Extrusion of Cu/ZnO catalysts for the single-stage gas-phase processing of dimethyl maleate to tetrahydrofuran

Article abstract of DOI:10.1016/S0021-9517(03)00157-X

The extruded catalysts consisting of copper, zinc oxide, and γ-alumina were used in the fixed-bed gas-phase conversion of dimethyl maleate at 453-493 K and at 0.2-2.4/hr. The extruded catalysts were much more active than would be expected from the low values of their Cu surfaces areas, this discrepancy being most pronounced at ≥ 473 K. The C₄-products formed over the catalysts were GBL, 1,4-butanediol(BDO), and tetrahydrofuran (THF). In all cases, the THF selectivity increased with increasing conversion and temperature. The molar ratio BDO/GBL produced over pure Cu/ZnO approximately equaled the thermodynamic equilibrium value at 493 K and 2.5 MPa. Through choice of composition and pore texture of the extrudates, it was possible to achieve THF yields of 100% (basis the C₄-products) at 2.5 MPa in a single stage.

Full text of DOI:10.1016/S0021-9517(03)00157-X

Journal of Catalysis 218 (2003) 419–426
www.elsevier.com/locate/jcat
Extrusion of Cu/ZnO catalysts for the single-stage gas-phase processing
of dimethyl maleate to tetrahydrofuran
Steffen Peter Müller, Martin Kucher, Christoph Ohlinger, and Bettina Kraushaar-Czarnetzki ∗
Institute of Chemical Process Engineering CVT, University of Karlsruhe, Kaiserstrasse 12, D-76128 Karlsruhe, Germany
Received 13 January 2003; revised 6 March 2003; accepted 10 March 2003
Abstract
Bifunctional catalysts consisting of copper, zinc oxide, and γ -alumina have been produced by means of extrusion. The effects of for-
mulation and forming conditions on morphology, pore texture, and copper surface area are discussed. These extruded catalysts have been
employed in the fixed-bed gas-phase conversion of dimethyl maleate at temperatures between 453 and 493 K and space velocities rang-
−1
ing from 0.2 to 2.4 h . Copper catalyzes the hydrogenation and hydrogenolysis reactions yielding γ -butyrolactone, and 1,4-butanediol.
γ -Alumina provides weakly acidic sites which promote the subsequent dehydration of 1,4-butanediol to tetrahydrofuran. Coupled products
are methanol and water. Through choice of composition and pore texture of the extrudates it is possible to achieve tetrahydrofuran yields of
100% (basis the C -products) at a low pressure (2.5 MPa) in a single stage.
4
2003 Elsevier Inc. All rights reserved.
Keywords: Catalyst forming; Extrusion; Pore-size distribution; Maleic acid; Hydrogenation; Hydrogenolysis; Copper catalyst; Tetrahydrofuran
1. Introduction
and from the corresponding dialkyl esters, e.g., dimethyl
maleate. The subsequent hydrogenation or hydrogenolysis
steps, respectively, are catalyzed by heterogeneous noble
metal, nickel, or copper catalysts. Most existing processes
are carried out in three-phase slurry reactors at pressures
between 5 and 30 MPa and temperatures between 420 and
570 K, and the preferred feedstock is the anhydride [5]. In
contrast, a gas-phase process is feasible only with the dialkyl
esters of maleic acid as a feed because the boiling points
of the anhydrides, in particular of the intermediate succinic
anhydride, are too high in the pressure range of interest. De-
spite the requirement of an additional esterification step up-
flow, a gas-phase process for the production of BDO would
be very attractive because established fixed-bed reactor tech-
nology can be utilized, and lower pressures (between 2 and
7 MPa) can be applied. So far, only one gas-phase process
has been established by Davy McKee Ltd. [6] which consists
of two stages. A potential problem in gas-phase operation
comprises fouling and plugging of the fixed-bed by poly-
meric material [7,8]. The polymeric deposits (polyesters)
can be formed upon trans-esterification of dialkyl esters of
maleic or succinic acid with the product BDO.
Tetrahydrofuran (THF) is an important solvent for many
polymers, and it is used as a monomer in the manufac-
ture of polytetramethylene ether glycol (PTMEG), a precur-
sor for polyurethanes, elastic fibers, molded elastomers, and
copolyesters [1]. In view of the strongly growing markets for
PTMEG, a doubling of the production of THF in the period
between 1995 and 2004 has been forecasted [2].
THF is obtained through the dehydration of 1,4-butane-
diol (BDO) in the presence of mineralic acid or acidic ion
exchangers. The major part of the BDO, in turn, is still
produced via the Reppe process [3] using acetylene and
formaldehyde as the feedstocks. New manufacturing routes,
however, will be based on economically more attractive raw
materials. Among these, maleic acid anhydride, which is
available from the direct catalytic oxidation of n-butane or
other linear C4-hydrocarbons with air, will play a dominant
role [4].
As shown in the reaction scheme in Fig. 1, the forma-
tion of BDO is possible both from maleic acid anhydride
We recently investigated the conditions at which fouling
and plugging by the viscous by-products occur. Our results
indicate that the polyester formation is initiated by capillary
*
Corresponding author.
E-mail address: bettina.kraushaar@ciw.uni-karlsruhe.de
(B. Kraushaar-Czarnetzki).
0021-9517/03/$ – see front matter
2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0021-9517(03)00157-X

420
S.P. Müller et al. / Journal of Catalysis 218 (2003) 419–426
Fig. 1. Scheme of the hydrogenation and hydrogenolysis reactions of maleic acid derivatives to GBL and BDO, and of the subsequent dehydration to THF.
condensation of reactants, in particular of BDO, in the pores
of the catalyst. Condensation can be prevented by applying
high molar feed ratios of hydrogen to dimethyl maleate, i.e.,
by dilution of the organic reactants. We calculated the min-
imum H2/maleate feed ratios required for safe and stable
processing in a wide range of reaction pressures and tem-
peratures and validated these calculated limits by processing
experiments [9]. Hence, the single-stage gas-phase hydro-
genation of dimethyl maleate in a fixed-bed reactor is a fea-
sible option. Depending on the pore-size distribution of the
catalyst, however, the required dilution with hydrogen may
be extremely high.
Here, we report on novel extruded catalysts with bifunc-
tional properties which provide a solution to this problem.
In addition to the hydrogenating function, the catalysts com-
prise weakly acid sites which catalyze selectively the de-
hydration of BDO to THF. Because BDO is consumed,
together with dimethyl succinate being the most critical
compound with respect to capillary condensation, and be-
cause THF exhibits considerably lower Antoine constants
than the other C4-reactants [9], it will be possible to ap-
ply much lower H2/maleate feed ratios without running into
the regime of fouling by polymer formation. The other in-
teresting aspect of this catalyst represents the possibility of
producing THF from dimethyl maleate in a single rather than
two stages, thereby using the most convenient, i.e., fixed-
bed, reactor technology.
The catalysts used in our study contain copper and zinc
oxide (Cu/ZnO). The active sites for hydrogenation and
hydrogenolysis reactions are provided by metallic copper.
The role of ZnO as an adsorbent for succinic anhydride has
been elucidated recently by Herrmann and Emig who in-
vestigated the hydrogenation of maleic acid anhydride in
the liquid (slurry) phase [10,11]. The gas-phase conversion
of dimethyl maleate has been studied by Turek and co-
workers [8,12]. These authors found that the hydrogenolysis
of dimethyl succinate rather than the hydrogenation of the
double bond in dimethyl maleate is the rate-limiting step.
The corresponding rate coefficient can be related directly
to the copper surface area [12]. At conversions of dimethyl
succinate above ca. 80%, the temperature- and pressure-
dependent thermodynamic equilibrium in the hydrogenation
of γ -butyrolactone (GBL) to BDO is attained over Cu/ZnO
catalysts [8,9]. The subsequent dehydration of BDO to THF
plays a minor role because Cu/ZnO does not exhibit suffi-
cient acidity.
In the catalysts reported here, the acid sites are provided
by γ -Al2O3 which, in addition, acts as mechanically strong
and porous binder for the embedded Cu/ZnO particles. In-
dustrial Cu/ZnO catalysts [13–15] as well as those described
in the literature are formed by pelleting. In contrast, our cata-
lysts are formed by extrusion. Because the extent of com-
pression and compaction is much lower in extrusion, the
resulting catalysts exhibit larger pores and a higher poros-
ity, and they allow for a higher effectiveness factors in the
conversion of dimethyl maleate when used in the sizes con-
venient for a technical fixed-bed reactor. To our knowledge,
this is the first report on the forming and application of
Cu/ZnO extrudates.
2. Experimental
2.1. Catalyst preparation
The carbonate precursor of the catalyst was prepared
by dropwise mixing of 1 M aqueous Na2CO3 (Merck,
purity > 99.9% m/m) solution with an aqueous solution
containing 0.5 M, each, copper and zinc nitrate (Merck,
Cu(NO3)2 · 3H2O, > 99.5% m/m and Zn(NO3)2 · 6H2O,
> 99.0% m/m) under vigorous stirring at 359 K and a con-
stant pH of 7. After cooling down, the precipitate was fil-
trated, washed several times with deionized water, and dried
at 353 K for 16 h.
To produce a paste for extrusion, the dried copper/zinc
carbonate precursor was mixed with boehmite (Sasol, im-
purities: Fe2O3 < 200 ppm, Na2O < 50 ppm, SiO2
<
250 ppm) and 22% m/m water in a rheo-kneader (Poly-
drive, Haake) at a constant temperature of 293 K and a rotor
speed of 40 min−1. Upon heating above 723 K, boehmite
(AlO(OH)) undergoes a phase transition to γ -Al2O3. The
proportions of metal carbonates and boehmite were chosen
such that the percentage (c) of copper/zinc carbonate pre-

S.P. Müller et al. / Journal of Catalysis 218 (2003) 419–426
421
cursor in the water-free extrudate according to
dimethyl maleate (purity: > 95% m/m) were fed into the re-
actor at a molar ratio of 250, a total pressure of 2.5 MPa,
and hydrogen flow rates varying from 150 to 1000 cm3/min
(STP). The resulting residence times (tmod) ranged between
0.8 and 13.5 s g/cm3, which is equivalent to space veloci-
ties (WHSV) between 0.2 and 2.4 h−1. The reaction tem-
peratures ranged from 453 to 513 K. On-line analysis was
carried out on a Hewlett Packard 5890 II gas chromatograph
equipped with a CP-Wax 52 CB column after pressure re-
lease and addition of nitrogen (purity: 99.996% m/m) as an
internal standard. The mean deviation in the product selec-
tivities reported ranges from 8% for low conversion levels
to 4% for full conversion.
(mCuCO + mZnCO
)
3
3
c = 100 ·
%
(1)
(mCuCO + mZnCO + mγ -Al O
)
3
3
3
2
amounted to 25, 50, and 75% m/m, respectively. Note that
symbols are explained at the end of this article. The paste
was transferred to a piston extruder and formed to cylin-
drical greenbodies (designation for ceramic bodies prior to
heat treatment) with a diameter of 2 mm and a length of
about 400 mm. The greenbodies were dried at room temper-
ature and cut into pieces of about 5 mm length. The particles
were transferred to a rotary kiln, heated at a rate of 2 K/min,
and calcined in air for 3 h at 823 K. At this stage, the metal
carbonates are transformed into the oxides, and the binder
particles build a strong matrix of γ -alumina.
3. Results and discussion
2.2. Characterization
3.1. Morphology of the extrudates
The morphology of particles of the zinc/copper carbon-
ate precursor was examined by means of scanning electron
microscopy (Hitachi S-4500). The particle-size distribution
of these precursor particles in aqueous suspension contain-
ing 1 g/L of solid was measured by means of laser diffrac-
tion (H9236, Sympatec Helos). Mercury intrusion (Auto-
pore III, Micromeritics) was used to characterize porosities
and pore-size distributions of the finished extrudates. Prior to
the measurements, the extrudates were dried for 2 h at 393 K.
The specific copper surface areas were measured by means
of nitrous oxide (N2O) decomposition at atmospheric pres-
sure after reducing the CuO in the extrudates with hydrogen
(purity: 99.999% m/m). The reduction was carried out with
3% H2 in helium (purity: 99.996% m/m) at a volumetric
flow rate of 300 cm3/min (STP). After raising the temper-
ature from 413 to 513 K at a rate of 20 K/h, the diluted
hydrogen stream was replaced stepwise by pure hydrogen
within 1 h. Finally, the samples were cooled to 333 K in
pure helium and exposed to a flow of 100 cm3/min (STP) of
0.1% N2O in helium. The copper surface areas were calcu-
lated from the consumed amount of nitrous oxide according
to the method described by Chinchen et al. [16].
Extrusion experiments were carried out with both the car-
bonate precursor of Cu and Zn and the oxides obtained after
calcination. In any case, the extrusion of the pastes was diffi-
cult. Many trials were necessary to identify the compositions
suitable to obtain stable and smooth extrudates, and these
compositions are reported in the experimental part. Eventu-
ally, the carbonate precursor was preferred as a raw material
because one additional calcination step can be avoided in
this way. A decisive factor for the quality of extrudates is
the rheological behavior of the paste. In the case of Cu/Zn-
containing material, very small variations in the water con-
tent ( 1% m/m) result in dramatic changes in the rheo-
logical properties. A low proportion of water results in a
highly viscous, brittle paste with low plasticity which can-
not be pressed through the nozzle of the extruder. When the
amount of water is too high, the greenbodies become fis-
sured and scaly as soon as they leave the nozzle (Fig. 2a).
Most beneficial for extrusion is a plastic paste which does
not display viscous flow in the nozzle but rather undergoes
deformation, thereby forming a thin liquid film at the wall
which enables slipping (wall slip behavior). As shown in
2.3. Catalytic experiments
The catalytic experiments were carried out in a continu-
ous flow unit equipped with a stainless-steel fixed-bed tubu-
lar reactor and on-line gas analysis. The catalyst bed had a
diameter of 13 mm and a length of approximately 40 mm
and consisted of extrudates diluted with nonreactive silicon
carbide particles exhibiting a mean diameter of 0.5 mm in a
mass ratio of 2:1. Upflow, the reactor void volume was filled
with pure SiC particles in order to ensure a plug flow profile
of the feed when contacting the catalyst bed. Prior to the ex-
periments, the catalysts were reduced in situ at atmospheric
pressure in the same way as described above for the mea-
surement of the copper surface. Hydrogen and evaporated
Fig. 2. Extruded strings produced from pastes containing boehmite, Cu-
and Zn-carbonate. (a) Greenbody with surface defects, water content of the
paste > 22% m/m. (b) Smooth greenbody, water content of the paste =
22% m/m. (c) Extruded string after calcination.

422
S.P. Müller et al. / Journal of Catalysis 218 (2003) 419–426
Fig. 4. Particle volume-size distribution q of the Cu/Zn-carbonate crystals.
3
Fig. 3. SEM micrograph of Cu/Zn-carbonate crystals (Cu:Zn = 1:1).
Fig. 2b, the resulting greenbodies are straight and exhibit
a smooth surface. Upon subsequent calcination, the color
changes from light greenish blue to brown because of the
transformation of CuCO3 into CuO (Fig. 2c).
3.2. Pore texture and strength
Extrudates can be conceived as formed agglomerates of
small particles. During calcination, at least a part of the par-
ticles representing the framework will form interconnecting
solid bridges, be it by condensation reactions or by a certain
degree of sintering. The voids in-between the particles rep-
resent the pores. Hence, porosity and pore-size distribution
of extrudates depend on the size distribution of the particles
representing the framework.
Fig. 5. Effect of the kneading time of the extrusion paste on the pore-size
distribution of the resulting, calcined extrudates with c = 75% m/m.
The boehmite particles used as a binder are of spherical
shape and exhibit a mean diameter of 45 nm. In contrast, the
particles of the Cu/Zn carbonate precursor are very large. As
shown in Fig. 3, the particles are large intergrowths of frag-
ile, plate-like crystals. Accordingly, the particle volume-size
distribution q3 (i.e., the derivative of the cumulative volu-
metric size distribution as a function of the equivalent diam-
eter) of the material is characterized by a bimodal population
(Fig. 4). The major part of the particles exhibits a mean di-
ameter of 16.4 µm. The second peak in the size distribution
resulting from particles of 2–3 µm is probably originating
from fragments of the intergrowths.
When extrudates are formed from such a material to-
gether with binder particles of much smaller size, a pore-size
distribution with three size maxima can be expected. This is
shown in Fig. 5 for three calcined extrudates of equal compo-
sition (c = 75% m/m). The smallest pores (6 nm) predomi-
nantly represent the voids between the binder particles. The
two types of macropores stem from voids between Cu/Zn
oxide particles. The difference between the samples refers to
the duration of kneading prior to extrusion. With increasing
kneading time, i.e., mechanical energy input, a larger por-
tion of the intergrowths of the carbonate precursor is broken
Fig. 6. Torque at the rotors of the kneader at a constant rotor speed of
40 min as a function of the kneading time for three paste compositions
(equal masses).
−1
to smaller particles and fragments. As a result, the size of
the macropores in the extrudates shifts to lower values with
increasing kneading time.
Obviously, the pore-size distribution in the extrudates
must also depend on the solid composition because the re-
spective particle sizes differ so much. In Fig. 6, three knead-
ing diagrams are depicted which were produced with equal

S.P. Müller et al. / Journal of Catalysis 218 (2003) 419–426
423
Table 1
Effect of paste composition and kneading time on the porosity of finished
(calcined) extrudates
Sample designation
c
Kneading time
(min)
Porosity
(% v/v)
(% m/m)
25/2
25
25
25
50
50
50
75
75
75
2
71.7
64.0
60.6
67.0
66.2
63.8
71.7
71.3
68.8
25/10
25/20
50/10
50/20
50/60
75/5
10
20
10
20
60
5
75/10
75/60
10
60
Fig. 7. Effect of the paste composition on the pore-size distribution of the
resulting, calcined extrudates after equal kneading times of 10 min. Light
gray line: sample 25/10, dark gray line: sample 50/10, and black line: sam-
ple 75/10 (see Table 1 for the designation).
masses of pastes containing equal amounts of water, but dif-
fering in the concentration (c) of Cu/Zn carbonate. This
diagram is useful because the input of mechanical energy is
represented by the area below the curves, i.e., the time inte-
gral of the torque at the rotors of the mixer at constant rotor
speed. It indicates that pastes containing a high amount of
binder are exposed to higher shear forces than pastes with
less binder and a higher value of c when mixed at the same
rotor speed. The corresponding mercury intrusion plots of
the calcined extrudates (kneading time 10 min in all cases)
are shown in Fig. 7. Large macropores are present only if
the solid content of Cu/Zn carbonate in the paste amounts
to 50% or higher. Again, it can be seen that the size of the
macropores shifts to lower values with increasing energy
input. In contrast, mesopores dominate when the amount
of binder is high, both because the small binder particles
are prevailing and because the high shear forces result in a
more pronounced breakdown of the Cu/Zn carbonate parti-
cles.
Porosities of a selection of finished extrudates are sum-
marized in Table 1. In accordance with the previous dis-
cussion it can be stated that the porosity decreases with in-
creasing proportion of small particles in the framework of
the extrudate which are either added as such or which are
produced by input of energy upon kneading.
A qualitative inspection of the mechanical properties by
manual breaking or squeezing of the extrudates revealed
that their strength increases with increasing amount of
γ -alumina. In addition, the kneading time has a pronounced
effect. For instance, sample 50/60 obtained with a knead-
ing time of 60 min exhibits a bulk crush strength (according
to the American standard ASTM D 4179) of 79.6 N, indi-
cating excellent suitability for application in a large-scale
fixed-bed reactor. In contrast, the same paste composition
after 5 min kneading time yields extrudates with an insuffi-
cient bulk crush strength of 7.1 N. These few data can only
give a flavor of the strong impact of formulation and forming
Fig. 8. Copper surface area of the extrudates as a function of the Cu/Zn
content.
conditions on the applicability of catalysts in industrial-size
reactors. We will devote more research to this issue in the
near future.
3.3. Copper surface and activity
As shown in Fig. 8, the copper surface area of the ex-
truded catalysts linearly increases with increasing copper
content. In the same graph, we have also added the copper
surface area of pure, nonextruded Cu/ZnO. This compari-
son indicates that the magnitude of the copper surface area
is negatively affected by the catalyst formulation and form-
ing. On the basis of earlier results reported by Schlander and
Turek [8], one should expect that the activity of the cata-
lysts in the conversion of dimethyl maleate follows the same
trend as the copper surface area. The following proportional-
ity should hold for equal modified residence time and equal
temperature:
−ln(1 − X) ∝ SCu
.
(2)
However, as shown in Fig. 9, our results indicate that the
extruded catalysts are much more active than would be ex-
pected from the low values of their copper surface areas,
this discrepancy being most pronounced at temperatures of
473 K or higher. It is important to note that the data of pure
Cu/ZnO stem from tiny particles, (crushed pellets) with di-
ameters ranging from 200 to 315 µm. Here, a negative effect
of diffusional limitations can be excluded. The extrudates,
in contrast, were all used in full size, i.e., with lengths of

424
S.P. Müller et al. / Journal of Catalysis 218 (2003) 419–426
(a)
(b)
(c)
Fig. 9. Conversion of dimethyl succinate (X
= 100% in all cases) as
maleate
a function of the residence time for three reaction temperatures. Full-size
extrudates of sample 25/10, white bars; sample 50/10, gray bars; and sam-
ple 75/10, black bars. Pure Cu/ZnO particles of diameter 200 to 315 µm,
squared bars.
5 mm and diameters of 2 mm. It is the presence of γ -alumina
which results in an enhancement of the DMS conversion.
This effect is not to be ascribed to a hydrogenating function
of this binder, but presumably to the fact that the hydrogena-
tion products GBL and BDO are rapidly withdrawn from
the respective equilibria because γ -alumina boosts the rate
of the subsequent dehydration.
Also within the series of full-size extruded catalysts, there
is no linear correlation between the activity (here expressed
as −ln(1 − X)) and the copper surface area. Fig. 9 shows
that the extrudates with the highest copper surface area (sam-
ple 75/10, c = 75% m/m) display almost the same activ-
ity as those exhibiting medium copper surface area (sample
50/10 with c = 50% m/m). Tests carried out with extrudates
of equal composition and different pore texture indicated
that the reaction rates are limited by pore diffusion when the
proportion of macropores is low. These preliminary results
show that the apparent kinetics are influenced by the pore
texture, and that the intrinsic activities of the extrudates are
a function of both the copper surface area and the proportion
of γ -alumina. A detailed kinetic analysis will be required,
first in the absence of transport limitations (crushed extru-
dates), and subsequently using extrudates exhibiting differ-
ent sizes and pore textures, to characterize quantitatively the
effect of mass transport in extrudates. We will report on this
work in the future.
Fig. 10. Selectivity to C -products versus dimethyl succinate conversion at
4
453, 473, and 493 K and p = 2.5 MPa. Circles, THF; triangles, GBL; and
squares, BDO. (a) Catalyst sample 25/10, (b) sample 50/10, and (c) sample
75/10.
of the C4-products according to
xi,C
4
ꢀ
Si =
(3)
x
i,C4
3.4. Catalyst selectivity
without taking the coupled production of methanol and wa-
ter into account. In Fig. 10, the integral selectivities to
C4-products obtained at three different temperatures are de-
picted. The experiments were carried out with the cata-
lyst samples 25/10, 50/10, and 75/10 (see Table 1 for
the designation). In all cases, the THF selectivity increases
with increasing conversion (i.e., increasing residence time)
and temperature. Because THF is produced in the acid-
catalyzed dehydration of BDO, the amount of weakly acidic
The C4-products formed over our catalysts are GBL,
BDO, and THF. Fouling by polyester deposition was not
observed. Other unwanted by-products such as 1-butanol or
carbon dioxide were formed in negligible traces only, and
the carbon mass balance was validated (100 2%) in the
experiments reported here. Hence, it is possible to define the
integral (reactor) selectivity on the basis of molar fractions

S.P. Müller et al. / Journal of Catalysis 218 (2003) 419–426
425
choice of particle size of the powders used, and by the in-
put of mechanical energy during formulation. It will be a
challenging task in the future to analyze quantitatively the
interaction among the activities of hydrogenation sites, acid
sites, and the diffusional properties of extruded catalysts, and
to adjust the balance between these properties for a desired
activity and product distribution.
Acknowledgment
Fig. 11. Selectivity to C -products at a dimethyl succinate conversion of
4
98%. T = 493 K, p = 2.5 MPa.
We thank Zwick/Roell GmbH for performing the mea-
surements of the bulk crush strengths according to standard
ASTM D 4179.
γ -Al2O3 in the catalyst also has an effect. As can be seen
in Fig. 10c, BDO is released only when the catalyst contains
a low amount of γ -Al2O3 (sample 75/10). Higher concen-
trations of γ -Al2O3 result in a complete consumption of
BDO.
Appendix A. Symbols and abbreviations
To illustrate more clearly the effect of the acidic binder, a
comparison is made in Fig. 11 between a catalyst consisting
of pure Cu/ZnO and the alumina-bound extrudates 50/10.
The selectivities to C4-products are plotted for the same con-
version of X = 98% at 493 K. The molar ratio BDO:GBL
produced over pure Cu/ZnO approximately equals the ther-
modynamic equilibrium value at this temperature and pres-
sure. Because of the relatively high temperature, a small
amount of THF is formed as well. However, the dehydration
does not disturb the equilibrium between GBL and BDO. In
contrast, this equilibrium is completely shifted toward BDO
because its subsequent dehydration is strongly accelerated
when the alumina-bound catalyst is applied. This result is
of technical importance because it means that the formation
of THF is not thermodynamically limited by the pressure-
dependent equilibrium reaction of GBL to BDO. Hence,
very low reaction pressures can be applied in a single-stage
process.
c
concentration of Cu- and Zn-carbonate in water-
free extrudates; see Eq. (1).
mass of species i.
mi
p
pressure, Pa.
SCu
Si
copper surface area, m2/g.
reactor selectivity to C4 product species i; see
Eq. (3).
temperature, K.
T
tmod
modified residence time, g s/cm3; defined as the ra-
tio of the catalyst mass (basis: total mass of the
calcined extrudates loaded into the reactor) to the
total volumetric flow at reaction conditions.
WHSV weight hourly space velocity, h−1; defined as the
ratio of the mass flow of dimethyl maleate into the
reactor to the catalyst mass (basis: total mass of the
calcined extrudates loaded into the reactor).
xi,C
X
molar fraction of C4 product species i.
conversion of dimethyl succinate (conversion of
dimethyl maleate is 100%).
4
4. Conclusions
GBL
γ -butyrolactone.
BDO 1,4-butanediol.
THF tetrahydrofuran.
We have demonstrated that it is possible to produce THF
yields of 100% from dimethyl maleate in a single stage by
using a fixed-bed of Cu/ZnO extruded with γ -alumina as a
binder. The reaction pressures can be very low as compared
to the two-stage process via BDO because the hydrogenation
equilibrium between the intermediates GBL and BDO is not
limiting when BDO is removed immediately by dehydration
over the acidic sites provided by the binder. In addition, it
can be anticipated that much less dilution with hydrogen will
be required to prevent fouling by polyester deposits in the
single-stage gas-phase process. This assumption, however,
must still be validated in future experiments.
References
[1] H. Müller, in: Industrial Organic Chemicals: Starting Materials and
Intermediates (Ullmann’s Encyclopedia), Vol. 8, Wiley–VCH, Wein-
heim, 1999, p. 4621.
[2] ECN, 14 February, 14 (2000).
[3] W. Reppe, E. Keyssner, US patent 2,232,867 (1941).
[4] M.L. Morgan, Chem. Ind. 3 (1997) 166.
[5] K. Weissermel, H.-J. Arpe, Industrial Organic Chemistry, Wiley–
VCH, Weinheim, 1997.
[6] M. Sharif, K. Turner, US patent 4,584,419 (1986).
[7] J. Kanetaka, T. Asano, S. Masamune, Ind. Eng. Chem. 62 (1970) 24.
[8] J.H. Schlander, T. Turek, Ind. Eng. Chem. Res. 38 (1999) 1264.
We have also shown that the forming of Cu/ZnO catalysts
by extrusion is an interesting alternative to pelleting. The re-
spective proportions of binder and hydrogenating function
can be varied, and the pore texture can be influenced by

426
S.P. Müller et al. / Journal of Catalysis 218 (2003) 419–426
[9] C. Ohlinger, B. Kraushaar-Czarnetzki, Chem. Eng. Sci., in press.
[14] R.-H. Fischer, F. Stein, R. Pinkos, M. Hesse, M. Sprague, M. Rösch, H.
Borchert, S. Schlitter, R.-T. Rahn, A. Weck, German patent application
DE 100 61 556 A1, 2002.
[15] H. Borchert, S. Schlitter, M. Rösch, R.-H. Fischer, R.-T. Rahn, A.
Weck, German patent application DE 100 61 553 A1, 2002.
[16] G.C. Chinchen, C.M. Hay, H.D. Vanderwell, K.C. Waugh, J. Catal.
103 (1987) 79.
[10] U. Herrmann, G. Emig, Ind. Eng. Chem. Res. 62 (1997) 2885.
[11] U. Herrmann, G. Emig, Chem. Eng. Technol. 21 (1998) 285.
[12] M. Mokhtar, C. Ohlinger, T. Turek, Chem. Eng. Technol. 24 (2001)
423.
[13] M.S. Spencer, in: M.V. Twigg (Ed.), Catalyst Handbook, Wolfe, Lon-
don, 1989, p. 17.