Multiple-field-coupled catalysis combined with photocatalysis is attracting increasing interest since it performs better than photocatalysis alone [1-3]. Recently, photothermal coupling has proved to be an efficient and reliable approach to improve the photocatalytic efficiency and promote the development of photothermal catalysis [4-9]. Understanding how a photothermal catalyst works requires knowledge of photocatalysis. However, present characterizations of photocatalyst systems at room temperature are insufficient to understand the effect of photothermal coupling in photothermal catalysis. In this regard, a photothermal coupled characterization technique is required to understand the mechanism of photothermal catalysis.

Photoconductivity measurement is a promising tool to detect photogenerated charge carriers, since a change in photocurrent is closely related to free carrier density [10]. Moreover, in situ thermal coupled photoconductivity can be easily achieved by integrating a heating unit in the photoconductivity test chamber. Pomoni et al. [11] studied the effect of temperature on the transient photoconductivity of a TiO₂ film. They suggested that the transient photoconductivity of a nanocrystalline TiO₂ film in vacuum was governed by the thermal release of electrons from the trap states in the bandgap. Liu et al. [12-14] successfully employed in situ temperature-dependent photoconductivity to understand the photocatalytic degradation of gaseous formic acid in air. They revealed that the electron transfer was determined by the interface transfer of electrons to O₂ rather than by the transport. Therefore, in situ thermal coupled photoconduction can be used to study the separation, recombination, and interfacial reactions of photogenerated charges, which in turn would be useful in understanding the mechanism of photocatalysis and photothermal catalysis.

The photocatalytic activity of semiconductor oxides can be successfully enhanced by crystal facet modulation [15, 16], surface modification [17], formation of Z-scheme [18] or S-scheme heterojunctions [19, 20], and introduction of oxygen vacancies (Ov) [21-24]. Of these methods, introduction of Ov not only extends the visible light absorption range of the semiconductor [25] but also inhibits carrier recombination [26]. Ov can also promote the adsorption of reactants and decrease the activation energy of the reaction [27]. More importantly, Ov has been found to endow the photocatalyst with photothermal catalytic activity. Thus, an Ov-type semiconductor oxide is deemed as an ideal model to study the photothermal catalytic process [28, 29].

In this study, the in situ thermal coupled photoconductivity of Ov-TiO₂ was used to correlate photoconductivity with the photothermal catalytic reduction of CO₂. The effects of temperature and atmosphere on the photoconductivity were investigated to understand the photothermal catalytic of CO₂ reduction. Moreover, Ov-TiO₂ samples with different Ov contents were compared on the basis of their photoconductivity and photothermal catalytic activity to reveal the role of Ov in enhancing CO₂ reduction. The aim of this study was to use photoconductivity to elucidate the mechanism of the photothermal catalytic reduction of CO₂.

To fabricate the TiO₂ film, commercial ST-01 TiO₂ (Japan, Ishihara Sangyo) was used to prepare a TiO₂ slurry. The detailed preparation is shown in the supporting information. The TiO₂ slurry was coated on fluorine-doped tin oxide (FTO) glass (2.5 cm × 2.5 cm) with an etched channel (77.52 μm × 2.5 cm), followed by temperature-programmed annealing in a muffle furnace up to 500 ℃. The TiO₂ film was then heated in a N₂/H₂ atmosphere (10% H₂, 90% N₂) at 150 and 350 ℃, respectively. This was denoted as H₂-150 and H₂-350 respectively. The TiO₂ film without the N₂/H₂ treatment was denoted as "Unannealed."

Scheme 1 shows the device structure used for the photoconductivity measurements. The test chamber was connected to an electrochemical workstation (Keithley 2400 Sourcemeter), which was used to observe the changes in photocurrent. A heating plate was integrated at the bottom of the chamber, and the temperature was controlled using a thermocouple. The chamber contained shielded wires connected to an external electrochemical workstation for the accurate measurement of photoconductivity. UV light (Hayashi LA-410UV lamp) was irradiated through a circular quartz window (diameter, 4.5 cm) from the top of the chamber, with a light intensity of 20 mW/cm².

Before the measurements, the chamber was pre-aerated by CO₂ or N₂ for 20 min to remove the air in the reactor. The flow rate of the gases was controlled by a digital gas mass flow meter and maintained at 2 L/min. Each set of measurements was commenced in a dark state for 0 to 20 s. The UV light source was then switched on at 20 s and turned off at 100 s. The measurements were conducted at room temperature and 120 ℃, respectively. The temperature of the different samples was monitored by an infrared camera. The infrared thermograms of the Unannealed, H₂-150, and H₂-350 samples are shown in figure S1. As can be seen, the temperature changes in the samples were primarily due to the heating of the external hot plate, while the irradiation of the external light resulted in only a small additional increase in temperature.

The photocurrent was measured and defined using the following formula [14]:

(1)

Here, U is the voltage applied by an electrochemical workstation, σ is the conductivity, L is the channel width, and A is the cross-sectional area of the TiO₂ film. The applied voltage, U, was set at 1.5 V. Since the photocurrent is proportional to conductivity, the change regularity of photocurrent represents a change in conductivity. Hence, the photocurrent is directly used and not converted to photoconductivity.

In the photocatalytic test, ST-01 film was put into an airtight reactor. After sealing the reactor, carbon dioxide gas was bubbled for 20 min to expel the air from the reactor (2 L/min). The film was then irradiated by 20 mW/cm² UV light for 3 h. A Shimadzu 2400 gas chromatograph was used to measure the yield of CO.

The photothermal catalytic experiment was similar to the above photocatalytic experiment. The schematic of the reactor is shown in Scheme 2. The difference between the two setups was that in the photothermal catalytic process, external heating was applied and the temperature was controlled at 120 ℃.

It is well reported that ST-01 is composed of an anatase phase [30, 31]. During the preparation of the TiO₂ slurry and the N₂/H₂ treatment, no significant structural changes in TiO₂ were observed in the X-ray diffraction (XRD) patterns (Fig. S2) as well as in the UV/vis spectra (Fig. 1(a)) and Raman spectra (Fig. 1(b)). However, electron spin resonance spectroscopy (ESR) (Fig. 1(c)) was able to distinguish three typical samples. The sample after hydrogen annealing showed a distinct signal at g = 2.002, where a higher annealing temperature led to a stronger signal. The signal at g = 2.002 is a characteristic feature of the ·O₂^- radicals generated from the interaction of the absorbed atmospheric O₂ with the surface Ov sites [21, 22, 32-34]. Therefore, Ov was successfully introduced into the H₂-150 and H₂-350 samples. Yan et al. [35] also proposed that a vacuum or a reducing atmosphere could produce Ov, which is consistent with our results. Integral processing on the ESR signals was performed to compare the Ov in different samples quantitively. The number of spin electrons over the Unannealed, H₂-150, and H₂-350 samples were 5.3×10^-9, 1.11×10^-8, and 3.24×10^-8 mol/g, respectively. Assuming that all spin electrons originated from Ov, the ratio of Ov concentration for the Unannealed, H₂-150, and H₂-350 samples was found to be 1 : 2.08 : 6.09.

Fig. 2 shows the catalytic reduction of CO₂ via photocatalysis and photothermal catalysis, respectively. It was observed that CO was the sole product, and no CH₄ was produced. This agrees with most reported results that a cocatalyst-free TiO₂ system is not suitable for CH₄ production due to the difficulty of multiple electron/proton transfer [36-40]. By carefully comparing CO yields, several interesting conclusions can be drawn. Firstly, thermal coupled photocatalysis for all samples significantly improved the CO yield in comparison with photocatalysis, suggesting that photothermal coupling is a powerful tool for enhancing CO₂ reduction with a photocatalyst. Secondly, for photocatalysis, the introduction of Ov, unfortunately, brings a lower CO yield, with the H₂-150 sample giving the lowest CO yield. This suggests that Ov is not conducive to enhanced photocatalysis. Contrary to this finding, the introduction of Ov enhances the CO yield for photothermal catalysis, with the H₂-150 sample providing the highest CO yield. There is no significant activity loss after four experimental runs (Fig. S3), which suggests that Ov is suitable for enhanced photothermal catalysis. Thus, photothermal catalysis was found to be superior to photocatalysis with regards to CO₂ reduction, and Ov was found to improve the activity of TiO₂ in photothermal catalysis.

As was discussed, in the photocatalytic reduction of CO₂, Ov helps adsorption of CO₂ and enhances CO yield. The CO₂ adsorption behavior for hydrogenated TiO₂ was then investigated. Fig. S4 shows that the CO₂ adsorption ability follows the order: H₂-350 > H₂-150. However, the photothermal catalytic activity of H₂-350 is lower than that of H₂-150. Thus, it can be deduced that the photothermal catalytic activity is not solely governed by CO₂ adsorption.

To investigate the causes for the enhanced activity of photothermal coupling and Ov, photoconductivity measurements at different conditions were performed. When comparing the photoconductivity at room temperature and 120 ℃ (Fig. 3, inset), it was found that all samples exhibit higher photoconductivity at 120 ℃. The increased photoconductivity is believed to result from the accelerated transfer of electrons from the bulk to the surface at elevated temperatures. Due to the increased number of electrons at the surface per unit time, the probability of electron/proton-coupled reduction of CO₂ is enhanced. This is consistent with the enhanced activity observed in photothermal catalysis that is not seen in photocatalysis (Fig. 2).

The decay of photoconductivity on the photoconductivity curve can provide valuable information. As shown in Fig. 3(a) and (b), both unannealed and H₂-150 samples displayed an accelerated decay with the increase in temperature in a CO₂ atmosphere. In principle, the decay of photoconductivity is induced by either charge recombination or a charge-involved surface reaction. Since the photoconductivity is measured in CO₂, the photogenerated electrons tend to be captured by CO₂, which partly contributes to decreased photoconductivity. In this regard, the accelerated decay means a higher capacity for CO₂ reduction with a catalyst at higher temperatures. Therefore, the Unannealed and H₂-150 samples have great potential to be activated at 120 ℃. However, the H₂-350 sample displays a slower decay at elevated temperatures, as shown in Fig. 3(c). It is due to the structural difference between the H₂-350 sample and the Unannealed and H₂-150 samples that a higher content of Ov is present in H₂-350 [41-43]. When considering the energy required for electron detrapping from the defect state to the conduction band by thermal excitation, the electrons trapped at a shallow level rather than a deep level are most likely to be excited to the conduction band [44-46]. It is well known that CO₂ reduction is a multi-electron reduction process. The electrons located at the conduction band of TiO₂ are thermodynamically allowed to reduce CO₂ to CO. However, the electrons trapped at a deep level are unlikely to participate in CO₂ reduction due to them being in lower energy state than that required for CO production. Therefore, the slower decay of photoconductivity in H₂-350 is believed to partly result from the weaker reduction ability of the electrons at a deep energy level. The results are consistent with the enhanced activity in the photothermal catalysis results of H₂-150 as compared to the photothermal catalysis results of H₂-350 in Fig. 2. A sample of H₂-550 was also prepared, and its photoconductivity was measured (Fig. S5). The decay trend of the photoconductivity of H₂-550 under the atmosphere of CO₂ and N₂ was the same as that of H₂-350. Hence, H₂-350 was regarded as a standard sample with a high concentration of Ov for a photoconductivity study.

To further prove that the faster decay of photoconductivity results from the catalytic reduction of CO₂ rather than charge recombination, photoconductivity measurements under a N₂ atmosphere was performed. As shown in Fig. 4, the decay of thermal coupled photoconductivity of the H₂-150 and the H₂-350 samples was slower in a N₂ atmosphere compared to a CO₂ atmosphere. The H₂-150 sample, in particular, shows a significant difference in the decay of photoconductivity under CO₂ as compared to the N₂ atmosphere. However, this slower decay of photoconductivity under a N₂ atmosphere is not observed for unannealed TiO₂. The mechanism diagram for this is shown in Scheme S1. According to the analysis of the changes in photoconductivity in the CO₂ atmosphere and N₂ atmosphere, the results can be interpreted in two ways. On the one hand, the charge recombination of the H₂-150 sample and the H₂-350 sample does not decrease upon thermal activation, since it benefits from the introduction of Ov. On the other hand, the photogenerated electrons trapped by Ov can be released to the conduction band. This detrapping of electrons can accelerate surface reactions.

Several conclusions can be drawn above from the photoconductivity test: (1) Thermal coupled photoconductivity is higher than room-temperature photoconductivity for the unannealed TiO₂ and Ov-TiO₂; (2) The accelerated decay of photoconductivity for Ov-TiO₂ in the simulated photothermal reaction implies a promoted charge-involved surface reaction; (3) The slower decay of photoconductivity for Ov-TiO₂ implies thermal detrapping of electrons at Ov.

To correlate thermal coupled photoconductivity with photothermal catalysis, the mechanism shown in Scheme 3 was proposed. Upon photoexcitation, the electrons of the valence band can be excited to the conduction band or trapped by shallow Ov located below the conduction band. The surface adsorbed CO₂ can be reduced by coupled electrons/protons. At this stage, the thermal coupled photoconductivity was increased in comparison to room temperature photoconductivity, due to the combined contribution of photo-induced excitation and thermal-induced detrapping of electrons. When the UV light is switched off, the trapped electrons at Ov gradually recombine with nearby holes at room temperature. The stored electrons seldom transfer to the surface and participate in CO₂ reduction, which leads to a slow electron consumption rate by CO₂. Therefore, a slower decay of photoconductivity is observed. At a higher temperature, the trapped electrons at Ov tend to escape to the conduction band and are then transferred to the surface, which leads to a faster electron consumption rate by CO₂. Hence, a faster decay of photoconductivity is observed.

In summary, thermal coupled photoconductivity was measured to obtain an understanding of the photothermal catalytic reduction of CO₂. Photothermal catalysis shows that the introduction of Ov can significantly enhance the CO yield over Ov-TiO₂. In comparison to room temperature photoconductivity, thermal coupled photoconductivity of highly active Ov-TiO₂ exhibits an accelerated decay in a CO₂ atmosphere and a decelerated decay in a N₂ atmosphere. With regards to the correlation between thermal coupled photoconductivity and photothermal catalysis, an accelerated electron transfer to CO₂ and the detrapping of electrons to the conduction band of TiO₂ are confirmed, which provides a reasonable explanation for the superior performance of Ov-TiO₂ in photothermal catalysis. This study on thermal coupled photoconductivity could represent a promising method for understanding photothermal catalysis.