Scribd Logo
Upload

Welcome to Scribd!

  • 97-JPPA - An Overview of Semiconductor Photocatalysis (A. Mills)

    Uploaded by

    Diego Pacherrez Gallardo
    122 views35 pages
    The semiconductor titanium dioxide (TiO2) features predominantly in past and present work on semiconductor photocatalysis; as a result, in most of the examples selected in this overwie to ilustrate various points the semiconductor is TiO2.

    Original Title

    97-JPPA_ an Overview of Semiconductor Photocatalysis (a. Mills)

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    The semiconductor titanium dioxide (TiO2) features predominantly in past and present work on semiconductor photocatalysis; as a result, in most of the examples selected in this overwie to ilustrate various points the semiconductor is TiO2.

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    122 views35 pages

    97-JPPA - An Overview of Semiconductor Photocatalysis (A. Mills)

    Uploaded by

    Diego Pacherrez Gallardo
    The semiconductor titanium dioxide (TiO2) features predominantly in past and present work on semiconductor photocatalysis; as a result, in most of the examples selected in this overwie to ilustrate various points the semiconductor is TiO2.

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    of 35
    ie ass oss ——_ araoe ic ELSEVIER Toa of trey ad Pail Chem 108 (197) 1-38 ern An overview of semiconductor photocatalysis Andrew Mills *, Stephen Le Hunte Deparment of Chemis, Unveiof Wales Suns, Snglton Po, Swansea SA2 SP, UK eyo: Seeondicir peas: Heeopetess aye 1. Introduction us) isan example of “Semiconductor particles as phoo- ‘The interest in heterogeneous photocatalysisisimtense and increasing, as shown by the number of publications on this theme which regularly appear in this journal, nd the fact that cover 2000 papers have been published on this topic since 1981, Thisartcleis an overview ofthe field of semiconductor photocatalysis: a brief examination of its 00s, achievements, and possible future. Scheme 1 identities some of the past and present guises of semiconductor photochemistry and associ ated major reviews [1-33]. The semiconductor titanium dioxide (TiO;) features predominantly in past and present ‘work on semiconductor photocatalysis; sa result, in mostof the examples selected in this overview to illustrate various points the semiconductor is TiO. ‘Most of the articles on semiconductor photocatalysis in this volume ft comfortably under atleast one ofthe heading in Scheme 1. Thus the bulk of the papers in tis volume fal largely under the heading of: “Semiconductor particles as ‘photocatalyst fr the oxidation of organi pollutants by oxy {gens ths isnot at all unepresentative and simply reflects the prevalent strong research ativity in this are; these papers, ‘include the work of: Singhal etal. (Photocatalytic degrada tion of ceylpyridinium over tanium dioxide powders), KE, (O'Shea et al. (Photocatalytic decomposition of organophos- jhonates in iradiated TiO, suspensions), M. Bouchy et al (Inhibition of the adsorption ad the photocatalytic degra- dation ofan organic contaminant in a TiO, aqueous suspen- sion), B. Viswanathan etal. (Synthesis, characterization and, photocatalytic properties of iton-doped TiO, catalysts), PR. Hills et al, (Photocatalytic detoxification of aqueous 1.4- dioxane) and Maurete etal. (Photocatalytic degradation of 2,4-dihydroxybenzoic aid in water: efficiency, optimization and mechanistic investigations) ‘The work of Richard etal. (Photocatalytic transformations of aromatic compounds in aqueous zine oxide suspensions effect of substrate concentrations onthe distribution of prod *Coneponding stor. Tel 1010-603097/$17.00 © 1997 Evie Seance S.A Allighs served PHrsi010-6030(97)00118-3 sensitizers for organic photosynthetic processes", and the work of B. Viswanathan et al. (Photocatalytic reduction of nitvite and nitrate ions to ammonia on M/TiO, catalysts) is an example of ‘Semiconductor panicles as photocatalysts, for the removal of inorganic pollutants” (Scheme 1) Finally, the work of Lobedank eta. (Sensitized photocatal- yflicoxidation of herbicides using natural sunlight) hasaclose association with recent workon “"Semiconductorcolloidsand panicles” (Scheme 1), a8 we shall see ‘The work of D. Kataks etal. (New catalysts in the photo- oxidation of water) and Navio etal. (UV photosssisted deg- radation of phenylin(IV) chlorides in water) are both exam- ples of homogeneous photochemistry (the latter work is not an example of photocatalysis) and fll ouside the scope of this overview. Although Scheme 1 forms the basis ofthe main part of this ‘overview, it seems appropriate to begin with a brief look at terms and basic principles, 2, Photocatalysis and photosynthesis ‘The term “photocatalyss™ is still the subject of some debate. For example itis argued [34] thatthe idea of a ‘photocatalysed reaction is fundamentally incorrect, since it implies that, in the reaction, light is acting as a catalyst, whereas it always act as a reactant which is consumed inthe chemical proces. In reality the term ~ photocataysis™ i ia Widespread use and is here o sta: i is not meant to, nor should it ever be used to, imply catalysis by light, but rather the “acceleration of a photoreaction by the presence of a catalyst". The term “‘photoreaction"’ is sometimes elabo- rated on as a “photoinduced” or “photosctivated” reaction, all to the same effect, The above definition of “photocata- lysis" includes the process of “photosensitization™, ie. process by which a photochemical alteration occurs in one chemical species as a reslt ofthe initial absorption of radi- ation by another chemical species called the photosensitizer. A. Mis, Le Hane ural of Photochemistry and Pho A Cems 108 (197) 1-15 Early work on smoondutorpotaenemitny = ‘Protoeacrochericl els or wae splting?> ‘Somicondutr paces 25 photcatasts forthe reductonostation of water and onorateceposhO* Semiconeutor paces 2s photocalayss or ne exdeten at organ potants by ange! = pholcstalts or the potas, ncting Semiconductor paricies Semiconductor pares Photodedtuctin of phetoyramiterapy ane ctocintectio’®228 ana tins torte photocatayte destruction of geneous crgane pats proto deodcurisaton& sat leaning phototime 2 pretodssceeion watt" Photoredutn of Ne = Photoslecrochemical aor elcrety producon®® Semicond Nanoaytaioe oe ‘oles ana parts 1 follows from the above that heterogeneous photocata: Isis involves photoreations which occur atthe surface of catalyst. Ifthe intial photoexcitation process occurs in an adsorbate molecule, which then ineracts with the ground state of the catalyst substrate, the process is referred to asa “catalysed photoreaction". If on the other hand, the initial, photoexcitation takes place in the catalyst substrate and the ‘hotoexcited catalyst then interacts with the ground state adsorbate molecule, the process is a “sensitized photorea- tion’, In most cases, heterogeneous photocatalysis refers to semiconductor pholocatalysis or semiconductor-sensitized ‘Photoreactons is sometimes argued tha the term “catalysis” should not be used unless it has been demonstrated thatthe tumover number (TN) for the process (i.e. the number of product ‘molecules per number of active sites) is greater than unity: if this is not demonstrated, the term “*heterogencous (oF semiconductor)-ssisted phocoreaction" should be used. In ‘most examples of heterogeneous photocaalysis, the value of ‘TN has not been demonstrated (0 be greater than unity because of the difficulty in determining the number of active sites on irradiation, ‘When titanium dioxide is used as the semiconductor sen- sitizer, itis sometimes assumed thatthe number of active ites, ‘ean be taken as the product of the surface density of OH ‘soups (typically 10'*-10"* em™*) and the specific surface area of TIO,. However, the later parameter is usually meas: tured using the powder in the dry state (the Brinauer EmmetTeller (BET) surface area), and the actual surface area of the TiO; photocatalyst dispersed in solution will ‘depend on the degree of aggregation. In addition, it does not follow that al the surface sites occupied by the OH groups will be necessarily active, These approximations aze likely to lead to a value for TN which is lower than the actual value thus, iby carrying out sueh calculations it cam be shown that "TN issill greater then unity it appears reasonable to assume ‘thatthe process under examination is a “catalysed”, rather than “assisted”, reaction. For most research conducted inthe area of semiconductor-asssted photoreactions, its assumed ‘that the process understudy isan example of semiconductor- sensitized heterogeneous photocatalyss, especialy if it ean be shown, orit is known, that there is no evidence of marked, loss in semiconductor photoactvity with extended use The overall process of semiconductor-sensitized photo- reactions can be summarized as follows AtD > AT4D* a hee where Ey isthe bandgep of the semiconductor (see below). If,inthe absence of semiconductor and light of energy greater than or equal to E,,, AG? for reaction (1) is negative, the A Mit, Le Hare Jounal of Photochemistry ond Phaaboogy A: Christy 108(1897) 1-35. 3 Exanpes of 0, sense photsyete and photostaytepocesen Deion 36 (met) Reference O10 >, +0, Wace spliting as hon 20,4 41,0~2CH,011+30, (Co nde eduton 01 (61 DN: F6HLO Nit +30, Nitogenedacon bss en 60, 68,0 >C,Hh,0, "60, Gree plat boys 2995 1a Photcatay 30:4 20H 044-+484,0 4200, Metic!mineraiation aot 1s HACC, +CHOH~IL0—2Au+s1CI+C0, Depostion af recoas meals [361 (40+ G04 12, +O, Blomas conven drogen La bo semiconductor-sensitized photoreaction is an example of photocatalyss (5). Alternatively, if AG? for reaction (1) is positive, the semiconductor sensitized photoreaction is an example of photosynthesis [5]. Reported examples [6,10,11,21,35-38] of the two different semiconductor- sensitized photoreactions are given i Table | ‘3. Basie principles ‘As indicated in Fig. 1, for many compounds, asthe number 1 of monomeric units'in a purtile increases, the energy necessary to photoexcite the particle decreases. In the limit when > 2000, iis possible to end up witha particle which exhibits the band electronic structure ofa semiconductor, as illustrated in Fig. 1, in which the highest occupied band (the valence band) and lowest unoccupied energy band (the conductance band) are separated by a bandgap Ey, 8 region devoid of energy levels in a perfect crystal ‘As indicated in Fig.2, activation of the semiconductor photocatalyst for reaction (1) isachieved through the absorp. tion ofa photon of ultra-bandeap energy, which results inthe ‘promotion of an electron e~ from the valence band (0 the FRED pmlc Molonde Cute rep Neto sue Semenduter 32000 N>>200 Fig Change inthe lctonc str of semicon compound as ‘hens of monomeric uns preset increases fos uty cher ‘ot moce ha 2000 (ter Re. [16] ‘conduction band, withthe concomitant generation of a hole 1h in the valence band. For a semiconductor photocatalyst to be efficient, the different interfacial electron processes involving e~ and h*, ie, reactions (c) and (4) in Fig 2 must compete effectively with the major deactivation proc: ‘esses involving e~-h* recombination, ic reactions (a) and (b)in Fg, 2 4, Early work on semiconductor photochemistry [1,2] In 1839, Becquerel (1) reported that a voltage and an electric current were produced when a silver chloride elec ‘wode, immersed in an electrolyte solution and connected to counter electrode, was illuminated with sunlight It was not, ‘until 1955, with the pioneering work of Brattain and Garret (39] on germanium semiconductor electrodes, that the origin (of this photovoltaic phenomenon, called the “"Beequerel cfect”, was understood and the mader era. of photoelectro chemistry was bor, It isnot usually recognized that many ‘examples of semiconductor photocaalysis had already been ‘reported in the literature by the early part of this century Most notably, the semiconductor zinc oxide had attracted a ‘reat deal of attention atthe turn ofthis century as a photo- sensitizer for the decomposition of organic compounds and. 8 a sensitizer for a number of inorganic photoreactions. “Much of thisearly work as been summarizedelsewhere [2] ‘As early 51929 itwas known thatthe pigment “titanium white”. TiO, was responsible for fading in paints [40]. and several major sties into the photoseasitizing ation of ‘TiO, followed [41]. The allure of TiO, asa pigment forthe paint industry stems from its high refractive index (3.87 for ‘utile and2.5-3 foranatase TiO, whereasthe efractive index for diamond is 2.42). Because ofits photosensitizing action, ‘TiO. appeared initially tobe unsuitableasa pigment for pain, since it caused chalking, ie. the photodegradation of the organic polymer binder of the paint [42]. However, this, undesirable feature was largely eliminated by coating the pigment with a hydrous layer ofan inert oxide, such a silica, alumina or ziconia. During the 19606, much ofthe theory of Semiconductor photoclectrochemistry was developed, most, 4 [A Mili, S Le Hue Journal of Photochem and Photobiology A: Chemistry 108 (1097) 1-35 by D Fig. 2 station afte major posses osu on semiconductor parle flowing secon excitation Eleceos-ole recombination canes a he surface (eatin (a) o ia the bale (eatin ()) ofthe semioofooor. tthe sua ofthe particle, photogenerated electns can rede an ecto secre A (ecin (c) sd pozeera hls can oxidize an ele donor D(eacto (6). The cmbaation of actions (e) ma (0 epee ‘emeonductrsentision othe peeral dot reaction (1) pen a tbe l(a aftr Rf {16 notably through the contributions of Gerischer [43] and Pleskoy and coworkers (44-26), S. Photoclectrochemical cells [3-9] {As indicated in Fig. 2, €~-h® pairs can be generated in a semiconductor by the absorption of Tight of energy greater than oF equal to Fy. In an n-semiconductor immersed in solution, an electric field forms spontaneously at the semi ‘conductor-elecrolyte interface;e~-h pairs generated inthe region ofthe electric field ie. the space-charge region, are separated efficiently rather than undergoing recombination ‘As a consequence, in an mtype semiconductor, the photo- generated electron moves into the bulk ofthe semiconductor, ‘where it can be transferred either through a wire toa second, ‘non-photoacive electrode (such as Pt) or through a surface site toa point where an electron acceptor A can be reduced, Aten A> 2 Inthe meantime, the photogenerated ole, under the influ cence of the electri field, migrates towards the surface of the semiconductor toa surface site where itcan oxidize a suitable electron donor D. ie. Dh? >D* 2) ‘The energetics of such a macrophotoelectrosynthetic cell for driving reaction (1) are illustrated in Fig. 3, and Table 2 lists some examples [3,47-S0]. From thermodynamic con- siderations, ifthe photogenerated electron in the conductance ‘band can reduce water, the potential ofthe band (eg) must be Tess than E(H™/H,); it also follows that, ifthe photo- ‘generated hole in the valence band can oxidize water, Evy ‘must be greater than E(O,/H,0). Fig. 4 illustrates the band positions of various semiconductors, including rutile andana- {ase TiO, and SfTiO,, and the redox potentials of relevant redox couples. The first photoelectrochemical cell for water spliting, i. 2H,0 2H, +O, (see Table 1), was reported by Fujishima and Honda [3] in 1972 using a rutile TiO, photoanode and PL counter electrode, This work came ata time when there was increasing intrest in developing arificial systems capa ble of converting sola to chemical or electrical energy. Such work was given greater impetus wit the oll crisis of 1973, and from it arose a substantial research effort into photo- semiconductor lecrae ig. 3. negates of amcrophotnymet cl or ding enc (1) sithereation 300, A. Mil. Le Hane /Loural of Posochemisry and Photobiology A: Chemistry 108 (1997) 135 s ‘rable? Examples of lifere macroptsonyntei eerste cll Semiconductor photelesrde Reason Av D-+A~+D" (eV) Referee TiO, H0+H.0~21i40, 30 Bi) smo, HO+H0-2H, +0, 2 Mose, $0, 28,0-H80. +H to PGake 2N, 468,098.30 2B 131 GAAS 300, 4,0 2C44,01120, a [sor electrochemical cells for water spliting and electricity production, It is not often recognized thatthe Pujishima and Honda ‘Ti0,/H,0/Px photeelectrochemical cell requires a degree of chemical or electrical bias to make it work [51]. This is not surprising given that E9(TiO,; rutile) is approximately 0.2 'V more positive than E(H*/H,) (see Fig. 4). Indeed, it appears likely thatthe original short-circuit photocurrents ‘observed in the Fujishima and Honda TiO,/#,0/Pt photo- electrochemical cell corresponded to the oxidation of water to oxygen atthe TiO; photeanode, but not the concomitant reduction of water o hydrogen atthe P-black cathode, rather the reduction of dissolved oxygen and/or some electrolyte impurity [4]; alteratvely, these workers may have used a ‘chemical bias (alkali in the TiO; photoanode half-cell and acid inthe Pt cathode half-ell) which was not mentioned in, their original paper (3). In contrast, Wrighton et al. [47] were able to show subsequently that the SeTiO,/H,0/Pt photoelectrolyss cll worked with no bis and with « 255 efficiency with respect to UV photons (Ey,=32 eV: (threshold) = 388 nm), but only 1% solar energy conver sion efficiency n, since the proportion of UY light in the solar spectrum is very low (ess than 5%). The reason for the suceess ofthe latter cell, compared with the Fujishima and Honda cell, is that in S‘TIOs, unlike nitile 10, Fey and Eye ese tel aut LT t ttt Fajieig Toy) straddle the hydrogen and oxygen evolution potentials, i. E(H* /H,) and E(O3/1,0) respectively [4] (see Fig. 4) ‘Although other r-type oxide semiconductors have been used as photoanodes forthe photoelectolysis of water, the inappropriate disposition of the conductance and valence bbands and/or a bandgap which sto large makes these mate~ rials ess than ideal for use inefficient solar photoelectolysis cells. In addition, its generally accepted that there is nota single non-oxide, n-type semiconductor which has been shown to be capable of sustaining the oxidation of water to ‘oxygen when illuminated sith ulia-bandgap High; all such semiconductors show a tendency towards photoanodic cor- rosion. Inthe case of p-type semiconductors, which are much rarer in nature, the bandgaps ae usually to0 small and most suffer serious stability problems. As a result, p-type semicon ductors are rarely used in semiconductor photocatalysis. “The energetics associated with typical photoeletrochem- ‘cal cell for electricity production are illustrated in Fig. 5. Tbe photogenerated eh” pair, produced as a result of the absorption of a photon of ultra-bandgap light, is separated by the electric field within the semiconductor. The photogener- sted holes migrate tothe surface ofthe semiconductor where they oxidize redox species A~ to A. For the photoelectr chemical cell to work efficiently, the photogenerated elec: ‘rons must move through an electrical citcuit to the counter IT] ito Fri 4. Valence and conduancelbnd postons for varices senicntistars, ad f elev ex coupes at HO In rer potoreduce a chemical {pects the conden an uth semexnfaior tbe more agate than te reduction pte of he cel species: o powcr a chemical ‘feces te poten ofthe valence tan fe semicondicior mist be mare pve tha the oan pte ofthe chemical pcs, 6 A. Mi, 5 Le Hane / Jounal of Phtochoisry and Philos A Chr 108 (199) 1-35 skerole amcor ace Fig. 5. Energetics a pel, general macrophieltoceria cl fe leery prediction employing an miype acmcondusr the hooters clectrode and there reduce A to A~. One ofthe major prob: Jems encountered in the development of an efficent photo- electrochemical cell for electricity production is to ensure thatthe semiconductor photoanode is stable towards pho- toanodic decomposition. It has already been mentioned that, low bandgap, non-oxide semiconductors, such as CdS, Gas and CASe, which are more suitable than large bandgap oxide ‘semiconductors for solar energy conversion, are prone to this, latter process. In order to ensure the kinetic stability of such ‘hotoanodes, a careful ehoice ofthe A/A~ couple must be ‘made so thatthe rte of oxidation of A by the photogenerated holes is much greater than that forthe semiconductor. Teble 3 lists a selection of photoelectrochemical cells for electricity proguction in which the semiconductors used were in single cexystal [52-55] or polycrystalline [56-S8] form: the later photoclectrochemical cells are more efficient, because of the reduced probability of e~-h* pair recombination at grain boundaries, but more expensive because of the greater dif cally in fabricating single-crystal semiconductors. The efficiency of polyerystalline photoelectrchemical cells can be improved by diffusing or co-adsorbing metal fons, such as Ru*, Os'* and Pb?*. It has been suggested [57] that the presence of such ions makes the surface and stain boundary recombination sates less deep and therefore less likely 10 mediate the efficiency lowering process of ¢--h* recombination. An alternative explanation offered by others [52] is that this approach to chemical modification of the semiconductor primarily improves the interfacial hole transfer rate. More recently, others have produced very ef > £(Os/H,O) (see Fig. 4), ie. there isa sufi: ciently large overpotential forthe reaction to proceed readily without an oxygen redox catalyst, Table 4 provides some ‘examples of the semiconductor. sensitized systems for water photoreduction [59-63] (incorporating a sacrificial electron donor), water photo-oxidation [61,64-67] (incorporating a ‘Some singles and polycrystalline ptotercemicl cll fr cick production" ‘Senicondecor Redox eecrlyte(D"/D) ” Reference Singl 1s (32) pie (3 rmat den PV" /008 mol dm? VS CL ns (53) mS: Til dn" NaBr/001 mol in”? Bs 6 Ga necase (028 mn dn? [K.Fe(CX1, 00128 ml dn? ws G8) [OGFECCN) 01 md” KCN pH 133 Pobertalinephotecectrochemica cls CBSecTEan 1 poled”? NaS! mold KOHL 19156) ane ol dn? Ke. dn K;Se1 mal rv? KOH, Ro, PR isn mcase {iol dn? NaS/1 al > NAOH 16 ts A Mils Le Hane Jura of Photochem an Photobiology A: Chemistry 18 (197) 1-5 7 Rhea Ho ocamse Be cata © — Fig.6 Schemaicilasaton fh commen om yp ystems fr th folowing semiconductor sensitized estone (a) phooeduction of wae ty a sare! electon doy, ich st etymotic ah (EDTA) entaoedy the praeneof aydogen cate chaplain. ‘onthe semeonbitr ssf: () photo oxdtonof watery ascii ‘ton aceptn,such spelt, poly enhanced hy the peace ‘fan open culate nani donde, onthe remiconductr= ‘rice: (6) ptodsociationf water wich appears be vie ules ‘a ou under foe onetins sacrificial electron acceptor) and water photodissocation [68-721 “There is no doubt that itis possible to use semiconductors to sensitize the photoreduction [59-63] end photo-oxidation {61,64-67) of water through the use of sacrificial electron donors and acceptors respectively. The photocleavage of ‘water is another matter. Most research groups have had to ‘resort to using large bandgap metal oxide semiconductors, such a anatase TiO, oF S¢TIO:, and forcing conditions of high temperature and/or reduced pressure fo observe both hydrogen and oxygen generation [11,6870]. The early reports [69] of a “bifunctional catalyst", RuO,/Ti0,/Pt, which photocleaved water efficiently (@(H,) =0.3 was reported; presumably ¢b(0;) =0.15, butte later i usually ‘not observed [73}) under ambient conditions, remain unsub- satiated and, with hindsight, doubtful. Indeed, it has now ‘been suggested [73] that any oxygen observed on radiation ‘of the RuO, /TiO,/P bifunctional eatalyst probably "leaked ‘in from outside". The claims that CaS can be used as @ Sensitizer for the photocleavage of water are highly dubious [71,72] given the well-recognized tendency of CdS 1 Undergo photoanodic corrosion; certainly a numberof major research groups have failed to reproduce these findings [67,73]. All the semiconductors listed in Table 4 are excel- leat sensitizers for the photoreduction of oxygen: it seems ‘most likely that, under ambient conditions, if the photos sociation of water does occur i will not be sustainable, since any oxygen generated will be photoreduced and so short, circuit the overall process eis worth making the following general points about sem iconductorsensitized systems for water photocleavage, Clearly, the observation of hydrogen evolution alone in such systems does not imply water cleavage and the concomitant ‘generation ofa stoichiometric amount of oxygen must not be ‘assumed, Great care must be taken when working on such systems, since hydrogen can be photogenerated using most semiconductorsifa sacrificial electron donor present likely sources ofthis sacrificial agent include electron donors. such as methanol or formaldehyde, used in the deposition of the PGM redox catalyst onto the surface of the semiconductor (repeated washing of such treated photocatalyst does not ensure that all the electron donor is removed) and any poly ‘mer support used ifthe semiconductor isin colloidal form. Even the observation ofthe evolution of oxygen can be the result of sir leaks; the oxygen/nitrogen ratio must not be taken asa guide fo assess how much of the oxygen observed, is photogenerated [67,73]. With all of these problems, it follows that many ofthe reports on semiconductor particle sensitized systems for water cleavage shouldbe viewed with 4 high degree of scepticism. A sustainable, reproducible, semiconductor panicle-based sensitized system for water cleavage is still tobe achieved, 7. Semiconductor particles as photocatalysts for the removal of organic and inorganic pollutants [12-20] ‘The first example in Table 4 is the oxidation of biomass ‘with the concomitant reduction of wate to hydrogen, ie Biomass+H,0 > CO,+H; @ a If experiments with this system had been carried out in aer- ated, rather than nitrogen-purged, solution the reaction under study would have been one of the first, well-documented ‘examples of the semiconductor-sensitzed photomineraliza tion of organic substrates by oxygen. The latter process may ‘be summarized a follows Organic pollutant+0, CO,+H,0+ mineral acids (5) ® |. Mis, Le Hate Journal of Phtochemisry end Photobiology A: Chemis 108 (1997) 1-35 ratios tarp of semiconlcor semi steno the pteedection, oon a pooditaciaten of water ‘Semicondctor ses systems forthe pleduto of water Hydoge alt aH) Reference Semicondocor Serial donor 0, ‘Biomass isuding grass, wood.age.seawced, P(A) 0a2-000% (39) To. Matias proneol P08 0088) (oo) me EDTA Rioim2e) 03 ts) cas EDTA eB : 02) cas * Ru; 28,0 : (631 Semicondoctor sensitised ystems forthe ao xian of ater Oxygen alyt ‘out Referee Semiconaucor Satie eqn To; ett ions ‘None, oF RB, eA, Co . rst) 0, ira None : tsa wo, Fell) ions ‘None oP Rh. Ro, Ru 2H0 003 {651 cas mau) P ei (6s cas incu 0, 0 - 5 Semiconaco-sesized yes forthe potato of water ‘Oxygen cyt Reference Semicontacor —__Hydgencats To, R Nore . (ss) To, PR R000 0 {9 smo, PL POs Ra, Re Co Nowe yi ( To; Nose a 0, cas, R040 ms i cs PLRhocle RU; H.0 wy [Ra(OH)(EDTA)031 i “The energetics and primary processes associated with the semiconductor sensitized photomineralization of organic substrates by oxygen, ie. reaction (5), are ilustrated in Fig. ‘The first clear recognition and implementation of semicon- ductor sensitizers for organic pollutant oxidative minerali- zation came in 1983 with the work of Ollis and coworkers 174,75] on the photomineralizaton of halogenated hydo- carbons, including trichloroethylene, dichloromethane, bloroform and cazbon tetrachloride, sensitized by TiO;-Sub- sequent work by these workers indicated that this approach ‘may be limited to non-aromatic compounds: however, these ‘concems were quickly dispelled through the work of others 76,77] using chlorobenzene and phenol asthe organic sub- strates. A beet list of some ofthe many (over 200 10 date) ‘organic substrates which can be mineralized via reaction (5) is given in Table 5; more exhaustive lst, with references, can be found elsewhere [16.78.79] ‘The semiconductor sensitizer for reaction (5) must be: 1. photoactive; 2. able wo utilize visible and/or near-UV light; 3. biologically and chemically inert; sings poke incre na my c= 40,087 00 Irion SS i. 7. ners and marge proces ote photo-oiéve mineralization of rane planta aca soon istolved oxen sensi by sevicontice panics. The eof xygen ithe one rces a cue! source of dete (24 (ee below: tay be hat may ofthe erent ‘hema pcs whch ca eat fom he ection of oxen such as HO, HO; H,0, ane poss OFT radials ply a majo role inthe oxation of {beers subst it saeay gene eogniae ta the tapped photogenerated le specs, TOM yeryimprtana he e roees A Mil, Ste tne / Sour of Photachonisry and Photobiology As Cher 1081997) 1-85 ° Tables Some examples of TO, sesied potomineraizaton of ogi subsets Cm uarple ‘hanes ‘Metue,uobtae, petane ete eylohenane pain Hidoaanes ‘Mono dtr and weracoanethane, irene, -4Rue-222elrethane ‘Alpe cable Methanol, tan spropy seo, aos, arose Alpha carboxylic cs Formic ethan, dimelethan,popeoi xa ade ‘lke Propeneyeabeiene Hasatenee Perchloreethes, | 2cclorthene, I -sborcethene Aromatics Benzeoe apie Hatoaromats CClarhenane, 2 dichlrobenzee, romabenzene iohuouromatcs S4Dichloreitchenene,dichloitebeaese Press Phenol, hyroginne catechol methyl, resect mel Topas 2 3.4.Cnlovophenl,peachloropeaol, repeal, +rophesat Aromat aon ashe Bera. tamiatenzte sal. allem and phyiroaybenolchoraydoxybezo aids Patines Fayedilee ply inl crise) (PVC) Sorat Sadi dbdecyslphate (SDS). plete lea soiam deel benzene alphonierimethy hohe, ictabtlaneonia oma Herbicides Mei vsloger, azine, sma, rmeton, pera, Betzon Pestes DDT. pean nde 4. photostable (ie. not lable to photoanodic corrosion for example); 5. inexpensive. By common consent, the semiconductor TiO, satisfies cr- teria (1)-(5) ands one of the best semiconductors for sen sitizing eaction (5) In order to improve the reproducibility of the results between groups. many have chosen to use a particular source of TiO,, wih a recognized high photocatal- yc activity, namely Degussa P2S TiO,, which is produced through the high temperature (greater than 1200 °C) flame hydrolysis of TiCl, inthe presence of hydrogen and oxygen, The TiO, is treated with steam to remove HCI which is also produced as part of the reaction. The product is 99.55 pure ‘TiO, (anatase: rutile ratio, 70: 30), which is non-porous, with cubic particles with rounded edges, The P25 TiO: pow er has a surface area of 50: 15 m? g-' and an average particle diameter by number count of 21 nm; 90% of the particles fallin the size range 9-38 um; however, the particles do not exist in isolation, but rather as ireducible complex. primary aggregates, typically approximately 0.1. wm in diameter It is no surprise that different samples of TiO, exhibit different photocatalytic activities towards the same organic substrate under otherwise identical reaction conditions 180,81. Such differences can be qualitatively atibuted 10 differences in. morphology, crystal phase, specific surface area, particle agaregate size and surface density of OH groups in the TiO, samples. However, it may be considered surprs- ing that a TiO; photocatalyst which is good at destroying one ‘organic substrate may be ineffective at destroying another, and that fora different sample of TIO, the situation can be reversed, For example, Degussa P25 TiO, exhibits a photo- activity for reaction (5) which is 50% and 149% of that of Tioxide A-HR TiO, (surface area only 12 m* g~', 100% anatase) when the organic substrates are phenol and hydro quinone respectively [81,82]. Such results highlight the importance of using standard form of Ti, suchas Degussa, 25, in order to make the general findings of one group reproducible by another The results of detailed laser flash photolysis experiments have enabled many of the primary processes associated with reaction (5), sensitized by TiO,, to be identified, together with their characteristic ime domains [16]; the reli ofthis, work are summarized in Table 6. In general, itis usually assumed that the organi pollutant substrate does not undergo direct hole transfer, although there is some evidence that this, process does occur in some cases (atleast fo some extent), but rater that oxidation takes place through a surface-bound Inydroxyl radical (Ti“OH'*). As we shal see later, and as Indicated in Fig. 7, the results of recent work on the semicon- ductor sensitized photomineralzation of thin organic films [24], withelecon acceptors other than oxygen, indicate that the role of oxygen in reaction (5) may be much greater than previously assumed, A full kinetic model for reaction (5) is far from complete and continued research in his area is more useful than further photomineralization studies on yet another, diferent, untried onganie substrate, Intheirinvestigationsand modelling of reaction (5).using "TiO, as the semiconductor sensitizes, Gerischer and Heller 182,83] have generated a substantial body of evidence indi- cating thatthe interfacial eleetron transfer process involving the reduction of oxygen (sce Table 5) isthe rate-limiting step inthe TiO,-sensitized photodegradation of organics. In order to improve the kinetics ofthis process, catalytic sites must be placed on the TiO, surface: these sites can be of the form of islands of Pe, These workers were able to show tha, inthe hotosenstized oxidation of 2,2-dichloropropionate by TIO;, (0.01 wi.% and 2 we.% loadings of the TIO, photocatalyst resulted in threefold and sevenfold improvements in the rte Recently, Lewis and coworkers [$4] have developed a flux- 0 A. Mil, SL Hume Surat of PhotzchomsrandPhaaboogy A Chast 108(1997) 1-25. abies ‘rma proces and socined caracrti tine domains inthe THO, sesion photomisrzaias ofogaic olan 16) Primary press Character tine TO.theoh te fe (von fa) Chai cai ping pees Ton-1> TOI") toss cas Stone > T0H) 100 ps sallow rapper) eboTh ST 10a (depp) ‘Charge carie recombination eH E> TOM") > TOK 100 1s (tow) bb Saions >on ons) [> TOH™ 1+ organo plluan-» >TIYOH + oxidizes pot 100 ns (tom) (80H) +0, > TOKO rs (very slow) ‘matching model of semiconductor photoc their work with a single-crystal rtile TiO; photoanode. In this model, three major carier decay pathways were idemti- fied, namely electron-hole recombination (which i signi cant), reduction of oxygen by photogenerated electrons and hole transfer tothe organic, water or any other acceptor. In this model all are processes are intimately related and one can be stated as being “rate determining". However, they ‘were also able to show thatthe presence ofa Pr catalyst on the surface of the TiO; photocatalyst should lead to an enhancement in the rate, as would the use of more readily reduced acceptors, such as [Fe(CN)¢)” However, the same workers also added, atthe end ofthe paper [84], a cautionary note about extending their findings, fn single-crystal rutile THO, to the anatase particulate T30, ‘hotocatalytic systems which most people work with, They note thatthe added redox agents can act as recombination sites, and 90 the overall photocatalytie rate could increase, decrease or remain unchanged in the presence of such rea ‘gens. Gerischer and Heller [82,83] were able to deposit Pa? ‘onto particulate TiO; to produce more setive photocatalysts In contrast, in an extensive investigation of TiO; photocats Iysts with PGM deposits, we found [85] that the presence of PGMs had very litle, if any, effect on the rate of photocata- Iysis, even at high (10 w1.%) loadings, and, if anything, ‘decreased the rate at very high loadings. This work provides a small insight into the complex, sometimes contradictory, nature of TiO, particulate photocatalytic systems and the resulting difficulty ia developing satisfactory general kinetic models. In general, the kinetics of photomineralization of organic substrates by oxygen, sensitized by TiO, on steady state iMluminaton ita Langmuir-Hinshelwood kinetic scheme, R= — Af] /dr=K(S)K(S)IS]VA+K(S)ES]) (6) where R is the initial rate of substrate removal, [S], isthe inital concentration ofthe organic substrate and, tradition- ally, K(S) is taken to be the Langmuir edsorption constant of species Son the surface of TIO; and &(S) is proportion: ality constant which provides a measure of the intrinsic rea. Livity of the photoactivated surface with S. Generally, the value of K(S) derived from a kinetic study is not dieectly ‘equivalent the dark Langmuir adsorption isotherm for S fon the semiconductor; the latter values are usually much smaller. Its ound that £(S) is proportional tf, where J, is the rate of light absorpdion and @ is a power term which is equal 100.5 or unity at high orlow light intensity respectively Itis also found that k(S) is proportional tothe fraction of O;, adsorbed on TIO, Le. f(0,), where £03) = Kol Og]/(1+ KoslO31) om Ko, is the Langmuir adsorption coefficient for O, om the semiconductor, which appears to be non-competitvely adsorbed at Ti" sites, Several kinetic models of semiconduc- tor photocatalysishave been developed which generate work: ing forms of Eqs. (6) and (7); most notably, Turchi and Olis 186] have proposed four possible different mechanistic schemes involving OH" radical atack of the organic substrate asthe rate Au> Pex Rh> Le> CoNi FeO ing the following met sl: ARNO, OCI, AUC, HPL, RBC! HCl, CU(NO,),, NINO, ): 4 e(NOS),) M1 1/20;4440-+MO,~ 2H" (A SMa” nd "(needs PUTIO,)) (105.106) 0542802420; (uo7} 2810 Be +30; (works bes wih PUTO.) {108.109} S0;+ 68,28, +40 +944,0 Cio} tal regulatory agencies have set a maximum beomate level in potable water of 10 ppb, although there are indications that this level will be reduced to 0.5 ppb very shorty. Typically, ‘potable water which has been treated with ozone will ave @ ‘bromate level in the range 3-68 ppb thus theresa great deal ‘of interest in inexpensive methods forthe removal of bromate jons in ozonated water. Ina recent paper [109], we esta: lished that TiO; was able to sensitize the photodecompostion of bromate, ie. opr +30, 2BrO; (2) atthe parts pe billion level (typically (BrOs ])=50 ppb) ‘using 254 nm light, and that pltiization ofthe TiO; pho- tocatlyst (0.5% PL w/w) enhanced the effectiveness of the photocatalystby a factor of 4.2 Research in the area of semiconductor sensitized photo- degradation of organic and inorganic materials has been cexiensve and, asa result, we have a greater understanding of the fundamental processes involved. Following on from this, work, some commercial products have already been gener ated, .g. small-scale TiO, photocatalytic reactor system for the destruction of organic pollutants in air and water is com secially available from Nutech Energy Systems (London, (Ont.), and SGE (UK Ltd) have launched ANATOC™, an analyser for the rapid determination of total organic carbon (TOC) in water, which utilizes semiconductor photocatalyst technology. Tt may appear surprising at fist that there are not more ‘commercial water purification systems available based on semiconductor photocatalysis given the impressive, exten sive range of organi pollutants which can be destroyed by this method and the fact that itcan be turned on and off with, the fick of a light switch, However, i should also be recog- nized that deactivation ofthe semiconductor photocatalyst is possible, especially by the deposition of solid inert oxides, such as iron(II) oxide, on the photocatalys’s surface and by the binding of stzongly adsorbed inert anions, such as phosphate ions, to the surface. In addition, the most active semiconductor photocatalysts such as iO;, require UV pho- tons and, therefore, are unlikely o find application in water purification systems using sunlight. In addition, the formal ‘quantum yields of most semiconductor sensitized photoca- talytc reactions associated with water purification ae less than 1% and, therefore, are quite inefficient. Funhermore, ‘major developments in this area are required, such as the production of more ative semiconductor photocatalysts, or photocatalysts which can utilize effectively, and without wear, visible light as well as UV light, i order to make this approach to water purification a realistic competitor to the ‘current traditional methods of chlorination, ozonation, air stripping and granulated activated carbon adsorption 8, Semiconductor particles as sensitizers for the pPhotoreduetion of nitrogen and carbon dioxide (and concomitant oxidation of water to oxygen) [6,21] If the photocleavage of water, sensitizedby semiconductor particle, under ambient conditions of temperature and pres: ‘sure, is Considered to be controversial, this ples into insig- rifcance when we examine the semiconductor-sensitized ‘Photoreduction of nitrogen and carbon dioxide, under ambi- tent conditions, which may be considered to be decidedly "unproven, despite some 20 years of research and many claims to the contrary ‘The photocatalytic synthesis of ammonia and oxygen from nitrogen and water, i. 2N,+6H,0 * “4NH, +30, «ay ‘was first claimed in 1977 by Schrauzer and Guth [115] using water vapour and generating gascous ammonia. Over the last 20 years, a plethora of semiconductor-sensitized photosys- tems for reaction (13), usually using liquid phase water and, purporting 10. generate dissolved ammonia, have been described [21]. Various semiconductors have been used, including TiO,, SrTiO, Fe,O,, FeO(OH), WO, and CdS: “ A. Mis, Le Hunt ural of Phtochemisrs an Photobiol A: Chemis 1081997) however, inal the systems studied, invariably the yields of ammonia have been very low, typically nanomoles, and around the limit of detection, Indeed, Davies etal. (21), in their outstanding and challenging review on the subject, state that with all the reports of ammonia symtbesis with Gian ‘based photocatalysts “no paper reports the photosynthesis of ‘ammonia from Water and nitrogen ata level that was shown, to be significantly diferent from levels of ammonia that, ‘might result fom contamination”. Contamination can come from breath (ammonia can be exhaled ata rate of up o 375 gh” ') and many laboratory sources. Forexample, theater ‘workers found that water placed in contact with silicone rubber septa, high vacuum grease, traces of acetone (com ‘monly used to clean glassware) and Teflon tape gave false positive indications of ammonia when subjected to the pop: lular Nessiee’s method for ammonia analysis (21) In their detailed examination of the literature associated with reaction (13), Davies etal. [21] found no report in which the experiments involving reaction (13) had been. carried out significant numberof ies in order to determine both the yield of ammonia and its standard deviation. In their ‘own studies of reaction (13), sensitized by Fe-doped TiO, these workers found no statistically significant experimental evidence for ammonia synthesis, despite previous reports ‘indicating that this system operates under a wide range of ‘experimental conditions [21,116-118]. It should also be pointed ou that the views of Davies and coworkers [21,116— 118], ic. that the achievement of reaction (13) remains, as Yet unproven through the lack of statistically significant resulls, are hotly contested by some of those working inthis field [119,120]. However, the ease made by Davies and coworkers [21,116-118} is strong and the call forthe use of Isotope ratio mass spectrometers in correctly designed, sig nificant experiments, repeatable by more than one research _10up, appears to be a reasonable approach to settle the mat ter, Time will ell whether 20 years of research inthis areas, up to this simple task. As Davies and coworkers [21,116— 118] point out “the concept of a major discovery, such as reaction (13) that remains continually on the borderline of demonstration is internally inconsistent” and that “results ‘which cannot be reproduced are of no scientific intrest ‘The photocatalytic reduction of carbon dioxide by semi- conductor particles, under ambient conditions of temperature and pressure, to reduced carbon products, such as methanol. ‘methane, formaldehyde and formic acid, in water, withthe ‘concomitant oxidation of water oxygen, ic CO, +H,0 * (CH,,CH,OH, HCHO, HCOOH) +0, (14) has boon less wel studied {6} compared with the reduction “of nitrogen, However the few reports of reaction (15) which do exist [6] have the common features of very low yields and the ceal possibilty of contamination, 1t should also be noted that, in most reported cases of reactions (13) and (14), the semiconductor photocatalyst has been pre-treated, e.g. by doping with iron or high tem- perature fring in an inert or reduced atmosphere, and this ‘may create reduction sites (such as Ti" in TiO, or SFTiO,) which may be able ro undergo a photoinduced, stoichiometric reaction with nitrogen or carbon dioxide in which the ammo- nia or reduced carbon products are generated, Such a process {not photocatalysis and yet would initially give the appear ance of i. The generation of Ti" sites, and their subsequent ‘hotoinduced stoichiometric reaction with carbon dioxide, certainly provides an alternative explanation to the results of ‘Hemminger etal. [121] on the reduction of carbon dioxide to methane by water vapour using a clean, reduced SrTiO, ngle crystal. In this system, methane production is low (quantum yield, 0.01%) and ceases after 10 min of irradi tion, buts restored after cleaning by gon ion bombardment and short periods of heating in oxygen. In the same system, methane production, “at rates comparable with the photo- assisted reaction" can also be observed in the dark using the same “ photocatalyst” heated to 450 K. [Lis also worthwhile noting that, although reactions (13) and (14) may no occur, the semiconductor sensitized reverse reactions do. Indeed, the oxidation of ammonia by oxygen, photocatalysed by ‘TiO,, t0 form nitrite and nitrate ions is a well-recognized process {110,118}, and the photomineralization of organies by oxygen, sensitized by TiO, isthe basis of a large amount ‘ofthe work described inthe previous section, was suggested as long agoas 1913 by Benjamin Moore [122] that “the fist step towards the origin of ife must have ‘been the synthesis of organie mater from inorganic by agency ‘of inorganic colloids acting as transformers or catalysts for radiant solar energy". This theme was followed up exten- sively some 28 years later by Dahr et al. [123] working on nitrogen fixation in soils. AS part ofthis work, Dhar etal [123] used naturally occurring minerals, such a hydroxides of iron and oxides of ion, titanium and manganese, as eat Iysts/photocatalysts for nitrogen fixation. The results of this ‘work are inconclusive as they do not eliminate the possibility ‘of contamination and the photodegradation of ritrogen-con- taining compounds. However, the theme of nitrogen fixation by inorganic photocatalysts was picked up by Schrauzer et al. [124] intheirworkon nitrogen fixation using desertsands.. ‘These workers suggested thatnitrogenisreduced photochem- {cally onthe surfaces of minerals containing titanium dioxide, and nitrogen fixed in this manner represents a significant contribution tothe nitrogen cycle, particularly in arid regions ‘of the carth [124]. Although positive results were reported bby Schrauzer eal [124] on some of the desert sands tested, Davies eta. [21] suggest thatthe results are inconclusive ‘and may be interpreted in terms of cither contamination or variations in background levels. Dunn etal, 125] have shown that pltinized TiO,, when placed in contact with a mixture of methane, ammonia and ‘water and iradiated with ultra-bandgap light, can generate small amounts of the amino acids glycine alanine, serine, A. Mil. Le Hate lara of Photochem and Phil aspartic cid and glutamic acid. Experiments with nitrogen- 1Sabelled ammonia showed that the nitrogen inthe amino acid originated from the ammonia rater tan from contamm- inans. Its tempting, therefore, to believe that such abiotic ‘rocesses may have played a role inthe production of amino {cid from the primordial atmosphere many millions of years ago. However, the same workers note [125) that the unpl tinized minerals, Fe,O, and ilmenite, do not photocatalyse the production of amino acids at detectable levels and that the primordial atmosphere [126] may not have been reduc: ing, but more oxidizing, containing CO, and Ns, 9. Semiconductor particles as sensitizers for organic photosynthetic processes (6,12,22,23] ‘The use of semiconductors as sensitizers for various ‘organic photosynthetic processes stems from the early work ‘of Kracutler and Bard [127] onthe use of pltinized TiO. as a sensitizer for the conversion of actie acid to methane and cearbon dioxide: an alternative photoinduced Kolbe reaction. ‘These workers were able to show subsequently thatthe con- ventional Kalbe reaction involving acetic acid 2CH,CO;H+2C,H,+2C0,+H, as) ‘could be photoinduced using aruile THO, photoanode twas ‘quickly recognized hat, if water was sed asthe solvent, reaction (5), the complete oxidative mineralization of the organic substrate, would usally predominate de othe gen eration of bighly oxidizing hydroxyl radicals. Thus most ‘work in the area of semicondoctorsensitzed organic photo Symesis hasbeen caried out either i an neat sovet, such a acetonitrile (MECN), or inthe net oreanie substrate, The semiconductor sensitized organic photosynthetic reactions which have been caried out to date can be ited imo the following broad categories: oxidations and oxidative cleav- pes, reductions, isomerizations and polymerizatons Most work in this area hasbeen carried out on oxidation and oxidative cleavage photoreactons sensitized by TiO and a number of examples are collected in Table 8 (127 135) For mos of the reactions in Table 8, AG" is negative, aud therefore they are technically examples of photocatalytic rather than photosynthetic, reactions (see above). An inter cating example of an oxidative cleavage photorestion sen- Siied by a Semiconductor isthe photo-oxidation of toluene to benzaldehyde and ten to benzoic aid, sensitized by TIO: in MeCN (131] (see Table 8) Fig, 9 illustratesthe observed temporal variations in {toluene}, [benzaldehyde] and {ben 2oie acid} on illumination ofthe system with ulc-bandgap ligh when a small amount sulphuric acid (50 pdm’; S mol dim) is added to the reaction mixare; inthe absence of he acid, the photosensitzed reaction is considerably slower. I isbelieved [131] thatthe HSO; anion promotes the reaction by mediating te oxidation ofthe organic substrate by react ing with the photogenerated hole to form the highly oxidizing radical HSO;, Further work using substituted toluenes fas A Chenin 105 (1997) 1-38 1s shown thatthe addition of small amount of sulphuric acid also enhances the kinetics of oxidation of the organi sub- strat and that the formation of benzo acid is favoured with more strongly electron-withdrawing substituents Examples of reduction, isomerization and polymerization photoreactions sensitized by TiO, are given in Table 9 136. 141]. Work in this area and in the field of oxidation and ‘oxidative cleavage photoreactions has shown thatthe product distribution can be markedly affected by a wide variety of Parameters, including the nature of the semiconductor (in particular, its conductance and valence band positions) ,crys- tal phase, panicle size, surface morphology, light intensity and presence of surface catalysts (such as Pt) [6,12,22.23] 10, Semiconductor particles and flms as sensitizers for ‘the destruction of gascous and thin film organic pollutants [24,25] The use of semiconductors as sensitizers for organic pho tosymthetie reactions in the condensed phase has also been extended to work in the gas phase. For example, Djeghri and ‘Teichner [142] reported the details of the oxidation of 2- methybutane by oxygen, sensitized by TiO. in which the main initial products were carbon dioxide (15%), ethanal (16%), acetone (36%), butanone (12%) and 3-methylbu- tanone (10%). However, in this work, only 2% ofthe initial substrate was converted. Many more studies have investi- ‘gated the intial selectivity of TiO, as a sensitizer for the photo-oxidation of gaseous orvanic substrates by oxygen. In most of the systems studied to date, although a relatively stable intermediate (or intermediates) is (are) often gener- sted, eventually all the initial gaseous organic substrate is completely degraded to carbon dioxide, water and mineral, acid, ie, Gaseous organic substrate +0, ———— imermediate(s) ————» CO,+H,0+ mineral acids (16) ‘Table 10 143-159} lists some.examplesor the use of TiO, ‘asa semiconductor sensitizer for the oxidation of gaseous ‘organie substrates by oxygen, together withthe major inter- ‘mediates and formal quantum eficencies, where the data are sven, For some simple, usually halogenated, organic sub- strates, such as trichloroethylene or perchloroethylene, FOE. values for reaction (16), sensitized by TiO;, approaching, unity have been reported and are taken a an indication ofthe operation of a radical chain reaction mechanism. In contrast, when other, non-halogenated organic substrates are used, such as benzene, toluene, xylene, ethylene propylene, ttra- rmethylethylene, acetone, propanol and ethanol, FQE values, fof less than 0.1 are found [143,145] ‘These FQE values are usually more than tenfold greater than those found forthe phocodestruction of the same organic 6 ‘A. Mis. Le Hue ural of Photochemistry and Photobiology A Cham 108(1997) 1-35. ‘Tbs ction and oxidative pasoreactons semszed by TO, Secor = Smtpeedtine emer Toma {GOs CO, hae hw Kee acm) fee Tolenymreeeante SECON Seceaor ghee) Aeacnciamece {tat 10, 9 eens wim) aloecane OO 10, n. mec, 11m) ee 1.1 diphenyl ethylene benzophenone (esto te sensitive to ing sation, with leston-vidving ‘roupedecresing and hsm ening eeu inate se) (rite) 0, MeCK, H80..05, vs) (ooen obi sbstes teed ince X=, pC, NO, Ll pF Me, pte, NO, m Clam, Rezo: a formation favented be more ectos wihavig tc obstae) To, <> oD H0.0, co oy Ho us oN say now. iHi,CH,08 C4, ,CHO (goo yes tained wih secondary Benzene tsa ‘Sphtic rronaileabols) (0% yield) To, Mec, 0; 13s) siphenylsuide oo—-oro Aipheny sulfoxide substrates dissolved in aqueous solution, e.g the FQE value {for trichloroethylene in aqueous solution is approximately 0.01 [145]. As a consequence, the use of semiconductor DPhotocatalysis as a method of purifying air has begun to attract increasing attention, Such a system could find appli- cation inthe following. 1. the remediation of contaminated soils and groundwater (through the use of air stripping, coupled with a photo catalytic ar purification process): 2. the purification of industrial gaseous effluent; 3. the treatment of indoor or closed system air. “The later is believed tobe a potentially very important market A. Mis. Le Hane Jour of Photochemistry an Phtbiloy A Cham 1081997) 8 ” Fig 9, Observed vations wih ine of te concentration lune (@) (Ge inka saber), enalehye (MI) (an itrmedne) a besa ‘sd (A) (he ma ial pret) afer prolonged main of 100m" ‘factoniecomasiag 10, (0 mg) O, OD saturation lve. rt) snd 4.50, (25 pal) 330°C. Oe eranesymtesiphotrections ested by TO; asthe public becomes more aware ofthe effect ofthe quality of indoor air, at work and at home, on general health. An important class of offensive odour chemicals are oxygenates, such as aldchydes, ketones and alcohols, many of which can bbe degraded via semiconductor photocatalysis, as indicted by the examples in Table 10, However, Ollis and coworkers {160-162}, and others, have highlighted possible weaknesses in the use of semicon: ductor photocatalysi as a means of purifying air. For exam ple, i is recognized that some organic substrates are very dificult to photomineralize completely (e.g. certain aromat- jes and halothanes), and that complete mineralization ofthe ‘organic substrate isnot always established unequivocally. In addition, there is te possibilty that this method of purfica- tion may generate a relatively stable intermediate (such as phosgene from trichloroethylene; see Table 10) which is ‘much more toxic than the initial organic substrate. Finally. and probably most crucially, the destruction of many organic substrates via reation (16) appears to generate byproducts, Semizondaer Reaction Solvent plus dives Referens educonrevctons Toyr on on -Aguecas NOS (361 CO3H a. ~ (63% yield) 2 7 at Mech 138) a as (01% viele) Satin To. (oi) ,CH-+ (CF (57 yl (a seeive Agia Mec ws Botan econ flee spines and poobie) 10, ees oe Men HO, NC (0) oot, 2c 7% yield) Poynton now ‘Paty metarynte—ply (ety! metre) Mecn wan 8 A. Mis. Le Hue / Jarl of Photochemistry and Photobiology A: Chemis 105 (197) 1-38 ‘Tie 10 tarps of TIO, snstize ouiaion potoreavtins of gaseous eganicubstes Semicontuoe Organic rire Major nero Comments FOE _Refeece Heogenaies To, cot. cic Nowe No degraaion cburved ° ns) To: ‘Tochoroeylene (TCE) Ditlracetaldeyde chloride TCBcanbemedtoiceacthe = sal (DEAC) and phosgene ere of dexration of ene from 01-01 LO ough chlorine aia chain acon TO,onslamina TCE DAC ad phosgene Chlrteaedcaton monoxidesre 05-08 [148] ‘so formes asian spared ona ph sarace area “Tio, plets cr Nopboigeneingeneedtythie 04-09 [146] oocyst To. Iobtane ‘Acer snd CO: Typ stu times ce o (eras scene than CO, produced nthe inal sages To, Metase None ike pocatalyssheded with — 1 ‘unpstscte,CO, CO, and #,0 ase profaced To. 1-Butnat ‘Bul (59%) and buns Low (2%) oxygen levels wee sed — soy To, Prana Dectone ‘po and methanol wee ho = si ‘Nie andi shy wee ‘served es major meres Alseoser To, Prpiosishe . ‘Theft of diferent suppers = asa} eta carbon) on the nts tre did can make» ator of ‘appoxinel ou dernce 10, Acealieryde Acetic aid Remoutsemiraspuen highly — 04-038 (183) sete fm pote asd “Tsongas fre Benzene . Lite egadationtbighievels = use (143000 ppm. bt speciale at omer evel (2200 pp Tio; “Tolaene : Depreof onsen 015-025, nie) Heroes TiOvonseclke Pyne : ont ssi ‘which lead tothe deactivation ofthe semiconductor photo- catalyst (usually TiO; ), For example, Peral and Ollis [160] hhave found that the photocatalysed degradation of certain ‘volatile organic substrates containing an Si or N heteroatom (Such as dimethylsiloxane and pyrrole respectively) deact- vates ireversibly a TiO, photocatalyst by the surface depo- sition of species which inhibit and deactivate photoactive sites. There is also evidence that some alcohols when used in reaction (16) willlead to photocatalyst deactivation, presum- ably through the formation of strongly adsorbing, deativat- ing carboxylic acids. Fig. 10 illustrates. the loss of photocatalytic activity observed by Peral and Olis [161] in the photodegradation of I-butanol by oxygen sensitized by ‘TiO,, The photocatalyst could ony be reactivated by contin ued illumination in lean flowing dry air It was suggested that a certain strongly adsorbed intermediate species was responsible forthe loss of photocatalytic activity and that, ‘with prolonged irradiation, the intermediate could be slowly oxidized [161]. In a recent paper, Sauer and Ollis [162] ‘surveyed the current literature on the semiconductor-sensi- ‘ized photodegradation of gaseous organics, and found that ‘hotocatalyst deactivation was almostalwaysfoundin single- pass, fited-bed photoreactors for cumulative contamination ‘conversions in excess of 1-10 equivalent monolayers, whereas no evidence of catalyst deactivation was found for batch reactor studies. The lack of evidence of photocatalyst "in sunlight, which implies that it may’ be possible ro use such fllms to remove, over a period of a day, the glare causing organic film, typically more than 100 nm, which deposits on automobile windscreens. thas also been proposed by Heller etal (24,176) that oil and organic spills on water canbe cleaned up through the use of TiO,-coated hollow glass or glass-ceramic microbeads, floating atthe sir-oil interface, as illustrated in Fig. 13. The ‘Ti0,-coated buoyant photocatalytic microbubblescreatedare typically 50-200 pm in diameter and are rendered oleophilc by reactive coating with an organosilicone, for example. Addition ofthe microbeads (2 2) toa thin layer of erude (2m) inasmall Pets dish causes the oi to rapidly ageregate around the oleoptilic photocatalytic microbeads and, with lengthy (2000 h) exposure to UV light (40 mW em™? of 365 nm light), the black crude oil is degraded, leaving the Fine, white sand-like microbeads floating onthe surface. Cal ‘culations show thatthe photocatalyst microbeads would be able to sensitize the photo-oxidation of a mass of oil equal to their own weight in 10-30 days under solar irradiation. The thickness of an oil layer on water is reduced, when fully, covered with bubbles and when irradiated by AM 1 sunlight, by about 5x 10-em each day. Heller etal, [24,176] believe that this rate of photo-oxidation is adequate to maintain the ‘water free of the oily sheen often seen around harbours and ‘where there is a lot of boat trafic. ‘Although the microbead photocatalyst are nt suitable for the removal of thick (2-10 em) layers of oil, they do help such layers to be burnt down toa thickness at which semi ‘conductor photocatalytic action could he applicable. There is A. Mil. Le Hunte/ Jura of Photochemistry and Photobiology A: Chem 108 (197) 1-38 a Fig. 13. Schama ilusrain of he potocsaticmirobobbles sy Hele tal. (24176 tsensze the psodesrion of hin ez ms fon ter, The ead re ply 50-20 pm in ameter. The pops Inecansm of photoes operting in thet tae ne ‘panded diagram feo he bead: The mechan ober: (1) ee tin f ©, ro stperoide by ptopesed leas the pogrered Satine rin om bene bee oe ‘ih easton wth Ot form an npnperoide pci} the eaton ‘ofthe reanopernde pin withthe uperoise rascal to fran oraaic terocde spice, whic desompoes 0 orm nerd Socompsiion produce which are the father aided 9 C0 evidence [177] to suggest that the proguets of the semicon- ‘ductor-sensitized photodegradation of oil are less harmful than those associated with the destruction of oil by weather- Jing; thisisnot surprising given tha he latter process produces, phenols, polyphenols and, eventually, tar. However, tomany, the photocatalyst itself represents a pollutant and it might be argued, a more permanent and visible pollutant than the sur- factans currently used to disperse oil spills In addition, seems probable, given the broad range of aromatic com- ‘pounds in ol, chat photocatalyst deactivation by UV-absorb ing, polymeric, photogenerated intermediates will be a problem. Thus much more work is needed on the environ: ‘mental impact and long-term efficacy of microbesd photo- ‘catalyst technology if this approach tothe cleaning up of oi and organic spills in environmental waters isto have a real furure ‘Schuitzgebel etal. [178] have also used & hydrophilic form of the photocatalyst microbubbles to investigate the air oxidation of r-octane, 3-octanol, 3-cctanone and n-octanoic acid in aqueous 0.5 mol dm~* NaCI with and without 2 10-moldm”* of FeCl, The major gascous-and organic _phase-soluble products per 100 einstein of 365 nm light, after 12 of imadiaton in ambient ai, for each of the organic substrates, with and without FeCl, ate summarized in Fig. 14, The observed products and intermediates, as well as, the inhibition ofthe degradation process by FeCl, have been. imerpreted by Schwitzgebel et al. [178] as evidence that not only holes, but also electrons and oxygen, participate inthe oxidative degradation process for hydrocarbon, aleohol, ketone and carboxylic acid organic substrates. In addition, the results suggest that the photogenerated holes, or OH radicals, though necessary, ate by themselves inadequate oxidizers. In panicular, these workers propose that the TiO; electron reduces O, to superoxide (pK, 48), ie. O,+e (ore +H") +05 (or 00H) an ‘the hole reacts with an aliphatic compound through a hydro. _gen abstraction process followed by reaction with dissolved ‘oxygen to form an organoperoxide radical, RH +h? (or OH) +R'+H™ (or H,0) ROO" (18) and the organoperoxide radical reats with the superoxide radical to form a tetraoxide, i. ROO'+ 05" (or OOH") +ROOOO- (or ROH) aay which decomposes to products. Inthe absence of oxygen, bot in the presence of FeCl, the yield of CO, and all the other ‘oxidation products of n-octane, -octanol, 3-octanone and n- coctanoic acd is reduced by at least one order of magnitude. ‘This finding is taken as proof that holes, or OH" radicals, are by themselves much less able to oxidize the reactants than a ‘combination of the holes, or OM radicals, and dissolved oxy: ‘gen. Its also suggested 178] thatthe reduced level of CO, ‘generated in ambient air with FeCl, is due to the capture of clectzons by ferric chloride and the oxidation of superoxide to oxygen, leading to a reduction inthe overall steady state ‘concentration of superoxide ‘The above ““Russell-like” mechanism {179} is simple and, if correct, is likely to find general application in many ofthe photocatalytic systems developed fo the photodegra- 31 Teexcange cet dent for his proces fow (10-"!-I0-" Am”) and. at res he Ptogesraed care caries T1, Ce) can pero hough he anosttred i witha deggie os Shrcuph hs combination proces ving» near quaniative charge clacton “TWO: (ez) = Pe”). Movement ofthe potenti lecton athe ck contact (1-10°"s) sa slow process thee is rove sac chug eo: however, kis ecient base the even slower aa of ces (3, alough paves (2) ould ener tinea twas ot forthe aperscavening ation odie nce (6) which sce “) Ile) +h o as 2541 [Nanocrystalline semiconductor electrodes have high sur- face area to volume ratio, and therefore a thin layer will, provide a much lager surface area fora monolayer of dye to adsorb fo than a flat, single-crystal semiconductor electrode, resulting in a larger LHE,. An expression forthe latter term, is as follows LHE, omer aa where Pis the number of moles of dye sensitizer in a mon- layer per square centimetre of fat surface, aA) is the absorption cross-section of the sensitizer (em* mol") and Ris the roughness factor forthe surface. In atypical moder, ‘dye-sensitized nanocrystalline photoelecttochemical cell [32], the sensitizers used have areas of approximately I nm*; thus P= 1.66% 10-! mol em? and (A) at Apu i equal to approximately 1.9% 10" em? mol”. Substituting these values for and oA) in Eq, (20), itfollows that, if the dye is adsorbed onto the surface of a perfectly flat single crystal of the semiconductor (R= 1), the best LHE, value expected ‘will be0.0072, whichis very inefficient. The roughness factor A for a nanocrystalline film of the semiconductor can be crudely estimated by assuming thatthe semiconductor par- ticles of diameter dare tacked on top of each other to create ‘an overall film of thickness b. Under these conditions Roabld en Ina typical nanocrystalline film (see Table 14), b= 10 ym and d= 15 nm; thus R= 2094 and LHE, = Iie. a much more efficient and viable photoelectrochemical cell than that based fn a single-crystal semiconductor. In practice, not all the surface area appears tobe utilized by the dye, and roughness factors of 780-1000 are usually reported: however, even, under these conditions, the value for LHE, will be greater than 0.99. ‘Avery bi proces, withan exchange curt of 1-001 Aer? ‘Avery estrone (10"* 5) ‘The sintering proces, ic. heating the semiconductor film 0 450-80°C, described for both the ansparent and eflect- ing dye-sensitized nanocrystalline photoelectrochemical cells in Table 14, produces electronic contact not only between particles, but also between the particles and the con= {ducting glass support. This electronic contactallows the pho- ‘oinjected electrons o percolate through the films othe back contact, The result isa spongectke structure, which is highly porous (typically $0% porous) and has 2 high surface area to volume ratio. A crude estimate ofthe degree of porosity (P (%)) ina nanocrystalline film can be gained by consid- ering it to comprise spherical particles stacked on top of one another. Under these conditions, P (4) =1~/6= 048, regardless ofthe film thickness, The pores between the par. ticles are interconnected and filed with eletrlyteso thatthe semiconductor-electrolyte interface is accessible throughout the whole film ‘The dye a the heart ofthese new dye-sensitized nanocrys- talline sandwich solar cells is is-X,bis(2.2"-bipytidyl-43'- dicarboxylate)ruthenium(I1) [31] where, for the most efficient solar cells, X=CN- (for the transparent cells) and SCN™ (for the most efficient (7)= 10%) reflecting solar cells); the general structure ofthese complexes is illustrated inFig. 2. Thisclas of dychas, fora longtime, been ignored, largely because of the much shorter lifetimes (typically less than 200 ns) compared with the more famous trsbipyridyt ‘analogues (typically greater than 600 ns). However, in pho- toelectrochemistr, charge injection can be much more rapid than even the shor lifetime ofthe cis-X.bis(2.2'-bipyridyl- 44¥-dicarboxylate)ruthenium(I}) compounds and, 3s & result these complexes can be very efficient sensitizers (see Table 15, process (1)). The four-ligand complex adsorbs ‘onto the surface of TiO, through two of the carboxylate ‘groups, which also provide a strong electonie coupling ‘between the charge transfer excited state ofthe dye and the 2 ‘A. Mis. Le Hut /Jural of Phtochenisry and Phtobilog A Chemis 1081997) 1-35 conduction band of the semiconductor. This coupling helps to ensure that the photoinduced electron injection step, pro- cess (1) in Table 15, is very rapid, with a quantum yield close to unity (30,33) ‘When adsorbed onto nanocrystalline TiO,, the visible absorption edge is shifted by about 200 mV. Thus the cis- 4i(thioeyanato)bis(2.2'-bipyridyl-4,4-dicarboxy- late)ruthenium( I) complex has an absorption edge at 708, ‘am in ethanolic solution, which is shifted to 800 nm when the dye is adsorbed onto TiO;, Fig. 23 illustrates the action spectrum of the nanocrystalline TiO, photoelectrochemical cal using cis-i(thiocyanato)bis(2.2'-bipyridy|-4.4'-dicar- boxylate)ruthenium(Il) as dye sensitizer. Comparing the action spectrum of the cis-di(thiocyanato)bis(2,2'-bipyri- dyl-4,4"-icarboxylate)ruthenium(Il)-sensitized nanocrys- talline TiO, photoelectrochemical cell with the spectral profile of sunlight (AM 1.5) as illustrated in Fig. 23, the ‘degree of overlap i striking. The action spectrum is broader ‘than the actual absorption spectrum of the dye, because of the high roughness factor, coupled with the high molar absorptivity of the dye at many wavelengths in the visible; the result isa light-harvesting efficiency of approximately unity at many wavelengths. Ifthe absorption edge ofthe cis X:bis(2,2-bipyridyl-44-dicarboxylate)ruthenium(M1) dye is shifted by 200 mV, it might be expected thatthe absorption ‘maxima will also be shifted tothe same extent, i.e. rom 534 X;bis(2.2'-bipyridyl-4.4-dicarboxylate)ruthenium (II) ‘complex for example. ‘Although the action spectra of the cis-X,bis(2.2'-bipyst- yl-4,4-dicarboxylate)ruthenium(I]) dye-sensitized nano- crystalline TIO, photoelectrochemical cells have been reported in some detail the actual visible absorption spectra ‘of the dyes inthe cll initially and as a function of time have not been reported, From the literature [30-33] on this cll, ‘we know that the absorption spectrum of the dye is shifted by 200 mV; however, when adjusted for this shift, itis not clear itis the same shape. A change in shape of the absorp- tion spectrum of the dye might give an indication of the formation of dimers, trimers, et, and any change in nature with prolonged use might provide an insight into the major 390 am isused, although light of A>420 nm appears tobe less effec tive in reducing the initial cell output. This intial loss in performance is associated with UV excitation of the TiO, nanocrystals and water contamination. After 2-20 days, no significant changes in the cell performance are observed and. the photocurrent is stable for months. The observed changes inthe visible absorption spectrum ofthe sensitizer with use, if any, may go some way to identifying in more detail the nature of these processes. In general, the cs-i( thioeyanato bis(2,2-bipyridyl-4 dicarboxylate)ruthenium( Il) dye sensitizers used in the new photoelectrochemical cells are considered to be very pho: testable. Although photoinduced ligand substitutions known for these complexes in solvents such as NN-dimethylfor- ‘mamide (DMF), the rates are much slower (1.4 10" times slower) than charge injection (approximately 1.4% 10" 7"). There appears to be no evidence of ligand substitution in the visible absorption spectrum ofthe dye after prolonged use. The oxidized form of the sensitizing dye (S" ) may aso undergo an internal self-destruction redox reaction; however, cyclic voltammetry studies indicat that this process will be very slow, onthe I s timescale, whereas the reduction of S* bby - will occuron the 10~*s timescale, as revealed by laser photolysis expeciments (33). In conclusion, the new nancerysalline photoelectrochem. ical cells appear to have excellent characteristics as solar to clecricl energy conversion devices, with efficiencies of 10% and predicted working lifetimes of 10 years of natural sun- light (30] (the report of "a lifetime consistent with about 108 turnovers foreach dye molecule” for such cells is pre- sumably a mistake and should, possibly, read “10* tumo- vers”) [31] Ifthe high efficiencies for ths type of eel and high dye tumover number can be validated by other research _r0ups, and ifthe production cost ae not to high, tis type of cell will represent a real challenge o existing solid state photovoltaic eels 14, Conclusions ‘The field of heterogeneous photocatalyss is very diverse, although most workin this area falls under one, or more, of the themes in Scheme I. Some ofthe earlier work on seri- ‘conductor photosystems has proved to be highly ireprodu- ibe; this has not helped the subject to develop as rapidly as it might have done and may have generated some degree of scepticism in the scientific community about subsequent developments in the field. However, as the research has moved away from systems which might workin the hands of specific research groups to those which do work inthe hands ofall research groups, 2 number of research themes have emerged which offer real potential for commercial develop ‘ment and merit much greater research. Of particular promise are the areas of: 1. semiconductor particles as photocatalyst forthe removal of organic pollutants [12-20]; 2, semiconductor particles and films 2s sensitizers for the estruction of gaseous and thin film organic pollutants (2425); 3, semiconductor particles as sensitizers for the photo-

    You might also like