Article

pubs.acs.org/est

Early Stage Formation of Iron Oxyhydroxides during Neutralization

of Simulated Acid Mine Drainage Solutions
Mengqiang Zhu,†,* Benjamin Legg,‡ Hengzhong Zhang,‡ Benjamin Gilbert,† Yang Ren,§
Jillian F. Banfield,†,‡ and Glenn A. Waychunas†
†
Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States
‡
Department of Earth and Planetary Science, University of California-Berkeley, Berkeley, California, 94720, United States
§
Advanced Photon Source, Argonne National Laboratory, Argonne, Illinois 60439, United States
*
S Supporting Information

ABSTRACT: The phases and stability of ferric iron products

formed early during neutralization of acid mine drainage
waters remain largely unknown. In this work, we used in situ
and time-resolved quick-scanning X-ray absorption spectros-
copy and X-ray diffraction to study products formed between 4
min and 1 h after ferric iron sulfate solutions were partially
neutralized by addition of NaHCO3 ([HCO3−]/[Fe3+] < 3).
When [HCO3−]/[Fe3+] = 0.5 and 0.6 (initial pH ∼ 2.1 and
2.2, respectively), the only large species formed were sulfate-
complexed ferrihydrite-like molecular clusters that were stable
throughout the duration of the experiment. When [HCO3−]/[Fe3+] = 1 (initial pH ∼ 2.5), ferrihydrite-like molecular clusters
formed initially, but most later converted to schwertmannite. In contrast, when [HCO3−]/[Fe3+] = 2 (initial pH ∼ 2.7),
schwertmannite and larger ferrihydrite particles formed immediately upon neutralization. However, the ferrihydrite particles
subsequently converted to schwertmannite. The schwertmannite particles formed under both conditions aggregated extensively
with increasing time. This work provides new insight into the formation, stability and reactivity of some early products that may
form during the neutralization of natural acid mine drainage.

■ INTRODUCTION
When exposed to air and water, metal sulfide deposits and coal
mines produce iron-rich sulfuric acid solutions that can contain
a broad range of toxic elements.1,2 These solutions, referred to
as acid mine drainage (AMD), degrade environmental quality.
Neutralization of AMD occurs when solutions mix with natural
waters, such as natural rivers and streams, during wetland
treatment,3 through reaction with rock surfaces, or following
lime addition in remediation processes. As a result, ferric iron
(Fe3+) hydrolyzes and precipitates as Fe oxyhydroxide
nanoparticles such as ferrihydrite and schwertmannite. These
Fe oxyhydroxide nanoparticles can transform to more
thermodynamically stable phases goethite and hematite with
time.
During the formation of Fe oxyhydroxide nanoparticles, toxic
elements in solution can be incorporated into the nanoparticle
structure. During or after formation, contaminants can also be
adsorbed onto nanoparticle surfaces or trapped in interfacial
regions between aggregated nanoparticles. Such processes
influence contaminant fate and transport.4 Nanoparticles also
can interact with microorganisms and their extracellular
polymers,5 and hence extracellular substances, and other
solution species may promote or retard particle growth,
transformation and aggregation. These processes can control
Fe oxyhydroxide phase stability and their behavior in the
environment.
It is generally believed that solvated Fe3+ undergoes
hydrolysis with subsequent formation of low-molecular-weight
Fe3+ species (hydrated and hydrolyzed monomers, dimers,
trimers, etc.).6 Fe oxyhydroxide particles may nucleate from the
polynuclear species6 and grow via atom-by-atom addition or
oriented aggregation.7 In addition to hydrolysis, olation and
oxolation processes are responsible for these condensation
processes.8 The resultant Fe oxyhydroxide phases depend on
anion type and pH. The impact of anions in phase organization
originates from their Fe3+ binding ability, and this is strongly
affected by pH. In general, weak Fe3+ binding anions, such as
NO3− and ClO4−, favor ferrihydrite formation. Stronger Fe3+-
binding ligands, such as Cl− and SO42‑, lead to formation of
akaganeite and schwertmannite, respectively, under acidic
conditions. Under neutral and alkaline pH conditions,
ferrihydrite tends to be the resultant phase because OH−
outcompetes Cl− and SO42‑ to bind Fe3+. In the presence of
very strong Fe3+-binding ligands, such as phosphate and silicate,
Received: March 30, 2012
Revised: June 25, 2012
Accepted: July 5, 2012
Published: July 5, 2012

© 2012 American Chemical Society 8140 dx.doi.org/10.1021/es301268g | Environ. Sci. Technol. 2012, 46, 8140−8147
Environmental Science & Technology
Fe oxyhydroxide particle formation can be substantially
impaired.9,10
Aqueous speciation in pure Fe3+ solutions is relatively well
understood. In recent years, much has been learned about
nanoparticle structure, growth, and phase transformation
behavior.11 However, a major gap in knowledge concerns the
nucleation pathway whereby aqueous monomeric Fe3+ species
are converted into molecular clusters and nanoparticles. Prior
work attempted to study early stage Fe oxyhydroxide formation
kinetics and mechanisms in partially neutralized Fe3+ solutions
using various methods, including titration, UV-vis spectroscopy,
wide-angle X-ray scattering diffraction (WAXS), small-angle X-
ray scattering (SAXS), extended X-ray absorption fine structure
(EXAFS) spectroscopy, electron microscopy, and analytical
centrifugation.6,12,13 Effects of anion type, ratio of neutralization
base to Fe3+, Fe3+ concentration, aging period, and temperature,
were examined. However limited knowledge was obtained
about the precipitate phase, average local atomic structure,
particle size and morphology, and aggregate structure. In
particular, due to the limited time-resolution of the
experimental methods, these investigations were confined to
late reaction stages (>30 min) when particle aggregation and
phase transformation comprise the major events. Although the
final precipitate phases could be identified in these studies, the
forms and structures of the early formed products in the
neutralization process remained largely unknown, and hence
their formation kinetics and mechanism are yet to be explored.
In the current study, we were able to investigate the formation
kinetics and phase identities of early formed products in
partially neutralized Fe3+-sulfate solutions using time-resolved
UV-vis spectroscopy, quick-scanning EXAFS (QEXAFS) spec-
troscopy, and synchrotron XRD, by combining these rapid
methods with a slowed neutralization process.
■ EXPERIMENTAL SECTION
Sample Preparation. All chemicals used were of analytical
grade. A 0.8 M Fe3+ stock solution ([SO42‑]/[Fe3+] = 1.5) was
prepared by adding 4.16 g of ferric iron sulfate powder
(containing 21−23% Fe by weight) to 16 mL deionized (DI)
water. The Fe solution was used within 48 h after preparation,
although there were no visible changes in color and turbidity
over 1 month. Reactions were initiated by slowly pumping 4
mL DI water, 0.2 M, 0.24 M, 0.4 M, or 0.8 M NaHCO3
solution at 5 mL·min−1 using a syringe pump, into a 30 mL
plastic bottle containing 4 mL of 0.4 M Fe3+ sulfate solution (2
mL 0.8 M Fe stock solution plus 2 mL DI water) to achieve
[HCO3−]/[Fe3+] molar ratios (represented by h) of 0, 0.5, 0.6,
1, and 2. An acidified Fe solution was prepared by adding 4 mL
of concentrated HNO3 to 4 mL of 0.4 M Fe3+ sulfate solution,
referred to as FeAcid. This solution is supposed to contain
mainly Fe(H2O)63+. The final concentration of total Fe3+ in all
solutions was 0.2 M. Vigorous stirring was enforced during the
base addition to minimize local oversaturation with respect to
Fe oxyhydroxide phases. Each base addition step took 48 s.
Upon completion, one additional minute of stirring was used to
remove CO2 gas resulting from NaHCO3 decomposition
during the neutralization reaction. Then, the solutions were
immediately transferred for the following time-resolved
analyses. Measurements of reaction time started from the first
drop of base addition. When the first data sets were recorded,
about 2−5 minutes had elapsed, depending on time-resolved
analytical approaches (see below). All experiments were
conducted at room temperature of 24−25 °C. Schwertmannite
8141
Article
and two-line ferrihydrite reference materials were synthesized
using the procedures described by Schwertmann and Cornell.14
UV-Vis Spectroscopy and pH Measurement. Each of
the mixed solutions was immediately transferred to a quartz
cuvette with 1 mm path length for time-resolved UV-vis spectra
collection using an Agilent UV-vis spectrophotometer (model
8453). UV-vis spectra were continuously recorded with 5 s of
integration time. Time-dependent pH values of the above
solutions were automatically recorded every 20 s. The pH
meter was calibrated using pH 1, 1.68, 2, and 4 buffer solutions.
Synchrotron X-ray Diffraction (SXRD). Time-resolved
SXRD 2-D images were collected at beamline 11-ID-B at the
Advanced Photon Source (APS). The high flux of the X-rays of
this beamline is the key for collecting high signal-to-noise ratio
XRD data from particles in dilute particle suspensions. A
detailed description for XRD collection and processing is
provided in the Supporting Information (SI-1).
QEXAFS Spectroscopy. Each of the mixed solutions was
transferred to a 2 mm thick, 15 mm long, and 10 mm wide cell
made from acrylic plastic plate. Both sides of the cell were
sealed with Kapton film to hold the solutions. QEXAFS spectra
were collected at one scan per 2 s in transmission mode. Fifteen
or 14 continuous scans of QEXAFS spectra were averaged to
improve counting statistics, resulting in a time-resolution of 30
s. The measurements were conducted at beamline X18A at the
National Synchrotron Light Source (NSLS), using a specially
equipped monochromator with a Si (111) crystal. The
monochromator was detuned by 35% with respect to I0 to
minimize higher order harmonic X-rays. The detailed beamline
and monochromator setup for the QEXAFS measurement can
be found elsewhere.15
Powder samples, including the Fe oxyhydroxide references
and the solid precipitates in the mixed solutions at h = 1 and 2,
were ground finely and uniformly spread on Scotch tape. The
tapes were folded to four or eight layers for EXAFS data
collection in transmission mode at beamline 4-1 at the Stanford
Synchrotron Radiation Lightsource (SSRL), using a Si (220)
monochromator crystal.
After background removal and normalization the EXAFS
spectra were converted into the k-weighted function, k3χ(k),
and Fourier transforms |χ(R)| were calculated over the k range
of 3−14 Å−1 using the Bessel-Kaiser window function. Linear
combination fitting (LCF) analysis of QEXAFS spectra was
performed over 3−12 Å−1 in k space. In the fits, the component
weight sum was forced to equal 1.0. The program Athena16 was
used for all of the above data reduction and processing. Further
EXAFS shell-by-shell fitting was performed using the SixPack
program suite17 to obtain the detailed local atomic environment
of iron atoms for representative samples.
Ferron Assay. Ferron (8-hydroxy-7-iodo-5-quinoline sul-
fonic acid) has been used for semiquantification of Fe3+
monomers, polymers and precipitates based on distinct
ferron−Fe complex formation rates. Fe3+ monomers are
complexed by ferron rapidly, but the breakdown of larger
Fe3+ species to form ferron−Fe complexes occurs much more
slowly.18 It has been assumed that reactions of monomeric Fe3+
species with ferron are complete within 1 min (denoted as
Fe_monomer).18 Species that form ferron−Fe complexes over
the next 12 h were interpreted as Fe polymers, and any Fe
species that remained unreactive after 12 h were considered as
stable precipitates.18 Following this method, we used the ferron
assay to estimate Fe3+ precipitation reaction progress during
neutralization of ferric salts by NaHCO3, based on Fe_mo-
dx.doi.org/10.1021/es301268g | Environ. Sci. Technol. 2012, 46, 8140−8147
Environmental Science & Technology
nomer fractions measured at different times as the neutraliza-
tion reaction proceeded. The specific experimental procedures
were described in the Supporting Information (SI-2). The
accuracy of ferron assay-based inferences was re-evaluated in
light of other results of this study (see below).
■ RESULTS
Color and Turbidity. The visible appearance of the
solutions immediately following bicarbonate addition was
dependent on the base to ferric iron molar ratio (h =
[HCO3−]/[Fe3+]). All h ≥ 0.5 solutions displayed red wine
color, and the color deepened as h increased. Moreover, the
cloudiness (or turbidity) of the h = 1 and h = 2 solutions
increased over time, whereas other solutions remained clear
during the experimental period. To monitor the turbidity
changes with time, the measured absorbance at 600 nm was
used as a turbidity indicator, as no samples had UV-vis
absorption bands at this wavelength (see below). Assuming that
Rayleigh scattering applies, increased absorbance was due to
increase of size and/or numbers of scattering objects.19 As
shown in Figure 1, absorbance at 600 nm increased with
Figure 1. UV-vis absorbance at 600 nm versus reaction time for Fe
sulfate solutions at various h values (h = [NaHCO3]/[Fe3+], i.e.,
neutralization ratio). Since no significant UV-vis absorption bands exist
at this wavelength, the measured absorbance was due to particle
scattering, indicating turbidity of the solutions.
increasing h when reaction time < 27 min, indicating particle
formation. For the h = 1 solution, the 600 nm absorbance
increased in the first ∼10 min (see inset) and dramatically after
∼26 min. For the h = 2 solution, an initially subtle decrease was
followed by a dramatic increase in scattering after ∼28 min
(Figure 1). The dramatic increases correlated with the
transitions from clear red wine-like solutions to cloudy yellow
suspensions. The cloudiness was due to the massive formation
of large particle aggregates.19 The aggregates then settled to
form precipitates that are referred to as Ppt.h1 and Ppt.h2 for h
= 1 and h = 2, respectively.
Fe3+ Molecular Clusters and Particles Determined by
UV-vis Spectroscopy. Fe3+ monomers, molecular clusters and
particles have different optical excitation features that can be
used for identification. Ligand-field or d → d transitions are
spin forbidden and thus are extremely weak in Fe 3+
monomers.20,21 By contrast, the ligand-field bands of Fe3+
clusters and particles are intense mainly due to antiferromag-
netic coupling between next neighboring Fe atoms.11,21 In
8142
Article
addition, ferric Fe species with an extended structure, such as
Fe oxyhydroxide particles, have a characteristic absorption
feature called electron pair transition (EPT) band.22 The EPT
band is significant for particles but very weak for molecular
clusters, and thus can be used to differentiate the two types of
Fe species.
Figure 2a−d shows UV-vis spectra of the solutions at
different h. For clarity, the spectra were presented in two
different wavelength ranges. The spectra of h = 1 and h = 2
solutions evolved with time whereas others remained the same.
The acidified solution (FeAcid) only had a very weak band at
∼806 nm (Figure 2b). The h = 0 solution had two broad
absorption bands centered at approximately 425 and 830 nm.
For the h = 0.5 and 0.6 solutions, three bands at about 430,
448, and 850 nm can be discerned (Figure 2b and d), and the h
= 0.6 solution bands are more intense. The barely changed
spectra of the above solutions indicate that Fe species in these
solutions were stable within the experimental time period. The
first spectrum of the h = 1 solution also showed the same three
bands as the h = 0.5 solution (Figure 2a and b). However, the
first two bands gradually disappeared with reaction time, and
the 850 nm band decreased its intensity and shifted its position
to 862 nm. Another band at 485 nm, not obvious in the first
spectrum of the h = 1 solution, emerged and became
increasingly pronounced (Figure 2b). This band did not exist
in the spectra of the h = 0, 0.5, and 0.6 solutions.
The observed bands at 425, 430, and 448 nm can be assigned
to 6A1 g → (4A1, 4Eg) transitions and the bands at 830, 850, and
862 nm to 6A1 g→ 4T1 g transitions.11,21 The substantial
existence of all these bands indicates formation of Fe clusters
or particles.21,23 The 485 nm band is assigned to EPT.22 The
absence of this band in the h = 0, 0.5, 0.6 and the initial h = 1
solutions suggests that the Fe species were molecular clusters.
The disappearance of the three bands of the h = 1 solution
shows that the Fe cluster was decreasing in abundance, whereas
the increasing EPT absorbance indicates that a Fe oxyhydroxide
phase was forming and increasing in abundance. The UV-is
spectrum of the filtered h = 1 solution after 2 h of reaction time
still showed the three bands, although they were much less
intense (Figure 2b and d), indicating that the Fe clusters had
not completely disappeared.
Regarding the h = 2 solution, the EPT band at 485 nm is
pronounced in the first collected spectrum, indicating that Fe
oxyhydroxide particles formed from the beginning. A band at
865 nm (assigned to 6A1 g→ 4T1 g) decreased in intensity with
time without significant position change. In addition, the
absorbance between 525 and 600 nm decreased with time,
suggesting disappearance of a metastable phase. The filtered h =
2 solution after 2 h had a weak peak at 850 nm besides the 485
nm band (Figure 2b and d), indicating the solution also had a
small amount of Fe clusters similar to those present in the h = 1
solutions. The lack of obvious 430 and 448 nm bands in this
case could be due to the overlap with the intense absorbance
tail (300−600 nm) from the larger particles.
Phase Analysis by Synchrotron XRD (SXRD). To
identify the phases of particles being formed, in situ SXRD
data were collected from the neutralized solutions. As shown in
Supporting Information Figure SI-1, the first XRD pattern of
the h = 1 solution cannot be firmly attributed to known phases
due to the low particle abundance and/or small particle size.
The first pattern of the h = 2 solution presented broad peaks,
indicating formation of very fine crystalline particles. With
increasing reaction time, the initial broad peaks of both h = 1
dx.doi.org/10.1021/es301268g | Environ. Sci. Technol. 2012, 46, 8140−8147
Environmental Science & Technology Article

Figure 2. UV-vis spectra of Fe sulfate solutions at various h values. The spectra of the h = 1 and 2 solutions were time dependent and the arrows
indicate the reaction directions with increased reaction time (2 min < t < 27 min). The “filtered” spectra were collected after filtration of the cloudy
solutions to remove solids. For closer inspection, the UV-vis spectra were presented in two different wavelength regimes (note that the absorbance
scales differ). The spectral amplitude of the h = 1 and 2 solutions increased over the whole wavelength range due to particle scattering, correlating
with the transition from clear to cloudy solutions due to onset of precipitation. The spectra collected after the solutions became cloudy (top black
curves in panels (a) and (c)) are not shown in (b) and (d).

Figure 3. Fe K-edge EXAFS spectra of HNO3 acidified Fe sulfate solution (FeAcid) (brown), the neutralized Fe sulfate solutions at different reaction
times and different h values (pink, green, cyan, blue), the final precipitates at h = 1 and 2 (red), with 2-line ferrihydrite (Fhy2L) and schwertmannite
(Schw) reference spectra. (a) k-weighted χ(k) and (b) the magnitude (|χ(R)|) of their Fourier transforms.

and h = 2 solutions sharpened (2.4 Å−1) and split (4 Å−1 and
5.5 Å−1), increasingly resembling those of the schwertmannite
reference phase. The SXRD results showed that the precipitates
(Ppt.h1 and Ppt.h2) formed in the two solutions were primarily
schwertmannite (Supporting Information Figure SI-1). How-
ever, XRD phase characterization for particles in solutions
based on poorly defined peaks has large uncertainties due to
solution interference. In the following two sections, QEXAFS
spectroscopic analyses, including EXAFS curve fitting and linear
8143
combination fitting (LCF), were used to further identify the
particle phases and quantify their amounts versus reaction time.
Local Atomic Environment of Fe. Fe K-edge EXAFS
spectroscopy provides information about the local atomic
environment around central Fe (within 6 Å). Figure 3 displays
the results of in situ QEXAFS spectroscopy acquired from
samples under the same conditions as the XRD data of
Supporting Information Figure SI-1. The peaks at 1.5−1.6 Å (R
+ΔR) in |χ(R)| are assigned to the nearest O shell at a Fe−O
distance of ∼2 Å, corresponding to Fe−O bond lengths. As the
dx.doi.org/10.1021/es301268g | Environ. Sci. Technol. 2012, 46, 8140−8147
Environmental Science & Technology
Figure 4. Molar fractions of ferrihydrite (red circles), schwertmanite (blue triangles), and h = 0.5 solutions (green triangles) were obtained from
linear combination fitting for (a) h = 1 and (b) h = 2. Values measured via the ferron assay are indicated by black boxes. Comparison of the inferred
molar fraction of ferrihydrite plus the h = 0.5 solution (open boxes) to the ferron assay values indicates that the ferron assay actually measures both
of these components at h = 1.
neutralization degree increased from the acidified solution to
the h = 2 solutions, the Fe−O bond length shortened, the O
peaks broadened, and the amplitudes decreased (Figure 3). The
bond length shortening can be due to the formation of particles
and clusters that usually have shorter Fe−O bonds in Fe−OH
and Fe−OFe moieties compared to those in Fe−OH2 of
monomeric Fe species. These changes also widen the overall
Fe−O bond length distribution and cause destructive
interference in the χ(k) functions and accordingly the reduced
O peak amplitudes in their Fourier transforms.24 The wider
Fe−O bond distribution is consistent with the increasing
Debye−Waller factors (σ2) obtained from fitting using a single
O shell (Supporting Information SI-3).
The peaks at ∼2.6 and ∼3.05 Å (R+ΔR)| are assigned to the
edge- and corner-sharing Fe shells at the Fe−Fe distances of
∼3.0 and ∼3.35 Å, respectively. Sulfate can complex with
Fe25,26 and sulfur scattering may slightly modify the peaks
corresponding to approximately these distances. As the
neutralization degree increases, the two Fe peaks become
more and more pronounced, indicating Fe polymerization.
FeAcid did not exhibit the edge-sharing Fe peak (Figure 3),
showing that the high concentration of acid suppressed
formation of Fe polymers. The pronounced peaks between
2.8 and 3.75 Å (R+ΔR) may result from both multiple-
scattering paths within the first O shell of Fe octahedra and
single-scattering paths from the O atoms in the second
hydration shell.27 The peak at 3.1 Å (R+ΔR) overlaps with
the corner-sharing Fe peak, making it hard to discern the Fe
shell when particle abundance is low, such as the spectra of the
h < 1 solutions and the initial spectrum of the h = 1 solution.
For the h = 1 and 2 solutions, with increasing reaction time,
the O shells did not change significantly, but the two Fe shells
underwent substantial increases in peak amplitude and shifts in
peak positions (representative spectra in Figure 3 and the
complete set in Supporting Information SI-4). As revealed by
further analysis described below, these changes are mainly due
to particle phase transformation.
Time-Dependent Evolution of Fe3+ Species. EXAFS
linear combination fitting (LCF) analysis can be used to
identify and quantify each species in a system containing mixed
Fe phases.28 We anticipate that the neutralized solutions
contain many different Fe species, and complexation by sulfate
8144
Article
introduces further diversity.25,26 As it is impossible to identify
and prepare LCF standards for all possible Fe species in the
neutralized solutions, we assume that the h = 0.6, 1, and 2
solutions can be represented by a combination of solutions at
lower neutralization ratios (representing monomeric Fe3+ and
small Fe3+ clusters) and Fe oxyhydroxide solid phases
approximating particles in aqueous solutions. FeAcid, the h =
0, 0.2, and 0.5 solutions, schwertmannite, Fhy2L, goethite,
lepidocrocite, hematite, Ppt.h1, Ppt.h2, and their combinations,
thus were tested with LCF analysis to determine if well-defined
general fits could be obtained with a minimum set of
components. It was found that a combination of the h = 0.5
solution, Ppt.h2 and Fhy2L gave the best fits with the least
number of components (fits in Supporting Information SI-6
and results in Figure 4). Use of Ppt.h2 and h = 0.5 solution only
in the fits decreased the goodness of fit (in terms of reduced χ2)
up to 33% (the first spectrum of the h = 1 solution), showing
necessity of the Fhy2L component.
The LCF analysis indicates that Ppt.h1 contained Fhy2L at a
phase fraction 17% and Ppt.h2 at 83%. The h = 0.6 solution
contained Fhy2L at 4% and the h = 0.5 solution at 96%, but this
was subject to large uncertainties due to the low abundance of
Fhy2L. Although the EXAFS spectrum of Ppt.h2 was almost
identical to that of schwertmannite reference phase (Supporting
Information SI-5), replacing Ppt.h2 with the schwertmannite
reference slightly decreased the fitting accuracies but gave
similar component fraction values. Henceforth, for clarity, we
use “schwertmannite” to represent “Ppt.h2” in further
discussion of the LCF results that were summarized in Figure 4.
The Fhy2L component accounted for ∼20% of Fe species
initially in the h = 1 solution whereas the schwertmannite
quantity was negligible (Figure 4a). With time, the Fhy2L
component fraction gradually decreased to ∼5% whereas
schwertmannite developed to 15−20% (Figure 4a). About
80% of the Fe could be represented by the h = 0.5 solution and
this quantity stayed almost constant with time, indicating
formation of schwertmannite from conversion of Fhy2L. For
the h = 2 solution, both Fhy2L and schwertmannite accounted
for ∼30%, respectively, initially (Figure 4b). With time, the
Fhy2L component gradually disappeared and the schwert-
mannite fraction increased to ∼70%. The h = 0.5 solution
component decreased from 40% to 30% with time.
dx.doi.org/10.1021/es301268g | Environ. Sci. Technol. 2012, 46, 8140−8147
Environmental Science & Technology
Figure 5. (a) The fraction of Fe species (Fe_monomer) that reacts with ferron within 1 min, and (b) the change in pH over time for h = 0 to h = 2
solutions. The empty symbols in (a) are replicates measured after filtration to remove precipitates. The solid symbols indicate the fraction of Fe
species measured by ferron assay for unfiltered solutions.
As the two Fe peaks of Fhy2L have lower amplitude and
longer corner-sharing Fe−Fe distance than those of schwert-
mannite (Figure 3), the relative changes of the two Fe phases
with time resulted in increasing amplitude of the two Fe peaks
and shortening of the corner-sharing Fe distance (Supporting
Information SI-4). For the same reason, the Fe peaks of the h =
1 solution were not significant initially, though Fhy2L
accounted for 20%.
Reaction Progress Measured by Ferron Assay and pH
Changes. In addition to the above EXAFS LCF analysis, the
ferron assay was also used to determine reaction progress of
particle formation. Figure 5a shows the time-dependent change
in the proportion of Fe3+ species that react with the ferron
probe within one minute, assumed to represent monomeric
ferric iron. Initially, almost all Fe was in the Fe_monomer
fraction for the h = 1 solution, whereas less than half of the Fe
was in this fraction for the h = 2 solution (Figure 5b). In
addition, the ferron-Fe formation rates of h = 1 and 2 solutions
gradually decreased with the increasing neutralization reaction
time (Supporting Information Figure SI-2). In contrast, for the
h = 0, 0.5, and 0.6 solutions, ferron−Fe complex formation
reaction completed almost instantaneously (data not shown)
and the rates did not change with the neutralization reaction
time, indicating that the solution was dominated by the
Fe_monomer fraction. It is noteworthy that for the h = 2
solution, a second abrupt decrease in Fe_monomer occurred at
∼30 min when the solution became cloudy, possibly due to a
second nucleation event.
Fe3+ oxyhydroxide particle formation is a hydrolysis and
condensation process, releasing H+. Therefore, the pH change
over reaction time reflects particle formation progress. As
Figure 5b shows, the pH of the h = 1 and 2 solutions slowly
decreased with the reaction time, suggesting particle formation.
■
DISCUSSION
According to the UV-vis results, Fe molecular clusters formed
initially in the h = 1 solution but transformed to Fe
oxyhydroxide phases over time; in contrast, the initially formed
Fe species in the h = 2 solution were not molecular clusters but
Fe oxyhydroxide particles, some of which decreased and others
increased in abundance with time. The QEXAFS LCF analysis
shows that the Fhy2L component was the only phase formed
initially in the h = 1 solution, whereas for the h = 2 solution,
both Fhy2L and schwertmannite phases were present initially;
the Fhy2L components in both solutions transformed to
8145
Article
schwertmannite over time. Combining the above two analyses
indicates that the initially formed Fhy2L component, detected
as molecular clusters in the h = 1 solution and as particles in the
h = 2 solution, transformed to schwertmannite over time. UV-
vis and EXAFS results are also consistent with the SXRD data
that schwertmannite particles were present initially in the h = 2
solution and not in the h = 1 solution, but the SXRD analysis
failed to identify the ferrihydrite particles. The XRD peak
sharpening and splitting seen in both solutions were due to the
transformation of the Fhy2L particles and clusters, showing
poorly defined XRD peaks, to schwertmannite particles
showing better defined peaks. The clusters in the h = 1
solution were too small to contribute much to the turbidity
measured at 600 nm (I ∼ d6/λ4). Therefore, the observed
gradual turbidity increase of the h = 1 solution was mainly due
to the formation of schwertmannite particles.
Transformation of Ferrihydrite to Schwertmannite.
The transformation of the ferrihydrite clusters and particles to
schwertmannite indicates that schwertmannite was the
thermodynamically more stable phase in the h = 1 and h = 2
solutions. This is anticipated, given the presence of sulfate
anions. However, the transformation in the h = 1 solution was
not complete (5% remained). In addition, ferrihydrite clusters
existed in the h = 0.5 and 0.6 solutions without transforming to
schwertmannite within the studied time period. These
observations indicate that ferrihydrite clusters were more stable
than schwertmannite in the presence of sulfate under more
acidic conditions, although their abundances were low. Given
the large structural difference of the two phases, the
transformation likely occurred via dissolution and reprecipita-
tion.
The EXAFS LCF result shows that 70% of Fe existed as
schwertmannite eventually in the h = 2 solution, greater than
the sum (60%) of the initial ferrihydrite and schwertmannite
fractions. A second nucleation event occurred at 30 min in the
h = 2 solution, as suggested by the sudden drop of the dissolved
Fe concentration in the ferron assay (Figure 5a). Therefore, the
additional 10% schwertmannite resulted from the formation of
new nuclei rather than growth of the existing particles. This
indicates that schwertmannite can directly form from dissolved
Fe species, and that ferrihydrite is not an indispensible
precursor for its formation.
According to a previous study using an electric field jump
relaxation kinetic technique,29 hydrolysis reactions of mono-
meric Fe3+ are extremely fast and should be complete within
dx.doi.org/10.1021/es301268g | Environ. Sci. Technol. 2012, 46, 8140−8147
Environmental Science & Technology
seconds. The slow pH decreases of the h = 1 and 2 solutions
must be due to other slower chemical reactions. In this case, the
pH decrease can be attributed to the phase transformation, that
is schwertmannite formation from ferrihydrite clusters and
particles leads to a net release of protons. Phase transformation-
induced pH decreases have been observed previously in
ferrihydrite transformation to goethite, hematite, and lepidoc-
rocite.30
Invalidation of Ferron Assay for Fe Monomers. The
LCF-determined soluble Fe fraction (represented by the h =
0.5 solution) in the h = 2 solution is basically consistent with
the ferron assay result (Figure 4b). The large discrepancy
between the two measurements for the h = 1 solution (Figure
4a) is removed if the ferron assay actually measures both Fhy2L
(clusters) and the Fe species approximated by the h = 0.5
solution. The finding that the Fhy2L-like clusters reacted with
ferron within one minute invalidates the ferron assay for
quantification of Fe monomers. There is less evidence for
reaction of ferron with the Fhy2L particles in the h = 2 solution
within one minute, consistent with the conclusion that the main
component was Fhy2L particles. The decreased ferron−Fe
complex formation rates with the neutralization reaction time
showed that the developing schwertmannite particles were less
reactive than ferrihydrite clusters and particles regarding
ferron−Fe formation.
The Ferrihydrite Components. The Fhy2L-like cluster in
the h = 1 solution might resemble the Fe13 motif in the
ferrihydrite structure, that is, [FeO4Fe12(OH)24(H2O)12]7+.31
This molecular cluster was proposed by Bradley and Kydd32
decades ago based on infrared spectroscopy, although few
studies followed this suggestion. These authors found this
cluster was unstable, but could be stabilized by sulfate binding.
The structure of the aluminum Keggin cluster (Al13),
[AlO4Al12(OH)24(H2O)12]7+, was determined after crystalliza-
tion of the cluster from a solution as a sulfate salt.33 It is not
surprising that sulfate binding could stabilize similar Fe13
molecular clusters in solution, possibly by preventing their
growth into larger ferrihydrite particles and their dissolution by
acid as well. The sulfate binding becomes weaker at higher
pH.34 Therefore, when adding more base for the h = 2 case,
OH− could outcompete sulfate to bind with Fe3+ facilitating the
growth of ferrihydrite clusters to large particles. This is
consistent with the fact that ferrihydrite can be synthesized
by rapidly increasing Fe3+ sulfate solution pH to pH 7. On the
other hand, the weak sulfate binding at h = 2 was not able to
protect ferrihydrite from reaction, and accordingly ferrihydrite
converted completely as the pH was still very acidic (pH <
2.65).
The Fhy2L-like cluster was also present in the h = 0.5 and 0.6
solutions, according to UV-vis spectroscopy. Since the h = 0.5
solution was one of the three components used in the LCF
analyses of the h = 0.6, 1, and 2 solutions, the abundances of
the Fhy2L-like clusters in these solutions were underestimated.
This means that the clusters were also present in the h = 2
solution, consistent with the UV-vis spectrum of the filtered
solution, although the majority of the ferrihydrite units were
particles.
Abundance of the Clusters. Assuming the cluster
contains 13 Fe atoms, based on the LCF-determined fractions
and UV-vis absorbance differences at 850 nm, we estimated its
absorption coefficient (μ) as 44.5 mol−1·cm−1 using the h = 1
and h = 0.5 results and 52.0 mol−1·cm−1 using the h = 0.6 and h
= 0.5 results, with an average value of 48.3 ± 3.7 mol−1·cm−1.
8146
Article
Using this value the cluster concentrations can be estimated as
2.4 ± 0.2, 3.1 ± 0.2, and 5.3 ± 0.4 mM in h = 0.5, 0.6 and the
initial h = 1 solutions, respectively, according to their absolute
absorbance at 850 nm.
Environmental Implications. For the first time, the
identities and abundances of iron molecular clusters and
particles formed early in ferric iron solution neutralization have
been determined. The result provides new insights into the
precipitation process and demonstrates the necessity of using in
situ fast diffraction and spectroscopic measurements for analysis
of rapid ferric iron oxyhydroxide particle formation processes.
The concentrations of ferric iron and sulfate used in this study
are comparable to those in AMD solutions, and the particle
phases identified are commonly found in AMD-impacted
environments. In addition, the various neutralization ratios are
applicable for natural and engineered AMD neutralization
processes. Therefore, the formation and phase transformation
processes revealed in this study are likely relevant in a variety of
AMD environments, though are most directly related to those
with minimal ferrous iron and low concentration of heavy
metal(loid)s and dissolved organic matter. Given that AMD can
contain a wide variety of toxic elements that are attenuated by
adsorption onto particle surfaces or through coprecipitation
during neutralization, knowledge of early stage particle
formation and transformation processes contribute to an
improved understanding of attenuation mechanisms.
■
*
ASSOCIATED CONTENT
S Supporting Information
SXRD, the ferron assay procedures, all time-series QEXAFS
spectra, QEXAFS shell-by-shell fitting results and comparisons
of QEXAFS spectra with their linear combination fits. This
material is available free of charge via the Internet at http://
pubs.acs.org.
■ AUTHOR INFORMATION
Corresponding Author
*Phone: (510) 643-9120; fax: (510) 486-5686; e-mail: mzhu@
lbl.gov.
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS
This work was supported by the Director, Office of Science,
Office of Basic Energy Sciences, Chemical Sciences, Geo-
sciences, and Biosciences Division, of the U.S. Department of
Energy under Contract No. DE-AC02-05CH11231. The
authors are grateful to beamline scientists Dr. Syed Khalid at
beamline X18B at NSLS, Brookhaven National Laboratory
(BNL), Dr. Karena Chapman, Dr. Olaf Borkiewicz and Dr.
Peter Chupas at beamline 11-ID-B at APS, Argonne National
Laboratory (ANL), and Dr. John Bargar at beamline 4-1 at
Stanford Synchrotron Radiation Lightsource (SSRL) for
providing technical help on data collection. Use of the National
Synchrotron Light Source, Brookhaven National Laboratory
was supported by the U.S. DOE Office of Science, Office of
Basic Energy Sciences, under Contract No. DE-AC02-
98CH10886. Use of APS was supported by the U.S. DOE
Office of Science under Contract No. DE-AC02-06CH11357.
Portions of this research were carried out at the Stanford
Synchrotron Radiation Laboratory, a national user facility
dx.doi.org/10.1021/es301268g | Environ. Sci. Technol. 2012, 46, 8140−8147
Environmental Science & Technology
operated by Stanford University on behalf of the U.S.
Department of Energy, Office of Basic Energy Sciences.
■
REFERENCES
(1) Evangelou, V. P.; Zhang, Y. L. A reviewPyrite oxidation
mechanisms and acid-mine drainage prevention. Crit. Rev. Environ. Sci.
Technol. 1995, 25 (2), 141−199.
(2) Burgos, W. D.; Borch, T.; Troyer, L. D.; Luan, F.; Larson, L. N.;
Brown, J. F.; Lambson, J.; Shimizu, M. Schwertmannite and Fe oxides
formed by biological low-pH Fe(II) oxidation versus abiotic
neutralization: Impact on trace metal sequestration. Geochim.
Cosmochim. Acta 2011, 76, 29−44.
(3) Johnson, D. B.; Hallberg, K. B. Acid mine drainage remediation
options: A review. Sci. Total Environ. 2005, 338 (1−2), 3−14.
(4) Waychunas, G. A.; Kim, C. S.; Banfield, J. F. Nanoparticulate iron
oxide minerals in soils and sediments: Unique properties and
contaminant scavenging mechanisms. J. Nanopart. Res. 2005, 7,
409−433.
(5) Chan, C. S.; De Stasio, G.; Welch, S. A.; Girasole, M.; Frazer, B.
H.; Nesterova, M. V.; Fakra, S.; Banfield, J. F. Microbial
polysaccharides template assembly of nanocrystal fibers. Science
2004, 303 (5664), 1656−1658.
(6) Flynn, C. M. Hydrolysis of inorganic iron(III) salts. Chem. Rev.
1984, 84 (1), 31−41.
(7) Banfield, J. F.; Welch, S. A.; Zhang, H.; Ebert, T. T.; Penn, R. L.
Aggregation-based crystal growth and microstructure development in
natural iron oxyhydroxide biomineralization products. Science 2000,
289 (5480), 751−754.
(8) Jolivet, J.-P.; Henry, M.; Livage, J., Metal Oxide Chemistry and
Synthesis: From Solution to Solid State; John Wiley & Sons Inc: New
York, 2000.
(9) Rose, J.; Manceau, A.; Bottero, J.-Y.; Masion, A.; Garcia, F.
Nucleation and growth mechanisms of Fe oxyhydroxide in the
presence of PO4 ions. 1. Fe K-Edge EXAFS study. Langmuir 1996, 12
(26), 6701−6707.
(10) Masion, A.; Doelsch, E.; Rose, J.; Moustier, S.; Bottero, J. Y.;
Bertsch, P. M. Speciation and crystal Chemistry of iron(III) chloride
hydrolyzed in the presence of SiO4 ligands. 3. Semilocal scale structure
of the aggregates. Langmuir 2001, 17 (16), 4753−4757.
(11) Cornell, R. M.; Schwertmann, U., The Iron Oxides: Structure,
Properties, Reactions, Occurrences and Uses, 2nd ed.; Wiley-VCH: New
York, 2003; p 703.
(12) Combes, J. M.; Manceau, A.; Calas, G.; Bottero, J. Y. Formation
of ferric oxides from aqueous solutions: A polyhedral approach by X-
ray absorption spectroscdpy: I. Hydrolysis and formation of ferric gels.
Geochim. Cosmochim. Acta 1989, 53 (3), 583−594.
(13) Bottero, J. Y.; Tchoubar, D.; Arnaud, M.; Quienne, P. Partial
hydrolysis of ferric nitrate salt. Structural investigation by dynamic
light scattering and small-angle x-ray scattering. Langmuir 1991, 7 (7),
1365−1369.
(14) Schwertmann, U.; Cornell, R. M., Iron Oxides in the Laboratory:
Preparation and Characterization, 2nd ed.; Wiley-VCH: New York,
2000.
(15) Khalid, S.; Caliebe, W.; Siddons, P.; So, I.; Clay, B.; Lenhard, T.;
Hanson, J.; Wang, Q.; Frenkel, A. I.; Marinkovic, N.; Hould, N.;
Ginder-Vogel, M.; Landrot, G. L.; Sparks, D. L.; Ganjoo, A., Quick
extended x-ray absorption fine structure instrument with millisecond
time scale, optimized for in situ applications. Rev. Sci. Instrum. 2010,
81, (1).
(16) Ravel, B.; Newville, M. ATHENA, ARTEMIS, HEPHAESTUS:
Data analysis for X-ray absorption spectroscopy using IFEFFIT. J
Synchrotron Radiat. 2005, 12, 537−541.
(17) Webb, S. SIXPack: A graphical user interface for XAS analysis
using IFEFFIT. Phys. Scr. 2005, T115, 1011−1014.
(18) Jiang, J. Q.; Graham, N. J. D. Observations of the comparative
hydrolysis/precipitation behaviour of polyferric sulphate and ferric
sulphate. Water Res. 1998, 32 (3), 930−935.
(19) Stotzel, J.; Lutzenkirchen-Hecht, D.; Frahm, R.; Santilli, C. V.;
Pulcinelli, S. H.; Kaminski, R.; Fonda, E.; Villain, F.; Briois, V.
8147
Article
QEXAFS and UV/Vis simultaneous monitoring of the TiO2-
nanoparticles formation by hydrolytic sol-gel route. J. Phys. Chem. C
2010, 114 (14), 6228−6236.
(20) Lopes, L.; de Laat, J.; Legube, B. Charge transfer of iron(III)
monomeric and oligomeric aqua hydroxo complexes: Semiempirical
investigation into photoactivity. Inorg. Chem. 2002, 41 (9), 2505−
2517.
(21) Rossman, G. R. Spectroscopic and magnetic studies of ferric
iron hydroxy sulfatesIntensification of color in ferric iron clusters
bridged by a single hydroxide ion. Am. Mineral. 1975, 60 (7−8), 698−
704.
(22) Smolakova, L.; Grygar, T.; Capek, L.; Schneeweiss, O.; Zboril,
R. Speciation of Fe in Fe-modified zeolite catalysts. J. Electroanal.
Chem. 2010, 647 (1), 8−19.
(23) Scheinost, A. C.; Chavernas, A.; Barron, V.; Torrent, J. Use and
limitations of second-derivative diffuse reflectance spectroscopy in the
visible to near-infrared range to identify and quantify Fe oxide minerals
in soils. Clay Clay Miner. 1998, 46 (5), 528−536.
(24) Webb, S. M.; Tebo, B. M.; Bargar, J. R. Structural
characterization of biogenic Mn oxides produced in seawater by the
marine Bacillus sp. strain SG-1. Am. Mineral. 2005, 90, 1342−1357.
(25) Majzlan, J.; Myneni, S. C. B. Speciation of iron and sulfate in
acid waters: Aqueous clusters to mineral precipitates. Environ. Sci.
Technol. 2004, 39 (1), 188−194.
(26) Dousma, J.; den Ottelander, D.; de Bruyn, P. L. The influence of
sulfate ions on the formation of iron(III) oxides. J. Inorg. Nucl. Chem.
1979, 41 (11), 1565−1568.
(27) Sakane, H.; Munoz-Pαez, A.; Diaz-Moreno, S.; Martinez, J. M.;
Pappalardo, R. R.; Sanchez Marcos, E. Second hydration shell single
scattering versus first hydration shell multiple scattering in M(H2O)63+
EXAFS spectra. J. Am. Chem. Soc. 1998, 120 (40), 10397−10401.
(28) Hansel, C. M.; Benner, S. G.; Fendorf, S. Competing Fe(II)-
induced mineralization pathways of ferrihydrite. Environ. Sci. Technol.
2005, 39 (18), 7147−7153.
(29) Hemmes, P.; Rich, L. D.; Cole, D. L.; Eyring, E. M. Kinetics of
hydrolysis of ferric ion in dilute aqueous solution. J. Phys. Chem. 1971,
75 (7), 929−932.
(30) Liu, H.; Li, P.; Zhu, M.; Wei, Y.; Sun, Y. Fe(II)-induced
transformation from ferrihydrite to lepidocrocite and goethite. J. Solid
State Chem. 2007, 180 (7), 2121−2128.
(31) Michel, F. M.; Ehm, L.; Antao, S. M.; Lee, P. L.; Chupas, P. J.;
Liu, G.; Strongin, D. R.; Schoonen, M. A. A.; Phillips, B. L.; Parise, J. B.
The structure of ferrihydrite, a nanocrystalline material. Science 2007,
316, 1726−1728.
(32) Bradley, S. M.; Kydd, R. A. Comparison of the species formed
upon base hydrolyzes of gallium(III) and iron(III) aqueous solutions -
the possibility of existence of an [FeO4Fe12(OH)24(H2O)12]7+
polyoxocation. J. Chem. Soc., Dalton 1993, 15, 2407−2413.
(33) Johansson, G.; Lundgren, G.; Sillén, L. G.; Söderquist, R. On the
crystal structure of a basic aluminium sulfate and the corresponding
selenate. Acta Chem. Scand. 1960, 14, 771.
(34) Fukushi, K.; Sverjensky, D. A. A surface complexation model for
sulfate and selenate on iron oxides consistent with spectroscopic and
theoretical molecular evidence. Geochim. Cosmochim. Acta 2007, 71
(1), 1−24.
dx.doi.org/10.1021/es301268g | Environ. Sci. Technol. 2012, 46, 8140−8147