Full text of DOI:10.1007/s004100000221

Contrib Mineral Petrol 02001) 140: 734±753
DOI 10.1007/s004100000221
F. V. Kaminsky á O. D. Zakharchenko á R. Davies
W. L. Grin á G. K. Khachatryan-Blinova
A. A. Shiryaev
Superdeep diamonds from the Juina area, Mato Grosso State, Brazil
Received: 29 May 2000 / Accepted: 30 October 2000 / Published online: 9 February 2001
Ó Springer-Verlag 2001
Abstract Alluvial diamonds from the Juina area in Mato and high in Ti 0Cr2O3<36.5 wt%, and TiO2>10 wt%).
Grosso, Brazil, have been characterized in terms of their Most ilmenite inclusions have low MgO contents, and
morphology, syngenetic mineral inclusions, carbon iso- some have very high 0up to 11.5 wt%) MnO contents.
topes and nitrogen contents. Morphologically, they are The rare ``olivine'' inclusions coexistingwith ferroperi-
similar to other Brazilian diamonds, showinga strong
clase have low Mg# 087±89), and higher Ca, Cr and Zn
predominance of rounded dodecahedral crystals. How- contents than typical diamond-inclusion olivines. They
ever, other characteristics of the Juina diamonds make are interpreted as inverted from spinel-structured 0Mg,
them unique. The inclusion parageneses of Juina Fe)2Si2O4. This suite of inclusions is consistent with
diamonds are dominated by ultra-high-pressure derivation of most of the diamonds from depths near
0``superdeep'') phases that di€er both from ``traditional'' 670 km, and adds ilmenite and relatively low-Cr, high-Ti
syngenetic minerals associated with diamonds and, in spinel to the known phases of the superdeep paragenesis.
detail, from most other superdeep assemblages. Ferro- Diamonds from the Juina area are characterized by a
periclase is the dominant inclusion in the Juina narrow range of carbon isotopic composition 0d13C=
diamonds. It coexists with ilmenite, Cr-Ti spinel, a phase )7.8 to ±2.5&), except for the one majorite-bearing
with the major-element composition of olivine, and SiO2. diamond 0d13C=±11.4&). There are high proportions of
CaSi-perovskite inclusions coexist with titanite 0sphene), nitrogen-free and low-nitrogen diamonds, and the
``olivine'' and native Ni. MgSi-perovskite coexists with aggregated B center is predominant in nitrogen-con-
TAPP 0tetragonal almandine-pyrope phase). Majoritic tainingdiamonds. These observations have practical
garnet occurs in one diamond, associated with CaTi- consequences for diamond exploration: Low-Mgolivine,
perovskite, Mn-ilmenite and an unidenti®ed Si-Mg low-Mgand high-Mn ilmenite, and low-Cr spinel should
phase. Neither Cr-pyrope nor Mg-chromite was found as be included in the list of diamond indicator minerals, and
inclusions. The spinel inclusions are low in Cr and Mg, the role of high-Cr, low-Ti spinel as the only spinel
associated with diamond, and hence as a criterion of
diamond grade in kimberlites, should be reconsidered.
F. V. Kaminsky 0&)
1KM Diamond Exploration, 815 Evelyn Drive,
West Vancouver V7 T 1J1, B.C., Canada
E-mail: felixvkaminsky@cs.com
Introduction
O. D. Zakharchenko á G. K. Khachatryan-Blinova
Institute of Diamonds Ltd., Ste. 155±5/10 Litovskii Blvd.,
Moscow 117593, Russia
The Juina area is a major diamond miningarea in Brazil
0Fig. 1). In the late 1980s±early 1990s, as many as 50,000
artisanal miners 0garimpeiros) worked this area, pro-
ducingup to 5±6 million carats of diamonds each year
accordingto local sources of information. Many of the
diamonds recovered duringthis period were larger than
100±200 carats, with the largest reported stone weighing
up to 480 ct 0Gaspar et al. 1998). There are at least 26
kimberlite pipes in the area. Some of them are sub-
economic with up to 0.5 ct t±1 diamond grade. However,
all diamond production has been from recent alluvial
and colluvial deposits.
F. V. Kaminsky á R. Davies á W. L. Grin
GEMOC Key Centre, Department of Earth and Planetary
Sciences, Macquarie University, Sydney 2109, Australia
W. L. Grin
CSIRO Exploration and Mining, P.O. Box 128,
North Ryde, NSW 1609, Australia
A. A. Shiryaev
Vernadsky Institute of Geochemistry and Analytical Chemistry,
Kosygin St. 19, Moscow 117975, Russia
Editorial responsibility: W. Schreyer

735
Fig. 1 Location and sketch map of the Juina area. Black circles
Diamond sample sites; black diamonds ®nds of large diamonds
0in carats); shaded area Cretaceous-Tertiary ``Chapada'' deposits
have documented the presence, in diamonds from the
Sao Luiz drainage, of a mineral inclusion paragenesis
that suggests derivation from the lower mantle at depths
of at least 660 km.
Diamonds from this area have attracted considerable We have studied a set of 475 diamonds, from 0.01±
attention because several studies by J.W. Harris, B. 5.3 ct in weight, from several drainages within the Juina
Harte and coworkers 0summarized by Harte et al. 1999) area 0Fig. 1). These include the Rio Sao Luiz [sample

736
Jena) with a beam condenser. Spectral resolution was 6±10 cm±1
Concentrations of A and B nitrogen centers were calculated ac-
cordingto the IR absorption coecients speci®ed by Boyd et al.
01994, 1995).
No crystal-structure determinations by X-ray di€raction have
been carried out on the mineral inclusions reported here. The
mineral identi®cations therefore are based solely on chemical
analysis, optical data and mineral associations.
.
area ``1''; this probably includes the source of the Sao
Luiz diamonds studied by Harte et al. 01999)], the Rio
Mutum 0sample area ``3''), the Rio Vermelho 0sample
area ``4''), and Corrigo Chicora 0sample area ``5''). The
sample numbers used in the text below and in the ta-
bles give the area number 0the ®rst part) and the stone
number 0the second part).
Morphologically, diamonds from the Juina area are
similar to other Brazilian diamonds, showinga strong
predominance of rounded dodecahedral crystals. How-
ever, other characteristics of the Juina diamonds, such as
their mineral inclusions, their carbon isotopic composi-
tion and their uniformly low nitrogen contents, combine
to make them unique with respect to other diamond
populations.
The data presented here expand the range of mineral
associations in diamonds from the transition zone and
lower mantle, which can be recognized as a ``super-
deep'' paragenesis, distinct from the more common
peridotitic and eclogitic parageneses found in diamonds
from the subcontinental lithosphere. They further
de®ne the carbon-isotope and nitrogen impurity char-
acteristics of diamonds from the transition zone and
lower mantle.
Diamond morphology and coloration
Morphologically, the Juina diamonds are very homo-
geneous. Dodecahedroids are most common 055±70%
of the diamonds in the sample), and diamonds of
irregular shape account for approximately one third of
the total sample. Minor morphological types include
octahedra 0about 5%) and diamonds with combina-
tions of forms 04±7.5%). Dodecahedroids are com-
plexly distorted. A distinctive feature of dodecahedroid
diamonds from the Juina area is that their surfaces
show only shagreen microrelief. This is in contrast to
the wide variety of surface features generally present
on diamonds from other regions, including placer
deposits in Minas Gerais. The crystals are commonly
smoky brown 0from pale, almost colorless, to dark
brown) and semitransparent with a silky luster. They
show abundant plastic deformation bands and etch
channels.
Analytical methods
Major-element data on mineral inclusions have been obtained
usingtwo instruments, and the source of each analysis is marked
in the data tables. Analyses produced at the Department of Earth
and Planetary Sciences, Macquarie University, were done on a
CAMEBAX SX50 electron microprobe usinga range of natural
Mineral inclusions in diamonds
and synthetic mineral standards. Similar data from this microprobe Examination of the syngenetic mineral inclusions in the
have been independently veri®ed by cross analysis in Australia
0CSIRO) and Norway 0Mineralogical-Geological Museum). Other
phases of the superdeep paragenesis, in contrast to most
analyses were carried out by the Institute of Diamonds usinga
Juina diamonds reveals that they are dominated by
other diamond suites worldwide. Fourteen di€erent
CAMEBAX instrument in a commercial laboratory.
phases, identi®ed by their chemical composition and
optical properties, were found as inclusions in the Juina-
area diamonds: ferropericlase 0fPer), MgSi-perovskite
0MgSiPvk), CaSi-perovskite 0CaSiPvk), CaTi-perovskite
0CaTiPvk), sphene 0CaTiSiO5), TAPP, SiO2, low-Mg
ilmenite, picroilmenite, low-Cr, high-Ti spinel, majoritic
garnet, native Ni, unidenti®ed Si-Mg phase, and a phase
with the major-element composition of olivine. Minor-
element data and mineral association 0see below) indi-
The trace-element analyses used in this report have been
obtained with a laser-ablation ICPMS microprobe 0LAM-ICPMS)
at the Department of Earth and Planetary Sciences, Macquarie
University. The LAM analyses were carried out usingmethods
described in detail by Norman et al. 01996, 1998). The instrument
uses an Nd-YAG infrared laser, frequency-quadrupled to produce
a beam of UV 0266-nm) light which is focused through a petro-
graphic microscope onto a polished grain or thin section in the
sample cell. The gas ¯owing through the cell carries the ablated
sample to a Hewlett-Packard 4500 ICPMS. The external standard
for the analyses reported here was the NIST 612 glass. Ca was used
as the internal standard for garnet analyses, and Mg or Al for the cate that this phase cannot originally have been olivine.
other phases. Typical ablation pits were 40 lm in diameter and up
to 50 lm deep. Detection limits for most elements are from 1 ppm
``MgSiPvk'', ``CaTiPvk'' and ``CaTiPvk'' inclusions,
to 10 ppb, and precision at these levels is typically better than 10%
In the absence of XRD data, it is possible that the
after retrograde phase transformations, no longer have
0Norman et al. 1996, 1998).
the perovskite structure, as suggested by Joswig et al.
01999).
Most carbon-isotope analyses were carried out at the Vernad-
sky Institute, Moscow, usingstandard techniques 0Galimov 1985).
Samples were converted to CO2 by oxygen released from CuO at a
temperature of 800° C in a vacuum system. The CO2 was analyzed
usinga MAT230 mass spectrometer with a typical precision of
0.1&. Isotopic composition is given in delta notation 0parts per
thousand) as the deviation from the PDB standard. Several sup-
plementary analyses were carried out at the Division of Petroleum
Resources, CSIRO 0North Ryde), usingmethods described previ-
ously by Davies et al. 01999).
Ferropericlase coexists in single diamonds with
ilmenite, spinel, ``olivine'' and SiO2. CaSiPvk coexists
with sphene, and MgSiPvk coexists with TAPP. Maj-
oritic garnet coexists with CaTiPvk and Mn-ilmenite.
Neither chromite nor Cr-pyrope garnet was found
as inclusions. A total of 68 mineral inclusions were
extracted by crackingfrom 48 Juina diamonds and
Infrared 0IR) spectra were obtained at the Institute of Dia-
monds in Moscow, usinga Specord M-80 spectrometer 0Karl Zeiss, analyzed 0Table 1).

737
Table 1 Mineral inclusions
in the Juina diamonds
Mineral inclusions and associations
Number of diamonds 0sample no.)
Monomineral inclusions
Fe-periclase
20 0incl. three grains in no. 3±15,
and two grains in no. 5±103)
Ilmenite
Picroilmenite
Spinel
CaSi-perovskite
``Olivine''
SiO2
Associations
8
3
3
1
1
2
Fe-periclase+ilmenite
Fe-periclase+spinel
Fe-periclase+``olivine''
Fe-periclase+SiO2
CaSi-perovskite+Ni
CaSi-perovskite+``olivine''
CaSi-perovskite+``titanite''
MgSi-perovskite+TAPP
Majorite+ilmenite+CaTi-perovskite
+Si-Mgphase
2 0no. 1±4, and 1±7)
1 0no. 5±17 with two spinel grains)
1 0no. 5±5)
1 0no. 5±102 ± touching)
1 0no. 5±104 with three CaSi-perovskite grains)
1 0no. 4±101)
1 0no. 5±101 ± touching)
1 0no. 4±104 ± touching)
1 0no. 5±108 with three majorite grains)
Total: 14 minerals
48 diamonds with 68 mineral inclusions
Ferropericlase
The trace-element analyses show that the fPer has low
average contents of the REE and other LILE 0large ion
Ferropericlase 0fPer) is the most abundant inclusion lithophile elements), but near-chondritic abundances of
phase in the diamonds studied. Twenty-eight ferroperi- several transition elements 0Sc, V, Cr, Mn, Co) and
clase inclusions were extracted from 25 diamonds. They elevated abundances of some HFSE 0high ®eld strength
have equidimensional to ¯attened, sometimes irregular elements; Ti, Zr, Nb, Ta), as well as U and Th 0Table 2,
shapes, and range in size from 0.05 to 0.45´0.30 mm. The Fig. 3). Mn and Cr contents are negatively correlated
chemical compositions of ferropericlase inclusions from with Mg#, whereas Na shows a weak positive correla-
the Juina area range in mg [=Mg/0Mg+Fe)] from 0.491± tion with Mg#. Li abundances are high 05±10x chon-
0.826 0correspondingto a variation in FeO from 62.42± drites). The abundances of the LILE and HFSE are
26.39%; Table 2, Fig. 2). This is within the wider range highly variable, with some analyses showing high levels,
0mg=0.38±0.85) reported for 24 ferropericlases from Sao and others very low levels.
Luiz 0Harte et al. 1999). Three ferropericlase grains from
a single diamond 0no. 3±15) show a signi®cant range in
mg from 0.609±0.746. Harte et al. 01999) report a range of Ilmenite
0.634±0.819 for two ferropericlase inclusions in a single
diamond from Sao Luiz. This shows that disequilibrium Ilmenite inclusions are 0.05 to 0.40´0.30 mm in size. The
amonginclusions in single diamonds is common not only
in the upper mantle, but in the lower mantle as well.
14 ilmenite grains analyzed can be subdivided into low-
Mgilmenite and picroilmenite 0Table 2). Low-Mgil-
The ferropericlases have almost stoichiometric 0Mg, menites, with MgO contents less than 0.82 wt%, are the
Fe)O composition, with some interestingfeatures in predominant type in the Juina diamonds 011 of the 14
minor-element contents 0Table 2). As a rule, the most ilmenites analyzed). Low-Mgilmenite inclusions were
iron-rich ferropericlases have higher MnO contents reported by Meyer and Svisero 01980) in Brazilian dia-
which reach 0.85 wt%. A maximum of 0.8% MnO has monds which were probably from the Juina area. Low-
been reported in the Sao Luiz ferropericlases by Harte Mgand hihg-Mn ilmenite inclusions have also been
et al. 01999). TiO2 contents are 0±0.07 wt% in most cases. found in a few Venezuelan diamonds 0Kaminsky et al.
In grain no. 1±4, ferropericlase associated with ilmenite 1997; Sobolev et al. 1998). One of the Mn-rich ilmenites
contains 0.24% TiO2, the highest value yet reported. The 0no. 1±7) from the Juina area occurs in the same dia-
highest Cr2O3 content is 2.26 wt% 0no. 5±5), similar to mond as a ferropericlase inclusion; another 0no. 5±108)
the 2.56% Cr2O3 reported by Harte et al. 01999). Na2O occurs together with three grains of majoritic garnet and
contents reach 1.35 wt% 0sample no. 5±21), similar to a CaTi-perovskite. Four of the 11 low-Mgilmenites
the 1.25% found by Harte et al. 01999). The Ni content from the Juina diamonds 0nos. 1±7, 1±8, 1±32, and 5±
of ferropericlase is positively correlated with MgO con- 108) have very high MnO contents 02±11 wt%). Several
tent, and NiO contents up to 1.37% are found in the of the high-Mn ilmenites are heterogeneous. In a large
ferropericlases with mg>0.7. Most non-Brazilian fPer inclusion from diamond no. 1±7, MnO ranges from 4.8±
inclusions have Ni contents at the upper end of the range 11.5%, and Mn is irregularly distributed within the
found in the Juina fPer inclusions 0Fig. 2).
grain. This pattern may suggest unmixing of the grain

738

739

740
into high- and low-Mn compositions, possibly related to
decompression 0see below).
Picroilmenites are less abundant in the Juina diamonds
0three inclusions, grains no. 1±3, 5±52, and 5±56). These
high-Mg ilmenites are similar to many picroilmenite
inclusions reported in diamonds from other regions, but
are quite low in chromium 00.54±0.66% Cr2O3).
The trace-element signatures of the picroilmenites are
similar to those of kimberlitic picroilmenites 0Grin
et al. 1997), with high contents of Nb, Ta, Zr, Hf, Cr and
Ni, but very low levels of REE and LILE. In contrast,
the Mn-bearingilmenites have egnerally chondritic
levels of REE, with a U-shaped chondrite-normalized
pattern 0Fig. 3), as well as signi®cant levels of Sr, Ba, Th
and U. Like the picroilmenites, they have high contents
of Nb, Ta, Zr and Hf, but they also have very low Cr
and Ni contents compared to kimberlitic ilmenites.
Spinel
Five spinel-group inclusions were extracted from four
Juina diamonds 0two inclusions were present in no. 5±17
in association with ferropericlase). They are 0.05 to
0.25 ´ 0.20 mm in size.
Spinel-group inclusions in diamonds from other re-
gions show a well-de®ned range of major oxide chemistry
with Cr2O3 between 56 and 72 wt% 0Sobolev 1971). The
Al2O3 content of chrome spinel inclusions worldwide
varies between 3 and 12 wt%, and TiO2 contents rarely
exceed 1 wt% 0Sobolev 1971; W.L. Grin, unpublished
data). These features have been considered as diagnostic
of chrome spinels associated with diamonds 0Sobolev
1971; Vaganov et al. 1999). Table 2 and Fig. 4 show the
chemical composition of ®ve spinel inclusions from the
Juina area compared with the compositional ®elds of
chrome spinel inclusions in diamonds from other regions
0Vaganov et al. 1999). The Juina inclusions have signi®-
cantly lower Cr2O3 contents 0<36.5 wt%), and higher
TiO2 0>10 wt%) and FeO contents 0>33 wt%).
Although their Fe2O3 contents 0calculated from stoichi-
ometry) are much higher than in other known diamond-
related spinels, Fe3+/Fe2+ is within the normal range.
The trace-element data emphasize the di€erences
between these spinels and those commonly found in
diamonds of lithospheric derivation. Their V and Ga
contents are about twice the maximum values found in
kimberlitic chromites 0Grin et al. 1994; W.L. Grin,
unpublished data), whereas their Co and Zn contents are
less than half the minimum values found in ``normal''
kimberlitic ilmenites. Their high Nb and moderate Zr
contents, however, are similar to those of kimberlitic
ilmenites 0Grin et al. 1994).
CaSi-perovskite 0CaSiPvk)
Six inclusions of CaSiPvk have been found. Three occur
in one diamond 0no. 5±104), coexistingwith native Ni.

741
Fig. 2 NiO, MnO, and Na2O abundances 0relative to Cr2O3, wt%
and Mg#) in ferropericlase inclusions in diamonds from Juina, Sao
Luiz 0Harte et al. 1999) and worldwide: from Sloan 0Otter and
Gurney 1989), Koefontein 0Moore et al. 1986), River Ranch,
Zimbabwe 0Kopylova et al. 1997), Birim, Ghana 0Stachel and
Harris 1997), Mwadui, Tanzania 0Stachel et al. 1999), Lac des Gras
0Davies et al. 1999; Davies, unpublished data), Lesteng-la-Terai
0McDade and Harris 1999), and Kankan, Guinea 0Stachel et al.
2000b)
CaSiPvk is in contact with sphene in no. 5±101, and with
``olivine'' 0see below) in no. 4±101.
The Juina CaSiPvk inclusions are essentially pure
CaSiO3, with only traces of other elements, except for
grain no. 4±103 which has higher NiO 00.09%), Na2O
00.12%), and K2O 00.07%). The CaSiPvk has REE
contents ranging from 10±100x chondrites, with an
enrichment in the LREE relative to the HREE 0Fig. 5).
These inclusions di€er from the CaSiPvk reported by
Harte et al. 01999) in havingno positive Eu anomaly,
and a greater relative enrichment in LREE. In this re-
gard they are more similar to the CaSiPvk reported from
Kankan, Guinea, by Stachel et al. 02000b). The CaSiPvk
also contains high but variable levels of Sr, Ba, Zr, Nb,
Th and U, but low levels of transition elements.
CaTi-perovskite 0CaTiPvk)
CaTi-perovskite 0CaTiPvk) is found in association with
ilmenite, majoritic garnet 0three grains), and an
unidenti®ed Si-Mgphase in diamond no. 5±108. To our
knowledge, this is the ®rst ®nd of perovskite as an
inclusion in diamond. Assumingit to be a syngenetic
inclusion, its occurrence together with majoritic garnet
indicates derivation from sublithospheric depths 0see
below). It has an almost stoichiometric composition,
with only minor SiO2 01.05%), Al2O3 00.64%), and
Na2O 00.33%).
Fig. 3 Averaged chondrite-normalized trace-element abundances
of ferropericlase, Mn-ilmenite, picroilmenite 0P-ilm), Cr-spinel and
0Fe, Mg)2SiO4 inclusions in Juina diamonds. The chondrite-
normalized values used here and in Figs. 5 and 7 are those
recommended by GERM 0www-ep.es.llnl.gov/germ/)
MgSi-perovskite 0MgSiPvk)
They show little variation in composition, though the
largest grain has a slightly lower Ca content. Native Ni
was also reported in Sao Luiz diamonds by Wildinget al.
01991), and in DO27 diamonds by Davies et al. 01999).
MgSi-perovskite 0MgSiPvk) was found in sample no. 4±
104 as a small 0100-lm) grain, in contact with a phase
interpreted to be TAPP 0Harte et al. 1999). The low Ni

742
Fig. 5 Averaged chondrite-normalized trace-element abundances
of CaSi-perovskite inclusions in Juina diamonds, compared to the
range of values in similar inclusions from Sao Luiz 0Harte et al.
1999) and Kankan, Guinea 0Stachel et al. 2000b)
from the garnet structure. In the present sample no
discrete grains of cpx were observed to coexist with the
majorite. The trace-element analysis of the majorite
shows a ¯at REE pattern, with signi®cant levels of Sr, Zr
and Nb but low contents of transition elements 0Fig. 7).
The REE pattern is distinctly di€erent from the LREE-
depleted patterns shown by majoritic garnets from the
Monastery Mine 0Moore et al. 1991), but similar ¯at
patterns were reported for Sao Luiz majorite inclusions
by Harte 01992), and for Kankan majorite inclusions
0Stachel et al. 2000b).
Fig. 4 Fetotal and Ti 0cations) versus Cr/0Cr+Al) of Cr-spinel in
diamonds from Juina and the literature 0shaded area; n=210)
content of this inclusion 0Table 3) is not consistent with
it beinga peridotitic enstatite. It is similar in FeO and
Al2O3 contents to the most iron-rich MgSiPvk reported
by Harte et al. 01999) and, like that example, it coexists
with an inferred TAPP inclusion. The inferred TAPP
inclusion analyzed here has higher TiO2 07.55 vs. 4.2%)
and lower Al2O3 017 vs. 20.2%) than that described by
Harte et al. 01999) as coexistingwith MgSiPvk and iron-
rich fPer, and also has a lower mg value 00.812 vs. 0.940).
Its identi®cation is based on its association with
MgSiPvk and its anisotropy 0both of which make it
unlikely to be garnet) as well as its composition, espe-
cially the low Ca and very low Cr, which are similar to
those reported by Harte et al. 01999).
``Olivine''
Three colorless inclusions with the composition of oli-
vine were found in the Juina diamonds. One of them
0no. 5±5) occurs in the same diamond as a ferropericlase
inclusion, though the two inclusions are not in contact.
Another 0no. 4±101) occurs in association with CaSiPvk.
This association con®rms the ``olivine'' as a member of
the lower-mantle assemblage, and implies that the ``oli-
vine'' may originally have been another high-pressure
phase 0see below).
Majorite garnet
Majorite garnet was found as three grains associated
Olivines occurringas diamond inclusions worldwide
with CaTiPvk in a single diamond 0no. 5±108). The 0Fig. 8) fall within well-de®ned limits in composition,
mean compositions of the three grains show a range in with 91±95 mol% forsterite and minor amounts of Cr2O3
CaO from 10.94±12.18%. However, two of the grains 00.05±0.15 wt%), CaO 00.02±0.03 wt%) and NiO 00.25±
are signi®cantly heterogeneous in composition, with 0.35 wt%). The three Juina inclusions have forsterite
CaO varyingin the raneg 9.75±12.1% in no. 5±108c
contents 087±89 mol%) lower than those of typical dia-
0n=5), and 9.4±12.0% in no. 5±108b 0n=2). In these mond-inclusion olivines, and at the low end of the range
spot analyses, CaO and MgO are inversely correlated. for olivine in mantle peridotites 0O'Reilly et al. 1997). All
TiO2 varies from 1.5±3.2%, and shows no correlation of the Juina ``olivine'' inclusions have high Ca contents
with CaO or MgO. XSi varies from 2.99±3.18 0Fig. 6), compared to typical mantle olivines. Two analyzed by
and it is positively correlated with CaO content. This LAM-ICPMS contain 1,300±1,900 ppm Ca, whereas the
pattern of compositional variation is similar to that one coexistingwith CaSiPvk contains 0.06% CaO 0EMP
described by Wildinget al. 01991) who interpreted it as analysis). Compared to typical mantle-derived olivines of
re¯ectinga very small-scale unmixingof clinopyroxene
similar FeO content, the Ni contents of the Juina olivines

743
are in the same range, whereas their Zn contents are
25±30% higher and their Mn contents are 50% lower
0O'Reilly et al. 1997). Compared to other diamond-
inclusion olivines worldwide, the Juina phases also have
high Cr contents 0Fig. 8). Contents of HFSE and REE
are below detection in most cases.
A similarly low-Mgolivine 0Fo 88.4) has been reported
as a diamond inclusion in the Canadian DO27 kimber-
lite pipe where numerous ferropericlase inclusions were
found 0Davies et al. 1999). A single olivine inclusion
with Fo78.8 composition was found in the Jagersfontein
kimberlite 0Chinn et al. 1998). Wildinget al. 01991)
reported an olivine inclusion of Fo85 composition in a
diamond from Sao Luiz. If our interpretation of this
phase is correct, these other reports of Fe-rich ``olivine''
from localities containingsuperdeep diamonds may also
represent an inverted superdeep phase. This suggestion
could be tested by trace-element analysis. Figure 8 shows
that some of these ``olivine'' inclusions have low Ni and
high Mn compared to olivine inclusions from localities
without fPer-bearingdiamonds.
Fig. 6 Al+Cr versus Si 0atomic proportions calculated from 12
oxygen) for majorite garnets from Juina, compared with worldwide
eclogitic garnet inclusions in diamond 0n=610). Crosses Experi-
mental data of Irifune 01987), labelled with pressure 0GPa)
SiO2
A SiO2 phase was found in three specimens 0nos. 4±102,
5±102, and 5±107). It is virtually pure SiO2. Minor NiO
00.06%) is present in this phase in sample no. 5±102
where SiO2 is associated with ferropericlase. Because of
this paragenesis, and by analogy with other SiO2 inclu-
sions found in the superdeep association, we assume that
this phase may originally have formed as stishovite.
Native Ni
Native Ni was found in association with CaSiPvk in
sample no. 5±104. It has approximately 1% Fe admix-
ture and minor amounts of Ti and Mn. The presence of
potassium 00.21%) is a remarkable 0and unexplained)
feature of this Ni grain. Native Ni was also reported
in Sao Luiz diamonds by Wildinget al. 01991), and in
DO27 diamonds by Davies et al. 01999).
Fig. 7 Averaged chondrite-normalized trace-element abundances
in majoritic garnet inclusions from Juina diamonds compared to
those from Kankan, Guinea 0Stachel et al. 2000a)
Si-Mgphase
An unidenti®ed Si-Mgphase was found in association
with ilmenite, majorite and perovskite in sample no. 5±
108. No elements other than Mgand Si were recorded in
the phase, which gives a low analytical sum suggesting
the presence of volatile components.
Sphene 0titanite)
Sphene 0titanite) was recorded as a large 0300-lm) grain
in association with CaSiPvk in sample no. 5±101. The
two phases are in contact, implyinga superdeep origin
for the titanite. This is in contrast to the recently re-
ported ``sphene'' inclusion, associated with magnetite, in
a Dokolwayo diamond 0Daniels and Gurney 1999). The
TiO2 content is low, which suggests the presence of other
elements 0the phase has not been analyzed by LAM-
ICPMS). We do not have crystallographic data on this
inclusion. Assumingthat it is a syngenetic inclusion, we
speculate that it may represent a high-pressure poly-
morph of CaTiSiO5, bearingthe same relationship to
titanite as the TAPP does to almandine-pyrope garnet.
Carbon-isotope composition of diamonds
Eighty-two representative diamonds from the Juina
area, 31 of them with mineral inclusions, have been
analyzed for carbon-isotope composition 0Table 4 and
Fig. 9). For some specimens two or more chips were
analyzed. These diamonds appeared to be homogeneous
in carbon isotopic composition, the di€erence between
the chips not exceeding0.4 &.

744

745
Fig. 8 NiO, MnO and Cr2O3 0wt%) versus Mg#, compared with
those of olivine inclusions in diamonds from the literature 0shaded
area; n=377) and 0Fe, Mg)2SiO4 inclusions in diamonds from
Juina and other superdeep diamond localities worldwide 0Rickard
et al. 1989; Stachel and Harris 1997; Davies et al. 1999; Harte et al.
1999; Stachel et al. 2000b; Davies, unpublished data). Points lying
outside the shaded area may have formed as the spinel-structured
c-Mg2SiO4 phase in the deep part of the transition zone
No correlation between carbon isotopic composition
and diamond morphology was found. This result is
consistent with the majority of previous studies 0e.g.,
Deines 1980).
With one exception, the diamonds studied are char-
acterized by a narrow range of carbon isotopic compo-
sition varyingfrom d13C=±2.5 to ±7.8&. No di€erences
related to inclusion type are apparent. These values fall
within the d13C range for diamonds of peridotitic
paragenesis 0d13C from ±1 to ±9&), and the mode at ±
4.9& d13C 0standard deviation ‹0.9&) is close to the
``mantle value'' of ±4.5 to ±5.0& 0e.g. Galimov 1968;
Mattey 1987; Deines 1992). A similar isotopic compo-
sition 0mode at d13C=±5&) was reported for Sao Luiz
diamonds of the superdeep paragenesis by Hutchison
et al. 01999). The data do not suggest any di€erences
in isotopic composition between the three localities
studied. The source material for most of the Juina dia-

746
Table 4 Carbon isotope composition of the Juina diamonds
Table 4 0Contd.)
d13C 0& PDB)
Diamond Morphology
no.
Inclusions
d13C 0& PDB)
Diamond Morphology
no.
Inclusions
Moscow CSIRO
Moscow CSIRO
Area 1. Rio Sao Luis
Area 5. Corrigo Chicora
1±1
1±2
1±4
1±5
1±6
1±7
1±8
Dodecahedroid
Dodecahedroid
Dodecahedroid
Uncertain
Dodecahedroid
Dodecahedroid
Combination O-D
Dodecahedroid
Dodecahedroid
Dodecahedroid
Uncertain
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Combination O-D
Octahedron
Dodecahedroid
Dodecahedroid
Octahedron
Ilm
Ilm
fPer+Ilm
Ilm
±6.3
±3.9
±4.1
±5.6
±5.0
±4.2
±5.3
±5.1
±5.7
±5.9
±4.1
±5.0
±5.0
±5.3
±4.6
±5.0
±4.9
±5.2
±5.3
±4.5
±3.9
±7.8
±6.9
±5.1
±5.0
5±20
5±21
5±26
5±29
5±30
5±34
Uncertain
±6.8
±4.7
±4.2
±4.4
±4.6
±6.1
Dodecahedroid
Combination O-D
Dodecahedroid
Uncertain
Aggregate of
octahedra
fPer
fPer+Ilm
Ilm
1±10
1±11
1±12
1±14
1±18
1±21
1±22
1±24
1±27
1±30
1±31
1±32
1±33
1±34
1±35
1±36
1±37
1±38
5±35
5±52
5±53
5±54
5±55
5±56
5±57
5±58
5±59
5±60
5±76
5±77
5±78
5±79
5±80
5±101
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Combination O-D
Twin of octahedra
Uncertain
±7.8
±4.3
±4.8
±5.0
±4.3
±4.9
±4.8
±4.6
±5.1
±5.2
±6.0
±5.2
±5.6
±4.9
±6.2
±4.4
Pilm
Spinel
Pilm
Ilm
fPer
Ilm
fPer
Ilm
Spinel
fPer
Uncertain
Dodecahedroid
Octahedron
Dodecahedroid
Dodecahedroid
fPer
CaSiPvk+
Spene
CaSiPvk+Ni
Maj+Ilm+
Pvk+SiMg
fPer
5±104
5±108
Uncertain
Uncertain
±4.3
±11.4
Area 4. Rio Vermelho
4±4
4±7
4±18
4±19
4±20
4±21
4±22
4±23
Dodecahedroid
Dodecahedroid
Dodecahedroid
Uncertain
Uncertain
Uncertain
Uncertain
Aggregate of
octahedra
Octahedron
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Dodecahedroid
Uncertain
Spinel
Ilm
fPer
±4.5
±4.5
±3.7
±6.8
±5.0
±5.3
±6.1
±4.4
monds appears to be isotopically homogeneous. No
statistically signi®cant di€erences were observed for di-
amonds with di€erent mineral inclusions, and stones
without inclusions. This supports the idea that the
source of carbon for all the lower-mantle diamonds from
the Juina area was isotopically homogenous. However,
the narrow range of d13C by itself cannot serve as an
indication of the lower-mantle paragenesis, narrow dis-
tributions of d13C havingbeen reported for diamonds
with ``normal'' inclusions 0e.g. Finsch and Premier;
Deines et al. 1989).
Diamond no. 5±108, which contained three majoritic
garnet inclusions, has d13C=±11.4, similar to values for
many eclogitic diamonds worldwide. Hutchison et al.
01999) also reported d13C values as low as ±12.4 in
``transition-zone'' diamonds from Sao Luiz, containing
majoritic garnet inclusions. These data suggest that most
of the superdeep diamonds from the Juina area belong
to a single population, but that the majorite-bearing
``transition-zone'' diamonds represent a di€erent popu-
lation with a di€erent source material, as was suggested
by Harte et al. 01999).
4±24
4±25
4±26
4±27
4±28
4±29
4±30
4±31
4±32
4±33
4±70
4±71
4±72
4±73
4±74
4±101
±4.8
±4.9
±4.9
±3.0
±4.4
±4.3
±2.9
±5.4
±5.1
±4.3
±4.3
±3.0
±2.5
±4.3
±3.9
±4.9
Uncertain
Dodecahedroid
Dodecahedroid
Uncertain
Uncertain
Uncertain
Dodecahedroid
Dodecahedroid
Uncertain
fPer
CaSiPvk+
``olivine''
CaSiPvk
MgSiPvk+
TAPP
4±103
4±104
Uncertain
Uncertain
±4.8
±5.1
Area 5. Corrigo Chicora
5±3
5±7
5±8
5±11
Combination O-D
Dodecahedroid
Combination O-D
Aggregate of
octahedra
±5.0
±5.1
±4.8
±4.5
fPer
A similar range of carbon-isotope composition has
been reported for diamonds with superdeep inclusions
from other localities. The diamonds with superdeep
inclusions from the Canadian DO27 pipe have a d13C
mode at ±4&, and three type-II diamonds from the same
pipe, inferred to be of the superdeep paragenesis, range
from ±3 to ±6& d13C 0Davies et al. 1999). The analysis of
5±13
5±16
5±17
Octahedron
±4.8
±5.6
±5.1
Dodecahedroid
Twin of
dodecahedroids
fPer+
spinel

747
Eighty percent of the diamonds containing nitrogen
show platelet degradation with no detectable IR peak.
Relatively weak hydrogen-related peaks were detected in
many samples. The intensity of the hydrogen-related
lines in the IR absorption of diamonds 03,107 and
1,405 cm±1) cannot be used for estimation of the hy-
drogen content of the diamond. As was demonstrated by
Sellschop 01992), and by Sweeney and Winter 01999), the
absolute hydrogen content of diamond is not correlated
with the IR absorption. It is likely that the above-men-
tioned lines arise from hydrogen-terminating broken
bonds at the internal surfaces 0cracks, inclusions, voids,
etc.). As was shown by Ki¯awi et al. 01996), hydrogen
may enter diamond in an optically inactive form, and
special annealingconverts only part of it to an optically
active form.
The FTIR characteristics of the Juina diamonds are
comparable to those of the Sao Luiz diamonds, where
more than 90% 025 of 36 studied) of ``lower-mantle''
diamonds are low-nitrogen type-II varieties 0average
of <20 at. ppm N), whereas eleven diamonds contain
aggregated B centers 0Hutchison et al. 1999). Type-II
diamond is also typical of ``superdeep'' diamonds from
the DO27 kimberlite in Canada, where the only type-I
diamond with superdeep inclusions contains nitrogen
aggregated to the pure IaB form 0Davies et al. 1999). A
single diamond with a ferropericlase inclusion from the
Letseng-la-Terai pipe, Tanzania is also type-II 0McDade
and Harris 1999).
At the Institute of Diamonds in Moscow, we have
studied many thousands of diamonds from Siberia,
Arkhangelsk, South Africa, Venezuela and Brazil.
Nitrogen-free and low-nitrogen diamonds comprise
approximately 1% of the diamond population in the
majority of kimberlite pipes. Only a few exceptions are
reported. These include the South African Premier
diamonds, which have a high proportion of nitrogen-
free diamonds 0Smirnov et al. 1986), and diamonds
from Jagersfontein 0approximately 60% type-IIa;
Chinn et al. 1998). In comparison to the worldwide
data, the Juina diamonds are distinguished by their low
contents of nitrogen and their strongly aggregated B
centers. The results of the IR analysis con®rm that the
superdeep, ``lower-mantle'' diamonds di€er substan-
tially in their nitrogen impurity characteristics from the
majority of typical kimberlite-related diamonds studied
previously.
Fig. 9 Carbon-isotope compositions of diamonds from Juina
0& PDB), compared to diamonds from Sao Luiz 0Hutchison et
al. 1999) and other localities hostingferropericlase 0Deines et al.
1991; McDade and Harris 1999; Hutchison et al. 1999) and other
superdeep phases 0Davies et al. 1999). LM Lower mantle; TZ
transition zone
the only diamond from the Letseng-la-Terai pipe with a
ferropericlase inclusion gave d13C=±3& 0McDade and
Harris 1999), and similar isotopic compositions have
been found for the Koefontein ``lower-mantle'' dia-
monds 0±5.0 to ±5.2& d13C; Deines et al. 1991).
Nitrogen impurities
Structural impurities in 98 Juina diamonds were studied
by infrared 0IR) spectroscopy. The Juina diamonds
contain a large proportion 0approximately 45%) of
nitrogen-free 0type-IIa) diamonds. A- and B-centers in
these diamonds do not exceed 20 at. ppm each. Among
the remaining55% of the stones 0which may be classed
as type-I diamonds), 80% are almost pure IaB with
B-center contents usually from 40±205 at. ppm, and up
to 460 ppm in one specimen. The A-center, if present
0in 20% of type-I diamonds), ranges from 21±42 at.
ppm. Only in a few samples does it reach 80±97 at. ppm.
Discussion and conclusions
An expanded lower-mantle diamond inclusion
paragenesis
The presence of lower-mantle minerals in kimberlites
was predicted by Ringwood and Lovering 01970).
However, the ®rst real evidence for this hypothesis ap-
peared only when the ®rst ferropericlase inclusions were
reported in diamonds from the Orroroo kimberlite,

748
South Australia, and from the Koefontein kimberlite, pipes 0Otter and Gurney 1989; Stachel et al. 1999;
South Africa 0Scott-Smith et al. 1984). These inclusions Davies et al. 1999), and in Guinean 0Harte et al. 1998;
were interpreted as formed under pressure exceeding Stachel et al. 2000a, 2000b) and Venezuelan 0Kaminsky
255 kbar, and the authors concluded, ``the depth of et al. 1997) placer deposits.
formation for some diamonds may be considerably
In the absence of the fPer+SiO2 assemblage, or other
greater than previously considered'' 0Scott-Smith et al. coexistingsuperdeep phases, fPer alone is not necessarily
1984, p. 138). Based on di€erent grounds 0density of diagnostic of lower-mantle conditions. However, most
nitrogen compounds in voidites), Bruley and Brown reported fPer inclusions from localities other than Juina/
01989) also suggested that some diamonds were formed Sao Luiz have compositions close to those predicted for
at pressures correspondingto depths of 350±600 km or high-pressure peridotitic assemblages from experimental
even deeper.
Harte, Harris and coworkers 0Wildinget al. 1991;
work 0Kesson and Fitz Gerald 1991), and to those
known to coexist with SiO2 or CaPvk 0Harte at al. 1999).
Harte and Harris 1993, 1994; Harte et al. 1994; Hutch- Such lower-mantle diamond inclusions are known now
ison et al. 1995; Harris et al. 1997; McCammon et al. in almost all diamondiferous provinces 0Table 5). In most
1997; Harte et al. 1998, 1999; Hutchison et al. 1999) provinces reports include only a few grains, but at least
developed this idea further, based on the study of dia- 15% of all diamond inclusions in the Canadian DO27
monds from the Sao Luiz placer. These diamonds were pipe are of the superdeep paragenesis 0Davies et al. 1999).
believed to be derived from the Cretaceous Aripuena
kimberlite province 0Wildinget al. 1991). We suppose
The lower-mantle paragenesis reported by Harte and
coworkers 0summarized by Harte et al. 1999) includes,
that these diamonds come from a placer in the southern besides ferropericlase 0fPer), two inferred perovskite
part of the Juina area 0Fig. 1). In addition to ferro- phases representing high-P polymorphs of MgSiO3
periclase, these authors found other superdeep mineral 0MgSiPvk) and CaSiO3 0CaSiPvk), a tetragonal poly-
inclusions, analyzed the IR and carbon-isotope charac- morph of pyrope garnet 0TAPP), and stishovite 0SiO2).
teristics of the Sao Luiz diamonds, and concluded that The coexistence of fPer+CaSiPvk+MgSiPvk+SiO2‹
these diamonds crystallized in the transition zone and TAPP in this assemblage is consistent with that pre-
the lower mantle.
dicted from experimental studies of peridotitic compo-
Single ferropericlase inclusions have been reported sitions at high pressure 0Kesson and Fitz Gerald 1991;
from the southern African Monastery, River Ranch and Harte et al. 1999). This assemblage represents the high-
Letseng-la-Terai kimberlite pipes 0Moore and Gurney pressure equivalent of the olivine, pyroxenes and garnet
1989; Kopylova et al. 1997; McDade and Harris 1999). of typical lithospheric mantle peridotites, at pressures
More recently, ferropericlase has been found in dia- greater than ca. 22 GPa. The assemblage fPer+Mg-
monds from Tanzanian and North American kimberlite SiPvk+SiO2 has been reported in diamonds from the
Table 5 Finds of ferropericlase inclusions in diamonds
No. Province
Locality 0pipe, country)
Reference
Number of grains
Association
1
2
3
4
5
6
7
Australia
Southern Africa
Orroroo, Australia
Koefontein, RSA
Monastery, RSA
Letseng-la-Terai, Lesotho
River Ranch, Zimbabwe
Placer deposit, Guinea
Kankan placer deposit,
Guinea
Scott-Smith et al. 01984)
Scott-Smith et al. 01984)
Moore and Gurney 01989)
McDade and Harris 01999)
Kopylova et al. 01997)
Harte et al. 01999)
2
4
1
1
1
Enst. MgSiPvk
Ecl-Ga
Western Africa
2
1
Joswiget al. 01999)
CaSiPvk+Opx
8
9
Eastern Africa
Siberia
Mwadui, Tanzania
Udachnaya, Russia
0Yakutia)
Stachel et al. 01999)
Zedgenizov et al. 01998)
1
2
10
11
Dianga, Russia 0Yakutia)
Placer deposit, Russia
0Yakutia)
Sobolev et al. 01999)
Sobolev et al. 01999)
2
2
Ecl-Ga
12
13
14
North America
South America
Sloan, USA
DO27, Canada
Guaniamo placer deposit,
Venezuela
Rio Sao Luiz placer deposit, Wildinget al. 01991)
Otter and Gurney 01989)
Davies et al. 01999)
Kaminsky et al. 01997)
1
1
8 015% incl.)
Ni+MgSiO3
15
16
36 024 anal.)
Brazil
Harte and Harris 01994)
Harte et al. 01994)
Hutchison et al. 01995, 1999)
Harte et al. 01998, 1999)
Enst, Ca-silicate
MgSiPvk, CaSiPvk,
TAPP, SiO2
Ilm, Sp, Ol, SiO2
Juina placer deposits, Brazil This work
28 0>40% incl.)
Total=93 081 anal.)

749
Slave Craton 0Davies et al. 1999). The diamonds studied lized, with extra Fe beingadded through core-mantle
for this report contain all of the inclusion phases pre- interaction near the ``D'' boundary. However, the pres-
viously reported by Harte et al. 01999, and references ence of phases such as ilmenite and Ti, Fe-rich Cr-spinel,
therein), and the unusual range of mg in the present suite and even the high Ti content of the TAPP phase, indicate
of ferropericlases reported by Harte et al. 01999) is that the rocks in which the Juina diamonds crystallized
replicated in the samples studied here. We therefore were also enriched in Ti which is not an obvious signature
consider that the ferropericlases found in this study are of core-mantle interaction. In this respect, they are sim-
of lower-mantle origin, and that phases coexisting with ilar to some types of highly metasomatized peridotites
them are of similar origin.
On this basis, we can add ®ve phases, each of which
known from the lithospheric mantle.
The available trace-element data on the lower-mantle
occurs together with ferropericlase and/or CaSiPvk in at phases from Sao Luiz 0Harte et al. 1999; Tables 2 and 3,
least one diamond, to the previously reported inclusion Figs. 3 and 5) suggest that, relative to the primitive
assemblage from this region: low-Mg manganiferous il- mantle, the parent rock also was signi®cantly enriched in
menite, titaniferous chrome spinel, Mg2SiO4, native Ni, Sr, Ba, LREE, HFSE, and possibly Th and U. In the
and titanite. Of the mineral inclusions recognized thus phases analyzed, Zr/Hf is higher than the chondritic
far in the Juina diamonds, only picroilmenite has not ratio, whereas Nb/Ta is lower. How this enrichment in
been con®rmed as belonging to the lower-mantle or LIL and HFSE elements has been produced near the
transition-zone assemblages.
660-km discontinuity is not clear. Mixingwith sub-
There are no experimental data on the stability of ducted material is a plausible model. A positive Eu
ilmenite at these pressures, but the ilmenite structure is anomaly found in the Sao Luiz CaSiPvk 0Harte et al.
stable in the MgSiO3 composition at pressures of ca. 1999) could support the idea of a subducted origin for
22 GPa 0Ringwood 1989; Fig. 4). The mineral associa- the protolith. However, the CaSiPvk inclusions analyzed
tions in the Juina diamonds suggest that a range of 0Mg, in this work do not show an Eu anomaly. The overall
Fe, Mn)TiO3 phases is also stable in this pressure range, carbon-isotope homogeneity of the lower-mantle dia-
given Ti-rich bulk compositions. Liu 01980a, b) synthe- monds does not require the addition of crustal material
sized a polymorph of MgAl2O4 stable at 20±25 GPa, and through subduction processes. The 13C-depleted carbon
Ringwood 01989) suggested that ``in the lower mantle, of the one majorite-bearingdiamond may provide
alumina will be contained in this, or a related phase''. evidence for the presence of such a recycled crustal
We suggest that the titaniferous spinel-group phase component within the transition zone. Further car-
found in the Juina diamonds 0coexistingwith fPer)
bon-isotope studies of majorite-bearingdiamonds from
represent such a high-pressure spinel polymorph. This the area, such as those studied by Wildinget al. 01991),
phase has not been observed in experimental studies of are required to test this suggestion.
peridotite compositions, but it may be stabilized in the
relatively Fe, Ti-rich bulk compositions represented by
the present inclusion suite. If this is the case, these spinel Transition-zone inclusions
phases may also provide a residence for Al in parts of
the lower mantle.
The occurrence in the Sao Luiz diamonds of majoritic
Olivine is not stable together with fPer in mantle garnets, many partially to completely unmixed to gar-
peridotite compositions, but a phase with olivine com- net+clinopyroxene, was documented by Wildinget al.
position and stoichiometry coexists with fPer in dia- 01991), and the two inclusions of this type found here
mond no. 5±5, and with CaSiPvk in diamond no. 4±101. show similar small-scale chemical variation, suggesting
However, a spinel structure is stable in the Mg2SiO4 partial unmixingof clinopyroxene from majorite. The
composition just above the 660-km discontinuity most Si-rich garnet compositions within diamond no.
0Ringwood 1989; Fig. 4). We therefore suggest that the 5±108 are consistent with derivation from depths of
relatively Fe-, Ca-, Cr-rich ``olivine'' observed in this 350±400 km 0Fig. 6; Irifune 1987). Given that some cpx
study may represent this c-Mg2SiO4 phase. An alterna- has already unmixed from this phase, this is a minimum
tive possibility is that it formed through the decom- estimate of the depth of diamond crystallization. The
pression reaction fPer+MgSiPvk=olivine. However, coexistence of this majoritic garnet with CaTi-perovskite
this would require that each inclusion originally repre- and Mn-ilmenite, and the high TiO2 content of the
sented a touchingpair of the two high-pressure phases.
Given the rarity of touchinginclusions in diamonds
generally, this appears unlikely.
majorite, indicate a Ti-rich environment of crystalli-
zation, and hence a link with the Ti-rich superdeep
assemblage discussed above. The occurrence of ilmenite,
Harte et al. 01999) noted that the wide range of mg in and especially the Mn-rich ilmenite, in both the ``lower-
the Sao Luiz ferropericlases contrasts with the narrow mantle'' and ``transition-zone'' assemblages is a further
range 00.866±0.890) found in most other localities 0Fig. 2), link between the two. We therefore suggest that dia-
and with that predicted by experimental studies of mond no. 5±108 may be derived from the deepest part of
peridotitic compositions 0Kesson and Fitz Gerald 1991). the transition zone, near the 660-km discontinuity, and
They suggested that this might imply a deep-mantle ori- that its majorite inclusions have partially unmixed dur-
gin for the protolith within which the diamonds crystal- ingascent. As noted above, the carbon-isotope compo-

750
sition of this diamond may indicate crystallization from growth temperature and is inversely proportional to ni-
recycled crustal material.
trogen solubility in the metal alloy. It is also known that
Ti or Zr serve as e€ective ``getters'' of nitrogen. The ti-
tanium-rich minerals observed in the Juina diamonds
might serve as an e€ective trap for nitrogen, thus de-
creasingits concentration in the diamond lattice.
Nitrogen impurities: implications for the source rocks
As noted above, the Juina superdeep diamonds di€er
from most suites of lithospheric diamonds in having
unusually low mean N contents 0with nearly half of the Tectonic implications
samples beingtype-II diamonds), and in the unusually
high degree of B-center aggregation in the diamonds The presence of lower-mantle diamonds is prima facie
that do contain N at measurable levels. They appear to evidence for plume tectonics, because 0in the absence of
share these characteristics with other diamonds of the rapid whole-mantle convection) a mechanism is required
superdeep paragenesis.
to bringthese samples from lower-mantle depths to a
It has been well demonstrated that the degree of level 0probably <250 km) where the eruption of kim-
nitrogen aggregation is a function of temperature and berlites or other magmas can bring them to the surface
time in both lines Ib±IaA 0Taylor et al. 1996) and IaA- 0Haggerty 1991, 1999; Grin et al. 1999). In most places
IaB 0Mendelssohn and Milledge 1995). The high pro- where superdeep diamonds occur, they represent a very
portion of IaB diamonds amongthe Juina diamonds
small proportion of the diamond population. The great
indicates that these diamonds had a prolonged high- majority of diamonds at these localities appear to be
temperature history, which resulted in the almost derived from the stable subcontinental lithosphere.
complete transformation of single-atomic and paired
The highest proportion of superdeep diamonds pre-
nitrogen centers into polyatomic complexes of B type. viously reported is from the DO27 kimberlite pipe, Slave
This suggests that diamond crystallization and mantle Province, where at least 15% of all diamond inclusions
storage occurred at high temperatures and under high are of the superdeep paragenesis, and perhaps 30% of all
stress.
Another interestingfeature of the Juina ``lower-
diamonds may belongto this paragenesis 0Davies et al.
1999, unpublished data). The lithospheric section be-
mantle'' diamonds is that eighty percent of the diamonds neath the Slave Province shows an unusually strong
containing nitrogen have degraded platelets with no compositional layering, with an upper strongly depleted
detectable peak. As was shown by Woods 01986), dia- layer separated from a lower layer of more typical
monds can be divided into two general types,these being Archean composition by a sharp boundary at 150‹10 km
``regular'', in which the intensity of platelet peak is 0Grin et al. 1999). Grin et al. 01999) therefore have
proportional to the intensity of absorption by B-defect, suggested that the lower part of the strongly layered
and ``irregular'', where this correlation does not exist. lithospheric mantle beneath the Slave Province may
Subsequent studies 0Evans et al. 1995) have revealed that consist of plume-generated material that rose from the
``irregular'' diamonds contain low amounts of platelets 660-km discontinuity. This model is intended to explain
but abundant dislocation loops and voidites. Evans et al. both the compositional layeringand the presence of
01995) suggested that, during their post-growth history, abundant superdeep diamonds. Gaul et al. 02000), by
``irregular'' diamonds underwent annealing under rela- mappingthe variation of olivine composition with depth,
tively high temperatures, which lead to the degradation have shown that similar layeringmay exist beneath the
of platelets and to the creation of voidites. The present South Australian and Kaapvaal cratons where superdeep
study indicates that many of the Juina diamonds belong diamonds have been reported, but not beneath the Dal-
to the ``irregular'' type. This implies that after crystalli- dyn kimberlite of Siberia where such diamonds appear to
zation a signi®cant fraction of these diamonds was be absent. Further studies are clearly required, but a link
subjected to annealingat high temperatures, probably between the presence of superdeep diamonds and litho-
exceedingthat of rgowth. It would be interestingto spheric structure, if substantiated, would suggest that
check for the existence of voidites in ``lower-mantle'' plume tectonics have contributed to the development of
diamonds by high-resolution TEM.
The reason for the low nitrogen contents of most su-
some lithospheric roots.
The Juina area is unique amongknown diamond
perdeep diamonds, from the Juina area as well from provinces in havinga marked predominance of lower-
other localities, is still a subject of debate. One of the mantle and transition-zone diamonds. The lack of
frequently discussed possibilities is that the rocks in the ``normal'' mineral inclusions 0such as pyrope garnet and
diamond protolith have low nitrogen contents. However, chromite, or eclogitic phases) in the Juina diamond
it should be kept in mind that the distribution coecient population, their homogeneous carbon-isotope compo-
of nitrogen between the rock and diamond is a complex sition, and the uniformly low nitrogen contents strongly
function of chemical composition and PT-fO2 conditions suggest that the overwhelming majority of the diamonds
of growth. For example, it has been convincingly in this area 0if not all of them) are of superdeep origin.
demonstrated 0e.g. Burns et al. 1999) that the nitrogen The Juina sample therefore may be considered as the
content of synthetic diamond strongly depends upon the end-member example of this paragenetic association.

751
publication no. 229 from the ARC National Key Centre for the
Geochemical Evolution and Metallogeny of Continents 0GE-
MOC).
The apparent absence of lithospheric diamonds, or
ultrama®c and eclogitic paragenesis, suggests that the
area did not possess
a deep lithospheric root,
extendinginto the diamond stability ®eld, at the time
the diamonds were brought to the surface. All dia-
monds in this area were derived from transition-zone
and lower-mantle material raised to the base of the
lithosphere, probably through the action of a plume,
before eruption of magmatic rocks that brought the
diamonds to the surface. The nature of the diamond-
inclusion minerals suggests that the diamond-bearing
layer thus formed may be of a signi®cantly di€erent
composition from the mantle commonly associated
with lithospheric diamonds. If this is the case, the
nature of ``indicator minerals'' associated with these
diamonds may also not conform to traditional
models.
References
Boyd SR, Ki¯awi I, Woods GS 01994) The relationship between
infrared absorption and the A defect concentration in diamond.
Philos MagB69:1149±1153
Boyd SR, Ki¯awi I, Woods GS 01995) Infrared absorption by the
B nitrogen aggregate in diamond. Philos Mag B72:351±361
Bruley J, Brown LM 01989) Quantitative electron energy-loss
spectroscopy microanalysis of platelet and voidite defects in
natural diamond. Philos MagA59:247±261
Burns RS, Hansen JO, Spits RA, Sibanda M, Welbourn CM,
Welch DL 01999) Growth of high purity large synthetic dia-
mond crystals. Diamond Related Mater 8:1433±1437
Chinn IL, Milledge HJ, Gurney JJ 01998) Diamonds and inclusions
from Jagersfontein kimberlite. Abstr 7th Int Kimberlite Conf,
Cape Town, pp 156±157
Daniels LM, Gurney JJ 01999) Diamond inclusions from the
Dokolwayo kimberlite, Swaziland. In: Gurney JJ, Gurney JL,
Pascoe MD, Richardson SH 0eds) Proc 7th Int Kimberlite
Conf, vol 1, Cape Town. Red Roof Design, Cape Town,
pp 143±147
Davies RM, Grin WL, O'Reilly SY 01999) Diamonds from the
deep: pipe DO27, Slave craton, Canada. In: Gurney JJ, Gurney
JL, Pascoe MD, Richardson SH 0eds) Proc 7th Int Kimberlite
Conf, vol 1, Cape Town. Red Roof Design, Cape Town,
pp 148±155
Deines P 01980) The carbon isotopic composition of diamonds:
relationship to diamond shape, color, occurrence and vapor
composition. Geochim Cosmochim Acta 44:943±961
Deines P 01992) Mantle carbon: concentration, mode of occur-
rence, and isotopic composition In: Schidlowski M, Golubic S,
Kimberley MM, McKirdy DM, Trudinger PA 0eds) Early
organic evolution: implications for mineral and energy
resources. Springer, Berlin Heidelberg New York, pp 133±146
Deines P, Harris JW, Spear PM, Gurney JJ 01989) Nitrogen and
13C content of Finsch and Premier diamonds and their impli-
cations. Geochim Cosmochim Acta 53:1367±1378
Deines P, Harris JW, Gurney JJ 01991) The carbon isotopic com-
position and nitrogen content of lithospheric and astheno-
spheric diamonds from the Jagersfontein and Koefontein
kimberlite, South Africa. Geochim Cosmochim Acta 55:2615±
2625
Evans T, Ki¯awi I, Luyten W 01995) Conversion of platelets into
dislocation loops and voidite formation in type IaB diamonds.
Proc R Soc Lond A449:295
Galimov EM 01968) Geochemistry of carbon isotopes 0in Russian).
Nedra PublishingHouse, Moscow
Galimov EM 01985) The relation between formation conditions
and variations in isotope composition of diamonds. Geochem
Int 0translated from Geokhimiya 8:1091±1118, 1984) 2201):118±
142
Gaspar JC, Teixeira NA, Steele IM 01998) Cathodoluminescence
of Juina diamonds. Abstr 7th Int Kimberlite Conf, Cape Town,
pp 242±244
Implications for diamond exploration
Superdeep diamonds may have a much wider distribu-
tion than is currently recognized. If these diamonds are
brought to the surface by very deep-seated processes, it
is not obvious that these processes will occur only be-
neath areas with thick continental roots, the areas tra-
ditionally regarded as prospective for diamond
exploration. Some practical consequences of this sug-
gestion may be important.
1. Diamondiferous kimberlites or related diamondifer-
ous rocks may be found not only within areas with
thick ancient lithosphere, but also in areas of rather
thin lithosphere, underlain by accreted deep material.
This may be a key to exploration for primary dia-
mond deposits in areas of placer deposits with ``un-
known'' primary sources.
2. Superdeep diamonds may be associated with indicator
minerals other than the ``classical'' set. Low-Mgoli-
vine, low-Mghigh-Mn ilmenite, low-Cr spinel and
other minerals should be included in the list of dia-
mond indicator minerals. In such areas, ``classical''
diamond indicator minerals such as pyrope garnet and
high-Cr chromite may be in the minority or absent.
3. The role of high-Cr, low-Ti chromite as the only
spinel associated with diamond, and as a criterion
of diamond grade in kimberlites, should be
reconsidered.
Gaul OF, Grin WL, O'Reilly SY, Pearson NJ 02000) Mapping
olivine composition in the lithospheric mantle. Earth Planet
Sci Lett 182:223±235
Grin WL, Ryan CG, Gurney JJ, Sobolev NV, Win TT 01994)
Chromite macrocrysts in kimberlites and lamproites: geo-
chemistry and origin. In: Meyer HOA, Leonardos OH 0eds)
Kimberlites, related rocks and mantle xenoliths. CPRM Spec
Publ 10A/93):366±377
Grin WL, Moore RO, Ryan CG, Gurney JJ, Win TT 01997)
Geochemistry of magnesian ilmenite megacrysts from southern
African kimberlites. Proc 6th Int Kimberlite Conf, Novosibirsk.
Russian Geol Geophys 38/2:421±443
Acknowledgements This work was ®nancially supported by
Diagem International Resource Corp., the GEMOC Key Centre,
and the Australian Research Council. Prof. S.Y. O'Reilly gave
extremely valuable support in organizing the research. F. Kamin-
sky thanks David Cohen, Paolo Andreazza and other Diagem of-
®cials for hospitality and assistance in the Juina area, Brazil.
Norman Pearson, Carol Larson and Ashwini Sharma provided
valuable assistance with the EMP and LAM-ICPMS analyses at
GEMOC, as did Anita Andrew with the C-isotope analyses at
CSIRO. We thank Ben Harte and Thomas Stachel for thoughtful
and constructive reviews which improved the manuscript. This is

752
Grin WL, Doyle BJ, Ryan CG, Pearson NJ, O'Reilly SY, Davies McCammon C, Hutchison MT, Harris JW 01997) Ferric iron
R, Kivi K, Van Achterbergh E, Natapov LM 01999) Layered
mantle lithosphere in the Lac de Gras area, Slave Craton:
composition, structure and origin. J Petrol 40:705±727
Haggerty SE 01991) Emplacement implications of ultra-deep
xenoliths and diamonds from the transition zone. Abstr 5th
Int Kimberlite Conf, Brazil, pp 157±159
content of mineral inclusions in diamonds from Sao Luiz: a
view into the lower mantle. Science 278:434±436
McDade P, Harris JW 01999) Syngenetic inclusion bearing dia-
monds from Letseng-la-Terai, Lesotho. In: Gurney JJ, Gurney
JL, Pascoe MD, Richardson SH 0eds) Proc 7th Int Kimberlite
Conf, Cape Town. Red Roof Design, Cape Town, pp 557±565
Haggerty SE 01999) Diamond formation and kimberlite-clan Mattey DP 01987) Carbon isotopes in the mantle. Terra Cognita
magmatism in cratonic settings. In: Fei Y, Bertka CM, Mysen 7:31±37
BO 0eds) Mantle petrology: ®eld observations and high pressure Mendelssohn MJ, Milledge HJ 01995) Geologically signi®cant
experimentation: a tribute to Francis R 0Joe) Boyd. Geochem information from routine analysis of the mid-infrared spectra
Soc Spec Publ 6:105±123 of diamonds. Int Geol Rev 37:95±110
Harris JW, Hutchison MT, Hursthouse M, Light M, Harte B Meyer HOA, Svisero DP 01980) Kimberlites and diamonds in
01997) A new tetragonal silicate mineral occurring as inclusions Brazil: windows to the upper mantle. An Acad Brasil Sci
in lower-mantle diamonds. Nature 387:486±488 52:819±825
Harte B 01992) Trace element characteristics of deep-seated Moore RO, Gurney JJ 01989) Mineral inclusions in diamond from
eclogitic paragenesis ± an ion microprobe study of inclusions in
diamonds. Abstr VM Goldschmidt Conf, pp A-48
Harte B, Harris JW 01993) Lower mantle inclusions in diamonds.
Abstr Eur Union Geosci, Strasbourg, pp 101 0review Kerr RA
01993) Science 261:1391)
the Monastery kimberlite, South Africa. In: Ross J, Jaques AL,
Ferguson J, Green DH, O'Reilly SY, Danchin RV, Janse AJA
0eds) Kimberlites and related rocks, vol 2. Proc 4th Int
Kimberlite Conf, Perth, 1986. Geol Soc Aust Spec Publ 14.
Blackwell, Carlton, pp 1029±1041
Harte B, Harris JW 01994) Lower mantle mineral associations Moore RO, Otter ML, Rickard RS, Harris JW, Gurney JJ 01986)
preserved in diamonds. Min Mag58A:384±385
Harte B, Hutchison MT, Harris JW 01994) Trace element char-
acteristics of the lower mantle: an ion microprobe study of in-
The occurrence of moissanite and ferro-periclase as inclusions
in diamonds. In: Ext Abstr 4th Int Kimberlite Conf 0Geol Soc
Aust Abstr 16), Perth, 1986, pp 409±411
clusions in diamonds from Sa
386±387
Harte B, Hutchison MT, Lee M, Harris JW 01998) Inclusions of
0Mg, Fe)O in mantle diamonds. Abstr 7th Int Kimberlite Conf,
Cape Town, pp 308±310
Harte B, Harris JW, Hutchison MT, Watt GR, WildingMC 01999)
Lower mantle mineral associations in diamonds from Sao Luiz,
Brazil. In: Fei Y, Bertka CM, Mysen BO 0eds) Mantle petrol-
Ä
o Luiz, Brazil. Min Mag58A:
Moore RO, Gurney JJ, Grin WL, Shimuzu N 01991) Ultra-high
pressure garnet inclusions in Monastery diamonds: trace
element abundance patterns and conditions of origin. Eur J
Mineral 3:213±230
Norman MD, Pearson NJ, Sharma A, Grin WL 01996) Quanti-
tative analysis of trace elements in geological materials by laster
ablation ICPMS: instrumental operatingconditions and cali-
bration values of NIST glasses. Geostand Newslett 20:247±261
ogy: ®eld observations and high pressure experimentation: a Norman MD, Grin WL, Pearson NJ, Garcia MO, O'Reilly SY
tribute to Francis R 0Joe) Boyd. Geochem Soc Spec Publ 6:125±
153
Hutchison MT, Harte B, Harris JW, Fitzsimmons I 01995) Infer-
ences of the exhumation history of lower mantle inclusions in
01998) Quantitative analysis of trace element abundances in
glasses and minerals: a comparison of laser ablation ICPMS,
solution ICPMS, proton microprobe, and electron microprobe
data. J Anal Atomic Spectrosc 13:477±482
diamonds. Abstr 6th Int Kimberlite Conf, Novosibirsk, pp 242± O'Reilly SY, Chen D, Grin WL, Ryan CG 01997) Minor elements
244
Hutchison MT, Cartigny P, Harris JW 01999) Carbon and nitrogen
in olivine from spinel peridotite xenoliths: implications from
thermobarometry. Min Mag61:257±269
compositions and physical characteristics of transition zone and Otter ML, Gurney JJ 01989) Mineral inclusions in diamonds from
lower mantle diamonds from Sao Luiz, Brazil. In: Gurney JJ,
Gurney JL, Pascoe MD, Richardson SH 0eds) Proc 7th Int
Kimberlite Conf, Vol 2, Cape Town. Red Roof Design, Cape
Town, pp 372±382
Sloan diatremes, Colorado-WyomingState Line kimberlite
district, North America. In: Ross J, Jaques AL, Ferguson J,
Green DH, O'Reilly SY, Danchin RV, Janse AJA 0eds) Kim-
berlites and related rocks, Vol 2. Proc 4th Int Kimberlite Conf,
Perth, 1986. Geol Soc Aust Spec Publ 14, Blackwell, Carlton,
pp 1042±1053
Irifune T 01987) An experimental investigation of the pyroxene-
garnet transformation and its bearing on the constitution of the
mantle. Earth Planet Sci Lett 45:324±336
JoswigW, Stachel T, Harris JW, Baur WH, Brey GP 01999) New
Ca-silicate inclusions in diamonds ± tracers from the lower
mantle. Earth Planet Sci Lett 173:1±6
Rickard RS, Harris JW, Gurney JJ 01989) Mineral inclusions
in diamonds from Kofontein mine. In: Ross J, Jaques AL,
Ferguson J, Green DH, O'Reilly SY, Danchin RV, Janse AJA
0eds) Kimberlites and related rocks, Vol 2, Their mantle crust
setting. Proc 4th Int Kimberlite Conf, Perth, 1986. Geol Soc
Aust Spec Publ 14, Blackwell, Carlton, pp 1054±1062
Kaminsky FV, Zakharchenko OD, Channer DMD, Blinova GK,
Bulanova GP 01997) Diamonds from the Guaniamo area,
Bolivar state, Venezuela. Mem VIII Congr Geologico Ven- Ringwood AE 01989) Constitution and evolution of the mantle.
ezolana. Soc Venezolana Geol, pp 427±430
In: Ross J, Jaques AL, Ferguson J, Green DH, O'Reilly SY,
Danchin RV, Janse AJA 0eds) Kimberlites and related rocks,
Vol 2. Proc 4th Int Kimberlite Conf, Perth, 1986. Geol Soc Aust
Spec Publ 14, pp 457±485
Kesson SE, Fitz Gerald JD 01991) PartitioningMgO, FeO, NiO,
MnO and Cr2O3 between magnesian silicate perovskite and
magnesiowustite: implications for the origin of inclusions in
diamond and the composition of the lower mantle. Earth Planet Ringwood AE, Lovering JF 01970) Signi®cance of the pyroxene-
ilmenite intergrowths among kimberlite xenoliths. Earth Planet
Sci Lett 7:371±375
Sci Lett 111:229±240
Ki¯awi I, Sittas G, Fisher D, Kanda H 01996) The creation of the
3107 cm±1 hydrogen absorption peak in synthetic diamond Scott-Smith BH, Danchin TV, Harris JW, Stracke KJ 01984)
single crystals. Diamond and Related Materials 5:1516±1518
Kopylova MG, Gurney JJ, Daniels LD 01997) Mineral inclusions
in diamonds from the River Ranch kimberlite, Zimbabwe.
Contrib Mineral Petrol 129:366±384
Liu L 01980a) The pyroxene-garnet transformation and its impli-
cation for the 200-km seismic discontinuity. Phys Earth Planet
Int 23:286±291
Kimberlites near Orroroo, South Australia. In: Kornprobst J
0ed) Kimberlites II: the mantle and crust-mantle relationships.
Elsevier, Amsterdam, pp 121±142
Sellschop JPF 01992) Nuclear probes in the study of diamonds. In:
Field J 0ed) The properties of natural and synthetic diamonds.
Academic Press, New York, pp 81±180
Smirnov GI, Kaminsky FV, Klyuyev YA 01986) Some features of
diamond crystals from the Premier kimberlite pipe, South
Africa 0in Russian). Mineralogicheskii J 8:69±74
Liu L 01980b) The mineralogy of an eclogitic earth mantle. Phys
Earth Planet Int 23:262±267

753
Sobolev NV 01971) On the mineralogy and diamond content of Stachel T, Harris JW, Brey GP, JoswigW 02000b) Kankan dia-
kimberlites. Russian Geol Geophys 3:70±79 monds 0Guinea) II: lower mantle inclusion parageneses.
Sobolev NV, Ye®mova ES, Channer DMD, Anderson PFN, Contrib Mineral Petrol 140:16±27
Barron KM 01998) Unusual upper mantle beneath Guaniamo, Sweeney RJ, Winter F 01999) Kimberlite as high-pressure melts: the
Guyana shield, Venezuela: evidence from diamond inclusions.
Geology 26:971±974
Sobolev NV, Ye®mova ES, Koptil VI 01999) Mineral inclusions
in diamonds in the northeast of the Yakutin diamondiferous
determination of segregation depth from major element chem-
istry. In: Gurney JJ, Gurney JL, Pascoe MD, Richardson SH
0eds) Proc 7th Int Kimberlite Conf, Vol 2, Cape Town. Red
Roof Design, Cape Town, pp 846±851
province. In: Gurney JJ, Gurney JL, Pascoe MD, Richardson Taylor WR, Canil D, Milledge HJ 01996) Kinetics of Ib to IaA
nitrogen aggregation in diamonds. Geochim Cosmochim Acta
60:4725±4733
SH 0eds) Proc 7th Int Kimberlite Conf, vol 2, Cape Town.
Red Roof Design, Cape Town, pp 816±822
Stachel T, Harris JW 01997) Syngenetic inclusions in diamond from Vaganov VI, Ilupin IP, Kocherov AI 01999) Chrome spinel as
Birim ®eld 0Ghana) ± a deep peridotitic pro®le with a history of indicators of diamond grade 0in Russian). Rudi Metall 5:15±31
depletion and re-enrichment. Contrib Mineral Petrol 127:336± WildingMC, Harte B, Harris JW 01991) Evidence for a deep origin
352
Stachel T, Harris JW, Brey GP 01999) Rare and unusual mineral
for Sao Luiz diamonds. Abstr 5th Int Kimberlite Conf, Brazil,
pp 456±458
inclusions in diamonds from Mwadui, Tanzania. Contrib Woods GS 01986) Platelets and the infrared absorption of type
Mineral Petrol 132:34±47 Ia diamonds. Proc R Soc Lond A407:219±238
Stachel T, Brey GP, Harris JW 02000a) Kankan diamonds 0Guin- Zedgenizov DA, Logviniva AM, Shatsky VS, Sobolev NV 01998)
ea) I: from lithosphere down to the transition zone. Contrib
Mineral Petrol 140:1±15
Inclusions in microdiamonds from some Yakutian kimberlite
pipes 0in Russian). Dokl Akad Nauk 359/1:74±79