ABE Condensation over Monometallic Catalysts: Catalyst Characterization and Kinetics

Article abstract of DOI:10.1002/cctc.201601507

Herein, we present work on the catalyst development and the kinetics of acetone-butanol-ethanol (ABE) condensation. After examining multiple combinations of metal and basic catalysts reported in the literature, Cu supported on calcined hydrotalcites (HT) was found to be the optimal catalyst for the ABE condensation. This catalyst gave a six-fold increase in reaction rates over previously reported catalysts. Kinetic analysis of the reaction over CuHT and HT revealed that the rate-determining step is the C?H bond activation of alkoxides that are formed from alcohols on the Cu surface. This step is followed by the addition of the resulting aldehydes to an acetone enolate formed by deprotonation of the acetone over basic sites on the HT surface. The presence of alcohols reduces aldol condensation rates, as a result of the coverage of catalytic sites by alkoxides.

Full text of DOI:10.1002/cctc.201601507

DOI: 10.1002/cctc.201601507
Full Papers
ABE Condensation over Monometallic Catalysts: Catalyst
Characterization and Kinetics
[
a, b, c]
[a, c, d]
[e]
[a, c]
Konstantinos A. Goulas,
Gorkem Gunbas,
Paul J. Dietrich, Sanil Sreekumar,
[c]
[b, c]
[c, e]
[a, c]
Adam Grippo, Justin P. Chen,
Amit A. Gokhale,
and F. Dean Toste*
Herein, we present work on the catalyst development and the
kinetics of acetone-butanol-ethanol (ABE) condensation. After
examining multiple combinations of metal and basic catalysts
reported in the literature, Cu supported on calcined hydrotal-
cites (HT) was found to be the optimal catalyst for the ABE
condensation. This catalyst gave a six-fold increase in reaction
rates over previously reported catalysts. Kinetic analysis of the
reaction over CuHT and HT revealed that the rate-determining
step is the CÀH bond activation of alkoxides that are formed
from alcohols on the Cu surface. This step is followed by the
addition of the resulting aldehydes to an acetone enolate
formed by deprotonation of the acetone over basic sites on
the HT surface. The presence of alcohols reduces aldol conden-
sation rates, as a result of the coverage of catalytic sites by
alkoxides.
Introduction
[1,2]
Emissions of CO from fossil fuel combustion have been a driv-
or hydrocarbon mixtures obtained by hydrodeoxygenation
2
ing force behind the increase in global temperatures over the
last 50 years. As a result, worldwide efforts have been made to
at high temperature and pressure. However, both of these
methods are hampered by problems with the vegetable oil
supply. In addition to the competition with food supplies, bio-
fuel yields per acre are low in the cases of rapeseed and soy-
reduce CO emissions, culminating in the Kyoto Protocol. As
2
part of its commitments under the Kyoto Protocol, the EU has
mandated a minimum bio-derived content in its diesel supply.
Currently, biodiesel production primarily relies on vegetable oil
to produce fatty acid methyl esters (FAME) derived by transes-
[3]
bean oil and sustainability issues related to land use have yet
[4]
to be addressed in the case of palm oil.
In our earlier work, we proposed a strategy combining fer-
mentation and chemical upgrading to yield biodiesel precursor
[1,2]
terification of triglycerides extracted from oleiferous seeds
[5,6]
molecules. In this sequence, sugars were fermented to a mix-
ture of acetone, butanol, and ethanol that was upgraded to
heavier ketones in a dehydrogenation-aldol condensation-hy-
drogenation reaction pathway. While the ABE condensation re-
action was catalyzed by a physical mixtures of Pd/C and K PO ,
[
a] Dr. K. A. Goulas, Prof. Dr. G. Gunbas, Dr. S. Sreekumar, Prof. Dr. F. D. Toste
Department of Chemistry
University of California, Berkeley
Berkeley, CA 94720 (USA)
3
4
E-mail: fdtoste@berkeley.edu
the water produced during the reaction tended to inhibit the
activity of potassium phosphate. Additionally, reaction rates
were relatively low (24 h required for complete conversion of
the starting material, corresponding to a turnover frequency of
[b] Dr. K. A. Goulas, J. P. Chen
Department of Chemical and Biomolecular Engineering
University of California, Berkeley
Berkeley, CA 94720 (USA)
À1
[
c] Dr. K. A. Goulas, Prof. Dr. G. Gunbas, Dr. S. Sreekumar, A. Grippo, J. P. Chen,
Dr. A. A. Gokhale, Prof. Dr. F. D. Toste
Energy Biosciences Institute
University of California, Berkeley
Berkeley, CA 94720 (USA)
about 0.03 h ) and the use of toluene as a solvent hampered
the applicability of this system in an industrial setting. Subse-
quently, Cu- and Pd-substituted hydrotalcite materials were
shown to be efficient catalysts for the ABE condensation.
These catalysts showed markedly improved recyclability com-
[d] Prof. Dr. G. Gunbas
[7,8]
Department of Chemistry
Middle East Technical University
Ankara 06800 (Turkey)
pared to the original Pd/C-K PO system;
however, the long
3
4
reaction times, on the order of 20 h (corresponding to a TOF
À1
of 0.06 h ), made this system impractical as an industrial cata-
[e] Dr. P. J. Dietrich, Dr. A. A. Gokhale
lyst. In this work, we describe both the improvement of these
catalysts by using batch reactor screening and instrumental
characterization techniques, such as X-ray absorption spectros-
copy, and kinetic studies on the optimal catalysts to determine
the reaction pathways.
BP Products North America
1
50 W. Warrenville Road
Naperville, IL 60563 (USA)
Supporting information and the ORCID identification number(s) for the
author(s) of this article can be found under http://dx.doi.org/10.1002/
cctc.201601507.
th
This manuscript is part of a Special Issue to celebrate the 50 annual
meeting of the German Catalysis Society. A link to the Table of Contents
will appear here once the Special Issue is assembled.
ChemCatChem 2017, 9, 1 – 10
1
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
there is excess hydrogen in the system, since the for-
mation of the C7 and C11 ketones does not require
extra hydrogen. The dehydrogenation of the alcohols
in the feed produces one mole of hydrogen, which is
consumed again in the hydrogenation of the a,b-un-
saturated ketone product of the aldol condensation.
Possible pathways for the formation of hydrogen in-
clude the Tishchenko reaction and decarbonylation
reactions. In the former the dehydrogenation of the
Scheme 1. The ABE condensation reaction.
Results and Discussion
2
Table 2. Yields of ABE reaction over Na- and Ca-basified SiO catalysts.
Reaction conditions: 2.3 mmol acetone, 3.7 mmol butanol, 1 mmol etha-
nol, 350 mg catalyst in 1 mL BuOH.
Rode et al. showed that alkali-substituted zeolites can be effi-
cient catalysts for aldol condensation reactions of butyralde-
[
9]
hyde. We therefore examined Pd-modified CsY and NaY
zeolites, as potential catalysts for ABE upgrading. The target
products (A–F) of the dehydrogenation/aldol condensation se-
quence are shown in Scheme 1.
Entry
Pd/10Na/SiO
A
B
C
D
E
F
–
Alcohols Overall
1
2
1.8 26.8 3.2 0.3 22
6.0
60.1
91.2
91.2
76.9
54.2
31.2
86.1
43.0
39.6
0.75Pd/15Na/15Ca/SiO2
4.8 34.8 5.1 0.6 39.8 1.4 4.7
2.5 32.3 6.6 0.5 42.1 2.2 5.0
4.1 41.0 4.0 0.9 22.0 0.2 4.7
2.9 34.2 1.7 0.5 10.7 – 4.2
0.5Pd/15Na/15Ca/SiO
2
0
1
1
0
0
0
.85Pd/17Na/17Ca/SiO
Pd/10Na/10Ca/SiO
Cu/10Na/10Ca/SiO
2
2
Table 1. Percent yield of ABE condensation products for the series of Pd-
modified, alkali-substituted zeolites. Reaction condition: 2.3 mmol ace-
tone, 3.7 mmol butanol, 1 mmol ethanol, 350 mg catalyst in 1 mL BuOH.
2
–
20.8 0.4 –
7.1 – 2.9
.5Pd/15Na/15Ca/SiO
.5Pd/15Na/15Ca/SiO
.5Pd/15Na/15Ca/SiO
2
, 1st 2.2 31.7 5.4 0.5 38.9 2.0 5.4
2
, 2nd 2.9 31.5 1.0 0.4 5.6 – 1.6
2
, 3rd 1.9 29.6 0.6 0.3 5.9 – 1.3
Entry
A
B
C
D
E
F
Total Alcohols Overall
1
2
5
5
5
Pd/NaY, 2008C
Pd/CsY, 2008C
Pd/CsY, 2008C
6.4 14.6
4.9 23.3
7.6 33.5
7.3 47
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
21.0
28.2
41.1
56.8
55.2
hemiacetal produced by the reaction of an aldehyde and an al-
cohol yields an ester. This pathway has been utilized for the
[
a]
Pd/CsY, 2158C
0.8 0.6 1.1
[
b]
Pd/CsY, 2408C
4.9 43.4 1.6 1.2 2.8
1.3
[11–13]
single-step production of ethyl acetate from ethanol.
In
[
a] 19 mg butylether formed. [b] 156 mg butylether formed.
the case of decarbonylation reactions, CÀC bond scission is
a necessary step in the production of hydrogen from organic
[14]
The yield to diesel-range precursors was consistently lower
maximum of 56.8%, Table 1) than that of the hydrotalcite-type
oxygenates during reforming over Pt and Pd-based catalysts.
(
The hydrotalcite-supported metal catalysts studied previous-
ly were made on low surface area supports, ranging from 6 to
materials, which took the reaction to full conversion. It is also
worth noting that higher temperatures gave higher yields of
etherification products. Ether formation from alcohols is a well-
known acid-catalyzed reaction, suggesting that there are resid-
ual acid sites on the zeolites. This is undesirable, since, in
a commercial process, these oxygenates would be subjected
to hydrodeoxygenation to form alkanes. In this process, ethers,
such as dibutyl ether, may undergo CÀO bond cleavage to
form butanes, which are unsuitable for blending with diesel
fuel and generally lower in value.
2
À1
26 m g , and we hypothesized that activity could be en-
hanced by increasing the surface area. To improve the reactivi-
ty, the hydrotalcite catalyst was calcined at 823 K prior to the
[
15]
ion exchange step. Consistent with previous studies,
this
treatment increased the surface area of the raw material up to
2
À1
200 m g , indicating the formation of the higher surface area
periclase structure. However, reaction studies did not show the
expected appreciable improvements in the yield of the reac-
tion (83% vs. 80% for the pre-calcined HT and the control, re-
spectively), or increased selectivity to the C11 + fraction of the
products. These results are likely a result of the fact that metal
oxide reverts to the lamellar structure upon exposure to
Silica-based materials were also investigated as catalysts for
the ABE condensation reaction. NaOH and CaO supported on
silica have been reported as active catalysts for aldol conden-
[
10]
[15]
sation reactions. Excellent yields were achieved upon sup-
porting Pd on the basified silica catalysts (Table 2). However,
the recyclability of these catalysts was rather poor (last three
entries in Table 2) in comparison to the previously reported hy-
drotalcite-based catalysts; the yield of ABE products in the
third cycle was less than half of that in the first cycle, com-
pared to a modest 10% loss of ABE yield in the case of hydro-
water during the metal impregnation as indicated by the
measured surface area of the finished metal-hydrotalcite
2
À1
material (26 m g ).
It is therefore clear, that to increase the surface area of the
catalyst it is necessary to retain the cubic structure of the peri-
clase material. This was achieved by co-precipitating the Cu,
Mg and Al hydroxycarbonates from a solution of the nitrates,
forming a Cu-doped hydrotalcite, and calcining it to form the
high-surface-area periclase structure. Our results (Table S1)
indicate that Cu-doped HT materials can be synthesized with
[
7]
talcite-based catalysts.
It is possible that the water formed in the reaction leaches
away the easily-soluble NaOH and Ca(OH) , and thereby
2
2
À1
deactivates the catalyst. We also noted the formation of alco-
hols in the product stream. This observation suggests that
surfaces areas between 140 and 200 m g after calcination at
[16]
823 K, in agreement with literature reports.
&
ChemCatChem 2017, 9, 1 – 10
www.chemcatchem.org
2
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
which group IB metals (Cu, Ag, Au) are supported on MgO-
SiO solids. The group IB metals act as redox sites, while the
2
MgO-SiO supports provide acidic and basic sites to catalyze
2
the condensation and dehydration reactions.
Similarly, in our case, Cu provides the sites for the dehydro-
genation reaction, while the MgO/Al O support provides basic
2
3
sites for the aldol condensation reaction of acetone with the in
situ generated reactive aldehydes.
Another technique employed for high-surface-area HT-sup-
ported catalysts is incipient wetness impregnation of metal ni-
[
18]
trates on HT. Therefore, 2.5% CuHT catalyst was prepared
using an incipient wetness impregnation method; the surface
2
À1
area of the sample was 198 m g . On testing against the
.5% CuHT catalyst prepared by coprecipitation, the catalyst
2
prepared by incipient wetness impregnation showed higher
yields (Table 3). We posited that co-precipitating the copper ni-
trate with the magnesium and aluminum nitrates places
copper in random positions throughout the structure and not
Figure 1. Effects of Cu loading on reaction yield and selectivity. Batch reac-
tor, 513 K, 18 h, 2.3 mmol acetone, 3.7 mmol butanol, 1 mmol ethanol,
on the surface, where it can be reduced by H and be active
2
350 mg catalyst in 1 mL BuOH. Dashed lines indicate qualitative trends.
for dehydrogenation reactions. To test this hypothesis, the cat-
alyst was characterized using X-ray absorption spectroscopy.
The X-ray absorption near edge structure (XANES) results
(Figure 2) are consistent with this hypothesis, as the white line
decreases in intensity and the edge is shifted to lower photon
energies. After treatment in a hydrogen/helium mixture at
The yields of the ABE condensation reaction using catalysts
prepared by coprecipitation are shown in Figure 1. An increase
in the yields of ketones A–F was observed with increasing Cu
content to approximately 2.5% Cu loading. Furthermore, the
selectivity to 2-ethylhexanol and its condensation product with
acetone, 4-ethyl-2-nonanone, also increased. Increasing Cu
content beyond 2.5% did not result in higher yields, but result-
ed in increased production of butyl butyrate and decreased
formation of 2-ethylhexanol and heavier condensation prod-
ucts. This undesirable effect is likely a result of the formation
of a hemiacetal over the basic sites of the HT and its subse-
quent dehydrogenation by the Cu, as shown in Scheme 2. Cu
catalysts are known to catalyze hemiacetal dehydrogenation,
0
523 K, we expect all accessible Cu to be reduced to a Cu spe-
[19]
cies, in accordance with literature TPR data. However, in the
case of these catalysts, a significant portion of Cu remains oxi-
dized, in a 2+ oxidation state, and in a coordination environ-
2
+
ment reminiscent of Cu(H O)6 , as evidenced by the shape of
2
the white line. Cu is therefore octahedrally coordinated in the
calcined catalyst, which suggests that Cu substitutes for Mg in
the Mg-Al oxide structure and is inaccessible to hydrogen or
the alcohols for reduction. This hypothesis is also consistent
[11–13]
[17b,c]
formed over basic or amphoteric supports, to form esters.
with literature reports on butadiene production catalysts
These results suggest that a balance of acid/base and metal
sites is necessary for the selective production of ABE condensa-
tion products. This is consistent with the results reported in
recent literature [17] for ethanol-to-butadiene conversion, in
and our own XPS data, which point to the existence of two
2
+
distinct Cu species (Figure S2).
To understand the reactivity aspects better, additional ex-
periments at short residence times (2 h) were conducted to
track the yield and conversion more systematically.
To keep the conditions closer to industrial processes,
butanol:acetone was limited to a range that was ex-
pected to be produced by the fermentation process-
es, and the use of large excess of butanol solvent
was discontinued.. While these conditions resulted in
lower conversions (Table 3), they are much more indi-
cative of the suitability of a given catalyst for eventu-
Scheme 2. Esterification of butanol and butyraldehyde.
Table 3. Effects of the solvent, the preparation method and the ratio of substrate to
catalyst on the conversion and the yield of 6-undecanone at 513 K, batch reaction.
Substrate is ABE mixture.
al industrial use. To further increase the aldol conden-
sation rates, we investigated the possibility of using
higher alkaline earth element oxides, since these
oxides are known to be more basic than MgO and
Total
yield
Yield E
[%]
Solvent
substrate/
catalyst
[mass]
Reaction
time
[h]
[14]
[
%]
Al O . Achieving high surface areas for these ele-
2 3
[20]
ments in their pure oxide form is challenging.
CuHT (IWI)
CuHT (Co-Pr)
CuHT (IWI)
CuHT (Co-Pr)
CuHT (Ion-exch)
89.6
73.0
47.5
37.8
24
60.4
35.5
13.8
12.8
2.2
Butanol
Butanol
None
None
None
1.25
1.25
2
2
2
20
20
2
2
2
Hence, these catalysts were supported on SiO , using
2
an incipient wetness impregnation of the nitrates,
and were calcined at 773 K, with the aim of forming
small metal oxide clusters on SiO . These materials
2
ChemCatChem 2017, 9, 1 – 10
www.chemcatchem.org
3
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Figure 2. X-ray absorption spectra for 2.5% Cu/HT catalysts and standards. A: Spectra recorded at room temperature in transmission mode after reduction at
523 K under a flowing 50% H
2
/He mixture. B: Spectra recorded in fluorescence (Calcined CuHT) and transmission (CuO and catalyst), before reduction, under
He at room temperature.
were then physically mixed with a commercial Cu/ZnO/Al O
2
3
catalyst prior to reaction to dehydrogenate the alcohols.
Table 4. Percent yields for alternative base screening for ABE condensa-
tion. 523 K: 4.6 mmol acetone, 7.4 mmol butanol, 2 mmol ethanol,
The results of these experiments are shown in Table 4. It is
350 mg basic catalyst in physical mixture with 50 mg Cu/ZnO/Al
2
O
3
. Reac-
clear that MO/SiO materials behaved poorly, since the product
2
tion time 2 hours.
yield was about a third of that of a HT catalyst tested under
the same conditions. A possible reason for the poor per-
formance of these materials is that they did not form the cubic
MO structure, as evidenced by their X-ray diffractograms which
show broad peak attributed to amorphous SiO2 (Figure S1).
The detrimental effects of the formation of silicates for CÀC
bond formation reactions were reported in the past for buta-
Entry
A
B
C
D
E
F
Total Alcohols Overall
BaO/SiO
MgO/SiO
SrO/SiO
CaO/SiO
2
–
–
–
–
–
–
–
–
–
–
–
0.6
–
–
0.1
–
–
–
–
–
–
–
–
–
ꢀ0
ꢀ0
2
2
0.8 6.7 0.2 0.2
1.9 15.9 0.3
2.7 17.8 2.6 0.7 11.9
1.9 13.4 1.6 0.4 7.8
8.0
2
–
0.6
1.8
1.1
3.1
2.4
5.1
0.2
4.0
19.3
37.5
26.2
52.8
44.8
53.1
26
SrO/MgO (1%)
SrO/MgO (2%)
CaO/MgO (10%) 2.0 12.2 6.3 1.1 28.1
CaO/MgO (5%)
CaO/MgO (1%)
Hydroxyapatite
Perovskite
[21]
diene formation reactions.
Perovskite and hydroxyapatite are known basic catalysts
that have demonstrated effectiveness for other base-catalyzed
2.5 15.5 4.2 0.7 19.5
2.2 10.7 7.1 1.0 27.0
2.4 21.3 0.5 0.3 1.3
0.5 22.7 0.6
0.5 6.0 1.6 0.5 20.0 1.8 28.3
–
–
[22]
reactions, such as ethanol oligomerization. Given this, a phys-
–
5.1
32.9
58.7
ical mixture with Cu-containing catalysts was tested in the ABE
condensation reaction. Unfortunately, the yields were not high
HT
&
ChemCatChem 2017, 9, 1 – 10
www.chemcatchem.org
4
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
enough (<40%) to warrant further investigations.
This is consistent with the observations of Sacia
[
23]
et al., who showed low activity of these catalysts
in the trimerization of methyl ketones at similarly
low temperatures.
Doping more basic elements into an MgO struc-
ture has been previously proposed as a strategy to
improve the basicity of the catalyst and, conse-
[
24,25]
quently, the aldol condensation reaction rate.
On preparation of such oxides from the calcination
[
26]
of the corresponding mixed oxalates,
mixed
oxides were the dominant phase in low heteroatom
substitution percentages (Figures S3 and S5).
However, increasing the doping gave rise to another
phase, possibly the mixed (M,Mg)CO carbonate. It is
3
known that oxalates decompose into carbonates
upon heating, with the carbonates decomposing to
the oxides upon further heating. TGA experiments
Figure 3. Effects of butanol pressure on aldol condensation rate. Gas-phase flow reactor,
0.4 kPa butyraldehyde, 2 kPa acetone, 2.5% Cu/HT catalyst, 473 K.
(Figure S4 and Figure S6) demonstrated an increase
in decomposition temperature with an increase in the atomic
number of the substituted group IIA element, suggesting an
increase in the fraction of carbonates. Since the carbonates are
less basic than the oxides, we hypothesized that the increasing
proportion of carbonates in the samples results in the ob-
served reduction of the reaction rate upon increasing the
doping over 1%.
a Langmuir-Hinshelwood rate model in which the butanol
competitively adsorbs on surface sites. This observation match-
es the ethanol inhibition effects reported by Young et al.
[29]
during ethanol Guerbet catalysis over TiO2. Also, Ho et al. re-
ported that ethanol adsorbs competitively over hydroxyapatite
[30]
catalysts during the same reaction. The apparent activation
energy for the aldol condensation reaction over HT is
À1
To understand the mechanism of the ABE reaction, kinetics
of the ABE condensation reactions over both HT and CuHT cat-
alysts were investigated using a gas-phase flow reactor. First,
transition-metal-free HT catalysts were used to study the mech-
anism of the aldol condensation. If acetone and butanol were
flowed over the HT catalyst, the cross-coupling prod-
uct yield at steady state was very low and the efflu-
ent was composed mainly of acetone coupling and
oligomerization products. This suggests that the
27 kJmol , as seen from the slope of Figure S8.
As shown in Scheme 3, aldol condensation reactions over
basic catalysts are generally believed to take place via the ab-
straction of a proton in the position alpha to the carbonyl to
form an enolate, followed by an attack of the electron-rich
[
27]
transfer (Meerwein-Ponndorf-Verley)
hydrogena-
tion from the butanol to the acetone is much slower
than the formation of the CÀC bond. On the contra-
ry, if the butanol feed was spiked with 5% (w/v)
butyraldehyde, the main product was the cross-cou-
pling product, 3-hepten-2-one, reflecting the fact
that butyraldehyde is a much stronger electrophile
than acetone.
These experiments also indicate that butanol ad-
sorbs competitively with acetone on the catalyst sur-
face. At a constant acetone and butyraldehyde pres-
sure, a reduction of rates as butanol pressure in-
creased was observed (Figure 3). This is consistent
with earlier reports, which describe a dissociative ad-
sorption of alcohols over metal oxide surfaces to
[
28]
form alkoxides.
In this way, unreactive alcohols
compete for surface sites with the reactive ketones
and aldehydes. To check for the consistency of this
model, a Langmuir-Hinshelwood rate law was fit to
the dependence of the reaction rate on the butanol.
Figure S7 shows that the inverse rate is proportional
to the butanol pressure, which is consistent with
Scheme 3. Condensation of acetone and butyraldehyde over a MgAlO surface.
ChemCatChem 2017, 9, 1 – 10
www.chemcatchem.org
5
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
enolate onto the electrophilic carbonyl carbon. Wang et al.
demonstrated that the enolate formation over anatase TiO2
was rate-determining of the aldol condensation reaction of
explained by the fact that butyraldehyde is a much stronger
electrophile than acetone, resulting in a much more facile CÀC
bond formation in the case of the ABE condensation. As such,
the rate-determining step in the sequence is the associative
desorption of the ketol or the dehydration of the ketol. To dis-
tinguish between those two possibilities, we prepared the
ketol ex situ and fed it over the HT catalyst. No ketol was de-
tected in the effluent of the reactor, indicating that the ketol
was rapidly decomposed over the catalyst, thereby ruling the
dehydration out as a rate-determining step. As such, we con-
clude that the concerted abstraction of a proton from the sur-
face to form the ketol is most likely the rate determining step
of the aldol condensation.
[
31]
propionaldehyde and acetone,
while Shylesh et al. found
that the enolate formation step was equilibrated during ace-
tone and other methyl ketone trimerization reactions over cal-
[
32]
cined hydrotalcite. In the ABE condensation, the regioselec-
tivity of the ABE products is consistent with the enolate forma-
tion primarily occurring from acetone. Only 2-heptanone and
its unsaturated counterpart, 3-hepten-2-one are observed. The
rate of the coupling reaction between acetone and butyralde-
hyde shows a first-order dependence on acetone (Figure 4A)
and the dependence on butyraldehyde is also first-order (Fig-
ure 4B). These observations support reversible enolate forma-
tion; an irreversible enolate formation step would result in an
aldol condensation rate independent of butyraldehyde
pressure.
For the CuHT catalysts, we studied a representative 2.5%
Cu/HT catalyst. The butanol dehydrogenation rate is propor-
tional to the butanol partial pressure at the low-pressure
regime (Figure 5). This is consistent with an earlier report for
isopropanol dehydrogenation over Cu/C catalysts, in which
Furthermore, the observed KIE with [D ]acetone (k /k =2.4)
6
H
D
[33]
is consistent with the rate-determining step for the aldol con-
Rioux and Vannice suggested that the rate-determining step
for alcohol dehydrogenation over Cu surfaces is the OÀH bond
activation. To test this hypothesis, we compared the dehydro-
genation rates for butanol and butanol-OD. We observed a KIE
densation involving the activation of a CÀH or OÀH bond. The
[32]
apparent contradiction with the results of Shylesh et al.,
who showed a KIE equal to 0.96 with [D ]acetone, can be
6
Figure 4. Dependence of cross-condensation rate on acetone pressure (A) and butyraldehyde pressure (B). 473 K, HT catalyst, 0.4 kPa butyraldehyde, 2 kPa
acetone.
&
ChemCatChem 2017, 9, 1 – 10
www.chemcatchem.org
6
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
of the method proposed by Putanov et al., whereby
magnesium oxalate is precipitated from a magnesium
[26]
acetate solution upon addition of oxalic acid. In our
experiments, we added a 0.83m solution of oxalic acid
(1.2 equivalents) to a 25 wt% magnesium acetate solu-
tion, spiked with calcium or strontium acetate. The re-
sulting milky suspension was aged for 16 h and subse-
quently the solids filtered and dried in vacuo at 343 K.
À1
Calcination of the mixed oxalates at 773 K (5 Kmin
h) afforded mixed oxides.
,
4
Coprecipitated Cu catalysts were prepared by modifying
[36]
the procedure outlined by Climent et al. In this proce-
dure, a solution of magnesium and aluminum nitrates
(
1.5m total metal concentration, 3:1 mol Mg:Al) was
mixed with a copper and aluminum nitrate solution
1.5m total metal concentration, 3:1 mol Cu:Al). The
(
resulting solution was heated to 333 K and an equal
volume of a solution of ammonium hydroxide (3.375m)
and ammonium carbonate (1m) was added to it drop-
wise. The resulting slurry was stirred overnight and the
solids filtered and washed with a volume of distilled
water equal to five times the volume of the solution.
Figure 5. Dependence of butanol dehydrogenation on butanol pressure. 2.5% Cu/HT
IWI, 473 K, 2.5 kPa acetone, balance He.
equal to 1.1, which is inconsistent with an OÀH bond scission
After drying at ambient air at 373 K, the solids were treated in am-
À1
as the rate-determining step. This suggests that the CÀH bond
activation is the rate-determining step for the dehydrogena-
tion of butanol on Cu surfaces. This is consistent with the TPD
studies of Bowker and Madix, who showed that O-H activation
and alkoxide formation occurs over Cu surfaces at low temper-
atures, as opposed to CÀH bond activation, which occurs at
bient air at 823 K for 4 h (ramp rate 1 Kmin ). NiHT and CoHT cata-
lysts were prepared in a similar way, using Ni(NO
Co(NO ·6H O, respectively, in the place of copper nitrate
hemipentahydrate.
) ·6H O and
3 2 2
)
3
2
2
Hydrotalcite-supported Cu catalysts were also prepared by incipi-
ent wetness impregnation of a solution of copper nitrate hemipen-
tahydrate into a mixed magnesium-aluminum oxide. This oxide
was prepared by calcination of synthetic hydrotalcite (Sigma–
[34,35]
higher temperatures.
À1
Aldrich) at 823 K for 4 h (ramp rate 1 Kmin ). Following impregna-
tion, the catalyst precursor was dried in ambient air at 373 K and
subsequently calcined at 823 K for 4 h (ramp rate 1 Kmin ). For
Conclusions
À1
In this work, we have examined a number of base-supported
metal catalysts for the condensation of acetone with butanol
and ethanol to form drop-in diesel fuel precursor ketones. Con-
siderable improvements of the reaction rates were achieved by
increasing the surface area of hydrotalcite-supported Cu cata-
lysts. Additionally, we showed that supporting Cu onto cal-
cined hydrotalcite materials by incipient wetness impregnation
gave catalysts that were tuned to give high selectivity to the
ABE ketones without extensive formation of undesirable ester
byproducts. Alcohol dehydrogenation rates over the Cu surfa-
ces are proportional to the alcohol pressure and do not show
any kinetic isotope effect with butanol-OD. This observation
suggests that the rate-determining step of the dehydrogena-
tion is the CÀH bond cleavage. On the other hand, the aldol
condensation of butyraldehyde with acetone proceeds over
the HT surface via an equilibrated enolate formation, followed
by a rate-limiting surface proton abstraction to form the ketol.
The ketol is then rapidly dehydrated to form the unsaturated
ketone.
the preparation of Pd/HT catalysts, a similar procedure was fol-
lowed, using palladium nitrate hydrate (Sigma Aldrich). For RuHT,
the calcined hydrotalcite was impregnated with an aqueous solu-
tion of RuCl and calcined, whereas for PtHT, the calcined hydrotal-
3
cite was impregnated with an aqueous solution of H PtCl and re-
2
6
À1
duced at 723 K in H for 2 h (ramp 2 Kmin ) after drying at 373 K
2
overnight. Titanium dioxide was prepared following a procedure
[33]
reported by Wang and Ying. A mixture of ethanol and water was
added dropwise to a titanium isopropoxide solution in ethanol.
The ratio of titanium isopropoxide to water was 1:100. After hy-
drolysis, the resulting suspension was aged for 16 h at ambient
temperature. The solids were separated by filtration and dried in
stagnant ambient air at 373 K for 16 h.
[37]
Hydroxyapatite was prepared according to Wang, et al. In this
process, a stoichiometric quantity of an ammonium hydrogen
phosphate solution was added dropwise at ambient temperature
to a calcium nitrate solution, for which pH was adjusted to 11 with
aqueous ammonium hydroxide solution. The slurry was aged at
3
63 K for 1 h and the solids were subsequently filtered and
washed with copious amounts of water. After that, they were treat-
ed in ambient air at 373 K for at least 16 h and subsequently treat-
À1
ed for 4 h at 573 K (ramp rate 5 Kmin ).
Experimental Section
Surface areas of the catalysts were measured by means of nitrogen
physisorption using a Micromeritics Tristar 3000 analyzer. The data
were analyzed using the BET and BJH methods for surface area
and pore size, respectively. The structure of the catalysts was
investigated by X-ray diffraction (XRD) and X-ray absorption
[
8]
Pd-modified
Y
zeolite
Pd were prepared based on methods described in detail else-
where in the literature. Preparation of mixed
M Mg O oxides, where M is Ca or Sr, was achieved by a variation
and
Na/Ca/SiO -supported
2
[
9]
x
1-x
ChemCatChem 2017, 9, 1 – 10
www.chemcatchem.org
7
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
spectroscopy (XAS). XRD experiments were performed using
Acknowledgements
a Bruker D8 instrument using a 2q-q geometry, scanning from
o
À
2
q=20 to 608, at a rate of 0.028s . Thermogravimetric analysis of
This research was funded by BP through the Energy Biosciences
Institute. MRCAT operations are supported by the Department of
Energy and the MRCAT member institutions. This research used
resources of the Advanced Photon Source, a U.S. Department of
Energy (DOE) Office of Science User Facility operated for the DOE
Office of Science by Argonne National Laboratory under Contract
No. DE-AC02-06CH11357. We also acknowledge NSF grant
the catalysts was performed using a TA8020 TGA instrument, using
a 5 Kmin ramp to 1123 K under air flow. Alumina crucibles were
used to hold 20–40 mg of sample.
À1
X-ray absorption spectroscopy experiments were performed on the
Sector 10, Materials Research Collaborative Access Team, Insertion
Device (ID) and bending magnet (BM) beamlines at the Advanced
Photon Source, Argonne National Lab. Experiments were conduct-
ed at the Pd K (24350 eV) and Cu K (8979 eV) edges. Experiments
were performed in either transmission (Pd and Cu) or fluorescence
1428149 for XPS analysis.
(
Cu) detection modes. A series of 3 ion chambers were used, with
Keywords: ABE · aldol condensation · copper · hydrotalcite ·
kinetic isotope effect
the third detector simultaneously measuring a reference foil with
all experiments for energy calibration. A Lytle detector was used as
the fluorescence detector. For transmission experiments, samples
were packed into self-supporting wafers, in a 6-well sample holder.
Sample loading was calculated to give an absorbance (m ) of <2.5
to avoid self-absorption effects, and an edge step (Dm ) between
[
1] “Triglycerides and Oils for Biofuels”: D. Y. Murzin, P. Mꢂki-Arvela, I. L. Si-
makova, Kirk-Othmer Encyclopedia of Chemical Technology, 2012, pp. 1–
x
14, John Wiley & Sons, Hoboken.
x
[
[
2] P. Mꢂki-Arvela, I. Kubickova, M. Snꢃre, K. Erꢂnen, D. Y. Murzin, Energy
Fuels 2007, 21, 30–41.
3] C. Somerville, Curr. Biol. 2007, 17, R115–R119.
0
.2 and 1.5. Fluorescence samples were packed into self-support-
ing wafers in a custom fluorescence holder that held the sample at
a 458 angle incident to the beam path. The detector was placed at
a 908 angle incident to the beam path.
[4] a) B. Wicke, V. Dornburg, M. Junginger, A. Faaij, Biomass Bioenergy 2008,
32, 1322–1337; b) L. P. Koh, D. S. Wilcove, Nature 2007, 448, 993–994.
[
5] a) T. J. Schwartz, B. J. O’Neil, B. H. Chanks, J. A. Dumesic, ACS Catal.
014, 4, 2060–2069; b) A. Deneyer, T. Renders, J. Van Aelst, S. Van den
X-ray photoelectron spectroscopy (XPS) experiments were per-
formed at the Surface Analysis Facility of the University of Dela-
ware. A Thermo Scientific Kalpha+ instrument, with an AlKa source
was used for the experiments. The analysis chamber pressure was
maintained at 5ꢁ10–8 mbar or lower and the spot size was
2
Soch, D. Gabriꢄls, B. Sels, Curr. Opin. Chem. Biol. 2015, 29, 40–48;
c) K. A. Goulas, F. D. Toste, Curr. Opin. Biotechnol. 2016, 38, 47–53;
d) T. J. Schwartz, B. H. Shanks, J. A. Dumesic, Curr. Opin. Biotechnol.
2016, 38, 54–62; e) L. Wu, T. Moteki, A. A. Gokhale, D. W. Flaherty, F. D.
Toste, Chem 2016, 1, 32–58; f) Y. Huo, H. Zeng, Y. Zhang, ChemSusChem
0
.4 mm in diameter. The samples were placed on conductive
2
016, 9, 1078–1080.
carbon tape and the analyses were conducted with the electron
flood gun on to minimize charging effects.
[
[
6] a) P. Anbarasan, Z. C. Baer, S. Sreekumar, E. Gross, J. B. Binder, H. W.
Blanch, D. S. Clark, F. D. Toste, Nature 2012, 491, 235–239; b) S. Sreeku-
mar, Z. C. Baer, A. Pazhamalai, G. Gunbas, A. Grippo, H. W. Blanch, D. S.
Clark, F. D. Toste, Nat. Protoc. 2015, 10, 528–537; c) K. A. Goulas, S. Sree-
kumar, Y. Song, P. Kharidehal, G. Gunbas, P. J. Dietrich, G. R. Johnson,
Y. C. Wang, A. M. Grippo, L. C. Grabow, A. A. Gokhale, F. D. Toste, J. Am.
Chem. Soc. 2016, 138, 6805–6812.
Samples were contained in 1’’ OD quartz tubes connected to Ultra-
torr fittings with welded ball valves at the end to control the gas
atmosphere. For fluorescence experiments, a similar 1’’ OD quartz
tube with a cone attached to the side was used. The samples were
then reduced in situ using a 50% H /He gas mixture at the prescri-
2
7] a) S. Sreekumar, Z. C. Baer, E. Gross, S. Padmanaban, K. A. Goulas, G.
Gunbas, S. Alayoglu, H. W. Blanch, D. S. Clark, F. D. Toste, ChemSusChem
bed temperature. Samples were held for 0.5–1 h (reducing time
was determined to not have an effect on the final state), at 523 K
after which the gas flow was switched to He (Ultra High Purity
grade) for 10–15 min. The samples were then cooled to room tem-
perature and scanned at room temperature under a static He
atmosphere.
2
014, 7, 2445–2448.
[8] For subsequently reported catalyst systems for ABE condensation: a) S.
Nahreen, R. B. Gupta, Energy Fuels 2013, 27, 2116–2125; b) S. B. Bankar,
S. A. Survase, H. Ojamo, T. Granstçm, Bioresour. Technol. 2013, 140, 269–
276; c) G. Xu, Q. Li, J. Feng, Q. Liu, Z. Zhang, X. Wang, X. Zhang, X. Mu,
ChemSusChem 2014, 7, 105–109; d) G. Onyestyꢅk, Gy. Novodꢅrszki, A. F.
Wellisch, A. Pilbꢅth, Catal. Sci. Technol. 2016, 6, 4516–4524; e) H. T. VO,
S. M. Yeo, D. Dahnum, J. Jae, C. S. Hong, H. Lee, Chem. Eng. J. 2016,
DOI: 10.1016/j.cej.2016.11.044.
In all reaction experiments, we used an E:A:B=1:3:6 (by weight)
ratio for ethanol, acetone and butanol mixing, since this is the
ratio produced by C. acetobutylicum during fermentation. Batch
reactions were conducted in pressure tubes (Q-Tubes), using a pro-
[
9] E. J. Rode, P. E. Gee, L. N. Marquez, T. Uemura, M. Bazargani, Catal. Lett.
1991, 9, 103–113.
[6,7]
cedure described in detail elsewhere.
Gas-phase reactions were
[
[
[
10] C. A. Hamilton, Appl. Catal. A 2004, 263, 63–70.
11] K. Inui, T. Kurabayashi, S. Sato, J. Catal. 2002, 212, 207–215.
12] A. B. Gaspar, F. G. Barbosa, S. Letichevsky, L. G. Appel, Appl. Catal. A
conducted in a tubular fused silica reactor with plug-flow hydrody-
namics. In this reactor, the catalyst was supported on a quartz frit.
Before reaction, the catalyst was sieved down to a size under
2
010, 380, 113–117.
1
80 mm to avoid mass transfer effects during reaction. The reactor
[13] N. Iwasa, N. Takezawa, Bull. Chem. Soc. Jpn. 1991, 64, 2619–2623.
[14] a) D. A. Simonetti, E. L. Kunkes, J. A. Dumesic, J. Catal. 2007, 247, 298–
306; b) G. W. Huber, R. D. Cortright, J. A. Dumesic, Angew. Chem. Int. Ed.
was enclosed in a furnace (ATS Pennsylvania) with aluminum in-
serts and the temperature was controlled by a PID controller
2
004, 43, 1549–1551; Angew. Chem. 2004, 116, 1575–1577.
15] D. P. Debecker, E. M. Gaigneaux, G. Busca, Chem. Eur. J. 2009, 15, 3920–
935 and references therein.
(
Watlow). Gas flow through the catalyst was regulated using mass
[
flow controllers (Parker). Liquid reactants were vaporized into the
gas stream through a syringe port in the heated transfer lines.
Their flow was regulated by a syringe pump (Legato 100).The efflu-
ent of the reactor was analyzed online using a gas chromatograph
3
[
[
16] V. Rives, S. Kannan, J. Mater. Chem. 2000, 10, 489–495.
17] a) S. Shylesh, A. A. Gokhale, C. D. Scown, D. Kim, C. R. Ho, A. T. Bell,
ChemSusChem 2016, 9, 1462–1472; b) C. Angelici, F. Meirer, A. M. J. van
der Eerden, H. L. Schaink, A. Goryachev, J. P. Hofmann, E. J. M. Hensen,
B. M. Weckhuysen, P. C. A. Bruijnincx, ACS Catal. 2015, 5, 6005–6015;
c) C. Angelici, M. E. Z. Velthoen, B. M. Weckhuysen, P. C. A. Bruijnincx,
ChemSusChem 2014, 7, 2505–2515; d) W. Janssens, E. V. Makshina, P. Va-
(
Shimadzu GC 2014) equipped with a flame ionization detector
and an HP-1 capillary column.
&
ChemCatChem 2017, 9, 1 – 10
www.chemcatchem.org
8
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
nelderen, F. De Clippel, K. Houthoofd, S. Kerkhofs, J. A. Martens, P. A.
Jacobs, B. F. Sels, ChemSusChem 2015, 8, 994–1008.
18] K. O. Christensen, D. Chen, R. Lødeng, A. Holmen, Appl. Catal. A 2006,
[26] P. Putanov, E. Kis, G. Boskovic, K. Lazar, Appl. Catal. 1991, 73, 17–26.
[27] P. S. Kumbhar, J. Sanchez-Valente, J. Lopez, F. Figueras, Chem. Commun.
1998, 535–536.
[
[
3
14, 9–22.
[28] M. A. Barteau, Chem. Rev. 1996, 96, 1413–1430 and references therein.
[29] Z. D. Young, S. Hanspal, R. J. Davis, ACS Catal. 2016, 6, 3193–3202.
[30] C. R. Ho, S. Shylesh, A. T. Bell, ACS Catal. 2016, 6, 939–948.
[31] S. Wang, K. Goulas, E. Iglesia, J. Catal. 2016, 340, 302–320.
[32] S. Shylesh, D. Kim, A. A. Gokhale, C. G. Canlas, J. O. Struppe, C. R. Ho, D.
Jadhav, A. Yeh, A. T. Bell, Ind. Eng. Chem. Res. 2016, 55, 10635–10644.
[33] R. M. Rioux, M. A. Vannice, J. Catal. 2003, 216, 362–376.
[34] M. Bowker, R. J. Madix, Surf. Sci. 1980, 95, 190–206.
19] S. Galvagno, C. Crisafulli, R. Maggiore, A. Giannetto, J. Schwank, J.
Therm. Anal. 1987, 32, 471–483.
[
[
[
[
20] G. Zhang, H. Hattori, K. Tanabe, Appl. Catal. 1988, 36, 189–197.
21] G. Natta, R. Rigamonti, Chim. Ind. 1947, 29, 239–243.
22] J. T. Kozlowski, R. J. Davis, ACS Catal. 2013, 3, 1588–1600.
23] a) E. R. Sacia, B. Madhesan, M. H. Deaner, K. A. Goulas, F. D. Toste, A. T.
Bell, ChemSusChem 2015, 8, 1726–1736; b) S. Sreekumar, M. Balakrishn-
an, K. A. Goulas, G. Gunbas, A. A. Gokhale, Y. L. Louie, A. Grippo, C. D.
Scown, A. T. Bell, F. D. Toste, ChemSusChem 2015, 8, 2609–2614; c) M.
Balakrishnan, E. R. Sacia, S. Sreekumar, G. Gunbas, A. A. Gokhale, C. D.
Scown, F. D. Toste, A. T. Bell, Proc. Natl. Acad. Sci. USA 2015, 112, 7645–
[35] M. Bowker, R. J. Madix, Surf. Sci. 1982, 116, 549–572.
[36] M. J. Climent, A. Corma, S. Iborra, A. Velty, J. Catal. 2004, 221, 474–482.
[37] K. Wang, G. J. Kennedy, R. A. Cook, J. Mol. Catal. A 2009, 298, 88–93.
7
649.
24] J. I. Di Cosimo, V. K. Dꢆez, C. R. Apesteguꢆa, Appl. Catal. A 1996, 137,
49–166.
25] W. Q. Shen, G. A. Tompsett, R. Xing, W. C. Conner, G. W. Huber, J. Catal.
012, 286, 248–259.
Manuscript received: November 23, 2016
Revised: January 9, 2017
[
[
1
Accepted Article published: January 9, 2017
Final Article published: && &&, 0000
2
ChemCatChem 2017, 9, 1 – 10
www.chemcatchem.org
9
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

FULL PAPERS
K. A. Goulas, G. Gunbas, P. J. Dietrich,
S. Sreekumar, A. Grippo, J. P. Chen,
A. A. Gokhale, F. D. Toste*
&& – &&
ABE Condensation over Monometallic
Catalysts: Catalyst Characterization
and Kinetics
I C-u: The development and the kinetics
of Cu on hydrotalcite as catalyst for ace-
tone-butanol-ethanol (ABE) condensa-
tion is described. These catalysts oper-
ate with increased rate over previously
reported catalysts. Kinetic analysis of
the reaction over CuHT and HT revealed
that the rate-determining step is the CÀ
H bond activation during oxidation of
alcohols on the Cu surface.
&
ChemCatChem 2017, 9, 1 – 10
www.chemcatchem.org
10
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!