Catalytic consequences of spatial constraints and acid site location for monomolecular alkane activation on zeolites

Article abstract of DOI:10.1021/ja808292c

The location of Bronsted acid sites within zeolite channels strongly influences reactivity because of the extent to which spatial constraints determine the stability of reactants and of cationic transition states relevant to alkane activation catalysis. Turnover rates for monomolecular cracking and dehydrogenation of propane and n-butane differed among zeolites with varying channel structure (H-MFI, H-FER, H-MOR) and between OH groups within eight-membered ring (8-MR) side pockets and 12-MR main channels in H-MOR. Measured monomolecular alkane activation barriers depended on catalyst and reactant properties, such as deprotonation enthalpies and proton affinities, respectively, consistent with Born-Haber thermochemical cycles that define energy relations in acid catalysis. Monomolecular alkane cracking and dehydrogenation turnovers occurred with strong preference on acid sites contained within smaller 8-MR pockets in H-MOR, while rates on sites located within 12-MR channels were much lower and often undetectable. This strong specificity reflects transition states that are confined only partially within 8-MR pockets; as a result, entropic gains compensate for enthalpic penalties caused by their incomplete containment to give a lower free energy for transition states within small 8-MR side pockets. These effects of entropy are stronger for dehydrogenation, with a later and looser transition state, than for cracking in the case of both propane and n-butane; therefore, selectivity can be tuned by the selective positioning or titration of OH groups within specific environments, the number of which was assessed in H-MOR by rigorous deconvolution of their infrared spectra. Specifically, cracking-to- dehydrogenation ratios for propane and n-butane were much smaller and terminal-to-central C-C bond cleavage ratios for n-butane were much larger on 8-MR than on 12-MR acid sites as a result of partial confinement, a concept previously considered phenomenologically as pore mouth catalysis. These marked effects of spatial constraints and of entropic factors on acid site reactivity and selectivity, also inferred for MFI from titration of OH groups by Na ⁺, have not been previously proposed or recognized and appear to be unprecedented in hydrocarbon catalysis. These findings and their conceptual interpretations open opportunities for the design of microporous solids by the rational positioning of acid sites within specific channel locations and with predictable consequences for catalytic rates and selectivities.

Full text of DOI:10.1021/ja808292c

Published on Web 01/15/2009
Catalytic Consequences of Spatial Constraints and Acid Site
Location for Monomolecular Alkane Activation on Zeolites
Rajamani Gounder and Enrique Iglesia*
Department of Chemical Engineering, UniVersity of California at Berkeley,
Berkeley, California 94720
Received October 22, 2008; E-mail: iglesia@berkeley.edu
Abstract: The location of Brønsted acid sites within zeolite channels strongly influences reactivity because
of the extent to which spatial constraints determine the stability of reactants and of cationic transition states
relevant to alkane activation catalysis. Turnover rates for monomolecular cracking and dehydrogenation of
propane and n-butane differed among zeolites with varying channel structure (H-MFI, H-FER, H-MOR)
and between OH groups within eight-membered ring (8-MR) side pockets and 12-MR main channels in
H-MOR. Measured monomolecular alkane activation barriers depended on catalyst and reactant properties,
such as deprotonation enthalpies and proton affinities, respectively, consistent with Born-Haber thermo-
chemical cycles that define energy relations in acid catalysis. Monomolecular alkane cracking and
dehydrogenation turnovers occurred with strong preference on acid sites contained within smaller 8-MR
pockets in H-MOR, while rates on sites located within 12-MR channels were much lower and often
undetectable. This strong specificity reflects transition states that are confined only partially within 8-MR
pockets; as a result, entropic gains compensate for enthalpic penalties caused by their incomplete
containment to give a lower free energy for transition states within small 8-MR side pockets. These effects
of entropy are stronger for dehydrogenation, with a later and looser transition state, than for cracking in the
case of both propane and n-butane; therefore, selectivity can be tuned by the selective positioning or titration
of OH groups within specific environments, the number of which was assessed in H-MOR by rigorous
deconvolution of their infrared spectra. Specifically, cracking-to-dehydrogenation ratios for propane and
n-butane were much smaller and terminal-to-central C-C bond cleavage ratios for n-butane were much
larger on 8-MR than on 12-MR acid sites as a result of partial confinement, a concept previously considered
phenomenologically as pore mouth catalysis. These marked effects of spatial constraints and of entropic
+
factors on acid site reactivity and selectivity, also inferred for MFI from titration of OH groups by Na , have
not been previously proposed or recognized and appear to be unprecedented in hydrocarbon catalysis.
These findings and their conceptual interpretations open opportunities for the design of microporous solids
by the rational positioning of acid sites within specific channel locations and with predictable consequences
for catalytic rates and selectivities.
-1
4,5
1
. Introduction
kJ mol ). Acid strength, reflected rigorously in these DPE
values, strongly influences activation barriers for reactions
involving cationic transition states that are stabilized by
electrostatic interactions with framework oxygens, which also
The design and selection of acidic zeolites for specific
reactions requires that we consider the consequences of the
number, structure, and acid strength of active sites, and of
intrachannel environments, within which sites stabilize adsorbed
intermediates and transition states relevant to chemical trans-
formations. At an early and defining moment in catalysis by
zeolites, cracking rates of n-hexane on H-MFI zeolites were
6,7
stabilize protons removed in deprotonation events. In contrast,
acid strength influences only weakly reactions requiring acti-
vated complexes with low charge, which are stabilized by
6,7
covalent bonds with the zeolite framework.
We have recently reported the remarkable enzyme-like
specificity of eight-membered ring (8-MR) channels in MOR
zeolites for CO insertion into methyl groups, an elementary step
1
,2
reported to depend only on the number of acid sites,
irrespective of Al content and, by inference, of the distribution
of Al among the 12 crystallographic T-sites in MFI. Quantum
chemical descriptions, however, indicated that the location of
the Al atoms in MFI influenced Brønsted acid strength and
8
that involves cationic transition states. In contrast, the exchange
of CD
4
with OH, which proceeds via essentially uncharged
covalent transition states, showed similar rates on Brønsted acid
3
reactivity. Later studies rigorously accounted for long-range
electrostatic interactions and found that deprotonation enthalpies
(
4) Eichler, U.; Br a¨ ndle, M.; Sauer, J. J. Phys. Chem. B 1997, 101, 10035.
(
DPE) were instead similar for all Al sites in MFI (1229-1240
(5) Sauer, J.; Sierka, M. J. Comput. Chem. 2000, 21, 1470.
(
(
6) van Santen, R. A.; Kramer, G. J. Chem. ReV. 1995, 95, 637.
7) Rigby, A. M.; Kramer, G. J.; van Santen, R. A. J. Catal. 1997, 170,
(
(
(
1) Olson, D. H.; Haag, W. O.; Lago, R. M. J. Catal. 1980, 61, 390.
2) Haag, W. O.; Lago, R. M.; Weisz, P. B. Nature 1984, 309, 539.
3) Kazansky, V. B. Acc. Chem. Res. 1991, 24, 379.
1.
(8) Bhan, A.; Allian, A. D.; Sunley, G. J.; Law, D. J.; Iglesia, E. J. Am.
Chem. Soc. 2007, 129, 4919.
1
958 9 J. AM. CHEM. SOC. 2009, 131, 1958–1971
10.1021/ja808292c CCC: $40.75  2009 American Chemical Society

Monomolecular Alkane Activation on Zeolites
A R T I C L E S
Table 1. Elemental Composition and Site and Structure Characterization of Zeolite Samples
a
a
b
c
c
d
d
zeolite
source
Si/Al ratio
Na/Al ratio
AlEF (%)
L
1
IR band area
L
2
IR band area
OH8-MR (%)
OH12-MR (%)
H-FER
H-MFI-1
H-MFI-2
Zeolyst
10.3
19
16.5
16.1
25
0.002
-
0.004
0.15
-
15
24
12
12
11
11
19
21
22
22
22
22
22
2.1
0.6
0.9
-
0.3
0.2
5.0
10.4
2.8
-
2.9
0.1
0.5
-
<0.1
<0.1
8.6
4.9
3.9
-
-
-
-
-
-
-
-
-
-
-
Chevron
Zeolyst
Zeolyst
Zeolyst
Zeolyst
Tosoh
Sud-Chemie
Zeolyst
Zeolyst
Zeolyst
Zeolyst
Zeolyst
H85Na15MFI-2
H-MFI-3
H-MFI-4
H-MOR-T
H-MOR-S
H-MOR-Z
40
8.9
-
-
-
0.001
0.001
0.001
0.17
0.27
0.41
0.55
78
60
56
36
27
20
13
22
40
44
64
73
80
87
10.1
10.0
10.0
10.0
10.0
10.0
H
H
H
H
83Na17MOR-Z
73Na27MOR-Z
59Na41MOR-Z
45Na55MOR-Z
-
-
-
-
-
-
a
Determined from elemental analysis (ICP-OES; Galbraith Laboratories). Extra-framework Al content (AlEF) determined from 27Al MAS NMR
b
c
-1
-1
spectra (details given in Section S.1 of the Supporting Information). Areas of L
1
(2224 cm ) and L
2
(2196 cm ) infrared bands for CO adsorbed at
3
1
(
23 K on Lewis acid sites (details shown in Section S.2 of the Supporting Information). Areas (× 10 ) are normalized to Si-O-Si overtone band areas
-1
d
8
2100-1750 cm ). Distribution of acid sites in MOR samples determined from infrared spectral band deconvolution.
sites present within 8-MR and 12-MR environments in MOR
zeolites. Strong effects of the local environment around
of ion pairs relevant for the late transition states prevalent in
monomolecular alkane reactions. Energy-entropy tradeoffs at
the transition state, mediated by local spatial constraints, are
essential contributors to the enzyme-like specificity of certain
locations within microporous solids for alkane activation on
Brønsted acid sites.
9
Brønsted acid sites are not evident in previous studies of
hydrocarbon catalysis, for which some studies have claimed no
1,2
effects of Al content and, by inference, of their location. Any
catalytic consequences of site location, however, are essential
to choose and synthesize inorganic microporous structures with
specific reactivity and selectivity.
2. Methods
Here, we report evidence for strong effects of local structure
on the catalytic cracking and dehydrogenation of alkanes on
acidic zeolites. These reactions can occur via bimolecular
2
.1. Catalyst Synthesis. MFI, FER, and MOR zeolite samples
+
(
Table 1) in their NH
4
form were treated in flowing dry air (2.5
3
-1 -1
cm g s , zero grade, Praxair) by increasing the temperature to
1
0,11
pathways involving carbenium ion chain cycles
or via
-1
+
+
7
4
73 K (0.0167 K s ) and holding for 4 h to convert NH to H .
monomolecular routes requiring pentacoordinated carbonium
ions at the transition state. Monomolecular routes prevail at low
H-zeolites were pelleted, crushed, and sieved to retain 180-250
µm (60-80 mesh) aggregates. Zeolite samples (∼14 g) were
1
2,13
+
alkane conversions and pressures
and form H
2
, smaller
exchanged with Na cations to varying extents using 0.5 L aqueous
alkanes, and the respective alkenes as primary products (pro-
solutions with different NaNO (99%, EMD Chemicals, 0.014-2.44
3
14
15,16
M) concentrations at 353 K for 12 h. The exchanged zeolite samples
were rinsed with 2 L of deionized water, isolated by filtration, and
treated in flowing dry air as described above. Si, Al, Na, and trace
metal contents (Table 1) were determined by inductively coupled
plasma optical emission spectroscopy (ICP-OES; Galbraith Labo-
ratories). The amounts of Al in framework and extra-framework
pane; n-butane
). Isotopic studies indicate that products of
monomolecular pathways initiate bimolecular routes mediated
17,18
by carbenium ions, which then propagate via hydride transfer.
Monomolecular alkane reactions are useful probes of the
catalytic consequences of confinement in acid catalysis, because
they form relatively unreactive primary products of cracking
27
locations (Table 1) were measured by Al MAS NMR spectros-
copy. Experimental methods and NMR spectra are shown in Section
S.1 of the Supporting Information.
(
e.g., CH from C
4
and C
2
H
4
3 8 3 6
H ) and dehydrogenation (e.g., C H
and H from C H ) events. Deviations from equimolar yields
2
3 8
provide a rigorous experimental kinetic marker for bimolecular
or secondary reactions.
2.2. Infrared Assessment of the Number and Location of
Intracrystalline Hydroxyl Groups and Lewis Acid Centers.
Infrared spectra were collected using a Nicolet NEXUS 670 infrared
spectrometer equipped with a Hg-Cd-Te (MCT) detector. Spectra
We interpret measured rate parameters here using thermo-
chemical cycles that rigorously define the contributions of
enthalpy and entropy to chemical reactions catalyzed by
Brønsted acids. This approach illustrates the essential role of
spatial constraints in the stabilization of adsorbed reactants and
were measured with 2 cm- resolution in the 4000-400 cm range
by averaging 64 scans. Self-supporting wafers (∼20-40 mg) were
sealed within a quartz vacuum infrared cell equipped with NaCl
1
-1
3
-1
windows. Samples were treated in flowing dry air (1.67 cm s ,
-
1
zero grade, Praxair) by heating to 723 K (0.033 K s ), holding
for 2 h, and then evacuating for at least 2 h at 723 K using a
diffusion pump (<0.01 Pa dynamic vacuum; Edwards E02) before
cooling to 303 K in vacuum and collecting spectra.
(
9) Bhan, A.; Iglesia, E. Acc. Chem. Res. 2008, 41, 559.
(
10) Greensfelder, B. S.; Voge, H. H.; Good, G. M. Ind. Eng. Chem. 1949,
4
1, 2573.
(
(
11) Thomas, C. L. Ind. Eng. Chem. 1949, 41, 2564.
Strong (L
1
) and weak (L ) Lewis acid centers were measured
2
12) Haag, W. O.; Dessau, R. M. Proceedings - International Congress on
Catalysis, 8th, Berlin, Verlag Chemie: Weinheim, 1984; Vol. 2, p 305.
13) Kotrel, S.; Kn o¨ zinger, H.; Gates, B. C. Microporous Mesoporous
Mater. 2000, 35-36, 11.
19
from infrared spectra of CO adsorbed at 123 K, and specifically
-
1
20
(
(
from CO bands at 2224 and 2196 cm , respectively. A constant
flow of liquid N was used to cool zeolites to 123 K before dosing
CO (UHP, Praxair) incrementally, without intervening evacuation;
spectra were collected 120 s after each CO dose. Integrated L and
2
14) Kwak, B. S.; Sachtler, W. M. H.; Haag, W. O. J. Catal. 1994, 149,
4
65.
1
(
(
(
15) Krannila, H.; Haag, W. O.; Gates, B. C. J. Catal. 1992, 135, 115.
16) Lercher, J. A.; van Santen, R. A.; Vinek, H. Catal. Lett. 1994, 27, 91.
17) Ivanova, I. I.; Pomakhina, E. B.; Rebrov, A. I.; Derouane, E. G. Topics
Catal. 1998, 6, 49.
(19) Cheung, P.; Bhan, A.; Sunley, G. J.; Law, D. J.; Iglesia, E. J. Catal.
2007, 245, 110.
(
18) Ivanova, I. I.; Rebrov, A. I.; Pomakhina, E. B.; Derouane, E. G. J.
Mol. Catal. A 1999, 141, 107.
(20) Benco, L.; Bucko, T.; Hafner, J.; Toulhoat, H. J. Phys. Chem. B 2004,
108, 13656.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009 1959

A R T I C L E S
Gounder and Iglesia
Table 2. Monomolecular Propane Cracking and Dehydrogenation Rate Constants (kmeas) and Cracking-to-Dehydrogenation (C/D) Rate
Ratios at 748 K and Measured Activation Energies (Emeas) and Entropies (∆Smeas) on Acidic Zeolites
3
+ -1 -1
-1
-1
-1 -1
k
meas (× 10 mol (mol H )
s
bar
)
E
meas (kJ mol
)
∆Smeas (J mol
K )
a
b
c
d
zeolite
cracking
dehyd
C/D ratio
cracking
dehyd
cracking
dehyd
H-MFI
H-FER
H-MOR-T
H-MOR-S
H-MOR-Z
2.0
6.2
2.0
1.3
1.4
2.1
3.2
3.0
1.9
2.2
0.9
2.0
0.7
0.7
0.7
158
157
160
167
160
200
195
189
192
198
-99
-91
-97
-93
-99
-54
-57
-71
-66
-56
a
Errors are ( 5 kJ mol-1
b
Errors are ( 7 kJ mol-1
c
Errors are ( 8 J mol-1 K-1
d
Errors are ( 10 J mol-1 K-1
.
.
.
.
L
2
band areas (normalized to those for Si-O-Si overtones;
zeolites (by up to factors of 4) and even among H-MOR samples
with similar Si/Al ratios but of different provenance (by up to
factors of 2). Measured activation energies for dehydrogenation
were consistently higher than for cracking (by 25-42 kJ mol )
3 8
on all samples, in contrast with activation energies for C H
-
1
2
100-1750 cm ) are proportional to the density of Lewis centers
(
per gram) in the sample and are shown in Table 1. The
characterization of Lewis acid centers by these infrared methods
is treated in more detail in Section S.2 of the Supporting
Information.
-1
dehydrogenation on H-MFI that vary widely among previous
2
.3. Catalytic Rates of Monomolecular Alkane Activation.
21-26
reports.
Measured activation entropies were also higher for
Steady-state alkane cracking and dehydrogenation rates were
measured in a tubular packed-bed quartz reactor (7.0 mm inner
diameter) with plug-flow hydrodynamics under differential condi-
tions (<2% conversion). Catalyst samples (0.02-0.2 g) were
supported on a coarse quartz frit, and the temperature was
maintained with a resistively heated three-zone furnace (Applied
Test Systems Series 3210). Each zone was controlled independently
by Watlow controllers (96 Series); catalyst temperatures were
recorded using K-type thermocouples contained within a thermowell
placed at the external surface of the quartz reactor.
dehydrogenation than for cracking, consistent with later transi-
tion states for dehydrogenation, in which the H-H bond is
2
7
nearly formed at the transition state. Rate parameters differ
among these zeolites because their transition states differ in their
energy and entropy relative to the gas-phase alkanes, as we
discuss later in the context of the relevant thermochemical
cycles; as a result, these rate parameters probe specific and subtle
effects of local environment on the structure and energy of these
transition states.
Catalysts were treated in a mixture of 5% O
2
in He (16.7 cm3
g-1 s , 99.999%, Praxair) at 803 K (0.0167 K s ) for 2 h before
catalytic measurements. The samples were then treated in He (16.7
-1
-1
Accepted pathways for monomolecular reactions of alkanes
+
-
(
B) on intrazeolite Brønsted acids (H Z ) are depicted sche-
3
-1 -1
matically in Scheme 1a. Alkanes adsorb from the extracrystalline
phase (B(g)) onto a Brønsted acid site within zeolite channels
(B(z)) in quasi-equilibrated and essentially nonactivated steps.
Transition state formalisms require that monomolecular rear-
rangements of adsorbed alkanes on Brønsted acid sites depend
cm g s , 99.999%, Praxair) for 0.5 h, while propane (10% C
9.999%, diluted in 5% Ar, 85% He, Praxair) or n-butane (10%
n-C 10, 99.99%, diluted in 5% Ar, 85% He, Praxair) reactants
3 8,
H
9
4
H
were sent via heated transfer lines held at 423 K to a gas
chromatograph (Agilent HP-6890GC) for calibration purposes.
Reactants and products were separated using GS-AL\KCl capillary
28
rigorously on the thermodynamic activity of reactants. These
(
0.530 mm ID × 50 m; Agilent) and HayeSep DB packed columns
treatments, shown in greater detail in Section S.4 of the
Supporting Information, give a rate expression of the form:
(
100-120 mesh, 10 ft.; Sigma-Aldrich); their respective effluents
were measured by flame ionization and thermal conductivity
detection.
†
γ
+
C
+
γ C
H Z H Z B
Bz
z
k T
Reactant pressures were changed by dilution with He (99.999%,
B
-∆G°
RgT
-
6
-4
-1
r )
exp
×
(1)
Praxair). Reactant flows were varied (10 -10 mol alkane g
h
(
)
γ†
-
1
s ) to probe primary and secondary pathways and any contributions
from bimolecular or secondary reactions. The absence of bimo-
lecular pathways and secondary reactions was confirmed by the
equimolar ratios of cracking products (1.0 ( 0.1) measured for
Monomolecular alkane activation pathways require high
temperatures, which favor low intrazeolite concentrations (CBz
)
+
and H sites that remain essentially unoccupied during catalysis
(evidence in Section S.3 of the Supporting Information).
Calorimetric, gravimetric, and infrared studies on various acidic
3 8 2 4 4 4 3 6 2 4 2 6
C H (C H /CH ) and n-C H10 (C H /CH4, C H /C H ) reactants,
which did not depend on space velocity, and by the absence of
hydrocarbons larger than the respective alkane reactants. Activation
energies and pre-exponential factors were obtained from rate
constants measured as a function of temperature (718-778 K).
Rates and selectivities measured after ∼100 ks on stream were
similar (within 5%) to those at the start of each experiment on all
catalyst samples. Transport corruptions of measured rates were ruled
out using Mears criteria; the detailed analysis is shown in Section
S.3 of the Supporting Information.
zeolites have shown that C
onto Brønsted acid sites with adsorption enthalpies that remain
constant up to saturation coverages at all H sites.
3
6
-C n-alkanes adsorb specifically
+
29-31
Thus,
(21) Narbeshuber, T. F.; Vinek, H.; Lercher, J. A. J. Catal. 1995, 157,
388.
(
(
(
22) Narbeshuber, T. F.; Brait, A.; Seshan, K.; Lercher, J. A. J. Catal.
1
997, 172, 127.
23) Wang, X.; Carabineiro, H.; Lemos, F.; Lemos, M. A. N. D. A.; Ribiero,
F. R. J. Mol. Catal. A 2004, 216, 131.
3
. Results and Discussion
.1. Assessment of Kinetic Parameters for Monomolecular
24) Bandiera, J.; Dufaux, M.; Ben Taarit, Y. Appl. Catal., A 1997, 148,
283.
3
(
(
25) Bandiera, J.; Ben Taarit, Y. Appl. Catal. 1990, 62, 309.
26) Xu, B.; Sievers, C.; Hong, S. B.; Prins, R.; van Bokhoven, J. A. J.
Catal. 2006, 244, 163.
Propane Cracking and Dehydrogenation Using Transition
State Theory and Thermochemical Cycles. Table 2 shows rate
constants (per H ) for propane cracking and dehydrogenation
and cracking-to-dehydrogenation rate ratios at 748 K on H-MFI,
H-FER, and H-MOR, together with activation energies and
entropies calculated from the temperature dependence of these
rate constants. Monomolecular rate constants differ among these
+
(27) Zheng, X.; Blowers, P. J. Phys. Chem. A 2005, 109, 10734.
(28) Madon, R. J.; Iglesia, E. J. Mol. Catal. A Chem. 2000, 163, 189.
(
29) Eder, F.; Stockenhuber, M.; Lercher, J. A. J. Phys. Chem. B 1997,
1
01, 5414.
(
30) Eder, F.; Lercher, J. A. J. Phys. Chem. B 1997, 101, 1273.
(31) Eder, F.; Lercher, J. A. Zeolites 1997, 18, 75.
1
960 J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009

Monomolecular Alkane Activation on Zeolites
A R T I C L E S
Scheme 1. (a) Reaction Sequence for Monomolecular Alkane
Measured activation energies and pre-exponential factors are
then given by:
Activation on Brønsted Acid Sites Located within Zeolite Channels
+
-
a
(
H Z ). (b) Thermochemical Cycle for Monomolecular Reactions
b
of Alkanes in Zeolite Channels
Emeas ) Eint + ∆Hads
(6)
(7)
(8)
ln(Ameas) ) ln(A ) + (∆Sads/R)
int
∆
Smeas ) ∆Sint + ∆Sads
where ∆Hads and ∆Sads are the enthalpy and entropy of
adsorption, respectively, and Emeas and Ameas (∆Smeas) are
measured rate parameters referenced to gas phase alkanes. Eint
and Aint (∆Sint) are intrinsic rate parameters for elementary
reactions of adsorbed alkanes relating the properties of adsorbed
21,33-36
reactants to the corresponding transition states.
the measured entropy of activation as:
We define
∆
Smeas ) R[ln(Ameas) - ln(k T/h)]
B
(9)
where Ameas is rigorously normalized by the number of acid sites
and by the number of bonds available for each type of reaction.
These measured activation entropies reflect a mathematical
manipulation that simply redacts Ameas in terms of thermody-
namic properties of reactants and transition states.
a
b
B denotes a reactant base; P denotes products. Measured activation
energies (Emeas) are related to zeolite deprotonation enthalpies (DPE), gas-
phase proton affinities (PA) and ion-pair stabilization energies within zeolite
channels (Estab). The dependence of intrinsic activation energies (Eint) on
adsorption enthalpies (∆Hads) and Emeas (eq 6) is also shown.
Table 3 shows intrinsic activation energies and entropies for
monomolecular propane cracking and dehydrogenation on
H-MFI, H-FER, and H-MOR; these values were estimated from
measured rate constants (Table 2), using eqs 6-9 and previously
2
9,30
reported adsorption enthalpies and entropies.
activation barriers for monomolecular C cracking were
similar on H-MFI, H-FER, and H-MOR (201-208 kJ mol ;
Table 3), in agreement with previous data for C cracking
Their magnitude and
insensitivity to zeolite structure also agree with previous reports
Intrinsic
3 8
H
Langmuir adsorption models accurately describe the siting of
alkanes onto Brønsted acid sites of uniform binding properties.
The concentrations (per active site) (C
-1
3 8
H
i
) of adsorbed alkanes
-1
21-26
on H-MFI (187-200 kJ mol ).
(
B
z
) and transition states (†) are related to their real (P
B
) or
hypothetical (P
†
) gas-phase pressures by:
2
6
for C
3 8
H cracking on H-MFI, H-MOR, H-BEA, and H-FAU
2
1,33,37
KiP
i
and for the cracking of larger C
3
-C20 n-alkanes on H-MFI.
C )
(2)
i
1
+ K P + K P
† †
Energies of monomolecular cracking transition states (reflected
in Emeas) relative to those of adsorbed reactants are independent
of chain size and of local channel environments, which appear
to stabilize adsorbed reactants and transition states identically.
As a result, intrinsic activation barriers (Eint) for cracking
reactions on different zeolite structures are similar within the
accuracy of the rate and adsorption data used to estimate them
B
B
in which the K
dynamic treatments of Langmuir surfaces give activity coef-
) for each species of the form (derived in detail in
i
terms are their adsorption constants. Thermo-
ficients (γ
i
Section S.4 of the Supporting Information):
γ ) a °(1 + K P + K P )
(3)
i
i
B
B
† †
(
Table 3), as long as adsorption and rate measurements probe
i
where a ° is the thermodynamic activity at the reference state;
activity coefficients (eq 3) include a configurational entropy term
the same intrachannel environment.
Intrinsic activation energies for monomolecular propane
dehydrogenation were also similar on H-MFI, H-FER, and
that reflects the number of configurations of N molecules
3
2
adsorbed onto M adsorption sites. Reaction rates are strictly
proportional to intrazeolite alkane concentrations (CBz) when
activity coefficients for alkane reactants (γBz) and transition
-1
H-MOR (229-245 kJ mol ; Table 3) and larger than for
-1
cracking (by 25-42 kJ mol ). These data contrast previous
†
states (γ ) cancel in eq 1; this occurs in this case, irrespective
reports of intrinsic dehydrogenation barriers that varied widely
-1
21-26
of their structure and solvation, because both species reside on
the same site and therefore have identical activity coefficients.
for C
3 8
H among various H-MFI studies (135-205 kJ mol )
-1
21-23
and which ranged from smaller barriers (by 60 kJ mol )
than for cracking to similar values
values than cracking barriers. Density functional theory (DFT)
studies on small T3 zeolite clusters concluded that transition
states are ∼60 kJ mol higher in energy for dehydrogenation
than cracking of propane. Figure 1 compares the temperature
dependence of propane cracking-to-dehydrogenation rate ratios
on H-MFI with DFT estimates and previous data.
activation barriers were reported in DFT studies, so we have
+
24,25
Moreover, CH+Z and γH+Z approach unity because H sites are
or even slightly larger
26
predominantly unoccupied during monomolecular reactions, for
which the rates given by eq 1 become:
2
7
-1
†
†
k T
B
∆S°
Rg
-∆H°
RgT
r )
exp
× exp
CBz
(4)
(
)
(
)
h
27
21,24,26
Intrazeolite concentrations (CBz) become proportional to
Only
2
7
external pressures (P
B
) and to the adsorption constants (K
B
) in
the low-coverage limit (eq 2), where reaction rates (eq 4) are
given by:
(33) Haag, W. O. Stud. Surf. Sci. Catal. 1994, 84, 1375.
(
34) van Bokhoven, J. A.; Williams, B. A.; Ji, W.; Koningsberger, D. C.;
Kung, H. H.; Miller, J. T. J. Catal. 2004, 224, 50.
r ) k K P
B B
(5)
int
(35) Babitz, S. M.; Williams, B. A.; Miller, J. T.; Snurr, R. Q.; Haag, W. O.;
Kung, H. Appl. Catal. A Gen. 1999, 179, 71.
(
32) Denbigh, K. The Principles of Chemical Equilibrium, 4th ed.;
(36) Bhan, A.; Gounder, R.; Macht, J.; Iglesia, E. J. Catal. 2008, 253, 221.
(37) Wei, J. Chem. Eng. Sci. 1996, 51, 2995.
Cambridge University Press: Cambridge, 1981.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009 1961

A R T I C L E S
Gounder and Iglesia
Table 3. Intrinsic Activation Energies (Eint) and Entropies (∆Sint) for Monomolecular Propane Cracking and Dehydrogenation on Acidic
Zeolites
-1
-1 -1
Eint (kJ mol
)
∆Sint (J mol
K
)
∆Hads (kJ mol-
1
)
∆Sads (J mol
-1 -1
K
)
cracking
a
dehyd
b
cracking
c
dehyd
d
zeolite
e
e
H-MFI
H-FER
H-MOR-T
H-MOR-S
H-MOR-Z
-45f
-102f
-109
204
205
202
208
201
245
244
229
234
239
3
17
47
51
13
18
28
-49
g
g
-41
-85
-12
g
g
-41
-85
-85
-8
g
g
-41
-14
a
Errors are ( 7 kJ mol-1
b
Errors are ( 9 kJ mol-1. c Errors are ( 10 J mol-1 K-1
. . Adsorption data reported in ref
Errors are ( 13 J mol-1 K-1
d
e
.
f
g
2
9. Adsorption data reported in ref 30. Adsorption data reported only for 12-MR channels of H-MOR in ref 29.
Scheme 1b. In the context of this thermochemical cycle
formalism, the following expressions relate measured and
intrinsic activation barriers to properties of the acid, the
molecules, and their ion pair at the transition state:
Emeas ) DPE + PA + Estab
(10)
(11)
Eint ) DPE + PA + Estab - ∆Hads
Deprotonation requires electron transfer from H atoms to the
zeolite lattice (Z), cleavage of H-Z bonds, and separation of
+
-
the H and Z fragments to noninteracting distances. DPE
values thus reflect contributions from homolytic H-Z bond
dissociation energies (BDE), the ionization energy (IE) of H
atoms, electron affinities (EA) of lattice oxygen atoms, and
+
-
electrostatic interactions that stabilize the ion pair (H -Z ) as
separation proceeds. Any variation in DPE values among zeolite
structures reflects predominantly the effects of the local environ-
ment on the electrostatic stabilization of the ion pair, because
BDE, IE, and EA values are likely to depend weakly on spatial
constraints. Hybrid quantum mechanical and interatomic po-
tential methods indicate that DPE values for Brønsted acid sites
Figure 1. Temperature dependence of monomolecular propane cracking-
to-dehydrogenation rate ratios on H-MFI. Differences in dehydrogenation
-1
4
at different T-sites in H-MFI (1229-1240 kJ mol ) and for
(Ede) and cracking (Ecr) activation energies are reflected in the slope. Symbols
the most stable Al location among H-MFI, H-FER, H-MOR
denote experimentally measured rate ratios in this study (b) and rate ratios
extrapolated from data reported in previous experimental studies (ref 21:
-1 5,40
zeolites (1218-1235 kJ mol )
are in fact similar. These
similar deprotonation enthalpies reflect the predominant con-
tributions from long-range interactions with the zeolite lattice
to the electrostatic stabilization of protons, which overwhelm
other contributions resulting from subtle variations in channel
geometry.
9
, ref 24: (, ref 26: 2) and in a theoretical study (ref 27: s).
used the pre-exponential factors (shown as ∆Smeas in Table 2)
measured in this study to estimate cracking-to-dehydrogenation
rate ratios from DFT activation barriers. All previous data show
weaker temperature effects on rate ratios than reported here and
predicted by theory (Figure 1). As we discuss below, higher
barriers for dehydrogenation relative to cracking are also
expected from the relative stabilities of protonated C-H and
Proton affinities reflect the energy required to protonate
+
neutral gas-phase molecules (B) to form BH complexes, which
differ for cracking and dehydrogenation events because the
C-C-H and C-H-H three-center/two-electron complexes
formed by protonation of C-C and C-H bonds in alkanes differ
3 8
C-C bonds in C H to form gas-phase structures resembling
4
1,42
those in the respective transition states for these two reactions.
We use Born-Haber thermochemical cycles (Scheme 1b) to
interpret the stabilization of reactants and transition states in
in energy. Theoretical estimates
of gas-phase carbonium ion
forms
3 8
energies indicate that protonation of C-H bonds in C H
C-H-H three-center/two-electron species that are less stable
3
8,39
acid catalysis by zeolites
and the contributions of intrac-
-1
by ∼20-40 kJ mol than the C-C-H analogues formed via
hannel environments and transition state structure and charge
to activation barriers. Measured activation energies (Emeas) reflect
the stability of carbonium ion-like transition states stabilized
within channels relative to gaseous reactants (B(g)). These
barriers depend on the deprotonation enthalpies (DPE) of
Brønsted acid sites, on proton affinities (PA) of gaseous
3 8
protonation of C-C bonds in C H .
Transition state energies reflect both electrostatic stabilization
+
-
of ion pairs that form as BH and Z interact and van der Waals
stabilization of the organic cation in the transition state complex.
Alkane adsorption enthalpies reflect predominantly van der
Waals stabilization by framework oxygens and, to a lesser
reactants, and on transition state stabilization energies (Estab
)
+
extent, any induced dipole interactions with H , where the latter
(
Scheme 1b). The relations between measured activation ener-
do not depend on the size of alkane reactants or zeolite
gies (Emeas), reactant adsorption enthalpies (∆Hads), and intrinsic
activation barriers (Eint) described by eq 6 are also depicted in
(
40) Nieminen, V.; Sierka, M.; Murzin, D. Y.; Sauer, J. J. Catal. 2005,
2
31, 393.
(
(
38) Aronson, M. T.; Gorte, R. J.; Farneth, W. E. J. Catal. 1986, 98, 434.
39) Macht, J.; Janik, M. J.; Neurock, M.; Iglesia, E. J. Am. Chem. Soc.
(41) Esteves, P. M.; Mota, C. J. A.; Ram ´ı rez-Sol ´ı s, A.; Hern a´ ndez-
Lamoneda, R. J. Am. Chem. Soc. 1998, 120, 3213.
2
008, 130, 10369.
(42) Collins, S. J.; O’Malley, P. J. Chem. Phys. Lett. 1994, 228, 246.
1
962 J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009

Monomolecular Alkane Activation on Zeolites
A R T I C L E S
Scheme 2. Relations between Measured Activation Energies
2
9-31
channels.
The positive charge (+0.9e) at the alkyl fragment
(Emeas), Adsorption Enthalpies (∆Hads), and Intrinsic Activation
in late monomolecular transition states is stabilized by long-
Energies (Eint)aAre Shown Schematically for Two Different Channel
6
,7,43,44
range electrostatic interactions with the zeolite framework,
rendering subtle effects of channel size inconsequential for Estab
Environments
,
the term that accounts for electrostatic stabilization in thermo-
chemical cycles; similar long-range electrostatic considerations
account for the weak dependence of DPE values on Brønsted
4
,5,40
acid site location or zeolite environment.
der Waals stabilization of adsorbed alkanes depends strongly
In contrast, van
45
on channel size and structure; by inference, such stabilization
is essential also for monomolecular transition states that
resemble in structure their adsorbed alkane precursors. As a
result, we conclude that differences in transition state stability
(
E
stab) among zeolite locations or structures are almost entirely
compensated by commensurate differences in reactant stability
∆Hads), as evidenced by intrinsic activation barriers (Eint) that
do not change with variations in the size or structure of zeolite
(
2
1,33,37
channels (Table 3) or n-alkane reactants.
The intrinsic activation barriers for alkane cracking and
dehydrogenation on a given zeolite (or a given location within
a zeolite) differ from each other primarily in the respective
affinities for protonation of an alkane at C-C or C-H bonds,
because DPE and ∆Hads are independent of the specific reaction
path for a given alkane and acid site (eq 11). The similar
magnitude (+0.9e) and extent of charge delocalization in the
organic fragment at cracking and dehydrogenation transition
a
Adsorbed reactants and transition states in location a are stabilized to
a greater extent than in location b.
2 5 4
C H and CH groups at the earlier transition states involved in
cracking reactions.
6
,7,43,44
Scheme 2 depicts a hypothetical reaction coordinate diagram
for acid sites within two environments with different spatial
constraints, as a result of differences in either the identity of
the zeolites or the siting of the acid within a specific location
in a given zeolite structure. The more constrained environment
in Scheme 2a stabilizes both adsorbed molecules and transition
states more effectively than the more open structure in Scheme
2b through concerted interactions that influence the adsorption
enthalpies of the reactants and the stabilization energies of the
states
in Estab) must also be insensitive to reaction path. The differences
in intrinsic activation barriers for C dehydrogenation and
suggest that electrostatic stabilization (contained
3 8
H
cracking (estimated from rate data) for a given zeolite (25-42
-1
kJ mol ; Table 3) must then reflect the calculated energy
differences between three-center/two-electron C-H-H bonds
(
dehydrogenation) and C-C-H bonds (cracking) in the gas-
-1
41,42
phase (∼20-40 kJ mol ),
consistent with the thermo-
chemical cycle in Scheme 1b and eq 11.
45,47
transition states,
assuming their complete containment within
Thermochemical cycles that relate transition state energies
to either gas-phase or adsorbed reactants (Scheme 1b) can also
be used to describe the contributions of various catalyst and
molecule properties to activation entropies. Measured activation
entropies for cracking and dehydrogenation reflect differences
in entropies between their respective transition states, because
both are referenced to the same gas-phase reactant. Measured
activation entropies for propane dehydrogenation are ∼40 J
the smaller environment. Thus, intrinsic activation barriers for
cracking (or dehydrogenation) are more weakly affected by local
environment than measured activation energies, which reflect
differences between transition states and gas-phase reactants.
Measured turnover rates therefore depend on transition state
energies and entropies, which depend, in turn, on the channel
environment that stabilizes transition states but not reactants in
the gas phase external to the zeolite.
H-MOR contains Al atoms that reside at four unique T-sites
and corresponding Brønsted acid sites that reside within two
distinct channel environments: 8-MR side pockets or 12-MR
main channels. The distribution of Brønsted acid sites among
-1
-1
mol
K
larger than for cracking on H-MFI, H-FER, and
H-MOR (Table 2), consistent with later transition states for
dehydrogenation, in which H-H bonds are nearly formed (0.079
2
7
nm vs 0.074 nm in H
2
(g)). The late transition states prevalent
2
7,43,44
in monomolecular alkane reactions
lead to hindered
8-MR and 12-MR locations can be measured from the frequency
4
6
rotations and rocking vibrations that increase entropy upon
formation of the transition state from adsorbed reactants. As a
result, intrinsic rate constants for monomolecular n-hexane
cracking are ∼50 times larger than for propane cracking on
and intensity of their antisymmetric vibrations in infrared spectra
-1
8
(
3550-3650 cm ). Such methods are unable to establish the
specific location of OH groups in H-MFI, however, because of
the greater diversity of T-sites and because intrachannel
environments are more uniform in local structure in MFI than
3
7
H-MFI, even though both reactions show similar intrinsic
activation barriers. We surmise that such late transition states
for dehydrogenation allow greater rotational and vibrational
48
in MOR. All acid sites in H-MFI reside within 10-MR
channels or their intersection, precluding the use of titrants of
varying size to probe local environment. In light of these
considerations, we choose to examine the catalytic consequences
of Al siting in the bimodal distribution of channel environments
3 7 2
freedom for C H and H fragments than for the corresponding
(
43) Zygmunt, S. A.; Curtiss, L. A.; Zapol, P.; Iton, L. E. J. Phys. Chem.
B 2004, 104, 1944.
(
(
44) Frash, M. V.; van Santen, R. A. Topics Catal. 1999, 9, 191.
45) Savitz, S.; Siperstein, F.; Gorte, R. J.; Myers, A. L. J. Phys. Chem. B
(47) Hunger, B.; Heuchel, M.; Clark, L. A.; Snurr, R. Q. J. Phys. Chem B
2002, 106, 3882.
1
998, 102, 6865.
(
46) Benson, S. W. Thermochemical Kinetics; Wiley: New York, 1976.
(48) Busca, G. Chem. ReV. 2007, 107, 5366.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009 1963

A R T I C L E S
Gounder and Iglesia
Table 4. Monomolecular Propane Cracking and Dehydrogenation Rate Constants (kmeas) and Cracking-to-Dehydrogenation Rate Ratios
+
Measured at 748 K and Measured Activation Energies (Emeas) and Entropies (∆Smeas) on Partially Na Exchanged MOR-Z Samples
3
+ -1 -1
-1
-1
-1 -1
k
meas (× 10 mol (mol H )
cracking
s
bar
)
E
meas (kJ mol
)
∆Smeas (J mol
K )
a
b
c
d
zeolite
dehyd
C/D ratio
cracking
dehyd
cracking
dehyd
H
H
H
H
H
100Na
83Na17MOR-Z
73Na27MOR-Z
59Na41MOR-Z
45Na55MOR-Z
0
MOR-Z
1.4
1.2
1.1
0.9
0.6
2.2
1.2
0.8
0.6
0.3
0.7
1.0
1.4
1.5
1.7
160
160
157
155
153
198
201
201
196
198
-99
-101
-105
-110
-115
-56
-58
-60
-69
-72
a
Errors are ( 5 kJ mol-1
b
Errors are ( 7 kJ mol-1
c
Errors are ( 8 J mol-1 K-1
d
Errors are ( 10 J mol-1 K-1
.
.
.
.
of H-MOR. Al siting was previously shown to strongly influence
the stability of charged transition states involved in CO insertion
8
into methyl groups but not symmetric and essentially neutral
9
transition states involved in exchange reactions.
3
.2. Catalytic Consequences of 8-MR and 12-MR Environ-
ments within MOR in Monomolecular Cracking and Dehydro-
genation of Propane. Monomolecular propane cracking and
+
dehydrogenation turnover rates (per H ) differ by factors of
∼
2 (Table 2) on H-MOR zeolites with similar Al content. This
range of reactivity does not reflect differences in the number of
extraframework Al species or Lewis acid centers among these
samples, as shown by the NMR and IR data in Sections S.1
and S.2, respectively, of the Supporting Information. Table 4
shows propane cracking and dehydrogenation rate constants (per
+
total H ) and cracking-to-dehydrogenation rate ratios at 748 K
on H-MOR samples partially exchanged with Na , together with
+
measured activation energies and entropies. Cracking and
+
dehydrogenation rate constants (per residual H ) decreased, and
+
their rate ratio increased as H species within H-MOR were
+
selectively replaced by Na . Measured activation entropies for
Figure 2. Infrared spectra (s) of (a) H-MOR-T (OH in 8-MR ) 78 (
5%) and (b) H-MOR-S (OH in 8-MR ) 60 ( 5%). Deconvoluted bands
(--) for 8-MR and 12-MR OH groups are shown along with their respective
band centers.
both cracking and dehydrogenation (eq 9; from rate constants
normalized by residual H sites) decreased monotonically with
the extent of Na exchange, but measured activation energies
+
+
remained essentially constant.
The number of OH groups within 8-MR and 12-MR environ-
ments in H-MOR was measured from the intensity of antisym-
-1
metric OH stretches (∼3550-3650 cm ) using reported
8
methods. Singular value decomposition was used to extract
principal component spectra for OH bands in 8-MR and 12-
-
1
-1
MR locations centered at 3592 cm and 3611 cm , respec-
8
tively. These values are similar to those reported earlier on
-1
-1 49-51
H-MOR (8-MR: ∼3590 cm ; 12-MR: ∼3610 cm )
and
to those for residual OH groups after selective titration of 12-
8
,49,50
MR sites by pyridine.
The contribution from each pure
component band was determined by least-squares regression
methods that allowed each pure component band center to shift
up to 3 cm- and assumed similar molar extinction coefficients
for OH groups in the two locations. The deconvoluted infrared
spectra for H-MOR-T and H-MOR-S are shown in Figure 2
1
+
8
and elsewhere for partially Na -exchanged H-MOR-Z samples.
The number of OH groups within 8-MR and 12-MR locations
is listed in Table 1 for all MOR samples used in this study.
+
Infrared spectra showed that Na cations selectively replace
8
,49,52
protons within constrained 8-MR side pockets in MOR
Figure 3), because more electropositive alkali cations appear
Figure 3. Number of residual protons (per Al) in 8-MR pockets (b) and
+
(
12-MR channels (2) of MOR-Z samples with varying Na content (per
8
Al) determined by infrared spectral band deconvolution.
(
49) Makarova, M. A.; Wilson, A. E.; van Liemt, B. J.; Mesters, C.; de
Winter, A. W.; Williams, C. J. Catal. 1997, 172, 170.
to be stabilized more effectively than H+ by the stronger
(
(
50) Datka, J.; Gil, B.; Kubacka, A. Zeolites 1997, 18, 245.
dispersive forces prevalent within such smaller environments.
51) Maache, M.; Janin, A.; Lavalley, J. C.; Benazzi, E. Zeolites 1995, 15,
+
Rate constants for C
H
3 8
dehydrogenation (per residual H )
5
07.
52) Veefkind, V. A.; Smidt, M. L.; Lercher, J. A. Appl. Catal., A 2000,
94, 319.
+
decreased even more strongly with Na exchange than for
cracking (Table 4), reflecting a greater specificity of 8-MR
(
1
1
964 J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009

Monomolecular Alkane Activation on Zeolites
A R T I C L E S
Table 5. Monomolecular Propane Cracking and Dehydrogenation Rate Constants (kmeas) at 748 K within 8-MR and 12-MR Locations of
a
MOR and Measured Activation Energies (Emeas) and Entropies (∆Smeas
)
b
3
+ -1 -1
-1
-1
-1 -1
reaction (location)
k
meas (× 10 mol (mol H )
s
bar
)
Emeas (kJ mol
)
∆Smeas (J mol
K )
cracking (8-MR)
cracking (12-MR)
dehydrogenation (8-MR)
dehydrogenation (12-MR)
2.0 ( 0.5
0.7 ( 0.4
3.2 ( 0.7
164 ( 5
151 ( 5
197 ( 7
-91 ( 9
-117 ( 14
-54 ( 11
c
c
c
n.d.
n.d.
n.d.
a
b
c
Uncertainties in regressed rate parameters are reported as twice the standard error. Rate parameters determined by least-squares regression. n.d.,
not detected.
pockets for dehydrogenation reactions compared with 12-MR
channels. Measured activation entropies became more negative
for both cracking and dehydrogenation as Na+ replaced H
within 8-MR side pockets (Table 4), demonstrating that the more
constrained environment provided by such pockets actually
stabilizes looser transition states than the larger 12-MR main
channels in MOR for both reactions.
+
We can extract from these data the specific contributions of
OH groups within 8-MR and 12-MR locations in H-MOR by
expressing measured rate constants for C H cracking and
3 8
dehydrogenation as linear combinations of rate constants in the
two locations:
k ) k8-MRX8-MR + k12-MRX12-MR
(12)
Here, X8-MR and X12-MR represent the fraction of all OH groups
located within 8-MR and 12-MR locations, respectively, and
k8-MR and k12-MR the corresponding monomolecular rate constants
in each environment. The latter were determined by regressing
overall rate constants on seven MOR samples with different
acid site densities and location; the details of the regression
analysis are shown in Section S.5 of the Supporting Information.
Figure 4. Propane dehydrogenation rates (per gram) at 748 K in H-MOR
and Na -exchanged MOR samples (MOR-Z: b MOR-S: 9, MOR-T: 2)
plotted against the density of Brønsted acid sites (per gram) in the 8-MR
pockets of MOR (closed symbols) and 12-MR channels of MOR (open
symbols). Inset shows propane dehydrogenation rates (per gram) plotted
against the total density of Brønsted sites (per gram).
+
Measured rate constants (shown at 748 K in Table 5) for C
cracking were larger (by factors of ∼3) in 8-MR than in 12-
MR environments; C dehydrogenation rates on OH groups
3 8
H
3 8
H
within 12-MR channels were undetectable within the accuracy
of these measurements.
Brønsted acid sites within the more constrained 8-MR side
pockets in H-MOR catalyze monomolecular cracking and
transition states. Propane cracking rates (per gram) were also
+
proportional to the number of H sites present within 8-MR
dehydrogenation reactions of C
sites of similar acid strength within the less constrained 12-
MR channels. Propane dehydrogenation rates (per gram) are
H
3 8
preferentially (Table 5) over
environments (per gram), but detectable cracking contributions
from OH groups in 12-MR channels cause a nonzero intercept
(Figure 5). The rigorous subtraction of the 12-MR OH contribu-
tions (the second term in eq 12) from measured rates led to a
strictly linear dependence of residual rates on the number of
OH groups within 8-MR pockets and to a zero intercept, as
shown by the dashed line in Figure 5.
These data demonstrate that propane cracking and dehydro-
genation occur preferentially within 8-MR pockets at 718-778
K on H-MOR. Yet, infrared bands of OH groups in 8-MR
5
3
+
strictly proportional to the number of residual H within 8-MR
environments (per gram) and do not depend on the number of
residual H+ within 12-MR channels (per gram) (Figure 4),
consistent with the exclusive reactivity of Brønsted acid sites
of OH groups within 8-MR pockets at the conditions of these
experiments (Table 5). Such remarkable specificity demonstrates
the essential role of local environment on the enthalpic or
entropic stabilization of dehydrogenation transition states. These
data also illustrate how reactivity, as reflected here in turnover
rates, varies significantly among the diverse T-site locations even
within a given zeolite structure. These trends resemble those
reported recently for CO insertion into adsorbed methyl groups
derived from dimethyl ether, which formed acetyl groups (and
ultimately methyl acetate after subsequent methoxylation with
dimethyl ether) only on sites contained within 8-MR environ-
3 8
pockets of H-MOR were not perturbed by contact with C H at
2
9
near ambient temperatures (323 K), suggesting that propane
molecules have restricted access to or weaker interactions with
1
3
54
8-MR compared with 12-MR OH groups. C NMR spectra
5
5
3 8
and theoretical simulations, however, have shown that C H
adsorbs preferentially within 8-MR H-FER channels of similar
size at ambient temperatures. Such specificity for adsorption
within 12-MR channels in H-MOR reflects stronger dispersive
forces within constrained environments that can fully contain
8
ments in H-MOR and H-FER.
4
5
The systematic increase in cracking-to-dehydrogenation rate
the adsorbate molecules and the consequently more negative
+
+
ratios (0.7 to 1.7; Table 4) as Na replaced H within 8-MR
environments reflects the stronger effects of local environment
on dehydrogenation transition states compared with cracking
(54) van Well, W. J. M.; Cottin, X.; de Haan, J. W.; Smit, B.; Nivarthy,
G.; Lercher, J. A.; van Hooff, J. H. C.; van Santen, R. A. J. Phys.
Chem. B 1998, 102, 3945.
(
55) van Well, W. J. M.; Cottin, X.; Smit, B.; van Hooff, J. H. C.; van
(
53) Br a¨ ndle, M.; Sauer, J. J. Am. Chem. Soc. 1998, 120, 1556.
Santen, R. A. J. Phys. Chem. B 1998, 102, 3952.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009 1965

A R T I C L E S
Gounder and Iglesia
Figure 5. Measured propane cracking rates (per gram) (closed symbols)
and residual cracking rates after correction for the contribution from 12-
MR sites (per gram) (open symbols) at 748 K in H-MOR and Na -
exchanged MOR samples (MOR-Z: b MOR-S: 9, MOR-T: 2) plotted
against the density of Brønsted acid sites (per gram) in 8-MR pockets. Inset
shows total propane cracking rates (per gram) plotted against the total density
of Brønsted sites (per gram).
Figure 6. Measured rate constants (kmeas) for monomolecular propane
cracking in 8-MR pockets (b) and 12-MR channels (9) and dehydrogenation
in 8-MR pockets (2) determined from least-squares regression analysis.
Dehydrogenation rate constants were not detected in 12-MR channels. Error
bars are shown as twice the standard error in regressed rate constants.
+
3 8
C H adsorption enthalpies on 8-MR pockets are not available,
adsorption enthalpies that reflect more effective van der Waals
but we can extend arguments and data indicating that intrinsic
activation barriers are independent of zeolite structure or location
(because they lead to equivalent stabilization of adsorbed
precursors and transition states) to estimate such adsorption
enthalpies from measured activation energies on 8-MR OH
groups. Adsorption enthalpies for C H within 8-MR H-MOR
2
9-31
contacts with framework oxygens.
Correlations that relate
adsorption enthalpies to the effective channel diameters for
4
5
different zeolites predict that C
3
H
8
would indeed adsorb more
strongly (by ∼12 kJ mol ) if fully contained within channels
.41 nm in diameter, such as those in 8-MR MOR pockets, than
within the larger channels (0.67 nm) formed by 12-MR
structures in MOR. C
-1
0
3
8
-
1
pockets would then be -30 kJ mol in order for intrinsic
H
3 8
molecules (∼0.65 nm in length)
cracking activation barriers to be the same as those determined
-1
extend, however, beyond the depth of the 8-MR pockets (∼0.37
on 12-MR H-MOR channels (∼190 kJ mol ). These data and
nm) and must protrude into connecting 12-MR channels
arguments lead us to conclude that C H adsorption is much
3
8
(
procedures used for estimating channel and molecular dimen-
weaker on 8-MR pockets in MOR than predicted based on
-1
sions are described in Sections S.6 and S.7, respectively, of the
Supporting Information). These dimensions preclude van der
Waals interactions between the entire propane molecule and the
channel size considerations (-53 kJ mol ; Section S.6 of the
Supporting Information). Thus, we conclude that both adsorbed
C H molecules and their monomolecular transition states are
3
8
framework oxygens in 8-MR pockets. As a result, C
3
H
8
adsorbs
only partially contained within shallow 8-MR pockets and
protrude into neighboring 12-MR channels. These data and
conclusions are consistent with measured transition state
entropies (∆Smeas), which are significantly higher on 8-MR than
on 12-MR acid sites (Table 5).
We conclude that transition states are less stable, in terms of
enthalpy, within 8-MR pockets, which can confine only in part
cracking and dehydrogenation transition states, than within 12-
MR channels, which provide more effective van der Waals
stabilization for both the transition states and the adsorbed
molecules. We note that partial containment destabilizes species
by decreasing van der Waals contacts with parts of the
molecules, but such contacts are not essential for electrostatic
stabilization of the positive charge (+0.9e) at the three-atom/
two-electron center in the transition state; in late transition states
for monomolecular alkane reactions, the positive charge is
localized at the carbon atom in the alkyl fragment left behind
preferentially within larger 12-MR channels, which interact with
the entire molecule, albeit more weakly than 8-MR channels
able to contain the entire molecule.
Regressed rate constants on acid sites within 8-MR and 12-
MR locations are shown as a function of temperature in Figure
6
; these data were used to estimate activation energies and
entropies within these two distinct MOR environments (Table
5
8
). Cracking turnover rates were larger on sites present within
-MR pockets than on acid sites of similar strength within 12-
MR channels, even though measured activation energies were
actually higher on 8-MR OH groups (by 13 kJ mol ),
apparently because of looser transition states with higher
-1
entropies (Table 5). The adsorption enthalpy of C
3
H
8
on 12-
-1
MR MOR channels was measured to be -41 kJ mol using
2
9
calorimetry, in agreement with the adsorption enthalpy
4
5
predicted from its correlation to channel size. Taken together
with the measured activation energy on 12-MR channels (151
2 4
upon elimination of the neutral molecule (H or CH ) from the
-1
kJ mol ), this enthalpy gives an intrinsic cracking activation
transition state. Turnover rates are higher for these partially
confined species, in spite of the concomitant enthalpic desta-
bilization caused by partial confinement, indicating that their
free energies are in fact lower than for similar species fully
barrier of 192 ( 7 kJ mol- for acid sites within 12-MR channels
1
(
3 8
eq 6). These data agree well with values for C H cracking on
-1
H-MFI, H-FER and H-MOR (201-208 kJ mol ; Table 3).
1
966 J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009

Monomolecular Alkane Activation on Zeolites
A R T I C L E S
contained within 12-MR channels. These findings illustrate the
dominant role of entropy in the stabilization of adsorbed species
and transition states within confined environments and further
extend our previous studies that concluded the exponential
increase in monomolecular cracking rates of C
3
6
-C n-alkanes
on H-MFI to reflect predominantly larger transition state
3
7
entropies.
The entropy component of transition state free energies, and,
by extension, the benefits of the trade-offs mediated by partial
containment, becomes increasingly significant for later transition
states involving dissociation and elimination of a fragment from
the activated complex, and causes atypically large pre-
1
5
17 -1
exponential factors (10 -10 s ) for gas-phase fission reac-
4
6
tions. As a result, the specificity of 8-MR channels is even
stronger for dehydrogenation than for cracking transition states,
2
7
because of the later nature of the former, as found experi-
mentally (Table 5). We were unable to detect dehydrogenation
contributions from 12-MR sites because of their low reactivity
at the conditions of our measurements, precluding a similar
analysis to that used above for 8-MR pockets to infer partial
containment and enthalpy-entropy trade-offs in cracking transi-
tion state free energies. The similar nature of cracking and
dehydrogenation transition states, however, leads us to conclude
that similar effects of partial containment are responsible for
the preferential stabilization of dehydrogenation transition states
within 8-MR pockets in MOR.
Figure 7. The dependence of propane cracking-to-dehydrogenation rate
ratios at 748 K on the fraction of Brønsted acid sites in 8-MR pockets of
MOR, predicted using regressed rate constants (Table 5) and eq 12, is given
by the solid curve. Data at 748 K are shown (MOR-Z: b MOR-S: 9, MOR-
T: 2) along with Na/Al ratios for MOR-Z samples.
The less negative transition state entropies (relative to
reactants in the gas phase) measured for OH groups within 8-MR
pockets (Table 5) reflect the strong effects of local environment
on cracking and dehydrogenation rate constants and on cracking-
to-dehydrogenation rate ratios and the concomitant changes in
these properties among H-MOR samples with different OH
distributions (Tables 2 and 4). The dependence of propane
cracking-to-dehydrogenation rate ratios (at 748 K) on the
fraction of all OH groups present within 8-MR H-MOR pockets
is shown in Figure 7, together with values predicted from eq
To our knowledge, the enzyme-like specificity of spatial
environments within 8-MR pockets in H-MOR or the marked
effects of partial confinement have not been previously recog-
nized or demonstrated for alkane activation reactions. These
findings resemble, however, the reported specificity of 8-MR
3
pockets for CO insertion into CH groups during dimethyl ether
8
carbonylation to methyl acetate at ∼450 K . The sole mecha-
nistic connection between these reactions and the alkane
activation reactions that we discuss here is the common
involvement of cationic transition states. Such specificity was
not observed for isotopic exchange reactions on zeolites, for
which transition states are symmetrical and essentially un-
12 and regressed rate constants (Table 5), which agree well with
data on H-MOR samples of different provenance and for
+
H-MOR samples in which Na exchange was used to selectively
titrate 8-MR OH groups.
9
We conclude that selectivity in alkane activation reactions
can be tuned using the location of acid sites within a given
zeolite structure. Temperature, which exploits differences in
activation energies between dehydrogenation and cracking
pathways, can also be used to control selectivity but relies solely
on intrinsic differences in relative stabilities of reactants
protonated at various positions or bonds; these relative intrinsic
barriers for cracking and dehydrogenation are therefore inde-
pendent of zeolite structures or channel location (Table 3), as
predicted by the thermochemical cycle in Scheme 1b. Our
findings provide an alternate strategy for selectivity control,
which exploits location-specific differences in entropy between
dehydrogenation and cracking transition states. To our knowl-
edge, the ability to exploit location and zeolite structure to tune
the selectivity of cracking and dehydrogenation pathways and
the interpretation of the resulting effects in terms of the dominant
effects of transition state entropies have not been previously
recognized in zeolite catalysis. These data and conclusions
provide design strategies for the placement of Al sites within
specific locations in order to promote specific reactions via
rigorous considerations of the position of their respective
transition states along the reaction coordinate.
charged. We conclude that this remarkable specificity and the
partial confinement effects responsible for it are general features
of chemical reactions occurring via transition states with highly
localized cationic centers within constrained environments. We
provide next evidence for the role of partial confinement on
the relative reactivity of various C-C bonds within alkanes in
monomolecular cracking reactions and demonstrate the ability
of H-MOR side pockets to selectively crack terminal C-C
bonds in n-butane as a result of partial confinement configurations.
3.3. Partial Confinement Effects and Preferential Terminal
Cracking Selectivity in n-Butane Activation on MOR. Concepts
of pore mouth catalysis, a form of partial confinement, have
been used as a phenomenological description of processes in
which molecules penetrate small zeolite channels only partially
as they react. Propane reactions in MOR may well fit this
empirical concept, because reactants cannot fully enter the 8-MR
pockets within which they preferentially react and remain in
part within connecting 12-MR channels (detailed treatments for
estimating channel and molecular dimensions shown in Sections
S.6 and S.7, respectively, of the Supporting Information). The
5
6
(56) Degnan, T. F. J. Catal. 2003, 216, 32.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009 1967

A R T I C L E S
Gounder and Iglesia
+
Table 6. Monomolecular n-Butane Cracking and Dehydrogenation Rate Constants (kmeas) (per total H ) and Rate Ratios at 748 K
Experimentally Measured on MOR-Z Samples and Estimated within 8-MR and 12-MR Locations of MOR
3
+ -1 -1
-1
k
meas (× 10 mol (mol H )
s
bar
)
rate ratios
terminal/central C-C cracking
zeolite
terminal C-C cracking
central C-C cracking
dehyd
cracking/dehyd
H
H
H
H
H
100Na
83Na17MOR-Z
73Na27MOR-Z
59Na41MOR-Z
0
MOR-Z
13.0
10.2
6.9
6.4
4.0
3.2
2.9
2.7
2.4
1.6
4.1 ( 2.0
1.8 ( 1.2
13.8
12.8
4.9
4.4
2.1
31.5 ( 6.0
n.d.
4.0
3.5
2.6
2.6
2.6
5.0
n.d.
1.1
1.0
2.0
2.0
2.6
0.9
45Na55MOR-Z
a
8
-MR
20.8 ( 5.4
n.d.
a
b
b
b
b
1
2-MR
n.d.
a
Rate parameters for 8-MR and 12-MR locations determined by least-squares regression. Uncertainties in regressed rate parameters are reported as
b
twice the standard error. n.d., not detected.
Table 7. Measured Activation Energies (Emeas) and Entropies (∆Smeas) for Monomolecular n-Butane Cracking and Dehydrogenation on MOR
Samples and Estimated Values within 8-MR and 12-MR Locations of MOR
-1
-1 -1
Emeas (kJ mol
)
∆Smeas (J mol
K )
a
a
b
c
c
d
zeolite
terminal C-C cracking
central C-C cracking
dehyd
terminal C-C cracking
central C-C cracking
dehyd
H
H
H
H
H
100Na
83Na17MOR-Z
73Na27MOR-Z
59Na41MOR-Z
0
MOR-Z
156
160
139
148
153
152
145
146
139
159 ( 11
134 ( 10
203
-86
-96
-36
227
213
231f
n.d.
215 ( 11
n.d.
-83
-98
-4
-114
-102
-110
-72 ( 23
-108
-31
-8
-108
f
45Na55MOR-Z
145
-120
n.d.
e
8
-MR
163 ( 15
-86 ( 19
-126 ( 22
-13 ( 17
e
f
f
f
f
1
2-MR
n.d.
n.d.
n.d.
a
Errors for experimentally determined cracking Emeas are ( 6 kJ mol-1
b
.
Errors for experimentally determined dehydrogenation Emeas are ( 10 kJ
mol-
15 J mol
1
.
c
Errors for experimentally determined cracking ∆Smeas are ( 9 J mol
-1
K
-1
d
.
Errors for experimentally determined dehydrogenation ∆Smeas are
-1
-1
e
(
K . Rate parameters for 8-MR and 12-MR locations determined by least-squares regression. Uncertainties in regressed rate parameters
f
are reported as twice the standard error. n.d., not detected.
resulting transition states would preferentially activate terminal
C-C and C-H bonds, but such preferences lead to identical
products for propane reactants. Any terminal C-C bond
activation preference arising from partial confinement would
be apparent for n-butane reactants, but specific activation of
terminal C-H bonds may be impossible to detect because of
fast hydride shifts on both 8-MR and 12-MR acid sites. Indeed,
equilibrated n-butene isomers were detected at all conditions
on all catalysts (data shown in Section S.8 of the Supporting
Information). We use here terminal-to-central C-C bond
cleavage rate ratios to probe how molecules access active sites
within 8-MR and 12-MR environments and to demonstrate the
effects of partial confinement on n-butane dehydrogenation and
cracking turnover rates.
5
on H-MOR (718-778 K; Figure 8b) and of n-C H12 cracking
22
selectivities on H-MFI (750-810 K). The absence of enthalpic
preferences for the cleavage of one of the two types of C-C
bonds in n-butane via late carbonium ions as transition states
is consistent with the similar gas-phase protolytic cracking
enthalpies for the various C-C bonds in n-decane estimated
5
7
by theory. The terminal-to-central C-C bond cracking
+
selectivity decreased monotonically as H sites within 8-MR
+
pockets were selectively replaced by Na (Table 6), suggesting
that terminal bonds are activated preferentially on 8-MR OH
sites, as expected from partial confinement of n-butane within
these side pockets.
+
Monomolecular rate constants for H sites within 8-MR and
12-MR locations were estimated from measured rate constants
+
Rate constants of terminal and central C-C bond cracking
(per H ) on the five MOR-Z catalysts (using eq 12 and
+
and dehydrogenation of n-C
4
H
10 (per residual H ) and their rate
treatments similar to those in Section S.5 of the Supporting
Information). These rate constants (733-763 K; shown at 748
K in Table 6) were used to determine activation energies and
entropies on acid sites within these two environments (Table
ratios are shown at 748 K in Table 6 for five MOR samples
+
with varying Na content; the corresponding measured activa-
tion energies and entropies are shown in Table 7. Figure 8 shows
the temperature dependence of monomolecular rate constants
3 8 4
7). As in the case of C H , n-C H10 cracking and dehydroge-
+
(
(
per residual H ) for terminal and central C-C bond cleavage
nation rate constants were larger on 8-MR than on 12-MR acid
sites. The rates of dehydrogenation and of terminal C-C
cracking on 12-MR acid sites were too small to be estimated
accurately from overall rate constants, because both reactions
occurred much faster on 8-MR acid sites. Measured terminal-
to-central cracking rate ratios increased from essentially zero
on 12-MR acid sites to 5.0 on sites within 8-MR pockets (Table
Figure 8a) and of their rate ratio (Figure 8b) on MOR samples
+
with 56% (H100Na
sites within 8-MR pockets. As Na selectively replaced H
0
MOR-Z) and 13% (H45Na55MOR-Z) of H
+
+
within 8-MR locations, terminal and central C-C bond cracking
+
and dehydrogenation rate constants (per total H ) decreased,
while cracking-to-dehydrogenation rate ratios concurrently
increased (Table 6, Figure 8). These trends resemble those for
propane cracking and dehydrogenation, suggesting that the
specificity of OH groups in 8-MR pockets for both reactions,
and especially for dehydrogenation, is not restricted to the
4 10
6). These data indicate that the partial confinement of n-C H
reactants into 8-MR pockets leads to a configurational preference
for cracking at terminal C-C bonds in n-butane.
Measured activation energies for central C-C bond cracking
smaller C
3
H
8
reactants. Measured activation energies for
4
of n-C H10 (Table 7) were higher on sites within 8-MR pockets
-
1
terminal and central C-C bond cracking are similar to each
other on all samples (Table 7), consistent with the lack of
(159 ( 11 kJ mol ), where this reaction actually occurs at
4
temperature effects on n-C H10 terminal cracking selectivities
(57) Hunter, K. C.; East, A. L. L. J. Phys. Chem. A 2002, 106, 1346.
1
968 J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009

Monomolecular Alkane Activation on Zeolites
A R T I C L E S
+
Figure 8. (a) Measured rate constants (kmeas) (per total H ) for terminal C-C bond cracking in n-butane ((, )) and central C-C bond cracking in n-butane
9, 0) on H-MOR-Z (closed symbols) and H45Na55MOR-Z (open symbols). (b) Dependence of terminal-to-central C-C bond cracking rate ratios of n-butane
on temperature on H-MOR-Z (b) and H45Na55MOR-Z (O).
(
Table 8. Monomolecular Propane Cracking and Dehydrogenation
Rate Constants (kmeas) and Cracking-to-Dehydrogenation (C/D)
Rate Ratios Measured at 748 K on Different H-MFI Samples and
-
1
higher rates, than in 12-MR channels (134 ( 10 kJ mol ), as
4 10
also found for C cracking. Adsorption enthalpies of n-C H
H
3 8
-1
29
Measured Activation Energies (Emeas
)
in 12-MR channels of H-MOR (-50 kJ mol ) lead to an
estimated intrinsic activation barrier for central C-C bond
cracking of 184 ( 12 kJ mol- (using eq 6); this barrier is
cracking on 12-MR
channels of H-MOR (192 ( 7 kJ mol ) and in previous studies
for cracking of C -C n-alkanes on H-MFI (194-198 ( 14 kJ
mol ). If we assume that intrinsic activation barriers for
central C-C bond cracking within 8-MR pockets were also
similar to these values in 12-MR channels, then we estimate
n-butane adsorption energies of -25 kJ mol- from eq 6,
indicating that n-butane binding is weaker than expected from
3
+ -1 -1
–1
-1
k
meas (× 10 mol (mol H )
s
bar )
E
meas (kJ mol )
1
a
C/D ratio cracking dehyd
b
zeolite
cracking
dehyd
similar to that reported here for C
3
H
8
H-MFI-1
H-MFI-2
2.0
6.3
3.9
4.4
1.5
2.1
3.9
1.5
3.5
0.8
0.9
1.6
2.7
1.3
1.9
158
155
150
150
150
200
204
193
200
194
-
1
3
6
H Na MFI-2
8
5
15
-1
21
H-MFI-3
H-MFI-4
a
Errors are ( 5 kJ mol-1
b
Errors are ( 7 kJ mol-1
.
.
1
of T-sites makes Al and OH distributions inaccessible to
experiment using existing tools.
-
1
complete confinement within 8-MR pockets (-59 kJ mol ;
Section S.6 of the Supporting Information). These findings
indicate that van der Waals interactions stabilize only part of
the n-butane molecule (0.83 nm length) and of its terminal
cracking transition state within shallow 8-MR pockets (0.37 nm
depth). Even for central C-C bond cleavage in n-butane,
turnover rates were higher in 8-MR than 12-MR locations, in
spite of weaker enthalpic stabilization of cracking transition
states in 8-MR pockets, because of concomitantly larger
activation entropies (Table 7). These compensating entropic
effects appear to reflect the partial protrusion of adsorbed
reactants and transition states into connecting 12-MR channels,
which decrease the free energy of the transition state, the
relevant thermodynamic property that describes its stability.
These transition state free energies, referenced to gas-phase
reactants, determine measured turnover rates at the conditions
of our study (eq 5) and account for the effects of reactant and
acid site environment on the rate and selectivity of alkane
activation reactions.
3
.4. Monomolecular Propane Activation on H-MFI Zeo-
lites. Rate constants for propane cracking and dehydrogenation
per H ) and cracking-to-dehydrogenation rate ratios at 748 K,
and measured cracking and dehydrogenation activation energies
are shown in Table 8 on H-MFI samples of different provenance
and acid site density. Measured activation barriers were 40-50
kJ mol higher for dehydrogenation than for cracking, as also
found for H-MOR. Consequently, cracking-to-dehydrogenation
rate ratios decreased with increasing temperature on all H-MFI
+
(
-1
+
samples. Cracking and dehydrogenation turnover rates (per H )
varied among these H-MFI samples (by up to factors of 5),
without any systematic trends with their Al content. These
turnover rates (or their ratios) did not correlate with the
concentration of extraframework Al atoms or Lewis acid centers,
as shown by the NMR and IR data in Sections S.1 and S.2,
respectively, of the Supporting Information.
These turnover rate differences (Table 8) suggest that
Brønsted acid sites at the 12 T-sites in MFI are not identical in
reactivity and that these samples differ in their distribution of
Al among these locations. In contrast, earlier studies reported
Next, we examine whether local spatial constraints also
influence reactivity via these entropic effects on transition state
stability as the zeolite structure changes. Specifically, we address
the role of intrachannel environment on Brønsted acid site
reactivity for H-MFI zeolites, for which the structural diversity
1,2,33
n-hexane cracking activities that did not depend on Al content
and inferred that the location of the sites were also inconse-
quential for reactivity. These data would also be consistent with
location-specific reactivity but for a set of samples in which Al
J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009 1969

A R T I C L E S
Gounder and Iglesia
is similarly distributed among T-sites for all Al contents, a
toward more rational design and selection strategies for mi-
croporous catalysts with specific catalytic properties.
33
possibility mentioned in the original article, but largely ignored
thereafter. In spite of similar acid strength, the reactivity of
Brønsted acid sites in MFI depends on location, as shown here
for H-MOR (Section 3.2) and appears to account for the turnover
rate differences among our H-MFI samples (Table 8). We
conclude that previous studies in which n-hexane cracking rates
4. Conclusions
+
Turnover rate (per H ) differences for monomolecular
cracking and dehydrogenation of propane and n-butane with
changes in zeolite structure (H-MFI, H-FER, and H-MOR) and
acid site location (H-MOR: 8-MR side pocket, 12-MR main
channel) reflect the strong dependence of cationic transition state
free energy on local channel environment. In agreement with
Born-Haber thermochemical cycles that define energy relations
in acid catalysis, intrinsic activation barriers for both mono-
1
,2,33
were independent of Al content
may have used H-MFI
samples with a similar distribution of acid sites among locations
of different reactivity. We cannot comment further about these
discrepancies or about the extent to which bimolecular pathways
prevailed during these previous studies, because precise experi-
mental conditions were not given and the cracking data were
reported as catalytic activities (arbitrary units) that were likely
measured under integral reactor conditions.
-1
molecular propane cracking (201-208 kJ mol ) and dehydro-
-1
genation (229-245 kJ mol ) were similar on H-MFI, H-FER,
+
+
and H-MOR samples and were consistently larger for dehy-
The replacement of just 15% of H sites in H-MFI-2 by Na
decreased C cracking and dehydrogenation turnover rates
(
-1
drogenation (by 25-42 kJ mol ). The insensitivity of these
barriers to zeolite structure reflects similar zeolite deprotonation
enthalpies and commensurate differences in the stabilization of
transition states and reactants by different channel environments.
Transition states for dehydrogenation are higher in energy than
for cracking, reflecting respective affinities for protonation at
C-C and C-H bonds in gas-phase alkanes, and are higher in
entropy than for cracking, consistent with the later and looser
transition states for the former pathways as suggested by theory.
3 8
H
+
per H ) (by ∼1.5 and ∼2.5 factors, respectively) and increased
their selectivity ratios (from 1.6 to 2.7) (Table 8), but measured
activation energies for neither cracking nor dehydrogenation
+
were affected. These effects of Na exchange on monomolecular
+
C H
3 8
cracking and dehydrogenation turnover rates (per H ) on
H-MFI resemble those attributed here in the case of H-MOR to
+
+
the preferential replacement of H by Na at locations where
cracking and dehydrogenation catalysis is most effective (Table
4
). These effects of alkali exchange are also similar to those
Monomolecular cracking and dehydrogenation reactions of
propane and n-butane occurred predominantly on Brønsted acid
sites located within 8-MR side pockets of H-MOR, a conse-
quence of spatially constrained environments that allow only
partial containment of reactants and transition states. Partial
transition state confinement results in entropy gains that
compensate for concomitant enthalpy losses and decrease
transition state free energies. Such strong effects of channel
environment and, by extension, acid site location on reactivity
allowed for systematic and precise control of cracking-to-
dehydrogenation selectivities and of terminal-to-central C-C
bond cleavage selectivities by selective titration of OH groups
5
8
59
4 6
reported for i-C H10 and n-C H14 cracking on H-USY, in
which cracking rates were rendered undetectable upon titration
of just 20-33% of the H species associated with framework
Al atoms. Our data show that Na cations preferentially replace
Brønsted acid sites with the highest reactivity for reactions
involving cationic transition states. We conclude also that
electropositive Na cations preferentially reside within smaller
channel environments, where entropy-enthalpy trade-offs result-
ing from partial confinement are most consequential. In MFI,
however, the structural diversity of T-site locations precludes
rigorous assessment of Al siting and prevents unequivocal and
specific interpretations of the effects of local environment on
cracking and dehydrogenation of alkanes.
+
+
+
+
in 8-MR pockets of MOR with Na .
These findings reflect the broad range of reactivities likely
to prevail among acid sites located within different channels of
the same zeolite structure, shown explicitly for MOR samples
of varying provenance and acid site distribution and consistent
with data obtained on different MFI samples. In what appears
to be a consideration specific to and consequential for acid
catalysis by zeolite, channel environments influence the forma-
tion of cationic transition states, more fundamentally than simple
considerations of size and shape, through their solvation of
transition states and mediation of compromises in enthalpy and
entropy factors. These findings and their conceptual interpreta-
tions offer specific design and selection strategies for mi-
croporous solids of specific channel structure and acid site
location with predictable consequences for acid catalysis.
The preferential reactivity of specific channel environments
reported here for alkane activation on Brønsted acid sites and
earlier for CO insertion into surface methyls appear to represent
8
general features of catalytic reactions involving cationic transi-
tion states. Spatial constraints imposed by zeolite channels,
frequently proposed to select transition states based simply on
their size and shape, play a more fundamental and consequential
role in acid catalysis via solvation of cationic transition states
and specifically via its mediation of enthalpy and entropy factors,
predominantly in ways that favor entropic stabilization even at
the expense of enthalpic penalties. We expect that similar effects
will prevail for bimolecular alkane reaction pathways that
propagate via hydride transfer and ꢀ-scission and oligomeriza-
tion cycles, channel environments permitting the formation of
the bulkier transition states involved, because these pathways
Acknowledgment. The authors thank Dr. Stacey I. Zones
(Chevron) for the MFI samples. We thank Dr. Zones along with
Prof. Aditya Bhan (University of Minnesota at Twin Cities), Josef
Macht (University of California at Berkeley), and Prof. Johannes
A. Lercher (Technische Universit a¨ t M u¨ nchen) for helpful discus-
sions. We also thank Dr. Sonjong Hwang (California Institute of
Technology) and Dr. Chul Kim (California Institute of Technology)
6
,7
also require the formation of cationic transition states. Our
findings about the dominant role of entropy and of partial
containment provide a conceptual path forward toward a more
rigorous assessment of local environment effects on transition
state stability and therefore on site reactivity and selectivity and
2
7
for collecting the Al NMR spectra reported here. Finally, we
acknowledge with thanks the financial support from Chevron
Energy Technology Company.
(
58) Beyerlein, R. A.; McVicker, G. B.; Yacullo, L. N.; Ziemiak, J. J. Abstr.
Pap.sAm. Chem. Soc. 1986, 191, 7-PETR.
(
59) Fritz, P. O.; Lunsford, J. H. J. Catal. 1989, 118, 85.
1
970 J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009

Monomolecular Alkane Activation on Zeolites
A R T I C L E S
Supporting Information Available: Experimental methods
and Al NMR spectra for zeolite samples; experimental methods
and IR spectra for zeolite samples upon adsorption of CO at
MOR; estimation of adsorption parameters in 8-MR pockets of
MOR; estimation of 8-MR MOR pocket depth and kinetic
diameters of molecules; kinetic data for butene isomer distribu-
tions on MOR samples. This material is available free of charge
via the Internet at http://pubs.acs.org.
27
1
23 K; assessment of transport corruptions using Mears criteria;
detailed derivations of rate laws and activity coefficients using
transition state treatments for thermodynamically nonideal
systems; description of least-squares regression methods used
for rate constant estimation in 8-MR and 12-MR locations of
JA808292C
J. AM. CHEM. SOC. 9 VOL. 131, NO. 5, 2009 1971