To face the growing concerns about the shortage of fossil fuels, the importance of CO serving as C1 building block to produce compounds has attracted increasing attentions. The Fischer-Tropsch synthesis (FTS) is a well-known industrial-scale process to produce liquid products using syngas, a mixture of CO and H₂ [1-5]. Transition metal catalysts play vital roles in the FTS process to drive the crucial steps, such as the reductive cleavage of CO and C-C coupling, to proceed under relatively milder conditions [6]. The C-C coupling step is crucial to obtain multi-carbon products [6-8], while it is difficult to control the length of carbon chain and a wide distribution of hydrocarbons is generally produced in the final outputs. A fundamental understanding on the factors that can affect the C-C coupling reaction is of great importance to guide the design of advanced catalysts and then shift a specific reaction toward a desirable product. However, the details on related elementary steps are obscure because the information on chemical bonding is difficult to obtain in real-life catalysis.

A specific chemical reaction is closely related to a limited number of atoms. In this case, atomic clusters can be the ideal active sites on heterogeneous catalysts [9-21]. The fascinating feature of gas-phase atomic clusters is that cluster reactions can be performed under isolated conditions to eliminate the effects from solvent, defects, impurity, and so on, and then the nature of bond formation and cleavage behaviors can be obtained at a strictly molecular level. The polymerization of surface CH$ _x $ units is the generally accepted process to obtain multi-carbon compounds [5]. It has also been demonstrated that carbon can induce the reconstruction of catalysts [22-27] in the FTS process and the surface with terminated metal carbide is more reactive and can function as precursors to couple with CO [28, 29]. Thus, metal carbide can represent the active sites on the surface of practice catalysts to investigate the C-C coupling mechanism. Moreover, transition metal carbides have attracted growing attention as important catalytic materials for a variety of energy-related reactions in the field of heterogeneous catalysis [30]. In the field of gas-phase studies, the geometrical and electronic structures of metal carbide species have been extensively studied [31, 32], in contrast, the reactivity was relatively rarely observed. The reactions of FeC$ _n $$ ^- $ ($ n $ = 3, 6), Ta₂C₄$ ^- $, MoC$ ^+ $, and Mo₂C₂$ ^- $ species with CH₄ [33-37], Ta₂C₄$ ^- $ and TaC$ _n $$ ^+ $ ($ n $ = 1-4) with N₂ [38, 39], MoC₃$ ^- $ with C₂H$ _6 $ [40], and TaC$ _n $$ ^- $ ($ n $ = 1-14) with D₂ [41] have been reported, while the related reaction of metal carbide species with CO is not available, to our knowledge. In this study, the reactions of the CuC₃H$ ^- $ and CuC₃$ ^- $ cluster anions with CO have been investigated by using mass spectrometry, photoelectron imaging spectroscopy, and quantum chemical calculations. The experimental results showed that the CuC₃H$ ^- $ cluster can induce more effective and selective coupling of the C₃H$ ^- $ moiety with CO to generate ion product COC₃H$ ^- $ than the reaction of CuC₃$ ^- $ with CO. The crucial roles of the attached H atom on CuC₃H$ ^- $ to improve C-C coupling have been rationalized by using theoretical calculations.

The CuC₃H$ ^- $, CuC₃D$ ^- $, and CuC₃$ ^- $ cluster anions were generated by laser ablation of a $ ^{63} $Cu disk in the presence of 0.5%CH₄/CD₄ seeded in a He carrier gas with the backing pressure of 5.0 standard atmospheres. The cluster ions of interest (CuC₃H$ ^- $, CuC₃D$ ^- $, and CuC₃$ ^- $) were mass-selected using a quadrupole mass filter (QMF) and then entered into a linear ion trap (LIT) reactor, where they were cooled by collisions with a pulse of He gas and then interacted with a pulse of CO for around 1.3 ms and 2.3 ms for CuC₃H$ ^- $ (CuC₃D$ ^- $)+CO and CuC₃$ ^- $+CO, respectively. The temperature of cooling gas (He), reactant gases (CO), and the LIT reactor was around 298 K. The cluster ions ejected from the LIT were detected by a reflector time-of-flight mass spectrometer (TOF-MS). The details of running the TOF-MS [42], the QMF [43], and the LIT [44] can be found in our previous work. To calculate the reaction efficiencies (the possibilities of reaction upon each collision), the collision rate constants were calculated on the basis of the surface charge capture model developed in Ref.[45].

The photoelecron imaging spectroscopy (PEIS) experiments were carried out with a separate TOF-MS equipped with a laser-ablation source, a cryogenic ion trap, and a photoelectron imaging spectromter. The CuC₃H$ ^- $ clusters were generated according to the procedure described in the reactivity experiment. After generation, the cluster anions were directed into a linear octopole ion trap. The ion trap was held at 10 K and filled with pulsed bursts of precooled buffer gas of He. The ions were traped in the ion trap for around 80 ms and then extracted from the trap into a TOF-MS. After mass selection by a mass gate, the clusters reached the interaction region, where they were photo-detached with a wavelenth-tunable laser beam delivered from an optical parametric oscillator laser source. In the room temperature PEIS, the ion trap ran at the ion-guided mode. The kinetic energies or velocities of the photo-detached electrons were measured by the photoelectron imaging spectromerter. The details about performing the experiment can be found in a previous study [46].

Density functional theory (DFT) calculations using the Gaussian 09 [47] program were carried out to study the mechanisms on the reactions of CuC₃H$ ^- $ and CuC₃$ ^- $ cluster anions with CO. The TPSS functional has been demonstrated to perform well for the Cu-related system [48], thus the results by TPSS [49] were given throughout this work. The 6-311+G(d) basis sets were used for all the atoms. The relaxed potential energy surface scans were used to obtain good guess structures for the intermediates (Is) and the transition states (TSs) along the pathways. The TSs were optimized by using the Berny algorithm method [50]. Vibrational frequency calculations were performed to check that each I or TS has zero or only one imaginary frequency, respectively. Intrinsic reaction coordinate [51, 52] calculations were performed so that each TS connects two appropriate local minima.

The TOF mass spectra for the interactions of laser-ablation generated and mass-selected CuC₃H$ ^- $, CuC₃D$ ^- $, and CuC₃$ ^- $ cluster anions with CO are shown in FIG. 1. For the interaction of CuC₃H$ ^- $ with CO (FIG. 1(B, C)), the single product COC₃H$ ^- $ can be clearly identified, indicating that CuC₃H$ ^- $ can react with CO to evaporate a neutral Cu atom (reaction (1)). This channel was further confirmed by isotope labeling experiment (FIG. 1(E)). For the reaction of CuC₃$ ^- $ with CO (FIG. 1(G)), the evaporation of Cu atom can also take place to generate product COC₃$ ^- $ (reaction (2)), whereas CO adsorption products CuC₃(CO)$ ^- $ and CuC₃(CO)₂$ ^- $ also appeared. This experiment revealed that the attachment of a H atom on CuC₃H$ ^- $ can obviously lead to a higher product selectivity.

$ \begin{eqnarray} &&{\rm{CuC}}_3{\rm{H}}^ - + {\rm{CO}} \to {\rm{COC}}_3{\rm{H}}^ - + {\rm{Cu}} \end{eqnarray} $  (1)

$ \begin{eqnarray} &&{{\rm{CuC}}_3}^ - + {\rm{CO}} \to {{\rm{COC}}_3}^ - + {\rm{Cu}} \end{eqnarray} $  (2)

The pseudo-first-order rate constants ($ k_1 $) on the reactions of CuC₃H$ ^- $, CuC₃D$ ^- $, and CuC₃$ ^- $ with CO have been well fitted and the results are shown in FIG. S1 in supplementary materials. The fitted results indicate that about 31% of experimentally generated CuC₃H$ ^- $ is inert toward CO and the inert component for CuC₃D$ ^- $ is about 37%. The generation of cluster isomers with very different reactivity has frequently been observed in laser-ablation experiments [53-58]. The determined rate constants for the reactive component of CuC₃H$ ^- $ and CuC₃D$ ^- $ are about (3.4$ \pm $1.1)$ \times $10$ ^{-12} $ cm$ ^3\cdot $molecule$ ^{-1}\cdot $s$ ^{-1} $ and (4.8$ \pm $1.4)$ \times $10$ ^{-12} $ cm$ ^3\cdot $molecule$ ^{-1}\cdot $s$ ^{-1} $, corresponding to the reaction efficiencies of about (0.5$ \pm $0.1)% and (0.7$ \pm $0.2)% by using the surface charge capture model [45], respectively. The rate constant for reaction CuC₃$ ^- $+CO is about (0.8$ \pm $0.2)$ \times $10$ ^{-12} $ cm$ ^3\cdot $molecule$ ^{-1}\cdot $s$ ^{-1} $ and the reaction efficiency is about (0.10$ \pm $0.03)%. The comparable rate constants for reactions CuC₃H$ ^- $+CO and CuC₃D$ ^- $+CO indicate that the hydrogen atom transfer could not be the rate-determining step. The much lower rate constant for reaction CuC₃$ ^- $+CO emphasizes that the attached H atom on CuC₃H$ ^- $ can significantly improve the coupling of CO with the C₃H moiety in CuC₃H$ ^- $.

The DFT calculations were carried out to investigate the structures of the CuC₃H$ ^- $ and CuC₃$ ^- $ clusters and the mechanisms on the reactions of CuC₃H$ ^- $ and CuC₃$ ^- $ with CO. As shown in FIG. 2(A) and FIG. S2 (in the supplementary materials), several low-lying CuC₃H$ ^- $ isomers have been predicted. The DFT calculations show that the linear CuC₃H$ ^- $ isomer (IS1) in the doublet electronic state is the lowest-lying structure, while single-point energy calculations using the high level [CCSD(T)] [59] method suggest that IS4 is the most stable CuC₃H$ ^- $ isomer. The internal conversion among different isomers has been performed (FIG. 2(B) and FIG. S3 in the supplementary materials) and the results reveal that the high barriers may prevent their intra-conversion. Thus, isomers IS1-IS5 may all be generated in the cluster source. In contrast, the linear structure in the triplet electronic state with a terminal Cu atom was unambiguously identified as the lowest-lying isomer for CuC₃$ ^- $.

To further characterize the structure of CuC₃H$ ^- $, the anion photoelectron imaging spectroscopy experiments were performed to gain more information, and the results are shown in FIG. 3. For the spectra collected at 520 nm at 298 K (FIG. 3(A)) and 10 K (FIG. 3(B)), a well-resolved vibrational progression with a peak spacing of about 403 cm^-1 can be observed. These spectra are in agreement with the Franck-Condon (FC) simulated profile of isomer IS4 (FIG. 3(F), $ ^1 $A$ \leftarrow $$ ^2 $A) at 298 K. Moreover, the relatively weak bonds with binding energy of 1.48 eV to 2.28 eV in the 520 nm spectrum collected at 298 K are generally consistent with the simulation of isomer IS3 (FIG. 3(E)), which may also be generated in the experiment. The FC simulated profile of isomer IS2 (FIG. 3(D)) can reproduce the experimental spectrum collected at 700 nm at 298 K (FIG. 3(A)) very well, suggesting the possible generation of IS2 in the cluster source. The 380 nm spectrum collected at 298 K has a broad band centered at 2.82 eV (FIG. 3(A)). This spectrum can assign the population of IS1, which has a calculated vertical electron detachment energy (VDE) of 2.80 eV ($ ^1 $A$ \leftarrow $$ ^2 $A) (FIG. 3(C)) and matches the experimental positions well. In our experiment, it is difficult to estimate the possible presence of isomer IS5 because of its large VDE value of 3.50 eV.

The reactions of low-lying CuC₃H$ ^- $ isomers (IS1-IS5) with CO were all calculated (FIG. 4(A) and FIGs. S4 and S5 in supplementary materials) and the results demonstrated that the positive barriers on the reactions of IS4 and IS5 with CO make both reactions kinetically hindered, thus, IS4 and IS5 may account for the inert component of CuC₃H$ ^- $ generated in the experiment. The main text below focused on the pathway on reaction CuC₃H$ ^- $(IS1)+CO (FIG. 4(A)) because the pathways on reactions IS2/IS3+CO are very similar to or share the same part with pathway CuC₃H$ ^- $(IS1)+CO. CO can be trapped by the Cu site of CuC₃H$ ^- $(IS1) with a binding energy of 0.65 eV to form intermediate 1 (I1). Then the H atom jumps onto the nearby C site to overcome a small barrier of 0.20 eV (I1$ \rightarrow $I2). In the next step, a second Cu-C bond forms favorably to generate I3 (I2$ \rightarrow $I3), from which C-C coupling is ready to take place to suffer from a higher barrier of 0.76 eV (I3$ \rightarrow $I4). Then the neutral Cu atom is evaporated to give rise to product COC₃H$ ^- $. In contrast, for reaction CuC₃$ ^- $+CO, the C-C coupling ($ ^3 $I6$ \rightarrow $$ ^1 $I7) and the Cu atom release steps are more difficult to proceed without the presence of such H atoms (FIG. 4(B)). Theoretical calculations strongly support the experimental results that the CuC₃H$ ^- $ cluster reacts faster with CO (FIG. 1).

CO adsorption and dissociation on heterogeneous catalysts was generally accepted as the initial steps in the FTS process, then followed by the C-C coupling. The H-assisted pathway in CO activation has been proposed theoretically and experimentally because of the very high availability of surface hydrogen [60-64]. Recent progress demonstrated that direct coupling of CO with the terminal Mo carbide serves as the initial step in FTS [28], and the sequential addition of hydride that generates from the heterolytic cleavage of H₂ promotes greatly these early steps. Herein, the fundamental study on the H-assisted C-C coupling chemistry is not only scientifically interesting but also practically important to tailor catalysts. The pathway in FIG. 4(A) shows that the H atom seems to be a spectator and does not take part directly in the C-C coupling and the Cu-release steps. While further analysis evidences that the attached H atom is vital to change the geometrical and electronic structure of the CuC₃H$ ^- $ cluster during the reaction. Though the existence of multiple intermediates and transition states along the pathway on reaction CuC₃H$ ^- $+CO, the crucial steps, C-C coupling (I3$ \rightarrow $TS7$ \rightarrow $I4) and Cu-release (I5$ \rightarrow $P1), are energetically more favorable to proceed. In the presence of such H atoms, it is favorable to form a planar intermediate I3, which is flexible to couple with CO through the terminal C atom to surpass a barrier of 0.76 eV (FIG. 4(A)). This flexibility of the C₃-chain can be reflected by the frequently and easily changed C-C bond lengths along the pathway. In sharp contrast, the linear intermediate $ ^3 $I6 is rather rigid and C-C bond is reluctant to be formed between CO and the opposite C₃-chain, and a high barrier of 1.50 eV has to be crossed laboriously ($ ^3 $I6$ \rightarrow $$ ^3 $TS9$ \rightarrow $$ ^1 $I7, FIG. 4(B)). This fact emphasizes the capability of H atom to reconstruct the geometrical structure of the reaction system toward a more reactive pattern.

Natural charge analysis reveals that the H atom can also drive the charge redistribution on reaction system CuC₃HCO$ ^- $. In the final intermediate I5, the C atom that anchors the Cu atom is much less negatively charged (-0.68 e) with respect to that (-1.18 e) in CuC₃CO$ ^- $ ($ ^1 $I7, shown in FIG. 5). The leading result is that such C atom in $ ^1 $I7 is more attractive toward the positively charged Cu atom (+0.31 e), and the higher barrier of 2.56 eV ($ ^1 $I7$ \rightarrow $P₂) increases the difficulty for the Cu atom to struggle from CuC₃CO$ ^- $, while the nearly neutral Cu (-0.07 e) atom in I5 can be easily evaporated. This can also be reflected by the much longer Cu-C bond (194 pm) in I5 than that (181 pm) in $ ^1 $I7. The DFT-calculated VDEs for the final products COC₃H$ ^- $ and COC₃$ ^- $ are 2.61 and 3.70 eV, respectively, indicating that the COC₃ moiety is more attractive toward electrons. Thus, the Cu atom is positively charged in $ ^1 $I7 and gives rise to a tight COC₃-Cu interaction. Moreover, the reorganization energy calculated for COC₃H$ ^- $ (0.08 eV) is much smaller than that for COC₃$ ^- $ (0.17 eV), indicating that the presence of such H atoms is greatly helpful to prevent the structure of the final product COC₃H$ ^- $ from reorganization during the process of Cu atom evaporation and then makes the reaction energetically favorable. The capability of H atom to reshape the geometrical and electronic structure of atomic clusters has also been previously proposed [65]. This study may provide new insights into the paramount roles of surface attached hydrogen in the real-life FTS process to affect the C-C coupling reaction.

The reactions of the CuC₃H$ ^- $ and CuC₃$ ^- $ cluster anions with CO have been investigated by using mass spectrometry and density functional theory calculations. The structure of CuC₃H$ ^- $ was further characterized by using photoelectron imaging spectroscopy experiment. The experimental results showed that CO can couple with the C₃-moiety in CuC₃H$ ^- $ and CuC₃$ ^- $ to give rise to ion products COC₃H$ ^- $ or COC₃$ ^- $, respectively. The attached H atom can greatly increase the reactivity and product selectivity on the reaction of CuC₃H$ ^- $ with CO. Theoretical studies revealed that such H atom can depress the barriers on the C-C coupling and Cu-release steps for CuC₃H$ ^- $+CO, the crucial steps of both reactions. Direct coupling of CO with the terminal metal carbide on real-life catalysts has been scarcely demonstrated as the initial steps in the FTS process. This fundamental study is of great importance to permeate this process and the vital roles of surface attached hydrogen to assist C-C coupling were obtained.

Supplementary materials: Fitted rate constants on the reactions of CuC₃H$ ^- $, CuC₃D$ ^- $, and CuC₃H$ ^- $ with CO are available. DFT-calculated low-lying isomers for the CuC₃H$ ^- $ and CuC₃$ ^- $ clusters and the potential energy profiles for CuC₃H$ ^- $+CO are also given.

This work was supported by the National Natural Science Foundation of China (No.21773254) and the Youth Innovation Promotion Association Chinese Academy of Sciences (No.2016030).