aerobic and anaerobic tio2-photocatalysed purifications of waters containing organic pollutants

Catalysis Today 53 (1999) 145–158

Aerobic and anaerobic TiO2-photocatalysed purifications of waterscontaining organic pollutants

Joseph Cunningham∗, Ghassan Al-Sayyed, Petr. Sedlak, John Caffrey

Chemistry Department, University College Cork, Cork, Ireland

Abstract

Developments reported by European research groups in removal of dilute organic pollutants from water under the combinedaction of TiO2 particles and solar-illumination are briefly outlined. Instances are given of poor correlation between trendsactually observed for selected TiO2*-photocatalysed conversion in aerobic conditions and trends expected within the context oftwo-parameter Langmuir–Hinshelwood-type rate expressions. The necessity of recognising important kinetic roles of severalparameters is emphasised not only for aerobic, but also for anaerobic conditions, e.g. TiO2*-photocatalysed conversions ofchloromethane pollutants in deoxygenated aqueous solutions. Comparisons between initial TiO2-sensitized photocatalyticdegradation (TiO2*-PCD) rates versus dark-equilibrated [C]eq profiles for onestrongly-adsorbedand oneweakly-adsorbedpollutant in aerated TiO2 suspensions at concentrations<100 ppm imply that, under high photon flux, the former can be morestrongly inhibited in consequence of: (i) chemisorption-induced depletions of surface OH− groups; (ii) adsorbate-enhancedhole–electron recombinations on the TiO2* surfaces (iii) mass transport limitations within the TiO2 particle aggregates.Insights into the nature of strongly chemisorbing pollutants (e.g. hydroxybenzoic acids or catechol) onto TiO2(P25) wereobtained by DRIFTS measurements upon freeze-dried TiO2 powder samples after their dark-equilibration with 3–100 ppmsolutions. ©1999 Elsevier Science B.V. All rights reserved.

Keywords:TiO2 photocatalysis; Adsorption isotherm; Chelated adsorbate; DRIFTS spectra; Monochlorophenols; Hydroxybenzoic acid

1. Introduction

Upon receiving the invitation to present a paper onthe above topic at this European Research Confer-ence I felt it would be appropriate to commence withbrief accounts of important recent progress achievedby other European groups in applying TiO2-sensitizedphotocatalytic degradation (TiO2∗-PCD) of various or-ganic contaminants in polluted waters [1–11]. Hope-fully, this introduction will servenot only to indicatethe wide range of organic pollutants whose removalfrom waters can now be effected thanks to innovative

∗ Corresponding author. +353-21-902454; fax: +353-21-274097E-mail address:[email protected] (J. Cunningham)

developments in photocatalytic reactor design and inthe utilization of solar illumination,but also to in-dicate where needs exist for significant revision ofsome frequently utilised mechanistic ideas concerningTiO2

∗-PCD. Subsequent parts of the paper will out-line results from research at University College Corkwhich has been vectored towards identifications of theseveral parameters which affect TiO2*-PCD rates andefficiencies. Implications of the results in respect ofimproved mechanistic ideas are also considered.

An outline representation is given in Scheme 1 ofhow absorption of UV-photons (λ ≤ 390 nm) may re-sult, via TiO2

∗ exciton formation (not shown), in sepa-ration of some excitons into e−/Se, a conduction-bandelectron localised upon Se, an electron-trapping

0920-5861/99/$ – see front matter ©1999 Elsevier Science B.V. All rights reserved.PII: S0920-5861(99)00109-1

146 J. Cunningham et al. / Catalysis Today 53 (1999) 145–158

Scheme 1. Outline of the reaction sequences initiated by near-UV illumination producing photogenerated electrons (reactions as per upperbranch) and photogenerated holes (reactions as per lower branch) originated at surfaces of TiO2 particles/aggregates suspended in aeratedaqueous solutions containing pollutant RCl.

surface site or species,plus h+/Sh a valence bandhole localised upon Sh, a hole-trapping surface site orspecies. The low efficiency of such localization pro-cesses relative to fast recombination of non-trapped orpartially trapped TiO2∗excitons, is mainly responsiblefor the low photonic efficiencies normally obtained inTiO2 sensitized PCD.

The sequence of post-primary events representedon the upper branch of Scheme 1 as originating frome−/Se typify those usually considered to occur at mi-crointerfaces between photoactivated TiO∗

2 particlesand aerated or oxygenated aqueous solutions: viz.electron-capture by O2 at the interface, conversion ofO2

− to HO2, and oxidative attack by the latter uponpollutant species. Gerischer and Heller have expressedthe view [3] that this electron-trapping role of O2 canbe the rate- and efficiency-determining-process, andmany investigators have demonstrated that rigorousdeoxygenation of aqueous solution/TiO∗

2-suspensionsresults in complete loss of capability of many systemsfor TiO2

∗-PCD (but see later for a notable exception).The lower branch of Scheme 1 represents one

version of the sequence of oxidative events initi-ated by those h+/Sh locations which, aided by theelectron-trapping role of O2, escape recombinationwith e−/Se, viz. rapid transfer of h+/Se to one OH−

group within the well-hydroxylated TiO2 surfacelayer, yielding an OH radical which oxidatively at-tacks pollutant molecules at or close to that layer.That simplified representation will not provide a fullyadequate representation of the sequence of oxidativeevents in cases where direct transfer of h+/Sh topollutant molecules adsorbed within the hydrated sur-face layer serve as an additional alternative primarystep towards oxidised product [4,5]. In his studies ofone such system Richard determined the relative ex-tents to which three possible oxidative products from4-hydroxy benzyl-alcohol HBA, – two attributableto OH-radical attack and one to h+ attack – wereproduced by TiO2∗-PCD versus their production byFenton’s reagent [5]. He concluded that positive holesand hydroxyl radicals had different regioselectivitiesin TiO2

∗-PCD of HBA, hydroquinone being producedby h+-attack, dihydroxybenzyl alcohol by OH attackand 4-hydroxy benzaldehyde via attacks by both h+and OH.

In respect of oxidative degradations of a wide rangeof organic pollutants in aerated waters, much valu-able technical progress has been achieved using so-lar illumination to drive TiO2

∗-PCD within the con-text of Scheme 1. Information on several of those pro-cesses have been provided by Dr. Blanco in his lead-in

J. Cunningham et al. / Catalysis Today 53 (1999) 145–158 147

Scheme 2. Complex network of TiO2∗-photogenerated primary transformations of atrazine (A) in aerated aqueous suspensions, showingalso the identities of intermediates formed initially and during their transformations ultimately to non-toxic cyanuric acid (X) [8].

presentation to session – and by various workers inseveral posters at this meeting. Consequently I shallconcentrate on just a few examples, which not onlytypify the impressive progress achieved but can alsoserve to illustrate the need for improved understandingof the basic processes involved. My first pair of exam-ples relate to solar-illumination induced TiO2

∗-PCDof the herbicide atrazine and the wood-preservativepentachlorophenol [6,7]. Run-off from uses of thoseand structurally-related toxic materials have, in someareas, resulted in their build-up to unacceptable levelsin the water table. Removal of toxic atrazine from suchwaters was effected by TiO2∗-PCD, whereby atrazinewas converted to non-toxic cyanuric acid in accor-dance with the following equation: C8H14ClN5 + 15/2

O2 → C3N3(OH)3 + 5CO2 + 5H2O [6]. A large solarplant photocatalytic system was used featuring oneHelioman unit through whose illuminated phototubesa suspension containing 200 mg TiO2 l−1 was recir-culated at 4000 l h−1 under near-UV-intensity equiva-lent to six suns [6]. The on-site confirmation of suc-cess in thus decreasing atrazine concentration to lev-els well below limits specified in EC legislation, albeitwith just 2% efficiency of absorbed near-UV-photonsin effecting removal of the lateral alkyl side chains,had been preceded by very detailed chemical analy-ses to identify whatever intermediates developed fromUV-illumination of aerated (Atrazine)aq/TiO(P25) sus-pensions at various stages [8]. Such information wasthen used to develop Scheme 2 as a representation of


the complex network of individual reactions via whichTiO2

∗-PCD of atrazine to cyanuric acid proceeded.An important benefit from establishment of that com-plex network, on the basis of detailed chemical anal-yses, was the reassurances it provided in respect ofnon-persistence of intermediates (non-toxic or other-wise) by virtue of their rapid consumption en-routeto non-toxic cyanuric acid as sole product. On theother hand, the possibility seemed remote that kinet-ics of TiO2

∗-PCD of atrazine (AZ) proceeding viasuch a complex network of reactions could receive aphysically meaningful representation on the basis of atwo-parameter equation such as

−d[CAZ]

dt∗= kLH

KAZCeq

1 + KAZCeq(1)

A common interpretation of the two parametersKAZ andkLH in such a Langmuir–Hinshelwood-typerate-expression would be that they correspondrespectively to (i) an equilibrium constant foradsorption–desorption of atrazine between its aque-ous solution and the TiO2 surface, and (ii) a specificrate-constant for reaction between adsorbed atrazineand sites/species upon TiO2 activated as per the lowerhalf of Scheme 1.

The low probability that just two such param-eters, having the indicated physical signficances,could be the only factors controlling TiO2∗-PCDprocesses involving complex reaction networks sim-ilar to that illustrated in Scheme 2 was criticallyexamined by Minero et al [9] in respect of solarillumination-induced TiO2∗-PCD of pentachlorophe-nol PCP. In one strand of their critical analysis theycompared parameters obtained from double-reciprocalplots of (extent of PCP adsorption)−1 versus[PCP]eq

−1 and of (initial TiO2∗-PCD rate)−1 ver-

sus (initial [PCP])−1 to deduce experimentally-basedvalues forKeq and kLH. The second strand in theirapproach involved modelling the kinetics expectedon the basis of TiO2∗-PCD proceeding via a reactionnetwork of intermediate reactions of comparable com-plexity to Scheme 2. From this double-stranded ap-proach they concluded that “whilst experimental datacan be fit by an approximate kinetic solution whichhas the analytical form of a Langmuir-Hinshelwoodequation, nevertheless the parameters of such workingmodel equation have been demonstrated not to havephysical significance”. Minero arrived at a similar

conclusion from detailed kinetic analysis of anotherTiO2

∗-PCD process [10]. A healthy scepticism is thusappropriate in respect of values ofkLH and Kads pa-rameters deduced solely on the basis of L–H-based,double-reciprocal-type plots of (rates of aqueous pol-lutant removal by TiO2∗-PCD)−1 versus (aqueouspollutant concentration)−1.

Recent advances in two qualitatively different uti-lizations of TiO2

∗-PCD for water-depollution alsomerit mention in this section, namely: (a) demon-strations that chloromethanes can successfully bedegraded by UV-illumination of oxygen-free aque-ous solutions having TiO2 suspended therein, and (b)photoactivation of TiO2 immobilized upon varioussupports in contact with aerated/oxygenated aqueoussolutions which contain pollutants.

(a) Indications that TiO2∗-PCD of waters pollutedby chloromethanes can proceed by photoactivated pro-cesses additional to those represented in Scheme 1 fol-low from recent reports by Calza et al. of their successin achieving TiO2

∗-PCD of chloromethanes indeoxy-genatedaqueous suspensions [11,12]. Stoichiometryof those processes has been represented by the follow-ing Eqs. (2)

CCl4(aq) + 2H2O(l) → CO2(g) + 4HCl(aq) (2A)

CHCl3(aq) + 2H2O(l) → HCOOH(aq) + 3HCl(aq)

(2B)

CH2Cl2(aq) + H2O(l) → HCHO+ 2HCl (2C)

2CH2Cl2(aq) + 3H2O(l)

→ HCOOH(aq) + CH3OH(aq) + 4HCl(aq) (2D)

The fact that carbon has +4 oxidation number onboth sides of Eq. (2)(A) points to one major mecha-nistic difference between that anaerobic TiO2

∗-PCDprocess and others proceeding by redox reactions inaerated suspensions in accordance with Scheme 1,since those latter usually resulted in increase of theoxidation number of carbon to +4. Pellizzetti andco-workers therefore referred to the conversions inEq. (2)(A–D) being formally hydrolyses or dispro-portionations. Another difference from Scheme 1,which could be implied from the observed occur-rence of the above equations in the absence of O2,


was that the electron-trapping role assigned to O2in the former is replaced by electron-trapping rolesof the chloromethanes. For example, the initial pro-cesses proposed in TiO2

∗-PCD of CCl4 included thefollowing:

CCl4 + e− → CCl3 + Cl− (3A)

CCl4 + 2e− → CCl−3 + Cl− (3B)

CCl4 + 2e− + H+ → CHCl3 + Cl− (3C)

Painstaking chemical analyses of all species presentin illuminated oxygen-free suspensions after variousperiods of illumination resulted in identifications ofall intermediates and made possible time-profiles fortheir production and eventual disappearance during theTiO2

∗-PCD of CCl4, CHCl3 and CH2Cl2 [11]. Calzaet al. thus developed a complex network of reactionevents – including attack by OH, and substitution re-actions e−/–Cl− and OH−/–Cl− and H2O/–HCl; plusthe corresponding reverse processes – with which toaccount for the intermediates observed and pathwaysfollowed en-route to the product stoichiometries indi-cated in Eq. (2)(A–D).

(b) An important practical advantage soughtthrough the use of immobilised TiO2 as the photoini-tiator in TiO2

∗-PCD, as distinct from TiO2 particlessuspended in a polluted aqueous phase, is avoidanceof the need for filtration to remove the suspendedTiO2 particles before return of the depolluted waterto desired re-use. Progress via Scheme 1 should ap-ply, provided that the aqueous phase circulating overthe immobilised TiO2 is oxygenated or aerated. Fullaccounts have been given by Bauer at this conferenceof wide-ranging success in depollution of variouslycontaminated waters by flowing them over a speciallydesigned photoreactor featuring TiO2 immobilisedupon surfaces of a large bundle of thin cylindricalquartz supports. Efficiency of delivery/penetration ofthe UV-illumination to TiO2 thus immobilized is en-hanced relative to its poorer penetration into aqueoussuspensions containing TiO2 particles.

Other recent applications of immobilised TiO2 lay-ers have involved their deposition onto flat surfaceswhich are both transparent and electrically conduct-ing, thereby allowing the combination to serve asan optically transparent electrode (OTE). [13–15].Under UV-illumination from the non-TiO2-coated

side, whilst a suitable positive bias is applied, theOTE tends to withdraw some electrons from theUV-illuminated TiO2 layer thereby substitutingfor the electron-trapping role of O2 (cf. Scheme1), and allowing studies of TiO2∗-PCD relatedprocesses in deoxygenated solutions. ConverselyTiO2

∗-photogenerated holes experience repulsion to-wards the TiO2//aqueous-solution interface and canenhance oxidative degradation in deoxygenated sys-tems [13,14]. Tada et al. have explored the relation-ship between thickness of the TiO2 layer immobilisedupon the OTE and the effectiveness of reverse-sideillumination in causing a measurable physical changeat the outermost surface of the layer [15]. Maximumeffectiveness occurred for layer thickness ca. 100 nm,above which it decreased. They attributed such de-crease to increased percentage loss of photoinducede− and h+ by virtue of increased recombinationswithin greater layer thicknesses. They argued that avalue of 2× 10−2 cm2 s−1 estimated from their resultsfor diffusion of primary photogenerated charge carri-ers were in fairly good agreement with correspond-ing literature values and consistent with lifetimesτ ∼ 100 ns [16].

2. Experimental

2.1. Materials

The titanium powder used for measurements wasDegussa TiO2(P25) powder specified as having 80/20anatase/rutile composition, primary particle size∼30 nm and BET surface area 50 m2 g−1. Surfaceimpurities were removed by overnight preoxidationunder a flow of high-purity, dried O2 at 823 K priorto measurements of the adsorption or photocatalyticproperties at 295 K. HPLC-grade water was used inpreparation of all solutions. Hydroxybenzoic acids(HBA) or phenols (PhOH) were the highest purityavailable from Aldrich.

2.2. Concentration dependence of equilibriumextents of adsorption at 295 K

Following introduction of 200 mg of precal-cined TiO2 plus 100 ml of (HBA)aq or (PhOH)aq


into stoppered and blackened Erhlenmeyer flasks,these were shaken overnight to ensure attainmentof equilibrium-dark adsorption. The unadjustedpH of the suspensions evidenced pH values of5.5± 0.5. A (TiO2 + water) reference was likewisedark-equilibrated. Filtration through Millipore 0.4 or0.2m membrane filters next morning was used to sep-arate the adsorption dark-equilibrated solutions fromthe TiO2 powder. Concentrations of HBA or PhOHin the filtrates were determined by HPLC. Valuesfor the amount removed from solution by virtue ofequilibrium extent of adsorption were then deducedfrom concentration decrease relative to an aliquot ofsolution not contacted with TiO2. They are expressedasns

2, mmol adsorbed per gram of TiO2.DRIFTS measurements upon aliquots of TiO2

powder separated by filtration (as above) from theirequilibrium-dark-equilibrated solutions were per-formed after the following work-up: TiO2 powderretained by the Millipore filter was freeze-dried un-der continuous pumping overnight at−40oC; then itwas permitted to warm-up to 295 K within its evac-uated freeze-drying container before transfer intothe sample holder of a diffuse reflectance accessory(PEDR) and measurement on an FTIR Spectrometer(Perkin–Elmer Paragon 1000). An identical set ofsteps was applied to TiO2 which had been equilibratedwith pure water. Difference-spectra (in Kubelka-Munkform) between that of the TiO2 reference and thosefrom TiO2 samples which had been dark-equilibratedwith a solution having HBA or PhOH concentrations3, 10, 30 or 100 ppm were obtained and compared.

2.3. Concentration and intensity dependence ofTiO2

∗-photoinitiated rates of decrease in [HBA]aqand [PhOH]aq from Ceq their dark-equilibratedconcentrations

Dark pre-equilibration was first established be-tween 30 ml of solution and 60 mg of precalcinedTiO2 powder in a stirred photoreactor. After flushingwith pure O2 or purified air, this was sealed. Provisionwas made for later syringe-extraction of small aliquotsof the stirred suspension via rubber septa sealedside arms. After≥1 h pre-equilibration of the stirredsuspension in the dark, illumination commencedwith photons havingl| = 365± 10 nm (selected by

interposing the appropriate narrow-band-pass filterbetween the transparent bottom window of the pho-toreactor and a 125 W medium-pressure Hg lamp).This yielded a flux of 2× 1018 photons min−1 inci-dent into the photoreactor volume. When so desireda medium flux of 5.6× 1017 photons min−1 or alow flux of 1.7× 1016 min−1 (as determined usingchemical actinometer solutions) were obtained byinterposing metal gauze screens in the light path. Af-ter various durations of illumination, samples of thecontinuously stirred and illuminated suspension weresyringe-extracted, filtered through 0.2mm membranesand the clear filtrate analysed by HPLC. Concentra-tions thus determined were plotted versus duration ofUV-illumination.

3. Results and interpretation

3.1. Substantial qualitative difference between ‘dark’adsorption isotherms of 3-chloro-4-hydroxybenzoicacid (CHBA) and monochlorophenols (ClPhOH)

The upper and lower parts of Fig. 1 respectivelyillustrate equilibrium dark adsorption isotherms at295 K for CHBA onto oxidatively precleaned surfacesof TiO2(P25) or for the three monochlorophenols insimilar conditions [17,19]. Comparison between themestablishes the following significant differences; (i)at Ceq concentrations≤ 50 ppm, values ofns

2 for theequilibrium extent of adsorption of the hydroxyben-zoic acid at eachCeq were clearly much larger thanfor the chlorophenols, as evidenced in particular by at-tainment of a plateau value equivalent to 100mmol/gmTiO2 at [CHBA]eq ∼40 ppm, whereas plateau valuesfor the chlorophenols in the same range were almosttwo-orders-of-magnitude lower; (ii) at higherCeq val-ues the isotherms in both figures evidence an approx-imately linear increase inns

2 above their respectiveplateau values, which classical adsorption–desorptionmodels might interpret as evidence for weaker addi-tional adsorption into layers outside the TiO2/aqueoussolution monolayer [20]. Net extent of equilibriumadsorption of the CHBA nevertheless remained atleast one order-of-magnitude greater than those ofthe monochlorophenols across the range studied[17,18].


Fig. 1. Equilibrium ‘dark’ adsorption isotherms data at 295 K(ns

2 = amount adsorbed per gram TiO2) versusCeq, equilibriumaqueous phase concentration, between precalcined TiO2(P25) andaerated aqueous solutions of: (a) 3-chloro-4-hydroxybenzoic acid(CHBA); and (b) monochlorophenols 2-ClPhOH,h; 3-ClPhOH,X; and 4-ClPhOH,5.

3.2. Inverse ratio between initial TiO2∗-PCD rates,−R∗

i of CHBA and ClPhOH

The following assumptions are implicit in thewidespread, but not universal, acceptance of somevalidity in applying to TiO2

∗-PCD processes takingplace in aqueous TiO2suspensions, rate equationssimilar to those developed by Langmuir, Hinshel-wood, Hougen and Watson in respect of heteroge-neously catalysed reactions at Gas/Solid interfaces[21]: firstly, that reaction events with rate coefficientkLH, involving active species/sites photogenerated atthe TiO2

∗/aqueous solution interface and some pol-

lutant species adsorbed thereon, constitute the rate-and efficiency-determining process: and secondly,that adsorption–desorption equilibria between pollu-tant species and the TiO2 surface are sufficiently fastas to be non-rate-determining. However, Eq. (4) be-low shows that they may influence the overall rates,-d[C]/dt of TiO2

∗-PCD processes via magnitude of theequilibrium constant,Kads, for adsorption–desorptionof pollutant onto the TiO2 surface.

−d[Ceq]

dt= kLH

KadsCeq

1 + KadsCeq(4)

On the basis of rate equations of that form, andthe much larger value of Kads evidenced for CHBAthan for monochlorophenols in Fig. 1, a much largerTiO2

∗-PCD rate was to be expected for−d[CHBA]/dtthan for −d[ClPhOH]/dt; – unless compensated forby larger inverse changes in the relative magnitudesof kLH in those two classes of pollutant. Evidence foroperation of some such inverse change, causing theapparentkLH for TiO2

∗-PC removal of CHBA to bevery muchlower than for the monochlorophenols [22],is contained within Figs. 2 and 3 and discussed in thenext paragraphs.

Each data-point on plots within Figs. 2 and 3orginated from a separate TiO2

∗-PC run, duringwhich several samples were syringe-extracted fromthe continuously-stirred suspension after selected du-rations of illumination. HPLC analysis of filtrate fromthe extracted samples yielded values for [CHBA]aqor [ClPhOH]aq remaining after each duration, whichwere plotted versus illumination time. Initial slopesdetermined from such plots, termed−Ri

∗, were inturn represented graphically versus initialCeq valuesto obtain the profiles shown in Figs. 2 and 3. It washoped that−Ri

∗ values evaluated this way, whilstTiO2

∗-induced decreases fromCeq were ≤10%,would minimise adsorption or kinetic interferencesfrom photoinduced intermediates or products. Com-parisons made between−Ri

∗ values for CHBA inFig. 2 and−Ri

∗ values in Fig. 3 for ClPhOH, ob-tained at equivalentCeq values and under identicalphoton flux, reveal that in almost all situations−Ri

∗for the poorly-adsorbing ClPhOH was substantiallyhigher than−Ri

∗ for the well-adsorbing CHBA. In-verse changes in the apparentkLH rate coefficients,sufficient to overcome the greater adsorption ofCHBA, could be inferred from such comparison. One


Fig. 2. Plots of initial rates,−R∗i (CHBA), for TiO2

∗-induceddecrease from dark pre-equilibrated aqueous-phase concentrationsof CHBA, under ‘high’ flux of 365 nm photons (A); ‘moderate’flux (B) and ‘low’ flux (C) for aerated TiO2 suspensions, havingnatural (unadjusted) pH, or with pH adjusted to 10 by (NaOH)aq.

Fig. 3. Plots of initial rates,−Ri∗(4-CP), for TiO2

∗-induced de-crease from dark pre-equilibrated aqueous phase concentrations of4-ClPhOH under ‘high’ flux of 365 nm photons (A), ‘moderate’flux (B) and ‘low’ flux (C). All data for aerated TiO2 suspensionsat natural pH, except plot (ii) in (A) which relate to preoxygenatedsuspensions.


such process previously proposed [23] is a capabilityof CHBA species chemisorbed onto TiO2 to serveas locations capable of enhancing the efficiency of(e− + h+) recombination events relative to that atTiO2/aqueous solution interfaces onto which suchchemisorption did not occur.

Another unexpected difference between the−Ri∗

versusCeq profiles for CHBA in Fig. 2 and ClPhOH inFig. 3 was the occurrence of regions ofnegative-orderdependence uponCeq for the former, but not forthe latter. However, it was only under highest fluxused, 2× 1018 photons min−1, that negative-orderdependence of−Ri

∗(CHBA) upon Ceq operated be-tween 100 ppm and the lowestCeq values, (cf. Fig.2a). Under moderate flux, positive-order was brieflyevidenced atCeq up to 5 ppm before a changeoverto negative-order at 5–80 ppm (Fig. 2b). Under thelowest flux used, negative-order was not observed.A not-uncommon cause for onset of negative-orderbehaviour in heterogeneously catalysed reactions isblockage of active sites by some strongly adsorbedproduct or by-product. Blockage of TiO2∗-active sitesby photogeneratedproduct from TiO2

∗-PCD of ad-sorbed CHBA seemed a somewhat analagous possiblecause for the negative-order in Fig. 2a and b, espe-cially since no such blockage was evidenced underlow photon flux.

The fact that even under highest photon flux thepre-100 ppm negative-order dependence of−Ri

∗ uponCeq was succeeded atC≥ 100 ppm by a region ofzero-order dependence indicated that not all routesto TiO2

∗-photoinitiated removal of CHBA were sus-ceptible to such blockage by photogenerated product.On the other hand zero-order dependence predomi-nated the profile for variation of−Ri

∗ of poorly ad-sorbing ClPhOH uponCeq in high flux conditions (cf.Fig. 3a), with no region of negative-order. Further-more, that rate was ca. five-fold greater than even-tually attained by CHBA in its zero-order dependentrange. Our earlier hypotheses [23] in explanation ofthe latter qualitative differences were: (a) that in thecase of poorly adsorbing ClPhOH, thepredominantfate of photogenerated h+/Sh entities at the illumi-nated TiO2

∗/solution interface (cf. Scheme 1) wastheir reaction with OH− or H2O to yield OH radicals;and (b) that the latter were capable of diffusing out-ward into near-surface layers where their reaction withnon-adsorbed ClPhOH was assured. The latter hypoth-

esis is now disfavoured by more recent results fromexperiments of Meisel and Serpone [24] and Minero etal [25] indicating that active species (including OH(s))formed upon irradiation do not migrate in solution.On the other hand, organic species may/may-not mi-grate to the catalyst surface. It became desirable, there-fore, to consider whether contributory factors to thestrong differences between kinetic behaviour of thewell-adsorbing CHBA and poorly-adsorbing ClPhOHcould be differences: (I) in migration of the substratesto photoactive sites; and/or (II) between surface den-sities of OH −

(s) , capable of OH−(s) + h+ → OH(s),at microinterfaces between TiO2 and [ClPhOH]aq or[CHBA]aq.

(I): Light-scattering measurements upon aqueoussuspensions of TiO2(P25) point to the existence of1–5mm sized aggregates of the 30 nm sized primaryparticles [26,27]. Relative difficulties for in-diffusionof weakly-adsorbed and strongly-adsorbed pollutantspecies from bulk solution inward towards surfacesites on TiO2 particles deep within the aggregates,thus becomes another possible source of differencebetween strongly-chemisorbing hydroxybenzoic acidsand poorly-adsorbing monochlorophenols. Slow out-ward diffusion, operating in converse manner uponstrongly adsorbed photoproduct, may also become arate-limiting process through lengthening the time re-quired for vacating active TiO2∗ sites upon particlesdeep within the aggregates. Recently, Bouchy et al.have concluded from their detailed studies of timeprofiles for approach to equilibrium dark-adsorptionin TiO2 suspensions, that mass transport can indeedbe relatively slow through interparticle spaces withinsuch TiO2 aggregates e.g. with diffusion coefficientscomparable to liquid diffusion in zeolites [27]. In thelight of those findings, possible influences of suchinternal mass transport limitation upon efficiency ofTiO2

∗-PCD merit serious consideration, especially inrespect of the contrast between strongly negative-orderof −Ri

∗ versus [CHBA]eq @ 10–100 ppm (cf. plot Aof Fig. 2) and zero-order dependence for−Ri

∗ versus[4-CP]eq across the same range (cf. plot A of Fig. 3).Bearing in mind the low extent of adsorption of 4-CPonto TiO2 relative to the much more extensive darkadsorption of CHBA across the 10–100 ppm concen-tration range (cf. Fig. 1a), the occurrence of a pro-nounced negative-order effect only for TiO2

∗-PCD ofCHBA might be rationalised as follows: animpeding


effect of strongly adsorbed CHBA upon the internalmass transfer processes required to effect removal ofwell adsorbed photoproduct from within the aggre-gate out to the bulk solution, and conversely an im-peding effect of strongly adsorbed photoproduct uponin-diffusion of adsorbing CHBA species from bulk so-lution via interparticle spaces to vacated TiO2 siteswithin the aggregates.

(II): Dissatisfaction with the somewhat indirectnature of the solution-based evidence leading to theabove-noted rationalizations of differing behavioursof CHBA and 4-CP atCeq values in the range10–100 ppm, caused us to turn to measurements byDRIFTS in efforts to gaindirect insights into theextent/nature of adsorption of hydroxybenzoic acidsor phenols onto TiO2(P25) from aqueous solutionwith concentrations in that range. Literature resultsconcerning such points had come from the researchof Tunesi and Anderson [28] and of Martin et al.[29] who respectively reported results from in-situCylindrical Internal Reflection (CIR) and AttenuatedTotal Reflectance (ATR) FTIR techniques to obtaininsights into the species present at actual Aqueoussolution//TiO2 interfaces. Both investigations usedTiO2 coated as a thin layer onto the external sur-faces of a zinc selenide cylinder or trapezoid sur-rounded by aqueous solution containing the adsorbateof interest. Particularly strong characteristic IR ab-sorbances originating from pollutant species presentat the TiO2/aqueous solution interfaces were re-ported for 2-hydroxybenzoic acid (salicylic acid) [28]and for 4-chloro catechol [ClCT] from dilute aque-ous solutions [29]. These were attributed to probemolecules strongly chemisorbed onto hydroxylatedsurface Ti4+ sites in chelate-like fashions, denotedas bidentate-mononuclear or bidentate-binuclear(cf. Fig. 4a and b). Schematic representations ofsite-specific chemisorption processes proposed toresult from chelation of the catechol species ontoTiO2 are reproduced in Fig. 4A, from which it maybe noted that, for each ClCT complexed to the sur-face in bidentate-binuclear fashion, two hydroxideions previously present upon TiO2 are displaced.Chelate-like chemisorption of 2-hydroxybenzoicacid in mononuclear-bidentate manner onto sin-gle four-coordinate Ti4+ cations upon anatase hadearlier been proposed by Tunesi and Anderson[28].

The primary objective of our exploratory DRIFTSmeasurements was to discover whether insightssimilar to those just summarised from CIR andATR-FTIR measurements [28,29] might be obtainedby applying DRIFTS measurements to TiO2(P25)samples onto which salicylic acid or catechol hadbeen dark-adsorbed in conditions closely similar tothose employed in our determinations of adsorptionisotherms.

The first stage in our procedures towards obtainingDRIFTS spectra of salicylic acid or catechol stronglyadsorbed onto/into TiO2 particles/aggregates involvedovernight dark-equilibration between samples of pre-calcined TiO2 powder and aqueous solutions withoriginal concentrations 3–100 ppm. Following filtra-tion through a 0.2mm membrane, the TiO2 collectedthereon was freeze-dried overnight, before mountingthe fully dried powder in the diffuse reflectance ac-cessory and obtaining a DRIFTS spectrum therefromin standardised conditions. A fortuitous outcome ofthe freeze-drying work-up was the very fluffy natureof the resultant TiO2, which respresented a conditionparticularly suited for diffuse reflectance-based de-tection of adsorbates retained on the TiO2 particles.Fig. 5 illustrates a DRIFTS spectra thus obtainedfrom dark-equilibration of TiO2 with a 30 ppm so-lution of catechol (after subtracting out a referenceDRIFTS spectrum measured in identical conditionsupon TiO2 dark-equilibrated with pure H2O). A sat-isfactory match existed between the two principalfeatures evident in our DRIFTS spectra at 1486 cm−1

and 1261 cm−1 and features at 1484 and 1268 cm−1

in the IR-data reported by Martin et al [29] whoassigned theirs respectively to C=C normal modesand C–O vibrations in the doubly-deprotonated an-ion produced from ClCT following its chelation tothe surface in the manner of Fig. 4A. Making dueallowance for presence of the chloride substituent intheir ClCP [29], there is satisfactory correspondencebetween our DRIFTS and their ATR measurements.Such correspondence may even extend to include asmall but highly reproducible feature at 1328.7 cm−1

clearly evident in our DRIFTS spectra from TiO2dark-equilibrated with [CT]aq= 30, 10 or 3 ppm anda small feature just detectable in their publishedATR-FTIR from ClCT/TiO2 at ∼1320 cm−1.

Fig. 5b illustrates the more complex pattern ofIR features observed by our DRIFTS measurements


Fig. 4. Schematic representations (adapted from [28,29]) of catechol species from aqueous solution onto hydroxylated TiO2 (anatase) surfacevia mononuclear-bidentate-type bonding, A(i); or binuclear-bidentate-type bonding, A(ii); and in B(i) and B(ii) for 2-hydroxybenzoic acid(salicyclic acid).

upon TiO2 dark pre-equilibrated overnight with a10 ppm aqueous solution of 2-hydroxybenzoic acid,followed by filtration and freeze-drying. The follow-ing pairwise listing (DRIFTS versus CIR) of valuesfor peak positions attributable to adsorbate in the twomethods shows adequate correspondence to supporta conclusion that the indicated features originatedin each case from closely similar salicylate anionschelated to surface Ti4+ cations: 1570 versus 1575or 1538 cm−1, consistent with COO− stretches; 1450versus 1457 cm−1, consistent with C–O stretch; and1399/1367 doublet versus 1386/1367 doublet fromcoupling between COO− and C–O bonding. Furtherwork is needed, however, to identify origins of otherpeaks in DRIFTS spectrum (Fig. 5b).

Thus, relative to IR features reported in the litera-ture on the basis of CIR/ATR-FTIR measurementsupon chemisorbed species present at actual (salicylicacid)aq/TiO2 or (ClCT)aq/TiO2 interfaces [28,29],rather similar features were detected by our DRIFTS

measurements upon ‘dry’ (salicyclic acid)ads/TiO2and ‘dry’ (catechol)ads/TiO2 samples obtained byfreeze-drying TiO2(P25) powder samples separatedfrom the corresponding dark pre-equilibrated sus-pensions having initial concentrations as low as3 ppm. Initially this occasioned some surprise, inview of measures used to control ionic-strength in theATR-FTIR measurements [29], versus their completeabsence from our procedures leading to the DRIFTSspectra. However, if strength of the chelation-typebonds between four-coordinated Ti4+ cations andchemisorbed salicylate or ClCT ions were such as topredominate over all other surface interactions, thesimilarities in FTIR spectra from the two techniquescould be understood. Furthermore, the close similar-ity maintained in our studies between dark-adsorptionpretreatments prior to DRIFTS measurements thereon,or measurements of initial ratesR∗

i of TiO2∗-PCD

thereon, allows the following important inferenceto be drawn: namely, that the number of surface


Fig. 5. (a) DRIFTS spectrum from precalcined TiO2(P25), after dark pre-equilibration with aqueous solution containing 30 ppm catechol,recovery by filtration, and overnight freeze-drying. (b) DRIFTS spectrum from precalcinced TiO2(P25), after dark pre-equilibrated withaqueous solution containing 10 ppm 2-hydroxybenzoic acid, recovery by filtration and overnight freeze-drying.

hydroxyls remaining upon TiO2 pre-equilibrated withan adsorbate capable of chelating the surface in waysanalagous to those illustrated in Fig. 4 can be sub-stantially diminished relative to those remaining uponTiO2 pre-equilibrated with weakly adsorbing species

which do not chelate surface titanium ions in the man-ner of Fig. 4. The negative-order behaviour evidencein Fig. 2A for Ri

∗ of the well-adsorbing CHBA may,in part, originate from such an effect, since decreasedavailability of surface hydroxyls to capture photogen-


erated holes would diminish the availability of surfaceOH radicals for oxidation attack upon organics. Suchan effect would appear to be the converse of a recentsuggestion by Nosaka et al. [30] to the effect ‘thatthe larger the amount of surface OH groups, the morelikely the oxidation via surface-trapped holes’.

Also of interest in respect of limitations/capabilitiesof DRIFTS measurements upon Adsorbate/TiO2 sam-ples obtained after freeze-drying procedures werepreliminary assessments of (a) the lowest detectablecoverage of the freeze-dried TiO2 surface by salicy-late or catechol species held thereon in a site-specificmanner, as implied by the sharp IR features in Fig.5a and b; and (b) whether or not the proceduresmight point to the development of additional typesof adsorbed species at the higher extents of surfacecoverage anticipated from dark pre-equilibration with[SA]aqor [CT]aq� 10 ppm. Valuable informationconcerning those points at ‘wet’ interfaces had beendeduced by Martin et al. from their detailed analysesof ATR-FTIR spectra of 4-chlorocatechol [29]. Sinceour DRIFTS measurements TiO2 freeze-dried afterdark pre-equilibration with very dilute, 3 ppm aque-ous solutions of CT or SA reproduced the IR featuresin Fig. 5a and b, except for ca. five-fold decrease inpeak intensities, it could be concluded: (i) that sitesindistinguishable from those occupied from 10 ppmsolutions were likewise involved in site-specificchemisorption at 3 ppm; and (ii) that fractional cov-erage equivalent to one twentieth of the total surfacearea of the dark pre-equilibrated TiO2 remained de-tectable. On the other hand, DRIFTS spectra fromTiO2 dark pre-equilibrated with solutions having30 ppm or 100 ppm of catechol or salicylic acid didnot yield proportionate increases in intensities ofthe previously sharp IR features. Instead significantbroadening and slight shifts in band maxima, occurredas confirmed by DRIFTS ‘difference-spectra’ betweensamples exposed to 30 ppm and 100 ppm solutions.Heterogeneity of the TiO2 surface is thus indicated inrespect of chemisorption of salicylate ions or catecholfrom aqueous solutions in the 30–100 ppm range.Further work will be needed to ascertain whether suchheterogeneity of the TiO2(P25) might adequately beaccounted for on the basis of site-specific chemisorp-tions ontotwo such distinguishable sites [28,29], orrequire consideration of a wider range of chemisorp-tion sites.

4. Conclusions

As an antidote to the preoccupation with two-parameter, Langmuir-Hinshelwood-type mechanismsand rate expressions in much of the literature concern-ing TiO2

∗-PCD of aqueous-phase organic pollutants,the foregoing text and results stress the need for duecognisance to be taken of the intrinsically multivari-able character of the TiO2∗-PCD process. The stronginfluences of nature/strength of pollutant adsorptiononto TiO2 in respect of such interdependences isevidenced by the following:

1. Chemisorbed versus physisorbed nature of pollu-tant adsorption and, in the case of micron-sizedTiO2-aggregates in aqueous suspensions, largeconsequential differences in extent of internal-mass-transport limitations through interparticlespaces between 30 nm TiO2(P25) particles withinthe micron-sized aggregates.

2. In respect of the nature of primary conver-sions experienced by TiO2∗-photogeneratedcharge-carriers at monolayer TiO2

∗/aqueous so-lution micro-interfaces, these too can be moresignificantly affected by chemisorbed species e.g.through decreases in [OH−]s.

3. Under high flux of near-UV photons, verydifferent order-of-reaction is observed in ki-netics of post-primary redox processes dur-ing TiO2

∗-PCD over micron-sized TiO2 ag-gregates pre-equilibrated respectively withstrongly-adsorbing or weakly-adsorbing pollutantfrom 50± 40 ppm solutions. These differencescan be rationalised when conclusions (1) and (2)above are taken into account.

Acknowledgements

Those aspects of the reported work which were car-ried out at University College Cork benefited greatlyfrom support received under EC contracts EV4V0068and STEP 0106-C.

References

[1] D. Bahnemann, J. Cunningham, M.A. Fox, E. Pelizzetti, P.Pichat and N. Serpone, in: G.R. Helz, R.G. Zepp, D.G. Crosby


(Eds.), Aquatic and Surface Photochemistry, chap. 1, LewisPublishers, Boca Raton, Florida, 1994, pp. 261–316.

[2] L. Amalric, C. Guillard, P. Pichat, Res. Chem. Intermed. 20(1994) 579.

[3] H. Gerischer, A. Heller, J. Phys. Chem. 95 (1991) 5261.[4] S. Tunesi, M. Anderson, J. Phys. Chem. 95 (1991) 3399.[5] C. Richard, J. Photochem. Photobiol. A. Chem. 72 (1993)

179.[6] C. Minero, E. Pelizzetti, S. Malato, J. Blanco, Solar Energy

56 (1996) 411 and 421.[7] C. Minero, E. Pelizzetti, S. Malato, J. Blanco, Chemosphere

26 (1993) 2103.[8] E. Pelizzetti, C. Minero, V. Carlin, M. Vincenti, E. Pramauro,

Chemosphere 24 (1992) 2103.[9] C. Minero, E. Pelizzetti, S. Malato, J. Blanco, Solar Energy

56 (1996) 421.[10] C. Minero, Solar Energy Materials and Solar Cells 38 (1995)

421.[11] P. Calza, C. Minero, E. Pelizzetti, Environ. Sci. Technol. 31

(1997) 2198.[12] P. Calza, C. Minero, E. Pelizzetti, J. Chem. Soc. Faraday

Trans. 93 (1997) 3765.[13] K.A. Gray, U. Stafford, Res. Chem. Intern. 20 (1994) 835.[14] K. Vinodgopal, S. Hotchandani, P. Kamat, J. Phys. Chem. 97

(l993) 9040.[15] H. Tada, M. Tanaka, Langmuir 13 (1997) 360.[16] M. Graetzel, Energy Resources Through Photochemistry and

Catalysis, Academic Press, New York, 1983.

[17] J. Cunningham, G. Al-Sayyed, J. Chem. Soc. Faraday Trans.86 (1990) 3935.

[18] J. Cunningham, P. Sedlak, J. Photochem. Photobiol. A. Chem.77 (1994) 255.

[19] J. Cunningham, G. Al-Sayyed, S. Srijaranai, in: G.R.Helz, R.G. Zepp, D.G. Crosby (Eds.), Aquatic and SurfacePhotochemistry, chap. 22, Lewis Publishers, Boca Raton,Florida, 1994, p. 317.

[20] R.P. Hansen, R.P. Craig, J. Phys. Chem. 58 (1954) 212.[21] M. Boudart, G. Djega-Mariadassou, Kinetics

of Heterogeneous Catalytic Reactions, Princeton UniversityPress, Princeton, NJ, p. 92 and 106.

[22] J. Cunningham, P. Sedlak, Catal. Today 29 (1996) 309.[23] J. Cunningham, P. Sedlak, in: D.F. Ollis, H. Al-Ekabi (Eds.),

Photocatalytic Purification and Treatment of Air and Water,Elsevier, Amsterdam, 1993, pp. 67–81.

[24] D. Meisel, N. Serpone, J. Phys. Chem. 95 (1991) 199.[25] C. Minero, F. Catozzo, E. Pelizzetti, Langmuir 8 (1992)

481.[26] R.I. Bickley, T. Gonzalez-Carreno, J.S. Lees, L. Palmisano,

R.J.D. Tilley, J. Solid State Chem. 92 (1991) 178.[27] H.Y. Chen, O. Zahraa, M. Bouchy, F. Thomas, J.Y. Bottero,

J. Photochem. Photobiol. A. Chem. 85 (1995) 179.[28] S. Tunesi, M. Anderson, Langmuir 8 (1992) 487.[29] S. Martin, J.M. Kesselman, D.S. Park, N.S. Lewis, M.F.

Hoffmann, Environ. Sci. Technol. 30 (1996) 2535.[30] Y. Nosaka, N. Kishimoto, F. Nishino, J. Phys. Chem. 102

(1998) 10279.

aerobic and anaerobic tio2-photocatalysed purifications of waters containing organic pollutants

Documents