Properties of copper chromite catalysts in hydrogenation reactions

Article abstract of DOI:10.1006/jcat.1997.1832

A pretreatment involving a 4-h reduction under H₂ at 573 K, compared to either 473 or 673 K, gave the highest specific activity for furfural and crotonaldehyde hydrogenation over an unpromoted copper chromite catalyst. Both H₂ and CO chemisorption declined monotonically with increasing reduction temperature, but turnover frequencies also showed a maximum after a pretreatment at 573 K. DRIFTS identified a band, associated with CO adsorbed on Cu⁺¹ sites, which reached a maximum intensity after a 573 K pretreatment, while XRD patterns also provided evidence for a CuCrO₂ phase and showed a continuous increase in Cu⁰ as temperature increased. Selectivity to furfuryl alcohol ranged from 35-80% and reaction orders on furfural and H₂ were near unity. In contrast, crotonaldehyde hydrogenation gave primarily butyraldehyde with only around 3% crotyl alcohol and a near zero-order dependence on crotonaldehyde, although the H₂ dependence was still near first order. IR spectra under reaction conditions revealed adsorbed crotonaldehyde, provided evidence for an unsaturated alkoxide intermediate and detected no butyraldehyde on the surface. A simple Langmuir-Hinshelwood sequence nicely explained both reactions, it was consistent with the DRIFTS results obtained during crotonaldehyde hydrogenation, and the resultant rate expression gave meaningful entropies and enthalpies of adsorption for H₂ and crotonaldehyde.

Full text of DOI:10.1006/jcat.1997.1832

JOURNAL OF CATALYSIS 171, 406–419 (1997)
ARTICLE NO. CA971832
Properties of Copper Chromite Catalysts in Hydrogenation Reactions
R. Rao, A. Dandekar, R. T. K. Baker,1 and M. A. Vannice2
Department of Chemical Engineering, Pennsylvania State University, University Park, Pennsylvania 16802-4400
Received October 7, 1996; revised July 16, 1997; accepted July 21, 1997
tion as two test reactions to probe catalytic behavior and
to allow comparisons among these Cu-containing systems.
Furfural, C4H3OCHO, is an important compound in the
fragrance industry and the vapor-phase hydrogenation of
furfural is perhaps the best method of producing furfuryl al-
A pretreatment involving a 4-h reduction under H2 at 573 K,
compared to either 473 or 673 K, gave the highest specific activity
for furfuraland crotonaldehydehydrogenation over an unpromoted
copper chromite catalyst. Both H2 and CO chemisorption declined
monotonically with increasing reduction temperature, but turnover
frequencies also showed a maximum after a pretreatment at 573 K.
DRIFTS identified a band, associated with CO adsorbed on Cu+1
sites, which reached a maximum intensity after a 573 K pretreat-
ment, while XRD patterns also provided evidence for a CuCrO2
phase and showed a continuous increase in Cu0 as temperature in-
creased. Selectivity to furfuryl alcohol ranged from 35–80% and
reaction orders on furfural and H2 were near unity. In contrast,
crotonaldehyde hydrogenation gave primarily butyraldehyde with
only around 3% crotyl alcohol and a near zero-order dependence
on crotonaldehyde, although the H2 dependence was still near first
order. IR spectra under reaction conditions revealed adsorbed cro-
tonaldehyde, provided evidence for an unsaturated alkoxide inter-
mediate and detected no butyraldehyde on the surface. A simple
Langmuir–Hinshelwood sequence nicely explained both reactions,
it was consistent with the DRIFTS results obtained during cro-
tonaldehyde hydrogenation, and the resultant rate expression gave
meaningful entropies and enthalpies of adsorption for H2 and cro-
c
==
cohol. Furthermore, because of the presence of both C
O
==
and unsaturated C C bonds, furfural is a suitable com-
pound to test the ability of a catalyst to selectively hydro-
genate a carbonyl bond. Crotonaldehyde also contains a
==
carbonyl group as well as a conjugated C C bond, which
makes selective hydrogenation of the former more difficult.
Despite decades of use, however, the number of fundamen-
tal kinetic studies of hydrogenation reactions over copper
chromite catalysts reported in the open literature has been
limited, with only three pertaining to furfural (2–4) and one
referring to crotonaldehyde (5).
In addition, there have not been many studies charac-
terizing these copper chromite systems, especially the state
of the surface, and, although Cu+1 species have most fre-
quently been associated with activity in various hydrogena-
tion reactions, the participation of Cu0, and even Cu+2
,
has been proposed (5–15). Chudinov et al. investigated
changes in the oxidation state of a copper-chromium-zinc-
alumina catalyst during reduction in H2 between 323 and
673 K and reported that significant amounts of Cu0 be-
gan to form only above 573 K (6). Pillai examined the in-
fluence of a variety of pretreatments on copper chromite
used for the reductive alkylation of aniline with acetone
and concluded that Cu+1 species were the active sites in
the reaction (7). Herman and co-workers also suggested
that Cu+1 was the active species in copper chromite used
for methanol synthesis (8), and similar conclusions were
reached by Monnier et al., who proposed that the Cu+1
species existed in a crystalline CuCrO2 phase (9). Hubaut
and co-workersstudied the hydrogenation ofdieneson cop-
per chromite and again proposed that Cu+1 species consti-
tuted the active sites (5). Makarova and co-workers sug-
gested that, after pretreatment in hydrogen between 453
and 643 K, two phases coexisted in the catalyst, i.e., metal-
lic Cu and a cation-deficient, controlled-disordered spinel
(10). In contrast, Imura et al. have stated that the active
form of copper for methanol decomposition can vary, and
they proposed that Cu+2, Cu+1, and Cu0 species are present
tonaldehyde.
ꢀ 1997 Academic Press
INTRODUCTION
Copper chromite catalysts have been used extensively
for years in various industrial processes such as the par-
tial hydrogenation of vegetable oils and fatty acids as well
as the decomposition or dehydration of alcohols (1). The
utilization of these catalysts in hydrogenation reactions
stems from their capability to selectively hydrogenate car-
==
bonyl bonds while leaving unsaturated C C bonds virtu-
ally untouched. New restrictions now prevent used cop-
per chromite catalysts from being deposited in landfill sites,
thus providing an incentive for new replacement Cu cata-
lysts containing no chromium. Consequently, we are cur-
rently studying Cu dispersed on different types of carbon
and have chosen furfural and crotonaldehyde hydrogena-
1 Current address: Chemistry Department, Northeastern University,
Boston, MA 02115.
2 To whom inquiries should be sent.
406
0021-9517/97 $25.00
c
Copyright ꢀ 1997 by Academic Press
All rights of reproduction in any form reserved.

COPPER CHROMITE CATALYSTS
407
in different catalysts (11). Capece and co-workers deter-
Vapor-phase rate measurements were performed at at-
mined the oxidation states and surface composition of cop- mospheric pressure in a pyrex reactor which was placed
per chromite at various stages of catalytic use and after inside a heating mantle equipped with a temperature con-
reductive pretreatments, and they concluded that Cu+1 is troller (Omega, Model CN2011). The catalyst was reduced
the active species for double-bond isomerization while Cu0 in situ for 4 h at either 473, 573, or 673 K under flowing H2
is required for hydrogenation of conjugated dienes (12). and the reactor was then cooled to the reaction tempera-
Nishimura et al. have reported that olefin hydrogenation ture. A constant flow of vapor-phase furfural,
proceeds on metallic copper (13), and Stroupe reported
––
HC
CH
that copper chromite catalysts used in low temperature hy-
drogenations showed partial or complete reduction to Cu0
(14). Finally, Yurchenko et al. used XPS and XRD to ob-
serve Cu0 formation during reduction under H2 (15). Thus,
despite these characterization studies of copper chromite
catalysts, uncertainty still remains about the relationships
between pretreatment procedures, composition, and active
sites.
This paper describes the characterization of a commer-
cial, unpromoted copper chromite catalyst after a series
of different pretreatments. In particular, the effect of re-
duction temperature was examined by physical adsorption,
H2, CO, and N2O chemisorption, X-ray diffraction, and IR
spectroscopy of adsorbed CO, while the hydrogenation of
furfural and crotonaldehyde was used as a catalytic probe
of the reduced surface. A high resolution TEM examina-
tion of this catalyst after different reduction temperatures
is reported elsewhere (16). These investigations not only
provide a basis for comparison with Cu/carbon catalysts,
but also contribute additional insights into the surface sites
associated with their catalytic behavior.
k
k
–
HC
C CHO,
lO ,
==
or crotonaldehyde, CH3CH CHCHO, was established by
bubbling the H2 carrier gas through a saturator contain-
ing pure liquid furfural (Alfa, 99.99% ) or crotonalde-
hyde (Aldrich, 99.5+% ) which was kept in a constant-
temperature liquid bath (Fisher Scientific, Model 90). This
eliminated the problem of line blockage, presumably due
to polymerization, that was experienced with furfural when
it was introduced via a syringe pump. The partial pressure
of the organic reactant in the feed was controlled by chang-
ing the temperature of the bath. All stainless steel lines
after the bath, as well as the sampling valve of the gas chro-
matograph, were heated to ensure that no condensation of
either the reactant or the productsoccurred. Massflowrates
of hydrogen and helium (used as a diluent) were regulated
by mass flow controllers (Tylan, Model FC-260). Furfural
(FURFAL) hydrogenation was studied between 423 and
473 K, standard reaction conditions were 730 Torr H2 and
10 Torr furfural unless otherwise noted, and at 413 K the H2
partial pressure dependency was determined at a furfural
partial pressure of 10 Torr and the furfural pressure de-
EXPERIMENTAL
The catalyst used was a commercial, unpromoted cop- pendence was determined at a hydrogen partial pressure of
per chromite sample (Cu-1800P, Engelhard Corp.). One of 730 Torr. For the companion reaction, standard conditions
three pretreatments was utilized—reduction in 40 cm3 H2 were 35 Torr crotonaldehyde (CROALD) and 720 Torr H2,
(STP)/min (WHSV = 5 ꢁ 104 cm3/h ꢂ g) at 1 atm at either with Arrhenius plots obtained over a temperature range of
473, 573, or 673 K for 4 h, because a 4-h reduction period 333 to 453 K. A more detailed study of this reaction was
was reported to be optimum by Pillai (7). The state of the conducted, and partial pressure dependencies on croton-
surface was probed by H2 and CO chemisorption at 300 K aldehyde were obtained between 110 to 240 Torr at a H2
usinga stainlesssteeladsorption system givinga vacuum be- pressure of 400 Torr while dependencies on H2 were mea-
low 10ꢃ7 Torr at the sample (17). After the initial isotherm, sured over a range of 200 to 600 Torr at a crotonaldehyde
the sample was evacuated at 300 K for 30 min and a second pressure of 160 Torr, and these reaction orders were ac-
isotherm was measured to determine reversible adsorption. quired at 353, 373, and 393 K. Products from both reactions
The H2 (MG Ind., 99.999% ) and CO (Matheson, 99.99% ) were analyzed by a gas chromatograph (Hewlett–Packard,
were flowed through molecular sieve traps (Supelco) and Model 5730A) equipped with a 10-ft column packed with
Oxytraps (Alltech Asso.) for additional purification. BET Carbowax 20M on 80/100 Supelcoport. All kinetic mea-
surface areas were measured using Ar at 77 K.
surements were made at atmospheric pressure and under
XRD spectra were obtained ex situ using a Rigaku differential conditions as conversions were routinely below
Geigerflex diffractometer equipped with a CuKꢀ radiation 15% , except for one run to determine selectivity to furfural
source and a graphite monochromator. Each sample was alcohol at higher conversions.
given one of the pretreatments described above, and after-
The infrared studies were conducted with a Sirius 100
wards it was passivated by exposure to a flowing mixture of FTIR system (Mattson Instr.) using a DRIFTS cell (HVC-
1% O2 in He for 1 h prior to handling in air during XRD DPR, Harrick Sci. Corp.) that was extensively modified to
measurements.
allow in situ treatments up to 800 K under flowing gases

408
RAO ET AL.
(18, 19). Ultrahigh purity H2 and Ar (99.999% , MG Ind.)
were further purified by passing them through indicating
Oxytraps (Alltech Assoc.) and molecular sieve traps, and
precise gas flow rates were maintained using mass flow
controllers (Tylan Corp., Model FC-260). The recorded in-
terferograms typically consisted of 10,000 signal-averaged
scans obtained using a postamplifier gain of 8, an iris
setting of 50, and a resolution of 4 cmꢃ1. Each sample and
corresponding background interferogram was then
Fourier-transformed to its equivalent frequency compo-
nent spectrum and the ratio of these two spectra gave the
transmittance spectrum from which either the absorbance
or the Kubelka–Munk spectrum was obtained. Further
details of the data handling procedure have been described
elsewhere (20). After loading the copper chromite into the
DRIFTS reactor cell, a standard procedure was used to col-
lect the interferograms, all of which were obtained at 300 K.
The first interferogram was that of the initial “as-is” sample
at 300 K under flowing Ar (20 cm3 (STP)/min). The catalyst
was then subjected to one of the three in situ pretreatments
previously discussed then cooled to 300 K for the second
interferogram. Following this, the catalyst was exposed to
a mixture of 10% CO and 90% Ar at 300 K for 30 min and
then purged with pure Ar for 1 h. Interferograms were
collected before and after the purging step. Spectra were
also obtained during crotonaldehyde hydrogenation after
one of the three pretreatments under reaction conditions of
720Torr H2, 30Torr crotonaldehyde, and 333K. In each case
the reference spectrum was that obtained just prior to the
introduction of crotonaldehyde. Additional spectra were
taken after a 30-min purge with Ar at each temperature.
FIG. 1. Adsorption isotherms at 300 K for copper chromite after
reduction in hydrogen at 573 K: (a) CO, (b) H2; (᭹)—total adsorption,
(᭺)—reversible adsorption.
RESULTS
A typical set of chemisorption isotherms is shown in
Fig. 1, and Table 1 lists the BET surface areas as well as
both the total and irreversible hydrogen and CO uptakes
after each pretreatment temperature. The irreversible ad-
sorption is defined as the difference between the total (ini-
tial) and reversible uptakes. The BET measurements indi-
cate that after reduction in hydrogen, the available catalyst
surface area more than doubles from the initial value of
28 m2/g, and the surface area increased with higher pretreat-
ment temperatures. The H2 and CO uptakes were utilized
to count adsorption sites and provide turnover frequencies
on these catalysts.
The XRD patterns are given in Fig. 2. That of the unre-
duced catalyst was in general agreement with the pattern
TABLE 1
BET Surface Areas and CO and H2 Uptakes for Copper Chromite as a Function of the Pretreatment Temperature
H2 uptake (ꢁmol H2/g)a
CO uptake (ꢁmol/g)a
Reduction
temperature (K)
Surface
Wt. loss
(mg/g)
O uptake
(ꢁmol O/g)
area (m2/gm)
Total
Irreversible
Total
Irreversible
Untreated
473
573
28
58
66
72
—
—
94
50
26
—
25
8
—
95
57
32
—
20
6
—
16.1
20.7
31.3
1046
1408
1240
673
5
4
a At 100 Torr.

COPPER CHROMITE CATALYSTS
409
FIG. 2. XRD patterns for Cu chromite catalysts after: (a) He treatment at 473 K; (b) reduction at 473 K; (c) reduction at 573 K; and (d) reduction
at 673 K. The following phases are denoted: Cu0 (ꢄ), CuCr2O4 (᭞), CuCrO2 (+).
previously reported for copper chromite, with the domi- tively (22); however, no such bands were detected in these
nant peak at a 2 ꢂ value of 35.4ꢅ corresponding to cupric spectra.
chromite (2). After pretreatment in hydrogen, narrow, well-
Calculation of the Weisz–Prater number gave values of
defined peaks appeared at 43.0, 51.0, and 75.0ꢅ which cor- 0.05 and 0.012 for the most active catalysts during hydro-
respond to metallic Cu. A broad, weaker peak can be seen genation of furfural and crotonaldehyde, respectively, thus
at 36.5ꢅ, which is assigned to CuCrO2 based on Pillai’s work indicating the absence of diffusional limitations. Table 2A
(7). As the pretreatment temperature was increased from lists the specific activities for furfural hydrogenation at
473 to 673 K, the intensity of the peak at 36.5ꢅ slowly de- 413 K per square meter as well as on a turnover frequency
creased while the peaks for Cu0 intensified.
basis (based on chemisorbed H atoms) after each of the
IR spectra of CuCr2O4, referenced to the untreated sam- three different pretreatments. Initial runs indicated that
ple, were obtained in Ar after reduction at different temper- the catalyst pretreated at 573 K gave the highest rate, spe-
atures. They showed intensity losses in the regions of OH cificactivity, and turnover frequency;therefore, unlessmen-
stretching and bending, but no loss of bands which could tioned otherwise, the reported kinetic data were obtained
be associated with any structural or functional change oc- after this pretreatment. A thorough kinetic study of fur-
curring during the pretreatments. All the spectra discussed fural hydrogenation was very difficult to conduct because
henceforth are referenced to the spectrum of the corre- in all experiments significant catalyst deactivation was ob-
sponding pretreated catalyst prior to CO admission, unless served, in agreement with previous reports (2, 4), and the
stated otherwise. Upon exposure of the pretreated catalyst extent of this process was more severe at higher furfural
surfaces to a 10% CO/90% Ar mixture, a distinct absorp- partial pressures. The deactivation profile of the catalyst
tion band developed around 2115 cmꢃ1 in all the three cases used in the present study at 413 K is displayed in Fig. 4a
(Fig. 3). The gas-phase CO spectrum, which has a charac- and, as indicated, the reciprocal of the rate was linearly
teristic doublet at 2170 and 2121 cmꢃ1, has been subtracted dependent on time on-stream. Two possible reasons have
from these spectra. The absorption band at 2115 cmꢃ1 cor- been suggested for the deactivation: the formation of coke
responds to CO adsorbed linearly on Cu+1 sites (21, 22). and/or poisoning of the catalyst by either adsorbed furfural
Absorption bands at 2140–2160 cmꢃ1 and 2080–2105 cmꢃ1 or some product that is formed, or a change in the oxidation
can be associated with CO on Cu+2 and Cu0 sites, respec- state of the copper during the course of the reaction. The

410
RAO ET AL.
FIG. 3. IR spectra of CO adsorbed at 300 K on copper chromite after different reduction temperatures: total pressure = 750 Torr, PCO = 75 Torr,
balance Ar: (a) Tred = 473 K; (b) Tred = 573 K; (c) Tred = 673 K.
Arrhenius plot for the most active copper chromite catalyst
for furfural hydrogenation is shown in Fig. 4b, and it gives
an apparent activation energy of 11 ꢆ 2 kcal/mol, where the
TABLE 2
uncertainty is represented by 95% probability limits. For
A. Furfural Hydrogenation over Copper Chromite at 413 K,
this reaction, Borts and co-workers (4) report an activation
720 Torr H2, 10 Torr FURFAL
energy of about 21 kcal/mol; however, the marked deactiva-
tion they observed could easily affect this value and, in ad-
dition, it is possible that their catalyst composition was dif-
ferent from that used here. In accordance with other reports
(2, 4), the only two products observed were furfural alcohol
and 2-methylfuran, and Fig. 5 represents selectivity, defined
as mole furfural/(mole furfurol + mole 2-methylfuran), at
higher furfural conversions. Partial pressure dependencies
on hydrogen and furfural for furfural hydrogenation were
determined at 413 K, using the most active catalyst, i.e., that
after reduction at 573 K. In order to correct for the deacti-
vation occurring during these experiments, after the activity
was measured at each chosen set of conditions, the reaction
conditions were returned to a predefined set of standard
conditions and the activity was remeasured. The activities
at the different reaction conditions were then normalized
to those collected at these standard conditions to correct for
deactivation, and the results are shown in Fig. 6. Dependen-
cies from a power rate law fit gave a value of 0.9 for either
hydrogen or furfural. These results compare to the study of
Borts et al. (4) which reported a first-order dependence on
furfural but a second-order dependence on H2.
Pretreatment
temperature
(K)
Specific
activitya
Ratea
TOFb TOFc
Ea
(sꢃ1
) (kcal/mol) (% )
Sd
(ꢁmol/g ꢂ s) (ꢁmol/m2 ꢂ s) (sꢃ1
)
473
573
673
12.4
16.8
6.4
0.21
0.26
0.09
0.25 0.13
1.05 0.34
0.64 0.24
—
11 ꢆ 2
—
80
70
80
B. Crotonaldehyde Hydrogenation at 333 K, 725 Torr H2,
35 Torr CROALD
Pretreatment
temperature
(K)
Specific
activitye
(ꢁmol/g ꢂ s) (ꢁmol/m2 ꢂ s) (sꢃ1
Ratee
TOFb TOFc
Ea
S f
)
(sꢃ1
)
(kcal/mol) (% )
473
573
673
9.9
15.0
8.2
0.17
0.23
0.11
0.20 0.053 4.6 ꢆ 0.4 2.1
0.94 0.150 3.4 ꢆ 0.3 3.1
0.82 0.158 4.4 ꢆ 0.3 2.8
a Based on furfural disappearance.
b Based on irreversible hydrogen adsorption.
c Based on total hydrogen adsorption.
d Selectivity for furfuryl alcohol.
e Crotonaldehyde disappearance.
f Selectivity for crotyl alcohol.

COPPER CHROMITE CATALYSTS
411
FIG. 4. Furfural hydrogenation over copper chromite after reduction at 573 K, PFURFAL = 10 Torr, PH2 = 730 Torr. (a) Deactivation during reaction
at 413 K and (b) Arrhenius plot.
Activities and turnover frequencies (TOFs) at 333 K of H2 was near unity and that of crotonaldehyde was near
for crotonaldehyde hydrogenation are listed in Table 2B. zero.
Again, the CuCr2O4 sample reduced at 573 K gave the high-
Vapor-phase crotonaldehyde gave peaks at 2940, 2830,
est rate and specific activity. Based on irreversible hydrogen 2730, 1724, 1646, 1455, 1145, 1078, and 970 cmꢃ1, as shown
adsorption, the TOF was also maximized after a reduction in Fig. 9, and these peaks have been subtracted from the
at 573 K, whereas the TOF values were similar after reduc- other spectra in Fig. 9. The IR spectra of copper chromite
tion at either 573 or 673 K if total H2 adsorption is used under reaction conditions with H2 and crotonaldehyde are
to count sites. An Arrhenius plot for the sample reduced shown in Fig. 9. Three strong absorption bands developed
at 573 K is shown in Fig. 7; others are provided elsewhere at 1428, 1529, and 1704 cmꢃ1 along with weak bands near
(23). All plots gave activation energies close to 4 kcal/mol, 968, 1130, 1636, 2740, and 2960 cmꢃ1. The band at 1704
which are slightly lower than those reported over Pt cata- cmꢃ1, red-shifted by 20 cmꢃ1 from the vapor-phase species,
lysts (24). In contrast to furfural, no significant deactivation can be assigned to the carbonyl stretch frequency of unre-
was observed in any of these activity runs. Partial pressure acted crotonaldehyde adsorbed on the copper chromite sur-
dependencies for crotonaldehyde hydrogenation were ob- face. Similarly, the bands at 968, 1130, 1636, 2740, and 2960
tained using CuCr2O4 reduced at 573 K, and data, shown can also be assigned to weakly adsorbed crotonaldehyde
in Fig. 8, were obtained at three temperatures to facilitate (21, 25). Purging with Ar after reaction leaves two strong
kinetic modeling. At all temperatures, the reaction order bands at 1428 and 1529 cmꢃ1 and a weak band at 1660 cmꢃ1
,

412
RAO ET AL.
TABLE 3
Wavenumber (cmꢃ1) and Assignment of Bands Observed during
Crotonaldehyde (CROALD) Hydrogenation on Copper Chromite
Observed
Wavenumber
(cmꢃ1
)
Assignment
Reference
==
1704
C
O stretch in adsorbed CROALD
21
==
1636
1529
1428
Olefinic C C stretch in adsorbed
CROALD
21–25
==
C
C ꢃ interaction between
alcoholate and Cu+1
CH3 deformation modes, or possible
25
25
0
==
C
C ꢃ complex with Cu
–
1130
968
C C stretch in adsorbed CROALD
CH wagging vibrations in adsorbed
CROALD
26
25
2740, 2960
CHx stretches in adsorbed CROALD
21, 25
per chromite surface. It is important to note that no band at
1751 cmꢃ1, indicative of butyraldehyde, was detected. Ex-
posure of this sample to 30 Torr of butyraldehyde in Ar at
333 K produced no detectable bands, indicating very weak
adsorption of butyraldehyde on this surface.
FIG. 5. Selectivity to furfuryl alcohol in furfural hydrogenation over
copper chromite pretreated at 573 K.
as shown in Fig. 10. No IR investigation of this reaction
over any catalyst has been previously reported to provide
an assignment for a species with only these bands; how-
ever, bands at 1475 cmꢃ1 have been associated with ethy-
lenꢃe1ꢃ-complexes with Cu metal and bands around 1530
==
cm have been assigned to C C ꢃ bonds interacting with
Cu+1 cations (21, 26). However, bands around 1430 cmꢃ1
can also be associated with CH3 asymmetrical deformation
modes (25), and the weak bands at 2740 and 2960 cmꢃ1
–
are representative of C H stretches, with the former in-
dicative of that between H and a carbon atom attached to
a carbonyl group (25). If the oxygen atom interacts with
Cu+1 or Cu0 to form an alcoholate (an alkoxide), the 1704
cmꢃ1 band would not be present and a weaker band in the
1000–1100 cmꢃ1 region might be expected to occur (21).
The spectra in Fig. 10 do not disallow this possibility; thus
both the 1529 and 1428 bands can be assigned to an in-
termediate consisting of crotonaldehyde with one H atom
added to the carbonyl bond and the O atom interacting
with Cu+1 while the C C bond forms a ꢃ-complex with an-
==
other Cu+1 cation, although the possibility of the 1428 band
representing a ꢃ-complex with Cu0 cannot be discounted.
The absence of the C H 2830 cmꢃ1 band in these adsorbed
–
==
species probably reflects this interaction between the C
C
ꢃ bond and the Cu species. Table 3 lists the assignments of
FIG. 6. Pressure dependencies for furfural hydrogenation on copper
the IR bands associated with the species observed on a cop- chromite: (a) H2; (b) furfural. The fit is that obtained with Eq. [2].

COPPER CHROMITE CATALYSTS
413
FIG. 7. Arrhenius plot of crotonaldehyde hydrogenation over copper chromite after reduction at 573 K: PCROALD = 35 Torr; PH2 = 710 Torr.
DISCUSSION
the principal Cu0 peak indicate Cu crystallites on the order
of 12–15 nm. The DRIFTS results provide evidence only
Previous studies of copper chromite catalysts have shown for Cu+1 species at the surface, which reach a maximum
that catalytic activity is dependent upon the reduction tem- concentration after the 573 K pretreatment. The absence
perature. In their study of acetone hydrogenation to iso- of Cu+2 is not surprising, but the inability to detect CO ad-
propanol over copper chromite, Makarova and co-workers sorbed on Cu0 iscertainlysomewhat unexpected. The broad
reported that the maximum activity was obtained after re- 36.5ꢅ peak implies a low level of crystallinity, whereas the
duction at 573–593 K (10), and Pillai also found that cop- linewidth of the Cu0 peaks indicates larger metallic copper
per chromite reduced at 573 K showed the highest activity crystallites; thus one possible interpretation is the forma-
for the reductive alkylation of aniline with acetone (7). As tion of Cu0 particles that are almost completely covered by
shown in Table 2, the catalyst pretreated in H2 at 573 K an oxide phase, such as CuCrO2. The verification of Cu+1
showed the highest activity; therefore, our results agree sites and the coincidence of maximum specific activity with
with these previous studies and indicate that the same ac- maximum Cu+1 concentration is strong evidence that these
tive form(s) of Cu may be involved in all these reactions. sites are involved in the catalytic cycle, in agreement with
This would imply that the copper sites involved in the re- previousconclusions(7–10). However, H2 activation should
action reach a maximum concentration after this particular be more likely to occur on Cu0 surfaces, thus the presence
pretreatment. Comparison of the XRD patterns obtained of both Cu+1 and Cu0 species is probably required for opti-
for the catalyst after the three different pretreatment tem- mum performance.
peratures shows that the only identifiable peak that may
The effect of pretreatment temperature under flowing
be maximized after reduction at 573 K is that at 36.5ꢅ, H2 was monitored by a variety of adsorption techniques.
which has been associated with CuCrO2 and implies that The BET surface area, as determined by Ar physisorption,
the cuprous ion plays some role in the catalytic cycle. The doubled after pretreatment at 473 K, and continued to in-
broad peak at 36.5ꢅ in the XRD pattern is assigned to a Cu+1 crease with increasing reduction of temperature to reach
species, designated as CuCrO2, based on previous studies a value of 72 m2/g after pretreatment at 673 K, as shown
(7, 9). Its presence along with metallic copper is evident in in Table 1. This is presumably a consequence of the open-
each reduced sample, and the relative amount of Cu0 clearly ing of small pores due to the removal of strongly adsorbed
increases as reduction temperature increases, although the water plus dehydroxylation. The IR spectra and the increas-
fraction at the surface cannot be ascertained. Peak widths of ing weight loss with higher pretreatment temperature are

414
RAO ET AL.
FIG. 8. Pressure dependencies at different reaction temperatures for crotonaldehyde hydrogenation over copper chromite after reduction at 573 K:
(4), 353 K; (ᮀ), 373 K; (᭺), 393 K. (A) Dependence on H2, PCROALD = 160 Torr; (B) dependence on CROALD, PH2 = 400 Torr.
consistent with this proposal. The total uptake of either gen, presumably in the form of H atoms, is about dou-
H2 or CO at 100 Torr is very similar for each sample and ble that which strongly adsorb CO. Since H2 must be acti-
both values decline as pretreatment temperature increases. vated in the reactions examined, hydrogen chemisorption,
The irreversible adsorption follows the same pattern and especially the irreversibly adsorbed form, is probably the
the number of sites capable of strongly adsorbing hydro- best choice to calculate turnover frequencies, i.e., molecule

COPPER CHROMITE CATALYSTS
415
FIG. 9. DRIFT spectra of copper chromite under reaction conditions at T = 333 K, PCROALD = 30 Torr, PH2 = 720 Torr. Reduction temperature:
(a) 473 K; (b) 573 K; (c) 673 K; and (d) DRIFT spectrum of vapor-phase crotonaldehyde at 333 K. The reference spectrum in each case is that of the
sample in H2 prior to introduction of CROALD. Gas-phase CROALD bands have been removed.
reacted/s/site, where a H atom counts an adsorption site. Ir- this technique for determining metallic surface Cu atoms in
reversible hydrogen (and to a lesser extent CO) adsorption this system.
does occur on Cu0 surfaces at 300 K, but at less than mono-
The kinetic behavior of Cu chromite can now be exam-
layer coverages (27). Dissociative N2O adsorption at 363 K ined. Depending on the catalyst, vapor-phase hydrogena-
has been used to measure metallic Cu surface area based tion of furfural can give rise to a variety of products such
on the stoichiometry; N2O(g) +2 Cu0s → Cu2O +N2(g) (28), as furfuryl alcohol, 2-methylfuran, furan, tetrahydrofuran,
where Cus is a surface atom. The results in Table 1 show tetrahydrofurfuryl alcohol, and even ring-decomposition
that very large amounts of oxygen are adsorbed during this products such as pentanols and pentanediols. Furfural has
reaction, regardless of the pretreatment temperature, if all been reduced quantitatively to furfuryl alcohol with Pt ox-
the weight gain is attributed to O atoms. No correlation ide, provided that the reduction was stopped after the ad-
with the amount of Cu0 detected by XRD is observed. The dition of one molecular equivalent of hydrogen (31). On
oxygen coverages of 1.0–1.3 ꢁ 1019 O atoms/m2 represent further hydrogenation, the ring is first saturated to yield
very high fractional coverages and were not especially sen- tetrahydrofurfuryl alcohol, together with smaller amounts
sitive to the pretreatment. The catalytic decomposition of of ring-opening products, i.e., pentanediols and pentanols.;
N2O over CuO and Cr2O3 at higher temperature has been Ni-based catalysts lead to hydrogenation of the furan ring
associated with coordinatively unsaturated cations at the and the accompanying products (3, 32). Cu-based catalysts
surface (29, 30); thus N2O adsorption on these sites, either have been most widely used for the selective hydrogena-
==
with or without dissociation, is likely and appears to obviate tion of the C O bond in the side group while leaving

416
RAO ET AL.
FIG. 10. DRIFT spectra of copper chromite after crotonaldehyde hydrogenation and purging in Ar at 333 K; (a) TRED = 473 K, (b) TRED = 573 K,
and (c) TRED = 673 K.
==
the C C bonds in the furan nucleus intact. Bremner and and H2 are consistent with the first-order dependence on
Keeys made an exhaustive investigation of furfural hydro- furfural reported by Borts et al., but clearly differ from the
genation to furfuryl alcohol and 2-methylfuran over sev- second-order H2 dependence reported by this group (4).
eral copper catalysts, including copper-alumina and copper Very severe deactivation was also observed by Borts et al.
chromite, and they found that furfuryl alcohol was the pri- and, if not correctly accounted for, this could explain the
mary product below 473 K while 2-methylfuran was the pri- difference in H2 dependency, especially if nondifferential
mary product around 523 K (2). Seo and Chon also reported conversions were achieved.
that furfuryl alcohol was the primary product formed
The partial pressure data suggest that a simple bimolecu-
in furfural hydrogenation over a variety of copper-based lar surface reaction as a rate determining step (rds) is capa-
catalysts, but they did not mention any significant change ble of explaining the hydrogenation of furfural over this
in selectivity with temperature (3). Borts and co-workers in- copper chromite catalyst, and a Langmuir–Hinshelwood
vestigated the kineticsoffurfuralhydrogenation over a cop- model invoking either one type of site with competitive
per chromium catalyst and reported that furfuryl alcohol adsorption between furfural and hydrogen or two types of
was the main product with less than 3 wt% 2-methylfuran sites, i.e., one for each reactant, fits the data well. For exam-
being formed (4).
ple, with the former sequence, assuming that H2 activation
Despite the preference of copper chromite for hydro- requires dissociation, the steps are
genation of furfural to furfuryl alcohol and 2-methylfuran,
there is surprisingly little fundamental information on the
KH2
ꢀ
–
H2(g) + 2 S
2 H S
[1a]
[1b]
[1c]
kinetics and mechanism of this reaction over this catalyst,
and a definitive understanding of the active state of cop-
per involved in furfural hydrogenation does not exist. In
this study, an unpromoted copper chromite catalyst was re-
duced at either 473, 573, or 673 K and then examined for
furfural hydrogenation. As indicated in Fig. 5, at conver-
sions up to 60% , only furfural and 2-methylfuran are pro-
duced, and the selectivity to furfural alcohol ranges from 35
to 70% , consistent with the earlier studies discussed above.
The reaction orders near unity (i.e., 0.9) for both furfural
KF
ꢀ
–
C4H3OCHO(g) + S
C4H3OCHO S
K
ꢀ
k
–
–
–
C4H3OCHO S + H S
C4H3OCH2O S
–
–
–
C4H3OCH2O S + H S →C4H3OCH2OH S (rds) [1d]
Iꢀ/KA
–
C4H3OCH2OH S
C4H3OCH2OH(g) + S, [1e]
where S designates an active site.

COPPER CHROMITE CATALYSTS
417
Step (4) is the rate determining step (rds) and the
It has been suggested that this low selectivity could be
other four are in quasi-equilibrium. Should the adsorbed due to the strong adsorption of the allylic alcohol on cop-
furfuryl alcohol in step (5) react further with hydrogen per chromite and its subsequent high reactivity (39). Other
rather than desorb, 2-methylfuran is formed. If only ad- than desorb, the allylic alcohol can either isomerize to form
sorbed furfural and hydrogen are considered to be sig- the saturated aldehyde or further hydrogenate to form the
nificant surface species, the site balance is then L = [S] + saturated alcohol. In the present study, the in situ FTIR
–
–
[C4H3OCHO S] + [H S] and the derived rate expression is spectra suggested the formation of a possible unsaturated
alkoxide-type intermediate on the catalyst surface, clearly
verified the presence of adsorbed crotonaldehyde via the
ꢀ
ꢁ
2
r = kK KH KF PH PF/ 1 + K1/2 PH1/2 + KF PF
,
[2]
2
2
H2
2
==
==
frequency of the C O and C C stretches, and indicated
the absence of butyraldehyde on the surface. It has been
previously proposed that both butyraldehyde and crotyl al-
cohol can originate from this type of common intermediate
(24, 40). This adsorbed complex can react with adsorbed H
to give either butyraldehyde or crotyl alcohol. Since no bu-
tanolwasdetected in the product analysis, secondaryhydro-
genation activity can be assumed to be negligible at these
reaction temperatures giving low conversions. However, a
very low fraction of the final product corresponds to crotyl
alcohol, thus rapid isomerization of crotyl alcohol must oc-
cur in the adsorbed state to yield butyraldehyde. This is con-
sistent with the results of Hubaut et al. which demonstrated
a low hydrogenation activity of crotyl alcohol relative to its
isomerization over copper chromite (39).
which fits the data well, as shown in Fig. 6. The value of
the apparent rate constant (kK KH KF) is 1.8 ꢁ 10ꢃ5 ꢁmole/
2
(s ꢂ m2 ꢂ atm2), and the values of the equilibrium adsorption
constants are KH = 1.1 ꢁ 10ꢃ4 atmꢃ1 and KF = 1.4 atmꢃ1
2
which provide near first-order behavior for each reactant.
An equally good fit of the data is obtained with the 2-site
model because of the near first-order dependence on each
reactant (33); thus a choice between the two models cannot
be made.
In regard to crotonaldehyde hydrogenation, only the pa-
per by Hubaut et al. has addressed this reaction over copper
chromite catalysts (5), although Jenck and Germain did ex-
amine competitive hydrogenation of aldehydes and olefins
over copper chromite (34). Noller and Lin studied croton-
The in situ FTIR studies of crotonaldehyde hydrogena-
tion suggest that an unsaturated alkoxide-type species is
a detectable organic intermediate whose hydrogenation
rate could control the overall rate of reaction. Therefore, a
Langmuir–Hinshelwood model with the hydrogenation of
such an intermediate species as the rate determining step
(rds) was utilized to describe crotonaldehyde hydrogena-
tion over copper chromite. This approach, which assumes
all steps other than the rds to be quasi-equilibrated, gen-
erates the same general sequence of elementary steps as
shown previously, i.e.,
–
aldehyde hydrogenation over Cu Ni catalysts (35), and this
reaction over various Pt catalysts has been investigated re-
cently (36–38). Hubaut et al. reported that at 313 K and at-
mospheric pressure, selectivity to crotyl alcohol decreased
from 15 to 6% as the overall conversion of crotonaldehyde
increased from 0.1to 49% , and butanolwasobtained at con-
version levels above 10% . Apparent reaction orders of 0.75
for hydrogen and 0.17 for crotonaldehyde were reported at
313 K, but no activation energies, rates, or specific activities
were reported and the effect of pretreatment temperature
was not investigated.
In the present study, the partial pressure dependencies of
0.88 for hydrogen and 0.08 for crotonaldehyde are in good
agreement with the values of Hubaut et al. The sample re-
duced at 573 K exhibited the highest rate and specific activ-
ity, and it yielded an activation energy of 3.4 ꢆ 0.3 kcal/mol
between 333 and 453 K. Selectivities to crotyl alcohol at
333 K and a 10% conversion level were near 3% , which is
about half that reported by Hubaut et al. (5). This low se-
lectivity towards the allyl alcohol during hydrogenation of
an ꢀ, ꢄ-unsaturated aldehyde is somewhat surprising, given
the high selectivity of CuCr2O4 for diene hydrogenation (5).
However, it is consistent with the conclusion of Jenck and
KH2
ꢀ
–
H2(g) + 2 S
2 H S
[3a]
[3b]
[3c]
KCROALD
ꢀ
–
C3H6CO(g) + S
C3H6CO S
Kꢄ
ꢀ
–
–
–
C3H6CO S + H S
C3H6CHO S + S
k
–
–
–
C3H6CHO S + H S →C3H6CHOH S + S (rds) [3d]
KISO
ꢀ
–
C3H6CHOH S
C3H8CO(g) + S,
[3e]
Germain that the functional group selectivity in a conju- where S again represents an active site, k is a rate constant,
gated system is lower than that expected from the relative KH and KCROALD are adsorption equilibrium constants
2
reactivities of isolated double bonds and carbonyl groups while Kꢄ is the equilibrium constant for reaction [3c]. The
(34). The results obtained in this study are in general agree- assumptions that an alkoxide-like species is an adsorbed re-
ment with previous studies in that the highest selectivity of active intermediate, whose formation is given by step [3c],
3% for crotyl alcohol is much lower than that achieved with and that adsorbed butyraldehyde concentrations are very
some other catalysts such as Pt/TiO2 (24).
low, both of which are supported by our DRIFTS data, give

418
RAO ET AL.
TABLE 4
TABLE 5
Parameters Obtained from Rate Expression for
Crotonaldehyde Hydrogenation
Thermodynamic and Kinetic Parameters for
Crotonaldehyde Hydrogenation
Reaction temp.
(K)
k
KH
KCROALD
ꢃ1Hꢅad (kcal/mol)
1Sꢅad (cal/mol/K)
H2 CROALD
2
Erds
(kcal/mol)
(ꢁmol/s/m2)
(atmꢃ1
)
(atmꢃ1
)
Kꢄ
H2
CROALD
353
373
393
2.0
3.9
5.7
0.46
0.36
0.21
5.4
3.6
3.3
0.29
0.45
0.78
5.6
3.6
ꢃ17.2
ꢃ6.7
8.5
(Fig. 11), and these values are listed in Table 5. The values
of KH , KCROALD, and Kꢄ are such that a near first-order
a site balance of
2
dependency on H2 and a near zero-order dependency on
CROALD are obtained over the pressure range studied.
The enthalpies of ꢃ5.6 and ꢃ3.6 kcal/mol indicate weak
chemisorption on the copper chromite surface, and the en-
tropy losses of ꢃ17.2 and ꢃ6.7 e.u. correspond to values
close to that expected for the loss of one degree of transla-
tional freedom (41).
–
–
–
L = [S] + [H S] + [C3H6CO S] + [C3H6CHO S] [4]
and result in the rate expression:
kKH KCROALD Kꢄ PH PCROALD
2
2
r =
.
ꢀ
ꢁ
2
1+KH1/2 PH1/2 +KCROALD PCROALD +K1/2 KCROALD Kꢄ P1/2 PCROALD
H
H
2
2
2
2
[5]
The capability of this equation to fit all the experimental
results is excellent, as shown in Fig. 8, and the correla-
tion constants obtained for the four fitting parameters were
near unity. The parameters obtained from these linear re-
gressions are listed in Table 4, and their physical consis-
tency was checked by obtaining enthalpies and entropies
SUMMARY
Both the specific activity and the TOF for furfural and
crotonaldehyde hydrogenation reached their highest val-
ues over unpromoted copper chromite after reduction un-
der H2
at 573 K, compared to 473 and 673 K, whereas both
of adsorption for H2 and crotonaldehyde from the plots of H2 and CO chemisorption declined monotonically with in-
these adsorption constants versus reciprocal temperature creasing temperature. In situ DRIFTS spectra showed that
FIG. 11. Kinetic parameters from rate expression (Eq. [5]) for crotonaldehyde hydrogenation versus inverse temperature: ln KH (♦), ln KCROALD
2
(ᮀ), ln k (4), and ln Kꢄ (᭺). Units of k are ꢁmole/s/g cat.

COPPER CHROMITE CATALYSTS
419
a peak associated with CO adsorption on Cu+1 sites reached
a maximum after the 573 K pretreatment, and XRD pat-
terns also provided evidence for a Cu+1 phase (CuCrO2),
although they also showed a continuous increase in the Cu0
phase as reduction temperature increased. Consequently,
8. Mehta, S., Simmons, G. W., Klier, K., and Herman, R. G., J. Catal. 57,
339 (1979).
9. Monnier, J. R., Hanrahan, M. J., and Apai, G., J. Catal. 92, 119 (1985).
10. Makarova, O. V., Yur’eva, T. M., Kustova, G. N., Ziborov, A. V.,
Plyasova, L. M., Miyukova, T. P., Davydora, L. P., and Zaikovskii,
V. I., Kinet. Catal. 34, 683 (1993).
Cu+1 sites correlate with activity and appear to be involved 11. Imura, A., Inoue, Y., and Yasumori, I., Bull. Chem. Soc. Jpn. 56, 2203
(1983).
in the catalytic process, as proposed by previous workers,
12. Capece, F. M., Castro, V. D., Furlani, C., Mattogno, G., Fragale, C.,
Gargano, M., and Rossi, M., J. Elec. Spec. Rel. Phenom. 27, 119 (1982).
13. Nishimura, E., Inoue, Y., and Yasumori, I., Bull. Chem. Soc. Jpn. 48,
but the presence of Cu0 sites is probably also required, thus
providing one explanation for the maximum in activity that
occurs as a function of reduction temperature. Selectivity to
803 (1975).
furfural alcohol ranged from 35–70% , significant deactiva- 14. Stroupe, J. D., J. Am. Chem. Soc. 71, 569 (1949).
15. Yurchenko, E. N., Boronin, A. I., Ziborov, A. V., Korolkova, O. G.,
tion occurred, and reaction orders on furfural and H2 were
near unity after correction for deactivation. A straightfor-
ward Langmuir–Hinshelwood model described these data
well. In contrast to furfural, little deactivation was observed
during crotonaldehyde hydrogenation, a lower apparent
activation energy was observed, and the dependency on
the organic reactant was near zero order, although that on
H2 was still near first order. IR spectra taken under re-
action conditions indicated the presence of adsorbed cro-
Zubritskaya, N. G., and Playasova, L. M., Kinet. Catal. 33, 401 (1990).
16. Ma, J., Rodriguez, N. M., Vannice, M. A., and Baker, R. T. K., to be
published.
17. Na, B.-K., Walters, A. B., and Vannice, M. A., Appl. Spect. 140, 585
(1993).
18. Venter, J. J., and Vannice, M. A., Appl. Spec. 42, 1096 (1988).
19. Fanning, P. E., and Vannice, M. A., Carbon 31, 721 (1993).
20. Venter, J. J., and Vannice, M. A., Carbon 26, 889 (1988).
21. Davydov, A. A., “Infrared Spectroscopy of Adsorbed Species on the
Surface of Transition Metal Oxides,” Wiley, London, 1990.
tonaldehyde, provided evidence for an unsaturate alkox- 22. Choi, K. I., and Vannice, M. A., J. Catal. 131, 22 (1991).
23. Dandekar, A., Ph.D. thesis, Pennsylvania State University, in progress.
24. Sen, B., and Vannice, M. A., J. Catal. 115, 65 (1989).
25. Socrates, G., “Infrared Characteristic Group Frequencies,” Wiley,
ide intermediate, and detected no butyraldehyde. A simple
Langmuir–Hinshelwood reaction sequence incorporating
this intermediate gave an excellent fit of the data, along
Chichester, 1980.
with physically meaningful values for enthalpies and en-
tropies of adsorption that imply weak adsorption for both
crotonaldehyde and hydrogen.
26. Nakanishi, K., “Infrared Absorption Spectroscopy,” Holden-Day, San
Francisco, 1977.
27. Leon y Leon, C. A., and Vannice, M. A., Appl. Catal. 69, 269 (1991).
28. Chinchen, G. C., Hay, C. M., Vandervell, M. D., and Waugh, K. C.,
J. Catal. 103, 79 (1987).
ACKNOWLEDGMENT
29. Winter, E. R. S., J. Catal. 19, 32 (1970).
30. Kapteijn, F., Rodriguez-Mirasol, J., and Moulijn, J. A., Appl. Catal. B:
Environ. 9, 25 (1996).
31. Kaufmann, A., and Adams, J. C., J. Am. Chem. Soc. 45, 3029 (1923).
32. Elderfield, R. C. (Ed.), “Heterocyclic Compounds,” Vol. 1, p. 161,
Wiley, New York, 1950.
Support for this project was provided by the National Science Founda-
tion under Grant CTS-9415335.
REFERENCES
33. Rao, R., Ph.D. thesis, Pennsylvania State University, in progress.
1. Thomas, C. L., “Catalytic Processes and Proven Catalysts,” Academic 34. Jenck, J., and Germain, J. E., J. Catal. 65, 141 (1980).
Press, New York, 1970.
35. Noller, H., and Lin, W. M., J. Catal. 85, 25 (1984).
2. Bremner, J. G. M., and Keeys, R. K. F., J. Chem. Soc., 1068 (1947).
3. Seo, G., and Chon, H., J. Catal. 67, 424 (1981).
36. Raab, C. G., and Lercher, J. A., J. Mol. Catal. 75, 71 (1992).
37. Raab, C. G., and Lercher, J. A., Catal. Lett. 18, 99 (1993).
4. Borts, M. S., Gil’chenok, N. D., Ignat’ev, V. M., and Gurevich, G. S., 38. Coloma, F., Sepulveda-Escribano, A., and Rodriguez-Reinoso, F.,
Zh. Prikl. Khim. 59, 126 (1986).
Appl. Catal. A 123, L1 (1995).
5. Hubaut, T., Daage, M., and Bonnelle, J. P., Appl. Catal. 22, 231
(1986).
6. Chudinov, M. G., Kuznetsov, B. N., Alekseev, A. M., and Pak, G. P.,
Kinet. Catal. 29, 590 (1988).
39. Hubaut, R., Daage, M., and Bonnelle, J. P., Appl. Catal. 22, 243 (1986).
40. Simonik, J., and Beranek, P., Coll. Czech. Chem. Commun. 37, 353
(1972).
41. Vannice, M. A., Hyun, S. H., Kalpakci, B., and Liauh, W. C., J. Catal.
56, 358 (1979).
7. Pillai, R. B. C., Catal. Lett. 26, 365 (1994).