Steering the reaction pathway of syngas-to-light olefins with coordination unsaturated sites of ZnGaOx spinel

Significant progress has been demonstrated in the development of bifunctional oxide-zeolite catalyst concept to tackle the selectivity challenge in syngas chemistry. Despite general recognition on the importance of defect sites of metal oxides for CO/H2 activation, the actual structure and catalytic roles are far from being well understood. We demonstrate here that syngas conversion can be steered along a highly active and selective pathway towards light olefins via ketene-acetate (acetyl) intermediates by the surface with coordination unsaturated metal species, oxygen vacancies and zinc vacancies over ZnGaOx spinel−SAPO-34 composites. It gives 75.6% light-olefins selectivity and 49.5% CO conversion. By contrast, spinel−SAPO-34 containing only a small amount of oxygen vacancies and zinc vacancies gives only 14.9% light olefins selectivity at 6.6% CO conversion under the same condition. These findings reveal the importance to tailor the structure of metal oxides with coordination unsaturated metal sites/oxygen vacancies in selectivity control within the oxide-zeolite framework for syngas conversion and being anticipated also for CO2 hydrogenation.

and ZnGaOx_F) of this study in the selectivity and activity to methanol. Methanol can subsequently be converted into olefins over  Consequently, the methanol selectivity and activity look as a straightforward explanation of the OXZEO data presented here. Furthermore, the quality of the manuscript could be improved by amending a number of imprecise statements/language, not sufficiently defined methods, considerations later turning into conclusions etc. In general I find that the manuscript does not contain the novelty and quality needed to merit publication in Nature Communication.
Specific comments: Line 32-33: "SAPO-34 catalysis" I would say this is a phrase from a spoken language. Please re-phrase. Line 46-49: Please also report conversion levels. This is very important information, when comparing selectivity of these reactions. Line 51: "ZnGaO-SAPO-34" should it be "ZnGaOx-SAPO-34" Figure 1: Please provide scalebars for all parts of this figure. In the part of the figure text that covers (a,b), it is not stated what material the SEM images show. Line 83-86: The type of samples is not introduced before e.g. "Hydrothermal sample" is mentioned. The paper should contain a small intro paragraph with an introduction of the samples. Figure 2a: I presume that the XPS values represent measurements of the oxidic samples? Line 109-111: "..than the flake one …" represents spoken language and this sentence should be carefully rewritten. "ZnGaOx_NP-B with a surface Zn/Ga molar ratio of 1.2" In Table S1, the Zn/Ga molar ratio of this sample is 0.3. Line 116-117: Even though this study is about using ZnGaOx samples together with SAPO-34 for the OXZEO process, the purpose of this study is to unravel the different reactivities of dissimilar surface sites on the ZnGaOx particles. This calls for a thorough investigation also of the catalytic properties of the ZnGaOx samples themselves and this has only been done for two samples in this work (Table S3). This should be done for all ZnGaOx samples studied here because these studies seem very informative as the products observed are not processed further on the active sites of SAPO-34 and in that way more complexed to analyze and the ZnGaOx differences somewhat blurred. Taking the available activity data for the two tested pure ZnGaOx samples, it seems to me that they explain the data with ZnGaOx mixed with SAPO-34 well and provide an explanation that is different from that put forward by the authors: The ZnGaOx-NP sample is quite selective towards methanol and my rough estimate suggests that methanol is (at least) in equilibrium (please provide information in Table S3 of how close the methanol is to equilibrium), while this is not the case for ZnGaOx-F, which also forms more methane. Since methanol is at equilibrium for the ZnGaOx-NP sample the selectivity towards methanol is underestimated if the data in Table S3 is used to estimate the selectivity to methanol directly. If this sample is mixed with SAPO-34, the methanol is converted to olefins, so the data in Table S3 seems to show that the difference between the ZnGaOx-F and the ZnGaOx-NP samples is just the methanol selectivity and activity. Please comment. Line 119-120: "CO conversion is enhanced remarkably over ZnGaOx_NPSAPO-34 ( Figure S6a), indicating a shifted reaction equilibrium by the tandem catalysis of SAPO-34" This appears to be correct, but the data in Table S3 strongly suggests that it is the methanol, which is (known to be) converted to olefins over  Line 121-123: "By contrast, introducing SAPO-34 to ZnGaOx_F hardly affects the overall conversion (< 8%) and light olefin selectivity < 16% at a wide range of OX/ZEO ratios ( Figure S6b)." This is not surprising considering the above analysis based on methanol as the precursor for olefins. Figure 3: It seems that the sensitivity of the In-situ FTIR is not large enough for the ZnGaOx_F sample to judge whether the formate to acetyl ratio is the same as that for ZnGaOx_NP. Line 134-136: "They (formates) are most likely just spectators rather than active intermediates, because there are also weak formate species observed over ZnGaOx_F, which cannot be effectively converted by subsequent SAPO-34 catalysis to olefins". "Weak formate species observed over ZnGaOx_F, which cannot be effectively converted by subsequent SAPO-34 catalysis to olefins" seems to be correct because ZnGaOx_F is not effective in converting CO/H2 to methanol due to a low number of formates at the surface of ZnGaOx_F catalyst, but this does not make the formates spectators. To me formates are most likely a very important part of this catalyst system. Line 140-146: The data are well explained by the selectivity to methanol, which have been demonstrated to be formed over both ZnGaOx_F and ZnGaOx_NP so an additional ketene mechanism is not needed to explain the observations. Line 146-150: "The above results demonstrate that formation of ketene-acetate (acetyl) or formate species over ZnGaOx is critical for the reaction pathway towards final products, ..." This statement seems to be build upon two observations: (1) the yield over ZnGaOx/SAPO-34 mixtures and (2) some IR data where the sensitivity does not seem large enough for the ZnGaOx_F sample. Even if it could be demonstrated that no acetyl was observed at the surface of the ZnGaOx_F sample then the methanol yields in my opinion form a straightforward explanation strongly underpinned by literature. Figure 4: Figure text. If the CO2 signal up to 400°C is potted in Figure 4b then these numbers should be available in Table S4, where the integrated CO2 signals in other temperature ranges are stated. Why was H2 TPR not used instead of CO TPR? Line 166-167: Integrated CO2 signals below 400°C are apparently not given in Table S4 Figure S7: The fitting procedures should be described. How was the background determined for example? Line 174-175: Please indicate the position of the Mn impurity in Figure 4d. Line 209-211: "This is further validated by AP-XPS analysis, which reveals a decreased O/Ga molar ratio of ZnGaOx-NP (the inset of Figure 5a)." The oxygen is bonded to both Zn and Ga and since Zn/Ga is reduced then the O/Ga should also be reduced. If this is so, then a lowering of the O/Ga ratio appears inconclusive. Line 228: Fitting procedure including line shape should be given. It looks as if the fitting of the ZnGaOx_NP-A sample and the ZnGaOx_F samples in Figure S9(d,e) are not well-defined and that one could in another fit to the data in Figure S9(e) obtain more medium strength acid sites for the ZnGaOx-F sample. Table S2: What are the methanol and the HC selectivities compared to each other? Please include these in the table.
The bifunctional catalysis represented by OXZEO pioneered by the authors of this manuscript has made great breakthrough towards the direct conversion of CO to bulky feedstocks like ethylene and aromatics with high selectivity. Considerable research interests were previously devoted to tailoring the structures of oxide and zeolite components and the assemble manner. In this manuscript, the authors systematically compared and analyzed two different ZnGaOx spinel oxides and tested their direct syngas conversion to light olefines coupling with SAPO-34 zeolite component. They highlighted the importance of coordination unsaturated Ga3+ species and oxygen vacancies in oxide composite and achieved the steering of reaction pathway for syngas conversion. It is highly impressive that a high CO conversion (49.5%) is achieved over such bifunctional systems while keeping high selectivity to light olefins. The structures of both ZnGaOx spinel oxides were comprehensively characterized and the manuscript was well written. I recommend the acceptance of this work after the following issues are appropriately addressed.
1) Both hydrothermal and coprecipitation methods were employed to synthesize ZnGaOx samples with quite different structures and catalytic performance. Why the preparation methods lead to such striking difference?
2) From Table S3, it seems that both methanol and C2-C4 hydrocarbons could be produced by individual ZnGaOx nanoparticles. What is the reaction pathway for the production of these light hydrocarbons? Can the methanol-mediated pathway be completely excluded for the direct syngas conversion in bifunctional systems combing such ZnGaOx-NP and SAPO-34 composites? The authors proposed that the reaction occurs via the ketene-acetate (acetyl) intermediates by the coordination unsaturated Ga3+ and species and oxygen vacancies of ZnGaOx-NP/SAPO-34 systems.
3) The authors demonstrated that medium Lewis acid sites differentiate over two kinds of ZnGaOx oxides and analyzed by NH3-TPD approach. While both Bronsted and Lewis acid sites can be characterized by such NH3-TPD approach. Additional characterization methods may be required to analyze the distribution of surface acid sites. 4) How about the stability of the catalyst samples?

Reviewer #1 (Remarks to the Author):
Comment: Tandem catalysis is a key aspect in C1 chemistry and catalysis, for the conversion of CO,CO2,methanol,methane,etc

Response:
We thank the reviewer for the insightful comment. It is indeed important to avoid the limitation at the activity of the second step of zeolite catalysis while studying the catalysis over metal oxides. Therefore, we carried out experiments by varying the zeolite content of the bifunctional OXZEO catalysts from 25 wt.% to 67 wt.%. The data were added in Fig. S8 of the revised Supplementary Information. The space time yield (STY) of light olefins for ZnGaOx_F-SAPO-34 composites did not vary with the zeolite content, which indicates that the catalytic activity of ZnGaOx_F was intrinsically low and there were not enough intermediates generated over ZnGaOx_F. Therefore, the reaction equilibrium was not affected in the presence of much SAPO-34. There is not limitation at the activity of the second step of zeolite catalysis for ZnGaOx_F under our reaction conditions. By contrast, the yield of light olefins increases with the zeolite content in the bifunctional catalyst ZnGaOx_NP-SAPO-34 and the one containing 50 wt.% zeolite gave a highest STY of light olefins. Therefore, there is also no limitation at the activity of the second step for ZnGaOx_NP under our reaction conditions.

Action taken:
We have added Fig. S8c and the above discussions in the revised Supplementary Information.

Action taken:
We have added Fig. S6 and related discussion in the revised Supplementary Information.

Response:
We have carefully checked all references and updated the format.

Reviewer #2 (Remarks to the Author):
Comment: "Steering the reaction pathway of syngas-to-light olefins with coordination unsaturated sites of ZnGaOx spinel" by Na Li, Yifeng Zhu, Feng Jiao, Xiulian Pan, Qike Jiang, Jun Cai, Yifan Li, Wei Tong, Changqi Xu, Shengcheng Qu, Bing Bai, Dengyun Miao, Zhi Liu, and  Response: We argue that this manuscript contains the novelty and quality, which merit publication in Nature Communications, from the following point of view.
First of all, this ZnGaOx_NP gives a highest activity among the Cr-free metal oxides for syngas-to-light olefins. There are a number of metal oxides within the framework of OXZEO reported for syngas-to-light olefins. The highest conversion is obtained over Cr containing metal oxides (~50%) at an olefin selectivity of ~70-80%.
Due to the concerns of the toxicity of Cr, there are wide efforts seeking highly active Cr-free metal oxides. In this work, ZnGaOx_NP-SAPO-34 gives olefin selectivity as high as 75% at 49.5% CO conversion. This represents the highest yield among reported Crfree metal oxide in the literature (Table R1). This is also acknowledged by the reviewer #3.  This is made clearer in the caption of Fig. 2a and Table S1.
Comment: Line 109-111: "..than the flake one …" represents spoken language and this sentence should be carefully rewritten. "ZnGaOx_NP-B with a surface Zn/Ga molar ratio of 1.2" In Table S1, the Zn/Ga molar ratio of this sample is 0.3.

Response:
We thank the reviewer for this comment. We have rephrased the sentence: "all nanoparticle ZnGaOx_NP samples demonstrate much superior performance than ZnGaOx_F upon being physically mixed with SAPO-34 separately".
We apologize for the mistake. It should be ZnGaOx_NP with a Zn/Ga molar ratio of 1.2 instead of ZnGaOx_NP-B. This is corrected in the revised manuscript.  Figure S6b)." This is not surprising considering the above analysis based on methanol as the precursor for olefins.

Response:
As suggested by the reviewer, we have tested the catalytic performance of all ZnGaOx oxides. Considering the low activity of ZnGaOx oxides alone as the catalyst for syngas conversion, GC data were analyzed using carbon normalization method.
Furthermore, to get more reliable data with standard deviation values (errors), we synthesized and tested 3-5 batches for each ZnGaOx catalyst. The data are displayed in  is a classical catalyst for methanol-to-olefins.

Action taken:
Fig. S10 and the corresponding discussion are added in the revised manuscript.  (Fig. S9). It shows CO conversion correlates much better with the intensity of acetate species (Fig. 3b), whereas it does not correlate well with that of formate (Fig. S9b), indicating the crucial role of surface acetate species in syngas conversion. Since acetate species are essentially the adsorbed ketene, acetate/ketene could be active intermediates leading to olefins on ZnGaOx_NP-SAPO-34 composites.

Fig. 3. Surface intermediates over ZnGaOx oxides and the relationship with catalytic
performance. a In-situ FT-IR differential spectra of syngas conversion over H2-reduced ZnGaOx_NP (navy line) and ZnGaOx_F (brown line) at 400 ºC. b CO conversion as a function of acetate intensity at 1525 cm -1 of FT-IR spectra of different ZnGaOx samples in Fig. S9a.   Fig. S9. In-situ FT-IR differential spectra of syngas conversion over H2-reduced ZnGaOx samples. a Samples of ZnGaOx_NP, ZnGaOx_NP600, ZnGaOx_NP-A, and ZnGaOx_F. b Relationship between formate intensity around 1589 cm -1 and CO conversion.

Action taken:
We have added the catalytic performance of metal oxides and in-situ IR results in Fig.   3b, S9, and S10. The corresponding discussion has been added on pages 8 and 9 of the revised manuscript. Figure 4: Figure text. If the CO2 signal up to 400°C is potted in Figure 4b then these numbers should be available in Table S4, where the integrated CO2 signals in other temperature ranges are stated.

Response and action taken:
Thanks for this suggestion. We have revised Table S4 (Table S5 in the revised version) as suggested.

Comment: Why was H2 TPR not used instead of CO TPR?
Response: It was shown in our previous study that reduction of ZnCr2O4 (111) surface by CO is much more energy favored than by H2 (Science, 2016, 351, 1065). Therefore, we used CO-TPR instead of H2-TPR since our purpose was only to compare the reducibility between different oxides.
Nevertheless, we carried out H2-TPR experiments as suggested by the reviewer.
The data are added in Fig. S12. An obvious H2 consumption signal (m/z = 2) is observed for ZnGaOx_NP below the reaction temperature of 400 o C, in contrast to a much less H2 consumption for ZnGaOx_F, which only takes place at above 400 ºC. Therefore, H2-TPR also reveals a much higher reducibility of ZnGaOx_NP than that of ZnGaOx_F below 400 o C, consistent with CO-TPR results in Fig. 4a.

Action taken:
H2-TPR profiles are added in Fig. S12 of the revised Supplementary Information.

Comment:
Line 166-167: Integrated CO2 signals below 400°C are apparently not given in Table S4.

Response and action taken:
We have revised Table S4 (Table S5 of the revised version) by adding the integrated CO2 signals below 400 ºC.
Comment: Figure S7: The fitting procedures should be described. How was the background determined for example?

Response and action taken:
Shirley type background is used here for all samples. The detailed fitting parameters have been added in Table S4.

Response and action taken:
As suggested, we have indicated the signal of Mn 2+ impurity (six blue vertical lines) in   Figure 5a)." The oxygen is bonded to both Zn and Ga and since Zn/Ga is reduced then the O/Ga should also be reduced. If this is so, then a lowering of the O/Ga ratio appears inconclusive.

Response and action taken:
As displayed in Scheme R1 below, just removing Zn atom but not the adjacent O will result in a decreased Zn/Ga ratio but not O/Ga  Figure   S9(d,e) are not well-defined and that one could in another fit to the data in Figure S9(e) obtain more medium strength acid sites for the ZnGaOx-F sample.

Response and action taken:
As requested, we have added the fitting procedure in Table S7. The full width at half maximum, i.e., half-width (FWHM) and other parameters have been carefully controlled to ensure a reliable fitting and they are consistent for all ZnGaOx samples.
In addition, we showed fitting of all NH3-TPD profiles, especially for Fig. S9d and S9e (Figs. S16d and S16e in the revised version). Although there is some error, the trend is clear, consolidating the conclusion of this work. Taking ZnGaOx_F as an example, whatever the fitting parameters change, the number of medium strength acids remains lower than 0.01. Therefore, we believe that the result given in Fig. 5d is comparable and acceptable. We have updated the fitting results in Figs. S16, 5d and Tables S7, S8 of the revised Manuscript and Supplementary Information.  Table S8 lists the detailed integration parameters.

Response and action taken:
As requested, we have added the methanol selectivity in Table S2. Since HC selectivities can be obtained by deducing that of CO2 and methanol, we did not add HC selectivities but gave hydrocarbon distribution in Table S2. Response: We thank the review for the thoughtful comment. The reaction pathway for formation of C2-C4 hydrocarbons on individual ZnGaOx nanoparticles was also suggested in our recent quasi-in-situ solid state NMR DFT study (Catal. Sci. Technol., 2022, 12, 1289-1295. Surface C1 species could react with CO to form ketene. In the absence of zeotypes, ketene was converted to thermodynamically stable acetate species and further to C2-C4 hydrocarbons, which was proved by NMR in previous work and also in-situ FT-IR in this work ( Figure 3). Note that acetyl species are not a commonly However, they are clearly observed on ZnGaOx-NP catalysts by both ssNMR and in-situ FT-IR, indicating C-C coupling activity of ZnGaOx-NP. Furthermore, Fig. S9 shows that CO conversion of ZnGaOx-SAPO-34 bifunctional catalysts is well correlated with the IR intensity of surface acetate signal around 1525 cm -1 (Fig. S9, Fig. 3b). Therefore, surface acetate/ketene may be more active intermediates over these oxides.

Reviewer #3 (Remarks to the Author):
To elucidate the role of methanol for olefin formation over ZnGaOx_NP-SAPO-34, we further conducted the catalytic performance test for all ZnGaOx oxides without SAPO-34. Fig. S10 indicates that the methanol concentration over sole metal oxides does not correlate well with the corresponding hydrocarbon concentration in syngas conversion over the corresponding OXZEO catalysts. Therefore, methanol may not be a key intermediate for olefin synthesis over ZnGaOx-SAPO-34. However, considering that SAPO-34 is a classical catalyst for methanol-to-olefins and there is also methanol formation over these oxides, we think that methanol contribution cannot be excluded completely over the corresponding bifunctional OXZEO catalysts.

Action taken:
We have added metal oxides performance and in-situ IR results in Figs. 3b, S9 and S10.
The related discussion has been added on pages 8 and 9 of the revised Manuscript and Supplementary Information.

Comment: 3) The authors demonstrated that medium Lewis acid sites differentiate over two kinds of ZnGaOx oxides and analyzed by NH3-TPD approach. While both Bronsted
and Lewis acid sites can be characterized by such NH3-TPD approach. Additional characterization methods may be required to analyze the distribution of surface acid sites.
Response: Since pyridine (Py) is widely used as a probe molecule to distinguish the Brønsted and Lewis acid sites, we performed in-situ IR with Py adsorption on two kinds of ZnGaOx oxides (ZnGaOx_NP and ZnGaOx_F). As displayed in Fig. S15, upon exposing ZnGaOx_NP to Py (navy colored line), two bands assigned to Lewis acid sites appear at ~1450 and ~1605 cm -1 . However, no signal at ~1540 cm -1 is observed, where the Brønsted acid sites generally respond. Therefore, there are only Lewis acid sites on ZnGaOx_NP and it is reasonable to assign these NH3-TPD signals to the Lewis acid sites.
By contrast, the Lewis acid signal intensities are extremely low and also no Brønsted acid sites are observed for ZnGaOx_F (wine line), which is consistent with the few medium strength Lewis acid sites reflected by NH3-TPD in Fig. S11e.   Fig. S15. In-situ FT-IR differential spectra of pyridine adsorption over ZnGaOx_NP (navy line) and ZnGaOx_F (wine line).