Ozonized activated carbon as catalyst for MTBE-cleavage

Article abstract of DOI:10.1016/j.apcata.2011.02.018

The application of oxidized carbon materials for the cleavage of methyl-tert-butyl ether has been investigated under technical relevant conditions in a fully automated apparatus with paralleled tube reactors. The carbon spheres oxidized for 60 min (SAC60) are active for MTBE cleavage at 498 K and 6 bar(g). MTBE is selectively converted into isobutene (S > 99%) and MeOH (S > 99.9%) at 80% conversion. The catalytic activity of the oxidized activated carbon can be traced back to the surface groups generated by ozonization of the carbon as the untreated sample did not show any conversion of MTBE at all. Broensted acidic carboxylic acid and anhydride surface groups are responsible for the activity of oxidized carbon as was concluded from simulation of the ex situ DRIFT spectra of the applied materials. In contrast to commonly used amorphous alumosilicates, the modified carbon completely inhibited the production of the undesired by-product dimethylether. Therefore, as no Lewis-acid sites are present on SAC60 and none of the present surface groups (carboxylic acids, anhydrides, lactones and carbonyl groups) cause the dehydration of methanol into dimethylether, it can be concluded that DME formation most probably occurs purely on Lewis-acidic or on basic sites. The decrease in conversion over time-on-stream obtained for the oxidized carbon material is ascribed either to poisoning of present active sites or to a decarboxylation of the activated carbon.

Full text of DOI:10.1016/j.apcata.2011.02.018

Applied Catalysis A: General 397 (2011) 103–111
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Ozonized activated carbon as catalyst for MTBE-cleavage
A. Naua,∗, S. Kohlb, H.-W. Zanthoffa, H. Wiederholdc, H. Vogelb
a Evonik Industries GmbH, Paul-Baumann-Str. 1, 45772 Marl, Germany
b Technische Universität Darmstadt, Ernst-Berl-Institut, Petersenstraße 20, 64287 Darmstadt, Germany
c Evonik Industries GmbH, Rodenbacher Chaussee 4, 63457 Hanau, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
The application of oxidized carbon materials for the cleavage of methyl-tert-butyl ether has been inves-
tigated under technical relevant conditions in a fully automated apparatus with paralleled tube reactors.
The carbon spheres oxidized for 60 min (SAC60) are active for MTBE cleavage at 498 K and 6 bar(g). MTBE
is selectively converted into isobutene (S > 99%) and MeOH (S > 99.9%) at 80% conversion. The catalytic
activity of the oxidized activated carbon can be traced back to the surface groups generated by ozoniza-
tion of the carbon as the untreated sample did not show any conversion of MTBE at all. Broensted acidic
carboxylic acid and anhydride surface groups are responsible for the activity of oxidized carbon as was
concluded from simulation of the ex situ DRIFT spectra of the applied materials. In contrast to com-
monly used amorphous alumosilicates, the modified carbon completely inhibited the production of the
undesired by-product dimethylether. Therefore, as no Lewis-acid sites are present on SAC60 and none of
the present surface groups (carboxylic acids, anhydrides, lactones and carbonyl groups) cause the dehy-
dration of methanol into dimethylether, it can be concluded that DME formation most probably occurs
purely on Lewis-acidic or on basic sites. The decrease in conversion over time-on-stream obtained for the
oxidized carbon material is ascribed either to poisoning of present active sites or to a decarboxylation of
the activated carbon.
Received 9 November 2010
Received in revised form 3 February 2011
Accepted 17 February 2011
Available online 24 February 2011
Keywords:
MTBE cleavage
Broensted-acidity
Activated carbon
DRIFTS
Oxygen surface groups
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
methacrylate. This is due to the fact that the MTBE cleavage reac-
tion (1) is accompanied by several secondary reactions, mainly the
Cleavage of methyl-tert-butyl ether (MTBE) is a well known
technically applied method for the production of isobutene. Liq-
uid and gas phase operation have been proposed so far [1,2]. In
general, the reaction is carried out in the presence of solid acidic
catalysts:
dimerization of isobutene into 2,4,4-trimethylpentene-1 and 2,4,4-
trimethylpentene-2 (here: sum. as C8):
2i-C4H8 → C8H16
(2)
the dehydration of methanol with the formation of dimethyl ether
(DME):
C5H12O ↔ i-C4H8 + CH3OH
(1)
2CH3OH ↔ C2H6O + H2O
(3)
In literature, most attention has been paid to amorphous
silica–alumina [3], ion-exchange resins [4–6], alumina or silica
modified with various metal sulfates [7], supported phospho-
ric acid [8] and zeolites [9] as effective catalysts. Although high
selectivities are reported exceeding 99% towards isobutene and
99% towards methanol, a remarkable expense for purification is
required to remove the by-products ensuring high isobutene puri-
ties to satisfy the specification required for e.g., the subsequent
polymerization into isobutene rubber or the oxidation to methyl
and the hydration of isobutene with the formation of tert-butyl
alcohol (TBA):
i-C4H8 + H2O ↔ C4H9OH
(4)
All these main and side reactions occur in the presence of cat-
alytic sites, but so far very little is known what kind of acidity,
Broensted or Lewis, and what strength of these sites is necessary to
obtain optimum splitting conditions.
Catalyst development and optimization is therefore ongoing
aiming at lower side product formation and longer catalyst life
times as it is limited due to poisoning by side products and other
effects. New and more selective catalysts are supposed to enhance
the process with the objective of improving the economical and
ecological efficiency.
∗
Corresponding author. Tel.: +49 2365/49 9322.
E-mail addresses: asli.nau@evonik.com (A. Nau),
kohl@ct.chemie.tu-darmstadt.de (S. Kohl), horst-werner.zanthoff@evonik.com
(H.-W. Zanthoff), holger.wiederhold@evonik.com (H. Wiederhold),
vogel@ct.chemie.tu-darmstadt.de (H. Vogel).
0926-860X/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2011.02.018

104
A. Nau et al. / Applied Catalysis A: General 397 (2011) 103–111
The application of innovative carbon materials in this content is
The mass balance is checked for each analytical result obtained.
All results discussed below feature a deviation of less than 3%.
The modified residence time of MTBE was varied between 0.02
and 0.2 h gKat g−1 in order to visualize the catalytic performance
over the whole conversion range keeping temperature and pressure
constant.
In order to classify the activated carbon vs. industrially available
catalysts for MTBE cleavage, the performance of a commercially
available amorphous alumina–silica (AAS) prepared according to
[13], was used. AAS is already applied for MTBE cleavage in indus-
trial scale.
In order to study the time-on-stream behavior of the catalyst
samples, a fixed bed reactor (L = 600 mm, di = 11 mm) divided into
the vaporizer zone (330 mm), the reaction zone (40 mm) and the
outlet (230 mm) was designed. 2.25 g of the catalyst sample was
diluted (1:1) with granular quartz. Again, MTBE was fed as a liq-
uid by a mass flow controller and the residence time was varied
between 0.2 and 0.04 h gKat g−1. The reaction was carried out at
498 K and 6 bar(g). The off-gas of the reactor was analyzed as
described above.
a new approach to improve the process performance and to better
understand the complex reaction system of MTBE cleavage. Carbon
materials have been used for decades in heterogeneous catalytic
reactions. They help to increase the rate and to control the selectiv-
ity of many chemical reactions [11]. Oxidative surface modification
of activated carbon allows the generation of new functional acidic
groups on the carbon surface, e.g., carboxyl, quinone or phenol
groups, in order to adjust the surface chemistry properties. The
functional groups can participate in redox reactions such as the oxi-
dation of SO2 to SO3 (SO2 removal from flue gas). Activated carbon
can also be used as catalyst for the synthesis of phosgene from CO
and chlorine. Activated carbon has no Lewis acidity and therefore, it
is superior suited as catalyst support for the industrial synthesis of
products like vinyl acetate than other oxidic supports because the
undesired polymerization of the vinyl monomers does not occur
[12].
It was the aim of the present study, to generate new catalyti-
cally active materials by oxidation of activated carbon shaped as
spheres by ozone, and apply those as catalysts for MTBE cleavage
into isobutene and methanol. In the following, the results about
the catalytic performance of the activated carbon materials are
presented.
2.3. Characterization methods
Structural properties of the catalysts applied for MTBE cleavage
in the present study are reported in Table 1. The amount of acid
sites was determined by NH3–TPD, the total surface area and the
average pore size were calculated applying the BET-method; the
external surface area and the total pore volume were determined
by the t-plot method.
Scanning electron microscopy (SEM) micrographs were
obtained using a Zeiss 962 microscope operating at 30 and 25 kV,
respectively.
2. Experimental
2.1. Oxidative treatment of the activated carbon
Ozone was used as oxidizing agent to modify the surface of a
commercial, spherical activated carbon (SAC) with an internal sur-
face area of 1306 m2 g−1 and a particle size of 0.5–0.6 mm (received
from CarboTechAC GmbH, Essen). Oxidation was carried out in a
stainless steel fixed bed reactor (45 mm × 60 mm × 60 mm) con-
nected with an IR spectrometer (FTS 175 C, BIORAD) to analyze
the gas phase (CO, CO2, O3, and H2O). The fixed bed reactor was
equipped with four heating cartridges (per 200 W) and a thermo-
couple. For the oxidation, the reactor was filled ten times with
550 mgof theactivated carbonSAC. Thesamplewas treated at333 K
with 4000 ppm (L L−1) ozone in a N2/O2 (1:1.3) stream for 60 min
(sample herein after called SAC60). Ozone was received from an
ozonizer (COM-SD-30, Anseros, Tuebingen).
Diffuse reflectance FT-IR spectra (DRIFTS, optical resolution
4 cm−1) were collected at room temperature with a FT-IR spectrom-
eter (FTS 175 C, BIORAD) equipped with a DRIFTS cell [14]. During
the spectra collection the DRIFT cell was flushed with 165 mL min−1
nitrogen. The untreated activated carbon (SAC) was used as ref-
erence material to highlight the changes due to the oxidation of
the material and to receive a straight baseline for the subsequent
simulation of the spectra. The characteristic IR absorptions of the
oxygen surface groups overlap in the IR spectrum. To differentiate
between the functional oxygen groups on the activated carbon the
ex situ DRIFT spectra were simulated with single Gauss functions in
the region between 1700 and 1850 cm−1. The simulated spectrum
results from the superposition of the used four single Gauss func-
tions. The fit is optimized by varying the parameters half-width,
peak position and integral value of each single Gauss function and
thereby minimizing the sum of least squares.
The temperature programmed desorption experiments (TPDs)
were carried out in the reactor mentioned above. 250 mg of each
sample (SAC, SAC60, SACar and SAC60ar, whereas ‘ar’ indicates the
sample after the use as catalyst for MTBE cleavage in the paralleled
tube reactor experiments) were heated up to 773 K in a N2 flow
(165 mL min−1) at a heating rate of 10 K min−1. The end tempera-
ture was held for half an hour. The evolved gases were analyzed by
an attached IR gas cell. The elemental analysis of the carbon mate-
rial for C and H was carried out with a VarioEL analyzer, O was
calculated as difference to 100%.
2.2. Catalytic performance
The catalytic tests were carried out in a fully automated appara-
tus with paralleled tube reactors (L = 80 mm, di = 8 mm) at 498 K
and 6 bar(g). 0.2 g of catalyst was loaded to a reactor. The cat-
alyst samples were diluted (1:5) with granular quartz in order
to ensure an isothermal temperature distribution in the catalyst
bed. In one test series, two reactors were always loaded with
the same catalyst sample. The redundant use of the catalyst sam-
ples allows the verification of the reproducibility of the catalyst
performance.
MTBE was dosed as a liquid by a mass flow controller and was
vaporized in a tubular reactor (L = 200 mm, di = 24 mm) consisting
of 2 mm glass beads. Subsequently, the outlet gas was divided into
identical parallel streams, which were led into the paralleled tube
reactors, respectively. Via a selecting valve, the off-gas of one reac-
tor was led to a gas chromatograph that was connected online
(HP 6890). The streams of the other reactors were collected and
disposed of according to the safety regulations.
3. Results and discussion
The gas chromatograph was equipped with a flame ionization
detector and in series connected columns HP-1 methylsiloxane and
LowOx, which allow the separation of all known products of the
MTBE cleavage in 15 min.
3.1. Catalytic performance
Isobutene and methanol are the main products of the MTBE
cleavage over all tested catalysts. For AAS the selectivity towards

A. Nau et al. / Applied Catalysis A: General 397 (2011) 103–111
105
Table 1
Structural properties of catalysts applied for MTBE cleavage.
Catalyst
Amount of acid
sites (mmol g−1
BET-surface area
External surface area
Total pore volume
Average pore size
width (nm)
)
(m2 g−1
)
(t-plot) (m2 g−1
)
(cm3 g−1
)
SAC
SAC 60
0.003
0.053
1306
1217
0.57
0.54
0.57
0.54
1.78
1.77
2,0
1,5
1,0
0,5
0,0
SAC
SAC60
AAS
SAC
SAC60
AAS
5
4
3
2
1
0
0
20
40
60
80
100
0
20
40
60
80
100
MTBE Conversion / %
MTBE Conversion / %
Fig. 1. DME formation vs. MTBE conversion for untreated carbon spheres (SAC),
Fig. 3. TBA formation vs. MTBE conversion for untreated carbon spheres (SAC),
for 60 min oxidized carbon spheres (SAC60) and in comparison, amorphous
for 60 min oxidized carbon spheres (SAC60) and in comparison, amorphous
˙
−1
.
˙
alumina–silica (AAS) at 498 K, 6 bar(g) and mKat/V = 0.02–0.2 h g
g
alumina–silica (AAS) at 498 K, 6 bar(g) and mKat/V = 0.02–0.2 h gKat g−1
.
Kat
these products amounted to 99.95% and 95%, respectively. For
SAC60, almost same selectivities for isobutene can be reached
(99.1%) and methanol selectivities of nearly 100% are achieved. For
SAC only conversions below 10% were achieved with selectivities
for both products of 100%. More interestingly, the selectivities of the
by-products differ strongly for the above mentioned catalysts. The
catalytic performance of SAC, SAC60, and of the technically applied
amorphous alumina–silica is shown in Figs. 1–3. Mainly, the by-
products DME, oligomers and TBA are detected analytically. Other
components were below detection limit of the gas chromatograph
and are not further discussed. As it can be seen in the figures, the
by-product formation is on a low level for both catalyst systems.
Figs. 1–3 indicate that the untreated carbon spheres (SACs)
exhibit almost no conversion (<10%) and hence by-products cannot
be identified analytically.
The carbon spheres oxidized for 60 min (SAC60) are active for
MTBE cleavage at 498 K and 6 bar(g). With increasing residence
time, the MTBE conversion increases up to 90%.
In contrast, the amorphous alumina–silica exhibits higher con-
version rates under equal residence time conditions and reaches at
96–98% conversion chemical equilibrium (Keq = 20) [10].
Considering the selectivities towards DME for SAC60, DME for-
mation was below detection over the whole range of conversion
(see Fig. 1). The production of DME is completely inhibited in the
presence of the oxidized activated carbon.
In contrast, AAS shows increasing DME selectivity with increas-
ing conversion rate. At 80% conversion, AAS features almost 3%
selectivity towards DME.
Dimers, formed by oligomerization of isobutene, are another
critical secondary product. Fig. 2 shows the selectivity dependence
towards C8 products for the three tested catalysts on MTBE con-
version. Up to 40% conversion, no C8 can be detected for SAC60.
For higher conversions, the selectivity towards dimers increases
for SAC60. At 80% conversion, around 0.1% C8 selectivity is reached.
In contrast, AAS exhibits a nearly constant value for C8 selectiv-
ity with increasing conversion. For AAS, the C8 formation at 80%
conversion averages around 0.04%. Therefore, the formation rate
of dimers using SAC60 is in the range of the common amorphous
alumina–silica described in literature [6].
0,20
SAC
SAC60
AAS
0,15
0,10
0,05
0,00
The results of the TBA formation rate using the three tested cat-
alysts are shown in Fig. 3. It can be clearly seen again that SAC60
shows a completely different behavior compared to AAS. At low
conversion, SAC60 produces most TBA and the selectivity decreases
with increasing conversion. Similar selectivities towards TBA of
both catalysts are reached at around 80% conversion and averages
around 0.1–0.2%. AAS shows a constant TBA formation rate over the
whole conversion range. As expected, SAC shows negligible activity
and no detectable TBA formation.
It can be summarized that the untreated activated SAC does not
exhibit the necessary active sites for MTBE cleavage. Only after the
oxidation of the activated carbon, MTBE cleavage takes place under
the conditions mentioned above.
0
20
40
60
80
100
MTBE Conversion / %
Fig. 2. C8 formation vs. MTBE conversion for untreated carbon spheres (SAC),
for 60 min oxidized carbon spheres (SAC60) and in comparison, amorphous
˙
alumina–silica (AAS) at 498 K, 6 bar(g) and mKat/V = 0.02–0.2 h gKat g−1
.

106
A. Nau et al. / Applied Catalysis A: General 397 (2011) 103–111
Fig. 4. The reaction mechanism of MTBE cleavage into isobutene and methanol.
The reaction mechanism of MTBE cleavage can be traced back
a favored hydrolysis of MTBE and water into TBA takes place using
SAC60 as catalyst and this behavior is to be ascribed to the sur-
face groups. In general, the mechanism of the reaction can also be
explained by the available Broensted-acid sites, which donates a
proton for the partially negative charged oxygen atom of the ether
molecule. Herein, a methoxylation of the OH-groups occurs and in
the presence of water TBA and methanol are formed (Fig. 5).
The source of water for TBA formation out of MTBE can pre-
sumably be ascribed to the feed, which includes water. Obviously,
SAC60 and therefore Broensted-acid sites preferably catalyze the
reaction of MTBE with water into TBA and methanol than AAS. As
AAS has also other surface groups, like Lewis-acid sites as well as
basic sites, it might convert produced TBA directly into isobutene
and water or hinder the conversion of MTBE and water into TBA
and methanol. Further studies are necessary in order to get insight
into the differences of TBA formation for both catalyst systems. For
to available Broensted-acid sites, which donates a proton for the
partially negative charged oxygen atom of the ether molecule. The
attack results in the cleavage of methanol and isobutene (Fig. 4):
Comparing the catalytic performance of AAS with SAC60, it can
be seen, that the C8 formation is similar. In each case the selectivity
towards oligomers increases with increasing conversion.
Differences occur in the formation of DME which cannot be
detected for SAC60. Therefore, SAC60 does not provide the active
sites which are responsible for the reaction of methanol into DME
as in the case of AAS. As SAC60 exhibits only Broensted-acidic sites,
it can be concluded, that those do not form DME and hence only
Lewis-acidic or basic sites must be responsible for DME formation
out of methanol.
The TBA formation also differs for both considered catalyst sys-
tems. As TBA selectivity is high for SAC60, it can be assumed that
Fig. 5. The reaction mechanism of MTBE cleavage with water into TBA and methanol.

A. Nau et al. / Applied Catalysis A: General 397 (2011) 103–111
107
100
80
60
40
20
0
sion is obtained at the final residence time and is either ascribed to
poisoning of all present active sites or to a decarboxylation of the
activated carbon as discussed below.
SAC60
3.2. Characterization of the activated carbon catalyst
5 * RT
2 * RT
3.2.1. Structural properties of the tested catalyst systems
The comparison of the structural properties of SAC and SAC60
shows that the oxidative treatment influences these properties
by less than 10%. Only the amount of acid sites is significantly
enhanced after the oxidation treatment by ozone. The reference
system AAS exhibits a 6-time lower BET-surface area but an 8-
times higher amount of acid sites. Nevertheless, the amount of acid
sites is not an evident parameter for the activity of the catalysts
as the strength of the acid sites cannot be quantified. Furthermore,
the activities of both catalyst systems for MTBE cleavage cannot be
compared because of the rapid deactivation rate of SAC60.
In order to decide whether mass transfer limitation by pore dif-
fusion must be considered, the average pore size of a catalyst is
significant. The average pore size of SAC 60 is smaller than a MTBE
molecule (0.6 nm). Therefore, the reaction does not take place in the
pores. For AAS a mass transfer limitation by pore diffusion cannot
be excluded as the pore size is greater than 6 nm. However, experi-
mental tests showed that pore diffusion could also be neglected for
AAS.
RT
0
50
100
150
200
250
300
Time on stream / h
Fig. 6. Long-term stability of SAC60: MTBE conversion is plotted vs. runtime. Tests
are carried out at 498 K and 6 bar(g) in a fixed bed reactor, initial residence time
−1
(RT: 0.04 h g
g
); 2 × initial residence time (2 × RT: 0.08 h gKat g−1); 5 × initial res-
Kat
idence time (5 × RT: 0.2 h gKat g−1).
SAC60, the TBA selectivities decrease with increasing MTBE con-
version as the TBA cleavage into isobutene and water is favored.
In order to study the time-on-stream behavior of oxidized car-
bon spheres, SAC60 was tested for 300 h at 498 K and 6 bar(g) in a
fixed bed reactor. The results are shown in Fig. 6.
3.2.2. SEM
Fig. 7 shows SEM images of the untreated activated carbon SAC
and the oxidized material SAC60. Clear differences between the two
samples can be observed. While the sphericals of the untreated car-
bon are undamaged, the surface of the oxidized particles is covered
with fissures.
Starting with a residence time of 0.04 h gKat g−1 results in an ini-
tial conversion rate of 80%. The activity of the catalyst decreases
rapidly with time on stream so that the conversion rate is less
than a half of its initial value within 30 h. After 50 h the decrease
in the conversion rate decelerates and after 100 h a constant level
at approximately 20% is reached. The residence time is doubled
(0.08 h gKat g−1) in order to examine the time-on-stream behavior
of SAC60 on a high conversion level. The effect is an immediate
doubling of conversion. A further increase of the residence time
(0.2 h gKat g−1) leads to 70–80% conversion. However, the activity
of SAC60 decreases rapidly again so that after another 100 h the
conversion reexhibits 20%. During these experiments, the same
selectivity-conversion behavior towards DME, C8 and TBA was
observed as in the fully automated apparatus and therefore are not
shown here.
3.2.3. Ex situ DRIFT spectroscopy
The ex situ DRIFT spectra of SAC60, SAC60ar and SACar are
depicted in Fig. 8. The differences of the spectra compared to the
spectra of SAC are shown. First, the spectra of SAC60 and SAC60ar
will be discussed.
The spectrum of SAC60 consists of four peaks located at 3500,
1787, 1610 and 1431–1313 cm−1 referring to the groups generated
by ozonization of SAC.
The observed peaks of SAC60 and SAC60ar and the assignments
according to literature are listed in Table 2.
The initial rapid decrease of catalytic activity down to a constant
level of 20% might be ascribed to the adsorption of oligomers on
the active sites of the oxidized carbon material. Hence, a coking of
material is assumed to take place. With increasing residence time,
the conversion is increased. Anyway, a further decrease in conver-
Simulation of the band around 1780 cm−1 with a set of Gauss
functions revealed that the peak consists of four single peaks cen-
tered at 1723, 1764, 1792 and 1835 cm−1 (Table 3).
Literature describes that anhydrides appear in IR spectra as
a doublet peak with one at 1750–1790 cm−1 and the other at
Fig. 7. SEM images (fifty-fold magnification) of the untreated activated carbon (left) and the oxidized activated carbon (right).

108
A. Nau et al. / Applied Catalysis A: General 397 (2011) 103–111
Table 2
IR absorptions in the DRIFT spectra of SAC60 and SAC60ar, and assignments of functional groups.
IR absorptions SAC60 (cm−1
)
Structural group
Ref.
Changes in SAC60ar compared to SAC 60
1313–1431
1610
1787
C–C, C–H, and C–O
Measure for highly conjugated carbonyl/chinone groups
[16]
[15]
[15,17]
[16]
Slight increase
No significant change in intensity
Decrease
C
O streching in lactones, carboxylic acids, anhydrides
3500
O–H groups
Disappeared
Table 3
Results of the simulation of the IR absorption band around 1780 cm−1 in the ex situ DRIFT spectra of SAC60 and SAC60ar using four Gauss functions.
SAC60
SAC60ar
Assigned group
Peak (cm−1
)
Half-width (cm−1
)
Integral value (cm−1
)
Peak (cm−1
)
Half-width (cm−1
)
Integral value (cm−1
)
1723
1764
1792
1835
92
42
39
58
0.15
0.44
0.54
0.54
1743
1756
1782
1830
70
35
38
58
0.15
0.23
0.22
0.12
Lactone
Carboxylic acid
Anhydride
Anhydride
1830–1880 cm−1 [16]. Lactones are supposed to show a single peak
around 1730 cm−1 and carboxylic acids appear in the DRIFT spec-
trum about 1760 cm−1 [15,17]. The changes in the two spectra of
SAC60 can be ascribed to the exposure to the reaction (q.v. Section
3.2.3). The results of simulation show that the peaks assigned to
carboxylic acids and anhydrides decrease after the use as catalyst
while the intensity of lactones remains constant.
As described in literature, characteristic decomposition tem-
peratures of the different identified surface groups are: carboxylic
acids above 423 K, anhydrides 573 K and lactones 623 K [18] (see
Section 3.2.4). Only the decomposition of carboxylic acids due to
the reaction temperature of 498 K was expected. Decreasing anhy-
dride peaks could be explained by esterification with methanol
formed during the reaction. In case that the hereby formed car-
boxylic groups are thermally unstable under reaction conditions
the groups decompose and cannot be transformed into anhydride
groups again. A second way how anhydrides may be reduced could
be the hydrolysis of the groups during the reaction with traces of
water carried in the MTBE feed.
Table 4
CH-amounts from elemental analysis of SAC, SACar, SAC60 and SAC60ar. The amount
of O is calculated as the difference to 100%.
Sample
Elemental analysis, % (g g−1
C
)
H
O
SAC
96.9
92.7
87.0
90.8
0.8
2.6
1.4
2.6
2.4
SACar
SAC60
SAC60ar
4.8
11.6
6.6
cracked into methanol and isobutene. This assumption is sup-
ported by the fact that the molar ratio of H/O is 12:1 for MTBE
(C5H12O) and 12:1 for the increase of H and O in the elemental
analysis of SAC, as it can be seen in Table 3. The oxygen content
of sample SAC60 decreases due to decomposition of functional
groups in consequence of the reaction temperature. So, in this case
there is no indication of oxidation of the catalyst SAC60 during the
reaction.
The spectrum of the untreated activated carbon (SACar), while
totally inactive as catalyst, shows an unexpected formation of
absorption bands around 1300–1400 cm−1. This region includes
the characteristic IR absorptions of C–C, C–H and C–O bondings
[16]. Elemental analysis of SAC shows a doubling of oxygen from
2.4 to 4.8% (g g−1) and an even stronger increase of hydrogen
(Table 4). Probably, MTBE was adsorbed on SAC without being
3.2.4. Temperature programmed desorption
TPD experiments up to 773 K were carried out to analyze the
activated carbon before and after the use as catalyst. The IR gas
phase spectra of sample SAC60 and SAC60ar depending on tem-
perature are depicted in Figs. 9 and 10.
At temperatures around 373 K water (1400–1900 cm−1) is
evolved from SAC60. Increasing the temperature leads to the
release of CO2 (2340 cm−1) that is at a maximum around 450 K.
The gray region in Fig. 9 indicates the stage of TPD where the tem-
0,01
SAC60
SAC60ar
SACar
4500
4000
3500
3000
2500
2000
1500
1000
Wavenumber / cm-1
Fig. 8. Ex situ DRIFT spectra of the activated carbon samples before (SAC60) and
Fig. 9. IR spectra of the gas phase during TPD of SAC60. Black: 298–773 K, and gray:
after (SACar and SAC60ar) use as catalyst (reference: SAC).
temperature held at 773 K.

A. Nau et al. / Applied Catalysis A: General 397 (2011) 103–111
109
Fig. 10. IR spectra of the gas phase during TPD of SAC60ar. Black: 298–513 K, light
gray: 513–613 K, dark gray: 613–773 K, and gray: temperature held at 773 K.
Fig. 12. IR spectrum of the gas phase during TPD of SAC60ar (dark gray region in
Fig. 8).
perature is held at 773 K and shows the declining desorption of CO
and CO2.
For SAC60ar (Fig. 10) four different phases during TPD were
marked by color assigned as follows: 298–513 K (black), 513–613 K
(light gray), 613–773 K (dark gray) and the stage of TPD where
TPD spectrum as well. It can be seen in Fig. 10 that the desorption
of the C8 by-products starts when temperature rises above 600 K.
That means that a noticable amount of the C8 by-products remains
on the catalyst during the reaction. The blocking of active sites on
SAC60 by the C8 by-products (coking) during the reaction could be
one reason beside the decomposition of functional surface groups
for the desactivation of the catalyst.
Comparing the characteristic IR absorption of DME at
1200–1050 cm−1 [18] with the IR spectra in Fig. 10 no DME is
detected. Therefore, it is unlikely that eventually formed DME was
adsorbed on the activated carbon and hence could not be mea-
sured by GC during the MTBE cleavage experiments. However, DME
decomposition or oxidation during TPD cannot be excluded.
CO, CO2 and H2O are evolved during temperature programmed
desorption of the samples. Figs. 13–15 show the evolution of these
gases depending on temperature.
The CO2 liberation (Fig. 13) of sample SAC60 reaches a max-
imum around 450 K and can be assigned to the decomposition
of carboxylic acids. The further CO2 release is traced back to the
decomposition of more stable carboxylic acids, anhydrides and
lactones and only decreases slowly from 500 K. The untreated car-
bon SAC does not show the maximum at 450 K and an observable
desorption of CO2 begins at 500 K. After the reaction CO2 desorp-
tion of both samples initially starts at temperatures above 550 K
with a maximum at 650 K. The difference of evolved CO2 quantities
temperature is held at 773 K (gray). From 513 K CO (2140 cm−1
)
and CO2 (2340 cm−1) are evolved from the sample. When the
temperature rises above 613 K other gases with IR absorptions at
2970–2880 cm−1, 1465 and 1385 cm−1 occur additionally in the gas
phase (assignment see below).
Fig. 11 shows one IR spectrum of the black region in Fig. 8.
The spectrum consists of one broad band around 3000–2900 cm−1
,
a peak at 2835 cm−1 and four bands at 1470, 1360, 1210 and
1100 cm−1. These wavenumbers refer to the characteristic IR
absorptions of MTBE [19]. Most likely the MTBE was adsorbed dur-
ing the reaction and is evolved by means of TPD.
One representative spectrum (taken from the dark gray region
in Fig. 10) is depicted in Fig. 12. It shows that during the TPD experi-
ment gases other than CO, CO2, H2O and MTBE are released as well.
The gray highlighted bands can be assigned to the known C8 by-
products 2,4,4-trimethylpentene-1 and 2,4,4-trimethylpentene-2
which are formed by the dimerization of isobutene. Their charac-
teristic IR absorptions are specified in Table 5.
The sample spectrum shows both the bands at 2970–2880 cm−1
and the peaks at 1465 and 1385 cm−1. The characteristic band for
the 1-en isomer at 3080 cm−1
( C–H stretching mode) occurs in the
0,30
0,25
0,20
0,15
0,10
0,05
0,00
2,5
SAC
before reaction
2,0
1,5
1,0
0,5
0,0
SAC60
300
400
500
600
600
700
2,5
2,0
1,5
1,0
0,5
0,0
SACar
SAC60ar
after reaction
TReaction
300
400
500
700
4000
3500
3000
2500
2000
1500
1000
Temperature / K
Wavenumber / cm-1
Fig. 13. Quantity of CO2 evolved during TPD of the samples before (top) and after
Fig. 11. IR spectrum of the gas phase during TPD of SAC60ar (black region in Fig. 7).
(below) use as catalyst.

110
A. Nau et al. / Applied Catalysis A: General 397 (2011) 103–111
Table 5
IR absorptions of the C8 by-products 2,4,4-trimethylpentene-1 and 2,4,4-trimethylpentene-2 in the region 4000–1000 cm−1 [19] and wavenumbers of observed absorptions.
Wavenumber of 2,4,4-trimethylpentene-1 (cm−1
)
Wavenumber of 2,4,4-trimethylpentene-2 (cm−1
)
Observed wavenumber (cm−1
)
3080
–
3076
2970, 2940, 2880
1800 (very weak)
1650
1465, 1385
1200
2970, 2940, 2880
–
1650 (very weak)
1465, 1385
1200
2977, 2934, 2883
Interfered with water
Interfered with water
1453, 1387
–
1080
500–600 K could be explained as mentioned above (see CO2
release).
2,0
before reaction
SAC
1,5
1,0
0,5
0,0
SAC60
SAC shows an increasing CO evolution from 600 K, too, but with
much lower amounts of CO than SAC60. Interestingly, the untreated
carbon SAC evolves after the application as catalyst more CO and
CO2 at 675 K than before. The amount of CO approximately equals
the amount of CO2 (see Fig. 13). The only oxygen surface group
known to us that decomposes in this range of temperature to equal
amounts of CO and CO2 is the anhydride group. However, no expla-
nation how and why anhydrides should be formed on the untreated
carbon through the reaction can be offered here.
One can also conjecture that the CO2 and CO originate from the
cleavage of adsorbed MTBE on the activated carbon at high temper-
atures via the reaction pathway, wherein the produced methanol is
dehydrogenatedso thatformaldehydeandotherby-products asCO,
CO2, and H2O are formed. Hereby, the increase in CO and CO2 des-
orption at temperatures above 600 K for SACar could be explained,
as well. It is supposable that trace amounts of methanol, which
could not be detected with the gas chromatograph during reaction,
are also formed using SAC.
300
400
500
600
600
700
2,0
1,5
1,0
0,5
0,0
SACar
SAC60ar
after reaction
TReaction
300
400
500
700
Temperature / K
Fig. 14. Quantity of CO evolved during TPD of the samples before (top) and after
(below) application as catalyst.
Before using SAC60 as catalyst for the MTBE cleavage the sam-
ple evolves H2O with maxima at 330 and 430 K. Water liberation
decreases to nearly zero above 600 K (Fig. 15). The untreated carbon
SAC does not release any water at all. After the reaction both sam-
ples only show a weak peak at approximately 350 K. In contrast to
the samples before the reaction, at temperatures above 550 K water
is evolved from the samples after the use as catalyst. This water has
to originate from the performed reaction. Similar to the CO and CO2
profiles, the water desorbing above 600 K is also an indication for
MTBE adsorbed on the probes SACar and SAC60ar, which is decom-
posed among other things into the by-products CO, CO2 and water.
As 40–100 ppm water is introduced by the feed, the detected water
can also originate from the reactant.
between SAC60 and SAC60ar at temperatures of 500–600 K could
be explained by the esterification with methanol or hydrolysis of
anhydrides due to water carried in the feed as mentioned before.
These anhydride groups may have formed carboxylic acids which
already decomposed during the reaction.
Before performing the MTBE cleavage reaction SAC60 liber-
ates CO with a maximum at 450 K and a further increase at
600 K (Fig. 14). According to literature, CO evolved at 450 K can
be ascribed to the decomposition of instable ˛-substituted car-
bonyl groups whereas more stable carbonyl groups decompose
above 650 K [20,21]. The explanation of the differences in the CO-
profile of SAC60ar compared to SAC60 at temperatures around
The characterization of the activated carbon material identifies
the active sites for MTBE cleavage and the reason for the decrease
in conversion with time. As the reaction temperature is 498 K and
the decarboxylation starts at 350 K (see Fig. 9), the carboxylic acids
seem to be the main active sites for MTBE cleavage. Carboxylic
acids are decomposed with time-on-stream and the conversion
decreases (see Fig. 6).
8
SAC
before reaction
6
4
2
0
SAC60
4. Conclusions
300
400
SACar
500
600
600
700
8
6
4
2
0
after reaction
An oxidized spherical activated carbon was applied as catalyst
for MTBE cleavage in the vapor phase. The results show that MTBE
was cracked with high selectivity into methanol and isobutene. The
catalytic activity of the oxidized activated carbon can be traced
back to the surface groups generated by ozonization of the car-
bon because the untreated sample did not show any conversion
of MTBE at all. In contrast to commonly used amorphous alu-
mosilicates, the modified carbon inhibited the production of the
by-product dimethylether. Simulation of the ex situ DRIFT spec-
tra of the material revealed that IR absorption bands associated
with carboxylic acids and anhydrides decreased noticeably after
TReaction
SAC60ar
300
400
500
700
Temperature / K
Fig. 15. Intensity of IR absorption of water released during TPD. Top: activated
carbon samples before the use as catalyst. Below: samples after the use as catalyst.

A. Nau et al. / Applied Catalysis A: General 397 (2011) 103–111
111
the use as catalyst while lactone and carbonyl groups stayed con-
References
stant. That means that carboxylic acid and anhydride groups are
responsible for the activity of SAC60. Furthermore, it can be con-
cluded that none of the present surface groups (carboxylic acids,
anhydrides, lactones and carbonyl groups) cause the dehydration of
methanol into dimethylether. That means that MTBE cleavage can
only occur on Broensted-acid sites. Furthermore, it is known, that
no Lewis-acid sites are present on SAC60. Therefore, it is assumed,
that the DME formation most probably occurs on Lewis-acidic sites.
Amorphous alumina–silica has both types of acidic sites and hence
DME formation cannot be completely inhibited by the application
of those materials.
Unfortunately, the ozonized activated carbon sample shows a
rapid deactivation behavior under the applied reaction conditions.
Nevertheless, the results may help to improve commonly used,
long-term stable catalysts by the characterization of the active
groups on the carbon catalyst which are necessary for MTBE cleav-
age but do not form dimethylether.
[1] D. Takashi, T. Toru, EU Pat. 0068 785 (1982).
[2] B. Juguin, J. Miquel, M. Hellin, B. Torck, US Pat. 4 395 580 (1983).
[3] R. Trotta, I. Miracca, Catal. Today 34 (1997) 447–455.
[4] J. Tejero, F. Cunill, S. Manzano, Appl. Catal. A 38 (1988) 327–340.
[5] G.M. Maksimov, I.V. Kozhevnikov, React. Kinet. Catal. Lett. 39 (1988) 317–322.
[6] F. Cunill, J. Tejero, J.F. Izquierdo, Appl. Catal. A 34 (1987) 341–351.
[7] D. Takashi, T. Tokumaru, EU 0068 785 A1 (1982).
[8] K. Kida, T. Aoyama, T. Uchiyama, US Pat. 4 521 638 (1985).
[9] S. Beran, J. Bolcha, K. Nedomova, B. Wichterolova, React. Kinet. Catal. Lett. 26
(1984) 437–439.
[10] P.B. Meunier, P. Chaumette, Oil Gas Sci. Technol. 46 (1991) 361–387.
[11] L.R. Radovic, F. Rodriguez-Reinoso, Chemistry and Physics of Carbon: A Series
of Advances, Carbon Materials in Catalysis, vol. 25, 1997, pp. 243–358.
[12] Ullmann‘s Encyclopedia of Industrial Chemistry, Activated Carbon, seventh ed.,
Wiley, New York, 2010.
[13] S. Houbrechts, D. Maschmeyer, H.-W. Zanthoff, W. Büschken, M. Winterberg, F.
Nierlich, S. Santiago Fernandez, W. Luh, DE Patent, no. 102006040434A1 (2008).
[14] K. Krauß, M. Fehlings, A. Drochner, H. Vogel, Chem. Eng. Technol. 23 (2000)
319–322.
[15] Q.L. Zhuang, T. Kyotani, A. Tomita, Energy Fuels 8 (1994) 714–718.
[16] M. Hesse, H. Maier, B. Zeeh, Spektroskopische Methoden in der organischen
Chemie, 5, Auflage, Thieme Verlag Stuttgart, 1995.
[17] P.E. Fanning, M.A. Vannice, Carbon 31 (1993) 721–730.
[18] U. Zielke, K.J. Hüttinger, W.P. Hoffman, Carbon 34 (1996) 983–998.
[19] Ansyco Analytische Systeme und Componenten GmbH, Karlsruhe (2010).
[20] J. Surygala, R. Wandas, E. Sliwka, Fuel 72 (1993) 409–411.
[21] C. Moreno-Castilla, F. Carrasco-Marin, A. Mueden, Carbon 35 (1997)
1619–1626.
Acknowledgement
We acknowledge Dr. Christian Böing (Evonik Oxeno GmbH) for
supporting this project.