Grafting nanometer metal/oxide interface towards enhanced low-temperature acetylene semi-hydrogenation

Metal/oxide interface is of fundamental significance to heterogeneous catalysis because the seemingly “inert” oxide support can modulate the morphology, atomic and electronic structures of the metal catalyst through the interface. The interfacial effects are well studied over a bulk oxide support but remain elusive for nanometer-sized systems like clusters, arising from the challenges associated with chemical synthesis and structural elucidation of such hybrid clusters. We hereby demonstrate the essential catalytic roles of a nanometer metal/oxide interface constructed by a hybrid Pd/Bi2O3 cluster ensemble, which is fabricated by a facile stepwise photochemical method. The Pd/Bi2O3 cluster, of which the hybrid structure is elucidated by combined electron microscopy and microanalysis, features a small Pd-Pd coordination number and more importantly a Pd-Bi spatial correlation ascribed to the heterografting between Pd and Bi terminated Bi2O3 clusters. The intra-cluster electron transfer towards Pd across the as-formed nanometer metal/oxide interface significantly weakens the ethylene adsorption without compromising the hydrogen activation. As a result, a 91% selectivity of ethylene and 90% conversion of acetylene can be achieved in a front-end hydrogenation process with a temperature as low as 44 °C.

Bi2O3. Unfortunately, this claim is not indisputably supported by the characterization.
8) The authors discuss the shifts observed in the K-edge XANES in the context of d-state filling, yet Ledge is far more sensitive since promoted core electrons are populating d-states directly. The choice of the K-edge must be motivated.
9) The DFT model development has not been performed in an exhaustive fashion. It appears the model is motivated by the need to promote an idea rather than a clear connection with reality. Specifically, both top and bottom of the Bi2O3 model slab are Bi terminated producing a highly reduced model that will more readily allow for electron density transfer to Pd. The bottom of the slab should be stoichiometrically terminated by oxygen. The termination of the top of the slab needs to be determined via ab initio thermodynamics. If this were done, it is almost assured that the most promising termination is oxygen terminated. Then the authors could determine the reducibility of the surface using H2 to motivate the production of a oxygen-lean surface termination. This more thorough computational study is critical since the experimental characterization do not clearly motivate a oxygen-lean Bi2O3 surface.
10) The literature support for highly reduced oxide surfaces promoting the reduction of supported metals needs to be scrutinized carefully. Many studies that promoted this view were performed under the naturally reducing conditions of UHV surface science. Many decades of follow-up studies aimed to understand the effect of non-model catalyst synthesis and pretreatment conditions and contact with the ambient. These studies motivated that oxygen-lean oxide surfaces are not stable and likely fully oxidized. Even a handful of UHV STM studies illustrated that oxygen-vacancy laden surfaces were not stable even under a well-baked UHV system condition and what were claimed to be oxygen vacancies were actually hydroxyl groups from H2O dissociating rapidly on the reduced oxide surface. This study is unfortunately promoting a less than realistic metal-oxide interface and must work to thoroughly support their claim that the Bi2O3 surface is metal terminated. Unfortunately, it appears that they have not achieved this. 11) Calling the catalyst of focus a "model catalyst" is quite misleading. It is quite ill-defined and the source of its catalytic activity is still unclear.
12) The authors need to show the catalytic performance of the PdBi IMC catalyst for comparison.
13) The effect of Bi could be a simple site blocking mechanism. This seems to be an effect that is completely avoided in the discussion of the catalyst. Was this effect directly investigated? It could be by simply producing a Pd/TiO2 catalyst and then adding Bi to the catalyst with an appropriate reduction pre-treatment procedure afterwards. 14) How is Bi oxidized to Bi2O3 in the synthesis method? There appears to be no elevated temperature oxidation or reduction pretreatment applied after initial catalyst synthesis. 15) No information is provided for sample handling before XPS, XRD, or XAS measurements. This information must be provided so the reader can predict the chemical state of the catalyst and how ambient may have affected composition.
16) At a minimum, the DFT model should be motivated significantly and the critical feature of surface termination be discussed directly. This is a vital feature of the model system and the electron density transfer to the metal. Maybe even add the oxygens to the bottom of the slab and show that the charge transfer still occurs.
Reviewer #3 (Remarks to the Author): This paper investigates a nanometer metal/oxide interface -hybrid Pd/Bi2O3 -in selective hydrogenation of acetylene, with promising catalytic performances. An original structural model for the catalytic system is proposed, which is supported by several arguments, but which can also be discussed. 1-There are several possible phases for the 1:1 BiPd IMC, for which the shortest Pd-Bi distance ranges from 2.83 A to 2.86 A [Canadian Mineralogist 28 (1990)  However, the authors talk about the Pd1.0Bi/TiO2-IMC, suggesting that they identified a specific phase, which is unfortunately not defined with the conventional way (space group and lattice parameters, instead of PDF#29-0238).
They invoked the shift in R-space (0.06 A, Fig. S9b) to assess that the Pd-Bi distances are not the same in Pd1.0Bi/TiO2 and in Pd1.0Bi/TiO2-IMC. However, it does not exclude that two different IMCs phases are formed.
2-Information about the crystal structure of Bi2O3 is also lacking. There are several phases with the Bi2O3 stoichiometry (Mat. Letters 64 (2010) 2247-2250, Chem. Phys. Letters 378 (2003 395-399, PRB 82 (2010) 024106, PRB 83 (2011) 214102, Acta Cryst. C44 (1988 587-589) but the authors do not provide any information about the phase. Experimental lattice parameters are given in the section "Computational methods", but no reference is given, suggesting that it corresponds to the measurements done by the authors. However, this is not clear for me. In addition, the choice of the (100) orientation is not really well-argued. How is it consistent with the assertion that most Pd clusters are identified to bond and hybridize with Bi2O3 clusters without any fixed orientation relationship or facet preference?
3-According to the experimental measurements, the Bi-Pd distance is larger in Pd1.0Bi/TiO2 than in Pd1.0Bi/TiO2-IMC. If the assumptions made about the structures of Pd1.0Bi/TiO2 and Pd1.0Bi/TiO2-IMC are correct, different Bi-Pd distances calculated by DFT, using the model already built for Pd1.0Bi/TiO2 and a model that the authors can build for Pd1.0Bi/TiO2-IMC, based on the crystal structure determined by XRD, could support this assumption. Such a calculation would also provide information about the charge transfer in the Pd1.0Bi/TiO2 system and in the model built for Pd1.0Bi/TiO2-IMC, which can be useful to strengthen the analysis of the spectroscopic data. Fig. 4b provides a reaction path for the hydrogenation reaction, (i) on the surface the Pd1.0Bi/TiO2 model provided by the authors and (ii) on Pd(111). The values of the barriers are given for Pd1.0Bi/TiO2 in the supplemental materials (Tab. S3), but not for Pd(111). The latter could have been compared with the literature, to assess the reliability of the computational method (constrained minimization method to locate the transition state structures).

General response:
We sincerely thank the editor and all reviewers for their valuable feedback that we have based on to improve the quality of our manuscript. The reviewer comments are listed below in italicized font and specific concerns have been numbered.

Reviewer #1 (Remarks to the Author):
In this paper, the author has synthesized a hybrid Pd/Bi2O3 cluster loaded on the TiO2 support, which present good performance on the selective hydrogenation of acetylene to ethylene. Multiple characterizations have been applied to prove the hybrid structure, and the existence of intra-cluster electron transfer. Here I still have some question about this paper before publication.
1. The previous reviewers have referred my concern about the XPS and XAFS results, it seems the author had enough evidence to prove the interface structure of Pd and Bi2O3 now, while, I still doubt about the uniformity of this structure, as we can see from the elemental mapping of Pd1.0/Bi2O3/TiO2. There are some individual Pd clusters can be seen, which cannot be ignored by the author, this may also be the reason for the perfect performance of this catalyst for acetylene hydrogenation.
Author response: We agree with the reviewer that there could be some individual Pd clusters in the Pd1.0/Bi2O3/TiO2. But the amount of these individual Pd clusters is quite small. The uniformity of the Pd-Bi2O3 hybrid clusters is evidenced by elemental mapping, ADF-STEM images, XAFS, and XPS.
(1) The concern of the existence of individual Pd clusters may due to the color selection for elemental mappings of Pd1.0/Bi2O3/TiO2 in Fig. 1e. Generally, Human vision is most sensitive to a green color. In the revised Figs. 1e (see below), we used green for Bi and red for Pd. In addition, more elemental mappings were presented in Fig. S4. As can be seen from these figures, most Pd are spatially correlated to Bi, suggesting the Pd grafted Bi2O3 hybrid cluster structure. Individual Pd clusters in the Pd1.0/Bi2O3/TiO2 are quite rare.
(2) Careful inspection on several representative ADF-STEM images of Pd1.0/Bi2O3/TiO2 (Figs. 1c, d and S5) allows the discrimination of Pd-and Bicontaining hemi-clusters from their distinct contrasts for adjacent clusters. The usually smaller hemi-clusters featuring less bright contrast (marked by circle) are directly attached to larger and brighter ones assigned to Bi2O3 clusters, further confirming that Pd is grafted onto the surface of Bi2O3 clusters. Notably, most such Pd clusters are identified to bond and hybridize with Bi2O3 clusters without any fixed orientation relationship or facet preference, probably due to the ultra-small size and large strain of the clusters. Two representative ADF-STEM images of Pd-Bi2O3 nanoclusters with well-resolved Pd-and Bi-containing hemi-clusters are shown in Figs. 1c and d, of which the image contrasts and FFT patterns closely resemble the simulated ones of artificially constructed hybrid cluster models with different orientation relationship between two types of hemi-clusters made of Pd and α-Bi2O3 structures respectively.
The above observations unambiguously validate the proposed Pd-Bi2O3 hybrid structural model.
(3) Generally, due to the size effect, the XANES absorption edge of cluster samples will shift to higher energy (blue-shift) than that of foil (Ref.: J. Phys. Chem. C 2017, 121, 361-374). In this study, however, Pd1.0/Bi2O3/TiO2 shows a marked red-shift of Pd Kedge XANES as compared with Pd foil and Pd/TiO2 (Fig. 2e). Similarly, Pd1.0/Bi2O3/TiO2 also exhibits a downshift of Pd 3d and 3p XPS that is opposite to the particle-size-induced shift (Fig. S8). The opposite trend arises from the Bi2O3-to-Pd intra-cluster electron transfer and therefore suggests that most Pd clusters are grafted on the surface of Bi2O3 clusters.
Taking all these results together, it is conclusive that the individual Pd clusters are minority and only make minor contribution to the good performance of Pd1.0/Bi2O3/TiO2 for acetylene hydrogenation. As suggested by the reviewer, we discussed the presence and the influence of individual Pd clusters in the revised manuscript as follows.
Followed by the secondary deposition of Pd, the clusters of Bi2O3 intergrown with Pd particles can be found according to the low-magnification ADF-STEM images (Fig.   S1).
Careful inspection on several representative ADF-STEM images of Pd1.0/Bi2O3/TiO2 (Figs. 1c, d and S5) allows the discrimination of Pd-and Bicontaining hemi-clusters from their distinct contrasts for adjacent clusters. The usually smaller hemi-clusters featuring less bright contrast (marked by circle) are directly attached to larger and brighter ones assigned to Bi2O3 clusters, further confirming that Pd is grafted onto the surface of Bi2O3 clusters. Notably, most such Pd clusters are identified to bond and hybridize with Bi2O3 clusters without any fixed orientation relationship or facet preference, probably due to the ultra-small size and large strain of    As suggested by the reviewer, we prepared PdAg3/TiO2 and tested its catalytic performance in acetylene hydrogenation. PdAg3/TiO2 shows similar catalytic performance as PdAg3/Al2O3. Specifically, the C2H4 selectivity at 95% C2H2 conversion is ~ 10%, which is much lower than that of Pd1.0/Bi2O3/TiO2 (84.5%) and Pd0.2/Bi2O3/TiO2 (88.8%). It is important to highlight that the comparison with PdAg3/Al2O3 is not meant to show the different promote effect of Bi and Ag. Instead, it helps us to evaluate the potential of Pd1.0/Bi2O3/TiO2 in industrial application. To compare the different promoting effect of Ag and Bi, we also synthesized Pd1.0Ag/TiO2 with the same method for the synthesis of Pd1.0/Bi2O3/TiO2. Catalytic test suggests that Pd1.0Ag/TiO2 over-hydrogenates acetylene to ethane at even room temperature, with a C2H4 selectivity of −167%. This result further proves the superior catalytic performance of Pd1.0/Bi2O3/TiO2 and their unique Pd-Bi2O3 hybrid-cluster structure.
The catalytic performance of PdAg3/Al2O3 obtained in this study is similar to the literature results (Ref.: U.S. patent 20200094226 A1;Chem. Commun. 2013, 49, 8350-8352). As noted by the reviewer, it is a good catalyst for this reaction, but the C2H4 selectivity at high C2H2 conversion still has room for improvement. In industry, the traces of acetylene in excess ethylene must be reduced to a very low level (typically < 15 ppm) before the downstream ethylene polymerization. A good hydrogenation catalyst should therefore be able to clean up C2H2 (typically, C2H2 conversion > 99%) with a good C2H4 selectivity. That's why we raise the reaction temperature of PdAg3/Al2O3 and compare its C2H4 selectivity at a high C2H2 conversion. We agree with the reviewer that the hydrogenation reaction is related a lot with the temperature.
As suggested by the reviewer, we added the following text into the revised manuscript to illustrate why we choose PdAg3/Al2O3 as a reference catalyst.
Both Pd/TiO2 and well-established Pd1Ag3/Al2O3 catalysts 28,43 were evaluated for comparison with Pd1.0/Bi2O3/TiO2. The composition and synthesis procedure of the PdAg3/Al2O3 catalyst is the same as OleMax@251, a widely used industrial catalyst for acetylene hydrogenation 43 .

It seems the role of the interface is not reflected in the paper, except enrich the electron of Pd. Meanwhile, based on the DFT, the reaction was happened on the surface of Pd, instead of the interface.
Author response: The interface plays important role in both the formation and the function of Pd-Bi2O3 hybrid nanoclusters. From a structural perspective, the unique interaction between interfacial Pd and Bi atoms allows the selective deposition of Pd onto Bi2O3 nanoclusters and therefore the formation of Pd-Bi2O3 hybrid clusters. While from a functional perspective, the Pd-Bi bonding and associated charge transfer at the interface allow the modulation of the electronic structure of Pd to optimize the bonding strength of reaction intermediates (e.g., ethylene). It is important to note that in this study, most Pd atoms are grafted on the surface of Bi2O3 clusters (please also refer to our reply to Comment 1). These interfacial Pd atoms are therefore essential parts of the interface. The hydrogenation reaction was happened on these interfacial Pd atoms and therefore reflected the role of the interface. The Pd8-cluster model used in DFT calculations was constructed based on the experimental observations. It turns out that the interfacial Pd atoms are negatively charged and show weaker adsorption towards ethylene than Pd(111), which consequently leads to higher C2H4 selectivity.
These results suggest that a simple site blocking mechanism cannot explain the beneficial effect of Bi in this study. PdBi/TiO2 composed of PdBi IMC (Fig. S9) shows 100% conversion of C2H2 and negative selectivity towards ethylene (−1319%) at room temperature. The strong exothermic effect of unselective acetylene hydrogenation eventually leads to a runaway temperature up to 58.5 °C. These results exclude PdBi IMC as the active site for Pd1.0/Bi2O3/TiO2 and further indicates the critical role of the nanometer Pd/Bi2O3 interface in the catalytic selectivity of Pd.
As suggested by the reviewer, we added more discussion to illustrate the importance of the interface.

The author declaimed that the hybrid clusters feature an intra-cluster electron transfer towards Pd and enables much weaker ethylene bonding, more direct evidence should be given to prove this. For example, C2H4-pulse or C2H4 TPD (detected by MS).
The microcalorimetric study cannot reflect both the adsorption strength and the adsorption capacity at the same time. Typically, there are three kinds of peaks in TPD patterns. The peak at 50 ~ 90 °C represents weakly adsorbed π-bonded ethylene, which consequently desorbs without decomposition. The peak centered at 105 ~ 150 °C represents di-σ-bonded ethylene, which undergoes decomposition followed by the recombination with hydrogen to produce ethylene as well as ethane. The peak above 250 °C corresponds to triply bound species, i.e., ethylidyne. As shown in Fig. S12, Pd/TiO2 exhibits a large peak centered at ~ 115 °C, suggesting that it mainly adsorb ethylene via a strong di-σ-bonding. In the case of Pd1.0/Bi2O3/TiO2, the intensity of the peak at 115 °C is significantly reduced, accompanied by the emergence of a peak at ~ 65 °C. These results suggest that the adsorption configuration of C2H4 is changed from strong di-σ-bonding to weak πbonding, which likely originates from the low Pd-Pd coordination and the Bi2O3-to-Pd intra-cluster electron transfer in Pd1.0/Bi2O3/TiO2. The change of adsorption mode eventually leads to reduced adsorption capacity of Pd1.0/Bi2O3/TiO2, as evidenced by the microcalorimetric study.
In addition to Fig. S12, the following discussion was added into the revised manuscript.
These observations can be attributed to the unique geometric and electronic structure of hybrid cluster, similar to the results reported in the alloying of Pd. 47,48 The low Pd-Pd coordination and the Bi2O3-to-Pd intra-cluster electron transfer likely change the adsorption configuration of C2H4 from stable ethylidyne to weak π-bonded C2H4 and promote the desorption of ethylene as the desired the product. To confirm this hypothesis, we further performed the temperature programmed desorption (TPD) of ethylene by monitoring the mass signal of m/e = 27 (Fig. S12). According to the literatures, the peak at ~ 65 °C could be assigned to weak π-bonded ethylene species, which readily desorb without decomposition. 49 The peak centered at ~ 115 °C originates from di-σ-bonded ethylene, which undergoes decomposition followed by the recombination of surface hydrocarbon species and hydrogen to produce ethylene and ethane. 49 Compared with Pd/TiO2, Pd1.0/Bi2O3/TiO2 exhibits a much weaker peak at ~ 115 °C but a significantly larger peak at 65 °C. These results confirm the adsorption configuration of C2H4 is changed from the strong σ-bonding for Pd/TiO2 to weak πbonding for Pd1.0/Bi2O3/TiO2.

Review Overview
In general, there is merit in the study and the potential for high impact, but the authors focus on a fine microscopic detail that is quite difficult to verify experimentally. Not that the focus is too fine, but that the phenomenon is hard to prove. Likewise, computationally proving the presence of a reduced metal-oxide interface is also not trivial given the role of material synthesis and atomic H spill-over effects. General response: We sincerely thank the reviewer for valuable comments and suggestions. In the revised manuscript, we have made considerable effort to address the proposed issues from both experimental and theoretical perspectives. The point-bypoint response to the reviewer's comments are listed below.

Specific Issues
1) The catalyst naming should be adjusted such that it is indicative of the materials composition, e.g., Pd/Bi2O3/TiO2.
Author response: As suggested by the reviewer, we have changed the name of Pd1.0Bi/TiO2 to Pd1.0/Bi2O3/TiO2 in the revised manuscript and supporting information.
2) The EXAFS signal at 2.6 Ang also corresponds to Pd-Pd scattering. Are the authors ascribing this signal to Pd-Bi bonds as well? The oxidation test that illustrated the loss of this feature could simply be explained through the formation of PdO and the loss of Pd-Pd scattering paths. See: 10.1039/C8CP00517F. Author response: It is important to highlight that EXAFS is an element-specific analysis because the energy required to excite a core electron is determinate of the element being analyzed. In this manuscript, the discussion of the signal at 2.6 Å is focused on the Bi L3-edge EXAFS spectra of Pd1.0/Bi2O3/TiO2. It reflects only the coordination environment of Bi rather than Pd. Therefore, this signal does not correspond to Pd-Pd scattering. The loss of this feature in the oxidation test cannot be explained through formation of PdO and the loss of Pd-Pd scattering.
3) The claim of a Pd-Bi bond could be motivated computationally by performing a surface termination study of Bi2O3. However, it is suggested that this analysis would show an oxygen-terminated Bi2O3 and that Pd is not in direct contact with Bi. However, this bond is likely not a critical feature in producing electron density polarization of Pd if the surface is heavily metal terminated. The extra electron density will go where it is most energetically favorable anyways.
Author response: We sincerely thank the reviewer for valuable comments and suggestions. In the revised manuscript, we added more experimental evidence and DFT calculations to illustrate the rationality and the origin of the Pd-Bi bond.
(1) Experimental. It is important to note that the Pd-Bi2O3 hybrid clusters model was constructed mainly based on experimental observations. The Pd-Bi bond is evidenced by the characteristic peak at ~ 2.6 Å of Bi L3 EXAFS. The Pd-Bi bond length obtained from the fitting of Pd K and Bi L3 edge is 2.79 ± 0.03 Å, which is significantly smaller than the Pd-Bi bond length in Pd-O-Bi moieties (~ 3.5 Å). This result suggests that the observed spatial correlation of Pd-Bi pairs arises from the direct bonding between Pd-and Bi-terminated clusters. We agree with the reviewer that oxygen-terminated Bi2O3 is generally more stable than Bi-terminated Bi2O3.
Pd directly deposited onto Bi2O3 likely bonds with O rather than Bi. To confirm this hypothesis, we synthesized a Pd/Bi2O3 sample by directly depositing Pd onto commercial Bi2O3 support. Interestingly, no characteristic peak at ~ 2.6 Å was observed in the Bi L3 EXAFS of Pd/Bi2O3, excluding the presence of Pd-Bi bond (please see below the new Figure 2c). In this study, Pd is in direct contact with Bi mainly because of the unique synthetic procedure of Pd1.0/Bi2O3/TiO2. Pd1.0/Bi2O3/TiO2 was synthesized by a stepwise photodeposition method, followed by a mild oxidation of Bi in the air. In brief, Bi 3+ was reduced by the photogenerated electrons to produce Bi 0 clusters on TiO2. Subsequently, Pd was deposited onto Bi/TiO2 to form Pd/Bi/TiO2. When the sample was exposed to the air, Bi was spontaneously oxidized into Bi2O3 because the Gibbs free energy of the reaction (2Bi(s) + 1.5O2(g) = Bi2O3(s) (298−544 K), ∆G Θ (J mol -1 ) = −574887 + 99.71T) is minus. The procedure was schematically illustrated in the new Figure 1a. The Pd-Bi bond preserves during the mild oxidation of Bi to Bi2O3 at room temperature occurred after the deposition of Pd onto Bi/TiO2, as evidenced by the characteristic Bi-Pd peak in Bi L3 EXAFS (Fig. 2c). More interestingly, 36% of this peak preserves even after oxidation in air at a higher temperature of 150 °C   (2) DFT calculations. As suggested by the reviewer, we calculated surface energies of O-and Bi-terminated Bi2O3(100) surfaces, which are listed in Table R1. It turns out that O-termination is thermodynamically more stable than Bi-termination. This is consistent with the argument of this reviewer and the experiments further performed in our work.  (100) surface are readily removed under reaction conditions, and the Bi-layer would be exposed. Therefore, we used the Bi-terminated Bi2O3(100) for further studies.
In terms of the electron transfer between Bi and Pd, we also added more DFT calculations (please also refer to Comment 9 for details). Following the reviewer's suggestion in Comment 9, we considered a Bi2O3 slab where its bottom is terminated by O and the top terminated by Bi. The new Bi2O3-Pd8 structure is shown in Fig. S16.
In addition, we used the model in Fig. S16 to conduct Bader charge analysis, and the data are given in Fig. S17. One can find from the figure that the charge-transfer effect of the new Bi2O3-Pd8 cluster structure follows the same trend as we proposed in the original manuscript (see Fig. S11), i.e., the Bi2O3 transfers electrons to Pd.  Part of the above discussion was added into the supporting information to illustrate the model development. In addition to new Figs. 1a, 2c, S16 and S17, we also added the following discussion into the revised manuscript.
In brief, Bi 3+ was reduced by the photogenerated electrons to produce Bi 0 clusters on TiO2. Subsequently, Pd was deposited onto Bi/TiO2 to form Pd/Bi/TiO2. When the sample was exposed to the air, Bi was spontaneously oxidized into Bi2O3 because the Gibbs free energy of the reaction is minus. The procedure was schematically illustrated in the Fig. 1a.
Photo-deposition procedure is critical to ensure the formation of direct Pd-Bi bonding in Pd/Bi2O3 clusters as depicted in Fig. 1a (Fig. 2c). In this study, photo-deposition procedure ensures the Pd cluster deposited on reduced Bi 0 clusters, and allows the formation of Pd-Bi bonding during the synthesis as depicted in Figure   1. The Pd-Bi bond preserves during the mild oxidation of Bi to Bi2O3 at room temperature occurred after the deposition of Pd onto Bi/TiO2, as evidenced by the characteristic Bi-Pd peak in Bi L3 EXAFS (Fig. 2c). More interestingly, 36% of this peak preserves even after oxidation in air at a higher temperature of 150 °C Model of Bi2O3 supported Pd8 cluster catalyst was built according to the experimental characterization results, and the structure of which is shown in Fig. 4a (please see the details of model development in supporting information).  (Fig. 2c), which further confirms that the peak at ~ 2.6 Å is a characteristic peak for Pd-Bi bond. In this study, Pd is in direct contact with Bi mainly because of the unique synthetic procedure of Pd1.0/Bi2O3/TiO2.

4) The fluxionality of the small clusters even under XAS conditions may affect
Pd1.0/Bi2O3/TiO2 was synthesized by a stepwise photo-deposition method, followed by a mild oxidation of Bi in the air. In brief, Bi 3+ was reduced by the photogenerated electrons to produce Bi 0 clusters on TiO2. Subsequently, Pd was deposited onto Bi/TiO2 to form Pd/Bi/TiO2. When the sample was exposed to the air, Bi was spontaneously oxidized into Bi2O3 because the Gibbs free energy of the reaction (2Bi(s) + 1.5O2(g) = Bi2O3(s) (298−544 K), ∆G Θ (J mol -1 ) = −574887 + 99.71T) is minus. The procedure was schematically illustrated in the new Fig. 1a. The Pd-Bi bond preserves during the mild oxidation of Bi to Bi2O3 at room temperature occurred after the deposition of Pd onto Bi/TiO2, as evidenced by the characteristic Bi-Pd peak in Bi L3 EXAFS (Fig. 2c).
More interestingly, 36% of this peak preserves even after oxidation in air at a higher temperature of 150 °C (Pd1.0/Bi2O3/TiO2-ox). These results clearly indicate the good stability of Pd-Bi bond in Pd1.0/Bi2O3/TiO2 and suggest Pd supported on a Bi terminated Bi2O3 as the structural model for Pd1.0/Bi2O3/TiO2 in the hydrogenation reaction conditions. Taking all these results together, the structure models used in DFT calculations are reliable.
As suggested by the reviewer, the following discussion was added into the revised manuscript.
In brief, Bi 3+ was reduced by the photogenerated electrons to produce Bi 0 clusters on TiO2. Subsequently, Pd was deposited onto Bi/TiO2 to form Pd/Bi/TiO2. When the sample was exposed to the air, Bi was spontaneously oxidized into Bi2O3 because the Gibbs free energy of the reaction is minus. The procedure was schematically illustrated in the Fig. 1a.
Photo-deposition procedure is critical to ensure the formation of direct Pd-Bi bonding in Pd/Bi2O3 clusters as depicted in Fig. 1a. Theoretically, Bi2O3 favours an oxygen-termination. Pd supported on Bi2O3 would be in direct contact with O rather than Bi. We prepared a Pd/Bi2O3 sample by directly depositing Pd onto commercial Bi2O3 support. Interestingly, no characteristic peak at ~ 2.6 Å was observed in the Bi L3 EXAFS of Pd/Bi2O3, excluding the presence of Pd-Bi bond (Fig. 2c). In this study, photo-deposition procedure ensures the Pd cluster deposited on reduced Bi 0 clusters, and allows the formation of Pd-Bi bonding during the synthesis as depicted in Figure   1. The Pd-Bi bond preserves during the mild oxidation of Bi to Bi2O3 at room temperature occurred after the deposition of Pd onto Bi/TiO2, as evidenced by the characteristic Bi-Pd peak in Bi L3 EXAFS (Fig. 2c). More interestingly, 36% of this peak preserves even after oxidation in air at a higher temperature of 150 °C Model of Bi2O3 supported Pd8 cluster catalyst was built according to the experimental characterization results, and the structure of which is shown in Fig. 4a (please see the details of model development in supporting information).  In addition to new Fig. S10, we also included the following discussion into the manuscript.

8) The authors discuss the shifts observed in the K-edge XANES in the context
Moreover, Pd1.0/Bi2O3/TiO2 shows a much weaker Pd L3 white line intensity at ~3173 eV than Pd/TiO2 (Fig. S10), which indicates an enhanced d-band filling. Author response: We sincerely thank the reviewer for valuable comments and suggestions. We developed the Pd8-cluster model based on the idea that the model used in calculations should be consistent with experimental observations. The obtained interatomic distance between Pd and Bi on Pd8-cluster structure is found to be similar with experimental values. Hence the model successfully represents the local Pd-Bi environment in the experiment. Furthermore, the reason we used the slab without Ooccupied bottom terminations was to keep the surface layer identical to the bottom layer and to avoid possible effect of dipole moment on energy calculations.

9) The DFT model development has not been performed in an exhaustive
As suggested by the reviewer, we also calculated the Bi2O3 structure with the Oterminated bottom slab (please also refer to Comment 3). The structure is given in Fig.   S16, and the electron-transfer result is given in Fig. S17. One can find from the figure that the charge-transfer effect of the new Bi2O3-Pd8 cluster structure follows the same trend as we proposed in the original manuscript (see Fig. S11), i.e., the Bi2O3 transfers electrons to Pd. In addition, we further compared the stability between the stoichiometric Bi2O3 and Oxygen-lean Bi2O3 under reducing conditions. The transition between the stoichiometric Bi2O3 (BixO1.5x) to oxygen-lean Bi2O3 (BixOy) structure can be regarded as the reaction: BixO1.5x + (1.5x − y)H2 → BixOy + (1.5x − y)H2O And the free energy free of the reaction can be obtained by: Using the above methods, we found it is spontaneous to reduce the O-terminated layer of stoichiometric Bi2O3(100) surface with H2. The result demonstrates the stability of the Bi-terminated Bi2O3(100) surface under reducing conditions.
As suggested by the reviewer, we added more DFT calculations and discussion into the revised supporting information.
10) The literature support for highly reduced oxide surfaces promoting the reduction of Author response: We sincerely thank the reviewer for valuable comments. It helps us to understand the origin of Pd-Bi bonding in Pd1.0/Bi2O3/TiO2. As suggested by the reviewer, we added more experimental evidence and DFT calculations to support the direct Pd-Bi bonding in Pd1.0/Bi2O3/TiO2. Please also refer to Comments 3, 7, and 9. In this study, the direct Pd-Bi bonding likely results from the unique synthesis procedures.
Because the mild oxidation of Bi to Bi2O3 occurred after the deposition of Pd onto Bi/TiO2, the Pd-Bi bond preserves.

11) Calling the catalyst of focus a "model catalyst" is quite misleading. It is quite ill-
defined and the source of its catalytic activity is still unclear.
Author response: As suggested by the reviewer, we have changed the name of the catalysts in the revised manuscript. The main catalyst is named Pd1.0/Bi2O3/TiO2 to illustrate that Pd is in contact with Bi2O3 to form hybrid clusters. The source of its catalytic activity is now clearly related to the Pd-Bi2O3 hybrid clusters.

12) The authors need to show the catalytic performance of the PdBi IMC catalyst for comparison.
Author response: As suggested by the reviewer, we showed the catalytic performance of the PdBi IMC catalyst in the revised supporting information and mentioned it in the revised manuscript as follows.
PdBi/TiO2 composed of PdBi IMC (Fig. S9) shows 100% conversion of C2H2 and negative selectivity towards ethylene (−1319%) at room temperature. The strong exothermic effect of unselective acetylene hydrogenation eventually leads to a runaway temperature up to 58.5 °C. These results exclude PdBi IMC as the active site for Pd1.0/Bi2O3/TiO2 and further indicates the critical role of the nanometer Pd/Bi2O3 interface in the catalytic selectivity of Pd.
13) The effect of Bi could be a simple site blocking mechanism. This seems to be an effect that is completely avoided in the discussion of the catalyst. Was this effect directly investigated? It could be by simply producing a Pd/TiO2 catalyst and then adding Bi to the catalyst with an appropriate reduction pre-treatment procedure afterwards.
Author response: It is important to note that in the synthesis of Pd1.0/Bi2O3/TiO2, Bi was deposited onto TiO2 before the deposition of Pd. The reduction pretreatment temperature is also too low to (only 100 °C) drive the migration of Bi2O3 onto Pd. catalysts. This will be our next project.
As suggested by the reviewer, the following discussion was added into the revised manuscript.
These results suggest that a simple site blocking mechanism cannot explain the beneficial effect of Bi in this study.
14) How is Bi oxidized to Bi2O3 in the synthesis method? There appears to be no elevated temperature oxidation or reduction pretreatment applied after initial catalyst synthesis.
Author response: In this study, Bi was mildly oxidized to Bi2O3 when the sample was exposed to the air. As suggested by the reviewer, we revised Fig. 1a and added more discussion to illustrate the oxidation of Bi as follows.
In brief, Bi 3+ was reduced by the photogenerated electrons to produce Bi 0 clusters on TiO2. Subsequently, Pd was deposited onto Bi/TiO2 to form Pd/Bi/TiO2. When the sample was exposed to the air, Bi was spontaneously oxidized into Bi2O3 because the Gibbs free energy of the reaction is minus. The procedure was schematically illustrated in the Fig. 1a. Figure 1a. Schematic illustration of the synthetic procedures.

15) No information is provided for sample handling before XPS, XRD, or XAS
measurements. This information must be provided so the reader can predict the chemical state of the catalyst and how ambient may have affected composition.
Author response: As suggested by the reviewer, we provided the information for sample handling before XPS, XRD and XAS measurements as follows.
The catalyst was activated by H2 at 100 °C for 1 h and then cooled to room temperature in N2 prior to the catalytic reaction and characterizations.
16) At a minimum, the DFT model should be motivated significantly and the critical feature of surface termination be discussed directly. This is a vital feature of the model system and the electron density transfer to the metal. Maybe even add the oxygens to the bottom of the slab and show that the charge transfer still occurs.  S9b). More importantly, a massive Bi 0 peak at 156.6/162.0 eV (Fig. S9d) is observed in the Bi4f of PdBi/TiO2-IMC but is absence in that of Pd1.0/Bi2O3/TiO2 (Fig. S8b).
We have tried to identify the orientation of Pd bonding with Bi2O3 clusters by ADF-STEM ( Fig. 1d and S1c). However, due to the ultra-small size and large strain of the clusters, most Pd clusters are identified to bond and hybridize with Bi2O3 clusters without any fixed orientation relationship or facet preference. We chose the (100) surface in the calculations because of several reasons. Firstly, (100) surface is a lowindex surface that normally has large proportions among the overall crystal surfaces, which makes this surface representative. Secondly, the Bi2O3(100) surface has a layered structure, and the Bi-terminated surface can be exposed under reducing conditions. We have also demonstrated the stability of Bi-terminated Bi2O3(100) surface with thermodynamic analyses regarding surface reduction. See the reply to comment 9) of referee 2. Thirdly, we found that only such Bi-terminated Bi2O3(100) surface can give rise to consistent Pd-Bi distances with those determined experimentally. In addition, we believe that the charge transfer from Bi2O3 to Pd is the key of the high reactivity of Pd/Bi2O3 catalysts, and the negatively charged Pd atoms on Bi2O3 behaves differently with respect to Pd substance. We have shown in computations that the acetylene hydrogenation reaction takes place on Pd sites rather than at Bi2O3 site, and hence the reactivity behavior of Pd-cluster during acetylene hydrogenations is possibly independent with surface orientations. We have added discussion in the manuscript to clarify this point.
As suggested by the reviewer, we have added more discussion into the revised manuscript and supporting information to illustrate the structure of Bi2O3 and the choice of (100) orientation.
Interestingly, close inspection of individual Bi cluster on the TiO2 support by using aberration-corrected annular dark field scanning transmission electron microscopy (ADF-STEM) indicates that the Bi cluster has an ordered α-Bi2O3 structure with a highly distorted lattice, which exhibits relatively weak diffuse peaks in the FFT (Fig.   1b). This is consistent with the literature result, which indicates that the mild oxidation of Bi 0 in the air is thermodynamically spontaneous and usually forms monoclinic α- Author response: As suggested by the reviewer, we built the model of Pd1.0Bi/TiO2-IMC based on the crystal structure determined by XRD (i.e., Sobolevskite, PDF#29-0238). The space group of the hexagonal PdBi-IMC is P63/mmc(194) and the lattice parameters are a = b = 4.22 Å, c = 5.709 Å, α = β = 90°, and γ=120 °. Upon DFT optimization, the Pd-Bi distance is calculated to be 2.91 Å (Table S3) As suggested by the reviewer, we added the following discussion into the revised manuscript.
The Pd-Bi pair distribution function of the Pd8 cluster structure is shown in Fig.   S12. In this figure the dominant peak appears at ~2.75 Å, which is smaller than that in the PdBi intermetallic model (~ 2.91 Å, Table S3). In addition, Bader charge analysis suggests that Pd atoms in PdBi IMC model are more electron-rich (average charge of −0.36 e) than those in Pd-Bi2O3 hybrid clusters model (−0.21 e). These results are consistent with Pd K-edge EXAFS and Pd3d XPS shown in Fig. S9, which therefore validate the reliability of the Pd8-cluster model. Fig. 4b  Author response: As suggested by the reviewer, we compared the barriers of acetylene hydrogenation to ethane on Pd(111) that were reported by different groups and obtained with different transition state searching methods. As shown in Table S5, the calculated barriers in the current work are generally consistent with reported values, demonstrating that our calculated results are reliable. In addition, we further calculated the vibrational frequency based on the transition state structures on Pd(111). All the transition states were characterized to possess only one imaginary frequency along the bond formation of hydrogenation reactions. These results further demonstrate the reliability of the constrained minimization method that we used in this work.

4.
In addition to revised Table S5, we also added the following discussion into the manuscript.
The calculated barriers are generally consistent with reported values ( Table S5), demonstrating that our calculated results are reliable. In addition, we further calculated the vibrational frequency based on the transition state structures on Pd(111). All the transition states were characterized to possess only one imaginary frequency along the bond formation of hydrogenation reactions. These results further demonstrate the reliability of the constrained minimization method that we used in this work.  , 2013, , 264-276 b: Science, 320, 5881, 2008, , 1320, -1322