Applied Catalysis, 24 (1986) 145-161 145

Elsevier Science Publishers B.V., Amsterdam-Printed in The Netherlands

KINETICS OF ETHYLENE OXIDATION OVER A SILVER CATALYST IN THE PRESENCE OF DI-

CHLOROETHANE

L. PETROV, A. ELIYAS and D. SHOPOV

Institute of Kinetics and Catalysis, Bulgarian Academy of Sciences, str. Acad. G.
Bonchev, bl. 11, Sofia 1113, Bulgaria.

(Received 13 November 1985, accepted 13 March 1986)

ABSTRACT
The kinetics of ethylene oxidation over a silver catalyst in the presence of
dichloroethane has been studied by the circulation flow method with a gradientless
reactor. Dichloroethane inhibits the total oxidation reaction to a greater extent
than the partial oxidation reaction. In this way it promotes the selectivity for
ethylene oxide. The promoter was added to the feed gas stream in trace quantities -
0.2 to 28 ppm. The experiments were carried out under steady-state conditions, at
atmospheric pressure and in the interval 240 - 292°C. Additional factors that could
influence the selectivity, temperature, feed composition and degree of conversion
of ethylene, were also investigated. It was established that the influence of
the conversion is negligible. A linear model of the influence of the dichloroethane
concentration on the selectivity could be applied to most series of the experiments.
The selectivity-promoting effect of dichloroethane is stronger at higher temperatures.
The selectivity is most sensitive to promoter concentration when the feed gas
stream has the composition 50% 02, 25% C2H and 25% Ar. A kinetic model consisting
of two rate equations - for the partial an ! for the complete oxidation is proposed,
including a dependence on the partial pressure of dichloroethane. The model describes
satisfactorily the influence of all factors on the selectivity. The role of di-
chloroethane is to inhibit the dissociative adsorption of oxygen on the silver
surface.

INTRODUCTION
The partial oxidation of ethylene over silver catalysts is always accompanied
by the highly exothermic complete oxidation of ethylene:

-1
I. C2H2 + l/2 O2 = C2H40 + 28 kcal mol
II. C2H4 +30 = 2 co2 + 2 H20 + 316 kcal mol -1
2

The second reaction is undesired as it results in losses of expensive ethylene

and it hampers the maintenance of the optimum catalyst bed temperature. The heats
of epoxidation and combustion at 298°K were reported previously [I]. Suppression
of the parasitic reaction, i.e., increased selectivity for ethylene oxide, is
achieved either by adding compounds or electronegative elements, Cl, S, Se, Te
and P, to the catalyst upon its preparation or by introducing trace amounts of
chlorinated organic compounds into the feed gas stream [2,3] or both. The industrial

0166-9834/86/$03.50 0 1986 Elsevier Science Publishers B.V.

146

gas-phase promoter employed is usually 1,2-dichloroethane, C2H4CL2 (WE), and its

concentration is normally 1 ppm.
Silver catalysts, whose selectivity is promoted by chlorine and its compounds,
have been the subject ot numerous investigations [Z-15]. Force and Bell [3] reported
that the presence of DCE causes a reduction in the oxidation rate and an increase
in the selectivity. In their opinion, this is due to a reduction in the adsorption
of ethylene. Kamensky et al. [43 proposed kinetic equations involving parameters
which are functions of the content of selectivity promoier. The same authors
conclude [53 that the activity and selectivity of an industrial siJver catalyst
increase when its surface is covered by l-7% chloride. Ayame et al. [6] claimed
that by the addition of sodium chloride to a silver catalyst the selectivity is
increased to 85-87%. Ionov et al. [7] described the kinetics of ethylene epoxidation
over chlorine-promoted silver by a model which is a special case of a more general
model. Kamensky and Petrov [8] proposed a generalized kinetic model for both
promoted and unpromoted silver catalysts that could be of use in the modelling
of industrial reactors. Park et al. [S-IO] reported a selectivity of 94% for a
chlorine-promoted silver catalyst. Talaeva et al. [IZ] proposed a method to account
for the influence of the DCE concentration in the gas phase on the rates of epoxi-
dation and combustion. Kharson et al. [I31 proposed a kinetic model of ethylene
oxidation over a silver catalyst promoted with calcium chloride. Despite these
studies, the quantitative description of the effect of DCE on the selectivity is
unsatisfactory.
The aim of the present paper is to study quantitatively the inhibiting effect
of DCE on the epoxidation and combustion of ethylene over a silver catalyst,
described in [16], as well as its promotion of the selectivity. Other factors that
could possibly influence the selectivity have also been examined, in order to
propose a kinetic model of the process in the presence of DCE.

EXPERIMENTAL

Apparatus
The kinetic measurements were conducted in a glass circulation flow system
with a gradientless reactor at atmospheric pressure. The system and the gas
chromatographic (GC) analysis have beendescribed in detail [16]. The only difference
in the apparatus is the presence of an additional gas line for DCE dosing. The
addition of DCE in the desired trace quantities (of the order of 1 ppm) to the
feed gas stream causes considerable experimental difficulties when the flow-rates
of the reagents are small. A traditional saturator using Ar bubbling through liquid
DCE resulted in DCE concentrations of much greater than 300 ppm even at tempera-
tures below 0°C. At such concentrations the catalyst was completely poisoned,
both partial and total oxidation were inhibited and no conversion of ethylene was
observed. Another saturator was much more successful (Figure 1). A capillary of
0.1 ccm (C) graduated in microlitres was filled with liquid DCE and immersed in
FIGURE 1 Dosage of the feed gas stream with I,Z-dichloroethane. M: mass flow
controller, S: flow splitter, F: four-way valves, C: capillary, T: thermostat,

FM: flow meter.

a thermostat (T) maintaining a constant temperature of about 10°C. Argon was passed
over the surface of the DCE at a very low flow-rate, 10 or 20 ml per hour. Such
low flow-rates were achieved by means of a flow splitter (S) and a mass flow
controller (M) and measured by a flow meter (FM). The four-way valves (F) are used
to isolate the DCE line from the reactor or to direct the argon flow through the
flow meter in order to measure its rate. The consumption of DCE, carried by the
argon flow into the reactor, is thus easily measured from the lowering of the
liquid level in the capillary over a definite period of time. By multiplying the
volume of liquid DCE consumed by its specific density and dividing by the time
interval during which this volume was consumed, we obtain the DCE flow-rate.
The DCE concentration in the feed may be calculated by dividing the DCE flow-rate
by the sum of all flow-rates

"dce
C
dce =
(2)
u. + ue + u,

where Uo, Ue, U, and Udce are the flow-rates of oxygen, ethylene, argon and
dichloroethane, respectively. In fact Udce in the denominator is neglected as

Udce << uo + u, + u,.

Conditions
The concentration of DCE in the feed gas flow was varied from 0.2 to 28 ppm.
At these concentrations no conversion of ethylene was observed at 21O"C, so the
kinetic measurements were carried out at catalyst bed temperatures of 240, 264
148

and 292°C (our intention was to reproduce, if possible, the conditions of our
previous study in the absence of DCE [16]). At each temperature three different
feed compositions were studied:

I. 25.0% o*, 50.0% C2H4, 25.0% Ar + C2H4C12 (ppm)

II. 33.3% o*, 33.3% C2H4, 33.4% Ar + C2H4C12 (ppm)
III. 50.0% o*, 25.0% C2H4, 25.0% Ar + C2H4C12 (ppm)

DCE will be called the "promoter", in the sense of "promoting the selectivity".
In the first set of experiments the argon flow-rate over the liquid DCE in the
capillary was IO ml h-'; then all the experiments were repeated at 20 ml Ar h-l.
The two flow-rates resulted in different promoter concentrations in the feed,
At a fixed catalyst bed temperature, feed composition and argon flow-rate
-1
through the saturator, the contact time was varied from 158 to 1901 g-cat. h mol .
The reaction proceeding in the kinetic region was monitored as in [19]. The
system reaches a steady state in half an hour, but the stationary catalyst activity
in the presence of DCE is achieved only 15-20 hours after the beginning of DCE
dosage. The inhibiting effect of DCE is reversible - IO-15 hours after dosage
the initial high catalytic activity is restored completely.
The degree of ethylene conversion varied from 2 to 45%.

Catalyst
The catalyst had already been characterized [16]. The fixed bed reactor
contained 25 g of catalyst. It was established by oxygen chemisorption at 200°C
that the area of the silver surface was 0.05 m2 g-cat -' which is 35.7% of the
2 -1
total specific surface area of the catalyst -0.14 m g .

Reproducibility of the experiments

For each set of fixed temperature, feed composition and contact time, after the
system had reached a steady state and the catalyst a stationary activity, a
series of IO GC analyses of the converted gas mixture was carried out. The average
concentration for each substance was calculated as well as the deviation for each
GC analysis. The deviation did not exceed 8%. Some of the experiments were repeated
weeks later to test their reproducibility - the average deviation was about 11%.

RESULTS
The calculations of the partial pressures of the reactants and products, the
rates for substances and the rates along the two linearly independent routes as

well as the extent of conversion and the selectivity from the chromatograms

required practically no modifications of the computer programs andweredescribed

previously [16].
FIGURE 2 Concentrations of products versus the contact time at 264°C and feedstock
composition: 25% 02, 50% C2H4, 25% Ar. 1 - ethylene oxide concentration for the
DCE-promoted catalyst, 2 - ethylene oxide concentration for the unpromoted catalyst,
3 - carbon dioxide concentration for the promoted catalyst, 4 - carbon dioxide
concentration for the unpromoted catalyst.

TABLE 1
Product concentration decrease over the promoted and unpromoted catalysts.

-1
T Flow-rates/ml min Promoted cat. Unpromoted cat. Concn. decrease

/"C o2 C,,/% Cc/% Ceo/% Cc/% AC,,/% Act/%

C2H4

240 5 10 3.01 3.75 5.79 7.68 48.0 51.2

240 10 10 2.86 4.93 4.04 9.55 29.2 48.4
240 10 5 3.85 5.20 5.08 7.31 24.2 28.9
264 5 10 4.72 6.74 5.76 11.74 18.1 42.6
264 IO IO 3.77 5.28 7.30 12.52 48.4 57.8
264 10 5 5.23 7.36 8.67 14.96 39.7 50.8
292 5 10 3.19 3.52 3.48 12.69 8.3 72.3
292 10 IO 2.29 3.52 3.97 14.83 42.3 76.3
292 IO 5 6.35 8.72 9.52 22.52 33.3 61.3

Figure 2 presents a comparison of the kinetic curves for the DCE-promoted and
the unpromoted catalysts (W is the mass of catalyst in grams, Fe the ethylene
flow-rate in moles per hour) for one series of experiments. It is seen that under
the same conditions (temperature, 0 2:C2H4 feed ratio and contact time) the pro-
moted catalyst is less active for both partial and complete oxidation, so in
this sense DCE is an inhibitor. Nevertheless the decrease in ethylene oxide con-
centration is smaller compared with that of carbon dioxide (see also Table 1) -
tg:;:::::::::

40 42 44 46 40 SO 40 42 44 46 46 SO

C’b” !W?TIAL PRESSURE , I$ X flw2(afm) C$4 PARTIAL PRESSURE, P,x10w2 (dm)

FIGURE 3 Rates of partial oxidation of ethylene for the WE-promoted catalyst

R(l)p and the unpromoted catalyst R(1)" as a function of the ethylene partial
pressure Pe at 264°C and feedstock composition 25% 02, 50% C2H4 and 25% Ar.

FIGURE 4 Rates of complete ethylene oxidation for the DCE-promoted catalyst R(2)p
and unpromoted catalyst R(2)" as a function of the ethylene partial pressure Pe
at 264°C and feedstock composition 25% 02, 50% C2H4 and 25% Ar.

the inhibiting effect of DCE is therefore selective and leads to selectivity

promotion. At any contact time the selectivity of the promoted catalyst for ethylene
oxide is higher than that of the unpromoted catalyst. Figure 2 as well as Table 1
illustrate the promoting effect of DCE on the selectivity.
The concentration decrease aCi in Table 1 has been calculated in the following

way

C; - cy
.I00
ACi = (3)
CY

where the index "u" denotes the'unpromoted catalyst" and the index "p" the pro-
moted catalyst; i = 1 is ethyiene oxide (eo), i = 2 is carbon dioxide (c) in
Table
II .

If we consider the rates R(1) of the partial oxidation and R(2) of the complete
oxidation as functions of the ethylene partial pressure at a fixed temperature
and feed composition and compare these dependences for the promoted and the un-
promoted catalysts, then we find that the decrease in rate of the complete oxida-
tion is greater than that of the partial oxidation - Figures 3 and 4. This effect
FIGURE 5 Selectivity S versus the degree of conversion of ethylene AX,: l -

240°C and feed composition 25% 02, 50% C2H4, 25% Ar, 5 ppm DCE. A - at 264°C
and feed composition 33.3% 02, 33.3% C2H4, 33.4% Ar and 9 ppm DCE. 0 - at 292°C
and feed composition 50% 02, 25% C2H4, 25% Ar, 2 ppm LICE.

FIGURE 6 Selectivity as a function of DCE concentration at 24O*C and feedstock

composition 25% 02, 50% C2H4 and 25% Ar: a - experimental selectivity Se. 0 -
model selectivity Sm.

is'more marked at higher ethylene partial pressures , which correspond to higher

promoter concentrations. In the presence of DCE even at the highest temperature

(292°C) the rate of the partial oxidationR(l) insomeexperiments is still higher

than that of the complete oxidation R(2). It has been shown [163 that in the
absence of this promoter the rate of the complete oxidation increases faster with
increasing temperature and at higher temperatures exceeds the epoxidation rate.
This fact is further evidence in support of the effectivity of the selectivity
promoter.
One of the main objects of this work was to study the influence of different
factors on the selectivity of the catalyst for ethylene oxide. This is a point

of great practical importance. The factors that could be expected to influence

the selectivity are the promoter concentration, the temperature, the feed composition
and possibly the degree of ethylene conversion AX,. As in our previous work C161,

the degree of conversion proved to be of no importance - this time for the,

selectivity in the presence of the promoter (DCE) - Figure 5. Thus the reaction

picture is simplified to a certain extent as we could fix any two of the remaining
factors and study the influence of the third on the selectivity.
The analysis of the experimental data at various feed ratios 02:C2H4 in the
152

TABLE 2
Selectivity at different feed ratios 02:C2H4 and at fixed temperature and DCE
concentration

T/"C DCE concent- Feed ratio Experimental Model

ration/ppm : C2H4 selectivity/% selectivity/%
O2

292 2 2 1 55.6 54.1

292 2 1 1 54.5 54.5
292 2 1 2 55.5 54.2
240 5 2 1 53.8 52.5
240 5 1 1 48.0 52.5
240 5 1 2 52.0 52.1

presence of the DCE-promoted catalyst at constant temperature and promoter con-

centration was very difficult. Two series of selectivities are presented in Table
2 - the influence of the feed composition on the selectivity of the chlorinated

catalyst is not as clearly expressed as was the case with the unpromoted catalyst
C161.
The most important factor influencing the selectivity was the concentration

of the promoter DCE in the feed gas stream. Using two different flow-rates of Ar
through the saturator - 10 and 20 ml h-' and a total flow-rate through the reactor

varying from 1200 to 10800 ml h-' , we obtained DCE concentrations from 0.2 to 28
ppm. Figure 6 shows the increase in selectivity S with increasing DCE concentration

in the feed gas stream Cdce at constant temperature and feed ratio 02:C2H4 for
one series of experiments. It was established that for most series of experiments

this dependence may be approximated to:

S = a Cdce + b

The slopes "a" and theintercepts "b" of the plots depend on the reaction tempera-

ture and unexpectedly (taking into consideration the data from Table 2) also on
the feed composition. The physical meaning of "b" is the selectivity in the absence

of DCE:

S = b (Cdce = 0) (5)

We can ascribe also a physical meaning to "a" - the sensitivity of the selectivity

to DCE concentration. The sensitivity was highest at 292°C (for two of the three
feed compositions) and at a feed ratio 02..C2H 4 = 2:l (for all temperatures) - under

these conditions the slopes have the highest values. These values for the model

selectivities are presented in Table 8. This result is in accord with our previous

results C161.
 153

TABLE 3
Selectivity at different temperatures, and a fixed DCE concentration and feed
ratio 02:C2H4

T/"C DCE concent- Feed ratio Experimental Model

ration/ppm : C2H4 selectivity/S selectivity/%
O2

240 3 2:l 47.4 50.7

264 3 2:l 52.6 53.5
292 3 2:l 58.1 56.1
240 4 I:1 46.6 51.5
264 4 I:1 51.8 54.3
292 4 I:1 56.9 57.1

The influence of temperature on the selectivity at a fixed DCE concentration

and 0 :C H feed ratio is shown in Table 3. The data in this table prove con-
2 24
vincingly that the selectivity promoting effect increases with increasing tempera-
ture. A possible explanation is the increase in Cl - AES signal up to about 650°K
[22] - our experiments were carried out in the temperature interval 240-292'C.

KINETIC MODEL

A computer program, based on the Nelder-Mead algorithm [16], was used for the
selection of the best kinetic equation for each route, as well as for the estimation

of the kinetic constants. The sum of the squared residuals (absolute errors) was

used as a minimization criterion. To the best of our knowledge, none of the previous-
ly published kinetic models of ethylene oxidation includes a dependence on the
DCE partial pressure, although chlorine and its compounds have been widely used
as promoters of selectivity [Z-15,17]. The presence of DCE or another Cl-containing
compound, used as promoter of selectivity, is reflected in the values of the kinetic
constants, compared with those values for the unpromoted catalyst, or by introducing
parameters - functions of the content of selectivity promoter [4,7,8,10,17].
Talaeva et al. [I21 included in their kinetic model an exponential function of the
surface coveraae of the selectivitv oromoter.
Llsins the Nelder-Mead prooram, we tested a varietv ,of published equations as
well as our own model of the unpromoted catalvst [16], for which we introduced
a term for the dependence on the DCE partial pressure. The criteria for dis-
crimination between these rival models were the same as in [163. The best result
was achieved with our kinetic equations [16], involving additional DCE terms

in the numerator:

k7
klPoPe - k2PoPePdce
R(1) =
1 + k5Po + ksPe
154

TABLE 4
Values of the preexponential factors and activation energies

Preexponential Activation energy

factor

Ea/cal mol-'
kO

kl 6.867 9,260
k2 1.073 x IO2 10,413
k3 1.062 x 10 8.792
k4 3.959 x 10 9,473
k5 1.346 x 10'6 9,610
k6 3.109 x lo-* 15,044

TABLE 5
Values of the kinetic constants at different reaction temperatures.

240°C 264°C 292°C

kl 7.807 x IO -4 1.171 x 10-3 1.801 x 1O'3

k2 3.941 x IF3 6.220 x 1O-3 1.009 x 1o-2
k3 1.910 x 10-3 2.808 x lO-3 4.224 x 10-3
k4 3.654 x 1O-3 5.534 x 10-3 8.591 x 1o-3
k5 1.668 x 10-2 1.094 x 10-2 7.006 x lO-3
.k6 7.949 x 1o-2 4.112 x 1o-2 2.045 x lo-*

k8
k3PoPe - k4PoPePdce
R(2) = (7)
1 + k5Po + k6Pe

The quantities in equations (6) and (7) are measured in the following units:

CP,I = CP,I = CPdcel = atm

-1 -'
[R(l)1 = [R(Z)] = mol h g-cat

Ck,l = Ck,l = mol h-' g-cat-' atm -2

-1 -1 -3
Ck,] = Ck4] = mol h g-cat atm

Ek,] = [k6] = atm-'

The reaction orders with respect to DCE - k7 and kg are dimensionless. Their
values are k7 = 0.19, kg = 0.07 (at all reaction temperatures). The preexponentia?

factors and activation energies corresponding to the kinetic constants k,-k6 are
given in Table 4. The kinetic constants are temperature dependent in accordance

with the Arrhenius equations

 155

TABLE 6
95% confidence intervals for the preexponential factors, activation energies and
reaction orders.

Preexponential factor, k, -1
Activation energies/Cal mol
Lower limit Upper limit Lower limit Upper limit

6.476 7.258 9,195 9,325

94.2 120.4 10,282 10,544
10.36 10.88 8,766 8,818
38.40 40.78 9,440 9,506
I.195 x Io-6 1,497 x 10-6 8,880 10,340
2.699 x IO-8 3.519 x 1o-8 13,690 16,398
0,181 a.199
0.068 0.072

k = ko.eSEa/RT (for kl-k4) (8)

k = k,. eEa/RT (for k5 and k6) (9)

where k, is the preexponential factor and Ea is the activation energy. Using

equations (8-9) and the data in Table 4 we can obtain the values of the kinetic
constants of equations (6-7) at our reaction temperatures (Table 5). The average
deviation of the model for the partial oxidation is 19% and for the complete
oxidation is 25%.

The 95% confidence intervals for the preexponential factors and activation
energies corresponding to the kinetic constants kl-k6 as well as for the reaction

orders with respect to DCE - k7 and kg are given in Table 6. It was not possible
to calculate the confidence intervals with the Nelder-Mead algorithm, so we tested
equations (6-7) by another computer program, based on an algorithm proposed by

Marquardt [18]. We obtained the same values for the kinetic constants plus their
95% confidence intervals.
The adequacy of our model (6-7) is illustrated in Figures 7 and 8, where the

model functions R(1) = f,(P,) and R(2) = f2(Pe) at constant temperature and feed-
stock composition are juxtaposed with the experimental data - it is seen that the

agreement between the experimental and model values is quite good,

The numerical analysis of the terms in the numerators of equations (6-7) drew

attention to the fact that the constants in the DCE terms kiPoPePdLe are greater
than the ones in the first terms: k2 % 5kl, k4 2 2k3 (Table 5). Nevertheless,

the DCE terms are smaller than the terms kjPoPe because of the smaller values of
0.19 0.07
the multiplicands Pdce respectively:
and 'dce
156

a
- 2
E

5 I
24 25 26 27 26 29

qH# www t=ws5uE, P,xW2(atm) cgH4 PARTIAL PRESSURE, P,X~~“fod

FIGURE 7 Comparison of the model ethylene epoxidation rate ( 0 - R(l)m) with

the experimental rate ( l - R(l)e) as a function of the ethylene partial pressure
Pe at 240°C and feedstock composition: 25% 02, 50% C2H4 and 25% Ar.

FIGURE 8 Comparison of the model ethylene combustion rate ( O- R(2)m) with the
experimental rate ( l - R(2)e) as a function of the ethylene partial pressure Pe
at 264°C and feed composition: 33.3% 02, 33.3% C2H4 and 33.4% Ar.

0.07
= 0.2 ppm + Pdce = 2.10W7 atm + PO*"
dce = 0.0534 < Pdce = 0.3398
'dce
'dce = 2.0 ppm -f Pdce = 1.9 x IO -6 atm + Pdce
0.19 = 0.0819 < Pdce
0.07 = 0.3978

'dce = 24.4 ppm -f Pdce = 2.3 x IO -5 atm + Pdce

0.19 = 0.1318 < P&z' = 0.4739

These examples show that the rate of complete oxidation R(2) equation (7) will
be inhibited to a greater extent than the rate of epoxidation R(1) equation (6).
In the absence of the selectivity promoter (Pdce = O), the DCE term becomes equal
to zero and the kinetic equations have the same form as they had for the unpro-

moted catalyst - somehow the kinetic constants have different values. So we can
ascribe a physical meaning to the DCE term - the value by which the corresponding
rate will be decreased - AR(I) and AR(Z). The decrease in rate of three different
promoter concentrations are presented in Table 7. The estimates AR(J) are only

tentative as our kinetic model (6-7) is valid only for the promoted catalyst
(0.2 to 28 ppm DCE) and the extrapolation to Cdce = 0 may not be justified. For

the unpromoted catalyst, the model described previously [I61 is applicable.

 157

TABLE 7
Rate decreases at different DCE concentrations

cdce P
dce pO ‘e
T AR(I) AR(2)

/ppm /atm /atm /atm /"C 1% /%

0.2 2 x 10 -7 0.280 0.284 264 28.4 67.3

2.0 1.9 x IO6 0.383 0.143 264 43.5 78.6
24.4 2.3 x 1O-5 0.187 0.431 240 66.6 90.9

Selectivity equations
The kinetic equations (6-7) lead to the following expression for the model
selectivity Sm of the DCE-promoted catalyst:

0.19
R(1) _ kl - k2Pdce
sm = (10)
R(i) + R(2) ' "1 + (k3 - k4P;;z7)
(kl - k2Pdce

According to equation (10) the selectivity depends on the temperature (through

the constants ki) and on the partial pressure of the promoter Pdce (atm). This
equation is a complex non-linear function with respect to Pdce (at constant tempera-
ture). Nevertheless, the model dependence of the selectivity on the promoter con-

centration Cdce (ppm) at a fixed temperature and feed composition may be approxi-
mated to a straight line (Figure 6). The slopes and the intercepts of these plots

are somewhat different from those of the experimental plots and this difference
illustrates the deviation of the model. The adequacy of our model may also be
judged by the juxtaposition of experimental and model selectivities in Tables 2
and 3. The linear behaviour of a non-linear equation could be explained by the
fact that the first derivative of the function (IO) with respect to Pdce (at to =
constant) depends weakly on the partial pressure of the promoter under our
experimental conditions, So this partial derivative bears some resemblance to the
first derivative of a linear function y' = constant. Table 8 lists the values of
the slopes of the model selectivity versus Cdce at a fixed temperature and feed
composition.

The data in Table 8 have already been discussed. Our attention is attracted

by the fact that equation (IO) does not reflect the partial pressures of the
reactants - oxygen (PO) and ethylene (P,). In fact these pressures cancel as they
are present in all terms both in the numerator and in the denominator of equation
(10). We derived another selectivity equation (11) which takes account of the
influence of the feed ratio oxygen/ethylene on the selectivity as evidenced by
the data in Table 8
158

TABLE 8
Values of the slopes of the model selectivity as a function of the promoter con-
centration.

Feed ratio Catalyst bed temperature/"C

: C2H4 240 264 292
O2

I:2 0.34 0.58 0.87

I:1 0.50 0.90 1.23
2:l 0.94 2.00 1.32

6(k, - k2P;;Ag)
sm = (11)
(5 + Fe) (k, - bPdce
L
O”‘) + 6F,(k3 - k4P;;17)

where Fe and F. are the ethylene and the oxygen flow-rates. This equation (11)
has been derived in the same manner as the corresponding one in [IS], using the
kinetic model (6-7) and the empirical selectivity formula

S = 6ACeo.100/(5ACeo + 2AC,)

where aCi is the difference between the concentrations of the reagent i in the
converted gas mixture and in the feed. If the ethylene and the oxygen flow-rates
are equal - Fe = F. then Fe/F0 = 1 and equation (11) reduces to equation (IO),
so the latter is a special case of the more general equation (II). The average

deviation for the model selectivity equation (11) is + 6%.

Catalyst deactivation

The various types of catalyst deactivation have been discussed [20,21]. The
deactivation of silver catalysts promoted by chlorinated compounds may be related

to the "side-by-side" deactivation, representative of feeds containing poisonous

impurities [20]. This deactivation, as has already been established, is reversible.

The activity of a pellet at any time is defined [20] as the ratio of the rate
at which the pellet converts a reactant A and the rate of reaction of A with a
fresh pellet, The rate at which the activity "a" decreases with time "t" is then

expressed by:

k Cnad (13)
-$= dp

where Cp is the poison concentration, kd is the rate constant for the deactivation

and "d" is the deactivation order [20]. In our case - steady-state conditions -

the activity does not change with time:

 159

TABLE 9
Values of critical DCE partial pressures and concentrations

Reaction 240°c 264°C 292°C

P P
dce 'dce dce 'dce 'dce 'dce
/atm lppm /atm /ppm /atm /ppm

Epoxidation 1.99 x 10-4 208 1.53 x 1o-4 160 1.15 x 1o-4 120
Combustion 9.44 x l@ 99 6.18 x 1O-5 65 3.93 x 1o-5 41

- $ = 0 a = constant (14)

In this case the activity is a function only of the promoter content in the feed

(C dce Or 'dce) as it is obvious that the degree of deactivation will be greater

when the promoter content is greater. We expect that at a certain DCE concentration
the catalyst is completely poisoned. Now let us define the catalyst activity for
epoxidation (a,) and for combustion (a,):

a, =
CR(l)lp and a2 = CR@)l’ (15)
CKi)-iO [W

where [R(J)]' is the rate in the presence of a promoter having a partial pressure
= p and [R(J)pthat in the absence of promoter, i.e., Pdce = 0. So ai = 1
'dce
for the unprompted catalyst and ai = 0 at a definite partial pressure of promoter
poisoning the catalyst completely - [R(J)lp’= 0. We expect these pressures Pdce =
p' to be specific for the epoxidation and the combustion reactions and to vary
with the temperature. Our task is to estimate these characteristic promoter partial
pressures for all our reaction temperatures on the basis of the kinetic model.
Substituting in the kinetic equations (6-7) P = p and Pdce = 0 and the expressions
dce
(15), we obtain:

0.19 =I- k4 0.07

k2 and (16)
a2 ’ 'dce
al 5
= ’ - -5 * Pdce

Taking the values of ki at the different catalyst bed temperatures from Table
5, we calculate the "critical".Pdce for complete catalyst deactivation and the
corresponding promoter concentrations (Table 9). The model predicts (Table 9)
poisoning of the combustion reaction at lower promoter concentrations. This means
that at Cdce > CELik and Cdce < C$: we expect 100% selectivity. Probably
the ethylene oxide yield in this range would be very low and would not have any

industrial importance. The complete poisoning of the catalyst at promoter concent-

rations above 30-O ppm was experimentally established in this study (see page 3).
160

The higher the temperature the lower is the critical content of promoter - this
fact can also be explained by the increase in Cl - AES signal up to about 650°K [ZZ],
as for the increase in selectivity with increasing temperature.

Mechanism
We previously proposed [16] a mechanism, corresponding to an empirical kinetic
model for the unpromoted catalyst. We have now added a new step, the dissociative
adsorption of DCE. This should be reversible, bearing in mind our experimental
results and those of others, but at the same time the equilibrium should be
shifted strongly towards the adsorption:

C2H4C12 + 2 Z = 2 ZCl + C2H4 (17)

SO we have derived from our previous mechanism [16] two new mechanisms - one

including reversible adsorption of DCE and the other - irreversible, all the
other steps remaining the same. We tried to derive corresponding kinetic models,
using Temkin's general equation for the rate of a steady-state reaction as in [16]

but the solution of the systems of equations obtained was beset with insurmountable
mathematical difficulties. This was probably because the stoichiometric numbers
for the DCE adsorption step are equal to zero in both reaction routes as the
adsorption of DCE does not lead to the formation of products, The role of DCE is
to occupy a portion of the active centres on the silver surface,.thus inhibiting

the dissociative adsorption of oxygen:

o*+zz = 2zo (18)

This explanation was proposed by Kilty et al. [14]. It is the only logical

explanation if we accept the mechanism, according to which adsorbed molecular

oxygen produces ethylene oxide and adsorbed atomic oxygen produces CO2:

+ C2H40 + ZO (19)
z"2 + C2H4

,._
6 ZO + C2H4 + 2 CO2 + 2 H20 + 6 Z (20)

We previously [16] obtained a kinetic model supporting this mechanism. The:'kinetic

equations (6-7) for the promoted catalyst are derived by the modification of this
model, so they can also be regarded as evidence in favour of this mechanism.

CONCLUSIONS
The kinetics of the selective oxidation of ethylene over a silver catalyst,

promoted with dichloroethane, may be adequately described by rate equations,

involving the partial pressure of the'promoter. The kinetic model for the unpro-
 161

moted catalyst may be modified by the introduction of additional terms in the

numerator, describing the inhibiting effect of the selectivity promoter when present.
For promoter concentrations of the order of ppm the reaction order with respect
to dichloroethane is found to be less than 0.5.
The selectivity of the catalyst for ethylene oxide is not influenced by the
degree of conversion of ethylene.
Dichloroethane is both an inhibitor of the oxidation of ethylene and promoter
of the selectivity for ethylene oxide. The influence of DCE concentration in the
feed on the selectivity may be satisfactorily described by a linear model (up
to 30 ppm). The promoting effect of DCE on the selectivity increases with increasing
temperature. A feed having an oxygen/ethylene ratio = 2:l results in selectivities
most sensitive to DCE concentration.
The adsorption of DCE on the silver surface is a reversible process - if the
addition of the selectivity promoter to the feed is discontinued the adsorbed
chlorine is gradually desorbed by the reactants and products and the initial high
catalytic activity is restored. Under steady-state conditions, the activity of
the catalyst is a function only of the content of selectivity promoter in the
feed, which leads to a simple form for the catalyst deactivation equations.
The role of dichloroethane is to inhibit the dissociative adsorption of oxygen,
thus inhibiting the complete oxidation of C2H4.

REFERENCES

1 Kirk-Othmer, Encyclopedia of Chemical Technology, 2nd. ed., ~01.8, Wiley-

Interscience, New York (1965).
2 V.E. Dstrovskii, N.V. Kul'kova, V.L. Lopatin and M.I. Temkin, Kinet. Catal.,
3 (1962) 160.
E.L. Force and A.T. Bell, J. Catal., 44 (1976) 175.
0. Kamensky, D. Bonchev, N.V. Kul'kova and M.I. Temkin, Kinet. Catal., 19
(1978) 633.
5 0. Kamensky, N.V. Kul'kova and D. Bonchev, React. Kinet. Catal. Lett., 7
(1977) 481.
6 A. Ayame, N. Takeno and H. Kanoh, J. Chem. Sot., Chem. Commun., (1982) 617.
7 Yu. V. Ionov, E.M. Temkina, D. Kamensky, N.V. Kul'kova and M.I. Temkin, Proc.
3rd Sov. Conf. Kin.Heter. Cat. React.. 1 (1980) 201.
8 0. Kamensky and L. Petrov, Bulg. Acad: SC;. - Commun. Dept. Chem., 16 (1983) 86.
D.W. Park, S. Ghazali and G. Gau, Appl. Catal., 6 (1983) 175.
I"0 S. Ghazali, D.W. Park and G. Gau, Appl. Catal., 6 (1983) 195.
11 A. Ayame, T. Kimura, M. Yamaguchi, H. Miura, N. Takeno and H. Kanoh, 3. Catal.,
79 (1983) 233.
I.G. Talaeva, L.A. Vasilevich, A.K. Avetisov, B.B. Chesnokov and A.I. Gel'
bshtein, Proc. 3rd Sov. Conf. Kin. Heter. Cat. React., 2(Q;80) 441.
M.S. Kharson. A. Kh. Mamedov and S.L. Kioerman. Kinet. 25 (1984) 353.
P.A. Kilty, N.C. Rol and W.M.H. Sachtler', Proc: 5th In,t. Con;;. Catal., Palm
Beach, 2 (1973) 929.
H.H. Voge and C.R. Adams, Adv. Catal., 17 (1967) 151.
L. Petrov, A. Eliyas and D. Shopov, Appl. Catal., 18 ( 1985) 87.
C-T. Campbell and B.E.Koel, J. Catal., 92 (1985) 272.
0. Marquardt, J. Sot. Ind. Appl. Math., 11 (1963) 2.
T.E. Corrigan, Chem. Eng. Fundam., 199 (1955).
0. Levenspiel, J. Catal., 25 (1972) 265.
P. Forzatti, G. Buzzi-Ferraris, M. Morbidelli and S. Carra, Intern. Chem. Eng.,
24 (1984) 60, 232, 441.
C.T. Campbell, Journal of Catalysis (in press).