Sub-Millisecond Laser-Irradiation-Mediated Surface Restructure Boosts the CO Production Yield of Cobalt Oxide Supported Pd Nanoparticles

The catalytic conversion of CO2 into valuable commodities has the potential to balance ongoing energy and environmental issues. To this end, the reverse water–gas shift (RWGS) reaction is a key process that converts CO2 into CO for various industrial processes. However, the competitive CO2 methanation reaction severely limits the CO production yield; therefore, a highly CO-selective catalyst is needed. To address this issue, we have developed a bimetallic nanocatalyst comprising Pd nanoparticles on the cobalt oxide support (denoted as CoPd) via a wet chemical reduction method. Furthermore, the as-prepared CoPd nanocatalyst was exposed to sub-millisecond laser irradiation with per-pulse energies of 1 mJ (denoted as CoPd-1) and 10 mJ (denoted as CoPd-10) for a fixed duration of 10 s to optimize the catalytic activity and selectivity. For the optimum case, the CoPd-10 nanocatalyst exhibited the highest CO production yield of ∼1667 μmol g−1catalyst, with a CO selectivity of ∼88% at a temperature of 573 K, which is a 41% improvement over pristine CoPd (~976 μmol g−1catalyst). The in-depth analysis of structural characterizations along with gas chromatography (GC) and electrochemical analysis suggested that such a high catalytic activity and selectivity of the CoPd-10 nanocatalyst originated from the sub-millisecond laser-irradiation-assisted facile surface restructure of cobalt oxide supported Pd nanoparticles, where atomic CoOx species were observed in the defect sites of the Pd nanoparticles. Such an atomic manipulation led to the formation of heteroatomic reaction sites, where atomic CoOx species and adjacent Pd domains, respectively, promoted the CO2 activation and H2 splitting steps. In addition, the cobalt oxide support helped to donate electrons to Pd, thereby enhancing its ability of H2 splitting. These results provide a strong foundation to use sub-millisecond laser irradiation for catalytic applications.


Introduction
The thriving anthropologic activities result in increased atmospheric carbon dioxide (CO 2 ) levels, which, consequently, has led to detrimental effects on the environment [1]. Therefore, establishing efficient CO 2 utilization strategies is imperative. In this context, the catalytic transformation of carbon dioxide (CO 2 ) into beneficial fuels and commodities is of potential interest [2,3]. Such an approach not only mitigates atmospheric CO 2 concentration but also circumvents energy issues. Given that CO 2 can be utilized as a C 1 feedstock for producing various fuels, such as carbon monoxide (CO), methane (CH 4 ), and formic acid (HCOOH), the selective CO 2 hydrogenation (reduction/conversion) to CO via the reverse water-gas shift (RWGS) reaction is the most promising path because of its well-known industrial applications [4][5][6]. However, the competitive CO 2 methanation (where the CO is further hydrogenated to produce methane (CH 4 )) severely hampers the rate of RWGS reaction, thereby suppressing the production yield of CO [7]. In addition, because of the endothermic behavior of the RWGS reaction, the exothermic CO 2 methanation reaction is more favourable at low temperatures [8]. Therefore, it can be concluded that the development of highly active and selective catalysts for an RWGS reaction that can be operated at low temperatures is an urgent need for establishing a zero-carbon economy.
The RWGS reaction follows two main possible pathways: the redox and the associative pathways [9,10]. For the redox pathway, the CO 2 molecule first dissociates into *CO and *O intermediates. There are two possibilities in this step: (i) the catalyst being oxidized due to *O atoms from the dissociation of the CO 2 molecule or (ii) the CO 2 dissociates at the oxygen vacancy (O V ) sites to produce *CO. On the basis of the CO 2 dissociation mechanism, the hydrogen (H 2 ) molecule can reduce the catalyst or create oxygen vacancies (H 2 + M-O → H 2 O + M; here, M is the notation for metal). Mostly, the catalytic materials comprising reducible oxides follow the redox pathway for completing the RWGS reaction. [11] In contrast, the associative mechanism follows a completely different pathway, where the H 2 molecule first dissociates into *H atoms, which subsequently react with adsorbed *CO 2 molecules to form intermediate species, such as formate (*HCOO) or carboxyl (*COOH), and they eventually decompose to yield *CO and *OH. As reported in the literature, reducible oxide-supported noble metals follow the associative pathway [12]. For instance, Zhao et al. revealed the size effect of Pt nanoparticles (NP)s on CeO 2 support towards RWGS reaction [13]. On the basis of the aforementioned discussion, it can be concluded that the RWGS reaction is sensitive to the type of catalyst, and to achieve the optimal CO selectivity and production yield at low temperatures, it is imperative to develop highly efficient catalysts that have two adjacent active sites for CO 2 activation and H 2 dissociation. In detail, the catalytic materials with a single reaction site for CO 2 reduction lead to quick failure due to the severe coke effect, while catalytic materials with single reaction sites for H 2 dissociation are inactive towards CO 2 . Consequently, potential synergy between two adjacent reaction sites (for CO 2 activation and H 2 dissociation) is desired for optimum product yield. For instance, our previous study demonstrated the potential collaboration between neighbouring reaction sites of tetrahedral symmetric nickel oxide (NiO T ) and Pd in bimetallic NiO T Pd-T nanocatalysts, where NiO T and Pd synergistically trigger the CO 2 activation and H 2 dissociation, respectively [14]. In addition, the previously published literature reported that surface O V sites in the catalyst supports as well as in the active metal can boost the CO 2 activation step during CO 2 hydrogenation, and, thus, the catalytic performance can be improved [15,16]. The aforementioned arguments suggest that the surface atomic configuration is the cornerstone for catalytic activity; therefore, exploration of next-generation techniques to manipulate the surface atomic arrangement is important. In this context, thermal annealing has been frequently employed for controlling the heteroatomic intermixing on the surface as well as the subsurface domains of nanocatalysts for improving the catalytic performance in various reactions; nonetheless, the longer operational time and high energy input make it highly reluctant to use on a commercial scale [17]. In contrast, pulsed laser irradiation with controllable duration and high energy photon flux has emerged as a potential technique for desired atomic manipulation to design the material with optimum functionality [18]. Moreover, this method enables the opportunity to restructure the nanocatalyst surface within a sub-nanometer range, promoting strong heteroatomic intermixing and enhancing catalytic activity. In our previous study, multiple metal-to-metal oxide heterogeneous Nanomaterials 2023, 13, 1801 3 of 12 interfaces have been formed in a trimetallic system (CuNiPd) for enhanced CO 2 reduction performance by using sub-millisecond laser annealing [19].
By keeping the aforementioned scenarios in view, herein, we have fabricated Pd NPs on the cobalt oxide support (hereafter denoted as CoPd) for high-performance RWGS reactions. Furthermore, a pulsed laser beam with per-pulse energies of 1 mJ and 10 mJ was used for a fixed duration of 10 s to manipulate the surface and/or sub-surface atomic arrangements of as-prepared CoPd nanocatalyst to improve the catalytic performance towards RWGS reaction. For the optimum case, when the per-pulse energy of the pulsed laser beam was 10 mJ (hereafter denoted as CoPd-10), some cobalt oxide (CoOx) atoms migrated on the surface of Pd nanoparticles (defect sites of Pd). With such an atomic rearrangement, the CoPd-10 nanocatalyst delivered the CO production yield of~1667 µmol g −1 catalyst , with CO selectivity of~88% at 573 K, which was enhanced 41% as compared with pristine CoPd (~976 µmol g −1 catalyst ). The results of the physical investigations and electrochemical analysis indicated that the potential synergism between surface-anchored atomic CoOx species and adjacent Pd active sites boosted the CO production yield of the CoPd-10 nanocatalyst, where CoOx and Pd reaction sites simultaneously promoted the CO 2 activation and H 2 splitting. We envision that the obtained results could serve as a basis for developing catalysts with improved activity and selectivity for the RWGS reaction.

Materials and Methods
The CoPd nanocatalysts were prepared by a sequential and vigorous wet chemical reduction method. To improve the metal-support interaction and achieve better dispersion, before synthesis, the surface functionalization of the catalyst support (i.e., carbon black (UR-XC72, UniRegion Bio-Tech, Palo Alto, CA, USA)) was achieved via acid treatment. [20] Subsequently, in the first step, 3 g of 2 wt.% acid-treated carbon black (hereafter denoted as active carbon (AC)) solution (i.e., 60 mg of AC) was dispersed in 3.06 g of 0.1 M cobalt (III) chloride (99%, Sigma-Aldrich Co., St. Louis, MO, USA) (i.e., the weight ratio of Co to AC was 30 wt.%) and stirred at 600 rpm for 6 h at room temperature (solution A). In the second step, 0.01 g of sodium borohydride (NaBH 4 ; 99%, Sigma-Aldrich Co.) in 5 mL of D.I. water was instantly dropped to solution A (i.e., Co 3+ ions adsorbed on AC surface) to reduce the Co 3+ ions on the AC surface (i.e., formation of Co-AC) (solution B). Finally, the, 3.06 g of a palladium (Pd) precursor solution (i.e., 0.1 M solution containing 0.306 mmol of Pd metal ions (PdCl 2 , 99%, Sigma-Aldrich Co.)) was added to solution B. In this step, the Pd 2+ ions were reduced on CoOx support via NaBH 4 added in the second step (the amount of NaBH 4 in the second step was measured for reducing both Co 3+ and Pd 2+ ions), and the CoPd nanocatalyst was formed. The final product was sequentially washed with acetone, isopropyl alcohol (IPA), and DI water several times, then dried at 70 • C for 24 h. As-prepared CoPd nanocatalyst was further subjected to sub-millisecond pulsed laser irradiation with per-pulse energies of 1 mJ and 10 mJ to restructure the surface of the material. Hereafter, the CoPd nanocatalysts irradiated with 1 mJ and 10 mJ per-pulse energies are denoted as CoPd-1 and CoPd-10, respectively. In this study, a laser beam with the wavelength of 976 nm and 850 µs pulses was generated from the diode laser. Further details of laser setup are provided in supplementary information.

Physical Characterizations
The structural characteristics of as-prepared CoPd nanocatalysts were examined by cross-referencing the results of high-resolution transmission electron microscopy (HRTEM), X-ray diffraction (XRD), X-ray absorption spectroscopy (XAS), and X-ray photoelectron spectroscopy (XPS). The HRTEM images were collected at National Tsing Hua University, Taiwan. The XRD patterns were obtained at the beamline of BL-01C2 in the National Synchrotron Radiation Research Center (NSRRC), where the wavelength of incident X-rays was 0.688 Å (18 keV). The XAS was carried out at beamlines BL-17C and 01C1 in NSRRC, Taiwan, whereas the XPS spectra were measured at beamline BL-24A1 in NSRRC, Taiwan.

Electrochemical Characterizations and Gaseous Product Analysis
The CO-stripping voltammograms were collected using a three-electrode system, where a glassy carbon electrode, Pt wire, and Ag/AgCl electrode were used as the working, counter, and reference electrodes, respectively. The adsorption of CO on the surface of the catalyst was performed by purging CO into 0.5 M H 2 SO 4 at 0.05 V (vs. RHE) for 20 min. Then, the CO stripping voltammetry was measured between −0.10 and 1.20 V (vs. RHE) in N 2 -saturated 0.5 M H 2 SO 4 solution at a scan rate of 50 mVs −1 . Finally, the catalytic performances of CoPd nanocatalysts towards CO 2 conversion were evaluated by using the previously reported protocol [14,19].

Structural Properties of CoPd Nanocatalysts
The crystal structure and surface atomic arrangements of pristine CoPd and laserirradiated CoPd nanocatalysts were revealed by HRTEM. Figure 1a shows the HRTEM image of the pristine CoPd nanocatalyst. Accordingly, the majority of Pd domains were covered by a thin layer of amorphous CoOx (denoted in a white square; region (a-1)), while the minority of the Pd domains were exposed to the surface (denoted by yellow circles; region (a-2)). Such characteristics were obvious due to the high extent of galvanic replacement reaction between the Co atoms and Pd 2+ ions (Co + Pd 2+ → Co 3+ + Pd 0 ), followed by redeposition of the residual Pd 2+ and Co 3+ ions [21]. These observations were further confirmed by the Fourier transformation (FFT), inverse Fourier transform (IFT) patterns, and their corresponding line histograms, where ring-like FFT patterns in inner lattices and fuzzy patterns in outer space (a-1), respectively, corresponded to the existence of polycrystalline Pd NPs covered by amorphous CoOx [22]. In addition, symmetrically aligned bright spots in the FFT pattern of the region (a-2) suggested the formation of locally ordered Pd NPs. Moreover, the line histograms determined that the interlayer (d)-spacing of regions (a-1) and (a-2) were 0.189 and 0.222, respectively, which corresponded to the Co 3 O 4 (130) (mp-1271793) and Pd (111), which is in good agreement with the aforementioned observations. Furthermore, a significant surface restructure was observed when the CoPd nanocatalyst was exposed to the laser with 1 mJ per-pulse energy (i.e., CoPd-1). As shown in Figure 1b, the majority of the Pd domains were exposed to the surface (denoted by yellow circles; region (b-2)) and could be attributed to the removal of the surface amorphous CoOx layer due to laser irradiation. However, some of the amorphous CoOx domains were still present on the surface (denoted in a white square; region (b-1)) due to limited per-pulse energy (i.e., 1 mJ). These scenarios were cross-referenced by the FFT patterns and linehistogram-determined d-spacing, where symmetrical aligned bright spots in both regions indicated the formation of long-range ordered structures due to the removal of surface oxide. Further raising the per-pulse energy to 10 mJ led to the formation of a completely different nanoarchitecture for the CoPd-10 nanocatalyst (Figure 1c), where the surface oxide layer was completely removed, and long-range ordered Pd NPs with twin boundaries (denoted by red circles) were formed on the surface. Moreover, the higher per-pulse energy (i.e., 10 mJ) induced a high extent of atomic migration; therefore, some of the Co-atoms from the CoOx support were deposited in the defect sites of Pd NPs (denoted by yellow squares). Meanwhile, the yellow arrows denotes surface defects of Pd NPs. In this way, atomic CoOx-species-decorated Pd NPs were formed on the CoOx support underneath. These scenarios were further confirmed by FFT patterns and d-spacing values, where similar FFT patterns and d-spacing values were observed on the whole surface, confirming the uniform distribution of decorated CoOx species.
CoOx support underneath. These scenarios were further confirmed by FFT patterns and d-spacing values, where similar FFT patterns and d-spacing values were observed on the whole surface, confirming the uniform distribution of decorated CoOx species. The effect of laser irradiation with different per-pulse energies on the local atomic and electronic properties of CoPd nanocatalyst was explored by cross-referencing the results of the Co K-edge and Pd K-edge XAS analyses. Figure 2a shows the X-ray absorption near-edge spectroscopy (XANES) spectra of pristine and laser-irradiated CoPd nanocatalysts, while the XANES spectra of standard Co foil and CoO were compared for reference. The XANES spectra at Co K-edge showed three main regions, including the pre-edge "R", the position of the inflection point (IS), and the intensity of the absorption edge (or whiteline) (HS), which, respectively, corresponded to the local geometry around the Co atoms, the oxidation state, and the extent of occupied/unoccupied state (due to electronic interaction with neighbouring atoms) of the targeting atom [23,24]. As shown in Figure 2a, the CoPd, CoPd-1, and CoPd-10 nanocatalysts exhibited similar features in all three regions, suggesting the unchanged local atomic and electronic properties of CoOx in CoPd after laser irradiation. In addition, the inflection point position of experimental nanocatalysts, The effect of laser irradiation with different per-pulse energies on the local atomic and electronic properties of CoPd nanocatalyst was explored by cross-referencing the results of the Co K-edge and Pd K-edge XAS analyses. Figure 2a shows the X-ray absorption near-edge spectroscopy (XANES) spectra of pristine and laser-irradiated CoPd nanocatalysts, while the XANES spectra of standard Co foil and CoO were compared for reference. The XANES spectra at Co K-edge showed three main regions, including the pre-edge "R", the position of the inflection point (I S ), and the intensity of the absorption edge (or whiteline) (H S ), which, respectively, corresponded to the local geometry around the Co atoms, the oxidation state, and the extent of occupied/unoccupied state (due to electronic interaction with neighbouring atoms) of the targeting atom [23,24]. As shown in Figure 2a, the CoPd, CoPd-1, and CoPd-10 nanocatalysts exhibited similar features in all three regions, suggesting the unchanged local atomic and electronic properties of CoOx in CoPd after laser irradiation. In addition, the inflection point position of experimental nanocatalysts, similar to that of CoO, implied that the Co was present in the form of CoO in all sam-ples. Figure 2b shows the Fourier-transformed extended X-ray absorption fine structure (FT-EXAFS) spectra of experimental samples at Co K-edge; the corresponding structural parameters are summarized in Table 1, and the fitting curves are shown in Figure S1. The peaks "P" and "Q" in Figure 2b correspond to the Co-O and Co-Co/Pd bond pairs. Meanwhile, as listed in Table 1, the Co-O bond pairs exhibited a higher coordination number (CN) as compared with the Co-Co and Co-Pd bond pairs, confirming that the Co was present in oxidized form, which is in good agreement with the aforementioned results. An even closer inspection of model-simulated fitting results revealed that CoPd-10 exhibited the highest CN for the Co-O bond pair (CN Co-O = 2.84), suggesting that Co atoms were exposed to the surface at the highest extent. These scenarios are in good agreement with the HRTEM results, where atomic CoOx species were observed on the surface of Pd NPs. Furthermore, Figure 2c shows the XANES spectra of experimental samples compared with the Pd foil and AC-supported Pd NPs, where peaks A and B correspond to the 1s → 5p-4f electron transitions [25,26]. Accordingly, the inflection point position of the experimental samples, similar to that of the Pd foil, confirmed that Pd was present in metallic form. Moreover, the lowest white-line intensities (H A and H B ) implied the lowest empty state in Pd 5p/4f orbitals due to the highest extent of electron relocation from Co to Pd in the CoPd-10 nanocatalyst [27]. Figure 2d shows the FT-EXAFS spectra of experimental samples at Pd K-edge; the fitting parameters are summarized in Table 1, and the fitting curves are shown in Figure S2. Accordingly, nearly similar CN for Pd-O, Pd-Co, and Pd-Pd bond pairs indicateed that the Pd domains were less affected by laser irradiation. However, the lowest radial distance for the Pd-Co bond pair (R Pd-Co = 3.102) again confirmed that the CoOx species were decorated in the defect sites of Pd NPs in CoPd-10. Moreover, the CN for the Pd-O bond pair was higher for CoPd-1 and CoPd-10 as compared with pristine CoPd, indicating a certain extent of Co oxide anchored on the metallic Pd defect sites. similar to that of CoO, implied that the Co was present in the form of CoO in all samples. Figure 2b shows the Fourier-transformed extended X-ray absorption fine structure (FT-EXAFS) spectra of experimental samples at Co K-edge; the corresponding structural parameters are summarized in Table 1, and the fitting curves are shown in Figure S1. The peaks "P" and "Q" in Figure 2b correspond to the Co-O and Co-Co/Pd bond pairs. Meanwhile, as listed in Table 1, the Co-O bond pairs exhibited a higher coordination number (CN) as compared with the Co-Co and Co-Pd bond pairs, confirming that the Co was present in oxidized form, which is in good agreement with the aforementioned results. An even closer inspection of model-simulated fitting results revealed that CoPd-10 exhibited the highest CN for the Co-O bond pair (CNCo-O = 2.84), suggesting that Co atoms were exposed to the surface at the highest extent. These scenarios are in good agreement with the HRTEM results, where atomic CoOx species were observed on the surface of Pd NPs. Furthermore, Figure 2c shows the XANES spectra of experimental samples compared with the Pd foil and AC-supported Pd NPs, where peaks A and B correspond to the 1s → 5p-4f electron transitions [25,26]. Accordingly, the inflection point position of the experimental samples, similar to that of the Pd foil, confirmed that Pd was present in metallic form. Moreover, the lowest white-line intensities (HA and HB) implied the lowest empty state in Pd 5p/4f orbitals due to the highest extent of electron relocation from Co to Pd in the CoPd-10 nanocatalyst [27]. Figure 2d shows the FT-EXAFS spectra of experimental samples at Pd K-edge; the fitting parameters are summarized in Table 1, and the fitting curves are shown in Figure S2. Accordingly, nearly similar CN for Pd-O, Pd-Co, and Pd-Pd bond pairs indicateed that the Pd domains were less affected by laser irradiation. However, the lowest radial distance for the Pd-Co bond pair (RPd-Co = 3.102) again confirmed that the CoOx species were decorated in the defect sites of Pd NPs in CoPd-10. Moreover, the CN for the Pd-O bond pair was higher for CoPd-1 and CoPd-10 as compared with pristine CoPd, indicating a certain extent of Co oxide anchored on the metallic Pd defect sites.   To gain more insight into the electronic interaction between the Co and Pd domains and to elucidate the binding energies of the constituting elements, the XPS was employed. Figure 3a,b show the comparative XPS spectra of experimental samples at Pd-3d and Co-2p orbitals. As shown in Figure 3a, the doublet peaks at~335.6 eV and~340.9 eV, respectively, corresponded to the Pd-3d 5/2 and Pd-3d 3/2 orbitals, where the lower binding energy was observed for CoPd-1 and CoPd-10 samples at Pd-3d orbitals as compared with pristine CoPd. Such characteristics suggest to some extent the electron relocation from Co-to-Pd atoms and are consistent with the aforementioned Pd K-edge XAS analysis. The results of XPS analysis at the Co-2p core level (Figure 3b) confirmed these scenarios, where the CoPd-1 and CoPd-10 exhibited higher binding energy at Co-2p orbitals as compared with pristine CoPd. These results integrally confirmed the electron localization from Co-to Pd atoms in laser-irradiated samples.  To gain more insight into the electronic interaction between the Co and Pd domains and to elucidate the binding energies of the constituting elements, the XPS was employed. Figure 3a,b show the comparative XPS spectra of experimental samples at Pd-3d and Co-2p orbitals. As shown in Figure 3a, the doublet peaks at ~335.6 eV and ~340.9 eV, respectively, corresponded to the Pd-3d5/2 and Pd-3d3/2 orbitals, where the lower binding energy was observed for CoPd-1 and CoPd-10 samples at Pd-3d orbitals as compared with pristine CoPd. Such characteristics suggest to some extent the electron relocation from Co-to-Pd atoms and are consistent with the aforementioned Pd K-edge XAS analysis. The results of XPS analysis at the Co-2p core level (Figure 3b) confirmed these scenarios, where the CoPd-1 and CoPd-10 exhibited higher binding energy at Co-2p orbitals as compared with pristine CoPd. These results integrally confirmed the electron localization from Co-to Pd atoms in laser-irradiated samples. The CO-stripping analysis was utilized to explain the effect of laser irradiation on the surface chemical identities of the CoPd nanocatalyst. Figure 4 shows the CO stripping curves of CoPd, CoPd-1, and CoPd-10 samples compared with the control samples (Pd-AC and Co-AC). Accordingly, the nearly flattened current responses for AC-supported The CO-stripping analysis was utilized to explain the effect of laser irradiation on the surface chemical identities of the CoPd nanocatalyst. Figure 4 shows the CO stripping curves of CoPd, CoPd-1, and CoPd-10 samples compared with the control samples (Pd-AC and Co-AC). Accordingly, the nearly flattened current responses for AC-supported Co NPs (i.e., the Co-AC) suggested their inert behaviour towards CO molecules [23]. Meanwhile, the AC-supported Pd NPs (i.e., Pd-AC) exhibited a sharp CO-oxidation peak (P) at~0.955 V vs. NHE, which corresponded to the CO-oxidation from the closely packed (111) facet. However, the absence of CO-oxidation peaks at lower potentials suggested that AC-supported Pd NPs had less selectivity between open and closed facets for Cooxidation. For pristine CoPd, a slight offset of the main CO oxidation peak P indicated the reduced energy barrier for CO oxidation as compared with Pd-AC and could be attributed to some extent to the electron localization from Co-to-Pd atoms due to the electronegativity difference and lattice mismatch (consistently conformed with the XPS spectra of pristine CoPd (Figure S3)). Moreover, the presence of an additional peak Q referred to the CO oxidation at the low energy barrier reaction sites (e.g., heterogenous Co-Pd interface or CoPd alloys due to the limited extent of heteroatomic intermixing). Meanwhile, the suppression of the CO-oxidation peak could be attributed to the surface coverage of Pd sites by a thin layer of amorphous CoOx, as consistently proved by the flattened CO-stripping curve of Co-AC (i.e., inert behaviour of CoO x towards CO molecules). For CoPd-1 and CoPd-10 samples, the absence of a peak (Q) suggested the removal of the surface oxide layer after laser irradiation. Such scenarios were further confirmed by the increased intensity of the CO-oxidation peak for CoPd-1, where very less atomic CoOx species were decorated on the surface of Pd NPs due to limited atomic migration (less per-pulse energy of 1 mJ). On the other hand, the CoPd-10 sample showed a CO-oxidation peak at the lowest potential as compared with the CoPd and CoPd-1 nanocatalysts, which could be attributed to the highest extent of electron relocation from Co-to-Pd atoms and is in good agreement with the XAS and XPS results. Meanwhile, the suppressed peak intensity of CoPd-10 as compared with CoPd-1 corresponded to the high density of atomic CoOx species on the surface of Pd NPs due to a high per-pulse energy and, thus, a higher extent of atomic migration. Co NPs (i.e., the Co-AC) suggested their inert behaviour towards CO molecules [23]. Meanwhile, the AC-supported Pd NPs (i.e., Pd-AC) exhibited a sharp CO-oxidation peak (P) at ~0.955 V vs. NHE, which corresponded to the CO-oxidation from the closely packed (111) facet. However, the absence of CO-oxidation peaks at lower potentials suggested that AC-supported Pd NPs had less selectivity between open and closed facets for Cooxidation. For pristine CoPd, a slight offset of the main CO oxidation peak P indicated the reduced energy barrier for CO oxidation as compared with Pd-AC and could be attributed to some extent to the electron localization from Co-to-Pd atoms due to the electronegativity difference and lattice mismatch (consistently conformed with the XPS spectra of pristine CoPd (Figure S3)). Moreover, the presence of an additional peak Q referred to the CO oxidation at the low energy barrier reaction sites (e.g., heterogenous Co-Pd interface or CoPd alloys due to the limited extent of heteroatomic intermixing). Meanwhile, the suppression of the CO-oxidation peak could be attributed to the surface coverage of Pd sites by a thin layer of amorphous CoOx, as consistently proved by the flattened CO-stripping curve of Co-AC (i.e., inert behaviour of CoOx towards CO molecules). For CoPd-1 and CoPd-10 samples, the absence of a peak (Q) suggested the removal of the surface oxide layer after laser irradiation. Such scenarios were further confirmed by the increased intensity of the CO-oxidation peak for CoPd-1, where very less atomic CoOx species were decorated on the surface of Pd NPs due to limited atomic migration (less per-pulse energy of 1 mJ). On the other hand, the CoPd-10 sample showed a CO-oxidation peak at the lowest potential as compared with the CoPd and CoPd-1 nanocatalysts, which could be attributed to the highest extent of electron relocation from Co-to-Pd atoms and is in good agreement with the XAS and XPS results. Meanwhile, the suppressed peak intensity of CoPd-10 as compared with CoPd-1 corresponded to the high density of atomic CoOx species on the surface of Pd NPs due to a high per-pulse energy and, thus, a higher extent of atomic migration. By cross-referencing the outcomes of the aforementioned physical and electrochemical characterization, the atomic structures of pristine CoPd, CoPd-1, and CoPd-10 nanocatalysts were proposed and are shown in Scheme 1. Accordingly, the pristine CoPd By cross-referencing the outcomes of the aforementioned physical and electrochemical characterization, the atomic structures of pristine CoPd, CoPd-1, and CoPd-10 nanocatalysts were proposed and are shown in Scheme 1. Accordingly, the pristine CoPd nanocatalyst comprised the CoOx-supported Pd nanoparticles with a thick oxide layer on the surface due to the high extent of galvanic replacement reaction between Co atoms and Pd 2+ ions (Co + Pd 2+ → Co 3+ + Pd 0 ), followed by redeposition of residual Pd 2+ and Co 3+ ions. Furthermore, the CoPd nanocatalyst was subjected to pulsed laser irradiation with the per-pulse energy of 1 mJ (i.e., CoPd-1). In this case, due to the relatively lower per-pulse energy of 1 mJ, the atomic migration was limited; thus, the CoOx species was deposited in the defect sites of Pd NPs to a lesser extent, while an unconformable thin oxide layer still existed on the surface of the CoPd-1 nanocatalyst. When the per-pulse energy was increased to 10 mJ, the surface oxide layer was completely removed from the surface of the CoPd-10 nanocatalyst, and a high density of atomic CoOx species were anchored in the defect sites as well as on the surface of the Pd NPs. nanocatalyst comprised the CoOx-supported Pd nanoparticles with a thick oxide layer on the surface due to the high extent of galvanic replacement reaction between Co atoms and Pd 2+ ions (Co + Pd 2+ → Co 3+ + Pd 0 ), followed by redeposition of residual Pd 2+ and Co 3+ ions. Furthermore, the CoPd nanocatalyst was subjected to pulsed laser irradiation with the per-pulse energy of 1 mJ (i.e., CoPd-1). In this case, due to the relatively lower per-pulse energy of 1 mJ, the atomic migration was limited; thus, the CoOx species was deposited in the defect sites of Pd NPs to a lesser extent, while an unconformable thin oxide layer still existed on the surface of the CoPd-1 nanocatalyst. When the per-pulse energy was increased to 10 mJ, the surface oxide layer was completely removed from the surface of the CoPd-10 nanocatalyst, and a high density of atomic CoOx species were anchored in the defect sites as well as on the surface of the Pd NPs. The catalytic performances of experimental nanocatalysts were evaluated across the temperature range from room temperature (RT) to 573 K at ambient pressure in a gas chromatography (GC) system equipped with a PDHID detector under a flowing reaction gas of H2/CO2 (3/1). Figure 5a shows the CO production yield of experimental nanocatalysts in the reaction gas of CO2:H2 = 1:3. Accordingly, the pristine CoPd nanocatalyst was chemically inert toward CO2 until reaching a temperature of 423 K, while the CO production yields for laser-irradiated CoPd-1 and CoPd-10 nanocatalysts were 6.70 µmol g −1 catalyst and 7.01 µmol g −1 catalyst, respectively, at 423 K, suggesting the laser-irradiation-mediated surface restructure successfully decreased the onset temperature of the CoPd nanocatalyst for CO production by 50 °C. The laser-irradiated CoPd-1 and CoPd-10 nanocatalysts exhibited higher CO production yields across the temperature range as compared with the pristine CoPd, where the CoPd-10 nanocatalyst achieved the highest CO production yield of ∼1667 µmol g −1 catalyst at 573 K. This value reflected improvements of ∼41% and ∼33% as compared with the pristine CoPd (∼976 µmol g −1 catalyst) and CoPd-1 (∼1114 µmol g −1 catalyst) nanocatalysts, respectively. In addition, the CoPd-10 nanocatalyst attained CO selectivity as high as ∼80%. Such a high catalytic performance at high temperature range was obvious because H2 splitting on Pd reaction sites increased at high temperatures. In this case, the hydrogenation rate of adsorbed CO2 (i.e., formation of *COOH) increased, thus increasing the CO production yield. The results of physical characterizations along with the electrochemical analysis suggested that such an improved catalytic performance of CoPd-10 nanocatalyst originated from the synergistic cooperation between the surface-anchored atomic CoOx species and adjacent Pd domains. Thus, the Pd reaction sites were favourable for H2 splitting, while adjacent atomic-scale metal oxide species with possible oxygen Scheme 1. The schematic representation for the atomic structures of CoPd, CoPd-1, and CoPd-10 nanocatalysts.
The catalytic performances of experimental nanocatalysts were evaluated across the temperature range from room temperature (RT) to 573 K at ambient pressure in a gas chromatography (GC) system equipped with a PDHID detector under a flowing reaction gas of H 2 /CO 2 (3/1). Figure 5a shows the CO production yield of experimental nanocatalysts in the reaction gas of CO 2 :H 2 = 1:3. Accordingly, the pristine CoPd nanocatalyst was chemically inert toward CO 2 until reaching a temperature of 423 K, while the CO production yields for laser-irradiated CoPd-1 and CoPd-10 nanocatalysts were 6.70 µmol g −1 catalyst and 7.01 µmol g −1 catalyst , respectively, at 423 K, suggesting the laser-irradiation-mediated surface restructure successfully decreased the onset temperature of the CoPd nanocatalyst for CO production by 50 • C. The laser-irradiated CoPd-1 and CoPd-10 nanocatalysts exhibited higher CO production yields across the temperature range as compared with the pristine CoPd, where the CoPd-10 nanocatalyst achieved the highest CO production yield of ∼1667 µmol g −1 catalyst at 573 K. This value reflected improvements of ∼41% and ∼33% as compared with the pristine CoPd (∼976 µmol g −1 catalyst ) and CoPd-1 (∼1114 µmol g −1 catalyst ) nanocatalysts, respectively. In addition, the CoPd-10 nanocatalyst attained CO selectivity as high as ∼80%. Such a high catalytic performance at high temperature range was obvious because H 2 splitting on Pd reaction sites increased at high temperatures. In this case, the hydrogenation rate of adsorbed CO 2 (i.e., formation of *COOH) increased, thus increasing the CO production yield. The results of physical characterizations along with the electrochemical analysis suggested that such an improved catalytic performance of CoPd-10 nanocatalyst originated from the synergistic cooperation between the surface-anchored atomic CoOx species and adjacent Pd domains. Thus, the Pd reaction sites were favourable for H 2 splitting, while adjacent atomic-scale metal oxide species with possible oxygen vacancies promoted CO 2 adsorption, followed by reduction. [28] Therefore, it can be concluded that the Pd and adjacent atomic scale CoOx reaction sites synergistically triggered the H 2 dissociation and CO 2 activation steps during the CO 2 reduction reaction (CO 2 RR). Furthermore, as shown in Figure 5b, the experimental nanocatalysts showed a much lower CH 4 production yield as compared with the CO production yield in the reaction gas, which confirmed the suppressed competitive CO 2 methanation process, and, therefore, the high CO-selectivity was achieved. The plausible reaction mechanism of the RWGS reaction on the surface of the CoPd-10 nanocatalyst is shown in Figure 5c. vacancies promoted CO2 adsorption, followed by reduction. [28] Therefore, it can cluded that the Pd and adjacent atomic scale CoOx reaction sites synergistically tr the H2 dissociation and CO2 activation steps during the CO2 reduction reaction (C Furthermore, as shown in Figure 5b, the experimental nanocatalysts showed a lower CH4 production yield as compared with the CO production yield in the r gas, which confirmed the suppressed competitive CO2 methanation process, and fore, the high CO-selectivity was achieved. The plausible reaction mechanism RWGS reaction on the surface of the CoPd-10 nanocatalyst is shown in Figure 5c.

Conclusions
Rational manipulation of surface atomic arrangements is a cardinal performa termining factor for heterogeneous catalysts. Herein, we used a sub-millisecond laser annealing technique with different per-pulse energies to manipulate the surfa figuration of cobalt oxide supported Pd (CoPd) nanoparticles for enhanced reverse gas shift (RWGS) reaction. For the optimum case, when the per-pulse energy was the CoPd-10 nanocatalyst exhibited the highest CO production yield of ∼1667 µm alyst at 573 K, with the CO selectivity as high as ∼80%. This CO production yield hanced ∼41% and ∼33% as compared with the pristine CoPd (∼976 µmol g −1 cata CoPd-1 (∼1114 µmol g −1 catalyst) nanocatalysts, respectively. The results of physical gations, electrochemical analysis, and gas chromatography (GC) results indicated enhanced catalytic activity and selectivity of the CoPd-10 nanocatalyst originate

Conclusions
Rational manipulation of surface atomic arrangements is a cardinal performancedetermining factor for heterogeneous catalysts. Herein, we used a sub-millisecond pulsed laser annealing technique with different per-pulse energies to manipulate the surface configuration of cobalt oxide supported Pd (CoPd) nanoparticles for enhanced reverse water-gas shift (RWGS) reaction. For the optimum case, when the per-pulse energy was 10 mJ, the CoPd-10 nanocatalyst exhibited the highest CO production yield of ∼1667 µmol g −1 catalyst at 573 K, with the CO selectivity as high as ∼80%. This CO production yield was enhanced ∼41% and ∼33% as compared with the pristine CoPd (∼976 µmol g −1 catalyst ) and CoPd-1 (∼1114 µmol g −1 catalyst ) nanocatalysts, respectively. The results of physical investigations, electrochemical analysis, and gas chromatography (GC) results indicated that the enhanced catalytic activity and selectivity of the CoPd-10 nanocatalyst originated from the potential synergy between surface-anchored atomic CoOx species and neighbouring Pd reaction sites, which, respectively, promoted the CO 2 activation and H 2 splitting. Briefly stated, the obtained results are expected to mark a step ahead in designing high-performance nanocatalysts for various redox reactions by using the pulsed laser technique.