Macrokinetics of Ethylene Epoxidation over A-type Silver Catalyst

Liang Rujun; Li Jianwei

(1. State Key Laboratory of Chemical Resource Engineering, Beijing University of Chemical Technology, Beijing 100029; 2. Yanshan Branch, Beijing Research Institute of Chemical Industry, SINOPEC, Beijing 102500)

Macrokinetics of Ethylene Epoxidation over A-type Silver Catalyst

Liang Rujun1,2; Li Jianwei1

(1. State Key Laboratory of Chemical Resource Engineering, Beijing University of Chemical Technology, Beijing 100029; 2. Yanshan Branch, Beijing Research Institute of Chemical Industry, SINOPEC, Beijing 102500)

By taking the surface chemical reactions as the rate-controlling step, a possible reaction mechanism for ethylene epoxidation to synthesize ethylene oxide over the A-type silver catalyst was developed, while it was assumed that the epoxidation reaction would take place between ethylene and the un-dissociated adsorbed oxygen O2aon the solo active sites, while the deep oxidation would occur between ethylene and the dissociated adsorbed oxygen Oaon the adjacent multi-active sites. In order to describe the effect of 1,2-C2H4Cl2(EDC) inhibitor on the ethylene epoxidation process, the reversible reactions between EDC and vinyl chloride (VC) on the active sites of silver catalyst was introduced. According to the assumed mechanism, the hyperbolic macro-kinetic model of ethylene epoxidation over the A-type silver catalyst was established, and the macrokinetic experiments were carried out in an internal-recycle gradientless reactor operating at a pressure of 2.1 MPa and a temperature in the range of 217.8—249.0 ℃, with the gas composition (molar fraction) consisting of 15.82%—34.65% C2H4, 2.55%—7.80% O2, 0.88%—6.15% CO2, 0.15—2.61 μmol/mol of 1,2-C2H4Cl2and 0.14—1.28 μmol/mol of C2H3Cl. By means of the Simplex Optimal Method, the parameters of the macrokinetic models were estimated. Statistical test showed that the macrokinetic models developed for the A-type silver catalyst agree well with the experimental results.

silver catalyst; ethylene epoxidation; reaction mechanism; macrokinetics

1 Introduction

At present, the silver catalyst is the only industrial catalyst for manufacture of ethylene oxide via direct oxidation of ethylene in the presence of air or pure oxygen. Although the reaction kinetics of ethylene oxidation over silver catalyst has been studied and applied in industry for decades, many arguments still remain which include some essential questions such as the reaction mechanisms and the reaction kinetic behaviour description regarding the complexity of ethylene epoxidation process and the limitation of research methods or means. These arguments also result in a great difference that is even conflicting in published literature on the ethylene epoxidation kinetic behavior description relating to both of the kinetic model form and the kinetic parameter values. Klugherz, et al.[1]assumed that the reaction takes place between adsorbed ethylene and adsorbed oxygen species on active sites that are dissociated from a lower oxygen layer on metallic silver. Ethylene and oxygen, as well as the reaction products, compete for adsorption sites on the catalyst surface. However, Metcalf, et al.[2]proposed a kinetic model in which the rate-controlling step was a bimolecular surface reaction between the adsorbed ethylene and the molecular oxygen. Most of their experimental data are complied with this model for the formation of ethylene oxide and carbon dioxide. It also succeeds in predicting the competition of active sites among water, carbon dioxide, and the reactants. Petrov, et al.[3]developed an empirical kinetic model with an E-R type mechanism. According to this model, the adsorbed molecular oxygen reacts on ethylene to produce ethylene oxide, whereas the atomic oxygen is responsible for the complete oxidation reaction. Borman, et al.[4]assumed that a bimolecular surface reaction was the rate-controlling step. This bimolecular surface reaction undergoes an L-H type mechanism. In this reaction, the adsorbed ethylene would react upon the dissociated oxygen to produce ethylene oxide, carbon dioxide,and water. Moreover, the two reactions were assumed to proceed over different catalytic sites. Stoukides, et al.[5]supposed that ethylene and ethylene oxide compete for the same sites on the catalyst surface. But Al-Saleh, et al.[6]believed that carbon dioxide has an inhibitory effect on the reaction rates of both ethylene oxide and carbon dioxide. Lafarga, et al.[7]investigated experimentally the kinetic network of ethylene epoxidation over a cesiumdoped silver catalyst supported on α-Al2O3pellets in a differential reactor, the result indicated that both reactants and products can influence the reaction rates; however, the influence of products is negligible in their dual-site L-H type kinetic mechanism models because of the small partial pressure of the product in the reactor. Borman[4]and Lafarga[7]suggested that their experimental data did not obey the E-R type kinetics. Carucci, et al.[8]developed two L-H type models for the partial oxidation of ethylene. In these competitive models, the surface reaction between the adsorbed ethylene and molecular or dissociated oxygen is considered as the rate-controlling step. Their observation supported the competitive adsorption of ethylene and molecular oxygen on the surface[8].

In this paper, we proposed a possible reaction mechanism for ethylene epoxidation over the A-type silver catalyst developed by the Yanshan Branch of Beijing Research Institute of Chemical Industry of China. The corresponding kinetic model was deduced and the macrokinetic experiments were carried out in a gradientless reactor with internal recycle. The nonlinear optimization analysis using the Simplex Optimal Method showed that the macrokinetic experimental data can be predicted by the kinetic model. The reaction mechanism was proved to be highly acceptable and reliable.

2 Reaction Mechanism

The reaction scheme in ethylene oxidation over the silver catalyst is proposed as follows:

These parallel and consecutive reactions form a complicated reaction network. Epoxidation (Eq. 1) and complete combustion (Eq. 2) are assumed to occur simultaneously and compete with each other. This kinetic study is mainly focused on the competition between reactions given in Eq. 1 and Eq. 2.

Studies[9]have shown that characteristics of the silver catalyst are ascribed to the types of adsorbed oxygen species.

Figure 1 Reaction scheme of ethylene oxygenation over a silver catalyst

In general, three types of oxygen are accepted in the reaction system, i.e., the adsorbed molecular oxygen (O2α), the adsorbed atomic oxygen (Oα), and the subsurface oxygen (Osub). It is elusive to find out that which type of oxygen is actually involved in the ethylene epoxidation process. An early study[10]assumed that the interaction between gaseous ethylene and O2αproduces ethylene oxide while the interaction between gaseous ethylene and Oαproduces carbon dioxide. But the later studies[11-13]suggested that the adsorbed ethylene reacts upon Oαto form an oxametallacycle intermediate. This intermediate is commonly generated in the course of ethylene epoxidation, ethylene combustion, and ethylene oxide combustion.

Based on the mechanistic study[14]on direct ethylene epoxidation, the elementary reactions and the reaction pathways are speculated as follows:

According to this mechanism, it is assumed that both O2αand Oαare the key active oxygen species existing in the reaction process. For the epoxidation reaction, the gaseous ethylene reacts upon O2αto form ethylene oxide (E-R type mechanism). For the complete combustion reaction, the gaseous ethylene first reacts on Oαto form oxametallacycle intermediate (Ag-O-C2H4); subsequently, the oxametallacycle intermediate interacts with four adjacent Oαspecies respectively to produce carbon dioxide and water (L-H type mechanism). For the kinetic equations, many predecessors’ studies are deduced strictly using the Langmuir adsorption isotherm theory as an origin, indicating that the sum of coverage of all active sites is 100% (i.e., θAg-O+θAg-O2+θAg-O-C2H4+θAg-CO2+θAg-H2O+θV=1). An obvious limitation of these models is that the significant inhibitory influence of 1,2-C2H4Cl2(EDC) is not taken into account. As noted earlier[15], the decomposition of EDC would lead to the formation of adsorbed chlorine, the accumulation of which could result in the formation of bulk silver chloride and in turn could change the topology of the active surface and reduce the catalyst activity. Campbell[16]believed that the addition of chlorine reduces the probability of the fragmentation process by blocking the necessary vacant Ag sites due to the formation of the oxametallacycle intermediate, resulting in an improved selectivity of ethylene epoxidation. Using the near ambient pressure X-ray photoelectron spectroscopy, Rocha[17]demonstrated that EDC not only alters the number of active sites, but also modifies the chemical nature of a given site, resulting in electronic modifications induced by adjacent atoms. The increase in selectivity is related to the rebalance between the electrophilic and the nucleophilic oxygen species, which constitutes the catalyst active sites for the selective and nonselective oxidation pathways. It has also been concluded that competitive adsorption exists between EDC and O2on the catalyst active sites[18]. Li, et al.[19]supposed that EDC is decomposed to form Cl-ions during the epoxidation reaction as evidenced by TPD-MS, XPS, LITOFM, and other analyses. The Clspecies were found to be adsorbed on the surface of silver catalyst in two modes, namely: the strongly and the weekly adsorbed ones. It controls the intensity and concentration of the adsorbed oxygen atoms that play a key role in the complete oxidation of ethylene. It also expels the EO formed and thus decreases the chance of isomerization of EO. Meanwhile, the C1-ions adsorbed on the surface of silver catalyst can react on ethylene to form EDC. The Cl-species can reach a steady state on the surface of silver catalyst as EDC is decomposed and formed so that the catalytic reaction progresses continuously.

By taking into account the above-mentioned phenomena together, the effects of EDC cannot be neglected in the ethylene epoxidation process. So, in order to reflect the function of EDC in the reaction kinetics, the reversible reactions between 1,2-C2H4Cl2and C2H3Cl on the active sites of silver catalyst are considered as follows:

Thus, according to the Langmuir adsorption isotherm theory,

Since the surface reaction is the rate-controlling step, the reaction rates of the epoxidation reaction and the total combustion reaction are deduced as follows:

3 Experimental

3.1 Catalyst description

The A-type silver catalyst used hereby is provided by the Yanshan Branch of Beijing Research Institute of Chemical Industry of China. However, the detailed characterization of the silver catalyst is confidential. The catalyst must be pretreated because of its gradual activity change over time. Therefore, the catalyst is pre-exposed to reaction over 1 000 h prior to the macrokinetic experiment in a wall-cooled packed bed reactor until its steady-state catalytic activity is achieved.

3.2 Equipment

The macrokinetic experiments were accomplished in a Berty reactor manufactured by the Autoclave Engineers (USA). This internal-recycle gradientless reactor had a basket with an inner diameter of 19.0 mm and a height of 24.7 mm. 5.6143 g of the A-type silver catalyst were loaded into the basket. The reaction system was initially pressurized to 2.1 MPa with nitrogen and was remained under that pressure for a while to test the system integrity. Figure 2 demonstrates the apparatus used in the experiments. The reactor basket temperature was adopted as the reaction temperature. The reaction temperature was controlled by an electrical heater surrounding the reactor and monitored by two thermocouples located at the top and bottom of the reactor basket, respectively. Inside the reactor, a motorized fan above the basket was used for the internal recirculation of the reaction gas to minimize the inter-particle mass and heat transfer resistance. The flow rates of the different feed gases were metered by mass flow controllers. The concentrations of the mixed feed gas and effluent gas were measured by an industrial online mass spectrometer. The concentrations of ethylene and carbon dioxide were continuously monitored using an infrared analyzer while the concentration of oxygen was determined in real time by a magnetic oxygen analyzer. The SUPCON DCS system was applied in all the control loops in this experimental system.

Figure 2 Flow scheme of the equipment used for macrokinetic experiments

3.3 Analytical methods

The feed gas mixture and the effluent gas mixture were analyzed continuously by a Prima δB online mass spectrometer equipped with an SEM detector. The macrokinetic assays were carried out under steady-state conditions at a pressure of 2.1 MPa and a temperature ranging between 217.8—249.0 ℃. The feedstock composition (molar fraction) consisted of 15.82%—34.65% of ethylene, 2.55%—7.80% of oxygen, 0.88%—6.15% of CO2and 0.15—2.61 μmol/mol of 1,2-C2H4Cl2inhibitor and 0.14—1.28 μmol/mol of C2H3Cl. The experiments lasted over 4 hours for each macrokinetic measurement. The activity of the catalyst was examined again at the end of the run to confirm that the catalyst worked under a steady state.

Under the above-mentioned conditions, a total of 80 data points of macrokinetic measurements were obtained by the orthogonal experiments. Some of these data are summarized in Table 1.

Table 1 Experimental data of macrokinetics over A-type silver catalyst

4 Results and Discussion

4.1 Estimation of the kinetic parameters

An appropriate kinetic model is critical for understanding the reaction mechanism. The model should be able to give a good mathematical correlation between reaction rates and the influencing factors that include temperature, pressure, flow rate, and concentrations of feedstock components. In order to provide the best correlation between the experimental results and the proposed model for the A-type silver compound catalyzed reactions, all 80 kinetic data points were applied to determine the parameters in the reaction rate equations as shown above. A nonlinear least-squares regression method was employed to find out the best reaction rate expression that could fit all these A-type silver compound catalyzed macrokinetic experimental data. According to the law of mass conservation in the Berty reactor, the epoxidation reaction rate and complete combustion reaction rate were calculated as follows:

By using the calculation of the experimental epoxidation reaction rate in Eq. (14) and the experimental complete combustion reaction rate expressed by Eq. (15), the parameters in all the reaction rate expressions were determined by means of the Simplex Optimal Method. Mathematically, the problem was to find the parameters that gave the minimum deviation between the experimental reaction rates and the corresponding values calculated by the models. This task was performed by a nonlinear programming method in MATLAB. The sum of the squared residuals corresponding to a normal error distribution was used by Eq. (16) as the objective function to perform the minimization:

The values of the pre-exponential factors, activation energies and kinetic constants in Eq. (12) and Eq. (13) are as follows:

4.2 Testing of the rate equations

Although the objective functions are convergent and all the parameters are obtained as shown above, the fit of the model to the experimental results must need further evaluation under four standards:

(1) All the parameters need to satisfy the laws of physics and chemistry. For example, a model should be rejected if it gives implausible activation energy or negative preexponential factors.

(2) Kinetic models need to pass the analysis of the standard deviation using the ρ2correlation index and F test. The confidence intervals of the estimates were calculated by the Matlab program. The kinetic models were considered to be feasible if ρ2>0.9 and F >10F0.01in the statistics test.

(3) The sum of deviation between calculated and experimentally measured reaction rates should be close to zero.

(4) The average relative errors, as defined in Eq. (17), should be small enough (<20%) to meet the needs of the macrokinetic study.

where M is the total number of kinetic experiments.

Using these standards, the macrokinetic models have been tested statistically and the results are summarized in Table 2. The average relative errors of the epoxidation reaction and combustion reaction were 9.22% and 14.73%, respectively. The results showed that the macrokinetic model met all the requirements.

Table 2 Statistical test of macrokinetic models for A-type silver catalys

Figure 3 Residual distribution among main reaction rate of experiment and that of model predictions

Figure 4 Residual distribution among side reaction rate of experiment and that of model predictions

Figure 5 Comparison of main reaction rate among experiment and model predictions

The parity plots for the macrokinetic model shown in Figures 3—6 further demonstrated that the proposed macrokinetic model is the best one to describe the A-type silver catalyst involved reactions. The symmetry of the residual distribution in the above Figures reflected the high suitability of the model. Since the residual is a function of the independent variables in the model, including temperature, pressure, flow rate, and concentrations of feedstock components, a significant correlation between the residual and any independent variable would indicate an inadequacy in the model. In an inadequate model, the residual distribution will be asymmetric. In the parity plots for the macrokinetic model, the residual distribution also indicates that the errors of the experiments are random. No correlations between the residual and any independent variable have been observed. Upon taking into account the above-mentioned facts together, our results support the conclusion that this macrokinetic model of the A-type silver catalyst is not only most suitable for fitting the experimental data, but is also highly acceptable and reliable.

Figure 6 Comparison of side reaction rate among experiment and model predictions

5 Conclusions

A possible reaction mechanism for ethylene epoxidation over the A-type silver catalyst was developed with an emphasis on the adsorbed molecular oxygen (O2α) and the adsorbed atomic oxygen (Oα) which should be the key active oxygen species in the ethylene epoxidation process. The coverage priority of reactive sites of 1,2-C2H4Cl2(EDC) inhibitor was introduced, and based on the reversible reactions between EDC and VC on the active sites of silver catalyst, the hyperbolic macrokinetic models involving the effect of EDC and VC for A-type silver catalyst were established.

The parameters of the macrokinetic models were evaluated by means of the Simplex Optimal Method. Statistical test results illustrated that the macrokinetic models for the A-type silver catalyst could fit the experimental data very well. In addition, it is also highly acceptable and reliable. This study has indicated that the apparent activation energy of the combustion reaction (96.484 kJ/mol) is greater than that of the epoxidation reaction (74.464 kJ/mol).

Nomenclature:

F——molar flow rate, mol/h

k——reaction rate constant, mol/(kg·h·Pa2)

K——adsorption equilibrium constant

m——catalyst mass in reactor, kg

R——production rates of reaction, mol/(kg·h)

Rg——gas constant, 8.314J/(mol·K)

y——mole fraction

P——gas partial pressure, Pa

T——temperature, K

δ——active site

θ——coverage of active sites

Subscripts

1——epoxidation reaction

2——combustion reaction

c——calculated

e——experimental

EDC——1,2-dichloroethane

EO——ethylene oxide

VC——vinyl chloride

in——inlet of reactor

out——outlet of reactor

[1] Klugherz P D, Harriot P. Kinetics of ethylene oxidation on a supported silver catalyst[J]. AIChE J, 1971, 17(4): 856-866

[2] Metcalf P L, Harriot P. Kinetics of silver-catalyzed ethylene oxidation[J]. Industrial & Engineering Chemistry Process Design and Development, 1972, 11(4): 478-484

[3] Petrov L, Eliyas A, Shopov D. A kinetic model of steady state ethylene epoxidation over a supported silver catalyst[J]. Applied Catalysis, 1985, 18(1): 87-103

[4] Borman P C, Westerterp K R. An experimental study of the kinetics of the selective oxidation of ethene over a silver on α-alumina catalyst[J]. Industrial & Engineering Chemistry Research, 1995, 34(1): 49-58

[5] Stoukides M, Pavlou S. Ethylene oxidation on silver catalysts: Effect of ethylene oxide and of external transfer limitations[J]. Chemical Engineering Communications, 1986, 44(1-6): 53-74

[6] Al-Saleh M A, Al-Ahmadi M S, Shalabi M A. Kinetic study of ethylene oxidation in a Berty reactor[J]. Chemical Engineering Journal, 1988, 37(1): 35-41

[7] Lafarga D, Al-Juaied M A, Bondy C M, et al. Ethylene epoxidation on Ag-Cs/α-Al2O3catalyst: Experimental results and strategy for kinetic parameter determination[J]. Industrial & Engineering Chemistry Research, 2000, 39(7): 2148-2156

[8] Carucci J R H, Halonen V, Eranen K, et al. Ethylene oxide formation in a microreactor: from qualitative kinetics to detailed modeling[J]. Industrial & Engineering Chemistry Research, 2010, 49(21): 10897-10907

[9] Kim Y C, Park N C, Shin J S, et al. Partial oxidation of ethylene to ethylene oxide over nanosized Ag/α-Al2O3catalysts[J]. Catalysis Today, 2003, 87(1-4): 153-162

[10] Kilty P A, Sachtler W M H. The mechanism of the selective oxidation of ethylene to ethylene oxide[J]. Catalysis Reviews, 1975, 10(1): 1-16

[11] Stegelmann C, Schiodt N C, Campbell C T, et al. Microkinetic modeling of ethylene oxidation over silver[J]. Journal of Catalysis, 2004, 221(2): 630-649

[12] Gleaves J T, Sault A G, Madix R J, et al. Ethylene oxidation on silver powder: A tap reactor study[J]. Journal of Catalysis, 1990, 121(1): 202-218

[13] Stegelmann C, Stoltze P. Microkinetic analysis of transient ethylene oxidation experiments on silver[J]. Journal of Catalysis, 2004, 226(1): 129-137

[14] ?zbek M O, ?nal I, van Santen R A. Chlorine and caesium promotion of silver ethylene epoxidation catalysts[J].Chem Cat Chem, 2013, 5(2): 443-451

[15] Berty J M. Ethylene oxide synthesis[M]// Leach B E. Applied Industrial Catalysis. Orlando: Academic Press, 1983: 207-238

[16] Campbell C T. Chlorine promoters in selective ethylene epoxidation over Ag (111): A comparison with Ag (110)[J]. Journal of Catalysis, 1986, 99(1): 28-38

[17] Rocha T C R, Haevecker M, Knop-Gericke A, et al. Promoters in heterogeneous catalysis: The role of Cl on ethylene epoxidation over Ag[J]. Journal of Catalysis, 2014, 312:12-16

[18] Zhang Z X, Zhang L R. Reaction mechanism of 1,2-dichloroethane in ethylene epoxidation[J]. Petrochemical Technology, 2003, 32(1): 14-16 (in Chinese)

[19] Li J T, Cai L F. Effect of a micro quantity of 1,2-dichloroethane on the epoxidation of ethylene and characterization of the adsorbed species[J]. Journal of Molecular Catalysis (China), 1992, 6(2): 113-119 (in Chinese)

New Progress in Carbonylation of Low-carbon Olefins

The CAS Lanzhou Institute of Chemical Physics (LICP) has realized hydroaminocarbonylation of various amines with olefins. The high-efficiency synthesis of amide compounds is still one of the important research areas in organic chemistry. The hydroaminocarbonylation of olefins catalyzed by transition metals is an important method for synthesis of amides. However, this kind of reaction can only produce catalytically reactive substances under acidic condition, because the strong basicity of aliphatic amines can inhibit the formation of amides.

The research group headed by Prof. Huang Hanmin of LICP by making use of the weak alkalinity of aminal has designed the palladium catalyzed hydroaminocarbonylation of olefins involving aminal and water to selectively obtain a series of aliphatic amides. Furthermore on the basis of previous work they have also designed a cocatalyst system involving small organic molecules and transition metals to realize hydroaminocarbonylation of olefins and amines to selectively yield chain-like amides. This catalytic system has solved a scientific gut issue associated with hydroaminocarbonylation of olefins.

Technology for Coal Hydrogenation-Gasification with Co-production of Methane and Aromatics Developed by Xin’ao Group

The Xin’ao Group has developed the technology for hydrogenation and gasification of coal with co-production of methane and aromatics in order to realize the effective and clean conversion of coal to natural gas coupled with co-production of aromatic components with high add-value. Currently a pilot unit with a coal throughput of 5 t/d has been set up, and the company plans to construct an industrial demonstration unit with a capacity of 400 t/d within the next three years.

According to this technology the pulverized coal at a temperature of 800—1 000 ℃ and under a pressure of 5-10 MPa reacts on hydrogen to yield methane, aromatic oil products and semi-coke. The test results have shown that the methane content in gas is greater than 80% and the aromatic oil yield is in the range of 10%—15%. It is told that compared to the traditional coal tar products, this technology can provide products with high aromatic components that are mainly composed of benzene and naphthalene. At present besides the process for toluene alkylation with methanol, the technique for benzene alkylation with methanol is also technically viable. The mixed aromatics obtained from alkylation of benzene with methanol can be used as the blending component for producing high-octane gasoline or can be further used for manufacture of para-xylene.

date: 2015-02-12; Accepted date: 2015-04-19.

Professor Li Jianwei, E-mail: lijw@ mail.buct.edu.cn.