i An update to this article is included at the end

Earth and Planetary Science Letters 288 (2009) 291–300

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e p s l

The effect of water on the electrical conductivity of olivine aggregates and its
implications for the electrical structure of the upper mantle
Takashi Yoshino ⁎, Takuya Matsuzaki, Anton Shatskiy, Tomoo Katsura
Institute for Study of the Earth's Interior, Okayama University, Misasa, Tottori 682-0193, Japan

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

Article history: The electrical conductivity of San Carlos olivine aggregate of various water content was measured at a
Received 14 March 2009 pressure of 10 GPa in a Kawai-type multi-anvil apparatus. Conductivity measurements were performed on
Received in revised form 16 September 2009 two sets of samples to determine the effect on conductivity of water in olivine: 1) a hydrogen-doped sample
Accepted 19 September 2009
and 2) a hydrogen-undoped sample. To minimize water escape from the hydrogen-doped samples, the
Available online 12 October 2009
conductivity measurement was carried out below 1000 K. Three conduction mechanisms were identified
Editor: L. Stixrude from the Arrhenian behavior of the undoped samples, which include a small amount of water. A change in
the activation enthalpy indicated that the dominant conduction mechanism changed from proton
Keywords: conduction to small polaron conduction with increasing temperature. At temperatures above 1700 K, the
electrical conductivity activation enthalpy exceeds 2 eV suggesting that the dominant mechanism of charge transport would be
olivine ionic conduction. The conductivity increased with increasing water content. The activation enthalpy for
upper mantle proton conduction tends to decrease slightly with increasing water content. The activation enthalpy
water
determined for each run had similar values (~ 0.9 eV). Taking the water concentration dependence of
activation enthalpy into account for proton conduction, all data were fitted to the electrical conductivity
formula σ=σ0Iexp[−EI/kT]+σ0Hexp[−EH/kT]+σ0PCWexp[−(E0−αC1/3 W )/kT], where σ0 represents a pre-
exponential term, CW is the water content in weight percent, E is the activation enthalpy, E0 is the activation
enthalpy for proton conduction at very low water concentration, α is the geometrical factor, k is the
Boltzmann constant, T is absolute temperature and subscripts I, H and P denote ionic, hopping (small
polaron) and proton conductions, respectively. The conductivity jump at the 410 km discontinuity (olivine–
wadsleyite transition) is much smaller than that previously predicted. Since the contribution of proton
conduction to the bulk electrical conductivity decreases with increasing temperature the high conductivity
anomaly at the top of the asthenosphere cannot be explained by olivine hydration.
© 2009 Elsevier B.V. All rights reserved.

1. Introduction
Nominally anhydrous minerals of the Earth's mantle (olivine and
its high-pressure polymorphs as well as pyroxenes and garnet) can
store water as the hydroxyl group together with intrinsic point
defects in their structure (e.g., Kohlstedt et al. 1996; Bolfan-Casanova
et al., 2000). Even if the water concentration is very low the presence
of hydrogen in nominally anhydrous minerals significantly affects a
variety of physical and chemical properties such as plastic deforma-
tion, melting temperature and seismic velocity (e.g., Karato et al.,
1986; Karato, 1990; Hirth and Kohlstedt 1996; Mei and Kohlstedt
2000; Jacobsen et al., 2004). Therefore, estimating the water content
of the Earth's mantle is essential for understanding the dynamics and
geochemical evolution of the Earth.
⁎ Corresponding author.
E-mail address: tyoshino@misasa.okayama-u.ac.jp (T. Yoshino).
Since olivine is the most abundant mineral phase in most models of
the upper mantle the effect of hydrogen incorporation into olivine on its
physical properties has been substantially investigated. Although the
water content of natural olivine is generally 100 ppm by weight or less
(Bell and Rothman, 1992), hydrous olivine synthesized at 12 GPa and
1373 K contains a considerable amount of water, 1510 ppm by weight
(Kohlstedt et al., 1996). Results from two recent experiments
demonstrated that the maximum solubility of H2O in olivine exceeds
0.5 wt.% at conditions close to the 410 km discontinuity (8900 ppmw;
Smyth et al., 2006 and 6250 ppmw; Litasov et al., 2007). However, these
hydration experiments only provided an upper boundary for the actual
water content of the upper mantle. To assess the water content of the
upper mantle a determination of physical properties as a function of
water content should be compared with geophysical observations.
Electrical conductivity can be used to quantify the water content of
the Earth's mantle because it is thought to be sensitive to small
amounts of water in nominally anhydrous minerals (Karato, 1990).
Recent in situ laboratory measurements of electrical conductivity for
hydrous olivine containing various amounts of water under high-

0012-821X/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsl.2009.09.032
292 T. Yoshino et al. / Earth and Planetary Science Letters 288 (2009) 291–300
pressure conditions have been carried out to estimate the water
content in the Earth's upper mantle (Yoshino et al., 2006; Wang et al.,
2006). Yoshino et al. (2006) reported that although the presence of
water greatly enhances the electrical conductivity of olivine this effect
is not large enough to explain the high conductivity at the top of the
asthenosphere as observed from the East Pacific Rise (Evans et al.,
2005). In contrast, Wang et al. (2006) reported that hydrous olivine
showed much higher conductivity and a larger temperature depen-
dence than those given by Yoshino et al. (2006). Wang et al. (2006)
concluded that the hydrous olivine is responsible for the high con-
ductivity at the top of the asthenosphere.
Wang et al. (2006) measured conductivity of hydrous olivine
aggregates at relatively high temperatures (up to 1273 K). However,
conductivity measurements of hydrated minerals at high temperatures
should lead to incorrect measurements because at those high
temperatures small polaron (hopping) conduction contributes to the
bulk conductivity and also because dehydration is unavoidable due to
fast hydrogen diffusion at high temperatures (N1000 K). In contrast,
Yoshino et al. (2006) measured electrical conductivity at low tempera-
tures (b1000 K) in which the hydrogen diffusion rate is sufficiently
slow. If the insulation resistance is high enough to measure sample
conductivity at such low temperatures this method is useful to obtain
relatively accurate conductivity values and activation enthalpy for
proton conduction. By developing this technique, Yoshino et al. (2008)
investigated the effect of water on the conductivity of polycrystalline
wadsleyite and ringwoodite, which are high-pressure polymorphs of
olivine. They successfully separated contributions of hopping from
proton conduction and reported lower conductivity values than had
been previously reported (Huang et al., 2005).
Yoshino et al. (2006) did not determine the electrical conductivity
of hydrous olivine as a function of water content. Estimation of the
bulk conductivity of hydrous olivine aggregate using single crystal
data requires some assumptions. In addition, measurements were
conducted using a Ni–NiO buffer which has an oxygen fugacity that is
too high for the deep upper mantle. To directly estimate water content
in the upper mantle from electromagnetic observations we deter-
mined electrical conductivities of a series of olivine aggregates with
various amounts of water at a pressure of 10 GPa in a Kawai-type
multi-anvil apparatus. For the hydrogen-doped olivine aggregates
conductivity measurements were performed at relatively low tem-
peratures of up to 1000 K to minimize water loss. To distinguish
between proton and small polaron conduction mechanisms, conduc-
tivities of hydrogen-undoped olivine were measured at up to 2000 K.
By combining the present experimental results with the geophysically
determined conductivity–depth profile of the upper mantle, we
evaluate the water content in the upper mantle and the influence of
water on the lithosphere–asthenosphere boundary beneath the
Archean lithospheric root of the upper mantle.
2. Experimental methods
Starting materials were prepared from San Carlos olivine crystals
with a composition of (Mg0.91Fe0.09)SiO4. The initial water content of
the crystals was less than 0.005 wt.%. The optically inclusion-free
crystals were ground with an agate mortar in acetone to obtain a fine
powder (b1 µm). For the hydrogen-undoped runs, the olivine powder
with a trace amount of orthopyroxene powder to control the silica
activity was used to measure the electrical conductivity directly. To
synthesize the hydrogen-doped olivine, the olivine powders were
mixed with a talc and brucite mixture (1.4:1 by weight) with 50 µm
thick Mo foil and welded onto Pt capsules with 1.8 mm inner
diameters. This weight ratio yields water plus approximately a 9:1
ratio of olivine to enstatite after dehydration and reaction and buffers
silica activity. Mixtures of olivine and talc and brucite with different
mixing ratios were prepared to obtain samples containing various
amount of water (H2O = 1 and 2 wt.%). The mixtures were kept at
10 GPa and at temperatures ranging from 1373 to 1623 K for a few
hours. After the run the presence of H2O was confirmed by bubbling
water and gas during the process of breaking the capsule. The
recovered sample consisted of olivine with a trace amount of enstatite
was observed by micro-focused X-ray diffractometer and scanning
electron microscope. The Mg number of the hydrogen-doped sample
is 92.5, which is slightly higher than that of the starting olivine
powder (Mg# = 91). A disk with a thickness of 1.0 mm and a diameter
of 1.5 mm for the conductivity measurement was cored using an
ultrasonic drilling machine to preclude the reaction zone between the
Mo foil and the sample.
Experiments were performed using a 5000-ton Kawai-type multi-
anvil apparatus. Pressure was generated by eight tungsten carbide
cubes with a 6 mm truncation edge length. A disk sample of hydrous
olivine was placed in a MgO single crystal sleeve. The use of a MgO
single crystal without a grain boundary ensures high background
resistance and enables the measurement of electrical conductivity at
low temperatures. Molybdenum disk electrodes with a diameter of
1.5 mm were placed in contact with a sample so that oxygen fugacity
was controlled on the Mo/MoO2 buffer, which is similar to the iron–
wüstite buffer (Xu et al., 1998). Two sets of WRe3–WRe25 thermo-
couples were mechanically connected to each Mo electrode on the
sample and were insulated from the LaCrO3 furnace by Al2O3 and MgO
insulators. The assembly consisted of a Cr-doped MgO pressure
medium with a 14 mm octahedral edge length ZrO2 thermal insulator
and a cylindrical LaCrO3 furnace. The cell design for the conductivity
measurement is shown in Fig. 1.
A reference resistance (106 Ω) was connected to the sample in
series and a sinusoidal signal with an amplitude of 1 V and a frequency
0.1 Hz was applied to the circuit. The sample and reference voltage
were measured simultaneously. The sample impedance was obtained
from the reference resistance and the complex ratio of voltage on the
sample to that on the resistance. The sample resistance was calculated
by assuming a simple parallel circuit of resistance and capacitance
although the effect of capacitance is very small. Sample conductivity
was calculated from the sample resistance and dimensions.
In the low frequency (0.1 Hz) measurements described above the
sample impedance could be overestimated because of contributions
from the electrochemical reaction at the sample-electrode interface and
from the chemical potential gradient induced by the diffusion of protons
in an electrical potential gradient. Hence, impedance spectroscopy was
also conducted as well as a low frequency measurement in two separate
runs to confirm the suitability of our measurement technique.
Impedance spectroscopic measurements were carried out using a
Solartron 1260 impedance Gain-Phase Analyzer combined with a
Solartron 1296 interface, which renders it possible to measure a very
high impedance material (up to 1014 Ω). The cell assembly was
essentially the same as those for the low frequency measurement.
Complex impedances were obtained at frequencies ranging from 1 MHz
to 0.1 or to 0.01 Hz. An equivalent circuit was composed of a sample
resistance and capacitance in parallel and was fit to the experimental
data for the determination of sample resistance. The applied voltage is
fundamentally 1 V except for where the sample has high resistance.
Sample water content was determined by Fourier-transform
infrared (FTIR) spectroscopy with unpolarized light using about 0.5-
mm thick samples in the region of ~ 100 × 100 µm. A Paterson calibration
was used to calculate the water content from infrared absorption
(Paterson, 1982). It has recently been claimed that the Paterson
calibration underestimates the hydrogen concentration in olivine by a
factor of 3–3.5 (Bell et al., 2003; Hirschmann et al., 2005). Although our
results could overestimate the effect of water on the electrical
conductivity of olivine we used values obtained from the Paterson
calibration as the water content because unpolarized light was used for
the FTIR analysis. We did not detect any special variation of the
hydrogen concentration for any sample larger than experimental errors,
nor detect a change in the shape of the absorption peak.
 T. Yoshino et al. / Earth and Planetary Science Letters 288 (2009) 291–300 293

Fig. 1. Schematic cross-section of high-pressure cell assembly for conductivity measurements in the Kawai-type multi-anvil press.

After the conductivity measurement, the recovered samples were

confirmed to be olivine by X-ray diffraction. A reaction between the Mo
electrode and the samples was not observed by scanning electron
microscopy. FTIR spectra showed the presence of water for each sample.
The hydrogen-undoped samples that experienced temperatures higher
than 1700 K contained measurable amounts of water. This water is
considered to come from the surrounding pressure medium at high
temperatures as was the case for single crystal olivine (Yoshino et al.,
2006). The FTIR spectra of olivine with a certain amount of water
(N0.01 wt.%) displayed two strong absorption bands at 3567 and
3611 cm− 1 (Fig. 2), which is identical to that observed in previous
studies for polycrystalline olivine by non-polarized analysis (e.g.,
Kohlstedt et al., 1996). In contrast, no change in water content during
the conductivity measurement was observed for those hydrogen-doped
samples that experienced temperatures less than 1000 K (Table 1).

3. Experimental results

Fig. 3 shows complex impedance spectra acquired for the hydrogen-

undoped olivine sample at 10 GPa and at temperatures ranging from
1000 to 1500 K. With increasing temperature the radius of the
impedance arc decreases. In the frequency range used for complex
impedance spectroscopy, impedance spectrum shows a near complete
semicircular pattern in the complex plane due to grain interior
conduction. Measurement for the hydrogen-doped olivine sample
(Fig. 3b) shows that the first arc shape is semicircular at high frequencies, Fig. 2. Unpolarized FTIR spectrum of olivine aggregates. Thick and thin lines denote FTIR
but the arc has a short tail toward the high real impedance at frequencies spectra before and after conductivity measurement, respectively. (a) The sample with the
lower than 10 Hz. These behaviors might represent contributions from maximum hydrogen concentration (5K1118). Note that the spectrum shows a higher
an electrode reaction. In any case, the impedance components such as a background after the conductivity measurement. (b) The hydrogen-doped sample (5K1221).
Note that there is no change of water content in each sample beyond errors, nor change of the
flow of neutral hydrogen-bearing molecules down the chemical absorption peak shape. (c) The spectrum of the hydrogen-undoped sample (5K1223), which
potential gradient generated by electrode reactions are negligible contains a small amount of water after the conductivity measurement. (d) The spectrum of a
compared with the grain interior conduction defined by the first arc. sample with the minimum water content (5K1055).
294 T. Yoshino et al. / Earth and Planetary Science Letters 288 (2009) 291–300

Table 1
Summary of runs.

Run no. Samplea H2O H2O T logσ H*. Remarks

wt.%b wt.%c (K) (S/m) (eV)

H/106Si H/106Si

5K1055 SC 0.005(0) 800–1400 0.83(7) 1.11(0)

80(3) 1400–1750 3.34(8) 1.77(2)
1750–2000 4.73(19) 2.25(7)
5K1126 SC 0.008(1) 700–1200 0.82(19) 1.10(3)
130(31) 1350–1650 2.44(10) 1.47(3)
1700–2000 4.47(10) 2.14(4)
5K1081 SC 0.009(4) 1550–1750 3.08(10) 1.61(3)
150(49)
d
5K1223 SC 0.012(6) 550–1350 0.87(6) 1.06(1)
200(87) 1400–1800 2.44(19) 1.51(6)
1K484 SC 0.022(3) 673–973 1.06(8) 0.98(1)
340(40)
5K1061 SCTB 0.042(5) 0.040(4) 600–800 0.91(21) 0.88(3)
690(80) 660(60)
5K1079 SCTB n.d. 0.066(3) 500–1000 0.93(16) 0.87(0)
1100(60)
5K1112 SCTB 0.166(17) 0.l59(42) 500–900 1.31(10) 0.88(1)
2700(270) 2600(700)
5K1122 SCTB n.d. 0.160(7) 550–900 2.44(12) 1.03(2)
2600(110)
5K1221d SCTB 0.179(12) 500–850 1.29(11) 0.85(1)
2900(320)
5K1118 SCTB 0.289(26) Dehydration
4700(430)
a
Starting materials used for the conductivity measurement. SC: San Carlos olivine powder with tiny amount of orthopyroxene. SCTB: the pre-synthesized hydrous olivine
aggregates.
b
Water content (in weight %) of synthesized hydrous olivine aggregate before the conductivity measurement experiment, determined by FTIR spectroscopy. Parentheses denote
pne standard deviation for five analyses.
c
Water content (in wt.%) after the conductivity measurement.
d
The conductivity were determined by impedance spectroscopic analyses.

When the resistances defined by both methods were plotted in the same
Arrhenius diagram, it was difficult to identify the lines separately.
In the first cycle, the sample was heated to 2000 K and 800–1000 K
for the hydrogen-undoped and -doped samples, respectively. After
the second cycle, heating and cooling were repeated until reproduc-
ibility was confirmed. During each cycle above 500 K, the temperature
was changed in 50 K increments and electrical conductivity was
measured at each temperature step. For the hydrogen-doped sample,
two typical examples of conductivity measurements are shown in
Fig. 4. In one case (Fig. 4a), the logarithmic conductivity linearly
increases and decreases with decreasing and increasing reciprocal
temperature, respectively. The obtained conductivity values were
reproducible after the first cooling. FTIR spectroscopy suggests that
the water content in the olivine was the same before and after the
measurement. In another case, in which the sample had the maximum
water content (~0.3 wt. %) in this study (Fig. 4b), the logarithmic
conductivity also increased and decreased with decreasing and
increasing reciprocal temperature, respectively, up to 900 K. How-
ever, once the sample was heated to 950 K the values of conductivity
increased by a factor of 2. After it was heated to 1000 K the
conductivity showed a much smaller temperature dependence than
before. This small temperature dependence is similar to that of the
conductivity of ionic water (Roberts, 2002). The FTIR spectrum of this
sample shows a higher background compared with that before the
conductivity measurement (Fig. 2). Therefore, the doped hydrogen
was released from the olivine in these runs. This observation suggests
that we cannot measure the conductivity of a hydrous mineral if grain
boundary water is present. Samples that were not repeatable are
excluded in the following discussion.
Fig. 5 shows the electrical conductivity of an olivine aggregate with
various amounts of water as a function of reciprocal temperature. The
absolute conductivity values increase with increasing water content.
The conductivity values determined from impedance spectroscopy
agree with those obtained from the low frequency conductivity
measurement. It suggests that contributions from the electrochemical
reaction at the sample-electrode interface and from the chemical
potential gradient induced by the diffusion of protons in an electrical
potential gradient are negligibly small in the investigated frequency
range compared with the sample impedance when the data were
plotted in an Arrhenius diagram. Therefore, the conductivity data
obtained from the different methods were treated equally in the
following data analysis.
All samples examined in this study behave as semiconductors.
Conductivity (σ)–temperature (T) relationships were expressed by
the Arrhenian formula:


H
σ = σ0 exp
ð1Þ
kT
where σ0 is a pre-exponential factor, E is the activation enthalpy, k is
the Boltzmann constant and T is absolute temperature. For the
hydrogen-undoped samples Arrhenius formula can be divided into
three regimes; high, intermediate and low temperature regimes. In
contrast, for the hydrogen-doped samples, a single activation
enthalpy can be defined in the investigated temperature range (up
to 1000 K). The pre-exponential term and activation enthalpy for each
temperature regime are summarized in Table 1.
The activation enthalpy of the hydrogen-undoped samples
increases from 1.1 to 2.2 eV as temperature increases (Fig. 6). Three
straight portions (activation enthalpy) in the Arrhenius plot are
recognizable at different temperature ranges. Below 1300 K, the
activation enthalpy is around 1.1 eV. In the temperature range
between 1300 and 1750 K, corresponding to the typical physical
conditions of the upper mantle, activation enthalpy increases to
~ 1.6 eV. A high activation enthalpy exceeding 2 eV appears above
1750 K. The temperature at which the activation enthalpy changes
increases with decreasing water content. The effect of water content
becomes smaller in the high temperature region compared with the
 T. Yoshino et al. / Earth and Planetary Science Letters 288 (2009) 291–300
Fig. 3. Complex impedance spectra showing the semicircular pattern of the real vs.
imaginary components of complex impedance at frequencies ranging from 1 MHz to
0.1 Hz at the temperatures indicated. (a) Impedance spectra of the hydrogen-undoped
olivine sample measured at 10 GPa and temperatures ranging from 1000 to 1500 K.
(b) Impedance spectra of the hydrogen-doped olivine sample measured at 10 GPa and
temperatures ranging from 500 to 800 K.
relatively dry samples. The pre-exponential terms determined at high
temperatures are higher than those determined at low temperatures
(Table 1). For the hydrogen-doped samples the activation enthalpy is
almost constant (0.9 eV) in the investigated temperature region (up
to 900 K). These values are similar to those determined from hydrous
single crystal olivine (0.73–0.98 eV: Yoshino et al., 2006). The pre-
exponential terms determined generally increase with increasing
water content (Table 1).
4. Discussion
4.1. Charge transport mechanisms
This study demonstrates that the electrical conductivity of olivine
aggregate shows a strong water-content dependence. At all tempera-
295
Fig. 4. Electrical conductivity of hydrogen-doped olivine aggregate as a function of
reciprocal temperature. Lines and dashed lines indicate heating and cooling cycles,
respectively. (a) Electrical conductivity of samples with 0.16 wt.% H2O (5K1122). Note
that the trace of 2nd heating–cooling cycle is consistent that of the 1st heating–cooling
cycle. (b) Electrical conductivity of samples with the maximum water content
(5K1118). Conductivity value was initially high (10− 3 S/m) and increased each cycle.
After heating at 1000 K, conductivity values were nearly constant along the 3rd cooling,
suggesting a presence of grain boundary water.
tures the conductivity of olivine with higher water content has a
larger value than those with lower water content. For the hydrogen-
undoped samples three conduction mechanisms can be identified.
Electrical conductivity (σ) of hydrous iron-bearing silicate minerals
could be expressed by the summation of three charge transport
mechanisms;
σ = σi + σh + σp ð2Þ
where subscripts i, h and p denote ionic, hopping (small polaron) and
proton conduction mechanisms. Ionic conduction is characterized by
the migration of magnesium-site vacancies. Relatively high activation
enthalpy (N2 eV) observed at high temperatures (N1750 K) could be
296 T. Yoshino et al. / Earth and Planetary Science Letters 288 (2009) 291–300

derived from an intrinsic mechanism such as ionic conduction, which

occurs at higher temperatures through the creation of cation
vacancies. In fact, a change in conduction mechanism for olivine
from small polaron to ionic conduction at around 1600 K and at
atmospheric pressure has been proposed by Schock et al. (1989).
In the intermediate temperature range between 1300 and 1700 K,
transportation of small polarons (electron-hole hopping between Fe2+
and Fe3+) has been considered the dominant conduction mechanism
for olivine and its high-pressure polymorphs (Schock et al., 1989; Xu
et al., 1998; Yoshino et al., 2006; 2008). The activation enthalpies
determined from each data point by Eq. (1) ranged from 1.47 to 1.77 eV.
These values are slightly higher than those for the small polaron
conduction of olivine (1.35–1.47 eV), wadsleyite (1.49 eV) and ring-
woodite (1.37 eV) with a composition of (Mg0.91,Fe0.09)2SiO4 and as
determined from measurements under high-pressure conditions (Xu
et al., 1998; Yoshino et al., 2006; 2008). For small polaron conduction,
charge transfer occurs by electron-hole hopping between ferric and
ferrous ions (Iyenger and Alcock, 1970). This reaction can be expressed
using the Kröger–Vink notation (Kröger and Vink, 1956) as

•
FeMe + h = Fe•Me
x
Below 1300 K, the activation enthalpy ranges from 0.85 to 1.10 eV.
An increase in conductivity with water content suggests that proton
conduction is the dominant conduction mechanism in this temper-
ature region. A relatively lower activation enthalpy is consistent with
previous reports of proton conduction (Yoshino et al., 2006; 2008).
For proton conduction, charge transfer occurs by proton hopping
amongst point defects. The pre-exponential factor for proton
conduction increases with increasing water content. Dependence of
the water content on electrical conductivity should follow the
equation because electrical conductivity depends on the number
(N) of electric charge carriers per unit volume;
σ = Nzeμ
where z is the charge number, e is the charge of electron and μ is
mobility. On the other hand, the activation enthalpy for proton
conduction tends to decrease with increasing water content as was
observed for wadsleyite and ringwoodite (Yoshino et al., 2008). If the
average distance of the neighboring proton is isotropic the activation
enthalpy is expected to be proportional to the cubic root of the
hydrogen concentration. This relationship can be approximated by an
equation for n-type semiconductors (Debye and Conwell, 1954),
ð3Þ
Fig. 5. Electrical conductivity of olivine with various amounts of water as a function of
reciprocal temperature. The symbols indicate raw data for each sample with different
water contents. Open and closed symbols represent results for the hydrogen-undoped and
-doped samples, respectively. Circle and cross symbols denote results from low frequency
measurement (0.1 Hz) and impedance spectroscopy (1 MHz to 0.1 Hz), respectively. Gray
thick lines indicate results of each crystallographic axis for hydrous single crystal olivine
with 0.02 wt.% H2O under Ni–NiO buffer (Yoshino et al., 2006). Parentheses denote each
crystallographic axis. Abbreviations: XSD00: conductivity data of San Carlos olivine
aggregates at 10 GPa under Mo–MoO2 buffer from Xu et al. (2000); WMXK06:
conductivity data of hydrous olivine aggregate at 4 GPa under Ni–NiO buffer from Wang
et al. (2006); C06: the latest model of olivine electrical conductivity at 0.1 MPa under QFM
(quartz–fayalite–magnetite) and IW (iron–wüstite) buffers from Constable (2006).
Taking the three different conduction mechanisms into account,
the resultant electrical conductivity of hydrous iron-bearing olivine
can be expressed as follows:
ð4Þ !


 1=3
H H H0
αCW
σ = σ0i exp
i + σ0h exp
h + σ0p CW exp
:
kT kT kT
ð7Þ

N 0
1=3
HA A = HA
αðNA Þ ð5Þ

N

where N− A is the number of proton per unit cell, HA is the value of HA at
A

a certain value of N− 0
A , HA is the activation enthalpy observed at very low
hydrogen concentrations and α is a constant accounting for geometrical
factors. Taking into account the concentration dependence of the pre-
exponential factor as defined by Eq. (4), the resultant electrical
conductivities of hydrous olivine can be expressed as follows:

1= 3
!
H 0
αCW
σp = σ0p CW exp
; ð6Þ
kT

where σ0P is a pre-exponential factor for proton conduction, CW is the

water content in weight percent, α is a geometrical factor, and H0 is the
activation enthalpy observed at very low water content. If the Fig. 6. Electrical conductivity of hydrogen-undoped olivine as a function of reciprocal
geometrical factor (α) were zero, the slopes for proton conduction in temperature. The dashed, solid thick and solid thin lines indicate the estimated ionic,
Figs. 5–7 would necessarily be constant. small polaron and proton conduction mechanisms, respectively.
 T. Yoshino et al. / Earth and Planetary Science Letters 288 (2009) 291–300
Fig. 7. Fitted electrical conductivity of olivine as a function of reciprocal temperature for
various water contents. The solid thick lines indicate the fitting results obtained from
Eq. (7).
The fitting parameters are shown in Table 2. Variations of con-
ductivity values and activation enthalpy as a function of the water
content were determined from Eq. (7) and can explain the present
data (Fig. 7).
The calculated dependence of water content on activation
enthalpy for proton conduction in olivine is quite small, as is the
case for wadsleyite (Yoshino et al., 2008; Manthilake et al., 2009). For
olivine the activation enthalpy for proton conduction (~0.9 eV) is
distinctly lower than that (1.14–1.86 eV) obtained from hydrogen
diffusion experiments (Mackwell and Kohlstedt, 1990) as was
observed for single crystal olivine (Yoshino et al., 2006). In hydrogen
diffusion experiments by Mackwell and Kohlstedt (1990) hydrogen
was incorporated into a hydrogen-free single crystal whereas in this
study hydrogen diffuses into olivine with an equilibrium concentra-
tion. For hydrogen diffusion experiments the migration of protons
must be accompanied by the creation of a Mg vacancy within the
olivine crystal. The resultant activation enthalpy for hydrogen
diffusion into the hydrogen-free crystal would become higher.
4.2. Comparison with previous work
The absolute conductivity values at high temperatures above
1400 K in this study are almost identical to those previously measured
at 0.1 MPa and 4 GPa using a starting composition of San Carlos
olivine (e.g., Constable et al., 1992; Xu et al., 1998). Using the same
oxygen buffer (Mo–MoO2), Xu et al. (2000) reported a positive
activation volume for electrical conductivity of olivine based on
conductivity measurements at pressures of 4, 7 and 10 GPa.
Conductivities obtained at 10 GPa using their data are lower than
the values obtained in this study. Because the experiments of Xu et al.
(2000) were conducted under various pressure conditions in a single
run repeated heating might be caused by iron loss to the electrode
although they denied this. On the other hand, electrical conductivity
Table 2
Parameter values.
Logσ0I HI* logσ0H HH* logσ0P HP0
(S/m) (eV) (S/m) (eV) (S/m) (eV)
4.73(53) 2.31(7) 2.98(85) 1.71(4) 1.90(44) 0.92(4)
297
values of dry single crystal olivine measured at 3 GPa (Yoshino et al.,
2006) are in agreement with the present results of the relatively dry
samples. These comparisons suggest that the pressure effect for small
polaron conduction in olivine is negligible. Constable (2006) recently
proposed a new model of dry olivine electrical conductivity as a
function of temperature at atmospheric pressure and designated the
model “Standard Electrical Olivine 3 (SEO3).” Conductivity values of
SEO3 are consistent with the present results.
Yoshino et al. (2006) reported the proton conduction of hydrous
single crystal olivine with 0.01–0.02 wt.% H2O using a Ni–NiO buffer.
The present data showed lower conductivity values and lower
activation energies for proton conduction compared with samples
with the same amount of water (Fig. 5). This may be because the
magnitude of proton conduction has a significant positive dependence
on ƒO2. The number of point defects to accommodate proton in olivine
increases with increasing ƒO2 (e.g., Grant et al., 2007). It may also
decrease the potential barrier of proton migration. Since mantle ƒO2
might decrease towards the transition zone (e.g., McCammon, 2005),
olivine conductivity due to proton conduction could be close to that
obtained from the present data in the vicinity of the 410 km
discontinuity. In addition, a positive activation volume for proton
conduction may result in the activation enthalpy increasing at higher
pressure. The other cause of high conductivity values obtained from
the hydrous single crystal is the possibility that the water content
estimated for the single crystal was underestimated because the
water content was determined from the FTIR spectra parallel to a
single crystallographic axis corresponding to the axis that was used
for the electrical conductivity measurements.
Wang et al. (2006) reported electrical conductivity for proton
conduction in hydrous olivine aggregate using the Ni–NiO buffer.
However, their conductivities are mostly higher than those obtained in
the present work. They reported that the amount of water in the
hydrous olivine decreased during their measurements and also found
some fluid in the grain boundary in their recovered samples. These
observations could be explained if electrical conductivity were partly
shunted through an interstitial fluid rather than purely through a
hydrous olivine. The temperature dependence shown in their figure
might indicate the amount of interstitial fluid during the measurement.
In this picture their conductivity values increased with increasing initial
water content in their hydrous olivine because the amount of released
fluid phase increased. Wang et al. (2006) proposed that magnitude of
proton conduction is proportional to around C2/3 W (0.62). When the
conductor phase distributed in layers along the grain boundary, the
effective conductivity maybe proportional to the two thirds power of
water content (Waff, 1974). This hypothesis is also the case for the
electrical conductivity of hydrous wadsleyite and ringwoodite mea-
sured by the same group (Huang et al., 2005). Yoshino et al. (2008)
reported much lower conductivity and much smaller temperature
dependence for hydrous wadsleyite and ringwoodite.
In view of thermochemistry, hydrous olivine must coexist with a
finite amount of aqueous fluid under equilibrium conditions accord-
ing to the following reaction:
Mg2
x H2x SiO4 = ð1
xÞ Mg2 SiO4 + xMgSiO3 + 2xH2 O:
If no fluid phase exists, the reaction proceeds in the right direction.
Under temperature conditions where the dehydration reaction can
proceed, the fluid phase must exit. To measure the electrical
conductivity of hydrous olivine without effects of the fluid phase we
have to kinetically prevent this reaction by maintaining low
temperatures, which is an approach adopted in this study.
4.3. Laboratory-based conductivity–depth profiles in the upper mantle
α
Electrical conductivity profiles can provide constraints on the thermal
0.16(2)
and chemical states of the mantle. To construct the conductivity–depth
298 T. Yoshino et al. / Earth and Planetary Science Letters 288 (2009) 291–300

profile as a function of water content in olivine we considered

distribution of conductivity in a wide depth range across a depth of
410 km, corresponding to the olivine–wadsleyite transition. In order to
constrain the upper limit for electrical conductivity of hydrous olivine,
the maximum water solubility as a function of depth based on the Eq. (1)
given by Hirschmann et al. (2005) was used for the calculation. Because
the water solubility of olivine used by Hirschmann et al. (2005) is three
times higher than that based on the calculation method proposed by
Paterson (1982), which we used in this study, we applied one-third of the
value of the water content used by Hirschmann et al. (2005) as the water
content of olivine. Electrical conductivity data of wadsleyite as a function
of water content by Yoshino et al. (2008) was used to consider the
conductivity jump at the depth of 410 km. Effects of grain size and
additional phases on the bulk rock conductivity were ignored to simplify
the model. The temperature distributions are largely controlled by the
conductive heat transfer in the lithosphere. The geotherm was
determined from thermobarometry of mantle xenoliths for the conti-
nental mantle (Rudnick et al., 1998), and derived from surface heat flow
for the oceanic mantle (e.g., Turcotte and Schubert, 1982). The geotherm
in the deep regions was considered to be adiabatic, which was taken from
Katsura et al. (2004). The oxygen fugacity of the transition zone is
assumed to be close to that of the iron–wüstite buffer (McCammon,
2005). Recently Rohrback et al. (2007) proposed that Fe metal is stable in
the upper mantle below at least 250 km depth. Since the Mo–MoO2
buffer is thought to be similar to the iron–wüstite buffer (Xu et al., 1998),
the obtained conductivity data based on Eq. (7) can be used directly for
the deep region (N250 km depth) of the upper mantle. Oxygen fugacity
in the shallower part of the upper mantle is considered to be relatively
high and generally close to that of the quartz–fayalite–magnetite (QFM)
buffer, based on studies of mantle peridotites (e.g., McCammon, 2005).
Because electrical conductivity increases with oxygen fugacity, assuming
such a reducing condition would yield an underestimation of conduc-
tivity values. The pressure dependence on electrical conductivity of
olivine is ignored.
Fig. 8 illustrates conductivity–depth profiles for oceanic and
continental upper mantles based on the present laboratory data Fig. 8. Laboratory-based conductivity–depth profiles as a function of water contents
(Yoshino et al., 2008) and the above assumptions. The conductivity– compared with the previous model (XSPR98: Xu et al., 1998). (a) Conductivity–depth
profiles for the oceanic mantle geotherm. (b) Conductivity–depth profiles for the
depth relation is strongly controlled by the thermal structure of the
continental mantle geotherm. Solid line indicates the dry mantle model. Thick short
mantle. The temperature abruptly increases with increasing depth in dashed lines denote the electrical conductivity of olivine and wadsleyite (Yoshino et al.,
the lithospheric mantle because of the conductive heat transfer. The 2008) as a function of water content (0.01, 0.1 and 1.0 wt.%). The water content of
geotherms change from the conductive one to the adiabat one at 1.0 wt.% is only shown in the stability field of wadsleyite. Thick long dashed line
depths around 60 km for the oceanic mantle (Fig. 8a) and over 200 km indicates the upper bound of the electrical conductivity of the hydrous olivine. Thin
dashed line represents the conductivity–depth model of Xu et al. (1998).
for the continental mantle (Fig. 8b). According to the change in the
heat transfer mechanism, the conductivity values abruptly increase
with increasing depth to around 10− 2 S/m in the shallower regions of study; Constable, 2006) demonstrated that olivine has similar
the mantle, and then gently increase in the deeper regions of the conductivity to wadsleyite (Yoshino et al., 2008; Manthilake et al.,
mantle. 2009) at pressure corresponding to a depth of 400 km. The effect of
The presence of a small amount of water in olivine affects its water on the electrical conductivity of olivine is more pronounced
electrical conductivity and this considerably affects the conductivity than the effect of high-pressure olivine polymorphs (wadsleyite and
structure of the upper mantle. As the water content in olivine ringwoodite). Assuming that both olivine and wadsleyite contain
increases from 0.01 to 0.1 wt.% the conductivity values of the mantle 0.1 wt.% H2O the conductivity decreases slightly across the depth of
adiabat or lithospheric regions increase by a half or one order of the olivine–wadsleyite transition. The reason for the more pro-
magnitude, respectively. If the water content in olivine is less than nounced effect of water on the electrical conductivity of olivine than
0.01 wt.% the water content cannot be determined from geophysical of wadsleyite would be explained in view of accommodations of
observations because the contribution of proton conduction to the hydrogen in their crystal structures. The maximum water content in
bulk conductivity becomes negligible. In addition, for the shallow olivine (0.89 wt.%: Smyth et al., 2006) is much smaller than that of
upper mantle, the estimation of water content in olivine would be wadsleyite (3.3 wt.%: Smyth, 1987). Lower accommodation of
even more difficult because the conductivity predicted from the hydrogen in the olivine structure than in the wadsleyite structure
maximum H2O solubility in olivine decreases with decreasing depth. may indicate that hydrogen is much less bounded in the olivine
The dry model shows no distinct conductivity jump at a depth of structure and can easily migrate within the crystal lattice.
410 km while the previous model showed a large conductivity jump
(by 2 log units) at this depth (Xu et al., 1998). Yoshino et al. (2008) 4.4. Implications of the conductivity–depth profiles in the upper mantle
claimed a conductivity jump by 0.5 log unit. This is because they used
the conductivity values given by (Xu et al., 2000), which show the To estimate the water content of olivine in the upper mantle some
negative pressure dependence of the electrical conductivity of olivine. geophysical observations were compared with our model (Fig. 9).
However, recent studies on electrical conductivities of olivine (this Several recent one-dimensional (1D) electrical conductivity profiles
 T. Yoshino et al. / Earth and Planetary Science Letters 288 (2009) 291–300
Fig. 9. Comparison of laboratory data on electrical conductivity of hydrous olivine as a
function of water contents with the geophysically observed conductivity–depth profiles
beneath the Pacific (a) and stable craton (b). NE Pacific (Lizarralde et al., 1995); EPR:
eastern part of the Eastern Pacific Rise (Evan et al., 2005); Pacific (Kuvshinov et al.,
2005); Philippine Sea (Seama et al., 2007); Superior Province (Schultz et al., 1993);
Carty Lake (Neal et al., 2000); Slave craton (Jones et al., 2003).
reveal a considerable lateral variation in the upper mantle. The
conductivity profiles, ranging from 100 to 250 km depth, are largely
higher for the mantle beneath the Northern Pacific and the Eastern
Pacific Rise (Lizarralde et al., 1995; Evans et al., 2005) than those for the
continental mantle beneath the stable cratons or continents (Schultz
et al., 1993; Olsen, 1998; Neal et al., 2000; Jones et al., 2003; Tarits et al.,
2004). This difference is considered to be because of a variation in the
lithosphere's thickness (e.g., Hirth et al., 2000; Jones et al., 2003).
Seismic studies show that the thickness of the oceanic lithosphere
ranges from 0 km at the mid-ocean ridges to 65 km for the old oceanic
plates (Gaherty et al., 1999). In contrast, the thickness of the continental
lithosphere beneath the Archean craton is believed to be much thicker
~ 400 km (Jordan, 1975; 1988) and recent studies have indicated a 200–
300 km thickness (e.g., Gung et al., 2003). The lithosphere–astheno-
sphere boundary as determined from seismological studies seems to be
consistent with the depth at which conductivity changes. Hirth et al.
(2000) concluded that the electrical conductivity in the deep litho-
sphere beneath the stable craton (Canadian shield) is consistent with a
reduced water content compared with the tectonically activated or
oceanic mantle, which may play a role in stabilizing the Archean
lithosphere against convective erosion.
Most models beneath the Pacific show a high conductivity anomaly
in the asthenosphere except for the latest semi-global reference model
299
beneath the Pacific (Kuvshinov et al., 2005). The conductivity values
near the Eastern Pacific Rise showing high conductivity parallel to the
plate motion (Evans et al., 2005) are too high to account for the olivine
hydration. Yoshino et al. (2006) also demonstrated that the magnitude
of conductivity anisotropy and conductivity values obtained from the
conductivity measurement of hydrous single crystal olivine were not
likely to account for the conductivity anomaly at the top of the
asthenosphere. The presence of a partial melt aligned parallel to the
plate motion would be required. The conductivity–depth model
beneath Hawaii (Lizarralde et al., 1995) yields higher conductivity
values than our model accounted for by considering the maximum
solubility of H2O in olivine at around the 200 km depth. This discrepancy
suggests that this region has a 150 K higher temperature than the
normal geotherm because of the presence of a hot plume beneath
Hawaii (Fukao et al., 2001). The semi-global model by Kuvshinov et al.
(2005) showed lower conductivity values. A reduction of temperature
and/or a presence of olivine with lower iron content are required to
explain this conductivity values. The conductivity profile beneath the
Philippine Sea (Seama et al., 2007) shows higher conductivity compared
with our dry model. In this case, olivine hydration can occur because the
subducting slab beneath the Philippine plate can supply water to the
wedge mantle. If this region had a lower temperature than the normal
geotherm a significant amount of water in excess of 0.1 wt.% might exist
in olivine. For example, at a temperature 200 K lower than the normal
geotherm the predicted water content in olivine would be around
0.25 wt.%.
Conductivity values of the upper mantle beneath continents or the
stable craton (Europe and Canadian shield) are relatively lower than
conductivity values beneath the Pacific (Schultz et al., 1993; Olsen,
1998; Neal et al., 2000; Jones et al., 2003; Tarits et al., 2004).
Conductivity values tend to become constant with increasing depth at
around 250 km. The region shallower than 250 km depth is considered
to be the continental lithospheric mantle. An abrupt change in
conductivity of the lithosphere is in agreement with that of the thermal
gradient as derived from the thermal structure beneath the stable craton
rather than the depletion of water in the lithosphere as pointed out by
Hirth et al. (2000). This is because the effect of water on electrical
conductivity is much smaller than the temperature effect. The region
shallower than 150 km depth has significantly higher conductivity
values than our model predicts even if maximal olivine water content is
assumed. The presence of aqueous fluid or a graphite film along grain
boundaries would explain these high conductivity values. Conductivity–
depth profiles beneath the stable craton, showing no conductivity jump
at the 410 km discontinuity, are consistent with our dry model.
Therefore, the olivine–wadsleyite transition is not a significant
boundary for the conductivity structure of the continental upper mantle.
5. Concluding remarks
1. Three conduction mechanisms were identified from the Arrhenian
behavior of the olivine aggregates. At temperatures above 1700 K,
the activation enthalpy exceeds 2 eV suggesting that the dominant
mechanism of charge transport would be ionic conduction. Below
1700 K, small polaron conduction would be dominant conduction
mechanism. The conductivity increased with increasing water
content by proton conduction. Eq. (7) provides a quantitative
description of electrical conductivity for olivine containing hydro-
gen to temperatures down to ~ 500 K (Fig. 7). There is an upper
limit to water solubility and therefore to conductivity due to the
proton conduction at any temperature.
2. In the upper asthenosphere, temperatures are too low for even
olivine containing water up to the solubility limit to be an adequate
conductor. Therefore, another conduction mechanism, e.g., partial
melt, fluids or graphite, must be invoked to enhance conductivity to
obtain observed lithospheric or upper asthenospheric conductivities.
300 T. Yoshino et al. / Earth and Planetary Science Letters 288 (2009) 291–300
Acknowledgments
We are grateful to E. Ito, D. Yamazaki, H. Utada, K. Baba and M.
Ichiki for the discussions. The manuscript benefited from the
constructive reviews of two anonymous reviewers. This work was
supported by grant-in-aids for scientific research, nos. 18740280 and
13440164 to TY and TK, respectively, from the Japan Society for
Promotion of Science. It was also supported by the COE-21 program of
the Institute for Study of the Earth's Interior, Okayama University.
References
Bell, D.R., Rothman, G.R., 1992. Water in the earth's mantle: the role of nominally
anhydrous minerals. Science 255, 1391–1397.
Bell, D.R., Rossman, G.R., Maldener, J., Endisch, D., Rauch, F., 2003. Hydroxide in olivine:
a quantitative determination of the absolute amount and calibration of the IR
spectrum. J. Geophys. Res. 108, 2105.
Bolfan-Casanova, N., Keppler, H., Rubie, D.C., 2000. Water partitioning between
nominally anhydrous minerals in the MgO–SiO2–H2O system up to 24 GPa:
implications for the distribution of water in the Earth's mantle. Earth Planet. Sci.
Lett. 182, 209–211.
Constable, S., 2006. SEO3: a new model of electrical conductivity. Geophys. J. Int. 166,
435–437.
Constable, S., Shankland, T.J., Duba, A., 1992. The electrical conductivity of an isotropic
olivine mantle. J. Geophys. Res. 97, 3397–3404.
Debye, P.P., Conwell, E.M., 1954. Electrical properties of N-type germanium. Phys. Rev.
93, 693–706.
Evans, R.L., Hirth, G., Baba, K., Forsyth, D., Chave, A., Mackie, R., 2005. Geophysical
evidence from the MELT area for compositional controls on oceanic plates. Nature
437, 249–252.
Fukao, Y., Widiyantoro, S., Obayashi, M., 2001. Stagnant slabs in the upper and lower
mantle transition region. Rev. Geophys. 39, 291–323.
Gaherty, J.B., Kato, M., Jordan, T.H., 1999. Seismological structure of the upper mantle: a
regional comparison of seismic layering. Phys. Earth Planet. Inter. 110, 21–41.
Grant, K.J., Brooker, R.A., Kohn, S.C., Wood, B.J., 2007. The effect of oxygen fugacity on
hydroxyl concentrations and speciation in olivine: implications for water solubility
in the upper mantle. Earth Planet. Sci. Lett. 261, 217–229.
Gung, Y., Romanowicz, B., Panning, M., 2003. Global anisotropy and the thickness of
continents. Nature 422, 707–711.
Hirschmann, M.M., Aubaud, C., Withers, A.C., 2005. Storage capacity of H2O in
nominally anhydrous minerals in the upper mantle. Earth Planet. Sci. Lett. 236,
167–181.
Hirth, G., Kohlstedt, D.L., 1996. Water in the oceanic upper mantle: implications for
rheology, melt extraction and the evolution of the lithosphere. Earth Planet. Sci.
Lett. 144, 93–108.
Hirth, G., Evans, R.L., Chave, A.D., 2000. Comparison of continental and oceanic mantle
electrical conductivity: in the Archean lithosphere dry? Geochem. Geophys.
Geosyst. 1 2000GC000048.
Huang, X., Xu, Y., Karato, S., 2005. Water content in the transition zone from electrical
conductivity of wadsleyite and ringwoodite. Nature 434, 746–749.
Iyenger, G.N.K., Alcock, C.B., 1970. A study of semiconduction in dilutic magnesio-
wüstite. Philos. Mag. 21, 293–304.
Jacobsen, S.D., Smyth, J.R., Spetzler, H., Holl, C.M., Frost, D.J., 2004. Sound velocities and
elastic constants of iron-bearing hydrous ringwoodite. Phys. Earth Planet. Int. 143–
144, 47–56.
Jones, A.G., Lezaeta, P., Ferguson, I.J., Chave, A.D., Evans, R.L., Garcia, X., Spratt, J., 2003.
The electrical structure of the Slave craton. Lithos 71, 505–527.
Jordan, T.H., 1975. The continental lithosphere. Rev. Geophys. Space Phys. 13, 1–12.
Jordan, T.H., 1988. Structure and formation of continental tectosphere. J. Petrol. 29,
11–37.
Karato, S., 1990. The role of hydrogen in the electrical conductivity of the upper mantle.
Nature 347, 272–273.
Karato, S., Paterson, M.S., Fitz Gerald, J.D., 1986. Rheology of synthetic olivine
aggregates—influence of grain-size and water. J. Geophys. Res. 91, 8151–8176.
Katsura, T., Yamada, H., Nishikawa, O., Song, M., Kubo, A., Shinmei, T., Yokoshi, S., Aizawa,
Y., Yoshino, T., Walter, M.J., Ito, E., 2004. Olivine–wadsleyite transition in the system
(Mg, Fe)2SiO4. J. Geophys. Res. 109, B02099. doi:10.1029/2003JB002438.
Kohlstedt, D.L., Keppler, H., Rubie, D.C., 1996. Solubility of water in the α, β and γ phases
of Mg 2SiO4. Contrib. Mineral. Petrol. 123, 345–357.
Kröger, F.A., Vink, H.A., 1956. Relations between the concentrations of imperfections in
crystalline solids. In: Seitz, F., Turnball, D. (Eds.), Solid State Physics, vol. 3. Academic
Press, NY, pp. 307–435.
Kuvshinov, A., Utada, H., Avdeev, D., Koyama, T., 2005. 3-D modelling and analysis of
Dst C-responses in the North Pacific Ocean region, revisited Geophys. J. Int. 160,
505–526.
Litasov, K.D., Ohtani, E., Kagi, H., Jacobsen, S.D., Ghosh, S., 2007. Temperature
dependence and mechanism of hydrogen incorporation in olivine at 12.5–
14.0 GPa. Geophys. Res. Lett. 34, L16314. doi:10.1029/2007GL030737.
Lizarralde, D., Chave, A.D., Hirth, G., Schultz, A., 1995. A Northern Pacific mantle
conductivity profile from long-period magnetotelluric sounding using Hawaii to
California submarine cable data. J. Geophys. Res. 100, 17,837–17,854.
Mackwell, S.J., Kohlstedt, D.L., 1990. Diffusion of hydrogen in olivine: implications for
water in mantle. J. Geophys. Res. 95, 5079–5088.
Manthilake, G., Matsuzaki, T., Yoshino, T., Yamashita, S., Ito, E., Katsura, T., 2009.
Electrical conductivity of wadsleyite as a function of temperature and water
content. Phys. Earth Planet. Int. 174, 10–18.
McCammon, C., 2005. The paradox of mantle redox. Science 308, 807–808.
Mei, S., Kohlstedt, D.L., 2000. Influence of water on plastic deformation of olivine
aggregates, part I: diffusion creep regime. J. Geophys. Res. 105, 21,457–21,469.
Neal, S.L., Mackie, R.L., Larsen, J.C., Schultz, A., 2000. Variations in the electrical
conductivity of the upper mantle beneath North America and the Pacific Ocean.
J. Geophys. Res. 105, 8229–8242.
Olsen, N., 1998. The electrical conductivity of the mantle beneath Europe derived from
C-responses from 3 to 270 hours. Geophys. J. Int. 133, 298–308.
Paterson, M.S., 1982. The determination of hydroxyl by infrared absorption in quartz,
silicate glasses and similar materials. Bull. Mineral. 105, 20–29.
Roberts, J.J., 2002. Electrical properties of microporous rock as a function of saturation
and temperature. J. Appl. Phys. 91, 1687–1694.
Rohrback, A., Ballhaus, C., Golla-Scindler, U., Ulmer, P., Kamenetsky, V.S., Kuzmin, D.V.,
2007. Metal saturation in the upper mantle. Nature 449, 456–458.
Rudnick, R.L., McDonough, W.F., O'Connell, R.J., 1998. Thermal structure, thickness and
composition of continental lithosphere. Chem. Geol. 145, 395–411.
Schock, R.N., Duba, A.G., Shankland, T.J., 1989. Electrical conduction in olivine. J. Geophys.
Res. 94, 5829–5839.
Schultz, A., Kurtz, R.D., Chave, A.D., Jones, A.G., 1993. Conductivity discontinuities in the
upper mantle beneath a stable craton. Geophys. Res. Lett. 20, 2941–2944.
Seama, N., Baba, K., Utada, H., Toh, H., Tada, K., Ichiki, M., Matsuno, T., 2007. 1-D
electrical conductivity structure beneath the Philippine Sea: results from an
oceanic bottom magnetotelluric survey. Phys. Earth Planet. Inter. 162, 2–12.
Smyth, J.R., 1987. Mg2SiO4: a potential host for water in the mantle? Am. Mineral. 72,
1051–1055.
Smyth, J.R., Frost, D.J., Nestola, F., Holl, C.M., Bromiley, G., 2006. Olivine hydration in the
deep upper mantle: effect of temperature and silica activity. Geophys. Res. Lett. 33,
L06612. doi:10.1029/2006GL026194.
Tarits, P., Hautot, S., Perrier, F., 2004. Water in the mantle: results from electrical
conductivity beneath the French Alps. Geophys. Res. Lett. 33, L15301. doi:10.1029/
2003GL019277.
Turcotte, D.L., Schubert, G., 1982. Geodynamics. John Wiley & Sons, Inc., New York.
p. 450.
Waff, H.S., 1974. Theoretical considerations of electrical conductivity in partial molten
mantle and implications for geothermometry. J. Geophys. Res. 79, 4003–4010.
Wang, D., Mookherjee, M., Xu, Y., Karato, S., 2006. The effect of water on the electrical
conductivity of olivine. Nature 443, 977–980.
Xu, Y., Poe, B.T., Shankland, T.J., Rubie, D.C., 1998. Electrical conductivity of olivine,
wadsleyite and ringwoodite under upper-mantle condition. Science 280, 1415–1418.
Xu, Y., Shankland, T.J., Duba, A.G., 2000. Pressure effect on electrical conductivity of
mantle olivine. Phys. Earth Planet. Int. 118, 149–161.
Yoshino, T., Matsuzaki, T., Yamashita, S., Katsura, T., 2006. Hydrous olivine unable to
account for conductivity anomaly at the top of the asthenosphere. Nature 443,
973–976.
Yoshino, T., Manthilake, G., Matsuzaki, T., Katsura, T., 2008. Dry mantle transition zone
inferred from electrical conductivity of wadsleyite and ringwoodite. Nature 451,
326–329.
 Update
Earth and Planetary Science Letters
Volume 391, Issue , 1 April 2014, Page 135–136

DOI: https://doi.org/10.1016/j.epsl.2014.02.016
 Earth and Planetary Science Letters 391 (2014) 135–136

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

Corrigendum

Corrigendum to “The effect of water on the electrical conductivity of

olivine aggregates and its implications for the electrical structure in
the upper mantle” [Earth Planet. Sci. Lett. 288 (2009) 291–300]
Takashi Yoshino ∗ , Takuya Matsuzaki 1 , Anton Shatzkiy 2 , Tomoo Katsura 3
Institute for Study of the Earth’s Interior, Okayama University, Misasa, Tottori 682-0193, Japan

a r t i c l e i n f o

Article history:
Received 17 December 2013
Accepted 8 January 2014
Available online 14 February 2014
Editor: L. Stixrude

The authors noticed that water content represented by a unit of H/106 Si in Table 1 of the above article is one order of magnitude
smaller.
The correct values should be as following:

Table 1
Summary of runs.

Run No. Samplea H2 O wt.%b H2 O wt.%c T log σ H ·∗ Remarks

H/106 Si H/106 Si (K) (S/m) (eV)
5K1055 SC 0.005(0) 800–1400 0.83(7) 1.11(0)
800(30) 1400–1750 3.34(8) 1.77(2)
1750–2000 4.73(19) 2.25(7)
5K1126 SC 0.008(1) 700–1200 0.82(19) 1.10(3)
1300(310) 1350–1650 2.44(10) 1.47(3)
1700–2000 4.47(10) 2.14(4)
5K1081 SC 0.009(4) 1550–1750 3.08(10) 1.61(3)
1500(490)
5K1223d SC 0.012(6) 550–1350 0.87(6) 1.06(1)
2000(870) 1400–1800 2.44(19) 1.51(6)
1K484 SC 0.022(3) 673–973 1.06(8) 0.98(1)
3400(400)
5K1061 SCTB 0.042(5) 0.040(4) 600–800 0.91(21) 0.88(3)
6900(800) 6600(600)
5K1079 SCTB n.d. 0.066(3) 500–1000 0.93(16) 0.87(0)
11000(600)
5K1112 SCTB 0.166(17) 0.l59(42) 500–900 1.31(10) 0.88(1)
27000(2700) 26000(7000)
5K1122 SCTB n.d. 0.160(7) 550–900 2.44(12) 1.03(2)
26000(1100)
(continued on next page)

DOI of original article: http://dx.doi.org/10.1016/j.epsl.2009.09.032.

* Corresponding author at: Institute for Study of the Earth’s Interior, Okayama University, 827 Yamada, Misasa, Tottori 682-0193, Japan.
E-mail address: tyoshino@misasa.okayama-u.ac.jp (T. Yoshino).
1
Current address: Center for Advanced Marine Core Research, Kochi University, Nankoku, Kochi, 783-8502, Japan.
2
Current address: V.S. Sobolev Institute of Geology and Mineralogy, Russian Academy of Science, Siberian Branch, Novosivirsk 630090, Russia.
3
Current address: Bayerisches Geoinstitut, University of Bayreuth, Bayreuth 95447, Germany.

http://dx.doi.org/10.1016/j.epsl.2014.02.016
0012-821X/© 2014 Elsevier B.V. All rights reserved.
136 T. Yoshino et al. / Earth and Planetary Science Letters 391 (2014) 135–136

Table 1 (continued)

Run No. Samplea H2 O wt.%b H2 O wt.%c T log σ H ·∗ Remarks

H/106 Si H/106 Si (K) (S/m) (eV)
5K1221d SCTB 0.179(12) 500–850 1.29(11) 0.85(1)
29000(3200)
5K1118 SCTB 0.289(26) dehydration
47000(4300)
a
Starting materials used for the conductivity measurement. SC: San Carlos olivine powder with tiny amount of orthopyroxene. SCTB: The pre-synthesized hydrous olivine
aggregates.
b
Water content (in weight %) of synthesized hydrous olivine aggregate before the conductivity measurement experiment, determined by FTIR spectroscopy. Parentheses
denote the standard deviation for five analyses.
c
Water content (in wt.%) after the conductivity measurement.
d
The conductivity were determined by impedance spectroscopic analyses.

However, the fit lines from the least squares regression in Fig. 7 are correct, and the fitting parameters in Table 2 are also correct. The
conclusions of the study were not changed.
The author apologizes for any inconvenience caused.