J. Phys. Chem. 1992, 96, 8480-8485
Mordenlte Acldity: Dependence on the Si/AI Ratio and the Framework Aluminum
Topology. 2. Acidity Investlgations
Helmut Stach,* Jochen Jiinchen, Hans-Georg Jerschkewitz, Ursula Lohse, Barbara Parlitz,
Central Institute o f Physical Chemistry, 0- 11 99 Berlin, Germany
and Michael Hunger
Department of Physics, University o f Leipzig, Leipzig, Germany (Received: March 16, 1992;
In Final Form: June 19, 1992)
The acidity of a series of mordenites dealuminated by acid refluxing was studied by calorimetric measurements of the NH3
chemisorption, proton magic angle spinning nuclear magnetic mnance spectroscopy, and temprature-programmed desorption
of NH3 and NH4+ ion exchange. The number of bridged OH group determined by ‘H MAS NMR corresponds to the
number of tetrahedrally coordinated Al atoms. Starting with a low A I content (a high Si/Al ratio), the number of the strong
acidic sites increases with increasing A1 number of the framework up to 4.6 A1 atoms/u.c. (Si/Al ratio of 9.5) and then
decreases. The decrease of the strong acidity above 4.6 A1 atoms/u.c. may be explained by the appearance of an aluminum
atom in the second coordination sphere of the Si-OH-A1 group. The experimentally found value of the maximum (at m
= 0.096) in the curve of the strong acidity versus the aluminum molar fraction coincides with a value calculated by Barthomeuf
from the topological A1 density of mordenite, thus confirming her theoretical concept of zeolite acidity.
Correlations between the catalytic activity and the acid strength
of H-mordensites are often found. Investigations of the acidity
were not always performed precisely, because only a few methods
allow determination of the acid strength of microporous catalysts
exactly. We investigated dealuminated H-mordenites (with Si/Alf
ratios between 7 and 48) using several methods to characterize
their physicochemical properties (see part 1). In this part, results
of the investigations of the H-mordenite acidity are presented using
four different methods to determine the number of acid sites.
Adsorption calorimetry was employed to measure the acid strength
and acid strength distribution of the H-mordenites studied ac-
’ Exdmge Capncity MepslwmeaQ. The ammonium
exchange was performed three times at 343 K as a batch process
with an excess of 0.2 N NH4N03 solution (pH = 5 . 6 ) . The
amount of the exchanged NH4 ions was obtained from the NH3
content using the Kjeldahl method. The results are given in Table
V (column 2) of part 1.
Temperaturehogrammed Desorption (TPD) of Ammonia.
Prior to the measurements the samples were activated at 773 K
in a helium stream for 1 h. After adsorption of ammonia from
a helium stream (3 vol %) at 393 K the samples were washed with
helium and heated under a stream of helium with a heating rate
of 12 K/min. The desorbed amount of NH3 was measured by
a thermal conductivity detector as well as by titration with 0.1
N H2S04. The TPD curves are depicted in Figures 7 and 8 and
the calculated amounts of desorbed ammonia are given in Table
V of part 1.
Calorimetric Measurements. For the measurement of the
differential molar heat of chemisorption of ammonia a microca-
lorimeter of Calvet-type was used. It was connected with a
volumetric sorption apparatus thus enabling the measurement of
the evolving heat and the adsorbed amount at the same time. All
samples were evacuated (p < 1 mPa) at elevated temperatures
(653 K) for about 15 h prior to the calorimetric experiments.
All measurements were made at 423 K. Earlier our systematic
studies’ of temperature dependence of the ammonia adsorption
showed, in accordance with literature data,* that at lower tem-
peratures not all strong acidic sites were obtained and at a much
*Corresponding author. Present address: Adlershofer Umweltschutz-
technik und Forschungsgesellschaft mbH, Rudower Chaussee 5, 0-1 199
higher temperature decomposition of the formed NH4+ ions could
occur. Adsorption equilibrium was attained within 4-24 h de-
pending on the amount adsorbed. It was controlled by the heat
transition (thermokinetic curve) as well as the mass transition
(pressure recording by a capacity membrane manometer of
Baratron type). The mean error of the heat of adsorption
amounted to approximately 2%. The calorimetric curves are
presented in Figure 3.
R d t S
Temperature-hogrammed NH3 Desorption. Figures 1 and 2
present some results on the temperature- programmed desorption
of ammonia. In accordance with data in the literat~re,~~
TPD spectra two separate peaks were found. While the maximum
temperature of the first peak (low-temperature peak) is practically
independent of the degree of dealumination, the maximum tem-
perature of the second (high-temperature peak) decreases with
decreasing Al content of the framework (see Table V or part 1).
A decrease of the peak maximum temperatures in the TPD spectra
can be carrelated with a decrease in the acid strength: Therefore,
it may be concluded that with a greater degree of dealumination
the acid strength of the dealuminated mordenites diminishes.
By investigating dealuminated zeolites of the HZSM-5 type,
we found that the spectra of the NH3 TPD also consist of two
peaks.’ A comparison of the total number of strong acid sites
(measured calorimetrically) with the TPD results showed that only
the population of the high-temperature peaks corresponded to the
calorimetric data. Thus it could be concluded that the low-tem-
perature peak is due not to strong but to weak acid sites. This
seems justifiable in the case of the mordenites, too. As will be
shown later, the number of strong acid sites derived from the
high-temperature peak and from calorimetric measurements of
ammonia adsorption are indeed in good agreement.
As is seen in Figure 2, the high-temperature peaks are not
symmetric and possess some shoulders especially for mlites with
a high A1 content in the framework. Therefore, it appears that
ammonia was released from more than two different types of sites
with different activation energy of desorption, i.e., different acid
Similar results were found by Karge et a1.4 after investigating
dealuminated mordenites by TPD of ammonia and pyridine.
Using a Gaussian distribution function for the probability density
of the activation energy of desorption and under special as-
sumptions, a deconvolution of the spectra was possible. Best results
were obtained assuming four different types of sites (two Lewis
and two Bransted sites) differing in number and acid strength.
0022-3654/92/2096-8480$03.00/0 0 1992 American Chemical Society
Mordenite Acidity. 2
The Journal of Physical Chemistry, Vol. 96, No. 21, 1992 8481
473 673 073
Figure 1. Ammonia TPD spectra of HM-0, HM-Ob and HM-Oc (- - -,
HM-0, - HM-Ob; -.-, HM-OC).
Figure 2. Ammonia TPD spectra of selected dealuminated mordenites
(-, HM-Ob,..., HM-2,-.-, HM-4b, ---,
Though the experimental TPD curves are well fitted by the
mathematical procedure mentioned, it seems possible that the
calculated most frequent activation energies for the ammonia
desorption not only reflect the acid strength but may also be
influenced by diffusion (as it is the case in nearly all experiments
which use dynamic methods).
More precise information on the acidity (especially about the
acid strength) should be obtainable by direct exact measurement
of the interaction energy of the acid sites with basic molecules,
for example, with ammonia.
c.brimetric MepslpemeaQ. In Figure 3, the differential molar
heats of ammonia chemisorption are plotted as a function of the
adsorbed number of the basic molecules. In preliminary exper-
iments' and in accordance with the literature:v8 it was found that
a temperature of 423 K was sufficient to obtain reproducible data.
Therefore, a l l measurements were performed at that temperature.
It follows from Figure 3 that most of the curves of the dealu-
minated mordenites show a continuous decrease of the differential
molar heats of ammonia adsorption with increasing adsorbed
amount. This indicates the existence of a rather broad distribution
of adsorption sites with different interaction energy, i.e., different
acid strength. The same conclusion was derived (vide supra) from
the TPD investigation.
Of special interest is the curve of the heat of ammonia ad-
sorption on Na-mordenite (Figure 3). In this sample all acid sites
are neutralized by the Na+ ions with the consequence that the
calorimetrically measured interaction energies are smaller if
1 ~-~ I
Figure 3. Calorimetric differential molar heats of ammonia adsorption
at 423 K (A, NaM; A, HM-Ob; m, HM-2; 0,
HM-4b; 0, 0,
compared with the H form of the mordenites. The curve is
characterized by a broad step at about 80 kJ/mol in accordance
with data given in ref 9. Comparing the curves depicted in Figure
3 with those found in calorimetric measurements of ammonia
adsorption on synthesized mordenites (with nearly the same Si/AI
ratio) a very similar behavior can be seen.2 We conclude that
an adsorption heat of about 80 kJ/mol corresponds to the in-
teraction of the ammonia molecules with the oxygen atoms of the
framework (and in the case of NaM as well with the sodium
cations), i.e., to the physical adsorption. Diminishing the number
of aluminum atoms in the framework results in a decreasing field
strength and hence in a reduced heat of physical adsorption as
is indeed seen in Figure 3.
Heats of adsorption >80 kJ/mol are due to the interaction of
ammonia with acid sites."l0 The heat of adsorption is proportional
to the acid strength: the greater the heat, the higher the acid
strength. The number of the chemisorbed ammonia molecules
is determined by the intercept of the heat curve with the 80 kJ/mol
mark. As is seen in Figure 3 the strength of the acid sites depends
on the Si/Al ratio. The initial heat of chemisorption amounts
to about 160 kJ/mol for the sample HM-Ob, decreases with rising
Si/Al ratio to about 140 kJ/mol, and then increases again.
Calorimetric data in the literature for the initial heats of adsorption
in the same system are still higher and reach values of about
160-170 kJ/mo1?.8J"12 These differences may be due to different
experimental conditions or differences in the investigated mor-
A comparison of the initial heat of chemisorption of ammonia
on dealuminated H-mordenites with those found by us for HY
zeolites13 reveals a substantial difference. The interaction energy
of ammonia and therefore the acid strength of the mordenites at
low coverages are higher than the corresponding data of the HY
molecular sieves. This agrees with the well-known behavior of
both zeolites in catalytic reactions1"18 and the conclusions drawn
from the investigations of the acidity by different methods.I9 If
the results of the acid strength of dealuminated mordenites here
given are compared with our measurements of the acidity of
dealuminated HZSM-5 zeolites:0 it can be found the differences
will be small. Concerning the overall initial acid strengths the
H forms of the dealuminated zeolites investigated may be arranged
in the following rank HM > HZSM-5 > HY.
Mluenee o f the Si/& Ratio on the Total Number of Acid Sites.
In Figure 4 the total number of the acid sites (derived by TPD,
NH, exchange, o r calorimetric measurements) of the dealuminated
mordenites are plotted against the tetrahedral aluminum content
of the unit cell. Additionally, calorimetric data of Tsutsumi et
aL2 are included. Our results show a linear relationship coinciding
with the diagonal. The authors attribute the systematic deviations
of the results given in ref 2 from the diagonal to a partial con-
version of Bransted into Lewis sites, or an extraction of aluminum
atoms from the framework. From the strong linearity of our
results in the figure, it can be concluded that every acid site
corresponds to a framework aluminum atom and that all acid sites
Molecuies / u c
8482 The Journal of Physical Chemistry, Vol. 96, No. 21, 1992
Stach et al.
Figure 4. Total number of acid site (per unit cell) dependence on the Allv
content/u.c. of H-mordenites from different methods of measurement (0,
calorimetry; 0, calorimetry data from ref 2; A, TPD; 0, NH4+-exchange
TABLE I: CompuisOn of tbe TOW Number of Br00sted Acid Sites
Obtained from Different E x p e r i m e n t a l Methods
sample (mmol/g) (mmol/g) (mmol/g) (mmol/g) N,4v/Al[
HM-Ob 2.04 1.93
HM-1 1.61 1.80
HM-2 1.56 1.67
HM-3 1.26 1.40
HM-4b 0.86 0.81
HM-6 0.38 0.39
' Ammonium ion exchange capacity. Temperature programmed
desorption of ammonia.
Calorimetric measurement of ammonia
chemisorption. 'H Magic angle spinning nuclear magnetic resonance.
e Average value from Naa, N.6, N : .
N,e N P d
are of Bransted type. The same conclusion was drawn from our
measurements of the acidity of dealuminated Y zeolites using four
independent methods.13 Applying the 'H and 27Al MAS NMR
technique, Heifer and mworkersZ1 also proved the proportionality
between the decrease of framework aluminum and the number
of bridged hydroxyl groups acting as Bransted sites in Y zeolites.
But besides the data presented in ref 2, there are several further
investigations on dealuminated and synthesized mordenites with
different Si/Al ratio, showing that the N,/Alf ratio is not close
to 1 .O (as in our results, see Table I) but smaller. This follows
from Figure 5 on comparing the results of ammonia TPD and of
calorimetric measurements of the adsorption heat of ammonia
by several a u t h o r ~ , ~ , ~ * ~ with our data.
As follows from Figure 5, the total numbers of acid sites we
measured are larger (for the same number of A1 atoms per unit
cell) than those found by the authors mentioned. This is not
surprising in the c a s e of TPD measurements? because the authors
reported only the number of acid sites corresponding to the
high-temperature peak. The smaller total number of acid sites
derived from calorimetric measurements in refs 2 and 9 may be
due to difficulties in the correct determination of the Si/AI ratio
and the preparation of the H forms of aluminum-rich mordenites.
AllV / U.C.
Figure 5. Comparison of the total number of acid sites (mmol/g) as a
function of the AI'" content/u.c. of H-mordenites from different mea-
surements (own data: V, calorimetry; V, TPD, 0, 'H MAS NMR .,
calorimetry, ref 9; B, calorimetry, ref 2; 0,
TPD, ref 3).
While Sawa et al.3 reported that above an A1 atom content of
5.6 per U.C. the total number of acid sites decreases with increasing
amount of A1 atoms as did Klyachko et al.? Tsutsumi et a1.2 did
not find such an effect. We did not investigate H-mordenita with
more than 6 A1 atoms per unit cell. But one has to bear in mind
that the samples (reported in ref 3) with an A1 content above 5.6
(which is equal to a Si/Al ratio above 7.8) do not belong to the
same dealumination series but are of different origin.
Influence of the Si/Alf Ratio on the Acid Strength. From
theoretical investigations on zeolites with different Al content per
unit cell, an increase of the acid strength with rising Si/Al ratio
can be p r e d i ~ t e d . ~ ~ ~ ~ ~ M ~ r t i e r , ~ ~
negativity equilization, calculated the average partial charge of
the protons in the H forms of several zeolites (faujasites) and found
that the charge changed from 0.12 (Si/Al = 1) to 0.14 (Si/AI
= 2.5) and 0.18 (Si/Al = infinity). The increase of the ionic
character of the H atoms was explained by the relative increase
of the Si content of the framework, having a higher electroneg-
ativity (compared with Al), and thus shifting the electron density
from the proton to the oxygen atom. Therefore, the OH bond
in the Si-richer zeolites becomes more ionic and the acid strength
Quantum chemical calculations also allow the conclusion that
with rising Si/Al ratio the average T U T bond angles increase
as well. The consequence is a decrease of the deprotonation energy
of the hydroxyl group which results in a higher acidity of the
Following the models of Mortier and Sauer, we have to conclude
that the acid strength rises with increasing Si/Al ratios, Le., with
decreasing A1 content of the framework in the dealuminated
mordenites. The higher acid strength of the samples with a lower
Alf content should be reflected by a higher heat of ammonia
chemisorption and by a larger number of the acid sites w i t h a high
chemisorption heat. As follows from Figure 3, we find qualitatively
such a behavior. To compare the above-mentioned conclusions
with the results of our calorimetric measurements more quanti-
tatively, we defined three groups of sites with different acid
strengths (with q > 80 Icl/mol, q > 100 kJ/mol and q > 120
using the principles of electro-
TABLE II: Distribution of the Acid Strength of Dealuminated Mordenites
N . 6
'Total number of acid sites with 9 > 80 kJ/mol. bNumber of acid sites with 80 < 9 € 100 kJ/mol. <Number of acid sites with q > 100 kJ/mol.
dNumber of acid sites with 100 € q < 120 kJ/mol. 'Number of acid sites with 9 > 120 kJ/mol.
Mordenite Acidity. 2
47 23 15 11
Aif atoms ~u.c.
cell of dealuminated H-mordenites (0, N, with Q > 80 kJ/mol; A, N,
with Q > 100 kJ/mol; 0,
N, with Q > 120 kJ/mol; 0, data from ref 9 ) .
6. Acid strength distribution dependence on A I " content per unit
W/mol; see Table 11) and determined the corresponding number
of acid sites for every group. Whereas the theoretical investigations
considered an ideal crystal structure, our measurements concerned
the real structure of the mordenite samples with the consequence
that we discovered an overlapping of the sites with different acid
strength resulting in a broad distribution of the acid strength.
In Figure 6 the acid strength distribution dependence of the
dealuminated mordenites on the Si/AI ratio is depicted. (Included
also are results given in ref 8. These data agree well with our
finding.) As follows from Figure 6, the number of strong acid
sites indeed rises with increasing Si/Al ratio. But this corre-
spondence is valid only up to a Si/Al ratio of about 9. Above
this value the number of strong acid sites diminishes. An anal-
ogous dependence of the strong acid sites on the Si/Al ratio was
also found in our calorimetric investigations of dealuminated HY
~eolites.'~ It seems to be necessary that the theoretical studies
consider the local environment of the hydroxyl groups in the
framework more precisely, especially the arrangement of the Al
atom having a dominant influence on the acidity.
Following the model of Barthomeuf,2B the curves of the strong
acid sites in Figure 6 may be explained as follows: At a high Si/AI
ratio and a consequently low aluminum content, in the unit cell
nearly all AI atoms are isolated, possessing a high acid strength
(expressed by a heat of chemisorption with q > 120 kJ/mol). With
increasing number of AI in the lattice, the number of the strong
acid sites rises. If there are A1 atoms instead of Si atoms in the
second coordination sphere, the number of strong acid sites di-
minishes, although the total number of acid sites increases further.
The effect of next-nearest Al atoms on the acidity was interpreted
earlier by (i) an influence of close neighbors as in solution or (ii)
an overall effect comparable to collective properties in solution.z9
From quantum chemical studies it follows that due to the closeness
of the AI atoms, the Si-OH-A1 angles change, with the conse-
quence that the acidity decreases.30
Topological Density and Acid Strength. Several models were
developed to explain the influence of the A1 atom in the second
coordination sphere on the a~idity.~*-~~
that strong acidity is connected with isolated A1 atoms and (ii)
that strong acidity reaches a maximum value at that Si/Al ratio
of the lattice at which no next A1 neighbor exists in the second
coordination sphere. It should be of special interest to consider
the structure and topology of the zeolites since the arrangement
and distribution of the neighbors of a given Si-OH-A1 group
determines the acidity of the zeolites, as was pointed out by
From a practical point of view as well as from theoretical
interest, it is therefore important to determine the limiting Si/Al
ratio below which aluminum atoms should have no aluminum as
next nearest neighbors (NNN). 29Si MAS NMR spectroscopy
The main ideas are (i)
The Journal of Physical Chemistry, Vol. 96, No. 21, 1992 8483
5 10 15
m x 10'
Figure 7. Dependence of the number of strong acid sites on the mole
fraction of tetrahedrally coordinated Al atoms (0, own calorimetric data;
0, data from ref 9 ) .
seems to be the only experimental method to give information on
the Al distribution in the T sites of a zeolite. But its application
becomes difficult in the case of dealuminated materials, as has
been shown (cf. part 1 of this paper).
Barthomeuf showedz8 that the topology of zeolites gives in-
formation on the way in which the tetrahedra are interconnected
and thus how to determine the limiting Si/Al ratios. She started
with the topological density of the T atoms (TDz-5). These values
can easily be calculated from the coordination sequences.35 (In
mordenite the number of T atoms around an A104 tetrahedron
in the second layer amounts to 12, in faujasites only to 9.) Its
value is taken over the second to the fifth coordination layer of
tetrahedra around a T site divided by the maximum number of
T atoms. The TD2-5 values are independent of chemical com-
position, structure symmetry and unit cell.% Multiplying the TDS5
value by the molar fraction of aluminum (m), an aluminum
topological density may be derived (TDAJ. A limiting value of
TDd (limit TDJ can be calculated for zeolites below which none
of the A104 tetrahedra will have another A104 as next-nearest
neighbor. Under the assumption that in any zeolite structure the
limiting aluminum topological density is the same and using an
experimentally determined value of the limit TDAl= 2.65 X 10-*
(for faujasites), it is possible to calculate the limiting values of
the aluminum molar fraction (limit "")
structures. T h i s procedure allows one to calculate in advance the
limiting Si/AI ratio below which A1 atoms should have no alu-
minum as next-nearest neighbors.
Barthomeuf calculated a limiting m " N
mordenites. This number corresponds excellently with our ex-
perimental finding as follows from Figure 7.
In Figure 7 the number of very strong acid sites (q > 120
W/mol) is plotted against the aluminum molar fraction of the
dealuminated mordenites. Included are also results given by
Klyachko et al.9 Their calorimetric measurements of ammonia
chemisorption were performed on synthetic and dealuminated
mordenites. The agreement between both sets of calorimetric
measurements is very good. (It could be argued that only one
measured point may not be enough to prove the decrease in
curvature of strong acid sites, but as follows from Figure 7 the
independently measured value in the critical range of composition
is strong evidence, as well as the complete curve given by Klyachko
et a1.9 for the decrease of the number of strong acid sites with
diminishing Si/Al ratio.)
Moreover, Barthomeuf predicted a linear increase of the ef-
fective strong acidity on mordenite from m = 0 to limiting mNm
and then a decrease to zero for the idealized composition of the
unit cell. As follows from Figure 7 our results completely fulfill
this prediction, thus proving the validity of the theoretical model
Acid Stremgth and Catalytic Activity. Hydrogen mordenites
are frequently studied catalysts in the reactions of paraffin
for other zeolite
value of 0.096 for
8484 The Journal o f Physical Chemistry, Vol. 96, No. 21, 1992
Stach et al.
\ ‘ i’”
m x IO2
F w 8. Comparison of catalytic parameters with the number of strong
acid site dependence on the mole fraction of tetrahedrally coordinated
AI atoms (curve 1, rate of isomerization; curve 2, rate of disproportion-
ation in conversion of o-xylene, from ref 43; 0, own calorimetric data;
0, calorimetric data from ref 9).
cracking, skeletal isomerization of hydrocarbons, and transfor-
mations of alkylaromatics (reviews in refs 12, 36, and 37). This
is mainly due to the uniquely high acid strength of the mordenites
resulting in a higher rate of hydrocarbon reactions compared with
other zeolites. Further important properties are the high chemical
and thermal stability and the shapeselective behavior toward large
All the above-mentioned reactions involve the generation of
carbenium ions, including the participation of Bransted sites.
Correlations between the catalytic activity and the acid strength
were therefore expected and indeed found. In many cases these
correlations are reported in an indirect way, Le., not between the
catalytic activity and the acid strength, but the dealumination
degree or the Si/AI ratio, due to the fact that investigations of
the acidity were not always performed.
In ref 12 studies of the influence of the dealumination on the
mordenite catalytic activity in the cracking of n-octane are
presented. The activity increases with rising degree of dealu-
mination, passes through a maximum at Si/Al ratio of about 10,
and then decreases upon further dealumination. The authors
qualitatively explain the activity curve by the increase of the acid
strength and the simultaneous decrease of the acid site concen-
tration upon further dealumination. As has been shown, we found
a maximum in the concentration of the strong acid sites at a Si/AI
ratio of 9.5 (see Figure 6 ) , thus being in a very good agreement
with the catalyst possessing a maximum catalytic activity.
Maxima in the curves of catalytic activity for dealuminated
mordenites with the Si/A1 ratio between 8.5 and 10 are also
reported for the liquid-phase hydrolysis of ethyl acetate3* and the
isomerization of C&
Direct relations between the catalytic activity of alkylaromatics
and the number and strength of acid sites in aluminum deficient
mordenites were found by Gokhman et a1.4O (toluene dispropor-
tionation correlated with the total number of OH groups),
Giordano et al.41 (conversion of m-xylene), and Karge et al.42
(linear relationship between the rate of ethylbenzene dispropor-
tionation and the density of Bransted sites).
Very interesting results were presented by Minachev et al.43
They investigated the o-xylene conversion over dealuminated
mordenites with a broad range of Si/AI ratios. The rate of
isomerization (curve 1) and disproportionation (curve 2) are
plotted against the aluminum molar fraction in Figure 8. The
concentration of the very strong acid sites derived from our own
measurements and from the literature is included. As may be
seen, both curves follow the course of the very strong acid sites,
and the maximum in the curves of the catalytic activity corre-
sponds to the maximum of the strong Bransted acidity.
To investigate the acidity of the samples, four different methods
were used: adsorption calorimetry of ammonia, ammonia TPD,
ammonium ion exchange, and ‘H
methods give values for the total number of Bransted sites in good
agreement, while the mean error of the ‘H
surements amounts to about 20%. It was found that the total
number of bridged hydroxyl groups rises linearly with increasing
number of aluminum atoms in the lattice and that the number
of chemisorbed (or exchanged) ammonia molecules per unit cell
corresponds to the number of Alf atoms.
In contrast, the curve of the number of strong acid sites (derived
from calorimetric measurements) as a function of aluminum
content of the framework passes through a maximum at AIf =
4.6 (Si/Al = 9.5). By considering the aluminum topological
density of different zeolites Barthomeuf predicted a limiting value
of Si/Al = 9.4 for the maximum number of strong acid sites in
mordenites, this being in excellent agreement with our experi-
mental data. This obviously confirms her theoretical concept.
Comparing the dependence of the catalytic activity of alumi-
num-deficient H-mordenites in the conversion of hydrocarbons
on the Si/Al ratio, maximum values of the activity are found in
the region of Si/AI = 8.5-10, this being in good agreement with
the composition of the H-mordenite sample investigated with the
maximum number of strong acidic sites.
MAS NMR. The first three
MAS NMR mea-
Acknowledgmenr. The preparation of the mordenite samples
by Dr. K. J. Waghamare, NCL Pune (India), is gratefully ac-
Regi~hy NO. AI, 7429-90-5; NH,, 7664-41-7; NHd’, 14798-03-9.
References and Notes
(1) Wendt, R.; Jinchen, J., unpublished results.
(2) Tsutsumi, K.; Nishimiya, K. Thermochim. Acta 1989, 43, 299.
(3) Sawa, M.; Niwa, M.; Murakami, Y. Zeolites 1990, 10, 532.
(4) Karge, H. G.; Dondur, V. J. Phys. Chem. 1990, 94, 765.
(5) Meyers, B. L.; Fleisch, T. H.; Ray, G. Y.;
Catal. 1988, 110, 82.
(6) Chandawar, K. H.; Kulkarni, S. B.; Ratnasamy, P. Appl. Catal. 1982,
(7) ahlmann, G.; Jerschkewitz, H.-G.; Lischke, G.; Parlitz, B.; Richter,
M.; Eckelt, R. Z. Chem. 1988, 28, 161.
(8) Klyachko, A. L.; Brueva, T. R.; Kapustin, G. I.; Rubinstein, A. M.
Acta Phys. Chem. 1978, 24, 183.
(9) Klyachko, A. L.; Kapustin, G. I.; Brueva, T. R.; Rubinstein, A. M.
Zeolites 1987, 7, 119.
(10) Auroux, A.; Bolis, V.; Wierzchowski, P.; Gravelle, P. C.; Vedrine, J.
C. J. Chem. Soc., Faraday Trans. 1979, 75, 2544.
(11) Senderov, E. E.; Bychkov, A. M.; Mishin, J. V.; Klyachko, A. L.;
Beyer, H. K. In Zeolites: Facts, Figures, Future; Jacobs, P. A., van Santen,
R. A., Eds.; Elsevier: Amsterdam, 1989; p 355.
(12) Mishin, J. V.; Bremer, H.; Wendlandt, K.-P. In Catalysis on Zeolites;
Kallo, D., Minachev, Ch. M., Eds.; Budapest, 1988; p 231.
(13) Lohse, U.; Parlitz, B.; Patzelova, V. J. Phys. Chem. 1989,93, 3677.
(14) Sohn, J. R.; DeCanio, St. J.; Fritz, P. 0.; Lunsford, J. H. J. Phys.
Chem. 1986, 90,4847.
(15) Lombardo, E. A.; Gaffney, T. R.; Hall, W. K. In Proc. 9th Inr. Congr.
Catal.; Phillip, M. J., Ternan, M., Eds.; Calgary, 1988; Vol. I, p 412.
(16) Topchieva, K. W.; Xo Shu Thoang Activity and Physic0 Chemical
Properties of High Silica Zeolites and Zeolite Containing Catalysts; Univ-
ersity of Moscow: Moscow, 1976; p 100, and references therein.
(17) Minachev, Ch. M.; Isakov, J. I.; Achmetov, J. S. Isu. Akad. Nauk
SSSR Ser. Chim. 1976, 108 and references therein.
(18) Pine, L. A.; Maher, P. J.; Wachter, W. A. J. Catal. 1984,85, 476.
(19) Naccache, C.: Cheng Fan Ren; Couduvrier, G. In Zeolites: Facts,
Figures, Future; Jacob, P. A., van Santen, R. A., Eds.; Elsevier: Amsterdam,
1989; p 661.
(20) Stach, H.; Janchen, J.; Lohse, U. Catal. Lett. 1992, 13, 389.
(21) Freude, D.; FrBhlich, T.; Hunger, M.; Pfeifer, H.; Scheler, G. Chem.
Phys. Lett. 1983, 98, 263.
(22) Wachter, E. A. In Proc. 6rh Con$ Zeolites; Olson, D., Bisio, A., Eds.;
Butterworth: London, 1984; p 141.
(23) Melchior, M. T.; Vaughan, D. E.; Jacobson, A. J. J. Am. Chem. Soc.
(24) Thomas, J. M. In Proceedings of the 8th International Congress on
Catalysis; Verlag Chemie: Frankfurt, 1984; Vol. 1, p 31.
(25) Mortier, W. J. J. Catal. 1978, 55, 138.
(26) Kassab, E.; Seiti, K.; Allavena, M. J. Phys. Chem. 1988, 92, 6705.
(27) Sauer, J. In Modelling of Structure and Reactivity in Zeolites;
Catlow, C. R. A., Vetrival, R., Fds.; Academic Press, to be published.
Miller, J. T.; Hall, J. B. J.
(28) Barthomeuf, D. Mater. Chem. Phys. 1987, 17.49.
(29) Barthomeuf, D.; Beaumont, J. J. Catal. 1973, 30, 288.
J. Phys. Chem. 1992, 96,8485-8488
(30) Dwyer, J.; OMalley, P. J. In Keynotes in Energy-Related Catalysis;
Kaliaguine, S., Ed.; Studies in Surface Sciences and Catalysis; Elsevier:
Amsterdam, 1988; Vol. 35, p 5.
(31) Dempsey, E. J. Catal. 1975, 39, 155 and references therein.
(32) Mikovski, R. J.; Marshall, J. F. J. Catal. 1976, 41, 1700.
(33) Peters, A. W. ACS Symp. Ser. 1984,248,201 and references therein.
(34) Baegley, B.; Dwyer, J.; Fitch, F. R.; Mann, R.; Walters, J. J. Phys.
Chem. 1984,88, 1744.
(35) Meier, W. M. J. Solid Stare Chem. 1979, 27, 149.
(36) Poutsma. M. L. In Zeolite Chemistry and Catalysis; Rabo, J. A., Ed.;
ACS Monograph; American Chemical Society: Washington, D.C., 1976; p
(37) Barbidge, B. W.; Keen, J. M.; Eyles, M. K. Adv. Chem. Ser. 1971,
(38) Namba, S.; Hosonuma, N.; Yashima, T. J. J. Catal. 1981, 72, 16.
(39) Koradia, P. B.; Kiovsky. J. R.; Asim, M. Y. J. Catal. 1980, 66, 92.
(40) Gokhan, B. Kh.; Gorodetskaya, I. V.; Kaganova, E. M.; Korchagina,
I. A.; Tisovsky, G. I. In Proc. 1st All-Union Conference on the Application
o f Zeolites in Catalysis; Novosibirsk, 1976; p 143.
(41) Giordano, N.; Bart, C. J.; Vitarelli, P.; Cavallaro, S.; Ottana, R. In
Proc. v r h Int. Symp. Heterogeneous Catal. Varna 1983, 1, 417.
(42) Karge, H. G.; Weitkamp, J. Chem.-Ing. Technol. 1986, 58, 946.
(43) Minachev, Ch. M.; Shpiro, E. S.; Mishin, J. V.; Mathe, T.; Antoshin,
B. V. Isv. Akad. Nauk. SSSR Ser. Chim. 1983, 2682.
Infrared Study of Ion Beam Deposited Hydrogenated AiN Thin Films. 2. Effect of
Xiao-Dong Wang, K. W. Hipps, and Ursula Mazur*
Department of Chemistry and Materials Science Program, Washington State University, Pullman,
Washington 99164-4630 (Received: January 9, 1992)
Thin films of amorphous hydrogenated AlN (AlNH) were prepared by ion beam reactive sputter deposition. Both hydrogenation
and deuteration experiments were preformed. Far-IR and mid-IR spectra obtained from these f i l m s are presented and vibrational
assignments are given. Two new bands are observed at 434 and 501 cm-' in the far-IR region. AI-NH3 and AI-N2 are
found to be present in hydrogenated AlN films.
There is a continuing growth in technological applications of
AlN thin films and ceramics. These materials are chemically inert,
nontoxic, posses high thermal conductivity, and have excellent
piezoelectric and dielectric properties.'-" Of special interest is
hydrogenated aluminum nitride, A1N:H. The presence of hy-
drogen improves the electrical and mechanical properties of this
material.s-7 Although there have been several investigations
addressing structural properties of stoichiometric A1N8-I0 only
a limited number of publications deal with molecular studies of
AlN:H.S*7*11 We have recently reported the mid-IR vibrational
spectra and assignments of polycrystalline AIN and amorphous
hydrogenated aluminum nitride thin
present refined assignments for AlN:H based on deuterium isotopic
substitution experiments. We also report and assign the far-IR
spectrum of the AI-N2 complex observed in A1N:H films.
Fclm Fabrication. Ion beam and thermally deposited f i l m s were
made in a cryopumped ultrahigh vacuum (UHV) chamber capable
of attaining x10-9 Torr without baking. The chamber was pumped
to Torr prior to all depositions. Aluminum nitride films
50 nm thick were deposited onto unheated substrates by reactive
ion beam sputtering at a rate of 0.02 nm/s. The ion gun was a
2.5-cm low-energy source purchased from Ion Tech, Inc. A n FM
4574 gas blender from Linde was used to mix nitrogen and hy-
drogen gases. The gas mixtures consisted of 100% N2, 15%
H2:85% N2, 25% H2:75% N2, 35% H2:65% N2, 15% D2:85% N2,
In this report we
and 25% D2:75% N2. Ion beam sputtering was performed at 2
X lo4 Torr total pressure. The target employed was a 3-in.
aluminum disk. A quartz balance was used to measure the de-
position rate as well as the thickness of the sputtered films. An
Edwards Anavac quadrupole mass spectrometer was used to
monitor the atmosphere in the chamber.
Nitride specimens were deposited onto a Au/Al base fabricated
on glass substrates. Deposition of the Au/Al film was accom-
plished through resistive evaporation. An A1 film, 200 nm thick,
was deposited onto glass from an aluminum-covered tungsten wire.
The AI film was then covered with 80 nm of gold. This gold
overlayer was evaporated from an aluminum oxide source.
Materirls. The substrates used were Coming glass micrcwcope
slides. These were cleaned in a solution of 4 1 HN03/H202 prior
to use. Concentrated HN03 and 30% H202 were used. Metals
of 99.99+% purity were used for all our deposition experiments.
All gases used were ultrapure grade. Deuterium gas was pur-
chased from Matheson and was 99% isotopically pure.
FT-IR. Reflectance spectra of AlN films were obtained with
an IR/98 R-IR spectrometer using 83O grazing incidence ra-
diation. The incident light was not polarized. The mid-IR spectra
were acquired with 8-cm-' resolution and are the result of 1024
scans. The far-IR data were produced by adding 6048 scans, each
of which was collected with 4-cm-I resolution. A helicoil source,
and either a gennanium-coated KBr (mid-IR) or a 6-pm Mylar
(far-IR) beam splitter were employed. A nitrogen-cooled MCT
detector and a helium-cooled bolometer (Si/SiGe system from
Infrared Associates) were used for the mid- and far-IR data
acquisition, respectively. All spectra were obtained at room
A diffuse reflectance spectrum was obtained from high-purity
(99.9%) A1N mixed with KBr (1:s). A KBr:TGS detector and
germanium-cuated KBr beam splitter were employed. Data were
taken at 4-cm-I resolution. The Kubelka-Munk transformation
was performed on the reflectance data.
Because there is a large contribution to the far-IR region by
the tail on the AI-N mode, we used background subtraction on
all the far-IR data. In essence a low order polynomial was fit
to the data and then subtracted from it.
Results and Discussion
The upper curve in Figure 1 depicts the vibrational spectrum
in the 200-4000-cm-' region of the nitride film made with 15%
H2 in the sputtering gas mixture. The bottom trace displays the
far- and mid-IR reflectance spectra obtained from an AlN film
fabricated with 15% D2 and 85% nitrogen. The far-IR portions
of the spectra have been scaled by a factor of 20. The positions
of the vibrational bands present in these spectra are collected in
0 1992 American Chemical Society