Structural requirements and reaction pathways in condensation reactions of alcohols on MgyAlOx catalysts

Article abstract of DOI:10.1006/jcat.1999.2734

A study of the effect of composition and of surface characteristics on alcohol-coupling reactions on Mg_yAlO_x catalysts using C₂H₅OH or ¹³CH₃OH/1-¹²C₃H₇OH mixtures as reactants showed that the biomolecular and consecutive character of the condensation reactions were affected by the catalyst acid-base properties and by the chemical nature of the alcohols and steric factors. The rates and product selectivity for C₂H₅OH or CH₃OH/C₃H₇OH reactions strongly depended on the chemical composition of Mg_y/AlO_x samples. The chemical composition affected the acid-base properties of Mg_yAlO_x samples. The rate of alcohol dehydration to ethers and olefins increased with increases in Al content. Al-rich Mg_yAlO_x samples contained a high density of Al³⁺-O^2- site pairs and of moderate strength basic sites, the combination of which promoted ethylene or propylene formation from primary alcohols via E₂ elimination pathways. The competitive dehydration to form ethers involved the adsorption of two alcohol molecules on neighboring acid sites offering various acid-base properties. On Mg_yAlO_x samples, the active acid sites for ether formation were possibly the Al³⁺ cations and the basic sites were the neighboring O^2- ions. The dehydrogenation of alcohols to aldehydes involved the initial alkoxy intermediate formation on weak Lewis acid-strong Broensted base site pairs. The synthesis of C₂H₄O or C₃H₆O was favored on Mg-rich Mg_yAlO_x samples, which contained a larger number of property positioned Al³⁺ Lewis acid sites and Mg²⁺-O^2- basic pairs required for hydrogen abstraction steps leading to alkoxy intermediates. Pure MgO showed lower dehydrogenation rates than Mg-rich Mg_yAlO_x samples. Aldol condensation reactions on Mg_yAlO_x samples also involved the formation of a carbanion intermediate on Lewis acid-strong Broensted base pair sites and produced products with a novel C-C bond, e.g., n-C₄H₈O and iso-C₄H₈O. Reactions leading to condensation products were also favored on Mg-rich samples.

Full text of DOI:10.1006/jcat.1999.2734

Journal of Catalysis 190, 261–275 (2000)
doi:10.1006/jcat.1999.2734, available online at http://www.idealibrary.com on
Structural Requirements and Reaction Pathways in Condensation
Reactions of Alcohols on Mg AlO Catalysts
y
x
�
�
J. I. Di Cosimo, C. R. Apestegu ´ı a, M. J. L. Gin e´ s,† and E. Iglesia†
�
Instituto de Investigaciones en Cat a´ lisis y Petroqu ´ı mica (INCAPE), Santiago del Estero 2654, (3000) Santa Fe, Argentina;
and †Department of Chemical Engineering, University of California at Berkeley, Berkeley, California 94720
E-mail: capesteg@fiqus.unl.edu.ar
Received July 8, 1999; accepted October 19, 1999
ucts containing a new C–C bond such as n-C4H8O (or n-C4H9OH)
The effect of composition and of surface properties on alcohol- and iso-C4H8O (or iso-C4H9OH). Reactions leading to condensa-
coupling reactions was studied on MgyAlOx catalysts using C2H5OH tion products were also favored on Mg-rich samples, but they took
13
12
or CH3OH/1– C3H7OH mixtures as reactants. Samples with place at much slower rates than those of the corresponding dehy-
Mg/Al ratios of 0.5–9.0 were obtained by thermal decomposi- drogenation reactions to aldehydes. This reflected the bimolecular
tion of precipitated hydrotalcite precursors. The nature, density, and consecutive character of condensation reactions, which are af-
and strength of surface basic sites were obtained by temperature- fected not only by the catalyst acid–base properties but also by the
1
3
12
programmed desorption (TPD) of CO2 and by CO2/ CO isotopic chemical nature of the alcohols and steric factors.
�c 2000 Academic
switch methods, whereas the acid site densities were measured by
TPD of NH3. The catalyst ability for activating H–H bonds was
Press
Key Words: alcohol coupling reactions; MgyAlOx catalysts; base–
investigated by performing H2–D2 steady-state equilibration reac- acid catalysis.
tions. Isotopic tracer studies were carried out in order to probe chain
growth pathways in the synthesis of isobutanol. The rates and prod-
uct selectivity for C2H5OH or CH3OH/C3H7OH reactions strongly
depended on the chemical composition of MgyAlOx samples. In
INTRODUCTION
Copper-based catalysts are widely used to produce
turn, the chemical composition affected the acid–base properties of
MgyAlOx samples by modifying surface acid and base site densities linear or branched higher alcohols from synthesis gas.
and the distribution of strength for such sites. The rate of alcohol de- Cu/ZnO/Al O catalysts selectively form methanol from
2
3
hydration to ethers and olefins increased with increasing Al content. CO/CO /H mixtures at low pressures (<5 MPa) and tem-
2
2
3+
2�
Al-rich MgyAlOx samples contained a high density of Al � O site
pairs and of moderate strength basic sites, the combination of which
promoted the formation of ethylene or propylene from primary al-
cohols via E2 elimination pathways. The competitive dehydration
to form ethers involved the adsorption of two alcohol molecules
on neighboring active sites offering different acid–base properties.
On MgyAlOx samples, the active acid sites for ether formation were
peratures (<550 K) (1). Other methanol synthesis catalysts
also promote chain growth and the formation of higher al-
cohols from synthesis gas. For example, Cu-Co/ZnAl2O4
catalysts give mixtures of methanol and linear C2–C6 alco-
hols (2, 3). Alkali species, especially those of Cs, increase
the rate of formation of branched alcohols (4, 5). Isobutanol
(2-methyl-1-butanol) is preferentially formed because it is
a kinetic endpoint of aldol-coupling reactions catalyzed by
3+
probably the Al cations, whereas the basic sites were the neighbor-
ing O2 ions. The abundance of surface Al –O pairs accounted
�
3+
2�
for the high ether formation rates observed on Al2O3 and Al-rich basicsiteson these catalysts. The synthesisofbranched alco-
MgyAlOx samples. The dehydrogenation of alcohols to aldehydes holsfrom H –CO mixtureshave also been reported recently
2
(
C2H4O or C3H6O) involved the initial alkoxy intermediate forma- on K- and Cs-promoted Cu Mg CeO catalysts (6–8).
z y x
tion on weak Lewis acid–strong Brønsted base site pairs. The syn-
thesis of C2H4O or C3H6O was favored on Mg-rich MgyAlOx samples
because these samples contained a much larger number of properly
Isobutanol is a useful chemical and a potential precursor
to isobutene and methyl-tert-butyl ether. Isobutanol syn-
thesis requires several catalytic functions: (i) hydrogena-
tion sites to form methanol and (ii) sites for C1 to C2 chain
growth and C2+ formation via different mechanisms cata-
lyzed by basic sites aided by Cu sites (4, 9). Methanol
synthesis pathways on Cu have been widely studied, and
there exists general agreement about the requirement for
positioned Al3 Lewis acid sites and Mg –O basic pairs, which
are required for hydrogen abstraction steps leading to alkoxy in-
termediates. Pure MgO showed lower dehydrogenation rates than
Mg-rich MgyAlOx samples because the predominant presence of
+
2+
2�
isolated O2� hindered formation of alkoxy intermediates by alcohol
dissociative adsorption. Aldol condensation reactions on MgyAlOx
samples involved also the formation of a carbanion intermediate Cu metal sites and for CO2 as a “co-catalyst” for the reac-
on Lewis acid–strong Brønsted base pair sites and yielded prod- tion of H –CO mixtures (10, 11). Chain growth pathways
2
261
0021-9517/00 $35.00
Copyright �c 2000 by Academic Press
All rights of reproduction in any form reserved.

2
62
Di COSIMO ET AL.
leading to branched alcohols on metal–base catalysts are relative abundance of basic and acid sites required for chain
more complex, depend on the detailed catalyst composi- growth pathways.
tion (4, 9), and remain less clear. Specifically, identification
of rate-determining steps in the overall reaction sequence
for isobutanol synthesis and the specific basic sites required
EXPERIMENTAL
for such steps remain unclear.
1. Catalyst Preparation
Hydrotalcite-derived Mg–Al mixed oxides catalyze reac-
tions requiring carbanion intermediates, such as aldol con-
densation reactions of aldehydes and ketones (12–14), side-
chain alkylations (15), Knoevenagel condensations (16),
cyanoethylation of alcohols (17), and double-bond iso-
merization of alkenes (18). The surface acid–base proper-
ties of Mg/Al hydrotalcites decomposed in nitrogen and
their catalytic properties depend on chemical composition
and on the method used to decompose the hydrotalcite
precursors (14, 18, 19). Intermediate Mg/Al ratios lead to
optimum catalytic properties for several reactions, but the
optimum Mg/Al ratio for each reaction depends on the
specific requirements for basic site density and strength.
Schaper et al. (18) reported that Mg–Al hydrotalcites
treated in air at 773 K show stronger basic sites and higher
Mg–Al hydroxycarbonate precursors with Mg/Al atomic
ratios (y) of 0.5, 1, 5, and 9 were prepared by co-
precipitation. An acidic solution of the metal nitrates with a
total [Al + Mg] cation concentration of 1.5 M was contacted
with an aqueous solution of KOH and K2CO3 at a constant
pH of 10. The two solutions were simultaneously added
dropwise to 300 ml distilled water kept at 333 K in a stirred
batch reactor. The resulting precipitates were aged for 2 h
at 333 K in their mother liquor and then filtered, washed
thoroughly with 300–500 ml deionized water at 373 K, and
dried at 348 K overnight. The residual potassium content
in all dried samples was below 0.1 wt% . Hydrotalcite pre-
cursors were decomposed in N2 at 673 K for 4 h in order to
obtain the corresponding Mg–Al mixed oxides (MgyAlOx).
Pure MgO and Al2O3 were prepared following the same
procedure.
1
-pentene double-bond migration rates than pure MgO.
Corma et al. (19) observed maximum isopropanol dehydro-
genation rates on air-treated hydrotalcites with an Mg/Al
ratio of 3. The rate of self-condensation of acetone to �,�-
unsaturated ketones reaches a maximum on Al-rich sam-
2
. Catalyst Characterization
The crystalline phases in the precipitated hydrotalcites
ples (Mg/Al < 1) (14), apparently as a result of contribu- before and after decomposition were determined by X-ray
tions from aldol condensation pathways catalyzed by both diffraction (XRD) using a Shimadzu XD-Dl diffractome-
basic and acidic sites.
ter and Ni-filtered Cu K� radiation. BET surface areas (Sg)
In a previous study (20), we examined the structural were measured by N physisorption at its boiling point us-
2
purity, site density, surface composition, and acid–base ing a Quantachrome Nova-1000 sorptometer. Elemental
properties of Mg–Al catalysts prepared by decomposition compositions were measured by atomic absorption spec-
of hydrotalcite precursors in nitrogen. Here, we describe troscopy (AAS).
the effect of composition and of surface properties on
CO2 adsorption site densities and binding energies
13
alcohol coupling reactions, using C2H5OH or CH3OH/ were obtained from temperature-programmed desorption
12
1
- C3H7OH as reactants. The objectives of these kinetic (TPD) of CO preadsorbed at room temperature. Samples
2
3
and isotopic tracer studies were to establish the reaction (50 mg) were treated in He (� 100 cm /min) at 723 K for
steps required for alcohol chain growth and to ascertain 0.3 h and exposed to 0.1% CO /0.1% Ar/He (Cambridge
2
the sites and structures required for the efficient catalysis Isotope Laboratories Inc.) for 0.15 h at room temperature
of such steps.
2
(RT). Weakly adsorbed CO was removed by flowing He
The nature, density, and strength of acid and basic sites at RT, and the temperature was then increased to 723 K
are probed by temperature-programmed desorption of at 30 K/min. CO concentrations in the effluent were mea-
2
NH3 and CO2 coupled with infrared spectra of adsorbed sured by mass spectrometry (MS).
CO2. On Mg–Al mixed oxides, both the surface acid–base
Basic site densities were also measured by 13CO / CO
12
2
2
properties and the catalytic activity and selectivity depend isotopic switch methods that give the number of sites avail-
on chemical composition. Mg-rich samples form predom- able for reversible adsorption–desorption of CO at typi-
2
inantly aldehydes via dehydrogenation reactions, whereas cal reaction temperatures, as described in detail elsewhere
Al-rich catalysts favor dehydration to ethers. Coupling pro- (8, 20). Samples were exposed to a 0.1% 13CO /0.1% Ar/He
2
ducts are formed via consecutive aldol condensation steps (Cambridge Isotope Laboratories, Inc.) stream at 573 K
from aldehydic intermediates. Dehydration, dehydrogena- until 13CO reached a constant concentration in the efflu-
2
12
tion, and coupling reactions proceed through different ent (0.5 h). Then, the flow was switched to 0.1% CO /He
2
mechanisms and require different surface active sites. (Cambridge Isotope Laboratories, Inc.). The resulting de-
13
The effect of chemical composition on the selectivity of crease in the surface concentration of CO , as it was
2
MgyAlOx samples is interpreted in terms of changes in the replaced by 12CO , was followed by mass spectrometric
2

CONDENSATION REACTIONS OF ALCOHOLS ON MgyAlOx
263
analysis of 13CO2 and Ar concentrations in the effluent a gradientless recirculating batch reactor, at atmospheric
stream as a function of the time elapsed after the isotopic pressure (21). Samples (38.0 mg) were treated in vacuum
switch. Ar was used as an internal standard in order to cor- by heating to 723 K at 3.5 K/min and holding for 2 h. The
rect for gas holdup and hydrodynamic delays within the temperature was then decreased to 573 K, and the reactants
apparatus.
were introduced along with a small amount of methane
Acid site densities were measured by TPD of NH3 pre- (used as an inert internal standard in order to ensure
adsorbed at room temperature. Samples (150 mg) were accurate mass balances). Reactant mixtures consisting of
3
12
13
treated in He (� 100 cm /min) at 723 K for 0.5 h and exposed
C3H7OH/ CH3OH/CH4/He (1.4/2.7/2.7/94.5 kPa) were
to a 0.93% NH3/He stream until saturation coverages were circulated continuously through the catalyst bed at about
reached. Weakly adsorbed NH3 was removed by flowing He 300 cm /min. The circulating stream was sampled after var-
3
at RT for 0.5 h. Temperature was then increased to 773 K ious contact times by syringe extraction. Reactants and
at 10 K/min, and the NH3 concentration in the effluent was productswere analyzed usinga dual-detector gaschromato-
measured by mass spectrometry.
graph (Hewlett-Packard, Model 5890 II Plus; FID/TCD de-
tectors). Gas samples were analyzed by flame ionization
after separation with a 5% phenyl methyl-silicone capil-
lary column (HP-1, 50 m, 0.32 mm diameter, 1.05 �m film
thickness) and by thermal conductivity after separation in
a Porapak Q packed column (1.8-m length, 0.32-cm diam-
eter). Mass spectrometry was used after chromatographic
separations in order to confirm the chemical identity of indi-
vidual chromatographic peaks and to measure the number
3
. Catalytic Reactions
.1. H2–D2 steady-state equilibration reaction on
3
MgyAlOx. H2–D2 equilibration reactions were carried
out in a differential packed bed reactor. Decomposed
hydrotalcite samples (150 mg) were treated in flowing He
3
(
(
� 100 cm /min) at 723 K for 0.5 h. Then, an H2/D2/Ar
3
5% /5% /90% ) mixture was introduced at 105 cm /min
13
and position of C within reactants and products (Hewlett-
while the sample was at room temperature, and the temp-
erature was increased at 10 K/min to 773 K. H2, D2, and
HD concentrations in the effluent were measured con-
tinuously by mass spectrometry. Calibrations for H2, HD,
and D2 were obtained from the signals at 2, 3, and 4 amu
using the reactant mixture and an isotopically equilibrated
mixture as standards.
Packard, Model 5890 II Plus GC; Hewlett-Packard, Model
13
5
972 Mass Selective Detector). The C content and the
isotopomer distributions of reactants and products were
obtained from mass spectrometry data using matrix decon-
13
volution methods that account for natural C abundance
and for mass fragmentation patterns (22).
RESULTS AND DISCUSSION
3
.2. Reactions of ethanol on MgyAlOx. Ethanol reac-
tions were carried out in a differential fixed-bed reactor at
73 K and atmospheric pressure. Samples were sieved to
1
. Catalyst Characterization
5
The precipitated hydrotalcite precursors showed diffrac-
retain particles with 0.35–0.42 mm diameter for catalytic
measurements and treated in N2 at 673 K for 1 h before re-
action in order to remove adsorbed H2O and CO2. C2H5OH
tion patterns consistent with the presence of a crystalline
hydrotalcite structure (ASTM 14-191). The diffraction pat-
tern for the sample with an Mg/Al ratio of 9 also showed
an additional Mg(OH)2 phase (ASTM 7-239).
The physiochemical properties of the hydrotalcite sam-
ples after decomposition at 673 K are shown in Table 1,
(
Merck, ACS, 99.8% purity) was introduced via a syringe
pump and vaporized into flowing N2 to give a N2/C2H5OH
molar ratio of 10. Reaction rates and selectivities were in-
dependent of particle size in the range between 0.15 and
0
.59 mm; thus, diffusional restrictions do not corrupt mea-
TABLE 1
sured reaction rates. Reaction rates are defined on an areal
basis as the moles of ethanol converted per unit of time and
per unit ofBET surface area. Catalytictestswere conducted
Composition, Surface Area, and XRD Characterization
of MgyAlOx Samples
at contact times (W/F0
) between 4 and 120 g h/mol
C2H5OH
Compositiona
Al/(Al + Mg)
(molar)
BET
surface area
C2H5OH. The concentrations of unreacted C2H5OH and
of reaction products in the reactor effluent were measured
at 0.5-h intervals for 10 h using an ATI Unicam 610 chro-
Phases detected
by XRD
2
Sample
(m /g)
matograph equipped with a 0.2% Carbowax 1500/80-100 MgO
0.00
0.11
0.18
0.47
0.65
1.00
191
114
184
231
298
388
MgO
MgO
MgO
Mg9AlOx
Carbopack C column and flame ionization detection.
Mg5AlOx
Mg1AlOx
Mg0.5AlOx
Al2O3
3
.3. Cross-coupling reactions of 13CH3OH/1– C3H7OH
12
MgO
MgO + MgAl2O4
mixtures on MgyAlOx. Cross-coupling reactions of
amorphous
13
13
18
CH3OH ( C, 99% ; O, <1% ; Icon Services Inc.) with
unlabeled 1-C3H7OH (Aldrich, 99.7% ) were carried out in
a
Bulk composition measured by AAS.

2
64
Di COSIMO ET AL.
which includes chemical composition, BET surface area, density of kinetically accessible active sites in MgO; MgO
2
and crystalline phases for MgO, Al2O3, and MgyAlOx sam- adsorbs 0.38 �mol CO2/m , but sample Mg9AlOx with only
2
ples. The Al/(Al + Mg) atomic ratios are very similar to 6.2% wt Al adsorbs 0.17 �mol CO2/m . X-ray photoelec-
those present in the precursor solution, suggesting that Mg tron spectroscopy showed significant Al surface enrichment
and Al salts precipitate completely during synthesis. X-ray in MgyAlOx samples with low Al/(Al + Mg) ratios (<0.20)
diffraction patterns show the presence of a poorly crys- (20). The presence of more electronegative amorphous
talline MgO periclase phase in all the MgyAlOx samples AlOx domains or isolated species on MgO blocks CO2 ad-
prepared by decomposition of hydrotalcite precursors. The sorption sites, and it is likely to account for the strong effect
Mg0.5AlOx sample contains an additional MgAl2O4 spinel of Al on the density of surface sites. At higher Al/(Al + Mg)
phase, but crystalline AlOx phases were not detected in any ratios (0.2–0.5), the nucleation of a separate stable Al-rich
of the MgyAlOx samples.
phase provides a thermodynamic sink for Al3 cations, and
+
Mg–Al hydroxycarbonate precursors evolve CO2 and the density of basic sites increases as Al migrates from MgO
H2O during thermal decomposition; this rapid gas evolu- surfaces to the Al-rich phase to reach a second maximum
tion leads to the formation of a well-connected network for the Mg1AlOx sample.
2
3+
of small pores with significant surface area (100–300 m /g).
In this composition range, the Al cations located in the
The intimate contact between Mg and Al cations in the MgO framework create a defect in order to compensate the
hydrotalcite structure is preserved during decomposition positive charge generated, and the adjacent oxygen anions
and leads to the formation of well-mixed MgyAlOx binary become coordinatively unsaturated; as a result, the solid
oxides. Their surface areas increase with increasing Al con- solution displays higher basicity. In Al-rich samples (sam-
tent, apparently because the carbonate content and thus ple Mg0.5AlOx), the nucleation of a separate bulk MgAl2O4
the amount of CO2 formed during decomposition are pro- spinel lacking basic surface oxygens causes the basic site
portional to the Al content (23).
density to decrease again at higher Al contents. We have
13
12
Basic site densities measured by CO2/ CO2 isotopic ex- also observed this decrease in basic site density at high Al
change at 573 K on MgO, Al2O3, and MgyAlOx samples are contents by XPS, and chemisorption and TPD of CO2 (20),
shown in Table 2. These measurements reflect the number as also reported by others (24, 25) for a more narrow com-
of adsorption sites that can bind CO2 reversibly at typical positional range.
temperatures of catalytic reactions. This method does not
The distribution of binding sites with basic properties
count basic sites that are strong enough to bind CO2 irre- was determined by temperature-programmed desorption
versibly at this temperatures or sites that bind molecules too (TPD) of CO2 preadsorbed at RT on MgO, MgyAlOx, and
weakly and desorb CO2 with characteristic times shorter Al2O3. The TPD profile for Mg1AlOx (Fig. 1) shows that
than those detected in these measurements (� 2 s). The CO2 desorbs in three overlapping peaks, as also observed
2
surface density (per m ) and the total number (per g) of on all other samples except Al2O3. These features are re-
sites capable of reversible CO2 chemisorption are higher ferred to as �- (low-temperature), �- (intermediate temper-
on pure MgO than on MgyAlOx samples and very low on ature), and � - (high-temperature) peaks in the discussion
pure Al2O3 (Table 2). Small amounts of Al decrease the that follows. The relative contributions from these three
TABLE 2
Catalyst Characterization: Acid–Base Properties
Temperature-programmed desorption of CO2 and NH3
b
Areas under CO2 TPD peaks
Areas under NH3 TPD
Exchangeable
2
Total evolved
CO2
(�mol/m )
2
Total evolved
NH3
(�mol/m )
(
�mol/m )
peaks (�mol/m )
a
CO2 at 573 K
2
c
c
c
2
Bd
Ld
2
Catalyst
(�mol/m )
�
�
�
MgO
0.38
0.17
0.18
0.33
0.17
0.04
0.06(3.5)
0.32(19.8)
0.44(37.3)
0.17(37.6)
0.37(44.5)
0.32(44.2)
0.18(51.4)
1.25(76.7)
0.50(42.7)
0.21(45.7)
0.28(33.6)
0.24(32.6)
0.00(0.0)
1.63
1.17
0.46
0.83
0.73
0.34
0.18
0.49
0.21
0.53
0.42
0.50
0.30
0.32
0.64
1.04
0.97
0.84
0.48
0.81
0.84
1.57
1.39
1.34
Mg9AlOx
Mg5AlOx
Mg1AlOx
Mg0.5AlOx
Al2O3
0.23(20.0)
0.08(16.7)
0.18(21.9)
0.17(23.2)
0.16(48.6)
a
Measured by the 13CO2/ CO2 isotopic switch method.
12
b
c
d
Parenthesis are the percentage of contribution of each peak.
, low-temperature peak; �, middle-temperature peak; � , high-temperature peak.
B, Brønsted peak; L, Lewis peak.
�

CONDENSATION REACTIONS OF ALCOHOLS ON MgyAlOx
265
the adsorbed CO2 detected byisotopicexchange methodsat
73 K. This explains the low basic site density obtained from
5
2
exchange measurements on Al2O3 (0.04 �mol/m ; column 2
in Table 2). The total amount of CO2 evolved during TPD
(
column 6 in Table 2) follows compositional trends similar
to those observed for the kinetically accessible adsorbed
CO2 measured by isotopic exchange. The total surface den-
sity of adsorbed CO2 from TPD decreases from a maximum
2
value on MgO (1.63 �mol/m ) to a minimum value as small
amounts of Al are added, and then it increases to a second
2
maximum (0.83 �mol/m ) on the Mg1AlOx sample.
The acid properties of MgO, MgyAlOx, and Al2O3 sur-
faces were probed by TPD of NH3 preadsorbed at room
temperature. Previous work by Shen et al. (26) on the NH3
adsorption on Mg–Al mixed oxides showed that MgyAlOx
samples contain Brønsted and Lewis acid sites. These au-
thors assigned the Lewis acid sites to Al–O–Mg species
located within a MgO structure and containing the Al3
+
cations predominantly in octahedral sites. NH3 adsorption
on Lewis sites is stronger than on Brønsted acid sites pro-
vided by surface OH groups. Lewis acid sites adsorb NH3
FIG. 1. TPD profiles of CO2 and NH3 on Mg1AlOx. Deconvolution
of CO2 TPD trace: �, low-temperature peak; �, middle-temperature
peak; � , high-temperature peak. Deconvolution of NH3 TPD trace: B, by coordinating the nitrogen free electron pair with Mg
Brønsted peak; L, Lewis peak. Heating rates: 30 K/min (CO2 TPD), and Al cations, whereas on Brønsted sites a surface proton
1
0 K/min (NH3 TPD).
+
is donated to NH3 with formation of NH . In agreement
4
with the results obtained by Shen et al. (26), the NH3 TPD
profile on Mg1AlOx (Fig. 1) shows that NH3 desorbs in two
overlapping peaks, as also observed for all other MgyAlOx
samples and pure MgO and Al2O3. The low- (B-peak) and
high-temperature (L-peak) peaks are assigned to Brønsted
and Lewis surface acidic sites, respectively. The NH3 sur-
face densities for Brønsted and Lewis sites were obtained
by deconvolution and integration of the TPD traces and
are presented in Table 2, column 9. The total adsorbed NH3
surface density is low on MgO and Mg-rich samples, but it
increases with increasing Al content reaching a maximum
types of adsorbed CO2 were obtained by deconvolution of
the experimental TPD profiles and integration of the three
desorption features. Although CO2 desorption follows a
first-order kinetics, we have used for simplicity Gaussian
functions for the deconvolution of the TPD traces. The ad-
sorbed CO2 surface densities are shown in Table 2 for all
samples.
The complex TPD profiles suggest that MgO and
MgyAlOx surfaces are nonuniform and contain several
types of adsorbed CO2. Infrared spectra of CO2 adsorbed
on MgO and MgyAlOx (20) detected three distinct ad-
sorbed structures, located on surface sites with different
structuresand basicstrengths. Bicarbonatesform on weakly
basic OH groups, bidentate carbonates adsorb on Mg–O
site pairs with accessible cations, and unidentate carbon-
ates on strongly basic surface O2 anions. These combined
IR and TPD data led us to assign the desorption peaks to
CO2 species adsorbed on OH groups (�-peak), Mg-O pairs
2
at 1.57 �mol/m for the Mg1AlOx sample. The density of
2+
Lewis acid sites, consisting of Mg cations on pure MgO,
2
increases from 0.30 �mol/m on MgO to about 1.00 on Al-
rich MgyAlOx samples (Table 2, column 8). Pure MgO ex-
hibits the lowest density of Brønsted acid sites (Table 2,
column 7).
�
2
. H2–D2 Steady–State Equilibration Reactions on MgO,
Al2O3, and MgyAlOx Catalysts
2�
(
�-peak), and O anions (� -peak). With this assignment,
the data on Table 2 show that pure MgO contains the
2
highest absolute (1.25 �mol CO2/m ) and relative (76.7% )
The isotopic equilibration of gaseous H2–D2 mixtures,
amounts of strongly basic O2 sites. The relative concen- aswellasthe exchange ofgas–phase D2 with surface H, have
tration of surface OH groups and Mg–O pairs in MgyAlOx been employed for investigating the surface reactivity,
increases with increasing Al content, leading to a net de- state, and content of surface hydrogen either on mixed ox-
crease in the average basic strength of surface sites. Al2O3 ides (27, 28) or on metallic catalysts (29).
�
is an amphoteric oxide and lacks strong basic sites, but
Here, H2–D2 isotopic equilibration reactions were used
it contains surface sites capable of binding CO2 with low in order to explore the ability of Mg–Al oxides to activate
and medium binding energy, corresponding to surface OH H–H bonds. Areal H2–D2 equilibration rates (rHD, �mol/
2
groups and Al–O site pairs. Weakly bonded bicarbonate min m ) were measured as a function of temperature on
species desorb at low temperatures and do not account for MgO, Al2O3, and MgyAlOx samples, and the obtained rHD

2
66
Di COSIMO ET AL.
and Al2O3). These results are consistent with previous re-
ports, which have shown that the activation energy for
H2/D2 equilibration on Al2O3 is significantly higher than
on MgO (31, 32).
Pure MgO and Al2O3 catalyze H2–D2 equilibration at
low temperatures (27, 28) but show negligible catalytic
activity after complete surface dehydroxylation (33, 34),
thereby suggesting that surface hydrogen stabilized at OH
groups participates in the exchange reaction. Nakano et al.
(
35), proposed that the active sites on MgO are Mg–O
pairs, which dissociate H2 heterolytically to form magne-
+
sium hydride and H species in surface OH groups. Boudart
et al. (33) proposed instead that active sites on MgO con-
sist of electron-deficient paramagnetic centers vicinal to
surface OH groups with no direct participation of metal
cations. These later authors showed that for temperatures
lower than 300 K, pure MgO catalyzes the H2–D2 equili-
bration reaction but is almost inactive for exchanging sur-
face hydrogen with gas-phase D2. In the case of alumina,
Kazansky et al. (36) proposed that the heterolytic H2 disso-
ciation occurs on unstable Al –O pairs with subsequent
formation of surface aluminum hydrides. They also showed
that at temperatures lower than 300 K Al2O3 catalyzes
H–D exchange reactions, but it does not dissociate H2 signi-
ficatively. From previous reports and the results described
above, it appears that the higher activity of Mg-rich sam-
ples in H2–D2 equilibration reactions (Fig. 3) reflects the
higher rate of H2 dissociation on Mg–O. The role of sur-
face hydroxyl groups in H2–D2 equilibration on partially
dehydroxylated materials is probably to promote the mi-
gration of adsorbed H(D) species from H2(D2) dissoci-
ation to form HD by recombination of hydrogen atoms
FIG. 2. HD formation rate for H2–D2 equilibration reactions as a
function of temperature on MgO, Al2O3, and MgyAlOx samples [P =
+
�
0
D2
1
1
01.3 kPa, H2 :D2 :Ar = 5 :5 :90, W/F = 11.6 g h/mol, heating rate =
0 K/min].
vs T sigmoid-shaped curves are shown in Fig. 2. Magnesium-
rich samples (MgO and Mg9AlOx) are clearly more ac-
tive than samples with the highest Al content (Mg1AlOx,
Mg0.5AlOx, and Al2O3). At the temperature of our alco-
hol conversion catalytic studies reported below (573 K),
the Mg9AlOx sample gave the highest H2–D2 equilibration
rate.
H2–D2 equilibration rates depend on the surface concen-
tration of H and D and on the distance away from isotopic
equilibrium (30). For Langmuir–Hinshelwood mechanisms
via dissociative H2 and D2 adsorption, the HD formation
rate for a 5% D2/5% H2/Ar reactant mixture is given by
H2 + D2 ⇔ 2HD
�
�
�
�
0
.5 0.5
0.5
rHD = ks P
P
� PHD /K
= 0.05 ks 1 � 2XD2 , [1]
H2 D2
2
where ks (�mol/atm min m ) is the apparent first-order ki-
netic constant per unit of surface area, K is the equilibrium
constant (K � 4 throughout the experimental temperature
range), and XD is the D2 conversion. From the design equa-
2
tion for an isothermal plug–flow reactor, we obtain
Z
Z
XD2
XD2
W
d XD2
d XD2
=
=
. [2]
0
F
0.5SgrHD
0.025Sgks(1 � 2XD2 )
D2
0
0
The ks values on all catalysts were obtained by integrating
Eq. [2] using rate data obtained at XD values lower than
2
0
.1. Figure 3 shows the values of ks as a function of com-
position at 348, 373, and 398 K. Rate constants on MgO
and Mg9AlOx samples are about three orders of magnitude
FIG. 3. Apparent kinetic constant ks for H2–D2 equilibration reac-
tions as a function sample composition. The ks were obtained by integrat-
larger than on the Al-rich samples (Mg1AlOx, Mg0.5AlOx, ing Eq. [2] using D2 conversion values lower than 0.1 (see text).

CONDENSATION REACTIONS OF ALCOHOLS ON MgyAlOx
267
dissociated on separate dissociation sites. A proper combi-
nation of H2(D2) dissociation sites and H–D exchange sites
would then enhance the catalyst activity for H2–D2 equili-
bration reactions. Thus, the Mg9AlOx sample displays the
highest H2–D2 equilibration rate probably because of an
optimum balance between the number of surface Mg–O
pairs and surface OH groups
3
. Ethanol Reaction Rates and Pathways on MgO, Al2O3,
and MgyAlOx Catalysts
Dehydrogenation and coupling reactions of ethanol
(
C2H5OH) on mixed oxides form many products and pro-
ceed via complex reaction pathways that may require spe-
cific combinations of acid and basic sites. The predominant
products of ethanol reactions on Mg–Al oxides were ac-
etaldehyde (C2H4O), n-butanol (n-C4H9OH), diethylether
FIG. 5. Product distribution for ethanol conversion reactions on
0
0
Mg1AlOx. Initial product yields (� ) and ethanol conversion (X
)
i
C
H OH
(
C2H5OC2H5), and ethylene (C2H4). Minor amounts of n-
2
5
as a function of contact time [573 K, 101.3 kPa total pressure].
butyraldehyde (n-C4H8O), ethyl acetate (CH3COOC2H5),
and 2-propanol (2-C3H7OH) were also formed on MgO
and MgyAlOx catalysts with low Al content. Other coupling C H O, as well as minor amounts of ethyl acetate, were also
2
4
products, such as acetone (CH3COCH3) and 2-pentanone observed. Formation of n-C H O was detected only initially
4
8
(
CH3COCH2CH2CH3), were detected only in trace on this sample.
amounts. Diethylether and ethylene were the only prod-
ucts detected on the pure Al2O3 sample.
The effect of contact time on the product distribution
on Mg AlO was determined in order to identify primary
1
x
Ethanol conversion rates decreased with time-on-stream and secondary reaction pathways. The observed deacti-
on both MgO and MgyAlOx samples (to � 50% of ini- vation, however, required that each data point be ob-
tial rates after 10 h), but no deactivation was observed tained on a fresh catalyst and that initial product yields
on the Al2O3 sample. Figure 4 shows ethanol conversions be obtained by extrapolating to initial time-on-stream us-
0
i
(
XC H OH ) and product yields (�i; mol of product i/mol of ing semi-logarithmic plots. These initial yields, � , and the
2
5
ethanol in feed) on Mg1AlOx as a function of time-on- corresponding initial ethanol conversion values are shown
stream. Diethyl-ether and n-butanol were the main prod- as a function of contact time in Fig. 5. The local slopes of
ucts formed on Mg1AlOx; significant amounts of C2H4 and the curves in Fig. 5 give the rate of formation of each prod-
uct at a specific ethanol conversion and residence time.
The nonzero initial slopes for acetaldehyde, diethylether,
and ethylene show that they form directly from ethanol
via dehydrogenation, coupling followed by dehydration,
and dehydration reactions, respectively. Diethylether and
ethylene yields increase monotonically with increasing res-
idence time, because they react slowly in secondary reac-
tions, but acetaldehyde reaches a maximum concentration
as it converts to larger oxygenates in secondary conden-
sation reactions with increasing residence time. The initial
zero slope of the n-butyraldehyde yield curve is consistent
with its formation via the secondary condensation of pri-
mary acetaldehyde products. n-Butanol becomes the pre-
dominant product at longer contact times. The sigmoidal
shape of the n-butanol yield curve (Fig. 5) and its low but
nonzero initial slope suggest that n-butanol forms via two
or more parallel pathways, such as direct condensation of
ethanol and sequential paths involving hydrogenation of
the n-butyraldehyde formed via acetaldehyde condensa-
FIG. 4. Product distribution for ethanol conversion reactions on
Mg1AlOx. Product yields (�i) and ethanol conversion (XC2
H
5OH ) as a tion steps. Ethyl acetate was also detected in low concen-
function of time [573 K, 101.3 kPa total pressure, N2/ethanol = 10, W/ trations among reaction products (yield curves not shown
0
F
= 46 g h/mol].
C2H OH
in Fig. 5).
5

2
68
Di COSIMO ET AL.
FIG. 6. Reaction network for ethanol conversion reactions.
These results and previous literature reports (4, 8, 20, 37, ciallyisobutanol(2-methyl-1-propanol) on MgyAlOx. Prod-
0
3
8) lead us to propose the reaction network described in uct yields (�i = ni /n , where ni is the moles of product i
T
0
T
Fig. 6. In this scheme, ethanol is converted on MgyAlOx and n is the total initial moles of reactants) are shown in
samples predominately via dehydrogenation, aldol cou- Fig. 7 for Mg1AlOx as a function of Wt/n , where W is
0
T
pling, and dehydration reactions. Acetaldehyde forms via the catalyst weight and t the reaction time. The local slope
ethanol dehydrogenation (step III). Aldol intermediates for each product in Fig. 7 gives its rate of formation at
formed from acetaldehyde condensation (step IV) de- a specific value of reactant conversion and contact time.
hydrate to form �,�-unsaturated aldehydes (crotonalde- The predominant reaction products are propionaldehyde
hyde), which rapidly hydrogenate to n-butyraldehyde (step (C3H6O), methyl-propyl ether (CH3OC3H7), isobutanol
VIII) using H2 formed in ethanol dehydrogenation steps. (iso-C4H9OH), dimethyl ether (CH3OCH3), propylene
This aldol intermediate can also decompose to acetone or (C3H6), di-n-propyl ether (C3H7OC3H7), and isobutyralde-
2
-propanol via reverse aldol coupling (steps X and XIV) hyde (iso-C4H8O). Other minor products are methyl
after an intramolecular hydride shift leading to the keto
form (step VII) (8). n-Butanol can form via direct con-
densation of ethanol (8) (step VI) or by hydrogenation
of n-butyraldehyde (step IX), but the latter pathway is
thermodynamically unfavored at the low H2 concentrations
prevalent during ethanol reactions. Ethylene is formed by
monomolecular dehydration of ethanol (step I). Diethyl-
ether is formed in bimolecular dehydration steps (step II),
and it can react further to form alkanes and alkenes
(
step XV) on acid sites, using pathways similar to those
reported for dimethylether and methanol conversion to
hydrocarbons. Ethylacetate forms via Tischenko-type re-
actions between aldehydic and alkoxide adsorbed species
(
step V) (39, 40).
. Reaction Pathways of 13CH3OH/1– C3H7OH Mixtures
12
4
on MgO, Al2O3, and MgyAlOx Samples
FIG. 7. Product distribution for reactions of 13CH3OH/1–12C3H7OH
mixtures. Product yields and methanol conversions as a function of
13
12
Isotopic tracer studies using CH3OH/1– C3H7OH mix-
tures were carried out in order to probe chain-growth path-
0
parameter Wt/n . Recirculating reactor data for Mg1AlOx catalyst [573 K,
T
1
01.3 kPa total pressure, W = 0.038 g, 2.7 kPa methanol, 1.4 kPa
ways involved in the synthesis of higher alcohols, espe- 1-propanol, balance He].

CONDENSATION REACTIONS OF ALCOHOLS ON MgyAlOx
269
propionate, butenes, 2-methyl-1pentanol (iso-C5H11OH),
TABLE 3
and 3-pentanone (C2H5(CO)C2H5).
13C-Distribution in Reactants and Products of 13
CH3OH–C3H7OH
On Mg1AlOx, propionaldehyde was the main reaction
product; dehydrogenation reactions occurred faster than
dehydration or chain growth. The nonzero initial slopes
for C3H6O, CH3OC3H7, CH3OCH3, C3H6, and C3H7OC3H7
show that these products form directly from the methanol
Reactions on Mg1AlOx
Number of 13C (% )
2
Compound
0
1
3
4
and 1-propanol reactants. Isobutyraldehyde yields, how- CH3OH
1.9
98.1
99.8
4.7
5.8
1.5
0.0
98.0
98.0
98.1
1.6
0.2
94.6
94.2
98.0
10.6
1.3
—
0.0
0.0
0.4
0.0
0.4
89.4
0.7
1.0
—
0.3
0.0
0.3
0.0
0.0
—
—
—
—
0.0
0.0
0.0
—-
0.0
—
C3H7OH
ever, show a zero initial slope. As in the case of 1-butanol
formation from ethanol (Fig. 5), the low but nonzero initial
slope for isobutanol formation from 1-propanol/methanol
mixtures (Fig. 7) suggests that the reaction involves par- CH OC H
7
C3H6O
iso-C4H9OH
iso-C4H8O
3
3
allel paths. In this case, such paths include the direct con- CH3OCH3
C3H7OC3H7
C3H6
0.0
0.0
densation of methanol-derived surface intermediates with
-propanol and/or with the propionaldehyde formed in pri-
1.0
1
mary 1-propanol dehydrogenation reactions.
Note. 573 K, 101.3 kPa total pressure, 2.7 kPa methanol, 1.4 kPa 1-
The 13C-content in 1-propanol and methanol and in the propanol, balance He, 2.8 h reaction time.
major products of their reactions was measured by mass
spectrometryafter chromatographicseparation usinga cap-
illary column. The measured isotopic contents in reaction
products formed on Mg1AlOx are shown in Table 3. The
1
1
2
3
3 7
C H OH (step IV). Propylene shows only about 1%
C enrichment, as expected from its formation via
12
results of these isotopic tracer studies (Table 3) confirmed monomolecular dehydration of 1- C H OH (step I). The
3
7
the reaction sequence described in Fig. 8 and previously in- small amounts of 2-methyl-1-pentanol formed by self-
12
ferred from the observed effects of residence time on prod- condensation of 1- C H OH (steps XIII, XIV, and XV)
3
7
uct yields (Fig. 7).
-Propanol and all reaction products derived di-
rectly from 1-propanol are predominantly unlabeled (Ta- shows that they involve 13CH OH as a reactant. Dimethyl-
also do not show 13C contents above natural abundance.
13
1
The presence of C in the other reaction products
3
13
ble 3). Propionaldehyde forms by dehydrogenation of ether molecules contain predominantly two C-atoms
- C3H7OH (step VIII) and shows no detectable C- ( CH O CH ), because they form via dehydration of
1
2
13
13
13
1
3 3
1
3
13
enrichment. Di-n-propyl ether is also mainly unlabeled,
3 1
CH OH (step II). The identity of the C intermedi-
and it forms by bimolecular dehydration reactions of 1- ate required is unclear, and adsorbed formyl or methoxy
FIG. 8. Reaction network for methanol/1-propanol reactions.

2
70
Di COSIMO ET AL.
species and gas-phase formaldehyde have been proposed
8). We did not detect formaldehyde among reaction prod-
(
13
ucts because of its high reactivity, but the presence of CO
among products suggests that CH3OH dehydrogenation
13
13
to H2 CO is likely to occur as an intermediate step in the
decarbonylation of methanol to 13CO (steps V and VI).
Some dimethylether molecules (about 10% , Table 3) con-
tain only one 13C; this is higher than expected from the
number of methanol precursors with C. It appears that un-
labeled methanol and singly labeled dimethylether can also
form via a 12C1 intermediate derived from C–C bond cleav-
age in aldol intermediates formed from two C3H6O, after
the latter undergo intramolecular hydrogen transfer (aldol-
keto isomerization) and reverse aldol reactions (steps XIII,
XVI, and XVII).
12
12
When products form via cross-coupling reactions, the po-
sition of the 1 C-atom can reveal mechanistic details not
available from the 13C-content alone. The most abundant
cross-coupling products were methyl-propyl ether, isobu-
tyraldehyde, and isobutanol. The single 13C in methyl-
3
FIG. 9. Mass spectra of (a) unlabeled isobutanol and (b) isobutanol
obtained during reaction of CH3OH/1– C3H7OH mixtures on Mg1AlOx
1
3
12
propyl ether was located at the methyl carbon and di- [573 K, 101.3 kPa total pressure, 2.7 kPa methanol, 1.4 kPa 1-propanol,
balance He, 2.8 h reaction time].
rectly bonded to the ether linkage, as expected from
the bimolecular dehydration of 1-propanol and methanol
+
(
step III).
ter the transfer of two hydrogen atoms (CH3–O H2) (42).
Isobutyraldehyde and isobutanol contained predomi- The peak at 31 amu is less intense than in linear alcohols
nantly one 13C (94.2% , 94.6% ; Table 3). Their similar C- because the isopropyl fragment formed from isopropanol
contents suggest that they arise from a common reaction (43 amu) is branched and leads to a much more stable pos-
intermediate, which in turn forms via condensation reac- itively charged fragment. The fragment at 15 amu in the
13
13
tions of one C1 species derived from CH3OH and one C3 unlabeled isobutanol spectrum corresponds to the methyl
12
species derived from 1- C3H7OH (e.g., steps IX and X, XI group.
or step VII). The amounts of unlabeled isobutyraldehyde
Figure 9b shows the mass spectrum for isobutanol formed
and isobutanol detected are very similar, and they corre- from 13CH3OH/1– C3H7OH mixtures on Mg1AlOx and
12
12
spond to those expected from the presence of some C in containing mostly singly labeled molecules. This spectrum
methanol and some 13C in 1-propanol, both of which arise contains peaks at 31 and 33 amu, as in the unlabeled spec-
from reverse aldol reactions.
trum, and very small peaks at 32 and 34 amu, which re-
The fragmentation pattern of isobutyraldehyde products flect the presence of 13C in the OH-containing fragments.
shows that the 13C is located predominantly in the CH3 Isopropyl fragments in the isobutanol products appear at
group (16 amu), with a much smaller but detectable amount 44 amu, consistent with the presence of a single 13C in
in the CH == O (29, 30 amu) fragment. The presence of this fragment. The spectrum in Fig. 9b also shows that
13
C in the aldehydic fragment confirms the formation of the methyl group of isopropyl fragments contains about
13
some isobutyraldehyde via dehydration of intermediates 50% of C (peak at m/z = 16). These results show that
(
(
step XIX) after isomerization from aldol to keto forms the 13C in labeled isobutanol is located in the methyl
13
step XVIII). These pathways involve the retention of the group of CH3CH(CH3)–CH2OH species. In contrast with
13
oxygen atom in CH3OH within the aldehyde condensa- the findings for iso-C4H8O fragmentation, the amount of
13
tion product; they have been shown to occur during the CH3CH(CH3) CH2OH formed in step XX is negligible
synthesis of branched alcohols from CO and H2 on Cs– (32 or 34 amu peaks in Fig. 9b). The absence of significant
Cu/ZnO (41) and K–CuzMg5CeOx (8, 21).
amounts of iso-C4H9OH labeled at the �-C suggests that
The location of the 13C-atom in isobutanol can be mea- cross-coupling reactions of methanol/1-propanol mixtures
sured by comparing the spectra for unlabeled isobutanol on MgyAlOx samples occur mainly via a normal Guerbet-
(
(
Fig. 9a) with that of the isobutanol formed during reaction type condensation (43, 44) (step VII or IX, X, and XI),
Fig. 9b). Unlabeled isobutanol shows a peak at 31 amu, and that the contribution of the aldol-keto equilibrium
+
corresponding to (CH2OH) fragments formed by cleav- (steps XVIII, XIX, and XX) to iso-C4H9OH formation is
age of the C–C bond in the �-position to the oxygen atom negligible, probably because of the consecutive nature of
(
42). This peak gives rise to another peak at 33 amu af- step XX.

CONDENSATION REACTIONS OF ALCOHOLS ON MgyAlOx
271
FIG. 10. Product distribution for ethanol conversion reactions on MgO, Al2O3, and MgyAlOx samples. Initial formation rates as a function of
the sample composition: (a) products of dehydrogenation and condensation reactions, (b) products of dehydration reactions. (573 K, 101.3 kPa total
0
�
pressure, N2/ethanol = 10, X
= 5% ).
C2H OH
5
5
. Sample Composition, Surface Base Property,
and Catalytic Performance
increases both dehydrogenation and condensation rates,
but dehydrogenation (to form C2H4O or C3H6O) becomes
the predominant reaction on Mg-rich samples (Figs. 10a
C2H5OH and CH3OH/C3H7OH mixtures undergo dehy- and 11a). The required dehydrogenation steps occur much
drogenation, dehydration, and condensation reactions on faster than chain growth reactions, which form higher al-
MgyAlOx samples. C2H5OH and CH3OH/C3H7OH conver- cohols and other oxygenates. For C H OH reactions, n-
2
5
sion rates were measured for MgO, Al2O3, and MgyAlOx C H OH is the main coupling product and it forms at the
4
9
samples at 573 K. Figures 10 and 11 show the product highest rate on Mg AlO and Mg AlO (Fig. 10a). Similarly,
9
x
1
x
distribution for dehydrogenation and coupling reactions iso-C H OH is the predominant condensation product of
4
9
(
Figs. 10a and 11a) and for dehydration reactions (Figs. 10b CH OH/C H OH cross-coupling reactions and shows max-
3 3 7
and 11b). A comparison of the data in Figs. 10 and 11 sug- imum rates on Mg AlO and Mg AlO (Fig. 11a). Even
9
x
0.5
x
gests a similar effect of chemical composition on the prod- though dehydration rates to olefins and ethers increase
uct distribution from C2H5OH and CH3OH/C3H7OH re- with increasing Al content (Figs. 10b and 11b), none of
actants. While MgO and Mg-rich MgyAlOx mixed oxides the samples was very active for olefin synthesis (C H or
2
4
behave as solid basic catalysts and convert short-chain lin- C H ) and ethers were the predominant dehydration prod-
3
6
ear alcohols predominantly into aldehydes or higher alco- ucts (C H OC H , Fig. 10b, or CH OCH , CH OC H ,
2
5
2
5
3
3
3
3
7
hols, Al-rich MgyAlOx samplesand Al2O3 mainlydehydrate and C H OC H , Fig. 11b). Pure Al O is about two or-
3
7
3
7
2
3
alcohols to ethers and to a lesser extent to olefins.
y
ders of magnitude more active than any of the Mg AlOx
Pure MgO is active for dehydrogenation and coupling re- samples in alcohol dehydration reactions, but it does
actions but shows negligible dehydration activity. The intro- not form condensation products such as n-C H OH and
4
9
duction of small amounts of Al3 ions into the MgO matrix iso-C H OH.
+
4
9
FIG. 11. Product distribution for methanol/1-propanol conversion reactions on MgO, Al2O3, and MgyAlOx samples. Initial formation rates as
a function of the sample composition: (a) products of dehydrogenation and condensation reactions, (b) products of dehydration reactions. (573 K,
1
01.3 kPa total pressure, W = 0.038 g, 2.7 kPa methanol, 1.4 kPa 1-propanol, balance He.)

2
72
Di COSIMO ET AL.
In summary, the data shown in Figs. 10 and 11 show acid–Brønsted base pair site of balanced strength (like the
3+
2�
that both the total reaction rates and the product selec- Al –O pairs in Al2O3). This mechanism leads to the
tivity for C2H5OH or CH3OH/C3H7OH conversion reac- formation of olefins without involving ionic intermediates
tions depend on the chemical composition of the MgyAlOx (45, 46). In contrast, E1cB pathways involve surface alkoxy
samples. In turn, the chemical composition affects the acid– intermediate and both strongly basic sites and weak Lewis
2+
2�
base properties of MgyAlOx samples by modifying surface acid sites (e.g., Mg –O pairs in MgO) (46). Formation of
acid and base site densities and the distribution of strength the olefin takes place by �-H elimination from carbanion
for such sites. The data shown in Table 2 and the infrared intermediates.
spectra of adsorbed CO2 reported elsewhere (20) show that
Pure MgO and Mg-rich samples contain strong basic sites
MgyAlOx catalysts exhibit three types of basic sites, isolated (Table 2) consisting of O2 anions, which catalyze alcohol
�
2�
2+ 2�
O
ions, Mg –O pairs, and OH groups. Pure MgO con- dehydration to olefins via E1cB pathways. The rate of al-
2�
tains the highest concentration of strongly basic O , but cohol dehydration to olefins increases with increasing Al
as the Al content increases, the relative concentrations of content on MgyAlOx catalysts, because the incorporation
the other weaker basic sites increases; the density of acid of the more electronegative Al3 ions increases the den-
+
sitesconcurrentlyincreases. Lewisacid sitesprovided byco- sity and strength of acid sites (Table 2) and thus decreases
ordinatively unsaturated Mg2 and Al centers bind basic the activation energy for �-H abstraction (47). Aluminum-
+
3+
molecules more strongly than Brønsted acid sites provided rich MgyAlOx samples contain a relatively high density
3+
2�
by OH groups at the surface of MgyAlOx samples (Fig. 1). of Al –O pairs, a higher number of acidic sites, and a
The results of isotopic switch measurements (Table 2) also lower density of basic sites (Table 2). Therefore, alcohol de-
3+
2�
show that the contribution from Al –O pairs to Al2O3 hydration to C2H4 or C3H6 on these samples is more likely
surface basicity is not significant. The rest of this paper at- to proceed through an E2 elimination mechanism. Pure alu-
2
3
tempts to bring together into a cogent structure–function mina, a typical E2 catalyst, is 10 –10 times more active in
relationship the effects of sample composition on surface olefin synthesis from ethanol or 1-propanol than MgyAlOx
acid–base properties and on the rates of each major reac- samples.
tion pathway involved in alcohol conversion reactions on
MgyAlOx samples.
Both ethanol and methanol/1-propanol mixtures lead to
ether/olefin molar ratios greater than one over the entire
Dehydration of alcohols to olefins on MgyAlOx samples range of composition of MgyAlOx samples, and these ra-
can proceed through the two elimination mechanisms (E2 tios increase with increasing Al content. Dehydration of
and E1cB) shown in Scheme 1. The E2 elimination is a primary alcohols to olefins is usually the predominant reac-
single-step concerted mechanism, in which the OH group tion on strongly acidic oxides, which contain small, highly
and the �-H are simultaneously abstracted by a Lewis charged cations (38), whereas ether formation takes place
on less acidic oxides (37, 48, 49). Alcohol dehydration to
olefins has a higher activation energy than the competitive
dehydration to ethers, and it is favored at high reaction tem-
peratures. MgyAlOx samples are weakly acidic (Table 2),
and alcohol reactions were studied at relatively low temper-
atures; both of these contribute to the predominant forma-
tion of ethers on the Al2O3 and Al-rich MgyAlOx samples
of this study.
Ether formation is a second-order reaction that involves
the adsorption of two alcohol molecules on neighboring ac-
tive sites offering different acid–base properties (20, 37). As
shown in Scheme 2, one alcohol molecule adsorbs through
the oxygen of the OH group on a Lewis acid site, which
is also involved in olefin formation via an E2 mechanism
(
37). The other alcohol molecule adsorbs on a basic site
via the hydroxylic hydrogen and forms an incipient sur-
face alkoxide. On our MgyAlOx samples, the active acid
sites for ether formation from primary alcohols are proba-
bly the Al3 cations or even the acidic OH groups, as also
+
proposed for zeolites (50), whereas the basic sites are the
neighboring O2 ions. The abundance of surface Al –O
2�
�
3+
pairs accounts for the high ether formation rates observed
SCHEME 1. Propylene formation mechanisms in methanol/1-pro-
panol conversion reactions.
on Al2O3 and Al-rich MgyAlOx samples.

CONDENSATION REACTIONS OF ALCOHOLS ON MgyAlOx
273
gated Lewis acid pairs to stabilize the alkoxy intermediates.
As a result, pure MgO shows lower dehydrogenation rates
than Mg-rich MgyAlOx catalysts.
The compositional trends observed for H2–D2 equilibra-
tion rates (Fig. 3) and alcohol dehydrogenation (Figs. 10a
and 11a) on MgO and Mg-rich samples are similar, sug-
gesting that both reactions involve rate-determining steps
requiring similar surface sites. In contrast, on Al-rich sam-
ples the aldehyde formation rate increases with increasing
Al content, while H2–D2 equilibration rates remain rela-
tively independent of composition. These results suggest
that H2(D2) dissociation and O–H or C–H bond activation
benefit from the presence of Al–O–Mg sites on Mg-rich
MgyAlOx samples, but that on more electronegative
Al-rich catalysts only the C–H bond activation can be
SCHEME 2. Methyl-propyl ether formation mechanism in methanol/
-propanol conversion reactions.
1
3+
2�
Only the symmetrical ether (C2H5OC2H5) can form from efficiently performed. Its is probable that Al –O pairs
ethanol, whereas three ethers, two symmetrical (CH3OCH3 contribute to aldehyde formation on Al-rich MgyAlOx
and C3H7OC3H7) and one asymmetrical (CH3OC3H7), can samples. For instance, it has been reported that on Al2O3
form from CH3OH + C3H7OH mixtures. Ether formation activated at low temperatures, alcohol dehydrogenation
involves adjacent adsorption of two alcohol molecules, and is largely determined by the density of structural defects,
reaction rates and selectivities depend sensitively on the specifically Al vacancies (53). Al-rich MgyAlOx samples
molecular structure of the alcohols involved. As the alco- contain a high density of Al–O pairs with unsaturated
hol carbon chain grows, steric factors cause the ether pro- surface aluminum cations, which can form aldehydes via
duction to decrease (51). This explains the rate formation the abstraction of the �-hydrogen as a hydride, as reported
order CH3OCH3 > C2H5OC2H5 � C3H7OC3H7 (Figs. 10b for ZnO and Cr2O3 (54).
and 11b).
The dependence of dehydrogenation and condensation
In summary, dehydration reactions to either olefins or rateson catalyst composition (Figs. 10a and 11a) are qualita-
ethers on MgyAlOx samples appear to occur on Lewis acid– tively similar, suggesting that both reactions require com-
2+
2�
3+
2�
Brønsted base sites (Mg –O or Al –O pairs) or on mon intermediates and similar acid–base surface proper-
weakly basic Brønsted sites (OH groups), both of which are ties. Condensation reactions also proceed via a combined
abundant in Al-rich samples and on Al2O3.
mechanism in which the role of Al–O pairs is to increase
Dehydrogenation of alcohols to aldehydes (C2H4O or the rate of formation of alkoxide intermediates and of the
C3H6O) is a typical base-catalyzed reaction (38, 45, 52), primary aldehyde products (Scheme 4). Pure alumina, how-
the site requirements for which have been disussed previ- ever, does not form n-butanol or isobutanol, suggesting that
2+
ously for ethanol dehydrogenation reactions (20). Mg –
2�
O
pairs with moderately basic oxygens play an important
role, because weak Lewis acid–strong Brønsted base site
pairs are required for hydrogen abstraction steps leading to
alkoxy intermediates. As shown in Scheme 3, dehydrogena-
2+
2�
tion starts by alcohol chemisorption on Mg –O site pairs,
which cleave O–H bonds to form surface alkoxide interme-
diates bound to the Mg2 acid center. The �-hydrogen in
the alkoxide group is then abstracted by a neighboring basic
site in order to form adsorbed aldehydes. Thus, synthesis of
C2H4O or C3H6O, from ethanol or 1-propanol, respectively,
prevails on basic oxides such as Mg-rich MgyAlOx samples
+
2+
2�
with abundant Mg –O site pairs (Table 2). These sam-
ples show Al surface enrichment, and the surface Al species
provide additional Lewis centers to stabilize alkoxide inter-
mediates. These results suggest that Al–O–Mg species may
be particularly effective and abundant sites for dehydro-
genation reactions on Mg-rich samples. In contrast, pure
2�
MgO contains a high density of isolated O ions, which
are strongly basic, but which lack sufficiently strong conju- 1-propanol conversion reactions.
SCHEME 3. Propionaldehyde formation mechanism in methanol/

2
74
Di COSIMO ET AL.
SCHEME 4. Isobutyraldehyde and isobutanol formation mechanism in methanol/1-propanol conversion reactions.
Al–O pairs, which can activate the C–H bond, do not pro-
mote the formation of C–C bonds required for aldol con-
densation reactions.
CONCLUSIONS
The rates and product distributions for C2H5OH or
Reactions leading to n-C4H9OH (or n-C4H8O) and iso- CH3OH/C3H7OH reactions are strongly influenced by the
C4H9OH (or iso-C4H8O) are much slower than ethanol and composition of MgyAlOx samples prepared by decomposi-
1
-propanol dehydrogenation reactions (Figs. 10a and 11a). tion of hydrotalcite precursors. The rate of alcohol dehydra-
This reflects the bimolecular nature of condensation reac- tion to ethers and olefins increases markedly with increas-
tions, which require not only the proper acid–base pair sites ing Al content. Al-rich MgyAlOx samples contain a high
3+
2�
but also a particular surface atom arrangement that can density of Al –O site pairs and of moderate-strength
accommodate vicinal adsorbed species. This may also ex- basic sites, the combination of which promote the forma-
plain why the rate of formation of iso-C4H9OH via cross- tion of ethylene or propylene from primary alcohols via
coupling reactions of CH3OH and 1-C3H7OH (Fig. 11a) is E2 elimination pathways. The parallel dehydration path-
lower than the rate of formation of n-C4H9OH in the self- ways to form ethers involve two adjacent alcohol-derived
coupling reaction of ethanol (Fig. 10a), because the former adsorbed species interacting with neighboring acid–base
requires the adjacent adsorption of C1 species and bulkier pair sites. On MgyAlOx samples, the acid sites required for
C3-derived species. Furthermore, substitution of the �-C ether formation probably consist of Al3 cations, while the
+
2�
in the aldehydic intermediates of aldol condensation re- basic sites are provided by the corresponding O anions.
3+
2�
actions (C3H6O in Scheme 4) defines the stability of the The abundance of surface Al –O pairs accounts for the
carbanion formed and also explains the higher reactivity high ether formation rates observed on Al2O3 and Al-rich
toward condensation of C2H4O compared to C3H6O. Also, MgyAlOx samples.
as stated previously, formaldehyde was not detected among
The dehydrogenation of alcohols to aldehydes (C2H4O
dehydrogenation products. Noller et al. (55) have shown or C3H6O) requires weak Lewis acid–strong Brønsted base
using temperature-programmed reactions of pre-adsorbed site pairs in order to form and stabilize alkoxide inter-
methanol that on basic oxides, such as MgO, formaldehyde mediates. The synthesis of C2H4O or C3H6O is favored
is decomposed to CO and H2 immediately upon forma- on Mg-rich MgyAlOx samples because these samples con-
3
+
tion. In contrast, formaldehyde desorbs before decompo- tain a much larger number of properly positioned Al
2+
2�
sition on more acidic oxides. Therefore, on strongly basic Lewis acid sites and Mg –O basic pairs, which cata-
oxide surfaces formaldehyde-type intermediates would be lyze C–H and O–H activation steps in alcohol dehydro-
present in much lower concentrations, a situation that disfa- genation reactions. Mixed Al–O–Mg sites are also active
vors the cross-coupling reactions leading to isobutanol and for the mechanistically related heterolytic dissociation of
isobutyraldehyde (step IX in Fig. 8).
H2 (and D2) molecules in H2–D2 isotopic equilibration

CONDENSATION REACTIONS OF ALCOHOLS ON MgyAlOx
275
reactions, which are also favored on Mg-rich MgyAlOx 18. Schaper, H., Berg-Slot, J. J., and Stork, W. H. J., Appl. Catal. 54, 79
1989).
(
samples.
Aldol condensation reactions on MgyAlOx samples in-
volve also the formation of a carbanion intermediate on
Lewis acid–strong Brønsted base pair sites, and conse-
1
2
9. Corma, A., Forn e´ s, V., and Rey, F., J. Catal. 148, 205 (1994).
0. Di Cosimo, J. I., Diez, V. K., Xu, M., Iglesia, E., and Apesteguia, C. R.,
J. Catal. 178, 499 (1998).
2
1. Gin e´ s, M. J. L., and Iglesia, E., J. Catal. 176, 155 (1998).
quently they are also favored on Mg-rich samples. The for- 22. Price, G. L., and Iglesia, E., Ind. Eng. Chem. 28, 839 (1989).
2
2
2
3. Reichle, W. T., Kang, S. Y., and Everhardt, D. S., J. Catal. 101, 352
mation rates of n-C4H8O and n-C4H9OH or iso-C4H9OH
and iso-C4H8O are slower than dehydrogenation reactions
because in bimolecular aldol condensations not only the
acid–base surface properties are relevant but also the sur-
face atom arrangement needed to adsorb adjacent species
and the steric factors related to the chemical nature of the
parent alcohols and aldehydic intermediates.
(
1986).
4. Nakatsuka, T., Kawasaki, H., Yamashita, S., and Kohjiya, S., Bull.
Chem. Soc. Jpn. 52, 2449 (1979).
5. Mc Kenzie, A. L., Fishel, C. T., and Davis, R. J., J. Catal. 138, 547
(1992).
6. Shen, J., Tu, M., and Hu, C., J. Solid State Chem. 137, 295 (1998).
7. Eley, D. D., and Zammit, M. A., J. Catal. 21, 366 (1971).
8. Eley, D. D., and Zammit, M. A., J. Catal. 21, 377 (1971).
9. Tsuchiya, S., J. Catal. 20, 1 (1971).
2
2
2
2
3
ACKNOWLEDGMENTS
0. Chelseske, F. J., Wallace, W. E., and Hall, W. K., J. Phys. Chem. 63, 505
(
1963).
This work was supported by the U.S. Department of Energy, the
National Science Foundation (U.S.A.), and CONICET (Argentina). The
authors thank Dr. M. Xu for his technical assistance with the CO2 isotopic
switch experiments. M. J. L. Ginesacknowledgesthe Universidad Nacional
del Litoral, Santa Fe, Argentina, for a fellowship for post-doctoral studies
at the University of California at Berkeley.
3
3
3
1. Steele, L. R., and Copper, W. H., J. Catal. 6, 147 (1966).
2. Lunsford, J. H., J. Phys. Chem. 66, 2591 (1962).
3. Boudart, M., Delbouille, A., Derouane, E. G., Indovina, V., and
Walters, A. B., J. Am. Chem Soc. 94, 6622 (1972).
4. Hattori, H., Tanaka, Y., and Tanabe, K., J. Am. Chem Soc. 98, 4652
3
(
1976).
35. Nakano, Y., Yamaguchi, T., and Tanabe, K., J. Catal. 80, 307 (1983).
REFERENCES
36. Kazansky, V. B., Borovkov, V. Y., and Kustov, L. M., in “Proceed-
ings 8th International Congress on Catalysis, Berlin, 1984,” III-3.
Dechema, Frankfurt-am-Main, 1984.
37. Jain, J. R., and Pillai, C. N., J. Catal. 9, 322 (1967).
38. Kibby, C. L., and Hall, W. K., J. Catal. 29, 144 (1973).
1
. Schneider, M., Kochloefl, K., and Bock, O., Eur. Patent Appl. 152, 809,
985.
. Sugier, A., and Freund, E., U.S. Patent 4,112,110, 1978.
1
2
3
. Di Cosimo, J. I., Marchi, A. J., and Apestegu ´ı a, C. R., J. Catal. 134, 594 39. Takezawa, N., Hanamaki, C., and Kobayashi, H., J. Catal. 38, 101
(
1992).
(1975).
4
5
6
7
8
9
. Nunan, J. G., Bogdan, C. E., Klier, K., Smith, K. J., Young, C., and
Herman, R. G., J. Catal. 116, 195 (1989).
. Forzatti, P., Tronconi, E., and Pasquon, I., Catal. Rev. Sci. Eng. 33, 109
40. Smith, K. J., and Anderson, R. B., Can. J. Chem. Eng. 61, 40 (1983).
41. Nunan, J. G., Bogdan, C. E., Herman, R. G., and Klier, K., Catal. Lett.
2, 49 (1989).
42. Beynon, J. H., Saunders, R. A., and Williams, A. E., in “The Mass
Spectra of Organic Molecules.” Elsevier, New York, 1968.
43. Ueda, W., Ohshida, T., Kuwabara, T., and Morikawa, Y., Catal. Lett.
12, 97 (1992).
44. Ueda, W., Ohowa, H., Iwasaki, K., Kubawara, T., Ohshida, T., and
Morikawa, Y., in “Acid–Base Catalysis II.” Kodansha-Elsevier, Tokyo,
1994.
45. Tanabe, K., Misono, M., Ono, Y., and Hatori, H., in “New Solid Acid
and Bases.” Kodansha-Elsevier, Tokyo, 1989.
(
1991).
. Apestegu ´ı a, C. R., Soled, S. L., and Miseo, S., U.S. Patent 5,387,570,
995.
. Apesteguia, C. R., De Rites, B., Miseo, S., and Soled, S. L., Catal. Lett.
4, 1 (1997).
1
4
. Xu, M., Gin e´ s, M. J. L., Hilmen, A., Stephens, B. L., and Iglesia, E., J.
Catal. 171, 130 (1997).
. Xu, M., and Iglesia, E., Catal. Lett. 51, 47 (1998).
1
0. Bowker, M., Hadden, R. A., Houghton, H., Hyland, J. N. K., and
Waugh, K. C., J. Catal. 109, 263 (1988).
1. Neophytides, S. G., Marchi, A. J., and Froment, G. F., Appl. Catal. A
46. Gervasini, A., Fenyvesi, J., and Auroux, A., Catal. Lett. 43, 219 (1997).
47. Parrott, S. L., Rogers, J. W., and White, J. M., Appl. Surf. Sci. 1, 443
(1978).
1
8
6, 45 (1992).
1
1
2. Reichle, W. T., J. Catal. 94, 547 (1985).
48. Padmanabhan, V. R., and Eastburn, F. J., J. Catal. 24, 88 (1972).
3. Rao, K. K., Gravelle, M., Sanchez Valente, J., and Figueras, F., J. Catal. 49. Luy, J. C., and Parera, J. M., Appl. Catal. 26, 295 (1986).
73, 115 (1998). 50. Jacobs, P. A., Tielen, M., and Uytterhoeven, J. B., J. Catal. 50, 98 (1977).
4. Di Cosimo, J. I., Diez, V. K., and Apestegu ı´ a, C. R., Appl. Clay Sci. 13, 51. Simonik, J., and Pines, H., J. Catal. 24, 211 (1972).
1
1
4
33 (1998).
52. Canesson, P., and Blanchard, M., J. Catal. 42, 205 (1979).
53. Wang, J. A., Bokhimi, X., Novaro, O., Lopez, T., Tzompantzi, F.,
Gomez, R., Navarrete, J., Llanos, M. E., and Lopez-Salinas, E., J.
Molec. Catal. 137, 239 (1999).
1
1
5. Velu, S., and Swamy, C. S., Appl. Catal. 119, 241 (1994).
6. Corma, A., Iborra, S., Primo, J., and Rey, F., Appl. Catal. 114, 215
(
1994).
1
7. Kumbhar, P., Sanchez Valente, J., and Figueras, F., J. Chem. Soc. Chem. 54. Burwell, R. L., Catal. Rev. 7, 25 (1972).
Commun., 1091 (1998). 55. Noller, H., and Ritter, G., J. Chem. Soc. Faraday Trans. I 80, 275 (1984).