Single-atom Sn-Zn pairs in CuO catalyst promote dimethyldichlorosilane synthesis

Abstract Single-atom catalysts are of great interest because they can maximize the atom-utilization efficiency and generate unique catalytic properties; however, much attention has been paid to single-site active components, rarely to catalyst promoters. Promoters can significantly affect the activity and selectivity of a catalyst, even at their low concentrations in catalysts. In this work, we designed and synthesized CuO catalysts with atomically dispersed co-promoters of Sn and Zn. When used as the catalyst in the Rochow reaction for the synthesis of dimethyldichlorosilane, this catalyst exhibited much-enhanced activity, selectivity and stability compared with the conventional CuO catalysts with promoters in the form of nanoparticles. Density functional theory calculations demonstrate that single-atomic Sn substitution in the CuO surface can enrich surface Cu vacancies and promote dispersion of Zn to its atomic levels. Sn and Zn single sites as the co-promoters cooperatively generate electronic interaction with the CuO support, which further facilitates the adsorption of the reactant molecules on the surface, thereby leading to the superior catalytic performance.


INTRODUCTION
As a result of their maximum atom utilization and unique electronic properties, single-atom catalysts have shown superior catalytic properties in a wide variety of reactions compared to conventional nanoparticle catalysts [1][2][3][4][5][6][7][8][9]. Thus they have received increasing research interest in recent years. At present, the stabilization of these single atoms under harsh reaction conditions, such as elevated temperatures and pressures [10][11][12], is still the main concern but can potentially be well-addressed by making use of uniform defects of underlying supports as anchoring sites [13]. The defects, like dopants and atom vacancies, also have the potential to alter the coordination environment and charge distribution on the surface [14], thus further improving the catalytic activity. To our knowledge, in recent years, the primary attention has been paid to the surface oxygen vacancies on supports [13,15,16], while the research activ-ities on how to make use of cation vacancies to ensure stable anchoring are still not enough.
On the other hand, promoters, which can further enhance the catalytic performances of many catalysts and are therefore of great importance in catalysis [17][18][19], are rarely studied in their singleatom forms for catalytic reactions. Therefore, preparing catalysts with single-sited promoters that possess similar advantages to single-atom catalysts, such as the structural simplicity and homogeneity [20], should be of great interest. These single-sited promoters may not only help to elucidate their real promotion mechanism in catalytic reactions, but also open up a new path to optimize catalyst performance. For instance, Wang et al. [21] reported that incorporating single-site Sn on TiO 2 as the promoter could create more oxygen vacancies on its surface, leading to the improved catalytic activity and selectivity in nitroarene hydrogenation. Very recently, it has been demonstrated that the doping of CeO 2 with single-atom Ni as the promoter is an effective means to generate oxygen vacancies, which promote the selective hydrogenation of acetylene to ethylene [22]. Because two or more kinds of promoters are often used in one industrial catalyst [17,23], the exploration of the preparation of two single-site promoters and the synergistic effect between them on catalytic reaction is of great interest in catalysis. However, due to the difficulty in the synthesis, there has been no such report so far.
Here we report the synthesis of a new catalyst consisting of atomically dispersed Sn and Zn copromoters on the CuO surface (denoted as Zn 1 -Sn 1 /CuO, where '1' represents the single atom, and the same is applied hereafter). Direct experimental evidence shows that single-site Sn is incorporated into the lattice of CuO catalysts to generate Cu 2+ vacancy sites, which further serve as anchoring sites to stabilize single-site Zn. Density functional the-50 nm

Synthesis and characterization of Zn 1 -Sn 1 /CuO catalyst
CuO and Sn 1 /CuO were synthesized via a facile hydrothermal treatment according to the previously reported method with some modifications [24]. Both the samples exhibit morphologies of nanosheets with an average thickness of about 600 nm (Fig. S1 in the online supplementary material). X-ray diffraction (XRD) patterns ( Fig.  S2 in the online supplementary material) show that, compared with those of pure CuO, the diffraction peaks of Sn 1 /CuO slightly shift to higher angles. Furthermore, no peaks of Sn species are detected in Sn 1 /CuO, indicating that Sn has been doped into the lattice of CuO and is highly dispersed [25]. This phenomenon is because the isomorphous substitution of Cu ions with Sn atoms having a smaller ionic radius would lead to a lattice contraction. It is further verified by a high-resolution transmission electron microscopy (HRTEM) image (Figs S3 and S4 in the online supplementary material), which does not show any lattice corresponding to Sn species [26], but does show the formation of numerous holes on the surface of Sn 1 /CuO (marked by the red circles). Furthermore, it is found that the brighter spots in the image (Fig. 1A) represent Sn atoms, which are identified by using Z contrast in aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC HAADF-STEM) as the atomic numbers of Sn and Cu are Z = 50 and 29, respectively, which are noticeably different [27]. Subsequently, a simple wet impregnation method [28] was employed to anchor the single-site Zn on defective-rich Sn 1 /CuO to form xZn 1 -Sn 1 /CuO (x refers to weight ratios of Zn to CuO; detailed information of the synthesis is shown in the online supplementary material).

RESEARCH ARTICLE
As shown in Fig. S5 Fig. S10) and inductively coupled plasma optical emission spectrometry (ICP-OES) analysis are basically consistent with the feeding ratios of the metal precursors in the synthesis ( Table  S1 in the online supplementary material) [30]. The coordination states and local structure of Sn and Cu were investigated by X-ray absorption spectroscopy (XAS) analysis. It should be noted that owing to the extremely low amount of Zn relative to Cu, the signal of Zn is shielded by that of Cu, and thus it is hard to obtain the structure information of Zn. As shown in Fig. 2A, the Cu K-edge Xray absorption near edge structure (XANES) spectra suggest that there is no noticeable change in the valence state of Cu in Sn 1 /CuO and 0.1Zn 1 -Sn 1 /CuO as compared with that of CuO. The Fourier transform extended X-ray absorption fine structure (FT-EXAFS) curves ( Fig. 2B) show that all the samples of CuO, Sn 1 /CuO and 0.1Zn 1 -Sn 1 /CuO exhibit one main peak at 1.56Å, which is ascribed to the contribution of the first shell of Cu-O. However, compared with that in CuO, the peak at about 2.50Å attributed to the second shell of Cu-O contribution is obviously enhanced in both Sn 1 /CuO and 0.1Zn 1 -Sn 1 /CuO, suggesting the change of the local environment of Cu in these two samples. As shown in  Table S2 in the online supplementary material, the best-fitting result is that the isolated Sn atoms are coordinated with four O atoms with the mean bonding length of 1.95Å. Further, X-ray photoelectron spectroscopy (XPS) measurements ( Fig. S13 in the online supplementary material) show that the binding energy of the Cu 2p 3/2 peak in CuO and Sn 1 /CuO is located at 933.48 eV, which corresponds to Cu 2+ in CuO [31], in good agreement with the above XAS results.
After incorporation with Zn atoms, the binding energy of Cu 2p 3/2 peak shifts to the lower-energy side in comparison with that of CuO, and this shift is observed obviously in 0.1Zn 1 -Sn 1 /CuO, indicating an increase of the electron density on the Cu atoms with the coexistence of Sn and Zn atoms [32,33]. It evidences that there is an interaction between Sn and Zn atoms, and when the Zn content is 0.1 wt% relative to CuO, the interaction is the strongest. In spite of that, the spectra of Sn 1 /CuO and xZn 1 -Sn 1 /CuO exhibit Sn 3d 5/2 peaks with similar binding energies at 486.5 eV, which could be deconvoluted into two peaks at 486.3 eV and 487.3 eV, corresponding to Sn 2+ and Sn 4+ , respectively (Figs S14-S17 in the online supplementary material). Besides, the peak of Zn 2p 3/2 at about 1021.2 eV belongs to Zn 2+ (Fig. S18 in the online supplementary material) [34]. The above results demonstrate the successful synthesis of a CuO catalyst with both Sn and Zn single atoms as co-promoters, and there generates strongly cooperative interaction between single-site Sn and Zn, which leads to the change in the electronic structure of the CuO catalyst.

Catalytic synthesis of dimethyldichlorosilane via the Rochow reaction
A vital industrialized reaction (Scheme S1 in the online supplementary material), the Rochow reaction, was used to evaluate the catalytic properties of xZn 1 -Sn 1 /CuO and Sn 1 /CuO. This reaction generates methylchlorosilanes as the main products, such as methyltrichlorosilane (MeSiCl 3 , M1), dimethyldichlorosilane (Me 2 SiCl 2 , M2), trimethylchlorosilane (Me 3 SiCl, M3), together with trace amounts of other Si-containing com-pounds. Among them, M2 is the most desired as it is the most important monomer used for the synthesis of the organosilicon polymers. Although many Cu-based catalysts have been reported to date [35][36][37], there is still room to improve the M2 selectivity and yield. It is expected that Zn 1 -Sn 1 /CuO with its unique structure might exhibit an outstanding catalytic property. For comparison purposes, the four physical mixtures were also obtained, including (1)    M2 selectivity are increased to 41.6% and 88.7% (entry 5), 13.8 and 2.7 times higher than those of CuO, respectively. It is noticed that the catalysts of CuO-0.1%Sn, CuO-0.1%Zn, CuO-0.1%Sn-0.1%Zn and Sn 1 /CuO-0.1%Zn also display a certain degree of improvement in catalytic performance compared with pure CuO (entries 7-10). However, these values are still much lower than those of 0.1Zn 1 -Sn 1 /CuO (Fig. 3A), even using a large amount of commercial Sn and Zn nanoparticles (5 wt% of CuO, see Fig. S23 and Table S3 in the online supplementary material), suggesting that 0.1Zn 1 -Sn 1 /CuO possesses an incredibly high activity for this reaction. Moreover, compared with CuO and Sn 1 /CuO (Tables S4-S6 in  These results indicate that the 0.1Zn 1 -Sn 1 /CuO catalyst possesses higher catalytic activity, selectivity and stability toward M2 than the other catalysts. As analyzed in the following section, this much-enhanced catalytic performance is associated with the cooperative interactions between the co-doped atomically dispersed promoters (Zn and Sn) and CuO, as well as the anchoring effect caused by defects. The waste contact masses (unreacted residue) after a 24 h reaction with the highest selectivity toward M2 for all the catalysts were characterized. From XRD analysis (Fig. S25 in the online supplementary material), it is found that the Cu x Si species, which is considered to be the real catalytic active phase [38,39], was formed and its strongest peak intensity was obtained on 0.1Zn 1 -Sn 1 /CuO, showing that 0.1Zn 1 -Sn 1 /CuO has the strongest ability to generate the active Cu x Si species. Scanning electron microscopy (SEM) observation (Fig. S26 in the online supplementary material) confirms that the shapes of the Zn 1 -Sn 1 /CuO catalyst remained unchanged, and there was an occurrence of Si etching during the catalytic process. Among all the catalysts, the extent of the Si etching was most severe on 0.1Zn 1 -Sn 1 /CuO. These results confirm that through controlling the surface structure of 0.1Zn 1 -Sn 1 /CuO, the formation of the Cu x Si active phase can be enhanced.

Density functional theory calculation
The DFT calculations support that Sn and Zn dopants are dispersed atomically on the CuO(110) surfaces. The calculated results show that for two Sn atoms to substitute two Cu atoms on CuO(110), the tendency is for substitution to occur at two Cu sites that are far from each other. Compared to the system of substituting two nearby Cu atoms, the energy for substituting two far-away Cu atoms is 0.14 eV lower (Fig. S27A in the online supplementary material). We also find that Sn energetically prefers to occupy a surface Cu site rather than a bulk Cu site. The energy difference between the surface and bulk sites is −1.76 eV/atom (E surface = −459.28 eV while E bulk = −457.52 eV).
Similarly, Zn atoms also prefer to separate from each other, but the energy difference of 0.08 eV for replacing two Cu atom by Zn atoms is smaller than that of the Sn doping case (supplementary Fig. S27B). We also find that Zn tends to occupy a Cu site that locates next to Sn if Sn exists on the surface, namely Sn and Zn atoms prefer to form pairs on CuO(110). The calculated result shows that forming a Sn-Zn pair is energetically 0.06 eV lower than separating Sn and Zn far away (supplementary Fig. S27C). Thus Zn prefers to fill in the nearby Cu vacancies caused by Sn doping. These theoretical results are consistent with the experimental observations. Figure 4A shows the atomic structure of CuO(110). We calculated the formation energies of a Cu vacancy for a clean CuO(110) surface (Fig. 4B), an Sn-doped CuO(110) surface (Fig. 4C) and a Zn-doped CuO(110) surface (Fig. S28 in the online supplementary material). It is found that on the Sn-doped CuO(110) surface, the formation energy is 0.78 eV lower than that on the clean CuO(110), while on the Zn-doped CuO(110) surface, it is 0.13 eV higher than that on the clean CuO(110). These results confirm that it is easier to form Cu vacancies in the Sn-doped case. The Sn doping facilitates the formation of Cu vacancies on CuO, and the latter can anchor Zn and achieve its single-atomic dispersion (Fig. 4D).
According to the previous reports [38,39], the catalytic reaction involves two stages: one is the transformation of catalyst into the Cu x Si active phase, and the other is the adsorption and activation of gaseous MeCl on the Cu x Si active phase to form gaseous products. Our previous work [40] has demonstrated that the rapid generation of free Cu atoms should be the crucial step to form Cu x Si, which is closely related to the dissociative adsorption strength of MeCl. Therefore, the dissociative adsorption behaviors of MeCl on the clean CuO(110) (Fig. 4E), Sn-doped CuO(110) (Fig. S29 in the online supplementary material), Zn-doped CuO(110) (Fig. S30 in the online supplementary material) and Sn-Zn pair-doped CuO(110) (Fig. 4F) were further investigated by DFT calculations. CH 3 tends to adsorb on the O top site, and the adsorption energy is only slightly affected by doping, while the strongest adsorption occurs for Sn doping (Table  S7 in the online supplementary material). However, we found a significant energy difference for Cl adsorption, and the most substantial adsorption occurs on Sn-doped CuO(110). Interestingly, the dissociative adsorption of MeCl is energetically unfavorable for clean Cu(110), but turns out to be energetically favorable upon doping, especially in the case of Sn-Zn pair and Sn-doping. Such an enhancement effect will facilitate the formation of copper chloride species and Cu atoms subsequently. As a result, the diffusion of Cu atoms to the surface of the Si matrix is enhanced, and thus the formation of the Cu x Si active phase promoted.
The electron transfers for CuO(110) with Sn or Zn doping and Sn-Zn pair doping are calculated as well, and the results are summarized in Table S8 in the online supplementary material. As discussed above, the doped Sn atoms can help to form surface Cu vacancies and make it possible to form Sn-Zn pairs on the surface. The results indicate that for Sn and Sn-Zn pair doped cases, the oxygen atoms bonded with Sn and Zn gain more electrons than those in clean CuO(110). In turn, the oxygen atoms gain fewer electrons from Cu atoms. Especially in the Sn-Zn pair doped case, Cu atoms transfer the fewest electrons to oxygen atoms, which means that Cu atoms have more valence electrons compared to the undoped CuO(110) surface. These theoretical results are consistent with the XPS measurements that electron density of the Cu atoms becomes higher for 0.1Zn 1 -Sn 1 /CuO relative to Sn-doped and clean CuO(110). Both experimental and theoretical results indicate that Sn and Zn interact with each other and have cooperative effects on the catalytic performance. Therefore, we can understand why Zn 1 -Sn 1 /CuO exhibits the best performance among various measured catalysts.

CONCLUSION
In summary, a novel CuO catalyst with atomically dispersed Sn and Zn as co-promoters has been successfully prepared via a facile hydrothermal method followed by wet impregnation. In the synthesis, single-site Sn is first incorporated into the lattice of CuO catalysts during hydrothermal treatment to generate a large number of surface Cu vacancies, which can then be used to anchor Zn atoms. This novel catalyst is highly active and stable towards M2 synthesis in the Rochow reaction. DFT calculation further confirms that the single-site Sn facilitates the generation of Cu vacancies, which can capture Zn and realize a single-atomic dispersion of Zn. The synergistic interaction between single-site Sn and Zn leads to the change in the electronic structure of the CuO catalyst, which promotes the adsorption of reactant MeCl and formation of the Cu x Si active phase, thereby leading to the enhanced catalytic performance. This work provides a new understanding of the synergistic effect among various promoters and will offer avenues to the design of new copromoters in catalysts for industrial reactions.

Reagents and chemicals
All the chemicals were of analytical grade and used without further purification. These chemicals include stannic chloride pentahydrate (

Materials preparation
Synthesis of Sn 1 /CuO, Zn 1 /CuO and CuO Sn 1 /CuO was synthesized by a reported method with some modifications [24]. First, 24.96 g (0.1 mol) of CuSO 4 ·5H 2 O and 0.0186 g (5.32 × 10 −2 mmol) of SnCl 4 ·5H 2 O were dissolved in 100 mL of deionized water under vigorous stirring and in an ice-water bath to form a homogeneous blue solution. Then, 200 mL of NaOH solution (1.2 mol/L) was slowly added, and the mixture was continuously stirred for 15 min. After being refrigerated (3 • C) for 24 h, the mixture was sealed and maintained at 130 • C for 18 h. Finally, the product was filtered after cooling, washed with distilled water and ethanol several times, and dried at 60 • C for 8 h. The synthesis of Zn 1 /CuO followed with the same procedures except that SnCl 4 ·5H 2 O was replaced by ZnCl 2 . Similarly, the CuO sample was also synthesized following the same procedures but without adding SnCl 4 ·5H 2 O.

Synthesis of xZn 1 -Sn 1 /CuO
The xZn 1 -Sn 1 /CuO samples were prepared by using the impregnated method, where x refers to the weight ratio of zinc to CuO. First, 4.00 g of Sn 1 /CuO was added in 100 mL of ethanol to obtain suspension A. A desirable amount of ZnCl 2 (0.0042 g, 0.0084 g and 0.0168 g, corresponding to 0.05 wt%, 0.1 wt% and 0.2 wt% relative to CuO, respectively) was dissolved in 20 mL of distilled water to obtain solution B. Subsequently, solution B was slowly added into the suspension A under stirring and the obtained mixture was continuously stirred for 150 min. The product was filtered, washed with distilled water and ethanol several times, and dried at 60 • C overnight under vacuum. Finally, the resulting powder was calcined at 400 • C for 3 h to obtain xZn 1 -Sn 1 /CuO.

Catalytic tests
The catalyst test was performed using a fixed-bed reactor. A 10.00 g amount of silicon powder (150 mesh, provided by Tangshan Sanyou Group Co. Ltd) was mixed homogeneously with 0.50 g of the prepared catalyst to form the contact mass, which was then loaded into the glass reactor. The reactor system was initially purged by methyl chloride (CH 3 Cl or MeCl, Zhejiang Guoya Gas Co., Ltd) at 20 • C for 0.5 h. Afterwards, the temperature was raised to 325 • C (5 • C min −1 ) for reaction for 24 h. The flow rate of MeCl was 25 mL min −1 . The gas product was cooled into liquid phase with a circulator bath controlled at −20 • C by a programmable thermal circulator (GDH series, Ningbo Xinzhi Biological Technology Co., Ltd). Gas chromatography (Agilent Technologies GC-7890A, KB-201 capillary column (60 m), thermal conducting detector (TCD)) was used to quantitatively analyze the products, which were mainly composed of methyltrichlorosilane (MeSiCl 3 , M1), dimethyldichlorosilane (Me 2 SiCl 2 , M2), trimethylchlorosilane (Me 3 SiCl, M3), methyldichlorosilane (MeSiHCl 2 , M1H), dimethylchlorosilane (Me 2 SiHCl, M2H), low boiler compounds (LB) and high boiler compounds (HB). The Si conversion and M2 selectivity are calculated as follows: Si conversion: C Si (%) = m Si,before − m Si,after m Si,before × 100 (2) Here, m Si, before and m Si, after in Formula (1) represent the mass of Si powder before and after the reaction, respectively, and m in Formula (2) is the mass of the products (as a percentage; peak area calibrated with response factor).

Characterization
XRD analysis was performed on a PANalytica X'Pert PRO MPD using CuKα radiation (k = 1.5418Å) at 40 kV and 40 mA. The size and shape of the as-prepared samples were observed using a cold field-emission SEM (SU8020, HITACHI, Japan) and a field-emission TEM (Tecnai G 2 F20 S-TWIN, FEI, USA), and a HAADF-STEM (JEM-ARM200F, JEOL, Japan) operated at 200 kV. The contents of Zn and Sn were determined by ICP-OES (Optima 5300DV, Perkin Elmer, USA) analysis. XPS spectra (Model VG ESCALAB 250 spectrometer, Thermo Electron, UK) were recorded using an AlKα X-ray source (hm = 1486.6 eV) radiation to analyze the surface chemical composition of samples. This reference gave BE values with an error within ±0.1 eV. Atomic force microscopy (AFM) (Multi-Mode 8, BRUKER, Germany) was used to observe the relative roughness of the sample surface.
The X-ray absorption fine structure (XAFS) spectra data (Cu K-edge) were collected at 1W1B station in Beijing Synchrotron Radiation Facility (BSRF, operated at 2.5 GeV with a maximum current of 250 mA). The data were collected in transmission mode. All samples were pelletized as disks of 13 mm diameter with 1 mm thickness using graphite powder as a binder. The acquired EXAFS data were processed according to the standard procedures using the ATHENA and ARTEMIS implemented in the IFEFFIT software packages. The fitting detail is described below.
The EXAFS spectra were obtained by subtracting the post-edge background from the overall absorption and then being normalized with respect to the edge-jump step. Subsequently, the χ (k) data of Fourier were transformed to real (R) space using a hinging window (dk = 1.0Å −1 ) to separate the EXAFS contributions from different coordination shells. To obtain the quantitative structural parameters around central atoms, least-squares curve parameter fitting was performed using the ARTEMIS module of IFEFFIT software packages.
The following EXAFS equation is used: S 0 2 is the amplitude reduction factor, F j (k) is the effective curved-wave backscattering amplitude, N j is the number of neighbors in the jth atomic shell, R j is the distance between the X-ray absorbing central atom and the atoms in the jth atomic shell (backscatterer), λ is the mean free path inÅ, φ j (k) is the phase shift (including the phase shift for each shell and the total central atom phase shift), σ j is the Debye-Waller parameter of the jth atomic shell (variation of distances around the average R j ). The functions F j (k), λ and φ j (k) are calculated with the ab initio code FEFF8.2. The coordination numbers of model samples are fixed as the nominal values. The obtained S 0 2 is fixed in the subsequent fitting, while the internal atomic distances R, Debye-Waller factor σ 2 , and the edge-energy shift E 0 are allowed to run freely.

Computational methods
All calculations were performed within the framework of DFT as implemented in the Vienna Ab initio Simulation Package (VASP) code [41][42][43]. The electron-ion interaction is described using the projector augmented wave method [44,45]. We employ the generalized gradient approximation (GGA) in the Perdew-Burke-Ernzerhof (PBE) form for the exchange-correlation functional [46]. The CuO(110) surface is simulated by a slab model constructed with the theoretical equilibrium lattice constants. The vacuum thickness is 15Å. The atoms in the top four layers are allowed to relax until the forces on those atoms are less than 0.02 eVÅ −1 , while the atoms in the bottom two layers are fixed at the bulk lattice sites. The energy cutoff for the plane wave basis is 500 eV for all our calculations. The Brillouin zone integration is sampled with the 4 × 4 × 1 k-point mesh by the Monkhorst-Pack scheme [47].

SUPPLEMENTARY DATA
Supplementary data are available at NSR online.