Effects of electrodeposition synthesis parameters on the photoactivity of nanostructured tungsten trioxide thin films: Optimisation study using response surface methodology

ARTICLE IN PRESS

JID: JTICE

[m5G;January 8, 2016;7:0]

Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–9

Contents lists available at ScienceDirect

Journal of the Taiwan Institute of Chemical Engineers journal homepage: www.elsevier.com/locate/jtice

Effects of electrodeposition synthesis parameters on the photoactivity of nanostructured tungsten trioxide thin films: Optimisation study using response surface methodology Tao Zhu a, Meng Nan Chong a,b,∗, Yi Wen Phuan a, Joey D. Ocon c, Eng Seng Chan a a School of Engineering, Chemical Engineering Discipline, Monash University Malaysia, Jalan Lagoon Selatan, Bandar Sunway, Selangor Darul Ehsan 46150 Malaysia b Sustainable Water Alliance, Advanced Engineering Platform, Monash University Malaysia, Jalan Lagoon Selatan, Bandar Sunway, Selangor Darul Ehsan 46150 Malaysia c Laboratory of Electrochemical Engineering (LEE), Department of Chemical Engineering, University of the Philippines Diliman, Quezon City 1101 Philippines

a r t i c l e

i n f o

Article history: Received 6 June 2015 Revised 3 December 2015 Accepted 9 December 2015 Available online xxx Keywords: Tungsten trioxide Photoelectrocatalysis Electrodeposition Response surface methodology Box–Behnken Photoelectrochemical water splitting

a b s t r a c t The main aim of this study was to synthesize and characterise nanostructured tungsten trioxide (WO3 ) thin films via electrodeposition and subsequently, optimise the electrodeposition synthesis parameters using response surface methodology (RSM). Statistical Box–Behnken RSM design was used to investigate and optimise the effects of four independent electrodeposition synthesis parameters, namely: deposition time, precursor tungsten (W) concentration, annealing temperature and pH. In addition, the synergistic interaction between different electrodeposition synthesis parameters was identified and quantified in enabling a higher photoactivity achievable by nanostructured WO3 thin films. Resultant nanostructured WO3 thin films were characterised using field-emission scanning electron microscopy (FE-SEM), X-ray diffraction (XRD) and photocurrent density measurements under one-Sun irradiation. From the electrodeposition synthesis process, it was found that there was a gradual increase in the nanocrystallites WO3 size from 30 nm to 70 nm when the annealing temperature was varied between 400 °C and 600 °C. XRD results verified the existence of the same photoactive phase of monoclinic WO3 with increasing annealing temperature with the preferred growth orientation along the {002} planar. Whilst from the Box–Behnken RSM design, it was found that the optimum deposition time, precursor W concentration, annealing temperature and pH were: 60 min, 0.15 mol/L, 600 ◦ C, and pH 1.0, respectively. The highest photocurrent density of 120 μA/cm2 was measured at 1 V (versus Ag/AgCl) for nanostructured WO3 thin film synthesized at the optimum conditions as informed by the Box–Behnken RSM. Further analysis and validation of the Box–Behnken RSM model using analysis of variance (ANOVA) revealed that the RSM-derived statistical predictive model was robust, adequate and representative to correlate the various electrodeposition synthesis parameters to photocurrent density. This study highlights the importance to systematically optimise the electrodeposition synthesis parameters in order to achieve a higher photocurrent density on nanostructured WO3 thin film for sustainable hydrogen production from photoelectrochemical water splitting reaction. © 2015 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction Photoelectrochemical (PEC) water splitting is attractive owing to its potential for solar energy conversion without the need of energy derived from declining fossil fuel supplies [1]. To date, much of the researches being carried out are focused on solar energy conversion

∗ Corresponding author at: School of Engineering, Chemical Engineering Discipline, Monash University Malaysia, Jalan Lagoon Selatan, Bandar Sunway, Selangor Darul Ehsan 46150 Malaysia, Tel.: +60 3 5514 5680; fax: +60 3 5514 6207. E-mail address: [email protected], [email protected] (M.N. Chong).

through PEC water splitting, which converts solar into renewable energy resource in the form of storable hydrogen energy [2]. During the PEC water splitting process, water molecule is photolysed into hydrogen (H2 ) and oxygen (O2 ) via the aid of semiconductor metaloxide photocatalyst. Thus, the presence of semiconductor photocatalyst plays an important role in the solar-to-hydrogen energy conversion during the PEC water splitting process. Among the common photocatalysts used for PEC water splitting process, tungsten trioxide (WO3 ) is increasingly used as a semiconductor photocatalyst for oxidative decomposition of water owing to its smaller bandgap energy of 2.6 eV [3]. Moreover, WO3 has received a great deal of attention due to its good resistance against photo-corrosion, high stability

http://dx.doi.org/10.1016/j.jtice.2015.12.010 1876-1070/© 2015 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Please cite this article as: T. Zhu et al., Effects of electrodeposition synthesis parameters on the photoactivity of nanostructured tungsten trioxide thin films: Optimisation study using response surface methodology, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2015.12.010

JID: JTICE 2

ARTICLE IN PRESS

[m5G;January 8, 2016;7:0]

T. Zhu et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–9

in acid solution and the extended absorption into visible-light spectrum that rendered a better photoactivity. In the last 2 decades, WO3 has been extensively studied from various fundamental science perspectives as well as a wide application in PEC water splitting process [4]. From the open literatures, various synthesis methods have been reported for the fabrication of nanostructured WO3 thin films with bespoke physicochemical, optical, electronic and PEC properties, as well as improved photoactivity [5]. These include sputtering, thermal evaporation, chemical vapour deposition, sol-gel, electrodeposition and hydrothermal synthesis methods [6]. Most of these synthesis methods, however, are not suitable for scale-up processing and commercialisation [7]. Recently, the electrodeposition synthesis method has received much attention due to the advantages of low capital cost, ambient temperature and pressure conditions, direct control of film thickness and the possibility of scale-up processing and commercialisation [8]. During the electrodeposition process, different electrodeposition synthesis parameters such as deposition time, precursor W concentration, annealing temperature and pH are known to have dominant and synergistic effects on the eventual photoactivity of nanostructured WO3 thin films, which were reviewed in our previous communication [7]. For instances, varying the deposition time can lead to different film thicknesses while a more stable crystal structure will be formed via heat treatment under different annealing temperature. Similarly, it is also known that both pH and precursor W concentration will have significant effects on the morphology and particle size. Typically, in order to optimise the electrodeposition synthesis conditions for nanostructured WO3 thin films, the experimentalbased one-factor-at-a-time (OFAT) optimisation approach could be employed to verify the different optimum synthesis parameters. The OFAT approach has been traditionally used to achieve higher efficiencies by varying one independent experimental factor or parameter at a time while keeping the other independent factors constant. However, the major disadvantage of the OFAT approach is that it cannot depict the synergistic and interactive effects among the electrodeposition synthesis parameters and verify the optimum synthesis parameters without vast experimentation efforts [9]. Recently, the application of response surface methodology (RSM) is gaining immense attention as a robust statistical technique used for experiments design, predictive model development and evaluation of synthesis parameters and their interactions, as well as optimisation to yield the desirable response surfaces [10]. Generally, there are a number of RSM types that are widely used in the materials synthesis experiment design and optimisation, including: D-optimal [11], Central Composite Design (CCD) [12] Box–Behnken Design (BBD) [12] and others [13]. When compared among these RSM types, BBD statistical design is an independent, rotatable or nearly rotatable, quadratic design. The design combinations are placed at the midpoints of the edges and at the centre of the process space. This design requires less experimental runs and time to optimise parameters, as well as able to predict the optimal conditions to obtain high quality results from the experiments performed [11,12]. Thus, the main aim of this study was to synthesize and characterise nanostructured WO3 thin films via electrodeposition and subsequently, optimise the electrodeposition synthesis parameters by using RSM. Previously, we had investigated the effect of heat treatment (i.e., annealing temperature) through the OFAT approach on nanostructured WO3 thin films and found that the highest photocurrent of 35 μA/cm2 was achievable at 600 °C. This study constitutes the foremost study to systematically investigate and optimise the synthesis parameters for nanostructured WO3 thin films via the electrodeposition synthesis route. Statistical Box–Behnken RSM design was used to investigate and optimise the effects of four independent synthesis parameters, namely: deposition time (X1 ), precursor W concentration (X2 ), annealing temperature (X3 ) and pH (X4 ). Resul-

tant nanostructured WO3 thin films were characterised using fieldemission scanning electron microscopy (FE-SEM), X-ray diffraction (XRD) and photocurrent density measurements under one-Sun irradiation. For the RSM design, the measured photocurrent densities were assigned as the response outputs for the developed and validated Box–Behnken model. This study highlights the importance to systematically optimise the electrodeposition synthesis parameters in order to achieve a higher photocurrent density and thus, the photoactivity for nanostructured WO3 thin films. 2. Materials and methods 2.1. Preparation of tungsten precursor solution In this study, all the chemicals were used as received without further purification. Hydrogen peroxide 30% (H2 O2 ) was obtained from HmbG Chemicals, USA. Tungsten (W) powder with particle size of 325 meshes was purchased from Chem Soln, USA. Platinum (Pt) black (≥ 99.97%) with particle size ≤ 20 μm was also supplied by ChemSoln, USA. All other miscellaneous chemicals were purchased from Merck, USA. Initially, the precursor solution was prepared by dissolving varying quantity of W powder in 50 mL of H2 O2 and the reaction was allowed to continue up to 24 h. Thereafter, the excess H2 O2 was decomposed by adding small amount of Pt black. The solution was further heated at 60 ◦ C until no effervescence was evident. Then, the precursor solution was diluted to 50 mM via the addition of 150 mL of 50/50 (v/v) water/2-propanol. The function of propanol-2-ol was to extend the stability of precursor solution by preventing the precipitation of amorphous WO3 -based hydrated phase [13]. 2.2. Synthesis of nanostructured WO3 thin films The electrodeposition synthesis of nanostructured WO3 thin films was performed at room temperature using a conventional three-electrode electrochemical cell system (Metrohm, Netherland). Fluorine-doped tin oxide (F-SnO2 ) glass slide (Chem Soln, USA; 2.5 cm × 2.0 cm size) was used as the working electrode (WE) after being cleaned with acetone and water; while a Pt rod was used as the counter electrode (CE) and Ag/AgCl (4 M KCl) as the reference electrode (RE). All the potentials used in the experiments were made reference to the Ag/AgCl (4 M KCl). During the electrodeposition synthesis, the effective immersion area of F-SnO2 was fixed constant at 1.5 cm × 2.0 cm. The applied potential between WE and RE was – 0.45 V controlled by the Autolab potentiostat/galvanostat (Metrohm, Switzerland). F-SnO2 was rinsed using distilled water, followed by drying using clean air for 20 min (i.e., heating and cooling rates of 10.0 ◦ C/min and 2.5 ◦ C/min, respectively). After the electrodeposition process, the as-deposited WO3 film was removed from the suspension and annealed at 400–600 ◦ C for 1 h to facilitate the phase transformation of amorphous WO3 into nanocrystalline WO3 , as well as to enhance the adhesion strength between the nanostructured WO3 thin film and F-SnO2 electrode. 2.3. Characterisation of nanostructured WO3 thin films X-ray diffraction (XRD) was used to determine the in-situ phase composition of nanostructured WO3 thin films (Philips X’pert Materials Powder Diffractometer; Cu Kα radiation; 45 kV; 40 mA). While the microstructure and chemical analysis was examined by using fieldemission scanning electron microscopy (FESEM; FEI Nova NanoSEM; uncoated samples; secondary electron emission; accelerating voltage 5 kV). The PEC properties of nanostructured WO3 thin films were measured in a dark-controlled and ambient temperature condition using the same Autolab potentiostat/galvanostat system. The only difference was that the peroxy-tungstic acid (PTA) electrolyte solution was replaced by sodium acetate (CH3 COONa) aqueous

Please cite this article as: T. Zhu et al., Effects of electrodeposition synthesis parameters on the photoactivity of nanostructured tungsten trioxide thin films: Optimisation study using response surface methodology, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2015.12.010

ARTICLE IN PRESS

JID: JTICE

[m5G;January 8, 2016;7:0]

T. Zhu et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–9 Table 1 Experimental levels for each independent parameter based on Box– Behnken RSM design. Variables

Symbol

–1

0

1

Deposition time (min) Precursor concentration (mol/L) Annealing temperature (◦ C) pH

X1 X2 X3 X4

30 0.05 400 0.5

60 0.1 500 1.0

90 0.15 600 1.5

Table 2 Box–Behnken RSM designed experiments along with the experimental and predicted photocurrent densities. Run

solution (0.1 mol/L) for the electrodeposition synthesis process and measurement of PEC properties, respectively. During the measurement of PEC properties, illumination from a halogen lamp restricted at a frequency of 0.05 Hz was used as the light source. The intensity of incident light was measured using a digital light meter and was found to be 100 mW/m2 (i.e., one-Sun equivalent) when the light source-tosample distance was set at 10 cm, while the linear potentiodynamic voltammetry was applied at a scan rate of 5 mV/s for the measurement of PEC properties. 2.4. Experimental design and statistical analysis The Box–Behnken statistical experiment design was used to investigate the effects of four independent electrodeposition synthesis parameters on response outputs (i.e., photocurrent densities) based on 29 set of experiments. The four independent parameters were deposition time (X1 ), precursor W concentration (X2 ), annealing temperature (X3 ) and pH (X4 ). The low, centre and high levels of each parameter were designated as –1, 0, and +1, respectively as illustrated in Table 1. The experimental levels selected for every parameter were based on the previous studies [7]. For instance, Yang et al. [14] found that the optimum pH range for the synthesis of mesoporous WO3 films lies between pH 0.8 and 1.1. In this study, the dependent parameter (or response output) set for the Box–Behnken RSM design was the measured photocurrent density. The four independent parameters together with the dependent parameter were fitted to an empirical quadratic polynomial predictive model as shown in Eq. (1):

Y = β0 +



βi Xi +



βi j Xi X j +



βii Xi2

(1)

where Y is the dependent (or response output); Xi (i = 1,2,3,4) are the controlling independent parameters; and β 0 , β i (i = 1,2,3,4) and β ij (i = 1,2,3,4; j = 1,2,3,4) are the model coefficient parameters. The multivariate regression analysis and optimisation of response surfaces were performed using DesignExpert® software (Version 8.0.6 StatEase, Inc., Minneapolis, USA). Analysis of variance (ANOVA) of the experimental data was performed. 3. Results and discussion 3.1. Model fitting and analysis of variance (ANOVA) The experimental results obtained from Box–Behnken RSM design under varying electrodeposition synthesis conditions were analysed by using the DesignExpert® software. Table 2 shows the Box–Behnken RSM designed experiments and the experimental and predicted results (photocurrent density). From Table 2, the Box–Behnken experimental outcomes were fitted using an empirical second-order polynomial equation as shown in Eq. (2):

Y = +62.20 − 2.33X1 + 5.5X2 + 43.33X3 − 3.5X4 − 3.50X1 X2 −2.5X1 X3 + 1.0X1 X4 + 7.00X2 X3 + 1.0X2 X4 − 1.0X3 X4 −11.27X12 − 12.27X22 + 6.98X32 − 31.52X42

3

(2)

where Y is the photocurrent density (μA/cm2 ); X1 , X2 , X3 and X4 are the coded parameters for deposition time, precursor concentration, annealing temperature and pH, respectively.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

X1

1.00 0.00 0.00 0.00 1.00 –1.00 –1.00 0.00 –1.00 0.00 1.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 –1.00 –1.00 0.00 0.00 –1.00

X2

0.00 –1.00 0.00 1.00 0.00 0.00 1.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 1.00 –1.00 –1.00 0.00 1.00 0.00 0.00 –1.00 1.00 –1.00 0.00 0.00 –1.00 0.00

X3

X4

0.00 0.00 –1.00 0.00 0.00 –1.00 0.00 0.00 0.00 1.00 1.00 1.00 1.00 –1.00 0.00 0.00 1.00 0.00 0.00 –1.00 0.00 –1.00 0.00 0.00 0.00 0.00 0.00 –1.00 1.00

1.00 –1.00 1.00 –1.00 –1.00 0.00 0.00 0.00 1.00 1.00 0.00 –1.00 1.00 0.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 –1.00 0.00 1.00 0.00 –1.00 0.00 0.00 0.00

Photocurrent density (μA/cm2 ) Experimental

Predicted

35 12 70 11 57 18 20 76 69 110 96 55 10 65 56 38 10 19 25 15 9 10 89 87 20 39 68 35 10

33 10 75 10 55 13 15 80 65 108 106 59 5 63 60 43 6 17 23 12 7 8 95 89 18 37 64 28 9

Table 3 ANOVA results for the coefficients of parameters of quadratic model for electrodeposition synthesis of nanostructured WO3 thin films using Box–Behnken RSM design. Parameter

Coefficient

Standard error

F-value

P-value

X1 X2 X3 X4 X1 X2 X1 X3 X1 X4 X2 X3 X2 X4 X3 X4 X1 2 X2 2 X3 2 X4 2

3.075E–004 –9.713E–003 9.992E–003 9.595E–003 3.518E–003 –1.697E–003 –2.237E–003 2.967E–003 1.460E–003 3.250E–005 –0.015 –1.736E–003 –2.735E–003 –3.819E–003

1.637E–003 1.056E–003 1.056E–003 1.056E–003 1.056E–003 1.830E–003 1.830E–003 1.830E–003 1.830E–003 1.830E–003 1.830E–003 1.437E–003 1.437E–003 1.437E–003

0.085 2.8 333.47 6.9 3.70 0.86 1.50 2.63 0.64 3.155E–004 102.71 1.46 3.62 7.06

0.2832 0.1163 < 0.0001 0.0198 0.0751 0.3693 0.2416 0.1271 0.4382 0.9861 < 0.0001 0.2470 0.0778 0.0187

Table 3 shows the coefficients of model parameters, significance of which was determined by both the F-value and P-value, respectively. According to Sahoo and Gupta [15], the smaller the P-value or the larger the F-value, the greater will be the contribution of corresponding model term toward the response output [16]. Thus, from this study, it was concluded that the significance of each parameter in electrodeposition synthesis process is: annealing temperature > pH > precursor W concentration > deposition time. ANOVA can be applied to determine the fitness of the proposed empirical quadratic model. From Table 3, the ANOVA results showed the significance of the proposed second-order equation model as evidenced from the high F-value of 89.47 and low P-value of < 0.0001. Generally if the P-value is less than 0.05, this indicates that the design model is significant and applicable to be used as statistical predictive model. Thus, the low P-value in our study indicated that the fitted second-order empirical model is valid. On the other hand, it was also

Please cite this article as: T. Zhu et al., Effects of electrodeposition synthesis parameters on the photoactivity of nanostructured tungsten trioxide thin films: Optimisation study using response surface methodology, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2015.12.010

ARTICLE IN PRESS

JID: JTICE 4

[m5G;January 8, 2016;7:0]

T. Zhu et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–9 Table 4 ANOVA results of the response surface quadratic model for photocurrent densities as the response outputs. Source

Sum of squares

Degree of freedom

Mean square

F-value

P-value

Model Residual Lack of fit Pure error Total

4.954E–003 1.875E–004 1.231E–004 6.433E–005 5.141E–003 R2 = 0.9635 Precision = 16.561

14 14 10 4 28 Adj R2 = 0.9271 CV = 6.23%

3.539E–004 1.339E–005 1.231E–005 1.611E–005 Predict R2 = 0.8426

26.42

< 0.0001

0.76

0.6689

found that there is only a low chance (i.e., 0.01%) of occurrence of the model due to noise as evidenced from the high F-value. Another parameter, coefficient of determination (R2 ), which is defined as the ratio of explained variation to the total variation and also a measure of the degree of fitness is shown in Table 4. In addition, the R2 -value is also the proportion of variability in the response output, which is accounted for by the regression analysis. In this instance, the closer the R2 -value to unity, the better empirical model fitting to the practical results can be concluded. Moreover, the smaller of the R2 -value indicated the less relevant of dependent parameter in the fitted second-order empirical model. In our study, the correlation coefficient (R2 ) value and the adjusted correlation coefficient (R2 adj) value of the fitted quadratic model are 0.9635 and 0.9271, respectively. Therefore, it can be concluded that the model predicted photocurrent densities are comparable to those measured experimentally owing to the close to unity for both R2 and R2 adjvalues. Fig. 1 shows the residual analysis, which is used to ensure the feasibility between the statistical assumptions and practical experiment data. From Fig. 1(a), the normal probability of residuals ensures that the standard deviations between predicted and experimental photocurrent densities are in a reasonable value range, indicating that there is no abnormal experimental result. The plot of residuals against predicted photocurrent densities by using the fitted quadratic model is shown in Fig. 1(b). From Fig. 1(b), it can be observed that the practical results are distributed randomly within the range of internally studentized residuals of ±3.00, implying that the fitted model is appropriate and the availability of variance assumption. Fig. 1(c) shows the comparison between the models predicted and experimental measured photocurrent densities, which shows that they are in good agreement. 3.2. Screening of parameters and response surface analysis Fig. 2 shows the three-dimensional (3-D) response surface plots that illustrate the effects of the four chosen independent electrodeposition synthesis parameters (i.e., deposition time, precursor W concentration, annealing temperature and pH) and their interactions on the response outputs (i.e., photocurrent densities) of nanostructured WO3 thin films. From Fig. 2 (a–c), it can be observed that the photocurrent densities increase proportionally when the deposition time increases. This positive effect may be attributed to the linear relationship between deposition time and film thickness of nanostructured WO3 thin films. Previously, Kwong et al. [13] also reported such a positive linear trend between the deposition time and film thickness. The observed trend might be owing to the time-dependent mass transfer, diffusion and growth of W ions on the F-SnO2 surfaces, as well as the increasing photoconversion due to the decrease in optical indirect bandgap that improved light absorption of the nanostructured WO3 thin films. Miller et al. [17] also discussed on the photocurrent density as a function of film thickness. They found that a further increase in the film thickness can lead to a reduction in photocurrent density owing to the increased film resistivity [18], excessive charge carriers, increased electron diffusion distance [19], increased charge

carriers equilibration time and increased volume of grain boundaries (i.e., as electron-hole recombination sites) [20]. As for the effect of precursor W concentration, a similar positive linear trend was observed between the precursor W concentration and photocurrent densities as shown in Fig. 2 (a–e). This observation can be explained by the following Eqs. (3) and (4): + 2W + 10H2 O2 → W2 O2− 11 + 2H + 9H2 O

(3)

+ − W2 O2− 11 + (2+x )H +xe → 2WO3 + (2+x )/2H2 O+ (8 − x )/4O2

(4) where W2 O11 2− is [(O2 )2 W(O)OW(O)(O2 )2 ]2− and (O2 ) denotes a peroxide ligand [7]. During the electrodeposition synthesis process, a higher precursor W concentration containing W2 O11 2− anions will result in a higher electrodeposition current density. This was due to the higher conductivity in precursor W solution as a result of higher W ions concentration in the electrolyte solution. Since identical F–SnO2 electrodes were used as the working electrodes, the higher initial deposition current in the electrolyte solution was found as a result of higher W ions concentration. Once the external electrical potential was applied, it was found that the deposition current density decreases rapidly to approximately halves of its initial value and reached a plateau at 1.65, 1.85 and 2.1 mA/cm2 , respectively. In this instance, the deposition current densities are due to the different precursor W concentrations used. In this instance, the rapid decrease in deposition current density prior to its stabilisation was largely a function of film thickness and area before the impingement and percolation processes take place. Both the nanostructured WO3 thin films and F-SnO2 electrodes would determine the baseline reduction in the deposition current density owing to their semi-conductive and conductive natures, respectively. The growth of semiconductor WO3 grain reduces the amount of exposed surface area of the conducting FTO electrode, which reduces the deposition current density. Fig. 2 (b,d,f) shows the effect of annealing temperature on photocurrent densities of nanostructured WO3 thin films. It was observed that the photocurrent densities increase with the increase in annealing temperature. Fig. 3 shows the FE-SEM images of nanocrystalline WO3 thin films annealed at 400–600 °C with the as-deposited amorphous WO3 film as the experimental control. From Fig. 3, it can be observed that the surface morphology of nanocrystalline WO3 films was significantly impacted by the annealing treatment. When the surface morphology of the as-deposited WO3 film was examined, it can be seen that the metal oxides WO3 display an irregular particle sizing that can be due to its amorphous nature. When the annealing temperature was increased up to 500 °C, a cracked film was evident that could be linked to the weakened adhesion and contact forces between WO3 particles. This is due to the drying stresses arising from surface dehydration under high temperature [21]. Previously, Rahman et al. [22] discussed that the presence of surface cracks might be beneficial in improving the photoactivity of nanostructured thin films since they allow the electrolytes to reach the F–SnO2 surfaces due to the increase in film/electrolyte interfacial area.

Please cite this article as: T. Zhu et al., Effects of electrodeposition synthesis parameters on the photoactivity of nanostructured tungsten trioxide thin films: Optimisation study using response surface methodology, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2015.12.010

JID: JTICE

ARTICLE IN PRESS

[m5G;January 8, 2016;7:0]

T. Zhu et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–9

5

Fig. 1. (a) Normal probability plot of residuals; (b) Plot of residuals against predicted responses, (c) Predicted against actual values plot for photocurrent densities (i.e., as response outputs).

In addition, the effect of annealing temperature on the phase structure and composition was also investigated by using the XRD method. Fig. 4(a) shows four different XRD spectra of nanostructured WO3 thin films, namely: as-deposited WO3 film and annealed WO3 films at 400 ◦ C, 500 °C and 600 ◦ C, respectively. From the XRD spectra, three distinct XRD peaks were observed at 23.1, 23.6 and 24.4° that correspond to the planar WO3 of {002}, {020} and {200}. In addition, the XRD analysis also confirmed the presence of photoactive monoclinic WO3 crystal structure for the annealed films at 400 °C, 500 °C and 600 °C but with different preferred orientation. Moreover, the existence of a strong XRD characteristic peak along the planar {002} after annealing treatment at 600 °C indicates a preferential orientation of WO3 nanocrystal growth in the {002} direction. This observation conformed to the previous finding as reported in the literatures [4,23,24]. Fig. 4(b) shows the dependence of photocurrent

density on the applied potential in 0.1 mol/L of aqueous CH3 COONa solution under controlled stepwise illumination. In this instance, the photocurrent density was increasing with the increase in applied potential owing to a higher band-bending at the photoanode-electrolyte interface. From Fig. 4, it can be concluded that the monoclinic WO3 phase annealed at 600 °C with preferred orientation of {002} generates a higher photocurrent density than the monoclinic WO3 phase annealed at 500 °C with a preferred orientation of {200}. In comparison, the nanostructured WO3 thin film annealed at 600 °C possesses closely uniform and homogeneously-shaped WO3 nucleus that attribute to a higher photocurrent density and thus, higher photoactivity when compared to the similar thin film annealed at 500 °C. Finally, Fig. 2 (c,e,f) examines the relationship between the pH of precursor W solution on photocurrent density during the electrodeposition synthesis of nanostructured WO3 thin films. It was found

Please cite this article as: T. Zhu et al., Effects of electrodeposition synthesis parameters on the photoactivity of nanostructured tungsten trioxide thin films: Optimisation study using response surface methodology, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2015.12.010

JID: JTICE 6

ARTICLE IN PRESS

[m5G;January 8, 2016;7:0]

T. Zhu et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–9

Fig. 2. Response surface plots illustrating the effects of (a) precursor W concentration and deposition time; (b) deposition time and annealing temperature; (c) deposition time and pH; (d) pH and W concentration; (e) pH and W concentration and; (f) pH and annealing temperature on the photocurrent densities.

Please cite this article as: T. Zhu et al., Effects of electrodeposition synthesis parameters on the photoactivity of nanostructured tungsten trioxide thin films: Optimisation study using response surface methodology, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2015.12.010

JID: JTICE

ARTICLE IN PRESS

[m5G;January 8, 2016;7:0]

T. Zhu et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–9

7

Fig. 3. FE-SEM images of nanocrystalline WO3 thin films annealed under different temperature at the precursor concentration 0.05 mol/L, pH 1.0 and deposition time of 90 min: (a) as-deposit; (b) 400 °C; (c) 500 °C; (d) 600 °C.

that the photocurrent density increases at a lower pH region (i.e., pH < X) while decreases at a higher pH region (i.e., pH < Y). This was in agreement with the finding by Yang et al. [14], where they found that the optimum pH range to prepare mesoporous WO3 films is between pH 0.8 and pH 1.1. Thermodynamically, WO3 (yellow coloured) was reported to be stable under acidic environment where tungstate cations may be formed in strong acidic electrolytes. Fig. 5 shows that each applied potential during the electrodeposition synthesis process represents a different region in the Pourbaix diagram. The Pourbaix diagram shows the potential at which both chemical and electrochemical reactions may occur on an electrode surface in a specific electrolyte as a function of pH [25]. In addition, the thermodynamic stability of chemical species at various potentials and pH regions can also be ascertained from the Pourbaix diagram. In general, W forms a wide range of oxides and the most stable oxide is WO3 . Other oxides formed on W are mostly the results of thermal oxidation of W or reduction of WO3 . According to the Pourbaix diagram, the surface of W at pH 2.0 should be passivated through the formation of chemical oxide species of WO2 , W2 O5 or WO3 . The fast nucleation and growth of WO3 at low pH region of pH 1.0 were indicated by the production of thicker and denser nanostructured WO3 film as compared to the higher pH region. Also the different pH of precursor W solution plays a very critical role in tailoring the surface morphology of the eventual nanostructured WO3 films formed. By increasing the acid concentration in the precursor W solution, the WO3 film was easily electrodeposited onto the F–SnO2 surfaces.

3.3. Optimisation and validation of the RSM-derived predictive model In order to enhance the photoactivity of nanostructured WO3 thin films, the RSM-derived predictive model obtained in this study was employed to optimise the four independent parameters (Xi ) during the electrodeposition synthesis process. Considering the model utilised in our study is an empirical basis, three levels of all the

Fig. 4. (a) XRD patterns of nanostructured WO3 thin films annealed at 400, 500 and 600 ◦ C, respectively. (b) Measured photocurrent densities of nanostructured WO3 thin films annealed at 400, 500 and 600 ◦ C, respectively

process parameters in the experimental design were applied to obtain the predicted optimum of photocurrent. The optimum condition for the maximum possible photocurrent density was found out to be: deposition time of 60 min, precursor W concentration of 0.15 mol/L, annealing temperature of 600 ◦ C and pH 1.0. Fig. 6 shows the FESEM image of nanocrystalline WO3 thin film prepared at the RSMpredicted optimum condition. From Fig. 6, it can be measured the average WO3 nanocrystals size is around 80 nm. At the optimum condition, the predictive model estimated the photocurrent density to be 128 μA/cm2 . In order to confirm the validity of the proposed predictive model, three replicates of validation experiments were conducted under the optimum condition. The average experimental measured photocurrent density was found to be 120 μA/cm2 , which is very close to the model predicted value of 128 μA/cm2 . Hence, the optimum condition as determined by using the RSM-derived predictive model was successfully validated and it was confirmed that the predictive model can be used to optimise the photocurrent density of nanostructured WO3 thin films as prepared by electrodeposition process.

Please cite this article as: T. Zhu et al., Effects of electrodeposition synthesis parameters on the photoactivity of nanostructured tungsten trioxide thin films: Optimisation study using response surface methodology, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2015.12.010

JID: JTICE 8

ARTICLE IN PRESS

[m5G;January 8, 2016;7:0]

T. Zhu et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–9

Fig. 5. (a) Pourbaix diagram for W (25 ◦ C, [WO4 2 – ] 10– 6 mol/L). The line a expresses the reduction equilibrium of water E0a = –0.0591 pH [V] at a hydrogen or oxygen pressure of 1 atm. The line b expresses the oxidation equilibrium of water E0b = 1.228–0.0591 pH[V] (Reprinted from [25]); (b)(c)(d) FE-SEM image of WO3 thin films with different pH value (0.5,1.0.1.5).

thickness of WO3 thin films, respectively. The optimum condition of the four independent parameters were found and experimentally validated to be: deposition time of 60 min, precursor W concentration of 0.15 mol/L, annealing temperature of 600 °C and pH 1.0. At the optimum condition, the predicted and experimental measured photocurrent densities were found to be 128 μA/cm2 and 120 μA/cm2 , respectively. Thus, this concludes that the RSM-derived predictive model provides a very good insight and estimation of the optimum photocurrent density without the necessity of substantial experimentation efforts. Acknowledgement

Fig. 6. FE-SEM image of nanocrystalline WO3 thin film prepared at the RSMpredicted optimum condition of 60 min deposition time, precursor W concentration of 0.15 mol/L, annealing temperature of 600 ◦ C and pH 1.0.

4. Conclusion In this study, four independent electrodeposition synthesis parameters for nanostructured WO3 thin films were systematically investigated by using the Box–Behnken RSM design. These parameters are the deposition time, precursor W concentration, annealing temperature and pH of the precursor W solution. ANOVA results showed that the annealing temperature was the most significant parameter followed by pH, precursor W concentration and the deposition time. XRD results showed that the as-deposited amorphous WO3 thin film can be transformed into monoclinic WO3 crystal structure with different preferred growth orientation after annealing treatment at 400 °C–600 °C. Both surface morphology and final reduction products of the films were found to be significantly affected by the pH of precursor W solution. It was found that the W concentration and deposition time will result in different deposition rate of film growth and

The authors are grateful to the financial support provided by the eScience fund (Project No: 03-02-10-SF0121) from Ministry of Science, Technology and Innovation (MOSTI), Malaysia. Similar gratitude also goes to the Advanced Engineering Platform and School of Engineering of Monash University Malaysia. Reference [1] Bettini LG, Della Foglia F, Milani P, Piseri P. Nanostructured carbon substrate improves the photoelectrochemical water splitting activity of cluster-assembled TiO2 thin films. Int J Hydrogen Energy 2015;40:6013–20. [2] Bartels JR, Pate MB, Olson NK. An economic survey of hydrogen production from conventional and alternative energy sources. Int J Hydrogen Energy 2010;35:8371–84. [3] Bignozzi CA, Caramori S, Cristino V, Argazzi R, Meda L, Tacca A. Nanostructured photoelectrodes based on WO3 : applications to photooxidation of aqueous electrolytes. Chem Soc Rev 2013;42:2228–46. [4] Liu Y, Li Y, Li W, Han S, Liu C. Photoelectrochemical properties and photocatalytic activity of nitrogen-doped nanoporous {WO3 } photoelectrodes under visible light. Appl Surf Sci 2012;258:5038–45. [5] Zhu T, Chong MN. Prospects of metal–insulator–semiconductor (MIS) nanojunction structures for enhanced hydrogen evolution in photoelectrochemical cells: A review. Nano Energy 2015;12:347–73. [6] Ni M, Leung MKH, Leung DYC, Sumathy K. A review and recent developments in photocatalytic water-splitting using for hydrogen production. Renew Sustain Energy Rev 2007;11:401–25. [7] Zhu T, Chong MN, Chan ES. Nanostructured tungsten trioxide thin films synthesized for photoelectrocatalytic water oxidation: A review. ChemSusChem 2014;7:2974–97.

Please cite this article as: T. Zhu et al., Effects of electrodeposition synthesis parameters on the photoactivity of nanostructured tungsten trioxide thin films: Optimisation study using response surface methodology, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2015.12.010

JID: JTICE

ARTICLE IN PRESS

[m5G;January 8, 2016;7:0]

T. Zhu et al. / Journal of the Taiwan Institute of Chemical Engineers 000 (2016) 1–9 [8] Kwong WL, Qiu H, Nakaruk a, Koshy P, Sorrell CC. Photoelectrochemical properties of WO3 thin films prepared by electrodeposition. Energy Proc 2013;34:617– 26. [9] Khamis SS, Lajis MA, Albert RAO. A sustainable direct recycling of aluminium chip (AA6061) in hot press forging employing response surface methodology. Proc CIRP 2015;26:477–81. [10] Chong MN, Jin B, Chow CWK, Saint CP. A new approach to optimise an annular slurry photoreactor system for the degradation of Congo Red: Statistical analysis and modelling. Chem Eng J 2009;152:158–66. [11] Wu J, Yu D, Sun H, Zhang Y, Zhang W, Meng F, et al. Optimizing the extraction of anti-tumor alkaloids from the stem of Berberis amurensis by response surface methodology. Ind Crops Prod 2015;69:68–75. [12] Chong MN, Zhu HY, Jin B. Response surface optimization of photocatalytic process for degradation of Congo Red using H-titanate nanofiber catalyst. Chem Eng J 2010;156:278–85. [13] Kwong WL, Savvides N, Sorrell CC. Electrodeposited nanostructured WO3 thin films for photoelectrochemical applications. Electrochim Acta 2012;75:371–80. [14] Yang B, Li H, Blackford M, Luca V. Novel low density mesoporous WO3 films prepared by electrodeposition. Curr Appl Phys 2006;6:436–9. [15] Sahoo C, Gupta AK. Optimization of photocatalytic degradation of methyl blue using silver ion doped titanium dioxide by combination of experimental design and response surface approach. J Hazard Mater 2012;215–216:302–10. [16] Mohammadi MM, Vossoughi M, Feilizadeh M, Rashtchian D, Moradi S, Alemzadeh I. Effects of electrophoretic deposition parameters on the photocatalytic activity of TiO2 films: Optimization by response surface methodology. Colloids Surfaces A Physicochem Eng Asp 2014;452:1–8.

9

[17] Eric LM, Bjorn M, Daniela P, Richard R. Optimization of hybrid photoelectrodes for solar water-splitting. Eletrochem Solid-State Lett 2006;9:248–50. [18] Yang B, Barnes PRF, Bertram W, Luca V. Strong photoresponse of nanostructured tungsten trioxide films prepared via a sol-gel route. J Mater Chem 2007;17:2722– 9. [19] Vidyarthi VS, Hofmann M, Savan A, Sliozberg K, König D, Beranek R, et al. Enhanced photoelectrochemical properties of WO3 thin films fabricated by reactive magnetron sputtering. Int J Hydrogen Energy 2011;36:4724–31. [20] Hong SJ, Jun H, Lee JS. Nanocrystalline WO3 film with high photo-electrochemical activity prepared by polymer-assisted direct deposition. Scr Mater 2010;63:757– 60. [21] Nah Y-C, Ghicov A, Kim D, Schmuki P. Enhanced electrochromic properties of selforganized nanoporous WO3 . Electrochem Commun 2008;10:1777–80. [22] Rahman G, Joo O-S. Photoelectrochemical water splitting at nanostructured α Fe2 O3 electrodes. Int J Hydrogen Energy 2012;37:13989–97. [23] Patrick E, Orazem ME, Sanchez JC, Nishida T. Corrosion of tungsten microelectrodes used in neural recording applications. J Neurosci Methods 2011;198:158– 71. [24] Liu X, Wang F, Wang Q, Liu X, Wang F, Wang Q. Nanostructure-based WO3 photoanodes for photoelectrochemical water splitting. Phys Chem Chem Phys 2012;14:7894–911. [25] Martínez-Rivera MC, Berry BW, Valentine KG, Westerlund K, Hay S, Tommos C. Electrochemical and structural properties of a protein system designed to generate tyrosine Pourbaix diagrams. J Am Chem Soc 2011;133:17786–95.

Please cite this article as: T. Zhu et al., Effects of electrodeposition synthesis parameters on the photoactivity of nanostructured tungsten trioxide thin films: Optimisation study using response surface methodology, Journal of the Taiwan Institute of Chemical Engineers (2016), http://dx.doi.org/10.1016/j.jtice.2015.12.010