ARTICLE IN PRESS
Solar Energy Materials & Solar Cells 90 (2006) 2429–2437 www.elsevier.com/locate/solmat
Photo-electrochemical properties of oxide semiconductors on porous titanium metal electrodes Kazuhiro Sayamaa,, Takashi Oib, Ryu Abea, Masatoshi Yanagidaa, Hideki Sugiharaa, Yasukazu Iwasakib a
Energy Technology Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), Central 5, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan b Nissan Research Center, Nissan Motor Co., LTD., 1 Natsushima, Yokosuka, Kanagawa 237-8523, Japan Available online 19 April 2006
Abstract For the purpose of producing hydrogen using solar energy, we investigated the potential of porous titanium metal sheet (PTMS) with high surface area for use as the basal plates for various types of oxide semiconductor photo-electrodes. The TiO2 photoelectrodes were prepared by oxidation of PTMS and flat titanium metal sheet (FTMS). The photocurrents of the TiO2/PTMS electrodes were always higher than TiO2/FTMS under the same oxidation conditions. The reflectance of PTMS was lower than FTMS over the entire wavelength spectrum, suggesting that the scattered light was absorbed more effectively on the former. A nanocrystalline WO3 layer-loaded PTMS electrode (WO3/PTMS) showed a high photocurrent compared to WO3/FTMS, suggesting that PTMS is highly suitable as basal plates for semiconductor photoelectrodes. r 2006 Elsevier B.V. All rights reserved. KeyWords: Photoelectrode; Porous titanium sheet; TiO2; WO3; Semiconductor
1. Introduction Since the report of the Honda–Fujishima effect on TiO2 photoelectrodes [1,2], hydrogen production using solar energy has been widely investigated using various types of oxide semiconductor photoelectrodes. In the early stages, single crystal, oxidized metal plates and sintered pellets made from oxide powders were studied for conventional photoelecCorresponding author. Fax: +81 29 861 4760.
E-mail address:
[email protected] (K. Sayama). 0927-0248/$ - see front matter r 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.solmat.2006.03.028
ARTICLE IN PRESS 2430
K. Sayama et al. / Solar Energy Materials & Solar Cells 90 (2006) 2429–2437
trodes [3–10]. Recently, it has been reported that nano crystalline and porous semiconductor photoelectrodes such as TiO2, WO3, Fe2O3 and BiVO4 on conducting glass basal plates showed excellent incident photon-to-current conversion efficiencies (IPCE) for water decomposition into H2 and O2 under external bias [11–20]. This porous thin film structure with high surface area is considered important for efficient water splitting reactions using solar energy. Generally, basal plates for nanocrystalline semiconductor electrodes are made of conducting glass. When light is irradiated through the conducting glass into the nanocrystalline semiconductor film, the diffusion length of electrons towards the back contact is short. However, the conductivity of conducting glass is not sufficient for practical enlargement of electrode area, and it breaks easily. There were some reports on the nanocrystalline TiO2 film on basal plates of flat titanium metal [21]. In the present study we investigated the possible use of porous titanium metal sheet (PTMS) as basal plate for nanocrystalline semiconductor electrodes. PTMS has a wider conductive surface area to carry semiconductor particles compared to flat conducting glass or flat titanium metal sheet (FTMS). PTMS is highly conductive, thin, light and flexible, and allows electrolyte to penetrate through it to the backside.
2. Experimental 2.1. Preparation of semiconductor electrodes Both PTMS and FTMS were mainly provided from Sumitomo Titanium Co. The PTMS was prepared by the doctor’s blade method using spherical titanium metal powder (average diameter: 25 mm) nd binders, and heated in vacuum [22]. Fig. 1 shows top view of scanning electron microscope (SEM) photograph of PTMS. The average pore size was 10 mm, and the porosity was 37%. The pores penetrate from the front to the back. The Ti metal spheres were melted at the boundaries, and interfacial discordance was not observed. The sheet resistances of both PTMS and FTMS were very small (o10 mO/sq). A home-made
Fig. 1. SEM micrographs of porous titanium metal sheet (PTMS).
ARTICLE IN PRESS K. Sayama et al. / Solar Energy Materials & Solar Cells 90 (2006) 2429–2437
2431
PTMS was also prepared from titanium metal powder (Nilako Co., powder particle size below 45 m) by pressing at 20 MPa at room temperature. The TiO2 photoelectrodes were prepared by oxidation of the metal sheets in air at 400–800 1C for 1 h (TiO2/PTMS and TiO2/FTMS). The WO3 colloid was prepared by a modified method from tungstic acid in polyethylene glycol solution (PEG) following a procedure detailed elsewhere [11,12]. The nanocrystalline WO3 film electrodes (WO3/ PTMS and WO3/FTMS) were prepared by spreading a portion of solution on various basal plates and annealing at 500 1C for 1 h. 2.2. Characterization The photo-electrochemical measurements were conducted using a potentiostat (BAS, 600) and a Pyrex glass cell. A Pt wire and an Ag/AgCl electrode (+0.21 V vs. NHE, pH ¼ 0) were used as the counter and reference electrodes, respectively. For the electrolyte solutions, 0.5 M aqueous solutions of Na2SO4 or H2SO4 were used as for the TiO2 or WO3 photoelectrodes, respectively. The light source was a xenon lamp (500 W) used with or without various filters. The Xe light power in the UV region (wavelengtho400 nm) without filter was ca. 4.3 times stronger than the standard solar light (AM-1.5G, 100 mW/ cm2, Japan Industrial Standard) in the same wavelength region (4.3 sun condition in UV region). The area irradiated using a mask was 0.28 cm2. 3. Results and discussion 3.1. TiO2 photoelectrodes prepared by oxidation of titanium metal sheets. Table 1 shows the relationship between the photocurrent and the preparation temperature of various electrodes. The photocurrents measured for TiO2/PTMS prepared by calcination of titanium metal sheets in air were always higher than TiO2/FTMS under the same oxidation conditions. The photocurrent of PTMS calcined at 700 1C (1.43 mA/ cm2) was 3.7 times higher than that at 400 1C (0.39 mA/cm2). The best efficiency for both TiO2/PTMS and TiO2/FTMS was obtained at 700 1C oxidation for 1 h. The IPCE of the TiO2/PTMS was ca. 50% at 360 nm. The photocurrent decreased significantly at 800 1C. The anodic dark current was very low at less than +1.0 V (vs.Ag/AgCl). The anodic photocurrent, observed under UV light with an applied potential, increased with increasing applied positive potential. The evolution of H2 gas bubbles was observed on the Pt counter electrode. We also confirmed that the photocurrents of FTMS provided by another company (Niraco Co.) as well as a homemade PTMS prepared by the press method from Ti powder, were similar to those of standard titanium metal sheets (FTMS and PTMS from Sumitomo Titanium Co.) (Table 1,c,d)). The results clearly suggest that the porous and undulated structure contributed to the high photocurrent efficiency. Fig. 2 shows the SEM photographs of TiO2/PTMS oxidized at 700 and 800 1C. The thickness of the oxide layer of TiO2/PTMS at 700 1C, estimated from the contrast micrograph, was ca. 0.6–1 mm, and the network of metal connections for the passage of electrons was maintained. On the other hand, the thickness of the oxide layer at 800 1C increased to 3–5 mm, and thus the metal network was disconnected (Fig. 2(b)). Peeling off of the oxide film from the metal interface was also observed in a few places. High resolution measurement of the cross-section of the TiO2/PTMS layer at 800 1C shows
ARTICLE IN PRESS 2432
K. Sayama et al. / Solar Energy Materials & Solar Cells 90 (2006) 2429–2437
Table 1 Photocurrent of the titanium metal sheets calcined at various temperatures Calcination temperature (1C)
Ti metal sheet
Photocurrenta (mA/cm2)
Oxide film thicknessb (mm)
Photocurrent ratio of PTMS/FTMS
400
PTMS FTMS PTMS FTMS PTMS FTMS PTMS FTMS Pressed-PTMSc Nilaco-FTMSd Etched-FTMSe PTMS FTMS
0.39 0.25 0.43 0.28 1.14 0.34 1.43 0.53 1.65 0.51 1.07 0.08 0.05
– – 0.1–0.2 0.1–0.2 0.2–0.3 0.2–0.3 0.6–1 0.6–1
1.6
500 600 700
800
3–5 3–5
1.6 3.4 2.7
1.6
a
0.8 V vs. Ag/AgCl. Measured on SEM micrograph of cross-section. c Prepared by pressing from Ti metal powder. d Provided by Nilaco Co. e FTMS etched by acidic solution with mask, and then calcined. b
numerous pores in the oxide layer, with especially large ones located at the metal-oxide interface. The oxide film easily peeled off at the interface and the large pores may inhibit the electron transfer from the oxide layer to the metal. The thicknesses of the oxide layers on FTMS oxidized at 700 and 800 1C were similar to that of TiO2/PTMS. Many pores and cracks were also observed at the metal-oxide interface at 800 1C. In the XRD patterns of the TiO2/PTMS electrodes calcined at various temperatures, only the metal peaks were observed below 500 1C, and a small shoulder peak at 39.91, assigned to Ti2O, was observed at 600 1C. Three phases, namely Ti metal, Ti2O and TiO2 (rutile) were present at 700 1C. At 800 1C the metal peaks disappeared, and the main phase was rutile. The XRD patterns of TiO2/FTMS were similar to those of TiO2/PTMS. The anatase phase of TiO2 was not observed in any of the electrodes. The XRD pattern of Ti2O was very similar to that of Ti metal, with a slight difference in lattice constant. We believe that the presence of Ti2O has a role of connecting between the titanium metal and the TiO2 layer. The decrease in photocurrent at high calcination temperatures could be attributed to the disconnection of the metal network and the peel-off of the oxide layer, as mentioned above. Moreover, the thickness of the oxide layer is also important. A thick oxide layer has an advantage in light absorption, and therefore, photocurrent increased with calcination temperature up to 700 1C. On the other hand, electrons have a longer diffusion path to the metal in the thicker oxide layer, and this may cause a decrease in photocurrent due to the increased possibility of charge recombination. The light absorption and the large surface area could explain the larger photocurrent of TiO2/PTMS compared to TiO2/FTMS. Fig. 3 shows the reflectance of TiO2/PTMS and TiO2/FTMS calcined at 700 1C measured using integrating sphere to gather the scattered
ARTICLE IN PRESS K. Sayama et al. / Solar Energy Materials & Solar Cells 90 (2006) 2429–2437
2433
Fig. 2. SEM micrographs of the TiO2/PTMS interface oxidized at: (a) 700 1C; (b) 800 1C.
40
Reflectance (%)
35 30
(b)
25 20 15 10
(a)
5 0 300
350
400 wavelength (nm)
450
500
Fig. 3. The reflectance of TiO2/PTMS (a) and TiO2/FTMS (b) calcined at 700 1C measured using an integrating sphere to gather the scattered light.
ARTICLE IN PRESS 2434
K. Sayama et al. / Solar Energy Materials & Solar Cells 90 (2006) 2429–2437
light. The reflectance of TiO2/PTMS calcined at 700 1C was always lower than that of TiO2/FTMS in all wavelength region for all electrodes. We infer that the incident photons were effectively collected by the porous and undulated surface structure, because the incident light was multi-scattered in the hollow regions. The maximum photocurrent was obtained on the electrodes calcined at 700 1C, whereas the maximum photocurrent ratio of PTMS/FTMS (3.4) was obtained at 600 1C (Table 1). It is suggested that the photocurrent conversion efficiency of PTMS is adequate even at lower temperature with thinner oxide thickness. The incident angle of photon was almost perpendicular in the case of the FTMS, on the other hand, the incident angles were varied in the case of the PTMS. The charge carriers are generated near the surface by the slanting irradiation, therefore, the diffusion length of the minority carrier to the surface of the PTMS is shorter than that of the FTMS. The short distance over which the minority carriers need to diffuse might contribute to the high IPCE. Furthermore, it might be one of important factors that PTMS has a higher surface area than FTMS. The densities of charge carriers per surface area decrease with increasing the surface area, therefore, the possibility for charge recombination on the TiO2/PTMS electrode might decrease. A similar behavior was also reported in the case of titanium oxide electrodes with high surface area prepared from titanium– aluminum alloy [23]. In order to observe the effect of an undulated surface structure, such a structure was prepared by etching FTMS with acidic solution with a photomask. The peak-to-peak distance was 70 mm, and the depth was ca. 25 mm. The photocurrent of TiO2/FTMS after etching and calcination at 700 1C was higher than without etching (Table 1e). The results suggest that the undulated surface structure has a beneficial effect on the photocurrent due to the effective light absorption and the slanting incident angle, similar to TiO2/PTMS. Khan et al. reported a very efficient carbon-doped TiO2 electrode using FTMS as the basal plate [24]. It is considered that the efficiency of the carbon-doped TiO2 electrode will be improved by using the undulated FTMS after etching or the PTMS as the basal plate. 3.2. Photo-electrochemical properties of PTMS covered with nanocrystalline WO3 particles The band gap of the WO3 semiconductor (2.7 eV) is smaller than that of TiO2 (ca. 3.0 eV), and WO3 can utilize visible light more effectively. A highly efficient WO3 photoelectrode using conducting glass as the basal plate was reported [11,12]. When we use PTMS as the basal plate for a WO3 electrode, the electron transfer from WO3 to Ti metal may be obstructed if a thick TiO2 layer forms between WO3 and the Ti metal. We investigated various titanium metal sheet electrodes covered with WO3 nano particles. Fig. 4 shows SEM photographs of WO3/PTMS electrode calcined at 500 1C. The WO3 particles were small disks, ca. 30 nm in thickness and 200 nm in diameter; they were stacked and covered the PTMS surface. The WO3 colloidal solution was viscous; therefore, the WO3 layer was present mainly at the outer surface of PTMS. The thickness of the WO3 layer after calcination at 500 1C was around 3 mm on average, and it was thicker in the hollow space of PTMS. The photocurrent increased significantly by covering with WO3, as shown in Table 2 (runs (3) and (4)). The photocurrent of WO3/PTMS (run (4)) was 4.3 times higher than that of run (2). The TiO2 electrodes could not utilize visible light, but photo-response under visible light was clearly observed with the WO3 particles. The results suggest that the electron transfer from WO3 particles to the titanium metal was effective. On the other
ARTICLE IN PRESS K. Sayama et al. / Solar Energy Materials & Solar Cells 90 (2006) 2429–2437
2435
Fig. 4. SEM micrographs of the WO3/PTMS electrodes calcined at 500 1C.
Table 2 Photocurrent of various titanium metal sheet electrodes covered with WO3 particles Run
(1) (2) (3) (4) (5)
Photoelectrodea
FTMS-500 1C PTMS-500 1C FTMS-WO3 coating-500 1C PTMS-WO3 coating-500 1C PTMS-500 1C-WO3 coating-500 1C
Photocurrent (mA/cm2)b UV+visible lightc
visible lightd
0.29 0.43 0.71
o0.01 o0.01 0.11
1.82
0.42
0.43
0.04
a
Calcination at 500 1C for 1 h. At 0.8 V vs. Ag/AgCl. c Without any filter. d With UV cut filter (l (wavelength)4420 nm). b
hand, regarding the WO3 coverage after calcination of the PTMS (run (5)), the photocurrent was not high compared to run (4). An oxide layer, ca. 150-nm thick, was observed on the surface of both PTMS and FTMS after calcination at 500 1C (runs (1) and (2)). The XPS measurements indicate that the ratio of Ti:O was the same as standard TiO2 (atomic ratio of Ti/O ¼ 2); therefore, the surface after calcination at 500 1C was TiO2, not Ti2O or TiO2x. The surface resistance after 500 1C calcination was very high (420 kO/sq). Pure TiO2 is an insulator, and its conduction band potential is more negative than that of WO3 [4]. Therefore, the electrons in the WO3 conduction band cannot transfer to the titanium metal over a thick and pure TiO2 layer. The low photocurrent of run (5) shows that the presence of the TiO2 layer significantly influenced the electron transfer from WO3.
ARTICLE IN PRESS 2436
K. Sayama et al. / Solar Energy Materials & Solar Cells 90 (2006) 2429–2437
In the case of runs (3) and (4), a titanium oxide layer was not observed on the titanium metal surface with SEM. But a very thin oxide layer (ca. 25 nm) was observed with high resolution TEM. The titanium oxide layer on WO3/PTMS after calcination at 500 1C was 6 times thinner than on PTMS without WO3 after calcination at 500 1C, suggesting that the presence of the WO3 layer significantly prevented the formation of the titanium oxide layer. The composition of the thin oxide layer (pure TiO2, Ti2O or TiO2x) is not fully understood. But it is speculated that it might not be TiO2 because the contrast of the thin layer looked different from that of pure TiO2 layer. If the oxide layer on WO3/PTMS has many oxygen defects (Ti2O orTiO2x) due to the shortage of oxygen supply through the WO3 layer, we speculate that the electron transfer from WO3 to titanium metal may take place effectively through the Ti2O or TiO2x layer having a high conductivity. The photocurrent of WO3/PTMS was higher than that of WO3/FTMS (Table 2). The WO3 particles located in the hollow space may absorb the scattered light effectively. Moreover, the adhesion of the WO3 layers to the titanium sheet is very important for photocurrent efficiency. In the case of the WO3/FTMS, partial cracks at the titanium surface were observed by SEM, and the WO3 layer easily peeled off by scratching with a finger. On the other hand, scratching or bending of the sheet could hardly peel-off the WO3 layer on PTMS, indicating that the undulated and porous surface of this metal sheet contributed to the strong adhesion of the WO3 semiconductor layer. 4. Summary and conclusions For TiO2 photoelectrodes prepared by calcination of titanium metal sheets in air, the photocurrents on PTMS were always higher than on FTMS under the same oxidation conditions. The best efficiency (IPCE ¼ 50% at 360 nm) was obtained for 700 1C oxidation for 1 h. The reflectance of TiO2 on PTMS was lower than on FTMS over the whole wavelength spectrum, suggesting that the scattered light was absorbed effectively on PTMS. The WO3/PTMS also showed a high photocurrent compared to the WO3/FTMS. The presence of the TiO2 layer had a significant influence on the electron transfer from WO3. All the results in this study suggested that PTMS is very suitable for use as basal plate for semiconductor photoelectrodes. References [1] A. Fujishima, K. Honda, Nature 238 (1972) 37. [2] A. Fujishima, K. Honda, S. Kikuchi, J. Chem. Soc. Japan 72 (1969) 108. [3] M. Wrighton, A. Ellis, P. Wolczanski, D. Morse, H. Abrahamson, D. Ginley, J. Am. Chem. Soc. 98 (1976) 2774. [4] H. Maruska, A. Ghosh, Sol. Energy 20 (1978) 443. [5] A. Ghosh, H. Maruska, J. Electrochem. Soc. 124 (1977) 1516. [6] M. Butler, D. Ginley, J. Mater. Sci. 15 (1980) 1. [7] J. Turner, M. Hendewerk, J. Parmeter, D. Neiman, G. Somorjai, J. Electrochem. Soc. 131 (1984) 1777. [8] K. Rajeshwar, P. Singh, J. Dubow, Electrochim. Acta 23 (1978) 1117. [9] Y. Matsumoto, M. Omae, K. Sugiyama, E. Sato, J. Phys. Chem. 91 (1987) 577. [10] G. Li, L. Bicelli, G. Razzini, Sol. Energy Mater. 21 (1991) 335. [11] C. Santato, M. Ulmann, J. Augustynski, J. Phys. Chem. B 105 (2001) 936. [12] C. Santato, M. Odziemkowski, M. Ulmann, J. Augustynski, J. Am. Chem. Soc. 123 (2001) 10639. [13] T. Lindgren, H. Wang, N. Beermann, L. Vayssieres, A. Hagfeldt, S. Lindquest, Sol. Energy Mater. Sol. Cells 71 (2002) 231.
ARTICLE IN PRESS K. Sayama et al. / Solar Energy Materials & Solar Cells 90 (2006) 2429–2437 [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
2437
G. Zhao, H. Kozuka, H. Lin, M. Takahashi, T. Yoko, Thin Solid Films 340 (1999) 125. Y. Wang, H. Cheng, Y. Hao, J. Ma, W. Li, S. Cai, Thin Solid Films 349 (1999) 120. S. Khan, J. Akikusa, J. Phys. Chem. B 103 (1999) 7184. K. Sayama, A. Nomura, Z. Zou, R. Abe, Y. Abe, H. Arakawa, Chem. Commun. (2003) 2908. S. So¨dergren, A. Hagfeldt, J. Olsson, S. Lindquist, J. Phys. Chem. 98 (1994) 5552. K. Sayama, J. Augustynski, H, Arakawa, Chem. Lett. (2002) 994. A. Wahl, J. Augustynski, J. Phys. Chem. B 102 (1998) 7820. A. Wahl, M. Ulmann, A. Carroy, J. Augustynski, J. Chem. Soc. Chem. Commun. (1994) 2277. T. Ogasawara, K. Tada, K. Azuma, T. Onishi, J. Jpn. Soc. Powder Metall. 49 (2002) 829. A. Kudo, M. Steinberg, A. Bard, A. Campion, M. Fox, T. Mallouk, S. Webber, J. White, J. Electrochem. Soc. 137 (1990) 3846. [24] S. Khan, M. Al-Shahry, W. Ingler Jr., Science 297 (2002) 2243.
