Kinetics of acetic acid synthesis from ethanol over a Cu/SiO2 catalyst

Article abstract of DOI:10.1016/j.apcata.2011.05.030

The dehydrogenation of ethanol via acetaldehyde for the synthesis of acetic acid over a Cu based catalyst in a new process is reported. Specifically, we have studied a Cu on SiO₂ catalyst which has shown very high selectivity to acetic acid via acetaldehyde compared to competing condensation routes. The dehydrogenation experiments were carried out in a flow through lab scale tubular reactor. Based on 71 data sets a power law kinetic expression has been derived for the description of the dehydrogenation of acetaldehyde to acetic acid. The apparent reaction order was 0.89 with respect to water and 0.45 with respect to acetaldehyde, and the apparent activation energy was 33.8 kJ/mol. The proposed oxidation of acetaldehyde with hydroxyl in the elementary rate determining step is consistent with these both. Density Functional Theory (DFT) calculations show the preference of water cleavage at the Cu step sites. In light of this, an observed intrinsic activity difference between whole catalyst pellets and crushed pellets may be explained by the Cu crystal size and growth rate being functions of the catalyst particle size and time.

Full text of DOI:10.1016/j.apcata.2011.05.030

Applied Catalysis A: General 402 (2011) 69–79
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Kinetics of acetic acid synthesis from ethanol over a Cu/SiO catalyst
2
a,∗
a
b
a
c
Bodil Voss , Niels Christian Schjødt , Jan-Dierk Grunwaldt , Simon Ivar Andersen , John M. Woodley
a
Haldor Topsøe A/S, Nymøllevej 55, DK-2800 Kgs. Lyngby, Denmark
Institute for Chemical Technology and Polymer Chemistry, Karlsruhe Institute of Technology (KIT), Kaiserstr. 12, 76128 Karlsruhe, Germany
Department of Chemical and Biochemical Engineering, Technical University of Denmark (DTU), DK-2800 Kgs. Lyngby, Denmark
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
The dehydrogenation of ethanol via acetaldehyde for the synthesis of acetic acid over a Cu based cata-
lyst in a new process is reported. Specifically, we have studied a Cu on SiO2 catalyst which has shown
very high selectivity to acetic acid via acetaldehyde compared to competing condensation routes. The
dehydrogenation experiments were carried out in a flow through lab scale tubular reactor. Based on 71
data sets a power law kinetic expression has been derived for the description of the dehydrogenation of
acetaldehyde to acetic acid. The apparent reaction order was 0.89 with respect to water and 0.45 with
respect to acetaldehyde, and the apparent activation energy was 33.8 kJ/mol. The proposed oxidation of
acetaldehyde with hydroxyl in the elementary rate determining step is consistent with these both. Den-
sity Functional Theory (DFT) calculations show the preference of water cleavage at the Cu step sites. In
light of this, an observed intrinsic activity difference between whole catalyst pellets and crushed pellets
may be explained by the Cu crystal size and growth rate being functions of the catalyst particle size and
time.
Received 1 February 2011
Received in revised form 18 May 2011
Accepted 19 May 2011
Available online 2 June 2011
Keywords:
Ethanol dehydrogenation
Acetic acid
Cu crystal
Kinetics
Step sites
©
2011 Elsevier B.V. All rights reserved.
1
. Introduction
As is well-known, but not industrialised, acetic acid may also be
produced from ethanol in a single non-oxidative reaction step [3].
The non-oxidative dehydrogenation of ethanol to acetic acid may
be described through the following reactions according to Eqs. (1)
and (2).
Acetic acid is a bulk chemical which today is produced in
amounts exceeding 10 million tons per year worldwide. The estab-
lished production route has been dominated by the carbonylation
of methanol which only relies on fossil sources [1]. In recent years,
◦
CH CH OH ꢀ CH CHO + H
ꢀH = 68.7 kJ/mol
(1)
(2)
3
2
3
2
under the consideration of CO neutrality and independence of fos-
2
◦
sil sources, the utilisation of biomass for chemical production has
regained significant interest. The dehydrogenation of ethanol to
acetic acid over Cu catalysts has been known for about a century [2],
but ever since the highly selective carbonylation process was dis-
covered the carbonylation route has attracted far the most interest
due to its comparably cheap feedstock. Industrially the utilisation
of ethanol as a feedstock for acetic acid production has survived
the competition from the fossil feedstock based carbonylation pro-
cess where cheap ethanol is available in large amounts. Less than
CH CHO + H O ꢀ CH COOH + H
ꢁH = −24.8 kJ/mol
3
2
3
2
Both reactions are equilibrium limited, implying that uncon-
verted ethanol and acetaldehyde should be separated from the
acetic acid product and recycled in an industrial process.
Depending on the catalyst activity the further esterification of
ethanol with acetic acid may take place:
CH CH OH + CH COOH ꢀ CH COOCH CH + H O
(3)
and butanol or other condensation side-products may be produced:
2CH CH OH → C H OH + H O (4)
3
2
3
3
2
3
2
1
0% of the world’s capacity of acetic acid is produced according to
the route: dehydrogenation of ethanol to acetaldehyde followed
by partial oxidation of acetaldehyde to acetic acid [2]. If not care-
fully operated, however, the two step ethanol route is hampered
by a high loss of feedstock to the parasitic CO2 production in the
oxidation step.
3
2
4
9
2
Many Cu catalysts, supported or non-supported, are able to con-
vert ethanol to acetic acid. Especially Cr promoted Cu catalysts have
been studied due to their regeneration and stabilising abilities, but
the carcinogenic effects of Cr in its hexavalent form makes this pro-
moter less desirable. The kinetics of the first conversion (Eq. (1))
has been investigated on unsupported Cu as well as Cr promoted
Cu by Tu et al. [4], where it was found that the dehydrogenation
of ethanol to acetaldehyde is a first order reaction in ethanol. An
apparent activation energy of the non-promoted unsupported Cu
∗ Corresponding author. Tel.: +45 45272000; fax: +45 45272999.
E-mail address: bov@topsoe.dk (B. Voss).
0
926-860X/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2011.05.030

7
0
B. Voss et al. / Applied Catalysis A: General 402 (2011) 69–79
catalyst of 51 kJ/mol was found. However, the data analysis by Tu
et al. [4] appears to have been conducted without allow for the fact
that the dehydrogenation reaction is equilibrium limited. Herein,
it was assumed that the fraction of converted ethanol should be
read as the fraction of ethanol converted divided by equilibrium
ethanol conversion fraction, which would be the normal param-
eter subject to graphical analysis with equilibrium reactions. This
would explain the high degree of conversion reported for ethanol.
The chemical equilibrium for this reaction has been discussed [5,6].
However, in an overall ethanol to acetic acid process high cat-
alyst selectivity to the desired product may be of even greater
importance than a high activity, due to simpler product recov-
ery section. We have found that Cu on SiO2 is a highly selective
dehydrogenation catalyst presumably due to its support neutral-
ity, thereby inhibiting e.g. acid or base catalysed dehydration and
condensation reactions.
2.2. Catalyst characterisation
Theoretical DFT calculations were made in-house by using a
Dacapo calculator on the simulation of a three-layer thick copper
slab. The water binding energy was calculated by subtracting the
energy of the relaxed copper slab and the water in the gas phase
from the total energy of the water adsorbed on the copper slab.
X ray diffraction (XRD) was conducted in an X ray diffractometer
with Cu K␣1 radiation of a wave length of 1.54 A˚ .
In situ reduction was conducted in Transmission Electron Micro-
scope (TEM) on crushed catalyst powder in a Titan ETEM apparatus.
The microscope was operated at 300 kV and tuned to a flat infor-
mation transfer out to 20 mrad using the supporting SiN membrane
window before conducting the experiment. The hydrogen source
was Alphagas H . Imaging took place before and during exposure
2
to app. 10 mbar H2 at 608 K.
The kinetics for the second conversion (Eq. (2)) on Cu has not
been reported to our knowledge. Parts of the pathway and mech-
anism have been elucidated through conversion of intermediates,
surface studies and isotopic labelling [7–11]. Furthermore, support
for the mechanisms may be found in similar chemical conver-
sions, where the difference between the reactions is merely the
nature of the radical attached to the functional group. Cu cata-
lysts are known to catalyse the dehydrogenation of methanol as
well [12–14]; herein C1 is designated as the set of elementary
reactions occurring on a Cu surface active in methanol reforming,
while C2 has been designated as the set of elementary reactions
involved in Cu surface catalysed ethanol reforming. Comparison
between part of the C1 and C2 mechanisms on Cu has been sug-
gested [7,8,10]. The C1 mechanisms are interconnecting methanol,
water and formic acid in parallel while the C2 mechanisms are
interconnecting ethanol, water and acetic acid. The common ele-
mentary surface reactions involved in water dissociation on Cu sites
have furthermore been discussed [15–17].
In order to support the process development of the new acetic
acid process, for example to properly design the synthesis reactors
and establish the optimal reaction conditions, a kinetic model is
needed. In this paper we have focused on the kinetic order of water
and acetaldehyde in the conversion of acetaldehyde with water to
acetic acid and hydrogen (Eq. (2)) and the correlations between the
Cu crystal size and activity. DFT calculations have been made on the
dissociation of water on Cu slabs. A kinetic discussion is made on
the basis of mechanistic considerations. A power law expression is
used to model the experimental data.
An EXAFS study on in situ reduction of a 100–150 micron sieve
fraction of the crushed catalyst was conducted at the beamline X1
at HASYLAB. QEXAFS scans were recorded using a Si(1 1 1) double-
crystal monochromator in continuous screening mode between
8500 and 9700 eV. The energy was calibrated using a Cu foil. The
data were reported on an in situ cell as described in reference [18]
and background removal and data fitting were conducted with the
Win XAS software [19].
2.3. Catalyst testing
Dehydrogenation of ethanol mixtures was conducted in an
8 mm reactor tube installed in a ventilated temperature controlled
oven. The catalyst was loaded either as (a) whole pellets (12 g cylin-
drical h × d: 5 mm × 6 mm) in a single-pellet-string configuration
with 3 mm glass balls as separators or (b) as crushed pellets (4 g,
1–3 mm sieve fraction) diluted with SiC in a 1:1 ratio on volume
basis. The catalyst was activated in each case prior to the experi-
ment with 5% H2 in N2 at 523 K and a flow rate of 900 Nml/min for
7 h. The liquid reactant was fed by means of a high precision pump
(HPLC pump, LDC Analytical, ConstaMetric 3200) to an evaporator
installed inside the ventilated oven. Preheated N2 carrier gas was
added from a mass flow controller at a ratio of either 333 or 666 Nml
per ml of liquid feed to the evaporated feed before being introduced
to the reactor. The inlet temperature and the reactor wall temper-
ature were the same. The molar feed ratios of ethanol to water in
the liquid feed were 60:40, 50:50, 40:60 and 20:80. In addition, 10%
of an acetaldehyde and water mixture in a 50:50 molar ratio was
co-fed in some experiments. The temperatures ranged from 553 to
6
13 K and the operating pressure was slightly above atmospheric.
The liquid products were sampled in a condenser cooled to
58 K by a cooling circuit. The gas flow rate at the exit of the
2
. Experimental
2
2
.1. Catalyst preparation
condenser was measured by means of a flow meter and the con-
densate flow rate was calculated. Online GC was not obtainable due
to the corrosivity of the product, potentially damaging the sample
port (previous experience). The composition of the condensate was
determined by means of a gas chromatograph (GC) with a flame
ionisation detector (FID). Based on the measured composition and
the assumption that acetaldehyde would be the only condensable
escaping the condenser in significant amounts (established by anal-
ysis), a complete calculated mass balance could be established. The
water and the acetic acid concentration in the condensate were
double checked by titration.
Due to the hard to handle acetaldehyde the variation of acetalde-
hyde vs. water in the feed was in most cases obtained indirectly by
feeding ethanol and water in different ratios, having observed that
the dehydrogenation reaction of ethanol to acetaldehyde (Eq. (1))
is much faster than the dehydrogenation of acetaldehyde (Eq. (2)).
Minute side products (<0.2 mol%) count ethyl acetate, diethoxy
ethane and ppm concentrations of n-butanol. All the unloaded cat-
The Cu/SiO2 catalyst used for this kinetic study was prepared
by precipitation of Cu(NO3)2 with K CO in a suspension of HSA-
2
3
2
silica (specific surface area = 348 m /g) at pH 6.0. The precipitate
was ripened at 333 K for 1 h, then filtered and washed with hot
water until the filtrate had a conductivity of less than 0.1 mS/cm,
and finally it was dried at 353 K. The powder (45.7 wt% Cu loading,
0
5
made from the catalyst powder, which were then calcined at 623 K
for 2 h before use. In part of the experiment a sample of crushed
catalyst pellets from a 1–3 mm sieve fraction was used in place of
whole pellets.
Different batches of Cu/SiO2 were made on the same recipe.
One sample was used for equilibrium data, one for the prediction
of reaction order. The whole pellets all originated from the same
batch.
.24 wt% K) was mixed with 3.3 wt% graphite as a lubricant and
mm × 6 mm (h × d, height × diameter) cylindrical pellets were

B. Voss et al. / Applied Catalysis A: General 402 (2011) 69–79
71
alyst samples were analysed by means of XRD. The samples were
passivated in 1% O /99% N at room temperature before unloading.
2
2
For each series of kinetic experiments carried out on whole pel-
lets for a given feed composition the temperature and flow rates
were varied to provide kinetic data at varying contact times, and at
each temperature the first flow rate was repeated in the end before
changing the temperature in order to follow the deactivation of the
catalyst. Each data point was compensated for deactivation assum-
ing that the rate equations of both Eqs. (1) and (2) were influenced
to the same degree, based on the Eq. (2) activity. The degree of
deactivation was 0–50% dependent on temperature.
The resulting kinetic data sets comprise numbers for feed and
effluent compositions, catalyst mass, inert mass, reactor length,
temperature, pressure and flow rate. To include thermal transport
effects and efficiently combine the analysis of the two consecutive
conversions according to Eqs. (1) and (2) the kinetic data obtained
were evaluated based on the data sets by contemporary integration
in a one-dimensional wall cooled plug flow reactor model on the
suggested power-law kinetics. The one-dimensional reactor soft-
ware is developed in-house but in principle any one-dimensional
reactor modelling software with appropriate integration accuracy
may be used for this purpose. With the above mentioned one-
dimensional reactor model it was possible to perform integration
over the catalyst mass for each of the experiments, n = 1, 2, 3, . . .,
Nexp, where Nexp is the total amount of proper experiments, thereby
calculating the product outlet concentrations.
Fig. 1. EXAFS spectra obtained during in situ reduction of 50 mg 100–150 micron
sieve fraction Cu/SiO2 catalyst in 5% H2 (flowing at a rate of 10 Nl/(min g)) at room
temperature (298 K, RT), 488, 498 and 579 K. The spectrum at 579 K corresponds to
Cu(0).
of the Cu crystals along the [2 0 0] direction (Cu D[2 0 0]) just after
reduction, Fig. 3b shows an XRPD spectrum for a specific catalyst
sample, and Fig. 3c shows the initial acetic acid formation rate of
whole catalyst pellets compared with crushed catalyst pellets. Fur-
ther sintering was observed after a number of hours on stream for
whole pellets and crushed catalyst pellets.
In-house DFT calculations show that water preferably disso-
ciates on Cu(2 1 1) surfaces, or step sites, the density of which
decreases with the Cu particle size. The Cu(2 1 1) surface calcula-
tions give an energy barrier of 1.3 eV (125 kJ/mol) for the water
dissociation reaction.
An error function was defined with respect to all the data sets
evaluated and to both the acetaldehyde and the acetic acid outlet
concentrations in the individual data sets. The error function (Eq.
(
5)) expresses the sum of the squared sum of the relative deviation
between the calculated outlet concentrations, ycalc, of acetaldehyde
and acetic acid according to Eqs. (1) and (2) and the corresponding
measured concentrations, yexp, at the outlet of the reactor.
3
.2. Pathway and equilibriums
In order to establish the pathways and equilibriums of the reac-
Nexp
ꢁ
ꢂ
2
tion system of Eqs. (1) and (2) a reaction profile experiment was
initially made at high to very low liquid weight hourly space veloc-
ities, LWHSV (g/(g h)), expressing the load of liquid feed per time
unit on the catalyst mass.
ꢀ
ꢀ
ycalc,i − yexp,i
yexp,i
SSQ =
(5)
i=HAc, HOAc
n=1
By means of the Complex Box method and by varying the kinetic
coefficients and activation energies in the suggested kinetic expres-
sions used for calculating the outlet concentrations, a minimisation
of the error object function value and a set of the correspondingly
optimised variables were obtained.
Fig. 4 shows the reaction profile found for the conversion of a
feed mixture with the molar ratio of ethanol to water of 40:60.
The concentration profile in Fig. 4 (ethanol = EtOH, water = H O,
2
hydrogen = H2, acetaldehyde = HAc, acetic acid = HOAc) with
acetaldehyde going through a maximum confirms acetaldehyde as
a typical intermediate. As may be gathered, the reaction rate of the
3
. Results
3.1. Catalyst characterisation results
XANES and EXAFS showed that the reduction of the CuO in the
fresh catalyst sample the Cu/SiO2 catalyst started at 488 K and was
finalised at 510 K. Fig. 1 shows a selection of the EXAFS spectra
recorded during the in situ reduction of the test sample.
Obviously complete reduction of Cu(II) to Cu (0) is obtained
within a narrow temperature window leaving copper in its metallic
state.
The formation of Cu particles on the surface of the SiO2 support
upon activation is supported by the analysis of the TEM images.
Fig. 2 shows the Cu crystals formed on the SiO2 surface when sub-
jected to 11 mbar H2 at 608 K.
Further investigation of the Cu crystal size by XRD (Scherrer
equation) showed that substantially larger Cu crystals were formed
on the whole pellets during reduction (160 A˚ compared to 110 A˚ on
crushed pellets). Likewise the growth rate of the Cu crystals seems
to be influenced by the size of the catalyst particle. Sintering due
to water formed during reduction may be the explanation for the
larger Cu crystals found on the whole pellets. Fig. 3a shows the size
Fig. 2. TEM image showing Cu crystals of average size>100 A (dark circular) on SiO2
˚
◦
support (brighter shades) after in situ reduction at 335 C in 10.5 mbar H2.

7
2
B. Voss et al. / Applied Catalysis A: General 402 (2011) 69–79
Fig. 4. The reaction profile of a 40:60 ethanol/water mixture diluted with N2 to
5
5
0 mol% fed into a bed of crushed Cu/SiO2 catalyst at an outlet temperature of
88 K at close to atmospheric pressure. Minor side-products e.g. ethyl acetate and
n-butanol have been neglected in the figure. The concentrations (mol%), of the
acetaldehyde (HAc) and acetic acid (HOAc) products may be found on the right
abscissa.
where pH , p
and pCH3CH2OH are the partial pressures of
2
CH3CHO
hydrogen, acetaldehyde and ethanol, respectively, and pꢂ is the
gas standard state pressure (1.013 bar = 1 atm). At low LWHSV the
approximate reaction quotient as calculated by Eq. (6) for the
smoothed data at 588 K is about 0.7, at a total operating pressure
of 1.05 bar.
The degree of conversion of acetaldehyde with water to acetic
acid and hydrogen according to Eq. (2) was in all kinetic experi-
ments so low that the reaction was far from equilibrium.
Feeding a mixture of acetaldehyde, being an intermediate, and
water to the catalyst gave no ethyl acetate side-product, while
acetic acid was produced, and the production rate of acetic acid
was higher than if a mixture of ethanol and water was used as a
feed. The result indicates that acetic acid may be produced directly
from acetaldehyde and water, and that this is the primary route to
acetic acid from ethanol and water.
3.3. Kinetic modelling
While the differential reactor type is the most preferred for
kinetic investigations, sound approaches may be obtained also
through integral reactors as long as a broad range of low to high
degrees of conversions are obtained for a variation of feeds at dif-
ferent temperatures. With two consecutive reactions and minor
side-reactions taking place, the smoothest analysis is obtained by
using computer modelling. The reaction rate according to the reac-
tion in Eq. (2) is so low that diffusion limitation is not controlling
the observed rates even on whole catalyst pellets of the size used
in our experiments. In such case, the application of a single-pellet
string reactor type is favourable in that the reactor may be mod-
elled more easily [20]. With a limited reactor volume and a slow
reaction in question the degree of conversion would be too low for
modelling of crushed catalyst at realistic ethanol partial pressures.
By applying a data fitting program using a non-linear optimi-
sation method on the sum of squared relative errors between the
observed and modelled product concentrations the kinetic param-
eters may be fitted.
Fig. 3. (a) The Cu crystal size D[2 0 0] analysed by ex situ XRPD after reduction and
operation. The resulting Cu crystal size depends on the catalyst particle size and
time on stream. HOS is hours on stream. (b) An ex situ XRPD for a 1.4 mm Cu/SiO2
catalyst particle reduced in 5% H2 and operated for 61 h on a 40:60 ethanol/water
molar feed diluted to 50% by nitrogen. (c) Arrhenius plot of the initial rates of acetic
acid formation vs. inverse absolute temperature for a 40:60 ethanol/water feed.
dehydrogenation reaction of ethanol (Eq. (1)) is much faster than
the dehydrogenation of acetaldehyde with water to acetic acid
and hydrogen (Eq. (2)). Furthermore, the profile seems to level out
at low LWHSV, indicating that the equilibrium is approached. In
a kinetic evaluation the reaction rate is typically corrected with
(
1 − ˇ) for considering a reaction approaching equilibrium, if it
describes a first order conversion, where ˇ is the reaction quotient
divided by the corresponding equilibrium constant. Therefore, the
equilibrium temperature function had to be estimated for this
purpose.
Present experiments were conducted at initial ethanol partial
pressures 0.11–0.26 bar and water partial pressures 0.18–0.45 bar.
Acetaldehyde was in some cases co-fed at an initial partial pressure
of 0.05 bar. In order to investigate its influence on the acetic acid for-
mation rate the acetaldehyde partial pressure at low contact time
The approximate reaction quotient, Qappr, may be expressed as:
pH2 · pCH3CHO
Qappr =
(6)
pCH
· pꢂ
3CH2OH

B. Voss et al. / Applied Catalysis A: General 402 (2011) 69–79
73
Fig. 6. The modelled conversions and the radial averaged temperature (lines) and
the experimental values for acetaldehyde (square) and acetic acid (cross); y is indi-
cated in mol%. The partial pressure of ethanol = 0.26 bar and the partial pressure of
water = 0.39 bar in the feed. The feed/wall temperature = 573 K.
Fig. 5. Log–log plots of the acetaldehyde space-time yield (STY) vs. the average
ethanol concentration and the acetic acid STY vs. the average water concentration,
respectively, at 563 K and atmospheric pressure as a simple approach for determin-
ing the reaction orders in the kinetic expressions for Eqs. (1) and (2) for crushed
pellets. Averages are arithmetic inlet to outlet.
ethanol to acetaldehyde (Eq. (1)) was assumed to be of first order
in ethanol in the further data analysis.
According to the same method and based on the same data the
log–log plotting of the STY of acetic acid, STY HOAc, vs. the aver-
age water vol% was made (Fig. 5) in order to indicate the reaction
order with respect to water in Eq. (2). As mentioned above, during
these experiments the acetaldehyde concentration in the effluent
was almost constant. If as an approximation the acetaldehyde con-
centration profile is assumed constant in this series of experiments,
we find a reaction order of water of 1.05, i.e. a close to first order
dependence.
was further varied indirectly by varying the initial ethanol partial
pressure. The temperature was varied between 553 and 613 K.
3.3.1. Crushed catalyst pellets
As mentioned above, the operation of the experimental reac-
tor at close to isothermal conditions may be obtained with a bed
of crushed catalyst pellets diluted with SiC, which has high ther-
mal conductivity, 360 W/mK. Such experiments were conducted in
order to observe the reaction order effects while minimising the
effect of non-isothermal behaviour.
In one part of the experiment – on crushed catalyst – the
ethanol/water ratio was fixed while its feed rate was changed keep-
ing the nitrogen carrier gas flow rate and the temperature constant.
Hereby, the partial pressures of the water and ethanol feed were
changed, while the exit acetaldehyde concentration was almost
constant.
The influence of the two varied reactants was investigated by
plotting the space-time yields (STYs) of the products vs. the average
of the reactants concentration log–log in a series of experiments
conducted at 563 K, where the degree of conversion is rather low,
hereby approaching differential analysis. Fig. 5 shows the plots to
determine the reaction order of Eq. (1) with respect to ethanol and
the reaction order of Eq. (2) with respect to water.
3
.3.2. Whole pellets
Modelling an example of a conversion profile based on a first
order reaction rate for Eq. (1) confirms that isothermal operation
is not achieved. Fig. 6 shows the modelled conversion and tem-
perature profiles for an example, where y is the concentration of
acetaldehyde and acetic acid in mol%, respectively.
However, by means of the fact that ꢀt approaches 0 at low con-
tact times the initial rates, rini, defined as the reaction rate of the
feed composition of acetic acid formation for an infinitely low con-
centration of acetaldehyde may approximately be established from
the depiction of yHOAc vs. 1/SV irrespective of the non-isothermal
behaviour. The rini values are found as the tangent slopes in 1/SV = 0.
Fig. 7 shows the log–log plot of experimental results of a series
of feeds on whole pellets where the ethanol partial pressure was
kept constant while the water partial pressure was varied. In effect,
this variation of water partial pressure at constant ethanol implies
a constant acetaldehyde concentration at a given, low contact time,
as the Eq. (1) reaction rate is very high and is assumed being depen-
dent on ethanol alone.
An approximation of the reaction order of the first rate equation
(
Eq. (1)) with respect to ethanol was obtained as the slope of the
Note that both on crushed and sieved catalyst and whole catalyst
pellets the reaction order of water is close to 1.
log–log depiction of the acetaldehyde STY, STY HAc, vs. the average
ethanol concentration (arithmetic inlet to outlet) in vol%. The STY
HAc value includes the amount of acetaldehyde which had further
been converted to acetic acid. The reaction order with respect to
ethanol is found to be 0.8 by this analysis. Apart from not being an
actual but rather an average reaction rate, the STY does not com-
pensate for the approach to equilibrium. ˇ for Eq. (1) ranged from
In a further series of experiments acetaldehyde was co-fed lead-
ing to a remarkable increase of the initial reaction rate to acetic
acid. Therefore, overall a rate expression for Eq. (2) is suggested
depending on both water, with a close to first order dependence,
and acetaldehyde partial pressures.
0.05 to 0.15.
In the study by Tu et al. [4] conducted at 523–583 K the conver-
3.4. Parameter fitting
sion of ethanol to acetaldehyde was found to take place according
to a first order reaction in ethanol. Based on this literature evidence
and the above finding, the rate equation for the dehydrogenation of
As mentioned, the conversion of ethanol to acetaldehyde was
assumed to be a first order reaction with respect to ethanol. The

7
4
B. Voss et al. / Applied Catalysis A: General 402 (2011) 69–79
Fig. 7. The initial acetic acid formation rate vs. the partial pressure of water at
toven = 573 K and constant ethanol pressure (0.26–0.265 bar) in the feed on whole
catalyst pellets.
power law expression in accordance herewith may be expressed
as follows:
ꢃ
ꢄ
−
E1
−rCH
= A · exp
· pCH3CH2OH · (1 − ˇ1)
(7)
3CH2OH,1
1
RT
where A1 is the pre-exponential factor in mol/(g h bar), E is the
1
apparent activation energy in kJ/mol, pCH3CH2OH is the partial pres-
sure of ethanol in bar, and ˇ1 is the observed reaction quotient
divided by the equilibrium constant for Eq. (1).
Even with a certain reduction of the reaction rate due to diffusion
limitation, the order of the reaction remains 1.
The reaction according to Eq. (2) was given a power law rate
equation, with a dependence of acetaldehyde and water:
ꢃ
ꢄ
−
E2
−
rCH3CHO,2 = A2 · exp
· pCH3CHOꢃCH3CHO · pH2OꢃH2
O
· (1 − ˇ2)
(8)
Fig. 8. Parity plots of calculated vs. measured concentration of the products
acetaldehyde and acetic acid based on whole catalyst pellets. Upper: acetalde-
hyde. Lower: acetic acid. The plot points are sorted in three series corresponding to
the reactor wall/feed temperature: diamonds 553 K, filled squares 573 and crosses
593 K.
RT
ꢄꢃ
where A2 is the pre-exponential factor in mol/(g h bar ), E2 is the
apparent activation energy in kJ/mol, pCH3CHO is the partial pressure
of acetaldehyde in bar, pH2O is the partial pressure of water in bar,
ꢃCH
and ꢃH2O are the reaction orders with respect to acetalde-
3CHO
hyde and water, and ˇ2 is the observed reaction quotient divided
by the equilibrium constant for Eq. (2). The approach to equilibrium
expressed by ˇ2 was in all cases almost negligible.
Using a one-dimensional reactor modelling program the above
power law expressions were evaluated.
The objective here was to study the kinetics for the conversion
of acetaldehyde with water to acetic acid and hydrogen. As shown
above in Fig. 6, the reactor was not operated completely isother-
mally. The integral analysis of the two rate expressions, Eqs. (7) and
(8), was however accounting also for the actual temperature pro-
In the rate expression, Eq. (7), the pre-exponential factor, A1,
file in the test reactor. From above parity plots Fig. 8 it is seen that
no systematic deviation occurs with reactor wall/feed temperature
variation. This indicates a good validity of the apparent activation
energy found for the studied conversion kinetics.
For comparison, further kinetic data as to conversion, selectivity
and stability of the Cu/SiO2 catalyst are enclosed in Appendix A.
and the apparent activation energy, E , were allowed to vary, while
1
in the rate expression, Eq. (8), both the pre-exponential factor, A2,
the apparent activation energy E2 and the exponents, ꢃCH3CHO and
ꢃH2O, were defined as variables.
The optimised parameters found for the 71 data sets with
ethanol partial pressures 0.11–0.26 bar, water partial pressures
0
.18–0.45 bar and acetaldehyde co-fed at up to 0.05 bar, diluted
with N2 carrier gas, and the temperatures ranging from 553 to
13 K, were as follows:
4. Discussion
6
4.1. Side-reactions and equilibrium
A = 3.53E mol/(g h bar) E1 = 43.7 kJ/mol
1
3
Using the present Cu/SiO2 catalyst, only very limited side-
A2 = 20.4 mol/(g h bar1.34)
reaction conversion to ethyl acetate and butanol via Eqs. (3) and
4) have been observed for the catalyst tested. In literature such
(
ꢃCH
= 0.45 ꢃH = 0.89 E2 = 33.8 kJ/mol
3CHO
2O
side-products are typically reported [7,9].
At higher temperatures the ethanol may dehydrate to ethylene
[21]. However, no such side-production of ethylene was observed
in the temperature range 553–613 K.
Fig. 8 shows the parity plots for acetaldehyde and acetic acid
respectively sorted in 3 selected reactor wall/inlet temperatures.

B. Voss et al. / Applied Catalysis A: General 402 (2011) 69–79
75
Regarding the ethanol dehydrogenation (Eq. (1)), the approx-
imate reaction quotient value of about 0.7 was found in present
investigation at 588 K for low LWHSV, approaching equilibrium.
This value corresponds closely to the equilibrium constant of 0.8
at 588 K obtained theoretically from the Stull et al. [5] thermo-
chemical values as calculated by Happel et al. [6]. Happel argues
that the departure from ideal behaviour is insignificant for Eq.
Adsorbed oxygen or hydroxyl needed for these oxidations may
be obtained by the dissociation of adsorbed water to OH* and H* on
the Cu surface, followed by 2 adsorbed hydroxyls forming adsorbed
water and atomic oxygen, steps XIII–XVI. Dihydrogen gas is formed
by desorption of two adsorbed hydrogen, step XVII. The dissociation
of water, step XIV, is known to be slow [13] and rate determining
under typical water-gas shift conditions.
(
1) at the conditions given allowing the direct comparison of the
Accordingly, the reaction mechanism of ethanol dehydrogena-
tion with water to acetic acid and hydrogen via acetaldehyde (Eqs.
(1) and (2)) may be described through the discussed elementary
steps of steps I–XVII.
approximate reaction quotient (Eq. (6)) at low LWHSV with the
equilibrium constant value. The corresponding standard entropy
and enthalpy of formation of acetaldehyde proposed by Stull et al.
[
5] are 264.2 J/(mol K) and −166.4 kJ/mol. Happel et al. [6] made a
series of experiments based on the approaching of the Eq. (1) equi-
librium from both sides at temperatures 456–533 K. They compared
their own experimental values against various literature thermo-
chemical and experimental data for this reaction. Happel et al. [6]
also found that their equilibrium data agreed best with the thermo-
dynamic values reported by Stull et al. [5]. These standard entropy
and enthalpy values for acetaldehyde were therefore used in our
data evaluation.
4.3. Rate determining step
The dissociation of water on Cu surfaces (step XIV) is known [13]
to be rate determining under typical water gas shift conditions in
conjunction with the methanol reforming, C1 system.
We may note, however, that whereas a high activation barrier
(84 kJ/mol) for the dehydrogenation of formaldehyde into formyl is
predicted by Shustorovich [14] for the C1 system, the ready forma-
tion of adsorbed acetyl CH CO* from acetaldehyde was observed in
3
4.2. Mechanism
the C2 system on a Cu surface [11].
Furthermore, it is reported by Iwasa and Takezawa [7] that in the
C1 system the dehydrogenation step of methanol to formaldehyde
is rate determining over its further conversion to methyl formate or
carbon dioxide, via formate. For the C2 system the dehydrogenation
step to acetaldehyde is much faster than its further steps to acetic
acid.
As the formation rate of acetic acid is higher from acetaldehyde
than from ethanol, the alternative overall ethoxy route via step VII
seems to not be the dominant. It is likely that the rate determin-
ing step is either of the oxidations in step VI, step VIII or step IX,
or the dissociation of water, step XIV. Either which oxidant, O* or
OH*, reacts with adsorbed acetaldehyde or acetyl, water must dis-
sociate via elementary step XIV on the Cu surface to provide for
these.
Up to now no kinetic model, empirical or microkinetic, for
the conversion of acetaldehyde to acetic acid has been suggested
in the literature. But some of the elementary steps have been
reported. From one side it is difficult to determine a mecha-
nism from an empirical kinetic expression. The other way round
with some mechanistic information it gets easier to propose an
empirical expression with greater validity. Reference is made
to the Appendix B where mechanistic considerations are made
based on the literature [7–13], comprising the corresponding
C1 mechanisms. Appendix B also contains Table 1 showing the
considered elementary reaction steps I–XVII on the Cu surface.
Reference to these steps I–XVII in Table 1 are hereinafter made
directly.
As learned in Appendix B, the formation of CH CHOO* may,
3
with a basis in the C1 elementary steps, be explained through
the oxidation of an adsorbed acetaldehyde with adsorbed O*, step
4.4. Reaction order
VIII or adsorbed OH* to H* and CH CHOO*, step IX, or it may
In analysing the Langmuir–Hinshelwood type reaction rate
3
be explained through a direct ethoxy route: ethanol dissociation
into adsorbed ethoxy and H*, the ethoxy being oxidised with
expressions for a proposed rate determining step, the denominator
ꢅ
term (1 +
adsorption terms) expresses the reduction of activ-
adsorbed O* to CH CHOO* and adsorbed H*, see steps I, II + VII.
ity due to occupation of free sites by adsorbed species. Askgaard
et al. [12] report that the coverage of the free Cu sites in the C1
system highly depends on the pressure of the active components.
At 2 bar and 500 K the relative abundance of free sites is 0.88,
3
The CH CHOO* then abstracts ␣-hydrogen, step X, and the acetate
3
formed reacts with adsorbed hydrogen to acetic acid which desorbs,
steps XI–XII.
Table 1
Elementary reaction steps suggested for the dehydrogenation of ethanol via acetaldehyde to acetic acid on Cu, steps VII–XI based on C1 elementary reactions on Cu. * signifies
a free Cu site.
Elementary surface reaction
Type
Reference
I
II
III
IV
V
VI
VII
VIII
IX
CH3CH2OH(g) + * = CH3CH2OH*
CH3CH2OH* + * = CH3CH2O* + H*
CH3CH2O* + * = CH3CHO* + H*
CH3CHO* = CH3CHO(g) + *
CH3CHO* + * = CH3CO* + H*
CH3CO* + OH* = CH3COOH* + *
CH3CH2O* + O* = CH3CHOO* + H*
CH3CHO* + O* = CH3CHOO* + *
CH3CHO* + OH* = CH3CHOO* + H*
CH3CHOO* + * = CH3COO* + H*
CH3COO* + H* = CH3COOH* + *
CH3COOH* = CH3COOH(g) + *
H2O(g) + * = H2O*
Ethanol adsorption
[8,11]
[8,11]
[8,11]
[7,8,11], this work
[11]
[11]
[12]
[12]
[12]
Hydroxyl hydrogen abstraction
Ethoxy ␣-hydrogen abstraction
Acetaldehyde desorption
Acetaldehyde ␣-hydrogen abstraction
Acetyl hydroxyl oxidation
Ethoxy oxidation
Acetaldehyde oxidation
Acetaldehyde hydroxyl oxidation
Acetate formation
Acetic acid formation
Acetic acid desorption
Water adsorption
Water dissociation
X
XI
[12]
[12]
XII
XIII
XIV
XV
XVI
XVII
[7,12], this work
[7,12,16,17]
[7,12,16,17]
[7,12,16,17]
[7,12,16,17]
[7,12,16,17], this work
H2O* + * = OH* + H*
2OH* = H2O* + O*
OH* + * = O* + H*
2H* = H2(g) + 2*
Hydroxyl disproportionation
Hydroxyl dissociation
Dihydrogen formation

7
6
B. Voss et al. / Applied Catalysis A: General 402 (2011) 69–79
balanced primarily by coverage of adsorbed hydrogen. It may be
argued then that at 0.5 bar of reactants and products and 553 K,
i.e. four times lower pressure and a higher temperature, the free
sites are by far the most abundant species. Assuming that the free
sites are dominant, the adsorption terms become negligible, leav-
ing basically a rate equation equal to the numerator term of the full
Langmuir–Hinshelwood expression.
Whereas the conversion of acetaldehyde to acetic acid is rather
slow the dehydrogenation of ethanol to acetaldehyde is so fast that
pore diffusion limitation could influence the observed rate. Being a
first order reaction intrinsically one would, however, still expect
a first order behaviour in ethanol for the ethanol dehydrogena-
tion under the influence of pore diffusion limitation, the observed
activation energy being somewhat lower, depending on the pore
branching.
We find an observed activation energy of 44 kJ/mol for the
ethanol dehydrogenation, Eq. (1). Tu et al. finds an intrinsic activa-
tion energy for ethanol dehydrogenation to acetaldehyde on pure
Cu of 51 kJ/mol. The two values are basically consistent keeping in
mind that our value is slightly lowered due to the diffusion limi-
tation in our whole pellet experiments. Shustorovich and Bell [14]
calculates the activation energy barrier of the formation of methoxy
from adsorbed methanol to be around 38 kJ/mol. The energy of
hydrogen abstraction from ethanol should be similar; thus the
calculated value by Shustorovich and Bell is consistent with the
activation energies for Eq. (1) found experimentally.
The approximate determination of the apparent reaction order
of 0.8 (see Fig. 5) in ethanol is consistent with the finding by Tu
et al. [4] who report the reaction order with respect to ethanol to
be 1.
Guan and Hensen [15] investigated the ethanol dehydrogena-
tion on a silica supported gold catalyst. The reaction rate was found
to be solely first order with respect to ethanol. The first order
ethanol dependence without the dependence of hydrogen is con-
sistent with a Langmuir–Hinshelwood expression with negligible
adsorption terms assuming that the first hydrogen abstraction is
the rate determining step [22].
where the observed numerator has a reaction order of 1 both with
respect to water and acetaldehyde. Furthermore a reaction order
with respect to hydrogen of −½ is found assuming acetaldehyde
oxidation with adsorbed hydroxyl, OH*, to be rate determining,
while the hydrogen reaction order was found to be −1 for acetalde-
hyde oxidation with atomic oxygen, O*, or acetyl oxidation with
adsorbed hydroxyl as the rate determining step, respectively.
Elementary steps XIII–XVII were studied on a Cu[1 1 1] surface
by Phatak et al. [16] and it was found that the coverage of OH* is a
factor of 106 as high as the O* coverage. In a more recent study by
Chen et al. [17] it was found that hydroxyl even binds stronger on
Cu[3 2 1] step sites. Therefore, most likely the oxidation takes place
by means of adsorbed hydroxyl, OH*.
As mentioned, the intrinsic reaction rate model has hydrogen
dependence as well. Hydrogen was varied as a process parameter
however being proportional to the acetaldehyde partial pressure
in most cases. Assuming that the rate determining step is the oxi-
dation of adsorbed acetaldehyde with hydroxyl, the corresponding
Langmuir–Hinshelwood expression may explain the observed near
0.5 dependence of acetaldehyde if the apparent reaction order
reflects in reality an overall acetaldehyde and hydrogen depen-
dence and if the adsorption terms are truly negligible. Eq. (9)
expresses the forward Langmuir–Hinshelwood reaction rate for
Eq. (2) being far from equilibrium, based on the above assump-
tion.
k · (pH2O · pCH3CHO)/p1
/2
H
2
−rCH
=
(9)
3CHO,L−H,forward
ꢅ
2
(
1 +
adsorption terms)
where k is the resulting apparent rate constant and p CHO, pH2O,
CH3
pH2 are the partial pressures of acetaldehyde, water and hydro-
gen.
Assuming the pathway would go through acetyl oxidation with
hydroxyl, the overall acetaldehyde and hydrogen reaction order of
the numerator alone is 0, which does not correspond well with
the overall dependence of acetaldehyde and hydrogen found by
optimisation.
For the acetaldehyde conversion, Eq. (2), we find an apparent
reaction order of water close to unity. However, a dependence of
acetaldehyde was found as well. A reaction order for the reaction
in Eq. (2) close to unity for water (see Fig. 5) may reflect that the
elementary step XIV is rate limiting. Under this assumption, a reac-
tion order higher than 0 with respect to acetaldehyde reflects that
the dissociation of water in step XIV is not much slower than is
the oxidation of acetaldehyde or acetyl. However, the dissociation
energy of adsorbed water on step sites was found to be 125 kJ/mol
by means of DFT calculations, which is far from the apparent acti-
vation energy 34 kJ/mol as found experimentally for Eq. (2). Wang
et al. have reported water dissociation energies of 100–123 kJ/mol
for the Cu(1 1 0), Cu(1 0 0) and Cu(1 1 1) surfaces [23]. Due to this
discrepancy, we do not believe that water dissociation is the rate
limiting step in Eq. (2).
Colley et al. found that the activation energy for the dehydro-
genation of adsorbed acetaldehyde to acetyl is much smaller than
the desorption of acetaldehyde leading to low concentrations of
acetaldehyde in the gas phase. We do, however, see very high con-
centrations of acetaldehyde in our reaction system where adsorbed
acetaldehyde or acetyl possibly reacts with dissociated water, O*
or OH*. This indicates that the acetaldehyde or acetyl oxidations
Furthermore, a positive dependence of the total pressure would
be expected from any of these terms. However, increasing the oper-
ating pressure by a factor of 10 caused serious deactivation of
the catalyst, presumably due to sintering, so this information was
veiled by an overlaying deactivation phenomenon.
The dependence in acetaldehyde (and hydrogen) could thus be
explained by the oxidation of adsorbed acetaldehyde with hydroxyl
as the primary pathway and the rate determining step, in agree-
ment with the suggestion by Iwasa and Takezawa [7]. Citing Ovesen
et al. [13] different mechanisms can lead to the same overall kinetic
expression. But based on more indications, the two most likely
pathway candidates finally down-selected actually have different
reaction order sets, of which one set has a superior fit with the
power-law expression found. Best consistence is therefore found
for acetaldehyde oxidation with hydroxyl.
The validity of the derived power-law rate equation for the
conversion of acetaldehyde (Eq. (8)) may be limited to feeds with
approximately equimolar amounts of acetaldehyde and hydrogen.
4.5. Reduction influence
(
Table 1, steps VI, VIII or IX) have considerable activation barriers.
As mentioned, in aiming for more isothermal operation some
experiments were conducted on crushed catalyst under which
conditions the catalyst seemed to have a 3 times higher intrin-
sic activity for the conversion of acetaldehyde to acetic acid. The
discrepancy cannot be explained by diffusion limitation, as judged
from the observed reaction rate, reaction order and the apparent
activation energy for the two catalyst sizes tested.
Therefore, it is most likely that either the oxidation with O* or OH*
of adsorbed acetaldehyde or hydroxyl oxidation of acetyl is rate
determining.
The assumption of acetaldehyde or acetyl oxidation
being the rate determining step does indeed result in
Langmuir–Hinshelwood expression, based on one site type,
a

B. Voss et al. / Applied Catalysis A: General 402 (2011) 69–79
77
Fig. 9. (a) STYHOAC, (b) ethanol conversion and (c–f) product selectivities vs. time on stream for reaction conditions LWHSV = 0.4–0.5 g/(g h) for a 40:60 ethanol:water feed at
◦
tbed = 290–300 C.
Using the information from the DFT calculations on Cu surfaces
the cleavage of water primarily takes place on step sites. Accord-
ing to Guan and Hensen [15], studying dehydrogenation of ethanol
on gold surfaces, the density of step sites is expected to depend
inversely on the particle size above a certain optimum nanoparticle
size; and for constant Cu amount the Cu surface area also depends
inversely on Cu crystal size.
Above best consistence with experiments was found for
acetaldehyde oxidation with hydroxyl as the rate determining step,
having a high dependence of water in the corresponding rate equa-
tion. If Cu step sites are the predominant active sites for the cleaving
of water the primary coverage of hydroxyl species used in the oxi-
dation of acetaldehyde is also on the step sites. The almost doubling
of the Cu crystal size may then, together with the inverse depen-
dence of the density of Cu step sites, account for the reduction of
the reaction rate with a factor of 3.
is induced under reduction. It may be argued that reduction water
plays a role.
5. Conclusion
Cu/SiO2 is a selective and well-suited catalyst for acetic acid
synthesis from ethanol, the Cu being present as metallic copper
particles on the silica support as confirmed by TEM, EXAFS and
XRD. DFT calculations show the preferential cleavage of water at
Cu step sites.
The reaction quotient of 0.7, approximating the equilibrium con-
stant, at 588 K for ethanol dehydrogenation to acetaldehyde and
hydrogen was found to agree well with the equilibrium constant
value estimated from the thermochemical data reported by Stull
et al. [5].
The suggested reaction order of 1 with respect to ethanol for the
dehydrogenation of ethanol to acetaldehyde was supported by the
experiments in present study. The apparent activation energy of the
non-intrinsic kinetic expression for dehydrogenation of ethanol to
acetaldehyde is as expected slightly lower than the intrinsic value
reported in literature for pure Cu.
The XRD on the catalyst reveal that the catalyst is influenced
by a sintering phenomenon during reduction such that Cu crystals
of almost double size are obtained for whole pellets as compared
to crushed pellets (1–3 mm sieve fraction). Similarly, the growth
rate during operation is affected but the most predominant effect

7
8
B. Voss et al. / Applied Catalysis A: General 402 (2011) 69–79
An intrinsic empirical kinetic power law expression for the
The initial mechanism for the abstraction of hydrogen from
ethanol to obtain acetaldehyde has been studied in the literature
[8,11].
The initial conversion of ethanol to acetaldehyde is suggested to
take place via steps I and II followed by another hydrogen abstrac-
tion step, see step III. Steps I and II in combination and step III
were shown in temperature dependent ethanol adsorption studies
dehydrogenation of acetaldehyde to acetic acid was derived. The
apparent reaction order of water close to unity and the apparent
reaction order of acetaldehyde and hydrogen overall of 0.45 tenta-
tively suggest that hydroxyl based oxidation of acetaldehyde can
be the rate determining step.
The relationship between the Cu crystal size and the intrinsic
activity found is consistent with the suggestion of the hydroxyl
oxidation as the rate determining step, and the preferential cleaving
of water on step sites as found by DFT.
(TDSS) on Cu/Cr O3 by Colley et al. [11]. The dominant abstrac-
2
tion of ␣-hydrogen from adsorbed ethoxy on Cu in step III was
confirmed by Chung et al. [8] by isotopic labelling.
As observed in our experiment acetaldehyde desorbes as an
intermediate, step IV.
Acknowledgments
Furthermore, Colley et al. found that acetaldehyde easily dehy-
drogenates into acetyl, step V, and that acetyl reacts with ethoxy
to ethyl acetate. Very small amounts of ethyl acetate were found
in our work in the conversion of ethanol over Cu/SiO2 indicating a
low coverage of either ethoxy or acetyl. Assuming a considerable
coverage of acetyl the reaction of an adsorbed acetyl with hydroxyl
We thank Berit Hinnemann and Burcin Temel, Haldor Topsøe
A/S, for their efforts on DFT calculations as well as Charlotte Ovesen,
Haldor Topsøe A/S, for mechanistic discussions, Robin Christensen
and Xenia Faber for XRPD analysis and Stig Helveg, Haldor Topsøe
A/S, for in situ TEM analysis of the Cu/SiO2 catalyst.
This work is part of an industrial PhD study under the CHEC
and PROCESS Research Centers at the Department of Chemical and
Biochemical Engineering at the Technical University of Denmark in
collaboration with Haldor Topsøe A/S, financed by Haldor Topsøe
A/S and the Danish Council for Technology and Innovation.
Finally we acknowledge HASYLAB at DESY, Hamburg, for beam-
time and DANSCAT and the EU for financial support of the beamtime
to CH COOH* may be suggested, step VI.
3
Iwasa and Takezawa [7] suggest the nucleophilic addition of
water (OH) to adsorbed acetaldehyde as the pathway for acetic
acid formation. They further made comparisons to the methanol
system studied over Cu based catalysts and found several similari-
ties. Therefore, in screening for more possible pathways inspiration
has been found in the corresponding C1 system conversions, i.e.
methanol (MeOH) reforming (MeOH + H O = 3H + CO ) or synthe-
(
Contract AII3-CT-2004-506008).
2
2
2
sis being catalysed on Cu surfaces in conjunction with the water gas
shift reaction (CO + H O = H + CO ). Support to the valid compari-
Appendix A.
2
2
2
son of the methanol and ethanol systems of reaction pathways over
Cu catalyst may be found in a work by Shimada et al. [10] where the
oxidation of formaldehyde and acetaldehyde, were studied. Chung
et al. [8] expects analogies between ethanol and methanol dehy-
drogenation mechanisms due to their identical distance between
the ␣-hydrogen and the oxygen under the consideration of config-
urational constraints.
In this appendix we provide further considerations on the exper-
imental basis and some additional catalytic data.
In a first round, the Cu/SiO2 catalyst was tested as crushed
pellets sieve fraction diluted with SiC in order to establish as
close to isothermal as possible conditions for equilibrium experi-
ments. However, running on crushed catalyst results in low particle
Reynolds Number (Rep) down to 10 (Rep = d G/ꢅ, where d is the
The elementary reaction pathways for methanol and shift active
Cu catalysts have been studied by Ovesen et al. [13] and Askgaard
et al. [12]. In the C1 system, the elementary steps underlying the
synthesis of methanol over Cu catalyst, where the side-product
formaldehyde is found, is explained by the dissociation of adsorbed
h
h
hydraulic diameter, G is the gas mass velocity per cross section of
empty tube and ꢅ is the fluid viscosity) in the 8 mm reactor with
the catalyst mass and flow rate limitations given. Therefore, due to
the dependence on proper modelling most of the kinetic experi-
ments were conducted with whole pellets in a single-pellet string
configuration at Rep ≈ 60–250, making a modelling of the con-
version reasonably trustworthy. Furthermore, from the observed
reaction rate of acetaldehyde conversion to acetic acid (based on the
Weisz–Prater Modulus) no significant internal diffusion limitations
are expected.
oxidised formaldehyde hydrate, H COO*, on a free site to form
2
HCHO* and O* by Askgaard et al. [12], the methanol synthesis tak-
ing place in a C1 system, while ethanol is a C2 compound. Thus, in
the reverse direction, the oxidation of the adsorbed formaldehyde
to the oxidised formaldehyde hydrate is implicitly suggested in the
C1 system, and the further steps of its decomposition to H* and
adsorbed formate, and the final decomposition of formate to CO2
and H2 were verified at 470 K. As opposed to formate as a product
in the C1 system acetate in the C2 system is a very stable compound
which may easily be hydrogenated to acetic acid.
Fig. 9a–f shows the activity (STY), conversion and product selec-
tivity vs. hours on stream for whole pellets.
Appendix B.
Following the suggested pathways for formaldehyde to formate
by Askgaard et al. the formation of oxidised acetaldehyde hydrate
In this appendix we have a discussion on the suggested reaction
mechanisms based on literature findings.
CH CHOO* is assumed possible, being the precursor to acetate and
3
As to the reaction pathway Inui et al. [9] suggest that over
a Cu–Zn–Zr–Al–O catalyst acetic acid is produced from ethanol
through acetaldehyde in agreement with Eq. (1) but that acetalde-
hyde reacts to hemiacetal and further to ethyl acetate, which then
hydrolyses to acetic acid in disagreement with Eq. (2). Based on our
work it is assumed that acetic acid is produced primarily through
the reaction of acetaldehyde with water over a Cu catalyst, which
was also supported by Iwasa and Takezawa [7].
Breaking the conversions into elementary steps the likelihood
of the suggested reaction pathway may be elucidated.
Table 1 shows a survey of elementary reactions established or
proposed as parallels to elementary reactions for C1 conversions.
Herein * signifies a free Cu surface site.
acetic acid.
References
[
1] K. Weissermel, H.-J. Arpe (Eds.), Industrial Organic Chemistry, fourth ed., Wiley,
Weinheim, 2003, pp. 171–174.
[2] W. Foerst, Ullmanns Encyklopädie, 6. Band, Urban & Scharzenberg, München-
Berlin, 1955, p. 783.
[
3] Y. Tomiaki, S. Masami, JP57102835.
[
4] Y.-J. Tu, C. Li, Y.-W. Chen, J. Chem. Technol. Biotechnol. 59 (1994) 141–147.
[5] D.R. Stull, E.F. Westrum, G.C. Sinke, The Chemical Thermodynamics of Organic
Compounds, Wiley, New York, 1969, pp. 439–440.
[
[
6] J. Happel, J.C. Chao, R. Mezaki, J. Chem. Eng. Data 19 (2) (1974) 110–112.
7] N. Iwasa, N. Takezawa, Bull. Chem. Soc. Jpn. 64 (1991) 2619–2623.
[8] M.-J. Chung, S.-W. Han, K.-Y. Park, S.-K. Ihm, J. Mol. Catal. 79 (1993) 335–345.

B. Voss et al. / Applied Catalysis A: General 402 (2011) 69–79
79
[
9] K. Inui, T. Kurabayashi, S. Sato, N. Ichikawa, J. Mol. Catal. A 216 (2004) 147–156.
10] T. Shimada, K. Sakata, T. Homma, H. Nakai, T. Osaka, Electrochim. Acta 51 (2005)
06–915.
[16] A.A. Phatak, W.N. Delgass, F.H. Ribeiro, W.F. Schneider, J. Phys. Chem. C 113
(2009) 7269–7276.
[17] C.-S. Chen, T.-W. Lai, C.-C. Chen, J. Catal. 273 (2010) 18–28.
[18] J.-D. Grunwaldt, M. Caravati, S. Hanneman, A. Baiker, Phys. Chem. Chem. Phys.
6 (2004) 3037–3047.
[
9
[
[
11] S.W. Colley, J. Tabatabaei, K.C. Waugh, M.A. Wood, J. Catal. 236 (2005) 21–33.
12] T.S. Askgaard, J.K. Nørskov, C.V. Ovesen, P. Stoltze, J. Catal. 156 (1995)
2
29–242.
[19] T. Ressler, J. Synchrotron Radiat. 5 (1998) 118–122.
[20] D.S. Scott, W. Lee, J. Papa, Chem. Eng. Sci. 29 (1974) 2155–2167.
[21] S. Freni, N. Mondello, S. Cavallaro, G. Cacciola, V.N. Parmon, V.A. Sobyanin, React.
Kinet. Catal. Lett. 71 (2000) 143–152.
[
13] C.V. Ovesen, B.S. Clausen, B.S. Hammershøi, G. Steffensen, T. Askgaard, I. Chork-
endorff, J.K. Nørskov, P.B. Rasmussen, P. Stoltze, P. Taylor, J. Catal. 158 (1996)
1
70–180.
[
[
14] E. Shustorovich, A.T. Bell, Surf. Sci. 253 (1991) 386–394.
15] Y. Guan, E.J.M. Hensen, Appl. Catal., A 361 (2009) 49–56.
[22] R.M. Rioux, M.A. Vannice, J. Catal. 216 (2003) 362–376.
[23] G.-C. Wang, J. Nakamura, J. Phys. Chem. Lett. 1 (2010) 3053–3057.