i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 6 ( 2 0 1 1 ) 6 5 1 9e6 5 2 8

Avai lab le a t www.sc iencedi rec t .com

journa l homepage : www.e lsev ier . com/ loca te /he

Efficient H2 production by water-splitting usingindiumetin-oxide/V-doped TiO2 multilayer thinfilm photocatalyst

R. Dholam*, N. Patel, A. Miotello

Dipartimento di Fisica, Universita degli Studi di Trento, I-38123 Povo (Trento), Italy

a r t i c l e i n f o

Article history:

Received 29 November 2010

Received in revised form

28 February 2011

Accepted 7 March 2011

Available online 6 April 2011

Keywords:

Photocatalytic water-splitting

Hydrogen generation

V-doped TiO2

RF-magnetron sputtering

Visible light irradiation

Multilayer

Charge recombination

* Corresponding author. Tel.: þ39 3478267397E-mail address: rupal@science.unitn.it (R

0360-3199/$ e see front matter Copyright ªdoi:10.1016/j.ijhydene.2011.03.028

a b s t r a c t

In order to sensitize TiO2 in visible light and to reduce photo-induced charge recombination,

the multilayer films of IndiumeTin Oxide (ITO)/V-doped TiO2 were synthesized by radio-

frequency magnetron sputtering. V-doped TiO2 thin films showed red shift in TiO2 absorp-

tion edge with increasing dopant concentration and, most importantly, the dopant energy

levels are formed in the TiO2 band gap due to V5þ/V4þ ions as confirmed by UVeVisible and

XPS spectra. Multilayer films with different numbers of ITO/V-doped TiO2 (6 at.%) bilayers

(namely, 2-, 3-, 4-, 5-, 6- and 7-bilayers) were deposited, in order to reduce the charge

recombination rate, by keeping the total thickness of TiO2 constant in each multilayer film.

In multilayer films, when exposed to visible light the photocurrent increases as function of

the number of bilayers by reaching the maximum with 6-bilayers of ITO/V-doped TiO2. The

measured enhanced photocurrent is attributed to: 1) ability of V-doped TiO2 to absorb visible

light, 2) number of space-charge layers in form of ITO/TiO2 interfaces inmultilayer films, and

3) generation of photoelectrons just in/or near to the space-charge layer by decreasing the V-

doped TiO2 layer thickness. The reduced charge recombination rate in multilayer films was

also confirmed by the photocurrent kinetic curves. The superior photocatalytic efficiency of

the 6-bilayers film is also reflected in hydrogen production rate through water-splitting: we

obtained indeed 31.2 mmol/h of H2 production rate.

Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights

reserved.

1. Introduction splitting such as: i) high resistant to corrosion and photo-

It is extremely important to use clean and renewable energy

sources such as solar and wind for hydrogen production

which is presently produced (about 95%) from fossil fuels by

procedures that lead to increase in greenhouse gases [1].

Photocatalytic water-splitting by using solar energy could

contribute to the solution of environmental and energy issues

related to the hydrogen production [2].

TiO2 semiconductor provides several advantages over other

materialsasaphotocatalyst forhydrogenproductionbywater-

.. Dholam).2011, Hydrogen Energy P

corrosion in aqueous media, ii) cheap and easily available, iii)

environmentally clean, iv) energy band edges which are well-

matched with the redox level of water, and v) electronic

properties that can be varied by just changing the lattice

defects chemistry or the oxygen stoichiometry [3]. In addition,

TiO2 is also the most desired material for self-cleaning tech-

nology and in purification of toxic compounds from polluted

water and air [4e6]. However, because of thewide energy band

gap (w3.2 eV) of TiO2, only a small fraction of solar spectrum,

UV light (3e5% of total), can be used for photocatalytic

ublications, LLC. Published by Elsevier Ltd. All rights reserved.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 6 ( 2 0 1 1 ) 6 5 1 9e6 5 2 86520

processes [7]. Thus, the efficient utilization of the solar spec-

trum (mainly a good portion of visible light) is one of the

important subjects for developing the future generation of

TiO2-based photocatalysts. Doping TiO2 with foreign ions is

one of the most promising strategies for sensitizing TiO2 to

visible light by forming impurity levels in the forbidden gap [8].

DopingTiO2with 3d-transitionmetals (V, Cr,Mn, Fe, Co, and

Ni) is the best approach to decrease the band gap or to establish

energy levels inside the band gap leading to significant visible

light absorption for TiO2 [9e13]. However, these doping ions

alsocontribute in reducing the lifetimeofphoto-inducedcharge

carriers in doped TiO2. Thus an increased recombination rate of

charge carriers in doped TiO2 implies a reduction of the photo-

catalytic activity. Choi et al. [10] investigated the photocatalytic

activity of TiO2 doped with 21 different metal ions and proved

that both dopant type and concentration are the important

aspects to control the recombination processes. Doping with

small concentration of transition metals (lower than 1 at.%) in

TiO2 successfully reduced the recombination processes by

introducing traps for electrons and/or holes [11]. However, the

low concentration doping does not cause significant increase in

theabsorptionof the visible light inTiO2. Anefficientnarrowing

of theTiO2bandgap isachievedwithhighdopantconcentration

(about 5e10 at.%), but at the same time it introduces new

recombination centers for the photo-generated charges thus

increasing recombination processes [14]. From previous

consideration, it becomes important to solve the problem of

recombination for these moderately doped TiO2.

By using redox couple as electrolyte inside the photo-elec-

trochemical cell can reduce the recombination processes to

some extent on the surface of semiconductor [15,16]. In addi-

tion, application of an external bias can also offer better charge

separation. But both these solutions are not convenient for

commercial purposes related to solar energy conversion. More

convenient route to suppress the recombination processes is to

produce specific charge states to favor electrons and holes

trapping while having appropriate spatial separation. This can

be accomplished by coupling two semiconductor layers such

that the edge of the conduction band of the first semiconductor

is lower than that of the second one. If the semiconductor

layers are not much thick, then the photo-generated electrons

in one layer are easily injected into the second one [17]. In

addition, a thin space-charge layer of a few tens of nanometers

is formed near the interface of the semiconductors to make

their Fermi level equal [18]. The junction electric field is built up

in this space-charge layer of the electrode and is able to provide

the driving force to displace the photo-generated electrons

from one semiconductor to another thus favoring the electron-

hole separation. Thus creating number of interfaces or space-

charge layers in the photo-electrode provides better solution

for the problem related to recombination of charges. Many

TiO2-based coupled systems such as CdS/TiO2 [19], WO3/TiO2

[20], SnO2/TiO2 [21], SiO2/TiO2 [22] etc., have been used in the

past as photocatalysts. However, none was reported using

transition-metal -doped TiO2 in the coupled system where the

recombination rates are paramount.

In our previous work [23], we succeeded in reducing the

recombination rates of photo-generated charges by adopting

the ITO/Cr-doped TiO2 multilayer film structure with a number

of bilayers. Now in the present work we further check the

efficiency of the multilayer structure in increasing both charge

separation and charge-transfer, by using a different dopant

metal, namely V. TiO2 filmswere dopedwithV-metal, using co-

sputtering deposition technique, to enhance the visible light

absorption efficiency while depositing ITO/TiO2 multilayers for

reduction of the charge recombination rates. The photo-

catalytic efficiency of the multilayer-based TiO2 photo-elec-

trode was tested by measuring hydrogen production through

water-splitting in photo-electrochemical cell.

2. Experimental

RF-magnetron sputteringwasused to synthesize undoped- and

V-doped TiO2 thin films. Pure TiO2 disc (purity equal to 99.99%)

and Ar gas (purity equal to 99.99%) were used as sputtering

target and working gas, respectively. Before deposition, high

vacuumwith base pressure of<3� 10�5 Pawas attained, while

during deposition Ar pressure of 0.8 Pa was maintained in the

chamber. V-doped TiO2 films were synthesized by partially

covering the TiO2 target surfacewith small V-metal pellets. The

number of pellets on the TiO2 target was varied in order to

obtainV-doped filmsof four different atomic concentrations. In

order to remove any surface contamination the TiO2 target was

pre-sputtered for 20 min before film deposition. The TiO2 films

were sputter-deposited on both glass and Si (100) substrates at

room temperature using RF power of 150W. The sample-target

distance was kept constant at 5.5 cm for deposition of all the

samples. After deposition, no post annealing was performed.

TiO2 thin films were deposited on a conducting indiumetin

oxide (ITO) layer which provides electrical back contact to test

photocatalytic activity. The ITO layer, of about 80 nm, was

deposited on the glass slide before the deposition of TiO2 film

using magnetron sputtering with RF power of 100 W and

working Ar gas pressure of 0.8 Pa. Pre-sputtering of the ITO

target was also conducted to remove possible surface contam-

ination. Multilayer film was prepared by sequential in-situ

sputtering of ITO layer and then of V-doped TiO2: this forms

a single bilayer. Similarly, several numbers of bilayers (2-, 3-, 4-,

5-, 6- and 7-bilayers) were deposited in order to study the effi-

ciency of the multilayer structure in reducing recombination

ratesofholes andelectrons. The total thickness ofV-dopedTiO2

was kept constant, about 750 nm, in all the multilayer films by

controlling the sputtering deposition time. Thus, as the bilayer

number increases, the thickness of the V-doped TiO2 decreases

in each bilayer to maintain the constant total thickness of all

TiO2 films. The thickness of ITO (w80 nm) was kept constant in

all the multilayer films irrespective of number of bilayers. The

surfaces of all the films, that will be in contact with the elec-

trolyte, were always covered with pure TiO2 of w100 nm to

protect the metal ions from leaching out in the electrolyte.

The structural characterization of themultilayeredV-doped

TiO2 films was carried out by X-ray diffraction (XRD) technique

(Cu Ka radiation, l ¼ 1.5414 A�) in Bragg-Brentano (qe2q)

configuration. Surface electronic states of the photocatalyst

were established by using X-ray photoelectron spectroscopy

(XPS). X-ray photoelectrons spectra were acquired using an

SCIENTAESCA200 instrumentequippedwithamonochromatic

AlKa (1486.6 eV)X-Ray source and ahemispherical analyzer.No

electrical charge compensation was required to perform XPS

10 20 30 40 50 60 70

Inten

sit

y (A

rb

. u

nit

s)

2θ (Degree)

Pure TiO2

V 2 at%

V 4 at%

V 5 at%

V 6 at%

Anatase

Rutile

Fig. 1 e XRD spectra of undoped- and V-doped TiO2 films

deposited by RF-magnetron sputtering with different

dopants concentration.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 6 ( 2 0 1 1 ) 6 5 1 9e6 5 2 8 6521

analysis. Optical measurements in the UV and visible range

were performed using a Bruker IFS66 spectrometer equipped

with reflection and transmission units with near normal inci-

dence of the incoming beam. The wavelength range between

250 nm and 750 nmwas used to obtain the absorbance spectra

of samplesdepositedonglassslides.Thesurfacemorphologyof

the TiO2 samples was characterized by means of a scanning

electron microscope (SEM-FEG, JSM-7001F, JEOL) and atomic

composition was analyzed by Energy Dispersive Spectroscopy

(EDS, INCA PentaFET-x3) attached to SEM.

Photocatalytic activity ofmultilayer-V-dopedTiO2filmswas

tested by measuring both photocurrent and hydrogen produc-

tion rates by water-splitting. Photo-electrochemical cell con-

structed with TiO2 as photo-anode, Pt mesh as cathode, and

aqueous electrolytes was employed to measure photocatalytic

activity. In order to establish a chemical bias we used two

chambers containing electrolytes with different pH values. The

TiO2 photo-anode was kept in contact with a NaOH (1 M) solu-

tion, while the cathodewas immersed in aH2SO4 (1M) solution.

AsaltbridgecontainingNaCl (1M)wasusedtoprovideapath for

ion conduction between the two solutions. The photo-anode

was exposed to visible light, generated by a 250 W tungsten

halogen lamp, to measure both open-circuit photo-voltage and

closed-circuit photocurrent. Reactor is made up of borosilicate

glass, thus no UV filter was used and reactant water solution

acts as filter for IR radiation. Oxygen from the reaction chamber

was completely removed by purging the chamber with pure Ar

gas (99.9%) for 1 h before performing measurements. The

separate evolution of H2 originated by the water-splitting

process at the cathode was measured on-line, as a function of

time, by using a gas chromatographer (GC, Agilent MIRCOGC-

3000A). Finally, the hydrogen evolution was measured in both

ON- and OFF- light regimes.

3. Results and discussion

3.1. Characterization of photocatalyst

The surface morphology, as observed by SEM, of V-doped TiO2

films appears quite flat, smooth, and without major defects,

while the cross-section of the V-doped films shows a typical

dense columnar structure. By compositional line scan using

EDS, inthesampledvolume, theuniformdistributionofV-metal

in theTiO2 filmwas confirmedwhich is clearly connected to the

Table 1e Crystal size and intensity peak (A(101) and R(110)) ratienergy level values for undoped- and V-doped TiO2 films depoconcentrations.

V concentrationin TiO2 film (at. %)

Crystal size (nm) Intenof ana

Anatase(101)

Rutile(110)

Pure TiO2 45 6

2 at. % 39 12

4 at. % 35 11

5 at. % 37 12

6 at. % 39 10

adopted co-deposition procedure. The V concentration of about

2 � 0.2, 4 � 0.3, 5 � 0.4 and 6 � 0.4 at.% was obtained in four

different TiO2 films as confirmed by EDS.

Fig. 1 shows the XRD spectra of V-doped TiO2 film of

different concentration synthesized on amorphous glass by

sputter deposition. The crystal grain sizes of the anatase and

rutile phase for all the samples are calculated by the

DebyeeScherrer equation from peak A(101) at 25.2� and R(110)

at 27.5� respectively, and summarized in Table 1. The intensity

ratio of the anatase (101) to rutile (110) peak is also reported in

Table 1. The peaks observed in the spectra of undoped TiO2

thin film are mainly due to the anatase phase while some

additional reflexes of rutile phase could be identified in the

form of broad peak. Crystal size (Table 1) suggests that rutile

phase for pure TiO2 is almost amorphous with very finely

dispersed grain, while anatase phase is in nano-crystallized

form. By increasing the doping concentration of vanadium,

the intensity of anatase diffraction peak gradually decreases

and peak corresponding to rutile increases. The TiO2 doped

with high concentration of vanadium (w5 and w6 at.%)

exhibits peak of rutile phase prevailing over the anatase

phase. The reflexes of the rutile phase are now shifted to lower

2q values, as compared to those of pure TiO2 film. This indi-

cates a slight increase in the spacing value between the lattice

o of anatase/rutile phase, TiO2 optical band gap, and dopingsited by RF-magnetron sputtering with several V-

sity ratiotase/rutile

TiO2 optical bandgap values (eV)

Doping energylevel values (eV)

1.14 3.25 � 0.1 e

1.12 3.17 � 0.1 2.96 � 0.1

1.01 3.10 � 0.1 2.83 � 0.1

0.91 3.05 � 0.1 2.75 � 0.1

0.81 2.85 � 0.1 2.50 � 0.1

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 6 ( 2 0 1 1 ) 6 5 1 9e6 5 2 86522

planes which might be caused by the insertion of V in the

lattice. No characteristic peaks due to vanadium oxides (V2O5

or VO2) were detected in XRD pattern. These results indicate

that V ions are incorporated into the TiO2 lattice and promote

the anatase-to-rutile phase transformation. Zhao et al. [24]

have reported that upon increasing the vanadium concen-

tration, the anatase-to-rutile transformation was promoted

due to a lattice distortion of anatase caused by vanadium

incorporation, which favors transformation to rutile.

XPS was carried out to confirm the concentrations and to

determine the chemical states of V in the samples. Fig. 2a and

b show the XPS spectra of Ti2p and V2p electronic levels,

respectively, for pure TiO2 and V-doped TiO2 (w6 at.%). For pure

TiO2, two peaks, namely Ti2p3/2 and Ti2p1/2, appear at 459.35 and

465.09 eV that correspond to þ4 valence state of Ti in the

samples. For V-doped TiO2, the Ti2p peaks position was slightly

shifted by 0.35 eV towards positive binding energy value thus

proving that V ions are incorporated into TiO2 lattice and influ-

encing the local chemical state of Ti4þ ions. To achieve local

charge balance in the TiO2 lattice, some of the Ti ions may

acquire higher oxidation state by releasing electrons, a process

that explains theXPSpeak shift for V-dopedTiO2. XPS spectrum

of V2p3/2 level was deconvoluted into two peaks having binding

energy values of 517.3 and 516.2 eV attributed to V5þ and V4þ

states of vanadium respectively. This indicates that V species

exist in the lattice of TiO2 in the formofV5þ andV4þwith higher

525 520 515 510

V4+2p3/2

Binding Energy (eV)

In

te

ns

ity

(A

br. U

nits

)

V5+2p3/2

V5+2p1/2

V2p

470 465 460 455 450

Ti4+2p

3/2

Ti4+2p

1/2

Ti2p

Pure TiO2

V-doped TiO2

In

ten

sity (A

br.U

nits)

Binding Energy (eV)

a

b

Fig. 2 e X-ray photoelectrons spectra of (a) Ti2p and (b) V2p

levels for pure TiO2 and V-doped TiO2 (6 at.%).

content of V5þ ions as indicated by the area under the peak in

XPS spectrum (Fig. 2b).

The optical properties of the pure and V-doped TiO2 were

studied by measuring the absorption spectra ranging from UV

(250 nm) to visible (750 nm) wavelengths and presented in

Fig. 3. Using Tauc plot (Fig. 4), i.e. (ahy)1/2 versus (hy) (where, a is

the absorbance, h is Planck’s constant, and y is frequency), the

band gap energies were deduced by extrapolating the linear

region of the plot to intersect the photon energy axis: the

obtained values are reported in Table 1. The figures clearly

show a slight shift in the absorption band edge towards longer

wavelengths when increasing the V concentration in the TiO2

film. Most importantly the tail of the absorption band is

observed, indicating the existence of additional energy levels in

the band gap. This can be related to the charge-transfer from

the d orbital of V5þ and V4þ to the conduction band of TiO2. For

VeTiO2, V5þ and V4þ ions are incorporated into the TiO2 lattice

which forms three kinds of isolated impurity energy levels just

below the bottom of the TiO2 conduction band (CB) [25].

Because of these impurity energy levels in the band gap, the

electrons in the valence band can be excited to the impurity

energy levels and then subsequently excited to the CB by

absorption of visible light. This means that these impurity

energy levels are beneficial for extending the absorption spec-

trum wavelength towards the visible light region.

3.2. Photocurrent measurement

The above results indicates that 6 at.% V-doped TiO2 shows the

maximum lowering of the photon-absorbed energy due to the

impurity energy levels in visible range. Thus photocatalytic

activity measurements were carried out only with 6 at.%

V-dopedTiO2film (henceforth thisfilmwill bedesignatedasV6-

dopedTiO2 film) bymaking comparisonwith undopedTiO2. V6-

doped TiO2 film (about 750 nm thick) was deposited on an ITO

layer (80 nm), previously deposited on glass substrate. In order

to protect the film from possible corrosion induced by contact

with the electrolyte the obtained bilayer was covered with thin

layer of pure TiO2 (100 nm) and later this single-bilayer filmwas

used as the photo-electrode for water-splitting experiments.

350 400 450 500

Ab

so

rb

an

ce (A

rb

.u

nit)

Wavelength(nm)

V-2 at% V-4 at% V-5 at% V-6 at% TiO 2

Fig. 3 e UVeVIS absorption spectra of undoped- and

V-doped TiO2 films deposited by RF-magnetron sputtering

with different dopants concentration.

Fig. 4 e Tauc plot of undoped- and V-doped TiO2 films

deposited by RF-magnetron sputtering with different

dopants concentration.

Fig. 5 e SEM Images, in back scattering mode, of multilayer

film with 6-bilayers of ITO/V6-doped TiO2.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 6 ( 2 0 1 1 ) 6 5 1 9e6 5 2 8 6523

The undoped TiO2 film with total thickness of 850 nm was

similarly deposited on an ITO layer (80 nm) for the comparison.

The ITO layer provides the necessary electrical path for the

photo-generated electron from TiO2 to Pt electrode. Most

importantly the ITO/TiO2 stack film is able to partially limit the

recombination of photo-generated holes and electrons (hþ and

e�) because the conduction band edge of ITO (w�4.5 eV) is at an

energy value lower than that of TiO2 (w�4.0 eV) thus making

favorable for the injection of electrons from the later into the

former.

The photo-induced voltage and current are measured for

undoped- and V6-doped TiO2 in light-ON regime in photo-

electrochemical cell with open- and closed-circuit configura-

tion, respectively. The open-circuit photo-voltage value

results more than w1.4 V, for both these TiO2 samples, which

is greater than the required voltage (1.23 V) to break the water

molecule. However, the photocurrent measured using V6-

doped TiO2 (w53 mA) is negligible as compared to pure TiO2

films (w290 mA) in the light-ON regime. This is mainly due to

recombination of the photo-generated charges on defects

associated to the dopant V5þ/V4þ ions.

To efficiently suppress the charge recombination rates it is

important to immediately remove the photoelectron from the

generation site before it recombines with the holes through

radiative or non-radiative processes. This fast separation can

be achieved by coupling two semiconductors with appropriate

conduction energy levels. For instance, as mentioned previ-

ously, in the ITO/TiO2 coupled system, the conduction band

edge of ITO (w�4.5 eV) is at an energy value lower than that of

TiO2 (w�4.0 eV) thus making favorable for the injection of

electrons from the later into the former. Once the photoelec-

tron from TiO2 is injected in ITO, it has a very low probability

of back transfer and, as a result, electrons and holes are

definitely separated. However, to be efficient, ITO should be

located near the electron generation site to provide the driving

force for the photoelectron transfer.

Following the above strategy, we have depositedmultilayer

films with different numbers (2-, 3-, 4-, 5-, 6- and 7-bilayers) of

ITO/V6-doped TiO2 bilayers and finally covering with thin

layer of pure TiO2 (100 nm) the surface layer that will be in

contact with the electrolyte. The bottom portion of the overall

multilayers (see Fig. 7b below) is also in contact with the

electrolyte through the pure TiO2 (100 nm) layer (while the

corresponding top portion is electrically connected to cathode

of the photo-electrochemical cell). In all the multilayer films

the total thickness of V6-doped TiO2 was kept constant

(750 nm) by decreasing the thickness of each V6-doped TiO2

layer when increasing the number of the bilayers. The thick-

ness of ITO (w80 nm) was kept constant in all the multilayer

films irrespective of number of bilayers.

The cross-section SEM image, in back scattering mode, of

a multilayer film with 6-bilayers of ITO/V6-doped TiO2 is

presented in Fig. 5. The white layer of ITO can be easily

distinguished from the dark layer of V6-doped TiO2 and for

this reason both the periodicity and continuity of the layers

are clearly visible. The Fig. 6 shows the thickness of single

V6-TiO2 layer, measured through cross-section SEM images,

as a function of number of bilayers. The figure confirms that as

the number of bilayers increases, the thickness of single

V6-doped TiO2 layer decreases.

ITO is a transparent conducting oxide with a direct band

gap in the range of 3.9e4.2 eV which is significantly higher

than the absorption edge of V6-doped TiO2 layer thus ITO

deposited on the top of TiO2 layer does not affect the amount

of photons reaching the TiO2 layer. This feature was experi-

mentally confirmed by measuring the absorbance of the V6-

doped TiO2eITO bilayers, with ITO on top, in the UVeVisible

range: the spectra showed similar red shift in the band edge as

that observed with single layer.

Photo-catalytic activity of themultilayer filmswas tested by

using ITO/V6-doped TiO2 multilayer films as photo-anode in

photo-electrochemical cell and measuring photo-voltage and

photocurrent for light-ON regime. Each ITO layer in the multi-

layer films is electrically connected to theplatinumcathodeand

so as soon as the electrons are injected from the V6-doped TiO2

layers into the ITO layers, they are immediately transferred to

the cathode. The measured photo-voltage of all the multilayer

films (more than 1.4 V)was always greater than that required to

split the water molecule. Fig. 6 reports the measured photo-

current as a function of number of bilayers. During the

measurements, the same surface area (13.75 cm2) is exposed to

the light for allmultilayer films.As thenumberof ITO/V6-doped

Fig. 6 e Photocurrent and thickness of each V6-doped TiO2

layer as a function of number of bilayers.

Fig. 7 e Schematic representation of the mechanism involving th

film with 4-bilayers of ITO/V-doped TiO2.

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 6 ( 2 0 1 1 ) 6 5 1 9e6 5 2 86524

TiO2 bilayers increases, the photocurrent increases and reaches

themaximumvaluewith6-bilayers (w800mA).However,with7-

bilayers thephotocurrentdecreases tothevalueofw280mA.The

photocurrent value measured with 6-bilayers V6-doped TiO2

film(w800mA) isabout2.75 timeshigher thanthatobtainedwith

pure TiO2 (w290 mA) with same total thickness of TiO2. These

results are consistent with that reported in our previous work

for ITO/Cr-doped TiO2 multilayer films where, similarly,

photocurrent increases with number of bilayers and reaches

maximumwith 6-bilayers film [23].

3.3. Electron flow mechanism

The above result can be explained on the basis that when

semiconductors having different energy bands are brought in

contact with each other, a band bending occurs on both sides

near the interface to make Fermi level equal. This creates an

e electron flow in: (a) single-bilayer film, and (b) multilayer

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 6 ( 2 0 1 1 ) 6 5 1 9e6 5 2 8 6525

interface space-charge layer having thickness of several tens of

nanometers where a large electric field is developed. When

photo-electrons are generated in this interface region (or very

near) they are instantaneously pushed out to the adjacent

semiconductor due to the driving force provided by the electric

field and the electron-hole separation is definitely obtained,

a process well known in pen junction semiconductors. Simi-

larly in the present ITO/TiO2 system, the interface region pres-

entselectrical features thatprovideanefficient route for the eeh

separation. Hence by adopting a multilayer structure, we are

abletoproducemanyinterfacesof ITO/TiO2whichestablishfast

transport channels along with efficient eeh separation.

A scheme is reported in Fig. 7 to understand in detail the

mechanismof the electronflow in themultilayer filmsandhow

the number of bilayers affects the generation of the photocur-

rent. The photocurrentmeasuredwith single bilayer of ITO/V6-

doped TiO2 film (Fig. 7a) is very low due to the presence of only

one interface of ITO/V6-doped TiO2 with the thick layer of

V6-doped TiO2 (750 nm) over it (Fig. 7a). Most of the photo-

electronsaregenerated in thebulkof thedopedTiO2where they

have to travel some distance within the TiO2 itself before

entering the space-charge region of the interface to feel the

driving electric field for final injection into ITO. In addition, the

probability of eeh recombination is much higher in V6-doped

TiO2due to thepresenceofrecombinationsitesassociated to the

V5þ/V4þ ions. On contrary, the photocurrent increases signifi-

cantly from w53 mA (for single-bilayer film) to w456 mA by

depositing 4-bilayers film. In this film, the previous V6-doped

TiO2 layer having thickness of 750 nm was divided into four

layers of w185 nm by introducing three more ITO layers. The

resulting film now contains seven ITO/TiO2 interfaces where

space-charge layer is formedand thedecreased thicknessof the

V6-doped TiO2 layers assures that the photo-generated elec-

trons have to travel a short distance into TiO2 (the unshielded

region) before being injected into the ITO layer where recombi-

nation is definitely hindered (Fig. 7b). Another relevant point to

be considered is that in single-bilayer film the photo-electrons,

generated in the V6-doped TiO2 layer, have to move only along

one direction towards the ITO layer placed at the one end of the

layered structure. On the contrary, in multilayer film each

V6-doped TiO2 layer is sandwiched between the two ITO layers

and so the photo-electrons canmove along either of two direc-

tions toavoid recombination (Fig.7b).All thesereasons favor the

increment of the photocurrent in the 4-bilayers film.

Linear increment in the photocurrent as increase in the bila-

yersnumberwasobserved.This ismainlyattributed to thebetter

conditions for charge separation and transport with reduced

number of recombination processes: here we have to further

underline the key role of number of ITO/TiO2 interfaces. The

measured photocurrent reaches the maximum value (w800 mA)

with 6-bilayers film; the measured photocurrent is one order of

magnitude higher than that measured with single-bilayer film

(w53mA). For6-bilayersfilm, thethicknessofV6-dopedTiO2 layer

(125 nm) in each bilayermay be lower than the total thickness of

the space-charge layer of the pertinent interface. Thus photo-

electrons generated in the V6-doped TiO2 are already in space-

charge region where the electric field provides them the driving

force to be instantaneously injected into the ITO layers: this

means that the 6-bilayers film creates the best conditions for

charge separation. In addition, this film is able to absorb more

visible light than pure TiO2 due to the V5þ/V4þ impurity energy

levels, hence generating higher amount of photoelectrons. With

7-bilayersfilmthephotocurrentdecreases (Fig.6);wesuggest this

might be due to the poor crystalline degree of very thin films of

TiO2 (92 nm). Takahashi et al. [26] andNasr et al. [17] showed the

increase in photocurrent by decreasing the TiO2 layer thickness

attaining maximum photocurrent value at particular thickness.

Below this thickness the photocurrent decreases drastically and,

as indicated by the authors, this is attributed to the low crystal-

linity achievedwith thinner films. For themomentwith our XRD

setup it is difficult to acquire the spectra of such a thin layer of

TiO2.However,wewilladdress thisopenproblemwithdedicated

experiments in the future.

3.4. Photocurrent kinetic studies

The better charge separation and reduced recombination

processes observed with multilayer films was further analyzed

by studying the photocurrent kinetics curve. Fig. 8A shows the

schematicdiagramofphotocurrent transient curve. Initially the

anodic photocurrent spike (I(i)) is obtained after the exposure of

light which ismainly due to the separation of the electron-hole

pairs bymovement ofholes towards the semiconductor surface

where they are trapped or reduced by the species in the elec-

trolyte, while the electrons are transported to the back contact.

After I(i) has been attained, the photocurrent decays with time,

owing to the charge recombination processes, until a steady-

state photocurrent (I(st)) is reached. The holes reaching the

semiconductorsurfacemay, insteadofcapturingelectronsfrom

the electrolyte, accumulate at the surface and recombine with

electronsfromtheconductionband, i.e. thedecay isdetermined

by the rate at which minority carriers trapped at surface states

capturemajoritycarriers [27].Thephotocurrenttransientcanbe

defined by the following kinetic equation [28]:

D ¼ exp

��ts

�(1)

Where D is defined as

D ¼ IðtÞ � IðstÞIðiÞ � IðstÞ (2)

Here, s is the transient timeconstant, I(t) is the current at time

t, I(i) is the current at t ¼ 0, and I(st) is the stationary current.

Fig. 8B, C, andD show the plot of lnD vs. time for pure TiO2 film,

single bilayer, and 6-bilayers of ITO/V6-doped TiO2 films,

respectively. The slope of this plot provides swhich is related to

the time for charge recombination processes in the films. The

measured s for V6-doped TiO2 with single-bilayer film is about

three times lower than that for the undoped TiO2. Specifically,

thephotocurrentdecayswithinafewseconds inV-dopedTiO2 is

due to recombinationof thephoto-generatedchargesondefects

associated to the V5þ/V4þ ions. On the contrary, with 6-bilayers

the transient time constant is about two times greater than that

forundopedTiO2filmwhere there arenodefect sites. Thisagain

proves that by introducing space-charge regions in form of ITO/

TiO2 interfaces and by decreasing the thickness of TiO2 layer

favors charge separation while recombination processes are

suppressed even if recombination sites such as V5þ/V4þ are still

present.

5 10 15 20

-0,8

-0,7

-0,6

-0,5

-0,4

-0,3

-0,2

ln

D

Time (sec)

Pure TiO2

Slope = -0.0235

= 42.5 sec

B

50 55 60 65 70 75

-1,4

-1,2

-1,0

-0,8

-0,6

-0,4

-0,2

C

ln

D

Time (sec)

TiO2+V-6 at%

Single Bilayer

slope= -0.06365

= 15.7 sec

light off

t(f)

I ph(A

)

Time (sec)

I(i)

I(st)

t(i)

light on

Pure TiO2

A

0 10 20 30 40 50

-1,1

-1,0

-0,9

-0,8

-0,7

-0,6

-0,5DTiO

2+V-6 at%

6-Bilayers

ln

D

Time (sec)

slope= -0,01147

= 87.2 sec

Fig. 8 e (A) Schematic representation of the photocurrent transient curve. Normalized plot of current-time dependence for:

(B) pure TiO2, (C) single bilayer, and (D) 6-bilayers of ITO/V6-doped TiO2 multilayer film.

Table 2 e Photo-voltage and photocurrent measured fordifferent numbers of ITO/TiO2 bilayers by keepingconstant the thickness of each single bilayer.

Number of ITO/TiO2

bilayersPhoto-voltage

(V)Photo-current

(mA)

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 6 ( 2 0 1 1 ) 6 5 1 9e6 5 2 86526

3.5. Effect of number of bilayers of equal thickness

The effect of the number of bilayers on the photocurrent of

V-doped TiO2 is studied. For this purpose, different number of

ITO/TiO2 bilayerswasdepositedbykeeping constant thickness

of the V6-doped TiO2 and ITO layers (125 and 80 nm respec-

tively). Thephoto-voltage andphotocurrentmeasured in light-

ON regime with these multilayer films are reported in Table 2.

All the multilayer films were able to produce photo-voltage

greater than that required to split the water irrespective of

number of bilayers. On the other hand, the photocurrent

increases with the number of bilayers and reaches a nearly

constant value for the highest number of bilayers (6 and 7).

A saturation effect is clearly expected on physical basis but

what is important here is that the photocurrent value depends

not only on thickness but also on the number of bilayers, i.e. on

the ability of the multilayer structure to efficiently avoid

charges recombination. Here, another important point is that

with 7 bilayers there is not a dramatic decrease of the photo-

current as observed in Fig. 6. This supports the point that the

low crystallinity achievedwith very thin films is deleterious to

photocurrent generation.

1 1.446 230

2 1.458 390

4 1.403 567

6 1.460 799

7 1.424 786

3.6. Hydrogen production measurement

The H2 generation by photocatalytic water-splitting was

studied in photo-electrochemical cell using V-doped TiO2

multilayer films. Fig. 9 shows the H2 evolution, as function of

time, from water-splitting by irradiating multilayer films

(single, 4-, and 6- bilayer of ITO/V6-doped TiO2) with light

(250 W tungsten halogen lamp) which contains mostly visible

light and includes a very small portion of UV light (1%). The

amount of the photocatalytically produced H2 increases line-

arly with the exposure time. The H2 production completely

stops after termination of light irradiation: this proves that H2

is only produced photocatalytically. The constant production

rate observed in the present case is simply due to the employ

of TiO2 photocatalyst in form of thin film in an apparatus

where H2 and O2 evolve separately. As the number of bilayers

increase the H2 generation rate increases thus showing the

same trend as that obtained for the photocurrent values, and

Fig. 9 e Hydrogen evolution, as a function of time, from

water-splitting in photo-electrochemical cell containing

two electrolytes of different pH by using photo-anode as

sputter-deposited multilayer films with single, 4- and

6-bilayers of ITO/V6-doped TiO2 (see procedure in ref. [27]).

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 6 ( 2 0 1 1 ) 6 5 1 9e6 5 2 8 6527

maximum rate is obtained with 6-bilayers. The obtained high

generation rate is again due to better photo-generated charge

separation induced by the space-charge layers in form of ITO/

TiO2 interfaces. The H2 generation rate was measured to be

31.2 � 0.1 mmol/h for 6-bilayers based V6-doped TiO2 multi-

layer film and is about 2.5 times higher than that measured

with pure TiO2 film (12.5 � 0.1 mmol/h) as reported in ref [29].

During the measurement, the same surface area (13.75 cm2) is

exposed to the light for all the three films. The greater value of

the H2 generation rate obtained with V6-doped TiO2 multi-

layer film is definitely attributed both to the visible light

absorption which is higher than pure TiO2 and to reduced

charge recombination processes. This result is consistentwith

our previous work [23] on Cr-doped TiO2 films, which also

showed significant enhancement in H2 generation rate

(24.4 mmol/h) by adopting multilayer structure for better

charge separation and transfer.

However, a comment must be included now to explain the

hydrogen production results. Indeed, while the photo-gener-

ated electrons are easily collected by the established electrical

contact to produce the measured photocurrent, the positive

holes that are produced inside the V-doped TiO2 layers sand-

wiched by the ITO films, may arrive at the TiO2 irradiated

surface that is in contact with water. Since the area interacting

with electrolyte at the bottom portion (see again Fig. 7b) of the

TiO2 film (total thickness of 850 nm) is negligible as compared

to the irradiated surface, we suggest that holesmigrate through

the bottom deposited TiO2 layer towards the irradiated surface

were theH2Omolecules are get oxidized.Weplan to investigate

this point as well as the role of the interface between V-doped

TiO2 and pure TiO2 in dedicated experiments.

4. Conclusions

TiO2 doped with V ions (w6 at.%) shows an increase in the

optical absorption efficiency due to the formation of impurity

energy levels in the band gap. However, in photo-electro-

chemical cell, negligible photocurrent is measured with ITO/

V-doped TiO2 (6 at.%) single bilayer, a problem that we attri-

bute to the increased recombination rate of the photo-gener-

ated charges. To reduce the charge recombination rate we

prepared multilayer films with different numbers of ITO/V6-

doped TiO2 bilayers (namely, 2-, 3-, 4-, 5-, 6- and 7-bilayers) by

keeping the total thickness of TiO2 constant. The reduced

thickness of the V-doped TiO2 film, deposited on ITO, signifi-

cantly contributes to reduce the charge recombination rates.

This is because the generated photoelectrons, traveling into

TiO2 film of limited thickness, rapidly enter the space-charge

interface of the ITOeTiO2 films from where they are instan-

taneously injected into the ITO layer and then removed

towards the cathode of the photo-electrochemical cell: here

Hþ reduction occurs. When the multilayer film is exposed to

visible light, we observe that the photocurrent increases as

function of the number of bilayers. Themaximumvalue of the

photocurrent is obtained with 6-bilayers of ITO/V6-doped

TiO2. The enhanced photocurrent is attributed to both higher

absorption of visible light by V-doped TiO2 and to the number

of space-charge ITO/TiO2 interfaces in multilayer films. With

six bilayers, the H2 production rate obtained through water-

splitting is about 31.2 mmol/h.

Acknowledgments

We thank Nicola Bazzanella for SEM-EDS analysis, Cristina

Armellini for XRD analysis, Lucia Calliari for XPS analysis and

Damiano Avi for hydrogen production measurement. The

research activity is financially supported by “Fondazione

Cassa di Risparmio di Trento e Rovereto”.

r e f e r e n c e s

[1] Cheekatamarla PK, Finnerty CM. Reforming catalysts forhydrogen generation in fuel cell application. J Power Sources2006;160:490e9.

[2] Momirlan M, Veziroglu TN. Current status of hydrogenenergy. Renew Sustain Energy Rev 2002;6:141e79.

[3] Bak T, Nowotny J, Rekas M, Sorrell CC. Photo-electrochemicalhydrogen generation from water using solar energy. Int JHydrogen Energy 2002;27:991e1022. Material-related aspects.

[4] Zhang H, Quan X, Chen S, Zhao H, Zhao Y. Fabrication ofphotocatalytic membrane and evaluation its efficiency inremoval of organic pollutants from water. Sep Purif Technol2006;50:147e55.

[5] Yamashita H, Harada M, Misaka J, Takeuchi M, Neppolian B,Anpo M. Photocatalytic degradation of organic compoundsdiluted in water using visible light-responsive metal ion-implanted TiO2 catalysts: Fe ion-implanted TiO2. Catal Today2003;84:191e6.

[6] Anpo M, Dohshi S, Kitano M, Hu Y, Takeuchi M, Matsuoka M.The preparation of highly efficient titanium oxide- basedphotofunctional materials. Annu Rev Mater Res 2005;35:1e27.

[7] Fujishima A, Zhang X, Tryk DA. TiO2 photocatalysis andrelated surface phenomena. Surf Sci Rep 2008;63:515e82.

[8] Dholam R, Patel N, Adami M, Miotello A. Hydrogenproduction by photocatalytic water-splitting using Cr- or

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 6 ( 2 0 1 1 ) 6 5 1 9e6 5 2 86528

Fe-doped TiO2 composite thin films photocatalyst. Int JHydrogen Energy 2009;34:5337e46.

[9] Ni M, Leung MKH, Leung DYC, Sumathy K. A review andrecent developments in photocatalytic water-splitting usingTiO2 for hydrogen production. Renew Sustain Energy Rev2007;11:401e25.

[10] Choi WY, Termin A, Hoffmann MR. The role of metal iondopants in quantum-sized TiO2: correlation betweenphotoreactivity and charge carrier recombination dynamics.J Phys Chem 1994;84:13669e79.

[11] Litter MI, Navıo JA. Photocatalytic properties of iron-dopedtitania semiconductors. J Photochem Photobio A Chem 1996;98:171e81.

[12] Khan MA, Woo SI, Yang OB. Hydrothermally stabilized Fe(III)doped titania active under visible light for water splittingreaction. Int J Hydrogen Energy 2008;33:5345.

[13] Zhu J, Deng Z, Chen F, Zhang J, Chen H, Anpo M, et al.Hydrothermal doping method for preparation of Cr3þ-TiO2

photocatalysts with concentration gradient distribution ofCr3þ. Appl Catal B Environ 2006;62:329e35.

[14] Radecka M, Rekas M, Trenczek-Zajac A, Zarrzewska K.Importance of the band gap energy and flat band potentialfor application of modified TiO2 photoanodes in waterphotolysis. J Power Sources 2008;181:46e55.

[15] Kozlova EA, Korobkina TP, Vorontsov AV, Parmon VN.Enhancement of the O2 or H2 photoproduction rate in a Ce3þ/Ce4þeTiO2 system by the TiO2 surface and structuremodification. Appl Catal A Gen 2009;367:130e7.

[16] Abe R, Sayama K, Arakawa H. Significant influence of solventon hydrogen production from aqueous I3

�/I� redox solutionusing dye-sensitized Pt/TiO2 photocatalyst under visible lightirradiation. Chem Phys Lett 2003;379:230e5.

[17] Nasr C, Kamat PV, Hotchandani S. Photoelectrochemistry ofcomposite semiconductor thin films. Photosensitization ofthe SnO2/TiO2 coupled system with a ruthenium polypyridylcomplex. J Phys Chem B 1998;102:10047e56.

[18] Morrison SR. Electrochemistry at semiconductor andoxidized metal electrodes. New York: Plenum Press; 1984.

[19] Chi Y, Fu H, Qi L, Shi K, Zhang H, Yu H. Preparation andphotoelectric performance of ITO/TiO2/CdS composite thinfilms. J Photochem Photobio A Chem 2008;195:357e63.

[20] Akurati KK, Vital A, Dellemann JP, Michalow K, Graule T,Ferri D, et al. Flame-made WO3/TiO2 nanoparticles: Relationbetween surface acidity, structure and photocatalyticactivity. Appl Catal B Environ 2008;79:53e62.

[21] Shang J, Yao W, Zhu Y, Wu N. Structure and photocatalyticperformances of glass/SnO2/TiO2 interface composite film.Appl Catal A Gen 2004;257:25e32.

[22] Beyers E, Biermans E, Ribbens S, De Witte K, Mertens M,Meynen V, et al. Combined TiO2/SiO2 mesoporousphotocatalysts with location and phase controllable TiO2

nanoparticles. Appl Catal B Environ 2009;88:515e24.[23] Dholam R, Patel N, Santini A, Miotello A. Efficient indium tin

oxide/Cr-doped-TiO2 multilayer thin films for H2 productionby photocatalytic water-splitting. Int J Hydrogen Energy 2010;35:9581e90.

[24] ZhaoG,KozukaH,LinH,YokoT.Sol-gelpreparationofTi1-xVxO2

solid solution film electrodes with conspicuous photoresponsein the visible region. Thin Solid Films 1999;339:123e8.

[25] Zhou W, Liu Q, Zhu Z, Ji Z. Preparation and properties ofvanadium-doped TiO2 photocatalysts. J Phys D Appl Phys2010;43:035301e6.

[26] Takahashi M, Tsukigi K, Uchino T, Yoko T. Enhancedphotocurrent in thin film TiO2 electrodes prepared by solegelmethod. Thin Solid Films 2001;388:231e6.

[27] Hagfeldt A, Lindstrom H, Sodergren S, Lindquist SE.Photoelectrochemical studies of colloidal TiO2 films: theeffect of oxygen studied by photocurrent transient. JElectroanal Chem 1995;381:39e46.

[28] Redecka M, Wierzbicka M, Komornicki S, Rekas M. Influenceof Cr on photoelectrochemical properties of TiO2 thin films.Physica B 2004;348:160e8.

[29] Dholam R, Patel N, Adami M, Miotello A. Physically andchemically synthesized TiO2 composite thin films forhydrogen production by photocatalytic water splitting. Int JHydrogen Energy 2008;33:6896e903.