Manipulating interstitial carbon atoms in the nickel octahedral site for highly efficient hydrogenation of alkyne

Light elements in the interstitial site of transition metals have strong influence on heterogeneous catalysis via either expression of surface structures or even direct participation into reaction. Interstitial atoms are generally metastable with a strong environmental dependence, setting up giant challenges in controlling of heterogeneous catalysis. Herein, we show that the desired carbon atoms can be manipulated within nickel (Ni) lattice for improving the selectivity in acetylene hydrogenation reaction. The radius of octahedral space of Ni is expanded from 0.517 to 0.524 Å via formation of Ni3Zn, affording the dissociated carbon atoms to readily dissolve and diffuse at mild temperatures. Such incorporated carbon atoms coordinate with the surrounding Ni atoms for generation of Ni3ZnC0.7 and thereof inhibit the formation of subsurface hydrogen structures. Thus, the selectivity and stability are dramatically improved, as it enables suppressing the pathway of ethylene hydrogenation and restraining the accumulation of carbonaceous species on surface. Design and synthesis of non-noble metal catalysts is crucial for highly-efficient hydrogenation. Here, the interstitial sites in nickel are manipulated with zinc and filled by dissociated carbon atoms, which drastically improve the selectivity/stability in acetylene hydrogenation reaction.

I nterstitial sites of transition metal catalyst are found to be occupied by dissociated atoms under a high chemical potential of reactant molecules [1][2][3][4] . Definitely, the interstitial atoms enable modifying electronic/geometric properties of surface atoms, which directly affects its Fermi level and density of states. Thus, the catalytic behaviors were tuned. For instance, the subsurface O in Cu nanoparticles (NPs) (~10 nm) impairs a positive charge to the Cu surface and thereof improves the catalytic performance in methanol steam reforming reaction 5 . More interestingly, the interstitial atoms are clarified to directly participate in reactions and afford the distinct reaction pathways from the surface-adsorbed atoms. So it leads to diverse products, reaction kinetics, and mechanisms. As evidenced by Ceyer et al. 2 , the interstitial H in Ni subsurface is found to be active for hydrogenation adsorbed CH 3 to CH 4 whereas the surfaceadsorbed H remains unreactive with CH 3 species. Initially, the interstitial site manipulated catalysis is considered to affect a few metal catalysts and associated reactions. More transition metalcatalyzed reactions, however, are controlled by the generation of metastable interstitial atoms [6][7][8][9] . Thus, it is of great interests to explore the control of interstitial site atoms in transition metal catalysts for tuning their catalytic performances [10][11][12] .
Acetylene selective hydrogenation, a well-known industrial reaction, with which a switchable selectivity towards distinct products is determined by the alternation of interstitial atoms in Pd catalyst 1,[13][14][15] . Experimental studies demonstrated that carbon or hydrogen atoms in the interstitial sites of Pd subsurface govern the selectivity to alkene or alkane 1,16 . Similar results were also observed in Ni, which is widely adopted as industrial hydrogenation catalyst (e.g., Raney Ni) for high activity and low cost in other heterogeneous reactions. However, it suffers from the drawbacks of low selectivity and stability in alkyne selective hydrogenation reaction 17,18 . Some investigations demonstrate that interstitial H atoms can readily hydrogenate adsorbed C 2 H 4 to C 2 H 6 . It drives C 2 H 4 from underneath with an orientation to the rehybridized π orbitals, although the surface H atoms remain unreactive 19 . These results reveal that it is the occupation of interstitial sites with carbon atom that is responsible, instead of hydrogen atoms, in Ni group catalysts. Following this behavior, the selectivity would be improved in hydrogenation. Thus, the formation of subsurface interstitial atoms was thereof focused on. Unfortunately, it is highly kinetically dependent and suffers from an input of external energies to stabilize; it remains challenging for the atomic-level control of desired atoms 14,20 . The common strategy to stabilize the metastable interstitial structure lies at construction of an encapsulated shell. Following this approach, PdC x @C structure was successfully prepared with the removal of the graphite layers after external procedures 21 . However, it is even difficult for the formation of Ni-based interstitial carbon structure because of the usually required high temperatures. The dissolved carbon atoms within Ni lattices are readily segregated on the surface. As a result, it offers the formation of graphite layers or even carbon nanotubes/nanofibers (CNTs/CNFs) under reaction conditions. Following the removal of carbon shells is even more challenging, during which Ni is readily oxidized and the interstitial carbon structure is damaged due to the oxyphilic nature of Ni. That is to say, it is crucial to modulate Ni environments in order to accommodate carbon atoms with directly exposed active surfaces instead of applying external process.
In general, the interstitial atoms prefer to locate in octahedral hole than at tetrahedral sites in view of the larger space with more available coordination atoms 22 . The radius of octahedral (R oct ) site in face-centered cubic (FCC) metals can be calculated by following the equation R oct = 0.414R M (R M represents the metallic radius of metal atom) 23 . Compared with Pd (0.569 Å), the smaller R oct in Ni (0.517 Å) leads partially to both a higher carbon dissolving temperature and an easier segregated carbon atom on surface. Therefore, a structural transformation from Ni (FCC) to Ni 3 C (hexagonal close-packed structure) occurs to accommodate a certain amount of carbon atoms, together with uncontrollable formation of graphitic layers 24,25 . Theoretical studies have evidenced that the expansion of metal lattice parameters leads readily to inclusion and diffusion of atoms at interstitial sites [26][27][28] . Second metal introduction can effectively tune the lattice parameter of Ni 29,30 . In addition, the electronic structure should not be changed in order to maintain the capability of Ni to dissociate adsorbed molecules. In Ni-Zn systems a Zn addition may expand the lattice parameters of Ni and generally leave a slightly modified electronic structure 30,31 .
Herein, R oct increases from 0.517 to 0.524 Å, provided if 25 at% Zn is introduced to Ni to form Ni 3 Zn, with a marginal change of electronic structure. Simple impregnation method and hydrogen/ acetylene treatment (the procedures are illustrated in Supplementary Fig. 1) readily generate the interstitial carbon structure Ni 3 ZnC 0.7 at low temperature, characterized by integrating in situ and ex situ techniques. It demonstrates that the supported Ni 3 ZnC 0.7 catalyst exhibits high selectivity at different ratios of H 2 /alkyne and an excellent stability in acetylene hydrogenation reaction. The dramatic improvement is ascribed to two issues: the interstitial carbon modulation of Ni surface/subsurface structures and the suppressed accumulation of carbonaceous fragments on the surface.

Results
Incorporation of carbon atom into Ni. Carbon materials with abundant defects, since providing additional chemical potential of carbon, are selected to probably diffuse and stabilize carbon in the interstitial sites of supported Ni-based structure 32 . In order to well disperse Ni and Zn ions, oxidized carbon nanotube (oCNT) is optimized for its high surface area and adjustable homogeneous functional sites 33 . Other generally adopted supports such as highsurface-area Al 2 O 3 is also an ideal candidate, but the desired structure has been not obtained ( Supplementary Fig. 2). For the preparation process, Ni or stoichiometric Ni/Zn salts were uniformly impregnated on oCNT ( Supplementary Fig. 3) and subjected to a temperature ramp under hydrogen atmosphere. During the reduction process, Ni promoted the reduction of Zn ions and contributed to the formation of Ni 3 Zn structure at 500°C under hydrogen atmosphere, as evidenced by in situ X-ray diffraction (XRD) experiments in Supplementary Fig. 4. A supported Ni catalyst was also reduced at 500°C for reference. The lattice parameter of Ni increased from 3.562 to 3.582 Å via the formation of Ni 3 Zn structure, as shown by the diffraction peak shifting from 44.5°to 43.7°of (111) plane in XRD patterns ( Supplementary Fig. 5). The radius of octahedral space is expanded from 0.517 to 0.524 Å (Fig. 1a) calculated by using the aforementioned equation.
Previous studies based on Ni-ZnO system have revealed that the formation of Ni-Zn alloy does not change the electronic structure of Ni much 31 . In addition, theoretical calculation exhibits that there is little hybridization between Ni 3d and Zn 3d states in Ni 3 Zn 30 . Thus, it seems that the electronic structure and the capability to dissociate hydrocarbon molecules of Ni have merely altered in Ni 3 Zn structure, as it is the initial step to form the interstitial carbon structure. Therefore, temperature-programmed surface reaction (TPSR) experiments were conducted to determine the dissociation temperature of acetylene on the surface of Ni and Ni 3 Zn. Acetylene was adopted in that it is the most reactive carbon source due to its highest change of Gibbs free energy in forming metal carbide structure 34 . The results in Fig. 1b, c exhibit that Ni 3 Zn structure maintain the strong capability to dissociate acetylene with an only 15°C delay compared to Ni. Although Ni exhibits the strong capability to dissociate acetylene, the XRD results ( Fig. 1d) revealed that the Ni NPs maintained in the FCC structure after 1.0 vol% C 2 H 2 /He treatment until 500°C. The dissociated carbon atoms tend to segregate from the Ni NPs and form graphitic layers or even CNTs/CNFs instead of the stable carbides. It has been observed that instead of Ni carbide formation, it is carbon encapsulation that prefers to occur to Ni carbide formation in Ni NPs supported on oxidized CNTs (Ni/oCNTs). TEM investigations confirm the encapsulation of graphitic layers on the FCC Ni NPs (Supplementary Fig. 6). The morphology of Ni NPs is greatly changed after the acetylene treatment, which is well consistent with the reported dynamic behavior and elongation of Ni NPs during CNT growth 35 . Ni 3 Zn exhibits a slightly reduced capability to dissociate acetylene in comparison with Ni. Interestingly, the carbon atom could readily dissolve and diffuse in Ni 3 Zn and form the Ni 3 ZnC 0.7 structure at low temperature under 1.0 vol% C 2 H 2 / He atmosphere. The lattice parameter of Ni 3 Zn is further enlarged with the carbon inclusion. As indicated by the diffraction peaks in XRD patterns (Fig. 1e), they shift left and also consistent well with Ni 3 ZnC 0.7 structure (a = b = c = 3.66 Å, JCPDS No. 28-0713).
Microstructural clarification of Ni 3 Zn and Ni 3 ZnC 0.7 . The crystal structures of Ni 3 Zn alloy were identified through XRD measurements: it has the expanded interstitial sites and the generated Ni 3 ZnC 0.7 structure with carbon incorporation. Following the microstructural features before and after carbon atom inclusion, an aberration-corrected TEM equipped with energydispersive X-ray spectroscopy (EDX) elemental analysis and electron energy loss spectroscopy (EELS) is applied. Firstly, highangle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images demonstrate the morphology of the catalysts. As shown in Fig. 2a, the formed Ni 3 Zn NPs are uniformly distributed on oCNT and exhibit an average particle size of 9.1 ± 1.8 nm. Through the dissociation of adsorbed acetylene and the dissolution of carbon atoms in the Ni 3 Zn structure, the particle size of Ni 3 ZnC 0.7 NPs increased slightly (9.8 ± 2.9 nm) ( Fig. 2b). High-resolution TEM images and corresponding fast Fourier transform (FFT) (Fig. 2c, d) Fig. 7) directions, respectively. The integrated pixel intensities are measured on the basis of the average lattice spacing over ten atomic layers (Fig. 2i). As a result, the average pristine Ni 3 Zn (111) (Fig. 2g, h). That is to say, a homogeneous distribution of Ni and Zn existed. Carbon (K-edge) EDX maps in Fig. 2h reveal the existence and uniform distribution of C in Ni 3 ZnC 0.7 NPs. It is different from Ni 3 Zn structure (Fig. 2g), and explicitly evidence the incorporation of carbon atom after acetylene treatment. The proportions of the atomic ratio in the Ni 3 ZnC 0.7 NPs are also confirmed via EDX line scan analysis ( Supplementary Fig. 8).
Subsequently, the body-centered carbon atom is identified by using EELS. The carbon K-edge EEL spectrum of Ni 3 ZnC 0.7 (Fig. 2j) exhibits three peaks in a range from 290 to 300 eV (291, 294, and 300 eV). Thus, it reveals a unique coordination structure of carbon atom and distinct from the sp 2 structure of carbon in the oCNTs. EELS signal of carbon is not found for Ni 3 Zn NP ( Supplementary Fig. 9), which is consistent with the EDX results. Catalytic performance in acetylene hydrogenation. Microscopic and spectroscopic results explicitly demonstrate that carbon atoms penetrate and occupy the interstitial site of Ni 3 Zn. It is supposed to inhibit the hydrogenation pathway of interstitial hydrogen and increase the selectivity towards ethylene. Thus, the selectivity is detected toward ethylene in acetylene hydrogenation pathway. Meanwhile, the phase transition process from Ni 3 Zn to Ni 3 ZnC 0.7 is monitored via an in situ XRD experiment (Fig. 3a) under a 1.0 vol% C 2 H 2 /He atmosphere. As shown in Fig. 3a, the diffraction peaks of the Ni 3 Zn (111) and (200) planes decreased gradually, whereas the intensity of Ni 3 ZnC 0.7 (111) and (200) peaks increased until a complete transformation of Ni 3 Zn to Ni 3 ZnC 0.7 occurred with a temperature preservation at 200°C during 1 h. During the subsurface structure changed from hydrogen to carbon atoms, the selectivity towards ethylene increased highly from 30.6% to 94.5% (Fig. 3b), while the conversion of acetylene decreased from 99.2% to 56.3%. In general, the intermediate interstitial structure is metastable and depend kinetically on the chemical potential of desired elements 14 . Herein, the Ni 3 ZnC 0.7 /oCNT catalyst exhibits a superior and stable selectivity for a variety of H 2 /C 2 H 2 ratios (Fig. 3c) due to the removal of the subsurface interstitial H hydrogenation pathway, which was evidenced by a series of H 2 temperatureprogrammed desorption experiments of Ni/oCNT, Ni 3 Zn/oCNT, and Ni 3 ZnC 0.7 /oCNT catalysts ( Supplementary Fig. 10). For a ratio in the range of 3-15, Ni 3 ZnC 0.7 /oCNT yields a 99.9-86.1% selectivity towards ethylene, and a 0.1-13.9% selectivity towards ethane in the conversion range of 9.7-93.5%. Generally, the reaction probabilities between adsorbed H and intermediate ethylene would increase as the coverage of activated H increases with the H 2 concentration. Also, it is the highly exothermic feature of ethylene hydrogenation that leads further to the increase of ethane selectivity and uncontrollable temperature runaway 13,36 . Therefore, the high selectivity of the Ni 3 ZnC 0.7 catalyst within a wide H 2 /C 2 H 2 ratio range is industrially important, which allows for stable operation under fluctuated conditions. In addition, the selectivity control under a constant high conversion is challenging but desirable to eliminate the trace amount acetylene in ethylene. As shown in Fig. 3d, the obtained Ni 3 ZnC 0.7 /oCNT catalyst exhibits 94% selectivity toward ethylene under 99% conversion in a 720 min reaction, which is superior to Ni/oCNT catalyst (27% selectivity at the initial 99% conversion). The Ni/oCNT catalyst exhibits a rapidly reduced activity and poor stability under reaction conditions. This is supposed to the segregation and accumulation of carbonaceous fragments on the Ni surface and consistent with previous studies 17, [37][38][39] . Obviously, the decreased conversion of alkyne for Ni catalyst cannot meet the industrial requirements and may poison the following polymerization catalysts. However, the Ni 3 ZnC 0.7 / oCNT catalyst exhibits excellent stability in comparison with Ni. That is, the introduction of carbon atoms not only inhibits the unselective hydrogenation pathway but also regulates the electronic structure of Ni. Therefore, it suppresses the further dissociation of adsorbed acetylene molecules. Thus, the stability  of Ni 3 ZnC 0.7 /oCNT catalyst is significantly improved in comparison with Ni ( Supplementary Fig. 11). The reaction carbon balance of Ni 3 ZnC 0.7 /oCNT catalyst was evaluated in the absence of ethylene. As shown in Supplementary Fig. 12 Supplementary Fig. 17, the fresh and spent Ni 3 ZnC 0.7 /oCNT catalysts exhibit a similar initial weight loss at 395 o C, which is supposed to the combustion of oCNT. The weight loss occurs at 240°C of the spent Ni 3 Zn/ oCNT, however, which is much lower than that of fresh Ni 3 Zn/ oCNT. It indicates the strongly adsorbed carbonaceous deposition on the surface of Ni 3 Zn after 10 h reaction at 100°C. These TGA results further prove that the carbonaceous accumulation is highly suppressed on the surface of Ni 3 ZnC 0.7 /oCNT catalysts.
These results indicate that the introduction of carbon atoms to the interstitial sites of Ni enables an efficient suppression of the over-hydrogenation reaction pathway and the carbonaceous accumulation on the surface (Supplementary Fig. 18). As a result, it affords the Ni 3 ZnC 0.7 /oCNT with high activity and selectivity, i.e., a stable catalyst in acetylene hydrogenation reaction. In addition, the selectivity control in acetylene hydrogenation is more challenging under high-pressure condition, which should be carefully taken into account 45,46 .
Electronic properties of Ni with Zn and C atom inclusion. The variation of the electronic structure of Ni caused by the charge transfer between C and Ni in Ni 3 ZnC 0.7 is demonstrated in Fig. 4. After calibration with reference samples (Ni foil and Ni oxide), the X-ray absorption near edge structure (XANES) spectra of the Ni K-edge were conducted for the Ni 3 Zn/oCNT and Ni 3 ZnC 0.7 / oCNT catalysts. It clearly shows that Ni is indeed electrondeficient in Ni 3 ZnC 0.7 compared with Ni and Ni 3 Zn (Fig. 4a). Furthermore, in situ XANES experiments reveal the evolution of the electronic structure of Ni from Ni 3 Zn to Ni 3 ZnC 0.7 . The intensity of the white line (Fig. 4b) increases from Ni 3 Zn gradually to Ni 3 ZnC 0.7 , which suggests Ni electron deficiency with the incorporation of carbon. The charge transfer distribution maps (Fig. 4c) 1.66). As a result, Ni has an electron richness of 0.138 per atom. After C incorporation, an obvious charge transfer occurs from Ni to C, rendering Ni with an electron deficiency of −0.133 per atom (Fig. 4d). Since the interaction is negligible between Zn and C (Fig. 4c), there is no charge transfer in this case. Therefore, the positive charge and the lower d-band center ( Supplementary Fig. 19) of Ni in Ni 3 ZnC 0.7 suppresses the dissociation of acetylene and weakens the adsorption of acetylene/ ethylene, which contributes to the decreased polymerization products. In short, it leads to improved stability for highly efficient acetylene hydrogenation.

Discussion
This work demonstrates that highly efficient hydrogenation of alkyne could be obtained via introducing carbon atoms within Ni lattice containing an expanded interstitial space. With an addition of 25% Zn, the interstitial space of Ni is expanded with a wellmaintained capability to adsorb and dissociate hydrocarbon molecules. Thus, the dissociated carbon atoms enable readily penetrating into Ni 3 Zn to offer the formation of Ni 3 ZnC 0.7 structure under acetylene contained atmosphere at 200°C. Both XRD and TEM investigations confirm the lattice parameter expansion after the inclusion of Zn and carbon. EDX and EELS analysis explicitly reveal the existence of carbon in Ni 3 ZnC 0.7 . Combined with in situ XANES experiments and DFT calculations, the introduced carbon atoms coordinate with the neighboring Ni atoms and impair a positive charge to Ni. As a result, the interstitial occupation in Ni 3 ZnC 0.7 suppresses the reaction pathway of ethylene hydrogenation and increases the selectivity drastically under hydrogen-rich/poor conditions. That is to say, it can effectively bypass the temporal metastable state variation under a fluctuation reaction condition. More importantly, the formed Ni 3 ZnC 0.7 exhibits a highly improved stability compared with Ni. As a result, the dissociation and accumulation of carbonaceous species are inhibited on the surface. Hence, our work provides a proof-of-concept to control of interstitial sites in heterogeneous metal catalyst. Moreover, such a new solution enables improving the selectivity and stability for selective acetylene hydrogenation.

Methods
Catalysts preparation. The multiwall CNTs was purchased from Shandong Dazhan Nano Materials Co., Ltd, China. The CNTs were firstly treated with HCl and then oxidized with concentrated HNO 3 at 140°C for 2 h (oCNT). The Ni 3 Zn/ oCNT catalyst was synthesized with a typical impregnation method. For the 5.0 wt % Ni 3 Zn/oCNT, 101 mg Ni(NO 3 ) 2 and 33.7 mg Zn(NO 3 ) 2 were dissolved in 40 mL ethanol, then 374 mg oCNT was added into the solution and the suspension was ultrasonic dispersed for 1 h to obtain a homogeneous distribution. The ethanol was evaporated at 50°C using a rotary evaporator. After calcination at 100°C for 2 h in static air, the reduction process was conducted at elevated temperature with a heating rate of 10°C min −1 from 30 to 500°C for 2 h in 50.0 vol% H 2 in Ar with the flow rate of 100 mL min −1 . The Ni 3 Zn/oCNT catalyst was obtained after cooling to room temperature. The Ni 3 ZnC 0.7 /oCNT catalyst prepared by switching the gas atmosphere to 1.0 vol% C 2 H 2 /He at 200°C for 1 h after the reduction of Ni 3 Zn/oCNT. The 5.0 wt% Ni/oCNT, 2.0 wt% Pd/oCNT and PdAg/oCNT catalysts were also synthesized with the same impregnation method and reduced at 500°C.
XRD investigations. The in situ XRD measurements were conducted on a PANalytical X'pert PRO diffractometer with XRK900 in situ chamber using CuKα radiation, operated at 40 kV and 30 mA. The 10.0 wt% loading of Ni and Zn on oCNT was chosen for a clear peak evolution considering that 5.0 wt% loading sample would cause low intensity of the diffraction peaks and lead to ambiguity in peak recognition in the in situ XRD experiments. The 10.0 wt% Ni 3 Zn/oCNT sample was transferred to an in situ chamber with the beryllium window. Then the H 2 /He flow at 10 mL min −1 was introduced after the He flushing, the temperature elevated with a rate of 10°C min −1 from room temperature to 500°C while the flow is steady. The XRD patterns were recorded at desired temperatures from 10°t o 90°. After the reduction, the chamber was cooled down to room temperature and flushed with helium. Then the acetylene was imported and the evolution of the Ni 3 Zn structure to Ni 3 ZnC structure was examined at elevated temperature to 200°C under 1.0 vol% C 2 H 2 /He flow (10 mL min −1 ). The ex situ XRD patterns were performed on a Rigaku D/max 2400 diffractometer (CuKα radiation, λ = 0.15418 nm).
X-ray absorption spectroscopy investigation. Theoretical calculation. The DFT calculations were performed with Generalized Gradient Approximation using the Perdew-Burke-Ernzerhof (PBE) exchangecorrelation function. By using WIEN2k code, the charge density difference was calculated with a plane-wave cutoff parameter of RK max = 7 and 1000 points in the whole Brillouin zone. The bader charge analysis and geometric relaxation of Ni 3 Zn was implemented using the Vienna ab-initio simulation package (VASP). The kinetic energy cutoff of the plane-wave was set to be 400 eV, and the Brillouin zone was sampled by Monkhorst-Pack meshes of 9 × 9 × 9 grid. Atomic position and lattice parameters of Ni 3 Zn were geometrically relaxed until changes in the total energy and force were less than 10 −4 eV and 5 × 10 −2 eV Å −1 , respectively.
Catalytic tests. The acetylene selective hydrogenation in the excess of ethylene (1.5, 3.0, 4.5, 6.0, and 7.5 vol% H 2 , 20.0 vol% C 2 H 4 , 0.5 vol% C 2 H 2 , helium as balance) was performed in a fixed-bed quartz micro-reactor under atmospheric pressure. The Ni 3 Zn/oCNT was in situ reduced and transformed to Ni 3 ZnC 0.7 /oCNT at 200°C under reaction conditions. For the stability test, the Ni 3 ZnC 0.7 /oCNT sample was in situ generated after reduction and carburization processes following 20 mL min −1 50 vol% H 2 /He treatment at 500°C for 2 h and 200°C 1.0 vol% C 2 H 2 /He for 1 h before the reaction. Ni/oCNT catalyst was also in situ reduced at 500°C for 2 h under 50 vol% H 2 /He atmosphere before the test. The conversion (Conv) and selectively (Sele) were calculated as follows: Conv C 2 H 2 ¼ C C 2 H 2 ;in À C C 2 H 2 ;out C C 2 H 2 ;in 100%; Sele C 2 H 4 ¼ 1 À C C 2 H 6 ;out C C 2 H 2 ;in À C C 2 H 2 ;out ! 100%; where C in represents the acetylene concentration in the feed gas and C out represents the different products in the outlet gas. Because of high concentration of the ethylene and the minor changes caused by the hydrogenation of acetylene was not easy to be detected, the consumption of the acetylene was assumed to be either hydrogenated to ethylene or hydrogenated to ethane.

Data availability
More experimental details and additional data can be found in the Supplementary  Information (Supplementary Figs. 1-19 and Notes 1-4). All the relevant data are available from the corresponding author upon reasonable request. Source data are provided with this paper.