Solar-driven photocatalytic conversion of CO 2 into fuels has attracted a lot of interest; however, developing active catalysts that can selectively convert CO 2 to fuels with desirable reaction products remains a grand challenge. For instance, complete suppression of the competing H 2 evolution during photocatalytic CO 2 -to-CO conversion has not been achieved before. We design and synthesize a spongy nickel-organic heterogeneous photocatalyst via a photochemical route. The catalyst has a crystalline network architecture with a high concentration of defects. It is highly active in converting CO 2 to CO, with a production rate of ~1.6 × 10 4 μmol hour −1 g −1 . No measurable H 2 is generated during the reaction, leading to nearly 100% selective CO production over H 2 evolution. When the spongy Ni-organic catalyst is enriched with Rh or Ag nanocrystals, the controlled photocatalytic CO 2 reduction reactions generate formic acid and acetic acid. Achieving such a spongy nickel-organic photocatalyst is a critical step toward practical production of high-value multicarbon fuels using solar energy.

Here, we design a model metal-organic CO 2 reduction catalyst, with Ni 2+ ions as active metal centers, TPA as a rigid linker, TEG as a soft linker, and dimethylformamide (DMF) as a solvent, via laser-induced solution reactions. The as-synthesized catalyst, labeled as Ni(TPA/TEG), has a crystalline network architecture with considerable defects and performs nearly 100% selective gas production (CO over H 2 evolution) with a high CO production rate of ~1.6 × 10 4 μmol hour −1 g −1 . Further metal decorations (that is, Rh and Ag) of the Ni(TPA/TEG) catalyst lead to controlled photocatalytic CO 2 reduction reactions that generate formic acid and acetic acid.

When designing catalysts for CO 2 reduction, the material’s ability to capture the CO 2 molecules is another significant consideration ( 24 ). Metal-organic frameworks (MOFs) with high surface area and tunable pores have been used for gas capture and heterogeneous catalysis ( 25 , 26 ). Typically, MOFs have highly ordered crystalline structures constructed by coordinating metal ions or clusters with rigid organic linkers, most often the aromatic carboxylic acid molecules ( 27 ), such as terephthalic acid (TPA). In light of MOF structure design, we replace part of the rigid linkers (for example, TPA) in traditional MOFs with soft molecules (for example, TEG) by laser, considering the comparable molecular length of TEG to TPA (fig. S1). When the TEG molecules, which lack essential carboxylic groups for the perfect framework construction, are weaved into the metal-TPA framework, their substitution of TPA linkers may frustrate the growth of highly ordered MOF crystals, resulting in disordered and defective metal-organic hybrids for effective CO 2 fixation.

We recently developed a laser-chemical method and synthesized active transition metal hydroxide catalysts with a high concentration of defects for water oxidation ( 23 ). Specifically, we used an unfocused infrared laser to initiate the reactions between transition metal ions and triethylene glycol (TEG) and obtained a series of metal hydroxide–TEG composites with a distorted layered structure ( 23 ). This disordered structure enhances the accessibility of water molecules to the active sites and enables efficient electrocatalysis of alkaline water oxidation ( 23 ). Such a laser-chemical strategy may be applied to the discovery of many other catalysts, for instance, novel nanostructured metal-organic heterogeneous catalysts for CO 2 reduction reaction.

Rapid fossil fuel consumption induces environmental burden and energy crisis ( 1 – 3 ). Excessive anthropogenic CO 2 emission is a significant concern because of its hastening impact on climate change ( 4 – 6 ), acidification of ocean ( 7 ), crop yield reduction ( 8 ), extinction of animal species ( 9 ), and damage to human health ( 10 , 11 ). Removal of excessive CO 2 from the atmosphere ( 12 ), particularly converting CO 2 to fuels using solar energy, is currently a global research endeavor ( 13 – 15 ). Discovering novel catalysts that can reduce the stable CO 2 molecules and convert them to liquid fuels with high activity and selectivity is essential ( 13 , 14 ). To date, despite the progress that has been made in investigating the photocatalytic reduction of CO 2 ( 15 – 19 ), controlling the reaction to yield a specific product among many possible reaction species, including CO, H 2 , CH 4 , and formic acid, remains a great challenge ( 16 , 20 , 21 ). Finding photocatalysts that can efficiently convert CO 2 to CO and largely suppress other competing photocatalytic reactions, such as H 2 evolution, would be a critical step forward toward practical solar-to-fuels conversion for the production of high-value multicarbon fuels ( 15 , 17 , 22 ).

RESULTS AND DISCUSSION

Structure determination of the Ni(TPA/TEG) catalyst As shown in Fig. 1A, the Ni(TPA/TEG) composite forms a disordered spongy network structure, in which Ni, O, and C are uniformly distributed (fig. S2). In comparison, the solution without TEG, Ni(TPA) only, forms large particles (Fig. 1B). A three-dimensional electron tomographic reconstruction of the spongy Ni(TPA/TEG) architecture reveals various mesopores in the structure (Fig. 1C and movie S1), which closely resembles the pore features identified from the N 2 physisorption measurements (fig. S3). Figure 1D shows a typical transmission electron microscopy (TEM) image of the spongy Ni(TPA/TEG) composite, where defective lattices with a d-spacing of 1.02 nm are captured. To further interpret the structure of Ni(TPA/TEG) composite, we acquire a scanning nanobeam diffraction data set using an electron beam with a size of ~3 nm, a total beam current of ~5 pA, and an exposure time of 0.5 s, where the electron beam damage to the metal-organic material has been evidently minimized (Fig. 1E). Single-crystalline diffraction patterns along the [100] and [111] orientations of the Ni(TPA/TEG) composite are captured from two different regions of the spongy network (Fig. 1F and movies S2 and S3), showing an orthorhombic structure similar to that of the Ni(TPA) particles (fig. S4). Changes of the diffraction patterns are observed from movies S2 and S3, indicating defects (that is, grain boundaries) in the spongy Ni(TPA/TEG) catalyst (fig. S5). Fig. 1 Structure of the laser-chemical tailored spongy Ni(TPA/TEG) catalyst. (A) Scanning TEM (STEM) images and energy-dispersive x-ray spectroscopy (EDX) mapping of the spongy Ni(TPA/TEG) nanostructure. (B) STEM image of the Ni(TPA/TEG) particles. (C) Three-dimensional tomographic reconstruction of a fraction of spongy Ni(TPA/TEG) composite (movie S1). (D) TEM image of the spongy Ni(TPA/TEG) nanostructure. The inset high-resolution TEM image displays the defective (020) lattices [d (020) = 1.02 nm] of an orthorhombic crystal. (E) Scanning electron nanodiffraction series taken from the Ni(TPA/TEG) particle by a scanning nanoprobe with an electron beam size of ~3 nm. The probe step size is 10 nm with an exposure time of 0.5 s at each step and a total beam current of ~5 pA. (F) Diffraction patterns showing the [100] and [111] orientations of the orthorhombic Ni(TPA/TEG) composite (movies S2 and S3). The dimensions of the diffraction patterns are 11.9 nm−1 × 11.9 nm−1. To verify that the soft TEG molecules have been incorporated into the Ni(TPA) framework (Fig. 2A), we compare the structure of laser-synthesized Ni(TPA) and Ni(TPA/TEG) composites in detail. The x-ray diffraction (XRD) pattern (Fig. 2B) shows that the Ni(TPA) composite has an orthorhombic structure where the Ni-TPA units construct the framework (fig. S4). The spongy Ni(TPA/TEG) has a crystal structure similar to that of Ni(TPA), but slight differences exist in the peak positions and widths of the x-ray lines, which may result from the exchange of linkers and solvent molecules in the structure (28). In the Fourier transform infrared (FTIR) spectra (Fig. 2C), we can see that both samples have clear bands of ν(COO−) (1375 and 1575 cm−1) and ring breathing (815 cm−1) from the TPA linkers. Characteristic bands of δNCO (690 cm−1) and ν(CO) (1685 cm−1) from DMF molecules are found in Ni(TPA), indicating that DMF molecules may reside in the MOF cavities and coordinate with Ni2+ through carbonyl groups (28). Meanwhile, distinct bands of ν(OH) (1065 and 3374 cm−1) related to TEG are found exclusively in Ni(TPA/TEG), and no DMF bands are detected, indicating that TEG molecules exist in Ni(TPA/TEG); DMF molecules that originally occupied the Ni(TPA) framework cavities prefer to leave. The EDX spectrum (Fig. 2D) also shows that no nitrogen (from DMF) can be detected in Ni(TPA/TEG). Thermogravimetric analysis (TGA) curves in Fig. 2E, both displaying three stages of mass losses, indicate differences between the two samples. For Ni(TPA), the mass loss measurements of 6, 18, and 50% are from H 2 O, DMF, and TPA, respectively (28); for Ni(TPA/TEG), the mass losses of 16, 26, and 26% are from H 2 O, TEG, and TPA, respectively. The large differences in the mass loss of TPA in the two samples (50% versus 26%) indicate that soft TEG molecules have replaced part of the rigid TPA linkers, causing the formation of a spongy Ni(TPA/TEG) network. Because of the varying chemical environment of Ni2+, the Ni2p peaks in the x-ray photoelectron spectroscopy (XPS) spectra (Fig. 2F) shifted to the right in Ni(TPA/TEG) compared to Ni(TPA). Fig. 2 Comparison of laser-chemical tailored Ni(TPA/TEG) and Ni(TPA) composites. (A) Proposed design strategy of the disordered spongy Ni(TPA/TEG) composite by introducing soft Ni-TEG building units into a Ni(TPA) framework through laser-chemical reaction. XRD patterns (B), FTIR spectra (C), EDX spectra (D), TGA curves (E), and XPS spectra (F) of the laser-chemical tailored Ni(TPA/TEG) and Ni(TPA) composites. a.u., arbitrary units. We also find that it is more effective to use lasers to cross-link the soft TEG and rigid TPA molecules together with the Ni2+ centers than to use a traditional heating process (see the Supplementary Materials). Laser irradiation appears to produce Ni-TEG building units that are indispensable for the spongy Ni(TPA/TEG) network construction (Fig. 2A). Because of the soft character of TEG molecules, various inhomogeneous configurations of Ni-TEG units can be generated. Mismatches between the soft Ni-TEG units and the rigid Ni-TPA units may introduce considerable defects, leading to the formation of disordered spongy Ni(TPA/TEG).

Evaluation of the photocatalytic activity for CO production We apply the as-synthesized Ni-organic composites for visible light–driven photocatalytic CO 2 reduction in a solvent mixture of acetonitrile/water [considering the high solubility of CO 2 in acetonitrile (29)] under mild reaction conditions (20°C and 400 torr of CO 2 ), with triethanolamine (TEOA) as a sacrificial reducing agent and Ru(bpy) 3 Cl 2 ·H 2 O as a photosensitizer (18, 30). Five samples (figs. S6 to S8), that is, Ni(TPA) (L), Ni(TPA) (H), Ni(TPA/TEG) (L), Ni(TPA/TEG) (H), and Ni(TEG) (L), synthesized by both laser irradiation (L) and traditional heating (H), are examined. Figure 3A shows the CO evolutions from these five Ni-organic catalysts in a 6-hour photocatalytic reaction. The spongy Ni(TPA/TEG) (L) composite shows the highest activity, and the amount of CO is 95.2 μmol after a 2-hour reaction, giving a CO production rate of 15,866 μmol hour−1 g−1, which is several times higher than that from other samples. The total amount of CO produced on the spongy Ni(TPA/TEG) catalyst in 6 hours reaches 136.9 μmol (Fig. 3A), giving a turnover number of 11.5 for the 6-hour reaction (table S1). The CO production rate is also superior compared with many other reported heterogeneous CO evolution photocatalysts to the best our knowledge (17), such as the Co 3 O 4 platelets with [Ru(bpy) 3 ]Cl 2 as a photosensitizer (3523 μmol hour−1 g−1) (18), the sensitized TiO 2 particles with enzyme as a cocatalyst (300 μmol hour−1 g−1) (31), and the sensitized BaLa 4 Ti 4 O 15 particles with Ag as a cocatalyst (22 μmol hour−1 g−1) (16). Note that the soluble homogeneous metal complex catalysts, which have also been investigated for controlled CO 2 reduction (32–34), are not categorized here for comparison. Fig. 3 Conversion of CO 2 to CO by photocatalysis. (A) CO evolution on five Ni-based catalysts with different combinations of TPA, TEG, and DMF. The composites synthesized by laser-chemical approach are labeled with “L”; the ones synthesized by traditional heating method are marked with “H.” (B) CO production on different amounts of the Ni(TPA/TEG) catalyst. (C) Average yield of CO in the first 2 hours for five recycling tests. (D) MS of 12CO (blue lines) and 13CO (red lines) produced on the spongy Ni(TPA/TEG) catalyst by using 12CO 2 and 13CO 2 as gas sources, respectively. m/z, mass/charge ratio. (E) Comparison of CO evolution on five laser-synthesized M(TPA/TEG) (M = Ni, Co, Cu) catalysts. (F) Comparison of H 2 evolution on the five M(TPA/TEG) catalysts. By testing the 2-hour yield of CO on different amounts of the Ni(TPA/TEG) catalyst, we obtain a roughly linear relationship between the amount of evolved CO and the amount of the catalyst (Fig. 3B). However, kinetically, we found that the CO production rate actually decreases with the increase in the amount of the catalyst (fig. S9), where 1.0 mg of the Ni(TPA/TEG) catalyst gives a CO production rate of ~26,620 μmol hour−1 g−1 in the same solution, indicating that more electrons generated from the photosensitizer molecules could have been transferred to the catalytic active sites. We have also tested the reusability of the spongy Ni(TPA/TEG) catalyst upon each 2-hour photocatalysis, where the catalyst has kept its activity and selectivity after recycling (Fig. 3C). It also exhibits excellent structural stability, and no obvious structural change is found after 24 hours of photocatalysis (fig. S10). To confirm the origin of the as-produced CO, we use isotopic 13CO 2 as feedstock gas for the photocatalytic reduction and examine the products by gas chromatography–mass spectrometry (GC-MS). A major signal at a mass/charge ratio of 29 on the mass spectrum corresponding to 13CO (Fig. 3D, red lines) appears, which confirms that the as-detected CO originates from the CO 2 gas source (fig. S11). Transition metal ions with switchable electronic states have long been considered promising active sites for diverse photocatalytic or electrocatalytic reactions, such as water splitting (35–37), CO oxidation (38), and CO 2 reduction (39). Consequently, we use the laser-chemical method to synthesize four additional samples and compare different metal ions, that is, Ni2+, Co2+, and Cu2+ (fig. S12), as active sites for the photocatalytic CO 2 reduction. Results show that the spongy Ni(TPA/TEG) catalyst is still the most active catalyst for CO evolution, and the activity worsens with the incorporation of Co and Cu ions (Fig. 3E). The amount of CO generated on the Ni(TPA/TEG) catalyst is almost two times that of the CO generated from Co(TPA/TEG) in a 6-hour photocatalytic reaction. The Cu(TPA/TEG) catalyst barely generates any CO, which is distinctly different from the metal Cu catalyst that has superior activity for CO evolution from the electrocatalytic reduction of CO 2 (40).

Evaluation of the CO production selectivity We find that CO (CO 2 + H 2 O + 2e−→CO + 2OH−) is the only detectable gas product from the photocatalytic CO 2 reduction on the TPA-containing Ni-organic catalysts. H 2 evolution (2H 2 O + 2e−→H 2 + 2OH−), usually acting as a major competing reaction in the CO 2 reduction system for many transition metal–based catalysts (18, 20), has been completely suppressed (table S1). Thus, a near 100% selectivity of CO production (over H 2 evolution) is achieved. Note that no other potential competing gas products, such as CH 4 and C 2 H 4 (17), have been detected in our experiments either. For comparison, we have detected considerable amounts of H 2 from other laser-synthesized M(TPA/TEG) (M = Ni, Co, Cu) catalysts (Fig. 3, E and F, and fig. S13), and CO selectivity values of 96, 70, 78.2, and 4.8% were measured for NiCo(TPA/TEG), Co(TPA/TEG), NiCoCu(TPA/TEG), and Cu(TPA/TEG), respectively. We have also detected a fair amount of H 2 (25.7 μmol) from the hydroxylated TPA-free Ni(TEG) catalyst in addition to the CO evolution (96.5 μmol) after a 6-hour reaction (fig. S14), which is analogous to the Ni-based hydroxides for practical H 2 evolution from the electrocatalysis of water (36).

Tuning the selectivity for liquid fuels production from CO 2 Furthermore, considering the potential of noble metal electrodes for CO 2 reduction (41, 42), we decorate the spongy Ni(TPA/TEG) with noble metal nanocrystals, that is, Rh and Ag, in pursuit of tuning the selectivity of liquid fuels production from the photocatalytic CO 2 reduction (15, 21). Figure 4 (A to C) shows the Ag-decorated Ni(TPA/TEG) catalyst, where Ag nanocrystals with an average diameter of ~6 nm are well dispersed on the Ni(TPA/TEG). We examine the liquid products after a 6-hour reaction on three catalysts, that is, undecorated Ni(NPA/TEG), Rh-decorated Ni(TPA/TEG), and Ag-decorated Ni(NPA/TEG) (Fig. 4D). For the Ni(NPA/TEG) catalyst without any decoration, we measure formic acid (HCOOH) with a concentration of 29.2 μM and acetic acid (CH 3 COOH) with a concentration of 72.5 μM in addition to CO. With the decoration of Rh and Ag nanocrystals, the amounts of CO decrease drastically, whereas the amounts of liquid products significantly increase. Formic acid (HCOOH) with a concentration of 313.5 μM is mainly obtained on the Rh-decorated Ni(NPA/TEG) catalyst, and CH 3 COOH with a concentration of 195.6 μM is the major product for the Ag-decorated Ni(NPA/TEG) catalyst, where the origin of CH 3 COOH has been confirmed from the flowing CO 2 source by MS (fig. S15). Note that CO, formic acid, and acetic acid reflect the overall product distribution of the photocatalytic CO 2 reduction reaction; no other liquid products, such as methanol, ethanol, or propanol, are detected in this experiment. Fig. 4 Generation of liquid products on metal-decorated Ni(TPA/TEG) composites. Low-magnification (A) and high-resolution (B) TEM images of the Ni(TPA/TEG) composite decorated with Ag nanocrytals. (C) EDX mapping of the as-prepared Ni(TPA/TEG)-Ag composite. (D) Comparison of the amount of the products (CO, HOOH, and CH 3 COOH) generated from photocatalytic CO 2 reduction on Ni(TPA/TEG), Ni(TPA/TEG)-Rh, and Ni(TPA/TEG)-Ag catalysts.

Enhanced production of liquid fuels from photocatalytic reduction of CO By assuming that the evolved CO may be further consumed for the production of acids, we also conduct control experiments using CO, instead of CO 2 , as the gas feedstock for the photocatalytic reduction reaction. As a result, largely enhanced yields of acids are obtained. For instance, the amount of HCOOH evolved from the CO reduction is 24 times higher than that from the CO 2 reduction on the Ni(NPA/TEG) catalyst (Fig. 5A), and the amount of CH 3 COOH produced from the CO reduction reaction for the Ag/Rh-decorated Ni(NPA/TEG) catalyst exhibits a sixfold increase over the results from the CO 2 reduction reaction (Fig. 5B and fig. S16). Besides the increase in acid production, another important C2 product, that is, ethanol [with a concentration of 270.6 μM for Ni(NPA/TEG)-Rh and 262.2 μM for Ni(NPA/TEG)-Ag], has also emerged from the 6-hour photocatalytic CO reduction reaction. Fig. 5 Comparison of the liquid products generated from photocatalytic CO 2 reduction reactions (CO 2 RR) and CO reduction reactions (CORR) on two catalysts. (A) Ni(TPA/TEG). (B) Ni(TPA/TEG)-Ag.