Surface Complexation at the TiO2(anatase)/Aqueous Solution Interface: Chemisorption of Catechol

JOURNAL OF COLLOID AND INTERFACE SCIENCE 177, 122–131 (1996)Article No. 0012

Surface Complexation at the TiO2 (anatase)/Aqueous SolutionInterface: Chemisorption of Catechol

RAUL RODRIGUEZ, MIGUEL A. BLESA, AND ALBERTO E. REGAZZONI1

Departamento Quımica de Reactores, Comision Nacional de Energıa Atomica, Av. del Libertador 8250, 1429-Buenos Aires, Argentina

Received January 3, 1995; accepted May 12, 1995

when adsorption is driven chemically and involves the ex-Catechol adsorbs at the TiO2 (anatase) /aqueous solution inter- change of bound water molecules, or OH0 , for the adsorbing

face forming inner-sphere surface complexes. The UV–visible dif- ligand (12–15); inner-sphere surface complexation is lim-ferential reflectance spectrum of surface titanium–catecholate ited to the adsorption of complexing species. Recent spectro-complexes presents a band centered at 420 nm which corresponds

scopic evidence (16–18) and EXAFS (19, 20) characteriza-to the ligand to metal charge transfer transition within the surfacetion of adsorbed anions provide sound grounds for the wide-complexes. At pH values below pKa1 , the surface excess of catecholspread use of the surface complexation approach.is almost insensitive toward pH and presents a Langmuirian de-

The surface sites have been customarily described as elec-pendence with the concentration of uncomplexed catechol. Thetroneutral GMeOH surface groups, a notation that accountsratio Gmax :NS (NS being the measured density of available OH

surface groups) indicates a prevailing 1 to 2 ligand exchange ad- for their amphoteric nature and the ability of GMe to ex-sorption stoichiometry. In the range pH § pKa1 , the catechol change ligands. In this description, however, all surface sitessurface excess decreases markedly with increasing pH. Formation are assumed to be energetically equivalent; differences aris-of 1 to 1 surface complexes produces an excess of negative surface ing from the dissimilar coordination environments of thecharge that is revealed by the shift of the iep to lower pH values. different crystal faces that may be exposed to the aqueousThe reported data, which are supplemented with information on

phase are lumped into overall surface equilibrium constants.the charging behavior of TiO2 suspended in indifferent electrolyteThe more general description that includes surface coordi-solutions, are interpreted in terms of the multi-site surface com-nated water molecules (7, 8, 21), i.e., GMe(OH)m(OH2)n ,plexation model. In this model, two types of surface OH groupsalso presents this limitation. To overcome this liability,are considered: GTiOH1/30 and GOH1/3/ . Although both surfaceHiemstra et al. (22–24) have recently proposed a multi-sitegroups undergo protonation–deprotonation reactions, only

GTiOH1/30 are prone to chemisorption. q 1996 Academic Press, Inc. protonation model in which the acidity and density of theKey Words: surface complexation; chemisorption; TiO2; anatase; different types of OH surface groups are dictated by the

catechol. crystal structure of the exposed hydroxylated surfaces. Theconcepts underlying this model further imply that not everysite would be prone to chemisorption, because the suscepti-

INTRODUCTION bility of each type of OH surface group towards ligand ex-change should depend on its coordination number.

The surface complexation concept is one of the most suc- Clearly, multisite complexation models should offer acessful approaches to the rationalization of the colloid chem- more comprehensive picture of the chemisorption phenom-istry of metal oxides immersed in water (1–6). In this out- ena. In this paper we present a detailed study of the adsorp-look, dissolved solutes interact with partially hydrated sur- tion of catechol (1,2-dihydroxybenzene) at the TiO2 (ana-face metal ions in a way that parallels solution chemistry. tase)/solution interface and probe the applicability of theThus, surface charge develops through the protonation–de- multisite surface complexation model to describe the adsorp-protonation of water molecules bound to surface metal ions, tion of weak acids at metal oxide/aqueous solution inter-and adsorbing ligands form outer- and/or inner-sphere surfaces.face complexes (7, 8) . Outer-sphere surface complexation The studied system was selected because catechol is atakes place when electrostatic forces are the dominant contri- diprotic weak acid that prevails in the undissociated formbution to the Gibbs adsorption energy (9–11). Formation

over a wide pH range (pKa1 Å 9.2; pKa2 Å 13.0) and formsof inner-sphere surface complexes, on the other hand, ensues

very stable complexes with aqueous Ti( IV) (25–27) . Inaddition, TiO2 aqueous suspensions behave as ideal sys-

1 To whom correspondence should be addressed. tems; indeed, they have served to establish the basis of

1220021-9797/96 $12.00Copyright q 1996 by Academic Press, Inc.All rights of reproduction in any form reserved.

/ m4147$3853 11-29-95 17:24:18 coidas AP: Colloid

123CATECHOL CHEMISORPTION ONTO ANATASE

various models for ionic adsorption, e.g., the site-binding TABLE 1Adsorption of Hydroxyl Ions from KOH Solutionsmodel (9, 28, 29) . Furthermore, dissociative water chemi-

sorption on the prevailing crystal faces of both rutile and[OH0] 103 GOH

0 106 [OH0]/GOH0anatase yields two types of OH surface groups (30–33) :

(mol dm03) (mol m02) (m2 dm03)hydroxyls bond to penta-coordinated titanium surface ions( i.e., GTiOH 1/30 ) and protons bond to di-coordinated oxo 4.58 2.19 2091surface ions ( i.e., GOH 1/3/) . 6.30 2.41 2614

9.49 2.54 3736

EXPERIMENTAL

of dissolved Ti(IV) –catecholate complexes in the superna-Titanium dioxide, P-25 TiO2, was from Degussa Corp.tants was always below 2 1 1007 mol dm03 , as indicatedThis solid was thoroughly washed by dialysis until the con-by the zero absorbance measured at 389 nm (cf. Refs. (25)ductivity of the supernatant solution was the same of water,and (27)) .vacuum dried, and stored in a desiccator, in the dark. The

In selected cases, the filtered off solids were washed withBET specific surface area, determined from the N2 adsorp-water and vacuum-dried, and the UV–visible diffuse re-tion isotherm at 77 K, was 51.4 m2 g01 . The XRD patternflectance spectra of the powders were recorded in a Shi-of the powder indicated that it is mainly anatase, the contentmadzu UV-210A spectrometer.of rutile being less than ca. 10%.

All adsorption experiments were carried out under normalAll other reagents were analytical grade and were usedlaboratory illumination at 298.0 { 0.2 K in thermostatedwithout further purification. Solutions were made using de-borosilicate glass vessels. In all cases, O2 and CO2 wereionized water obtained from an E-pure apparatus (conductiv-removed from the systems by bubbling N2 previouslyity less than 1005 S m01) .scrubbed through an alkaline pyrogallol solution. During theThe Brønsted acid–base behavior of the P-25 TiO2 sur-experiments, TiO2 particles were kept in suspension using aface was characterized following the well-known fast titra-magnetic stirrer and a teflon-coated magnetic bar; to breaktion procedure. About 0.3 g of TiO2 were suspended in 0.1aggregates, the suspensions were briefly ultrasonicated usingdm3 of KCl solutions of desired concentration and titrateda titanium probe. Cellulose nitrate membranes with a porewith 0.1 mol dm03 HCl and 0.1 mol dm03 KOH using asize of 0.20 mm were used for filtration.Metrohm 761 DMS Titrino and a 0.001 dm3 burette; pH

The electrophoretic mobilities of P-25 TiO2 particles sus-values were recorded 3 min after each addition of titrant.pended in 1002 mol dm03 KCl solutions were measured atComparison with blank titration curves yields the difference298.0{ 0.2 K as a function of pH and catechol concentration(GH/ 0 GOH0) .in a Mark II Rank Brothers apparatus using a thermostatedAdditional base adsorption experiments were performedcylindrical microelectrophoresis cell. Before the measure-to measure the density of OH surface groups. Weightedments, the suspensions were left standing overnight underamounts of TiO2 were suspended in KOH solutions of knownan O2–CO2-free atmosphere.concentration and allowed to equilibrate for 15 min. After

removing the solid by filtration, the concentration of base inRESULTSthe supernatant solutions was determined potentiometrically.

Each experiment was run in triplicate; experiments carriedA key parameter in the description of ionic adsorptionout at longer equilibration times yielded the same results.

is NS , the number of available OH surface groups, whichTo measure catechol adsorption isotherms, typically 0.8determines the monolayer (maximum) coverage. Resultsg of TiO2 were suspended in 0.05 dm3 of freshly preparedfrom the set of KOH adsorption experiments carried out tocatechol solutions of known concentration and the pH wasquantify NS are collected in Table 1. These data were fittedadjusted to prefixed values by the addition of measuredto a linearized Langmuir adsorption isotherm, e.g.,amounts of HCl or KOH. These suspensions were left to

equilibrate for 30 min (sequential experiments indicated thatcatechol adsorption reached equilibrium within 10 min); Ì([OH0] /GOH0)

Ì[OH0]Å 1

NS

Å const, [1]during this period, pH remained constant within {0.1. Thesolid was filtered off and the concentration of catechol in

from which NS Å 2.97 1 1006 mol m02 (or 1.79 sites nm02)the supernatant solutions was determined spectrophotometri-was obtained; note that under the present conditions (GOH0cally at 275 nm. Catechol surface excesses were calculated

by solving the mass balance of the systems; in all cases, 0 GH/) É GOH0 . This figure is in close agreement with thefluoride exchange capacity (1.7 sites nm02) measured for P-blank experiments were performed. Dissolution during the

course of the experiments was negligible; the concentration 25 TiO2 at pH 7.8 (34) and the maximum attainable acetate


124 RODRIGUEZ, BLESA, AND REGAZZONI

as further discussed below. Therefore, it seems plausible thatthe increased consumption of base at the higher hydroxideconcentrations may reflect a nucleophilic attack of OH0 ontitanium sites. In fact, participation of OH0 ions in the ratedetermining step of Ti(IV) dissolution from TiO2 films inalkaline media has already been diagnosed (37).

The interfacial properties of P-25 TiO2 immersed in indif-ferent electrolyte aqueous solutions is presented in Fig. 1.Figure 1A depicts proton adsorption data as a function ofpH and electrolyte concentration; these data are expressedas surface charge density values, i.e., s0 Å F(GH/ 0GOH0) . Figure 1B shows the pH dependence of the electroki-netic potential (z) of P-25 TiO2 suspended in 1002 mol dm03

KCl; z potentials were derived from experimental electro-phoretic mobility data following the procedure proposed byO’Brien and White (38). The point of zero charge (pH0 Å6.50 { 0.05), which coincides with the isoelectric point( iep) , is in excellent agreement with previously reportedvalues (39–47); s0 and z values are also in the range offormer data.

Catechol adsorption isotherms at three fixed pH valuesare presented in Fig. 2. Within the pH range 3.6 £ pH £6.0, the surface excess is almost independent of pH andpresents a Langmuirian dependence with the concentrationof free catechol; the Langmuir parameters are KL Å 8.2 1103 mol01 dm3 and Gmax Å 1.25 1 1006 mol m02 . Thesevalues are in good concordance with those reported by Moseret al. (48) for the adsorption of aromatic acids from hydro-alcoholic solutions onto P-25 TiO2. While only two parame-

FIG. 1. pH dependence of the surface charge density (A) and z potentialters are needed to describe the adsorption isotherms depicted(B) of TiO2 (anatase) particles immersed in KCl solutions: (l) 0.1 molin Fig. 2, the Langmuirian constants describing ionic adsorp-dm03 ; (j) 0.01 mol dm03 ; (m) 0.001 mol dm03 ; data points in (B) corre-

spond to 0.01 mol dm03 KCl; T Å 298 K; solid lines are model calculations tion are necessarily conditional; indeed, they may be pH(see text) . dependent. In fact, the affinity of catechol for the TiO2 sur-

face decreases markedly at pH values larger than ca. 8.5.adsorption density (2.0 sites nm02) (32); in both cases, a This is illustrated by Fig. 3, in which adsorption data ob-one to one ligand exchange stoichiometry may be assumed.The reported value of NS also agrees well with those (ca.2.4 sites nm02) derived by Boehm (32, 35) and by van Veenet al. (34) from hydroxide adsorption experiments carriedout under conditions similar to those of the present work.

At higher hydroxide concentrations (ca. 0.1 mol dm03) ,Boehm (32, 35) and van Veen et al. (34) observed higherOH0 consumptions, from which they derived larger NS val-ues (in the order 4.4–5.7 sites nm02) ; as a possible explana-tion, the existence of OH surface groups with markedly dif-ferent acidic behaviors was postulated. However, the mea-sured density of chemisorbed water is 2.3 H2O moleculesnm02 (36). Thus, 4.6 sites nm02 would require the ionizationof all protons derived from chemisorbed water. It has beennoted previously (8) that nondissociatively coordinated wa-

FIG. 2. Catechol adsorption isotherms for different pH values: (l)ter molecules can behave only as monoprotic acids. More- 3.65; (j) 4.25; (m) 6.00; no added background electrolyte; T Å 298 K;over, it has been shown that dissociative chemisorption of symbols are experimental data points; solid lines are model calculations

(see text) .H2O on TiO2 produces only one acidic OH group (22–24),



FIG. 3. Catechol surface excess as a function of pH. Total catecholconcentration: 0.001 mol dm03 ; available TiO2 surface area: 822.4 m2 dm03 ; FIG. 5. Differential diffuse reflectance spectrum of surface titanium–no added background electrolyte; T Å 298 K; symbols are experimental catecholate complexes; in arbitrary absorbance units.data points; the solid line corresponds to model calculations (see text) .

Complexation of surface titanium ions by catechol is dem-tained from experiments carried out at a constant total cate-onstrated by the changes in the UV–visible diffuse re-chol concentration and at a fixed surface to volume ratio areflectance spectrum of TiO2. The spectrum of TiO2 displaysplotted as a function of pH; note that G values presented inthe characteristic absorption edge at l É 400 nm, but, in thethe form of Fig. 3 are sensitive to the experimental condi-presence of adsorbed catechol, it presents a broad shoulder attions, e.g., they depend on the selected surface to volumeca. 440 nm and a tail that extends to about 700 nm; the latterratio.resembles the absorption spectrum of the colloidal-TiO2/Adsorption of catechol produces an excess of negativecatechol system reported by Moser et al. (48). The differ-surface charge, thus the iep of TiO2 particles moves to lowerence spectrum, which is presented in Fig. 5, shows a bandpH values (Fig. 4); at pH values apart from iep, however,centered at 420 nm that can be assigned to the intramolecularz-potential values are only slightly sensitive to catechol con-ligand to metal charge transfer transition within the surfacecentration. This effect is characteristic of anionic adsorption,titanium(IV) –catecholate complexes. This band is 30 nmand indicates, as already diagnosed for the Fe3O4/H3BO3 red-shifted as compared to that appearing in the absorptionand ZrO2/H3BO3 systems (15), that surface complexationspectrum of aqueous Ti(cat)20

3 complexes (lmax Å 389 nm)enhances the acidity of catechol. Quasi-equivalent coadsorp-(27). Even though such shifts seem to be common whention of protons (49), which is responsible for the insensitiv-comparing the spectra of ‘‘analogous’’ aqueous and surfaceity of catechol surface excess toward pH in the range 3.6–complexes (50), a conclusive explanation of the observed8.5 (Fig. 3) , does not suffice to counterbalance the builddifferences cannot yet be forwarded; differences in the coor-up of negative charge that causes the observed shift.dination environments of surface and aqueous Ti(IV) ions,and coupling between the electronic levels of bulk oxide andsurface complexes might, in principle, be invoked.

DISCUSSION

The Surface Site Density

The use of multi-site complexation models (22–24) re-quires the identification of the types of OH surface groupsthat determine the colloid chemistry of the oxide sample.Rigorously, such a task implies the knowledge of the actualcrystal faces composing the particle habit. This is not usuallythe case. In principle, equilibrium crystal shape may be as-

FIG. 4. The influence of catechol adsorption on z potential vs pH pro- sumed, although particles of nano-crystalline powders, suchfiles; catechol equilibrium concentration: (j) 5 1 1003 mol dm03 ; (l) 5

as the P-25 TiO2, may expose different crystal faces, includ-1 1004 mol dm03 ; (s) in the absence of catechol; [KCl] Å 0.01 moling the less stable ones. Keeping this drawback in mind, wedm03 ; T Å 298 K; symbols are experimental data points; dotted lines are

model calculations (see text) . will assume, for simplicity, that the behavior of the TiO2



(anatase)/solution interface is dominated by the properties expression is lacking (51).2 Nevertheless, the surface specia-tion can be calculated on the basis of a suitable electricalof GTiOH1/30 and GOH1/3/ surface groups that are cre-

ated upon hydroxylation of freshly cleaved (001) and (011) double layer (EDL) model by solving the set of equationsthat describes the charge and potential distribution across thesurfaces (32, 33).

The acid–base character of these sites is represented by interface. This set, which is implicitly based on the Gouy–Chapman–Stern–Grahame (GCSG) structure of the inter-two surface equilibria:face, is presented in Table 2. Although simpler EDL models(e.g., the constant capacitance model) can accurately de-GTiOH2/3/

2 S GTiOH1/30 / H/ ; K intA [2]

scribe the charging behavior of metal oxide surfaces (52),GO2/30 / H/ S GOH1/3/ ; K int

B . [3] those based on the GCSG structure account better for electro-kinetic potentials (7, 15).

These equations emphasize the Brønsted acidic nature of To solve the set of equations presented in Table 2 theGTiOH2/3/

2 (A sites) and the basic character of GO2/30values of K int

A , K intB , fK/ , fCl0 , C1 , and C2 are required; as

(B sites) . K intA and K int

B , the intrinsic (potential independent) discussed before, NA and NB are related to NS which wassurface equilibrium constants, are given by assessed experimentally. The actual number of unknown pa-

rameters is however less, for the surface equilibrium con-stants are related through

K intA Å {GTiOH1/30}[H/]

{GTiOH2/3/2 }

exp(0Fc0 /RT ) [4]

pH0 Å12

(log K intB 0 log K int

A ) , [16]K intB Å {GOH1/3/ }

{GO2/30}[H/]exp(Fc0 /RT ) , [5]

where { } denotes surface concentration, which will be and equivalent specific adsorption of background electrolyteexpressed in mol m02 . c0 is the surface potential. ions requires

Equations [2] and [3] also stress that dissociatively che-misorbed water behaves as a monoprotic acid; deprotonation

fK/ Å fCl0 . [17]of GTiOH1/30 , if possible, may only occur in extremelyalkaline media (23, 24). Thus, provided that electrostaticeffects are neglected, the maximum number of titrable sur- Dissociative chemisorption of water imposes a further con-face protons, NS Å 1.79 OH nm02 (or 2.97 1 1006 mol straint (cf. Eqs. [2] and [3]):m02) , must represent the total density of A sites. The massbalance for A sites at any pH is therefore

K intA § 1/K int

B . [18]

NA Å NS Å {GTiOH2/3/2 } / {GTiOH1/30}. [6]

In the single-site description of metal oxide surfaces, allmodel parameters, with the exception of C2 , can be deter-The electroneutrality condition in the chargeless dehydratedmined through the appropriate treatment of s0 vs pH and NSbare surface requires that NS must also equal the total densitydata (9–14, 53, 54). In the present description of the P-25of B sites. Thus, the mass balance for B sites isTiO2 surface, however, linearization of s0 vs pH is not possi-ble, and unknown model parameters must be derived fromNB Å NS Å {GO2/30} / {GOH1/3/ }. [7]the best fit to the data sets presented in Fig. 1; note that, inmodeling s0 vs pH and z vs pH data, the identity, z Å cd ,These simple ideas are essential to avoid the predictionis assumed to be valid. Nonlinear least-squares fitting wasof unrealistic monolayer coverages.performed with the aid of a BASIC program written by theauthors. Since capacity values must be independent of theThe TiO2/Indifferent Electrolyte Solution Interfaceactual description of the surface sites, previously reported

Before discussing the chemisorption of catechol at the values of C1 and C2 were used to reduce the number ofTiO2 surface in terms of the multisite surface complexationmodel, a brief description of the TiO2/ indifferent electrolyte

2 Notice that a Nernstian pH dependence of c0 is incompatible with anysolution interface will be presented. In principle, the distribu-surface site model (see, for instance, Ref. (51)); if c0 were a Nernstian func-

tion of surface sites, hence the charging behavior of the TiO2 tion of pH, the ratios {GTiOH2/3/2 }/{GTiOH1/30} and {GOH1/3/ }/

surface, could be predicted if an independent expression {GO2/30} would be constant and independent of pH, viz. s0 would beinvariant (cf. Eqs. [9] – [11]) .relating c0 and pH was available. Unfortunately, such an



TABLE 2Equations Describing the TiO2/Indifferent Electrolyte Solution Interface According to the GCSG Structure

of the Interfacial Region

Surface mass balance [6]NA Å NS Å {åTiOH2/3/2 } / {åTiOH1/30}

[7]NB Å NS Å {åO2/30} / {åOH1/3/}

Electroneutrality [8]s0 / sb/ sd Å 0

Surface charge density [9]s0 Å s/0 0 s00

[10]s/0 Å F[23{åTiOH2/3/

2 } / 13 {åOH1/3/}]

[11]s00 Å F[23 {åO2/30} / 1

3 {åTiOH1/30}]

Countercharge at the b planea [12]sb Å

s00 KK/ [K/]exp(0Fcb /RT)

1 / KK/[K/]exp(0Fcb /RT)

0 s/0 KCl0[Cl0]exp(Fcb /RT)

1 / KCl0[Cl0]exp(Fcb /RT)

Countercharge at the d planeb [13]sd Å 0.1174 I1/2 sinh (0Fcd /2RT)

Charge-potential relationships [14]c0 0 cb Å s0/C1

[15]cb 0 cd Å 0sd /C2

a The constants KK/ and KCl

0 are related to the specific adsorption potentials (fi) through Ki (mol01 dm3) Å 0.018 exp(0fi /RT).b For aqueous solutions at 298 K; sd is given in C m02; I is the ionic strength in mole dm03.

parameters that require optimization.3 The model parameters the point of zero charge is insensitive to electrolyte concen-tration.that best describe the P-25 TiO2/ indifferent electrolyte solu-

tion interface are listed in Table 3. The agreement betweenComplexation of Surface Titanium Ions by Catecholmodel calculations and experimental data is noteworthy (see

Fig. 1) . The analogy between the UV–visible spectra ofThe calculated distribution of surface sites is depicted in Ti ( IV ) –catecholate aqueous complexes reported by

Fig. 6 as a function of pH and ionic strength. As imposed by Borgias et al. ( 27 ) and that shown in Fig. 5 is anotherthe constraint [18], the OH surface groups with the smallest example of the correspondence between surface and so-charge number prevail. However, as the ionic strength in- lution chemistry. Adsorbed catecholate anions bind sur-creases, the fractions of GTiOH2/3/

2 and GO2/30 become face titanium ions in a process that involves the substitu-increasingly important; notice that the surface speciation at tion of surface coordinated OH0 ( or water molecules ) .

Since the surface complexation approach describes che-

TABLE 3Model Parameters Describing the P-25 TiO2/Indifferent

Electrolyte Solution Interface

log logNs C1 C2

(nm02) K intA K int

B fK/ /RT fCl

0 /RT (F m02) (F m02)

1.79 05.38a 7.60a 04.36a 04.36a 1.40b 0.23c

a Obtained by fitting s0 vs pH and z vs pH data under the constraintsimposed by Eqs. [16]–[18].

b Taken from Refs. (28, 29).c Taken from Ref. (41).

3 In principle, K intA and K int

B can also be estimated independently from thecorrelation found by Hiemstra et al. (22). However, the estimated valuespredict a pH0 that would range between 5.8 and 4.2, depending on the FIG. 6. Distribution of A and B surface sites in the absence of catechol;

ionic strengths are 0.1 mol dm03 (dotted lines) and 0.001 mol dm03 (solidexposed crystal face. We have therefore chosen to leave K intA (or K int

B ) asan adjustable parameter. lines) .



misorption as a fully equilibrated and totally reversible of 1 to 1 surface complexes (species I) would imply thatsurface titanium ions adopt the uncommon hepta-coordina-process, only the A sites ( e.g., GTiOH 1 / 30 ) should be

prone to reversible ligand exchange. In fact, the partici- tion. On the other hand, surface titanium ions retain theirnormal coordination environment in species II (Eq. [20]) .pation of B sites ( e.g., GOH 1 / 3/ ) in a ligand exchange

surface reaction must be impeded, at least kinetically, Thus, although the distortion of the TiO6 octahedra in speciesII must increase to accommodate for the mismatch betweenby the large energetic requirement involved in the rupture

of titanium-oxo bonds. Also, any ligand exchange in- the O–O distance in the catecholate ion and the surface Ti–Ti distance (in the most occurring (001) and (011) anatasevolving B sites must be highly irreversible, because sur-

face reconstruction (a sluggish process ) should be ex- cleavage planes, this is 3.78 A) , the formation of species IIwould be expected to predominate, particularly at pH valuespected to follow ligand desorption. Incidentally, a com-

bined attack of ligands ( nucleophilic attack ) and protons below pKa1 (cf. Eqs. [19] and [20]) . In fact, surface equilib-rium [20] accounts for both the prevailing adsorption stoi-( electrophilic attack ) on metal-oxo bonds usually leads

irreversibly to dissolution (6 ) ; as mentioned, this pro- chiometry (indicated by the ratio Gmax :NS Å 0.42) and thenoted pH dependence of the adsorption density (Figs. 2 andcess may be involved in OH0 ‘‘adsorption’’ in very basic

media. Therefore, the chemisorption of catechol (or other 3) . Formation of species I is, however, essential to accountfor the increased negative surface charge (i.e., the shifts incomplexing anions ) at the TiO2 / aqueous solution inter-

face will be limited by the availability of A surface sites. iep shown in Fig. 4) . Even though other adsorption modescould also be invoked (see, e.g., Refs. (12–14) and (49)) ,The information provided by the spectrum in Fig. 5

is, however, insufficient to identify the possible Ti –cate- those depicted by Eqs. [19] and [20] suffice to describe ourexperimental results. The corresponding surface equilibriumcholate surface complexes, which therefore have to be

inferred from the observed adsorption behavior. In prin- constants areciple, two surface complexation equilibria may be postu-lated:

K intI Å {I}[H/]

{GTiOH1/30}[H2L]exp(0Fc0 /RT ) [22]

K intII Å

{II}{GTiOH1/30}2[H2L]

, [23]

where H2L denotes catechol. As is usual for surface isocou-lombic reactions, {II} does not depend on surface potential;obviously, any change in c0 may influence {II} through itseffect on the coupled surface equilibria. A quadratic depen-dence of {II} with the surface concentration of GTiOH1/30

(Eq. [23]) was chosen to represent the need of finding twovicinal A sites available for the formation of a 2 to 1 surfacecomplex. It has been suggested, however, that a linear rela-tionship between {II} and {GTiOH1/30} is also consistentwith this requirement (55). Stumm and co-workers (13, 14)have pointed out that the most realistic value of the exponent(n) on {GTiOH1/30} in Eq. [23] must lie in the range 2§ n § 1, and showed that a reasonable description of thestability of binuclear surface complexes can be offered bysetting n Å 1. Instead, we chose to set n Å 2, for it allows fora consistent description of the equilibrium between species Iand II, i.e.,

II / H2L S 2I / 2H/ . [24]

ßTiOH1/32 1

4/32

ßTi 1 H1 1 H¤O [19]

OH

OH

O

O(I)

2ßTiOH1/32 1

The catechol adsorption density is thus given by the sum

G 5 {I} 1 {II}.

2/32

1 2H¤O. [20]

[21]

OH

OH

ßTiO

ßTiO(II)

The same choice was made by Muller and Sigg (55) toreport bS

2 , the stability constant of the (GFeOO)2Pb surfacecomplex.Adsorption in a 1 to 1 GTi:catechol ratio (Eq. [19]) has

already been proposed by Moser et al. (48). In terms of the The set of equations presented in Table 2 also describesthe TiO2/catechol aqueous solution interface. Equations [6]present representation of the OH surface groups, formation



TABLE 4Surface Complexation Equilibria at the P-25 TiO2/Catechol Aqueous Solution Interface

Surface equilibrium log K int

B sites åO2/30 / H/ S åOH1/3/; K intB 7.60

A sites åTiOH1/30 / H/ S åTiOH2/3/2 ; 1/K int

A 5.38

åTiOH1/30 / H2L S åTiL4/30 / H/ / H2O; K int1

a 03.00

2 å TiOH1/30 / H2L S (åTi)2L2/30 / 2H2O; K intII

a 3.20

a Eqs. [19] and [20] were rewritten for simplicity.

and [11] were appropriately modified to include the surface ence of ionic strength. This is illustrated by Fig. 7, wherethe contributions of species I and II to the adsorptioncomplexes I and II:density are plotted as a function of pH and electrolyteconcentration. The marked change of ÌG /ÌpH ( pH õ

NA Å NS Å {GTiOH2/3/2 } / {GTiOH1/30}

pKa1 ) is a consequence of the increased screening effectdue to the swamping electrolyte; it favors protonation of/ {I} / 2{II} [6 *]GTiOH 1 / 30 sites at pH õ pH0 (cf. Fig. 6 ) and enhancesthe contribution of species I at pH values within pH0 and

s00 Å FF23

{GO2/30} / 13

{GTiOH1/30} pKa1 . In a thorough study of the adsorption of catecholon g-Al2O3 in 0.1 mol dm03 NaClO4 , Kummert and

/ 43

{I} / 23

{II} G [11 *]

The values of K intI and K int

II that best characterize the chemi-sorption behavior of catechol onto P-25 TiO2 are includedin Table 4, where the surface equilibria that take place inthis system are summarized. The model parameters listed inTable 3 should not be affected, in principle, by catecholadsorption.

Model calculations presented in Figs. 2 and 3 demon-strate that the multisite surface complexation model ac-counts excellently for the observed chemisorption behav-ior. The agreement between predicted and experimentalz potentials is, however, less satisfactory (Fig. 4 ) . Al-though the dependence of iep with catechol concentrationis fairly well reproduced, the model underestimates theslopes Ìz /ÌpH, particularly in the vicinity of iep. Ad-sorbed catechol may affect the inner structure of the elec-trical double layer ; adsorbed ligands should alter thelocus of specifically adsorbed counterions, an effect thatreflects itself essentially on C1 . A suitable modificationof the internal region of the interface, e.g., following theideas introduced by Smit (56, 57 ) , may improve theperformance of the model.

FIG. 7. The influence of ionic strength on the speciation of surfaceAlthough there is no implicit relationship between the titanium–catecholate complexes: (A) no added background electrolyte; (B)

ligand adsorption density and the concentration of indif- in the presence of 0.1 mol dm03 KCl. Total catechol concentration: 0.001mol dm03 ; available TiO2 surface area: 822.4 m2 dm03 ; T Å 298 K.ferent electrolyte, the model predicts a noticeable influ-



2. Blesa, M. A., Regazzoni, A. E., and Maroto, A. J. G., Mater. Sci.Stumm (13 ) report positive values of ÌG /ÌpH ( pH õForum 29, 31 (1988).p Ka1 ) . This is in contrast with the trend shown in Fig.

3. Dzombak, D. A., and Morel, F. M. M., ‘‘Surface Complexation Mod-3. However, the dissimilar behaviors may be interpreted eling, Hydrous Ferric Oxide.’’ Wiley–Interscience, New York, 1990.on the basis of the predicted influence of ionic strength; 4. Stumm, W., ‘‘Chemistry of the Solid–Water Interface.’’ Wiley–Inter-TiO2 and g-Al2O3 may also behave differently toward science, New York, 1992.

5. Stumm, W., Colloids Surf. A 73, 1 (1993).catechol adsorption due to their different complexation6. Blesa, M. A., Morando, P. J., and Regazzoni, A. E., ‘‘Chemical Disso-chemistry (note also the different values of pH0 ) .

lution of Metal Oxides.’’ CRC Press, Boca Raton, FL, 1994.7. Regazzoni, A. E., Blesa, M. A., and Maroto, A. J. G., J. ColloidComparison with the Single-Site Approach

Interface Sci. 122, 315 (1988).8. Blesa, M. A., Maroto, A. J. G., and Regazzoni, A. E., J. ColloidThe outstanding ability of the multisite surface com-

Interface Sci. 140, 287 (1990).plexation model to describe the chemisorption behavior9. Davis, J. A., James, R. O., and Leckie, J. O., J. Colloid Interface Sci.

of weak acids at metal oxide / aqueous solution interfaces 63, 408 (1978).is largely shared by the single-site versions of the surface 10. Davis, J. A., and Leckie, J. O., J. Colloid Interface Sci. 74, 32 (1980).

11. James, R. O., and Parks, G. A., in ‘‘Surface and Colloid Science’’complexation approach. In fact, the surface equilibria(E. Matijevic, Ed.) , Vol. 12, Chap. 2. Plenum, New York, 1982.depicted by Eqs. [19 ] and [20 ] are compatible with the

12. Stumm, W., Kummert, R., and Sigg, L., Croat. Chem. Acta 53, 291single-site description (only a minor change in notation(1980).

is required ) and, if the appropriate surface equilibrium 13. Kummert, R., and Stumm, W., J. Colloid Interface Sci. 75, 373constants are used, the adsorption of catechol at the P- (1980).

14. Sigg, L., and Stumm, W., Colloids Surf. 2, 101 (1981).25 TiO2 / solution interface can also be accounted for on15. Blesa, M. A., Maroto, A. J. G., and Regazzoni, A. E., J. Colloidthe basis of any suitable single-site surface complexation

Interface Sci. 99, 32 (1984).model.16. Tejedor-Tejedor, M. I., and Anderson, M. A., Langmuir 2, 203

The noticeable influence of ionic strength anticipated (1986).by the present model (Fig. 7 ) is also predicted by the 17. Zeltner, W. A., and Anderson, M. A., Langmuir 4, 469 (1988).single-site models that are based on GCSG structure of 18. Tunesi, S., and Anderson, M. A., Langmuir 8, 487 (1992).

19. Hayes, K. F., Roe, A. L., Brown, G. E., Jr., Hodgson, K. O., Leckie,the electrical double layer (15, 58 ) ; in the latter descrip-J. O., and Parks, G. A., Science 238, 783 (1987).tion, the effect of indifferent electrolyte concentration is

20. Mulcahy, F. M., Fay, M. J., Proctor, A., Houalla, M., and Hercules,accounted for by surface ion-pairing equilibria involvingD. M., J. Catal. 124, 231 (1990).

charged surface complexes and coadsorbed counterions. 21. Pulfer, K., Schindler, P. W., Westall, J. C., and Grauer, R., J. ColloidThe implications of the different EDL structures that are Interface Sci. 101, 554 (1984).

22. Hiemstra, T., van Riemsdijk, W. H., and Bolt, G. H., J. Colloidassumed by single-site surface complexation models hadInterface Sci. 133, 91 (1989).already been discussed in the literature (7, 58 ) .4

23. Hiemstra, T., de Wit, J. C. M., and van Riemsdijk, W. H., J. ColloidAlthough both the single-site and the multisite surfaceInterface Sci. 133, 105 (1989).

complexation models offer an excellent description of the 24. Hiemstra, T., and van Riemsdijk, W. H., Colloids Surf. 59, 7 (1991).chemisorption of catechol at the P-25 TiO2/aqueous solution 25. Sommer, L., Collect. Czech. Chem. Commun. 28, 2102 (1963).

26. Martell, E. A., and Smith, R. M., ‘‘Critical Stability Constants’’ Vol.interface, the internal coherence of all the model assumptions3. Plenum, New York, 1976.favors the use of the present multisite approach that does

27. Borgias, B. A., Cooper, S. R., Koh, Y. B., and Raymond, K. N.,not require the use of an increased number of adjustableInorg. Chem. 23, 1009 (1984).

parameters. 28. Yates, D. E., Levine, S., and Healy, T. W., J. Chem. Soc., FaradayTrans. 1 70, 1807 (1974).

ACKNOWLEDGMENTS 29. Yates, D. E., Ph.D. Thesis, University of Melbourne, Australia, 1975.30. Jones, P., and Hockey, J. A., Trans. Faraday Soc. 67, 2679 (1971).

Partial support by Fundacion Antorchas is gratefully acknowledged. The 31. Jackson, P., and Parfitt, G. D., Trans. Faraday Soc. 67, 2469 (1971).authors are members of CONICET. 32. Boehm, H. P., Disc. Faraday Soc. 52, 264 (1971).

33. Doremieux-Morin, C., Enriquez, M. A., Sanz, J., and Fraissard, J., J.Colloid Interface Sci. 95, 502 (1983).REFERENCES

34. van Veen, J. A. R., Veltmaat, F. T. G., and Jonkers, G., J. Chem.Soc., Chem. Commun. 1656 (1985).1. Hohl, H., Sigg, L., and Stumm, W., in ‘‘Particulates in Water’’

35. Herrmann, M., and Boehm, H. P., Z. Anorg. Chem. 368, 73 (1969).(M. C. Kavanaugh and J. O. Leckie, Eds.) , Advances in Chemistry36. Boehm, H. P., and Herrmann, M., Z. Anorg. Chem. 352, 156 (1967).Series, Vol. 189, p. 1. Am. Chem. Soc., Washington, DC, 1980.37. Allard, K. D., and Heusler, K. E., J. Electroanal. Chem. 77, 35

(1977).38. O’Brien, R. W., and White, L. R., J. Chem. Soc., Faraday Trans. 24 It is important to stress that these implications can only be assessed

74, 1607 (1978).properly through the detailed study of the influence of ionic strength on39. Schindler, P. W., and Gamsjager, H., Disc. Faraday Soc. 52, 286anion chemisorption. In view of the scarcity of data, we have undertaken

this task which will be reported in a future communication. (1971).



40. Schindler, P. W., and Gamsjager, H., Kolloid Z. Z. Polym. 250, 759 49. Borghi, E. B., Morando, P. J., and Blesa, M. A., Langmuir 7, 1652(1972). (1991).

41. Foissy, A., M’Pandou, A., Lamarche, J. M., and Jaffrezic-Renault, 50. Regazzoni, A. E., and Blesa, M. A., Langmuir 7, 473 (1991).N., Colloids Surf. 5, 363 (1982). 51. Blesa, M. A., and Kallay, N., Adv. Colloid Interface Sci. 28, 111

42. M’Pandou, A., and Siffert, B., J. Colloid Interface Sci. 102, 138 (1988).(1984).

52. Westall, J. C., and Hohl, H., Adv. Colloid Interface Sci. 12, 26543. Furlong, D. N., Wells, D., and Sasse, W. H. F., J. Phys. Chem. 89,

(1980).626 (1985).53. Smit, W., and Holten, C. L. M., J. Colloid Interface Sci. 78, 1 (1980).44. Janssen, M. J. G., and Stein, H. N., J. Colloid Interface Sci. 109, 50854. Regazzoni, A. E., Ph.D. Thesis, Universidad Nacional de Tucuman,(1986).

Argentina, 1984.45. Girod, G., Lamarche, J. M., and Foissy, A., J. Colloid Interface Sci.121, 265 (1988). 55. Muller, B., and Sigg, L., J. Colloid Interface Sci. 148, 517 (1992).

46. Miller, N. P., and Berg, J. C., Colloids Surf. 59, 119 (1991). 56. Smit, W., J. Electrochem. Soc. 132, 2172 (1985).47. Avena, M. J., Ph.D. Thesis, Universidad Nacional de Cordoba, Argen- 57. Smit, W., J. Colloid Interface Sci. 109, 295 (1986).

tina, 1993.58. Hayes, K. F., Papelis, C., and Leckie, J. O., J. Colloid Interface Sci.48. Moser, J., Punchihewa, S., Infelta, P. P., and Gratzel, M., Langmuir

125, 717 (1988).7, 3012 (1991).


Documents

Surface Complexation at the TiO2(anatase)/Aqueous Solution Interface: Chemisorption of Catechol