Photocatalytic Activity of Ga2O3 Supported on Al2O3 for Water Splitting and CO2 Reduction.

We have examined the photocatalytic activity of Ga2O3 supported on Al2O3 (Ga2O3/Al2O3 catalyst) without a noble metal cocatalyst for water splitting and reduction of CO2 with water under UV light irradiation by changing the loading amount of Ga2O3. All prepared Ga2O3/Al2O3 catalysts show photocatalytic activities for both water splitting and CO2 reduction, and their activities are significantly improved compared to those of nonsupported Ga2O3 and Al2O3. The water splitting is dominated for Ga2O3/Al2O3 with less than 1.0 vol % of Ga2O3 loaded, whereas the CO2 reduction, for higher Ga2O3-loaded samples (2.6, 4.2 vol %). Crystalline structure characterizations of Ga2O3/Al2O3 catalysts indicate that active sites for both reactions are different. The water splitting proceeds on nanometer-sized Ga2O3 rods dispersed on an Al2O3 support consisting of a little distorted α-Ga2O3 phase. On the other hand, the CO2 reduction proceeds on sub-micrometer-sized Ga2O3 particles consisting of mixed phases of α-Ga2O3 and γ-Ga2O3 or with appearance of boundaries between the α and γ phases, which plays a critical role. Al2O3 used as the support of the Ga2O3 particles does not seem to play an important role in the photocatalytic CO2 reduction.

Introduction

Recently, Ga2O3 has attracted a lot of interest as a photocatalyst for water splitting and CO2 reduction with water and various efforts have been paid to improve its photocatalytic activity.[1−4] Yamamoto et al. have reported improvement of photocatalytic activity for CO2 reduction with water under UV irradiation using Ag as a cocatalyst.[4−6] Teramura et al. have succeeded in improving photocatalytic activity for the CO2 reduction by cation (Zn, Pr, or Yb) doping into Ga2O3 with a Ag cocatalyst.[7−9] However, both the mechanism and the role of the Ag cocatalyst for these photocatalytic reactions have not been well understood. Although Kato et al.[10−12] have succeeded in nitrogen doping into Ga2O3 to induce photocatalytic activity under visible light irradiation, the activity has remained at a lower level.

In the present study, we have focused on geometrical or morphological effects of Ga2O3 particles supported on Al2O3 (referred as Ga2O3/Al2O3 hereafter) for water splitting and CO2 reduction under UV light irradiation without using the Ag cocatalyst. The reasons for utilization of Al2O3 as the support are 2-fold: (1) It increases the surface area of Ga2O3,[13,14] as evidenced by the observation that Ga2O3/Al2O3 was used for the removal of NO.[15,16] (2) Al2O3 hardly shows photocatalytic activity for both water splitting and CO2 reduction. Ga2O3 supported on Al2O3 (Ga2O3/Al2O3 samples) was produced by loading of Ga2O3 particles on γ-Al2O3 particles by an impregnation method.[17] By changing the loading amount of Ga2O3, i.e., changing the mass or volume ratio of Ga2O3 and Al2O3, their photocatalytic activities for both water splitting and CO2 reduction were examined. All prepared Ga2O3/Al2O3 samples have shown significantly higher photocatalytic activity compared to that of pure Ga2O3 (nonsupported Ga2O3). The cause of the improvement is discussed considering detailed characterization using X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), and X-ray absorption fine structure (XAFS) and morphological observation using field emission scanning electron microscope (FE-SEM).

Results

Photocatalytic Reduction of CO2 with Water over the Ga2O3/Al2O3 Samples

Figure compares production rates of H2, O2, and CO for all samples that were taken after 5 h of the reaction test showing a nearly steady state. They are also given in Table . The production rates of H2, CO, and O2 almost retained the stoichiometric ratio.[18] Although we prepared the samples with a wt % base; hereafter, we have converted wt % to vol % and discussed all things on the basis of vol % for an easy discussion and understanding hereafter. As seen in the figure, the H2 production rates of all samples were higher than those with unsupported Ga2O3 and the rates increased with the loaded amount of Ga2O3 and after reaching the maximum, decreased with increasing the loaded amount of Ga2O3. The CO production rate changed in a completely different way from that of H2 production. This indicates that H2 production and CO production proceeded independently, which is discussed later. Although all samples except pure Al2O3 were active for the photocatalytic CO2 reduction, samples with the loaded amount of Ga2O3 less than 1 vol % showed a lesser CO production rate and CO selectivity remained small compared to those of unsupported Ga2O3 (100 vol % Ga2O3/Al2O3). Samples loaded with more than 2 vol % Ga2O3 showed a higher reaction rate than that of unsupported ones. Ga2O3/Al2O3 (2.6 vol %) showed the maximum CO production rate, which was 2.7 times larger than that of the unsupported Ga2O3. This is a significant improvement of the photocatalytic activity of Ga2O3 without a cocatalyst.

Figure 1

Production rates of CO (gray), H2 (white), and O2 (black) and selectivity toward CO evolution in the photocatalytic conversion of CO2 with H2O over Ga2O3/Al2O3 together with those for α-Ga2O3 and γ-Ga2O3 after 5 h of the reaction test.

Table 1

Summary for Production Rates of H2, O2, and CO; Ga2O3 Crystal Structures; and Brunauer–Emmett–Teller (BET) Specific Surface Area for All Prepared Samples

sample Ga2O3/Al2O3 ratio
wt %	vol %	mol %	CO production rate (μmol/h)	H2 production rate (μmol/h)	O2 production rate (μmol/h)	Ga2O3 crystalline structure	BET specific surface area (m2/g)
0	0	0	0.033	1.2	0.4		189.0
5	0.15	2.7	1.0	32.5	17.2	α	167.3
10	0.32	5.7	1.4	49.2	28.2	α	134.6
20	0.73	12	0.7	30.4	15.9	α	130.7
40	2.6	27	4.4	34.2	19.9	α, γ	112.9
60	4.2	45	3.3	24.4	12.2	α, γ	120.8
100	100	100	1.6	13.0	6.9	α, β, γ	77.3
	α-Ga2O3		0.17	14.7	5.6
	γ-Ga2O3		0.39	2.9	1.5

Characterization

Crystalline Structure

Figure A shows changes of XRD patterns with Ga2O3 loading amounts plotted against a diffraction angle of 2θ for all prepared samples. Figure B shows the magnified XRD patterns. As seen in Figure B, the two peaks within 42–50° attributed to Al2O3 did not show any shifts.[19] This suggests that loaded Ga2O3 hardly dissolved into Al2O3, although Ga2O3 and Al2O3 are fully dissolvable with each other according to the phase diagram.[20] Hence, in Figure are shown difference XRD spectra, i.e., the intensity of XRD peaks of Al2O3 subtracted from the observed intensity of XRD peaks of the samples. As indicated in the figure, most of the XRD peaks of Ga2O3 were assigned to either α, β, or γ phase. This confirms that Ga2O3 hardly dissolved into Al2O3 and vice versa but was loaded or deposited on Al2O3.[21]

Figure 2

(A) XRD patterns of Ga2O3/Al2O3 samples: (a) 0 vol % Ga2O3/Al2O3, (b) 0.15 vol % Ga2O3/Al2O3, (c) 0.32 vol % Ga2O3/Al2O3, (d) 0.73 vol % Ga2O3/Al2O3, (e) 2.6 vol % Ga2O3/Al2O3, (f) 4.2 vol % Ga2O3/Al2O3, and (g) 100 vol % Ga2O3/Al2O3. (B) Enlarged XRD patterns of all samples at 2θ ranging from 42 to 50°.

Figure 3

XRD patterns of the Ga2O3 phase for Ga2O3/Al2O3 samples given by subtracting the XRD intensity of pure Al2O3 from that of the Ga2O3/Al2O3 sample: (a) 0.15 vol % Ga2O3/Al2O3, (b) 0.32 vol % Ga2O3/Al2O3, (c) 0.73 vol % Ga2O3/Al2O3, (d) 2.6 vol % Ga2O3/Al2O3, (e) 4.2 vol % Ga2O3/Al2O3, and (f) 100 vol % Ga2O3/Al2O3.

Results of the XRD analysis are summarized as follows: lower Ga2O3-loaded samples (less than 1 vol % Ga2O3/Al2O3) consisted of a single phase α-Ga2O3. However, the lowest Ga2O3-loaded sample (0.15 vol %) does not seem well crystallized but is close to amorphous or its particle size would be very small. The crystallinity of Ga2O3 particles on Al2O3 became better on increasing the loading amount. Higher Ga2O3-loaded samples (more than 2 vol %) consisted of mixed phases α-Ga2O3 and γ-Ga2O3. It should be noted that unsupported Ga2O3 (100% vol % Ga2O3/Al2O3) consisted of three phases α, β, and γ.

Figure A shows k3-weighted Ga K-edge EXAFS spectra. Fourier transformation was performed on each EXAFS spectrum in the range from 3 to 12 Å–1, and the radial structure function (RSF) was obtained as shown in Figure B. In RSFs, the first peak appearing at 1–2 Å is assigned to the backscattering from an adjacent oxygen atom to a Ga atom (Ga–O bond). The second peak at around 2.7 Å mainly showed the presence of the second-neighboring gallium atoms (Ga–(O)–Ga bond).[10,17,22] It should be noted that 0.15 vol % Ga2O3/Al2O3 showed a high-intensity peak at around 1.5 Å. However, the peak intensity is too high to be caused simply by Ga–O bonds, suggesting some contribution of O bond to Al (Al–O–Ga). Probably because the deposited Ga2O3 particles are too small to be completely crystallized, their crystal structure, i.e., bonding lengths of Ga–O in surface layers neighboring to the Al2O3 support, should be distorted owing to a difference in atomic distances of Ga–O and Al–O. The RSF of lesser Ga2O3-loaded samples (under 0.73 vol %) showed a shoulder at around 3.5 Å, and a broad peak appeared at around 4–5 Å. Since both were clearly observed in α-Ga2O3, the Ga2O3 crystal phase should be dominated with the α phase, agreeing with the result of XRD. However, the peak of the first coordination region is much higher than that of α-Ga2O3. This could be attributed to the distortion of the α-Ga2O3 by the supporting Al2O3. The RSF of higher Ga2O3-loaded samples (above 2.6 vol %) showed a shoulder peak at around 0.8 Å in addition to the 3.5 Å shoulder. Since the 0.8 Å shoulder or peak was prominent in γ-Ga2O3, higher Ga2O3-loaded samples are confirmed to consist of α and γ phases, as observed by the XRD.

Figure 4

(A) Ga K-edge EXAFS and (B) Fourier transforms of EXAFS for Ga2O3/Al2O3 samples, α-Ga2O3 and γ-Ga2O3, (a) 0.15 vol % Ga2O3/Al2O3, (b) 0.32 vol % Ga2O3/Al2O3, (c) 0.73 vol % Ga2O3/Al2O3, (d) 2.6 vol % Ga2O3/Al2O3, (e) 4.2 vol % Ga2O3/Al2O3, and (f) 100 vol % Ga2O3/Al2O3.

Figure shows field emission scanning electron microscope (FE-SEM) images of the samples taken in a backscattered electron mode, of which the contrast well corresponds to the weight of constituent atoms of the sample. The images clearly distinguish columnar-like Ga2O3 particles (around 100 nm in length and a few nanometer in width) from supporting Al2O3 ones with micrometer size. One can see that for lesser Ga2O3-loaded samples (less than 1 vol %), number density of the Ga2O3 particles increased on increasing the loaded Ga2O3 amount, whereas for higher loaded samples, the particle size grew and certain areas of Al2O3 were hindered. These images are quite consistent with the characterization of XRD and EXAFS.

Figure 5

FE-SEM images of (a) 0.32 vol % Ga2O3/Al2O3, (b) 0.73 vol % Ga2O3/Al2O3, (c) 2.6 vol % Ga2O3/Al2O3, and (d) 4.2 vol % Ga2O3/Al2O3.

Diffuse Reflectance UV–Vis Spectra

Figure compares diffuse reflectance UV–vis spectra for all samples. On increasing the loaded amount of Ga2O3, the absorption edge shifted to the longer wavelength region and the shifts were saturated above 2.6 vol % very close to 100 vol % Ga2O3/Al2O3. However, for lower Ga2O3-loaded samples (less than 1 vol %), the red shift of their absorption edge was not appreciable. A small amount of dissolution of Al2O3 having wider band gap into nanometer-sized Ga2O3 rods would result distortion of their α-Ga2O3 phase and consequently inhibit their band gap narrowing.

Figure 6

Diffuse reflectance UV–vis spectra of Ga2O3/Al2O3 samples: (a) 0.15 vol % Ga2O3/Al2O3, (b) 0.32 vol % Ga2O3/Al2O3, (c) 0.73 vol % Ga2O3/Al2O3, (d) 2.6 vol % Ga2O3/Al2O3, (e) 4.2 vol % Ga2O3/Al2O3, and (f) 100 vol % Ga2O3/Al2O3.

Table summarizes production rates of H2, O2, and CO, together with crystal structures and BET specific surface areas for all samples tested. The surface area decreased with the increase of the loading amount of Ga2O3 except that of 4.2 vol % Ga2O3/Al2O3.

Chemical Nature of Ga in Surface Layers

XPS Analysis

Figure A shows Ga 3d XPS spectra for the samples. The binding energy of the XPS spectra were corrected, referring to the Al 2p XPS peak shown in Figure B. For lesser Ga2O3-loaded samples (less than 1 vol %), the intensity of Ga 3d peaks was very weak. Nevertheless, the intensity gradually increased on increasing the loaded amount of Ga2O3. For higher Ga2O3-loaded samples (higher than 2 vol %), the Ga 3d peak became significant. The intensity of Al 2p XPS peaks decreased continuously without changing their shapes and positions. From the peak intensities of Ga 3d and Al 2p, the atomic fractions of Ga near the surface region of all samples were determined and plotted against the loaded amount of Ga2O3 in vol %. The figure clearly shows that for lesser Ga2O3-loaded samples, the atomic fraction linearly increased with the volume fraction of Ga2O3. This indicates that the coverage of Ga2O3 on Al2O3 linearly increased. Over 2 vol % loading, the coverage became large enough to hinder some parts of the Al2O3 surface, as seen in Figure , and accordingly, the Ga atom fraction near the surface significantly increased, as appears in Figure .

Figure 7

(A) Ga 3d XPS peaks. (B) Al 2p XPS peaks. (a) 0.15 vol % Ga2O3/Al2O3, (b) 0.32 vol % Ga2O3/Al2O3, (c) 0.73 vol % Ga2O3/Al2O3, (d) 2.6 vol % Ga2O3/Al2O3, and (e) 4.2 vol % Ga2O3/Al2O3.

Figure 8

Atomic fraction of Ga near the surface region of Ga2O3/Al2O3 samples determined by XPS.

X-ray Absorption Near Edge Structure (XANES) Analysis

Figure shows Ga L3-edge XANES spectra. The peak appearing near 1123 eV is caused by a Ga–O–Al bond.[16,23,24] For lesser Ga2O3-loaded samples, this peak was relatively high compared with other peaks. This indicates that some interaction between Ga2O3 and Al2O3 occurred in lesser loaded samples, as already suggested in the EXAFS analysis. For higher Ga2O3-loaded samples, the peak intensity at 1123 eV became less and the spectrum transformed to that of Ga2O3, showing loaded Ga2O3 as well crystallized. Thus, the result of XANES analysis is quite consistent with that of EXAFS and XPS analyses.

Figure 9

Ga L3-edge XANES spectra of all prepared samples. The spectrum of β-Ga2O3 is also given for comparison. (a) 0.15 vol % Ga2O3/Al2O3, (b) 0.32 vol % Ga2O3/Al2O3, (c) 0.73 vol % Ga2O3/Al2O3, (d) 2.6 vol % Ga2O3/Al2O3, (e) 4.2 vol % Ga2O3/Al2O3, and (f) 100 vol % Ga2O3/Al2O3. The peaks appearing at 1123 eV are assigned to be caused by Ga–O–Al bond formation.

Discussion

As indicated in Figure and Table , all Ga2O3/Al2O3 samples showed higher photocatalytic activity compared to that of unsupported Ga2O3 for both water splitting and CO2 reduction without a cocatalyst. Furthermore, the water splitting seems to proceed in a different way from the CO2 reduction. In the following discussion, we separately discuss the mechanism and the active sites for the water splitting and CO2 reduction.

It should be noted that BET surface area was the largest for Al2O3 whereas the least for Ga2O3. As depicted in FE-SEM images (Figure ), the surface of the loaded Ga2O3 particles seems smoother than that of Al2O3. Consequently, the BET surface area decreased on increasing the loaded amount of Ga2O3 or surface coverage of Al2O3 by Ga2O3 and was not directly correlated to the photocatalytic activities. This means that the role of Al2O3 in the photocatalytic reaction is not straightforward but just supporting Ga2O3 and assisting dispersive precipitation of small Ga2O3 particles.

Water Splitting

The water splitting was dominated for lesser Ga2O3-loaded samples (less than 1.0 vol %). In these samples, columnar Ga2O3 particles (around 10–300 nm in length and less than 10 nm in width) consisting of the α phase were precipitated on supporting Al2O3 particles (larger than micrometer). On increasing the loaded amount of Ga2O3, its areal density increased without changing its shape (see Figure ). This indicates that the active sites for water splitting should be on the surface of the precipitated α-Ga2O3 of which the deposited density increased with the loaded amount of Ga2O3. As EXAFS and XANES analyses showed, the crystalline structure of the α phase of the Ga2O3 particles is likely distorted by supporting Al2O3. Such small distortion in the α phase could be effective for photocatalytic water splitting.

The H2 production rates in Figure are magnified in Figure . Since the production rate continuously decreased on increasing the deposited amounts of Ga2O3, smaller sizes of the α-Ga2O3 particles, hence probably having larger distortion, are much effective for the water splitting. This supports that distortion of the α phase is a critically important factor for the catalytic activity.

Figure 10

H2 production rate against the loaded amount of Ga2O3 given in vol %.

For higher Ga2O3-loaded samples, precipitated particle sizes grew with the loaded amount, resulting in a decrease of the surface-to-volume ratio of the particles and consequently decreasing the number of active sites as a whole. In addition, the larger precipitated particles for the larger Ga2O3-loaded samples were dominated with α and γ phases (see Table ). In other words, larger sizes of Ga2O3 particles prefer to take thermally more stable γ and β phases than the α phase and accordingly they lose the active sites for water splitting.

From all above observations, for water splitting under UV irradiation, we can conclude that the higher photocatalytic activity of Ga2O3 particles supported on Al2O3 is caused by dispersive precipitation of nanometer-sized Ga2O3 rods consisting of a little distorted α phase on the Al2O3 support.

It should be mentioned that because the UV–vis absorption of these less Ga2O3-loaded samples were weak and their band gap remained narrow (see Figure ), a number of electron–hole pairs that enable CO2 reduction are not likely produced enough under the present UV condition. This does not change the above conclusion.

CO2 Reduction

Active sites for CO2 reduction seem totally different from those for water splitting. In Figure , CO production rates and CO selectivity are replotted. One can clearly see that lesser Ga2O3-loaded samples were inactive for photocatalytic reduction of CO2. Their activities were less than those of the unsupported Ga2O3. Other samples active for the CO2 reduction consisted of two phases α-Ga2O3 and γ-Ga2O3. Thus, we can conclude that for the CO2 reduction, larger Ga2O3 particles consisting of the α and γ phases play an important role, whereas the α-Ga2O3 phase in nanorods dispersed on Al2O3 is inactive.

Figure 11

CO production rate and selectivity toward CO evolution in the photocatalytic reduction of CO2 with H2O for the Ga2O3/Al2O3 samples together with those of nonsupported Ga2O3, α-Ga2O3, and γ-Ga2O3.

It is also important to note that 2.6 vol % Ga2O3/Al2O3 showed the maximum activity. As seen in Figure (XRD spectra), on increasing the loaded amount of Ga2O3, the γ phase became dominant and crystallized well. Since the catalytic activity of both unsupported α-Ga2O3 and γ-Ga2O3 was far less than that of the supported ones, well-crystallized single-phase (α or γ) particles on Al2O3 are not likely to have high catalytic activity. Since the BET surface areas of the active samples were similar to each other, the difference of the activity is not likely caused by the surface area or the surface roughness. This indicates that crystallinity of the Ga2O3 particles on Al2O3 must play an important role, i.e., coexistence of both α and γ phases would be the key. Stronger UV–vis absorption and narrower band gap of the larger Ga2O3-loaded samples compared to those of less loaded samples would also contribute to enhance the activity.

Boundaries between the two different phases of Ga2O3 were claimed as active sites for photocatalytic CO2 reduction.[25−28] The present results also suggest the importance of the boundary. The ratio of the integrated length of all boundaries to the integrated surface areas of all particles is 1/r (2πr/πr2 = 1/r) and decreases on increasing the radius (r). This could be a reason for the activity reduction for higher Ga2O3 contents seen in Figure .

Conclusions

We have examined photocatalytic activity of Ga2O3 loaded on an Al2O3 support (Ga2O3/Al2O3 catalyst) for water splitting and CO2 reduction with water under UV light irradiation without a noble metal cocatalyst. Since on changing the loaded amount of Ga2O3, the geometrical structure or morphology of loaded Ga2O3 could be controlled, effects of these changes on the photocatalytic activity for water splitting and CO2 reduction were investigated.

All Ga2O3/Al2O3 samples showed higher photocatalytic activity compared to that of unsupported Ga2O3 for both water splitting and CO2 reduction without a cocatalyst. The water splitting seems to proceed in a different way from the CO2 reduction. The former is preferred by lower Ga2O3-loaded samples (less than 1 vol % Ga2O3), whereas the latter, by higher loaded ones (2.6 and 4.2 vol % Ga2O3).

For lesser Ga2O3-loaded samples, nanometer-sized Ga2O3 rods consisting of α-Ga2O3 phase were dispersively precipitated on Al2O3 support. On increasing the loading amount, the areal density of the precipitated rods increased without appreciable change in their shape. Correspondingly, the production rate of H2 linearly increased with the amount of loaded Ga2O3. The α-Ga2O3 phase of the nanorods is very likely distorted by the influence of supporting Al2O3 both in chemical nature and crystallinity and works as active sites for the water splitting.

For higher Ga2O3-loaded samples, active for the CO2 reduction, sub-micrometer-sized Ga2O3 particles were deposited on the Al2O3 support, which were consisted of α-Ga2O3 and γ-Ga2O3 and were not likely influenced by Al2O3. The mixed phases of α and γ in the sub-micrometer particles play an important role; in particular, the appearance of boundaries between the α and γ phases is very likely the key.

It should be noted that the role of Al2O3 in the photocatalytic reaction is not straightforward but supporting Ga2O3 and assisting dispersive precipitation of small Ga2O3 particles.

Experimental Section

Preparation of Ga2O3/Al2O3 Photocatalyst Samples

Ga2O3/Al2O3 photocatalyst samples were prepared by an impregnation method. Ga(NO3)3·8H2O (Kishida Chemical Co. Ltd. purity 99.0%) and 1.0 g γ-Al2O3 (Sumitomo Chemical Co. Ltd. purity 99.99%) were added to 200 mL of distilled water and stirred with a magnetic stirrer in air and dried up, followed by calcination at 823 K for 4 h, resulting in the Ga2O3/Al2O3 samples. The loaded amounts of Ga2O3 were 5, 10, 20, 40, and 60 wt %. Pure Al2O3 and unsupported Ga2O3 samples (referred as 0 and 100 wt %, respectively) were also prepared with a similar procedure. Unsupported Ga2O3 samples having single phases α-Ga2O3, β-Ga2O3, and γ-Ga2O3 were also prepared for comparison. α-Ga2O3 was obtained by calcining Ga2O3·nH2O (Mitsuwa Chemicals Co., Lid Ga2O3 78.8%) at 823 K for 5 h.[29] β-Ga2O3 (purity 99.99) was purchased from Kojundo Chemical Laboratory Co. Ltd. For the preparation of γ-Ga2O3, Ga(NO3)3·8H2O was dissolved in ethanol (approximately 3 g of the reagent in 50 mL of the solvent), and then an ethanol solution of 28 vol % aqueous ammonia (volume ratio of ethanol/aqueous ammonia = 1) was added slowly under continuous stirring at room temperature until no further precipitates were formed. The resultant precipitates were filtered, washed with ethanol, and vacuum-dried in a desiccator. Finally, the obtained solid was calcined at 773 K for 5 h to produce γ-Ga2O3.[30]

Crystalline structures of all Ga2O3/Al2O3 samples were examined by X-ray diffraction (XRD) analysis. XRD patterns of the samples were recorded on a MiniFlex600 (Rigaku) using Cu Kα as a radiation source with an operating voltage of 40 kV and current of 15 mA. The XRD patterns were collected at 2θ angles of 20–70°. The 2θ step size was 0.02°, and the scanning rate was 10°/min.

To examine the surface composition of the sample, XPS measurements were carried out at room temperature under vacuum using ESCA 3400 (Shimadzu). Mg Kα was used as an X-ray source with an electron acceleration voltage of 10 kV and a current of 20 mA. Ga K-edge EXAFS and Ga L3-edge XANES were recorded with the beam line of BL5S1 at the Aichi Synchrotron Radiation Center and BL2A at UVSOR, Institute for Molecular Science in Japan, respectively. Ga K-edge EXAFS and Ga L3-edge XANES were obtained by a transmission mode and by a sample current method or total electron yield mode, respectively. UV–vis diffuse reflectance spectra were recorded at room temperature using a spectrometer (JASCO V-670). The spectrum of Ba2SO4 was used as reference.

Morphology of the samples was observed by a field emission scanning electron microscope (FE-SEM, JSM-6500F, JEOL Ltd.) with a backscattered electron mode under an acceleration voltage of 15 kV. All samples were subjected to BET specific surface area measurements at 77 K (liq. N2 temperature) using Monosorb (Quantachrome). Before the BET measurements, all samples were heated at 573 K for 3 h under a nitrogen atmosphere.

Photocatalytic CO2 Reduction with Water (CO2 Reduction and Water Splitting Tests)

Photocatalytic CO2 reduction with H2O under UV light irradiation was tested for 0.1 g of one of the samples set in a fixed-bed flow reactor cell under CO2 gas flow. The UV light intensity was 35 mW/cm2 in the range of 254 ± 10 nm. Before the test, the sample was irradiated with a 300 W Xe lamp for 1 h under CO2 gas flow with the flow rate of 20.0 mL/min. Then, the reduction test was started, introducing a NaHCO3 aqueous solution (1.0 M) of 10.0 mL and CO2 gas with a flow rate at 3.0 mL/min under the UV light irradiation. The reaction products (CO, H2, and O2) were analyzed by a gas chromatograph equipped with a thermal conductivity detector.