2890 J. Phys. Chem.

B 1998, 102, 2890-2896

Characterization of Acid Sites in Zeolitic and Other Inorganic Systems Using Solid-State
31P NMR of the Probe Molecule Trimethylphosphine Oxide

Edward F. Rakiewicz,† Alan W. Peters,* and Richard F. Wormsbecher

Washington Research Center, W.R. Grace and Company, 7500 Grace DriVe, Columbia, Maryland 21044

Kevin J. Sutovich and Karl T. Mueller*

Department of Chemistry, The PennsylVania State UniVersity, 152 DaVey Laboratory,
UniVersity Park, PennsylVania 16802
ReceiVed: January 15, 1998; In Final Form: February 18, 1998

The ability to determine the types and concentrations of acid sites in zeolites and fluid-catalytic cracking
systems is important for an increased understanding of structure/performance relationships in these materials.
Currently, a variety of thermal methods exist that allow quantitative measurement of the Brønsted acid site
concentration. In addition, numerous spectroscopic methods using probe molecules are available for qualitative
and quantitative detection of both Brønsted and Lewis acid sites. In studies utilizing solid-state NMR
spectroscopy, probe molecules containing 31P nuclei present substantial advantages over probes isotopically
enriched with 13C and 15N nuclei. These advantages include increased sensitivity and chemical shift dispersion.
While a number of phosphorus-based experiments have probed the interaction of trimethylphosphine with
solid acid catalysts, initial studies of the more stable trimethylphosphine oxide (TMPO) have only been reported
on amorphous silica-alumina surfaces. We now report the successful completion of TMPO studies of acid
sites in several systems including γ-alumina; HY, USY, and dealuminated Y zeolites; and a silica-alumina
catalyst with an aluminum concentration of 13%. Comprehensive and consistent assignments to particular
types of sites are made for all resonance lines in the 31P MAS NMR signals from TMPO. Based on results
from dehydroxylated γ-alumina, new chemical shift assignments are made for the TMPO/Lewis acid complex.
The assignments of 31P resonances from molecules not directly associated with nearby 27Al nuclei (such as
crystalline or physisorbed TMPO species) are supported using 1H/31P/27Al triple-resonance NMR methods.
The concentrations of Brønsted acid sites from the NMR results are compared with concentrations obtained
from isopropylamine/temperature-programmed-desorption measurements, and substantial agreement between
the methods is found.

Introduction need exists for methods that are both discriminating and
quantitative with respect to populations of Lewis and Brønsted
Acidic forms of high-surface-area materials are important for acid sites in catalytic systems.
a number of catalytic reactions, including fluid-catalytic cracking
Nuclear magnetic resonance (NMR) spectroscopy is a power-
of petroleum to produce useful fractions with a desired range
ful analytic tool, providing quantitative measurement of popula-
of molecular weights. Important for understanding the chem-
tions of nuclei discriminated by their chemical shift and other
istry associated with catalysis is the structural characterization
parameters related to the local structural environment. A variety
and quantification of acid sites. Methods for the characterization
of solid-state 13C and 15N NMR methods have been applied to
and quantification of acidic sites in solids are well-documented
the study of solid acidity. Many of the probe molecules used
in the literature,1-14 and most of these approaches involve the
in these studies are amines such as n-butylamine,5 pyridine,4
loading of a weakly basic probe molecule onto the solid
and ammonia.7 However, these studies are hampered by low
followed by spectroscopic and/or thermal characterization. One
sensitivity and/or limited chemical-shift ranges of the observed
of the most commonly used techniques is pyridine/diffuse-
nuclei. 13C and 15N are the NMR-active isotopes used in these
reflectance IR spectroscopy (DRIFTS)1-3 which measures the
studies, and these species have natural abundances of 1.1% and
qualitative amounts of Lewis and Brønsted acid sites in a given
0.4%, respectively. Therefore, isotopic labeling is generally
material. Another common method used to characterize acid
necessary. Both 13C and 15N also have moderate to low
sites is isopropylamine/temperature-programmed desorption
gyromagnetic ratios, and this further contributes to their low
(IPA/TPD)10 which provides a quantitative measure of Brønsted
NMR sensitivity.
sites. However, this method is not capable of providing
information on Lewis sites since it depends on the protonation The use of phosphorus-containing bases in conjunction with
of the IPA molecule and subsequent formation of propylene. A solid-state 31P NMR6,8,9,12,13 overcomes the most difficult
experimental limitations associated with 13C and 15N studies.
† Current address: PPG Industries, Inc. 4325 Rosanna Drive Allison Park, The 31P isotope is 100% abundant in nature, and it also has a
PA 15101. larger gyromagnetic ratio than 15N and 13C. The large isotropic
* To whom correspondence should be addressed. chemical shift range and full chemical shift anisotropy of 31P
S1089-5647(98)00808-6 CCC: $15.00 © 1998 American Chemical Society
Published on Web 03/31/1998
Acid Sites in Zeolitic and Inorganic Systems J. Phys. Chem. B, Vol. 102, No. 16, 1998 2891

provide a wealth of information on the local electromagnetic TABLE 1: Summary of the Physical Properties of Samples
interactions of the nucleus and can therefore be applied directly in This Study
in the study of phosphorus environments. silica-
A number of studies of the acidity of zeolites and related γ-alumina HY USY low-Al Y alumina
oxides have been performed using trimethylphosphine (TMP) chemical analysis
as a probe molecule for solid-state 31P NMR.6,9,12,13 This % Al2O3 100 21.81 21.82 9.23 12.78
% Na2O 0 5.01 4.40 0.17 0.046
relatively strong Lewis base (pKa ) 5.3 in H2O) has shown % SiO2 0 72.94 73.80 90.46 86.95
substantial promise as a probe of acidic sites in solids. Si/Al 0 2.8 2.9 8.3 6.8
Formation of a protonated TMPH+ adduct is characterized by nitrogen adsorption
an isotropic chemical shift of approximately -3 ppm relative total surface area 201 874 744 731 497
zeolite surface area NA 866 697 668 NAa
to aqueous H3PO4 and scalar coupling of approximately 500
matrix surface area NA 9 47 63 NA
Hz between 1H and 31P nuclei.15 Lewis-bound TMP exhibits a X-ray diffraction
considerable upfield shift relative to the Brønsted site, and, if unit cell (nm) NA 2.466 2.4525 2.436 NA
an aluminum is involved in the bonding, 31P/27Al scalar coupling no. of Al per unit cell NA 54.14 38.59 19.58 NA
results in a well-characterized multiplet pattern. Recently, Kao Si/Al NA 2.55 3.98 8.80 NA
and Grey16,17 have used 31P/27Al transfer of populations in a
NA ) not applicable.
double resonance (TRAPDOR) NMR18 to study the binding of
trimethylphosphine on dehydroxylated zeolite HY, demonstrat- was prepared by hydrothermal dealumination of a sample of
ing the use of novel double-resonance NMR methods as probes NaNH4Y for 2 h at 700 °C. The low-aluminum Y was prepared
of the binding of molecules on surfaces and in zeolites. by two cycles of hydrothermal dealumination followed by a
When TMP is introduced in excess, the room-temperature low-pH (2.9) ammonium sulfate wash. The silica-alumina
dynamics of chemical exchange between bound species and the sample was prepared from a sodium silicate/sodium aluminate
liquidlike TMP molecules complicate the identification and slurry by the addition of acid. The resulting gel was dried, and
enumeration of acidic sites, and therefore the temperature must the sodium was removed by washing with ammonium sulfate
be lowered in order to investigate the binding. TMP is a highly solution, followed by a final drying of the sample. In all cases
flammable air-sensitive liquid at room temperature and requires residual ammonia was removed by calcination at 650 °C before
great care in its handling. Studies on amorphous silica-alumina treatment with TMPO. The samples prepared for this study
indicate that the use of trialkylphosphine oxides offers several were characterized for atomic composition and other physical
advantages for the study of acidic binding sites.8 These oxides properties, and a listing of a number of relevant results is
are solids at room temperature, they are not susceptible to presented in Table 1.
oxidation, and titration of the molecules onto acidic solids can Prior to characterization by solid-state NMR, samples were
be accomplished using solution-state chemistry. However, there prepared following a uniform procedure for loading with TMPO.
are no comprehensive published reports describing the use of For each sample studied, approximately 0.8 g of solid was
trialkylphosphine oxides to systematically study other interesting calcined at 600-800 °C for 4 h in a round-bottom flask. After
catalytic materials such as zeolites and aluminas. lowering the oven temperature to 150 °C, the flask was capped
In this paper we report our progress on the use of trimeth- with a rubber septum. A measured amount of a dry TMPO/
ylphosphine oxide (TMPO) and solid-state 31P NMR to study CH2Cl2 solution of known concentration was added to the flask
the acidic nature of solids, specifically zeolites and aluminas. via a gastight syringe, and the mixture was agitated overnight
The studies provide an unambiguous assignment of Lewis acid under N2 on a mechanical shaker. The solvent was removed
sites based on results from TMPO adsorbed directly onto a under vacuum and the flask transferred into a glovebox where
sample of dehydroxylated Catapal alumina. Samples containing the product was placed into a vial and sealed. For analysis by
Brønsted acid sites have also been characterized by 31P MAS solid-state NMR, the prepared solid samples were packed into
NMR of TMPO probes. A sample of ultrastable Y zeolite was dry MAS rotors of the Chemagnetics Pencil design. Either 5.0
further characterized using triple-resonance 1H/31P/27Al NMR or 7.5 mm rotors with sealed endcaps were used to hold the
experiments to confirm a number of the specific assignments. samples within the NMR probes.
Quantification of spectral intensities of resonances assigned to For determination of the quantitative number of acid sites
Brønsted sites is used to calculate the concentration of Brønsted using NMR measurements, it is necessary to first measure the
acidity in the solids, and these results are compared with those amount of phosphorus in a known mass of a particular sample.
obtained using IPA/TPD. A study of the 31P NMR spectra of Phosphorus quantification was carried out by sample digestion
TMPO in samples at various stages of rehydration also followed by standard ICP analysis. A 0.1 g sample of the zeolite
demonstrates the need for careful and consistent interpretation was dissolved in 10 mL of a 1:1:1 mixture of HF, HNO3, and
of experimental results. HCl until clear. A further 60 mL of a 5.3% boric acid solution
was added to the cool solution, and the ICP analysis was carried
Experimental Section out using matched phosphorus standards. To compare the NMR
results with another, well-accepted method for analyzing the
A representative collection of a number of acidic solids was concentration of Brønsted acid sites, a slightly modified
chosen for study. The samples are a dehydroxylated γ-alumina, version19 of the IPA/TPD method described by Kofke et al.10
a protonated form of NaY zeolite (HY), an ultrastable Y (USY) was used.
zeolite, a low-aluminum Y zeolite, and an amorphous silica- Solid-state 31P MAS and 1H/31P CPMAS NMR spectra were
alumina sample containing 13% aluminum. A commercial obtained on both a Chemagnetics CMX Infinity 400 and a
sample of Catapal A alumina, obtained from Condea (Houston, Chemagnetics CMX 200 spectrometer operating with 31P
TX), was calcined for 3 h at 705 °C to prepare the γ-alumina Larmor frequencies of 161.880 or 81.025 MHz, respectively.
sample. The HY faujasite sample was prepared from NaY by Typical spinning speeds of 5-8 kHz were used, and nominal
ammonium sulfate exchange of the sodium, and the composition 90° pulses were 4 µs in length. Recycle delays of 10-30 s
of the ammonium form is given in Table 1. A sample of USY were necessary to allow full recovery of equilibrium magnetiza-
2892 J. Phys. Chem. B, Vol. 102, No. 16, 1998 Rakiewicz et al.

tion. The number of scans collected for each spectrum varied

between 500 and 10 000, and the 31P chemical shifts were
referenced to an external sample of solid phosphomolybdic acid
(Fisher Scientific Co.) which was assigned a shift value of -6
ppm with respect to aqueous 85% H3PO4.
Solid-state NMR experiments were also performed on a
home-built 400 MHz NMR spectrometer utilizing a Tecmag
Libra system for pulse programming and signal digitization. The
static magnetic field strength of the spectrometer is 9.4 T,
corresponding to resonance frequencies for the 1H, 31P, and 27Al
species of 399.995, 161.927, and 104.229 MHz, respectively.
Spectra were acquired using a Chemagnetics 5.0 mm triple-
resonance CPMAS NMR probe with rotor frequencies between
5 and 10 kHz. The rotor frequencies were stabilized to (2 Hz
using an active-feedback-spin-rate controller system from Che-
magnetics. Solid-state rotational-echo double-resonance (RE-
DOR)20 experiments were performed utilizing 31P and 27Al spin
pairs,21,22 with initial cross-polarization from 1H spins in this
triple-resonance 1H/31P/27Al REDOR experiment. Triple-
resonance TRAPDOR experiments18 were also performed in a
similar manner. The use of proton polarization as the initial
source of the signal allows for a shorter recycle delay in the Figure 1. (a) The single-pulse solid-state 31P MAS NMR spectrum of
triple-resonance experiments (compared to 31P/27Al double- crystalline TMPO. The isotropic peak appears at 39 ppm, along with
resonance), since the T1 values for the protons are considerably a well-defined set of spinning sidebands. (b) The single-pulse solid-
shorter than relaxation times for the 31P nuclei. A TRAPDOR state 31P MAS NMR spectrum of crystalline TMPO with the addition
NMR signal is obtained as the difference of the signals from of approximately 1 µL of water to the sample rotor. A new isotropic
resonance appears at 41 ppm. Sample spinning rates in both spectra
two separate NMR experiments: The first is a spin-echo MAS are 5.0 kHz.
signal obtained from the 31P spins after initial cross-polarization
from the 1H nuclei. In a second experiment, the 31P spin-echo Addition of a small amount (approximately 1 µL) of water
is formed in an analogous manner, but during this time a long to solid TMPO contained within an NMR rotor produces a new
31P resonance at 41 ppm, lacking a broad sideband pattern, as
radio-frequency pulse is applied at the Larmor frequency of the
27Al nuclei. The interaction between the nuclear electric shown in Figure 1b. The characteristic pattern from crystalline
quadrupole moment of the 27Al nuclei and local gradients in TMPO also remains, representing the fraction of the sample
the electric field causes a time-dependent change in the spin unaffected by the water.
energy levels under MAS conditions, and the long radio- When a small amount of TMPO is dissolved in excess
frequency pulse causes mixing and scrambling of coherences deionized distilled water (pH 7), the 31P resonance from TMPO
among these energy levels. If 31P nuclei are close in space to appears at 53 ppm. A single peak also appears at 53 ppm in a
the 27Al nuclei (usually within 5 Å), there is a substantial slightly basic solution and also when the solution has a starting
dipole-dipole intereaction between the heteronuclei, and the pH of 6. This single peak at 53 ppm remains until the pH of
result of the time-dependent modulation in 27Al energy levels the initial solution is dropped below a pH of 1. Eventually,
is the dephasing of the 31P spin-echo signal. The signal will the resonance moves downfield, reaching 83 ppm in a concen-
only be diminished for 31P close in space (dipole-coupled) to trated (12 M) HCl solution. For comparison, the chemical shift
27Al spins, and therefore the TRAPDOR difference signal will of TMPO in DMSO is 37 ppm while that of TMPO in
display only resonances from 31P nuclei close to 27Al nuclei. chloroform is reported as 38 ppm.6
Spectral editing of the resonances in the REDOR experiment Catapal Alumina. The first acidic sample investigated was
occurs in a similar manner, except that in REDOR experiements a dehydroxylated Catapal alumina with two different loadings
rotor-synchronized 180° pulses are applied to the heterospins of TMPO per gram of solid. This form of alumina (γ-alumina)
during the formstion of the spin-echo signal. is a representative source of Lewis acid sites, as hydroxyl groups
are removed from the surface during high-temperature treat-
Results ment.25 The heat treatment to above 700 °C, as described above,
is sufficient to remove over 90% of surface hydroxyl species.25
Crystalline and “Hydrated” Trimethylphosphine Oxide. The solid-state 31P MAS NMR spectra from a sample of
A sample of crystalline TMPO (purchased from Alfa and γ-alumina loaded with 0.78 and 2.3 mmol of TMPO/g of solid
recrystallized from a solution in dry CH2Cl2) was analyzed by are shown in Figure 2. At low TMPO loading (Figure 2a), there
31P MAS NMR, and a wide sideband pattern is observed is a major signal in the spectrum at 37 ppm. At higher loading
corresponding to a large 31P chemical shift anisotropy (Figure (Figure 2b) a second peak appears with an isotropic chemical
1a). The parameters describing the chemical shift anisotropy shift of 39 ppm and a wide sideband pattern matching that of
were determined from a Herzfeld-Berger analysis of the crystalline TMPO. The close correspondence of the wide
spinning sideband pattern.23,24 The measured parameters are pattern to that observed from crystalline TMPO demonstrates
∆σ ) -170.9, η ) 0.44, and σiso ) 39.4. In our studies of that crystalline TMPO readily forms in these systems in a bulk
solid acids, any crystalline TMPO present will show a similar phase. The 31P NMR spectra of both TMPO/alumina samples
pattern if the crystalline material is contained within a bulk also show the presence of a broad minor signal (with a
crystalline phase. Comparison of calculated sideband intensity discernible sideband pattern) at approximately 53 ppm, while
patterns at a given spinning speed will allow positive identifica- all other peaks in the spectrum correspond to spinning sidebands
tion of this phase. of the resonances at 37 and 53 ppm.
Acid Sites in Zeolitic and Inorganic Systems J. Phys. Chem. B, Vol. 102, No. 16, 1998 2893

Figure 3. Single-pulse solid-state 31P MAS NMR spectrum of HY

sample containing 2.70 mmol of TMPO/g of solid. The sample spinning
rate is 6.0 kHz.

Figure 2. Single-pulse solid-state 31P MAS NMR spectra of Catapal

alumina samples with: (a) 0.78 mmol of TMPO/g of solid, (b) 2.3
mmol of TMPO/g of solid, and (c) 2.3 mmol of TMPO/g of solid after
exposure to an atmosphere with 25% relative humidity for 1 h. Sample
spinning rates in all spectra are 5.0 kHz.

The effect of the introduction of water to the TMPO/alumina

system was accomplished by exposing the sample with 2.3 mmol
of TMPO/g of solid to a controlled atmosphere with 25%
relative humidity. The 31P MAS NMR spectrum (Figure 2c)
shows two distinct features: a broad resonance at around 53
ppm and an isotropic peak with a wide sideband pattern. The
relative sideband intensities, but not the intensity of the isotropic
peak, match the intensities of the sidebands in the crystalline
pattern of Figure 1. The enhanced isotropic peak at 39 ppm in
the pattern must therefore be the sum of resonances from both Figure 4. Single-pulse solid-state 31P MAS NMR spectra of a USY
crystalline TMPO and 31P in a different environment than the sample containing 2.70 mmol of TMPO/g of solid: (a) prior to exposure
crystalline sample. The resonance at 37 ppm is absent in the to atmosphere, (b) after exposure to an atmosphere with 25% relative
spectrum acquired after rehydration. humidity for 1.5 h, and (c) after hydrating in a saturated atmosphere
for 2 h.
Zeolite HY. The 31P NMR spectrum of a sample of zeolite
HY loaded with 2.70 mmol of TMPO/g of solid is shown in a wide band of signals appears (Figure 4a). Each signal has a
Figure 3. Isotropic resonances appear in three distinct regions measurable sideband pattern, but the overlap and spectral
of the spectrum. The first is a sharp peak at 39 ppm, with the crowding precludes direct comparison with the crystalline
same well-defined sideband pattern as seen in the crystalline TMPO signal of Figure 1a. Isotropic peaks are observed at
TMPO and nonhydrated TMPO/alumina samples. Broader 65, 55, 44, and 39 ppm. After the sample was gradually exposed
resonances are observed at 65 ppm and around 55 ppm, with to a controlled atmosphere (20% relative humidity) for 2 h, the
small but measurable spinning sidebands at this moderate (6 peak at 65 ppm decreased in intensity, while the peak at 55
kHz) rotational frequency. ppm increased in relative intensity (Figure 4b). After the sample
Steam-Calcined Zeolite HY (USY). In the 31P MAS NMR was placed in a saturated atmosphere (100% relative humidity)
spectrum of a sample of steam-calcined HY impregnated with for 1 h, the 31P MAS spectrum (Figure 4c) shows no resonance
a moderate loading of base (2.70 mmol of TMPO/g of solid), at 39 ppm and an increase in relative intensity for the peak at
2894 J. Phys. Chem. B, Vol. 102, No. 16, 1998 Rakiewicz et al.

Figure 7. Single-pulse solid-state 31P NMR spectrum of an amorphous

Figure 5. 1H/31P/27Al triple-resonance TRAPDOR spectrum of USY silica-alumina (13% Al) with 2.5 mmol of TMPO/g of solid. The
after treatment with 2.70 mmol of TMPO/g of solid. The difference sample spinning rate is 6.0 kHz.
spectrum shown was obtained with 10 rotor cycles on each side of the
180° refocusing pulse and a spinning speed of 6.25 kHz.
intensity of the spinning sidebands and the absence of the broad
peak in the CP spectrum (Figure 6b) indicates once again that
there are actually two peaks present, similar to the observation
from the rehydrated Catapal aluminum spectrum (Figure 2c).
Silica-Alumina (13% Al). The solid-state 31P MAS NMR
spectrum for the silica-alumina sample with 2.5 mmol of
TMPO/g of solid is shown in Figure 7. A large 31P signal with
few or no spinning sidebands appears at 41 ppm. An additional
broad peak appears at 63 ppm, similar to that observed in the
faujasite samples.

Discussion
Lewis Acid Sites. The isotropic chemical shift for a TMPO/
Lewis complex in acidic solids has been reported by Baltusis
et al. as 53 ppm based on a spectrum of TMPO on amorphous
silica-alumina (75% silica, 25% alumina by weight). They
also reported a similar shift of 58 ppm for 31P from a solution
of 1:1 TMPO/AlCl3 in CH2Cl2 measured at -60 °C.8 In our
studies, a dehydroxylated sample of Catapal alumina was chosen
as a representative solid acid containing a predominant popula-
tion of Lewis acid sites. Figure 2a displays a strong resonance
at 37 ppm from TMPO absorbed onto a Catapal alumina.
Therefore, we assign this resonance at 37 ppm, with a minor
sideband pattern, to TMPO/Lewis acid sites.
In this same sample, the assignment of the minor peak at 53
Figure 6. Solid-state 31P MAS NMR spectra of a dealuminated Y
zeolite with 2.74 mmol of TMPO/g of solid: (a) single-pulse spectrum; ppm is not as clear. According to Peri,25 as hydroxyl groups
(b) 1H-31P CPMAS spectrum acquired with a contact time of 3 ms. are removed from the surface of alumina, the underlying
The sample spinning rate is 6.0 kHz. aluminum cations become exposed to the surface and form
Lewis sites. The hydroxyl groups that remain on the surface
44 ppm. The ratio of peak intensities at 65 and 55 ppm remains could, in principle, behave as weak proton donors. Further
approximately constant after the initial rehydration procedure. insight is gained by exposing the sample containing the higher
The original TMPO/USY sample (before hydration) was also loading of 2.3 mmol of TMPO/g of solid to the atmosphere
characterized using a 1H/31P/27Al triple-resonance TRAPDOR and again examining the 31P MAS NMR signal. The peak at
NMR experiment. The results of this experiment (Figure 5) 37 ppm disappears from the spectrum, along with its associated
demonstrate which phosphorus resonances arise from complexes sideband pattern. Based on our assignment of this resonance
where the TMPO molecules are in close proximity to aluminum to Lewis-bound TMPO molecules, this result is expected given
atoms. Resonances are observed in the TRAPDOR difference the high sensitivity of Lewis sites to aqueous attack. The
spectrum at 65 and 55 ppm, while no TRAPDOR intensity is signature signal for crystalline TMPO is still present, with an
seen at 44 and 39 ppm. Therefore, the peaks at 65 and 55 ppm isotropic chemical shift of 39 ppm and the same CSA pattern,
must arise from 31P nuclei in spatial proximity to 27Al nuclei. but with the difference that the peak at 39 ppm is of improper
REDOR experiments were also performed, and the REDOR intensity. Therefore, another peak must be contributing to the
effect for the sites at 65 and 55 ppm was approximately the signal at 39 ppm, and this peak does not have a strong pattern
same, indicative of similar 31P/27Al distances. of spinning sidebands. Therefore, we assign this narrower
Low-Aluminum Y Zeolite. Solid-state 31P NMR spectra of resonance to either “solution-like” TMPO formed in the presence
a low-aluminum Y sample loaded with 2.74 mmol of TMPO/g of water (see below) or physisorbed TMPO which would be
of solid are shown in Figure 6. As in the HY and USY samples, weakly bound and undergoing rapid exchange with more mobile
signals appear at 65 and 55 ppm. Although the two signals in species. Importantly, the broad resonance at 53 ppm is still
the 40 ppm region are not resolved in Figure 6a, the low present, indicative of sites not affected by partial rehydration
Acid Sites in Zeolitic and Inorganic Systems J. Phys. Chem. B, Vol. 102, No. 16, 1998 2895

of the sample. As shown below, we identify this resonance TMPO resonance when coordinated to Lewis sites in the Catapal
with H+ complexed (Brønsted-bound) TMPO species. alumina sample. We have also shown that an enhancement of
Brønsted Acid Sites. The 31P NMR signal from TMPO in a peak at 53 ppm (but with a noticeably different spinning
association with the Lewis acid sites on the surface of a sideband pattern) occurs upon partial rehydration of zeolite
dehydroxylated alumina shows a resonance at 37 ppm. Moving samples. This is indicative of at least a partial contribution to
to studies of systems with potential Brønsted acid sites, a peak this peak of signals from TMPO molecules in rapid motion and
at 65 ppm in the 31P MAS NMR spectrum of TMPO has been closely associated with water in the system.
reported and assigned in the literature to a Brønsted acid site However, not all TMPO molecules in contact with water
by Baltusis et al. in their study of an amorphous silica-alumina contribute resonances at 53 ppm. The most striking example
sample (as described above).8 It is also reported26 that the occurs when a small amount of water is added to a sample of
spectrum of TMPO in contact with H+ resin reveals a 31P crystalline TMPO, and an isotropic peak arises at 41 ppm.
resonance with an isotropic shift of 72 ppm. Our study of Recent ab initio quantum mechanical calculations27 may explain
TMPO dissolved in 12 M HCl also confirms a large downfield and illuminate the difference in 31P chemical shift observed
shift (83 ppm) for protonated TMPO. Therefore, the assignment when comparing TMPO dissolved in an excess of water and a
of a peak at 65 ppm to a Brønsted complex is unambiguous. pure TMPO sample with a small amount of water present. In
The form of this site is presumably a TMPOH+ species attached systems with isolated water molecules, the minimum surface
to a bridging oxygen, similar to the TMPH+ species associated potential calculated for the oxygen atom in water is made
with Brønsted acid sites in previous studies.6 Further, in significantly more negative (by about 20 kcal mol-1) if two
aqueous solution with an excess of water and a pH above 1, a molecules of TMPO form a complex with each water molecule.
single 31P resonance appears at 53 ppm representing the average In this energetically preferred orientation, the P-O double bonds
chemical shift for 31P in a TMPO molecule undergoing rapid are each also involved in hydrogen bonding with the water.
exchange of an acidic proton in solution. This should be When water is present in excess, the complexation and therefore
compared to an isotropic shift of 37 ppm from TMPO in DMSO local electromagnetic environment will differ. The 31P chemical
and 38 ppm reported for TMPO in CDCl36 where no protonation shift will undergo a change, which our studies show is
equilibrium is possible. approximately 12-14 ppm downfield.
In solid acids, peaks in the 50-55 ppm range are seen as a Consistent with previous results obtained using trimeth-
minor component in the TMPO/alumina sample and as major ylphosphines as probes,6 our work shows that there are at least
components in the spectra from TMPO in association with HY, two different types of Brønsted complexes formed when probe
USY, and dealuminated Y zeolites. In previous studies, these molecules are added to acidic materials. Although there are
peaks have been assigned to Lewis-complexed or possibly many factors that influence the 31P chemical shift in these
physisorbed TMPO species.14 Since TMPO is a solid at room complexes, one possible explanation may be that there are
temperature, we do not expect a large amount of free TMPO in varying degrees of proton transfer to the TMPO that would result
the unhydrated samples. A peak at 53 ppm is a major in variations of the proton-oxygen bond length of the complex.
constituent of the spectrum from HY, and this sample of HY Assuming this to be the case, it may be possible to relate the
zeolite is not expected to possess an appreciable amount of chemical shift of the TMPO/Brønsted complex with the strength
Lewis acidity due to the minimal amount of nonframework of the acid site. For instance, if we assume that a shift value
aluminum in the sample. Solid-state 27Al MAS NMR (not of 72 ppm measured for TMPO in contact with H+ resin
shown) confirms the lack of substantial amounts of NMR-visible represents a TMPO/Brønsted complex with a very short H- - -
octahedrally coordinated aluminum atoms with moderately sized O bond, we can then tentatively assign the signals with chemical
quadrupolar-coupling constants. shifts of 65 and 55 ppm to TMPO/Brønsted complexes with
The assignments of the resonances at 37 ppm to a Lewis acid longer H- - -O bonds. Preliminary results obtained in our
site and at 53 ppm to a Brønsted site are further supported by laboratories studying TMPO adsorbed onto ZSM-5 zeolites show
consideration of chemical shift perturbations observed in TMP additional, and supposedly more tightly bound, Brønsted acid
and TMPO binding experiments. The chemical shift of 31P in sites giving rise to chemical shift values of 75 ppm.
liquidlike TMP (the major form of TMP at room temperature The spectra from a rehydrated USY system also contain peaks
when in excess in zeolite systems) is reported as -67 ppm when tentatively attributed to physisorbed and mobile TMPO species.
in association with acidic solids, while that of neat TMP is -63 These assignments are corroborated with 1H/31P/27Al REDOR
ppm.6 A large change is observed in the TMP 31P NMR and TRAPDOR NMR experiments. The experimental TRAP-
chemical shift upon complexation to a Brønsted site (∆δ ) 65 DOR results (Figure 5) clearly demonstrate that the phosphorus
ppm) forming TMPH+.6 In TMPH+, the proton is directly nuclei in the two TMPO/acid site complexes (65 and 55 ppm)
bonded to the phosphorus and should result in a larger change are coupled via a through-space direct dipolar interaction to
of the electronic environment (and hence 31P chemical shift aluminum atoms. The signals attributed to physisorbed (43
values) compared to TMPOH+, where the proton is one bond ppm) and crystalline (39 ppm) TMPO do not appear in the
removed from the phosphorus. The chemical shift of crystalline spectrum, demonstrating that they are not closely associated with
TMPO, the major form of TMPO at room temperature, is 39 aluminum species.
ppm, and the largest shift we measure for a protonated TMPO Brønsted Site Quantification. The results in this study
molecule has a chemical shift difference of 44 ppm. Using a suggest that the chemical shift range for TMPO/Brønsted acid
similar argument for the Lewis sites, ∆δ for the TMPO system complexes is from approximately 50 to 75 ppm. If these
should again be much smaller than the value obtained when assignments are made in a dry atmosphere and then used to
considering TMP, which is reported in the literature as 11 ppm.6 calculate absolute concentrations of Brønsted sites, the Brønsted
This would not be the case for a 31P peak from TMPO at 53 acid site concentrations as measured by 31P NMR should
ppm (∆δ ) 14 ppm), although when observed previously in compare favorably with measurements from other analytic
TMPO/silica-alumina this peak was assigned to a TMPO/Lewis methods such as IPA/TPD. Assigning all resonances between
acid complex.8 In fact, we measure an upfield shift of the 50 and 70 ppm to Brønsted acid sites, and calculating the
2896 J. Phys. Chem. B, Vol. 102, No. 16, 1998 Rakiewicz et al.

TABLE 2: Summary of the Brønsted Acid Site assistance in sample preparation, J. Burnham for her ICP work,
Concentrations Measured with IPA/TPD and 31P NMR and Dr. J. Frye for his helpful discussions and reports of
Brønsted site concn from Brønsted site concn from unpublished data. This report is also based upon work supported
sample IPA/TPD (µmol/g)a 31P NMR (µmol/g)b
by the National Science Foundation under Grant DMR-9458053
HY 1750 1458 and an Arnold and Mabel Beckman Foundation Young Inves-
USY 520 662 tigator Award to K.T.M.
low-Al Y 610 824
silica-alumina 310 327 References and Notes
Error estimated at (10% of reported value. b Error estimated at
a
(1) Basila, M. R.; Kantner, T. R.; Rhee, K. H. J. Phys. Chem. 1964,
(5% of reported value. 68, 3197-3207.
(2) Hughes, T. R.; White, H. M. J. Phys. Chem. 1967, 71, 2192-
concentration of acid sites per gram of material from the 2201.
(3) Eberly, P. E., Jr. J. Phys. Chem. 1968, 72, 1042-1047.
phosphorus concentrations measured with ICP analysis, the (4) Liang, S. H. C.; Gay, I. D. J. Catal. 1980, 66, 294-300.
concentrations of Brønsted acid sites were calculated and are (5) Dawson, W. H.; Kaiser, S. W.; Ellis, P. D.; Inners, R. R. J. Am.
reported in Table 2. For comparison, the IPA/TPD results are Chem. Soc. 1981, 103, 6780-6781.
also provided in Table 2. For the HY sample, the concentration (6) Lunsford, J. H.; Rothwell, W. P.; Shen, W. J. Am. Chem. Soc. 1985,
107, 1540-1547.
by NMR is measured as 83% of the concentration measured (7) Majors, P. D.; Raidy, T. E.; Ellis, P. D. J. Am. Chem. Soc. 1986,
by IPA/TPD. This sample contained the largest number of 108, 8123-8129.
Brønsted sites per gram in our studies. The results for the USY (8) Baltusis, L.; Frye, J. S.; Maciel, G. E. J. Am. Chem. Soc. 1986,
108, 7119-7120.
and dealuminated Y samples are also in relatively close (9) Baltusis, L.; Frye, J. S.; Maciel, G. E. J. Am. Chem. Soc. 1987,
agreement. Finally, in the silica-alumina sample, the numbers 109, 40-46.
obtained by NMR and IPA/TPD are the same, within the (10) Kofke, T. J. G.; Gorte, T. J.; Kokotailo, G. T.; Farneth, W. E. J.
experimental error of both techniques. Catal. 1989, 115, 265.
(11) Peters, A. W. Instrumental methods of FCC catalyst characterization.
In Fluid Catalytic Cracking: Science and Technology; Magee, J. S.,
Conclusion Mitchell, M. M., Jr., Eds.; Elsevier: New York, 1993; Vol. 76, pp 183-
221.
We have demonstrated that it is indeed possible to use solid- (12) Sheng, T.-C.; Gay, I. D. J. Catal. 1994, 145, 10-15.
state 31P NMR of the probe molecule TMPO to determine the (13) Sang, H.; Chu, H. Y.; Lunsford, J. H. Catal. Lett. 1994, 26, 235-
types and amounts of acid sites on a variety of zeolitic and 246.
(14) Maciel, G. E.; Ellis, P. D. NMR Characterization of Silica and
amorphous materials. The concentrations of Brønsted sites Alumina Surfaces. In NMR Techniques in Catalysis; Bell, A. T., Pines, A.,
determined by this method are in good agreement with those Eds.; Marcel Dekker: New York, 1994; Vol. 55, pp 231-309.
determined via IPA/TPD. The 31P NMR chemical shift range (15) Lunsford, J. Top. Catal. 1997, 4, 91-98.
(16) Kao, H.-M.; Grey, C. P. Chem. Phys. Lett. 1996, 259, 459-464.
for Brønsted complexes was determined to be 50-75 ppm. (17) Kao, H.-M.; Grey, C. P. J. Am. Chem. Soc. 1997, 119, 627-628.
Contrary to previous reports, we report that the chemical shift (18) van Eck, E. R. H.; Janssen, R.; Maas, W. E. J. R.; Veeman, W. S.
of a TMPO/Lewis acid complex is 37 ppm based on our studies Chem. Phys. Lett. 1990, 174, 428-432.
(19) Juskelis, M.; Peters, A. W.; Slanga, J. P.; Roberie, T. G. J. Catal.
of dehydroxylated alumina. The quantitation of Lewis acid 1992, 138, 391.
strengths and sites is important for an increased understanding (20) Gullion, T.; Schaefer, J. J. Magn. Reson. 1989, 81, 196-200.
of the catalytic nature of these systems, and the newly assigned (21) Fyfe, C. A.; Mueller, K. T.; Grondey, H.; Wong-Moon, K. C. Chem.
signal can in principle be used to determine the amount of Lewis Phys. Lett. 1992, 199, 198-204.
(22) Fyfe, C. A.; Mueller, K. T.; Grondey, H.; Wong-Moon, K. C. J.
acidity in a given material. This is the subject of current studies Phys. Chem. 1993, 97, 13484-13495.
in our group. Further studies of 27Al/31P distances (as well as (23) Herzfeld, J.; Berger, A. E. J. Chem. Phys. 1980, 73, 6021-6030.
1H/31P interactions) at Brønsted acid sites are also in progress. (24) Eichele, K.; Waylishen, R. E. HBA 1.2 Dalhousie University, 1997.
(25) Peri, J. B. J. Phys. Chem. 1965, 69, 211-220.
(26) Frye, J. S., personal communication.
Acknowledgment. The authors wish to thank J. Swain for (27) Kariuki, B. M.; Harris, K. D. M.; Philp, D.; Robinson, J. M. A. J.
conducting the IPA/TPD experiments, A. Nadjadi for his Am. Chem. Soc. 1997, 119, 12679-12680.