The selective epoxidation of conjugated olefins containing allylic substituents and epoxidation of propylene in the presence of butadiene

Article abstract of DOI:10.1016/j.jcat.2004.04.032

The epoxidation of isoprene (2-methyl-1,3-butadiene) and piperylene (1,3-pentadiene), both conjugated olefins containing allylic methyl groups, has been conducted using conventional, CsCl-promoted, Ag/α-Al ₂O₃ catalysts. Selectivities to the allylic olefin epoxide isomers are over 20% and are much higher than expected due to the presence of the conjugated olefin structure. The epoxidation of propylene over the same catalyst under the same conditions is only 2.5%. The epoxidation of propylene in the presence of butadiene also yields PO in much higher selectivities. The presence of as little as 1% C₄H₆ in the reaction feedstream increases the selectivity to PO from 2.5% to over 40% at the expense of overall activity for C₃H₆ conversion. The upper limit of selectivity to PO in the presence of C₄H₆ appears to be approximately 50%, suggesting an upper limit for the effectiveness of this methodology. Epoxidation of C₄H₆ alone on similar Ag catalysts indicates that the consecutive reaction of EpB to CO ₂/H₂O is strongly limited by the presence of excess C ₄H₆ in the feedstream. In addition, the selectivity to EpB is directly proportional to the amount of C₄H₆ in the reaction feed stream. Selectivities >90% are obtained only when there is sufficient C₄H₆ in the reaction feedstream to control the concentration of the reactive Ag-O surface. For C₄H₆ epoxidation, all CO₂/H₂O is formed by a consecutive reaction pathway from EpB; there is no parallel pathway for the direct formation of CO₂/H₂O from C₄H₆. Using the selective epoxidation of C₄H₆ as the model for understanding the enhancement in selectivity for allylic olefin epoxide formation, the most likely reason for improved selectivities is that strongly adsorbed C₄H₆ (or other conjugated olefins) limits the ensemble size of contiguous Ag-O surface sites. These ensembles are too small for PO combustion, but not too small for PO formation.

Full text of DOI:10.1016/j.jcat.2004.04.032

Journal of Catalysis 225 (2004) 374–380
www.elsevier.com/locate/jcat
The selective epoxidation of conjugated olefins containing allylic
substituents and epoxidation of propylene in the presence of butadiene
John R. Monnier,∗,1 Kimberly T. Peters, and Gary W. Hartley
Research Laboratories, Eastman Chemical Company, Kingsport, TN 37660, USA
Received 22 December 2003; revised 29 April 2004; accepted 29 April 2004
Available online 25 May 2004
Abstract
The epoxidation of isoprene (2-methyl-1,3-butadiene) and piperylene (1,3-pentadiene), both conjugated olefins containing allylic methyl
groups, has been conducted using conventional, CsCl-promoted, Ag/α-Al O catalysts. Selectivities to the allylic olefin epoxide isomers are
2
3
over 20% and are much higher than expected due to the presence of the conjugated olefin structure. The epoxidation of propylene over the
same catalyst under the same conditions is only 2.5%. The epoxidation of propylene in the presence of butadiene also yields PO in much
higher selectivities. The presence of as little as 1% C H in the reaction feedstream increases the selectivity to PO from 2.5% to over 40% at
4
6
the expense of overall activity for C H conversion. The upper limit of selectivity to PO in the presence of C H appears to be approximately
3
6
4 6
50%, suggesting an upper limit for the effectiveness of this methodology. Epoxidation of C H alone on similar Ag catalysts indicates that
4
6
the consecutive reaction of EpB to CO /H O is strongly limited by the presence of excess C H in the feedstream. In addition, the selectivity
2
2
4 6
to EpB is directly proportional to the amount of C H in the reaction feed stream. Selectivities > 90% are obtained only when there is
4
6
sufficient C H in the reaction feedstream to control the concentration of the reactive Ag–O surface. For C H epoxidation, all CO /H O is
4
6
4
6
2
2
formed by a consecutive reaction pathway from EpB; there is no parallel pathway for the direct formation of CO /H O from C H . Using the
2
2
4 6
selective epoxidation of C H as the model for understanding the enhancement in selectivity for allylic olefin epoxide formation, the most
4
6
likely reason for improved selectivities is that strongly adsorbed C H (or other conjugated olefins) limits the ensemble size of contiguous
4
6
Ag–O surface sites. These ensembles are too small for PO combustion, but not too small for PO formation.
2004 Elsevier Inc. All rights reserved.
Keywords: Epoxidation; Allylic olefins; Propylene; Butadiene; Propylene oxide; Epoxybutene; Silver catalyst; Ensemble effect
1. Introduction
is produced as the minor weight fraction during the epox-
idation reaction. The second indirect method is known as
the chlorohydrin process, where propylene is reacted with
hypochlorous acid to form propylene chlorohydrin. Follow-
ing reaction with Ca(OH)2, propylene oxide and CaCl2 are
produced. This reaction is both corrosive and environmen-
tally unfriendly as well as generating large amounts of salt
by-products, although this process has been very well in-
tegrated during its many years of commercial practice. As
a result of these limitations, there has been a great deal of
recent activity in exploring different alternatives for the di-
rect epoxidation of nonallylic olefins, including propylene,
by molecular oxygen. These alternatives include epoxida-
tion of propylene by in situ generated H2O2 from H2 and
O2 using size-specific gold crystallites supported on TiO2
catalysts [3–5] as well as H2O2 generated in situ over Pd–Pt
catalysts supported on TS-1 titanium silicalite [6,7]. The in
The direct epoxidation of allylic olefins, most notably
propylene, has been one of the most elusive goals in catal-
ysis research. Currently, over eight billion lb/year of propy-
lene oxide are produced annually by either of two indirect
routes [1,2]. The first is the so-called peroxidation process,
whereby organic hydroperoxides or organic peracids are re-
acted with propylene in the presence of an early row transi-
tion metal complex to form equimolar amounts of propylene
oxide and organic alcohols or acids, respectively. The po-
tential drawback with this approach is that propylene oxide
*
Corresponding author. Fax: (803) 777-8265.
E-mail address: monnier@engr.sc.edu (J.R. Monnier).
Current address: Department of Chemical Engineering, University of
1
South Carolina, Columbia, SC 29208.
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2004.04.032
2004 Elsevier Inc. All rights reserved.

J.R. Monnier et al. / Journal of Catalysis 225 (2004) 374–380
375
situ H2O2 catalysts and processes are thus far too inactive,
unstable, and nonselective for commercial application, al-
though further research and development continues in these
new areas.
For the past several years, we have studied the epoxida-
tion of conjugated, nonallylic olefins over promoted silver
catalysts supported on fused α-Al2O3 [18–21]. These cat-
alysts are more typical of silver catalysts normally used
for olefin epoxidation. Silver weight loadings are typically
between 5 and 15%, promoter loadings are between 500
and 1500 ppm (not 1–5% as in the case of the catalysts of
Gaffney and Bowman), and the support is fused α-Al2O3. In
the case of butadiene epoxidation, we have observed that the
selectivity for 3,4-epoxy-1-butene (EpB) from butadiene is
directly proportional to the concentration of butadiene in the
gas-phase feed stream. This is different from the situation
for epoxidation of ethylene to ethylene oxide, where the se-
lectivity to ethylene oxide is typically inversely proportional
to the concentration of ethylene in the feed [22,23]. The
situation for ethylene oxide formation is the expected rela-
tionship of feed concentration to selective product formation
where there is a parallel reaction pathway for nonselective
product formation (CO2/H2O in the case of olefin epoxida-
tion). Vannice and co-workers [24] have also observed that
the integral heats of adsorption of butadiene on O-covered
Ag catalysts are much stronger than ethylene adsorption on
the same surfaces, presumably due to the conjugated olefin
structure of butadiene. These results, coupled with the direct
relationship of EpB selectivity and butadiene feed concen-
tration, strongly suggest that adsorbed C4H6 stabilizes EpB
or its precursor against over oxidation to CO2 and H2O. One
explanation is that strongly bound butadiene on the Ag–O
surface may dilute the Ag surface into ensembles of Ag–O
too small for CO2/H2O formation, but not EpB formation.
In this paper, we present the results for the epoxida-
tion of olefins that contain both a conjugated olefin struc-
ture and a methyl group allylic to one of the C=C double
bonds [25]. These olefins examined in this study are isoprene
(2-methyl-1,3-butadiene)and piperylene (1,3-pentadiene).If
the strongly bound, conjugated olefin structure exerts the
same influence as it does during butadiene epoxidation, then
it may be possible to increase the selectivity for the allylic
epoxide isomer. By analogy, we also present data for the
epoxidation of propylene in the presence of various con-
centrations of butadiene in the propylene and O2-containing
feedstream to see whether selectivity to propylene oxide can
also be enhanced by the presence of butadiene. Finally, se-
lect data for the selective epoxidation of C4H6 are presented
to illustrate the relationship of C4H6 feed concentration to
the selectivity for EpB formation.
Conventional, supported silver catalysts have also been
tested for the epoxidation of allylic olefins such as propy-
lene. However, due to the reactivity of the allylic C–H bonds
(bond dissociation energy of allylic C–H in propylene is
77 kcal/mol), abstraction of the allylic hydrogen is preferred
over the addition of adsorbed oxygen to the C=C bond, thus
precluding the formation of propylene oxide [8,9]. More re-
cently, however, data have been published suggesting that
a new class of heavily modified, very high weight loading
silver catalysts show unexpected selectivity for propylene
oxide formation. Gaffney and co-workers [10–12] state that
catalysts containing 30–60% Ag, 0.5–3.0% K, 0.5–1.0% Cl
with balance CaCO3 give propylene oxide at selectivities as
high as 40–50% at temperatures of approximately 250 ◦C.
Likewise, Bowman and co-workers [13,14] evaluated a cat-
alyst containing approximately 60% Ag and high promoter
loadings of Group IIA salts, mixed with sodium silicate,
and found that at 180 ◦C the selectivity to propylene was
47% at 3.7% propylene conversion (10% C3H6 in feed and
GHSV = 200 h−1). Most recently, Jin et al. [15] report that
when a gas stream containing 15.6% C3H6, 12.2% O2, and
balance N2 was passed over a catalyst containing 50% Ag
and 50% MoO3 at 400 ◦C and GHSV = 4500 h−1, the selec-
tivity to propylene oxide was 53.1% selectivity at a propy-
lene conversion of 2.1%.
It is difficult to rationalize the performance of any of
the catalysts containing such high silver loadings in terms
of conventional supported catalysts, since, in many cases,
the support is present as a minor component. In addition,
the alkali or alkaline earth modifiers are present at much
higher concentrations than the ppm levels commonly asso-
ciated with promoted silver catalysts. It is more realistic to
consider these catalysts as more or less uniform mixtures of
silver, modifier, and inert support and to think of the changes
in catalytic properties as being due to the well-known ligand
and/or ensemble effects of catalysis [16,17]. In the case of
these Ag catalysts, ensemble theory predicts that catalytic
behavior may be influenced by either the electronic proper-
ties of silver sites interfaced with the support (ligand effect)
or by the sizes of the contiguous arrays of the silver surface
atoms resulting from the dilution of the silver surface by the
modifier and/or the support (ensemble effect). For example,
if the nonselective combustion of propylene to CO2/H2O re-
quires a different sized silver ensemble from that needed for
propylene oxide formation, then ensemble effects could play
a critical role in selective propylene oxide formation. The
ligand effect would require that the alkali or alkaline earth
additive or support influences the electronic nature of the
adjacent Ag surface sites. These types of electronic modi-
fications are usually associated with very small metal crys-
tallites, or even transition metal complexes, not large, bulk
Ag crystallites.
2. Experimental
The CsCl-promoted catalysts used in this study were
prepared using methods that have been described ear-
lier [18,19]. The CsCl loadings were determined using
AA-ICP and the Ag weight loading was maintained at
12% (wt) for all catalysts. The catalyst support was fused
α-Al2O3 rings, type SA-5562, which were supplied by

376
J.R. Monnier et al. / Journal of Catalysis 225 (2004) 374–380
Norton Corporation. The surface area of the support was
0.7–0.8 m2/g with a total pore volume of 0.55 cm3/g
and a median pore diameter of 7 µm. The Ag crystallite
diameters were measured from scanning electron micro-
graphs and automatic particle counting/measuring software.
The average Ag crystallite diameter was determined to be
0.14 µm; assuming hemispherical particle shapes, the Ag
surface site density was calculated to be approximately
2.7 × 1018 sites/g. No allowances were made for Cs and
Cl coverage on the Ag surface when estimating the Ag sur-
face site concentrations. Before being evaluated, all catalysts
were ground and sieved to give particle diameters of 0.4–
0.8 mm.
Fig. 1. Effect of C H conversion on selectivity to EpB. Catalyst is
4
6
Catalyst evaluations were made using the reactor and GC
analytical system previously described [21], except that the
CsCl-promoted, Ag/α-Al O . Promoter loading is 900 ppm Cs. Re-
2
3
◦
action temperature = 205 C; feed is 9% C H , 18% O , balance
4
6
2
−1
1
4
n-C H + 2 ppm 2-chlorobutane at GHSV = 12,000 h
.
reactor was constructed of -inch o.d. stainless steel embed-
4 10
ded in a 1-inch o.d. brass shell to help maintain isothermal
conditions during catalyst evaluation studies. Catalyst bed
temperatures were measured by means of a movable C/A
thermocouple (0.020-inch o.d.) inside an axial, thin-walled
stainless-steel thermowell running the full length of the re-
actor. Tylan FC-260 mass-flow controllers were used to reg-
ulate and maintain flows of C3H6, C4H6, O2, CH4, n-C4H10,
and He in the various reaction protocols. The addition of
gas-phase compositions of isoprene (bp = 34 ◦C), pipery-
lene (bp = 44 ◦C), and 1,5-hexadiene (bp = 59.5 ◦C) was
accomplished using vapor-liquid equilibrium (VLE) satura-
tors that were maintained at the desired temperatures using
close system refrigeration/heaters. By using a combination
of the temperature of the VLE saturator (and incorporation
into the appropriate Antoine equation), the flow rate of He
sweep gas through the VLE, and the overall flow rate of
the reactor feed stream, the desired feedstream compositions
were made.
(a)
3. Results and discussion
Results for steady-state epoxidation of C4H6 to EpB are
shown in Fig. 1 to illustrate the nature of the reaction net-
work for CO2 and H2O formation. The plot of C4H6 con-
version versus selectivity to EpB (as a catalyst activates to
steady-state performance) intersects the y axis very near
100% selectivity, indicating that essentially all combustion
occurs consecutively from EpB, or its precursor, and that
there is no parallel pathway from C4H6 for the formation of
CO2/H2O. These data are consistent with the earlier work of
Monnier and co-workers [9,26]. The data in Fig. 2 show the
effects of C4H6 feed concentrations on catalyst performance,
most notably the selectivity for EpB formation. Fig. 2a indi-
cates that at all reaction temperatures, the selectivity to EpB
is higher for the feed stream containing 17% C4H6, relative
to the feed containing 5% C4H6. The plot for 5% C4H6 feed
concentration shows a sharp transition toward even lower
selectivities at temperatures greater than 190 ◦C. This is be-
cause of the lower residual feed concentrations of C4H6 as
(b)
Fig. 2. Effect of C H feed concentration on the selectivity to EpB. Cat-
4
6
alyst is CsCl-promoted, Ag/α-Al O ; promoter loading is 900 ppm Cs.
2
3
Feed conditions are % C H as stated, 18% O , balance n-C H + 2 ppm
4
6
2
4 10
−1
2-chlorobutane at GHSV = 20,000 h . (a) Selectivity versus reaction tem-
perature. (b) Mole fraction EpB versus reaction temperature.
conversion increases; the conversion of C4H6 increases from
20 to 35% as the reaction temperature proceeds from 190
to 210 ◦C, further confirming the dependence of EpB selec-
tivity on C4H6 feed concentration. The results in Fig. 2b
confirm that the lower selectivity values for the 5% C4H6
feed composition are not due to higher EpB concentrations.
Since the data in Fig. 1 show that CO2/H2O are formed con-
secutively from EpB, selectivity should decrease as the con-
centration of EpB in the feed stream increases. Fig. 2b shows

J.R. Monnier et al. / Journal of Catalysis 225 (2004) 374–380
377
Fig. 4. Epoxidation of piperylene and 1,5-hexadiene over CsCl-promoted,
◦
Ag/α-Al O ; Cs loading is 700 ppm. Reaction temperature is 210 C and
2
3
feedstream is 12% olefin, 17% O , balance n-C H + 2 ppm 2-chlorobu-
2
4 10
−1
tane at GHSV = 6200 h
.
Fig. 3. Epoxidation of isoprene over CsCl-promoted, Ag/α-Al O ; Cs
loading is 700 ppm. Reaction temperature is 205 C and feedstream
2
3
◦
is 12% olefin, 17% O , balance n-C H + 2 ppm 2-chlorobutane at
2
4 10
C=C bonds. Since olefin epoxidation is considered to be due
to the electrophilic addition of an adsorbed O atom to a C=C
bond [23], the CH3-substituted C=C bond should be intrin-
sically more reactive, thus potentially canceling any steric
effect.
−1
GHSV = 6200 h . Epoxidation of butadiene and propylene are also in-
cluded for comparison with the results for isoprene. Conversion of olefin is
denoted as ‘C’ and molar selectivity to olefin epoxide as ‘S’.
that the effect of C4H6 feed concentration more than off-
sets the higher EpB concentration; the selectivity values are
higher for the 17% C4H6 feed concentrations, even though
the EpB concentrations are greater. It is noteworthy that even
though C4H6 feed concentrations have increased by more
than threefold (from 5 to 17% C4H6), the amount of EpB
produced is only slightly higher, again consistent with the
approximate zero-order C4H6 kinetics previously noted [9].
The results in Fig. 2a do suggest, however, that it is essential
to maintain C4H6 concentrations at the end of a reactor tube
high enough to keep selectivity to EpB at desired levels.
The results in Fig. 3 compare the epoxidations of buta-
diene, propylene, and isoprene at 205 ◦C under steady-state
reaction conditions for a CsCl-promoted, Ag catalyst. The
catalyst shows the expected results for epoxidation of both
C4H6 and C3H6. The selectivity of 2.6% for propylene ox-
ide (PO) formation is typical for conventional Ag catalysts,
either promoted or unpromoted. Epoxidation of isoprene
yields both isomers; 2-methyl-3,4-epoxy-1-butene (2-Me-
EpB) represents the nonallylic epoxide and is more similar
to 3,4-epoxy-1-butene (EpB), while 3-methyl-3,4-epoxy-1-
butene (3-Me-EpB) is the allylic epoxide and is more similar
to propylene oxide. Interestingly, the selectivity for the for-
mation of 3-Me-EpB is 8.5 times higher than for PO forma-
tion during C3H6 epoxidation. As noted in Fig. 3, the only
assumption in the data is that the selectivity to 2-Me-EpB
is the same as that for EpB, which seems reasonable given
the structural similarities of the two epoxides. This assump-
tion is necessary to establish CO2 accountability for each
isomeric epoxide. Although not shown in Fig. 3, the total se-
lectivity for the formation of both epoxides was 56.9%, with
the balance being CO2/H2O. Further, the reactivities of both
allylic and nonallylic C=C bonds in isoprene are similar, in-
dicating that steric effects are not controlling reactivity of
the C=C bonds. If steric limitations are present, they may
be offset by the increased electron density of the substituted
Similarly, the results in Fig. 4 compare epoxidation activ-
ities for piperylene and 1,5-hexadiene at 210 ◦C using the
same catalyst as in Fig. 3. Epoxidation of 1,5-hexadiene
was conducted to determine whether it was necessary for
the two C=C bonds to be conjugated for the enhanced al-
lylic epoxidation to take place. No epoxide formation was
observed for this di-olefin; only CO2/H2O was formed. Con-
version of 1,5-hexadiene declined with time on-line and the
results shown are for activity after approximately 1 h of re-
action time. Interestingly, as activity declined, a small peak
corresponding to 1,3,5-hexatriene (from GC/MS analysis)
was observed, indicating that oxidative dehydrogenation of
1,5-hexadiene had taken place. This compound, which is
highly conjugated, is apparently very strongly adsorbed on
the Ag–O surface and leads to catalyst deactivation. How-
ever, during the initial time on-line period, no epoxide was
formed, indicating that conjugated olefins are required for
allylic epoxidation to occur. It is also obvious that conju-
gated olefins can be so strongly adsorbed that all reaction
is suppressed; of course, there may be some concentration
of such conjugated olefins where epoxidation is allowed to
take place. As in Fig. 3, both isomeric epoxides of pipery-
lene are formed; 4,5-epoxy-2-pentadiene is the nonallylic
epoxide and is obviously similar to EpB, while 3,4-epoxy-
1-pentene is the allylic epoxide and represents the propylene
oxide analog. As before, the selectivity for the formation of
the allylic analog is approximately 8.5 times greater than
for PO formation as shown in Fig. 3. Catalyst performance
for the epoxidation of piperylene is very similar to that for
isoprene, although reaction temperature is 5 ◦C higher. The
total selectivity for formation of both epoxides was 56.0%,
very similar to the results for the epoxidation for isoprene.
In order to determine whether the positive effect of the
conjugated olefin structures in Figs. 3 and 4 toward allylic
epoxidation was limited to intramolecular conjugation, or

378
J.R. Monnier et al. / Journal of Catalysis 225 (2004) 374–380
(a)
Fig. 5. Effect of C H on epoxidation of propylene. Catalyst is CsCl-
4
6
promoted, Ag/α-Al O ; Cs loading is 580 ppm and reaction temperature
2
3
◦
is 210 C. Feed composition is 6% C H , 12% O , variable C H balance
3
6
2
−1
4 6,
CH + 1 ppm ethyl chloride at GHSV = 6000 h
.
4
was general to the existence of any conjugated olefin in the
reaction feed stream, varying amounts of C4H6 were added
to a feed stream containing C3H6, O2, CH4 diluent, and
1 ppm ethyl chloride at a reaction temperature of 210 ◦C.
These results, which are summarized in Fig. 5, show that
C4H6 has a similar effect on increasing selectivity to PO at
the expense of overall activity for C3H6 conversion. As the
feed concentration of C4H6 increases from zero to 0.9 mol%,
the selectivity to PO increases from approximately 2.5 to
40%. The concurrent decrease in C3H6 conversion indicates
that C4H6 is much more strongly adsorbed on the Ag–O
surface sites than C3H6, consistent with the observations
of Vannice and co-workers [24]. At C4H6 levels as low as
0.2 mol%, the conversion of C3H6 is lowered by more than
50% even though the ratio of C3H6/C4H6 = 30. Clearly, the
strongly adsorbed C4H6 is limiting both selective and nons-
elective reactions.
(b)
Fig. 6. Reactivity of C H during the C H and C H co-feed experiments.
Catalyst and reaction conditions are the same as for Fig. 5. (a) Relationship
of C H conversion to selectivity for EpB. (b) Comparison of C H and
4
6
4
6
3 6
4
6
4 6
C H reaction rates during co-feed experiments.
3
6
To illustrate even further the powerful effect of C4H6
on selectivity for EpB formation, the plots in Fig. 6 show
C4H6 conversion and EpB selectivity during the C3H6 and
C4H6 cofeed experiments. When C4H6 conversion is 80%
or greater, the selectivity to EpB ranges from 25% (at 100%
C4H6 conversion) to 65% (at 80% C4H6 conversion of a
1% C4H6 feed composition). Selectivity values to EpB of
> 90% are attained only when the concentration of C4H6
in the feed is increased to 6%. At 20% C4H6 conversion of
a 6% C4H6 feed composition, the concentration of C4H6 at
the reactor exit is approximately 4.8%. The curves in Fig. 6b
show again that the conjugated structure of C4H6 greatly
suppresses overall conversion of C3H6 and that the rate of
reaction of C3H6 is completely suppressed for C4H6 concen-
trations greater than 2% in the feed. The curve representing
C4H6 reaction rate as a function of C4H6 feed concentration
also illustrates the transition from first-order to zero-order
C4H6 dependency as the C4H6 concentration increases to
values commonly used for EpB production [9,18].
Fig. 7. Mechanistic scheme for epoxidation of C H to PO showing parallel
3
6
and consecutive pathways for formation of CO and H O.
2
2
Figs. 1 and 2 indicate that nonselectivity arises from the con-
secutive oxidation of EpB and/or its precursor. Thus, the
effect of C4H6 on selectivity is limited to controlling the
extent of the consecutive oxidation of EpB, not in suppress-
ing a parallel pathway for the formation of CO2/H2O from
a different type of adsorbed C4H6. If we argue that the ef-
fect of butadiene (or other conjugated olefin structures) is
manifested similarly during C3H6 (or other allylic olefin)
epoxidation, then the presence of strongly bound C4H6 sup-
presses the overoxidation of PO to CO2/H2O. This is best
viewed by considering the mechanistic scheme shown in
Fig. 7, for the Ag-catalyzed reaction of C3H6 with O2 to
form PO and CO2/H2O. The formation of the reaction prod-
ucts is shown as a parallel-consecutive reaction network, in
the same way that the reaction of C2H4 and O2 is commonly
represented [23].
To more clearly identify the effect of conjugated olefins
on allylic olefin epoxidation, it is helpful to examine again
the effect of C4H6 toward C4H6 epoxidation. The data in

J.R. Monnier et al. / Journal of Catalysis 225 (2004) 374–380
379
Referring to Fig. 7, the selectivity to PO or allylic epox-
ides is enhanced by lowering the rate of the reactions having
rate constants k2 and k3. From analogy to C4H6 epoxida-
tion over similar promoted Ag catalysts, the parallel path-
way with rate constant k2 is not preferentially affected by
the presence of conjugated olefins, although from the over-
all lower activity in the presence of C4H6, it is obvious that
the strongly adsorbed C4H6 is adsorbed on sites leading di-
rectly to both PO formation (k1) and CO2/H2O formation
(k2). The apparent upper level of selectivity in Fig. 5 is ap-
proximately 50% at vanishing C3H6 conversion, very similar
to the corresponding values reported by Gaffney and co-
workers [10–12] and Bowman and co-workers [13,14] for
their heavily modified, high weight loading Ag catalysts.
This suggests that the parallel pathway for CO2/H2O for-
mation occurs at approximately the same rate as PO forma-
tion under normal reaction conditions, thus leading to 50%
selectivity. The observed limit of 50% selectivity for PO for-
mation, obtained by either modification of the Ag surface
during catalyst preparation (Gaffney and Bowman) or by
cofeeding conjugated olefins, possibly suggests a real upper
limit of performance of Ag catalysts for allylic olefin epoxi-
dation. It is also interesting to note that the selectivity values
for formation of the allylic epoxide isomers in Figs. 3 and 4
were limited to 22–23%, considerably below the values ob-
served for PO in the presence of C4H6. This may simply
mean that the intramolecular conjugated olefin structure is
less capable of site blocking Ag–O sites from the allylic moi-
ety than C4H6, which is smaller and is a separate molecule
from C3H6.
allylic –CH3 group by forming Ag ensembles too small
for PO combustion, but not too small for PO formation.
While changing Ag ensemble sizes by adsorption of con-
jugated olefins is desirable for increasing selectivities for
allylic epoxides, ensemble formation greatly decreases the
concentration of available Ag–O surface sites, resulting in
an overall lowering of catalytic activity. It is tempting to also
assume this ensemble effect is occurring for the catalysts
prepared and evaluated by Gaffney and co-workers [10–12],
Bowman and co-workers [13,14], and Jin et al. [15] but di-
rect evidence is lacking.
This model for conjugated olefin addition is not limited to
adsorbed butadiene or other simple conjugated olefins, but is
likely valid for many other olefins containing extended C=C
compounds, such as aromatic compounds. Indeed, Law and
Chitwood [29] claimed in one of the earliest patents for EO
production that aromatic compounds such as benzene and
xylene could be added to the feedstream to improve selec-
tivity for EO formation. The authors claimed that the mod-
erating effect of aromatic compounds was not as long-lived
as was the effect from Cl addition to the Ag catalyst, con-
sistent with strong, reversible adsorption of such aromatic
compounds on the Ag surface. More recently, Evans and
Chipman [30] have also claimed that aromatic compounds
can be added to reaction feedstreams for EO production to
lower the rate of combustion of EO to CO2 and H2O. Thus,
C4H6 may not even be the preferred conjugated olefin ad-
ditive for enhancing selective allylic olefin epoxide forma-
tion. Regardless, the presence of a conjugated olefin, either
intramolecular, as in isoprene or piperylene, or added sep-
arately, as in the case of butadiene, gives much higher se-
lectivities for the formation of allylic epoxides relative to
conventional, supported silver catalysts.
The surface reaction steps summarized in Fig. 8 illus-
trate the proposed role of adsorbed conjugated olefins in
suppressing the combustion of allylic olefin epoxide precur-
sors. The adsorbed PO species is meant only to represent
an adsorbed PO precursor, the structure of which may actu-
ally more closely resemble an oxametallocycle intermediate
of the type proposed by Barteau and co-workers [27,28].
The upper reaction step represents the Ag–O surface state in
the presence of preferentially adsorbed C4H6 (or other ad-
sorbed conjugated olefin, such as isoprene or piperylene),
which limits accessibility of Ag–O species to the reactive
4. Conclusions
The presence of conjugated olefin structures in feed-
streams containing allylic olefins or allylic olefin moieties
results in dramatic increases in selectivities for allylic olefin
epoxides. This enhancement in selectivity is related to the
strength of adsorption of conjugated olefins on Ag–O sur-
faces relative to the more weakly adsorbed allylic olefins.
For the case of C3H6 epoxidation in the presence of as little
as 1% C4H6 in the reaction feedstream, selectivity to PO is
increased from 2.5% to over 40% at the expense of overall
activity for C3H6 conversion. The upper limit of selectiv-
ity to PO in the presence of C4H6 appears to be approxi-
mately 50%, suggesting an upper limit for the effectiveness
of this methodology, since the parallel pathway for direct
combustion of C3H6 to CO2/H2O is not preferentially in-
hibited. Epoxidation of C4H6 on similar Ag catalysts indi-
cates that the consecutive reaction of EpB to CO2/H2O is
strongly limited by the presence of excess C4H6 in the feed-
stream. Rather, the selectivity to EpB is directly proportional
to the amount of C4H6 in the reaction feed stream. Selec-
Fig. 8. Proposed surface intermediate schemes showing effect of strongly
adsorbed C H on lowering combustion of adsorbed PO or PO precursor.
4
6

380
J.R. Monnier et al. / Journal of Catalysis 225 (2004) 374–380
tivities > 90% are obtained only when there is sufficient
C4H6 in the reaction feedstream to control the concentra-
tion of the reactive Ag–O surface. For C4H6 epoxidation, all
CO2/H2O is formed by a consecutive reaction pathway from
EpB; there is no parallel pathway for the direct formation
of CO2/H2O from C4H6. Using the selective epoxidation of
C4H6 as the model for understanding the enhancement in se-
lectivity for allylic olefin epoxide formation, the most likely
reason for improved selectivities is that strongly adsorbed
C4H6 (or other conjugated olefins) limits the ensemble size
of contiguous Ag–O surface sites. These ensembles are too
small for PO combustion, but not too small for PO forma-
tion. The conjugated olefin structure and allylic olefin can
also be coexistent within the same molecule for this desir-
able ensemble effect to occur. Epoxidation reactions of both
isoprene and piperylene give approximately 10 times more
allylic epoxide isomer products than epoxidation of propy-
lene alone.
[4] Y.A. Kalvachev, T. Hayashi, S. Tsubota, M. Haruta, J. Catal. 186
(1999) 228.
[5] G.C. Bond, D.T. Thompson, Catal. Rev.-Sci. Eng. 41 (1999) 319.
[6] R. Meiers, U. Dingerdissen, W.F. Holderich, J. Catal. 176 (1998) 376.
[7] R. Meiers, W.F. Holderich, Catal. Lett. 59 (1999) 161.
[8] M.A. Barteau, R.J. Madix, J. Am. Chem. Soc. 105 (1983) 344.
[9] J.R. Monnier, Stud. Surf. Sci. Catal. 110 (1997) 135.
[10] R. Pitchai, A.P. Kahn, A.M. Gaffney, US patent 5,686,380 (1997), to
ARCO Chemical Technology, L.P.
[11] A.M. Gaffney, C.A. Jones, R. Pitchai, A.P. Kahn, US patent 5,698,719
(1997), to ARCO Chemical Technology, L.P.
[12] A.M. Gaffney, A.P. Kahn, R. Pitchai, US Patent 5,703,254 (1997), to
ARCO Chemical Technology, L.P.
[13] R.G. Bowman, J.I. Maj, H.W. Clark, G.E. Hartwell, S.R. Bare, WO
Appl. 98/00414 (1998), to Dow Chemical Company.
[14] H.W. Clark, J.I. Maj, R.G. Bowman, S.R. Bare, G.E. Hartwell, WO
Appl. 98/00415 (1998), to Dow Chemical Company.
[15] G. Jin, L. Guanzhong, G. Yanglong, Y. Guo, J. Wang, X. Liu, Catal.
Lett. 87 (2003) 249.
[16] R. Van Hardeveld, F. Hartog, Surf. Sci. 15 (1969) 189.
[17] W.H. Flank, H.C. Beachell, J. Catal. 8 (1967) 316.
[18] J.R. Monnier, P.J. Muehlbauer, US patent 4,897,498 (1990), to East-
man Chemical Company.
[19] J.R. Monnier, P.J. Muehlbauer, US patent 4,950,773 (1990), to East-
man Chemical Company.
Acknowledgments
[20] J.L. Stavinoha, J.R. Monnier, D.M. Hitch, T.R. Nolen, G.L. Oltean,
US patent 5,362,890 (1994), to Eastman Chemical Company.
[21] J.R. Monnier, J.W. Medlin, M.A. Barteau, J. Catal. 203 (2001) 362.
[22] R.A. Van Santen, C.P.M. De Groot, J. Catal. 98 (1986) 530.
[23] R.A. Van Santen, H.P.C.E. Kuipers, Adv. Catal. 35 (1987) 265.
[24] M.V. Badani, J.R. Monnier, M.A. Vannice, J. Catal. 206 (2002) 29.
[25] J.R. Monnier, K.T. Peters, US patent 6,388,106 (2002), to Eastman
Chemical Company.
The authors gratefully acknowledge the efforts of Ms.
Libby Cradic in helping prepare this manuscript for publica-
tion and thank Eastman Chemical Company for permission
to publish this material. J.R.M. also appreciates his faithful
Black Labrador Retriever Bruno for staying by his side while
the manuscript was being prepared.
[26] J.W. Medlin, J.R. Monnier, M.A. Barteau, J. Catal. 204 (2001) 71.
[27] J.W. Medlin, M.A. Barteau, J.M. Vohs, J. Mol. Catal. A 163 (2000)
129.
References
[28] J.W. Medlin, A.B. Sherrill, J.G. Chen, M.A. Barteau, J. Phys. Chem.
B 105 (2001) 3769.
[29] G.H. Law, H.C. Chitwood, US patent 2,279,469 (1942), to Carbide
and Carbon Chemicals Corporation.
[1] D.L. Trent, Kirk-Othmer Encyclopedia of Chemical Technology,
vol. 20, Wiley, New York, 1996.
[30] W.E. Evans, P.I. Chipman, US Patent 6,373,925 (2002), to Shell Oil
Company.
[2] Chem. Eng. News 22 October (2001) 19–20.
[3] T. Hayashi, K. Tanaka, M. Haruta, J. Catal. 178 (1998) 566.