Reaction Mechanism of Ethylene Oxide at Various Oxygen / Ethylene Oxide Ratios in an RF Cold Plasma Environment

An innovative method was used to simulate ethylene oxide (EO) oxidation in an RF plasma reactor. The objective of this work was to simulate the stable species mole fraction profiles measured in a flowing plasma system at constant temperature and pressure. The mechanism involved participation of 36 species in 140 elementary reactions. Sensitivity analysis was also performed to identify the order of significance of reactions in the mechanism of the model s predictions. The results show that the main reactions for EO decomposition changed with a varying O2/EO ratio in the plasma system. That is to say, the most important reaction to the O2/EO ratio of zero was the electron dissociation reaction of EO, C2H4O + eCH3CHO + e-. While, the most influential reaction for EO decomposition at O2/EO ratio of 5.0 was the formation reaction of HO2, which forms OH radicals, then enhances the decomposition of C2H4O by the reaction, C2H4O + OH = C2H3O + H2O.


INTRODUCTION
Ethylene oxide (EO) is widely used in chemical plants and hospitals to synthesize chemical intermediates and sterilize contaminated instruments.Controlling EO emission is of great importance, since it can be emitted into the environment through processing and ventilation of sterilization equipment, causing exposure risks to humans through the inhalation of polluted air..In a previous study (Liao et al., 2001), an RF plasma reactor with glow discharge was used to decompose EO-containing gas.The effects of plasma operational parameters for the EO decomposition, the profile of final products, and the fraction of total-carbon converted into CO 2 and CO were investigated in an EO/O 2 /Ar plasma system.And the O 2 /EO ratio, varying from 0 to 5, was chosen to represent the reaction conditions of oxygen-lean and oxygen-rich, and to see how oxygen influences the product profile in the EO/O 2 /Ar plasma system.
According to these results, the amount of oxygen needed for proceeding a suitable plasma reaction can be chosen, while the EO is drawn out from a closed reaction system, or a sterilization equipment.
Various types of plasma models have been proposed to provide an interpretation of diagnostic measurements and to understand the effects of operating parameters (Lister, 1992;Meyyappan and Govindan, 1995;Bose et al., 1999).Basic components for a complete plasma model should include: 1) an electric model describing discharge physics for charged species; 2) a plasma chemistry model, including fluid flow; and 3) a surface model, describing reactions at the substrate and chamber wall.While plasmas of gas mixtures are currently used in numerous industrial applications, modeling studies usually focus on plasma of a single gas.Few modeling studies of mixed-gas plasmas exist due to their complexity.This paper describes an innovative method used to simulate the reaction mechanism of EO in the study of Liao et al (Liao et al., 2001).The mechanism involves 36 species in 140 elementary reactions.
The objective of our work was to simulate the stable species mole fraction profiles measured in flowing plasma system at constant temperature and pressure.The Boltzmann equation was solved first to calculate the electron energy distribution function (EEDF).Mass and energy conservation equations were then solved to calculate the electron temperature and electron density as a function of feed composition.
Combining the neutral reactions (Pitz and Westbrook, 1986;Dagaut et al., 1996) with the electronneutral reactions, we constructed a complete mechanism for EO decomposition in the RF plasma reactor.
By applying this mechanism and running the CHEMKIN II package (Kee et al., 1993), we obtained the calculated concentrations of products in the plasma reactor outlet.To test the accuracy of the model, the calculated results were compared with experimental measurements taken from literature (Liao et al., 2001).Via the sensitivity analysis by using SENKIN II program (Lutz et al., 1992), we identified the rank order of significance of reaction in the mechanism.

PLASMA MODEL
The plasma model includes three parts described as model assumption, charged species model and neutral species model.

MODEL ASSUMPTION
The reactor used in this study is a cylinder-type RF power supply reactor.In the operational range of this experiment, the species and their behavior in the plasma are assumed as follows.The species in the EO/O 2 /Ar plasma will include charged species, neutral radicals and stable molecules.
A. In the plasma reactor, the ambipolar diffusion velocity of charged species is higher than that of convection velocity, and the ion-electron recombination in plasma volume is negligible.
Therefore, the electron density distribution in cylindrical plasma of radius R and length L is: where J 0 is the zero-order Bessel function, and n e0 is the maximum electron density along the axis of the plasma (Roth, 1995).The average value of electron density distribution is used in the model to calculate the electron-neutral reaction rate.
B. The convection velocity of neutral species is higher than that of the diffusion velocity; therefore, the neutral species in the plasma reactor can be assumed as plug flow reactor (PFR).In this model, the concentration of neutral species in the plasma reactor is a function of the position in the axis of reactor length.
C. This plasma reactor is an isothermal reactor, both gas and electron temperatures are homogeneous.

CHARGED SPECIES MODEL
In this model, both mass and energy balances of charged species will be used to calculate the electron temperature (T e ) and electron density (n e ).Both Te and ne will be used for the calculation of reaction rate constant of electrons.
In the case of EO/O 2 /Ar plasma, the charged species in this model will be Ar Therefore, the formation rate of electrons is as follows.
where f j j the inlet gaseous mixture.
After formation, the positive ions will diffuse to the sheath in both radial and axial directions by ambipolar diffusion and then be neutralized by electrons (Lieberman and Lichtenburg, 1994;Roth, 1995).The ko (frequency factor) and Ea (apparent threshold energy) are calculated from the electronimpact-ionization of mixed feeding gas by Arrhenius plot (details in Section 2-1-3).Knowing ko, Ea, P, Tg, a , R and L, we can calculate the electron temperature Te.

Ar e Ar
In the RF plasma, when the operational pressure is between 10 and 30 torr, the electron temperature is much higher than the gas temperature.The gas temperature at the outlet of the plasma reactor was measured and used as the input parameter for this model.Also, we obtained the electron temperature by using the electron mass balance.Therefore, we can find the electron density (n e ) by an energy balance equation as follows.
0= P RF -P ev (16) is the fraction of total input power absorbed by the plasma reactor.P RF is the applied RF power and P ev is the energy consumption rate of electron due to the electron-neutral collision.
In the plasma reactor, the collision between electron and neutral gas molecule can be divided into elastic and non-elastic collision.The non-elastic collision included ionization, dissociation and vibrational excitation reactions, etc.Therefore, the energy consumption rate in the plasma reactor can be expressed as follows.
P RF =P ela +P ion +P dis +P vib =n e (P' ela +P' ion +P' dis +P' vib ) where P i ' is the energy consumption rate of electrons i llision.
Bolsig software (Wei et al., 2000;Chen et al., 1999) can find the electron energy distribution function (EEDF) and the energy consumption of unit electrons at different types of collision reactions, as well as the collision frequency at different E/N values.

NEUTRAL SPECIES MODEL
When the EO/O 2 /Ar mixing gas enters the plasma zone of the reactor, it can form neutral species such as free radicals and stable molecules.The composition of gas stream in the plasma zone will vary with the length in the gas flow direction.If we neglect the diffusion phenomenon of gas molecules, the neutral species can be described by using the PFR model.
where F i is the molar flux of i ; Z is the axial coordinate of cylinder-type reactor; A is the cross section of cylinder-type reactor; ij is the stoichiometric The distribution of electron energy in the plasma reactor can be calculated by Bolsig software as a function of E/N (reduced field; E is the electric field and N is the neutral density).After EEDF are calculated by running Bolsig software, the rate constants of electron-neutral reactions are obtained by calculating the following equation.
where k is the rate constant of electron reaction; is the electron energy; is the velocity of electrons; ( ) is the distribution function of the cross section; and f( ) is the distribution function of electron energy.
For each electron-neutral reaction, the dependence of rate constant on electron temperature is fitted to the Arrhenius form (k=k 0 e -Ea/Te ).

Modeling Description
The fundamental simulation for neutral reaction in the plasma reactor is based on the following principles (Kee et al., 1993): 1) Thermochemical theory; 2) Reaction kinetics, including Transition State Theory; and 3) The proper and accurate thermodynamic data.
The mechanism of the elementary reactions set describing the decomposition of EO are presented in this paper.This elementary reactions set consists of 140 elementary reactions and the available data obtained from previous references (Pitz and Westbrook, 1986;Dagaut et al., 1996).The rate parameters containing three Arrhenius coefficients (pre-exponential factor, A; temperature exponent, n; and activation energy, Ea) for the forward reaction paths are also based on those studies.Reverse reaction rates are calculated from a detailed balancing between the forward and reverse rates through the use of the equilibrium constant.In all cases, rate parameters are consistent with reaction thermodynamics.
Combining these neutral reactions with the previous electron reactions, we can construct a complete mechanism for EO decomposition in the plasma reactor.Calculations can be conducted assuming a system where the pressure and temperature are constant, and performed over a cylindrical domain with a diameter of 4.14 cm and a height of plasma length.The pressure of the reaction in the RF plasma reactor was set to 20 Torr and the initial temperature was set as the gas temperature, which is measured at the plasma reactor .

SENSITIVITY ANALYSIS
The major reaction channels responsible for the decomposition and formation of species were identified by sensitivity analysis.Sensitivity analysis involves quantitative information in how the rate coefficients affect the reaction conditions.Sandia SENKIN computer code was used for the calculation of sensitivity coefficients (Dean et al., 1991).Reaction pathways responsible for the decomposition and formation of species were then determined from the calculations of individual reaction rates and first-order normalized where Z j is the concentration of species j, and A i is the pre-exponential constant of the forward branch of the ith elementary reaction (Won and Bozzelli, 1992).

KINETIC PARAMETERS OF ELECTRON-NEUTRAL REACTIONS
In this study, the rate constants were obtained in two ways: 1) by estimation using the threshold energy, maximum cross-section and averaged electron temperature; and 2) by calculation integrating the total cross-section function and the electron energy distribution function.
The total cross-section for the dissociation reaction of CH 4 collided with electrons can be found in the Winters et al. (1975Winters et al. ( , 1979)).By applying the calculation process described in the previous section, we get the total dissociation rate constant.Then, the rate ratio of three following CH 4 dissociation reactions suggested by Fan et al. (1999) This implies that the rate constant ratio of the above three reactions are 2.5:2.5:1.By using this ratio, individual rate constants can be obtained.Note (1) Reaction mechanism and rate constants expressed as: k A T n exp(-Ea / RT). ( 2 Combining the neutral reactions (Pitz and Westbrook, 1986;Dagaut et al., 1996) with the electronneutral reactions, we construct a complete mechanism, shown in Table 1, for EO decomposition in this RF plasma reactor.The rate constants of electron-neutral impact dissociation reactions at various O 2 /EO ratios are shown in Table 2.

COMPARISON OF THE SIMULATED RESULTS WITH THE EXPERIMENTAL DATA
Calculations were conducted assuming a system where the pressure and temperature are constant, and were performed over a cylindrical domain with a diameter of 4.14 cm and a height of plasma length.The pressure of the reaction in the RF plasma reactor was set to 20 Torr and the initial temperature was set as the gas temperature, which was measured at the outlet of the plasma reactor.By applying the above mechanism and running the CHEMKIN II package (Kee et al., 1993), we obtained the calculated concentrations of residual reactant and products in the outlet of the plasma reactor.
Comparison of experimental and calculated decomposition fraction of EO for various EO/O 2 ratios are shown in Figure 1.It shows that the simulation results are in good agreement with the experimental data.Figure 2 shows the model prediction results of product concentrations.It also shows that the model prediction results agree well with the experimental results.Figure 2(b) shows the model predictions for carbon monoxide is overestimated at 0 of O 2 /EO ratio.A possible reason for this overestimate of carbon monoxide is the lack of reactions in our mechanism to species of higher molecular weight, such as polyhydrocarbon species and soot.In this circumstance, the carbon contained in the reactant might lead to the formation of CO product.This is in agreement with predictions on polyaromatic compound formation (Stein and Fahr, 1985).At some high temperature levels in EO/Ar plasma decomposition, the radicals and unsaturated molecules begin to combine leading, ultimately, to soot or highly carbonized structures.Krestinin (2000) and others (Vuitton et al., 2001) think that the soot may be formed via the acetylene pathway; i.e. polyyne model and the diacetylene, C 4 H 2, is recognized as a soot precursor.
Without thermo data for polyyne, C 2n H 2 , the model was not introduced in this work.However, the concentration of carbon monoxide between the experimental and modeled data was still within 18% of difference.

SENSITIVITY ANALYSIS
According to the sensitivity analysis method described, we can obtain the sequence of the sensitivity coefficients in the plasma reaction.In addition, the major reaction channels responsible for the decomposition and formation of species have been identified.Those results are presented below.

FIVE MOST IMPORTANT REACTIONS RELATED TO C 2 H 4 O IN THE EO/AR PLASMA SYSTEM
The most important five reactions involving the decomposition of C 2 H 4 O, in sequence, are shown in Table 3.Each sensitivity coefficient value, which is responsible for the corresponding reaction, follows each reaction.

THE REACTION PATHWAYS IN THE EO/AR AND EO/O 2 /AR PLASMA SYSTEM
By summarizing the sequence of formation and dissipation reactions of the major species measured in the plasma reactor, we built up two reaction pathways, one for the EO/Ar plasma system and the other for the EO/O 2 /Ar (O 2 /EO=5.0)plasma system.They are shown in Figure 3 and Figure 4, respectively.
As seen in Figure 3, when no O 2 is added, the major decomposing route for EO, C 2 H 4 O+e -

CONCLUSIONS
Results showed that the calculated decomposition fraction of EO and concentrations of products agree very well with the experiment data.In addition, the decomposition reactions for EO were changing with varying O 2 /EO ratios in the complex plasma system.The most important reaction with an O 2 /EO ratio of zero was the electron dissociation reaction of EO, C 2 H 4 O + e - CH 3 CHO + e -.However the most significant reaction with an O 2 /EO ratio of 5.0 was the formation reaction of HO ) A unit = cm 3 /(mol sec) for bimolecular reactions, 1/sec for unimolecular reactions.s. (3) Ea unit = cal/mol.(4) T unit = k.(5) Est.= Estimated in this study.(6) Cal.= Calculation in this study.

Figure 1 .
Figure 1.The comparison of calculated data with experimental data on the decomposition fraction of EO at various O 2 /EO ratios (input power = 30 W; operational pressure = 20 torr; Eo feeding concentration = 2 total gas flow rate = 100 sccm).

Figure 2 .FIVE
Figure 2. Comparison of modeling data with experimental data on the product concentrations at various O 2 /EO ratios (input power = 30 W; operational pressure = 20 torr; Eo feeding concentration = 2% total gas flow rate = 100 sccm).

Table 1 .
The elementary reaction mechanism for the EO/O 2 /Ar plasma.
Table 3 also presents results from the sensitivity analysis on the model for the relative importance of C 2 H 4 O decomposition reactions in EO/Ar plasma system.Both equations of C 2 H 4 O+e-=CH 3 CHO+e-and C 2 H 4 O+H=C 2 H 3 O+H 2 with significant negative sensitivity coefficients are the most important decomposition reactions over the operation regime.In a whole view, Table 3 shows that those two reactions are the dominant decomposition paths for C 2 H 4 O in the condition in which no O 2 was added.However, both equations of CH+H=C+H 2 and CH 3 +H+M=CH 4 +M with larger positive sensitivity coefficients will inhibit the decomposition of C 2 H 4 O.The reason is that the reactions CH+H=C+H 2 and CH 3 +H+M=CH 4 +M will compete with H with the reaction C 2 H 4 O+H=C 2 H 3 O+H 2 .So, enhancing the reactions of CH+H=C+H 2 and CH 3 +H+M=CH 4 +M will weaken the C 2 H 4 O decomposition reaction C 2 H 4 O+H=C 2 H 3 O+H 2 .

Table 3 .
The most important five reactions related to C 2 H 4 O in EO/Ar plasma.

Table 4 .
The most important five reactions related to C 2 H 4 O in EO/O 2 /Ar plasma.
Table 4 presents results from sensitivity analysis on the model for the relative importance of C 2 H 4 O decomposition reactions in the EO/O 2 /Ar plasma system.Although the equation, H+O 2 +M=HO 2 +M, seems not to have much relation with the decomposition of EO, it is the most important decomposition reaction of EO over the operation regime.The reason for this is that the reactions H+O 2 +M=HO 2 +M first produce more HO 2 radicals, then HO 2 proceeds with the reaction HO 2 +H = OH+OH further to form more OH radicals; finally, OH radicals participate in the reaction, C 2 H 4 O+OH=C 2 H 3 O+H 2 O, to decompose more C 2 H 4 O.To conclude, the reaction H+O 2 +M=HO 2 +M enhances the decomposition reaction of EO =CH 3 CHO+e -, produces the intermediate, CH 3 CHO.Then, CH 3 CHO dissipates through two major routes, one makes it form CH 3 and CHO radicals, and the other leads to the formation of the other intermediate, CH 3 CO.Furthermore, the CH 3 radical goes around a reaction loop to form a significant amount of C 2 H 6 and a perceivable amount of C 2 H 4 and C 2 H 2 , or terminates with radical H to form a small amount of CH 4 .As seen in Figure4, when a large amount of O 2 is added, the major decomposing route for EO shifts to the reactions C 2 H 4 O+H=C 2 H 3 O+H 2 and C 2 H 4 O+OH=C 2 H 3 O+H 2 O.Then, C 2 H 3 O dissipates to form the stable products, CO 2 and H 2 O. Furthermore, because of the loop reaction for CH 3 to form C 2 H 6 has been stopped.So, instead of forming C 2 H 6 , CH 3 forms the intermediate CH 2 O; then it reacts further to form the final product CO 2 .The reaction pathways of EO/ Ar plasma system.
2 , which forms OH radicals further, then enhances the decomposition of C 2 H 4 O by the reaction, C 2 H 4 O + OH = C 2 H 3 O + H 2 O.The detail reaction pathways for decomposition of EO at various O 2 /EO ratios in the RF plasma reactor tell us that the CH 3 radical goes around a reaction loop to form significant amount of C 2 H 6 at zero of O 2 /EO.And the loop reaction for CH 3 to form C 2 H 6 has been stopped at 5.0 of O 2 /EO, and instead of forming C 2 H 6 , CH 3 forms the intermediate CH 2 O, then it reacts further to become the final product CO 2 .