Single-atom catalysts reveal the dinuclear characteristic of active sites in NO selective reduction with NH3

High-performance catalysts are extremely required for controlling NO emission via selective catalytic reduction (SCR), and to acquire a common structural feature of catalytic sites is one key prerequisite for developing such catalysts. We design a single-atom catalyst system and achieve a generic characteristic of highly active SCR catalytic sites. A single-atom Mo1/Fe2O3 catalyst is developed by anchoring single acidic Mo ions on (001) surfaces of reducible α-Fe2O3, and the individual Mo ion and one neighboring Fe ion are thus constructed as one dinuclear site. As the number of the dinuclear sites increases, SCR rates increase linearly but the apparent activation energy remains almost unchanged, evidencing the identity of the dinuclear active sites. We further design W1/Fe2O3 and Fe1/WO3 and find that tuning acid or/and redox properties of dinuclear sites can alter SCR rates. Therefore, this work provides a design strategy for developing improved SCR catalysts via optimizing acid-redox properties of dinuclear sites.


I. Reply to the reviewers
We thank the editor and the reviewers for carefully reviewing our manuscript. We have revised the manuscript carefully, according to the reviewers' comments. Below is a point-by-point response to the reviewers' comments.

Reviewer #1
This paper reported on single-site catalysts of Mo 1 /Fe 2 O 3 , W 1 /Fe 2 O 3 , and Fe 1 /WO 3 for NH 3 -SCR. Basically, the manuscript was written well, but there are considerable issues in the present form. I recommend publishing after a major revision. I strongly request the below to improve this manuscript. Figure S4).

The authors described that 'Blue shade represents a decrease of the surface defect oxygen after the Mo loading' in Supporting Information. However, this explanation is
confusing. XPS can reflect the state of oxygen on the catalyst. The defect oxygen will not be observed directly by XPS. If the words of 'a decrease of the surface defect oxygen' means that oxygen occupied the lattice, the intensity should become higher for Mo 1 /Fe 2 O 3 . Reply 1.1: Thank the reviewer for the good comments. We agreed that the explanation about the O 1s XPS is confusing, which possibly originates from the different definitions of defect oxygen. Some researchers defined the oxygen vacancy as defect oxygen (ACS Appl. Mater. Interfaces, 2014, 6, 12505−12514;ACS Appl. Mater. Interfaces, 2016, 8, 33765-33774), while others defined surface oxygen with the low coordination number as the defect oxygen (Phys. Chem. Chem. Phys., 2000, 2, 1319-1324Electrochim. Acta, 2017, 229, 229-238;ACS Sustainable Chem. Eng., 2019, 7, 12117−12124). We agreed with the reviewer that it is more suitable to define oxygen vacancy as the defect oxygen, which thus cannot be observed directly by XPS.
As a result, we modified the manuscript as "As further analyzed by the selected-area intensity surface plot and the corresponding structural model ( Figures 1H and S3)" and deleted the Figure S4 in the revised Supporting Information. Figure 1F. Figure 1F is not meaningful. In addition, the color of Mo and O was similar. In this case, the overlap of Fe and Mo is enough to show the dispersion of Fe species as Figure 1F.

The overlap of O in
Reply 1.2: According to the reviewer's suggestion, we have deleted the overlap of O in Figure 1F, and thus the dispersions of Mo and Fe species are clearly displayed in the revised Figure 1F. to 'single-atom' in the revised manuscript. As a consequence, the 'dinuclear Mo 1 -Fe 1 site' is used as the catalytically active site in the SCR process, which is consistent with the recent research (Angew. Chem. Inter. Ed., 2019, 58, 12609-12616 Figure S7) with the Mo L 3 -edge X-ray absorption spectra ( Figure 2B). Although the oxidation states of Mo species are associated with the Mo=O bonding model, the oxidation states of Mo in MoO 6 motif with Mo=O bonds might be +5 or +6 (Appl. Surf. Sci., 1989, 40, 179-181;Appl. Catal. B, 1998, 3, 245-258;Science, 2015, 348, 686-690).
Generally, MoO 6 motif is the octahedral structure in the α-MoO 3 crystal, as confirmed by the Mo L 3 -edge X-ray absorption spectra of α-MoO 3 in Figure 2B.
Meanwhile, the FeO 6 motif in the α-Fe 2 O 3 crystal is also the octahedral structure  (001) surface, though we did not know the accurate geometric structure.
According to the reviewer's suggestion, we have added the in situ DRIFT spectra collected at the SCR reaction temperature of 250 o C into the revised Supporting Information. As shown in Figure S10, the Lewis acid sites can be observed during the SCR reaction process. Accordingly, the manuscript has been revised as "Thus, each isolated Mo ion due to the formation of a MoO 6 H species can provide one Brønsted acid site 23 , which can transform to the Lewis acidic site during the SCR reactions 24 ( Figure S10)."  Reply 1.6: As the reviewer suggested, we have recorded the NH 3 and NO conversions of Mo 1 /Fe 2 O 3 together with -Fe 2 O 3 in the SCR process. As shown in Figure R2A, the conversions of NH 3 and NO over Mo 1 /Fe 2 O 3 are almost the same as each other, in the whole temperature range. We further recorded the concentrations of N 2 O ( Figure   R2B) and the concentrations of NO 2 can be ignored, from which we calculated the N 2 selectivity as shown in Figure R2C. The results demonstrated that Mo 1 /Fe 2 O 3 has high catalytic activity and selectivity. In contrast, for -Fe 2 O 3 , the conversion of NO decreased rapidly and became lower than that of NH 3 above 300 o C, as shown in Figure R2D, which suggested that the NH 3 was oxidized to NO by -Fe 2 O 3 at high temperatures. Meanwhile, as the temperature increases, N 2 O concentration over -Fe 2 O 3 increases ( Figure R2B), leading to the low selectivity to N 2 ( Figure R2C).
We have added these results into the revised Supporting Information as Figure S12.   Figure R3), which were also added in the revised Supporting Information as Figure S24. Figure R3.  Figure S13, and we revised the manuscript as: "and the Mo 1 /Fe 2 O 3 also showed excellent H 2 O and/or SO 2 durability ( Figure S13)".  improved. For example, between pg 3 and 4 the use of "plenty of" should be modified.
Reply 2.1: Thank the reviewer for the good comment. We have replaced "plenty of" by "abundant" in the revised manuscript.

Comment 2.2:
The STEM data in Figure 4D is not as convincingly assigned as was seen for Figure 1g. Specifically, it seems like many more W species may exist in neighboring sites just based on scattering intensity. This should be clarified with some discussion.

Reply 2.2:
As the reviewer suggested, we have taken the image intensity line scans in the areas, where W species seem to exist in neighboring sites. The intensity profiles revealed that the most W exists as isolated single atoms ( Figure R5). We have added the result into the revised Supporting Information as Figure S19. It might also be the existence of adjacent W species, if so, these adjacent W species should not make a great contribution to our conclusion owing to a low activity in SCR reactions, as testified by the very low conversion on the two adjacent W sites of WO 3 in our work ( Figure S26).   2019,9,12,[10899][10900][10901][10902][10903][10904][10905][10906][10907][10908][10909][10910][10911][10912]. This comment is not meant to take away from the current paper, just to suggest that the authors clarify the structure of the sites described here compared to other recent reports.

Reply 2.3:
As the reviewer mentioned, on the Mo 1 /Fe 2 O 3 or W 1 /Fe 2 O 3 surfaces, each isolated Mo or W ion connects to three neighboring Fe ions bridged by the lattice oxygen ( Figure S2), and thus one active catalytic site (ACS) might contain more than one Fe ions besides one Mo or W ion. To rule out this possibility, we anchored single Fe ions on (001) surfaces of γ-WO 3 nanosheets with a square morphology to achieve a single-atom Fe 1 /WO 3 catalyst ( Figure S25). As the number of the single Fe ions increases, X NO over Fe 1 /WO 3 linearly increases, the trend of which is the same as that of W 1 /Fe 2 O 3 ( Figure 4E). As a consequence, only single Fe ion is enough for one ACS of Mo 1 /Fe 2 O 3 or W 1 /Fe 2 O 3 . According to the reviewer's suggestion, we have also cited the paper (ACS Catal., 2019, 9, 12, 10899-10912) as Reference 17 to clarify the concept of dinuclear metal sites, as described in the revised manuscript: "An ideal structure is to fabricate dinuclear metal sites on supports 17 , which allows it to function as dual sites catalyzing SCR reaction, but it is a formidable task to synthesize such a catalyst."