Synthesis of Cobalt Oxide on FTO by Hydrothermal Method for Photoelectrochemical Water Splitting Application

: Cobalt oxide thin ﬁlms were successfully grown directly on ﬂuorine-doped tin oxide glass substrates through a simple, green, and low-cost hydrothermal method. An investigation into the physicochemical characteristics and photoelectrochemical (PEC) properties of the developed cobalt oxide thin ﬁlm was comprehensively performed. At various annealing temperatures, different mor-phologies and crystal phases of cobalt oxide were observed. Microﬂowers (Co 3 O 4 ) and microﬂowers with nanowire petals (Co 3 O 4 /CoO) were produced at 450 ◦ C and 550 ◦ C, respectively. Evaluation of the PEC performance of the samples in KOH (pH 13), Na 2 SO 4 (pH 6.7), and H 2 SO 4 (pH 1) revealed that the highest photocurrent − 2.3 mA cm − 2 generated at − 0.5 V vs. reversible hydrogen electrode (RHE) was produced by Co 3 O 4 (450 ◦ C) in H 2 SO 4 (pH 1). This photocurrent corresponded to an 8-fold enhancement compared with that achieved in neutral and basic electrolytes and was higher than the results reported by other studies. This promising photocurrent generation was due to the abundant source of protons, which was favorable for the hydrogen evolution reaction (HER) in H 2 SO 4 (pH 1). The present study showed that Co 3 O 4 is photoactive under acidic conditions, which is encouraging for HER compared with the mixed-phase Co 3 O 4 /CoO.


Introduction
Hydrogen (H 2 ) is an ideal future energy carrier to replace fossil fuels. H 2 as fuel in a fuel cell system has been proven to be safe, clean, and environmentally friendly, and its only byproduct is pure water [1]. H 2 can be extracted from water molecules through photoelectrochemical (PEC) water-splitting using direct solar energy, which has been actively studied globally [2,3]. Both components, water and sunlight, are abundant and available everywhere. PEC water-splitting requires a semiconductor material as a photoelectrode. This photoelectrode is exposed to sunlight and is the site for the initiation of the catalytic reaction. The results depend on the nature of the semiconductor. The n-type semiconductors, such as titanium dioxide (TiO 2 ), act as a photoanode; the water molecule is oxidized to produce oxygen (O 2 ) gas. Meanwhile, p-type semiconductors, such as Cu 2 O, act as a photocathode. The proton (H + ) is reduced, and H 2 is generated [4].
A few requirements need to be fulfilled by photoelectrodes to achieve an efficient PEC water-splitting process. The bandgap energy should be greater than 1.23 eV and less than 3 eV to allow the utilization of the spectral range of visible light [5]. To electrochemically split the water molecule, the band edges of the photoelectrode need to be in straddle position to the water molecule redox potential, which means the conduction band (CB) of the material should lay at a more negative potential than the water reduction potential. Meanwhile, the valence band (VB) position should lay at a more positive potential to the

Characterization
The crystallinity of the sample was determined using X-ray diffraction (XRD) with Cu Kα radiation at a wavelength of 0.15406 Å (Siemens X-ray Diffractometer D5000, Munich, Germany). The surface morphology of the samples was observed using a field emission scanning electron microscope (FESEM) (JEOL JSM-7600F, Tokyo, Japan) and high-resolution transmission electron microscopy (HRTEM) (JEOL JEM-2100F, Japan). The optical properties of the samples were examined using a UV-Vis-NIR spectrophotometer (UV-3101PC Shidmadzu, Kyoto, Japan). Details of the chemical composition and binding energy were obtained with an X-ray photoelectron spectrometer (Fourier Kratos Analytical Axis Ultra DLD/2009, Manchester, UK).

Photoelectrochemical Measurement and Stability Test
Photochemical (PEC) measurements were performed with a three-electrode configuration that comprised the samples, saturated Ag/AgCl, and platinum as the working, reference, and counter electrodes, respectively. These cables were connected to a potentiostat (Versa stat, Ametex, Oak Ridge, TN, USA). To evaluate the effect of the pH of the electrolytes on the photocurrent generation, different solutions of 0.5 M KOH (pH 13), 0.5 M Na 2 SO 4 (pH 7), and 0.5 M H 2 SO 4 (pH 1) were prepared as the electrolytes. Prior to PEC measurement, the electrolytes were degassed with N 2 for 10 min. The photocurrent measurements were conducted under simulated solar light illumination at a power density of 100 mW cm −2 with a Xenon lamp as the light source (ORIEL). Under similar conditions, Mott-Schottky data were collected under illumination at a frequency of 1000 Hz. The electrochemical impedance of the samples was evaluated using electrochemical impedance spectroscopy (EIS) by scanning from 1 Hz to 100,000 Hz using a similar potentiostat. The applied potential was converted into the reversible hydrogen electrode (RHE) scale using the Nernst equation; E RHE = E Ag/AgCl + 0.05196pH + 0.1976.

Phase and Crystal Structure Analysis
The cobalt oxide thin film on FTO was fabricated using a sequential hydrothermal method and annealed under inert conditions. The product from the hydrothermal process that was not annealed (named as a pre-annealed sample) was also analyzed to confirm the first product produced by the hydrothermal method. XRD analysis was utilized in this study to fully understand the phase of cobalt oxide produced according to the annealing temperature. The XRD chromatogram in Figure 1 shows that cobalt (III) oxide (Co 3 O 4 ) was formed on the FTO after the hydrothermal reaction (pre-annealed), as indicated by the diffracted peaks of (111), (220), (311), (511), and (440), which corresponded to the plane of the cubic crystal structure with the spinel Co 3 O 4 phase (JSPD NO 00-042-1467). In addition, as the sample was later subjected to an annealing treatment of 450 • C, the intensity of the peaks belonging to the Co 3 O 4 phase markedly increased, indicating an increase in crystallinity. Interestingly, at 550 • C, a mixed phase of Co 3 O 4 and CoO was observed at approximately 36 • , 45 • , and 59 • , indicating that oxidation of Co 3 O 4 to CoO occurred. According to previous studies [28], increasing the annealing temperature decreases the amount of oxygen in the films, which might be explained by the increase in purity and crystallinity of the as-deposited Co 3 O 4 (pre-annealed) and after annealing at 450 • C, and by the loss of oxygen in the form of water vapor during the annealing treatment. To confirm the composition of pre-annealed 450 °C and 550 °C samples, X-ray photoelectron spectroscopy (XPS) analysis was performed as displayed in Figure 2. The preannealed and the annealed 450 °C samples have similar full survey spectra. The Co 2p spectra for the pre-annealed appeared at the value of binding energy for 778 ± 0.5 and 800 ± 0.5 eV, whereas the annealed 450 °C appeared at the value of binding energy for 778 ± 0.5 and 793 ± 0.5 eV. The spectra of the pre-annealed are present in a broad area compared to the Co 2p spectra for the annealed 450 °C. The broad value of the pre-annealed perhaps is caused by water or solvent contaminant. A similar trend was also reported by Chen et al. 2020 on the annealing effect on MoS2 [29]. The 2p 3/2 and 2p 1/2 peaks separation for both pre-annealed and annealed 450 °C samples are around ± 10 eV. According to H. Jadhav et al., between the 2p 3/2 and 2p 1/2 peaks, an energy difference of ~15 eV is characteristic of the Co3O4 phase [30]. The major peak in Co 2p 3/2 for the sample annealed at 450 °C de-convoluted into two peaks centered at 776.96 and 778.90 eV, which corresponded to Co 3+ and Co 2+ species, respectively. For the Co 2p spectrum of the 550 °C thin film, the Co 3+ and Co 2+ species were located at 777.07 and 778.95 eV, respectively. In addition, the presence of CoO was confirmed by the indication of a low appearance of satellites [31]. The Co 2p spectrum of single-phase Co3O4 showed a clear image of weaker peaks compared with that associated with CoO in the mixed phase of Co3O4/CoO [32]. Figure 2c shows the O 1s XPS spectra of the pre-annealed, annealed 450 °C, and annealed 550 °C samples. All samples produce oxygen peaks in different binding energy and height. The oxygen peaks of pre-anneal sample present at 529 and 530.6 eV. Meanwhile, the position oxygen peaks of annealed 450 °C and 550 °C samples are in lower binding energy, that is, at 527 and 528.6 eV. They differed only in the slight shifting in intensity and the peak position, thereby indicating that some oxygen atoms bind with different oxidation states of Co. To confirm the composition of pre-annealed 450 • C and 550 • C samples, X-ray photoelectron spectroscopy (XPS) analysis was performed as displayed in Figure 2. The pre-annealed and the annealed 450 • C samples have similar full survey spectra. The Co 2p spectra for the pre-annealed appeared at the value of binding energy for 778 ± 0.5 and 800 ± 0.5 eV, whereas the annealed 450 • C appeared at the value of binding energy for 778 ± 0.5 and 793 ± 0.5 eV. The spectra of the pre-annealed are present in a broad area compared to the Co 2p spectra for the annealed 450 • C. The broad value of the preannealed perhaps is caused by water or solvent contaminant. A similar trend was also reported by Chen et al. (2020) on the annealing effect on MoS 2 [29]. The 2p 3/2 and 2p 1/2 peaks separation for both pre-annealed and annealed 450 • C samples are around ± 10 eV. According to H. Jadhav et al., between the 2p 3/2 and 2p 1/2 peaks, an energy difference of~15 eV is characteristic of the Co 3 O 4 phase [30]. The major peak in Co 2p 3/2 for the sample annealed at 450 • C de-convoluted into two peaks centered at 776.96 and 778.90 eV, which corresponded to Co 3+ and Co 2+ species, respectively. For the Co 2p spectrum of the 550 • C thin film, the Co 3+ and Co 2+ species were located at 777.07 and 778.95 eV, respectively. In addition, the presence of CoO was confirmed by the indication of a low appearance of satellites [31]. The Co 2p spectrum of single-phase Co 3 O 4 showed a clear image of weaker peaks compared with that associated with CoO in the mixed phase of Co 3 O 4 /CoO [32]. Figure 2c shows the O 1s XPS spectra of the pre-annealed, annealed 450 • C, and annealed 550 • C samples. All samples produce oxygen peaks in different binding energy and height. The oxygen peaks of pre-anneal sample present at 529 and 530.6 eV. Meanwhile, the position oxygen peaks of annealed 450 • C and 550 • C samples are in lower binding energy, that is, at 527 and 528.6 eV. They differed only in the slight shifting in intensity and the peak position, thereby indicating that some oxygen atoms bind with different oxidation states of Co. Appl. Sci. 2021, 11, x FOR PEER REVIEW 5 of 13

Structure and Morphology Characterization
The structure and morphology of the formed thin films were tested and confirmed by FESEM analysis. Figure 3 shows the morphology of the photoelectrode prepared at different annealing temperatures, namely, (a) pre-annealing, (b) 450 °C, and (c) 550 °C. The pre-annealed photoelectrode surface formed a uniformly grown flower-like structure. The high magnification FESEM images in Figure 3a revealed the formation of microscale flowers, which are composed of micro-sized petals with an average thickness of ~6 µm. According to Zhu et al., urchin-like spheres are composed of Co3O4. When the annealing temperature increased to 450 °C, the color of the photoelectrode changed from purple to black. FESEM observation revealed that the Co3O4 produced a microflower that started to decompose ( Figure 3b). With increasing annealing temperature, the Co3O4 changed to mixed phase Co3O4.CoO, and the morphology of the sample changed to microflowers with less thick nanowire petals ( Figure 3c) [33].
Details of the microstructures of Co3O4 and mixed Co3O4.CoO were examined using HRTEM, and specifics were examined with selected area electron diffraction (SAED) (Figure 4). Lattice fringes of Co3O4 annealed at 450 °C are shown in Figure 4(a1-a3). The SAED of Co3O4 shown in Figure 4(a3) reveals the ring and spot pattern in a well-defined manner, demonstrating that the Co3O4 sample was highly crystalline. However, for the mixed Co3O4.CoO annealed at 550 °C, the SAED showed a small spot that made rings, which were designated as polycrystalline. According to Andrews et al., (1971) each spot arising from brag was reflected from individual crystalline [34]. As shown in Figure 4(b3), a SAED ring demonstrated the presence of a mixed phase that was well-blended at this phase. The first ring indicated Co3O4, whereas the second ring indicated CoO.

Structure and Morphology Characterization
The structure and morphology of the formed thin films were tested and confirmed by FESEM analysis. Figure 3 shows the morphology of the photoelectrode prepared at different annealing temperatures, namely, (a) pre-annealing, (b) 450 • C, and (c) 550 • C. The pre-annealed photoelectrode surface formed a uniformly grown flower-like structure. The high magnification FESEM images in Figure 3a revealed the formation of microscale flowers, which are composed of micro-sized petals with an average thickness of~6 µm. According to Zhu et al., urchin-like spheres are composed of Co 3 O 4 . When the annealing temperature increased to 450 • C, the color of the photoelectrode changed from purple to black. FESEM observation revealed that the Co 3 O 4 produced a microflower that started to decompose (Figure 3b). With increasing annealing temperature, the Co 3 O 4 changed to mixed phase Co 3 O 4 ·CoO, and the morphology of the sample changed to microflowers with less thick nanowire petals (Figure 3c) [33]. Appl. Sci. 2021, 11, x FOR PEER REVIEW 6 of 13

Optical Measurements
The optical properties of the photoelectrode samples were analyzed using a UV-Vis-NIR spectrophotometer. The optical absorption of the prepared photoelectrode samples and their corresponding Kubelka-Munk plots are shown in Figure 5, in which the pre-  Figure 4(a3) reveals the ring and spot pattern in a well-defined manner, demonstrating that the Co 3 O 4 sample was highly crystalline. However, for the mixed Co 3 O 4 ·CoO annealed at 550 • C, the SAED showed a small spot that made rings, which were designated as polycrystalline. According to Andrews et al., (1971) each spot arising from brag was reflected from individual crystalline [34]. As shown in Figure 4(b3), a SAED ring demonstrated the presence of a mixed phase that was well-blended at this phase. The first ring indicated Co 3 O 4 , whereas the second ring indicated CoO.

Optical Measurements
The optical properties of the photoelectrode samples were analyzed using a UV-Vis-NIR spectrophotometer. The optical absorption of the prepared photoelectrode samples and their corresponding Kubelka-Munk plots are shown in Figure 5, in which the pre-annealed samples showed an absorption region at 400-600 nm. Accordingly, the corresponding K-M plots exhibited a band gap value of 4.30 eV (Figure 5b). The pre-annealed sample exhibited absorption at much lower wavelengths and a large band gap, because the preannealed sample was not activated. For the sample prepared at 450 • C, the intensity of the light absorption was greater than that of the pre-annealed sample, which covered a wider range of the light spectrum, ranging from 200 to 900 nm (Figure 5a). Within this light spectrum range, two main broad peaks were identified, between 200-500 and 580-900 nm. Correspondingly, two main broad peaks were identified with two band gaps of 2.00 and 2.50 eV (Figure 5b). This observation suggested that a higher temperature was required to affect the phase transformation of the pre-annealed sample. This suggestion was supported by the XRD analysis result, in which an identical phase of Co 3 O 4 , as identified in the pre-annealed sample, was also detected in the sample prepared at 450 • C. However, after annealing, the band gap values significantly shifted to smaller values. This shifting pattern of the band gap value could be an indication of the intermediate phase transformation. In general, Co 3 O 4 is known as "Co 2 O 3 /CoO," which represents two constituents in the material. Therefore, these constituents might contribute to the creation of two optical absorption peaks. The first band (500-700 nm) can be referred to as the O 2− → Co 2+ charge transfer process of CoO, whereas the second absorption band (250-500 nm) can be attributed to the O 2− → Co 2+ charge transfer of Co 2 O 3 [35,36].

Optical Measurements
The optical properties of the photoelectrode samples were analyzed using a UV-Vis-NIR spectrophotometer. The optical absorption of the prepared photoelectrode samples and their corresponding Kubelka-Munk plots are shown in Figure 5, in which the pre- annealed samples showed an absorption region at 400-600 nm. Accordingly, the co sponding K-M plots exhibited a band gap value of 4.30 eV (Figure 5b). The pre-annea sample exhibited absorption at much lower wavelengths and a large band gap, beca the pre-annealed sample was not activated. For the sample prepared at 450 °C, the int sity of the light absorption was greater than that of the pre-annealed sample, which c ered a wider range of the light spectrum, ranging from 200 to 900 nm (Figure 5a). Wit this light spectrum range, two main broad peaks were identified, between 200-500 a 580-900 nm. Correspondingly, two main broad peaks were identified with two band g of 2.00 and 2.50 eV (Figure 5b). This observation suggested that a higher temperature w required to affect the phase transformation of the pre-annealed sample. This suggest was supported by the XRD analysis result, in which an identical phase of Co3O4, as id tified in the pre-annealed sample, was also detected in the sample prepared at 450 However, after annealing, the band gap values significantly shifted to smaller values. T shifting pattern of the band gap value could be an indication of the intermediate ph transformation. In general, Co3O4 is known as "Co2O3/CoO," which represents two c stituents in the material. Therefore, these constituents might contribute to the creation two optical absorption peaks. The first band (500-700 nm) can be referred to as the O 2 Co 2+ charge transfer process of CoO, whereas the second absorption band (250-500 n can be attributed to the O 2− → Co 2+ charge transfer of Co2O3 [35,36]. As the annealing temperature was increased to 550 °C, the intensity of absorpt significantly decreased; only one absorption peak that signaled between 200 and 450 remained. Extrapolation of K-M plots on the x-axis found that the band gap value w narrowed to a single band gap value of 3.1 eV (Figure 5b). When the as-prepared Co was annealed with a nitrogen gas flow, and the temperature was increased, Co 2+ was p tected against oxidation, thereby ensuring the production of CoO nanowire. Oxygen cancies occurred during annealing in an oxygen-deficient atmosphere, and this format was associated with the energy band gap. A narrowing of the band gap was the resul the increase in oxygen vacancies, and hence, the increase in the visible light absorptio As the annealing temperature was increased to 550 • C, the intensity of absorption significantly decreased; only one absorption peak that signaled between 200 and 450 nm remained. Extrapolation of K-M plots on the x-axis found that the band gap value was narrowed to a single band gap value of 3.1 eV (Figure 5b). When the as-prepared Co 3 O 4 was annealed with a nitrogen gas flow, and the temperature was increased, Co 2+ was protected against oxidation, thereby ensuring the production of CoO nanowire. Oxygen vacancies occurred during annealing in an oxygen-deficient atmosphere, and this formation  (Figure 6c). Sample 450 • C shows better stability compared to sample 550 • C. Under constant simulated solar illumination and fixed current density of 0.4 mA cm −2 , the 450 • C sample was stable until 4000 s of irradiation time before slowly decreasing. Meanwhile, the 550 • C sample was only stable for less than 500 s and decreased drastically.

Electrochemical Impedance Studies
Electrochemical impedance analyses (EIS) of the as-prepared samples were conducted to qualitatively examine the charge transfer resistance (Rct) through Nyquist plotting, as shown in Figure 7a. The pre-annealed sample had high resistance compared to the annealed samples at 450 • C and 550 • C. The Rct was evaluated based on the size of the semicircle, wherein the smallest size of the semicircle represented the lowest Rct. The effect of electrolyte on the Rct was further conducted to the annealed 450 • C sample, where the smallest semicircle was produced when EIS testing was conducted in H 2 SO 4 solution. Correspondingly, the extent of Rct evaluated herein explained the results obtained in the measurement of photocurrent generation, as shown in Figure 6. In this case, the lowest Rct improved the efficiency of charge transfer from the semiconductor to the electrolyte and vice versa, consequently yielding the highest photocurrent.
electrons. The present study suggested that the recombination rates of electrons and holes were remarkably reduced, leading to enhanced photocurrent generation. Furthermore, a chronoamperometry stability test was conducted for samples of 450 °C and 550 °C in 0.5 M H2SO4 electrolyte (Figure 6c). Sample 450 °C shows better stability compared to sample 550 °C. Under constant simulated solar illumination and fixed current density of 0.4 mA cm −2 , the 450 °C sample was stable until 4000 sec of irradiation time before slowly decreasing. Meanwhile, the 550 °C sample was only stable for less than 500 s and decreased drastically.

Electrochemical Impedance Studies
Electrochemical impedance analyses (EIS) of the as-prepared samples were conducted to qualitatively examine the charge transfer resistance (Rct) through Nyquist plotting, as shown in Figure 7a. The pre-annealed sample had high resistance compared to the annealed samples at 450 °C and 550 °C. The Rct was evaluated based on the size of the semicircle, wherein the smallest size of the semicircle represented the lowest Rct. The effect of electrolyte on the Rct was further conducted to the annealed 450 °C sample, where the smallest semicircle was produced when EIS testing was conducted in H2SO4 solution. Correspondingly, the extent of Rct evaluated herein explained the results obtained in the measurement of photocurrent generation, as shown in Figure 6. In this case, the lowest Rct improved the efficiency of charge transfer from the semiconductor to the electrolyte and vice versa, consequently yielding the highest photocurrent.

Mott-Schottky (M-S) Plot
M-S Analysis was done to study the charge carrier concentration of samples. The implementation of M-S analysis on annealed samples of 450 °C and 550 °C was performed in H2SO4 (pH 1). The charge carrier was analyzed from the flat band, wherein flat band is the potential at which the potential drop between the electrode surface and the bulk is zero. This can establish the position of semiconductor energy band with respect to the redox potential. The intrinsic value of flat band (VFB) and carrier density (NA) for both samples can be extracted from the x-intercept and slope of the plot between the reciprocal of the square 1/C 2 and the bias potential (Figure 7c), in accordance with the Mott-Schottky equation, as follows (Equation (1) where e is the electronic charge, ε is the relative permittivity of CoO (12.9), εo is the permittivity of vacuum, kB is Boltzmann's constant, and T is the absolute temperature [38]. A negative slope from all samples implied a p-type semiconductor. The extrapolation of slopes from the M-S plot (Figure 7) when integrated into the gradient of Equation (1) gives the NA. The interception of slopes at x-axis represents the Vfb. In this case, the calculated value of NA was 1.24 × 10 11 and 3.99 × 10 16 for 450 °C and 550 °C samples.

Mott-Schottky (M-S) Plot
M-S Analysis was done to study the charge carrier concentration of samples. The implementation of M-S analysis on annealed samples of 450 • C and 550 • C was performed in H 2 SO 4 (pH 1). The charge carrier was analyzed from the flat band, wherein flat band is the potential at which the potential drop between the electrode surface and the bulk is zero. This can establish the position of semiconductor energy band with respect to the redox potential. The intrinsic value of flat band (V FB ) and carrier density (N A ) for both samples can be extracted from the x-intercept and slope of the plot between the reciprocal of the square 1/C 2 and the bias potential (Figure 7c), in accordance with the Mott-Schottky equation, as follows (Equation (1)) [37] 1 where e is the electronic charge, ε is the relative permittivity of CoO (12.9), ε o is the permittivity of vacuum, K b is Boltzmann's constant, and T is the absolute temperature [38]. A negative slope from all samples implied a p-type semiconductor. The extrapolation of slopes from the M-S plot ( Figure 7) when integrated into the gradient of Equation (1) gives the N A . The interception of slopes at x-axis represents the V FB . In this case, the calculated value of N A was 1.24 × 10 11 and 3.99 × 10 16 for 450 • C and 550 • C samples. However, the valence band (V B ) of the samples was determined using the following equation (Equation (2)) where N V is the effective density of states in the valence band (V B ). The N V can be expressed as follows (Equation (3)) Incorporating the band gap energy (Eg) determined earlier (Figure 7), the conduction band (C B ) of the samples can be determined [39]. The analyzed data of V FB , V B , Eg, and C B are arranged as depicted in Figure 8. For this purpose, the potential measurements in Ag/AgCl were converted to the RHE using the following equation (Equation (4)) where E 0 Ag/AgCl = 0.197 V at 25 • C. To electrochemically split the water molecule, the band edges of the photoelectrode need to be in straddle position to the water molecule redox potential, which means the C B of the material should lie at a more negative potential than the water reduction potential. Meanwhile, the V B position should lie at a more positive potential to the water oxidation potential. Based on the calculation, the resultant band edges of C B and V B of both samples are as illustrated in Figure 8. The energy band structure of sample prepared at 450 • C (Co 3 O 4 ) is shown in Figure 8; the C B and V B are −1.04 eV and 0.98 eV. The band edges in this study were found at the more negative potential of our previous study; the C B and V B of Co 3 O 4 are estimated to be −0.55 and 1.10 eV, respectively [12]. Both the C B and V B located were at a more negative potential of H + to H 2 and oxidation potential of H 2 O to O 2 , respectively [35]. A similar trend was also found in the sample prepared at 550 • C. Given that the bandgap of the sample prepared at 550 • C was bigger than the sample prepared at 450 • C, the C B position falls at more negative potential than the 450 • C sample. Meanwhile, the V B position is not much different. This result showed that both cobalt oxides resulting in this study do not straddle to the redox potential of water splitting, but it can be used to produce hydrogen only [40].

Conclusions
This work provided the necessary support for the direct growth of cobalt oxide on FTO thin films, the effect of annealing on the phase transformation of cobalt oxide, and the effect of electrolyte on the PEC water-splitting application. The annealing temperature affected the phase of cobalt oxide thin films and changed the morphology from microflower Co3O4 to urchin-like Co3O4/CoO mixed phase. A single phase of CoO could not be produced by direct growth on FTO through the hydrothermal method due to the sensitivity of FTO to temperatures higher than 550 °C. In addition, the electrolyte affected the performance of PEC water splitting. H2SO4 resulted in a higher photocurrent density than Na2SO4 and KOH. According to the measured photocurrent generation, the highest value was contributed by the single-phase sample of Co3O4 from the annealed at 450 °C. The stability test also showed that the Co3O4 from the annealed at 450 °C is more stable than the mixed-phase produced by the annealed at 550 °C. This study found that the bandgap and band edges of the cobalt oxide thin films produced from the direct hydrothermal method slightly different from the powder results, in which powder cobalt oxide produces a more appropriate band structure to the PEC water-splitting application.

Conclusions
This work provided the necessary support for the direct growth of cobalt oxide on FTO thin films, the effect of annealing on the phase transformation of cobalt oxide, and the effect of electrolyte on the PEC water-splitting application. The annealing temperature affected the phase of cobalt oxide thin films and changed the morphology from microflower Co 3 O 4 to urchin-like Co 3 O 4 /CoO mixed phase. A single phase of CoO could not be produced by direct growth on FTO through the hydrothermal method due to the sensitivity of FTO to temperatures higher than 550 • C. In addition, the electrolyte affected the performance of PEC water splitting. H 2 SO 4 resulted in a higher photocurrent density than Na 2 SO 4 and KOH. According to the measured photocurrent generation, the highest value was contributed by the single-phase sample of Co 3 O 4 from the annealed at 450 • C. The stability test also showed that the Co 3 O 4 from the annealed at 450 • C is more stable than the mixed-phase produced by the annealed at 550 • C. This study found that the bandgap and band edges of the cobalt oxide thin films produced from the direct hydrothermal method slightly different from the powder results, in which powder cobalt oxide produces a more appropriate band structure to the PEC water-splitting application.

Conflicts of Interest:
The authors declare no conflict of interest.