1 Introduction

Obtaining clean and sustainable energy sources has become increasingly important with the extensive use of non-sustainable fossil fuels, the depletion of their reserves, and their possible environmental harm [1, 2]. Solar power is a renewable energy resource that possesses a theoretical potential of 1.2 × 105 TW, which is considerably greater than those of other renewable energy sources, such as wind, wave, hydropower and biomass [3]. Solar energy can be converted into chemical fuels, such as hydrogen fuel, through a photo-electrochemical (PEC) cell [4] as a promising alternative for carbon-based fuels, which was established by Fujishima and Honda in 1972 [5]. Photocatalytic hydrogen production is possible under ambient conditions with the use of irradiated semiconductors and mixtures of water and alcohol as electrolyte [6]. Nevertheless, the application of PEC cells is limited by their low solar energy conversion efficiency to hydrogen because no materials meet the requirements for water splitting without applying an external bias [7, 8]. Applying an additional internal or external bias is a practical approach to drive water redox reactions, overcome overvoltage losses and enhance PEC cell performance. The application of an electrical bias from fossil fuel sources or a chemical bias based on pH differences is undesirable due to its limitations [9]. The combination of photovoltaic (PV) and PEC cells is a practical strategy to drive water redox reactions and convert solar energy into hydrogen. Dye-sensitized solar cells (DSSCs) are an attractive PV device owing to their low manufacturing cost, high photoanode stability and capacity to generate sufficient electricity for unassisted solar water splitting.

To date, visible-light active materials, such as WO3 and Fe2O3, are the main photoanodes for PEC in the PEC/DSSC system [3, 10,11,12]. However, these photocatalysts are mostly used for water photooxidation process due to their positive conduction band (CB) edges compared with the hydrogen redox potential [13,14,15]. TiO2 is a potential candidate for PEC cell because of its unique features, including low cost, non-toxicity, water stability and negative CB position [16]. However the photocatalytic of TiO2-based PEC cells is still far practical approach because of their predominant absorption in the UV region, low electron mobility, and slow hole kinetics [17, 18]. Therefore, the modification of these PEC cells by shifting the absorbance edge to the visible region and diminishing the charge carrier recombination rate has the positive effect on the photocatalytic performance [19, 20]. Many research works have been reported to develop different materials for photosplitting of water in PEC cells but less attention has been paid to the optimization of the reaction’s conditions of PEC cells for photocatalytic hydrogen production in the PEC/DSSC system. Understanding the relationship between PEC/DSSC system characteristics and solar hydrogen production is important in designing an efficient solar system. In our previous study, we optimized solar hydrogen production over bimetallic Cu-/Ni-doped TiO2 thin film in the PEC/DSSC system by varying the preparation parameters, including the molar ratios of water, acetic acid, Cu to titania precursor, calcination temperature and duration, and total metal loading [21, 22].

In the present study, we systematically investigated the effects of fabrication parameters of 5 mol% Cu–Ni/TiO2 thin film such as sintering temperature and thickness and reaction parameters (electrolyte concentrations and applied voltages) on solar-driven hydrogen production in the PEC/DSSC system. Detailed analyses were carried out to examine the variations in charge carrier transfer resistance, photocurrent density and flat band potential in the PEC/DSSC system. Finally, the stability of the photoanode under the optimum reaction conditions was examined for continuous illumination of four days.

2 Experimental

2.1 Preparation of nanostructured 5 mol% Cu–Ni/TiO2 photocatalysts

The 5 mol% Cu–Ni/TiO2 photocatalyst was synthesized with sol–gel hydrothermal method using the following procedure: a solution was prepared by mixing titanium tetraisopropoxide, anhydrous ethanol, glacial acetic acid (1:8:4 ratio), Ni(NO3)2·6H2O, and Cu(NO3)2·3H2O (Cu:Ni:6:4 ratio with the total metal loading of 5 mol%). The prepared solution was added dropwise to a mixture of water and ethanol (32:4 ratio). The resulted hydroxide precipitates were directly transferred to a Teflon-lined stainless-steel vessel for hydrothermal treatment at 180 °C for 12 h. Finally, the products were separated and dried at 105 °C and calcined at 450 °C for 2 h. The details of the optimization processes of Cu–Ni/TiO2 photocatalysts are reported in our previous research works [21,22,23].

2.2 Experimental setup and procedures

2.2.1 Fabrication of 5 mol% Cu–Ni/TiO2 photoanode

The 5 mol% Cu–Ni/TiO2 thin film (1 × 1 cm2) was fabricated using polyester screen material of interconnected photocatalyst particles on the surface of conductive glass (F-doped tin oxide) substrate. The details of preparation screen printable paste are explained in our previous work [23]. A small strip of conductive silver ink as the ohmic contact was printed on the top of printed pasted and followed with sintering process and finally a copper wire was soldered on the silver strip. The prepared photoanode was immersed in a KOH solution with 5 vol% glycerol in a designed glass reactor (150 mL) for solar hydrogen measurement and photo-electrochemical study.

2.2.2 Fabrication of DSSC

The photoanode of PEC cell was connected in series to a fabricated DSSC (0.7 V), including a double-layered sensitized TiO2 photoanode with N719 dye, a printed counter electrode with platinum paste (dyesol) and iodide/triiodide redox couple as the electrolyte. A 500 W halogen lamp was utilized as a light source for visible-light radiation (λ > 400 nm) with a light intensity of 100 mW cm−2. The produced hydrogen was collected around the Pt rod in a reversed burette during 2 h and the gaseous products were identified by using multiple gas analyzer (SRI Instruments 8610-0071).

2.2.3 Photo-electrochemical measurements

The photo-electrochemical characteristics were studied with a conventional three-electrode configuration using a computer-controlled potentiostat (Autolab PGSTAT302N, Metrohm) with frequency response analyzer (FRA) module and NOVA software version 1.10.1.9. Electrochemical impedance spectroscopy (EIS) measurement was performed in the range of 0.1 Hz–100 kHz and AC amplitude of 10 mV in the dark using Z-Plot/Z-View to control instrument operation and data acquisition. Mott–Schottky (M–S) data were collected in the dark at a frequency of 1.0 kHz and a signal amplitude of 10 mV [24, 25]. The extrapolation on the X-axis of the M–S plot shows the flat band potential (Vfb). Donor density (ND) can be calculated from the slope (m) of the linear region of the M–S plot by Eq. (1)

$${N_{\text{D}}}=2/\bar {e}{\varepsilon _0}\varepsilon m$$
(1)

where ē is an electronic charge unit (1.6 × 10−19 C), Ɛ0 is the permittivity of the free space charges (8.86 × 10−12 Fm−1), Ɛ is the dielectric constant of anatase TiO2 (48 Fm) [26, 27]. The photocurrent density versus applied potential (I–V) was measured with linear sweep voltammetry (LSV) procedure for data acquisition at potential from − 0.5 to 1.0 V versus Ag/AgCl with a scan rate of 20 mV s−1 [21,22,23].

3 Results and discussion

3.1 Photocatalyst characterization

Physicochemical characteristics of the powder form of 5 mol% Cu–Ni doped TiO2 have been described in our previous publication [21]. Briefly, photocatalyst consists of anatase nanocrystals with the average crystal sizes of 11.37 nm and its particles are almost spherical shape with average size of 11.54 nm. In addition, it had the active surface area of 76.41 m2 g−1 with a band gap energy of 2.48 eV, which is capable of harvesting photons in the visible region of the solar spectrum. We reported that the accumulative hydrogen over this photocatalyst was more than other amounts of total metal loadings (10 and 15 mol%) [21]. In the following sections, the influence of sintering temperatures, photoanode thicknesses, electrolyte concentrations, applied voltages on the photo-electrochemical properties, and photocatalytic hydrogen production of 5 mol% Cu–Ni doped TiO2 thin film are discussed in details.

3.2 Effect of sintering temperature

Photoanode paste compositions are compacted by sintering, allowing injected electrons to be collected at the underlying conductive substrate. Therefore, sintering temperature optimization is desirable to ensure good adherence of the photoanode on the substrate, eliminating the existence of organic residues in the paste and improving the connection for efficient charge transport among the nanocrystallites constituting the thin film. Figure 1a shows the thermal decomposition of the paste components and the photoanode before and after sintering. It reveals that all organic compounds were decomposed until 400 °C and the thin film showed thermal stability. Figure 1b displays that the photoanode sintered at 400 °C produced the highest accumulative hydrogen (7.8 mL) and the amount of produced hydrogen was reduced by increasing the sintering temperature. The effect of different sintering temperatures on the XRD pattern (supplementary information, Fig. S1) of 5 mol% Cu–Ni/TiO2 photoanode depicts that anatase TiO2 (JCPDS 21-1272) was not affected by the sintering temperature. Increasing sintering temperature slightly shifted TiO2 anatase peaks to a high 2θ degree without phase transition from anatase to rutile. However, the calculated crystallite sizes increase from 11.19 (400 °C) to 12.97 (500 °C) [22]. The FESEM images (Fig. S2) indicate that the increasing sintering temperature led to increase the degree of agglomeration particle sizes and reduce the porosity of the photoanode.

Fig. 1
figure 1

a Thermal decomposition of paste components and photoanode before and after sintering and b hydrogen production at different sintering temperatures (four layers printed 5 mol% Cu–Ni/TiO2, electrolyte: 1 M KOH and 5 vol%, T: 25 °C)

Full size image

EIS measurements were conducted for each photoelectrode at the open circuit potential to monitor the interaction between the electrode and electrolyte interface and charge transfer resistances. Figure 2a illustrates the typical Nyquist plots of all sintered thin films by denoting the x- and y-axis to the real part of the measured impedance (Z′) and the negative number of imaginary part of the measured impedance (−Z″), respectively. The circuit equivalent fittings of the EIS data are presented in Table 1.

Fig. 2
figure 2

a EIS spectra, b M–S plots and c I–V curves at different sintering temperatures (thickness: 4L, electrolyte: 1 M KOH and 5 vol%, T: 25 °C)

Full size image
Table 1 Circuit equivalent fitting data at different sintering temperatures
Full size table

The impedance spectra demonstrated that the charge transfer resistance (Rct) increased and the electron life-time significantly reduced after increasing the sintering temperature up to 500 °C. Increasing charge transfer resistance shows that the presence of several surface states can trap the electron during the EIS measurement. These trapped electrons are difficult to transfer to the interface photoanode and electrolyte [28]. Increasing the Rct value beyond 400 °C can be explained by the collapse of porous structure and production of some new surface states (mainly from the surface atoms), which can control the electron transfer through the surface state and increase recombination rate [29]. Furthermore, the collected data from M–S plots (Fig. 2b) confirm that donor density was reduced and flat band shifted to a positive potential with the negative influence on the reduction of water to hydrogen [30]. In addition, I–V curves versus applied potential (Fig. 2c) reveals that increasing sintering temperature reduced the photocurrent density on the electrode/electrolyte interface due to high Rct and recombination rate.

3.3 Photoanode thickness

Figure 3 shows the effect of photoanode thickness (number of printed layers; 2L, 3L, 4L, and 5L) on the hydrogen production. Cross section FESM images (Fig. S3) of each photoanode shows thicknesses of 12.25, 18.59, 24.8, and 31.5 µm for 2L, 3L, 4L, and 5L, respectively. The accumulative hydrogen production increased as the photoanode thickness was increased from 2 to 4L. However, the total produced hydrogen reduced up to 5 mL by exceeding the number of printing layers beyond the optimum level. A similar trend was reported by Michael et al. for 4-chlorophenol degradation [31].

Fig. 3
figure 3

Effect of different photoanode thicknesses on hydrogen production (thickness: 2–5L, electrolyte: 1 M KOH and 5 vol%, T: 25 °C)

Full size image

EIS spectra (Fig. 4a) and its fitting data (Table 2) confirm that the electron transport resistance decreased when the thin film thickness was increased up to 4L. The Rct value reduced from 2 to 4L because the interfacial adhesion between Cu–Ni/TiO2 nanoparticles and FTO substrate was enhanced in the alkaline solution [32]. Increasing photoanode thickness generated more excited electrons due to better light absorption and more contribution of the Cu–Ni/TiO2 photocatalyst on the photocatalytic reaction [33]. The Rct increased and the electron lifetime reduced when the photocatalyst layers were beyond the optimum thickness possibly due to the long path for the photogenerated electrons and the high electrolyte penetration in the photoanodes [34,35,36]. In addition, M–S plots (Fig. 4b) show flat band potentials slightly shifted to a more negative potential by increasing the thickness up to 4L. In addition, increasing donor density by thickness until 4L (Table 2) confirmed the increase in electron transfer from the bulk to the electrode surface and the decrease in resistance at the electrode/electrolyte interface [37].

Fig. 4
figure 4

a EIS spectra and b M–S plots for different photoanode thicknesses (thickness: 2–5L electrolyte: 1 M KOH and 5 vol%, T: 25 °C)

Full size image
Table 2 Circuit equivalent fitting results for different photoanode thicknesses
Full size table

3.4 Electrolyte concentration

Electrolyte concentration is another important parameter in PEC cells; it controls the ionic transfer in solution and electrical resistance of electrolytes [38,39,40]. Normally, high electrolyte concentration can enhance the PEC cell performance up to the optimum level, as shown in Fig. 5. Photocatalytic hydrogen production was increased from 0.5 (3.4 mL) to 2 M (8.5 mL). However, increasing KOH concentration beyond the optimum level (2 M) has a negative effect on hydrogen production.

Fig. 5
figure 5

Influence of different electrolyte concentrations on hydrogen production [thickness: 4L (electrolyte: 0.5–3) M KOH and 5 vol%, T: 25 °C]

Full size image

The concentration of the supporting electrolyte should be sufficiently high (> 0.5 M) to avoid large ohmic voltage losses across the electrolyte. Figure 6a and Table 3 represent the Nyquist spectra and the fitting data of photocatalyst performance at different KOH solutions. The results confirmed that Rct and RE (intersection of the EIS spectrum with Z′) decreased when the KOH concentration was increased up to 2 M.

Fig. 6
figure 6

a EIS spectra, b I–V curves and c M–S plots in different KOH solutions [thickness: 4L (electrolyte: 0.5–3) M KOH and 5 vol%, T: 25 °C]

Full size image
Table 3 Circuit equivalent fitting results with different KOH concentrations
Full size table

However, the photocurrent densities decreased when the KOH concentration exceeding 2 M (Fig. 6b) due to high electrolyte resistance (RE) and Rct. The increasing RE results in voltage loss at the system as expected by Eq. (2) [41, 42].

$${V_{{\text{loss}}}}=I \times {R_{\text{E}}},$$
(2)

Furthermore, the M–S plots shown in Fig. 6c demonstrate that the electrolyte concentration can shift the flat band potential to a negative potential until 2 M. However, a slightly positive shift in the flat band potential was observed in the electrolyte with 3 M KOH, which had a negative effect on hydrogen production. The 2 M KOH solution was selected as the optimum electrolyte concentration for further study.

3.5 Applied voltage

Figure 7a reveals solar hydrogen production as a function of applied voltage (0.7, 3.4, 5.5, 9.5, and 16.5 V) under the optimum reaction conditions. The maximum produced hydrogen under optimum conditions over the applied voltage of 0.7 eV was 8.5 mL, whereas the produced hydrogen sharply increased until 24.6 mL at the applied voltage of 3.4 V. Increasing voltage until 16.5 V reduced the hydrogen production to 19.6 mL. The optimum applied voltage of 3.4 V [1.9 V vs. reduced hydrogen electrode (RHE)] agrees with the practical reported voltage between 1.6 and 2 V versus RHE [9]. Improving the photocatalytic performance of system were related to increasing photocurrent density, minimizing the charge carrier recombination and charge transfer resistance in the photoanode up to optimum level [40, 41]. However, increasing the applied potential beyond the optimum level has the negative effect of hydrogen production in which reaction rate remained almost constant by applying more potential. High-applied potential can fully separate photogenerated charge carriers and increase photocurrent saturation on the surface of photoanode, resulting in high charge transfer resistance and recombination rates as confirmed in EIS spectra (Fig. 7b). The Rct value sharply reduced from 22.45 to 7.45 Ω by increasing voltage from 0.7 to 3.4 V (Table S1) due to improved separation of the photogenerated charge carriers and more photoexcited electrons on the surface of the electrode and the electrolyte, saturating the photocurrent on the surface of the photoanode with the negative effect on hydrogen production. However, increasing applied voltage from 5.5 to 16.5 V were caused the Rct increased from 6.08 to 22.28 V (Table S1). High-applied voltage causes increasing the number of hydrogen bubbles on the surface of Pt electrode and electrical resistance, diminishing the electron transfer and increasing the ohmic loss of the whole system [38]. The maximum hydrogen production difference between the highest and lowest levels of applied voltage was 16.5 mL. Thus, this parameter can be considered as the most effective parameter among the applied parameters. However, further optimization of hydrogen production is limited due to its maximum effect around the practical voltage of the PEC cell in the range of 1.6–2 V. We compared the photocatalytic performance of 5 mol% Cu–Ni/TiO2 photoanode under the optimum conditions with those of photoanodes in relevant past works in PEC/DSSC system. However, the comparison was difficult because different photoanodes and reaction conditions were used. Therefore, data were scaled to a light intensity of 100 mW cm− 2, a geometric area of 1 cm2 and an applied voltage of 3.4 V. The scaled data of reported hydrogen production rate by Lee et al. [43] and Shi et al. [7] were 3.35 mL h−1 cm−2 over WO3 and 4.72 mL h−1 cm−2 over WO3/BiVO4, respectively in the PEC/DSSC system. The fabricated 5 mol% Cu–Ni/TiO2 photoanode with a hydrogen production rate of 12.45 mL h−1 cm−2 can facilitate the transportation of electron to the photoanode/electrolyte interface and enhance PEC/DSSC performance. The better performance of this photocatalyst confirms that CB edges of bulk WO3 and BiVO4 are excessively positive for hydrogen production compared with water reduction potential.

Fig. 7
figure 7

a Hydrogen production and b EIS spectra of different applied voltages [thickness: 4L (electrolyte: 2 M KOH and 5 vol%, T: 25 °C)]

Full size image

3.6 Stability testing of photoanode

The long-term performance of the PEC/DSSC system was tested by immersing the modified 5 mol% Cu–Ni/TiO2 photoanode with the thickness of 4L in 2 M KOH under the applied voltage of 3.4 V. Figure 8 shows the activity of the optimized photocatalyst during 4 days of illumination from 8 am to 8 pm.

Fig. 8
figure 8

Trend of hydrogen production for 4 days [thickness: 4L (electrolyte: 2 M KOH and 5 vol%, T: 25 °C)]

Full size image

Table 4 shows the volume of produced hydrogen and the photo-voltage of the system for each day with the total accumulative hydrogen production of 443.4 mL. The hydrogen production rate gradually declined from the first to the fourth day, although the system voltage slightly increased at the second and third days. The increase in system voltage may be related to the increase in DSSC voltage for long-term irradiation as reported by Xue et al. [44].

Table 4 Total hydrogen production volume, rate and photo-voltage for 4 days
Full size table

The rate of hydrogen production reduced during 4 days in the closed system due to the increase in electrolyte concentration, decrease in the number of H+ ions in the solution, high resistance on the Pt surface, and reverse reaction between oxygen and hydrogen. Furthermore, the observed oxygen bubbles around the photoanode on the fourth day indicated the reduced amount of glycerol as a sacrificial agent in the solution. Therefore, the slow oxidizing process of OH− increased the charge carrier recombination rate and decreased the hydrogen production rate.

4 Conclusion

The effect of important experimental parameters on the analytical performance of the PEC/DSSC system was investigated. Optimization process demonstrates that the photoanode thin film fabrication controls the electrochemical properties of photocatalysts, which in turn affects the photocatalytic activity. Furthermore, no linear trend was observed between hydrogen production and reaction parameters in the PEC/DSSC system. The optimization of the reaction parameters demonstrated that the sintered 5 mol% Cu–Ni/TiO2 photoanode at 400 °C with the thickness of 4L in 2 M KOH solution under the applied voltage of 3.4 V produced the maximum hydrogen of 24.9 mL. The optimization results confirm that applied voltage exerts the most significant effect on the photocatalytic hydrogen production among the parameters used. Highly stable photocatalyst in the PEC/DSSC system can generate sufficient voltage for photocatalysis to produce 443.4 mL of hydrogen after 4 days. The optimization of the preparation condition for the photoanode and reaction parameters introduce cost-effective and less complex system to improve solar hydrogen production.