Structural and electronic evolution of the As(OH)3 molecule in high temperature aqueous solutions: An x-ray absorption investigation

Article abstract of DOI:10.1063/1.1785150

The geometrical and electronic structure of the arsenious acid molecule As(OH)₃ in aqueous solutions has been investigated by x-ray absorption spectroscopy (XAS) within extended x-ray absorption spectroscopy (EXAFS) and x-ray absorption near edge structure (XANES), using realistic first-principle calculations in the latter case. This investigation was performed on aqueous solutions of arsenious acid from ambient to supercritical conditions (P = 250 and 600 bars, T ≤500°C) using a new optical cell. The analysis of the XAS spectra is consistent with (1) a constant As-O distance, (2) an opening of the O-As-O angles within the C_3V pyramidal structure in the range 30-200°C. This structural evolution comes along with a small decrease of the partial charges of the atoms in the As(OH)₃ molecule. The explanation invoked for both structural and electronic modifications observed is the weakening of the interactions, through hydrogen bonds, between the As(OH)₃ complex and water molecules. This is a fingerprint of the similar weakening of hydrogen bonding interactions in the solvent itself.

Full text of DOI:10.1063/1.1785150

Structural and electronic evolution of the As(OH ) 3 molecule in high temperature
aqueous solutions: An x-ray absorption investigation
Denis Testemale, Jean-Louis Hazemann, Gleb S. Pokrovski, Yves Joly, Jacques Roux, Roger Argoud, and
Olivier Geaymond
Citation: The Journal of Chemical Physics 121, 8973 (2004); doi: 10.1063/1.1785150
View online: http://dx.doi.org/10.1063/1.1785150
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/121/18?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Electronic structure and optical properties of CdSxSe1−x solid solution nanostructures from X-ray absorption
near edge structure, X-ray excited optical luminescence, and density functional theory investigations
J. Appl. Phys. 116, 193709 (2014); 10.1063/1.4902390
In situ near-edge x-ray absorption fine structure spectroscopy investigation of the thermal defunctionalization of
graphene oxide
J. Vac. Sci. Technol. B 30, 061206 (2012); 10.1116/1.4766325
Local atomic and electronic structures and ferroelectric properties of PbZr0.52Ti0.48O3: An x-ray absorption
study
Appl. Phys. Lett. 99, 042909 (2011); 10.1063/1.3607475
Geometric and electronic structures of NO dimer layers on Rh(111) studied with near edge x-ray absorption fine
structure spectroscopy: Experiment and theory
J. Chem. Phys. 127, 024701 (2007); 10.1063/1.2751156
Extended x-ray absorption fine structure study of Al x Ga (1−x) N films
Appl. Phys. Lett. 70, 2108 (1997); 10.1063/1.118963
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.104.242.103 On: Sat, 29 Nov 2014 15:22:54

JOURNAL OF CHEMICAL PHYSICS
VOLUME 121, NUMBER 18
8 NOVEMBER 2004
Structural and electronic evolution of the As„OH…3 molecule in high
temperature aqueous solutions: An x-ray absorption investigation
Denis Testemalea)
Laboratoire de Cristallographie CNRS-Grenoble, UPR5031, 25 avenue des Martyrs, BP 166 38042 Grenoble
Cedex 09, France and SNBL/ESRF, 6 rue Jules Horowitz, BP 220 38043 Grenoble
Jean-Louis Hazemann
Laboratoire de Cristallographie CNRS-Grenoble, UPR5031, 25 avenue des Martyrs, BP 166 38042 Grenoble
Cedex 09, France
Gleb S. Pokrovski
´
´
Laboratoire de Geochimie, UMR 5563, CNRS-OMP-Universite Paul Sabatier, 38 rue des 36 Ponts,
´
31400 Toulouse, France and Institut des Sciences de la Terre d’Orleans (ISTO), UMR6133,
´
´
1A rue de la Ferollerie, 45071 Orleans Cedex 02, France
Yves Joly
Laboratoire de Cristallographie CNRS-Grenoble, UPR5031, 25 avenue des Martyrs, BP 166 38042 Grenoble
Cedex 09, France
Jacques Roux
´
´
´
Institut des Sciences de la Terre d’Orleans (ISTO), UMR6133, 1A rue de la Ferollerie, 45071 Orleans
Cedex 02, France
Roger Argoud and Olivier Geaymond
Laboratoire de Cristallographie CNRS-Grenoble, UPR5031, 25 avenue des Martyrs, BP 166 38042 Grenoble
Cedex 09, France
͑Received 5 May 2004; accepted 29 June 2004͒
The geometrical and electronic structure of the arsenious acid molecule As͑OH)3 in aqueous
solutions has been investigated by x-ray absorption spectroscopy ͑XAS͒ within extended x-ray
absorption spectroscopy ͑EXAFS͒ and x-ray absorption near edge structure ͑XANES͒, using
realistic first-principle calculations in the latter case. This investigation was performed on aqueous
solutions of arsenious acid from ambient to supercritical conditions (Pϭ250 and 600 bars, T
р500 °C) using a new optical cell. The analysis of the XAS spectra is consistent with ͑1͒ a constant
As-O distance, ͑2͒ an opening of the O-As-O angles within the C3V pyramidal structure in the range
30–200 °C. This structural evolution comes along with a small decrease of the partial charges of the
atoms in the As͑OH)3 molecule. The explanation invoked for both structural and electronic
modifications observed is the weakening of the interactions, through hydrogen bonds, between the
As͑OH)3 complex and water molecules. This is a fingerprint of the similar weakening of hydrogen
bonding interactions in the solvent itself. © 2004 American Institute of Physics.
͓DOI: 10.1063/1.1785150͔
I. INTRODUCTION
Two situations must be considered: either the solute spe-
cies are ions or neutral molecules. In the former case, the
study of ion hydration-shell structures under high tempera-
ture and high pressure conditions gives information on the
molecular reorientation taking place in the solvent,2 and on
the evolution of solvent permittivity ⑀ that controls coulom-
bic interactions.3 The present study belongs to the latter case
where the solute under consideration is a neutral complex.
The molecule of aqueous arsenious acid As͑OH)3 is a very
good candidate as a probe: it possesses a nonzero dipole
moment that may evolve with the solvent permittivity ⑀ and
three OH groups likely to establish hydrogen bonds with
solvent water molecules.
The understanding of the processes of molecular interac-
tions in fluids at elevated pressures and temperatures is of
crucial importance, in particular, in the case of aqueous
solutions.1,2 The growing interest in these processes stems
from both the fundamental and applied insights that their
understanding brings to the geochemical, ͑bio͒chemical, and
industrial communities. It must be emphasized in the case of
aqueous solutions that the interparticle interactions under el-
evated temperature and pressure conditions are very sensitive
to the structure of the solvent and, in particular, to the hy-
drogen bonding strength. As a consequence, studying the
structural and electronic evolution of solute particles in aque-
ous solutions is an efficient indirect method to get informa-
tion on the solvent itself.
Among different techniques ͑x-ray diffraction, isotopic
substitution neutron scattering, Raman, IR, and NMR spec-
troscopies͒, x-ray absorption spectroscopy is particularly
powerful to obtain local information on the geometrical and
a͒
Electronic mail: denis.testemale@esrf.fr
0021-9606/2004/121(18)/8973/10/$22.00
8973
© 2004 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.104.242.103 On: Sat, 29 Nov 2014 15:22:54

8974
J. Chem. Phys., Vol. 121, No. 18, 8 November 2004
Testemale et al.
electronic structure of any compound. Thus many studies
using this tool have been dedicated to the structural charac-
terization of the hydration of cations and anions of metals
and nonmetallic elements at elevated pressure and tempera-
ture conditions.2–20 Although the extended x-ray Absorption
Fine Structure ͑EXAFS͒ analysis is widely used to get infor-
mation on the structure of molecular aqueous complexes ͑see
Ref. 21 and references therein͒, quantitative analysis of the
x-ray Absorption Near Edge Structure ͑XANES͒ is rather
new for aqueous systems. Recently, advances in this field
have been made with the appearance of different codes based
on the full multiple scattering scheme. For instance, Benfatto
et al.22 developed a code that has been used to model the
geometry of ion hydration shells.19,20 In the present study, we
used another approach with the code FDMNES ͑Ref. 23͒ that
calculates the absorption cross section beyond the muffin-tin
approximation ͑realistic calculations͒. We show that such an
approach is compulsory in the case of the nondense and
asymmetrical structure of the aqueous As͑OH)3 molecule.
Furthermore, an experimental cell has been developed that
enables the acquisition of both transmission and fluorescence
spectra of high quality for low concentration samples ͑ab-
sorber concentration р0.3m): these conditions allow to re-
duce the solute-solute interactions that may exist between
solute molecules.24,25
A first EXAFS analysis of the 0.3m solutions was re-
ported in Ref. 27. The cell used for these measurements has
been described in detail elsewhere.3,18,27–29 For the 0.05m
samples, a High Pressure ͑HP͒/High Temperature͑HT͒ setup,
designed recently in the Laboratoire de Cristallographie
͑CNRS-Grenoble͒, was used. It will be described in details in
a further publication.30 This cell design is compatible with
many different x-ray based techniques ͑XAS in both trans-
mission and fluorescence modes, small angle x-ray scatter-
ing, inelastic x-ray scattering, and visible Raman scattering,
and it operates in a wide range of temperature and pressure
(Tр600 °C and Pр600 bars for aqueous samples, and T
р1700 °C and Pр2000 bars for liquid metals and metallic
glasses͒. In the present work, several scans have been re-
corded for each sample, both in transmission and fluores-
cence modes at the same time, at each pressure and tempera-
ture. Reversibility has been checked at each T-P point: the
absorption jump at the As K edge, proportional to the sample
density, was found to be perfectly reversible on increasing
and decreasing temperature, as well as the shape of the
XANES and EXAFS spectra.
B. EXAFS analysis procedure
The EXAFS data processing was conducted with a set of
programs taken from the UWXAFS project.31 This software
allows to use the cumulants method32 and take into account
the multiple scattering contributions in the signal. The stan-
dard analysis procedure is followed. First, the EXAFS oscil-
lations ␹(k) are extracted by substraction of the baseline
with the AUTOBK program;33 the baseline is defined by the
low R components in the Fourier transform of the above
edge oscillating part of the spectra. Then, the analysis of
these oscillations is performed with FEFF 8.0 and FEFFIT.34,35
FEFF 8.0 calculates the scattering functions ͑amplitude and
phase͒ and the mean free paths of single and multiple scat-
tering paths within the structure of the cluster under consid-
eration. The geometrical parameters of these paths ͑number
N of oxygen neighbors, distance r0 between As and O atoms
and Debye–Waller parameter ␴2) are extracted by fitting the
experimental spectra with FEFFIT.
Section II describes briefly the experimental and analyti-
cal methods employed in the present study to collect and
model the experimental and calculated x-ray absorption
spectroscopy spectra. In Sec. III, we present the results of
both EXAFS and XANES analyses. Finally, Sec. IV ad-
dresses the discussion concerning the relation between the
evolution of the structure of As͑OH)3 and the properties of
water itself at elevated conditions of pressure and tempera-
ture.
II. EXPERIMENTAL AND ANALYTICAL METHODS
A. Cell design and spectra acquisition
XAS spectra at the As K edge ͑11 867 eV͒ were obtained
at the CRG-IF BM32 beamline at the European Synchrotron
Radiation Facility ͑ESRF, Grenoble, France͒. This beamline
is installed on a bending magnet ͑0.8 T͒. It is constituted by
a parabolic Ni-coated mirror, a Si ͑111͒ double-crystal mono-
chromator and an experimental setup for different XAS ac-
quisition modes ͑transmission and fluorescence͒. The spot
size ͑200ϫ300 ␮m2͒ is kept constant on the sample using the
dynamical sagittal focusing of the second crystal of the
monochromator.26 The spectra were recorded at the As K
edge in both transmission and fluorescence modes: the
former through two Si diode detectors located above and
below the sample on the x-ray beam and the latter with a 30
Ge elements Canberra detector.
The samples studied were 0.3m and 0.05m ͑moles per
kg of solvent͒ arsenious acid aqueous solutions. They were
prepared by dissolving As2O3 ͑Merck Normapur͒ in de-
ionized water ͑18 M⍀͒ at 90 °C. The spectra were recorded
at 250 and 600 bars and temperatures from 30 to 500 °C for
the 0.3m As solutions, and at 250 bars and temperatures from
30 to 350 °C for the 0.05m sample.
C. XANES simulations
The XANES part of the absorption spectrum extends up
to 50–60 eV beyond the threshold. It is known to be highly
sensitive to the three-dimensional structural environment
around the absorbing atom. Moreover, the photoelectron act-
ing as a localized probe of the electronic environment, it is
expected that XANES spectra will be sensitive to the evolu-
tion of the local charges with temperature. We used the
FDMNES package,23 which runs within the real space cluster
approach. The absorption cross section is calculated option-
ally under two procedures:
͑1͒ The first one uses the multiple scattering ͑MS͒
theory where the cross section is calculated in terms of path
operators solving the Lippman–Schwinger equation. It uses
the muffin-tin ͑MT͒ approximation for the shape of the po-
tential where this one is spherically averaged in the atoms
and constant between them.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.104.242.103 On: Sat, 29 Nov 2014 15:22:54

J. Chem. Phys., Vol. 121, No. 18, 8 November 2004
Structural and electronic evolution of aqueous As(OH)3
8975
TABLE I. Structural parameters of the first As coordination shell derived from EXAFS fits ͑reference As2O3
and As͑III͒ aqueous samples from ambient up to supercritical conditions͒. The Fourier parameters are k weight-
ing ͑3.8–11.5 Å͒ Hanning apodisation window, and ͑0.9–2.8 Å͒ real space fitting range. The reduction factor S02
is deduced from a fit of the arsenolite reference spectra, and fixed to 0.83 for aqueous samples. ͑f͒ means the
corresponding parameter is fixed during the fit. The goodness of the fits is considered via the R factor and the
uncertainties are calculated through an inversion of the curvature matrix ͑Ref. 35͒ ͑numbers between brackets͒.
␳
water (g cmϪ3
)
␴2(10Ϫ3 Å2)
1.9 ͑4.2͒
R factor
r0 (Å)
N0
⌬E0 (eV)
Arsenolite As2O3
1.787(f)
3(f)
10.7
0.004
As͑OH)3 , 0.3m, 250 bars
30 °C
1.00
0.97
0.88
0.63
0.51
0.13
1.77 ͑0.01͒
1.77 ͑0.01͒
1.77 ͑0.01͒
1.77 ͑0.01͒
1.76 ͑0.01͒
1.76 ͑0.01͒
3.0 ͑0.2͒
3.0 ͑0.2͒
3.0 ͑0.2͒
3.0 ͑0.2͒
3.0 ͑0.3͒
3.0 ͑0.3͒
1.2 ͑0.9͒
1.6 ͑1.0͒
2.0 ͑1.1͒
2.6(f)
2.8 ͑1.6͒
3.3 ͑1.6͒
12.3 ͑1.1͒
12.3 ͑1.1͒
13.0 ͑1.2͒
12.7 ͑1.4͒
11.0 ͑1.8͒
11.4 ͑1.7͒
0.007
0.007
0.008
0.012
0.016
0.014
100 °C
200 °C
350 °C
375 °C
425 °C
As͑OH)3 , 0.3m, 600 bars
30 °C
1.02
0.98
0.79
0.67
0.34
1.77 ͑0.01͒
1.77 ͑0.01͒
1.77 ͑0.01͒
1.77 ͑0.01͒
1.78 ͑0.01͒
3.0 ͑0.2͒
3.0 ͑0.2͒
2.9 ͑0.2͒
2.9 ͑0.2͒
2.9 ͑0.2͒
1.2 ͑0.9͒
1.3 ͑0.9͒
1.7 ͑1.1͒
1.8 ͑1.0͒
2.5(f)
12.6 ͑1.0͒
12.5 ͑1.0͒
12.5 ͑1.2͒
13.4 ͑1.2͒
13.9 ͑1.4͒
0.007
0.006
0.008
0.008
0.013
100 °C
300 °C
375 °C
500 °C
As͑OH)3 , 0.05m, 250 bars
30 °C
1.00
0.97
0.88
0.63
0.51
1.75 ͑0.01͒
1.75 ͑0.01͒
1.75 ͑0.01͒
1.76 ͑0.01͒
1.76 ͑0.01͒
3(f)
3(f)
3(f)
3(f)
3(f)
1.4 ͑0.5͒
1.9 ͑0.5͒
1.9 ͑0.6͒
2.0 ͑0.6͒
2.0 ͑0.6͒
12.4 ͑1.1͒
12.0 ͑1.3͒
11.6 ͑1.4͒
12.7 ͑1.5͒
12.4 ͑1.6͒
0.011
0.011
0.013
0.016
0.018
100 °C
200 °C
350 °C
375 °C
͑2͒ The second procedure uses the finite difference
As-O-H angle͒, which are expected to be of great importance
despite their weak scattering power. Note that the As-O dis-
tance was first carefully evaluated by the EXAFS analysis
͑following section͒. The electronic parameters are the partial
charges on the atoms ͑defined in terms of relative orbital
occupancy of the As, O, and H atoms͒. Several models of
partial charges were tested through the FDM calculations. It
should be pointed that the inspection of charge evolution, by
comparing experimental absorption spectra and realistic cal-
culations, is an efficient method which has been developing
for recent years in complicated chemical systems.37,38 These
local charges are found to be critical in the present case.
¨
method ͑FDM͒ to solve the Schrodinger equation. Its main
advantage is the possibility to have a totally free potential
shape, thus getting rid of the muffin-tin limitations.
Both procedures make further a Lorentzian energy-
dependent convolution of the spectrum to account for the
multielectronic and inelastic phenomena occurring in the ab-
sorption process. At the Fermi level, the Lorentzian width is
due to both the interaction with the core hole and the mono-
chromator resolution ͑respectively, 2.1 eV ͑Ref. 36͒ and
about 2 eV at the As K edge on the interface beamline ͑IF͒ at
ESRF͒. It was fixed at the total value of 3 eV full width at
half maximum ͑FWHM͒ for photoelectron energy below 10
eV. For higher photoelectron energy, between 10 and 40 eV,
the onset of plasmons collective interactions increase the
Lorentzian width up to 7 eV. It is worth noting here that the
same energy-dependent width was used in all the present
XANES calculations.
The multiple scattering approach which uses the muffin-
tin approximation is much less computer expensive than the
FDM method. It is well suited for symmetrical and dense
structures but is less satisfactory for highly asymmetrical or
less dense structure found in many molecular species. In this
case FDM method should be preferred as discussed thor-
oughly in Ref. 23. Nevertheless, MS calculations are per-
formed first to assess the influence of the geometrical param-
eters on the spectra ͑defining the geometry of the cluster͒ and
get a reasonable approximation. Then FDM calculations are
conducted in order to specify the value and the evolution
with temperature of the geometrical and electronic param-
eters. These geometrical parameters are the As-O distance,
the O-As-O angle, and the location of the H atoms ͑i.e.,
III. RESULTS
A. Results from EXAFS analysis
Several studies have shown that the ϩ3 oxidation state
of arsenic is stable in most high temperature hydrothermal
fluids characterized by reducing conditions.39–42 Available
solubility and spectroscopic data indicate that the neutral hy-
droxide complex As͑OH)3 is the dominant species in both
high density ͑␳Ͼ0.4 g cmϪ3͒ and low density ͑␳Ͻ0.4
g cmϪ3͒ hydrothermal fluids.27,41,42 Under the assumption
that individual and covalent As͑OH)3 species are largely
dominant in solution,27,42,43 a single shell model was used.
The EXAFS parameters derived in this study are summa-
rized in Table I. We checked the influence of including hy-
drogen atoms in the structure, but for the sake of brevity the
corresponding results are not presented because they neither
improved the fit nor altered it significantly. It stems from the
fact that the hydrogen atoms show up in the EXAFS signal
only when they are placed in focusing positions,18,44–46
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.104.242.103 On: Sat, 29 Nov 2014 15:22:54

8976
J. Chem. Phys., Vol. 121, No. 18, 8 November 2004
Testemale et al.
which is not the case for the As͑OH)3 molecule. The influ-
ence of multiple scattering on the fit was also tested and
found to be too weak to affect the structural parameters de-
rived from single-shell fitting. All the results are in fair
agreement with previous EXAFS works on identical ars-
enious acid solutions.27,43,47
2
The evolution of the value of ␴ consists of a slight
increase with temperature but its value remains low ͑from
1.2ϫ10Ϫ3 Å2 at 30 °C to 3.3ϫ10Ϫ3 Å2 at 425 °C 250
bars͒.48 Neither the As-O distance nor the number of oxygen
neighbors change with temperature and pressure within er-
rors. The number of oxygen neighbors remains around 3 in
the whole temperature range and the distance values are be-
tween 1.76 and 1.77 Å at 250 bars and between 1.77 and
1.78 Å at 600 bars. The 0.05m solutions are very close to
those 0.3m solutions, except for the value of the As-O dis-
tance. Indeed, the mean value seems to be smaller ͑about
1.75 Å͒ for the 0.05m solutions ͑to be compared to the mean
value of 1.77 Å for the 0.3m solutions͒.
B. Results from XANES analysis
1. Experimental spectra
The evolution of the XANES part of the experimental
spectra as a function of temperature is presented in Fig. 1. In
this figure, spectra are normalized to the density of water. It
is worth identifying and describing the specific features of
these spectra because the simulation part of the work will
consist of reproducing them. The first obvious observation is
the evolution of the white line with temperature: its intensity
increases regularly while its position does not change at both
pressures. The second point is the existence, at both pres-
sures, of a resonance at about 8 eV above the threshold ͑in-
dicated by the arrow on Fig. 1͒. The evolution of this feature
with temperature is pressure independent: its amplitude is
higher at 30 °C, rapidly decreases with increasing tempera-
ture, and becomes undetectable above 100 °C. Due to the low
intensity of this feature above 200 °C, it is difficult to con-
clude about its evolution at higher temperatures. In Fig. 1,
only three temperatures per pressure were drawn: 30 °C,
100 °C, and the higher temperature.
FIG. 1. Raw transmission XANES spectra of arsenious acid solutions
(0.3m) ͑a͒ at 250 bars ͑30 °C, ᭺; 100 °C, ᮀ; 425 °C, ᭞͒, ͑b͒ at 600 bars
͑30 °C, ᭺; 100 °C, ᮀ; 500 °C, ᭞͒. ͑c͒ raw fluorescence XANES spectra of
arsenious acid solutions (0.05m) at 250 bars ͑30 °C, ᭺; 100 °C, ᮀ; 375 °C,
᭝͒. The spectra are scaled to the solvent density values. The vertical arrow
denotes the resonance at 8 eV above the edge ͑see text͒. For reasons of
clarity, only three temperatures are plotted. In the inset, the resonance region
is plotted for the lowest and the highest temperature spectra.
2. XANES simulations
a. Muffin-tin Multiple Scattering ͑MT͒ versus Finite Dif-
ference Method ͑FDM͒. As it was already mentioned, the
relevant geometrical parameters of our realistic XANES cal-
culations are the O-As-O angle, the As-O distance, and the
As-O-H angle. They need to be constrained to reasonable
values by MS calculations ͑see Sec. II C͒. Determination of
the structure of the As͑OH)3 cluster was recently derived
from quantum mechanical ab initio calculations,49,50 in
agreement with Raman spectroscopy measurements.24,42
These ab initio calculations predicted that the molecule has a
pyramidal structure, with an apical arsenic atom and three
OH groups at the basis, with an As-O distance around 1.75
Å, O-As-O angles between 97° and 99°, an O-H distance of
0.97 Å, and As-O-H angles about 120°. These angles coupled
with the EXAFS mean As-O distance derived in the present
study were a starting point for the MS muffin-tin calcula-
tions. By comparing these preliminary basic simulations with
the experimental spectra in the EXAFS region, it appeared
that these geometrical values were not far from reality, since
they generate spectra similar to the experimental ones ͑am-
plitude and positions of the oscillations͒ in the ͑50–100 eV͒
range. For brevity these MS simulations are not detailed in
this study. It is worth pointing out here that these MS muffin-
tin calculations can only be considered as a first step. Indeed,
the main experimental feature ͑the resonance 8 eV above the
edge, see Fig. 1͒ can only be reproduced within FDM calcu-
lations, as it is demonstrated in Fig. 2. The inefficiency of the
Muffin-tin approximation is due to the strong asymmetry of
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.104.242.103 On: Sat, 29 Nov 2014 15:22:54

J. Chem. Phys., Vol. 121, No. 18, 8 November 2004
Structural and electronic evolution of aqueous As(OH)3
8977
TABLE II. Values of the partial atomic charges on the atoms in the
As͑OH)3 cluster. The Mulliken, AIM, and CHELPG sets of charges are
calculated with the GAUSSIAN 98 package ͑Ref. 51͒.
Charge set
Mulliken charge values
AtomsInMolecule charge values ͑AIM͒
CHELPG charge values
‘‘Electronegativity’’ charge values
͑refered to as ‘‘E charges’’͒
E charges ϫ1.5
As
O
H
ϩ1.5
Ϫ0.65
Ϫ1.23
Ϫ0.58
Ϫ0.34
ϩ0.15
ϩ0.62
ϩ0.42
ϩ0.27
ϩ1.83
ϩ0.48
ϩ0.21
ϩ0.315
ϩ0.42
Ϫ0.51
Ϫ0.68
ϩ0.405
ϩ0.54
E charges ϫ2
͑As-O͒ϭ1.77 Å, angle ͑O-As-O͒ϭ98°, and angle ͑As-O-H͒
ϭ120°. The different charges tested are displayed in Table II.
The first two sets of partial charges ͑Mulliken and AIM͒
were obtained by ab initio quantum calculations within a
Mulliken’s population analysis ͑GAUSSIAN 98 program51͒.
This type of charges was successfully used with FDMNES.23
An additional set of charges, fitted to the electrostatic poten-
tial and the dipole of the cluster, was calculated with the
CHELPG routine.71 Finally, a set of lower charges, refered to
as electronegativity charges in Table II, and labeled E else-
where, was calculated on Sanderson’s principle of balancing
electronegativities52 in the neutral cluster, using the elec-
tronegativity scale of Allred and Rochow.53
FIG. 2. ͑a͒ raw and ͑b͒ convoluted simulated spectra ͑distance As-Oϭ1.77
Å, O-As-Oϭ98°, As-O-Hϭ120° and E charges, see Table II͒, within finite
difference method FDM and Muffin-tin multiple scattering ͑MT MS͒ ap-
proaches.
As it can be seen in Fig. 3, a very good agreement with
the experimental spectrum is obtained with the E charges.
Using either the Mulliken or the AIM charges, of the same
order of magnitude, the convoluted spectra appeared to be
very different from the experimental ones. The reason is that
in the corresponding raw simulated spectra, i.e., before
Lorentzian convolution, the white lines were systematically
too low and too sharp to be discernible in the convoluted
spectra ͑the white lines FWHMs were inferior to 1 eV
whereas the width of the convoluting Lorentzian is 3 eV at
the Fermi level͒. This is apparent in Fig. 4 which shows the
raw simulated spectra, and the respective convoluted ones,
for three different sets of charges: Mulliken, AIM, and E
charges. One can see that the magnitude of the atomic
charges affects significantly the amplitude, width, and posi-
tion of the white line. However, the energy gap between the
white line and the specific resonance is 8 eV in the three
calculations, supporting the assumption that this feature is
essentially related to geometrical parameters. It is worth not-
ing, at this stage, that the ab initio quantum calculations with
GAUSSIAN 98 are made for an isolated As͑OH)3 molecule in
vacuum.
the As͑OH)3 molecule and the very low energy position of
the resonance considered ͑8 eV͒. Furthermore, still concern-
ing the low energy part of the spectra, we shall pay attention
to the good value of the convolution width, which was fixed
a priori based on experimental and intrinsic considerations
͑see Sec. II C͒. This is visible in Fig. 3: the experimental and
simulated spectra have the same white linewidth. Below, for
the FDM calculations, we examine the influence of each pa-
rameter on the shape of the XANES simulated spectra and
compare it with the experimental ones.
b. Atomic partial charges. The good order of magnitude
of the atomic partial charges must be estimated first, in order
to further vary the geometrical parameters. This estimation
was done using FDM calculations within the same structure
of As͑OH)3 as for the preliminary MS calculations: distance
That is the reason why the Mulliken’s charges ͑as well as
the AIM charges͒, although successfully implemented in the
FDMNES program,23 are not appropriate in the present case.
Indeed, the presence of the solvating environment around the
covalent As͑OH)3 neutral complex influences strongly the
magnitude of the partial charges, as compared to an isolated
molecule. It is confirmed by the magnitude of the CHELPG
charges, which are close to the E charges: the CHELPG
charges are fitted to the electrostatic dipole of the isolated
As͑OH)3 complex evaluated by the Hartree–Fock calcula-
tions; but this dipole value is always overestimated by this
FIG. 3. Experimental spectrum ͑0.3 m; 250 bar; 30 °C͒ and FDM convo-
luted simulated spectrum ͑distance As-Oϭ1.77 Å, O-As-Oϭ98°, As-O-H
ϭ120°, E charges͒. The experimental spectrum is scaled to fit both the
preedge and afteredge ͑Ϫ20͒–͑Ϫ5͒ eV and 40–60 eV regions, respectively͒.
Note that the white line and the 8 eV feature are well reproduced in the
calculated spectrum.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.104.242.103 On: Sat, 29 Nov 2014 15:22:54

8978
J. Chem. Phys., Vol. 121, No. 18, 8 November 2004
Testemale et al.
FIG. 6. Convoluted FDM simulated spectra ͑distance As-Oϭ1.77 Å, As-O-
Hϭ120° and E charges͒ for O-As-O angles varying from 92° to 104° with a
2° step. For clarity, the 92° dashed line and the 104° solid line are shown in
bold.
configurations conserve a symmetry C3V for the As͑OH)3
cluster ͑all the bonds of the cluster belong to the three planes
of symmetry͒. Keeping the O-As-O angle at 98° and the
As-O distance at 1.77 Å, the As-O-H angle was varied from
180° ͑As, O, and H atoms aligned͒ to 120°. Figure 6 shows
the resulting convoluted spectra. It appears that with increas-
ing the As-O-H angle the resonance at 8 eV shrinks and the
intensity of the white line increases. Both effects are similar
to the experimental spectra evolution observed with increas-
ing temperature ͑see Fig. 1͒. But the energy domain over
which the simulated spectra do not superpose is much larger
in Fig. 6 than in the experimental spectra in Fig. 1. In par-
ticular, between 20 and 30 eV above the white line, the simu-
lated spectra are not similar, in contrast to the experimental
spectra. Consequently, the evolution of the experimental
spectra with temperature cannot be explained by an evolu-
tion of the hydrogen position. For this reason, in further
simulations, it has been chosen to fix the As-O-H angle to
120°, the value resulting from the structure optimization us-
ing GAUSSIAN 98.51 Our choice is reinforced by the good
agreement between the experimental spectrum at ambient
conditions and the simulated spectrum for O-As-O angle of
98° and As-O-H angle of 120° ͑see Fig. 3͒, and between
experimental42 and calculated49 Raman frequencies for
As͑OH)3 aqueous species with the same structure as used in
the present study. However, it is worth noting the high sen-
sitivity of the FDM simulated spectra to the position of H
atoms, thus the small disagreement between experiment and
simulation observed about 20 eV above the white line in Fig.
3 is most likely influenced by the H position ͑for example,
twisting of the O-H bond out of the C3V symmetry, or
change of the O-H distance͒.
FIG. 4. ͑a͒ raw and ͑b͒ convoluted FDM simulated spectra ͑distance As-O
ϭ1.77 Å, O-As-Oϭ98° and As-O-Hϭ120°͒ for three different sets of
charges ͑see Table II͒.
method50 and can thus be considered closer to the real dipole
value of nonisolated As͑OH)3 complex in its aqueous polar-
izable medium.
Once the right order of charges magnitude was deter-
mined, its influence on the spectra was investigated more
precisely. In Fig. 5, three convoluted simulated spectra are
plotted ͑O-As-Oϭ98° and As-O-Hϭ120°͒, corresponding to
three different subsets of E charges: E charges, E charges
ϫ1.5, and E charges ϫ2. Although the intensity of the white
line in Fig. 5 diminishes with increasing the charges and its
position is shifted regularly towards higher energy ͑a 1.7 eV
shift between E charges and E charges ϫ2͒, it can be seen
that the ‘‘8 eV’’ resonance is not modified by the charges
magnitude ͑neither its amplitude, nor its position͒.
c. Hydrogen atoms positions. The hydrogen atoms can
be placed in several configurations, defined by the distance
O-H ͑kept fixed at 0.97 Å, see above͒ and the value of the
As-O-H angle. For simplicity, it was assumed that all the
d. O-As-O angle. The spectra simulated with the set of E
charges, for an As-O-H angle of 120° are plotted in Fig. 7 for
various values of the O-As-O angle. Their comparison with
the experimental spectra plotted in Fig. 1 shows that the
evolution of the XANES features observed in the experiment
with increasing temperature is similar to the changes in the
simulated spectra with increasing the O-As-O angle. First,
the white line intensity increases. Then, the amplitude of the
resonance at 8 eV above the edge is reduced, and in contrast
to what is observed when varying the hydrogen position, the
rest of the spectra superposes perfectly. However, the varia-
FIG. 5. Convoluted FDM simulated spectra ͑distance As-Oϭ1.77 Å, O-As-
Oϭ98°, and As-O-Hϭ120°͒ for three different set of charges: E charges, E
charges ϫ1.5 and E charges ϫ2.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.104.242.103 On: Sat, 29 Nov 2014 15:22:54

J. Chem. Phys., Vol. 121, No. 18, 8 November 2004
Structural and electronic evolution of aqueous As(OH)3
8979
ted in Fig. 8͑a͔͒. In Fig. 8͑b͒, the corresponding simulated
convoluted spectra are plotted. They correspond to two dif-
ferent distances ͑1.75 and 1.77 Å͒ for O-As-O angleϭ98°
and As-O-H angleϭ120°. It appears that both spectra are
similar ͑resonances shape and position͒, except for the inten-
sity of the white line which is lower for the shorter distance.
The comparison between experimental spectra and simula-
tions leads to the conclusion that the As-O distance is smaller
for the low concentration solutions. This XANES result re-
inforces the similar conclusion drawn from the EXAFS
analysis.
FIG. 7. Convoluted FDM simulated spectra ͑distance As-Oϭ1.77 Å, O-As-
Oϭ98°, and E charges͒ for three different hydrogen positions ͑keeping the
C3V symmetry͒: As-O-Hϭ180°, 140°, and 120°.
IV. DISCUSSION
A. Covalent As-O bond
The EXAFS results of this study show that the radial
structure of As environment in the neutral As͑OH)3 complex
does not change within errors in a wide temperature ͑30–
500 °C͒ and pressure ͑250–600 bar͒ range. This result is
similar to existing studies dedicated to the evolution of the
structure of strongly bonded metallic complexes of Mo and
W in near-critical or supercritical aqueous solutions: WO42Ϫ
up to 400 °C in Ref. 9 and MoO24Ϫ up to 150 °C in Ref. 55.
tion of the O-As-O angle has another effect, apparent in Fig.
7: the white line position shifts towards higher energies by
about ϩ0.5 eV when the O-As-O angles increases from 92°
to 104°. Such an energy shift, large enough to be experimen-
tally detected, is, however, not observed in the experimental
spectra.
e. As-O distance. Since the solute concentration seems to
affect the As-O distance ͑see Sec. III A͒, we checked the
effect of varying this parameter on the XANES simulated
spectra. Figure 8͑a͒ shows the experimental XANES spectra
of 0.3m and 0.05m arsenious acid solutions, at 30 and
375 °C. It is clearly visible that the amplitude of the white
line is smaller at lower concentration. This effect is observed
for all temperatures, and for both fluorescence and transmis-
sion spectra, ruling out any self-absorption effect54 ͓fluores-
cence and transmission spectra of 0.05m solutions have been
measured simultaneously and are identical, but for clarity
reasons only fluorescence scans at two temperatures are plot-
2
The weak evolution with temperature of the value of ␴
found for As͑OH)3 is also in agreement with those studies.
2
Indeed, for the 0.3m samples, the ␴ values vary from 1.2
ϫ10Ϫ3 Å2 at 30 °C/250 bars to 3.3ϫ10Ϫ3 Å2 at 425 °C/250
bars and are of the same order of magnitude as in those
2
works ͑for tungstate ions ␴ increases from 1.5ϫ10Ϫ3 Å2 at
20 °C/1 bars to 2.2ϫ10Ϫ3 Å2 at 400 °C/450 bars,9 and for
molybdate ions55
␴
values are around 5ϫ10Ϫ3 Å2). It is
2
worth comparing these values to those found for hydrated
ions in aqueous high temperature solutions3–5,7,8,10,14,18
where the typical values of the Debye–Waller ͑DW͒ factor
of the hydration shell are much higher. For monovalent ions,
the DW factor is higher than 20ϫ10Ϫ3 Å2 ͓see for example
BrϪ-H2O ͑Refs. 3, 10, and 11͒, Cuϩ-H2O ͑Ref. 8͒, and
Agϩ-H2O ͑Ref. 14͒ shell parameters͔. Due to a reinforce-
ment of the coulombic interactions with charge of cations,
the DW factor is smaller ͑but still remains much higher than
that for As͑OH)3) for solvated divalent or trivalent metallic
cations ͓around 8ϫ10Ϫ3 Å2 for Zn2ϩ-H2O ͑Ref. 3͒, for
Ni2ϩ-H2O ͑Ref. 10͒ and for La3ϩ-H2O ͑Ref. 4͔͒. Concern-
ing the number of neighbors, no reduction of this parameter
is observed at supercritical conditions for the As͑OH)3 mol-
ecule in water. This strongly suggests the absence of signifi-
cant disorder in the As-O bond within the As͑OH)3 mol-
ecule, in contrast with what is observed in the EXAFS
studies of ions hydration shells in such conditions.5 These
comparisons demonstrate the strong covalent character of the
As-O bonding in the As͑OH)3 molecule, as it was stressed in
Ref. 27. Besides, the present small variation of the DW fac-
tor with temperature can be interpreted in terms of molecular
forces within the bond. In their study of the temperature
dependence of the mean square displacement56 Beni et al.
2
report an analytical relation in which ␴ is proportional to
the temperature and to the inverse of the molecular force
field term. Considering this relation, and the weak depen-
dence of ␴ upon temperature for the As-O shell, one can
FIG. 8. ͑a͒ experimental XANES spectra for 0.3m and 0.05m arsenious acid
solutions at 30 and 375 °C. ͑b͒ FDM convoluted simulated XANES spectra
of the As͑OH)3 cluster with two As-O distances ͑1.75 and 1.77 Å͒.
2
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.104.242.103 On: Sat, 29 Nov 2014 15:22:54

8980
J. Chem. Phys., Vol. 121, No. 18, 8 November 2004
Testemale et al.
conclude that the constant As-O distance found here is con-
sistent with a strong covalent force in the As-O bond.
groups in the higher concentration solutions, that would lead
to an elongation of the As-O distance.24 But a systematic
determination of this distance with varying the concentration
should be made in order to confirm this effect. Recent
EXAFS experiments on the germanic acid at hydrothermal
conditions do not show the same tendency for the Ge-O dis-
tance in the Ge͑OH)4 complex. The Ge-O distance remains
the same within 0.004 Å in a wide range of Ge concentra-
tions (0.02m–0.09m).
B. Evolution of intermolecular interactions
Because the As͑OH)3 molecule is neutral, the coulombic
interactions with the dipoles of water molecules are limited.
Hence, the existence of an outer hydration shell strongly
bonded to the cluster is very improbable, thus explaining the
high stability of the As-O distance when the water structure
changes with increasing temperature. Nevertheless, as indi-
cated in the introduction, the existence of three OH groups
with hydrogen bonding affinity suggests the existence of hy-
drogen bonding association between the As͑OH)3 molecule
and the solvent water molecules not detectable by EXAFS.
Such an interaction is most likely to evolve in elevated con-
ditions of temperature and pressure, having for consequence
the modification of the internal structure of the molecule.
Based on comparisons between experimental and simu-
lated spectra, XANES analysis leads to the conclusion that
the O-As-O angle increases with increasing temperature.
This increase starts between 30 and 100 °C and seems to stop
above 200 °C. This is in agreement with previous Raman
results.24,42 In these studies, the authors measured the Raman
signal from As͑OH)3 solutions at different concentrations.
For the low concentration samples ͑0.02–0.5 mol kgϪ1͒, the
form of the spectra does not change in the temperature range
20–250 °C, except for a small shift ͑by a few cmϪ1͒ of the
two Raman bands (a1 and a3) towards lower wave numbers
͑cmϪ1͒. Under the assumption that the molecule symmetry
remains C3V when the temperature is increased, a shift of the
band towards low wave numbers can be explained by an
increase of the angles.57
The ‘‘opening’’ of the As͑OH)3 cluster with temperature,
below 200 °C, can be explained by a reduction of the inter-
actions through H bonds between the complex As͑OH)3 and
the water molecules. The cluster, which is constrained in the
tetrahedral arrangement of water molecules at low tempera-
ture, is released when the temperature increases and the hy-
drogen bonds weaken. The electrostatic repulsive interac-
tions between oxygen atoms of the As͑OH)3 molecule can
thus result in the enlargement of the pyramidal structure,
ending in a more symmetrical configuration ͑closer to a regu-
lar tetrahedron͒ consistent with the sp3 hybridization of the
outer electronic orbitals of the As atom. This conclusion
based on structural analysis is also strongly supported by the
evolution of the apparent hydration number of arsenious
acid, derived from As2O3 solubility measurements in
H2O-inert organic compound solvents.58 Thermodynamic
predictions based on these data indicate that As͑OH)3
•4H2O is the dominant species in a water-rich solvent below
100 °C, but this complex is progressively replaced by
As͑OH)3 with increasing temperature, showing the reduction
of the interactions between the cluster As͑OH)3 and water
molecules.
C. Evolution of atomic charges
The XANES modeling carried out in this work predicts
that the partial atomic charges in the As͑OH)3 complex may
change with increasing temperature. Indeed, as it was dem-
onstrated by the XANES analysis, the white line position in
the experimental spectra does not change with increasing ei-
ther T or P ͑see Fig. 1͒, which is in apparent disagreement
with the small shift of this position caused by the variation of
the O-As-O angle as predicted by XANES modeling ͑see
Fig. 7͒. The evolution of the white line with increasing tem-
perature can thus be interpreted by the competition between
a widening of the O-As-O angle and a reduction of the
atomic charges. Indeed, the former tends to increase the
white line intensity and to shift it towards higher energy, and
the latter also increases the white line intensity but shifts it
towards lower energy. A counterbalance of both effects on
the white line results in a global increase of its intensity
while its position does not change, which is the experimental
observation.
The weakening of the atomic partial charges is likely to
stem from the decrease of hydrogen bonding at elevated tem-
perature conditions. Indeed, the reduction of hydrogen bond-
ing interactions between solvent molecules themselves is re-
sponsible for the decrease of the water permittivity ͑see next
paragraph͒, resulting in the global reduction of the charges of
the solute atoms immersed in the solvent. Furthermore, the
joint reduction of hydrogen bonding interactions between
water molecules and the OH groups of As͑OH)3 , already
responsible for the increase of the O-As-O angle ͑see above͒,
may diminish the additional molecule polarizability, and
hence the atomic charges of As, O, and H atoms.
D. Temperature dependence
As proposed just above, the structural and electronic
evolution of As͑OH)3 is caused by the reduction of the hy-
drogen bonding interactions both between the OH groups of
the molecules and the water molecules ͑modification of the
local hydration͒ and between the water molecules themselves
͑resulting in the decrease of ⑀͒. It is thus interesting to ob-
serve if the temperature dependence of the modifications in
As͑OH)3 is the same as the temperature dependence of hy-
drogen bondings and ⑀ evolution.
The destruction of the tetrahedral arrangement of water
molecules with increasing temperature is responsible for the
decrease of the permittivity ⑀ which collapses from 80 at
ambient conditions down to 6 at the critical point, 374 °C
221 bars.59–62 Indeed, the high value of ⑀ at ambient condi-
tions, compared to simple liquids, is explained, within the
The reduction of the As-O distance in the low concen-
tration solutions, determined by EXAFS analysis and con-
firmed by XANES modeling, cannot be explained at present.
A possible interpretation is the existence of solute-solute in-
teractions by hydrogen bonding interactions between the OH
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.104.242.103 On: Sat, 29 Nov 2014 15:22:54

J. Chem. Phys., Vol. 121, No. 18, 8 November 2004
Structural and electronic evolution of aqueous As(OH)3
8981
Kirkwood theory,63 by the dipole-dipole interaction term re-
sulting from the hydrogen bonding association. Several stud-
ies have been dedicated to experimental estimation of water
permittivity60–62,64 and spectroscopic observation of hydro-
gen bonding in water ͑see Ref. 65 and references therein͒ at
high temperature and pressure conditions. For example, the
permittivity value ⑀͑0͒, which is around 80 at ambient con-
ditions, is about 40 at 200 °C, and about 20 at 300 °C, at
pressures in the 100–500 bars range.64 In the RMN study of
Hoffmann and Conradi,65 the authors measure the variation
of the mean number of hydrogen bonds in subcritical and
supercritical water. At 250 bars they found that, in compari-
son with room temperature, there are about 70% ͑60% and
45%͒ of hydrogen bonds at 100 °C ͑200 and 300 °C, respec-
tively͒. Their results are in very good agreement with previ-
ous NMR and Raman measurements of hydrogen bonding.
Those experimental studies also reveal a really weak pres-
sure dependence at these low temperature/high density con-
ditions. Many theoretical studies tend to support the experi-
mental observations.66–70 Jedlovsky and Richardi compare
different water models’ viability to reproduce water proper-
ties from ambient to supercritical conditions.66 Concerning
the permittivity value at HP/HT conditions, these calcula-
tions reveal, at 300 °C, a strong decrease of ⑀͑0͒ to 1/4 the
ambient value. In Ref. 69, a geometrical and topological
study of the hydrogen-bonding structure, based on RMC cal-
culations, indicates a notable modification of its tetrahedral
arrangement from 150 °C upwards. It is thus evident that the
hydrogen bonded network of water and the corresponding
value of the permittivity rapidly evolve with increasing tem-
perature mostly in the 30–200 °C range.
of the atomic partial charges magnitude. The changes in the
solvent structure, such as the weakening of the hydrogen
bonded network concomitant with the decrease of the solvent
dielectric constant, are invoked to explain the structural and
electronic modifications in the As͑OH)3 cluster with increas-
ing temperature: the interactions between the As hydroxide
complexes and water molecules originating from hydrogen
bonding weaken, thus releasing the molecule from the hydro-
gen bonded network. This release enables the molecule to
open its structure to a more tetrahedrallike configuration con-
sistent with the sp3 hybridization of the As electronic orbit-
als. At the same time, the decrease of the bulk water permit-
tivity justifies the reduction of the partial atomic charges. All
these effects take place at both concentrations (0.3m and
0.05m) as soon as the temperature is increased, the major
changes occuring between 30 and 200 °C. This is in agree-
ment with several theoretical and experimental studies on
water showing a strong reduction of its dielectric properties
in this temperature range. The combination of EXAFS analy-
sis with XANES realistic simulations is demonstrated to be a
powerful tool to get precise information on both geometrical
and electronic ͑partial charges͒ structures of aqueous com-
plexes and on the solvent itself, in high temperature/high
pressure fluids. This HP/HT study would not have been pos-
sible without the development of an efficient experimental
setup that allows to obtain high-quality transmission and
fluorescence spectra.30
ACKNOWLEDGMENTS
The authors thank J.-C. Soetens for his strong help with
the various GAUSSIAN 98 calculations, and J.-M. Beny for the
Raman calculations of the normal modes frequencies.
In our study, the increase of the O-As-O angle and the
decrease of the atomic charges occur below 200 °C, and do
not depend on pressure. That is in accordance with the rapid
evolution of the solvent properties mentioned just above. For
temperatures above 200 °C, changes in the XANES spectra
become undetectable. This might imply that either the struc-
tural and electronic evolution of As͑OH)3 stops, or a further
evolution of both angles and charges go on but is undetect-
able within the resolution of the methods. No unequivocal
conclusion can be drawn at this point.
1 A. G. Kalinichev, Reviews in Mineralogy and Geochemistry ͑Mineralogi-
cal Society of America, Washington, D.C., 2001͒, Vol. 42, p. 83.
2 A. Filipponi, S. DePanfilis, C. Oliva, M. A. Ricci, P. D’Angelo, and D. T.
Bowron, Phys. Rev. Lett. 91, 165505 ͑2003͒.
3 V. Simonet, Y. Calzavara, J.-L. Hazemann, R. Argoud, O. Geaymond, and
D. Raoux, J. Chem. Phys. 117, 2771 ͑2002͒.
4 A. J. Anderson, S. Jayanetti, R. A. Mayanovic, W. A. Bassett, and I.-M.
Chou, Am. Mineral. 87, 262 ͑2002͒.
5 G. Ferlat, A. S. Miguel, J. F. Jal, J. C. Soetens, P. A. Bopp, I. Daniel, S.
Guillot, J. L. Hazemann, and R. Argoud, Phys. Rev. B 63, 134202 ͑2001͒.
6 J. L. Fulton, D. M. Pfund, S. L. Wallen, M. Newville, E. A. Stern, and Y.
Ma, J. Chem. Phys. 105, 2161 ͑1996͒.
V. CONCLUSION
7 J. L. Fulton, M. M. Hoffmann, and J. G. Darab, Chem. Phys. Lett. 330,
300 ͑2000͒.
The structure of the As͑OH)3 covalent molecule in aque-
ous solutions was determined by X-Ray absorption spectros-
copy from ambient to supercritical temperatures. Both
EXAFS analysis and XANES simulations were used to ob-
tain information on the evolution of the geometry of As en-
vironment ͑distances, angles͒ and its electronic configuration
͑atomic partial charges͒ with temperature and pressure. The
EXAFS analysis shows the constancy of the As-O distance,
which remains about 1.77 Å in 0.3m solutions in the whole
temperature range 30–500 °C at 250 and 600 bars, and about
1.75 Å in 0.05m solutions in the temperature range 30–
375 °C, at 250 bars. These new data confirm that the As-O
bonding has a strong covalent character, which keeps up un-
der the supercritical temperature conditions. New XANES
simulations reveal an enlargement of the O-As-O angles
when temperature increases, coupled with a small reduction
8 J. L. Fulton, M. M. Hoffmann, J. G. Darab, B. J. Palmer, and E. A. Stern,
J. Phys. Chem. A 104, 11651 ͑2000͒.
9 M. M. Hoffmann, J. G. Darab, S. M. Heald, C. L. Yonker, and J. L. Fulton,
Chem. Geol. 167, 89 ͑2000͒.
10 M. M. Hoffmann, J. G. Darab, B. J. Palmer, and J. L. Fulton, J. Phys.
Chem. 103, 8471 ͑1999͒.
11 R. A. Mayanovic, A. J. Anderson, W. A. Bassett, and I.-M. Chou, Chem.
Phys. Lett. 336, 212 ͑2001͒.
12 R. A. Mayanovic, A. J. Anderson, W. A. Bassett, and I.-M. Chou, J. Syn-
chrotron Radiat. 6, 195 ͑1999͒.
13 D. M. Pfund, J. G. Darab, J. L. Fulton, and Y. Ma, J. Phys. Chem. 98,
13102 ͑1994͒.
14 T. M. Seward, C. M. B. Henderson, J. M. Charnock, and B. R. Dobson,
Geochim. Cosmochim. Acta 60, 2273 ͑1996͒.
15 T. M. Seward, C. M. B. Henderson, J. M. Charnock, and T. Driesner,
Geochim. Cosmochim. Acta 63, 2409 ͑1999͒.
16 S. L. Wallen, B. J. Palmer, D. M. Pfund, and J. L. Fulton, J. Phys. Chem.
101, 9632 ͑1997͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.104.242.103 On: Sat, 29 Nov 2014 15:22:54

8982
J. Chem. Phys., Vol. 121, No. 18, 8 November 2004
Testemale et al.
17 S. L. Wallen, B. J. Palmer, and J. L. Fulton, J. Chem. Phys. 108, 4039
͑1998͒.
45 B. Lengeler, Phys. Rev. Lett. 53, 74 ͑1984͒.
46 K. R. Wilson, J. G. Tobin, A. L. Ankudinov, J. J. Rehr, and R. J. Saykally,
Phys. Rev. Lett. 85, 4289 ͑2000͒.
18 V. Simonet, Y. Calzavara, J.-L. Hazemann, R. Argoud, O. Geaymond, and
D. Raoux, J. Chem. Phys. 116, 2997 ͑2002͒.
47 I. J. Pickering, R. C. Prince, M. J. George, R. D. Smith, G. N. George, and
D. E. Salt, Plant Physiol. 122, 1171 ͑2000͒.
19 M. Benfatto, P. D’Angelo, S. DellaLonga, and N. V. Pavel, Phys. Rev. B
65, 174205 ͑2002͒.
48 Debye-Waller parameters have been evaluated by Recursion-Method cal-
culations performed within FEFF 8 program for an AsO3 cluster: these
calculations yield a 0.0017 ͑resp. 0.0022͒ value at 30 °C ͑resp. 300 °C͒.
The small disagreement with our experimental determination ͑0.0012 and
0.0017 at respective temperatures͒ is due to the small k range that we
used.
20 P. D’Angelo, M. Benfatto, S. DellaLonga, and N. V. Pavel, Phys. Rev. B
66, 064209 ͑2002͒.
21 A. Filipponi, J. Phys.: Condens. Matter 13, 23 ͑2001͒.
22 M. Benfatto and S. DellaLonga, J. Synchrotron Radiat. 8, 1087 ͑2001͒.
23 Y. Joly, Phys. Rev. B 63, 125120 ͑2001͒.
24 R. Gout, G. S. Pokrovski, J. Schott, and A. Zwick, J. Raman Spectrosc.
28, 725 ͑1997͒.
49 J. A. Tossell, Geochim. Cosmochim. Acta 61, 1613 ͑1997͒.
50 J.-C. Soetens ͑private communication͒.
25 Note that for such low concentrations, the acquisition of diffraction pat-
terns S(q) is impossible.
51 M. J. Frisch, G. W. Trucks et al., GAUSSIAN Inc. ͑2001͒.
52 R. T. Sanderson, Science 114, 670 ͑1951͒.
26 J.-L. Hazemann, K. Nayouf, and F. D. Bergevin, Nucl. Instrum. Methods
Phys. Res. B 97, 547 ͑1995͒.
53 J.-P. Jolivet, De La Solution a L’axyde, Savoirs Actuels ͑CNRS, Paris,
1994͒.
27 G. S. Pokrovski, I. V. Zakirov, J. Roux, D. Testemale, J.-L. Hazemann, A.
Bychkov, and G. Golikova, Geochim. Cosmochim. Acta 66, 3453 ͑2002͒.
28 Y. Soldo, J.-L. Hazemann, D. Aberdam, M. Inui, K. Tamura, D. Raoux, E.
Pernot, J.-F. Jal, and J. D. Philon, Phys. Rev. B 57, 258 ͑1998͒.
29 K. Tamura, M. Inui, and S. Hosokawa, Rev. Sci. Instrum. 66, 1382 ͑1995͒.
30 D. Testemale, O. Geaymond, R. Argoud, and J.-L. Hazemann ͑unpub-
lished͒.
54 This has been however checked by quantitative determination of the self-
absorption effect by use of the FLUO software ͑FLUO:correcting XANES for
self-absorption in fluorescence measurements͒ by Daniel Haskel, Univer-
sity of Washington. For 0.05m solutions of heavy arsenic atom in light
water molecules matrix, the effect is found negligible.
55 J. F. W. Mosselmans, P. F. Schofield, J. M. Charnock, C. D. Garner, R. A.
D. Pattrick, and D. J. Vaughan, Chem. Geol. 127, 339 ͑1996͒.
56 G. Beni and P. M. Platzman, Phys. Rev. B 14, 1514 ͑1976͒.
57 J.-M. Beny ͑ISTO-CNRS͒ made calculations of the normal Raman modes
of the molecule AsF3 , analogous to As͑OH)3 and showed that an opening
of the molecule results in a shift of the a1 band towards lower wavenum-
bers.
31 E. A. Stern, M. Newville, B. Ravel, Y. Yacoby, and D. Haskel, Physica B
208–209, 117 ͑1995͒.
32 G. Bunker, Nucl. Instrum. Methods Phys. Res. 207, 437 ͑1983͒.
33 M. Newville, P. Livins, Y. Yacoby, J. J. Rehr, and E. A. Stern, Phys. Rev.
B 47, 14126 ͑1993͒.
34 A. L. Ankudinov, B. Ravel, J. J. Rehr, and S. D. Conradson, Phys. Rev. B
58, 7565 ͑1998͒.
58 G. S. Pokrovski, J.-M. Beny, and A. V. Zotov, J. Solution Chem. 28, 1307
͑1999͒.
35 M. Newville, B. Ravel, D. Haskel, J. J. Rehr, E. A. Stern, and Y. Yacoby,
Physica B 208–209, 154 ͑1995͒.
59 E. U. Franck, The Physics and Chemistry of Aqueous Ionic Solutions
͑Reidel, Dordrecht, 1987͒.
36 M. O. Krause and J. H. Oliver, J. Phys. Chem. Ref. Data 8, 329 ͑1979͒.
37 Y. Joly, D. Cabaret, H. Renevier, and C. R. Natoli, Phys. Rev. Lett. 82,
2398 ͑1999͒.
60 D. G. Archer and P. Wang, J. Phys. Chem. Ref. Data 19, 371 ͑1990͒.
61 D. P. Fernandez, Y. Y. Mulev, A. R. Goodwin, and J. M. H. Sengers, J.
Phys. Chem. Ref. Data 24, 33 ͑1995͒.
38 Y. Joly, S. Grenier, and J. E. Lorenzo, Phys. Rev. B 68, 104412 ͑2003͒.
39 N. N. Akinfiev, A. V. Zotov, and A. P. Nikonorov, Geochem. Intl. 29, 109
͑1992͒.
62 M. Uematsu and E. U. Franck, J. Phys. Chem. Ref. Data 9, 1291 ͑1980͒.
63
¨
H. Frohlich, Theory of Dielectrics ͑Oxford University Press, London,
40 G. R. Helz, J. A. Tossel, J. M. Charnock, R. A. D. Pattrick, D. J. Vaughan,
and C. D. Garner, Geochim. Cosmochim. Acta 59, 4591 ͑1995͒.
41 G. S. Pokrovski, S. Kara, and J. Roux, Geochim. Cosmochim. Acta 66,
2361 ͑2002͒.
1958͒.
64 K. Okada, Y. Imashuku, and M. Yao, J. Chem. Phys. 107, 9302 ͑1997͒.
65 M. M. Hoffmann and M. S. Conradi, J. Am. Chem. Soc. 119, 3811 ͑1997͒.
66 P. Jedlovszky and J. Richardi, J. Chem. Phys. 110, 8019 ͑1999͒.
67 B. D. Bursulaya and H. J. Kim, J. Chem. Phys. 110, 9646 ͑1999͒.
68 Y. Guissani and B. Guillot, J. Chem. Phys. 98, 8221 ͑1993͒.
69 P. Jedlovszky, J. P. Brodholt, F. Bruni, M. A. Ricci, A. K. Soper, and R.
Vallauri, J. Chem. Phys. 108, 8528 ͑1998͒.
42 G. S. Pokrovski, R. Gout, A. V. Zotov, J. Schott, and J.-C. Harrichoury,
Geochim. Cosmochim. Acta 60, 737 ͑1996͒.
43 Y. Arai, E. J. Elzinga, and D. L. Sparks, J. Colloid Interface Sci. 235, 80
͑2001͒.
44 P. D’Angelo, A. DiNola, M. Mangoni, and N. V. Pavel, J. Chem. Phys.
104, 1779 ͑1996͒.
70 A. K. Soper, F. Bruni, and M. A. Ricci, J. Chem. Phys. 106, 247 ͑1996͒.
71 C. M. Breneman and K. B. Wiberg, J. Comput. Chem. 11, 361 ͑1990͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
131.104.242.103 On: Sat, 29 Nov 2014 15:22:54