1. Introduction

Since the 1970s, PEC cells made of various semiconductors and oxides have been used to generate hydrogen. Honda and Fujishima used TiO₂ as working electrodes to split water into hydrogen and oxygen under ultraviolet light illumination [1,2]. However, several problems are encountered during the PEC process. One of the materials, such as GaAs, is easily corroded in acidic or alkaline solution [3,4]. There is also a limited UV component in the terrestrial solar spectrum which leads to poor solar-to-hydrogen conversion efficiency [5]. The band-edge potentials of In_xGa_1-xN-based materials not only satisfy the conditions for splitting water [6], but they are also potentially resistant to aqueous solutions [7]. Therefore, the use of PEC cells with working electrodes made of In_xGa_1-xN-based materials is a promising approach to convert solar power to hydrogen gas. This is because In_xGa_1-xN materials with a band gap ranging from 0.7 to 3.4 eV can fit most terrestrial solar spectra [8,9]. In the present study, PEC cells with working electrodes made of n-GaN epitaxial layers grown on sapphire were studied. Although the use of n-GaN working electrodes facilitates the splitting by PEC cells of water into hydrogen and oxygen at zero bias, the photocurrent density is not high enough to produce adequate hydrogen gas within a short time [10]. In fact, a moderate bias on the PEC cells is necessary to obtain a higher photocurrent density and faster hydrogen gas generation rate. Previous studies have shown that a proper arrangement of metal contacts on the working electrodes is beneficial to hydrogen generation from water splitting [6]. Waki et al. indicated that a patterned n-GaN epitaxial layer formed through selective area regrowth with metal stripes can be employed for direct water photoelectrolysis [6]. The metal stripes functioned to eliminate current crowding, thereby improving the photocurrent. However, the metal stripes were directly immersed in the electrolyte. Therefore, they were increasingly eroded during the electrochemical reactions. Our previous studies showed that working electrodes with finger-type ohmic contacts made of Cr/Au bilayer metal, as well as the metal contacts, were covered with SiO₂ protection layer to protect the metal from direct contact with the electrolyte [11]. However, the life span of the immersed metal ohmic contacts remains inadequate. The opaque metal deposited on the semiconductor surface also leads to a reduction in light incident in the GaN layer. Indium tin oxide (ITO) is a well-known transparent and conductive optical film. ITO/n-GaN contacts exhibit ohmic property when the carrier concentration of n-GaN is close to 1 × 10¹⁹/cm³ [12,13]. In the present study, ITO was used as immersed finger-type ohmic contacts on n-GaN working electrodes. The ITO film can be easily eroded in aggressive aqueous solutions, such as HNO₃ and NaOH [14], so the electrolyte used was 1M NaCl aqueous solution, which is not an aggressive aqueous solution for most oxide materials. Further, the degradation of ITO in the electrolyte mostly occurs during cathodic polarization [14].

2. Experiments

Si-doped GaN (n-GaN) epitaxial layers with a carrier concentration of ~1 × 10¹⁹/cm³ were grown on (0001) sapphire substrates. To clarify whether the immersed finger-type ITO ohmic contacts can lead to an enhancement in photocurrent density and hence the gas generation rate, two different n-type GaN working electrodes were designed for the PEC experiments. One was GaN working electrode with finger-type ITO ohmic contacts covered with SiO₂ layer to protect the ohmic contacts from contact with the electrolyte and thus generate leakage current (labeled as PEC-1), as shown in Fig. 1(a) . For comparison, GaN working electrodes with ITO ohmic contacts aside from the working area (labeled as PEC-2) were also prepared, that is, the ITO ohmic contacts were not immersed in the electrolyte, as shown in Fig. 1(b). For PEC-1, the width of the ITO stripe contacts was 20 μm, and the spacing between stripes was 200 μm. The schematic diagram and the photograph of an area on the PEC-1 are shown in Fig. 1(a) and 1(c), respectively. The inset of Fig. 1(c) shows the schematic structure of the ITO ohmic contact and SiO₂ protection layer stacked on the GaN layer. Further, the results for two other samples, PEC-3 and PEC-4, were demonstrated for comparison with PEC-1 and PEC-2. Similar to PEC-1, PEC-3 had an immersed finger-type ohmic contact. However, the ohmic contacts were made of a bilayer Cr/Au metal insted of ITO. The structure of PEC-4 was similar to PEC-2 but the ohmic contacts onto n-GaN were bilayer Cr/Au metal insted of ITO. Potentiostat (Autolab-PGSTAT128N) was used to supply the external bias and to measure the current density for the evaluation of the electrical properties of the PEC cells. A platinum (Pt) wire was used as the counterelectrode. A 300 W Xe lamp was used as light source, and 1 mol/L NaCl was used as the electrolyte. In this study, the PEC cells were constructed to be similar to the Hoffman apparatus for measuring the total volume of generated gases. The composition of generated gases was analyzed by a gas chromatograph (Agilent-6850). Based on the total volumes and composition of the generated gases, we thus could get the volumes of H₂ and N₂ gases generated from the PEC process.

 

Fig. 1 Schematic diagram of the photoelectrochemical cells (a) PEC1, (b) PEC2 (c) picture taken at a local area of PEC1. The inset of Fig. 1(c) shows the schematic structure of ITO/SiO₂ staked layers on the GaN.

Download Full Size | PDF

3. Results and discussions

Figure 2 shows the photocurrent density as a function of external bias voltage (V_CE). V_CE was the applied voltage between the n-GaN working electrode and the Pt counterelectrode. As shown in Fig. 2, the photocurrent density of PEC-1 is higher than that of PEC-2. The enhancement in photocurrent density in PEC-1 can be attributed to the distance between neighboring ITO ohmic contacts, which is small enough to increase the collection efficiency of the photo-generated electrons by the ITO ohmic contacts. In short, the photo-generated electrons can reach the ITO ohmic contacts before the electrons recombine with holes or charged defects in the GaN layer. Compared with PEC-1, the recombination of photo-generated carriers with one another or the charged defects is more likely to occur before the carriers reach the ohmic contact of PEC-2. The photocurrent densities of PEC-1 were at least 50% higher than those of PEC-2. This result can be attributed to the photo-generated electrons in PEC-1 which entail less transit time compared with PEC-2 before they reach the ohmic contacts, whereas the transit time is inversely proportional to the probability of recombination of the electrons with holes or charged defects. In PEC-2, the applied fields are difficult to distribute uniformly over the entire working area, thereby leading to a longer transit time of the electrons compared with PEC-1. Similarly, PEC-3 had a Cr/Au finger ohmic contact on the n-GaN working electrodes, so the photocurrent densities of PEC-3 were higher than those of PEC-4. Notably, the photocurrent densities of PEC-1 were slightly higher than those of PEC-3. This result can be attributed to the transparent ITO film that allowed more incident photons to reach the GaN layer compared with PEC-3 with opaque metal contacts (Cr/Au) on the n-GaN surface. To examine the stability of ITO finger contacts, the photocurrent was measured for one hour with V_CE = 0.8V applied to the samples. Hydrogen gas at the Pt counter electrode was also collected. As shown in Fig. 3 , the photocurrent density of PEC-1 and PEC-3 was higher than that of PEC-2 and PEC-4 because the former had immersed ITO finger-type and Cr/Au finger-type ohmic contacts on the GaN surface, respectively. The photocurrent density of PEC-1 was slightly higher than that of PEC-3. This result was consistent with the bias-dependent photocurrents, as shown in Fig. 2, and can be attributed to the transparent characteristic of the immersed ITO finger ohmic contacts. To evaluate the stability of the contacts, images taken from the surface of the working electrodes using scanning electron microscope (SEM) and optical microscope (OM) are shown in Fig. 4 . Figures 4(a) and 4(b) correspond to PEC-1 before and after PEC measurement, respectively. The insets of Figs. 4(a) and 4(b) show the top-view of the OM images. Figures 4(c) and 4(d) correspond to PEC-3 before and after PEC measurement, respectively. The insets of Fig. 4(c) and 4(d) show the top-view of the OM images. No observable change can be seen from the ITO fingers after PEC measurements. By contrast, a part of the Cr/Au finger contacts was lifted off the n-GaN layer after the PEC measurements, as shown in the inset of Fig. 4(d). Although the photocurrent of PEC-3 was almost the same as that of PEC-1, as shown in Fig. 2 and Fig. 3, the poor stability of the contacts in the working electrodes of PEC-3 led to a reduction in the life span of the PEC cells. Based on these SEM images, the Cr/Au metal layer was peeled off the GaN layer. This result can be attributed to the relatively poor adhesion of Cr/Au on GaN. As a result, the OH^- ions might diffuse through the SiO₂ layer into the GaN/Cr/Au interface where the PEC process occurred. The PEC reactions at the GaN working electrode are both water oxidization and GaN decomposition, simultaneously [8]. The reactions occurred at the GaN/Cr/Au interface resulted in the Cr/Au metal contacts to peel off the GaN layer. This is probably the reason why the Cr/Au metal layer did not resist the corrosion of the electrolyte during the PEC process. However, the ITO film was firmly cohered on the GaN layer, as shown in Fig. 4(a) and 4(b), even if the ITO/n-GaN working electrodes were immersed into the NaCl electrolyte for a long time. There are two main issues in hydrogen generation by PEC water splitting. One is how to improve the hydrogen generation rate, and another one is how to prevent the photoelectrodes from the oxidation reactions. It had been revealed that the addition of CH₃OH into the electrolyte could prevent the GaN from oxidation [15]. In this study, we focused on the immersed finger-type ITO contacts to improve the hydrogen generation rate.

 

Fig. 2 Typical photocurrent density-bias curves of the experimental PEC cells.

Download Full Size | PDF

 

Fig. 3 Photocurrent densities as a function of time when the PEC cells were biased at V_CE = 0.8 V.

Download Full Size | PDF

 

Fig. 4 SEM images of the cross-section of PEC1 (a) before and (b) after photoelectrochemical measurements. (c) and (d) are the SEM images of the cross-section of PEC3 before and after photoelectrochemical measurements, respectively.

Download Full Size | PDF

Table 1 shows the gas generation rate and the total reaction charge of the four samples. The gas generation rates of PEC-1 and PEC-3 were both higher than those of PEC-2 and PEC-4. In addition, the gas generation rate of PEC-1 was slightly higher than that of sample PEC-3. Therefore, the gas generation rates obtained from the fabricated samples were consistent with the results on photocurrent density (shown in Fig. 3). In theory, the total charges of the electrochemical reaction are proportional to the product (i.e., hydrogen). The percentages of hydrogen in the collected gas might also be of interest. Based on the analysis of the gas chromatograph, the collected gas included 67% H₂ and 33% N₂. Nitrogen gas primarily came from the etching reaction that occurred at the n-GaN/electrolyte interface during the PEC process. As shown in Table 1, the gas generation rate is indeed proportional to the amount of the total reaction charge, indicating that PEC-1 and PEC-3 can generate more hydrogen gas than PEC-2 and PEC-4. However, whether the immersed ITO finger contacts in PEC-1 contributed photocurrent remains to be elucidated. Theoretically, the conduction band edge of ITO is assumed to be more positive than the reduction level of the hydrogen ion [16,17]. Therefore, it cannot split water under illumination without any applied electrical potential. Huang et al. reported that in the scanning of potentials ranging from −1.0 to 1.5 V, after the completion of reduction current peak, a reduction in hydrogen ions does not occur. In other words, the reduction of hydrogen ions can be observed in potentials lower than −1.0 V [18]. The voltages from −0.4 to 2 V were applied in the present study. Therefore, the immersed ITO finger contacts cannot generate hydrogen in this bias region. To clarify this point further, PEC cells using ITO films and 1M NaCl aqueous solution as the working electrodes and electrolyte, respectively, were also prepared for measuring the photocurrents. The ITO films with a thickness of 280 nm were deposited on a sapphire substrate. The photocurrents taken from the PEC cells were lower than those of PEC-1 and PEC-2. This result indicates that the reaction rate of oxidation in the ITO contacts was extremely low. In short, the ITO contacts in the 1M NaCl aqueous solution under bias and UV illumination were resistant to erosion.

Table 1. The Results of Solar-to-Hydrogen Conversion Efficiency, Gas Generation Rate, and Total Reaction Charge at V_CE = 0.8 V

View Table

In addition to the adhesion issue, this phenomenon shows that the ITO/n-GaN working electrodes are more stable than the Au/Cr/n-GaN electrodes. Therefore, the use of ITO films associated with GaN-based materials for the formation of working electrodes is a promising method to generate hydrogen by photoelectrolysis from water.

4. Conclusions

The photocurrent density and hydrogen generation rate of n-type GaN photoelectrodes, with and without immersed finger-type ITO (or Cr/Au) ohmic contacts, were symmetrically compared. At an external bias of 0.8 V, the n-GaN working electrodes with ITO finger contacts exhibited a 25% enhancement in gas generation rate compared with those with Cr/Au finger contacts. The marked enhancement in photocurrent can be attributed to the transparent ITO contacts which facilitated the introduction of relatively more photons into the GaN layer. Further, no significant corrosion was observed in the ITO layer after the PEC process compared with the Cr/Au finger contacts which were significantly peeled from the GaN layer. These results indicate that the use of n-GaN working electrodes with finger-type ITO contacts is a promising approach for PEC cells.