Isotope Effects in Photocatalysis: An Underexplored Issue

In order to improve the performance of well-established photocatalysts and to develop new potential photocatalyst materials, an understanding of the underlying mechanisms of photocatalytic reactions is of the utmost importance. An often neglected method for studying the mechanism is the investigation of isotope effects. Although just a few studies related to isotope effects exist, it has been shown to be a powerful tool for exploring mechanisms of photocatalytic processes. Most of the reports are focused on TiO2, which is the most studied photocatalyst, while there is a lack of data for other photocatalyst materials. This mini-review represents an overview of research utilizing isotope effects in the area of photocatalysis. The benefits and the importance of these studies will be highlighted, and the potential for these processes to be applied for the study of further photocatalytic reactions and different photocatalyst materials will be shown.


■ INTRODUCTION
Semiconductor photocatalysis is a versatile technology that has been applied to a broad range of applications from treatment of contaminated water and air to energy conversion and storage. 1a−e In designing and developing this process for practical commercial applications, it is critical to have a robust understanding of fundamental mechanistic processes that are occurring on the surface of the semiconductor material. 1e−j A broad range of physical and chemical methods have been used in developing our understanding of these surface processes over the past four decades. 1j−o One area that has not been applied to the same extent is the application of isotope effects to probe photocatalytic processes and mechanisms.
For this purpose, it is always necessary to perform two sets of experiments. In one experimental run one species (e.g., the photocatalyst) is labeled, and in a further run the same reaction is performed but with the same species unlabeled. This allows a comparison of both runs to make conclusions related to the mechanism of the investigated reaction.
The labeling of reactant molecules or the photocatalyst allows us to study the transfer of atoms between both species. For a correct interpretation of the obtained results, several points need to be considered. Besides a reaction during illumination, also a reaction in the dark needs to be performed since it might be possible that an exchange of atoms occurs spontaneously between the photocatalyst and the reactant molecules, which could lead to wrong conclusions for the reaction under illumination. Further, it is important to consider all possible reaction products that might be produced. For example, if CO 2 is the reaction product and the incorporation of 18 O is expected, it is necessary to detect C 16  In the case of the investigation of solvent isotope effects, the effect of the exchange of H 2 O by D 2 O on the rate constant is investigated. It can be recommended not only to perform experiments in pure H 2 O or D 2 O but also to consider mixtures of both solvents.
This paper explores the areas in which these techniques have been used successfully to date and also looks at the scope for more extensive application of such methods to studying the photocatalytic process promoted by semiconductor materials. In this mini-review, works from our groups as well as works from other groups are considered.

■ ISOTOPIC LABELING OF SEMICONDUCTOR PHOTOCATALYST MATERIALS AND TARGET MOLECULES FOR REACTION
Isotopic labeling is a method to investigate the incorporation of atoms from reactant molecules into the surface of the photocatalyst and vice versa during a reaction, which allows the mechanism of chemical processes on surfaces to be followed. In the case of TiO 2 as a photocatalyst, oxygen labeling (exchange of 16  which proves the incorporation of 18 O 2− into the surface of TiO 2 . Furthermore, they were able to confirm that only one surface oxygen anion is involved in the exchange at a time. The proposed mechanism of the oxygen isotopic exchange (OIE), as described by Pichat et al., 3 is depicted in Figure 1. After the excitation with light, with an energy equal to or higher than the band gap energy of TiO 2 , electron−hole pairs are formed. The conduction band electrons are able to reduce 18  •− radical anion, while Murata et al. 5 were able to confirm the formation of this species. Equation 1 describes the reaction mechanism based on the formation of an O 3 •− intermediate, 6 which thus rules out a direct involvement of a superoxide species as proposed by Pichat et al. 3 Courbon et al. 2 found that the activity for the photooxidation of isobutane correlates with the activity for the OIE and thus concluded that an O •− species is involved in both reactions, which supports the mechanism.
The isotopic-labeled oxygen can be used also simultaneously with other unlabeled molecules to investigate the effect of the OIE. For example, Liao et al. 7 reported that during UV illumination of a TiO 2 surface no oxygen exchange between 18 O 2 and adsorbed CO occurs, while for adsorbed CO 2 , CO 3 , and HCOO species an oxygen exchange was observed.
A further approach is to use directly isotopic-labeled reactant molecules instead of 18  Besides isotopic labeling of the reactant molecules, a further method is to label the photocatalyst. 18  O), which confirms the incorporation of lattice oxygen into the products. Henderson 14 also investigated the same reaction at an 18 O-enriched (110) TiO 2 crystal in the dark. Similar as for the (100) crystal, the transfer of lattice oxygen to the product molecules was observed. Bogdanoff and Alonso-Vante 11 studied the photoelectrooxidation of formic acid in the presence of 18 O-enriched TiO 2 , but no incorporation of 18 O was observed in the detected CO 2 molecules. Civišet al. 15 reported for formic acid in contact with isotopically pure Ti 18 O 2 during illumination that no oxygen exchange occurs between the oxygen atoms of Ti 18 O 2 and the formic acid during adsorption and decomposition since strongly bonded formate species inhibit the exchange. Although no exchange takes place, C 16    with adsorbed HCl and water, no OIE was observed. Civišet al. 12 investigated the same interface in the dark and during illumination. As shown in Figure 3, an involvement of oxygen vacancies in the OIE reaction in the dark was found, while the whole process was found to be very fast. One oxygen atom of each C 16   quadrupole mass spectrometry (QMS) signal ratio as compared to the dark could be detected, which turned back to the initial value after switching off the light ( Figure 4). In contrast, by applying Ti 16 O 2 instead, independent from the illumination, no change in the QMS signal ratio appeared. Since in the case of Ti 18 O 2 16 O 18 O was evolved, it could be concluded that the photooxidation of water proceeds via a bridging oxygen from the lattice of TiO 2 , which is incorporated in the oxygen molecules. In the initial step, a 2-fold-coordinated (symbol: >) bridging oxygen (>O br 2− ) (eq 2) or a 2-fold-coordinated protonated bridging oxygen (>OH br − ) (eq 3) reacts with a photogenerated hole, leading to the formation of a 1-fold coordinated (symbol: −) bridging oxygen radical (−O br

•−
) and a 1-fold coordinated bridging hydroxyl radical (−OH br • ), respectively. The further steps yielding molecular oxygen according to the water redox photooxidation (WRP) mechanism are described elsewhere. 17 Melchers et al. 18 employed Ti 18 O 2 to analyze the mechanism of the anaerobic acetaldehyde degradation during illumination.
In a previous study of the same authors, the incorporation of lattice oxygen into acetate after the adsorption of acetaldehyde was proposed, which resulted in the formation of CO 2 and CH 4 . 19 The comparison of the C 16 Figure 5). Consequently, the incorporation of lattice oxygen into the product molecules, and thus the proposed mechanism ( Figure 6), could be proven. Montoya et al. 20 studied the anaerobic oxidation of benzene in aqueous solutions in the presence of Ti 18 O 2 and Ag + as electron scavenger. Ti 18 O 2 with an unlabeled hydrated surface (Figure 7), employing H 2 16 O, was used to distinguish between two possible reaction pathways. Either hydroxyl radicals are generated from adsorbed water species ( 16 OH ads • ) or lattice     anion can also act as a hole scavenger, which finally results in the formation of a phenol molecule that contains an 16  O molecule (eq 12). Ag + ions are able to react with the free photogenerated electrons, which leads to the formation of metallic silver (eq 13). In eq 14, the complete process is summarized, which shows the exchange of an 18 ■ SOLVENT ISOTOPE EFFECTS IN PHOTOCATALYSIS Solvent isotope studies have been used extensively by organic chemists for decades. In this process, the relative rates of a reaction are compared, when they are carried out in normal water and deuterated water or "heavy" water. The solvent isotope effect (SIE) is the ratio of the rate constant in the "heavy" water (k D ) to that observed in normal water (k H ): Typically the reactions in the heavy water solvent are significantly slower, and hence the rate constants are lower than those observed in normal water. The slower reaction rate in the deuterated solvent is due to the fact that the deuterated solvent has a lower vibrational zero-point energy, and hence a greater activation energy is required to dissociate the OD bond   ACS Omega http://pubs.acs.org/journal/acsodf Mini-Review compared to OH bonds. Consequently, the rates are slower in deuterated solvents, which are involved in reactions. Cunningham and Srijaranai were the first to report the use of this technique for a semiconductor photocatalytic process in 1988. 22 In their investigation of the photocatalytic degradation of isopropanol (IPA) using TiO 2 , they observed a primary solvent isotope effect of 3. It was proposed that the reduced rate of IPA destruction in D 2 O was a result of the lower quantum efficiency for the formation of OD • radicals on the TiO 2 surface in the heavy water solvent. As a result of this, there would be a lower number of OD • radicals on the TiO 2 surface, which would be available to attack the isopropanol. On the basis of this proposal, they suggested that the photogeneration of hydroxyl radicals was the rate-determining step for the photocatalytic process.
Robertson et al. also observed a solvent isotope effect of 3 for the photocatalytic destruction of the cyanobacterial toxin microcystin-LR using a P25 TiO 2 photocatalyst. 23a The solvent isotope effect observed by Cunningham for the decomposition of IPA was the same as that reported by Robertson for the cyanotoxin, considering the substantial difference in structure and the molecular mass of the two substrates. Robertson suggested that this was also a confirmation of Cunningham and Srijaranai's proposal that the hydroxyl radical generation on the photocatalyst surface was also the rate-determining step for the photocatalytic reaction.
In a subsequent study, Robertson and co-workers investigated the solvent isotope effect on the degradation of microcystin-LR (MC-LR) and another cyanobacterial chemical metabolite, geosmin (GSM), using a Hombikat K01/C TiO 2 photocatalyst. 23b In this case a solvent isotope effect of 1.5 was observed for microcystin and geosmin, 23b which was approximately 50% lower than that found in the previous studies by Robertson et al. and Cunningham and Srijaranai (Table 1).
In this study, Robertson et al. proposed that the solvent isotope effect observed for both molecules was mediated via hydroxyl radicals, generated from the subsequent reduction of the superoxide radical anion, produced at the conduction band. After being generated, the superoxide would be hydrated or deuterated by the solvent to form a hydroperoxide ion (eq 16). The hydroperoxide ions may then interact to form hydrogen peroxide, which would then generate OH • (or OD • ) radicals following an electron transfer from the conduction band again. This may be rate determining since O 2 has to be generated at the conduction band prior to the interaction with the solvent and the subsequent formation of OD • or OH • species (eqs 17 and 18).  18) It was suggested that the observed solvent isotope effect could be a result of the rate of the reaction of the solvent with superoxide species rather than by the rate of reaction of OH • (or OD • ) on the microcystin or geosmin. If the isotope reaction depended on this latter reaction, one would expect it to be the same no matter what photocatalyst or species being oxidized was utilized. An interesting observation is the fact that the solvent isotope effect is approximately 3 for P-25 and approximately 1.5 for K01/C. They also suggested that since similar kinetic solvent isotope effects were observed for different substrate molecules on the same photocatalyst materials the interaction of the solvent with the photocatalyst and the rate of oxidation of the solvent were probably the rate-determining steps for the photocatalytic reaction, as opposed to conduction band reduction of oxygen as previously proposed by Gerischer and Heller. 24 Furthermore, Robertson et al. proposed that the reason the kinetic solvent isotope effect observed in this subsequent work was smaller than that in their previous study and Cunningham's work was due to the fact that different photocatalyst materials were employed, 23b and hence the effect was likely to be dependent on the photocatalyst material.
Belhadj et al. used solvent isotope effects to study the adsorption of water and deuterium oxide on TiO 2 surfaces in the dark and under UV(A) irradiation using in situ ATR-FTIR spectroscopy under aerobic and anaerobic conditions. 25a Under dark conditions in a mixture of H 2 O and D 2 O solvents, an isotopic exchange was found to occur on the surface of the TiO 2 material. Following irradiation with UV(A) light, the quantity of both OH and OD groups was found to be increasing in the presence of molecular oxygen. Additionally, hydroperoxide was generated through a photocatalytic process under aerobic conditions, which was believed to be produced as a result of the reduction of molecular oxygen adsorbed at the TiO 2 surface by the photogenerated conduction band electrons, as opposed to being generated via water oxidation from valence band holes. It was also demonstrated from the spectroscopic studies that under conditions where the percentage of H 2 O was significantly less than that of D 2 O there was an exchange of solvent groups on the TiO 2 surface with the OD − ions, having a stronger adsorption affinity to the photocatalyst compared to the OH − ions. Following illumination with UV light, both OH and OD groups were generated on the photocatalyst surface in the presence of oxygen. The generation of these groups also increased the hydrophilicity of the TiO 2 surface. If the experiment was conducted under either a nitrogen or argon atmosphere, there was no evidence of the formation of OH and OD groups, and the hydrophilicity was inhibited (Figure 9). This result indicated that under UV irradiation oxygen played a critical role in both the photocatalytic response and the photoinduced hydrophilicity.
In a subsequent study, the adsorption and photocatalytic degradation of acetate on TiO 2 surfaces was investigated in H 2 O and D 2 O by both attenuated total reflection Fourier transformed infrared spectroscopy (ATR-FTIR) and EPR spectroscopy. 25b Different interactions between the adsorbed acetate and OD groups resulted from the isotopic exchange on the TiO 2 surface following adsorption of D 2 O. The interaction

ACS Omega
http://pubs.acs.org/journal/acsodf Mini-Review of the acetate with the TiO 2 surface was found to be strongly influenced by the pH, and a range of surface complexes with the acetate were observed to form. Under acidic conditions, the formation of a bidentate structure involving two distinct Ti atoms appeared to be the preferred complex structure. At pH values close to the point of zero charge for the TiO 2 , the acetate favored a monodentate complex, formatted through adsorption to the positively charged TiO 2 anatase material ( Figure 10). Following irradiation with UV(A) light, hydroxyl radicals were observed under alkaline conditions, while methoxy radicals were generated under acidic conditions. Two different degradation pathways were proposed for the acetate under acidic and alkaline conditions (Figure 11), which were supported by the experimental studies performed using ATR-FTIR and EPR spectroscopy. Overall, the results of the EPR study suggested that under alkaline conditions acetate degradation was mainly promoted by attack by valence band generated hydroxyl radicals. Under acidic conditions, the degradation appeared to occur via direct oxidation via photogenerated valence band holes.
Solvent isotope studies were also used for the investigation of the simultaneous photocatalytic degradation of formaldehyde and hydrogen evolution on a platinized TiO 2 material under an oxygen-free atmosphere. 25c Using QMS and ATR-FTIR spectroscopy for analysis, the main reaction products obtained from the photocatalytic degradation of 20% formaldehyde were hydrogen and carbon dioxide in a ratio of 2 to 1. From the solvent isotope study, it was found that the rate of mineralization of formaldehyde to CO 2 is significantly reduced with increasing concentration of D 2 O. Following the investigation of the solvent isotope effect on the system using ATR-FTIR analysis, it was proposed that the formaldehyde oxidation was promoted by attack by OD • radicals, formed from the reaction with the photogenerated valence band hole. This reaction generated a surface-adsorbed deuterated formic acid (HCOOD), which subsequently underwent further oxidation by valence band holes in a photo-Kolbe-type reaction. The photogenerated conduction band electrons were proposed to simultaneously reduce H + and D + , originating from both formaldehyde and D 2 O, to form molecular HD. The yield of the HD gas was found to be strongly influenced by the solvent and was maximized when the ratio of H 2 O:D 2 O was 20%:80%. The proposed mechanism for the simultaneous hydrogen production and formaldehyde oxidation in the presence of D 2

■ SCOPE FOR RESEARCH DEVELOPMENT AND FOCUS
Each of the studies considered in this paper related to the labeling of photocatalyst materials and target molecules have utilized TiO 2 as the photocatalyst. This is also the case for the studies dealing with solvent isotope effects. Although TiO 2 is the most studied photocatalyst material, there has only been a relatively small number of reports related to isotope effects. As has been detailed above, the use of these isotope studies has enabled important insights to be gained for photocatalytic reactions on TiO 2 , but there is still significant scope for further studies using these techniques. For future research, while also investigating the isotope effects of the photocatalytic decomposition of other substrates on TiO 2 materials, it would be important to extend the studies to other photocatalyst materials as well. For example, Fe 2 O 3 , WO 3 , CdS, C 3 N 4 , and SrTiO 3 are known as photocatalytically active materials, while the labeling of the catalyst or reactant molecules would allow us to get deeper insights into the corresponding mechanisms using these materials.
With respect to TiO 2 itself, there are many more reaction mechanisms that should be investigated based on solvent isotope effects. In particular, the involvement of terminal hydroxyl or oxygen radicals in the photocatalytic mechanism, as one of the initial steps, might be either established or refuted.
The technique could also be used for kinetic studies, particularly in the case of rapidly decomposing intermediates, which may be more easily followed in the heavy water solvent.

■ CONCLUSION
The investigation of isotope effects represents a powerful tool in the area of photocatalysis to study the mechanisms of the reactions occurring on the surface of photocatalyst materials. Using this technique, it is possible to elucidate whether photocatalyst surface atoms are transferred into product molecules, while also the incorporation of atoms from reactant molecules into the surface of the photocatalyst can be observed. Furthermore, by exchanging H 2 O by D 2 O, the ratio of the rate constants between both solvents can be followed to investigate whether the generation of hydroxyl radicals is the rate-determining step of a reaction. Using D 2 O as a solvent has a further advantage since it allows the determination of whether hydrogen atoms in product molecules originate from reactant molecules or from the solvent. Consequently, it is important to perform such studies to allow the determination of the mechanistic pathway of the photocatalytic process. It should be noted that isotope effect studies are, however, not enough as the sole process to provide such detail, but in combination with other tools, they can provide important information on such processes.
In conclusion, it is clear that the application of isotope studies is a versatile and useful tool for studying photocatalytic reactions; however, the technique has been only applied in a relatively small number of investigations. There is therefore great scope for the further application of this technique in the field of semiconductor photocatalysis, and it is anticipated that this will be an area of growing interest within the photocatalytic research community over the next few years.
Technology, Xi'an (Peoples Republic of China). His research topics including photocatalysis, photoelectrochemistry, solar chemistry, and photochemistry focused on the synthesis and physical-chemical properties of semiconductor and metal nanoparticles. His more than 500 publications have been cited over 61 000 times (h-index: 99). He is a Visiting Research Professor at Queens University Belfast, DeTao Master of Photocatalysis, Nanomaterials and Energy Applications, and holds a Guest Professorship at Tianjin University (China) and several visiting professorships in China, Malaysia, Saudi Arabia, and Britain.
Prof. Dr. Peter K.J. Robertson is Professor of Energy and Environmental Engineering at Queen's University Belfast. His particular expertise is in the area of photocatalytic technology for both energy and water sustainability. This work has encompassed basic research on photocatalysis through to pilot process development for water treatment and for solar energy conversion and storage. He is also a member of the UK Engineering and Physical Science Research Council's (EPSRC) Scientific Advisory Committee on Energy and the SUPERGEN High Level Group. He is also a member of sub Panel 8, Chemistry for REF2021 the UK National research assessment exercise. He is Chair of the Royal Irish Academy's Physical, Chemical and Mathematical Sciences Committee. He holds visiting professor positions at the University of St Andrews, Ulster University, and Robert Gordon University. He is a Chartered Engineer, Chartered Scientist, and Chartered Chemist, a Fellow of the Royal Society of Chemistry, the Institute of Chemistry in Ireland, the Energy Institute, and an Associate Fellow of the Institution of Chemical Engineers.