FTIR Spectrum of Phenocryst Olivine as an Indicator of Silica Saturation in Magmas

Abstract

Fourier Transform infrared (FTIR) absorption spectra of hydroxyl were measured on olivine phenocrysts from hydrous basaltic melts that originated in island-arc tectonic settings. The basaltic melts encom-pass a wide range of silica activities from orthopyroxene-saturated hypersthene-normative to nepheline-normative compositions. The intensities and wavenumber placement of hydroxyl absorption bands correlate with the degree of silica saturation of the parent melt from which the olivine crystallized. Olivines from silica-undersaturated nepheline-normative melts absorb IR radiation in the wavenumber range 3430–3590 cm-1 (Group 1). In contrast, olivines from orthopyroxene-saturated boninitic melts exhibit hydroxyl absorption bands in the wavenumber range 3285–3380 cm-1 (Group 2). Olivines crystallized at intermediate silica activities exhibit a combina-tion of the two groups of hydroxyl IR bands, where the proportion of Group 2 bands increases with increasing silica saturation of the parent melt. The positions of hydroxyl absorption peaks observed here for natural samples are consistent with previous measurements on experimentally annealed olivines. Thus protonation experiments can be employed to make spectroscopically dry olivine structures visible by IR, yielding information on the silica saturation of the parental magmas. Hydroxyl concentrations in the studied olivines were estim-ated to be 1–2 ppm, corresponding to an olivine–melt partition coefficient of �(1�.0 �+/- 0�.3) �x10�4.

Full text

FTIR Spectrum of Phenocryst Olivine as an
Indicator of Silica Saturation in Magmas
S. MATVEEV
1,2
*, M. PORTNYAGIN
3
, C. BALLHAUS
2
, R. BROOKER
4
AND C. A. GEIGER
5
1
DEPARTMENT OF EARTH AND ATMOSPHERIC SCIENCES, UNIVERSITY OF ALBERTA,
1–26 EARTH SCIENCES BUILDING, EDMONTON, ALTA., TG6 2E3, CANADA
2
INSTITUT FU
¨R MINERALOGIE, UNIVERSITA
¨TMU
¨NSTER, CORRENSSTR. 24, 48149 MU
¨NSTER, GERMANY
3
LEIBNIZ INSTITUTE FOR MARINE SCIENCES, DYNAMICS OF THE OCEAN FLOOR, WISCHHOFSTR. 1–3, 24148 KIEL,
GERMANY
4
DEPARTMENT OF EARTH SCIENCES, UNIVERSITY OF BRISTOL, QUEENS ROAD, BRISTOL BS8 1RJ, UK
5
INSTITUT FU
¨R GEOWISSENSCHAFTEN, UNIVERSITA
¨T KIEL, LUDEWIG-MEYN-STRAßE 10, 24118 KIEL, GERMANY
RECEIVED SEPTEMBER 15, 2003; ACCEPTED OCTOBER 7, 2004
ADVANCE ACCESS PUBLICATION DECEMBER 3, 2004
Fourier Transform infrared (FTIR) absorption spectra of hydroxyl
were measured on olivine phenocrysts from hydrous basaltic melts that
originated in island-arc tectonic settings. The basaltic melts encom-
pass a wide range of silica activities from orthopyroxene-saturated
hypersthene-normative to nepheline-normative compositions. The
intensities and wavenumber placement of hydroxyl absorption bands
correlate with the degree of silica saturation of the parent melt from
which the olivine crystallized. Olivines from silica-undersaturated
nepheline-normative melts absorb IR radiation in the wavenumber
range 3430–3590 cm
1
(Group 1). In contrast, olivines from
orthopyroxene-saturated boninitic melts exhibit hydroxyl absorption
bands in the wavenumber range 3285–3380 cm
1
(Group 2).
Olivines crystallized at intermediate silica activities exhibit a combina-
tion of the two groups of hydroxyl IR bands, where the proportion of
Group 2 bands increases with increasing silica saturation of the
parent melt. The positions of hydroxyl absorption peaks observed
here for natural samples are consistent with previous measurements on
experimentally annealed olivines. Thus protonation experiments can
be employed to make spectroscopically dry olivine structures visible by
IR, yielding information on the silica saturation of the parental
magmas. Hydroxyl concentrations in the studied olivines were estim-
ated to be 1–2 ppm, corresponding to an olivine–melt partition
coefficient of (1003) 10
4
.
KEY WORDS: nominally anhydrous minerals; olivine; water; mantle; silica
activity; melt inclusions
INTRODUCTION
Following the recognition that nominally anhydrous
rock-forming silicates can incorporate small amounts of
hydrogen (Beran, 1970; Wilkins & Sabine, 1973), much
work has been undertaken to document the hydroxyl
concentrations and possible hydroxyl substitution
mechanisms in olivine using Fourier Transform infrared
(FTIR) spectroscopy (e.g. Beran & Putnis, 1983; Aines &
Rossman, 1984; Miller et al., 1987; Mackwell &
Kohlstedt, 1990; Bai & Kohlstedt, 1993; Kohlstedt et al.,
1996; Kohn, 1996; Libowitzky & Rossman, 1997; Ingrin
& Skogby, 2000; Matveev et al., 2001; Lemaire et al.,
2004), and also using computational methods ( Wright &
Catlow, 1994; Braithwaite et al., 2003). FTIR spectro-
scopic studies of natural olivine show that hydroxyl
stretching bands generally occur in two wavenumber
regions, one between 3430 and 3630 cm
1
and the other
between 3285 and 3380 cm
1
. These are, following the
work of Bai & Kohlstedt (1993), referred to as Group 1
and Group 2 hydroxyl bands, respectively. In addition,
synthetic pure forsterite and rare natural samples
show lower-frequency hydroxyl bands ranging down to
3100 cm
1
(Miller et al., 1987; Demouchy & Mackwell,
2003; Lemaire et al., 2004).
Matveev et al. (2001) have shown experimentally
that the IR absorption of dissolved hydroxyl relates to
the silica activity (a
SiO
2
) at which the natural olivine
*Corresponding author. Present address: Department of Earth and
Atmospheric Sciences, University of Alberta, 1–26 Earth Sciences
Building, Edmonton, Alta., TG6 2E3, Canada. Telephone: þþ1 780
492 3191. Fax: þþ1 780 492 2030. E-mail: smatveev@ualberta.ca
#The Author 2004. Published by Oxford University Press. All
rights reserved. For Permissions, please email: journals.permissions@
oupjournals.org
JOURNAL OF PETROLOGY VOLUME 46 NUMBER 3 PAGES 603–614 2005 doi:10.1093/petrology/egh090

crystallized or finally equilibrated. When olivine is equilib-
rated with magnesiow€uustite (i.e. at low a
SiO
2
), the hydroxyl
stretching bands occur between 3430 and 3630 cm
1
(Group 1). On the other hand, when olivine is equilibrated
with orthopyroxene (i.e. at high a
SiO
2
), the hydroxyl bands
occur at lower energies between 3285 and 3380 cm
1
(Group 2). Hydroxyl in olivine could be associated with
the presence of cation vacancies whose concentrations
are largely controlled by a
SiO
2
(Stocker & Smyth, 1978;
Nakamura & Schmalzried, 1983), whereby with decreas-
ing a
SiO
2
the concentration of silicon vacancies increases
and the concentration of metal vacancies decreases.
Under this premise, Matveev et al. (2001) assigned
Group 1 absorption bands to hydroxyl groups associated
with silicon vacancies (i.e. a hydrogarnet-type substitu-
tion) and Group 2 absorption bands to hydroxyl asso-
ciated with octahedral M-site (Fe, Mg) vacancies. Thus
the possibility of locating hydrogen in silicon vacancies
might stabilize these defects and promote their formation
during olivine crystallization at low a
SiO
2
, even though the
concentration of silicon vacancies in the anhydrous oli-
vine structure might be regarded as insignificant (e.g.
Nakamura & Schmalzried, 1983; Mackwell et al., 1988).
These experimental results are supported by recent
computer simulations undertaken for the pure forsterite
(Mg
2
SiO
4
) crystal structure (Braithwaite et al., 2003). The
simulations were made to determine the energetics asso-
ciated with various possible hydroxyl substitutional
mechanisms, including both cation vacancies and inter-
stitials (nominally unoccupied crystallographic positions).
Braithwaite et al. concluded that hydroxyl groups should
occur in either vacant octahedral or tetrahedral sites so as
to produce neutral defect complexes, which are energe-
tically favorable relative to other substitution mechan-
isms. The simulations indicate that hydroxyl stretching
bands occur in two wavenumber regions. Their calcu-
lated spectra show that in the case where Si is replaced by
four H
þ
atoms, four hydroxyl stretching bands occur at
relatively high wavenumbers, whereas in the case where
two H
þ
atoms substitute for one Mg, two hydroxyl bands
occur at lower energies. The calculations agree with the
experimental observations of Lamaire et al. (2004), which
showed a significant effect of a
SiO
2
on the FTIR spectrum
of hydroxyl-bearing synthetic forsterite. These results are
in line with the experimental observations of Matveev
et al. (2001), although quantitative comparisons of
OH stretching frequencies measured on pure forsterite
with data obtained for natural samples are not possible
because of the effects of Fe and trace elements, such as Ti
(Berry et al., 2004), on the defect structure of olivine.
In this paper, we test the postulate of Matveev et al.
(2001) that the hydroxyl substitution mechanism could
serve as an indicator of the silica activity at which the
crystal structure of natural olivine equilibrates. To do this,
we measured the FTIR absorption spectra of hydroxyl in
olivine phenocrysts from a range of primitive mafic vol-
canic rocks. From bulk-rock compositions as well as the
composition of olivine melt inclusions, the parental melts
were all water-bearing and cover a wide range in a
SiO
2
.
SAMPLES
The basaltic lavas from which olivines were separated are
listed in Table 1. The eight samples are volcanic rocks
from subduction-related tectonic settings and include
lavas from (1) Troodos, Cyprus, (2) Avacha Volcano,
Kamchatka, Russia, and (3) Mount Mahimba, New
Georgia Archipelago, Solomon Islands. All the samples
are primitive in the sense that the melts from which
olivine phenocrysts crystallized could be in Fe–Mg
exchange equilibrium with typical mantle olivine (Fo
88–92
).
In the most silica-rich samples from Cyprus, olivine is
joined by orthopyroxene, and in the samples from
Avacha volcano and the Solomon Islands clinopyroxene
is the early liquidus mineral along with olivine.
Compositions of basaltic lavas and homogenized oli-
vine melt inclusions (e.g. Danyushevsky et al., 2002) are
plotted in Fig. 1 on projections from the olivine (a) and
diopside (b) apices onto the base of the ‘basalt tetrahe-
dron’. The molar concentrations of normative olivine
(Ol), diopside (Di), quartz (Qtz), jadeite ( Jd), Ca-Tscher-
mak pyroxene (CaTs), and leucite (Lc) were calculated
following procedures given by Falloon & Green (1987)
(Tables 2 and 3). To quantify the silica saturation degree
of the primary melts we use projections of plotted composi-
tions on the ( Jd þCaTs þLc)(Qtz) join parameter-
ized as the Qtz/(Qtz þJd þCaTs þLc) molar ratio,
subsequently referred to as the Silica Saturation Index
(SSI) (Tables 2 and 3). The SSI is independent of changes
in normative olivine and diopside concentrations, and
thus remains largely unchanged during magma fractio-
nation or during reaction of melt with host olivine. Thus
the SSI of melt inclusions or basaltic lavas adequately
reflects the silica-saturation degree of the primary melt.
Another advantage of SSI is that unlike a
SiO
2
it can be
easily calculated for magmas where olivine is not coexist-
ing with orthopyroxene or magnesiow€uustite (compare
Carmichael, 2002). A negative correlation between SSI
and TiO
2
concentration in both basaltic lavas and melt
inclusions is consistent with the origin of the Troodos
primary magmas from variably depleted sources as a
result of different degrees of melting (Cameron, 1985;
Sobolevet al., 1993;Falloon & Danyushevsky,2000) (Fig. 2).
Lavas from the Troodos ophiolite and
Mamonia complex, Cyprus
The Troodos ophiolite and disaggregated ophiolitic
blocks in the Mamonia complex in Cyprus are
tectonically exhumed fragments of Upper Cretaceous
604
JOURNAL OF PETROLOGY VOLUME 46 NUMBER 3 MARCH 2005

oceanic-like crust interpreted to have been formed in a
supra-subduction zone (Pearce, 1975; Cameron, 1985;
Rautenschlein et al., 1985). They represent the ‘birth’
and ‘youth’ stages of island-arc volcanism (Shervais,
2001). During the initial stages of subduction the astheno-
spheric mantle sources produced high-magnesian
hypersthene- to quartz-normative basaltic melts including
island-arc tholeiites and refractory high-CaO boninites
(Cameron, 1985; Sobolev et al., 1993). The presence of
water and hydroxyl in pillow-rim glasses and in melt
inclusions in phenocrysts suggests involvement of water
during mantle melting (e.g. Rautenschlein et al., 1985;
Sobolev & Chaussidon, 1996). In the samples studied,
bulk-rock MgO ranges from 22 to 327 wt %, and the
Fig. 1. Compositions of basaltic lavas (&) and average compositions of melt inclusions (*) shown on projections onto the base of the normative
‘basalt tetrahedron’ (Falloon & Green, 1987) from olivine (left diagram) and diopside (right diagram) apices. The outlined field is the entire
compositional range of melt inclusions. Relative proportions of ( Jd þCaTs þLc) and (Qtz) components define the Silica Saturation Index (SSI)
of magmas. Silica-undersaturated magmas with SSI <06 crystallize olivine exhibiting only the Group 1 hydroxyl absorption bands. Olivine from
orthopyroxene-saturated magmas with SSI >075 exhibits only the Group 2 hydroxyl bands. Shaded sector corresponds to the SSI from 060 to
075 where olivines exhibit both Group 1 and Group 2 absorption bands.
Table 1: Sample description
Sample Location Geochemical type Phenocryst mineralogy Fo range (mol %) Reference
MAM-25 Mavrokolimbos Dam,
Mamonia complex, Cyprus
High-Ca boninite Ol þCrSp þ(Opx
*
)87
.091.5 Portnyagin et al. (1996)
TRD-64 Kalavasos village,
Limassol Forest complex, Cyprus
High-Ca boninite Ol þOpx þCrSp 89.091.5 Sobolev et al. (1993)
TRD-41 Margi village, Troodos ophiolite,
Cyprus
Transitional to
high-Ca boninite
Ol þCrSp 89.093.5 Sobolev et al. (1993)
TRD-39 Pedieous River valley,
Troodos ophiolite, Cyprus
Transitional to
high-Ca boninite
Ol þCpx þCrSp 88.392.5 Sobolev et al. (1993)
TRD-75 Margi village, Troodos ophiolite,
Cyprus
Transitional to
high-Ca boninite
Ol þCrSp 89.291.2 Sobolev et al. (1993)
TRD-150 Analiondas village,
Troodos ophiolite, Cyprus
Low-K tholeiite Ol þCpx þCrSp 81.791.7 Portnyagin et al. (1997)
AV-2 Avacha volcano, Kamchatka, Russia Calc-alkaline basalt Ol þCpx þPl þCrSp 85.291.0 Portnyagin et al. (2004)
SS-1 Mt. Mahimba, Kolo Caldera,
New Georgia Island, Solomon Islands
Picrite Ol þCpx þCrSp 9092 Rohrbach (2003)
*
Opx found as inclusions in olivine.
605
MATVEEV et al. FTIR SPECTRUM OF OLIVINE

lavas contain up to 50 vol. % of 05–3 cm diameter
euhedral olivine phenocrysts. The unusually MgO-rich
samples were shown by Sobolev et al. (1993) to be olivine
cumulates in an evolved melt matrix.
Previously reported compositions of rocks and homo-
genized melt inclusions from forsteritic olivine pheno-
crysts (>Fo
88
) are summarized in Tables 2 and 3. Water
contents are based on ion-probe analysis of homogenized
melt inclusions and may be as high as 25 wt % in the
most silica-rich, refractory boninitic melts. Further details
concerning the petrography of the samples and the melt
inclusion data have been given by Sobolev et al. (1993),
Portnyagin et al. (1996, 1997), Sobolev & Chaussidon
(1996) and Portnyagin (1998).
Lava from Avacha volcano
Basaltic lava AV-2, from Avacha volcano in the
Kamchatka arc, with 15 wt % MgO (Tables 2 and
3) contains about 35 vol. % of large (up to 2 cm) pheno-
crysts of magnesian olivine (Fo
91–80
) and clinopyroxene
(Mg-number 92–73). This represents one of the most
primitive basalts of the Quaternary volcanic series of
the Kamchatka arc. Based on the compositions of homo-
genized melt inclusions in olivine, the parental magmas
were strongly silica undersaturated under crustal condi-
tions, and were probably derived from a refractory
asthenospheric mantle source that was refertilized prior
to, or during, partial melting by a subduction-related
component rich in H
2
O, CO
2
, and incompatible
trace elements. In contrast to the parental melt composi-
tion inferred from melt inclusions, the host lava is
hypersthene-normative and was interpreted to be a
mush of olivine and clinopyroxene crystals in an evolved
andesitic matrix (Portnyagin et al., 2004).
Lava from the Solomon Islands
Sample SS-1 is an olivine clinopyroxene-phyric lava
with 24 wt % MgO, derived from Mt. Mahimba, Kolo
Caldera, New Georgia Island. The sample is a cumulate
derived from a highly oxidized olivine–hypersthene-
normative parental melt with 132 wt % MgO and
489 wt % SiO
2
. Rohrbach (2003) identified two
Table 2: Average compositions of rocks
Sample MAM-25 TRD-64 TRD-41 TRD-39 TRD-75 TRD-150 AV-2 SS-1
wt %
SiO
2
45.66 42.81 44.64 42.13 41.72 44.93 50.89 48.90
TiO
2
0.16 0.12 0.26 0.18 0.23 0.33 0.52 0.61
Al
2
O
3
7.26 4.93 6.97 4.95 5.16 10.12 9.68 11.06
FeO 9.19 8.62 8.17 9.05 9.15 8.21 8.14 11.40
MnO 0.20
.15 0.15 0.17 0.16 0.16 0.16
MgO 26.48 32.52 28.34 31.44 32.73 22.00 16.10 13.2
CaO 5.15 3.68 5.01 3.35 2.61 8.09 11.56 9.75
Na
2
O0
.23 0.18 0.63 0.25 0.11 0.46 1.57 1.80
K
2
O0
.05 0.05 0.25 0.08 0.07 0.06 0.36 0.83
P
2
O
5
0.04 0.02 0.03 0.03 0.04 0.04 0.09
Cr
2
O
3
0.23 0.29 0.35 0.25 0.25 0.14
LOI 4.00 5.79 4.33 7.94 7.58 5.79 0.75
Total 98.65 99.14 99.11 99.80 99.79 100.30 99.82 97.35
Normative components
Jd 0.017 0.013 0.046 0.019 0.008 0.034 0.128 0.128
Lc 0.002 0.002 0.012 0.004 0.004 0.003 0.017 0.039
CTTs 0.005 0.003 0.007 0.005 0.007 0.009 0.015 0.017
CaTs 0.149 0.102 0.120 0.099 0.107 0.200 0.138 0.139
Di 0.057 0.048 0.077 0.038 0.000 0.122 0.297 0.228
Ol 0.582 0.707 0.600 0.707 0.742 0.466 0.239 0.317
Qtz 0.188 0.125 0.137 0.128 0.133 0.167 0.166 0.132
SSI 0.771 0.762 0.698 0.760 0.770 0.679 0.638 0.563
Data source references are given in Table 1. Normative components (molar fractions): Jd ¼NaAlSi
2
O
6
,Lc¼KAlSi
2
O
6
,
CTTs ¼CaTiAl
2
O
6
, CaTs ¼CaAl
2
SiO
6
,Di¼Ca(Mg,Fe,Mn)Si
2
O
6
,Ol¼(Mg,Fe,Mn)
3
Si
15
O
6
, Qtz ¼Si
3
O
6
. SSI (Silica
Saturation Index) ¼Qtz/(Qtz þCaTs þLc þJd) (see text for details). LOI, loss on ignition.
606
JOURNAL OF PETROLOGY VOLUME 46 NUMBER 3 MARCH 2005

generations of primitive olivine phenocrysts (Fo
90–92
),
one with 01 and the other with >03 wt % CaO, of
which only the high-CaO generation is considered to
have been in equilibrium with the CaO content of
the parent melt. The low-CaO olivines are probably
xenocrysts from the lithospheric upper mantle. Con-
sequently, FTIR spectra were collected from only the
high-CaO olivines.
PROTONATION EXPERIMENTS
To make the defect structure of anhydrous olivine
visible by IR spectroscopy, olivine can be protonated
experimentally (Bai & Kohlstedt, 1993). Unlike equilibra-
tion experiments in which the olivine defect structure is
largely controlled by buffered a
SiO
2
(e.g. Bai & Kohlstedt,
1993; Kohlstedt et al., 1996; Matveev et al., 2001), the
protonation experiments are designed to keep the origi-
nal olivine defect structure intact. Therefore, pressure–
temperature conditions and run durations must be such
that the relaxation time of cation point defect equilibra-
tion significantly exceeds the run time of an experiment.
To optimize experimental conditions, we have used the
diffusion coefficients for hydrogen and metal vacancies
reported by Kohlstedt & Mackwell (1998) and references
therein. We also assumed that the diffusion rate of silicon
vacancies does not exceed that of metal vacancies
Table 3: Representative compositions of melt inclusions in olivine
MAM-25 TRD-64 TRD-39 TRD-41 TRD-75 TRD-150 AV-2 SS-1
n: 9 14 15 60 28 103 15 1
wt %
SiO
2
53.12 0.62 52.98 0.58 52.49 0.50 52.30 0.52 52.00 0.46 51.42 0.17 49.26 0.87 n.d.
TiO
2
0.19 0.02 0.20 0.01 0.38 0.04 0.38 0.02 0.48 0.03 0.53 0.02 0.99 0.16 n.d.
Al
2
O
3
12.12 0.28 12.24 0.46 11.53 0.42 12.80 0.39 13.27 0.37 15.58 0.17 14.42 0.73 n.d.
FeO 5.80 0.22 6.18 0.28 7.34 0.32 7.04 0.28 6.73 0.23 6.20 0.11 6.20 0.20 n.d.
MnO 0.13 0.02 0.14 0.02 0.16 0.01 0.14 0.01 0.13 0.01 0.14 0.01 0.11 0.01 n.d.
MgO 14.66 0.80 15.17 0.79 14.13 1.08 14.72 1.31 13.21 0.91 9.97 0.30 11.59 1.09 n.d.
CaO 11.09 0.63 11.35 0.36 10.31 0.48 9.91 0.44 10.37 0.27 12.83 0.17 13.81 0.94 n.d.
Na
2
O0
.55 0.15 0.42 0.03 1.30 0.05 1.50 0.10 1.77 0.10 1.40 0.04 2.65 0.17 n.d.
K
2
O0
.09 0.03 0.08 0.01 0.16 0.02 0.16 0.02 0.19 0.04 0.11 0.01 0.68 0.07 n.d.
H
2
O1
.73 0.03 1.56 0.17 1.85 0.19 1.71 0.09 1.63 0.06 1.36 0.05 <0.1 n.d.
Total 99.48 0.47 100.32 0.40 99.65 0.59 100.67 0.33 99.79 0.29 99.53 0.16 99.70 0.27
Normative components
Jd 0.038 0.010 0.029 0.002 0.104 0.004 0.091 0.007 0.124 0.006 0.098 0.003 0.209 0.015
Lc 0.004 0.001 0.003 0.001 0.007 0.001 0.007 0.001 0.009 0.002 0.005 0.001 0.030 0.002
CTTs 0.005 0.001 0.005 0.001 0.010 0.001 0.010 0.001 0.013 0.001 0.014 0.001 0.025 0.002
CaTs 0.230 0.004 0.235 0.008 0.204 0.009 0.186 0.007 0.202 0.006 0.265 0.003 0.183 0.010
Di 0.191 0.025 0.192 0.012 0.165 0.019 0.204 0.013 0.185 0.008 0.216 0.005 0.289 0.029
Ol 0.257 0.018 0.267 0.019 0.278 0.023 0.262 0.030 0.244 0.019 0.170 0.004 0.169 0.013
Qtz 0.275 0.012 0.267 0.009 0.232 0.011 0.239 0.010 0.224 0.008 0.232 0.003 0.096 0.019
SSI 0.751 0.007 0.749 0.005 0.687 0.006 0.715 0.006 0.667 0.006 0.654 0.003 0.371 0.048
Selected trace element compositions (ppm)
Cr 2400 3080 2700 2500 2860 1450 1240
Ni 581 1435 1243 1005 931 581 360
Co 76 103 105 93 108 76 48
Zn 59 64 61 66 66 59
Zr14 4251416 7 35
Errors calculated at 95% confidence level for ninclusions analysed. Data source references are given in Table 1. Normative
components (molar fractions): Jd ¼NaAlSi
2
O
6
,Lc¼KAlSi
2
O
6
, CTTs ¼CaTiAl
2
O
6
, CaTs ¼CaAl
2
SiO
6
,Di¼
Ca(Mg,Fe,Mn)Si
2
O
6
,Ol¼(Mg,Fe,Mn)
3
Si
1.5
O
6
, Qtz ¼Si
3
O
6
. SSI (Silica Saturation Index) ¼Qtz/(Qtz þCaTs þLc þJd)
(see text for details). n.d., not determined.
607
MATVEEV et al. FTIR SPECTRUM OF OLIVINE

(Mackwell et al., 1988; Matveev et al., 2001). Experiments
were performed in a piston-cylinder apparatus and held
at 1000C and 2 GPa for 4 h. At these run conditions, we
calculate that hydrogen may penetrate and protonate a
05–1 mm diameter olivine grain (the typical grain size of
our olivine separates), with the defect structure remaining
intact.
For the protonation experiments we have chosen
olivines from Solomon Islands basalt (SS-1) in which
the primary hydroxyl content was found to be below
the detection limit of the FTIR analysis. To ensure
that the olivine defect structure did not change in the
course of protonation we performed two a
SiO
2
-buffered
experiments: in one, olivine crystals were embedded in
periclase powder; in the other, an orthopyroxene powder
was used. Dry experimental charge ( 150 mg) along
with 25 ml of water were welded in 5 mm outer dia-
meter platinum capsules. Only fluid-saturated experi-
ments that released water upon recovery are reported in
this study.
ANALYTICAL PROCEDURE
Olivine grains were separated from the host rocks or
removed from the protonation experiments and polished
on both sides to give crystal platelets of 100–600 mm
thickness. The studied crystals were anhedral, so their
crystallographic orientation could not be visually deter-
mined prior to polishing. To have a range of crystal
orientations available, at least 10 crystals per sample
were separated from each rock sample and prepared for
FTIR analysis.
For the FTIR measurements, the crystal platelets were
placed in an IR microscope attached to a Nicolet 800
spectrometer. Spectra were collected in transmission
mode in regions free of cracks or inclusions using an
unpolarized beam. The size of the measuring spot was
defined by choosing either 75 or 100 mm diameter aper-
tures. The IR spectra were collected in the wavenumber
range from 600 to 6000 cm
1
with a resolution of 4 cm
1
.
The IR microscope was kept inside a plastic box and
purged with dried nitrogen gas [see Jamtveit et al. (2001)
for technical details]. The IR spectra were measured after
holding the crystals in a dry nitrogen gas atmosphere for
at least 12 h, resulting in low spectral noise in the wave-
numbers above 3600 cm
1
, where IR stretching of water
vapour occurs (e.g. Bernath, 2002).
ANALYTICAL RESULTS
The FTIR absorption spectra were examined in two
wavenumber ranges: from 3100 to 3700 cm
1
, where
hydroxyl bands absorb; and from 1600 to 2100 cm
1
,
where second-order Si–O overtones occur. The latter
are used to estimate the crystal orientation.
The FTIR spectra exhibit hydroxyl absorption bands
consistent with the Group 1 and Group 2 classification of
Bai & Kohlstedt (1993). Group 1 bands were observed
between 3430 and 3590 cm
1
. The rims of experimen-
tally protonated crystals show a further absorption band
at 3615 cm
1
, which is probably due to high f
H
2
O
in the
experiments. The Group 2 hydroxyl bands occur
between 3285 and 3380 cm
1
. At room temperature,
these bands have FWHM (full width at half maximum)
of 20–40 cm
1
.
Second-order Si–O overtones were used to deduce the
crystallographic orientation of the studied olivine grains
following Jamtveit et al. (2001), but it should be noted that
the directions [100] and [010] were swapped as they
appear to have been wrongly assigned for the commonly
accepted olivine unit-cell settings, where b>c>a.For
several olivine grains, spectroscopically estimated orien-
tations were confirmed by single-crystal X-ray diffracto-
metry (XRD; Fig. 4; Table 4). Below we compare FTIR
spectra that were measured with the IR beam parallel to
[010]. Such spectra are characterized by the strongest
absorption of unpolarized IR radiation in both the
Group 1 and Group 2 wavenumber regions (e.g.
Kohlstedt et al., 1996).
Olivines from Cyprus and Kamchatka
IR absorption spectra of olivine from the Cyprus lavas
are summarized in Fig. 3a. Olivines from orthopyroxene-
bearing samples MAM-25 and TRD-64 absorb in the
Group 2 wavenumber range (spectra 1–4), whereas
absorption in the Group 1 range is at the limit of the
Fig. 2. Correlation between TiO
2
content and SSI of basaltic lavas (&)
and melt inclusions (*). Error bars reflect compositional variations
of melt inclusions in each sample and are calculated at the 95%
confidence level.
608
JOURNAL OF PETROLOGY VOLUME 46 NUMBER 3 MARCH 2005

FTIR detection. Olivines from the orthopyroxene-
free Cyprus samples show both the Group 2 and the
Group 1 bands (spectra 5–8). The proportion of
Group 2 bands decreases progressively in the sample
sequence TRD-39, TRD-41–TRD-75–TRD-150, so
that olivine from the tholeiitic sample TRD-150 shows
mainly Group 1 hydroxyl bands (spectrum 8). The FTIR
spectra of the olivine phenocrysts from the Avacha lava
AV-2 contain only Group 1 bands (Fig. 3b).
Placement and intensity of hydroxyl bands in the FTIR
spectra of olivines correlate well with SSI calculated for
basaltic lavas and melt inclusions (Fig. 1). SSI is highest
for the samples in which olivine exhibits Group 2 hydro-
xyl bands and lowest for the samples in which olivine
exhibits mainly Group 1 bands. To better illustrate the
influence of silica saturation on IR absorption, we have
integrated the Group 1 absorption bands from 3466 to
3620 cm
1
and the Group 2 bands from 3260 to
3412 cm
1
, and then determined the absorption intensity
ratio A
int, ratio
¼A
int, Group 1
/(A
int, Group 1
þA
int, Group 2
),
where A
int
denotes integrated absorption coefficients.
The ratio systematically increases with decreasing SSI
of basaltic lavas (Fig. 5a) and melt inclusions (Fig. 5b
and c). Lesser scatter observed for melt inclusions
Table 4: Crystallographic orientation of olivine polished
planes calculated from XRD measurements and data
from Inorganic Crystal Structure Database (ICSD) for
forsteritic olivine (83-1535)
Sample hkl
TRD-64-7 2.26 1.11 0.15
TRD-150-111 5 10 1
TRD-150-9 0 1 0
TRD-150-112 0.31 4.68 0.95
TRD-150-5 0.09 1.33 1.24
TRD-150-6 0.16 0.95 1.34
TRD-64-6 0.29 1.78 2.69
TRD-75-6 0.97 1.36 2.51
We assume unit cell parameters a¼5.990, b¼10.226, c¼
4.769 and Pmnb symmetry group. Calculations were per-
formed using POWD-12þþ software (1997) (Smith et al.,
1982). Samples are listed in the same order as in Fig. 4.
Samples with orientation of polished planes approaching
(100), (010) and (001) are shown in bold in the table, and the
corresponding spectra are highlighted in Fig. 4.
Fig. 3. FTIR absorption spectra of olivine phenocrysts from (a)
Troodos lavas, Cyprus, and (b) a sample from Avacha volcano,
Kamchatka, Russia: 1, 2, MAM-25; 3, 4, TRD-64; 5, TRD-39; 6,
TRD-41; 7, TRD-75; 8, TRD-150; 9, 10, AV-2. Absorption coeffi-
cients are normalized to 1 cm sample thickness and offset to stack the
spectra such that the degree of silica saturation of the parent melt
decreases from top to bottom (see text). The spectra were measured
with an IR beam parallel to [010].
Fig. 4. FTIR spectra of Si–O overtones measured on olivine grains
whose orientation was confirmed by XRD. Bold lines are the spectra
for which the polished planes are close to the principal crystallographic
orientations. The spectra are stacked to illustrate the gradual change in
the overtone spectrum with changing orientation.
609
MATVEEV et al. FTIR SPECTRUM OF OLIVINE

illustrates that melts trapped in olivine preserve the pri-
mary composition better than the host lavas. Based on
the FTIR results obtained for the Cyprus samples, the
transition from the Group 1 dominating spectrum to
the Group 2 dominating spectrum occurs within the
hypersthene-normative field of parental magmas
and in the relatively narrow range of SSI between
06 and 075.
Water solubility in olivines from the Cyprus lavas was
estimated from the Group 1 and Group 2 integrated
absorption coefficients using the calibration of
Libowitzky & Rossman (1997). The weighted mean
wavenumber for the Group 1 hydroxyl bands was located
at 3500 cm
1
and for Group 2 at 3330 cm
1
. Because
FTIR spectra were measured with the unpolarized beam
parallel to [010], an additional orientation factor g¼05
was applied (Mackwell & Kohlstedt, 1990; Lemaire et al.,
2004). Water concentrations are given in Table 5, and for
olivines from the Cyprus lavas fall in the concentration
range from 1 to 2 ppm. However, it should be noted
that the poorly constrained value of orientation factor
(g03–05; Paterson, 1982; Mackwell & Kohlstedt,
1990), spectral noise and unknown olivine matrix correc-
tion of the calibration make these calculations rather
qualitative. Using water concentrations measured on
homogenized melt inclusions in olivine from the Cyprus
lavas as a proxy for water content in the primary melt, an
olivine–melt partition coefficient for H
2
O can be estim-
ated as K
d
(1003) 10
4
. Because the baselines
were positioned at the flanks of each group of absorption
bands, the broad plateau of underlying absorption
commonly attributed to small amounts of molecular
water in submicroscopic inclusions (Miller et al., 1987;
Matveev et al., 2001) did not contribute to the total
calculated water content. Thus water contents reported
here are somewhat lower than those that would result
from integration of the entire hydroxyl absorption area
(e.g. Matveev et al., 2001). The obtained partition coeffi-
cient is in good agreement with the coefficient from Hirth
& Kohlstedt (1996) calculated for shallow mantle and
crustal conditions (<300 MPa), and consistent with a
shallow depth of olivine crystallization.
Hydroxyl in olivine from AV-2 absorbs IR only in
the Group 1 wavenumber range, which is consistent
with the highly silica-undersaturated compositions of
the melt inclusions (Fig. 1b). Low water concentrations
Fig. 5. Integrated absorption coefficient ratio A
int, ratio
¼A
int, Group 1
/
(A
int, Group 1
þA
int, Group 2
) vs SSI of rocks (a) and olivine melt
inclusions (b, c). Vertical error bars are calculated at the 95% con-
fidence level and reflect the compositional range of melt inclusions in
each sample. Error bars in terms of the A
int, ratio
represent the 95%
confidence interval calculated from integrated Group 1 and Group 2
absorbances. These errors include contributions from imperfect IR
beam orientation, spectral noise, and natural variations in A
int, ratio
.
Nepheline-normative and quartz–hypersthene-normative fields are cal-
culated according to the CIPW scheme. Linear regression (dashed line)
calculated for rocks and melt inclusion from Cyprus samples illustrates
the SSI range in which the Group 2 absorption bands replace the
Group 1 hydroxyl absorption bands. The respective compositional
variations are shown in Fig. 1.
610
JOURNAL OF PETROLOGY VOLUME 46 NUMBER 3 MARCH 2005

in AV-2 melt inclusions (<01 wt %) might be indi-
cative of higher pressures of olivine crystallization
compared with that of Cyprus olivines, and thus a
somewhat higher olivine–melt H
2
O partition coeffi-
cient (Hirth & Kohlstedt, 1996). Alternatively, water
could have been lost from the melt inclusions during
ascent, as a result of decrepitation (Portnyagin et al.,
2004).
Experimentally protonated olivines from
Solomon Islands picrites
FTIR spectra of experimentally protonated olivine from
sample SS-1 (Mt. Mahimba, Solomon Islands) were
measured from core to rim with 100 mm step incre-
ments using a beam diameter of 100 mm (Fig. 6). Both
experiments produced olivine grains with increased water
contents. The resulting spectra show a 100 mm wide
rim whose defect structure corresponds to the a
SiO
2
of
the buffer used (the lowest spectra in Fig. 6a and b). In
contrast, the cores have lower water contents, but all
show the spectra features related to low silica activity
regardless of the buffer used. Hydroxyl incorporation in
the rim may involve dissolution–precipitation mechan-
isms (e.g. Matveev et al., 2001), whereas core protonation
occurs probably according to reduction–oxidation
reactions as described by Kohlstedt et al. (1997). Good
correlation between FTIR spectra measured on olivine
cores and the low a
SiO
2
implied by the SSI calculated for
SS-1 basalt (Tables 2 and 3, Fig. 5a) implies that the
protonated structure of initially anhydrous olivine can
be used to estimate the degree of silica saturation of the
parent melts.
DISCUSSION
The experiments of Matveev et al. (2001), and the result-
ing FTIR spectra of hydroxyl-bearing mantle olivines,
allow clear discrimination between the effects of a
SiO
2
and other important parameters such as pressure, tem-
perature, water and oxygen fugacities, and olivine major
and trace element compositions. Olivine grains separated
from the same sample at equal P,T,a
H
2
O
, and f
O
2
exhibited Group 1 OH bands when experimentally equi-
librated with magnesiow€uustite (low a
SiO
2
) and Group 2
OH bands when equilibrated with orthopyroxene (high
a
SiO
2
). The IR data on olivines from hydrous basaltic
melts obtained in this study support the experimental
results of Matveev et al. (2001), confirming that
natural hydroxyl speciation in olivine is also largely
controlled by a
SiO
2
.
In natural samples, the influence of f
O
2
and trace ele-
ment composition on the FTIR spectrum of olivine may
Table 5: Analytical results
No. of
grains (points)
A
int, ratio
H
2
O (ppm)
MAM-25 4(4) <0.05 1.14 0.18
TRD-64 4(4) <0.05 1.45 0.11
TRD-41 3(4) 0.29 0.06 1.84 0.11
TRD-39 5(4) 0.29 0.04 1.86 0.08
TRD-75 3(2) 0.57 0.05 1.56 0.08
TRD-150 3(3) 0.64 0.07 2.08 0.51
AV-2 2(2) 10
.46 0.1
SS-1 3(6) 1 b.d.l.
Integrated spectra were measured with the IR beam
direction parallel to the [010] crystallographic axis. Errors
are calculated at the 95% confidence level from data
measured on the reported number of grains and points.
b.d.l., below detection limit.
Fig. 6. FTIR spectra of olivine phenocrysts from the Solomon Islands
picrite SS-1, experimentally protonated in periclase (a) and enstatite (b)
matrices. Absorbances are normalized to 1 cm sample thickness. The
lowermost spectra in (a) and (b) are measured on olivine rims, which
equilibrated with the respective a
SiO
2
buffer. All other spectra are from
cores that preserved their pre-experiment defect structures but became
protonated during the experiment. These spectra are taken along
traverses with 100 mm spot intervals.
611
MATVEEV et al. FTIR SPECTRUM OF OLIVINE

often be hard to isolate from the effect of a
SiO
2
and grain
orientation inconsistencies. As f
O
2
controls concentration
of Fe
3þ
in olivine, it may also affect the capacity of olivine
to store hydrogen during natural or experimental proto-
nation (Kohlstedt et al., 1996). In the FTIR spectra of
natural olivines f
O
2
appears to affect the intensities of
individual OH bands (e.g. Matsyuk & Langer, 2004,
and references therein), but because ferric iron is likely
to occur in complexes where it substitutes in neighbour-
ing tetrahedral and octahedral positions (Nakamura &
Schmalzried, 1983) it is unlikely that f
O
2
will significantly
affect the A
int, ratio
.
The effect of trace impurities and the associated extrin-
sic defects in olivine is yet more cryptic. Relative concen-
trations of trace elements were estimated from the melt
inclusion compositions (Table 3) assuming similar
olivine–melt partition coefficients for the variety of the
studied rocks. The A
int, ratio
systematically changes only
with the Ti content of the melt inclusions and thus the
likely Ti content of the olivine (Nikogosian & Sobolev,
1997; Canil & Fedortchouk, 2001). However, the experi-
mental data of Matveev et al., (2001) suggest that the
spectra features assigned to high and low a
SiO
2
can be
reproduced in olivines with identical Ti contents and are,
therefore, not primarily controlled by this parameter.
Thus the apparent correlation between olivine FTIR
spectrum and Ti content is caused by decreasing TiO
2
concentration with increasing a
SiO
2
of the olivine parent
melt (Fig. 2). Nevertheless, the effect of Ti on the olivine
hydroxyl speciation is significant; the data of Berry et al.
(2004) suggest that Ti in olivine strongly affects the exact
position of Group 1 peaks.
Another important effect is f
H
2
O
(P,T,a
H
2
O
), which
affects the solubility of hydroxyl in the olivine structure.
At higher f
H
2
O
not only do the intensities of both Group 1
and Group 2 bands increase, but also the number of
resolvable OH bands, particularly at relatively higher
frequencies (compare the rim and core spectra in Fig. 4;
Kohlstedt et al., 1996; Matveev, 1997; Matveev et al.,
2001; Matsyuk & Langer, 2004). However, there is
no indication that f
H
2
O
notably affects the A
int, ratio
(Matveev, 1997; Matveev et al., 2001).
Considering a
SiO
2
as a key variable controlling the fre-
quency of OH IR absorption in olivine, we suggest that
the FTIR spectrum of olivine can be used to deduce the
a
SiO
2
at crystallization or final equilibration. The good cor-
relation between IR spectra and the composition of melt
inclusions or host lavas implies that the defect structure of
the studied olivines has survived changes in pressure–
temperature and even a
SiO
2
conditions during ascent.
CONCLUSIONS
The FTIR spectra of olivine phenocrysts in basaltic lavas
correlate well with the degree of silica saturation of the
parent melt and therefore a
SiO
2
. Liquidus olivines crystal-
lized from nepheline-normative basaltic melts have OH
bands between 3430 and 3590 cm
1
(Group 1). Olivines
that coexist at magmatic temperatures with orthopyrox-
ene have OH bands in the wavenumber range from 3285
to 3380 cm
1
(Group 2). Olivines from orthopyroxene-
undersaturated hypersthene-normative basaltic melts
exhibit both groups of hydroxyl absorption bands, with
the proportion of Group 2 bands increasing with increas-
ing SSI. The compositions of melt inclusions correlate
better with the FTIR spectra of olivine than the com-
position of the host basaltic lava, and thus more
accurately preserve information on the silica saturation
of primary melts.
FTIR results from this study are consistent with the
experimental results of Matveev et al. (2001), implying
that natural hydroxyl occurrence in olivine is similar to
that imposed on natural olivine in high-pressure experi-
ments. Therefore spectroscopically ‘dry’ olivine defect
structures that have lost hydrogen during magma ascent
can be re-protonated experimentally to reveal the a
SiO
2
at
their last vacancy equilibration. Hydrogen diffusion and
reduction of ferric iron in the olivine structure to OH and
ferrous iron are relatively rapid, such that the original
defect structure may survive a short experimental run
time, and may be made visible by subsequent IR spectro-
scopy to reveal the original a
SiO
2
conditions.
ACKNOWLEDGEMENTS
We thank the lapidary workshop at M€uunster University
for sample preparation. Financial support by the DFG
(Deutsche Forschungsgemeinschaft) through grants Ba
964/16-1 and Ge 659/11-1 (to C.B. and C.G.), the
European Commission IHP Programme grant, which
allowed IR analyses at the University of Bristol (to
S.M.), as well as BMBF (Bundesministerium f€uur Bildung
und Forschung) funded KOMEX-2 project and RFBR
(Russian Foundation for Basic Research) through grant
03-0564629 (to M.P.) are gratefully acknowledged. Ear-
lier discussions with A. Sobolev (MPI f€uur Chemie, Mainz)
were invaluable for structuring the study. We also thank
R. W. Luth, T. Chacko (University of Alberta) and
J. Harris (University of Glasgow) for their comments,
which helped to improve the manuscript. We thank
J. Loens (M€uunster University) for performing single-
grain XRD analyses.
REFERENCES
Aines, R. D. & Rossman, G. R. (1984). Water in minerals? A peak in
the infrared. Journal of Geophysical Research 89, 4059–4071.
Bai, Q. & Kohlstedt, D. L. (1993). Effects of chemical environment on
the solubility and incorporation mechanism for hydrogen in olivine.
Physics and Chemistry of Minerals 19, 460–471.
612
JOURNAL OF PETROLOGY VOLUME 46 NUMBER 3 MARCH 2005

Beran, A. (1970). U
¨ber (OH)-Gruppen in Olivin. (The OH-groups in
olivine.) Anzeiger der
€
OO
sterreichischen Akademie der Wissenschaften,
Mathematisch Naturwissenschaftliche Klasse 106, 73–74.
Beran, A. & Putnis, A. (1983). A model of the OH positions in olivine,
derived from infrared-spectroscopic investigations. Physics and
Chemistry of Minerals 9, 57–60.
Bernath, P. F. (2002). The spectroscopy of water vapour: experiment,
theory and applications. Physical Chemistry and Chemical Physics 4,
1501–1509.
Berry, A., Hermann, H. & O’Neill, H. (2004). The water site in mantle
olivine. Geochimica et Cosmochimica Acta 68, A36.
Braithwaite, J. S., Wright, K. & Catlow, C. R. A. (2003). A theoretical
study of the energetics and IR frequencies of hydroxyl defects in
forsterite. Journal of Geophysical Research 108, ECV 1–ECV 9.
Cameron, W. E. (1985). Petrology and origin of primitive lavas from
the Troodos ophiolite, Cyprus. Contributions to Mineralogy and Petrology
89, 239–255.
Canil, D. & Fedortchouk, Y. (2001). Olivine–liquid partitioning of
vanadium and other trace elements; with applications to modern
and ancient picrites. Canadian Mineralogist 39, 319–330.
Carmichael, I. S. E. (2002). The andesite aqueduct; perspectives on the
evolution of intermediate magmatism in west–central (105–99
degrees W ) Mexico. Contributions to Mineralogy and Petrology 143,
641–663.
Danyushevsky, L., McNeill, A. W. & Sobolev, A. V. (2002).
Experimental and petrological studies of melt inclusions in
phenocrysts from mantle-derived magmas: an overview of techni-
ques, advantages and complications. Chemical Geology 183, 5–24.
Demouchy, S. & Mackwell, S. (2003). Water diffusion in synthetic iron-
free forsterite. Physics and Chemistry of Minerals 30, 486–494.
Falloon, T. J. & Danyushevsky, L. V. (2000). Melting of refractory
mantle at 15, 2 and 25 GPa under anhydrous and H
2
O-
undersaturated conditions; implications for the petrogenesis of
high-Ca boninites and the influence of subduction components on
mantle melting. Journal of Petrology 41, 257–283.
Falloon, T. J. & Green, D. H. (1987). Anhydrous partial melting of
MORB pyrolite and other peridotite compositions at 10 kbar:
implication for the origin of primitive MORB glasses. Mineralogy and
Petrology 37, 181–219.
Hirth, G. & Kohlstedt, D. L. (1996). Water in the oceanic upper
mantle: implications for rheology, melt extraction and the evolution
of the lithosphere. Earth and Planetary Science Letters 144, 93–108.
Ingrin, J. & Skogby, H. (2000). Hydrogen in nominally anhydrous
upper-mantle minerals; concentrations levels and implications.
European Journal of Minerals 12, 543–570.
Jamtveit, B., Brooker, R., Brooks, K., Larsen, L. M. & Pedersen, T.
(2001). The water content of olivines from the North
Atlantic Volcanic Province. Earth and Planetary Science Letters 186,
401–415.
Kohlstedt, D. L. & Mackwell, S. J. (1998). Diffusion of hydrogen and
intrinsic point defects in olivine. Zeitschrift f€uur Physikalische Chemie—
International Journal of Research in Physical Chemistry and Chemical Physics
2, 147–162.
Kohlstedt, D. L., Keppler, H. & Rubie, D. C. (1996). Solubility of
water in the alpha, beta and gamma phases of (Mg,Fe)
2
SiO
4
.
Contributions to Mineralogy and Petrology 123, 345–357.
Kohn, S. C. (1996). Solubility of H
2
O in nominally anhydrous
mantle minerals using H-1 MAS NMR. American Mineralogist 81,
1523–1526.
Lemaire, C., Kohn, S. C. & Brooker, R. A. (2004). The effect of silica
activity on the incorporation mechanisms of water in synthetic
forsterite: a polarised infrared spectroscopic study. Contributions to
Mineralogy and Petrology 147, 48–57.
Libowitzky, E. & Rossman, G. R. (1997). An IR absorption calibration
for water in minerals. American Mineralogist 82, 1111–1115.
Mackwell, S. J. & Kohlstedt, D. L. (1990). Diffusion of hydrogen in
olivine: implications for water in the mantle. Journal of Geophysical
Research 95, 5079–5088.
Mackwell, S. J., Dimos, D. & Kohlstedt, D. L. (1988). Transient creep
of olivine: point-defect relaxation times. Philosophical Magazine A 57,
779–789.
Matsyuk, S. S. & Langer, K. (2004). Hydroxyl in olivines from mantle
xenoliths in kimberlites of the Siberian platform. Contributions to
Mineralogy and Petrology 147, 413–437.
Matveev, S. (1997). CHO fluids at upper mantle conditions. Ph.D.
thesis, Universit€aat zu K€ooln, Cologne.
Matveev, S., O’Neill, H. S. C., Ballhaus, C., Taylor, W. R. &
Green, D. H. (2001). Effect of silica activity on OH

IR spectra of
olivine: implications for low-a
SiO
2
mantle metasomatism. Journal of
Petrology 42, 721–729.
Miller, G. H., Rossman, G. R. & Harlow, G. E. (1987). The natural
occurrence of hydroxide in olivine. Physics and Chemistry of Minerals 14,
461–472.
Nakamura, A. & Schmalzried, H. (1983). On the nonstoichiometry
and point defects of olivine. Physics and Chemistry of Minerals 10,
27–37.
Nikogosian, I. K. & Sobolev, A. V. (1997). Ion-microprobe analysis of
melt inclusions in olivine: experience in estimating the olivine–melt
partition coefficients of trace elements. Geochemistry International 35,
119–126.
Paterson, M. (1982). The determination of hydroxyl by infrared
absorption in quartz, silicate glasses and similar materials. Bulletin de
Mineeralogie 105, 20–29.
Pearce, J. A. (1975). Basalt geochemistry used to investigate past
tectonic environments on Cyprus. Tectonophysics 25, 41–67.
Portnyagin, M. V. (1998). Origin of mantle derived magmas in supra-
subduction zones: a case study of the Troodos Ophiolite (Cyprus).
Ph.D. thesis, Vernadski Institute, Moscow, 230 pp. (in Russian).
Portnyagin, M. V., Magakyan, R. & Schminke, H. U. (1996).
Geochemical variability of boninite magmas; evidence from
magmatic inclusions in high magnesian olivine from lavas of
southwestern Cyprus. Petrologiya 4, 250–265.
Portnyagin, M. V., Danyushevsky, L. V. & Kamenetsky, V. S. (1997).
Coexistence of two distinct mantle sources during formation of
ophiolites: a case study of primitive pillow-lavas from the lowest part
of the volcanic section of the Troodos Ophiolite, Cyprus.
Contributions to Mineralogy and Petrology 128, 287–301.
Portnyagin, M. V., Plechov, P. Yu., Matveev, S. V. & Mironov, N. L.
(2004). Petrology of the ‘avachites’— high-magnesian basalts from
Avachinsky volcano (Kamchatka): II. Melt inclusions in olivine.
Petrology (in press).
Rautenschlein, M., Jenner, G. A., Hertogen, J., Hofmann, A. W.,
Kerrich, R., Schmincke, H. U. & White, W. M. (1985). Isotopic and
trace element composition of volcanic glasses from the Akaki
Canyon, Cyprus; implications for the origin of the Troodos
Ophiolite. Earth and Planetary Science Letters 75, 369–383.
Rohrbach, A. (2003). Petrologische Untersuchungen an Pikriten der
Salomonen-Inseln. Diploma thesis, Westf€aalische Wilhelms
Universit€aat, M€uunster, 61 pp.
Shervais, J. W. (2001). Birth, death, and resurrection: the life cycle of
suprasubduction zone ophiolites. Geochemistry, Geophysics, Geosystems,
45, paper number 2000GC000080.
Smith, D. K., Nichols, M. C. & Zolensky, M. E. (1982). POWD10, a
FORTRAN IV program for calculating X-ray powder diffraction
patterns—Version 10. University Park: Pennsylvania State
University, 103 pp.
613
MATVEEV et al. FTIR SPECTRUM OF OLIVINE

Sobolev, A. V. & Chaussidon, M. (1996). H
2
O concentrations in
primary melts from supra-subduction zones and mid-ocean ridges:
implications for H
2
O storage and recycling in the mantle. Earth and
Planetary Science Letters 137, 45–55.
Sobolev, A. V., Portnyagin, M. V., Dmitriyev, L. V., Tsameryan, O. P.,
Danyushevskiy, L. V., Kononkova, N. N., Shimizu, N. & Robinson, P.
(1993). Petrology of ultramafic magmas and associated rocks of the
Troodos Massif, Cyprus. Petrologiya 1, 379–412 (in Russian).
Stocker, R. L. & Smyth, D. M. (1978). Effect of enstatite activity and
oxygen partial pressure on the point-defect chemistry of olivine.
Physics of the Earth and Planetary Interiors 16, 146–156.
Wilkins, R. W. T. & Sabine, W. (1973). Water content of some
nominally anhydrous silicates. American Mineralogist 58, 508–516.
Wright, K. & Catlow, C. R. A. (1994). A computer simulation study
of (OH) defects in olivine. Physics and Chemistry of Minerals 20,
515–518.
614
JOURNAL OF PETROLOGY VOLUME 46 NUMBER 3 MARCH 2005