The Co2+Adsorption Properties of Al2O3, Fe2O3, Fe3O4, TiO2, and MnO2Evaluated by Modeling with the Frumkin Isotherm

JOURNAL OF COLLOID AND INTERFACE SCIENCE 195, 192–202 (1997)ARTICLE NO. CS975148

The Co2/ Adsorption Properties of Al2O3, Fe2O3, Fe3O4, TiO2, andMnO2 Evaluated by Modeling with the Frumkin Isotherm

Hiroki Tamura,* ,1 Noriaki Katayama,† and Ryusaburo Furuichi*

*Laboratory of Materials Chemistry, Graduate School of Engineering, Hokkaido University, Sapporo 060, Japan;and †Asahikawa National College of Technology, Asahikawa 071, Japan

Received May 27, 1997; accepted August 19, 1997

the initial step in the incorporation (1–4). This radioactiveAdsorption of Co(II) ions on metal oxides is related to radioac- pollution causes an exposure problem to plant workers and

tive 60Co(II) (de)contamination of nuclear power plants, Co(II) results in large amounts of bulky radioactive waste when theion retention in soils as a plant nutrient, concentration of Co(II) plants are closed after their use. To prevent the radioactivein deep-sea manganese nodules, and other applications. Here, the

contamination of nuclear power plants, deionization of cool-amount of adsorbed Co(II) on metal oxides was measured as aing water is carried out, and here inorganic deionizers includ-function of the pH and concentration of Co(II) ions, and theing metal oxides have advantages over organic ones in heatadsorption properties of metal oxides were evaluated with a modeland radioactivity resistance (5) . Large amounts of Co(II)that considers simultaneous (1:1) and (1:2) exchange reactionsare also concentrated in deep-sea manganese nodules, pre-between Co2/ aqua ions and surface hydroxyl protons obeying the

Frumkin isotherm. The possibility of participation of mono- and sumably through adsorption on iron and manganese oxidespolynuclear Co(II) hydroxo complexes in the adsorption was ex- common in soils (6–9). The Co(II) retention in soils dueamined, and it was suggested that these species play no role under to the adsorption on soil metal oxides is related to soil fertil-the conditions here. From the model parameters, it was found that ity (10–14) and is also critically important when siting andthe Co2/ adsorption ability of metal oxides increases in the order designing repositories for radioactive waste and generalAl2O3 õ Fe2O3 õ TiO2 õ Fe3O4 õ MnO2, showing a good correla- metal waste, since in case of leakage the underground migra-tion to the electronegativity Xi of the lattice metal ions of the

tion of Co(II) ions may be stopped by such adsorption (15).oxides. The Co2/ adsorption was divided into two processes: (1)To solve problems described above, it is important todeprotonation of surface hydroxyl sites and (2) bonding of Co 2/

know the quantitative Co(II) adsorption properties of metalto the deprotonated sites with a negative charge. With increasingoxides. With knowledge of quantitative properties, it is pos-Xi , process 1 increases possibly due to the decrease in the donorsible to design Co(II) adsorption processes and to controlelectron density responsible for covalent bonds with protons, while

process 2 changes only slightly. It was suggested that process 2 is the processes under optimum operating conditions for partic-due to ionic bond formation (‘‘electrostatic contact adsorption’’ ) , ular purposes. A model of the Co(II) adsorption on metalwhich is independent of the donor electron density, and the corre- oxides may be useful for a quantitative evaluation of thelation of the overall process to Xi found here was ascribed to properties.process 1 above. q 1997 Academic Press A number of investigations have been made for modeling

Key Words: Co(II) ion; metal oxides; adsorption; electronegativ- the adsorption of Co(II) as well as other divalent heavyity; Frumkin isotherm; modeling.

metal ions on metal oxides. Kozawa (16) studied the ion-exchange adsorption of Zn(II) on manganese dioxide, andnamed the process ‘‘surface chelation.’’ The products of

INTRODUCTION heavy metal ion adsorption on metal oxides have been called‘‘surface complexes’’ by other investigators, and the Lang-

The Co(II) adsorption on metal oxides has been the object muir adsorption isotherm, derived from the mass-action law,of study in relation to many topics. In the cooling systems has been applied for surface complexation as an equilibriumof nuclear power plants, the radioactivity level increases model (1, 2, 5, 12, 13, 17–20). However, this isothermduring operation, mainly due to 60Co(II) incorporation in holds only in limited ranges even at very low surface cover-oxide scales formed by corrosion in turbines and piping,

ages, due to suppression of the reaction with increasing cov-where the adsorption of Co(II) on scale surfaces may be

erage. The limited range of application has been ascribed tothe effect of the electric surface potential due to electric

1 To whom correspondence should be addressed. charges resulting from surface (de)protonation and com-

1920021-9797/97 $25.00Copyright q 1997 by Academic PressAll rights of reproduction in any form reserved.

AID JCIS 5148 / 6g34$$$401 11-04-97 11:28:19 coidas

193Co(II) ADSORPTION PROPERTIES OF METAL OXIDES

plexation. To describe this, a layered distribution of electric The Co(II) retention in soils may also be assessed by themodel, which would serve soil science, soil fertility evalua-charges in the oxide/solution interphase (electrical double

layer, EDL) has been assumed, the relationship between the tion, waste disposal technology, and others.surface charge and potential derived, and an electric potentialterm introduced in the model equations (21–28). MODEL

The EDL models are being used in many investigations,but they show a number of shortcomings when applied to Stoichiometry‘‘real’’ oxide samples found in environments and industry:

The Co(II) adsorption on metal oxides in the presence of(a) To meet the assumed EDL structure of the interphase,a monovalent anion, X 0 , can be described by (1:1) andthe EDL models require a well-defined oxide/solution in-(1:2) exchange reactions between Co2/ aqua ions and sur-terphase and a smooth surface with uniform chemical com-face protons of acid hydroxyl sites, {OH(a):position and with regular crystal structure. Such an ideal

interphase, however, is seldom realized in real oxide sam-{OH(a) / Co2/ / X 0 S {OCo/rX 0 / H/ , [1]ples. (b) At the same time, the validity of the EDL models

cannot be substantiated experimentally because the surface 2{OH(a) / Co2/ S ({O)2Co / 2H/ . [2]potential cannot be measured directly or controlled exter-nally (21). As a result, the EDL parameter values obtained The reaction products, {OCo/ and ({O)2Co, are termedby fitting the models to experimental data do not necessarily surface complexes, and the charge on the {OCo/ is coun-have the significance assigned to them. (c) Nonelectrical terbalanced by delgo adsorption of X 0 .lateral interactions between interphase species, such as geo- These adsorption reactions were derived from the estab-metric, chemical, and other interactions, are not considered lished facts and observed behaviors summarized in the fol-in the EDL models. However, the steric (geometric) hin- lowing: (a) Under the experimental conditions of this inves-drance (29) is very likely to affect the adsorption of cations tigation, the predominant Co(II) species in solution wason metal oxides, since small surface protons are replaced by Co2/ aqua ions. Minor CoOH/ hydroxo complexes andthe larger cations. Further, the chemical lateral interaction polynuclear Co(II) hydroxo complexes were examined ashas been suggested as due to two-dimensional association adsorption species by evaluating adsorption affinities ofof fatty acids adsorbed on mercury to explain the deviation CoOH/ and analyzing the slopes of the adsorption isothermsfrom the Langmuir isotherm (30). measured in this investigation. The results indicated that

Elsewhere (4, 31), the adsorption of Co(II) on spherical neither the mono- nor the polynuclear hydroxo complexesmagnetite particles as a model corrosion product of iron and can be the adsorption species, and so the predominant Co2/

steels and on manganese dioxide as a battery material was aqua ions were assumed to be the adsorption species. Detailssuccessfully modeled by considering simultaneous (1:1) and will be described in the Results and Discussion. (b) The(1:2) exchange reactions with surface protons which obey number of protons released per Co2/ adsorbed was measuredthe Frumkin isotherm. The Frumkin isotherm assumes a ther- by titration to be in the range between 1 and 2 (7, 14,modynamic relation to express lateral interactions instead 17, 32), indicating simultaneous (1:1) and (1:2) exchangeof assuming idealized EDL structures of the oxide/solution reactions with surface protons. Similar stoichiometric ratiosinterphase and can embody all causes of suppression for have also been found for other divalent heavy metal ions andinterphases that are not well-defined. oxides (33–35). (c) The surface charge on oxide particles

The purpose of this investigation is to model the Co(II) measured by electrophoresis changed from negative to posi-adsorption on various real metal oxide samples with the tive as the amount of adsorbed Co2/ increased (charge rever-Frumkin isotherm. With such model equations and model sal) (2, 8, 9, 14, 36), indicating that the positive charge ofparameters, the extent of adsorption for a given metal oxide Co2/ is donated to the oxides in the formation of surfacecan be calculated as a function of the pH and concentration species (specific adsorption).of Co(II) . The model calculations may then serve a numberof applications: In the removal of Co(II) from the cooling Material Balancewater of nuclear power plants by adsorption, the most effec-

For the acid surface hydroxyl group sites, the sum of thetive metal oxide and optimum adsorption conditions candensities (mol m02) of sites unoccupied and occupied bybe evaluated (selected) theoretically. In the development ofthe (1:1) and (1:2) surface complexes is given bymetal oxide–base inorganic deionizers, the model parame-

ters can be used to characterize and evaluate the performance[{OH(a)] / [{OCo/rX 0] / 2[({O)2Co] Å Ns /2,of the deionizers in a quantitative manner. This characteriza-

tion research may help to establish details of the factorswhich govern the ion adsorption properties of metal oxides. [3]

AID JCIS 5148 / 6g34$$$401 11-04-97 11:28:19 coidas

194 TAMURA, KATAYAMA, AND FURUICHI

u1 and u2 are the coverages (dimensionless) of (1:1) and(1:2) surface complexes on the acid hydroxyl sites,

u1 Å [{OCo/rX0] / (Ns /2) , [8]

u2 Å 2[({O)2Co]/(Ns /2) , [9]

and f1 and f2 are proportionality constants (energy mol01) .The linear terms with respect to the coverages express theeffect of the above mentioned lateral interactions, identicalto the assumption for the Frumkin isotherm (38).

The interaction between the (1:1) and (1:2) exchangereactions and the effects from the ion pairs ({O0

rM/ andFIG. 1. Model of (1:1) and (1:2) surface complexes of divalent heavymetal ions, M2/ , formed by exchange with protons of acid-type surface {OH/

2 rX0) are omitted in Eqs. [4] and [5], and the reasonhydroxyl groups on metal oxide, {OH(a), in the presence of a monovalent for this will be discussed in the section ‘‘Examination ofanion, X0 . Model Assumptions’’ below.

The equilibrium conditions of the reactions, Eqs. [1] andwhere Ns is the density (mol m02) of all the hydroxyl sites[2] , are derived from Eqs. [4] and [5] at DG1 Å 0 andand half of Ns is the cation exchange capacity (37). In theDG2 Å 0,presence of a supporting electrolyte, MX, surface ion pairs,

{O0rM/ and {OH/

2 rX 0 , also form with the acid andbase hydroxyl sites, {OH(a) and {OH(b), as have been exp(0DG 71 /RT ) Å K 71 Å K1 exp(B1u1) , [10]modeled in our previous paper (37). Such ion pairs occupy

exp(0DG 72 /RT ) Å b 72 Å b2 exp(B2u2) , [11]at most a few percent of the ion-exchange capacity in thepH range 3–7, where the Co(II) adsorption data were ob-tained, and so {O0

rM/ is omitted in Eq. [3] . where K 71 and b 72 are equilibrium constants; K1 and b2 arethe concentration ratios, Q1 and Q2 , at equilibrium; and B1Equilibrium Conditionsand B2 are the lateral interaction constants (dimensionless) ,

As illustrated in Fig. 1, the (1:1) surface complex has a B1 Å f1 /RT and B2 Å f2 /RT . The equilibrium constants ofcharge of /1, the (1:2) complex is neutral overall, but the ion exchange reactions K71 and b 72 can be regarded as stabil-adsorbed ions and oxide ion sites may display dipole charac- ity constants of the (1:1) and (1:2) surface complexes. Here,teristics, resulting in local charges. Hence, there is repulsion the concentration ratios K1 and b2 decrease with increasingbetween surface complexes of the same kind due to the same u, because the products of the concentration ratios and thesign of the charges. Further, it is very likely that the larger exponential terms (K1 exp(B1u1) and b2 exp(B2u2)) mustsize of Co2/ which replaces H/ at hydroxyl sites causes remain unchanged as equilibrium constants. The exponentialsteric hindrance. All such lateral interactions, electrical, geo- terms, thus, express suppression of ion exchange by alreadymetrical, and others, suppress the ion exchange and will formed surface complexes and they are activity coefficientplay an increasingly larger role as the surface coverage of ratios of interphase species, as was described in the previousexchange sites increases, this suppression is apparent in the papers (37, 39). The values of B1 and B2 are measures ofmeasured adsorption isotherms in the ‘‘Adsorption Iso- the suppression effect due to the lateral interaction betweentherms’’ section below. The Gibbs free energy changes in

interphase species.the reactions, Eqs. [1] and [2], may then be described by

According to the model, the amount of adsorbed Co2/

per unit surface area, G (mol m02) , is given byDG1 Å DG 71 / RT ln Q1 / f1u1 , [4]

DG2 Å DG 72 / RT ln Q2 / f2u2 . [5] G Å [{OCo/rX0] / [({O)2Co]. [12]

Here, DG 71 and DG 72 are standard Gibbs free energy changesThe model equations [10] and [11] are mathematically(energy mol01) ; Q1 and Q2 are the concentration ratios

identical to the ‘‘constant capacitance’’ model (21–28),(mol01 m3 and m01 , respectively, by expressing the concen-which assumes a specific type of EDL. However, the modeltration of solution species in mol m03) ,presented here is based on the Frumkin isotherm and does

Q1 Å [{OCo/rX 0][H/] / ([{OH(a)] not specify any particular physical or molecular EDL fea-tures of the solid/solution interphase. Hence, the suppression1 [Co2/][X 0]) , [6]or lateral interaction constants B1 and B2 are not simply EDLparameters; they quantify phenomenological thermodynamicQ2 Å [({O)2Co][H/]2 / ([{OH(a)]2[Co2/]) . [7]

AID JCIS 5148 / 6g34$$$401 11-04-97 11:28:19 coidas


relations between species at the interphase and may be Determination of Model Parameterstermed ‘‘thermodynamic’’ parameters. As a result, the model

The model Eqs. [3] , [10], [11], and [12] were appliedembodies electrical, geometrical, chemical, and other lateralto the experimental data, and the model parameters wereinteractions.determined by the nonlinear least-squares method (41, 42).The procedure is as follows:

MATERIALS AND METHODS

(a) With K71 , b 72 , B1 , and B2 as parameters, initial valuesMaterials of the four parameters are assumed.

(b) The measured values of [Co2/] , [H/] , and [NO03 ]The manganese dioxide (MnO2) is the International Com-

corresponding to a measured adsorption density, Gmeas , aremon Sample IC12 for batteries, which consists of nearlyintroduced into Eqs. [10] and [11].spherical particles with diameters up to several tens of mm,

(c) The initial value of [{OH(a)] is assumed.and alumina (Al2O3) is the reference catalyst JRC-ALO-4,(d) Equations [10] and [11] are solved successively byboth as described in the previous paper (37). Magnetite

the Newton–Raphson method (43) to obtain [{OCo/(Fe3O4) was synthesized as spherical particles with a diame-rNO0

3 ] and [({O)2Co].ter of 1.0 mm by partial oxidation of iron(II) hydroxide(e) The values of [{OH(a)] , [{OCo/rNO0

3 ] , andgel with nitrate (4) . Titania (TiO2) aggregates of smaller[({O)2Co] are introduced into Eq. [3] , and the sum isparticles with diameters less than 1 mm and hematiteexpressed by j.(Fe2O3) , rectangular with dimensions several to 10 mm, are

(f ) If j is not equal to Ns /2, a secondary approximatedboth commercial reagents (Kanto Chemical Co., Tokyo).value of [{OH(a)] , [{OH(a)]new , is obtained fromThe oxide particle surfaces were cleaned with water for[{OH(a)]old 1 (Ns /2) /j.Al2O3 and Fe3O4, and for MnO2, TiO2, and Fe2O3, first with

(g) Steps d–f are repeated till convergence with Eq. [3]0.1 mol dm03 HNO3 and then water. The specific surfaceto obtain the densities of the respective surface species forarea, SBET , of the samples was measured by the BET methodthe parameter values assumed in step a.with N2 gas adsorption as described in the previous paper

(h) The calculated surface density, Gcalc , is obtained from(37). The density of surface hydroxyl groups on oxide sam-Eq. [12] as the sum of [{OCo/rNO0

3 ] and [({O)2Co],ples was measured by the Grignard reaction method (37) forand the square of the deviation, (Gcalc 0 Gmeas )

2v, is calcu-MnO2, TiO2, and Al2O3 and by the saturated deprotonationlated; here v is the weight of data, being G02

meas .method in alkali solution (4, 40) for Fe3O4 and Fe2O3. The(i) For all the Gmeas , steps b–h are repeated, and thesalt of Co(II) is a commercially available nitrate and was

sum of the squares of the deviation, S(Gcalc 0 Gmeas )2v, isused without further purification.

calculated.( j) Parameter values are systematically changed, andMeasurement of Co2/ Adsorption

steps b–i are repeated to find the optimum values of theparameters, those that give the least value of the error sum.Metal oxide powder samples were dispersed in Co(II)

solutions containing 0.1 mol dm03 NaNO3 with different pHNonlinear regression analysis was carried out with a pro-adjusted with HNO3 and NaOH solutions. In the pH and

gram by Yamaoka (44), and steps a–h were performed withCo(II) concentration ranges used here (pH °7.5, [Co(II)]a subroutine developed by us.° 1003 mol dm03) , Co2/ aqua ions are the predominant

species in solution. The suspensions were shaken at 257C toRESULTS AND DISCUSSIONreach equilibrium (3–12 h), after the separation of solid

and solution the equilibrium pH was measured, and the equi-Adsorption Isothermslibrium concentration of Co2/ in solution, [Co2/] (mol

dm03) , was determined by the radioactive tracer method forThe specific surface area SBET and the density of surface

magnetite (4) and by atomic absorption spectrometry forhydroxyl groups Ns of the metal oxides which were used to

the other oxides (31, 40). The amount of adsorbed Co2/define the surface densities of adsorbed Co2/ , G, and of free

per unit surface area, G(mol m02) , is given byadsorption sites, [{OH(a)] , are listed in Table 1. Theresults for TiO2 were obtained in this investigation, and the

G Å ([Co2/]T 0 [Co2/]) 1 V /S , [13] other values are from previous reports (4, 31, 40).Figures 2 to 6 show the Co2/ adsorption density G as a

function of the equilibrilum concentration [Co2/] and pHwhere [Co2/]T is the total concentration of Co2/ added, Vis the solution volume (dm3), and S is the surface area of (adsorption isotherms). Under the experimental conditions

here, the predominat Co(II) species is Co2/ aqua ion. Thethe oxides (m2) .

AID JCIS 5148 / 6g34$$$401 11-04-97 11:28:19 coidas


FIG. 4. Adsorption isotherm for Co2/ on TiO2 with different pH atFIG. 2. Adsorption isotherm for Co2/ on MnO2 with different pH at

ionic strength 0.1 M (NaNO3) and 257C (points, experimental; curves,ionic strength 0.1 M (NaNO3) and 257C (points, experimental; curves,

calculated) .calculated) .

interactions and is expressed by the Gibbs free energy linearlog G increases with log [Co2/] with a slope of one at veryterms with respect to u (Eqs. [4] and [5]) . The increase inlow G. The slope of one is the largest in the whole regionG with pH suggests an exchange of Co2/ with surface pro-of the isotherms, suggesting that the monomeric Co2/ intons (Eqs. [1] and [2]) , and no participation of CoOH/

solution reacts with the surface as assumed by Eqs. [1] andmononuclear hydroxo complexes will be described below.[2] ( if multinuclear Co(II) complexes were adsorbed, the

slope would be steeper than one depending on the degreeModel Parametersof polymerization as described below). With increasing G,

the slope starts to decrease even at surface densities several The optimum values of the model parameters (stabilityorders of magnitude lower than saturation, about 1005 mol and suppression constants) are shown in Table 2. The curvesm02 from the cation-exchange capacity for the (1:1) surface in the Figs. 2–6 are the calculated best fit of the model to thecomplexation (Ns /2) in Table 1. The slope decreases over measured data points, and the agreement is good, apparentlya wide [Co2/] range, indicating a suppression of adsorption showing the soundness of the model assumptions.with increasing G, and hence with increasing coverage u. In The stability constants K 71 and b 72 differ for different ox-our model, this suppression was assumed to be due to lateral ides, indicating different adsorption properties. The suppres-

sion constant B2 is larger than B1 for any oxide, and the(1:2) surface complexes suppress adsorption more strongly

FIG. 5. Adsorption isotherm for Co2/ on Fe2O3 with different pH atFIG. 3. Adsorption isotherm for Co2/ on Fe3O4 with different pH ationic strength 0.1 M (NaNO3) and 257C (points, experimental; curves, ionic strength 0.1 M (NaNO3) and 257C (points, experimental; curves,

calculated) .calculated) .

AID JCIS 5148 / 6g34$$$401 11-04-97 11:28:19 coidas


TABLE 2Values of Stability and Suppression Constants of Co2/ Surface

Complexes with Metal Oxides at Ionic Strength 0.1 mol dm03

(NaNO3) and 257C

Surfacecomplex Log K71

a Log b72b Log B1 Log B2

Co2/–MnO2 04.32 2.48 1.08 1.67Co2/–Fe3O4 04.98 01.02 0.80 1.47Co2/–TiO2 05.53 00.91 0.94 2.26Co2/–Fe2O3 06.32 00.75 1.62 2.23Co2/–Al2O3 06.62 03.02 1.88 2.47

a K 71 (mol01 m3).b b72 (m01).

FIG. 6. Adsorption isotherm for Co2/ on Al2O3 with different pH atously, and there is the interaction between the (1:1) andionic strength 0.1 M (NaNO3) and 257C (points, experimental; curves,

calculated) . (1:2) surface complexes (hetero interaction), i.e., a surfacecomplex suppresses or enhances the formation of the otherdue to electric, steric, or other reasons, and so affects the total

than the (1:1) complexes. An interpretation of these differ- adsorption density G. However, this interaction is ignored inences from a microscopic point of view will be given below the model as seen from Eqs. [4] and [5] and hence Eqs.in the sections ‘‘Co2/ Adsorption Properties and Electroneg- [10] and [11] due to the following reasons. In pH regionsativity of Metal Oxides’’ and ‘‘Lateral Interaction Properties away from the intersect (Fig. 7) , one exchange reactionof Metal Oxides.’’ dominates (Fig. 7 is shown in log scale) because of the

different pH dependence. The minor component affects theExamination of Model Assumptions

formation of the major component little, the effect of theWith the constants established here (Table 2), the densi- major component on the minor component may be large,

ties of (1:1) and (1:2) Co2/ surface complexes on Fe3O4 but the contribution of the minor component to G is little,(as an example) were back-calculated as a function of pHwith the model equations for an equilibrium Co2/ concentra-tion of 1007 mol dm03 (Fig. 7) . The sum of the densitiesof the (1:1) and (1:2) complexes gives the Co2/ adsorptiondensity G, and this is also shown in the figure together withthe experimental data points. Again, the calculated G curvereproduces the experimental data well. The (1:2) complexdensity has a higher pH dependence than that of (1:1) com-plexes due to the larger number of protons participating inthe (1:2) reaction. With increasing pH, the slopes of thecurves decrease, this is due to suppressive lateral interactionbetween surface complexes as assumed in the model.

The (1:1) and (1:2) surface complexes form simultane-

TABLE 1Specific Surface Area SBET and Hydroxyl Site Density Ns

of Metal Oxide Samples

Sample SBET (m2 g01) Ns (mol m02)

MnO2 (IC12) 80.0 2.35 1 1005 (Grignard)TiO2 (Kanto) 21.5 1.16 1 1005 (Grignard)Fe3O4 (spheres) 1.73 1.08 1 1005 (NaOH)

FIG. 7. Calculated densities of (1:1) and (1:2) Co2/ –Fe3O4 surfaceFe2O3 (Kanto) 15.9 2.38 1 1005 (NaOH)complexes as a function of pH for an equilibrium Co2/ concentration ofAl2O3 (ALO-4) 155 3.20 1 1005 (Grignard)1007 M at ionic strength 0.1 M (NaNO3) and 257C.

AID JCIS 5148 / 6g34$$$401 11-04-97 11:28:19 coidas


and so the effect of the hetero interaction on G needs not age ¢ ca. 10%) obtained with solutions containing 1002–1004 M Co(II) at pH 7.6–10.7.be considered.

If such multinuclear species were in equilibrium withHetero interaction between the surface complexes and theCo2/ in solultion under the conditions of this investigation,ion pairs ({O0

rM/ or {OH/2 rX0) was also ignored for

the log G vs log [Co2/] curves in Figs. 2–6 would have athe following considerations. These ion pairs cause electricslope larger than one, since the surface density of multinu-charges on the metal oxide surface due to their charged sites,clear species is proportional to its monomer concentrationbut they occupy at most a few percent of the ion-exchangewith the power of the number of monomers in the polymer.capacity as described above. Further, the Co2/ adsorptionHowever, the slope observed here is less than one, and thebehavior is little affected by the surface charge due to ionmultinuclear complexes do not seem to play a role in thepairs as described next. Figure 7 shows that the measuredadsorption.Co2/ adsorption density G increases steadily with pH across

The highest surface density in the adsorption isothermspH 6.5, the point of zero charge (pzc) of Fe3O4 (4, 37),measured here is about 1 mmol m02 (Figs. 2–6), correspond-and no dramatic change in the Co2/ adsorption density ising to the EXAFS study condition, but most surface densityobserved when the surface charge changes from positive todata scatter down to the minimum 10010 mol m02 (a surfacenegative at pzc. The other oxides also adsorbed Co2/ evencoverage of ca. 0.001%). The formation of multinuclearin the pH range where their surfaces are positively charged.complexes becomes sharply less favorable with the decreaseThis small effect on Co2/ adsorption of the surface chargein the concentration, which is evidenced for Co(II) also bydue to ion pairs may be ascribed to the very high adsorptionan EXAFS study (49). Further, the pH values used hereaffinity of divalent heavy metal ions compared with simplewere 7.0 or below for Co(II) concentrations higher thanmonovalent ions, and Co2/ ions are adsorbed irrespective of1004 M. All these conditions are unfavorable for the forma-the surface charge due to simple monovalent ion adsorption.tion of hydroxo polymers and may explain why the multinu-As a result, it may be concluded that the model hereclear complexes were not adsorbed in this study.assuming only the interaction between surface complexes of

The Co2/ concentration levels of the cooling water ofthe same kind is operationally sufficient and adequate tonuclear power plants and natural waters are very low (ppt oranalyze and reproduce the observed G.below). Modeling the adsorption in such low concentration

Here, the adsorption species of Co(II ) is assumed to beranges is important and necessary in various applications as

the mononuclear aqua ion Co2/ , but in other investigations,mentioned in the Introduction, although the direct evidence

it has been assumed that the adsorption takes place throughfor the molecular structure of surface complexes forming at

hydrolyzed species, e.g., CoOH/ (45) . However, this spe- such low concentrations is difficullt to obtain by EXAFS orcies only becomes significant in solutions at pH above 9 other spectroscopic methods due to sensitivity limits. There-(46) , and in the pH range of the adsorption experiments fore, instead of direct evidence, a model deduced from estab-here (°7.5) , CoOH/ is a minor component. As a result, an lished facts and observed behaviors was used to explain theexplanation of the measured amount of adsorption through results here.CoOH/ would require a very high reactivity of this species(equilibrium constant ) . In a previous paper (31) , the ex- Computation of Extent of Adsorptionperimental data for Co(II ) adsorption on MnO2 were ana-

The extent of adsorption on a metal oxide for given pH,lyzed with such a mechanism, and the equilibrium constanttotal concentrations of Co2/ , [Co2/]T , amounts (surfacefor CoOH/ was calculated to be 1010 –10 11 times largerareas) of the metal oxide, S , and solution volumes, V, canthan that of Na/ adsorption. However, both CoOH/ andbe computed by solving Eqs. [3] and [10] – [12] togetherNa/ are monovalent cations, and such a very large differ-with the following material balance equation for Co2/ , inence in the equilibrium constants makes the mechanisman iterative way.difficult to accept.

The present model is thermodynamical, and the goodness[Co2/]T Å [Co2/] / GS /V [14]of fit to adsorption data shows the soundness of the model

with respect to the reaction stoichiometry and the thermody-namic relations of adsorption. However, the molecular struc- The fraction of adsorbed Co2/ obtained for a total Co2/

ture of the surface complexes is only indirectly evidenced concentration of 1004 mol dm03 and an oxide surface area/by a thermodynamic approach. solution volume ratio (S /V ) of 80 m2 dm03 is plotted against

Recently, EXAFS studies show that Co(II) are adsorbed pH for the metal oxide samples (Fig. 8) . Figure 8 compareson Al2O3, TiO2, and SiO2 as multinuclear sorption com- the ability of the metal oxides to adsorb Co2/ ions for theplexes (47–50). These multinuclear species were detected given condition on the unit surface area basis, and the ability

order is MnO2 ú Fe3O4 ú TiO2 ú Fe2O3 ú Al2O3.at surface densities of 1 mmol m02 or above (surface cover-

AID JCIS 5148 / 6g34$$$401 11-04-97 11:28:19 coidas


corrosion scale appear to explain the radioactive contamina-tion of cooling systems in nuclear power plants as being dueto the incorporation of radioactive 60Co2/ into this oxidethrough adsorption.

The K 71 and b 72 terms may be regarded as complex con-stants, since the adsorption consists of two processes: ( i)deprotonation of acid type surface hydroxyl groups and (ii)adsorption of Co2/ to the deprotonated sites with a negativecharge.

Process i can be evaluated by the exchange of a monova-lent cation with a surface proton, such as

{OH(a) / Na/ S {O0rNa/ / H/ . [16]

The values of the equilibrium constant for this reaction K 7aFIG. 8. Calculalted fractions of adsorbed Co2/ on metal oxides as areported in the previous paper (37) are shown as a functionfunction of pH for an initial Co2/ concentration of 1004 M and an oxide

surface area S to solution volume V ratio of 80 m2 dm03 at ionic strength of Xi in Fig. 10 (bottom), the value for Fe2O3 was obtained0.1 M (NaNO3) and 257C. in this investigation. That K7a increases with Xi was interpre-

ted to be due to the electron density on donor oxide ionsdecreasing with increasing Xi and the surface hydroxyl pro-

These model calculations would enable the selection oftons held less tightly by covalent bonds, which leads to an

the most effective metal oxide, the optimum pH, and metalenhancement of deprotonation.

oxide dose to remove or recover a specified amount of Co2/

That Fe3O4 shows a larger K 7a value than expected fromions from a solution, and it appears to be applicable to water

the correlation (Fig. 10, bottom) was not pointed out in theand waste water treatment technology. The model calcula-

previous paper (37) but may be explained as follows. Thetions may further be applied to assess the capacity and ability

Xi value for Fe3O4 was calculated with a Z of /2.7, anof soils to retain Co2/ ions, which may serve in soil fertility

average value of the oxidation number of Fe in Fe3O4. How-evaluation, design of repositories for nuclear waste and

ever, with air, Fe(II) ions on the surface of Fe3O4 are oxi-heavy metal waste in general, and other applications.

dized to Fe(III) as suggested by M. A. Blesa, (personalcommunication at the 70th Colloid and Surface Science

Co2/ Adsorption Properties and Electronegativity ofSymposium, Potsdam, NY, 1996), and an appropriate value

Metal Oxidesof Z for the Fe3O4 surface would be larger than /2.7. Hence,the Xi value of the Fe3O4 surface could be larger than thatThe stability constants K 71 and b 72 determined here, ex-

pressing the Co2/ adsorption properties of oxides, were plot-ted against the electronegativity of lattice metal ions of ox-ides Xi (Fig. 9) , as for the monovalent ion adsorption onmetal oxides in the previous paper (37). The Xi is given bythe following equation from Tanaka et al. (51):

Xi Å (1 / 2Z )X0 [15]

where Z is the valence of ions and X0 is the electronegativityof the corresponding elemental metals. This equation wasdeveloped from the relation Xi Å (ÌI /ÌZ ) , where I is theionization potential of the metals and is a polynomial of Z .

Except for Fe3O4, both K 71 and b 72 increase and correlatewell with Xi . This correlation may provide guidelines todescribe and predict the different Co2/ adsorption abilitiesof metal oxides and metal oxide deionizers. The large K71and b 72 for MnO2 strongly suggest that adsorption plays animportant role in the concentration of cobalt in manganese FIG. 9. Log K 71 and log b 72 vs electronegativity Xi of lattice metal ions:

K71 (s) , b 72 (l) .nodules. The large K 71 and b 72 for Fe3O4 among oxides in a

AID JCIS 5148 / 6g34$$$401 11-04-97 11:28:19 coidas


donor electron density with Xi , suggests that the interactionbetween Co 2/ and the deprotonated hydroxyl sites on theseoxides is not a covalent bond formation, although this typeof bond is commonly assumed for divalent heavy metal ionadsorption (52). It may be suggested that the adsorption isdue to ionic bond formation between the large electriccharge on Co2/ and the negative charge of 01 on the depro-tonated hydroxyl sites {O0 . The adsorption of Co 2/ ionsis specific as described in the section ‘‘Stoichiometry,’’ andat least in the direction of the bonds there must be removalof free water molecules or dehydration of ions and deproto-nated sites to allow direct contact between Co2/ and {O0 .This may be termed ‘‘electrostatic specifc (contact) adsorp-tion’’ (53).

The larger K 7*1 and b 7*2 values for Al2O3 indicate astronger Co 2/ attraction by the {O0 sites of this oxide.Here, the donor electron density on {O0 may be higherbecause the lattice Al3/ ion is a very hard acid (54) anddoes not accept donor electrons from surrounding (hydr)-oxide ions . It is likely that the Co2/ adsorption on Al2O3

is due to both the ionic and covalent bond formation. How-ever, for the other oxides with less hard acid lattice metalions, the donor electron density may be lower, and the

FIG. 10. Log K7a (bottom), log K 7*1 (middle) , and log b 7*2 (top) vs covalent bond may not play a role in the adsorption aselectronegativity Xi of lattice metal ions. discussed above.

Lateral Interaction Properties of Metal Oxidesobtained here, and then the points for Fe3O4 would shifttoward the correlation curve. Further, surface hydroxyl

For all the oxides, B2 ú B1 (Table 2), indicating that thegroups formed by air-oxidation may exhibit large reactivities(1:2) surface complex formation is subject to a strongerdue to the following reasons. The lattice structure of thesuppression than the (1:1) complex. If the Co(II) surfaceoxidized magnetite surface (all Fe(III) ions) must be differ-complexes are completely polarized, i.e., d/ Å /2 and d0ent from that of the underlying Fe3O4, and stress caused byÅ 01 (Fig. 1) , then the (1:2) surface complex has localthe adjacent different lattices may give rise to higher reactiv-charges of 01 1 2 Å 02 on one side and /2 on the otherity, resulting in the large value of K 7a .side. While the (1:1) complex has local charges of 01 andProcess ii can be expressed by/2, and the extra positive charge /1 is balanced symmetri-cally by an adsorbed X0 ion. It is likely that a more symmet-

{O0rNa/ / Co2/ / X0 S {OCo/rX0 / Na/ , [17]

rical charge distribution causes less electrostatic repulsion,2{O0

rNa/ / Co2/ S ({O)2Co / 2Na/ . [18] and this may explain the smaller B1 than B2 .The B1 and B2 values are different for different oxides,

If the equilibrium constants for these reactions are repre- suggesting that the lateral interactions are affected by surfacesented by K 7*1 and b 7*2 , then K 71 Å K 7a K 7*1 and b 72 Å properties of the oxides, like morphology, geometry, chemi-(K 7a ) 2b 7*2 . The values of K 7*1 and b 7*2 were obtained with cal composition, crystal structure, pore structure, and others.these relations and are plotted against Xi in Fig. 10 (middle The BET specific surface areas of the metal oxides, exceptand top). The K7*1 and b 7*2 values decrease with Xi , but for Fe3O4, are much larger than those estimated from thethe change is only slight except for Al2O3. It is now apparent densities of the oxides and from the modal diameters of thethat the increases in K 71 and b 72 with Xi (Fig. 9) are due to oxide particles observed by electron microscopy (Table 1).the larger increase in K 7a (Fig. 10, bottom) compared with The Fe3O4 sample here was shown to be nonporous (4) , butthe changes in K 7*1 and b 7*2 (Fig. 10, middle and top). The the other metal oxides have internal surfaces possibly duelarge K 71 and b 72 for Fe3O4 deviating from the correlations to microstructures, such as pores, crevices, and flaws (55).are also due to the above mentioned large value of K7a . On such internal surfaces, interactions from ions adsorbed

on opposite walls (across-pore interaction) (39), in additionThe small Xi dependence of K 7*1 and b 7*2 with Fe3O4,Fe2O3, TiO2, and MnO2, irrespective of the decrease in the to interactions between adjacent ions, would suppress ad-

AID JCIS 5148 / 6g34$$$401 11-04-97 11:28:19 coidas


10. Tillar, K. G., Hodgson, J. F., and Peech, M., Soil Sci. 95, 392 (1963).sorption. It may be due to this across-pore interaction that11. McKenzie, R. M., Aust. J. Soil Res. 5, 235 (1967).the lateral interaction constants are different for different12. Grimme, H., Z. Pflanzenernaehr. Bodenkd. 121, 58 (1968).

oxides and the metal oxides with large BET specific surface 13. Forbs, E. A., Posner, A. M., and Quirk, J. P., J. Soil Sci. 27, 154areas have larger B1 and B2 values than Fe3O4 (Table 2). (1976).

It will be necessary to characterize the surface properties 14. Loganathan, P., Burau, R. G., and Fuerstenau, D. W., Soil Sci. Soc.Am. J. 41, 57 (1977).of oxides in more detail for further evaluation of these lateral

15. Brusseau, M. L., and Zachara, J. M., Environ. Sci. Technol. 27, 1937interaction constants.(1993).

16. Kozawa, A., J. Electrochem. Soc. 106, 552 (1959).SUMMARY 17. Morgan, J. J., and Stumm, W., J. Colloid Sci. 19, 347 (1964).

18. Loganathan, P., and Burau, R. G., Geochim. Cosmochim. Acta 37, 1277(1973).The results of this investigation may be summarized as

19. Swallow, K. C., Hume, D. N., and Morel, F. M. M., Environ. Sci. Tech-shown below.nol. 14, 1326 (1980).

20. Wang, Z., and Stumm, W., Netherlands J. Agric. Sci. 35, 231 (1987).(a) Reevaluation and successful application of the Frum-21. Westall, J., and Hohl, H., Adv. Colloid Interface Sci. 12, 265 (1980).kin isotherm to model the Co(II) adsorption properties of22. James, R. O., and Parks, G. A. in ‘‘Surface and Colloid Science’’ (E.various ‘‘real’’ metal oxide samples found in environments

Matijevi’c, Ed.) , Vol. 12, p. 119. Plenum, New York, 1982.and industry; 23. Westall, J. C., (Chap. 4) , Bousse, L., and Meindl, J. D., (Chap. 5) ,

(b) Determination of adsorbing Co(II) as Co2/ aqua ions Chan, D. Y. C., (Chap. 6) , Hayes, K. F., and Leckie, J. O., (Chap. 7) ,Honeyman, B. D., and Leckie, J. O., (Chap. 9) , Yasunaga, T., andfor the conditions of this investigation, excluding mono- andIkeda, T., (Chap. 12), in ‘‘Geochemical Processes at Mineral Surfaces’’polynuclear Co(II) hydroxo complexes;(J. A. Davis and K. F. Hayes, Eds.) , ACS Symposium Series 323.(c) Establishment of correlations between the equilib-American Chemical Society, Washington, DC, 1986.

rium constants for adsorption and the electronegativity of 24. Westall, J. C., (Chap. 1) , Schindler, P. W., and Stumm, W., (Chap.lattice metal ions Xi ; 4) , in ‘‘Aquatic Surface Chemistry’’ (W. Stumm, Ed.) . John Wiley &

Sons, New York, 1987.(d) Elucidation of c as being due to the enhancement of25. Dzombak, D. A., and Morel, F. M. M., ‘‘Surface Complexation Model-deprotonation of surface hydroxyl sites (covalent bond

ing,’’ Chap. 2. John Wiley & Sons, New York, 1990.break) by decreasing donor electron densities with increas-26. Davis, J. A., and Kent, D. B., (Chap. 5) , Sposito, G., (Chap. 6) , Schin-

ing Xi ; dler, P. W., (Chap. 7) , in ‘‘Mineral–Water Interface Geochemistry’’(e) Proposal of an ‘‘electrostatic specific (contact) ad- (M. F. Hochella, Jr., and A. F. White, Eds.) , Reviews in Mineralogy,

sorption’’ of Co2/ to deprotonated surface hydroxyl sites Vol. 23. Mineralogical Society of America, Washington, DC, 1990.27. Stumm, W., ‘‘Chemistry of the Solid–Water Interface,’’ Chap. 3. Johnwith negative charge, this process being independent of the

Wiley & Sons, New York, 1992.donor electron density which varies with Xi ; and28. Morel, F. M. M., and Hering, J. G., ‘‘Principles and Applications of(f ) Suggestion of ‘‘across-pore interaction’’ in micro- Aquatic Chemistry,’’ Chap. 8. John Wiley & Sons, New York, 1993.

pores of oxides to explain the large lateral interactions for 29. Jin, X., Talbot, J., and Wang, N.-H. L., AIChE J. 40, 1685 (1994).oxides with large BET specifc surface areas. 30. Ulrich, H.-J., Stumm, W., and Cosovic, B., Environ. Sci. Technol. 22,

37 (1988).31. Katayama, N., Tamura, S., Tamura, H., and Furuichi, R., Denki KagakuACKNOWLEDGMENT

62, 251 (1994).32. Katayama, N., Dr. Eng. Thesis, Hokkaido University, Japan, 1996.

H.T. expresses thanks to Nippon Steel Corporation, Japan, for financial33. Katayama, N., Tamura, H., and Furuichi, R., Bunseki Kagaku 39, 547

support.(1990).

34. Gadde, R. R., and Laitinen, H. A., Anal. Chem. 46, 2022 (1974).35. Kinniburg, D. G., and Jackson, M. L., in ‘‘Adsorption of Inorganics atREFERENCES

Solid–Liquid Interfaces’’ (M. A. Anderson and A. J. Rubin, Eds.) ,Chap. 3. Ann Arbor Science Publishers, Ann Arbor, MI, 1981.1. Tewari, P. H., Campbell, A. B., and Lee, W., Can. J. Chem. 50, 1642

36. James, R. O., and Healy, T. W., J. Colloid Interface Sci. 40, 53 (1972).(1972).37. Tamura, H., Katayama, N., and Furuichi, R., Environ. Sci. Technol.2. Tewari, P. H., and Lee, W., J. Colloid Interface Sci. 52, 77 (1975).

30, 1198 (1996).3. Blesa, M. A., Larotonda, R. M., Maroto, A. J. G., and Regazzoni, A. E.,38. Gileade, E., Kirowa-Eisner, E., and Penciner, J., ‘‘Interfacial Electro-Colloids Surf. 5, 197 (1982).

chemstry,’’ p. 82. Addison-Wesley, Reading, MA, 1975.4. Tamura, H., Matijevi’c, E., and Meites, L., J. Colloid Interface Sci.39. Tamura, H., Kudo, M., and Furuichi, R., Anal. Chim. Acta 271, 30592, 303 (1983).

(1993).5. Venkataramani, B., Venkateswarlu, K. S., and Shankar, J., J. Colloid40. Tamura, H., Ohkita, K., Katayama, N., and Furuichi, R., Bunseki Ka-Interface Sci. 67, 187 (1978).

gaku 43, 831 (1994).6. Goldberg, E. D., J. Geol. 62, 249 (1954).41. Meites, L., CRC Crit. Rev. Anal. Chem. 8, 1 (1979).7. Murray, J. W., Geochim. Cosmochim. Acta 39, 505 (1975).42. Rustling, J. F., and Kumosinski, T. F., ‘‘Nonlinear Computer Modeling8. Murray, J. W., Geochim. Cosmochim. Acta 39, 635 (1975).

of Chemical and Biochemical Data,’’ pp. 18–31. Academic Press, San9. Schenk, C. V., Dillard, J. G., and Murray, J. W., J. Colloid InterfaceSci. 95, 398 (1983). Diego, CA, 1996.

AID JCIS 5148 / 6g34$$$401 11-04-97 11:28:19 coidas


43. Press, W. H., Flannery, B. P., Teukolsky, S. A., and Vettering, W. T., 49. Chisholm-Brause, C. J., Brown, G. E., Jr., and Parks, G. A., ‘‘Proceed-‘‘Numerical Recipes,’’ pp. 254–259. Cambridge Univ. Press, Cam- ings of the 6th International Conference on X-Ray Absorption Finebridge, 1986. Structure’’ (S. S. Hasnain, Ed.) , Paper No. 65. Ellis-Horwood Publish-

44. Yamaoka, K., ‘‘Yakubutsu Tainaidotai Kaisekiho,’’ pp. 124–125. Nan- ers, Chichester, U.K., 1991.kodo, Tokyo, 1984. 50. Towle, S. N., Bargar, J. R., Brown, G. E., Jr., and Parks, G. A., J.

45. James, R. O., and Healy, T. W., J. Colloid Interface Sci. 40, 42 (1972). Colloid Interface Sci. 187, 62 (1997).46. Kragten, J., ‘‘Atlas of Metal–Ligand Equilibria in Aqueous Solution,’’ 51. Tanaka, K., and Ozaki, A., J. Catal. 9, 1 (1967).

p. 173. John Wiley & Sons, New York, 1978.52. Stumm, W., ‘‘Chemistry of the Solid–Water Interface,’’ p. 21. John

47. Brown, G. E., Jr., in ‘‘Mineral–Water Interface Geochemistry’’ (M. F.Wiley & Sons, New York, 1992.

Hochella and A. F. White, Eds.), Reviews in Mineralogy, Vol. 23, pp.53. Bockris, J. O’M., and Khan, S. U. M., ‘‘Surface Electrochemistry,’’ p.338–341. Mineralogical Society of America, Washington, DC, 1990.

135. Plenum, New York, 1993.48. O’Day, P. A., Brown, G. E., Jr., and Parks, G. A., ‘‘Proceedings of the54. Basolo, F., and Pearson, F., ‘‘Mechanisms of Inorganic Reactions,’’6th International Conference on X-Ray Absorption Fine Structure’’

2nd ed., p. 25. John Wiley & Sons, New York, 1967.(S. S. Hasnain, Ed.) , Paper No. 64. Ellis-Horwood Publishers, Chiches-ter, U.K., 1991. 55. Kozawa, A., Prog. Batteries Solar Cells 7, 327 (1988).

AID JCIS 5148 / 6g34$$$401 11-04-97 11:28:19 coidas

Documents

The Co2+Adsorption Properties of Al2O3, Fe2O3, Fe3O4, TiO2, and MnO2Evaluated by Modeling with the Frumkin Isotherm