Chemical Geology 410 (2015) 53 62
Contents lists available at ScienceDirect
Chemical Geology
j ournal homepage: www.el sevier.com/l ocate/chemgeo
Arsenate and phosphate adsorption on ferrihydrite nanoparticles.
Synergetic interaction with calcium ions
a, b
N
Juan Antelo , Florencio Arce , SarahFiolb
a
Department of Soil Science and Agricultural Chemistry, University of Santiago de Compostela, 15782 Santiago de Compostela, Spain
b
Department of Physical Chemistry, University of Santiago de Compostela, 15782 Santiago de Compostela, Spain
a r t i c l e i n f o a b s t r a c t
Article history:
The geochemical behaviour of phosphate and arsenate ions in soil and aquatic systems is determined by the pres-
Received 8 January 2015
ence of mineral surfaces and major ions. Information about the distribution of oxyanions over the solid and solu-
Received in revised form 8 June 2015
tion phases is essential for understanding the transport, bioavailability and toxicity of these compounds in the
Accepted 9 June 2015
environment. Here, we studied the adsorption of both arsenate and phosphate on ferrihydrite nanoparticles in
Available online 12 June 2015
the presence of calcium ions. The presence of calcium ions enhanced the retention of these oxyanions on ferrihy-
drite and vice versa. The arsenate calcium and phosphate calcium multi-component systems were described
Keywords:
using a mechanistic surface complexation model. Use of this type of model enables prediction of the solution
Ferrihydrite
and surface speciation, along with analysis of oxyanion mobility in relation to environmental conditions. We
Phosphate
Arsenate were able to calibrate the charge distribution model with the macroscopic data obtained for the single-
Calcium
component systems, thus obtaining surface complexation constants for later use to simulate multi-component
Surface complexation modelling
systems. The mutual interactions between arsenate and calcium were successfully described with these param-
CD model
eters, indicating that changes in the electrostatic forces at the solid/solution interface caused the observed
enhanced adsorption. However, adsorption in the phosphate calcium system was underestimated with the
parameters obtained for the single-component systems, indicating that additional mechanisms or processes
should be considered. Formation of insoluble mineral phases was ruled out, but the inclusion of a phosphate
calcium ternary surface complex improved the modelling predictions.
© 2015 Elsevier B.V. All rights reserved.
1. Introduction biotoxicity. Usually the concentration of arsenic in natural waters is
below the guideline value proposed by the World Health Organization
The presence of phosphorus and arsenic in surface waters and for drinking water, 10 źg/l (WHO, 2011), although concentrations
groundwaters is considered a major environmental problem world- above this value are not uncommon and its concentration can reach
wide. Phosphorus is an essential nutrient for plant growth in soils up to hundreds of mg/l in systems affected by acid mine drainage
and often limits algal growth and eutrophication in surface waters (Nordstrom, 2002; Smedley and Kinniburgh, 2002). The chemical be-
(Klapper, 1991). Although phosphorus in freshwaters is not consid- haviour of phosphorus and arsenic is similar and both elements tend
ered to be directly toxic to humans and animals, elevated concentra- to form analogous species such as phosphate and arsenate; however,
tions can adversely affect the ecosystems and ultimately produce the biogeochemical behaviour of these elements is different. Thus,
anoxic conditions. Usually the contamination of surface waters by although phosphate can be considered as a macronutrient, arsenate is
phosphorus is caused by a combination of over-fertilization and very toxic at low concentrations.
agricultural land runoff, which may lead to concentrations above For a better knowledge of how these two species are distributed be-
the recommended limit proposed by USEPA, 0.1 mg/l. On the other tween the solid and solution phases in natural systems, it is essential to
hand, arsenic is a very toxic element that may be present at high con- understand how they interact with the reactive constituents present in
centrations in freshwater environments due to weathering processes these systems. Both arsenate and phosphate display a relatively strong
or to anthropogenic sources as mining and agricultural activities affinity for iron mineral oxides (Raven et al., 1998; Gao and Mucci,
(Smedley and Kinniburgh, 2002). The major health and ecological 2003; Antelo et al., 2005; Kanematsu et al., 2010; Carabante et al.,
issues of arsenic are related to its carcinogenicity, phytotoxicity and 2012). Ferrihydrite is one of the most reactive and most common natu-
rally occurring iron oxides. Its occurrence has been reported at near-
neutral pH conditions in a variety of redox-active environments, such
as soils and sediments, or freshwater and marine settings (Jambor and
N Corresponding author.
E-mail address: [e-mail] (J. Antelo). Dutrizac, 1998). The oxidation and dissolution of Fe-bearing sulphide
http://dx.doi.org/10.1016/j.chemgeo.2015.06.011
0009-2541/© 2015 Elsevier B.V. All rights reserved.
54 J. Antelo et al. / Chemical Geology 410 (2015) 53 62
minerals associated with mine wastes and acid mine drainage also re- 2.2. Ferrihydrite synthesis and characterization
sult in the formation of ferrihydrite, along with other secondary iron ox-
ides. Due to its abundance, large surface area and high reactivity it is Two-line ferrihydrite was synthesized following the procedure pro-
considered a key mineral for metal sequestration and plays a crucial posed by Cornell and Schwertmann (1996). Briefly, a 0.1 M Fe(NO3)3
role in the cycling of iron and trace elements in natural systems. The rel- solution was adjusted to pH 8.0 through dropwise addition of a 1 M
evance of this iron hydroxide has been made evident from its ability to NaOH solution under continuous stirring and N2 bubbling. The ferrihy-
adsorb different contaminants on its surface and to control the transport drite dispersion thus obtained was aged at 20 °C for 48 h and then dia-
of ions in soils and aqueous systems (Dzombak and Morel, 1990). lyzed until the conductivity was lower than 10 źS/cm. The dispersion
A better understanding of the adsorption of oxyanions on this iron was then freeze-dried to obtain a dry powder. Information found in
oxide has become increasingly important in recent years (Harrington the literature indicates that no mineralogical alterations are produced
et al., 2010; Das et al., 2011; Wang et al., 2013; Neupane et al., 2014; on mineral oxides after freeze-drying (Lee et al., 2002).
Zhu et al., 2014). Retention of oxyanions can be affected by the presence The ferrihydrite sample was previously characterized by powder X-
of other ions, which may enhance or suppress the adsorption process. A ray diffraction and transmission electron microscopy using a Phillips
good knowledge of the interactions and effects of co-existing elements PW1710 diffractometer and a Philips CM-12 microscope, respectively
(e.g. calcium) in soils and aquatic systems is essential for modelling (Antelo et al., 2010). The diffractogram revealed two broad peaks,
and predicting the mobility of phosphate and arsenate. The presence which confirm the presence of 2-line ferrihydrite, while microscopic
of positively charged ions such as calcium may enhance the adsorption images showed that ferrihydrite particles were heavily aggregated.
of negatively charged ions such as phosphate or arsenate on mineral The surface area, measured by N2 adsorption with an ASAP 2000
surfaces above pH 7 (Rietra et al., 2001; Stachowicz et al., 2008; Arai, V3.03 Micromeritics instrument, was 229 m2/g. This value is in agree-
2010; Kanematsu et al., 2013). Most of the aforementioned studies ment with surface areas reported by Cornell and Schwertmann (1996)
focus on goethite, which has become a model mineral surface for ad- for different ferrihydrite samples. As the surface area values obtained
sorption studies. Limited studies can be found in the literature devoted by BET measurements are of limited reliability for ferrihydrite, the
to integration of macro- and microscopic information to obtain a de- actual surface area of the ferrihydrite sample was obtained by compar-
tailed mechanistic description of the competitive and synergetic effects ison of the surface charge determined by modelling calculations and the
found in multi-component systems containing ferrihydrite (Gustafsson, experimental surface charge (Davis and Leckie, 1978). The surface area
2003; Swedlund et al., 2003; Tiberg et al., 2013; Neupane et al., 2014). obtained by this method (350 m2/g) (Antelo et al., 2010) is lower than
As stated above, information about the enhancement of arsenate and the theoretical values proposed for fresh ferrihydrite (range 600
phosphate adsorption on some iron oxides and clays due to the pres- 750 m2/g) (Dzombak and Morel, 1990). The point of zero charge
ence of calcium can be found in the literature, however discussion (PZC) was previously determined by potentiometric titrations (8.7 Ä…
about multi-component systems simultaneously involving environ- 0.1) (Antelo et al., 2010). Although the value is slightly higher than
mentally relevant concentrations of calcium, phosphate or arsenate those usually reported for freshly precipitated ferrihydrites, it is consis-
and ferrihydrite is not available to our knowledge. Thermodynamic de- tent with other PZC values reported in the literature (Raven et al., 1998;
scription of the surface reactivity of ferrihydrite is crucial for developing Hofmann et al., 2005; Wang et al., 2013). Freshly precipitated and not
models that predict the fate of geochemically and environmentally rigorously de-carbonated ferrihydrite samples show consistently low
relevant species in settings where this mineral oxide is formed. PZC values (7.9 8.1), whereas dialyzed and N2-purged samples after
In this study, we investigated the effect of calcium on the adsorption synthesis tend towards higher values (8.6 8.7).
of arsenate and phosphate on 2-line ferrihydrite nanoparticles. We
studied the binding processes for arsenate, phosphate and calcium 2.3. Arsenate adsorption to ferrihydrite
separately, prior to investigating arsenate and phosphate adsorption
in the presence of calcium ions. Arsenate, phosphate and calcium con- Arsenate adsorption isotherms for ferrihydrite were obtained at
centrations in this study have been selected to simulate the levels that pH 4.5, 7.0 and 9.0 to cover a wide range of pH and to allow comparison
these species can reach in polluted systems, such as waters affected by with phosphate adsorption data. For this purpose, batch experiments
mining activities or settings under the influence of intense agricultural were conducted with 1 g/l suspensions in 20 ml of 0.1 M KNO3. For
activities. Furthermore, the arsenate concentrations employed enable each data point, an arsenate solution (KH2AsO4) of the desired initial
direct comparison with phosphate data, which have been taken for concentration was prepared at the desired pH and ionic strength. The
the single-component system from Antelo et al. (2010). We also solution was then added to the ferrihydrite suspension and the pH
evaluated the capacity of the Charge Distribution (CD) model to pre- was adjusted by addition of 0.1 M HNO3 or KOH solutions. The KOH so-
dict the behaviour of these oxyanions in multi-component systems. lution was freshly prepared for each experiment from a non-carbonated
Finally, we highlight the importance of obtaining surface parame- stock solution and maintained under N2 atmosphere. Preliminary kinet-
ters for thermodynamic description of the adsorption behaviour of ic experiments showed that a contact time of 24 h was required to
ferrihydrite for future application of modelling parameters to field achieve equilibrium. During the equilibration period, the samples
data. were continuously shaken with a magnetic stirrer, and the pH was peri-
odically measured and readjusted, where necessary. Special care was
taken to prevent the presence of CO2, by maintaining the ferrihydrite
2. Materials and methods suspensions in an N2 atmosphere. After 24 h, the pH was recorded and
the ferrihydrite suspensions were filtered through 0.22 źmmembrane
2.1. Reagents and materials filters. A colorimetric method was used to determine the concentration
of arsenate in solution by UV visible spectroscopy (Lenoble et al., 2003),
All chemicals were of Merck p.a. quality, and the water used in the and the amount adsorbed was calculated as the difference between the
experiments was ultrapure (Milli-Q water-purification system; resistiv- total arsenate concentration and the arsenate concentration in the su-
ity 18 M© cm at 25 °C) and CO2 free. A-grade glassware was used in the pernatant. To ensure that no ferrihydrite particles were present in the
preparation of stock solutions. Plastic material was used for ferrihydrite final filtered solution, aliquots of the filtered solution were acidified
synthesis and adsorption experiments in order to avoid silica contami- with a 1 M solution of HNO3, and the iron concentration in solution
nation. In addition, all glassware and plastic materials were washed was measured after 24 h by atomic absorption spectroscopy (Perkin
with 10% HNO3 and rinsed with ultrapure water to prevent contamina- Elmer 1100B). The results showed that iron concentrations were
tion by metals. below the detection limit (<0.01 mg/l).
J. Antelo et al. / Chemical Geology 410 (2015) 53 62 55
Adsorption envelope experiments were also carried out to deter- above-mentioned methods. The concentration of adsorbed ions was cal-
mine the amount of arsenate adsorbed to the ferrihydrite surface as a culated as the difference between the initial and final concentrations in
function of pH (within the range 4.0 10.0 in 0.5 1.0 pH increments) the solution.
and ionic strength (0.01, 0.1 and 0.5 M in KNO3). Arsenate and ferrihy-
drite were mixed in 50 ml polyethylene bottles to provide a final 2.6. CD model
arsenate concentration of 0.60 mM and a ferrihydrite concentration of
1 g/l. Separate bottles were used for each pH value studied. The pH of Application of surface complexation models (SCMs) to describe the
the suspensions was adjusted and any change in the pH during the adsorption behaviour of ferrihydrite may be challenging because of
experiment was corrected by adding small volumes of 0.1 M HNO3 or the limited and contradictory information about the crystalline struc-
KOH. The samples were shaken for 24 h, and the pH values were record- ture of ferrihydrite. It was recently shown that both the charging behav-
ed at the end of the experiment. Arsenate concentration in solution was iour and ion adsorption could be modelled for ferrihydrite by using
measured following the above-mentioned procedure. goethite as a proxy (Hiemstra and van Riemsdijk, 2009; Hiemstra
Each experiment was carried out in duplicate (at least) to confirm et al., 2009; Villalobos and Antelo, 2011; Tiberg et al., 2013). Among
the reproducibility (uncertainty <2%). Polyethylene flasks were used many SCMs that have been applied so far, the CD model has become
in the adsorption experiments in order to avoid contamination by the most popular for describing the surface reactivity of crystalline
silicates, and the temperature was maintained at 25 Ä… 1 °C in all exper- and amorphous mineral oxides (Hiemstra and van Riemsdijk, 1996;
iments. The pH measurements were conducted after the combined glass Hiemstra and van Riemsdijk, 2006). This model combined with the Ex-
electrode (Radiometer GK-2401C) was calibrated with standard buffer tended Stern concept for the description of the solid/solution interface
solutions. has been used to describe the experimental data and to provide a micro-
scopic interpretation of adsorption in the single- and multi-component
2.4. Calcium adsorption to ferrihydrite systems.
Optimization of the parameters required for describing the arsenate,
A similar procedure to that described above was carried out to obtain phosphate and calcium binding to the ferrihydrite surface with the CD
the calcium adsorption edges on ferrihydrite. Batch experiments were model was carried out with the ECOSAT (Keizer and van Riemsdijk,
conducted at different pH values in separate vials with 1 g/l ferrihydrite 1998) speciation program and the FIT program (Kinniburgh, 1993).
suspensions in 20 ml of 0.1 M KNO3. A calcium solution (0.1 M Ca(NO3) The solution species and their equilibrium constants were formulated
) was added to the suspensions to yield three different concentrations as summarized in Table S1 (Supplementary data). Surface complexation
2
(0.5, 2 and 6 mM), and the pH of the suspensions was adjusted to the parameters specific for ferrihydrite and for the solid/solution interface
desired pH value (within the range 4.0 10.5 in 0.7 1.0 pH increments) (such as surface area, site density, capacitance or protonation constants)
by addition of 0.1 M HNO3 or KOH solutions. During the 24 h equilibra- are shown in Table 1.
tion period, the suspensions were continuously shaken and the pH was
periodically measured and readjusted, when necessary. Special care was 3. Results and discussion
taken to prevent the presence of CO2, by maintaining the ferrihydrite
suspensions in an N2 atmosphere. The pH was registered after 24 h 3.1. Arsenate adsorption
and the suspensions were then filtered through 0.22 źm membrane fil-
ters. The concentration of calcium in solution was measured by induc- Fig. 1 shows the arsenate adsorption isotherms in 0.1 M KNO3 at dif-
tively coupled plasma optical emission spectrometry (ICP-OES) ferent pH values and the adsorption envelopes at the different ionic
(Optima 3300 DV, Perkin Elmer). The amount of calcium adsorbed strengths for an initial arsenate concentration of 0.6 mM. The figure
was calculated as the difference between the concentration of calcium also illustrates the CD model predictions and the arsenate surface speci-
initially added to the suspensions and the concentration measured in ation as a function of pH, according to the modelling calculations. No
solution at the end of the experiment. difference was observed between adsorption on freeze-dried ferrihy-
drite or ferrihydrite stored as a wet paste (Fig. S1, Supplementary
2.5. Adsorption experiments in multi-component systems data). Arsenate adsorption decreased gradually and continuously as
the pH of the systems increased, as expected from previous findings re-
Phosphate and arsenate adsorption experiments were carried out in garding oxyanion adsorption on amorphous and crystalline iron oxides
the presence of calcium at different pH (5.0 10.0) and constant ionic (Salazar-Camacho and Villalobos, 2010; Villalobos and Antelo, 2011;
strength. A constant initial concentration (0.6 mM) of phosphate or ar- Kanematsu et al., 2013). The point of zero charge of the ferrihydrite
senate was used at each pH, whereas different volumes of calcium solu- particles is 8.7, indicating that at low pH the iron hydroxyl groups are
tion were added to the suspensions to yield an initial concentration of mainly positively charged, which will favour the interaction (via ligand
calcium between 0.3 mM and 6 mM. Additionally, adsorption experi- exchange) between the arsenate ions and the iron surface groups.
ments with initial arsenate concentrations of 0.1, 0.3, 0.45 and 0.8 mM The effect of ionic strength is rather low at pH < 5.0, whereas an in-
were conducted at two calcium loadings, 0.7 and 6 mM. Appropriate crease in the ionic strength at pH > 5.0 produced an increase in the ad-
volumes of distilled water and 1 M KNO3 solution were added to ensure sorption of arsenate ions. This effect, previously observed for phosphate
an ionic strength of 0.1 M in all experiments. The final volume in the binding on goethite and ferrihydrite (Rahnemaie et al., 2007; Antelo
suspensions for each data point was 20 ml. To minimize precipitation et al., 2010), is usually attributed to changes in the electrostatic poten-
of calcium phosphate or calcium arsenate, calcium was added to ferrihy- tial at the solid/solution interface. An increase in the ionic strength pro-
drite suspensions (1 g/l) that had been pre-equilibrated at the desired duces a decrease in the electrostatic repulsion between the charged
pH with phosphate or arsenate for 2 h. Any change in pH after the surface and the arsenate ions, favouring the adsorption process at rela-
mixing was corrected by addition of 0.1 M HNO3 or KOH solutions. tively high pH. Ions that form inner-sphere complexes, such as arsenate
The suspensions were equilibrated for 24 h, and the pH was periodically ions, are directly coordinated to surface hydroxyl groups and may not
monitored and readjusted, when necessary, during the experiment. compete, or may compete relatively weakly with electrolyte ions.
After the equilibration period, the suspensions were filtered and the As stated above, arsenate adsorption on ferrihydrite was simulated
solutions were analysed for phosphate, arsenate and calcium. The using the CD model, which requires molecular scale information to con-
phosphate concentration was measured spectrophotometrically by the strain the nature and the charge distribution of the surface complexes
molybdenum blue method (Murphy and Riley, 1962), whereas the formed at the solid/solution interface. Previous spectroscopic and mo-
concentrations of arsenate and calcium were determined by the lecular studies have provided very useful information about the identity
56 J. Antelo et al. / Chemical Geology 410 (2015) 53 62
Table 1
Surface species and CD model parameters for H+, K+, NO- and PO-3 binding to ferrihydrite, estimated using the Extended Stern layer model and considering C1 =0.74F/m2 and C2 =
3 4
0.93 F/m2. "z0, "z1, and"z2 represent the change in the charge (or charge distribution) in the 0-, 1-, and 2-planes, respectively.
Surface reactions `FeOH `Fe3O "z0 "z1 "z2 H+ K+ NO- PO-3 log K
3 4
`FeOH-1/2 1 0 0 0 0 0 0 0 0 0.00
`FeOH+1/2 1 0 +1 0 0 1 0 0 0 8.70
2
`FeOH-1/2ï"K+ 1 0 0+10 0 1 0 0  1.16
`FeOH+1/2ï"NO- 10 +1  1 0 1 0 1 0 7.74
2 3
`Fe3O-1/2 0 1 0 0 0 0 0 0 0 0.00
`Fe3OH+1/2 0 1 +1 0 0 1 0 0 0 8.70
`Fe3O-1/2ï"K+ 0 1 0+10 0 1 0 0  1.16
`Fe3OH+1/2ï"NO- 01 +1  1 0 1 0 1 0 7.74
3
`Fe2O2PO-2 2 0 +0.46  1.46 0 2 0 0 1 27.78
2
`Fe2O2POOH- 2 0 +0.63  0.63 0 3 0 0 1 32.09
Surface site density of the singly and triply coordinated groups is set to Ns,1 =6sites/nm-2 and Ns,3 =1.2sites/nm-2, respectively. Surface area is set to 350 m2/g. Values for the surface
complexation constants were taken from Antelo et al. (2010).
of the surface species that arsenate ions form on iron oxide surfaces. of the predominance of bidentate surface complexes, the presence
Studies by Waychunas et al. (1993) and by Sherman and Randall of protonated monodentate surface complexes was also considered.
(2003) indicated that bidentate surface complexes are thermodynami- Inclusion of this complex, along with the bidentate complexes, did
cally and kinetically favoured over monodentate surface complexes. not significantly improve the modelling predictions (see r2 values in
The EXAFS study conducted by Morin et al. (2008) confirmed that arse- Table 2). The fitting results showed that we were not able to distinguish
nate is predominantly bound to iron oxide surfaces as bidentate corner between both model scenarios, possibly due to the predominance of the
sharing complexes, regardless of the nature of the iron oxide mineral non-protonated bidentate surface complex in both model options. As
(goethite, ferrihydrite, lepidocrocite, or maghemite). The recent litera- the main objective was to achieve good predictions of the experimental
ture reports some controversy about the nature of the arsenate surface data for arsenate and then to analyse the suitability of the complexation
complexes, especially after the thorough study of Loring et al. (2009). constants in a multi-component system including arsenate and calcium,
These authors combined EXAFS information with data obtained by IR no additional calculations or efforts were conducted to identify the
spectroscopy, and they suggested that monodentate coordination may exact nature of the surface complexes occurring at the ferrihydrite
be the dominant geometry for arsenate binding on iron oxides. surface.
In the present study, two bidentate surface complexes (protonated Fig. 1c shows the abundance of arsenate surface species as a function
and non-protonated) were considered in the modelling calculations of pH for an arsenate loading of 0.6 mM at 0.1 M ionic strength accord-
conducted to describe the adsorption of arsenate on ferrihydrite. As a ing to the CD model. Under these conditions, the dominant surface
simplification, it was assumed that all the singly coordinated groups species at intermediate to high pH is the non-protonated bidentate
present in ferrihydrite show the same reactivity against arsenate ions, complex. At the lower pH values, the protonated bidentate complex
and no distinction was made between corner sharing and edge sharing contributes less to the arsenate adsorption than the non-protonated
surface complexes. A second modelling scenario, which included the bidentate complex, although it makes a greater contribution at pH < 3.
formation of singly-protonated monodentate surface complexes, was The bidentate complexes predominate when the protonated
also considered, as these complexes may also occur at the ferrihydrite monodentate complex is considered (Fig. S2, Supplementary
surface. Neither scenario considered the formation of surface ion-pair data). The inclusion of the monodentate complex accounts for ~15% of
complexes, i.e. K AsO4, since their inclusion does not improve the the total arsenate adsorbed at acidic pH values and its contribution de-
modelling predictions (Rahnemaie et al., 2007). The surface reactions creases with the increase of pH. On the other hand, for the protonated
for the three surface complexes considered can be formulated as surface complexes the ratio between monodentate and bidentate com-
follows: plexes decreases when the arsenate loading increases. These results are
partially in agreement with the findings of Waychunas et al. (1993),
0 1
a"2FeOH-1=2 þ 2Hþþ AsO4-3Ć! a"Fe2O2-1þ"z AsO2"z þ 2H2O ð1Þ
who found that the contribution of the monodentate complex only
accounted for 30% of the adsorbed arsenate at the most favourable
0 1
a"2FeOH-1=2 þ 3Hþþ AsO4-3Ć! a"Fe2O2-1þ"z AsOOH"z þ 2H2O ð2Þ conditions. The relative abundance of the monodentate surface com-
plex does not change with ionic strength, whereas the abundance of
þ"z0
1
bidentate complexes is ionic strength dependent (Fig. S3, Supplementa-
a"FeOH-1=2 þ 2Hþþ AsO4-3 Ć! a"FeO-1=2 AsO2OH"z þ H2O ð3Þ
ry data).
The only parameters allowed to vary in the fitting were the com- Finally, a comparison can be established between arsenate and
plexation constants of the arsenate surface species. The complexation phosphate adsorption on ferrihydrite. Considering the similar chemical
constants previously obtained by Stachowicz et al. (2006) for the ad- behaviour shown by both oxyanions, a similar affinity on ferrihydrite
sorption of arsenate on goethite were used as initial estimates. The particles may be expected. The adsorption observed for both oxyanions
charge distribution coefficients "z0 and "z1 were also taken from the confirms this behaviour (Fig. S4, Supplementary data). On the other
same study. These values may not be very accurate because they were hand, the surface complexation constants calculated for the oxyanions
derived from molecular orbital and density functional theory calculations are comparable (Table 1 and option I in Table 2). The slight differences
and were corrected to take into account the electrostatic dipole effect between these constants are consistent with the differences found in
induced by the introduction of charge at the solid/solution interface. the solution chemistry. According to the thermodynamic data available
Arelativelygoodfit to the experimental results was achieved by for aqueous speciation of arsenate and phosphate (Table S1, Supple-
optimizating the complexation constants of the protonated and mentary data), protonation and ion-pair formation constants are
non-protonated bidentate complexes (Fig. 1a andFig. 1b). The sur- systematically higher for the phosphate species than for the arsenate
face complexation parameters for arsenate are shown in Table 2. Al- analogues. The same tendency can be observed for the surface complex-
though these results are consistent with the spectroscopic evidence ation constants.
J. Antelo et al. / Chemical Geology 410 (2015) 53 62 57
-2.9 7.0
6 mM Ca
a
6.0
-3.3
5.0
4.0
-3.7
3.0
2 mM Ca
2.0
-4.1 pH = 4.5
pH = 7.0
0.5 mM Ca
1.0
pH = 9.0
-4.5 0.0
-8.0 -7.0 -6.0 -5.0 -4.0 -3.0 -2.0 3 5 7 9 11
log [AsO4]sol (mol/l)
pH
Fig. 2. Effect of the pH on the equilibrium calcium concentration in solution after adsorp-
0.7 tion on ferrihydrite. Solid and dotted lines correspond to the results of the modelling
scenarios I and II, respectively.
b
0.6
0.5
3.2. Calcium adsorption
0.4
Fig. 2 shows that calcium adsorption on ferrihydrite only occurs at
0.3
pH > 8.5. An almost horizontal edge was observed for the three calcium
loadings, indicating that no adsorption takes place within the pH range
0.2
I = 0.5 M
4.0 8.5. A notable decrease in the calcium concentrations in solution
I = 0.1 M
0.1 only was observed beyond pH 8.5. This pH is close to the PZC (8.7)
I = 0.01 M
and as the mineral surface develops an increasingly negative charge at
0.0
pH > PZC, attractive electrostatic interactions will favour calcium ion ad-
2 4 6 8 10 12
sorption. The adsorption of calcium is relatively weak compared with
pH
that of heavy metals such as copper, lead, and zinc and is consistent
with the findings reported by Dzombak and Morel (1990). These au-
thors observed a correlation between the hydrolysis constant and the
100
surface complexation constant. According to this correlation, the diva-
c
lent cation Ca+2, which has a relatively low hydrolysis constant
80
(Table S1), is expected to adsorb weakly onto the surface of amorphous
Fe2O2AsO2
and crystalline iron oxides. The relatively low affinity that iron oxides
60
show for calcium ions was also observed by Stachowicz et al. (2008)
when using goethite as adsorbent material (SSA 98 m2/g; PZC 9.2).
HAsO4
More recently, Kanematsu et al. (2013) conducted adsorption experi-
40
Fe2O2AsOOH
ments at two calcium loadings with a porous adsorbent material that
contained ~90% nanogoethite (SSA 158.1 m2/g; PZC 8.5). These authors
20
observed calcium adsorption above pH 8.0, and adsorption only reached
H2AsO4
AsO4
100% at the lower loading, 0.03 mM.
0
To describe the adsorption of calcium on ferrihydrite using the CD
3 4 5 6 7 8 9 10
model, as previously discussed, molecular scale information is needed.
pH
As far as we are concerned, no spectroscopic information is available
with respect to the nature of the calcium surface complexes occurring
Fig. 1. Arsenate adsorption on ferrihydrite: (a) adsorption isotherms for 0.1 M KNO3;
on iron oxides. Nevertheless, surface speciation may be inferred through
(b) adsorption envelopes at initial arsenate concentration of 0.6 mM; (c) surface (dashed
those studies that applied realistic surface complexation models, such as
lines) and solution (dotted lines) speciation for an arsenate loading of 0.6 mM according to
the CD model or the extended triple layer model, to the adsorption of
the CD model predictions. Solid lines represent the CD model simulations based on option
I scenario (bidentate surface complexes). calcium ions onto goethite (Stachowicz et al., 2008; Hiemstra et al.,
Table 2
Surface species and complexation constants estimated with the CD model for arsenate binding on ferrihydrite.
Surface species `FeOH `Fe3O "z0 "z1 "z2 H+ AsO-3 log K
4
Option I: bidentate surface complexes
`Fe2O2AsO-2 2 0 +0.47  1.47 0 2 1 27.36 Ä… 0.07
2
`Fe2O2AsOOH- 2 0 +0.58  0.58 0 3 1 31.03 Ä… 0.45
2
r = 0.914
Option II: bidentate and (protonated) monodentate surface complexes
`Fe2O2AsO-2 2 0 +0.47  1.47 0 2 1 27.35 Ä… 0.09
2
`Fe2O2AsOOH- 2 0 +0.58  0.58 0 3 1 30.81 Ä… 0.75
`FeOAsO2OH-3/2 1 0 +0.30  1.30 0 2 1 24.76 Ä… 1.02
2
r = 0.916
Standard deviations are given for the fitted parameters.
4 ads
sol
[Ca]
(mmol/l)
log [AsO ]
(mol/l)
4 ads
[AsO ]
(mmol/l)
4
% AsO species
58 J. Antelo et al. / Chemical Geology 410 (2015) 53 62
2010; Kanematsu et al., 2013). In the present study, two different sce- the calculations. Considering the results of this modelling scenario and
narios were considered for the modelling of calcium adsorption using the previously mentioned risk of using too many surface complexes,
the CD model. In the first scenario, we considered the previous informa- an optimization process will be required to describe the single- and
tion available for goethite, whereas some additional considerations multi-component systems with the same set of constants.
were assumed in the second scenario to improve the simulations for
the single-component and multi-component systems. 3.2.2. Modelling scenario II: optimization of the calcium complexation
constants
3.2.1. Modelling scenario I: calcium complexation constants for goethite In this second scenario, an attempt was made to improve the de-
Different surface complexes have been postulated for the interaction scription of calcium adsorption by optimization of the surface complex-
between calcium ions and synthetic and natural iron oxides (Rietra ation constants and the charge distribution at the solid/solution
et al., 2001; Rahnemaie et al., 2006; Stachowicz et al., 2008; Hiemstra interface. Optimization was performed by including all the calcium ad-
et al., 2010). In these studies, the adsorption of calcium ions has been sorption data obtained in the present study from the single-component
described by combining bidentate and monodentate surface complexes systems (ferrihydrite Ca) and also from the multi-component systems
and considering the existence of both inner- and outer-sphere com- (ferrihydrite AsO4 Ca and ferrihydrite PO4 Ca).
plexes. In the present study, the adsorption of calcium on ferrihydrite Initially, only the complexation constants for the inner-sphere com-
was initially explained by considering the calcium surface complexes plexes were optimized. Although the description of calcium adsorption
proposed by Stachowicz et al. (2008) for goethite. On the basis of previ- was improved relative to the previous modelling scenario, adsorption in
ous EXAFS studies with strontium (Axe et al., 1998; Sahai et al., 2000), multi-component systems was still overestimated. We tried to correct
these authors proposed two inner- and two outer-sphere monodentate the excess charge in the 1-plane by transferring part of the charge to
complexes. The same surface complexes have been used by Hiemstra the surface plane. When strong hydrogen bonds are formed between
et al. (2010) to describe phosphate behaviour on field samples, although the surface sites and the outer-sphere complex, a small charge attribu-
the values of the surface complexation constants were slightly different. tion to the surface plane is feasible (Rietra et al., 2001; Rahnemaie
In both cases, the total charge of calcium in the outer-sphere complexes et al., 2006). In our case, a small transfer of charge from the 1-plane to
was located at the 1-plane, and no charge was assigned to the 0-plane. the 0-plane ("z0 =+0.2and"z1 = +1.8) was not sufficient to predict
Awareness should be taken on the risk of using a large number of the mutual effects of calcium and arsenate/phosphate in the multi-
surface complexes in the modelling that might lead to a meaningless component systems.
description of the system. This situation can be especially relevant in The best fits were obtained when the outer-sphere complexes were
the case of species, such as calcium, that show low affinity for the min- excluded from the model. Analysis of the distribution of the calcium sur-
eral surface. The model simulations are less sensitive to changes in the face species reveals that the outer-sphere complexes were not relevant
surface complexation constants and this situation becomes more and the hydrolysed (inner-sphere) form is dominant under the condi-
noticeable when only one dataset is available. tions studied (Fig. S5, Supplementary data). Outer-sphere complexation
The surface reactions for calcium can be formulated as follows: cannot be completely ruled out, although its inclusion did not improve
the modelling simulations. The contribution of these complexes is
"z0 1
slightly significant in the pH range 8.5 9.5, where the calcium adsorbed
a"FeOH-1=2 þ Caþ2Ć! a"FeOH-1=2þ Ca"z ð4Þ
is lower than 10%. The outer-sphere complexes would predominate
"z0 1 over the other surface species at pH below 8.5, where adsorption barely
a"FeOH-1=2 þ H2O þ Caþ2Ć! a"FeOH-1=2þ CaOH"z þ Hþ ð5Þ
reaches 1%. Although most of the modelling work on calcium adsorption
conducted so far considers outer-sphere complexes, Rahnemaie et al.
1
a"FeOH-1=2 þ Caþ2Ć! a"FeOH-1=2 Ca"z ð6Þ
(2006) have also discussed the possibility of describing calcium adsorp-
tion to goethite by considering only inner-sphere complexes. Recalcula-
1
a"Fe3O-1=2 þ Caþ2Ć! a"Fe3O-1=2 CaOH"z þ Hþ ð7Þ
tion of the goethite Ca and goethite PO4 Ca systems studied by
Stachowicz et al. (2008) using their model parameters, without the
As can be observed in Fig. 2, use of the set of constants proposed by inclusion of the outer-sphere complexes, yielded an equally good de-
Hiemstra et al. (2010) (Table 3) yielded acceptable prediction of the scription of their experimental results (Figs. S6 and S7, Supplementary
behaviour of calcium ions, although the amount of calcium in solution data).
is underestimated above pH 9. When these model parameters were Optimization of the complexation constants (Table 3) for the inner-
later applied to describe phosphate and arsenate adsorption in the pres- sphere surface species was required along with relocation of the excess
ence of calcium, the adsorption of both oxyanions was considerably charge in the 1-plane by not considering the formation of the outer-
overestimated. The discrepancy between experimental data and model- sphere complexes. Application of the parameters obtained improved
ling results was attributed to the inclusion of outer-sphere complexes in the prediction of the calcium adsorption in the single-component
Table 3
Surface species and complexation constants estimated with the CD model for calcium binding on ferrihydrite.
Surface species `FeOH `Fe3O "z0 "z1 "z2 H+ Ca+2 log K
Scenario I: Two inner- and two outer-sphere surface complexesa
`FeOHCa+3/2 1 0 +0.31 +1.69 0 0 1 3.23 Ä… 0.05
`FeOHCaOH+1/2 1 0 +0.31 +0.69 0  11 6.42 Ä… 0.08
`FeOH-1/2ï"Ca+2 1 0 0 +2 0 0 1 1.80 Ä… 0.95
`Fe3O-1/2ï"Ca+2 0 1 0 +2 0 0 1 1.80 Ä… 0.95
2
r = 0.992
Scenario II: Two inner-sphere surface complexesb
`FeOHCa+3/2 1 0 +0.31 +1.69 0 0 1  0.89 Ä… 0.35
`FeOHCaOH+1/2 1 0 +0.31 +0.69 0  11 7.20 Ä… 0.15
2
r = 0.998
a
Values taken from Hiemstra et al. (2010).
b
Standard deviations are given for the fitted parameters.
J. Antelo et al. / Chemical Geology 410 (2015) 53 62 59
system (Fig. 2). This modelling scenario also yields adequate prediction The effect of the arsenate loading has been evaluated separately at
of calcium adsorption in the multi-component systems, which will be constant calcium concentration and constant pH (Fig. 3b). As expected,
discussed in the following section along with the effect that calcium increasing the As/Ca ratio at constant calcium loading produced a higher
ions have on oxyanion adsorption. The lower calcium complexation adsorption of arsenate on ferrihydrite. The observed results for the ad-
constants obtained for ferrihydrite, compared with those proposed by sorption of arsenate are sensitive to the calcium loading in the system.
Hiemstra et al. (2010) for goethite, are consistent with the differences The ferrihydrite surface is close to saturation when the calcium loading
observed for the arsenate or phosphate constants in both iron oxides. is relatively low and the maximum adsorption capacity was ~0.45 mmol
AsO4/l, whereas it is evident that arsenate did not reach saturation
when the calcium loading is relatively high and the maximum adsorp-
3.4. Arsenate and calcium adsorption in multi-component systems tion capacity of the ferrihydrite increased (>0.55 mmol AsO4/l). The
good performance of the CD model, which is able to predict the en-
The arsenate adsorption results for the multi-component systems hancement of arsenate adsorption on ferrihydrite in the presence of cal-
are shown in Fig. 3 for a 0.6 mM arsenate loading, along with the predic- cium, is discussed below.
tions of the CD model for the different cases studied. Results obtained Fig. 4a and b shows the equilibrium concentration of calcium ions in
for two other arsenate loadings, 0.1 and 0.3 mM, are shown in Fig. S8 the presence of arsenate and ferrihydrite. Comparison of the results in
(Supplementary data). For purposes of comparison, the arsenate ad- the absence and in the presence of arsenate, analysed at two different
sorption level in the absence of calcium is also shown. The presence of calcium loadings (2 and 6 mM), indicated that calcium adsorption is en-
calcium ions enhanced the adsorption of arsenate ions on ferrihydrite hanced at pH > 7.5 when arsenate is present in the system. Initially, in
at relatively high pH values (Fig. 3a). At pH < 8.0 no influence of calcium the absence of arsenate, calcium adsorption only took place at
was observed, due to the weak adsorption of this cation. The degree to pH > 8.5. Binding of phosphate or arsenate ions on iron mineral surfaces
which the adsorption is enhanced depends on the concentration of cal- strongly decreases the surface charge and shifts the isoelectric point to
cium in the system, i.e. at pH ~10.2 a concentration of 0.3 mM slightly lower pH values (Antelo et al., 2005). Therefore, adsorption of arsenate
affects the arsenate adsorption, whereas a concentration of 6 mM in- ions onto the ferrihydrite nanoparticles stimulated changes in the elec-
creased the arsenate adsorption by up to 30%. Adsorbed calcium ions trostatic forces at the solid/solution interface and favoured the adsorp-
are positively charged and will promote the adsorption of the negatively tion of the positively charged calcium ions.
charged forms of the arsenate ions through changes in the electrostatic The mutual effect of arsenate and calcium ions was simulated using
potential. As the calcium loading on ferrihydrite increases, the repulsive the CD model, with the complexation parameters previously obtained
electrostatic interactions between the negatively charged surface (at in the single-component experiments (Tables 2 and 3). As stated
relatively high pH values) and the arsenate ions decrease.
7.0
0.7 a
6.0
a
6 mM Ca
0.6
5.0
4.0
0.5
3.0
2 mM Ca
no Ca
0.4
2.0
0.3 mM Ca
0.7 mM Ca
1.0
0.3 1 mM Ca
2 mM Ca
0.0
6 mM Ca
0.2
6 7 8 9 10 11
5 6 7 8 9 10 11
pH
pH
1.2
0.3 mM Ca
0.6
b
b
0.7 mM Ca
1.0
0.5
1mMCa
0.8
0.6 mM AsO4
0.4
0.6
0.3
0.4
0.2
0.2
0.1 [Ca] = 6 mM
[Ca] = 0.7 mM 0.0
0.0
6 7 8 9 10 11
0.0 0.3 0.6 0.9 1.2
pH
As/Ca
Fig. 4. Calcium behaviour in the ferrihydrite AsO4 Ca multi-component system. a) Equi-
Fig. 3. Arsenate adsorption experiments in the ferrihydrite AsO4 Ca multi-component librium calcium concentrations in solution in the presence (open symbols) and absence
system. a) Effect of calcium in the arsenate adsorption to ferrihydrite; b) effect of ar- (closed symbols) of arsenate; b) Effect of the pH on the equilibrium calcium concen-
senate loading at two fixed calcium concentrations. Lines correspond to the CD model tration (lower loadings) in the presence of arsenate. Lines correspond to the CD
predictions. model predictions.
sol
4 ads
[Ca]
(mmol/l)
[AsO ]
(mmol/l)
sol
4 ads
[Ca]
(mmol/l)
[AsO ]
(mmol/l)
60 J. Antelo et al. / Chemical Geology 410 (2015) 53 62
0.7
above (Section 3.2), consideration of outer-sphere complexes for calci-
a
um, along with inner-sphere complexes, introduced an excess of posi-
tive charge on the 1-plane and led to overestimation of the adsorption
0.6
of arsenate in the multi-component systems. The optimized complexa-
tion parameters obtained for calcium in modelling scenario II, which
0.5
excluded the formation of outer-sphere complexes, led to successful
predictions of both the arsenate and calcium adsorption in the multi-
no Ca
0.4
component systems. In order to consider in the modelling calculations
0.3 mM Ca
the possible formation of insoluble arsenic-calcium mineral phases,
0.7 mM Ca
0.3 1mMCa
solubility products, recently compiled by Nordstrom et al. (2014), for
2mMCa
different mineral phases (haidingerite, CaHAsO4·H2O, log Ksp =  4.79;
6mMCa
tricalcium-arsenate, Ca3(AsO4)2·3H2O, log Ksp =  21.14; johnbaumite,
0.2
Ca5(AsO4)3OH, log Ksp =  40.12) were included in these simulations.
5 6 7 8 9 10 11
Saturation indices were calculated for these mineral phases, and the re-
pH
sults indicated that under the conditions analysed the multi-component
systems were undersaturated and therefore no precipitation was
7.0
expected. Only at pH > 9 and for the highest Ca/AsO4 ratio the effect of
b
precipitation was slightly significant, although its contribution to the ar-
6.0
senate retention is lower than 10%. The adequate modelling predictions
6 mM Ca
showed that the electrostatic changes produced in the solid/solution
5.0
interface, when both ions were present, had a major role in the mutual
4.0
effect produced between arsenate and calcium. Nevertheless, the possi-
ble formation of ternary complexes cannot be ruled out. Finally, model-
3.0
ling simulations of the multi-component systems were also conducted 2 mM Ca
with the generalized two-layer model, using the complexation parame- 2.0
ters proposed by Dzombak and Morel (1990). Although the simulations
1.0
predicted an arsenate adsorption enhancement in the presence of calci-
um ions, the amount of arsenate adsorbed was not accurately predicted
0.0
for the entire range of calcium loadings (Fig. S9, Supplementary data).
6 7 8 9 10 11
pH
3.5. Phosphate and calcium adsorption in multi-component systems
The mutual effect produced by phosphate and calcium on their ad-
1.2
0.3 mM Ca
sorption is shown in Fig. 5. Qualitatively, this effect is similar to that c
0.7 mM Ca
discussed above for the arsenate calcium multi-component system, 1.0
i.e. it is a synergetic interaction. The presence of calcium enhanced phos- 1mMCa
0.8
phate adsorption on ferrihydrite nanoparticles and vice versa, although
0.6 mM PO4
the effect of calcium seems to be greater for phosphate than for arse-
0.6
nate. This result is consistent with the solution chemistry of both
oxyanions, since slightly higher Ca ion-pair formation constants have
0.4
been reported for phosphate (Table S1, Supplementary data). At pH
~9.3, when the concentration of calcium present in the system is
0.2
1 mM, adsorption of phosphate and arsenate reached 83% and 74%, re-
spectively, whereas at pH ~10.2 the corresponding levels were 88%
0.0
and 66%. On the other hand, the concentration of calcium in solution
6 7 8 9 10 11
was slightly lower in the presence of phosphate in comparison with
pH
the system containing arsenate (Fig. S10 in the Supplementary data).
Initially, the enhanced adsorption was explained by the changes pro-
Fig. 5. Adsorption experiments in the ferrihydrite PO4 Ca multi-component system.
duced in the electrostatic forces at the solid/solution interface following
a) Effect of calcium in the phosphate adsorption to ferrihydrite; b) equilibrium calcium
the adsorption of phosphate and calcium ions. However, in contrast to
concentrations in solution in the presence (open symbols) and absence (closed symbols)
the successful modelling prediction of the arsenate calcium interac- of phosphate; c) effect of the pH on the equilibrium calcium concentration (lower
loadings) in the presence of phosphate. Lines correspond to the CD model predictions.
tions, the CD model underestimated the adsorption of phosphate in
the presence of calcium when the constants obtained in the single-
component systems were used (Fig. S11 in the Supplementary data).
Phosphate adsorption was only adequately described for the lower amount of phosphate immobilized (adsorbed and precipitated) in
calcium loading, indicating that changes in the electrostatic forces the system was predicted at the higher Ca/PO4 ratios (Fig. S12 in
were not the only mechanism responsible for the adsorption en- the Supplementary data). Considering the discrepancy that arises
hancement at intermediate to high calcium loadings. Precipitation from the assumption of the formation of phosphate calcium pre-
of insoluble mineral phases or the formation of ternary complexes cipitates, and taking into account the experimental conditions
at high pH may also cause the observed effects. Saturation indices (phosphate pre-equilibration periodreduces the presence of phos-
were calculated for the most insoluble phosphate calcium mineral phate in solution before calcium addition) and the kinetics of the
(hydroxyapatite, Ca5(PO4)3OH, log Ksp =  58.2) and the results in- hydroxyapatite formation, which requires the formation of the
dicated that precipitation may occur under the conditions analysed. more soluble tricalcium-phosphate (Ca3(PO4)2, log Ksp =  28.9)
However, if precipitation was included in the calculations no fur- as a precursor (Weng et al., 2011), the precipitation of these solid
ther improvement in the prediction of the experimental results phases was eventually ruled out. For this reason, ternary complexes
was achieved. On the contrary, a considerable reduction in the were considered in the modelling calculations. In the absence of
4 ads
[PO ]
(mmol/l)
sol
[Ca]
(mmol/l)
sol
[Ca]
(mmol/l)
J. Antelo et al. / Chemical Geology 410 (2015) 53 62 61
spectroscopic data, a monodentate ternary surface complex, `FeOPO3Ca, Adsorption decreased with increasing pH, due to the decrease in posi-
was proposed. This approach was used as a modelling exercise to assess tively charged surface sites at the solid/solution interface. The effect of
the effect that the formation of ternary complexes had on the adsorption. ionic strength was only significant at intermediate and high pH, indicat-
The surface reaction and stoichiometry for the proposed ternary surface ing the formation of inner-sphere surface complexes. The arsenate ad-
complex were as follows: sorption behaviour was successfully described with the CD model
under the different conditions studied. The arsenate surface complexes
0 1
a"FeOH-1=2 þ Hþþ PO4-3 þ Caþ2Ć! a"FeO-1=2þ"z PO3Ca"z þ H2O ð8Þ
were selected according to the spectroscopic and molecular data report-
ed in the literature. The model predictions indicated that the dominant
The charge distribution over the 0- and 1-plane was calculated by
surface species were bidentate complexes, although the presence of
combination of the actual values of the monodentate non-protonated
monodentate surface species could not be ruled out.
phosphate surface complex (FeOPO3; "z0 = +0.22, "z1 =  2.22) and
" Adsorption of calcium onto the ferrihydrite surface is relatively weak,
the inner-sphere monodentate calcium surface complex (FeOHCa;
showing similar behaviour to that observed for the adsorption of calci-
"z0 =+0.31, "z1 = +1.69). Since the complexation constants for
um and magnesium on goethite. Retention of calcium ions was only
the single-component systems had already been obtained, the only pa- relevant at pH > 8.5, which was the expected result taking into account
rameter fitted in this modelling exercise was the complexation constant
the relatively low hydrolysis constant for calcium ions and the relatively
of the ternary surface complex.
high PZC found for the ferrihydrite nanoparticles. The CD model ade-
The inclusion of the ternary complex in the modelling calculations
quately describes the calcium adsorption considering the same type of
yielded adequate prediction of the phosphate adsorption at the dif- surface complexes previously proposed for goethite. However, the pres-
ferent calcium loadings. The optimized complexation constant was
ence of outer-sphere complexes produced an excess charge in the elec-
log KFeOPO3Ca = + 20.5. Description of the calcium adsorption did
trostatic 1-plane, and then the adsorption of oxyanions in the multi-
not improve considerably, considering that rather good modelling
component systems was overestimated. Exclusion of these complexes
simulation was already obtained without the inclusion of the ter- not only improved the predictions in the multi-component systems,
nary surface complex. The surface speciation for the phosphate but also yielded an equally good prediction of the calcium adsorption
calcium multi-component system is shown in Fig. S13 and Fig.
in the single-component systems.
S14 (Supplementary data). The ternary complex is the dominant
" The presence of calcium ions enhanced the adsorption of both arsenate
phosphate surface species at intermediate to high pH (>8.0) for a
and phosphate on ferrihydrite. The degree to which calcium ions en-
calcium loading of 1 mM or higher. On the other hand, its contribu- hanced the oxyanion adsorption depended on the pH, calcium loading
tion to the calcium surface speciation is relevant at intermediate pH
and the nature of the oxyanion. A similar synergetic interaction was
values (7.0 9.0), while the dominant surface species at high pH
observed when the adsorption of calcium was studied in the presence
(>9.0) is the hydrolysed monodentate complex. Although the
of these oxyanions. The enhanced adsorption may be explained by
modelling simulations yielded good prediction of the experimental
changes in the electrostatic forces at the solid/solution interface,
data, this does not imply the existence of the ternary surface com- although the presence or the formation of ternary surface complexes
plex proposed. Another likely mechanism for the enhancement of
cannot be ruled out.
phosphate adsorption is surface precipitation, which cannot be en- " The mutual interaction between arsenate and calcium was successfully
tirely excluded as the formation of insoluble phases in the solution
described using the CD model, with the surface complexation constants
was. Concentration of the ionic species near the ferrihydrite surface
previously derived in the single-component systems. On the other
might be larger than concentrations in the bulk solution. This situation
hand, the phosphate calcium interactions were only successfully
might lead to conditions where surface precipitation is occurring. Sur- described at the lower calcium loading. At higher calcium loadings,
face precipitation may also account for the differences observed be- the formation of ternary complexes or surface precipitates may also
tween the phosphate and arsenate in the multi-component systems.
be involved in enhancing the adsorption of phosphate ions.
Solubility products for the phosphate calcium minerals are usually
lower than those for the analogous arsenate calcium minerals. There-
fore, in the surface region, oversaturation with respect to the solid- Acknowledgements
phases considered may be achieved with lower concentrations in the
This work was supported through the Ministerio de Educación y
case of phosphate compared with arsenate. Nevertheless, both surface
Ciencia under research project CTM2011-24985 and by the Xunta
precipitation and ternary surface complexation involved multilayer
complex formation and can be considered as similar mechanisms. Fu- Galicia under research project EM2013/040. The authors are grateful
to Pilar Bermejo and Paloma Herbello of the Department of Analytical
ture spectroscopic studies should be carried out to clarify the real nature
Chemistry, Nutrition and Bromatology of the University of Santiago de
and structure of ternary complexes or surface precipitates of phosphate
Compostela (USC) for the ICP-OES measurements. The authors also
and calcium at relatively high pH.
thank María Santiso of the Department of Soil Science and Agricultural
Chemistry for assistance in the AAS measurements. The associate editor,
4. Conclusions
Carla Koretsky, and four anonymous reviewers are gratefully acknowl-
edged for their feedback and constructive comments, which have great-
The results of the present study demonstrate that the presence of a
ly contributed to improve this manuscript.
major ion such as calcium enhances the adsorption of arsenate and
phosphate on ferrihydrite nanoparticles. Retention of these oxyanions
on mineral surfaces is important for controlling their transport and bio-
Appendix A. Supplementary data
availability in soil and aquatic systems. Improved knowledge of multi-
component systems is critical to enable prediction of the mobility of nu-
Supplementary data to this article can be found online at http://dx.
trients and contaminants in natural systems. On the basis of the results
doi.org/10.1016/j.chemgeo.2015.06.011.
of the experimental and modelling studies carried out, the following
conclusions were reached:
References
" Qualitatively, the adsorption of arsenate on ferrihydrite showed simi-
Antelo, J., Avena, M., Fiol, S., López, R., Arce, F., 2005. Effects of pH and ionic strength on
lar trends to those found for different ionic species (phosphate,
the adsorption of phosphate and arsenate at the goethite water interface. J. Colloid
arsenate, chromate& ) on mineral surfaces with relatively high PZC. Interface Sci. 285, 476 486.
62 J. Antelo et al. / Chemical Geology 410 (2015) 53 62
Antelo, J., Fiol, S., Pérez, C., MariÅ„o, S., Arce, F., Gondar, D., López, R., 2010. Analysis of Loring, J.S., Sandström, M.H., Norén, K., Persson, P., 2009. Rethinking arsenate coordina-
phosphate adsorption onto ferrihydrite using the CD-MUSIC model. J. Colloid tion at the surface of goethite. Chem. Eur. J. 15, 5063 5072.
Interface Sci. 347, 112 119. Morin, G., Ona-Nguema, G., Wang, Y., Menguy, N., Juillot, F., Proux, O., Guyot, F., Calas, G.,
Arai, Y., 2010. Effects of dissolved calcium on arsenate sorption at the kaolinite water Brown Jr., G.E., 2008. Extended X-ray absorption fine structure analysis of arsenite
interface. Soil Sci. 175, 207 213. and arsenate adsorption on maghemite. Environ. Sci. Technol. 42, 2361 2366.
Axe, L., Bunker, G.B., Anderson, P.R., Tyson, T.A., 1998. An XAFS analysis of strontium at the Murphy, J., Riley, J.P., 1962. Amodified single solution method for the determination of
hydrous ferric oxide surface. J. Colloid Interface Sci. 199, 44 52. phosphate in natural waters. Anal. Chim. Acta 27, 31 36.
Carabante, I., Grahn, M., Holmgren, A., Kumpiene, J., Hedlund, J., 2012. Influence of Neupane, G., Donahoe, R.J., Arai, Y., 2014. Kinetics of competitive adsorption/desorption of
Zn(II) on the adsorption of arsenate onto ferrihydrite. Environ. Sci. Technol. 46, arsenate and phosphate at the ferrihydrite water interface. Chem. Geol. 368, 31 38.
13152 13159. Nordstrom, D.K., 2002. Worldwide occurrence of arsenic in groundwater. Science 296,
Cornell, R.M., Schwertmann, U., 1996. The Iron Oxides: Structure, Properties, Reactions, 2143 2145.
Occurrences and Uses. VCH Publishers, New York. Nordstrom, D.K., Majzaln, J., Königsberger, E., 2014. Thermodynamic properties for
Das, S., Hendry, M.J., Essilfie-Dughan, J., 2011. Effects of adsorbed arsenate on the rate of arsenic minerals and aqueous species. In: Bowell, R.J., Alpers, C.N., Jamieson, H.E.,
transformation of 2-line ferrihydrite at pH 10. Environ. Sci. Technol. 45, 5557 5563. Nordstrom, D.K., Majzaln, J. (Eds.), Reviews in Mineralogy & Geochemistry  Arsenic:
Davis, J.A., Leckie, J.O., 1978. Surface ionization and complexation at the oxide/water Environmental Geochemistry, Mineralogy, and Microbiology. The Mineralogical Soci-
interface II. Surface properties of amorphous iron oxyhydroxide and adsorption of ety of America, Washington, DC, pp. 217 255.
metal ions. J. Colloid Interface Sci. 67, 90 107. Rahnemaie, R., Hiemstra, T., van Riemsdijk, W.H., 2006. Inner- and outer-sphere complex-
Dzombak, D., Morel, F.M.M., 1990. Surface Complexation Modelling: Hydrous Ferric ation of ions at the goethite solution interface. J. Colloid Interface Sci. 297, 379 388.
Oxide. John Wiley & Sons, New York. Rahnemaie, R., Hiemstra, T., van Riemsdijk, W.H., 2007. Geometry, charge distribution,
Gao, Y., Mucci, A., 2003. Individual and competitive adsorption of phosphate and arsenate and surface speciation of phosphate on goethite. Langmuir 23, 3680 3689.
on goethite in artificial seawater. Chem. Geol. 199, 91 109. Raven, K.P., Jain, A., Loeppert, R.H., 1998. Arsenite and arsenate adsorption on ferrihydrite:
Gustafsson, J.P., 2003. Modelling molybdate and tungstate adsorption to ferrihydrite. kinetics, equilibrium, and adsorption envelopes. Environ. Sci. Technol. 32, 344 349.
Chem. Geol. 200, 105 115. Rietra, R.P.J.J., Hiemstra, T., van Riemsdjik, W.H., 2001. Interaction between calcium and
Harrington, R., Hausner, D.B., Bhandari, N., Strongin, D.R., Chapman, K.W., Chupas, P.J., phosphate adsorption on goethite. Environ. Sci. Technol. 35, 3369 3374.
Middlemiss, D.S., Grey, C.P., Parise, J.B., 2010. Investigation of surface structures by Sahai, N., Carroll, S.A., Roberts, S., O'Day, P.A., 2000. X-ray absorption spectroscopy of
powder diffraction: a differential pair distribution function study on arsenate sorp- strontium(II) coordination: II. Sorption and precipitation at kaolinite, amorphous
tion on ferrihydrite. Inorg. Chem. 49, 325 330. silica, and goethite surfaces. J. Colloid Interface Sci. 222, 198 212.
Hiemstra, T., van Riemsdijk, W.H., 1996. A surface structural approach to iron adsorption: Salazar-Camacho, C., Villalobos, M., 2010. Goethite surface reactivity: III. Unifying arsenate
the charge distribution (CD) model. J. Colloid Interface Sci. 179, 488 508. adsorption behaviour through a variable crystal face  site density model. Geochim.
Hiemstra, T., van Riemsdijk, W.H., 2006. On the relationship between charge distribution, Cosmochim. Acta 74, 2257 2280.
surface hydration, and the structure of the interface of metal hydroxides. J. Colloid Sherman, D.M., Randall, S.R., 2003. Surface complexation of arsenic(V) to iron(III) (hydr)
Interface Sci. 301, 1 18. oxides: structural mechanism from ab initio molecular geometries and EXAFS spec-
Hiemstra, T., van Riemsdijk, W.H., 2009. A surface structural model for ferrihydrite: I. Sites troscopy. Geochim. Cosmochim. Acta 67, 4223 4230.
related to primary charge, molar mass, and mass density. Geochim. Cosmochim. Acta Smedley, P.L., Kinniburgh, D.G., 2002. A review of the source, behaviour and distribution
73, 4423 4436. of arsenic in natural waters. Appl. Geochem. 17, 517 568.
Hiemstra, T., van Riemsdijk, W.H., Rossberg, A., Ulrich, K.U., 2009. A surface structural Stachowicz, M., Hiemstra, T., van Riemsdijk, W.H., 2006. Surface speciation of As(III) and
model for ferrihydrite: II Adsorption of uranyl and carbonate. Geochim. Cosmochim. As(V) in relation to charge distribution. J. Colloid Interface Sci. 302, 62 75.
Acta 73, 4437 4451. Stachowicz, M., Hiemstra, T., van Riemsdijk, W.H., 2008. Multi-competitive interaction of
Hiemstra, T., Antelo, J., Rahnemaie, R., van Riemsdijk, W.H., 2010. Nanoparticles in natural As(III) and As(V) oxyanions with Ca2+, Mg2+, PO3-, and CO2- ions on goethite.
4 3
systems I: the effective reactive surface area of the natural oxide fraction in field sam- J. Colloid Interface Sci. 320, 400 414.
ples. Geochim. Cosmochim. Acta 74, 41 58. Swedlund, P.J., Webter, J.G., Miskelly, G.M., 2003. The effect of SO4 on the ferrihydrite ad-
Hofmann, A., van Beinum, W., Meeussen, J.C.L., Kretzschmar, R., 2005. Sorption kinetics of sorption of Co, Pb, and Cd: ternary complexes and site heterogeneity. Appl. Geochem.
strontium in porous hydrous ferric oxide aggregates II. Comparison of experimental 18, 1671 1689.
results and model predictions. J. Colloid Interface Sci. 283, 29 40. Tiberg, C., Sjöstedt, C., Persson, I., Gustafsson, J.P., 2013. Phosphate effects on copper(II)
Jambor, J.L., Dutrizac, J.E., 1998. Occurrence and constitution of natural and synthetic and lead(II) sorption to ferrihydrite. Geochim. Cosmochim. Acta 120, 140 157.
ferrihydrite, a widespread iron oxyhydroxide. Chem. Rev. 98, 2549 2585. Villalobos, M., Antelo, J., 2011. Aunified surface structural model for ferrihydrite: proton
Kanematsu, M., Young, T.M., Fukushi, K., Green, P.G., Darby, J.L., 2010. Extended triple charge, electrolyte binding, and arsenate adsorption. Rev. Int. Contam. Ambient. 27,
layer modeling of arsenate and phosphate adsorption on a goethite-based granular 139 151.
porous adsorbent. Environ. Sci. Technol. 44, 3388 3394. Wang, X., Li, W., Harrington, R., Liu, F., Parise, J.B., Feng, X., Sparks, D.L., 2013. Effect of fer-
Kanematsu, M., Young, T.M., Fukushi, K., Green, P.G., Darby, J.L., 2013. Arsenic(III, rihydrite crystallite size on phosphate adsorption reactivity. Environ. Sci. Technol. 47,
V) adsorption on a goethite-based adsorbent in the presence of major co-existing 10322 10331.
ions: modeling competitive adsorption consistent with spectroscopic and molecular Waychunas, G.A., Rea, B.A., Fuller, C.C., Davis, J.A., 1993. Surface chemistry of ferrihydrite:
evidence. Geochim. Cosmochim. Acta 106, 404 428. part 1. EXAFS studies of the geometry of coprecipitated and adsorbed arsenate.
Keizer, M.G., van Riemsdijk, W.H., 1998. ECOSAT: Equilibrium Calculation of Speciation Geochim. Cosmochim. Acta 57, 2251 2269.
and Transport. Technical Report, Department Soil Science and Plant Nutrition. Weng, L., Vega, F.A., van Riemsdijk, W.H., 2011. Competitive and synergistic effects in pH
Wageningen Agricultural University, Wageningen, The Netherlands. dependent phosphate adsorption in soils: LCD modeling. Environ. Sci. Tecnol. 45,
Kinniburgh, D.G., 1993. FIT: user guide. Technical Report WD/93/23. British Geological 8420 8428.
Survey, Keyworth, UK. WHO, 2011. Guidelines for Drinking-water Quality. 4th ed. World Health Organization,
Klapper, H., 1991. Control of Eutrophication in Inland Waters. Ellis Horwood, Chichester. Geneva.
Lee, G., Bigham, J.M., Faure, G., 2002. Removal of trace metals by coprecipitation with Fe, Zhu, M., Northrup, P., Shi, C., Billinge, S.J.L., Sparks, D.L., Waychunas, G.A., 2014. Structure
Al and Mn from natural waters contaminated with acid mine drainage in the of sulfate adsorption complexes on ferrihydrite. Environ. Sci. Technol. Lett. 1, 97 101.
Ducktown Mining District Tennessee. Appl. Geochem. 17, 569 581.
Lenoble, V., Deluchat, V., Serpaud, B., Bollinger, J.C., 2003. Arsenite oxidation and arsenate
determination by the molybdene blue method. Talanta 61, 267 276.