Tuning Oxygen Vacancies in Ultrathin TiO2 Nanosheets to Boost Photocatalytic Nitrogen Fixation up to 700 nm

Yunxuan Zhao, Yufei Zhao, Run Shi, Bin Wang, Geoffrey I. N. Waterhouse, Li-Zhu Wu, Chen-Ho Tung, and Tierui Zhang*

Dinitrogen reduction to ammonia using transition metal catalysts is central to both the chemical industry and the Earth’s nitrogen cycle. In the Haber–Bosch process, a metallic iron catalyst and high temperatures (400 °C) and pressures (200 atm) are necessary to activate and cleave NN bonds, motivating the search for alternative catalysts that can transform N2 to NH3 under far milder reaction conditions. Here, the successful hydrothermal synthesis of ultrathin TiO2 nanosheets with an abundance of oxygen vacancies and intrinsic compressive strain, achieved through a facile copper-doping strategy, is reported. These defect-rich ultrathin anatase nanosheets exhibit remarkable and stable performance for photocatalytic reduction of N2 to NH3 in water, exhibiting photoactivity up to 700 nm. The oxygen vacancies and strain effect allow strong chemisorption and activation of molecular N2 and water, resulting in unusually high rates of NH3 evolution under visible-light irradiation. Therefore, this study offers a promising and sustainable route for the fixation of atmospheric N2 using solar energy.

Y. X. Zhao, Dr. R. Shi, Prof. L.-Z. Wu, Prof. C.-H. Tung, Prof. T. Zhang Key Laboratory of Photochemical Conversion and Optoelectronic Materials Technical Institute of Physics and Chemistry Chinese Academy of Sciences Beijing 100190, P. R. China E-mail: tierui@mail.ipc.ac.cn Y. X. Zhao, Prof. T. Zhang Center of Materials Science and Optoelectronics Engineering University of Chinese Academy of Sciences Beijing 100190, P. R. China Prof. Y. F. Zhao State Key Laboratory of Chemical Resource Engineering Beijing University of Chemical Technology Beijing 100029, P. R. China Dr. B. Wang Beijing Research Institute of Chemical Industry Sinopec Group Beijing 100013, P. R. China Prof. G. I. N. Waterhouse School of Chemical Sciences The University of Auckland Auckland 1142, New Zealand The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adma.201806482. DOI: 10.1002/adma.20180648

Ammonia (NH3) is one of the most important commodity chemicals in today’s chemical industry, representing a key feedstock for the synthesis of urea, ammonium nitrate, nitric acid, and various nitrogen-containing compounds.[1] NH3 finds further uses in the energy sector as an energy carrier.[2] Industrially, NH3 is produced by N2 hydrogenation through the Haber–Bosch process, which consumes ≈2% of global energy used annually by humans and releases huge amounts of greenhouse gases into the atmosphere (CO2 from fossil fuel combustion to provide energy for the process, and CO2 released from the steam methane reforming-water gas shift reactions used to make the H2 feedstock for the N2 reduction reaction).[3] To circumvent these problems, enormous research effort has been directed toward the discovery of alternative catalysts that can adsorb and activate N2 for hydrogenation under much milder conditions than those used in the Haber–Bosch process. Recently, research in the field of heterogeneous catalysis has shifted focus toward technologies and processes that are green and sustainable.[3,4] In the case of NH3 production, replacement of H2 with H2O as a reducing agent is highly desirable though finding a catalyst which can efficiently convert N2 and 3H2O to 2NH3 and 1.5O2 in a stable manner has proven a massive technical obstacle.[5]

Introducing solar energy or electricity rather than thermal energy into nitrogen fixation could provide lower energy input and lower carbon footprint technologies for N2 fixation.[6] Among these, photocatalytic nitrogen fixation is considered to be one of the most promising technologies for directly producing ammonia from nitrogen and water under ambient conditions. Schrauzer and Guth and co-workers first reported that TiO2-based photocatalysts were capable of driving nitrogen fixation.[7] Subsequently, various semiconductors including Fe2O3, [8] WO3, [9] BiOBr,[10] and others have been investigated in relation to their N2 fixation performance under visible light.[11] In addition, various biomolecules (such as [Fe–Fe] nitrogenase[12] and CdS:MoFe proteins[13]) have also been explored for N2 fixation. However, most photocatalysts studied to date exhibit weak light absorption above 500 nm, severely handicapping their solar spectrum-utilization efficiency for photocatalysis.[14] From the point of effective utilization of solar energy, it is imperative to develop new semiconductor photocatalysts with wide absorption range (ideally up to 700 nm) to allow efficient N2 fixation under ambient conditions.

Recently, defect engineering in semiconductor photocatalysts has attracted attention as a means of providing abundant active sites for optimizing reactant (e.g., N2 and H2O) adsorption and lowering activation energy barriers for challenging chemical transformations.[15] Recently, it was demonstrated that ultrathin CuCr-layered double hydroxides (LDHs) nanosheets containing abundant oxygen vacancies (VO) exhibited outstanding N2 reduction activities at wavelengths up to 500 nm. As the thickness of the LDH nanosheets was reduced, the N2 adsorption properties and photoinduced charge transport in the sheets were improved due to the higher concentration of VO. [14] However, the defect concentration could not be fully optimized as it was inherently linked to morphology control, with the minimum LDH thickness synthetically achievable being ≈2.4 nm. Accordingly, alternative synthetic strategies need to be developed that allow precise modulation and control of the VO concentration in semiconductor photocatalysts (ideally independent of morphology). Minimal literature data are presently available in this area, particularly for semiconductor photocatalysts operating at room temperature.

In this work, we demonstrate that the concentration of VO in ultrathin TiO2 nanosheets can be precisely controlled by doping with copper ions (Scheme 1), leading to greatly improved N2 photofixation performance under visible-light irradiation and near ambient conditions. Defect-rich TiO2 nanosheets containing 6 mol% Cu (denoted herein as 6%-TiO2 nanosheets) exhibited remarkable and stable performance toward photofixation of N2 to NH3 in water under UV–vis and visible light, with the activity extending up to 700 nm (NH3 evolution rates were 1.54 µmol g−1 h−1 at 600 nm and 0.72 µmol g−1 h−1 at 700 nm with corresponding quantum yields (QY) of 0.08% and 0.05%, respectively). Detailed structure analysis and density functional theory (DFT) calculations allow the excellent activity to be attributed to Jahn–Teller distortions and around the Cu dopant which create abundant VO and introduce compressive strain into the TiO2 nanosheets, which act synergistically to promote N2 adsorption and activation while facilitating fast charge separation under UV–vis or visible irradiation. This work demonstrates that bulk doping of TiO2 with Cu ions offers a new pathway for optimizing the concentration of VO for efficient dinitrogen photoreduction to ammonia Ultrathin TiO2 nanosheets containing different amounts of copper ions (denoted here as X%-TiO2, where X = mol Cu/mol Ti) were successfully synthesized using a facile hydrothermal method. X-ray diffraction (XRD) patterns for the X%-TiO2 nanosheets and a bulk anatase counterpart (denoted as Bulk-TiO2) are shown in Figure S1 (Supporting Information), all of which were consistent with anatase TiO2 (JCPDS-21-1272). With Cu doping, the anatase (101) reflection for the X%-TiO2 nanosheet samples shifted to higher 2θ angles (Figure 1a; Figure S1b, Supporting Information), which can be attributed to the introduction of lattice strain. Even the 0%-TiO2 nanosheets showed a negative strain compared with Bulk-TiO2, which can be rationalized in terms of the formation of oxygen defects as reported previously.[16] Quantitative energy-dispersive X-ray (EDX) (Figure S2, Supporting Information) analyses revealed that the actual Cu loading in the X%-TiO2 nanosheets was almost identical to the nominal loading in hydrothermal synthesis step. Transmission electron microscopy (TEM) examination of the 6%-TiO2 nanosheet sample, which is shown below to have the highest activity for photocatalytic N2 fixation, revealed a lateral sheet size of ≈10 nm and a thickness of ≈3.5 nm (Figure S3a,b, Supporting Information), in good agreement with the atomic force microscopy findings (Figure S4, Supporting Information). Lattice spacings of 0.19 and 0.19 nm corresponding to the (200) and (020) facets, respectively, could be discerned by high-resolution TEM (HRTEM), revealing the TiO2 nanosheets exposed a high proportion of reactive (001) facets.[17] The corresponding fast Fourier transform (FFT) image also confirmed the coexistence of (020) and (200) planes of TiO2 (the inset of Figure S3c in the Supporting Information). TEM-EDX mapping of the 6%-TiO2 nanosheets (Figure S3d, Supporting Information) confirmed a uniform distribution of Ti, O, and Cu throughout the nanosheets. The 1%-TiO2, 3%-TiO2, and 8%-TiO2 nanosheets all possessed similar morphologies to the 6%-TiO2 nanosheets (Figure S5, Supporting Information). The Bulk-TiO2 reference sample contained sintered spherical particles of average size 80–130 nm (Figure S6, Supporting Information).

Taking the above XRD, HRTEM, and TEM-EDX data into consideration, the shift in the position of the (101) reflection for X%-TiO2 with increasing X can be taken as evidence for a constricted lattice compared with Bulk-TiO2. Using the Voigt method, the lattice strain in each of the X%-TiO2 nanosheets (X = 0, 1, 3, 6, and 8) was calculated, with the analysis yielding values of −0.421%, −0.601%, −0.760%, −0.892%, −0.840%, respectively, compared with Bulk-TiO2 with the lattice parameter of a = b = 3.785 Å, and c = 9.513 Å (Table S1, Supporting Information). The data reveal that 6%-TiO2 nanosheets possessed the most compressive strain (−0.892%). Since Cu2+ ions (0.73 Å) are larger than Ti4+ ions (0.64 Å), the compressive strain seen for X%-TiO2 must be due to copper doping on titanium sites along with different concentrations of VO, [16] which was confirmed and discussed in detail below.

To probe the local atomic structure and cation coordination in X%-TiO2 resulting from doping Cu ions into the TiO2 nanosheets, X-ray absorption fine structure (XAFS) was employed (Figure S7, Supporting Information; Figure 1b,c). Ti K-edge X-ray absorption near-edge structure (XANES) spectra for Bulk-TiO2 and X%-TiO2 possessed three typical pre-edge peaks associated with the anatase phase (denoted as P1, P2, and P3), corresponding to quadruple-allowed 1s → 3d transitions,[18] providing further evidence for the formation of pure anatase samples (Figure S7a,b, Supporting Information; Figure 1b).[19] The increased intensity of P2 peak seen for the X%-TiO2 nanosheets is associated with severe Ti site distortions. The 6%-TiO2 sample gave the most intense P2 signal among all samples analyzed in this work, further indicating that the 6%-TiO2 contained the most severe structural distortions (the intensity of the P2 signal decreased in the order 6%-TiO2 < 3%-TiO2 < 8%-TiO2 < 1%-TiO2 < 0%-TiO2 < Bulk-TiO2, revealing that Cu doping was the probable origin of the distortion).[20] The corresponding R-space spectra are shown in Figure 1c. The Ti K-edge EXAFS spectra for X%-TiO2 nanosheets display two characteristic peaks assigned to the Ti–O shell and Ti-metal shell (Ti/Cu), respectively. On increasing the Cu concentration in TiO2 from 0% to 6%, the intensity of Ti–O shell peak gradually decreased and shifted to shorter distances (see the inset of Figure 1c). Table S2 (Supporting Information) reveals that when the amount of Cu dopant was increased from 0% to 6%, the Ti–O coordination number decreased from 5.98 for 0%-TiO2 to 5.52 for 6%-TiO2. The Bulk-TiO2 reference sample afforded a Ti–O coordination number of 6.0, typical for octahedrally coordinated Ti in TiO2. The data for the 8%-TiO2 nanosheet sample were anomalous, with a Ti–O coordination number of 5.92. This may be the result of new phases forming (such as CuOx), which is supported by the appearance of a new peak around 2.6 Å in Figure S8a (Supporting Information). Raman spectroscopy (Figure S9, Supporting Information) provided further evidence for local structural changes and distortions in the TiO2 nanosheets with increasing copper doping.[16,21] Ti K-edge EXAFS was used to probe the average Ti oxidation state in the TiO2 samples, in comparison with spectra from different Ti reference materials (Figure S10a,b, Supporting Information). As expected, the Ti K-edge spectra were consistent with Ti4+ being the dominant titanium state at all Cu concentrations. Table S3 (Supporting Information) reveals that when the amount of Cu dopant was increased from 0% to 6%, the oxygen vacancy concentration increased from 0.985% for 0%-TiO2 to 8.621% for 6%-TiO2. The above characterization data reveal that copper doping into anatase TiO2 nanosheets leads to lattice distortion and strain effects due to concomitant formation of VO. Abundant VO (Figure 1d) could be expected to enhance N2 adsorption on TiO2, as noted by other groups.[10,22]

Efficient light absorption and charge separation are key requirements for efficient photocatalytic performance.[14] Figure 2a shows UV–vis diffuse reflectance spectra (UV-DRS) for Bulk-TiO2 and the X%-TiO2 nanosheets. Bulk-TiO2 showed only absorption below 400 nm due to charge transfer process from valence band (VB) O 2p orbitals to conduction band (CB) Ti 3d orbitals. The 0%-TiO2 nanosheets showed an additional feature from 400 nm to longer wavelengths due to VO which lead to the formation of colored Ti3+ centers.[23] With increasing copper doping, the color of the samples transformed from white to dark green (the inset of Figure 2a), with the features above 400 nm becoming progressively stronger with Cu doping up to 6%, evidence for increased VO defect states in the TiO2 crystal lattice.[21c] Furthermore, the absorption edges for the X%-TiO2 nanosheets red shifted from 400 to 700 nm as X increased, which is explained by the presence of VO and also 2Eg → 2T2g transitions from O to Cu atoms.[21c] All the X%-TiO2 nanosheets exhibited a near-infrared absorption band near 800 nm, attributable to d-d transitions of the dopant Cu ions.[21c] The excellent light absorption ability of the X%-TiO2 nanosheets, especially in the range 400–800 nm, was expected to enhance photocatalytic performance and is explored below.

Recently, various TiO2 materials have been investigated for photocatalytic nitrogen fixation, though their performance is generally poor with rates of NH3 production typically below 10 µmol g−1 h−1 . [22] Introducing defects such as VO into TiO2 is expected to provide more active sites for N2 adsorption and thereby enhance photocatalytic performance. The performance of the ultrathin X%-TiO2 nanosheets for N2 fixation were evaluated under UV–vis irradiation, using both the Nessler’s regent and ion chromatography to quantify NH3 evolved (Figure S11, Supporting Information). As shown in Figure 2b, all the X%-TiO2 samples were active for N2 fixation, with the rates of NH3 evolution following the order 6%-TiO2 (78.9 µmol g−1 h−1 ) > 3%-TiO2 (64.6 µmol g−1 h−1 ) > 8%-TiO2 (57.12 µmol g−1 h−1 ) > 1%-TiO2 (45.2 µmol g−1 h−1 ) > 20%-TiO2 (23.33 µmol g−1 h−1 ) > 0%-TiO2 (15.3 µmol g−1 h−1 ), which were all higher than rates found for Bulk-TiO2 (0.34 µmol g−1 h−1 ) and Cu2O (2.64 µmol g−1 h−1 ) (Figure S12, Supporting Information). Comparing the data for Bulk-TiO2 with that of 0%-TiO2 (15.3 µmol g−1 h−1 ), the ≈50-fold higher rate for 0%-TiO2 can be explained by the higher specific surface area and the abundance of VO in the nanosheets, which cooperatively enhance N2 adsorption and activation.[10,22] The enhanced photocatalytic activity with Cu doping up to 6% can be rationalized in terms of the increased visible absorption in the range 400–800 nm linked to the increased concentration of VO. Further, increasing the Cu doping level above 6% was not beneficial (e.g., 8%-TiO2), since the formation of extra phases (such as CuOx) on the TiO2 surface may cover defect sites, reducing the concentration of VO available for N2 adsorption. The photocatalytic data confirm that controlled doping of TiO2 nanosheets with Cu ions can greatly enhance the activity of TiO2-based photocatalysts for N2 fixation (comparing 6%-TiO2 and 0%-TiO2, the enhancement factor was 5-fold), with extra VO introduced via Cu doping being primarily responsible for this performance enhancement.

To confirm that the NH3 evolved during the photocatalytic tests originated from N2 (rather than some other nitrogen source such as nitrate ions), further experiments were conducted using 15N2. The resulting product 15NH4 + was detected by the indophenol blue method,[22] and further analyzed by high-resolution mass spectroscopy (HRMS) (Figure 2c). The product solution obtained using 15N2 as the reactant exhibited a strong mass spectroscopy peak at m/z 199. Further, the 15N:14N abundance ratio was clearly higher when 15N2, rather than 14N2, was used as the source of nitrogen. The experiments thus confirm that N2 was indeed the source of NH3 evolved during the photocatalytic tests.

Since the X%-TiO2 (X > 0) nanosheets absorbed strongly in the near-infrared region (600–800 nm) (Figure 2a), the photocatalytic performance of 6%-TiO2 was investigated under monochromatic light at 600 and 700 nm. Ion chromatography (Figure S11b, Supporting Information; Figure 2d) was used to quantify NH3 evolution and establish the structure–activity relationship. Under 600 and 700 nm irradiation, no NH3 was formed in the absence of N2, whereas under a N2 flow the concentration of NH3 increased linearly with irradiation time. NH3 evolution rates for 6%-TiO2 were 1.54 µmol g−1 h−1 under 600 nm irradiation and 0.72 µmol g−1 h−1 under 700 nm irradiation, respectively (Figure S13, Supporting Information). A charge transfer process involving OCu bonding induced under 600 nm irradiation was confirmed by low-temperature electron paramagnetic resonance (EPR) spectroscopy (Figure S14a, Supporting Information), which is likely responsible for the excellent performance under near infrared excitation (λ > 600 nm). The QY for 6%-TiO2 under near monochromatic light irradiation closely matched the absorption spectrum of the semiconductor (Figure 2e). The QY of 6%-TiO2 was calculated to be 0.74% at 380 nm, 0.23% at 420 nm, 0.08% at 600 nm, and 0.05% at 700 nm. No NH4 + was detected via ion chromatography under 800 nm irradiation. The excellent photocatalytic stability of 6%-TiO2 was verified over 5 testing cycles, with no obvious decrease in the activity observed for N2 photofixation (Figure S15, Supporting Information). Further, no change in the oxidation state of either Cu or Ti was found for 6%-TiO2 after the five cycles (Figure S16, Supporting Information). The data confirm the outstanding photocatalytic stability of 6%-TiO2. Table S4 (Supporting Information) compares the photocatalytic performance of 6%-TiO2 for nitrogen fixation with various other semiconducting materials. The activity of 6%-TiO2 exceeds all TiO2-based materials reported to date, with superior performance up to 700 nm.

To probe the activation of N2 on the surface of 6%-TiO2, in situ diffuse reflectance infrared Fourier transformation (DRIFT) spectroscopy was used to follow the time-dependent change of the N-containing functional groups on the photocatalyst under UV–vis irradiation (Figure 2f).[10,14] DRIFT spectra for N2 adsorption on 6%-TiO2 are shown in Figure S17 (Supporting Information). The adsorption band (assigned to NHx species) becomes intensified with N2 exposure,[24] suggesting that N2 can be efficiently adsorbed and activated on the surface of 6%-TiO2. Following 60 min irradiation, absorption bands appeared in the range 3000–3700 and 1200–1700 cm−1 . The peak at 3555 cm−1 is attributed to a N–H stretching mode,[10] whereas bands at 1557 and 1300 cm−1 are characteristic for adsorbed NH3. With increasing irradiation time, the band at 1415 cm−1 assigned to adsorbed NH4 + intensified.[10] In addition, the signal of strongly chemisorbed N2 or H2O molecules can be detected at 1614 cm−1. [25] The DRIFT data thus provide strong evidence that NN tripe bonds can be activated on 6%-TiO2 leading to the formation of NH4 + species under light irradiation. For NH3 evolution from N2 and H2O, O2 was expected to form as a byproduct (e.g., N2 + 3H2O → 2NH3 + 1.5 O2). The rates of O2 (59.1 µmol g−1 h−1 ) and NH3 (78.9 µmol g−1 h−1 ) evolved during our tests were very close to theoretical ratio of 3:4, with no N2H4 detected as a byproduct of N2 photofixation (Figure S18, Supporting Information), revealing that water acts as the main proton source during photocatalytic reduction of N2 reduction to NH3. The photocatalytic activity for H2 production was investigated and the rate of H2 production is about 0.0264µmol h−1 for 20mg 6%-TiO2 nanosheets (1.32µmol g−1 h−1 ) under UV–vis irradiation. Compared with the activity for H2 production, the performance of nitrogen fixation can reach 78.9 µmol g−1 h−1 with almost no N2H4 and NO3 − (Figure S19, Supporting Information) being detected, demonstrating a high selectivity for N2 reduction to NH3.

As mentioned above, the 6%-TiO2 nanosheets demonstrated superior photocatalytic N2 reduction activity compared with its counterparts, with the origin of this superior activity being related to the TiO2 nanosheet defect structure, electronic structure and the resulting photoinduced charge transfer properties. To further understand the origin of the outstanding performance for 6%-TiO2, the existence of VO in X%-TiO2 was probed by EPR and X-ray photoelectron spectroscopy (XPS).[19] From Figure S14b,c (Supporting Information), it can be seen that Bulk-TiO2 is EPR silent, whereas 0%-TiO2 shows distinctive EPR signals at approximately g = 1.998 and g = 2.004 assigned to VO and Ti3+, respectively, confirming the coexistence of VO and Ti3+ in 0%-TiO2 (in agreement with the XRD, XAFS, and Raman findings above). The 6%-TiO2 nanosheets showed an additional triplet around 2587 Gauss corresponding to Cu2+ (Figure 3a).[26] XPS data for 0%-TiO2 and 6%-TiO2 showed characteristic Ti 2p peaks for Ti4+, though in the case of 6%-TiO2 these were shifted by about 0.16 eV to lower binding energy compared with the same peaks for 0%-TiO2 (Figure 3b). This small shift likely results from the presence of additional VO and Ti3+ in 6%-TiO2, in accordance with the EXAFS data above. XPS with both Cu+ and Cu2+ states identified by spectral deconvolution (Figure S8b, Supporting Information).

In order to investigate photoinduced charge transfer in X%- TiO2, we performed photocurrent and time-resolved photoluminescence decay measurements on 0%-TiO2 and 6%-TiO2. The data in Figure 3c show that the photocurrent density of 6%-TiO2 was much larger than that of 0%-TiO2, suggesting enhanced electron and hole separation in 6%-TiO2 following photoexcitation. The 6%-TiO2 nanosheets had longer average decay times (≈1.37 ns) compared with 0%-TiO2 (≈1.09 ns) (Figure 3d; Table S5, Supporting Information), demonstrating that VO act as trapping sites for photogenerated electrons thereby increasing the lifetime of charge carriers. The data indicate that moderate Cu doping into TiO2 crystal facilitates effective separation of electron–hole pairs, leading to enhanced nitrogen activation and high photoreduction rates.

Mott–Schottky (MS) analyses and VB-XPS spectra were collected to obtain the flat band potentials for 0%-TiO2 and 6%-TiO2. The flat band potential of 6%-TiO2 was calculated to be −0.25 V versus the normal hydrogen electrode (NHE), which is slightly more negative than that determined for 0%-TiO2 (−0.2 V vs NHE) (Figure 3e). The MS plot for 0%-TiO2 and 6%-TiO2 were further analyzed and the number of charge carriers was calculated. From the slope of the MS plot, the carrier concentration of 0%-TiO2 and 6%-TiO2 were estimated to be 1.02 × 1017 and 1.58 × 1017 cm−3 , with the difference probably due to the action of extra surface states in the 6%-TiO2 nanosheets capturing and immobilizing the carriers. Moreover, we also investigated the charge transfer resistance by electrochemical impedance spectroscopy (EIS) measurements, with and without illumination. The diameter of the semi-circle in a Nyquist plot is proportional to the charge-transfer resistance of a sample, thus providing valuable information on charge transfer processes. As shown in Figure S20a (Supporting Information), the semi-circle for the 6%-TiO2 nanosheets is much smaller than that of 0%-TiO2 nanosheets under same testing conditions, indicative of very efficient electron transfer in the 6%-TiO2 nanosheets. Intriguingly, the two samples exhibited a smaller semi-circle diameter under light irradiation than that in the dark, suggesting improved electron transport under light illumination.[27] Linear voltammetry data for 0%-TiO2 and 6%-TiO2 nanosheets were conducted and the results were presented in Figure S20b (Supporting Information). Enhancement of the current density was achieved in N2-saturated solutions, which could be taken as further evidence that the 6%-TiO2 nanosheets were efficient for activation of nitrogen molecules. From the UV-DRS measurements (Figure S21, Supporting Information), the bandgap for the 6%-TiO2 nanosheets was estimated to be ≈3.0 eV with the VBM located at 2.75 eV (Figure 3f).[28] From a thermodynamic viewpoint, N2 photoreduction to NH3 is more favored over VO-doped ultrathin TiO2 than pristine TiO2.

It is well established that the electronic structure of semiconductor photocatalysts strongly influences their photocatalytic performance.[29] Accordingly, DFT calculations were used to examine the bandgap and density of states (DOS) of TiO2 (denoted as TiO2-Pure), VO-doped TiO2 (denoted as TiO2-VO) and Cu-doped TiO2 with VO and 0.8% compressive strain (denoted as TiO2-VO-Strain). Figure S22 (Supporting Information) shows the crystal structure models used for the DFT calculations. TiO2-Pure showed a bandgap of 3.15 eV between the VBM (mainly composed of O 2p orbitals) and the CBM (composed of Ti 3d orbitals) (Figure S23, Supporting Information). For TiO2-VO (Figure S24, Supporting Information), a narrower bandgap (2.79 eV) was observed and a new mid gap defect state appeared, consistent with previous studies.[14] After introducing VO and Cu with a strain effect, the electron density around the O atoms became much larger than that around the Ti or Cu atoms, indicating that electrons accumulate around the O atoms (Figure 4a). An obvious defect level (containing contributions from Cu 3d and O 2p orbits) was formed and divided the bandgap into two distinct sections (Figures S25 and S26, Supporting Information), providing strong evidence for a new electron transfer path (O–Cu1+ or O–Cu2+) in TiO2-VO-Strain. This new path would explain the photocatalytic activity observed for 6%-TiO2 under 600 and 700 nm illumination. It is envisaged that the defect levels introduced by Cu doping would serve as electron-trapping sites for accelerated transfer of photogenerated electrons from TiO2-VO-Strain (i.e., conduction band electrons) to π* antibonding orbitals of molecular N2, thereby promoting N2 activation and reaction.[14,30] For nitrogen fixation on the surface of TiO2, the optimized adsorption energy of N2 is a critical consideration.[29] Figure 4b shows adsorption energies for N2 on the surface of TiO2-Pure, TiO2-VO, and TiO2-VO-Strain, corresponding to the optimized adsorption structures (Figure S27, Supporting Information). The adsorption energy trend is TiO2-VO-Strain (−0.37 eV) > TiO2-VO (−0.25 eV) > TiO2-Pure (−0.17 eV), suggesting that adsorption was most favorable on the surface of TiO2-VO-Strain, a finding supported by studying the nitrogen absorption–desorption isotherms and N2 temperature-programmed desorption data for the 6%-TiO2, 0%-TiO2, and Bulk-TiO2 samples, respectively (Figure S28, Supporting Information). The increased adsorption energy may lead to improved charge transfer between TiO2 and N2, while also weakening the NN triple bond (cf. the NN length on TiO2- VO-Strain was 1.163 Å, compared with 1.155 Å on TiO2-Pure to 1.160 Å on Ti-VO) (Figure 4c). Moreover, the Gibbs free energy for nitrogen fixation is also important.[31] In photocatalysis, the generally accepted mechanism for N2 fixation is sequential hydrogenation of adsorbed N2. [10] Thus, after N2 molecular adsorption, the subsequent hydrogenation step leading to N2H* is considered as the most difficult and significant step in whole N2 fixation process. N2 hydrogenation to N–NH* required a reaction energy of 2.115 and 0.893 eV on TiO2-Pure and TiO2- VO (Figure 4d; Table S6, Supporting Information), respectively, whereas only 0.365 eV was required on TiO2-VO-Strain.[31c] Subsequent N–NH* hydrogenation to N–NH2* required a reaction energy of only 0.223 eV on the surface of TiO2-VOStrain. The data indicate that the sluggish Volmer step on TiO2 can be greatly accelerated after the introduction of VO and strain engineering (both of which exist in 6%-TiO2), resulting in enhanced N2 activation and fixation.

In summary, ultrathin TiO2 nanosheets with different concentrations of VO were successfully synthesized through a facile copper-doping method which exploited Jahn–Teller distortions. The obtained X%-TiO2 nanosheets exhibited a wide solar absorption range and enhanced electron–hole pairs separation efficiency compared with pristine TiO2, resulting in the superior performance for N2 fixation in water. The introduction of Cu as a dopant in TiO2 nanosheets creates additional VO and introduces substantial compressive strain, modifications which enhance N2 adsorption and provide a low energy pathway to NH3. High NH3 evolution rates were thus obtained, especially for 6%-TiO2 which afforded rates of 78.9 µmol g−1 h−1 under full solar irradiation, and 1.54 and 0.72 µmol g−1 h−1 under 600 and 700 nm monochromatic excitation, respectively. Results guide the development of new high-performance semiconductor photocatalysts with controllable VO concentrations and strain engineering for solar ammonia synthesis.

Supporting Information

Supporting Information is available from the Wiley Online Library or from the author.

Acknowledgements

Y.X.Z., Y.F.Z., and R.S. contributed equally to this work. The authors are grateful for financial support from the National Key Projects for Fundamental Research and Development of China (2017YFA0206904, 2017YFA0206900, and 2016YFB0600901), the National Natural Science Foundation of China (51825205, 51772305, 51572270, U1662118, 21871279, and 21802154), the Beijing Natural Science Foundation (2191002, 2182078, and 2194089), the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB17000000), the Royal Society-Newton Advanced Fellowship (NA170422), the International Partnership Program of Chinese Academy of Sciences (GJHZ1819), the Beijing Municipal Science and Technology Project (Z181100005118007), the K. C. Wong Education Foundation, the Young Elite Scientist Sponsorship Program by CAST (YESS), and the Youth Innovation Promotion Association of the CAS. G.I.N.W. acknowledges funding support from the Energy Education Trust of New Zealand and the MacDiarmid Institute for Advanced Materials and Nanotechnology.