Isotopic transient analysis of the ethanol coupling reaction over magnesia

Article abstract of DOI:10.1016/j.jcat.2012.11.014

Isotopic transient analysis of ethanol coupling to butanol over MgO in a fixed-bed reactor at 673 K revealed a surface coverage of adsorbed ethanol equivalent to about 50% of the exposed MgO atomic pairs. DRIFTS of ethanol reaction at 673 K confirmed that the surface was populated primarily with adsorbed ethoxide and hydroxide, presumably from the dissociative adsorption of ethanol. The coverage of reactive intermediates leading to butanol was an order of magnitude lower than that of adsorbed ethanol, and about half the surface base sites counted by adsorption of CO₂. The intrinsic turnover frequency for the coupling reaction at 673 K determined by isotopic transient analysis was 0.04 s^-1, which is independent of any assumptions about the nature of the active sites. Although the ethanol coupling reaction appears to involve aldol condensation of an aldehyde intermediate, the high coverage of ethanol under steady-state conditions apparently inhibits unproductive CC coupling reactions that deactivate the catalyst at high temperature.

Full text of DOI:10.1016/j.jcat.2012.11.014

Journal of Catalysis 298 (2013) 130–137
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Isotopic transient analysis of the ethanol coupling reaction over magnesia
⇑
Theodore W. Birky, Joseph T. Kozlowski, Robert J. Davis
Department of Chemical Engineering, University of Virginia, 102 Engineers Way, Charlottesville, VA 22904-4741, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Isotopic transient analysis of ethanol coupling to butanol over MgO in a fixed-bed reactor at 673 K
revealed a surface coverage of adsorbed ethanol equivalent to about 50% of the exposed MgAO atomic
pairs. DRIFTS of ethanol reaction at 673 K confirmed that the surface was populated primarily with
adsorbed ethoxide and hydroxide, presumably from the dissociative adsorption of ethanol. The coverage
of reactive intermediates leading to butanol was an order of magnitude lower than that of adsorbed eth-
Received 14 August 2012
Revised 5 November 2012
Accepted 8 November 2012
anol, and about half the surface base sites counted by adsorption of CO
2
. The intrinsic turnover frequency
Keywords:
ꢀ1
for the coupling reaction at 673 K determined by isotopic transient analysis was 0.04 s , which is inde-
pendent of any assumptions about the nature of the active sites. Although the ethanol coupling reaction
appears to involve aldol condensation of an aldehyde intermediate, the high coverage of ethanol under
steady-state conditions apparently inhibits unproductive CAC coupling reactions that deactivate the cat-
alyst at high temperature.
Guerbet reaction
Acetaldehyde
Butanol
Carbon dioxide adsorption
Microcalorimetry
IR spectroscopy
Ethanol
Ó 2012 Elsevier Inc. All rights reserved.
Magnesia
1
. Introduction
Although numerous synthetic routes exist to produce saturated
a CAC bond, and hydrogenation of the resulting unsaturated prod-
uct [2–8]. Although alternative mechanisms have been proposed
[9–11], the Guerbet reaction likely requires a catalyst that is mul-
tifunctional since OAH, CAH and CAC bonds are transformed in
the sequence. Catalysts such as magnesia [6,9,12], hydroxyapatite
[13–15], MgAAl mixed oxides [5,8,16–18], alkali-metal-loaded
zeolites [11], and transition metals in the presence of basic com-
pounds [3,4,9,16,18–23] have demonstrated activity in the Guerbet
coupling of alcohols.
A quantitative basis for evaluating activity for the Guerbet reac-
tion is lacking since there is no good method to independently
count the actual active sites on the catalysts. Calculating a rate
based on the surface area of a solid catalyst is usually a good
way to begin quantifying activity, but it is quite likely that only a
small fraction of the catalyst surface is involved in such a complex
reaction as Guerbet coupling. One of the most common methods to
quantify the basic site density on a solid catalyst is to adsorb a gas-
long-chain alcohols, a particularly attractive one involves the so-
called Guerbet reaction or coupling of two shorter chain alcohols
over basic catalysts. The Guerbet reaction has historically been
used for the production of long-chain branched alcohols due to
the plethora of available primary and secondary alcohols that can
be used as reactants. However, recently, it has received attention
for the possible upgrading of short, readily available linear alco-
hols. For example, the small molecule ethanol is an oxygenated
transportation fuel additive that is produced in large quantities
by fermentation of sugars, but has a significantly lower energy
density than gasoline and is readily soluble with water. Butanol,
on the other hand, has a higher energy density than ethanol and
is less hydrophilic, thus making it an attractive oxygenated fuel
additive that can be produced from ethanol via the Guerbet reac-
tion over solid base catalysts.
phase acid such as CO
2
. However, it is not clear that sites titrated
The Guerbet reaction or alcohol coupling was first investigated
over a century ago after its namesake Marcel Guerbet published
the original paper in the field on the conversion of 1-butanol into
by CO at low temperature would be involved in a reaction that oc-
curs at temperatures several hundreds of kelvins higher. Moreover,
a distribution of affinities for adsorption usually exists on a solid
2
2
-ethylhexan-1-ol in 1899 [1]. The reaction is generally believed
2
catalyst so that not all sites counted by a probe such as CO , which
to occur through a series of elementary steps that include alcohol
dehydrogenation to form an aldehyde or ketone intermediate, al-
dol condensation of the aldehyde or ketone intermediates to form
is known to adsorb in different modes, would have the same
likelihood for participating in the Guerbet reaction.
To overcome these limitations, we have chosen to use an
isotopic transient method to quantify the turnover frequency for
ethanol coupling to butanol over a commercially available MgO
catalyst at 673 K. In this technique, often referred to as
⇑
Corresponding author.
E-mail address: rjd4f@virginia.edu (R.J. Davis).
0
021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.11.014

T.W. Birky et al. / Journal of Catalysis 298 (2013) 130–137
131
steady-state isotopic transient kinetic analysis (SSITKA), a catalytic
reaction that is running at steady-state experiences a step change
in the isotopic content of the reactant. After the step change, the
isotopically labeled reactant will gradually appear in the product,
without disturbing the steady-state rate of reaction. By evaluating
the time dependence of the isotope incorporation into the product,
the coverage of reactive intermediates and the intrinsic turnover
frequency of the catalytic cycle can be determined. The SSITKA
technique was developed several decades ago [24–27] and a com-
plete description of the method can be found in a review on the to-
pic by Shannon and Goodwin [28]. The goal of this study is to
determine fundamental parameters from SSITKA (steady-state
coverages and intrinsic turnover frequency) for ethanol coupling
ammonia synthesis and CO oxidation [29–35]. A schematic illus-
tration of the modified system is provided in Fig. 1. Prior to loading
200 mg of the MgO catalyst (UBE Industries, Grade 500 Å, BET sur-
2
ꢀ1
face area = 34.9 m g ) into the reactor, the sample was pressed,
crushed, and sieved to give particles between 106 and 180 m.
The sample was supported on a quartz wool plug in the stainless
l
ꢀ1
steel reactor and heated in situ at 10 K min to 773 K in flowing
3
ꢀ1
2
N (50 cm min ) and held at 773 K for 1 h before cooling to the
reaction temperature of 673 K. To initiate the reaction, unlabeled
anhydrous ethanol (Sigma Aldrich, 99.5%) was added to the flowing
N stream (ultrahigh purity from Praxair, additionally purified by
2
passage through a Supelco OMI-2 purifier) via a saturator main-
tained at 299 K in a temperature-controlled water bath. A small
3
ꢀ1
and relate these to adsorption microcalorimetry of CO
spectroscopy of adsorbed ethanol.
2
and IR
flow of argon gas (1 cm min , ultrahigh purity from Praxair)
was added to the reactant stream to monitor the gas-phase hold-
up of the system, as illustrated in Fig. 1. A separate saturator was
1
3
13
13
loaded with C-labeled ethanol ( CH
3 2
CH OH, from Cambridge
2
. Experimental methods
Isotopes) and placed in the same water bath as the unlabeled eth-
1
3
anol saturator. The C-labeled ethanol was received with a sub-
stantial amount of water (5.89 wt.%), so 3A molecular sieves
2.1. Reactor system
(
Sigma Aldrich), dehydrated at 523 K in N
2
, were used to remove
The reactor system is a slightly modified version of the one used
in prior work to perform isotopic transient experiments during
the water. Additional dehydrated 3A molecular sieves were added
Fig. 1. Schematic of the reactor system used for isotopic transient analysis of ethanol reactions on MgO.

1
32
T.W. Birky et al. / Journal of Catalysis 298 (2013) 130–137
to both of the saturators to ensure the heat transfer and gas–liquid
contacting profiles were the same. The back-pressure regulators on
both the reactor line and the vent line were adjusted to give a total
system pressure of 1.3 atm, with a mole fraction of ethanol equal to
The instrument is a heat flow calorimeter with two cells that are
inserted into a large aluminum block maintained at 303 K. One cell
functioned as a sample cell and the other one served as a reference.
A catalyst sample was first heated to 773 K and evacuated to a
pressure less than 10 Pa. The sample was then cooled and al-
lowed to thermally equilibrate with the system for 2 h prior to
adsorption of carbon dioxide (GT&S Welco, Scientific Grade). Initial
dosing pressures of adsorbate ranged from 10 Pa to 250 Pa, and
each dose was allowed to equilibrate with the sample for 15 min.
ꢀ2
6
7
%. Three different reactant flow rates were used (25, 50, and
5 cm min ) to explore the influence of reactant and product
3
ꢀ1
re-adsorption on the isotopic transient responses. It should be
noted that there was some small conversion of ethanol to acetalde-
hyde in an empty reactor tube at 673 K, but no butanol was ever
detected in the absence of catalyst.
The reactor was operated differentially (low ethanol conver-
sion) at a stationary state, which allowed for an isotope jump in
the reactants and products to be monitored with time. The rate
of reaction and selectivity to different products were evaluated
from a sample of gas removed from a sample port located after
the catalyst bed and injecting it into a gas chromatograph (Agilent
2
.3. Diffuse reflectance FT-IR spectroscopy of adsorbed ethanol
The vibrational spectrum of ethanol adsorbed on MgO after var-
ious treatments was collected on a BioRad FTS-60A FT-IR spec-
trometer equipped with a Harrick Praying Mantis accessory and
high-temperature reaction chamber for diffuse reflectance infrared
Fourier transform spectroscopy (DRIFTS) studies of catalysts. The
MgO sample was first mixed with KBr powder (5 wt% MgO) and
loaded into the DRIFTS cell. Prior to adsorption of ethanol, the sam-
ple was heated at 10 K min in He (ultrahigh purity, Praxair, fur-
ther purified by passage through a Supelco OMI-2 purifier) flowing
at 30 cm min to a temperature of 773 K and held at that temper-
ature for 1 h prior to cooling to the adsorption temperature. The He
stream (30 cm min ) was then passed through a saturator con-
taining anhydrous ethanol (Sigma Aldrich, 99.5%) maintained at
7
890A, equipped with a PoraPLOT Q-HT 25 m, 320 lm ID, column
connected to a flame ionization detector). The major products from
the reaction were acetaldehyde (from ethanol dehydrogenation),
butanol (from ethanol coupling), and ethene (from ethanol dehy-
dration). Since crotonol is sometimes reported as a product from
Guerbet coupling of ethanol and the PoraPLOT column used for
determination of reaction rate did not separate crotonol from
butanol, we used a GC equipped with an Agilent DB-WAX 30 m,
ꢀ1
3
ꢀ1
3
ꢀ1
5
5
8
30
0 cm min
.0. Since the amount of crotonol formed was low under the
lm ID, column to check for crotonol formation. At 673 K and
3
ꢀ1
reactant flow, the butanol to crotonol ratio was
2
95 K and the He/ethanol mixture passed over a bed of 3A molec-
ular sieves prior to entering the DRIFTS cell. All spectra were aver-
aged from 100 scans collected at a resolution of 4 cm
conditions of the study, we ignored its presence in the mass
spectrometer. The selectivity to various products reported in this
work is a carbon-based selectivity. For example, the selectivity to
butanol, Sbutanol, is defined as
ꢀ1
.
The adsorption of ethanol was studied in two different modes,
1) adsorption followed by stepwise temperature-programmed
(
desorption (STPD) and (2) reaction at high temperature. For the
STPD experiment, the pretreated MgO catalyst was cooled to
3
4
ꢁ Ratebutanol
S
butanol ¼ 4
03 K, exposed to the flowing He/ethanol stream for 15 min, and
ꢁ Ratebutanol þ 2 ꢁ RateAcetaldehye þ 2 ꢁ RateEthene
purged with pure He for 15 min prior to acquisition of spectra.
The temperature was then increased stepwise while flowing He
and waiting 15 min at each new temperature prior to collection
of the IR spectra. The averaged spectrum collected at each temper-
ature was referenced to a background spectrum of MgO collected
at that same temperature. In this manner, the progressive desorp-
tion of ethanol from MgO could be followed by DRIFTS. For the
evaluation of DRIFTS during reaction of ethanol, a spectrum of
the sample was collected after the ethanol/He mixture flowed over
a pretreated MgO sample at 673 K for 1 h. In this case, ethanol was
not purged from the cell, but was continually fed to the heated cat-
alyst during the collection of the spectra. Two background spectra
were used in this case, one of flowing ethanol at 673 K in the sam-
ple cell containing only KBr was used to remove the contribution of
gas-phase species and another which was MgO in KBr at 673 K
with no ethanol in the gas phase.
where Rate is defined as moles of product formed per unit time per
unit surface area.
The ethanol, argon, and butanol concentrations were continu-
ously monitored by an online mass spectrometer (Balzers-Pfeiffer
Prisma 200) during the reaction. After the steady-state reaction
was achieved, the four-port valve (see Fig. 1) rapidly switched
the reactant stream from unlabeled ethanol to labeled ethanol.
Since the Ar tracer was not added to the labeled ethanol steam,
the evolution of the Ar signal in the mass spectrometer (m/
e = 40) with time was used to quantify the overall gas-phase re-
sponse of the system. The complexity of the fragmentation pat-
terns of unlabeled and labeled ethanol, acetaldehyde (a primary
product), and butanol prevented the resolution of all products in
the mass spectrometer. After comparing the fragmentation pat-
terns of ethanol (MW = 46) and acetaldehyde (MW = 44), we chose
to follow the most intense peak of ethanol, which was a fragment
at m/e = 31. Thus, the labeled ethanol was monitored at m/e = 32.
Acetaldehyde was not followed because of the large signal from
3
. Results and discussion
2
the N carrier gas at m/e = 28, and other overlapping peaks with
3
.1. Steady-state reaction
ethanol and other products. Butanol was the main coupling prod-
uct and was followed at m/e = 56 and 60 for unlabeled and com-
pletely labeled butanol, respectively. We also tried to follow m/
e = 58 to derive information on bi-labeled butanol formation (prod-
uct of coupling between labeled and unlabeled ethanol), but we
were unable to derive any useful results.
The ethanol coupling reaction over MgO produces three main
products under the conditions used here, acetaldehyde, butanol,
and ethene. Table 1 summarizes the influence of flow rate on the
conversion of ethanol and the distribution to various products over
the MgO catalyst. The fractional conversion of ethanol from Table 1
is plotted in Fig. 2 as a function of the inverse flow rate, which is
proportional to the reactor space time. The linearity of the plot,
which passed through the origin, confirmed that the reactor was
operated differentially with respect to ethanol. Since the mass of
the catalyst was held constant, the slope of the line in Fig. 2 (con-
version versus 1/F) is directly proportional to the reaction rate.
2
2.2. Adsorption microcalorimetry of CO on MgO
Adsorption microcalorimetry experiments were completed on
the same home built instrument that has been described previ-
ously by Bordawekar et al. and used in previous work [36–39].

T.W. Birky et al. / Journal of Catalysis 298 (2013) 130–137
133
Table 1
Product distribution during steady-state reaction of ethanol over MgO at 673 K.
Total flow
rate
Ethanol
conversion ethanol
(%)
Rate of
Selectivity (%)
Acetaldehyde Butanol Ethene
3
ꢀ1
(
cm min
)
conversion
ꢀ
2 ꢀ1
(
mol m
s )
ꢀ
ꢀ
ꢀ
8
25
50
75
23
13
7
4.7 ꢁ 10
5.3 ꢁ 10
4.6 ꢁ 10
52
66
73
34
20
13
14
14
13
8
8
0
.25
0
.2
0
0
.15
0
.1
Fig. 3. Isotopic transient results following the switch from unlabeled ethanol to
doubly labeled 13C-labeled ethanol at a total flow rate of 75 cm3 min during
ꢀ1
2
reaction of 6% ethanol in N over 0.2 g of MgO at 673 K and 1.3 atm. The transient
for the argon tracer in the unlabeled ethanol stream illustrates the gas-phase hold-
up the reactor.
.05
0
0
0.02
0.04
0.06
between the unlabeled ethanol and argon is attributed to adsorp-
tion of ethanol on the catalyst surface. The integral between the
normalized unlabeled ethanol transient, Fethanol, and the argon
transient, FAr, in Fig. 3 is the characteristic time constant for etha-
-
3
1/F (min cm )
Fig. 2. Linear relationship of the fractional ethanol conversion as a function of the
inverse of the reactant flow rate (proportional to reactor space time) confirmed
differential reactor behavior.
nol adsorption, sethanol, at a particular flow rate. The value of s
ethanol
can be combined with the molar flow rate of ethanol exiting the
reactor to determine the total amount of ethanol residing on the
surface of MgO at steady-state conditions, Nethanol, according to
the following equations:
Acetaldehyde was the major product with all three flow rates, but
its selectivity decreased as the conversion of ethanol increased.
This behavior is consistent with the sequential nature of the
Guerbet reaction in which acetaldehyde is produced directly from
ethanol, whereas butanol is produced from reaction of acetalde-
hyde. Ethene appears to be a minor side product of ethanol conver-
sion over MgO at all of the flow rates examined.
Z
1
sethanol
¼
ðFethanol ꢀ FArÞdt
ð1Þ
0
N
ethanol ¼ Flow Rateethanol
ꢂ
s
ethanol
ð2Þ
Small amounts of the expected intermediates of acetaldehyde
condensation were also observed in the product mixture. For
example, crotonol was seen exiting the reactor at 673 K and
Table 2 summarizes the values of
s
ethanol and Nethanol for the three
volumetric feed rates used in this study. The ethanol coverage var-
ꢀ6
ꢀ2
ꢀ6
ied from 4.6 ꢁ 10 mol m to 5.1 ꢁ 10 , which is a fairly narrow
range given a factor of three difference in flow rate. Based on the
periclase structure of MgO and its known lattice parameter
3
ꢀ1
5
1
0 cm min reactant flow, but its concentration was less than
5% of that for butanol. Trace amounts of crotonaldehyde were also
detected at the conditions of the study. Since crotonol and croton-
aldehyde were insignificant side products of the reaction, the
selectivity and rate associated with these intermediates were not
explicitly quantified at the various flow conditions and were ig-
nored in the isotopic transient analysis.
4
.21 Å, the maximum density of MgAO atomic pairs exposed on
ꢀ6
ꢀ2
the (100) surface is calculated to be 9.37 ꢁ 10 mol m , suggest-
ing that about half of the available MgAO adsorption sites are cov-
ered by ethanol under steady-state Guerbet coupling conditions.
3.3. Analysis of product transients
12
12
13
13
3
.2. Switch of CH
3 2 3 2
CH OH to CH CH OH
The mass spectrometer at the reactor exit was also programmed
During the steady-state reaction, the unlabeled ethanol feed
to monitor the isotopic transients associated with the butanol
product. An example set of transients, for the decay of unlabeled
butanol and the rise of labeled butanol in the product stream, after
stream was switched to a doubly 1 C labeled ethanol feed stream
without disturbing the steady state. Fig. 3 presents the normalized
isotopic transients in the ethanol concentration measured at the
exit of the reactor following the switch. The curves for the labeled
and unlabeled ethanol are inverted from each other and intersect
at a relative concentration of 0.5, which indicates the concentra-
tion of ethanol in the reactor (both labeled and unlabeled) was
constant throughout the switch. The argon tracer that was in-
cluded in the feed stream, with unlabeled ethanol, is also included
in Fig. 3 to illustrate the gas-phase hold-up of the system. Since Ar
is assumed to pass through the packed bed reactor without adsorb-
ing on the catalyst, the difference in normalized transients
3
Table 2
Coverage of ethanol on MgO during steady-state Guerbet reaction at 673 K
determined from isotopic transient analysis.
Total flow rate
sethanol (s)
Coverage of ethanol
3
ꢀ1
ꢀ2
(
cm min
)
Nethanol (mol m
)
ꢀ
ꢀ
6
6
25
50
75
32
13
7.8
5.1 ꢁ 10
4.7 ꢁ 10
4.6 ꢁ 10
ꢀ6

1
34
T.W. Birky et al. / Journal of Catalysis 298 (2013) 130–137
1
.8
.6
70
60
50
40
30
20
10
0
0
0
Butanol
Argon
0
0
.4
.2
0
1
1
2C Butanol
3C Butanol
Ethanol
0.04
0
100
200
time (s)
300
400
0
0.02
0.06
-
3
1/F (min cm )
Fig. 4. Isotopic transient results of the product butanol following the switch from
Fig. 5. Influence of the volumetric flow rate of the feed stream on the measured
values of for butanol production (solid triangles) and ethanol adsorption (solid
circles) on MgO at 673 K and 1.3 atm during Guerbet coupling of ethanol.
1
3
unlabeled ethanol to doubly labeled C-labeled ethanol at a total flow rate of
s
ꢀ1
7
1
5 cm3 min during reaction of 6% ethanol in N
2
over 0.2 g of MgO at 673 K and
.3 atm. The transient for the argon tracer in the unlabeled ethanol stream
illustrates the gas-phase hold-up the reactor.
In isotopic transient experiments, linear extrapolation of the
s
values for product molecules to infinite flow rate (zero space time)
is often used to find the intrinsic time constant for the reaction, free
of artifacts from re-adsorption. A complete description of how to
minimize the influence of product re-adsorption in the analysis of
isotopic transients can be found in the review by Shannon and
Goodwin [28] and the work of McClaine and Davis [34].
a switch from unlabeled to labeled ethanol feed is provided in
Fig. 4. The transient curves for the unlabeled and labeled butanol
product are similar in shape to those for the ethanol signals
(
Fig. 3), but the butanol transient is significantly delayed from that
of ethanol. Butanol is a product of reaction, so the characteristic
time for producing butanol, butanol, and the surface coverage of
In the results reported here, extrapolating sethanol to infinite flow
s
rate (zero space time) gave a slightly negative value, suggesting
that a linear extrapolation might not be the best method in this
case. Moreover, the product butanol is not actually a primary reac-
tion product and its rate of formation depends on the level of inter-
mediate concentration of acetaldehyde in the reactor. Table 3
shows the significant change in the coverage of reactive intermedi-
ates leading to butanol at the different flow rates (i.e., different
levels of conversion), and Fig. 6 illustrates the nearly pseudo-
first-order dependence of butanol formation on the acetaldehyde
concentration during these experiments. The similarity of the
slopes of the two lines in Fig. 5 is consistent with the fact that both
ethanol and butanol are short-chain primary alcohols that would
likely interact with the MgO in a similar manner. Thus, we
reactive intermediates that lead to butanol, Nbutanol, can be derived
from the transient results. In particular, the characteristic time for
butanol production is calculated from the following integral in
Fig. 4:
Z
1
s
butanol
¼
ðFbutanol ꢀ FArÞdt
ð3Þ
0
where Fbutanol and FAr are the normalized transient responses of the
butanol and Ar tracer, respectively. The surface coverage of reactive
intermediates is given by:
N
butanol ¼ Ratebutanol
ꢂ
s
butanol
ð4Þ
where Ratebutanol is the formation rate of butanol in units of
ꢀ
2
ꢀ1
ꢀ2
mol m
Table 3 summarizes the values of
three flow rates used in this study. The value of
flow rate, similar to the case of ethanol as given in Table 2. Fig. 5
compares the flow-rate dependence of ethanol and butanol, which
s
, so that the units of Nbutanol are mol m
butanol and Nbutanol for the
butanol varied with
.
s
9
8
7
6
5
4
3
2
s
s
s
s
reveals that both alcohols (reactant and product) experienced
re-adsorption on the MgO catalyst prior to exiting the reactor. As
the flow rate increased, the influence of alcohol re-adsorption
decreased and therefore lowered the values of
isotope switches. Extrapolation of the line for
s
derived from the
ethanol to infinite
s
flow rate (origin of the x-axis in Fig. 5) passes nearly through the
origin, demonstrating the expected lack of re-adsorption of
reactant molecules at exceedingly high flow rates.
Table 3
1
0
Coverage of reactive intermediates leading to butanol on MgO during steady-state
Guerbet reaction of ethanol at 673 K determined from isotopic transient analysis.
0
.0
50.0
100.0
150.0
200.0
Total flow rate Rate of butanol
s
butanol (s) Coverage of intermediates
3
ꢀ1
ꢀ2
(
cm min
)
formation
to butanol Nbutanol (mol m
)
Acetaldehyde Concentration
ꢀ
2
ꢀ1
-1
(mol m
s
)
( mol L )
8.1 ꢁ 10ꢀ
9
60
36
32
4.9 ꢁ 10
ꢀ7
25
50
75
ꢀ
9
ꢀ7
Fig. 6. Correlation of butanol production rate to the exit concentration of
acetaldehyde measured at three different flow rates during the differential
conversion of ethanol at 673 K over MgO at 1.3 atm.
5.2 ꢁ 10
1.9 ꢁ 10
3.1 ꢁ 10ꢀ9
ꢀ7
1.0 ꢁ 10

T.W. Birky et al. / Journal of Catalysis 298 (2013) 130–137
135
assumed the influence of alcohol re-adsorption is nearly equivalent
for the two molecules and simply subtracted the corresponding
values of sethanol (only alcohol re-adsorption) from sbutanol (reaction
at higher pressures signifies a weaker interaction with the surface
typical of physisorption. The chemisorption uptake of CO on the
UBE MgO, calculated by extrapolating the physisorption regime
2
ꢀ6
ꢀ2
plus re-adsorption) at each flow rate to arrive at an average intrin-
of the isotherm at 303 K to zero pressure, was 1.0 ꢁ 10 mol m
Fig. 7b summarizes the differential heat of adsorption as a function
of CO coverage on the MgO catalyst. The initial heat of adsorption
(ꢀ
interaction with the surface, but weakens to about 50 kJ mol at
.
0
sic time constant,
s
, of 25 s for the ethanol coupling reaction
butanol
0
at 673 K on MgO. This value of
s
is similar to that obtained
2
butanol
ꢀ1
by simple linear extrapolation of the space time dependence of
butanol to zero space time (within about 30%), which is a more con-
DHads) was 135 kJ mol , which is consistent with a fairly strong
ꢀ1
s
ꢀ
6
ꢀ2
ventional way to evaluate the intrinsic time constant. An estimate
the saturation coverage of 1.0 ꢁ 10 mol m
.
of the intrinsic turnover frequency per active site, TOF, is the reci-
The coverage of ethanol and intermediates leading to butanol
during the Guerbet reaction (Tables 2 and 3, respectively) can be
0
ꢀ1
procal of
s
, which in this case is 1/(25 s) = 0.04 s for butanol
butanol
formation from ethanol coupling over MgO. This turnover fre-
quency for butanol formation is quite similar to the value reported
compared to the amount of strongly held CO
> 50 kJ mol ) on MgO. During the Guerbet reaction at 673 K, the
coverage of ethanol was found to be five times greater than the
2
(ꢀ
DH
ads
ꢀ1
ꢀ1
by Tsuchida et al. (0.030 s ) for ethanol coupling over MgO at
58 K and 20% conversion [14]. Although the basis of their turn-
over frequency was the number of CO adsorption sites at 523 K,
6
2
CO adsorption capacity, but about half of the number of exposed
2
MgAO pairs at the catalyst surface. Evidently, the adsorption of
ethanol on the metal oxide surface does not require strongly basic
sites, presumably because ethanol can be easily activated to form
adsorbed ethoxy and hydrogen at 673 K. The surface coverage of
intermediates leading to butanol during the Guerbet reaction
which is a somewhat arbitrary choice, their overall magnitude
matches quite well with the one determined here from isotopic
transient analysis, which does not require an independent deter-
mination of active site density.
2
was substantially lower than the CO adsorption capacity, which
might suggest that strong basic sites facilitate the formation of
butanol, presumably through an aldol condensation type coupling
reaction.
3
.4. CO
2
adsorption microcalorimetry
The coverage of surface intermediates leading to butanol ranged
ꢀ7
ꢀ7
ꢀ2
between 2.8 ꢁ 10 and 6.6 ꢁ 10 mol m , which is about an or-
der of magnitude lower than ethanol coverage during the same
reaction. If the Guerbet coupling of ethanol proceeds through a
sequential reaction path in which aldol condensation of acetalde-
hyde is a key step, then base sites are likely to be important. Thus,
microcalorimetry of CO adsorption was used to characterize the
2
basic adsorption sites on the MgO used in this study.
Results from adsorption of carbon dioxide onto the MgO cata-
lyst are presented in Fig. 7. The adsorption isotherm in Fig. 7a re-
vealed a rapid rise in uptake at low pressure, consistent with
3.5. Diffuse reflectance FT-IR spectroscopy of adsorbed ethanol
To corroborate the results from isotopic transient analysis,
DRIFTS of ethanol adsorbed onto MgO was performed. Fig. 8 pre-
sents the IR spectra of adsorbed ethanol during stepwise tempera-
ture-programmed desorption (STPD) in the DRIFTS cell. Two
regions shown in the figure are from 3300 to 2650 and from
ꢀ1
1
300 to 1000 cm , which correspond to the CH
2
/CH
3
and CACAO
stretching regions, respectively, for ethanol and adsorbed ethoxide.
The STPD of ethanol was performed to calibrate the major adsorp-
tion modes and the relative strength of adsorption. Two IR peaks in
2
chemisorption of CO on the MgO surface, whereas the low uptake
the CACAO stretching region were observed. One peak at 1066–
1
1
0
0
.5
.0
.5
.0
ꢀ1
1
058 cm corresponds closely to the gas-phase CACAO stretch
(
a)
in ethanol and is attributed to molecularly adsorbed ethanol. It
should be noted that there is also a contribution from one mode
of the CACAO stretch from dissociatively adsorbed ethanol at a
similar position.
A
second major peak observed at 1132–
ꢀ1
1
119 cm is attributed solely to deprotonated ethanol, or ethox-
ide species, coordinated with a cationic Mg surface atom. These
peak assignments correspond well to previously reported studies
of adsorbed ethanol [40,41]. By 673 K in the STPD experiment
(Fig. 8c), nearly all of all the molecularly adsorbed ethanol had des-
0
100
200
orbed from the MgO, leaving signatures of the two CACAO modes
ꢀ
1
of ethoxide at 1122 and 1058 cm . The CH
2
and CH
3
modes are
Pressure (Pa)
ꢀ
1
also shown in the region 3300–2650 cm . The two peaks at
ꢀ1
1
1
1
1
60
40
20
00
2953 and 2920 cm are attributed to CH
3
stretches in ethoxide,
stretch in
stretches from adsorbed eth-
oxide are red shifted from corresponding CH and CH stretches of
(
b)
ꢀ1
whereas the peak at 2847 cm is attributed to a CH
2
2 3
ethoxide. The observed CH and CH
2
3
gas-phase ethanol. Most of the ethoxide desorbed from the catalyst
by 713 K.
8
6
4
2
0
0
0
0
0
Fig. 9 shows three IR spectra recorded at 673 K in flowing etha-
nol with various solid substrates and various backgrounds. The pur-
pose of the experiments in Fig. 9 was to interrogate the adsorbed
species on MgO in the presence of flowing ethanol at 673 K in an ef-
fort to mimic the conditions of the Guerbet reaction. The spectrum
in Fig. 9a presents essentially the gas-phase signal from ethanol at
0
.0
0.5
1.0
1.5
-
2
Surface Coverage (μmol m )
6
73 K since MgO was not present in the cell. (Fig. 9b and c are the
spectrum for ethanol flowing over MgO at 673 K and the spectrum
of adsorbed species on MgO with the gas phase subtracted, respec-
Fig. 7. Adsorption microcalorimetry of carbon dioxide on MgO at 303 K. (a)
adsorption isotherm of CO on MgO; (b) differential heat of adsorption of CO on
MgO determined from microcalorimetry.
2
2
tively). The gas-phase ethanol spectrum in Fig. 9a reveals two CH
3

1
36
T.W. Birky et al. / Journal of Catalysis 298 (2013) 130–137
The CACAO region in Fig. 9a shows features exclusively from
the gas-phase CACAO stretches, whereas Fig. 9b contains contri-
ꢀ
1
butions from ethanol in the gas phase (1060 cm ) as well as from
(
(
(
(
d)
c)
b)
a)
ꢀ1
surface ethoxide (1120 cm ). When the gas-phase spectrum of
ethanol was removed, Fig. 9c, the two CACAO stretches associated
with surface bound ethoxide were clearly revealed, indicating that
under Guerbet reaction conditions, adsorbed ethoxide is the major
adsorbed species identified by DRIFTS on the MgO surface.
Since results in Fig. 9 were obtained at ethanol coupling condi-
tions, a feature associated with adsorbed aldol intermediates such
as the C@C stretch present in crotonol or crotonaldehyde (which
ꢀ1
should appear near 1523 cm [42]) might be expected. However,
no IR feature was observed in that region in Fig. 9b. Although there
are stretches observed in that region in Fig. 9c, the background for
that particular scan was gas-phase ethanol and KBr, so residual
surface carbonates on MgO would still be present in the spectrum.
ꢀ1
The lack of a C@C stretch in the 1500–1600 cm region of Fig. 9b
suggests very little ethanol coupling must have occurred during
the DRIFTS experiment, presumably because of the very low con-
version of ethanol in the cell. A very small amount of acetaldehyde
ꢀ1
is detected in Fig. 9a and b (1711 cm ), which is consistent with a
very low conversion of ethanol observed in the DRIFTS cell.
Information on the surface hydroxyls can be elucidated in the
ꢀ1
region from 3000 to 3700 cm . Fig. 9b shows the influence of
ethanol adsorption on MgO since the background used for that
spectrum consisted of the thermally pretreated MgO surface. The
ꢀ
1
OH stretch of ethanol was observed at 3675 cm and the stretch
3
200 3000 2800 1300 1200 1100 1000
ꢀ1
of surface OH groups on MgO was seen at 3760 cm . The band
position of the surface OH groups on MgO agrees well with that
Fouad et al. [42]. The small increase in the surface OH stretch in
Fig. 9b was likely the result of ethanol dissociative chemisorption,
but we cannot rule out the potential role of water produced from
ethanol dehydration. Fig. 9c cannot be used to evaluate the OH
stretching region since the background contribution from MgO
was not subtracted (only gas-phase ethanol and KBr were removed).
-1
Wavenumber (cm )
Fig. 8. DRIFTS of ethanol adsorbed on MgO at room temperature followed by
heating in He to: (a) 473 K, (b) 573 K, (c) 673 K, and (d) 713 K. Spectra are offset for
clarity.
stretches at 2981 and 2969 cmꢀ1 and one CH
The stretches observed in this same region in Fig. 9c, which involves
the species adsorbed on MgO in the flowing gas mixture but with
the contributions from the gas-phase ethanol subtracted, are
952, 2925, and 2845 cm . Those peak positions correspond to
the stretches, two CH and one CH , observed in the STPD of ad-
sorbed ethoxide (Fig. 8c), within the resolution of the spectrometer.
The observed red shifts of the CH and CH stretches from the gas
phase agree with the predicted red shifts by Branda et al. for ethox-
ide on MgO [40]. In that work, red shifts of 25–40 cm are expected
for the CH stretches upon dissociative adsorption and we observed
shifts of 29 and 44 cm . Likewise, Branda et al. predicted a 25–
ꢀ1
2
stretch at 2909 cm
.
3.6. Implications for the Guerbet reaction on MgO
ꢀ1
2
3
2
The results from isotopic transient analysis of the Guerbet
reaction of ethanol, CO2 adsorption microcalorimetry, and DRIFTS
of adsorbed ethanol provide a consistent picture of the Guerbet
reaction of ethanol on MgO. First, the selectivity of the reaction
as a function of reactant flow rate suggests that acetaldehyde is a
reaction intermediate for the formation of butanol. Indeed, the ob-
served rate of butanol formation was correlated strongly with the
acetaldehyde concentration in the gas phase (Fig. 6). Those results
are consistent with the idea that the CAC bond-forming step in the
3
2
ꢀ1
3
ꢀ1
ꢀ
1
1
2
30 cm shift for CH upon adsorption and we recorded a red shift
of 64 cm [40].
ꢀ
1
(c)
-1
3
1675 cm
ethoxide
+
2
-
3
760 cm
-
CH CH O-Mg O
3
(b)
gas phase
ethanol
-1
3
675 cm
(a)
4
000
3500
3000
2500
2000
1500
1000
-1
Wavenumber (cm )
Fig. 9. DRIFTS of ethanol in He at 673 K in the presence of: (a) KBr, with KBr in He at 673 K as a background; (b) MgO in 95 wt.% KBr, with MgO in 95 wt.% KBr in He at 673 K as
a background; (c) MgO in 95 wt.% KBr, with KBr in ethanol and He at 673 K as a background. Spectra are offset for clarity.

T.W. Birky et al. / Journal of Catalysis 298 (2013) 130–137
137
Guerbet coupling reaction is an aldol-condensation step, which is
well known to occur on basic catalysts. However, aldol addition
reactions occur very readily on MgO, as illustrated by many earlier
published papers in the area [43–53]. For example, the TOF for ace-
Science, U.S. Department of Energy, Grant No. DE-FG02-
95ER14549.
References
ꢀ1
tone addition over MgO at 273 K is 0.05 s , based on the results
from Zhang et al. [49], assuming a base site density on MgO of
[
[
[
1] M. Guerbet, C. R. Acad. Sci. Paris (1899) 1002.
2] A.J. O’Lenick, J. Surfactants Deterg. 4 (2001) 311–315.
3] C. Carlini, M.D. Girolamo, M. Marchionna, M. Noviello, A.M.R. Galletti, G.
Sbrana, J. Mol. Catal. A – Chem. 184 (2002) 273–280.
ꢀ6
ꢀ2
1
.0 ꢁ 10 mol m (as seen in Fig. 7). Thus, aldol condensation
proceeds at reasonable rates even at temperatures as low as
73 K. In an attempt to explore the aldol condensation of acetone
[
[
4] M.J.L. Gines, E. Iglesia, J. Catal. 176 (1998) 155–172.
5] J.I. Di Cosimo, C.R. Apesteguía, M.J.L. Ginés, E. Iglesia, J. Catal. 190 (2000) 261–
2
over MgO at similar gas-phase concentrations and temperatures
as those used here for Guerbet coupling, we observed many cou-
pling products and very rapid deactivation of the catalyst. The
presence of ethanol in the gas phase evidently preserves the cata-
lyst activity by inhibiting access of the aldehyde intermediates to
the active sites on MgO and/or lowering the base strength of the
surface oxide by forming surface hydroxide. Our results from isoto-
pic transient analysis, CO adsorption, and DRIFTS confirm that the
2
coverage of adsorbed ethanol, primarily as ethoxide, is nearly 50%
of the exposed MgAO surface sites and well beyond the adsorption
capacity of chemisorbed CO , a typical probe of basic metal oxides.
2
2
75.
[6] A.S. Ndou, N.J. Coville, Appl. Catal. A: Gen. 275 (2004) 103–110.
7] E. Iglesia, D.G. Barton, J.A. Biscardi, M.J.L. Gines, S.L. Soled, Catal. Today 38
[
[
(
1997) 339–360.
8] J.I. Di Cosimo, V.K. D´ıez, M. Xu, E. Iglesia, C.R. Apestegu ´ı a, J. Catal. 178 (1998)
499–510.
[9] A.S. Ndou, N. Plint, N.J. Coville, Appl. Catal. A: Gen. 251 (2003) 337–345.
10] K.W. Yang, X.Z. Jiang, W.C. Zhang, Chin. Chem. Lett. 15 (2004) 1497–1500.
11] C. Yang, Z.Y. Meng, J. Catal. 142 (1993) 37–44.
12] W. Ueda, T. Kuwabara, T. Ohshida, Y. Morikawa, J. Chem. Soc. Chem. Commun.
(1990) 1558–1559.
[
[
[
[
[
13] T. Tsuchida, S. Sakuma, T. Takeguchi, W. Ueda, Ind. Eng. Chem. 45 (2006)
8
634–8642.
14] T. Tsuchida, J. Kubo, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, J. Catal. 259
(2008) 183–189.
15] S. Ogo, A. Onda, K. Yanagisawa, Appl. Catal. A: Gen. 402 (2011) 188–195.
16] C. Carlini, M. Marchionna, M. Noviello, A.M.R. Galletti, G. Sbrana, F. Basile, A.
Vaccari, J. Mol. Catal. A – Chem. 232 (2005) 13–20.
In contrast, the coverage of reactive intermediates leading to buta-
nol during ethanol coupling at 673 K is nearly an order of magni-
tude lower than that of the adsorbed ethoxide (see Tables 2 and
[
[
3
). Thus, we speculate that Guerbet coupling of ethanol on MgO oc-
[17] M. León, E. Díaz, S. Ordóñez, Catal. Today 164 (2011) 436–442.
[
[
[
[
18] C. Carlini, C. Flego, M. Marchionna, M. Noviello, A.M.R. Galletti, G. Sbrana, F.
Basile, A. Vaccari, J. Mol. Catal. A – Chem. 220 (2004) 215–220.
19] C. Carlini, A. Macinai, M. Marchionna, M. Noviello, A.M.R. Galletti, G. Sbrana, J.
Mol. Catal. A – Chem. 206 (2003) 409–418.
20] C. Carlini, M. Di Girolamo, A. Macinai, M. Marchionna, M. Noviello, A.M.R.
Raspolli Galletti, G. Sbrana, J. Mol. Catal. A – Chem. 200 (2003) 137–146.
21] C. Carlini, M.D. Girolamo, A. Macinai, M. Marchionna, M. Noviello, A.M.R.
Galletti, G. Sbrana, J. Mol. Catal. A – Chem. 204 (2003) 721–728.
curs on a primarily ethoxide covered surface through a sequence of
steps that involves dehydrogenation to acetaldehyde, aldol con-
densation to form C4 compounds, followed by hydrogenation or
hydrogen transfer to the butanol final product. The high coverage
of ethoxide provides a ready source of atomic hydrogen on the sur-
face as well as inhibits the subsequent aldol condensation of C4
intermediates to C6 and heavier compounds that would deactivate
the catalyst at 673 K. Since aldol condensation occurs readily on
MgO, even at room temperature, the high temperatures required
for Guerbet coupling are needed to dehydrogenate the ethoxide
to the acetaldehyde intermediate, which is consistent with other
works that show adding a hydrogen transfer catalyst allows the
reaction to occur at lower temperatures [3,4,16,18–22,54].
[22] M.N. Dvornikoff, M.W. Farrar, J. Org. Chem. 22 (1957) 540–542.
[
[
[
23] T.C. Hsiao, S.D. Lin, Catal. Lett. 119 (2007) 72–78.
24] P. Biloen, J. Mol. Catal. 21 (1983) 17–24.
25] C.O. Bennett, in: A.T. Bell, L.L. Hegedus (Eds.), Catalysis under Transient
Conditions, ACS Symposium Series, American Chemical Society, Washington,
DC, 1982, pp. 1–32.
[
[
26] J. Happel, Chem. Eng. Sci. 33 (1987) 1567.
27] M. Depontes, G.H. Yokomizo, A.T. Bell, J. Catal. 104 (1987) 147–155.
[28] S.L. Shannon, J.G. Goodwin, Chem. Rev. 95 (1995) 677–695.
[
[
29] J.T. Calla, R.J. Davis, J. Catal. 241 (2006) 407–416.
30] J.T. Calla, M.T. Bore, A.K. Datye, R.J. Davis, J. Catal. 238 (2006) 458–467.
4
. Conclusions
[31] J.T. Calla, R.J. Davis, J. Phys. Chem. B 109 (2005) 2307–2314.
[
32] S. Siporin, R. Davis, W. Raróg-Pilecka, D. Szmigiel, Z. Kowalczyk, Catal. Lett. 93
2004) 61–65.
33] S.E. Siporin, R.J. Davis, J. Catal. 222 (2004) 315–322.
(
The Guerbet coupling of gas-phase ethanol to butanol was cat-
[
alyzed by MgO at 673 K. The coverage of adsorbed ethoxide, which
was identified by in situ DRIFTS, was determined from isotopic
transient analysis to be 5 ꢁ 10 mol m , which is about half of
the exposed MgAO pairs based on the measured surface area of
the catalyst assuming a surface atom density equivalent to that
on the (100) plane of MgO. The surface coverage of intermediates
[34] B.C. McClaine, R.J. Davis, J. Catal. 211 (2002) 379–386.
[35] B.C. McClaine, R.J. Davis, J. Catal. 210 (2002) 387–396.
[36] S.V. Bordawekar, E.J. Doskocil, R.J. Davis, Langmuir 14 (1998) 1734–1738.
ꢀ6
ꢀ2
[
[
37] S.V. Bordawekar, R.J. Davis, J. Catal. 189 (2000) 79–90.
38] J.T. Kozlowski, M.T. Aronson, R.J. Davis, Appl. Catal. B: Environ. 96 (2010) 508–
515.
[
[
39] E.J. Doskocil, S.V. Bordawekar, R.J. Davis, J. Catal. 169 (1997) 327–337.
40] M.M. Branda, A.H. Rodríguez, P.G. Belelli, N.J. Castellani, Surf. Sci. 603 (2009)
ꢀ7
ꢀ7
ꢀ2
leading to butanol ranged from 2.8 ꢁ 10 to 6.6 ꢁ 10 mol m
,
1
093–1098.
[41] R.N. Spitz, J.E. Barton, M.A. Barteau, R.H. Staley, A.W. Sleight, J. Phys. Chem. 1
1986) 4067–4075.
42] N.E. Fouad, P. Thomasson, H. Knözinger, Appl. Catal. A: Gen. 196 (2000) 125–
33.
[43] M.J. Climent, A. Corma, V. Fornés, R. Guil-Lopez, S. Iborra, Adv. Synth. Catal. 344
2002) 1090–1096.
depending on the conversion of ethanol. Since the primary product
from the reaction over MgO was acetaldehyde, we concluded that
butanol is a secondary product of acetaldehyde conversion, pre-
sumably through a CAC bond-forming aldol-condensation step.
The coverage of reaction intermediates leading to butanol was be-
(
[
1
(
[
[
44] W. Ji, Y. Chen, H.H. Kung, Appl. Catal. A: Gen. 161 (1997) 93–104.
45] J. Tai, R.J. Davis, Catal. Today 123 (2007) 42–49.
2
low the surface base site density evaluated by CO adsorption,
ꢀ6
ꢀ2
which was measured to be 1.0 ꢁ 10 mol m . The intrinsic turn-
over frequency of the catalytic cycle on MgO leading to the forma-
tion of butanol at 673 K was determined from isotopic transient
analysis to be 0.040 s . The rather low turnover frequency of this
reaction at 673 K is likely related to the difficulty associated with
dehydrogenating ethanol to acetaldehyde over MgO.
[46] M. Bailly, C. Chizallet, G. Costentin, J. Krafft, H. Lauron-Pernot, M. Che, J. Catal.
35 (2005) 413–422.
2
[
[
47] J. Oviedo, J.F. Sanz, Surf. Sci. 397 (1998) 23–33.
48] S. Lippert, W. Baumann, K. Thomke, J. Mol. Catal. 69 (1991) 199–214.
ꢀ
1
[49] G. Zhang, H. Hattori, K. Tanabe, Appl. Catal. 36 (1988) 189–197.
[50] G. Zhang, H. Hattori, K. Tanabe, Appl. Catal. 40 (1988) 183–190.
[
51] J.F. Sanz, J. Oviedo, A. Márquez, J.A. Odriozola, M. Montes, Angew. Chem. Int.
Ed. 38 (1999) 506–509.
[
52] J. Li, J. Tai, R.J. Davis, Catal. Today 116 (2006) 226–233.
Acknowledgments
[53] J. Lercher, H. Noller, G. Ritter, J. Chem. Soc. Faraday Trans. 1 (1981) 621–628.
54] K. Koda, T. Matsu-ura, Y. Obora, Y. Ishii, Chem. Lett. 38 (2009) 838–839.
[
This work was supported by the Chemical Sciences, Geosciences
and Biosciences Division, Office of Basic Energy Sciences, Office of