Solid phosphoric acid catalyst: A multinuclear NMR and theoretical study

Article abstract of DOI:10.1021/ja9813461

The synthesis, structure, and acid function of solid phosphoric acid (SPA) catalyst were studied in detail. ³¹P and ²⁹Si MAS NMR and X-ray powder diffraction identified the following crystalline silicon phosphate phases in SPA: Si₅O(PO₄)₆, hexagonal-SiP₂O₇, Si(HPO₄)₂·H₂O, and SiHP₃O10. The acidity of SPA is due to a liquid or glassy solution of phosphoric acid oligomers supported on the silicon phosphate phases. ¹⁵N MAS NMR of adsorbed pyridine-¹⁵N and ¹³C MAS NMR of adsorbed acetone-2- ¹³C showed Bronsted acid sites and no Lewis acid sites. ¹H→¹⁵N→³¹P and ¹H→¹³C→³¹P double cross polarization MAS NMR of the probe molecules provided a rare opportunity to use NMR to unambiguously localize chemisorption sites; the probe molecules are complexed to phosphoric acid and pyrophosphoric acid but not to the silicon phosphate phases. In situ NMR of the oligomerization of propene on SPA suggests that propene quantitatively reacts with phosphoric acid and its oligomers to form isopropyl phosphate, and formation of this very stable intermediate accounts for the lower olefin oligomerization activity of SPA relative to acidic zeolites. Theoretical calculations including geometries at B3LYP/6-311+G(d,p) and chemical shifts at GIAO-MP2/tzp/dz were used to model complexation of acetone or propene to SPA, and these support our conclusions.

Full text of DOI:10.1021/ja9813461

8502
J. Am. Chem. Soc. 1998, 120, 8502-8511
Solid Phosphoric Acid Catalyst: A Multinuclear NMR and
Theoretical Study
Thomas R. Krawietz, Ping Lin, Karen E. Lotterhos, Paul D. Torres, Dewey H. Barich,
Abraham Clearfield, and James F. Haw*
Contribution from the Department of Chemistry, Texas A&M UniVersity, P.O. Box 300012,
College Station, Texas 77842-3012
ReceiVed April 20, 1998
Abstract: The synthesis, structure, and acid function of solid phosphoric acid (SPA) catalyst were studied in
detail. 31P and 29Si MAS NMR and X-ray powder diffraction identified the following crystalline silicon
phosphate phases in SPA: Si5O(PO4)6, hexagonal-SiP2O7, Si(HPO4)2‚H2O, and SiHP3O10. The acidity of
SPA is due to a liquid or glassy solution of phosphoric acid oligomers supported on the silicon phosphate
phases. 15N MAS NMR of adsorbed pyridine-15N and 13C MAS NMR of adsorbed acetone-2-13C showed
Brønsted acid sites and no Lewis acid sites. 1Hf15Nf31P and 1Hf13Cf31P double cross polarization MAS
NMR of the probe molecules provided a rare opportunity to use NMR to unambiguously localize chemisorption
sites; the probe molecules are complexed to phosphoric acid and pyrophosphoric acid but not to the silicon
phosphate phases. In situ NMR of the oligomerization of propene on SPA suggests that propene quantitatively
reacts with phosphoric acid and its oligomers to form isopropyl phosphate, and formation of this very stable
intermediate accounts for the lower olefin oligomerization activity of SPA relative to acidic zeolites. Theoretical
calculations including geometries at B3LYP/6-311+G(d,p) and chemical shifts at GIAO-MP2/tzp/dz were
used to model complexation of acetone or propene to SPA, and these support our conclusions.
The application of modern experimental and theoretical
been proposed that the acid function is incidental to a redox
mechanism.15
methods is providing an understanding of the structure and
function of the solid acid catalysts used in chemical processes.
Considerable progress has been made in the characterization
and reaction chemistry of isolated Brønsted sites in alumino-
silicate zeolites,1-8 but less is known about other solid acid
catalysts. Supported liquid-phase (SLP) catalysts are second
only to zeolites in numbers of materials, processes, patents, and
journal publications. SLP catalysts can be very straightforward
materials, liquid acids dispersed on metal oxide supports.9 But
consider sulfuric acid dispersed on zirconia, sulfated zirconia.10
This alkylation and cracking catalyst resists both characterization
and consensus; there is little agreement regarding surface
structure,11 nature,12 and strength of acidity,13,14 and it has also
For our first NMR and theoretical investigation of an SLP
catalyst we selected solid phosphoric acid (SPA). This catalyst
was developed in the 1930s and is used to this day in a number
of processes including propene oligomerization and alkylation
of benzene with propene to form cumene.16-22 Remarkable for
a highly successful solid acid catalyst, the literature reports no
previous systematic investigation of SPA using modern spec-
troscopic or theoretical methods and especially no solid-state
NMR studies. Here we report a thorough investigation of SPA
using multinuclear NMR, X-ray powder diffraction, and ab initio
theoretical methods. A large number of SPA samples were
prepared with different synthesis conditions.23-25
We find that only ca. 60% of the phosphorus in SPA is
present as free acids, and much of that is oligomerized. In
agreement with previous reports,23,24 we confirm the presence
* To whom correspondence should be addressed.
(1) Haw, J. F.; Nicholas, J. B.; Xu, T.; Beck, L. W.; Ferguson, D. B.
Acc. Chem. Res. 1996, 29, 259-267.
(2) van Santen, R. A.; Kramer, G. J. Chem. ReV. 1995, 95, 637-660.
(3) Bra¨ndle, M.; Sauer, J. J. Am. Chem. Soc. 1998, 120, 1556-1570.
(4) Haase, F.; Sauer J. J. Am. Chem. Soc. 1995, 117, 3780-3789.
(5) Haw, J. F.; Xu, T.; Nicholas, J. B.; Goguen, P. W. Nature 1997,
389, 832-835.
(6) Goguen, P. W.; Xu, T.; Barich, D. H.; Skloss, T. W.; Song, W.;
Wang, Z.; Nicholas, J. B.; Haw, J. F. J. Am. Chem. Soc. 1998, 120, 2650-
2651.
(7) Blaszkowski, S. R.; van Santen, R. A. J. Am. Chem. Soc. 1996, 118,
5152-5153.
(8) Blaszkowski, S. R.; van Santen, R. A. J. Am. Chem. Soc. 1997, 119,
5020-5027.
(9) Villadsen, J.; Livbjerg, H. Catal. ReV.-Sci. Eng. 1978, 17, 203-272.
(10) Song, X.; Sayari, A. Catal. ReV.-Sci. Eng. 1996, 38, 329-412.
(11) Adeeva, V.; de Haan, J. W.; Janchen, J.; Lei, G. D.; Schunemann,
V.; van de Ven, L. J. M.; Sachtler, W. M. H.; van Santen, R. A. J. Catal.
1995, 151, 364-372.
(12) Clearfield, A.; Serrette, G. P. D.; Khazi-Syed, A. H. Catal. Today
1994, 20, 295-312.
(13) Hino, M.; Kobayashi, S.; Arata, K. J. Am. Chem. Soc. 1979, 101,
6439-6441.
(14) Tabora, J. E.; Davis, R. J. J. Am. Chem. Soc. 1996, 118, 12240-
12241.
(15) Farcasiu, D.; Ghenciu, A.; Li, J.-Q. J. Catal. 1996, 158, 116-127.
(16) (a) Ipatieff, V. U.S. Patent No. 1,993,512, 1935. (b) Ipatieff, V.
U.S. Patent No. 1,993,513, 1935.
(17) Ipatieff, V.; Schaad, R. E. U.S. Patent No. 2,275,182, 1942.
(18) Ipatieff, V.; Schaad, R. E. U.S. Patent No. 2,157,208, 1939.
(19) McMahon, J. F.; Bednars, C.; Solomon, E. AdV. Pet. Chem. Refin.
1963, 7, 284-321.
(20) Jones, E. K. AdV. Catal. 1956, 8, 219-238.
(21) Weisang, E.; Engelhard, P. A. Bull. Soc. Chim. Fr. 1968, 1811-
1820.
(22) Oblad, A. G.; Mills, G. A.; Heinemann, H. In Catalysis; Emmett,
Ed.; Reinhold: New York, 1958; Vol. VI, pp 341-406.
(23) Durand, B.; Lenzi, M.; Boulle´, A. Bull. Soc. Chim. Fr. 1972, 2,
442-456.
(24) Cavani, F.; Girotti, G.; Terzoni, G. Appl. Catal. A: Gen. 1993, 97,
177-196.
(25) Ipatieff, V. N.; Schaad, R. E.; Shaney, W. B. Catalytic Polymeri-
zation of Olefins; Oxford University Press: London, 1953; Vol. V, Part II.
S0002-7863(98)01346-8 CCC: $15.00 © 1998 American Chemical Society
Published on Web 08/05/1998

Solid Phosphoric Acid Catalyst
J. Am. Chem. Soc., Vol. 120, No. 33, 1998 8503
With the exception of SiHP3O10, the crystalline components of SPA
were synthesized in >95% purity, as measured by quantitative 31P MAS
NMR. Si5O(PO4)6 was synthesized by heating an evacuated quartz
ampule containing 5.0 g of silica and 11.8 g of phosphorus pentoxide
to 858 K for 24 h. Si(HPO4)2‚H2O was prepared by heating a slurry
of 5.0 g of silica and 19.2 g of 85% phosphoric acid to 473 K in a
shallow-form dish for 336 h. The hexagonal polymorph of SiP2O7 was
prepared by heating a mixture of 5.0 g of silica with 28.8 g of 85%
phosphoric acid to 873 K for 48 h. 31P and 29Si MAS NMR and powder
XRD were used to confirm the identity and purity of each silicon
phosphate phase.
SPA catalyst samples were prepared in 70 mL of shallow-form,
quartz-coated alumina dishes (Coors) from viscous pastes consisting
of 5.0 g of silica and 28.8 g of 85% phosphoric acid (Si:P mol ratio of
1.0:3.0). This mixture was heated to 573 K for 48 h to form “as-
prepared SPA”.
Cold water washes of SPA catalyst were used to remove phosphoric
acid and oligomerized phosphoric acids without hydrolysis of silicon
phosphate phases. Typically, 10 mL of deionized water/g of catalyst
was vigorously stirred for 10 min, followed by vacuum filtration and
drying the resultant solid in an oven at 473 K for 10 min. 31P MAS
NMR showed that >98% of phosphoric acid is removed from the
catalyst without degradation of the silicon phosphate phases.
NMR Spectroscopy. 1H, 13C, 15N, 29Si, and 31P MAS NMR spectra
were obtained on either a Chemagnetics CMX-360 or a home-built
200 MHz instrument. All spectra were externally referenced with
acetone (1H, 2.11 ppm relative to 1H in tetramethylsilane), HMB (13C,
methyl at 17.4 ppm relative to 13C in tetramethylsilane), ammonium-
15N nitrate (15N, ammonium at -359 ppm relative to 15N in ni-
tromethane), sodium hexafluorosilicate (29Si, -188.6 ppm relative to
29Si in tetramethylsilane), or (2S,3S)-chiraphos (31P, upfield resonance
at -13.3 ppm relative to 85% phosphoric acid). All Bloch decay
spectra reported used quantitative conditions. 1Hf13Cf31P and
1Hf15Nf31P double cross polarization spectra were acquired at 198
K on a triple resonance probe from Otsuka Electronics spinning 7.5
mm zirconia rotors. All spectra presented used active spin speed
control.
of crystalline silicon orthophosphate (Si5O(PO4)6) and hexagonal
silicon pyrophosphate (SiP2O7) in SPA. We extend the
identification of silicon phosphates in SPA to include silicon
hydrogen phosphate monohydrate (Si(HPO4)2‚H2O)26,27 and
silicon hydrogen tripolyphosphate (SiHP3O10).26 We also
synthesized the silicon phosphates as either pure or majority
phases and surveyed their NMR behavior, which is exceptional
due to the presence of octahedral silicon in these phases.
The nature and strength of the acid sites on SPA was
investigated by measuring the NMR spectra of common probe
molecules and in situ NMR1,28 of propene oligomerization.29
Propene oligomerization on SPA occurs at much higher tem-
peratures than on zeolite acid catalysts, because it first reacts
with phosphoric acid to form isopropyl phosphate. The probe
molecule studies reveal that all of the acid sites on SPA are
Brønsted; there are no Lewis sites. 1Hf15Nf31P and
1Hf13Cf31P double cross polarization experiments prove that
the basic probe molecules are complexed to phosphoric acid
and its oligomers. The chemical shift of acetone is larger on
SPA than on zeolite HZSM-5, suggesting that SPA is the
stronger acid. This interpretation is counter to the conventional
ordering of acid strengths based on propene oligomerization
activity. Quantitative conversion of propene to isopropyl
phosphate on SPA accounts for the difference in oligomerization
activity on the two media. We used theoretical methods to better
understand the acid strength of SPA. We optimized a number
of geometries of acetone or propene complexed to species
known to be present in SPA and then calculated theoretical
chemical shift tensors. These calculations provide several lines
of evidence that the acid strength of SPA is greater than that of
zeolites.
Experimental Section
Materials. Pyridine-15N (98+% 15N), acetone-2-13C (99% 13C), and
5.8 N ammonium-15N hydroxide (98% 15N) were purchased from
Cambridge Isotopes. Propene-1-13C (99% 13C) and propene-2-13C (99%
13C) were purchased from CDN Isotopes. Ammonium-15N nitrate (99%
15N) was purchased from MSD Isotopes. 2-Propanol-2-13C (99% 13C)
was purchased from Isotec. Kieselgu¨hr (91.9% SiO2), electronic grade
silicon(IV) oxide (Puratrem, 99.999%), sodium hexafluorosilicate
(99%), and (2S,3S)-chiraphos (99%) were purchased from Strem
Chemicals. Hexamethylbenzene (HMB), H3PO4 (98%), pyrophosphoric
acid (85+%), and acetone (spectrophotometric grade, 99.5%) were
purchased from Aldrich. H3PO4 (85%) and phosphorus pentoxide
(99%) were purchased from EM Science. Ammonium-15N dihydrogen
phosphate was synthesized from the dropwise addition of 5.0 g of 5.8
N ammonium-15N hydroxide to an ice-cooled solution of 3.34 g of 85%
phosphoric acid diluted in 10 mL of deionized water. The water was
evaporated on a stirred hot plate, and the remaining solid was dried in
an oven at 423 K.
Acid-washed kieselgu¨hr was prepared with 10 mL of 0.1 M HCl/g
of kieselgu¨hr to remove excess Fe2O3. The slurry was then filtered
and dried at 473 K. This procedure was carried out a total of three
times. Following acid leaching, the kieselgu¨hr contained 0.65 wt %
Fe, as determined by ICP/AE (Galbraith Labs).
Silicon Phosphate Synthesis. Crystalline silicon phosphates were
synthesized with 85% phosphoric acid and either acid-washed kiesel-
gu¨hr or electronic grade silicon(IV) oxide as sources of phosphate and
silica, respectively. The silica source was dried in air at 473 K prior
to use. After preparation, all materials were immediately ground with
quartz coated ceramic utensils and sealed to prevent adsorption of
atmospheric moisture.
X-ray Diffraction. X-ray powder diffraction patterns of silicon
phosphates were obtained on a Seifert-Scintag PAD V automated
diffractometer with Cu KR radiation. The X-ray source was an anode
operating at 40 kV and 30 mA with a copper target and filtered with
nickel foil (λ ) 1.5418 Å). Data were collected between 5° and 62°
in 2θ with a step size of 0.04°. The scan rate was 2.4 s/step. The
measured X-ray diffraction patterns were compared with those reported
in the JCPDS database.
Computations. All structures were fully optimized with density
functional theory (DFT) with use of the 6-311+G(d,p) basis set.
Becke’s three-parameter hybrid method,30 using the LYP correlation
functional,31 was used in all DFT calculations. Harmonic frequency
analyses were performed on geometries optimized at the B3LYP/6-
31G(d) level of theory with use of analytical second derivatives.
The method of gauge-invariant atomic orbitals (GIAO)32,33 was used
for chemical shift calculations. We used Alrich’s34 tzp {51111/311/
1} (with 6 Cartesian d orbitals) on carbon and oxygen atoms, tzp
{5121111/51111/1} on phosphorus, and dz {31} on the hydrogens in
all of the chemical shift calculations. 13C chemical shifts were
calculated at the GIAO-RHF/tzp/dz level of theory and referenced
against tetramethylsilane calculated at the same level. All of the above
calculations were performed within the Gaussian 94 program package.35
For small clusters, we also performed chemical shift calculations at
the GIAO-MP2/tzp/dz level of theory36 with ACES II.37
Results
31P and 29Si MAS NMR of SPA Catalysts. We first studied
catalyst samples synthesized using kieselgu¨hr as the silica source
(30) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652.
(31) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789.
(32) Ditchfield, R. Mol. Phys. 1974, 27, 789-807.
(33) Wolinski, K.; Hinton, J. F.; Pulay, P. J. Am. Chem. Soc. 1990, 112,
8251-8260.
(34) Schafer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1992, 100,
5829-5835.
(26) Lelong, B. Ann. Chim. 1964, 9, 229-260.
(27) Mudrakovskii, I. L.; Mastikhin, V. M.; Shmachkova, V. P.;
Kotsarenko, N. S. Chem. Phys. Lett. 1985, 120, 424-426.
(28) Xu, T.; Haw, J. F. Top. in Catal. 1997, 4, 109-118.
(29) Haw, J. F.; Richardson, B. R.; Oshiro, I. S.; Lazo, N. L.; Speed, J.
A. J. Am. Chem. Soc. 1989, 111, 2052-2058.

8504 J. Am. Chem. Soc., Vol. 120, No. 33, 1998
Krawietz et al.
Figure 1. 80.8 MHz 31P MAS Bloch decay spectra of SPA samples
heated under vacuum for 2 h at the indicated temperature: (a) 298, (b)
413, (c) 453, (d) 493, (e) 533 K. All spectra were acquired at 298 K
and are the result of 12 transients utilizing a 60 s recycle delay and a
spinning speed of 5000 Hz. See text for assignments.
Figure 2. 80.8 MHz variable-temperature 31P MAS Bloch decay
spectra of a SPA sample evacuated at 373 K for 2 h. Phosphoric acid
(ca. 2 ppm) and oligomerized acids (-12 and -24 ppm) are in a glassy
state at 248 K and below, and in a solution state at higher temperatures.
Broadening at the highest temperatures studied may reflect chemical
exchange. All spectra are the result of 12 transients utilizing a 60 s
recycle delay and a spinning speed of 5000 Hz.
as in industrial preparations.16,17 At 298 K several of the 31P
resonances were greater than 10 ppm in width, and those
assigned to phosphoric acid and its oligomers broadened with
increasing temperature so as not to be detected at 398 K and
above. The total integrated intensity of the MAS spectra
strongly deviated from the Curie law at higher temperatures.
Elemental analysis showed that SPA prepared with kieselgu¨hr
contained 3600 ppm iron. We observed similar temperature-
dependent 31P NMR line widths for a 0.036 M FeCl3 (3400
ppm Fe3+) solution in 85% H3PO4. To obtain NMR results
free of paramagnetic broadening, we synthesized SPA samples
using high purity, electronic grade silicon(IV) oxide. X-ray
powder diffraction (vide infra) showed that the crystalline phases
were identical for catalysts prepared from either silica source,
and other than paramagnetic broadening, the NMR results were
identical as well. All NMR results for SPA catalysts reported
below were obtained with use of electronic grade silicon(IV)
oxide.
effect. These downfield signals are due to phosphoric acid (ca.
2 ppm), terminal (-12 ppm), and internal (-24 ppm) phosphate
groups in phosphoric acid oligomers.38 These assignments were
confirmed by solution NMR studies of washes of the catalyst
samples with anhydrous ethanol, MAS NMR studies of solid
98% H3PO4 and pyrophosphoric acid, and other standard
procedures. The average chain length of the phosphoric acid
species on the catalysts, nav, can be determined from the
integrated intensities of the three signals:
nav ) (AH PO + Aterm. + Aintern.)/(AH PO + 0.5Aterm.
)
3
4
3
4
As-prepared SPA has an average chain length of 2.0 (pyro-
phosphoric acid) and samples evacuated at 453 K and above
have average chain lengths approaching 3.0 (tripolyphosphoric
acid). The concentrations of very strong phosphoric acid
solutions are commonly reported on a percent P2O5 basis, and
the acidic phase of as-prepared SPA corresponds to ap-
proximately 80% P2O5.
Figure 1 shows 31P MAS spectra measured at 298 K for five
catalysts that were heated under vacuum at various temperatures
prior to sealing in MAS rotors under an anhydrous N2
atmosphere. Processes using SPA require careful control of
the water content in the feed,9,24 and the large changes in the
downfield signals in the NMR spectra as a function of drying
conditions motivate the molecular-level interpretation of this
Figure 2 reports variable-temperature 31P MAS spectra of an
SPA sample that was treated by evacuation at 373 K. The
signals due to phosphoric acids broaden considerably at tem-
peratures below 298 K. The pure acids are low-melting solids
(e.g., 100% H3PO4 melts at 315 K), and mixtures of phosphoric
and pyrophoshoric acids are viscous liquids at 298 K. The low-
temperature NMR behavior in Figure 2 is consistent with the
freezing out of various conformations and intermolecular
interactions as in polymers below their glass transitions.
Residual dipolar broadening is also a potential contributor to
line broadening at reduced temperature. The signals due to
(35) Gaussian 94, Revision E.2. Frisch, M. J.; Trucks, G. W.; Schlegel,
H. B.; Gill, P. M. W.; Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; T.
Keith; Petersson, G. A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham,
M. A.; Zakrzewski, V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.;
Stefanov, B. B.; Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala,
P. Y.; Chen, W.; Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts,
R.; Martin, R. L.; Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart,
J. P.; Head-Gordon, M.; Gonzalez, C.; Pople, J. A.; Pittsburgh, PA, 1995.
(36) Gauss, J. Chem. Phys. Lett. 1992, 191, 614-620.
(37) ACES II, an ab initio quantum chemical program system: Stanton,
J. F.; Gauss, J.; Watts, J. D.; Lauderdale, W. J.; Bartlett, R. J.
(38) Crutchfield, M. M.; Callis, C. F.; Irani, R. R.; Roth, G. C. Inorg.
Chem. 1962, 1, 813-817.

Solid Phosphoric Acid Catalyst
J. Am. Chem. Soc., Vol. 120, No. 33, 1998 8505
Figure 4. 71.5 MHz 29Si MAS Bloch decay spectra. (a) SPA contains
silicon in both tetrahedral (-119 ppm) and octahedral (ca. -210 to
-217 ppm) environments. (b) Si(HPO4)2‚H2O contains only octahe-
drally coordinated silicon at -210 ppm. Unreacted silica ([) is also
present at -109 ppm. (c) Si5O(PO4)6 shows isotropic peaks at -119,
-213, and -217 ppm in the 2:2:1 intensity ratio predicted from the
crystal structure. Quantitative 29Si spectra in each case required 256
transients and a 20 s recycle delay. All spectra were acquired with a
spinning speed of 5000 Hz.
Figure 3. 80.8 MHz variable-temperature 31P CP MAS spectra of a
SPA sample evacuated at 373 K for 2 h. Relative cross polarization
efficiencies, estimated by the comparison of these spectra with the
corresponding Bloch decay spectra (cf., Figure 2), reflect both the
presence or absence of protons and the extent of dynamical averaging.
All spectra are the result of 64 transients with a 1 ms 1Hf31P contact
time utilizing a 10 s recycle delay and a spinning speed of 5000 Hz.
Asterisks denote spinning sidebands.
phases for NMR characterization. Figure 4 reports 29Si MAS
NMR spectra of our preparations of SPA, silicon hydrogen
phosphate monohydrate, (Si(HPO4)2‚H2O), and silicon ortho-
phosphate, (Si5O(PO4)6); the 29Si spectra are also compared with
that of a SPA sample. A similar procedure yielded the SiHP3O10
assignment. The upfield 29Si chemical shifts (ca. 210-220 ppm
in Figure 4) are from octahedrally coordinated silicon.39 For
example, Clearfield and co-workers previously used powder
methods to solve the crystal structure of Si5O(PO4)6 and
determined that the unit cell contains five silicon atoms: two
equivalent tetrahedral sites, two equivalent octahedral sites, and
a second octahedral site.40 The 29Si spectrum of Si5O(PO4)6 is
thus completely assigned by comparison to the crystal structure.
Silicon pyrophosphate (SiP2O7) is a minority species in SPA,
but its characterization caused us a disproportionate amount of
trouble due to the existence of a number of polymorphs.41-45
We synthesized pure samples of the four well-established
polymorphs and found that their 31P spectra (Figure 5) were in
agreement with their solved crystal structures. Our NMR work
has found only the hexagonal polymorph of SiP2O7 in SPA.
Principal components of the 31P chemical shift tensors were
determined by the method of Herzfeld and Berger46 for most
phosphoric acids also broaden above 448 K; possible dynamics
at high temperatures may include chemical exchange of
phosphate units between various oligomers, but we did not
investigate this any further.
The upfield signals in Figures 1 and 2 are due to crystalline
silicon phosphate phases formed by chemical reaction between
phosphoric acid and silica. The evidence for these assignments
is described below. These signals are due to silicon orthophos-
phate (Si5O(PO4)6, -44 ppm) and smaller amounts of silicon
hydrogen phosphate monohydrate (Si(HPO4)2‚H2O, -31 ppm)
and silicon hydrogen tripolyphosphate (SiHP3O10, -35 ppm).
A very small amount (typically <1%) of hexagonal silicon
pyrophosphate (SiP2O7, -52 ppm) can also be observed in the
31P MAS spectra of many SPA samples.
The structure and dynamics of the components of SPA
catalyst are also reflected in the signal intensities observed in
1Hf31P cross polarization spectra (Figure 3). The signals due
to phosphoric acids effectively cross polarize only at low
temperatures; at 298 K and above these species are in a
liquidlike environment. The relative cross polarization efficien-
cies of the various silicon phosphate phases reflect the presence
or absence of protons; thus, the signals due to Si(HPO4)2‚H2O
and SiHP3O10 are greatly emphasized relative to the quantitative
Bloch decay spectra (Figure 2) and the peak at -44 ppm due
to Si5O(PO4)6 is by comparison attenuated. The latter phase
must contain some hydrogen as impurities or inclusions;
otherwise there would be no cross polarization response, as we
observe for the pure compound.
(39) Weeding, T. L.; de Jong, B. H. W. S.; Veeman, W. S.; Aitken, B.
G. Nature 1985, 318, 352-353.
(40) Poojary, D. M.; Borade, R. B.; Clearfield, A. Inorg. Chim. Acta
1993, 208, 23-29.
(41) Bissert, G.; Liebau, F. Naturwissenschaften 1969, 56, 212.
(42) Poojary, D. M.; Borade, R. B.; Campbell, F. L., III; Clearfield, A.
J. Solid State Chem. 1994, 112, 106-112.
Using X-ray powder diffraction measurements, other workers
have previously reported that SPA contains silicon orthophos-
phate (Si5O(PO4)6).23,24 A careful examination of our X-ray
powder data on SPA samples suggested other specific silicon
phosphate phases, which we then synthesized as pure or majority
(43) Liebau, F.; Bissert, G.; Koppen, N. Z. Anorg. Allg. Chem. 1968,
359, 113-134.
(44) Makart, H. HelV. Chim. Acta 1967, 50, 399-405.
(45) Hartmann, P.; Jana, C.; Vogel, J.; Ja¨ger, C. Chem. Phys. Lett. 1996,
258, 107-112.
(46) Herzfeld, J.; Berger, A. E. J. Chem. Phys. 1980, 73, 6021-6030.

8506 J. Am. Chem. Soc., Vol. 120, No. 33, 1998
Krawietz et al.
Figure 5. 80.8 MHz 31P MAS Bloch decay spectra of four polymorphs
of SiP2O7. In each case the parenthetical information denotes reaction
temperature, reaction time, and initial Si:P molar ratio. (a) cubic-SiP2O7
(1273 K, 4 h, 1.0:3.0) shows that many of the 11 crystallographically
distinct phosphorus nuclei are resolved. (b) monoclinic-SiP2O7 (1273
K, 4 h, 1.0:4.0) shows two resonances at -46.9 and -49.1 ppm. (c)
tetragonal-SiP2O7 (973 K, 4 h, 1.0:1.0) shows two resonances at -46.7
and -54.3 ppm. (d) hexagonal-SiP2O7 (773 K, 48 h, 1.0:3.0) shows a
single resonance at -52.7 ppm. Each spectrum is the result of 512
transients utilizing a 2 s recycle delay and a spinning speed of 5000
Hz. Asterisks denote spinning sidebands.
Figure 6. 36.5 MHz 15N MAS and 145.6 MHz 31P MAS spectra of a
SPA sample evacuated at 298 K followed by adsorption of 1.6 mmol/g
pyridine-15N. The 15N MAS spectrum (298 K shown) has a single
resonance at -182 ppm, indicative of coordination to a Brønsted site.
We used the 15N spin on pyridine as means of determining which
phosphorus atoms in the catalyst were close in space to the adsorbed
base. To obtain efficient 1Hf15Nf31P double cross polarization (4 ms
1Hf15N contact, 6 ms 15Nf31P contact), we measured the 31P MAS
NMR spectra at 153 K using a spinning speed of 3600 Hz. This
experiment shows unambiguously that the pyridine is complexed to
phosphoric acid and its oligomers, not to the silicon phosphate phases.
1Hf31P cross polarization and 31P Bloch decay spectra acquired under
similar conditions are shown for comparison. f, 2, and b denote
spinning sidebands of phosphoric acid, terminal groups of oligomerized
acid, and silicon orthophosphate, respectively.
of the silicon phosphate phases and are reported in the
Supporting Information.
NMR Characterization of Acid Sites and Reaction Mech-
anism. The above structural characterization of SPA suggests
little potential for Lewis sites on this catalyst; nevertheless, there
are anecdotal accounts of Lewis acidity for silicon phosphates.
We investigated this hypothesis using the 15N MAS spectrum
of adsorbed pyridine47 as a profile of SPA acidity. We studied
a number of samples with various pyridine loadings and
measured the 15N spectra over a range of temperatures to guard
against dynamical averaging. The pyridine studies revealed only
Brønsted sites and no evidence of Lewis sites. Pyridine-15N
on SPA provides the NMR spectroscopist an unusually straight-
forward opportunity to locate an adsorbate molecule on a
complex heterogeneous catalyst. Figure 6 reports various CP/
MAS spectra measured at 153 K of a sample of 1.6 mmol/g of
pyridine-15N on a SPA sample prepared by evacuation at 298
K for 30 min. In these spectra the spinning speed was
deliberately set to a low value, 3600 Hz, to improve cross
polarization efficiency, with the consequence that the 31P MAS
spectra are crowded with spinning sidebands. Nevertheless, the
isotropic 31P signals for phosphoric acid (ca. 0 ppm), terminal
phosphates (-12 ppm), and several silicon phosphate phases
are clearly resolved. Quantitative 15N Bloch decay MAS NMR
shows that all of the pyridine is protonated (-182 ppm).48 Figure
6 also reports the 1Hf15Nf31P double cross polarization
spectrum of this sample. Cross polarization is a “through space”
method of polarization transfer, and its rate is highly dependent
upon distance. The double cross polarization spectrum is
expected to report only those 31P spins that are within a few
angstroms of a 15N spin. Indeed, only phosphoric acids are
observed when the magnetization is transferred though pyridine.
The adsorbed base is not associated with the silicon phosphate
phases.
Most of the processes that use SPA convert propene into other
products. This olefin and associated intermediates and products
have been the subject of detailed in situ NMR studies of
reactions on zeolites and metal halide acids.29,49 Propene is
readily adsorbed into the channel structures of zeolites, even at
the lowest temperatures. On acidic zeolites, propene forms a
π complex with the Brønsted site at cryogenic temperatures,
which is observed as a downfield shift of the C-2 resonance.
The zeolite studies also show that propene reversibly forms a
framework bound isopropoxyl species below room tempera-
ture.29,49 This process results in 1, 3 13C label scrambling, but
not 1, 2, 3 13C scrambling as is seen on frozen superacids50
(47) Haw, J. F.; Chuang, I.-S.; Hawkins, B. L.; Maciel, G. E. J. Am.
Chem. Soc. 1983, 105, 7206-7207.
(48) Xu, T.; Kob, N.; Drago, R. S.; Nicholas, J. B.; Haw, J. F. J. Am.
Chem. Soc. 1997, 119, 12231-12239.
(49) Nicholas, J. B.; Xu, T.; Haw, J. F. Top. Catal. 1998, 6, 141-149.
(50) Nicholas, J. B.; Xu, T.; Barich, D. H.; Torres, P. D.; Haw, J. F. J.
Am. Chem. Soc. 1996, 118, 4202-4203.

Solid Phosphoric Acid Catalyst
J. Am. Chem. Soc., Vol. 120, No. 33, 1998 8507
Figure 8. Summary of loading-dependent 13C isotropic chemical shifts
of the carbonyl carbon of acetone-2-13C on a SPA sample prepared by
evacuation at 298 K. Measurements were performed at 153 K to obtain
spectra in the slow-exchange regime. Equivalent loading is defined as
the ratio of acetone molecules to phosphate units in phosphoric acid
and its oligomers.
NMR study carried out in a sealed rotor. In some in situ
experiments (not shown) we observed low concentrations of
free propene immediately prior to the onset of oligomerization.
This presumably reflects the decomposition of isopropyl phos-
phate at higher temperatures.
Acetone-2-13C is widely applied as a probe molecule for the
study of solid acids.48,51 In the case of Brønsted sites on zeolites,
the shift is commonly interpreted in terms of conventional
notions of relative acid strength. For the case of zeolite HZSM-
5, a strongly acidic zeolite, this value is ca. 223 ppm. We
observed that acetone has a shift of 228.5 ppm when a very
low loading is adsorbed on SPA evacuated at 298 K, and values
as high as 231 ppm were obtained for samples evacuated at
higher temperatures. On solid 98% H3PO4 and solid 85+%
pyrophosphoric acid, we found acetone shifts of 228.1 and 229.8
ppm, respectively.
The possibility of a role for the crystalline components in
the downfield shifts observed for acetone on SPA was ruled
out by several experiments. We used a literature procedure24
to remove the phosphoric acid component of SPA with cold
water washings followed by drying and observed that acetone
has a chemical shift of only 213 ppm on such materials. The
distance dependence of magnetization transfer by cross polariza-
tion was exploited to establish unambiguously that basic probe
molecules on SPA are complexed to phosphoric acid species
and are not associated with the mineral phases. 1Hf13Cf31P
double cross polarization experiments were carried out for
acetone-2-13C on SPA. These were analogous to the
1Hf15Nf31P double cross polarization experiments with py-
ridine-15N reported in Figure 6. The 1Hf13Cf31P experiments
(reported in the Supporting Information) show that acetone on
SPA is complexed to phosphoric acid species and is not
associated with the silicon phosphate phases.
Figure 7. 90.5 MHz 13C in situ MAS NMR study of the reactions of
propene-2-13C on a sample of SPA prepared by evacuation at 373 K
followed by adsorption at 153 K. Propene is excluded from the catalyst
at low temperatures, and reacts with the catalyst to form isopropyl
phosphate (C-2 at 80 ppm) at 273 K and above. Oligomerization occurs
at 398 K.
that reversibly form a free isopropyl carbenium ion. On zeolites
propene oligomerizes in the vicinity of room temperature; on
SPA propene oligomerization occurs at higher temperatures.
This has been used to argue that SPA is a weaker acid than
zeolites.
We carried out a number of in situ NMR studies of propene
oligomerization on SPA using methods very similar to our
zeolite studies. Figure 7 reports representative results from in
situ studies of propene-2-13C on SPA prepared by evacuation
at 373 K. A signal at 134 ppm is observed for propene on
SPA at low temperature; this is almost identical to the solution
or gas-phase value. For comparison, a chemical shift of 146
ppm was observed for π complexes of propene on zeolite
HZSM-5. We carried out a number of experiments to establish
the environment of unreacted propene on SPA at low temper-
atures; for example, we observed no spinning sidebands and
little cross polarization efficiency. We conclude that propene
is not dissolved into the glassy phosphoric acid phase at low
temperature. At 273 K propene began to react with the catalyst
to form isopropyl phosphate. This reaction was complete at
298 K. The phosphate ester was very mobile at 298 K, but
when the sample was cooled to 173 K, we readily observed
spinning sidebands and a strong cross polarization response for
this species. Similar in situ experiments with propene-1-13C
We explored the loading dependence of the 13C isotropic shift
of acetone on SPA evacuated at 298 K (Figure 8) and found,
not surprisingly, that SPA has a distribution of apparent acid
strength, especially at higher loadings. 13C NMR measurements
were performed at 153 K to eliminate chemical exchange
between acetone in different coordination environments. At
lower loadings, there was a single resonance at ca. 228-229
ppm. At higher loadings, a second resonance appeared at ca.
222 ppm. Figure 8 suggests that acetone may form complexes
with different species present in the concentrated phosphoric
acid solution that is the acidic component of SPA.
(not shown) established that propene does not undergo 1, 3 13
C
label scrambling prior to formation of isopropyl phosphate, as
is the case for the reversible formation of surface-bound alkoxyl
species in zeolites. The irreversible nature of isopropyl
phosphate formation at these low temperatures also underscores
the stability of this species. The ester is indefinitely stable on
SPA at moderate temperatures, but it reacts at approximately
398 K to yield a complex mixture of products characteristic of
“conjunct polymerization”. The complex mixture of reaction
products generated is a result of the long reaction times in an
(51) Xu, T.; Munson, E. J.; Haw, J. F. J. Am. Chem. Soc. 1994, 116,
1962-1972.

8508 J. Am. Chem. Soc., Vol. 120, No. 33, 1998
Krawietz et al.
Ab Initio Calculations. We used theoretical methods to
model the complexation of either acetone or propene with some
of the species expected to be present in a concentrated
phosphoric acid solution. The acetone study was more detailed,
because we were especially interested in trying to understand
the large chemical shift of acetone on SPA. For the best
imaginable calculation of this type, one would like to explicitly
include several solvation shells of phosphoric and pyrophos-
phoric acid as well as long-range electrostatics, but this level
of treatment is not currently feasible. Instead, we do what is
most commonly done for calculations of adsorption on zeo-
lites: use computationally manageable clusters to approximate
the acid site and model interactions as if in the gas phase.
We considered the obvious cases of acetone complexed to
either one phosphoric acid or one pyrophosphoric acid; as
demonstrated below, we found two isomers of each. We also
considered acetone complexed to two phosphoric acid mol-
ecules. A less obvious choice was acetone complexed to
protonated phosphoric acid, H4PO4+. We were motivated to
consider the latter by the work of R. A. Munson, who carried
out a detailed study of ionic equilibria in phosphoric acid.52
Two equilibria were considered in that investigation, autopro-
tolysis
-
2H3PO4 a H4PO4+ + H2PO4
and self-dehydration
Figure 9. Complexes of acetone with one phosphoric acid molecule
via one hydrogen bond (1) and two hydrogen bonds (2) and the complex
of acetone with protonated phosphoric acid (3 and 4). Geometries were
optimized at the B3LYP/6-311+G(d,p) level of theory.
3H3PO4 a H3O+ + H4PO4+ + H2P2O7
2-
Munson used cryoscopic and EMF measurements to determined
the composition of 100% H3PO4 and obtained molal concentra-
The geometries of eight structures, modeling acetone on acidic
species in SPA, are reported in Figures 9 and 10, and Figure
11 reports geometries of the lower energy structures modeling
complexation of propene. Several additional structures were
also calculated to obtain deprotonation energies, the energies
of various thermodynamic reference states, or for chemical shift
calculations. These included H4PO4+, two isomers of H3PO4,
the phosphoric acid dimer, two isomers of pyrophosphoric acid,
the deprotonated states of the acids, acetone, protonated acetone,
and propene. Geometries of these latter species as well as some
higher energy complexes of acetone or propene not reported in
Figures 9-11 are collected in the Supporting Information.
Energies for the more important geometries, including thermal
corrections, are compiled in Table 1.
+
-
tions for the ionic species of mH4PO4 ) 0.54, mH2PO4
)
0.26, and mH3O+ ) mH2P2O7 ) 0.28. Since 100% H3PO4
is formally 10.2m in itself, the concentration of H4PO4+ reported
by Munson corresponds to 5.3% of the phosphate groups. Ionic
equilibria have not been quantified for phosphoric acid solutions
more concentrated than 100% (P2O5 > 72.4%), and we are
2-
+
unable to cite a literature reference specifying the H4PO4
concentration expected for solutions in the concentration range
of SPA. However, since SPA has a high concentration of
pyrophosphoric acid, the following equilibrium should also be
important.
2-
2H3PO4 + H4P2O7 a 2H4PO4+ + H2P2O7
The geometries in Figures 9 and 10 all show conventional
hydrogen bonding between hydroxyl groups on phosphoric acid
species and acetone. Structures 1 and 3 also suggest weak
C-H-O hydrogen bonding. Structures 1 and 2 (Figure 9) are
the two geometries we found for acetone hydrogen bonded to
a single molecule of phosphoric acid. The monodentate
geometry 1 was the most stable at -9.13 kcal/mol relative to
isolated acetone and phosphoric acid at the same level of theory,
and the bidentate structure 2 was less stable at -6.58 kcal/mol.
We searched for other stable states including, for example,
zwitterionic states of acetone and phosphoric acid, but these
spontaneously reverted to the geometries shown. We also found
two complexes with H4PO4+; 3 was the more stable, but the
bidentate structure 4 was only 5.60 kcal/mol higher. The
oxygen-oxygen distance in structure 3, 2.41 Å, indicates an
exceptionally strong hydrogen bond. Also, the proton is closer
to acetone than to phosphoric acid; this is the only species we
observed for which proton transfer to acetone could be claimed.
Partial proton transfer occurs although the proton affinity of
Thus, although we are uncertain of the concentration of H4-
PO4 in SPA, we are satisfied that it is sufficiently high to
warrant inclusion in the theoretical study.
+
In common with recent studies of adsorbates on zeolite
Brønsted sites53,54 we used density functional theory (DFT) to
determine the geometries of various complexes of acetone or
propene with phosphoric acids followed by ab initio chemical
shift calculations using the gauge-invariant atomic orbital
(GIAO) approach. All structures were calculated by using DFT
with the B3LYP exchange-correlation functional and the
6-311+G(d,p) basis set. This level of theory has, in our
experience, provided reliable models of hydrogen bonding and
acid-base chemistry at manageable computational expense.
(52) Munson, R. A. J. Phys. Chem. 1964, 68, 3374-3377.
(53) Nicholas, J. B.; Haw, J. F.; Beck, L. W.; Krawietz, T. R.; Ferguson,
D. B. J. Am. Chem. Soc. 1995, 117, 12350-12351.
(54) Beck, L. W.; Xu, T.; Nicholas, J. B.; Haw, J. F. J. Am. Chem. Soc.
1995, 117, 11594-11595.

Solid Phosphoric Acid Catalyst
J. Am. Chem. Soc., Vol. 120, No. 33, 1998 8509
Figure 11. Complexes of propene with either phosphoric acid (8 and
9) or protonated phosphoric acid (10) and the reaction product isopropyl
phosphate (11). Geometries were optimized at the B3LYP/6-311+G-
(d,p) level of theory.
Figure 10. Complexes of acetone with either two molecules of
phosphoric acid (5) or one of pyrophosphoric acid (6 and 7). Geometries
were optimized at the B3LYP/6-311+G(d,p) level of theory.
characterized by a hydrogen bond between a hydroxyl group
and the π cloud of the olefin. The optimized geometry of
isopropyl phosphate 11, known to be the reaction product
between propene and phosphoric acid, is included in Figure 11
for completeness.
isolated H3PO4 (198.0 kcal/mol at B3LYP/6-311+G(d,p)) is
slightly higher than for isolated acetone at the same level of
theory (195.5 kcal/mol).
Figure 10 reports the geometries we obtained for acetone with
either two phosphoric acid molecules, 5, or acetone with one
molecule of pyrophosphoric acid, 6 and 7. Species 5 is
characterized by a chelated hydrogen bond, and there is
precedent for such bonds in the literature.55 We also searched
for a geometry in which the phosphoric acid dimer formed a
conventional two-center hydrogen bond to acetone, but we were
unable to converge any such structure. We obtained structures
for two 1:1 complexes of acetone and pyrophosphoric acid;
calculations of 1:2 complexes were not attempted on the basis
of the large computational resources necessary to obtain the
smaller structures. 6 is the more stable species at -12.48 kcal/
mol relative to the isolated molecules, but 7 is similar at -11.57
kcal/mol. Pyrophosphoric acid hydrogen bonds to acetone more
strongly than does phosphoric acid; this is reflected in the
reaction enthalpies as well as by the internuclear coordinates,
and is also consistent with the relative gas-phase deprotonation
energies of the acids.
Chemical Shift Calculations. Chemical shift calculations
provide a bridge between experimental measurements and
theoretical geometries. We prefer the method of gauge-invariant
atomic orbitals (GIAO) and Ahlrichs’ basis sets for chemical
shift calculations. As was found previously by other work-
ers,56,57 treatment of electron correlation with the GIAO-MP2
method of Gauss is essential for accurate 13C chemical shifts
for carbenium ions and related charged species. Unfortunately,
GIAO-MP2 calculations rapidly become prohibitive with in-
creasing numbers of basis functions. We obtained MP2/tzp/dz
shifts for acetone, protonated acetone, propene, and complexes
1-4 and 8-11, but the numbers of basis functions or primitives
required for structures 5-7 exceeded the program limitations
for ACES II, and these could not be obtained at MP2. However,
we were able to calculate the 13C shifts at GIAO-RHF/tzp/dz
for all of 1-11. Calculated shifts are reported in Table 2 for
acetone complexes and Table 3 for propene complexes.
In common with our other recent work58 we find that RHF
shift calculations for the carbonyl carbon of acetone systemati-
cally overestimate the MP2 result. Previous work strongly
We found four stable geometries with very similar energies
for propene complexed with phosphoric acid. Figure 11 reports
the structures of the two lowest energy species, 8 and 9, while
the other two are reported in the Supporting Information. We
also found one complex between H4PO4+ and propene 10, which
is also reported in Figure 11. All five propene complexes were
(56) Xu, T.; Barich, D. H.; Torres, P. D.; Haw, J. F. J. Am. Chem. Soc.
1997, 119, 406-414.
(57) Xu, T.; Torres, P. D.; Barich, D. H.; Nicholas, J. B.; Haw, J. F. J.
Am. Chem. Soc. 1997, 119, 346-405.
(58) Barich, D. H.; Nicholas, J. B.; Xu, T.; Haw, J. F. Theoretical Studies
of the 13C Chemical Shift Tensor of Acetone Complexed with Brønsted
and Lewis Acids. In J. Am. Chem. Soc., submitted for publication.
(55) Frisch, M. J.; Pople, J. A.; Del Bene, J. E. J. Phys. Chem. 1985,
89, 3664-3668.

8510 J. Am. Chem. Soc., Vol. 120, No. 33, 1998
Krawietz et al.
Table 1. Electronic Energies, Zero-Point Energies, and Thermal Corrections to Energies of Adsorption Complexesa
species and
complexes
electronic
energyb
thermal
ZPEc
correctiond
total energye
Acet
-193.218179
-193.539655
-644.281722
-1288.601407
-1288.097171
-1212.106318
-643.753473
-1211.616124
-644.606086
-837.517854
-837.514107
-837.876308
-837.868218
-1481.840734
-1405.347238
-1405.346300
-117.945586
-118.243955
-762.234433
-762.234222
-762.573519
-762.250797
0.084058
0.096374
0.048162
0.098237
0.086860
0.075028
0.036827
0.063897
0.059277
0.134598
0.135259
0.144526
0.145489
0.186289
0.160708
0.161417
0.080089
0.088458
0.129756
0.129863
0.140045
0.133896
0.005425
0.005413
0.006313
0.013234
0.012554
0.009874
0.005658
0.009135
0.006331
0.012766
0.012415
0.012922
0.012797
0.018870
0.016537
0.016335
0.004073
0.004247
0.011975
0.011938
0.011841
0.010284
-193.128696
-193.437868
-644.227247
-1288.489936
-1287.997757
-1212.021416
-643.710988
-1211.543092
-644.540478
-837.370490
-837.366433
-837.718860
-837.709932
-1481.635575
-1405.169993
-1405.168548
-117.861424
-118.151250
-762.092702
-762.092421
-762.421633
-762.106617
Acet‚H+
H3PO4
(H3PO4)2
H3PO4‚H2PO4
H4P2O7
H2PO4
H3P2O7
-
-
-
+
H4PO4
H3PO4‚Acet (1)
H3PO4‚Acet (2)
H4PO4+‚Acet (3)
H4PO4+‚Acet (4)
(H3PO4)2‚Acet (5)
H4P2O7‚Acet (6)
H4P2O7‚Acet (7)
C3H6
C3H6‚H+
H3PO4‚C3H6 (8)
H3PO4‚C3H6 (9)
H4PO4+‚C3H6 (10)
i-(C3H7)H2PO4 (11)
a All energies reported are in hartrees. b Calculated at the B3LYP/6-311+G(d,p)//B3LYP/6-311+G(d,p) level of theory. c Zero-point energies
were obtained without scaling at 298.15 K and 1 atm at the B3LYP/6-31G(d)//B3LYP/6-31G(d) level of theory. d Thermal corrections were calculated
at 298.15 K and 1 atm at the B3LYP/6-31G(d)//B3LYP/6-31G(d) level of theory. e Total energy ) E0 + ZPE + thermal correction.
Table 2. Theoretical 13C Isotropic Chemical Shiftsa (ppm) for the
C-2 Carbon of Acetone for Acetone, Protonated Acetone, and
Acetone Complexes Formed with H3PO4, H4PO4+, (H3PO4)2, and
H4P2O7
Table 3. Theoretical 13C Isotropic Chemical Shiftsa (ppm) for
Propene, Propene Complexed with H3PO4 and H4PO4+, and
Isopropyl Phosphate
RHF-GIAO/tzp/dz//
MP2-GIAO/tzp/dz//
B3LYP/6-311+G(d,p) B3LYP/6-311+G(d,p)
RHF-GIAO/tzp/dz//
MP2-GIAO/tzp/dz//
B3LYP/6-311+G(d,p) B3LYP/6-311+G(d,p)
C3H6
C1
C2
C3
C1
C2
C3
C1
C2
C3
120.4
140.9
19.7
117.3
152.2
19.5
131.8
158.2
32.9
115.1
166.3
20.5
115.1
133.3
21.8
113.1
143.8
22.0
117.2
139.3
21.9
113.3
156.9
23.0
Acet
Acet‚H+
208.1
263.7
223.7
228.7
250.4
241.0
233.1
228.4
225.9
194.5
256.8
210.1
215.4
239.9
229.9
H3PO4‚C3H6 (8)
H3PO4‚C3H6 (9)
H3PO4‚Acet (1)
H3PO4‚Acet (2)
H4PO4+‚Acet (3)
H4PO4+‚Acet (4)
(H3PO4)2‚Acet (5)
H4P2O7‚Acet (6)
H4P2O7‚Acet (7)
H4PO4+‚C3H6 (10) C1
C2
C3
a Isotropic chemical shifts are reported relative to TMS calculated
at the same level of theory. The absolute shieldings (in ppm) of TMS
are 193.2 for RHF and 198.7 for MP2.
i-(C3H7)H2PO4 (11) C1
22.6
66.9
20.5
24.9
76.7
25.6
C2
C3
a Isotropic chemical shifts are reported relative to TMS calculated
at the same level of theory. The absolute shieldings (in ppm) of TMS
are 193.2 for RHF, and 198.7 for MP2.
suggests that MP2 shifts are reliable, even for “difficult” cases
such as carbenium ions,50,56,57 while RHF calculations are
reliable for trends if not exact values. Acetone complexed to a
single phosphoric acid yields MP2 shifts of 210.1 ppm for 1
and 215.4 ppm for 2; these values are well below acetone in
either SPA (up to ca. 231 ppm) or zeolite HZSM-5 (223 ppm).
of isopropyl phosphate, 76.7 ppm, is in good agreement with
the 80 ppm result in Figure 7.
+
The structures for acetone complexed to H4PO4 have MP2
Discussion
shifts of 239.9 and 229.9 ppm for 3 and 4, respectively. These
shifts bracket the largest experimental values for acetone on
SPA.
Solid phosphoric acid is more complex than phosphoric acid
supported on a silica source. Chemical reactions occur in the
preparation of SPA to form various silicon phosphate phases;
thus the true support material in SPA is a mixture of crystalline
reaction products. SPA does not contain Lewis sites, and the
Brønsted acid sites on this catalyst arise from a viscous solution
or glass (depending on temperature and composition) of
phosphoric acid and phosphoric acid oligomers. The average
chain length of these acids is strongly sensitive to preparation
and handling of the catalyst and this accounts for the water
sensitivity of SPA in commercial processes. The acidity of the
phosphoric acid phase is not enhanced by the crystalline
components.
Only RHF shift calculations were feasible for 5-7, and these
were interpreted relative to the RHF shifts for 1-4. Complex-
ation of acetone with two molecules of phosphoric acid, 3,
produces a slightly larger shift than complexation with one
molecule of acid, 2. Complexation with one molecule of
pyrophosphoric acid, 6 and 7, results in shifts little different
from complexation with one phosphoric acid.
The results in Table 3 predict that C-2 of propene should
show an appreciable downfield shift when dissolved in the acidic
phase of SPA, but recall that we observed a value of 134 ppm,
much closer to the gas-phase value. The calculated C-2 shift

Solid Phosphoric Acid Catalyst
J. Am. Chem. Soc., Vol. 120, No. 33, 1998 8511
The theoretical results in Figure 11 and Table 3 suggest that
propene is not dissolved in the acidic phase of SPA at low
temperatures, in agreement with our interpretation of in situ
studies such as those in Figure 7. Complexation of propene
with phosphoric acid or other, more acidic species in SPA should
result in appreciable downfield shifts of the C-2 resonance,
which are not observed. Propene is excluded from the glassy,
acidic phase of SPA at low temperatures, and when it does enter
SPA, at 273 K and above, it reacts to form isopropyl phosphate.
The in situ NMR identification of isopropyl phosphate supports
the classical mechanism for olefin oligomerization on SPA.25
Ipatief was able to isolate phosphate esters using similar reaction
conditions and showed that these produced oligomeric hydro-
carbons upon heating.59
calculated deprotonation energies of the clusters considered here
support the idea that SPA is a strong acid.
The geometries of the acetone complexes in Figures 9 and
10 show trends in the degree of proton transfer that generally
reflect the deprotonation energies of the acids, which are in turn
reflected in chemical shift trends. The MP2 shifts of the
complexes of acetone with H4PO4+, 239.9 ppm for 3 and 229.9
ppm for 4, bracket the largest shift we observed for acetone on
SPA, 231 ppm. We conclude that the acetone chemical shift
reliably ranks the relative acid strengths of SPA and zeolites;
SPA is the stronger acid.
On zeolites, propene is present either as free olefin, π
complexes with the acid site, or lower concentrations of
framework-bound alkoxyl species.29,49 The latter species is
analogous to the phosphate ester formed on SPA in that each is
a covalently bound complex of propene and the acid site. The
essential difference in the propene oligomerization activity of
the two catalysts is the extent to which this intermediate forms.
On the zeolite, the alkoxyl is reversibly formed as a minority
species, and other propene molecules are available to react with
it to form oligomers. On SPA, the reaction to form isopropyl
phosphate is quantitative at 298 K. At higher temperatures,
the ester is in equilibrium with olefin, and the position of this
equilibrium moderates the oligomerization activity.
SPA is widely believed to be a weaker acid than zeolites,
and we gave much thought to this comparison. The acid
strength of SPA is a straightforward problem with a rigorous
solution. The acid strengths of solutions and melts of phos-
phoric acid have been previously characterized over a very large
concentration range by using the Hammett acidity function. At
298 K H0 is -4.70 for 70% P2O5 and -6.96 for 85% P2O5.60
Like other mineral acids, the Hammett acidity of phosphoric
acid has a modest temperature dependence, becoming slightly
larger (less negative) at higher temperatures. At 298 K our SPA
samples have Hammett acidities in the range of -5 to -7.
Although the acid strength of SPA does not approach that of
100% H2SO4 (H0 ) -12), SPA is unquestionably a strong acid.
Unfortunately, the Hammett formalism cannot be applied to
zeolites with the same rigor as for phosphoric acid, but the
comparison of acid strengths can be made in other ways.
Experimental measurements of the deprotonation energies of
zeolites are in the range of 284-318 kcal/mol,61 and values
near 296 kcal/mol are commonly used in theoretical studies to
model isolated Brønsted sites.62 The theoretical energies in
Table 1 permit us to calculate the deprotonation energies of
some of the species we used to model SPA; smaller values
correspond to stronger acids. For a single phosphoric acid
molecule in the gas phase the deprotonation energy is fairly
large, 325.4 kcal/mol. However, the dimer of phosphoric acid
has a value of 310.3 kcal/mol, indicating a significantly stronger
acid than the monomer. Our calculations also show that a single
pyrophosphoric acid is a strong gas-phase acid at 301.6 kcal/
mol. We are unable to model a pyrophosphoric acid dimer,
but consideration of the phosphoric acid dimer suggests that
the deprotonation energy of the former would be well below
the 296 kcal/mol value used to model zeolites theoretically. Not
Conclusions
SPA is a supported liquid phase catalyst, a highly concen-
trated phosphoric acid solution on a support of crystalline silicon
phosphate reaction products. The acid function of SPA is purely
Brønsted in nature, and it is entirely due to the phosphoric acid
phasesthe crystalline component functions purely as a support.
The acid strength of SPA exceeds that of typical zeolite
catalysts; this is the reverse of the olefin oligomerization activity.
Propene oligomerization proceeds through analogous intermedi-
ates on the two catalysts, but the phosphate ester formed on
SPA is so stable that this reaction goes to completion and higher
temperatures are required to establish an equilibrium with free
olefin. The differential stabilities of the intermediates in propene
oligomerization on zeolites vs SPA highlight an important aspect
of catalysis: the more active catalysts avoid pathways with
exceptionally stable intermediates.
Acknowledgment. The NMR component of this work was
supported by the U.S. Department of Energy (DOE) Office of
Basic Energy Sciences (BES) (Grant No. DE-FG03-93ER14354)
and the theoretical component was supported by the National
Science Foundation (CHE-9528959).
+
surprisingly, H4PO4 is a particularly strong acid with a
deprotonation energy of 198.0 kcal/mol. While we are not able
to fully model a concentrated phosphoric acid solution, the
Supporting Information Available: Spectra of acetone
adsorbed on SPA; structures of neutral and protonated acetone,
propene, and phosphoric acid; structures of phosphoric acid
related species; structures of complexes of propene and phos-
phoric acid; T1 data for silicon phosphates; additional theoretical
energies (6 pages print/PDF). See any current masthead page
for ordering information and Web access instructions.
(59) Pines, H.; Ipatieff, V. N. Catalytic Reactions at High Temperatures
and Pressures; Macmillan: New York, 1936.
(60) Rochester, C. H. Acidity Functions; Academic Press: New York,
1970.
(61) Datka, J.; Boczar, M.; Rymarowics, P. J. Catal. 1988, 114, 368-
376.
(62) Brand, H. V.; Curtiss, L. A.; Iton, L. E. J. Phys. Chem. 1993, 97,
12773-12782.
JA9813461