Defect-phase engineered NiTi-TiO₂ enabling near-unity selective photocatalytic CO₂-to-methanol conversion

Abstract

Scaling up methanol yields by artificial photosynthesis at a modest cost remains thermodynamically challenge. Designing concerted reaction sites to control intermediate evolution and stimulate proton-coupled electron transfer (PCET) is necessary. Here we show a nickel-titanium-based catalyst that achieves near-millimolar hourly methanol yields with 99.79% selectivity and a solar-to-chemical conversion efficiency of 2.23%. This catalyst is synthesized through one-step etching of NiTi-layered double hydroxide, which generates abundant unsaturated sites, along with the in-situ formation of amorphous TiO₂. Revealed by in-situ characterizations, these defect-rich units effectively suppress the formation of undesirable carbonate while promoting the favorable *COOH intermediate. Furthermore, theoretical simulations confirm this *COOH boost facilitates the production of *CO and accelerates the PCET steps. This work significantly advances efficient methanol production by artificial photosynthesis and offers fundamental insights into controlling reaction pathways for renewable fuel synthesis.

Introduction

Artificial photosynthesis is a promising pathway toward achieving carbon neutrality while producing renewable feedstocks^1,2. Among the various products of CO₂ photoreduction, methanol (CH₃OH) holds significant economic value, ranking as the fourth-largest chemical raw material globally³. However, current CH₃OH production rates in water systems are limited to a few dozen micromoles. Although increased energy input could increase output, such demanding conditions often correspond to high costs. Therefore, how to improve CO₂ photoreduction rates and selectively at a gentle cost is crucial.

CO₂ can be transformed into complex intermediate species upon photocatalysis, including *CO, *CO₃²⁻, *COOH and CH_xO, etc⁴. The formation and evolution of these intermediate species play crucial roles in determining the selectivity and photoactivity of the final products⁵. To increase CH₃OH production, promoting the formation of *CO species and accelerating proton-coupled electron transfer (PCET) reactions are essential. However, stabilizing these favorable species and regulating their transformation dynamics pose significant challenges.

Over the years, various strategies have been employed to increase the protonation of CO₂ in conjunction with electron transfer^6,7. Effective approaches involve regulating the interface microelectric field or creating synergistic sites for CO₂ activation and hydrolysis dissociation^8,9. Reports have shown that low-coordinated atoms or defective sites are more active in catalytic reactions, effectively promoting the PCET reactions, owing to their distinct chemical environments^10,11. In particular, the dangling bonds at the edges can modulate the electron dynamics and interfacial reactions, thereby improving the CO₂ reduction performance¹².

Layered double hydroxides (LDHs), which consist of divalent and trivalent metal cations, exhibit compositional flexibility and the potential to create various edge vacancy sites^13,14. Researchers have successfully constructed abundant oxygen vacancies (V_O) and metal vacancy (V_M) by developing ultrathin nanosheets, which significantly enhance photoactivity^15,16,17. However, the synthesis of ultrathin materials is often time-consuming and suffers from stability issues. Capitalizing on the pH-sensitive properties of LDHs, Cai et al. successfully utilized polyoxometallic acids to etch NiFe-LDHs, generating substantial Ni, Fe, and O vacancies¹⁸. Additionally, Luo et al. demonstrated that alkali etching of NiZnTiAl-LDH precursors enabled the creation of Zn and Al vacancies, thereby enhancing CO₂ photocatalytic performance¹⁹. Despite these advancements, it remains challenging to control the vacancy selectivity. Furthermore, the relationships between specific defect sites, CO₂ activation intermediates, and catalytic performance have not been sufficiently established. Thus, engineering effective unsaturated sites with practical feasibility, while simultaneously facilitating PCET reactions to optimize the distribution of intermediates and clarify the reaction mechanisms, remains a challenging goal.

In this study, we successfully engineered effective unsaturated sites within a NiTi-LDH matrix via a wet chemical etching method. This approach facilitated phase transformation, allowing the construction of defective NiTi-TiO₂ units (D-NTL/TO). The optimal catalyst exhibited CH₃OH production rates of 0.97 mmol g⁻¹ h⁻¹ with nearly 100% selectivity without any sacrificial agents or photosensitizers. Our investigations revealed that amorphous TiO₂ (TO) was uniformly dispersed on the defective NTL (D-NTL), which promoted the separation of photogenerated electron-hole pairs. In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) and theoretical calculations suggested that D-NTL/TO units optimized the distribution of CO₂ intermediate species and accelerated the PCET reactions, specifically promoting the formation of *COOH species for subsequent conversion to CH₃OH. This work presented an efficient strategy for enhancing the photoconversion of CO₂ into CH₃OH, indicating significant advancements in mimicking natural photosynthesis.

Results

Morphology and structure analysis

NiTi layered double hydroxides (LDHs) were synthesized via a urea homogeneous coprecipitation method, followed by etching with an oxidizing nitric acid solution to create edge defect sites. X-ray diffraction (XRD) was employed to identify the crystal structure of the catalysts, as illustrated in Fig. 1a. The diffraction pattern of the NTL sample displayed distinct Bragg reflections at approximately 12.04°, 24.19°, 33.71°, 34.56°, 37.61°, 59.51°, and 60.83°, corresponding to the (003), (006), (012), (015), (018), (110), and (113) crystal planes, respectively. These reflections are indicative of a hydrotalcite structure and consistent with both the simulated XRD patterns and the reference PDF card (Supplementary Fig. 1-3). Following the acid-etching treatment, a noticeable decrease in the intensity of these diffraction peaks was observed, suggesting decreased crystallinity. Additionally, the (003) lattice plane progressively shifted toward higher 2θ values, suggesting a decrease in the interlayer spacing from 7.34 Å in NTL to 7.09 Å in D-NTL/TO-2 by acid etching, as calculated through the Bragg equation (Supplementary Table 1). The TO sample completely lacked characteristic diffraction peaks associated with the structure of hydrotalcite. However, new peaks were emerged at approximately 25.41°, 37.91°, 47.52°, 54.47°, and 63.10°, which were attributed to the (101), (004), (200), (105), and (204) crystal facets, respectively, characteristic of anatase TiO₂. The Raman spectra of these catalysts are depicted in Supplementary Fig. 4. The band observed at 871 cm⁻¹ was associated with HO-CO₃²⁻ units, which decreased in intensity for D-NTL/TO-2 compared with NTL. The continued etching process led to the emergence of new bands at 152, 404, 414 and 636 cm⁻¹, indicative of the anatase TiO₂^20,21.

Fig. 1: Structural and morphological characterization.

a XRD patterns of the synthesized catalysts. b O 1 s XPS spectra and c EPR spectra of bulk NTL and acid-etched derivatives. d Normalized Ni K-edge XANES spectra of NTL and D-NTL/TO-2. e, f The k²-weighted FT profiles in R-space from Ni K-edge and Ti K-edge EXAFS spectra for NTL and D-NTL/TO-2, respectively. g SEM image and h SAED pattern of NTL. i SEM image, j SAED pattern, k HRTEM image, l AC-TEM image with corresponding 3D surface plot and atomic profile of white-dotted area, and m EDS-STEM element mapping of D-NTL/TO-2. Source data for Fig. 1a-f are provided as a Source Data file.

In the thermogravimetric coupled with mass spectrometry (TG-MS) analysis, carbonate ions were identified as the main interlayer anions in NTL (Supplementary Fig. 5). Following acid etching, the content of CO₃²⁻ ions decreased, whereas the content of interlayer NO₃^- ions increased, as demonstrated by the Ar-temperature programmed desorption coupled with mass spectra analysis (Ar-TPD-MS, Supplementary Fig. 6). The O 1 s X-ray photoelectron spectroscopy (XPS) spectra revealed three peaks at binding energies of 530.45, 531.25 and 532.66 eV, which were assigned to lattice oxygen, hydroxyl groups (-OH), and hydroxyl vacancies (OH_v), respectively (Fig. 1b)^18,22. Notably, the concentration of OH_v in D-NTL/TO-2 was significantly greater than that in NTL. Furthermore, the O 1 s XPS spectrum of TO shifted to a lower binding energy, accompanied by an enhanced contribution of lattice oxygen, indicating the phase transition from LDH to TiO₂. In the Ni 2p XPS spectra (Supplementary Fig. 7a), the Ni 2p_3/2 band of D-NTL/TO-2 shifted to a higher binding energy, suggesting the potential formation of Ni vacancies (Ni_v)²³. The Ti 2p XPS spectra revealed 5.67 eV between the Ti 2p_1/2 and Ti 2p_3/2 bands, suggesting the presence of Ti⁴⁺. The binding energy gradually shifted to higher values, indicating a phase transition or electron transfer due to alterations in the chemical environment at the Ti sites (Supplementary Fig. 7b). Additionally, electron paramagnetic resonance (EPR) measurements revealed signals at g = 2.00 (Supplementary Fig. 8) and g = 1.95 (Fig. 1c), confirming the presence of OH_v and Ni_v, respectively^24,25. Note that the fluctuation amplitudes of D-NTL/TO-2 were greater than those of NTL and D-NTL/TO-1, which should be attributed to increased distortions resulting from the acid etching process. The N₂ adsorption-desorption experiments suggested a progressive increase in the specific surface area from NTL to D-NTL/TO-2, followed by a decrease upon completion of the phase transition to TiO₂ (Supplementary Fig. 9 and Table 2).

To further investigate the electronic and coordination structure in etched sample, X-ray absorption fine structure (XAFS) measurements were performed. As shown in Fig. 1d, the Ni K-edge XANES spectra demonstrated that the absorption edge of D-NTL/TO-2 shifted to higher energy relative to NTL, coupled with enhanced white-line intensity. These observations suggested an increased oxidation state of Ni (Ni^2+δ, 0<δ < 1), consistent with the XPS findings. The corresponding k²-weighted Fourier transform (FT) of Ni K-edge EXAFS spectra (Fig. 1e) revealed two characteristic peaks at 1.56 Å and 2.75 Å for NTL, corresponding to first-shell Ni-O bonds and second-shell Ni-O-Ni/Ni-O-Ti coordination, respectively. Notably, D-NTL/TO-2 demonstrated shortened bond lengths compared to the pristine NTL, which could be attributed to defect-mediated structural reorganization and the decreased ionic radius of Ni resulting from its higher oxidation state. Similarly, Ti K-edge EXAFS spectra displayed FT peaks at 1.43 Å and 2.64 Å, assigning to first-shell Ti-O and second-shell Ti-O-Ni/Ti-O-Ti coordination, respectively (Fig. 1f). The D-NTL/TO-2 sample showed broader Ti spectral peaks compared to NTL, suggesting increased structural disorder. This conclusion was further supported by wavelet transform analysis of k²-weighted EXAFS signals at both Ni and Ti K-edges, which showed attenuated second-shell contributions, confirming disruption of long-range order (Supplementary Fig. 10). Quantitative χ(R)-space spectral fitting was conducted to analyze local coordination environments. The Ni-O coordination number decreased from 6.51 in NTL to 4.93 in D-NTL/TO-2, while the Ti-O coordination number reduced from 5.36 to 3.72 after etching (Supplementary Tables 3-4). The k²-weighted XAFS χ(k) versus reciprocal wave vector plots confirmed data reliability (Supplementary Figs. 11-12). This analysis further confirmed the formation of structural defects in NTL following the etching process.

Scanning electron microscopy (SEM) images revealed the typical nanosheet structure of NTL (Fig. 1g). The Ni/Ti ratio in NTL was determined to be approximately 3.40 via energy-dispersive X-ray spectroscopy (EDS, Supplementary Fig. 13). High-resolution transmission electron microscopy (HRTEM) and selected-area electron diffraction (SAED) patterns revealed that the (012) and (110) facets of NTL were exposed, with corresponding lattice spacings of 0.25 and 0.15 nm, respectively (Fig. 1h). Interestingly, the etching treatment applied to NTL led to significant shrinkage of the nanosheets (Fig. 1i). The Ni/Ti ratios gradually decreased from 1.14 in D-NTL/TO-1 (Supplementary Fig. 14) to 0.53 in D-NTL/TO-2 (Supplementary Fig. 15) and 0.25 in D-NTL/TO-3 (Supplementary Fig. 16). The HRTEM image of D-NTL/TO-2 revealed that the lattice structure was damaged, resulting in the formation of numerous defects (Fig. 1k). Additionally, cationic vacancies were identified for the D-NTL/TO-2 catalyst, as indicated by aberration-corrected transmission electron microscopy (AC-TEM) image and the corresponding intensity profiles of the framed regions in Fig. 1l. The corresponding SAED pattern (Fig. 1j) displayed blurred diffraction rings for D-NTL/TO-2, suggesting a decrease in crystallinity. Importantly, the SAED pattern revealed the coexistence of both the LDH and TiO₂ phases within the D-NTL/TO-2 catalyst. It could be deduced that TiO₂ formed locally in situ during the acid etching process because of the lattice confinement effect, resulting in high dispersion in the defective NTL. Scanning transmission electron microscopy (STEM) and the corresponding EDS images suggested a uniform distribution of Ni and Ti (Fig. 1m). In contrast, the TO system exhibited a near-complete absence of Ni, with the morphology transitioning from a nanosheet to a granular form (Supplementary Fig. 17). HRTEM analysis revealed distinct lattice fringes with a spacing of 0.19 nm, corresponding to the TiO₂ (200) facet (Supplementary Fig. 18). Additionally, inductively coupled plasma optical emission spectroscopy (ICP-OES) was conducted on these samples, which suggested that the Ni content gradually decreased during the acid etching process, as displayed in Supplementary Table 5. As mentioned above, the etching process for NTL induced abundant OH_v and Ni_v sites within the lamellar structure, simultaneously leading to the in-situ formation of TiO₂.

Photocatalytic CO₂ conversion performance

The photocatalytic CO₂ reduction performance of these catalysts was evaluated in H₂O media free from organic sacrificial agents or photosensitizers with the setup in Supplementary Fig. 19. Gas chromatography (GC) analysis revealed the production of CO and CH₃OH, where the CH₃OH yield gradually increased as the etching process was applied to NTL (Supplementary Figs. 20-21). Among these catalysts, the D-NTL/TO-2 catalyst afforded the highest CO₂-to-CH₃OH production rate, reaching 970.76 μmol g⁻¹ h⁻¹ (Fig. 2a, Supplementary Table 6), with CO₂ conversion rate of approximately 0.2%. ¹H Nuclear Magnetic Resonance (¹H NMR) and ion chromatography (IC) indicated the absence of other liquid products such as HCHO or HCOOH, demonstrating that CH₃OH was the sole liquid product with a selectivity of 99.79% (Supplementary Fig. 22). Furthermore, the ¹H NMR spectra (Supplementary Fig. 23) of the reaction products obtained with different catalysts demonstrated consistent trends in methanol production rates with GC analysis.

Fig. 2: Evaluation of the photocatalytic CO₂ reduction performance.

a Product yields over different photocatalysts, in which error bars represent the standard deviation derived from three independent measurements. b Mass spectrum result of ¹³CO₂ labeled test over D-NTL/TO-2. c AQY values under various wavelengths of light irradiation over D-NTL/TO-2. d, e The performance comparasion of photocatalytic CO₂ reduction in our work with that in the previous literature. Source data for Fig. 2a-e are provided as a Source Data file.

To trace the carbon source for CH₃OH synthesis, a ¹³CO₂-labeling experiment was conducted. As depicted in Fig. 2b, mass spectrum analysis revealed signals at m/z values of 30, 32 and 33, which confirmed the formation of ¹³CH₃OH²⁶. The carbon source was further verified through isotopic labeling experiment using ¹³C NMR and ¹H NMR analysis. The ¹³C NMR spectrum displayed a characteristic peak at δ 49.2 ppm corresponding to the carbon atom in ¹³CH₃OH (Supplementary Fig. 24a), while the ¹H NMR spectrum revealed doublet signals at δ 3.39 and 3.16 ppm, demonstrating significant ¹³C-¹H coupling interactions of ¹³CH₃OH (Supplementary Fig. 24b). Furthermore, a control experiment in which CO₂ was substituted with Ar demonstrated no detectable production of CH₃OH (Supplementary Fig. 25). These findings illustrated that CO₂ served as the carbon source for CH₃OH production. Additionally, the removal of any one of the components, such as photocatalysts, H₂O or light irradiation, from the reaction system resulted in cessation of CH₃OH generation, suggesting the critical roles of these factors in the photocatalytic CO₂ reaction. The optimal D-NTL/TO-2 sample exhibited nearly constant photocatalytic activity throughout the cycle tests (Supplementary Fig. 26). The morphology, crystal structure, and optical properties of the used D-NTL/TO-2 remained unchanged after the reaction (Supplementary Figs. 27-29). The apparent quantum yield (AQY) of D-NTL/TO-2 was determined to be approximately 17.24% at 420 nm, as illustrated in Fig. 2c.

Recent reports, as summarized in Fig. 2d and details in Supplementary Table 7, have shown that the production rates of photocatalytic CO₂ reduction in pure H₂O systems are typically limited to micromolar levels. Catalysts that yield higher outputs generally contain precious metals (such as Au), sacrificial agents (such as triethanolamine), or severe conditions (high pressure/temperature or H₂ reductant). Moreover, more than 50% of the developed photocatalysts tend to convert CO₂ into gaseous products, including CO and CH₄ (Fig. 2e), which poses challenges for separation from gaseous CO₂. In contrast, liquid products, such as CH₃OH, are more readily separable and play a significant role in the synthesis of commodity chemicals. In this study, the 970.76 μmol g⁻¹ h⁻¹ CH₃OH production rates with 99.79% selectivity was achieved without employing precious metal, sacrificial agents, or harsh conditions. Under visible-light (λ ≥ 420 nm) irradiation, the optimal CH₃OH yield achieved 203.42 μmol g⁻¹ h⁻¹ (Supplementary Fig. 30). The solar-to-chemical (STC) conversion efficiency (η%) was calculated to be 2.23%, which surpassed most recently reports^{27,28,29,30,31}.

Mechanistic insights into the CO₂ photoreduction process

Ultraviolet-visible diffuse reflectance spectroscopy (UV-vis DRS) was employed to investigate the light-harvesting capabilities of the synthesized catalysts (Supplementary Fig. 31a). NTL demonstrated significant light absorption at wavelengths below 500 nm, which can be attributed to ligand-to-metal charge transfer (LMCT)³². Additionally, two distinct absorption peaks were observed at 673 nm and 752 nm, corresponding to the d-d transitions of Ni²⁺ ions within an octahedral coordination environment³³. After acid etching, the D-NTL/TO-1 and D-NTL/TO-2 samples exhibited decreased intensities for the absorption peaks at 673 nm and 752 nm, suggesting a reduction in the Ni content. Furthermore, the adsorption edge shifted slightly toward shorter wavelengths, suggesting alterations in the electronic structure. In contrast, the TO sample demonstrated characteristic light-absorption properties associated with TiO₂³⁴, which further supported the phase transformation following acid etching. The band gap energies (E_g) of NTL and TO were determined to be 2.92 eV and 3.35 eV, respectively, as calculated via Tauc’s equation (Supplementary Fig. 31b-c). Mott-Schottky plots were constructed to evaluate the conduction band potential (E_CB), which was found to be −1.85 eV for NTL and −1.00 eV for TO, as measured against the normal hydrogen electrode (vs. NHE), as shown in Supplementary Fig. 32a. Consequently, the valence band potentials (E_VB) of NTL and TO could be calculated to be 1.07 and 2.35 eV, respectively, enabling the construction of energy band diagrams for NTL and TO (Supplementary Fig. 32b).

Density functional theory (DFT) calculation yielded work function (Φ) values of 2.28 eV for NTL (Fig. 3a) and 7.05 eV for TO (Fig. 3b). These results suggested that the Fermi level (E_F) of TO was lower than that of NTL, which could facilitate spontaneous electron transfer from NTL to TO and thereby align the Fermi levels in the D-NTL/TO heterojunction³⁵. The interfacial charge density differences were theoretically calculated to further determine the accumulation of electrons and holes. As illustrated in Fig. 3c, d, NTL exhibited positive charge accumulation, whereas TO displayed negative charge accumulation, which provided clear evidence for the formation of an internal electric field (IEF) directed from NTL to TO. Upon light irradiation, electrons can be excited from the valence band to the conduction band. Driven by the IEF, electrons should migrate from the conduction band of TO to the valence band of NTL. To further verify this charge migration mechanism, in situ irradiated XPS (ISI-XPS) spectra were performed. After 10 minutes of irradiation, the Ni 2p spectra exhibited a negative binding energy shift (Supplementary Fig. 33a), indicative of electron accumulation. Conversely, the Ti 2p spectra showed a slight positive shift under illumination (Supplementary Fig. 33b), suggesting electron depletion. These results confirmed that D-NTL and TO function as electron acceptor and donor species, respectively, during photocatalytic process. This charge transfer mechanism was consistent with the characteristics of an S-type heterojunction³⁶, as depicted in Fig. 3e.

Fig. 3: Energy band structure and charge migration mechanism.

a, b DFT calculations on work function of NTL and TO. c, d Planar-averaged electron density difference from DFT calculations, in which yellow and cyan regions represent negative and positive electron density, respectively. e Energy band structure of NTL and TO (top panel) and charge migration diagram in D-NTL/TO (bottom panel). f Photocurrent profiles, g time-resolved PL spectra, and h internal electric field intensities of NTL and D-NTL/TO-2 catalysts. Source data for Fig. 3a-c and f-h are provided as a Source Data file.

The separation and migration dynamics of electron-hole (e⁻/h⁺) pairs in the NTL and D-NTL/TO-2 catalysts were evaluated through photochemical characterization. As depicted in Fig. 3f, D-NTL/TO-2 exhibited a significantly higher photocurrent intensity than NTL did, along with a markedly lower interfacial resistance (Supplementary Fig. 34a), indicating the improved charge separation performance of the D-NTL/TO-2 sample. This conclusion was further supported by the photoluminescence (PL) spectra, which exhibited a quenched fluorescence signal for the D-NTL/TO-2 sample (Supplementary Fig. 34b). Additionally, time-resolved photoluminescence (TRPL) spectra (Fig. 3g) revealed that D-NTL/TO-2 had a longer average carrier lifetime than did NTL. The proportion of the radiative recombination parameter τ1 decreased from 54.50% for NTL to 39.51% for D-NTL/TO-2, suggesting greater availability of free electrons in the latter^37,38. Furthermore, the IEF, which serves as the driving force for charge migration, was quantitatively estimated via the model developed by Kanata et al.^39,40. The IEF was found to depend on the surface voltage and charge density of a catalyst. Following the methodology detailed in Supplementary Fig. 35a, the carrier concentration for D-NTL/TO-2 was determined to be 2.46 times greater than that of NTL. Moreover, open-circuit potential tests (Supplementary Fig. 35b) revealed surface voltages of 0.036 V for NTL and 0.096 V for D-NTL/TO-2. Thus, the IEF intensity of D-NTL/TO-2 could be calculated to be 2.58 times greater than that of NTL, as illustrated in Fig. 3h, which could be the reason for the enhanced charge separation efficiency.

The distribution and evolution of reaction intermediates during CO₂ photoreduction on catalysts were investigated via in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS). In the NTL sample (Fig. 4a), the absorption peaks at 1290 and 1619 cm⁻¹ were described to the vibrations of carbonate (*CO₃²⁻)^41,42. The peaks at 1420 and 1670 cm⁻¹ were belonged to adsorbed carbamic acid (*COOH) and H₂O species, respectively. In the D-NTL/TO-1 and D-NTL/TO-2 samples, several new IR bands emerged (Fig. 4b, c). Specifically, peaks at 1524, 1420 and 1331 cm⁻¹ were identified as *COOH species, which were formed by proton-coupled electron transfer (PCET) reactions (CO₂ + H⁺ + e^- → *COOH)^43,44. Furthermore, the IR bands at 1134/1368, 1090 and 1048 cm⁻¹ should correspond to *CHO, *CH₂O and *CH₃O species, respectively, which are also closely associated with the PCET process^45,46. Notably, the IR bands of the TO were significantly different from those of the other samples (Supplementary Fig. 36), which could be attributed to the phase transition from LDH to TiO₂. Several IR peaks at 1383/1571, 1619, and 1670 cm⁻¹ were observed, which could be described to *COOH, *CO₃²⁻ and H₂O, respectively.

Fig. 4: In situ DRIFTS monitoring of photocatalytic CO₂ reduction.

(a₁-c₁) In situ DRIFTS spectra NTL, D-NTL/TO-1, and D-NTL/TO-2 during co-adsorption of CO₂/H₂O vapor in dark for 30 min followed by light irradiation for 30 min. (a₂-c₂) Corresponding time-dependent DRIFTS color maps. (a₃-c₃) Relative intensities of key reaction intermediates during in situ DRIFTS testing. Source data for Fig. 4a-c are provided as a Source Data file.

The corresponding IR maps are shown in Fig. 4a₂-c₂, and the variations in the intensities of the IR peaks during the in-situ DRIFTS tests are presented in Fig. 4a₃-c₃. The NTL sample significantly stimulated the formation of *CO₃²⁻ and H₂O species (Fig. 4a₂-a₃). After etching, there was a gradual decrease in the proportion of *CO₃²⁻ species, accompanied by an increase in the concentration of *COOH intermediates (Fig. 4b₃-c₃). It is well established that *COOH species play a critical role in facilitating the conversion of CO₂ into CO, CH₃OH, and other products. Additionally, the detection of *CHO, *CH₂O, and *CH₃O species indicated the continuous hydrogenation of *CO species, which was essential for CH₃OH production. These findings suggested that the acid etching process applied to NTL significantly altered the distribution of intermediate species. The *CO₃²⁻ species were inhibited, whereas the *COOH species and the continuous hydrogenation of *CO species were promoted, which should be favorable for CO₂ reduction to CH₃OH.

The CO₂ capture capacities of NTL, D-NTL/TO-2 and TO were evaluated via pulsed chemisorption, and the results are presented in Fig. 5a. NTL demonstrated the highest CO₂ adsorption capacity, reaching 0.11 mmol g⁻¹. Combined with the in-situ DRIFTS results, it was determined that the adsorbed species mainly corresponded to carbonate. Following acid etching, a decrease in the CO₂ adsorption capacity was observed, with the intermediate species mainly transforming to *COOH. Compared with both the NTL and D-NTL/TO-2 catalysts, the TO sample exhibited a significantly lower CO₂ capture capacity. Thus, CO₂ tended to competitively adsorb onto the D-NTL component rather than the TO. Moreover, we can reasonably speculate that the photocatalytic performance was strongly related to the adsorbed species on the catalyst surface rather than merely the amount of adsorption.

Fig. 5: CO₂ adsorption behavior and reaction pathways.

a Pulsed chemisorption of CO₂ for NTL, D-NTL/TO-2 and TO catalysts. b Top and side view of adsorbed CO₃²⁻ species on NTL, with a calculated adsorption energy of −2.72 eV. c DFT-calculated pathway for the conversion of CO₃²⁻ species to *CO over NTL. d Gibbs free energy profiles for CO₂ reduction to CO, HCOOH, HCHO, CH₃OH and CH₄ on D-NTL/TO. e Schematic illustration of the CO₂ activation mechanism on NTL and D-NTL/TO. Source data for Fig. 5a, c, d are provided as a Source Data file.

DFT calculations were conducted to clarify the intrinsic relationships among the active sites, intermediate species and underlying reaction mechanism. The top and side views of the model for the *CO₃²⁻ species upon NTL are displayed in Fig. 5b. The Gibbs free energy (ΔG) for the adsorption of *CO₃²⁻ on NTL was determined to be -2.72 eV, which explained why a greater proportion of carbonate species formed on NTL, as observed in the in-situ DRIFTS measurements. However, the transformation from *CO₃²⁻ species to *CO species was found to be significantly more challenging, with the rate-determining step exhibiting ΔG of 2.52 eV (Fig. 5c), with detail models displaying in Supplementary Fig. 37. These findings are likely responsible for the decreased photoactivity of NTL.

The adsorption energy of *CO₃²⁻ on TO was calculated to be −1.82 eV (Supplementary Fig. 38), which was higher than that on NTL (−2.72 eV). Moreover, the formation of *COOH upon TO was also thermodynamically favorable, exhibiting a ΔG of −0.16 eV (Supplementary Figs. 39-40). These theoretical results showed good consistency with the in-situ DRIFTS measurements, where characteristic vibrational bands of both *CO₃²⁻ and *COOH intermediates were detected.

Based on the in-situ DRIFTS measurements, the D-NTL/TO sample mainly contained *COOH species, which should be associated with the newly formed reaction sites. DFT calculations were performed on the hydroxyl and vacancy sites for the D-NTL/TO sample. As illustrated in Fig. 5d, the ΔG for the conversion from CO₂ to *COOH at the D-NTL/TO-OH sites were measured at −0.21 eV, which was significantly greater than the value at the defect sites, which was −1.59 eV. These observations indicated that defect sites promoted the conversion of CO₂ to *COOH species, supporting the findings from in situ DRIFTS measurements. The *COOH species should undergo a PCET process to form *CO species. The subsequent PCET reactions of *CO led to the formation of CH_xO, ultimately resulting in the generation of CH₃OH.

The ΔG for the transformation from *COOH to *CO at defect sites for the D-NTL/TO sample was only 0.3 eV, indicating that *CO formation should occur readily on this catalyst. The pathway to either CO or higher-value products was determined by the desorption or hydrogenation of *CO species. The ΔG for the conversion from *CO to CO was calculated to be 2.01 eV, whereas the ΔG for the formation of *CHO at defect sites was only 0.31 eV. These findings suggested that the formation and subsequent hydrogenation of *CO species were thermodynamically more favorable than their desorption into gaseous CO, which accounts for the lower CO product yield. The ΔG for *COOH hydrogenation to HCOOH was found to be less favorable compared to its dehydration to *CO. Similarly, the ΔG for *HCHO desorption into HCHO exceeded that for its further hydrogenation to *CH₂OH. Additionally, the ΔG for converting *CH₂OH to *CH₂ was significantly greater than that for producing CH₃OH. These theoretical results elucidated the greater selectivity for CH₃OH over CO, HCHO, HCOOH, and CH₄ in the D-NTL/TO system. The reaction intermediates at the defect sites for the D-NTL/TO sample are detailed in Supplementary Fig. 41.

By integrating experimental and theoretical investigations, it was revealed that the pristine NTL surface readily generated abundant carbonate species, which exhibited limited conversion into crucial *CO species (Fig. 5e). Acid etching treatment induced structural reconstruction, forming a D-NTL/TO heterostructure with coexisting metal vacancies and in-situ generated amorphous TiO₂. This structural evolution induced three synergistic effects: First, the coordinatively unsaturated metal sites served as electron-rich centers that strongly interacted with CO₂ molecules to promote their activation (Supplementary Fig. 42). Second, the D-NTL/TO-2 heterostructure exhibited a 2.58-fold enhancement of the built-in electric field compared to the pristine NTL, which effectively improving charge separation efficiency, thereby promoting the subsequent multi-electron reduction processes. Most importantly, this structure regulation enabled enhanced proton-coupled electron transfer, allowing each CO₂ molecule at defect sites to simultaneously receive protons and electrons. This concerted transfer process effectively suppressed the formation of unfavorable carbonate intermediates, while selectively driving the *COOH formation with thermodynamically spontaneous process. Furthermore, the stepwise proton-coupled electron transfer process stabilized *CHO/*CH₂O intermediates while inhibiting competing pathways that lead to CO, HCOH, HCOOH, and CH₄ byproducts. These combined effects ultimately achieved methanol selectivity of 99.79%. This study combined heterostructure-enhanced charge separation with defect-engineered reaction pathway control, overcoming traditional limitations in C₁ product selectivity and providing efficient insights into directional CH₃OH synthesis.

Discussion

In summary, the etching of the NiTi-layered double hydroxide facilitated the formation of metal and hydroxy vacancies, along with the in-situ generation of amorphous TiO₂, resulting in the development of a D-NTL/TO catalyst. The optimal catalyst achieved millimolar-level methanol production rates with nearly 100% selectivity without any sacrificial agents or photosensitizers. Characterization studies demonstrated that the in situ formed TiO₂ was uniformly distributed on the surface of the defective NTL, enhancing the internal electric field and promoting the efficient separation of electron-hole pairs. In situ DRIFTS and DFT calculations indicated that the vacancy sites optimized the PCET reactions and distribution of intermediate species. Notably, the formation of carbonate species was suppressed, whereas *COOH species were favored. This further facilitated the generation of *CO species that were subsequently hydrogenated to produce CH₃OH. This work presented a feasible pathway to modulate intermediate species, thereby achieving significant advancements in natural photosynthesis

Methods

Chemicals and materials

Ni(NO₃)₂·6H₂O (≥99%), tetrabutyl titanate (≥99%) and urea (≥99.5%) were purchased from Shanghai Aladdin Biochemical Technology Co. Ltd. Nitric acid was obtained from China National Pharmaceutical Group Chemical Reagent Co. Ltd. The high-purity CO₂ (99.999%), H₂ (99.9999%), Ar (99.999%), air (99.999%), and N₂ (99.999%) were purchased from Nanjing Wenda Special Gas Co. Ltd. All chemicals were used without further purification.

Synthesis of catalysts

The NiTi-LDHs (NTL) was prepared as the method below. Firstly, Ni(NO₃)₂·6H₂O (2.18 g, 7.5 mmol), tetrabutyl titanate (0.85 g, 2.5 mmol) and urea (1.65 g, 27.5 mmol) were dissolved in 100 mL of deionized water. The mixture was vigorously stirred at 85 °C for 12 h, followed by an additional stirring at 100 °C for another 12 h, with reflux in an oil bath. After cooling to 25 °C, the resulting suspension was centrifuged, washed with deionized water, and dried at 60 °C overnight to obtain NiTi-LDH, denoted as NTL. The acid etching method was employed to create vacancies sites within the NTL matrix. Initially, nitric acid was dissolved in deionized water to prepare a 0.15 mol L⁻¹ acid solution. Subsequently, the as-prepared NTL was dispersed in 20 mL acid solution and subjected to sonication for 30 min, followed by a reaction at 90 °C for 2 h in a water bath. The resultant samples, denoted as D-NTL/TO, was collected by centrifugation and dried at 60 °C overnight. The amounts of NTL using in the reaction were 1000 mg, 700 mg, 500 mg and 300 mg, resulting products designated as D-NTL/TO-1, D-NTL/TO-2, D-NTL/TO-3 and TO, respectively.

Characterization

XRD test was conducted on a Bruker D2 PHASER system with Cu Kα radiation. Raman spectra were performed on a HORIBA apparatus with model of HR EVO NANO. N₂ adsorption-desorption plots were obtained through Micromeritics ASAP 2460. TG was conducted on a Mettler Toledo 851e thermobalance system at temperature range of 50–600 °C with heating rate of 10 °C min⁻¹. EPR was carried out on a Bruker A300 spectrometers. The Ni and Ti contents were analyzed by ICP-OES (PerkinElmer Optima 2100DV). XPS tests of the catalysts were performed on a Thermo ESCALAB 250 spectrometer with a mono-chromated Al K-alpha source with a power of 150 W. The in-situ irradiated XPS measurement was performed using the same experimental setup, with the exception that a 300 W Xenon lamp was employed as the irradiation source, positioned at 20 cm from the photocatalyst surface. The morphology was analyzed via TEM (Tecnai G2 F30) and SEM (Hitachi S4700). For TEM, the sample was ultrasonically dispersed in ethanol for 10 min. Then the supernatant was dropped on a copper screen, and dried for characterization. As for SEM, the powder sample was smeared on conducting resin. AC-TEM images were conducted on a JEOL JEM-ARM300F equipment. ¹H NMR and ¹³C NMR spectra were recorded on a Bruker Biospin AVANCE III HD spectrometer operating at 600 MHz, using deuterated acetonitrile (CD₃CN) as the solvent. Ion chromatography was performed on a DIONEX ICS-600 equipment. The XAFS data for the Ni K-edge and Ti K-edge were collected at the BL11B beamline of the Shanghai Synchrotron Radiation Facility (SSRF). The corresponding samples were measured in transmission mode. UV–vis DRS was conducted on a Shimadzu UV-2600 spectrophotometer with a wavelength range of 220–800 nm using BaSO₄ as the reference. Photoelectrochemical characterizations including EIS, photocurrent and Mott-Schottky measurements were conducted by a CHI 660E workstation. PL spectra and the TRPL spectra were performed under 532 nm by using a Labram-HR800-type spectrophotometer. Details regarding the working electrode preparation and TRPL curve fitting can be found in Supplementary Note 1. Ar-TPD was conducted on an AutoChem II 2920 apparatus instrument (Micromeritics, America) coupled with an in-situ mass spectrum (Hidden). CO₂ pulse chemisorption experiments were performed on the same instrument to determine the CO₂ adsorption rates of catalysts. The evolution process of CO₂ intermediates was tested by in-situ DRIFTS, using a Nicolet IZ10 spectrometer equipped with a Harrick Scientific DRIFTS cell. More details for the CO₂ adsorption and activation experiments were found in Supplementary Note 2.

Photocatalytic performance test

The CO₂ reduction tests were performed in a sealed stainless-steel reactor featuring a top-mounted quartz window for illumination. Typically, 50 mg of photocatalysts powder was uniformly dispersed in 1 mL of water via 30-min ultrasonication to obtain a homogeneous suspension and introduced into reactor. Prior to illumination, the system was evacuated to remove residual gases, followed by the injection of high-purity CO₂ to 0.6 MPa. The reaction was performed by 300 W Xe light irradiating (425 mW cm⁻²) for 4 h at 80 °C. After the reaction, the product was detected by gas chromatography (GC 9860) equipped with flame ionization detector (FID). Moreover, the liquid-phase product after cooling was analyzed by ¹H NMR (Bruker Biospin AVANCE III HD spectrometer) operating at 600 MHz, using deuterated acetonitrile (CD₃CN) as the solvent. For sample preparation, 200 μL of the liquid-phase solution was combined with 750 μL of CD₃CN, centrifuged at 12000 rpm for 5 min, and 600 μL of the supernatant was transferred to an NMR tube. All spectra were performed at 298 K with the following parameters: 16 scans, a 30° pulse angle, 2.73 s acquisition time, a relaxation delay of 1 s, and 0.3 Hz line-broadening applied during processing. The labeled ¹³CO₂ experiment was performed by the similar process, except that the ¹³CO₂ was used.

CO₂ conversion rate

Prior to the photocatalytic CO₂ reduction reaction, high-purity CO₂ gas was introduced into a sealed 400 mL reactor and pressurized to 0.6 MPa at 25 °C. The molar quantity of CO₂ in the reactor was determined using the ideal gas law (PV = nRT), yielding approximately 0.097 mol of CO₂. Under optimal conditions, the D-NTL/TO-2 catalyst producing 194.15 μmol of CH₃OH and 1.21 μmol of CO over a 4-h reaction. Thus, the CO₂ conversion rate was calculated to be approximately 0.2% using the following equation:

$${{\rm{CO}}}_{2}\,{\rm{Conversion}}=\left(\frac{{\rm{Moles}}\,{\rm{of}}\,{{\rm{CO}}}_{2}\,{\rm{consumed}}}{{\rm{Initial}}\,{\rm{moles}}\,{\rm{of}}\,{{\rm{CO}}}_{2}}\right)\times 100\%$$

(1)

CH₃OH selectivity

The CH₃OH selectivity was calculated using the following equation:

$${\rm{Selectivity}}=\left(\frac{{{\rm{6N}}}_{{{\rm{CH}}}_{3}{\rm{OH}}}}{\sum ({{\rm{6N}}}_{{{\rm{CH}}}_{3}{\rm{OH}}}{+{\rm{2N}}}_{\rm{CO}})}\right)\times 100\%$$

(2)

where N_CH3OH is molar amount of CH₃OH produced, ∑(N_CH3OH+N_CO) is the total molar number of all detected CO₂-derived products (CH₃OH and CO). The corresponding data is showed in Supplementary Table 6.

STC conversion efficiency

The STC conversion efficiency (η%) was calculated via the following equation:

$${\rm{STC}}=\frac{[\Delta {\rm{G}}]\times (\left[\right. \Delta {{\rm{N}}}_{{\rm{product}}})}{{\rm{I}}\times {\rm{S}}\times {\rm{T}}}\times 100\%$$

(3)

where ΔG is the change of Gibbs free energy of the product relative to the reactants, ΔN is the amount of substance of the product, I is the incident light power density, S is the irradiated area, T is the reaction time.

AQY calculation

The AQY was calculated based on the following equation.

$${\rm{AQY}}(\%)=\frac{{6\times {\rm{N}}}_{{\rm{a}}} \times [{{\rm{CH}}}_{3}{{\rm{OH}}}]+2\times {\rm{N}}_{{\rm{a}}} \times [{\rm{CO}}]}{{\rm{I}}\times {\rm{A}}\times \frac{{\rm{\lambda }}}{{\rm{hc}}}\times {\rm{t}}}\times 100\%$$

(4)

where [CH₃OH] and [CO] are the molar number of CH₃OH and CO, respectively produced over time t. N_a is Avogadro’s number, I denotes the incident solar irradiance on the illuminated area A, λ is the wavelength (specifically 380, 420, 450, 500, and 550 nm), h is Planck’s constant and c is the speed of light.

Computational details

All the simulations were carried out using spin-polarized DFT with generalized gradient approximation of Perdew-Burke-Ernzerhof (PBE) as implemented in VASP code⁴⁷. The interaction between core and electrons was described using the projector-augmented wave approach with cut-off energy of 500 eV⁴⁸. A Hubbard U term was employed for the Ni and Ti 3 d orbitals (U_eff = 6.4 eV and 4.1 eV) to rectify the self-interaction error of the GGA^48,49,50. The van-der Waals interactions were corrected by the zero damping DFT-D3 method of Grimme⁵¹.

The initial Ni(OH)₂ model was obtained from the Materials Project (mp-27912) with lattice parameters a = b = 3.165 Å, c = 4.677 Å, α = β = 90°, γ = 120° in the P-3M1 space group. A 6×6×1 supercell of Ni(OH)₂ was constructed by cleaving the optimized (001) facet of the Ni(OH)₂ unit cell (post-optimization parameters: a = b = 3.157 Å, c = 8.872 Å, α = β = 90°, γ = 120°), with a 15 Å vacuum layer added along the Z-axis to minimize interlayer interactions. For the NTL model (Supplementary Fig. 43), substitution of Ni²⁺ ions with Ti⁴⁺ ions was performed on the fully optimized Ni(OH)₂ slab, with substitution ratios consistent with ICP-OES characterization results. Complementary structural analysis including XRD and HRTEM revealed the predominant presence of anatase-phase TiO₂ with (101) facet exposure, prompting the construction of a TiO₂ slab model using optimized anatase unit cell parameters (a = b = 3.776 Å, c = 9.486 Å, α = β = γ = 90°). The D-NTL/TO heterojunction was subsequently engineered through interfacial coupling between the anatase TiO₂ and the NTL model (Supplementary Fig. 44, Supplementary Data 1). Surface defects were intentionally introduced into the longitudinal NTL-TO heterojunction structure to mirror experimental defect configurations identified in HRTEM and XAFS characterizations.

The periodic boundary condition was employed in all directions. Before the calculation, a reasonable initial magnetism was set in the INCAR file, and the result file was carefully checked to ensure that the total magnetic moment was reasonable.

The work function (Φ) was defined as below:

$$\Phi=-{\rm{e}}*{\rm{ESP}}-{{\rm{E}}}_{{\rm{Fermi}}}$$

(5)

where −e is the charge of an electron, ESP is the electrostatic potential in the vacuum nearby the surface, and E_Fermi is the Fermi level inside the material, which is confirmed by calculating the projected density of states (PDOS) data.

To fully explore the energy changes along the reaction coordinate for different sites, the energy of the intermediate structure of CO₂ reduction was calculated using VASP. To ensure reasonable results, the INCAR file was set with an appropriate initial magnetism, and the result file was carefully checked to confirm that the total magnetic moment was reasonable. The ΔG was calculated as the difference between the Gibbs free energy of the reactant and product, which was determined using the equations provided in Supplementary Table 8.

The ΔG in the generation of H₂O was calculated as follows:

$$\Delta {\rm{G}}={{\rm{G}}}_{{\rm{product}}}+{{\rm{G}}}_{{\rm{H}}2{\rm{O}}}-{{\rm{G}}}_{{\rm{reactant}}}-{{\rm{G}}}_{{\rm{H}}}+{\rm{kT}}\mathrm{ln}10 \cdot {\rm{pH}}-{\rm{eU}}$$

(6)

The ΔG in the reduction of the reactant was calculated as follows:

$$\Delta {\rm{G}}={{\rm{G}}}_{{\rm{product}}}-{{\rm{G}}}_{{\rm{reactant}}}-{{\rm{G}}}_{{\rm{H}}}+{\rm{kT}}\mathrm{ln}10 \cdot {\rm{pH}}-{\rm{eU}}$$

(7)

The computational standard hydrogen electrode (SHE) approximation was applied so that the Gibbs free energy of H· could be replaced with the Gibbs free energy of 0.5 H₂ at U = 0 V vs SHE. The pH and bias terms eU and kTln10·pH were both set to 0, as the Gibbs free energy analysis was performed under standard conditions (pH = 0, T = 298.15 K) and U = 0. Therefore, the simplified equation was as below.

The ΔG in the generation of H₂O was calculated as follows:

$$\Delta {\rm{G}}={{\rm{G}}}_{{\rm{product}}}+{{\rm{G}}}_{{\rm{H}}2{\rm{O}}}-{{\rm{G}}}_{{\rm{reactant}}}-{1/2{\rm{G}}}_{{\rm{H}}2}$$

(8)

The ΔG in the reduction of reactant was calculated as follows:

$$\Delta {\rm{G}}={{\rm{G}}}_{{\rm{product}}}-{{\rm{G}}}_{{\rm{reactant}}}-{1/2{\rm{G}}}_{{\rm{H}}2}$$

(9)

All the calculation models in this work were visualized by VESTA (Version 3.5.8). The input file and result processing were done using the vaspkit package⁵².