Co-infiltration and dynamic formation of Pd₃ZnC_x intermetallic carbide by syngas boosting selective hydrogenation of acetylene

Abstract

Transition metal carbide shows excellent performance in selective hydrogenation of acetylene, however, the carburization of Pd-based intermetallic compounds remains infeasible. Here we report the successful synthesis of an unprecedented Pd₃ZnC_x intermetallic carbide, via co-infiltration of zinc and carbon in one-step carburization by syngas. Utilizing state-of-the-art in situ characterizations and theoretical calculation, we unveil the dynamic evolution of Pd₃ZnC_x during carburization, forming a Pd₃Zn like cubic phase carbide structure. A unique transitional state (Pd_t) with low content of Zn/C co-infiltration is clearly identified facilitating phase transition and sustain incorporation of carbon and zinc at elevated temperatures. The Pd₃ZnC_x carbide shows by far the best catalytic performance in the selective hydrogenation of acetylene with a high selectivity (>90%) even at a high H₂/C₂H₂ ratio. Our results therefore provide a co-infiltration strategy and dynamic insights for the one-step synthesis of Pd based intermetallic carbides, towards high-performance intermetallic compound for selective hydrogenation of acetylene.

Introduction

Selective hydrogenation of acetylene is an important process widely used in ethylene industry for removing acetylene impurities^1,2. Pd is the most active component for this reaction yet suffers low selectivity at high acetylene conversions and poor stability for long-term operations¹. To address these challenges, various Pd-based alloys and intermetallic compounds (Pd–Ag, Pd–Au, Pd–Cu, Pd–Ga, Pd–Zn)^3,4,5,6,7 have been developed. The isolation of Pd sites by a second relatively inert metal substantially improves the ethylene selectivity and suppresses the side reactions of oligomerization/coking as the C–C coupling requires at least four contiguous Pd sites⁸. For example, the spatial arrangement of Pd atoms in PdZn alloy allows moderately strong σ-bonding of acetylene, which bridges over the two Pd atoms in the Pd–Zn–Pd linkage⁷, and weak π-bonding with ethylene that facilitates high selectivity in acetylene semi-hydrogenation.

Various surface/subsurface carbon species can be generated in situ from alkynes/alkenes and have been reported to play an important role during selective hydrogenation. The coke formation is one major reason for deactivation^2,8,9, while the subsurface or interstitial carbon can greatly suppress the deep hydrogenation which improves the selectivity¹⁰. Recently, transition metal carbides (Pd¹⁰ or Ni^11,12) have attracted tremendous interests due to their excellent performance in selective hydrogenation of acetylene. The interstitial carbon can significantly suppress the formation of hydride species (PdH_x)^10,12 and strongly modulate the geometric and electronic structure of the active metal atom. The modified electronic state of the central atom by interstitial carbon¹¹ thus weakens the adsorption of acetylene on the surface and alleviates catalyst deactivation by suppressing polymerization products. To date, however, the carburization of Pd-based intermetallic compounds by alkynes/alkenes remains infeasible likely due to their excellent structural stability and fully occupied interstitial sites, which is thus highly desirable but of grand challenges.

Carbon monoxide as a process modifier can usually exist or be intentionally added in trace amounts in acetylene hydrogenation reaction gas in the industrial operating condition for front-end or tail-end process^1,9,13,14. Carbon monoxide can improve the selectivity of ethylene by poisoning the active center and weakening the adsorption of ethylene. Meanwhile, CO has been validated to enable a variety of structural changes on binary alloys, e.g. segregation¹⁵, phase separation^16,17, carburization^18,19,20,21, under a reaction environment. However, the role of carbon monoxide as a structural assistant in the dynamic carburization of Pd-based intermetallic catalysts remains elusive.

Herein, we developed a co-infiltration strategy via one-step carburization by syngas (CO + H₂) and reported an unprecedented Pd₃ZnC_x intermetallic carbide in Pd-based intermetallic compound family, as high-performance catalyst for selective hydrogenation of acetylene. Utilizing state-of-the-art in situ characterizations and theoretical calculation, we revealed a significant role of CO in dynamic formation of Pd₃ZnC_x intermetallic carbides. A unique transitional state (Pd_t) with multi-domain structures was identified that evolved continuously from 120 to 350 °C as a critical intermediate facilitating the sustained insertion of Zn and C at elevated temperatures. The catalytic testing and theoretical calculations further revealed that Pd₃ZnC_x was the highly active phase for the selective hydrogenation of acetylene, significantly higher than the highest reported value of PdZn intermetallic compound at high selectivity (>90%) by lowering the initial hydrogenation barrier and promoting ethylene desorption.

Results

Preparation and identification of Pd₃ZnC_x intermetallic carbide

The carburization of PdZn intermetallic compounds (β₁-PdZn) by alkynes/alkenes remains infeasible likely due to its excellent structural stability and weak C–H bond dissociation on Pd–Zn–Pd fragments⁷. Instead, we proposed a co-infiltration strategy via one-step carburization using syngas (10%CO–50%H₂–40%He) to achieve the Pd₃ZnC_x intermetallic carbides (see preparation procedure in Fig. 1a). It is worth noting that the stepwise heating procedure (up to 400 °C) is very critical since the transition of palladium and other interstitial species, i.e. H, C and Zn has a relatively narrow temperature window (see detailed temperature ramps in Supplementary Discussion).

Fig. 1: Preparation and identification of Pd₃ZnC_x.

a Preparation procedure of Pd₃ZnC_x by syngas. b STEM images of Pd₃ZnC_x and the corresponding particle size distribution. c Aberration-corrected HAADF-STEM image of Pd₃ZnC_x; the corresponding EDS line scan (d), and the EDS elemental mapping of Pd L-edge (green) (e), Zn K-edge (purple) (f), C K-edge (yellow) (g). Quasi in situ XPS spectra of Pd 3d (h) and C 1s (i) core levels for PdO/ZnO/Al₂O₃ fresh sample, PdZn/ZnO/Al₂O₃ and Pd₃ZnC_x/ZnO/Al₂O₃. j Enlarged XRD patterns of Pd₃ZnC_x/ZnO/Al₂O₃, Pd₃Zn/Al₂O₃ (α-PdZn), PdZn/ZnO/Al₂O₃ (β₁-PdZn), PdC_x/Al₂O₃ and Pd/Al₂O₃ for comparison. k Linear correlation between the C:Pd ratio and shift of (111) diffraction peak relative to that of Pd₃Zn, predicted by DFT calculation (Diffraction peak shift = −5.88 × C − 0.01; R² = 0.999).

The aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) reveals the atomic structure of Pd₃ZnC_x nanoparticles (NPs) after syngas treatment (Fig. 1b–g). As shown in Fig. 1b, the as-formed NPs are uniformly distributed on the Al₂O₃ support with an average particle size of 16.1 ± 1.4 nm (Fig. 1b inset). The energy dispersive spectroscopy (EDS) mapping (Fig. 1e–g) reveals a uniform distribution of Pd (L-edge), Zn (K-edge) and C (K-edge) within the selected NP. The EDS line scan (Fig. 1d and Fig. S3c) further confirms the presence of both zinc and carbon on the selected NP. The average atomic ratio of Pd:Zn is ~3:1 based on the quantitative analysis results (Fig. S3b, d), while the carbon content from EDS can be less accurate due to the interference with surface carbon deposit. Therefore, we denoted the new intermetallic carbide phase as Pd₃ZnC_x, and the carbon content can be determined by the XRD peak shift as discussed in the latter section. The lattice spacing of the Pd₃ZnC_x NPs is 0.222 nm (Fig. 1c), which is slightly larger than 0.220 nm of (111) plane in cubic phase Pd₃Zn alloy (α-PdZn), and 0.219 nm of (101) plane in tetragonal β₁-PdZn alloy (see detailed analysis in Fig. S1). The lattice expansion is observed quite often in transition metal carbides as the carbon atoms are filled into the interstitial position in the alloy NPs^11,22. The EEL spectrum in Fig. S5 also reveals a similar feature (280–300 eV) to previously reported Ni₃ZnC_0.7¹¹, suggesting its carbide nature rather than amorphous carbon deposit.

Quasi in situ XPS experiments (Fig. 1h, i) further reveal the formation of Pd₃ZnC_x intermetallic carbides. The PdO/ZnO/Al₂O₃ fresh sample (denoted as PdO–ZnO) was pretreated at 400 °C in hydrogen (denoted as PdZn) and in syngas (denoted as Pd₃ZnC_x) respectively and then directly transferred for XPS analysis without exposure to the air. The Pd 3d XPS spectra (Fig. 1h) show a prominent reduction of PdO (fresh sample) after H₂ reduction at 400 °C and the PdZn intermetallic alloy is evidently identified with a typical Pd 3d_5/2 binding energy of 335.9 eV, in agreement with the reported value of multilayer PdZn alloy²³, along with emerging metallic Zn signal in Zn LMM Auger spectra (Fig. S6). By contrast, the Pd₃ZnC_x sample obtained by syngas pretreatment shows a Pd 3d_5/2 binding energy of 335.4 eV, slightly lower than that for PdZn phase, but higher than that of Pd metal, suggesting its alloy composition. The carbide phase is unambiguously confirmed by C 1s spectra (Fig. 1i), with a characteristic carbide component at 283.6 eV²⁴.

The phase structure of Pd₃ZnC_x was further validated by XRD, as shown in Fig. 1j and Supplementary Fig. S8. The as-formed Pd₃ZnC_x intermetallic carbide shows a cubic structure with (111) and (200) diffraction peaks similar to that of Pd₃Zn (α-PdZn), suggesting a carburization on cubic Pd₃Zn (α-PdZn) rather than tetragonal β₁-PdZn. The slight shift to lower diffraction angle is thus attributed to the carbon infiltration into Pd₃Zn alloy, similar to the XRD shift between Pd and PdC_x. The content of carbon in Pd₃ZnC_x can thus be quantified by the shift of Pd₃Zn (111) diffraction peak in XRD, similar to the previously reported method for PdC_x^24,25. Accordingly, the average carbon content in the bulk carbide can be estimated to be Pd₃ZnC_0.03. Based on DFT calculations, there is a clear linear correlation between the C:Pd ratio and the shift of (111) diffraction peak relative to that of Pd₃Zn (Fig. 1k and Table S1). Specifically, when the C:Pd ratio is 0.01, the calculation predicted shift in the diffraction peak of (111) between Pd₃ZnC_0.03 and Pd₃Zn is 0.07°, which is very close to the 0.05° of the shift observed experimentally between Pd₃ZnC_x and Pd₃Zn (Fig. 1j). According to the XRD results, Pd₃Zn shift to higher diffraction angle (lattice contraction) than Pd metal due to Zn replacement, while Pd₃ZnC_x shifts to lower diffraction angle than Pd₃Zn which is attributed to the lattice expansion due to interstitial carbon insertion. The overall effect thus results in a similar Pd–Pd distance of Pd₃ZnC_x to Pd metal. This is in line with the EXAFS fitting results (Figs. S11–14) that Pd₃ZnC_x shows the Pd–Pd(Zn) distance of 2.73 Å which is closer to Pd metal and PdC_x, rather than β-PdZn (2.85 Å)⁷(Supplementary Table S2).

In situ observation of dynamic phase transition during carburization

In situ, XRD experiments were performed to gain insights into the dynamic phase transition during Pd₃ZnC_x formation. As shown in Fig. 2a and Fig. S15, PdO phase was initially observed on the fresh PdO/ZnO/Al₂O₃ sample after calcination in air, with characteristic diffraction peaks at 41.93° and 54.79° for PdO(110) and (112), respectively. Upon heating (rate: 5 °C/min) in a syngas atmosphere to 90 °C, a phase transition from PdO to PdH_x hydrides (β-PdH_x) was clearly observed, exhibiting a characteristic β-PdH_x(111) diffraction at 38.92°²². The hydrogen content in β-PdH_x can be estimated by the peak shift through Vegard’s law²⁶ which is around x = 0.489~0.583. With increasing temperature, the β-PdH_x hydrides became thermodynamically unstable, and decompose into α-PdH_x, characterized by the upshift of XRD peak to 39.98° due to lower hydrogen content in α-PdH_x²⁷. Meanwhile, a transitional state (Pd_t) emerged at above 120 °C showing continuous upshifts of XRD diffraction peak to a higher degree. It thus implied a sustained process for carbon and Zn infiltration in a wide temperature window (120–350 °C). Eventually at 400 °C, a complete formation of Pd₃ZnC_x was achieved showing a typical diffraction peak at 40.25°, to a slightly lower angle of Pd₃Zn (40.30°) (see typical XRD patterns in Fig. 1j and Fig. S8). The thermal effect during this process can also be ruled out as thermal expansion usually leads to an opposite shift to lower angle (Fig. S16).

Fig. 2: In situ observation of the dynamic phase transition during Pd₃ZnC_x formation.

a In situ XRD patterns of PdO/ZnO/Al₂O₃ sample under syngas atmosphere at elevated temperatures ranging from 30 to 400 °C. Quasi in situ XPS spectra of Pd 3d (b) and Zn LMM (c) of PdO/ZnO/Al₂O₃ sample under syngas atmosphere at elevated temperatures. The BE of Pd 3d_5/2 for Pd⁰ is 334.8 eV in this work.

Using quasi in situ XPS, we further track the evolution of Pd and Zn chemical state during carburization in syngas. For Pd 3d core level (Fig. 2b), the fresh PdO/ZnO/Al₂O₃ sample shows a typical PdO state with Pd 3d_5/2 at 336.6 eV. At above 90 °C, the PdO was initially reduced showing a prominent metallic Pd component emerging at 334.8 eV, and remained metallic state up to 200 °C. The upshift of Pd 3d_5/2 to 335.4 eV was observed at 400 °C which is attributed to the formation of Pd alloy²³ at this critical temperature. In the case of Zn, here we employ Zn LMM Auger band (Fig. 2c) because of high sensitivity to its chemical nature^28,29. The fresh sample presents a typical ZnO feature with Zn LMM at 989.4 eV. Initial reduction of ZnO occurs at 120 °C exhibiting a characteristic twin-peak feature of Zn metal at 993.5 and 996.7 eV. The small fraction of the metallic Zn component suggests a low content of Zn insertion in the intermediate Pd_t phase at this temperature. The gradually intensified signal of Zn metal (twin-peak feature) thus reflects a sustained infiltration of Zn with increasing temperature, which is in line with the consecutive upshift in XRD results. Until 400 °C, complete alloying/carburization of Pd₃ZnC_x was thus obtained as an indicator of abundant Zn metal (Fig. 2c) and carbide formation (Fig. 1i). It should be noted that ZnO is excess during this process, as indicated by the prominent diffraction peak of ZnO in the XRD (Fig. 2a and Fig. S15) and quasi in situ XPS results (Fig. 2c). Only a small amount of Zn is reduced, and co-infiltrated into the Pd₃ZnC_x, reflecting a inhibitive role of syngas in forming β₁-PdZn alloy.

Environmental TEM was carried out to directly visualize the phase transition of the PdO/ZnO/Al₂O₃ during carburization (Fig. S17). Initially, PdO NPs were observed in the fresh sample under high vacuum conditions at 30 °C (Fig. S17a), with an average d-spacing of 0.260 nm for PdO(101). The formation of PdH_x was observed immediately after introducing a syngas at 30 °C, with a characteristic d-spacing of 0.236 nm for β-PdH_x(111), which remained stable up to 80 °C (Fig. S17b). Compared to metallic Pd NPs the lattice spacing was expanded attributed to the incorporation of H into the Pd NPs. Further increasing the temperature (80–120 °C) leads to a contraction of Pd lattice to 0.226 nm (Fig. S17c), which is very close to the lattice constant of transitional Pd_t species determined by XRD. After fully carburized at 400 °C, the Pd₃ZnC_x phase was eventually formed, showing a typical d-spacing of 0.222 nm for Pd₃ZnC_x as observed ex situ in Fig. S17d.

Combining the in situ XRD, quasi in situ XPS and ETEM results, we unambiguously demonstrated phase transition of PdO → PdH_x → Pd_t → Pd₃ZnC_x during the carburization with syngas. The PdO NPs were first reduced at 40–90 °C and transformed to β-PdH_x hydrides due to H infiltration. Further increasing temperature, the β-PdH_x became thermally unstable up to 120 °C and decomposed gradually to α-PdH_x. Along with the α-PdH_x formation, the growth of intermediate Pd_t species simultaneously occurred at 120 °C with initial insertion of a small amount of metallic Zn as verified by quasi in situ XPS. The gradual upshift of XRD peaks thus indicates the slow and sustained process of Zn/C infiltration that evolves continuously in a temperature range from 120 to 350 °C. The fully carburized Pd₃ZnC_x phase will not be achieved until 350 °C when intense insertion of zinc and carbon occurs.

The critical role of intermediate Pd_t species in promoting Pd₃ZnC_x formation

The intermediate Pd_t transitional state obtained by pretreating the PdO/ZnO/Al₂O₃ with syngas at 200 °C was further characterized. The HAADF-STEM image in Fig. 3a resolves a polycrystalline state of Pd NPs, with co-existence of multi-domains of α-PdH_x (D₍₁₁₁₎ = 0.233~0.235 nm)²⁷ and Pd_t (D₍₁₁₁₎ = 0.223~0.225 nm) growing in between α-PdH_x domains. Although the lattice of Pd_t is very close to that of Pd metal or Pd₃Zn, the incorporation of both carbon and zinc can be validated by EDS line scan (Fig. S18) despite a low amount. The initial incorporation of Zn was also verified by Zn LMM Auger spectra (Fig. 2c). Therefore, we can conclude that the Pd_t species is an intermediate state during transition from PdH_x to Pd₃ZnC_x with low Zn and C infiltration.

Fig. 3: Identification of Pd_t intermediates at the polycrystalline α-PdH_x domain boundaries and mechanistic insights for co-infiltration.

a High-resolution STEM images of Pd_t at 200 °C and the corresponding local FFTs are displayed; crystal planes assigned to α-PdH_x (red dots), the crystal plane indexed to Pd_t (blue dots); yellow dash lines indicate the α-PdH_x/Pd_t interface; black arrows show the growth direction of Pd₃ZnC_x, which is perpendicular to the α-PdH_x(111) planes. b In situ XRD patterns of PdO/ZnO/Al₂O₃ (i) and PdO/Al₂O₃ (ii) treatment by syngas from 100 to 400 °C. c TPSR characterization for PdO/ZnO/Al₂O₃ (red line) and PdO/Al₂O₃ (gray line) under syngas atmosphere monitored by XRD-MS. d Schematic diagram of the continuous growth of Pd_t at the α-PdH_x domain boundaries. The silver ball represents the Pd with the H coordination environment. The royal blue ball represents the Pd with the Zn coordination environment. The lake blue ball represents the Pd with the Zn and C coordination environment.

This initial formation of Pd_t intermediates was clearly resolved occurring on the boundaries between polycrystalline α-PdH_x domains, as depicted by yellow dots in Fig. 3a. The five different domains (R1~R5) on the surface can be identified as face-centered cubic (FCC) structured α-PdH_x hydrides by fast Fourier transform (FFT), showing interplanar spacing D₍₁₁₁₎ = 0.233~0.235 nm characteristic for α-PdH_x(111) with slightly varied H content. Between the two neighboring crystal boundaries, the belt-like Pd_t domain (highlighted by blue) was clearly resolved which separates the adjacent α-PdH_x territories. This strongly indicates the incorporation of carbon and zinc initiated on the domain boundaries of polycrystalline α-PdH_x, and preferentially grows along PdH_x<111> direction (marked with black arrows) when the Pd_t domain expanded. This continuing transition of polycrystalline α-PdH_x to Pd_t and eventually to Pd₃ZnC_x also explains the consecutive upshift of Pd diffraction peak in XRD, rather than a trade-off of two crystalline phases.

At 120 °C, the abundant grain boundaries of α-PdH_x were formed during the decomposition of β-PdH_x hydrides that facilitated the subsequent insertion of Zn and carbon (α-PdH_x + H₂ + ZnO → α-Pd₃Zn + H₂O). The Zn LMM Auger spectra (Fig. 2c) validated the surface enrichment of reduced Zn with the increase of temperature. During this process, α-PdH_x and Pd_t domains coexisted until a fully carburized Pd₃ZnC_x formed at 350 °C.

To elucidate the critical role of Pd_t in the carburization, we designed a control experiment by treating a PdO/Al₂O₃ catalyst (without ZnO) by syngas with the identical procedure. As shown by in situ XRD in Fig. 3b (ii), the phase transition of PdO/Al₂O₃ catalyst can be identified as PdO → β-PdH_x → α-PdH_x. Specifically, after Pd hydride was decomposed at 130 °C, the diffraction peak remained at 40.00°, suggesting the preference and stability of Pd metal rather than carbides under this condition. Herein, the formation of PdC_x carbide is not favored in the absence of Zn, which can be rationalized by the higher capability of H₂ dissociation on metallic Pd surface that inhibits the formation of surface/subsurface C* species from CO³⁰. In contrast, the Zn infiltration during initial carburization of PdO–ZnO (Pd_t phase), despite in a very small amount, can further open the channel for the infiltration of interstitial carbon.

On the other hand, carburization of the pre-synthesized PdZn intermetallic alloy (β₁-PdZn) is also greatly inhibited due to excellent structural stability of PdZn alloy in syngas and semi-hydrogenation reaction conditions, as revealed by in situ XRD results (Fig. S15). Fully occupied interstitial site by Zn incorporation can substantially suppress the infiltration of C* species, which explains the infeasibility of previous attempts for the carburization of PdZn alloy^7,31. Our results thus provide a co-infiltration strategy of both zinc and carbon during initial stage, promoting carbon infiltration on Pd₃Zn (α-PdZn) as a practical approach to synthesize Pd-based intermetallic carbide.

To trace the origin of interstitial carbon, we further performed temperature-programmed surface reaction (TPSR) by coupling a mass spectrometer to the XRD setup (XRD-MS, see “Methods” section). The gaseous products of CO₂ (m/z 44), H₂O (m/z 18), CH₄ (m/z 15), CO (m/z 28) were monitored in real-time during carburization by syngas, as shown in Fig. 3c and Supplementary Figs. S19, 20. Different from monometallic Pd/Al₂O₃, a prominent peak of CO₂ formation was found on PdO/ZnO/Al₂O₃ catalyst in a rather low-temperature regime at 120–200 °C during the initial formation of Pd_t. It thus suggests the source of carbon for the initial infiltration, likely via disproportionation reaction of CO, i.e. 2 CO→CO₂ + C. CO dissociation (disproportionation) on Pd has been earlier reported on stepped Pd(112)³² and undercoordinated Pd sites^20,21. We thus speculated that the enriched dislocations at Pd–Zn_surface/PdH_x phase boundaries could plausibly serve as active sites for this low-temperature CO disproportionation, promoting abundant surface C* species for co-infiltration into Pd_t intermediates. Besides, the CO₂ signal at high-temperature range (>300 °C) can be attributed to the side reaction (water–gas shift reaction) during CO hydrogenation (see Fig. S19).

To further validate the feasibility of the low-temperature disproportionation reaction, we employed DFT calculations to evaluate the reaction-free energies for the formation of Pd₃ZnC_x by reacting Pd₃Zn with CO via Pd₃Zn + 2x CO → Pd₃ZnC_x + x CO₂, with variable carbon content (x, 0.01~1). As shown in Supplementary Table S3, by decreasing C:Pd ratio from 1 to 0.01, a trend becomes evident where lower C:Pd ratio renders the disproportionation reaction more favorable. Notably, at a C:Pd ratio of 0.13, the reaction becomes spontaneous with free energy of −0.18 eV. Particularly for our experimental results where the C:Pd ratio is 0.01, the reaction-free energy is only −0.26 eV that further underscores the thermodynamic feasibility of the new phase formation (Pd₃ZnC_0.03).

Based on the above-mentioned results, we can thus depict an overall picture for the formation of Pd₃ZnC_x intermetallic carbide (Fig. 3d). In syngas, PdO reduction first occurs to generate thermodynamically stable β-PdH_x species (Eq. 1). With increasing temperature (>120 °C), the formation of polycrystalline α-PdH_x species during β-PdH_x decomposition facilitates the reduction of a small amount of vicinal ZnO, forming Pd–Zn_surface surface alloys (likely Pd₃Zn) initiated on the α-PdH_x domain boundaries (Eq. 3)²⁷. The enriched dislocations at Pd–Zn_surface/PdH_x boundaries thus catalyze the low-temperature CO disproportionation reaction (Eq. 4) providing carbon source for Pd_t intermediates (Eq. 5). The intermediate Pd_t continuously expands starting from the domain boundaries and opens up the channel for sustained infiltration of both Zn and C toward Pd₃ZnC_x (Eq. 6). Besides, a semi-quantitative calculation of trend comparisons from PdH to Pd₃Zn and then Pd₃ZnC_x for each transition was carried out by DFT simulation. Figure S21 shows that it is thermodynamically spontaneous for the formation from PdH to Pd₃Zn, and Pd₃Zn to Pd₃ZnC, which further validates the rationality of Pd₃ZnC_x generation.

$${{\rm{PdO}}}+{{{\rm{H}}}}_{2}\to {{\rm{\beta }}}-{{{\rm{PdH}}}}_{{{x}}}+{{{\rm{H}}}}_{2}{{\rm{O}}}$$

(1)

$${{\rm{\beta }}}-{{{\rm{PdH}}}}_{{{x}}}\to {{\rm{\alpha }}}-{{{\rm{PdH}}}}_{{{x}}}+{{{\rm{H}}}}_{2}$$

(2)

$${{\rm{\alpha }}}-{{{\rm{PdH}}}}_{{{x}}}+{{{\rm{H}}}}_{2}+{{{\rm{ZnO}}}}_{{{\rm{suface}}}}\to {{{\rm{H}}}}_{2}{{\rm{O}}}+{{\rm{Pd}}}-{{{\rm{Zn}}}}_{{{\rm{surface}}}}$$

(3)

$$2{{\rm{CO}}}\;\mathop{\longrightarrow}^{{{\rm{Pd}}}-{{{\rm{Zn}}}}_{{{\rm{surface}}}}/{{{\rm{PdH}}}}_{x}}{{{\rm{CO}}}}_{2}+{{\rm{C}}}$$

(4)

$${{\rm{Pd}}}-{{{\rm{Zn}}}}_{{{\rm{surface}}}}+{{\rm{C}}}\to {{{\rm{Pd}}}}_{{{\rm{t}}}}$$

(5)

$${{{\rm{Pd}}}}_{{{\rm{t}}}}\cdots \cdots \to {{{\rm{Pd}}}}_{3}{{{\rm{ZnC}}}}_{{{x}}}$$

(6)

Enhanced catalytic performance of Pd₃ZnC_x in acetylene hydrogenation

The catalytic performance of the unprecedented Pd₃ZnC_x intermetallic carbides was tested for selective hydrogenation of acetylene in an excess of ethylene under two conditions with different H₂/C₂H₂ ratios, by comparing with the conventional PdZn catalyst (Fig. 4). In the tail-end method, the H₂/C₂H₂ ratio is close to the stoichiometric ratio, as presented in Fig. 4a. The Pd₃ZnC_x shows significantly higher activity (acetylene conversion) than PdZn in a wide range of temperature (40 °C~120 °C), while maintaining high ethylene selectivity (>90%). In the front-end method, the hydrogen content is relatively higher, and we choose H₂/C₂H₂ = 4 as shown in Fig. 4b. Notably at high H₂/C₂H₂ ratio of 4 (Fig. 4b), the Pd₃ZnC_x still exhibits superior activity in low-temperature range (40 °C~120 °C), and maintains high ethylene selectivity (>80%) at high acetylene conversion (120 °C~160 °C). As can be seen, the Pd₃ZnC_x has excellent ethylene selectivity under both tail-end and front-end conditions. In contrast, the PdZn catalyst, however, has a sharp decline in ethylene selectivity below 30% at high acetylene conversion under the front-end condition.

Fig. 4: Catalytic performance of Pd₃ZnC_x in acetylene selective hydrogenation reaction.

Acetylene conversion (bar plot) and ethylene selectivity (line plot) of the Pd₃ZnC_x (red) and PdZn (black) catalysts in selective hydrogenation of acetylene in an excess of ethylene, with different H₂/C₂H₂ ratio. a 2% C₂H₂, 2.2% H₂ and 80% C₂H₄ in balance with He (H₂/C₂H₂ = 1); b 2% C₂H₂, 8% H₂ and 80% C₂H₄ in balance with He (H₂/C₂H₂ = 4). Reaction conditions: space velocity = 5000 mL g⁻¹ h⁻¹; pressure = 0.5 MPa. c Durability test on Pd₃ZnC_x for 200 h at 120 °C, and STEM images after the durability test as inset. Pd loading: 0.08 wt%. Feed gas: 2% C₂H₂, 2.2% H₂ and 80% C₂H₄ in balance with He. Reaction conditions: space velocity = 5000 mL g⁻¹ h⁻¹; pressure = 0.5 MPa. d Catalytic performance of Pd₃ZnC_x and other reported transition metal catalysts in literatures for the semi-hydrogenation of acetylene (Data from Table S4). e DFT calculated surface model of PdZn and Pd₃ZnC_x (top view). f Free energy profiles of acetylene hydrogenation on PdZn and Pd₃ZnC_x catalysts.

Furthermore, the mass activity of Pd₃ZnC_x catalyst was calculated using the method proposed by Osswald et al.³³, and compared with other transition metal catalysts in the literature^{34,35,36,37,38,39,40,41}. As displayed in Fig. 4d and Table S4, Pd₃ZnC_x showed the highest selectivity at high acetylene conversions far beyond a vast variety of transition metal catalysts. Meanwhile, Pd₃ZnC_x also exhibits strikingly high activity-per-mass, 1.4 times that of the PdZn intermetallic compounds, which is among the highest reported values under the acetylene-hydrogenation industrial condition operating <120 °C, and remarkably higher turnover frequencies (TOF), ~2.9 times that of conventional PdZn catalyst (Table S5).

The durability of the Pd₃ZnC_x catalyst was further evaluated at 120 °C for 200 h, as shown in Fig. 4c. An excellent durability was observed showing only a slight decrease in selectivity from 97% to 91% after the 200 h time-on-stream test with stable acetylene conversion as high as 94%. Both fresh (Fig. 4c) and spent (Fig. S22) Pd₃ZnC_x catalysts present an excellent structural stability after long-term durability test, showing 0.222 nm spacing for Pd₃ZnC_x(111). Compared with PdZn alloy and other Pd catalysts in literatures, this unprecedented Pd₃ZnC_x intermetallic carbide has shown superior activity, selectivity and stability for the selective hydrogenation of acetylene.

The superior activity of Pd₃ZnC_x can be attributed to its excellent H₂ dissociation ability. The H₂–D₂ exchange experimental results (Fig. S23) showed the formation of HD (deuterium hydride) over the Pd₃ZnC_x at as low as 25 °C, implying a favored hydrogen dissociation even at room temperature. On the contrary, the production of HD was not favored and rapidly declined for PdZn at the same temperature. Figure S24 showed the formation of HD over the Pd₃ZnC_x at the whole test temperature range from 25 to 200 °C, implying a favored hydrogen dissociation on Pd₃ZnC_x sample. The enhanced H₂ dissociation ability of Pd₃ZnC_x can be attributed to the electron-rich Pd sites in Pd₃ZnC_x^42,43, and thus facilitates hydrogenation of acetylene.

DFT calculations further reveal the catalytic superiority of Pd₃ZnC_x in the acetylene hydrogenation reaction. Free energy profiles in Fig. 4f demonstrate that the first hydrogenation barrier on Pd₃ZnC_x (0.58 eV) is lower than that of PdZn (0.77 eV), promoting the hydrogenation activity. The desorption of ethylene on Pd₃ZnC_x is spontaneous, which accounts for its high selectivity of ethylene. Furthermore, the higher barrier for deep hydrogenation on Pd₃ZnC_x (0.93 eV) also inhibits further reactions, thereby further enhancing the selectivity toward ethylene. Calculations of the reaction pathways were also conducted for other comparative catalysts, including Pd and Pd₃Zn (Fig. S25). The results similarly underscore the exceptional catalytic performance of Pd₃ZnC_x. Besides, high H₂/C₂H₂ ratio condition was also considered for Pd₃ZnC_x catalyst with H* coverage calculation in Fig. S28 (Pd₃ZnC_x–H catalyst). Based on the thermodynamically preferred reaction pathway (Fig. S29), we performed calculations of the complete acetylene hydrogenation reaction profiles on Pd₃ZnC_x–H catalyst (Fig. S30), which reveals that under H* coverage, the reaction barrier of Pd₃ZnC_x–H for the first hydrogenation step is still lower than PdZn, proving the superior catalytic performance of Pd₃ZnC_x. It should be mentioned that a common disproportionation elementary step in the hydrogenation of acetylene, that is, the instead of generating CH₂CH₂, the hydrogenation of *CHCH₂ to *CHCH₃, is also computationally compared (Fig. S31). Both thermodynamic and kinetic results show that for Pd₃ZnC_x, *CHCH₂ will prefer to generate CH₂CH₂ rather than *CHCH₃. Higher barrier of C–C coupling was also found on Pd₃ZnC_x (Figs. S32, S33), suggesting its high resistance to coke formation. Moreover, the energy barrier for H₂ dissociation on Pd₃ZnC_x (Fig. S34) is significantly lowered compared to PdZn. This indicates that Pd₃ZnC_x is also conducive to facilitating H₂ dissociation, thereby promoting the occurrence of selective hydrogenation of acetylene⁴⁴.

The origin of the superior catalytic performance of Pd₃ZnC_x was further investigated. Due to the electronegativity differences among Pd (2.20), Zn (1.65) and C (2.55), there is a significant charge transfer from Zn to Pd, resulting in an electron enrichment of 0.16 e⁻ for Pd in Pd₃Zn, and from Pd to C, adding 0.09 e⁻ upon carbon incorporation. This charge redistribution is also confirmed by charge density difference (Fig. S35), leading to an increased electron density on Pd and a lower d-band center (−2.23 eV) in Fig. S36, weakening C₂H₄ adsorption. Projected Density of States (PDOS) analysis (Fig. S34) indicates a broadened d-band and more active electrons near the Fermi level in Pd₃ZnC_x, enhancing acetylene hydrogenation and H₂ dissociation. The adsorption configurations are shown in Fig. S37. Geometrically, the most stable C₂H₂ adsorption configuration is at the fcc site on Pd₃ZnC_x surfaces, with each C atom bridging two Pt sites, enhancing C₂H₂ adsorption. C₂H₄ shows a favorable π-bonding mode, similar to PdZn (Fig. S38), indicating geometric favorability for Pd₃ZnC_x.

Discussion

Using co-infiltration strategy, we reported an unprecedented intermetallic carbide of Pd₃ZnC_x in one-step carburization of PdO–ZnO via syngas. State-of-the-art in situ characterizations explicitly revealed a dynamic phase transition as PdO → PdH_x → Pd_t → Pd₃ZnC_x. The initial formation of Pd_t transitional state with low Zn and C content was identified as a crucial intermediate promoting phase transition and sustainable infiltration of zinc and carbon toward Pd₃ZnC_x formation. Comparing to conventional PdZn intermetallic catalysts and other reported Pd-based catalysts, the Pd₃ZnC_x intermetallic carbide shows by far the best catalytic performance for the selective hydrogenation of acetylene, maintaining high selectivity (>90%) and excellent stability (~200 h) at high acetylene conversion. The insertion of interstitial C enables the electron enrichment of Pd₃Zn sites that further facilitate ethylene desorption and the hydrogen dissociation resulting in its enhanced catalytic performance. Our results thus developed a co-infiltration strategy for one-step synthesis of Pd₃ZnC_x intermetallic carbide, as a practical approach to synthesizing high-performance Pd-based intermetallic carbide toward selective hydrogenation of acetylene.

Methods

Chemicals and materials

Palladium (II) diacetylacetonate (34.7 wt% Pd) was purchased from Alfa Aesar Shanghai Co., Ltd.; zinc nitrate hexahydrate was purchased from Damao Tianjin Co., Ltd.; toluene was purchased from Sinopharm Chemical Reagent Co., Ltd.; α-Al₂O₃ (surface area of 5.4 m² g⁻¹) was purchased from Aluminum Industry Shangdong Co., Ltd. All the reagents were of analytical grade and used as-received without further purification.

Sample preparation

ZnO/Al₂O₃ catalyst was synthesized by incipient wet impregnation of Al₂O₃ powder with Zn(NO₃)₂ as precursor. Briefly, 546.0 mg Zn(NO₃)₂·6H₂O was dissolved in 1.3 g DI H₂O to form a clear solution, and to this solution 2.0 g Al₂O₃ powder was added. The resultant slurry was allowed for standing at room temperature overnight, and then dried at 120 °C for 12 h and calcined at 400 °C for 3 h. The Zn loading of the ZnO/Al₂O₃ was determined to be 5.5 wt% by inductively coupled plasma atomic emission spectrometry (ICP-OES).

The Pd was supported on ZnO/Al₂O₃ through a wet impregnation method. In detail, 286.5 mg (5 wt% Pd) or 4.6 mg (0.08 wt% Pd) palladium diacetylacetonate was dissolved in 60 g toluene under ultrasonication for several minutes. Then, 2.0 g ZnO/Al₂O₃ powder was added to the solution and stirred for 30 min to obtain a homogeneous dispersion. After removing toluene by vacuum distillation with a rotary evaporator at 60 °C, the samples were calcined at 400 °C for 3 h and served as the fresh catalyst for further treatment or testing. The low-loading sample (0.08 wt% Pd) with an average particle size of 5 nm was used for catalytic testing and H₂–D₂ exchange reaction, while the high-loading sample (5 wt%) with an average particle size of 16 nm was used for in situ experiments to gain a satisfied signal/noise ratio. ZnO was excessive compared with Pd.

Material characterizations

Aberration-corrected HAADF-STEM

Aberration-corrected HAADF-STEM imaging was performed on a JEOL JEM-ARM200F equipped with a probe Cs-corrector working at 200 kV. For the HAADF imaging, a convergence angle of ~23 mrad and a collection angle range of 68–174 mrad were adapted for the incoherent atomic number imaging. The elemental composition as well as distribution were analyzed with an energy dispersive X-ray analyzer (EDS, EX-230 100 m² detector) equipped on the microscope. The sample was ultrasonically dispersed in ethanol for 15–20 min, and then a drop of the suspension was dropped on a copper TEM grid coated with a thin holey carbon film.

Environmental TEM

Environmental TEM experiments were carried out on the Titan Themis G3 ETEM (Thermo Scientific Company) in the electron microscopy center at the Dalian Institute of Chemical Physics Chinese Academy of Sciences, which was working at 300 kV with a Cs corrector for parallel imaging (CEOS GmbH) and a measured resolution of better than 1.0 Å. The sample was loaded on a Si₃N₄ membrane and heated in situ using a micro-electromechanical system (MEMS) based sample holder (FEI NanoEX).

Inductively coupled plasma optical emission spectrometer (ICP-OES)

Pd and Zn contents of the samples were determined by ICP-OES. Before the experiment, a microwave digestion system (Anton Paar Multiwave 3000) was used to dissolve ~100 mg of PdO/ZnO/Al₂O₃ sample, which was then analyzed with the ICP-OES method (PerkinElmer 7300DV) to obtain the loading of Pd and Zn.

In situ X-ray diffraction (XRD)

In situ XRD measurements were conducted on a PANalytical Empyrean diffractometer with in situ reactor (XRK900) using CuKα radiation, operated at 40 kV and 40 mA. The PdO/ZnO/Al₂O₃ sample with 5 wt% Pd was transferred to an in situ reactor with the beryllium window. Then the syngas (10%CO–50%H₂–40%He) or 50%H₂–50%He flow at 50 mL min⁻¹ was introduced after the inert gas flushing, with a heating rate of 5 °C min⁻¹ from room temperature to 400 °C. The XRD patterns were recorded at desired temperatures from 36° to 60°. All XRD data were measured multiple times on different batches of samples which were consistent and reproducible.

XRD-MS setup

The gas composition during in situ XRD was monitored by an online mass spectrometer (Pfeiffer OmniStar) to the outlet of the in situ reactor. The detection point of mass spectrometry was located 150 cm down flow of the XRD in situ reactor, and excess exhaust gas was discharged from the exhaust pipe through a three-way valve.

Quasi in situ X-ray photoelectron spectroscopy (XPS)

Quasi in situ XPS was performed on a ThermoFisher ESCALAB 250Xi+ equipped with a gas cell directly mounted to the sample chamber. The quasi in situ measurements enable direct sample transfer under UHV condition for XPS analysis after a certain treatment without exposure to air. All the spectra were recorded using monochromated X-ray irradiation Al Kα (hv = 1486.7 eV) at room temperature. The peak positions were calibrated by adjusting the C 1s peak to 284.8 eV and Al 2p peak to 74.5 eV. All the samples were treated in the gas cell with syngas (10%CO–50%H₂–40%He) or 50%H₂–50%He flow at 50 mL min⁻¹.

X-ray absorption fine structure (XAFS) spectra

XAFS spectra at Pd K-edge of the samples were measured at beamline 14 W of the Shanghai Synchrotron Radiation Facility (SSRF) in China. The output beam was selected by Si (311) monochromator. The energy was calibrated by the Pd foil. Before measurement, the samples were reduced at 400 °C under syngas (Pd₃ZnC_x) or under hydrogen (PdZn), and cooled to room temperature. Then the samples were transferred to glove box without exposure to air. The samples were sealed in Kapton films in the glove box before XAFS measurement. The data were collected at room temperature under transmission mode by using solid state detector. Athena software package was employed to process the EXAFS data.

H₂–D₂ exchange reaction

H₂–D₂ exchange reaction was evaluated at a Micromeritics Chemisorption Analyzer AutoChem II equipped with a Pfeiffer OmniStar mass spectrometer. In total, 100 mg catalyst was first reduced in 10%CO–50%H₂–40%He or 50%H₂–50%He at 400 °C for 2 h and then was cooled down to room temperature in He. The catalyst was then tested for H₂–D₂ exchange reaction under a standard condition: a gas mixture of H₂ and D₂ with a flow rate of 18 and 14 mL/min, respectively.

Catalytic testing

The selective hydrogenation of acetylene over Pd-based catalysts was tested in a fixed-bed flow reactor. The total flow rate was kept at 4.17 ml min⁻¹. The amount of Pd-based catalysts was 50 mg. Before the reaction test, all the catalysts were reduced in 10%CO–50%H₂–40%He or 50%H₂–50%He at 400 °C for 2 h, and then cooled down to reaction temperature. The catalytic performance was evaluated in a gas feed of 2% C₂H₂, 2.2% H₂ and 80% C₂H₄ in balance with He for H₂/C₂H₂ = 1; and 2% C₂H₂, 8% H₂ and 80% C₂H₄ in balance with He for H₂/C₂H₂ = 4. Space velocity = 5000 mL g⁻¹ h⁻¹; pressure = 0.5 MPa. Both the inlet and outlet gases were analyzed online with an Agilent Technologies 7890B gas chromatograph equipped with a flame ionization detector (FID) packed with a capillary column (Agilent HP-PLOT Al₂O₃ S; 50 m × 0.530 mm). To accurately quantify the concentration of gas components, the relatively inert CH₄ gas was used as the internal standard doped in the reaction atmosphere. The acetylene conversion and ethylene selectivity were calculated according to equations:

$${{{\rm{C}}}}_{2}{{{\rm{H}}}}_{2}\;{{\rm{conversion}}}\;(\%)=\frac{{{{\rm{C}}}}_{2}{{{\rm{H}}}}_{2}\left({{\rm{in}}}\right)-{{{\rm{C}}}}_{2}{{{\rm{H}}}}_{2}\left({{\rm{out}}}\right)}{{{{\rm{C}}}}_{2}{{{\rm{H}}}}_{2}\left({{\rm{in}}}\right)}\times 100$$

(7)

$${{{\rm{C}}}}_{2}{{{\rm{H}}}}_{4}\;{{\rm{selectivity}}}\;(\%)=\left(1-\frac{{{{\rm{C}}}}_{2}{{{\rm{H}}}}_{6}\left({{\rm{out}}}\right)-{{{\rm{C}}}}_{2}{{{\rm{H}}}}_{6}\left({{\rm{in}}}\right)}{{{{\rm{C}}}}_{2}{{{\rm{H}}}}_{2}\left({{\rm{in}}}\right)-{{{\rm{C}}}}_{2}{{{\rm{H}}}}_{2}\left({{\rm{out}}}\right)}\right)\times 100$$

(8)

C₂H₂(in) and C₂H₂(out) are the concentrations of acetylene at the inlet and outlet of the reactor, respectively. C₂H₆(in) and C₂H₆(out) are the concentrations of ethane at the inlet (unavoidable impurity in feed gas) and outlet of the reactor, respectively.

The TOF value was calculated based on the surface moles of Pd. The metal dispersion of Pd for Pd₃ZnC_x and PdZn samples was 24.7 and 21.1%, respectively, detected by CO pulse chemisorption.

Calculation methods

We conducted calculations using the Vienna ab initio Simulation Package with the projector augmented wave method and the generalized gradient approximation in the form of the Bayesian error estimation functional with van der Waals corrections (BEEF–vdW) for periodic boundary conditions^45,46,47. The phase structures were built with different C ratios, optimized, and analyzed using XRD spectra. The cutoff energy for the plane-wave basis set was 400 eV, with fixed bottom two layers and a 15 Å separation along the Z direction to minimize slab interactions. A 0.02 eV/Å criterion was used for atomic force convergence during geometry optimization. K-point sampling was done with a 3 × 3 × 1 mesh using the Monkhorst–Pack method⁴⁸. The basic modeling structures are provided on GitHub (https://github.com/TJU-ECAT-AI/Pd3ZnCx-DFT-Structures).

Transition states (TS) were found using the CI-NEB technique with six intermediate images, followed by refinement using the dimer method with a force tolerance of 0.02 eV/Å^49,50. A normal mode analysis confirmed the TS structures. Free energy reaction profiles were calculated to include entropy and ZPE. We used the equation S_ads = 0.7S_gas − 3.3R to calculate the adsorbate entropy⁵¹, with S_gas obtained from a thermodynamic database using heat capacity extrapolation for unlisted temperatures.