Kinetics of Hydro-isomerization of n-Hexane over Platinum Containing Zeolites

Article abstract of DOI:10.1006/jcat.1997.1779

Kinetic hydro-isomerization experiments at atmospheric pressure using n-hexane as reactant were performed. Four zeolites with different pore structures but intrinsically equal acid strength (mordenite, ZSM-5, ZSM-22, and β) were used. Adsorption and diffusion effects were found to mainly determine the differences in activity of these zeolites. The general observation is that the higher the adsorption enthalpy of n-hexane, the higher the activity per acid site and the lower the activation energy. The latter effect is not very pronounced when the order of reaction in n-hexane is low. The true activation energy of isomerization with respect to the adsorbed alkoxy state is found to be approximately 125 kJ/mol. ZSM-22 and mordenite were both relatively inactive compared to the other zeolites. In the case of ZSM-22 this was due to pore-mouth catalysis resulting in a 5 to 20% usage of the acid sites. The relatively low activity of mordenite occurs since two-thirds of the acid sites, those located in the side-pockets, are not accessible to n-hexane.

Full text of DOI:10.1006/jcat.1997.1779

JOURNAL OF CATALYSIS 171, 77–84 (1997)
ARTICLE NO. CA971779
Kinetics of Hydro-isomerization of n-Hexane over Platinum
Containing Zeolites
A. van de Runstraat, J. A. Kamp, P. J. Stobbelaar, J. van Grondelle, S. Krijnen, and R. A. van Santen
Department of Inorganic Chemistry and Catalysis, Faculty of Chemical Engineering, Eindhoven University of Technology,
P.O. Box 513, 5600 MB Eindhoven, The Netherlands
Received November 27, 1996; revised March 1, 1997; accepted May 20, 1997
The hydro-isomerization mechanism can be divided in
Kinetic hydro-isomerization experiments at atmospheric pres- (de)hydrogenation, protonation and isomerization steps.
sure using n-hexane as reactant were performed. Four zeolites with This bifunctional mechanism is given schematically in Fig. 1
different pore structures but intrinsically equal acid strength (mor-
(
1). These reactions take place in the adsorbed state. Ac-
denite, ZSM-5, ZSM-22, and �)wereused. Adsorption and diffusion
effects were found to mainly determine the differences in activity
of these zeolites. The general observation is that the higher the ad-
sorption enthalpy of n-hexane, the higher the activity per acid site
and the lower the activation energy. The latter effect is not very
pronounced when the order of reaction in n-hexane is low. The
true activation energy of isomerization with respect to the adsorbed
alkoxy state is found to be approximately 125 kJ/mol. ZSM-22 and
mordenite were both relatively inactive compared to the other ze-
cording to Kazansky, adsorption of an alkene to an acid
site of a zeolite gives an alkoxy species (2, 3). This species is
characterized by a covalent bond between an oxygen atom
from the zeolite lattice and a carbon atom of the alkene. The
elementary isomerization reaction step involves in this case
a carbenium ion-like transition state; the n-alkoxy species
are the stable intermediates.
After a short summary of current thinking in hydro-
olites. In the case of ZSM-22 this was due to pore-mouth catalysis isomerization kinetics, details of the experiments and their
resulting in a 5 to 20% usage of the acid sites. The relatively low results will be provided. Under Discussion we will evaluate
activity of mordenite occurs since two-thirds of the acid sites, those
the results of optimizing the platinum content and the re-
located in the side-pockets, are not accessible to n-hexane.
�c 1997
sulting “key” samples. The relation between performance
of the zeolites and their adsorption behavior will be high-
lighted.
Academic Press
2
. METHODS
1
. INTRODUCTION
A Summary of Hydro-isomerization Kinetics
In this paper we will report the results of a study of the
catalytic activity of acidic zeolites. It will be determined
When it is assumed that the isomerization reaction step
whether differences are solely due to differences in intrinsic is rate determining (ideal bifunctional behavior) and the
zeolite proton properties or differences in adsorption char- hydrocarbon concentrations inside the zeolite are in equi-
acteristics. We used the hydro-isomerization of n-hexane librium with the gas phase, the overall rate of isomerization
as a model reaction since its low deactivation rate enabled is given by (4)
study of the reaction under true steady-state conditions.
Catalysts had to be chosen such that the platinum function
was not rate determining. Hexane was chosen as reactant
because of its low cracking rate. The rate of isomerization
could therefore be approximated by the rate of n-hexane
conversion.
�
�
pnC6
Kdehydr � Kprot �
p
H
2
R = kiso
�
�
p
nC
6
1
+ Kdehydr � Kprot �
pH
2
�
�
��
�
pnC
6
�
kiso Kdehydr � Kprot �
,
[1]
pH2
Four zeolites with a Si/Al ratio larger than 10 were cho-
sen. The catalysts had the same intrinsic acid strength, as in which R is the rate of reaction, Kdehydr is the equilib-
1
determined by H NMR and IR spectroscopy. Therefore, rium constant of dehydrogenation, Kprot is the equilibrium
comparison of their activities normalized per acid site was constant of protonation, kiso is the rate constant of isomer-
useful. From the data obtained on the optimized (“key”) ization, pnC is the partial pressure of n-hexane, and pH is
6
2
samples an intrinsic activation energy of isomerization in the partial pressure of hydrogen. The order of reaction � is
zeolites could be deduced. equal to or smaller than 1.
77
0
021-9517/97 $25.00
Copyright �c 1997 by Academic Press
All rights of reproduction in any form reserved.

7
8
VAN DE RUNSTRAAT ET AL.
TABLE 2
Pore Structures of Zeolites Chosen
One-dimensional Three-dimensional
Medium pore
Large pore
ZSM-22
Mordenite
ZSM-5
�-Zeolite
FIG. 1. Schematic representation of Weisz’ bifunctional mechanism.
+
+
6
n-C6, n-hexane; n-O6, n-hexene; n-C , n-carbenium ion; i-C , iso-
6
carbenium ion; i-O6, iso-hexene; i-C6, iso-hexane.
and 1Hads, n=C is the adsorption enthalpy of n-hexene. In
6
this equation, the protonation enthalpy of hexene from the
gas phase (actually, n-alkoxy formation) is expressed as
When the zeolite micropore is almost completely cov-
ered with reactant, at high pressure or when a long alkane
is used, Eq. [1] is no longer valid (5, 6). Then an expres-
sion as proposed by Steijns and Froment applies (5). This
expression implies no equilibrium between hydrocarbons
in the micropores and gas phase. The high micropore filling
causes severe diffusion inhibition. The general form of this
expression is
the sum of 1Hprot, ads and 1Hads, n=C . It is assumed that the
6
adsorption enthalpies of n-hexene and n-hexane are equal,
which is a reasonable assumption (9, 10).
Substitution of Eq. [3] into Eq. [1] allows for the sub-
stitution of (1 � �n-alkoxy) by �. Equation [3] will be used
to determine the true activation energy of isomerization
from experimental data. The values used for 1Hdehydr and
1Hprot, ads are respectively � 114.8 and 80 kJ/mol. The first
value has been obtained from thermodynamics (11). The
latter value is a computed value (12).
PAR(1) � pn alkane
-
R = kiso
,
pH + PAR(2) � pH � pn alkane + PAR(3) � pn alkane
2
2
-
-
[2]
Catalyst Treatment
in which PAR(1), PAR(2), and PAR(3) are parameterscon-
sisting of equilibrium and adsorption constants. The term
PAR(3) � pn-alkane represents Langmuir adsorption of prod-
uct alkanes. This equation has been confirmed for n-decane
isomerization over a Pt/USY catalyst (6).
Expected orders at different conditions can be found in
Table 1. We have developed a kinetic modeling scheme that
reproduces the two extreme situations and enables simula-
tions in the intermediate regime (7). When one takes the
derivative of the natural logarithm of Eq. [1] toward the
reciprocal of the temperature, one obtains (8)
Four zeolites were used that were representative of
four prototype pore structures (Table 2). The Na�-zeolite,
NaZSM-5, and NaZSM-22 samples were kindly donated
by Exxon Chemicals in Machelen, Belgium. The Na-
mordenite was kindly donated by Shell Research and Tech-
nology Centre in Amsterdam, The Netherlands.
Conversion to the protonated forms was carried out
through ammonium nitrate exchange and subsequent de-
composition in nitrogen at 783 K and calcination at 773 K.
Platinum loading of the zeolites was then performed by ion
exchange with a platinum tetra-ammonium hydroxide so-
lution. The filtrate was checked for platinum by UV–VIS
[3] to establish the true loading of the zeolite (13). The dried
sample was then treated in one of the following ways:
�
Eact, iso = Eact, app + (1 � �n alkoxy) � � 1Hprot, ads
-
�
�
1Hads, n=C + 1Hdehydr ,
6
in which Eact, app is the apparent activation energy, Eact, iso is
the activation energy of n-alkoxy isomerization, 1Hdehydr is
the enthalpy of dehydrogenation, 1Hprot, ads is the protona-
tion enthalpy (alkoxy formation) from the adsorbed state,
(
a1) Calcination at 723 K, reduction at 673 K and passi-
vation at room temperature. Such a sample is referred to as
an ex situ pretreated sample.
(
a2) The pure ex situ pretreated sample was embedded
in a silica matrix (Aerosil Degussa 380) by making a gel in
which the zeolite particles were dispersed. Such a sample is
referred to as an embedded sample.
TABLE 1
Expected Orders of Reaction under Different Conditions
(
b) In case of a sample that was pretreated in situ, calci-
Equation [1]
Equation [2] nation and reduction were performed inside the reactor.
pn-alkane
pH2
n
m
n
m
Sieve fractions of 125–500 �m were used in the reactor.
�
High
High
Low
Low
High
Low
High
Low
0
0
1
� 1
0
� 1
Invalid
Catalyst Characterization
�
� �
� 1
Acidity. The most important characteristics of the zeo-
lites in the acidic form are given in Table 3. Two types of IR
spectra were recorded as published by J a¨ nchen et al. (14).
1
Note. n, order in n-alkane; m, order in hydrogen.

HYDRO-ISOMERIZATION OF n-HEXANE OVER ZEOLITES
79
TABLE 3
when the points deviate at a conversion lower than approx-
imately 80% (19).
Results of Characterization of the Acidic Zeolites Used
The Kinetic Experiments
Analysis
HMOR
HBEA
HTON
HMFI
Equipment. n-Hexane (Acros 99+% purity), delivered
by a HP1050 HPLC pump, was evaporated into flowing hy-
drogen or a hydrogen/nitrogen mixture. The gas flows were
�
1
IR: �(OH) [cm ]:
1
3608
1069
4.1
3609
1078
4.3, 5.1
� 0.9, 53.4
17.0
3600
1072
4.4
53.4
44.7
81.5
3611
1073
4.2
�
1
� CD3CN [cm ]:
1
H NMR, � [ppm]:
2
7
Al NMR, � [ppm]:
� 1.0, 54.5
� 1.0, 55.2 controlled by thermal mass-flow controllers. The mixture
Si/Alframework ratioa
12.5
40.7
82
was then flowed through a fixed-bed, continuous flow reac-
�
1Hads [kJ/mol]
71.9
70
tor at atmospheric pressure. Quartz reactors with an inter-
a
Determined by 29Si MAS NMR.
nal diameter of 4 and 8 mm were used. Reaction products
were analyzed on-line using a HP5890 Series II gas chro-
matograph equipped with a FID detector and a Chrompack
Both the frequency of the OH stretch vibration and its
shift upon acetonitrile adsorption are a measure of intrinsic
Brønsted acid strength. Since the shift always results in two
broad bands, the shift (in cm ) based on the weighted av-
erage of these new bands is given. The higher the position
of the OH peak and the smaller the shift upon acetonitrile
adsorption, the lower the acid strength per proton.
fused silica column with a Al2O3/KCl coating.
Conditions. Standard conditions were 513 K, 0.03 ml/
min liquid flow of n-hexane, and a 145 Nml/min flow of hy-
drogen. The WHSV was tuned by using an amount of cata-
lyst (between 50 and 200 mg) which yielded a conversion
below 10% under these standard conditions. A hydrogen/
n-hexane molar ratio higher than 20 was used to minimize
deactivation by coking.
�
1
1
Since the H chemical shift for the acid sites and the OH
stretch vibrations as well as the shift upon acetonitrile ad-
sorption were the same for all zeolites used it may be con-
cluded that they all had the same acid strength.
Catalyst pretreatment. Each catalyst was subjected to
an aging period. After a pretreatment at 723 K (mordenite
and �) or 523 K (ZSM-5 and ZSM-22) in flowing hydrogen
the catalyst was submitted to 20 h of standard reaction con-
ditions. During this treatment a deactivation period and
a slight increase in the 2-MP/3-MP ratios were observed.
All catalysts lost approximately 25% of their initial activ-
ity. In the case of �-zeolite, a strong increase in selectivity
for dimethylbutanes and a decrease in cracking products
were observed. This indicates that deactivation of the metal
function is the main reason for the initial deactivation. Af-
ter stabilization, the catalysts could be used for up to 1000 h
Crystal size and pore volume. The crystal size was de-
termined by SEM. The mordenite and �-zeolite samples
consisted of agglomerated crystals. The mordenite agglom-
erates were up to 20 �m in size and the zeolite � agglom-
erates up to 200 �m. We learned an important feature of
mordenite by comparing the BET pore volume (0.2 ml/g)
and n-hexane adsorption capacity (0.07 ml/g). These fig-
ures may indicate that only one-third of the sites inside the
mordenite pores can be reached by n-hexane.
Calorimetry. The 1Hads values given in Table 3 are the on stream without further loss of activity.
calorimetrically measured heats of adsorption of n-hexane.
Between measurements the catalyst was reactivated by
The uncertainty of these values is less than 1 kJ/mol. Gen- an in situ hydrogen pretreatment. This resulted temporarily
erally, a smaller pore size led to higher adsorption enthalpy. in an approximately 10% higher activity, which was leveled
The heat of adsorption of �-zeolite is relatively low. An ex-
trapolation of Configurational-Bias Monte Carlo calcula-
tions on all-silica �-zeolite confirmed the value of approx-
imately 70 kJ/mol (15), which is probably due to the low
density of � (16, 17). The value of ZSM-22 was not mea-
sured here but was taken from results of Eder (18).
TABLE 4
Characterization of the Platinum Function of the ex Situ
Pretreated Samples
Liquid phase hydrogenationa
Chemisorption
�
Sample
qTOFPt [min 1]
Xdev [% ]
H/Pt
Platinum function. In addition to hydrogen chemisorp-
tion, we used liquid phase hydrogenation experiments of
0
1
.5 wt% Pt/HMOR
wt% Pt/HMOR
3.8
3.4
3.8
2.3
75
55
40
80
1.4
2.0
2.0
1.0
1
-hexene to characterize the platinum particles. The results
of both methods obtained on some representative samples 2 wt% Pt/HMOR
are given in Table 4. A quasi turnover frequency (qTOFPt)
3
wt% Pt/HMOR
is calculated by taking into account all platinum atoms 1.5 wt% Pt/HBEA
19.8
19.8
1.8
55
60
85
1.6
1.6
2.0
0
0
.5 wt% Pt/HMFI
.7 wt% Pt/HTON
present. Xdev stands for the conversion level of deviation
from the straight line of conversion versus time. This level
is a measure of mass-transfer limitations within the zeo-
a
Solvent n-heptane, atmospheric hydrogen pressure, room tempera-
lite. Mass-transfer limitations of 1-hexene play a larger role ture.

8
0
VAN DE RUNSTRAAT ET AL.
out, depending on the zeolite type, in a 4- to 10-h period. A
partial cleaning of the platinum and acid sites was assumed
to be responsible for the high initial activity.
Definition of activity. We defined a turnover frequency
(
TOF) as the number of moles reactant converted per mole
of Brønsted acid site per hour. It appeared, however, from
our characterization experiments that only one-third of the
acid sites in mordenite can be reached by n-hexane (20).
It might therefore be argued whether in this case the true
TOF must be calculated using only one-third of the sites.
FIG. 3. (a) Selectivities and (b) 2-MP/3-MP ratios on mordenites and
two � samples versus platinum to acid site ratio. (� ) Hydrocracking (mor-
denite). (᭜) DMBs (mordenite). (᭝) DMBs (�). (᭹) 2-MP/3-MP (mor-
These values are given in parentheses where needed. On denite). (᭺) 2-MP/3-MP (�).
average, the experimental error amounts to 10% .
In Fig. 3, the hydrocracking selectivity (Fig. 3a) and
the 2-methylpentane/3-methylpentane (2-MP/3-MP) ratio
(Fig. 3b) are plotted versus the platinum to acid site ra-
tio. The amount of 2,2-dimethylbutane (22-DMB) was very
low and difficult to detect (its retention time was almost
equal to that of methylcyclopentane, which was a contami-
nant in the feed n-hexane). Therefore, the dimethylbutanes
ratio could not always be used as a means to characterize
the platinum function. The sum of these products was a
more reliable source of information and is also plotted in
Fig. 3a. Table 5 lists the kinetic data.
3
. RESULTS
Diffusion Limitations
Experiments using both 0.5 and 2.0 wt% Pt/Hmordenite
catalysts showed that external diffusion limitations could
be excluded when a total flow of 150 Nml/min was used. In-
ternal diffusion limitations were shown to be irrelevant in
experiments using a set of three embedded ZSM-5 samples
with crystal sizes ranging from 0.6 to 12 �m. From this we
could estimate a lower limit for the effective diffusion coef-
�
14
2
ficient of 8 � 10 m /s at 513 K (21). This is approximately
the same value as obtained by Haag et al. at 538 K (22). The
same type of experiment was performed on two ZSM-22
catalysts with crystal lengths of 4 (TON(4)) and � 1�m
Selectivities
The overall selectivities towards isomerization were usu-
ally very high. Only up to 6 mol% cracking products were
found at conversions below 10% . ZSM-5 and the 0.5 wt%
(
TON(1)). Here a relative efficiency of 0.28 was found for
the larger crystals. This means that, at least for TON(4)
crystals, internal diffusion limitations do play a role.
�
even showed an isomerization selectivity of over 99% .
However, the isomerization selectivity of the ZSM-22 cata-
lysts was only 80% . In Table 6, the selectivities to the four
hexane isomers as a function of zeolite are given at stan-
Determination of the Optimum Platinum Loading
Sufficient platinum had to be present to establish the hex-
ane/hexene equilibrium and make the reaction on the acid
sites rate determining. The optimum platinum loading was
determined by using both ex situ pretreated and embedded
mordenites as well as two � samples. The relative activities
of these catalysts as a function of platinum to acid site ratio
are given in Fig. 2.
TABLE 5
Kinetic Data as a Function of Platinum Loading
Eact, app
[kJ/mol]
Eact, iso
[kJ/mol]
Zeolite
na
mb
1
1
2
3
1
1
2
2
3
1
1
/2% MOR, embedded
% MOR, embedded
0.42
0.32
0.24
0.31
0.34
0.24
0.14
0.13
0.19
0.53
� 0.2
� 0
106
101
111
114
96
122
113
120
126
109
104
112
111
122
135
c
% MOR, embedded
% MOR, embedded
� 0
c
� 0.23
� 0
/2% MOR, in situ
% MOR, in situ
% MOR, in situ
% MOR, ex situ
% MOR, ex situ
/2% BEA, ex situ
.6% BEA, ex situ
0.1
95
� 0
107
106
115
116
� 0
� 0.25
� 0.24
d
e
d
e
d
e
0.5 (0.69 )
� 0.29 113 (99 ) 131 (123 )
a
1
6
.95 kPa < pnC6 < 4.29 kPa, pH2 = 85.7 kPa.
b
c
d
e
8.7 kPa < pH2 < 98.0 kPa, pnC6 = 3.43 kPa.
FIG. 2. Relative activity of mordenite catalysts and two � samples as
a function of platinum to acid site ratio. (᭜) Embedded Mordenite. (᭹)
Ex situ pretreated mordenite. (� ) Ex situ pretreated �-zeolite.
Pretreated at 523 K.
493 K < T < 506 K.
493 K < T < 533 K.

HYDRO-ISOMERIZATION OF n-HEXANE OVER ZEOLITES
TABLE 6
81
Product Selectivities (mol%) for the “Key” Zeolites
a
at Standard Conditions
Selectivity for
MOR BEA TON(1) TON(4) MFI (<1 �m)
2
3
2
2
2
2
-MP
-MP
-MP/3-MP
3-DMB
2-DMB
50.5
33.3
1.5
10.5
1.7
53.4
34.2
1.6
3.7
2.5
55.6
19.5
2.9
0
0
n.a.
45.4
32.6
2.7
0
0
n.a.
73.9
22.2
3.4
0.3
0
2-DMB/23-DMB
0.2
0.7
n.a.
Note. n.a., not applicable.
T = 513 K, 0.03 ml/min liquid n-C6, 145 Nml/min H2, conversion � 5% .
FIG. 4. Arrheniusplot of BEA (1.6 wt% Pt/H�). (᭹) Experimental
points, Eact, low T = 133.3 kJ/mol, Eact, high T = 67.3 kJ/mol.
a
dard conditions. The 2-MP/3-MP and 2,2-dimethylbutane/
,3-dimethylbutane (22-DMB/23-DMB) ratios are also
listed. These ratios are 1.5 and 2.5 at equilibrium respec-
tively. The 2-MP/3-MP ratio increased from 3.4 to 3.6 to 3.8
with increasing MFI crystal diameter. On the TON samples
the opposite trend was observed.
4. DISCUSSION
2
Determination of the Optimum Platinum Loading
The activities of the ex situ pretreated and embedded
samples were comparable. The in situ pretreated samples
were less active, which is probably due to the lower disper-
sion of platinum resulting in rate limitation by the metal
function. Since in the latter case the zeolites were pressed
into agglomerates before destruction of the platinum com-
plex, the mass transport of the desorbing ammonia will be
hindered. This has been shown to lead to low dispersions
Kinetics of the “Key” Samples
+
All four “key” zeolites had a Pt/H ratio of 0.067. These
samples will now be indicated by their IZA three-letter
codes (23). Their activities per acid site at standard condi-
tions and the corresponding WHSV as well as the kinetic
data are listed in Table 7. All orders of reaction were found
to be relatively independent of temperature.
(
13, 24).
Mordenite. In Table 5, the orders of reaction in n-hex-
ane for different mordenites are listed. Since these orders
are higher on the embedded type catalysts, these morden-
ites have a smaller active site coverage. On both types a
decrease in n-hexane orders with increasing loading and
a hydrogen order of approximately 0 can be observed up
to a platinum content of 2 wt% . The former observation
implied that the zeolite coverage and, therefore, the usage
of the available acid sites increased. The measured orders
agree with Eq. [2] in the case of low hydrogen pressure and
high hexane pressure. We found a delicate balance between
the rate of the isomerization and deactivation effects. De-
tails of catalyst preparation and pretreatment determine
whether the order in hydrogen will be positive or negative.
The solid line (embedded Pt/mordenite) in Fig. 2 shows
that the metal function is no longer rate determining at
The Arrhenius plot measured on BEA was curved
Fig. 4). In the lower-temperature range (493 to 506 K)
the activation energy was 113 kJ/mol, while in the higher-
temperature range (521 to 533 K) a value of 67 kJ/mol was
obtained. Table 7 contains the low-temperature data.
(
TABLE 7
a
Activities under Standard Conditions and Kinetic Data
of the “Key” Zeolites
WHSV
TOF
Eact, app
[kJ/mol] [kJ/mol]
Eact, iso
Sample
[g/gh] [mol/molh] nb
mc
d
MFI, ex situ
TON(1), ex situ 13.5
TON(4), ex situ 10.0
12.6
54
7.9
2.2
0.51 � 0.25
103
103
86
127
136
133
d
d
+
0.7
1.0
� 0
� 0
a Pt/H ratio of 0.067. This conclusion is supported by the
observed suppression of secondary acid-catalyzed reactions
leading to a more bifunctional behavior of the catalyst (25).
Moreover, the 2-MP/3-MP ratio decreased with increasing
MOR,
embedded
BEA, ex situ
8.1
9.7 (29.1e) 0.24 � 0
111d
120
25.9
27.8
0.5f
� 0.29
113f
131
+
metal loading and reached its equilibrium value at a Pt/H
a
T = 513 K, 0.03 ml/min liquid n-C6, 145 Nml/min H2, conversion � 5% . ratio of 0.067.
b
c
d
e
1
.95 kPa < pnC < 4.29 kPa, pH2 = 85.7 kPa.
6
The dashed line shows a different trend because the
3 wt% sample showed a strong increase in hydro-isomer-
ization activity. We attribute this to local destruction of the
mordenite framework caused by growth of large platinum
particles (26, 27). The presence of large metal particles was
6
4
8.7 kPa < pH2 < 98.0 kPa, pnC6 = 3.43 kPa.
93 K < T < 533 K.
This takes into account that only one-third of the acid sites can be
reached by n-C6.
f
4
93 K < T < 506 K.

8
2
VAN DE RUNSTRAAT ET AL.
confirmed by a relatively low H/Pt and a low qTOFPt in the apparent activation energy is similar to the one obtained
liquid phase hydrogenation (Table 4). Framework destruc- on the BEA sample in the lower temperature range (be-
tion facilitates molecular diffusion or increases the number low 518 K). This strongly indicates that the reaction on the
of reachable acid sites. The former explanation also means higher loaded sample is diffusion limited in the higher tem-
that the reaction on the lower loaded samples might be lim- perature range. Since this catalyst is the most active of all
ited by single-file diffusion problems. The decrease of dif- (in mol/g � h), this is a real possibility. Falsification of the or-
fusion limitations is evidenced by the value of Xdev in the der in the reactant is a consequence of diffusion limitations
liquid phase hydrogenation. This value decreased with in- (31):
ntrue � 1 d ln �
creasing metal loading; the 3 wt% sample showed a higher
value.
The orders of reaction also demonstrated that the 3 wt%
nobserved = ntrue
+
[4]
2
d ln �
Since d ln �ꢀd ln � has a value between 0 and � 1, and order
will be observed that is higher than the true order. Mea-
surements indeed showed a higher order at higher loading.
The 2-MP/3-MP ratio selectivity decreased at increasing
platinum loading. An opposite trend was observed for the
sum of the dimethylbutanes which was mainly due to an in-
crease in 22-DMB. According to Blomsma et al., 22-DMB
cannot be formed by a dimerization and cracking mecha-
nism and can, therefore, be used to monitor monomolecular
activity versus bimolecular activity on �-zeolite (32). Our
observation is in favor of an increase of the former mecha-
nism. The cracking activity increased from a 0.5 to a 1.6 wt%
Pt/�. This was mainly due to an increase in acid cracking.
This is possibly due to the high residence times caused by
diffusion limitations.
sample was a special case. Opposite to the trend observed
earlier, the n-hexane order is higher than that of the 2 wt%
sample. Since the hydrogen and n-hexane orders are now
almost equal, except for the sign, the catalyst shows ideal
bifunctional behavior (Eq. [1]) with equilibration between
micropore and gas phase. We thus see a change from one
extreme at lower loading to another extreme at higher
loading.
The true or intrinsic activation energies of isomerization,
calculated in Table 5, range from 100 to 125 kJ/mol. When
a catalyst does not conform to Eq. [1], the isomerization
may not be the (only) rate determining step and Eq. [3]
cannot be used. This is the case with the 1 wt% in situ pre-
treated catalyst because of its positive order in hydrogen.
In all other cases there seems to be a trend that a low appar-
ent activation energy is “compensated” by a high order in The “Key” Zeolites
n-hexane. This observation is in agreement with Eq. [3].
Activity. When adsorption effects play a decisive role in
Still, two sets of intrinsic isomerization activation ener-
gies of around 111 and 125 kJ/mol can be distinguished in
Table 5.
determining the activity, MOR and BEA should be equally
active. The same must be true for MFI and TON. BEA
is, however, three times more active than MOR (Table 7).
When one corrects for only one-third usage of the acid
sites in MOR, the BEA and MOR are equally active. MOR
might, however, experience single-file diffusion problems.
On the other hand, we discussed that the reaction on BEA
is probably diffusion limited at 513 K. The ex situ pretreated
Kazansky et al. calculated by quantum chemistry a value
of approximately 120 kJ/mol for the activation energy of
deprotonation (dissociation of the alkoxy species) to the
adsorbed state (28). We estimated the activation energy
of isomerization from the adsorbed state to be approxi-
mately 130 kJ/mol (29, 30) and we thus see the same dif-
ference in energies (10 kJ/mol) as observed in Table 5. The
lower value then represents the activation energy of depro-
tonation affected by limited hydrogenation activity and the
higher value the true intrinsic activation energy of isomer-
ization on a zeolite.
3
wt% Pt/H mordenite has an approximately 11 times
2
higher TOF than the MOR catalyst, which still cannot ac-
count for the factor of three difference in activity. The low
noncorrected TOF of MOR then indicates that both site
accessibility and diffusion play a role.
The TON(4) samples are very inactive compared to
�
-Zeolite. A check for the optimum platinum loading MFI. This is consistent with pore-mouth catalysis. This phe-
was also performed using �-zeolite. The TOF increased by nomenon was first proposed by Martens et al. on the basis of
+
a factor of 1.5, from 0.5 to 1.6 wt% Pt/H� (Pt/H ratio of the peculiar isomerization selectivity of this zeolite in the
0
.067). This effect was less pronounced than on the morden- reaction of n-decane and Molecular graphics calculations
ite catalyst, which would confirm importance of single-file (33). Pore-mouth catalysis means that only part of the sites
diffusion on the latter zeolite. is taking part in the reaction and is in fact an extreme case
The Arrhenius activation energy changed as a function of single-file diffusion limitation. Indeed the TON(1) crys-
of temperature for the BEA sample (Fig. 4). This might be tals are almost 4 times more active than the TON(4) crys-
explained by a change in coverage of reactive intermediates tals, which would be consistent. The activity of TON(1) is,
and thus the order of reaction in n-hexane. The assumption however, still not as high as would be expected from its ad-
of Arrhenius behavior is then no longer valid. The 0.5 wt% sorption enthalpy. Assuming that the activity per active site
Pt/H�, however, showed a straight Arrhenius plot. This of TON is equal to that of MFI, the sites that contribute to

HYDRO-ISOMERIZATION OF n-HEXANE OVER ZEOLITES
83
highly dispersed inside the zeolite crystals, as evidenced by
a high H/Pt (Table 4) and a low qTOFPt (due to diffusion
problems). The metal particles are, therefore, difficult to
reach by the reactant and an acid (cracking) mechanism is
observed: the main cracking product is propane (4). When,
however, the platinum is poorly dispersed very high iso-
merization selectivities can be obtained (34). Either expla-
nation can account for the lowered cracking on the smaller
TON crystals.
FIG. 5. Energy scheme of true and apparent activation energies (� �
). Eact, app, apparent activation energy; Eact, true, true activation energy;
0
1Hads, enthalpy of adsorption; R(g), gas phase reactant; R(ads), adsorbed
reactant; P(g), gas phase product; P(ads), adsorbed product.
Selectivity of the large pore zeolites. MOR and BEA
converted n-hexane to the MPs in their equilibrium ratio.
The selectivity toward 22-DMB was low since the formation
of this isomer involves a secondary carbenium ion, through
either isomerization of monomethylpentanes or a methyl
shift of 23-DMB (35). Formation of the latter involves a
tertiary carbenium ion. 22-DMB can even be formed by
direct transformation of n-hexane when the reactant is iso-
merized twice before it is hydrogenated on the platinum.
Thismaybe the case at lowplatinum loadingor zeoliteswith
a one-dimensional pore system, such as mordenite (36).
the activity can be estimated. Only 4% of the acid sites are
then active on the 4 �m crystals; for the 1-�m crystals this
is 15% .
1
2
Table 7 shows that the TOF of MFI is a factor of 5 (1 )
higher than for MOR and a factor of 11 higher than for
2
BEA. It may be concluded that a higher adsorption en-
thalpy leads to a higher activity, as long as one excludes
the TON data. This phenomenon might be explained as
is schematically shown in Fig. 5 for an almost empty zeo-
lite. The true activation energy of isomerization is the same
Orders of reaction. The two zeolites with a three-
for all zeolites. Equation [3] and Table 7 also show that, dimensional pore system, BEA and MFI, showed similar
for the same order of reaction, the apparent activation en- orders. The order of reaction in n-hexane of 0.69 measured
ergy for isomerization is lowest using the highest adsorption on BEA might be obtained under external diffusion limited
enthalpy (MFI). Kinetic computer simulations, published conditions. The fact that the absolute value of the order in
elsewhere, led to the same conclusion (7). The overall ef- hydrogen is lower than the order in n-hexane is not con-
fect will be small due to the low orders in the reactant.
sistent with either Eq. [1] or [2]. Since we found the same
phenomenon in our kinetic simulations that did not contain
deactivation, deactivation does not have to be invoked as
an explanation. We believe from the simulations that slow
hydrogenation of the isomer hexenes is a more probable
explanation (7).
Selectivity of the medium pore zeolites. The MFI and
TON catalysts showed product shape selectivity since the
DMBs were formed in very low quantities (if any). Also the
2
-MP/3-MP ratio was far from equilibrium on both catalysts
and might be explained in two ways:
The TON(4) sample appeared to be almost empty, since
1. 3-MP is just a little bit bulkier than 2-MP. The effective the order of reaction in n-hexane in this case was 1. Con-
size of 3-MP (4.4 � 5.8 A˚ ) is very close to that of a ZSM-5 sidering the high adsorption enthalpy we believe that the
pore (5.3 � 5.6 A˚ or 5.1 � 5.5 A˚ ) (22). From its structure zeolite was actually almost completely filled and most sites
one expects 2-MP to be smaller.
were blocked by reactant molecules. Hence only a few sites
2. Monobranched products can only adsorb with the lin- were responsible for the reaction, which supported the con-
ear part of the chain into the TON catalyst (33, 34). This clusion of pore-mouth catalysis. The order in n-hexane on
results in a higher adsorption enthalpy for 2-MP than for the TON(1) sample was lower, which indicated that more
3
-MP.
sites are used in this reaction. This enhanced usage of acid
sites was already calculated from the activity data. Using
Since the 2-MP/3-MP ratio increased on MFI and de-
�
n-alkoxy � 1 � n, coverages of 0 and 30% can be estimated.
creased on TON with increasing crystal size, two different
mechanisms must be operable to establish the relatively
high 2-MP/3-MP ratios. A size effect (explanation 1) is more
probable on MFI since it is increasingly difficult for the
bulkier product to diffuse out of a crystal at increasing crys-
From the activitydata valuesofrespectively4and 15% were
calculated. This is a reasonable agreement considering the
crudeness of estimation.
Activation energies. The measurements listed in Table 7
tal size. Explanation 2 is more pronounced on TON since showed that increasing adsorption enthalpy of a zeolite led
larger crystals possess fewer pore mouths to perform the to lower apparent activation energy and higher activity. We
reaction.
already concluded in the discussion for the optimum plat-
Another special feature of the TON-type catalysts is the inum content that the intrinsic activation energy on a zeo-
low isomerization selectivity. This might be due to pore- lite is approximately 125 kJ/mol. This value is confirmed by
mouth reaction or high residence times. The platinum is Table 7. The thus calculated true activation energy of

8
4
VAN DE RUNSTRAAT ET AL.
isomerization is much higher than the one in the superacid
solution (approximately 30 kJ/mol) (30). This is due to the
fact that in reactions catalyzed by acidic zeolites alkoxy
species are the reactive intermediates. The covalent C–O
bond between carbon chain and zeolite lattice must be (par-
tially) dissociated before isomerization.
8. Van Santen, R. A., and Niemantsverdriet, J. W., “Chemical Kinetics
and Catalysis,” p. 49. Plenum Press, New York, 1995.
9. Stach, H., Lohse, U., Thamm, H., and Schirmer, W., Zeolites 6, 74
(
1986).
1
1
0. Lerchert, H., and Schweitzer, W., in “Proceedings of the International
Conference on Zeolites, Reno” (D. Olson and A. Bisio, Eds.), p. 210.
Guildford-Butterworth, 1983.
1. “API 44 Tables,” American Petroleum Institute Research Project,
Vols. V and VI, Thermodynamic Research Center, Texas A&M Uni-
versity, 1968.
5
. CONCLUSIONS
1
1
1
2. Kazansky, V. B., Frash, M. V., and Van Santen, R. A., Appl. Catal. 146,
225 (1996).
Differences in activity of different zeolites are due not
only to differences in intrinsic acid strength but also to
a combination of adsorption and diffusion effects. We
demonstrated this using four zeolites that were intrinsically
3. Van den Broek, A. C. M., Van Grondelle, J., and Van Santen, R. A.,
J. Catal. 167, 417 (1997).
4. J a¨ nchen, J., Van Wolput, J. H. M. C., Van de Ven, L. J. M., De Haan,
J. W., and Van Santen, R. A., Catal. Lett. 39, 147 (1996).
equally acidic. Their activities per acid site could be there- 15. Bates, S. P., and Van Santen, R. A. [Unpublished results]
1
1
1
6. Meier, W. M., Olson, D. H., and Baerlocher, Ch., “Atlas of Zeolite
Structure Types.” Elsevier, Amsterdam, 1996.
7. J a¨ nchen, J., Stach, H., Uytterhoeven, L., and Mortier, W. J., J. Phys.
Chem. 100, 12489 (1996).
fore be directly compared. A higher adsorption enthalpy
led to a lower apparent activation energy and, thus, a higher
activity per acid site. On the zeolites under investigation the
activation energy of the elementary isomerization reaction
8. Eder, F. [Personal communication]
with respect to a n-alkoxy intermediate was the same for 19. Madon, R. J., O’Conell, J. P., and Boudart, M., AIChE 24, 904
(
1978).
all structures and amounted to approximately 125 kJ/mol.
This value is much higher than in the superacid solution be-
cause of the energy needed to break or lengthen the bond
between the alkoxy carbon and zeolite lattice oxygen.
2
2
0. This conclusion is supported by: Stockenhuber, M., Eder, F., and
Lercher, J. A., Stud. Surf. Sci. Catal. 97, 495 (1995).
1. Van der Pol, A. J. H. P., Verduyn, A. J., and Van Hooff, J. H. C., Appl.
Catal. 92, 113 (1992).
ZSM-22 showed pore-mouth catalysis due to its one- 22. Haag, W. O., Lago, R. M., and Weisz, P. B., Faraday Disc. Chem. Soc.
7
2, 317 (1982).
dimensional pore system with relatively small pores. This
zeolite was therefore relatively inactive. Both ZSM-5 and
ZSM-22showed shape selectivitywhile no dimethylbutanes
were formed and the 2-MP/3-MP ratio was not at equilib-
rium. The latter was at equilibrium on both large-pore zeo-
lites, mordenite and �. The reaction on �-zeolite may have
become diffusion limited at higher temperatures due to its
high activity per gram. The mordenite data indicate the pos-
sibility of single-file diffusion and a limited accessibility of
the acid sites.
2
3. Meier, W. M., Olson, D. H., and Baerlocher, Ch., “Atlas of Zeolite
Structure Types,” Elsevier, Amsterdam/New York, 1996.
4. Sachtler, W. M. H., and Zhang, Z., Adv. Catal. 39, 129 (1993).
5. Gianetto, G., Alvares, F., Ribeiro, F. R., P e´ rot, G., and Guisnet, M.,
in “Guidelines for Mastering the Properties of Molecular Sieves”
2
2
(
D. Barthomeuf, E. G. Derouane, and W. H o¨ lderich, Eds.), NATO
ASI Series B, Vol. 221, p. 355. Plenum Press, New York, 1990.
6. Carvill, B. T., Lerner, B. A., Adelman, B. J., Tomczak, D. C., and
Sachtler, W. M. H., J. Catal. 144, 1 (1993).
7. Jaeger, N. I., Jaeger, A. L., and Schulz-Ekloff, G., J. Chem. Soc. Faraday
Trans. 87, 1251 (1991).
2
2
2
8. Kazansky, V. B., Frash, M. V., and Van Santen, R. A., Appl. Catal. 146,
225 (1996).
ACKNOWLEDGMENT
29. Van de Runstraat, A., Stobbelaar, P. J., Van Grondelle, J., Anderson,
B. G., Van IJzendoorn, L. J., and Van Santen, R. A., Stud. Surf. Sci.
Catal. 105, 1253 (1996).
30. Brouwer, D. M., in “Chemistry and Chemical Engineering of Catalytic
Processes” (G. C. A. Schuit and R. Prins, Eds.), p. 137. Noordhoff,
Alphenaanden Rijn, Rockville, MD, 1980.
The work described in this paper was funded by the Dutch Organiza-
tion for Scientific Research (NWO) through its Foundation for Chemistry
(
SON).
31. Languasco, J. M., Cunningham, R. E., and Calvelo, A., Chem. Eng.
Sci. 27, 1459 (1972).
REFERENCES
3
2. (a) Blomsma, E., Martens, J. A., and Jacobs, P. A., J. Catal. 155, 141
(1995); (b) Blomsma, E., Martens, J. A., and Jacobs, P. A., J. Catal. 159,
323 (1996); (c) Blomsma, E., Martens, J. A., and Jacobs, P. A., Stud.
Surf. Sci. Catal. 105, 909 (1996).
1
. Weisz, P. B., in “Advances in Catalysis and Related Subjects, Vol. 13”
D. D. Eley, P. W. Selwood, and P. B. Weisz, Eds.), p. 157. Academic
Press, London, 1963.
(
2
. Kazansky, V. B., in “Acidity and Basicity of Solids, Theory, Assess- 33. Martens, J. A., Parton, R., Uytterhoeven, L., Jacobs, P. A., and
ment and Utility” (J. Fraissard and L. Petrakis, Eds.), p. 335. Kluwer
Academic, Dordrecht, 1994.
. Kazansky, V. B., and Senchenya, I. N., J. Catal. 119, 108 (1989).
. Ribeiro, F., Marcilly, C., and Guisnet, M., J. Catal. 78, 267 (1982).
. Steijns, M., and Froment, G. F., Ind. Eng. Chem. Prod. Res. Dev. 4, 660
Froment, G. F., Appl. Catal. 76, 95 (1991).
34. Martens, J. A., Souverijns, W., Verrelst, W., Parton, R., Froment, G. F.,
and Jacobs, P. A., Angew. Chemie 107, 2726 (1995).
35. (a) Li, C., and Zhu, Z., Fuel Sci. Technol. 9, 1103 (1991); (b) Marin,
G. B., and Froment, G. F., Chem. Eng. Sci. 37, 759 (1982).
36. Allain, J. F., Magnoux, P., Schulz, Ph., and Guisnet, M., in “Proceed-
ings of the DGMK-Conference: Catalysis on Solid Acid and Bases”
(J. Weitkamp and B. L u¨ cke, Eds.), p. 219. German Society for
Petroleum and Coal Science and Technology, 1996.
3
4
5
(
1981).
6
7
. Froment, G. F., Catalysis Today 1, 455 (1987).
. Van de Runstraat, A., Van Grondelle, J., and Van Santen, R. A., Ind.
Eng. Chem. Res., in press.