Design of CuCs-doped Ag-based Catalyst for Ethylene Epoxidation

Ⅰ.   INTRODUCTION

Ethylene oxide (EO) is one of the important chemical derivatives of ethylene. Ethylene oxide is a key chemical reaction intermediate, which can be further converted into ethylene glycol, surfactants, ethanolamine, and other important chemical products [1]. Industrially, ethylene oxide is produced selectively by the epoxidation reaction. The key to this reaction is the use of Ag catalysts, whose special role improve the selectivity of ethylene epoxidation [2, 3]. Until now, many researchers have tried to improve its catalytic performance by various methods, such as doping promoters [4-13], constructing alloys [14-16], modifying carriers [17-20], modifying Ag particles morphology [21-27], etc. Among these methods, the doping of trace promoters is one of the most effective ways, so it is significant to study the promoter-doped Ag-based catalysts to obtain more efficient catalytic performance.

The promoters generally doped to Ag-based catalysts are Cs, Re, Cl, etc. Among them, the effect of Cs has become a hot research topic [6, 9, 28-37]. Reviewing the effect of Cs, it is divided into two categories, namely the geometric effect and the electronic effect. The geometric effect suggests that Cs occupies non-selective active sites, thus inhibiting the occurrence of non-selective reactions and improving the EO selectivity. Atkins et al. [31] and Waugh et al. [32] found that Cs inhibit oxygen adsorption on the stepped surface to prevent the formation of strong adsorption Ag-O bonds, with an EO selectivity of 33% on this surface. Cs has no effect on oxygen adsorption on the Ag(111) surface, which has 57% EO selectivity. Amorim et al. [33] suggested that Cs would promote the dispersion of Ag particles, allowing the formation of an Ag film on the surface of the carrier. In addition, it increased the number of lattice defects and improved the oxygen storage capacity, thus affecting the catalytic performance. Recent studies have also suggested that the effect of Cs involves the electronic effect. Linic et al. [34] proposed that the effect of Cs is to stabilize the transition state of EO formation through the electric field/dipole interaction effect rather than the transition state of combustion reaction process. Diao et al. [5] employed X-ray photoelectron spectroscopy and found that Cs decreases the binding energy of Ag to promote the desorption of adsorbed EO. Ren et al. [6] found that Cs weakens the adsorption strength of oxygen atoms, lowering the activation energy of oxometallacycle (OMC) intermediate formation and EO, as well as increasing the activation energy of byproducts acetaldehyde (AA) formation. Our team studied the effect of Cs and found that Cs causes charge redistribution on the Ag surface, which affects the adsorption capacity of the adsorbate and thus the EO selectivity [36]. Moreover, Salaev et al. [37] found that Cs and Re promoters tend to form CsReO$_x$ species. The individual promoters promote each other to adjust the electronic state of surface oxygen and improve the EO selectivity. In addition, Cu has also been doped to Ag catalysts for ethylene epoxidation [36, 38-41]. Linic et al. [38] found experimentally and computationally that CuAg catalysts for ethylene epoxidation increase EO selectivity. Jankoiak et al. [39] also found that CuAg catalysts exhibit excellent catalyst performance for ethylene epoxidation in a wide range of feed compositions. Recently, our team investigated the effect of CuAg composition on ethylene epoxidation and found a volcano-like selectivity trend with increasing Cu content [41].

The doping of both Cs and Cu into Ag catalysts for ethylene epoxidation has also been studied [12, 42, 43]. Jankoiak et al. [42] studied Cs doping into Cu-Ag catalysts for ethylene epoxidation and found that the catalytic performance of CsCu-Ag catalysts was superior to that of Ag catalysts. Our team has also conducted a theoretical study on the catalytic performance of CuCs-doped Ag catalysts for ethylene epoxidation. It was found that the CuCs-doped Ag catalysts were more selective than Ag catalysts on both Ag(111) and Ag(110) surfaces [43]. In addition, our team further studied the Ag catalytic performance for ethylene epoxidation between CuCs-doped oxygen-reconstructed Ag (Orec-Ag) surface and pure oxygen-reconstructed Ag (Orec-Ag) surface under the reaction condition. The predicted result indicated that the performance of CuCs-doped Orec-Ag catalysts on Ag(111) and Ag(100) were also better than that of pure Orec-Ag catalysts on Ag(111) and Ag(100) [12]. However, various theoretical computational studies investigating promoter-doped Ag catalysts have focused on surface models for ethylene epoxidation. Instead, during the actual ethylene epoxidation reaction, the Ag catalyst is presented in a particle shape, so the catalytic performance prediction of this surface model may not be particularly robust. Therefore, it is necessary to investigate the effect of CuCs-doped Ag catalysts on the performance of ethylene epoxidation reaction at the particle level.

In this work, we combined theoretical calculations with experiments to investigate the effect of CuCs-doped Ag catalysts on ethylene epoxidation. According to the previous work of our team [12], two types of Ag surface models namely metallic and oxygen-reconstructed surface were constructed to represent Ag catalyst surface at low and high oxygen pressure. The performance of CuCs-doped Ag catalysts on ethylene epoxidation at the particle level was comprehensively explored by density functional theory (DFT), the Wulff construction theory [44], and micro-kinetic analysis. Further experimental evaluation results were used to verify the theoretical calculation results. Moreover, the samples were characterized to analyze the reasons for the improved catalytic performance.

Ⅱ.   COMPUTATIONAL AND EXPERIMENTAL METHODS

A   Computational method

DFT calculations was carried out by using the Vienna ab initio Simulation Package (VASP) code. The pseudopotential used the Perdew-Burke-Ernzerhof (PBE) function. After a series of tests to determine the energy cutoff with 450 eV. The structural optimization process was performed using the BFGS algorithm, and the convergence standard was 0.2 eV/Å. The convergence criterion of self-consistent field (SCF) iteration was 10$^{-5}$ eV.

Two types of Ag surface models were constructed. One was the metallic p(3$\times$3) Ag(111) and p(3$\times$3) Ag(100) surface. The other was oxygen-reconstructed p(4$\times$4) OrecAg(111) and p(2$\times$2) OrecAg(100) surface. The promoter CuCs double dopants were doped into the above mentioned surface. Each surface contains five layers, in which the bottom three layers of atoms are fixed, and the top two layers of atoms relax. To eliminate the dipole moment and surface-surface interaction, a vacuum layer of 15 Å thickness was inserted in the $z$-directed. The $k$-point gird of the p(3$\times$3) Ag(100), p(3$\times$3) Ag(111), p(4$\times$4) OrecAg(111) and p(2$\times$2) OrecAg(100) were 3$\times$3$\times$1, 3$\times$3$\times$1, 3$\times$3$\times$1, and 6$\times$6$\times$1, respectively.

The surface free energy ($\gamma$) is calculated by the following equation:

In which, $G_{ {\rm{total}}}$ is the total Gibbs free energy of the surface system. $G_{ {\rm{metal\; bulk}}, i}$ is the Gibbs free energy of a single metal $i$ atom in the bulk structure. $N_{ {\rm{metal}}, i}$ is the number of atoms of metal $i$ on the surface. $A$ is the surface area. $N_ {\rm{O}}$ is the number of O atoms on the surface. $\mu_ {\rm{O}}$ is the chemical potentials of oxygen atoms, which is related to the temperature and pressure of the reaction system [45]:

in which, $p^0$ represents standard atmospheric pressure, $k_ {\rm{B}}$ represents Boltzmann's constant, $\mu_{ {\rm{O}}_2 }$($T, p^0$) is the chemical potentials of molecular oxygen at temperature $T$ under standard atmospheric pressure. The total energy of molecular oxygen $E_{ {\rm{O}}_2}$, is calculated as the energy difference between bulk Ag oxide ($E_{ {\rm{Ag}}_2 {\rm{O}}}$), bulk Ag ($E_{ {\rm{Ag}}}$), and the experimental value of the heat of formation ($H_{ {\rm{Ag}}_2 {\rm{O-bulk}}}^ {\rm{f}}$).

B   Catalyst preparation, evaluation, and characterization

Pure Ag catalyst and CuCs-doped Ag catalyst were prepared by the excessive impregnation method, according to the method of the patent of Ref.[46]. First, Ag oxalate (Ag$_2$C$_2$O$_4$) was added to a 3:1 mol/L mixed solution of ethylenediamine (EDA) and deionized water to form an impregnating solution with a certain concentration of Ag. The $\alpha$-Al$_2$O$_3$ support (produced by Lanzhou Chemical Research Center) was added to the impregnation solution. After the immersion was completed, the immersion liquid was leached. Then, the catalyst was activated in a tube furnace at 300 ℃ for 10 min to prepare a 15 wt% pure Ag catalyst. The preparation of 15 wt% CuCs-doped Ag catalyst adopted the same steps as described above. The difference was that the appropriate amount of cesium nitrate standard solution and copper nitrate standard solution was added when configuring the immersion solution.

The evaluation of the catalysts was carried out inside micro evaluation equipment with a stainless steel tube of 4 mm inner diameter. The reaction airspeed was set to be 4500 h$^{-1}$. The component of the gas going inside and outside the equipment was analyzed by Prima online mass spectrometer. The component of gas feed was C$_2$H$_4$ 30 mol%, O$_2$ 7.5 mol%, CO$_2$ 1.55 mol%. In all evaluation processes, an amount of 0.1-0.5 ppm inhibitor dichloroethane (EDC) was added. N$_2$ was utilized to be the balance gas, the pressure of the reaction was set to be 2.1 MPa. The temperature was recorded when the EO concentration of the outlet gas stream was attained at 2.5 mol%. The selectivity $S$ is calculated by

The scanning electron microscopy (SEM) test was performed on used Hitachi S4800 scanning electron microscope to take micrographs of the surface of the Ag catalyst sample and analyze the particle morphology distribution on the surface of the catalyst. X-ray diffraction (XRD) test used Japanese Rigaku Ultima Ⅳ instrument, with Cu k$\alpha$ radiation. The scanning range was 5°-90° and the scanning speed was 1°/min. Crystallite size was calculated by the Scherrer equation to the (220) reflection of the Ag phase.

here, $K$ is the shape factor (0.89), $\lambda$ is the X-ray wavelength, $\beta$ is the full-width at half maximum and $\theta$ is the Bragg angle of the Ag(220).

X-ray photoelectron (XPS) test used a Thermo Scientific K$\alpha$ instrument, the excitation source was Al K$\alpha$ rays, and the binding energy of the A1 2p line was 74.5 eV. Oxygen-temperature-programmed desorption (O$_2$-TPD) test used AutoCheml Ⅱ 2920 instrument. The 0.2 g catalyst sample was exposed to 30 mL/min He for 1 h, and then the temperature was raised to 550 ℃ for 1 h, cooled to 210 ℃, switched to 10 O$_2$/He gas flow for oxygen adsorption for 1 h. Then temperature droped to room temperature at He. After the baseline was stable, the temperature increased from room temperature to 800 ℃ for oxygen desorption.

Ⅲ.   RESULTS AND DISCUSSION

A   Theoretical catalytic performance at particle model

Previous studies by our team predicted that catalytic performance of CuCs-doped Ag catalysts is superior to pure Ag catalysts on the surface model [12, 43]. However, the Ag catalyst is particle-shaped during the actual reaction. And the catalyst morphology is affected by the reaction environment atmosphere [47]. Therefore, in this section, we focused on the particle model to investigate the difference in catalytic performance between CuCs-doped Ag catalysts and pure Ag catalysts.

First, based on the previous study [12], we constructed two types of surface models. One was a metallic surface at low oxygen pressure, whose surface was not oxidized and may represent a catalyst in the initial stage of the reaction. The other one was an oxygen-reconstructed surface at high oxygen pressure, whose surface was reconstructed by oxygen and maybe represent a catalyst in active stable stage of the reaction. As the reaction continues, oxygen will deeply oxidize the Ag surface, and the Ag surface is reconstructed by oxygen. This oxygen-reconstructed surface was also observed experimentally [48-50]. This could be the catalyst for the active stable stage of the reaction.

The DFT was employed to calculate the surface free energy of each surface model (as shown in Table S1 in Supplementary materials (SM)). The surface free energy information at each temperature point was obtained, as shown in FIG. 1(a, b). For the metallic Ag surface, the surface free energy of both Ag(111) and Ag(100) surfaces decreased after CuCs doping. Similarly, for the oxygen-reconstructed Ag surface, the surface free energy of Orec-Ag(111) and Orec-Ag(100) surfaces also decreased after CuCs doping. This indicates that the CuCs-doped surfaces are more stable. In addition, for any type of Ag surface, the surface free energy of the Ag(111) surface is smaller than that of the Ag(100) surface. This indicates that the Ag(111) surface is more stable and it can be predicted that the particle model in a real environment will expose more Ag(111) surface.

Figure  1.  Surface free energy of Ag surface at (a) metallic surface and (b) oxygen-reconstructed surface at 480, 490, 500, 510, and 520 K. (c) Particle model of pure Ag, CuCs-doped Ag, pure OrecAg, and CuCs-doped OrecAg at 500 K, respectively. The green face is (111) surface, and the yellow face is (100) surface. (d) Particle boundary density (particle edge length to area ratio) of the Pure OrecAg and CuCs-doped OrecAg particle model at 15, 20, 25, and 30 nm.

DownLoad: Full-Size Img PowerPoint

Further, we constructed the corresponding particle models by Wulff construction theory. FIG. 1(c) show the corresponding particle models at 500 K. The shapes of the particles at other temperature points are similar. When constructing the particle model, we defined the particle size as the distance from the (111) surface to the center of the particles. Based on the previous experimental findings that large silver particles are mainly composed of 25-30 nm multiple primary silver crystallites [25], we constructed a 25 nm particle model for the following calculating on catalytic performance. In addition, we constructed Pure OrecAg and CuCs-doped OrecAg particle models of 15, 20, 25, and 30 nm, respectively. As shown in FIG. 1(d), we found that the particle boundary density decreases as the particle model size increases. Theoretical calculations and experiments show that the EO selectivity on Ag(111) and Ag(100) surfaces are superior to that of stepped surfaces [12]. Therefore, we believe that larger particle models have less edges and vertexes, and larger size Ag particles will have better catalytic selectivity.

After the particle models were constructed, we counted the area percentage of the Ag(111) surface to the whole particle area in each particle model, as shown in Table S3 in SM. The reaction energy information on each surface model was referred to the previous results of our team as shown in Tables S4 and S5 in the SM [12, 51]. The reaction rate information on the surface models was acquired by using pervious study method [12].

Firstly, The micro-kinetic model for ethylene epoxidation on Ag surface models is given as the following:

Then, according to the reaction network, the micro kinetic analysis was simplified by assuming that the dissociated adsorption of O$_2$ and adsorption of C$_2$H$_4$ can be in kinetic equilibrium. The rate of these reactions can be written as the following:

where ${{{K}}_1}$=$\exp {\left( {\frac{{ - \Delta{{G}}_1}}{{{{kT}}}}} \right)}$ and ${{{K}}_2}$=${{\rm{exp}}} {\left( {\frac{{ - \Delta {{G}}_2}}{{{{kT}}}}} \right)}$ are the equilibrium constants for R1 and R2, $p$(O$_2$) and $p$(C$_2$H$_4$) are the partial pressures of O$_2$ and C$_2$H$_4$, $k_i^+$ and $k_i^-$ represents the forward and the backward reaction constant for R$i$ ($i$=1, 2), and $\Delta G_1$=$G_{ {\rm{ad}}}$(O) and $\Delta G_2$=$G_{ {\rm{ad}}}$(C$_2$H$_4$) are adsorption free energy of atomic oxygen and C$_2$H$_4$ with O pre-adsorption.

The reaction rates of R3, R4, R5, R6, R7, and R8 ($r_3$, $r_4$, $r_5$, $r_6$, $r_7$, and $r_8$) can be calculated from the following equations:

where ${{{k}}_{m{i}}}$=${\frac{{{{kT}}}}{{{h}}}}\exp \left( {\frac{{ - {\Delta {G_i}}}}{{{{kT}}}}} \right)$, and $\Delta G_{ {\rm{a}}i}$ is the activation free energy for reaction of R$i$ ($i$=3, 4, 5, 6, 7, 8).

Next, by applying the "steady-state'' approximation, ${\frac{{{{\rm{d}}}{{\theta}_{{{\rm{OMC}}}}}}}{{{{\rm{dt}}}}}}$=0, ${\frac{{{{\rm{d}}}{{\theta}_{{{\rm{EO}}}}}}}{{{{\rm{dt}}}}}}$=${\frac{{{{\rm{d}}}{{\theta}_{{{\rm{AA}}}}}}}{{{{\rm{dt}}}}}}$=0

The sum of the coverage of adsorbed O, C$_2$H$_4$, OMC, EO, AA, and free adsorption sites ($\theta_*$) should be equal to 1.

Similar to the example for the rate on the particle model mentioned in the study of Yi et al. [52]. After the rates of EO formation on the surface models are obtained, we combine the above particle model information to predict the rates of EO formation at the particle level. The rate of EO formation at particle level is as follows:

Here, $S$ and $q_k$ represents the total surface area of a single particle and the area ratio of the crystal plane $k$ on the particle, and $r_{ {\rm{EO}}}^k$ represents the rate of EO formation on crystal plane $k$ per unit area. Furthermore, the EO selectivity ($S_{ {\rm{EO}}}$) and the apparent activation energy ($E_ {\rm{a}}$) of the catalyst at particle level are deduced.

Thus, the EO selectivity at the particle model for different catalysts was acquired under reaction conditions ($p$$_{ {\rm{O}}_2}$=0.3 bar, and $p$$_{ {\rm{C}}_2 {\rm{H}}_4}$=0.1 bar, $T$=480, 490, 500, 510, and 520 K, respectively). As shown in FIG. 2(a, b), for metallic Ag catalyst in the initial stage of reaction, the CuCs doping into the pure Ag catalyst increased the EO selectivity from 41.93%-42.66% to 60.72%-61.49%. For the oxygen-reconstruction Ag catalyst in active stable stage of reaction, the selectivity of CuCs-doped Ag catalyst (61.95%-62.75%) is also greater than that of pure Ag catalyst (43.99%-44.46%). This shows that at any temperature point and reaction stage, the EO selectivity of CuCs-doped Ag catalyst is better than that of the pure Ag catalyst. Compared to our previous works on the surface model, the EO selective results on the particle model are closer to those on the Ag(111) surface model. The main reason is that the particle model exposes much Ag(111) surface and the EO formation rate on the Ag(111) surface is greater than that on the Ag(100) surface.

Figure  2.  (a, b) The EO selectivity of each particle model for pure Ag catalysts and CuCs-doped Ag catalysts at 480, 490, 500, 510, and 520 K, respectively. (c, d) Arrhenius plots for determination of the apparent activation energy $E_ {\rm{a}}$ of $R_{ {\rm{EO}}}$ at particle model.

DownLoad: Full-Size Img PowerPoint

In addition, the apparent activation energy ($E_ {\rm{a}}$) of the catalyst at the particle model was determined by Arrhenius plots, as shown in FIG. 2(c, d). For metallic Ag catalyst, the apparent activation energies of pure Ag catalysts and CuCs-doped Ag catalysts are -522.71 kJ/mol and -484.33 kJ/mol, respectively. For oxygen-reconstruction Ag catalyst, the apparent activation energy value (-461.37 kJ/mol) of the CuCs-doped Ag catalyst was also less than that of the pure Ag catalyst (-504.27 kJ/mol). The lower activation energy value indicates a higher activity. Therefore, CuCs-doped Ag catalysts have better activity compared to pure Ag catalysts, and the temperature required for reaction proceeding of CuCs-doped Ag catalysts will be lower.

B   Experimental evaluation of catalyst performance

The theoretical study successfully predicted that CuCs-doped Ag catalyst has better activity and EO selectivity than pure Ag catalyst. We further carried out experiments to verify our prediction results. The pure Ag catalyst and CuCs-doped Ag catalyst samples were prepared by the impregnation method. Through the evaluation and characterization of the catalyst, the performance difference of the catalyst was analyzed. The evaluation results of pure Ag catalyst and CuCs-doped Ag catalyst are shown in FIG. 3. The EO selectivity over pure Ag catalyst is 76.15% under the condition of EDC contained in the component of gas feed. The EO selectivity increases by 1.27% to 77.42% when doped with CuCs on Ag catalyst in the same reaction environment, which is consistent with the theoretical calculation predicting. In addition, when EO concentration of outlet gas stream is 2.5 mol%, the reaction temperature of pure Ag catalyst (216.3 ℃) is higher than that of CuCs-doped Ag catalyst (211.7 ℃), indicating that the activity of CuCs-doped Ag catalyst is higher than that of pure Ag catalyst. The experimental results are also consistent with the results predicted by theoretical prediction, that is, the CuCs-doped Ag catalyst has lower apparent activation energy and better activity than the pure Ag catalyst.

Figure  3.  (a) EO selectivity and (b) reaction temperature of pure Ag catalysts and CuCs-doped Ag catalysts under the condition of EDC contained in the component of gas feed and at 2.5 mol% EO concentration of outlet gas stream.

DownLoad: Full-Size Img PowerPoint

C   Promoter effect of CuCs double dopant

The SEM image and size distribution of the pure Ag catalyst and CuCs-doped Ag catalyst, are shown in FIG. 4. In FIG. 4 (a, b), the smooth and black areas are $\alpha$-Al$_2$O$_3$ support, and the smaller, off-white is the loaded Ag particles. It can be seen intuitively that the size of Ag particles for CuCs-doped Ag catalyst is relatively uniform, and the size is slightly larger than that of pure Ag catalyst. For a more accurate analysis, we counted the particle size of the corresponding Ag catalyst, and the statistical results are shown in FIG. 4 (c, d). The histogram clearly shows that the Ag particle size of the pure Ag catalyst is 86.69$\pm$25.67 nm, while the Ag particle size of the CuCs-doped Ag catalyst is 102.50$\pm$19.87 nm. This indicates that CuCs doping increases the size of Ag particles and makes the size of Ag particles more uniform. Ren et al. [6] and Hassani et al. [30] studied the effect of Cs on the Ag catalyst. They found that Cs can promote the dispersion of Ag particles and make the size distribution of Ag particles more uniform. In addition, under almost the same loading conditions, the doping of Cs induces a decrease in the size of Ag particles [30]. Dellamorte et al. [5] studied the changes in Ag morphology induced by Cu promoter and found that Cu doping lead to an increase in the size of Ag particles. In this work, we found here that after doping CuCs in Ag catalyst, the size of the Ag particles increases, and the size distribution is relatively uniform. It suggests that Cu and Cs dopant have played their advantages, respectively. Cu promotes the growth of the size of Ag particles, while Cs regulates the distribution of Ag particle size more uniformly. Moreover, our theoretical study above found that the large-size particle model has fewer step surfaces, and the step surfaces may be beneficial for the occurrence of non-selective reactions. Therefore, CuCs-doped Ag catalysts with larger particles and uniform size distribution may be attributed to the better catalytic performance for ethylene epoxidation.

Figure  4.  SEM images, and corresponding histograms for (a, c) pure Ag/$\alpha$-Al$_2$O$_3$ catalyst and (b, d) CuCs-doped Ag/$\alpha$-Al$_2$O$_3$.

DownLoad: Full-Size Img PowerPoint

FIG. 5(a) shows the XRD diffractogram of $\alpha$-Al$_2$O$_3$, pure Ag/$\alpha$-Al$_2$O$_3$ catalyst, and CuCs-doped Ag/$\alpha$-Al$_2$O$_3$ catalyst. The peaks marked at 38.1°, 44.6°, and 64.8° represent Ag(111), Ag(200), and Ag(220), respectively. Others represent the exposed surface of $\alpha$-Al$_2$O$_3$ supports. By enlarging the XRD diffractogram of the Ag(220) region, as shown in FIG. 5 (b) and (c), we calculated the primary crystallite size of Ag particles. The primary crystallite size of the pure Ag particles and the CuCs-doped Ag particle is 23.6 nm and 25.7 nm, respectively. The number of primary Ag crystallites was estimated by combining the primary crystallite size and the average particle size from SEM statistics. The estimation results indicate that CuCs-doped Ag particles with an average size of 102.50 nm consist of approximately 60 primary Ag crystallites. The Ag particles of pure Ag with an average size of 86.69 nm consist of about 45 primary Ag crystallites. Therefore, CuCs-doped Ag catalyst is rich in more grain boundaries, which can activate more oxygen to react with ethylene, thereby improving the catalytic performance of Ag catalyst [25].

Figure  5.  (a) XRD characterization. (b, c) The corresponding magnification of the XRD diffractogram of the Ag(220) region for pure Ag/$\alpha$-Al$_2$O$_3$ catalyst, and CuCs-doped Ag/$\alpha$-Al$_2$O$_3$ catalyst, respectively.

DownLoad: Full-Size Img PowerPoint

As shown in FIG. 6(a), XPS results show that the binding energy of Ag 3d$_{5/2}$ of pure Ag catalyst is 367.3 eV, while the binding energy of Ag 3d$_{5/2}$ of CuCs-doped Ag catalyst is 367.5 eV. Doping CuCs into the Ag catalyst transfer the Ag 3d$_{5/2}$ BE to a higher value. Many previous studies have shown that single doping of Cs into the Ag catalyst reduces the Ag 3d$_{5/2}$ BE [5, 6]. Based on the previous reliable research results, we considered that if our catalyst sample is only doped with Cs, the Ag 3d$_{5/2}$ BE will also decrease. Our XPS results here show that the Ag 3d$_{5/2}$ BE of CuCs-doped Ag catalyst is greater than the Ag 3d$_{5/2}$ BE of pure Ag catalyst. This shows that the addition of Cu not only offsets the reduction of the Ag 3d$_{5/2}$ BE derived from Cs dopant but also further increases the Ag 3d$_{5/2}$ BE. We inferred that the increase in the Ag 3d$_{5/2}$ BE caused by Cu is similar to the increase in the Ag 3d$_{5/2}$ BE caused by the introduction of Re into the CsAg catalyst [5]. Therefore, we suggested that Cu and Cs still play their respective advantages. Among them, the role of Cs is to improve the desorption of EO and prevent the isomerization of EO into CO$_2$, which further improves the selectivity of EO [29]. The role of Cu is to enhance the electrophilicity of adsorbed atomic oxygen and promote the attack toward electron-rich C=C bonds, which also enhance the selectivity of EO [5, 53].

Figure  6.  (a) The XPS characterization and (b) the O$_2$-TPD characterization for pure Ag/$\alpha$-Al$_2$O$_3$ catalyst and CuCs-doped Ag/$\alpha$-Al$_2$O$_3$ catalyst, respectively.

DownLoad: Full-Size Img PowerPoint

The O$_2$-TPD results further verified that the increase of electrophilicity of adsorbed atomic oxygen is promoted on CuCs-doped Ag catalysts. It can be seen in FIG. 6(b) that at higher temperature points more lattice atomic oxygen is desorbed from CuCs-doped Ag catalyst than pure Ag catalyst [10, 31]. In addition, the oxygen adsorption concentration on CuCs-doped Ag catalysts is greater than that of pure Ag catalysts both at low-temperature and high-temperature oxygen desorption peak. We thought that a higher oxygen adsorption concentration may favor the catalyst activity and thus reduce the required reaction temperature.

Ⅳ.   CONCLUSION

In summary, theoretical calculations and experiments were employed to design CuCs-doped Ag catalysts for ethylene epoxidation. Theoretical calculations predicted that the selectivity and activity of CuCs-doped Ag catalysts at the particle model were superior to those of pure Ag catalysts. The CuCs-doped Ag catalysts showed excellent catalytic performance when the catalysts were in either the initial or active stable stage of the reaction. Further experiments verified the theoretical calculation results. Moreover, the SEM characterization revealed that the effect of Cu increases the Ag particle size, while the effect of Cs promotes the uniform dispersion of Ag particles. Thus CuCs-doped Ag catalysts have larger Ag particles size and are enriched with grain boundaries, which are beneficial for the catalytic performance. The XPS and O$_2$-TPD results revealed that Cu and Cs exert their respective advantages to promote the selectivity of the catalysts. The effect of Cs promotes the desorption of EO, while the effect of Cu increases the electrophilicity and coverage of the adsorbed atomic oxygen.

Supplementary materials: Structural information for each surface model, surface free energy of the initial stage and the active stable stage, ratio of the (111) surface area to the entire particle area, EO selectivity of particle models are available.

Ⅴ.   ACKNOWLEDGMENTS

This work is supported by PetroChina Innovation Foundation (2019D-5007-0403).