Kinetics of Ethylene Epoxidation on a Promoted Ag/α-Al2O3 Catalyst—The Effects of Product and Chloride Co-Feeds on Rates and Selectivity

Article abstract of DOI:10.1002/chem.201801356

The overall chloriding effectiveness factor (Z*), defined as the ratio of ethyl chloride concentration in parts per million to the sum of ethylene and ethane concentration in mole percent multiplied by a weighting factor to account for their efficacy in removing chlorine-adatoms from the surface, was used as a parameter to account for the effects of chlorine on the kinetics of ethylene epoxidation on a highly promoted 35 wt % Ag/α-Al₂O₃ catalyst. An increase in O₂ order (≈0.7 to 1) and a decrease in C₂H₄ order (≈0.5 to <0) with increasing Z* (Z=2.1, 3.4, 5.2, and 8.9) was observed implicating kinetic relevance of O₂ activation on chloride-promoted silver catalysts. Carbon dioxide co-feed (1–5 mol %) was found to promote ethylene oxide selectivity as CO₂ co-feed reversibly inhibits CO₂ synthesis rates (?0.6 order) more than ethylene oxide synthesis rates (?0.49 order) at all Z* values. Ethylene oxide and CO₂ rates were found to be invariant with ethylene oxide (0–0.5 mol %) and acetaldehyde (0–1.7 ppm) co-feeds, suggesting that there is minimal product inhibition under reaction conditions. A model involving a common reaction intermediate for ethylene oxide and carbon dioxide synthesis and two types of atomically adsorbed oxygen species—nucleophilic and electrophilic oxygen—is proposed to plausibly describe the observed reaction rate dependencies and selectivity trends as a function of the chloriding effectiveness.

Full text of DOI:10.1002/chem.201801356

A Journal of
Accepted Article
Title: Kinetics of ethylene epoxidation on a promoted Ag/α-Al2O3
catalyst - the effects of product and chloride co-feeds on rates
and selectivity
Authors: Cha-Jung Chen, James W. Harris, and Aditya Bhan
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201801356
Link to VoR: http://dx.doi.org/10.1002/chem.201801356
Supported by

Chemistry - A European Journal
10.1002/chem.201801356
Kinetics of ethylene epoxidation on a promoted Ag/α-Al
2
O catalyst
3
–
the effects of product and chloride co-feeds on rates and selectivity
Cha-Jung Chen, James W. Harris, and Aditya Bhan*
Department of Chemical Engineering and Materials Science
University of Minnesota - Twin Cities
4
21 Washington Avenue SE
Minneapolis, Minnesota 55455
USA
E-mail addresses: chenx990@umn.edu (C.-J. Chen), harr1707@umn.edu (J. W. Harris), and
abhan@umn.edu (A. Bhan)
*
Corresponding author
1
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Abstract
Overall chloriding effectiveness factor (Z*), defined as the ratio of ethyl chloride concentration
in parts per million to the sum of ethylene and ethane concentration in mole percent multiplied
by a weighting factor to account for their efficacy in removing chlorine-adatoms from the
surface, was used as a parameter to account for the effects of chlorine on the kinetics of
ethylene epoxidation on a highly promoted 35 wt% Ag/α-Al
order (~0.7 to 1) and a decrease in C order (~0.5 to <0) with increasing Z* (Z*= 2.1, 3.4,
.2, and 8.9) was observed implicating kinetic relevance of O activation on chloride-promoted
silver catalysts. Carbon dioxide co-feed (1 – 5 mol%) was found to promote ethylene oxide
selectivity as CO co-feed reversibly inhibits CO synthesis rates (–0.6 order) more than
ethylene oxide synthesis rates (–0.49 order) at all Z* values. Ethylene oxide and CO rates
2
O
3
catalyst. An increase in O
2
2
H
4
5
2
2
2
2
were found to be invariant with ethylene oxide (0 – 0.5 mol%) and acetaldehyde (0 – 1.7 ppm)
co-feeds, suggesting that there is minimal product inhibition under reaction conditions. A
model involving a common reaction intermediate for ethylene oxide and carbon dioxide
synthesis and two types of atomically adsorbed oxygen species – nucleophilic and electrophilic
oxygen – is proposed to plausibly describe the observed reaction rate dependencies and
selectivity trends as a function of the chloriding effectiveness.
2
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
1
. Introduction
Organic chlorides act as promoters for Ag-catalyzed olefin epoxidation reactions.[1–8]
Industrial catalysts for selective oxidation of ethene to ethylene oxide (EO) are typically Ag/α-
2
-
Al
2
O
3
formulations promoted with anions (e.g., Cl, SO
4
) and cations (e.g., Cs, Re, Mo, Li) in
combination with continuously fed chloride promoters (e.g., ethyl chloride), which catalyze
[
9,10]
EO synthesis at ~90% carbon selectivity.
The introduction of chlorine by five doses of
dichloroethane at 0.05 cm (He: dichloroethane = 40:1) to an unpromoted 12 wt% Ag/α-Al
catalyst at 523 K and ambient pressure measured under transient flow conditions resulted in
3
2
O
3
[
5]
an increase in EO selectivity from ~43% to 81%. Density functional theory (DFT)
calculations by Lambert and co-workers[ showed that the energy barrier for EO formation
on a Ag(111) surface covered with Cl (0.25 monolayer (ML)) and O (0.25 ML) is 0.39 eV less
than that on a clean Ag(111) surface, which is 0.07 eV lower than that for acetaldehyde
11]
formation, a precursor to CO , suggesting that the promotional effect of chlorine on EO
2
selectivity over silver catalysts is due to an electronic effect. Campbell and co-workers
demonstrated using combined surface science studies at ultra-high vacuum (UHV) conditions
(
X-ray photoelectron spectroscopy, Low-energy electron diffraction, and Auger electron
spectroscopy) and ethylene oxidation reactions in a batch microreactor at sub-ambient
pressures that apparent O (increases from 0.2 to 0.9) and C (ranges between 0.4 and 0.8)
reaction orders and apparent activation energies (decreases from ~50 to ~8 kJ mol ) for EO
and CO formation vary with increasing chlorine coverage (0 – 0.7) on single crystal Ag(110)
and Ag(111) surfaces.
Ag/α-Al
catalysts (~453-473 K) during epoxidation reactions,[
2
2
H
4
-
1
2
[
2,3]
Ethane, butane, and 3,4-epoxy-1-butene remove adsorbed Cl from
12,13]
2
O
3
suggesting the
concentration of an organic chloride co-feed is not solely responsible for modulating the
surface chlorine coverage on Ag catalysts useful for olefin epoxidation. Further, the presence
of an organic chloride in the feed without an alkane moderator (e.g. ethane) results in
[
14,15]
irreversible poisoning of ~90% of the reaction rate over Ag surfaces.
An overall chloriding effectiveness factor (Z*, Eq. (1)) was developed by The Dow
[
16]
Chemical Company
to account for both chlorine deposition by organic chlorides and
chlorine removal by hydrocarbons. The numerator in the definition of Z* reflects the effective
chloriding concentration of all the organic chlorides present in the feed stream in terms of ethyl
chloride concentration in ppmv, i.e., l ppmv of ethylene dichloride is equivalent of 2 ppmv of
ethyl chloride, and l ppmv of vinyl chloride is equivalent of 1 ppmv of ethyl chloride. The
denominator in the definition of Z* signifies the dechloriding efficacy of non-chlorine
containing hydrocarbons in the feed stream represented as its molar percent multiplied by a
weighting factor, an empirical value determined to represent its equivalent dechloriding
effectiveness as ethane, e.g., the weighting factor for ethane is one, ethylene is 0.01, and
[
9]
methane is 0.002. A similar parameter, Q, was developed by Shell, wherein Q represents the
ratio of the effective molar quantity of gas-phase reaction modifier to the effective molar
quantity of hydrocarbons present in the feed stream. In this work, we use Z* as a descriptor of
chloriding effectiveness for ethylene oxidation to systematically probe the effect of reactant
and product co-feeds on catalytic rates and selectivity as a function of Z*.
3
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
퐸푡ℎ푦푙 푐ℎ푙표푟푖푑푒 푒푞푢푖푣푎푙푒푛푡 (푝푝푚푣)
퐸푡ℎ푎푛푒 푒푞푢푖푣푎푙푒푛푡 (푚표푙 푝푒푟푐푒푛푡)
푍∗ =
Eq. (1)
Force and Bell[17] showed that the rate of ethylene oxidation to either EO or CO
silica supported silver catalyst (1 atm, 443 K – 493 K) was first order in C pressure and
over a
2
2
H
4
half order in O pressure. Similar results were reported for non-promoted single crystal silver
2
crystals[ and silver sponge catalysts at sub-ambient and ambient pressures at temperatures
18]
[19]
ranging from 435–540 K. These results suggest that a common reaction intermediate, likely an
[
20–25]
oxametallacycle as inferred from reports by Barteau and coworkers,
is involved in
ethylene oxidation on silver catalysts, however, it is still unclear how the extent of chlorine
modification affects (i) branching of the common reaction intermediate and thereby selectivity,
and (ii) the identity and concentration of adsorbed oxygen species.
The precise nature of the oxygen species required for selective oxidation of ethylene has
been extensively studied over the past 50 years. Early reports from surface science and
microreactor studies by Campbell and co-workers suggested that molecular oxygen was
involved in ethylene epoxidation.[
2,26,27]
Since that time, Lambert and Grant reported that
Ag(111) single crystal surfaces exposed to oxygen and subsequently heated to evolve adsorbed
, but not oxygen adatoms, could be exposed to ethylene to form ethylene oxide during
temperature-programmed reactions and in the same quantity as if the O removal step was not
Reactions of nitrous oxide and ethylene form ethylene oxide over a Ag sponge
O
2
2
[
28]
performed.
[
29]
catalyst (507 K), and sequential pulses of oxygen and ethylene react to form ethylene oxide
over Ag powder (523 K) despite the time between pulses (25-400 ms) exceeding the time
required for O
desorption (~0.4 ms),[ suggesting atomic O adsorbed on Ag catalysts reacts
30]
2
1
8
with ethylene to form ethylene oxide. Reactions of ethylene and O
2
over Ag powder pre-
1
6
16
18
treated with O
2
(475 K) formed C
2
H
4
O before C
2
H
4
O during temperature programmed
1
6
surface reaction experiments, demonstrating that reaction with surface O is faster than with
gas-phase O and thus that atomic oxygen is relevant for ethylene epoxidation. Likewise,
1
8
[31]
2
computational studies have shown that reaction coordinates involving oxametallacycle
transition states formed from ethylene and atomic oxygen accurately predict observed EO
[
21–23]
selectivities over unpromoted Ag powders (~40%, ~450 K).
The combined experimental
and theoretical evidence supports the conclusion that oxygen adatoms, and not adsorbed
molecular oxygen, are required for ethylene epoxidation over Ag catalysts.
While there is broad agreement regarding the involvement of O-adatoms in ethylene
epoxidation catalysis, a variety of atomic oxygen species are present on Ag catalysts during
reaction. X-ray photoelectron spectra (XPS spectra) of a polycrystalline Ag foil exposed to 95
Pa O
2
and 5 Pa C
2
4
H at 470 K evidence two types of atomic oxygen species, either nucleophilic
(
O
nucl) or electrophilic (Oelec), which give rise to XPS peaks with O1s binding energies (BEs)
[
32]
of ~528.4 eV and ~530.5 eV, respectively. Ethylene adsorption and subsequent temperature
programmed surface reaction on a polycrystalline silver foil prepared with Onucl that was
1
8
16
16
entirely O and Oelec that was entirely O resulted exclusively in formation of C
2
H
4
O and
, suggesting only Oelec participates in selective oxidation. Both species were required
to form ethylene oxide however, as ethylene did not adsorb to a silver foil prepared without
18
[32]
C O
2
[
32]
O
nucl
.
In situ XPS studies and DFT calculations on Ag(110) showed that two different
4
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
oxygen species can be active for EO formation: (i) the aforementioned electrophilic oxygen
species (Oelec, BE(O1s) ~530.5 eV), as well as (ii) an atomic oxygen species on an
unreconstructed silver surface (BE(O1s) ~527.9 eV).[33,34] These unreconstructed O atoms are
highly reactive and become unstable at oxygen coverages greater than 0.02 monolayers (ML),
[33]
and thus could be undetectable by in situ techniques under ethylene epoxidation conditions.
We build on these precepts in the literature regarding different oxygen species on the surface
of Ag catalysts to postulate a role of surface chlorine coverage on the relative densities of the
types of adsorbed oxygen species and consequently on the rate and selectivity of ethylene
epoxidation.
In this report, we employ the chloriding effectiveness value (Z*) as a parameter to evaluate
the extent of chlorine modification under ethylene oxidation conditions, and we report the
kinetics of ethylene oxidation on a highly promoted 35 wt% Ag/α-Al
2.1 – 8.9) at ~5.3 bar and 513 K under excess ethylene (C :O molar ratio 10:1), conditions
similar to those used in industrial practice. We show that, for both EO and CO formation, the
oxygen reaction order increases from ~0.7 to 1 and the ethylene reaction order decreases from
0.5 to subzero as the extent of chlorine modification (Z*) increases. Carbon dioxide inhibits
both EO and CO production while also increasing EO selectivity, and the CO reaction orders
for both EO and CO synthesis are invariant with Z* and less than zero. The observed reaction
2
O
3
catalyst at various Z*
(
2
H
4
2
2
~
2
2
2
rate dependencies and selectivity trends as a function of the chloriding effectiveness can be
rationalized in terms of changes in the identity and concentration of oxygen species present on
the surface.
5
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
2
. Materials and Methods
2
.1. Catalyst composition and site density
A promoted 35 wt% Ag/α-Al catalyst provided by The Dow Chemical Company was
used throughout this work. The reported specific surface area is 1.25 m g and the promoters
[
10]
2
O
3
2
-1
2
-
include, Cs (628 ppm), Li (33 ppm), Na (42 ppm), Re (440 ppm), [SO
4
] (151 ppm), and Mn
(115 ppm). The catalyst was pelletized and sieved between 40-80 mesh (177 – 420 μm).
35]
36]
Surface silver site density was measured using N
2
O adsorption-decomposition,[
[
performed in a recirculating batch reactor system identical to that reported in a prior report.
Briefly, 0.525 g of 35 wt% Ag/α-Al
inner diameter), pretreated in H (Minneapolis Oxygen, 99.999%) to 513 K with a ramping
rate of 0.12 K s and held for 20 h, then cooled to 443 K in H flow before dosing 15.9 μmol
of N O (medical grade, 99%, Matheson). The effluent was sampled and separated using an
2
O (40-60 mesh) were loaded to a quartz tube (0.4 cm
3
2
-1
2
2
Agilent 6890 gas chromatograph (GC) with an HP-Plot/Q column (30.0 m × 320 μm × 0.20
μm) and quantified using a thermal conductivity detector (TCD).
2
.2. Kinetics of ethylene epoxidation
Measurements of ethylene oxidation kinetics were conducted in a stainless steel tubular
reactor (McMaster-Carr, OD 6.36 mm, ID 4.572 mm) loaded with 0.020 g of 35 wt% Ag/α-
Al (40-60 mesh), which was inserted in between the two split-tube copper sleeves heated
2
O
3
by two cartridge heaters (Omega Engineering Inc., CIR-2100, 600 W) regulated by a Watlow
temperature controller (96 Series). Reactions were conducted at ~5.3 bar total pressure and 513
3
-1
-3
K with a GHSV of ~500,000 cm h cmcat . The gases used were O
ethylene (Matheson, ≥99.999%), ethyl chloride (80 ppm in balance He, Matheson, Certified
Grade), ethane (10% in balance He, Matheson, Gravimetric Grade), CO (10% in balance He,
2
(Matheson, >99.997%),
2
Matheson, Gravimetric Grade), ethylene oxide (1.5% in balance He, Matheson, Certified
Grade), acetaldehyde (300 ppm in balance He, Airgas, Primary Standard Grade), and He
(Minneapolis Oxygen, 99.997%). All gases other than He, which was fed through a moisture
trap prior to use, were used without further purification and introduced to the reactor via mass
flow controllers. The overall chloriding effectiveness factor (Z*) defined in Eq. (1) was used
as a parameter to represent the extent of chloride promotion during kinetic studies. The influent
reaction mixture at the reference condition at the four Z* conditions investigated (Z*= 2.1, 3.4,
5
.2, 8.9) was comprised of 3 mol% O
various ethyl chloride (C Cl) concentrations (1.5, 2.4, 3.8, and 6.3 ppm for Z*= 2.1, 3.4, 5.2,
and 8.9) to achieve the target Z* value (balance in He). The range of concentrations
investigated in O , C , CO , EO, and acetaldehyde co-feeds is summarized in Table 1.
Carbon dioxide (1 mol%) was co-fed in the influent mixture to ensure CO concentrations
varied <10% across the catalyst bed, as CO was reported to have inhibitive effects on both
EO and CO
synthesis on both promoted and non-promoted silver catalysts.[
2
, 0.4 mol% C
2
H
6
, 30 mol% C
2
H
4
, 1 mol% CO , and
2
2
H
5
2
2
H
4
2
2
2
17,37]
2
The
composition of the reactor effluent was analyzed by a thermal conductivity detector connected
in series with a flame ionization detector using an online gas chromatograph (GC, Agilent 7890)
with a HP-Plot Q column (HP-Plot Q, 30 m × 320 μm × 20 μm). Ethylene conversion was <
1
.5 % in all cases, and the carbon mass balance (carbon in divided by moles carbon out) always
closed to within ±5%, which is within the error of quantification by GC and of the measured
6
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
mass flow rates. The temperature gradient across the catalyst bed was estimated as < 1 K using
[
38]
GradientCheck. The parameters used for the estimation are reported in Section S.1 of the
Supporting Information (SI).
The patent literature reports that fresh samples of promoted silver catalysts are subject to
[
39,40]
days-long “break-in” periods during the start-up phase of the operation.
A 51-hour start-
up procedure, involving holding the reactor temperature at 483 K for 35 h, and subsequently
increasing the temperature to 489 K for 16 h before raising the temperature to 491 K for steady
state measurements for 70+ hours, was implemented in a recent report by Monnier and co-
[
41]
workers,
during which EO rate and selectivity are transient. We addressed the slow
evolution of the catalyst under reaction conditions by standardizing our kinetic measurements
and analysis approach as follows: (i) a reference condition was used in every kinetic study and
this condition was restored between each new condition (gas phase pressures or temperatures)
in order to account for the activation/deactivation of the catalyst with time on stream, (ii) each
condition of temperature or pressure was maintained for 12-hours to allow the reaction rates
to reach an apparent “steady state” before changing any reaction parameter, (iii) the reaction
orders and the effect of chlorine modifications were assessed based on adjusted rates and
selectivity (Eqs. 2 and 3), i.e., the measured rates/selectivity (Rnew) were adjusted in reference
to the rates/selectivity measured at the reference condition after 12 hours on stream (R
the rates/selectivity measured at the reference condition prior to the new reaction condition
new-1). An example of the EO and CO synthesis rates measured as a function of time on
stream using this approach is shown in Figure S3.
1
) and
(R
2
ꢃ푎푡푒ꢄ1
푅ꢀꢁꢂ푎푑푗 = 푅ꢀꢁꢂ푛푒푤 × ꢃ푎푡푒
Eq. (2)
ꢄꢅꢆꢇ−1
푆ꢂꢈꢂꢉꢁꢊꢋꢊꢁꢌ푎푑푗 = (푆ꢂꢈ푛푒푤 ꢍ 푆ꢂꢈꢃ푛푒푤ꢎꢏ) + 푆ꢂꢈꢃꢏ Eq. (3)
3
. Results and Discussion
3
.1. Site density probed by N
2
O adsorption-decomposition
Nitrous oxide adsorption-decomposition has been used previously to quantify surface Ag
[
35,42]
sites on pre-reduced Ag catalysts.
The adsorption stoichiometry is considered to be one
oxygen atom per surface Ag atom based on the observation that adsorption-decomposition of
35]
N
2
O on pre-reduced Ag powder resulted in ~ 0.9 mol O adsorbed per mol surface Ag.[ The
amount of N O converted as a function of reaction time while recirculating dilute N O in a
batch reactor is shown in Figure 1. The amount of dinitrogen formed as a function of reaction
time correlates with the amount of N O consumed; the disappearance of N O was employed
to quantify the extent of N O adsorption-decomposition. Total oxygen uptake was found to be
5.2 μmol gcat , which corresponds to a 0.47% surface silver dispersion, assuming N O to Ag
ratio (O/Ag) to be unity. This value is comparable to that reported by Monnier and co-
2
2
2
2
2
-1
1
2
[
41]
[43]
workers and Vannice and coworkers measured using O chemisorption at 443 K on non-
2
promoted Ag/α-Al
2
O
3
catalysts (~0.2 – 0.95%).
7
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
3
.2. Effect of Z* on EO rates and selectivity
The effect of chlorine modification on EO formation rates and selectivity on the promoted
3
5 wt% Ag/α-Al
2
3
O catalyst was investigated by systematically varying Z* from 2.1 to 8.9,
with Z* = 2.1 taken as the reference condition, as this value falls within the optimal range of
Z* (Z* = 1-15) reported in the patent literature.[ The EO formation rate decreased by ~40%
16]
while CO
CO rates were restored upon returning to the reference condition (Z* = 2.1, Figure S3),
demonstrating that chlorine modification reversibly inhibits EO and CO formation rates in the
presence of ethane co-feed, while modification by organic chloride co-feeds in the absence of
2
rate decreased by ~63% as Z* was varied from 2.1 to 8.9 (Figure 2). Both EO and
2
2
[
14,15]
alkane moderators has been reported to be irreversible.
formation rate than the EO formation rate as Z* increased from 2.1 to 8.9 resulted in an increase
in EO selectivity (from 81.7% to 86.6%) and a decrease in CO selectivity (18.2% to 13.4%).
These results mirror those reported for chlorine modification of single-crystal,
Stronger inhibition of the CO
2
2
[
2,3]
pure
via pre-adsorption or dosing of an organic chloride
over a wide range of reaction conditions. The same effects of chlorine on EO and CO rates
and selectivity were observed at space velocities of ~250,000 cm h cmcat (~2 % conversion,
Figure S4) and ~100,000 cm h cmcat (~4% conversion, Figure S5). These effects of chlorine
modification on promoted silver catalysts parameterized in terms of Z* to consider both
deposition and removal of Cl-adatoms, demonstrate that (i) EO selectivity is promoted as a
result of disproportionate inhibition in epoxidation and total oxidation rates by Cl-containing
species, and (ii) chlorine modification reversibly inhibits ethylene oxidation rates.
[
4,8]
[5–7]
silver, and supported silver catalysts
2
3
-1
-3
3
-1
-3
3
.3. Kinetics for ethylene epoxidation
The O reaction order for ethylene oxidation was measured by varying the O
concentration non-monotonically between 2 – 7 mol%.
to increase monotonically from ~0.7 to ~1 as Z* was varied from 2.1 to 8.9 (Figure 3) for both
2
2
inlet
reaction order was found
[
1,16,39]
The O
2
[
3]
EO and CO
of O was noted to decrease by 4.6 kJ mol with increasing chloride coverage (0< θCl <0.65)
on chlorine-modified Ag(111) as measured by thermal desorption spectra of O . The selectivity
for EO and CO was invariant with O concentration at all Z* conditions (Figure 4), as well as
with C pressure and temperature (vide infra), observations that suggest that the two
products form from a common intermediate and that branching from this intermediate to EO
or CO occurs after the kinetically relevant step(s). Microkinetic models of ethylene
2
formation, consistent with a report by Campbell, in which the heat of adsorption
-1
2
2
2
2
2
H
4
2
epoxidation over Ag(111) including oxygen dissociation and formation of an oxametallacycle
intermediate, formed from adsorption of ethylene on oxygen adatoms, accurately capture
experimentally measured trends in EO formation rates and selectivity as a function of oxygen
[
44]
and ethylene pressure.
Experimental evidence for plausible oxametallacycle intermediates
[
45]
over Ag catalysts was presented by Medlin et al., who showed that a stable oxametallacycle
[
46]
was formed when Ag(110) was exposed to 1-epoxy-3-butene at 300 K and by Jones et al.
who showed 2-iodoethanol reacts at 263 K on Ag(110) to form a stable oxametallacycle with
O stoichiometry. Linic and Barteau proposed two types of oxametallacycle intermediates
as plausible candidates for the common intermediate for EO and CO production over Ag
C
2
H
4
2
catalysts on the basis of DFT calculations of non-promoted Ag15 clusters, each composed of
8
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
an atomic oxygen bound to two carbons on a metal surface ( –O–C–C– )[
21–23]
including either
[
47]
two Ag atoms (Ag-O-C-C-Ag, OMME), as was postulated by Jones et al., or one Ag atom
20]
(
Ag-O-C
The ethylene oxide selectivity at Z* = 8.9 was found to exceed the theoretical upper bound
for EO selectivity (85.7%), reached under the assumption that only molecular oxygen is
2
, OME), as was observed by Medlin et al.[
formation.[
2,48]
If only molecular
involved in EO formation and atomic oxygen results in CO
is incorporated into EO, then EO formation from ethylene and O
2
O
2
2
should result in deposition
of an oxygen atom on the catalyst surface. Stoichiometric removal of oxygen adatoms in
combustion pathways requires formation of six molecules of EO and consumption of seven
O).[
49]
Early reports by Campbell
molecules of C
and co-workers
2
H
4
(6O
2
+ 7C
2
H
4
→ 6C
2
H
4
O + 2CO
2
+ 2H
2
[
2,26,27]
showed from their combined surface science studies at UHV conditions
and ethylene oxidation reactions in a batch microreactor at sub-ambient pressures that adsorbed
molecular oxygen on a single silver crystal surface is active for ethylene oxidation. Grant and
[
28]
Lambert, however, argued that atomically adsorbed oxygen (O
oxidation, as the yield of EO and CO was found to be approximately proportional to O
coverage and independent of O2(a) coverage on a Ag(110) surface. Experimental work by Noel
a
) participates in ethylene
2
a
[
29]
and Cant
showed that atomic oxygen deposited on a Ag sponge via N O adsorption-
2
decomposition is active for ethylene epoxidation; similar observations were noted by Roberts
[
50]
and Madix for partial oxidation of norbornene to norbornene oxide by atomically adsorbed
oxygen on Ag(110). The monotonic increase in O reaction order with increasing Z* and the
identical pressure dependence of selective and unselective oxidation pathways on O partial
pressures reported here suggests that (i) an atomic oxygen species derived from molecular O
is involved in the kinetically relevant step resulting in a kinetic dependence on gas phase
oxygen pressure; (ii) a common reaction intermediate is involved in EO and CO formation;
formation of this intermediate is the kinetically-relevant step, and this intermediate forms EO
and CO via branching in steps that occur after the rate-limiting step; and (iii) the surface
2
2
2
2
2
coverage of oxygen-containing species decreases as Z* increases, resulting in reaction orders
that increase from 0.7 to unity with increasing Z*.
The effect of C
by varying C concentration between 20–40 mol%, while varying the C
accordingly to maintain a constant Z* value. EO and CO synthesis rates were found to have
2
H
4
concentration on EO and CO
2
rates at each Z* value was investigated
2
H
4
2
H
5
Cl concentration
2
nearly identical ethylene pressure dependencies at each Z* investigated, while pressure
dependencies varied as a function of Z* (Figure 5). The nearly identical reaction orders for
both EO and CO2 as a function of ethylene pressure at each Z* resulted in EO and CO
selectivities that were invariant with ethylene concentration (Figure S6). Similar to the effect
of oxygen pressure, the constant EO and CO selectivity with varying C pressure suggests
that EO and CO form via a common intermediate. Common reaction orders for EO and CO
formation were reported previously by Force and Bell in a batch recycle system (101 kPa,
2
2
2
H
4
2
2
[
17]
[
18]
4
43 K – 493 K) on a silica supported silver catalyst, by Campbell on single crystal Ag(110)
[
19]
and Ag(111) surfaces (~22 kPa, 490-539 K), and by Cant and Hall on silver sponge catalysts
101 kPa, 430-480 K). The apparent activation energies (Eapp) for both EO and CO synthesis
(
2
measured between 508 K – 523 K were found to be similar at Z*=8.9 (~211.8 ± 24.9) and at
Z*=2.1 (~104.6 ± 7.0) (Table 2, Figures S7 and S8), consistent with prior microkinetic
[
44]
modeling and experimental work for ethylene oxidation on non-promoted silver surfaces
9
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
and on a Ag-Cs/α-Al
2
O
3
catalyst[51]. The nearly identical reaction orders and apparent
activation energies for EO and CO
intermediate is required for EO and CO
synthesis is not reflected in the measured kinetic parameters.
2
synthesis in this study demonstrate that, a common reaction
2
synthesis, and that branching for EO and CO
2
At Z* = 2.1, EO and CO synthesis rates are positive order (~0.5) in ethylene pressure
Figures 5 and 6). Ethylene order decreases to zero for Z* values of 3.4 and 5.2. Ethylene
inhibits EO and CO formation rates at higher ethylene concentration (> 30 mol%) at Z*=8.9
Figures 5 and 6). This result is consistent with prior reports that noted a decrease in ethylene
reaction order (from ~0.8 to ~0.5) for EO formation with increasing chloride coverage (θCl
from 0 < θCl <0.3) and ascribed this observation to an increase in the heat of adsorption of C
with increasing chlorine coverage as assessed from thermal desorption spectra of ethylene on
2
(
2
(
,
2
H
4
Ag(110) surfaces pre-adsorbed with atomic chlorine.[
of adsorption for C with increasing Z* can be rationalized based on an increase in the
concentration of electrophilic oxygen or Ag, which favors the adsorption of an electron donor
2,26]
We postulate that the increasing heat
2
H
4
[52,53]
(
Lewis base) like C
2
H
4
.
Increased density of electrophilic oxygen on Ag powder with
increasing Cl coverage during ethylene epoxidation at 0.3 mbar, 503 K, was reported by Rocha
[8]
et al., suggesting increased density of Cl-adatoms could be responsible for the increased
coverage of C at high Z*. Alternatively, Campbell and Paffett reported increases in the
work function of Ag after adsorption of chlorine on Ag(110), suggesting significant negative
2
H
4
δ+
[27]
charge on Cl atoms presumably accompanied by formation of Ag sites.
coverage up to 0.3 monolayers led to increased C adsorption at 134 K as quantified by
temperature programmed desorption, suggesting increased numbers of C adsorption sites
were formed by the presence of Cl on the surface, which the authors posit are electrophilic Ag
Increasing Cl
2
H
4
2
H
4
[27]
sites capable of accepting π-electrons from C
2
H
4
.
These results suggest that chlorine
modification alters the adsorption properties of ethylene by reducing the electron density of
the Ag surface or of surface oxygen adatoms. We develop a plausible mechanistic description
of chloriding effectiveness and its effect on reaction rate dependencies on O
terms of electrophilic and nucleophilic oxygen adatoms in Section 3.4.
2
and ethylene in
EO selectivity increased by ~2% as CO
at all Z* conditions as a result of CO co-feed inhibiting total combustion (~ –0.6 order in CO
formation) more than EO formation (~ –0.49 order in EO formation) (Figures 7 and 8). A
2
co-feed concentration increased from 1 to 5 mol%
2
2
[
37]
similar effect of CO
2
promotion on EO selectivity has also been reported by Metcalf et al.
on a Ag/α-Al
2
O
3
catalyst in a packed bed reactor, and by Force and Bell[17] on a silica supported
silver catalyst in a batch recycle system at ambient pressure and 493 K, without chlorine
modification. We propose four possible non-exclusive mechanisms for the observed promotion
in EO selectivity by CO
2
co-feed. First, the promoting effect of CO co-feed on EO selectivity
2
was ascribed to an ensemble effect[ with the postulate that ethylene combustion requires a
2]
larger ensemble of available sites with adsorbed oxygen than ethylene epoxidation. Second,
CO
2
could preferentially adsorb on oxygen vacancy sites in a surface silver oxide layer where
ethylene combustion preferentially occurs during ethylene oxidation as postulated by Ӧzbek
[
54]
et al. based on DFT calculations. Third, the observed increase in EO selectivity with CO
pressure could result from loss of atomically adsorbed nucleophilic oxygen required for total
oxidation as CO likely preferentially adsorbs to nucleophilic oxygen to form a carbonate
species as observed in infrared spectra of Ag/SiO
2
2
[55]
2
and by EELS and XPS on Ag(110) single
1
0
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
crystals.[56,57] Finally, it is also possible that CO
acts similarly to adsorbed Cl by drawing
2
electron density away from the surface, resulting in increased electrophilicity of O adatoms in
the vicinity of adsorbed CO and increased selectivity to EO. Consistent with this hypothesis,
we note that a monotonic increase in EO selectivity is observed when a modified form of Z*
is considered in which CO acts as an electronic promoter, akin to chlorine adatoms on the
surface (Figure S10). Our results on promoted 35 wt% Ag/α-Al with co-feed of an organic
chloride are consistent with these reports in the literature and highlight that EO selectivity can
be tuned by co-feeding CO and that a kinetic description of EO synthesis must include
inhibition of oxidation rates by CO
2
2
2
O
3
2
2
.
Carbon dioxide was found to reversibly inhibit ethylene epoxidation as both EO and CO
synthesis rates were restored when CO co-feed concentration was restored to the reference
condition (1 mol% CO , Figure S9). The reaction orders for both EO and CO synthesis (~ –
.6 order in CO and ~ –0.49 order in EO) were found to be invariant with Z* (Figure 6),
implying that adsorption properties of CO on the highly promoted silver catalyst are
unchanged in the range of ethyl chloride co-feed concentrations investigated. The invariant
CO reaction order with Z* can be rationalized by preferential and proportional CO adsorption
on the nucleophilic oxygen required for ethylene combustion across the range of Z*
values investigated. CO adsorption may also occur on nucleophilic oxygen adatoms
neighboring oxametallacycle intermediates, occupying adjacent sites required to form
2
2
2
2
0
2
2
2
2
[17,44,58]
2
[
27]
ensembles of surface oxygen adatoms required for ethylene combustion, thus preventing
further oxidation reactions.
The effect of EO co-feed on EO and CO synthesis rates was examined by varying EO co-
2
feed concentration between 0.0–0.5 mol%, which is comparable to that measured from the
reactor effluent at the reference condition (< 0.1 mol% of EO). EO co-feed had no effect on
EO and CO
negligible amount of EO co-fed was converted to CO
However, varying the space velocity resulted in a concurrent decrease in EO selectivity (0.5 –
2
rates and selectivity at Z* values of 2.1 and 8.9 (Figure 9), indicating that a
2
under these reaction conditions.
1
~
.6%) and increase in CO selectivity (0.5 – 1.6%) as space velocity was decreased from
500,000 cm h cmcat to ~100,000 cm h cmcat (Figure 10). There are two plausible
2
3
-1
-3
3
-1
-3
reasons for the observed change in EO and CO
presence of a sequential reaction pathway for EO combustion to CO
conditions investigated, and/or (ii) a change in the branching ratio of the common reaction
2
selectivity as space time increases: (i) the
[19,23]
2
under the reaction
[59]
intermediate with space time similar to an observation by van den Reijen et al. wherein EO
selectivity, measured on non-promoted Ag/α-Al catalysts with silver particle sizes between
0–200 nm at 453–543 K, decreases with increasing ethylene conversion. We note that only
.6–1.8% of EO formed has to undergo combustion to achieve the observed 0.5–1.6% increase
selectivity, which is well below ~10% EO combustion measured by Cant and Hall[19]
2
O
3
2
0
in CO
2
on an unsupported silver sponge formulation without chlorine modification at 443 K and
ambient pressure.
Acetaldehyde co-feed concentrations (0 – 1.7 ppm) comparable to the amount of
acetaldehyde measured at the reactor effluent (0.17 – 0.32 ppm) were found to have no effect
on EO and CO rates and selectivity at Z* values of 2.1 and 8.9 (Figure 11). Although there is
2
no effect of acetaldehyde co-feed on ethylene oxidation on the highly promoted silver catalyst,
acetaldehyde conversion between 5–80% was observed (Table S3) in the co-feed experiments,
1
1
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
suggesting that some acetaldehyde is being converted, most likely to CO
2
.[23,54,60,61] A similar
trend was observed in the space time study, in which acetaldehyde selectivity (0.02–0.06%) at
all Z* conditions was found to decrease with increasing space time (Figure 12). These results
clearly demonstrate that, on a promoted silver catalyst, (i) acetaldehyde co-feed does not affect
the kinetics of ethylene epoxidation, and (ii) acetaldehyde is most likely converted to CO
2
under ethylene epoxidation conditions, and it does not contribute to an increase in CO
selectivity due to the low concentration of acetaldehyde in the system.
2
3
.4. Discussion of possible roles of Cl on surface oxygen speciation
While the kinetic study presented here does not provide direct evidence as to the nature of
the oxygen species present under reaction conditions, we discuss here our results in context of
[
62,63]
spectroscopic and computational results reported in the literature. Schlögl and co-workers
demonstrated that the yield of EO, monitored by proton-transfer reaction mass spectrometry,
correlates with the amount of electrophilic oxygen (O1s/Ag3d5/2) measured in-situ using XPS
on a polycrystalline Ag foil at 420 K. A concurrent increase in the atomic percentage of Cl
(from 1.7 to 7.9%, O decreased from 16.7 to 14.9% with balance Ag, determined by XPS with
a probing depth of ~0.6 nm), in EO selectivity (from 9.5% to 55%), and in the ratio of
electrophilic (O1s binding energy 530-531 eV) to nucleophilic (528-529 eV) oxygen species
(
from 1.24 to 2.07) was observed as the number of ethylchloride (C
2
H Cl) pulses was varied
5
-1
from zero to three during ethylene epoxidation on a commercial Ag powder (~120,000 h
space velocity, 0.3 mbar total pressure, 503 K).[ Subsequent studies under UHV conditions
8]
demonstrated that a third species, with O1s binding energy < 528 eV, was formed on Ag(110)
at <0.04 monolayer oxygen coverage which rapidly (<180 s) recombined to form nucleophilic
oxygen with increasing exposure time.[
33]
DFT calculations predicted this species to be
unreconstructed oxygen adatoms which could form EO and acetaldehyde via an OMME-type
intermediate, though with low selectivity to EO.[ Bukhtiyarov et al. showed on the basis
of experiments with isotopically labeled oxygen over a silver foil that a nucleophilic oxygen
species with O1s binding energy of 528.4 eV participated solely in total oxidation, while an
electrophilic oxygen species with an O1s binding energy of 530.5 eV performed ethylene
epoxidation. On the basis of these results, it seems likely that three oxygen-containing species
may exist on the surface of Ag during ethylene oxidation: an unreconstructed oxygen species
with low surface residence time, nucleophilic oxygen, and covalently bound electrophilic
oxygen. We postulate that the relative concentration of these types of atomically adsorbed
33]
[32]
oxygen on the catalyst potentially depends on Z*, the concentration of electrophilic O
a
increases while the concentration of nucleophilic O
Z* increases from 2.1 to 8.9.
a
and/or unreconstructed O decreases as
a
The observed increase in O
to a decrease in O surface coverage as a result of (i) an increase in chloride coverage, and/or
ii) an increase in the concentration of electrophilic surface oxygen. We hypothesize that as the
concentration of electrophilic oxygen increases, the adsorption of an electron donor (Lewis
2
order with increasing Z* (Figures 3 and 6) could be ascribed
x
(
[
52]
base) such as ethylene is favored over a Lewis acidic molecule such as O
2
.
This would result
in increased reaction orders for O with increasing Z*, as we report here (Figure 6).
2
Alternatively, as mentioned in Section 3.3, increasing chloride coverage may lead primarily to
δ+
changes in the electron density of surface Ag, forming electrophilic Ag sites neighboring
[
2]
adsorbed Cl. These electrophilic Ag sites would then be less likely to adsorb electron-
1
2
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
deficient gas phase O
2
, resulting in increased oxygen reaction orders as a function of Z*, as
[
2]
has been reported on Ag(110). The observed decrease in ethylene reaction order as a function
of Z* (Figure 6) could be explained by increased coverages of electrophilic oxygen adatoms
or by an increased density of electrophilic Ag sites, both of which bond with an electron-
[
8]
δ+
[
52]
donating adsorbate such as ethylene stronger than their neutral or nucleophilic counterparts.
δ+
Indeed, Ag sites were proposed to explain increased ethylene heats of adsorption over Ag(110)
as Cl coverage increased from 0 to 0.3.[
2,26]
The observed increase in EO selectivity as Z* increases (Figure 2) would further suggest
that electrophilic oxygen adatoms are more selective than nucleophilic oxygen adatoms for EO
formation, in agreement with the findings of Bukhtiyarov et al.[ The changes in selectivity
32]
δ+
are not explained simply on the basis of increased density of electrophilic Ag sites, which
can explain changes in the observed reaction orders for formation of EO and CO but not
2
variation in the branching to the two products that apparently occurs after the kinetically
relevant step(s). A third possible explanation for the observed changes in selectivity as a
function of Z* is that oxygen may coexist on Ag surfaces as nucleophilic oxygen atoms and
oxygen radicals, as the predicted energies for these two states were only separated by ~4 kJ
-1
[64,65]
mol on the basis of DFT calculations performed for Ag
3
cluster models.
Carter and
Goddard proposed oxyradical species as the selective species for ethylene epoxidation as it
[
64,65]
was predicted to be further from the surface compared to the nucleophilic oxygen,
implying the possibility of a one-electron process that is not accurately explained via an acid-
[53]
base formalism. Nucleophilic oxygen and chlorine would compete for 4-coordinate sites in
Ag(110), while the oxyradical is only present at high coverages in 3-coordinate sites on
Ag(110), thus the selectivity would increase with chlorine coverage as chlorine prevents
[
64]
binding of unselective nucleophilic oxygen at 4-coordinate sites.
The epoxidation
.
mechanism proposed involves radical attack of the π bond in ethylene, forming a Ag-O-C-C
radical intermediate that rapidly collapses to form the epoxide.[64] However, these calculations
were based on Ag
3
clusters rather than periodic or large cluster (e.g., >560 atoms in the case
of Au ) calculations that more accurately predict binding energies and intermediates for large
~0.1 μm diameter) Ag particles by avoiding finite size and quantum effects possible for
[
66]
(
smaller clusters.
Aided by and in addition to the extensive body of literature on silver-catalyzed ethylene
epoxidation, we report kinetics of EO formation on anion- and chlorine- promoted catalysts.
Kinetics are reported as a function of Z*, a parameter accounting for Cl-adatom deposition by
C
2
5
H Cl and Cl-adatom removal by C
2
H
6
and C
2
H . We further posit a model that plausibly
4
explains observed trends in O and C
2
2
H
4
reaction orders and EO selectivity with Z* as resulting
from surface-chlorine-induced variations in the speciation and number of surface oxygen
adatoms. Ongoing investigations attempt to assess changes in Cl and O coverages with Z* in
order to further specify the role of C
2
H
6
, C
2
H
4
, and C
2
5
H Cl in moderating surface coverages.
4
. Conclusion
The effect of chlorine modification on the kinetics of ethylene epoxidation for a highly
promoted Ag/α-Al catalyst is reported with the overall chloriding effectiveness factor (Z*)
used as a parameter to represent the extent of chlorine modification of the catalyst.
2
O
3
1
3
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Ethylchloride reversibly inhibits both EO (by up to ~40%) and CO
2
(up to ~63%) synthesis
rates while promoting EO selectivity from 81.7% to 86.6% as Z* increases from 2.1 to 8.9,
reflecting a moderate but nevertheless significant change in selectivity given the large volumes
of EO produced on an annual basis (~25 million tons in 2013).[67] Chlorine modification also
results in changes in O
and ~0.5 to subzero, respectively, presumably due to changes in the surface electronic
properties. The near first order dependence on O pressure and measured EO selectivity
85.7% indicate that O activation is involved in the kinetically-relevant step and that atomic
2
and C
2
H
4
reaction orders for both EO and CO
2
formation of ~0.7 to 1
2
>
2
oxygen is involved in EO formation after the rate limiting step. Reaction orders for carbon
dioxide, which reversibly inhibits both epoxidation (–0.49) and total combustion (–0.6) rates,
were found to be invariant with Z*, demonstrating that CO adsorption properties are
2
unchanged with Z*. Co-feeds of EO (0 – 0.5 mol%) and acetaldehyde (0 – 1.7 ppm) at
concentrations pertinent to those measured from the reactor effluent without co-feed were
found to have no effect on the rates and selectivity of EO and CO formation in all Z*
2
conditions investigated. We propose that the effects of chlorine modulation on rates and
selectivity during ethylene epoxidation on the surface of promoted silver catalysts can be
described in terms of a change in the identity and concentration of surface oxygen adatoms.
1
4
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Acknowledgements
We acknowledge financial support from The Dow Chemical Company. We thank Mr. Linh
Bui for assistance with N O experiments.
2
1
5
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
References
[
[
[
[
[
[
[
[
[
1] A. M. Lauritzen, A Process for Producing Ethylene Oxide, 1989, EP0352850A1.
2] C. T. Campbell, B. E. Koel, J. Catal. 1985, 92, 272.
3] C. T. Campbell, J. Catal. 1986, 99, 28.
4] E. L. Force, A. T. Bell, J. Catal. 1976, 44, 175.
5] D. Law, M. Bowker, Catal. Today 1991, 10, 397.
6] R. M. Lambert, R. L. Cropley, A. Husain, M. S. Tikhov, Chem. Commun. 2003, 1184.
7] L. Petrov, A. Eliyas, D. Shopov, Appl. Catal. 1986, 24, 145.
8] T. C. R. Rocha, M. Hävecker, A. Knop-Gericke, R. Schlögl, J. Catal. 2014, 312, 12.
9] J. R. Lockemeyer, R. C. Yeates, D. Reinalda, Method for Improving the Selectivity of a
Catalyst and a Process for the Epoxidation of an Olefin, 2012, US8148555.
10] H. Shibata, A. G. Basrur, S. Gopal, M. H. McAdon, A. C. Y. Liu, L. Zhang, E. R. Frank,
Method of Formulating Alkylene Oxide Catalyst in Relation to Catalyst Reference
Properties, 2014, US 2014/0323295 A1.
[
[
[
[
[
11] D. Torres, F. Illas, R. M. Lambert, J. Catal. 2008, 260, 380.
12] J. R. Monnier, J. L. Stavinoha, G. W. Hartley, J. Catal. 2004, 226, 321.
13] J. M. Berty, Chem. Eng. Commun. 1989, 82, 229.
14] A. Ramirez, J. L. Hueso, H. Suarez, R. Mallada, A. Ibarra, S. Irusta, J. Santamaria,
Angew. Chemie Int. Ed. 2016, 55, 11158–11161.
[
[
15] J. T. Jankowiak, M. A. Barteau, J. Catal. 2005, 236, 379–386.
16] L. Zhang, A. C. Liu, M. Habenschuss, Method of Achieving and Maintaining Specified
Alkylene Oxide Production Parameter with a High Efficiency Catalyst, 2010, US
2
010/0267972 A1.
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
17] E. L. Force, A. T. Bell, J. Catal. 1975, 40, 356.
18] C. T. Campbell, J. Catal. 1985, 94, 436.
19] N. W. Cant, W. K. Hall, J. Catal. 1978, 52, 81.
20] J. W. Medlin, M. A. Barteau, J. M. Vohs, J. Mol. Catal. A Chem. 2000, 163, 129–145.
21] J. W. Medlin, M. A. Barteau, J. Phys. Chem. B 2001, 105, 10054.
22] S. Linic, M. A. Barteau, J. Catal. 2003, 214, 200.
23] S. Linic, M. A. Barteau, J. Am. Chem. Soc. 2003, 125, 4034.
24] S. Linic, J. W. Medlin, M. A. Barteau, Langmuir 2002, 18, 5197–5204.
25] S. Linic, M. A. Barteau, J. Am. Chem. Soc. 2002, 124, 310–317.
26] C. T. Campbell, M. T. Paffett, Surf. Sci. 1984, 139, 396.
27] C. T. Campbell, M. T. Paffett, Appl. Surf. Sci. 1984, 19, 28.
28] R. B. Grant, R. M. Lambert, J. Catal. 1985, 92, 364–375.
29] Y. S. Yong, N. W. Cant, J. Catal. 1990, 122, 22.
30] J. T. Gleaves, A. G. Sault, R. J. Madix, J. R. Ebner, J. Catal. 1990, 121, 202–218.
31] R. A. van Santen, C. P. M. de Groot, J. Catal. 1986, 98, 530–539.
32] V. I. Bukhtiyarov, I. P. Prosvirin, R. I. Kvon, Surf. Sci. 1994, 320, 2.
33] E. A. Carbonio, T. C. R. Rocha, A. Y. Klyushin, I. Pis, E. Magnano, S. Nappini, S.
Piccinin, A. Knop-Gericke, R. Schlogl, T. E. Jones, et al., Chem. Sci. 2018.
34] T. E. Jones, T. C. R. Rocha, A. Knop-Gericke, C. Stampfl, R. Schlögl, S. Piccinin, ACS
Catal. 2015, 5, 5846–5850.
[
1
6
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
[
[
[
[
[
35] J. J. F. Scholten, J. A. Konvalinka, F. W. Beekman, J. Catal. 1973, 28, 209–220.
36] L. Bui, R. Chakrabarti, A. Bhan, ACS Catal. 2016, 6, 6567.
37] P. L. Metcalf, P. Harriott, Ind. Eng. Chem. Process Des. Dev. 1972, 11, 478.
38] D. A. Hickman, J. C. Degenstein, F. H. Ribeiro, Curr. Opin. Chem. Eng. 2016, 13, 1.
39] W. E. Evans, J. M. Kobe, M. Matusz, Process for Improving the Selectivity of an EO
Catalyst, 2012, US 2012/0277447 A1.
[
40] W. E. Evans, J. R. Lockemeyer, D. M. Rekers, A. Jan Te Raa, T. Wermink, Method for
the Start-up of an Epoxidation Process and a Process for the Epoxidation of an Olefin,
2
006, US7102022 B2.
[
[
41] W. Diao, C. D. Digiulio, M. T. Schaal, S. Ma, J. R. Monnier, J. Catal. 2015, 322, 14.
42] S. R. Seyedmonir, D. E. Strohmayer, G. L. Geoffroy, M. A. Vannice, H. W. Young, J. W.
Linowski, J. Catal. 1984, 87, 424–436.
[
[
[
[
43] C.-F. Mao, M. A. Vannice, Appl. Catal. A Gen. 1995, 122, 61.
44] C. Stegelmann, N. C. Schiødt, C. T. Campbell, P. Stoltze, J. Catal. 2004, 221, 630.
45] J. W. Medlin, M. A. Barteau, J. M. Vohs, J. Mol. Catal. A Chem. 2000, 163, 129–145.
46] G. S. Jones, M. Mavrikakis, M. A. Barteau, J. M. Vohs, J. Am. Chem. Soc. 1998, 120,
3
196–3204.
47] G. S. Jones, M. Mavrikakis, M. A. Barteau, J. M. Vohs, J. Am. Chem. Soc. 1998, 120,
196–3204.
[
3
[
[
[
[
48] W. M. H. Sachtler, C. Backx, R. A. van Santen, Catal. Rev. 1981, 23, 127.
49] P. A. Kilty, W. M. H. Sachtler, Catal. Rev. 1974, 10, 1–16.
50] J. T. Roberts, R. J. Madix, J. Am. Chem. Soc. 1988, 110, 8540.
51] D. Lafarga, M. A. Al-Juaied, C. M. Bondy, A. Varma, Ind. Eng. Chem. Res. 2000, 39,
2
148.
[
[
52] H. Metiu, S. Chre, Z. Hu, B. Li, X. Sun, J. Phys. Chem. C 2012, 116, 10439.
53] J. T. Roberts, R. J. Madix, in Surf. React. (Ed.: R.J. Madix), Springer, Berlin, Heidelberg,
Berlin, Heidelberg, 1994, p. 5.
[
[
[
[
[
[
54] M. O. Özbek, R. A. van Santen, Catal. Letters 2013, 143, 131.
55] E. L. Force, A. T. Bell, J. Catal. 1975, 38, 440.
56] C. T. Campbell, M. T. Paffett, Surf. Sci. 1984, 143, 517–535.
57] E. M. Stuve, R. J. Madix, B. A. Sexton, Chem. Phys. Lett. 1982, 89, 48–53.
58] S. A. Tan, R. B. Grant, R. M. Lambert, J. Catal. 1987, 64, 54.
59] J. E. van den Reijen, S. Kanungo, T. A. J. Welling, M. Versluijs-helder, T. A. Nijhuis, K.
P. de Jong, J. Catal. 2017, 356, 65.
[
60] D. A. Bulushev, E. A. Paukshtis, Y. N. Nogin, B. S. Bal’zhinimaev, Appl. Catal. A Gen.
1
995, 123, 301.
[
[
61] E. M. Cordi, J. L. Falconer, Appl. Catal. A Gen. 1997, 151, 179.
62] V. V. Kaichev, V. I. Bukhtiyarov, M. Hävecker, A. Knop-Gercke, R. W. Mayer, R.
Schlögl, Kinet. Catal. 2003, 44, 432.
[
63] V. I. Bukhtiyarov, A. I. Nizovskii, H. Bluhm, M. Hävecker, E. Kleimenov, A. Knop-
Gericke, R. Schlögl, J. Catal. 2006, 238, 260.
[
[
[
64] E. A. Carter, W. A. Goddard, Surf. Sci. 1989, 209, 243–289.
65] E. A. Carter, W. A. Goddard, J. Catal. 1988, 112, 80–92.
66] J. Kleis, J. Greeley, N. A. Romero, V. A. Morozov, H. Falsig, A. H. Larsen, J. Lu, J. J.
Mortensen, M. Dułak, K. S. Thygesen, et al., Catal. Letters 2011, 141, 1067–1071.
1
7
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
[
67] A. Ghannadzadeh, Clean Technol. Environ. Policy 2017, 19, 2145.
1
8
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Tables
Table 1. The range of concentrations investigated to determine pressure dependencies of EO and
CO synthesis rates on O , C , CO , EO, and acetaldehyde partial pressure in this work.
2
2
2
H
4
2
Co-feed
Concentration
2 – 7%
O
2
C
2
H
4
20 – 40%
0 – 5%
CO
EO
CH CHO
2
0 – 0.45%
0 – 1.7 ppm
3
Table 2. Apparent activation energies of EO and CO
at Z* values of 2.1 and 8.9 between 508–523 K.
2
formation on promoted 35 wt% Ag/α-Al
2
O
3
-1
E
app (kJ mol )
Z*=2.1
Z*=8.9
EO formation 96.4 ± 11.6 211.9 ± 24.3
CO
formation 112.7 ± 2.3 211.6 ± 25.5
2
1
9
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Figures
Figure 1. N
The catalyst (0.525 g) was treated in pure H
Initial N O amount introduced to the system was 15.9 μmol.
2
O adsorption-decomposition on the promoted 35 wt% Ag/α-Al
2
O
3
catalyst at 443 K.
-1
2
at 513 K for 20 h with a ramping rate of 0.12 K s .
2
Figure 2. Adjusted rates (Left) and adjusted selectivity (Right) for EO (■), CO
2
(◆), and water
△) synthesis as a function of Z* at ~5.3 bar and 513 K with a GHSV of ~500,000 cm h cmcat-3,
3
-1
(
3
-1
at a total flow rate of ~2.78 cm s , and 0.020 gcat. Feed = O
CO (1%)/ C
2
(3%)/ C
2
H
4
(30%)/ C
2
H
6
(0.4%)/
2
2
5
H Cl (1.5; 2.4; 3.8; 6.3 ppm for Z*= 2.1, 3.4, 5.2, and 8.9)/ He balance (mol%).
2
0
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Figure 3. Adjusted rates for EO (■), CO
2
(◆), and water (△) synthesis as a function of O
2
concentration at Z* values of (a) 2.1, (b) 3.4, (c) 5.2, and (d) 8.9 at ~5.3 bar and 513 K with a
3
-1
-3
3 -1
GHSV of ~500,000 cm h cmcat , at a total flow rate of ~2.78 cm s , and 0.020 gcat. Feed = O
2 – 7%)/ C (30%)/ C (0.4%)/ CO (1%)/ C Cl (1.5; 2.4; 3.8; 6.3 ppm for Z*= 2.1, 3.4,
.2, and 8.9)/ He balance (mol%).
2
(
5
2
H
4
2
H
6
2
2
H
5
2
1
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Figure 4. Adjusted selectivity for EO (■), CO
2
(◆) synthesis as a function of O
2
concentration at
Z* values of (a) 2.1, (b) 3.4, (c) 5.2, and (d) 8.9 at ~5.3 bar and 513 K with a GHSV of ~500,000
3
-1
-3
3 -1
cm h cmcat , at a total flow rate of ~2.78 cm s , and 0.020 gcat. Feed = O
(0.4%)/ CO (1%)/ C Cl (1.5; 2.4; 3.8; 6.3 ppm for Z*= 2.1, 3.4, 5.2, and 8.9)/ He balance
mol%).
2
(2 – 7%)/ C
2
H (30%)/
4
C
(
2
H
6
2
2
H
5
2
2
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Figure 5. Adjusted rates for EO (■), CO
2
(◆), and water (△) synthesis as a function of C
2
H
4
concentration at Z* values of (a) 2.1, (b) 3.4, (c) 5.2, and (d) 8.9 at ~5.3 bar and 513 K with a
3
-1
-3
3
-1
GHSV of ~500,000 cm h cmcat , at a total flow rate ~2.78 cm s , and 0.020 gcat. Feed = O
3%)/ C (20 – 40%)/ C (0.4%)/ CO (1%)/ C Cl (1.5; 2.4; 3.8; 6.3 ppm for Z*= 2.1, 3.4,
.2, and 8.9 at 30 mol% of C )/ He balance (mol%).
2
(
5
2
H
4
2
H
6
2
2
H
5
2
H
4
2
3
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Figure 6. Reaction orders for O
2
, C
2
H
4
, and CO
synthesis; (◆) C
orders for CO at Z* values of 2.1, 3.4, 5.2, and 8.9. Reaction orders
2
in EO and CO
2
synthesis from ethylene oxidation;
(
■) O
2
orders for both EO and CO
2
2
H
4
orders for both EO and CO synthesis; (●)
2
CO
2
orders for EO and (○) CO
2
3
-1
-3
were measured at ~5.3 bar and 513 K with a GHSV of ~500,000 cm h cmcat , at a total flow rate
3
-1
of ~2.78 cm s , and 0.020 gcat. Feed = O
2
(3%)/ C
2
H
4
(30%)/ C
2
H
6
(0.4%)/ CO
2
(1%)/ C
2
5
H Cl
(1.5; 2.4; 3.8; 6.3 ppm for Z*= 2.1, 3.4, 5.2, and 8.9)/ He balance (mol%).
2
4
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Figure 7. Adjusted rates for EO (■), CO
2
(◆), and water (△) synthesis as a function of CO c
2
concentration at Z* values of (a) 2.1, (b) 3.4, (c) 5.2, and (d) 8.9 at ~5.3 bar and 513 K with a
3
-1
-3
3 -1
GHSV of ~500,000 cm h cmcat , at a total flow rate of ~2.78 cm s , and 0.020 gcat. Feed = O
3%)/ C (30%)/ C (0.4%)/ CO (1 – 5%)/ C Cl (1.5; 2.4; 3.8; 6.3 ppm for Z*= 2.1, 3.4,
.2, and 8.9)/ He balance (mol%).
2
(
5
2
H
4
2
H
6
2
2
H
5
2
5
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Figure 8. Adjusted selectivity for EO (■) and CO
2
(◆) synthesis as a function of CO
2
concentration at Z* values of(a) 2.1, (b) 3.4, (c) 5.2, and (d) 8.9 at ~5.3 bar and 513 K with a
3
-1
-3
3 -1
GHSV of ~500,000 cm h cmcat , at a total flow rate of ~2.78 cm s , and 0.020 gcat. Feed = O
3%)/ C (30%)/ C (0.4%)/ CO (1 – 5%)/ C Cl (1.5; 2.4; 3.8; 6.3 ppm for Z*= 2.1, 3.4,
.2, and 8.9)/ He balance (mol%).
2
(
5
2
H
4
2
H
6
2
2
H
5
2
6
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Figure 9. Adjusted rates (Top) and adjusted selectivity (Bottom) for EO (■), CO (◆), and water
2
(
~
△) synthesis as a function of EO concentration at Z* values of (a) and (c) 2.1; (b) and (d) 8.9 at
5.3 bar and 513 K with a GHSV of ~500,000 cm h cmcat , at a total flow rate of ~2.78 cm s ,
3
-1
-3
3 -1
and 0.020 gcat. Feed = O
2
(3%)/ C
2
H
4
(30%)/ C
2
H
6
(0.4%)/ CO
2
(1%)/ EO (0 – 0.5%)/ C
2
5
H Cl (1.5;
6
.3 ppm for Z*= 2.1 and 8.9)/ He balance (mol%).
2
7
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Figure 10. Adjusted selectivity for EO (■) and CO
2
(◆) synthesis as a function of space time at
Z* values of (a) 2.1, (b) 3.4, (c) 5.2, and (d) 8.9 at ~5.3 bar and 513 K. Feed = O
30%)/ C (0.4%)/ CO (1%)/ C Cl (1.5; 2.4; 3.8; 6.3 ppm for Z*= 2.1, 3.4, 5.2, and 8.9)/ He
balance (mol%). A total flow rate of ~2.78 cm s , and 0.020 gcat at 2 × 10 cmcat h cm ; a total
2
(3%)/ C
2
H
4
(
2
H
6
2
2
H
5
3
-1
-3
-3
3
-3
3
-1
3
-3
3 -
flow rate of ~2.78 cm s and 0.040 gcat at 4 × 10 cmcat h cm ; a total flow rate of ~2.22 cm s
1
-2
3
-3
,
and 0.080 gcat at 1 × 10 cmcat h cm .
2
8
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801356
Figure 11. Adjusted rates (Top) and adjusted selectivity (Bottom) for EO (■), CO (◆), and water
2
(
8
△) synthesis as a function of CH
3
CHO concentration at Z* values of (a) and (c) 2.1; (b) and (d)
3
-1
-3
.9 at ~5.3 bar and 513 K with a GHSV of ~500,000 cm h cmcat , at a total flow rate of ~2.78
3
-1
cm s , and 0.020 gcat. Feed = O
ppm)/ C
2
(3%)/ C
2
H
4
(30%)/ C
2
H
6
(0.4%)/ CO
2
(1%)/ CH CHO (0 – 1.7
3
2
H
5
Cl (1.5; 6.3 ppm for Z*= 2.1 and 8.9)/ He balance (mol%).
2
9
This article is protected by copyright. All rights reserved.