Influence of thermal treatment on the photoelectrochemical behaviour of WO3 photoanodes electrochemically grown

Solar Energy Materials 11 (1985)419-433 North-Holland, Amsterdam

419

INFLUENCE OF THERMAL TREATMENT ON THE PHOTOELECTROCHEMICAL BEHAVIOUR OF WO 3 PHOTOANODES ELECTROCHEMICALLY GROWN * F. DI QUARTO, A. DI PAOLA, S. PIAZZA and C. SUNSERI Istituto di Ingegneria Chimica, Universitit di Palermo, Viale delle Scienze, 90128 Palermo, Italy Received 6 August 1984; in revised form 25 October 1984 The influence of anodizing parameters and thermal treatments on the photoelectrochemical behaviour of the corrosion layers grown on tungsten is presented and discussed. Very significant changes are observed in the photocurrent spectra as well as in the photocharacteristics of anodic oxide films grown in different conditions. Large increases in the measured photocurrent and in the resistance against photocorrosion are observed after thermal treatment. An explanation of these findings is suggested according to the experimental results. The influence of morphology, composition and crystallographic structure on the photoelectrochemical behaviour of different tungsten oxide photoanodes is presented and discussed by taking into account possible changes in the solid-state properties of the films. The possible formation of defective structures during a long-term photoelectrolysis experiment is also discussed.

1. Introduction

The practical application of photoelectrochemical (PEC) cells for solar energy conversion is related to the solution of several problems in order to obtain: (a) inexpensiveness of the photosensitive materials as well as of the electrodes preparation; (b) good match of the semiconducting electrodes band gap with the solar spectrum; (c) long durability against the photocorrosion. Although in the last years a substantial progress has been made in the understanding of the photoelectrochemical behaviour of semiconductor/liquid junctions, many aspects of the behaviour of amorphous and polycrystalline semiconductors still need to be investigated. In this frame we have investigated the influence of different methods of preparation on the behaviour of tungsten oxides photoanodes. The band gap of these oxides is reported in the literature [1-8] to range between 2.5 and 3.4 eV, depending on the crystallinity degree and stoichiometric composition of the oxide. For this reason tungsten oxides seem unsatisfactory for practical applications, although some recent papers in the field of powder-photocatalysis have raised new interest for these materials [9,10].

* Part of this work was presented at the 33th ISE Meeting, Lyon, September 1982.

0165-1633/85/$03.30 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

420

F. Di Quarto et al. / wo~ photoanodes

Apart from the practical application in PEC cells, both the semiconducting [11] and the photoelectrochemical behaviour [12] of these oxides present some features which make these materials interesting for a more general understanding of the photoelectrochemical behaviour of the semiconductor/electrolyte junctions. Many papers [8,12-16] have shown that the photoelectrochemical behaviour of tungsten oxides is widely affected by the method of preparation, the thermal annealing and the ageing effects under illumination. Anyway no systematic analysis has been performed on the combined effects of these variables. The aim of this work is to get more information on the influence of these variables on the photoelectrochemical behaviour of tungsten oxides. Moreover some explanations for the observed behaviour will be proposed by taking into account the possible changes in the solid-state properties of the oxide layers.

2. Experimental The electrodes were obtained from spectrographically pure tungsten sheets and rods. The methods of surface preparation have been described elsewhere [7,18]. The anodic films were grown in H3PO 4, HzSO 4 and HNO 3 solutions at a constant anodizing current density of 8 mA cm 2 until various final voltages were reached. The annealing process of the films was performed in a quartz tube under a preheated argon stream. Each film was characterized by X-ray analysis, scanning electron microscopy and photoelectrochemical measurements before and after annealing. Photoelectrochemical and capacitance measurements were performed in a quartz electrochemical cell containing a 0.5 M H2SO 4 solution. The experimental apparatus has been described elsewhere [8]. The photoelectrochemical measurements were made with a monochromator exit slit-width of 3 mm. The photocurrent spectra were corrected for the lamp emission. The counter electrode was a 10 cm 2 Pt foil and the reference electrode was Hg/Hg2SO4/0.5 M HzSO 4 (mercurous sulphate electrode, MSE). Solutions were prepared from distilled water and analytical grade reagents. Experiments were carried out at room temperature (25°C), unless otherwise stated.

3. Results and discussion In previous papers [17,18] we have shown the possibility of preparing different tungsten oxide films by galvanostatic anodization of metal electrodes in suitable conditions. In these works it has been shown that morphology, crystallinity, chemical composition and thickness of these oxides can be controlled by a careful choice of the anodization parameters (current density, temperature and electrolyte composition, final anodization voltage, anodization time). Moreover, a very good reproducibility of the characteristics of the corrosion layers was observed in these experiments. In this paper we have chosen to investigate the photoelectrochemical behaviour of three different anodic oxide layers which are representative of the different types of anodic films grown on tungsten in different anodization conditions. A detailed

hydrated porous layers

anhydrous porous layers

barrier type films

Structure (main phase)

amorphous

monoclinic

monoclinic

triclinic triclinic

triclinic

orthorhombic

triclinic

triclinic

Oxide layer

WO 3

WO 3

WO 3

WO 3 WO 3

WO 3

WO3.H20

WO 3

WO 3

Table 1 O p t i c a l b a n d g a p s o f d i f f e r e n t t u n g s t e n o x i d e layers

1 N H 2 S O 4 at 7 0 ° C for 1 h 1 N H2SO 4 at 70°C for 1 h

1 N H 2 S O 4 at 7 0 ° C for 1 h

1 N H N O 3 for 10 m i n

1 N H N O 3 f o r 10 m i n 1 N N N O 3 f o r 10 m i n

0 . l N H 3PO4 before breakdown 0.1 N H 3 P O 4 before breakdown 0.1 N H 3 P O 4 before breakdown

Anodic oxidation

T = 3 5 0 ° C for 3 h under argon T = 3 5 0 ° C for 17 h under argon

-

T = 3 5 0 ° C for 3 h under argon T = 3 5 0 ° C for 3 h under argon

T - 3 5 0 ° C for 3 h under argon T = 3 5 0 ° C for 3 h under argon

-

Thermal treatment

-

-

-

photoelectrolysis for 48 h in 1 N H 2 S O 4 at ~ = 380 n m

-

photoelectrolysis for 18 h in 1 N H 2 S O , , at A = 300 n m

-

-

Other treatment

2.6_+0.05

2.6 + 0 . 0 5

2.7 3.05_+0.1

2.55 _+ 0.05

2.55 + 0.05 2.55 + 0 . 0 5

2.75 + 0.05

2.75 + 0 . 0 5

3.05 +_0.1

Optical b a n d g a p (eV)

,...a

z~

e,

F. Di Quarto et al. / WO~ photoanodes

422

I

d

I

.,~

I

®~"

'~~ =~

¢0 . ~ O

C

~i,,

22_

a

10

.....

~

I

t

I

20

30

40

2# Fig. 1. X-ray diffraction patterns for different tungsten oxide fihns: (a) amorphous film obtained in 0.1 N H ~PO 4 solution: (b) monoclinic WO3 obtained after annealing of the amorphous film: (c) triclinic WO3 anodized in 1 N HNO~ solution until breakdown: (d) orthorhombic W O 3 - H 2 0 obtained in 1 N H 2SO4 at 70°C.

discussion of the morphology of these layers has been reported in previous papers [16-18] but some results will be mentioned in the following. In fig. 1 are shown the X-ray diffraction patterns of tungsten oxide layers grown in different experimental conditions. For a better comparison table 1 reports the morphology, the crystallographic structure, the method of preparation, the treatment and the optical band gap of the different films. Since the determination of the exact chemical composition of tungsten oxides is very difficult also for WO 3 single crystals [2,11], the identification of polycrystalline corrosion layers made by X-ray analysis must be taken with some caution as for the exact stoichiometry of the layers. More information on this aspect will be presented and discussed in the following. 4. C r y s t a l l i z e d a n o d i c barrier f i l m s

In the figs. la and l b are reported the X-ray diffraction patterns of a compact barrier-type anodic film before and after thermal treatment under argon atmosphere

F. Di Quarto et al. / w o ¢ 4

I

I

photoanodes I

I

423

800

a

3

8OO

2

400

1

200

E

(J

.i

0

300

300

400

400

O

nm Fig. 2. Photocurrent action spectra of a film grown in 0.1 N H2SO 4 solution until 70 V, before (curve a) and after (curve b) thermal treatment.

at 350°C. Due to the annealing the amorphous films (a-WO3) crystallize originating a monoclinic WO 3 phase. The influence of crystallization on the solid-state properties of the WO 3 anodic films was evidenced by: (a) a noticeable decrease in the measured optical band gap value, E~ p (see table 1 and ref. [16]); (b) a large change in the donor concentration, No; (c) an increase in the measured photocurrent accompanied by a change of the current-voltage characteristics of the photoelectrodes. In fig. 2 are reported the photocurrent action spectra of amorphous and crystallized electrodes. From the (IphhP) °5 vs. hp plots, by assuming non-direct and indirect optical transitions, for the amorphous and the crystallized electrode, respectively, an optical band gap of 2.75 eV was found for the crystalline monoclinic WO 3 electrodes, whilst a larger value of the band gap (3.05 eV) was obtained for the amorphous films. According to the theory of amorphous semiconductors [19], a possible explanation for the higher optical band gap observed in a-WO 3 films has been suggested [16] by taking into account the existence of localized states near the mobility edges, due to the absence of a long-range order. In fig. 3 are reported Mott-Schottky plots obtained at different frequencies for a crystallized film. By assuming a roughness factor equal to unity and a dielectric constant value equal to that one measured for a-WO 3 (Cox=44 [17]), donor concentrations of 7 × 1020 and 8.6 × 1020 cm -3 were measured at a frequency of 160 Hz in the low and high electrode potential range, respectively. The increase of the slope at higher frequencies (1 kHz and 5 kHz) could be attributed to the decrease of the Cox value, in agreement with the behaviour reported for a-WO 3 films [20]. By

[: Di Quarto et al. / W O ¢ photoanode,~

424

10.0

/// ///

7.5

I

0

o//

E

U

u.

-

5.0

:3_

0 2.5

tO

0.0 r

I 0.0

,

I

,

1.0 U~ ,

I

2.0

,

I

3.0

V vs M S E

Fig. 3. M o t t - S c h o t t k y plots at three different frequencies of a film grown in 0.1 N H 3 P O 4 solution until 100 V and annealed for 3 h at 350°C under argon atmosphere. O : 160 Hz: D: 1 kHz; ~: 5 kHz.

assuming a constant Cox value, the N o values obtained at 5 kHz are 4 × 1 0 20 and 7.8 x 10 2o cm 3. These N D values are about three orders of magnitude larger than the donor concentrations estimated for an amorphous film before annealing [7], and are independent of the initial film thickness. No direct influence on the N D value was observed increasing the annealing time until 17 h. The large increase in the N D values seems to suggest that during the annealing process a chemical reaction between the metallic substrate and the oxide film a n d / o r a loss of oxygen can occur. The final result of these processes could be the increase of oxygen vacancies, Vo, and their migration into the films, in agreement with the following reactions: T~350°C

xW + WO 3

~

(1 + x ) W O 3 + 3xVo(oxide )

(1)

at the metal/oxide interface, and T = 350°C

WO3

~

WO3_, + ½ x O :

at the oxide/gas interface.

(2)

425

F. Di Quarto et a L / WOe photoanodes l

I

r

I 0.5

~ 1.0

I 1.5

I

1.0

0'7 ~

0.50

0.2E

0.0 0.0

U,

V

~

I 2.0

2.5

MSE

Fig. 4. Iph VS. U E plots at A = 300 nm. The experiments were performed on the same electrode of fig. 3, before (solid curve) and after (dashed curve) a photoelectrolysis of 18 h at UE = 2 V and A = 300 nm.

According to Berak and Sienko [2,11] higher donor concentrations can be related to more defective WO3 phases. It must be stressed that, as a result of the crystallization, meaningful changes were observed in the chemical reactivity of the anodic barrier films. In fact, well crystallized WO 3 barrier films showed very good stability in acidic solutions both in the dark and under illumination. Moreover, a sharp change in the dissolution rate after film crystallization was observed by immersing the electrodes in 1 N N a O H solution. In this last solution amorphous barrier films dissolved instantaneously, whilst crystallized films showed a much lower dissolution rate. The increase in the donor concentration suggests that, during the annealing, some non-stoichiometric phase is developed at the surface of the films, so lowering their chemical reactivity. For tungsten oxides single crystals Sienko and Berak [11] found that a donor concentration of about 10 20 cm -3 corresponded to a sample stoichiometry ranging between WO2.987 and WO2.995. Moreover, from their work it comes out that identical X-ray diffraction patterns could be assigned to WO 3 as well as to WO2.987 samples. In fig. 4 are reported typical photocurrent vs. electrode potential curves for a crystallized barrier-type anodic film, before and after prolonged photoelectrolysis performed at constant potential and under monochromatic irradiation. In comparison with the photoresponse exhibited by the amorphous films [8], both photocharacteristics show meaningful differences, as for the overall efficiency of light conversion as well as for the dependence of iph upon U z. In particular, for a-WO 3 films a straight-line behaviour was found in the plots of .2 .4/3 lph against U E at ~ >/270 nm, whilst a linear dependence was observed in the tph vs. U z plots at ~ = 230 nm. In that work it was suggested that the mechanisms of transport of photogenerated carriers in the extended states region (~ = 230 nm) or in

426

F. Di Quarto et aL / WO¢ photoanodes 4

i

r

I

I

I

o J

%

3

oJ jo

oJ

D/

oJ

J

2

/°

1

0.0

0.5

1.0 U~

V ~

1.5

2.0

2.5

MSE

Fig. 5. Iph vs. U E plots obtained at different wavelengths. The experiments wcre performcd on the same electrode of fig. 3. Electrode surface: 0.053 cm2. rn: X = 300 nm; zx: ), = 380 nm.

the localized states (?~ >/270 nm) could explain the different dependences o f iph upon U E observed at different wavelengths. For crystallized films different mechanisms must be invoked in order to explain the iph VS. U E plots. For a crystalline semiconductor under constant illumination, according to several authors [21-28], a quite general but simplified relationship can be used to relate the measured photocurrent to the electrode potential. By taking into account all possible kinetic controls, it is possible to write: iph= e ~ P 0 S t - - ~

1

1+

Lh e x p ( - ~ x ~ ° U~2E- UF~) '

(3)

where q% is the photon flux at the semiconductor surface after correction for the reflection losses, S t is the rate of holes reaction at the semiconductor/electrolyte interface, S r is the overall recombination rate. The term within brackets is, essentially, the G~rtner's term [21], which takes into account the supply of minority carriers coming from the bulk of semiconductor (diffusion term) and from the space-charge region (migration term). In eq. (3) ~ is the light absorption coefficient, 0 the width of the space-charge region at U E - - UFB= 1 V, UFB the flat band X Sc potential and L h the diffusion length of holes. The limits and the usefulness of a photoelectrochemical method in determining the UFB value have been pointed out by several authors [29]. The best way to get reliable UFB values from photoelectrochemical measurements is to extrapolate to zero photocurrent theoretical relationships between iph and U E, from the high electrode potential region. In fact for high U E values the ratio S t / ( S t + S r) can be assumed to be constant and equal to unity [25-28]. The use of the photocurrent onset potential as UFB value may be uncorrect if high recombination rate of minority carriers occurs at the photoelectrode/electrolyte

F. Di Quarto et al. / WO3 photoanodes

427

interface. On the other hand, Albery et al. [30] have shown that a dependence of iph upon the square root of the electrode potential could indicate that the controlling kinetic factor is the recombination of minority carriers within the space-charge region. 2 vs. UE plots at According to the previous considerations, in fig. 5 are reported Iph different wavelengths for the crystallized film of fig. 4 before photoelectrolysis. Two straightline behaviours can be observed at each wavelength at low (U E ~< 1.2 V) and high (U E > 1.2 V) electrode potentials, respectively. Moreover, the straight lines obtained in the high-potential region converge to the same point of the potential axis, which should be a measure of the UFB value. This value of electrode potential is in very good agreement with the onset photocurrent potential (see fig. 4), but it is slightly more positive than the value obtained from the Mott-Schottky plots (see fig. 3). On the other hand, the extrapolation of the straight lines obtained in the low potentials region gives different intersection values which are more positive than the onset photocurrent potential. All these findings can be qualitatively explained by looking back at the different terms of eq. (3). In fact, according to several theoretical treatments of the electrolyte/semiconductor junctions under illumination, at sufficiently high band bending the behaviour of the iph VS. UE curves is essentially determined by the G~rtner term of eq. (3) [24,26,27,29]. Both the linear dependence of the /ph .2 VS. UE curves and the common intersection voltage suggest that in the high electrode potential region the migrative term gives a much higher contribution with respect to the diffusive one. Moreover, the extrapolated potential value is in fairly good agreement with the previously reported UFB values of WO 3 electrodes [7]. The linear region of tph -2 VS. UE plots at lower potentials seems to be related with a kinetic control due to the recombination of minority carriers in the depletion layer

[301. It is interesting, moreover, to compare the iph VS. UE plots obtained before and after prolonged photoelectrolysis under constant potential and monochromatic irradiation (see fig. 4). The modification of the photocharacteristics seems to occur in the direction of an increase of the diffusive term with respect to the migration term. This interpretation would imply that under illumination an increase of donor concentration occurs so that the width of the space-charge region decreases in presence of a practically constant film thickness. The constancy of the film thickness was inferred on the basis of the constancy of the interference colour. This aspect will be further developed in the following. It must be stressed that after the annealing process a very large increase was always observed in the stability of the electrodes against the photocorrosion.

5. Porous anodic films

In fig. 6 are reported the photocharacteristics of an electrode annealed after anodization in H N O 3 solution until the breakdown voltage, before and after

428

[ : Di Quarto et al. / 1.5

i

w o s photoanodes

i

i

i

a

? E u

i

1.0

0.5

b

0.1

0.5

1.0 U~

,

V

1.5 ,,~

2.0

MSE

Fig. 6. iph VS. UL plots at 2~= 3~0 nm for a film grown in 1 N HNO~ solution for 10 min and annealed for 3 h at 350°C under argon atmosphere. The photoresponses were measured before (curve a) and after (curve b) a photoelectrolysis of 46 h at U E = 2 V and ~ = 380 n m

prolonged photoelectrolysis, The X-ray diffraction patterns of this kind of layers b e f o r e t h e a n n e a l i n g p r o c e s s a r e s h o w n in fig. l c . D u e t o t h e c o m p l e x m o r p h o l o g y of the layers (an external thick porous triclinic layer on an underlying monoclinic b a r r i e r film [16,18]), t h e c a p a c i t a n c e m e a s u r e m e n t s a r e n o t u s e f u l to c h a r a c t e r i z e t h e s e films. A s r e p o r t e d in fig. 6a, l a r g e r p h o t o c u r r e n t s w e r e o b s e r v e d d u e t o t h e l a r g e r a b s o r p t i o n o f light. A s h a r p d e c r e a s e in t h e m e a s u r e d p h o t o c u r r e n t a n d a '

'

'

'

b 0.3

1.8

'?,

i

] 0.1

0.0

I, 300

400

~ 300

500

,

r 400

, "-..-JO.O 500

nm

Fig. 7. Photocurrent action spectra of triclinic WO 3 films grown in 1 N HNO3 solution. (a) before annealing; (b) after 3 h of annealing at 350°C under argon atmosphere; (c) after annealing and 46 h of photoelectrolysis at UE = 2 V and ~ = 380 nm.

F. Di Quarto et al. / WQ¢ photoanodes

429

relative increase of the diffusion term is still observed in the higher electrode potential region after photoelectrolysis (fig. 6b). Fig. 7 shows the photocurrent spectra obtained soon after the anodization process as well as after crystallization and long-term photoelectrolysis. By assuming indirect optical transitions, the plots o f (iphht.') °5 VS. h v give always the same optical band-gap value (Eg°pt = (2.55 _+ 0.05) eV) regardless of the treatments (see table 1). After the photoelectrolysis experiment no relevant changes of the electrode surface were visible by electron microscopy and, moreover, the film thickness was apparently unchanged. Large modifications were, instead, observed in the photocurrent action spectra after different treatments. As previously reported [16], the main changes in the spectra before and after annealing are related to the disappearance of the amorphous barrier film, underlying the thick polycrystalline layer formed by prolonged anodization above the breakdown voltage. It was suggested that the external crystalline layer determines the optical absorption edge, whilst the underlying amorphous film controls the transport of the photogenerated carriers. After crystallization the photoelectrochemical behaviour of the film changes as a consequence of the large increase in the mobility of photocarriers in the underlying barrier film. After prolonged photoelectrolysis at constant potential and under monochromatic irradiation further changes in the composition of the triclinic external layer can be suggested in order to explain the change both in the spectrum and in the photocharacteristic. As a possible explanation we suggest the removal of oxygen from the oxide surface under illumination and the formation of crystallographic shear planes. According to this hypothesis, under illumination the following reactions occur at the electrode surface with photons having energy higher than F °pt" --g

-

WO3 + h v ~ WO 3 + h + + e - ,

(4a)

WO 3 + 2xh + + x H 2 0

(4b)

--~ WO3_ x + 2 x H + + xO 2.

This last reaction of surface defect formation competes with the reactions of water photoelectrolysis and photocorrosion: 2h + +

H20

~

2H + + 0.5 0 2,

2h + + WO 3(surface) + solv ~ WO22+(sol) + 0.5 02.

(5) (6)

Eqs. (4b) and (6) are analogous, but they differ for the final product which is a defective structure in the former case, and a solvated cation in the latter one. It is very interesting to observe that eq. (6) is favoured if the WO 3 electrode is in the amorphous state, whilst eq. (4b) is favoured for crystalline WO 3 layers * The hypothesis of formation of shear planes can be supported by the following considerations:

* However, in the photoelectrolysis experiment on crystalline layers, it was estimated that more than 90% of the circulated charge was spent for oxygen evolution according to eq. (5).

F. Di Quarto et al. / WO3 photoanodes

430 I

1

1200

a c

800 E

¢J

t¢

2

b 400

o

300

300

400 I

400

500

,nm

Fig. 8. Photocurrent action spectra of WO 3. H 2 0 anodic films grown in 1 N H2SO 4 solution at 70°C. (a) before annealing: (b) after 3 h of annealing: (c) after 17 h of annealing.

(a) in WO3_, oxides the existence of the so called "Magn61i shear planes" which originate from the removal of oxygen from stoichiometric WO 3 regions is widely accepted [2]; (b) it has been shown theoretically [31] that the release of oxygen from the surface of oxides having a crystallographic structure analogous to WO 3 oxides can be "more facile when accompanied by the structural rearrangement resulting in crystallographic shear"; (c) in WO 3_x oxides, Berak and Sienko [2,11] reported an increase of free carriers with increasing oxygen vacancies, together with a drop in the carriers mobility "as the number of oxygen vacancies becomes large enough to form Magn61i shear planes". All these findings can help explain the changes observed both in the iph VS. UE plot and in the photocurrent spectrum of WO 3 electrodes after prolonged photoelectrolysis. In fact, the shape of the iph VS. U E curve (see curve b of fig. 6) should indicate that the diffusive term is increased with respect to the migration term as a result of the increased contribution of the photocarriers coming from the bulk of the semiconductor. A reduction of the space-charge region due to the higher N D values is in agreement with the above mentioned findings. However, this increment of the diffusive component occurs contemporarily to the mobility drop of the photocarriers and this causes a substantial decrease in the total measured photocurrent. The modification of the spectrum at X < 400 nm (see figs. 7b and c) could be ascribed to an increase of the electronic states density associated with these defective regions, where a higher electron-hole recombination rate occurs. It is interesting to compare our results with those reported by Ulman and Augustynski [12], relatively to the effect of a prolonged photoelectrolysis on the photocurrent spectra and photocharacteristics. As a consequence of the photoelec-

F. Di Quarto et al. / WO 3 photoanodes

7~

f

I

431

I

I

I 1.5

I 2.0

b

500

25¢

o

o.o

I 1.0

o.5

u~

V

,,

MSE

Fig. 9. iph vs. U E plots at ~, = 380 nm of the films of fig. 8 after different annealing times. (a) after 3 h; (b) after 17 h.

trolysis at high c.d. ( i p h >~ 10 m A cm-2), these authors found an improvement of the photocurrent efficiency and a modification of the shape of the spectrum different from that reported in fig. 7. This finding is quite understandable taking into account that at high illumination levels the rate of oxygen evolution according to eq. (5) can suppress or invert reactions (4b). In fig. 8a is shown the photocurrent action spectrum of a film grown in 1 N H2SO 4 at 70°C for 1 h, which has been identified (see fig. ld) as orthorhombic WO 3 • H20. After annealing a dehydration of the layer was observed with formation of triclinic WO 3 (see table 1). Figs. 8b and c show the influence of the annealing time on the shape of the spectra, whilst fig. 9 shows the effect of the annealing time on the photocharacteristics of the electrodes. From figs. 8 and 9 it comes out that longer annealing time have favourable effect on the overall efficiency of light conversion, specially in the long wavelength region. The large increase in the measured photocurrent at ?~-- 380 nm seems attributable once more to the formation of non-stoichiometric WO 3 regions. However, a better crystallization and a not too high defect concentration conspire in favouring higher overall efficiency in specimens annealed for a longer time. From a practical point of view, better efficiency was usually found for specimens anodized in H N O 3 solution at room temperature rather than with electrodes prepared at 70°C. This fact can be explained by taking into account the different morphology of films formed in different experimental conditions [18]. The more spongy structure of WO 3 • H20, which remains unchanged after annealing, could result in a layer with ohmic losses higher than the more compact WO 3 layers formed in H N O 3 solution at room temperature.

F. Di Quarto et al. / WO~ photoanodes

432

6. Conclusions It has been shown that different photoelectrochemical responses can be obtained from WO 3 electrodes prepared in different ways. This fact is essentially related to the formation of an amorphous barrier film or a crystalline layer. The favourable effect of the annealing at 350°C under argon atmosphere is essentially due to the crystallization of the amorphous barrier film. Due to crystallization a change in the optical band gap is observed for amorphous barrier films, from about 3 to about 2.7 eV. A decrease in the --g F °pt value seems to occur also for the transformation: WO 3 • H:O(orthorhombic) --, WO3(triclinic ). Lower --g F~°pt values (2.55 eV) were measured for thicker triclinic WO 3 electrodes prepared in H N O 3 solutions at room temperature. Large changes both in the photocharacteristics and in the photocurrent spectra were observed after prolonged photoelectrolysis. The formation of defective WO 3 ., regions, with the appearance of Magn61i shear planes was suggested in order to explain such modifications. The experimental findings have indicated that the annealing time and the anodization procedure are the most important parameters which govern the photoelectrochemical response of WO 3 photoanodes.

References [1] [2] [31 [4] [5] [6] [7] [8]

[9] [10] [11] [12]

[13] [14] [15] [16] [171 [18] [19]

S. Sawada, J. Phys. Soc. Japan 11 (1956) 1237. J.M. Berak and M.J. Sienko, J. Solid State Chem. 2 (1970) 109. S.K. Deb, Phil. Mag. 27 (1973) 801. Z.M. Hanafi and M.A. Khilla, Z. Phys. Chem. N.F. 89 (1974) 230. M.A. Butler, J. Appl. Phys. 48 (1977) 1924. A. Nakamura and S. Yamada, Appl. Phys. 24 (1981) 55. F. Di Quarto, A. Di Paola and C. Sunseri, Electrochim. Acta 26 (1981) 1177. F. Di Quarto, G. Russo, C. Sunseri and A. Di Paola, J. Chem. Soc. Faraday Trans. I 78 (1982) 3433; F. Di Quarto, S. Piazza, G. Russo and C. Sunseri, Extended Abstracts 33th ISE Meeting, Lyon (1982) p. 843. J.R. Darwent and A. Mills, J. Chem. Soc. Faraday Trans. I1 78 (1982) 359. J. Kiwi and M. Gr~tzel, ibid. 78 (1982) 931. M.J. Sienko and J.M. Berak, The Chemistry of Extended Defects in Non-metallic Solids, eds. g. Eyring and M. O'Keeffe (North-Holland, Amsterdam, 1970) p. 541 M. Ulmann and J. Augustynski, in: Photoelectrochemistry: Fundamental Processes and Measurements Techniques, eds. W.L. Wallace, A.J. Nozik, S.K. Deb and R.H. Wilson (The Electrochemical Society, New York, 1982) p. 663; J. Appl. Phys. 54 (1983) 6064. G. Hodes, D. Cahen and J. Manassen, Nature 260 (1976) 312. K.L. Hardee and A.J. Bard, J. Electrochem. Soc. 124 (1977) 215. W. Gissler and R. Memming, ibid. 124 (1977) 1710. F. Di Quarto, S. Piazza and C. Sunseri, in: Passivity of Metals and Semiconductors, ed. M. Froment (Elsevier Science Publishers, Amsterdam, 1983) p. 497. F. Di Quarto, A. Di Paola and C. Sunseri, J. Electrochem. Soc. 127 (1980) 1016. A. Di Paola, F. Di Quarto and C. Sunseri, Corrosion Sci. 20 (1980) 1067, 1079. N.F. Mort and E.A. Davis, Electronic Processes in Non-crystalline Materials (Clarendon Press, Oxford, 1979).

F. Di Quarto et al. / WO 3 photoanodes

[20] [21] [22] [23] [24] [25] [26] [27] [28] [29]

433

A. Mansingh, M. Sayer and J.B. Webb, J. Non-Crystalline Solids 28 (1978) 123. W.W. Gartner, Phys. Rev. 116 (1959) 84. M.A. Butler, J. Appl. Phys. 48 (1977) 1914. R.H. Wilson, ibid. 48 (1977) 4292. H. Reiss, J. Electrochem. Soc. 125 (1978) 937. J. Reichmann, Appl. Phys. Lett. 36 (1980) 574. F. El Guibaly and K. Colbow, J. Appl. Phys. 53 (1982) 1737. R.U.E. 't Lam and D.R. Franceschetti, Mater. Res. Bull. 17 (1982) 1081. S.K. Kovach and A.T. Vas'ko, Electrokhimiya 19 (1983) 252. Faraday Discussions of the Chemical Society no. 70: Photoelectrochemistry (The Royal Society of Chemistry, London, 1980). [30] W.J. Albery, P.N. Barlett, A. Hamnett and M.P. Dare-Edwards, J. Electrochem. Soc. 128 (1981) 1492. [31] E. Broclawik and J. Haber, J. Catalysis 72 (1981) 379.