. American Mineralogist 85 (11–12), 1804–1807.

Frost, D.J., Langenhorst, F., 2002. The effect of Al2O3 on Fe–Mg partitioning betweenmagnesiowüstite and magnesium silicate perovskite. Earth and Planetary ScienceLetters 199 (3–4), 227–241.

Gasparik, T., Wolf, K., Smith, C.M., 1994. Experimental determination of phase relations inthe CaSiO3 system from 8 to 15 GPa. American Mineralogist 79 (11–12), 1219–1222.

Galimov, E.M., 2005. Redox evolution of the Earth caused by a multi-stage formation ofits core. Earth and Planetary Science Letters 233 (3–4), 263–276.

Harte, B., 2010. Diamond formation in the deep mantle: the record of mineral inclu-sions and their distribution in relation to mantle dehydration zones. MineralogicalMagazine 74 (2), 189–215.

Harte, B., Harris, J.W., 1994. Lower mantle mineral association preserved in diamonds.Mineralogical Magazine 58A, 384–385.

Harte, B., Harris, J.W., Hutchison, M.T., Watt, G.R., Wilding, M.C., 1999. Lowermantle mineralassociations in diamonds fromSao Luiz, Brazil. In: Fei, Y., Bertka, C.M.,Mysen, B.O. (Eds.),Mantle Petrology: Field Observations and High Pressure Experimentation: A Tribute toFrancis R. (Joe) Boyd: Geochemical Society Special Publication No. 6, pp. 125–153.

Harris, J.W., Hutchison, M.T., Hursthouse, M., Light, M., Harte, B., 1997. A new tetrago-nal silicate mineral occurring as inclusions in lower mantle diamonds. Nature 387(6632), 486–488.

Hayman, P.C., Kopylova, M.G., Kaminsky, F.V., 2005. Lower mantle diamonds from RioSoriso (Juina, Brazil). Contributions to Mineralogy and Petrology 149 (4), 430–445.

Helffrich, G.R., Wood, B.J., 2001. The Earth's mantle. Nature 412 (6846), 501–507.Hirose, K., Fei, Y., Ma, Y., Mao, H.-K., 1999. The fate of subducted basaltic crust in the

Earth's lower mantle. Nature 397 (6714), 53–56.Hutchison, M.T., Hurtshouse, M.B., Light, M.E., 2001. Mineral inclusions in diamonds:

associations and chemical distinctions around the 670-km discontinuity. Contribu-tions to Mineralogy and Petrology 142 (2), 119–126.

Izraeli, E.S., Harris, J.W., Navon, O., 2001. Brine inclusions in diamonds: a new uppermantle fluid. Earth and Planetary Science Letters 187 (3–4), 323–332.

Irifune, T., 1994. Absence of an aluminous phase in the upper part of the Earth's lowermantle. Nature 370 (6485), 131–133.

Irifune, T., Ringwood, A.E., 1993. Phase transformations in subducted oceanic crust andbuoyancy relationships at depths of 600–800 km in the mantle. Earth and Plane-tary Science Letters 117 (1–2), 101–110.

Irifune, T., Koizumi, T., Ando, J.I., 1996. An experimental study of the garnet–perovskitetransformation in the system MgSiO3–Mg3Al2Si3O12. Physics of the Earth and PlanetInterior 96 (3–4), 147–157.

Javoy, M., Kaminski, E., Guyot, F., Andrault, D., Sanloup, C., Moreira, M., Labrosse, S., Jambon,A., Agrinier, P., Davaille, A., Jaupart, C., 2010. The chemical composition of the Earth:enstatite chondrite models. Earth and Planetary Science Letters 293 (3–4), 259–268.

Joswig, W., Stachel, T., Harris, J.W., Baur, W.H., Brey, G.P., 1999. New Ca-silicate inclu-sions in diamonds—tracers from the lower mantle. Earth and Planetary ScienceLetters 173 (1–2), 1–6.

Kamenetsky, V.S., Sharygin, V.V., Kamenetsky, M.B., Golovin, A.V., 2006. Chloride–carbonate nodules in kimberlites from the Udachnaya pipe: alternative approach tothe evolution of kimberlite magmas. Geochemistry International 44 (9), 935–940.

Kaminsky, F.V., Belousova, E.A., 2009. Manganoan ilmenite as kimberlite/diamond indi-cator mineral. Russian Geology and Geophysics 50 (12), 1212–1220.

Kaminsky, F.V., Wirth, R., 2011. Iron carbide inclusions in lower-mantle diamond fromJuina, Brazil. The Canadian Mineralogist 49 (2), 555–572.

Kaminsky, F.V., Zakharchenko, O.D., Griffin, W.L., Channer, D.M. DeR, Khachatryan-Blinova, G.K., 2000. Diamond from the Guaniamo area, Venezuela. The CanadianMineralogist 38 (6), 1347–1370.

Kaminsky, F.V., Zakharchenko, O.D., Davies, R., Griffin, W.L., Khachatryan-Blinova, G.K.,Shiryaev, A.A., 2001. Superdeep diamonds from the Juina area, Mato Grosso State,Brazil. Contributions to Mineralogy and Petrology 140 (6), 734–753.

Kaminsky, F.V., Khachatryan, G.K., Andreazza, P., Araujo, D., Griffin, W.L., 2009a. Super-deep diamonds from kimberlites in the Juina area, Mato Grosso State, Brazil. Lithos112S (2), 833–842.

Kaminsky, F., Wirth, R., Matsyuk, S., Schreiber, A., Thomas, R., 2009b. Nyerereite andnahcolite inclusions in diamond: evidence for lower-mantle carbonatitic magmas.Mineralogical Magazine 73 (5), 797–816.

Kaminsky, F.V., Sablukov, S.M., Belousova, E.A., Andreazza, P., Tremblay, M., Griffin,W.L., 2010. Kimberlitic sources of super-deep diamonds in the Juina area, MatoGrosso State, Brazil. Lithos 114 (1–2), 16–29.

Kennedy, C.S., Kennedy, G.C., 1976. The equilibrium boundary between graphite anddiamond. Journal of Geophysical Research 81 (B14), 2467–2470.

Kerschhofer, L., Scharer, U., Deutsch, A., 2000. Evidence for crystals from the lowermantle: baddeleyite megacrysts of the Mbuji Mayi kimberlite. Earth and PlanetaryScience Letters 179 (2), 219–225.

Kesson, S.E., Fitz Gerald, J.D., 1991. Partitioning of MgO, FeO, NiO, MnO and Cr2O3 be-tween magnesian silicate perovskite and magnesiowustite: implications for the or-igin of inclusions in diamond and the composition of the lower mantle. Earth andPlanetary Science Letters 111 (2–4), 229–240.

Kesson, S.E., Fitz Gerald, J.D., Shelley, J.M.G., Withers, R.L., 1995. Phase relations, struc-ture and crystal chemistry of some aluminous silicate perovskites. Earth and Plan-etary Science Letters 134 (1–2), 187–201.

Klein-BenDavid, O., Wirth, R., Navon, O., 2006. TEM imaging and analysis of microinclu-sions in diamonds: a close look at diamond-growing fluids. American Mineralogist91 (2–3), 353–365.

Kogarko, L.N., Plant, D.A., Henderson, C.M.B., Kjarsgaard, B.A., 1991. Na-rich carbonateinclusion in perovskite and calzirtite from the Guli Intrusive Ca-carbonatite, PolarSiberia. Contributions to Mineralogy and Petrology 109 (1), 124–129.

Kopylova, M.G., Gurney, J.J., Daniels, L.D., 1997. Mineral inclusions in diamonds fromthe River Ranch kimberlite, Zimbabwe. Contributions to Mineralogy and Petrology129 (4), 366–384.

Kubo, A., Suzuki, T., Akaogi, M., 1997. High pressure phase equilibria in the systemCaTiO3–CaSiO3: stability of perovskite solid solutions. Physics and Chemistry ofMinerals 24, 488–494.

Lee, K.K.M., O'Neill, B., Panero, W.R., Shim, S.H., Benedetti, L.R., Jeanloz, R., 2004. Equationsof state of the high-pressure phases of a natural peridotite and implications for theEarth's lower mantle. Earth and Planetary Science Letters 223 (3–4), 381–393.

Leineweber, A., Jacobs, H., Hüning, F., Lueken, H., Kockelmann, W., 2001. Nitrogen or-dering and ferromagnetic properties of ∈−Fe3N1+ x (0.10≤x≤0.39) and ∈−Fe3(N0.80C0.20)1.38. Journal of Alloys and Compounds 316, 21–38.

Litasov, K., Ohtani, E., Langenhorst, F., Yurimoto, H., Kubo, T., Kondo, T., 2003. Watersolubility in Mg-perovskite and water storage capacity in the lower mantle. Earthand Planetary Science Letters 211 (1–2), 189–203.

Liu, L.-g., 2002. An alternative interpretation of lower mantle mineral associations indiamonds. Contributions to Mineralogy and Petrology 144 (1), 16–21.

Logvinova, A.M., Wirth, R., Fedorova, E.N., Sobolev, N.V., 2008. Nanometre-sized miner-al and fluid inclusions in cloudy Siberian diamonds: new insights on diamond for-mation. European Journal of Mineralogy 20 (3), 317–331.

Lord, O.T., Walter, M.J., Dasgupta, R., Walker, D., Clark, S.M., 2009. Melting in the Fe–Csystem to 70 GPa. Earth and Planetary Science Letters 284 (1–2), 157–167.

McCammon, C., 1997. Perovskite as a possible sink for ferric iron in the lower mantle.Nature 387 (6634), 694–696.



McCammon, C., Hutchison, M.T., Harris, J.W., 1997. Ferric iron content of mineral inclu-sions in diamonds from São Luiz: a view into the lower mantle. Science 278 (5337),434–436.

McCammon, C.A., Stachel, T., Harris, J.W., 2004. Iron oxidation state in lower mantlemineral assemblages. II. Inclusions in diamonds from Kankan, Guinea. Earth andPlanetary Science Letters 222 (2), 423–434.

McDade, P., Harris, J.W., 1999. Syngenetic inclusion bearing diamonds fromLetseng-la-Terai,Lesotho. Proceedings of the VIIth International Kimberlite Conference, vol. 2. Red RoofDesign, Cape Town, pp. 557–565.

McDonough, W.F., Sun, S.-s., 1995. The composition of the Earth. Chemical Geology 120(3–4), 223–253.

Moore, R.O., Gumey, J.J., 1985. Pyroxene solid solution in garnets included in diamonds.Nature 318 (6046), 553–555.

Moore, R.O., Otter, M.L., Rickard, R.S., Harris, J.W., Gurney, J.J., 1986. The occurrence ofmoissanite and ferro-periclase as inclusions in diamond. 4th International Kimber-lite Conference Extended Abstracts, Perth: Geological Society of Australia Ab-stracts, vol. 16, pp. 409–411.

Navon, O., Izraeli, E.S., Klein-BenDavid, O., 2003. Fluid inclusions in diamonds—the car-bonatitic connection. 8th International Kimberlite Conference, Long AbstractFLA_0107.

Neal, C.R., Haggerty, S.E., Sautter, V., 2001. “Majorite” and “silicate perovskite” mineralcompositions in xenoliths from Malaita. Science 292 (5519), 1015a.

Newton, M.G., Melton, C.E., Giardini, A.A., 1977. Mineral inclusions in the Arkansas di-amond. American Mineralogist 62 (5–6), 583–586.

Ono, S., 1999. High temperature stability of phase “Egg”, AlSiO3(OH). Contributions toMineralogy and Petrology 137 (1–2), 83–89.

Ono, S., Hirose, K., Murakami, M., Isshiki, M., 2002. Post-stishovite phase boundary inSiO2 determined by in-situ X-ray observations. Earth and Planetary Science Letters197 (3–4), 192–197.

Otter, M.L., Gurney, J.J., 1989. Mineral inclusions in diamond from the Sloan diatremes,Colorado–Wyoming State Line kimberlite district, North America. In: Ross, J., et al.(Ed.), Kimberlites and related rocks. : Proceedings of the Fourth International Kim-berlite Conference, Perth 1986, vol. 2. Blackwell, Carlton, pp. 1042–1053.

Ringwood, A.E., Irifune, T., 1988. Nature of the 650-km seismic discontinuity: implica-tions for mantle dynamics and differentiation. Nature 331 (6152), 131–136.

Safonov, O.G., Perchuk, L.L., Litvin, Y.A., 2007. Melting relations in the chloride–carbon-ate–silicate systems at high-pressure and the model for formation of alkalic dia-mond-forming liquids in the upper mantle. Earth and Planetary Science Letters253 (1–2), 112–128.

Sakai, T., Ohtani, E., Terasaki, H., Miyahara, M., Nishijima, M., Hirao, N., Ohishi, Y., Sata,N., 2010. Fe–Mg partitioning between post-perovskite and ferropericlase in thelowermost mantle. Physics and Chemistry of Minerals 37, 487–496.

Sano, A., Ohtani, E., Kubo, T., Finakoshi, K., 2004. In situ X-ray observation of decompositionof hydrous aluminum silicate AlSiO3OH and aluminum oxide hydroxide d-AlOOH athigh pressure and temperature. Journal of Physics and Chemistry of Solids 65,1547–1554.

Saxena, S.K., 2010. Thermodynamic modeling of the Earth's interior. Elements 6 (5),321–325.

Schärer, U., Berndt, J., Deutsch, A., 2011. The genesis of deep-mantle xenocrystic zirconand baddeleyite megacrysts (Mbuji-Mayi kimberlite): trace-element patterns.European Journal of Mineralogy 23, 241–255.

Schmidt, M.W., Finger, L.W., Ross, R.J., Dinnebier, R.E., 1998. Synthesis, crystal struc-ture, and phase relations of AlSiO3OH, a high-pressure hydrous phase. AmericanMineralogist 83 (7–8), 881–888.

Scott Smith, B.H., Danchin, R.V., Harris, J.W., Stracke, K.J., 1984. Kimberlites nearOrroroo, South Australia. In: Kornprobst, J. (Ed.), Kimberlites I: Kimberlites and Re-lated Rocks. Elsevier, Amsterdam, pp. 121–142.

Shim, S.H., Duffy, T., Shen, G., 2000. The stability and P–V–T equation of state of CaSiO3

perovskite in the Earth's lower mantle. Journal of Geophysical Research 105 (B11),25955–25968.

Sobolev, N.V., Kaminsky, F.V., Griffin, W.L., Yefimova, E.S., Win, T.T., Ryan, C.G., Botkunov,A.I., 1997. Mineral inclusions in diamonds from the Sputnik kimberlite pipe, Yakutia.Lithos 39 (3–4), 135–157.

Sobolev, N.V., Yefimova, E.S., Channer, D.M.D., Anderson, P.F.N., Barron, K.M., 1998.Unusual upper mantle beneath Guaniamo, Guyana shield, Venezuela: evidencefrom diamond inclusions. Geology 26 (11), 971–974.

Sobolev, N.V., Yefimova, E.S., Koptil, V.I., 1999. Mineral inclusions in diamonds in theNortheast of the Yakutian diamondiferous province. Proceedings of the VIIth Inter-national Kimberlite Conference, vol. 2. Red Roof Design, Cape Town, pp. 816–822.

Solovova, I.P., Ryabchikov, I.D., Girnis, A.V., Pedersen, A., Hansteen, T., 2002. Reducedmagmatic fluids in basalt from the island of Disko, central West Greenland. Chem-ical Geology 183, 365–371.

Stachel, T., Harris, J.W., Brey, G.P., 1999. REE patterns of peridotitic and eclogitic inclu-sions in diamonds from Mwadui (Tanzania). Proceedings of the VIIth InternationalKimberlite Conference, vol. 2. Red Roof Design, Cape Town, pp. 829–835.

Stachel, T., Harris, J.W., Brey, G.P., Joswig, W., 2000. Kankan diamonds (Guinea) II:lower mantle inclusion parageneses. Contributions to Mineralogy and Petrology140 (1), 16–27.

Stachel, T., Harris, J.W., Aulbach, S., Deines, P., 2002. Kankan diamonds (Guinea) III:δ13C and nitrogen characteristics of deep diamonds. Contributions to Mineralogyand Petrology 142 (4), 465–475.

Stixrude, L., Lithgow-Bertelloni, C., 2007. Influence of phase transformations on lateralheterogeneity and dynamics in Earth's mantle. Earth and Planetary Science Letters263 (1–2), 45–55.

Tappert, R., Stachel, T., Harris, J.W., Shimizu, N., Brey, G.P., 2005a. Mineral inclusions indiamonds from the Slave Province, Canada. European Journal of Mineralogy 17 (3),423–440.

Tappert, R., Stachel, T., Harris, J.W., Muehlenbachs, K., Ludwig, T., Brey, G., 2005b. Dia-monds from Jagersfontein (South Africa): messengers from the sublithosphericmantle. Contributions to Mineralogy and Petrology 150 (5), 505–522.

Tappert, R., Foden, J., Stachel, T., Muehlenbachs, K., Tappert, M., Wills, K., 2009. The di-amonds of South Australia. Lithos 112S, 806–821.

Vanpeteghem, C., Ohtani, E., Kondo, T., Takemura, K., Kikegawa, T., 2003. Compressibil-ity of phase Egg AlSiO3OH: equation of state and role of water at high pressure.American Mineralogist 88 (10), 1408–1411.

Van Rythoven, A.D., Schulze, D.J., 2009. In-situ analysis of diamonds and their inclu-sions from the Diavik Mine, Northwest Territories, Canada: mapping diamondgrowth. Lithos 112S, 870–879.

Wang, W., Gasparik, T., Rapp, R., 2000. Partitioning of rare earth elements betweenCaSiO3 perovskite and coexisting phases: constraints on the formation of CaSiO3

inclusions in diamond. Earth and Planetary Science Letters 181 (3), 291–300.Wilding, M.C., 1990. Unpublished Ph.D. Thesis. University of Edinburgh, UK, 281 pp.Wilding, M.C., Harte, B., Harris, J.W., 1991. Evidence for a deep origin for the Sao Luiz

diamonds. Fifth International Kimberlite Conference Extended Abstracts, Araxa,pp. 456–458.

Wirth, R., Vollmer, C., Brenker, F., Matsyuk, S., Kaminsky, F., 2007. Nanocrystalline hy-drous aluminium silicate in superdeep diamonds from Juina (Mato Grosso State,Brazil). Earth and Planetary Science Letters 259 (3–4), 384–399.

Wirth, R., Kaminsky, F., Matsyuk, S., Schreiber, A., 2009. Unusual micro- and nano-inclusions in diamonds from the Juina Area, Brazil. Earth and Planetary Science Let-ters 286 (1–2), 292–303.

Withers, A.C., Essene, E.J., Jhang, Y., 2003. Rutile/TiO2II phase equilibria. Contributionsto Mineralogy and Petrology 145 (2), 199–204.

Wood, B.J., 2000. Phase transformations and partitioning relations in peridotite underlower mantle conditions. Earth and Planetary Science Letters 174 (3–4), 341–354.

Yu, Y.G., Wentzcovitch, R.M., Vinograd, V.L., Angel, R.J., 2011. Thermodynamic proper-ties of MgSiO3 majorite and phase transitions near 660 km depth in MgSiO3 andMg2SiO4: a first principles study. Journal of Geophysical Research 116, B02208.doi:10.1029/2010JB007912.

Zedgenizov, D.A., Logviniva, A.M., Shatsky, V.S., Sobolev, N.V., 1998. Inclusions inmicrodiamonds from some Yakutian kimberlite diatremes. Dokladi AkademiiNauk 359 (2), 204–208.

Zedgenizov, D.A., Yefimova, E.S., Logvinova, A.M., Shatsky, V.S., Sobolev, N.V., 2001.Ferropericlase inclusions in a diamond microcrystal from the Udachnaya kimber-lite pipe, Yakutia. Doklady Akademii Nauk 377 (3), 319–321.

Zhang, J., Li, B., Utsumi, W., Liebermann, R.C., 1996. In situ X-ray observations of thecoesite-stishovite transition: reversed phase boundary and kinetics. Physics andChemistry of Minerals 23, 1–10.


Author's personal copy Mineralogy of the lower mantle: A review of 'super-deep' mineral inclusions...

Documents

Transcript of Author's personal copy Mineralogy of the lower mantle: A review of 'super-deep' mineral inclusions...