On the chemistry of ethanol on basic oxides: Revising mechanisms and intermediates in the lebedev and guerbet reactions

Article abstract of DOI:10.1002/cssc.201402632

A common way to convert ethanol into chemicals is by upgrading it over oxide catalysts with basic features; this method makes it possible to obtain important chemicals such as 1-butanol (Guerbet reaction) and 1,3-butadiene (Lebedev reaction). Despite their long history in chemistry, the details of the close inter-relationship of these reactions have yet to be discussed properly. Our present study focuses on reactivity tests, in situ diffuse reflectance infrared Fourier transform spectroscopy, MS analysis, and theoretical modeling. We used MgO as a reference catalyst with pure basic features to explore ethanol conversion from its very early stages. Based on the obtained results, we formulate a new mechanistic theory able to explain not only our results but also most of the scientific literature on Lebedev and Guerbet chemistry. This provides a rational description of the intermediates shared by the two reaction pathways as well as an innovative perspective on the catalyst requirements to direct the reaction pathway toward 1-butanol or butadiene.

Full text of DOI:10.1002/cssc.201402632

CHEMSUSCHEM
FULL PAPERS
DOI: 10.1002/cssc.201402632
On the Chemistry of Ethanol on Basic Oxides: Revising
Mechanisms and Intermediates in the Lebedev and
Guerbet reactions
Alessandro Chieregato,[a, c] Juliana Velasquez Ochoa,[a] Claudia Bandinelli,[a, d]
Giuseppe Fornasari,[a] Fabrizio Cavani,*[a, c, d] and Massimo Mella*[b]
A common way to convert ethanol into chemicals is by up-
grading it over oxide catalysts with basic features; this method
makes it possible to obtain important chemicals such as 1-bu-
tanol (Guerbet reaction) and 1,3-butadiene (Lebedev reaction).
Despite their long history in chemistry, the details of the close
inter-relationship of these reactions have yet to be discussed
properly. Our present study focuses on reactivity tests, in situ
diffuse reflectance infrared Fourier transform spectroscopy, MS
analysis, and theoretical modeling. We used MgO as a reference
catalyst with pure basic features to explore ethanol conversion
from its very early stages. Based on the obtained results, we
formulate a new mechanistic theory able to explain not only
our results but also most of the scientific literature on Lebedev
and Guerbet chemistry. This provides a rational description of
the intermediates shared by the two reaction pathways as well
as an innovative perspective on the catalyst requirements to
direct the reaction pathway toward 1-butanol or butadiene.
Introduction
Over the past decade, applications of ethanol as a platform
chemical have aroused new interest because of both environ-
mental concerns connected with petrochemical processes and
new economic opportunities seen in bio-based feedstocks.[1–5]
Indeed, bio-ethanol has been demonstrated to be a promising
and “green” reactant for both biofuels and biochemicals, par-
ticularly if it is derived from non-food crops and lignocellulosic
materials (second-generation biomass).[6–8] Despite the long
history of ethanol in the chemical industry, the mechanism
behind its catalytic upgrading on mixed oxide catalysts with
basic features is still a subject of debate, as shown by the pres-
ence of several alternative mechanisms proposed in the litera-
ture.[1,9–20] Ethanol can react on these materials to form various
important chemicals, among which are butadiene (through the
Lebedev reaction on bifunctional acid–base catalysts) and 1-
butanol (by the Guerbet route).
We reviewed the literature on the pathways proposed to
justify the mentioned products, and it is noticeable that they
share many features and a few key intermediates. We analysed
the nature of the catalysts involved, and it emerges that a key
role is played in the product formation by basic or dehydro-
genating components such as MgO, CaO, Mg-Al-O (hydrotal-
cites), ZnO, Cr2O3, Ta2O5 and NiO.[9–19] However, there is still
a lack of detail on the specific role played by each catalyst
component at every step along the reaction pathway. For in-
stance, MgO is a good catalyst for the Guerbet reaction and is
also often present in catalysts for the Lebedev process; it was
claimed as a major component of the industrial catalysts used
in the 1940s to produce butadiene.[21] A few Mg-based cata-
lysts reported in the literature are listed in Table S1; these data
stress that typical Guerbet and Lebedev products or intermedi-
ates have been reported in the same downstream molecular
pool and their respective distributions are related to reaction
conditions and catalyst features.
[a] A. Chieregato,+ J. Velasquez Ochoa,+ C. Bandinelli, Prof. G. Fornasari,
Prof. F. Cavani
Dipartimento di Chimica Industriale “Toso Montanari”
Alma Mater Studiorum Universitꢀ di Bologna
Viale del Risorgimento, 4, 40136 Bologna (Italy)
E-mail: fabrizio.cavani@unibo.it
[b] Dr. M. Mella
The reaction network generally accepted in the literature for
both reactions considers acetaldehyde as the first product
formed from ethanol and its subsequent condensation to 3-hy-
droxybutanal (acetaldol; Scheme 1).
Dipartimento di Scienze ed Alta Tecnologia
Universitꢀ degli Studi dell’Insubria
Via Valleggio 9, 22100 Como (Italy)
E-mail: massimo.mella@uninsubria.it
[c] A. Chieregato,+ Prof. F. Cavani
Hydrogenation is believed to occur by H-transfer from etha-
nol [Meerwein–Ponndorf–Verley (MPV) reaction], even though
the hydrogen generated in situ might also play a role. Howev-
er, other mechanistic options have been reported: acetaldol
may undergo reduction by ethanol to form 1,3-butanediol,[19]
the dehydration of which possibly leads to either crotyl alcohol
or 3-butene-2-ol. Notably, both compounds can be further de-
hydrated into butadiene. Additionally, a few authors have hy-
CIRI Energia e Ambiente
Alma Mater Studiorum Universitꢀ di Bologna
[d] C. Bandinelli, Prof. F. Cavani
Consorzio INSTM, Firenze
Research Unit of Bologna
[+] These authors contributed equally.
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cssc.201402632.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2014, 7, 1 – 13 1
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
imise the yield of C4 compounds. In this situation, however,
poor carbon balance may be observed during the initial period
of reactivity because of severe heavy-compound deposition on
the catalyst surface.[23,24]
The catalytic behaviour of ethanol on MgO at 2508C for dif-
ferent residence times is depicted in Figure 1. At contact times
Scheme 1. Generally accepted pathway for ethanol transformation on basic
oxides: Lebedev and Guerbet routes.
pothesised that the formation of C4 compounds might be the
result of a reaction between two molecules of ethanol or one
molecule of ethanol and one of acetaldehyde.[9,22]
Despite the sometimes detailed studies and the many tech-
niques used, a definitive consensus on the most likely pathway
has yet to be reached. Recently, Meunier et al.[20] managed to
rule out aldol condensation as the main path to 1-butanol for-
mation by reactivity tests on hydroxyapatite catalysts and ther-
modynamic calculations, at least on heterogeneous catalysts at
high temperature (350–4108C).
In this paper, we report a mechanistic study performed on
MgO as a model catalyst (i.e., one that shows pure basic prop-
erties) aimed to elucidate the key intermediates of ethanol
conversion on mixed oxide catalysts with basic features. The
existing correlation between the Lebedev and the Guerbet re-
actions was further analysed with the goal to gain a more
complete view of the tangled mechanisms. Reactivity tests and
in situ diffuse reflectance infrared Fourier transform spectrosco-
py (DRIFTS) and MS were combined with computational stud-
ies, which allowed us to indicate, for the first time, additional
details that led toward an alternative view of the complete
mechanism for the processes involved. In particular, we dis-
cussed the key factors that govern the product distribution
and reaction intermediates for the interconnected and com-
petitive reactions that lead, alternatively, to 1-butanol/butenes
or to butadiene.
Figure 1. Ethanol conversion at 2508C on MgO. Symbols: ethanol conversion
~
^
*
(
); yields: acetaldehyde (+), 1-butanol ( ), crotyl alcohol ( ), 3-buten-1-ol
&
(ꢁ), others ( ). Others: 2-pentanol, 2-butanol and 3-buten-2-ol. Figure S1
shows the same results plotted as selectivity towards the products as a func-
tion of ethanol conversion.
lower than 0.4 s (calculated at the reaction T) the main product
detected was acetaldehyde with traces of 1-butanol. At higher
contact times, the formation of 1-butanol, 2-buten-1-ol (crotyl
alcohol) and 3-buten-1-ol occurred, which was accompanied
by the formation of other trace compounds. Neither C4 alde-
hydes (3-hydroxybutanal, crotonaldehyde and butyraldehyde)
nor C4 olefins (butenes and butadiene) were detected.
To confirm the lack of C4 aldehydes, reactivity experiments
were performed at 2508C with 0.6 s residence time for 24 h;
the collected products were analysed by GC–MS and ESI-MS
and presented no trace of these compounds. Indeed, the ab-
sence of 3-hydroxybutanal and crotonaldehyde poses impor-
tant questions on the reaction mechanism as these are the
generally accepted intermediates for C4-compound formation
from ethanol. As the lack of C4 aldehydes might be also attrib-
uted to a fast conversion to crotyl alcohol (by MPV reduction)
and then 1-butanol, these results are not exhaustive or able to
rule out their role in the mechanism; further tests were thus
deemed necessary and performed (vide infra). Nevertheless, it
is important to highlight the direct formation of 1-butanol as
a kinetic primary product, which goes against the mechanism
accepted widely that postulates Guerbet alcohol formation by
the in situ reduction of crotonaldehyde. This unexpected fea-
ture was also recently pointed out by other authors.[1,20]
Results and Discussion
Ethanol reactivity
After an initial screening to determine the dependence of cata-
lyst performance on temperature, 250 and 4008C were found
to be the most representative temperatures (T) to evaluate the
reaction pathways. Indeed, the lowest T value seemed to be
the minimum needed to observe a detectable conversion of
ethanol under the reaction conditions used, whereas MgO
showed a good activity and a relatively limited yield to heavy
products at higher T. As we wanted to investigate the very
early stages of the reaction, we scanned the catalytic behav-
iour through the first hour of reaction and at uncommonly low
contact times in comparison to those normally applied to max-
The products of ethanol transformation at 4008C are shown
in Figure 2. In order of abundance, ethylene was formed in
large amounts, followed by acetaldehyde and 1-butanol, and
finally butadiene as a consecutive product (additional informa-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&
2
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Figure 2. Ethanol conversion at 4008C on MgO. Symbols: ethanol conversion
Figure 3. Crotyl alcohol conversion at 4008C on MgO. Symbols: crotyl alco-
!
~
^
*
(
); yields: ethylene ( ), acetaldehyde (+), 1-butanol ( ), 1,3-butadiene ( ),
~
^
*
hol conversion ( ); yields: 1-butanol ( ), acetaldehyde (+),1,3-butadiene ( ).
&
others ( ). Figure S2 shows the same results plotted as selectivity towards
the products as a function of ethanol conversion.
The distribution of the other products detected is shown in Figure S6.
tion on minor products is reported in the Supporting Informa-
tion). Notably, ethylene formation was not expected on pure
basic oxides and its presence was found previously not to be
due to the result of a gas-phase homogeneous reaction. Also,
acetaldehyde and 1-butanol appear to be kinetic primary prod-
ucts, even though selectivity extrapolation towards zero con-
tact time may be ambiguous at such a high temperature.
to verify this possibility (Figure S5, and Figures 3 and S6, re-
spectively). 1-Butanol was the major product detected; this
could be obtained by MPV reduction of the carbonyl group in
butanal, which was formed in a relatively large amount[18,19,22]
by tautomery of the alkenol obtained by the isomerisation of
the double bond in crotyl alcohol. The MPV reduction can
occur through the reaction with crotyl alcohol itself (which is
dehydrogenated to crotonaldehyde), a hypothesis confirmed
by a catalytic test performed by co-feeding crotyl alcohol and
ethanol (molar ratio=1:1) (Table 1). This did not show any sig-
nificant difference from the test performed by feeding crotyl
alcohol alone. Overall, the distribution of the products ob-
tained by reacting crotyl alcohol on MgO was very similar to
that obtained from ethanol.
Reactivity of proposed intermediates
To better comprehend the chemistry of ethanol condensation,
we fed a few molecules proposed previously as possible inter-
mediates during the formation of C4 compounds.
The products obtained at 4008C on feeding 1,3-butanediol
(1,3-BDO), suggested to form from either the MPV reduction of
acetaldol or the reaction of ethanol with acetaldehyde,[4] are
shown in Figure S3. As 1,3-BDO is much more reactive than
ethanol, its residence time was kept lower than that of ethanol
to maintain a low conversion. The two main products were
ethanol and acetaldehyde; methyl vinyl ketone (MVK) and
methyl ethyl ketone (MEK) were also formed, as well as minor
amounts of other compounds, which include aromatics. Inter-
estingly, the presence of ethanol and acetaldehyde suggests
that 1,3-BDO undergoes a reverse addition. The catalytic be-
haviour of 1,3-BDO was also studied at 2508C (Figure S4) and
similar results were obtained. Only minor changes were ob-
served in the product distribution even if ethanol and 1,3-BDO
were co-fed in an equimolar amount; specifically, a slight in-
crease in the yields of MVK, 2-butanone and aromatic com-
pounds was noted. All in all, this evidence and the fact that we
did not detect the formation of 1,3-BDO during ethanol trans-
formation, suggest that this compound does not form under
our reaction conditions. Any adsorbed intermediate compound
that desorbs into the gas phase as 1,3-BDO would mainly de-
compose into ethanol and acetaldehyde.
Further catalytic tests were performed to elucidate addition-
al key points in the mechanism. First, we examined the conver-
sion of acetaldol (3-hydroxybutanal) under different conditions
as it has been accepted for decades that this molecule is the
key intermediate toward C4 molecules obtained by the aldol
route, although it was not detected in the product mixture in
our experiments or reported in the literature. From the results
obtained at both 250 and 4008C (Table 1), it is possible to see
that the dehydration of the aldol to crotonaldehyde was only
a minor reaction. Indeed, the main reaction observed was the
reverse aldolisation to acetaldehyde, which indicates clearly
that the aldol condensation is thermodynamically unfavourable
at these temperatures. As no significant changes in the prod-
uct distribution appeared even if ethanol and the aldol were
co-fed, any option that presents acetaldol as a key intermedi-
ate in C4-compound formation could either be ruled out or its
role limited only to a minor one. This idea has gained theoreti-
cal support recently.[20]
In conclusion, the main results obtained from reactivity ex-
periments may be summarised as follows:
Crotyl alcohol (2-buten-1-ol) was also suggested as a key in-
termediate.[12,13] Thus we reacted it on MgO at 250 and 4008C
a) the main primary product of ethanol transformation was
acetaldehyde; immediately after this, the main products
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&
3
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
Table 1. Main catalytic results for potential key reaction intermediates in ethanol conversion on MgO (in the gas phase).
Feed
T
[8C]
t
Conversion[a]
[%]
Yield [%][a]
[c]
[c]
[s]
C2H4O[b],[c]
C4H6O[c]
C4H9OH[c]
C4H6
C2H4
MVK[c]
C4H8O[c]
C4H10O[c]
acetaldol
acetaldol/ethanol
acetaldol
3-buten-2-ol
3-buten-2-ol
250
250
400
250
400
400
0.5
0.5
0.2
0.0
0.3
0.4
100
100
100
14
58
97
32 (+7)
36 (+7)
11
–
–
9
8
2
–
–
1
tr
tr
–
–
–
–
–
tr
tr
1
1
–
–
–
–
tr
2
tr
tr
tr
tr
5
tr
tr
tr
tr
10
tr
–
–
–
1
9
–
crotyl alcohol/ethanol
8
18
–
[a] If two different products are co-fed, conversion and yield refer only to the C4 compounds. [b] The yield obtained as paraldehyde is shown in parenthe-
[
sis. c] C2H4O, acetaldehyde; C4H6O: crotonaldehyde; C4H9OH: 1-butanol; C4H6: 1,3-butadiene; C2H4: ethylene; MVK: methyl vinyl ketone; C4H8O: 2-buta-
none; C4H10O: 2-butanol. tr: trace amount (yield<1%). Note: t stands for contact time.
obtained were 1-butanol and crotyl alcohol. Conversely, bu-
tanal, crotonaldehyde and butadiene were all kinetically
consecutive compounds (compared to acetaldehyde);
b) 3-hydroxybutanal (acetaldol), crotonaldehyde and gas-
phase 1,3-BDO are not key intermediates during the etha-
nol transformation to C4 compounds over MgO;
After adsorption at 308C, the alcohol feed was stopped and
He was allowed to flow to remove the ethanol adsorbed physi-
cally. Then, the temperature program was started (up to 4008C
at 108Cminꢀ1) and desorption profiles were registered simulta-
neously by DRIFTS and MS (Figure 4 and Figure S7, respectively).
In a recent paper, Davis et al.[11] reported a similar study that
focussed on ethanol adsorption on MgO. Similar to our case,
the bands at n˜ =1058–1066 cmꢀ1 were attributed to molecular
ethanol; moreover, the band at n˜ =1119–1132 cmꢀ1 was indicat-
ed to be caused by ethoxy species that disappear if T was in-
creased from 200 to 4408C. This allowed us to detect other spe-
cies; for instance, we noticed bands that appear in the spectra
at n˜ =1718 and 1143 cmꢀ1 as T was increased to 1508C. Al-
though the first of the two bands can be attributed to the C=O
stretch of acetaldehyde,[26] the second one corresponds to a spe-
cies with a peculiar CꢀO stretch that is neither from ethoxy nor
molecular ethanol. We anticipate that DFT calculations (vide in-
fra) will allow us to tentatively attribute this band to the CꢀO
stretch of a carbanion species formed by the abstraction of one
H+ from the methyl group in ethanol (this hypothesis is devel-
oped in the following sections). This band seems to reach
a maximum intensity and then shifts towards lower wavenum-
bers in the same way as some of the CꢀH vibration bands.
To have a clearer view of the bands with increasing intensity
or that disappear with increasing T and highlight correlations
among them, the different spectra were deconvoluted by
using the built-in OPUS software fitting function, and the in-
tensity of a few selected bands was plotted against T
(Figure 5). This analysis confirmed that the first species ob-
served were ethanol (bands at n˜ =2972 and 1270 cmꢀ1, with
the maximum intensity at room temperature) and ethoxy
(band at n˜ =1105 and 2925 cmꢀ1 with the maximum intensity
at 1008C). Thereafter, acetaldehyde appeared at 1008C (bands
at n˜ =1718 and 2813 cmꢀ1, also typical for aldehydes). In the
middle range of temperatures (100–2008C), we notice bands
at n˜ =2957 and 1653 cmꢀ1; the latter was reported to be char-
acteristic of the C=O stretch of acyl or acetyl species.[27] Upon
increasing the temperature beyond 2008C, the acetyl band in-
tensity decreased and the band attributed tentatively to a carb-
anion species began to shift towards lower wavenumbers
(until n˜ =1124 cmꢀ1 at T>3008C). The intensity of the ethoxy
bands decreased with no shift of the corresponding wavenum-
c) the intermediate compound that shows a product distribu-
tion closest to that shown by ethanol was crotyl alcohol.
DRIFTS
In situ DRIFTS experiments were performed to understand the
way ethanol adsorbs on the surface of MgO and how it is fur-
ther transformed with temperature (Figure 4). At room temper-
ature (T=308C), the observed bands correspond to un-dissoci-
ated H-bonded and O-bonded ethanol (a broad band between
n˜ =3500–3000 cmꢀ1 for the OHn stretch and small bands at n˜ =
1377 and 1270 cmꢀ1 that correspond to OHd and CH3d) and
surface ethoxide (i.e., the product of ethanol dissociative ad-
sorption), the bands of which at n˜ =1062 and 1103 cmꢀ1 can
be assigned to coupled CꢀC and CꢀO stretching modes. Bands
at n˜ =2972, 2928 and 2876 cmꢀ1 are caused by CH3n(a), CH2n(a)
and CH3n(s) stretching modes.[25]
Figure 4. DRIFT spectra of ethanol adsorption on MgO and desorption/trans-
formation with increasing temperature.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&
4
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
nol was also present), hydrogen, methane and CO2. The latter
may be formed both by the decomposition of surface carbo-
nates and by the “reforming” of ethanol or other products be-
cause of the water generated in the reaction medium (e.g., by
ethanol dehydration).
With no traces of acetaldol or crotonaldehyde revealed in
the spectra and, instead, the formed intermediates that resem-
ble crotyl alcohol, the in situ DRIFTS study appears to confirm
that the aldol condensation is not the key mechanism to lead
to the formation of C4 compounds from ethanol on basic
oxide materials.
To conclude our experimental investigation, a set of experi-
ments that involve feeding ethanol continuously on MgO at
4008C was performed while the mass signal of the desorbed
products was monitored (so-called “operando mode”, Figur-
es S9–S11). These tests differed from those that involved low-
temperature adsorption because the effective contact time of
reactants is much shorter. In this respect, in the continuous
feeding mode, reaction conditions were certainly closer to
those achieved in the flow reactor, albeit they may make the
identification of reaction intermediates more difficult because
of their fast transformation. Nevertheless, experiments at
4008C generated spectra that are similar to those recorded
using the TPD approach at lower temperature, and thus sup-
port the hypothesis that intermediates observed during experi-
ments under such conditions are likely precursors for the for-
mation of the final products (see detailed discussion in the
Supporting Information). To support this conclusion, the feasi-
bility of the reaction pathways that involve the unknown inter-
mediate (identified tentatively as a carbanion) adsorbed on
MgO was further investigated by DFT calculations.
Figure 5. Ethanol on MgO. IR band intensity as a function of temperature.
bers. This assignment was further confirmed by the behaviour
of the intensity of the bands (Figure 5) and by the adsorption
experiment performed at high temperature (Figure S9), which
shows that at 4008C the band at n˜ =1124 cmꢀ1 was prevalent,
whereas those attributed to ethoxy species were less intense.
Therefore, it can be suggested that the acetaldehyde/acetyl
and the carbanion may react together to form adsorbed crotyl
alcohol as the band at approximately n˜ =1620 cmꢀ1 (which ap-
pears after 2508C) was also observed if crotyl alcohol was fed
over MgO (Figure S8). Conversely, bands of adsorbed crotonal-
dehyde were not observed except perhaps if T was higher
than 3508C (Figure S8). Later on, the formation of consecutive
products (such as crotonaldehyde, acetone, butanal and buta-
diene) was detected as indicated by bands at n˜ =1672 and
1649 cmꢀ1 characteristic of C=O and C=C stretching, and by
the mass desorption profiles recorded during the temperature-
programmed desorption (TPD) experiment (Figure 4).
Theoretical modelling
Additional clues that emerge from the TPD experiments are
that 1-butanol is a primary product (especially at low tempera-
ture), that butadiene formation starts only if T>2508C and
that crotonaldehyde is observed only in traces at high temper-
atures. Acetyl species may also decompose into CO and
methyl, which may form acetone (band at around n˜ =
1649 cmꢀ1, T>3508C)[28] according to Equations (1), (2) and (3).
As a result of the complexity of the tangled reaction set that
involves ethanol on basic oxides, our computational study at-
tempted to, at least, clarify the very early stages of ethanol
conversion. We did this by focusing on likely transient species
and their production mechanism from gas-phase ethanol. A
few reactive pathways that may lead to the products detected
experimentally were also investigated and linked to the experi-
mental results. The technical details of our theoretical investi-
gation are given in the Experimental Section.
CH3CHO ! CH3COadsþHads
CH3COads ! CH3adsþCOads
CH3COadsþCH3ads ! CH3COCH3
ð1Þ
ð2Þ
ð3Þ
Alcohol dehydrogenation
As the first step, we investigated the dehydrogenation reaction
that generates the detected aldehydes using an approach vali-
dated previously.[29] The reaction (DE) and transition state (TS)
energetics for ethanol dehydrogenation with respect to two al-
ternative energy zeros are shown in Table 2; the optimised sta-
tionary points are shown in Figure 6.
A lower barrier (by 4.4 kcalmolꢀ1) and DE are calculated if
ethanol dehydrogenation takes place close to the O3C site.
The energetic ordering shown in Table 2 changes if we refer
to the gas-phase ethanol/MgO cluster as energy zero because
of the adsorption energies. Nevertheless, both catalytic sites
With respect to lighter species, the formation and partial ad-
sorption of carbonates in the range of temperatures consid-
ered is evidenced by both the broadening of the band at
around n˜ =1600 cmꢀ1 and the CO2 detected in the TPD and
that the ethanol observed at T<2508C may be related to both
the desorption of the molecularly adsorbed fraction and the
recombination of the ethoxy species with protons. Other prod-
ucts observed were ethylene (only at low temperature if etha-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&
5
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
The idea of a carbanion formation by the deprotonation of
the methyl group was considered previously by Iglesia and
Gines,[30] albeit from an adsorbed aldehyde. In their study, the
catalyst was a basic oxide (MgCeOx) modified with Cu and K,
and they used 12C2H5OH/13C2H4O reactant mixtures to establish
that condensation reactions can proceed by the direct reaction
of ethanol without the intermediate formation of gas-phase
acetaldehyde. This means that acetaldehyde condensation may
occur and that ethanol direct condensation contributes to the
formation of the products. Besides, they discussed that the
condensation reactions become easier than dehydrogenation
in presence of a metal (e.g., Cu) as the metal helps to remove
the hydrogen as H2.
Table 2. Alcohol adsorption, reaction and TS energies for the process
that leads to the dehydrogenation of ethanol and methanol on Mg10O10
and for the formation of a carbanion by methyl deprotonation in ethano-
l.[a]
Alcohol/site
Adsorption energy
DE[b]
[kcalmolꢀ1
Barrier[b]
[kcalmolꢀ1
]
[kcalmolꢀ1
]
]
Dehydrogenation
EtOH/Mg3C
EtOH/O3C
MeOH/O3C
MeOH/Mg3C
39.5
23.9
25.1
40.4
38.1 (ꢀ1.4)
29.4 (5.5)
37.0 (11.9)
45.9 (5.5)
44.7 (5.2)
40.4 (16.5)
44.5 (19.4)
51.4 (11.0)
Carbanion formation
EtOH/Mg3C
EtOH/O3C
31.3 (ꢀ8.2)
33.4 (ꢀ6.1)
36.4 (12.5)
36.7 (12.8)
The suggestion of carbanion formation on the MgO surface
rationalises several experimental observations, the simplest
one of which is the production of ethylene. We have located
two TS that lead from the Mg3C-adsorbed carbanion to ethyl-
ene and dissociated water (see Figure 9 for the reactant/TS
structures) with fairly low barrier heights (3.7 and 6.5 kcal
molꢀ1). Such a finding places the barrier to be surmounted
during the carbanion formation plus dehydration sequence at
4.8 kcalmolꢀ1 below the barrier that leads to acetaldehyde
plus H2. Thus, the dehydration of ethanol is a competitive pro-
cess compared to its dehydrogenation even onto the purely
basic MgO. Besides, the carbanion normal mode at n˜ =
1165 cmꢀ1 also helps to assign the IR feature at n˜ =1143 cmꢀ1,
the evolution of which is in line with the reactivity expected
from such species (vide supra, DRIFTS Section and Figure S12,
which show the simulated IR spectrum of the carbanion). Simi-
larly, the position of the two peaks around n˜ =1220 cmꢀ1 in
the spectrum shown in Figure S12 correlates well with the lo-
cation of a shoulder that appears and then disappears in the
DRIFT spectrum upon increasing the temperature. Notably, the
relative intensities in the n˜ =1143–1220 cmꢀ1 region are not re-
produced well by the calculations. We consider that this find-
ing is most likely because the shoulder modes involve the dis-
placement of the proton on MgO that interacts directly with
the anionic carbon atom; the effect of such motion on the mo-
lecular dipole may be estimated improperly by DFT due to of
shortcomings in the description of electronic correlation. In-
stead, the lack of any feature in the DRIFT spectra around n˜ =
1380 cmꢀ1 (Figure S12), a mode connected to the wagging of
OH-bearing CH2 in the carbanion, remains for the moment un-
clear.
[a] The zero of the energy for the processes is assumed to be the ad-
sorbed alcohol with the OH dissociated. [b] In parentheses, there are en-
ergetic quantities that refer to the gas-phase alcohols plus Mg10O10 as re-
actants.
should be considered as active thanks to the high temperature
(4008C) employed during the reaction, and the Mg3C site is
likely to produce more aldehydes, comparatively, thanks to the
lower barrier from the feed reactants.
Enol formation
The enolic form of acetaldehyde would be important if the C4
formation proceeded through the aldolic pathway, and DFT
calculations indicate that C4 formation on Mg3C should release
energy (2 kcalmolꢀ1) and require a low barrier to be surmount-
ed (6.6 kcalmolꢀ1; see Figure 7 for the optimised structures). It
is thus likely that a fast equilibrium may exist between the two
molecules.
Carbanion formation
The basic MgO surface sites may cleave a CꢀH bond hetero-
lythically in the ethanol methyl group, and our calculations
(Table 2 and Figure 8) indicate that such a process generates
a carbanion 31.3 and 36.4 kcalmolꢀ1 above the adsorbed etha-
nol in the vicinity of the Mg3C and O3C sites, respectively. De-
protonation close to the Mg3C site generates a product that is
6.8 kcalmolꢀ1 lower in energy than acetaldehyde plus ad-
sorbed H2 compared to MgO/gaseous ethanol; the TS barriers
for such processes also have the same order of energy. The
carbanion produced close to the O3C site, instead, is higher by
7.0 kcalmolꢀ1 than acetaldehyde/adsorbed H2, although the TS
barriers have the same order as that close to Mg3C. Notably,
the barrier for carbanion formation near O3C is just 0.3 kcal
molꢀ1 above the energetic requirement for the reaction.
With the quantitative results discussed, it is possible to draw
a partial conclusion with respect to the importance of the al-
dolic path toward C4 formation on MgO. The higher energetic
cost to be paid to produce two molecules of acetaldehyde
compared to a single carbanionic species or a carbanion and
an acetaldehyde ought to make the surface concentration of
the acetaldehyde/enol pair substantially less than that of the
carbanion/acetaldehyde couple (or the carbanion/ethanol pair,
vide infra). In turn, this suggests that the aldolic pathway is
less likely than the one that involves the carbanion. Additional
arguments against the aldolic scheme come from the fact that
the adsorbed aldol lies substantially higher in energy than the
adsorbed 1,3-BDO and that our DFT calculations predict the al-
The data reported in Table 2 suggest that both Mg3C and
O3C sites may produce the two intermediates, albeit with dif-
ferent activities. In fact, the global energy profiles suggest that
carbanion formation on the Mg3C site ought to be the most
likely process, followed by ethanol dehydrogenation on the
same site and carbanion formation on O3C.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&
6
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
dolic condensation to be nearly
energy neutral in the gas phase
(see Figure S12 and related dis-
cussion). Notably, this finding is
valid whatever the catalyst em-
ployed and suggests that the
presence of reduced metals
(e.g., see work by Iglesia and
Gines[30] and the butadiene cata-
lyst by Ivanova et al.[31]
) may
only favour the formation of
acetaldehyde to be consumed
along paths alternative to the al-
dolic one.
Carbanion reactivity
As suggested, the in situ genera-
tion of a carbanion from ethanol
may open alternative channels,
among which are the attack
to the acetaldehyde carbonyl
group and to the hydroxyl-bear-
ing carbon atom in another eth-
anol molecule. The first reaction
would lead to the alkoxide pre-
cursor of 1,3-BDO, whereas 1-bu-
tanol would be obtained by the
SN2-like attack of the Mg-coordi-
nated and activated ethanol. The
energetics for these two pro-
cesses are provided in Table 3,
and the relevant structures are
shown in Figure 10. Although
both processes are exergonic
with respect to the reactants be-
cause of the formation of a CꢀC
bond, the carbonyl attack ap-
pears to be substantially more
facile than the SN2-type reaction
because of the electronic struc-
ture of the aldehydic group and
the fact that the charge-bearing
carbon atom in the carbanion
has to detach from the stabiliz-
ing lattice Mg to approach the
OH-bearing carbon atom in eth-
anol (Figure 10).
Figure 6. Reactant, TS and product (from left to right) of the alcohol (a and b, ethanol; c and d, methanol) dehy-
drogenation process on MgO near the Mg3C (a and c) and O3C (b and d) sites.
With the zero set as two
vapour ethanol molecules plus
the MgO clusters, the energy
profiles place the products of
the ethanol/carbanion and acet-
aldehyde/carbanion reactions at
ꢀ54.1 and ꢀ21.2 kcalmolꢀ1, re-
spectively. Thus, the adsorbed 1-
butanol plus dissociated water is
Figure 7. Reactant, TS and product for the enol formation from acetaldehyde near the Mg3C site. Note the pres-
ence of two near-degenerate isomers for the enol coordinated onto the Mg3C site.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&
7
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
compounds are likely to be pro-
duced by the dehydration of the
deprotonated 1,3-BDO by carb-
anion formation (Figure 10) as
discussed in the Supporting In-
formation.
An additional consequence of
the fact that adsorbed 1,3-BDO
lies only 21.2 kcalmolꢀ1 below
the common energy zero is the
possible retro-dissociation into
ethanol and acetaldehyde, a pro-
cess that requires only ꢀ0.3 kcal
molꢀ1 on MgO and is apparent
in both Figures S3 and S4. Im-
portantly, this conclusion does
not contradict the proposed
mechanism of 3-buten-1-ol and
crotyl alcohol formation (Sup-
porting Information), as these
are formed in the vicinity of
corner sites. Feeding 1,3-BDO di-
rectly onto MgO is instead likely
Figure 8. Reactant, TS and product for the carbanion formation from ethanol near the Mg3C (top) and O3C
(bottom) sites. Notice the presence of two near-degenerate isomers for the carbanion coordinated onto the
Mg3C site.
to lead toward terrace adsorption, at least initially; in such a sit-
uation, the diol may gather the energy to decompose easily.
Table 3. TS barrier heights and energetics of the carbanion attack to the
acetaldehyde carbonyl and hydroxyl-bearing carbon atom in ethanol
gauged from the co-adsorbed reactive species in the vicinity of the
corner site.[a]
Guerbet versus Lebedev reactions: Common features and
differences in mechanisms
Reactants
TS barrier[b]
DE[b]
[kcalmolꢀ1
]
[kcalmolꢀ1
]
From the data collected so far by reactivity tests, DRIFTS, TPD,
MS and theoretical modelling, it is possible to suggest a new
mechanism able to explain not only the evidence reported in
this work but also the data published in many decades of liter-
ature on both the Lebedev and Guerbet reactions.
acetaldehyde/carbanion
ethanol/carbanion
acetaldehyde/enol
11.4 (19.1)
41.0 (28.9)
7.2 (6.3)
ꢀ28.9 (ꢀ21.2)
ꢀ42.0 (ꢀ54.1)
ꢀ10.6 (ꢀ11.5)
[a] The energetics of the stationary points for the aldolic pathway are
also shown. [b] The energies that refer to two ethanol molecules in the
gas phase plus the MgO cluster are shown in parentheses.
First, ethanol is adsorbed on the basic oxide surface and dis-
sociates into acetaldehyde and hydrogen; this is a rate-deter-
mining step and it requires a relatively high temperature
(>2008C) to take place. Adsorbed acetaldehyde is in equilibri-
um with the related enolic form. At the same time, ethanol
can be adsorbed on the catalyst surface as an ethoxide onto
low-coordination sites, although this anion always remains in
equilibrium with its un-dissociated alcoholic form. This hap-
pens especially near O3C (the alcoholic oxygen atom is coordi-
nated on a Mg4C site), and the molecular ethanol may under-
go a proton abstraction from the a-carbon atom to generate
a carbanion, the negative charge of which is stabilised by the
more stable than the dissociated 1,3-BDO because of the dehy-
drogenation energy (Table 2). On the same energy scale, the
TS barrier for the two processes becomes 28.9 and 19.1 kcal
molꢀ1, respectively; hence, either the two pathways have a sim-
ilar activity, at least based on the energy requirement alone, or
the process for 1,3-BDO is somewhat faster than that for 1-bu-
tanol. This suggestion agrees well with the results shown in
Figure 1, in which a slightly lower yield is obtained for 1-buta-
nol than for 3-buten-1-ol plus crotyl alcohol. Importantly, these
Figure 9. Reactants and TS for the dehydration of ethanol starting from the Mg3C-adsorbed carbanion.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&
8
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
duces much more ethylene than
magnesium oxide.[15] Importantly,
the aldol condensation (which is
endergonic by 33.8 kcalmolꢀ1 ac-
cording to the level of theory
used)[1] can be definitely ruled
out as the main path in the Leb-
edev process as the encounter of
two molecules of acetaldehyde is
hindered because of their scarce
concentration on the surface
(the Tishchenko dimerisation to
ethyl acetate is slow for the
same reason) and the large pre-
vailing concentration of ad-
sorbed ethanol. This is activated
either as ethoxide or carbanion
species. A reaction between the
former and acetaldehyde is
indeed possible, even though it
would yield the hemiacetal (a
possible precursor of ethyl ace-
tate, which we observed only in
traces). Its formation, however, is
thermodynamically hindered at
high temperatures.
The second reaction is the
main path followed by ethanol
to produce Guerbet alcohols,
that is, mainly 1-butanol. This re-
action is, therefore, aided by cat-
alysts able to stabilise the carb-
anion (but not too much to
avoid ethylene formation) and
by the lack of acetaldehyde on
Figure 10. Reactants, TS and products of the reaction between the carbanion and an adsorbed ethanol molecule
(top) or acetaldehyde (middle). Also shown, are the two carbanions obtained from 1,3-BDO (bottom).
cations present in the oxide lattice as the near-sp3-type orbital
that contains the lone pair electrons points towards one of
them. The carbanion may undergo three different reactions,
and each path is facilitated or hindered depending on temper-
ature, catalyst acid–base features and the likelihood of molecu-
lar encounter (vide infra):
a vicinal site. These conditions may be reached by decreasing
the temperature to limit ethanol dehydrogenation to acetalde-
hyde (see different selectivity to 1-butanol in Figures 1 and 2)
and/or by tuning catalyst features. For example, hydroxyapa-
tite[15] shows the highest yield to 1-butanol because, as point-
ed out by the authors, Lewis acid sites (Ca2+) and basic sites
(O2ꢀ) have a greater atomic distance than MgO, which might
play an important role in the likelihood of molecular encoun-
ter. Moreover, for hydroxyapatite with a high Ca/P ratio, the
acid sites needed to perform alcohol dehydration are almost
nil, which avoids consecutive reactions that might possibly
take place from 1-butanol or crotyl alcohol, of which the latter
is a competitive path.
a) the hydroxyl group present in the carbanion reacts with
the H+ dissociated previously (adsorbed on the catalyst sur-
face) to produce water and ethylene, which desorb in the
gas phase (route 1 in Scheme 2);
b) the carbanion reacts with another ethanol molecule by an
SN2-like attack to lose a water molecule to produce 1-buta-
nol directly (route 2 in Scheme 2);
Route 3 leads to 1,3-butadiene from ethanol, that is, the
Lebedev reaction; the latter process has been demonstrated
clearly to require complex catalysts able, firstly, to dehydrogen-
ate to acetaldehyde, secondly, to condense C2 intermediates
and, lastly, to dehydrate the formed C4.[1,18] Moreover, it is well
known that acetaldehyde has a beneficial effect on 1,3-buta-
diene yield, which is why this aldehyde is recycled to the reac-
tor in the industrial process.[32] All this evidence is confirmed
by our model. Acetaldehyde must react with the ethanol carb-
c) the carbanion reacts with an acetaldehyde molecule to pro-
duce an adsorbed species directly, which desorbs to pro-
duce either crotyl alcohol (route 3 in Scheme 2 or 3-buten-
1-ol and a water molecule. Alkenols may then dehydrate to
1,3-butadiene or rearrange into 3-buten-2-ol.
As far as the first path is concerned, the harder the cation is,
the more stable the carbanion; this is why calcium oxide pro-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&
9
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
tance of two active sites (one
bound to the aldehyde and the
other to the carbanion) must be
right to facilitate the process.
Lastly, dehydration properties
are essential to dehydrate the in-
termediate alkenol by shifting
the equilibrium to 1,3-butadiene.
In the case of the purely basic
system investigated here, the
yield and selectivity to butadiene
are much lower than those ob-
served with mixed acid–base
systems (e.g., Mg/Si/O). Howev-
er, the presence of acid sites is
required only to perform the
consecutive steps that concern
alcohol dehydration (e.g., to
lead to 1,3-butadiene), whereas
they are not involved in the gen-
eration of primary products and
intermediates shared by both
the Lebedev and the Guerbet re-
actions.
The dehydration of alcohols
on acid systems involves energy
barriers similar to those calculat-
ed with MgO;[33] this means that
with bifunctional catalysts, an
additional, acid-catalysed path-
way leads to higher butadiene
(and butene) selectivity and less
by-product formation than with
a
purely basic catalyst, with
which several other side reac-
tions contribute to the formation
of a wide spectrum of products.
Finally, the mechanism pro-
posed by Meunier et al.[20] for the
Guerbet reaction, based mainly
on thermodynamic calculations,
has analogies with the mecha-
nism proposed here. In particular,
it was reported that at least two
reaction pathways take place si-
multaneously, the main pathway
Scheme 2. Hypothesised reaction pathways that lead to ethylene (route 1), 1-butanol (Guerbet reaction, route 2)
and crotyl alcohol, precursor of butadiene (Lebedev reaction, route 3).
anion to form crotyl alcohol as the key intermediate, and the
precursor for butadiene formation by further dehydration. The
final and kinetically consecutive dehydration of the alkenol to
butadiene is likely to involve a mechanism quite similar to that
observed for both ethanol and 1,3-butanediol dehydration.
The reaction between acetaldehyde and ethanol is in com-
petition with the direct production of 1-butanol, which is why
an external source of aldehyde is needed to decrease the
probability of the reaction of the carbanion with ethanol and
increase the selectivity to butadiene. The condensation fea-
tures of catalysts must be carefully tuned as the atomic dis-
involves the condensation of two ethanol molecules with no in-
termediate gaseous compounds (so-called “direct” route),
whereas a minor “indirect” route involves the condensation of
ethanol with acetaldehyde (formed from ethanol dehydrogena-
tion) to form butenol, which is converted subsequently to buta-
nol by hydrogen transfer from a sacrificial ethanol molecule.
Conclusions
A new mechanism for the transformation of ethanol into C4
compounds over MgO, as a model basic catalyst, has been
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&10
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
proposed based on the results of reactivity tests, in situ diffuse
reflectance infrared Fourier transform spectroscopy and theo-
retical modelling. This approach made it possible to draw a uni-
fying picture of the different routes that lead to either 1-buta-
nol (a transformation known as the Guerbet reaction), buta-
diene (the main product in the Lebedev process) or ethylene.
The suggested mechanism discards the key role of both acetal-
dol and crotonaldehyde as the reaction intermediates, usually
accepted to be shared between the two pathways that lead
either to the alcohol or the diolefin. Conversely, we have found
that the two reactions involve different intermediate species.
Crotyl alcohol and 3-buten-1-ol are the key intermediates of
the Lebedev process and precursors for butadiene formation.
However, the reaction between ethanol and its activated form
(the carbanion) might explain the formation of the Guerbet al-
cohol as a kinetic primary product.
The mechanism suggested here also explains the formation
of dehydration compounds even on purely basic materials.
Furthermore, the mechanism proposed does not include H-
transfer (Meerwein–Ponndorf–Verley) reactions from ethanol to
any reaction intermediate.
Experimental Section
Catalyst preparation and characterisation
The catalyst was prepared by precipitation from an aqueous solu-
tion that contained the corresponding nitrate salt, Mg(NO3)2·6H2O.
The solution was added dropwise to another solution that con-
tained Na2CO3 dissolved in distilled water. As the first solution was
added to the second one, the pH was adjusted continuously to
keep it close to 10.0. Under these conditions, the precipitation of
Mg(OH)2 occurred. The obtained slurry was stirred for 40 min. The
precipitate was separated from the liquid by filtration and washed
with distilled water (6 L) at 408C. The solid was then dried at
110 8C overnight and calcined at 4508C for 8 h in air.
The reactions involved are summarised in Scheme 3, which
show in a simplified scheme the stoichiometries for the forma-
tion of main products; an overview of all reactions involved is
shown in Figure S14, as inferred from reactivity experiments
performed on both ethanol and the intermediates.
The surface area was measured by the BET single-point method (N2
adsorption at the temperature of liquid N2) by using a Sorpty 1750
Fisons Instrument. The surface area of the calcined sample was
On the basis of these results and on the hypothesis suggest-
ed, one might expect MgO to be a good catalyst in crotyl alco-
hol dehydration to butadiene if the alkenol is fed directly to
the reactor. However, this was not the case as shown by the
results in Figure 3. This is because, in addition to dehydration,
many other reactions occur on adsorbed crotyl alcohol, such
as double-bond isomerisation, dehydrogenation on the enol
species, which result in a variety of products.
86 m2 gꢀ1
.
Bench-scale plant tests
Reactivity experiments were performed by using a continuous-flow
reactor made of glass that operated under atmospheric pressure.
Catalyst (0.10–1 g) was loaded in pellet (30–60 mesh) form. Resi-
dence time (calculated as the ratio between catalyst volume [mL]
and total gas flow [mLsꢀ1], which was measured at the reaction
temperature) varied. The inlet feed molar ratio was constant and
set at 2 mol% organic+98% nitrogen.
Downstream products were fed to an automatic sampling system
for GC analysis. This was performed by using an Agilent-5890 in-
strument equipped with a flame ionisation detector (FID) and
a thermal conductivity detector. Agilent chromatographic columns
HP-5 (50 m, 0.20 mm) and HP-plot Al2O3-KCl (30 m, 0.50 mm) were
used for product separation. Qualitative analysis were performed
by bubbling effluent gas through two in-series abatement devices,
which were filled with acetone (but in some cases either water or
dichloromethane was used) and held at a temperature of 0–28C.
Compounds were identified by GC–MS and the injection of pure
reference standards for the comparison of retention times in the
GC column.
DRIFTS–MS
The sample was pre-treated at 4508C in a He flow (10 mLminꢀ1
)
for 45 min to remove any molecules adsorbed on the material.
Then the sample was cooled to RT, and ethanol was fed at
0.6 mLminꢀ1. Subsequently, it was left to flow until weakly ad-
sorbed ethanol was evacuated, and then the temperature program
started (until 4508C at 108Cminꢀ1). The following selected MS sig-
nals were monitored continuously with time (and temperature):
m/z 2, 16, 25, 28, 29, 30, 31, 40, 41, 43, 44, 45, 56, 58, 59, 60, and
61. By combining the information obtained from several different
m/z signals, it was possible to obtain unambiguous information on
the various products formed. The IR apparatus used was a Bruker
Vertex 70 with a Pike DiffusIR cell attachment. Spectra were record-
Scheme 3. Stoichiometries of the main reactions that occur at low tempera-
ture (top) and high temperature (bottom) if ethanol is fed over MgO.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&
11
&
ÞÞ
These are not the final page numbers!

CHEMSUSCHEM
FULL PAPERS
www.chemsuschem.org
[6] S. Kim, B. E. Dale, Biomass Bioenergy 2004, 26, 361–375.
[7] S. Liu, H. Lu, R. Hu, A. Shupe, L. Lin, B. Liang, Biotechnol. Adv. 2012, 30,
785–810.
ed by using a mercury cadmium telluride (MCT) detector after 128
scans with 2 cmꢀ1 resolution. The mass spectrometer was an
EcoSys-P from European Spectrometry Systems. The second type
of experiment was performed by pre-treating the sample in the
same way and then setting the temperature at 4008C before the
ethanol feed was started (30 min at 0.6 mLminꢀ1). As stated briefly
in the discussion, the spectra were deconvoluted by using the
built-in OPUS software fitting function. The model was optimised
using Lorentzian peak shapes and the Levenberg–Marquardt algo-
rithm. We also divided each spectrum at each temperature into
three frequency ranges to fit them separately. The criteria to
accept them was a residual root-mean-square (RMS) error lower
than 0.0009.
[8] V. Menon, M. Rao, Prog. Energy Combust. Sci. 2012, 38, 522–550.
[9] A. S. Ndou, N. Plint, N. J. Coville, Appl. Catal. A 2003, 251, 337–345.
[10] E. V. Makshina, W. Janssens, B. F. Sels, P. a. Jacobs, Catal. Today 2012,
198, 338–344.
[11] T. W. Birky, J. T. Kozlowski, R. J. Davis, J. Catal. 2013, 298, 130–137.
[12] M. Leꢃn, E. Dꢄaz, S. OrdꢃÇez, Catal. Today 2011, 164, 436–442.
[13] M. Leꢃn, E. Dꢄaz, A. Vega, S. OrdꢃÇez, A. Auroux, Appl. Catal. B 2011,
102, 590–599.
[14] V. V. Ordomsky, V. L. Sushkevich, I. I. Ivanova, J. Mol. Catal. A 2010, 333,
85–93.
[15] T. Tsuchida, J. Kubo, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, J.
Catal. 2008, 259, 183–189.
[16] T. Tsuchida, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, Ind. Eng.
Chem. Res. 2008, 47, 1443–1452.
[17] H. Jones, E. Stahly, B. Corson, J. Am. Chem. Soc. 1949, 71, 1822–1828.
[18] G. Natta, R. Rigamonti, Chim. Ind. 1947, 29, 195–200.
[19] W. M. Quattlebaum, W. J. Toussaint, J. T. Dunn, J. Am. Chem. Soc. 1947,
69, 593–599.
[20] J. Scalbert, F. Thibault-Starzyk, R. Jacquot, D. Morvan, F. Meunier, J.
Catal. 2014, 311, 28–32.
[21] B. B. Corson, H. E. Jones, C. E. Welling, J. A. Hinckley, E. E. Stahly, Ind.
Eng. Chem. 1950, 42, 359–373.
DFT calculations
The modelling of the processes involved in the reactivity of etha-
nol on MgO was performed by employing gas-phase electronic
structure calculations by using the Gaussian 09[34] suites of codes
at the DFT level. Specifically, we used the B3LYP DFT functional[35]
together with the 6-31+ +G(d,p) basis set, a level of theory em-
ployed commonly for similar tasks.[36,37]
[22] C. Yang, Z. Meng, J. Catal. 1993, 142, 37–44.
[23] W. J. Toussaint, J. T. Dunn, D. R. Jackson, Ind. Eng. Chem. 1947, 39, 120–
125.
As performed previously,[29] MgO nanocrystals were modelled using
a cluster approach mainly with Mg10O10 as the model of the crystal
corners (Mg3C and O3C sites; Figure 9). Whenever edge-proximity
effects may have introduced a bias in the energetic profile studied,
the largest Mg16O16 cluster (Figure 10) was used to limit polarisa-
tion-induced artefacts. The geometrical parameters of the clusters
(right angles and dMgO =2.1084 ꢂ) were kept frozen[36] while the de-
grees of freedom of the adsorbed species were optimised. In many
cases, putative TS structures were generated by means of relaxed
scans along chosen reaction coordinates (e.g., atomꢀatom distan-
ces) and subsequently optimised fully by using analytical second
derivatives of the energy surface. All stationary points were charac-
terised using frequency calculations. As these interesting processes
involve heavy atoms or groups, no correction for zero point effects
was introduced. Relative TS and product energies are provided
with respect to the reactants of the specific process under investi-
gation and with respect to the species fed into the reaction envi-
ronment (i.e., the MgO clusters and ethanol molecules). The
second type of energy scale allows a direct comparison between
alternative reaction pathways.
[24] G. Szçllçsi, M. Bartꢃk, Appl. Catal. A 1998, 169, 263–269.
˝
[25] M. Dçmçk, M. Tꢃth, J. Raskꢃ, A. Erdohelyi, Appl. Catal. B 2007, 69, 262–
272.
[26] C. Resini, S. Cavallaro, F. Frusteri, React. Kinet. Catal. Lett. 2007, 90, 117–
126.
[27] A. M. da Silva, K. R. de Souza, L. V. Mattos, G. Jacobs, B. H. Davis, F. B.
Noronha, Catal. Today 2011, 164, 234–239.
[28] M. G. Rosmaninho, F. C. C. Moura, Appl. Catal. B 2012, 115 – 116, 45–52.
[29] M. M. T. Pasini, A. Lolli, S. Albonetti, F. Cavani, J. Catal. 2014, 317, 206–
219.
[30] M. J. L. Gines, E. Iglesia, J. Catal. 1998, 176, 155–172.
[31] V. L. Sushkevich, I. I. Ivanova, V. V. Ordomsky, E. Taarning, ChemSusChem
2014, 7, 2527–2536.
[32] W. J. Toussaint, J. T. Dunn, US Pat. 2,421,361, 1947.
[33] M. J. Janik, J. Macht, E. Iglesia, M. Neurock, J. Phys. Chem. C 2009, 113,
1872–1885.
[34] Gaussian 09, Revision B.01, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E.
Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Men-
nucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian,
A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara,
K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O.
Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr., J. E. Peralta, F. Ogliaro,
M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Ko-
bayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyen-
gar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B.
Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann,
O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin,
K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg,
S. Dapprich, A. D. Daniels, ꢅ. Farkas, J. B. Foresman, J. V. Ortiz, J. Cio-
slowski, D. J. Fox, Gaussian, Inc., Wallingford CT, 2009. .
[35] A. D. Becke, J. Chem. Phys. 1993, 98, 5648.
Acknowledgements
CIRI Energia e Ambiente (Universitꢀ di Bologna) is acknowledged
for the PhD grant to A.C. Consorzio INSTM (Firenze) is acknowl-
edged for the PhD grant to C.B.
Keywords: alcohols · alkenes · density functional calculations ·
magnesium · reaction mechanisms
[36] A. G. Pelmenschikov, G. Morosi, A. Gamba, S. Coluccia, G. Martra, E. A.
Paukshtis, J. Phys. Chem. 1996, 100, 5011–5016.
[37] M. Cavalleri, A. Pelmenschikov, G. Morosi, A. Gamba, S. Coluccia, G.
Martra, Oxide-Based Systems at the Crossroads of Chemistry, Second Inter-
national Workshop, Elsevier, 2001.
[1] C. Angelici, B. M. Weckhuysen, P. C. A. Bruijnincx, ChemSusChem 2013, 6,
1595–1614.
[2] J. A. Posada, A. D. Patel, A. Roes, K. Blok, A. P. C. Faaij, M. K. Patel, Biore-
sour. Technol. 2013, 135, 490–499.
[3] S. I. Mussatto, G. Dragone, P. M. R. Guimar¼es, J. P. A. Silva, L. M. Car-
neiro, I. C. Roberto, A. Vicente, L. Domingues, J. A. Teixeira, Biotechnol.
Adv. 2010, 28, 817–830.
[4] M. Balat, H. Balat, Appl. Energy 2009, 86, 2273–2282.
[5] P. C. A. Bruijnincx, B. M. Weckhuysen, Angew. Chem. Int. Ed. 2013, 52,
11980–11987; Angew. Chem. 2013, 125, 12198–12206.
Received: July 7, 2014
Revised: August 28, 2014
Published online on && &&, 0000
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&12
&
ÞÞ
These are not the final page numbers!

FULL PAPERS
Lebedev and Guerbet: what’s the dif-
ference? The multifaceted approach
used to study the mechanism of the
Lebedev and Guerbet reactions indi-
cates that the two processes share the
same anionic intermediate but evolve
along different reaction pathways,
which avoid the thermodynamically
hampered aldolic route.
A. Chieregato, J. Velasquez Ochoa,
C. Bandinelli, G. Fornasari, F. Cavani,*
M. Mella*
&& – &&
On the Chemistry of Ethanol on Basic
Oxides: Revising Mechanisms and
Intermediates in the Lebedev and
Guerbet reactions
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 0000, 00, 1 – 13
&13
&
ÞÞ
These are not the final page numbers!