JOURNAL OF PETROLOGY
VOLUME 54
NUMBER 8
PAGES 1555^1583
2013
doi:10.1093/petrology/egt023
Melting and Phase Relations of Carbonated
Eclogite at 9^21GPa and the Petrogenesis of
Alkali-Rich Melts in the Deep Mantle
EKATERINA S. KISEEVA1*, KONSTANTIN D. LITASOV2,3,
GREGORY M. YAXLEY1, EIJI OHTANI4 AND
VADIM S. KAMENETSKY5
1
RESEARCH SCHOOL OF EARTH SCIENCES, AUSTRALIAN NATIONAL UNIVERSITY, CANBERRA, ACT 0200, AUSTRALIA
2
V. S. SOBOLEV INSTITUTE OF GEOLOGY AND MINERALOGY, SIBERIAN BRANCH, RUSSIAN ACADEMY OF SCIENCE,
NOVOSIBIRSK, 630090, RUSSIA
3
NOVOSIBIRSK STATE UNIVERSITY, NOVOSIBIRSK, 630090, RUSSIA
4
DEPARTMENT OF EARTH AND PLANETARY MATERIAL SCIENCE, FACULTY OF SCIENCE, TOHOKU UNIVERSITY,
SENDAI 980-8578, JAPAN
5
ARC CENTRE OF EXCELLENCE IN ORE DEPOSITS AND SCHOOL OF EARTH SCIENCES, UNIVERSITY OF TASMANIA,
HOBART, TAS. 7001, AUSTRALIA
RECEIVED JUNE 15, 2012; ACCEPTED MARCH 15, 2013
ADVANCE ACCESS PUBLICATION MAY 6, 2013
The melting and phase relations of carbonated MORB eclogite have
been investigated using the multi-anvil technique at 9^21 GPa and
1100^19008C. The starting compositions were two synthetic mixes,
GA1 and Volga, with the CO2 component added as CaCO3 (cc):
GA1 þ10%cc (GA1cc) models altered oceanic crust recycled into
the convecting mantle via subduction, and Volga þ 10%cc (Volgacc) models subducted oceanic crust that has lost some of its siliceous
component in the sub-arc regime (GA1 minus 6·5 wt % SiO2). The
subsolidus mineral assemblage at 9 and 13 GPa includes garnet,
clinopyroxene, magnesite, aragonite, a high-pressure polymorph of
TiO2 (only at 9 GPa) and stishovite (only at 13 GPa). At 17^
21 GPa clinopyroxene is no longer stable; the mineral assemblage consists predominantly of garnet with subordinate magnesite (only at
17 GPa), Na-rich aragonite, stishovite, Ca-perovskite (mostly at
21 GPa), and K-hollandite (mostly at 17 GPa). Na-carbonate with
an inferred composition (Na,K)2(Ca,Mg,Fe)(CO3)2 was present
in Volga-cc at 21 GPa and 12008C. Diamond (or graphite) crystallized in most runs in the GA1cc composition, but it was absent in experiments with the Volga-cc composition. In Volga-cc, the solidus
temperatures are nearly constant between 1200 and 13008C over the
entire pressure range investigated. In GA1cc, the solidus is located at
similar temperatures at 9^13 GPa, but at higher temperatures of
1300^15008C at 17^21 GPa. The difference in solidi between the
GA1cc and Volga-cc compositions can be explained by a change in
Na compatibility between 13 and 17 GPa as omphacitic clinopyroxene
disappears, resulting in the formation of Na-carbonate or Na-rich
melt in Volga-cc. The solidus temperature in GA1cc also increases
with increasing pressure as a consequence of carbonate reduction
and diamond precipitation, possibly brought on either via progressive
Fe2þ^Fe3þ transition in garnet at higher pressures or by a decrease
of the activity of the diopside component in clinopyroxene. The lowdegree melts are highly alkalic (K-rich at 9^13 GPa and Na-rich at
17^21 GPa) carbonatites, changing towards SiO2-rich melts with
increasing temperature at constant pressure. The solidi of both compositions remain higher than typical subduction pressure^temperature (P^T) profiles at 5^10 GPa; however, at higher pressures the
flat solidus curve of carbonated eclogite may intersect the subduction
P^T profile in the Transition Zone, where carbonated eclogite can
produce alkali- and carbonate-rich melts. Such subduction-related
alkali-rich melts can be potential analogues of kimberlite and
*Corresponding author. Department of Earth Sciences, South Parks
Road, Oxford, OX13AN, UK; E-mail: kate.kiseeva@earth.ox.ac.uk
ß The Author 2013. Published by Oxford University Press. All
rights reserved. For Permissions, please e-mail: journals.permissions@
oup.com
JOURNAL OF PETROLOGY
VOLUME 54
carbonatite melt compositions and important agents of mantle metasomatism and diamond formation in theTransition Zone and in cratonic roots. Melting of carbonated eclogite produces a garnet-bearing
refractory residue, which could be stored in the Transition Zone or
lower mantle.
KEY WORDS: high-pressure experiments; MORB eclogite; mantle;
Transition Zone; carbonate metasomatism; kimberlite formation;
diamonds
I N T RO D U C T I O N
Subducted slabs of oceanic lithosphere, containing pelagic
sediments and hydrothermally altered basalts (MORB)
formed at a mid-ocean ridge by sea-floor spreading, are
one of the major sources for geochemical heterogeneity in
the mantle, transporting incompatible trace elements and
water. The amount of carbon in the primordial and
modern Earth, and the magnitude of carbon fluxes between the mantle, the crust, the hydrosphere and the atmosphere are highly uncertain (e.g. Zhang & Zindler,
1993; Sleep & Zahnle, 2001; Dasgupta & Hirschmann,
2010). Large amounts of carbon may be introduced into
the mantle by subduction of oceanic crust, which may contain 43 wt % CO2 in its uppermost few hundred metres
(Alt & Teagle, 1999; Staudigel, 2003). Some of the subducted material may undergo partial melting in the subarc regime, releasing the most incompatible and volatile
components back to the surface via arc magmatism, or
later by contributing to MORB, hotspot or continental
magmatism. However, both thermal modelling of subducting slabs and thermodynamic and experimental constraints on slab dehydration and decarbonation
indicate that decomposition of carbonate-bearing species
occurs at much higher depths than that of water-bearing
species, allowing preferential subduction of some slab carbonate relative to hydrous species (e.g. Yaxley & Green,
1994; Bebout, 1995; Poli & Schmidt, 1995; Kerrick &
Connolly, 2001). Consequently, slabs should transport
H2O-poor and carbonate-rich eclogite deep into the
Earth’s mantle.
Most previous experimental studies on carbonated
eclogite have been performed at pressures 10 GPa
(Hammouda, 2003; Dasgupta et al., 2004, 2005; Yaxley &
Brey, 2004; Gerbode & Dasgupta, 2010; Kiseeva et al.,
2012). These studies reported a variety of solidus temperatures and shapes, attributed to compositional differences
in the starting mixes, such as Na2O/CO2, Mg# [molar
Mg/(Mg þ Fe)], Ca# [molar Ca/(Ca þ Mg þ Fe)], the
abundances of alkali components, and minor but variable
amounts of water.
The solidi of carbonated eclogite at higher pressures
from 10 to 32 GPa have been reported only for simplified
chemical systems, such as Na-CMAS þ 5% CO2 by
NUMBER 8
AUGUST 2013
Litasov & Ohtani (2010) and CMAS þ 20% CO2 by
Keshav & Gudfinnsson (2010). A previous study of representative carbonated eclogites at 3·5^5·5 GPa (Kiseeva
et al., 2012) showed the great importance of minor components, especially alkalis. Small amounts of K2O and P2O5
in subducted MORB can significantly decrease its solidus
temperature. The present study is the first to investigate at
P410 GPa a complex natural composition, which includes
the additional and potentially highly influential components FeO and K2O.
The focus of this study is to determine the phase relations (and particularly solidus temperatures) in the deep
upper mantle and Transition Zone (9^21GPa, corresponding to a depth interval of 180^600 km) of carbonated
eclogite, modelling deeply subducted, altered MORB. The
effects of variable alkali and SiO2 contents on solidus temperatures and phase compositions are examined. The results are applied to the stability of different carbonbearing phases in the deep mantle and their roles in
mantle melting and metasomatism and generation of kimberlitic and alkaline magmas.
E X P E R I M E N TA L A N D
A N A LY T I C A L P RO C E D U R E S
Starting composition
Two eclogite compositions (GA1 and Volga) were used as
starting materials (Table 1). The GA1 composition represents altered oceanic basalt (MORB) and is somewhat enriched in alkalis compared with fresh MORB
compositions (Yaxley & Green, 1994). Volga is identical to
GA1, but with 6·5% less SiO2. To both compositions, 10 wt
% of pure CaCO3 (cc) was added, producing GA1 þ10%
CaCO3 (GA1cc) and Volga þ10% CaCO3 (Volga-cc).
The GA1cc composition models subducted, altered, mafic
oceanic crust. Volga-cc models subducted altered mafic
crust, which may have lost a siliceous component during
dehydration and/or silicate melting in the subduction
zone. Altered oceanic crust contains typically no more
than 3 wt % CaCO3. The enhanced carbonate proportions in the current experiments were designed to aid in
the detection of carbonate phases in experimental run
products. Details of the starting material preparation,
phase relations and mineral assemblage at 3·5^5·5 GPa for
GA1cc have been given by Kiseeva et al. (2012). The preparation of the Volga-cc composition was identical to that for
GA1cc.
Experimental techniques
The experiments were conducted using a 3000 ton Kawaitype multianvil apparatus at Tohoku University, Sendai,
Japan. For experiments at 9^13 GPa, the truncated edge
length (TEL) of the tungsten-carbide anvils was 5·0 mm,
and for experiments at 17^21GPa the TEL was 3·5 mm.
1556
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Table 1: Compositions of experimental mixes from this and
other experimental studies
GA1cc Volga-cc Y
P (GPa): 9–21
9–21
W&T O&M
L&O1 K&G
L&O2
3–20
2–27
10–19
18–28
10–32
SiO2
45·32
42·22
49·71
53·53
51·11
50·06
TiO2
1·34
1·43
1·71
1·44
1·76
1·47
Al2O3
14·88
15·91
15·68
14·85
Cr2O3
—
—
14·86
15·39
0·05
—
—
FeOT
8·85
9·46
9·36*
7·92
10·30
MnO
0·15
0·14
0·18
0·16
—
MgO
7·15
7·64
8·43
7·64
7·68
CaO
14·24
14·85
11·73
9·12
Na2O
3·14
3·36
2·76
K2O
0·40
0·42
0·23
P2O5
0·14
0·15
CO2
4·40
4·40
H2O
Total
—
—
100·00 100·00
0·02
9·61
12–25
30·80
50·02
—
—
4·02
16·59
—
—
—
—
—
—
7·59
22·49
14·82
11·23
11·28
21·23
11·49
2·64
2·94
2·43
—
1·31
0·13
0·17
—
—
—
—
—
—
—
—
—
—
—
—
21·46y
—
—
99·81
98·66 102·01 100·00 100·00
2·00
2·00
—
2·08
5·00
—
100·00
*Additional 0·95 wt % Fe2O3 included in the value.
yCO2 measured by difference.
GA1cc and Volga-cc, this study; Y, Yasuda et al. (1994); W&T,
Wang & Takahashi (1999); O&M, Okamoto & Maruyama
(2004); L&O1, Litasov & Ohtani (2005); K&G, Keshav &
Gudfinnsson (2010); L&O2, Litasov & Ohtani (2010).
Each experimental charge contained two capsules with the
GA1cc and Volga-cc compositions (Fig. 1a). The assembly
design was similar to that of Litasov & Ohtani (2009a,
2009b) with two minor modifications: (1) an MgO insulator (instead of BN) was used to separate the capsules from
the LaCrO3 heater; (2) ZrO2 (instead of MgO) was used
as a spacer on top of each capsule to separate it from the
Mo electrodes. It was also used as the pressure transmitting material. The size of the sample chamber before compression was 1·4 and 0·8 mm3 for TEL 5·0 and 3·5 mm,
respectively.
Temperature was monitored with a W97Re3^W75Re25
thermocouple located at the centre of the furnace, between
the two capsules. The temperature gradient in the runs
did not exceed 508C across the sample, according to twopyroxene thermometry determined for special temperature
gradient experiments at 4^6 GPa (Litasov & Ohtani,
2009a). Pressure was calibrated based on in situ synchrotron X-ray diffraction experiments at the ‘SPring-8’ facility
using the gold pressure scale after Dorogokupets &
Dewaele (2007) and Sokolova et al. (2013). Both room-temperature and high-temperature (1200^16008C) data with
durations of greater than 60 min were used for this calibration. The pressure uncertainty was determined to be less
than 1GPa. This calibration was confirmed by laboratory
measurements using semiconductor to metal transitions at
room temperature and at high temperatures (16008C)
using the forsterite ! wadsleyite and wadsleyite !
Fig. 1. Optical (a) and back-scattered electron (BSE) (b^d) images of the experimental runs. (a) Recovered and cut in half experimental charges
G1400-21 and V1400-21 [read as starting composition (G indicates GA1cc,V indicates Volga-cc), temperature,14008C, and pressure, 21 GPa]". (b)
Run G1300-21with large garnet, diamond and carbonate (in the top side of the capsule) crystals surrounded by a fine-grained matrix of similar composition material. No melt is observed. (c) RunV1300-21 with Grt crystals surrounded by melt pools. (d) Run G1300-21. Run G1300-21. Magnified
view of part of the run shown in (b). Large garnet and stishovite crystals surrounded by a fine-grained matrix of similar material.
1557
JOURNAL OF PETROLOGY
VOLUME 54
ringwoodite transitions in Mg2SiO4 (Litasov & Ohtani,
2009a, 2009b).
Au75Pd25 capsules were used as sample containers,
which were found to be the best material to avoid a hydrogen flux into and out of the capsule during the experiments
(Nishihara et al., 2006). Sample parts were fired in the
oven at 8508C and pyrophyllite gaskets were heated at
2308C for several hours prior to the experiment.
Encapsulated starting mixtures were dried in the oven at
3008C for 1h before final sealing by arc welding. These
procedures minimized penetration of hydrogen into the
sample chamber during the experiments.
Experiments were conducted at 9, 13, 17 and 21GPa,
over a range of temperatures between 1100 and 18008C.
After recovery, the Au^Pd capsule was cut in two using
a 0·15 mm thick diamond saw and petroleum benzene cutting fluid to preserve water-soluble phases. One half was
then mounted into epoxy resin and rough polished on
abrasive paper under petroleum benzene (Fig. 1a). The
samples were then reimpregnated with epoxy resin under
vacuum, followed by final polishing with oil-based diamond paste.
Analytical techniques
Run products were analysed using both wavelength- and
energy-dispersive (WDS and EDS) spectroscopy. All
phases were analysed using a JEOL 6400 scanning electron
microscope fitted with an energy-dispersive detector at
the Centre for Advanced Microscopy, ANU. Spectra were
acquired using a 15 kVaccelerating voltage, 1 nA beam current, and an acquisition time of 120 s. Garnet and clinopyroxene were also analysed on a Cameca SX100 at the
University of Tasmania, using a beam current of 30 nA
and accelerating voltage of 15 kV.
The compositions of garnets and pyroxenes measured
using both EDS and WDS differ by less than 5%, consistent with the work of Spandler et al. (2010), who compared
multiple EDS and WDS analyses (using the electron
microprobe at James Cook University) obtained from the
same phases in experimental run products. The reported
values of garnet and clinipyroxene are averages of both
WDS and EDS analyses.
For crystalline phases, a 1 mm beam with an excitation
diameter of about 1·5 mm was used. To obtain the most
representative composition, at least 10 grains of each
phase were analyzed in each experiment, and only those
analyses close to the theoretical cation sum were accepted.
For the majority of melt analyses a larger area scan was
used. Most of the quenched melts present in the runs
were highly heterogeneous, so as many area scans as possible were performed on each melt-bearing run.
Detection limits were 0·1^0·2 wt %. Analyses were obtained for most melts and mineral phases; however, it
was not always possible to precisely analyse some very
fine-grained accessory phases and the extremely
NUMBER 8
AUGUST 2013
heterogeneous melt patches present in some runs. Massbalance calculations were carried out for each of the
experiments.
Raman spectroscopy was used for identification of
carbon allotropes (graphite or diamond) and carbonates.
The Raman spectra were obtained using a Jasco NRS2000 microspectrometer at Tohoku University. A microscope was used to focus the excitation laser beam (the
488 nm lines of a Princeton Instruments Ar þ laser) on
the sample surface. Spectra were collected for 120^240 s,
using a laser operating at 12^20 mW and a beam 1 mm in
diameter.
R E S U LT S
A summary of all run conditions and calculated phase proportions for GA1cc and Volga-cc are given in Table 2.
Representative phase compositions are listed in
Tables 3^8. The observed phase assemblages were used to
construct an experimental P^T phase diagram (Fig. 2).
Most runs produced well-crystallized, chemically homogeneous mineral assemblages from the glass starting material. The homogeneity of phase compositions in most runs
indicates a close approach to chemical equilibrium.
Evidence for disequilibrium was observed in some of the
lowest temperature runs, with incorrect garnet stoichiometry, manifested by cation sums (12 oxygens per formula
unit) substantially less than eight.
Solidus position, phase assemblage and
types of melt
The solidus temperature at a given pressure was bracketed
based on the presence of visible quenched melt products
and mass-balance calculations. In the case of the lowest degrees of melting an additional compositional criterion is
useful to distinguish stable mineral and metastable
quenched phases. Across the entire range of the experiments, the quenched melt products are a mixture of carbonate and silicate components, whereas the solid
carbonates are homogeneous and free from silicate
components.
The estimated solidus is between 1250 and 13008C at
9 GPa and between 1200 and 13008C at 13 GPa for both
starting compositions (Fig. 2). Solidus temperatures are
estimated to lie between 1200 and 13008C at 17 and 21GPa
for the Volga-cc composition and between 1300 and
14008C at 17 GPa for the GA1cc composition. Experiments
G1300-21 and G1400-21 [read as starting composition (G
indicates GA1cc), temperature, 14008C, and pressure, 21
GPa] exhibit unusual textures (Fig. 1b and d), distinguished by garnet, diamond and to a lesser extent large
crystals of stishovite and K-hollandite (high-pressure analogue of KAlSi3O8) (10^20 mm) surrounded by a finegrained (1 mm) matrix of the same minerals. In this case,
1558
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Table 2: Experimental results and run conditions
T (8C)
dT
P
D (h)
Exp. no.
Phases present (GA1cc)
Exp. no.
Phases present (Volga-cc)
1050
50
9
72
G1050-9
Grt(42), Cpx(45), Arag(8), Mst(1·5), Co(3·5), TiO2
V1050-9
Grt(26), Cpx(62), Arag(6·5), Mst(3), Co(2), TiO2
1200
20
9
48
G1200-9
Grt(42), Cpx(45), Arag(5), Mst(4), Co(3·5), TiO2
V1200-9
Grt(49), Cpx(39), Arag(3), Mst(7), Co(1·5), TiO2
1250
60
9
48
G1250-9
Grt(48), Cpx(35), CMss(10), Co(6·5), TiO2
V1250-9
Grt(45), Cpx(44), CMss(10), Co, TiO2
1400
20
9
24
G1400-9
Grt(46), Cpx(40), Co(4), TiO2, LCarb(10)
V1400-9
Grt(54), Cpx(35), Co, TiO2, LCarb(10·5)
1200
20
13
80
G1200-13
Grt(84), Cpx(4), Arag(5·5), Mst(4), St(2·5)
V1200-13
Grt(83·5), Cpx(5), Arag(8), Mst(2), St(1), K-Holl
1300
20
13
48
G1300-13
Grt(70), Cpx(12), St(6), LSi-Carb(12)
V1300-13a
Grt(78), Cpx(8), St(2), LCarb(11·5)
1300
20
13
48
V1300-13b
Grt(77), Cpx(11), St(2), LCarb(10·5)
1400
20
13
24
G1400-13
Grt(69), Cpx(13), St(6), LSi-Carb(12)
V1400-13
Grt(73), Cpx(15), St(1), LCarb(11)
1550
70
13
12
G1550-13
Grt(59), Cpx(19), St(8), LSi-Carb(13)
V1550-13
Grt(68), Cpx(16), St(4), LCarb(12)
1100
40
17
48
G1100-17
Grt(89), Arag(9), Mst(1), K-Holl(1), St, CPv
V1100-17
Grt(88), Arag(9), Mst(1), St(1), K-Holl, CPv
1200
20
17
48
G1200-17
Grt(87·5), Arag(10), Mst(1), K-Holl(1), St, CPv
V1200-17
Grt(85), Arag(10), St(3), K-Holl(2), Mst, CPv
1250
70
17
48
G1250-17
Grt(81), Arag(9·5), St(9), Mst, K-Holl
V1250-17
Grt(83), Arag(3·5), St(5·5), K-Holl, Mst, LSi-Carb(7·5)
1400
20
17
24
G1400-17
Grt(74), St(10·5), LSi-Carb(15·5)
V1400-17
Grt(81), St(3), LSi-Carb(16)
1500
70
17
12
G1500-17
Grt(73·5), St(8·5), LSi-Carb(18)
V1500-17
Grt(81), St(2·5), LSi-Carb(17)
1200
20
21
48
G1200-21
Grt(89·5), Arag(9·5), St, K-Holl, CPv
V1200-21
Grt(84·5), Arag(7·5), St(3), K-Holl(1), CPv(1·5),
1300
20
21
48
G1300-21
Grt(79), Arag(9), St(10), K-Holl(2), CPv
V1300-21
Grt(80), St(8), CPv(1·5), LCarb(10·5)
Na-Carb(2·5)
1400
20
21
16
G1400-21
Grt(78), Arag(9·5), St(12·5), CPv
V1400-21
Grt(78·5), St(10), CPv, LCarb(11)
1650
80
21
16
G1650-21
Grt(79·5), St(8), LSi-Carb(12·5)
V1650-21
Grt(82), St(6), CPv, LCarb(12)
1900
100
21
12
G1900-21
Grt(80), St(7), LSi-Carb(14)
V1900-21
Grt(83), St(4), LCarb(13)
dT, estimated temperature gradient. D (h), duration of the experiment in hours. We did not identify the structure of the
TiO2 phase in the experiments and followed results by Withers et al. (2003) and Sato et al. (1991) for rutile or TiO2 II and
further phase transitions. Numbers in parentheses are wt % of the phase, extracted from mass-balance calculations. For
phases with no wt % value included, it is considered to be 51 wt %. Grt, garnet; Cpx, clinopyroxene; Arag, aragonite;
Mst, magnesite; Co, coesite; St, stishovite; CMss, calcite–magnesite solid solution; K-Holl, K-hollandite; CPv,
Ca-perovskite; Na-Carb, Na-carbonate; LCarb, carbonate melt; LSi-Carb, silicate–carbonate melt.
large grains of Na-rich aragonite are usually found segregated to the edge of the capsule, although it is not possible
to rule out their presence within the matrix. These runs
are considered to be subsolidus and hence the GA1cc solidus at 21GPa is located at a slightly higher temperature
than 14008C. However, there is a possibility that a small,
undetected melt fraction is present within the fine-grained
matrix.
The subsolidus phase assemblages for both starting materials at 9^13 GPa consist of garnet, clinopyroxene, carbonates (aragonite and magnesite, or calcite^magnesite
solid solution), a high-pressure polymorph of TiO2, coesite
or stishovite, graphite or diamond (only in GA1cc runs)
and K-hollandite (only in the V1200-13 run). Na-rich aragonite appeared in the V1300-13 run, whereas no alkalirich carbonates were observed in GA1cc runs at 9^13 GPa.
At 17^21GPa, the phase assemblages consist of garnet,
stishovite, K-hollandite, magnesite (only at 17 GPa), Narich aragonite and Ca-perovskite. Carbonate with a high
Na content (around 20 wt % Na2O) was detected in a subsolidus V1200-21 run.
Phase relations and compositions
Major phases
Garnet is the major phase in all the experiments
(Fig. 3a^f). Its modal proportion increases from 40% at
9 GPa (except V1050-9, which crystallized only 26%
garnet) to 70% at 13 GPa and 80% at 17^21GPa. The
grain size differs significantly and varies from 55 mm in
low-temperature 9 GPa runs to 40 mm in 17^21GPa runs.
In most experiments at 9 and 13 GPa, garnet occurs as
well-shaped, equant grains, often containing inclusions of
clinopyroxene and coesite or stishovite (Fig. 3a^c). At 17
and 21GPa, large fractured crystals of garnet occupy most
of the experimental charge (Figs 1c and 3d^f), with accessory phases (usually stishovite and K-hollandite) as inclusions and intergranular carbonate or melt. At all
pressures, with increasing temperature the number of inclusions in garnet decreases and the grains become larger
and more compositionally homogeneous.
As in previous studies (Yasuda et al., 1994; Litasov &
Ohtani, 2010), high-pressure garnet is generally characterized by Si in excess of 3·00 cations per 12-oxygen formula
1559
JOURNAL OF PETROLOGY
VOLUME 54
NUMBER 8
AUGUST 2013
Table 3: Compositions of experimental garnet
GA1cc
T (8C):
P (GPa):
1200
9
n:
4
1250
9
s
SiO2
38·66
0·86
1·49
0·16
TiO2
21·29
0·72
Al2O3
FeO
9·68
1·18
MnO
0·14
0·17
MgO
7·91
0·53
0·26
CaO
16·82
0·71
0·26
Na2O
0·13
0·01
K2O
0·41
0·09
P2O5
Total
97·23
1·70
Mg#
59·32
3·87
Atoms per 12-oxygen formula unit
Si
2·96
Ti
0·09
Al
1·92
Fe
0·62
Mn
0·01
Mg
0·90
Ca
1·38
Na
0·10
K
0·01
P
0·03
Total
8·02
1400
9
1300
13
1400
13
1550
13
5
s
4
s
8
s
6
s
7
s
42·67
1·25
19·36
11·14
0·17
8·14
13·85
1·69
0·20
0·23
98·70
56·61
1·02
0·14
0·76
0·88
0·08
0·28
0·26
0·59
0·09
0·09
1·01
1·28
41·27
1·77
19·53
10·57
0·22
8·18
16·33
0·80
0·14
0·18
99·00
57·98
0·42
0·08
0·13
0·08
0·01
0·07
0·26
0·13
0·04
0·04
0·34
0·33
43·00
1·99
19·01
10·29
0·20
8·67
14·79
1·76
b.d.l.
0·25
99·95
60·01
0·76
0·13
0·44
0·25
0·10
0·25
0·32
0·09
—
0·05
1·36
0·63
43·27
1·97
19·85
10·19
0·18
8·76
15·09
1·64
b.d.l.
0·17
101·12
60·49
0·37
0·16
0·58
0·06
0·01
0·17
0·24
0·11
—
0·03
0·71
0·36
42·31
1·71
19·25
10·05
0·20
8·68
14·61
1·63
b.d.l.
0·16
98·60
60·62
0·40
0·11
0·50
0·23
0·12
0·05
0·22
0·14
—
0·06
0·64
0·61
3·20
0·07
1·71
0·70
0·01
0·91
1·11
0·25
0·02
0·01
7·99
3·10
0·10
1·73
0·66
0·01
0·92
1·32
0·12
0·01
0·01
7·98
3·18
0·11
1·66
0·64
0·01
0·95
1·17
0·25
—
0·02
7·99
3·16
0·11
1·71
0·62
0·01
0·95
1·18
0·23
—
0·01
7·98
3·17
0·10
1·70
0·63
0·01
0·97
1·17
0·24
—
0·01
7·99
1400
21
1650
21
GA1cc
T (8C):
P (GPa):
1250
17
n:
6
1400
17
s
SiO2
44·32
1·16
1·60
0·04
TiO2
Al2O3
18·43
0·58
FeO
9·61
0·32
MnO
0·15
0·12
MgO
8·29
0·31
0·25
CaO
13·45
2·40
0·11
Na2O
0·10
0·05
K2O
0·23
0·05
P2O5
Total
98·59
2·06
Mg#
60·59
0·80
Atoms per 12-oxygen formula unit
Si
3·29
Ti
0·09
Al
1·62
Fe
0·60
Mn
0·01
Mg
0·92
Ca
1·07
Na
0·35
K
0·01
P
0·01
Total
7·97
1500
17
1300
21
5
s
6
s
5
s
7
s
5
s
43·20
1·73
20·26
9·46
0·16
8·88
14·82
1·97
b.d.l.
0·22
100·70
62·57
0·34
0·06
0·22
0·06
0·01
0·06
0·18
0·09
—
0·03
0·44
0·29
42·93
1·69
19·45
9·01
0·18
8·48
14·45
2·10
b.d.l.
0·28
98·56
62·65
0·54
0·14
0·29
0·17
0·05
0·14
0·19
0·11
—
0·11
0·44
0·75
43·32
1·54
19·28
8·92
0·20
8·55
14·36
2·18
b.d.l.
0·23
98·58
63·07
0·49
0·07
0·24
0·23
0·09
0·10
0·19
0·04
—
0·07
1·06
0·45
42·60
1·46
19·70
9·69
0·16
8·42
14·73
1·92
b.d.l.
0·31
98·99
60·78
1·32
0·23
1·28
0·52
0·07
0·37
1·08
0·62
—
0·06
1·11
1·15
44·09
1·60
20·16
9·18
0·16
8·84
15·00
2·01
b.d.l.
0·19
101·21
63·17
0·32
0·03
0·08
0·09
0·01
0·09
0·06
0·08
—
0·01
0·27
0·31
3·15
0·10
1·74
0·58
0·01
0·97
1·16
0·28
—
0·01
8·00
3·19
0·09
1·71
0·56
0·01
0·94
1·15
0·30
—
0·02
7·99
3·22
0·09
1·69
0·55
0·01
0·95
1·14
0·31
—
0·01
7·99
3·17
0·08
1·73
0·60
0·01
0·93
1·17
0·28
—
0·02
8·00
3·19
0·09
1·72
0·56
0·01
0·95
1·16
0·28
—
0·01
7·98
(continued)
1560
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Table 3: Continued
Volga-cc
T (8C):
P (GPa):
1050
9
n:
3
V1300-13a
1200
9
s
SiO2
39·51
0·41
1·07
0·32
TiO2
20·29
0·71
Al2O3
FeO
12·90
0·58
MnO
0·38
0·06
MgO
6·75
0·67
0·48
CaO
16·60
0·73
0·10
Na2O
0·14
0·05
K2O
0·36
0·16
P2O5
Total
98·73
0·82
Mg#
48·16
1·38
Atoms per 12-oxygen formula unit
Si
3·02
Ti
0·06
Al
1·83
Fe
0·82
Mn
0·02
Mg
0·77
Ca
1·36
Na
0·11
K
0·01
P
0·02
Total
8·03
1250
9
1400
9
1200
13
1300
13
6
s
4
s
5
s
2
s
7
s
41·42
1·70
20·02
12·58
0·33
6·85
16·99
1·11
0·10
0·19
101·29
49·24
0·44
0·09
0·21
0·30
0·02
0·23
0·34
0·16
0·04
0·02
0·31
0·51
40·41
1·27
21·12
14·60
0·38
8·62
13·72
0·59
b.d.l.
0·15
100·86
51·26
0·40
0·06
0·25
0·30
0·01
0·26
0·21
0·16
—
0·04
0·09
0·57
41·60
1·48
19·96
12·51
0·32
8·65
13·57
1·02
b.d.l.
b.d.l.
99·10
55·20
0·37
0·17
0·29
0·25
0·01
0·25
0·15
0·24
—
—
0·66
0·51
45·05
1·71
17·59
11·11
0·23
7·73
13·05
2·61
0·22
0·18
99·47
55·28
0·99
0·10
0·30
0·08
0·04
0·74
1·46
0·12
0·03
0·11
0·04
2·19
43·35
2·00
18·89
10·50
0·20
8·64
14·63
1·79
b.d.l.
0·21
100·21
59·45
0·43
0·13
0·60
0·29
0·04
0·31
0·61
0·06
—
0·05
1·22
1·32
3·08
0·10
1·75
0·78
0·02
0·76
1·35
0·16
0·01
0·01
8·02
3·01
0·07
1·85
0·91
0·02
0·96
1·10
0·09
—
0·01
8·02
3·12
0·08
1·76
0·78
0·02
0·97
1·09
0·15
—
—
7·99
3·34
0·10
1·54
0·69
0·01
0·85
1·04
0·37
0·02
0·01
7·98
3·20
0·11
1·64
0·65
0·01
0·95
1·15
0·26
—
0·01
7·98
1200
17
1250
17
V1300-13b
T (8C):
P (GPa):
1300
13
n:
2
1400
13
s
SiO2
42·81
0·64
2·02
0·07
TiO2
Al2O3
18·50
0·15
FeO
10·92
0·22
MnO
0·35
0·12
MgO
8·16
0·22
0·22
CaO
14·21
1·70
0·02
Na2O
b.d.l.
—
K2O
0·11
0·10
P2O5
Total
98·78
0·92
Mg#
57·09
1·17
Atoms per 12-oxygen formula unit
Si
3·21
Ti
0·11
Al
1·63
Fe
0·68
Mn
0·02
Mg
0·91
Ca
1·14
Na
0·25
K
—
P
0·01
Total
7·97
1550
13
1100
17
5
s
5
s
4
s
5
s
9
s
43·87
1·92
19·11
11·45
0·28
8·60
14·48
1·70
b.d.l.
0·13
101·53
57·21
0·56
0·07
0·43
0·26
0·00
0·49
0·69
0·18
—
0·03
0·18
0·87
42·46
1·69
19·63
11·19
0·31
8·64
14·31
1·56
b.d.l.
0·11
99·90
57·89
0·99
0·08
0·53
0·22
0·04
0·12
0·55
0·17
—
0·01
1·86
0·32
45·05
1·56
17·49
10·63
0·25
8·31
11·43
3·08
0·29
0·19
98·27
58·19
0·80
0·10
0·29
0·31
0·04
0·48
1·43
0·51
0·14
0·02
0·68
1·38
44·32
1·84
18·91
11·15
0·26
8·55
13·33
2·27
0·12
0·22
100·95
57·71
0·93
0·38
0·29
0·18
0·01
0·41
1·20
0·32
0·06
0·03
0·21
0·80
43·09
1·79
18·83
10·86
0·27
8·02
14·24
1·96
b.d.l.
0·21
99·26
56·81
0·98
0·09
0·39
0·14
0·07
0·22
0·27
0·08
—
0·04
0·69
0·59
3·20
0·11
1·64
0·70
0·02
0·94
1·13
0·24
—
0·01
7·98
3·15
0·09
1·72
0·69
0·02
0·95
1·14
0·22
—
0·01
8·00
3·36
0·09
1·54
0·66
0·02
0·92
0·91
0·45
0·03
0·01
8·00
3·24
0·10
1·63
0·68
0·02
0·93
1·04
0·32
0·01
0·01
7·99
3·21
0·10
1·65
0·68
0·02
0·89
1·14
0·28
—
0·01
7·99
(continued)
1561
JOURNAL OF PETROLOGY
VOLUME 54
NUMBER 8
AUGUST 2013
Table 3: Continued
Volga-cc
T (8C):
P (GPa):
1400
17
n:
6
1500
17
s
5
SiO2
42·78
0·64
42·89
1·88
0·14
1·69
TiO2
19·56
0·48
19·28
Al2O3
FeO
10·07
0·42
10·19
MnO
0·20
0·07
0·25
MgO
8·00
0·29
8·32
CaO
15·12
0·58
13·96
1·87
0·16
2·02
Na2O
b.d.l.
—
0·33
K2O
0·18
0·04
0·16
P2O5
Total
99·66
0·95
99·07
Mg#
58·60
0·99
59·27
Atoms per 12-oxygen formula unit
Si
3·17
3·20
Ti
0·10
0·09
Al
1·71
1·69
Fe
0·62
0·64
Mn
0·01
0·02
Mg
0·88
0·92
Ca
1·20
1·11
Na
0·27
0·29
K
—
0·03
P
0·01
0·01
Total
7·99
8·01
1200
21
1300
21
1400
21
1650
21
1900
21
s
6
s
5
s
7
s
4
s
5
s
0·54
0·05
0·25
0·15
0·01
0·15
0·23
0·09
0·21
0·02
0·55
0·35
44·88
1·23
18·02
11·41
0·25
8·48
12·70
2·29
0·21
0·19
99·65
56·94
0·74
0·15
0·75
0·16
0·11
0·37
0·67
0·28
0·07
0·07
0·97
1·20
42·76
1·44
20·50
11·01
0·27
8·59
14·78
1·55
b.d.l.
0·21
101·10
58·16
0·80
0·14
0·50
0·16
0·01
0·23
0·64
0·15
—
0·03
0·16
0·36
42·24
1·46
20·43
10·96
0·23
8·35
14·77
1·52
b.d.l.
0·25
100·20
57·58
0·41
0·03
0·50
0·22
0·06
0·36
0·38
0·09
—
0·07
1·20
0·60
43·56
1·68
19·39
10·53
0·26
8·36
14·64
2·03
b.d.l.
0·15
100·60
58·58
0·18
0·05
0·59
0·38
0·04
0·17
0·65
0·19
—
0·01
1·36
0·55
44·34
1·29
19·51
8·47
0·25
8·90
14·80
2·25
b.d.l.
0·15
99·96
65·21
0·41
0·04
0·32
0·41
0·01
0·10
0·34
0·08
—
0·03
0·23
1·32
3·32
0·07
1·57
0·71
0·02
0·93
1·01
0·33
0·02
0·01
7·98
3·13
0·08
1·77
0·67
0·02
0·94
1·16
0·22
—
0·01
8·00
3·12
0·08
1·78
0·68
0·01
0·92
1·17
0·22
—
0·02
8·00
3·20
0·09
1·68
0·65
0·02
0·91
1·15
0·29
—
0·01
8·00
3·24
0·07
1·68
0·52
0·02
0·97
1·16
0·32
—
0·01
7·99
b.d.l., below the detection limit (taken as 0·1 for all EDS measured values).
unit (p.f.u.), and significant amounts of Na (0·09^
0·45 p.f.u.) and low Al contents (1·54^1·92 p.f.u.). Within
the 9^13 GPa pressure interval the amount of Si p.f.u. in
garnet sharply increases, from 3·00^3·05 in some of the
9 GPa runs to 3·15^3·20 Si p.f.u. in most of the 13 GPa
runs (Fig. 4b), indicating significant increase in the majorite component in garnet. The strong positive correlation
between Si and Na is consistent with the Na-majorite
(Na2MgSi5O12) substitution (e.g. Presnall et al., 1978;
Dymshits et al., 2013). Garnet in subsolidus experiments
contains higher Si and Na contents than in partially
molten experiments. The Na content of garnet slightly decreases with increasing temperature and degree of melting.
The amount of Na in garnet generally increases with pressure, but the rate of increase becomes lower as pressure increases (Figs 4a, b and 5a).
Clinopyroxene is the dominant phase in the GA1cc
composition at 3·5^5·5 GPa (see Kiseeva et al., 2012).
The amount of clinopyroxene decreases from 45^55 at
5 GPa to 35^40% at 9 GPa (except V1050-9 run with estimated 62 modal % of clinopyroxene). At 13 GPa the amount
of clinopyroxene drops steeply and reaches 4^19%
(Fig. 4a). The modal proportion of clinopyroxene increases
with increasing temperature.The clinopyroxene-out boundary in both compositions lies just above 13 GPa, because in
subsolidus runs at 13 GPa only a few clinopyroxene grains
were observed. Further evidence for this is provided by the
garnet composition. The amount of the majorite component
ingarnet, manifestedby Sicationsp.f.u., doesnot increase significantly over the pressure interval between13 and 21GPa.
Clinopyroxene is Na-rich and with increasing pressure
the amount of Na increases from around 0·29^0·50 Na
p.f.u. (4·3^8·0 wt % Na2O) at 9 GPa to 0·59^0·80 Na p.f.u.
(8·7^12·2 wt % Na2O) at 13 GPa. It does not show any significant correlation with temperature, and does not
change significantly across the solidus (Fig. 5b). However,
the Na content of clinopyroxene correlates with Na in
garnet (Fig. 4a). The amount of M2þ cations (Mg, Fe, Ca,
Mn) in clinopyroxene decreases with increasing
pressure (Fig. 6a), whereas the amount of Al does not
exceed 0·7^0·8 cations p.f.u. (Fig. 6b). Excess of Si (up
to 0·05 Si cations p.f.u. over 2·00) in clinopyroxene
1562
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Table 4: Compositions of experimental clinopyroxene
GA1cc
T (8C):
P (GPa):
1050
9
n:
11
1200
9
s
SiO2
50·95
0·33
1·43
0·12
TiO2
17·23
0·21
Al2O3
FeO
8·03
0·56
MnO
0·13
0·01
MgO
7·68
0·31
CaO
10·23
0·54
4·28
0·34
Na2O
0·55
0·04
K2O
0·22
0·02
P2O5
Total
100·73
0·63
Mg#
63·05
1·32
Atoms per 6-oxygen formula unit
Si
1·81
Ti
0·04
Al
0·72
Fe
0·24
Mn
0·00
Mg
0·41
Ca
0·39
Na
0·29
K
0·02
P
0·01
Total
3·94
1250
9
1400
13
n:
6
1200
13
1300
13
3
s
10
s
8
s
2
s
7
s
55·20
0·78
16·95
3·65
b.d.l.
6·20
9·55
8·02
0·43
b.d.l.
100·80
75·35
1·45
0·16
0·48
0·90
—
0·11
0·62
0·61
0·05
—
0·09
4·17
53·50
0·92
16·15
4·67
b.d.l.
6·90
10·98
6·38
0·44
0·14
100·08
72·54
1·12
0·10
0·38
0·50
—
0·17
0·44
0·34
0·07
0·04
0·57
1·76
54·58
0·80
14·16
3·71
b.d.l.
7·42
11·65
6·57
0·44
b.d.l.
99·32
78·13
0·58
0·08
0·74
0·31
—
0·25
0·29
0·23
0·03
—
0·45
1·02
55·14
0·81
17·36
3·64
b.d.l.
4·78
7·67
8·65
0·35
b.d.l.
98·39
70·05
0·32
0·00
0·06
0·06
—
0·17
0·32
0·25
0·06
—
0·30
1·07
57·28
0·36
17·55
2·34
b.d.l.
4·67
6·88
10·52
0·18
b.d.l.
99·77
78·06
1·13
0·01
0·48
0·19
—
0·14
0·44
0·33
0·05
—
1·03
1·01
1·92
0·02
0·70
0·11
—
0·32
0·36
0·54
0·02
—
3·99
1·89
0·02
0·67
0·14
—
0·36
0·42
0·44
0·02
0·00
3·97
GA1cc
T (8C):
P (GPa):
1400
9
1·94
0·02
0·59
0·11
—
0·39
0·44
0·45
0·02
—
3·98
1·95
0·02
0·73
0·11
—
0·25
0·29
0·59
0·02
—
3·96
1·99
0·01
0·72
0·07
—
0·24
0·26
0·71
0·01
—
4·00
Volga-cc
1550
13
s
SiO2
58·99
1·45
0·40
0·02
TiO2
Al2O3
17·17
0·83
FeO
2·36
0·11
MnO
b.d.l.
—
MgO
4·99
0·10
CaO
7·03
0·50
10·21
0·31
Na2O
0·13
0·06
K2O
b.d.l.
—
P2O5
Total
101·27
1·24
Mg#
79·04
0·59
Atoms per 6-oxygen formula unit
Si
2·01
Ti
0·01
Al
0·69
Fe
0·07
Mn
—
Mg
0·25
Ca
0·26
Na
0·68
K
0·01
P
—
Total
3·97
1050
9
1200
9
1250
9
1400
9
6
s
8
s
8
s
8
s
9
s
56·66
0·50
16·46
2·49
b.d.l.
5·42
7·42
9·74
0·11
b.d.l.
98·80
79·50
1·23
0·11
0·46
0·10
—
0·33
0·37
0·29
0·04
—
1·21
0·43
47·49
1·36
17·62
9·40
0·22
7·58
11·43
4·31
0·58
0·18
100·16
58·98
0·36
0·16
0·39
0·36
0·01
0·28
0·61
0·23
0·03
0·03
0·52
0·59
52·19
0·75
16·08
6·04
0·11
6·17
11·26
6·94
0·55
0·13
100·21
64·75
2·10
0·11
0·44
1·02
0·04
0·12
1·17
0·79
0·09
0·02
0·61
3·51
54·67
0·73
14·77
4·67
b.d.l.
7·02
11·19
6·90
0·31
b.d.l.
100·26
72·83
1·65
0·32
1·04
0·46
—
0·35
0·13
0·36
0·15
—
0·64
2·29
55·08
0·59
14·77
4·48
b.d.l.
6·49
9·85
7·21
0·27
b.d.l.
98·74
72·09
0·97
0·07
0·33
0·32
—
0·18
0·28
0·30
0·03
—
0·57
1·24
1·99
0·01
0·68
0·07
—
0·28
0·28
0·66
0·01
—
3·99
1·73
0·04
0·76
0·29
0·01
0·41
0·45
0·30
0·03
0·01
4·01
1·87
0·02
0·68
0·18
0·00
0·33
0·43
0·48
0·02
0·00
4·02
1·93
0·02
0·61
0·14
—
0·37
0·42
0·47
0·01
—
3·98
1·96
0·02
0·62
0·13
—
0·34
0·38
0·50
0·01
—
3·97
(continued)
1563
JOURNAL OF PETROLOGY
VOLUME 54
NUMBER 8
AUGUST 2013
Table 4: Continued
Volga-cc
V1300-13a
V1300-13b
T (8C):
1200
1300
1300
1400
1550
P (GPa):
13
13
13
13
13
n:
5
SiO2
58·27
0·73
56·69
1·77
56·90
0·75
59·45
0·79
58·52
TiO2
0·21
0·03
0·39
0·07
0·32
0·03
0·39
0·02
0·37
0·02
Al2O3
17·69
0·54
16·29
1·29
16·76
0·31
16·93
0·45
17·30
0·48
s
0·35
s
9
4·74
MgO
2·64
0·23
4·43
0·28
3·98
0·07
4·91
0·17
5·02
0·36
CaO
3·98
0·33
6·73
1·08
6·13
0·54
6·78
0·25
6·80
0·51
Na2O
12·23
0·13
10·50
0·49
10·45
0·30
10·01
0·33
10·35
0·33
0·24
0·28
0·11
K2O
b.d.l.
—
0·25
P2O5
b.d.l.
—
b.d.l.
—
0·09
b.d.l.
—
b.d.l.
—
3·02
b.d.l.
0·06
—
3·02
0·74
b.d.l.
—
3·07
s
8
MnO
b.d.l.
0·24
s
8
FeO
—
3·33
s
2
b.d.l.
b.d.l.
—
0·11
b.d.l.
—
b.d.l.
0·17
—
0·01
—
Total
99·76
0·69
98·61
2·45
97·87
0·82
101·48
0·36
101·49
0·31
Mg#
49·75
1·01
70·34
1·37
69·77
0·24
74·34
0·69
74·74
0·52
Atoms per 6-oxygen formula unit
Si
2·03
2·00
2·02
2·02
2·00
Ti
0·01
0·01
0·01
0·01
0·01
Al
0·73
0·68
0·70
0·68
0·70
Fe
0·14
0·10
0·09
0·09
Mn
—
—
—
—
0·09
—
Mg
0·14
0·23
0·21
0·25
0·26
Ca
0·15
0·26
0·23
0·25
0·25
Na
0·83
0·72
0·72
0·66
0·69
K
0·00
0·01
0·01
0·00
P
Total
—
4·02
—
4·01
—
—
3·99
3·96
from this and other studies at 13^19 GPa (Fig. 6b) can be
potentially explained by the additional clinopyroxene
component NaMg0·5Si0·5Si2O6, synthesized by Gasparik
(1988).
Minor and accessory phases
Crystalline carbonate (magnesite, aragonite, alkali-bearing carbonates) is present at all pressures (Fig. 7a, c and
d). It usually occurs as relatively large (up to 25 mm) subhedral crystals interstitial to garnet (or clinopyroxene at
9^13 GPa). Where both phases are present, magnesite usually occurs along aragonite crystal boundaries (Fig. 7a).
Crystalline carbonates coexisting with low-degree silicate^carbonate melts were observed in only one experiment (V1250-17).
The type andcomposition of carbonate present varies with
runtemperature and pressure (Fig.8). At 9 GPa andtemperatures of 12008C, in both GA1cc and Volga-cc, nearly pure
0·00
—
3·99
aragonite (91·5^95·4 mol % CaCO3) coexists with magnesite
(72·5^75·6 mol % MgCO3) that has significant CaCO3
(10·9^12·1mol %) and FeCO3 (17·5^19·0 mol %) components. At higher temperatures, in G1250-9 andV1250-9, only
a single carbonate of siderite and magnesite-bearing calcite
composition (61·4^66·1mol % CaCO3, 21·3^27·7 mol %
MgCO3, 8·1^9·7 mol % FeCO3) is present. This is also the
case for the experiments at13 GPa. In subsolidus experiments
G1200-13 and V1200-13 two carbonates, aragonite and magnesite, are observed. However, in the V1300-13 experiment
(in contrast to G1200-13), the carbonate is no longer pure
CaCO3 but contains significant Na-, K-, Mg- and Fe-bearing
components. Although the structure of the crystallized carbonate has not been determined, its composition is similar to
the aragonite crystallized at high pressures in alkali-carbonatite systems (Litasov et al., 2013).
In most experiments at 17 GPa, pure aragonite was not
observed and Na-rich aragonite (4·60^8·42 wt % Na2O;
1564
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Table 5: Compositions of experimental carbonate
GA1cc
Phase:
T (8C):
P (GPa):
Mst
1200
9
Arag
1200
9
CMss
1250
9
Mst
1200
13
Arag
1200
13
Arag
1100
17
n:
2
s
5
s
8
s
4
s
5
s
2
s
SiO2
TiO2
Al2O3
FeO
MnO
MgO
CaO
Na2O
K2O
P2O5
CO2
Total
0·16
b.d.l.
0·18
12·55
b.d.l.
29·31
6·09
0·22
b.d.l.
b.d.l.
51·49
100
0·06
—
0·04
0·42
—
1·74
1·27
0·07
—
—
0·79
b.d.l.
b.d.l.
b.d.l.
b.d.l.
b.d.l.
b.d.l.
51·29
b.d.l.
b.d.l.
b.d.l.
48·71
100·00
—
—
—
—
—
—
0·90
—
—
—
1·02
0·95
0·23
0·65
5·81
b.d.l.
8·58
37·08
0·20
1·15
0·27
45·08
100·00
0·53
0·13
0·27
0·89
—
0·56
0·69
0·11
0·26
0·05
1·31
3·15
b.d.l.
1·17
11·09
b.d.l.
32·22
2·68
0·48
b.d.l.
b.d.l.
49·21
100·00
0·96
—
0·60
0·59
—
0·92
2·00
0·12
—
—
1·29
b.d.l.
b.d.l.
b.d.l.
0·41
b.d.l.
b.d.l.
52·14
b.d.l.
b.d.l.
b.d.l.
47·46
100·00
—
—
—
0·18
—
—
0·75
—
—
—
0·80
0·25
b.d.l.
b.d.l.
3·41
b.d.l.
5·03
37·01
5·37
0·53
0·12
48·29
100·00
0·21
—
—
0·21
—
0·62
0·48
1·65
0·01
0·01
1·92
GA1cc
Phase:
T (8C):
P (GPa):
Mst
1200
17
Arag
1200
17
Arag
1250
17
Arag
1200
21
Arag
1300
21
Arag
1400
21
n:
1
3
s
5
s
5
s
3
s
5
s
SiO2
TiO2
Al2O3
FeO
MnO
MgO
CaO
Na2O
K2O
P2O5
CO2
Total
0·63
0·10
1·01
9·46
b.d.l.
33·21
1·84
0·60
b.d.l.
0·09
53·06
100·00
b.d.l.
b.d.l.
b.d.l.
3·72
b.d.l.
5·90
37·89
5·58
0·57
0·15
46·19
100·00
—
—
—
0·40
—
0·23
0·76
1·65
0·05
0·07
0·83
0·44
b.d.l.
b.d.l.
2·94
b.d.l.
5·51
36·04
8·06
0·53
b.d.l.
46·48
100·00
0·28
—
—
0·28
—
0·26
0·35
0·19
0·05
—
0·56
0·41
b.d.l.
0·14
3·27
b.d.l.
5·36
36·92
6·09
0·43
b.d.l.
47·39
100·00
0·24
—
0·06
0·35
—
0·72
1·63
0·96
0·02
—
1·83
b.d.l.
b.d.l.
b.d.l.
5·12
b.d.l.
4·80
33·67
6·56
0·53
b.d.l.
49·32
100·00
—
—
—
0·54
—
0·49
0·97
0·43
0·08
—
1·03
0·18
b.d.l.
b.d.l.
4·12
b.d.l.
5·93
36·23
6·16
0·99
0·17
46·21
100·00
0·10
—
—
0·46
—
0·32
1·27
0·94
0·54
0·08
1·03
Volga-cc
Phase:
T (8C):
P (GPa):
Mst
1050
9
Arag
1050
9
n:
1
2
SiO2
TiO2
Al2O3
FeO
MnO
MgO
CaO
Na2O
K2O
P2O5
CO2
Total
0·19
b.d.l.
0·30
13·64
0·23
30·48
6·79
0·30
b.d.l.
b.d.l.
48·07
100·00
0·15
b.d.l.
b.d.l.
b.d.l.
b.d.l.
b.d.l.
53·52
b.d.l.
b.d.l.
b.d.l.
46·33
100·00
Mst
1200
9
Arag
1200
9
CMss
1250
9
Mst
1200
13
s
1
2
s
3
s
3
s
0·21
—
—
—
—
—
0·24
—
—
—
0·03
0·42
0·01
0·17
12·74
0·23
29·21
6·79
0·30
0·04
0·02
50·07
100·00
b.d.l.
b.d.l.
0·12
b.d.l.
b.d.l.
0·10
52·82
b.d.l.
b.d.l.
b.d.l.
46·97
100·00
—
—
0·06
—
—
0·03
1·81
—
—
—
0·69
b.d.l.
0·11
0·54
6·95
0·22
11·15
34·43
0·27
1·18
b.d.l.
45·16
100·00
—
0·05
0·06
0·84
0·04
0·35
0·72
0·08
1·58
—
1·49
1·03
b.d.l.
0·37
9·80
b.d.l.
33·89
1·54
0·31
b.d.l.
b.d.l.
53·06
100·00
0·15
—
0·13
0·23
—
0·50
0·16
0·05
—
—
0·94
(continued)
1565
JOURNAL OF PETROLOGY
VOLUME 54
NUMBER 8
AUGUST 2013
Table 5: Continued
Volga-cc
Phase:
T (8C):
P (GPa):
Arag
1200
13
n:
4
SiO2
TiO2
Al2O3
FeO
MnO
MgO
CaO
Na2O
K2O
P2O5
CO2
Total
0·69
b.d.l.
b.d.l.
2·92
b.d.l.
4·51
36·25
6·41
1·50
b.d.l.
47·72
100·00
Mst
1100
17
Arag
1100
17
Arag
1200
17
Mst
1250
17
Arag
1250
17
s
1
4
s
3
s
2
s
7
s
0·62
—
—
0·47
—
0·54
1·75
1·82
0·04
—
0·99
2·20
b.d.l.
0·94
9·54
0·19
34·57
1·16
0·49
b.d.l.
b.d.l.
50·91
100·00
0·26
b.d.l.
b.d.l.
3·62
b.d.l.
4·87
35·14
4·60
0·38
0·13
51·02
100·00
0·11
—
—
0·72
—
1·36
1·78
0·59
0·01
0·04
3·91
0·31
0·16
b.d.l.
3·89
b.d.l.
6·15
36·47
6·20
0·42
0·17
46·23
100·00
0·00
0·09
—
0·38
—
0·48
1·86
1·26
0·03
0·08
1·22
0·34
b.d.l.
b.d.l.
9·47
b.d.l.
38·51
1·22
0·70
0·02
b.d.l.
49·74
100·00
0·02
—
—
0·40
—
4·04
0·07
0·27
0·03
—
4·32
b.d.l.
b.d.l.
b.d.l.
3·21
b.d.l.
5·42
34·05
8·42
0·49
b.d.l.
48·40
100·00
—
—
—
0·16
—
0·36
1·09
0·81
0·14
—
0·75
Volga-cc
Phase:
T (8C):
P (GPa):
Arag
1200
21
Na-Carb
1200
21
n:
4
s
2
s
SiO2
TiO2
Al2O3
FeO
MnO
MgO
CaO
Na2O
K2O
P2O5
CO2
Total
b.d.l.
b.d.l.
b.d.l.
3·90
b.d.l.
5·65
38·08
6·38
0·27
b.d.l.
45·72
100·00
—
—
—
0·76
—
0·93
1·70
0·38
0·03
—
0·00
1·26
b.d.l.
b.d.l.
2·13
b.d.l.
3·48
13·57
21·02
3·77
b.d.l.
54·78
100·00
0·85
—
—
0·30
—
0·03
1·03
0·33
0·23
—
0·01
CO2 content was calculated from the mass-balance calculations and EDS totals.
0·38^0·53 wt % K2O; 4·89^5·90 wt % MgO; 2·94^3·89 wt
% FeO) coexists with small amounts of magnesite
(1 modal %) that usually has about 13·2 mol % FeCO3
and 2·1^3·2 mol % CaCO3. No magnesite was found in
experiments at 21GPa, leaving Na-rich aragonite as the
only solid carbonate, except in run V1200-21, where Nacarbonate
with
the
approximate
composition
(Na,K)2(Ca,Mg,Fe)(CO3)2 (Fig. 7c) was detected. Owing
to the small grain size and the presence of multiple, tiny inclusions in garnet, we do not exclude the possibility that
unobserved magnesite is a subsolidus phase at 21GPa.
Stishovite or coesite (at 9 GPa) is an accessory phase at
all pressures in both starting compositions, reaching 10
modal % at 21GPa in G1300-21 and V1400-21. It is usually
present as inclusions in garnet or clinopyroxene, and often
occurs as quench crystals in carbonate-rich melts. At
17^21GPa, the amount of stishovite is higher than at
9^13 GPa. At subsolidus conditions it is anhedral to subhedral (Fig. 1d), whereas with increasing temperature the
crystals become euhedral and slightly elongated. In some
of the experiments, stishovite consists of pure SiO2.
However, up to 5 wt % Al2O3 was observed in stishovite
1566
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Table 6: Compositions of experimental coesite, stishovite and K-hollandite
Composition:
GA1cc
GA1cc
GA1cc
GA1cc
GA1cc
GA1cc
Volga-cc
Volga-cc
Volga-cc
Phase:
Co
St
St
St
St
St
St
St
St
T (8C):
1250
1200
1400
1300
1400
1650
1550
1250
1200
P (GPa):
9
13
17
21
21
21
13
17
21
SiO2
97·20
92·92
94·31
95·62
92·14
96·57
96·39
97·03
89·56
TiO2
b.d.l.
0·83
b.d.l.
0·20
0·35
0·31
0·41
0·19
1·74
Al2O3
0·61
1·87
1·67
1·11
4·31
2·99
1·23
0·45
0·44
FeO
0·50
0·14
0·19
1·11
b.d.l.
0·32
0·66
0·54
0·78
MnO
b.d.l.
0·13
b.d.l.
0·10
0·03
b.d.l.
b.d.l.
b.d.l.
b.d.l.
MgO
0·26
0·24
0·19
0·30
0·77
0·59
0·70
0·13
0·54
CaO
0·48
1·00
0·56
0·34
1·30
1·49
1·22
0·57
2·27
Na2O
b.d.l.
0·80
1·42
b.d.l.
0·23
0·21
0·27
0·41
1·11
K2O
b.d.l.
0·32
0·65
b.d.l.
0·25
b.d.l.
b.d.l.
b.d.l.
0·50
P2O5
b.d.l.
0·21
b.d.l.
b.d.l.
b.d.l.
b.d.l.
b.d.l.
b.d.l.
0·11
Total
99·05
98·46
98·99
98·78
99·38
102·48
100·88
99·32
97·05
Composition:
Volga-cc
GA1cc
GA1cc
GA1cc
Volga-cc
Volga-cc
Volga-cc
Volga-cc
Phase:
St
K-Holl
K-Holl
K-Holl
K-Holl
K-Holl
K-Holl
K-Holl
T (8C):
1650
1200
1250
1300
1200
1100
1200
1250
P (GPa):
21
17
17
21
13
17
17
17
63·81
SiO2
96·43
63·79
67·51
65·13
66·01
62·45
63·24
TiO2
0·46
0·72
0·38
0·37
0·27
1·82
1·11
0·47
Al2O3
0·97
18·58
18·05
18·11
18·32
17·44
18·77
19·01
FeO
0·35
0·77
b.d.l.
0·54
0·62
0·88
0·62
0·32
MnO
0·12
b.d.l.
b.d.l.
b.d.l.
b.d.l.
b.d.l.
b.d.l.
b.d.l.
b.d.l.
MgO
0·47
0·73
b.d.l.
0·20
0·27
0·61
0·18
CaO
1·28
1·67
0·91
0·89
0·70
1·90
1·08
0·61
Na2O
0·92
0·96
0·98
1·09
0·42
0·76
0·86
0·90
K2O
0·48
13·16
13·87
13·67
15·39
11·71
13·72
14·87
P2O5
b.d.l.
Total
101·48
b.d.l.
100·39
b.d.l.
101·69
b.d.l.
b.d.l.
100·00
from some experiments, although because of the small
grain size it is hard to measure the composition precisely.
Quenched melt pools exhibit a high proportion of crystalline SiO2. No other SiO2-bearing phase is present in
the quenching products. Usually, these crystals are small
and elongated.
K-hollandite is observed mainly in experiments at
17^21GPa, although small amounts were also observed at
subsolidus conditions at 13 GPa (i.e. V1200-13). At pressures
of 17 and 21GPa (although more abundant at 17 GPa),
K-hollandite is the most common accessory phase in the
subsolidus runs in both compositions. It persists to 21GPa
but in lesser amounts and as smaller crystals. Usually it
102·00
0·27
97·84
b.d.l.
99·58
b.d.l.
99·97
forms eudredral, elongated (up to 20 mm long) inclusions
in garnet (Fig. 3e). The amount of K2O in all the measured
K-hollandite crystals varies between 11·7 and 15·4 wt %;
other components include CaO (0·61^1·90 wt %),
TiO2 (0·27^1·82 wt %), Na2O (0·42^1·09 wt %) and FeO
(0^0·89 wt %).
Ca-perovskite was observed at 21GPa. It is more
abundant in experiments with the Volga-cc starting material. It usually forms tiny (1^3 mm) well-shaped cubic
crystals. Its small grain size prevents precise analysis.
Nevertheless, all the values are consistent and range
within 25·2^37·7 wt % TiO2, 28·8^40·4 wt % CaO and
18·0^29·3 wt % SiO2. The main impurities are Al2O3
1567
JOURNAL OF PETROLOGY
VOLUME 54
NUMBER 8
AUGUST 2013
Table 7: Compositions of experimental Ca-perovskite
Composition:
GA1cc
Volga-cc
T (8C):
1300
1400
1200
1200
1300
1400
1400
P (GPa):
21
21
17
21
21
21
21
n:
3
1
5
s
s
4
s
2
s
s
2
s
2
SiO2
26·55
0·85
18·02
9·97
23·03
5·62
29·26
28·78
2·23
26·10
0·70
26·10
0·70
TiO2
26·51
1·67
34·70
11·43
37·73
6·51
25·17
26·82
1·85
26·82
1·70
26·82
1·70
Al2O3
4·98
1·23
1·69
5·91
3·13
0·65
5·56
3·19
1·14
3·55
0·63
3·55
0·63
FeO
1·47
0·46
1·48
1·28
2·63
0·69
2·28
1·03
0·25
1·26
0·33
1·26
0·33
MnO
0·12
0·07
0·11
0·08
b.d.l.
b.d.l.
b.d.l.
MgO
1·35
0·48
0·63
1·92
0·84
0·46
1·54
0·67
0·52
0·83
0·42
0·83
0·42
CaO
37·54
1·67
40·40
6·95
28·82
1·00
34·73
37·92
1·83
39·75
1·34
39·75
1·34
Na2O
1·16
0·29
2·44
0·87
2·49
0·26
1·08
1·31
0·63
1·32
0·84
1·32
0·84
K2O
0·33
0·07
0·52
0·25
1·33
0·17
0·38
0·27
0·19
0·38
0·29
0·38
P2O5
b.d.l.
b.d.l.
b.d.l.
Total
100·00
—
b.d.l.
100·00
—
b.d.l.
—
—
100·00
100·00
(which may be related to overlapping crystals), Na2O and
FeO. The abundance of CaO and to a lesser extent TiO2
and Na2O in Ca-perovskite increases with increasing temperature, whereas the amounts of SiO2, Al2O3 and MgO
decrease.
Most of the experiments with the GA1cc starting material, at both subsolidus and above-solidus conditions, contained accessory graphite or diamond crystals (Figs 1b, 7b
and 9a, b). Crystal size ranged from 10 to 40 mm. At
9^13 GPa, grain shapes were anhedral or sometimes
rounded. Under subsolidus conditions the grains had a
clear basal cleavage indicating that they were graphite. At
17^21GPa, the shape of this phase appeared more crystalline, and the hardness of the grains while polishing indicated the formation of diamond for both subsolidus and
supersolidus runs. The presence of diamond in some
runs (e.g. G1400-21) was verified by laser Raman
spectroscopy.
Experimental melts and their compositions
All the supersolidus experiments in this study contain 7·5^
18% melt. Melts did not quench to a glass but instead usually formed heterogeneous pools of quenched silicate and
carbonate phases, interstitial between coarser residual
crystals of garnet and other phases. The compositions of
these metastable quench crystals could not generally be
determined precisely because of their very fine grain size.
(Figs 1c and 3c, d, f). In some cases, melt partially segregated to distinct zones in the capsules.
100·00
—
—
b.d.l.
b.d.l.
100·00
—
—
b.d.l.
b.d.l.
—
0·29
—
100·00
All of the melts produced are carbonate-rich (25^46%
CO2). The amount of CO2 in the melt has been estimated
from analytical totals that deviate from 100%, as well as
by mass-balance calculations. As previously reported
(Litasov & Ohtani, 2010), the melt composition evolves
from carbonatitic near the solidus to a more siliceous composition with increasing temperature and degree of melting. The heterogeneity of the produced melts is manifested
mainly by high variations in SiO2, Al2O3 and CaO contents (Table 8). Low-degree melts in both compositions
at 9 GPa are very similar to the solid carbonate
compositions (Fig. 8), but contain significant amounts of
TiO2 (2·67^3·18 wt %) and SiO2 (1·82^4·64 wt %). The
melts for GA1cc and Volga-cc differ slightly at 13 GPa,
with different proportions of SiO2, Al2O3 and CaO. The
amount of alkali components, TiO2, FeO and MgO in
both melts is similar. All the melts at 17 GPa for both compositions (except V1250-17) are silicate^carbonate at relatively high (15·5^18%) degrees of melting. Similar to runs
at 13 GPa, melt produced by the GA1cc composition at
21GPa is more SiO2-rich, whereas melts of the Volga-cc
starting material contain much higher concentrations of
CaO. The only low-degree melt (7·5% melting) that
was analyzed (in experiment V1250-17) coexists with solid
magnesite and aragonite and is alkali-rich (14·8 wt %
Na2O and 3·6 wt % K2O).
The Ca# of the melts decreases slightly and Mg# increases slightly with increasing pressure and increasing
degree of melting (Fig. 10a and b). However, the Ca#
for melts at 17 GPa is lower than for melts at 21GPa.
1568
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Table 8: Compositions of experimental melts
GA1cc
T (8C):
1400
1300
1400
1550
1400
P (GPa):
9
13
13
13
17
17
Type of melt:
Carb
Si-Carb
Si-Carb
Si-Carb
Si-Carb
Si-Carb
n:
6
s
s
5
s
3
s
5
1500
s
6
s
2
SiO2
4·64
1·96
17·28
16·29
18·67
2·94
20·35
7·44
25·90
2·76
25·97
2·24
TiO2
2·67
0·61
0·82
0·26
1·46
0·55
0·54
0·33
0·90
0·19
1·00
0·02
Al2O3
1·48
0·88
3·38
2·95
2·66
2·13
1·62
1·64
4·05
2·32
5·75
4·16
FeO
5·62
0·68
5·14
0·47
5·72
0·56
5·33
0·56
4·81
0·59
5·03
MnO
b.d.l.
0·16
0·08
0·19
0·18
0·10
0·03
b.d.l.
—
b.d.l.
—
0·21
—
MgO
4·82
0·29
5·97
1·00
5·09
0·38
6·01
0·91
5·27
0·74
5·55
0·67
CaO
31·96
1·88
22·06
2·71
22·00
1·91
25·88
3·69
18·95
1·08
20·94
4·60
Na2O
0·78
0·29
5·12
0·54
5·41
1·03
2·56
1·55
9·06
1·00
8·57
2·14
K2O
2·89
1·77
2·12
0·14
1·43
0·36
1·20
0·84
1·92
0·36
1·88
0·47
P2O5
1·40
0·22
CO2
43·74
2·71
Total
100·00
100·00
100·00
100·00
100·00
100·00
b.d.l.
38·11
—
15·29
0·38
0·07
0·48
0·17
37·00
0·00
35·83
4·60
b.d.l.
—
29·04
3·73
0·25
0·07
25·06
0·00
Element ratios
Mg#
60·46
67·45
61·31
66·79
66·14
66·28
CaO/SiO2
6·89
1·28
1·18
1·27
0·73
0·81
K2O/Na2O
3·69
0·41
0·26
0·47
0·21
0·22
Volga-cc
V1300-13a
V1300-13b
GA1cc
T (8C):
1650
1400
1300
1300
1400
P (GPa):
21
9
13
13
13
13
Type of melt:
Si-Carb
Carb
Carb
Carb
Carb
Carb
n:
4
s
s
5
s
5
s
1
1550
s
7
s
8
SiO2
23·48
1·84
1·82
1·09
14·57
3·04
7·63
1·89
1·20
3·25
2·80
TiO2
1·17
0·17
3·18
1·34
0·65
0·21
0·21
0·73
0·24
0·93
0·48
Al2O3
0·91
0·50
0·63
0·42
2·24
1·35
0·20
0·73
0·54
1·04
1·18
FeO
3·17
0·24
7·00
0·21
4·72
0·27
5·25
7·13
1·20
7·96
MnO
b.d.l.
0·24
0·16
b.d.l.
0·11
b.d.l.
—
—
—
b.d.l.
1·36
—
MgO
3·88
0·26
5·18
0·83
4·72
0·47
5·39
5·90
1·46
6·74
CaO
18·84
0·64
32·66
1·05
23·53
1·50
28·71
30·64
1·42
33·00
2·04
3·49
Na2O
10·32
0·46
0·82
0·16
6·68
0·63
5·34
3·72
1·25
5·82
3·90
2·01
K2O
2·59
0·32
1·82
0·57
3·02
0·27
3·34
1·83
0·91
2·38
P2O5
0·36
0·03
1·22
0·38
0·37
0·05
0·32
0·63
0·20
0·76
0·34
CO2
35·30
1·34
45·42
1·54
39·49
2·95
43·50
46·79
1·39
38·12
0·00
Total
100·00
100·00
100·00
100·00
100·00
100·00
Element ratios
Mg#
68·57
56·87
64·04
64·66
59·61
60·17
CaO/SiO2
0·80
17·90
1·61
3·76
16·18
10·15
K2O/Na2O
0·25
2·23
0·45
0·63
0·49
0·41
(continued)
1569
JOURNAL OF PETROLOGY
VOLUME 54
NUMBER 8
AUGUST 2013
Table 8: Continued
Volga-cc
T (8C):
1250
1400
1500
1300
1400
1650
P (GPa):
17
17
17
21
21
21
21
Type of melt:
Si-Carb
Si-Carb
Si-Carb
Carb
Carb
Carb
Carb
n:
6
s
s
4
s
4
s
4
s
5
1900
s
9
s
5
SiO2
15·20
8·35
28·08
4·60
28·25
14·08
1·98
0·49
4·50
3·08
5·02
3·61
9·58
3·52
TiO2
0·54
0·13
0·78
0·16
0·89
0·23
0·44
0·19
0·49
0·38
1·53
1·11
3·24
0·41
Al2O3
0·62
0·23
5·37
0·75
3·13
3·37
0·35
0·11
0·95
0·57
0·97
0·45
2·77
1·60
FeO
4·06
0·46
3·83
0·56
4·45
1·01
4·73
0·68
5·42
1·62
6·91
1·09
5·77
0·62
MnO
b.d.l.
0·14
0·04
0·19
0·10
0·12
0·04
MgO
4·18
0·60
4·83
0·55
5·72
1·26
6·52
0·41
5·45
1·08
5·75
0·72
5·18
0·45
CaO
15·99
1·63
16·34
1·65
19·65
4·40
26·32
0·77
24·95
1·55
27·95
1·66
21·96
3·25
Na2O
14·79
1·16
8·56
0·99
7·91
1·72
10·85
0·50
12·42
1·52
9·67
2·82
11·38
3·37
K2O
3·62
0·34
1·97
0·33
3·10
1·35
3·23
0·11
3·49
0·46
3·41
0·69
3·60
0·62
P2O5
0·11
0·04
0·14
0·05
0·38
0·13
b.d.l.
0·19
0·07
0·58
0·14
1·39
0·25
CO2
40·90
4·73
30·10
0·00
26·52
8·21
42·00
0·00
38·00
0·00
35·00
0·00
Total
100·00
100·00
100·00
—
b.d.l.
—
b.d.l.
—
b.d.l.
45·60
—
—
0·36
100·00
100·00
100·00
100·00
Element ratios
Mg#
64·74
69·19
69·61
71·08
64·20
59·72
61·53
CaO/SiO2
1·05
0·58
0·70
13·33
5·54
5·57
2·29
K2O/Na2O
0·24
0·23
0·39
0·30
0·28
0·35
0·32
CO2 content was calculated from the mass-balance calculations and EDS totals. The normalized values of melts totals
have 0·0 standard deviation for CO2, compared with those that have not been normalized.
Fig. 2. Experimental P^T phase diagram for GA1cc and Volga-cc. Abbreviations as in Table 2. Y, solidus of dry eclogite by Yasuda et al. (1994);
HF, solidus of dry eclogite by Hirose & Fei (2002); K, solidus for GA1cc in Au^Pd capsules at 5 GPa (Kiseeva et al., 2012). Gr-D indicates graphite^diamond transition (Kennedy & Kennedy, 1976). Circles indicate experimental runs. Diamonds inside the circles indicate the presence of
diamond or graphite in GA1cc runs.
1570
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Fig. 3. BSE images of the experimental runs. (a) Run V1400-9. Mineral assemblage at 9 GPa above the solidus. (b) Run V1200-13 subsolidus
mineral assemblage at 13 GPa. (c) Run G1300-13 showing heterogeneously distributed melt and patches of carbonate melt separated from
areas of silicate^carbonate melt. (d) Run G1400-17. Heterogeneous melt pools. (e) Run G1100-17. K-hollandite crystal included in Grt. (f) Run
G1650-21. Heterogeneous silicate^carbonate melt at 21GPa. Carbonate-rich matrix with quenched CAS and stishovite crystals.
Unlike Keshav & Gudfinnsson (2010), in this study any
increase in the Mg content of the melt with increasing
pressure is not observed. The amount of alkali components in the melt increases dramatically with increasing
pressure, and the Na/K ratio increases up to the point of
K-hollandite saturation, and then subsequently decreases
(Fig. 11a and b).
DISCUSSION
Solidus of carbonated eclogite and
comparison with previous studies
Experimental data on MORB-like compositions at pressures above 8^10 GPa are limited. Studies on volatile-free
MORB and K-rich MORB compositions (Yasuda et al.,
1994; Wang & Takahashi, 1999; Hirose & Fei, 2002)
1571
JOURNAL OF PETROLOGY
VOLUME 54
NUMBER 8
AUGUST 2013
Fig. 4. Na content of experimentally crystallized garnet and clinopyroxene. (a) Na distribution amongst the major phases within an eclogite assemblage. Dashed line indicates 1^1 ratio; numbers indicate pressure in the experimental runs reported in literature. Arrows indicate pressure
and the amount of clinopyroxene in the experiments with GA1cc and Volga-cc compositions. (b) Na vs Si in experimental garnet. Pt-Gr, experiments in Pt^graphite capsule; Au-Pd, experiments in Au^Pd capsule (see Kiseeva et al. 2012). O&M, hydrous MORB (Okamoto &
Maruyama, 2004); L&O, carbonated MORB (Litasov & Ohtani, 2010); Y, dry MORB (Yasuda et al., 1994).
reported an increase in solidus temperatures from 16008C
at 8 GPa to 2100^22008C at 20 GPa. The melts in these
studies are silica-rich, with 50^60 wt % SiO2 at pressures
below 20 GPa (Yasuda et al., 1994; Wang & Takahashi,
1999). At higher pressures the amount of SiO2 in the
melts decreases to 44^48 wt % (Wang & Takahashi,
1999; Hirose & Fei, 2002). The amount of alkali components in the low-degree melts is strictly governed by the
1572
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Fig. 5. Na2O content of the experimentally crystallized phases. (a)
Garnet. (b) Clinopyroxene.
stability of clinopyroxene and K-hollandite, as in the present study.
The solidus of hydrous MORB has been estimated by
Okamoto & Maruyama (2004) to lie near 12008C at
19 GPa, which is similar to the GA1cc and Volga-cc solidi
and about 10008C below the dry MORB solidus at the
same pressure. Litasov & Ohtani (2005) reported a much
higher solidus temperature for hydrous MORB at
18^28 GPa, only 50^1008C lower than the dry MORB
solidus. Those researchers suggested that a small amount
of supercritical fluid was present in the runs even at the
lowest temperatures (10008C), and reported an ‘apparent
solidus’ based on extensive melting of the main silicate
phases, which occurred above 20008C at 20 GPa. Similar
to the melts produced from dry MORB compositions, the
melts reported by Litasov & Ohtani (2005) are silica-rich.
There are only two published experimental studies on carbonated eclogite compositions at pressures above 10 GPa,
both in simplified systems: CMAS þ 20% CO2 (Keshav &
Gudfinnsson, 2010) and Na-CMAS þ 5% CO2 (Litasov &
Ohtani, 2010). The solidus of carbonated eclogite at 10^
20 GPa was reported to lie about 400^5008C below the dry
eclogite solidus (Litasov & Ohtani, 2010). At 10 GPa, it was
about 50^1008C higher than the carbonated eclogite solidus
at 9 GPa of Dasgupta et al. (2004) (Fig.12).
Keshav & Gudfinnsson (2010) referred to their melts as
‘calcio-carbonatites’, with a substantial increase in the
MgCO3 component from low pressures (12^16 GPa) to high
pressures (20^25 GPa); the amount of SiO2 in the melts does
not exceed 2·4 wt %. Litasov & Ohtani (2010) faced problems
with determination of partial melt compositions at 10·5 and
16·5 GPa owing to possible coexistence of both carbonatitic
and carbonate-rich silicate melts, which indicate liquid immiscibility or heterogeneity across the sample. The Na2O
content in either of these melts does not exceed 3·1wt %
within the stability field of clinopyroxene, and is up to 7·2 wt
% Na2O at higher pressures, where clinopyroxene is no
longer stable. This is similar to the melting style of GA1cc
and Volga-cc, which both exhibit a dramatic increase in the
alkali contents of the melts at pressures above clinopyroxene
stability. Both studies reported magnesite (or magnesite and
aragonite together) coexisting with carbonatitic melt.
Unlike at lower pressures (55^6 GPa), all of the MORB
solidi except Volga-cc show a gradual increase in solidus
temperature with increasing pressure. However, the
slopes of the various solidi in P^Tspace are highly variable
(Fig. 12). Volatile-free eclogite solidi are very steep and
linear up to the point of clinopyroxene disappearance, but
significantly less steep at higher pressures (Yasuda et al.,
1994; Wang & Takahashi, 1999; Hirose & Fei, 2002). A
steep linear solidus was also reported for CMAS þ 20%
CO2 (Keshav & Gudfinnsson, 2010) and MORB þ 2%
H2O (Litasov & Ohtani, 2005). On the other hand, most
alkali-bearing, carbonated eclogites and peridotites
(Ghosh et al., 2009; Litasov & Ohtani, 2009b, 2010) display
essentially flat solidi with increasing pressure from 10 to
20^30 GPa, in good agreement with the solidi of GA1cc
and Volga-cc determined in this study.
The probable cause for the dramatic differences in solidus temperatures between different carbonated eclogite
compositions lies in similar compositional parameters
identified in lower pressure studies (Dasgupta et al., 2004,
2005), which include Ca#, amount of CO2 and H2O, and
CaO/MgO and Na2O/CO2 ratios. Litasov et al. (2013) considered the true solidus of hydrogen-free carbonated eclogite and peridotite to be strongly influenced by the amount
of alkalis and placed it at temperatures similar to this
study. Those researchers also reported a possible negative
slope from 15 to 21GPa for the Na-carbonatite solidus,
which is similar to the solidus determined for Volga-cc.
The main host for K in different starting compositions
differs significantly. In more K-rich compositions, such as
1573
JOURNAL OF PETROLOGY
VOLUME 54
NUMBER 8
AUGUST 2013
Fig. 6. Compositions of experimentally crystallized clinopyroxene. (a) Sum of divalent cations as a function of pressure. (b) Al content as a
function of Si content. Labels as in Fig. 4.
dry CO2-bearing pelite with 2·21wt % K2O in the starting mixture, a K-bearing phase that incorporates most of
the bulk K2O has been reported by Grassi & Schmidt
(2011a, 2011b) across the entire P^T range studied (e.g.
T ¼ 900^15508C; P ¼ 5·5^23·5 GPa). Those researchers
observed K-feldspar, which crystallized at 59 GPa and
was followed by K-hollandite at 49 GPa, with other
phases such as clinopyroxene, carbonate or garnet containing very small amounts of K (usually 1wt %). K-feldspar was also reported at 900 and 10008C and 3 GPa by
Tsuno & Dasgupta (2012) for carbonated pelite composition with 1·99 wt % K2O in the starting mixture.
1574
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Fig. 7. C-bearing phases at different P^T conditions. (a) Run G1200-9. Aragonite and magnesite. (b) Run G1400-13. Diamond or graphite
within carbonate melt. (c) Run V1200-21. Na-carbonates coexisting with Na-rich aragonite. (d) Run G1400-21. Na-rich aragonite within finegrained matrix.
For carbonated eclogite compositions, like those studied
by Litasov et al. (2013), the low solidus temperatures may
be attributed mainly to high alkali/CO2 ratios and to
the stability of alkali carbonates. Although at pressures
higher than around 6 GPa in the simple KAlSi3O8 or
(K,Na)AlSi3O8 systems, K-feldspar was noted to transform to wadeite þ kyanite þ coesite (Urakawa et al., 1994;
Yagi et al., 1994), no potassium-bearing crystalline phase
was detected in most experimental runs on eclogite and
peridotite systems between the stability fields of sanidine
and K-hollandite (Wang & Takahashi, 1999; Ghosh et al.,
2009). Hence within the pressure range 5^13 GPa potassium is highly incompatible in silicate phases, and under
anhydrous conditions any potassium would partition into
the melt or a carbonate phase if not incorporated into
clinopyroxene. This was well documented by Wang &
Takahashi (1999), who reported 6·64 wt % K2O at 5 GPa
in the melt and up to 1·9 wt % K2O in clinopyroxene at
about 7 GPa. In CO2-bearing starting compositions, a Kcarbonate phase may be a more plausible host for K
than clinopyroxene; however, this needs more
clarification.
With increasing pressure up to the clinopyroxene-out
phase boundary at 14^16 GPa, Na becomes more compatible in clinopyroxene. The more Na is in the system, the
more jadeitic clinopyroxene is formed. The disappearance
of clinopyroxene roughly coincides with the appearance of
K-hollandite, which changes the compatibility of Na and
K in opposite senses. K-hollandite incorporates all the K,
whereas Na becomes highly incompatible until the stability
fields of NAL and CF phases (425 GPa) are reached
(Hirose & Fei, 2002; Litasov & Ohtani, 2005). Although
some Na2O can be accommodated in majoritic garnet, in
the studied eclogitic systems the N2O concentration in
garnet does not exceed 2^3 wt % (Yasuda et al., 1994;
Wang & Takahashi, 1999; Litasov & Ohtani, 2005) and
goes up to 3·1wt % Na2O for majoritic garnet in experiment V1100-17. Thus, in the same manner as K, the ‘excess’
Na either fluxes the formation of low-degree melts, or, in
carbonated systems, it partitions into Na-rich crystalline
carbonates. This suggests that the Na- and K-bearing carbonated eclogite solidus will be largely controlled by the
melting of Na- and K-bearing carbonate phases, which
presumably are K-carbonates at 9^13 GPa and Na-
1575
JOURNAL OF PETROLOGY
VOLUME 54
NUMBER 8
AUGUST 2013
Fig. 8. Compositions of crystalline carbonate, melts, garnet and clinopyroxene.
carbonates at 17^21GPa. This may be also one of the reasons for the significant difference in GA1cc and Volga-cc
solidus temperatures at 17^21GPa. Whereas most of the
carbon in GA1cc partitions into Na-rich aragonite and diamond, no diamond was observed in Volga-cc, and Na-carbonate (with, perhaps, a lower solidus temperature than
Na-rich aragonite) crystallized. This is consistent with the
more sodic character of the clinopyroxene and melts
formed in experiments on the Volga-cc composition.
Carbonate-rich melts in GA1cc and Volga-cc persist up to
the highest temperatures of the experiments, with relatively minor participation of silicates in the melting process. This explains the low melt productivity (518%) over
the large range of P^T conditions of the present
experiments.
The alkali/CO2 ratio of the bulk-rock and low solidus
temperatures can also affect the stability fields of the
main phases, including the transformation from eclogite
to garnetite. In carbonate-bearing systems, this transformation should occur at lower pressures (15 GPa) relative
to carbonate-free systems, because of the increased partitioning of Na into Na-aragonite with increasing pressure.
The absence of clinopyroxene in Na-CMAS þ 5% CO2 experiments at 16·5 GPa (Litasov & Ohtani, 2010) may indirectly indicate the presence of an additional Na-bearing
phase. In contrast, in dry and hydrous eclogite compositions, clinopyroxene can be stable to higher pressures, in
the range of 16^19 GPa (Yasuda et al., 1994; Okamoto &
Maruyama, 2004). This has important implications for
mantle melting and density profiles. Owing to the low
melting temperatures of Na-bearing carbonates (511508C
between 10 and 21GPa; Litasov et al., 2010, 2013; this
study), melting of carbonated eclogite may commence in
the deep upper mantle or at the very top of the Transition
Zone. This will effectively remove at least some of the carbonate from the system at depths within the upper part of
the Transition Zone.
Stability of carbon-bearing phases in the
deep mantle
Most experimental studies of carbonate stability in the
mantle show that at 5^9 GPa dolomite breaks down to aragonite plus magnesite (Martinez et al., 1996; Luth, 2001;
Sato & Katsura, 2001; Buob et al., 2006; Morlidge et al.,
2006). Although there is poor agreement regarding where
this reaction occurs in the CaO^MgO^CO2 system at
lower pressures, most experimental studies place it around
12008C at 9 GPa. This study demonstrates that this reaction for compositionally complex natural basaltic compositions at 9 GPa occurs at temperatures similar to those in
simplified compositions (i.e. between 1200 and 13008C).
At temperatures below 13008C, pure aragonite and magnesite are present, whereas at higher temperatures calcite^
magnesite solid solution is observed. In more complicated
1576
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Fig. 9. Carbonate inclusions in diamond or graphite. SEM images of
G1200-9 run.
Na- and K-bearing systems at pressures 513 GPa, alkalirich carbonates either may coexist with magnesite (as
inferred from this study) or may form a solid solution
between alkali-rich and calcite-rich carbonates (13 GPa;
this study) with an estimated composition of
or
(Na,K)2Ca2(CO3)3^
(Na,K)2Ca(CO3)2^CaCO3
CaCO3. The amount of K- and Na- component in Na-rich
aragonite is buffered by coexisting K-hollandite and clinopyroxene (Grassi & Schmidt, 2011a, 2011b).
Despite some variation in the Na vs K content of Na-rich
aragonite, the proportions of alkali components relative
to (Ca þ Mg þ Fe) remain constant from 13 to 21GPa
(Fig. 8). The fluctuations observed at 17 GPa are within
analytical error, given that Na content may be underestimated during SEM analysis. This may indicate (1) that
the capacity of aragonite to incorporate alkali components
or to create a solid solution with alkali-carbonates is limited and does not depend on pressure, and (2) the formation of an alkali-bearing carbonate with a different
structure.
The detailed study of the structure of this alkali-bearing
carbonate is beyond the scope of this study. In any case,
both hypotheses are commensurate with the observation
of Na-carbonate crystals Na2(Ca,Mg,Fe)(CO3)2 (Fig. 7c)
in run V1200-21 coexisting with Na-rich aragonite. The
excess of Na and K that could not be incorporated into
majoritic garnet or alkali-rich carbonate triggered crystallization of additional Na-carbonate under subsolidus
conditions. Recently, Na^Ca carbonate containing
10·1^11·0 wt % Na2O and 34·4^38·6 wt % CaO was reported by Grassi & Schmidt (2011b) in a carbonate-bearing marine sediment bulk composition at 16^23·5 GPa and
1200^14008C, and more Na-rich carbonate with c.
20·8 wt % Na2O and c. 36 wt % CaO crystallized at
22^23·5 GPa and 1350^14008C. (K,Na)2Ca4(CO3)5 and
(K,Na)2(Mg,Fe,Ca)(CO3)2 carbonates formed at 21GPa
in alkali carbonatite starting mixture have been reported
by Litasov et al. (2013). Similar K2Mg(CO3)2 carbonate
has also been synthesized at 8 GPa and 12008C in the
study of Brey et al. (2011).
Another important observation is that alkali-rich carbonates tend to form a solid solution with calcium carbonate
rather than magnesite. The apparent absence of magnesite
at 21GPa could be the result of incorporation of all the Mg
into majorite and Na-rich aragonite. Unlike the previous interpretation that with increasing pressure to the Transition
Zone and lower mantle magnesite remains the only stable
carbonate (Takafuji et al., 2006; Isshiki et al., 2004; Litasov,
2011), there is a possibility that if Ca-bearing rocks of eclogitic
paragenesis are also present at that depth, aragonite (or
alkali-bearing aragonite in the case of high bulk alkali contents) may become stable as well.
The complex phase relations of carbonates at uppermantle^Transition Zone pressures are made more complex
by the presence of diamond or graphite. To our knowledge,
this is the first experimental demonstration of diamond
crystallization in a carbonated MORB composition. The
diamond aggregates coexist with either compositionally
variable crystalline carbonates (Fig. 9) or carbonate melts
(Fig. 7b). The fact that diamonds crystallized only in the
GA1cc bulk composition and not the Volga-cc bulk composition is of particular interest. A possible explanation
for diamond crystallization in GA1cc is the oxidation of
ferrous iron in silicate garnet as a consequence of the
increased stability of the andradite component (Simakov,
2006). However, the composition and modal proportions
of garnet in GA1cc and Volga-cc at 17^21GPa are almost
the same.
Unfortunately, it was not possible to directly measure
Fe3þ in the garnets from this study, although their cation
sums are consistent with most, if not all, iron being present
as Fe2þ. Another reason would be partial contamination
of the sample by hydrogen derived from cell assembly
parts during the experiment. This is not likely because
then diamond formation would have been expected in
both compositions, which is not the case.
1577
JOURNAL OF PETROLOGY
VOLUME 54
NUMBER 8
AUGUST 2013
Fig. 10. Compositions of experimental melts as a function of pressure.
(a) Ca#. (b) Mg#.
Fig. 11. Compositions of experimental melts as a function of pressure.
(a) Molar Na/K ratio. (b) (Na þ K) in mol %. (Grt þ Cpx) and Grt
indicate stability field of the phases coexisting with melts.
The following reaction governing diamond formation in
eclogite systems has been proposed by Luth (1993):
the two charges is the same. Thus, diamond versus carbonate stability will depend on the activity of the diopside
component. The fact that the diopside component in
GA1cc clinopyroxene from subsolidus experiments at
9 GPa is 5^10 mol % lower than the Volga-cc clinopyroxene at the same P^Tconditions is consistent with this analysis. However, the influence of clinopyroxene composition
on diamond formation at constant oxygen fugacity needs
further experimental investigation.
CaMgðCO3 Þ2 þ2SiO2 ¼ CaMgSi2 O6 þ 2C þ O2
dolomite þ coesite ¼ diopside þ diamond þ O2
Applying this reaction to the subsolidus experiments at
9 GPa, in which magnesite and aragonite coexist, results in
CaCO3 þ MgCO3 þ 2SiO2 ¼ CaMgSi2 O6 þ 2C þ O2 :
The equilibrium constant for this reaction is
Melting of carbonated eclogite in the deep
mantle
aCaMgSi2 O6 a2C f O2
K¼
:
aCaCO3 aMgCO3 a2SiO2
Given the presence of both coesite or stishovite and carbonates in all the experiments, it is possible to conclude
that the denominator can be assumed to be unity for
almost pure carbonate components. Because GA1cc and
Volga-cc compositions are run simultaneously (in the
same experiment), we assume that the oxygen fugacity in
There is general agreement, based on experimental studies
and on studies of natural rocks, that substantial amounts
of carbonate survive subduction beyond the sub-arc
regime and into the deeper upper mantle (Kerrick &
Connolly, 2001; Dasgupta & Hirschmann, 2010). Our results agree with previous studies and demonstrate that the
solidus of the carbonated mafic component of the subducting slab is at temperatures above most subduction
1578
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Fig. 12. Comparison of GA1cc and Volga-cc solidi with other experimental studies and mantle and subduction geotherms (field marked with
bold lines). OG, oceanic geotherm; CG, cratonic geotherm. Grey field indicates an approximate mantle adiabat with an average of
Tp ¼13158C (McKenzie & Bickle, 1988; McKenzie & O’Nions, 1991). Black and coloured lines mark the following solidi: WT, solidus of dry
eclogite by Wang & Takahashi (1999); KG, solidus of carbonated eclogite by Keshav & Gudfinnsson (2010); H, solidus of carbonated eclogite
by Hammouda (2003); D, solidus of carbonated eclogite by Dasgupta et al. (2004); YB, solidus of carbonated eclogite by Yaxley & Brey (2004);
LO, solidus of carbonated eclogite by Litasov & Ohtani (2010). With light grey lines the following solidi are marked: D, solidus of carbonated
peridotite by Dasgupta & Hirschmann (2006); L, solidus of carbonated peridotite by Litasov & Ohtani (2009b). Subduction geotherms are compiled from van Keken et al. (2002) and Syracuse et al. (2010). Other labels as in Fig. 2.
geotherms in the upper mantle, although it may intersect
subduction zone geotherms in the Transition Zone
(Fig. 12). Unlike all previous studies on MORB-like compositions by Yasuda et al. (1994), Wang & Takahashi (1999),
Hirose & Fei (2002), Keshav & Gudfinnsson (2010) and
Litasov & Ohtani (2010), the solidi of GA1cc and Volga-cc
are significantly below the mantle adiabat, at least to
21GPa.
Low-velocity seismic anomalies on top of the 410 km discontinuity in the vicinity of subducted slabs have been reported in some geophysical studies (Revenaugh & Sipkin,
1994; Song et al., 2004). Low P-wave velocity zones have
also been observed above the subducted Pacific slab along
almost its entire descending path into the Transition Zone,
in the depth range of 250^500 km (Zhao & Ohtani, 2009).
One possible interpretation of this may be partial melting
of the slab associated with dehydration, decarbonation or
their combined effects. Being located at the top and therefore hotter region of the subducted slab, carbonate-rich
(and perhaps some H2O-bearing) rocks are most likely to
be first in the melting sequence. On melting, they may
yield alkali-rich calcio-dolomitic melts across the whole
range of investigated pressures. The concentration of
alkali components in these melts will be dependent on
their content in the bulk-rock, on Na and K compatibilities, and on the degree of partial melting.
Na and K compatibilities in MORB mineral phases
(garnet and clinopyroxene) at higher pressures have been
addressed by many experimental studies (Wang &
Takahashi, 1999; Spandler et al., 2008; Ghosh et al., 2009).
Sodium can be incompatible, partitioning into the melt
relative to clinopyroxene at fairly low pressures of
53 GPa (e.g. Blundy & Dalton, 2000; Dasgupta et al.,
2005; Yaxley & Brey, 2004). It may also be incompatible at
pressures beyond the clinopyroxene stability field (i.e.
P415 GPa). However, between 4^5 and 15 GPa,
sodium is compatible in clinopyroxene because of the high
stability of jadeite. Therefore low-degree, highly sodic
melts are unlikely to form in carbonate eclogite in the
depth range of 90^400 km. Given that the subducting
mafic oceanic crust is expected to have Na/K41, K-rich
and Na-poor low-degree melts may form in this
90^400 km depth range. Conversely, at Transition Zone
pressures, the melts are likely to have higher Na/K ratios
because of the disappearance of clinopyroxene from the
system.
1579
JOURNAL OF PETROLOGY
VOLUME 54
Thus, if partial melting of carbonate begins at
250^300 km depth (7·5^9 GPa) the melts produced in the
top part of the subducted slab would have K-rich carbonate
compositions. These melts will segregate easily from the
eclogitic residue owing to their low density and low viscosity, removing potassium and other incompatible elements
from the residual slab. The next pulse of partial melting
will most probably start at the eclogite^garnetite transformation, which in carbonated eclogite systems will
begin near the top of the Transition Zone (about
400^450 km depth), corresponding to pressures of around
14^15 GPa. Low-degree melts are expected to be extremely
Na-rich, consistent with the sodic nature (14·8 wt %
Na2O, Na2O/K2O ¼ 4·09) of the low-degree partial melt
(7·5%) in experiment V1200-17. It is likely that most potassium has been removed from the system before these
depths, during the lower pressure melting pulse described
above. Being of low viscosity, these Na-rich melts will rapidly segregate from the garnetite residue, metasomatizing
the surrounding peridotite, or even fluxing its partial melting, or forming diamond by ‘redox freezing’ (Rohrbach &
Schmidt, 2011). It should be noted that the eclogite^garnetite transformation occurs at different pressures depending
on bulk-rock composition.
After considerable partial melting and removal of alkalis
from the carbonated eclogite system, some carbon may
still be preserved in the rock as diamond. This argument
is supported by the formation of diamonds at 17 GPa in
equilibrium with alkali-rich carbonate melt, and by the
presence of carbonate inclusions in diamonds from the
Transition Zone^lower mantle (Brenker et al., 2007). The
amount of diamond preserved in the rock will depend on
the bulk-rock composition and on oxygen fugacity.
Overall, we conclude that alkali-bearing carbonated
eclogite in the subducted slab will lose most of its volatile,
carbonate and alkali components during multiple partial
melting events upon descent, leaving only refractory eclogite with small amounts of carbon (stored as diamond) and
water (in normally anyhdrous minerals) as it approaches
the Transition Zone and the top of the lower mantle. This
assumption is supported by low Na contents (1·5 wt %
Na2O) in majorite inclusions in diamonds (Collerson
et al., 2010) although at the pressures of majorite stability
this phase is capable of holding much higher amounts of
Na (Yasuda et al., 1994; Litasov & Ohtani, 2005).
Findings of various high-pressure minerals as inclusions
in natural diamonds (Harte, 2010) reinforce the results of
this study. Here we show that diamonds can crystallize
from Ca-rich carbonate melts (with various amount of
alkali components) that are produced by low-degree melting of carbonated MORB at all the pressures from 5 to
21GPa (Kiseeva et al., 2012; this study). These melts are
compositionally similar to those proposed as parental to
the Udachnaya-East kimberlites (Kamenetsky et al., 2004;
NUMBER 8
AUGUST 2013
Litasov et al., 2010; Sharygin et al., 2013) and other kimberlites worldwide (Kamenetsky et al., 2009), and can also be
involved in the carbonatite magmatism.
CONC LUSIONS
We have investigated experimentally the melting and
phase relations of two MORB eclogite compositions with
4·4% CO2, at temperatures of 1100^19008C and pressures
of 9^21GPa. The solidus temperatures are above the subduction geotherm but below the estimated mantle adiabat.
1580
(1) The main subsolidus mineral assemblage consists of
garnet, coesite or stishovite, clinopyroxene (9, 13 GPa)
and carbonate. Over the range of P^Tconditions studied, carbon-bearing phases include the following:
magnesite and aragonite or calcite^magnesite solid
solution (similar to dolomite composition) at 9 GPa;
magnesite and aragonite (GA1cc) or magnesite and
Na-rich aragonite (Volga-cc) at 13 GPa; magnesite
and Na-rich aragonite at 17 GPa; Na-rich aragonite
(GA1cc, Volga-cc) and Na-carbonate (Volga-cc) at
21GPa; diamond or graphite at 9^21GPa (GA1cc).
(2) Na-rich aragonite is an alternative to clinopyroxene as
a host for K and Na at pressures greater than 13 GPa.
(3) In the Volga-cc bulk composition, the solidus curve is
almost flat and falls between 1200 and 13008C over
the entire investigated pressure range. In the GA1cc
bulk composition, the solidus is located at similar
temperatures at 9^13 GPa, but lies at higher temperatures (1300^15008C) at 17^21GPa.
(4) The difference in solidi between the GA1cc and Volgacc bulk compositions is related to a change in Na compatibility between 13 and 17 GPa, owing to the disappearance of omphacitic clinopyroxene, resulting in
the formation of Na-bearing carbonate in the Volgacc, to carbonate reduction and diamond precipitation,
induced either by progressive Fe2þ^Fe3þ transition in
garnet with pressure or by influence of the diopside
component in the clinopyroxene, in the GA1cc bulk
composition.
(5) Low-degree melts in both compositions are alkalirich. The amount of alkalis in the melts increases significantly with pressure, and is buffered by the presence of clinopyroxene and K-hollandite in the system.
(6) Two melting pulses are proposed for subducted slabs
carrying carbonated eclogite in their upper sections.
The first melt pulse at 250^300 km depth, or
8^9 GPa, will produce K-rich carbonatite melts,
whereas the second melt pulse (near the top of the
Transition Zone, at 400^450 km depth, or
14^15 GPa) will produce very Na-rich carbonatite
melts. Some of the carbon will still survive in the
form of diamond and graphite.
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
(7) Because of their low viscosity, the resulting carbonatite melts are assumed to segregate from the main
eclogite body at depths above the Transition Zone,
allowing refractory carbon-bearing eclogite to be
stored in the Transition Zone or lower mantle. These
melts can be involved in generation of such magmas
as kimberlitic or carbonatitic
AC K N O W L E D G E M E N T S
We wish to thank Hugh O’Neill and Robert Rapp for valuable suggestions on improving this paper. We also thank
Raj Dasgupta and Oleg Safonov for constructive reviews.
The experimental work was performed during Internship
to K.K. as a part of the 21st Century Center-of-Excellence
program ‘Advanced Science and Technology Center for
the Dynamic Earth’ at Tohoku University. The authors
gratefully acknowledge Hidenori Terasaki for his help in
setting up experiments. Karsten Goemann (Central
Science Laboratory at University of Tasmania) is thanked
for assistance with electron microprobe analyses. Frank
Brink and Hua Chen (Centre for Advanced Microscopy
at ANU) assisted with the SEM analyses.
FUNDING
E.S.K. was funded by an ANU Postgraduate Scholarship.
The research was partly funded by an Australian
Research Council Discovery Grant to G.Y.
R EF ER ENC ES
Alt, J. C. & Teagle, D. A. H. (1999). The uptake of carbon during alteration of ocean crust. Geochimica et Cosmochimica Acta 63, 1527^1535.
Bebout, G. E. (1995). The impact of subduction-zone metamorphism
on mantle^ocean chemical cycling. Chemical Geology 126, 191^218.
Blundy, J. & Dalton, J. (2000). Experimental comparison of trace
element partitioning between clinopyroxene and melt in carbonate
and silicate systems, and implications for mantle metasomatism.
Contributions to Mineralogy and Petrology 139, 356^371.
Brenker, F. E., Vollmer, C., Vincze, L., Vekemans, B., Szymanski, A.,
Janssens, K., Szaloki, I., Nasdala, L., Joswig, W. & Kaminsky, F.
(2007). Carbonates from the lower part of transition zone or even
the lower mantle. Earth and Planetary Science Letters 260, 1^9.
Brey, G. P., Bulatov, V. K. & Girnis, A. V. (2011). Melting of K-rich carbonated peridotite at 6^10 GPa and the stability of K-phases in
the upper mantle. Chemical Geology 281, 333^342.
Buob, A., Luth, R. W., Schmidt, M. W. & Ulmer, P. (2006).
Experiments on CaCO3^MgCO3 solid solutions at high pressure
and temperature. American Mineralogist 91, 435^440.
Collerson, K. D., Williams, Q., Kamber, B. S., Omori, S., Arai, H. &
Ohtani, E. (2010). Majoritic garnet: a new approach to pressure estimation of shock events in meteorites and the encapsulation of sublithospheric inclusions in diamond. Geochimica et Cosmochimica Acta
74, 5939^5957.
Dasgupta, R. & Hirschmann, M. M. (2010). The deep carbon cycle
and melting in Earth’s interior. Earth and Planetary Science Letters
298, 1^13.
Dasgupta, R., Hirschmann, M. M. & Withers, A. C. (2004). Deep
global cycling of carbon constrained by the solidus of anhydrous,
carbonated eclogite under upper mantle conditions. Earth and
Planetary Science Letters 227, 73^85.
Dasgupta, R., Hirschmann, M. M. & Dellas, N. (2005). The effect of
bulk composition on the solidus of carbonated eclogite from partial
melting experiments at 3 GPa. Contributions to Mineralogy and
Petrology 149, 288^305.
Dorogokupets, P. I. & Dewaele, A. (2007). Equations of state of MgO,
Au, Pt, NaCl-B1, and NaCl-B2: Internally consistent high-temperature pressure scales. High Pressure Research 27, 431^446.
Dymshits, A. M., Bobrov, A. V., Bindi, L., Litvin, Y. A., Litasov, K.
D., Shatskiy, A. F. & Ohtani, E. (2013). Na-bearing majoritic
garnet in the system Na2MgSi5O12^Mg3Al2Si3O12 at 11^20 GPa:
phase relations, structural peculiarities and solid solutions.
Geochimica et Cosmochimica Acta 105, 1^13.
Gasparik, T. (1988). The stability of pyroxene in the transition zone.
Chemical Geology 70, 61.
Gerbode, C. & Dasgupta, R. (2010). Carbonate-fluxed melting of
MORB-like pyroxenite at 2·9 GPa and genesis of HIMU ocean
island basalts. Journal of Petrology 51, 2067^2088.
Ghosh, S., Ohtani, E., Litasov, K. D. & Terasaki, H. (2009). Solidus of
carbonated peridotite from 10 to 20 GPa and origin of magnesiocarbonatite melt in the Earth’s deep mantle. Chemical Geology 262,
17^28.
Grassi, D. & Schmidt, M. (2011a). Melting of carbonated pelites at 8^
13 GPa: generating K-rich carbonatites for mantle metasomatism.
Contributions to Mineralogy and Petrology 162, 169^191.
Grassi, D. & Schmidt, M. W. (2011b). The melting of carbonated pelites from 70 to 700 km depth. Journal of Petrology 52, 765^789.
Hammouda, T. (2003). High-pressure melting of carbonated eclogite
and experimental constraints on carbon recycling and storage in
the mantle. Earth and Planetary Science Letters 214, 357^368.
Harte, B. (2010). Diamond formation in the deep mantle: the record of
mineral inclusions and their distribution in relation to mantle dehydration zones. Mineralogical Magazine 74, 189^215.
Hirose, K. & Fei, Y. W. (2002). Subsolidus and melting phase relations
of basaltic composition in the uppermost lower mantle. Geochimica
et Cosmochimica Acta 66, 2099^2108.
Isshiki, M., Irifune, T., Hirose, K., Ono, S., Ohishi, Y., Watanuki, T.,
Nishibori, E., Takata, M. & Sakata, M. (2004). Stability of magnesite and its high-pressure form in the lowermost mantle. Nature
427, 60^63.
Kamenetsky, M. B., Sobolev, A. V., Kamenetsky, V. S., Maas, R.,
Danyushevsky, L. V., Thomas, R., Pokhilenko, N. P. & Sobolev, N.
V. (2004). Kimberlite melts rich in alkali chlorides and
carbonates: A potent metasomatic agent in the mantle. Geology 32,
845^848.
Kamenetsky, V. S., Kamenetsky, M. B., Weiss, Y., Navon, O.,
Nielsen, T. F. D. & Mernagh, T. P. (2009). How unique is the
Udachnaya-East kimberlite? Comparison with kimberlites from
the Slave Craton (Canada) and SW Greenland. Lithos 112,
334^346.
Kennedy, C. S. & Kennedy, G. C. (1976). Equilibrium boundary between graphite and diamond. Journal of Geophysical Research 81,
2467^2470.
Kerrick, D. M. & Connolly, J. A. D. (2001). Metamorphic devolatilization of subducted oceanic metabasalts: implications for seismicity,
arc magmatism and volatile recycling. Earth and Planetary Science
Letters 189, 19^29.
Keshav, S. & Gudfinnsson, G. H. (2010). Experimentally dictated stability of carbonated oceanic crust to moderately great depths in
the Earth: Results from the solidus determination in the system
1581
JOURNAL OF PETROLOGY
VOLUME 54
CaO^MgO^Al2O3^SiO2^CO2. Journal of Geophysical Researchç
Solid Earth 115, doi: 10.1029/2009JB006457.
Kiseeva, E. S., Yaxley, G. M., Hermann, J., Litasov, K. D.,
Rosenthal, A. & Kamenetsky, V. S. (2012). An experimental study
of carbonated eclogite at 3·5^5·5 GPaçimplications for silicate
and carbonate metasomatism in the cratonic mantle. Journal of
Petrology 53, 727^759.
Litasov, K. D. (2011). Physicochemical conditions for melting in the
Earth’s mantle containing a C^O^H fluid (from experimental
data). Russian Geology and Geophysics 52, 475^492.
Litasov, K. D. & Ohtani, E. (2009a). Phase relations in the peridotite^
carbonate^chloride system at 7·0^16·5 GPa and the role of chlorides
in the origin of kimberlite and diamond. Chemical Geology 262,29^41.
Litasov, K. D. & Ohtani, E. (2009b). Solidus and phase relations of
carbonated peridotite in the system CaO^Al2O3^MgO^SiO2^
Na2O^CO2 to the lower mantle depths. Physics of the Earth and
Planetary Interiors 177, 46^58.
Litasov, K. D. & Ohtani, E. (2005). Phase relations in hydrous MORB
at 18^28 GPa: implications for heterogeneity of the lower mantle.
Physics of the Earth and Planetary Interiors 150, 239^263.
Litasov, K. & Ohtani, E. (2010). The solidus of carbonated eclogite in
the system CaO^Al2O3^MgO^SiO2^Na2O^CO2 to 32 GPa and
carbonatite liquid in the deep mantle. Earth and Planetary Science
Letters 295, 115^126.
Litasov, K. D., Sharygin, I. S., Shatskiy, A. F., Ohtani, E. &
Pokhilenko, N. P. (2010). Experimental constraints on the role of
chloride in the origin and evolution of kimberlitic magma. Doklady
Earth Science 435, 1648^1653.
Litasov, K. D., Shatskiy, A. F., Ohtani, E. & Yaxley, G. M. (2013). The
solidus of alkaline carbonatite in the deep mantle. Geology 41, 79^82.
Luth, R. W. (1993). Diamonds, eclogites, and the oxidation state of the
Earth’s mantle. Science 261, 66^68.
Luth, R. W. (2001). Experimental determination of the reaction aragonite þ magnesite ¼ dolomite at 5 to 9 GPa. Contributions to
Mineralogy and Petrology 141, 222^232.
Martinez, I., Zhang, J. Z. & Reeder, R. J. (1996). In situ X-ray diffraction of aragonite and dolomite at high pressure and high temperature: evidence for dolomite breakdown to aragonite and
magnesite. American Mineralogist 81, 611^624.
Mckenzie, D. & Bickle, M. J. (1988). The volume and composition of
melt generated by extension of the lithosphere. Journal of Petrology
29, 625^679.
McKenzie, D. & O’Nions, R. K. (1991). Partial melt distributions from
inversion of rare-earth element concentrations. Journal of Petrology
32, 1021^1091.
Morlidge, M., Pawley, A. & Droop, G. (2006). Double carbonate breakdown reactions at high pressures: an experimental
study in the system CaO^MgO^FeO^MnO^CO2. Contributions to
Mineralogy and Petrology 152, 365^373.
Nishihara, Y., Shinmei, T. & Karato, S. (2006). Grain-growth kinetics
in wadsleyite: effects of chemical environment. Physics of the Earth
and Planetary Interiors 154, 30^43.
Okamoto, K. & Maruyama, S. (2004). The eclogite^garnetite transformation in the MORB þ H2O system. Physics of the Earth and
Planetary Interiors 146, 283^296.
Poli, S. & Schmidt, M. W. (1995). H2O transport and release in subduction zones^çexperimental constraints on basaltic and andesitic
systems. Journal of Geophysical ResearchçSolid Earth 100,
22299^22314.
Presnall, D. C., Dixon, S. A., Dixon, J. R., O’Donnell, T. H.,
Brenner, N. L., Schrock, R. L. & Dycus, D. W. (1978). Liquidus
phase relations on join diopside^forsterite^anorthite from 1
atm to 20 kbarçtheir bearing on generation and crystallization
NUMBER 8
AUGUST 2013
of basaltic magma. Contributions to Mineralogy and Petrology 66,
203^220.
Revenaugh, J. & Sipkin, S. A. (1994). Seismic evidence for
silicate melt atop the 410 km mantle discontinuity. Nature 369,
474^476.
Rohrbach, A. & Schmidt, M. W. (2011). Redox freezing and melting in
the Earth’s deep mantle resulting from carbon^iron redox coupling.
Nature 472, 209^212.
Sato, H., Endo, S., Sugiyama, M., Kikegawa, T., Shimomura, O. &
Kusaba, K. (1991). Baddeleyite-type high-pressure phase of TiO2.
Science 251, 786^788.
Sato, K. & Katsura, T. (2001). Experimental investigation on dolomite
dissociation into aragonite þ magnesite up to 8·5 GPa. Earth and
Planetary Science Letters 184, 529^534.
Sharygin, I. S., Litasov, K. D., Shatskiy, A. F., Golovin, A. V.,
Ohtani, E. & Pokhilenko, N. P. (2013). Melting experiments on kimberlite of Udachnaya-East pipe at 3^6·5 GPa and 90015008Q.
Doklady Earth Science 448, 200^205.
Simakov, S. K. (2006). Redox state of eclogites and peridotites from
sub-cratonic upper mantle and a connection with diamond genesis.
Contributions to Mineralogy and Petrology 151, 282^296.
Sleep, N. H. & Zahnle, K. (2001). Carbon dioxide cycling and implications for climate on ancient Earth. Journal of Geophysical Researchç
Planets 106, 1373^1399.
Sokolova, T. S., Dorogokupets, P. I. & Litasov, K. D. (2013).
Self-consistent pressure scales based on the equations of state for
ruby, diamond, MgO, B2-NaCl, as well as Au, Pt and other
metals to 4 Mbars and 3000 K. Russian Geology and Geophysics 54,
181^199.
Song, T. R. A., Helmberger, D. V. & Grand, S. P. (2004). Low-velocity
zone atop the 410-km seismic discontinuity in the northwestern
United States. Nature 427, 530^533.
Spandler, C., Yaxley, G., Green, D. H. & Rosenthal, A. (2008). Phase
relations and melting of anhydrous K-bearing eclogite from 1200
to 16008C and 3 to 5 GPa. Journal of Petrology 49, 771^795.
Spandler, C., Yaxley, G., Green, D. H. & Scott, D. (2010).
Experimental phase and melting relations of metapelite in the
upper mantle: implications for the petrogenesis of intraplate
magmas. Contributions to Mineralogy and Petrology 160, 569^589.
Staudigel, H. (2003). Hydrothermal alteration processes in the oceanic
crust. In: Holland, H. D. & Turekian, K. K. (eds) Treatise on
Geochemistry. Oxford: Elsevier^Pergamon, pp. 511^535.
Syracuse, E. M., van Keken, P. E. & Abers, G. A. (2010). The global
range of subduction zone thermal models. Physics of the Earth and
Planetary Interiors 183, 73^90.
Takafuji, N., Fujino, K., Nagai, T., Seto, Y. & Hamane, D. (2006).
Decarbonation reaction of magnesite in subducting slabs at the
lower mantle. Physics and Chemistry of Minerals 33, 651^654.
Tsuno, K. & Dasgupta, R. (2012). The effect of carbonates on near-solidus melting of pelite at 3 GPa: Relative efficiency of H2O and
CO2 subduction. Earth and Planetary Science Letters 319, 185^196.
Urakawa, S., Kondo, T., Igawa, N., Shimomura, O. & Ohno, H.
(1994). Synchrotron radiation study on the high-pressure and hightemperature phase relations of KAlSi3O8. Physics and Chemistry of
Minerals 21, 387^391.
van Keken, P. E., Kiefer, B. & Peacock, S. M. (2002). High-resolution
models of subduction zones: implications for mineral dehydration
reactions and the transport of water into the deep mantle.
Geochemistry, Geophysics, Geosystems 3, doi:10.1029/2001GC000256.
Wang, W. Y. & Takahashi, E. (1999). Subsolidus and melting experiments of a K-rich basaltic composition to 27 GPa: implication for
the behavior of potassium in the mantle. American Mineralogist 84,
357^361.
1582
KISEEVA et al.
CARBONATED ECLOGITE PHASE RELATIONS
Withers, A. C., Essene, E. J. & Zhang, Y. X. (2003). Rutile/TiO2 II
phase equilibria. Contributions to Mineralogy and Petrology 145,
199^204.
Yagi, A., Suzuki, T. & Akaogi, M. (1994). High-pressure transitions in
the system KAlSi3O8^NaAlSi3O8. Physics and Chemistry of Minerals
21, 12^17.
Yasuda, A., Fujii, T. & Kurita, K. (1994). Melting phase relations of an
anhydrous mid-ocean ridge basalt from 3 to 20 GPa: implications
for the behavior of subducted oceanic crust in the mantle. Journal
of Geophysical ResearchçSolid Earth 99, 9401^9414.
Yaxley, G. M. & Brey, G. P. (2004). Phase relations of carbonate-bearing eclogite assemblages from 2·5 to 5·5 GPa: implications for
petrogenesis of carbonatites. Contributions to Mineralogy and Petrology
146, 606^619.
Yaxley, G. M. & Green, D. H. (1994). Experimental demonstration
of refractory carbonate-bearing eclogite and siliceous melt in
the subduction regime. Earth and Planetary Science Letters 128,
313^325.
Zhang, Y. X. & Zindler, A. (1993). Distribution and evolution of
carbon and nitrogen in Earth. Earth and Planetary Science Letters
117, 331^345.
Zhao, D. P. & Ohtani, E. (2009). Deep slab subduction and dehydration and their geodynamic consequences: evidence from seismology
and mineral physics. Gondwana Research 16, 401^413.
1583