Bifunctional condensation reactions of alcohols on basic oxides modified by copper and potassium

Article abstract of DOI:10.1006/jcat.1998.2009

Alcohol dehydrogenation and condensation reactions are involved in chain growth pathways on Cu/MgCeO_x promoted with potassium. These pathways lead to the formation of isobutanol with high selectivity via reactions of higher alcohols with methanol-derived C₁ species in reaction steps also relevant to higher alcohol synthesis from CO/H₂ mixtures at higher pressures on K-Cu/MgCeO_x catalysts. Ethanol reactions on K-CU_yMg₅CeO_x show that both Cu and basic sites participate in alcohol dehydrogenation and aldol condensation steps leading to n-butyraldehyde and acetone. Chain growth occurs by condensation reactions involving a metal-base bifunctional aldol-type coupling of alcohols. Reactions of ¹²C₂H₅OH-¹³C₂H ₄O mixtures show that direct condensation reactions of ethanol can occur without requiring the intermediate formation of gas phase acetaldehyde. Reactions of C₂H₅OH/D₂ mixtures show that Cu sites increase the rate of aldol condensation by introducing recombinative desorption sites that remove hydrogen atoms formed in C-H activation steps leading to the unsaturated aldol-type species required for chain growth. Reactions of acetaldehyde and ¹³C-labeled methanol lead predominantly to 1-¹³C-propionaldehyde and 2-¹³C-isobutyraldehyde, both of which lead to isobutanol during CO/H₂ reactions. Mixtures of propionaldehyde and ¹³C-labeled methanol lead to singly-labeled isobutyraldehyde. Chain growth to C₂₊ alcohols occurs via addition of a methanol-derived C₁ species to adsorbed oxygen-containing intermediates. The gradual appearance of ¹³C in the unlabeled reactant within these mixtures shows that aldol coupling reactions are reversible. Reverse aldol condensation reactions after intramolecular hydride transfer lead to the formation of acetone from ethanol. Isobutyraldehyde is a preferred end-product of aldol-type chain growth reactions of alcohols because it lacks the two α-hydrogens required for subsequent chain growth. 998 Academic Press.

Full text of DOI:10.1006/jcat.1998.2009

JOURNAL OF CATALYSIS 176, 155–172 (1998)
ARTICLE NO. CA982009
Bifunctional Condensation Reactions of Alcohols on Basic Oxides
Modified by Copper and Potassium
Marcelo J. L. Gines and Enrique Iglesia1
Department of Chemical Engineering, University of California at Berkeley, Berkeley, California 94710
Received September 9, 1997; revised January 12, 1998; accepted January 20, 1998
(1–5). These catalysts produce isobutanol with relatively
low rates and selectivity at lower temperatures and pres-
sures. Alkali-modified methanol synthesis catalysts pro-
duce higher alcohols from CO/H2 mixtures at lower tem-
perature (583 K) and pressures (7.6 MPa) (6–9). These
catalysts produce a mixture of linear and branched alcohols
and maximum isobutanol productivities are 48.6 g/kg cat-h
(8). Recently, K Cu Mg CeOx materials have been shown
to catalyze higher alcohol synthesis at low temperatures
and pressures (10–13). These bifunctional catalysts consist
of Cu crystallites supported on high surface area basic ox-
ides based on K-modified MgO.
Alcohol dehydrogenation and condensation reactions are
involved in chain growth pathways on Cu/MgCeOx promoted
with potassium. These pathways lead to the formation of isobu-
tanol with high selectivity via reactions of higher alcohols with
methanol-derived C1 species in reaction steps also relevant to
higher alcohol synthesis from CO/H2 mixtures at higher pressures
–
–
on K Cu/MgCeOx catalysts. Ethanol reactions on K CuyMg5CeOx
show that both Cu and basic sites participate in alcohol dehydro-
genation and aldol condensation steps leading to n-butyraldehyde
and acetone. Chain growth occurs by condensation reactions
–
–
–
involving a metal-base bifunctional aldol-type coupling of alcohols.
12
Reactions of C2H5OH 13C2H4O mixtures show that direct
–
Isobutanol synthesis requires the initial formation of
methanol and higher linear alcohols and their subse-
quent chain growth leading to 2-methyl-branched alco-
hols with high selectivity (3 and references therein). Cu
metal sites catalyze methanol synthesis and hydrogenation–
dehydrogenation reactions of aldehydes and alcohols (14,
15). Solids based on MgO catalyze condensation reactions
of alcohols and aldehydes with high selectivity (16, 17).
Also, MgO is an active and selective catalyst for Guerbet
condensation reactions of ethanol can occur without requiring the
intermediate formation of gas phase acetaldehyde. Reactions of
C2H5OH/D2 mixtures show that Cu sites increase the rate of aldol
condensation by introducing recombinative desorption sites that
–
remove hydrogen atoms formed in C H activation steps leading
to the unsaturated aldol-type species required for chain growth.
Reactions of acetaldehyde and 13C-labeled methanol lead predomi-
nantly to 1-13C-propionaldehyde and 2-13C-isobutyraldehyde, both
of which lead to isobutanol during CO/H2 reactions. Mixtures of
propionaldehyde and 13C-labeled methanol lead to singly-labeled
isobutyraldehyde. Chain growth to C2+ alcohols occurs via addition
of a methanol-derived C1 species to adsorbed oxygen-containing
intermediates. The gradual appearance of 13C in the unlabeled re-
actant within these mixtures shows that aldol coupling reactions are
reversible. Reverse aldol condensation reactions after intramolecu-
lar hydride transfer lead to the formation of acetone from ethanol.
Isobutyraldehyde is a preferred end-product of aldol-type chain
growth reactions of alcohols because it lacks the two ꢀ-hydrogens
c
–
condensation reactions of primary C2 C5 alcohols with
methanol (18, 19).
The mechanism of methanol synthesis from CO2/CO/H2
mixtures has been extensively studied (20–25). Methanol
synthesis involves the formation of a formate species via
the reaction of CO2 with H-adatoms, followed by a rate-
determining hydrogenolysis step (22–25). Bowker et al.
(22) reported that the mechanism by which methanol is
made on Cu/ZnO/Al2O3 from CO2/H2/He and CO2/CO/H2
feeds is by the co-adsorption of hydrogen and carbon
dioxide on the copper component of the catalyst, form-
ing surface formate species, which undergo hydrogena-
tion/hydrogenolysis reactions to yield methanol. Robbins
et al. (24) showed that from CO/CO2/H2 feeds the rate of
methanol formation on Cu/SiO2 catalysts is proportional
to Cu surface area, suggesting that metallic Cu is the ac-
tive catalytic component. In CO2-free feeds, the synthe-
sis rate, however, is not proportional to the Cu surface
area, indicating that other types of catalytic sites are in-
required for subsequent chain growth.
ꢀ 1998 Academic Press
1. INTRODUCTION
Isobutanol is a useful intermediate in the synthesis of
methyl-tert-butyl-ether after its dehydration to isobutene
and further etherification with methanol. Currently avail-
able catalysts for the synthesis of isobutanol from CO/H2
mixtures with high productivity (>100 g/kg cat-h) require
high temperatures (750–800 K) and pressures (10–30 MPa)
1 Corresponding author. E-mail address: iglesia@cchem.berkeley.edu. volved. They suggested that surface hydroxyls might be
155
0021-9517/98 $25.00
c
Copyright ꢀ 1998 by Academic Press
All rights of reproduction in any form reserved.

156
GINES AND IGLESIA
required for the initial conversion of CO to adsorbed
formate. Using diffuse reflectance infrared Fourier trans-
form (DRIFTS) and temperature-programmed reaction
spectroscopies (TPRS), Neophytides et al. (25) found that
methanol synthesis on a Cu/ZnO/Al2O3 commercial cata-
lyst occurs via the rate-limiting hydrogenation of copper
formate. Recently, it has been proposed that methanol syn-
thesis from carbon oxide/hydrogen mixtures involves the
generation of a formate species on the copper surface,
the formate species then migrates onto ZnO, and is sub-
sequently hydrogenated to methanol (26). Temperature-
programmed infrared (TPIR) spectroscopy of copper for-
mate/ZnO showed the migration of surface formate species
from copper to ZnO at temperatures below 430 K and the
beginning of copper formate decomposition above 430 K
(26).
The formation of C2+ linear alcohols during CO hydro-
genation remains the least understood step and the rate of
formation of ethanol appears to limit the overall rate of
chain growth (4, 10). Several mechanisms have been pro-
SCHEME 1. Homogenous aldol condensation of aldehydes (33).
–
posed for the formation of the initial C C bond:
in the C2 to C3 step (Scheme 3) (7, 8). The retention of
oxygen associated with the C1 intermediate is believed to
be caused by the strong bonding of the alkyl oxygen in the
di-oxygenated RCH(H2COꢁ)CHO intermediate anion to
Cs+ cations, which prevents hydrogenation of the H2COꢁ
group, while allowing complete hydrogenation of the free
–
(a) C C bond formation by CO insertion into metal–
carbon bonds in adsorbed formaldehyde species, observed
on ZnO and ThO2 catalysts (27);
(b) Coupling of two C1 aldehydic intermediates on car-
bides (CaC2, Na2C2, CeC2, and LaC2) (28);
(c) Methanol and ethanol formation from a common in-
termediate on CuO/ZnO/Al2O3 (29);
(d) Direct carbonylation of surface C1 species (CH3OH,
H2CO, and CH3O ) on Cu/ZnO/Al2O3 and Cs Cu/ZnO
catalysts (6, 30).
Ourrecentexperimentsusing13CO/CH3OH/H2 mixtures
have shown that ethanol is formed directly from CO at
short residence time on K Cu0.5Mg5CeOx catalysts (31),
but via methanol coupling reactions on Cs Cu/ZnO/Al2O3
–
–
CHO group to CH3 (7).
Here, we report kinetic and isotopic tracer studies
of alcohol dehydrogenation and condensation reactions,
with emphasis on chain growth pathways on Cu/MgCeOx
solids promoted with potassium. Our results show that di-
rect condensation reactions of ethanol occurs without the
intermediate formation of gas phase acetaldehyde, that Cu
sites increase the rate of aldol condensation by introduc-
ing recombinative desorption sites that disposal hydrogen
ꢁ
–
–
–
catalysts (6, 32).
After initial C C bond formation steps, higher alcohol
–
mixtures form via aldol condensation reactions of surface-
bound aldehydic species (10). Aldol-type condensation re-
actions of alcohols in the liquid phase proceed via the
following reaction steps (Scheme 1): alcohol dehydrogena-
tion to form aldehydes (or ketones), ꢀ-hydrogen abstrac-
tion to form enolate species, nucleophilic addition of eno-
late to carbonyl species followed by dehydration to form
ꢀ,ꢁ-unsaturated aldehyde (or ketone), and hydrogenation
==
==
of the C C and C O bonds to form the corresponding
alcohol (33).
On CuO/ZnO/Al2O3, linear 1-alcohols containing n car-
bons condense mainly to esters with 2n carbon atoms,
ketones with 2n-1 carbons, and ketones with 2n atoms
(Scheme 2) (34). On Cu/ZnO, the aldol-condensation reac-
–
tions occur via the normal mechanism, but on Cs Cu/ZnO
the oxygen associated with the C1 carbon source is retained
SCHEME 2. Pathways to higher oxygenates (34).

BIFUNCTIONAL CONDENSATION REACTIONS
157
N2O/pulse) into a He stream. The number of chemisorbed
oxygen atoms was obtained from the amount of N2O con-
sumed or N2 produced. A saturation coverage of 0.5 : 1
O : Cus was used to estimate Cu surface area and disper-
sion (35–37). Cu dispersion is defined as the ratio of surface
Cu (Cus) to the total number of Cu atoms in the catalyst.
The density of basic sites was determined using a
13CO2/12CO2 exchange method at 573 K. This method pro-
vides a direct measure of the number of sites that can
reversibly interact with CO2 at typical reaction tempera-
tures. Detailed descriptions and experimental procedures
and methods are reported elsewhere (13, 38).
2.3. Catalytic Evaluation and Mechanistic Studies
SCHEME 3. Normal and reversal aldol condensation according to
Nunan et al. (7).
2.3.1. Ethanol reactions. Ethanol reactions were car-
riedoutinagradientlessrecirculatingbatchreactorat573K
and 101.3 kPa. Catalyst samples (18.0 mg) were first re-
duced in H2 (10% H2/He, 175 cm3/min, 5 K/min) at 623 K
for0.5h. Thetemperaturewasloweredto573Kandethanol
was introduced along with a small amount of methane,
which was used as an unreactive internal standard in or-
der to ensure accurate mass balances (C2H5OH/CH4/He =
4.0/2.7/94.6 kPa; C2H5OH, Fisher Scientific, A.C.S. certi-
fied; CH4, Matheson, ultra high purity). H2 was purified by
passing through a catalytic purifier (Matheson, Model 64-
1008). Traces of water were removed from H2 and helium
using molecular sieves (Matheson, Model 452 : 4A). CH4
was used without further purification. The reactant mixture
was circulated continuously through the catalyst bed and
reaction products were sampled by syringe extraction after
various contact times. Reactants and products were ana-
lyzed using a dual detector gas chromatograph (Hewlett-
Packard, Model 5890 II Plus; TCD/FID detectors). One
sample was injected into a 5% phenyl-methyl-silicone cap-
illary column (HP-1, 50 m, 0.32 mm diameter, 1.05 ꢂm film
thickness) and the other into a packed column (Porapak Q,
–
atoms formed in C H activation steps, and that chain
growth to C2+ alcohols occurs via addition of a C1 species
to adsorbed oxygenates. Based on these findings, we pro-
pose that chain growth occurs by condensation reactions
involving a metal-base bifunctional mechanism.
2. EXPERIMENTAL METHODS
2.1. Catalyst Synthesis
–
–
Cu Mg CeOx samples were prepared by coprecipitation
of mixed metal nitrate solutions with an aqueous solution of
KOH (2M) and K2CO3 (1M) at 338 K and a constant pH of
9 in a well-stirred thermostated container (13). The solids
formed were filtered, washed thoroughly with 300–500 cm3
of deionized water at 333 K, and dried at 353 K overnight.
Dried samples were treated in flowing air at 723 K for 4 h
in order to form the corresponding mixed metal oxides.
Residual potassium in all dry samples was less than 0.1 wt%
before alkali impregnation. K was introduced by incipi-
ent wetness of the oxidized samples using 0.25M K2CO3
aqueous solutions (K2CO3: Fisher Scientific, A.C.S. Certi-
fied). The properties of the synthesized samples are listed in
Table 1.
TABLE 1
Composition, Surface Area, and Basic Site Density
of Mixed Metal Oxides
2.2. Catalyst Characterization
BET
surface
area
Cu
content content
Ka
Exchangeable
CO2 at 573 K
Powder X-ray diffraction (XRD) spectra were obtained
using a D5000 Siemens Diffractometer and monochromatic
Cu-Kꢀ radiation. Total surface areas were determined us-
ing the single point BET method by measuring N2 phy-
sisorption at 77 K using a continuous flow Quantasorb Sur-
face Area Analyzer (Quantachrome Corp.). Bulk catalyst
compositions were measured by atomic absorption spec-
troscopy (AAS).
Cub
Sample
(wt%) (wt%) (m2/g) dispersion (10ꢁ6 mol/m2)
Mg5CeOx
Mg5CeOx
Cu0.5Mg5CeOx
Cu0.5Mg5CeOx
Cu7.5Mg5CeOx
0
0.01
0.8
0.1
1.0
1.2
188
155
167
147
92
—
—
0.23
0.14
0.047
0.95
1.7
1.2
2.3
3.3
0
7.7
7.7
49
a Bulk composition measured by atomic absorption.
Cu surface areas in reduced samples were measured by
titrating Cu surface atoms with N2O (Matheson, ultra high
b Dispersion calculated from the ratio of the number of surface Cu
(determined by N2O decompositon at 363 K (35, 36, 37)) to the total
purity) at 363 K. N2O was introduced as pulses (1.60 ꢂmol number of Cu atoms in the catalyst.

158
GINES AND IGLESIA
1.8 m length, 0.32 cm diameter). A thermal conductivity de- spectrometry using a matrix deconvolution method that ac-
tector (TCD) was used to measure the concentration of CO, counts for the natural 13C abundance and mass fragmenta-
CO2, and C2H5OH in the effluent from the packed column. tion patterns of each molecule (39).
A flame ionization detector (FID) was used to measure
2.3.4. Reactions of C2H5OH/D2 mixtures. Reactions of
the concentrations of all organic compounds eluting from
C2H5OH/D2 mixtures were carried out in a gradientless re-
the capillary column. Products were also analyzed by mass
circulatingbatchreactorinordertoprobetherateofrecom-
spectrometry after capillary chromatography in order to
binative hydrogen desorption during ethanol coupling re-
confirm the chemical identity and to determine the isotopic
actions. Catalyst samples (20.0 mg) were treated in H2 (10%
H2/He, 175cm3/min, 5K/min)at623Kfor0.5h. Thetemper-
compositions of reactants and products (Hewlett-Packard,
Model 5890 II Plus GC; Hewlett-Packard, Model 5972 Mass
ature was then decreased to 573 K and the reactant mixture
Selective Detector).
wasintroducedalongwithasmallamountofmethane(reac-
Reaction rates are reported as turnover rates, defined as
the moles of reactant converted to products per unit time
tion mixture: C2H5OH/CH4/D2/He = 3.6/2.4/27.0/68.3 kPa;
C2H5OH, Fisher Scientific, A.C.S. certified; CH4, Mathe-
and per g-atom surface site. The local slope of turnover-
son, ultra high purity; D2, C.P. Matheson). Products were
time plots gives the turnover rate for the formation of each
sampled by syringe extraction from the recirculating stream
product. Turnovers (for reactants) or site-yields (for each
at different contact times and analyzed as described in Sec-
product) are defined as the moles of reactant converted
tion 2.3.1. The deuterium content in reactants and prod-
to a given product per gram-atom surface sites in the re-
ucts was measured by mass spectrometry using previously
actor. When divided by the fractional Cu dispersion, this
reported deconvolution methods (39).
becomes a site-time yield for each product, at least for re-
actions with rate-determining elementary steps requiring
Cu surface atoms. Also, the total surface area, the MgCeOx
surface area (estimated from the difference between the to-
2.3.5. Cross-coupling reactions of 13CH3OH/C2H4O and
13CH3OH/C3H6O mixtures. Cross-coupling reactions
of 13CH3OH (13C, 99%,18O, <1%, Icon Services Inc.)
with acetaldehyde (Fisher Scientific, Reagent Grade)
tal surface area and the copper surface area) and the num-
and propionaldehyde (Aldrich, 97%) were carried out
ber of basic sites from CO2 exchange were used in order
using the following reactant mixture: 12C2H4O(12C3H6O)/
to normalize reaction rates. The choice among these val-
13CH3OH/CH4/He = 7.3/4.0(4.0)/2.7/87.3 kPa. 13CH3OH
ues depends on which sites or surfaces are required for
was purified by several freeze-pump-thaw cycles before
the rate-determining elementary step in the overall cata-
catalytic experiments. Reactions were conducted at 573 K
lytic sequence.
and 101.3 kPa in a recirculating reactor unit. Mass spec-
trometry analysis after chromatographic separation was
2.3.2. Reactions of acetaldehyde. Condensation and hy-
drogenation reactions of acetaldehyde were examined on
used to confirm the identity of each reaction product and
–
–
0.8 wt% K Mg5CeOx, 1.0 wt% K Cu0.5Mg5CeOx, and
to determine 13C content, position, and distributions in
reactants and products.
–
1.2 wt% K Cu7.5Mg5CeOx. The catalyst (33.0 mg) was first
reduced in H2 (10% H2/He) at 623 K for 0.5 h. The reac-
tion was carried out at 573 K. The reactant mixture was:
C2H4O/CH4/He = 6.6/2.7/92.0 kPa (C2H4O, Fisher Scien-
tific, Reagent Grade; CH4, Matheson, ultra high purity).
Reaction products were analyzed by the procedures de-
scribed in the previous section.
3. RESULTS AND DISCUSSION
3.1. Structural and Textural Properties
–
–
–
of K Cu Mg CeOx Catalysts
Thecomposition, surfacearea, Cudispersion, anddensity
2.3.3. Reactions of C2H5OH/13C2H4O mixtures. Com- of basic sites for mixed metal oxides are shown in Table 1.
petitive reactions between 12C2H5OH and 13C2H4O were Copper dispersion decreased with increasing K loading
–
–
examined on 0.8 wt% K Mg5CeOx, 1.0 wt% K Cu0.5 (Table 1), because the presence of alkali decreases
–
Mg5CeOx, and 1.2 wt% K Cu7.5Mg5CeOx catalysts in or- Mg5CeOx surface area, blocks surface Cu atoms, and in-
der to probe direct pathways from ethanol to condensation hibits Cu2+ reduction to Cu metal (13). The addition of K
products and the role of acetaldehyde intermediates. The also decreased the total surface area of Cu0.5Mg5CeOx sam-
catalyst (22.0 mg) was reduced in 10% H2 (balance He) ples (Table 1). The copper dispersion in Cu0.5Mg5CeOx is
at 623 K for 0.5 h. Reactions were conducted at 523 K 0.23 (Table 1); this value is among the highest reported
and 101.3 kPa in the recirculating reactor unit. The reac- in the literature for Cu-based methanol and higher al-
tant mixture was introduced along with a small amount of cohol synthesis catalysts. The density of basic sites ob-
methane (reaction mixture: 13C2H4O/12C2H5OH/H2/CH4/ tained by isotopic CO2 exchange methods on MgO at
He = 2.9/1.1/2.9/2.4/92.0 kPa; 13C2H4O: Isotec Inc., 13C: 573 K is 0.38 ꢂ 10ꢁ6 mol/m2; this is lower than the value
99%; C2H5OH: Fisher Scientific, A.S.C. Certified; CH4: of 18.3 ꢂ 10ꢁ6 mol/m2 reported for the surface density of
Matheson, ultra high purity). 13C content and isotopic dis- oxygen atoms in MgO (40). This is not unexpected because
tributions in reactants and products were analyzed by mass the 13CO2/12CO2 isotopic switch method detects only those

BIFUNCTIONAL CONDENSATION REACTIONS
159
basic sites that interact reversibly with CO2 at 573 K in typ- n-butyraldehyde, and other oxygenates. Acetaldehyde
ical catalytic turnover times (13, 38). The basic site density reaches a maximum concentration at intermediate contact
increased from 1.2 to 2.3 ꢂ 10ꢁ6 mol/m2 with increasing K times during ethanol reactions and then decreases grad-
loadingfrom0.1to1.0wt%. SmallamountsofCeOx inMgO ually. The acetaldehyde concentration at this point corre-
(Mg/Ce = 5at.)increasedthedensityofbasicsites(Table1). sponds to the acetaldehyde concentration calculated from
Mg(OH)2 and Ce(OH)4/CeO2 phases were identified by the thermodynamic equilibrium of the chemical reaction:
X-ray diffraction on dried precipitate samples. The poor
C2H5OH ꢀ C2H4O + H2.
sample crystallinity and the coincidence of diffraction lines
for Ce oxide and hydroxide phases make it difficult to de-
termine conclusively the form of Ce in dried samples. No
layered hydrotalcite-type structures were detected by X-
ray diffraction. Samples calcined at 723 K exhibited sep-
arate MgO and CeO2 phases. Diffraction lines provided
no evidence for MgCeOx mixed oxides. CuO crystallites
were not detected, suggesting that the Cu component is
well dispersed, either as small crystallites or as a solid so-
lution within the CeO2 lattice. Mixed Cu/Ce oxide solid so-
lutions can form during co-precipitation and remain after
high temperature oxidative treatments (41, 42).
The small decrease in acetaldehyde concentration at long
contact times could be caused by further reaction of
acetaldehyde to secondary products and/or by the reversal
of the ethanol-acetaldehyde equilibrium as ethanol is con-
verted to chain growth products. Acetone and n-butyral-
dehyde were the predominant condensation products. The
non-zero initial slopes on acetone and n-butyraldehyde
curves (Fig. 1) are inconsistent with their formation
only in sequential reactions of gas phase acetaldehyde
intermediates such as
C2H5OH ꢀ C2H4O → Condensation products.
3.2. Ethanol Reactions
Therefore, condensation products are either formed direct-
ly by ethanol self-condensationorfrom acetaldehyde, but at
reaction rates independent of acetaldehyde concentration.
Methyl-ethyl ketone (by acetaldehyde self-condensation
with intramolecular hydride shift before dehydration)
(7, 8), 2-pentanone (by acetaldehyde-acetone condensa-
tion), 2-propanol (by hydrogenation of acetone), 1-butanol
(by hydrogenation of n-butyraldehyde), and ethyl acetate
(by ethanol-acetaldehyde) were also detected in much
smaller concentrations than n-butyraldehyde and acetone
among reaction products.
Ethanol dehydrogenation and coupling reactions were
studied in order to probe reaction pathways involved
in branched and linear higher alcohol syntheses from
methanol, ethanol, and propanol intermediates formed ini-
tially during CO hydrogenation reactions at high pres-
sures. These studies also address the roles of Cu, K, and
–
Mg Ce oxides in ethanol coupling reactions. The experi-
–
ments were carried out on Mg5CeOx, 0.8 wt% K Mg5CeOx,
–
Cu0.5Mg5CeOx, 1.0 wt% K Cu0.5Mg5CeOx, and 1.2 wt%
–
K Cu7.5Mg5CeOx at atmospheric pressure.
Acetaldehyde was the predominant product of reactions
The formation of all detected products of ethanol
–
of pure ethanol on 1.0 wt% K Cu0.5Mg5CeOx (Fig. 1).
–
–
reactions on K Cu Mg5CeOx catalysts can be described
by the reaction pathways shown in Scheme 4. Acetalde-
hyde is formed via ethanol dehydrogenation (step I).
Self-condensation of acetaldehyde (or surface acetalde-
hyde formed from dissociative adsorption of ethanol and
sequential ꢀ-hydrogen abstraction) leads to the forma-
tion of n-butyraldehyde (steps II–IV) and methyl-ethyl
ketone (steps II–V) (34). Enolate species can convert to
the keto form via intramolecular H transfer (step VI);
methyl-ethyl ketone can then be formed by dehydration-
dehydrogenation reactions of this keto form (step VII).
Acetone can form via two pathways: (i) by reaction of
aldol intermediates with surface lattice oxygen to form the
ꢁ-keto acid followed by decarboxylation (steps VIII–IX)
and/or (ii) by reverse aldol reactions of the keto form of re-
action intermediates (step X). Formaldehyde formed in the
latter reaction decomposes rapidly to CO and H2. Surface
lattice oxygen consumed by reaction (IX) can be replen-
ished by water and CO2 formed in condensation reactions;
2-pentanone is formed by condensation of aldol-type
intermediates formed in acetone-acetaldehyde reactions
(steps XI–XII). Diethyl ether was only detected in trace
Dehydrogenation reactions occur much faster than chain
growth;thelatterreactionleadstotheformationofacetone,
FIG. 1. Site yields as a function of contact time on 1.0 wt%
–
K Cu0.5Mg5CeOx in ethanol reactions (573 K, 101.3 kPa total pressure,
4.0 kPa ethanol, balance He).

160
GINES AND IGLESIA
–
–
SCHEME 4. Reaction scheme for ethanol reactions on K Cu Mg5CeOx catalysts.
amounts on Mg5CeOx catalysts, and it is formed on CeOx and the copper surface area), the number of basic sites from
acid sites by dehydration of ethanol. Ethyl acetate can form CO2 exchange, or the Cu metal surface area of each sample
by reactions of ethanol with acetaldehyde (step XIII) (34). in order to normalize reaction rates.
Ethanol reaction rates and the selectivities to acetalde-
Initial dehydrogenation rates were very low on both
–
hyde, n-butyraldehyde, and acetone were measured on Mg5CeOx and 0.8 wt% K Mg5CeOx catalysts and ethanol
–
–
K Cu MgCeOx catalysts with a range of Cu and K con- conversion reached a limiting value of about 9% after
tents in order to examine the role of Cu and basic sites on 0.33 h; this conversion value is much lower than the
dehydrogenation and condensation reactions. The results predicted equilibrium conversion (90%) at these reaction
are shown in Table 2. Initial reaction rates were obtained conditions. The decrease in ethanol conversion rates with
from the slope of the site-yield plots at short contact times increasing conversion may reflect (1) the formation of
using the total surface area, the MgCeOx surface area (es- surface polymeric species from acetaldehyde and/or (2) an
timated from the difference between the total surface area inhibition effect of CO2 formed in ethanol condensation on
TABLE 2
Effects of Cu- and K-loading on Ethanol Consumption and Product Formation on Mg5CeOx
Rates of formation
Ethanol dehydrogenation
Areal rate Turnover rate
Acetone + 2-Propanol
Areal rate Turnover rate
Butyraldehyde + 1-Butanol
Cu
(wt%)
K
(wt%)
Areal rate
Turnover rate
a
b
c
d
c
d
r1
r1
r2
r2
r3
r3
0
0
7
7
49
0
4.0 ꢂ 10ꢁ9
3.4 ꢂ 10ꢁ9
3.6 ꢂ 10ꢁ7
2.4 ꢂ 10ꢁ7
9.4 ꢂ 10ꢁ7
—
—
0.24
0.23
0.24
7.5 ꢂ 10ꢁ11
9.7 ꢂ 10ꢁ11
3.0 ꢂ 10ꢁ9
4.3 ꢂ 10ꢁ9
4.4 ꢂ 10ꢁ8
7.9 ꢂ 10ꢁ5
5.6 ꢂ 10ꢁ5
2.2 ꢂ 10ꢁ3
1.8 ꢂ 10ꢁ3
1.0 ꢂ 10ꢁ2
2.8 ꢂ 10ꢁ11
1.6 ꢂ 10ꢁ10
4.5 ꢂ 10ꢁ10
7.6 ꢂ 10ꢁ10
1.2 ꢂ 10ꢁ9
2.9 ꢂ 10ꢁ5
9.4 ꢂ 10ꢁ5
3.4 ꢂ 10ꢁ4
3.2 ꢂ 10ꢁ4
2.8 ꢂ 10ꢁ4
0.8
0.1
1.0
1.2
a r1 is the rate of ethanol consumption, in mol/m2 total ꢃ s.
b Turnover rates per Cu surface atom, in sꢁ1
.
c r2 and r3 are the rates of product formation, in mol/m2 MgCeOx ꢃ s.
d Turnover rates per accessible site from 13CO2/12CO2 in mol C2H5OH/mol basic site ꢃ s; basic sites measured by 13CO2/12CO2 isotopic switch method
(13, 38).

BIFUNCTIONAL CONDENSATION REACTIONS
161
base-catalyzed dehydrogenation reactions on Mg5CeOx. Dehydrogenation rates were much higher on Cu0.5Mg5CeOx,
–
–
1.0 wt% K Cu0.5Mg5CeOx, and 1.2 wt% K Cu7.5Mg5CeOx catalysts than on Mg5CeOx, suggesting that Cu sites catalyze
ethanol dehydrogenation to acetaldehyde. The presence of K decreased areal ethanol dehydrogenation rates, because
K decreased Cu dispersion. Ethanol dehydrogenation turnover rates (normalized by the number of exposed Cu atoms)
on Cu0.5Mg5CeOx, 1.0 wt% K Cu0.5Mg5CeOx, and 1.2 wt% K Cu7.5Mg5CeOx catalysts were 0.24, 0.23, and 0.24 sꢁ1
,
–
–
respectively. Thus, ethanol dehydrogenation steps require exposed Cu atoms for the rate-determining step. Surface Cu
atoms become unavailable for both N2O decomposition and alcohol dehydrogenation when covered with K species or as
Cu crystallites become larger. Dehydrogenation turnover rates were not affected by Cu crystallite size or by the titration
of a fraction of the Cu surface with KOx species.
Aldol coupling chain growth rates are lower on Mg5CeOx than on Cu0.5Mg5CeOx because of the lower acetaldehyde
concentrations when Cu sites are not present. K increases the rate of base-catalyzed aldol coupling reactions to acetone
and butyraldehyde (Table 2), even though steady-state acetaldehyde concentrations on Cu0.5Mg5CeOx and 1.0 wt%
–
K Cu0.5Mg5CeOx catalystsaresimilaronthesetwocatalysts. Thus, itappearsthatthehigherbasicsitedensitiesdetermined
by 13CO2/12CO2 isotopic switch measurements indeed lead to higher rates of base-catalyzed aldol condensation reactions.
The total rates of base-catalyzed aldol coupling reactions, when normalized by the number of accessible basic sites,
are similar on Cu0.5Mg5CeOx (2.5 ꢂ 10ꢁ3) and K Cu0.5Mg5CeOx (2.2 ꢂ 10ꢁ3). Aldol condensation rates on 1.2 wt%
–
–
K Cu7.5Mg5CeOx (49 wt% Cu, 0.047 Cu dispersion) are, however, much higher than on catalysts with lower Cu content,
even though the density of basic sites is similar on these two samples. The difference in condensation rates suggests that
Cu sites are involved in rate-determining steps required for condensation reactions, even though such Cu sites are present
in sufficient number to ensure ethanol-acetaldehyde thermodynamic equilibrium. The lower aldol coupling chain growth
reaction rates observed on Cu-free catalysts, which is caused not only by the low concentration of required acetaldehyde
intermediates but also by the absence of Cu sites required in condensation steps, is consistent with this proposal.
A sequential reaction scheme, where ethanol dehydrogenates to form gas phase acetaldehyde, which then undergoes
self-condensation reactions into products, is not considered, because it is not consistent with the sharp initial increase in
product site-yield curves (Fig. 1). On Cu containing catalysts, the steady-state acetaldehyde concentrations are similar.
The effect of copper suggests a bifunctional mechanism for aldol condensation on metal-promoted basic oxides; this
mechanistic proposal is discussed in detail below.
The proposed mechanism involves the initial dissociative adsorption of ethanol on MgO to form ethoxide and hydrogen
species (43, 44). Hydrogen species can then be removed by migration to Cu sites, recombination with another hydrogen
adatom, and desorption as H2:
–
Additional C H bond cleavage events in ethoxide species can then occur and the hydrogen atoms formed are transferred
to oxygen ions and form surface acetaldehydic species:
In this scheme, hydrogen atoms migrate from basic to Cu sites and recombine to form H2. As a result, basic oxygen
–
sites become available for another hydrogen abstraction step. H H recombination rates increase with increasing ratio
–
–
of surface Cu atoms to basic sites, which is 0.31 on K Cu7.5Mg5CeOx and 0.07 on K Cu0.5Mg5CeOx, at similar basic

162
GINES AND IGLESIA
–
site density. The high ratio of Cu to basic oxide surface area on K Cu7.5Mg5CeOx leads to higher aldol-condensation rates
than on catalysts with lower ratios, without a detectable effect on the (equilibrium) concentration of acetaldehyde.
Adsorbed acetaldehyde species can desorb as acetaldehyde or react with other surface species to form aldol conden-
sation products:
Surface aldol species can undergo condensation reactions via the pathways shown in Scheme 4. In these reactions,
the presence of Cu sites increases the rate of hydrogen removal and reintroduction and provides the hydrogen atoms
required for hydrogenation of unsaturated aldehyde species (steps IV, VII, and XII, Scheme 4). The non-zero initial
slopes of acetone and n-butyraldehyde turnover-time plots (Fig. 1) are consistent with the direct condensation of ethanol
reactants. These bifunctional condensation pathways are qualitatively similar in kinetic structure to those proposed for
–
bifunctional C H activation of alkanes on cation-modified zeolites (45–47).
The ratio of C3 to C4 oxygenates measured on the different catalysts gives the selectivity of the decomposition of
keto species to form acetone (Scheme 4, step X) compared to aldol dehydration–hydrogenation of such species to
form n-butyraldehyde (Scheme 4, steps III and IV). On Cu-free catalysts this ratio is <3, and the total condensation
rate is ꢄ10ꢁ10 mol/m2-s. At high Cu concentrations, the C3/C4 selectivity ratio increases to 37, while the total conden-
sation rate increases to ꢄ10ꢁ8 mol/m2-s. Thus, in addition to increasing the surface concentration of aldol species via
removal of adsorbed hydrogen atoms, Cu appears to enhance hydrogen transfer steps required for enol-keto intercon-
version (Scheme 4, step VI); as a result, Cu sites increase the concentration of keto species and the rate of step X
(Scheme 4).
3.3. Reactions of Acetaldehyde
In Section 3.2, we proposed that condensation reactions could proceed (i) via direct reactions of ethanol without the
intermediate formation of gas phase acetaldehyde molecules, or (ii) through consecutive reactions including the formation
of gas-phase acetaldehyde, with zero order kinetics in acetaldehyde concentration. In order to test the second possibility,
–
catalytic activity and product yields for acetaldehyde reactions on 1.0 wt% K Cu0.5Mg5CeOx were measured at several
acetaldehyde pressures.
Acetone and crotonaldehyde are the predominant products of acetaldehyde reactions (Fig. 2). Carbon dioxide,
n-butyraldehyde, and 2-pentanone were also detected in small concentrations among reaction products. Ethanol
–
forms by hydrogenation of acetaldehyde using H-adatoms formed in C H activation steps or H2 molecules des-
–
orbed by Cu sites after C H activation. Aldol-type condensation of acetaldehyde forms crotonaldehyde, which then
hydrogenates to form n-butyraldehyde. Acetaldehyde can also polymerize to give dimers and trimers (Scheme 5)
and/or by sequential aldol condensations leading to high molecular weight aldehydes (48). These polymeric species are
likely to remain on the catalyst surface and may account for the poor overall mass balances observed in these experiments

BIFUNCTIONAL CONDENSATION REACTIONS
163
FIG. 3. Acetaldehyde reactions. Effect of the acetaldehyde initial par-
FIG. 2. Acetaldehyde reactions. Site yields as a function of contact
–
tial pressure over the initial rates. Acetaldehyde conversion as a function
time on 1.0 wt% K Cu0.5Mg5CeOx (573 K, 101.3 kPa total pressure, 8.0 kPa
–
of contact time on 1.0 wt% K Cu0.5Mg5CeOx (573 K, 101.3 kPa total pres-
acetaldehyde pressure, balance He).
sure, 3–15 kPa acetaldehyde pressures, balance He).
(carbon selectivity totals less than 45%). Hydrogen is also
produced by decomposition of the formaldehyde formed in
retro-aldol reactions leading to acetone (Scheme 4).
Initial reaction rates at different initial partial pressures
of acetaldehyde are shown in Fig. 3. Condensation, hydro-
genation, and total conversion rates are approximately first
order in acetaldehyde concentration. These results show
that the initial reaction nonzero slope for condensation
products in ethanol reactions (Fig. 1) cannot reflect zero-
order kinetics for aldehyde condensation. Instead, it sug-
gests that direct reactions of ethanol account for a signif-
icant fraction of the condensation products formed from
ethanol reactants.
less pronounced than for reactions of ethanol. These re-
sults show that Cu sites also increase rates of C H bond
activation in acetaldehyde.
–
Acetone is a condensation product, while ethanol
is a hydrogenation product. The molar ratio between
ethanol/acetone is a measurement of the catalyst ability
to remove H-atoms from the surface by recombinative
desorption compared to its ability to catalyze condensation
reactions. On Mg5CeOx catalysts, ethanol is produced
selectively over acetone. The catalyst surface is saturated
with H-atoms from the abstraction of ꢀ-hydrogens from
acetaldehyde on basic sites. Without Cu, it is more probable
Figure 4 shows the molar ratio of ethanol to acetone and
of C4 oxygenates (crotonaldehyde and n-butyraldehyde) to
acetoneamongthereactionproductsandtheinitialconden-
–
sation rates for acetaldehyde reactions on K CuyMg5CeOx
catalysts. Aldol coupling chain growth rates increase with
increasing copper surface area. The increase, however, is
FIG. 4. C2 to C3 and C4 to C3 molar ratios on acetaldehyde reac-
tions and initial acetaldehyde condensation rates as a function of Cu sur-
–
SCHEME 5. Formation of dimer and trimer compounds from acetalde-
face area on K CuyMg5CeOx catalysts (573 K, 101.3 kPa total pressure,
hyde.
6.6 kPa acetaldehyde, balance He).

164
GINES AND IGLESIA
13
13
13
13
–
FIG. 5.
C-content in reactants and reaction products ( C2H4O
–
C-content in reactants and reaction products ( C2H4O
FIG. 7.
–
–
C2H5OH H2 reactionmixtureson0.8wt%K Mg5CeOx)(523K, 101.3kPa
total pressure, 2.9 kPa acetaldehyde pressure, 1.1 kPa ethanol pressure,
2.9 kPa hydrogen pressure, balance He).
–
–
C2H5OH H2 reaction mixtures on 1.2 wt% K Cu7.5Mg5CeOx) (523 K,
101.3 kPa total pressure, 2.9 kPa acetaldehyde pressure, 1.1 kPa ethanol
pressure, 2.9 kPa hydrogen pressure, balance He).
3.4. Reactions of C2H5OH/13C2H4O Mixtures
that the H-atoms are used to hydrogenate acetaldehyde
to ethanol. When Cu sites are present, ethanol selectivity
decreases markedly because these H-atoms recombine to
form H2. On the other hand, acetone is formed selectivity
on Cu-containing catalysts instead of n-butyraldehyde or
crotonaldehyde. This shows that reverse aldol reaction
steps from C4 intermediates (Scheme 4) become less likely
as Cu is introduced into the catalysts.
As discussed in Section 3.2, it appears that ethanol
undergoes condensation reactions directly, without re-
quiring its intermediate conversion to gas phase acetal-
dehyde. Also, condensation rates increase with increas-
ing acetaldehyde concentration. Chung et al. (49), using
CH313CH2OH/C2O4H mixtures on Cu/ZnO catalysts,
–
showed that acetone can be produced by C C bond cleav-
age in C4 species formed from two aldehydic species and
–
n-butyraldehyde by C O bond cleavage in C4 species
formed from ethanol and acetaldehyde. In order to ex-
amine the role of ethanol and to detect available di-
rect pathways for ethanol condensation, competitive reac-
tions between 12C2H5OH and 13C2H4O were carried out
–
–
on 0.8 wt% K Mg5CeOx, 1.0 wt% K Cu0.5Mg5CeOx, and
–
1.2 wt% K Cu7.5Mg5CeOx catalysts. The carbon-13 con-
tents in reactants and products as a function of contact time
are shown in Figs. 5, 6, and 7.
On all catalysts, the 13C-content in both ethanol and ac-
etaldehyde changes with contact time, as the products reach
chemical and isotopic equilibrium. The rate of ethanol-
acetaldehyde equilibration is much higher on 1.2 wt%
–
K Cu7.5Mg5CeOx than on the other two catalysts. When
copper is present, hydrogenation–dehydrogenation reac-
tions occur more rapidly, in agreement with the results
shown in Section 3.2. On Cu-free catalysts (Fig. 5), the
13C-content in both ethanol (30%) and acetaldehyde (80%)
reach values that differ markedly from those predicted at
isotopic equilibrium (70%). This is consistent with the rapid
deactivation observed during ethanol reactions on Cu-free
catalysts (Sections 3.2, 3.3).
13
13
–
C-content in reactants and reaction products ( C2H4O
FIG. 6.
–
–
C2H5OH H2 reaction mixtures on 1.0 wt% K Cu0.5Mg5CeOx) (523 K,
101.3 kPa total pressure, 2.9 kPa acetaldehyde pressure, 1.1 kPa ethanol
pressure, 2.9 kPa hydrogen pressure, balance He).

BIFUNCTIONAL CONDENSATION REACTIONS
165
FIG. 8. 13C-distribution of products in 13C2H4O C2H5OH H2 reactions on 1.0 wt% K Cu0.5Mg5CeOx as a function of contact time (523 K, 101.3 kPa
–
–
–
total pressure, 2.9 kPa acetaldehyde pressure, 1.1 kPa ethanol pressure, 2.9 kPa hydrogen, balance He).
13
The 13C-distributions in the products of C2H4O
zero conversion. Table 3 also shows the predicted binomial
–
C2H5OH H2 mixtures on 1.0 wt% Cu0.5Mg5CeOx are 13C distributions when n-butyraldehyde forms only from
shown in Fig. 8. The 13C-content in condensation products ethanol, only from acetaldehyde, or from both ethanol and
changes slightly with contact time, but the isotopic con- acetaldehyde with equal probability. These data (Fig. 8,
tent in ethanol and acetaldehyde (Figs. 6 and 8) change Table 3) suggest that n-butyraldehyde and acetone are
significantly. These data may reflect (i) the rapid forma- formed from both reactants (ethanol and acetaldehyde).
tion of surface polymer species from the starting mixture In contrast, the 13C-content in crotonaldehyde formed
–
–
and the subsequent formation of all products only from on 0.8 wt% K Mg5CeOx at relatively low contact times
this surface polymer pool or (ii) a similar reactivity of (Table 3) shows that it is produced predominantly from
ethanol and acetaldehyde in condensation products, lead- acetaldehyde.
ing to products that contain the (constant) average 13C-
Ethanol adsorbs dissociatively on MgO to form surface
content within the remaining reactant pool. The first pos- ethoxide and hydrogen species (43, 44). In the absence of
–
sibility is unlikely because the decomposition of surface the Cu sites that provide a H H recombination function,
polymer species would lead to C1, C2, C3, etc. species and surface ethoxide and hydrogen recombine preferably and
not only n-butyraldehyde and acetone. The second ex- desorb as ethanol, instead of remaining as adsorbed ethox-
planation seems more plausible. On Cu-containing cata- ide species that then lose a ꢁ-H to give surface aldehydic
lysts, hydrogenation–dehydrogenation steps (ethanol =
acetaldehyde + H2) are fast and the 13C-content in acetone
and n-butyraldehyde extrapolated to zero contact time is
about 80%. If these products formed only from acetalde-
hyde, they should contain only 13C, and if they formed only
from ethanol, they should be unlabeled. This 80% value
shows that ethanol and acetaldehyde contribute to products
in proportion to their concentration in the reactant pool.
The 13C-composition of the reactant mixture is 73%, which
is less than the 80% observed for 13C-content in products.
The 13C-contents in products decreases slightly with con-
tact time, similar to the 13C-content decrease observed in
acetaldehyde (Figs. 6 and 8). These findings suggest that
product formation rates from acetaldehyde are slightly
higher than from ethanol, but that direct condensa-
tion of ethanol occurs readily at these reaction condi-
tions.
TABLE 3
13C-distribution of Prodcuts (mole fraction) in 13C2H4O C2H5OH
–
–
–
H2 reactions on 1.0 wt% Cu0.5Mg5CeOx and 0.8 wt% K Mg5CeOx
Predicteda
n-Butyraldehydeb Crotonaldehydeb
Number
Acetal-
1.0 wt%
0.8 wt%
of 13
C
Ethanol dehyde Both Cu0.5Mg5CeOx
K Mg5CeOx
–
0
1
2
3
4
1.00
0
0
0
0
0.08
0
0.39
0
0.07
0
0.30
0
0.02
0
0.12
0
0
0
0
0
1.00
0.53
0.63
0.86
Note. 523 K, 101.3 kPa total pressure, 2.9 kPa acetaldehyde pressure,
1.1 kPa ethanol pressure, 2.9 kPa hydrogen, balance He.
a Calculated by assuming the n-butyraldehyde or crotonaldehyde are
formed exclusively from ethanol, or from acetaldehyde or from both with
identical probability.
Table 3 shows the 13C-content in n-butyraldehyde formed
–
on 1.0 wt% K Cu0.5Mg5CeOx and in crotonaldehyde form-
ed on 0.8 wt% K Mg5CeOx, obtained by extrapolating to
b Experimental values extrapolated to zero conversion.
–

166
GINES AND IGLESIA
SCHEME 6. Competitive reactions of 13C2H4O C2H5OH H2 mixtures.
–
–
SCHEME 7. Ethanol reactions and hydrogen removal steps.
species and condense with other surface species. Some H-
atoms are used to hydrogenate adsorbed aldehydic species,
which leave the surface as labeled ethanol (Scheme 6). In
this situation, it is more likely that two adsorbed acetalde-
hyde molecules react with each other leading to labeled
containing a statistical mixture of D-atoms from D2 and
H-atoms from ethanol. The deuterium content in reaction
products, therefore, measures the rate of communication
between the surface and gas phase hydrogen pools. In con-
trast, irreversible recombinative desorption steps would
lead to reaction products containing predominantly pro-
tium, because the slow communication between gas phase
(D2) and surface hydrogen pool leads to predominant pres-
aldol intermediates.
12
These isotopic tracer studies of C2H5OH 13C2H4O mix-
–
tures suggest that condensation reaction can proceed via
two parallel pathways: direct reactions of ethanol with-
out the intermediate formation of gas phase acetaldehyde
molecules and sequential dehydrogenation to form ac-
etaldehyde followed by condensation steps (Scheme 6).
These results show that “dehydrogenation” and conden-
sation steps of ethanol can occur on the same type of active
site, an acid-base site pair on MgO. This proposal is consis-
tent with the promoting effect of Cu on condensation rates,
with the initial non-zero slope of product concentration
–
ence of protium atoms formed in C H activation steps on
the catalyst surface.
The role of Cu sites in ethanol reactions was exam-
ined by measuring the initial rates of D-atom incorpora-
tion into unreacted ethanol and into reaction products.
These data are shown as a function of contact time in
–
–
Fig. 9 on 0.8 wt% K Mg5CeOx, 1.0 wt% K Cu0.5Mg5CeOx,
curves observed in ethanol reactions, and with the 13C con-
–
and 1.2 wt% K Cu7.5Mg5CeOx catalysts. The initial rates
12
tent in the condensation products of C2H5OH 13C2H4O
–
of incorporation of deuterium into ethanol, acetaldehyde,
and acetone as a function of Cu surface area were calcu-
lated from the initial slopes in Fig. 9 and the results for
mixtures.
–
K CuyMg5CeOx catalysts are summarized in Table 4. The
3.5. Reactions of C2H5OH/D2 Mixtures
distribution of deuterium atoms in unreacted ethanol, ac-
etaldehyde, and acetone products of ethanol/D2 mixtures
at different contact times are shown in Fig. 10.
The rate of appearance of deuterium in unreacted
ethanol and in reaction products (Fig. 9 and Table 4)
increases with increasing Cu content on Mg5CeOx cata-
lysts. The deuterium fraction is much lower on Mg5CeOx
–
K
As mentioned above, ethanol reactions on
CuyMg5CeOx lead to acetaldehyde, acetone, and n-butyral-
dehyde as predominant products. Acetaldehyde and
ethanol condensation reactions occur on basic sites, but
condensation rates increase with increasing density of
Cu surface atoms (Table 2). Hydrogen abstraction steps
(involved in condensation reactions) appear to be limited
by the rate of hydrogen removal via recombinative desorp-
tion. This desorption process occurs rapidly on Cu surface
atoms, but it is slow on metal-free MgO or MgCeOx.
Hydrogen migration and recombinative desorption on
Cu sites can then increase aldol condensation rates by
removing a surface bottleneck that tends to reverse the
steps that form unsaturated aldol-type species required for
chain growth.
In order to probe the reversibility of hydrogen
adsorption-desorption pathways (Scheme 7), ethanol re-
actions were carried out in the presence of gas phase D2.
The recombinative hydrogen disposal steps can be quasi-
equilibrated or irreversible steps during ethanol coupling
reactions. Quasi-equilibrated desorption steps would re-
sult in the formation of reaction products and reactants
TABLE 4
Reactions of C2H5OH/D2 Mixtures
Cu surface
Ethanol
Acetaldehyde Acetone
area
Catalyst
(m2/gcat
)
rD
rtb
rD
rtb
rD
rtb
a
a
a
Mg5CeOx
Cu0.5Mg5CeOx
Cu7.5Mg3CeOx
0
10.4
21.9
4.5 78.2
129 514 200
925 683 601
1.4
57.9 0.02 0.38
469 6.7 5.6
585 78 27
Note. Effect of Cu Surface Area on the Initial Rates of Incorporation
of Deuterium into Unreacted Ethanol and Products.
a rD is the rate of deuterium incorporation in unreacted ethanol and
ꢂ107, in mol D2/g-cat ꢃ s.
b rt is the rate of ethanol consumption or product formation ꢂ107,
in mol/g-cat ꢃ s.

BIFUNCTIONAL CONDENSATION REACTIONS
167
FIG. 9. Deuterium fraction in unreacted ethanol (a) and in acetaldehyde (b), acetone (c), and n-butyraldehyde (d) formed from ethanol/D2
–
mixtures (ꢄ20% H, ꢄ80% D) on K CuyMg5CeOx catalysts (573 K, 101.3 kPa total pressure, 3.6 kPa ethanol pressure, 27 kPa D2 pressure, balance He).
than on Cu-containing catalysts, suggesting that Cu sites ucts increases with increasing contact time. All deuterium
–
increase the rate of communication between surface and distributions are binomial, suggesting that every C H (and
–
gas phase hydrogen pools by providing a site for dissocia- O H) bond is broken and reformed several times during
tive D2 chemisorption and for recombinative desorption of the formation of one product molecule (while the precur-
–
H-atoms formed in C H bond activation steps occurring on sor molecule remains adsorbed on the surface) or that a
basic sites. desorbed molecule returns to the surface and exchanges
–
At short contact times (Fig. 10), gas phase molecules (and sequentially with the surface H D pool. The latter process
adsorbed species in equilibrium with them) have relatively leads to the observed increase in deuterium content with
low deuterium contents. Singly-deuterated species are the contact time.
most abundant, suggesting that gas phase D2 contributes
Initial rates of ethanol dehydrogenation and of deu-
weakly to the surface hydrogen pool. This surface hydro- terium incorporation into ethanol (in ethanol/D2 mixtures)
–
–
gen pool contains predominantly H-atoms formed by C H on K CuyMg5CeOx catalysts are shown in Fig. 11 as a
bond activation steps required in ethanol dehydrogenation function of Cu surface area. The initial rate of ethanol
reactions. The deuterium content in reactants and prod- dehydrogenation increases linearly with Cu surface area

168
GINES AND IGLESIA
–
FIG. 10. Reactions of C2H5OH/D2 mixtures on 1.0 wt% K Cu0.5Mg5CeOx catalyst. Deuterium distributions in unreacted ethanol and acetaldehyde
and acetone products at various contact times (0.3 ks, 1.2 ks, 6.0 ks) (573 K, 101.3 kPa total pressure, 3.6 kPa ethanol pressure, 27 kPa D2 pressure,
balance He).
(measured by N2O decomposition methods); therefore, de- for the rate-determining step in condensation reactions. If
hydrogenation turnover rates are not affected by Cu crys- hydrogen removal is rate-determining in ethanol dehydro-
tallite size and involve a rate-determining step occurring genation and condensation reactions, Cu may provide the
on Cu surface atoms. At similar steady-state acetaldehyde required site for the recombinative desorption of H-atoms
–
concentrations, condensation rates to form acetone and formed in C H bond activation reactions on MgO acid-base
n-butyraldehyde also increase with increasing Cu surface sites. The removal of H species from basic oxygen atoms on
area (Fig. 11), even though basic site densities measured by MgO surfaces frees such sites for the required activation of
–
CO2 isotopic switch methods are very similar on these cata- thenextC H bond. Theincreaseintherateofdeuteriumin-
lysts (Table 1). Therefore, Cu metal sites are also required corporation into unreacted ethanol with increasing amount

BIFUNCTIONAL CONDENSATION REACTIONS
169
FIG. 12. Methanol-acetaldehyde reactions. Site yields as a function of
–
contact time on 1.0 wt% K Cu0.5Mg5CeOx (573 K, 101.3 kPa total pressure,
8.0 kPa methanol, 4.0 kPa acetaldehyde, balance He).
FIG. 11. Initial rates of ethanol dehydrogenation and condensation
and of incorporation of deuterium into ethanol (in ethanol/D2 mixtures)
–
as a function of Cu surface area on K CuyMg5CeOx catalysts (573 K,
2-propanol and 1-butanol, after subsequent hydrogenation
during CO/H2 reactions. 2-Propanol and 1-butanol cannot,
however, form isobutanol precursors (such as isobutyralde-
hyde and propionaldehyde) by condensation reactions dur-
ing CO hydrogenation. 13C-tracer studies of methanol re-
action with acetaldehyde or propionaldehyde were carried
out in order to examine reaction pathways that lead to the
formation of the C3 and C4 oxygenate precursors required
for isobutanol formation via aldol-type reactions of alco-
101.3 kPa total pressure, 3.7 kPa ethanol (27 kPa D2), balance He).
of Cu surface atoms is consistent with faster rates of H2
adsorption–desorption steps when Cu is present. When
H-atoms are not rapidly removed, they tend to reverse the
–
C H bond activation step that forms the unsaturated aldol-
type species required for chain growth. As more Cu sites are
introduced, hydrogen coverages decrease and the surface
favors ꢀ-H abstraction surface steps.
hols and aldehydes with methanol-derived C1 species.
13
Cross-coupling site yields for CH3OH 12C2H4O mix-
–
3.6. Cross-Coupling Reactions of 13CH3OH/C2H4O
and 13CH3OH/C3H6O Mixtures
tures on Cu0.5Mg5CeOx are shown in Fig. 12. The initial
rates (site-time yields per total 9-atom Cu) of methanol
conversion were higher than that for the conversion of ac-
etaldehyde to ethanol and to condensation products. The
13CH3OH component in the feed reacted predominantly
to form 13CO and H2; this reaction reverses methanol
synthesis because of the unfavorable methanol synthesis
thermodynamics at atmospheric pressures (but not during
Ethanol is a useful and simple probe molecule to test the
role of individual components in metal–base bifunctional
catalysts for isobutanol synthesis. Ethanol reactions, how-
ever, lead to acetone and n-butyraldehyde, which form only
TABLE 5
13C-Distribution in Products of 13CH3OH C2H4O Reactions
–
–
on 1.0 wt% K Cu0.5Mg5CeOx
Number of carbon-13
0
1
2
3
4
Carbon dioxide
Acetaldehyde
Ethanol
Propionaldehyde
n-Butyraldehyde
Isobutyraldehyde
11.4
99.4
100
4.4
97.5
0
88.6
0.6
0
91.6
2.5
10.1
—
0
0
4.1
0
—
—
—
0
0
20.6
—
—
—
—
0
70.3
0
SCHEME 8. Reaction scheme of the interconversion of aldol and keto
Note. 573 K, 101.3 kPa total pressure, 8.0 kPa methanol, 4.0 kPa
species.
acetaldehyde, balance He.

170
GINES AND IGLESIA
CO hydrogenation at high pressures). The ethanol reac-
tion product was predominantly unlabeled (Table 5) and
it formed by hydrogenation of acetaldehyde using Cu sites
and hydrogen produced via methanol decomposition.
Propionaldehyde contains predominantly one 13C atom,
suggesting that it is formed by condensation of acetalde-
hyde with reactive formyl or methoxy-type surface species
or with gas phase formaldehyde derived from methanol.
The analysis of the mass fragmentation pattern showed
that the labeled carbon is present in either the CH3 or
CHO fragments of propionaldehyde. The methyl group
contains 35% 13C (at contact times of 2.7 h), suggesting
that aldol condensation reactions can occur with reten-
tion of either one of the two oxygen atoms in aldol-keto
species as proposed by Nunan et al. (7, 8). They used
the term “oxygen retention reversal” to indicate the oxy-
gen atom in the C1 species (formaldehyde) remains in
SCHEME 9. Reaction scheme for methanol-acetaldehyde reactions.
the final product, whereas in normal aldol condensation tion pathways are reversible. COx and H2 (from methanol
the oxygen of the growing intermediate remains in the fi- decomposition and water–gas shift reaction), isobutyralde-
nal product. Both normal and oxygen retention reversal hyde, 1-propanol, and methyl propionate were the main re-
–
types of aldol condensation occurred on K CuyMg5CeOx action products. Other products included methyl formate,
catalysts. These types of reactions are caused by in- 3-pentanone (from propionaldehyde self-condensation fol-
terconversion of aldol and keto species via H-transfer lowed by decarboxylation), and 2-methyl-pentaldehyde
(Scheme 8).
(from propionaldeyde self-condensation). 2,2-dimethyl bu-
Isobutyraldehyde contained predominantly two 13C tyraldehyde (from cross coupling of isobutyraldehyde and
atoms, suggesting that it forms by condensation of pro- C1 species) was not observed, indicating that isobutyralde-
pionaldehyde (with one 13C) with a C1 species formed hyde (precursor to isobutanol) is a preferred end-product
from methanol (Table 5). The CO2 product of CH3OH
13
–
of chain growth reactions. Neither ethanol nor acetalde-
12C2H4O mixtures on Cu0.5Mg5CeOx was predominantly hyde were detected. Scheme 10 shows a reaction scheme for
labeled and formed from 13CO (via 13CH3OH decomposi- propionaldehyde-13C-methanol coupling reactions consis-
tion) via water-gas shift reactions (Table 5). Acetone and tent with the results presented in this section. Independent
n-butyraldehyde showed no significant 13C-enrichment studies of alcohol reaction pathways during CO/H2 reac-
and they formed only by self-condensation reactions of tions at high pressures [4.5 MPa, 573–593 K, CO/H2 = 1] are
acetaldehyde or ethanol, as also observed in reactions also in agreement with the pathways reported here (50).
of pure ethanol (Scheme 4). This also accounts for the
The results obtained from quantitative analysis of the iso-
presence of small amounts of unlabeled CO2. Only traces topic content during propionaldehyde-13C-methanol cou-
–
of labeled ethanol were detected among reaction products, pling reactions on 1.0 wt% K Cu0.5Mg5CeOx are shown in
suggesting that methanol coupling reactions, previously Table 6. CO2 molecules were predominantly labeled
proposed as a route for C1 to C2 chain growth during (88% 13C) and formed from methanol-derived 13CO via
higher alcohol synthesis (6), do not readily occur at these water–gas shift reactions. Unlabeled CO2 was formed via
low pressures. In contrast with high-pressure CO hydro- decarboxylation reactions that also form 3-pentanone
genation conditions, our reaction conditions favor the (Scheme 10). Propionaldehyde acquires some 13C during
decomposition of methanol and minimize its bimolecular reaction. The increase in the 13C-content of propionalde-
chain-growth reactions.
hyde molecules as a function of contact time can be ex-
These isotopic tracer studies are consistent with the se- plained by Scheme 11. The proposed pathways include the
quence of steps shown in Scheme 9. Reverse aldol con- formation of surface aldol species (from propionaldehyde
densation reactions of aldol and keto species (formed from and C1 intermediates), keto-enol transformation (by
13C1 aldehyde-type intermediates and acetaldehyde) form intramolecular H-transfer) and reverse aldol condensation
1-13C labeled acetaldehyde, which in turn can lead to propi- from the keto form, leading to propionaldehyde and C1
onaldehyde with 2-13C atoms, isobutyraldehyde with 3-13C species.
atoms, and traces of labeled n-butyraldehyde.
Isobutyraldehyde is predominantly labeled with one 13C
Reactions of propionaldehyde-13C-methanol mixtures (Table 6), suggesting that it forms most frequently by cross-
confirm these condensation pathways. These studies also coupling reactions between 13C-labeled methanol and pro-
show that aldol formation and dehydration–decarbonyla- pionaldehyde. Isobutyraldehyde with two 13C forms by

BIFUNCTIONAL CONDENSATION REACTIONS
171
TABLE 6
13C-Distribution in Reactants and Products of 13CH3OH C3H6O
–
–
Reactions on 1.0 wt% K Cu0.5Mg5CeOx
Time (min)
0
5
10
20
Propionaldehyde
0
1
2
3
1.000
0.000
0.000
0.000
0.000
0.999
0.001
0.000
0.000
0.000
0.998
0.002
0.000
0.000
0.001
0.996
0.004
0.000
0.000
0.002
%
13C
Isobutyraldehyde
0
1
2
3
4
0.000
0.978
0.034
0.000
0.000
0.252
0.000
0.958
0.054
0.000
0.000
0.243
0.000
0.941
0.069
0.000
0.000
0.262
SCHEME 11. Reaction scheme of the labeled propionaldehyde pro-
duction.
%
13C
4. CONCLUSIONS
Methyl propionate
The addition of K increases the density and strength of
0.988 basic sites and the areal rates of aldol condensation chain
0
1
2
3
4
0.033
1.025
0.000
0.002
0.000
0.228
0.023
0.000
0.000
0.000
0.242
growth reactions. Ethanol dehydrogenation is a structure-
insensitive reaction that requires exposed Cu sites for its
rate-determining elementary steps. Cu sites also play a crit-
ical role in condensation reactions. Alcohol chain growth
reactions occur via bifunctional (metal-base) pathways re-
%
13C
Note. 573 K, 101.3 kPa total pressure, 7.7 kPa methanol, 3.5 kPa propi-
onaldehyde, balance He.
–
quiring both Cu and basic sites. C H bond activation steps
on acid–base site pairs on basic oxides are coupled kineti-
cally with recombinative desorption on Cu sites during re-
aldol condensation reactions of labeled methanol (13C1
aldehyde-type intermediate) with singly-labeled propi-
onaldehyde formed by reverse aldol condensation re-
actions. Methyl propionate also contains predominantly
one 13C, showing that it forms via reactions of methanol
(methoxide) with propionaldehyde by Tischenko-type re-
actions (34). Both carbon atoms of methyl formate were
labeled, indicating that it is formed from two molecules of
methanol.
–
actions of alcohols on K CuyMg5CeOx catalysts.
12
Isotopic tracer studies using C2H5OH 13C2H4O reac-
–
tant mixtures confirm that condensation reactions can pro-
ceed via direct reactions of ethanol without the interme-
diate formation of gas phase acetaldehyde molecules. The
effect of Cu surface area on the rate of deuterium incor-
poration into the unreacted ethanol is consistent with the
proposed bifunctional mechanism, in which Cu sites in-
crease the rate of aldol condensation by removing surface
hydrogen that tends to reverse the step forming the unsat-
urated aldol-type species required for chain growth. Iso-
topic tracer studies of alcohol coupling reactions suggest
that chain growth occurs by the addition of a methanol-
derived C1 species to adsorbed oxygenates. These studies
also show that aldol condensation reactions of alcohols are
reversible on metal-base bifunctional catalysts at these re-
action conditions.
ACKNOWLEDGMENTS
This work was supported by Division of Fossil Energy of the United
States Department of Energy under Contract DE-AC22-94PC94066. The
authorsthankDr. M. XuandMr. ZhengjieHufortheirtechnicalassistance
with the CO isotopic switch experiments and Drs. M. Xu and A. M.
2
Hilmen for frequent discussion and helpful suggestions. M. J. L. Gines
acknowledges the Universidad Nacional del Litoral, Santa Fe, Argentina
for a postdoctoral fellowship.
SCHEME 10. Reaction scheme for methanol–propionaldehyde reac-
tions.

172
GINES AND IGLESIA
24. Robbins, J. L., Iglesia, E., Kelkar, C. P., and DeRites, B., Catal. Lett.
REFERENCES
10, 1 (1991).
1. Keim, W., and Falter, W., Catal. Lett. 3, 59 (1989).
2. Keim, W., D.E. Patent 3810421 (1988).
25. Neophytides, S. G., Marchi, A. J., and Froment, G. F., Appl. Catal. A
86, 45 (1992).
3. Lietti, L., Forzatti, P., Tronconi, E., and Pasquon, I., J. Catal. 126, 401 26. Joo, O. S., Jung, K. D., Han, S. H., Uhm, S. J., Lee, D. K., and Ihm,
(1990).
S. K., Appl. Catal. A 135, 273 (1996).
4. Forzatti, P., Tronconi, E., and Pasquon, I., Catal. Rev. Sci. Eng. 33, 109
(1991).
5. Sofianos, A., Catal. Today 15, 119 (1990).
27. Mazanec, T. J., J. Catal. 98, 115 (1986).
28. Fox, J. R., Pesa, F. A., and Curatolo, B. S., J. Catal. 90, 127 (1984).
29. Elliot, D. J., and Pennella, F., J. Catal. 114, 90 (1988).
6. Nunan, J. G., Bogdan, C. E., Klier, K., Smith K. J., Young, C.-W., and 30. Vedage, G. A., Himelfarb, P. B., Simmons, G. W., and Klier, K., ACS
Herman R. G., J. Catal. 113, 410 (1988). Symp. Ser. 279, 295 (1985).
7. Nunan, J. G., Bogdan, C. E., Herman, R. G., and Klier, K., Catal. Lett. 31. Xu, M., and Iglesia, E., Catal. Lett., in press.
2, 49 (1989). 32. Xu, M., and Iglesia, E., unpublished results.
8. Nunan, J. G., Bodgan, C. E., Klier, K., Smith, K., Young, C., and 33. Loudon, G. M., “Organic Chemistry,” 2nd ed., Benjamin/Cummings,
Herman, R. G., J. Catal. 116, 195 (1989).
Redwood City, CA, 1988.
9. Nunan, J. G., Herman, R. G., and Klier, K., J. Catal. 116, 222 (1989).
34. Elliot, D. J., and Pennella, F., J. Catal. 119, 359 (1989).
10. Slaa, J. C., van Ommen, J. G., and Ross, J. R. H., Catal. Today 15, 129 35. Chinchen, G. C., Hay, C. M., Vandervell, H. D., and Waugh, K. C.,
(1992). J. Catal. 103, 79 (1987).
11. Apesteguia, C. R., Soled, S. L., and Miseo, S., U.S. Patent 5,387,570 36. Iglesia, E., and Boudart, M., J. Catal. 81, 204 (1983).
(1995), assigned to Exxon Research and Engineering Co.
12. Apesteguia, C. R., DeRites, B., Miseo, S., and Soled, S. L., Catal. Lett.
44, 1 (1997).
37. Scholten, J. J., and Kovanlinka, J. A., Trans. Faraday Soc. 65, 2454
(1969).
38. Xu, M., and Iglesia, E., J. Phys. Chem. 102, 961 (1998).
13. Xu, M., Gines, M. J. L., Hilmen, A. M., Stephens, B. L., and Iglesia, 39. Price, G. L., and Iglesia, E., Ind. Eng. Chem. 28, 839 (1989).
E., J. Catal. 171, 130 (1997). 40. McKenzie, A. L., Fishel, C. T., and Davis, R. J., J. Catal. 138, 547 (1992).
14. Chinchen, G. C., Denny, P. J., Jennings, J. R., Spencer, M. S., and 41. Lamonier, C., Bennani, A., Duysser, A., and Wrobel, G., J. Chem.
Waugh, K. C., Appl. Catal. 36, 1 (1988).
Soc., Faraday Trans. 92(1), 131 (1996).
15. Waugh, K. C., Catal. Today 15, 51 (1992).
42. Soria, J., Solid State Ionics 63–65, 755 (1993).
43. Takezawa, N., Hanamaki, C., and Kobayashi, H., J. Catal. 38, 101
(1975).
44. Parrott, S. L., Roger, W., and White, J. M., Applications of Surf. Sci. 1,
443 (1978).
45. Mole, T., Anderson, J. R., and Creer, G., Appl. Catal. 17, 141
(1985).
46. Biscardi, J. A., and Iglesia, E., Catal. Today 31, 207 (1996).
47. Iglesia, E., Barton, D. G., Biscardi, J. A., Gines, M. J. L., and Soled,
S. L., Catal. Today 38, 339 (1997).
48. Norman, R., and Coxon, J. M., Principles of Organic Synthesis, 3rd ed.,
Blackie Academic, Glasgow/London, 1993.
49. Chung, M. J., Moon, D. J., Kim, H. S., Park, K. Y., and Ihm, S. K.,
J. Molec. Catal. A: Chemical 113, 507 (1996).
50. Hilmen, A.-M., Xu, M., Gines, M. J. L., and Iglesia, E., Appl. Catal.,
in press.
16. Tsuji, H., Yagi, F., Hattori, H., and Kita, H., J. Catal. 148, 759 (1994).
17. Radlowski, C. A., and Hagen, G. P., U.S. Patent 5,159,125, assigned to
Amoco Corporation, Chicago, IL.
18. Ueda, W., Kuwabara, T., Ohshida, T., and Morizawa, Y., J. Chem. Soc.,
Chem. Comm. 1558 (1990).
19. Ueda, W., Kuwabara, T., Ohshida, T., and Morizawa, Y., Catal. Lett.
12, 97 (1992).
20. Chinchen, G., Mansfied, K., and Spencer, M., CHEMTECH (1990).
21. Nakamura, J., Uchijima, T., Kanai, Y., and Fujitani, T., Catal. Today
28, 223 (1996).
22. Bowker, M., Hadden, R. A., Houghton, H., Hyland, J. N. K., and
Waugh, K. C., J. Catal. 109, 263 (1988).
23. Chauvin, C., Saussey, J., Lavalley, J. C., Idriss, H., Hindermann,
J. P., Kiennemann, A., Chaumette, P., and Courty, P., J. Catal. 121,
56 (1990).