Catalytic cracking, dehydrogenation, and aromatization of isobutane over Ga/HZSM-5 and Zn/HZSM-5 at low pressures

Article abstract of DOI:10.1002/kin.10073

Isobutane cracking, dehydrogenation, and aromatization over Ga/HZSM-5 and Zn/HZSM-5 has been investigated in a Knudsen cell reactor and the kinetics of the primary reaction steps for isobutene and propene formation have been accurately determined, Although cracking is the dominant reaction channel, with propene and methane being primary products, methane formation is significantly less than propene formation. This indicates that a proportion of the cracking proceeds via Lewis acid attack at C-C bonds, and not just via alkanium ion formation at Bronsted acid sites. This is particularly apparent over Zn/HZSM-5. Intrinsic rate constants for cracking, calculated from the rate of propene formation, are k_3a(°C^-1) = 10^18.3±1.1 exp(-148±6 kjmol^-1/RT) Ga/HZSM-5 k_3a(°C^-1) = 10^14.8±1.1 exp(-114±6 kjmol^-1/RT) Zn/HZSM-5 and for dehydrogenation, calculated from the rate of isobutene formation, are k_3b(°C^-1) = 10^18.0±1.1 exp(-148±6kjmol^-1/RT) Ga/HZSM-5 k_3b(°C^-1) = 10^13.5±1.1 exp(-101 ± 6kjmol^-1/RT) Zn/HZSM-5 Large preexponential factors for cracking and dehydrogenation over Ga/HZSM-5 indicate that either the coverage of active sites is significantly less than the coverage of exposed sites or the intrinsic reaction step involves a large entropy change between reactant and transition state. For Zn/HZSM-5 the small preexponential factors suggest either small entropy changes during activation, perhaps initiated by Lewis acid sites, or a steady-state distribution of active and exposed sites is rapidly reached. Differences in intrinsic activation energies may reflect the ratio of Lewis and Bronsted acid sites on the respective catalyst surfaces. Aromatization is more prolific over Ga/HZSM-5 than over Zn/HZSM-5 under the low-pressure conditions.

Full text of DOI:10.1002/kin.10073

Catalytic Cracking,
Dehydrogenation, and
Aromatization of Isobutane
over Ga/HZSM-5 and
Zn/HZSM-5 at Low
Pressures
YANPING SUN, TREVOR C. BROWN
Chemistry, University of New England, Armidale, NSW, 2351, Australia
Received 28 December 2001; accepted 17 April 2002
DOI 10.1002/kin.10073
ABSTRACT: Isobutane cracking, dehydrogenation, and aromatization over Ga/HZSM-5 and
Zn/HZSM-5 has been investigated in a Knudsen cell reactor and the kinetics of the primary re-
action steps for isobutene and propene formation have been accurately determined. Although
cracking is the dominant reaction channel, with propene and methane being primary products,
methane formation is significantly less than propene formation. This indicates that a propor-
tion of the cracking proceeds via Lewis acid attack at C C bonds, and not just via alkanium ion
formation at Bronsted acid sites. This is particularly apparent over Zn/HZSM-5. Intrinsic rate
constants for cracking, calculated from the rate of propene formation, are
k
k
3a( C 1) = 1018.3 1.1 exp( 148 6 kJ mol 1/RT) Ga/HZSM-5
3a( C 1) = 1014.8 1.1 exp( 114 6 kJ mol 1/RT) Zn/HZSM-5
and for dehydrogenation, calculated from the rate of isobutene formation, are
k
k
3b( C 1) = 1018.0 1.1 exp( 148 6 kJ mol 1/RT) Ga/HZSM-5
3b( C 1) = 1013.5 1.1 exp( 101 6 kJ mol 1/RT) Zn/HZSM-5
Large preexponential factors for cracking and dehydrogenation over Ga/HZSM-5 indicate that
either the coverage of active sites is significantly less than the coverage of exposed sites or the
Correspondence to: Dr. Trevor Brown; e-mail: tbrown3@metz.
une.edu.au.
Contract grant sponsor: Australian Research Council.
2002 Wiley Periodicals, Inc.
c

468
SUN AND BROWN
intrinsic reaction step involves a large entropy change between reactant and transition state.
For Zn/HZSM-5 the small preexponential factors suggest either small entropy changes during
activation, perhaps initiated by Lewis acid sites, or a steady-state distribution of active and ex-
posed sites is rapidly reached. Differences in intrinsic activation energies may reflect the ratio
of Lewis and Bronsted acid sites on the respective catalyst surfaces. Aromatization is more pro-
C
lific over Ga/HZSM-5 than over Zn/HZSM-5 under the low-pressure conditions.
Periodicals, Inc. Int J Chem Kinet 34: 467–480, 2002
2002 Wiley
on clean surfaces are minimized and a range of tem-
peratures can be investigated over a short period of
time, without an internal standard. Although these
techniques have the effect of minimizing both sec-
ondary reactions and heat fluctuations the likely re-
action scheme remains complex, as it involves several
steps:
INTRODUCTION
Aromatization rates of small alkanes over HZSM-5
are greatly enhanced by the introduction of gallium or
zinc cations. For industrial applications Ga-containing
zeolites are generally preferred over Zn/HZSM-5 for
alkane aromatization as zinc is eluted from the lat-
ter catalyst during alkane conversion [1–3]. Over pure
HZSM-5, cracking and dehydrogenation are the domi-
nant reaction pathways. At low temperatures these de-
composition steps are initiated by proton donation from
the pure zeolite to the alkane to form highly energetic
alkanium ions [4]. A number of studies have aimed
to determine the role of ion exchanged or impregnated
metal cations in cracking, dehydrogenation and aroma-
tization. The experimental evidence suggests that both
zinc and gallium species are either directly or indirectly
responsible for increasing the rate of dehydrogenation
of the alkane reactant [5]. One proposal is that hydride
abstraction in the vicinity of Ga [6] and Zn [7] sites is
theinitialrate-determiningstep. Othershaveconcluded
that Zn and Ga cations are the driving force for dehy-
drogenation by combining the H-atoms, which form
during the activation of C H at Bronsted acid sites [8].
This occurs by the cyclic reduction and oxidation of Zn
and Ga species [9]. All other reaction steps occur on
Bronsted acid sites within the zeolite micropores, [5]
although it has been suggested that zinc may also be
involved in oligomerization of the alkenium ions and
dehydrogenation of cycloalkenes [7].
(i) Adsorption of gas-phase reactant R on the ex-
posed surface of the catalyst Sexp
k1
→
R(g) + Sexp
S exp(R)
(R1)
k
1
Direct adsorption onto active sites is not in-
cluded, as most of the catalytic activity is within
the micropores of the zeolite structure.
(ii) Reactant surface diffusion to and from active
sites A
k2
→
Sexp(R) + A
A(R) + Sexp
(R2)
k
2
(iii) Intrinsic reaction at the active sites to form ad-
sorbed products A(P)
k3
A(R) → A(P)
(R3)
In this paper the kinetics of the initial steps in the
catalytic cracking, dehydrogenation, and aromatization
of isobutane over Ga/HZSM-5 and Zn/HZSM-5 have
been investigated in a Knudsen cell reactor. Under low-
pressure conditions the first steps in a heterogeneous
reaction can generally be observed because gas-phase
and surface bimolecular reactions are minimized. Fur-
thermore this technique allows for the determination
of accurate rate parameters as the gas/surface collision
frequency may be accurately calculated and can be var-
ied by eitheradjustingthe flowrate, the exposedsurface
area of the catalyst, or the exit aperture area [4]. Also
by including temperature programming, temperature-
fluctuations resulting from nonequilibrium adsorption
This step my also involve several steps, such as
alkanium ion formation, rearrangement to alke-
nium ions and formation of alkenes.
(iv) Product diffusion from active sites to the ex-
posed surface
k4
A(P) + Sexp → S exp(P) + A
(R4)
(v) Product desorption from the exposed surface
k5
→
Sexp(P)
P (g) + Sexp
(R5)
k
5

CATALYTIC CRACKING, DEHYDROGENATION, AND AROMATIZATION OF ISOBUTANE
469
(vi) Escape from the Knudsen cell through the exit
ratio of the area of the reactor walls (Sv) to the area
of the exit aperture (Sea). Measured exit aperture areas
can be corrected by Clausing factors [10] to take into
account the finite width of the exit aperture, but for the
current reactors this factor is estimated to be one.
Zeolyst International (Kansas City, USA) supplied
the ammonium ZSM-5 powder (CBV 3024G) and con-
version to HZSM-5 was achieved by heating in air at
500 C. The original zeolite has a unit cell formula of
NH5.48(Si90.52Al5.48)O190.9 and N2 BET surface area
aperture with rate constant kep
kep
P(g) → Mass Spectrometer
(R6)
As a first approximation it is assumed that steps (R1)
and (R3) are rate determining. This means that a steady
state is quickly reached between the coverage of the
exposed and active sites (step (R2)), and that adsor-
bate diffusion (step (R3)) and desorption (step (R5))
are rapid. Also the escape rate constant kep must be
significantly larger than the intrinsic rate constant k3.
Larger than expected Arrhenius preexponential factors
fortheintrinsiccrackinganddehydrogenationofisobu-
tane over HZSM-5 have been previously attributed to
an inequality in the coverage of exposed and active
sites [4].
of 400
10 m2 g 1. Zn/HZSM-5 and Ga/HZSM-5
were prepared by ion exchange with HZSM-5 at 343 K
for 24 h in solutions of either 0.0050 M zinc nitrate or
0.0050 M gallium nitrate. The exchanged zeolites were
then dried in flowing air at 383 K for 24 h and then at
773 K for another 24 h. All substrates were pressed
binder-free and crushed to particles of 0.15–0.23 mesh
size. The Ga and Zn content was determined by X-ray
fluorescence and found to be 0.5 wt% in both cases. X-
ray diffraction patterns for pure HZSM-5, Zn/HZSM-5
and Ga/HZSM-5 were similar. This indicated that the
metals are finely dispersed and that no dealumination
or damage to the zeolite structure has occurred during
ion exchange. Isobutane, isobutene, propene, methane,
benzene, toluene, and xylene were used as supplied
without further purification.
EXPERIMENTAL
The Knudsen cell reaction system has been described in
detail elsewhere [4]. Briefly, isobutane passes through
a variable leak valve (Granville–Phillips), through 240
mm of 1 mm ID capillary glass tubing and into the
Pyrex reaction cell containing the solid substrate. Re-
actant molecules then experience collisions with the
walls of the reactor and substrate, prior to escaping
through the exit aperture and directly into a quadrupole
mass spectrometer. Reactor and substrate geometry de-
termine the average number of collisions within the
Knudsen cell. Flow rate is monitored before each ex-
periment by recording the pressure decrease from the
pressure gauge situated, in a known volume, next to the
variable leak valve. The detection system is a Hewlett
Packard 5995 quadrupole mass spectrometer. Details
of the two reactors used in the analysis are listed in
Table I. Collision numbers (Z) are calculated from the
RESULTS
Temperature Programmed
Reaction Profiles
A careful analysis of the mass spectra that resulted
from the exposure of isobutane to Zn/HZSM-5 and
Ga/HZSM-5 at temperatures ranging from 200 to
500 C indicated that the only products were methane,
propene, isobutene, benzene, toluene, and xylene
(molecular hydrogen could not be accurately detected).
Small amounts of ethene could also be evolved, partic-
ularly over Ga/HZSM-5, but quantification was dif-
ficult because of the interference of characteristic
mass spectral peaks from major products. Other pos-
sible products, ethane, propane, n-butanes, and pen-
tanes were not observed. When isobutane was flowed
through the reaction cell in the absence of catalyst
no reaction was apparent over the temperature range.
For the kinetic analysis gas-phase species were char-
acterized by m/e = 13 (methane), m/e = 36 (propene),
m/e = 56 (isobutene), m/e = 58 (isobutane), m/e = 78
(benzene), m/e = 91 (toluene), and m/e = 106 (xylene).
Isobutane had peaks at m/e = 56, m/e = 36, and m/e =
13, which were measured at the commencement of
each temperature program when no reaction was
Table I Key Physical Parameters, Calculated in the
Absence of a Catalyst, for the Two Knudsen Cell
Reactors
Reactor 1
Reactor 2
Volume
Internal surface
area
Exit aperture
area
Collisions/
molecule
Residence
time (s)
8830 5 mm3
2355 2 mm2
7414 5 mm3
2320 2 mm2
0.95 0.03 mm2
2480 200
1.54 0.03 mm2
1510 200
0.25(M/T)1/2
0.13(M/T)1/2

470
SUN AND BROWN
apparent. Isobutene had peaks at m/e = 36 and m/e =
13, while propene contributed to m/e = 13. For the
aromatics xylene had peaks at m/e = 91 and m/e = 78,
while toluene contributed to m/e = 78. For these prod-
ucts calibration factors were determined using the pure
gas or vapor over a range of flow rates. These are listed
in Table II. Also listed are mass-spectral conversion
factors (␣(abundance s molecule 1)), which convert
flow rate to abundance for the gaseous species. Accu-
rate conversion factors for benzene, toluene, and xy-
lene could be easily determined because a broad range
of vapor flow rates was difficult to control and mea-
sure. Hence, conversion factors for each of the aromat-
ics were calculated from a reactant and product mass
balance.
rate became nonlinear between 490 and 500 C and so,
for kinetic analysis, data in this region were discarded.
Typical profiles for isobutane over Ga/HZSM-5 and
Zn/HZSM-5 are plotted in Figs. 1 and 2. Here the
1
heating rate is 5 C min and flow rates into the reac-
1
tor were 4.73 1014 and 5.61 1014 molecules s
respectively. For both substrates the isobutane escape
rate continuously declined from 300 C, while prod-
ucts began to appear from this temperature. Because
of different reactant and product sensitivities to the
characteristic mass spectral peaks, more informative
profiles were obtained by plotting fractional yield and
decomposition against temperature. Fractional decom-
position of isobutane and fractional yields of products
were calculated from
I
IC H
4 10
C4H10
I
IC H
4 10
IC H
IC H
3 6
C4H10
4
8
fDecomposition
=
=
(3)
(4)
I
C4H10
␣
␣
C4H8
␣
C3H6
C4H10
␣
IProduct
C4H10
IBenzene
IToluene
IXylene
fYield
= 2
+
+
␣
Xylene
(1)
␣
Product I
C4H10
␣
␣
Toluene
Benzene
and are plotted in Figs. 3–6. Only 60% of isobu-
tane decomposed over Zn/HZSM-5 at 500 C, while
under similar conditions almost 90% was lost over
Ga/HZSM-5. Product evolution rates decreased in the
order propene > isobutene > methane > benzene
> toluene > xylene for Zn/HZSM-5 catalysis and
propene > methane > isobutene toluene > benzene
> xylene for Ga/HZSM-5. That is, for both systems
propene was the dominant product and xylene had the
lowest yield, while evolution order for the other four
products was dependent on the catalyst. Propene was
also the dominant product for isobutane decomposition
over pure HZSM-5 [4].
where IC H and IC H are mass-spectral abundances
4
10
4
10
characteristic of isobutane before and after reaction, re-
spectively. Rearranging leads to the following expres-
sion for calculating the conversion factor for benzene
IXylene
IToluene
2 IBenzene
+
+
a
a0
␣
Benzene
=
(2)
IC4
IC3
H
H
(I I)
8
6
␣
C
␣
C
␣
C
3
H
H
H
6
4
10
4
10
The two experimentally measured constants a and a0
are the ratios of calibration constants ␣Toluene/␣Benzene
and ␣Xylene/␣Benzene, respectively. For toluene a = 0.50
0.05 and for p-xylene a0 = 0.30 0.05.
Eight temperature-programmed experiments were
carried out for isobutane flowing over Zn/HZSM-5 and
Ga/HZSM-5. Flow rates ranged from 2.16 1014 to
2.90 1015 molecules s 1 (isobutane reactor pressure
ranged from 1–20 Pa) and the heating rate was varied
Kinetics of Alkane and Alkene Formation
The methodology used to determine rate parame-
ters for alkane and alkene formation during cracking
and dehydrogenation has been previously described in
1
between 2 and 5 C min 1. For 5 C min the heating
Table II Calibration Factors for the Mass Spectrometer
1
␣ (I s mol
)
Corrected Abundance
Isobutane
Isobutene
Propene
(1.2 0.2) 1014
(1.8 0.6) 1015
(4.2 0.5) 1013
(1.6 0.1) 1014
I58
I56 = I56 (0.138 0.003) I58
I36 = I36 (0.0461 0.0017 ) I58 (0.0115 0.0011) I56
I13 = I13 (0.00610 0.0008 ) I58 (0.009 0.003) I56
(0.49 0.03) I36
Methane
Benzene
Toluene
Xylene
(3.0 0.9) 1015
(1.5 0.4) 1015
(8.9 0.3) 1014
I78 = I78 (0.162 0.003) I106 (0.0026 0.0005) I91
I91 = I91 (2.11 0.18) I106
I106

CATALYTIC CRACKING, DEHYDROGENATION, AND AROMATIZATION OF ISOBUTANE
471
Figure1 Correctedmass-spectralabundancesforGa/HZSM-5decompositionofisobutane( m/e=58), formationofisobutene
(4 m/e = 56), propene ( m/e = 36), methane ( m/e = 13), and aromatics as indicated, during temperature programming.
1
1
Initial isobutane flow rate is 4.73 1014 molecules s and heating rate is 5 C min
.
detail [4]. Assumed rate determining steps are ad-
sorption and desorption of isobutane and surface
reaction.
abundance of a mass spectral peak uniquely character-
istic of a primary product is proportional to the mass-
spectral calibration factor and the product flow rate,
k1
dnProduct
→
C4H10 + Sexp
S exp(C4H10)
(R1)
IProduct = ␣Product
(5)
k
1
dt
k3a
A(C4H10) → A(C3H6 + CH4)
(R3a)
(R3b)
The product flow rate may be calculated from the colli-
sion number Z, the average reactant flow rate and a ratio
of rate constants determined by assuming a steady-state
coverage of isobutane:
k3b
A(C4H10) → A(C4H8 + H2)
Here adsorbed species are in parentheses. The rate con-
1
stants k 1, k3a, and k3b have units s site 1, while
dnProduct
dt
k3a or 3b
dn
C4H10
k1PC H , where P is partial pressure, is the isobu-
=
Z
(6)
4
10
k
1 + k3a + k3b
dt
tane/catalyst surface collision frequency. Rate con-
stants for surface diffusion to or from active sites have
been previously estimated to be of the order 4 106–
The collision number is the ratio of the exposed catalyst
surface area to the exit aperture area (SHZ,e/Sea), while
the reactant flow rate is calculated from the mean of the
1
3
109 s [4] and so as a first approximation these
steps are considered not to be rate determining. The

472
SUN AND BROWN
Figure 2 Corrected mass-spectral abundances for Zn/HZSM-5 catalyzed decomposition of isobutane ( m/e = 58), formation
of isobutene (4 m/e = 56), propene ( m/e = 36), methane ( m/e = 13), and aromatics as indicated, during temperature
1
1
programming. Initial isobutane flow rate is 5.61 1014 molecules s and heating rate is 5 C min
.
abundances entering and leaving the Knudsen cell
surface reaction step is
ProductSHZ,e A3a or 3b
E1
␣
k3a or 3b exp
ProductSHZ,e
␣
+ k3a +RkT3b
IProduct
=
IProduct
=
␣
C4H10
Sea
A
1
␣
C4H10
Sea
k
1
E1 + E
E3a or 3b
1
I
+ IC H
C4H10
4
10
exp
(7)
RT
2
I
+ IC H
C4H10
4
10
The activation energy for adsorption (E1) of isobutane
onto the zeolite surface was included in these calcula-
tions. The geometric surface area of the catalyst [11,12]
was used as an estimate for the exposed surface area of
the catalyst and could have introduced errors of a factor
of ten in the reported Arrhenius preexponential factors.
A value of 330 mm2 for SHZ,e was used in all calcula-
tions, which represents ca. 14% of the internal surface
area of the Knudsen cell reactor. Further simplification
of Eq. (7) is possible by noting that the surface reac-
tion rate constants (k3a) and (k3b) are much smaller than
the isobutane desorption rate constant (k 1). Hence the
full rate expression with Arrhenius parameters for each
(8)
2
The unknowns in this expression are the preexponential
factors and activation energies. These are determined
by taking the natural logarithm of Eq. (8) and rear-
ranged to give
!
IC H + IC H
␣ProductSHZ,e
4
10
4
10
ln
2IProduct
␣
Sea
C4H10
A
E1
E
1 + E3a or 3b
1
= ln
+
(9)
A3a or 3b
RT

CATALYTIC CRACKING, DEHYDROGENATION, AND AROMATIZATION OF ISOBUTANE
473
Figure 3 Fractional isobutane decomposition ( ) over Ga/HZSM-5 and apparent yields of isobutene (4), propene ( ) and
methane ( ), calculated from abundances plotted in Fig. 1.
A plot of the left-hand side of Eq. (9) against 1/RT pro-
duced a straight line with slope equal to E1
isobutane cracking over pure HZSM-5 this ratio was
less than 2:1, which is not considered to be significant
[4]. In this system it was assumed that the intrinsic re-
action step is dominated by the formation of alkanium
ions [(H3C)3CH2]+ at Bronsted sites [HO]
E
+
1
E3a or 3b and an intercept of ln(A 1/A3a or 3b). This
plot is shown in Figs. 7 and 8 for the temperature-
programmed experiments depicted in Figs. 1 and 2. For
all Arrhenius plots overall conversions of 10–50% were
used. Average Arrhenius preexponential factors and ac-
tivation energies with standard deviations determined
from all temperature-programmed isobutane decompo-
sitions over Ga/HZSM-5 and Zn/HZSM-5 systems are
listed in Table III.
(H3C)3CH + [HO] → [(H3C)3CH2]+ + [O]
which subsequently rearrange to alkenium ions releas-
ing hydrogen and methane.
[(H3C)3CH2]+ → [(H3C)3C]+ + H2
→ [(H3C)2CH]+ + CH4
DISCUSSION
The observations of particular significance in the
temperature-programmed reaction profiles obtained
for both catalysts are the low yields of methane (Figs.
3 and 5). In the case of isobutane cracking over
Ga/HZSM-5 the ratio of propene to methane was ca.
2:1 while for Zn/HZSM-5 the ratio was ca. 8:1. For
It is then assumed that the molecular species proceed to
the mass spectrometer, while the alkenium ions rapidly
eliminate a proton to form propene or isobutene. How-
ever, as yields of methane are low over all HZSM-5
catalysts, particularly Zn/HZSM-5, a different mecha-
nism must be invoked. In this case the rate determining

474
SUN AND BROWN
Table III Measured Arrhenius Rate Parameters for the Cracking and Dehydrogenation of Isobutane Over HZSM-5 [4],
Ga/HZSM-5 and Zn/HZSM-5. Preexponential Factors have been Adjusted for the Heating Rate
Apparent
Intrinsic
log(␤A 1/A3)
(log( C s 1))
E1
(kJ mol
E
+ E3
log(A3/␤)
(log( C 1))
E3
(kJ mol
1
1
1
)
)
HZSM-5
Cracking
Dehydrogenation
Ga/HZSM-5
5.30 (12)
4.81 (12)
121.7 (1.7)
123.4 (1.7)
17.3 (1.1)
16.8 (1.1)
170 (6)
172 (6)
Cracking
5.72 (12)
6.29 (12)
4.80 (10)
5.96 (10)
93.3 (1.6)
99.1 (1.4)
88.7 (1.2)
99.6 (1.2)
Correcteda
18.3 (1.1)
18.0 (1.1)
148 (6)
148 (6)
Dehydrogenation
Correcteda
Zn/HZSM-5
Cracking
2.85 (8)
2.76 (8)
1.60 (8)
1.54 (8)
67.3 (1.2)
65.7 (1.2)
53.6 (1.2)
52.5 (1.2)
Correcteda
14.8 (1.1)
13.5 (1.1)
114 (6)
101 (6)
Dehydrogenation
Correcteda
aCorrected parameters to account for the loss of alkenes to form aromatics.
Figure 4 Fractional apparent yields over Ga/HZSM-5 of toluene, benzene, and xylene calculated from the corrected abundances
plotted in Fig. 1.

CATALYTIC CRACKING, DEHYDROGENATION, AND AROMATIZATION OF ISOBUTANE
475
Figure 5 Fractional isobutane decomposition ( ) over Zn/HZSM-5 and apparent yields of isobutene (4), propene ( ), and
methane ( ), calculated from abundances plotted in Fig. 2.
step may be associated with zeolite Lewis acid sites
[AlO]+ or the added metal sites [6] where extraction
of hydride ions and electron attack of C C bonds can
occur [13]. For example,
initiating step is dominant. Bimolecular surface pro-
cesses are already apparent with the formation of aro-
matics over both Ga/HZSM-5 and Zn/HZSM-5. For
isobutanetransformationsoverH-mordeniteithasbeen
demonstrated that the primary products are propane,
n-butane and isopentane, involving the formation of
a C8 alkenium ion; methane is not a major prod-
uct [14]. Over the two HZSM-5 catalysts the only
observed primary products from the Knudsen cell
are methane, propene and isobutene. Hence it is un-
likely that a bimolecular step is the dominant initiat-
ing reaction step for cracking and dehydrogenation,
and so does not explain the observed low yield of
methane.
(H3C)3CH + [AlO]+ → [(H3C)3C]+ + [AlOH]
→ [(H3C)2CH]+ + [AlOCH3]
The hydrogen and methyl groups are bonded to the
surface and can only desorb as molecular hydrogen or
methane if they subsequently come into contact with
a proton. Recombinative desorption of H-atoms as H2
has been shown to occur at Zn and Ga cations [8].
Under low-pressure conditions H-atom coverage will
be low and so recombination may be limited. As a
consequence the effective rates of CH4 and H2 diffu-
sion and desorption (steps (R4) and (R5)) may be rate
determining.
Cracking, Dehydrogenation, and
Aromatization over Ga/HZSM-5
From Fig. 1 it is apparent that after an initial exponen-
tial increase product evolution rates begin to plateau
at about 370 C or 50% conversion. This slowing of
An alternative to this proposed mechanism to ex-
plain the low methane yields is that a bimolecular

476
SUN AND BROWN
Figure 6 Fractional apparent yields over Zn/HZSM-5 of toluene, benzene, and xylene calculated from the corrected abundances
plotted in Fig. 2.
the change in rate is because isobutane pressure has
decreased to a point where adsorption and transport to
active sites is rate limiting. The decline is also due to the
formation of aromatics. Temperature-dependent frac-
tional yields shown in Fig. 4 indicate that the ratio of
toluene to benzene is approximately 2:1, while toluene
to xylene is 2:1 at low temperatures but increases to
5:1 at 500 C. Statistically the maximum ratio is 2:1
if the aromatics are produced from only two alkene
precursors; isobutene and propene or related adsorbed
complexes. That is, one isobutene and one propene
molecule interact to give toluene, two isobutenes give
xylene and two propenes give benzene. For xylene fur-
ther decomposition to toluene and to benzene may
be affecting the measured ratios at higher tempera-
tures. It is reasonable to assume that surface alkenes
or associated alkenium ions are precursors to aromatic
formation [8].
in Eq. (9) according to
␣
1
Propene
=
(10)
IPropene
2IBenzene
Benzene
IToluene
IPropene,c
+
+
␣
Propene
␣
␣
Toluene
Benzene contributions are doubled because two
propene molecules are lost to benzene formation. Anal-
ogous corrections were applied to the isobutene abun-
dance to obtain adjusted rate parameters (Table III).
The result was that the activation energy for isobutene
1
formation increased by 10.8 kJ mol and propene
formation increased by 5.8 kJ mol 1. Preexponential
factors for both pathways also increased. Corrections
could also be applied to methane formation, as the con-
version of xylene to toluene possibly included the re-
lease of methane. However, the correction was small
and could not be accurately measured.
Evolutionratesofisobuteneandpropenemustthere-
fore be corrected to account for the loss due to aromatic
formation. This is achieved by adjusting ␣Product/IProduct
Apparent activation energies for cracking (propene
formation) and dehydrogenation (isobutene formation)
over Ga/HZSM-5 were similar. This was also the case

CATALYTIC CRACKING, DEHYDROGENATION, AND AROMATIZATION OF ISOBUTANE
477
Figure 7 Arrhenius-type plots of ln(k 1/k3) against 1/RT for isobutane dehydrogenation and cracking over Ga/HZSM-5.
for isobutane reactions over HZSM-5, and indicates
that the relative reaction rate, between cracking and
dehydrogenation, is controlled by the preexponential
factors. Intrinsic rate parameters for the cracking and
dehydrogenation reaction step can be estimated from
reportedadsorptionenthalpiesandfromtypicaldesorp-
tion preexponential factors. The enthalpy of adsorp-
tion of isobutane on HZSM-5 and silicate has been
measured using a range of techniques and found to be
48.7 4.7 kJ mol 1 [15]. Typical preexponential fac-
tors for physisorbed systems are A 1 = 1012 1 s 1 [16].
From the intrinsic rate parameters listed in Table III,
it may be seen that for Ga/HZSM-5 the cracking pre-
exponential factor is larger than the dehydrogenation
value by (2.1 1.7). It has been previously noted that
a factor of three between preexponential factors can be
associatedwithdifferencesinreactionpathdegeneracy.
For cracking of isobutane three identical C C bonds
are available for Lewis acid attack, while only one hy-
drogen atom, attached to the tertiary carbon, is avail-
able for hydride abstraction. A difference in the kinet-
ics is that the activation energies over Ga/HZSM-5 are
1
23 kJ mol less than over HZSM-5, while preexpo-
nential factors are a factor of 10 larger. Smaller activa-
tion energies disagree with reported observations that
gallium decreases the acid strength of the HZSM-5 ac-
tive sites [17], but the increased number of Lewis acid
sites [6] may contribute to a lowering of the energy
barrier.
Two explanations for the large preexponential fac-
tors have previously been suggested [4]. One is that
the coverage of isobutane on the external surface of the
catalyst is greater than the coverage in the micropores.
That is, preexponential factors for unimolecular reac-
tions where the entropy difference between the reactant
1
and transition-state is negligible are typically 1013
s

478
SUN AND BROWN
Figure 8 Arrhenius-type plots of ln(k 1/k3) against 1/RT for isobutane dehydrogenation and cracking over Zn/HZSM-5.
[18]. Unitsfortheintrinsicpreexponentialfactorslisted
in Table III are degrees centigrade geometric surface
coverage per active site coverage. Hence, if A2a or A2b
is much larger than 1013 then active site coverge
geometric surface coverage. It has been previously es-
timated that the time constant for diffusion to active
sites at 200–500 C is in the range 10 7–10 10 s, which
is much shorter than the residence time in the cell and
the reaction time constants [4]. There may, however,
be blockages within the zeolite or inaccessible active
sites that slow the diffusion rate. A second possibility
is that there is a large entropy increase when form-
ing the transition-state. As alkanium ion formation is
the dominant pathway for Ga/HZSM-5 catalyzed de-
composition, entropy increase can be associated with
extra vibrational motions caused by the insertion of
the proton from Bronsted acid sites into the activated
complex.
Cracking, Dehydrogenation, and
Aromatization over Zn/HZSM-5
Fractional yields and decomposition plotted in Fig. 5
show that propene and methane formation and isobu-
tane loss increase exponentially over the temperature
range. Isobutene formation rate slows from 440 C,
which shows that the dehydrogenation mechanism and
hence rate are different from the cracking kinetics.
As aromatic yields are much lower than for
Ga/HZSM-5, trends in these data are ambiguous.
The expected statistical distribution of two toluene
molecules for each benzene and xylene is not observed.
By again assuming that propene and isobutene forma-
tion is a precursor to aromatic formation, cracking and
dehydrogenation rate parameters are corrected by ad-
justing ␣Product/IProduct using Eq. (10) and the equiv-
alent for isobutene. Corrected activation energies and

CATALYTIC CRACKING, DEHYDROGENATION, AND AROMATIZATION OF ISOBUTANE
479
preexponential factors were listed in Table III. Because
of the small aromatic yields the rate parameters are not
significantly affected by this adjustment.
Bronsted acid sites on the respective catalysts.
Lewis acid sites may dominate for Zn/HZSM-
5, while for Ga/HZSM-5 both Lewis and Bron-
sted acid sites contribute to the decomposition.
Negligible reaction entropy differences are ex-
pected if the transition state is not larger or more
complex than the reactant. For Lewis acid ini-
tiated reaction, involving hydride abstraction or
electron attack of C C bonds, the structure of
the transition-state is likely to be less complex
thanalkaniumionformation. Continuingthisar-
Intrinsic activation energies for cracking were 13
1
2 kJ mol higher than for dehydrogenation over
Zn/HZSM-5 and preexponential factors varied by a
factor of (20
2). Unlike for pure HZSM-5 and
Ga/HZSM-5, both rate parameters controlled the rel-
ative rates of cracking and dehydrogenation of isobu-
tane with the Zn2+ exchanged HZSM-5. This is further
evidence that either different mechanisms or different
rate determining steps are involved for the two reaction
pathways. The very small rates of methane formation
for this catalyst suggest that cracking, at least, is dom-
inated by Lewis acid attack.
In comparison with Ga/HZSM-5 isobutane reac-
tions, activation energies were significantly lower for
Zn/HZSM-5; 33 3 kJ mol 1 and 47 2 kJ mol 1 for
cracking and dehydrogenation respectively. Ono and
Kanae [19] have shown that Zn/HZSM-5 has weaker
acid sites than Ga/HZSM-5, which is an apparent con-
tradiction to the activation energies listed in Table III.
However, our results show that overall yields of all
products are lower for Zn/HZSM-5, which is reflected
in the significantly lower preexponential factors. Dif-
ferences between intrinsic activation energies obtained
for pure HZSM-5, Ga/HZSM-5 and Zn/HZSM-5 catal-
ysis may therefore be attributed to the ratio of Lewis
acid to Bronsted acid sites.
gument a larger intrinsic preexponential factor
1.1
for propene formation (1014.8
C
1), when
compared with isobutene (1013.5
C
1), is
1.1
consistent with a larger transition-state for
cracking.
CONCLUSION
For isobutane decomposition over pure HZSM-5,
Ga/HZSM-5, and Zn/HZSM-5, the dominant initial re-
action pathway was cracking to form propene. Dehy-
drogenation was also a significant reaction, having a
rate constant two to three times lower than the compet-
ing cracking reaction step. Activation energies were
similar for cracking and dehydrogenation on the same
catalyst for HZSM-5 and Ga/HZSM-5, and so the dif-
ference in rate was primarily due to differences in
reaction path degeneracy. For isobutane catalysis on
Zn/HZSM-5 a large difference in preexponential fac-
tors and activation energies were calculated for the
competing reaction steps, which suggested a pathway
other than alkanium ion formation is initiating decom-
position.
The low rates of methane evolution over both
Ga/HZSM-5 and Zn/HZSM-5 indicate that Lewis acid
sites initiate cracking and dehydrogenation. That is,
electron attack of C C bonds and hydride abstraction
are the primary reaction steps leaving hydrogen and
methyl groups directly bonded to the catalyst surface.
At low pressures and hence low coverages, bimolecular
formation of H2 and CH4 is likely to be limited. Bron-
sted acid sites also contributed to isobutane decompo-
sition, but for Zn/HZSM-5 this reaction pathway was
minor. Differences between intrinsic activation ener-
gies over different catalysts reflect the ratio of the two
types of acid sites on the surface.
Two possible explanations for the intrinsic preex-
ponential factors for Zn/HZSM-5 cracking and de-
hydrogenation, which are small when compared to
Ga/HZSM-5 and pure HZSM-5 values, are proposed:
(i) The active surface area coverage is similar to
the external surface area coverage. Hamid et
al. [20] observed that ion-exchanged Ga3+ ini-
tially tends to concentrate on the outer surface of
HZSM-5, but reductive reaction and oxidative
regeneration leads to a redistribution of gallium
in the zeolite. Because of the smaller coordina-
tion sphere of Zn2+, ion exchange of this diva-
lent cation occurs directly onto intrazeolite sites
[9]. Such a distribution may allow reactants and
products to more freely diffuse into the microp-
ores of Zn/HZSM-5, and so more rapidly reach a
steady-state. External Ga3+ may block or hinder
diffusionintothemicropores. Similarblockages
are not expected for pure HZSM-5, but acid sites
not associated with metal cations may be less
accessible.
Differences in intrinsic preexponential factors may
be related to entropy differences between the reactant
and transition-state. For Lewis acid attack entropy dif-
ferences were small, while the formation of alkanium
ions resulted in a large entropy change. An alternative
explanation for the preexponential factors for isobutane
(ii) The entropy difference between reactants and
transition-state is negligible. This may be a con-
sequence of the concentrations of Lewis and

480
SUN AND BROWN
decomposition is that the ratio of exposed-site cover-
age to active-site coverage is high for Ga/HZSM-5, but
low for Zn/HZSM-5. Blockages caused by large Ga3+
ions in the micropores of the zeolite may hinder the
rate at which a steady-state coverage is reached.
The formation of aromatics was more prolific on
Ga/HZSM-5 than on Zn/HZSM-5 at low isobutane
pressures during temperature-programming. This was
simply a consequence of the higher alkene yields over
Ga/HZSM-5. All experiments verified that precursors
to benzene, toluene and xylene were propene and
isobutene or corresponding alkenium ions.
5. Guisnet, M.; Gnep, N. S. Appl Catal 1996, 146, 33.
6. Kwak, B. S.; Sachtler, W. M. H. J Catal 1994, 145, 456.
7. Rojasova, E.; Smieskova, A.; Hudec, P.; Zidek, Z. React
Kinet Catal Lett 1999, 66, 91.
8. Biscardi, J. A.; Iglesia, E. J Catal 1999, 182, 117.
9. Biscardi, J. A.; Iglesia, E. Catal Today 1996, 31, 207.
10. Iczkowski, R. P.; Margrave, J. L.; Robinson, S. M. J Phys
Chem 1963, 67, 229.
11. Keyser, L. F.; Moore, S. B.; Leu, M.-T. J Phys Chem
1991, 95, 5496.
12. Underwood, G. M.; Li, P.; Usher, C. R.; Grassian, V. H.
J Phys Chem A 2000, 104, 819.
13. Marczewski, M. J Chem Soc, Faraday Trans 1 1986, 82,
1687.
14. Bearez, C.; Avendano, F.; Chevalier, F.; Guisnet, M. Bull
Soc Chim France 1984, 3, 346.
15. Nijhuis, T. A.; van-den-Broeke, L. J. P.; Linders, M. J.
G.; Makkee, M.; Kapteijn, F.; Moulijn, J. A. Catal Today
1999, 53, 189.
Yanping Sun is appreciative of UNERS and OPRS awards.
BIBLIOGRAPHY
16. Pitt, I. G.; Gilbert, R. G.; Ryan, K. R. J Chem Phys 1995,
102, 3461.
1. Giannetto, G.; Monque, R.; Galiasso, R. Catal Rev-Sci
Eng 1994, 36, 271.
2. O’Connor, C. T. In Handbook of Heterogeneous Cataly-
sis; Ertl, G.; Knozinger, H.; Weitkamp, J. (Eds.); VCH:
Weinheim, 1997, Vol. 4, pp. 2069.
3. Raichle, A.; Moser, S.; Traa, Y.; Hunger, M.; Weitkamp,
J. Catal Commun 2001, 2, 23.
4. Yanping, S.; Brown, T. C. J Catal 2000, 194, 301.
17. Choudhary, V. R.; Mantri, K.; Sivadinarayana, C. Mi-
croporous Mesoporous Mater 2000, 37, 1.
18. Benson, S. W. ThermochemicalKinetics, 2nded.;Wiley:
New York, 1976.
19. Ono, Y.; Kanae, K. J. Chem Soc, Faraday Trans 1991,
87, 669.
20. Hamid, S. B.; Derouane, E. G.; Demortier, G.; Riga, J.;
Yarmo, M. A. Appl Catal A 1994, 108, 85.