Scholarly article on topic 'A review on the effects of TiO2 surface point defects on CO2 photoreduction with H2O'

A review on the effects of TiO2 surface point defects on CO2 photoreduction with H2O Academic research paper on "Nano-technology"

CC BY-NC-ND
0
0
Share paper
Academic journal
Journal of Materiomics
OECD Field of science
Keywords
{"CO2 photoreduction" / "TiO2 " / "Solar fuel" / Nanostructure / "Point defects" / "Oxygen Vacancy"}

Abstract of research paper on Nano-technology, author of scientific article — Huilei Zhao, Fuping Pan, Ying Li

Abstract Photocatalytic reduction of CO2 with water by photocatalysts such as TiO2 to produce solar fuels is an attractive approach to alleviate the environmental influences of greenhouse gases and in the meantime produce valuable carbon-neutral fuels. Among the materials properties that affect catalytic activity of CO2 photoreduction, the point defect on TiO2 is one of the most important but not frequently addressed and well understood in the literature. In this review, we have examined the major influences of TiO2 point defects on CO2 photoreduction with H2O, by changing the catalysts' gas adsorption capabilities, optical properties, and electronic structures. In addition, the performances of various defective TiO2 toward CO2 photoreduction are summarized and compared in terms of productivity, selectivity, and stability. We hope this review can contribute to understanding the mechanism of CO2 photoreduction on defective TiO2 and provide insights to the design of highly efficient defect-rich TiO2 to boost the CO2 utilization.

Academic research paper on topic "A review on the effects of TiO2 surface point defects on CO2 photoreduction with H2O"

Accepted Manuscript

A review on the effects of TiO2 surface point defects on CO2 photoreduction with H2O

Huilei Zhao, Fuping Pan, Ying Li

S2352-8478(16)30100-9 10.1016/j.jmat.2016.12.001

Reference: JMAT 85

To appear in: Journal of Materiomics

Received Date: 4 September 2016

Accepted Date: 5 December 2016

Please cite this article as: Zhao H, Pan F, Li Y, A review on the effects of TiO2 surface point defects on CO2 photoreduction with H2O, Journal of Materiomics (2017), doi: 10.1016/j.jmat.2016.12.001.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

A review on the effects of TiO2 surface point defects on CO2 photoreduction with H2O

Huilei Zhao, Fuping Pan, Ying Li* Department of Mechanical Engineering, Texas A&M University, College Station, TX 77845, USA

Corresponding Author Email: yingli@tamu.edu

Abstract

Photocatalytic reduction of CO2 with water by photocatalysts such as TiO2 to produce solar fuels is an attractive approach to alleviate the environmental influences of greenhouse gases and in the meantime produce valuable carbon-neutral fuels. Among the materials properties that affect catalytic activity of CO2 photoreduction, the point defect on TiO2 is one of the most important but not frequently addressed and well understood in the literature. In this review, we have examined the major influences of TiO2 point defects on CO2 photoreduction with H2O, by changing the catalysts' gas adsorption capabilities, optical properties, and electronic structures. In addition, the performances of various defective TiO2 toward CO2 photoreduction are summarized and compared in terms of productivity, selectivity, and stability. We hope this review can contribute to understanding the mechanism of CO2 photoreduction on defective TiO2 and provide insights to the design of highly efficient defect-rich TiO2 to boost the CO2 utilization.

Contents

Abstract..................................................................................................................................................1

1. Introduction........................................................................................................................................3

2. Point defects of TiO2............................................................................................................................4

2.1 Oxygen vacancies...........................................................................................................................4

2.2 Dopants induced impurity states....................................................................................................6

2.3 Characterization methods to identify defects.................................................................................8

3 Properties of defective TiO2................................................................................................................11

3.1 Gas adsorption ability of defective TiO2........................................................................................11

3.1.1 CO2.......................................................................................................................................11

3.1.2 H2O.......................................................................................................................................13

3.2. Band structure and optical property of defective TiO2.................................................................14

3.3 Charge transfer in defective TiO2..................................................................................................17

4. Catalytic performance of defective TiO2 in CO2 photoreduction with H2O..........................................20

4.1 Productivity.................................................................................................................................26

4.2 Selectivity....................................................................................................................................27

4.3 Stability.......................................................................................................................................28

Acknowledgement................................................................................................................................30

References............................................................................................................................................30

1. Introduction

Photocatalytic reduction of CO2 has a rather promising future in alleviating the current fossil fuel shortage and urgent environmental problems because it can harvest and store solar energy, mitigate CO2 emissions, and produce valuable fuels.1 Due to the thermodynamically stable nature of the CO2 molecule and energetically unfavorable nature of photoreduction processes, advanced photocatalysts are required to facilitate the CO2 photoreduction and boost the solar energy conversion efficiency. Among the variety of photocatalysts reported in the literature, TiO2 is widely used for photodegradation of organic compounds, water splitting for hydrogen production, and conversing CO2 to fuels such as CO, CH4, CH3OH, etc.2'4 Compared to other photocatalysts, TiO2 has the advantages of low cost, low toxicity, high chemical and thermal stability, and abundant availability.5'6 However, realizing TiO2 assisted CO2 photoreduction with commercialized devices faces a great difficulty, owing to the complex chemical reaction pathways and unrevealed underlying mechanism.7 Besides, TiO2 itself suffers several limitations for CO2 photoreduction. First, TiO2 has a large band gap, leading to the low solar energy usage efficiency. Second, the fast charge recombination rate in TiO2 causes the poor photocatalytic performance8. Third, relatively insufficient CO2 adsorption capability results in the low production rate of solar fuels.

Recent studies show that engineering defects in TiO2 could be one possible solution to address the above three limitations.910 There are, in general, four categories of defects in crystals based on the defect's dimension: point, line, planar, and volume defects. Yang and his coworkers pointed out the majority of the defect in nanoscale photocatalysts are point and volume defects.8 Because most of the published studies about defective TiO2 as a photocatalyst focus on point defects, this review only discusses the effects of point defects of TiO2 on CO2 photoreduction.

It is widely demonstrated that defect-rich TiO2 has a superior activity than defect-free TiO2.11 The improved photocatalytic performance of TiO2 may be attributed to the following three aspects. First, tailoring the surface defects can change surface properties of TiO2, resulting in different adsorption and desorption capabilities of reactants, intermediates, and products. Second, light adsorption ability is closely related to the band gap structure, which is affected by the type, position and concentration of defects. With a reduced band gap or addition of sub-bands, TiO2 can respond to visible light, increasing light adsorption efficiency. Third, disorders at the surface or subsurface could become traps for electrons or holes, suppressing the recombination of electrons and holes and thus facilitating their separation. For example, oxygen vacancies, which are crucial to the activity of TiO212, play different roles on CO2 photoreduction. Both experimental and computational studies have demonstrated that oxygen vacancies can facilitate CO2 adsorption and activation.1314 Moreover, defects on the surface can also act like charge

traps to separate the charges. For example, trapped charges like holes in the surface defects are ready to react with the electron donors, promoting the photocatalytic reaction. However, defects in the bulk, which could also be charge traps, usually become recombination centers, because those trapped charges are not available for the photoreaction process.15 Also, some X-ray photoelectron spectroscopy (XPS) studies showed the shift in the band edge of TiO2 owing to oxygen vacancies which result in the reduction of the band gap, extending the light absorption edge of TiO2.16

Although the importance of surface defects has already been widely recognized, systematic analyses of their effects on the photocatalytic performance of TiO2, especially for CO2 photocatalytic reduction, are not available in the literature. This review gives an overview of the surface point defects on TiO2 that consists of five sections. In section 1 we have introduced the background and rationale of this review. Section 2 describes the categories of point defects in TiO2 and characterization methods to visualize or quantify the defect sites. Section 3 explains how defects on TiO2 could affect CO2 photoreduction by changing gas adsorption capabilities, optical properties, and charge generation and transfer. Section 4 presents the recent progresses in the application and performance of defective TiO2 for CO2 photoreduction. The last section summarizes the challenges in CO2 photoreduction and future opportunities in this field.

2. Point defects of TiO2

Point defects in TiO2 include oxygen vacancies, Ti interstitials, Ti vacancies, impurities, and defects at interfaces. The focus of this review is on the oxygen vacancies and impurities, which are the most common point defects reported in the literature. Removing one of the neutral oxygen atoms from the lattice forms an oxygen vacancy, resulting in the excess electrons filling into the empty states of the Ti ions, and forming Ti3+ species. Oxygen vacancies, especially on rutile (110) and anatase (101) are extensively studied because this type of defects can be generated or engineered at a mild condition. Ti interstitials are formed when Ti atoms migrate from the surface to the lattice interstitial sites at a high temperature, or after a long time thermal treatment.17

2.1 Oxygen vacancies

Oxygen vacancies (VO) are most frequently discussed in the literature with regard to the defects

of TiO2, because VO is relatively easier to generate, compared to Ti interstitial. As mentioned in the

previous section, VO is formed by ejecting an oxygen from the lattice of TiO2. Different methods were

reported to prepare oxygen deficient TiO2. One popular approach is through post thermal treatment in an

inert or reductive gas. Hydrogen thermal treatment is a usual method to produce vacancy-rich TiO2, with

possible formation of H interstitial. Black TiO2 generated by thermal annealing in H2 atmosphere has been brought to attention by Chen et al. in 2011.18 In that work, a disordered surface with about 1 nanometer thickness was formed at 200 °C in H2 at 20 bar. Liu et al. also successfully synthesized defective TiO2 with 0, 1, and 5% Cu loading using in situ thermal treatment in He or H2 at 1 atm. 19 As shown in the Fig. 1, the color of the material varied with the Cu concentration and the treatment gas environment. The color change was due to the formation of VO/Ti3+ defects and the reduction of Cu2+ to Cu+ or Cu0. The higher concentration of Cu, the darker the color of the sample. H2 pretreated Cu-TiO2 had darker color compared to those treated in He, because thermal pretreatment in reduction gas H2 can form a higher concentration of VO and possible addition of H interstitials. During the thermal treatment in inert or reductive gases, the formation of defects is closely related to the oxygen partial pressure.20 Another method to produce VO is to use NaBH4 as the reductive agent and grind TiO2 with NaBH4 in a mortar, the mixture of which is then thermally treated in Ar.21

Fig. 1 Appearance of the un-pretreated TiO2 and Cu-TiO2 samples (denoted as UP) and those thermally treated in He or H2. (Reprinted with permission from ref 19. Copyright 2013 Elsevier Science)

It is reported that UV irradiation could induce surface defects on the hydrophilic surface of TiO2 since the defects are believed to be the active sites for dissociative adsorption of H2O.22-28 However, Mezhenny et al.29 concluded that UV irradiation cannot produce additional oxygen vacancies after investigation by scanning tunneling microscope (STM) imaging. Defect sites were statistically counted after UV irradiation on the pre-cleaned TiO2 (110) surface, indicating the UV did not help to generate additional surface defects on TiO2 within the statistical error. Moreover, the formation energy of an oxygen vacancy defect was calculated to be ~7 eV, greater than the UV irradiation energy.30 The theoretical calculation precluded the possibility of UV irradiation-induced oxygen vacancies, which

agreed with a study carried out by Zubkov et al.31 However, defect formation on the TiO2 surface is made easier by introducing reducing gas molecules such as CH3OH8 under UV irradiation.

TiO2 has three kinds of polycrystalline phases: anatase, rutile, and brookite.32 Different crystal phases have different symmetries, close stacking planes, slip directions, theoretical crystal densities and available interstitial positions per unit cell. Therefore, defect distribution and density are altered in different crystal structures. Because the brookite phase has a more complex structure compared to anatase and rutile, brookite is more likely to have defects.33 This is consistent with the order of the formation energy of VO on TiO2 surface: 5.52 eV (brookite) < 5.58 eV (anatase) < 5.82 eV (rutile).34 Liu et al.13 conducted CO2 photoreduction on both pristine and defective TiO2 (anatase, brookite, and rutile) prepared by thermal annealing in He. The defective brookite exhibited the best activity among three defective polycrystals, which might be attributed to the higher density of VO in defective brookite crystal phase.

The formation of surface VO is also related to the exposed plane of certain TiO2 crystals. The most stable surface (101) of anatase is found to have lower tendency to have VO than (110) surface of rutile, because of the lower stability of adjacent 4-fold Ti3+ on anatase (101) than 5-fold Ti3+ on rutile (110).35 Morgan and Waston discovered that the oxygen vacancy formation energy on each facet of rutile was in the order (100) > (110) > (001) > (101) using DFT calculations, i.e. theoretically, surface (101) in rutile is more energetically favorable to have Vo than other rutile facets.36 Therefore, engineering TiO2 with a specific exposed facet or mixed facets is important to control defect formation on the surface. However, the defect formation on specific exposed facets and crystal phase is rarely investigated, particularly for the rutile phase due to the difficulty in synthesizing pure brookite. More researches are suggested to be launched in this direction.

2.2 Dopants induced impurity states

As mentioned in the introduction, the application of TiO2 is hindered by the insufficient utilization of sunlight due to the large band gap of TiO2 (~3.2 eV for anatase and 3.0 eV for rutile) and the fast recombination rate of photogenerated charge carriers. Doping is one of the frequently used methods to narrow down or introducing impurity levels inside the band gap of TiO2, and possibly improve the separation of charge carriers. Doping is a process of adding foreign atoms into the crystal lattice.37 Various metal38-42 or nonmetal43-46 elements are used as dopants. As for TiO2, dopants could substitute the original Ti or O atom, or be placed at the interstitial position depending on the valence, radius, and electronic properties of the dopants. For example, for carbon-doped TiO2, carbon atoms are likely to locate at interstitial or substitutional46 positions in TiO2 crystals, while iodine doping usually causes

substitution of Ti by iodine.47 Impurities introduced into the lattice by doping could lead to different substitutional48 and interstitial49,50 point defects. In addition, oxygen vacancies and Ti3+ may form to maintain the charge balance. Because of doping, the original stacking pattern is disturbed. The change of bond length as well as the bond angles, results in the alteration of the interatomic interaction. Matthisa et al. claimed that doping of N could lower the formation energy of oxygen vacancy.51 Dopant sites are different depending on the choice of the dopant.52 As shown in Fig. 3, (TiO2)6 has different relaxation structures depending on the dopant type: C, N, or S. Though S are O are both in group VI, the larger radius of S makes S ready to substitute a 2-fold oxygen site. N as a group V element tends to substitute higher or equal to 3-fold oxygen sites. While C does not substitute 4-fold Ti sites, it replaces oxygen at 4fold sites.

Fig. 3 (Ti02)6 nanoparticles with lowest energy anion substitution structures, (a): C0 dopant, (b): N0 dopant, and (c): S0 dopant. Purple dots represent Ti, red O, gray C, light blue N, and yellow S. C0 , N0 , and S0 are Kroger-Vink notations, where, C, N, or S sits at the O lattice site, and x represents the null charge. (Reprinted with permission from ref52. Copyright 2010 ACS)

Table. 1 Band gap energies and band structure changes of doped TiO2 materials

Metal Band gap Band structure Nonmetal Band gap Band structure

dopant (eV) change dopant (eV) change

Fe V 3.0-3.253,54 2.857 2.5-2.759 CB: -0.19 eV VB: 2.81 eV 55 C S N 3.254, 2.2256 3.254, 2.62.958 3.054, 2.93.060 Additional electronic states introduced by C, S, or N above the VBM

Nd 2.67-3.1261 Impurity level near the CBM B 2.9462 Continuous states:

Cu 3.0-3.1538 - I 3.0-3.0547 5p and/or 5s orbitals of I5+ and O 2p orbitals of the VB

Co 2.19-3.0863 Higher VBM and CBM

In 3.19-3.2464 -

Ag 2.74-2.8865 -

The introduction of dopants to the original crystal lattice of TiO2 can introduce sub-band or continuous bands near/adjacent the original bands, which makes the modified TiO2 responsible to visible light. The band gap energy varies with different dopants (Table. 1). Meanwhile, dopants also play the role of mediators of interfacial charge transfer, or recombination centers, which is related to the concentration of the dopants.66 The effect of dopants on the charge carriers' transfer is discussed in Section 3.2.

2.3 Characterization methods to identify defects

Appropriate characterization methods are required to investigate defect properties regarding its type, location, density and so on. Unlike most other nanomaterial charaterization methods, TiO2 defect identification is heavily depending on atomic level microscopic technologies, because of the unique requirements of atomic scale modifications. To directly observe the crystal changes, scanning tunneling microscopy (STM)67-72 and high-resolution transmission electron microscopy (HR-TEM) are commonly used. STM uses a tunneling current through a gap between a crystal surface and a fine probe tip. Different from other scanning probe methods, for example, atomic force microscopy (AFM), STM gives electronic band information through current signals across the surface, which can be further calculated to find defect aspects.29'73'74 STM obtains images with a typical resolution of 0.1 nm lateral in length and 0.01 nm in depth. STM can be used not only to show the location of defects but also to monitor the reactions

occurring at or near the defect sites. As shown in Fig. 4, the O vacancy in Ti02 (110) face was observed, and STM recorded the corresponding diffusion path. With exposure to 02, the oxygen vacancy diffusion was recorded as a movie, and Fig. 4B and 4C are two consecutive STM images extracted from the movie. Fig. 4D is the difference STM image by subtracting Fig. 4C from Fig. 4B. Based on the STM images, a simple imaginable diffusion mechanism can be proposed, and that is, a neighboring O atom moving into the vacancy, leading to oxygen vacancy diffusion. By comparing through frames of the vacancy diffusion, the frequency and density of O vacancies are studied, which proves the defect behavior in C02 reduction reaction. Similar to STM, HR-TEM can also demonstrate the distortion of the lattice. HR-TEM directly observes atom position in a crystal and shows crystal defects.75 Also, by calculating the crystal constants, physical properties of as-prepared Ti02 crystals can be found.

Fig. 4 (A) Ball model of the Ti02 (110) surface. A bridging O vacancy is marked by a circle. The arrow denotes the observed vacancy diffusion pathway. (B and C) Two consecutive STM images (D8.5 s/frame). (D) Difference image, in which (C) is subtracted from (B). Bright protrusions indicate the presence of vacancies in (B), whereas dark depressions indicate the new vacancy positions in (C). (Reprinted with permission from ref74. Copyright 2003 Science)

Photon-electron interaction based techniques study Ti02 defects through collecting chemical bonding and electronic band information, for example, Fourier transform infrared spectroscopy (FTIR)76, X-ray photoelectron spectroscopy (XPS or ESCA)77'78, photoluminescence (PL)79'8", X-ray diffraction (XRD) and Raman81. Among those methods, FTIR is a low-cost and fast way to study chemical behaviors of defects on the surface. FTIR is usually used to investigate the surface properties of Ti02, and can also

be used to identify the defect such as Ti3+ caused by Vo, because the products from the reaction of Ti3+ with other gases (H2O, CO2, CO etc.) are different from those of Ti4+. An example is shown in Fig. 5 76 that after thermal pretreatment in He, FTIR detected the Ti3+ bonded OH group on the surface of TiO2, while before pretreatment, only Ti4+ bonded OH group was detected. XPS as a surface chemistry analytical technique is used to investigate the surface species having different valance electrons. Extensive XPS assisted studies have been launched on detecting oxygen vacancies, Ti3+ identification and quantifications.77'78 PL can be used to detect the photon emitting due to electron and hole recombination after photo-irradiation. Strong PL at near infrared (~840 nm) of rutile after UV irradiation might be assigned to the recombination of electrons to surface hydroxyl group or oxygen vacancies.8283 For example, Shi and coworkers pointed out the PL observed at the visible range is due the recombination assisted by oxygen vacancies.84 In addition, Raman and XRD are commonly used as indirect methods to prove the existence of defects. The changes before and after the formation of the defects can be observed by the changing of photon renounce modes in Raman or peaks in XRD patterns.

Fig. 5 Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) spectra of OH groups in the region of 3000-3800 cm-1 for (a) unpretreated TiO2 and (b) TiO2 thermally pretreated in helium (anatase: A, rutile: R and brookite: B). (Reprinted with permission from ref.13 Copyright 2004 ACS Publications)

Many other methods are useful for the identification of defects. Positron annihilation lifetime spectra (PALS) is a useful approach to obtain defects information related to the size, density, and location by investigating the interaction of positrons and electrons. Positrons injected into solids annihilate fast by reacting with free electrons unless there are defects near the injection site, because the positrons can be

attracted to the defect, where the electron density is relatively low. 85-87 In PALS, the lifetime t (ti, t2...) gives the information about the size of the defects (usually a larger defect size on the surface and a smaller in the bulk). Meanwhile, the relative intensity I (Ii, I2...) of the current provides the information of the defect density. I1/I2, for example, could give the information of the ratio of the bulk defect density versus the surface defect density.85'86'88-92 The short lifetime t1 is attributed to the free annihilation of the positron in defect free TiO2, while some small voids or shallow positron trap (such as oxygen vacancies in ZnO) could elongate t1.85-87 The lifetime t2 is much longer than t1, due to the larger oxygen vacancy clusters (dimer, trimer or larger).88 The longer lifetime t2 is resulted from the lower annihilation rate due to the relatively lower density of electron at these defects.8890 Electron paramagnetic resonance (EPR)9394 is another method to identify the oxygen vacancy, due to the paramagnetic properties caused by unpaired electrons of oxygen vacancy.9596 Some published works also use molecular beam sputtering with thermal desorption spectroscopy (TDS) to study TiO2 defects.

Without instrumentational assistance, visual observation alone may tell the tendency of defect formation, which is helpful in experimental studies. For example, the color change was regularly related

13 19 32 97

to the lattice imperfections (e.g. oxygen vacancies and impurities). , , , Straumanis et al. reported the color change of rutile from yellowish white to bluish black corresponding to the O/Ti ratio from 2.000 to 1.98398, and the color changed to darker indicated the increase in defect concentration. In addition, theoretical calculations are conducted to facilitate the understanding of the properties of defective TiO2, which is discussed in Section 3.

3 Properties of defective TiO2

3.1 Gas adsorption ability of defective TiO2 3.1.1 CO2

The CO2 molecule is relatively stable and chemically inert due to its electronic and geometric properties.99 The C=O bond has a higher dissociation energy of 750 kJ mol-1 compared to C-H bond (430 kJ mol-1) and C-C bond (~336 kJ mol-1).100 Activation of CO2 before reduction by accepting photogenerated electrons is very important. CO2 adsorption on TiO2 is considered as an activating process. CO2 adsorbed on TiO2 can have different geometries. Linear CO2 geometry was reported by Henderson et al. 101 on pristine TiO2. However, the CO2 adsorption capability on the defect-free TiO2 surface is relatively low, and STM researches showed CO2 was bent when adsorbed on the defect sites such as Ti3+/Vo.102

Oxygen vacancies offer chemical adsorption sites for CO2.68,103 Vo favors the dissociation of H2O into OH groups, which is helpful for the adsorption of CO2 on the surface of TiO2.104 Theoretical calculations also demonstrated that point defects favor bent conformations of CO2 and help stabilizing surface CO2-, which is believed to be the intermediate for CO formation.105 Acharya and coworkers used STM to demonstrate the weak adsorption of CO2 on VO-free surface but strong bonding between CO2 and the bridging Vo.106 The group of Pipornpong identified the different configurations of the perfect or oxygen deficient anatase surface (001) absorbing CO2 molecules. The adsorbed CO2 at VO could oxidize VO to perfect site again, and meanwhile, CO2 was reduced to CO.107

Yin et al.108 reported that oxygen vacancies could provide excess electrons and adsorption sites for CO2, and presented the adsorption configurations of CO2 at VO sites on rutile (110) surface (detailed description of each configuration is shown in Fig. 6). The binding energy calculated for each configuration showed the VO-4 had the highest binding energy indicating VO-4 was the most stable configuration. VO-1, the linear geometry, which is also the most reported for defect-free TiO2 had slightly lower binding energy compared to VO-4. The meta-stable configurations were VO-2 and VO-3, the binding energy values of which were lower than VO-1 and VO-4.

• V0.2 one O of C02 adsorbed at VQ site and the other O

adsorbed at Ti5f in the plane.

• v0.4

one O atom of C02 adsorbed at V0 site and the C atom adsorbed with the 02f near the V0 site.

Fig. 6. Different adsorption configurations of CO2 at VO sites on rutile (110) surface. (Adapted with permission from ref.108. Copyright 2016 Nature)

Similarly, Andino et al.14 performed a computational study on the interaction of CO2 with perfect and oxygen deficient brookite (210) surface. They conducted simulations for four different configurations

quite similar to Vq_2 except for the C ator linked to 03f in the plane

of CO2 on brookite surfaces, and the adsorption energies of CO2 adsorbed on the perfect and oxygen deficient brookite are compared in Table 2. Similar approaches were performed on anatase surfaces reported by He et al.105 The calculated results demonstrated that an oxygen deficient surface is thermodynamically favorable to adsorb CO2 than a defect-free surface. Civis and coworkers synthesized isotopically pure Ti18O2 by hydrolysis with heavy oxygen water.109 They depicted the process of C16O2 adsorption on vacuum annealed Ti18O2 as follows: C16O2 + VO-Ti-18O^Ti-18OC16O16O.

Table 2 Calculated adsorption energy (in eV) of CO2 adsorbed on the perfect and oxygen deficient brookite (Adapted with permission from ref.14 Copyright 2012 ACS Publications)

Model Perfect brookite Oxygen-deficient brookite

A -0.44 -1.53

B -0.50

C -0.30 -0.51

D -0.72 A^ ^/-1.59

A: Linear adsorption configuration

B: Bent adsorption configuration with one side on a Ti(5-fold)-O(2-fold) bond on the surface

C: Bridging bidentate binding configuration with two O atoms bridging two 5-fold Ti atoms and the C atom pointing upward without forming any bond with the surface atoms.

D: Two O atoms bridging two 5-fold Ti atoms and C atom pointing downward and forming a C---O bond with the 3fold O atom on the surface

3.1.2 H2O

The adsorption of H2O during the photoreduction of CO2 with H2O vapor is also very important, because H2O reacts with holes and provides H+. As early as 1989, Richard et al.110 investigated the interactions between H2O with defective (annealing in an ultra-high vacuum) and near-perfect (annealing in O2) TiO2. The defects helped water dissociation on the surface of defective TiO2, while dissociative adsorption of water helped the defect formation in near-perfect TiO2. Philip et al. launched DFT calculations on the reaction of water with the (110) plane of rutile, and demonstrated that defect sites benefited the absorption and dissociation of water on the surface of TiO2.111 When water dissociates at the oxygen vacancy site, the H atom is removed and the O atom from the remaining OH group heals the original VO site.104112 Under UV irradiation and in the presence of CO2 and H2O vapor, more bridged CO2-, HCO3-, and HCOOH species were formed and more dissociative H2O adsorption on defective TiO2 were detected than on defect-free TiO2, according to an in situ DRIFTS study.13 CO2-, HCO3-, and HCOOH were suggested to be the intermediates for CO and CH4, the final reaction production. Fig. 7 gives an example of different H2O adsorption behaviors on oxygen vacancies and on perfect TiO2 sites. H2O adsorbs on the Ti5c sites (1), and diffuses along a Ti row (2). One dominant theory is that the adsorbed H2O is in its molecular form, but another theory is that it may dissociate to a bridging OH group

(OHb) and a terminal OH group (OHt) (6). In Fig. 7, the process of H2O adsorption on oxygen vacancies is as follows. An H2O molecule approaches (3), adsorbs on (4) and dissociates to OHb on a VO (5).

(¿>H20

! OH from H20 (¡) OHb = + H

(H20)2 dimer

#o • ®oV W w W g

Fig. 7 Schematic of the reactions of H2O on reduced TiO2 (110). (Reprinted with permission from ref Copyright 2015 ACS Publications)

3.2. Band structure and optical property of defective TiO2

One of the most important benefits of creating defects in TiO2 crystals is the manipulation of exciton flow during the photoreduction process. In most literature, the improved photoactivity is explained by more active sites at the crystal surface, the presence of oxygen vacancies and Ti3+ species, and functionalization by active groups. It can also be interpreted as an improved photon-electron excitation process and a controlled charge transfer process.

Light absorption of general TiO2 crystals is limited in ultraviolet bands due to its band gap of around 3.2 eV.114 The defective TiO2 usually has a color rather than just white13, which comes from the extended light adsorption ability in the visible light region. Defective TiO2 tends to have a relatively narrower band gap than defect-free ones, as shown in Fig. 8c and Fig. 9, with a shift of electronic bands or a smaller band gap.

9 6 3 E-E (eV)

E-Ep(eV)

3.0 eV~

Blank sample Hydrogenated Ti02

Fig. 8 (a) Synchrotron VB spectra of TiO2 with blank sample and hydrogenated; (b) The total DOS and partial DOS of hydrogenated TiO2 with H staying at VO which corresponds to the minimum energy state; (c) Schematic illustration of the DOS of hydrogenated TiO2, as compared with the blank sample. (Reprinted with permission from ref.10 Copyright 2014 AIP)

Many studies show oxygen vacancies are the invisible agent on the surface of TiO2, which extends the light absorption edge of TiO2 to the visible light region,115 because the oxygen vacancies rise the local states under the edge of the conduction band. Oxygen-vacant states with one or two electrons are generated due to the missing of an oxygen atom in the bulk or on the surface.9 Atomic rearrangements and surface reconstruction could also be triggered by the formation of oxygen vacancies. Excess electrons are found in the oxygen vacancy states. Some works combined with theoretical and experimental studies give proofs of this. An earlier work10 showed hydrogenation treatment reduced the band gap of TiO2, because of the rise of the defect band near the top of the valance band and the unchanged conduction band, as shown in Fig. 8. This band gap change was attributed to not only the formation of oxygen vacancies but also the introduction of H interstitials. The increase of the H2 pressure during the hydrogenation process gave rise to the concentration of defects in hydrogenated TiO2, and the band gap structure could be tailored by changing the concentration of H2. However, another model of the hydrogenated TiO2 band gap is the formation of a defect band near the bottom of the conduction band. Naldoni and his coworkers116 found a slightly negative band shifting of CB, tailoring of VB and the localization of oxygen vacancy states by comparing the density of states (DOS) of Degussa P25 and black TiO2, which are shown in Fig. 9. The nanoparticle structures in Fig. 9 were built based on the data of Synchrotron X-ray powder diffraction (SXRPD), Cathodoluminescence spectroscopy (CL), UV-vis spectroscopy, and XPS analysis. The proposed model of VO localized states in black TiO2 are at 0.7-1.0 eV below the conduction band minimum (CBM). Electrons can migrate not only from both VB and VO localized states to CB, but

also from the VB to the VO localized states. All the band structure changes caused the visible and near-infrared (vis-NIR) absorption of black Ti02. In addition, a theoretical calculation of nitrogen-doped Ti02 using spin-polarized DFT proved the absorption of light with wavelength from 400 to 500 nm due to the dopant N, and the oxygen vacancies dominated the adsorption above 500 nm.16

(a) (b)

reference (P25)TiO; black TiO,

Fig. 9 Schematic of nanoparticle's structure and DOS for (a) Ti02 P25 Degussa and (b) black Ti02. (Reprinted with permission from ref 116. Copyright 2012 ACS)

Besides oxygen vacancies, impurity levels induced by doping can also change the band structure of Ti02. Using the optimized stoichiometric model of Ti02, the electronic band structures and the densities of states (DOS) of defective Ti02 can be calculated.95'117 Usually band gap calculated by DFT is relatively smaller than that calculated by experiments, e.g. UV-vis spectra because of the discontinuity of exchange- correlation energy.118 However, the estimated band structure by DFT is still helpful to better understand the defect related properties. Asahi et al. calculated densities of states (DOSs) of anatase Ti02 with doping of C, N, F, P, or S anions, by the full-potential linearized augmented plane wave (FLAPW) formalism using the local density approximation (LDA). As shown in Fig.10, different anions doped into the lattice of TiO2 by substituting O, leading to the various band structure change of TiO2. The band gap was narrowed by N dopants the most due to the p states of N dopants mixing with O 2p states. Both C and P dopants formed impurity states deep in the band gap of TiO2. Based on the calculated results, it is helpful to design modified TiO2 materials with suitable band structure to benefit specific redox reactions.

B Ti02

kl ! F-doped ■i i

l/| N-dopcd

ft ______U C-doped

! S-doped

_, ¡/v. A P-doped

i Ni-doped

Jl Ni+s-doped

Fig. 10 (A) Total DOSs of doped TiO2 and (B) the projected DOSs into the doped anion sites, calculated by FLAPW. N doping at an interstitial site (Ni-doped) and at both substitutional and interstitial sites (Ni+s-doped) (Reprinted with permission from ref.119 Copyright 2001 Science)

3.3 Charge transfer in defective TiO2

Electron-hole pairs generated by photo-illumination usually have a short lifetime before they recombine. Controlling and manipulating the behaviors of charges and mitigating charge recombination are essential to reach and maintain a high performance of prepared catalysts. Recombination could occur either in the bulk or on the surface. Experimental and theoretical studies have shown that oxygen vacancies can affect the recombination of electrons and holes, by trapping charge carriers in the defect sites, which could hinder electrons and holes meeting with each other and thus increase the regional reaction rate.120 In terms of the location of the trapped charges, some believe VO is the electron trap, and

3+ 121 122

Ti is the hole trap, but the others believe the opposite . However, some researchers also demonstrated that the defect sites could act as the recombination center of the charge carriers. Kong et al.12 used PALS to characterize defective TiO2 and found that the surface defects helped charge separation, while the defects in the bulk acted as the recombination center.

-¿^TilLc—Tpt-O- ¿It — 0 -Ti^- □—Ti34— O—

¡0.002 1321

2GB012TO 1240, 1320 1200

rayinwitffirftim"

— O ^Ti4— C—{j4 — O— _ (3)

HCu^ y^? i&Q-Bid

Fig. 11. A possible mechanism for spontaneous dissociation of CO2 on defective Cu(I)/TiO2-x catalyst in the dark at room temperature, and DRIFTS analysis (inserted). (Reprinted with permission from ref. 123 Copyright 2012 ACS Publications)

The electrons trapped on the surface VO can help the CO2 adsorption, activation, and dissociation. Liu. et al.123 demonstrated the spontaneous CO2 dissociation to CO on the surface of defective Cu(I)/TiO2-x using in situ DRIFTS analysis, and the possible mechanism is shown in Fig. 11. CO2- species was first generated upon an electron attachment to CO2 at the oxygen vacancy site, and thereafter spontaneously dissociated into CO on the surface of Cu(I)/TiO2-x even in the dark. Furthermore, the defect sites helped the dissociative chemisorption of H2O, forming OH groups, which helped the chemisorption of CO2 and the formation of bicarbonates and CO2- species.

Fig. 12 illustrates possible charge transfer pathways in defective TiO2 summarized from the most acceptable theories. Photo-excited electrons and holes can migrate to the surface of TiO2, participating in redox reactions. Oxygen vacancies in the bulk play the role as charge recombination centers, while surface oxygen vacancies serve as electron traps, which can facilitate the reduction reaction.

^ V0 on the surface # VQ in the bulk

Q Photoexcitation of e and h+ Q Reduction by e Q Oxidation by h+

Q Recombination at V0 in the bulk Surface recombination Qj e trapped at surface V0

Fig. 12 Charge transfer pathways in oxygen-deficient Ti02 under photo-illumination

The energy difference between the conduction band of the photocatalyst and the reduction potential of the electron acceptor is the driving force for the transport of electrons, which affects the migration of charges.8 Narrowing band gap may result in the decline of the redox activity due to the decrease of the driving force caused by the positive shifting of the VB edge or the negative shifting of the CB edge, which might be a drawback of defective modification.124

The third commonly studied effect is impurity scattering, which is caused by the dopant ions in the lattice of Ti02. The dopants ions have fixed charges, and they can change the movement direction of charge carriers by Coulombic force. An electron flow can be scattered by a donor or acceptor ion, which is also true to holes. Impurity scattering may lower the relaxation time and the mobility.125 The impurity

scattering mobility of charge carriers is proportional to n , where T is the temperature, Na and Nd are

the numbers of the acceptor ions and donor ions, respectively.126 The scattered charges may increase the rate of recombination and decrease overall performance, which can be used to explain the poor performance of Ti02 with relatively high doping level. Furthermore, one oxygen vacancy provides two excess electrons, and one Ti interstitial provides four excess electrons.127 The increasing electron density caused by the defects may improve the C02 photoreduction process.

4. Catalytic performance of defective TiO2 in CO2 photoreduction with H2O

VB h h

C02, H+

CO, CH4, CH3OH, HCHO, HCOOH, etc.

02/ H+

Fig. 13 Schematic illustration of the photocatalytic process in the CO2 photoreduction with H2O

CO2 photoreduction with H2O can be understood in two steps. First, CO2 and H2O molecules are adsorbed onto catalyst surface, and this process requires high surface area and low physical and chemical energy barriers. Second, the photocatalytic process induces excitons from photons and initiates the photoreduction process. The second step in detail can be described in the following sub-steps (shown in Fig. 13): (1) photo-illumination on the surface of photocatalyst induces the generation of charge carriers, i.e. electron-hole (e- -h+) pairs; (2) the excited electrons in the conduction band (CB) of photocatalyst could migrate to the surface and reduce CO2 to solar fuels (e.g., CO, CH4, CH3OH, HCOOH), and (3) the holes left in the valence band (VB) of the photocatalyst could oxidize H2O into oxygen.128 Possible reactions involved in the photocatalytic reduction of CO2 with H2O on TiO2 and thermodynamic redox potentials of various compounds are listed in Table 3.

Table 3. Possible reactions involved in photocatalytic reduction of C02 with H20 on Ti02 and associated redox potentials

Reaction Equation Redox pairs Redox potential'1"9"131 (V vs. NHE @ pH=7)

2H20 + 4h+ ->4 H+ + 02 H20/02 +0.82

C02 + Qe~ + 8H+ CHA + H20 C02/CH4 -0.24

C02 + 6e~ + 6H+ CH3OH + H20 C02/CH30H -0.38

2H+ + 4e~ H2 H+/H2 -0.41

C02 + 4e~ + 4H+ HCHO + H20 C02/HCHO -0.48

C02 + 2e~ + 2H+ CO + H20 co2/co -0.53

C02 + 2e~ + 2H+ HCOOH 02/HC00H -0.61

C02 + e~ C02 co2/co2 -1.9

The reaction pathways of C02-to-fuel conversion and the product selectivity are quite complicated and poorly understood. Two possible pathways are proposed for C02-to-CH4 formation, i.e. (1) formaldehyde pathway: C02 HCOOH -> HCHO -> CH3OH -> CH4 and (2) carbine pathway: C02 CO -> C ■ -> CH:i ■ -> CH3OH/CH4.7 ns u2 The C02-to-C0 formation is possibly through single-electron transfer pathway C02 ->■ C02 -> CO or through two-electron transfer pathway C02 -> CO.133 For the single electron process it is difficult to form C02 since the C027C02 potential at -1.9 V (vs. normal hydrogen electrode (NHE)) needs the conduction band edge to be very negative.7 Whereas, the conduction band edge of defect-free Ti02 is -0.5 V43 (vs. NHE, pH 7.0). The valence band edge of Ti02 locates at +2.7 V, and thus, theoretically it can oxidize water to form H+ and 02.

Although desorption of the products is not one of the focuses in this review, it is important to the performance of Ti02 during C02 photoreduction processes, because the rate of product desorption has an influence on the number of surface active sites of the catalyst. Desorption activation energy of CO on most reduced surface is 11.7 kcal mol1, larger than 10.5 kcal mol"1 on the oxidized Ti02 surface.134 As a result, the oxygen deficient Ti02 needs a relatively higher temperature (350 K) to desorb CO compared to oxidized Ti02 (225 K). In another word, C02 reduction on oxygen deficient Ti02 needs a relatively higher temperature than defect-free Ti02.

The drawbacks of Ti02 as a photocatalyst (large band gap, fast charge recombination, etc.) that

we discussed in the introduction section could be partially mitigated by engineering defects in Ti02. To

date, oxygen deficient Ti02, or metal/nonmetal doped Ti02 have been widely applied in photocatalytic

C02 reduction with H20. Table 4 summarizes the photoreduction performance of various types of

defective Ti02 in terms of products yields, selectivity, and catalyst stability, which are correlated with the catalyst preparation methods and materials properties.

The photocatalytic activity for C02 photoreduction is mainly reported by two methods: (1) the amount (|imol) or rate (|imol/h) of C02 reduction products or that normalized by catalyst mass (|imol/g or (imol/h/g), and (2) the quantum yield (QY) or quantum efficiency (QE). For the convenience of comparison among different studies, the production rate in |imol/h/g is most frequently reported in the literature. However, the photocatalytic activity of typically has a nonlinear relationship with the mass or concentration of photocatalysts.135 This is because the light absorption capacity of photocatalysts highly depends on the mass (or thickness) and dispersion of the photocatalysts. In addition, because the catalyst particle sizes and morphologies are different, the same mass amount of catalysts may have drastically different surface areas and number of active sites. As a result, some studies report production rate normalized by surface area. The mass of catalyst to apply, however, is a more important factor in large-scale applications. In consideration of all the above factors, the authors suggest to use production rate in (imol/h while reporting the catalyst mass as a more appropriate method to evaluate C02 photoreduction performance.

Quantum yield (QY) and quantum efficiency (QE) are interchangeably used when reporting C02 photoreduction performance.21136"139 Two standards, external quantum yield (EQY) and internal quantum yield (IQY) are typically used, according to the following equations.

Y(N„ X number of reaction product)

EQY = ---r. -]—-- X 100%

number of incident photons

X number of reaction product) ^^ number of absorbed photons 100 /o

where Ne is the number of electrons required to produce a reaction product. For example, two electrons are required for one molecule of CO production while eight electrons for CH4, and thus Ne is 2 and 8 for CO and CH4, respectively. EQY is sometimes referred to as apparent quantum efficiency (AQY). EQY is normally lower than IQY because not all incident photons are absorbed. However, when a laser or monochromatic light is used as the photo-excitation source, EQY and IQY values could be close. IQY excludes the variations caused by different light sources used; however, IQY may be misleading if a poor light absorber is used. EQY is more practical when evaluating catalytic performance driven by sunlight. On the other hand, none of the QY calculations takes the catalyst mass into consideration, which provides little information on reaction scalability.

In summary, the authors suggest to use production rate (^mol/h) and QY while reporting the catalyst mass to evaluate the photocatalytic performance of photocatalysts. In addition, other factors such as light source, reactor design, and reaction conditions can also affect the values of production rate and QY, as listed in Table 4. The authors suggest in future research in the field of CO2 photoreduction, a standard reference material such as TiO2 P25 to be tested in each study and compared against the new catalysts developed in that study. The differential performance between P25 and developed catalysts may be a more appropriate evaluation standard.

Table 4. A list of defective TiO2 prepared by different methods and their catalytic performance for CO2 photoreduction with H2O.

Defect Type I: VO (Oxygen vacancy)

Photocatalyst Defect formation method CO2 photoreduction condition Production rate & selectivity Stability Materials advancement Ref.

Mass Reaction gas Light source Time ^mol/h* QY* *

1 at.% Cu-TiO2- x In situ thermal treatment in He or H2 50 mg 2ml/min CO2+H2O continuous flow 150 W solar simulator 6.5 h H2 pretreated: CO=0.192 CH4=0.034 N/A Gradually dropped to zero Cu valence and oxygen vacancy 19

He pretreated: CO=0.092 CHt=0.004 N/A

4% Cu@TiO2 Annealing at 450°C for 2 h in high vacuum 20 mg CO2 generated from NaHCO3/ H2SO4 reaction Visible light of 500 W Xe lamp (X > 400 nm) 6 h CO=0.162 CH4=0.027 N/A Production rate was stable within 6 h photoreaction test Metallic Cu core and black TiO2 with oxygen vacancy 140

TiO2-x w/ exposed {001} /{101} facets Solid-state method through NaBH4 reduction 40 mg 4 ml/min CO2+H2O continuous flow UV: 100 W mercury vapor lamp; Visible: 450 W Xe lamp (X > 400 nm) 5 h UV: CO=0.44 Vis: CO=0.208 UV: CO=0.31% Vis: CO=0.134% N/A Heterojunction between facets and the oxygen vacancies 21

Brookite TiO2-x Thermal treatment in He 100 mg 2 ml/min. CO2+H2O continuous flow 150 W solar simulator 6 h CO+CH4=0.315 N/A N/A Oxygen vacancy and Ti3+ 13

TiO2-x TIIP with H2O2 refluxing at 90°C for 10 h 100 mg Closed system with 20 psi CO2+H2O gas Solar simulator at 80 mW cm-2 6 h CO=0.0157 CH4=0.0041 CO =0.0141% CH4 =0.0148% Stable after 106 days Oxygen vacancy 13

Defect Type II: Dopants

Photocatalyst Defect formation method CO2 photoreduction condition Production rate & selectivity Stability Materials advancement Ref.

Mass Reaction gas Light source Time ^mol/h* QY* *

TiO2/Cu-TiO2 and TiO2/Fe-TiO2 double layered film Solvothermal treatment of CuCl2 (or FeCl2) and TTIP 200 mg CO2: H2O (1:2) continuous flow 6.0 W/cm2 mercury lamp with a wavelength at 365 nm 8 h TiO2/Cu-TiO2: CH4 =4.375 TiO2/Fe-TiO2: CH4=55 N/A Not stable on TiO2/Cu-TiO2, but stable on TiO2/Fe-TiO2 Cu or Fe doping 141,142

10%In-TiO2 Sol gel method of indium nitrate with TTIP 250 mg CO2 +H2O+He continuous flow 500 W mercury flash lamp 2 h CO=58.75 CH4=143.75 N/A N/A In doping 64

N-TiO2 nanotube heated in 500 °C Hydrothermal method using hexamethylene tetramine 150 mg CO2 bubbled in NaOH solution Visible light of 500W tungsten-halogen lamp 12 h HCOOH=156 CH3OH=14 HCHO=12 N/A Production rate dropped to near zero after 12 h irradiation N doping 143

10%I-TiO2 Hydrolysis of TTIP in iodic acid solution followed by hydrothermal 200 mg Closed system CO2+H2O mix gas Visible light of 150 W solar simulator 3.5 h CO=0.48 N/A Stable production rate in 3.5 h I doping 47

7%Ag-TiO2 Sol-gel method of TTIP and AgNO3 100 mg 100 ml CO2 saturated solution 8W Hg lamp (254 nm) 24 h CH4=0.033 CH3OH =0.0075 N/A N/A Ag doping 65

2.5 at.% Co-TiO2 Sol-gel method with cobalt acetate and tetrabutyl titanate 100 mg Closed system with 3 mL DI H2O and 80 kPa CO2 gas Visible light of 300 W Xe arc lamp with an L-42 glass filter 6 h CO=0.194 CH4=0.009 O2=0.0133 H2=0.074 N/A N/A Co doping 63

Cu and C co-modified TiO2 Sol-gel 125 mg CO2 bubbled in NaOH solution Peak light wavelength at 253.7 nm 12 h CO=1.25 CH4=0.3125 N/A N/A Cu and C co-modification 144

10%In-TiO2 Sol-gel method of In(NO3)3 and TiO2 552 mg CO2+H2O+He continuous flow 200 W mercury lamp peak at 252 nm 10 h CO=481 CH,=27.7 CO=0.1% CH4=0.022% Stable in 10 h irradiation In doping 139

* Production rates are converted to ^mol/h from the original values reported in the references

** QY = internal quantum yield

4.1 Productivity

As we discussed in section 3, by virtue of the possible advantages of defects for the photocatalytic performance of TiO2, the productivity of CO2 photocatalytic conversion can be greatly improved as a consequence of (1) the enhanced adsorption of the gas molecules at the defect sites, (2) the enhanced light absorption resulted from the defect-assisted band structure change, and (3) the enhanced charge separation. As demonstrated by Liu et al.50, the in situ thermally treated TiO2 in He facilitated the formation of oxygen vacancies. After the thermal treatment, oxygen deficient TiO2 showed 2.2, 10.3, and 3.9 times higher CO production and 13.3, 8.2, and 5 times higher CH4 production on anatase, brookite, and rutile (Fig. 14a), respectively. The enhanced dissociation of the adsorbed CO2 on the defective surface of TiO2 was also evidenced by in situ FTIR investigation. Ag doped TiO2 also showed enhanced CO2 photocatalytic conversion (Fig. 14b). The yield of CH4 and CH3OH increases when modifying the TiO2 by Ag is caused by two mechanisms: up to 5% of Ag in TiO2 the Ag impurity band inside the TiO2 bandgap decreases the absorption edge and increases electron-hole generation; above 5% of Ag in TiO2 Ag metallic clusters are formed in TiO2 crystals with Shottky barrier at the metal-semiconductor interface, which spatially separates electron and holes and decreases their recombination.

Fig. 14 Production of CO and CH4 on the three unpretreated and He pretreated TiO2 polymorphs for 6 h photo-illumination (a), and production of CH4 and CH3OH on Ag doped TiO2 for 24 h photo-illumination. (Reprinted with permission from ref13, copyright 2012 ACS, and ref65, copyright 2010 Elsevier Science.)

4.2 Selectivity

CO2 photoreduction is a multi-electron process that involves a series of reduction steps as outlined in Section 3. The reported products of CO2 photoreduction with water on TiO2 are CH4, CO, H2, CH3OH, etc. Selectivity of products is related to the band position relative to the redox potentials. The possible reactions involved in CO2 photoreduction with water on TiO2 photocatalyst and their reduction potentials are listed in the Table 3. Defective TiO2 has an altered band position or inter-band defect sites, which lead to different selectivity of products. Generally, two electrons are needed to form CO, while CH4 formation consumes eight electrons. Thus, selectivity of the products also depends on the local density of electrons. For example, reduced Cu-TiO2 had higher CH4 production than TiO2, because Cu modification improved the TiO2 conductivity and density of electrons on the surface.19 Wang et al.63 synthesized Co-doped ordered-mesoporous TiO2, donated as Co-OMT-x, where x (1-8) represented the molar ratio Co:Ti (0.2, 0.5, 1, 2.5, 5, 10, 15, and 20%). These photocatalysts were tested for CO2 photoreduction under visible light illumination, with CO, CH4, H2 and O2 as the products (Fig. 15a). The production of CO and CH4 are shown in Fig. 15b. The band structures of the samples are drawn in Fig. 15c. Both the CB and the VB edges of Co doped TiO2 shifted negatively with increase of the Co/Ti ratio, resulting in the decreasing of the band gap energy, which contributed to a higher visible light absorption and a higher CO+CH4 production. In addition, the increasing of Co dopant concentration may cause the raising of the local electron density, which led to the increase in CH4 selectivity. Above a certain Co concentration, Co3O4/Co-doped TiO2 nanocomposites with oxygen vacancies were obtained, which also caused the improvement of CH4 generation.

Visible-light

Co-doped TiO:

^^Visible-lightC02+h*

I,-- K""1

^^^^ "y CO+CH,

h* h" I

i 1. >

is = 2.

«TeT °

jfS s?

COfCOj

H/h,O ^CH(OHi co; chi/co]

Redox potentials

Fig. 15 (a) Schematic illustration of the basic mechanism of the TiO2 photocatalytic process. (b) Comparison of the photocatalytic activity of the samples with common photocatalysts under visible light, calculated according to the amounts evolved in 6 h, (c) Schematic illustration of the band structures of the samples (Adapted with permission from ref63. Copyright 2015 RSC)

4.3 Stability

Stability is another important aspect that reflects the overall and cycling performance of TiO2. Although some studies reported better photocatalytic activity on defective TiO2 than pristine TiO2, there is very few research discussing its catalytic stability. The activity of TiO2 often starts to drop after a few hours during the reaction of CO2 photoreduction with H2O vapor, which was probably due to the competing adsorption of H2O over CO2 or the slow desorption of the intermediates/products during reaction on the surface.53 145 As discussed in Section 3.1.1, for the surface of defective TiO2, the defect sites may help the dissociative chemisorption of H2O, forming OH groups, which help the chemisorption of CO2 and the formation of bicarbonate and the CO2- species. Therefore, engineering the surface of TiO2 with defect sites might be able to help prevent the decline of the photocatalytic CO2 conversion activity. However, the VO sites are likely healed by the adsorbed H2O and CO2 during photocatalytic reactions.

Hence, how to maintain the population of defect sites and prolong the catalyst lifetime is of interest for future investigation.

5. Conclusions and future work

In this paper, we have reviewed the recent advances on the effects of TiO2 surface point defects on the CO2 photoreduction with H2O to produce solar fuels. It is highlighted that the defective TiO2 materials exhibit superior performance than pristine TiO2 for CO2 photoreduction. The possible reasons for the enhanced photocatalytic activity are explained. First, defects on the surface of TiO2 provides can enhance dissociative adsorption of CO2 and H2O molecules. Second, the change of the electronic band structure of TiO2 caused by the introduction of the defect can improve the sunlight absorption efficiency. Third, the defect induced charge traps can hinder the charge recombination, and the defects may help the migration of electrons to the surface. Although enormous progresses have been made, as depicted in this review, there is still a long way ahead before the defective TiO2 could be successfully used to push CO2 photocatalysis toward practical application with both economic and environmental benefits. To achieve this goal, the following directions are highly recommended for future research.

5.1 Understanding the origin of point defects on TiO2

It is still in a debate whether the defect has a positive or negative impact on the photocatalytic performance of TiO2. The impact of the defect on TiO2 photocatalytic performance varies with the density, location, and type of the point defect. A high concentration of defects is more likely to hinder the CO2 photoreduction by acting as the recombination center. Varied types of defects can behave significantly different. Besides, the same type of defects at different locations can have different functions. Therefore, it is important to investigate the origin of the defect by applying appropriate characterization methods and combining with the photocatalytic activity results to optimize the structure of the defective TiO2.

5.2 Exploring CO2 photoreduction mechanism involving point defects

The reaction pathways of CO2 photoreduction are very complicated, and point defects are involved in each step of the reaction process. Thus, the real role of the defects in the CO2 photocatalytic reduction with H2O needs to be more deeply understood and systematically investigated in the future. In situ STM analysis during the CO2 photocatalytic reaction will be useful to monitor the defect movement and explore the mechanism of the enhanced photocatalytic activity.

5.3 Constructing high-performance defect-rich TiO2 photocatalysts

Based on the knowledge of the above two aspects, the next step is to design and synthesize low-cost, high-performance defective TiO2. Though VO facilitates the dissociation of H2O and helps CO2 adsorption, the VO sites may be consumed during the reaction. Moreover, the binding energy of O2 on VO sites are higher than that of CO2, so it is less likely for CO2 to adsorb on VO in the presence of O2.105 Meanwhile, the VO sites can be healed by adsorbed O2146-148 and thus become unavailable for the CO2 photoreduction process. Thus, another critical issue to address is the recovery of vacancy sites during the photocatalytic process. Defect-rich TiO2 with unique geometric and electronic structures is highly desired to enhance the photocatalytic activity and stability, which is important toward the realization of practical application of CO2 reduction to solar fuels.

Acknowledgement

The authors acknowledge the financial support from National Science Foundation CAREER Award (CBET#1538404).

References

(1) (2)

8398-8406.

52, 7372-7408.

(8) Yang, X.; Salzmann, C.; Shi, H.; Wang, H.; Green, M.; Xiao, T. J. Phys. Chem. A 2008, 112, 10784-10789.

(9) Pan, X.; Yang, M.; Fu, X.; Zhang, N.; Xu, Y. Nanoscale 2013, 5, 3601-3614.

(10) Mo, L.; Bai, Y.; Xiang, Q.; Li, Q.; Wang, J.; Ibrahim, K.; Cao, J. Appl. Phys. Lett. 2014, 105, 202114.

(11) Zhang, H.; Liang, C.; Liu, J.; Tian, Z.; Wang, G.; Cai, W. Langmuir 2012, 28, 3938-3944.

Song, C. Catal. Today 2006, 115, 2-32.

Centi, G.; Perathoner, S. Catal. Today 2009, 148, 191-205.

Yang, C.; Yu, Y.; van der Linden, B.; Wu, J.; Mul, G. J. Am. Chem. Soc. 2010, 132,

Anpo, M.; Yamashita, H.; Ichihashi, Y.; Ehara, S. J. Electroanal. Chem. 1995, 396, 21-

Ni, M.; Leung, M.; Leung, D.; Sumathy, K. Renew. Sust. Energ. Rev. 2007, 11, 401-425. Kumar, S.; Devi, L. J. Phys. Chem. A 2011, 115, 13211-13241.

Habisreutinger, S.; Schmidt-Mende, L.; Stolarczyk, J. Angew Chem. Int. Ed. Engl. 2013,

(12) Brandao, F.; Pinheiro, M.; Ribeiro, G.; Medeiros-Ribeiro, G.; Krambrock, K. Phys. Rev. B 2009, 80.

(13) Liu, L.; Zhao, H.; Andino, J.; Li, Y. ACSCatal. 2012, 2, 1817-1828.

(14) Rodriguez, M.; Peng, X.; Liu, L.; Li, Y.; Andino, J. J. Phys. Chem. C 2012, 116, 1975519764.

(15) Yan, J.; Wu, G.; Guan, N.; Li, L.; Li, Z.; Cao, X. Phys. Chem. Chem. Phys. 2013, 15, 10978-10988.

(16) Lin, Z.; Orlov, A.; Lambert, R.; Payne, M. J. Phys. Chem. B 2005, 109, 20948-20952.

(17) Li, Y.; Aschauer, U.; Chen, J.; Selloni, A. Acc. Chem. Res. 2014, 47, 3361-3368.

(18) Chen, X.; Liu, L.; Yu, P.; Mao, S. Science 2011, 331, 746-750.

(19) Liu, L.; Gao, F.; Zhao, H.; Li, Y. Appl. Catal., B 2013, 134, 349-358.

(20) Samiee, M.; Luo, J. Mater. Lett. 2013, 98, 205-208.

(21) Liu, L.; Jiang, Y.; Zhao, H.; Chen, J.; Cheng, J.; Yang, K.; Li, Y. ACS Catal. 2016, 6, 1097-1108.

(22) Wang, R.; Hashimoto, K.; Fujishima, A.; Chikuni, M.; Kojima, E.; Kitamura, A.; Shimohigoshi, M.; Watanabe, T. Nature 1997, 388, 431-432.

(23) Wang, R.; Sakai, N.; Fujishima, A.; Watanabe, T.; Hashimoto, K. J. Phys. Chem. B 1999, 103, 2188-2194.

(24) Wang, R.; Hashimoto, K.; Fujishima, A.; Chikuni, M.; Kojima, E.; Kitamura, A.; Shimohigoshi, M.; Watanabe, T. Adv. Mater. 1998, 10, 135-138.

(25) Sakai, N.; Fujishima, A.; Watanabe, T.; Hashimoto, K. J. Phys. Chem. B 2001, 105, 3023-3026.

(26) Miyauchi, M.; Nakajima, A.; Watanabe, T.; Hashimoto, K. Chem. Mater. 2002, 14, 28122816.

(27) Sun, R.; Nakajima, A.; Fujishima, A.; Watanabe, T.; Hashimoto, K. J. Phys. Chem. B 2001, 105, 1984-1990.

(28) Wang, L.; Baer, D. R.; Engelhard, M. H.; Shultz, A. Surf Sci. 1995, 344, 237-250.

(29) Mezhenny, S.; Maksymovych, P.; Thompson, T.; Diwald, O.; Stahl, D.; Walck, S.; Yates, J. Chem. Phys. Lett. 2003, 369, 152-158.

(30) Bouzoubaa, A.; Markovits, A.; Calatayud, M.; Minot, C. Surf. Sci. 2005, 583, 107-117.

(31) Zubkov, T.; Stahl, D.; Thompson, T.; Panayotov, D.; Diwald, O.; Yates, J. J. Phys. Chem. B 2005, 109, 15454-15462.

(32) Sekiya, T.; Ichimura, K.; Igarashi, M.; Kurita, S. J. Phys. Chem. Solids 2000, 61, 12371242.

(33) Mo, S.; Ching, W. Phys. Rev. B 1995, 51, 13023.

(34) Pan, H.; Gu, B.; Zhang, Z. J. Chem. Theory Comput. 2009, 5, 3074-3078.

(35) Hebenstreit, W.; Ruzycki, N.; Herman, G.; Gao, Y.; Diebold, U. Phys. Rev. B 2000, 62,

R16334.

(36) Morgan, B.; Watson, G. J. Phys. Chem. C 2009, 113, 7322-7328.

(37) Neamen, D. Semiconductor Physics and Devices: Basic Principles; McGraw-Hill Education, 2012.

(38) Colon, G.; Maicu, M.; Hidalgo, M.; Navio, J. Appl. Catal., B 2006, 67, 41-51.

(39) Kim, S.; Hwang, S.; Choi, W. J. Phys. Chem. B 2005, 109, 24260-24267.

(40) Wang, C.; Böttcher, C.; Bahnemann, D.; Dohrmann, J. J. Mater. Chem. 2003, 13, 23222329.

(41) Zhao, X.; Yin, J. Chem. Mater. 2002, 14, 2258-2263.

(42) Dvoranova, D.; Brezova, V.; Mazur, M.; Malati, M. Appl. Catal., B 2002, 37, 91-105.

(43) Sakthivel, S.; Kisch, H. Chemphyschem 2003, 4, 487-490.

(44) Ohno, T.; Mitsui, T.; Matsumura, M. Chem. Lett. 2003, 32, 364-365.

(45) Yu, J.; Yu, J.; Ho, W.; Jiang, Z.; Zhang, L. Chem. Mater. 2002, 14, 3808-3816.

(46) Yang, K.; Dai, Y.; Huang, B.; Whangbo, M. J. Phys. Chem. C 2009, 113, 2624-2629.

(47) Zhang, Q.; Li, Y.; Ackerman, E.; Gajdardziska-Josifovska, M.; Li, H. Appl. Catal., A 2011, 400, 195-202.

(48) Asahi, R.; Morikawa, T.; Ohwaki, T.; Aoki, K.; Taga, Y. Science 2001, 293, 269-271.

(49) Morris Hotsenpiller, P.; Bolt, J.; Farneth, W.; Lowekamp, J.; Rohrer, G. J. Phys. Chem. B 1998, 102, 3216-3226.

(50) Diwald, O.; Thompson, T.; Zubkov, T.; Goralski, E.; Walck, S.; Yates, J. J. Phys. Chem. B 2004, 108, 6004-6008.

(51) Batzill, M.; Morales, E.; Diebold, U. Phys. Rev. Lett. 2006, 96, 026103.

(52) Shevlin, S.; Woodley, S. J. Phys. Chem. C 2010, 114, 17333-17343.

(53) Zhao, H.; Liu, L.; Andino, J.; Li, Y. J. Mater. Chem. A 2013, 1, 8209-8216.

(54) Chen, X.; Burda, C. J. Am. Chem. Soc. 2008, 130, 5018-5019.

(55) Wu, Z.; Dong, F.; Liu, Y.; Wang, H. Catal. Commun. 2009, 11, 82-86.

(56) Park, J.; Kim, S.; Bard, A. Nano Lett. 2006, 6, 24-28.

(57) George, S.; Pokhrel, S.; Ji, Z.; Henderson, B.; Xia, T.; Li, L.; Zink, J.; Nel, A.; Mädler, L. J. Am. Chem. Soc. 2011, 133, 11270-11278.

(58) Li, H.; Zhang, X.; Huo, Y.; Zhu, J. Environ. Sci. Technol. 2007, 41, 4410-4414.

(59) Klosek, S.; Raftery, D. J. Phys. Chem. B 2001, 105, 2815-2819.

(60) Sakthivel, S.; Janczarek, M.; Kisch, H. J. Phys. Chem. B 2004, 108, 19384-19387.

(61) Li, W.; Wang, Y.; Lin, H.; Shah, S.; Huang, C.; Doren, D.; Rykov, S.; Chen, J.; Barteau, M. Appl. Phys. Lett. 2003, 83, 4143-4145.

(62) Deng, L.; Chen, Y.; Yao, M.; Wang, S.; Zhu, B.; Huang, W.; Zhang, S. J. Sol-Gel Sci. Technol. 2010, 53, 535-541.

(63) Wang, T.; Meng, X.; Liu, G. G.; Chang, K.; Li, P.; Kang, Q.; Liu, L. Q.; Li, M.; Ouyang, S.; Ye, J. H. J. Mater. Chem. A 2015, 3, 9491-9501.

(64) Tahir, M.; Amin, N. Appl. Catal., B 2015, 162, 98-109.

(65) Koci, K.; Mateju, K.; Obalova, L.; Krejcikova, S.; Lacny, Z.; Placha, D.; Capek, L.; Hospodkova, A.; Solcova, O. Appl. Catal., B 2010, 96, 239-244.

(66) Choi, W.; Termin, A.; Hoffmann, M. J. Phys. Chem. 1994, 98, 13669-13679.

(67) Lira, E.; Wendt, S.; Huo, P.; Hansen, J.; Streber, R.; Porsgaard, S.; Wei, Y.; Bechstein, R.; L^gsgaard, E.; Besenbacher, F. J. Am. Chem. Soc. 2011, 133, 6529-6532.

(68) Lee, J.; Sorescu, D.; Deng, X. J. Am. Chem. Soc. 2011, 133, 10066-10069.

(69) Gong, X.; Selloni, A.; Batzill, M.; Diebold, U. Nat. Mater. 2006, 5, 665-670.

(70) He, Y. Y.; Tilocca, A.; Dulub, O.; Selloni, A.; Diebold, U. Nat. Mater. 2009, 8, 585-589.

(71) He, Y. Y.; Dulub, O.; Cheng, H.; Selloni, A.; Diebold, U. Phys. Rev. Lett. 2009, 102,

106105.

(72) Kong, M.; Li, Y.; Chen, X.; Tian, T.; Fang, P.; Zheng, F.; Zhao, X. J. Am. Chem. Soc. 2011, 133, 16414-16417.

(73) Leng, Y. In Materials Characterization; John Wiley & Sons (Asia) Pte Ltd: 2008, p 145170.

(74) Schaub, R.; Wahlstrom, E.; Ronnau, A.; Lagsgaard, E.; Stensgaard, I.; Besenbacher, F.

Science 2003, 299, 377-379.

(75) Tsai, M.; Chen, S.; Shen, P. Nano Lett. 2004, 4, 1197-1201.

(76) Hayden, B.; King, A.; Newton, M. J. Phys. Chem. B 1999, 103, 203-208.

(77) Jing, L.; Xin, B.; Yuan, F.; Xue, L.; Wang, B.; Fu, H. J. Phys. Chem. B 2006, 110, 17860-17865.

(78) Pan, J.; Maschhoff, B.; Diebold, U.; Madey, T. J. Vac. Sci. Technol, A 1992, 10, 24702476.

(79) Iijima, K.; Goto, M.; Enomoto, S.; Kunugita, H.; Ema, K.; Tsukamoto, M.; Ichikawa, N.; Sakama, H. J. Lumin. 2008, 128, 911-913.

(80) Wang, X.; Feng, Z.; Shi, J.; Jia, G.; Shen, S.; Zhou, J.; Li, C. Phys. Chem. Chem. Phys. 2010, 12, 7083-7090.

(81) Pandey, M.; D'cunha, R.; Tyagi, A. J. Alloys. Compd. 2002, 333, 260-265.

(82) Knorr, F.; Zhang, D.; McHale, J. Langmuir 2007, 23, 8686-8690.

(83) Nakamura, R.; Nakato, Y. J. Am. Chem. Soc. 2004, 126, 1290-1298.

(84) Shi, J. Y.; Chen, J.; Feng, Z.; Chen, T.; Lian, Y.; Wang, X.; Li, C. J. Phys. Chem. C 2007, 111, 693-699.

(85) De La Cruz, R.; Pareja, R.; Gonzalez, R.; Boatner, L.; Chen, Y. Phys. Rev. B 1992, 45,

(86) Sanyal, D.; Banerjee, D.; De, U. Phys. Rev. B 1998, 58, 15226.

(87) Dutta, S.; Chattopadhyay, S.; Jana, D.; Banerjee, A.; Manik, S.; Pradhan, S.; Sutradhar, M.; Sarkar, A. J. Appl. Phys. 2006, 100, 114328.

(88) Liu, X.; Zhou, K.; Wang, L.; Wang, B.; Li, Y. J. Am. Chem. Soc. 2009, 131, 3140-3141.

(89) Murakami, H.; Onizuka, N.; Sasaki, J.; Thonghai, N. J. Mater. Sci. 1998, 33, 5811-5814.

(90) Dannefaer, S.; Bretagnon, T.; Kerr, D. J. Appl. Phys. 1993, 74, 884-890.

(91) Li, Y.; Huang, J.; Peng, T. Y.; Xu, J.; Zhao, X. ChemCatChem 2010, 2, 1082-1087.

(92) Chakraverty, S.; Mitra, S.; Mandal, K.; Nambissan, P.; Chattopadhyay, S. Phys. Rev. B 2005, 71, 024115.

(93) Berger, T.; Sterrer, M.; Diwald, O.; Knözinger, E.; Panayotov, D.; Thompson, T.; Yates, J. J. Phys. Chem. B 2005, 109, 6061-6068.

(94) Li, M.; Hebenstreit, W.; Diebold, U.; Tyryshkin, A.; Bowman, M.; Dunham, G.; Henderson, M. J. Phys. Chem. B 2000, 104, 4944-4950.

(95) Di Valentin, C.; Finazzi, E.; Pacchioni, G.; Selloni, A.; Livraghi, S.; Paganini, M.; Giamello, E. Chem. Phys. 2007, 339, 44-56.

(96) Di Valentin, C.; Pacchioni, G.; Selloni, A.; Livraghi, S.; Giamello, E. J. Phys. Chem. B

2005, 109, 11414-11419.

(97) Liu, L.; Pitts, D.; Zhao, H.; Zhao, C.; Li, Y. Appl. Catal., A 2013, 467, 474-482.

(98) Straumanis, M.; Ejima, T.; James, W. Acta Crystallogr. 1961, 14, 493-497.

(99) Freund, H.; Roberts, M. Surf Sci. Rep. 1996, 25, 225-273.

(100) Chang, X.; Wang, T.; Gong, J. Energy Environ. Sci. 2016.

(101) Henderson, M. Surf. Sci. 1998, 400, 203-219.

(102) Burghaus, U. Catal. Today 2009, 148, 212-220.

(103) Tan, S.; Zhao, Y.; Zhao, J.; Wang, Z.; Ma, C.; Zhao, A.; Wang, B.; Luo, Y.; Yang, J.; Hou, J. Phys. Rev. B 2011, 84, 155418.

(104) Thompson, T.; Yates, J. Chem. Rev. 2006, 106, 4428-4453.

(105) He, H. Y.; Zapol, P.; Curtiss, L. J. Phys. Chem. C 2010, 114, 21474-21481.

(106) Acharya, D.; Camillone, N.; Sutter, P. J. Phys. Chem. C 2011, 115, 12095-12105.

(107) Pipornpong, W.; Wanbayor, R.; Ruangpornvisuti, V. Appl. Surf. Sci. 2011, 257, 1032210328.

(108) Yin, W.; Wen, B.; Bandaru, S.; Krack, M.; Lau, M.; Liu, L. Sci. Rep. 2016, 6.

(109) Kavan, L.; Zukalova, M.; Ferus, M.; Kurti, J.; Koltai, J.; Civis, S. Phys. Chem. Chem. Phys. 2011, 13, 11583-11586.

(110) Kurtz, R.; Stockbauer, R.; Madey, T.; Roman, E.; Desegovia, J. Surf. Sci. 1989, 218, 178200.

(111) Lindan, P.; Harrison, N.; Gillan, M. Phys. Rev. Lett. 1998, 80, 762-765.

(112) Bikondoa, O.; Pang, C.; Ithnin, R.; Muryn, C.; Onishi, H.; Thornton, G. Nat. Mater. 2006, 5, 189-192.

(113) Petrik, N.; Kimmel, G. J. Phys. Chem. C 2015, 119, 23059-23067.

(114) Dette, C.; Perez-Osorio, M.; Kley, C.; Punke, P.; Patrick, C.; Jacobson, P.; Giustino, F.; Jung, S.; Kern, K. NanoLett. 2014, 14, 6533-6538.

(115) Pacchioni, G. Chemphyschem 2003, 4, 1041-1047.

(116) Naldoni, A.; Allieta, M.; Santangelo, S.; Marelli, M.; Fabbri, F.; Cappelli, S.; Bianchi, C.; Psaro, R.; Dal Santo, V. J. Am. Chem. Soc. 2012, 134, 7600-7603.

(117) Nambu, A.; Graciani, J.; Rodriguez, J.; Wu, Q.; Fujita, E.; Sanz, J. F. J. Chem. Phys.

2006, 125, 094706.

(118) Lin, Z.; Orlov, A.; Lambert, R.; Payne, M. J. Phys. Chem. B 2005, 109, 20948-20952.

(119) Asahi, R.; Morikawa, T.; Ohwaki, T.; Aoki, K.; Taga, Y. science 2001, 293, 269-271.

(120) Tsui, L.; Zangari, G. J. Electrochem. Soc. 2014, 161, D3066-D3077.

(121) Papageorgiou, A.; Beglitis, N.; Pang, C.; Teobaldi, G.; Cabailh, G.; Chen, Q.; Fisher, A.; Hofer, W.; Thornton, G. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 2391-2396.

(122) Mercado, C.; Knorr, F.; McHale, J.; Usmani, S.; Ichimura, A.; Saraf, L. J. Phys. Chem. C 2012, 116, 10796-10804.

(123) Liu, L.; Zhao, C.; Li, Y. J. Phys. Chem. C 2012, 116, 7904-7912.

(124) Mrowetz, M.; Balcerski, W.; Colussi, A.; Hoffmann, M. J. Phys. Chem. B 2004, 108, 17269-17273.

(125) Motta, C.; El-Mellouhi, F.; Sanvito, S. Sci. Rep. 2015, 5.

(126) Walton, D.; Plenum Press: 1991.

(127) Morgan, B.; Watson, G. J. Phys. Chem. C 2010, 114, 2321-2328.

(128) Liu, L.; Li, Y. Aerosol Air Qual. Res. 2014, 14, 453-469.

(129) Inoue, T.; Fujishima, A.; Konishi, S.; Honda, K. Nature 1979, 277, 637-638.

(130) Yahaya, A.; Gondal, M.; Hameed, A. Chem. Phys. Lett. 2004, 400, 206-212.

(131) Zhou, Y.; Tian, Z.; Zhao, Z.; Liu, Q.; Kou, J.; Chen, X.; Gao, J.; Yan, S.; Zou, Z. ACS Appl. Mater. Inter. 2011, 3, 3594-3601.

(132) Dhakshinamoorthy, A.; Navalon, S.; Corma, A.; Garcia, H. Energy Environ. Sci. 2012, 5, 9217-9233.

(133) He, H.; Zapol, P.; Curtiss, L. Energy Environ. Sci. 2012, 5, 6196-6205.

(134) Linsebigler, A.; Lu, G.; Yates Jr, J. J. Chem. Phys. 1995, 103, 9438-9443.

(135) Davydov, L.; Pratsinis, S.; Smirniotis, P. Environ. Sci. Technol. 2000, 34, 3435-3442.

(136) Wang, W.; An, W. J.; Ramalingam, B.; Mukherjee, S.; Niedzwiedzki, D.; Gangopadhyay, S.; Biswas, P. J. Am. Chem. Soc. 2012, 134, 11276-11281.

(137) Yaghoubi, H.; Li, Z.; Chen, Y.; Ngo, H.; Bhethanabotla, V.; Joseph, B.; Ma, S.; Schlaf, R.; Takshi, A. ACS Catal. 2014, 5, 327-335.

(138) Baeissa, E. Ceram. Int. 2014, 40, 12431-12438.

(139) Tahir, M.; Amin, N. Appl. Catal., A 2013, 467, 483-496.

(140) Zhao, J.; Li, Y. X.; Zhu, Y.; Wang, Y.; Wang, C. Appl. Catal., A 2016, 510, 34-41.

(141) Park, M.; Kwak, B.; Jo, S. W.; Kang, M. Energy Convers. Manage. 2015, 103, 431-438.

(142) Do, J.; Im, Y.; Kwak, B.; Kim, J.; Kang, M. Chem. Eng. J. 2015, 275, 288-297.

(143) Zhao, Z.; Fan, J.; Wang, J.; Li, R. J. Catal. Commun. 2012, 21, 32-37.

(144) Yan, Y.; Yu, Y.; Cao, C.; Huang, S.; Yang, Y.; Yang, X.; Cao, Y. Crystengcomm 2016, 18, 2956-2964.

(145) Smith, S.; Li, Z.; Chen, L.; Dohnalek, Z.; Kay, B. J. Phys. Chem. B 2014, 118, 80548061.

(146) Henderson, M.; Epling, W.; Perkins, C.; Peden, C.; Diebold, U. J. Phys. Chem. B 1999, 103, 5328-5337.

(147) Lu, G.; Linsebigler, A.; Yates Jr, J. J. Chem. Phys. 1995, 102, 4657-4662.

(148) Rusu, C.; Yates, J. Langmuir 1997, 13, 4311-4316.

Highlights

• Point defects, mainly oxygen vacancies and dopants in TiO2 photocatalysts are reviewed on their effects on CO2 photoreduction with H2O.

• Diverse characterization methods are available to investigate the type and nature of point defects.

• Point defects can affect TiO2 properties including gas adsorption ability, electronic band structures, and charge transfer dynamics.

• Photocatalytic performance for CO2 photoreduction is correlated with various types of defective TiO2 materials.

• Recommendations for future research on defective TiO2 are provided.

Professor Ying Li, Texas A&M University, Email: yingli@tamu.edu. Dr. Ying Li is an Associate Professor and Pioneer Natural Resources Faculty Fellow III in the Department of Mechanical Engineering at Texas A&M University. He received his Ph.D. degree in Environmental Engineering Sciences at the University of Florida. His research interests include nanomaterials, catalysis and photocatalysis, solar energy conversion, CO2 utilization, batteries, water treatment, air quality and aerosol technology. He received the US National Science Foundation CAREER Award in 2013 to study CO2 photoreduction to solar fuels.

Dr. Fuping Pan is currently a postdoctoral research associate with Prof. Ying Li in the Department of Mechanical Engineering at Texas A&M University. He received his Ph.D. degree in Physical Chemistry from Lanzhou Institute of Chemical Physics, Chinese Academy of Sciences, in 2016. His current research is focused on the synthesis of functionalized nanoporous carbon materials and their applications in CO2 conversion.

Huilei Zhao received her BS degree in chemistry in 2011 from Shandong University, Jinan, China. She is currently a PhD student of Mechanical Engineering at Texas A&M University, under the direction of Prof. Ying Li. Her research interests involve the synthesis and characterization of TiO2 based nanomaterials for photocatalytic conversion of CO2 to solar fuels.