Atom manufacturing of photocatalyst towards solar CO₂ reduction

Among various vacancies species, V_Os are most easily formed in various oxygen-containing compounds due to their relatively low formation energies. Because the CO₂ molecule can be preferentially adsorbed by these V_Os by filling the oxygen atom of CO₂ into V_Os, therefore V_Os are widely investigated to promote photocatalytic CO₂ conversion over various photocatalysts [82, 83]. For instance, Zhao et alprepared V_O-rich ZnAl-layered double hydroxide (ZnAl-LDH) nanosheets by reducing the lateral size of ZnAl-LDH from 5 μm to 30 nm, followed by lessening the nanosheet thickness into a few layers via a bottom-up process of inverse microemulsion and subsequently controllable hydrolysis (figure 4(a)) [84]. The stable reverse emulsion was composed by sodium dodecylsulfate, isooctane, and water with 1-butanol as a co-surfactant. Benefitting from the decreased lateral size, V_Os were successfully introduced into ZnAl-LDH, which could not only result in abundant coordinatively unsaturated Zn ions (denoted as Zn⁺-V_O) but also serve as electron trapping sites accompanying by offering a new defect energy level, thereby promoting the separation and transfer of photogenerated charge carriers. Particularly, the V_O-rich ZnAl-LDH exhibited more favorable for CO₂ and H₂O adsorption because of the higher adsorption energy for both CO₂ and H₂O than those of defect-free LDH (figure 4(b)). In addition, density functional theory (DFT) calculations revealed a concentrated charge density at the Zn atoms, indicating that the active electrons were generated from Zn⁺-V_O defective site in LDH, which was good for facilitating the interaction between active electrons and the reactant, thereby improving the efficiency of photocatalytic CO₂RR. Based on the above virtues, V_O-rich ZnAl-LDH (denoted as ZnAl-1) delivered a superior CO formation rate of 7.6 μmol g⁻¹ h⁻¹, much higher than that of the bulk counterpart (ZnAl-3, ∼0 μmol g⁻¹ h⁻¹) under the same operation condition (figure 4(c)). Besides bottom-up processing, top-down approach is an alternative to synthesis nanosheet architecture, which is susceptive to introducing vacancy-type defects. Using BiOBr-based lamellar precursors, Wu et al synthesized BiOBr atomic layers via ultrasonication for 24 h, which were further treated by ultraviolet irradiation to introduce V_Os on BiOBr (figure 4(d)) [82]. The introduced V_Os were favorable for improving both light harvesting and separation of photo-generated electron–hole pairs.

Zoom In Zoom Out Reset image size

Furthermore, the negatively charged surface of V_O modified BiOBr atomic layers was beneficial for CO₂ adsorption compared with pristine BiOBr atomic layers and bulk sample. Benefited from the aforementioned advantages, the V_O–BiOBr atomic layers exhibited a superior CO evolution rate of about 87.4 μmol g⁻¹ h⁻¹, which was much higher than those of pristine atomic layers and bulk counterpart (figure 4(e)). In situ Fourier-transform infrared spectroscopy was performed to identify photocatalytic CO₂RR process, revealing that the presence of V_Os was conductive to stabilize the COOH* intermediate according to the more COOH* groups on V_O–BiOBr, which could effectively decrease the activation energy of CO₂ and promote CO evolution.

The aforementioned methods suffer from tedious soft-template process, which goes against large-scale preparation of photocatalysts. In contrast, one-pot thermal treatment is a promising strategy to fabricate photocatalysts on mass production. Shi and coworkers developed a molten-salt way to synthesize B-doped BiOCl using B₂O₃ as both molten salt and doping precursor (figure 4(f)) [85]. The resultant sample possessed a homogeneous B-doping feature from the surface to bulk, accompanied by the formation of V_Os. Surface B doping could induce surface reconstruction of BiOCl, resulting in V_O formation by extracting adjacent hydroxyl, thereby bringing intimate B–V_O associates (BiOCl–B–V_O). Photocatalytic CO₂RR test showed that the BiOCl presented the highest CO evolution rate of 83.64 μmol g⁻¹ h⁻¹ without any scavengers or cocatalysts (figure 4(g)), which was attributed to the synergistic effect between B doping and V_O. More importantly, BiOCl–B–V_O sample possessed a high selectivity of more than 98% for CO generation, much exceeding those of BiOCl and BiOCl–V_O. According to corresponding DFT calculation, the B–V_O associates could lengthen the C–O bond and contribute to a CO₂ adsorption energy of −0.89 eV, resulting in efficient CO₂ activation. For CO₂ to CO conversion, the formation of *COOH typically requires high energy barriers and has been regarded as the rate-limiting step. Compared with an individual V_O, B–V_O associates delivered a much lower energy barrier of 1.14 eV to form *COOH, which was subsequently dissociated with a proton to develop *CO and H₂O (figure 4(h)). Afterwards, *CO was released from BiOCl–B–V_O surface rather than further hydrogenation to form CHO* for CH₄ generation because of the exothermic of *CO desorption. This suggested CH₄ formation was not a favorable pathway over BiOCl–B–V_O sample. Additionally, the competitive hydrogen generation was also largely suppressed. Therefore, the introduced B–V_O associates could mediate BiOCl for both enhancing photocatalytic CO₂RR activity and selectivity, restrict the competitive hydrogen evolution, and improve the stabilization of *COOH for selective CO evolution (figure 4(i)).

Considering the less energetically demands of CH₄ and CO, photocatalytic CO₂RR into liquid products, such as methanol and formic acid, is more sustainable and valuable for CO₂ reutilization. Rational atom manufacturing of suitable semiconductors can achieve CO₂ activation towards liquid solar fuels. Hou et al proposed the V_Os confined in phosphate-doped Bi₂WO₆ atomic layers (V_O–PO₄–BWO) as an ideal photocatalyst for triggering CO₂ conversion, which was synthesized by thermal treatment of PO₄ oxoanion doped Bi₂WO₆ under H₂/Ar mixture gas [86]. Particularly, some pits could be distinctly observed in the HRTEM image (figure 5(a)), suggesting the successful elimination of the oxygen atom connecting with the Bi atom and demonstrating the formation of V_Os confined in PO₄–BWO layers (figure 5(b)). In addition, the stronger EPR signal of V_O–PO₄–BWO further confirmed the introduction of V_Os, which was beneficial for reducing the bandgap and enhancing light absorption property. As a result, the V_O–PO₄–BWO delivered a methanol yield rate of 157 μmol g⁻¹ h⁻¹, outperforming those of V_O–BWO, PO₄–BWO, BWO atomic layers, and bulk BWO (figure 5(c)). The superior performance for photocatalytic CO₂ conversion of V_O–PO₄–BWO could be ascribed to both the PO₄ doping and introduced V_Os. The presence of V_Os could trap the photogenerated electrons in the CB, thereby bringing about the effective charge transfer and separation over the V_O–PO₄–BWO sample. In principle, photocatalytic CO₂RR usually follows a multi-electron transfer pathway with proton assistance to overcome the high energies, thus suppressing the formation of carbonate radical (CO₂ ⁻) to generate long-chain chemicals with more than three carbons. Rational design of V_Os also offers an alternative strategy to allow CO₂ activation through a single-electron process to the formation of long-chain chemicals. For instance, Chen et al prepared Bi₂O₃ atomic layers through in situ oxidation of exfoliated Bi nanosheets by use of its high surface energy (figure 5(d)) [68]. After that, V_Os could be successfully introduced by ultrasonication treatment. Interestingly, the short oxidation time promoted the formation of higher concentration of V_Os than that of long oxidation time. The distinct lattice disorder originated from V_Os induced by the unsaturated coordination of Bi can be clearly observed from HAADF-STEM image, confirming that V_O defective structures were successfully developed in Bi₂O₃. Moreover, the V_O-rich-Bi₂O₃ exhibited a lower photoluminescence (PL) emission intensity than that of V_O-poor-Bi₂O₃, indicating that the V_Os could suppress the recombination of photogenerated electrons and holes. Subsequently, the photocatalytic CO₂ and CH₃OH conversion to dimethyl carbonate were carried out and showed a conversion yield of above 18% over V_O-rich-Bi₂O₃, which was nine-fold higher than that of V_O-poor-Bi₂O₃, implying the V_Os provided additional active sites for triggering CO₂ reduction (figure 5(e)). The V_Os confined in the surface of Bi₂O₃ nanosheets could supply abundant unsaturated coordination sites for CO₂ adsorption, which was further activated by the localized electrons on the V_Os to the generation of •CO₂ ⁻. Meanwhile, CH₃OH was also inclined to be adsorbed on V_O-rich-Bi₂O₃ nanosheets with more negative adsorption energy than that of vacancy-free counterpart (figure 5(f)). Benefiting from the above advantages associated with the improved separation efficiency of photogenerated carriers, the V_O-rich-Bi₂O₃ nanosheets could give a superior CO₂ conversion performance toward dimethyl carbonate with nearly 100% selectivity.

Zoom In Zoom Out Reset image size

Besides V_Os, other types of anionic vacancies have also been widely studied to enhance the corresponding photocatalytic performance toward CO₂ conversions, such as nitrogen vacancies (V_Ns), sulfur vacancies (V_Ss), and carbon vacancies (V_Cs). For instance, Tu et al prepared V_Ns modified PCN nanosheets for photocatalytic CO₂ conversion [45]. The density of V_Ns could be tuned by varying the post-treatment temperature at 475, 500, 525, and 550 °C. In comparison with the negligible EPR signal of pristine bulk PCN, the intensity of the EPR signal gradually enhanced as the post-processing temperature increased, indicating the improvement of two-coordinated V_Ns confined in PCN frameworks, which could offer unpaired electrons to adjacent sp² carbon atoms with aromatic rings of PCN. The photocatalytic CO₂RR tests demonstrated that the CO generation rate was enhanced as the increase of V_Ns from bulk pristine PCN to PCN-525, delivering a superior CO evolution rate of 6.21 μmol h⁻¹ for PCN-525, which was nearly 4.2 folds higher than that of bulk PCN. Noticeably, an excessive V_Ns could result in a slightly inferior CO generation rate (PCN-550), which might be attributed to the recombination of photo-generated electrons and holes. Benefiting from the introduced V_Ns, the formed middle bandgap states could not only promote the excitation of more photo-generated electrons and holes, but also serve as a temporary reservoir for trapping photo-generated electrons, thereby accelerating the separation of photo-generated electrons and holes for favoring the remarkable photocatalytic activity.

Given the role of V_Ns as recombination centers for photo-generated carriers, the integration of dopants and V_Ns may be an effective alternative strategy for avoiding drawbacks of V_Ns. For this matter, Wang et al developed a B, K co-doped PCN associated with controllable V_Ns over thermal polymerization of dicyandiamide with KBH₄ (KBH–PCN), in which the K doped in vacancy site could promote the electron transfer between CO₂ and PCN [87]. Meanwhile, the B dopants were helpful for keeping the necessary reduction potential for triggering CO₂ photoreduction. Moreover, the presence of V_Ns could lower the bandgap for boosting light-harvest capability. The CO₂ temperature-programmed desorption results elucidated that the optimized 3% KBH–PCN delivered the superior CO₂ adsorption capability with higher desorption temperature, which could be attributed to both the improved number and strength of Lewis basic sites. According to the density of states (DOS) of PCN and KBH–PCN, the K and B dopants did not significantly participate in developing the middle gap states. Photocatalytic CO₂ reduction with H₂O demonstrated that the 3% KBH–PCN offered the highest CO₂ conversion rates of 5.93 and 3.16 μmol g⁻¹ for CH₄ and CO production, respectively, after 5 h irradiation, which were approximately 1.61 and 5.27 folds of pristine PCN (figure 6(a)).

Zoom In Zoom Out Reset image size

Besides V_Ns, V_Cs can be also engineered in PCN skeleton, which could serve as adsorption sites for capturing CO₂ because of the left unsaturated nitrogen, enabling the activation of CO₂. For instance, Yang et al introduced V_Cs on PCN using a constant temperature steam etching approach to selectively remove carbon atoms of PCN skeleton [88]. After that, the surface C/N ratio of PCN decreased from 0.859 to 0.705, indicating the successful removal of carbon atoms toward V_C creation. The positron annihilation spectroscopy (PAS) was employed to reveal the surface and bulk defects in materials. In comparison with pristine PCN, the PCN-500-4 possessed an increased τ2 value, suggesting that numerous V_Cs were presented onto the PCN surface. Correspondingly, the PCN-500-4 sample also delivered an enhanced EPR signal originated from the unpaired electrons around the V_Cs. In addition, the time-resolved PL spectra revealed a longer lifetime of the photo-generated carriers after V_Cs introduction, implying that the V_Cs could capture the photo-generated electrons and thereby prolong the life-span of charges for participation in CO₂ conversion. The optimized V_C-modified PCN gave a CO evolution rate of 21.5 μmol h⁻¹, which was much higher than that of pristine PCN (1.2 μmol h⁻¹). DFT calculations demonstrated that *CO₂ was preferentially trapped in the location near the two NH₂ groups generated by carbon removal with no obvious geometric differences. The required energy for the rate-determining step of the hydrogenation of *CO₂ to *COOH could be lowered by about 0.15 eV in V_C modified PCN, thereby promoting photocatalytic CO₂ reduction.

Transition metal sulfide is a broad class of layered structures that possess unique optical and thermoelectric properties, which can be exfoliated into atomically 2D nanosheet structure via atom manufacturing. Using Cu₂O as a precursor, Wang et al synthesized sulfur vacancy modified Cu₃SnS₄ (V_S–Cu₃SnS₄) via a simple hydrothermal treatment [89]. A series of characterizations confirmed the successful introduction of V_S in Cu₃SnS₄, which promoted electron accumulation on surrounding Sn(II) atoms. Accordingly, the CB position negatively shifted with enhanced reduction capability of electrons. As a result, the optimized V_S–Cu₃SnS₄ gave a CH₄ generation rate of 22.65 μmol g⁻¹ h⁻¹ with about 83.1% selectivity. Compared with CH₄, C2 products are more valuable for CO₂ conversion, but still remains a great challenge. To this end, the AgInP₂S₆ single atomic layers (AgInP₂S₆ SAL) was fabricated using ultrasound exfoliation of the bulk counterpart, in which V_Ss could be successfully introduced by etching process in H₂O₂ solution [90]. Photocatalytic CO₂RR was performed with water vapor under light irradiation. For both bulk sample and vacancy-free AgInP₂S₆ SAL, CO was the major product. A small amount of C₂H₄ was also detected for AgInP₂S₆ SAL, which was formed via a 12 electrons and 12 protons process. After the rational introduction of V_Ss, C₂H₄ became the dominant product with a yield of 44.3 μmol g⁻¹ associated with an impressive selectivity of ∼73% (figure 6(b)). In the meantime, both CO and CH₄ were also detected but at lower concentrations compared with vacancy-free SALs, suggesting that the presence of V_Ss in SALs preferentially enabled C–C coupling process between C1 intermediates toward C2 generation rather than isolated them into C1 gases. DFT calculations were carried out to investigate the catalytic selectivity mechanism toward C₂H₄ on V_S–AgInP₂S₆ surface. It is well known that the CO₂ hydrogenation to generate *COOH is the key step for CO evolution. Further coupling a pair of proton/electron with *COOH promoted the formation of CO and H₂O. For vacancy-free AgInP₂S₆, the generated *CO suffered from a low adsorption energy of −0.07 eV, indicating a facile extraction of *CO from catalyst surface to become CO gas and thereby presenting a high CO selectivity. After introducing V_Ss, the exposed Ag atom around the V_S exhibited charge accumulation, which could anchor the *CO via chemical adsorption way with an adsorption energy of −0.25 eV. On one hand, the adsorbed *CO could be further protonated to generate a series of intermediates to participate in the following reaction. On the other hand, other generated *CO on catalyst surface diffused to V_Ss and coupled with these intermediates to form C₂H₄. As shown in figure 6(c), three possible unsaturated reaction intermediates were proposed. In comparison with other two coupling processes, *CO–CHOH coupling possessed a relatively lower energy barrier of 0.84 eV. Therefore, C₂H₄ evolution would follow *CO–CHOH coupling associated with hydrogenation. Notably, the protonation of *CO toward *COH was the rate-determining step during C₂H₄ generation process based on the whole energy diagram (figure 6(d)).

Similar to anionic vacancies, cationic vacancies can also bring about significant electronic structure variation with respect to perfect ones, which is beneficial for improving the corresponding catalytic activities. Particularly, cationic vacancies seem to be preferable compared with anionic vacancies because of the susceptible deactivation of anionic vacancies induced by oxygen species from the environment. Di et al prepared BiOBr ultrathin nanosheets with tunable Bi vacancies (V_Bi) using controllable ionic liquid-assisted synthesis under ambient conditions [91]. The aberration-corrected STEM-HAADF was employed to uncover the surface microstructure of V_Bi–BiOBr nanosheets (figure 7(a)), demonstrating that partial surface Bi atoms were missing, which confirmed the successful formation of V_Bi. On the contrary, the BiOBr nanosheets sample exhibited an ordered lattice and surface atomic configuration. V_Bi–BiOBr nanosheets delivered an apparently red-shifted absorption edge compared to BiOBr nanosheets. In particular, benefiting from the Bi vacancies, the V_Bi–BiOBr nanosheets displayed a tail peak range from 430 nm to 500 nm. The bandgaps of V_Bi–BiOBr and BiOBr nanosheets were estimated to be 2.65 eV and 2.77 eV, respectively. Combined with x-ray photoelectron spectroscopy (XPS) valence band (VB) spectra, the VB and CB potentials of V_Bi–BiOBr were shifted from 2.01 eV and −0.76 eV to 1.57 eV and −1.08 eV (vs NHE at pH = 7), respectively, compared with BiOBr nanosheets (figure 7(b)). The negatively shifted CB potential of V_Bi–BiOBr contributed to higher reducibility of photo-generated electrons and thus favored the photocatalytic CO₂ conversion process. The V_Bi–BiOBr delivered a photocatalytic CO evolution rate of 20.1 μmol g⁻¹ h⁻¹ with a 98% selectivity, which was almost 3.8 folds higher than CO generation rate of BiOBr nanosheets (figure 7(c)). The enhanced photocatalytic activity could be attributed to the association of the tunable electronic structure, optimized CO₂ adsorption/activation, and CO desorption process for V_Bi–BiOBr sample. It should be noted that the V_Bi–BiOBr still retained an efficient photocatalytic activity after fourth cycling tests, implying its superior stability.

Zoom In Zoom Out Reset image size

Zn vacancies (V_Zn) are another case of common cationic vacancies. For instance, Jiao et al successfully synthesized one-unit-cell thick ZnIn₂S₄ layers with tunable vacancy concentrations through a simple hydrothermal way. TEM and atomic force microscope (AFM) results verified the sheet-like morphology with an average thickness of about 2.46 nm. The Zn vacancies could be well observed in the HAADF-STEM image shown in figure 7(d), and the corresponding line intensity profiles confirmed that the vacancies were V_Zn (figure 7(e)) [92]. EPR and PAS further manifested the successful preparation of one-unit-cell ZnIn₂S₄ atomic layers with distinct V_Zn concentrations. Benefiting from the presence of V_Zn, the V_Zn-rich one-unit-cell ZnIn₂S₄ layers possessed the distinctly enhanced carrier separation efficiency compared with the V_Zn-poor sample, as demonstrated by surface photovoltage (SPV) spectroscopy in figure 7(f). Furthermore, the abundant V_Zn also favored light-harvesting, CO₂ adsorption capability, and improved surface hydrophilicity, which could promote the photocatalytic performance of ZnIn₂S₄. As a result, the V_Zn-rich one-unit-cell ZnIn₂S₄ layers delivered the photocatalytic CO₂RR toward CO formation rate of 33.2 μmol g⁻¹ h⁻¹, which was about 3.6-fold higher than that of V_Zn-poor counterparts, signifying the vital role of V_Zn in fully optimizing the photocatalytic performance. The corresponding DFT calculations unveiled that the V_Zn could not only largely enrich charge density at the adjacent sulfur atoms to enable the electrons photoexcitation easier but also serve as electron capture centers to promote charge separation. Of note, the presence of abundant V_Zn could endow a more negatively charged surface, which potentially improved the CO₂ adsorption capabilities. ZnS is also a promising photocatalyst for CO₂ conversion with a relatively high CB position at −1.04 V (vs NHE at pH = 7). However, highly crystalline ZnS seriously suffers from insufficient active sites, resulting in a thwarted photocatalytic CO₂ conversion capability. Pang et al created V_Zn on the surface of ZnS via an acid-etching strategy, which could effectively avoid the adverse influence on photo-generated electron transfer induced by the bulk defects [93]. As DFT calculation shown in figure 7(g), the S atoms in the V_Zn vicinity presented the lower charge density relative to pristine ZnS based on (001) surface slabs. The steady-state PL spectroscopy was also conducted and revealed that the peak intensities at 580 nm, which represented the donor–acceptor luminescence band associated with V_Zn, increased as the acid increasing, implying more V_Zn were formed by stronger acidic etching (figure 7(h)). Subsequently, photocatalytic CO₂ conversion was explored without any cocatalyst. As shown in figure 7(i), for pristine commercial ZnS, only H₂ and a small quantity of CO could be monitored without any liquid product. In contrast, the V_Zn modified ZnS processed by various sulfuric acid delivered significantly improved photocatalytic activity with formate as the main product when the pH of the acid solution was adjusted to 1.2. Particularly, a selectivity up to 86.6% for HCOOH could be achieved for acid-treated ZnS (pH = 0.2). In situ attenuated total reflection-infrared spectra were collected to identify CO₂RR over V_Zn–ZnS, demonstrating the generation of HCOO* under light irradiation. And then HCOO⁻ was formed after the second electron transfer, which subsequently desorbed from V_Zn–ZnS surface. DFT calculations clarified that the adsorption energy of CO₂ was lower than that of the competitive proton, suggesting that the presence of V_Zn was not favorable for hydrogen evolution. Furthermore, compared with the higher adsorption energy of *CO intermediate that promoted CO generation, the photocatalytic CO₂RR towards formate possessed a more facile energy barrier.

By varying the temperature of hydrothermal treatment, Gao et al successfully tuned the vanadium vacancy (V_V) density on orthorhombic-BiVO₄ (o-BiVO₄) atomic layers with the single-unit-cell thickness (figure 8(a)) [94]. During the process, the BiVO₄-based lamellar structures, which was formed by the reaction between the added VO₄ ³⁻ and the Bi³⁺ of the BiCl₄ ⁻-CTA⁺ (cetyltrimethylammonium) lamellar hybrids, could spontaneously exfoliate into o-BiVO₄ ultrathin nanosheets. The AFM image shown in figure 8(b) disclosed an average thickness of 1.28 nm, which was consistent with a unit cell thickness along the [001] direction. PAS was also conducted to relatively quantify the V_V concentration. As shown in figure 8(c), the V_V-rich o-BiVO₄ sample possessed a longer positron annihilation time at V_V and higher V_V density than those of V_V-poor o-BiVO₄, resulting in a new defect level for enhancing light harvesting. In addition, SPV spectroscopy was performed to unveil the dynamic features of photogenerated carriers, which delivered an improved SPV intensity for V_V-rich o-BiVO₄, ensuring a higher separation efficiency of photogenerated electrons and holes, which was further verified by the time-resolved fluorescence spectra. Benefiting from the above virtues, the V_V-rich o-BiVO₄ ultrathin nanosheets showed a CH₃OH generation rate of 398.3 μmol g⁻¹ h⁻¹, which was about 1.4 times higher than that of V_V-poor o-BiVO₄ (figure 8(d)), suggesting the pivotal role in promoting the photocatalytic performance of Vv. Furthermore, the V_V-rich o-BiVO₄ also exhibited a higher apparent quantum efficiency with a maximum of 5.96% at 350 nm compared with the V_V-poor o-BiVO₄ sample (figure 8(e)).

Zoom In Zoom Out Reset image size

In addition, taking advantage of the middle energy level created by vacancies, direct utilization of wide bandgap semiconductors as photocatalysts for CO₂ conversion becomes possible. In this regard, Hu and coworkers prepared a series of In(OH)₃ with a large number of indium vacancies (V_In) via a simple hydrothermal way using mixed solvent comprised of ethylenediamine and water, of which the volume ratio of ethylenediamine/water was varied from 0/1 to 1/1, 2/1, 3/1, 4/1 and 5/1 [95]. The HRTEM image shown in figure 8(f) obviously indicated indiscernible lattice fringes in the circle regions originated from the presence of vacancies. On the contrary, the In(OH)₃ synthesized using water as the solvent delivered a smooth surface with a cubic structure. UV–Vis absorption spectra were collected to investigate the effects of vacancies on the light-harvesting capability. As depicted in figure 8(g), the 0/1-In(OH)₃ sample showed single absorption at the ultraviolet region with a calculated bandgap of more than 5.2 eV. After introducing V_In, the In(OH)₃ samples not only possessed improved light absorption in the range of 260–350 nm and presented a crucial tailing absorption at the wavelength ranging from 480 nm to 650 nm, but also possessed CO₂ chemisorption capability, which was enhanced as the vacancy content increased. Furthermore, the photocatalytic CO₂ conversion capabilities were determined under visible light irradiation at 200 °C using water vapor as a reducing agent, illustrating that all V_In-modified samples delivered much higher H₂, CO and CH₄ production rates than that of pristine In(OH)₃. Among them, the 4/1-In(OH)₃ sample processed the highest photocatalytic CO₂ conversion activity with superior cycling stability. Benefiting from the middle energy level induced by V_In, the In(OH)₃ could be excited by visible light over a two-step process. The electrons in VB could be excited to defect energy level, which was further excited to the CB of In(OH)₃ for CO₂ conversion, leaving photo-generated holes on the VB for water oxidation (figure 8(h)).

Similar to the multiple heteroatom dopants to offer more catalytic sites relative to single ones, the combination of two or more types of vacancies would be an intriguing way to promote photocatalytic performance owing to the existence of more active sites and the synergistic effects between multiatomic vacancies. Therefore, exploring the association of various vacancies is extremely significant and imperative. Li et al reported dual anionic vacancies of oxygen and chlorine in BiOCl (BiO_1−x Cl_1−y ) via a simple hydrolysis approach using the extracting solutions separately derived from green tea and spinach as reductant, denoted as BiO_1−x Cl_1−y -T and BiO_1−x Cl_1−y -S, respectively [96]. Figure 9(a) showed the Fourier transformed profile of x-ray absorption fine structure (XAFS), which revealed that BiOCl-T delivered a shorter Bi–O bond length relative to pristine BiOCl derived from the lattice local contraction of the V_Os, while the Bi–Cl bond length was longer than that of BiOCl due to the Cl vacancies. Furthermore, the pristine BiOCl exhibited no EPR signals because of no defects in BiOCl. In contrast, both BiOCl-T and BiOCl-S samples exhibited an obvious signal peak at g = 2.0, which was attributed to V_Os. Especially, the peak intensity of BiOCl-S was stronger than that of BiOCl-T, validating a higher V_Os content in BiOCl-S. UV–Vis absorption spectroscopy was performed to reveal the optical harvest capabilities of samples, displaying that BiOCl-T and BiOCl-S gave visible responses extending to 500 nm, whereas the pristine BiOCl only possessed slight absorption in the visible region (figure 9(b)). It should be noted that both BiOCl-T and BiOCl-S had two new absorption peaks centered at 430 nm and 660 nm due to the formation of two defect energy levels. Base on the above, the calculated bandgaps of BiOCl, BiOCl-T, and BiOCl-S were 3.07 eV, 2.66 eV, and 2.31 eV, respectively. With dual vacancies introduced, both BiOCl-T and BiOCl-S samples presented largely boosted photocatalytic CO₂RR toward CO, even the CH₄ production rate decreased. The CO selectivity for BiOCl-T and BiOCl-S were 2.2 and 2.5 folds higher than that of pristine BiOCl, which could be attributed to the presence of V_Os and Cl vacancies (figure 9(c)). DFT calculation demonstrated that the introduced dual vacancies could efficiently tune the electronic structure of BiOCl, resulting in downward shifted position of CB and VB with a decreased bandgap.

Zoom In Zoom Out Reset image size

In another study, Yang et al constructed a Z-scheme heterostructure photocatalyst with interfacial dual vacancies, of which the V_Ns and V_Os were originated from PCN and TiO₂, respectively [97]. The SEM image shown in figure 9(d) clearly revealed V_Os–TiO₂ nanotube arrays with V_Ns–PCN intimate deposition. The room temperature ESR spectra demonstrated an apparently enhanced ESR signal for V_Ns–PCN/V_Os–TiO₂ compared with PCN/TiO₂ without vacancies (figure 9(e)). Of note, the interaction between V_Os and V_Ns might deliver a slightly reduced ESR signal with respect to V_Ns–PCN/TiO₂. Furthermore, the low-temperature ESR spectra exhibited a distinct signal at g = 1.97 (inset in figure 9(e)), confirming the existence of V_Os in V_Ns–PCN/V_Os–TiO₂. Benefits from the interfacial dual vacancies, the V_Ns–PCN/V_Os–TiO₂ possessed a broad light adsorption band edge in the range of 300–800 nm, accompanied by a superior visible light absorption intensity, indicating the alterant band structure of PCN and TiO₂. With the emergence of dual vacancies, the as-prepared sample achieved visible light yields of 13.4 and 3.0 μmol g⁻¹ for CO and CH₄, respectively, significantly outperforming the performance of pristine PCN/TiO₂. In addition, the dual vacancies modified photocatalyst also possessed good stability after four cycling tests (figure 9(f)).

In contrast to dual anionic vacancies, Di et al prepared freestanding Bi₂MoO₆ ultrathin nanosheets with surface 'Bi–O' vacancy pairs using hydrothermal way assisted with polyvinyl pyrrolidone as a surfactant [98]. During the process, layered BiOBr ultrathin nanosheets were initially formed, which served as a sacrificial template to guide the fabrication of Bi₂MoO₆ nanosheets over the ion-exchange approach (figure 10(a)). The aberration-corrected STEM-HAADF image clearly showed the dim sites of Bi associated vacancies. Surface V_Os could be confirmed by low-temperature EPR analysis with the typical signal at g = 2.0. It is well accepted that the vacancies could serve as centers for reserving photo-generated electrons, thereby promoting the separation of photo-generated charge carriers. As displayed in figure 10(b), the time-resolved fluorescence spectra revealed that the Bi₂MoO₆ ultrathin nanosheets with 'Bi–O' vacancy pairs possessed a longer radiative lifetime of photo-generated charge carriers than that of bulk Bi₂MoO₆, which could be ascribed to the faster charge transfer derived from the atomically ultrathin framework as well as dual vacancies. Profiting from the successful introduction of dual vacancy, the Bi₂MoO₆ ultrathin nanosheets with 'Bi–O' vacancy pairs showed an absorption tail that extending to ∼520 nm, which was further verified by DFT calculations. As shown in figure 10(c), the enhanced DOS at the VB edge, as well as newly formed DOS in the forbidden band, could be observed in 'Bi–O' vacancy modified Bi₂MoO₆, implying the higher carrier concentration and much easier excitation of electrons into CB. As a result, the 'Bi–O' vacancy modified Bi₂MoO₆ delivered a CO evolution rate of 3.62 μmol h⁻¹ g⁻¹, which was much higher than that of the bulk counterpart (1.42 μmol h⁻¹ g⁻¹). In addition, the defective sample also had excellent cycling stability originated from no obvious deterioration after three cycle tests.

Zoom In Zoom Out Reset image size

The accurate control of dual vacancies concentration is also an effective strategy to mediate the selectivity of photocatalytic CO₂RR. Tan et al synthesized a series of Ni–Al layered double hydroxide (NiAl-LDH) with various thicknesses ranging from 27 nm (b-NiAl-LDH) to 5 nm (f-NiAl-LDH) and 1 nm (m-NiAl-LDH) [99]. The XAFS displayed in figure 10(d) revealed the decreased energy locations of Ni K-edge in the arrangement of b-NiAl-LDH, f-NiAl-LDH, and m-NiAl-LDH, showing a reduced average oxidation state of +1.6 for Ni in m-NiAl-LDH, which was further confirmed by soft XAS tests. The coordination number of Ni was analyzed and exhibited decrease of the first Ni–O shell along the sequence of b-NiAl-LDH, f-NiAl-LDH, and m-NiAl-LDH, manifesting a slightly greater degree of structural distortion as the thickness decreases, which might be induced by the hydroxyl defect. In addition, the PAS was also performed and further verified that the m-NiAl-LDH possessed the most surface defects in three samples. Resulted from the abundant surface vacancies, the m-NiAl-LDH exhibited a lowered bandgap of 1.78 eV, which promoted light harvesting in a wider spectrum. By using NiAl-LDH samples as photocatalysts with the addition of Ru(bpy)₃Cl₂•6H₂O under visible light above 400 nm, the selectivity of CH₄ could reach 16.54% for m-NiAl-LDH compared with the 0.11% of b-NiAl-LDH. While the H₂ evolution was validly suppressed from a selectivity of 43.84% for b-NiAl-LDH to 13.26% for m-NiAl-LDH. Furthermore, the m-NiAl-LDH possessed CH₄ and CO yields of 168 μmol h⁻¹ g⁻¹ and 712 μmol h⁻¹ g⁻¹, respectively, greatly surpassing the performance of b-NiAl-LDH. More intriguingly, when the light irradiation was extended to above 600 nm, the CH₄ selectivity can be enhanced to 70.3% with the completely suppressive H₂ evolution (figures 10(e)–(f)). Whereas, the selectivity of H₂ was still about 18% for b-NiAl-LDH. To unveil the correlation between the selectivity and light wavelength of m-NiAl-LDH, DFT with Hubbard corrected calculations were conducted and demonstrated that the synergistic effect derived from metal and hydroxyl defects could effectively reduce the Gibbs free energy barrier of CO₂RR toward CH₄ (0.127 eV) with a driving force of 0.313 eV.

In summary, vacancy engineering is an effective strategy to tune photocatalytic performance by offering abundant active sites, altering reaction mechanisms or pathways to boost the catalytic activity. There are mainly three types of vacancies, anionic, cationic, and dual vacancies, which could be developed to optimize the electronic structure and effectively accelerate photocatalytic CO₂ conversion rate. Although many efforts have been devoted to engineering vacancies for enhancing photocatalytic CO₂, the current challenges and barriers, such as effectively engineering strategies and remarkably catalytic stability, are still urgent to be dealt with. In addition, the relationship between vacancies and photocatalytic activities remains ambiguous, which limits the further enhancement of the photocatalytic performance. Therefore, developing efficient vacancy engineering strategies and advanced characteristic techniques associated with theoretical simulation are significantly required to further unveil the key roles of vacancies during the photocatalytic process and promote practical applications.