TiO2表面光催化基元过程

郭 庆 周传耀 马志博 任泽峰 樊红军 杨学明,*

（1中国科学院大连化学物理研究所，分子反应动力学国家重点实验室，辽宁 大连 116023；2北京大学量子材料科学中心和物理学院，北京 100871）

TiO2表面光催化基元过程

郭 庆1周传耀1马志博1任泽峰2樊红军1杨学明1,*

（1中国科学院大连化学物理研究所，分子反应动力学国家重点实验室，辽宁 大连 116023；2北京大学量子材料科学中心和物理学院，北京 100871）

在过去的几十年里，得益于二氧化钛（TiO2）作为光催化剂在光催化分解水、污染物降解方面的潜在应用，人们对TiO2光催化剂的开发、改良以及TiO2表面光催化机理的基础研究方面都投入了巨大的精力。因此，在超高真空环境下，利用不同的实验和理论方法，人们对TiO2表面（特别是金红石TiO2（110）表面）的热催化和光催化过程进行了大量的研究，以此来获得上述重要反应中的一些机理性的信息。本文中，将从TiO2的物质结构以及电子结构开始，然后着重介绍TiO2表面光生电荷（电子和空穴）的传输、捕获以及电子转移动力学方面的进展。在此基础上，总结了甲醇在金红石TiO2（110）、TiO2（011）以及锐钛矿TiO2（101）表面光化学基元反应过程的一些实验结果。这些结果不仅能增进我们对表面光催化基元过程的认识，同时也能激励我们进一步去研究表面光催化基元过程。最后，基于现有光化学实验结果，简短地讨论了我们对光催化反应机理的一点看法，并提出了一个可能的光催化模型，这可以引起人们对光催化反应机理更全面的思考。

二氧化钛；光催化；电子-空穴分离；非绝热过程；基态势能面

1 Introduction

Today, most of the energy we use comes from fossil fuels created over several millions of years, and they are being consumed much more rapidly, moreover burning fossil fuels leads to pollution and many environmental impacts. Thus, developing renewable and environmentally friendly energy sources is of the greatest importance for human beings and is one of the biggest challenges facing the scientific community. The most important renewable and clean energy source on the earth is the solar energy. Solar energy contains light in a wide spectral range, in particular infrared, visible, and ultraviolet light, which can be converted into electrical energy and clean chemical. For example, when sunlight is used for the water splitting reaction to produce hydrogen and oxygen, clean hydrogen energy can be provided for combustion with little pollution. Thus, it is our priority to develop ecologically clean, environmentally harmonious, and sustainable, safe, and energy efficient chemical technologies using solar energy1. Therefore, heterogeneous photocatalysis has expanded rapidly within the last four decades, having undergone various developments especially in relation to energy and environment.

Heterogeneous photocatalysis, which can be described as the acceleration of photoreaction in the presence of a catalyst, is initiated by the photoexcitation of the phototocatalyst to generate pairs of electrons and holes. The charge carriers are then used to drive the chemical reactions on the surface of photocatalyst. These processes must avoid causing chemical changes for the photocatalysts2. Understanding the fundamental processes in heterogeneous photocatalysis is important for this field of study. Generally, heterogeneous photocatalysis on TiO2photocatalysts is driven by photogenerated electrons or holes, suggesting that the reactions occur on the excited electronic state after photogenerated electrons or holes transfer to reactants on the surface. Recently, however, detailed studies of photocatalysis on TiO2single crystal surfaces suggest that photocatalytic reactions may also possibly occur on the ground state, which may be driven by the energy from electron-hole recombination. It is therefore useful for us to review these studies to seek a more in-depth understanding of the entire heterogeneous photocatalysis process.

In the contexts of history and research, interest in heterogeneous photocatalysis can be traced back to many decades since the discovery of photocatalytic water splitting on TiO2by Fujishima and Honda in 19723. From this time, extensive research has been carried out to produce hydrogen from water in oxidation reduction reactions using a variety of semiconductor catalyst materials4. Among the many different photocatalysts, TiO2, as a model of semiconductor photocatalyst, has been most widely investigated because of its exceptionally efficient photoactivity, high chemical stability, low cost, as well as its low toxicity for both humans and the environment. In recent years, interest in photocatalysis has been extended to the use of semiconductor materials as photocatalysts to reduce or oxidize reagents to obtain useful target products including hydrogen and hydrocarbons, and to remove pollutants and bacteria on wall surfaces and in air and water5.

Even though extensive studies have been focused on the characterization of photocatalytic processes and the development of photocatalytic materials, studies of fundamental mechanisms about detailed surface photocatalytic processes are still lacking. Fundamental issues need to be investigated systematically in typical photocatalytic reaction processes. Firstly, we need to learn how the energy of excitons generated by photon excitation drives a surface chemical reaction. Secondly, we need to investigate the elementary steps of a photocatalytic chemical reaction and thus the whole photocatalytic chemical process. Only when these processes are understood, a clear physical picture of surface photocatalysis at the molecular level can be achieved. A deeper understanding of photocatalysis at the molecular level is crucial for the future development of the photocatalysis field. In this review, we will provide you with a comprehensive overview of the mechanisms and dynamics of photocatalytic reactions on a few TiO2surfaces and TiO2-based photocatalysts. A detailed discussion on a possible new picture of photocatalysis from a different perspective is also provided at the end of this review.

2 TiO2semiconductor photocatalysts

2.1 Crystal structures of TiO2

TiO2exists mainly in three different crystalline habits: rutile, anatase, and brookite, all of which can be described in terms of distorted TiO6octahedra with different symmetries or arrangements （Fig.1）6. The different characteristics of the Ti―O bonds play a very important role in the structural and electronic features of different phase forms of TiO27. The detailed knowledge of properties of TiO2has been comprehensively reviewed by Diebold8. Rutile is the most stable form, while anatase and brookite are the metastable ones and are readily transformed to rutile when heated. Li and co-workers9–11found that the transformation of TiO2anatase nanoparticles to TiO2rutile nano-particles occurs via a bulk phase transformation mechanism. The transformation of brookite to rutile undergoes brookite to anatase transition first followed by transforming to rutile12,13. Furthermore, the particle size of the initial TiO2plays an important role in the transformation of anatase and brookite to rutile. For example, at temperatures ranging between 325 and 750 °C, when the particle sizes > 35 nm, rutile is the most stable, however, anatase is the most stable phase with particle sizes < 11 nm, and brookite is the most stable phase between 11 and 35 nm14.

Fig.1 Representations of the TiO2anatase （a）， rutile （b）， and brookite （c） forms

On the basis of the different physical/chemical properties of the three polymorphs of TiO2, they have been enormously applied in different fields. Whereas, generally, the majority of the research on TiO2is mainly performed on rutile and anatase phases. To date, research works on TiO2have been carried out for yielding H2from water splitting to biomass reforming and industrial waste reforming. One limitation of TiO2photocatalyst is its wide band gap （i.e., band gap for bulk materials: rutile 3.02 eV corresponding to 410 nm and anatase 3.20 eV corresponding to 384 nm）, which makes TiO2very sensitive to the UV light, whereas, the UV light in natural sunlight represents only 5%–8% of the solar spectrum. However, TiO2serves as an ideal model semiconductor photocatalyst for understanding and exploring the elemental processes and mechanisms of photocatalysis, which might be useful for the development of more efficient photocatalysts for photocatalytic reactions such as water splitting and CO2reduction.

The photocatalytic activities of rutile and anatase forms of TiO2are quite different. Because of the lower thermodynamic stability of anatase, the synthesis of anatase single crystals has only been accomplished by few laboratories15. For TiO2nanocrystalline, anatase TiO2usually has a much higher surface area than rutile, led to enhanced adsorption capability and more active sites （i.e., oxygen vacancies）. In addition, due to the existence of more oxygen vacancies on anatase, it can generate more efficient charge separation than rutile. Due to these advantages, the photocatalytic activity of anatase is usually much higher than rutile5. However, on anatase single crystal surfaces （such as anatase TiO2（101）16and TiO2（001）17）, the surface point defects can hardly be produced through high temperature UHV annealing, whereas, with the same treatment, rutile single crystal surfaces can be highly reduced. Therefore, the physical/ chemical properties of a material may be quite different from its various phase structures, as well as the particle sizes.

2.2 Electronic structures of TiO2

Photocatalysis can be described as the acceleration of photoreaction in the presence of a catalyst18. Fig.2 shows a typical model that is comprised of the essential features in the process of photocatalytic water splitting to produce hydrogen and oxygen with a photocatalyst； it also describes the preferred energy band level alignment between energy band edges of the photocatalyst and the reactant to drive a redox reaction via charge transfer from the catalyst to the reactant19.

Fig.2 Schematic presentation of the typical light-induced decomposition of water molecules wherein （i） hydrogen is produced via a reduction process at the conduction band assisted by a photoelectron and （ii） oxygen is produced via an oxidation process assisted by a photoinduced hole in the valence band of the photocatalyst

In general, the valence band （VB） and conduction band （CB）of rutile, anatase, and brookite TiO2are mainly constructed by O 2p and Ti 3d states, respectively. Furthermore, a detailed molecular orbital bonding structure of TiO2not only serves as aclear insight in the chemical bonds formed between O and Ti of TiO2but also provides an excellent tool to visualize the hybridization between the different energy levels within the cation and anion. Asahi et al.20have investigated the chemical bonding in anatase TiO2using first-principles calculations carefully, and then a molecular orbital bonding structure of anatase TiO2could be further constructed from these orbitals （Fig.3）, it now more clearly shows that （i） the valence band of anatase TiO2is comprised of O pπ（higher energy region）, pσand pπ（intermediate energy region）, and pσ（lower energy region）； the conduction band is comprised of Ti 3d and 4s, and the lower energy regions of the conduction band are constructed by the degenerate eg-like and 3-fold t2g-like states resulting from the crystal field splitting of Ti 3d. Similarly, a simplified molecular orbital bonding structure for rutile TiO2has been summarized by van de Krol and Grätzel21, as shown in Fig.4.

Fig.3 Detailed molecular orbital bonding diagram of anatase TiO2

Fig.4 Simplified molecular orbital diagram of rutile TiO2summarized by van de Krol

3 Charge transport and trapping

A photocatalytic reaction is typically initiated by a photo adsorption process in the bulk （i.e., subsurface）. However, in contrary to metals with a continuum of electronic states, TiO2has a wide band gap between the top of the VB and the bottom of the CB, where no electron energy level is available. The band structure, including the positions of CB and VB and the bandgap （denoted as Eg）, is one of the important properties for TiO2photocatalyst, and other semiconductor photocatalysts as well, because it determines the photocatalytic property （including light absorption property and the redox capability） of a semiconductor photocatalyst. Once photo excitation occurs across the band gap, the electrons in VB will be excited to CB, leaving the holes in VB. This electron-hole pair generation process in TiO2can be illustrated as follows:

Fig.5 Important processes in the surface photocatalysis processes: photoexcitation of electron-hole pair， charge transfer processes， bulk and surface recombination processes， electron and charge induced chemistry on surfaces

These photogenerated charge carriers may further be involved in the following three possible processes: （i） the charge carriers successfully diffuse to the surface of TiO2, and then flow to adsorbed reactants or to the solvent （processes （1） and（3） in Fig.5）； （ii） the charge carriers are trapped by the surface and/or bulk defect sites of TiO2； （iii） the separated charge carriers recombine and release the energy in the form of photon or heat in bulk and/or on the surface （processes （2） and （4） in Fig.5）. Usually, the latter two processes are referred as deexcitation processes, because the charge carriers do not have thechance to drive photocatalytic reactions. Only the charge carriers that transfer to the reactants could initiate photocatalytic reactions. The surface and bulk defect sites of TiO2may serve as the recombination centers for the charge carriers, which will reduce the efficiency of the photocatalytic reaction.

As mentioned above, photocatalytic reactions occur on the surface, photo generated charge carriers must tranfer to the surface and be stabilized on the surface for the electron/hole transfer to the reactants （and not for recombination events）. This situation raises questions about how charge carriers transfer through a lattice to a surface. What are the relative timescales for charge separation, thermalization, transport and trapping What are the typical charge trapping surface sites However, characterizations of exciton （electron/hole） thermalization, lifetimes, transport kinetics, trapping and quenching are all difficult tasks22.

3.1 Charge separation

Generally, the efficiency of a photocatalyst can be evaluated by the photonic efficiency ζ, which is defined as the rate of the products formation divided by the incident photon flux23. Unfortunately, the ζ of TiO2, as well as other semiconductor photocatalysts is found to be rather small. In fact, the ζ measured by time-resolved spectroscopic methods demonstrates that about 90% of the photogenerated charge carriers recombine rapidly after excitation. This is suggested to be the main reason for the relatively low ζ （ < 10%） for most semiconductor-based photocatalytic reactions.

Therefore, a series of approaches are carried out to enhance the efficiency of charge separation in TiO2, such as supported charge trap, heterojunction, and bulk dopant. Of course, the efficiency of charge separation in TiO2may be determined by the energy of the charge carriers, as well as temperature. For example, ‘hot’ electrons and deep holes may be more likely to separate than charge carriers generated with near-bandgap excitation. Whereas, based on the work of Berger et al.24, higher temperatures can decrease the efficiency of charge separation at some trapping sites.

3.2 Charge thermalization

It is well known that the higher the potential energy of the electron （or hole） the more reductive （or oxidative） capacity there is22. Thus, desired electron （or hole） transfer reactions can be selected by tuning the energy of charge carriers. However, thermalization of charge carriers is rapid. For example, using two-photon photoemission （2PPE） spectroscopy, Gundlach and coworkers25,26found that thermalization of the injected electrons from two adsorbed dyes adsorbed to rutile TiO2（110） occurs on the 10 fs timescale. This result illustrates that excess potential energy of hot electrons is lost to the lattice via strong coupling, leading to reduce the potential advantage. Photoemission study of Ag clusters on rutile TiO2（110）27as a function of injection electron energy and photoluminescence studies of rutile TiO2（110）28,29with 3.35 eV photon irradiation shows that a constant energy of the emitted photons is observed （at and below the band-gap energy of 3.05 eV） at both studies, independent of the energy of the exciting electron or photon. This also provides additional evidence for the rapid thermalization of electrons. Furthermore, experimental investigations on photo stimulated desorption of O2rutile TiO2（110） surface indicate that the photodesorption yield and translational energy of O2are independent of the excitation photon energy, but instead depending on the photon flux and adsorption coefficient30,31. Using time-resolved terahertz spectroscopy, Turner et al.32found that the thermalization of injected electrons from a surface R535 dye to P25 nanoparticles to the CB edge occurs on the～300 fs timescale.

Conversely, hole thermalization is always accompanied by electron thermalization, which makes it difficult to investigate hole thermalization. Through examining the rate of salicylate photolysis on suspended TiO2nanoparticles （≤ 5 nm） as a function of excitation energy, Grela and coworkers33found that higher energy photons are always more efficient, leading to higher quantum yields. Similarly, these authors showed that the quantum yields for 3-nitrophenol photooxidation34were greater for higher energy photons as well （Fig.6）. These results were in accord with the action spectroscopy study of ‘hot’ hole oxidation of various aromatic molecules on colloidal TiO2by this group35. Thus, these authors suggested that hole transfer can precede hole thermalization in TiO2.

Fig.6 Quantum yields for photooxidation of 3-nitrophenol （NP） over suspended TiO2as a function of “excess photon energy” （defined as the photon energy in excess of the TiO2bandgap energy）

Using femtosecond transient reflecting grating （TRG） method, Morishita and coworkers36showed that the hole transfer time at TiO2/SCN–interfaces is on the 110–690 fs timescale, which is much faster than the theoretical estimation （1.8 ns by Marcus theory） and previous result （5 μs by Bahnemann et al.37）, indicating that the non-thermalized holes in the VB of TiO2may be involved in interface reactions. Similarly, the lifetimes of electron and hole trapping in nanocrystalline films of anatase （particles of ≤ 2 nm） have been measured by Tamaki et al.38,39using transient absorption method. On the basis of their assignments of the absorption spectrum of trapped holes, theseauthors proposed that in some cases hole trapping preceded hole thermalization by up to 100 ps. However, the charge carrier thermalization has no concern with chemical reactions actually.

3.3 Electron trapping

Electron is usually reported to be trapped as localized Ti3+sites in the TiO2lattice. After being trapped, electrons will lose some energy. The energy loss is estimated in a range between 0.1 and 1 eV22,40, which agrees with previous photoemission results for the surface of TiO2single crystal with electronic defects8. However, so far, to identify the electron trap sites is still a long standing issue. Initially it is thought that electron trap sites are located on the TiO2surfaces22. Whereas, trapping electron in the bulk also is observed later41,42. Some theoretical calculations43,44claim that electrons prefer trapping in the bulk（subsurface） rather than trapping at the surface. Conversely, other theoretical work45proposes that the under-coordinated Ti cation sites on surfaces are the most stable electron trap sites. Similarly, other groups find that electron traps locate at TiO2-TiO2particle interfaces46or at grain boundaries47. The electron trap stabilities of TiO2can be affected by the pore size. For example, Planelles and Movilla48found that at low temperature, when the pore size < 14 nm, the effect from the pore size will be observed, with the trapping stability depending mostly on the pore diameter and the chosen permittivity, as well as temperature.

Although a good electron trap ability can promote hole-mediated photochemistry efficiently, electron trapping inhibits some processes depending on rapid electron transport （such as photovoltaic application）. Thus, rapid electron detrapping is more important in these applications. Electron detrapping is usually considered to two different type: （1） polaronic hopping, which refers as a ‘trap-to-trap’ hopping process49； （2） real （complete）detrapping, in which the localized trapped electrons are promoted into conduction band and become real free electrons. Detrapping can be achieved via thermal activation50–52, or by nonthermal activation such as sub-bandgap light excitation53,54. Usually, thermal activation can easily induce both polaronic hopping and real detrapping. Based on sensitization studies at room temperature on TiO2films by van de Lagemaat and coworkers55, they pointed out that the average thermal detrapping time is on the order of 10 ns. Conversely, non-thermal detrapping is mainly controlled by the optical properties of the trap state. According to transient absorbance （TA） studies, Shkrob and Sauer53reported that either 532 nm or 1064 nm light could induce the electron detrapping in colloidal anatase TiO2. Beermann54and Komaguchi56et al. also observed the similar results. While, Komaguchi and coworkers investigated the Ti3+electron paramagnetic resonance （EPR） signals in thermally reduced anatase, rutiles, and mixture P25 during and after visible light irradiation （Fig.7）. Upon irradiation, for reduced anatase and rutile, the EPR signals decreased obviously during visible light irradiation, which was resulted from detrapping of electrons from the localized Ti3+to free delocalized conduction band electrons. After the light was removed, the EPR signals restored for both anatase and rutile. However, in the case of P25, the EPR signal increased after visible light irradiation. These authors proposed that when activated, the electron detrapped from the anatase component in the P25, but retrapped by the rutile, giving a stronger signal than anatase at the same spin density.

ESR: electron spin resonance. Reprinted with permission from Ref.56. Copyright 2006 Elsevier.

3.4 Hole trapping

On the basis of various EPR studies22, a surface Ti4+－O–site is suggested to be the most likely site for hole trapping, where the hole stays at an under-coordinated surface oxygen atom. However, it is still unclear which is the preferred trapping site between bridged oxygen and oxo oxygen. Additionally, some groups have also found evidence for subsurface hole traps40,57,58. Recently, Kerisit et al.59carried out calculations on the sites of hole （and electron） trapping at the unrelaxed rutile TiO2（110）surface. They found that holes preferred being trapped in the near-surface region, and for comparing, electrons favored subsurface sites, which is in accord with the early findings by Shapovalov et al.60.

A dye sensitization and electron injection could provide a good method for electrons trapping investigation, but for hole, it is hard to achieve a facile method to inject holes into the TiO2VB. Therefore, most of investigation focused on the band-toband excitation process, since it is crucial for understanding TiO2photocatalytic chemistry. Although optical spectroscopic methods have been used for detecting trapped holes based on comparisons with and without exposure to hole scavengers, the excited state level of an optical transition is not known. For example, by time resolved transient absorption studies, Tamai and coworkers61pointed out that the timescale of a hole trapping at the surface of anatase nanoparticles is ～50 fs. Similarly, Tamaki and coworkers40,62assigned the spectral region at ～500 nm to excitation of trapped holes from excitation of nanocrystallinefilms of anatase （diameter of particles ≤ 2 nm） using the same method. These authors also estimated a hole trapping timescale of ～200 fs for ‘hot’ holes, and then, the trapped holes thermalized over the next 100+ ps.

Pineering work on quantifying concentration of hole traps in the near-surface region of a reduced rutile TiO2（110） surface were first carried out by Thompson and Yates63by varying the photon flux and using photodesorption yield of18O2to monitor the rate of hole delivery to the surface （Fig.8）. At low fluxes（branch A）, the photodesorption yield of18O2was significantly lower than at high fluxes （branch B）, which is speculated to arise from preferential trapping of holes away from adsorbed O2molecules, presumably in the subsurface. At Fhv（crit） （Fhvis photon flux）, the holes transferring to the surface fills all the hole traps, resulting in a more efficient branch B photodesorption process. By assuming that the fluence of photons at Fhv（crit） is entirely consumed by trap-filling by holes, an upper limit for the hole trap density of ～2.5 × 1018cm–3（～0.003% of the atomic sites in the bulk lattice） is estimated by these authors. While, Berger and coworkers64found that reduced TiO2, formed by vacuum annealing is less efficient in hole trapping than oxidized TiO2, indicating that these bulk traps are related to reduced centers resulting from vacuum annealing.

Fig.8 Dependence of the18O2photodesorption yield on the incident UV photon flux for18O2adsorbed on the rutile TiO2（110） surface at 110 K

3.5 Charge recombination

Similar to all the other semiconductors, the fast electron-hole recombination is a main reason to limit the photocatalytic activity of TiO2. Although the recombination process is undesirable, understanding this process can provide a deep insight into the charge carrier dynamics in TiO2. The recombination can be classified into two types: irradiative and non-irradiative recombination. The irradiative recombination usually associates with light emission. Non-irradiative pathway involves in energy release via phonon emission. Generally, the non-radiative recombination is more favorable than the radiative recombination. While, the ratio between the two recombination types can be strongly affected by chemical adsorbates （i.e., electron or hole scavengers） or additives （such as supported metal particles, dopants or interfacial heterojunctions）.

Radiative decay, also called photoluminescence （PL）, is usually at sub-band gap energies owing to carrier relaxation and trapping. As a result, PL spectrum is a good medium that can provide valuable information in the energetic distribution of sub-bandgap states and the dynamics of TiO2related to relaxation and trapping of charge carrier. For both rutile and anatase, the PL spectra are detected from the visible to the near-IR wavelength range. Whereas, the PL for anatase is related to the decay of self-trapped excitons, and, for rutile, to intrinsic sites（such as defects, surface atoms or impurities） or to surface bound species65–68. For example, Knorr and coworkers69found that the character PL peak for the anatase nanocrystalline is located at ～540 nm, while the character PL peak of the rutile nanocrystalline is in the near-infrared range centered at 840 nm. Using transient absorption （TA）, time-resolved PL, and photoconductivity （PC） measurements, the electron and hole relaxation dynamics in both rutile and anatase TiO2single crystals have been investigated by Yamada and Kanemitsu70. For anatase crystals, due to its indirect-gap band structure, the interaction between electrons and holes is very weak. Therefore, the decay for the electrons in anatase occurs on the microsecond time scale, while the lifetime of holes is on the nanosecond time scale, indicating the presence of multiple carrier trapping processes. Whereas, the decay for electrons and holes in rutile occurs on a few nanoseconds timescale. These authors also proposed that the long electron lifetime is related to the high photoactivity of anatase. Furthermore, using time-resolved PL spectroscopy, Dozzi et al.71found that with the increasing long lifetime component of the PL signal, the photoactivity of the anatase TiO2materials doped with fluorine increases, indicating that long-living charge carriers are beneficial in photoactivity.

As mentioned above, the non-radiative recombination is dominating in nature for TiO2, which occurs upon the rapid release of heat, making it difficult to be direct measure. Thus, it is usually detected by indirect methods, such as time-resolved photoacoustic spectroscopy （TRPAS）. Using this method, Leytner and Hupp72found that about 60% of all trapped electron-hole pairs in their colloidal anatase samples recombine on the time scale of about 25 ns, releasing 154 kJ·mol–1of energy as heat. Conversely, irradiative recombination is not expected to contribute significantly to non-radiative recombination.

As in the case of photoluminescence, surfaces, defects, adsorbates, and impurities all likely play roles in promoting or inhibiting charge recombination22. However, temperature can strongly affect the ratio between irradiative and non-irradiative recombination in TiO2. For example, Murakami and coworkers73studied the photoluminescence decay profile of the anatase TiO2（001） film epitaxial growth on LaAlO3（100）, and pointed out that recombination at higher temperature （ > 200 K）will prefer the non-radiative owing to the more active lattice motion. From above studies, it is clear that non-irradiative recombination is a main pathway for excited state energy relaxation, it is therefore interesting to question whether this largeamount of energy released via phonon coupling will have a significant effect on surface chemical reactions. If it does have an effect, it means that ground state surface reaction of adsorbed molecules is possible, which is obviously very different from the hole or electron induced reaction mechanisms. It is therefore very necessary to investigate this important issue of how charge recombination will affect the surface chemical reactions.

4 Electron transfer dynamics

Electron transfer is an interfacial phenomenon between a surface-trapped charge carrier and a chemisorbed or physisorbed species, which is a fundamental process that is relevant to many applications in heterogeneous photocatalysis. The electron transfer process is essentially interpreted as a donor-acceptor（D-A） model initiated by a photo-excitation event. There are commonly four types of electron transfer processes. The first type is electron transfer from the CB of TiO2to an electron acceptor （1） in Fig.5）； the second （3） in Fig.5） is hole transfer from the VB of TiO2to a hole acceptor （i.e., electron transfer from an electron donor into a VB hole state）； the third involves electron transfer from a donor into the CB of TiO2； and the fourth involves hole transfer from a donor into the VB of TiO2, which is not commonly observed for TiO2（i.e., electron transfer from a TiO2VB state to an unoccupied state of an acceptor）. The first two types of electron transfer correspond to photo excitation occurring in TiO2and the last two correspond to photo excitation occurring in an adsorbate or attached entity （such as dopants or impurities）. It is worth noting that not all electron transfer processes lead to chemistry （i.e., old bond brokenness and new bond formation） or involve ground state configurations.

4.1 TiO2conduction band to electron acceptor

Generally, a photocatalytic redox reaction can be divided into two half reactions, the oxidation and reduction half reactions. A reduction reaction may be initiated by an electron in the CB transferring to the lowest unoccupied molecular orbital（LUMO） of an adsorbed species. However, the coupling of the CB electron to the LUMO of the adsorbed species should be strong enough and energetically down-hill to inhibit back-electron transfer to the higher DOS in the TiO2CB. For example, Weitz and coworkers74,75have investigated the substrate-mediated photodesorption of CH3X （X = I, Br） on rutile TiO2（110）. These authors found that the CB electrons generated from bandgap excitation of TiO2transfer to these adsorbed CH3X molecules do not drive appreciable photochemical reaction. On the contrary, an Antoniewicz-type desorption process occurs due to the rapid back-electron transfer to the surfaces76. As shown in Fig.9, starting from the neutral ground state （and following the dashed lines）, an electron attachment causes a Frank-Condon transition from the ground state of the adsorbatesubstrate system to antibonding states, which leads to the excited/ground ion state. The ion created in the excitation process sees an attractive image potential. The new equilibrium position of the ionized adsorbate is considerably closer to the surface than the ground-state adsorbate equilibrium position. Thus, the newly formed ion relaxes toward the surface, and is neutralized via back-electron transfer to the surface, which is described by a vertical jump from the upper to the lower curve. During the vertical jump process, the kinetic energies （KES） of the ion and the neutral adsorbate are the same, so the total energy of the neutral adsorbate is the kinetic energy before neutralization plus the potential energy of the potential energy at the position of the neutralization. If the total energy is greater than the adsorption energy, the adsorbate will desorb from the surface after breaking the adsorption bond. The desorption events strongly depend on the potential-energy surfaces of ion and molecule, and the lifetime of excited states as well.

Fig.9 Schematic model of the Antoniewicz model for excitation-induced desorption from surfaces The vertical dashed arrow corresponds to the initial charge transfer event, placing the adsorbate on an ‘ionized adsorbate’’ potential energy surface. Prompt re-neutralization back to the ‘‘neutral’’ adsorbed potential energy surface results in the adsorbate retaining sufficient kinetic energy （KE） for desorption. Reprinted with permission from Ref.76. Copyright 1980 American Physical Society.

Although many studies have been focused on the mechanistic details of photoreduction processes on TiO2（such as H2O and CO2photoreduction）, there are only few dynamics studies focusing on a photoreduction process. The main limitation is the unavailability of suitable molecular markers that are sensitive to electron attachment. One of the convenient molecular markers is the methyl viologen divalent cation （MV2+）. Using TA method, Asahi and coworkers77found that electron transfer from photoexcited colloidal anatase TiO2to MV2+on the surface accomplishes in the few picoseconds to nanoseconds timescale. The timescales are comparable to the electron trapping times in TiO2, suggesting that the reduction of MV2+adsorbed on the surface is likely due to trapped electrons, and not free CB electrons. While, the similar studies carried out by Martino et al.78also support these conclusions. Although many dynamics works have been done on electron transfer from TiO2to an electron accepter by molecular sensitization studies, the dynamics of real photoreduction reactions （i.e., H2O reduction） are still notwell-understood.

4.2 Electron donor to TiO2valence band hole

Similar to CB electron photoreduction, VB hole photooxidation also initiates from a band-to-band photoexcitation in TiO2leading to charge carriers with a broad range of energies, generated near or at the TiO2surface. Compared with extensive dynamic studies of CB electron transfer to acceptor states of an adsorbed species, the dynamics of hole transfer on TiO2is less investigated. Usually, thiocyanate （SCN–）, as the most effective adsorbate （or hole scavenger）, is chosen to probe the dynamics of hole transfer. Colombo and coworkers79demonstrated that an electron transfer from the SCN–to a hole in the VB of TiO2effectively competes with electron-hole recombination on an ultrafast time scale. Similarly, Yang and Tamai61found that the timescale of the hole transfer from nanosized anatase TiO2colloid to SCN–is comparable to that of the hole trapping, which is estimated to be in the few femtoseconds timescale （～ 50 fs）, indicating that the hole transfer process occurs on a similar timescale. The similar conclusion is further verified by Morishita36and Furube80et al.

However, the timescale of hole transfer varies remarkably based on the different kinds of hole scavengers. For example, using I–as a hole scavenger, Rabani and coworkers81found that hole transfer from packed films of 5 nm TiO2particles to I–occurs within less than 10 ns. In contrast, the adsorbed alcohols（methanol or isopropanol） are more efficient for recombination after trapping holes. Other groups have observed reactive trapped holes in TiO2with adsorbed alcohols. Shkrob and coworkers82found that on the timescale of the 355 nm laser pulse width （3.3 ns）, 50%–60% of the generated holes react with chemisorbed diols and carbohydrates on TiO2nanoparticle （pH = 4, ～4.6 nm diameter） rather than recombine. While, the examination of hole transfer to glycerol bound to TiO2nanoparticle by Shkrob and Sauer53obtained the similar results.

On the basis of above studies, the hole transfer rates seem not to be the key factor for the efficient photooxidations on TiO2. For example, the single electron transfer from SCN–to a VB hole is fast and efficient. However, actually, an overall reaction essentially requires many charge transfer processes, and not all the steps are as fast and efficient as SCN–photooxidation. And every electron transfer step in a photooxidation reaction is governed by the properties of VB hole and the electron donor. Thus, how to identify and improve the slow and inefficient steps is still a key challenge.

4.3 Electron donor to TiO2conduction band

The dynamics of electron transfer from an electron donor to the TiO2CB has been extensively investigated. Normally, a dye molecule is chosen as an electron donor. With dye sensitization, the molecular dye acts as the ‘photocatalyst’, efficiently absorbing light to generate charge carriers, and then the excited electrons inject into the TiO2CB with high efficiency. Because of the well-known optical properties and energy levels of dye chromophores and the ability to spectroscopically track the injected electron, ultrafast dynamic studies of the electron transfer processes with dye/TiO2systems are tailor-made. A series of results22reveal that the coupling of electronic states between the excited electronic states of dyes and the continuous TiO2CB are very strong, leading to electron transfer between the two are very fast with injection yields of ～100% for many different dyes. In some cases, due to the strong coupling, back-electron transfer processes are also fast （see the work of Lian et al.83）, whereas, in almost all cases, the continuous CB takes over in thermalizing the injected electron, so that the process does not occur.

4.4 TiO2valence band to acceptor hole

Up to now, no examples of studies about hole generated in an adsorbed species injection into the VB of TiO2have been reported. In this case, the highest occupied molecular orbital（HOMO） of the chosen dyes should be below the energy level of the TiO2VB edge, and the electronic excited states of dye should reside mid-gap （far below the energy level of the TiO2CB edge） to avoid electron transfer from dye to the TiO2CB. In early studies, Hagfeldt and coworkers84,85have constructed such a model solar cell system using NiO and proper dyes, but a similar model system has not been identified for TiO2. Similar to back-electron transfer for photoreduction, back-hole transfer for photooxidation between an adsorbate and the TiO2VB （i.e., a VB hole generated on TiO2transfers transiently to an adsorbate and flows back the TiO2VB） provides another potential ground for examining hole injection to the TiO2VB, which may be very universal in photooxidation with TiO2catalysts. However, reliable photochemical ‘markers’ are required for identifying and studying such a hole transfer process.

5 Mechanisms of methanol photocatalysis

The mechanism for a whole photocatalysis process is usually quite complicated. In this section, we will describe recent studies of detailed photocatalytic reactions of one important molecule on various TiO2surfaces: methanol （hole scavenger）, which are representative photoreduction reactions on TiO2.

Photocatalytic reactions of alcohols on TiO2are important in a number of technological applications. Due to the prominent use of alcohols in reforming reactions to produce H2, biomass conversion to fuels and useful synthetic chemicals, and oxidative remediation of organic wastes, great interest has been attracted to the reactivity of TiO2with alcohols, especially methanol （CH3OH） and ethanol. Since surface science studies on single crystal surfaces under UHV conditions could provide fundamental insights into these important processes, both thermal- and photo-chemistry on the TiO2especially rutile TiO2（110） surface have been investigated with a variety of experimental and theoretical approaches.

5.1 Adsorption of methanol on TiO2surfaces

On reduced rutile TiO2（110）, five prominent features at 150, 165, 295, 350, and 480 K appeared at the temperature pro-grammed desorption （TPD） spectra of CH3OH. The 150 and 165 K peaks were due to multilayer desorption, the peak at 295 K was assigned to the desorption of molecular CH3OH on Ti5csites, and the broad peak around 480 K was attributed to the recombinative desorption of dissociated CH3OH at Ov′s and other defect sites. Using high resolution electron energy loss spectroscopy （HREELS） and static secondary ion mass spectroscopy（SSIMS） spectra, Henderson et al.87proposed that the 350 K shoulder was assigned to CH3OH dissociated at non-defective sites of the surface, probably Ti5c, according to similar behavior of the 350 and 480 K peaks following electron bombardment. However, no obvious 350 K shoulder was observed in recent TPD studies of CH3OH on the reduced rutile TiO2（110）surface with gentle surface treatment88. The assignment of the 350 K shoulder is still inconclusive.

Petek et al.89,90provided some indirect evidence that part of CH3OH dissociatively adsorbed at Ti5csites. Using 2PPE, an empty wet electron state at ～ （2.3 ± 0.2） eV above Fermi level（EF） was detected on both reduced and stoichiometric rutile TiO2（110） surfaces. However, in the case of H2O, this wet electron state could only be observed on reduced rutile TiO2（110）91. While on the H2O covered stoichiometric rutile TiO2surface, this state totally disappears. By analogy with the properties of the excited electronic state at H2O/rutile TiO2（110） interface, those authors argued that the excited electronic state at CH3OH/rutile TiO2（110） interface is due to the partial dissociation of CH3OH on the surface.

A recent scanning tunnel microscope （STM） work by Zhang et al.92showed that CH3OH dissociated at the Ovsites spontaneously to form methoxy （CH3O） and hydroxyl on the neighboring Obsites. Further investigation found that molecular CH3OH can diffuse along the Ti5crows at room temperature （RT）, indicating that CH3OH molecularly adsorbs at the Ti5csites. Theoretical work about CH3OH adsorption on the surface93–95suggested the molecular state of CH3OH was nearly iso-energetic to the dissociated state with the molecular state slightly more stable, and the energy barrier for CH3OH dissociation was very small. This suggested that molecular adsorption of CH3OH on rutile TiO2（110） was more stable, which was consistent with previous STM observation at both liquid nitrogen temperature96and RT92. Theoretical result also showed that dissociative adsorption of CH3OH on Ovsites is more favorable than molecular adsorption by 0.5 eV94.

5.2 Photocatalytic chemistry of methanol on TiO2surfaces

CH3OH/TiO2is an important benchmark system because of the remarkable enhancement of CH3OH in photocatalytic hydrogen production with TiO2catalyst97, the potential applications of CH3OH in photocatalytic selective oxidation98, environmental photocatalysis99, and photocatalytic reforming reactions100. Meanwhile, as one of the simplest alcohols, CH3OH is often chosen as a probe for the fundamental studies of photocatalytic properties of oxide surfaces.

Recently, Using 2PPE, STM, and density functional theory（DFT） calculations, the methanol/rutile TiO2（110） system has been investigated by Zhou et al.96. These authors also detected an excited electronic state at about 2.4 eV above EFby irradiating 0.77 ML （monolayer）101,102of CH3OH covered TiO2（110）surface with 400 nm femtosecond laser. This was consistent with the previous 2PPE measurements by Petek and coworkers90, where the unoccupied excited state was assigned to be an intrinsic “wet electron state” on the CH3OH covered rutile TiO2（110） surface. In contrary, Zhou et al. found that the excited state was absent immediately after the laser irradiation（Fig.10A）. It increased as the laser irradiation time increased and saturated after a 15-min irradiation. The result unambiguously demonstrated that the excited state was a photo induced surface state rather than a wet electron state. The lifetime of the excited state was determined to be ～ 20 fs by time-resolved two-pulse correlation103, which was in accord with previous results90.

Fig.10 （A） 2PPE spectra for CH3OH adsorbed stoichiometric rutile TiO2（110） after the interface was exposed to the probe light for different periods； （B） time-dependent excited resonance signal integrated from （A） and the fractal-like kinetics model fitting

In order to understand this photoinduced process, further STM experiments were employed by Zhou et al.96to reveal the nature of the photochemical changes detected by 2PPE（Fig.11）. Fig.11A shows an typical STM image of a bare rutile TiO2（110） surface （Ovwas labeled as “BBOv” in the figure）, thebright and dark lines corresponded to the Ti5cand Obrows respectively, while the bright spots on the dark lines were the Ov′s. With 0.02 ML of CH3OH adsorption, the majority of the CH3OH molecules adsorbed on the Ti5csites （CH3OH/Ti5c） and imaged as clear bright round spots （Fig.11B）. These bright round spots could diffuse along the Ti5crow as a whole or desorb driven by the STM tip, indicating that CH3OH molecules molecularly adsorbed on the Ti5csites, which was consistent with previous STM results.92After a 10-minute UV （ < 400 nm）irradiation, most of the bright round spots became stretched（marked by black arrows in Fig.11C）. By manipulating one of these altered species （labeled “m1” in Fig.11C） with the STM tip, the stretched spots were separated into two parts. One of the components left on the Obsites was identified to be OHbgroups104, while the other was likely a CH3O on a Ti5csite（CH3O/Ti5c）. The result clearly indicated that CH3OH molecules were dissociated after UV irradiation.

Fig.11 STM images （acquired at bias of 1.0 V and set point current of 10 photocatalyzed dissociation of methanol pA， size of 7.3 nm × 7.3 nm）showing the photocatalyzed dissociation of methanol

Given the coverage difference between the 2PPE and STM experiments, additional 2PPE measurements at low CH3OH coverage were carried out by the authors. The similar results were observed when the CH3OH coverage was changed from 0.77 to 0.12 ML, implying the photochemical process was independent on the CH3OH coverage. Therefore, the photoinduced excited state detected by 2PPE arose from photodissociation of CH3OH on this surface. Additional DFT calculations on the electronic structure found that molecularly adsorbed CH3OH had little effect on the density of states （DOS） of Ti 3d, however, the adsorbate-substrate interaction became much stronger when CH3OH was dissociated. The strong interaction between CH3O and the Ti5cion resulted in the appearance of a new electronic state which was centered at 2.5 eV above the EF, consistent with the 2PPE measurements.

Zhou et al. then suggested that photodissociation of CH3OH was via O―H bond dissocaition with the H atom transfer to an adjacent Obsite,

Since the excited state arose from the dissociation of CH3OH, the irradiation time dependent excited state signal illustrated the kinetics of dissociation of CH3OH on the rutile TiO2（110） surface. The integrated time dependent excited state signal（Fig.10B） could not be described by a single exponential model, while a fractal-like kinetics model （equation （2）105,106simulated the data well.

where I0and I are signal intensities after reaction equilibrium and during reaction, respectively. k0is the rate at t = 1 s, and h is equal to 1 – ds/2, where dsis the spectral dimension of the heterogeneous reaction media. The fractal-like kinetics of photochemistry on TiO2surface was reported to result from the trapping and detrapping of charge carriers.

As a model system, photodissociation of CD3OD on TiO2has been carried out by Zhou et al.107to study the photocatalytic activities of the stoichiometric and reduced rutile TiO2（110）surfaces. On both surface, the excited state at about 2.5 eV above EFwas observed, whereas, the rise times of the excited state signal on the two surfaces were significantly different（Fig.12）. The increase of the excited state signal on the reduced surface was more than 10 times faster than that on the stoichiometric surface, which was attributed to the concentration of point defects on these two surfaces. These authors also suggested that the surface and/or subsurface defects could accelerate methanol photolysis on rutile TiO2（110） surface. Unfortunately, due to the difficulty in characterization of the density of the subsurface defects, it was difficult to unravel whether the surface defects or subsurface defects play a more important role in the acceleration of methanol photolysis on TiO2108. The observed higher photoactivity on reduced TiO2surface may be due to the enhanced light absorption via self Ti3+doping109.

As metioned above, dissociation of the O－H bond on CH3OH covered rutile TiO2（110） surfaces under UV irradiation can be demonstrated with 2PPE and STM methods. However,both these two methods were not able to identify the species on the surface. Recently, photolysis of CH3OH on rutile TiO2（110） has been investigated using TPD method by Shen and Henderson88,110,111. These authors proposed that the adsorption state of CH3OH was crucial to its photochemistry on TiO2. By co-adsorption of CH3OH and O2on rutile TiO2（110） to prepare CH3O covered surface,

Fig.12 Normalized time-dependent signal of the excited resonance of 0.77 ML CD3OD covered stoiciometric （olive circle） and reduced （blue circle） rutile TiO2（110） surface and the fractal-like kinetics model fitting （red line）

where OHtis a thermal hydroxyl group adsorbed on a Ti5csite. Henderson et al.88concluded that CH3O, rather than molecular CH3OH, was the photoactive species in CH3OH photolysis on TiO2. Their results also suggested that formaldehyde （CH2O）was produced from photochemistry of CH3O on the Ti5csites, which was initiated by defects, coadsorbed O/Ti5cor OHtgroups, but not by Ovsites, and not from photodissociation of molecular CH3OH/Ti5c.

Guo et al.112have extended the study of CH3OH photolysis using TPD method. These authors systematically investigated the photo-induced dissociation of partially deuterated methanol（CD3OH） on rutile TiO2（110） with 400 nm laser irradiation. After adsorption of 0.5 ML of CD3OH on rutile TiO2（110）, the observed CD3OH signal at m/z = 33 （Fig.13A） decreased monotonically with irradiation time, suggesting that the CD3OH molecules on the Ti5csites were dissociated. Concomitant to the decrease of the CD3OH, a new peak at 270 K appeared and increased in the TPD spectra of m/z = 32 （CD2O+） （Fig.13B）, which was due to the desorption of molecularly CD2O adsorbed on the Ti5csites. During CD2O formation, the dissociated H/D atoms transferred to adjacent Obsites, which was further confirmed by the characteristic recombinative desorption of bridging hydroxyls around 460 K.113,114

Fig.13 （A） Typical TPD spectra collected at m/z = 33 （CD2OH+）following different laser irradiation times. CD2OH+is formed by dissociative ionization of the desorbed parent CD3OH molecule in the electron-bombardment ionizer. （B） Typical TPD spectra collected at m/z = 32 （CD2O+） following different laser irradiation times. The m/z = 32 （CD2O+） signal has three components: the parent ion signal of formaldehyde （CD2O）， as well as the ion-fragment signals of the parent CD3OH molecule and the photocatalyzed CD3OD product.

DFT calculations on the ground state potential surface showed that the O－H dissociation of CH3OH was slightly endoergic （0.03 eV）, with a very small barrier （0.25 eV）. While, the C－H dissociation to produce CH2O and H was highly endoergic by 1.03 eV, with a high barrier of 1.57 eV （Fig.14）.

Similar energetics for the methanol/rutile TiO2（110） system have been also reported by Lang et al.115. A recent study on the reverse reaction of CH2O and Hbatoms by Mao et al.116observed spontaneous and STM tip induced recombination of CH2O and Hbatoms, suggesting the lower stability of the CH2O and Hbproducts. This was consistent with the energetics for the methanol/rutile TiO2（110） system mentioned above.

After a series of studies, Guo and co-workers112proposed that photolysis of CD3OH into CD2O at Ti5csites （CD2O/Ti5c） occurs in a two-step process, leaving H and D atoms on the adjacent Obsites:

These experimental results clearly demonstrated that molecular CH3OH is photodissociated on rutile TiO2（110） occurred on Ti5csites rather than CH3O.

Photodissociation of CH3OH on rutile TiO2（110） can provide a method to control the surface hydroxyls without affecting thesubsurface defects. Mao et al.117thus reinvestigated the long standing issue of the origin of band gap states in rutile TiO2（110） by measuring the correlation between the DOS of the band gap states and the concentration of surface hydroxyls. The intensity of band gap states measured by ultra-violet photoelectron spectroscopy （UPS） showed a linear relationship with surface hydroxyls characterized by TPD with a small intercept, suggesting that the band gap states is mainly contribute by surface defects.

Fig.14 Calculated energetics of the two-step dissociation of CD3OH on the rutile TiO2（110） surface

In addition to CH2O formation from CH3OH photolysis on rutile TiO2（110）, the secondary product, methyl formate（HCOOCH3）, was further observed by Guo et al.118on the 0.5 ML of CH3OH covered rutile TiO2（110） surface. As shown in Fig.15, for the first 10 minutes of irradiation, CH2O was formed rapidly to a maximum, while little HCOOCH3was produced. Longer irradiation times lead to a steady depletion of CH2O and a concomitant increase of that of HCOOCH3. Thus, the formation of HCOOCH3appeared to be directly correlated with the decrease of CH2O. Similar results were also observed by Phillips119and Yuan120et al.

According to previous catalytic investigations of CH3OH reforming to HCOOCH3over copper-based catalysts,121–123Guo et al.118suggested that HCOOCH3product was yielded through cross coupling of CH2O and CH3O:

While, Phillips et al.119proposed that HCOOCH3was formed through cross coupling of CH3O and a formyl （HCO） intermediate:

Recently, Lang and co-workers115have investigated the two pathways （reactions （6） and （7）） for CH3OH oxidation into HCOOCH3on both perfect and defect rutile TiO2（110） based on DFT calculations. The much smaller energy barrier for hemiacetal channel （CH3O + CH2O） suggested that CH3OH oxidation through the intermediate hemiacetal to produce HCOOCH3on both surfaces facilely proceeded.

Photoactivity of rutile TiO2（110） - （1 × 1） and TiO2（011） - （2 × 1） has recently been tested using CH3OH photolysis by Mao et al.124, with the latter has been reported to be of higher photoactivity towards photooxidation reactions125,126. In this work, CH2O, HCOOCH3products have been detected on both surfaces. Whereas, the rate of CH3OH dissociation on rutile TiO2（011） - （2 × 1） was only 42% of that on rutile TiO2（110） -（1 × 1）, contradicting with previous reports in aqueous environments.125

Conventionally, the effect of surface structure on the activity of a catalyst in heterogeneous reactions is greatly related to the percentage of under-coordinated atoms. The higher percentage of the under-coordinated surface atoms usually means the more reactive surface. The results observed by Mao et al. did not comply with this rule, since the coverage of Ti5con rutile TiO2（011）-（2 × 1） is twice of that on rutile TiO2（110） - （1 × 1）. In addition, this result was inconsistent with a previous electronic structure study which proposed that the electron trapping and electron-hole separation properties of rutile TiO2（011） - （2 × 1） surface were more efficient than that of rutile TiO2（110） - （1 × 1）due to the higher energy level of the band gap states of rutile TiO2（011） - （2 × 1）127. Through calculating the energetics of the CH3OH/TiO2system124, the authors found that the rate determining step of CH3OH photolysis on both surfaces were the C－H bond breaking, however, the barrier of this step was 0.2 eV higher on rutile TiO2（011） - （2 × 1）.

From above experimental studies, it is no doubt that photolysis of CH3OH on rutile TiO2（110） occurs through multiple elementary steps. After CH3OH dissociation, the dissociated H-atoms transfer to Obsites nearby, which prompts the question of whether molecular hydrogen could be formed during CH3OH photolysis. Recent experimental investigation on the photolysis of CD3OD on rutile TiO2（110） by Xu and co-workers128suggested that molecular hydrogen （D2） was not formed during CD3OD photolysis. Whereas, during TPD process, dissociated D-atoms on Obsites desorbed in the forms of D2O and D2（Fig.16）.

Fig.15 （A） TPD spectra acquired at m/z = 31 （CH3O+） after 0.5 ML CH3OH was adsorbed on rutile TiO2（110） at 120 K and irradiated at 400 nm for various times； （B） TPD spectra acquired at 30（CH2O+） after 0.5 ML CH3OH was adsorbed on rutile TiO2（110） at 120 K and irradiated at 400 nm for various times； （C） yields of CH3OH，CH2O， and HCOOCH3as a function of irradiation time，derived from data in Fig.15 （A） and （B）

As laser irradiation time increases, both the D2O and D2desorption signal increases and peaks gradually shift toward lower temperature. However, the desorption of D2starts from ～375 K, which is ～50 K higher than that of D2O, indicating that D2formation is more difficult than D2O formation. The yields of these two products suggested that D2formation is just a minor channel on this surface.

It is well known that the thermalization of charge carriers generated from UV irradiation of TiO2surface takes place in the hundred-fs scale. Then, photochemistry on TiO2is driven by separated electrons or holes located at the band edges. The reaction rate will strongly depend on the photon flux rather than irradiation wavelength22. As a result, excess potential energy of charge carriers was lost to the lattice through strong coupling with phonon modes of lattice, leading to reduce the potential advantage gained by the specificity in the absorption event. Xu and coworkers129have examined this idea by measuring the initial dissociation rate and the initial quantum yield of CH3OH on rutile TiO2（110） with two irradiation wavelengths, 355 and 266 nm using the yield of H-atom production through monitoring the recombinative desorption of H2O product. As shown in Fig.17, the initial rate of H2O formation was found to be strongly dependent on photon energy, with the initial rate being about 100 times higher at 266 nm than at 355 nm. This obvious result was clearly in conflict with the traditional electron-hole photocatalysis model that the excess potential energy of charge carriers in TiO2was relaxed to lattice via strong coupling with phonon modes rapidly, resulting in that photocatalysis was driven by the charge carriers with the same energy. On the basis of the result, the authors speculated that CH3OH photolysis maybe occur on the ground electronic state, where increased converted photon energy should be more efficient in driving chemical reactions.

Kavan and coworkers130found that anatase is the more photoactive polymorph for the photocatalytic hydrogen production from H2O oxidation. However, the availability of high quality and large size single crystal anatase is very limited in early years. As yet, only a few experimental studies131,132of CH3OH chemistry on a well-defined anatase TiO2（101） surface have been carried out. On reduced anatase TiO2（101）, five desorption peaks are observed at 142, 188, 270, 410, and 650 K in TPD spectra of CH3OH （Fig.18）131. Compared with the TPD spectra of CH3OH on rutile TiO2（110）, the 142 and 188 K peaks are attributed to multilayer desorption, the peak at 270 K is due to the desorption of CH3OH molecularly adsorbed on the Ti5csites, and the broad tail around 410 K is assigned to the recombinative desorption of dissociated CH3OH on defect sites. The 650 K peak is most likely due to the disproportionation of CH3O at Ti5csites112,spectra collected at m/z = 60as a function of irradiation time irradiation time. The m/z = 60signal is from the parent ion signal of HCOOCH3molecule.

Fig.16 （A） Typical TPD spectra collected at m/z = 20 （D2O+） following different laser irradiation times at 400 nm， the peak （marked with *） slightly below 300 K is attributed to the dissociative ionization signal of molecular adsorbed CD3OD in the electron-impact ionizer and impurity of D2O in CD3OD； （B） typical TPD spectra collected at m/z = 4 （D2+） following different laser irradiation times at 400 nm， the peak （marked with *） slightly below 300 K is attributed to the dissociative ionization signal of molecular adsorbed CD3OD in the electron-impact ionizer. The right figures （C， D）show TPD product yields for D2O and D2as a function of irradiation time， derived from data in Fig.16 （A） and （B）.

Fig.17 Laser irradiation time dependence of the water （from BBO）TPD yield at both 355 nm （red line） and 266 nm （dark blue line）photolysis from a 0.5 ML methanol covered rutile TiO2（110） surface

Fig.18 CH3OH TPD spectra （m/z =31） spectra from various exposures of CH3OH on the anatase TiO2（101） surface at 100 K

Fig.19 0.38 ML of CH3OH were dosed to the anatase TiO2（101）surfaces at 100 K. （A） Typical TPD spectra collected at m/z = 31（CH2OH+） as a function of irradiation time with a photon flux of 1.9 × 1017photons∙cm-2∙s-1. CH2OH+is formed by dissociative ionization of the desorbed parent CH3OH molecule in the electron-bombardment ionizer. （B） Typical TPD spectra collected at m/z = 30 （CH2O+） as a function of irradiation time irradiation time. The m/z = 30 （CH2O+）signal has two components: the parent ion signal of CH2O， as well as the ion-fragment signals of the parent CH3OH molecule. （C） Typical TPD

Recently, some preliminary study of CH3OH photolysis on anatase TiO2（101） has been done by Xu et al.132using TPD method. As shown in Fig.19, the 650 K peak in m/z = 30 and 31 TPD was depleted rapidly after irradiating the surface for 5 s while the 300 K peak kept unchanged implies that CH3O has a much higher reactivity than molecular CH3OH on this surface. Similar to the photochemistry of CH3OH on rutile TiO2（110）, photocatalytic products, CH2O and HCOOCH3, were also detected as laser irradiation time increased （Fig.19（B, C）, indicating that the mechanisms for the formation of these two products are similar to that on rutile TiO2（110）112. These authors also measured the dissociated H-atoms from CH3OH photolysis on anatase TiO2（101） by collecting TPD spectra of H2O and H2at m/z = 18 and 2, respectively （Fig.20（A, B）.

Fig.20 TPD spectra collected at m/z = 2， 18， from photocatalysis of 0.38 ML of CH3OH covered anatase TiO2（101） at 100 K with a photon flux of 1.9 × 1017photons∙cm-2∙s-1

However, a rather sharp peak at about 260 K appeared at TPD spectra of m/z = 18 with laser irradiation, which was due to desorption of molecular H2O adsorbed on Ti5csites（H2O/Ti5c）, and no obvious recombinative desorption of H2O was detected at higher temperature. Concomitant to the increase of the H2O on the surface, a broad methyl radical desorption peak stretching from 400–700 K was also detected at m/z = 15, keeping the same increasing rate with that of H2O. Referring to earlier theoretical studies of CH3OH/anatase TiO2（101）by Tilocca and Selloni133, theses authors suggested that the H2O formation is likely the result from the following thermally driven exchange reaction,

Fig.21 A newly proposed photocatalysis model based on nonadiabatic chemical dynamical processes and ground state reactions. In this model，photoexcited electron-hole pairs are nonadiabatically recombined to convert the excited electronic state energy to the ground state energy， which drives the chemical reactions on the ground state surface.

Similar to the behavior of D2formation on rutile TiO2（110）, H2formation from CH3OH photolysis on anatase TiO2（101） likely occurs via thermal recombination of H atoms on Obsites following photocatalytic dissociation of CH3OH. While, compared with H2formation on the rutile TiO2（110） surface, the formation of H2and H2O are comparable on the anatase TiO2（101） surface, indicating that the H2formation process is more efficient, and it is a major reaction channel.

Referring to the recent works,33–35,112,129one surprising fact is that the experimental results of methanol, water and other organic chemicals photooxidation can not be well explained by the traditional photocatalysis model. For example, the traditional photocatalysis model cannot explain why methanol can be dissociated on rutile TiO2（110） at 400 nm irradiation while water cannot112. The strong photon energy dependence of methanol photocatalysis on rutile TiO2（110） is also not well explainable129. It is therefore fair to question whether the traditional photocatalysis model can explain all photocatalytic processes. On the basis of the results of methanol and water photocatalys on rutile TiO2（110）, a new photocatalysis model （Fig.21）134based on nonadiabatic dynamics and ground state surface reactions was proposed by Yang and coworkers, which can qualitatively explain the new experimental results obtained recently for photocatalysis of water and methanol on rutile TiO2（110）. It is clear that surface dynamics is a crucial factor in photocatalysis, which is well described in this new model. However, it is necessary to point out that this new model proposed is derived from studies of photocatalysis of methanol and water on TiO2under vacuum conditions only, the applicability of this model to other photocatalysis systems or photocatalysis under realistic conditions needs to be rigorously tested in the future. In addition, theoretical studies of surface nonadiabatic processes and surface reaction dynamics are also urgently needed in order to gain clear and important insights to the fundamental understanding of photocatalysis processes.

6 Conclusions

Based on the fundamental studies of photocatalysis processes on TiO2performed using various surface science techniques, this review highlighted some of the most significant advances made on charge transfer dynamics and photolysis of methanol on TiO2surfaces. However, numerous opportunities and challenges still remain, for example,

（1） Understanding how additives （i.e., dopants, cocatalysts, etc.） influence the photon absorption, electron transfer and thermal/non-thermal chemistry on TiO2surfaces at the molecular level.

（2） More detailed knowledge about the effect of the charge transfer process on surface photocatalytic reactions.

（3） Detailed knowledge about the charge transfer dynamics and mechanistic studies of important ‘up-hill’ reactions such as water photooxidation at the molecular level.

（4） The role of recombination energy on the photocatalytic reaction processes. In addition to these big challenges of experimental photocatalysis studies, the development of theoretical understanding of photocatalysis processes at the fundamental level from chemical dynamics perspective is also very important and potentially rewarding. Up to date, more and more new photocatalysis phenomena cannot be well explained by traditional photocatalysis model. Although a new model based on nonadiabatic dynamics and ground state surface reactions was proposed by Yang and coworkers, it is proposed based on limited studies of photocatalysis of methanol and water on TiO2under vacuum conditions. Thus, the development of theoretical understanding to gain clear and important insights to the fundamental understanding of photocatalysis processes is still a big challenge.

（1）Schneider, J.； Matsuoka, M.； Takeuchi, M.； Zhang, J.； Horiuchi, Y.； Anpo, M.； Bahnemann, D. W. Chem. Rev. 2014， 114, 9919. doi: 10.1021/cr5001892

（2）Fujishima, A.； Zhang, X.； Tryk, D. A. Surf. Sci. Rep. 2008， 63, 515. doi: 10.1016/j.surfrep.2008.10.001

（3）Fujishima, A.； Honda, K. Nature 1972， 238, 37. doi: 10.1038/238037a0

（4）Nakata, K.； Fujishima, A. J. Photochem. Photobiol. C: Photochem. Rev. 2012， 13, 169. doi: 10.1016/j.jphotochemrev.2012.06.001

（5）Ma, Y.； Wang, X.； Jia, Y.； Chen, X.； Han, H.； Li, C. Chem. Rev. 2014， 114, 9987. doi: 10.1021/cr500008u

（6）Dambournet, D.； Belharouak, I.； Amine, K. Chem. Mater. 2010，22, 1173. doi: 10.1021/cm902613h

（7）Nosheen, S.； Galasso, F. S.； Suib, S. L. Langmuir 2009， 25, 7623. doi: 10.1021/la9002719

（8）Diebold, U. Surf. Sci. Rep. 2003， 48, 53. doi: 10.1016/S0167-572900100-0

（9）Zhang, J.； Li, M. J.； Feng, Z. C.； Chen, J.； Li, C. J. Phys. Chem. B 2006， 110, 927. doi: 10.1021/jp0552473

（10）Su, W. G.； Zhang, J.； Feng, Z. C.； Chen, T.； Ying, P. L.； Li, C. J. Phys. Chem. C 2008， 112, 7710. doi: 10.1021/jp7118422

（11）Shi, J. Y.； Chen, J.； Feng, Z. C.； Chen, T.； Lian, Y. X.； Wang, X. L.； Li, C. J. Phys. Chem. C 2007， 111, 693. doi: 10.1021/jp065744z

（12）Zhang, J.； Xu, Q.； Li, M. J.； Feng, Z. C.； Li, C. J. Phys. Chem. C 2009， 113, 1698. doi: 10.1021/jp808013k

（13）Xu, Q. A.； Zhang, J.； Feng, Z. C.； Ma, Y.； Wang, X.； Li, C. Chem. -Asian J. 2010， 5, 2158. doi: 10.1002/asia.201000249

（14）Zhang, H. Z.； Banfield, J. F. J. Phys. Chem. B 2000， 104, 3481.

（15）Yang, H. G.； Sun, C. H.； Qiao, S. Z.； Zou, J.； Liu, G.； Smith, S. C.； Cheng, H. M.； Lu, G. Q. Nature. 2008， 453, 638. doi: 10.1038/nature06964

（16）He, Y.； Dulub, O.； Cheng, H.； Selloni, A.； Diebold, U. Phys. Rev. Lett. 2009， 102, 106105. doi: 10.1103/PhysRevLett.102.106105

（17）Liang, Y.； Gan, S.； Chambers, S. A.； Eltman, E. I. Phys. Rev. B 2001， 63, 235402. doi: 10.1103/PhysRevB.63.235402

（18）Tachibana, Y.； Vayssieres, L.； Durrant, J. R. Nat. Photonics. 2012， 6, 511. doi: 10.1038/nphoton.2012.175

（19）Kapilashrami, M.； Zhang, Y.； Liu, Y. S.； Hagfeldt, A.； Guo, J. Chem. Rev. 2014， 114, 9662. doi: 10.1021/cr5000893

（20）Asahi, R.； Taga, Y.； Mannstadt, W.； Freeman, A. J. Phys. Rev. B 2000， 61, 7459. doi: 10.1103/PhysRevB.61.7459

（21）van de Krol, R.； Grätzel, M. Photoelectrochemical Hydrogen Production； Springer: Heidelberg, 2012； p 16.

（22）Henderson, M. A. Surf. Sci. Rep. 2011， 66, 185. doi: 10.1016/j.surfrep.2011.01.001

（23）Hoffmann, M. R.； Martin, S. T.； Choi, W.； Bahnemann, D. W. Chem. Rev. 1995， 95, 69. doi: 10.1021/cr00033a004

（24）Berger, T.； Sterrer, M.； Diwald, O.； Knozinger, E. ChemPhysChem 2005， 6, 2104.

（25）Gundlach, L.； Felber, S.； Storck, W.； Galoppini, E.； Wei, Q.；Willig, F. Res. Chem. Intermed. 2005， 31, 39. doi: 10.1163/1568567053146841

（26）Gundlach, L.； Ernstorfer, R.； Willig, F. Phys. Rev. B 2006， 74, 035324. doi: 10.1103/PhysRevB.74.035324

（27）Nilius, N.； Ernst, N.； Freund, H. J. Chem. Phys. Lett. 2001， 349, 351. doi: 10.1016/S0009-261401232-5

（28）Yamada, Y.； Kanemitsu, Y. Phys. Rev. B 2010， 82, 113103. doi: 10.1103/PhysRevB.82.113103

（29）Yamada, Y.； Kanemitsu, Y. Phys. Status. Solidi. C 2011， 8, 104. doi: 10.1002/pssc.201000642

（30）Sporleder, D.； Wilson, D. P.； White, M. G. J. Phys. Chem. C 2009， 113, 13180. doi: 10.1021/jp901065j

（31）Diwald, O.； Thompson, T. L.； Goralski, E. G.； Walck, S. D.；Yates, J. T., Jr. J. Phys. Chem. B 2004， 108, 52. doi: 10.1021/jp030529t

（32）Turner, G. M.； Beard, M. C.； Schmuttenmaer, C. A. J. Phys. Chem. B 2002， 106, 11716. doi: 10.1021/jp025844e

（33）Grela, M. A.； Brusa, M. A.； Colussi, A. J. J. Phys. Chem. B 1997， 101, 10986. doi: 10.1021/jp972172x

（34）Grela, M. A.； Colussi, A. J. J. Phys. Chem. B 1999， 103, 2614. doi: 10.1021/jp9829492

（35）Grela, M. A.； Brusa, M. A.； Colussi, A. J. J. Phys. Chem. B 1999， 103, 6400. doi: 10.1021/jp990952v

（36）Morishita, T.； Hibara, A.； Sawada, T.； Tsuyumoto, I. J. Phys. Chem. B 1999， 103, 5984. doi: 10.1021/jp984729u

（37）Bahnemann, D. W.； Hilgendorff, M.； Memming, R. J. Phys. Chem. B 1997， 101, 4265. doi: 10.1021/jp9639915

（38）Yoshihara, T.； Katoh, R.； Furube, A.； Tamaki, Y.； Murai, M.；Hara, K.； Murata, S.； Arakawa, H.； Tachiya, M. J. Phys. Chem. B 2004， 108, 3817.

（39）Tamaki, Y.； Furube, A.； Murai, M.； Hara, K.； Katoh, R.；Tachiya, M. J. Am. Chem. Soc. 2006， 128, 416. doi: 10.1021/ja055866p

（40）Tamaki, Y.； Furube, A.； Murai, M.； Hara, K.； Katoh, R.；Tachiya, M. Phys. Chem. Chem. Phys. 2007， 9, 1453. doi: 10.1039/b617552j

（41）Nakaoka, Y.； Nosaka, Y. J. Photochem. Photobiol. A 1997， 110, 299. doi: 10.1016/S1010-603000208-6

（42）Jenkins, C. A.； Murphy, D. M. J. Phys. Chem. B 1999， 103, 101.

（43）Ganduglia-Pirovano, M. V.； Hofmann, A.； Sauer, J. Surf. Sci. Rep. 2007， 62, 219. doi: 10.1016/j.surfrep.2007.03.002

（44）Kowalski, P. M.； Camellone, M. F.； Nair, N. N.； Meyer, B.；Marx, D. Phys. Rev. Lett. 2010， 105, 146405. doi: 10.1103/PhysRevLett.105.146405

（45）Qu, Z. W.； Kroes, G. J. J. Phys. Chem. B 2006， 110, 8998. doi: 10.1021/jp056607p

（46）Miyagi, T.； Kamei, M.； Mitsuhashi, T.； Ishigaki, T.； Yamazaki, A. Chem. Phys. Lett. 2004， 390, 399. doi: 10.1016/j.cplett.2004.04.042

（47）Kamei, M.； Miyagi, T.； Ishigaki, T. Chem. Phys. Lett. 2005，407, 209. doi: 10.1016/j.cplett.2005.03.075

（48）Planelles, J.； Movilla, J. L. Phys. Rev. B 2006， 73, 235350. doi: 10.1103/PhysRevB.73.235350

（49）Deskins, N. A.； Dupuis, M. Phys. Rev. B 2007， 75, 195212. doi: 10.1103/PhysRevB.75.195212

（50）Agrell, H. G.； Boschloo, G.； Hagfeldt, A. J. Phys. Chem. B 2004， 108, 12388. doi: 10.1021/jp037119p

（51）Mora-Sero, I.； Bisquert, J. Nano Lett. 2003， 3, 945. doi: 10.1021/nl0342390

（52）Barzykin, A. V.； Tachiya, M. J. Phys. Chem. B 2002， 106, 4356. doi: 10.1021/jp012957+

（53）Shkrob, I. A.； Sauer, M. C. J. Phys. Chem. B 2004， 108, 12497. doi: 10.1021/jp047736t

（54）Beermann, N.； Boschloo, G.； Hagfeldt, A. J. Photochem. Photobiol. A 2002， 152, 213. doi: 10.1016/S1010-603000236-8

（55）van de Lagemaat, J.； Frank, A. J. J. Phys. Chem. B 2001， 105, 11194. doi: 10.1021/jp0118468

（56）Komaguchi, K.； Nakano, H.； Araki, A.； Harima, Y. Chem. Phys. Lett. 2006， 428, 338. doi: 10.1016/j.cplett.2006.07.003

（57）Peiro, A. M.； Colombo, C.； Doyle, G.； Nelson, J.； Mills, A.；Durrant, J. R. J. Phys. Chem. B 2006， 110, 23255. doi: 10.1021/jp064591c

（58）Takahashi, H.； Watanabe, R.； Miyauchi, Y.； Mizutani, G. J. Chem. Phys. 2011， 134, 154704. doi: 10.1063/1.3578178

（59）Kerisit, S.； Deskins, N. A.； Rosso, K. M.； Dupuis, M. J. Phys. Chem. C 2008， 112, 7678. doi: 10.1021/jp8007865

（60）Shapovalov, V.； Stefanovich, E. V.； Truong, T. N. Surf. Sci. 2002， 498, L103.

（61）Yang, X. J.； Tamai, N. Phys. Chem. Chem. Phys. 2001， 3, 3393. doi: 10.1039/b101721g

（62）Tamaki, Y.； Furube, A.； Katoh, R.； Murai, M.； Hara, K.；Arakawa, H.； Tachiya, M. C. R. Chim. 2006， 9, 268. doi: 10.1016/j.crci.2005.05.018

（63）Thompson, T. L.； Yates, J. T., Jr. J. Phys. Chem. B 2005， 109, 18230. doi: 10.1021/jp0530451

（64）Berger, T.； Sterrer, M.； Diwald, O.； Knozinger, E.； Panayotov, D.； Thompson, T. L.； Yates, J. T., Jr. J. Phys. Chem. B 2005，109, 6061. doi: 10.1021/jp0404293

（65）Tang, H.； Levy, F.； Berger, H.； Schmid, P. E. Phys. Rev. B 1995， 52, 7771. doi: 10.1103/PhysRevB.52.7771

（66）Stevanovic, A.； Buettner, M.； Zhang, Z.； Yates, J. T., Jr. J. Am. Chem. Soc. 2012， 134, 324. doi: 10.1021/ja2072737

（67）Murakami, M.； Matsumoto, Y.； Nakajima, K.； Makino, T.；Segawa, Y.； Chikyow, T.； Ahmet, P.； Kawasaki, M.； Koinuma, H. Appl. Phys. Lett. 2001， 78, 2664. doi: 10.1063/1.1365412

（68）Shi, J.； Chen, J.； Feng, Z.； Chen, T.； Lian, Y.； Wang, X.； Li, C. J. Phys. Chem. C 2007， 111, 693. doi: 10.1021/jp065744z

（69）Knorr, F. J.； Mercado, C. C.； McHale, J. L. J. Phys. Chem. C 2008， 112, 12786. doi: 10.1021/jp8039934

（70）Yamada, Y.； Kanemitsu, Y. Appl. Phys. Lett. 2012， 101, 133907. doi: 10.1063/1.4754831

（71）Dozzi, M. V.； D′Andrea, C.； Ohtani, B.； Valentini, G.； Selli, E. J. Phys. Chem. C 2013， 117, 25586. doi: 10.1021/jp4095563

（72）Leytner, S.； Hupp, J. T. Chem. Phys. Lett. 2000， 330, 231. doi: 10.1016/S0009-261401112-X

（73）Murakami, M.； Matsumoto, Y.； Nakajima, K.； Makino, T.；Segawa, Y.； Chikyow, T.； Ahmet, P.； Kawasaki, M.； Koinuma, H. Appl. Phys. Lett. 2001， 78, 2664. doi: 10.1063/1.1365412

（74）Kim, S. H.； Stair, P. C.； Weitz, E. J. Chem. Phys. 1998， 108, 5080. doi: 10.1063/1.475914

（75）Kim, S. H.； Stair, P. C.； Weitz, E. Chem. Phys. Lett. 1999， 302, 511. doi: 10.1016/S0009-261400156-6

（76）Antoniewicz, P. R. Phys. Rev. B 1980， 21, 3811. doi: 10.1103/PhysRevB.21.3811

（77）Asahi, T.； Furube, A.； Masuhara, H. Chem. Phys. Lett. 1997，275, 234. doi: 10.1016/S0009-261400752-5

（78）Martino, D. M.； van Willigen, H.； Spitler, M. T. J. Phys. Chem. B 1997， 101, 8914.

（79）Colombo, D. P.； Bowman, R. M. J. Phys. Chem. 1996， 100, 18445. doi: 10.1021/jp9610628

（80）Furube, A.； Asahi, T.； Masuhara, H.； Yamashita, H.； Anpo, M. Res. Chem. Intermediat. 2001， 27, 177. doi: 10.1163/156856701745069

（81）Rabani, J.； Yamashita, K.； Ushida, K.； Stark, J.； Kira, A. J. Phys. Chem. B 1998， 102, 1689. doi: 10.1021/jp973411j

（82）Shkrob, I. A.； Sauer, M. C.； Gosztola, D. J. Phys. Chem. B 2004， 108, 12512. doi: 10.1021/jp0477351

（83）Wang, Y. H.； Hang, K.； Anderson, N. A.； Lian, T. Q. J. Phys. Chem. B 2003， 107, 9434.

（84）He, J. J.； Lindstrom, H.； Hagfeldt, A.； Lindquist, S. E. J. Phys. Chem. B 1999， 103, 8940. doi: 10.1021/jp991681r

（85）Borgstrom, M.； Blart, E.； Boschloo, G.； Mukhtar, E.； Hagfeldt, A.； Hammarstrom, L.； Odobel, F. J. Phys. Chem. B 2005， 109, 22928. doi: 10.1021/jp054034a

（86）Henderson, M. A.； Otero-Tapia, S.； Castro, M. E. Faraday Discuss. 1999， 114, 313. doi: 10.1039/a902070e

（87）Henderson, M. A.； Otero-Tapia, S.； Castro, M. E. Surf. Sci. 1998， 412/413, 252.

（88）Shen, M. M.； Henderson, M. A. J. Phys. Chem. C 2012， 116, 18788. doi: 10.1021/jp3046774

（89）Onda, K.； Li, B.； Zhao, J.； Petek, H. Surf. Sci. 2005， 593, 32. doi: 10.1016/j.susc.2005.06.044

（90）Li, B.； Zhao, J.； Onda, K.； Jordan, K. D.； Yang, J. L.； Petek, H. Science 2006， 311, 1436. doi: 10.1126/science.1122190

（91）Onda, K.； Li, B.； Zhao, J.； Jordan, K. D.； Yang, J.； Petek, H. Science 2005， 308, 1154. doi: 10.1126/science.1109366

（92）Zhang, Z. R.； Bondarchuk, O.； White, J. M. J. Am. Chem. Soc. 2006， 128, 4198. doi: 10.1021/ja058466a

（93）Bates, S. P.； Gillan, M. J.； Kresse, G. J. Phys. Chem. B 1998，102, 2017. doi: 10.1021/jp9804998

（94）de Armas, R. S.； Oviedo, J.； San Miguel, M. A.； Sanz, J. F. J. Phys. Chem. C 2007， 111, 10023.

（95）Zhao, J.； Yang, J. L.； Petek, H. Phys. Rev. B 2009， 80, 235416. doi: 10.1103/PhysRevB.80.235416

（96）Zhou, C.； Ren, Z. F.； Tan, S. J.； Ma, Z. B.； Mao, X. C.； Dai, D. X.； Fan, H. J.； Yang, X. M.； LaRue, J.； Cooper, R.； Wodtke, A. M.； Wang, Z.； Li, Z. Y.； Wang, B.； Yang, J. L.； Hou, J. G. Chem. Sci. 2010， 1, 575.

（97）Kawai, T.； Sakata, T. J. Chem. Soc. Chem. Commun. 1980， 694.

（98）Palmisano, G.； Augugliaro, V.； Pagliaro, M.； Palmisano, L. Chem. Commun. 2007， 3425.

（99）Muggli, D. S.； Odland M. J.； Schmidt, L. R. J. Catal. 2001， 203, 51. doi: 10.1006/jcat.2001.3311

（100）Chen, X.； Chen, S.； Guo, L.； Mao, S. S. Chem. Rev. 2010， 110, 6503. doi: 10.1021/cr1001645

（101）Li, Z.； Smith, R. S.； Kay, B. D.； Dohnálek, Z. J. Phys. Chem. C 2011， 115, 22534. doi: 10.1021/jp208228f

（102）Shen, M. M.； Acharya, D. P.； Dohnálek, Z.； Henderson, M. A. J. Phys. Chem. C 2012， 116, 25465. doi: 10.1021/jp309768b

（103）Zhou, C.； Ma, Z.； Ren, Z.； Wodtke, A. M.； Yang, X. Energy Environ. Sci.. 2012， 5, 6833. doi: 10.1039/c2ee21493h

（104）Cui, X. F.； Wang, Z.； Tan, S. J.； Wang, B.； Yang, J. L.； Hou, J. G. J. Phys. Chem. C 2009， 113, 13204. doi: 10.1021/jp901657u

（105）Klymko, P. W.； Kopelman, R. J. Phys. Chem-Us. 1983， 87, 4565. doi: 10.1021/j100246a006

（106）Kopelman, R. Science 1988， 241, 1620. doi: 10.1126/science.241.4873.1620

（107）Zhou, C.； Ma, Z.； Ren, Z.； Mao, X.； Dai, D.； Yang, X. Chem. Sci. 2011， 2, 1980. doi: 10.1039/c1sc00249j

（108）Haubrich, J.； Kaxiras E.； Friend, C. M. Chem. Eur. J. 2011， 17, 4496. doi: 10.1002/chem.201002588

（109）Zuo, F.； Wang, L.； Wu, T.； Zhang, Z. Y.； Borchardt, D.； Feng, P. Y. J. Am. Chem. Soc. 2010， 132, 11856. doi: 10.1021/ja103843d

（110）Shen, M. M.； Henderson, M. A. J. Phys. Chem. Lett. 2011， 2, 2707. doi: 10.1021/jz201242k

（111）Shen, M. M.； Henderson, M. A. J. Phys. Chem. C 2011， 115, 5886.

（112）Guo, Q.； Xu, C.； Ren, Z.； Yang, W.； Ma, Z.； Dai, D.； Fan, H.；Minton, T. K.； Yang, X. J. Am. Chem. Soc. 2012， 134, 13366. doi: 10.1021/ja304049x

（113）Lane, C. D.； Petrik, N. G.； Orlando, T. M.； Kimmel, G. A. J. Phys. Chem. C 2007， 111, 16319. doi: 10.1021/jp072479o

（114）Petrik, N. G.； Kimmel, G. A. J. Phys. Chem. C 2009， 113, 4451.

（115）Lang, X.； Wen, B.； Zhou, C.； Ren, Z.； Liu, L. M. J. Phys. Chem. C 2014， 118, 19859. doi: 10.1021/jp505674g

（116）Mao, X.； Wei, D.； Wang, Z.； Jin, X.； Hao, Q.； Ren, Z.； Dai, D.；Ma, Z.； Zhou, C. ； Yang, X. J. Phys. Chem. C 2015， 119, 1170.

（117）Mao, X.； Lang, X.； Wang, Z..； Hao, Q.； Wen, B.； Ren, Z.； Dai, D.； Zhou, C.； Liu, L.； Yang, X. J. Phys. Chem. Lett. 2013， 4, 3839. doi: 10.1021/jz402053p

（118）Guo, Q.； Xu, C.； Yang, W.； Ren, Z.； Ma, Z.； Dai, D.； Minton, T. K.； Yang, X. J. Phys. Chem. C 2013， 117, 5293.

（119）Phillips, K. R.； Jensen, S. C.； Baron, M.； Li, S. C.； Friend, C. M. J. Am. Chem. Soc. 2013， 135, 574. doi: 10.1021/ja3106797

（120）Yuan, Q.； Wu, Z.； Jin, Y.； Xu, L.； Xiong, F.； Ma, Y.； Huang, W. J. Am. Chem. Soc. 2013， 135, 5212. doi: 10.1021/ja400978r

（121）Domokos, L.； Katona, T.； Molnar, A. Catal. Lett. 1996， 40, 215. doi: 10.1007/BF00815285

（122）Liu, J. L.； Zhan, E. S.； Cai, W. J.； Li, J.； Shen, W. J. Catal. Lett. 2008， 120, 274. doi: 10.1007/s10562-007-9280-9

（123）Minyukova, T. P.； Simentsova, I. I.； Khasin, A. V.； Shtertser, N. V.； Baronskaya, N. A.； Khassin, A. A.； Yurieva, T. M. Appl. Catal. A-Gen. 2002， 237, 171. doi: 10.1016/S0926-860X00328-9

（124）Mao, X.； Wang, Z.； Lang, X.； Hao, Q.； Wen, B.； Dai, D.； Zhou, C.； Liu, L. M.； Yang, X. J. Phys. Chem. C 2015， 119, 6121.

（125）Ohno, T.； Sarukawa, K.； Matsumura, M. New J. Chem. 2002，26, 1167. doi: 10.1039/b202140d

（126）Takahashi, H.； Watanabe, R.； Miyauchi, Y.； Mizutani, G. J. Chem. Phys. 2011， 134, 154704. doi: 10.1063/1.3578178

（127）Tao, J. G.； Batzill, M. J. Phys. Chem. Lett. 2010， 1, 3200. doi: 10.1021/jz1013246

（128）Xu, C.； Yang, W.； Guo, Q.； Dai, D.； Chen, M.； Yang, X. J. Am. Chem. Soc. 2013， 135, 10206. doi: 10.1021/ja4030963

（129）Xu, C.； Yang, W.； Ren, Z.； Dai, D.； Guo, Q.； Minton, T. K.；Yang, X. J. Am. Chem. Soc. 2013, 135， 19039.

（130）Kavan, L.； Grätzel, M.； Gilbert, S. E.； Klemenz, C.； Scheel, H. J. J. Am. Chem. Soc. 1996， 118, 6716. doi: 10.1021/ja954172l

（131）Herman, G. S.； Dohnalek, Z.； Ruzycki, N.； Diebold, U. J. Phys. Chem. B 2003， 107, 2788. doi: 10.1021/jp0275544

（132）Xu, C.； Yang, W.； Guo, Q.； Dai, D.； Chen, M.； Yang, X. J. Am. Chem. Soc. 2014， 136, 602. doi: 10.1021/ja411020t

（133）Tilocca, A.； Selloni, A. J. Phys. Chem. B 2004， 108, 19314. doi: 10.1021/jp046440k

（134）Guo, Q.； Minton, T. K.； Yang, X. Chin. J. Catal. 2015， 36, 1649. ［郭 庆, Minton, T. K., 杨学明. 催化学报, 2015， 36, 1649.］ doi: 10.1016/S1872-206760935-4

Fundamental Processes in Surface Photocatalysis on TiO2

GUO Qing1ZHOU Chuan-Yao1MA Zhi-Bo1REN Ze-Feng2FAN Hong-Jun1YANG Xue-Ming1,*
（1State Key Laboratory of Molecular Reaction Dynamics, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, Liaoning Province, P. R. China；2International Center for Quantum Materials and School of Physics, Peking University, Beijing 100871, P. R. China）

Because of the potential applications of TiO2in photocatalytic hydrogen production and pollutant degradation, oνer the past few decades we haνe witnessed increasing interest in and effort toward deνeloping TiO2-based photocatalysts, and improνing the efficiency and exploring the reaction mechanisms at the atomic and molecular leνels. Because surface science studies on single crystal surfaces under ultrahigh νacuum （UHV） conditions can proνide fundamental insights into these important processes, both the thermo- and photo-chemistry on TiO2, especially on rutile TiO2（110） surfaces, haνe been extensiνely inνestigated with a νariety of experimental and theoretical approaches. In this reνiew, commencing with the properties of TiO2, we then focus on charge transport and trapping, and electron transfer dynamics. Next, we summarize recent progress made in the study of elementary photocatalytic chemistry of methanol on mainly rutile TiO2（110）, as well as in some studies on rutile TiO2（011） and anataseTiO2（101）. These studies haνe proνided fundamental insights into surface photocatalysis and stimulated new inνestigations in this exciting area. The implications of these studies for the deνelopment of new photocatalysis models are also discussed.

Titanium dioxide; Photocatalysis; Electron-hole separation; Nonadiabatic process; Ground-state potential energy surface

O643

10.3866/PKU.WHXB201512081

Received: November 6, 2015； Revised: December 8, 2015； Published on Web: December 8, 2015.

*Corresponding author. Email: xmyang@dicp.ac.cn； Tel: +86-411-84695174； Fax: +86-411-84675584.

The project was supported by the National Natural Science Foundation of China （21203189, 21321091, 21173212, 21403224, 21573225, 21322310）, National Key Basic Research Program of China （973） （2013CB834605）, Key Research Program of the Chinese Academy of Sciences （KGZD-EWT05）, and the State Key Laboratory of Molecular Reaction Dynamics, China （ZZ-2014-02）.

国家自然科学基金（21203189, 21321091, 21173212, 21403224, 21573225, 21322310）, 国家重点基础研究发展规划项目（973） （2013CB834605）,中国科学院重点部署项目（KGZD-EW-T05）以及分子反应动力学国家重点实验室开放课题（ZZ-2014-02）资助©Editorial office of Acta Physico-Chimica Sinica