X-ray absorption spectroscopy of aqueous aluminum-organic complexes

Article abstract of DOI:10.1021/jp909656q

Aqueous-phase X-ray absorption near-edge structure (XANES) spectra were collected on dissolved Al complexes with organic ligands, including desferrioxamine B, EDTA, acetohydroxamate, malate, oxalate, and salicylate. Spectral interpretations were made using the density functional theory-based modeling package StoBe. The goals of this work were to study the geometric and electronic structural characteristics of these complexes relative to Al(H ₂O)₆³⁺ and to examine the utility of the aqueous Al XANES technique as a tool for probing Al speciation and structure. In the case of EDTA, aqueous Fourier-transform infrared spectroscopy was also used to corroborate the structures of the Al(EDTA)^- and AlOH(EDTA) ^2- complexes. Synthetic XANES spectra calculated with StoBe reproduced the observed spectral differences between Al(H₂O) ₆³⁺, Al(dfoB)⁺, and Al(EDTA)^-. The narrower XANES feature observed for Al(dfoB)⁺ relative to Al(H ₂O)₆³⁺ can be attributed to a weaker splitting of the Al 3p - O 2p interactions in the former, while Al(EDTA)^- exhibits split Al 3p - ligand interactions that likely result from the mixed O/N coordination. In complexes with mixed aqua/organic-oxygen ligation (Al-acetohydroxamate, Al-malate, Al-oxalate, and Al-salicylate), spectra exhibit linear, systematic changes in peak width as a function of H₂O to organic ligand ratio in the Al coordination sphere. These results highlight the sensitivity of the aqueous Al K-edge XANES spectrum to coordination environment and demonstrate its utility as an experimental probe for future studies of Al speciation in complex solutions.

Full text of DOI:10.1021/jp909656q

6138
J. Phys. Chem. A 2010, 114, 6138–6148
X-ray Absorption Spectroscopy of Aqueous Aluminum-Organic Complexes
Michael B. Hay†,* and Satish C. B. Myneni‡,§
Department of CiVil and EnVironmental Engineering, Princeton UniVersity, Princeton, New Jersey 08544,
Department of Geosciences, 151 Guyot Hall, Princeton UniVersity, Princeton, New Jersey 08544, and Earth
Sciences DiVision, Lawrence Berkeley National Laboratory, Berkeley, California 94720
ReceiVed: October 8, 2009; ReVised Manuscript ReceiVed: February 18, 2010
Aqueous-phase X-ray absorption near-edge structure (XANES) spectra were collected on dissolved Al
complexes with organic ligands, including desferrioxamine B, EDTA, acetohydroxamate, malate, oxalate,
and salicylate. Spectral interpretations were made using the density functional theory-based modeling package
StoBe. The goals of this work were to study the geometric and electronic structural characteristics of these
3+
complexes relative to Al(H2O)6 and to examine the utility of the aqueous Al XANES technique as a tool
for probing Al speciation and structure. In the case of EDTA, aqueous Fourier-transform infrared spectroscopy
was also used to corroborate the structures of the Al(EDTA)- and AlOH(EDTA)2- complexes. Synthetic
3+
XANES spectra calculated with StoBe reproduced the observed spectral differences between Al(H2O)6
,
3+
Al(dfoB)+, and Al(EDTA)-. The narrower XANES feature observed for Al(dfoB)+ relative to Al(H2O)6
can be attributed to a weaker splitting of the Al 3p - O 2p interactions in the former, while Al(EDTA)-
exhibits split Al 3p - ligand interactions that likely result from the mixed O/N coordination. In complexes
with mixed aqua/organic-oxygen ligation (Al-acetohydroxamate, Al-malate, Al-oxalate, and Al-salicylate),
spectra exhibit linear, systematic changes in peak width as a function of H2O to organic ligand ratio in the
Al coordination sphere. These results highlight the sensitivity of the aqueous Al K-edge XANES spectrum to
coordination environment and demonstrate its utility as an experimental probe for future studies of Al speciation
in complex solutions.
Introduction
are used to study Al complexes from the perspective of the
ligand, particularly when a ligand of interest is highly sensitive
The aqueous chemistry of aluminum is of critical interest in
the geologic and environmental sciences. In the aqueous phase,
Al may be present in a variety of species depending on solution
composition, confounding its behavior in the environment. The
free Al3+ ion strongly binds to water in solution, forming a series
of monomeric and polymeric hydrolysis products depending on
pH,1 but equally important in natural aquatic environments are
the various complexes formed with organic and inorganic
ligands, such as fluoride, sulfate, silicate, small organic acids
(e.g., acetate, oxalate, and citrate), and organic macromolecules
such as humic substances.2-4 These ligand complexes control
Al dissolution kinetics, solubility, and mobility, thereby affecting
mineral weathering and diagenetic processes.4-6 The toxicity
of Al to plants and fish is also dependent on speciation, which
is important in regions of high dissolved Al content resulting
from acid mine drainage and acid deposition.7-10
to a given spectroscopic method. Infrared and Raman spec-
troscopies have been used to study Al complexes with small
organic acids and humic substances,13-17 and Raman spectros-
copy has also been applied to the study of aqua-Al complexes
and their hydration structures.18-20 Though useful, these tech-
niques are limited to complexes that contain IR- and Raman-
active vibrational modes. 17O, 1H, 13C, 31P, and 19F NMR have
also been effective in elucidating structural details of Al
complexes from the ligand perspective21-28 but are limited to a
subset of ligands that contain NMR-active nuclei sensitive to
Al complexation.
In general, a more complete understanding of metal-ligand
structure can often be gained by combining ligand-based
methods with complementary, element-specific techniques that
probe the metal. 27Al-NMR spectroscopy has been widely used
to study Al solution complexes, often in combination with
ligand-targeted NMR spectroscopy, vibrational spectroscopy,
and molecular modeling. The 27Al-NMR method has been
applied with varying success to virtually all classes of aqueous
Al complexes, including monomeric hydrolysis species,29-31
polymers,32,33 inorganic complexes (e.g., sulfate and fluoride),21,34
and organic complexes (e.g., lactate,35,36 oxalate,37 citrate,35,38
salicylate,36 amino acids,22 and EDTA35,39,40). However, the
method is limited due to the high sensitivity of the quadrupolar
27Al nucleus to electric field gradients that may arise from
geometric distortions and mixed ligation in the Al coordination
sphere. This effect can cause substantial peak broadening in
the 27Al-NMR spectrum, limiting one’s ability to study or even
detect many environmentally relevant Al complexes.
For these reasons, there has been a great deal of interest in
understanding the chemistry of aqueous Al, including its
speciation (i.e., the relative abundance and stoichiometries of
Al-ligand complexes in solution) and the chemical and electronic
structure of individual Al species, the latter being necessary for
understanding and predicting chemical behavior. Numerous
analytical techniques have been applied to the study of Al
speciation,11,12 with vibrational and nuclear magnetic resonance
(NMR) spectroscopies proving particularly useful in resolving
both aqueous speciation and structure. Many of these techniques
* To whom correspondence should be addressed. E-mail: mbhay@usgs.gov.
† Department of Civil and Environmental Engineering, Princeton University.
‡ Department of Geosciences, Princeton University.
§ Lawrence Berkeley National Laboratory.
10.1021/jp909656q  2010 American Chemical Society
Published on Web 05/05/2010

XAS of Aqueous Al-Organic Complexes
J. Phys. Chem. A, Vol. 114, No. 20, 2010 6139
Materials and Methods
Another promising element-specific technique is Al K-edge
X-ray absorption near-edge structure (XANES) spectroscopy. Al-
edge XANES has been used extensively to study the structure of
Al in solid phases, but due to the relatively short penetration depth
of X-rays in air at the Al K-edge (∼1550 eV), the experiments are
typically done under vacuum. Attempts have been made to
overcome this experimental complexity by designing sample
chambers and solution cells that minimize beam attenuation, and
Al-edge XANES has now been applied to the study of dissolved-
phase Al.41-45 Matsuo and co-workers, using a thin solution cell
for collection of transmission-mode spectra, have applied the
Al XANES technique in combination with discrete variational
ꢀR (DV-ꢀR) calculations to study aqueous Al(H2O)63+ and Al-
ethylenediaminetetraacetate (Al(EDTA)-).41,42 Discrete varia-
tional ꢀR (DV-ꢀR) calculations were run for spectral inter-
pretation, and on the basis of these results, the authors proposed
Materials and Sample Preparation. Reagent-grade chemi-
cals, including aluminum chloride hexahydrate (AlCl3 ·6H2O),
hydrochloric acid (HCl), sodium hydroxide (NaOH), malic acid
disodium salt (Na2 · malate), oxalic acid disodium salt (Na2 ·
oxalate), salicylic acid monosodium salt (NaH ·salicylate),
acetohydroxamic acid (aHA), desferrioxamine B mesylate
(dfoB), and ethylenediaminetetraacetic acid disodium salt
(Na2 ·EDTA) were purchased from Sigma-Aldrich. Aluminum
chloride and organic acid 100 mM stock solutions were prepared
in plastic bottles using deionized water (Milli-Q, 18 MΩ
resistivity). Organic stock solutions were prepared 1-2 days
in advance to allow complete dissolution. Aluminum-organic
solutions were prepared by addition of stock solutions to
deionized water, followed by slow NaOH addition to yield the
desired Al concentration (20 mM), Al-organic ratio, and pH.
This procedure ensured that the majority of the Al was in an
organic-complexed form before base addition. A pH of 4.0 was
chosen to maximize ligand complexation while preventing Al
polymer formation and precipitation. Higher pH Al-EDTA
solutions were prepared in the same manner, with higher EDTA
to Al solution ratios to ensure complete complexation. After
the mixing and base addition steps, solutions were analyzed by
spectroscopy within the same day. pH values were measured
using an Orion model 525A pH meter with Orion PerpHecT
ROSS Model 8203 and Orion 9107BN pH electrodes following
a three-point calibration with buffers at pH 4.0, 7.0, and 10.0.
XANES Spectroscopy. X-ray absorption spectra were col-
lected in the modified Soft X-ray Endstation for Environmental
Research (SXEER-2)43,44 on the Molecular Environmental
Sciences (MES) beamline 11.0.2 at the Advanced Light Source,
Lawrence Berkeley National Laboratory (Berkeley, CA). Fluo-
rescence-mode X-ray absorption spectra were collected on liquid
samples in a 1 atm He environment using either a GaAsP
photodiode or a photomultiplier tube outfitted with a phosphor
scintillator. Two types of solution cells were used in the
experiments: a Plexiglas solution cell fitted with a 100 nm thick
Si3N4 window and a vertically aligned plastic beverage straw,
plugged at the bottom with Teflon tape and with a small
rectangular slit cut into the side. The liquid sample was retained
in the straw by surface tension, allowing the X-ray beam to
contact the air-water interface at the rectangular slit.
3+
a distorted 6-coordinate geometry for Al(H2O)6 and a 5-co-
ordinate geometry for Al(EDTA)-. The Al XANES technique
was also used by Matsuo et al. to track the structural
transformation of Al during polymerization and precipitation
-
of gibbsite from Al(OH)4 on pH reduction, illustrating its
potential for monitoring chemical changes in real-time.45 In a
recent report,44 we presented a detailed analysis of the aqueous
Al(H2O)63+ fluorescence-mode XANES spectrum and its struc-
tural implications using density functional theory (DFT) based
XANES modeling, illustrating that the combined experimental
and modeling approach is effective in elucidating the electronic
structure of the complex. We provided a modified interpretation
of the XANES spectrum, arguing that it is in fact consistent
with the highly symmetric octahedral structure proposed on the
basis of ab initio geometry optimizations reported in the
literature.
In this work, we extend the use of the Al K-edge XANES
technique to the study of mixed-ligand aqueous Al species,
including organic complexes with oxalate, salicylate, malate,
acetohydroxamate (aHA), desferrioxamine B (dfoB), and EDTA.
Oxalate, salicylate, and malate are typical examples of the small
carboxylic acids abundant in natural waters and are representa-
tive of the metal-binding functional groups in humic sub-
stances.46 Desferrioxamine B, a trihydroxamate siderophore
secreted by microbes for acquiring Fe, is also abundant in many
soils,47 and aHA is often used as a simpler structural analogue
in chemical studies of dfoB.48,49 These particular organic acids
were chosen not only for their environmental relevance, but also
to systematically study the effects of mixed aqua-organic oxygen
ligation on the aqueous Al XANES spectrum in systems with
relatively straightforward aqueous speciation. The comparison
shows a clear, predictable narrowing trend in the Al XANES
spectrum as water molecules are replaced by organic oxygen
ligands in the Al coordination sphere. We also revisit the
Al(EDTA)- spectrum previously described by Matsuo,42 ex-
tending the study of this system to include the AlOH(EDTA)2-
complex dominant at higher pH. On the basis of aqueous
Fourier-transform infrared (FTIR) spectra and DFT-based
XANES calculations, an octahedral-hexadentate structure is
proposed for Al(EDTA)-, consistent with the literature con-
sensus. A similar octahedral structure is predicted for AlOH-
(EDTA)2-, with one acetate ligand displaced by a hydroxyl.
This work provides new insights on the geometric and electronic
structures of some geochemically relevant aqueous Al com-
plexes, while further demonstrating the utility of the XANES
technique as an additional tool for elucidating the speciation
and structure of aqueous-phase Al.
XANES spectra were collected between 1550 and 1600 eV,
with step sizes of 0.2 eV in the edge region between 1560 and
1573 and 0.5 eV below and above the edge. Baseline correction
and normalization were performed by fitting straight lines to
the spectra well below and above the edge, then rescaling the
spectra by zeroing the line slopes and setting pre-edge and
postedge line intercepts to 0 and 1, respectively. In some of the
overlay graphs, spectra were slightly rescaled based on peak
height to compare spectral shapes in the XANES region. Energy
calibration was performed periodically by collecting a spectrum
on a single-crystal corundum sample and shifting the first
XANES transition to 1567.5 eV, as described in ref 44.
Normalization and calibration were performed using WinXAS50
and Microsoft Excel.
Infrared Spectroscopy. Aqueous FTIR spectra were col-
lected in attenuated total reflection (ATR) mode using a Bruker
IFS 66v/S spectrometer, equipped with a liquid nitrogen cooled
mercury cadmium telluride (MCT) detector, with an aperture
setting of 4 mm. The sample spectra were collected on a ZnSe
trough-style ATR crystal, housed in a Spectra-Tech ATR unit.
IR spectra were obtained from an average of 5000 interferometer
scans. Spectra of clean (Milli-Q) water were collected before

6140 J. Phys. Chem. A, Vol. 114, No. 20, 2010
Hay and Myneni
and after sample collection for use in subtraction of ZnSe
Molden68), useful in qualitatively assessing the atomic orbital
contributions involved in bonding.
features and the water feature near 1640 cm-1
.
Calculations were run using the IGLO-III basis set69 on Al
and DZVP basis set70 on the ligands (C, H, O, and N). For most
of the complexes we studied, use of the IGLO-III basis set on
all elements would have exceeded the maximum number of
orbital Gaussians allowed in StoBe (version 1.0 for Microsoft
Windows), and in our previous study the IGLO-III/DZVP
combination performed well.44 The ionization potential (IP) was
recalculated as the difference in total energy between structures
with Al R-1s orbital occupations of 1 and 0. This new value
was typically lower than the IP calculated for the half-corehole
complex by 2.6-2.8 eV, depending on the complex. An initial
energy calibration was performed by shifting the half-corehole
IP and transition energies downward by this amount. Compari-
son with experimental spectra required an upward reshifting of
the transitions by 4.0-4.5 eV, depending on the complex. In
our previous study, this shift was found to vary with the basis
set of the absorbing element.44 In this study, as in the previous,
the absolute value of this shift is not of critical importance.
Speciation Calculations. The aHA acid dissociation constant
and stability constants for the Al-aHA system (T ) 25 °C,
ionic strength (µ) ) 0.2 M) were taken from Farkas et al.,71
and stability constants (not including acid dissociation constants)
for the Al-malate system (T ) 37 °C, ionic strength (µ) )
0.15 M) were taken from Venturini-Soriano and Berthon.72
Stability constants for all other species, including the malic acid
dissociation constants, were collected from the Martell and
Smith database73 and represent values at T ) 25 °C and varying
µ. Stability constants were included for Al dimer and trimer
species but not for Al13 or Al solids. As necessary, stability
constants were converted to overall formation constants (ꢀ) and
extrapolated to zero ionic strength using activity coefficients
calculated with the Davies equation. The formation constant
values used are given in the Supporting Information. Speciation
calculations were performed with ionic strength corrections
Calculation of X-ray Absorption Spectra. X-ray absorption
transition energies and oscillator strengths were calculated with
density functional theory, implemented in the StoBe software
package51 using the gradient corrected exchange functional of
Becke52 and correlation functional of Perdew.53 In StoBe, X-ray
absorption transition energies and probabilities (oscillator
strengths) are calculated using the transition-potential method.54
In this method, initial and final states are calculated using the
same potential; one that corresponds to a “transition state” in
which the R-1s orbital is assigned an electron occupation of
one-half. This procedure accounts for the majority of the
relaxation energy on core excitation, while simplifying the
calculation of transition energies and intensities.54,55 The DFT
level of theory implemented in StoBe, an improvement over
the traditional ꢀR-multiple scattering method with a “muffin-
tin” potential,54,56 has been used to accurately describe XANES
spectra at the O K-edge of water and ice (refs 57-59 and
references therein), the C, N, and O K-edges of numerous
organic molecules (refs 49, 60-64 and references therein), the
Fe L2/L3 edges of a ferrocene-labeled peptide,63 and recently
the Al K-edge of Al(H2O)63+ in our previous report,44 to name
a few examples. Advances in theoretical XANES modeling
continue to be made, with real-space multiple scattering and
time-dependent density functional theory (TDDFT) methods
being active areas of research (e.g., refs 65-67 and references
therein). StoBe was chosen in our work, however, as it is a
fast, relatively simple, well-established, commercially available,
and easily accessible molecular orbital-based method for semi-
quantitatively understanding orbital contributions involved in
XANES transitions.
Although the raw transition energies and intensities were
much more important in our interpretations of the experimental
XANES spectra, synthetic spectra were also generated from the
calculated transitions to qualitatively assess agreement between
theory and experiment. These synthetic spectra can be generated
within StoBe by assigning Gaussian peaks to each of the
calculated transitions, with peak areas fixed by the oscillator
strength. In StoBe, the widths of the Gaussian peaks are assigned
according to a continuous, piecewise linear function of energy
defined by four variables, E1, E2, w1, and w2. Specifically, a
constant width of w1 is assigned below energy E1, a constant
width of w2 is assigned above energy E2, and the width is varied
linearly between energies E1 and E2. These width and energy
parameters are typically chosen by the user to obtain a good fit
with an experimental spectrum. Increasing the peak width with
energy in this fashion accounts for two effects: the decrease in
lifetime of the excited state (and hence wider peak) with
increasing energy above the ionization potential, and the way
in which StoBe handles continuum-state transitions above the
ionization potential. These transitions are approximated in StoBe
by addition of a large, secondary basis set of diffuse orbitals,
such that the continuum is represented by a large but limited
series of discrete transitions. In order to represent the postedge
continuum in the spectra, these transitions require larger peak
widths than those representing the sharp bound state transitions
closer to the absorption edge. Although we do not necessarily
expect the increase in peak width to be piecewise linear, the
technique is a simple way to fit a synthetic spectrum with very
few parameters. The method has also proven adequate in several
previous studies (58 and references therein). Finally, the StoBe
program also calculates three-dimensional (3D) electron densi-
ties for each molecular orbital (viewed using the program
(Davies equation) using Visual MINTEQ version 2.50,74
a
Microsoft Windows version of MINTEQA2 version 4.0.75 The
pH 4 results (Table 1) were obtained by maintaining charge
balance in the calculation and iteratively adjusting the Na+
concentration until a pH of 4.0 was calculated. This procedure
closely mimicked the experimental procedure, which involved
NaOH addition with no attempts to fix ionic strength. In all
cases, the calculated Na+ content was similar to the amount of
NaOH required in the experiments to reach pH 4.
Results and Discussion
The XANES spectra of the aqueous Al-organic solutions
studied are shown in Figure 1, along with aqueous AlCl3
solutions at low and high pH (Figure 1f,g). The energy of the
first Al XANES peak is a good indicator of coordination number
for Al, and the peak at 1569.2 eV for 20 mM AlCl3 at pH 3.6
(Figure 1f) is consistent with the 6-coordination of the
Al(H2O)63+ complex (Figure 2a), the dominant solution species
at this pH.44 For comparison, the spectrum of a 20 mM AlCl3
solution adjusted to pH 12.3 is shown in Figure 1g. At this
pH, the dominant species in solution is the tetrahedral
-
Al(OH)4 complex (Figure 2b).30 The complex exhibits a
bound state transition at 1565.8 eV, consistent with 4-coor-
dinate Al-containing minerals that exhibit this peak near 1566
eV.76,77 With the exception of the high pH Al-EDTA solution
(Figure 1c, discussed below), the peak energies for the
Al-organic spectra are within 1569.1-1569.8 eV, just above
3+
that of the Al(H2O)6 complex. This is slightly above the

XAS of Aqueous Al-Organic Complexes
J. Phys. Chem. A, Vol. 114, No. 20, 2010 6141
TABLE 1: Chemical Speciation of the Al-Organic Solutions
Used in the XANES Experimentsa
20 mM AlCl3 + 20 mM aHA,b pH 4.0 (18.4 mM Na+)c
conc. (mM)
% Al
% aHA
Al3+
5.88
0.18
9.50
4.09
0.13
2.01
29.7%
0.9%
48.0%
20.7%
0.6%
Al(OH)2+
Al(aHA)2+
47.5%
20.5%
0.6%
+
Al(aHA)2
0
AlOH(aHA)2
H(aHA)0
10.0%
20 mM AlCl3 + 60 mM aHA, pH 4.0 (38.0 mM Na+)
conc. (mM)
% Al
% aHA
Al3+
0.14
2.95
15.54
0.86
0.51
22.39
0.7%
14.7%
77.7%
4.3%
Al(aHA)2+
4.9%
25.9%
1.4%
0.8%
37.3%
+
Al(aHA)2
Al(aHA)3
0
0
AlOH(aHA)2
2.5%
H(aHA)0
20 mM AlCl3 + 20 mM Na2Malate, pH 4.0 (24.8 mM Na+)
conc. (mM)
% Al
% Malb
Al3+
0.11
0.23
1.02
0.10
0.24
0.57
1.55
0.53
3.03
0.21
0.5%
1.1%
5.1%
1.0%
2.4%
5.7%
15.5%
7.9%
60.6%
Al(Mal)+
AlOH(Mal)0
Al2OH(Mal)3
1.1%
5.1%
1.5%
1.2%
2.8%
15.5%
10.6%
60.6%
1.0%
-
Al2(OH)2(Mal)2+
Al2(OH)3(Mal)+
-
Al2(OH)3(Mal)2
3-
Al3(OH)4(Mal)4-
Al4(OH)5(Mal)4
H(Mal)-
20 mM AlCl3 + 20 mM Na2Oxalate, pH 4.0 (3.2 mM Na+)
conc. (mM)
% Al
% Oxalb
Al3+
2.59
11.43
2.64
0.04
0.62
0.84
12.9%
57.1%
13.2%
0.2%
3.1%
12.5%
Figure 1. Al K-edge XANES spectra of aqueous AlCl3 and Al-organic
solutions. All spectra were collected on solutions containing 20 mM
Al at pH 4.0, unless indicated otherwise in the legend. Where specified,
ratios in the legend refer to the total Al to organic concentration ratio,
not the stoichiometry of the dominant solution species.
Al(Oxal)+
57.1%
26.4%
0.5%
3.1%
12.5%
-
Al(Oxal)2
Al(Oxal)3
3-
AlOH(Oxal)0
0
Al3(OH)3(Oxal)3
20 mM AlCl3 + 20 mM (NaH)Salicylate, pH 4.0 (18.6 mM Na+)
considered would affect the detailed percentages given in Table
1, it is unlikely that the true dominant species are different from
those indicated by the calculations.) Speciation results for each
solution are discussed in more detail below. The configurations
of some of the dominant solution complexes are illustrated in
Figure 2, with hypothetical configurations shown for complexes
with unknown or debatable structure, including the Al-malate
and Al-salicylate complexes. The speciation results and
XANES spectra for these solutions are described below,
beginning with a discussion of the Al(dfoB)+ complex, in which
the Al ion is fully complexed by organic O ligands (in this case,
O from CdO and N-O groups). This is followed by results
obtained for structures with more complex mixed-ligation. These
include Al(EDTA)-, which involves complexation by organic
O and N ligands, followed by cases involving sequential
replacement of organic O ligands by OH- and H2O.
Organic O Ligation: Al-dfoB. Desferrioxamine B (dfoB)
is a common, well-studied siderophore found in soils, which is
known to form a strong, symmetric, hexadentate aqueous
solution complex with Fe.61,78 The binding constant for Al,
although not as high as for Fe (log K ) 24.14 for Al vs 30.6
for Fe(III)),73 indicates that the Al-dfoB complex is quite
strong. The speciation calculation for the 20 mM AlCl3, 20 mM
dfoB solution at pH 4 indicates that 99.8% of the Al in solution
is bound within the Al(dfoB)+ species, with the terminal amine
conc. (mM)
% Al
% Salb
Al3+
1.61
0.05
17.88
0.31
0.06
0.10
1.27
8.0%
0.3%
89.4%
1.6%
0.6%
AlOH2+
Al(Sal)+
Al(Sal)2
89.4%
3.1%
0.6%
0.5%
6.3%
-
0
Al2(OH)2(Sal)2
H2(Sal)0
H(Sal)-
a Only species representing more than 0.5% of total Al or organic
are listed. b aHA ) acetohydroxamic acid, Mal ) malate, Oxal )
oxalate, Sal ) salicylate. c Amount of excess Na+ added in the
charge-balanced calculation to achieve a pH of 4.0.
observed range for octahedral coordination in Al-bearing solid
phases but is still consistent with 6-fold coordination for these
complexes.
Chemical speciation calculations were conducted to determine
the relative abundance of various Al-organic complexes present
for the solution conditions used in the XANES experiments.
Species abundances are listed on a percentage basis in Table 1
for illustrative purposes, but it is important to note that our
primary goal with these calculations was merely to assess which
species are most likely in the majority. (Although inaccuracies
in the reported stability constants for the Al-organic species

6142 J. Phys. Chem. A, Vol. 114, No. 20, 2010
Hay and Myneni
Figure 2. Structures of the aqueous Al species dominant in solution under the conditions investigated in the XANES experiments. Only one ligand
configuration is shown for each complex, although multiple configurations may exist for some species; for example, one of the two aHA ligands
in Al(aHA)2+ could be rotated to displace an axial water ligand. Two configurations are shown for Al(salicylate)+, since the structure of this species
is uncertain.
protonated below pH 9. We assume that the bonding geometry
is similar to that of the Fe-complex, with bidentate chelation
by all three of the hydroxamate ligands in the structure (Figure
2d), consistent with the DFT-optimized Al(dfoB)+ structure of
Domagal-Goldman et al.79 This is also consistent with the
0
bonding environment proposed for the Al(aHA)3 complex on
the basis of 27Al-NMR spectroscopy (Figure 2c).71
The XANES spectrum of the 1:1 Al-dfoB solution (Figure
1a) exhibits a strong feature at 1569.8 eV, appearing much
narrower in width than the Al(H2O)63+ feature. Peak broadening
in XANES spectra can often be attributed to transition splitting
resulting from geometric distortions. However, our previous
3+
DFT calculations indicated that the broadness of the Al(H2O)6
spectrum is caused by the presence of multiple Al 3p - O 2s/p
transitions, observed for both highly symmetric and distorted
3+
complexes.44 The spectral differences observed for Al(H2O)6
and Al(dfoB)+ are therefore assumed to be due to the differences
in ligands, rather than geometry effects.
Theoretical XANES spectral simulations were performed in
StoBe using Al(aHA)30 as a model for the Al(dfoB)+ complex.
The larger Al(dfoB)+ structure could not be run in StoBe with
the chosen basis sets due to size constraints, but the similarity
in coordination environments between the two complexes makes
0
Al(aHA)3 a very good model (Figure 2). To test exclusively
for ligand effects, the structure of the coordination sphere was
assigned the same highly symmetric geometry used in previous
calculations on Al(H2O)63+, with Al-O bond lengths set to 1.90
Figure 3. XANES spectrum of the Al(dfoB)+ complex compared with
StoBe calculation results for the structurally analogous Al(aHA)3
Å and O-Al-O bond angles set to 90°.44 Bond lengths and
angles within the acetohydroxamate ligand were assigned based
on the DFT-optimized, hydroxyl deprotonated cis-aHA structure
of Edwards et al.78,80 After a 4.4 eV correction (similar to the
4.6 eV correction required for Al(H2O)63+), the synthetic
spectrum produced from the calculated XANES transitions
closely reproduces the features observed in the Al(dfoB)+
spectrum (Figure 3). The results suggest that the sharp, narrow
XANES feature primarily results from one set of three degener-
ate transitions, labeled “i.” in the figure, exhibiting Al 3p - O
2p antibonding character (orbital plot i. in Figure 3). A smaller
contribution comes from transition set “ii.”, which also exhibits
Al 3p - O 2p antibonding character, but with apparently less
electron density on the Al-bound ligands.
0
complex. The black vertical lines represent energies of calculated Al
1s to valence orbital transitions (with line heights proportional to
transition probability); the thick gray vertical line is the calculated
ionization potential; and the synthetic spectrum calculated from the
transitions (thin black curve) overlays the experimental spectrum (gray
circles). Calculation results have been shifted +4.4 eV to achieve
alignment with experiment. One representative electron density image
is shown below the plot for each of the first two major transition
clusters, indicated by roman numerals. The synthetic spectrum was
generated using the following parameters (in eV): E1 ) 1573.7, E2 )
1581.7, w1 ) 4.3, w2 ) 10.
0
of the Al 3p - O 2p interaction is observed here for Al(aHA)3 ,
The Al(aHA)30 calculation results differ markedly from those
the effect is not as strong as with Al(H2O)63+. Further, no
0
obtained previously for Al(H2O)6
.
3+ 44 Although some splitting
significant Al 3p - O 2s contribution is observed for Al(aHA)3 ,

XAS of Aqueous Al-Organic Complexes
J. Phys. Chem. A, Vol. 114, No. 20, 2010 6143
highlighting important differences in the electronic structures
0
of the two complexes. The results for Al(aHA)3 may also be
compared with the results of Domagal-Goldman et al., who
assert based on natural bond orbital (NBO) calculations that
the Al-O bonds within the Al(dfoB)+ complex are exclusively
ionic.79 Our calculation results are not consistent with this
assertion, since the electron density images in Figure 3 do
suggest some Al 3p - O 2p orbital interaction. However, these
density plots are only a qualitative indicator of orbital interac-
tions and cannot be used to accurately assess the degree of
covalent bonding without further processing and validation. In
addition, the experimental spectrum itself is not necessarily
inconsistent with the idea that the Al-O bonds are ionic, since
an Al 1s - 3p transition without significant electron sharing
between atomic Al and O orbitals would also yield the single,
narrow XANES feature observed. It is important to note that
0
an energy-optimized Al(aHA)3 or Al(dfoB)+ structure might
exhibit different bond lengths and angles than the idealized
structure used here, which may have an influence on the XANES
calculation results. However, given the degree to which the
calculations reproduce the experimental spectrum, the idealized
0
Al(aHA)3 structure used here is assumed to be sufficient for
spectral interpretation.
Organic O/N Ligation: Al-EDTA. The XANES spectra
collected on Al-EDTA solutions at pH 4.0 and 7.5 are shown
in Figure 1, panels b and c, respectively. Speciation calculations
indicate that the Al(EDTA)- and AlOH(EDTA)2- complexes
are dominant in solution at these respective pH values (Figure
S2, Supporting Information). To ensure complete complexation
of Al, the solutions were prepared with greater proportions of
EDTA (30-40 mM) relative to the 20 mM Al. The Al(EDTA)-
XANES spectrum (Figure 1b) has a single peak centered at
1569.2 eV. It lacks the higher-energy shoulder present in the
Al(H2O)63+ spectrum (Figure 1f) but exhibits a slight shoulder
at 1567 eV, between the 4- and 6-coordinate regions highlighted
in the figure. The spectrum of the hydroxylated AlOH(EDTA)2-
complex at pH 7.5 (Figure 1c) has a lower energy peak at 1568.3
eV, which either includes or replaces the 1567 eV shoulder
observed for Al(EDTA)-. The emergence of these features
between the 4- and 6-coordinate regions may be the result of N
ligands bound directly to Al, or it may be due to geometric
factors, indicating either a distorted 6-coordinate or possibly a
5-coordinate geometry for these structures.
Figure 4. ATR-FTIR spectra of 20 mM Na2EDTA (a-c) and 20 mM
AlCl3 + 20 mM Na2EDTA (d, e) solutions as a function of pH. The
legends give the dominant aqueous species (discussed in the text and
Supporting Information), along with solution pH in parentheses.
less than 1 eV for the pH 7.5 spectrum, suggesting a possible
calibration error. Solution pH and calibration procedures were
not reported in their study.
We used aqueous infrared spectroscopy to further verify the
structures of the Al(EDTA)- and AlOH(EDTA)2- complexes.
The asymmetric stretch of the carboxylate group, νas, is highly
sensitive to H-bonding, protonation state, and metal binding,
and can be used to infer carboxylate binding modes in solids
and solution.84-86 This technique was used here to probe the
EDTA acetate group interactions with Al. Infrared spectra were
first collected on aqueous EDTA in the absence of Al, shown
in Figure 4a-c. The pH 11.3 spectrum (Figure 4a) exhibits a
single band at 1574 cm-1 corresponding to the νas of the free
carboxylates in EDTA4-, the dominant solution species at this
pH (Figure S1, Supporting Information). At pH 4.1 (Figure 4b),
the H2EDTA2- species is dominant (Figure S1), and the νas band
increases to 1618 cm-1 due to hydrogen bonding between the
free carboxylates and the protons in the structure, which are
believed to reside on the amine groups.87-89 As the pH is
lowered further and the EDTA acetate groups become proto-
nated, the νas band is replaced by a weaker carbonyl stretch at
1730 cm-1 (pH 0.44, Figure 4c). These values are consistent
with previous infrared results obtained for EDTA in D2O by
Nakamoto, et al.86,88
The coordination geometry of the Al(EDTA)- complex has
been a subject of debate in the literature. Proposed structures
have included 7-fold coordination involving the six EDTA
ligands and a water molecule,81 6-fold coordination involving
either the six EDTA ligands or a combination of EDTA ligands
and water molecules,35,39,40,82,83 and a 5-coordinate structure
involving four EDTA ligands and a water molecule.42 Of these
possibilities, the strongest evidence is in support of the
6-coordinate, hexadentate structure (full coordination by the
EDTA with no water ligands in the coordination sphere), which
has been proposed based on 27Al-NMR spectroscopy,39 13C- and
1H NMR spectroscopy,82 and ab initio calculations.83 The
5-coordinate structure proposed by Matsuo et al.42 was based
on Al-edge XANES spectroscopy, discrete variational ꢀR (DV-
ꢀR) XANES modeling, and unpublished NMR data. However,
their Al-EDTA spectrum differs from ours in important ways.
First, their XANES spectrum more closely resembles the pH
7.5 AlOH(EDTA)2- spectrum (Figure 1c) than the Al(EDTA)-
spectrum. Second, the peak-to-peak energy difference between
The aqueous FTIR spectra collected on equimolar (20 mM)
Al-EDTA solutions at pH 4.2 and 8.1 (Figure 4d,e) provide
insight into the mode of acetate binding in the Al(EDTA)- and
AlOH(EDTA)2- complexes. The Al(EDTA)- spectrum exhibits
a single νas band at 1654 cm-1, suggesting that all four of the
EDTA acetate groups are tightly bound to the Al, consistent
with the hexadentate binding mode previously proposed.39,82,83
The 80 cm-1 increase in νas relative to EDTA4- is consistent
3+
Al-EDTA and Al(H2O)6 in their study appears to be
approximately 3 eV, whereas our data indicate a difference of

6144 J. Phys. Chem. A, Vol. 114, No. 20, 2010
Hay and Myneni
with other general observations of metal binding by carboxylate
in solution,90-92 and the νas value of 1654 cm-1 is consistent
with trivalent metal-acetate interactions previously observed
in metal-EDTA solid salts.86,93 The AlOH(EDTA)2- spectrum
exhibits a νas band at 1637 cm-1, which is still consistent with
metal binding, but may be indicative of a slightly weaker
Al-O(-C) bond. This spectrum also exhibits a distinct shoulder
at 1592 cm-1, indicative of free or weakly H-bonded carboxy-
late. The ratios of the two peaks suggest that at most one of the
four EDTA acetate groups is detached from the Al in this
complex, presumably replaced by a hydroxyl group in the Al
coordination sphere. The infrared spectroscopy results therefore
suggest that both Al(EDTA)- and AlOH(EDTA)2- are 6-coor-
dinate. The structures proposed based on the IR evidence are
shown in Figures 2e and 2f.
For the XANES modeling of Al(EDTA)-, we used the
aqueous-phase structure recently calculated by Coskuner and
Jarvis,83 which was optimized using DFT (PBE/cc-pVTZ level
of theory) with the COSMO continuum solvation model. This
structure has the geometry shown in Figure 2e, with bond
lengths of 2.09 Å for Al-N, 1.88 Å for equatorial Al-O (i.e.,
those within the plane of the Al-N bonds) and 1.92 Å for axial
Al-O. Results obtained using this structure are shown in Figure
5a, including the calculated XANES transitions and the synthetic
spectrum (solid line), overlain with the experimental Al(EDTA)-
spectrum from Figure 1b (gray circles). A synthetic spectrum
was obtained that closely matched experiment after shifting the
calculated transitions 4.0 eV, which is slightly less than the 4.4
0
eV shift required for the Al(aHA)3 /Al(dfoB)+ comparison.
Strong transitions located near 1569 eV after correction
reproduce the main Al(EDTA)- peak, and a significant transition
near 1567 eV is present under the low-energy shoulder. All of
these transitions exhibit Al 3p character and varying amounts
of O 2p and N 2p character. The lower energy transition, labeled
transition “i.”, apparently contains slightly stronger N 2p
components relative to transition set “ii.” at higher energy, which
exhibits more O 2p character. This suggests that the broadening
of the spectrum and the appearance of the lower energy feature
are the result of mixed N/O ligation.
To investigate distortion effects, StoBe calculations were also
performed for an ideally symmetric Al(EDTA)- geometry
(Figure 5b). The structure was assembled by shifting the Al
and ligand positions to yield Al-O/Al-N bond lengths of 1.90
Å and bond angles of 90°, using the optimized structure as a
starting point. This structure exhibits similar transitions near
1569 eV, but the lower energy transition decreases in intensity.
This may suggest that distortion effects are controlling the
XANES features for this complex, but it could equally be argued
that optimum Al-O and Al-N bond lengths are the critical
factor, as this procedure required a significant shortening of the
Al-N bonds. Discussing ligand and distortion effects indepen-
dently in this case may therefore not be practical. An attempt
was also made to investigate AlOH(EDTA)2- with StoBe using
an approximate geometry assembled from the optimized
Al(EDTA)- structure. For this structure, one of the two axial
carboxylates was moved out of the coordination sphere by
rotating the acetate group 180° about the C-N bond, and an
OH- group was added in its place at a distance of 1.7 Å, yielding
a structure similar to the schematic in Figure 2f. XANES
calculation results for this structure are shown in Figure 5c,
shifted 4.0 eV for consistency with the Al(EDTA)- results. The
results do contain some weak transitions in the lower energy
region near 1568 eV, but they are not far enough above the
noise level to clearly account for the lower energy of the main
Figure 5. StoBe results for Al(EDTA)- and AlOH(EDTA)2- com-
plexes. (a) Calculated transitions (black vertical lines), ionization
potential (gray line), and synthetic spectrum (black curve) obtained
using the DFT-optimized Al(EDTA)- structure of Coskuner and Jarvis.
The synthetic spectrum (E1 ) 1573.3, E2 ) 1578.3, w1 ) 3.4, w2 )
10) overlays the experimental Al(EDTA)- XANES spectrum (gray
circles). Calculation results have been shifted +4.0 eV to achieve
alignment with experiment. Representative electron density images are
shown above the plot for the transition clusters indicated by roman
numerals. (b) Calculated transitions and ionization potential for an
Al(EDTA)- structure with a highly symmetric coordination sphere;
Al-ligand bond lengths of 1.9 Å and bond angles of 90°. (c) Calculated
transitions for an AlOH(EDTA)2- structure obtained via modification
of the structure in (a) as discussed in the text. The calculated ionization
potential occurs off the scale at 1558.6 eV. Results in (b) and (c) have
also been shifted 4.0 eV for consistency.
peak in the AlOH(EDTA)2- spectrum. It is clear that this
approximate, hypothetical structure does not capture the im-
portant features of the spectrum. A better analysis of this
spectrum could likely be obtained in the future by calculating
an optimized structure with inclusion of solvation waters.
Mixed Aqua/Organic Ligation: 1:3 Al-aHA and 1:1
Malate. In the 1:3 Al-aHA solution (20 mM AlCl3, 60 mM
aHA, pH 4), only ∼4% of the Al3+ is present in the Al(aHa)3
0

XAS of Aqueous Al-Organic Complexes
J. Phys. Chem. A, Vol. 114, No. 20, 2010 6145
+
aqueous complex. The majority of the Al is in the Al(aHA)2
complex, shown in Figure 2k, has been argued based on 27Al
NMR36,96 and 13C NMR spectroscopy.23 A monodentate, phenol-
protonated Al(H ·salicylate)2+ complex, however, has been
proposed as the dominant complex in low pH 1:1 Al-salicylate
solutions on the basis of infrared and UV resonance Raman
data collected at and below pH 3.8,16,97 as well as a reinterpreta-
tion of the published 27Al-NMR data collected at pH 3.3 on the
basis of chemical shift calculations.16,94 Potentiometry data on
this system are more consistent with the Al(salicylate)+ stoi-
chiometry (e.g., ref 98), but Trout and Kubicki argue that the
potentiometry analyses may not be applicable to the solution
conditions used in their resonance Raman study and/or may
contain inaccurate assumptions.16
form (78% total Al3+), with a significant fraction also in the
Al(aHA)2+ form (15% total Al3+, Table 1). Speciation of the
1:1 Al-malate solution (20 mM AlCl3, 20 mM Na2malate, pH
4) is more complicated due to the presence of polymeric species.
The calculation results suggest that this solution is dominated
-
-
by complexes such as Al4(OH)5(malate)4 , Al2(OH)3(malate)2 ,
and Al3(OH)4(malate)43-. The proportions listed in Table 1 are
approximate at best, since the Al-malate stability constants used
are only strictly valid at 37 °C.72 Regardless, we assume the
results are accurate enough to suggest that the majority of the
Al in solution is present in mixed H2O/OH/malate complexes.
Chelation by the malate in such complexes presumably accounts
for 2-3 of the 6 octahedral sites around the Al, with the
remaining 3-4 sites occupied by H2O and OH-. A purely
hypothetical example structure (not based on any theoretical or
experimental evidence) is shown in Figure 2d for the
For our particular set of solution conditions, some insight on
the 1:1 complex stoichiometry may be gained from the observed
changes in pH during solution preparation. The 20 mM 1:1
Al-salicylate solution was prepared by combining and diluting
100 mM stock AlCl3 and Na-H-salicylate solutions, followed
by slow adjustment to pH 4.0 with concentrated NaOH. This
procedure was repeated to study the detailed pH changes as a
function of NaOH addition (titration curve results are included
in the Supporting Information). Before adjustment to pH 4.0,
the pH of the 1:1 Al-salicylate solution had a measured value
of 2.51, whereas the individual AlCl3 and Na ·H·salicylate
stocks after dilution to 20 mM had pH values of 3.6 and 5.8,
respectively, clearly indicating the liberation of protons on
complex formation. Achieving a pH of 4.0 in the 1:1 solution
required addition of NaOH to a final concentration of ap-
proximately 18.5 mM. Assuming that the 1:1 Al-salicylate
complex is dominant at pH 4.0 as our speciation calculations
suggest, these results are consistent with the liberation of one
proton per complex, yielding the Al(salicylate)+ stoichiometry.
The close agreement between our pH titration curve and the
theoretical curve predicted from speciation calculations (SI,
Figure S3) suggests that our solution conditions are consistent
with published stability constants from pH 2.5 to 4.0. Within
this pH range, the 1:1 Al(salicylate)+ always makes up a
majority of the Al and salicylate in solution (SI, Figure S4),
although this does not rule out the possibility that the remainder
is present in a 1:1 Al(H ·salicylate)2+ complex not previously
considered in formulation of the stability constants. We therefore
assume, regardless of the arguments to the contrary based on
NMR predictions,16,94 that the Al(salicylate)+ complex is
dominant in this pH regime. Because the IR and resonance
Raman analyses suggest that the phenol is protonated in the
complex, an alternate possibility to the bidentate structure
(Figure 2k) is one in which an Al-bound water deprotonates
and is potentially stabilized by H-bonding between the hydroxyl
group and the protonated phenol (Figure 2l). This alternate
structure is referred to as the AlOH(H ·salicylate)+ species.
The XANES spectra for the three solutions (Figure 1e
overlay) are very similar, exhibiting a shape intermediate
between the broad, asymmetric spectrum of Al(H2O)63+ and the
narrower spectra of the organic complexes discussed above. In
the case of aHA, this may be partly due to the fact that
approximately half of the Al in solution is in fact a mixture of
Al(H2O)63+ and Al(aHA)2+. However, the Al(H2O)63+ proportion
in the Al-oxalate and Al-salicylate solutions is low enough
that the 1:1 Al-organic complexes are the primary contributors.
-
Al2(OH)3(malate)2 complex to illustrate what the ligand
distribution might look like. One could also imagine a structure
in which Al ions share OH- ligands, as in the Al dimer,
4+
Al2OH2(H2O)8
.
XANES spectra collected on the 1:3 Al-aHA and 1:1
Al-malate solutions are shown in Figure 1d. The overlay
illustrates that the two spectra are nearly identical, despite
differing organic ligands and complicated speciation. The spectra
are similar to the Al-dfoB spectrum, exhibiting a single feature
at 1569.8 eV, but are slightly wider. One common characteristic
of these two solutions is the ratio of water to organic-O ligands
in the inner coordination sphere of Al, which we suggest might
be responsible for the similarity of the two spectra. In the 1:3
Al-aHA solution, the majority of the Al is bound to two
oxygens from H2O and four from aHA. In the 1:1 Al-malate
solution, the majority of the Al is also bound to approximately
two (plus or minus one) oxygens from H2O, with the balance
consisting of organic O and OH-. On the basis of this
comparison, it is not yet clear if Al complexes containing
varying proportions of organic O and OH- can be resolved.
Mixed Aqua/Organic Ligation: 1:1 Al-aHA, Al-Oxalate,
and Al-Salicylate. In the equimolar Al-aHA, Al-oxalate, and
Al-salicylate solutions, between 50 and 90% of the Al is present
in 1:1 Al-organic complexes (Table 1). Smaller proportions
3+
of Al are also present as uncomplexed Al(H2O)6 and 1:2
Al-organic complexes. For the 1:1 Al-oxalate and Al-aHA
complexes, shown schematically in Figures 1i and 1j, bidentate
chelation is assumed. Bidentate chelation by aHA is likely, since
0
this binding mode has been observed in the Al(aHA)3 com-
plex.71 For oxalate, ab initio studies have demonstrated that the
bidentate Al(oxalate)+ species is more stable than possible
monodentate forms above pH 3.94,95 Kubicki et al.94 argued on
the basis of 27Al chemical shift calculations that a monodentate
form with a singly protonated oxalate would be more consistent
with the 27Al-NMR features observed by Thomas et al. on an
equimolar Al-oxalate solution at pH 3.3.37 However, a proto-
nated structure would be inconsistent with the pH titration results
reported by Thomas et al. for the same system, which indicate
that both protons are lost from oxalic acid on complexation with
Al.37 Further experimental evidence for the deprotonated bi-
dentate complex was provided by Clause´n et al., based on
EXAFS experiments with Ga as an Al analogue and a
comparison of FTIR results using Ga and Al at pH values below
4.0.15
The XANES spectra of the Al-organic complexes discussed
thus far exhibit a clear, linear trend with substitution of H2O
for organic-O ligands in the Al coordination sphere. Specifically,
as H2O is replaced by organic-O, the low-energy peak and broad,
high-energy shoulder characteristic of the Al(H2O)63+ spectrum
For salicylate, there is disagreement in the literature as to
whether the phenol group deprotonates at low pH on complex-
ation with Al. The phenol-deprotonated bidentate Al(salicylate)+

6146 J. Phys. Chem. A, Vol. 114, No. 20, 2010
Hay and Myneni
Figure 6. XANES spectra for the low pH AlCl3, Al-aHA, and
Al-dfoB solutions from Figure 1, overlaid for comparison. Descriptions
in the legend refer to solution compositions, not the dominant species
stoichiometries. The dominant species under the given solution condi-
tions are discussed in the text and Table 1.
begin to merge, creating a narrower and more symmetric feature
in the XANES region. This change with ligand substitution is
more clearly observed in Figure 6. In this figure, the XANES
results for the low pH AlCl3, Al-aHA, and Al-dfoB solutions
are overlaid, effectively illustrating spectral change as a function
of hydroxamate group substitution. Specifically, the high-energy
shoulder between 1571 and 1575 eV decreases in intensity, and
the main peak shifts up slightly to 1570 eV with the substitution
of water ligands by hydroxamate. To a large degree, these
Figure 7. Electron transitions (thin vertical lines) and ionization
potentials (thick gray vertical lines) for mixed aqua-organic Al
complexes as a function of hydroxamate substitution, calculated using
Stobe. Experimental XANES spectra from Figure 1 are shown for
comparison, with solution conditions indicated in parentheses in the
legend.
observations are reproduced by the XANES modeling. Figure
3+
7 shows the calculated electron transitions for Al(H2O)6
,
Al(aHA)2+, Al(aHA)2+, and Al(aHA)3 , compared with the
experimental spectra from Figure 6. Identical coordination shell
geometries were assigned to the four structures (Al-O ) 1.9
0
3+
Å, O-Al-O angle ) 90°). The Al(H2O)6 structure with 12
Summary and Conclusions
point-charge solvation waters in the second shell was used in
the comparison, since this structure better reproduced the
experimental spectrum (see Figure 5 in ref 44). In this structure,
two point-charge water molecules were assigned to each water
ligand in the complex in an H-bond accepting position.44 In the
Al-edge X-ray absorption near-edge structure (XANES)
spectra were collected on a series of Al-organic complexes to
study the effects of coordination environment on the spectra.
The purposes of these experiments were two-fold: (1) to reveal
insights into the geometric and electronic structures of the
specific complexes studied and (2) to illustrate more generally
the utility of the XANES technique as a tool for studying
dissolved Al speciation. The XANES spectra were interpreted
using the DFT-based software package StoBe.
+
Al(aHA)2+ and Al(aHA)2 structures, the second shell point
charges were included for the remaining water ligands and
removed for the water ligands replaced by aHA. Again, these
structures do not reflect the changes in coordination geometry
that likely occur with ligand exchange. Using these structures,
it was difficult to generate synthetic spectra that adequately
reproduced the experimental data, so synthetic spectra are not
shown for comparison. From a qualitative perspective, however,
the calculated transitions exhibit the types of changes observed
in the experimental spectra; the Al 3p - O 2p interactions
consolidate and move to lower energy, and the Al 3p - O 2s
interaction is gradually lost with exchange of H2O by aHA.
It is not clear if organic O and hydroxyl ligands are as easily
resolved in the XANES spectrum. It was observed above that
the 1:3 Al-aHA and 1:1 Al-malate solution spectra were very
similar, despite the presence of hydroxyl groups in the dominant
Al-malate complexes. Unfortunately, this observation makes
it difficult to resolve which of the two Al-salicylate structures
shown in Figure 2 is the true structure, based on XANES
arguments. In other words, although it is tempting to argue that
the bidentate structure is the most likely based on the spectral
similarity between Al-salicylate, Al-aHA, and Al-oxalate,
the Al-salicylate spectrum is not necessarily inconsistent with
the monodentate AlOH(H ·salicylate)+ structure.
Synthetic spectra generated using StoBe reproduced the
3+
observed differences in XANES spectra between Al(H2O)6
and the hexadentate Al-organic complex Al(dfoB)+. The
calculation results indicate that whereas the Al(H2O)63+ spectrum
is influenced by multiple Al 3p - O 2s and 2p interactions, the
Al(dfoB)+ spectrum is dominated by a single Al 3p - O 2p
transition. Calculated spectra also reproduced the XANES
spectrum for the hexadentate Al(EDTA)- complex, which
exhibited unique features attributable to the mixed O/N ligation
within the coordination sphere. Infrared spectroscopy evidence
was also presented that confirmed the hexadentate structure of
the complex. StoBe calculation results were less successful in
reproducing the AlOH(EDTA)2- features, which may have been
due to the lack of a more appropriate, energy-optimized
structure.
Spectra collected on mixed aqua-organic Al complexes
exhibited highly systematic trends. It was observed that Al
complexes involving different organic ligands (acetohydrox-
amate, malate, oxalate, and salicylate) yield very similar spectra

XAS of Aqueous Al-Organic Complexes
J. Phys. Chem. A, Vol. 114, No. 20, 2010 6147
(20) Gaspar, A. M.; Alves Marques, M.; Cabaco, M. I.; de Barros
Marques, M. I.; Kolesnikov, A. I.; Tomkinson, J.; Li, J.-C. J. Phys.:
Condens. Matter 2004, 16, 6343.
when the ratios of organic O to H2O in the coordination sphere
are similar. Further, the XANES spectra exhibit a steady
narrowing of the near-edge feature on progressive replacement
of H2O by organic O in the Al coordination sphere. This
observation was qualitatively reproduced in spectral calculations
on the Al-H2O-aHA system. Spectra collected on mixed
complexes believed to contain OH- in the coordination sphere
appeared similar to those that do not, suggesting that differences
arising from OH- and organic-O may be difficult to resolve
without further studies on structurally well-known complexes.
Overall, these results demonstrate that the XANES technique
has the sensitivity to resolve many structural aspects of aqueous
Al complexes. Used in combination with 27Al-NMR and other
ligand-based spectroscopic methods, this technique represents
a powerful tool in elucidating aqueous Al structure, and one
that will see greater use in the future as synchrotron-based tools
become more readily available and user-friendly.
(21) Martinez, E. J.; Girardet, J.-L.; Morat, C. Inorg. Chem. 1996, 35,
706.
(22) Kiss, T.; So´va´go´, I.; To´th, I.; Lakatos, A.; Bertani, R.; Tapparo,
A.; Bombi, G.; Martin, R. B. J. Chem. Soc., Dalton Trans. 1997, 1967.
(23) Yokoyama, T.; Abe, H.; Kurisaki, T.; Wakita, H. Anal. Sci. 1997,
13, 425.
(24) Nordin, J. P.; Sullivan, D. J.; Phillips, B. L.; Casey, W. H. Inorg.
Chem. 1998, 37, 4760.
(25) Yokoyama, T.; Murata, T.; Kinoshita, S.; Wakita, H. Anal. Sci.
1998, 14, 629.
(26) Swaddle, T. W.; Rosenqvist, J.; Yu, P.; Bylaska, E.; Phillips, B. L.;
Casey, W. H. Science 2005, 308, 1450.
(27) Tashiro, M.; Fujimoto, T.; Suzuki, T.; Furihata, K.; Machinami,
T.; Yoshimura, E. J. Inorg. Biochem. 2006, 100, 201.
(28) Jo´szai, R.; Purgel, M.; Pa´pai, I.; Wakita, H.; To´th, I. J. Mol. Liq.
2007, 131-132, 72.
(29) Akitt, J. W. Prog. NMR Spec. 1989, 21, 1.
(30) Moolenaar, R. J.; Evans, J. C.; McKeever, L. D. J. Phys. Chem.
1970, 74, 3629.
(31) Faust, B. C.; Labiosa, W. B.; Dai, K. H.; MacFall, J. S.; Browne,
B. A.; Ribeiro, A. A.; Richter, D. D. Geochim. Cosmochim. Acta 1995, 59,
2651.
(32) Bertsch, P. M.; Parker, D. R. In The EnVironmental Chemistry of
Aluminum, 2nd ed.; Sposito, G., Ed.; CRC Press: Boca Raton, FL, 1996;
Ch 4.
(33) Bi, S.; Wang, C.; Cao, Q.; Zhang, C. Coord. Chem. ReV. 2004,
248, 441.
(34) Akitt, J. W.; Farnsworth, J. A. J. Chem. Soc. Faraday Trans. 1985,
81, 193.
(35) Karlik, S. J.; Tarien, E.; Elgavish, G. A.; Eichhorn, G. L. Inorg.
Chem. 1983, 22, 525.
(36) Thomas, F.; Masion, A.; Bottero, J. Y.; Rouiller, J.; Montigny, F.;
Gene´vrier, F. EnViron. Sci. Technol. 1993, 27, 2511.
(37) Thomas, F.; Masion, A.; Bottero, J. Y.; Rouiller, J.; Gene´vrier, F.;
Boudot, D. EnViron. Sci. Technol. 1991, 25, 1553.
(38) Lakatos, A.; Ba´nyai, I.; Decock, P.; Kiss, T. Eur. J. Inorg. Chem.
2001, 461.
(39) Iyer, R. K.; Karweer, S. B.; Jain, V. K. Magn. Reson. Chem. 1989,
27, 328.
(40) Yokoyama, T.; Kurisaki, T.; Kinoshita, S.; Matsuo, S.; Wakita, H.
Anal. Sci. 2000, 16, 647.
(41) Matsuo, S.; Wakita, H. Struct. Chem. 2003, 14, 69.
(42) Matsuo, S.; Shirozu, K.; Tateishi, Y.; Wakita, H. AdV. Quantum
Chem. 2003, 42, 407.
Acknowledgment. This project was supported by grants from
NSF (Chemical Sciences-EMSI program) and from the BES,
DOE (Geosciences). M. B. H. also acknowledges financial
support from the EPA STAR and NSF graduate research
fellowships. The Advanced Light Source is supported by the
Director, Office of Science, Office of Basic Energy Sciences,
of the US Department of Energy under Contract No. DE-AC02-
05CH11231. We thank Tolek Tyliszczak, Hendrik Bluhm, Mary
Gilles, and David Shuh at Beamline 11.0.2, Advanced Light
Source, for assistance with experimental design and setup. We
also thank Kate Campbell, Alessandra Leri, and two anonymous
reviewers for helpful comments on the manuscript.
Supporting Information Available: Stability constants used
in speciation calculations, concentration versus pH diagrams for
the EDTA and Al-EDTA systems, and pH titration results and
speciation diagram for the Al-salicylate system. This informa-
tion is available free of charge via the Internet at http://
pubs.acs.org.
(43) Hay, M. B. Advances in the Molecular-Scale Understanding of
Geochemical Processes: Carboxyl Structures in Natural Organic Matter,
Organosulfur Cycling in Soils, and the Coordination Chemistry of Aqueous
Aluminum. Ph.D. Dissertation, Princeton University: 2007.
(44) Hay, M. B.; Myneni, S. C. B. J. Phys. Chem. A 2008, 112, 10595.
(45) Matsuo, S.; Wakita, H. AdV. Quantum Chem. 2008, 54, 193.
(46) Stevenson, F. J.; Cole, M. A. Cycles of Soil: Carbon, Nitrogen,
Phosphorus, Sulfur, Micronutrients, 2nd ed.; John Wiley & Sons: New York,
1999.
References and Notes
(1) Baes, C. F.; Mesmer, R. E. The Hydrolysis of Cations; John Wiley
& Sons: New York, 1976.
(2) Neal, J. J. Hydrology 1995, 164, 39.
(3) Driscoll, C. T.; Postek, K. M. In The EnVironmental Chemistry of
Aluminum, 2nd ed.; Sposito, G., Ed.; CRC Press: Boca Raton, FL, 1996;
Ch 9.
(47) Powell, P. E.; Cline, G. R.; Reid, C. P. P.; Szanislo, P. J. Nature
1980, 287, 833.
(4) Vance, G. F.; Stevenson, F. J.; Sikora, F. J. In The EnVironmental
Chemistry of Aluminum, 2nd ed.; Sposito, G., Ed.; CRC Press: Boca Raton,
FL, 1996; Ch 5.
´
(48) Farkas, E.; Enyedy, E. A.; Cso´ka, H. Polyhedron 1999, 18, 2391.
(49) Edwards, D. C.; Myneni, S. C. B. J. Phys. Chem. A. 2006, 110,
(5) Fein, J. Geology 1991, 19, 1037.
11809.
(6) Wilson, M. J. Clay Minerals 2004, 39, 233.
(7) Burrows, W. D.; Hem, J. D. Crit. ReV. EnViron. Control 1977, 7,
167.
(50) Ressler, T. J. Synchrotron Rad. 1998, 5, 118.
(51) Hermann, K.; Pettersson, L. G. M.; Casida, M. E.; Daul, C.;
Goursot, A.; Koester, A.; Proynov, E.; St-Amant, A.; Salahub, D. R.;
Carravetta, V.; Duarte, H.; Godbout, N.; Guan, J.; Jamorski, C.; Leboeuf,
M.; Malkin, V.; Malkina, O.; Nyberg, M.; Pedocchi, L.; Sim, F.; Triguero,
L.; Vela, A. StoBe-deMon Version 1.0; StoBe Software: 2002.
(52) Becke, A. D. Phys. ReV. A 1988, 38, 3098.
(8) Driscoll, C. T.; Baker, J. P.; Bisogni, J. J.; Schofield, C. L. Nature
1980, 284, 161.
(9) Kinraide, T. B. J. Exp. Bot. 1997, 48, 1115.
(10) Gensemer, R. W.; Playle, R. C. Crit. ReV. EnViron. Sci. Technol.
1999, 29, 315.
(53) Perdew, J. P. Phys. ReV. B 1986, 34, 7406.
(11) Bi, S.; Yang, X.; Zhang, F.; Wang, X.; Zou, G. Fresenius J. Anal.
Chem 2001, 370, 984.
(54) Triguero, L.; Pettersson, L. G. M.; Ågren, H. Phys. ReV. B 1998,
ˇ
58, 8097.
(12) Se`ane`ar, J.; Milae`ie`, R. Anal. Bioanal. Chem. 2006, 386, 999.
¨
(55) Slater, J. C.; Johnson, K. H. Phys. ReV. B 1972, 5, 844.
(56) Sto¨hr, J. NEXAFS Spectroscopy; Springer-Verlag: Berlin, 1992.
(57) Nilsson, A.; Ogasawara, H.; Cavalleri, M.; Nordlund, D.; Nyberg,
M.; Wernet, P.; Pettersson, L. G. M. J. Chem. Phys. 2005, 122, 154505.
(58) Cavalleri, M.; Na¨slund, L.-Å.; Edwards, D. C.; Wernet, P.;
Ogasawara, H.; Myneni, S.; Ojama¨e, L.; Odelius, M.; Nilsson, A.; Pettersson,
L. G. M. J. Chem. Phys. 2006, 124, 194508.
(59) Smith, J. D.; Cappa, C. D.; Messer, B. M.; Drisdell, W. S.; Cohen,
R. C.; Saykally, R. J. J. Phys. Chem. B 2006, 110, 20038.
(60) Kolczewski, C.; Pu¨ttner, R.; Plashkevych, O.; Ågren, H.; Staemmler,
V.; Martins, M.; Snell, G.; Schlachter, A. S.; Sant’Anna, M.; Kaindl, G.;
Pettersson, L. G. M. J. Chem. Phys. 2001, 115, 6426.
(13) Persson, P.; Karlsson, M.; Ohman, L.-O. Geochim. Cosmochim.
Acta 1998, 62, 3657.
(14) Loring, J. S.; Karlsson, M.; Fawcett, W. R.; Casey, W. H. Geochim.
Cosmochim. Acta 2000, 64, 4115.
¨
(15) Clause´n, M.; Ohman, L.-O.; Axe, K.; Persson, P. J. Mol. Struct.
2003, 648, 225.
(16) Trout, C. C.; Kubicki, J. D. J. Phys. Chem. A 2004, 108, 11580.
(17) Elkins, K. M.; Nelson, D. J. Coord. Chem. ReV. 2002, 228, 205.
(18) Brooker, M. H.; Tremaine, P. R. Geochim. Cosmochim. Acta 1992,
56, 2573.
(19) Rudolph, W. W.; Mason, R.; Pye, C. C. Phys. Chem. Chem. Phys.
2000, 2, 5030.

6148 J. Phys. Chem. A, Vol. 114, No. 20, 2010
Hay and Myneni
(61) Edwards, D. C.; Myneni, S. C. B. J. Phys. Chem. A 2005, 109,
10249.
(62) Messer, B. M.; Cappa, C. D.; Smith, J. D.; Drisdell, W. S.;
Schwartz, C. P.; Cohen, R. C.; Saykally, R. J. J. Phys. Chem. B 2005, 109,
21640.
(63) Wilks, R. G.; MacNaughton, J. B.; Kraatz, H.-B.; Regier, T.;
Moewes, A. J. Phys. Chem. B 2006, 110, 5955.
(64) Wilks, R. G.; MacNaughton, J. B.; Kraatz, H.-B.; Regier, T.; Blyth,
R. I. R.; Moewes, A. J. Phys. Chem. A 2009, 113, 5360.
(65) Besley, N. A.; Peach, M. J. G.; Tozer, D. J. Phys. Chem. Chem.
Phys. 2009, 11, 10350.
(77) Ildefonse, P.; Cabaret, D.; Sainctavit, P.; Calas, G.; Flank, A. M.;
Lagarde, P. Phys. Chem. Miner. 1998, 25, 112.
(78) Edwards, D. C.; Nielsen, S. B.; Jarzecki, A. A.; Spiro, T. G.;
Myneni, S. C. B. Geochim. Cosmochim. Acta 2005, 69, 3237.
(79) Domagal-Goldman, S. D.; Paul, K. W.; Sparks, D. L.; Kubicki,
J. D. Geochim. Cosmochim. Acta 2009, 73, 1.
(80) Edwards, D. C., Wesleyan College, personal communication, 2008.
(81) Hovey, J. K.; Tremaine, P. R. J. Phys. Chem. 1985, 89, 5541.
(82) Jung, W. S.; Chung, Y. K.; Shin, D. M.; Kim, S. D. Bull. Chem.
Soc. Jpn. 2002, 75, 1263.
(83) Coskuner, O.; Jarvis, E. A. A. J. Phys. Chem. A 2008, 112, 2628.
(84) Deacon, G. B.; Phillips, R. J. Coord. Chem. ReV. 1980, 33, 227.
(85) Tackett, J. E. Appl. Spectrosc. 1989, 43, 483.
(86) Nakamoto, K. Infrared and Raman Spectra of Inorganic and
Coordination Compounds, Part B: Applications in Coordination, Organo-
metallic, and Bioinorganic Chemistry; John Wiley: New York, 1997.
(87) Kula, R. J.; Finley, C. M.; Sawyer, D. T.; Chan, S. I. J. Am. Chem.
Soc. 1963, 85, 2930.
(88) Nakamoto, K.; Morimoto, Y.; Martell, A. E. J. Am. Chem. Soc.
1963, 85, 309.
(89) Sawyer, D. T.; Tackett, J. E. J. Am. Chem. Soc. 1963, 85, 314.
(90) Hug, S. J.; Bahnemann, D. J. Electron Spectrosc. Relat. Phenom.
2006, 150, 208.
(91) Quile`s, F.; Burneau, A.; Gross, N. Appl. Spectrosc. 1999, 53, 1061.
(92) Strathmann, T. J.; Myneni, S. C. B. Geochim. Cosmochim. Acta
2004, 68, 3441.
(93) McConnell, A. A.; Nuttall, R. H.; Stalker, D. M. Talanta 1978,
25, 425.
(66) De Francesco, R.; Stener, M.; Fronzoni, G. J. Phys. Chem. C 2007,
111, 13554.
(67) Rehr, J. J. Radiat. Phys. Chem. 2006, 75, 1547.
(68) Schaftenaar, G.; Noordik, J. H. J. Comput. Aided Mol. Des. 2000,
14, 123.
(69) Kutzelnigg, W.; Fleischer, U.; Schindler, M. In NMR-Basic
Principles and Progress; Diehl, P., Ed.; Springer: New York, 1990; Vol.
23, p 165.
(70) Godbout, N.; Salahub, D. R.; Andzelm, J.; Wimmer, E. Can.
J. Chem. 1992, 70, 560.
(71) Farkas, E.; Kozma, E.; Kiss, T.; To´th, I.; Kurzak, B. J. Chem. Soc.,
Dalton Trans. 1995, 477.
(72) Venturini-Soriano, M.; Berthon, G. J. Inorg. Biochem. 2001, 85,
143.
(73) Martell, A. E.; Smith, R. M. Database 46: NIST Critically Selected
Stability Constants of Metal Complexes, Version 7 [Electronic Resource];
NIST Standard Reference Data; NIST:: Gaithersburg, MD, 1995.
(74) Gustafsson, J. P. Visual MINTEQ Version 2.50.; Stockholm,
Sweden, October2006; http://www.lwr.kth.se/English/OurSoftware/vminteq/
index.htm.
(75) Allison, J. D.; Brown, D. S.; Novo-Gradac, J. K. MINTEQA2/
PRODEFA2, A Geochemical Assessment Model for EnVironmental Systems:
User Manual Supplement for Version 4.0; US Environmental Protection
Agency: Washington, DC, 1998.
(94) Kubicki, J. D.; Sykes, D.; Apitz, S. E. J. Phys. Chem. A 1999,
103, 903.
(95) Aquino, A. J. A.; Tunega, D.; Haberhauer, G.; Gerzabek, M.;
Lischka, H. Phys. Chem. Chem. Phys. 2000, 2, 2845.
(96) Rakotonarivo, E.; Tondre, C.; Bottero, J. Y.; Mallevialle, J. Water
Res. 1989, 23, 1137.
(97) Biber, M. V.; Stumm, W. EnViron. Sci. Technol. 1994, 28, 763.
¨
(98) Ohman, L.-O.; Sjo¨berg, S. Acta Chem. Scand. A 1983, 37, 875.
(76) Fro¨ba, M.; Wong, J.; Rowen, M.; Brown, G. E.; Tanaka, T.; Rek,
Z. Physica B 1995, 209, 555.
JP909656Q