Published: September 08, 2011

r 2011 American Chemical Society 12930 dx.doi.org/10.1021/la201775c | Langmuir 2011, 27, 12930–12937

ARTICLE

pubs.acs.org/Langmuir

Ionic Strength Effects on Silicic Acid (H4SiO4) Sorption andOligomerization on an Iron Oxide Surface: An Interesting Interplaybetween Electrostatic and Chemical ForcesRossuriati Dol Hamid,†,‡ Peter J. Swedlund,*,† Yantao Song,† and Gordon M. Miskelly†

†School of Chemical Sciences, University of Auckland, Private Bag, 92019 Auckland, New Zealand‡School of Chemistry and Environmental Studies, Faculty of Applied Sciences, Universiti Teknologi MARA, 40450 Shah Alam,Selangor, Malaysia

1. INTRODUCTION

Surface charge is an important feature of metal oxide/aqueousinterfaces. Amphoteric groups on metal oxide surfaces developsurface charge through sorption or desorption of protons1 andother ions2,3 from solution. The chemistry of charged species atthe interface will be influenced by the electrostatic potential, andinsights into surface reactions can be obtained by studying howthese systems are affected by changes in ionic strength.4 Forexample, anions have been considered to be either electrostati-cally (outer-sphere) or covalently (inner-sphere) bound to oxidesurfaces depending on whether they can be displaced by a rise inionic strength.3,5 Spectroscopic studies of inorganic anionsadsorbed on oxide surfaces indicate that the situation is morecomplex, and for many anions both electrostatic and covalentcomplexes are present. For example, in situ attenuated totalreflectance (ATR) IR spectra of SO4

2- adsorbed on iron oxidegoethite indicate the presence of an electrostatic complex withdistorted tetrahedral symmetry at pH > 6, while at lower pH acovalent complex with symmetry of C2v or lower is alsoobserved.6 The proportion of adsorbed SO4

2- present as inner-sphere complexes increased with increasing ionic strength.6

Interestingly, even inorganic ligands with an overall neutral

charge, such as H3BO3 and H3AsO3, can form electrostatic com-plexes at metal oxide surfaces, and two rationales for this havebeen proposed. Either the electropositive central B or As atomhas been considered to act as a Lewis acid7�9 or the deprotonatedligand considered to act as a Lewis base.10 Spectroscopic studiesof the neutral ligand silicic acid (H4SiO4) adsorbed on iron oxidesurfaces have shown that electrostatic surface complexes are notformed11,12 which is chemically reasonable, since Si in H4SiO4 is4-coordinate compared to the 3-coordinate central atoms inH3BO3 and H3AsO3.

Many properties of iron oxides are influenced by the presenceof silicates including mineralogy,13 morphology,14 surface charge,15

and the availability of surface reactive sites.16 For these reasons,the interactions of silicates on iron oxide surfaces are important innumerous systems including catalytic supports,17 corrosion science,14

wastewater treatment,18 and many natural systems.19 The adsorp-tion of H4SiO4 has been studied on several iron oxide phasesincluding ferric precipitates,20,21 goethite,22,23 ferrihydrite,11,12,16

Received: May 11, 2011Revised: September 6, 2011

ABSTRACT:The effect of ionic strength on reactions at aqueous interfaces canprovide insights into the nature of the chemistry involved. The adsorption ofH4SiO4 on iron oxides at low surface silicate concentration (ΓSi) formsmonomeric silicate complexes with Fe�O�Si linkages, but as ΓSi increasessilicate oligomers with Si�O�Si linkages become increasingly prevalent. In thispaper, the effect of ionic strength (I) on both ΓSi and the extent of silicateoligomerization on the ferrihydrite surface is determined at pH 4, 7, and 10,where the surface is, respectively, positive, nearly neutral, and negativelycharged. At pH 4, an increase in ionic strength causes ΓSi to decrease at a given H4SiO4 solution concentration, while theproportion of oligomers on the surface at a given ΓSi increases. At pH 10, the opposite is observed; ΓSi increases as I increases, whilethe proportion of surface oligomers at a given ΓSi decreases. Ionic strength has only a small effect on the surface chemistry of H4SiO4

at pH 7, but at low ΓSi this effect is in the direction observed at pH 4 while at high ΓSi the effect is in the direction observed at pH 10.The pHwhere the surface has zero charge decreases from≈8 to 6 asΓSi increases so that the surface potential (Ψ) is positive at pH 4for all ΓSi and at pH 7 with low ΓSi. Likewise,Ψ < 0 at pH 10 for all ΓSi and at pH 7 with high ΓSi. The diffuse layer model is used tounravel the complex and subtle interactions between surface potential (Ψ) and chemical parameters that influence interfacial silicatechemistry. This analysis reveals that the decrease in the absolute value ofΨ as I increases causes ΓSi to decrease or increase whereΨis, respectively, positive or negative. Therefore, at a given ΓSi, the solution H4SiO4 concentration changes with I, and becauseoligomerization has a higher H4SiO4 stoichiometry coefficient thanmonomer adsorption, this results in the observed dependence ofthe extent of silicate oligomerization on I.

12931 dx.doi.org/10.1021/la201775c |Langmuir 2011, 27, 12930–12937

Langmuir ARTICLE

magnetite,23,24 hematite,23 andmaghemite.25 The adsorbedH4SiO4

is considered to be attached to the iron oxide surface as a bidentatecomplex in which an SiO4 tetrahedron shares corners with twoadjacent edge-sharing Fe octahedra. This structural model wasbased on extended X-ray absorption fine structure (EXAFS)spectroscopy,26 chemical modeling of the surface,22 and ATR-IR.11,12 The ATR-IR spectrum of H4SiO4 in solution27 has oneSi�O stretching mode, ν(Si�O), at 937 cm�1. Polymerizationlimits the concentration of monomeric H4SiO4 in solution to≈1.8 mM28 at 25 �C and produces a three-dimensional polymerphase, SiO2(am), with ν(Si�O) at 1100 cm�1.29,30 ATR-IRstudies of H4SiO4 adsorbing on ferrihydrite at low silicate surfaceconcentration (termed ΓSi and given here in units of moles ofadsorbed Si per mole of Fe) typically have ν(Si�O) at≈940 cm�1 characteristic of monomeric silicates with nobridging Si�O�Si bonds. As ΓSi increases, the position of theSi�O stretch shifts to≈1010 cm�1, which indicates formation ofbridging Si�O�Si linkages in discrete silicate oligomers andthis species dominates the surface at high ΓSi.

11,12 Only a smallamount of a three-dimensional SiO2(am) polymer phase withν(Si�O) at ≈1100 cm�1 is formed on the ferrihydrite surfacewhen the solution [H4SiO4] is below saturation with respect toSiO2(am). The IR for H4SiO4 adsorption and oligomerization onmagnetite24 show similar spectral features to those on ferrihy-drite discussed above. A structural model for the interfacialoligomerization has been proposed whereby linear trimers areformed by the insertion of a solutionH4SiO4 between sufficientlyclose adjacent adsorbed monomers (Figure 1).11

Swedlund et al.31 explored changes to H4SiO4 sorptionisotherms on ferrihydrite and the quantified silicate oligomeriza-tion as a function of pH and ΓSi and a surface complexationmodel was used to interpret the changes in surface chemistrywith pH and ΓSi.

31 The model involves reactions occurring on ageneric surface site (tFeOH), and the surface charge is con-sidered to reside in one plane. The surface potential (Ψ) isrelated to surface charge by electric double-layer theory and theequilibrium constant for the formation of a surface species with acharge of “n” includes a columbic term enFΨ/RT (where F isFaraday’s constant, R is the gas constant, and T is temperature inKelvin). Equation 1 gives an example of a surface reaction forwhich the expression for the equilibrium constant is given in eq 2([X] = molar concentration, and γX is the activity coefficient forspecies X) to illustrate the contribution of the chemical freeenergy and the electrostatic free energy to the equilibriumconstant. When I increases at a constant pH, the values of |Ψ|and γA will decrease. When the pH is below the point of zerocharge (PZC) of the oxide, thenΨ > 0 and the decrease in both

|Ψ| and γA will result in less adsorption of A2- at a higher I. Incontrast, at pH > PZC whenΨ < 0, a decrease in |Ψ| can meanthat there is more adsorption at higher I. This behavior has beenreported for the effect of I on PO4

3- andCO32- adsorption onto iron

oxides at pH conditions above and below the oxide’s PZC.32�34

tFeOH þ A2� þ Hþ S tFeA� þ H2O ð1Þ

K ¼ ½tFeA��½ tFeOH�ðγAÞ½A2��ðγHÞ½Hþ� e

�FΨ=RT ð2Þ

Swedlund et al.31 used reactions for silicate surface complexformation on ferrihydrite that were based on the structuralinterpretation of the ATR-IR spectra (Figure 1) but with asurface site stoichiometry coefficient of one for monomeradsorption for reasons discussed in section 3.3. The set ofequilibrium expressions for the surface reactions are given ineqs 3�8, and this model framework could describe the changesin the ATR-IR spectra and the distribution of Si between the solidand aqueous phases with a consistent set of model parameters.This model allowed for the unraveling of the complex interac-tions between the various parameters which determine the effectof pH on H4SiO4 adsorption and oligomerization.31

½ tFeOHþ2 � ¼ K1½ tFeOH�ðγHÞ½Hþ� e�FΨ=RT ð3Þ

½ tFeO�� ¼ K2½ tFeOH�ðγHÞ�1½Hþ��1 eFΨ=RT ð4Þ

½ tFeH3SiO40� ¼ K1, 0½H4SiO4�½ tFeOH� ð5Þ

½ tFeH2SiO4�1�

¼ K1,�1ðγHÞ�1½Hþ��1½H4SiO4�½ tFeOH� eFΨ=RT ð6Þ

½ tFe2H6Si3O100� ¼ K3, 0½H4SiO4�3½ tFeOH�2 ð7Þ

½ tFe2H4Si3O10�2�

¼ K3,�2ðγHÞ�2½Hþ��2½H4SiO4�3½ tFeOH�2 e2FΨ=RT

ð8ÞThe purpose of the current work is to measure the change in

the chemistry of H4SiO4 at the ferrihydrite surface in response tochanges in ionic strength. Insights into surface reactions can beobtained by studying how they are affected by changes in ionicstrength. In addition, the effect of ionic strength on the chemistryof oxide surfaces is important in many systems.35,36 This workexplores changes to theH4SiO4 isotherms (i.e., total amount of Siat the ferrihydrite surface) and also changes to the amount ofsilicate oligomerization on the ferrihydrite surface which isquantified from the ATR-IR spectra of the interfacial silicate.The effect of ionic strength on this system is examined at pH 4, 7,and 10 where the ferrihydrite surface charge is, respectively,positive, almost neutral, and negative. The beauty of this simplemodel approach is that it separates the chemical and electrostaticcomponents of the surface reactions while also minimizing thenumber of adjustable parameters. Some deficiencies of the modelare discussed in section 3.3.

2. METHODS

The methods used in this work were based on those from Swedlundet al.31 In brief, water used was 18.2 MΩ cm water that had also been

Figure 1. Model for the formation of surface oligomeric silicate specieson ferrihydrite. Tetrahedra are SiO4 and octahedra are FeO6. Modifiedfrom Swedlund et al.11

12932 dx.doi.org/10.1021/la201775c |Langmuir 2011, 27, 12930–12937

Langmuir ARTICLE

distilled, and all solutions and suspensions were kept under N2 toexclude CO2. The pH was adjusted with isothermally distilled HCl andlow carbonate NaOH. A carbonate-free stock solution of 1.66 mMH4SiO4 at pH 4 was prepared from amorphous SiO2 as previouslydescribed.12 Two-line ferrihydrite was prepared between 1 and 4 hbefore an experiment by raising the pH of a ferric nitrate (5 mM) andHNO3 (pH 2) solution to 8. The resulting suspensions were agedfor 1 h at pH ≈ 8, and the pH was then raised to 11 for 0.5 h todesorb carbonate. The suspensions were then rinsed twice at pH 11(by centrifuging and decanting) and then diluted to the desiredFe concentration (≈2.5 mM) in 0.01 or 0.1 MNaCl. The surface areaof the ferrihydrite after freeze-drying (240 m2 g�1) was measuredby the BET N2 adsorption method and falls within the reported rangeof values.37

Adsorption isotherms of H4SiO4 on ferrihydrite suspensions at pH 4,7, and 10 in 0.01 M NaCl were measured with a reaction time of 5 days.Drift in pH was typically less than 0.2 pH units, and this was manuallycorrected each day. Previous kinetic studies have shown that H4SiO4

adsorption by ferrihydrite is rapid over the first 1�2 h and then the rateof adsorption slows such that there is little change in adsorption after≈2days.16,38 Pure ferrihydrite is metastable and transforms to a more stablephase, usually goethite, over a time scale of weeks at neutral pHor days atpH over ≈10. In the presence of H4SiO4, the transformation offerrihydrite is inhibited and ferrihydrite can be stable for manymonths.39

At pH 10, the silicate-free ferrihydrite used for the ATR-IR backgroundwas aged for only 1 day because IR bands corresponding to goethite at≈900 and 800 cm�1 were observed after 5 days. Goethite IR bands werenot observed in any other samples equilibrated for 5 days.While changesin particle morphology or crystallinity over the course of the experimentscannot be definitively precluded, both the spectroscopic and macro-scopic evidence suggests that the ferrihydrite particles are stable overthis time.After the reaction, the suspensions were centrifuged and the super-

natant filtered (0.4 μm). The solution Si concentration (Sisol) wasmeasured using the molybdenum blue method and the surface concen-tration of Si was determined by the difference in Sisol from the totaladded Si. The total Fe concentration in HCl digests of aliquots ofsuspension was measured colorometrically using KSCN. Ferrihydritepastes were collected by centrifugation of the suspensions and applied toa diamond ATR-IR crystal, and the IR spectra were measured andprocessed as previously described.31 The spectra were analyzed bymultivariate curve resolution with alternating least squares (MCR-ALS)in Matlab (Mathworks) using the constraints of non-negative concen-tration and non-negative spectra.40 MCR-ALS analysis was used todetermine the optimum combination of the previously reported monomer,oligomer, and polymer spectra11 to describe each measured spectrumwhich was used to quantify the different surface silicate species.The diffuse layer model (DLM) was used to provide insights into the

effects of ionic strength on the interfacial chemistry of H4SiO4. The reac-tions for species formation used the components H+, H4SiO4,tFeOH,and Ψ. The H4SiO4 pKa’s of 9.84 and 13.20

41 were used, and solutionsilicate oligomers were not included because their concentrations wereless than 0.1% SiT under all conditions as calculated with the parametersfrom Felmy et al.28 All modeling used the Dzombak and Morel37

parameters for the ferrihydrite surface area (600 m2 g�1), adsorptionsite density (0.2 mol (mol Fe)�1), protonation constant (logK1 = 7.29),and deprotonation constant (log K2 =�8.93). The log K1,0 and log K1,�1

values for H4SiO4 neutral and anionic monomer adsorption, respec-tively, were 3.56 and �3.12.31 The log K3,0 and log K3,�2 values forH4SiO4 neutral and anionic trimer adsorption, respectively, were 15.33and 2.02.31 These parameters were developed to describe H4SiO4

adsorption to ferrihydrite using the same reaction time as used in thecurrent study. Activity coefficients of solution species were calculatedwith the Davies equation.42

3. RESULTS AND DISCUSSION

The effect of ionic strength on H4SiO4 adsorption wasinvestigated on the macroscopic scale with adsorption isothermsand also at the molecular scale using ATR-IR spectroscopy. Thecombination of the macroscopic and spectroscopic data with theDLM modeling allows increased understanding of the interac-tions of the various parameters by which ionic strength influencesthe extent of H4SiO4 adsorption and the silicate speciation on theferrihydrite surface.3.1. The H4SiO4 Isotherms. The H4SiO4 isotherms on

ferrihydrite at pH of 4, 7, and 10 with I = 0.01 M are shown inFigure 2 together with the corresponding data measured at I =0.1 M.31 Using the model parameters of Swedlund et al.,31 theDLM correctly predicted the observed trends in ΓSi as a functionof I and pH, and this is discussed in detail in section 3.3. At bothionic strengths, at a given Sisol, ΓSi was highest at pH 10 and lowestat pH 4. Previous studies have shown that H4SiO4 adsorptiondecreases as the pH increases above 10,16,31 and this is consistentwith the observation that ligand sorption onto iron oxides isat a maximum near the ligand pKa.

43�46

An increase in ionic strength shifted the H4SiO4 adsorptionisotherms in opposite directions at pH 4 and pH 10. At pH 4 overthe whole isotherm, the value of ΓSi at I = 0.01 M was almostdouble ΓSi at I = 0.1 M. In contrast, at pH 10, the value of ΓSi waslower at I = 0.01 M than at I = 0.1 M over the whole isotherm.There was very little effect of I on the H4SiO4 adsorptionisotherm at pH 7. The effect of I on ΓSi correlates with theferrihydrite surface potential (Ψ). The PZC of pureferrihydrite47 occurs at pH ≈ 8, and the values of Ψ and σcalculated using the DLM (eqs 3 and 4) are shown as a functionof pH in Figure 3. At any pH, apart from the PZC, an increase in Icauses the absolute magnitude of Ψ (|Ψ|) to decrease. Incontrast, the absolute magnitude of σ increases as I increasesbecause the decrease in |Ψ| favors protonation of surface sites atpH < PZC and deprotonation at pH > PZC. The adsorption andoligomerization of H4SiO4 on ferrihydrite causes a decrease in

Figure 2. Measured (points) and modeled (lines) H4SiO4 adsorptionisotherms. Data from this work at I = 0.01 M (open symbols and dashedline) compared to data with I = 0.1 M from Swedlund et al.31 (solidsymbols and solid lines). (a) pH of 4 and 10, (b) pH of 7. Dashed ellipsesindicate samples for which the IR spectra are shown in Figures 6 and 7.

12933 dx.doi.org/10.1021/la201775c |Langmuir 2011, 27, 12930–12937

Langmuir ARTICLE

both the measured and the DLMmodeled surface PZC (Figure 4)because of the greater acidity of the tSi�OH group comparedto the tFeOH group. For example, as ΓSi increased from 0 to0.42, the measured PZC decreases from 8 to 5.7, and a PZC ofpH 7 occurred with ΓSi ≈ 0.1.47 Therefore, the ferrihydritesurface will be positively charged for all data at pH 4 and data atpH 7 with lowΓSi, while the surface will be negatively charged forall data at pH 10 and data at pH 7 with high ΓSi. From Figure 2, it isclear that at a given Sisol ifΨ> 0, thenΓSi is higher at lower Iwhile ifΨ < 0 then ΓSi is lower at lower I. In general, the effect of I onH4SiO4 adsorption over the pH range 4�10 is similar to thepreviously reported effect of I on PO4

3- and CO32- adsorption onto

iron oxides at pH conditions above and below the oxide’s PZC.32�34

The isotherm data are discussed here by considering primarilythe effect of I on the negatively charged surface Si species. At agiven Sisol, the effect of I on an isotherm is due to the dependenceof ΓSi onΨ and the availability of surface sites (tFeOH). WhenΨ is positive, a decrease in |Ψ| will result in a decrease in ΓSi asobserved at pH 4 when I increases. When Ψ is negative, adecrease in |Ψ| will result in an increase in ΓSi as observed at pH10 when I increases. The small effect of I on ΓSi at pH 7 isconsistent with the low |Ψ| close to the ferrihydrite PZC. Thereis still, however, a trend in adsorption as a function of I at pH = 7.At the lowest Sisol point on the isotherm at pH 7, there ismarginally less adsorption at higher I, while as Sisol increasedthere is increasingly more adsorption at higher I. These smallchanges are consistent with the decrease in the ferrihydrite PZCas ΓSi increases as shown in Figure 4.31,47 At low ΓSi, when thePZC is above pH 7, adsorption at pH 7 would be favored at lowerI, while at higher ΓSi, when the PZC is below pH 7, adsorption atpH 7 would be favored at higher I.

Two reasons account for the greater effect of I on ΓSi at pH 4than at pH 10. Because the PZC of ferrihydrite is≈8.0, the valueof |Ψ| is greater at pH 4 than at pH 10, causing I to have a greatereffect on ΓSi. In addition, adsorption is also influenced by theavailability of surface sites. At any pH, apart from the PZC, as Iincreases, there is a decrease in the concentration of neutralsurface sites available to react with H4SiO4; that is, H

+ and OH�

compete more strongly for surface sites as I increases. As Iincreases at pH 4, the decreased availability of surface sites andthe decreased |Ψ| both contribute to a decrease in ΓSi. Incontrast, at pH 10, as I increases the decreased availability ofsurface sites and the decreased |Ψ| have opposing effects on ΓSi

so that the overall effect of I on the H4SiO4 adsorption isothermis less at pH 10 compared to pH 4.3.2. The IR Spectra of Silicate on the Ferrihydrite Surface.

The ν(Si�O) region of the ATR-IR spectra for H4SiO4 adsorbedand oligomerized on the surface of ferrihydrite pastes at pH of 4and 10 with I = 0.01 M is shown in Figure 5 together with theoptimized spectra for the pure Si surface species from Swedlundet al.11 The measured IR intensity depends on the contactbetween the ATR crystal and the sample which is variable.Therefore, the spectra were scaled so that the area of theν(Si�O) region equaled the ΓSi. The spectra show the samefeatures as those measured for H4SiO4 on ferrihydrite pastes at0.1 M31 and those measured with ferrihydrite films deposited onan ATR-IR crystal that was mounted in a flow cell.11 Themaximum IR absorbance at the lowest ΓSi at pH of 4, 7, and10 occurred at ≈945 cm�1, and as ΓSi increased the position ofthe maximum shifted toward ≈1009 cm�1. The IR absorbancefeature at low ΓSi is due to the presence of monomeric adsorbed

Figure 4. Measured47 (points) and modeled31 (lines) PZC of ferrihy-drite as a function of ΓSi.

Figure 5. ATR-IR spectra of H4SiO4 adsorbed and oligomerized on thesurface of ferrihydrite pastes. The legend gives values for ΓSi in molSi(mol Fe)�1. (a) Measured spectra at pH 4 and I = 0.01 M, (b)measured spectra at pH 10 and I = 0.01 M, and (c) optimized spectra ofpure surface Si species from Swedlund et al.11

Figure 3. Modeled surface charge (a) and surface potential (b) for apure ferrihydrite as a function of pH and ionic strength.

12934 dx.doi.org/10.1021/la201775c |Langmuir 2011, 27, 12930–12937

Langmuir ARTICLE

silicates, while the spectral feature at 1009 cm�1 is due to surfaceoligomers which become more prevalent on the surface as ΓSi

increases. The IR spectrum at the highest ΓSi at pH 10 is almostthe same as the optimized spectrum of the pure oligomericcomponent.The IR spectra in the ν(Si�O) region (750�1250 cm�1)

were fitted with MCR-ALS using the spectra for the surfacespecies as determined from a previous study with ferrihydritefilms.11 All of the measured spectra were reasonably welldescribed by a linear combination of the spectra for the mono-meric and oligomeric silicate species, with a small amount of thepolymeric silicate spectrum present only at the highest ΓSi valuesat each pH. At the optimum fit of the matrix of all spectra, 99.4%of the variance in the matrix of spectra was explained and thelargest residual for each spectrum was between 5 and 10% of themaximum IR absorption. This residual is slightly higher than thatobtained in our previous in situ studies with ferrihydrite filmsdeposited on the ATR crystal because spectra obtained frompastes are more variable. Figures 6 and 7 show typical results atpH 4 and 10, respectively, where the spectra from this work at I =0.01M are compared with spectra for ferrihydrite pastes at I = 0.1M from Swedlund et al.31 that had a similar ΓSi. At pH 4, withΓSi≈ 0.058, the maxima of the IR spectra were similar at 967 and956 cm�1 for I of 0.1 and 0.01 M, respectively. The spectrum atI = 0.01 M was narrower and had less IR absorption in the regionof 1000 cm�1 compared to the IR spectrum at I = 0.1 M. At pH10, with ΓSi≈ 0.12 (Figure 7), the maxima of the IR spectra wereat 954 and 991 cm�1 for I of 0.1 and 0.01 M, respectively, and inthis case the spectrum at I = 0.1 M was narrower and had less IRabsorption in the region of 1000 cm�1 compared to the spectrumat I = 0.01 M. From the shape of the spectra, it is clear that anincrease in I caused an increase in the proportion of oligomer onthe ferrihydrite surface at pH 4. The opposite effect was observedat pH 10 where an increase in I caused a decrease in theproportion of oligomer. Therefore, both the position of theH4SiO4 isotherm and the degree of surface silicate oligomeriza-tion were shifted in opposite directions at pH 4 and 10 inresponse to a change in I.The proportions of the ν(Si�O) region of the IR spectra that

were due to the monomer component are shown as a function of

ΓSi, pH, and I in Figure 8 together with the model predictionswhich are discussed in section 3.3. The trends in surfacecomposition as a function of ΓSi, pH and I are clearly revealed.At each pH, the proportion of the IR spectra that was due tomonomeric silicate decreased as ΓSi increased. The remainingarea of each spectrum was due to oligomeric silicates, apart fromthe highestΓSi samples at pH 4 and 10 for which, respectively, 6%and 7% of the area of the spectra were due to polymeric silicates.In all cases, there are more polynuclear species at a higher total Siconcentration, and it is also evident that at a given ΓSi there is moreoligomerization at a lower pH. The effect of pH on oligomerizationwas discussed in detail previously,31 and can be rationalized basedon the stoichiometry coefficients for forming surface monomers

Figure 6. Measured (points) and fitted (lines) ATR-IR spectra ofH4SiO4 adsorbed and polymerized on the surface of ferrihydrite pastesat pH 4 with ΓSi ≈ 0.058.

Figure 7. Measured (points) and fitted (lines) ATR-IR spectra ofH4SiO4 adsorbed and polymerized on the surface of ferrihydrite pastesat pH 10 with ΓSi ≈ 0.12.

Figure 8. Measured (points) and modeled (lines) percent of the Siadspresent as monomeric species. Data from this work at I = 0.01 M (opensymbols and dashed line) compared to data with I = 0.1 M fromSwedlund et al.31 (solid symbols and solid lines). (a) pH of 4 and 10, (b)pH of 7. Model parameters from Swedlund et al.31 Dashed boxesindicate samples for which the IR spectra are shown in Figures 6 and 7.

12935 dx.doi.org/10.1021/la201775c |Langmuir 2011, 27, 12930–12937

Langmuir ARTICLE

and oligomers as given in eqs 5�8. At a ΓSi of ≈0.1 on theisotherms with I = 0.01 M (Figure 2a), the corresponding valueof Sisol at pH 10 (≈4.8� 10�5 M) is much lower than that at pH4 (≈2.0� 10�4 M) because H4SiO4 adsorption is greater at pH10 than at pH 4. Monomer adsorption depends on [H4SiO4]while oligomerization depends on [H4SiO4]

3, and it is the largeincrease in [H4SiO4]

3 as the pH decreases from 10 to 4 at a givenvalue of ΓSi that leads to the greater amounts of oligomerizationat pH 4 compared to pH 10.3.3. Insights into the Silicate on Ferrihydrite System

Provided by DLM. The DLM parameters for silicate adsorptionand oligomerization on ferrihydrite obtained by Swedlund et al.31

were derived fromH4SiO4 isotherms on ferrihydrite suspensionsand IR spectra of ferrihydrite pastes at pH 4, 7, and 10 with I =0.1 M. The variables in the model are the proton stoichiometrycoefficients and equilibrium constants for the formation ofadsorbed monomeric silicates and trimeric silicates. These wereoptimized in Swedlund et al.31 using the measured concentra-tions of H+, Sisol, Sitot, Fetot, and monomeric silicate in conjunc-tion with the data for the PZC as a function of ΓSi fromSchwertmann and Fechter.31,47 A polymeric SiO2(am) phasewas not included in the model because there was insufficientdata to constrain silicate polymerization reactions. It is clear fromFigures 2 and 8 that the model of Swedlund et al.31 correctlypredicted the trends in the effect of I on the H4SiO4 isothermsand the percent of the adsorbed silicate present as monomers atpH 4, 7, and 10. The model tends to overestimate the percent ofmonomeric silicate at high ΓSi especially at pH 10. At high ΓSi,more complex polymerization is occurring as indicated by thepresence of SiO2(am) polymers in the IR spectra. The reason forthe discrepancy at pH 10 is not known, but in solution theoligomerization of silicates is favored at high pH and it may bethat more complex polymerization occurs at pH 10 than at alower pH.Before discussing the DLM model details, it is necessary to

emphasize that this model describes macroscopic properties ofthe system without necessarily providing structural informationof the species involved. This has been previously discussed indetail,31,48,49 and the main model simplifications are mentionedbriefly here. The DLM equations employed describe ligandadsorption as being monodentate, that is, at a single site, whereasthe IR and EXAFS spectra suggest a bidentate and binuclearcomplex. This discrepancy in surface site stoichiometry occursbecause the DLM by necessity underestimates the total numberof adsorption sites. This arises because charges are considered aspoint charges and the DLM therefore calculates unrealisticsurface charges at high surface concentrations of chargedspecies.50 Second, the DLM adsorption reactions include specieswith different degrees of protonation while the IR spectra for thedifferent species were independent of pH between pH 4 and10.11 Blesa et al.51 suggested that the deprotonation implied fromthe DLM model may occur at adjacent surface hydroxyl groupsrather than at the actual ligand. Their suggestion was based onthe differences in DLM adsorption constants for ligand com-plexes with different protonation states.At a fixed pH, the model parameters that influence the

concentration of the monomeric surface silicate species are[tFeOH] and [H4SiO4] plus eFΨ/RT for the anionic species(eqs 5 and 6). The concentrations of the trimeric surface silicatespecies depend on [tFeOH]2, and [H4SiO4]

3 plus e2FΨ/RT forthe anionic species (eqs 7 and 8). To unravel the complexinteractions between these parameters, we have plotted the

modeled values as a function of I for systems with constant ΓSi

and pH (Figures 9 and 10). Because the absolute values rangeover 14 orders of magnitude, we have plotted each parametervalue divided by its value at I = 0.01 M. The slopes of these linesreflect both the direction and degree that each parameterinfluences silicate speciation as I increases from 0.01 to 0.1 M.The changes in these parameters at pH 4 with ΓSi ≈ 0.05 are

shown in Figure 9a. The IR spectra of samples close to theseconditions are shown in Figure 6, and the corresponding datapoints are also indicated in Figure 2 (isotherms) and Figure 8(% monomer vs ΓSi). As I increases from 0.01 to 0.1 M, the valueof [H4SiO4] increases by a factor of 3.4 predominantly because ofthe effect of Ψ on the position of the H4SiO4 isotherms aspreviously discussed. For this reason, the parameter that changes

Figure 9. (a) Change in DLMparameters affecting adsorbed H4SiO4 asI increases from 0.01 to 0.1 M at pH 4 with ΓSi of 0.05 modeled usingeqs 3�8. Actual values of the parameters at I = 0.01 M are given in thelegend. (b) Modeled speciation of surface Si species modeled usingeqs 3�8.

Figure 10. (a) Change in DLM parameters affecting adsorbed H4SiO4

as I increases from 0.01 to 0.1M at pH 10 withΓSi of 0.12 modeled usingeqs 3�8. Actual values of the parameters at I = 0.01 M are given in thelegend. (b) Modeled speciation of surface Si species modeled usingeqs 3�8.

12936 dx.doi.org/10.1021/la201775c |Langmuir 2011, 27, 12930–12937

Langmuir ARTICLE

most significantly from I = 0.01 to 0.1 M at constant ΓSi is[H4SiO4]

3 which increases by a factor of 40. It is clear fromFigure 9a that the combination of the large increase in [H4SiO4]with I and the higher H4SiO4 stoichiometry coefficient foroligomerization than monomer adsorption causes the observedincrease in oligomers at the higher I at pH 4.The parameter eFΨ/RT decreases by a factor of 3 as I increases

at pH 4, which represents the decrease in the favorable electro-static energy of forming negative surface species asΨ decreases.The charge of the surface species is analogous to a stoichiometrycoefficient for this term so that the oligomer anion with z =�2 ismore greatly affected than the monomer anion with z = �1.Therefore, the direct effect ofΨ on surface silicate chemistry is toinhibit oligomerization; however, this direct effect is exceeded bythe indirect effect of Ψ on [H4SiO4] because [H4SiO4] affectsthe neutral and the anionic oligomerization reaction and alsobecause, compared to the monomer, the oligomer stoichiometrycoefficient for [H4SiO4] is larger than that for eFΨ/RT. Themodeled species concentrations for this system are shown inFigure 9b on a linear scale to aid comparison with the plot ofpercent of monomeric silicate (Figure 8). At I = 0.01 M, the twoanionic species are dominant, but they decrease at higher I whilethe neutral trimer increases substantially, causing oligomeriza-tion to be favored at the higher I.The silicate�ferrihydrite system at pH 10 with ΓSi of 0.12 is

shown in Figure 10. As I increases from 0.01 to 0.1M, the value of[H4SiO4] decreases by a factor of 2.8 predominantly because ofthe effect of Ψ on the position of the H4SiO4 isotherms aspreviously discussed. The parameter that changes most signifi-cantly from I = 0.01 to 0.1 M is [H4SiO4]

3 which decreases by afactor of 21. Therefore, at pH 10, the situation is similar butopposite to that at pH 4 in that it is the combination of the largedecrease in [H4SiO4] with I and the higher H4SiO4 stoichiom-etry coefficient for oligomerization than monomer adsorptionwhich causes the observed decrease in oligomers at the higher I atpH 10. The parameter eFΨ/RT increases by a factor of 3.4 as Iincreases at pH 10, which represents the decrease in theunfavorable electrostatic energy of forming negative surfacespecies as |Ψ| decreases. Because the oligomer anion hasz = �2 it is more greatly affected than the monomer anion withz = �1. Therefore, the direct effect of Ψ on surface silicatechemistry is to increase oligomerization; however, this directeffect is exceeded by the indirect effect ofΨ on [H4SiO4] whichdecreases oligomerization at high pH for the same reasonsdiscussed in relation to the model system at pH 4. The modeledspecies concentrations for this system are shown in Figure 10b.The two deprotonated anionic species are dominant at all I at pH10, but as I increases the monomeric anion increases while theanionic trimer decreases.

4. CONCLUSIONS

Ionic strength (I) has a significant impact on H4SiO4 chem-istry at the ferrihydrite surface via its effect on surface potentialand the availability of surface sites. When the surface potential ispositive, an increase in ionic strength causes the amount of Si onthe oxide surface at a given solution [H4SiO4] to decrease butincreases the extent of silicate oligomerization on the oxidesurface at a given ΓSi. When the surface potential is negative,the opposite effects are observed. The diffuse layer model is auseful predictive tool for this system, and was used to unravelthe interactions between surface potential (Ψ) and chemical

parameters that influence interfacial silicate chemistry. Thedecrease in the absolute value ofΨ as I increases directly causesthe Si surface concentration to decrease or increase where Ψ is,respectively, positive or negative. Therefore, at a given Si surfaceconcentration, the solution H4SiO4 concentration changes withI, and because oligomerization has a higher H4SiO4 stoichiom-etry coefficient than monomer adsorption, this results in theobserved dependence of the proportion of oligomers on I.

’AUTHOR INFORMATION

Corresponding Author*E-mail: p.swedlund@auckland.ac.nz. Fax: + 64 9 373 7422.

’REFERENCES

(1) Bourikas, K.; Hiemstra, T.; Van Riemsdijk, W. H. Langmuir2001, 17, 749–756.

(2) Zhang, Z.; Fenter, P.; Cheng, L.; Sturchio, N. C.; Bedzyk, M. J.;Predota, M.; Bandura, A.; Kubicki, J. D.; Lvov, S. N.; Cummings, P. T.;Chialvo, A. A.; Ridley, M. K.; Benezeth, P.; Anovitz, L.; Palmer, D. A.;Machesky, M. L.; Wesolowski, D. J. Langmuir 2004, 20, 4954–4969.

(3) Hayes, K. F.; Papelis, C.; Leckie, J. O. J. Colloid Interface Sci. 1988,125, 717–726.

(4) Dukhin, A.; Dukhin, S.; Goetz, P. Langmuir 2005, 21, 9990–9997.(5) Adachi, K.; Mita, T.; Yamate, T.; Yamazaki, S.; Takechi, H.;

Watarai, H. Langmuir 2010, 26, 117–125.(6) Peak, D.; Ford, R. G.; Sparks, D. L. J. Colloid Interface Sci. 1999,

218, 289–299.(7) Arai, Y.; Elzinga, E. J.; Sparks, D. L. J. Colloid Interface Sci. 2001,

235, 80–88.(8) Peak, D.; Luther, G. W.; Sparks, D. L. Geochim. Cosmochim. Acta

2003, 67, 2551–2560.(9) Goldberg, S.; Johnston, C. T. J. Colloid Interface Sci. 2001,

234, 204–216.(10) Sverjensky, D. A.; Fukushi, K. Geochim. Cosmochim. Acta 2006,

70, 3778–3802.(11) Swedlund, P. J.; Miskelly, G. M.; McQuillan, A. J. Langmuir

2010, 26, 3394–3401.(12) Swedlund, P. J.; Miskelly, G. M.; McQuillan, A. J. Geochim.

Cosmochim. Acta 2009, 73, 4199–4214.(13) Jambor, J. L.; Dutrizac, J. E. Chem. Rev. 1998, 98, 2549–2585.(14) Kwon, S. K.; Kimijima, K.; Kanie, K.; Suzuki, S.; Muramatsu, A.;

Saito, M.; Shinoda, K.; Waseda, Y. Corros. Sci. 2007, 49, 2946–2961.(15) Anderson, P. R.; Benjamin, M. M. Environ. Sci. Technol. 1985,

19, 1048–1053.(16) Swedlund, P. J.; Webster, J. G.Water Res. 1999, 33, 3413–3422.(17) Corrias, A.; Ennas, G.; Mountjoy, G.; Paschina, G. Phys. Chem.

Chem. Phys. 2000, 2, 1045–1050.(18) Davis, C. C.; Chen, H.-W.; Edwards, M. Environ. Sci. Technol.

2002, 36, 582–587.(19) Swedlund, P. J.; Sivaloganathan, S.; Miskelly, G. M.; Waterhouse,

G. I. N. Chem. Geol. 2011, 285, 62–69.(20) Doelsch, E.; Stone, W. E. E.; Petit, S.; Masion, A.; Rose, J.;

Bottero, J.-Y.; Nahon, D. Langmuir 2001, 17, 1399–1405.(21) Doelsch, E.; Rose, J.; Masion, A.; Bottero, J. Y.; Nahon, D.;

Bertsch, P. M. Langmuir 2000, 16, 4726–4731.(22) Hiemstra, T.; Barnett, M. O.; van Riemsdijk, W. H. J. Colloid

Interface Sci. 2007, 310, 8–17.(23) Jordan, N.; Marmier, N.; Lomenech, C.; Giffaut, E.; Ehrhardt,

J. J. J. Colloid Interface Sci. 2007, 312, 224–229.(24) Yang, X.; Roonasi, P.; Holmgren, A. J. Colloid Interface Sci.

2008, 328, 41–47.(25) Yang, X.; Roonasi, P.; Jolster�a, R.; Holmgren, A.Colloids Surf., A

2009, 343, 24–29.(26) Pokrovski, G. S.; Schott, J.; Farges, F.; Hazemann, J.-L. Geochim.

Cosmochim. Acta 2003, 67, 3559–3573.

12937 dx.doi.org/10.1021/la201775c |Langmuir 2011, 27, 12930–12937

Langmuir ARTICLE

(27) McIntosh, G. J.; Swedlund, P. J.; Sohnel, T. Phys. Chem. Chem.Phys. 2011, 13, 2314–2322.(28) Felmy, A. R.; Cho, H.; Rustad, J. R.; Mason, M. J. J. Solution

Chem. 2001, 30, 509–525.(29) Smith-Palmer, T.; Lynch, B. M.; Roberts, C.; Lu, Y. J. Appl.

Spectrosc. 1991, 45, 1022–1025.(30) Kieffer, S. W. Rev. Geophys. 1979, 17, 20–34.(31) Swedlund, P. J.; Hamid, R. D.; Miskelly, G. M. J. Colloid

Interface Sci. 2010, 352, 149–157.(32) Barrow, N. J.; Bowden, J. W.; Posner, A. M.; Quirk, J. P. Aust. J.

Soil Res. 1980, 18, 395–404.(33) Geelhoed, J. S.; Hiemstra, T.; Van, R. W. H. Geochim. Cosmo-

chim. Acta 1997, 61, 2389–2396.(34) Villalobos, M.; Leckie, J. O. J. Colloid Interface Sci. 2001,

235, 15–32.(35) Spiteri, C.; Van Cappellen, P.; Regnier, P.Geochim. Cosmochim.

Acta 2008, 72, 3431–3445.(36) Millward, G. E.; Moore, R. M. Water Res. 1982, 16, 981–985.(37) Dzombak, D. A.; Morel, F. M. M. Surface Complexation Model-

ing: Hydrous Ferric Oxide; John Wiley & Sons: New York, 1990.(38) Swedlund, P. J.; Miskelly, G. M.; McQuillan, A. J. Langmuir

2010, 26, 3394–3401.(39) Cornell, R.M.; Giovanoli, R.; Schindler, P.W.Clays ClayMiner.

1987, 35, 21–28.(40) Tauler, R.; Smilde, A.; Kowalski, B. J. Chemom. 1995, 9, 31–58.(41) Gustafsson, J. P. Visual MINTEQ, 2.51; KTH (Royal Institute

of Technology): Stockholm, Sweden, 2006.(42) Davies, C. W. J. Chem. Soc. 1938, 2093–2098.(43) Hingston, F. J.; Atkinson, R. J.; Posner, A. M.; Quirk, J. P.

Nature 1967, 215, 1459–1461.(44) Hingston, F. J.; Posner, A. M.; Quirk, J. P. J. Soil Sci. 1972,

23, 177–191.(45) Luxton, T. P.; Tadanier, C. J.; Eick, M. J. Soil Sci. Soc. Am. J.

2006, 70, 204–214.(46) Waltham, C. A.; Eick, M. J. Soil Sci. Soc. Am. J. 2002, 66,

818–825.(47) Schwertmann, U.; Fechter, H. Clay Miner. 1982, 17, 471–476.(48) Swedlund, P. J.; Webster, J. G.; Miskelly, G. M. Geochim.

Cosmochim. Acta 2009, 73, 1548–1562.(49) Song, Y.; Swedlund, P. J.; Singhal, N. Environ. Sci. Technol.

2008, 42, 4008–4013.(50) Hayes, K. F.; Redden, G.; Ela, W.; Leckie, J. O. J. Colloid

Interface Sci. 1991, 142, 448–469.(51) Blesa, M. A.; Weisz, A. D.; Morando, P. J.; Salfity, J. A.; Magaz,

G. E.; Regazzoni, A. E. Coord. Chem. Rev. 2000, 196, 31–63.