Structure and density of Mo and acid sites in Mo-exchanged H-ZSM5 catalysts for nonoxidative methane conversion

Article abstract of DOI:10.1021/jp990866v

Mo/H-ZSMS (1.0-6.3 wt % Mo; Mo/Al = 0.11-0.68) catalysts for CH₄ aromatization were prepared from physical mixtures of MoO₃ and H-ZSM5 (Si/Al = 14.3). X-ray diffraction and elemental analysis of physical mixtures treated in air indicate that MoO_x species migrate onto the external ZSM5 surface at about 623 K. Between 773 and 973 K, MoO_x species migrate inside zeolite channels via surface and gas phase transport, exchange at acid sites, and react to form H₂O. The amount of H₂O evolved during exchange and the amount of residual OH groups detected by isotopic equilibration with D₂ showed that each Mo atom replaces one H⁺ during exchange. This stoichiometry and the requirement for charge compensation suggest that exchanged species consist of (Mo₂O₅)²⁺ ditetrahedral structures interacting with two cation exchange sites. The proposed mechanism may provide a general framework to describe the exchange of multivalent cations onto Al sites in zeolites. As the Mo concentration exceeds that required to form a MoO_x monolayer on the external zeolite surface (～4 wt % Mo for the H-ZSM5 used), Mo species sublime as (MoO₃)_n oligomers or extract Al from the zeolite framework to form inactive Al₂(MoO₄)₃ domains detectable by ²⁷Al NMR. These (Mo₂O₅)²⁺ species reduce to form the active MoC_x species during the initial stages of CH₄ conversion reactions. Optimum CH₄ aromatization rates were obtained on catalysts with intermediate Mo contents (～0.4 Mo/Al), because both MoC_x and acid sites are required to activate CH₄ and to convert the initial C₂H₄ products into C₆₊ aromatics favored by thermodynamics.

Full text of DOI:10.1021/jp990866v

J. Phys. Chem. B 1999, 103, 5787-5796
5787
Structure and Density of Mo and Acid Sites in Mo-Exchanged H-ZSM5 Catalysts for
Nonoxidative Methane Conversion
Richard W. Borry III, Young Ho Kim, Anne Huffsmith, Jeffrey A. Reimer, and
Enrique Iglesia*
Materials Sciences DiVision, E.O. Lawrence Berkeley National Laboratory, and Department of Chemical
Engineering, UniVersity of California, Berkeley, California 94720
ReceiVed: March 12, 1999; In Final Form: May 7, 1999
Mo/H-ZSM5 (1.0-6.3 wt % Mo; Mo/Al ) 0.11-0.68) catalysts for CH4 aromatization were prepared from
physical mixtures of MoO3 and H-ZSM5 (Si/Al ) 14.3). X-ray diffraction and elemental analysis of physical
mixtures treated in air indicate that MoOx species migrate onto the external ZSM5 surface at about 623 K.
Between 773 and 973 K, MoOx species migrate inside zeolite channels via surface and gas phase transport,
exchange at acid sites, and react to form H2O. The amount of H2O evolved during exchange and the amount
of residual OH groups detected by isotopic equilibration with D2 showed that each Mo atom replaces one H+
during exchange. This stoichiometry and the requirement for charge compensation suggest that exchanged
species consist of (Mo2O5)2+ ditetrahedral structures interacting with two cation exchange sites. The proposed
mechanism may provide a general framework to describe the exchange of multivalent cations onto Al sites
in zeolites. As the Mo concentration exceeds that required to form a MoOx monolayer on the external zeolite
surface (∼4 wt % Mo for the H-ZSM5 used), Mo species sublime as (MoO3)n oligomers or extract Al from
the zeolite framework to form inactive Al2(MoO4)3 domains detectable by 27Al NMR. These (Mo2O5)2+ species
reduce to form the active MoCx species during the initial stages of CH4 conversion reactions. Optimum CH4
aromatization rates were obtained on catalysts with intermediate Mo contents (∼0.4 Mo/Al), because both
MoCx and acid sites are required to activate CH4 and to convert the initial C2H4 products into C6+ aromatics
favored by thermodynamics.
Introduction
(MoO3)n oligomers into zeolite channels.6 These authors con-
cluded that MoOx interacts with framework oxygens in H-ZSM5
The selective conversion of natural gas to higher hydrocar-
bons and aromatics remains an important industrial challenge.
The discovery of catalytic CH4 aromatization on Mo/H-ZSM51
was followed by several reports of near-equilibrium CH4
conversions at 973 K with high selectivity to benzene. 2-4 CH4
reaction rates increased with time on stream as dispersed MoOx
formed the active MoCx species.3 Ethylene and ethane are
formed on MoCx sites as primary products and they convert to
C6+ aromatics via oligomerization, cracking, and cyclization
reactions. These reactions require chain growth and dehydro-
genation steps that occur on Bronsted acid sites aided by
hydrogen desorption sites provided by MoCx species. The net
reaction rate is limited by the rate of CH4 activation and by the
approach to thermodynamic equilibrium of the overall methane
aromatization reaction.3
2-
at 973 K to form isolated monomolybdate species (MoO4
based on the appearance of infrared bands for ModO vibrations
in tetrahedral Mo6+
)
.
Surface migration of MoOx onto H-ZSM5 can occur at high
temperatures because lattice mobility within MoO3 becomes
possible above its Tammann temperature (534 K).9 Sublimation
of MoO3 becomes detectable above 623-673 K,10 and MoO3
reaches a vapor pressure of 56 Pa at 973 K11 as (MoO3)n
oligomers (n ) 2-5).12 H2O formed during decomposition of
AHM precursors can form MoO2(OH)2,12,13 which has a vapor
pressure of 4.9 Pa at 973 K.11
As isolated MoOx species migrate into zeolite channels via
gas phase or surface diffusion, they react with H+ atoms at
exchange sites to form (MoO2(OH))+ species, which can
condense with another one to form a (Mo2O5)2+ dimer and H2O:
In previous studies, Mo/H-ZSM5 catalysts were prepared by
slurry or incipient wetness impregnation of H-ZSM5 with
aqueous ammonium heptamolybdate (AHM; (NH4)6Mo7O24)
solutions, followed by treatment in air at 723-973 K.1-4 Large
aqueous molybdate ions do not exchange directly onto H-ZSM5
cation exchange sites during impregnation.5 Infrared spectros-
copy and differential thermal analysis6 showed that AHM
decomposed in air between 500-650 K to form MoO3 crys-
tallites on the outer surface of zeolite crystals.7,8 At 773 K, MoO3
infrared bands disappeared as MoO3 crystallites dispersed on
the external surface of ZSM5 crystals and then migrated as
Throughout this report, we refer to these dimers as (Mo2O5)2+
in spite of the fact that such cationic species are not known in
* Corresponding author. E-mail: iglesia@cchem.berkeley.edu.
10.1021/jp990866v CCC: $18.00 © 1999 American Chemical Society
Published on Web 06/25/1999

5788 J. Phys. Chem. B, Vol. 103, No. 28, 1999
Borry et al.
solution, because it is customary to think of exchanged species
in zeolites as cationic. The stable structural analog is the known
(Mo2O7)2- anion, consisting of two Mo centers with tetrahedral
symmetry, with two of the oxygen atoms residing at framework
positions in ZSM5. (MoO2(OH))+ species can also react with a
zeolite OH group to form a (MoO2)2+ cation bridging two acid
sites and water, a structure previously proposed for the exchange
of MoO2Cl2 onto H-Y zeolite.14
measured by mass spectrometry (Leybold Inficon, THP-TS200)
using heated transfer lines (373 K) and Ar as an internal
standard. Samples were held at 973 K for 0.5 h and then cooled
to 300 K. The number of residual OH groups in Mo/H-ZSM5
was measured from the evolution of HD and H2 by mass
spectrometry as samples were heated from 300 to 873 K (10
K/min, 2 min hold time) in 5% D2/Ar (100 cm3/min, Praxair,
>99.999%).16,17 Samples were then cooled to 300 K and the
experiment was repeated using 5% H2/Ar (100 cm3/min, Praxair,
>99.999%) while measuring HD and D2 evolution rates. The
Mo content was measured by atomic absorption (Galbraith
Laboratories) after treatment in air at 973 K.
27Al NMR spectra were collected using a home-built 400
MHz spectrometer (9.1 T magnetic field) at 104.2 MHz while
spinning at 4 kHz.18 Samples were hydrated at room temperature
by placing them in an empty desiccator containing liquid water.
Hydration of Al sites weakens quadrupole interactions that
broaden 27Al NMR lines in H-ZSM5.19,20
Similar condensation reactions of OH groups lead to the
extraction of Al ions from the framework, with the formation
of water and the disappearance of two Bronsted acid sites. In
Mo/H-ZSM5, extraframework Al atoms form small domains
of Al2O3 or Al2(MoO4)3; the latter species were detected by
27Al NMR in Mo/H-ZSM5 samples with high Mo content (>10
wt %).15
Steady-State CH4 Aromatization on Mo/H-ZSM5. CH4
reactions were carried out at 950 K in a tubular reactor (4 mm
i.d.) with plug-flow hydrodynamics (25 cm3/min, 1:1 CH4/Ar
(Praxair, >99.995%)). Mo/H-ZSM5 catalysts (1.0 g) were
loaded onto a porous quartz disk (1.0 cm i.d.) located within
this reactor. Temperatures were measured with a type K
thermocouple located inside a quartz sheath in contact with the
catalyst bed. The reactor effluent was sampled using transfer
lines held at 400 K,21 and the reactant and product concentrations
were measured by gas chromatography.22 Reported selectivities
are defined as the percentage of the CH4 converted appearing
as a given product. CH4 was used as the reference peak between
the flame ionization and thermal conductivity detectors and Ar
was used as an internal standard. The amount of carbon missing
within the measured products (1-10%) is reported as “carbon”
in the results, but it includes carbon consumed to form MoCx
and condensable products remaining in zeolite channels or in
transfer lines. The reported values for CH4 conversion, hydro-
carbon yields, and carbon balances were reproducible within
(0.5% (absolute), and the detection limits were 50 and 1 ppm
for thermal conductivity and flame ionization detectors, respec-
tively. Catalysts were treated in 20% O2/He (100 cm3/min,
Praxair, >99.999%) at 950 K for 2 h before catalytic reactions.23
Impregnation or ion exchange of H-ZSM5 with AHM
solutions leads initially to external MoO3 crystals during
treatment in air. Thus, it seemed possible to prepare Mo/H-
ZSM5 via exchange from intimate mixtures of MoO3 and
H-ZSM5 powders. This approach avoids AHM decomposition
products (N2, NH3, and H2O)6, which interfere with measure-
ments of the kinetics and the extent of exchange from the
amount of water evolved during synthesis. Here, we report
mechanistic details of the synthesis of MoOx/H-ZSM5 from
mixtures of MoO3 and H-ZSM5 powders, the structure of the
MoOx species formed during exchange, and the role and density
of MoOx species and of Bronsted acid sites in alkane reactions.
Results
Catalytic CH4 Aromatization on Mo/H-ZSM5. CH4 reac-
tion rates and selectivities measured on Mo/H-ZSM5 (4.0 wt
%) prepared from MoO3/H-ZSM5 mixtures (Figure 1) were
similar to those reported on samples prepared by impregnation
of H-ZSM5 with aqueous AHM (Table 1).3 During the initial
activation (∼1-2 h), CH4 conversion reached about 20%, but
carbon, H2, CO2, H2O, and CO were the most abundant initial
products. After 2 h, CO and CO2 were no longer detectable
and the carbon selectivity was below 7%. Benzene (∼70%
selectivity), toluene (∼5%), ethylene (4-8%), ethane (2-4%),
and naphthalene (10-20%) were formed at CH4 conversions
of 7-10%. Propene and o-xylene were also detected in trace
amounts (<2% selectivity). By comparison, a Mo/H-ZSM5 (2
wt %) sample prepared by impregnation of H-ZSM5 with
aqueous AHM3 gave 10% CH4 conversion and 70% selectivity
to benzene at 973 K (Table 1).
Experimental Section
Catalyst Synthesis and Characterization. H-ZSM5 was
prepared by aqueous exchange (repeated four times) of Na-
ZSM5 (Zeochem, Si:Al ) 14.3) with 1.0 M NH4NO3 (Fisher)
(∼10 g Na-ZSM5/L). The samples were dried at 400 K for 24
h and treated in dry air at 773 K for 24 h (10-15 g of cat., 50
cm3/min). Mo/H-ZSM5 was prepared from physical mixtures
of MoO3 (Johnson Matthey, 99.5% purity) and H-ZSM5; these
mixtures were ground together for about 0.1 h in an alumina
mortar and pestle. Powder X-ray diffraction patterns were
obtained in a Siemens Diffractometer D5000 using Cu KR
radiation (λ ) 1.5406 Å). Surface areas and pore volumes were
obtained from N2 physisorption at its boiling point (Autosorb
6; Quantachrome, Inc.), after treating samples at 773 K for 1 h
in flowing dry air (Medical grade, Praxair).
MoO3 Migration and Sublimation during Catalyst Syn-
thesis. Our initial measurements of the rate of MoOx migration
from the rate of H2O evolution during treatment in air were
influenced by MoO3 sublimation from the physical mixtures.
Heating MoO3/H-ZSM5 mixtures (4.0 wt % Mo) from 300 to
The effect of air treatment was examined using MoO3/H-
ZSM5 mixtures (0.3 g, 0-8 wt % Mo) dried at 623 K for 24 h
in 20% O2/Ar (100 cm3/min, Praxair, >99.999%). Samples were
heated at 10 K/min to 973 K and H2O evolution rates were

Mo-Exchanged H-ZSM5 Catalysts
J. Phys. Chem. B, Vol. 103, No. 28, 1999 5789
TABLE 1: Comparison of CH4 Aromatization on Mo/H-ZSM5 Prepared from Aqueous Impregnation of Molybdate Solution on
H-ZSM5 and Physical Mixture of MoO3/H-ZSM5
selectivity (%)
Mo content
(wt %)
Mo/Al
atomic ratio
reaction
temp (K)
GHSV
CH4
conversion (%)
solid
carbon
(h-1
800
)
C2H4
C6H6
C10H8
2.0a
4.0b
4.3e
0.37
0.42
0.45
973
950
950
10.0
10.2
1.7
3
2
17
70
66
27
20
18
4
5
12
52
750d
750d
a Incipient wetness impregnation of AHM/H-ZSM5.3 b Physical mixture of MoO3/H-ZSM5 (this study). c “Vapor-exchanged” preparation method
(this study, Figure 6). d Space velocity was calculated assuming catalyst packed density of 0.5 g cat./cm3 reactor volume.
Figure 2. H2O desorption from H-ZSM5 during first (a) and second
(b) TPO. The sample was heated at 10 K/min from 323 to 623 K, held
for 24 h, then heated at 10 K/min to 973 K (0.3 g, 100 cm3/min, 20%
O2/Ar, 101 kPa).
subsequent heating in air to 973 K led to Mo losses of less
than 10% (Table 2, program II). The H2O evolved during
exchange appeared as a well-defined peak between 623 and 973
K (as shown later in Figure 3). Holding at 773 K for 48 h during
treatment in air (Table 2, program III) eliminated Mo loss
(<0.1%), apparently because MoOx crystallites spread as
strongly interacting layers with much lower vapor pressure than
bulk MoO3. This method was used to prepare the Mo/H-ZSM5
samples used in the reaction studies (Figure 1, Table 1).
H2O Desorption during Oxidative Treatment of MoO3/
Figure 1. CH4 pyrolysis on 4 wt % Mo/H-ZSM5, (a, top) CH4
H-ZSM5 Mixtures. Figure 2 shows the H2O evolution rate
conversion, C2-C10 hydrocarbon product yield, and solid carbon yield
during temperature-programmed oxidation (TPO) of H-ZSM5
vs. time on stream, (b, bottom) carbon selectivity of gas-phase products
vs time on stream (1.0 g, 25 cm3/min, 50% CH4/Ar, 105 kPa, 950 K).
(using program II in Table 2). The peak at 300-500 K
corresponds to the desorption of water adsorbed on H-ZSM5
from ambient air (2.62 mol H2O/Al). The smaller peak between
673 and 973 K corresponds to zeolite dealumination (eq 3) by
loss of OH groups and extraction of Al atoms from tetrahedral
framework positions. A previous thermogravimetric study of
H-ZSM5 reached similar conclusions.24 Figure 2b shows H2O
evolution rates during a second TPO after cooling the sample
in dry 20% O2/Ar (100 cm3/min) to room temperature. The
physisorbed water peak is no longer detected and the dealumi-
nation peak was smaller and it appeared at higher temperature,
indicating that the zeolite structure was stabilized against further
dealumination by the initial treatment. The number of OH
groups desorbed as H2O during the first and second treatments
corresponds to 0.23 and 0.06 H/Al, respectively.
TABLE 2: Mo Content after Temperature-Programmed
Oxidation of MoO3/H-ZSM5 Mixtures (∼4 wt % Mo,
0.3-1.0 g, 100 cm3/min, 20% O2 in He)
Mo as prepareda Mo after TPOb
temperature program
(wt %)
(wt %)
I
300 to 973 K (10 K/min)
4.0
4.0
3.47
3.60
II 300 to 623 K (hold 24 h)
623 to 973 K (10 K/min)
III 300 to 623 K (hold 24 h)
623 to 773 K (10 K/min, 48 h)
773 to 950 K (10 K/min, 2h)
3.94
3.95
a Based on amount of Mo added to physical mixture. b Based on
elemental analysis of oxidized sample.
973 K (at 10 K/min) led to the sublimation of 13% of the Mo
as (MoO3)n or MoO2(OH)2 (Table 2, program I). The sublimed
MoOx appeared as small MoO3 crystallites on the walls after
the heated sample. When MoO3/H-ZSM5 mixtures were held
at 623 K for 24 h in order to remove physisorbed H2O and to
disperse MoOx species onto the external zeolite surface,
H2O evolution during TPO of a MoO3/H-ZSM5 physical
mixture (4.5 wt % Mo)25 shows that the amount of physisorbed
H2O is slightly smaller than on H-ZSM5 (1.80 mol H2O/Al),
but the second peak is much larger (Figure 3). This reflects the
anchoring of Mo at cation exchange sites to form Mo species
containing OH groups and the subsequent condensation of these

5790 J. Phys. Chem. B, Vol. 103, No. 28, 1999
Borry et al.
TABLE 3: Mo Content Measured by Atomic Absorption
Spectrometry before and after Oxidation of MoO3/H-ZSM5
Physical Mixtures of Varying Mo Concentration
(Temperature Program II in Table 2, 0.3 g, 100 cm3/min,
20% O2 in He)
target Mo
analyzed Mo
analyzed Mo
content content in physical
Mo/Al
content after Mo/Al
(wt %)
mixture (wt %) before TPO TPO (wt %) after TPO
1
2
4
8
1.1
2.2
4.0
7.9
0.11
0.22
0.42
0.87
1.0
2.0
3.6
6.3
0.11
0.20
0.37
0.68
TABLE 4: H2O Desorption (TPO Program II, 0.3 g, 100
cm3/min, 20% O2 in He) and Isotopic Exchange (10 K/min,
100 cm3/min, 5% (D2 or H2) in Ar) of MoO3/H-ZSM5
Mixtures
Mo content
H
H2O desorption
residual H content
balancef
H/Al
Figure 3. H2O desorption from MoO3/H-ZSM5 mixture (4 wt % Mo)
during TPO. The sample was heated at 10 K/min from 323 to 623 K,
held for 24 h, then heated at 10 K/min to 973 K (0.3 g, 100 cm3/min,
20% O2/Ar, 101 kPa).
after TPO Mo/Al
b
(wt %)
ratio H/Ala H/AlF H/Moc H/Ald
D/Ale
0
0
0.229
0
0.630
0.400
0.254
0.099
0.035
0.019
0.564
0.456
0.355
0.170
0.107
0.074
0.919g
0.803
0.829
0.856
0.772
1.0
2.0
3.6
6.3
4.3
0.11 0.347 0.118 1.11
0.20 0.474 0.245 1.22
0.37 0.686 0.457 1.23
0.68 0.665 0.436 0.64
0.45
a From integrated area of H2O desorption above 623 K (Figure 4).
b Subtract value in column 3 from that for H-ZSM5 (0 wt % Mo).
c Divide column 4 by Mo/Al ratio. d From D2(g)/O(s)-H isotopic
exchange (Figure 9). e From H2(g)/O(s)-D isotopic exchange (Figure
10). f Sum of columns 3 and 7. g Includes second TPO and D2(g)/
O(s)-H values (see text).
Figure 4. H2O desorption between 623 and 973 K from MoO3/
H-ZSM5 mixtures with varying Mo initial concentration (0.3 g, 100
cm3/min, 20% O2/Ar, 101 kPa, heat at 10 K/min).
OH groups with those in H-ZSM5 or in neighboring Mo cations
(eqs 1 or 2). The small difference in H2O desorption rates below
623 K between H-ZSM5 (Figure 2) and MoO3/H-ZSM5
mixtures (Figure 3) is caused by OH groups bound to MoO3,
and not by Mo exchange. This conclusion is confirmed by the
observation that H2O evolution rates between 623 and 773 K
did not depend on Mo loading (Figures 4 and 14), suggesting
that Mo cations migrate into channels and exchange only above
773 K, as also proposed by others.26 Figure 4 shows that the
amount of H2O formed between 773 and 973 K increased with
increasing Mo content. Some Mo sublimes from mixtures with
more than 2 wt % Mo (Table 3) when treated by program II
(Table 2); this was confirmed by elemental analyses before and
after air treatment and by the MoO3 residue formed at the exit
of the treatment cell.
The number of H2O molecules (per Al) formed during air
treatment is shown in Table 4. If the loss of OH groups via
recombination and Al extraction from framework positions
occurred to the same extent in Mo/H-ZSM5 and H-ZSM5
samples, H2O desorption values on Mo/H-ZSM5 can be
corrected by subtracting the amount of H2O desorbed from pure
H-ZSM5 (0.23 H/Al, Table 4). These corrected values are
reported as H/AlF (AlF ) framework aluminum). This assump-
Figure 5. Number of H desorbed as H2O per Al atom during treatment
of MoO3/H-ZSM5 mixtures in air to 973 K.
tion was confirmed using 27Al NMR to measure the change in
intensity of the peak corresponding to tetrahedral (framework)
Al centers (as discussed later). These data show that for samples
with less than 4 wt % Mo, MoOx species exchange with H+
with a stoichiometry of about one Mo per H (Figure 5, slope )
1.24 H/Mo). For H-ZSM5 with a Si/Al ratio of 14.3, the number
of cation exchange sites is not sufficient to accommodate all
Mo atoms in the 8 wt % Mo sample, and the excess MoOx
sublimes above 773 K during air treatment.
Vapor-Exchange Synthesis from Separated MoO3 and
H-ZSM5 Powders. In a separate experiment, Mo/H-ZSM5
samples were prepared by flowing 20% O2/He (100 cm3/min)
over bulk MoO3 (0.185 g), and contacting the (MoO3)n vapor
with a bed of H-ZSM5 (1.509 g; Figure 6). The Mo content
would be 7.3 wt % if all the MoO3 remained in the exchanged

Mo-Exchanged H-ZSM5 Catalysts
J. Phys. Chem. B, Vol. 103, No. 28, 1999 5791
Figure 6. Preparation of “vapor-exchanged” Mo/H-ZSM5 sample (1.5
g, 50 cm3/min, 20% O2/He, 101 kPa, 950 K, 96 h).
TABLE 5: N2 Physisorption at 77 K of MoO3/H-ZSM5
Mixtures after TPO and TPR Treatment in Air and H2
(D2) to 973 K
Figure 7. 27Al NMR spectra of MoO3/H-ZSM5 physical mixtures after
TPO and TPR treatments to 973 K (*) 4.3 wt % Mo sample prepared
via MoO3 vapor exchange at 950 K, 96 h.
Mo concn
(wt %)
BET surface area
(m2/g zeolite)
micropore volume
(cm3/g zeolite)
0
272
278
266
218
132
201
0.128
0.119
0.120
0.107
0.059
0.084
1.1
2.0
3.6
6.3
4.3a
decreased with increasing Mo content. 27Al NMR lines corre-
sponding to crystalline Al2(MoO4)3 appeared in the 6.3 wt %
Mo/H-ZSM5 sample.27 The Al2(MoO4)3 crystallites are too small
to be detected by X-ray diffraction in this sample, but they
become detectable in Mo/H-ZSM5 samples with higher Mo
content (15 wt %).15
a Prepared via MoO3 vapor exchange at 950 K, 96 h (see text).
Thus, it appears that the unusual shape of the H2O desorption
curve for 6.3 wt % Mo/H-ZSM5 reflects the extraction of Al
from the framework to form Al2(MoO4)3 and the concomitant
collapse of the zeolite pore structure. H2O desorption rates from
mixtures initially containing 4 and 8 wt % Mo were identical
up to about 850 K (Figure 4), indicating that MoOx-induced
extraction of framework Al did not occur below 850 K. This
agrees with 27Al NMR data for 15 wt % Mo/H-ZSM5, which
showed Al2(MoO4)3 in samples treated in air at 873 or 973 K,
but not in those treated at 773 K.15 The ratio of OH groups
removed as H2O to Mo atoms in the 6.3 wt % Mo sample is
only 0.64; this value is very similar to that expected for Al2-
(MoO4)3 (0.67), but it is smaller than in Mo/H-ZSM5 samples
with lower Mo content.
samples after treatment in air. All the MoO3 powder sublimed
after 96 h at 950 K, contacted the H-ZSM5 sample, and either
exchanged at ZSM5 sites or deposited as light green crystals
(bulk MoO3) on colder reactor walls above the zeolite bed. The
resulting Mo/H-ZSM5 contained 4.3 wt % Mo. These results
showed that surface diffusion is not required for MoOx migration
during treatment in air at high temperature (>873 K) and that
vapor-phase transport of (MoO3)n oligomers can occur during
synthesis of Mo/H-ZSM5 from physical mixtures. CH4 aroma-
tization reaction rates were very low on this sample. At a space
velocity of 750 h-1, CH4 conversion was only 1.7% (cf. 4.3 wt
% Mo sample; Table 1).
Zeolite Structural Changes during Thermal Treatment
of MoO3/H-ZSM5 Mixtures. The unusual shape of the H2O
evolution rates on the 8 wt % Mo/H-ZSM527 (Figure 4)
suggested some structural differences between this sample and
the others, which we explored using powder X-ray diffraction
(XRD), N2 physisorption, and 27Al NMR. N2 physisorption
results were used to estimate surface areas and micropore
volumes (Table 5). Neither diffraction nor N2 physisorption data
showed any structural degradation of the zeolite in samples with
1.0 or 2.0 wt % Mo relative to pure H-ZSM5, but both methods
detected a modest loss of crystallinity in the 3.6 and 4.3 wt %
Mo samples, and significant destruction of the zeolite framework
in the 6.3 wt % sample.27 A loss of surface area with increasing
Mo content was previously reported also for MoOx/ZSM5
prepared by AHM slurry methods.28
Isotopic Equilibration of D2 with OH Groups in Mo/H-
ZSM5 Samples. The isotopic equilibration of surface O-H
groups with D2 was carried out after treating Mo/H-ZSM5 in
air at 973 K (procedure II, Table 2). This method was used
previously in order to determine the number of OH groups in
H-ZSM5 and cation-exchanged H-ZSM5.16
D2/O-H exchange data are shown in Figure 8 for 3.6 wt %
Mo/H-ZSM5 (prepared by program II; Table 2). The total
number of H atoms removed as HD and H2 corresponds to 0.1
H per Al. Figure 9 shows HD desorption rates during D2/O-H
exchange for all samples (0-6.3 wt % Mo; prepared by
exchange in air at 973 K for 0.5 h). The number of O-H groups
decreased with increasing Mo content, as O-H groups were
increasingly replaced by Mo during exchange (eq 1). The data
in Figure 9 also show that Mo species catalyze D2 dissociation,
the rate-determining step in D2/O-H exchange, and thus
27Al NMR spectroscopy can probe the local structure of Al
cations in Mo/H-ZSM5 samples. Tetrahedral Al centers in the
zeolite framework bonded to terminal O-H groups have a
chemical shift of 56 ppm; extraframework octahedral Al centers
(in ZSM5 and in Al2O3) show no shift (δ ) 0 ppm), while
crystalline Al2(MoO4)3 (octahedral Al, tetrahedral Mo) shows
a line at -13 ppm.15 27Al NMR spectra for MoOx/H-ZSM5
samples (Figure 7) showed that the density of unperturbed
framework Al centers (corresponding to Bronsted acid sites)

5792 J. Phys. Chem. B, Vol. 103, No. 28, 1999
Borry et al.
Figure 10. Rate of HD desorption during H2(g)/O-D(s) isotopic
Figure 8. Production rate of H2, HD, and D2O during temperature-
programmed D2(g)/O-H(s) isotopic exchange on 3.6 wt % Mo/
H-ZSM5 (0.3 g, 100 cm3/min 5% D2/Ar, 101 kPa).
exchange on Mo/H-ZSM5 (0.3 g, 100 cm3/min 5% D2/Ar, 101 kPa).
Figure 11. Number of acid sites (H) or framework Al (AlF) remaining
per total Al for series of Mo/H-ZSM5 samples after TPO and TPR
treatment to 973 K (0.5 h) measured by 27Al NMR, D2/O-H, and H2/
O-D (0.45 Mo/Al sample prepared via MoO3 vapor exchange at 950
K, 96 h).
Figure 9. Rate of HD desorption during D2(g)/O-H(s) isotopic
exchange on Mo/H-ZSM5 (0.3 g, 100 cm3/min 5% D2/Ar, 101 kPa).
decrease the temperature required for D2-OH exchange by
about 200 K.
The formation of D2O during D2/O-H exchange above 500
K (Figure 8) reflects some reduction of Mo6+ species. This
process influences the calculated exchange values, because both
H2O and HDO form via D2O exchange with OH groups. H2O
and HDO contain H atoms from surface O-H groups, but their
concentrations and isotopic contents are difficult to measure
because of water adsorption and exchange on the heated walls
of the cell and the transfer lines. These effects were shown to
be small by cooling the samples to 300 K after D2(g)/O(s)-H
exchange, and then heating to 973 K in flowing H2; the
measured amounts of HD and D2 formed (Figure 10, denoted
as H2(g)/O(s)-D exchange) gave densities of residual OD
groups that were only slightly different from those measured
from the initial D2-OH exchange.
D2/O-H and H2/O-D exchange data show that each Mo
added to H-ZSM5 replaces 1.0 ( 0.2 OH groups during
exchange from physical mixtures at 973 K (Figure 11).29
Tetrahedral Al atoms containing an OH group (AlF) give a 27Al
NMR line that decreases in intensity as Mo content increases.
These data allow us to calculate the fraction of the Al centers
that interact with MoOx species instead of OH groups (Figure
7), using a value of 0.92 for the AlF/Al ratio in a H-ZSM5
sample pretreated in air at 773 K for 1 h (Table 4).30 The results
from D2-OH, H2-OD, and 27Al NMR are in excellent
agreement (Figure 11); they show that each Mo replaces one
OH group, as expected for (Mo2O5)2+ dimers bridging two
cation exchange sites.
The sum of the OH groups removed as H2O during exchange
and those remaining (from D2/O-H data) should correspond
to the total number of OH groups in the starting H-ZSM5
samples (Table 4). On H-ZSM5, the two TPO experiments and
the D2-OH exchange add to a value of 0.92 OH/Al (Table 4).30
All Mo samples had values of 0.77-0.86 total OH/Al, in
reasonable agreement with the value of 0.92 measured on
H-ZSM5 (Table 4).
Discussion
X-ray diffraction, N2 physisorption, and 27Al NMR methods
show that after treatment in air at 973 K, MoOx/H-ZSM5

Mo-Exchanged H-ZSM5 Catalysts
J. Phys. Chem. B, Vol. 103, No. 28, 1999 5793
Figure 14. H2O desorption rate during TPO of MoO3/H-ZSM5
mixtures at 740 and 840 K (from Figure 4). (*) Mo/Al ratios are as
prepared. Mo concentrations after TPO are shown in Table 3.
Figure 12. Formation of Mo/H-ZSM5 active sites from MoO3/H-ZSM5
physical mixtures.
vibrations overcome this binding and bilayers slide over each
other until they anchor at Al sites on external ZSM5 surfaces.38
MoOx species may also migrate into zeolite channels at these
temperatures, but they do not react with O-H groups, as shown
by the low rate of H2O formation below ∼750 K and by the
lack of dependence of this rate on the Mo content (Figure 14).
After the external ZSM5 surface is covered completely with
a MoO3 layer above 773 K, further spreading cannot decrease
the surface energy. In our study, samples with more than 4 wt
% Mo contain more Mo than that required to form a MoO3
monolayer on the external surface area of the ZSM5 crystals
used. The excess MoOx can sublime as (MoO3)n, as expected
from the MoO3 vapor pressure and from the observed loss of
Mo in samples with high Mo content (Table 3). Excess Mo
can also migrate into ZSM5 channels as isolated MoOx species
(as shown by the disappearance of XRD peaks and by the
observed increase of H2O desorption rate above 773 K; Figure
14) or extract Al from the ZSM5 framework to form Al2(MoO4)3
(detected in the 8 wt % Mo sample by 27Al NMR and confirmed
by the unusual shape of the H2O desorption curve for this sample
above 850 K; Figure 4).
The strong Al-O-Mo interactions that lead to the spreading
of MoO3 on external zeolite surfaces also prevent MoO3
sublimation. Thus, physical mixtures with less Mo than that
required to form a single MoO3 layer (1-2 wt % Mo) do not
lose Mo by sublimation. The mixture containing 4 wt % Mo
required treatment in air at 773 K for 24 h in order to attain
full spreading of a MoOx layer and to avoid sublimation during
exchange (Table 2, program III). Even after this treatment, the
8 wt % Mo/H-ZSM5 sample lost about 20% of its MoO3 by
sublimation (Table 3). This shows that the maximum MoO3
coverage for the external (mesoporous) surface area of the ZSM5
used (24 m2/g) corresponds to about 4 wt % Mo.39
Figure 13. Structure of bulk MoO3 and solid-state migration on
H-ZSM5 surface.
samples prepared from MoO3-H-ZSM5 physical mixtures lead
to exchanged Mo species similar to those obtained from aqueous
exchange methods. Low pH conditions may alter the density
and nature of external OH groups on zeolite crystals,31 which
can lead to MoOx binding and may account for the high external
Mo concentrations detected by X-ray photoelectron spectroscopy
on samples prepared by aqueous exchange.26,31 Methane conver-
sion rates and selectivities are similar on samples prepared by
the two methods, suggesting that active site densities and
aromatization pathways are similar in MoOx/H-ZSM5 samples
prepared from physical mixtures (this study) and in those
prepared using aqueous exchange.1,3,6,28,32,33
Migration of MoOx Species during Treatment in Air. Our
results are consistent with the processes depicted schematically
in Figure 12. Below 623 K, X-ray diffraction shows that MoO3
and H-ZSM5 crystallites remain intact. Between 623 K and 773
K, MoO3 crystallites disappear as MoOx migrates onto external
zeolite surfaces. Kno¨zinger, et al.34 have proposed that MoO3
spreads on Al2O3 surfaces via an “unrolling carpet” mechanism.
Above 600 K, MoO3 layers migrate until they reach AlOx sites,
where they anchor and become immobile (Figure 13). This
process decreases surface energies as MoO3 forms strong bonds
with Al2O3; similar processes were not detected on SiO2.35 These
data and the stability of Al2(MoO4)3 suggest that Al-O-Mo
bonds are stronger than Si-O-Mo bonds.36
H2O evolution rates (per Al or g-catalyst) during air treatment
of MoO3/H-ZSM5 mixtures at high temperatures (Figure 4)
reflect the rate of replacement of H+ with cationic Mo oxo
species. This rate of exchange in turn depends on the mobility
of migrating Mo species and on the number of ZSM5 channels
available for transport. Below 740 K, H2O evolution rates are
independent of Mo content (Figure 14), indicating that MoOx
species do not undergo the condensation reactions that anchor
them at exchange sites. Between 750 and 850 K, H2O evolution
rates become proportional to the Mo content for samples with
MoO3 consists of MoO6 octahedra with displaced central Mo
atoms.37 MoO6 octahedra form bilayers with van der Waals
interlayer binding (Figure 13). Between 623 and 773 K,

5794 J. Phys. Chem. B, Vol. 103, No. 28, 1999
Borry et al.
1.0-3.6 wt % Mo (Mo/Al ) 0.11-0.37; Figure 14). These data
suggest that zeolite channels become accessible to the external
MoO3 surface layer in proportion to the amount of Mo spread
as a layer on the external zeolite surface (at least up to 3.6 wt
% Mo). At higher Mo contents, H2O desorption rates remain
constant as temperatures increase from 750 and 850 K,
suggesting that the external MoO3 exceeds one monolayer and
that all channel openings become accessible to MoO3; then,
condensation reactions become limited by rate of Mo migration
within ZSM5 channels. These data are consistent with the
scheme in eq 5 and with expected formation of a single MoO3
layer in the 3.6 wt % Mo samples for the external area available
in the H-ZSM5 crystals used in the physical mixtures.
corresponding to tetrahedral Mo6+ centers were detected previ-
ously but incorrectly assigned to (MoO4)2- monomers bridging
two exchange sites,6 a stoichiometry that is not consistent with
our results. The size of (Mo2O7)2- ditetrahedra in tetra-n-
butylammonium dimolybdate (∼5.7 Å)40 and simple geometrical
considerations based on a ZSM5 crystal structure with randomly
distributed Al ions showed that about 60% of the Al atoms in
H-ZSM5 (Si/Al ) 14) can reside sufficiently close to another
Al to allow Mo dimers to bridge Al pairs.41 The exclusive
presence of (Mo2O5)2+ dimers is then possible at all Mo/Al ratios
below 0.6 (i.e., <5 wt % Mo for our Si/Al zeolite ratio).
27Al NMR shows that all Al centers in vapor-exchanged Mo/
H-ZSM5 have been displaced from framework crystallographic
positions (Figure 7). Thus, either Al centers are interacting with
(Mo2O5)2+ dimers or they have been extracted from the
framework via condensation of two O-H groups at next nearest
neighbor Al atoms. All Al-OH species in H-ZSM5 ultimately
find the next nearest neighbor required to anchor a Mo dimer
or to dehydrate during the extended treatment in air (96 h at
950 K) required for vapor-phase exchange. The apparent high
density of Al site pairs in this sample may reflect the migration
On the sample with the highest Mo content (7.9 wt % Mo in
mixture; 6.3 wt % after exchange), multiple MoO3 layers can
form on external zeolite surfaces. Above 843 K, AlOx species
can be extracted from the aluminosilicate and dissolved into
liquidlike MoO3 layers, within which they can react to form
stable Al2(MoO4)3 domains. Al2(MoO4)3 was detected by 27Al
NMR and X-ray diffraction15 in samples with high Mo content.
In our study, Al2(MoO4)3 was not detected in samples with less
than 4 wt % Mo, because MoO3 monolayers apparently lack
the local three-dimensional structure and the thermodynamic
incentive required to stabilize Al in Al2(MoO4)3 structures.
The extraction of AlOx species from the aluminosilicate
framework to form Al2(MoO4)3 causes the structural collapse
detected by X-ray diffraction and N2 physisorption measure-
ments in the Mo/H-ZSM5 sample with 6.3 wt % Mo.27 High
temperatures alone do not cause the observed structural collapse,
because “vapor-exchanged” (4.3 wt % Mo) samples were treated
at 950 K for 96 h without significant loss of crystallinity and
27Al NMR did not detect any Al2(MoO4)3. It appears that high
local MoOx concentrations are required in order to extract Al
from the aluminosilicate framework. The sample with 6.3 wt
% Mo (7.9 wt % Mo in starting mixture) and the vapor-
exchanged sample were exposed to similar MoOx concentrations
(7.9 wt % vs 7.3 wt % Mo), but vapor-exchanged samples were
contacted by (MoO3)n (n ∼ 3-4) oligomers at a pressure of
about 25 Pa (250 ppm),11 while the 8 wt % Mo sample contained
significant interfacial area between liquidlike bulk MoO3 and
zeolite surfaces. Thus, local MoO3 concentrations are lower
during vapor exchange synthesis (Figure 6), because (MoO3)n
oligomers adsorb onto the ZSM5 surface, migrate within zeolite
channels, and bind irreversibly at sites where they form Mo-
O-Al bonds before high local MoO3 concentrations are reached.
Exchange of Cationic Mo Oxo Species via Condensation
Reactions. The amount of water evolved during exchange and
the number of O-H groups remaining after exchange show that
each migrating MoOx replaces one H+ in H-ZSM5 (Figures 5
and 11). These data and the required charge balance suggest
that (Mo2O5)2+ dimers interacting with two exchange sites form
during exchange (eq 1), which acquire the ditetrahedral structure
of (Mo2O7)2- dimers when two framework oxygen atoms
provide the anchoring sites for such dimers. Infrared bands
of Al centers by exchange with Si atoms above 773 K,42
a
process that stops when thermodynamically stable Al-O-Al
or Mo-O-Al bonds form via condensation reactions. The
required migration can occur via “hopping” into lattice vacancies
or defects, via thermally-induced exchange of Si with Al, or
by H2O-assisted diffusion pathways. When Al atoms reside
momentarily near another one (as a next nearest neighbor or
across a channel), their OH groups react to form H2O and an
O2- vacancy, or exchange with two (MoO2OH)+, which can
then condense to form a (Mo2O5)2+ dimer containing a bridging
oxygen atom. Either event stabilizes Al-O-Al or Al-O-Mo
site pairs slightly detached from framework positions, distorted
from tetrahedral symmetry, and invisible by 27Al NMR.
27Al NMR data (Figure 7) confirmed the nature of the
exchanged Mo species and the details of the migration and
exchange processes. The intensity of the line at 56 ppm,
corresponding to tetrahedral Al center in the ZSM5 framework,
decreased as the Mo content increased for samples with less
than 4 wt % Mo; this decrease in intensity was not compensated
by the detection of new or more intense NMR lines. At higher
Mo contents, the framework Al line was not detected, but strong
lines corresponding to Al2(MoO4)3 appear. No NMR lines were
detected (δ ) -416 and 543 ppm) in vapor-exchanged (4.3 wt
%) Mo/H-ZSM5 samples. This shows that vapor exchange
creates neither Al2O3 nor Al2(MoO4)3 in samples with this Mo
content. The anchoring of (Mo2O5)2+ dimers appears to form
Mo-O-Al bonds that distort the tetrahedral structure of
framework Al centers, leading to the typical broadening of lines
for quadrupolar Al nuclei with decreasing symmetry.43 The Al
symmetry at exchange sites bonded to (Mo2O5)2+ dimers differs
markedly from that observed in known Al compounds. Al3+
centers have tetrahedral coordination in aluminosilicate frame-
works and octahedral symmetry in γ-Al2O3. Al2(MoO4)3 has

Mo-Exchanged H-ZSM5 Catalysts
J. Phys. Chem. B, Vol. 103, No. 28, 1999 5795
Al ratios, even though the zeolite Si/Al ratio varied over a wide
range (12.5-50) in these samples.
Conclusion
CH4 conversion rates on catalysts prepared from mixtures of
MoO3 and H-ZSM5 powders are very similar to those obtained
on samples prepared via aqueous impregnation-exchange
methods. The simpler synthesis from physical mixtures allows
measurements of the kinetics of formation and of the stoichi-
ometry of exchanged MoOx species. H2O evolved during heating
reflects the kinetics of condensation pathways that anchor MoOx
at zeolite exchange sites. Isotopic equilibration between D2 and
residual OH groups after exchange gives the fraction of H+
species replaced by exchanged Mo oxo cations. The mechanism
of exchange involves the initial formation of an external MoO3
monolayer on zeolite crystals via surface migration at 623-
773 K. When the Mo content exceeds that required to form a
monolayer (∼4-5 wt % Mo in the H-ZSM5 used), MoOx
species are lost as (MoO3)n oligomers via sublimation or as
unreducible and inactive Al2(MoO4)3 domains via reactions with
framework Al atoms. Between 773 and 973 K, surface and gas-
phase transport paths lead to migration of MoOx species into
zeolite channels and to reactions with OH groups to form MoO2-
(OH)+ species that condense quickly to form H2O and a strong
Mo-O-Al anchoring bond. The amount of H2O evolved during
exchange and the number of H+ remaining after exchange are
consistent with the replacement of one H+ by each exchanged
Mo. This stoichiometry and the requirement for charge neutrality
during exchange lead to the proposal that Mo species exist as
(Mo2O5)2+ ditetrahedra interacting with two zeolite exchange
sites. These (Mo2O5)2+ dimers reduce and carburize during
methane reactions to form the active MoCx sites required for
catalytic C-H bond activation. Maximum rates are observed
on catalysts with intermediate Mo/Al ratios (∼0.4), because both
exchanged cations and residual Bronsted acid sites are required
for the conversion of methane to C6+ aromatics.
Figure 15. Literature reports of CH4 conversion, normalized by the
maximum conversion reported in each study, as a function of Mo/Al
ratio. The solid lines connect data from a single study.
octahedral Al centers and tetrahedral Mo centers, while (Mo2O5)2+
dimers located at two cation exchange sites would have distorted
tetrahedral Al centers bound to distorted tetrahedral Mo6+; it is
this required distortion that leads to the broadening of NMR
lines after exchange or dehydroxylation of zeolites.
Formation of Active Sites during CH4 Aromatization. The
predominant formation of COx, H2O, and H2 during initial
contact of MoOx/H-ZSM5 with CH4 at 950 K (Figure 1) and
the increase in hydrocarbon formation rates with increasing time
on stream suggest that active sites are created by the reduction
and carburization of (Mo2O5)2+ species that are not active for
methane conversion to hydrocarbons. Sites in (Mo2O5)2+ dimers
react with C-H bonds in CH4, but they cannot desorb the
resulting fragments, which form instead MoCx species via
subsequent dehydrogenation steps. Catalytic CH4 activation
begins only after reduced Mo species are passivated by the
formation of these carbide species, which are then able to desorb
reaction products and complete a turnover. The high melting
point and the low volatility of Mo suboxides and carbides11
suggest that migration is unlikely to occur after reduction-
carburization of exchanged (Mo2O5)2+. These conclusions are
consistent with detailed characterization studies reported else-
where.44,47
Acknowledgment. R.B. was supported by a National Sci-
ence Foundation Fellowship. Y.-H.K. was supported by the
Korean Science and Engineering Foundation (KOSEF). The
project was funded by the Federal Energy Technology Center
of the U.S. Department of Energy (DE-AC03-76SF00098) under
the technical supervision of Dr. Daniel Driscoll.
Exchanged (Mo2O5)2+ species are precursors to CH4 activa-
tion sites in Mo/H-ZSM5 catalysts. Therefore, CH4 conversion
rates increase with increasing extent of Mo exchange during
synthesis. Reaction pathways, however, also require Bronsted
acid sites, provided by the remaining H+ species in ZSM5. These
acid sites catalyze chain growth and cyclization reactions of
the initial ethylene products formed in CH4 activation steps;
thus, acid sites shift the product distributions toward aromatics,
which are favored over alkenes by thermodynamics. As a result,
the fraction of the Al sites exchanged by Mo (Mo/Al), and not
the Mo content, becomes the relevant parameter determining
catalytic rates. Our H-ZSM5 samples (Si/Al ) 14.3) lose about
40% of the framework Al sites during air treatment at 973 K;
for these samples, the optimum Mo concentration for CH4
aromatization is about 0.4 Mo/Al (Figure 15) or about 0.7 Mo/
AlF if only framework Al are assumed to stabilize H+. In these
samples, the remaining Bronsted acid sites are sufficient to
catalyze C2H4 aromatization reactions to near-equilibrium levels.
Figure 15 shows our data and those from previous stud-
ies,1,6,28,45,46 as CH4 reaction rates vs Mo/Al ratio. All studies
detect maximum relative rates on catalysts with 0.3-0.5 Mo/
References and Notes
(1) Wang, L.; Tao, L.; Xie, M.; Xu, G. Catal. Lett. 1993, 21, 35.
(2) Solymosi, F.; Erdohelyi, A.; Szoke, A. Catal. Lett. 1995, 32, 43.
(3) Wang, D.; Lunsford, J. H.; Rosynek, M. P. Top. Catal. 1996, 3,
289.
(4) Liu, S.; Dong, Q.; Ohnishi, R.; Ichikawa, M. J. Chem. Soc., Chem.
Commun. 1997, 1455.
(5) Stakheev, A. Y.; Khodakov, A. Y.; Kustov, L. M.; Kazansky, V.
B.; Minachev, Kh. M. Zeolites 1992, 12, 866.
(6) Xu, Y.; Shu, Y.; Liu, S.; Huang, J.; Guo, X. Catal. Lett. 1995, 35,
233.
(7) Dong, Y.; Liu, S.; Zhang, Q.; Liu, J.; Yang, K. Acta Pet. Sin. (Pet.
Process. Sect.) 1992, 8, 66.
(8) Xu, Y.; Liu, S.; Wang, L.; Xie, M.; Guo, X. Catal. Lett. 1995,
135, 135.
(9) Satterfield, C. N. Heterogeneous Catalysis in Industrial Practice,
2nd ed.; McGraw-Hill, Inc.: New York, 1991.
(10) Brewer, L. Molybdenum: Physico-chemical properties of its
compounds and alloys; International Atomic Energy Agency: Vienna, 1980;
Vol. 7.
(11) Knacke, O., Kubaschewski, O., Hesselmann, K., Eds. Thermo-
chemical Properties of Inorganic Substances II, 2nd ed.,; Springer-Verlag:
Berlin, 1991; Vol. 2.
(12) Vahel, A.; Hubener, S.; Eichler, B. Radiochim. Acta 1995, 69, 233.

5796 J. Phys. Chem. B, Vol. 103, No. 28, 1999
Borry et al.
(13) Vahel, A.; Hubener, S.; Dressler, R.; Eichler, B.; Turler, A.
Radiochim. Acta 1997, 78, 53.
(14) Minming, H.; Howe, R. F. J. Catal. 1987, 108, 283.
(15) Liu, W.; Xu, Y.; Wong, S.-T.; Wang, L.; Qiu, J.; Yang, N. J. Mol.
Catal. A 1997, 120, 257.
(16) Biscardi, J. A.; Meitzner, G. D.; Iglesia, E. J. Catal. 1998, 179,
192.
(17) H-ZSM5 samples (0% Mo) were heated to higher temperatures (973
K) in order to ensure complete isotopic exchange.
(18) The relaxation delay was 1.0 s, the sweep width was 50 kHz, and
all chemical shifts were referred to a dilute Al(NO3)3 solution. A 90°/180°
echo pulse sequence was used and data were collected for 0.35 h. Increasing
the relaxation delay between acquisitions did not increase the 27Al NMR
signal, indicating that 1.0 s was sufficient to allow full relaxation of all Al
spin states. In this case, the NMR signal peak area, integrated using Peakfit
(Jandel Scientific), is proportional to the number of Al centers represented
in that peak.
(19) Kentgens, A. P. M.; Scholle, K. F. M. G. J.; Veeman, W. S. J.
Phys. Chem. 1983, 87, 4357.
(20) Deng, F.; Du, Y.; Ye, C.; Wang, J.; Ding, T.; Li, H. J. Phys. Chem.
1995, 99, 15208.
(21) Experiments using unheated lines overestimated the amount of
carbon deposited on the catalyst because of condensation of naphthalene
and heavier products in the gas transfer lines.
(22) HP 6890 gas chromatograph (Hewlett-Packard) equipped with a
Carboxen 1000 packed column (3.2 mm × 2 m, Supelco) connected to a
thermal conductivity detector and with a HP-1 capillary column (methyl
silicone, 0.32 mm × 50 m × 1.05 mm, Hewlett-Packard) connected to a
flame ionization detector.
(23) All gases were >99.99% purity (from Praxair and Matheson) and
were passed through molecular sieve (13X and 5A) traps before use. Flow
rates were metered using electronic mass flow controllers (Porter Instru-
ments).
(24) Wolthuizen, J. P.; van den Berg, J. P.; can Hoof, J. H. C. In
Catalysis by Zeolites, Imelik, B., Ed.; Elsevier Scientific Publishing Co.:
Amsterdam, 1980.
(25) The initial concentration of Mo for this sample was 4.0 wt %, but
after temperature-programmed treatment in air up to 973 K the sample
contained 3.6 wt % Mo (Table 3).
(26) Wang, D.; Lunsford, J. H.; Rosynek, M. P. J. Catal. 1997, 169,
347.
(27) The initial Mo concentration for this sample was 8 wt %, but after
temperature-programmed treatment in air the sample contained 6.3 wt %
Mo (Table 3).
was during the initial D2(g)/O(s)-H exchange. One explanation for this
discrepancy is that the mass fragmentation pattern of the D2O reduction
peak during the first D2(g)/O(s)-H exchange contains some H2O.
(30) This value is less than unity because some acid sites were lost during
the conversion of NH4-ZSM5 to H-ZSM5 at 773 K. This sample also lost
additional O-H groups during the initial D2(g)/O(s)-H exchange, because
it required heating to 973 K in order to complete the isotopic exchange.
The H/Al values from the second H2(g)/O(s)-D exchange were used for
Mo-containing samples in the calculations shown in Table 4, because this
led to a more accurate hydrogen atom balance.
(31) Zhang, J.-Z.; Long, M. A.; Howe, R. F. Catal. Today 1998, 44,
293.
(32) Lunsford, J. H.; Rosynek, M. P.; Wang, D. Presented at the 4th
International Natural Gas Symposium, Kruger National Park, South Africa,
1995.
(33) Solymosi, F.; Szoke, A. Symposium on Heterogeneous Hydrocarbon
Oxidation, 211th National Meeting, American Chemical Society (New
Orleans, LA), 1996; p 147.
(34) Mestl, G.; Kno¨zinger, H. Langmuir 1998, 14, 3964.
(35) Regalbuto, J. R.; Ha, J.-W. Catal. Lett. 1994, 29, 189.
(36) The fact that MoOx spreading occurs on external ZSM5 surfaces
suggests that the Al content in H-ZSM5 has a high enough surface density
of Al sites to form (MoOx)n oligomers that bridge Al species on the external
surface. Si-O-Mo bonds do not appear to be strong enough to survive air
treatment at 950 K, since MoO3 vapor exchange preparation (Figure 6)
results in only 4.3 wt % Mo, which is much less than expected if the external
ZSM5 surface area was covered and framework OH groups were exchanged
with Mo.
(37) Andersson, G.; Magneli, A. Acta Chem. Scand. 1950, 4, 793.
(38) MoO3 may eventually spread into a monolayer, but below 773 K
the weak bonds between bilayers are more likely to break than the strong
Mo-O-Mo bonds within bilayers.
(39) This Mo content (4 wt %) corresponds to about 19 m2/g for a MoO3
bilayer or 38 m2/g for a monolayer.
(40) Pope, M. T. Heteropoly and Isopoly Oxometalates; Inorganic
Chemistry Concepts; Jorgensen, C. K., Ed.; Springer-Verlag: Berlin, 1983.
(41) Feng, X. B.; Hall, W. K. Catal. Lett. 1997, 46, 11.
(42) The Tammann temperature of quartz SiO2 is 940 K, so mobility of
ions in the O-Si-O lattice of ZSM5 should occur between 673 and 973
K.
(43) Engelhardt, G.; Michel, D. High-resolution solid-state NMR of
silicates and zeolites; Wiley, New York, 1987.
(44) Borry, R. W.; Li, W.; Meitzner, G.; Iglesia, E. J. Catal. submitted
for publication.
(28) Chen, L.; Lin, L.; Xu, Z.; Li, X.; Zhang, T. J. Catal. 1995, 157,
190.
(29) In principle, the number of D atoms counted during H2(g)/O(s)-D
exchange should equal the number of H atoms detected during D2(g)/
O(s)-H exchange. The latter, however, was larger for all samples containing
Mo, because more D2 was formed during H2(g)/O(s)-D exchange than H2
(45) Xu, Y.; Liu, W.; Wong, S.-T.; Wang, L.; Guo, X. Catal. Lett. 1996,
40, 207.
(46) Pinglian, T.; Zhusheng, X.; Tao, Z.; Laiyuan, C.; Liwu, L. React.
Kinet. Catal. Lett. 1997, 61, 391.
(47) Borry, R. W.; Lu, E. C.; Kim, Y. H.; Iglesia, E. Stud. Surf. Sci.
Catal. 1998, 119, 403.