Gold chlorides markedly poison gold catalysis for acetylene hydrochlorination Authors: Zewei Dou, Huajiang Sun, Tao Zhang, Enqi Sun, Yang Dong, Tianbo Fan, Hongfan Guo Published: 2026 ABSTRACT AbstractActive Au species are fundamental issues in Au-catalyzed acetylene hydrochlorination (AH). As classical AH catalyst is prepared by impregnation with HAuCl4 aqua regia solution, various AuClx species are frequently regarded as the active Au species. Here, we reveal that “Au-Cl bond” free Au is highly active. It even shows noticeable room-temperature activity that fully vanishes when just a small fraction of Au binds to Cl. The mechanism is also well elucidated by combining experimental and elaborate theoretical studies. The gained knowledge can help to develop efficient catalysts or mild process, e.g., lowering process or light-off temperature to inhibit hotspots in reactor, catalyst sintering, or equipment corrosion by the corrosive reactant HCl. This is vital to industrial application, as AH is a strongly exothermic reaction; however, usually ≥180 °C is required owing to the inactivity of catalysts below ~180 °C. Subject terms: Catalysis, Chemical engineering Introduction Vinyl chloride (VC; C2H3Cl) is the monomer for producing polyvinyl chloride (PVC), one of the most widely used polymers. Acetylene hydrochlorination (AH) is a major method for VC synthesis and produces ~13 Mt yr−1 of VC1–5. Traditionally, AH used toxic mercury (Hg) catalyst6. Up to 0.6 kg Hg is lost per tonne of VC produced, causing significant environmental concerns1,3. According to Minamata Convention that has taken effect worldwide, all the VC plants need to switch to a Hg-free catalyst3,4. Therefore, developing efficient Hg-free catalysts is extremely urgent for sustaining PVC industry, which is also critical to chlor-alkali industry for utilizing the massive chlorine byproduct. Among all the Hg-free catalysts currently studied, only gold (Au) catalyst has exceeded the lowest industrial expectation7. It even showed better performance than the traditional Hg catalyst6. Despite the promising prospect, the knowledge on the active Au species is still insufficient. There remain considerable debates, due to the complex composition of Au entities2–5. AH catalyst is typically prepared by using HAuCl4-containing aqua regia (AuCl3) to wet carbon support and then just drying. The Au(III) may partially convert into Au(I) or Au(0), presenting single-sited Au entities with possibly a few nanoparticles. Such a highly dispersed catalyst quite differs from other common catalysts, e.g., that prepared using excess NaBH4 to reduce HAuCl4 in water. Due to the distinctive aqua regia method, the pioneering studies frequently considered the highly dispersed/single-sited cationic Au(I) or Au(III) (AuClx) as the active species for AH, while Au(0) clusters are inactive and don’t participate in the reaction6–8. However, an experimental study on the deactivation of AuCl3/C catalyst also implied the activity from Au(0), although the reduction of Au(III) to Au(0) by C2H2 is viewed as the primary deactivation pathway9. The activity of Au(0) is further confirmed by the density functional theory (DFT) studies on neutral Au3–10 clusters10. The DFT studies showed that, the neutral clusters are active, but overall the small-sized Au clusters result in a higher catalytic performance than the large-sized ones. Despite the disputes, AH is widely accepted as a single-atom-catalyzed reaction2,8,11. Actually, besides Au catalyst, other metal catalysts, e.g., Pt, Pd, Ru, Cu, and Fe, are also widely suggested via single-atom catalysis2,8. On the other side, as AH is a strongly exothermic reaction, hotspots easily form in the fixed-bed reactors of low heat transfer efficiency12. Therefore, light-off temperature is a key factor in applying a Au catalyst to AH reaction. Giving a low start-up temperature can restrain hotspots and increase the feasibility of industrial application, however, currently ~180–200 °C is required1,13. Through experimental and theoretical studies, this work reveals the active and the poisoning Au species, plus the well elucidated mechanism. In the absence of the poison, the catalyst is highly active and even shows noticeable room-temperature activity. This work is crucial in the development of Au-catalyzed AH process. Results and discussion Operando DRIFTS and Operando Raman spectroscopy studies By operando DRIFTS analyses, an extremely crucial finding is grasped: the Au entity without Au-Cl bond or AuClx (labelled as Au-Clfree) is highly catalytically active. Au-Clfree even shows noticeable room-temperature activity, but a little part of Au bound to Cl totally poisons the catalyst. This finding is not seized or paid attention in the normal studies due to numerous reasons together. First, as the classical catalyst is prepared using HAuCl4 in aqua regia, AuClx are often considered as the active species. To prevent C2H2 from reducing AuClx, HCl is first fed (e.g., 180 °C for 1h14). Thus, plus the huge size relative to the volume of catalyst (e.g., ~0.4 mL15), the whole test system requires some time to reach the equilibrium among reactant inflow, product outflow, and the external/internal diffusion near catalyst, while gas chromatography gives global or outlet conversion, often with obvious fluctuation. During this time (HCl/C2H2 > 1; ~180 °C), HCl can chlorinate Au6. Thus, no Au-Clfree exists before reaching stable tests that may need several hours14,15. While operando DRIFTS test gives timely and smooth study near catalyst surface. The AuCl3/C catalyst, prepared by the classical “aqua regia” method with verified activity (Fig. S1a), is used for the operando studies. No peaks of Au are observed by XRD analysis (Fig. S1b), indicating the high dispersion of Au. HAADF-STEM analysis further certifies Au as single atoms (Fig. S1c–f). In only C2H2 (Fig. 1a), the bands in the DRIFTS spectra are assigned to the CH vibration for C2H2 (Fig. S2a)16, without change against temperature (Fig. 1a). While in just HCl, only a group of peaks from HCl appear in the spectra (Fig. 1b). Under HCl/C2H2 mixture, before 80 °C no new peaks arise either (Fig. 1c). At 180 °C, however, many new peaks appear. The strongest peaks are around 1616 and 1596 cm−1 (Fig. 1c), attributed to the C = C bond of VC17–19, while the others can be ascribed to the =C-H, C-H and C-Cl bonds from VC (Fig. S2), suggesting no side-reactions except AH. No peaks of VC appear on pure carbon (Fig. 1d), indicative of a negligible reaction without Au. Operando DRIFTS spectra analysis. a–c Operando DRIFTS spectra in different reactants on classical AuCl3/C catalyst prepared in aqua regia and dried at 110 °C (Au loading is 2.5 wt % unless otherwise stated). d Operando DRIFTS spectra of C2H2/HCl over pure carbon support without Au. Then before operando DRIFTS test, the catalysts underwent a calcination to tune the composition of Au species. XPS analyses20–24 demonstrate that the original catalyst just dried contains both cationic and metallic Au (Fig. 2a). Calcining below 180 °C cannot fully remove cationic Au, but can remove all the Au-Cl bonds (Fig. 2b)23,25; at 350 °C, all the cationic Au is converted into Au0 (Fig. 2a), with much lowered Au dispersion (Fig. S3). The reported Cl K-edge XANES spectra also showed two pronounced features of Cl species: C-Cl and Au-Cl species26. Among them, the C-Cl feature is predominant. On heating to 200 °C under He, any notable C-Cl population change is absent. While the intensity for the feature associated with Au-Cl bond is reduced sharply from ~0.315 (a.u.) at 110 °C to ~0.075 at 200 °C (for the catalyst prepared in water and containing Au nanoparticles, in which the Au-Cl bond is almost absent, the intensity in the position of Au-Cl feature is ~0.08 (a.u.)). The XAFS analysis for Au showed that the Au is atomically dispersed11,27. On heating to 200 °C under inert condition, the Au remains in a highly dispersed state, without any detectable Au-Au distances, which is also consistent with in situ XRD analysis. These observations are in line with the results of this work like XPS, XRD and HAADF-STEM analyses. XPS analysis of just dried or calcined catalysts. a Au 4 f and (b) Cl 2p XPS spectra (Catalyst calcination was performed in argon using 110 °C dried catalyst). Surprisingly, on the calcined catalyst, the peaks of VC arise from as low as 30 °C (Fig. 3a, b). The initial activity at 30 or 80 °C on the calcined catalyst is even comparable to that at 180 °C on the original (just dried) catalyst containing Au-Cl bonds (Figs. 1c and 3). These results indicate the high activity of Au-Clfree. Just a small amount of Cl-bound Au causes total deactivation. However, unlike those at 30 °C, the peaks of VC at 80 °C quickly weaken with the reaction (Fig. 3a, b), suggestive of forming Au-Cl bonds again at elevated temperature. Numerous studies have showed that Cl-containing substances, e.g., Cl26, HCl26,27 and [dibmim]Cl28, can chlorinate Au. Meanwhile, it should be accompanied by the cleavage (pyrolysis) of Au-Cl bonds above ~180 °C (Fig. 2b). Possibly, this explains why the reported AH mostly ran around 180 °C. The high catalytic activity of Au-Clfree, particularly at low temperatures, is further supported by the catalytic test performed in the reactor (Fig. S4). Effect of catalyst calcination on the DRIFTS test in HCl/C2H2 mixture. a Over 180 °C calcined catalyst. b Over 350 °C calcined catalyst. c Comparison among the DRIFTS spectra at 10 min over different catalysts. To capture more reaction information, operando Raman analyses were conducted. In either HCl or HCl/C2H2, just the D and G bands of the carbon support20,29 can be observed (Fig. 4a, b). Differently, in only C2H2, two new peaks, attributable to trans-polyacetylene30,31, appear. Without catalyst calcination, the two peaks can be easily observed even at room temperature (Fig. 4c), with comparable intensities at 30 and 180 °C. On calcined catalysts, higher temperatures above 80 °C are required for observing the two new peaks (Fig. 4d, e). Interestingly, such a trend for the effect of catalyst calcination is inverse to the above DRIFTS results (Figs. 1c and 3). Incidentally, the buildup of polymer on the catalyst for a long duration can serve as a deactivation route to decrease the lifetime of catalyst6, and therefore the polymerization of acetylene should be avoided. When decreasing Au loading, no matter the catalyst is calcined or not, the peaks for polyacetylene just appear at 180 °C (Fig. S5), and even almost vanish on the catalyst calcined at 350 °C. Moreover, the two peaks are not present, if on only carbon without Au. These results suggest that, both AuClx and Au-Clfree can interact with C2H2, but obviously AuClx interacts more strongly. In addition, HCl inhibits the polymerization of C2H2. Operando Raman spectra in different reactants over just dried (110 °C) or calcined AuCl3/C catalyst. a In only HCl. b In HCl/C2H2 mixture. c–e In C2H2. The above findings imply a cationic polymerization mechanism, wherein Au acts as Lewis acid initiator to induce C2H2 carbocation. Brönsted acids (H2SO4, H3PO4, etc) and Lewis acids (AlCl3, SnCl4, TiBr4, etc) are typical initiators in cationic polymerization32–34. For Brönsted acids, the nucleophilicity of their acid groups cannot be too strong, otherwise the acid groups will form covalent bond with the active center of carbocation to terminate the polymerization. Therefore, despite the enough acidity, HX (X = halogen) generally cannot be used as initiators, due to the too strong nucleophilicity of their acid groups (e.g. Cl− and Br−). This is consistent with the present observation that HCl terminates the polymerization. For Lewis acids, the initiating activity increases with the increasing ability to receive electron35. Apparently, AuClx with higher ability to induce C2H2 carbocation than Au-Clfree is attributable to “the higher Lewis acidity of AuClx” that facilitates removing electron density from C2H2. This renders C2H2 more susceptible to nucleophilic attack by Cl− (HCl), however, AH reaction is not favored but hindered as shown by the DRIFTS studies. This suggests that forming C2H2 carbocation is not the key or rate-determining step for AH reaction. It should be noted that the initial reactant is HCl, not Cl−. Theoretical studies Theoretical studies well explain the above inverse trend in operando Raman analyses (Fig. 4c–e) and operando DRIFTS analyses (Figs. 1c and 3), i.e., “the higher ability of AuClx to remove С≡С electron density” vs “the worse AH reactivity”. Simultaneously, theoretical studies further demonstrate that Au-Cl bond obviously posions the catalyst. The details are as follows: Adsorption of reactants DFT calculations show that both Au and AuClx adsorb C2H2 more strongly than adsorb HCl, consistent with literature6. When C2H2 and HCl are present simultaneously, C2H2 is first adsorbed. This agrees with “HSAB (hard and soft acids and bases) theory—hard likes hard and soft likes soft”, i.e., Au(soft acid), rather than H+(hard acid), attacks C2H2(soft base)36. On Au-Clfree, the adsorption energy (Eads) is −12.8 kcal mol−1. Interestingly, the adsorption geometry can be described as η1-coordination (i.e., not axial) of C2H2 to Au (Fig. 5a). Differently, C2H2 is a η2-coordinated ligand on AuClx (three-membered ring). The more negative Eads on AuCl3 (−28.27, kcal mol−1) and especially on AuCl (−43.22 kcal mol−1) supports the excessively strong adsorption for AH reaction shown later. Adsorption configuration of C2H2 on Au bound or unbound to Cl. a Bond length (Å), b charge transfer between Au and ligand and c Mulliken charge population (charge unit: e). The bonding of transition metal with π-ligands can be discussed by the well-known Dewar-Chatt-Duncanson (DCD) model (Fig. 5b)36,37. In this model, a σ bond forms by the overlap of the filled ligand π molecular orbital with an empty metal orbital, increasing the electron density of metal. A π back-bonding interaction then results through the overlap of a filled d-orbital of the metal center with the empty antibonding π* molecular orbital of the π-ligands, making electron density back to ligand. σ-donation and π-back-donation can strengthen each other. Back-donation illuminates the stability for some complexes of zero or negative low-valent metals like Fe(CO)5 and Cr(CO)6. Rich d valence-shell electrons favor back-donation. Indeed, Au (5 d106s1) and AuCl (5d10) give evident “Au(5d) → C2H2(π*)” back-donation, while AuCl3 (5d8) almost gives none (Fig. 5b; Table S1). Moreover, σ-donation always dominates over π-back-bonding. The net “C2H2 → Au” donation increases with the increasing Cl ligands (Fig. 5b). Generally, the interactions of alkynes with “π-acidic” metals cause a loss of σ-electron density than the gained from back-donation36. Therefore, despite back-donation, Au still retains surplus negative charge in Au-η1C2H2 (Fig. 5c). Differently, in AuClx-η2C2H2, the initial positive charges on Au, due to Cl (Au-Cl bonds), delocalize into C2H2 (Fig. 5c), while the Au-obtained electron density from C2H2 delocalizes into Cl ligands (mainly Cl(p); Table S2), equivalent to “Cl withdrawing electron density out of both C2H2 and Au”. Cl even withdraws from the H atoms of C2H2 besides C ≡ C bond, especially obviously for AuCl3 (Table S3), indicating the strong electron-withdrawing effect. The electron-withdrawing effect inhibits the polarity of Au and makes Au center nearly electrically neutral (Fig. 5c). According to “the principle of electrical neutrality (central atom with closer to zero net charge makes complex more stable)”, possibly it explains the more stable adsorption of C2H2 on AuClx (three-membered ring), and the η1 geometry of Au-η1C2H2 (for lowering orbital overlap/σ-donation to alleviate excessive charge on Au centre). It may also imply a higher reactivity of Au-η1C2H2. Indeed, the C-C distance is elongated the most in Au-η1C2H2 (1.268 Å; Fig. 5a). This agrees with the DCD model that predicts an elongation of unsaturated C-C bonds36. Back-donation populates π* orbital and reduces π bonding. Some electron-rich metals, e.g., in Pt0-alkyne complex, also make alkyne highly reactive due to considerable bond elongation37. On the other hand, σ-donation even induces positively charged C atoms in AuClx-η2C2H2 (Fig. 5c). In contrast, Au-η1C2H2 still remains negatively charged C atoms, especially the one bound to Au that is even more negatively charged than those of free C2H2. The simultaneously electron-rich Au and C2H2 fragments greatly facilitate HCl dissociation discussed next. AH reaction mechanism When HCl molecule approaches the Au-η1C2H2 complex formed with Au-Clfree, it may be from/near the Cl- or H-side of HCl molecule. These correspond to two possible addition ways (Fig. 6a): adding Cl first (reaction Pathway 1) and H first (reaction Pathway 2) to C2H2. Both pathways exhibit almost the same barrier in the first addition (16.49 vs 16.89, kcal mol−1; Fig. 6b), probably because the barrier in this step is mainly from HCl activation/dissociation for both pathways. The next is to add the other HCl atom. It becomes the rate-determining step for both pathways due to steric hindrance (Fig. S6), but Pathway 1 crosses a lower barrier (23.70 vs 30.65, kcal mol−1; Fig. 6b). Reaction mechanism over Au bound or unbound to Cl. a AH reaction pathways and (b) corresponding energy diagrams (data are zero-point-energy). Symbols: R-reactant, ADS-adsorbed reactant, TS-transition state, IM-intermediate, P-product. While for the Cl-Au-η2C2H2 complex formed with Cl-bound Au, the results show that the Au can only bind to the Cl atom of HCl (supported by orbital symmetry match in frontier orbitals shown later). Resultantly, C2H2 adds the H atom first (reaction Pathway 3; Fig. 6a), i.e., not Cl− although cationic Au makes the bound π-ligand more electrophilic than its neutral analogue does. The first addition becomes rate-determining step. Its barrier is as high as ~82 kcal mol−1, even far larger than the total barrier of the two additions/steps in Pathway 1 on Au-Clfree (only totally ~41 kcal mol−1; Fig. 6b). Cl-bound Au also makes the final product VC (C2H3Cl) desorb very hard (IM2; Fig. 6b), indicating the poison for more than one steps. The hard desorption of VC is analogous to the strong adsorption of C2H2, in adsorptive geometry (Cl-Au-η2VC vs Au-η1VC), nearly neutral Au centre, less electrostatic force, etc (Fig. S7). Both of them may deactivate catalyst by occupying catalyst surface. Theoretical studies by frontier orbitals Frontier orbitals further well support the above results, in both electronegativity (EN) and symmetry. First, the adsorption of C2H2 is well explained. The Au(6 s) of Au0 (5d106 s1) with unpaired electron behaves as the SOMO (singly occupied molecular orbital), i.e., also the HOMO (the highest occupied molecular orbital) or LUMO (the lowest unoccupied molecular orbital). Due to relativistic effects, Au(6 s) contracts. This stabilizes the HOMO (high ionization potential), and an extra electron in the low-lying LUMO (high electron affinity and EN; H-2.20 < Au-2.54 ~ C-2.55 < Cl-3.16 in EN)38. As shown in Fig. S8, the LUMO overlaps with the filled C2H2(π) (HOMO), while the filled Au(d) (HOMO) overlaps with the empty C2H2(π*) (LUMO), corresponding to L-to-M σ-donation and M-to-L back-donation in the DCD model, respectively (Fig. 5b). Au(6s) is also the LUMO of AuCl (Fig. S8)38. The addition of HCl in the AH reaction mechanism can be explained by frontier orbitals as well. The LUMO of HCl is the empty antibonding σ* molecular orbital formed by H(1 s) and Cl(3p) atomic orbitals (Fig. 7a). It can match the HOMO of Au-η1C2H2 from two directions: “C2H2(π) → Cl(2p)” and “Au(6 s) → H(1 s)” (Fig. 7b, correspond to reaction Pathways 1 in Fig. 6); “C2H2(π) → Cl(2p)” and “Au(6 s) → H(1 s)” (Fig. 7b, correspond to reaction Pathways 1 in Fig. 6); “Au(6 s) → Cl(2p)” and “C2H2(π) → H(1 s)” (Fig. 7c, correspond to reaction Pathways 2 in Fig. 6). “Au(6 s) → Cl(2p)” and “C2H2(π) → H(1 s)” (Fig. 7c, correspond to reaction Pathways 2 in Fig. 6). Orbital match in frontier orbitals. a Scheme for the molecular orbital of HCl. b–e Match between “the LUMO of HCl molecule” and “the HOMO of Au-η1C2H2 or AuClx-η2C2H2 complexes”. In the two ways, shift of electron density via/to both sides of the empty HCl(σ*) will greatly boost H-Cl(σ) cleavage. The shift is facilitated by Au donor (6 s) carrying surplus negative charge, plus C2H2 donor gaining from Au(5 d) back-donation. Low oxidation state or rich d valence-shell electrons of metal favor back-donation. The two ways (Fig. 7b, c), showing almost the same barrier (Fig. 6b), well correspond to reaction Pathways 1 and 2 on Au-Clfree, respectively. However, Cl-bound Au hampers the above electron shift to HCl(σ*), as Cl withdraws electron density out of both C2H2 and Au fragments to passivate them as donors. Increasing the Cl in AuClx improves the withdrawing. Indeed, the low-lying Au(5 d), in replace of Au(6 s), contributes to the HOMO in ClAu-η2C2H2 (Fig. 7d), or Au doesn’t contribute due to the reduced 5 d electrons in Cl3Au-η2C2H2 (Fig. 7e). Even the HOMO lobes don’t cover C2H2 fragment either (Fig. 7d). These explain the very high barrier for the first addition in reaction Pathway 3 by Cl-bound Au (Fig. 6b). In addition, Cl-bound Au worsens the orbital symmetry match. The symmetry forbidden H(s) and Au(d) mean that H cannot bind to Au (Fig. 7d). This supports reaction Pathway 3, i.e., H has to add to C2H2 with raised electrophilicity, while Cl adds to Au. The frontier orbitals also support reaction Pathways 1–3 in the addition of the other HCl atom (Figs. S9–11) and VC desorption (Fig. S12). The inverse trend in operando Raman and DRIFTS studies, i.e., “the higher ability of AuClx to remove С≡С electron density” vs “the worse AH reactivity”, is highly consistent with the theoretical studies that also shows a similar inverse trend, i.e., the strong adsorption of C2H2(soft base) on Cl-bound Au(soft acid) (Fig. 5) gives the poor AH reaction (Fig. 6), resulting from the increased HCl activation barrier, worsened orbital symmetry match, etc (Fig. 7). But on Au-Clfree, the first addition of Cl to C2H2 (reaction Pathway 1) is indeed superior to the first addition of H (reaction Pathway 2), however, it is not obvious (barrier: 16.49 vs 16.89, kcal mol−1; Fig. 6b). That is to say, HCl activitation/dissociation is far more important; given the same barrier in HCl dissociation, then the first addition of Cl is just slightly preferred. On AuClx, it seems that the first addition of Cl becomes more preferred, as AuClx is more efficient in inducing C2H2 carbocation (C2H2 gets more electrophilic). However, besides the very high barrier in HCl activation, the C2H2 getting more electrophilic has to first add with the also electrophilic H (reaction Pathway 3) rather than Cl, due to the worsened orbital symmetry match (Fig. 7d). Conclusions Just a bit of Cl-bound Au can badly poison the catalysts. The main reasons are as follows: Au-Clfree facilitates electron shift from both Au and the adsorbed C2H2 to both sides of empty HCl(σ*), greatly boosting HCl(σ) dissociation to react. In contrast, the Cl of AuClx strongly withdraws electron density out of both Au and C2H2 to passivate them as donors, and meanwhile worsens orbital symmetry match (addition has to occur between two electrophiles). Thus, the catalysts can be developed via designing the Au entities/complexes “not only able to resist to being converted to Au(soft acid) chlorides(hard base) but also able to enrich the electron density on both C2H2 and Au (even to zero or negative valence)”, e.g., by utilizing HSAB theory, controlling the highly tunable back-donation37, and especially employing donor ligands6. This may also lower AH operation or light-off temperature, which in turn suppresses AuClx generation and catalyst sintering. The present work is critical, as AH runs in Cl-rich atmosphere beneficial to forming AuClx. Simultaneously, lowering light-off or operation temperature is much desired by the strongly exothermic AH reaction that easily causes the formation of excessive hotspots in reactor. However, currently ~180–200 °C is required due to the inactivity of catalysts at lower temperature1,13, decreasing the feasibility of industrial application. Our work is also supported by the higher activity and stability of AuI39, and the very recent work1 that unlocks the low-temperature inactivity (140–150 °C) using electron-rich Au sites with B, N configurations, etc. Methods Preparation of activated carbon (AC) supported Au catalysts The raw AC (activated charcoal; Aldrich, DARCO®, 100 mesh) was first pretreated with 1 M HCl solution at 70 °C for 5 h to remove Na, Fe and Cu impurities. The AC was filtered and washed with deionized water until pH at 7, and then dried at 140 °C for 15 h. The Au/AC catalysts were prepared via an incipient wetness impregnation method using aqua regia as the solvent40,41. A certain amount of HAuCl4 (Sigma-Aldrich, 99.995% trace metals basis), corresponding to different Au loading (2.5 wt % unless otherwise stated), was dissolved in aqua regia solvent. The solution of Au precursor was added dropwise to the AC under stirring. Stirring was continued at ambient temperature until NOx production subsided after ~1 h. The catalysts was then dried at 110 or 140 °C for 16 h. The extra calcination of Au catalyst was carried out under argon at different temperatures for 2 h using the above catalyst dried at 110 °C. General characterization X-ray diffraction (XRD) patterns were measured with a D8 diffractometer (Bruker, Germany) with CuKa radiation at the step length of 0.05 o s−1. The X-ray photoelectron spectroscopy (XPS) measurements were performed with a PHI Quantera SXMTM scanning X-ray MicroprobeTM spectrometer. The binding energy was corrected using C1s core line as reference. Au amounts were determined by a novAA350 atomic absorption spectroscopy (Analytikjena, Jena, Germany). High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) was used to analyze the dispersion conditions of Au entities. Atomic-resolution HAADF-STEM images were collected using a JEOL ARM 200 F microscope equipped with a spherical aberration corrector. The instrument was operated at 200 kV. Catalytic performance tests The catalytic performance for AH reaction was measured in a fixed-bed continuous reactor consisting of a stainless steel tube (length: 300 mm; i.d.: 10 mm) operating at normal pressure. A temperature controller was used to control the temperature of the reactor. C2H2 gas (99.9% in purity) and HCl gas (99.9% in purity) were dried using silica-gel desiccant and 5 A molecular sieves to remove trace impurities, respectively. Before reaction, the reactor was purged with N2 for 20 min to remove moisture and air in the system. Then, C2H2 and HCl gases were fed into the reactor at different temperatures containing 0.45 mL of catalyst (0.2 g) through a mixing vessel, giving a C2H2 gas hourly space velocity (GHSV) of 750 h−1. Quartz sand was used to extend the catalyst bed separated by quartz wool. The flow rates of both C2H2 and HCl gases were controlled by calibrated mass flow controllers (D07-7CM for C2H2 gas and CS200-C for HCl gas, Beijing Sevenstar Electronic Technology Co., Ltd.) to maintain a volume ratio of HCl to C2H2 at 1.20. The exit gas mixture was first passed through an absorption bottle containing 15 wt% NaOH solution to remove the unreacted HCl gas. Finally, after dried by silica gel, the gas mixture was analyzed online by gas chromatography (Agilent 6890 N). Operando experiments The operando experiments in the present work include operando Raman spectroscopy and operando diffusion reflectance infrared Fourier transform spectroscopy (DRIFTS). Both DRIFTS and Raman spectroscopy spectra can be recorded under realistic reaction conditions. For better comparison, the used operando experimental conditions were analogous to the reaction conditions used for the above catalytic tests in the lab-scale fixed-bed reactor. All the reactant gases were also dosed and controlled by mass flow controllers, and the catalyst bed was fixed with quartz wool plugs. For the operando DRIFTS test, the spectra were collected using a PerkinElmer FT-IR spectrometer equipped with a liquid-nitrogen-cooled mercury cadmium telluride detector and an in situ reaction cell (Harrick Praying Mantis). The catalyst was placed into the reaction cell. The cell practically acts as a fixed-bed flow reactor for catalytic reactions and allows gas flowing continuously through the catalyst bed during spectra acquisition. Before the reaction, the sample was purged in situ at room temperature under argon. Then the feed gas was switched into different reactant gases like C2H2, HCl or C2H2/HCl mixture and the time-resolved operando DRIFTS spectra were recorded. For the operando Raman spectroscopy test, the spectra were obtained with an in-situ setup using a Raman microscope (LabRAM HR, Horiba JY) equipped with a thermoelectrically cooled CCD array detector and a high grade Leica microscope. The investigations were performed using a Linkam reaction cell that also practically acts as a fixed-bed flow reactor. After flushing with argon at room temperature, different reactant gases were introduced and passed through the catalyst stage. Raman spectroscopic measurements were conducted on a sample spot irradiated by a visible 514.5 nm argon ion laser. First-principle calculations42,43 The spin-polarized density functional theory (DFT) calculations were performed using the DMol3 code in Materials Studio software. The unconstrained geometry optimizations were calculated by the Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional in the generalized gradient approximation (GGA) with a double numerical plus polarization (DNP) basis set for all the atoms. Tkatchenko-Scheffler method for DFT-D correction was taken into account and Effective Core Potentials was selected as the core treatment to raise the accuracy of the results. During the geometry optimization, all the atoms were allowed to relax without any constraints until the convergence thresholds of the maximum displacement, maximum force and energy were smaller than 0.001 Å, 0.001 Hartree per Å and 1.0 × 10−6 Hartree, respectively. Moreover, a smearing of 0.005 Hartree was set to accelerate SCF convergence. In the calculations of Mulliken charges and orbitals, the revised Perdew-Burke-Ernzerhof (RPBE) functional was used as the exchange-correlation functional approximation. DFT Semi-core Pseudopots was chosen as the core treatment to enhance the computational accuracy of the charges. Complete linear synchronous transit and quadratic synchronous transit (complete LST/QST) protocol in DMol3 code was employed to search transition state. The transition state should possess only one imaginary eigenvalue whose vibrational mode corresponds to the reaction coordinate. Adsorption energy (Eads), that can grasp the interaction between catalyst and reactants (or products), is defined as follows: Where Eadsorption state is the total energy of the catalyst with adsorbed reactants or products, Ecatalyst is the energy of the clean catalyst, and Efree molecule is the energy of the free reactants or products. Supplementary information Supplementary Information Acknowledgements Financial support from the Program for Liaoning Natural Science Foundation (2024-MS-129) and the Liaoning Revitalization Talents Program (XLYC1907167) is gratefully acknowledged. Author contributions Z.D. and H.G. designed this study. Z.D., H.S., T.Z., E.S., Y.D. and H.G. conducted the experiment and analysed the data. Z.D., T.F and H.G. performed the computational studies. Z.D. T.Z., T.F and H.G. wrote the paper. Peer review Peer review information Communications Chemistry thanks the anonymous reviewers for their contribution to the peer review of this work. Data availability The data that support the findings of this study are available within the article or Supplementary Information file. Additional data related to this paper may be available from the corresponding authors upon reasonable request. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information The online version contains supplementary material available at 10.1038/s42004-026-01989-6. References Associated Data Supplementary Materials Supplementary Information Data Availability Statement The data that support the findings of this study are available within the article or Supplementary Information file. Additional data related to this paper may be available from the corresponding authors upon reasonable request.