Chinese Journal of Chemical Physics  2018, Vol. 31 Issue (4): 547-554

The article information

Xiao Chen, Fang-liang Li, Guo Qing, Dong-xu Dai, Xue-ming Yang

Photoinduced Decomposition of Formaldehyde on Rutile TiO2(100)-(1×1)

Chinese Journal of Chemical Physics, 2018, 31(4): 547-554

http://dx.doi.org/10.1063/1674-0068/31/cjcp1806137

Article history

Accepted on: June 25, 2018
Photoinduced Decomposition of Formaldehyde on Rutile TiO2(100)-(1×1)
Xiao Chena,b, Fang-liang Lib,c, Guo Qinga, Dong-xu Daia, Xue-ming Yanga
Dated: Received on June 8, 2018; Accepted on June 25, 2018
a. State Key Laboratory of Molecular Reaction Dynamics, Dalian Institute of Chemical Physics, Chinese Academy of Science, Dalian 116023, China;
b. University of Chinese Academy of Sciences, Beijing 100049, China;
c. School of Physical Science and Technology, ShanghaiTech University, Shanghai 201210, China
*Author to whom correspondence should be addressed. Qing Guo, E-mail: guoqing@dicp.ac.cn
Abstract: We have investigated the photoinduced decomposition of formaldehyde (CH2O) on a rutile TiO2(100)-(1×1) surface at 355 nm using temperature-programmed desorption. Products, formate (HCOO-), methyl radical (CH3·), ethylene (C2H4), and methanol (CH3OH) have been detected. The initial step in the decomposition of CH2O on the rutile TiO2(100)-(1×1) surface is the formation of a dioxymethylene intermediate in which the carbonyl O atom of CH2O is bound to a Ti atom at the five-fold-coordinated Ti4+ (Ti5c) site and its carbonyl C atom bound to a nearby bridge-bonded oxygen (Ob) atom, respectively. During 355 nm irradiation, the dioxymethylene intermediate can transfer an H atom to the Ob atom, thus forming HCOO- directly, which is considered as the main reaction channel. In addition, the dioxymethylene intermediate can also transfer methylene to the Ob row and break the C-O bond, thus leaving the original carbonyl O atom at the Ti5c site. After the transfer of methylene, several pathways to products are available. Thus, we have found that Ob atoms are intimately involved in the photoinduced decomposition of CH2O on the rutile TiO2(100)-(1×1) surface.
Key words: Rutile TiO2(100)-(1×1)    Formaldehyde    Temperature-programmed desorption    Photoinduced decomposition
Ⅰ. INTRODUCTION

Titanium dioxide (Ti$\rm{O}_2$), is one of the most important metal oxides used in catalysis and photocatalysis [1-5]. Formaldehyde (C$\rm{H}_2$O) is a key species (reagent, intermediate, or product) in various catalytic and photocatalytic reactions, such as methanol (C$\rm{H}_3$OH) synthesis [6-8], C$\rm{H}_3$OH oxidation [9-13], and hydrocarbon production [14, 15]. In addition, C$\rm{H}_2$O is one of the main indoor air pollutions in our daily life. It has been reported that Ti$\rm{O}_2$-based catalysts are widely used in thermally catalytic and photocatalytic reactions involving C$\rm{H}_2$O. Therefore, it is of significant importance to gain an insightful understanding of the interactions of C$\rm{H}_2$O with Ti$\rm{O}_2$ surfaces.

The adsorption and reactions of C$\rm{H}_2$O on Ti$\rm{O}_2$ surfaces have been extensively studied both experimentally and theoretically [16-29]. It has been well-established that C$\rm{H}_2$O can adsorb on Ti$\rm{O}_2$ surfaces in two different configurations [19, 21-23, 25, 28, 29]. First, C$\rm{H}_2$O can weakly adsorb at a surface 5-fold-coordinated $\rm{Ti}^{4+}$ ($\rm{Ti}_\rm{5c}$) site in a monodentate configuration (${\eta}$$^{1}-\rm{CH}_2O), where it binds weakly via its O atom to the surface \rm{Ti}_\rm{5c} atom. Alternatively, it can also adsorb in a bidentate fashion ({\eta}$$^{2}$-$\rm{CH}_2$O), where it binds to the surface strongly with its O atom bound to a surface $\rm{Ti}_\rm{5c}$ atom, and its C atom bound to an adjacent bridging oxygen ($\rm{O}_\rm{b}$) atom. When Ti$\rm{O}_2$ surfaces contain bridging oxygen vacancy ($\rm{O}_\rm{v}$) sites, C$\rm{H}_2$O can adsorb at $\rm{O}_\rm{v}$ sites [20, 23, 29]. Further heating may result in the formation of other products via the carbon-carbon coupling reactions of two C$\rm{H}_2$O molecules, such as ethylene ($\rm{C}_2$$\rm{H}_4) [19, 21-23, 29]. In addition, the possibility of forming paraformaldehyde chains on rutile Ti\rm{O}_2(110) has also been reported by Wöll and coworkers [25, 30]. The photochemistry of C\rm{H}_2O has also been investigated extensively on various Ti\rm{O}_2 surfaces and formate (HCO\rm{O}^{-}) is observed as a main product [17-19, 27, 29, 31]. On rutile Ti\rm{O}_2(110), Xu and coworkers found that the photoinduced decomposition of C\rm{H}_2O could occur efficiently to produce HCO\rm{O}^{-}, methyl radical (C\rm{H}_3$$\cdot$) and $\rm{C}_2$$\rm{H}_4 in the absence of surface oxygen species [17]. It was proposed that although lattice oxygen atoms may not appear in HCO\rm{O}^{-} product, they are intimately involved in the photoinduced decomposition of C\rm{H}_2O on rutile Ti\rm{O}_2(110). Later, Cremer and coworkers reported that the formation of HCO\rm{O}^{-} is the dominate reaction channel and the efficiency of HCO\rm{O}^{-} formation on rutile Ti\rm{O}_2(110) with surface oxygen species is about 4 times larger than that without surface oxygen species [18]. On the reduced anatase Ti\rm{O}_2(001)-(1\times4) [27] and rutile Ti\rm{O}_2(011)- (2\times1) surfaces [29], the photolysis of C\rm{H}_2O also produces HCO\rm{O}^{-} as the main product in the absence of surface oxygen species. The rutile Ti\rm{O}_2(100)-(1\times1) surface is one of the common facets of rutile Ti\rm{O}_2, but the photochemistry of C\rm{H}_2O on the rutile Ti\rm{O}_2(100)-(1\times1) surface has not been reported and an insightful understanding of photocatalytic reactions of C\rm{H}_2O on the rutile Ti\rm{O}_2(100)-(1\times1) surface is thus lacking. In this work, we have investigated the photocatalytic reactions of C\rm{H}_2O on the rutile Ti\rm{O}_2(100)-(1\times1) surface at 355 nm with temperature-programmed desorption (TPD) method. Without irradiation, nearly no thermal reaction products are observed. Under UV irradiation, C\rm{H}_2O is mainly decomposed into HCO\rm{O}^{-}. While, C\rm{H}_3OH, C\rm{H}_3$$\cdot$, and $\rm{C}_2$$\rm{H}_4 are detected as minor products at elevated temperature during the TPD process. These results will help broaden the fundamental understandings of C\rm{H}_2O photochemistry on Ti\rm{O}_2 surfaces. Ⅱ. EXPERIMENTAL METHODS TPD experiments were carried out in an ultrahigh vacuum (UHV) chamber with a base pressure of 6\times$$10^{-11}$ Torr. Details of this TPD apparatus have been described elsewhere [10]. TPD signals were detected by a quadrupole mass spectrometer (Extrel). The third harmonic (355 nm) output of a diode pumped, solid state, Q-switched 1064 nm laser (Spectra-Physics) was used as the UV light resource for the photocatalytic reactions of C$\rm{H}_2$O in our experiments. The laser was operated with a pulse time of 12 ns and a repetition rate of 50 kHz. The average power of the light used in our experiments was approximately 20 mW, corresponding to $\sim$6.5$\times$$10^{16} photons/(\rm{cm}^2$$\cdot$s).

The rutile Ti$\rm{O}_2$(100) single crystal was purchased from Princeton Scientific Corp., with a size of 10 mm$\times$10 mm$\times$1 mm. The surface was cleaned by repeated cycles of $\rm{Ar}^+$ sputtering and annealing in $>$4$\times$$10^{-7} Torr \rm{O}_2 at 800 K. Then the ordering and cleanness of the sample were confirmed by a sharp (1\times1) low energy electron diffusion (LEED) pattern and Auger electron spectroscopy (AES), respectively. C\rm{H}_2O was obtained via the thermal decomposition of paraformaldehyde (95% purity, Sigma-Aldrich). Prior to experiments, the purity of C\rm{H}_2O was checked by our spectrometer. The purified C\rm{H}_2O was introduced into the surface through a calibrated molecular beam doser at about 120 K. TPD spectra were measured with a ramping rate of 2 K/s, and with the surface facing the mass spectrometer. Ⅲ. RESULTS A. The adsorption and reactions of C{\rm{H}}_\textbf{2}O on the rutile Ti\rm{O}_\textbf{2}(100)-(1\times1) surface Before TPD experiments, the surface condition was checked by LEED and water (\rm{H}_2O) TPD spectra, respectively. As shown in FIG. 1, a sharp (1\times1) LEED pattern for the rutile Ti\rm{O}_2(100)-(1\times1) surface is observed, which confirms the ordering of the surface. To further confirm the surface condition, TPD spectra of \rm{H}_2O at different coverages were subsequently collected. At the highest \rm{H}_2O coverage (4 ML, 1 ML=7.36\times$$10^{14}$ molecules/$\rm{cm}^2$), four main desorption features at 147, 167, 242, and 300 K are detected in our TPD spectra (FIG. 2), which is similar to previous results of $\rm{H}_2$O on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface [32, 33], indicating that a well-defined and unreconstructed (1$\times$1) surface has been obtained. On the basis of previous studies [32, 33], the 147, 167, 242, and 300 K peaks are due to desorption of $\rm{H}_2$O from multilayer (ice layer), second layer, molecular adsorption at $\rm{Ti}_\rm{5c}$ sites, and dissociative adsorption at $\rm{Ti}_\rm{5c}$ sites, respectively. Whereas, no desorption peak at higher temperature is detected, implying that no surface $\rm{O}_\rm{v}$ sites exist on the surface.

 FIG. 1 The LEED (low energy electron diffraction) pattern for the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface at $E_\rm{el}$=50 eV after the surface cleaning process was accomplished.
 FIG. 2 Typical spectra collected at mass-to-charge ($m/z$) of 18 ($\rm{H}_2$$\rm{O}^+) after the rutile Ti\rm{O}_2(100)-(1\times1) surfaces were dosed with various coverages of \rm{H}_2O at 120 K. Afterwards, we carried out experiments with C\rm{H}_2O adsorption on the rutile Ti\rm{O}_2(100)-(1\times1) surface. As shown in FIG. 3, TPD spectra were collected at a mass-to-charge ratio (m/z) of 30 (C\rm{H}_2$$\rm{O}^+$) after rutile Ti$\rm{O}_2$(100)-(1$\times$1) surfaces were dosed with different coverages of C$\rm{H}_2$O. At low coverages ($<$0.26 ML), a single desorption peak appears at $\sim$310 K and increases in intensity with increasing C$\rm{H}_2$O coverage, and its peak position nearly keeps unchanged. When the C$\rm{H}_2$O coverage is higher than 0.26 ML, an additional desorption peak appears at around 260 K. As C$\rm{H}_2$O coverage increases, the intensity of the 260 K peak increases, with its peak position shifting to lower temperature until 255 K. Although the 310 K peak is seriously overlapped with the 260 K peak, the occurrence of two desorption peaks shows the possibility of two different C$\rm{H}_2$O adsorption configurations on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface. Since the lowest surface temperature that we could achieve in this work is 120 K, the highest coverage of C$\rm{H}_2$O on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface is about 0.75 ML.

 FIG. 3 Typical spectra collected at mass-to-charge ($m/z$) of 30 (C$\rm{H}_2$$\rm{O}^+) after the rutile Ti\rm{O}_2(100)-(1\times1) surfaces were dosed with various coverages of C\rm{H}_2O at 120 K. Based on previous TPD results of C\rm{H}_2O on rutile Ti\rm{O}_\rm{2}(110) [17], the desorption of {\eta}$$^1$-$\rm{CH}_2$O gives a TPD peak at 280 K. Theoretical calculations [19, 20, 23-25] predict that the ${\eta}$$^1 configuration of C\rm{H}_2O has an adsorption energy of \sim0.7 eV, and the {\eta}$$^2$ configuration has an adsorption energy of $\sim$1.3 eV. C$\rm{H}_2$O may also adsorb at $\rm{O}_\rm{v}$ sites with an adsorption energy of about 0.89-0.99 eV. Thus, ${\eta}$$^{2}-\rm{CH}_2O is the more stable adsorption configuration. However, only the {\eta}$$^1$ configuration of C$\rm{H}_2$O can be formed on rutile Ti$\rm{O}_2$(110) after adsorption, and the transition from the ${\eta}$$^{1} configuration to the {\eta}$$^{2}$ configuration requires a long time [23]. In this work, the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface contains no $\rm{O}_\rm{v}$ sites, and thus C$\rm{H}_2$O can only adsorb at $\rm{Ti}_\rm{5c}$ sites. As shown in the C$\rm{H}_2$O TPD spectra (FIG. 3), the peak position of the 310 K peak is only about 55 K higher than that of the 255 K peak. According to the first-order thermal desorption model (with a typical pre-exponential factor value of $10^{13}$/s) [34], the difference between desorption energies of these two peaks is less than 0.2 eV. Therefore, the 310 and 255 K peaks could not be due to desorption of ${\eta}$$^{2}-\rm{CH}_2O and {\eta}$$^{1}$-$\rm{CH}_2$O, respectively. Conversely, the two desorption peaks may be due to desorption of ${\eta}$$^{1}-\rm{CH}_2O at \rm{Ti}_\rm{5c} sites. Because the population of \rm{Ti}_\rm{5c} sites on the rutile Ti\rm{O}_\rm{2}(100)-(1\times1) surface is about 1.5 times bigger than that on rutile Ti\rm{O}_2(110), the repulsive interaction between C\rm{H}_2O molecules adsorbed at \rm{Ti}_\rm{5c} sites on the rutile Ti\rm{O}_2(100)-(1\times1) surface will be much stronger than that on rutile Ti\rm{O}_2(110). In this case, the appearance of the 255 K peak is likely the result of the strong intermolecular repulsions between C\rm{H}_2O molecules adsorbed at \rm{Ti}_\rm{5c} sites. In addition to desorbed parent C\rm{H}_\rm{2}O, other possible desorption products were examined by comprehensively monitoring various signals of m/z=15, 18, 27, 28, 29, 30, 31, 32, and 46 (FIG. 4). No evidence of other reaction products was found. The phenomenon is also observed on anatase Ti\rm{O}_\rm{2}(101) [28]. But it is considerably different from the phenomenon observed on rutile Ti\rm{O}_\rm{2}(110) [19, 21, 22], rutile Ti\rm{O}_\rm{2}(100)-(2\times1) [29], and the reduced anatase Ti\rm{O}_\rm{2}(100)-(1\times4) surfaces [27]. On these surfaces [19, 21, 22, 27, 29], two C\rm{H}_\rm{2}O molecules can be coupled to form \rm{C}_\rm{2}$$\rm{H}_\rm{4}$, and surface $\rm{O}_\rm{v}$ sites are identified as the reactive sites for the formation of $\rm{C}_\rm{2}$$\rm{H}_\rm{4}. In our experiments, the rutile Ti\rm{O}_\rm{2}(100)-(1\times1) surface used is annealed in >4\times$$10^{-7}$ Torr $\rm{O}_\rm{2}$, and nearly no surface $\rm{O}_\rm{v}$ sites exist on the rutile Ti$\rm{O}_\rm{2}$(100)-(1$\times$1) surface after sample preparation. Therefore, it is reasonable that no thermal products are detected after C$\rm{H}_\rm{2}$O adsorption.

 FIG. 4 The rutile Ti$\rm{O}_2$(100)-(1$\times$1) surfaces were dosed with 0.52 ML of C$\rm{H}_\rm{2}$O at 120 K. Typical TPD spectra collected at $m/z$=15 (C$\rm{H}_3$$^+), 18 (\rm{H}_2$$\rm{O}^+$), 27 ($\rm{C}_2$$\rm{H}_3$$^+$), 28 ($\rm{C}_2$$\rm{H}_4$$^+$, C$\rm{O}^+$), 29 (HC$\rm{O}^+$), 30 (C$\rm{H}_2$$\rm{O}^+), 31(C\rm{H}_2OH^+), 32 (C\rm{H}_3OH^+), and 46 (HCOOH^+) following 0 min (red lines) and 20 min (blue lines) irradiation. B. The photocatalytic reactions of C\rm{H}_\rm{\textbf{2}}O on the rutile Ti\rm{O}_\rm{\textbf{2}}(100)-(1\times1) surface FIG. 5 shows TPD spectra acquired at mass-to-charge ratios (m/z) of 28 (C\rm{O}^{+}, \rm{C}_\rm{2}$$\rm{H}_\rm{4}$$^{+}), and 30 (C\rm{H}_\rm{2}$$\rm{O}^{+}$) after rutile Ti$\rm{O}_\rm{2}$(100)-(1$\times$1) surfaces were dosed with 0.52 ML of C$\rm{H}_\rm{2}$O and then irradiated by a laser at 355 nm for various durations. Before irradiation, the signal profiles of the 260 K peak and the 310 K peak from $m/z$=28 are exactly same as those from $m/z$=30, suggesting that both peaks are the results of dissociative ionization of desorbed parent C$\rm{H}_\rm{2}$O molecules in the electron-bombardment ionizer. As mentioned above, no other reaction products are detected, implying that this surface is thermally inactive for the reactions of C$\rm{H}_\rm{2}$O.

 FIG. 5 The rutile Ti$\rm{O}_2$(100)-(1$\times$1) surfaces were dosed with 0.52 ML of C$\rm{H}_2$O at 120 K. (a) Typical TPD spectra collected at $m/z$=30 (C$\rm{H}_2$$\rm{O}^+) following different laser irradiation times. (b) Typical TPD spectra collected at m/z=28 (\rm{C}_2$$\rm{H}_4$$^+, C\rm{O}^+) following different laser irradiation times. After UV irradiation, both C\rm{H}_\rm{2}O peaks at 260 and 310 K decrease monotonically as the laser irradiation time increases, suggesting that C\rm{H}_\rm{2}O molecules are either photo-desorbed or reacted to form other products. A TPD peak at \sim440 K becomes obvious in the TPD trace of m/z=30 after irradiation and keeps nearly unchanged with increasing irradiation time (FIG. 5(a)). Taking into account additional traces of m/z=29 (FIG. 4), this peak is also assigned to desorption of C\rm{H}_\rm{2}O. This result suggests that part of C\rm{H}_\rm{2}O molecules have become more strongly bound to the surface after irradiation. A similar phenomenon has been observed on rutile Ti\rm{O}_\rm{2}(110) [17], which may be due to the formation of {\eta}$$^{2}$-$\rm{CH}_\rm{2}$O after irradiation.

 $\rm{C}\rm{H}_\rm{2}\rm{O}_\rm{Ti5c}+\rm{O}_\rm{b}\rightarrow\rm{O}_\rm{b}-C\rm{H}_\rm{2}-\rm{O}_\rm{Ti5c}$ (1)

In our work, the 440 K peak is also likely due to the formation of ${\eta}$$^{2}-\rm{CH}_\rm{2}O. Based on previous work of aldehydes photochemistry on rutile Ti\rm{O}_\rm{2}(110) [35], the \rm{O}_\rm{Ti5c}-C\rm{H}_\rm{2}-\rm{O}_\rm{b} species is very photoactive, and can be easily decomposed into HCO\rm{O}^{-}. As a result, the amount of {\eta}$$^{2}$-$\rm{CH}_\rm{2}$O on the surface will not keep increasing with increasing irradiation time.

Concomitant to the decrease in the $\rm{CH}_2$O TPD peaks, the TPD signal for $m/z$=28 at around 580 K increases with increasing irradiation time (FIG. 5(b)). Considering the small adsorption energy of CO on the surface, the 580 K peak could only come from the thermal decomposition of other species. In order to determine the origin of this new feature, TPD traces were collected at a variety of $m/z$ ratios (FIG. 4). On the basis of the TPD results in FIG. 4, the $m/z$=28 signal at 580 K may come from three sources. Compared with previous results of HCOOH on rutile Ti$\rm{O}_2$(100) [36] and the cracking patterns of HCOOH observed in our mass spectrometer, the first two sources of $m/z$=28 signal at 580 K are HCOOH and HCO$\rm{O}^{-}$, both of which could be the products of $\rm{CH}_2$O photo-oxidation.

The large TPD signal seen in FIG. 5(b) indicates that HCO$\rm{O}^{-}$ may be an important photoinduced product. Compared to other products, the intensity of the 580 K peak is several times greater (see FIG. 4), strongly suggesting that HCO$\rm{O}^{-}$ is probably the major product of the photoinduced decomposition of $\rm{CH}_2$O on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface. The yields of CO (from HCO$\rm{O}^{-}$) and $\rm{CH}_2$O with increasing irradiation time are calculated and displayed in FIG. 6. About 0.19 ML of $\rm{CH}_\rm{2}$O is depleted after 20 min irradiation, whereas, only 0.023 ML of HCO$\rm{O}^{-}$ is produced, implying that the decrease of the $\rm{CH}_2$O signal is mainly the result of photoinduced desorption of $\rm{CH}_2$O during laser irradiation.

 FIG. 6 Yields of C$\rm{H}_2$O and CO as a function of laser irradiation time following the adsorption of 0.52 ML of C$\rm{H}_2$O on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surfaces at 120 K, derived from data in FIG. 5.

In order to produce HCO$\rm{O}^{-}$, the C atom of the $\rm{CH}_\rm{2}$O molecule must acquire a second O atom. According to previous work [17], without additional surface oxygen species, the second O atom is either from an $\rm{O}_\rm{b}$ row or from another adsorbed C$\rm{H}_2$O molecule. The direct photodissociation of $\rm{CH}_2$O to produce an O atom via the C=O bond cleavage is not possible because of the extremely low adsorption cross section of $\rm{CH}_2$O at 355 nm ($\sim$$10^{-20} c\rm{m}^{2}) [37]. Xu and coworkers [17] proposed that an O atom at the \rm{Ti}_\rm{5c} site (\rm{O}_\rm{Ti5c}) can be produced via the transfer of methylene from C\rm{H}_2O to an \rm{O}_b atom during UV irradiation through an intermediate adsorption structure consisting of dioxymethylene.  {\rm{O}_\rm{b}}-{\rm{CH}_2}-{\rm{O}_\rm{Ti5c}}\rightarrow{\rm{O}_\rm{b}}-{\rm{CH}_2}+\rm{O}_\rm{Ti5c} (2) Subsequently, the \rm{O}_\rm{Ti5c} atom may react with a nearby adsorbed \rm{CH}_2O molecule to form an \rm{O}_\rm{Ti5c}-\rm{CH}_2-\rm{O}_\rm{Ti5c} complex by heat or laser irradiation.  \rm{O}_\rm{Ti5c}+{\rm{CH}_\rm{2}}{\rm{O}_\rm{Ti5c}}\rightarrow{\rm{O}_\rm{Ti5c}}-{\rm{CH}_2}-\rm{O}_\rm{Ti5c} (3) Then the complex perhaps gives rise to HCO\rm{O}^{-} either by transferring an H atom to an \rm{O}_\rm{b} atom (H\rm{O}_\rm{b}), or ejecting an H atom to the vacuum.  {\rm{O}_\rm{Ti5c}}-{\rm{CH}_2}-{\rm{O}_\rm{Ti5c}}+{\rm{O}_\rm{b}}\rightarrow{\rm{HCOO}_\rm{Ti5c}}{^{-}}+\rm{HO}_\rm{b} (4)  {\rm{O}_\rm{Ti5c}}-{\rm{CH}_2}-{\rm{O}_\rm{Ti5c}}\rightarrow{\rm{HCOO}_\rm{Ti5c}}{^{-}}+\rm{H} (5) In addition, \rm{HCOO}_\rm{Ti5c}$$^{-}$ may combine with H$\rm{O}_\rm{b}$ to produce HCOOH.

 ${\rm{HCOO}_\rm{Ti5c}}{^{-}}+\rm{HO}_\rm{b}\rightarrow\rm{HCOOH}+\rm{O}_\rm{b}$ (6)

While, as shown in TPD trace of $m/z$=18 (FIG. 4), two main desorption features at 478 and 580 K are observed after irradiating the 0.52 ML C$\rm{H}_2$O covered rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface for 20 min. The 580 K peak may come from thermal decomposition of HCOOH and HCO$\rm{O}^{-}$ [36]. For the 478 K peak, no signals of higher $m/z$ ratios are detected at this temperature. Thus, this peak can be only due to the $\rm{H}_\rm{2}$O desorption. However, the desorption temperature is much higher than that of $\rm{H}_\rm{2}$O adsorbed at the $\rm{Ti}_\rm{5c}$ sites of rutile Ti$\rm{O}_2$(100)-(1$\times$1), and is similar to that of the recombinative desorption of $\rm{H}_2$O from hydroxyls on $\rm{O}_\rm{b}$ rows of rutile Ti$\rm{O}_\rm{2}$(110) [38]. Thus, the 478 K peak is also likely from recombinative desorption of $\rm{H}_2$O.

 $2\rm{HO}_\rm{b}\rightarrow\rm{H}_2O+\rm{O}_\rm{v}$ (7)

The formation of $\rm{H}_2$O via the reaction (7) verifies that the reaction channel involving a dioxymethylene intermediate is very possible to proceed on the surface, resulting in the formation of HCO$\rm{O}^{-}$ eventually.

The reaction channel consisting of a dioxymethylene intermediate is also supported by the observation of other products. As shown in FIG. 7(a), typical TPD spectra of $m/z$=27 ($\rm{C}_2$$\rm{H}_3$$^+$) were collected after irradiating the 0.52 ML C$\rm{H}_2$O covered rutile Ti$\rm{O}_2$(100)-(1$\times$1) surfaces for different time with a laser at 355 nm. Before irradiation, no signal is observed. As irradiation time increases, a peak at $\sim$580 K appears and increases in intensity. Based on previous works [17, 19, 21, 22], this peak is due to desorption of $\rm{C}_2$$\rm{H}_4 product, which is formed via the carbon-carbon coupling of two C\rm{H}_2O moleules. Clearly, \rm{C}_2$$\rm{H}_4$ product is another source for the $m/z$=28 signal at 580 K. This is very similar to the observation of C$\rm{H}_2$O photodecomposition on the rutile Ti$\rm{O}_2$(110) surface. On rutile Ti$\rm{O}_2$(110), the $\rm{C}_2$$\rm{H}_4 product arises from C\rm{H}_2O adsorbed at \rm{O}_\rm{v} sites to form a diolate (-OC\rm{H}_2C\rm{H}_2O-) species, which releases \rm{C}_2$$\rm{H}_\rm{4}$ at a relatively high temperature [21, 22]. But the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface used in our work contains no $\rm{O}_\rm{v}$ sites. Thus, the dioxymethylene species may act as the intermediate for the formation of $\rm{C}_2$$\rm{H}_4. After the transfer of methylene to \rm{O}_\rm{b} atoms, two \rm{O}_\rm{b}-C\rm{H}_2 groups may be coupled to form \rm{C}_2$$\rm{H}_4$.

 $2\rm{O}_\rm{b}-C\rm{H}_2\rightarrow\rm{C}_2\rm{H}_\rm{4}+2\rm{O}_\rm{b}$ (8)
 FIG. 7 The rutile Ti $\rm{O}_2$ (100)-(1 $\times$ 1) surfaces were dosed with 0.52 ML of C $\rm{H}_2$ O at 120 K. (A) Typical TPD spectra collected at $m/z$ =27 ( $\rm{C}_2$ $\rm{H}_3$ $^+$ ) following different laser irradiation times. (B) Typical TPD spectra collected at $m/z$ =15 (C $\rm{H}_3$ $^+$ ) following different laser irradiation time.

While, another desorption peak appears at 570 K in the TPD spectrum of $m/z$=15 (C$\rm{H}_3$$^+) after UV irradiation and increases with irradiation time (FIG. 7(b)). This peak is attributed to C\rm{H}_3 radical desorption, probably from \rm{O}_\rm{b} atoms. The formation of C\rm{H}_3$$\rm{O}_\rm{b}$ may also need the participation of $\rm{O}_\rm{b}$-C$\rm{H}_2$ groups.

 ${\rm{O}_\rm{b}-C\rm{H}_2}+\rm{HO}_\rm{b}\rightarrow\rm{O}_\rm{b}-C\rm{H}_3+\rm{O}_\rm{b}$ (9)
 $\rm{O}_\rm{b}-C\rm{H}_3\rightarrow \rm{CH}_3\cdot+\rm{O}_\rm{b}$ (10)

The appearance and increase of C$\rm{H}_3$$\cdot on \rm{O}_\rm{b} atoms further demonstrate that methylene groups can be transferred to \rm{O}_\rm{b} atoms during irradiation. Therefore, it can be concluded that the reaction channel involving the dioxymethylene intermediate is very likely to proceed on the rutile Ti\rm{O}_2(100)-(1\times1) surface. It is also noteworthy that the intensity of CO signal at 580 K is about 20 times bigger than that of \rm{C}_2$$\rm{H}_4$ signal, and about 40 times bigger than that of C$\rm{H}_3$$\cdot signal after 20 min irradiation (see FIG. 5 and FIG. 7). These results suggest that the dominant photocatalytic reaction is the formation of HCO\rm{O}^{-}. Meanwhile, at the first 1 min irradiation, nearly no \rm{C}_2$$\rm{H}_4$ and C$\rm{H}_3$$\cdot products are formed (FIG. 7), but the HCO\rm{O}^{-} product has been largely formed (bigger CO desorption signal in FIG. 5(b)). In other words, the formations of \rm{C}_2$$\rm{H}_\rm{4}$ and C$\rm{H}_3$$\cdot are not exactly coincident with the formation of HCO\rm{O}^{-} and the formation of HCO\rm{O}^{-} precedes the formations of \rm{C}_2$$\rm{H}_4$ and C$\rm{H}_3$$\cdot. As a result, the formation of HCO\rm{O}^{-} may not completely depend on the formation of the \rm{O}_\rm{Ti5c} atoms. On the contrary, the \rm{O}_\rm{b}-C\rm{H}_2-\rm{O}_\rm{Ti5c} intermediate is likely to decompose into HCO\rm{O}^{-} directly, transferring an H atom to the nearby \rm{O}_\rm{b} atom (\rm{HO}_\rm{b}).  \rm{O}_\rm{b}-C\rm{H}_2-\rm{O}_\rm{Ti5c}+\rm{O}_\rm{b}\rightarrow\rm{HCO}_\rm{b}\rm{O}_\rm{Ti5c}^-+\rm{HO}_\rm{b} (11) In addition to the increase in the m/z=28 signal at around 580 K, a new desorption peak appears at about 570 K in the TPD trace of m/z=31 (FIG. 8) and increases with increasing irradiation time. Taking into account additional TPD traces in FIG. 4, this peak is attributed to desorption of C\rm{H}_3OH. This phenomenon has been observed previously on rutile Ti\rm{O}_2(110) [19] and the reduced anatase Ti\rm{O}_2(100)-(1\times4) surface [27]. In Huang's experiment [19], they proposed that the occurrence of H-transfer between HCO\rm{O}^{-} and dioxymethylene led to the formation of methoxy species (C\rm{H}_3O). The disproportionation reaction between C\rm{H}_3O could lead to the formation of C\rm{H}_3OH and C\rm{H}_2O at elevated temperature. Previous results of C\rm{H}_2O on rutile Ti\rm{O}_2(001) also show that the coincident desorption of C\rm{H}_3OH and C\rm{H}_2O could occur at 370 and 550 K, respectively [16]. However, we did not observe the coincident desorption of C\rm{H}_3OH and C\rm{H}_2O at 570 K. Based on previous work on rutile Ti\rm{O}_2(110) [39], C\rm{H}_2O can recombine with \rm{HO}_\rm{b} to form C\rm{H}_\rm{3}OH again. On the reduced anatase Ti\rm{O}_2(100)-(1\times4) surface [27], Wang and coworkers also ascribed the formation of C\rm{H}_3OH to the reaction of C\rm{H}_2O with the produced H atoms during irradiation or during the heating process. Thus, on the rutile Ti\rm{O}_2(100)-(1\times1) surface, the formation of C\rm{H}_3OH may proceed as follows:  \rm{CH}_2\rm{O}_\rm{Ti5c}+\rm{HO}_\rm{b}\rightarrow\rm{O}_\rm{Ti5c}-C\rm{H}_3+\rm{O}_\rm{b} (12)  \rm{O}_\rm{Ti5c}-C\rm{H}_3+\rm{HO}_\rm{b}\rightarrow{\rm{CH}_3}\rm{OH} (13)  FIG. 8 Typical TPD spectra collected at m/z=31 (C\rm{H}_2OH^+) after the rutile Ti\rm{O}_2(100)-(1\times1) surfaces were dosed with 0.52 ML of C\rm{H}_2O following different laser irradiation time. While, during the TPD process, C\rm{H}_3$$\rm{O}_\rm{Ti5c}$ may also decompose into C$\rm{H}_3$$\cdot and \rm{O}_\rm{Ti5c}.  \rm{O}_\rm{Ti5c}-C\rm{H}_3\rightarrow\rm{CH}_3\cdot +\rm{O}_\rm{Ti5c} (14) Ⅳ. DISCUSSION Although C\rm{H}_2O is a very simple molecule, the photolysis of this molecule is very complicated. The products and reaction channels for C\rm{H}_2O decomposition on different Ti\rm{O}_2 surfaces [19, 21, 22, 27, 29] are very similar. For rutile Ti\rm{O}_2(100)-(1\times1) and anatase Ti\rm{O}_2(101) surfaces [28], however, no other thermal reaction products were detected during the TPD process due to the absence of \rm{O}_\rm{v} sites on the surfaces. This result is also an evidence that \rm{O}_\rm{v} sites are the reactive sites for the carbon-carbon coupling of C\rm{H}_2O to \rm{C}_2$$\rm{H}_4$. Among all these Ti$\rm{O}_2$ surfaces discussed above, the desorption temperature of C$\rm{H}_2$O is the highest on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface. That may be due to the fact that the densities of $\rm{Ti}_\rm{5c}$ and $\rm{O}_\rm{b}$ atoms on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface are much higher than those on the other Ti$\rm{O}_2$ surfaces. As a result, the carbonyl C atom of the C$\rm{H}_2$O molecule may interact with the $\rm{O}_\rm{b}$ atom more easily, which may result in the formation of a more stable adsorption configuration.

The rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface is thermally inactive for the reactions of C$\rm{H}_2$O, but it is photoactive for the reactions of C$\rm{H}_2$O. The main decomposition product is HCO$\rm{O}^{-}$, which is similar to the results obtained on other Ti$\rm{O}_2$ surfaces [17, 19, 27, 29]. However, the dominant reaction channels leading to HCO$\rm{O}^{-}$ formation may be different. On the rutile Ti$\rm{O}_2$(110) surface, the dominant reaction channel is consisted of the $\rm{O}_\rm{Ti5c}$ atoms formation via the transfer of methylene to the $\rm{O}_\rm{b}$ sites and the $\rm{O}_\rm{b}$-C$\rm{H}_\rm{2}-\rm{O}_\rm{Ti5c}$ species acts as the intermediate [17]. Whereas, on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface, the case is very different. As shown in FIG. 5 and FIG. 7, the intensity of CO signal at 580 K is about 20 times bigger than that of $\rm{C}_2$$\rm{H}_4 signal, and about 40 times bigger than that of C\rm{H}_3$$\cdot$ signal. These values are much bigger than the ratios of those produced from C$\rm{H}_2$O photoinduced decomposition on rutile Ti$\rm{O}_2$(110) [17]. As mentioned above, the formations of $\rm{C}_2$$\rm{H}_4 and C{\rm{H}_3}$$\cdot$ require the transfer of methylene to the $\rm{O}_\rm{b}$ sites forming $\rm{O}_\rm{b}$C$\rm{H}_\rm{2}$. After the transfer process, the $\rm{O}_\rm{Ti5c}$ atoms are produced simultaneously, which will take part in the following formation of HCO$\rm{O}^{-}$. Whereas, the formation of HCO$\rm{O}^{-}$ precedes the formation of $\rm{C}_\rm{2}$$\rm{H}_\rm{4} and C\rm{H}_3$$\cdot$ in this work. These results clearly indicate that $\rm{O}_\rm{Ti5c}$ is probably not necessary for the formation of HCO$\rm{O}^{-}$. Therefore, the dominant reaction channel on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface may be changed. The direct decomposition of $\rm{O}_\rm{b}$-C$\rm{H}_2-\rm{O}_\rm{Ti5c}$ into $\rm{HCO}_\rm{b}$$\rm{O}_\rm{Ti5c}$$^-$ and $\rm{HO}_\rm{b}$ becomes the main reaction channel, while the reaction of $\rm{O}_\rm{Ti5c}$ with C$\rm{H}_2$O adsorbed at $\rm{Ti}_\rm{5c}$ sites to produce HCO$\rm{O}^{-}$ becomes a minor reaction channel. Therefore, on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface, two possible reaction channels may lead to the formation of HCO$\rm{O}^{-}$ and the $\rm{O}_\rm{b}$-C$\rm{H}_2-\rm{O}_\rm{Ti5c}$ species maybe acts as the intermediate. The direct decomposition is considered as the dominant reaction channel and the lattice oxygen atom ($\rm{O}_\rm{b}$) is directly involved in the photoinduced decomposition of C$\rm{H}_2$O on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface.

Ⅴ. CONCLUSION

In this work, we have investigated the interactions of C$\rm{H}_\rm{2}$O with a rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface under UV irradiation. Experimental results show that the photodecomposition of C$\rm{H}_2$O on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface can occur easily under UV light irradiation. Before irradiation, only molecular desorption of C$\rm{H}_2$O can be detected during the TPD process. When irradiated by UV light, several photoinduced products are detected. HCO$\rm{O}^{-}$ is the major product, while $\rm{C}_2$$\rm{H}_4, C\rm{H}_3$$\cdot$, and C$\rm{H}_3$OH are minor products.

Our TPD investigation demonstrates that $\rm{O}_\rm{b}$ atoms play a very important role in the photoinduced decomposition of C$\rm{H}_2$O on the rutile Ti$\rm{O}_2$(100)-(1$\times$1) surface through an initial $\rm{O}_\rm{b}$-C$\rm{H}_2-\rm{O}_\rm{Ti5c}$ intermediate structure. Clear mechanisms have been delineated for the participation of lattice ($\rm{O}_\rm{b}$) atoms in the decomposition pathways, including their presence in one type of HCO$\rm{O}^{-}$ product. Our results are supposed to broaden the fundamental understanding of C$\rm{H}_2$O photochemistry on Ti$\rm{O}_\rm{2}$ surfaces.

Ⅵ. ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (No.21673235 and No.21403224), and the Youth Innovation Promotion Association CAS, and the Key Research Program of the Chinese Academy of Sciences.

Reference
 [1] A. L. Linsebigler, G. Q. Lu, and J. T. Yates, Chem. Rev. 95 , 735 (1995). DOI:10.1021/cr00035a013 [2] U. Diebold, Surf. Sci. Rep. 48 , 53 (2003). DOI:10.1016/S0167-5729(02)00100-0 [3] A. Fujishima, X. Zhang, and D. Tryk, Surf. Sci. Rep. 63 , 515 (2008). DOI:10.1016/j.surfrep.2008.10.001 [4] M. A. Henderson, Surf. Sci. Rep. 66 , 185 (2011). DOI:10.1016/j.surfrep.2011.01.001 [5] Q. Guo, C. Y. Zhou, Z. B. Ma, Z. F. Ren, H. J. Fan, and X. M. Yang, Chem. Soc. Rev. 45 , 3701 (2016). DOI:10.1039/C5CS00448A [6] K. Klier, Adv. Catal. 31 , 243 (1982). [7] M. Kurtz, J. Strunk, O. Hinrichsen, M. Muhler, K. Fink, B. Meyer, and C. Wöll, Angew. Chem. Int. Ed. 44 , 2790 (2005). DOI:10.1002/anie.200462374 [8] M. Behrens, F. Studt, I. Kasatkin, S. Kuhl, M. Havecker, F. Abild-Pedersen, S. Zander, F. Girgsdies, P. Kurr, B. L. Kniep, M. Tovar, R. W. Fischer, J. K. Norskov, and R. Schlogl, Science 336 , 893 (2012). DOI:10.1126/science.1219831 [9] M. F. Camellone, J. Zhao, L. Jin, Y. Wang, M. Muhler, and D. Marx, Angew. Chem. Int. Ed. 52 , 5780 (2013). DOI:10.1002/anie.201301868 [10] Q. Guo, C. B. Xu, Z. F. Ren, W. S. Yang, Z. B. Ma, D. X. Dai, H. J. Fan, T. K. Minton, and X. M. Yang, J. Am. Chem. Soc. 134 , 13366 (2012). DOI:10.1021/ja304049x [11] D. Wei, X. C. Jin, C. Q. Huang, D. X. Dai, Z. B. Ma, W. X. Li, and X. M. Yang, J. Phys. Chem. C 119 , 17748 (2015). DOI:10.1021/acs.jpcc.5b05074 [12] M. M. Shen, and M. A. Henderson, J. Phys. Chem. Lett. 2 , 2707 (2011). DOI:10.1021/jz201242k [13] X. C. Mao, Z. Q. Wang, X. Lang, Q. Q. Hao, B. Wen, D. X. Dai, C. Y. Zhou, L. M. Liu, and X. M. Yang, J. Phys. Chem. C 119 , 6121 (2015). DOI:10.1021/acs.jpcc.5b00503 [14] P. Biloen, and W. M. Sachtler, Adv. Catal. 30 , 165 (1981). [15] J. Zhu, J. Catal. 225 , 388 (2004). DOI:10.1016/j.jcat.2004.04.015 [16] H. Idriss, K. S. Kim, and M. A. Barteau, Surf. Sci. 262 , 113 (1992). DOI:10.1016/0039-6028(92)90464-H [17] C. B. Xu, W. S. Yang, Q. Guo, D. X. Dai, T. K. Minton, and X. M. Yang, J. Phys. Chem. Lett. 4 , 2668 (2013). DOI:10.1021/jz401349q [18] T. Cremer, S. C. Jensen, and C. M. Friend, J. Phys. Chem. C 118 , 29242 (2014). DOI:10.1021/jp5053908 [19] Q. Yuan, Z. F. Wu, Y. K. Jin, F. Xiong, and W. X. Huang, J. Phys. Chem. C 118 , 20420 (2014). DOI:10.1021/jp5061733 [20] Z. R. Zhang, M. R. Tang, Z. T. Wang, Z. Ke, Y. B. Xia, K. T. Park, I. Lyubinetsky, Z. Dohnlek, and Q. F. Ge, Top. Catal. 58 , 103 (2014). [21] K. Zhu, Y. B. Xia, M. R. Tang, Z. T. Wang, B. Jan, I. Lyubinetsky, Q. F. Ge, Z. Dohnálek, K. T. Park, and Z. R. Zhang, J. Phys. Chem. C 119 , 14267 (2015). DOI:10.1021/acs.jpcc.5b04781 [22] K. Zhu, Y. B. Xia, M. R. Tang, Z. T. Wang, I. Lyubinetsky, Q. F. Ge, Z. Dohn, á lek, K. T. Park, and Z. R. Zhang, J. Phys. Chem. C 119 , 18452 (2015). DOI:10.1021/acs.jpcc.5b05639 [23] H. Feng, L. M. Liu, S. H. Dong, X. F. Cui, J. Zhao, and B. Wang, J. Phys. Chem. C 120 , 24287 (2016). DOI:10.1021/acs.jpcc.6b08797 [24] L. M. Liu, and J. Zhao, Surf. Sci. 652 , 156 (2016). DOI:10.1016/j.susc.2015.12.036 [25] X. J. Yu, Z. R. Zhang, C. W. Yang, F. Bebensee, S. Heissler, A. Nefedov, M. R. Tang, Q. F. Ge, L. Chen, B. D. Kay, Z. Dohn, á lek, Y. M. Wang, and C. Wöll, J. Phys. Chem. C 120 , 12626 (2016). DOI:10.1021/acs.jpcc.6b03689 [26] D. W. Guan, R. M. Wang, X. C. Jin, D. X. Dai, Z. B. Ma, H. J. Fan, and X. M. Yang, Chin. J. Chem. Phys. 30 , 253 (2017). DOI:10.1063/1674-0068/30/cjcp1703030 [27] B. Luo, H. Q. Tang, Z. W. Cheng, Y. Y. Ji, X. F. Cui, Y. L. Shi, and B. Wang, J. Phys. Chem. C 121 , 17289 (2017). DOI:10.1021/acs.jpcc.7b04530 [28] M. Setvin, J. Hulva, H. H. Wang, T. Simschitz, M. Schmid, G. S. Parkinson, C. Di Valentin, A. Selloni, and U. Diebold, J. Phys. Chem. C 121 , 8914 (2017). DOI:10.1021/acs.jpcc.7b01434 [29] Z. M. Wang, F. Xiong, Z. Zhang, G. H. Sun, H. Xu, P. Chai, and W. X. Huang, J. Phys. Chem. C 121 , 25921 (2017). [30] H. Qiu, H. Idriss, Y. Wang, and C. Wö ll, J. Phys. Chem. C 112 , 9828 (2008). DOI:10.1021/jp801327b [31] Q. Yuan, Z. F. Wu, Y. K. Jin, L. Xu, F. Xiong, Y. Ma, and W. X. Huang, J. Am. Chem. Soc. 135 , 5212 (2013). DOI:10.1021/ja400978r [32] M. A. Henderson, Surf. Sci. 319 , 315 (1994). DOI:10.1016/0039-6028(94)90598-3 [33] M. A. Henderson, Langmuir 12 , 5093 (1996). DOI:10.1021/la960360t [34] T. L. Thompson, O. Diwald, and J. T. Yates, J. Phys. Chem. B 107 , 11700 (2003). DOI:10.1021/jp030430m [35] M. A. Henderson, N. A. Deskins, R. T. Zehr, and M. Dupuis, J. Catal. 279 , 205 (2011). DOI:10.1016/j.jcat.2011.01.021 [36] M. A. Henderson, J. Phys. Chem. 99 , 15253 (1995). DOI:10.1021/j100041a048 [37] R. Atkinson, D. L. Baulch, R. A. Cox, J. N. Crowley, R. F. Hampson, R. G. G. Hynes, M. E. Jenkin, M. J. Rossi, and J. Troe, Atmos. Chem. Phys. 6 , 3625 (2006). DOI:10.5194/acp-6-3625-2006 [38] R. T. Zehr, and M. A. Henderson, Surf. Sci. 602 , 1507 (2008). DOI:10.1016/j.susc.2008.02.031 [39] X. C. Mao, D. Wei, Z. Q. Wang, X. C. Jin, Q. Q. Hao, Z. F. Ren, D. X. Dai, Z. B. Ma, C. Y. Zhou, and X. M. Yang, J. Phys. Chem. C 119 , 1170 (2014).

a. 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, 大连 116023;
b. 中国科学院大学, 北京 100049;
c. 上海科技大学, 物质科学与技术学院, 上海 201210