Structure and Photocatalytic Activity of PdCrO_x Cocatalyst on SrTiO₃ for Overall Water Splitting

Abstract

The mechanism of PdCrO_x multi-component cocatalyst formation on SrTiO₃ was investigated using transmission electron microscopy, X-ray absorption fine structure spectroscopy and X-ray photoelectron spectroscopy. The PdCrO_x/SrTiO₃ samples were synthesized by a photodeposition method under UV light irradiation (λ > 300 nm) for various time periods (0–5 h). The fine structure and valence state of the Pd species of PdCrO_x nanoparticles were varied from Pd oxide to a mixture of metallic Pd and oxidized Pd species with an increase in the irradiation time. The overall water-splitting activity of PdCrO_x was strongly dependent on the photoirradiation time during deposition. Although longer photoirradiation time during preparation did not influence the H₂ evolution activity of PdCrO_x/SrTiO₃ from aqueous methanol solution, it was effective in suppressing the O₂ photoreduction activity, which is one of the backward reactions during overall water splitting.

1. Introduction

Photocatalytic decomposition of H₂O into H₂ and O₂, using a semiconductor powder, is one of the target reactions of artificial photosynthesis in producing a clean energy source [1,2,3,4]. Many strategies have been proposed to improve the efficiency of this reaction, such as doping [5], morphological modification [6], application of visible-light-absorbing materials [7,8,9,10], cocatalyst loading [11,12,13,14,15,16,17] and surface protection [18,19]. With respect to surface modification, the loading of metal or metal-oxide nanoparticles, which are referred to as cocatalysts, onto the surface of a semiconductor photocatalyst has been widely employed to enhance the activity toward photocatalytic water splitting. The cocatalyst nanoparticle reduces the activation energy for reduction/oxidation reactions and suppresses charge recombination of electron and hole pairs in a semiconductor, thereby enhancing the activity.

Several cocatalysts that consist of more than two metal species have recently been reported for water splitting, and some of them have been more efficient than their single component counterparts [9,10,11,12,13,14]. Among these multi-component cocatalysts, mixed oxides of Cr and a paired transition metal have been well studied for overall water splitting. The metal-Cr mixed compounds were reported as promoters for not only the hydrogen evolution reaction (HER) but also the oxygen evolution reaction (OER). Whereas RhCrO_x was reported to work as a HER cocatalyst on the surface of various semiconductor photocatalysts [14,15], FeCrO_x was recently reported to function as an OER cocatalyst on SrTiO₃ [17]. Therefore, metal-Cr mixed compounds are considered to be promising cocatalyst materials for water splitting.

Our group recently reported PdCrO_x nanoparticles as a HER cocatalyst [16]. However, it was difficult to synthesize an active PdCrO_x cocatalyst using a conventional impregnation method, and thus required a photochemical method instead (see Scheme 1). The PdCrO_x/semiconductor photocatalyst system is interesting from a synthetic inorganic chemistry and functional materials chemistry perspective. However, compared to a similar mixed oxide cocatalyst such as RhCrO_x, the structural characterization of photochemically-synthesized PdCrO_x has yet to be completed, and thus the impact of the preparation conditions on the formation of PdCrO_x remains unknown.

In this work, the structure of PdCrO_x formed on the surface of SrTiO₃ with various irradiation times was characterized using high resolution transmission electron microscopy (HR-TEM), extended X-ray absorption fine structure spectroscopy (EXAFS), X-ray absorption near edge structure (XANES) and X-ray photoelectron spectroscopy (XPS).

2. Results and Discussion

2.1. Deposition of PdCrO_x Nanoparticles

Figure 1 shows a typical time course for H₂ evolution observed during the photodeposition of PdCrO_x. After an induction period (ca. 1 h), stable H₂ evolution (ca. 63.3 μmol h⁻¹) was observed. The induction period between 0–1 h suggests that PdCrO_x cocatalyst deposition was accomplished within 1 h. In the absence of the Pd and Cr sources, H₂ production was observed at a moderate rate (0.2 μmol h⁻¹). The amount of Cr deposited in the product (PdCrO_x/SrTiO₃) was quantified by monitoring the concentration of the Pd and Cr sources in the solution using UV-visible absorption spectroscopy. Figure 2 shows the amount of Cr deposited on SrTiO₃ as a function of the photoirradiation time. It is noted that all of the Pd precursor added (0.5 wt% Pd) underwent inclusion on the surface of SrTiO₃, regardless of the photoirradiation time. Therefore, the amount of Pd in deposited PdCrO_x/SrTiO₃ was identical. The amount of deposited Cr was increased with the irradiation time. Even without photoirradiation, 0.2 wt% portion of Cr disappeared from the solution phase, which suggests that the Cr species is adsorbed onto the surface of SrTiO₃. After 1–5 h irradiation, the amount of loaded Cr reached almost a constant value of 0.7–0.8 wt%, which indicated that Cr had been photodeposited.

Figure 3 shows TEM images of the deposited PdCrO_x nanoparticles. These observations indicated that the deposited PdCrO_x nanoparticles are not only core/shell type particles (core: Metallic Pd, shell: Cr₂O₃) but also aggregates and isolated metallic Pd nanoparticles. The ratio of the deposited Pd and Cr was also different in each particle. In general, the photodeposition rate of metal nanoparticles (in other words, the rate of photoreduction of metal ions) on the semiconductor surface depends on the surface property of the semiconductor (e.g., crystal faces), as reported by Ohno et al. [20]. Because SrTiO₃ particles used in this work have featureless morphology, the precise control of the Pd⁰ photodeposition would be difficult, leading to the compositional deviation in the Pd/Cr ratio. Metallic and oxidized Pd species were loaded on SrTiO₃, as evidenced by XPS measurements (discussed later). The lattice fringe attributable to metallic Pd was observed in the 5 h sample (see 5 h (1)). Therefore, the central part of the nanoparticle consisted of metallic Pd. However, the location of the oxidized Pd species in the TEM images could not be detected.

2.2. XANES

Figure 4 shows Pd K-edge XANES spectra of PdCrO_x/SrTiO₃ prepared for various time periods (0–5 h). Data for Pd foil, PdO, and Pd/SrTiO₃ (prepared without Cr) are also shown for comparison. The spectral shapes for Pd foil and Pd/SrTiO₃ were very similar. Compared to metallic Pd (Pd foil and Pd/SrTiO₃), the absorption edges of PdCrO_x/SrTiO₃ were varied with respect to the irradiation time (Figure 4A). The Pd K-edge positions are located at higher energies, which indicates that their electronic density is lower than metallic Pd. Figure 4B shows an enlarged view that shows that the K-edge positions are shifted to higher energies in the following order: Pd foil ≈ Pd/STO < 15 min < 5 h ≈ 1 h ≈ 30 min < 0 h < PdO. The K-edge of the 0 h sample was located at the highest energy position among all the samples examined. The K-edge position was shifted to the lower energy side with an increase of the irradiation time to 15 min, which indicates that the adsorbed Pd precursor on SrTiO₃ was reduced by photoexcited electrons supplied from SrTiO₃ within 15 min. After 15 min, the K-edge position was shifted slightly to the higher energy side, which indicates that the electron density of the PdCrO_x nanoparticles is slightly decreased after 15 min, i.e., the Pd species were oxidized. These results indicate that the electronic state of the Pd species in the PdCrO_x nanoparticles is dependent on the photoirradiation time.

Cr K-edge XANES spectra of PdCrO_x/SrTiO₃ were also investigated, and the results are shown in Figure S1. In contrast to the Pd K-edge spectra, the Cr K-edge positions were similar, regardless of the photoirradiation time. Therefore, it is concluded that the electronic state of only Pd showed a significant change during photoirradiation.

2.3. EXAFS

Figure 5 shows the Fourier transforms of k³-weighted Pd K-edge EXAFS spectra. The peak of Pd foil at 2.5 Å is assigned to Pd–Pd bonding, and the peak of PdO at 1.5 Å is assigned to Pd–O. The 0 h sample showed a peak at 1.5 Å, which indicates the presence of Pd–O bonding. In the cases of the 0–5 h samples, the intensity of Pd–O bonding became relatively weak with an increase in the photoirradiation time. On the other hand, the intensity of the Pd–Pd bonding peak was increased. Finally, the 5 h sample showed peaks for both Pd–O and Pd–Pd, which indicates that the Pd species in PdCrO_x/SrTiO₃ transforms from an oxide-like state to a mixture of Pd⁰ and Pd oxide as photoirradiation progressed. This trend does not contradict the observation in the XANES analysis (Figure 4). Therefore, it is concluded that the Pd species is reduced from oxide to the metallic state, although not completely, during the photoirradiation process. Similar to the Cr K-edge XANES results, no noticeable change could be identified in the fine structure of Cr (Figure S1).

2.4. XPS

Figure 6 shows XPS Pd 3d spectra for the PdCrO_x/SrTiO₃ samples. All of the Pd 3d peaks can be deconvoluted into two peaks with binding energies of 337.1 and 335.2 eV, which are assigned to oxidized and metallic Pd species, respectively, although the binding energies are slightly different from those in the references (PdO: 336.5 eV; metallic Pd in Pd/SrTiO₃: 334.8 eV) [21]. The intensity ratio of the 335.2 eV peak becomes stronger relative to the 337.1 eV peak with an increase in the photoirradiation time, which is consistent with the trend observed in the EXAFS spectra (Figure 5). The details of the surface electronic states of PdCrO_x were previously investigated using XPS [16]. The XPS Pd 3d peak position of Pd species co-existing with Cr³⁺ was thereby determined to be higher than that of bulk PdO(II). In addition, the Pd 3d peak position and structure of PdCrO_x was different from the PdO(II) single oxide, and the surface absorbed Pd precursor. Therefore, the 337.1 eV peak is not a single Pd oxide or the Pd precursor but another Pd species that interacted strongly with Cr.

In the case of Cr, the Cr 2p peak in various PdCrO_x/SrTiO₃ samples was located in a position similar to Cr₂O₃, as shown in Figure S2. Therefore, the valence state of Cr on the surface of PdCrO_x/SrTiO₃ is independent of the photoirradiation conditions.

On the basis of the XAFS and XPS measurements, it is thus likely that the deposited core/shell nanostructures on SrTiO₃ consist of metallic Pd, oxidized Pd, and Cr^III (i.e., Cr₂O₃), where the core and shell are mainly formed by metallic Pd and Cr₂O₃, respectively, with oxidized Pd species being located at the core/shell interface. However, we could not observe the oxidized Pd species in the TEM images shown in Figure 3, due to the compositional inhomogeneity of the deposited nanoparticles.

2.5. Effect of PdCrO_x Nanoparticles deposited on SrTiO₃ upon the Photocatalytic Activity toward Overall Water Splitting

The as-prepared PdCrO_x/SrTiO₃ samples were used to conduct photocatalytic reactions.

Table 1. Photocatalytic overall water-splitting activities of PdCrO_x/SrTiO₃ prepared for various time periods (0–5 h)^a.

Entry	Sample	Loaded Cr/wt%	Rate of Gas Evolution/μmol h−1
			H2	O2
1	0 h	0.2	6.4	0.2
2	15 min	0.6	6.5	1.9
3	30 min	0.6	9.4	4.3
4	1 h	0.7	14.7	4.7
5	5 h	0.8	14.0	5.7
6	SrTiO3	-	0.1	N. D.
7	Pd/SrTiO3	-	3.4	0.3
8	Cr2O3/Pd/SrTiO3	0.5	5.3	1.7

^a Reaction conditions: Catalyst, 100 mg; pure water, 140 mL; metal precursors, Pd 0.5 wt%, Cr 1.0 wt%; light source, 300 W xenon lamp (λ > 300 nm).

shows the overall water-splitting activities of PdCrO_x/SrTiO₃ prepared for various time periods (0–5 h). The 0 h sample produced H₂ and O₂; however, the H₂/O₂ ratio was not stoichiometric (entry 1). The total amount of H₂ and O₂ was increased, and the stoichiometry was improved with an increase of the irradiation time (entries 2–5). Therefore, longer irradiation times were effective to maximize the promotional effect of PdCrO_x nanoparticles on photocatalytic water splitting.

Bare and only Pd-loaded SrTiO₃ were also used for comparison. Only H₂ was evolved at a very slow rate with the unmodified SrTiO₃ (entry 6); therefore, cocatalyst loading was necessary for overall water splitting. Pd-loaded SrTiO₃ produced H₂ and O₂ upon UV irradiation, although the H₂/O₂ ratio was far from stoichiometric (entry 7). As reported previously, molecular O₂ present in the reaction system can prevent stoichiometric water splitting on a semiconductor photocatalyst, because it can cause backward reactions [16]. Therefore, a suitable modification method such as Cr₂O₃-shell modification [13,19] is essential to suppress O₂-related backward reactions and efficiently allow overall water splitting to proceed. Both the stoichiometry and the amount of evolved gases were improved by the loading of a thin Cr₂O₃ shell on the Pd particles (entry 8). However, the activity of the Pd-core/Cr₂O₃-shell system was much lower than that of the optimized PdCrO_x/SrTiO₃. This provides clear evidence of the superior functionality of the optimally-synthesized PdCrO_x nanoparticles as cocatalysts. Our previous work has also shown that the photocatalytic activity of PdCrO_x/SrTiO₃ (equivalent to the present 5 h sample) for overall water splitting is recoverable, with negligible changes in the valence state of Pd and Cr [16].

In Cr-containing mixed metal (oxide) cocatalysts, the key factors to enhance the promotional effect on photocatalytic water splitting are improvement of the H₂ evolution activity and the suppression of backward reactions (including H₂/O₂ recombination and O₂ photoreduction) [14]. For example, Cr(III) oxide shells formed on noble metal nanoparticles (cores) prevents these backward reactions [19]. We have previously reported that PdCrO_x nanoparticles with a similar core/shell type structure had the same functionality [17].

The promotional effect of PdCrO_x nanoparticles on photocatalytic H₂ evolution was investigated using methanol as an electron donor. Note that, in this reaction, methanol is irreversibly oxidized, with no O₂ evolution [17]. As listed in

Table 2. Photocatalytic hydrogen evolution activity using PdCrO_x/SrTiO₃ prepared for various time periods (0–5 h)^a.

Entry	Sample	Loaded Cr/wt%	Rate of H2 Evolution/μmol h−1
1	0 h	0.2	12.5
2	15 min	0.6	57.4
3	30 min	0.6	44.2
4	1 h	0.7	50.0
5	5 h	0.8	47.4
6	Pd/SrTiO3	0	43.3
7	Cr2O3/Pd/SrTiO3	0.5	28.2

^a Reaction conditions: Catalyst, 100 mg; 10 vol% aqueous methanol solution, 140 mL; metal precursors, Pd 0.5 wt%, Cr 1.0 wt%; light source, 300 W xenon lamp (λ > 300 nm).

, all of the tested samples produced H₂. The 15 min sample, which exhibited relatively low activity for overall water splitting (see Table 1), gave the highest activity among the tested samples (entry 2), although the difference in activity among the PdCrO_x/SrTiO₃ samples was not very large. The Pd metal and Pd/Cr₂O₃ (core/shell) type samples exhibited lower activity (entries 6 and 7). A Cr₂O₃ shell formed on the surface of noble metal nanoparticles has been reported to prevent not only back reaction but also proton reduction to produce H₂ to a certain extent [19]. The H₂ evolution activity using Pd/SrTiO₃ was decreased by the stepwise photodeposition of Cr₂O₃ thin shell, as shown in Figure S3. The H₂ evolution activity of PdCrO_x/SrTiO₃ should be reduced with an increase of the irradiation time due to an increase of the amount of loaded Cr, because the reaction sites are covered with the Cr₂O₃ shells. However, the observed H₂ evolution activities were not simply decreased by Cr co-loading, as evident in Table 2. Therefore, it is suggested that interaction between Pd and Cr enhanced the intrinsic H₂ evolution activity, although the mechanistic detail is still under investigation. Nevertheless, the H₂ evolution activity of PdCrO_x is not significantly changed by the photoirradiation time. Therefore, it is worth considering that there is another factor that influences the overall water-splitting activity of PdCrO_x/SrTiO₃.

We also considered that backward reaction caused by evolved O₂ could influence the water-splitting activity. Therefore, two types of backward reaction (H₂/O₂ recombination and O₂ photoreduction) were examined.

Table 3. Recombination activity of H₂ and O₂ over PdCrO_x/SrTiO₃ (15 min, 5 h)^a.

Entry	Sample	Loaded Cr/wt%	Rate of Diminished Gases/μmol h−1
			H2	O2
1	15 min	0.6	0.1	trace
2	5 h	0.8	2.0	1.7
3	Pd/SrTiO3	-	24.6	12.2

^a Reaction conditions: Catalyst, 25 mg; 10 vol% aqueous methanol solution, 100 mL; metal precursors, Pd 0.5 wt%, Cr 1.0 wt%; dark conditions; initially contained gases, H₂ 400 μmol and O₂ 200 μmol.

compares the rates of H₂/O₂ recombination using the 15 min and 5 h samples, as well as Pd-loaded SrTiO₃. In contrast to Pd/SrTiO₃, which exhibited very fast H₂/O₂ recombination, the PdCrO_x loaded samples exhibited much slower recombination rates.

Oxygen photoreduction activity was also investigated during photocatalytic H₂ evolution in the presence or absence of O₂. The introduction of O₂ had a negative impact on the H₂ evolution activity. As detailed in

Table 4. Rates of H₂ evolution and O₂ consumption over PdCrO_x/SrTiO₃ samples (15 min and 5 h) in the presence or absence of O₂^a.

Entry	Sample	Loaded Cr/wt%	Atmosphere	Rate/μmol h−1
				H2 Evolution	O2 Consumption
1	15 min	0.6	Ar	37.5	-
2	15 min	0.6	Ar + O2	1.1	47.7
3	5 h	0.8	Ar	28.1	-
4	5 h	0.8	Ar + O2	5.5	23.6

^a Reaction conditions: Catalyst, 100 mg; 10 vol% aqueous methanol solution, 140 mL; metal precursors, Pd 0.5 wt%, Cr 1.0 wt%; light source, 300 W xenon lamp (λ > 300 nm); atmosphere condition, 3.8 kPa Ar or 1.9 kPa Ar + 1.9 kPa O₂.

, almost 97% of the initial activity was lost in the 15 min sample. Nevertheless, the 5 h sample still maintained approximately 20% of the initial activity, even in the presence of O₂. The difference in the H₂ evolution activity between the two samples is in qualitative agreement with the difference in the O₂ consumption activity. It is thus clear that O₂ photoreduction occurred more actively on the 15 min sample than on the 5 h sample. Therefore, more effective suppression of O₂ photoreduction could be realized by longer time photoirradiation during the preparation of PdCrO_x.

3. Materials and Methods

3.1. Synthesis of SrTiO₃

The SrTiO₃ was prepared using the polymerized complex method [6]. Titanium tetraisopropoxide (95.0+%, FUJIFILM Wako Pure Chemical, Osaka, Japan) was dissolved in methanol. Anhydrous citric acid (FUJIFILM Wako Pure Chemical, Osaka, Japan) was then added to the solution and heated at 353 K. SrCO₃ (99.9%, Kanto Chemical, Tokyo, Japan) and ethylene glycol (>99.5%, FUJIFILM Wako Pure Chemical, Osaka, Japan) were added to the solution. After heating at 423 K overnight for polymerization to proceed, the solution was heated in air at 573 K, and then at 823 K for 2 h to remove organic residues. The precursor was finally calcined at 1423 K for 2 h.

3.2. Cocatalyst Loading

The PdCrO_x cocatalyst was deposited on the surface of SrTiO₃ by a simultaneous photodeposition method, as reported previously [16]. An amount of 0.2 g of SrTiO₃ was dispersed in 100 mL of 10 vol% methanol aqueous solution containing 1.0 wt% Cr (vs. SrTiO₃) from K₂CrO₄ and 0.5 wt% Pd (vs. SrTiO₃) from Na₂PdCl₄·3H₂O (FUJIFILM Wako Pure Chemicals Co., Osaka, Japan) as precursors. The solution was completely degassed and irradiated at room temperature for various time periods of 0–5 h. Evolved gases were detected using gas chromatography (GC-3200, Shimadzu Co., Kyoto, Japan) with a thermal conductivity detector (TCD), an MS-5A column and argon as a carrier gas (GL Science Co., Tokyo, Japan).

The amount of K₂CrO₄ (i.e., deposited Cr) remaining in solution was quantified using UV-visible absorption spectroscopy. First, the PdCrO_x/SrTiO₃ powder was collected by filtration and dried overnight at 343 K. After the addition of 0.27 mmol of EDTA·2Na (99.5%, Dojin Chem., Tokyo, Japan) to the filtrate, the mixture was boiled for 0.5 h. The solution was analyzed using UV-visible absorption spectroscopy.

3.3. Characterization

Prepared samples were studied using TEM, XAFS and XPS. TEM observations and the combined EDX were conducted using a JEM-2010F apparatus (JEOL, Tokyo, Japan). The Pd and Cr K-edge XAFS spectra were collected at AR-NW10A of the Photon Factory Advanced Ring and BL-9A of the Photon Factory, respectively (Proposal no. 2014S2-006, Photon factory, Tsukuba, Japan). Standard samples (Pd foil, PdO, Cr foil, Cr₂O₃) were measured by the transmission method. PdO and Cr₂O₃ were diluted in boron nitride and compressed to form pellets. Various PdCrO_x/SrTiO₃ samples prepared for various photoirradiation times (0–5 h) were measured by the fluorescence method using a multichannel solid state detector. The photon energies were calibrated according to the X-ray absorption edge of Pd foil (24350 eV) and Cr foil (5989 eV). Analysis of the raw XAFS spectra was conducted using the Athena and Artemis programs [22]. The Fourier transforms of the k³-weighted EXAFS spectra were typically in the 3.0–11 Å⁻¹ region. The XPS spectra were obtained using an ESCA 3400 (Shimadzu, Kyoto, Japan) with a Mg anode. The binding energy of the C 1s peak (285.0 eV) that originated from adventitious carbon was used as the XPS reference.

3.4. Photocatalytic Reaction

Water-splitting reactions were conducted in a top-irradiation cell connected to a closed gas circulation system. An amount of 0.1 g of PdCrO_x/SrTiO₃ was dispersed in 140 mL of distilled water or 10 vol% aqueous methanol solution. The reactant solutions were degassed completely, followed by photoirradiation (Xe lamp, 300 W, 20 A, λ ≥ 300 nm). Evolved gases were analyzed using gas chromatography. The reproducibility of the gas evolution rate was within ~20%.

3.5. Backward Reactions

Water formation from H₂ and O₂ was conducted under similar conditions to that for the photocatalytic reaction. An amount of 0.025 g of PdCrO_x/SrTiO₃ was dispersed in pure water. The reactant solution and gas phase were degassed completely and purged with hydrogen and oxygen. The reduction of gases was observed under dark conditions. Photocatalytic O₂ reduction activity was investigated under similar experimental conditions to those used for the photocatalytic hydrogen evolution reaction with O₂.

4. Conclusions

Photochemical preparation of PdCrO_x multi-component nanoparticles on SrTiO₃ was studied to develop a new cocatalyst for overall water splitting. The valence state and fine structure of the Pd species in PdCrO_x were strongly dependent on the photoirradiation time during PdCrO_x deposition. The Pd species in PdCrO_x was changed from oxide to oxide/metal mixed phase with an increase of the photoirradiation time, which contributed to enhanced activity for overall water splitting. The results of photocatalytic reactions indicated that the effect of activity enhancement was due to the suppression of O₂-photoreduction, a backward reaction of overall water splitting.