Structural and Kinetics Understanding of Support Effects in Pd-Catalyzed Semi-Hydrogenation of Acetylene

Yueqing Co, Xiohu Ge, Yurou Li, Rui Si, Zhijun Sui, Jinghong Zhou, Xuezhi Dun,*,Xinggui Zhou

a State Key Laboratory of Chemical Engineering, East China University of Science and Technology, Shanghai 200237, China

b Shanghai Synchrotron Radiation Facility, Shanghai Institute of Applied Physics, Chinese Academy of Sciences, Shanghai 201800, China

Keywords:Acetylene semi-hydrogenation Reaction kinetics Support effects Electronic effects Pd local environment

ABSTRACT In this study, the support effects on the Pd-catalyzed semi-hydrogenation of acetylene have been investigated from the structural and kinetic perspectives.According to the results of kinetic analysis and X-ray photoelectron spectroscopy, hydrogen temperature-programmed reduction, temperature-programmed hydride decomposition, and in situ X-ray diffraction measurements, using carbon nanotubes as support for Pd nanocatalysts with various sizes instead of α-Al2O3 decreases the Pd0 3d binding energy and suppresses the formation of undesirable palladium hydride species, thus increasing the ethylene yield.Furthermore, X-ray absorption spectroscopy, high-resolution transmission electron microscopy, and C2H4 temperature-programmed desorption studies combined with density-functional theory calculations reveal the existence of a unique Pd local environment,containing subsurface carbon atoms,that produces positive geometric effects on the acetylene conversion reaction.Therefore,tailoring the Pd local environment and electronic properties represents an effective strategy for the fabrication and design of highly active and selective Pd semi-hydrogenation catalysts.

1. Introduction

Elucidating the mechanism of the Pd-catalyzed semihydrogenation of trace acetylene in the ethylene-rich feedstock used for polyethylene production is of primary scientific and industrial importance for the further catalyst modification and design [1-4]. This process is widely recognized as a remarkable structure-sensitive reaction that exhibits strong Pd size effects on the catalytic performance [5-7]. Kinetic and structural insights into such size effects reveal that the Pd(111) surface promotes the conversion of acetylene and formation of C4byproducts,while the edge sites catalyze the formation of ethane byproduct [8]. It has been also shown that small-sized Pd catalysts exhibit different electronic properties that strongly affect the reaction kinetics[6,8],demonstrating the existence of significant Pd particle size effects on the semi-hydrogenation of acetylene.

In addition to particle size effects, support effects also strongly influence the catalytic performance of Pd[9-13].In the majority of these studies, simultaneous changes in the size and electronic properties of supported Pd nanoparticles were detected.Moreover,when reducible and/or acidic metal oxides were used as supports,metal-support interface effects and the effects of support acidity on the coke formation were observed. For this reason, the weakly interactive and chemical inert α-Al2O3is commercially used as the support material [14,15]. The application of another weakly interactive carbon support revealed that its inherent weak acidity produced no apparent effects on catalyst deactivation [16], and the carbon-supported Pd catalyst exhibited strong positive effects on the semi-hydrogenation of acetylene as compared with those of the α-Al2O3-supported catalyst [8,17]. Moreover, the carbon support possesses other unique properties that distinguish it from the α-Al2O3support such as the high specific surface area for metal immobilization, tailorable carbon surface characteristics, and effective electron transport between the metal and the support[18-20]. Therefore, it is necessary to examine the structural and kinetic aspects of support effects for optimizing the catalyst design.

In this work, we utilized weakly interactive carbon nanotube(CNT) and α-Al2O3supports in the Pd-catalyzed semihydrogenation of acetylene to investigate their geometric and electronic effects on the catalytic performance of Pd nanoparticles without the interferences mentioned above. After excluding the influence of the size of Pd, the support effects on the modified Pd nanoparticles and their catalytic properties were examined by combining kinetic analysis and density-functional theory (DFT)calculations with various experimental techniques such as in situ X-ray diffraction (XRD), hydrogen temperature-programmed reduction (H2-TPR), temperature-programmed hydride decomposition (TPHD), C2H4temperature-programmed desorption (C2H4-TPD), X-ray photoelectron spectroscopy (XPS), and X-ray absorption spectroscopy. The resultant unique Pd local environment and electronic properties that were able to suppress the undesirable palladium hydride formation and enhance the catalytic performance of supported Pd nanoparticles were identified.

2. Experimental and theoretical methods

2.1. Catalyst preparation

Pd catalysts were prepared by an impregnation method using CNTs and α-Al2O3as the support materials. For this purpose, 2 g of commercial CNTs (Beijing Cnano Technology Limited, China)and α-Al2O3were impregnated with calculated amounts of chloropalladic acid based on the nominal Pd loading of 2 wt%.The resultant catalyst precursors were aged at 25 °C for 12 h and subsequently dried at 100 °C for 12 h. After that, the precursors were reduced at 160 °C for catalytic testing and characterization.The obtained CNT-supported and α-Al2O3-supported Pd catalysts were denoted as Pd/CNT and Pd/α-Al2O3, respectively.

2.2. Characterization

Transmission electron microscopy (TEM) and high-resolution TEM (HRTEM) observations were performed on a JEOL JEM-2100 transmission electron microscope (JEOL Ltd., Japan) with an acceleration voltage of 200 kV to determine Pd particle sizes and catalyst shapes. The obtained images are shown in Fig. S1 in Appendix A. The electronic properties of the catalysts were analyzed by XPS on a Kratos XSAM-800 spectrometer equipped (Kratos Analytical Ltd., UK) with an Al Kα anode. The binding energies of the Pd 3d XPS spectra obtained for all samples were calibrated with respect to the position of the C 1s peak (284.6 eV). Xray absorption fine spectra of Pd/CNT and Pd/α-Al2O3at the Pd K-edge (E0= 24 350 eV) were recorded on the BL14W1 beamline of the Shanghai Synchrotron Radiation Facility (China) utilizing Pd foil and PdO as the references. Data extraction and processing were conducted using the Athena and Artemis codes [21,22].

H2-TPR and TPHD experiments were performed on a Micrometrics Autochem 2920 chemisorption analyzer (Canada) equipped with a thermal conductivity detector(TCD)according to the procedure described in our previous work[8].In a typical H2-TPR experiment, approximately 200 mg of the tested sample was reduced inside a quartz U-type tube under a 10% H2/Ar atmosphere in the temperature range of 45-800 °C. The temperature was ramped at a heating rate of 10 °C·min-1. In a TPHD experiment, approximately 200 mg of the sample was reduced at 160 °C for 1 h and then cooled down to 45°C under H2/Ar atmosphere.Subsequently,the sample was purged with pure Ar gas for 45 min to remove the weakly adsorbed hydrogen species followed by the temperature increase from 45 to 400°C at a rate of 10°C·min-1.Hydrogen evolutions during the TPR and TPHD processes were analyzed by the TCD detector.Ethylene catalytic TPD studies were conducted using the procedure outlined in our previous work [19].

In situ XRD measurements were performed on a D8 ADVANCE diffractometer (Brucker, Germany) in the 2θ range of 38°-42° to characterize the formation of subsurface hydrogen species when the samples were first reduced and heated to 200 °C under N2gas for 30 min.After cooling to 25°C,in situ XRD studies were initiated. The sample was first kept under N2for 30 min to achieve stability. Subsequently, it was purged with 10% H2/Ar for 30 min followed by N2treatment for another 30 min (XRD patterns were continuously recorded during these steps).Afterwards,the sample was heated to 100 °C, and its XRD patterns were recorder again.After collecting data at 100 °C for 30 min, the sample was cooled down to finish the testing procedure.

2.3. Catalytic studies

The catalytic performances of the prepared catalysts for the semi-hydrogenation of acetylene were evaluated using a commercial stainless steel reactor (μBenchCAT, Altamira Instruments,USA).Approximately 20 mg of the sample mixed with quartz sand at a ratio of 1:10 was reduced in the reactor tube under H2atmosphere at 160°C for 2 h and then cooled down to the reaction temperature under N2atmosphere. Afterwards, a reactant mixture containing 1% of C2H2, 3% of H2, 20% of C2H4, and N2balance was introduced to the reactor at a flow rate of 120 mL·min-1.The reactants and products were analyzed by a 3000 Micro gas chromatograph (Inficon, Germany) equipped with a TCD detector. The following equations were employed to calculate the acetylene conversion and ethylene selectivity:

where Cidenotes the determined concentration of i component in the inlet and outlet streams. For example, CC2H2,inletand CC2H2,outletare the concentration of acetylene in the inlet and streams, respectively. XC2H2and SC2H4represent the conversion of acetylene and selectivity to ethylene, respectively. According to the results of our previous work[8],the Pd(111)surface contained catalytically active sites for acetylene hydrogenation. Herein, the reaction rates related to acetylene conversion were normalized with respect to the number of Pd(111) atoms rather than to the total Pd amount to determine turnover frequencies for the active sites (TOFactivesite) and compare their intrinsic catalytic activities(the corresponding calculation details are provided in Supplementary data and Table S1 in Appendix A). Some parameters of the Pd/CNT catalysts with different Pd particle sizes varied by adjusting the Pd loading were taken from our previous work [8].

2.4. DFT calculations

All DFT calculations were performed using the Vienna ab initio simulation package (VASP). The Kohn-Sham wave functions were expanded by a plane-wave basis set in the calculations with a cut-off energy of 500 eV. The projected augmented wave method was employed to define the interactions between the ionic core and valence electrons[23],and the Perdew-Burke-Ernzerhof functional based on the generalized gradient approximation was used to describe the exchange-correlation effects [24]. The Brillouin zone was sampled by the Monkhorst-Pack k-point mesh [25]. The 5 × 5 × 5 and 5 × 5 × 1 grids were used for the bulk and slab models, respectively. Geometry optimization was performed until the forces on each atom were below 0.05 eV·?-1and the total energy differences were less than 1×10-5eV.The transition state of the diffusion of a hydrogen atom from the surface to the subsurface was located by a dimer method [26]. The adsorption energy(Eads)was calculated as Eads=Eadsorbate/slab-Eadsorbate-Eslab,where Eadsorbate/slab, Eadsorbate, and Eslabwere the energies of the adsorbate on the slab, free adsorbate, and clean slab, respectively.

Bader analysis [27] was conducted to calculate electronic charges on various atoms in order to examine the electronic interactions between Pd and C, and the charge density difference (Δρ)isosurface was computed by the formula Δρ=Δρ(Pd(111)-C)-Δρ(Pd(111))-Δρ(C), where Δρ(Pd(111)-C), Δρ(Pd(111)), and Δρ(C) were the charge densities of the relaxed Pd(111) surface modified by the subsurface C atoms located at the octahedral sites[28,29], Pd atoms at the corresponding surface locations, and C atoms at the corresponding surface locations, respectively. For the electron transfer between the surfaces and adsorbed ethylene species, the charge density difference isosurface was calculated as Δρ=Δρ(molecule/surface)-Δρ(surface)-Δρ(molecule), where Δρ(molecule/surface), Δρ(surface), and Δρ(molecule) were the charge densities of the molecule/surface system, relaxed surface,and ethylene molecule in their final configurations, respectively.

3. Results and discussion

3.1. Support effects and their kinetic interpretation

Fig.1(a)shows the effects of both supports on the Pd-catalyzed semi-hydrogenation of acetylene. Herein, the TOFactivesitevalues were calculated from the corresponding numbers of Pd active sites(Pd(111) atoms [8]), which catalyzed the conversion of acetylene molecules. At close Pd particle sizes, the CNT support resulted in higher TOFactivesitemagnitudes as compared with those obtained for the α-Al2O3support, indicating the existence of strong electronic effects on the reaction kinetics. The XPS spectra of the Pd/CNT and Pd/α-Al2O3catalysts depicted in Fig. 1(b) show that the Pd/CNT catalyst exhibits a lower Pd03d5/2binding energy as compared with that of the Pd/α-Al2O3catalyst. As demonstrated in previous studies of the Pd electronic effects on the acetylene conversion reaction [9,30-32], the electron-rich Pd/CNT catalyst inhibits the adsorption of acetylene, making a larger number of active sites available for hydrogen activation and promoting acetylene hydrogenation. For this reason, the Pd/CNT catalyst has relatively high TOFactivesitevalues.

Fig.1. (a) Correlations of TOFactive site at 30 °C and Ea with the Pd particle size.(b) Pd 3d XPS spectra recorded for the Pd/CNT and Pd/α-Al2O3 catalysts (the original Pd/CNT catalyst data were taken from our previous work). (c) Acetylene conversion and (d) ethylene selectivity plotted as functions of the reaction temperature. (e) Ethylene selectivity, C4 selectivity, and ethane selectivity plotted as functions of the acetylene conversion adjusted by varying the reaction temperature (derived from the data presented in panels (c) and (d) and Fig. S3 in Appendix A). (a) Reproduced from Ref. [8] with permission of ACS Publications, ?2019.

To provide kinetic insights into Pd electronic effects, the resultant activation energy(Ea)is plotted as a function of the Pd particle size in Fig.1(a).The Eaof the Pd/α-Al2O3catalyst is lower than that of the Pd/CNT catalyst,which positively contributes to the catalytic activity of the former system.In contrast to Ea,the pre-exponential factor and concentration produce a stronger negative effect on the activity of the Pd/α-Al2O3catalyst (Fig. S2 in Appendix A),ultimately endowing the Pd/CNT catalyst with a higher activity.

Fig. 1(a) also shows that the smaller-sized Pd catalysts exhibit much higher TOFactivesitevalues than those of the larger-sized catalysts, which can be related to the difference in their electronic properties[8].Moreover,the last three Pd/CNT catalysts with close Pd03d binding energies possess similar Eamagnitudes, suggesting that the quantitative, not qualitative differences between the Pd active sites are mainly responsible for the observed differences in their reaction rates. In other words, these catalysts are located in the dominant region of geometric, not electronic effects. In contrast, the former two Pd/CNT catalysts and Pd/α-Al2O3catalyst are located in the dominant region of electronic effects.

The catalytic semi-hydrogenation of acetylene is both a wellknown structure-sensitive reaction and a sequential-parallel process[1,5-7,33].The Pd electronic effects on the acetylene conversions and product selectivities observed at various reaction temperatures are shown in Figs.1(c)-(e)and Fig.S3(Appendix A).The Pd/CNT catalyst exhibits higher acetylene conversion and ethylene selectivity as compared with those of the Pd/α-Al2O3catalyst,which can be due to the lower Pd03d5/2binding energy of the former system that increases the hydrogenation activity as discussed above and promotes the desorption of ethylene product [9,30-32].

Meanwhile,typical volcano relationships between the ethylene selectivity and reaction temperature are observed for both catalysts in Fig.1(d).Increasing the temperature changes the acetylene conversion;hence,a fair comparison of the product selectivities of the two catalysts should be based on the close acetylene conversions in the sequential-parallel process. Product selectivities are plotted against acetylene conversion in Fig.1(e),which shows that the ethylene selectivity of the Pd/CNT catalyst is higher than that of the Pd/α-Al2O3catalyst. The results obtained in previous studies [34] suggest that on the one hand, the increased ethylene selectivity can be related to the enhanced ethylene desorption at higher temperatures that prevents the over-hydrogenation of ethylene. On the other hand, the decreased ethylene selectivity likely originates from both the lower surface coverage of acetylene and higher hydrogen coverage leading to over-hydrogenation, which is supported by the inverse volcano curve shape obtained for the relationship between the ethane selectivity and the reaction temperature in Fig. 1(e).

Fig.1(e)also shows that the dependences of the ethylene and C4selectivities on acetylene conversion have similar shapes, which can be explained by their occupations of similar Pd active sites,such as those of the Pd(111)surface reported in our previous work[8]. Moreover, the relatively weak correlation between the C4selectivity and acetylene conversion obtained for the Pd/CNT catalyst is most likely caused by its lower Pd03d binding energy that decreases the adsorption energy of acetylene.

3.2.Support effects on the formation of undesirable palladium hydride phase (PdHx) by-products

Previous studies [9,28,35-40] have shown that the palladium hydride phase(PdHx)is easily formed under hydrogen atmosphere,which is detrimental to the semi-hydrogenation of alkynes,leading to over-hydrogenation and formation of alkanes. Therefore, the effects of the Pd/CNT and Pd/α-Al2O3catalysts on the formation of PdHxspecies were investigated by conducting H2-TPR measurements (Fig. 2(a)). Both catalysts exhibit negative hydrogen consumption peaks in the range of 100-150 °C, which can be related to the reduction of Pd species. Moreover, as compared with the Pd/CNT catalyst, the Pd/α-Al2O3catalyst produces an additional positive (hydrogen formation) peak at approximately 75 °C that likely originated from the decomposition of palladium hydride[17,41,42]. To verify this hypothesis, TPHD measurements were performed and the results are shown in Fig. 2(b). Here, the Pd/α-Al2O3catalyst exhibits one strong palladium hydride decomposition peak at approximately 75°C,while no such peaks are observed for the Pd/CNT catalyst, which was consistent with the results of H2-TPR studies presented above. The obtained data indicate that in contrast to the α-Al2O3support, the CNT support can suppress the formation of undesirable palladium hydride species; it also explains the higher selectivity of the Pd/CNT catalyst as compared with that of the Pd/α-Al2O3catalyst.

In situ XRD measurements were conducted to determine the support effects on the structural properties of supported Pd nanoparticles and achieve a better understanding of the remarkably different formation behaviors of undesirable palladium hydride species described above.Figs.2(c)and(d)show the evolutions of the XRD patterns of Pd/α-Al2O3and Pd/CNT under variable treatment conditions,respectively.The XRD patterns of the freshly reduced catalysts and the corresponding supports depicted in Fig. S4 (Appendix A) exhibit the characteristic Pd(111) diffraction peak (JCPDS 46-1043), which partially overlaps with the α-Al2O3peak. To illustrate the structural evolution of the Pd(111) peak under variable treatment conditions,planar projection transformations of the three-dimensional in situ XRD patterns were performed to yield the two-dimensional images displayed in Figs.2(c)and(d).Switching from N2to the H2-containing atmosphere at 30 °C apparently shifted the position of the Pd(111) diffraction peak of the Pd/α-Al2O3catalyst from 40.1°to 39.8°,which could be caused by the formation of α-PdHxspecies [43]. Unfortunately, it is difficult to obtain more accurate information regarding the α-PdHxformation at 38.8° [44,45] because the corresponding peak fully overlaps with the Al2O3(110) peak (JCPDS 81-1667). Furthermore,after switching back to the N2atmosphere and elevating the temperature to 100°C,the diffraction peak shifted back to 40.0°,which was likely caused by the PdHxdecomposition as suggested by the results of H2-TPR and TPHD studies presented in Figs. 2(a) and(b), respectively. In contrast, no visible Pd(111) diffraction peak shifts were detected for the Pd/CNT catalyst, indicating that PdHxformation was suppressed by the CNT support,which was in good agreement with the results of H2-TPR and TPHD analyses.Notably,the 2θ value obtained for the Pd/CNT catalyst (39.9°) was lower than that of the Pd/α-Al2O3catalyst(40.1°),corresponding to a larger Pd(111) lattice spacing. This phenomenon is related to the presence of subsurface carbon atoms in the CNT-supported Pd nanoparticles, which will be discussed in detail in Section 3.3.

3.3. Unique local environment of the Pd/CNT catalyst

Extended X-ray absorption fine structure(EXAFS)studies of the Pd/α-Al2O3and Pd/CNT catalysts were conducted to identify the difference in their local environments. Fig. 3(a) shows the Fourier transforms of the Pd K-edge EXAFS spectra recorded for these catalysts as well as for the reference PdO and Pd foil samples.The R-space spectrum of the Pd/α-Al2O3catalyst exhibits a peak at approximately 2.74°, which matches the Pd-Pd bond length[46-48]; however, its intensity is lower than that of the peak obtained for the reference Pd foil due to the finite size effect of Pd nanoparticles [49-51]. At similar sizes of supported metallic Pd nanoparticles, the Pd/CNT catalyst demonstrates a broad peak in the range of 1.8°-2.2° and a Pd-Pd peak with a much lower intensity, which is shifted to higher R values with respect to the position of the main Pd-Pd peak [46]. This phenomenon can be attributed to the existence of scattering interactions between Pd atoms and the first Pd-C shell [46,48]. The above-mentioned XRD patterns showing a small Pd lattice expansion in the Pd/CNT catalyst indicate the presence of carbon atoms in the Pd particle shell,which is schematically depicted in Fig. 3(a). This observation is consistent with the previously reported data suggesting that the thermal treatment of carbon-supported Pd catalysts leads to the dissolution of the carbon atoms belonging to the oxygencontaining groups of the carbon support (such as carboxylic ones)[52,53]into the Pd particles located at the octahedral sites[28,29].

The HRTEM and fast Fourier transform(FFT)images of the Pd/α-Al2O3and Pd/CNT catalysts presented in Figs. 3(b) and (c), respectively, show that both catalysts exhibit the characteristic diffraction patterns of metallic Pd and that the angle between the Pd(111) and Pd(100) planes in the Pd/CNT catalyst remains almost unchanged, indicating the presence of interstitial carbon atoms in Pd nanoparticles and absence of palladium carbide species.These results are in good agreement with the EAXFS and XRD data presented above. To examine the effects of interstitial carbon atoms on the semi-hydrogenation of acetylene, an energetically favorable Pd(111)-Csubsurfacemodel describing the subsurface carbon atoms located at the octahedral Pd sites [28,29] was constructed (Fig. 4(a)). It was further utilized to theoretically study the influence of these atoms on the hydrogen diffusion process and related formation of undesirable PdHxspecies due to the over-hydrogenation activity.In this model,the most stable subsurface H atoms below the clean Pd(111) surface are located at the octahedral sites,which match the locations of the most stable subsurface C atoms on the Pd(111)-Csubsurfacesurface blocking the favored H diffusion pathway. Alternatively, H atoms can also diffuse to other metastable (tetrahedral) subsurface sites; the corresponding diffusion barrier for this process is 0.51 eV, which is higher than that on the clean Pd(111) surface (0.41 eV), as shown in Fig. 4(b). Therefore, it can be concluded that unlike the α-Al2O3support,the CNT support generates subsurface C atoms at the octahedral sites of supported Pd nanoparticles and thus inhibits the formation of undesirable PdHxspecies, promoting the semihydrogenation of acetylene into ethylene.

Fig. 3. (a) Fourier transforms of the Pd K-edge EXAFS spectra recorded for the Pd/α-Al2O3 and Pd/CNT catalysts as well as for the PdO and Pd foil reference specimens. The right panel contains a schematic diagram of the CNT-supported Pd nanoparticles with subsurface carbon atoms.Typical HRTEM images of the(b)Pd/α-Al2O3 and(c)Pd/CNT catalysts with the corresponding (FFT) patterns displayed in the insets.

Fig.4. (a)The most stable adsorption configurations of the subsurface C and H atoms on the Pd(111)surface.(b)Energy barriers of the hydrogen diffusion from the Pd(111)surface to the subsurface over the Pd(111)-Csubsurface surface.

3.4. Origin of selectivity enhancement

As demonstrated in the previous sections of this work, the Pd/CNT catalyst produces a stronger positive effect on the formation of ethylene from acetylene as compared with that of the Pd/α-Al2O3catalyst,owing to the lower Pd03d binding energy of the former catalyst and its unique Pd local environment containing subsurface C atoms that suppresses the formation of undesirable PdHxspecies. To achieve a better understanding of the effects of the Pd electronic properties and local environment of the Pd/CNT catalyst on the ethylene selectivity enhancement, C2H4-TPD measurements were conducted for the Pd/α-Al2O3and Pd/CNT catalysts by investigating the adsorption/desorption characteristics of the ethylene product.As shown in Fig.5(a),the Pd/α-Al2O3catalyst exhibits three desorption peaks centered at 120, 240, and 490 °C.Previous C2H4-TPD studies revealed that the adsorbed C2H4easily decomposed into hydrogen atoms and C2-fragments at elevated temperatures[54].These three peaks can be attributed to the desorption of three different C2H4species and/or decomposed fragments (e.g., C2H3) from the top, bridge, and hollow sites based on the results of DFT calculations[28].In contrast,the Pd/CNT catalyst generates only one broad low-temperature desorption peak(Fig. 5(a)), corresponding to relatively weak ethylene adsorption that increases the ethylene selectivity.

Fig.5. (a)C2H4-TPD profiles of the Pd/CNT and Pd/α-Al2O3 catalysts.(b)The most stable adsorption configurations of ethylene on the Pd(111)and Pd(111)-Csubsurface surfaces with the corresponding adsorption energies.Top and front views of the charge density distributions of the(c)Pd(111)-Csubsurface surface and(d)adsorbed ethylene molecule.The isosurface value is 0.003 e·?-3. The red and green isosurfaces represent the accumulation and depletion of the electron density, respectively.

Finally,DFT calculations were performed to investigate the effect of the Pd local environment on ethylene adsorption. As shown in Fig. 5(b), both the Pd(111) and Pd(111)-Csubsurfacesurfaces favor ethylene adsorption at the bridge sites. The ethylene adsorption energy of the latter surface is significantly lower than that of the Pd(111)surface,owing to the presence of subsurface carbon atoms in the CNT-supported Pd nanoparticles[3,28].This can account for the higher ethylene selectivity and formation rate of the Pd/CNT catalyst mentioned above. To examine this issue in more detail,the charge density distribution of the Pd(111)-Csubsurfacesurface was compared with that of the configuration containing an adsorbed ethylene molecule(Figs.5(c)and(d)).The calculated electron charge density is highest on the Pd(111)-Csubsurfacesurface around its carbon positions and lowest at the palladium positions,which suggests the existence of an electron transfer from Pd atoms to the subsurface carbons. However, for the ethylene molecule adsorbed on the Pd(111)-Csubsurfacesurface, the subsurface carbon positions possess the lowest electron densities, indicating a back-donation of electrons to the palladium positions upon ethylene adsorption. This observation can help to better interpret the lower ethylene adsorption energy on the Pd(111)-Csubsurfacesurface as compared with that of the ethylene adsorbed on the Pd(111) surface. By taking into account the results described in Sections 3.2 and 3.3, the presence of subsurface carbon atoms produces two remarkable effects:suppression of the formation of undesirable PdHxspecies and enhancement of the desorption of the ethylene product due to its electron back-donation contribution. Notably, the results of XPS analysis showed that the lower Pd03d binding energy and high electron density of the Pd/CNT catalyst favored ethylene desorption and increased the ethylene selectivity [9,30-32]. Therefore, it can be concluded the CNT catalyst support not only possesses a lower Pd03d binding energy than that of the α-Al2O3support for positive electronic effects, but also creates a unique Pd local environment containing subsurface carbon atoms for positive geometric effects.

4. Conclusions

In this study, support effects on the Pd-catalyzed semihydrogenation of acetylene were investigated from the structural and kinetic perspectives. Compared with the α-Al2O3support, the CNT catalyst support reduced the Pd03d binding energy and suppressed the formation of PdHxspecies to enhance the reaction kinetics in terms of ethylene selectivity and formation rate.Meanwhile, the positive geometric effects produced by the Pd/CNT catalyst were attributed to its unique Pd local environment containing subsurface carbon atoms that promoted the electron back-donation to Pd atoms and inhibited the adsorption of ethylene.The obtained results can be used in the fabrication and design of highly efficient Pd semi-hydrogenation catalysts by tailoring the Pd local environment and electronic properties.

Acknowledgements

This work was financially supported by the Natural Science Foundation of China (21922803, 21776077, and 22008067), the Innovation Program of Shanghai Municipal Education Commission,the Shanghai Natural Science Foundation (17ZR1407300 and 17ZR1407500), the Program for Professor of Special Appointment(Eastern Scholar) at Shanghai Institutions of Higher Learning, the Shanghai Rising-Star Program (17QA1401200), the China Postdoctoral Science Foundation(2020M681202),the Open Project of State Key Laboratory of Chemical Engineering(SKLChe-15C03),the State Key Laboratory of Organic-Inorganic Composites (oic-201801007),and the Fundamental Research Funds for the Central Universities(222201718003). We also thank the BL14W1 beam line of Shanghai Synchrotron Radiation Facility for the XAFS measurement.

Compliance with ethics guidelines

Yueqiang Cao, Xiaohu Ge, Yurou Li, Rui Si, Zhijun Sui, Jinghong Zhou, Xuezhi Duan, and Xinggui Zhou declare that they have no conflict of interest or financial conflicts to disclose.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.eng.2020.06.023.