Mechanisms of iron oxide transformations

in hydrothermal systems
Tsubasa Otake
a,
, David J. Wesolowski
b
, Lawrence M. Anovitz
b,c
,
Lawrence F. Allard
b
, Hiroshi Ohmoto
a
a
NASA Astrobiology Institute & Department of Geosciences, The Pennsylvania State University, University Park, PA 16802, USA
b
Oak Ridge National Laboratory, Oak Ridge, TN 37831-6110, USA
c
Department of Earth and Planetary Sciences, University of Tennessee, Knoxville, TN 37996, USA
Received 17 August 2009; accepted in revised form 22 July 2010; available online 29 July 2010
Abstract
Coexistence of magnetite and hematite in hydrothermal systems has often been used to constrain the redox potential of
uids, assuming that the redox equilibrium is attained among all minerals and aqueous species. However, as temperature
decreases, disequilibrium mineral assemblages may occur due to the slow kinetics of reaction involving the minerals and uids.
In this study, we conducted a series of experiments in which hematite or magnetite was reacted with an acidic solution under
H
2
-rich hydrothermal conditions (T = 100250 C, P
H
2
= 0:055 MPa) to investigate the kinetics of redox and non-redox
transformations between hematite and magnetite, and the mechanisms of iron oxide transformation under hydrothermal con-
ditions. The formation of euhedral crystals of hematite in 150 and 200 C experiments, in which magnetite was used as the
starting material, indicates that non-redox transformation of magnetite to hematite occurred within 24 h. The chemical com-
position of the experimental solutions was controlled by the non-redox transformation between magnetite and hematite
throughout the experiments. While solution compositions were controlled by the non-redox transformation in the rst 3 days
in a 250 C experiment, reductive dissolution of magnetite became important after 5 days and aected the solution chemistry.
At 100 C, the presence of maghemite was indicated in the rst 7 days. Based on these results, equilibrium constants of non-
redox transformation between magnetite and hematite and those of non-redox transformation between magnetite and maghe-
mite were calculated. Our results suggest that the redox transformation of hematite to magnetite occurs in the following steps:
(1) reductive dissolution of hematite to Fe
2
(aq)
and (2) non-redox transformation of hematite and Fe
2
(aq)
to magnetite.
2010 Elsevier Ltd. All rights reserved.
1. INTRODUCTION
Hematite (a-Fe
III
2
O
3
) and magnetite (Fe
II
Fe
III
2
O
4
) are
the two most common iron oxides in near-surface environ-
ments, and iron is the most abundant redox-sensitive metal
in the Earths crust. Magnetite is stable under more reduc-
ing conditions than hematite, when the redox equilibrium is
attained among all aqueous species and minerals (Fig. 1).
The redox equilibrium is generally attained between miner-
als and solutions during the nucleation and precipitation of
iron oxides, as indicated by the observation that the miner-
alogy of iron (hydr)oxides (e.g., goethite) precipitating in
near-surface water bodies agrees with that predicted from
the measured Eh-pH values (e.g., Lindberg and Runnells,
1984). However, once the iron (hydr)oxides are subjected
to new conditions, redox equilibrium may not be estab-
lished between the iron oxides and surrounding water; this
is illustrated by the common occurrence of magnetite in
beach sands under oxic atmosphere and that of hematite
in anoxic water bodies (e.g., the Red Sea metalliferous sed-
iments, Pottorf and Barnes, 1983).
0016-7037/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.gca.2010.07.024

Corresponding author. Address: Department of Earth and

Planetary Materials Science, Graduate School of Science, Tohoku
University, Aoba 6-3, Aramaki, Aoba-ku, Sendai 980-8578, Japan.
Tel./fax: +81 22 795 3453.
E-mail address: totake@m.tohoku.ac.jp (T. Otake).
www.elsevier.com/locate/gca
Available online at www.sciencedirect.com
Geochimica et Cosmochimica Acta 74 (2010) 61416156
Coexistence of magnetite and hematite is observed in
various modern and ancient hydrothermal systems (e.g.,
Zierenberg and Shanks, 1983; Davidson, 1992; Gow
et al., 1994; Lynch and Ortega, 1997), and has often been
used to constrain the redox potential of hydrothermal u-
ids, because this parameter is crucial for determining the
capability of transporting other metals (e.g., Barnes,
1997). However, for the waterrock systems where redox
equilibrium is not attained between iron oxides and water,
the redox potential of groundwater/hydrothermal uids
cannot be constrained from the mineralogy of iron oxides
in the plumbing system. Therefore, one of the important
questions in hydrothermal geochemistry is the kinetics
and mechanisms of redox reactions between iron oxides
and water, H
2
and O
2
under hydrothermal conditions.
Transformation between magnetite and hematite has
been considered a redox reaction involving an oxidant
(e.g., O
2
; SO
4
2
) or a reductant (e.g., H
2
, organic com-
pounds), as for example:
2Fe
3
O
4(mt)
H
2
O 3Fe
2
O
3(hm)
H
2(g)
(1)
The attainment of equilibrium among magnetite, hematite,
and H
2
O would imply the attainment of redox equilibrium
for the following two redox reactions:
1=3Fe
3
O
4(mt)
2H

(aq)
1=3H
2(g)
Fe
2
(aq)
4=3H
2
O
(2)
1=2Fe
2
O
3(hm)
2H

(aq)
1=2H
2(g)
Fe
2
(aq)
3=2H
2
O
(3)
Reactions (2) and (3) represent reversible reductive dissolu-
tion/oxidative precipitation of magnetite and hematite,
respectively, and the equilibrium constants (1)(3) of theses
reactions are related by:
log K
1
= 6 log K
2
6 log K
3
= 6 log K
2
=K
3
(4)
Previous investigators have shown that the redox equilib-
rium between magnetite and hematite is sluggish under
low-temperature hydrothermal conditions. Kishima and
Sakai (1984) and Seyfried et al. (1987) conducted long-term
experiments with an initial excess of H
2
(P
H
2
> 0.1 MPa),
150C
5
0
-5
-10
-15
-20
2 4 6 8 0
Fe
2+
Hematite
(Fe
2
O
3
)
aFe
2+
= 10
-2
10
-4
10
-6
l
o
g

P
H
2

(
M
P
a
)
pH
Magnetite
(Fe
3
O
4
)
(b)
200C
5
0
-5
-10
-15
-20
2 4 6 8 0
Fe
2+
Hematite
(Fe
2
O
3
)
Magnetite
(Fe
3
O
4
)
aFe
2+
= 10
-2
10
-4
10
-6
l
o
g

P
H
2

(
M
P
a
)
pH
(c)
250C
5
0
-5
-10
-15
-20
2 4 6 8 0
Fe
2+
Hematite
(Fe
2
O
3
)
Magnetite
(Fe
3
O
4
)
aFe
2+
= 10
-2
10
-4
10
-6
l
o
g

P
H
2

(
M
P
a
)
pH
(d)
100C
5
0
-5
-10
-15
-20
2 4 6 8 0
Fe
2+
Hematite
(Fe
2
O
3
)
Magnetite
(Fe
3
O
4
)
aFe
2+
= 10
-2
10
-4
10
-6
l
o
g

P
H
2

(
M
P
a
)
pH
(a)
Fig. 1. Stability diagram of iron oxides as a function of pH and P
H2
at (a) 100 C, (b) 150 C, (c) 200 C, and (d) 250 C, considering the
phases, hematite and magnetite. Solid-solution boundaries are drawn for activity of Fe
2+
of 10
6
, 10
4
, and 10
2
. The diagrams are made
using Geochemists Workbench

(2002) using themo.com.v8.r6+.dat database, based on thermodynamic data from Baes and Mesmer (1976),
Helgeson et al. (1978) and Johnson et al. (1992). Shaded area indicates our experimental conditions at each temperature.
6142 T. Otake et al. / Geochimica et Cosmochimica Acta 74 (2010) 61416156
and demonstrated that redox equilibrium between hematite
and magnetite was established only after 80 days at 300 C.
Therefore, under the conditions where equilibria (2) and (3)
do not easily take place, transformations of iron oxides may
not occur through redox reactions. Ohmoto (2003) inferred
from preliminary experimental studies (Wesolowski et al.,
2000) that transformations between hematite and magnetite
at temperatures below ~300 C proceed through the follow-
ing redox-independent reaction:
Fe
3
O
4(mt)
2H

(aq)
Fe
2
O
3(hm)
Fe
2
(aq)
H
2
O (5)
In this reaction, hematite is formed by the removal of Fe
2+
from magnetite and magnetite is formed by the addition of
Fe
2+
to hematite; no reductant or oxidant is involved in the
reaction. However, this transformation requires either so-
lid-state recrystallization or dissolution/precipitation, since
these phases have dierent oxygen framework structures.
Ohmoto (2003) termed reaction (5) a non-redox transfor-
mation reaction of iron oxide.
Otake et al. (2007) conducted a series of laboratory
experiments to investigate the processes of non-redox
transformation of magnetite to hematite, and hematite to
magnetite, both under H
2
-rich hydrothermal conditions
at 150

C (P
H
2
= 0:053:5 MPa and pH = 26). Under
the high P
H
2
conditions of the experiments, only magnetite
should be thermodynamically stable (Fig. 1b; Baes and
Mesmer, 1976; Helgeson et al., 1978; Johnson et al.,
1992). However, we found that the dissolution of Fe
2+
from magnetite by proton attack caused the transforma-
tion of some magnetite crystals to larger, euhedral hema-
tite crystals, as indicated by the forward direction of
reaction (5). While equilibrium [Fe
2+
]/[H
+
]
2
ratios in terms
of reaction (5) were attained very rapidly (within 24 h) un-
der these conditions, redox equilibrium in terms of reaction
(2) or (3) was not attained, even after 16 days in the stirred
reactor. Although hematite was not expected to be present
in redox equilibrium under the experimental conditions,
the metastable mineral assemblage persisted and controlled
the chemical composition of the experimental solution dur-
ing the course of experiments. Acid/base titration experi-
ments demonstrated that non-redox transformation
between magnetite and hematite is reversible, and that
Fe
2
(aq)
is the dominant aqueous species under the experi-
mental conditions. This study suggested that the hema-
titemagnetite assemblage buered pH and Fe
concentration of uids through reaction (5), rather than
the redox-equilibrium reaction (1), under low-temperature
hydrothermal conditions. The transformation products
are chemically and structurally homogeneous and typically
occur as euhedral, 36 lm single crystals. Notably, pro-
duced hematite exhibits the unusual ditrigonal dipyramid
morphology.
The purposes of this study are (i) to determine the equi-
librium constants for non-redox equilibrium between mag-
netite and hematite under a wide range of temperature
conditions (i.e., 100250 C), (ii) to investigate the eect
of hydrogen pressure and pH change on the transformation
mechanisms of magnetite to hematite, and hematite to mag-
netite, (iii) to investigate whether a precursor mineral was
present for the formation of hematite during the non-redox
transformation of magnetite to hematite at a low tempera-
ture (i.e., 100 C), and nally (iv) to understand the mech-
anisms of iron oxide transformation occurring in natural
and laboratory systems under low-temperature hydrother-
mal conditions.
2. EXPERIMENTAL
2.1. Experimental apparatus and materials
The apparatus used in this study was the same as that
used in our previous study (Otake et al., 2007). The stirred,
hydrogen-electrode concentration cell consists of a heavy-
walled pressure vessel (stainless steel) containing two Teon
compartments that share a common head-space and are
joined by a porous Teon liquid junction (Mesmer et al.,
1970; Palmer and Hyde, 1993; Wesolowski et al., 2000).
Each compartment contains a PtH
2
electrode for in situ
pH measurement. This cell is tted with titrant delivery, l-
tered sample removal lines, and a hydrogen-permeable
membrane enabled continuous monitoring of the hydrogen
partial pressure.
Methods for preparation of the magnetite and hematite
powders and triuoromethanesulfonic (hereafter triic)
acid and NaTriate solutions used in the experiments are
described in Otake et al. (2007).
2.2. Experimental procedure
We ran a total of 11 series of experiments, which are di-
vided into the following types: (1) magnetite + acidic solu-
tion at 100250 C (runs #4, #7, #9, #15, #18, and #21),
(2) magnetite + acid/base titration at 150 and 200 C (runs
#5, #6, and #14), (3) hematite + acidic solution at 150 C
(run #11), and (4) hematite + acid/base titration at 150 C
(run #12). We use terms magnetite experiment and
hematite experiment, depending on the starting iron
oxide. Magnetite or hematite (0.31.2 g) was placed in
the cell with ~50 g of various concentrations (0.00025
0.02 mol/kg) of triic acid solution. NaTriate was added
to the solution for the adjustment of the ionic strength
(0.100.12). This background electrolyte, rather than
NaCl, was used to minimize the liquid junction term in
the calculation of pH from the measured cell potential,
while avoiding the complication of chloride complexation
of the dissolved iron species. The vessel was repeatedly
purged and then pressurized by high-purity hydro-
gen + argon (P
H
2
= 0:055 MPa) and rapidly (<1 h) heated
to the desired temperature (100, 150, 200, or 250 C), and
then thermostatted to 0.05 C. After the system reached a
steady state, as indicated by stabilized potential reading
(typically <0.5 mV change in a few hours), the stirring mo-
tor was turned o, and the solid permitted to settle for a
period of ~10 min. A solution sample was then withdrawn
from the experimental solution through a platinum frit
gold-welded to a platinum capillary tube immersed in the
test solution, then through a 0.2 lm PVDF syringe lter
axed to the titanium sampling valve outside the furnace.
About 1 g of solution was extracted and discarded, then 2
5 g were collected in a polypropylene syringe, pre-loaded
Iron oxide transformations in hydrothermal systems 6143
with 28 g of 0.2 wt.% HNO
3
solution. Stirring was then
resumed until the next sampling. In the rst series of exper-
iments, hydrogen pressure was increased during the runs.
In the second series of experiments, titrant solutions were
added to the experimental solution. Experiments were
run for 116 days. After the vessel was removed from the
furnace and rapidly cooled to ambient temperature, the so-
lid product was recovered from the cell, quickly rinsed with
distilled-deionized water and immediately dried under vac-
uum. Initial conditions and compositions of experimental
solutions and solids for each run are summarized in
Table 1.
2.3. Measurements and analyses
The pH of experimental solutions was determined by
monitoring the potential dierence between the Pt elec-
trodes immersed in the experimental and reference solu-
tions. The details of pH measurement are described in
previous publications (Wesolowski et al., 1998, 2000). Esti-
mated uncertainty of the pH measurements in this study is
less than 0.03. The pH in this study is dened as log[H
+
]
on the stoichiometric molal pH scale. A solution sample
withdrawn from the experimental solution was diluted to
0.110 ppm total Fe for analysis using ICP-AES (Thermo
Electron IRIS Intrepid II XSP) at ORNL. Precision better
than 1% of measured concentrations was routinely achieved
in repeated analyses of samples and standards. The associ-
ated standard errors (2r) are included as error bars in all
the gures showing Fe concentrations.
Solid samples were analyzed by SEM and XRD. SEM
images were obtained using a JEOL 6700 eld-emission
SEM at Penn State and a Hitachi S-4700 eld-emission
SEM at ORNL. XRD spectra of staring materials and
run products were obtained using a Bruker/Siemens
D5005 diractometer with CuKa X-ray radiation at
Table 1
Summary of experimental conditions and results.
Run# Initial
solution
Initial
solid
T (C) Duration
(days)
P
H2
(MPa) pH log[Fe]
log(mol/kg)
Run product
(mol fraction)
comments
1 0.001 m HTr
0.099 m NaTr
mt
*
150 4 ~0.1 ~4.5 ~3.3 100%mt
2 0.02 m HTr
0.085 m NaTr
mt 150 3.5 ~0.5 ~4.1 ~2.0 77%mt, 23%hm Acid titration
4 0.02 m HTr
0.085 m NaTr
mt 150 4 0.55 ~4.2 ~2.0 79%mt, 21%hm P
H2
change
5 0.001 m HTr
0.099 m NaTr
mt 150 9.5 ~1.5 4.45.8 2.6 to 5.3 95%mt, 5%hm Acid/base titration
6 0.02 m HTr
0.085 m NaTr
mt 150 13 ~0.7 4.27.3 2 to 3.6 95%mt, 5%hm Base titration
7 0.02 m HTr
0.085 m NaTr
mt 150 16 0.053.5 ~4.2 ~2.0 94%mt, 6%hm P
H2
change
8 0.02 m HCl
0.085 NaCl
mt 150 7 ~0.05 ~4.2 ~2.0 78%mt, 22%hm Cl system
9 0.02 m HTr
0.085 m NaTr
mt 200 11 0.055 ~3.6 ~2.0 89%mt, 11%hm P
H2
change
10 0.02 m HTr
0.085 m NaTr
mt 150 1 ~5 4.9 ~2.0 100%mt Short term
11 0.02 m HTr
0.085 m NaTr
hm
*
150 5 ~4 2.24.6 2.0 to 2.2 95%hm,
5%mt
12 0.02 m HTr
0.085 m NaTr
hm 150 15 ~4.5 2.05.3 2.0 to 3.5 80%hm, 20%mt Base titration
14 0.00025 m HTr
0.1m NaTr
mt 200 11 ~1 3.84.7 2.3 to 3.9 100%mt Acid/base titration
15 0.02 m HTr
0.085 m NaTr
mt 250 10 0.34 3.23.7 ~2.0 100%mt P
H2
change
18 0.001 m HTr
0.1 m NaTr
mt 100 21 ~3 3.76.4 ~3.1 Mt/Mh/
Fe(OH)
2
?
21 0.01 m HTr
0.1 m NaTr
mt 100 18 0.56 3.75.8 ~2.8 Mt/Mh/
Fe(OH)
2
?
P
H2
change
*
mt, magnetite; hm, hematite.
6144 T. Otake et al. / Geochimica et Cosmochimica Acta 74 (2010) 61416156
ORNL. Hematite and/or magnetite were the only identi-
ed minerals by XRD in all the starting materials and
run products. Note that maghemite (c-Fe
III
2
O
3
) and iso-
structural magnetite are not readily distinguishable from
the XRD spectra, particularly if maghemite is present in
small quantity relative to magnetite. However, maghemite
is thermodynamically unstable relative to hematite and
normally is observed only in low-temperature environ-
ments. Fractions of hematite and magnetite in the run
products at the end of each run were estimated from the
peak height ratio and those of standards with known frac-
tions of hematite and magnetite. The method is semi-quan-
titative and accuracy of the estimate is ~5 mol% as
mineral fractions, although some factors, such as dierence
in crystallinity or grain size, may cause the accuracy to be
worse. Standards and three run samples (runs #4, #7, and
#9) were also analyzed using a Rigaku DMAX-Rapid
Microdiractometer with MoKa X-ray radiation at Penn
State. In the estimate, peak area ratios were used instead
of peak height ratios. The results obtained at ORNL and
Penn State (PSU) were consistent, displaying a dierence
in the estimates of the three run samples of less than
3 mol% (#4: 21% (ORNL) vs. 21% (PSU), #7: 6% vs.
5%, and #9: 11% vs. 8%). Although the Mo source may
be superior for determining the fraction of hematite, we
used the Cu source results in the discussion for consistency.
We also calculated the fraction of hematite and magnetite
during hydrogen pressure change experiments (shown in
Fig. 2), based on the mineral fraction at the end of the runs
estimated by XRD and reduction rates of Fe
III
by hydro-
gen calculated in Section 4.1.7 at each temperature. Details
of the estimation method are described in Sections 4.1.4
and 4.1.7.
3. RESULTS
All the experimental data obtained, including those at
150 C (Otake et al., 2007), and those at 100, 200, and
250 C are presented in Supplementary data. The rst series
of experiments were intended to investigate the eect of
hydrogen pressure on the chemical compositions of experi-
mental solutions in magnetite experiments (runs #4, #7, #9,
and #15). No signicant change in pH or Fe concentration
was observed corresponding to the hydrogen pressure in-
crease of up to an order of magnitude at 150 C (Fig. 2a).
At the end of runs, both magnetite and hematite are identi-
ed in the run products by XRD analysis. The fraction of
hematite in the run products estimated from the XRD spec-
tra is 21 mol% for the 4-day run and 6 mol% for the 16-day
run. SEM images of the 4-day run product show that large
(~4 ~6 lm) ditrigonal dipyramidal crystals of hematite
formed (Fig. 3a), compared with the ~1 lm, euhedral mag-
netite starting material. Otake et al. (2007) conrmed that
the dipyramidal crystals are monolithic hematite by TEM
electron diraction patterns. In contrast, SEM images of
the long-term, 16-day run product show that hematite crys-
tals in the run product are slightly rounded and smaller, 3
5lm size (Fig. 3b).
When hydrogen pressure was increased from 0.05 to
~5 MPa after 6 days at 200 C (run #9), the pH of the solu-
tion slightly increased from 3.5 to 3.6 (Fig. 2b). No change
in Fe concentration was observed. Note that the total dis-
solved Fe concentrations in our experimental solutions
are much higher than the measured H
+
concentrations.
On the other hand, at 250 C, the pH spontaneously (i.e.,
without increasing hydrogen pressure) increased after
3 days, and reached the second steady state after 5 days at
pH 3.5 (Fig. 2c). Then, when hydrogen pressure was in-
creased from 0.2 to ~4 MPa after 6 days, the pH of the
solution increased from 3.5 to 3.7. XRD analysis indicates
that the run product at 200 C contains 11 mol% hematite.
The SEM images show that the hematite crystals at 200 C
are rounded and their grain size is smaller (23 lm; Fig. 3c)
than those found in the short-term, 4-day run product at
150 C. On the other hand, only magnetite (no hematite)
is identied in the 250 C run product by XRD and SEM
(Fig. 3d).
A second series of experiments were conducted to inves-
tigate the pH dependence of dissolved Fe speciation and
the transformation mechanisms of iron oxides. Otake
et al. (2007) performed magnetite + acid/base titration
experiments at 150 C, yielding the regression line of log[-
Fe] = 1.98pH + 6.28 in the log[Fe]pH diagram (thin so-
lid line in Fig. 4). The slope very close to 2 indicated that
the dominant species in the experimental solutions was
Fe
2+
. However, our new data obtained from a hema-
tite + base titration experiment (run #12) yielded a slope
of 1.4 in the log[Fe]pH diagram (bold solid line in
Fig. 4). In two runs (#11 and #12), hematite was used as
the starting material at 150 C (Fig. 5a). It should be
emphasized that, in these experiments, Fe was added only
as hematite (Fe
III
2
O
3
) into the system and no Fe
II
was ini-
tially present. It should be also noted that hematite starting
material was ner-grained (e.g., <0.5 lm; Otake et al.,
2007) than the magnetite starting material (e.g., ~1 lm).
We kept hydrogen pressure constant at ~5 MPa through-
out the experiments. Hematite experiments needed much
longer duration (~3 days) to reach the steady state, com-
pared with magnetite experiments. XRD analysis indicates
that the run products contain both hematite and magne-
tite, the latter in the form of much larger euhedral grains
clearly exhibiting the cubic crystal structure (cf. Otake
et al., 2007). Fractions of magnetite estimated by XRD
analysis are ~5 mol% for run #11, and 20 mol% for run
#12. At 200 C, acid/base titrations were performed in
magnetite experiments (Fig. 5b), yielding the regression
line of log[Fe] = 2.00pH + 5.44 in the log[Fe]pH dia-
gram (Fig. 6).
The third series of experiment was conducted to investi-
gate the possible presence of maghemite (c-Fe
III
2
O
3
) as a
precursor of hematite at a lower temperature, 100 C (runs
#18 and #21; Fig. 5c). While hydrogen pressure was kept
constant at ~3 MPa in run #18, it was increased from 0.8
to 6 MPa after 6 days in run #21. In both experiments, after
reaching the rst steady state at pH ~3.7 within 36 h, pH
gradually increased to ~6. While the pH increase occurred
spontaneously after 3 days in run #18, it was induced by the
hydrogen pressure increase after 8 days in run #21. No sig-
nicant change in Fe concentration was observed in either
experiment. XRD analysis indicates no hematite present
Iron oxide transformations in hydrothermal systems 6145
in the run products. This, however, does not preclude the
presence of maghemite, either as a separate phase or as
an iron-decient solid-solution with isostructural magne-
tite, as described in Section 2.3.
stage I stage II stage III stage I stage II stage III stage I stage III stage IV
6.0
7.0
8.0
9.0
-1.5
-1
-0.5
0
0.5
1
-1.5
-1
-0.5
0
0.5
1
p
H
l
o
g
P
H
2

(
M
P
a
)
l
o
g
[
F
e
]

(
m
o
l
/
k
g
)
(a) Magnetite + acid (150C) (b) Magnetite + acid (200C) (c) Magnetite + acid (250C)
1.5
2
3.5
4
~
~
1.5
2
3.5
4
~
~
E
s
t
i
m
a
t
e
d

m
i
n
e
r
a
l

f
r
a
c
t
i
o
n
Magnetite
Magnetite
Magnetite
Hematite
Hematite
Hematite
l
o
g
(
[
F
e
]
/
[
H
+
]
2
)
-2.02
-2.00
-1.98
-1.96
-1.94
-2.02
-2.00
-1.98
-1.96
-1.94
1.5
2
4
4.5
-1.5
-1
-0.5
0
0.5
1
H
2
added
H
2
added
H
2
added
Run #9 Run #15
~
~
-2.02
-2.00
-1.98
-1.96
-1.94
0
0.2
0.4
0.6
0.8
1
0 4 8 12 16
0
0.2
0.4
0.6
0.8
1
0 4 8 12
0
0.2
0.4
0.6
0.8
1
0 4 8 12
Time (day) Time (day) Time (day)
#4
#7
Run #4
Run #7
Mt-Hm
Mt dissolution
Mt-Hm
Mt dissolution
4.5
6.0
7.5
4.0
5.0
6.0
7.0
Mt-Hm
Mt dissolution
Fig. 2. Hydrogen pressure, solution compositions, and modeled mineral fractions of the hydrogen pressure change experiments at (a) 150 C,
(b) 200 C, and (c) 250 C runs. Magnetite was reacted with a 2.0 10
2
mol/kg triic acid + 8.5 10
2
mol/kg NaTriate solution in each
run. Hydrogen pressure was increased during the experiments. While no change in pH or Fe concentration was observed at 150 C, pH
increased corresponding to the P
H2
changes at 200 and 250 C. Dotted and Dashed-dotted lines are predicted solution compositions for non-
redox transformation between magnetite and hematite, and reductive dissolution of magnetite at the experimental hydrogen pressures,
respectively (Baes and Mesmer, 1976; Helgeson et al., 1978; Johnson et al., 1992; Shock et al., 1997). Shaded areas are interpreted as time
periods during which the solution compositions were not equilibrated. Modeled fractions of hematite and magnetite were estimated based on
the P
H
2
values and calculated eective rate constants that are obtained in Section 4.1.5. Moles of hematite formed by non-redox reaction were
estimated from pH change in the rst 24 h of each run. Hematite was formed with concomitant proton consumption in the rst ~24 h (Stage
I), then the fraction of hematite/magnetite reached the steady state values at low P
H2
pressures (Stage II). When P
H2
was increased, the
hematite fractions were signicantly decreased (Stage III). After all the hematite was consumed, chemical compositions of the solutions were
controlled by reductive dissolution of magnetite (Stage IV).
6146 T. Otake et al. / Geochimica et Cosmochimica Acta 74 (2010) 61416156
4. DISCUSSION
4.1. Non-redox and redox transformations between magnetite
and hematite in H
2
-rich hydrothermal solutions
4.1.1. Response of solution composition to P
H
2
change
To distinguish which of reactions (2), (3), or (5) controls
the chemical composition of the experimental solution at
each temperature, we conducted hydrogen pressure change
experiments during which P
H
2
was increased orders of mag-
nitude. Because hydrogen is involved in reductive dissolu-
tion of magnetite and hematite (i.e., reactions (2) and (3)),
the solution compositions resulting from the redox-equilib-
rium reactions are functions of hydrogen pressure:
log([Fe
2
[=[H

[
2
) = log Q
2
1=3 log P
H
2
(6)
and
log([Fe
2
[=[H

[
2
) = log Q
3
1=2 log P
H
2
(7)
where Q
2
and Q
3
are the stoichiometric molal solubility
quotients for reductive dissolution of magnetite and hema-
tite, respectively. On the other hand, the solution composi-
tions controlled by non-redox transformation between
magnetite and hematite are independent of hydrogen
pressure:
log([Fe
2
[=[H

[
2
) = log Q
5
(8)
where Q
5
is the solubility quotient for non-redox transfor-
mation between magnetite and hematite.
When Otake et al. (2007) varied hydrogen pressure from
0.05 to 5 MPa at 150 C, log([Fe]/[H
+
]
2
) values of the
experimental solution remained constant at 6.43 0.02
for 16 days (also shown in Fig. 2a). This clearly indicates
a lack of redox equilibrium during the experiments. The
log([Fe]/[H
+
]
2
) values agreed better with those predicted
for non-redox transformation of magnetite to hematite
(6.10), than those for reductive dissolution of magnetite
(7:57 1=3 log P
H
2
), and hematite (8:31 1=2 log P
H
2
),
based on available thermodynamic data (Baes and Mesmer,
1976; Helgeson et al., 1978; Johnson et al., 1992; Shock
et al., 1997). The Fe concentration and pH in titration
experiments at 150 C and P
H
2
=~ 1 MPa fall on the line:
log[Fe] = 1.98pH + 6.23 (Otake et al., 2007; also shown
in Fig. 4). This line was also more consistent with the line
predicted for non-redox transformation between magnetite
and hematite than that for reductive dissolution of magne-
tite or hematite under the P
H
2
conditions (Fig. 4). There-
fore, the results of both hydrogen pressure change
experiments and titration experiments suggest that the
chemical compositions of the experimental solution at
150 C are controlled by non-redox reaction (5).
At 200 C, log([Fe]/[H
+
]
2
) values of the experimental
solution were 5.11 at P
H
2
= 0:05 MPa, which were also
much closer to those calculated for the non-redox reaction
(4.98) than those predicted for the redox reactions at the
(c)
(a) (b)
(d)
Fig. 3. SEM images of (a) 150 C short-term run product (run #4), and (b) 150 C long-term run product (run #7), (c) 200 C run product
(run #9), and (d) 250 C run product (run #15). Scale bars are (a and b) 4 lm, and (c and d) 1 lm. The long-term run at 150 and 200 C run
show dissolution features of dipyramidal hematite crystals, compared with the short-term run at 150 C. No hematite crystals are observed in
250 C runs.
Iron oxide transformations in hydrothermal systems 6147
hydrogen pressure (i.e., reductive dissolutions of magnetite,
5.52, and hematite, 6.29; Fig. 2b). However, the log([Fe]/
[H
+
]
2
) values increased by ~0.2 log units when we increased
hydrogen pressure from 0.05 to 5.6 MPa. This increase in
log([Fe]/[H
+
]
2
), proportional to 1=10 log P
H
2
, is still much
less than the increase expected from reductive dissolution
of magnetite (proportional to 1=3 log P
H
2
) or hematite (pro-
portional to 1=2 log P
H
2
). The results suggest that the solu-
tion chemistry is controlled by the non-redox
transformation at 200 C. However, the slight response to
hydrogen pressure change appeared to reect that the redox
reaction started aecting the solution chemistry due to fas-
ter reaction rates, compared with 150 C.
In contrast, at 250 C, the log([Fe]/[H
+
]
2
) value was
~5.0 after 5 days at P
H
2
= 0:2 MPa, which was closer to
that calculated for reductive dissolution of magnetite
(4.72) at the experimental hydrogen pressure than that for
non-redox transformation of magnetite to hematite (4.07)
or reductive dissolution of hematite (5.55). When we in-
creased hydrogen pressure from 0.2 to 3.6 MPa, the
log([Fe]/[H
+
]
2
) value increased from 5.2 to 5.4 (Fig. 2c).
The results suggest that, at 250 C, reductive dissolution
of magnetite became important and aected the chemical
composition of the experimental solution. Note that, before
reaching ~5.2 after 5 days, log([Fe]/[H
+
]
2
) values rst
became stable at ~4.5 between 1 and 3 days. The temporary
steady state suggests that the solution chemistry was con-
trolled by the non-redox reaction between magnetite and
hematite only in the rst 3 days at 250 C.
4.1.2. Metastability of maghemite
Maghemite, which has a spinel structure and is often
formed as a result of oxidation of magnetite (Swaddle and
Oltmann, 1980), is thermodynamically unstable with respect
to hematite. However, because transformation of maghemite
to hematite is very slowat roomtemperature, maghemite is a
common mineral in nature, including weathered volcanic
rocks and some soils (Singer and Fine, 1989; Da Costa
et al., 1999;). Maghemite can be also formed by a reaction be-
tween magnetite and acidic solutions at a low temperature
(e.g., <150 C) by the forward directionof the following reac-
tion (Jolivet and Tronc, 1988; White et al., 1994):
Fe
3
O
4(mt)
2H

(aq)
Fe
2
O
3(mh)
Fe
2
(aq)
H
2
O (9)
This process can be viewed as leaching of Fe
2+
from mag-
netite by an acidic solution, leaving isostructural, but
iron-decient, maghemite behind. This is supported by
the studies of Jolivet and Tronc (1988) wherein room tem-
perature acid/base titrations of magnetite nanoparticles re-
sulted in reversible magnetitemaghemite transformations
(9) with no change in particle size or morphology. At low
temperature, maghemite may also form a solid-solution
with magnetite, such that within the same particle, some
portions of the particle are magnetite-rich, and other parts
maghemite-rich. The solubility quotient for reaction (9), Q
9
is described as follows:
log Q
9
= log([Fe
2
[=[H

[
2
) (10)
Recent calorimetric studies by Majzlan et al. (2003a,b) pro-
vide thermodynamic data for maghemite. According to
their results, the logQ
9
value is expected to be 5.13 at
100 C. During our 100 C experiments, we observed steady
states twice. The rst steady state occurred after 3 days in
run #18 and after 6 days in run #21. The log([Fe]/[H
+
]
2
) va-
lue (5.37) obtained during the rst steady state was signi-
cantly lower than predicted logQ
5
values at this
temperature (7.51), but are in reasonably good agreement
with the calculated value for logQ
9
values (Fig. 5c). This
observation indicates that maghemite was present during
the rst steady state in the 100 C experiments. The dier-
ence between the observed and predicted solubility quo-
tients for Q
9
corresponds to a change in the free energy
of formation of maghemite of only 2 kJ/mol at this temper-
ature, relative to the value of 736 kJ/mol for maghemite
reported by Majzlan et al. (2003a,b).
When both runs #18 and #21 reached to the second
steady state, the log([Fe]/[H
+
]
2
) values (e.g., 9.32 in run
#21) were consistent with those predicted for reductive dis-
solution of magnetite (e.g., 9.44; Fig. 5c), possibly indicat-
ing that reduction of Fe
III
in maghemite occurred within
14 days at 100 C. The apparent control of solution compo-
sitions by redox reactions and lack of hematite in the run
product may be explained by reductive dissolution of meta-
stable maghemite that occurred at a greater rate than that
of hematite in the higher temperature experiments:
-6
-5
-4
-3
-2
-1
4 4.5 5 5.5 6
Hematite + acid titration (150C)
l
o
g

[
F
e
]

(
m
o
l
/
k
g
)
pH
Run #11
Run #12 (bas e titration)
Non-redox transformation of Mt-Hm
Reductive dissolution of Mt (P
H2
=5MPa)
Reductive dissolution of Hm (P
H2
=5MPa)
Fig. 4. Solubility diagrams for titration experiments at 150 C.
Solid arrows indicate the pathways of titration in a hema-
tite + acid/base titration experiment. Open symbols are the results
of magnetite + hydrogen + acid/base titration experiments at
150 C, shown in Otake et al. (2007). While regression line (thin
solid line) produced by magnetite experiments has a slope of 2,
which was consistent with thermodynamically predicated non-
redox magnetite-hematite transformation line (dotted line), regres-
sion line produced by hematite experiment (bold solid line) has a
slope of 1.4. Dashed-dotted and long-dashed lines are the
predicted solubility lines for reductive dissolution of magnetite
and hematite at P
H2
= 5 MPa, respectively (Baes and Mesmer,
1976; Helgeson et al., 1978; Johnson et al., 1992; Shock et al.,
1997).
6148 T. Otake et al. / Geochimica et Cosmochimica Acta 74 (2010) 61416156
1=2Fe
2
O
3(mh)
2H

(aq)
1=2H
2(g)
Fe
2
(aq)
3=2H
2
O
(11)
Alternative possibility for yielding the high solubility quo-
tient toward the end of runs #18 and #21 is the precipita-
tion of Fe(OH)
2
on the surface of magnetite and
maghemite. There is a large uncertainty in the solubility
of Fe(OH)
2
predicated from the literature data, as shown
in Fig. 7. Ziemniak et al. (1995) postulated that Fe(OH)
2
controls the solubility of magnetite by covering its surface
at temperatures below 116 C under reducing conditions
(P
H
2
= 0:1 MPa at 25 C). If we consider the thermody-
namic properties of Fe(OH)
2
reported by Ziemniak et al.
(1995), it is possible that Fe(OH)
2
controlled the solution
composition at the second steady state observed in our
100 C experiments. However, other thermodynamic data
(e.g., Wagman et al., 1982) indicates much higher solubility
of Fe(OH)
2
at our experimental conditions, shown as the
upper boundary of Fe(OH)
2
stability eld in Fig. 7.
Fe(OH)
2
was not identied in our run products by SEM
or XRD analyses, but we did not conduct XPS studies of
the run products as was done by Ziemniak et al. (1995).
#18 #21
#18
#21
4
5
6
7
8
p
H
l
o
g
[
F
e
]

(
m
o
l
/
k
g
)
l
o
g
(
[
F
e
]
/
[
H
+
]
2
)
l
o
g
P
H
2

(
M
P
a
)
-1.5
-1
-0.5
0
0.5
1
1
2
3
4
5
6
-3.5
-3
-2.5
-2
-1.5
1
3
5
7
9
0 4 8 12 16
OH
-
added
OH
-
added H
+
added
H
2
added
H
2
added
H
2
added
-1.5
-1
-0.5
0
0.5
1
3
3.5
4
4.5
5
-4
-3.5
-3
-2.5
-2
0 4 8 12
-1.5
-1
-0.5
0
0.5
1
2
3
4
5
6
7
-4
-3.5
-3
-2.5
-2
Time (day) Time (day) Time (day)
Run #11
Run #12
Run #18
Run #21
Run #14
Mt-Hm
Mt dissolution
4
8
12
16
0 5 10 15 20 25
Hematite + base titration
(150C)
Magnetite + acid/base
titration (200C)
Magnetite + acid (100C)
(a) (b)
(c)
Mt-Hm
Mt dissolution
Hm dissolution
Mt-Mh
Mt-Hm
Mt dissolution
Fig. 5. Hydrogen pressure and solution compositions of (a) hematite experiments at 150 C, and magnetite experiments at (b) 200 C and (c)
100 C. Base and/or acid were added during the titration experiments. Predicted chemical compositions of the experimental solution are
indicated for non-redox transformation between magnetite and hematite (dotted lines), reductive dissolution of magnetite (dash-dotted lines),
reductive dissolution of hematite at the experimental hydrogen pressures (long-dashed lines), and non-redox transformation between
magnetite and maghemite (dashed lines), (Baes and Mesmer, 1976; Helgeson et al., 1978; Johnson et al., 1992; Shock et al., 1997). In hematite
experiments, Fe concentration decreased during base titrations. In magnetite titration experiment at 200 C, a base titration resulted in a
signicant change in the log([Fe]/[H
+
]
2
) value, which appeared to be controlled by the reductive dissolution of magnetite at the end of the run.
At 100 C, steady states were observed twice. The log([Fe]/[H
+
]
2
) values at the rst steady state were lower than those predicated for non-
redox transformation between magnetite and hematite, but were more consistent with those for non-redox transformation between magnetite
and maghemite. On the other hand, log([Fe]/[H
+
]
2
) values at the second steady state were close to those predicted for reductive dissolution of
magnetite. Shaded areas are interpreted as time periods during which the solution compositions were not equilibrated.
Iron oxide transformations in hydrothermal systems 6149
4.1.3. Equilibrium constants of non-redox transformations
Using the data obtained in hydrogen pressure change
experiments, we calculated the solubility quotients (Q
5
)
and equilibrium constants (K
5
) for non-redox transforma-
tion between magnetite and hematite at 150, 200, and
250 C (Fig. 8 and Table 2). Because solution compositions
were eventually aected by redox reactions at 200 and
250 C, we used only the pH data for the rst 3 days for
the calculation of logQ
5
and logK
5
values at these temper-
atures. Fig. 7 shows the calculated solubility quotients at
each temperature, which generally agreed well with those
calculated from available thermodynamic data, though
they dier from the predicted values by as much as 0.5
log units at 250 C. The discrepancy between observed
and predicted values for this equilibrium quotient increases
systematically with increasing temperature, which may be
related to approach of the system to rapid redox equilib-
rium at the higher temperatures, even during the rst
few days of reaction. Another potential source of error
may be the approximation method we employed to de-
scribe the activity coecient ratio of the reaction, in
which we assumed that the dependency of the ratio
(cFe
2+
)/(cH
+
)
2
on the ionic strength in NaTriate media
is identical to that obtained for the ratio (cZn
2+
)/(cH
+
)
2
of the well-studied ZnO
2
solubility reaction, conducted
by Wesolowski et al. (1998) using the same apparatus over
the same experimental conditions. Note that the data point
at 100 C indicates the solubility quotient for non-redox
transformation between magnetite and maghemite, which
is two orders of magnitude lower than that expected for
non-redox transformation between magnetite and hematite
at the same temperature.
4.1.4. Redox transformation of hematite to magnetite in long-
term experiments
The absence of hematite in the 250 C run product sug-
gests that reduction of Fe
III
occurred, which is consistent
with the interpretation of the solution chemistry (Sec-
tion 4.1.1). However, while the solution chemistry indicates
control by a redox-independent reaction (5) at 150 and
200 C, analyses of these run products suggest some reduc-
tion of Fe
III
also occurred at these temperatures. In a long-
term (16 days) magnetite experiment at 150 C (run #7),
the fraction of hematite in the run product estimated from
XRD spectra (6 mol%) is signicantly less than that esti-
mated from the dissolved Fe concentration and stoichiome-
try of the non-redox transformation (15 mol%). One mole of
magnetite is transformed to one mole of hematite upon
releasing one mole of Fe
2+
into the solution during the
non-redox transformation (i.e., reaction 5). Because
3.80 10
3
mol (0.88 g) of magnetite were initially added
and 5.82 10
4
mol of Fe were dissolved (1.06 10
2
mol/kg 0.0550 kg of solution), we expected 5.82 10
4
mol of hematite (15 mol% of the total solid) to be formed if
the non-redox transformation was the only reaction taking
place in the run. Similarly, the fraction of hematite is esti-
mated to be 25 mol% for a short-term, 4-day run at 150 C
(run #4) from the dissolved Fe concentration. However, in
this case, the estimate is relatively consistent with the
estimate based on the XRD analysis (21 mol%). Compared
with hematite crystals observed in the shot-term run
product (Fig. 3a), hematite crystals in the long-term run
product show more prominent dissolution features, such as
rounded edges or broken pieces of dipyramids (Fig. 3b).
However, stirring abrasion may also have been a factor in
the observed morphologic changes of the hematite reaction
products.
The fraction of hematite in the 200 C run product is
estimated as 11 mol% by XRD and 28 mol% by Fe concen-
tration. The SEM images show that hematite crystals are
rounded and their grain size is smaller (23 lm) than those
found in the 4-day run product at 150 C (Fig. 3c). There-
fore, the results of run product analyses (XRD and SEM) at
long-term experiments at 150 and 200 C indicate that
reductive dissolution of hematite occurred. A base titration
at 200 C and P
H
2
=~ 1 MPa (run #14) resulted in large pH
increase from 3.8 to 4.7 accompanied by a decrease in Fe
concentration from 5.3 to 1.5 10
3
mol/kg (Figs. 5b
and 6). However, the nal Fe concentration was much high-
er than that (~1 10
4
mol/kg) predicted from non-redox
transformation of hematite to magnetite. If non-redox
transformations between magnetite and hematite were com-
pletely reversible at 200C, 4.9 10
4
mol (22 mol% of to-
tal solid) of hematite would have remained in the run
product, based on the mass balance of added H
+
and
OH

and the stoichiometry of the non-redox transforma-

tion. However, no hematite was identied in the run prod-
uct and the solution chemistry was apparently controlled by
reductive dissolution of magnetite after the base titration.
-6
-5
-4
-3
-2
-1
3 3.5 4 4.5 5
Magnetite + acid/base titration (200C)
Run #14 (acid titration)
Run #14 (base titration)
Non-redox transformation of Mt-Hm
Reductive dissolution of Mt (P
H2
=1MPa)
Reductive dissolution of Hm (P
H2
=1MPa)
pH
l
o
g

[
F
e
]

(
m
o
l
/
k
g
)
Fig. 6. Solubility diagrams for titration experiments at 200 C.
Solid arrows indicate the pathways of titration. The bold solid line
with a slope of 2 is consistent with the acid titration data. When
base was titrated, the solubility was shifted toward that for
reductive dissolution of magnetite. Dashed-dotted and long-dashed
lines are the predicted solubility lines for reductive dissolution of
magnetite and hematite at P
H2
= 1 MPa, respectively (Baes and
Mesmer, 1976; Helgeson et al., 1978; Johnson et al., 1992; Shock
et al., 1997).
6150 T. Otake et al. / Geochimica et Cosmochimica Acta 74 (2010) 61416156
These observations suggest that at least 4.9 10
4
mol of
hematite were transformed to magnetite in 7 days.
The redox transformation of hematite to magnetite also
occurred in hematite experiments. In spite of no addition of
Fe
II
into the system, large euhedral magnetite crystals were
formed during the experiments. In run #11 (150 C, 5 days),
reaction between 2.9 10
3
mol (0.47 g) of hematite and
53 g of a 2.0 10
2
mol/kg acid solution produced
~5.5 10
4
mol of Fe
2+
and ~5 mol% of magnetite. From
these numbers, we can estimate that ~10 mol% of hematite
was reduced to Fe
II
in 5 days by reductive dissolution of
hematite to Fe
2+
(reaction (2)), and that ~20% of reduced
Fe
II
was consumed to form magnetite by non-redox trans-
formation of hematite to magnetite (reverse of reaction (2)).
4.1.5. A proposed model for non-redox and redox
transformations between magnetite and hematite
Based on the observations on changes in solution chem-
istry and mineralogy with time at dierent temperatures, we
propose a model for the mechanisms of non-redox and re-
dox transformations between magnetite and hematite under
H
2
-rich hydrothermal conditions. Our model is built on the
following assumptions: (i) magnetite is transformed to
maghemite by reaction with acidic solutions, accompanied
by release of Fe
2+
into solution; (ii) maghemite is trans-
formed to hematite at temperatures above 100 C, involving
dissolution/reprecipitation, whereas at 100 C the maghe-
mite persists for a period and then undergoes reductive
5MPa
0.05MPa
Temperature (C)
l
o
g
(
[
F
e
]
/
[
H
+
]
2
)
4
6
8
10
12
100 150 200 250
Experimental data
Non-redox transformation of Mt-Hm
Non-redox transformation of Mt-Mh
Reductive dissolution of Mt
Dissolution of Fe(OH)
2
Fig. 7. Calculated solubility quotients for (solid circle) non-redox transformation between magnetite and hematite at 150250 C (average of
all the data points of runs #2, #4, #5, #6, and #7 at 150 C, all the data points of run #9 and the rst seven data points of run #14 at 200 C,
and the rst three data points of run #15 at 250 C, respectively), and (solid square) transformation between magnetite and maghemite at
100 C (the rst three data point of run #21). The bars indicate standard deviation (2r) of the data at each temperature. Data plotted in
shaded areas in Figs. 2 and 5 and Supplementary data are excluded. Ranges of solubility quotient (solid diamond) for reductive dissolution of
magnetite at 100, 200, and 250 C, including all other data points in magnetite experiments, were also plotted. Predicted chemical
compositions of the experimental solutions for non-redox transformation between magnetite and hematite (dotted line), non-redox
transformation between magnetite and maghemite (dashed line), and reductive dissolution of magnetite (dashed-dotted line), at the
experimental hydrogen pressures, are shown (Baes and Mesmer, 1976; Helgeson et al., 1978; Johnson et al., 1992; Shock et al., 1997). Grey
shaded area is a possible equilibrium solubility range of Fe(OH)
2
. The lowest solubility curve was obtained from Ziemniak et al. (1995) and
the highest solubility curve was obtained from Wagman et al. (1982). At a lower temperature (e.g., 100 C), Fe(OH)
2
may have precipitated on
magnetite and controlled the chemical compositions of the experimental solution.
3
4
5
6
7
1.6 1.8 2 2.2 2.4 2.6
l
o
g
K
5
1000/T (K
-1
)
logK
5
= 4450/T - 4.38
Fig. 8. Calculated equilibrium constants of reaction (5) (i.e., non-
redox transformation between magnetite and hematite), based on
our experimental study, at 150250 C. Obtained R
2
value was
0.9998.
Iron oxide transformations in hydrothermal systems 6151
transformation to magnetite; (iii) magnetite crystals nucle-
ate by reaction of dissolved Fe
2+
and Fe(OH)
3(aq)
, and
hematite crystals rapidly form by precipitation of Fe(O-
H)
3(aq)
; (iv) reduction of Fe(OH)
3(aq)
by H
2
is slow, but
the rate increases as temperature increases. Our model fur-
ther indicates that Fe(OH)
3(aq)
is the dominant ferric iron
species in solution, but in most experiments was not a sig-
nicant species relative to Fe
2+
, shuttling rapidly between
solid phases during dissolution/precipitation processes.
The model is illustrated by the following chemical
reactions:
Fe
3
O
4(mt)
2H

(aq)
Fe
2
(aq)
Fe
2
O
3(mh)
H
2
O (12)
Fe
2
O
3(mh)
3H
2
O 2Fe(OH)
3(aq)
(13)
2Fe(OH)
3(aq)
Fe
2
O
3(hm)
3H
2
O (14)
Fe
2
O
3(hm)
3H
2
O 2Fe(OH)
3(aq)
(15)
2Fe(OH)
3(aq)
Fe
2
(aq)
Fe
3
O
4(mt)
2H

(aq)
2H
2
O
(16)
2Fe(OH)
3(aq)
H
2(g)
4H

(aq)
2Fe
2
(aq)
6H
2
O
(17)
When reaction (13) is fast, maghemite may not be recogniz-
able as a precursor of hematite. Then, magnetite is congru-
ently dissolved as Fe
2
(aq)
and Fe(OH)
3(aq)
by the combination
of reaction (12) and (13):
Fe
3
O
4(mt)
2H

(aq)
2H
2
O Fe
2
(aq)
2Fe(OH)
3(aq)
(18)
This reaction (18) may have occurred in magnetite experi-
ments above 100 C, because we did not observe any indi-
cations of maghemite present during the experiments.
However, the solution stoichiometries at T > 100 C indi-
cate rapid precipitation of released Fe(OH)
3(aq)
via reaction
(14) or reaction (16).
Combination of reactions (14) and (18) gives an overall
reaction of non-redox transformation of magnetite to
hematite (forward reaction (5)). The reaction mechanism
involving aqueous Fe
III
species explains the growth of large
euhedral hematite (Otake et al., 2007). Combination of
reactions (15) and (16) gives an overall reaction of non-re-
dox transformation of hematite to magnetite (reverse reac-
tion (5)). Combination of reactions (17) and (18) gives an
overall reaction of reductive dissolution of magnetite (reac-
tion (2)). Combination of reactions (15) and (17) gives an
overall reaction of reductive dissolution of hematite (reac-
tion (3)). Combination of reactions (15)(17) (3 reaction
(15) + 2 reaction (16) + reaction (17)) gives an overall
reaction of redox transformation of hematite to magnetite
(reaction (1)). In a hydrogen pressure change experiment
at 150 C (run #7), Fe concentrations and pH were con-
stant throughout the experiments (Fig. 2a), regardless of
whether or not reduction of Fe(OH)
3(aq)
occurred, as dis-
cussed in the previous section. This suggests that all Fe
2+
produced by reduction of Fe(OH)
3(aq)
(reaction (17)) was
reacted with hematite by non-redox transformation of
hematite to magnetite (reaction (5)) with no apparent
change in the overall solution chemistry.
On the other hand, in a hematite experiment at 150 C
(run#12), base titration yielded a slope of 1.4 in the
log[Fe]pH diagram (Fig. 4), indicating that Fe
2+
was not
the only dominant species in the solution. We speculate that
aqueous Fe
III
species (e.g., Fe(OH)
3(aq)
) were metastably
present in the experimental solution. If we assume Fe
2+
and Fe(OH)
3(aq)
were the two dominant species in the
experiment, ~30% of dissolved Fe would have been
Fe(OH)
3(aq)
based on the observed slope of 1.4. This is
much greater abundance than predicted at the equilibrium
P
[Fe
III
[=
P
[Fe
II
[ ratios (10
11
and 10
4
) under the experi-
mental conditions (Mesmer et al., 1970; Shock et al., 1997).
However, currently, there is no reasonable explanation why
the aqueous Fe
III
species were metastably present in this
hematite experiment.
4.1.6. Electron exchange on the surface of iron oxides
Aqueous Fe
II
species are known to facilitate dissolution
of Fe
III
-bearing minerals through surface adsorption and
electron exchange between Fe
II
and Fe
III
(e.g., Stumm
and Sulzberger, 1992; Pedersen et al., 2005, Yanina and
Rosso, 2008). Dissolution occurs by the following sequence:
Table 2
Calculated solubility quotients and equilibrium constants.
(a) Non-redox transformation of magnetitehematite (Fe
3
O
4(mt)
2H

(aq)
Fe
2
O
3(hm)
Fe
2
(aq)
H
2
O)
Solubility quotient
a
(logQ
5
)
Number of
samples
Uncertainity
(as 2r)
Ionic
strength
Activity correction
b
(logQ
5
/K
5
)
Equilibrium constant
a,c
(logK
5
)
150 C 6.4 37 0.2 0.12 0.25 6.1
200 C 5.3 14 0.3 0.12 0.30 5.0
250 C 4.5 3 0.1 0.12 0.38 4.1
(b) Non-redox transformation of magnetitemaghemite (Fe
3
O
4(mh)
2H

(aq)
Fe
2
O
3(hm)
Fe
2
(aq)
H
2
O)
Solubility quotient
a
(logQ
5
)
Number of
samples
Uncertainity
(as 2r)
Ionic
strength
Activity correction
b
(logQ
9
/K
9
)
Equilibrium constant
a
(logK
9
)
100 C 5.4 3 0.2 0.11 0.21 5.2
a
While logQ
5
and logQ
9
is dened as log([Fe
2
(aq)
[=H

(aq)
[
2
); log K
5
and logK
9
is dened as log(Fe
2
(aq)
=Fe
2
(aq)

2
).
b
Activity correction was made based on the assumption that dependency of the activity coecient ratio of reactants vs products on the ionic
strength is identical to that of well-studied ZnO solubility experiment by Wesolowski et al. (1998).
c
Regression line for logK
5
:logK
5
= 4450/T 4.38 (427K < T < 523K) (see Fig. 8).
6152 T. Otake et al. / Geochimica et Cosmochimica Acta 74 (2010) 61416156
BFe
III
OHFe
2
(aq)
= (Fe
III
OFe
II
)

(aq)
(19)
(adsorption of aqueous Fe
2+
on the Fe
III
-mineral surface)
= (Fe
III
OFe
II
)

= (Fe
II
OFe
III
)

(20)
(interfacial electron transfer)
(21)
(detachment of Fe
II
and Fe
III
from the surface =
dissolution)
where Fe
III
OH and (Fe
III
OFe
II
)
+
indicate sur-
face species. The rates of reactions (12), (15), and (18)
may be facilitated by the presence of Fe
2+
in the solution.
In both magnetite and hematite experiments, aqueous
Fe
2+
release into solution was associated with consumption
of H
+
in the initial solution. Hematite produced in magne-
tite experiments exhibits ditrigonal dipyramid morphology
(Fig. 3a), which is similar to hematite observed in a recently
study by Yanina and Rosso (2008). They reported the rapid
growth of monolithic hematite pyramids on the 001 basal
surface of hematite, with the c-axis perpendicular to the
001 surface, from anoxic solutions at room temperature,
in which Fe
2+
was the stable aqueous species. Yanina and
Rosso (2008) argued, based on novel manipulation of ex-
posed surfaces to the aqueous phase, that Fe
2+
ions adsorb
at the 001 surface and release electrons into the crystal,
leaving behind Fe
3+
ions that grow as the hematite pyra-
mids, in crystallographic registry with the underlying hema-
tite substrate. The electrons migrate to edge sites, where
they convert Fe
3+
ions of the bulk crystal surface into
Fe
2+
species, which are then released into solution to com-
plete the circuit. Based on similarity in the observed mor-
phology, the formation mechanism of hematite pyramids
proposed by Yanina and Rosso (2008) may be applicable
to formation of the hematite crystals observed in our
experiments.
4.1.7. Reduction kinetics of Fe
III
by H
2
According to the dierence in hematite fractions esti-
mated from XRD and Fe concentration, ~60% of hematite
initially formed through the non-redox reaction (~10 mol%
of total solid) was transformed to magnetite at 150 C after
16 days, and ~60% of hematite (~20 mol% of total solid)
was transformed to magnetite after 11 days at 200 C. At
250 C, hematite was completely transformed to magnetite
after 10 days, but likely transformed before 3.5 days when
the large pH increase from 3.2 to 3.6 (Fig. 2c) was observed.
The pH increase is interpreted as the transition from non-
redox transformation between magnetite and hematite to
reductive dissolution of magnetite as the predominant reac-
tion to control the solution chemistry. The transition would
have been caused by the complete redox transformation of
hematite to magnetite within 3.5 days. Based on the above
estimates, we calculated mineral fractions of magnetite and
hematite during the experiments (shown in Fig. 2). It was
assumed for the estimates that reduction kinetics of reac-
tion (17) is only dependent on hydrogen pressure by the rst
order, and that other parameters (e.g., pH, concentration of
aqueous Fe
III
species, surface area of hematite) are approx-
imately in a steady state, and therefore, can be treated as
constants. Then, we can estimate the rate equation as the
following:
R =
d[Fe
2
[
dt
= k
ef
P
H
2
(22)
where k
ef
= k [Fe(OH)
3(aq)
] [H
+
]
n
.
The rst order kinetics in terms of hydrogen pressure
refers to a previous study on reductive dissolution of goe-
thite by hydrogen (Fischer, 1987). The calculated rate con-
stants, k
ef
in units of mmol kg H
2
O
1
day
1
MPa
1
, were
0.2, 0.3, and 8 at 150, 200, and 250 C, respectively (Ta-
ble 3). Reduction rates of Fe
III
thus obtained in our exper-
iments are much faster than those obtained in previous
experimental studies (Kishima and Sakai, 1984; Seyfried
et al., 1987). They showed that redox equilibrium between
hematite and magnetite was established much more slowly
(~80 days at 300 C with an initial excess of H
2
,
P
H
2
> 0:1 MPa). This is presumably due to Pt electrodes
in our experiments catalyzing the reduction of aqueous
Fe
III
by hydrogen.
4.2. Implications
Regardless of the possible eect of Pt electrodes catalyz-
ing the reduction kinetics, our experimental results demon-
strated that non-redox transformations between magnetite
and hematite (and magnetite and maghemite) occurred over
a wide range of temperature and hydrogen pressure condi-
tions. Without an eective catalyst in natural systems, redox
kinetics may be much slower, and the non-redox transfor-
mations may control the chemical compositions of hydro-
thermal uids over much longer time scales. We propose
that transformation mechanisms of iron oxides revealed in
this study are applicable to various natural and laboratory
systems involving low-temperature hydrothermal uids,
which were also discussed partly in Otake et al. (2007).
Table 3
Calculated reduction rates of Fe
III
by H
2
.
Elapsed
time (day)
P
H2
(MPa) Hematite fraction
estimated by XRD (mol%)
Hematite fraction
estimated from [Fe] (mol%)
Rate constant (mmol kg
H
2
O
1
day
1
MPa
1
)
150 C (run #7) 16 0.053.5 6 15 0.2
200 C (run #9) 11 0.055 11 28 0.3
250 C (run #15) 10
*
0.34 0 27 8
*
All hematite is assumed to be transformed to magnetite after 3.5 days.
Iron oxide transformations in hydrothermal systems 6153
4.2.1. Transformation between hematite and magnetite in
various hydrothermal systems
Coexistence of hematite (or Fe
III
(hydr)oxides) and mag-
netite has been long recognized in various hydrothermal
environments: modern seaoor hydrothermal systems (Pot-
torf and Barnes, 1983; Zierenberg and Shanks (1983); Rona,
1984), volcanogenic massive sulde (Callow, 1967; Westen-
dorp et al., 1991) and oxide (Davidson, 1992; Duhig et al.,
1992) deposits, magnetite skarn-deposits (Buseck, 1966;
Zu rcher et al., 2001; Forster et al., 2004), Olympic Dam-type
(Fe oxideCuAuCo), porphyry Cu deposits (Gow et al.,
1994; Lynch and Ortega, 1997; Kolb et al., 2006), and
banded iron formations (BIFs; Ewers and Morris, 1981; La-
Berge et al., 1987; Pecoits et al., 2009). Some authors ques-
tioned whether redox equilibrium was attained between the
iron oxides, and whether enough oxidants/reducants were
present during the replacement of magnetite to hematite/
hematite to magnetite. In the Coxheath CuMoAu deposit,
Lynch and Ortega (1997) proposed the formation of tour-
maline by a reaction with albite and magnetite:
12NaAlSi
3
O
8(albite)
6Fe
3
O
4(mt)
6B(OH)
3(aq)
10H

(aq)
2NaFe
3
Al
6
(BO
3
)
3
Si
6
O
18
(OH)
4(schorl)
24SiO
2(qz)
6Fe
2
O
3(hm)
10H
2
O 10Na

(aq)
(23)
Although this reaction postulated to control the redox con-
ditions of the hydrothermal system by these authors, it is
also a redox-independent reaction; there is no change in
the oxidation state of iron. Therefore, the existence of mag-
netite and hematite may not have buered the P
O
2
, but in-
stead buered the pH.
For a seawaterbasalt interaction, Bischo and Dickson
(1975) carried out an experimental study reacting a basaltic
rock with a hydrothermal solution at 200 C and speculated
that the magnetite + hematite assemblage observed during
the experiment did not buer the P
O
2
, but buered the pH
and ferrous iron concentration in the uids, however, with-
out decisive evidence. SeyfriedandDing (1995) alsoindicated
that, in a theoretical treatment for understanding waterrock
interactions in mid-ocean ridges, the non-redox transforma-
tion is an important mass action reaction. Our current study
provides explicit evidence that non-redox transformations
betweenhematite (or Fe
III
(hydr)oxides) andmagnetite occur
under low temperature hydrothermal conditions, and, in
fact, commonly occur in natural environments.
Coexistence of magnetite and hematite is commonly ob-
served in typical oxide-type BIFs. Based on detailed geo-
chemical and mineralogical studies, Hoashi et al. (2009)
suggested that hematite is a primary mineral in ferruginous
cherts (i.e., low-grade oxide BIFs) of the 3.46 Ga Marble
Bar Chert, and that magnetite is a transformation product
of primary hematite. Several transformation processes have
been suggested for the transformation of hematite to magne-
tite in BIFs, including (a) reduction of hematite by organic
matter during metamorphism (e.g., Perry et al., 1973), (b)
biological iron reduction of hematite before diagenesis
(e.g., Johnson et al., 2008), (c) reaction between hematite
and siderite (Fe
II
CO
3
) during metamorphism (e.g., Burt,
1972). However, our experimental study showed that non-
redox transformation of hematite to magnetite is a plausible
process in these systems, by a reaction between primary
hematite and Fe
2+
-rich hydrothermal uids during early
diagenesis (Ohmoto, 2003; Otake et al., 2007). An important
implication in considering the origin of magnetite in BIFs is
that magnetite was likely formed under the conditions where
the system was far from redox equilibrium. Therefore, the
presence of magnetite in BIFs cannot be used to constrain
the atmospheric or oceanic oxygen levels during the deposi-
tion of BIFs. This contradicts the recent arguments by Ros-
ing et al. (2010) that magnetite and siderite in BIFs were in
near-equilibrium with atmospheric gas, and therefore, that
high atmospheric P
H
2
(log P
H
2
= 10
6
MPa) was sustained
during the deposition of BIFs.
4.2.2. Role of Pt as a catalyst of redox reactions
Previous experimental studies that investigated redox
reactions (e.g., solubility of redox-sensitive minerals, isotope
exchange reactions) encountered a diculty reaching the re-
dox equilibrium because of the slow reaction rates at
T < ~500 C. For example, Ohmoto and Lasaga (1982)
showed that redox equilibrium between SO
2
4(aq)
and
H
2
S
(aq)
, which is essentially the same as complete exchange
of sulfur isotopes between the species, is nearly attained (i.e.,
90% isotope exchange) in ~1 year at T = 250 C, pH 7, and
P
S = 10
2
m. The rates of CO
2(aq)
CH
4(aq)
redox reactions
are several orders of magnitude slower than those of
SO
2
4(aq)
H
2
S
(aq)
reactions (Ohmoto and Goldhaber, 1997).
Therefore, currently, solubility of many redox-sensitive
minerals (e.g., Cr
2
O
3
, UO
2
) under various redox conditions
is determined based on thermodynamic data obtained at
high temperatures and extrapolated to low-temperature
ranges, which could give a potential error in the solubility
data.
However, our experimental study surprisingly demon-
strates that redox equilibrium in the FeHO system was
attained in 3.5 day, much faster than reaction rates ob-
tained by Kishima and Sakai (1984) and Seyfried et al.
(1987) in a similar experimental set-up. The major dier-
ence between the two groups of experiments was the pres-
ence of Pt electrodes in our experimental system. The Pt
electrodes, which were coated with platinum black (i.e.,
nano-size Pt particles), were used for in situ pH measure-
ment. While the Pt electrodes catalyze the half-cell reaction
between hydrogen molecules and protons, they also appear
to catalyze reduction of Fe
III
by molecular hydrogen.
Therefore, we propose that, by adding nely-divided plati-
num in an experimental system, future experimental studies
will give us much more accurate solubility data of redox-
sensitive minerals as well as isotope fractionation factors
between species with various oxidation states (e.g., CO
2

CH
4
, N
2
NH
3
) in true redox equilibrium under low-tem-
perature conditions.
5. CONCLUSION
We conducted a series of experiments (100250 C) in
which hematite or magnetite was reacted with an acid solu-
tion under H
2
-rich hydrothermal conditions, with in situ
measurements of pH and partial hydrogen pressure (P
H
2
).
During the 116 day experiments, pH was varied from 1.7
6154 T. Otake et al. / Geochimica et Cosmochimica Acta 74 (2010) 61416156
to 7.3 and P
H
2
was varied from 0.05 to 5 MPa. The exper-
imental results are summarized as follows:
1. The formation of euhedral crystals of hematite in all
magnetite experiments at 150 C and some experiments
at 200 C, and euhedral crystals of magnetite in all
hematite experiments at 150 C, indicate that redox-
independent acidbase transformations between magne-
tite and hematite occurred.
2. Responses of log([Fe]/[H
+
]
2
) values to hydrogen pressure
changes indicate that, at 150 and 200 C, non-redox trans-
formation between magnetite and hematite controls the
solution chemistry. However, at 250 C, reductive disso-
lution of magnetite becomes important and aects the
solution chemistry. On the other hand, the presence of
maghemite was indicated from the chemical composition
of the experimental solution at 100 C in the rst 7 days.
3. A discordance in estimated fraction of hematite in the
150 and 200 C run products from the XRD spectra
and from the dissolved Fe concentrations suggests that
hematite formed through the non-redox process partially
transformed back to magnetite by a redox transforma-
tion. This redox transformation of hematite to magnetite
occurs in the following steps: (1) reductive dissolution of
hematite to Fe
2+
and (2) non-redox transformation of
hematite and Fe
2+
to magnetite. The reduction kinetics
was apparently facilitated by the Pt electrodes.
The results of this study suggest that non-redox trans-
formations between hematite magnetite occur in various
hydrothermal systems, particularly when redox kinetics
are sluggish under low-temperature conditions.
ACKNOWLEDGMENTS
The authors acknowledge H. Barnes, S. Brantley, P. Heaney, T.
Lasaga, K. Osseo-Asare, I. Johnson, and D. Bevacqua for valuable
comments on the early manuscript. The authors also acknowledge
J. Rosenqvist, M. Angelone, and J. Cantolina for technical assis-
tance. Comments by K.M. Rosso and two anonymous reviewers
are greatly appreciated. This project was supported by grants from
NASA Astrobiology Institute (NCC2-1057; NNA04CC06A) and
NSF (EAR-0229556) to H.O. D.J.W.s eort and a portion of
the eort of T.O. were supported by the U.S. Department of En-
ergy, Oce of Basic Energy, Geoscience Research Program, at
Oak Ridge National Laboratory, managed by UT-Battelle, LLC,
for the U.S. Department of Energy (DE-AC05-00OR22725).
APPENDIX A. SUPPLEMENTARY DATA
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/j.gca.2010.
07.024.
REFERENCES
Baes C. F. and Mesmer R. E. (1976) The Hydrolysis of Cations.
John Wiley & Sons.
Barnes H. L. (1997) Geochemistry of Hydrothermal Ore Deposits,
third ed. John Wiley & Sons Inc., New York.
Bischo J. L. and Dickson F. W. (1975) Seawater-basalt interaction
at 200 C and 500 bars: Implications for origin of sea-oor
heavy-metal deposits and regulation of seawater chemistry.
Earth. Planet. Sci. Lett. 25, 385397.
Burt D. M. (1972) The system FeSiCOH: a model for
metamorphosed iron formation. Carnegie Inst. Wash. Year
Book 19711972, 435443.
Buseck P. R. (1966) Contact metasomatism and ore deposition:
concepcion Del Oro, Mexico. Econ. Geol. 61, 97136.
Callow K. J. (1967) The geology of the Thanksgiving Mine, Baguio
District, Mountain Province, Philippines. Econ. Geol. 62, 472
481.
Da Costa A. C. S., Bigham J. M., Rhoton F. E. and Traina S. J.
(1999) Quantication and characterization of maghemite in
soils derived from volcanic rocks in Southern Brazil. Clays Clay
Miner. 47, 466473.
Davidson G. J. (1992) Hydrothermal geochemistry and ore genesis
of sea-oor volcanogenic copper-bearing oxide ores. Econ.
Geol. 87, 889912.
Duhig N. C., Stolz J., Davidson G. J. and Large R. R. (1992)
Cambrian microbial and silica-gel textures in silica iron exhalites
from the Mount Windsor Volcanic Belt, Australia their
petrography, chemistry, and origin. Econ. Geol. 87, 764784.
Ewers W. E. and Morris R. C. (1981) Studies of the Dales Gorge
member of the Brockman Iron Formation, Western Australia.
Econ. Geol. 76, 19291953.
Fischer W. R. (1987) Standard potential (E0) and iron (III) oxides
under reducing conditions. Z. Panzenerna hr. Bodenk. 150,
286289.
Forster D. B., Seccombe P. K. and Phillips D. (2004) Controls on
skarn mineralization and alteration at the Cadia Deposits, New
South Wales, Australia. Econ. Geol. 99, 761788.
Gow P. A., Wall V. J., Oliver N. H. S. and Valenta R. K. (1994)
Proterozoic iron oxide (CuUAuREE) deposits: Further
evidence of hydrothermal origins. Geology 22, 633636.
Helgeson H. C., Delany J. M., Nesbitt H. W. and Bird D. K. (1978)
Summary and critique of the thermodynamic properties of
rock-forming minerals. Am. J. Sci. 278, 229.
Hoashi M., Bevacqua D. C., Otake T., Watanabe Y., Hickman A.,
Utsunomiya S. and Ohmoto H. (2009) Primary haematite
formation in an oxygenated sea 3.46 billion years ago. Nat.
Geosci. 2, 301306.
Johnson C. M., Beard B. L., Klein C., Beukes N. J. and Roden E.
E. (2008) Iron isotopes constrain biologic and abiologic
processes in banded iron formation genesis. Geochim. Cosmo-
chim. Acta 72, 151169.
Johnson J. W., Oelkers E. H. and Helgeson H. C. (1992)
SUPCRT92: a software package for calculating the standard
molal thermodynamic properties of minerals, gases, aqueous
species, and reactions from 1 to 5000 bar and 0 to 1000 C.
Comp. Geosci. 18, 899947.
Jolivet J.-P. and Tronc E. (1988) Interfacial electron transfer in
colloidal spinel iron oxide. Conversion of Fe
3
O
4
cFe
2
O
3
in
aqueous medium. J. Colloid Interface Sci. 125, 688701.
Kishima N. and Sakai H. (1984) A simple gas analytical technique
for the Dickson-type hydrothermal apparatus and its applica-
tion to the calibration of MH, NNO and FMQ oxygen buers.
Geochem. J. 18, 1929.
Kolb J., Sakellaris G. and Meyer F. (2006) Controls on hydro-
thermal Fe oxideCuAuCo mineralization at the Guelb
Moghrein deposit, Akjoujt area, Mauritania. Miner. Deposita
41, 6881.
LaBerge G. L., Robbins E. I. and Han T.-M. (1987) A model for
the biological precipitation of Precambrian iron-formations
A: geological evidence. In Precambrian Iron-Formations (eds. P.
Iron oxide transformations in hydrothermal systems 6155
W. U. Appel and G. L. LaBerge). Theophrastus Publications,
Athens, pp. 6996.
Lindberg R. D. and Runnells D. D. (1984) Groundwater redox
reactions an analysis of equilibrium state applied to Eh
measurements and geochemical modeling. Science 225, 925
927.
Lynch G. and Ortega S. (1997) Hydrothermal alteration and
tourmalinealbite equilibria at the Coxheath porphyry Cu
MoAu deposit, Nova Scotia. Can. Mineral. 35, 7994.
Majzlan J., Grevel K. D. and Navrotsky A. (2003a) Thermody-
namics of Fe oxides: Part II. Enthalpies of formation and
relative stability of goethite (a-FeOOH), lepidocrocite (c-
FeOOH), and maghemite (c-Fe
2
O
3
). Am. Mineral. 88, 855859.
Majzlan J., Lang B. E., Stevens R., Navrotsky A., Woodeld B. F.
and Boerio-Goates J. (2003b) Thermodynamics of Fe oxides:
Part I. Entropy at standard temperature and pressure and heat
capacity of goethite (a-FeOOH), lepidocrocite (c-FeOOH), and
maghemite (c-Fe
2
O
3
). Am. Mineral. 88, 846854.
Mesmer R. E., Baes, Jr., C. F. and Sweeton F. H. (1970) Acidity
measurements at elevated temperatures: IV. Apparent dissoci-
ation product of water in 1 m potassium chloride up to 292. J.
Phys. Chem. 74, 19371942.
Ohmoto H. (2003) Nonredox transformations of magnetite
hematite in hydrothermal systems. Econ. Geol. 98, 157161.
Ohmoto H. and Goldhaber M. B. (1997) Isotopes of sulfur and
carbon. In Geochemistry of Hydrothermal Ore Deposits (ed. H.
L. Barnes), third ed. John Wiley & Sons, New York, pp. 517
611.
Ohmoto H. and Lasaga A. C. (1982) Kinetics of reactions between
aqueous sulfates and suldes in hydrothermal systems. Geo-
chim. Cosmochim. Acta 46, 17271745.
Otake T., Wesolowski D. J., Anovitz L. A., Allard L. F. and
Ohmoto H. (2007) Experimental evidence for non-redox
transformations between magnetite and hematite under H
2
-
rich hydrothermal conditions. Earth Planet. Sci. Lett. 257, 60
70.
Palmer D. A. and Hyde K. E. (1993) An experimental determina-
tion of ferrous chloride and acetate complexation in aqueous
solutions to 300 C. Geochim. Cosmochim. Acta 57, 13931408.
Pecoits E., Gingras M. K., Barley M. E., Kappler A., Posth N. R.
and Konhauser K. O. (2009) Petrography and geochemistry of
the Dales Gorge banded iron formation: paragenetic sequence,
source and implications for palaeo-ocean chemistry. Precam-
brian Res. 172, 163187.
Pedersen H. D., Postma D., Jakobsen R. and Larsen O. (2005) Fast
transformation of iron oxyhydroxides by the catalytic action of
aqueous Fe(II). Geochim. Cosmochim. Acta 69, 39673977.
Perry, Jr., E. C., Tan F. C. and Morey G. B. (1973) Geology and
stable isotope geochemistry of the Biwabiki Iron Formation.
Northern Minn. Econ. Geol. 68, 11101125.
Pottorf R. J. and Barnes H. L. (1983) Mineralogy, geochemistry,
and ore genesis of hydrothermal sediments from the Atlantis II
deep, Red Sea. Econ. Geol. Monogr. 5, 198223.
Rona P. A. (1984) Hydrothermal mineralization at Seaoor
spreading centers. Earth Sci. Rev. 20, 1104.
Rosing M. T., Bird D. K., Sleep N. H. and Bjerrum C. J. (2010) No
climate paradox under the faint early Sun. Nature 464, 744747.
Seyfried W. E. and Ding K. (1995) Phase equilibria in subseaoor
hydrothermal systems: a review of the role of redox, temper-
ature, pH, and dissolved Cl on the chemistry of hot spring uids
at mid-ocean ridges. In Seaoor Hydrothermal Systems: Phys-
ical, Chemical, Biological, and Geological Interactions (eds. S.E.
Humphirs, R.A. Zierenberg, L.S. Mullineaux and R.E. Thom-
son). Geophysical Monograph, vol. 91, American Geophysical
Union, Washington, DC. pp. 248272.
Seyfried W. E., Janecky D. R. and Berndt M. E. (1987) Rocking
autoclaves for hydrothermal experiments: II. the exible
reaction-cell system. In Hydrothermal Experimental Techniques
(eds. G. G. Ulmer and H. L. Barnes). John Wiley & Sons, New
York, pp. 216239.
Shock E. L., Sassani D. C., Willis M. and Sverjensky D. A. (1997)
Inorganic species in geologic uids: correlations among stan-
dard molal thermodynamic properties of aqueous ions and
hydroxide complexes. Geochim. Cosmochim. Acta 61, 907950.
Singer M. J. and Fine P. (1989) Pedogenic factors aecting
magnetic susceptibility of northern California soils. Soil Sci.
Soc. Am. J. 53, 11191127.
Stumm W. and Sulzberger B. (1992) The cycling of iron in natural
environments: considerations based on laboratory studies of
heterogeneous redox processes. Geochim. Cosmochim. Acta 56,
32333257.
Swaddle T. W. and Oltmann P. (1980) Kinetics of the magnetite
maghemitehematite transformation, with special reference to
hydrothermal systems. Can. J. Chem. 58, 17631772.
Wagman D. D., Evans W. H., Parker V. B., Schumm R. H., Halow
I., Bailey S. M., Churney K. L. and Nuttall R. L. (1982) The
NBS tables of chemical thermodynamic properties, selected
values for inorganic and c1 and c2 organic substances in SI
units. J. Phys. Chem. Ref. Data 11, 392.
Wesolowski D. J., Benezeth P. and Palmer D. A. (1998) ZnO
solubility and Zn
2+
complexation by chloride and sulfate in
acidic solution to 290 C with in-situ pH measurement.
Geochim. Cosmochim. Acta 62, 971984.
Wesolowski D. J., Machesky M. L., Palmer D. A. and Anovitz L.
M. (2000) Magnetite surface charge studies to 290 C from
in situ pH titrations. Chem. Geol. 167, 193229.
Westendorp R. W., Watkinson D. H. and Jonasson I. R. (1991)
Silicon-bearing zoned magnetite crystals and the evolution of
hydrothermal uids at the Ansil CuZn mine, Rouyn-Noranda,
Quebec. Econ. Geol. 86, 11101114.
White A. F., Peterson M. L. and Hochella, Jr., M. F. (1994)
Electrochemistry and dissolution kinetics of magnetite and
ilmenite. Geochim. Cosmochim. Acta 58, 18591875.
Yanina S. V. and Rosso K. M. (2008) Linked reactivity at mineral
water interfaces through bulk crystal conduction. Science 320,
218222.
Ziemniak S. E., Jones M. E. and Combs K. E. S. (1995) Magnetite
solubility and phase stability in alkaline media at elevated
temperatures. J. Solution Chem. 24, 837877.
Zierenberg R. A. and Shanks, III, W. C. (1983) Mineralogy and
geochemistry of epigenetic features in metalliferous sediment,
Atlantis II Deep, Read Sea. Econ. Geol. 78, 5772.
Zu rcher L., Ruiz J. and Barton M. D. (2001) Paragenesis,
elemental distribution, and stable isotopes at the Pen a Colorada
iron skarn, Colima, Mexico. Econ. Geol. 96, 535557.
Associate editor: Kevin M. Rosso
6156 T. Otake et al. / Geochimica et Cosmochimica Acta 74 (2010) 61416156