THE SILVER CATALYZED OXIDATION OF ETHYLENE

111. KINETICS OF ETHYLENE OXIDE OXIDATION'

ABSTRACT
A flow type apparatus was used for kinetic studies of t h e silver catalyzed
oxidation of ethylene oxide (EtO) by oxygen a t 274'C. Using N2 as diluent the
concentrations of 0 2 and ethylene oxide were varied independently frorn 9.9 t o
79y0 and 2.35 t o 9.4y0 respectively while a total pressure of 1 atmosphere was
maintained. Flow rates were varied t o give a range of contact times varying
frorn 0.06 t o 0.25 sec. I t was shown t h a t E t O is oxidized without previous
dissociation into CpHl a n d 0 2 . The dependence of the initial rate of oxidation
of EtO-on reactant concentrations excludes isomerization of EtO (to acetalde-
hyde) a s a main step in its oxidation, and a direct oxidation mechanism is
suggested. The results of a few experiments t o determine the extent of iso-
Can. J. Chem. Downloaded from cdnsciencepub.com by 198.245.107.67 on 03/05/21

merization of E t O t o acetaldehyde in the absence of oxygen are presented. No

steady state could be achieved but the results may be used semiquantitatively
t o support t h e belief that isomerization is not the rate determining step in the
oxidation of ethylene oxide.
INTRODUCTION
In order to further the establishment of a reaction mechanism for the
oxidation of C z H 4on a silver catalyst, Parts I and I1 of this series are here
supplemeilted by a study of the oxidation of ethylene oxide (EtO) and its
isomerization on the same catalyst sample.
EXPERIMENTAL
For personal use only.

T h e apparatus, catalyst sample, and determination of reaction products

have been described in Part 11. The source of the gases used and their purity
was specified in Part I.
Flow rates of EtO were controlled by a two-stage reducing valve followed
by three needle valves in series. Mild infrared heating of the cylinder head and
reducer ensured constancy of flow with downstream pressure of about 7
P.S.I.G. A dual range orifice meter was used (1-5 and 5-20 liters/hr.) filled
with Apiezon-B oil, and having a scale graduation accurate to 0.5%. This was
calibrated by mixing the EtO with a Nz stream of known Row rate to cover a
range of composition from 1.5 to 10% EtO. Samples were drawn into a gas
burette and the EtO content determined within 1% of the true value employ-
ing previously described techniques.
T o determine acetaldehyde samples were measured into a mercury burette
and then forced through a sintered glass bubbler containing a solution of
hydroxylamine hydrochloride.* T h e HC1 liberated by the equivalent amount
of aldehyde was titrated with NaOH from microburettes. Checks on the
efficiency of this method were made by vaporizing pure liquid acetaldehyde
Manuscript received November 5 , 1953.
Contribution from tlze Division of Applied Clzenzistry, National Research Council, Ottawa.
Issued as N.R.C. No. 3830.
National Rcsearclz Council Postdoclorate Fellow, 1951-53.
*Seven grants of ATH20H.HCl was dissolved i?z100 cc. of water. For eaclz determi~zatio~z 0.1 cc.
of this reagent w a s diluted with 10 cc. H?O, a drop of bromoplzenol blzle solz~tio?~
added, and litrated
with 0.005 N N a O H to llze jirst blz~ecolor, before absorption of aldehyde.
 ORZECHOWSKI A N D &facCGRMACK: OXIDATION OF ETHYLENE. III 433

into a mercury burette. After nleasuring the pressure, volume, and tempera-
ture of the sample, Nz was added to give 0.1% CH3CHO by volume. The
estimation of hydroxylamine l~ydrochlorideaccounted for a t least 75y0 of the
aldehyde present.
Polarographic n~easurements, kindly made by Dr. R. J. Cvetanovid,
showed that the aldehyde found in the products of EtO isomerization was
acetaldehyde.
As described in Parts I and 11, the catalyst was conditioned with respect
to CzH4oxidation a t the working temperatures, and kinetic runs a t 274' and
234°C. were made. Subsequently the investigation on EtO was started. I t was
realized (Part I) that the catalyst conditioned by ethylene oxidation was in a
very unsteady state with respect to EtO oxidation. On the other hand, as soon
as runs of the latter reaction were made, the catalyst tended to be conditioned
Can. J. Chem. Downloaded from cdnsciencepub.com by 198.245.107.67 on 03/05/21

by thein and behaved, then, inhibited with respect to ethylene oxidation.

An explanation of these phenomena is offered in Part I.
As was the case during the studyof ethylene oxidation, the obtaining of
reproducible results would therefore appear to necessitate catalyst condition-
ing a t the given reaction temperature with a standard gas mixture having
average composition respecting the range employed. Runs of short duration
were made, followed by the standard mixture. In this way frequent checks
of the catalyst's activity and the maintenance of its reference state were
achieved.
T h e symbols for feed composition are similar to those used previously
except that instead of E (ethylene) symbol X (ethylene oxide) is used. T h e
For personal use only.

arbitrary unit of gas mixture flow rate is 41.8 liters (N.T.P.)/hour. The symbol
[pXnOn] means 41.8 liters (N.T.P.)/hour of feed gas mixture having
composition 2.35 n% CZHIO; 9.90 m% 0 2 , the remainder being N z a t a total
pressure of 1 atmosphere.
T h e standard feed for the purpose of conditioning and checking was
1. x 2 0 ? .
In the preliminary stages of the work, seven measurements were made to
check the material balance of reactants and products (EtO and COz). Three
of thein iilcluded CzH4 determination. At a later stage analyses for acetalde-
hyde were made frequently, particularly for the composition X401, but
always with negative result.
Ethylene was determined as described in Part 11. T h e method did not allow
detection of less than 0.1-0.2 cc. (N.T.P.) as was shown by blank runs. T h e
results with l X z O zand 4X202, using gas samples of 250-500 cc. showed t h a t
the conversion of EtO to C2H4, if any, was below 1%.
The over-all balance was found to be between 97 and 101%* (without
counting the possibility of slight C2H4 formation).
Therefore all conversions were determined directly by COz analysis instead
of indirect determination by EtO analysis.
*There i s some indication Lhut during the firs1 40 min. afler admission of 1x202 to the catalyst
preconditioned by C2H4 oxidalion, the balance of E10, COz, and possibly C2H4 i s only abot~t90%.
Tlze data are not sz(ficient to explain or cottfir?tt this behauior.
 434 CANADIAN JOUKVAL OF CXEMISTRY. VOL. 32

TABLE I
O X ~ D . ~ T IOF
ONETIIYLENE O X I D E AT 274°C.

Kinetic
run
No.
--
Feed
composition
symbol
Time
operated
(hr.)
I .,??to
C02
Standard
check
No.
-
1a 1x101 0.65 10.7 -
Ib 1x101 1.65 10.3 -
1c lXl01 2.80 10.2 -
2a 2x101 0.65 6.27 A

2b 2x101 1.40 6.35 -

- 1x202 0.60 9.00 la
- 1x202 1.20 9.30 Ib*
- 1x202 1.25 8.86 2a
- lX?Oz 3.35 8.65 2b
Can. J. Chem. Downloaded from cdnsciencepub.com by 198.245.107.67 on 03/05/21

3a 1x104 0.65 12.85 -

3b 1x104 2.00 12.85 -
- 1X?01 0.65 9.46 3a
- 1X?02 17.00 9.00 3b
4a 2x104 0.65 8.12 -
4b 2x104 2.00 8.30 -
5a 1x10s 1 .OO 13.30 -
5b lX10a 1.75 13.05 A

Ga 2x10s 0.65 7.22 -

Gb 2x10s 1.35 7.40 A
For personal use only.

- 1x202 0.65 9.50 la

- 1x202 17.00 8.95 4b
7a 1X.,04 0.65 7.92 -
7b 1x404 1.35 8.25 -
7c 1x404 2.00 8.39 -
8a 2x404 0.65 4.51 -
8b 2x104 I .35 4.42 -
9a 4x404 0.65 2.84 -
9b 4x404 2.00 2.74 -
- 1X202 0.65 8.95 5a
- I x202 18.00 9.46 5b
- 1x202 18.75 8.70 5c
--
1x202 19.25 8.86 5d
- 1x202 20.00 8.66 5e
- 1x202 20.75 8.92 5f
- 1x202 22.00 8.86 5g*
- 1X202 1.50 9.70 6a
--
1x202 3.65 8.41 6b
- 1x202 20.50 8.70 6c
Ila 2X202 0.65 5.30 -
llb 2x202 1.35 5.34 -

12 4x202 0.65 3.04 -

*TIze apparatns was closed down (cooling to room tentp. i n N2 stream) after G I L C GIb,
~ 5g, and
lob.
 ORZECHOIVSKI A N D ilfacCORBIACK: OXIDATION 01.' ETIIYLENE I I I 435

T A B L E I (Co~zclz~ded)
I I
Icinetic Feed Time 7% Standard
run composition operated EtO t o check
No. symbol (hr.1 CO* No.
- --- - --
4x101 0.65 4.25 -
-
-

ia
7b
-

-
-
Can. J. Chem. Downloaded from cdnsciencepub.com by 198.245.107.67 on 03/05/21

8a
86
-
-
-
-
-

9a
9b
-
For personal use only.

-
-
-
-

10a
1Ob*
110
llb

*The apparatzrs was closed d o w ? ~(coo!ing to room temp. in Nz stream) after check l b , 5g, u?zd
1Ob.

RESULTS
1. Oxidation Kinetics at Constant Temperature
The influence of feed composition and ROW rate was studied after overnight
stabilization of the catalyst using mixture l X z O zin all cases.
Table I shows the results of kinetic runs and of the frequent checks for the
standard l X z O zfeed. Operation times are included in this table.
All checks for l X z 0 2indicated 8.5-9.5% conversioil to CO? and hence the
surface was assumed to be in the same reference state. One exception was the
check run 1 X 2 0 ? made after studying the composition X401, where CO2
conversion of 1X202 was found a s high as 11.5% during the fir-st few hours.
 436 CANADIAN JOUlWAL OF CHEMISTRY. VOL. 32

Hence the values for X 4 0 1are corrected to the normal reference state. This
behavior of oxygen deficient feed is very similar to the behavior observed
with similar ethylene-oxygen compositions. (Compare feed E d 0 1 in Parts I
and 11.)
Coilversions versus reaction time co-ordinate for different feed compositions
are plotted in Fig. 1.
Can. J. Chem. Downloaded from cdnsciencepub.com by 198.245.107.67 on 03/05/21

025 0.50 0.75

REACTION TIME COORDINATE
For personal use only.

FIG.1. Conversion of E t O to CO? and H20 a t 274'C. vs. reaction time for different feed
compositions.

The initial rate of reaction is given by

where V z = volunle fraction of E t O in the feed,

x = fraction of E t O converted to CO2 and H20,
M = catalyst quantity (assigned unity and constant throughout),
F = flow rate of feed gas mixture in previously defined arbitrary
units,
r = rate of reaction in the same units with M = 1.
Suffix (I means conditions a t zero reaction time when M/F = 0.
The initial rates are shown by the points on Fig. 2.
In Part IV, which deals with reaction mechanisn~,the reasons will be presen-
ted which led to a rate equation of the form
 ORZECHOFVSKZ A N D M a c C O R I I A C K : O X I D A T I O N O F E T H Y L E N E . I I I 437

where Px EtO partial pressure

=
Po partial pressure
= 0 2
A , B, C are constants.
'The constants were found by the method of least mean squares to give

(Partial pressures are expressed in mm. Hg.)

Fig. 2 includes as solid lines the values of rates calculated according to the
above equation showing good agreement between experimental and calculated
data.
Can. J. Chem. Downloaded from cdnsciencepub.com by 198.245.107.67 on 03/05/21
For personal use only.

O2 PARTIAL PRESSURE (mm. Hg

FIG.2. Initial rate of COnrH20 formati011 frorn EtO a t 274°C. a s a f u ~ ~ c t i oofn reactant
partial pressures.

I t is noteworthy that the eqt~ationobtained is very similar to the one

derived for CzH4oxidation (Part 11), except for the constant A.
2. Oxidation Temperature Dependence
A few measurements were made shortly after the temperature was chang-ed,
and are presented in Table II. (Times necessary for heating or cooling to the
required temperature were usually in the order of half an hour.) These are not
the steady values which would have been obtained after very prolonged opera-
 C A N A D I A N J O U R N A L O F CHE&IISTRY. VOL. 32

TABLE I1
TE~IPERATURE DEPENDENCE
OXIDATION
OF ETHYLENE OXIDE (FEED 1x202)

I I I
T , "C. 1 Time operated Sample No.
1 at T0C. (hr.)
Can. J. Chem. Downloaded from cdnsciencepub.com by 198.245.107.67 on 03/05/21

tion where slow processes have changed the nature of the catalyst (compare
Part I). Therefore they indicate more correctly the dependence of kinetics
on temperature. Checks a t 274OC. after operating a t other temperatures
included in Table I1 prove indeed that the reference state of the surface has
been maintained throughout.
3. Isomerization to Acetaldehyde
Immediately following the kinetic runs for EtO oxidation, the investigation
of EtO isomerization a t 274OC. was begun by feeding only EtO and N?. All
For personal use only.

data are presented in Table 111.

In Expt. 3, after 2.25 hr. operation, a check was made for the presence of
C2H4in the outgas. This, if any, corresponded to less than 0.6yo conversion of
EtO. The over-all balance of EtO was also within the limits of experimental
error.
In view of the absence of CrH4, the formation of COz seems surprising.
If the oxygen were supplied by deco~npositionof EtO, according to:
5 CzH40 --t 5 C2H4 2.5 0 + 2
CsH40 +
2.5 0 2 + 2 CO2 + 2 H20
then for each mole of EtO converted to COz, there would be 5 moles of EtO
converted to C2Ha. The conversion to COz being of the order of 2y0, this
would result in conversion to C2H4 of the order of 10yo, which is not the case.
Also, the adsorption capacity of the catalyst is far too small to account for
the missing C2H4.
I t is therefore concluded that the Nz enlployed contained srnall amounts of
oxygen which are responsible for COz formation; 0.2-0.5yo of 0 2 in nitrogen
might be sufficient quantitatively, based on the kinetics of EtO oxidation.
The conversion to C 0 2 increases with decreasing EtO pressure (compare
Expts. 6 and 8 in Table 111) which also collforms to the kinetics of EtO
oxidation but is not paralleled by the behavior of isomerization in the same
experiments.
 ORZECEIOFF'S'SRI A N D UocCOR!>fACK: O X I D A T I O N OF ETHYLEl\'E. III 139

TABLE 111
ISOMERIZATION
01: ETHYLENE OXIDE TO ACETALDEHYDE
--
% EtO converted
Flow Time
composition
syrnbol
operated
(hr.1
To
CHxCHO
-
To
CO? I Total
Expt. No.

0.40 2 - - i
0.85 2 - - i
1.50 2 - - 1
Can. J. Chem. Downloaded from cdnsciencepub.com by 198.245.107.67 on 03/05/21
For personal use only.

After 58 lzr. (1E202)

I I I I I

- --

DISCUSSION
1. Oxidation
Twigg (1) reported that EtO is oxidized 011 silver catalyst, but there is
little i~lfornlationconcer~li~lg
the kinetics of this reaction. Also there are
 440 C A N A D I A N J O U R N A L OF C H E M I S T R Y . V O L . 32

insufficient data relating to rate of C2H4 formation, especially a t small reaction

times.
There were, a priori, three possibilities:
1. EtO is oxidized without previous dissociation into C z H 4and 0 2 .
2. EtO is decomposed into C2H4 and OZ*followed by oxiclation of CzH4,
with no EtO being oxidizecl directly.
3. EtO unclergoes sin~ultaneousdirect oxidation and dissociation to O z
and C2H.I, the latter being subject to further oxidation.
The present d a t a substantiate statement (1) for the main reaction step,
since CZH4 did not occur in detectable quantities in the products.
The dependence of oxidation rate on the reactant partial pressures excludes
isomerization of E t O on the Ag surface a s a main step in its oxidation, because
there is no reason for isomerization to increase with oxygen pressure. Instead,
it is suggested that one of the following mechanisms is responsible for the
Can. J. Chem. Downloaded from cdnsciencepub.com by 198.245.107.67 on 03/05/21

formation of intermediates, which are then capable of undergoing rapid

oxidation.
(a) Reaction between EtO and an oxygen adatorn, adsorbed side by side.
(b) Collision of a gaseous E t O molecule with an oxjrgen adatom.
In accordance with this (see discussion below) the rate of EtO isomerization
to acetaldehyde is shown to be substantially lower than the rate of E t O
oxidation.
A private communication from industrial sources claimed that the rate of
EtO oxidation passes through maximum a t 260°C. This would imply that
mechanism ( a ) may be valid where E t O desorption rates become comparable
with oxidation rates a t approximately 260°C.
For personal use only.

We were unable to observe such an effect (Table 11) and therefore believe
that the mechanism (b) cannot be excluded.
The present data may be compared with those for CzH4 oxidation (Part 11).
I t transpires that the same factors favorable to high conversions of ethylene
to EtO in the initial stages of the reaction are also operative for the conversion
of E t O to COz. Those factors are: large oxygen partial pressure and low oxygen
acceptor pressure. Therefore a high p o , / p c , H 4 ratio does not necessarily
represent the best operating conditions for E t O production as was pointed out
previously (Part 11).
The present d a t a may be helpful in attempting to correlate the reaction
rate data for the over-all CzH4 oxidation as a function of contact time.
Before this can be done, however, more data are needed for the influence of
COz and HzO. Also the effect of the initial composition ratio P ~ , H ~ / P ~ , ~

should be investigated, which could be followed experimentally by using

carbon-labelled EtO. This method would provide further evidence for the
formulation of a reaction mechanism.
2 . Isomerization
The quantitative interpretation of the isomerization process is unfortu-
nately obscured by the presence of oxygen traces in the diluent N z and influenced
*Titis was actually found to occztr by Twigg ( 1 ) ulhen EtO i s adnzitled to orygelz free silver i n a
static system.
 ORZECIIOIVSKI AND ilfacCORBfXCK: O X I D A T I G N OF E T H Y L E N E . I I I 441

adversely by the difficulty of estimating low concentrations of acetalclellyde

in the products. Furtl~ernlore, the conversion level is not steady. Higher
conversions t o acetaldehyde (about 2%) are obtained immediately after
pretreatment by E t O oxidation (Expts. 1 and 3, Table 111) or by C2H4oxida-
tion (Expt. 11). Further, the isornerization decreases with operating time.
This confuses tlie interpretation of data but might indicate the formation of a
surface deposit. I t will be noted that the standard oxidation of ethylene. oxide
(Expt. 10) gives higher than normal values: 13.4% instead of 9%, when made
immediately after the series of isomerization runs. The same is true if C z H 4
oxidation is run after isomerization: initially, abnornlally high conversio~l~
both to COs and to EtO are obtained as described in Part I.'The adsorptioll
capacity of the catalyst used (about 1 cc. N.T.P.) seems too small to e x p l a i ~ ~
this excess conversion only on the basis of a deposit which burns on contact
Can. J. Chem. Downloaded from cdnsciencepub.com by 198.245.107.67 on 03/05/21

with 0 2 .
This behavior could be explained if we assume that feeding E t O wit11
practically no oxygen is effective as follows:
( a ) stably sorbed oxygen, inhibiting the reaction (see Part I) is removed;
(b) the surface is slowly covered by acetaldehyde residues.
When oxygen is admitted later, together with CzH4 or EtO, the acetalde-
hyde residues are burned off in~n~ediately. Tlle fixation of oxygen in stable
positions and the formation of oxidation residues (see Part I ) being slow, the
rate of oxidation during the first few hours will be greater than normal on this
cleaned surface.
Notwithstanding the complications clue to the unsteady level of isomeriza-
For personal use only.

tion and to the presence of oxygen, it appears t h a t iso~nerizationincreases

with increasing EtO pressure (comparing Expts. 3, 4, and 5 ; 6 ancl 8; 8 and 9).
This is expected since the rate of isomerization will be proportional to the
fraction of Ag surface covered by adsorbed EtO, and this must increase with
its pressure.
As was pointed out in the discussion of E t O oxidation, the kinetics of this
reaction do not confornl to the'hypothesis t h a t isomerization to acetaldehyde
is the rate determining stage during oxidation. T o support the independent
direct oxidation of E t O as the maill path of COz formation it should be
possible to show that isomerization of EtO t o acetaldehyde is slower than t h e
oxidation of E t O under comparable conditions.
Experiments 2 and 3 in Table I11 show t h a t initial conversion to acetalde-
hyde is 2.1Yo against about 9% conversion to Con, when oxygen is present.
Thus, less than 25Y0 of EtO oxidized undergoes isomerization. Moreover the
conditions of Expts. 2 and 3 are not strictly comparable because in the latter
case, the available free Ag surface is increased in the absence of oxygen, and
consequently measured values for isomerization could be actually higher than
those which occur during oxidation.
The COz conversions shown in Table I11 can be attributed without difficulty
to the direct oxidation of EtO by traces of oxygen. On the other hand, if it is
assumed that all COz shown in Table I11 arises by prior isomerizatioll followed
by oxidation due to traces of oxygen, the highest isonlerization values would be
 442 CANADIAN J O U R N A L OF CHEMISTRY. VOL. 32

3.6% (Expt. 11). This would still account for only 4070 of the EtO which is
oxidized uncler conlparable conditions.

REFERENCE
1. Twrcc, G. H. Proc. Roy. Soc. (London), A, 188: 92. 105; 123. 19.16.
Can. J. Chem. Downloaded from cdnsciencepub.com by 198.245.107.67 on 03/05/21
For personal use only.