Kinetics and site requirements of ether disproportionation on γ-Al2O3

Article abstract of DOI:10.1016/j.apcata.2015.06.008

Abstract Measured rates of methyl propyl ether (MPE), an asymmetric ether, conversion on γ-alumina at 623 K verify that hydration rates are negligible compared to disproportionation rates below 2.0 kPa of water. Steady state kinetic measurements establish that diethyl ether (DEE) disproportionation rates possess reaction orders between 0 and 1. A mechanism for DEE disproportionation in which ethanol monomers and reactive ethoxy species are the primary surface species is consistent with measured pressure dependencies. The intrinsic rate constant of DEE disproportionation is nearly identical to that of unimolecular ethanol dehydration, revealing the similarity in the rate-limiting steps of these two reactions. In-situ pyridine titration studies verify that DEE disproportionation and unimolecular ethanol dehydration possess similar site requirements and densities (0.3 and 0.2 sites nm^-2, respectively) while bimolecular ethanol dehydration occurs on a separate pool of catalytic sites.

Full text of DOI:10.1016/j.apcata.2015.06.008

Applied Catalysis A: General 502 (2015) 361–369
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Kinetics and site requirements of ether disproportionation on ␥-Al O
2
3
∗
Joseph F. DeWilde, Aditya Bhan
Department of Chemical Engineering and Materials Science, University of Minnesota—Twin Cities, 421 Washington Avenue SE, Minneapolis, MN 55455, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Measured rates of methyl propyl ether (MPE), an asymmetric ether, conversion on ␥-alumina at 623 K
verify that hydration rates are negligible compared to disproportionation rates below 2.0 kPa of water.
Steady state kinetic measurements establish that diethyl ether (DEE) disproportionation rates possess
reaction orders between 0 and 1. A mechanism for DEE disproportionation in which ethanol monomers
and reactive ethoxy species are the primary surface species is consistent with measured pressure depen-
dencies. The intrinsic rate constant of DEE disproportionation is nearly identical to that of unimolecular
ethanol dehydration, revealing the similarity in the rate-limiting steps of these two reactions. In-situ pyri-
dine titration studies verify that DEE disproportionation and unimolecular ethanol dehydration possess
Received 30 April 2015
Received in revised form 5 June 2015
Accepted 7 June 2015
Available online 19 June 2015
Keywords:
Ether disproportionation
Alkenes
−2
similar site requirements and densities (0.3 and 0.2 sites nm , respectively) while bimolecular ethanol
dehydration occurs on a separate pool of catalytic sites.
Kinetics
␥
-Alumina
Site requirements
© 2015 Elsevier B.V. All rights reserved.
1
. Introduction
ether hydration, the reverse reaction pathway of bimolecular alco-
hol dehydration, and (ii) direct ether disproportionation to form an
Alcohols dehydrate over acid catalysts, such as gamma-alumina
␥-Al O ) [1–7], to form olefins and water through a unimolecu-
lar pathway or ethers and water through a bimolecular pathway
8–21]. In previous investigations, pyridine inhibition measure-
olefin and an alcohol [26–29]. In this work, we identify the promi-
nent ether conversion pathway at 623 K as well as elucidate its
kinetics (including the prominent surface species present during
reaction) and mechanism.
(
2
3
[
ments identified that these parallel alcohol dehydration pathways
occur on non-equivalent pools of catalytic sites. Concurrently,
kinetic measurements have demonstrated that, at 488 K, alcohol
dehydration is inhibited by non-reactive multimers while at tem-
peratures greater than 623 K the surface is primarily covered with
reactive precursors [22–24]. Specifically, olefin production from the
dehydration of ethanol, 1-propanol, isopropanol, and isobutanol at
Negligible quantities of ether were observed from ethanol reac-
tions at 623 K on ␥-Al O3 at conversions greater than 99% in
2
investigations by Phung et al. [26], suggesting that ether conver-
sion pathways exist on ␥-Al O . This observation is consistent with
2
3
ethanol dehydration measurements presented in Fig. S1 in which
diethyl ether (DEE) selectivity over ␥-Al O3 at 623 K decreases
2
with increasing space time before becoming negligible at ethanol
conversions of ∼99%. Reactor effluent DEE partial pressure was
observed to go through a maximum with increasing contact time
4
88 K was observed to be inhibited by surface dimers composed
of alcohol and water while ether synthesis was inhibited by sur-
face trimer species at alcohol pressures larger than 4.2 kPa [24].
The proposed presence of surface multimers at low temperatures
is consistent with kinetic measurements of propylene synthesis
from 2-propanol dehydration at 373–433 K by De Morgues et al.
over a bed of ␥-Al O3 at 571 and 616 K in steady state kinetic mea-
2
surements of ethanol dehydration by Knözinger and Köhne [27].
The authors additionally observed that ethylene and ethanol were
synthesized in a nearly 1:1 stoichiometric ratio at <10% conver-
sion in kinetic measurements of ether conversion as a function of
increasing temperature, and thus conversion, leading them to pos-
tulate that ether disproportionation presents a significant ether
[
25] which demonstrated that the rate of propylene synthesis was
independent of 2-propanol pressure above 1.1 kPa while simulta-
neously being inhibited by co-fed water at water partial pressures
below 1.2 kPa. In addition to alcohol dehydration, ethers can also
convert on ␥-Al O through two possible reaction pathways: (i)
decomposition pathway on ␥-Al O3 [27]. The 1:1 stoichiometric
2
production of ethylene and ethanol from DEE conversion on ␥-
2
3
Al O was also observed in more recent investigations at 573 K by
2
3
Phung and Busca [28]. Similarly, Morávek and Kraus [29] observed
nearly equal steady state production of ethylene and ethanol with
a much smaller production rate of water from the conversion
of DEE on alumina at 523 K, consistent with a prominent ether
∗ Corresponding author. Fax: +1 612 626 7246.
E-mail addresses: dewil005@umn.edu (J.F. DeWilde), abhan@umn.edu
A. Bhan).
(
http://dx.doi.org/10.1016/j.apcata.2015.06.008
926-860X/© 2015 Elsevier B.V. All rights reserved.
0

3
62
J.F. DeWilde, A. Bhan / Applied Catalysis A: General 502 (2015) 361–369
similar site requirements and densities (0.3 and 0.2 sites nm 2
respectively) as ethylene synthesis from ethanol dehydration.
−
,
disproportionation pathway on alumina materials. Furthermore,
ethylene synthesis persisted after the DEE feed was stopped,
leading the authors to infer that ether disproportionation occurs
through the formation of persistent surface alkoxides that desorb
to form ethylene [29]. In more recent investigations, density func-
tional theory (DFT) calculations by Christiansen et al. [30] show
that DEE disproportionates on the ␥-Al O surface with an acti-
2. Experiments
2.1. Catalyst preparation
2
3
vation energy nearly identical to that of ethylene synthesis from
Kinetic measurements were carried out on ␥-Al O3 pur-
2
−1
ethanol dehydration (38 and 37 kcal mol , respectively), consis-
tent with the authors’ proposal that the rate-limiting step for the
two reactions is similar.
chased from the manufacturer (18HPa-150Catalox, BET surface
2
−1
3
−1
area = 141 m /g , pore volume = 0.786 cm g ). Two batches of
catalyst particles, one with particle sizes between 180 and 420 ␮m
40–80 mesh) for reactions of DEE, ethanol, and water mixtures and
one with particle sizes between 180 and 250 ␮m (60–80 mesh) for
measuring MPE kinetics and pyridine inhibition, were prepared by
DFT calculations (PW 91 functional) of the adsorption and dis-
(
sociation energies of water on a variety of surface sites on ␥-Al O3
2
by both Wischert et al. [31] and Digne et al. [32] noted that these
energies were dependent on the coordination of the surface alu-
minum atom and facet onto which the water molecule adsorbed,
demonstrating that the surface of ␥-Al O possesses a distribu-
pressing and subsequently sieving ␥-Al O3 powder. The desired
2
catalyst mass (∼1.0 mg) was obtained by individually counting
catalyst particles and adding them to a quartz tube reactor filled
with acid-washed quartz sand as described in previous reports
22,23]. The catalyst beds were then treated in air and subsequently
exposed to 2.2 kPa of water prior to reaction as described previously
23].
2
3
tion of active sites with varying acidities. Hendriksen et al. [33]
noted that heats of water immersion on alumina samples outgassed
[
◦
at different temperatures (100–600 C) decreased as the surface
hydration of the alumina surface increased, consistent with these
computational results. The presence of a distribution in acid site
strength was also experimentally observed in both (i) infrared spec-
troscopic measurements by Morterra and Magnacca [34] as well
as independently by Parry [35] and (ii) in 15N nuclear magnetic
resonance measurements by Ripmeester [36] of pyridine-exposed
[
2
.2. Steady state kinetic measurements
The reactor system described in previous reports [22,23] was
used to gather steady state kinetic measurements on ether conver-
sion. All experiments were performed at a reaction temperature of
␥
-Al O samples in which multiple (i) vibrational bands around
2 3
−
1
1
450 cm and (ii) peaks with chemical shifts 110 and 134 ppm
6
23 K using a mixture of liquid feeds, a He carrier gas (Grade 4.7,
were attributed to pyridine adsorption onto Lewis acid sites of vary-
ing strengths. Similarly, multiple infrared vibrational bands (2238,
3
−1
Minneapolis Oxygen Company) with a flowrate of 9.9 cm s at
ambient pressure, and a mixture of 25.0% CH4 with a balance of
−
200, and 2165 cm ) on CO-exposed ␥-Al O3 samples measured
2
1
2
3
−1
Ar (Minneapolis Oxygen Company) fed at 0.017 cm s at ambi-
ent pressure to act as an internal standard for gas chromatography
analysis. Ether conversions were kept to differential levels (<10%)
by using reactor beds containing 1.0 mg of catalyst.
by Zecchina et al. [37] verify the non-uniformity of the ␥-Al O
2
3
surface. Wischert et al. [38] further supported these conclusions
1
27
with {H} Al cross-polarization nuclear magnetic resonance mea-
surements in which a range of both the measured chemical shift
MPE (97% Sigma–Aldrich with 3% methanol as stabilizer) was fed
to and vaporized in a flowing He stream to maintain a pressure of
.8 kPa for kinetic measurements of asymmetric ether conversion.
(
(
between 10 and 70 ppm) and quadropolar coupling constants
between 5 and 38 MHz) were noted on ␥-Al O samples that were
2
3
0
treated in synthetic air at 573 and 773 K. The area of a peak at
23 ppm attributed to penta-coordinated aluminum atoms was
Methanol (99.9% Fisher Scientific) was also fed in these experi-
ments to maintain a partial pressure of 0.2 kPa throughout the
catalyst bed. Methanol and 1-PrOH dehydration kinetics were mea-
sured in independent investigations by feeding 2.4–16.8 kPa of
methanol and 1.4–9.4 kPa of 1-PrOH (99.9% Sigma–Aldrich). Sim-
ilarly, the kinetics of DEE conversion were measured by varying
the partial pressure of ethanol-stabilized DEE (98.1% DEE with 1.8%
ethanol Fisher Scientific) between 0.4 and 8.8 kPa. Pyridine inhibi-
tion measurements were carried out at 0.02–0.08 kPa of pyridine
∼
27
observed to decrease with BaO loading in Al nuclear magnetic
resonance measurements by Kwak et al. [16] on BaO/␥-Al O ;
2
3
peaks attributed to octahedral (0 ppm reference in this study)
and tetrahedral (∼59 ppm) aluminum atoms were unaffected by
the BaO loading. The rate of methanol dehydration to synthesize
DME decreased monotonically with BaO loading on these catalytic
systems [18], suggesting that these penta-coordinated sites on ␥-
Al O are active for ether synthesis. While olefin synthesis was not
2
3
(
99+%, Sigma–Aldrich) with a co-feed of non-stabilized DEE (99.9%
investigated in these studies, these experiments suggest that mul-
Sigma–Aldrich) at 1.4 kPa as well as, in an independent experiment,
a co-feed of 3.5 kPa of ethanol and 1.3 kPa of water. For all exper-
iments used to evaluate alcohol and ether conversion kinetics as
well as pyridine inhibition measurements, deionized water was
co-fed to establish feed partial pressures between 0.2 and 2.1 kPa.
The composition of the reactor effluent was determined using an
online gas chromatograph (GC) with previously described analyti-
cal protocols [22]. The 95% confidence intervals reported in tables
and figures were evaluated from subsequent GC measurements at
the same experimental conditions.
tiple types of potential catalytic sites exist on ␥-Al O . The diversity
2
3
and number of acid sites is expected to be a function of temperature
and water partial pressure and, based on our prior reports, results
in at least two distinct sets of catalytic centers, one which catalyze
unimolecular alcohol dehydration and dehydrogenation reactions
and another which catalyzes bimolecular alcohol dehydration reac-
tions [22,23]. We report in this study that ether disproportionation
occurs on sites that catalyze unimolecular alcohol conversion.
In this investigation, kinetic measurements of methyl propyl
ether (MPE), an asymmetric ether, verify that ether dispropor-
tionation, rather than ether hydration, is the predominant ether
conversion pathway on ␥-Al O3 at water partial pressures below
2.3. In-situ chemical titrations using pyridine
2
2
.0 kPa. Measured DEE disproportion rates are independent of co-
fed water partial pressure and are consistent with a proposed
mechanism in which reactive DEE-derived ethoxy groups and
inhibitory ethanol monomers are the prominent surface species.
In-situ pyridine inhibition experiments are used to affirm that
the catalytic sites responsible for DEE disproportionation possess
Steady state rates and transient profiles of the effluent com-
position during DEE conversion were measured using an online
mass spectrometer on 0.005–0.010 g of ␥-Al O at 623 K and with a
1.5 kPa DEE feed upon the introduction of 0.02–0.05 kPa of pyridine
to the feed. These profiles were used to determine the effective cat-
2
3

J.F. DeWilde, A. Bhan / Applied Catalysis A: General 502 (2015) 361–369
363
MPE and methanol to form dimethyl ether (DME) and 1-propanol.
The rate of this reaction is given by Eq. (3).
r
= rDME − rMeOHtoDME − (rDPE − rPrOHtoDPE − rMPE+PrOH)(3)
MPE+MeOH
rDME and rDPE are the net rates of DME and DPE synthesis, respec-
tively; rMeOHtoDME represents the rate of methanol dehydration to
synthesize DME. The term in the parentheses signifies the rate of
DME formation from the reaction of two MPE molecules to form
DPE and DME. The reaction sequence described in Eqs. (1)–(3) is
shown in Scheme 1 and Eqs. (4)–(11).
C H7OCH + H O → C H7OH + CH OH
(4)
(5)
3
3
2
3
3
C H7OCH → C H + CH OH
3
3
3
6
3
C H7OCH + CH OH → C H7OH + CH OCH
(6)
3
3
3
3
3
3
2
2
2
C H7OCH → C H7OC H7 + CH OCH
(7)
3
3
3
3
3
3
CH OH → CH OCH + H O
(8)
3
3
3
2
C H7OH → C H7OC H7 + H O
(9)
3
3
3
2
Scheme 1. Proposed reaction network for MPE conversion on ␥-Al2O3. The effluent
concentrations of compounds highlighted in boxes were measured experimentally,
while the other compounds were fed to the reactor. The reactions indicated by
dashed arrows signify reactions that were evaluated in independent kinetic mea-
surements.
C H7OH → C H + H O
(10)
(11)
3
3
6
2
C H7OH + C H7OCH → C H7OC H7 + CH OH
3
3
3
3
3
3
Eqs. (1)–(3) contain three rates that can be evaluated in inde-
pendent kinetic measurements: the unimolecular and bimolecular
dehydration rates of 1-propanol to synthesize propylene and DPE,
respectively, and the dehydration rate of methanol to DME.
The 1-propanol reaction order of the measured propylene syn-
thesis rate from the unimolecular dehydration of 1-propanol on
alyst mass that is active for DEE disproportionation in the reactor, as
described previously for ethanol dehydration [23]. The density of
active sites for DEE disproportionation was estimated by linearly
extrapolating the inhibition profile and determining the surface
density of pyridine necessary to completely deactivate the DEE
disproportionation rate, as shown in previous reports [22–24].
␥
-Al O at 623 K decreases from 0.6 to 0.4 as the 1-propanol pres-
2 3
sure increases from 1.4 to 9.4 kPa (Fig. 1a). The reaction order of DPE
synthesis from bimolecular 1-propanol dehydration also decreases
with increasing alcohol partial pressure (decreasing from 1.7 to 0.7
between 1.4 and 9.4 kPa of 1-propanol; Fig. 1b). The methanol order
of DME synthesis decreases from 0.9 to 0.7 as methanol pressure
is increased from 2.4 to 15.8 kPa (Fig. 1c). All dehydration rates at
2.4. Estimation of kinetic parameters
Bayesian statistical optimization techniques in the Athena
Visual Studio statistical software package (v14.2, W. E. Stewart and
M. Caracotsios) were employed to estimate reported kinetic param-
eters. Calculated 95% marginal highest posterior density intervals
were used to determine the reported uncertainties. Independent
measurements at the same reactant partial pressures acted as
experimental replicates for parameter estimation.
6
23 K were observed to be independent of water partial pressure
at water partial pressures up to 2.3 kPa (Fig. 2).
The dehydration kinetics of 1-propanol and methanol match
previous investigations of ethanol dehydration on ␥-Al O at 623 K
2
3
in which dehydration rates were found to be consistent with a
mechanism in which only reactive precursors (alcohol monomers
for olefin synthesis and alcohol dimers for ether synthesis) cover
the active sites during reaction [23]. The rate expressions shown in
Eqs. (12) and (13) can be derived assuming that the reactive precur-
sors prior to the rate-limiting step are the only prominent surface
species and are in quasi-equilibrium with the alcohol in the vapor
phase.
3
. Results and discussion
3.1. Kinetic measurements of MPE conversion
The product distribution of asymmetric ethers, such as MPE,
allow for the rates of ether disproportionation to be deconvoluted
from the rates of ether hydration because different products result
from the two reactions. The hydration of MPE yields both methanol
and propanol while MPE disproportionation will yield methanol
and propylene. The rates of ether hydration, rhyd, and ether dispro-
portionation, rdis, in a differential packed-bed reactor can therefore,
be evaluated using Eqs. (1) and (2) (Scheme 1).
kolefinKA1PAlcohol
rolefin
=
=
(12)
(13)
1
+ KA1PAlcohol
2
ketherKA1KA2
P
Alcohol
rether
2
1+KA1KA2
P
Alcohol
rolefin and rether are the olefin and ether synthesis rates. kolefin
and kether signify the intrinsic rate constants for unimolecular
and bimolecular alcohol dehydration, respectively. The equilibrium
constant to form a surface alcohol monomer is signified by KA1
while KA2 represents the equilibrium constant for the adsorption
of another alcohol molecule to the reactive monomer to form a
surface alcohol dimer. PAlcohol represents the alcohol partial pres-
sure in the catalyst bed. Table 1 summarizes the kinetic parameters
of Eqs. (12) and (13) for 1-propanol and methanol dehydration
estimated from the data presented in Figs. 1 and 2. Solid lines in
Figs 1 and 2 show the models presented Eqs. (12) and (13) with the
estimated kinetic parameters from Table 1. Figs. S2–S4 show the
rhyd = rPrOH + rPrOHtoDPE + rPrOH to C3H6 − rMPE+MeOH − rMPE+PrOH (1)
rdis = rC3H6 − rPrOH toC3H6
(2)
rC and rPrOH represent the net rates of propylene and 1-propanol
3H
6
synthesis, respectively; rPrOHtoDPE and rPrOHtoC3H6 are the rates of
bimolecular and unimolecular 1-propanol dehydration to synthe-
size dipropyl ether (DPE) and propylene, respectively r
MPE+PrOH
denotes the rate of DPE and methanol formation from the reaction
of MPE and 1-propanol; and r signifies the reaction rate of
MPE+MeOH

3
64
J.F. DeWilde, A. Bhan / Applied Catalysis A: General 502 (2015) 361–369
Fig. 1. (a) Propylene, (b) DPE, and (c) DME synthesis rates over 1.0 mg ␥-Al2O3 (total volumetric flowrate = 9.9 cm3
s
−1
) at 623 K as a function of fed (a,b) 1-propanol and (c)
methanol partial pressure with a co-feed of (a,b) 0.7 kPa or (c) 1.5 kPa of water. Solid lines show kinetic model fits to Eqs. (a) (12) and (b,c) (13).
parity, lag, and normal probability plots for the kinetic models for
each reaction.
(1.5 Pa) relative to the fed MPE (0.8 kPa) and methanol (0.2 kPa)
partial pressures, the reactants of the remaining kinetically relevant
reactions. Therefore, the rate of DPE synthesis from the reaction of
The rate of MPE hydration over ␥-Al O3 at 623 K, as defined
2
in Eq. (1), is negligible in reference to experimental variance and
uncertainties at water partial pressures below 2.1 kPa (Fig. 3).
The calculated rates of 1-propanol dehydration at the effluent 1-
propanol partial pressures were at least two orders of magnitude
smaller than all other rates in Eqs. (1)–(3); therefore, the rates of
MPE and 1-propanol, r
was neglected in this analysis; if
MPE+PrOH
this rate was taken into account, it would lower the already negli-
gible calculated hydration rate. The negligible ether hydration rates
lead us to conclude that, at water partial pressures below 2.1 kPa,
the measured rate of conversion of light ethers, such as DEE, are
primarily a result of ether disproportion. The conclusion that ether
disproportionation is the primary ether consumption pathway on
1
-propanol conversion throughout the bed were neglected in this
analysis. Methanol was co-fed to avoid concentration gradients of
methanol across the packed bed that could affect the observed rate
of the reaction of methanol and MPE to synthesize DME and 1-
propanol. The average effluent pressure of 1-propanol was small
␥-Al O3 is consistent with the 1:1 stoichiometric production of
2
ethanol and ethylene at low conversions from DEE conversion over
␥-Al O at 523–616 K observed previously by Knözinger and Köhne
2
3
Table 1
Kinetic parameters corresponding to models for unimolecular and bimolecular dehydration shown in Eqs. (12) and (13) for the synthesis of propylene, DPE, and DME from
alcohol dehydration on ␥-Al2O3 at 623 K estimated from the data presented in Figs. 1 and 2.
Propylene (Eq. (12))
Dipropyl Ether (Eq. (13))
Dimethyl Ether (Eq. (12))
kolefin
-1
kether
−2
kether
−2
KA1(kPa
)
KA1KA2(kPa
)
KA1KA2(kPa
)
/10 4molC2
−
s
−1 −1
g
)
(/10 molDPEs
−5
−1 −1
g
)
(/10 molDMEs
−4
−1 −1
)
g
(
H
6
6
.9 ± 0.9
0.32 ± 0.09
5.8 ± 1.5
0.020 ± 0.009
11 ± 2
0.022 ± 0.009

J.F. DeWilde, A. Bhan / Applied Catalysis A: General 502 (2015) 361–369
365
Fig. 2. (a) Propylene, (b) DPE, and (c) DME synthesis rates from the conversion of (a,b) 3.0 kPa of 1-propanol and (c) 5.2 kPa of methanol over 1.0 mg ␥-Al2O3 (total volumetric
flowrate = 9.9 cm3
s
−1
) at 623 K as a function of water partial pressure. Solid lines show model fits to Eqs. (a) (12) and (b,c) (13).
3.2. Steady state kinetic measurements of DEE disproportionation
The rate of DEE disproportionation at 623 K possesses a DEE par-
tial pressure order between 0 and 1, decreasing with increasing DEE
partial pressure (going from 0.6 to 0.2 as DEE partial pressure is
increased from 0.4 to 8.5 kPa; Fig. 4a). Additionally, DEE dispropor-
tionation rates decrease by a factor of 20% when the ethanol partial
pressure is increased from 0.3 to 0.5 kPa (DEE pressure = 4.3 kPa),
demonstrating that ethanol-derived surface species inhibit DEE
conversion at 623 K (Fig. 4a). This observation is consistent with
measured ethanol dehydration kinetics in which ethanol-derived
surface species were observed to be a kinetically relevant species
above 623 K [23]. As is the case with MPE disproportionation (Fig. 3)
and ethanol dehydration [23] at 623 K, DEE disproportionation
rates are independent of water partial pressure below 1.0 kPa of
water (Fig. 4b), verifying that water-derived surface species are not
Fig. 3. The conversion rate of 0.8 kPa of MPE through (᭹) disproportionation and (ꢀ)
hydration on 1.0 mg of ␥-Al2O3 at 623 K as a function of co-fed water partial pres-
sure. A co-feed of 0.2 kPa of methanol was added to avoid methanol concentration
gradients across the reactor.
stable on the catalytically active sites of ␥-Al O3 at 623 K.
2
A mechanism for the disproportionation of DEE that is consis-
tent with the steady state kinetic measurements shown in Fig. 4 is
presented in Scheme 2. In this mechanism, DEE first adsorbs onto
as well as Morávek and Kraus [27,29]. The kinetic dominance of
ether disproportionation at low water partial pressures (<2.1 kPa)
allows for the evaluation of the kinetics of DEE disproportionation
in absence of complications arising from multiple DEE conversion
pathways.
an empty catalytic site on the ␥-Al O3 surface. The DEE then dis-
2
sociates to form two C2 surface species, likely two surface ethoxy
species, one of them comprising a lattice oxygen atom. The surface

3
66
J.F. DeWilde, A. Bhan / Applied Catalysis A: General 502 (2015) 361–369
Fig. 4. DEE disproportionation rates over 1.0 mg of ␥-Al2O3 at 623 K as a function of (a) DEE partial pressure with a 0.5 kPa water co-feed and (b) co-fed water partial pressure
with a feed DEE partial pressure of 2.4 kPa. In the data presented in (a) ethanol was both independently fed and in the DEE feed solution as a stabilizing agent, the average
total fed ethanol partial pressure was (ꢀ) 0.1 kPa, (ꢀ) 0.2 kPa, (᭿) 0.3 kPa, and (᭹) 0.5 kPa. The solid lines show model fits to Eq. (14).
Table 2
Kinetic parameters for DEE disproportionation on ␥-Al2O3 at 623 K estimated using
the model presented in Eq. (14) and the data presented in Fig. 4.
kdis
KDEEKD
−1
KEtoH
−1
Parameter
Value
/10−4mols−1
−1
(
g
)
(kPa
)
(kPa
)
4.6 ± 0.5
0.78 ± 0.32
0.22 ± 0.17
kdis is the intrinsic rate constant for DEE disproportionation. KDEE
and KEtOH signify the equilibrium constants of adsorption for DEE
and ethanol, respectively. KD is the equilibrium constant for dissoci-
ation of an adsorbed DEE molecule into two surface ethoxy groups,
one containing an oxygen atom from the reactant molecule and
another containing a lattice oxygen atom. PDEE and PEtOH respec-
tively represent the partial pressures of DEE and ethanol. The values
for the kinetic parameters kdis and KDEE of Eq. (14) were estimated
from the data in Fig. 4 and are compiled in Table 2. The equilib-
rium constant for the adsorption of ethanol to form chemisorbed
species that are reactive for ethylene synthesis from dehydration,
Scheme 2. The proposed mechanism for DEE disproportionation on ␥-Al2O3.
ethoxy containing the lattice oxygen then undergoes decompo-
sition via cleavage of the C␤ H and C O bonds to form a gas
phase ethylene molecule and a surface-bound hydrogen adatom
in a manner analogous to the decomposition of ethoxy species in
the postulated mechanism for unimolecular ethanol dehydration
reported previously [22]. The surface-bound hydrogen adatom and
the remaining surface-bound ethoxy species recombine to synthe-
size ethanol and regenerate the catalyst surface.
−1
KEtOH, was evaluated previously (0.22 ± 0.17 kPa ) [23] and was
held constant in the parameter estimation for DEE disproportion-
ation (Eq. (14)). The solid line in Fig. 4 shows the fit of the kinetic
model presented in Eq. (14). A statistical analysis of the residual
errors of this kinetic model, including parity, lag, and normal prob-
ability plots, is presented in the supplementary information (Fig.
S5).
The rate-limiting step of the mechanism depicted in Scheme 2
is similar to that for ethylene synthesis from ethanol dehydra-
The value for the intrinsic rate constant for DEE dispropor-
−4
−1 −1
g , is approximately a factor
tionation, (4.6 ± 0.5) × 10 mol s
tion on ␥-Al O discussed in our previous reports [22,23]. In these
2
3
of two lower than the previously reported value for unimolecu-
reports, a primary kinetic isotope effect (KIE) for ethylene synthe-
sis was only observed when the C H bonds of the ethanol were
deuterated (rH/rD = 2.4 at 488 K and rH/rD = 1.9 at 623 K), verifying
that C H bond cleavage, likely that of a C␤ H bond, was rate-
limiting for ethylene synthesis. Similarly, Knözinger and Scheglila
lar dehydration of ethanol on ␥-Al O to form ethylene at 623 K,
2
3
−4
−1 −1
g [23]. We postulate that the compa-
(
9.4 ± 3.3) × 10 mol s
rable values in the rate constants for these steps correspond to
similarities in the proposed rate-limiting steps; in both reactions,
the C␤ H bond of a C2 surface species, postulated to be an ethoxy
group, is cleaved in the rate-limiting step to form ethylene [22]. This
result is supported by the previously mentioned DFT calculations
of Christiansen et al. [30] in which the calculated barrier for DEE
[
11] measured a primary KIE for the dehydration of tert-, sec-, and
iso-butanol only when the C␤ H bond of the reactant was deuter-
ated. Considering the rate-limiting step for ethylene formation is
the same for both ethanol dehydration and DEE disproportionation,
the rate equation for DEE disproportionation, rdis, shown in Eq. (14)
can be derived assuming that chemisorbed ethanol and surface
ethoxy species formed from dissociated DEE are the prominent sur-
face species on the active sites and are in quasi-equilibrium with
gas phase ethanol and DEE.
−1
disproportionation (38 kcal mol ) was nearly equivalent to that of
unimolecular ethanol dehydration (37 kcal mol ).
−1
The form of the rate equation presented in Eq. (14) is consis-
tent with an alternative mechanism in which adsorbed DEE, rather
than ethoxy groups formed by DEE dissociation, directly dispro-
portionates in the rate-limiting step to form ethylene, a surface
ethoxy group, and a surface hydrogen adatom. The direct decom-
position of ethers as well as alcohols was proposed by Roy et al.
[19] and Christiansen et al. [30,39] as the reaction pathway with
kdisKDEEKDPDEE
rdis = 1
(14)
+KDEEKDPDEE+KEtOHPEtOH

J.F. DeWilde, A. Bhan / Applied Catalysis A: General 502 (2015) 361–369
367
Fig. 6. Catalyst mass available for the disproportionation of 1.5 kPa of DEE over
3
−1
) as a function of time
0.005 g of ␥-Al2O3 at 623 K (total gas flowrate = 9.9 cm
s
after the introduction of 0.05 kPa of pyridine to the feed. The dashed line shows the
linear extrapolation used to determine the pyridine uptake necessary to completely
deactivate the rate of DEE disproportionation.
Fig. 5. The rate of (᭹) DEE disproportionation from the conversion of 1.4 kPa of DEE
as well as the independently evaluated ( ) ethylene, ( ) DEE, and ( ) acetaldehyde
synthesis rates from the conversion of 3.5 kPa of ethanol with a 1.3 kPa water co-feed
on ␥-Al2O3 at 623 K normalized to their rates in absence of pyridine as a function of
co-fed pyridine partial pressure. The dashed lines serve as a guide for the eye.
Table 3
Estimated pyridine surface density necessary to completely deactivate the sites for
DEE disproportionation over ␥-Al2O3 at 623 K determined from in-situ pyridine
titrations at different catalyst loadings and pyridine partial pressures. The reported
errors are 95% confidence intervals evaluated using independent titrations.
the lowest calculated (density functional theory; PW 91) activation
−
5
−1
)
Pyridine uptake (/10 mol g
Catalyst mass
0.010 g
−
1
energies (38 and 37 kcal mol for DEE and ethanol, respectively).
The direct decomposition of adsorbed ethanol, rather than the
decomposition of ethoxy groups, was also discussed as an alterna-
tive mechanism for unimolecular ethanol dehydration in previous
reports [22–24], demonstrating that the concluded similarities in
the rate-limiting steps of DEE disproportionation and unimolecular
ethanol dehydration are still valid with the alternative mechanism.
Both potential mechanisms are kinetically equivalent; thus, the
functional form of Eq. (14), the kinetic parameters presented in
Table 2, and the site requirement analysis discussed in Section 3.3
would be unaffected by the choice of the modeled mechanism.
Pyridine pressure
0.005 g
0.02 kPa
7.7 ± 0.5
7.2 ± 0.5
9.5 ± 1.5
9.0 ± 1.1
0.05 kPa
of catalytic sites responsible for ethylene synthesis to be 0.2 sites
−2
nm
[23]. The density of active sites was evaluated by analyz-
ing the decay in available catalyst mass for DEE disproportionation
after pyridine was introduced to the reactor feed using measured
reactor effluent compositions, Eq. (14), and previously evaluated
kinetics for ethanol dehydration [23]. To estimate the catalytic site
density from this transient deactivation profile, we assume that ini-
tially (far from equilibrium) each molecule of pyridine that enters
uniquely adsorbs onto one of the catalytic sites and that every
catalytic site for DEE disproportionation is equivalent. We then
determine the total pyridine uptake that would completely shut
down DEE disproportionation by extrapolating the initial linear
decay of the calculated available catalyst mass upon introduc-
tion of pyridine to the feed stream (example in Fig. 6). Using this
technique at different pyridine pressures and catalyst loadings,
the average pyridine uptake necessary to completely shut down
DEE disproportionation on ␥-Al O at 623 K was determined to
3.3. DEE disproportionation site requirements and density on
ꢀ
-Al O
2
3
In previous investigations, the degree of inhibition of mea-
sured steady state reaction rates by pyridine was used to probe
the site requirements of ethanol dehydration and dehydrogena-
tion on ␥-Al O . Specifically, pyridine inhibited DEE synthesis to a
2
3
lesser extent than ethylene or acetaldehyde synthesis from the con-
version of ethanol, demonstrating that unimolecular reactions of
ethanol occur on a separate pool of catalytic sites from bimolecular
ethanol dehydration [22,23].
2
3
−
5
−1
−2
be (8.4 ± 0.5) × 10 mol gcat , corresponding to 0.3 sites nm
The rate of DEE disproportionation at 623 K is inhibited by pyri-
dine to the same extent as ethylene and acetaldehyde synthesis
from ethanol dehydration and dehydrogenation and to a greater
extent than DEE synthesis from bimolecular ethanol dehydra-
tion (Fig. 5). This result demonstrates that DEE disproportionation
occurs on catalytic sites that adsorb pyridine equivalently to those
that are active for unimolecular ethanol dehydration and dehy-
drogenation, revealing the similar site requirements for these
chemistries. Furthermore, the different extents of inhibition of DEE
synthesis and disproportionation by pyridine verifies that the pool
of catalytic sites responsible for bimolecular ethanol dehydration
and, thus, its reverse reaction DEE hydration is not equivalent
to the pool responsible for DEE disproportionation. The similar
site requirements and rate constants for ethylene synthesis from
ethanol dehydration and DEE disproportionation reinforces the
mechanism proposed in Scheme 2 in which ethylene is formed in
a rate-limiting step that involves the cleavage of a C␤ H bond of
a surface ethoxy species as was the case for unimolecular ethanol
dehydration [22].
(Table 3). This estimation of site density represents an upper bound
because pyridine could adsorb onto non-active sites during the
titration. The comparable estimated catalytic site densities for DEE
disproportionation and ethylene synthesis from ethanol dehydra-
−
2
tion (0.3 and 0.2 sites nm , respectively) reinforce the conclusion
that the site requirements of these two reactions are similar [23].
The nonequivalent nature of the catalytic sites that catalyze DEE
disproportionation and those that are active for DEE synthesis from
ethanol dehydration on ␥-Al O is illustrated by the ∼6-fold differ-
2
3
ence in estimated catalytic site densities at 623 K (0.3 and 1.8 sites
−
2
nm for DEE disproportionation and synthesis, respectively) [23].
Christiansen et al. [39] calculated that energy barriers for ethy-
lene synthesis from ethanol dehydration differ among the facets of
␥-Al O (37, 30, and 28 kcal mol for the (1 0 0), (1 1 0), and (1 1 1)
facets, respectively) while the activation energies for DEE synthe-
sis did not differ substantially (35, 34, and 32 kcal mol for the
(1 0 0), (1 1 0), and (1 1 1) facets, respectively). The discrepancy in
the calculated activation energies for unimolecular and bimolecu-
lar alcohol dehydration on different facets of ␥-Al O is consistent
−
1
2
3
−
1
2
3
Previously, transient in-situ titrations of ethanol dehydration
over ␥-Al O at 623 K with pyridine estimated the surface density
with our conclusion that at least two non-equivalent pools of cat-
alytic sites exist on the ␥-Al O3 surface. These calculations allow
2
3
2

3
68
J.F. DeWilde, A. Bhan / Applied Catalysis A: General 502 (2015) 361–369
us to postulate the identity of the active surfaces for the two pools
of the catalytic sites. We propose that unimolecular reactions of
ethanol and DEE occur preferentially on higher-indexed facets of ␥-
Al O , (1 1 0) and (1 1 1), while bimolecular dehydration and ether
sure. A mechanism in which the rate-limiting step is ethylene
formation from a surface ethoxy group and surface-bound ethanol
monomers and DEE-derived ethoxy groups primarily cover the
catalytic sites describes the measured DEE rates. The estimated
intrinsic rate constants for DEE disproportionation and ethylene
2
3
hydration can occur on each of the three prominent facets ((1 0 0),
(
1 1 0), and (1 1 1)), reflected by our measured ∼6-fold difference
synthesis from ethanol dehydration [23] on ␥-Al O3 are within a
2
in catalytic site densities for DEE synthesis from ethanol dehydra-
tion and DEE disproportionation. Electron diffraction patterns and
transmission electron microscopy images of plate-like ␥-Al O par-
factor of two, demonstrating the similarities in the rate-limiting
steps of these reactions. DEE disproportionation rates are inhib-
ited by pyridine to the same extent as ethylene synthesis rates
2
3
ticles synthesized by Kovarik et al. [40] demonstrate that the (1 1 0)
and (1 1 1) facets dominate the external surfaces of these parti-
cles while the surfaces in the particle pores primarily comprise the
from ethanol dehydration on ␥-Al O3 and to a greater extent than
2
DEE synthesis rates from bimolecular dehydration. The estimated
density of catalytic sites responsible for DEE disproportionation
−
2
(
1 0 0) and (1 1 1) facets. Furthermore, the authors concluded, based
(0.3 sites nm ) is comparable to that of ethylene synthesis from
−
2
on observed surface reconstructions, that the surface energies of
these exposed facets increase in the following manner: (1 0 0),
ethanol dehydration (0.2 sites nm ). These results verify that (i)
unimolecular dehydration of ethanol and DEE disproportionation
possess similar site requirements, supporting the common rate-
limiting steps conjectured for each reaction, and (ii) bimolecular
ethanol dehydration and DEE disproportionation occur on non-
equivalent catalytic sites.
(
1 1 1), and (1 1 0) [40]. We propose that pyridine preferentially
adsorbs onto the facets with higher surface energies, (1 1 0) and
1 1 1). Therefore, the observation that ethylene and acetaldehyde
(
synthesis from ethanol dehydration and DEE disproportionation
are inhibited by pyridine to a greater extent than DEE synthe-
sis (Fig. 5) is consistent with our postulate that only these higher
Acknowledgements
energy surfaces catalyze unimolecular reactions on ␥-Al O3 while
2
bimolecular reactions can also be activated by the (1 0 0) surface.
We would like to thank The Dow Chemical Company and the
University of Minnesota Doctoral Dissertation Fellowship for their
financial support. We would also like to thank Mr. Marcello Herrera
and Mr. Minje Kang for helpful technical discussions.
2
The postulate that multiple pools of catalytic sites exist on ␥-Al O3
is consistent with the microkinetic model of ethanol dehydration
developed by Christiansen et al. [39] in which both DEE and ethy-
lene synthesis could not be described with complete accuracy using
only the (1 1 1) facet. Based on our postulate, we predict that transi-
tional alumina phases which predominantly feature the (1 1 1) and
Appendix A. Supplementary data
(
1 1 0) facets, such as ␩- or ꢁ-Al O [41], will produce ethylene with
2 3
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.apcata.2015.06.
higher selectivity than ␥-Al O . This prediction will be the subject
of future investigations.
2
3
0
08
One complication that may alter our prediction is the stability
of surface hydroxyl groups on each surface facet for each alumina.
In a previous report [23], we determined that the density of cat-
alytic sites that catalyze DEE synthesis from ethanol dehydration
increased from 0.1 to 1.8 sites nm as the reaction temperature
was increased from 488 to 623 K. We postulated that this increase
in site density was a result of a lower density of surface hydroxyl
groups at higher reaction temperatures. DFT calculations (PW91) by
Digne et al. [32,42] determined that the density of surface hydroxyl
References
[
1] B.C. Lippens, J.H. De Boer, Acta Crystallogr. 17 (1964) 1312–1321.
−2
[2] R.C.T. Slade, J.C. Southern, I.M. Thompson, J. Mater. Chem. 1 (1991) 563–568.
[
[
3] Y. Cesteros, P. Salagre, F. Medina, J.E. Sueiras, Chem. Mater. 11 (1999) 123–129.
4] Industrial minerals & rocks, in: J.E. Kogel, N.C. Trivedi, J.M. Barker, S.T.
Krukowski (Eds.), Society for Mining, Metallurgy, and Exploration, 7th ed.,
Englewood, CO, 2006.
[
[
5] I. Chorkendorff, J.W. Niemantsverdriet, Concepts of Modern Catalysis and
Kinetics, Wiley-VCH, 2007, pp. 193–195.
6] J.H. Kwak, C.H.F. Peden, J. Szanyi, J. Phys. Chem. C 115 (2011) 12575–12579.
groups on the (1 0 0) and (1 1 0) facets of ␥-Al O at 488 K was
8
2
3
−2
.8 and 11.8 OH nm , respectively; At 623 K, the calculated den-
[7] P. Kostestkyy, J. Yu, R.J. Gorte, G. Mpourmpakis, Catal. Sci. Technol. 4 (2014)
3861–3869.
−2
sity decreased to 0 and 8.9 OH nm , respectively, supporting our
explanation for the increased site density at higher temperatures.
The derived kinetics and site requirements for DEE dispro-
portionation in this report describe one of the necessary DEE
conversion pathways to model the reaction network of ethanol
[
[
8] H. Knözinger, Angew. Chem. Int. Ed. 7 (1968) 791–805.
9] H. Knözinger, H. Bühl, E. Ress, J. Catal. 12 (1968) 121–128.
[
10] H. Knözinger, A. Scheglila, A.M. Watson, J. Phys. Chem. 72 (1968) 2770–2774.
[11] H. Knözinger, A. Scheglila, J. Catal. 17 (1970) 252–263.
[
[
12] H. Knözinger, H. Bühl, K. Kochloefl, J. Catal. 24 (1972) 57–68.
13] E.C. DeCanio, V.P. Nero, J.W. Bruno, J. Catal. 135 (1992) 444–457.
dehydration on ␥-Al O to selectively form ethylene. These DEE
2
3
[14] B.C. Shi, B.H. Davis, J. Catal. 157 (1995) 359–367.
conversion pathways explain the negligible DEE yields from
[15] P.A. Clayborne, T.C. Nelson, T.C. DeVore, Appl. Catal. A 257 (2004) 225–233.
[
[
[
16] J.H. Kwak, J.Z. Hu, D.H. Kim, J. Szanyi, C.H.F. Peden, J.Catal. 251 (2007) 189–194.
17] H.A. Dabbagh, M. Zamani, B.H. Davis, J. Mol. Catal. A: Chem. 333 (2010) 54–68.
18] J. Kwak, D. Mei, C.F. Peden, R. Rousseau, J. Szanyi, Catal. Lett. 141 (2011)
649–655.
ethanol dehydration on ␥-Al O3 at very high conversions (>99%)
2
at 623 K observed by Phung et al. [26] as well as our studies shown
in Fig. S1. The reported kinetic model for DEE disproportionation
and those derived previously for ethanol dehydration [22–24] com-
prise a mechanistic picture capable of rigorously describing the
[
19] S. Roy, G. Mpourmpakis, D. Hong, D.G. Vlachos, A. Bhan, R.J. Gorte, ACS Catal. 2
(
2012) 1846–1853.
[
20] R.S. Schiffino, R.P. Merrill, J. Phys. Chem. 97 (1993) 6425–6435.
conversion pathways of ethanol on ␥-Al O .
[21] V.A. Matyshak, L.A. Berezina, O.N. Sil’chenkova, V.F. Tret’yakov, G.I. Lin, A.Y.
Rozovskii, Kinet. Catal. 50 (2009) 111–121.
2
3
[
22] J.F. DeWilde, H. Chiang, D.A. Hickman, C.R. Ho, A. Bhan, ACS Catal. 3 (2013)
98–807.
[23] J.F. DeWilde, C.J. Czopinski, A. Bhan, ACS Catal. 4 (2014) 4425–4433.
7
4
. Conclusions
[
[
[
24] M. Kang, J.F. DeWilde, A. Bhan, ACS Catal. 5 (2015) 602–612.
25] L. de Mourgues, F. Peyron, Y. Trambouze, M. Prettre, J. Catal. 7 (1967) 117–125.
26] T.K. Phung, A. Lagazzo, M.Á. Rivero Crespo, V. Sánchez Escribano, G. Busca, J.
Catal. 311 (2014) 102–113.
Steady state kinetic measurements of MPE conversion on
-Al O at 623 K confirm that ether hydration rates are negligi-
␥
2
3
ble compared to ether disproportionation rates at water partial
pressures below 2.0 kPa. Measured DEE disproportionation rates
possess a DEE reaction order between 0 and 1 (between 0.4 and
[27] H. Knözinger, R. Köhne, J. Catal. 5 (1966) 264–270.
[28] T.K. Phung, G. Busca, Chem. Eng. J. 272 (2015) 92–101.
[29] V. Morávek, M. Kraus, J. Catal. 87 (1984) 452–460.
[30] M.A. Christiansen, G. Mpourmpakis, D.G. Vlachos, ACS Catal. 3 (2013)
1965–1975.
8
.5 kPa of DEE) and are independent of co-fed water partial pres-

J.F. DeWilde, A. Bhan / Applied Catalysis A: General 502 (2015) 361–369
369
[
[
31] R. Wischert, P. Laurent, C. Copéret, F. Delbecq, P. Sautet, J. Am. Chem. Soc. 134
2012) 14430–14449.
32] M. Digne, P. Sautet, P. Raybaud, P. Euzen, H. Toulhoat, J. Catal. 226 (2004)
4–68.
[38] R. Wischert, P. Florian, C. Copéret, D. Massiot, P. Sautet, J. Phys. Chem. C 118
(2014) 15292–15299.
[39] M.A. Christiansen, G. Mpourmpakis, D.G. Vlachos, J. Catal. 323 (2015) 121.
[40] L. Kovarik, A. Genc, C. Wang, A. Qiu, C.H.F. Peden, J. Szanyi, J.H. Kwak, J. Phys.
Chem. C 117 (2013) 179–186.
[41] G. Busca, in: Friederike C. Jentoft (Ed.), Advances in Catalysis, Academic Press,
2014, pp. 319.
(
5
[
[
[
[
[
33] B.A. Hendriksen, D.R. Pearce, R. Rudham, J. Catal. 24 (1972) 82–87.
34] C. Morterra, G. Magnacca, Catal. Today 27 (1996) 497–532.
35] E.P. Parry, J. Catal. 2 (1963) 371–379.
36] J.A. Ripmeester, J. Am. Chem. Soc. 105 (1983) 2925–2927.
37] A. Zecchina, E.E. Platero, E.O. Areán, J. Catal. 107 (1987) 244–247.
[42] P. Digne, P. Sautet, P. Euzen, H. Toulhoat, J. Catal. 211 (2002) 1–5.