You are on page 1of 14

Geochimica et Cosmochimica Acta, Vol. 64, No. 19, pp.

3299 –3312, 2000


Copyright © 2000 Elsevier Science Ltd
Pergamon Printed in the USA. All rights reserved
0016-7037/00 $20.00 ⫹ .00
PII S0016-7037(00)00435-X

Forsterite surface composition in aqueous solutions: A combined potentiometric,


electrokinetic, and spectroscopic approach
OLEG S. POKROVSKY* and JACQUES SCHOTT
Géochimie: Transferts et Mécanismes, CNRS (UMR 5563)-OMP-Université Paul-Sabatier, 38 rue des Trente-Six Ponts, 31400 Toulouse, France

(Received November 5, 1999; accepted in revised form April 18, 2000)

Abstract—Surfaces of natural and synthetic forsterite (Fo91 and Fo100) in aqueous solutions at 25°C were
investigated using surface titrations in batch and limited residence time reactors, column filtration experi-
ments, electrokinetic measurements (streaming potential and electrophoresis techniques), Diffuse Reflectance
Infrared Spectroscopy (DRIFT), and X-ray Photoelectron Spectroscopy (XPS). At pH ⬍ 9, a Mg-depleted,
Si-rich layer (⬍20 Å thick) is formed on the forsterite surface due to a Mg2⫹ 7 H⫹ exchange reaction.
Electrokinetic measurements yield a pHIEP value of 4.5 corresponding to the dominance of SiO2 in the surface
layer at pH ⬍ 9. In contrast, surface titrations of fresh powders give an apparent pHPZC of about 10 with the
development of a large positive charge (up to 10⫺4 mol/m2 or 10 C/m2) in the acid pH region. This may be
explained by penetration of H⫹ into the first unit cells of forsterite surface. The surface charge of acid-reacted
forsterite is one or two orders of magnitude lower than that of unreacted forsterite with an apparent pHPZC at
around 6.5 and a pHIEP value of 2.1 which is close to that for amorphous silica and reflects the formation of
a silica-rich layer on the surface. XPS analyses indicate the penetration of hydrogen into the surface and the
polymerization of silica tetrahedra in this leached layer. At pH ⬎ 10, a Si-deficient, Mg-rich surface layer is
formed as shown by XPS analyses and the preferential Si release from the surface during column filtration
experiments. Copyright © 2000 Elsevier Science Ltd

1. INTRODUCTION (2000) to propose that forsterite dissolution was controlled by


the adsorption of hydronium ions on the surface. In contrast,
There are several arguments for revisiting the surface chem-
Luce et al. (1972) and Blum and Lasaga (1988) reported that a
istry and reactivity of forsterite in aqueous solutions. This
reversible exchange of 2H⫹ for Mg2⫹ occurred on the forsterite
mineral belongs to the olivine group which is a solid solution
surface during its dissolution under acidic conditions. Simi-
series (Fe, Mg, Mn, . . . , Me2⫹)2SiO4 with fayalite, forsterite,
larly, XPS analyses of acid-reacted fayalite, Fe2SiO4, (Schott
and tephroite being the pure Fe, Mg, and Mn end members,
and Berner, 1983; Schott and Berner, 1985) and tephroite,
respectively. Forsterite is a common mineral that exerts an
Mn2SiO4, (Casey et al., 1993a) evidenced an exchange reaction
important control on ultramafic and basaltic rock weathering.
between H⫹ and Fe2⫹ and H⫹ and Mn2⫹, respectively, indi-
As a result, it is an important buffer of surface water pH
cating the formation of leached layers at the surfaces of these
involved in atmospheric CO2 regulation (Barnes and O’Neil,
minerals. Comprehensive analysis of these data does not lead to
1969; Holland, 1978; Brady and Gislason, 1997). Numerous
a clear picture of the forsterite dissolution mechanism. Forster-
field observations (Banfield et al., 1990; Banfield et al., 1991)
ite dissolution can be controlled either by a simple protonation
indicate that forsterite weathering occurs in intimate contact
of its surface as is the case for oxides (Stumm, 1992) or by an
with secondary minerals where topotactic reactions proceed in
exchange reaction between H⫹ and constitutive metals (Mg or
small channels (⬃30 Å diameter) on the surface. Therefore,
Fe) as for other silicate minerals (Oelkers et al., 1994; Schott
knowledge of forsterite surface chemistry and rate limiting surface
and Oelkers, 1995). In order to distinguish between possible
reactions is required for modeling its overall chemical weathering
different mechanisms of forsterite dissolution, a detailed study
rate. Available data on its surface chemistry and dissolution mech-
of forsterite surface speciation and reactivity in aqueous solu-
anism, however, are still scarce and contradictory.
tions has been performed. This paper presents a description of
From surface titrations performed in batch reactors, Blum
the forsterite-water interface obtained by combining surface
and Lasaga (1988) and Wogelius and Walther (1991) found an
titrations, column filtration experiments, electrokinetic mea-
important charge build up (10⫺5–10⫺4 mol/m2) on the forster-
surements, and surface spectroscopy analyses. A companion
ite surface comparable to that developed on feldspar surfaces.
paper will focus on the interpretation of forsterite dissolution
At the same time, Westrich et al. (1993) and Casey et al.
experiments using the data on forsterite surface chemistry
(1993b), using energetic ion beam analyses, were not able to
obtained in this study.
detect H⫹ penetration into the forsterite lattice or the formation
of a cation-leached layer as is the case for other silicates.
Application of these results for interpreting forsterite dissolu- 2. EXPERIMENTAL
tion in acid solutions led Brady and Walther (1989), Wogelius 2.1. Mineral Samples
and Walther (1991), Oelkers (1999), and Rosso and Rimstidt
Forsterite (Fo91) from San Carlos (Arizona) obtained from Ward Co.,
and a synthetic forsterite (Fo100) were used in this study. Samples of
*Author to whom correspondence should be addressed both forsterites were analyzed with electron microprobe by the WDS
(oleg@lucid.ups-tlse.fr). technique using 15 KeV accelerating voltage. The composition of San
3299
3300 O. S. Pokrovsky and J. Schott

Table 1. Chemical composition (weight%) of the forsterite samples The pressure difference across the column (⌬P, mbar) together with the
used in this study. electrical potential developed between the two Ag-AgCl electrodes
(⌬E, mV) were measured as a function of time. At the end of each
Oxide Natural Fo91 Synthetic Fo100 experiment, a sample was taken for pH, [SiO2(aq)]tot, and [Mg2⫹]tot
determinations. The streaming potential (␨, mV) was determined from
SiO2 40.81 43.04 the slope of the plot of E against P according to the Helmholtz -
Al2O3 0.01 0.01 Smoluchowsky relation (Hunter, 1989):
Cr2O3 0.02 0.04
FeO 9.12 0.08 ⌬E/⌬P ⫽ ␨ 䡠 ␧/(4␲␩␭)
MnO 0.09 0.00
MgO 49.48 56.38 where ␧, ␩, and ␭ represent the dielectric constant, viscosity coefficient
CaO 0.08 0.55 and specific conductivity of the solution, respectively. Measurements
NiO 0.43 0.00 were performed at 25 ⫾ 2°C and under four to five different pressure
Total 100.05 100.10 gradients with fluid flow in different directions. At least 300 experi-
mental points (E vs. P) were collected for each run. The average
uncertainty in each experimental run was less than 5% whereas the
reproducibility of ␨ for similar solution compositions was about 10%.
Carlos forsterite and the synthetic forsterite as obtained from 10 elec- The forsterite powder was equilibrated with 1 L of solution for several
tron microprobe analyses is shown in Table 1. X-ray diffraction anal- hours before conducting the measurements. About 30 min of solution
ysis (XRD) confirmed that the samples are well-crystallized forsterite circulation was required for E to reach a stable value within ⫾0.2 mV
without any contaminant clay minerals. XRD analyses of reacted before starting the measurements. Most experiments were performed at
forsterite indicate that there was no compositional or structural change pH values from 3 to 11 at a constant ionic strength of 0.01 or 0.001 M
during the experiments. Both forsterite samples were ground to a fine (NaCl). Several measurements at higher ionic strength in other solu-
powder (1–10 ␮m) in an agate mortar. Portions of freshly ground tions (0.005 to 0.01 M MgCl2 and 0.001 to 0.01 M NaHCO3) were
synthetic Fo100 were annealed for 24 h at 500 and 900°C. The surface made to account for the effect of magnesium and bicarbonate ions on
areas of the resulting powders were measured by N2 adsorption using the Fo91 streaming potential.
the B.E.T. method. These fresh forsterite powders were aged at least 1 A microelectrophoremeter (“Zetaphoremetre II”, model Z3000
day in aqueous solutions at neutral pH prior to the surface titrations and SEPHY S.A.R.L.) was used to measure the electrophoretic mobilities
electrophoretic measurements. Two types of acid-reacted forsterite and zeta-potentials of forsterite particles at 25 ⫾ 1°C. Forsterite parti-
powders were obtained by immersion of a freshly ground powder in cles were suspended in aqueous solution (3.0 g/L) for 1 day. For each
solutions at pHs of 5 and 2 for 24 h. It should be noted that such an acid sample, approximately 30 mL of supernatant (containing mineral par-
treatment induced a notable increase in forsterite specific surface area, ticles of ⬍10 ␮m grain size) were used to fill the microelectrophoresis
consistent with similar observations made on acid-reacted wollastonite cell. Experiments were performed in solutions of different composition
(Rimstidt and Dove, 1986; Xie and Walther, 1994). The 100 –200 and with ionic strengths ranging from 0.001 to 0.01 M (NaCl, NaHCO3,
200 –500 ␮m size fractions were separated from freshly ground Fo91 MgCl2), [Mg2⫹]tot from 10⫺6 to 10⫺3 M, and pH from 2.7 to 11.4.
and cleaned ultrasonically for streaming potential measurements and Immediately after the electrokinetic measurements, pH was measured
column filtration experiments. For X-ray Photoelectron Spectroscopy and a sample was taken for [SiO2(aq)]tot and [Mg2⫹]tot determination.
(XPS) analyses, a 50 –100 ␮m fraction was sieved, cleaned ultrasoni- At least three electrophoretic mobility measurements were performed
cally in alcohol several times, and dried overnight at 50°C. The BET on each sample. The uncertainty ranged from 5 to 20%, being the
surface areas for the samples used in the different experiments of this highest near the isoelectric point. Electrophoretic mobilities were con-
study are given in Table 2. verted to ␨-potentials using the Smoluchowski equation.

2.2. Electrokinetic Measurements 2.3. Surface Titrations

A ZETACAD (CAD Inst., France) zeta-meter was used for stream- Surface titrations were conducted at 25 ⫾ 0.5°C under a N2 atmo-
ing potential measurements. This apparatus measures the electrical sphere in a batch reactor and a limited residence time (LRT) reactor
potential generated by the imposed movement of a solution through a described in Pokrovsky et al. (1999). Blank and sample titrations were
column of porous solid. This electrical potential has the same sign as performed by addition of HCl or NaOH in a background electrolyte of
that of the solid surface. In each experiment, one liter of solution was NaCl (1.0, 0.01, and 10⫺4 M). Titrations were performed on freshly
pumped using N2 gas pressure through the horizontally-oriented col- ground powders, grains annealed at 500 or 900°C, and acid-reacted
umn of forsterite powder (approx. 20 g) in a Pyrex tube 15 cm long and powders. For batch reactor experiments, 5–10 g of powder were sus-
1 cm in diameter containing two spiral Ag-AgCl electrodes at each end. pended in 100 mL of an electrolyte solution giving resulting surface

Table 2. Forsterite mineral samples used in this study.

Surface
Solid area, m2/g Type of experiments

Fo91 San Carlos ground (fresh) 2.78 Surface titration, electrophoresis,


XPS, DRIFT
Fo91 San Carlos acid-reacted 1 (pH 5) 1.92 Surface titration, XPS
Fo91 San Carlos acid-reacted 2 (pH 2.7) 4.27 Surface titration, electrophoresis,
XPS, DRIFT
Fo91 San Carlos ground, annealed 24 hr at 900°C 1.49 DRIFT
Fo100 synthetic ground (fresh) 1.19 Surface titration, DRIFT
Fo100 synthetic acid-reacted at pH 3 2.93 Surface titration
Fo100 synthetic ground, annealed 24 hr at 500°C 1.23 Surface titration, DRIFT
Fo100 synthetic ground, annealed 24 hr at 900°C 0.619 Surface titration, DRIFT
Fo91 San Carlos 50–100 ␮m, ultrasonically cleaned 0.800 Dissolution kinetics, XPS
Fo91 San Carlos 100–200 ␮m, ultrasonically cleaned — Streaming potential, column filtration
Fo91 San Carlos 200–500 ␮m, ultrasonically cleaned — Column filtration
Forsterite surface composition in aqueous solutions 3301

Table 3. Experimental conditions of forsterite surface titrations.

Type of Number Surface


No. reactor Type of solid of points area, m2/L I, M pH ⌺ Mg, M ⌺ Si, M

1 Batch Fresh Fo91 10 251 0.01 10.0–6.8 0.0024–0.043 1.9 ⫻ 10⫺4–0.0039


2 Batch Fresh Fo91 17 26 0.01 10.0–3.4 7.3 ⫻ 10⫺4–0.0044 2.4 ⫻ 10⫺4–0.0016
3 Batch Fresh Fo91 16 67 0.01 9.7–6.0 7.9 ⫻ 10⫺4–0.007 1.1 ⫻ 10⫺4–0.002
4 Batch Fresh Fo91 15 57 1.0 10.0–3.8 0.0013–0.009 2.6 ⫻ 10⫺4–0.0038
5 Batch Fresh Fo91 17 26 1.0 10.0–5.9 0.001–0.0032 3.4 ⫻ 10⫺4–0.0014
6 Batch Fresh Fo91 10 0.45 0.01 8.6–1.8 1.1 ⫻ 10⫺4–2.8 ⫻ 10⫺4 1.6 ⫻ 10⫺4–2.4 ⫻ 10⫺4
7 Batch Fo91 acid–reacted 1 14 161 0.01 9.5–4.5 3.7 ⫻ 10⫺4–0.0031 8.2 ⫻ 10⫺4–0.0019
8 LRT Fresh Fo91 22 36 0.01 11.2–8.0 5.6 ⫻ 10⫺5–0.0048 8.7 ⫻ 10⫺5–0.0023
9 LRT Fo91 acid–reacted 1 26 19 0.01 10.8–2.7 4.6 ⫻ 10⫺5–2.6 ⫻ 10⫺4 1.74 ⫻ 10⫺4–2.0 ⫻ 10⫺4
10 LRT Fo91 acid–reacted 2 38 242 0.01 11.3–2.0 4.0 ⫻ 10⫺5–3.2 ⫻ 10⫺4 6.2 ⫻ 10⫺4–7.4 ⫻ 10⫺4
11 LRT Fresh Fo100 32 43 0.01 11.2–3.0 7.0 ⫻ 10⫺5–0.01 2.8 ⫻ 10⫺5–4.2 ⫻ 10⫺3
12 LRT Fo100 annealed at 500°C 30 23 0.01 11.4–2.5 3.4 ⫻ 10⫺5–2.6 ⫻ 10⫺3 5.3 ⫻ 10⫺5–9.7 ⫻ 10⫺4
13 LRT Fo100 annealed at 900°C 9 24 0.01 10.8–5.9 2.8 ⫻ 10⫺5–7.6 ⫻ 10⫺4 3.6 ⫻ 10⫺5–2.6 ⫻ 10⫺4

The arrow indicates the direction of titrations for batch reactors.


LRT ⫽ limited residence time reactor.
Acid–reacted 1 ⫽ Fo91 powder treated at pH 5.
Acid–reacted 2 ⫽ Fo91 powder treated at pH 2.5.

areas from 120 to 250 m2/L. In LRT reactors, the forsterite surface area Non-monochromatic twin Al K␣ X-rays (h␯ ⫽ 1486.6 eV) were used
varied from 24 to 240 m2/L depending on the type of powder and as the excitation source at a power of 200 Watts. An analyzing pass
solid/solution ratio. The powder surface area did not change (within energy of 50 eV with a step size of 1 eV was used for survey scans. For
10%) during short term surface titration as shown by N2 B.E.T. mea- the regional narrow scans, a 20 eV analyzing pass energy with a step
surements of the reacted powder. The suspension was titrated by size of 0.1 eV was used. Sample preparation and analysis were identical
0.01–1.00 M HCl or NaOH in the 2 to 11.5 pH range. For both titration to those previously described in Schott et al. (1981). The relative
techniques, a potential drift of less than 1 mV/min at pH ⬍ 6 and 0.5 abundances of elements at the forsterite surface were obtained from
mV/min at 6 ⬍ pH ⬍ 11 was taken as a steady-state condition for each measured peak areas and Scofield sensitivity factors (Scofield, 1976)
titration point. In the batch reactor, the time between each new addition for Mg2s, Si2p, and O1s, determined on oxides. The Mg/Si atomic ratio
of acid or base varied from 10 to 40 min. With the LRT reactor, the in the near-surface region of each powder was determined by measur-
residence time of the suspension varied from 1 to 8 min depending on ing the intensities (areas) of the Mg2s and Si2p lines. These lines are
pH and the type of powder investigated. Once steady-state was reached, very close in energy and allow for the analysis of both elements to the
pH was recorded, and a solution sample taken and immediately filtered same depth in the sample. It must be noted that the XPS spectra probe
and stored for Mg and Si analysis. Altogether, fourteen titration runs only the most shallow surface layers of the mineral: 63% of the signal
were performed, with 15–30 experimental points for each run. The originates from less than 26 Å, and 95% from less than 78 Å (Hochella
experimental conditions are listed in Table 3. and Carim, 1988).

2.4. Column Filtration Experiments 2.7. Diffuse-Reflectance Infrared Fourier-Transformed (DRIFT)


Spectroscopy
Filtration experiments were carried out in columns (15–30 cm in
length with a diameter of 1 cm) filled with 15 to 30 g of ultrasonically- The diffuse reflectance (DRIFT) spectra were recorded on a Bruker
cleaned forsterite powder 100 –200 or 200 –500 ␮m in grain size. The IFS88 FTIR spectrometer with an MCT detector by means of a diffuse
columns were constructed from transparent plastic in order to check for reflectance attachment. All the accessories were from Harrick Scientific
homogeneity of particle packing and the absence of air bubbles. Solu- Co. The spectrometer was purged with CO2-free dry air (Balston
tions of different initial compositions passed through the column with Filter). 50 mg of dry powder were dispersed in 350 mg of KBr and
flow rates ranging from 0.5 to 6 mL/min, were filtered at the end of the lightly packed into a 5 mm diameter microsampling cup. The spectra
column through a 0.45 ␮m membrane filter and then collected for pH were taken at 4 cm⫺1 resolution by co-adding up to 200 scans in the
measurements and Mg and Si analysis. Such a design allows perfor- 4000 –500 cm⫺1 region. The unit of intensity was defined as ⫺log(R/
mance of experiments with characterized powder of high surface area R0) where R0 and R are the reflectivities of the system without and with
without interferences from adherent small particles, surface defects, the investigated sample, respectively. The contribution of atmospheric
and amorphous layers such as those produced by the grinding of water was always subtracted from the spectra.
samples for surface titration experiments.
3. RESULTS
2.5. Analyses
3.1. Electrokinetic Study
NIST buffers (pH ⫽ 4.008, 6.865, and 9.180 at 25°C) were used for
calibration of a combination pH-electrode (Schott Geräte H62). Preci- The values of ␨-potentials for fresh Fo91 measured by
sion of the pH measurement was ⫾0.002 units. Magnesium concen- streaming potential and electrophoresis techniques are listed in
tration was measured by flame atomic absorption spectroscopy with an Appendix A. Both methods gave similar values for the forster-
uncertainty of 1%. Total silica concentration was determined using the
molybdate blue method with an uncertainty of 2%. Alkalinity was ite isoelectric point, i.e., pHIEP⫽ 4.4 ⫾ 0.1 (Fig. 1). This value
determined following a standard HCl titration procedure with an un- is consistent with those determined by Deju and Bhappu (1966)
certainty of 1% and a detection limit of 5 䡠 10⫺5 M. (pHIEP⫽ 4.1 at 25°C) and Ishido and Mizutani (1981) (pHIEP⫽
5.3 at 40°C) from streaming potential measurements. As for
2.6. X-ray Photoelectron Spectroscopy (XPS) other silicates, forsterite zeta potential decreases with increas-
XPS analyses were conducted on an ESCALAB VG 220i-XL spec- ing pH; at above pH 8, however, ␨ becomes independent of pH.
trometer. A standard hemispherical detector was used in this study. Increase of ionic strength from 0.001 to 0.01 M and addition of
3302 O. S. Pokrovsky and J. Schott

3.2. Surface Titrations


The total amount of protons consumed during forsterite
titration ([Hs⫹]) is the sum of H⫹ consumption due to the
following reactions:

1. Stoichiometric dissolution of the bulk solid ([Hdis ]) releas-
ing both Mg and Si to the solution according to (for sim-
plicity, Fe is omitted):

Mg2SiO4(s) ⫹ 4H⫹(aq) ⫽ 2Mg2⫹(aq) ⫹ H4SiO4(aq)


(1)

2. Stoichiometric exchange of 2 H (aq) for one Mg atom on
the surface ([H⫹
ex]) releasing only Mg to the solution (Luce et
al., 1972; Blum and Lasaga, 1988):

2H⫹(aq)⫹ ⬎Mg⫺X 7 Mg2⫹(aq)⫹ ⬎2H⫺X (2)


where ⬎2H⫺X represents an exchanged surface site and
3. Adsorption/penetration of nH⫹ or OH⫺ in the surface

layer(s) ([Hads ]) resulting in surface charge development. This
Fig. 1. Influence of pH on the ␨—potential of fresh forsterite (San reaction should occur on ⬎2H⫺X exchanged sites :
Carlos Fo91).
nH⫹(aq)⫹ ⬎2H⫺X ⫽ ⬎Hn⫹2⫺X⫹ (3)
As will be shown below from the data obtained in the batch
bicarbonate ions (up to 10⫺3 M) at pH below 8 have little effect
surface titrations, the value of n in reaction (3) is close to 0.5.
on the ␨-potential. In contrast, the addition of Mg2⫹ leads to an
Consequently, proton mass balance can be expressed as
increase of ␨ at pH above 6. The isoelectric point of acid-
reacted Fo91 moves towards acidic pH (pHIEP ⫽ 2.1) as deter- 关H s⫹兴 ⫽ 关H dis

兴 ⫹ 关H ⫹ ⫹
ex兴 ⫹ 关H ads兴 (4)
mined by the electrophoresis technique (Fig. 2). This is in
agreement with a shift of pHIEP to more acid pHs observed for Monitoring of pH, and Mg and Si concentrations in each
other silicates following their acid treatment (Cases, 1966; titration step allows us to distinguish between these three
Cases, 1967; Parks, 1967). Note also that the values of ␨-po- reactions and also calculating the net consumption of H⫹/OH⫺
tential measured at pH ⬍ 2.1 for acid-reacted samples are much due solely to reaction (3). For example, the amount of protons
lower than those measured for fresh powder. This reflects a consumed by the adsorption/penetration reaction can be calcu-
lated by
strong decrease of acid-reacted surface charge and potential
compared to fresh forsterite. ⫹
[Hads ]⫽ [Hs⫹] ⫺ 10⫺pH/␥H⫹ ⫺ 4䡠⌬[SiO2]total⫺ 2⌬[Mg2⫹]exch
(5)
where ␥H⫹ is the activity coefficient of H⫹ calculated using
Davis equation,

关Hs⫹兴 ⫽ 共 ␯ i /V i) 䡠C HCl (6)


and

⌬[Mg2⫹]exch ⫽ ⌬[Mg2⫹]total ⫺ 2⌬[SiO2]total (7)


In these equations, ⌬[j] stands for the change of concentration
of j-th component compared to its initial value, ␯i and Vi are the
volumes of added titrant (HCl or NaOH) and total solution
volume in reactor at the time of i-th addition, respectively, and
CHCl is the molarity of titrant.
Finally, surface charge (coulombs/m2) is given by

␴app ⫽ F 䡠 [Hads ]/s (8)
where F is the Faraday constant (96493.5 couloumbs/equiva-
lent), s is the surface area of solid in the reactor and ␴app is the
surface charge density. At each titration point, the “extra” or
“missing” protons were normalized to the volume of solution
Fig. 2. Influence of pH on the ␨—potential of acid-reacted forsterite and surface area of solid in reactor. Magnesium and Si aqueous
(San Carlos Fo91). speciation were accounted for in these calculations and deter-
Forsterite surface composition in aqueous solutions 3303

Fig. 3. Total Mg and Si aqueous concentrations as a function of pH


during titration of fresh powder.

mined by means of the MINTEQA2 code (Allison et al., 1991).


For LRT experiments, the surface charge was calculated fol-
lowing the description given in Charlet et al. (1990) and Pok-
rovsky et al. (1999) after correcting for the dissolution and ion
exchange reactions. The surface charge (␴app) calculated this
way will be further referred to as “apparent surface charge” of
forsterite because it includes both the adsorption and penetra-
tion of H⫹ in the near-surface layers.
The results of forsterite surface titration allow characteriza-
tion of the ion-exchange reaction and H⫹/OH⫺ adsorption on
the surface. In both batch and LRT reactors, the concentrations
of Mg and Si during titration of fresh powder does not stay
constant: they increase with decreasing pH due to dissolution of
the bulk solid (Fig. 3). For most experiments performed in
batch reactors, a linear relationship is observed between Mg
Fig. 4. Total Mg and Si aqueous concentrations as a function of total
and Si concentration and H⫹ consumed during titration ([Hs⫹])
amount of H⫹ consumed for fresh powder titration. The slope [Mg]/
with a slope of 0.37 ⫾ 0.02 for Mg and 0.15 ⫾ 0.01 for Si (Fig. [Hs⫹] of 0.387 is consistent with reaction of 2.5 H⫹ with forsterite
4). Similar dependence between Mg, Si, and [Hs⫹] has been surface via exchange/dissolution and adsorption/penetration reactions.
reported by Luce et al. (1972). This fractional value can be
understood if one assumes that the protons consumed during
titration participate in three sequential reactions: i) dissolution ometry of reaction (1) when reactions (2) and (3) are neg-
of 1 mole of forsterite (reaction 1) with a Mg/H stoichiometry ligible.
of 2 and Si/H stoichiometry of 4, ii) Mg/H exchange reaction Similar plots for acid-reacted forsterite with high surface
having a Mg/H stoichiometry of 2 (reaction 2), and iii) adsorp- area revealed the same dependence between [Hs⫹] and Mg and
tion/penetration of 0.5 H⫹ on 1 mole of forsterite (n ⫽ 0.5 in Si concentration as for the fresh powders (Fig. 5), whereas the
reaction 3). The overall calculated stoichiometry for these pH-dependence for total Si and Mg concentration in the course
reactions, 1/(2 ⫹ 0.5) ⫽ 0.4, is in reasonable agreement with of titration in the LRT reactor was much weaker (Fig. 6).
the value 0.37 measured in the present study. Similarly, the Results of Fo91 surface titrations are presented in Figures 7
slope of the [SiO2(aq)]tot ⫺ [Hs⫹] dependence arising from and 8 where the measured apparent surface charge is plotted as
reactions (1–3) is equal to 1/(4 ⫹ 2 ⫹ 0.5) ⫽ 0.154 which is in a function of solution pH. It can be seen that the apparent pH
excellent agreement with the experimental value (0.15). In of zero charge point (pHPZC) for fresh forsterite is equal to
contrast, the titration of powders with very low surface area 10.0 ⫾ 0.2 whereas the corresponding value for the acid-re-
(0.45 m2/L in reactor, exp. No. 6 from Table 3) does not allow acted powder ranges from 6.5 to 8 depending on the intensity
seeing the effect of ion-exchange and surface adsorption of acid treatment (Fig. 8). Generally, the apparent surface
reactions on the total mass balance. As a result, the slopes charge of fresh forsterite is one order of magnitude higher than
[Mg2⫹]total/[Hs⫹] and [SiO2(aq)]tot /[Hs⫹] were measured to that of the acid-reacted mineral, achieving 10 C/m2 or 10⫺4
be 0.5 and 0.25, respectively, corresponding to the stoichi- mol/m2 at pH ⬃ 4. Such a high value, with regard to the
3304 O. S. Pokrovsky and J. Schott

Fig. 5. Total Mg and Si aqueous concentrations as a function of total


amount of H⫹ consumed for acid-reacted forsterite. Fig. 7. Surface charge of Fo91 corrected for dissolution and ion-
exchange reactions as measured in LRT reactor.

amount of total available surface sites on forsterite (i.e., 3.3 䡠


10⫺5 mol/m2 according to Wogelius and Walther, 1991), indi- charge although this effect is much weaker for forsterite than
cates that there is not only H⫹ adsorption on the surface but that reported for feldspar surfaces (Schott, 1990).
also H⫹ penetration into the near-surface region over at least The acid treatment of forsterite powder induced Mg deple-
one unit cell. Note that the apparent surface charge increases tion on the mineral surface as demonstrated by the high Mg/Si
with increasing ionic strength (Fig. 7). This indicates that the concentration ratio determined in the final acid solution. Sub-
penetration of of H⫹ occurs in some very hydrated, gel-like sequent dissolution of such acid-reacted powder released a
surface layer several units thick, where the solution counterion large excess of Si over Mg in solution. This is illustrated in
shielding might still be significant. Figure 10a where Mg and Si aqueous concentrations are plotted
The presence of iron in the forsterite structure does not as a function of surface area for acid-treated Fo91 in contact
appear to have an important effect on the charge development with solutions at pH 8.3. It can be seen that, compared to almost
as the value of ␴app for synthetic Fo100 was found to be close stoichiometric dissolution of the fresh mineral (Fig. 10b), the
to that of natural Fo91 (Fig. 9 and 7). Annealing of Fo100
powder at 500 and 900°C leads to a slight decrease of surface

Fig. 8. Experimental surface titration of the acid-reacted Fo91 at


Fig. 6. Total Mg and Si aqueous concentrations during titration of 25°C in 0.01 M NaCl using a LRT reactor. Surface charge is corrected
acid-reacted forsterite. for dissolution and ion-exchange reaction.
Forsterite surface composition in aqueous solutions 3305

Fig. 9. Experimental surface titration of synthetic Fo100 at 25°C in Fig. 11. pHimm for Fo91 as a function of added surface area in
0.01 M NaCl using a LRT reactor. solution. The equilibration time of the solid was 1 week.

dominance of Si over Mg in solution equilibrated with (s) (Sprycha, 1982). To check the effect of the value of s on
acid-reacted forsterite, demonstrates existence of a Mg-de- pHimm and pHPZC, a series of experiments with different
pleted, Si-enriched surface layer on the acid-reacted forster- amounts of solid was carried out. In Figure 11, the steady-state
ite. pHimm (after 1 week) is plotted as a function of the surface area
In analyzing our titration data, we followed the approach of of the immersed powder. It can be seen that forsterite pHimm
previous workers (Amrhein and Suarez, 1988; Wogelius and increases with surface area until the surface area reaches a
Walther, 1991; Stillings et al., 1995) and assumed that the value of ⬃20 m2/L, but when s increases further, the pHimm
apparent pHPZC is equal to the pHimm (pH immersion or pH of stays constant. As our titration experiments were performed
solution in contact with solid powder). It has been argued, with s ⬎ 25 m2/L (Table 3), the effect of various solid/
however, that the value of pHimm changes with the increase of solution ratios on pHimm, and consequently, pHPZC, was
solid/solution ratio or of total surface area of solid in solution negligible.

Fig. 10. Magnesium and Si concentrations in solution equilibrated for 24 hr with acid-reacted and fresh Fo91 at pH 8.3
and 9.0, respectively, as a function of solid surface area in the system. Preferential Si release from the solid indicates the
formation of a Si-rich layer on the acid-reacted forsterite.
3306 O. S. Pokrovsky and J. Schott

Table 4. Surface composition of forsterite samples with different histories.

Ols half-peak, O/Si Mg/Si


Sample eV atomic ratio atomic ratio

Initial powder 50–100 ␮m 2.4 4.0 1.80


Dissolution at pH ⫽ 1.8 2.6 3.3 1.47
Powder 2.8 m2/g titrated by HCl to pH 5.0 2.8 3.25 1.35
Exp. # 13, dissolution at pH ⫽ 2.8 2.7 3.2 1.66
Exp. # 8, dissolution at pH ⫽ 7.20 3.1 1.33
Exp. # 16, dissolution at pH ⫽ 8.8; 0.05 M NaHCO3 2.4 1.55
Exp. # 4, dissolution at pH ⫽ 10.9 2.3 4.6 2.08
Exp. # 19, dissolution at pH 10.8 2.00
Exp. # 20, dissolution at pH 11.0 1.91
Exp. # 17, dissolution at pH ⫽ 11.2; 0.03 M Na2CO3 1.80
Exp. # 3, dissolution at pH ⫽ 11.1; 0.02 M Na2CO3 2.4 4.6 1.89
Powder 2.8 m2/g in dist. H2O at pH 10 4.0 1.80

For convenience, the various surface compositions are normalized to the Mg/Si and O/Si ratio of unreacted
Fo91 taken as a standard value equal to 1.8 and 4.0, respectively (mineral stoichiometry).

3.3. Surface Spectroscopy evidence of H⫹ penetration in the near surface structure of


forsterite and/or the presence of bridging oxygens reflecting
The XPS results indicate that the near surface composition of
polymerization of isolated silica units. Such a polymerization
forsterite depends markedly on solution pH (Table 4). Mg/Si
and formation of amorphous SiO2 䡠 nH2O surface layer on the
atomic ratios at the surface of the reacted materials vary sys-
olivine grains reacted in 0.05 M sulfuric acid has been also
tematically, with Mg being depleted relative to Si in the acidic
observed by XPS spectroscopy (Seyama et al., 1996).
to weakly alkaline pH range (1.8 to 8.8) and slightly enriched
There is no broadening of the Si2p peak after reaction of
in alkaline solutions (pH ⬎ 10). Note that the results of our
forsterite in acidic solutions which is in contrast with the
XPS study are in full agreement with those of Fujimoto and
observations on acid-reacted tephroite (Casey et al., 1993a).
Velde (1990), Fujimoto et al. (1993), and Seyama et al. (1996)
The O/Si atomic surface ratio decreases for the samples reacted
who detected at low to medium pH a hydrogen-rich, Mg,
in acid solutions which provides further evidence of the poly-
Fe-leached surface layer on olivine using the same technique.
merization process (Hellmann et al., 1990). It should be noted
Our results are also in qualitative agreement with those of
that the altered layer at the forsterite-solution interface may be
Casey et al. (1993a) on the pH-dependence of tephroite
more than 1–2 unit cells thick as it can retract from large areas
(Mn2SiO4) surface composition as determined by XPS analy-
of the surface following its introduction in the high vacuum
sis. The O1s peak broadens during reaction at neutral to acidic
spectrometer chamber.
pH conditions as can be seen from O1s half-peak-height-width
The results of the DRIFT surface spectroscopy study of San
values listed in Table 4. A typical XPS O1s spectrum of two
Carlos Fo91 are presented in Figure 13. To our knowledge,
(starting material and acid-reacted forsterite) samples is shown
there is no forsterite reference reflectance spectra reported in
in Figure 12. As already suggested by Schott and Berner (1983)
for fayalite, this O1s half-peak-height-width increase provides the literature, thus only analogy with known compounds may
be used for interpreting the results. Freshly ground forsterite
exhibits a single sharp band at 3698 cm⫺1 which may be
associated with isolated ⬎MgOH or hydrogen-bonded surface
OH groups as is the case for MgO whose infrared spectra
contain both a sharp band at 3752 cm⫺1 and a broad band at
3610 cm⫺1 (Anderson et al., 1965). The intensity of the band at
3698 cm⫺1 in forsterite spectra decreases significantly after
acid treatment of the powder indicating that most of the Mg is
removed from the surface. Annealing of the powder at 500 and
900°C leads to the disappearance of this band similar to what is
observed for MgO. It should be also noted that the amount of
physically-adsorbed water responsible for the broad band cen-
tered at 3350 cm⫺1 increases significantly after acid treatment
of the sample. This may indicate H⫹ (H3O⫹) penetration into
the surface with formation of OH bonds within a thin micro-
porous gel surface layer as has been established by other
methods for silica (Tadros and Lyklema, 1968) and for
Fig. 12. Typical XPS O1s spectrum of initial and acid-reacted for- feldspars (Schott and Petit, 1987; Casey et al., 1989). Interest-
sterite. Broadening of O1s peak after reaction of forsterite in acidic
solutions provides evidence of H⫹ penetration in the near surface
ingly, the double broad band at 1420 –1464 cm⫺1 correspond-
structure of forsterite and/or the presence of bridging oxygens reflect- ing to ␯3 asymmetric stretching of the CO3 groups (White,
ing polymerization of isolated silica units. 1974), present in fresh forsterite, almost completely disappears
Forsterite surface composition in aqueous solutions 3307

Fig. 13. DRIFT spectra of San Carlos forsterite. Disappearance of the 3698 cm⫺1 band for acid-reacted Fo and increase
of the intensity of a broad band at around 3300 cm⫺1 is consistent with formation of a leached, Mg-exchanged,
H⫹-permeated layer on the surface of acid-reacted forsterite.

in the acid-reacted sample. This indicates the removal of car- precipitation of Mg(OH)2 could occur. Carbonate ions were
bonate (originating from fluid CO2 inclusions, Tingle et al., found to have similar effect on surface composition even at
1990) during acid treatment. lower pH values in solutions undersaturated with respect to
magnesite (⍀magnesite⫽ 0.07) (Fig. 14c). In contrast, acidifying
3.4. Column Filtration Experiments the reacting solution to pH 2–3 produces a spike in outlet Mg
concentration, indicating the dissolution of this Mg-rich layer
These experiments have been performed to precisely inves- and subsequent formation of a Si-rich surface layer; as the
tigate the stoichiometry of Mg and Si release from ultrasoni- volume of reacting fluid increases, stoichiometric Mg/Si ratio is
cally cleaned forsterite. The experimental setup does not allow achieved in the outlet solution. In alkaline solutions, stoichio-
quantifying the reaction rates so that only the time and pH metric Mg/Si release was never achieved within 1 day of
dependence of Mg/Si ratios measured in the outlet solution will experiment duration. These characteristic features were found
be discussed here. Results of several column filtration experi- to be reproducible for a wide variety of reacting solution
ments are presented in Figure 14 a– c where the outlet Mg/Si compositions, flow rates, and different powder size fractions.
molar ratio is plotted as a function of the solution volume that
passed through a column for different pHs. Fresh forsterite 4. DISCUSSION
powder, when first brought into contact with neutral or weakly
alkaline (pH ⫽ 9) aqueous solution, exhibits initial weak Mg The results obtained in this study indicate that Mg/Si ratios
preferential release and then stoichiometric dissolution. In- exhibited by forsterite surfaces in contact with solution are
creasing initial solution pH to 12.1 leads to a dramatic decrease always different from those of the bulk mineral. The surface of
of the Mg/Si ratio in the outlet solution (i.e., from 1.8 to 0.01, forsterite reacted in solutions at pH ⬍ 9 is depleted in magne-
Fig. 14a,b) indicating the formation of a Si-deficient, Mg-rich sium and enriched in silica. This silica-rich layer is formed by
surface layer. Note that all solutions were always undersatu- the exchange of two hydrogen ions for a magnesium atom at
rated with respect to brucite (⍀brucite ⫽ 0.13– 0.35) so that no the forsterite surface. The mass balance calculation deduced
3308 O. S. Pokrovsky and J. Schott

Fig. 14. Column-through filtration experiments. Mg to Si ratio in the outlet solution as a function of volume that passed
through the column, for different initial fluid compositions.

from the results of acid-reacted forsterite dissolution (Fig. 10) hydrogen penetration into the surface lattice of forsterite under
indicates that the thickness of this Mg-leached layer does not acidic conditions has also been detected using X-ray Absorp-
exceed 1–2 unit cells, i.e., 10 –20 Å. This explains the failure to tion Spectroscopy (XAS) and Elastic Recoil Detection Analysis
detect it using various techniques to measure hydrogen profiles (ERDA) observations (Wogelius and Fraser, 1996) as well as
with a depth resolution at the surface of about 100 Å (Westrich various surface analytical techniques based on energetic ion
et al., 1993; Casey et al., 1993b). In contrast, XPS and DRIFT beams (Petit et al. 1990a; b; Fujimoto et al., 1992; 1993).
techniques are sensitive enough to detect thin altered layer. Unlike for zeta-potential measurements, where only the first
This layer normally develops very fast as soon as forsterite is surface layer determines the electrokinetic properties and pHIEP
brought into contact with aqueous solutions having neutral or for the mineral/water interface, forsterite surface titration data
acidic pH. Its formation can explain the low value of forsterite indicate that several atomic layers are involved in the H⫹
pHIEP, 4.5 and 2.7 for fresh and acid-reacted powders, respec- uptake from the solution. Indeed, the calculated surface charge
tively, which is consistent with the dominance of silanol groups of fresh powder, corrected for dissolution and ion exchange
at the surface for pH ⬍ 9. reactions, reaches 10⫺4 mol/m2 (Fig. 7, 9), a value at least 3
Besides the Mg-H exchange reaction, penetration of protons times higher than the forsterite maximal positive surface site
in the form of H⫹ or H3O⫹ occurs into the first unit cells of density. Similar surface charge values were reported by Woge-
forsterite as confirmed by our XPS and DRIFT data. Such lius and Walther (1991). It can be estimated that at least 3
Forsterite surface composition in aqueous solutions 3309

atomic layers of fresh forsterite (one unit cell) are permeated by the formation of a Mn-rich surface layer. Oelkers and Schott
hydronium ions at pH ⬃ 4. Therefore, it can be concluded that (2000) observed a strong preferential Si over Mg release from
the exchange reaction allows protons and water molecules to the surface of enstatite (MgSiO3) reacted at pH 10 in solutions
penetrate deeper than the first surface layer thus yielding an undersaturated with respect to brucite. These observations to-
apparent surface charge value higher than the maximal value gether with the results of the present study indicate that the
estimated from crystallographic data and moving the pHimm. formation in alkaline solutions of such Si-depleted, cation-
(and, consequently, the pHPZC) to values much higher than that enriched layers is a general feature of pyroxenes and orthosili-
of the pHIEP. Proton uptake in the surface lattice is a general cates which should be taken into account when modeling their
feature of silicates already observed for feldspars and py- dissolution rates.
roxenes (see the recent review of Brantley and Chen, 1995;
Brantley and Stillings, 1996; Brantley and Stillings, 1997). The 5. CONCLUDING REMARKS
hydrogen-magnesium exchange reaction and the large proton The surface chemistry of forsterite was investigated using
uptake in the lattice prevent the interpretation of forsterite electrokinetic techniques, surface titrations, column experi-
surface titrations within the framework of the surface coordi- ments, and XPS and Infrared spectroscopy. All these methods
nation theory which does not account for the formation of indicate the formation at pH ⬍ 9 of a Mg-depleted, Si-rich
leached layers. altered layer via an exchange reaction between 2H⫹ and Mg2⫹
The value of ␴app for acid-reacted forsterite is an order of at the forsterite surface. The thickness of this leached layer does
magnitude lower than that for the fresh sample. Note that in not exceed 10 –20 Å. It leads to protonation and condensation
neutral and alkaline pH, acid-reacted forsterite behaves like of SiO4 tetrahedra on the surface. Besides this exchange and
amorphous silica with very low surface charge at neutral pHs polymerization reactions, penetration of H⫹ (H3O⫹) into the
and large surface charge development above pH 9 (Fig. 8). first unit cell occurs that leads to the development of apparent
Similar behavior has been reported for acid-reacted wollasto- high surface charge.
nite (Xie and Walther, 1994) which suggests that such titration In basic solutions of pH ⬎ 10, a unit cell thick, Si-depleted,
curves are a typical feature of acid-reacted silicates. Mg-rich layer forms on the surface due to Si preferential
For feldspars and wollastonite, it has been demonstrated that release into solution. As a result, it may be concluded that the
along with H⫹ penetration, condensation of silanol groups surface composition of forsterite reacting with aqueous solu-
occurs within the surface layer (Casey et al., 1988; Casey et al., tions is always different from its bulk composition. It is antic-
1993b; Hellmann et al., 1990). Results of our study are con- ipated that the knowledge of forsterite surface chemistry should
sistent with the polymerization of Si tetrahedra on the forsterite facilitate the rigorous description of the mechanisms control-
surface as suggested by the decrease of the O/Si surface atomic ling its dissolution kinetics and weathering in natural condi-
ratio and increase of the O1s peak width for the samples reacted tions.
at pH below 6 (Table 4). A similar conclusion was reached for
acid-reacted forsterite by Fujimoto and Velde (1990) and Acknowledgments—This manuscript benefited from insightful and very
Seyama et al (1996) from their XPS analyses. It is likely that constructive reviews by K. L. Nagy, D. Wesolowski, and two anony-
condensation reactions such as mous reviewers. We are grateful to M. Thibaut, G. Chatainier, M.
Molina, and F. Thomas for careful technical assistance with XRD,
XPS, B.E.T. analyses and electrokinetic measurements, respectively. E.
⬎ Si-OH ⫹ OH-Si ⬍ 3 ⬎ Si-O-Si ⬍ ⫹ H2O (12) Oelkers and R. Wogelius are thanked for the useful discussions in the
course of this study. This work was supported by a visiting research
are required to bring together the isolated Si tetrahedra in the position awarded to O. P. by CNRS.
forsterite structure and thus to build a silica-rich leached layer. Editorial handling: D. J. Wesolowski
In alkaline solutions, the development of a Si-depleted, Mg-
rich layer is detected on the surface of forsterite. This is REFERENCES
inferred from our XPS data showing an increase of Mg/Si and
O/Si ratios for the samples in contact with alkaline solutions Allison J. D., Brown D. S., and Novo-Gradac K. J. (1991)
MINTEAQA2/PRODEFA2, A geochemical assessment model for
(pH ⬎ 10, Table 4) and from preferential Si release in column environmental systems: version 3.0 user’s manual. U.S. EPA, Ath-
filtration experiments (Fig. 13, 14). Selective leaching of Si ens, GA.
from olivine surface at high pH was also reported by Fujimoto Amrhein C. and Suarez D. L. (1988) The use of a surface complexation
et al. (1993). Based on the results presented in Figures 13 and model to describe the kinetics of ligand-promoted dissolution of
anorthite. Geochim. Cosmochim. Acta 52, 2785–2793.
14, mass balance calculations were performed which indicate Anderson P. J., Horlock R. F., and Oliver J. F. (1965) Interaction of
that the thickness of the Mg-rich layer after 3 h of reaction does water with the magnesium oxide surface. Trans. Faraday Soc. 61,
not exceed half of unit cell. With prolongation of the reaction 2754 –2762.
time in basic solution, this layer grows further in thickness and Banfield J. F., Veblen D. R., and Jones B. F. (1990) Transmission
becomes detectable via XPS as can be seen from the data listed electron microscopy of subsolidus oxidation and weathering of for-
sterite. Contrib. Mineral. Petrol. 106, 110 –123.
in Table 4. The exact structure of this layer remains unclear but Banfield J. F., Jones B. F., and Veblen D. R. (1991) An AEM-TEM
it can be suggested that, as in the case of brucite or Mg-bearing study of weathering and diagenesis, Abert Lake, Oregon: I. Weath-
sheet silicates, it contains Mg octahedra linked together via ering reactions in the volcanics. Geochim. Cosmochim. Acta 55,
Mg-O-Mg bonds. 2781–2793.
Barnes I. and O’Neil J. R. (1969) The relationship between fluids in
It is worth noting that Casey et al. (1993a) reported an some fresh alpine-type ultramafics and possible modern serpentini-
increase of the Mn/Si ratio on the surface of tephroite reacted zation, Western United States. Geol. Soc. Amer. Bull. 80, 1947–
at pH 8 as determined by XPS analysis. This is consistent with 1960.
3310 O. S. Pokrovsky and J. Schott

Blum A. and Lasaga A. (1988) Role of surface speciation in the Oelkers E. H., Schott J., and Devidal J.-L. (1994) The effect of
low-temperature dissolution of minerals. Nature 331, 431– 433. aluminum, pH, and chemical affinity on the rates of aluminosilicate
Brady P. V. and Walther J. V. (1989) Controls on silicate dissolution dissolution reactions. Geochim. Cosmochim. Acta 58, 2011–2024.
rates in neutral and basic pH solutions at 25°C. Geochim. Cosmo- Parks G. A. (1967) Aqueous surface chemistry of oxides and complex
chim. Acta 53, 2823–2830. oxide minerals. In Equilibrium Concepts in Natural Water Systems.
Brady P. V. and Gı́slason S. R. (1997) Seafloor weathering controls on Amer. Chem. Soc. Adv. Chem. Ser. 67, 121–160.
atmospheric CO2 and global climate. Geochim. Cosmochim. Acta 61, Petit J. C., Della Mea G., Dran J. C., Magonthier M. C., Mando P. A.,
965–973. and Paccagnella A. (1990a) Hydrated-layer formation during disso-
Brantley S. L. and Chen Y. (1995) Chemical weathering rates of lution of complex silicate glasses and minerals. Geochim. Cosmo-
pyroxenes and amphiboles. In Chemical Weathering Rates of Silicate chim. Acta 54, 1941–1955.
Minerals (eds. A. F. White and S. L. Brantley). Rev. Mineral. 31, Petit J. C., Della Mea G., and Dran J. C. (1990b) Energetic ion beam
119 –17. analysis in the earth sciences. Nature 344, 621– 626.
Brantley S. L. and Stillings L. (1996) Feldspar dissolution at 25°C and Pokrovsky O. S., Schott J., and Thomas F. (1999) Processes at the
low pH. Am. J. Sci. 296, 101–127. magnesium-bearing carbonates/solution interface. I. A surface spe-
Brantley S. L. and Stillings L. (1997) Feldspar dissolution at 25°C and ciation model of magnesite. Geochim. Cosmochim. Acta 63, 863–
low pH—Reply. Am. J. Sci. 297, 1021–1032. 880.
Cases J. (1966) Sur la détermination du point de charge nulle des Rimsdidt J. D. and Dove P. M. (1986) Mineral/solution reaction rates
tectosilicates en milieu aqueux. C.R. Acad. Sci. Paris 262, serie C, in a mixed-flow reactor: Wollastonite hydrolysis. Geochim. Cosmo-
1456 –1458. chim. Acta 50, 2509 –2516.
Cases J. (1967) Les phénomènes physicochimiques à l’interface. Ap- Rosso J. J. and Rimstidt J. D. (2000) A high resolution study of
plication au procédé de la flottation. Unpublished Thesis, Université forsterite dissolution rates. Geochim. Cosmochim. Acta 64, 791– 811.
de Nancy. Schott J. (1990) Modeling of the dissolution of strained and unstrained
Casey W. H., Westrich H. R., and Arnold G. W. (1988) Surface multiple oxides: The surface speciation approach. In Aquatic Chem-
chemistry of labradorite feldspar reacted with aqueous solutions at ical Kinetics (ed. W. Stumm). Wiley, 337–365.
pH ⫽ 2, 3 and 12. Geochim. Cosmochim. Acta 52, 2795–2807. Schott J., Berner R. A., and Sjöberg E. L. (1981) Mechanism of
Casey W. H., Westrich H. R., Arnold G. W., and Banfield J. F. (1989) pyroxene and amphibole weathering—I. Experimental studies of
The surface chemistry of dissolving labradorite feldspar. Geochim. iron-free minerals. Geochim. Cosmochim. Acta 45, 2123–2135.
Cosmochim. Acta 53, 821– 832. Schott J. and Berner R. A. (1983) X-ray photoelectron studies of the
Casey W. H., Hochella M. F., Jr., and Westrich H. R. (1993a) The mechanism of iron silicate dissolution during weathering. Geochim.
surface chemistry of manganiferous silicate minerals as inferred Cosmochim. Acta 47, 2233–2240
from experiments on tephroite (Mn2SiO4). Geochim. Cosmochim. Schott J. and Berner R. A. (1985) Dissolution mechanisms of py-
Acta 57, 785–793. roxenes and olivines during weathering. In The Chemistry of Weath-
Casey W. H., Westrich H. R., Banfield J. F., Ferruzi G., and Arnold ering (ed. J. J. Drever). D. Riedel Publ. Co., 35–53.
G. W. (1993b) Leaching and reconstruction at the surfaces of dis- Schott J. and Petit J.-C. (1987) New evidence for the mechanisms of
solving chain-silicate minerals. Nature 366, 253–256. dissolution of silicate minerals. In Aquatic Surface Chemistry (ed.
Charlet L., Wersin P., and Stumm W. (1990) Surface charge of MnCO3 W. Stumm). Wiley, 293–318.
and FeCO3. Geochim. Cosmochim. Acta 54, 2329 –2336. Schott J. and Oelkers E. (1995) Dissolution and crystallization of
Deju R. A. and Bhappu R. B. (1966) A chemical interpretation of silicate minerals as a function of chemical affinity. Pure Appl. Chem.
surface phenomena in silicate minerals. Trans. A.I.M.E. 235, 329 – 67, 903–910.
332. Scofield J. H. (1976) Hartree-Slater subshell photoionization cross
Fujimoto K., Fukutani K., Tsunoda M., Yamashita H., and Kobayashi sections at 1254 and 1487 eV. J. Electron. Spectros. Related Phe-
K. (1992) Characterization of olivine-water interface using nom. 8, 129 –137.
1
H(15N,␣␥)12C resonant nuclear reaction. In Water-Rock Interaction Seyama H., Soma M., and Tanaka A. (1996) Surface characterization
(eds. Y. K. Kharaka and A. S. Maest), pp. 145–148 Balkema, of acid-leached olivines by X-ray photoelectron spectroscopy. Chem.
Rotterdam. Geol. 129, 209 –216.
Fujimoto K., Fukutani K., Tsunoda M., Yamashita H., and Kobayashi Sprycha R. (1982) Determination of electrical charge at Zn2SiO4/
K. (1993) Hydrogen depth profiling using 1H(15N,␣␥)12C resonant solution interface. Colloids Surfaces 5, 147–157.
nuclear reaction on water-treated olivine surfaces and characteriza- Stillings L. L., Brantley S. L., and Machesky M. L. (1995) Proton
tion of hydrogen species. Geochem. J. 27, 155–162. adsorption at an adularia feldspar surface. Geochim. Cosmochim.
Fujimoto K. and Velde B. (1990) Dissolution and hydration of olivine Acta 59, 1473–1482.
under hydrothermal conditions. EOS 71, 962. Stumm W. (1992) Chemistry of the Solid-Water Interface. Wiley.
Hellmann R., Eggleston C. M., Hochella M. F., Jr., and Crerar D. A. Tadros Th. F. and Lyklema J. (1968) Adsorption of potential-deter-
(1990) The formation of leached layers on albite surfaces during mining ions at the silica-aqueous electrolyte interface and the role of
dissolution under hydrothermal conditions. Geochim. Cosmochim. some cations. J. Electroanalyt. Chem. 17, 267–275.
Acta 54, 1267–1281. Tingle T. N., Hochella, M. F., Jr., Becker, C. H., and Malhotra R.
Hunter R. J. (1989) Foundation of Colloid Science, vol. 1, Clarendon (1990) Organic compounds on crack surfaces in forsterite from San
Press, Oxford. Carlos, Arizona, and Hualalai Volcano, Hawaii. Geochim. Cosmo-
Hochella M. F., Jr. and Carim A. H. (1988) A reassessment of electron chim. Acta 54, 477– 485.
escape depths in silicon and thermally grown silicon dioxide thin Westrich H. R., Cygan R. T., Casey W. H., Zemitis C., and Arnold
films. Surface Sci. Lett. 197, 260 –268. G. W. (1993) The dissolution kinetics of mixed-cation orthosilicate
Holland H. D. (1978) The Chemistry of the Atmosphere and Oceans. minerals. Amer. J. Sci. 293, 869 – 893.
Wiley. White, W. B. (1974) In The Infrared Spectra of Minerals (ed. V. C.
Ishido T. and Mizutani H. (1981) Experimental and theoretical basis of Farmer), Mineral. Soc. Amer.
electrokinetic phenomena in rock-water systems and its application Wogelius R. A. and Walther J. V. (1991) Olivine dissolution at 25°C:
to geophysics. J. Geophys. Res. 86, No. B3, 1763–1775. Effects of pH, CO2, and organic acids. Geochim. Cosmochim. Acta
Luce R. W., Bartlett R. W., and Parks G. A. (1972) Dissolution kinetics 55, 943–954.
of magnesium silicates. Geochim. Cosmochim. Acta 36, 35–50. Wogelius R. A. and Fraser D. G. (1996) Kinetics and mechanisms of
Oelkers E. H. (1999) A comparison of enstatite and forsterite dissolu- surface reactions at the olivine-aqueous fluid interface. Abstract of
tion rates and mechanisms. In Growth and Dissolution in Geosys- European Research Conf. Geochemistry of Crystal Fluids, Seefeld,
tems (eds. B. Jamveit and P. Meakin), Kluwer, 253–257. Austria.
Oelkers E. H. and Schott J. (2000) An experimental study of enstatite Xie Z. and Walther J. V. (1994) Dissolution stoichiometry and adsorp-
dissolution kinetics and the mechanism of pyroxene dissolution. tion of alkali and alkaline earth elements to the acid-reacted wollas-
Geochim. Cosmochim. Acta in press. tonite surface at 25°C. Geochim. Cosmochim. Acta 58, 2587–2598.
Forsterite surface composition in aqueous solutions 3311

Appendix A. Zeta-potential of olivine (␨) measured by streaming potential and electrophoresis techniques in
solutions of different compositions. Ionic strength was changed by adding NaCl, NaHCO3, HCl, NaOH, or
MgCl2.

Streaming potential technique

Solution
ionic strength ⌺Mg, M ⌺Si, M pH ␨, mV
⫺5 ⫺5
0.01 M HCl 3.9 ⫻ 10 4.1 ⫻ 10 2.09 28.1
0.01 M NaCl ⫹ HCl 6.0 ⫻ 10⫺5 5.0 ⫻ 10⫺5 2.20 28.0
0.01 M NaCl ⫹ HCl 6.3 ⫻ 10⫺5 5.4 ⫻ 10⫺5 3.07 23.8
0.01 M NaCl ⫹ HCl 9.3 ⫻ 10⫺5 5.1 ⫻ 10⫺5 3.29 26.1
0.01 M NaCl ⫹ HCl 4.3 ⫻ 10⫺5 4.0 ⫻ 10⫺5 3.46 14.6
0.01 M NaCl ⫹ HCl 2.7 ⫻ 10⫺5 3.1 ⫻ 10⫺5 4.25 5.7
0.01 M NaCl 1.5 ⫻ 10⫺5 2.2 ⫻ 10⫺5 4.80 ⫺5.0
0.01 M NaCl 1.2 ⫻ 10⫺5 2.1 ⫻ 10⫺5 5.40 ⫺17.2
0.01 M NaCl 1.1 ⫻ 10⫺5 1.5 ⫻ 10⫺5 6.71 ⫺30.0
0.01 M NaCl 6.7 ⫻ 10⫺6 2.0 ⫻ 10⫺5 6.72 ⫺36.5
0.01 M NaCl 2.9 ⫻ 10⫺5 1.9 ⫻ 10⫺5 7.29 ⫺47.2
0.01 M NaCl 3.5 ⫻ 10⫺6 1.8 ⫻ 10⫺5 7.62 ⫺46.2
0.01 M NaCl ⫹ NaOH 3.0 ⫻ 10⫺6 1.5 ⫻ 10⫺5 10.56 ⫺39.2
0.01 M NaCl ⫹ NaOH 3.7 ⫻ 10⫺6 2.1 ⫻ 10⫺5 10.94 ⫺39.7
0.01 M NaHCO3 5.6 ⫻ 10⫺6 2.0 ⫻ 10⫺5 8.41 ⫺39.1
0.01 M 0.001 5.1 ⫻ 10⫺5 6.94 ⫺24.6
0.006 M 0.002 2.0 ⫻ 10⫺5 6.79 ⫺20.4
0.009 M 0.003 2.0 ⫻ 10⫺5 8.86 ⫺16.9
0.015 M 0.005 2.0 ⫻ 10⫺5 7.13 ⫺15.5
0.001 M 3.0 ⫻ 10⫺4 2.5 ⫻ 10⫺5 6.61 ⫺22.0
0.001 M NaHCO3 6.1 ⫻ 10⫺6 1.9 ⫻ 10⫺5 8.88 ⫺44.4
0.001 M NaHCO3 4.5 ⫻ 10⫺6 1.9 ⫻ 10⫺5 7.99 ⫺56.3
0.001 M NaCl ⫹ HCl 3.5 ⫻ 10⫺5 3.3 ⫻ 10⫺5 3.54 22.7
0.001 M NaCl ⫹ HCl 3.2 ⫻ 10⫺5 3.5 ⫻ 10⫺5 3.60 21.0
0.001 M NaCl ⫹ HCl 4.2 ⫻ 10⫺5 2.8 ⫻ 10⫺5 3.63 17.0
0.001 M NaCl ⫹ HCl 2.6 ⫻ 10⫺5 3.2 ⫻ 10⫺5 3.80 9.8
0.001 M NaCl 2.3 ⫻ 10⫺5 3.0 ⫻ 10⫺5 4.45 ⫺0.4
0.001 M NaCl 1.2 ⫻ 10⫺5 2.3 ⫻ 10⫺5 5.10 ⫺8.2
0.001 M NaCl 6.4 ⫻ 10⫺6 2.1 ⫻ 10⫺5 6.32 ⫺39.0
0.001 M NaCl 1.1 ⫻ 10⫺5 2.2 ⫻ 10⫺5 6.43 ⫺42.6
0.001 M NaCl 6.0 ⫻ 10⫺6 2.0 ⫻ 10⫺5 6.45 ⫺41.5
0.001 M NaCl 1.7 ⫻ 10⫺5 2.7 ⫻ 10⫺5 6.65 ⫺47.0
0.001 M NaCl 4.8 ⫻ 10⫺6 1.9 ⫻ 10⫺5 7.69 ⫺53.0
0.001 M NaCl 4.8 ⫻ 10⫺6 2.0 ⫻ 10⫺5 7.68 ⫺48.8
0.001 M NaCl 4.2 ⫻ 10⫺6 2.0 ⫻ 10⫺5 8.06 ⫺60.4
0.001 M NaCl 4.6 ⫻ 10⫺6 1.8 ⫻ 10⫺5 8.13 ⫺49.4
0.001 M NaCl ⫹ NaOH 4.1 ⫻ 10⫺6 1.9 ⫻ 10⫺5 10.0 ⫺55.0
0.001 M NaCl ⫹ NaOH 3.6 ⫻ 10⫺6 2.0 ⫻ 10⫺5 10.6 ⫺50.0

Electrophoresis technique
Solution,
ionic strength ⌺Mg, M pH ␨, mV
0.01 M NaCl ⫹ HCl 3.9 ⫻ 10⫺5 3.53 1.50 ⫾ 0.5
0.001 M NaCl ⫹ HCl 3.2 ⫻ 10⫺5 3.80 12.5 ⫾ 1.5
0.01 M NaCl 1.4 ⫻ 10⫺5 4.70 ⫺4.5 ⫾ 0.5
0.01 M NaCl 1.0 ⫻ 10⫺5 5.46 ⫺17 ⫾ 1
0.01 M NaCl 3.1 ⫻ 10⫺6 9.06 ⫺30
0.01 M NaCl 2.8 ⫻ 10⫺6 9.08 ⫺39
0.01 M NaCl 3.0 ⫻ 10⫺6 9.32 ⫺40.5 ⫾ 1.5
0.01 M NaCl ⫹ NaOH 2.3 ⫻ 10⫺6 9.93 ⫺41.2
0.01 M NaCl ⫹ NaOH 2.8 ⫻ 10⫺6 11.2 ⫺38.7
0.01 M NaCl ⫹ NaOH 3.0 ⫻ 10⫺6 11.4 ⫺31
0.0015 M HCl 6.5 ⫻ 10⫺5 2.70 25 ⫾ 5
0.001 M HCl 5.8 ⫻ 10⫺5 3.02 20 ⫾ 5
0.001 M HCl ⫹ NaCl 5.1 ⫻ 10⫺5 3.24 20 ⫾ 4
0.001 M HCl ⫹ NaCl 4.0 ⫻ 10⫺5 3.42 17.0
0.001 M HCl ⫹ NaCl 3.4 ⫻ 10⫺5 3.71 11 ⫾ 1
0.001 M HCl ⫹ NaCl 3.2 ⫻ 10⫺5 3.81 6.8 ⫾ 2.3
0.001 M HCl ⫹ NaCl 2.4 ⫻ 10⫺5 4.27 ⫺1.7 ⫾ 0.5
0.001 M HCl ⫹ NaCl 2.2 ⫻ 10⫺5 4.30 3.0 ⫾ 2.0
0.001 M NaCl 4.1 ⫻ 10⫺6 6.69 ⫺28 ⫾ 2

(Continued)
3312 O. S. Pokrovsky and J. Schott

Appendix A. (Continued)

Electrophoresis technique

Solution,
ionic strength ⌺Mg, M pH ␨, mV
0.001 M NaCl 4.7 ⫻ 10⫺6 6.89 ⫺30.2 ⫾ 3.0
0.001 M NaCl 5.1 ⫻ 10⫺6 6.92 ⫺33.2 ⫾ 1.5
0.001 M NaCl 4.5 ⫻ 10⫺6 6.95 ⫺35 ⫾ 4
0.001 M NaCl 4.2 ⫻ 10⫺6 7.81 ⫺43.8 ⫾ 3.5
0.001 M NaCl 3.5 ⫻ 10⫺6 8.32 ⫺36 ⫾ 3
0.001 M NaCl 6.5 ⫻ 10⫺6 9.22 ⫺33.3
0.001 M NaCl 5.0 ⫻ 10⫺6 10.61 ⫺32.3

You might also like