Ultrahigh Photocatalytic CO2 Reduction Efficiency by Single Metallic Atom Oxide on TiO2


 Photocatalytic carbon dioxide (CO2) reduction is a sustainable and energy-consumption-free route to directly convert the greenhouse gas into chemicals. Given the vast amount of greenhouse gases, numerous efforts have been devoted to developing inorganic photocatalysts due to their stable, low-cost and environmental-friendly properties. However, more efficient titanium dioxide (TiO2) without noble metal or sacrifice/organic agent is highly desirable for CO2 reduction practical application, and it is also difficult and urgently in demand for TiO2 producing selectively valuable compounds, i.e. industrial chemicals and fuels. Here, we develop a novel “adatom at step” strategy via anchoring single tungsten atom oxide (STAO) site on intrinsic steps of classic TiO2 nanoparticles. The composition of single-sites can be controlled by tuning the ratio of adatom W5+ to neighboring Ti3+, resulting in significant CO2 reduction efficiency and selectively yield of carbon monoxide (CO) or methane (CH4) as main products. The W5+-dominated catalysts can achieve an ultrahigh photocatalytic CH4 production of 59.3 μmol/g/h, while the Ti3+-dominated catalysts can achieve a CO production of 181.4 μmol/g/h, which both exceed those of pristine TiO2 by more than one order of magnitude. The mechanism relies on the accurate control of atomic sites with high coverage and the subsequent excellent electron-hole separation along with favorable adsorption-desorption of intermediates on sites. This approach not only provides a novel strategy for inorganic catalytic single-sites with superior performance, but also identifies the rational design mechanisms of the efficient site with controllable production.


Introduction
Carbon fuel consumption and CO 2 greenhouse gas emissions are imperative for the sustainable development of human civilization, causing acceleration of energy shortages and global warming [1][2][3][4][5].
The mechanism of catalytically active sites is highly important to promote photocatalytic CO 2 reduction activity and particularly product selectivity. In the process of photocatalytic CO 2 reduction on TiO 2 , the electrons are generated by light irradiation and then transferred to the catalytically active sites to react with the adsorbed CO 2 molecules. Consequently, there are three main aspects to improving TiO 2 -based catalytic e ciency, namely including (1) intrinsic light absorption, (2) kinetic photocarrier transfer pathway and (3) intermediates adsorption-desorption and hydrogenation. Tremendous efforts have been made to optimize photocarrier separation and charge transfer pathway on TiO 2 . Recently for instance, a TiO 2 -graphene composite structure was found to improve high surface area for CO 2 adsorption and increase the anatase-TiO 2 light absorption region, resulting in increasing CO 2 reduction e ciency. [28] Surface vacancy-mediated Ti 3+ in TiO 2 can e ciently accelerate the adsorption and chemical activation of CO 2 . [29] In general, synergistic effect of these three aspects is important and stringent. Besides, these aspects also in uence each other. Ag/TiO 2 shows a strong interfacial coupling leading to an e cient separation of photogenerated carriers and a consequent enhanced selectivity between CH 4 and CH 3 OH. [30] Additionally, note that the hydrogenation of CO 2 plays a vital role in the design of thermocatalysts [16], rational design of active sites on photocatalysts are needed to further explored to accelerate CO 2 reduction. However, these inorganic TiO 2 -based photocatalytic performance are still not su cient for the practical applications. Meanwhile, owing to the lack of in-depth understanding of photocatalytic active sites at the atomic scale, most of the TiO 2 catalysts still suffer from non-controllable products (a C 1 mixture containing CO, CH 4 and CH 3 OH).
With the blooming development of single-atom catalysts and accurate atomic site con guration on catalysts, numerous active sites with controllable products have been investigated and gradually recognized by creating single-sites or introducing co-catalysts or dimer reactive sites [47][48][49][50][51][52][53][54][55][56][57][58][59], e. g. in thermo-and electro-catalytic active sites rational design, both isolated sites and the proper atomic con guration of supports are of great importance for CO 2 reduction activity, stability and selectivity [52].
However, as the most common intrinsic defects on TiO 2 surfaces and probably the predominant ones on TiO 2 nanoparticles [81,82], atomic steps on TiO 2 surface has never been reported as adatom substrate for single-sites anchoring. Therefore, it is imperative to explore single-site anchoring strategy at the atomic TiO 2 steps to realize high e cient CO 2 reduction via further optimizing the photocarrier separation and reaction pathway.
Herein, we develop single-tungsten-atom oxide (STAO) on TiO 2 (101) terrace edges, i.e. an oxygencoordinated tungsten atom (W 5+ ) anchored at TiO 2 atomic steps. The facile method of "adatom at step" can not only create novel active sites via anchoring single-atoms at steps, but also cause neighboring Ti 3+ formation. The rational design strategy of "adatom at step" is schematically illustrated in Figure 1a.
The densities of single W 5+ and Ti 3+ sites are optimized to achieve W 5+ -and Ti 3+ -dominant catalysts, yielding superior CH 4 and CO production with high selectivity, respectively. The controlled CH 4 production e ciency is highest one compared with other photocatalysts without noble-metal. Due to the variation of adatom elements and oxygen coordination, there exists a larger number of "adatom at step" sites for rational design and new mechanisms with exceptional catalytic performance could be uncovered.

Results And Discussions
Structural characterization of STAO at TiO 2 atomic steps The atomic morphology of STAO at step sites is illustrated via Cs-corrected scanning transmission electron microscopy high-angle annular dark eld (STEM-HAADF), indicating the STAO bright spot uniformly dispersed on TiO 2 nanoparticles (Fig. 1b). The elemental mapping analysis con rms that tungsten (W), titanium (Ti) and oxygen (O) are evenly dispersed (Fig. 1c). Moreover, Fig. 1b clearly demonstrates that W atoms are anchored above the vertical centers of Ti atoms, suggesting STAO sites are deposited at the steps with certain distortion as demonstrated in Fig. 1a. Since monoatomic-height steps at terraces constitute the most common defects on TiO 2 nanoparticle surface, chemical adsorption or new sites nucleation are more accessible at steps on metal oxide surfaces [76][77][78][79][80][81][82]. As shown in Figure  S1, the step of TiO 2 nanoparticles (approximately 5-10 nm) are identi ed mainly on (101) terrace in HAADF results. Furthermore, steps can be identi ed as mono-or bi-atomic steps at the edge of (101) terrace based on statistic of steps density in HAADF images (Fig. S1).
Since the amount of steps is su cient on TiO 2 surface, the density of the anchored STAO at steps can be controlled via depositing various STAO concentration, determined by inductively coupled plasma optical emission spectrometry (ICP-OES). It is found that the maximum tungsten density on TiO 2 can reach 3 wt% (Table S1). When considering the anchored STAO as a quasi-spherical model, the maximum number of W atoms covering on TiO 2 (101) surface is calculated as 8% approximately. Images of low-coverage (0.1%) and high-coverage (8%) STAO sites anchored on TiO 2 are presented in Figure S2.
To con rm the valence state of STAO sites at TiO 2 steps and their in uence on TiO 2 substrate, X-ray photoelectron spectroscopy (XPS) characterization was performed. Firstly for the uniformly dispersed tungsten, main peaks appearing as shoulders at 34.7 eV (W 4f 7/2) and 36.7 eV (W 4f 5/2) can be identi ed to the W 5+ doublet, while the peaks at 35.8 eV (W 4f 7/2) and 37.8 eV (W 4f 5/2) are due to the W 6+ doublet ( Fig. 1d) [63]. It can be noticed that the majority of W on TiO 2 surface exhibit 5+ valence state. Besides, Ti 2p banding energy (Fig. 1e) Fig. S3b. To be speci c, the densities of Ti 3+ are increasing from 0.76% to 9.5%, corresponding to 0% to 8%-coverage STAO ( Fig. S3c-d). In addition, Fig. S3f shows that the peak at 533.5 eV is ascribed to the Meanwhile, electron paramagnetic resonance (EPR) was applied to con rm the appearance of Ti 3+ . Fig.  1g shows that a strong EPR peak is observed at a g value of 1.95, which can be ascribed to the unpaired electrons induced by surface oxygen-vacancy mediated Ti 3+ . Moreover, as the STAO anchoring density increases, the Ti 3+ -O site EPR peak area also increases accordingly (Fig. S3e). Thus, EPR results further con rm that adatom at steps can induce Ti 3+ formation, and the ratio between STAO and Ti 3+ density can be well controlled. Hence, in the following section, Ti 3+ -dominant catalysts are used to describe the 0.1%-coverage STAO catalysts, and the catalysts with 8%-coverage STAO are regarded as W 5+ -dominant catalysts and labelled as STAO/TiO 2 for abbreviation.
To analyze the local environment of STAO more thoroughly, the STAO/TiO 2 are further examined by X-ray absorption near-edge structure (XANES) and extended X-ray absorption ne structure (EXAFS). Firstly, Fourier-Transformed (FT) EXAFS spectra (Figure 2a) shows one main peak at 1.6 Å of W-O in contrast to a W-W coordination peak at 2.6 Å shown in W foil, con rming the existence of single-atom W sites, consistent with HAADF and XRD ( Figure S4). The patterns of various density STAO (0.05, 0.5, 0.5, 1, 2 and 3 wt%)-anchored TiO 2 are not different from that of pristine anatase-TiO 2 , suggesting the STAO at step site has no periodic structure. However, when further increasing the amount of anchoring STAO, either a cluster or string structure ( Figure S5) appears along Ti stripe, reaching a limitation with 8% coverage (3 wt%) of STAO. Moreover, the W-O bonding (1.6 Å) state of the atomically dispersed STAO at TiO 2 steps is different from the bonding state of the W-O bonds in WO 3 (1.3 Å), suggesting that W-O bonds of STAO at step sites are slightly distorted and stretched by surface Ti-O bonds. Additionally, the EXAFS tting curve (Fig. 2c) con rms that the isolated W atom only coordinates with four oxygen atoms. As listed in Table 1, W-O bonds on STAO site are grouped into two types: longer one of 2.19 Å and shorter one of 1.81 Å. The XANES spectra are analyzed on the absorption of W L III -edge (Fig. 2b), where the white-line peak of STAO is located between the WO 2 and WO 3 , revealing that the W 5+ mainly presents. Combined with XPS results, the tetrahedron oxygen-coordinated W 5+ -O 4 sites are anchored at steps as schematically shown in Fig.   2d. Steps play a key role in catalysis, which could be further improved by anchoring STAO at the steps. A gassolid interfacial CO 2 photoreduction was applied to evaluate the photocatalytic e ciency of the STAOanchored anatase-TiO 2 photocatalysts. The photoreduction proceeded under mild conditions without any photosensitizer or sacri cial agents. Figure 3 shows CO 2 conversion products as CH 4 (Fig. 3a) and CO ( Fig. 3b)  photocatalysts, which exceeds those of pristine TiO 2 by more than one orders of magnitude. Notably, the selectivity is increased to 87.6% and 380 nm-induced apparent quantum yields on sites are 0.36 %, which is unprecedent in TiO 2 catalysts and comparative to the reported homogenous catalysts [39]. Meanwhile, for Ti 3+ -dominant catalysts, it can achieve 181.4 μmol g -1 h -1 (87.8% electron selectivity) with CO as main product (CO-max). Besides, Fig. 3c further demonstrates that a growth in CH 4 production along with a decline in CO production occurs with increasing STAO densities, suggesting that the STAO sites result in the utilization of C 1 mass and electrons for CH 4 production. The CH 4 e ciency is highest one in TiO 2based photocatalysts without noble-metal. Furthermore, the relationship between the CH 4 product and STAO sites density is linear from 0.5-3 wt% as shown in Fig. 3c-3d. Therefore, through tuning the sites density, the photocatalysts can exhibit controllable two main products of CH 4 and CO, which are both important chemicals in practical industry. Additionally, 5-times cycling utilization was performed to demonstrate its remarkable stability as the photocatalysts with superior e ciency ( Figure S6). To further illustrate its stability, the STAO at TiO 2 step sites after a long reaction were also examined by HAADF and XPS analyses ( Figure S7). We also assessed reference samples without CO 2 gas and without H 2 O gas for comparison ( Figure S8), showing that the CO 2 reduction and H 2 O splitting occur on the catalytic sites without any sensitizers or sacri ces. Although little extra surfactants are bene t for Ti 3+ appearance and might be involved in electron-hole separation [41,83], the little residue of that has not much in uence on CO or CH 4 reproductively generation in experiments. To illuminate the mechanism for the remarkably high e ciency of STAO at step sites, the photoelectric properties and kinetic pathway of photocarriers were systematically investigated. Photoluminescence (PL) quenching was conducted to reveal the separation of electron-hole pairs, as shown in Figure 4a.
Signi cant quenching of PL is observed in catalysts with the STAO at step sites, indicating that the recombination of electron-hole pairs is effectively suppressed upon STAO anchoring. The electronmigration pathway is veri ed as transferring from TiO 2 substrate to single-site due to the linear relationship between sites density and PL quenching degree as shown in inserted Fig. 4a. Furthermore, the photocurrent density with irradiation time increases as the density of STAO increases, further con rming the photoelectron separation and migration enhancement (Fig. 4b). Similarly, the charge migration improvement was also demonstrated by electrochemical impedance spectroscopy (EIS) analysis. Nyquist plots (Fig. 4c) show that the more STAO sites anchoring, the smaller of the impedance radii, indicating that the introduction of STAO sites greatly reduces the charge transfer resistance on catalysts.
A kinetic investigation of transient photoelectron's behavior on STAO sites was also accomplished. Transient orescence decay curves of various samples were collected ( Figure S9). Analysis of the curves with re-convolution tting manifests that the decays of each sample follows a bi-exponential model with t 1 and t 2 ( Fig. 4d and Table S3). The pristine TiO 2 exhibits a short lifetime of less than 200 ps due to the fast recombination of electron-hole. In contrast, "adatom at step" sites on W 5+ -dominant catalysts demonstrates t 1 =784 ps and t 2 =8576 ps, especially the t 2 enhancement from 31ps to 8576 ps is over 2 magnitudes, indicating the signi cant enhancement of electron-trapping on surface by W 5+ sites. Even in the Ti 3+ -dominant catalysts, the lifetime t 2 can be extended to 3000 ps obviously. Thus, STAO at step sites lead to an increased trapped-electron lifetime and adjust it in a large degree. The t 2 photoelectron lifetime (photoelectron on surface) is enhanced more obviously than that of t 1 (photoelectron in deep band) as presented in Table S4. The noteworthy increased t 2 lifetime via adatom at steps result from charge-transfer and charge-trapping at the surface sites, which could be bene cial for multi-electrons reactions. Besides, it is proven that further increasing STAO density (i.e. 4.5 wt%) would cause cluster formation, resulting in the decline of PL quenching ( Figure S10). It indicates that STAO at step play a vital role on charge-transfer and charge-trapping, which might be related with unique electronic states of STAO [63].
(2) CO 2 /CO adsorption-desorption and Gibbs free energy of intermediate To reveal the mechanism of products selectivity in STAO at TiO 2 steps, the molecular adsorptiondesorption was carried out, mainly considering CO and CO 2 . We calculated that the binding energy of CO 2 and CO adsorbed on Ti 3+ , Ti 4+ and W 5+ atom as shown in Table 2. For CO 2 adsorption, it indicates that W 5+ and Ti 3+ sites from STAO anchoring have stronger CO 2 binding energy than that of Ti 4+ site from pristine TiO 2 . The improved CO 2 adsorption on STAO at steps is further experimentally exhibited via temperature programmed desorption ( Figure S11), which show that the amount of CO 2 adsorption molecule is distinctly increased on STAO at steps (W 5+ and Ti 3+ ). Furthermore, via the in-situ diffuse re ectance infrared Fourier transform spectroscopy (DRIFTS), CO 2 and intermediates are vividly detected at the wavenumber from 1400-1800 cm -1 (Figure 5a and Figure S12). For CO adsorption, W 5+ sites have stronger CO binding energy than other sites, which is also consistent with DRIFTS results, where CO adsorption signal is clearly observed in Figure 5a. It demonstrates absorbed CO can be observed at 2065 and 2080 cm -1 , which decreases on Ti 3+ dominant catalysts and nearly absent on pristine TiO 2 (Ti 4+ ).

Gibbs free energy (ΔG) calculations were applied where all the presented intermediate states have been
optimized after considering several most possibilities pathways with the lowest energies. (Figure S13a).
Therefore, Ti 3+ and W 5+ sites are bene cial for CO production, and W 5+ sites are favorable for CH 4 production. Step by step, when one more H atom approaches to the adsorbed CO* and OH*, the formation of the rst H 2 O molecule is more desirable, leaving the CO* (step IV). The subsequent process is the CO reduction to CH 4 . One interesting phenomenon is the exchange of the adsorbed terminal point from C in OHC* (step V) to O in OHHC* (step VI). This process is quite important for the production of CH 4 because the exposed C atom could bind to maximum approaching H atoms readily. The intermediate OHHHC* (step VII) and O* + CH 4 (step VIII) just correspond to more and more H atoms approaching to the exposed C atom, extremely facilitating to form the nal production of CH 4 . As shown in Fig. 5b, the whole reaction from CO 2 to CH 4 is exothermic on STAO sites except for the formation of H 2 O, which possibly limiting the whole kinetics of CH 4 production. However, the endothermic feature from H 2 O formation can be easily overcome when involved with photons (step IX-X). Here, hole participation (simulating photon [85]) is intentionally involved, making the whole reaction process closed loop.
Interestingly, during the photocatalytic reduction reaction, rare amount of H 2 evolution is detected on W 5+dominant catalysts, indicating signi cant hydronation occurs on W 5+ sites. Moreover, Fig. 5a also show that the wavenumber at 1268, 1313, 2856 and 2900 cm -1 ascribed as the stretching of C-H bonds increase from 10 to 60 mins, indicating C-H formation are gradually improved on the W 5+ -dominant sites.
For Ti 3+ -dominant catalysts, as revealed in Figure S13b, CO hydronation on Ti 3+ has an uphill of 0.1 eV rather than -0.5 eV on W 5+ sites. Consistently, there is a ~1 eV barrier for the reaction from CO to CHO intermediate on Ti 3+ sites according to previous literatures [27,29,34], suggesting that CO desorption is more energy-favorable on Ti 3+ . Therefore, Ti 3+ sites are bene cial for CO 2 adsorption and CO releasing, while the speci c sites of W 5+ kinetically accelerates the CH 4 production without releasing CO in the reaction process.
In photocatalytic CO 2 reduction, the photogenerated charges are produced based on semiconductor photo-electric effect. Band diagram of STAO at TiO 2 steps are explored to con rm that the CO 2 reduction on STAO at step sites is thermodynamically allowed. Figure S14a-d shows that both the band edges of high-density STAO anchored catalysts downshift to lower energy level compared with low-density STAO band structure. Such a shift is considered to be more favorable for CO 2 reduction. The STAO-induced interfacial band structure near the Fermi level (E f ) was further veri ed by the at band potentials determined by the Mott-Schottky plots obtained through EIS measurements (Fig. S14e). It is noted that STAO at step sites can tune the electrochemical redox potential, leaving the photoexcited electrons energy level are still within CO and CH 4 redox potentials. Additionally, Figure S15 shows surface area of the whole sites relatively decrease as W 5+ induction, indicating total amount of sites is not vital for the enhanced e ciency. The improved e ciency and products selectivity could be ascribed to the adatom sites where the introduction of STAO at steps yielding the formation of W 5+ and Ti 3+ sites, where have (1) enhanced electron-hole separations; (2) optimized molecular adsorption-desorption; (3) Gibbs free energyfavorable pathway.
In summary, for the rst time, single-tungsten-atom-oxides were successfully anchored on steps of anatase-TiO 2 surface. The "adatom at step" sites achieve enhanced e ciency with selective products in the CO 2 reduction reaction. Upon the rational design, main products of CH 4 and CO are realized on W 5+dominant catalysts and Ti 3+ -dominant catalysts, respectively. The ultrahigh CH 4 production e ciency with 59.3 mmol/g/h is attained in STAO/TiO 2 (8% coverage). This superior photocatalytic performance is resulted from the enhanced photocarrier separation and adsorption-desorption as well as satisfactory intermediates. This study elaborates the "adatom at step" is an effective strategy in photocatalysts design and provides a facile approach by anchoring single-sites at substrate steps. Such method could be extended to other catalytic sites to reach the optimized e ciency and excellent electron selectivity and provide an effective way to develop novel materials and eventually new physics and chemistry. Photocatalyst characterization. HR-TEM, scanning transmission electron microscopy (STEM)-HAADF and energy-dispersive X-ray spectroscopy (EDX) elemental mapping characterization was performed by a Titan Themis G2 transmission electron microscope operated at 300 kV and equipped with a probe spherical aberration corrector. XRD data were acquired via a Bruker D8-Advance X-ray powder diffractometer with Cu Kα radiation (λ=1.5406 Å). ICP-OES measurements were conducted on an Agilent ICP-OES-730 analyser. The binding energies obtained in the XPS spectral analysis were corrected for specimen charging by referencing C 1s to 284.8 eV. The ESR spectra were measured on Bruker A300-10/12 ESR spectrometer. UV-Vis spectra were acquired with a Hitachi U-3900 UV-Vis spectrophotometer. PL spectroscopy was performed on a Hitachi F-7000 uorescence spectrometer with an excitation wavelength of 290 nm. The electrochemical measurements were performed in a conventional threeelectrode cell on a CHI-760E electrochemical workstation (Shanghai Chenhua Instrument Co., Ltd, China).

Methods
During the photocurrent measurement, an Ag/AgCl electrode was used as the reference electrode and a Pt nanosheets electrode acted as the counter electrode. The working electrodes were designed using resulting samples covered on the surface of uoride tin oxide (FTO) conductor glass. A quartz cell lled with 0.5 M Na 2 SO 4 (pH = 6.8) electrolyte was used as the measure system. For electrochemical impedance spectroscopy (EIS) measurements, the amplitude of the sinusoidal wave was 5 mV, and the frequency range from 100 kHz to 0.1 Hz. The W L III -edge X-ray absorption ne structure (XAFS) spectra of the sample and standards were collected at the beamline BL14W1 of the Shanghai Synchrotron Radiation Facility (SSRF). The white light was monochromatized by a Si (111) double-crystal monochromator and calibrated with W foil. The XAFS spectra of sample and standards were recorded in uorescence and transmission mode, respectively.   Catalytic performance of STAO anchored catalysts in the CO2 reduction reaction. Time-dependent production of a CH4 and b CO evolution in photocatalytic CO2 reduction by catalysts with various STAO density. c, Average photocatalytic e ciency of catalysts with different STAO sites density during 12 h of irradiation. The dashed line in the inset is the electron selectivity of CH4 and CO in the various catalysts. d, Electron reacted utilization rate of total, CO and CH4 along with the increase of anchored STAO.