Pp

MCa

b

c

a

ARRAA

KCZTPHPPP

1

tscisfeumo

l

0d

Applied Catalysis B: Environmental 107 (2011) 188– 196

Contents lists available at ScienceDirect

Applied Catalysis B: Environmental

journa l h o me pa ge: www.elsev ier .com/ locate /apcatb

hotocatalysis and photoelectrocatalysis using (CdS-ZnS)/TiO2 combinedhotocatalysts

aria Antoniadoua, Vasileia M. Daskalakib, Nikolaos Balisa, Dimitris I. Kondaridesb,∗,hristos Kordulis c, Panagiotis Lianosa,∗

Engineering Science Department, University of Patras, 26500 Patras, GreeceDepartment of Chemical Engineering, University of Patras, 26500 Patras, GreeceChemistry Department, University of Patras, 26500 Patras, Greece

r t i c l e i n f o

rticle history:eceived 15 March 2011eceived in revised form 30 June 2011ccepted 10 July 2011vailable online 20 July 2011

eywords:dSnSiO2

hotocatalysisydrogenhotoelectrocatalysishotoelectrochemical cellshotofuel cells

a b s t r a c t

Powdered composite CdS-ZnS photocatalysts of variable composition have been synthesized by a co-precipitation method and were used as photocatalysts to produce hydrogen and as photoelectrocatalyststo produce electricity. Results of catalyst characterization show that composite sulphide photocatalystsform solid solutions and that their band gap energy can be tuned between that of ZnS (3.5 eV) and thatof CdS (2.3 eV) by varying Cd (or Zn) content. The composite materials can photocatalytically producesubstantial quantities of molecular hydrogen in the presence of sulphide–sulfite ions as sacrificial electrondonors. Photocatalytic performance is significantly improved when small amounts of Pt crystallites aredeposited on the photocatalyst surface. The rate of hydrogen production over the Pt-free CdS-ZnS powdersdepends on Cd (or Zn) content and is generally much higher for the composite materials than for pure CdSor ZnS. Pure semiconductors were found to be very poor photocatalysts under the present experimentalconditions. Furthermore, two specific photocatalyst compositions, i.e., 67% and 25% CdS, gave maximumhydrogen production rates. An analogous behavior was observed when the same powders were usedto make photoanode electrodes since both the rate of hydrogen ion reduction and the current flow areproportional to the number of photogenerated electrons. Composite CdS-ZnS photocatalysts were also

applied by successive ionic layer absorption and reaction on TiO2 films deposited on FTO electrodes. Theobtained materials were used as photoanodes in a two-compartment photoelectrocatalysis cell filled witha basic electrolyte and with ethanol as sacrificial electron donor (fuel). The (CdS-ZnS)/TiO2 photoanodesdemonstrated a qualitatively similar behavior as CdS-ZnS photocatalysts. Thus 75%CdS–25%ZnS over TiO2

was a better electrocatalyst than 100%CdS over TiO2. When CdS-ZnS photocatalysts were combined withioned

titania, they mainly funct

. Introduction

In view of the increasing awareness on environmental issues,he increasing interest in the employment of renewable energyources and the need of recycling of waste materials, metal sulfidesombined with large band-gap semiconductors have been stud-ed as efficient photocatalysts for hydrogen production by waterplitting or by consumption of waste materials [1–11]. It has beenound that the combined photocatalyst by itself is not efficientnough unless it is associated with a co-catalyst like nanopartic-

late Pt or other noble metals. The promoting effect of dispersedetal crystallites is usually attributed to their ability to act as traps

f photogenerated electrons [12–17]. An essential requirement is

∗ Corresponding authors.E-mail addresses: dimi@chemeng.upatras.gr (D.I. Kondarides),

ianos@upatras.gr (P. Lianos).

926-3373/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.apcatb.2011.07.013

as visible-light-photosensitizers of this large band-gap semiconductor.© 2011 Elsevier B.V. All rights reserved.

that the work function of the metal is higher than that of the semi-conductor. When this condition is fulfilled, a Schottky barrier isdeveloped at the metal–semiconductor interface, which leads toan efficient charge separation and, therefore, to a decrease of therate of electron–hole recombination and other back reactions thatlimit system efficiency [12–17]. Dispersed metal crystallites mayalso function as classical thermal catalysts by enhancing the rateof “dark” catalytic reactions and/or selectivity to reaction products.For example, deposition of metal (e.g., Pt) and/or metal oxide (e.g.,RuO2) particles on TiO2 has been reported to improve kinetics of thewater-splitting reaction by decreasing the overpotentials of hydro-gen and oxygen evolution, respectively [14,15]. Even though, thebinary Pt/CdS catalyst has been shown to be very successful in sev-eral occasions [18–20], the ternary Pt-CdS-TiO2 catalyst does not

seem to be efficient at random combination [1,4,5]. In particular, ithas been shown that CdS and Pt should in some cases be spatiallyseparated. Thus in Ref. [1], Pt nanocrystallites were first depositedon titania nanoparticles and CdS nanoparticles were last deposited

sis B:

iilbidt

tteesS[ioCmefStetisp[

fiagtTfgcscpefipsscosfaialsts

2

2

A

M. Antoniadou et al. / Applied Cataly

n order to occupy separate sites on TiO2. It has been also found that,n order to make the ternary catalyst work, CdS/TiO2 and Pt had toie on a conductive electrode and be spatially separated [4]. Thisehavior can be explained by considering that spatial separation

s necessary in order to achieve vectorial electron transfer in theirection CdS → TiO2 → Pt (cf. SI in [5]) and avoid multi-vectorialransfer, which acts against efficiency [1].

The favorable electron transfer from photoexcited CdS to Pthrough TiO2 is not only demonstrated by the nanoparticulateernary catalyst but is also realized in photoelectrocatalyst (photo-lectrochemical) cells. In that case, TiO2 is deposited on the anodelectrode followed by deposition of CdS on the TiO2 films. Depo-ition of CdS can be easily achieved by employing the so-calledILAR (Successive Ionic Layer Absorption and Reaction) method21]. For this, the electrode bearing the titania film is successivelymmersed in a solution of a Cd2+-containing salt, then in a solutionf a S2− salt and so on. Both ions are absorbed and react giving smalldS nanoparticles (quantum dots) accommodated within titaniaesoporous structure. When the so formed CdS nanoparticles are

xcited by visible light, the photogenerated electrons are trans-erred onto the lower-lying conduction band of titania [1,2,11,22].ubsequently, photogenerated electrons move from the anode tohe cathode electrode, which may carry Pt or other noble metallectrocatalyst, through an external circuit. Electrons arriving athe cathode may then participate in reductive reactions, includ-ng molecular hydrogen production from hydrogen ions. Indeed, ashown in previous publications, hydrogen and/or electricity can behotoelectrocatalytically produced by using CdS/TiO2 photoanodes2,6–8].

In the present work, we deal with two interrelated matters. Therst one concerns a previously studied effect [23–26] of combining

large band gap semiconductor (ZnS) with a relatively small bandap semiconductor (CdS), a combination, which allows band gapuning within a substantial range of the visible part of the spectrum.his combination is reviewed in terms of either photocatalysisor hydrogen production or photoelectrocatalysis for electricityeneration. The second matter concerns the employment of theomposite semiconductor as photosensitizer of titania. Resultshow that combination of these two metal sulfides and furtherombination with titania offers an advantage with respect to sim-le CdS or CdS/TiO2 systems. The challenge is not only to developfficient solar photocatalytic systems that can produce hydrogenrom water, but also systems characterized by long term stabil-ty under irradiation. This is particularly important for CdS-basedhotocatalysts, which are known to be susceptible to photocorro-ion and, therefore, can operate efficiently only in the presence ofacrificial electron donors. Ideally, the latter should be low- or zero-ost waste materials. In this respect, results presented here werebtained with the use of either sulphide/sulfite ions or ethanol inolution. The former reagent is an undesirable waste product inossil fuel technology (e.g., oil refineries) whereas the second isn example of a biomass-derived compound that may be presentn wastes of biomass processing industries. Even though, certainspects of the behavior of the presently used combined photocata-ysts have been previously treated, the new techniques of materialsynthesis and deposition and the new dimensions and range ofheir application justifies the renewed [1–9] interest in theseystems.

. Experimental

.1. Materials

Unless otherwise indicated, reagents were obtained fromldrich and were used as received. The nanocrystalline titania

Environmental 107 (2011) 188– 196 189

was commercial Degussa P25 (specific surface area 50 m2 g−1).Millipore water was used in all experiments. SnO2:F transpar-ent conductive electrodes (FTO, Resistance 8 �/�) were purchasedfrom Pilkington, USA, Carbon Cloth, 20% wet proofing and Pt/CarbonBlack electrocatalyst (30% on Vulcan XC72) from BASF Fuel Cell,Inc., USA and Carbon Black, Vulcan XC72R, was a gift from CABOTCorporation.

2.2. Construction of CdS-ZnS/TiO2 photoanode electrodes

FTO electrodes were cleaned by first washing in soap, rinsedwith water and then sonicated in isopropanol, acetone and ethanol.Titania films were prepared by depositing on clean FTO electrodesa home-made paste based on pure titania Degussa P25, preparedaccording to the recipe given in Ref. [27]. Two layers were screenprinted (90 mesh) giving a uniform film, which, after calcinationin air at 550 ◦C, was about 8–10 �m thick [28]. The geometricalarea of each film was 3 × 4 = 12 cm2. Electric contact was made byusing an adhesive copper ribbon and a copper wire soldered onit. A picture of such an electrode can be seen in Ref. [29]. CdS-ZnS composite catalysts at various proportions were deposited onthe titania electrode by the SILAR method [21]. For this purpose,we used two aqueous solutions, one containing Cd(NO3)2·4H2Oor Zn(NO3)2·6H2O or mixtures of both, and the second containingNa2S·9H2O. When metal salt mixtures were used, the total concen-tration of metal ions and the corresponding concentration of sulfurions were 0.1 M. Other concentrations of salts have also been tried.The freshly prepared titania electrode was immersed for 5 min inthe metal salt solution, then copiously washed with triple-distilledwater, then immersed for 5 min in the Na2S·9H2O solution andfinally washed again. This sequence corresponds to one SILAR cycle.10 SILAR cycles were performed in all studied cases. Finally, theelectrode with deposited (CdS-ZnS)/TiO2 layer was dried, first ina N2 stream and then it was put for a few minutes in an oven at100 ◦C.

2.3. Synthesis of powdered photocatalysts

Composite CdS-ZnS photocatalysts of variable CdS content (0,20, 25, 33, 40, 60, 67, 75, 100 wt.%) were prepared by a co-precipitation method [30,31]. For the synthesis of 5 g of CdS, anamount of 10.7 g of Cd(NO3)2·4H2O was dissolved in 200 mL ofwater and heated at 70 ◦C under continuous stirring. A secondsolution was prepared by dissolving 8.31 g Na2S·9H2O in 200 mLof water and then mixed dropwise with the first. The resultingmixture was maintained under stirring at 70 ◦C for 30 min. Afterthis, the precipitant was filtered, washed three times with triplydistilled water and dried at 70 ◦C for 20 h. The resulting pow-der was crushed, sieved (dp < 0.63 �m) and stored in the darkfor subsequent use. Composite CdS-ZnS photocatalysts were pre-pared in a similar manner, with the use of appropriate amountsof Cd(NO3)2·4H2O, Zn(NO3)2·6H2O and Na2S in order to obtain thedesired composition in the final material.

A mixed (25%CdS–75%ZnS)/TiO2 photocatalyst with a (CdS-ZnS):TiO2 ratio of 1.0 (wt.% basis) was synthesized following thesame above method, by addition of the corresponding amount ofTiO2 powder (Degussa P25) in the salt-containing solution prior tothe addition of Na2S.

Platinum-promoted photocatalysts were prepared by impreg-nation of the synthesized CdS-ZnS and (CdS-ZnS)/TiO2 powders

with an aqueous solution of (NH3)2Pt(NO2)2 (Alfa). The impreg-nated support was dried at 110 ◦C for 24 h, ground, sieved andfinally reduced at 150 ◦C in H2 flow for 2 h. The nominal metalloading of all materials thus prepared was 0.5 wt.% Pt.

1 sis B: Environmental 107 (2011) 188– 196

2

wiwotoZut4

2

CBouwpni831mltuptatI3

2p

c[Xtaqcmait[iaotwtcums

90 M. Antoniadou et al. / Applied Cataly

.4. Construction of metal sulphide photoanode electrodes

Selected CdS-ZnS composite powders, obtained in Section 2.3,ere used to make pastes by slightly modifying the recipe given

n Refs. [32,33]. For this purpose, 1.0 g of metal sulphide powderas mixed with 0.1 g of CdCl2 and then further mixed with 1.0 g

f ethylene glycol. This mixture was first magnetically stirred andhen subjected to ultrasonic vibration until a uniform paste wasbtained. In the case of pure ZnS powder, an amount of 0.1 g ofnCl2 was used instead of CdCl2. This paste was screen printed,sing a 60 mesh screen, on clean FTO electrodes as above. Finally,he films were calcined in an Ar atmosphere for a few minutes at00 ◦C.

.5. Construction of the cathode electrode

The cathode electrode was made of carbon cloth with depositedarbon Black and Pt as described in previous publications [34,35].riefly, an amount of 0.246 g of carbon black was mixed with 8 mlf distilled water by vigorous mixing in a mixer (about 2400 rpm)ntil it became a viscous paste. This paste was further mixedith 0.088 ml polytetrafluorethylene (Aldrich, Teflon 60 wt.% dis-ersion in water) and then applied on a carbon cloth cut in theecessary dimensions. This has been achieved by first spread-

ng the paste with a spatula, preheating for a few minutes at0 ◦C and finally heating also for a few minutes in an oven at40 ◦C. Subsequently, the catalytic layer was prepared as follows:

g of Pt/carbon black electrocatalyst (30% on Vulcan XC72) wasixed with 8 g of nafion perfluorinated resin (5 wt.% solution in

ower aliphatic alcohols and water, Aldrich) and 15 g of a solu-ion made of 7.5 g H2O and 7.5 g isopropanol. The mixture wasltrasonically homogenized and then applied on the previouslyrepared carbon cloth bearing carbon black. The electrode washen heated at 80 ◦C for 30 min and the procedure was repeateds many times as necessary to load about 0.5 mg of Pt/cm2. Thehus prepared Pt/carbon cloth (Pt/CC) electrode was ready for use.ts dimensions were similar to those of the anode electrode, i.e.

× 4 = 12 cm2.

.6. Description of the installation used for monitoringhotocatalytic hydrogen production

In order to evaluate the performance of powdered photo-atalysts we used an installation described in detail elsewhere36,37]. Briefly, it consists of a solar light-simulating source (OsramBO 450 W), a quartz photoreactor and an on-line analysis sys-

em comprising a gas chromatograph (Varian 3800) and a CO2nalyzer (Binos) connected to the exit of the photoreactor. Theuartz photoreactor was of cubical shape and its top cover hadonnections for inlet/outlet of the sweep gas (argon). The chro-atograph was interfaced to a personal computer which enables

utomatic sampling of the reactor effluent at pre-selected timentervals via an electrically actuated gas sampling valve. The pho-on flow entering the reactor, measured by chemical actinometry36], was found to be 5.61 × 10−7 Einsteins s−1. In a typical exper-ment, 80 mg of photocatalyst powder was added into 60 mL ofn aqueous solution containing an inorganic (S2−/SO2−

3 ) and/orrganic (ethanol) sacrificial agent. The reactor was then sealed, theemperature was adjusted at 40 ◦C, and the system was purgedith argon to remove atmospheric oxygen from the reactor and

ubing. The reactor was then exposed to light (at t = 0) underontinuous stirring. Experiments were conducted under a contin-ous Ar flow through the reactor (20 cm3 min−1), which served aseans of collection and transfer of gaseous products to the analysis

ystem.

Fig. 1. Configuration of the photoelectrocatalysis cell used in the present work.

2.7. Description of the reactor used for photoelectrocatalysisexperiments

We used some different reactors applicable in each particularcase. Most data were obtained with the reactor having the basicconfiguration of Fig. 1. It was made of Plexiglas and comprised twocompartments of orthogonal shape separated by a silica frit (ROBU,Germany, porosity SGQ 5, diameter 25 mm, thickness 2 mm). Bothcompartments contained the same aerated electrolyte. The capac-ity of the anode compartment was 10 ml and that of the cathodecompartment 2 ml. The fuel (ethanol), when necessary, was addedonly in the anode compartment. UVA excitation of the photoan-ode was made by employing Black Light fluorescent tubes, peakingaround 363 nm. A home-made array of parallel tubes, each of 4 Wnominal power, was placed in front of the cell window, provid-ing UVA light of about 4.0 mW/cm2 at the position of the titaniafilm. Excitation was also made by a Xe lamp simulating solar radi-ation, as in the previous section (Section 2.6), using an Osram XBO450 W source. Light was filtered for the IR radiation using a cir-culating water filter to take away heat. The intensity of radiationat the position of the photocatalyst was in that case 55 mW cm−2.In all cases, the exciting radiation passed through the glass-FTOelectrode, which played the role of window. Photocurrent measure-ments as a function of the wavelength of the incident radiation weremade by filtering light with a monochromator (see Section 2.8).

2.8. Methods

Electrochemical measurements were carried out with an Auto-lab potentiostat PGSTAT128 N. All current–voltage curves weretraced at 20 mV/s. Photocurrent measurements as a function ofthe wavelength were made by using a homemade installationemploying the above mentioned Xe lamp using an Osram XBO450 W source, a Jobin-Yvon monochromator and current and light-measuring instruments. The current measured at each wavelengthwas corrected for the intensity of the incident radiation at eachwavelength. Radiation intensity for � < 400 nm was measured witha PMA 2100 Radiant Power meter (Solar Light Co), calibrated for theNear UV spectral range and for � ≥ 400 nm with an Oriel Radiant

Power meter (70260) calibrated for the visible range.

Powder photocatalysts were characterized with respect to theirspecific surface area (BET) employing nitrogen physisorption at thetemperature of liquid nitrogen. Phase composition and primary

M. Antoniadou et al. / Applied Catalysis B: Environmental 107 (2011) 188– 196 191

Table 1Specific surface area (SSA), mean primary crystallite size (d) and bandgap energy(Ebg) of the synthesized powdered CdS-ZnS photocatalysts.

CdS content (nominal) SSA (m2 g−1) d (nm) Ebg (eV)

(wt.%) (% mole)

0 0.0 107 4.8 3.520 14.4 74 3.8 3.025 18.4 68 3.7 2.933 24.9 76 3.2 2.840 31.0 78 3.4 2.760 50.3 81 3.3 2.6

cPb

apfepsf

3

3

iTpa(

lpttaa

FCa

600500400300

E

h

g

f

e

d

c

b

a

F(R

)

Wavelength (nm)

CdS(wt.%)

(a) 0(b) 20(c) 25(d) 33(e) 40(f) 60(g) 67(h) 100

Fig. 3. UV/vis absorption spectra obtained over CdS-ZnS photocatalyst powders of

67 57.8 67 3.8 2.5100 100 119 3.6 2.3

rystallite size were determined by X-ray diffraction (XRD) on ahilips P (PW 1830/40) powder diffractometer in the 2� scan rangeetween 20 and 80◦, using Cu Ka radiation.

Diffuse reflectance spectra (DRS) were recorded with the use of Varian Cary 3 spectrometer and transformed to a magnitude pro-ortional to the extinction coefficient through the Kubelka–Munkunction, F(R). Results obtained were used to estimate the bandgapnergy (Ebg) of the synthesized materials using the methodroposed by Tandon and Gupta [38], according to which Ebg corre-ponds to the point where the linear increase of the Kubelka–Munkunction starts (Fig. 3, trace a, vide infra).

. Results and discussion

.1. Photocatalyst characterization

Powdered composite CdS-ZnS catalysts have been character-zed by the BET method, and the obtained results are shown inable 1. It is observed that the specific surface area of compositehotocatalysts varies in the range of 67–81 m2 g−1, which is gener-lly lower than that of single-component CdS (119 m2 g−1) or ZnS107 m2 g−1).

X-ray diffraction patterns of the powdered CdS-ZnS photocata-ysts are presented in Fig. 2. It is observed that CdS exhibits broadeaks located at 2� of ca 26.6, 43.9 and 51.8◦ (trace h), which are

ypical of cubic CdS [1]. A similar pattern was obtained for ZnS, withhe corresponding reflections located at 28.6, 47.7 and 56.5◦ (trace). Composite materials exhibit peaks at intermediate diffractionngles, which shift progressively from those of ZnS to those of CdS

80706050403020

hgf

e

d

c

b

a

100

67

60

40

33

25

20

(311)(220 )(111 )

CdS(wt.%)

Inte

nsity

(a.u

.)

0

Diffraction angle (2 theta)

ig. 2. X-ray diffraction patterns obtained over CdS-ZnS photocatalysts of variabledS content. Solid and dotted lines denote reflections from single-component ZnSnd CdS powders, respectively.

variable CdS content. The method used to estimate bandgap energy is shown for ZnS(trace a).

with increase of CdS content from 0 to 100% (traces a–h). The factthat no doublets were observed for the XRD peaks of compositesamples indicates the absence of separate phases of CdS and ZnSand implies formation of homogeneous solid solutions. The dataof Fig. 2 were used to estimate the mean crystallite size of CdS-ZnS photocatalysts with the use of Scherrer’s equation, and theresults are listed in Table 1. It is observed that all samples containnanocrystallites with a mean primary size between 3.2 and 4.8 nm.

DRS spectra obtained over CdS-ZnS powdered photocatalysts ofvariable CdS content are shown in Fig. 3. It is observed that pure ZnS(trace a) absorbs light below ca 380 nm whereas pure CdS absorbslight up to ca 600 nm (trace h). Composite materials exhibit inter-mediate absorption characteristics (traces b–g) thereby providingadditional evidence for the formation of homogeneous solid solu-tions. The band gap energies (Ebg) estimated from the spectra ofFig. 3 are listed in Table 1. It is observed that Ebg decreases pro-gressively from 3.5 eV for ZnS to 2.3 eV for CdS with increase ofCdS content from 0 to 100%. When the powdered photocatalystswere used to make thin films on FTO electrodes (cf. Section 2.4),they produced similar absorption spectra as those of Fig. 3. There-fore, it may be assumed that the principle composite photocatalystcharacteristics are preserved when they are applied as thin films.

CdS-ZnS composite photocatalysts formed on TiO2 films by theSILAR method demonstrated similar spectroscopic characteristicswith those of powdered photocatalysts, as seen in Fig. 4. Onlythe visible part of the spectra is shown in that case, because theabsorbance by the titania film at wavelengths below 400 nm wasso intense that the spectra were recorded with large uncertainty.Pure ZnS did not absorb in the Visible. The longest wavelengthabsorption onset, observed with 100% CdS (Fig. 4, trace 5), wasmuch shorter than the corresponding value in the case of powderedCdS (Fig. 3, curve h). This is apparently due to quantum size effectsand it indicates that the metal sulphide nanoparticles supportedon titania are substantially smaller than in the case of powderedphotocatalysts. Indeed, Transmission Electron Microscopy mea-surements presented in a previous publication [39] indicated thatCdS nanoparticles were no larger than 2 nm, i.e. almost half thesize of CdS nanocrystallites obtained in the case of powdered pho-tocatalysts (cf. Table 1). The absorption onset suffered a blue shift

in the presence of ZnS, indicating a tuning of Ebg within an impor-tant wavelength range in the presence of increasing Zn content. Inorder to verify that this blue shift of the absorption onset is due tothe presence of Zn and not to the lower concentration of Cd ions in

192 M. Antoniadou et al. / Applied Catalysis B: Environmental 107 (2011) 188– 196

520500480460440420400

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1

54

32

F(R

)

wavelength (nm)

Fig. 4. Absorption spectra of CdS-ZnS composite semiconductor nanoparticlesd(C

tsufstt

ptsaoca

3p

3

hE

FncCr

0 20 40 60 80 1000.0

0.2

0.4

0.6

0.8

1.0(B)

Rat

e (μ

mol

e H

2 min

-1)

CdS content (wt.%)

1200100080060040020000.00

0.25

0.50

0.75(A) CdS(wt.%)

60

40

67

33

25

20

0100

Rat

e (μ

mol

e H

2 min

-1)

Irradiation time (min)

Fig. 6. (A) Rate of hydrogen evolution as a function of irradiation time obtained2− 2−

eposited on nanocrystalline TiO2 by the SILAR method at various combinations:1) 100% Zn; (2) 75%Zn–25%Cd; (3) 50%Zn–50%Cd; (4) 25%Zn–75%Cd; and (5) 100%d. Spectra were recorded against nanocrystalline TiO2 film as reference.

he precursor solutions, we recorded the data of Fig. 5 showing thepectroscopic characteristics of pure CdS formed on titania filmssing precursor solutions of lower concentration. The marked dif-erence between the absorption onsets of curves 1 and 3 clearlyhows that the shift observed in the presence of zinc is much largerhan the quantum size effect, the latter most probably explaininghe difference between curves 2 and 3.

In conclusion, results presented in this section show that com-osite CdS-ZnS nanocrystals do not contain separate phases of thewo components but they consist of composite nanoparticulateemiconductors of the type CdxZnyS (where 0 ≤ x ≤ 1, x + y = 1). Vari-tion of the x, y values results in tuning of the energy band gapf the composite material. This conclusion is valid for all studiedases, i.e. powdered photocatalysts, powders applied as thin filmsnd photocatalysts formed on titania films by the SILAR method.

.2. Photo-induced H2 production over powdered CdS-ZnShotocatalysts

.2.1. Effect of the CdS contentPowdered CdS-ZnS photocatalysts have been tested for

ydrogen production, in the absence of any other co-catalyst.xperiments were first conducted in the presence of S2−/SO2−

3 ions

540510480450420-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4 3

2

1

F(R

)

wavelength (nm)

ig. 5. Comparison between absorption spectra of CdS and composite CdS-ZnSanoparticles deposited on nanocrystalline TiO2 by the SILAR method at various pre-ursor solution combinations: (1) 0.025 M Cd2+–0.075 M Zn2+–0.1 M S2−; (2) 0.025 Md2+–no Zn2+–0.025 M S2−; and (3) 0.1 M Cd2+–no Zn2+-0.1 M S2−; Spectra wereecorded against nanocrystalline TiO2 film as reference.

over the indicated CdS-ZnS photocatalysts in the presence of S /SO3 ions (0.24 MNa2S/0.35 M Na2SO3) in solution. (B) Effect of CdS content on the steady-state rateof hydrogen production (data obtained from Fig. 6A).

(0.24 M Na2S/0.35 M Na2SO3) in solution. Results obtained are sum-marized in Fig. 6A, where the rate of hydrogen evolution in the gasphase (rH2 ) is plotted as a function of irradiation time. In all cases,rH2 initially increased with time and then reached a pseudo-steady-state. The plateau values are plotted in Fig. 6B as a function of theCdS content. It is observed that rH2 over bare CdS was very low. Thishas been previously attributed to the presence of low-lying sur-face states (acceptor levels) that are ineffective toward reductionof protons due to their less negative potential [24]. In the pres-ence of ZnS, the rate of hydrogen evolution is markedly enhancedin a manner, which depends on the composition of the mixed sul-phide photocatalyst. In particular, rH2 goes through two maxima formaterials containing 25 and 67 wt.% CdS. Similar results have beenreported by several other authors [23,24,26,40–42]. The beneficialeffect of the presence of ZnS has been attributed to the removalof inactive surface states [24] or to the blockage of these states byZn2+ ions which are in intimate contact with CdS [42]. It has beenalso proposed that the enhancement of photocatalytic efficiencyobserved for CdS-ZnS composites is due to a shift of the conductionband toward more negative potentials with increase of ZnS con-tent [23], which is supported by results of photoelectrochemical

measurements [40]. Indeed, the conduction band level (ECB) of ZnSlies at more negative potentials than that of CdS and this increasesthe difference (�E) between ECB and E(H2/H+), which is a (ther-modynamic) measure of the ability of photogenerated electrons

M. Antoniadou et al. / Applied Catalysis B: Environmental 107 (2011) 188– 196 193

100080060040020000.0

0.2

0.4

0.6

0.8

d

c

(a) CdS-ZnS(b) Pt/(CdS-ZnS)(c) Pt/(CdS-ZnS)/TiO2(d) Pt/TiO2

a

b

Rat

e (μ

mol

e H

2 min

-1)

Irradiation time (min)

FpZ

tpcooarw

a(NtsogtspifhdTc(tbac

3

(icitacpcp

1.00.50.0-0.5-1.0-1.5

0

5

10

15

20

25

30

35

804000

5

10

15

I sc (m

A)

% CdS

3

2

1

I (m

A)

V (Volts)

Fig. 8. I–V curves recorded with the cell of Fig. 1 under illumination with a Xe lamp.The photoanode was made of: (1) nanocrystalline titania alone; (2) CdS/TiO2; and

ig. 7. Rates of hydrogen evolution as a function of irradiation time obtained in theresence of S2−/SO2−

3 ions (4.8 mM Na2S/7.0 mM Na2SO3) in solution over (a) CdS-nS (25–75); (b) Pt/(CdS-ZnS); (c) Pt/(CdS-ZnS)/TiO2; and (d) Pt/TiO2 photocatalysts.

o reduce protons to H2. Apparently, the more than one differentarameters that simultaneously affect hydrogen production effi-iency may explain the appearance of two maxima in Fig. 6B insteadf a smooth variation. In the case of pure ZnS, the maximum ratef hydrogen production was again very low. This can be explainedlso by the existence of defects and by the fact that the excitingadiation has a small component in the UVA part of the spectrumhere ZnS absorbs light.

In Fig. 7, trace a, the rate of hydrogen production is plotteds a function of irradiation time for the best-performing CdS-ZnS25–75 wt.%) catalyst, in the presence of S2−/SO2−

3 ions (4.8 mMa2S/7.0 mM Na2SO3) in solution. In this experiment, the concen-

ration of S2−/SO2−3 ions was deliberately chosen to be 50 times

maller than that used in Fig. 6 in order to allow consumptionf these sacrificial agents in less than 24 h. It is observed that rH2oes through a maximum of ca 0.47 �mol min−1 at ca 350 min andhen abruptly decreases. This decrease of rH2 is due to the con-umption of the S2−/SO2−

3 ions, which, when present in solution,rotected the photocatalyst against photocorrosion by rapidly and

rreversibly removing photogenerated holes. It is well known that,or both thermodynamic and kinetic reasons [43], photogeneratedoles do not interact with sulphide ions belonging to the semicon-uctor lattice but preferentially oxidize sulphide ions in solution.his explains the excellent stability observed for CdS-ZnS photo-atalysts in the presence of high concentrations of S2−/SO2−

3 ionse.g., more than 20 h in Fig. 6A) and the loss of activity under condi-ions where these sacrificial agents were consumed (Fig. 7). It maye noted that the stability against photocorrosion of CdS-ZnS cat-lysts in the presence of high concentrations of S2− ions has beenonfirmed by atomic absorption spectroscopic analysis [26].

.2.2. Effects of addition of Pt and/or TiO2Dispersion of Pt (0.5 wt.%) on the CdS-ZnS photocatalyst surface

Fig. 7, trace b) results in a significant increase of the rate max-mum (0.78 �mol min−1 at ca 300 min) and a faster drop of rH2 ,ompared to those obtained over the Pt-free sample (trace a). Thiss expected, because the amount of sacrificial agents added in solu-ion was the same in both cases (4.8 mM Na2S/7.0 mM Na2SO3)nd, therefore, higher reaction rates should result in their faster

onsumption. Clearly, addition of Pt has a beneficial effect on thehotocatalytic performance of the mixed sulphide catalyst, whichan be attributed to suppression of electron–hole recombination byumping away photogenerated electrons and by acceleration of the

(3) (75%CdS-25%ZnS)/TiO2. Insert: variation of the short-circuit current Isc with thepercentage of CdS. The electrolyte was 0.5 M Na OH with 10% v. ethanol added inthe anode compartment.

(dark) catalytic reduction of protons toward H2. An intermediateperformance was observed over the platinized Pt/(CdS-ZnS)/TiO2catalyst (trace c). It should be noted, however, that this catalystcontains 50 wt.% TiO2 and therefore only half of the active CdS-ZnS component, compared with pure Pt/(CdS-ZnS) photocatalyst.However, a main reason of the lower efficiency in that case may alsobe the incompatibility of the three components appearing simul-taneously in the ternary photocatalyst, as it has been reported inprevious publications [1,4]. Furthermore, a rate curve was alsoobtained over Pt/TiO2 photocatalyst and is also shown in Fig. 7(trace d). It is observed that rH2 reaches a plateau value of ca0.04 �mol min−1, which is dramatically lower than that obtainedfor (CdS-ZnS)-containing samples (traces a–c). Interestingly, thisplateau value of rH2 is similar to that observed over the Pt/TiO2catalyst under conditions of photocatalytic splitting of pure water,i.e. in the absence of sacrificial agents in solution [36]. This clearlyindicates that photocatalytic activity of Pt/TiO2 is not practicallyinfluenced by the presence of S2−/SO2−

3 ions in solution, in con-trast to what was observed when organic compounds were usedas sacrificial agents [44]. Indeed, as it will be discussed here below,oxide photocatalysts alone or functionalized with metal sulfidesfunction better in the presence of organic sacrificial agents.

3.3. Photoelectrocatalysis using composite (CdS-ZnS)/TiO2photoanodes

Cells of the type depicted in Fig. 1 and described in Section2.7 have been employed by using photoanodes made of titaniananocrystalline films and SILAR-deposited CdS-ZnS, prepared asdescribed in Section 2.2. The cathode electrode was a carbon clothbearing Carbon Black and Pt electrocatalyst (Pt/CC), as describedin Section 2.5. The electrolyte was 0.5 M NaOH in both anode andcathode compartments. 10% v. ethanol was also added in the anodecompartment. Ethanol served as the fuel to be consumed in orderto run the cell (cf. Ref. [35]), i.e. it was the sacrificial electron donor.The relative efficiencies of the cell were compared by recordingcurrent–voltage (I–V) curves. Fig. 8, shows I–V curves recorded withphotoanodes made of various proportions of CdS and ZnS appliedon titania, under excitation by simulated solar radiation (Xe lamp).

The value of the short-circuit current Isc, i.e. the value of the currentat zero voltage, extensively varied from one sample to the other ascan be seen in Fig. 8 and also in Table 2. Isc was very modest inthe case of pure titania (trace 1 and last row of Table 2) but sig-

194 M. Antoniadou et al. / Applied Catalysis B: Environmental 107 (2011) 188– 196

Table 2Values of short-circuit current and open-circuit voltage obtained with the cell ofFig. 1, employing photoanodes made of (CdS-ZnS)/TiO2 as a function of the CdScontent.

CdS content (%) Xe lamp Black Light

Isc (mA) Voc (Volts) Isc (mA) Voc (Volts)

0 1.8 1.20 3.6 1.2025 9.2 1.25 6.9 1.2035 11.9 1.23 7.3 1.2050 8.3 1.23 6.1 1.2065 11.5 1.19 7.4 1.1775 14.8 1.21 8.7 1.16100 11.7 1.20 7.8 1.16

nImMdhtiagaimtasmviatp6f

tvteTUtwsrscvtbdawpsio

dc

5505255004754504254000.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

2

1

abso

rban

ce/c

urre

nt (a

.u.)

wavelength (nm)

of UV light incident on the photocatalyst and generated by theabove Xe lamp is no more than 2 mW cm−2 while the correspond-ing intensity of Black Light was about the double (see Section 2.7).As seen in the insert of Fig. 10, Isc increased by increasing CdS con-

1.00.50.0-0.5-1.0-1.5

0

5

10

15

20

25

30

10075502503

4

5

6

7

8

9

I sc (m

A)

% CdS 3

2

1

I (m

A)

V (Volts)

Fig. 10. I–V curves recorded with the cell of Fig. 1 under illumination with BlackLight. The photoanode was made of: (1) nanocrystalline titania alone; (2) CdS/TiO2;

Titania withoutmetal sulfides

1.3 1.20 3.0 1.25

ificantly increased in the presence of metal sulfides. The value ofsc increased by increasing the CdS content and demonstrated two

axima at 35 and 75% CdS, as can be seen in the insert of Fig. 8.aximum Isc was obtained in the case of 75% CdS. The depen-

ence of Isc on CdS content is qualitatively similar to that of theydrogen production rate of Fig. 6, even though, the positions ofhe maxima were obtained at slightly different CdS contents. Thiss an interesting finding that grants support to the validity of thebove data. It may be noted that both evolution of molecular hydro-en and current flowing through the external circuit depend on themount of photogenerated electrons. Therefore, it is not surpris-ng that both photocatalytic and photoelectrocatalytic processes

ay demonstrate similar behavior. The above data also suggesthat composite CdS-ZnS precipitated in an aqueous suspension ordsorbed on nanocrystalline titania demonstrate a qualitativelyimilar photo(electro)catalytic performance. The presence of twoaxima in the photocatalytic activity of CdS-ZnS photocatalysts of

ariable CdS content has been also reported by other authors. Fornstance, De et al. [45] and Roy and De [26] observed two maximat ca 20%CdS and 60%CdS, in qualitative agreement with results ofhe present study. Interestingly, other investigators reported theresence of only one of these maxima, at either 15%CdS [23] or0–80%CdS [25,31,41]. These differences most possibly originaterom the different synthesis conditions employed.

The higher current obtained in the case of CdS/TiO2 compared tohat of ZnS/TiO2 may be explained by considering that CdS absorbsisible light while ZnS is excited only by UVA radiation. The por-ion of UVA in simulated solar light is very small, therefore, thextent of ZnS excitation is limited under the present conditions.he same is true for pure titania, which is also excited only byVA radiation. Therefore, titania alone gave the lowest Isc. The fact

hat the combined (CdS-ZnS)/TiO2 system indeed responded to andas photoactivated by Visible light is demonstrated in Fig. 9. It is

een that the system does respond to Visible light and the responseoughly follows the absorption spectrum. The data in Fig. 9 corre-pond to a photoanode made of (75% CdS–25% ZnS)/TiO2 but similaronclusions were obtained with other CdS proportions. The obser-ation that variation of Isc with increase of CdS content from 0%o 100% was not monotonous but exhibited two maxima, coulde explained in a similar manner as in Section 3.2.1. It seems thatefect sites on pure CdS or pure ZnS are repaired in the presence of

limited quantity of the minority component but defects are not soell repaired when the two components are present at equal pro-ortions. This may explain the appearance of maxima at the twoides of the curve in Fig. 6B or in the insert of Fig. 8 and a minimumn between. Further investigation of this issue is beyond the scopef the present study and may be the subject of our future work.

The open-circuit voltage Voc, i.e. the voltage at zero current,id not substantially change in the presence of metal sulfides, asan be seen in Fig. 8 and in Table 2. Since the cathode potential is

Fig. 9. Comparison of the photocurrent response (1) and the absorption spectrum(2) in a cell where the photoanode was made of (75%CdS-25% ZnS)/TiO2.

not affected by whatever variations occur at the anode electrode itmay be concluded that the presence of metal sulfides did not affectthe anode potential either. Apparently, the anode potential is onlydetermined by the position of the valence band of nanocrystallinetitania. Metal sulfides then play the unique role of photosensitizerof titania. As a matter of fact, the value of Voc = 1.2 V is typical ofnanocrystalline titania photoanode in the presence of a basic elec-trolyte and alcohol as sacrificial electron donor [35].

IV traces were also obtained by UVA (Black Light) irradiationof the same as above system and the results are shown in Fig. 10and Table 2. The short-circuit current was lower under UVA exci-tation but it was still much higher in the presence of metal sulfidesthan for titania alone. As expected, since titania is the majoritycomponent and since it absorbs only UVA radiation, the currentobtained with pure titania was higher under Black Light radia-tion than with simulated solar light radiation. Indeed, the portion

and (3) (75%CdS–25%ZnS)/TiO2. Insert: variation of the short-circuit current Isc withthe percentage of CdS. The electrolyte was 0.5 M Na OH with 10% v. EtOH added inthe anode compartment (same installation as in Fig. 8 but excitation only by UVAradiation).

M. Antoniadou et al. / Applied Catalysis B: Environmental 107 (2011) 188– 196 195

TiO

2

3.2

eV

MS

2.3

-3.5

eV

e-

hνh+

TiO

2

3.2

hν

e-

h+

MS

2.3

-3.5

eV

TiO

2

3.2

eV

MS

2.3

-3.5

eV

e-

hνh+

TiO

2

3.2

eV

MS

2.3

-3.5

eV

e-

hνh+

TiO

2

3.2

hν

e-

h+

MS

2.3

-3.5

eV

TiO

2

3.2

hν

e-

h+

MS

2.3

-3.5

eV

Fig. 11. Illustration of the photosensitization of titania by absorption of light by theme

tFpsottteactom

poa

tueAbd

lovedpmhftbs

dopsIi

1.00.50.0-0.5-1.0-1.5-10

-5

0

5

10

15

20

25

30

4

3

2

1

I (m

A)

V (Volts)

Fig. 12. I–V curves recorded with photoanodes based on pure CdS-ZnS composites

the two cases, i.e. presence or not of titania and supports the con-

etal sulphide (MS) composite semiconductor (left) and the transfer of photogen-rated hole after excitation of titania (right).

ent and demonstrated two maxima at the same positions as inig. 8. The higher currents obtained in the presence of metal sul-hide combined photocatalysts, albeit smaller than in the case ofimulated solar light excitation, may be justified by suppressionf electron–hole recombination. Since the majority component isitania, it is naturally assumed that, in the present case, TiO2 ishe mainly excited semiconductor. The photogenerated hole in theitania valence band is expected to be transferred onto the morelectronegative valence band(s) of the composite metal sulphide,s depicted in Fig. 11. This separation of photogenerated chargearriers is expected to increase the number of free electrons andhus the flow of current in the external circuit. Combination thenf titania with metal sulphide semiconductors increases current noatter what the excitation wavelength is.An interesting feature of Fig. 10 is the appearance of oxidation

eaks at V = −49 V (curve 3) and V = −0.32 V (curve 2). This suggestsxidation of metal sulfides, a characteristic that is in line with thebove mentioned hole transfer process.

Similar to what was observed for experiments conducted withhe use of simulated solar light, the open-circuit voltage measurednder UVA excitation was not appreciably influenced by the pres-nce of metal sulfides or by CdS content of the photoanode (Table 2).s discussed above, this result indicates that Voc is governed mainlyy titania and that deposited metal sulfides do not play a role inefining electrode potential.

It may be noted here that (CdS-ZnS)/TiO2 combined photocata-yst functioned very well in a basic electrolyte and catalyzed ethanolxidation. Ethanol then acted as the sacrificial agent (fuel), by bothisible and UV excitation. Oxidation of ethanol may take placeither directly or through interaction with OH• radical interme-iates [46]. When titania is excited by UV excitation, part of thehotogenerated holes are transferred to the valence band of theetal sulphide combined semiconductor and part interacts with

ydroxyl ions. When metal sulphide is excited by visible light orunctionalized by hole transfer from excited titania, it is reasonableo assume that ethanol is directly oxidized by metal-sulphide-orne holes, since the valence band potential of the latter is nottrong enough to interact with hydroxyl ions [46].

In conclusion, results of this section show that metal sulfideseposited on TiO2 films form a composite material similar to thatbtained in the case of precipitated powdered materials. This com-osite acts as a visible light-sensitizer of titania and also assists

eparation of electron–hole pairs formed by UV excitation of TiO2.n both cases, a beneficial effect is obtained as to the flow of currentn the photoelectrochemical cell. Variation of CdS-ZnS composi-

using simulated solar light excitation: (1) 100% CdS; (2) 67% CdS–33% ZnS; (3) 25%CdS–75% ZnS; and (4) 100% ZnS. The electrolyte was 0.12 M Na2S and 0.175 Na2SO3

and the cell had one single compartment.

tion defines regions where maxima of current are observed. Thisis analogous to the appearance of maxima in the rate of hydrogenproduction, since hydrogen ion reduction and current flow are bothproportional to the number of photogenerated electrons.

3.4. Photoelectrocatalysis using composite CdS-ZnS photoanodes

The cell of Fig. 1 was also used with photoanodes made of com-posite CdS-ZnS powders without titania, prepared as described inSection 2.4. It was then found that no current or negligible currentwas produced in the presence of 0.5 M NaOH. By adding ethanolin the anode compartment a substantial current was obtained,ranging between 2.9 and 3.1 mA, in the case of pure CdS. The corre-sponding open-circuit voltage was 0.8 V. However, the current wasmuch higher when a mixture of 0.12 M Na2S and 0.175 Na2SO3 (cf.Section 3.2.1) was used as electrolyte instead of NaOH, withoutany organic additive, as can be seen in Fig. 12. This result pro-vides additional evidence that metal sulphide photocatalysts workbetter in a sulphide–sulfite electrolyte while oxide photocatalystswork better in a basic electrolyte. Since no distinction of electrolytebetween anode and cathode compartment was made in this lastcase, the following data were obtained by removing the separatormembrane, thus the cell was operated as a one-compartment cell.Fig. 12 shows that Isc was much larger in the case of CdS than for ZnS.This is expected, since excitation was made by simulated solar lightwhere the UV portion is very low, while ZnS, as already said, absorbsonly in the UV region. Composite CdS-ZnS photoanodes gave highercurrents with increase of CdS content, as expected. Interestingly,a maximum was again observed by combining 67% CdS with 33%ZnS (Fig. 12, trace 2). This maximum may be again explained in thesame manner as above, i.e. by considering the repairing capacity ofZn on CdS defect sites. A second local maximum was also observedby combining 25% CdS with 75% ZnS (trace 3). Once more then ithas been shown that both hydrogen ion reduction and current flow,which are proportional to the number of photogenerated electrons,are enhanced by combining CdS with ZnS.

The open-circuit voltages measured in the case of metal sulphidephotoanodes, i.e. in the absence of titania, were much lower thanin its presence, i.e. they were 0.7–0.9 V, similar to that obtainedwith 0.5 M NaOH + 10% v. Ethanol (0.8 V). This finding distinguishes

clusion of the previous section that when CdS-ZnS is deposited ontitania, the latter determines the photoanode potential. The aboveconclusions were also verified by studying photocurrents and pho-

1 sis B:

tr

tmcd

4

tchtomomitaIeacf

A

Rt

R

[

[

[[[[[

[[[[

[[[[[

[[

[

[

[[[[[

[[[[[

[

[

[

[

96 M. Antoniadou et al. / Applied Cataly

opotentials in a three electrode system, i.e. working, counter andeference (Ag/AgCl) electrodes (results not shown).

Summarizing, results of this section provide additional evidencehat combination of CdS with ZnS at various proportions provides

ore efficient photocatalysts than pure materials. When CdS-ZnSomposites are deposited on titania, the photoanode potential isetermined by the majority component, i.e. titania itself.

. Conclusions

Composite CdS-ZnS semiconductors can be employed as func-ional photocatalysts and photoelectrocatalysts either alone or inombination with nanocrystalline titania. In all studied cases, itas been found that composite semiconductors are better catalystshan pure CdS or pure ZnS. Optimal results, both for photocatalyticr photoelectrocatalytic purposes, were obtained for compositeaterials consisting of 67–75% CdS and 33–25% ZnS. When this

ptimized photocatalyst is deposited on nanocrystalline titania,aking the minority species, it acts as a photosensitizer of titania

n the visible and also assists electron–hole dissociation by UV exci-ation. The potential of photoanodes made of (CdS-ZnS)/TiO2 is notffected by the minority component but is defined by titania itself.n general terms, when CdS-ZnS composites alone are employed forither photocatalytic or photoelectrocatalytic purposes it is prefer-ble to use a sulphide–sulfite electrolyte. When the composites areombined with titania, they better fit basic electrolytes and organicuels, like ethanol, as sacrificial agents.

cknowledgement

Financial aid from a project financed by E.ON Internationalesearch Initiative is gratefully acknowledged. Responsibility forhe content of this publication lies with the authors.

eferences

[1] H. Park, W. Choi, M.R. Hoffmann, J. Mater. Chem. 18 (2008) 2379.[2] Y.-L. Lee, C.-F. Chi, S.-Y. Liau, Chem. Mater. 22 (2010) 922.[3] G. Wang, X. Yang, F. Qian, J.Z. Zhang, Y. Li, Nano Lett. 10 (2010) 1088.

[4] N. Strataki, M. Antoniadou, V. Dracopoulos, P. Lianos, Catal. Today 151 (2010)

53.[5] V.M. Daskalaki, M. Antoniadou, G. Li Puma, D.I. Kondarides, P. Lianos, Environ.

Sci. Technol. 44 (2010) 7200.[6] J. Bai, J. Li, Y. Liu, B. Zhou, W. Cai, Appl. Catal. B 95 (2010) 408.

[

[[

Environmental 107 (2011) 188– 196

[7] H. Wang, Y. Bai, H. Zhang, Z. Zhang, J. Li, L. Guo, J. Phys. Chem. C 114 (2010)16451.

[8] M. Antoniadou, P. Lianos, J. Photochem. Photobiol. A 204 (2009) 69.[9] C.-F. Chi, S.-Y. Liau, Y.-L. Lee, Nanotechnology 21 (2010) 025202.10] S.V. Tambwekar, D. Venugopal, M. Subrahmanyam, Int. J. Hydrogen Energy 24

(1999) 957.11] Y. Liu, H. Zhou, B. Zhou, J. Li, H. Chen, J. Wang, J. Bai, W. Shangguan, W. Cai, Int.

J. Hydrogen Energy 36 (2011) 167.12] O. Carp, C.L. Huisman, A. Reller, Prog, Solid State Chem. 32 (2004) 33.13] A.L. Linsebigler, G. Lu, J.T. Yates Jr., Chem. Rev. 95 (1995) 735.14] A.J. Bard, J. Phys. Chem. 86 (1982) 172.15] D. Duonghong, E. Borgarello, M. Grätzel, J. Am. Chem. Soc. 103 (1981) 4685.16] M. Ni, M.K.H. Leung, D.Y.C. Leung, K. Sumathy, Renew. Sust. Energy Rev. 11

(2007) 401.17] M. Anpo, M. Takeuchi, J. Catal. 216 (2003) 505.18] H. Yan, J. Yang, G. Ma, G. Wu, X. Zong, Z. Lei, J. Shi, C. Li, J. Catal. 266 (2009) 165.19] M. Sathish, R.P. Viswanath, Catal. Today 129 (2007) 421.20] M. Berr, A. Vaneski, A.S. Susha, J. Rodriguez-Fernandez, M. Doblinger, F. Jackel,

A.L. Rogach, J. Feldmann, Appl. Phys. Lett. 97 (2010) 093108.21] H.M. Pathan, C.D. Lokhande, Bull. Mater. Sci. 27 (2004) 85.22] D. Liu, P.V. Kamat, J. Phys. Chem. 97 (1993) 10769.23] J.-F. Reber, M. Rusek, J. Phys. Chem. 90 (1986) 824.24] H.-C. Youn, S. Baral, J.H. Fendler, J. Phys. Chem. 92 (1988) 6320.25] M.M. Taqui Khan, R.C. Bhardwaj, C. Bhardwaj, Int. J. Hydrogen Energy 13 (1988)

7.26] A.M. Roy, G.C. De, J. Photochem. Photobiol. A 157 (2003) 87.27] S. Ito, P. Chen, P. Comte, M.K. Nazeeruddin, P. Liska, P. Pechy, M. Gratzel, Prog.

Photovolt.: Res. Appl. 15 (2007) 603.28] M. Antoniadou, E. Stathatos, N. Boukos, A. Stefopoulos, J. Kallitsis, F.C. Krebs, P.

Lianos, Nanotechnology 20 (2009) 495201.29] M. Antoniadou, P. Bouras, N. Strataki, P. Lianos, Int. J. Hydrogen Energy 33 (2008)

5045.30] W. Schindler, H. Kisch, J. Photochem. Photobiol. A 103 (1997) 257.31] A. Koca, M. Sahin, Int. J. Hydrogen Energy 27 (2002) 363.32] S.-L. Fu, T.-S. Wu, M.-P. Houng, Sol. Energy Mater. 12 (1985) 309.33] V. Kumar, V. Singh, S.K. Sharma, T.P. Sharma, Opt. Mater. 11 (1998) 29.34] M. Antoniadou, D.I. Kondarides, D. Labou, S. Neophytides, P. Lianos, Sol. Energy

Mater. Sol. Cells 94 (2010) 592.35] M. Antoniadou, P. Lianos, Appl. Catal. B 99 (2010) 307.36] A. Patsoura, D.I. Kondarides, X.E. Verykios, Appl. Catal. B 64 (2006) 171.37] V.M. Daskalaki, D.I. Kondarides, Catal. Today 144 (2009) 75.38] S.P. Tandon, J.P. Gupta, Phys. Status Solidi 38 (1970) 363.39] N. Balis, V. Dracopoulos, E. Stathatos, N. Boukos, P. Lianos, J. Phys. Chem. C 115

(2011) 10911.40] M.F. Finlayson, B.L. Wheeler, N. Kakuta, K.-H. Park, A.J. Bard, A. Campion, M.A.

Fox, S.E. Webber, J.M. White, J. Phys. Chem. 89 (1985) 5676.41] N. Kakuta, K.H. Park, M.F. Finlayson, A. Ueno, A.J. Bard, A. Campion, M.A. Fox,

S.E. Webber, J.M. White, J. Phys. Chem. 89 (1985) 732.42] N. Kakuta, K.-H. Park, M.F. Finlayson, A.J. Bard, A. Campion, M.A. Fox, S.E. Web-

ber, J.M. White, J. Phys. Chem. 89 (1985) 5028.43] J. Sabaté, S. Cervera-March, R. Simarro, J. Gimenez, Int. J. Hydrogen Energy 15

(1990) 115.44] D.I. Kondarides, V.M. Daskalaki, A. Patsoura, X.E. Verykios, Catal. Lett. 122

(2008) 26.45] G.C. De, A.M. Roy, S.S. Bhattacharya, Int. J. Hydrogen Energy 21 (1996) 19.46] P. Lianos, J. Hazard. Mater. 185 (2011) 575.