Synthesis and characterization of the designed photothermal host and Ru-loaded MXene catalyst (Ru@m-Ti3C2Tx)To simultaneously achieve high solar absorption, low thermal emission, and activated catalytic process in the catalyst, we employed a two-step approach including a hard-templated method and subsequent impregnation, constructing the photothermal host with selective spectrum and porous structure (Fig. 2a–e, details seen in Methods)34,35,36. To create the ideal photothermal layer, we utilized negative pressure during suction filtration to create the flat Ti3C2Tx flake layers (Fig. 2e). Noted that the intrinsic selective spectrum of Ti3C2Tx is dependent on its structure (details seen in Fig. 3 below). As the presence of channels for advancing mass transfer of reactants and products is crucial for photothermal catalytic process, the PMMA spheres were first wrapped by Ti3C2Tx flakes via polar groups. Subsequently, PMMA spheres were removed through thermal evaporation to obtain a macroporous Ti3C2Tx (m-Ti3C2Tx) membrane, which contains numerous channels or porosities (Fig. 2b–e and Supplementary Figs. 2–7). After loading the Ru nanoparticle catalyst, we can finally achieve the Ru@m-Ti3C2Tx with a porous side composited of macroporous Ti3C2Tx spheres and a flat side of Ti3C2Tx flakes (thickness ~70 μm, Supplementary Fig. 8).Fig. 2: Characterizations of designed photothermal host and Ru-loaded MXene catalyst (Ru@m-Ti3C2Tx).a Schematic of the preparation process of Ru@m-Ti3C2Tx. b–e SEM image of PMMA (b), PMMA@Ti3C2Tx (c) and m-Ti3C2Tx (d, e). f TEM image of Ru@m-Ti3C2Tx. g HRTEM image of Ru@m-Ti3C2Tx and secondary electron image (inset). h HAADF-STEM image and corresponding elemental mapping images of Ru@m-Ti3C2Tx. i High-resolution HAADF-STEM image of a single Ru nanoparticle and the corresponding crystalline structure (inset).Fig. 3: Comparison of different photothermal effects caused by two different side structures.a Schematic of the high photothermal effect of the flat side due to the low εMIR (~21%). b The Ti3C2Tx (002) pole figure for the flat side of m-Ti3C2Tx. c The spectrum of the flat side of m-Ti3C2Tx within the 0.28–18.3 μm wavelength range, as well as the AM 1.5 G solar spectrum and the radiation spectrum of a blackbody at 673 K. d Schematic of the low photothermal effect of the porous side due to the high εMIR (~91%). e The Ti3C2Tx (002) pole figure for the porous side of m-Ti3C2Tx. f The spectrum for the porous side of m-Ti3C2Tx within the 0.28–18.3 μm wavelength range, as well as the AM 1.5 G solar spectrum and the radiation spectrum of a blackbody at 673 K. g The spectrum simulation for the flat side of m-Ti3C2Tx with varying thicknesses within the 0.28–20 μm wavelength range. h The spectrum simulation for the porous side of m-Ti3C2Tx with varying pore diameters within the 0.28–20 μm wavelength range. i The thermal simulations of different sides. j The measured temperatures from infrared images under different light powers.To characterize the designed photothermal host and the catalyst Ru@m-Ti3C2Tx, several measurements were performed. Inductively coupled plasma optical emission spectroscopy (ICP-OES) result reveals that the Ru content is ~0.75 wt%, which causes the diffraction peaks associated with metallic Ru could not be detected by X-ray diffraction (XRD) (Supplementary Fig. 6)37. The Ru 3d and Ru 3p peaks were clearly revealed by X-ray photoelectron spectroscopy (XPS), which confirms the presence of Ru in Ru@m-Ti3C2Tx (Supplementary Fig. 9)38. Figure 2f depicts the transmission electron microscopy (TEM) image of a typical Ru-Ti3C2Tx flake. The black dots are uniformly distributed on Ti3C2Tx flakes, which are Ru-particles prepared by the impregnation method. The high-resolution transmission electron microscopy (HRTEM) image at the edge of the layer indicates the lattice fringe spacing of ~1.3 nm, in agreement with the (002) plane of Ti3C2Tx (Fig. 2g). Meanwhile, the high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) image and corresponding elemental mappings confirm that C and Ti elements are distributed throughout the support while Ru mainly exists in the form of nanoparticles (Fig. 2h). To reveal the crystal structure, we further investigated a single Ru nanoparticle. The HAADF-STEM image shows that the distances between adjacent planes are 2.3 and 2.1 Å, corresponding to the lattice spacing of the (1-10) and (10-1) planes of metallic Ru, respectively. The result is consistent with the hexagonal close-packing crystal phase of Ru, thus the Ru nanostructures via this preparation process are considered to be crystalline Ru nanoparticles (Fig. 2i). Meanwhile, we probed the flat side of the Ru@m-Ti3C2Tx and the Ru@Ti3C2Tx catalyst, and we found that the Ru nanoparticle sizes on both sides typically fall within the range of 3–4 nm (Supplementary Figs. 10, 11). Of note, different metal nanoparticles can also be loaded onto our designed Ti3C2Tx host using a similar preparation method (see “Method” section for details) for different reactions.Detailed analysis of its unique photothermal effectTo verify the enhanced photothermal effect from low εMIR in the designed Ru@m-Ti3C2Tx membrane, a detailed analysis of the relationship among surface structure–spectrum–thermal was performed. Specifically, the spectral features of Ru@m-Ti3C2Tx strongly depend on the surface structure irradiated under sunlight, which could be identified by the crystal orientations of nanoflakes33,39. On the flat side, the overall εMIR of m-Ti3C2Tx (2.8–18.3 μm) is only ~21%, which remarkably prohibits it from thermal radiation loss, thus promoting localized high temperature on this side of the designed catalyst (Fig. 3a). Typically, as a two-dimensional material, the multilayer overlapping structure of Ti3C2Tx forms the crystallographic orientation of a certain characteristic crystal plane (002), which can be determined by X-ray pole figure40,41. For this measurement, the 2θ value was fixed and the m-Ti3C2Tx membrane was rotated from 0° to 360° at series of tilt angles from 0° to 75°. Figure 3b and Supplementary Fig. 12a show a (002) pole figure for the flat side of m-Ti3C2Tx, with one spot at a tilt angle of 0°. The out-of-plane orientation of flat side was detected, corresponding to the (002) plane. In this orientation, similar to the natural crystalline structure of Ti3C2Tx, the flat side of the m-Ti3C2Tx membrane shows an excellent spectral selectivity, which exhibits an 88% absorptance in the sunlight wavelength range and a εMIR of 21% (Fig. 3c).To further demonstrate the low εMIR on the flat side, we also performed the full-wave simulation using the finite-element method-based software package COMSOL Multiphysics (see Method for details). In the calculation model, periodic close-packing Ti3C2Tx spherical pores are used for simplicity. We used the permittivity data of Ti3C2Tx from previous work33. Meanwhile, we characterized the thickness of the top layer to be ~150 nm (Supplementary Fig. 13). Calculated emissivity spectra at the normal incidence for flat side with different pore diameters are shown in Fig. 3g and Supplementary Fig. 14. The relatively low εMIR is nearly independent with pore diameters. The phenomenon can be explained by the large real and imaginary permittivity of Ti3C2Tx. The infrared light is highly reflected from the top flat side due to the drastic impedance mismatching between reactant gas and Ti3C2Tx slabs, and the large imaginary permittivity of Ti3C2Tx also leads to the strong absorption of residual light even with hundreds of nanometers-thick slabs. As a result, light can hardly transmit to the porous Ti3C2Tx layer, and the low εMIR is nearly unchanged with different diameters of Ti3C2Tx pores.Furthermore, to understand the attenuation of the photothermal effect of conventional photothermal catalysts from high εMIR, the irradiated surface was experimentally reconstructed. As we all know, multiple scattering is a common phenomenon in optics, which increases emissivity at the corresponding waveband. The porous side which consists of massive ~5 μm macroporous Ti3C2Tx spheres increase the εMIR via multiple scattering, providing an in-situ control group (Fig. 3d). According to the (002) pole figure for the porous side of m-Ti3C2Tx, there is non-orientation on the porous side (Fig. 3e and Supplementary Fig. 12b). It could be caused by the large angular bending of the Ti3C2Tx flakes. Figure 3f shows that the overall εMIR for the porous side of m-Ti3C2Tx (2.8–18.3 μm) rises to ~91%, which significantly increases the thermal emission. According to the simulation, strongly enhanced εMIR is found for porous side. The near-unity εMIR is due to the multiple scattering effect generated in Ti3C2Tx pore layers (Fig. 3h). Although the shell of a single pore is thick, the multiple scattering effect effectively increases the time and strength of light-matter interaction, leading to near-unity and broadband emission in infrared wavelengths. As a result, the engineered porous side of the m-Ti3C2Tx membrane possesses no crystal orientation and exhibits high emission characteristics in the 0.28–18.3 μm wavelength range, with an absorptance of 92% in the sunlight wavelength range and an εMIR increasing to 91% (Fig. 3f).To further verify the enhanced photothermal effect from low εMIR, light–thermal conversion efficiency calculation, infrared images, and thermal simulation were employed to demonstrate the reduction of thermal loss. Benefit from the low εMIR, the light–thermal conversion efficiency (ηlight-th, at 698 K) of the flat side can reach a high value of 74.2% far higher than that of the porous side (32.2%) (details seen in Methods). This difference will be more obvious in the high-temperature region, as mentioned above. Under 2 W cm−2 light power, the highest temperatures at the center of the flat side can reach 742 K, while the porous side only reaches 656 K (Fig. 3i). This similar phenomenon also can be observed under other light powers. The temperature difference between the two sides increases with the light power (Supplementary Fig. 15). Furthermore, the infrared images also prove the temperature difference between the two different sides, and this difference also increases with the light power (Fig. 3j and Supplementary Fig. 16). The flat side of Ru@m-Ti3C2Tx possess higher heating rate than the porous side of Ru@m-Ti3C2Tx (Supplementary Fig. 17). Meanwhile, to systematically assess the photothermal process of the catalyst, we performed experiments and simulations to investigate the its cooling process. As a result, the cooling rate of the flat side of Ru@m-Ti3C2Tx is lower than that of the porous side of Ru@m-Ti3C2Tx in the high-temperature region, which attributes to the reduction of thermal radiation loss (Supplementary Figs. 18, 19 and Supplementary Note 3). The heat profile is also investigated to prove the uniform distribution of heat in the architecture (Supplementary Fig. 20 and Supplementary Note 4). Through the detailed analysis of surface structure–spectrum–thermal, it reveals that the enhanced photothermal effect originates from low εMIR on the flat side of the Ru@m-Ti3C2Tx membrane, which could serve as an ideal photothermal catalyst.Boosting photothermal catalytic performanceThis photothermal host design, composited of macroporous Ti3C2Tx spheres and Ti3C2Tx flakes, aims to enhance the photothermal effect via low εMIR and to provide channel for the mass transfer of reactants and products. To reveal the advantages of our catalyst design, CO2 hydrogenation performance was tested under full-Arc Xe lamp irradiation for the Ru@m-Ti3C2Tx membrane. Ti3C2Tx membrane and Ru@Ti3C2Tx membrane were also prepared and used as references (see Method for details). First, to clarify whether the photothermal host affects the Sabatier reaction, a series of control experiments were performed under light power 0.1–2 W cm−2 for pure Ti3C2Tx membrane. At different light powers, nearly no methane and intermediates can be detected, clearly indicating that Ti3C2Tx itself is not active to CO2 methanation (Supplementary Figs. 21–23). As shown in Supplementary Fig. 24, the catalytic activity of the Ru@Ti3C2Tx membrane increases with the light power. However, its CH4 production rate normalized by the mass of Ru only reached 128.5 mmol gRu−1 h−1 at 2 W cm−1, which could result from insufficient contact of the reactant with the Ru catalyst. In sharp contrast, the porous side of Ru@m-Ti3C2Tx (Ru@m-Ti3C2Tx-p) membrane, despite having higher εMIR, possesses much higher CH4 production rate (the average value of 733.2 mmol gRu−1 h−1 at 2 W cm−1) at each light power compared with Ru@Ti3C2Tx membrane, which could be attributed to rapid mass transfer of reactants as well as products (Fig. 4a, the blue line). The contribution of mass transfer has been further verified by simulation and in-situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) (Supplementary Figs. 25–29, Supplementary Table 4 and Supplementary Note 5).Fig. 4: Photothermal catalytic performance of the designed optically spectral catalyst and mechanism analysis.a CH4 production rate on the two different sides of the Ru@m-Ti3C2Tx catalyst. b CO2 conversion rate and selectivity on the flat side of the catalyst. c CO2 conversion rate and selectivity on the porous side of the catalyst. d Stability test for hydrogenation of CO2 at 2 W cm−2 without external heating. e Performance comparison of CO2 reduction among this work and previous works. f In-situ DRIFTS over flat side of Ru@m-Ti3C2Tx. g CO2 conversion rate on other catalysts with our design strategy. h Energy landscape for CO2 hydrogenation to CO and CH4 on the typical (0001) surface of Ru particles. To improve legibility, H2 is omitted from the labels after the initial state. All error bars represent the standard deviations of three independent measurements and the bars indicate mean values.Furthermore, the reduction of the εMIR can increase the photothermal effect and further improve the methane yield of photothermal CO2 reduction. As shown in Fig. 4a, the red line, catalytic activity based on the flat side of the Ru@m-Ti3C2Tx (Ru@m-Ti3C2Tx-f) increases with the light power, which outperforms Ru@Ti3C2Tx and Ru@m-Ti3C2Tx-p under all light powers, especially at high intensities. Under the highest power of 2 W cm−2, the average CH4 production rate of Ru@m-Ti3C2Tx-f reaches 3317.2 mmol gRu−1 h−1, which is over 4 times higher than that of Ru@m-Ti3C2Tx-p (Supplementary Fig. 30). On the other hand, to achieve the same production rate (733.2 mmol gRu−1 h−1), Ru@m-Ti3C2Tx-f could reduce about 0.4 W cm−2 power compared with that of Ru@m-Ti3C2Tx-p at 2 W cm−2. Furthermore, the high product selectivity for CH4 can be detected on both Ru@m-Ti3C2Tx-p and Ru@m-Ti3C2Tx-f, probably due to the same size of Ru nanoparticles which does not vary with the irradiated surface (Fig. 4b, c)42. In order to eliminate potential differences in catalytic activity due to the placement state of the catalysts, the Arrhenius plots of Ru@m-Ti3C2Tx-p and Ru@m-Ti3C2Tx-f were constructed to assess kinetic performance. The results demonstrate that no noticeable changes were observed in the apparent activation energies (Ea) of these two sides of catalysts, indicating that the reaction energy remains unchanged with the placement state (79.8 ± 4.2 kJ mol−1 for Ru@m-Ti3C2Tx-p and 76.5 ± 8.6 kJ mol−1 for Ru@m-Ti3C2Tx-f) (Supplementary Fig. 31).To demonstrate the high stability of our catalyst with the proposed design, we first performed 20-h consecutive testing and another 10-h consecutive testing for Ru@m-Ti3C2Tx-f (Fig. 4d). Between the two tests, the Ru@m-Ti3C2Tx-f was kept in argon gas at room temperature for two months. During the first 20-h test, both the CO2 conversion rate and product selectivity are almost no change. Surprisingly, the next 10-h test still exhibits stable selectivity in spite of the slightly fluctuating yield of CH4. The stable activity and selectivity can be attributed to the fact that the strong physical barrier between Ti3C2Tx layers resists the Ostwald ripening of Ru nanoparticles (see following post-catalysis characterizations). As shown in Supplementary Fig. 32, the hollow structure of Ru@m-Ti3C2Tx is nearly not changed and the overall structure was kept intact after 30 cycles of testing. Meanwhile, the absorptance of visible light still reaches 87.4%, and the emissivity in the mid-infrared region remains at 22.0% (Supplementary Fig. 33). Besides, the Ru nanoparticle still possesses good crystallinity and high dispersion, and its combination with the Ti3C2Tx has no obvious change (Supplementary Figs. 34, 35). The high stability may attribute to the demonstrated structural stability of the MXene analogue and the potential interaction between nanoparticle and Ti3C2Tx43,44. We compared the CH4 production rate to typical reports of CO2 methanation with or without an external heater (Fig. 4e and Supplementary Table 5). The results indicate that, at high optical power levels, whether based on the mass of Ru or the entire catalyst, our catalyst exhibits significantly higher production rates compared to the reported works, and it achieves a yield record among catalysts without active supports.Our generic design strategy is expected to serve the vast majority of photothermal catalytic reactions. In order to prove the efficient catalysis performance resulting from the reduction of the εMIR, Pd, and Ni catalysts were also loaded on to m-Ti3C2Tx as active metal catalysts. The low yields are obtained compared with Ru-based catalysts, which ascribed to the types of active metals testified by previous works15. However, both catalysts have several times the performance improvement reaching an excellent yield by the reduction of the εMIR without active supports (Fig. 4g). Furthermore, to demonstrate the generality of the Janus design towards various MXene materials, we also performed structural design on two other MXene, i.e. Ti3CNTx and Ti2CTx. As a result, both materials exhibit excellent selective spectra and complete porous structures (Supplementary Figs. 36, 37).To confirm the origin of the as-produced CH4, the 13C isotope labeling experiment was performed for the photothermal hydrogenation. The products were examined by gas chromatography-mass spectrometry (GC–MS). A major signal at a mass/charge ratio of 17 on the mass spectrum corresponding to 13CH4 appears, which verifies that the as-detected CH4 originates from the CO2 hydrogenation process (Supplementary Fig. 38). We also performed in-situ DRIFTS under reaction conditions to reveal the possible reaction mechanism of CO2 methanation on our Ru@m-Ti3C2Tx. The measurements were performed at 300 °C in the reaction gas mixture of 80 vol% H2/20 vol% CO2. As shown in Fig. 4f and Supplementary Fig. 39, the peak at 1649 cm−1 is characteristic of surface CO2* species, namely bicarbonate45,46. As the reaction proceeded, the formate species gradually emerged according to the peaks at 1585, 1393, 1374, and 1355 cm−1, which means the transformation of bicarbonate to formate47,48. Meanwhile, the peak at 1040 cm−1 for CH3O* is detected, demonstrating the further hydrogenation of HCOO*. Consequently, the signal intensities attributed to CH4 species (located at 3016 and 1304 cm−1) increased gradually as the reaction progressed. The peak assignments of the surface species are listed in Supplementary Table 6. The reaction diagram of CO2 hydrogenation on the (0001) surface of Ru was explored by using density functional theory (DFT) calculations to further reveal the reaction mechanisms (Fig. 4h)49,50. The adsorption and conversion of CO2 on the Ru site are more favorable than other sites. Although both reactions CO2* + H* → HOCO* and CO2* + H* → HCOO* are exothermic. The reaction CO2* + H* → HCOO* (ΔrG = −0.956 eV) is energetically more preferred than CO2* + H* → HOCO* (ΔrG = −0.555 eV). So, we conclude that HCOO* is the dominating species of CO2 reduction on Ru surface, which is in consistent with the in-situ DRIFTS observations. Whereafter, the C-O bond cleavage of HOCO* and the subsequent desorption at the catalytic site render the entire process highly energetically unfavorable (ΔrG = 2.355 eV). In sharp contrast, the C-O bond cleavage of HCOO*, followed by its successive hydrogenation until the final formation of CH4, results in the entire process highly energetically favorable (ΔrG = −0.989 eV), which is also consistent with the in-situ DRIFTS observations. Therefore, it can be inferred that CO2 hydrogenation on the surface of Ru exhibits good selectivity towards CH4.
