POROUS SOLIDS OF , ALUMINUM PHOSPHATE, AND SILICON PHOSPHATE

K. B. BABB, D. A. LINDQUIST, S. S. ROOKE, W. E. YOUNG, and M. G. KLEVE'" The University of Arkansas at Little Rock, Department of Chemistry and *Department of Biology, Little Rock, Arkansas 72204

ABSTRACT

Anhydrous sol-gel condensation of [(CH3CH20bPO] with (BCI3), triethyl aluminum [(CH3CH2bAl] or silicon tetrachloride [SiCI 4] in organic solvents led to rigid gels. The pore fluid of the gels was removed under supercritical conditions in a pressurized vessel to form porous solids. The condensation chemistry prior to the gel point was monitored by solution lH, 13C, 31p, and 11B NMR. The materials were then calcined at progressively higher temperatures to produce high surface area . Nitrogen gas physisorption was used to determine the surface areas, total pore volume, and average pore radius of the products. FT -IR was used to determine functional groups in the materials. The microstructure was also examined by scanning electron microscopy.

INTRODUCTION

The orthophosphate (MP04) compounds of boron and aluminum may be described as covalent network solids of oxygen bridged alternating P04 and M04 tetrahedra [1]. These phosphates are consequently structurally isomorphous with one or more of the various forms of silica (Si02) and also share similar chemical and physical properties with those of silica. Silicon phosphate [Si3(P04)4] is also a covalent product, but is not isomorphous with any silica. Instead, it may be a cyclo-chain structure of eight-membered rings connected through spiro atoms of silicon with "pockets" of pure silicate and phosphate tetrahedra [2]. The formation of silica by sol-gel routes has been intensively studied for various applications such as coatings and formation of high surface area materials [3], but comparatively little has been written on the sol• gel preparation of covalent phosphates [4-7]. Phosphates of aluminum, boron, and silicon were prepared in this work. Since phosphates of acidic metal cations have useful solid acid catalytic properties, it was desirable to prepare them with a high surface area. Supercritical solvent removal was used to preserve the fine pore structure. We chose non aqueous synthetic routes because most organic solvents have a considerably lower critical point than water. A second rationale for anhydrous conditions is the difficulty in making stoichiometric phosphate compositions from aqueous solutions. In water solutions, one may obtain some metal oxide phase in addition to the desired phosphate due to competing hydrolysis reactions.

EXPERIMENTAL

All solvents were dried under a dry nitrogen atmosphere by distillation from P20S in the case of chlorobenzene, potassium carbonate for acetone, and sodium benzophenone ketyl for

211

Mat. Res. Soc. Symp. Proc. Vol. 371 °1995 Materials Research Society pentane. The triethyl phosphate was also freshly distilled and all gel syntheses were conducted under a dry nitrogen atmosphere using Schlenk techniques. The BP04 in this work was synthesized using the method of Gerrard and Griffey [3] by reaction of triethyl phosphate with boron trichloride to yield boron phosphate and ethyl chloride as a byproduct. To prepare the boron phosphate, 8 mL (46 mmol) of triethyl phosphate «CH3CH20)3PO) was dissolved in 18 mL of chlorobenzene. The solution was cooled in an ice bath and a flask containing 4 mL (46 mmol) of boron trichloride (BCI3) was mated to the triethyl phosphate solution flask to allow condensation of the BCl3 vapor into the stirring phosphate solution over a period of about 3 hr. During this time, the flasks were closed off from nitrogen purge so that no BCI3 was lost from the system. The resulting solution of adduct was then aged at 60 cC for 8 hr. During this time, the flask was periodically vented to allow for escape of the ethyl chloride byproduct. The gel point occurred, on average, 1 hr after beginning heating at 60 QC. The gel was then allowed to age in a sealed flask at room temperature for 2 days. The AlP04 was prepared by a novel method using triethyl aluminum instead of the chloride compound due to the low solubility of aluminum chloride in organic solvents. In a flask equipped with a water condenser, a neat mixture of 10 mL (73 mmol) of (CH3CH2})Al and 12.4 mL of (CH3CH20})PO (73 mmol) were heated to 175 cC in an oil bath on a hot plate stirrer overnight. This formed an oligomeric oil. The oligomer (2 mL) was then dissolved in 24 mL of acetone and the solution cooled in a salt ice bath (-40 CC). Anhydrous NH3 was then bubbled vigorously through the solution with a syringe needle. After a few minutes the mixture gelled with a concomitant evolution of gas. The gel was then allowed to stand in a sealed flask at room temperature for two days. The silicon phosphate gel was prepared by reaction of triethyl phosphate with silicon tetrachloride in chlorobenzene solvent, similar to the method of Borisov and Voronkov [2]. In a 250 mL flask equipped with a water condenser under nitrogen purge, 8.69 mL (CH3CH20)3PO (51.2 mmol) was added to 30 mL chlorobenzene and stirred. Next, 4.4 mL

SiCl4 (38.4mmol) was added and the resulting solution gelled after standing 36 hr at room temperature. Prior to gellation, the flask was periodically vented to allow escape of the ethyl chloride byproduct. The gels were allowed to age for 7 days prior to solvent exchange. The aged gels then were transferred to a soxhlet apparatus and the chlorobenzene or acetone exchanged for pentane by reflux overnight (15 hr). While in the soxhlet, the gels were kept in an open top glass container so that fresh refluxing pentane could condense and pour off the top of the gels without draining the pore liquid from the gels during each emptying cycle of the soxhlet. The solvent exchange through the gels could be monitored visually because the translucent gels became more opaque upon substitution of the pentane for chlorobenzene or acetone. The pentane solvent exchanged gels were transferred to a one lit er bomb (Parr® Instrument Co. Model 4500), heated over a period of one hour to 210 cC, and held at this temperature under supercriticai conditions for 0.5 hr. During heating, the vessel attained a maximum pressure of approximately 1000 psi; the critical point of pentane is 196°C and 480 psi. The supercritical pentane then was vented from the bomb while maintaining the temperature at 210 cC. Portions of the prepared materials were subsequently heated in air in a muffle furnace at temperatures of 200 cC, 500 cC and 800 cC for one hour at each temperature. This heating was done in order to determine the effects of progressive calcining on each material's microstructure.

212 Characterization methods

The early stages of the condensation chemistry prior to the gel point could be monitored by multinuclear solution NMR. In the case of boron phosphate, the freshly prepared chlorobenzene adduct solution was examined by 1H, 13 C, 31 P, and 11 B FT -NMR (Bruker Instruments 250 MHz FT-NMR). The NMR spectra were collected from fresh samples in vacuum sealed NMR tubes containing COCl3 lock solvent. The chemistry up to the gel point was studied by heating the samples in a 50°C water bath for 5 minute intervals prior to each spectrum being taken. Infrared spectra of aerogels were obtained as Nujol mulls or KBr pellets (Perkin Elmer Model 1600 FTIR). Surface characterization of solids by nitrogen physisorption, including multipoint Brunauer-Emmett-Teller (BET) surface area, total pore volume, and average pore radius were determined using a Quantachrome® Model NOV A 1000 instrument. Samples were outgassed at 200°C overnight under oil pump vacuum prior to physisorption measurements. Multipoint BET data were collected using a 4 point adsorption isotherm (PIP 0= 0.05, 0.15, 0.25, and 0.35). The total pore volume was determined based on the quantity of N2 adsorbed onto the sample at a relative pressure PIPo=0.99. The average pore radius in the gels was calculated by dividing twice the pore volume by the surface area. This method of calculating the value of the average pore radius assumes that the pores have a cylindrical geometry. Scanning electron micrographs were taken using gold coated samples at 20KV accelerating voltage (ISI Model DS-130).

RESULTS AND DISCUSSION

Polymerization of the boron phosphate gel proceeds by condensation of B-O-P bonds with evolution of chloroethane [4]. Freshly prepared solutions of the metal chlorides and triethyl phosphate showed sharp NMR resonances attributable to a 1: 1 adduct [(CH3CH20hP=O:BCh] [8]. lIB resonance of the adduct was narrow due to the tetrahedral coordination of boron. NMR spectra collected after heating the sample in 50 QC water for 5 minutes exhibited the emergence of peaks ascribed to ethyl chloride and a diminution of adduct peaks as condensation began. The 31 P spectrum of the heated sample also showed an additional resonance at -20 ppm pertaining to formation of [(CH3CH20hP(O)OBCl2] by elimination of one equivalent of ethyl chloride from the adduct. After an additional 2 days of aging at room temperature only ethyl chloride could be seen in the NMR spectra as the boron and phosphorus became incorporated in the cross-linked gel. The structure of the oligomer formed by heating the neat mixture of (CH3CH2hAl and (CH3CH20hPO is not yet understood. The NMR data [9] indicates that ethyl groups remain on both aluminum and phosphorus atoms. However, integration of the IH resonances indicated that there were more ethyl groups on aluminum than on phosphorus (4:3 ratio) instead of the expected 1: 1 ratio. The quantity of gas evolved in oligomer formation is consistently 0.33 moles for one mole of each reagent. The gas is a mixture of ethene and butane as confirmed by gas phase FTIR. Regardless of the precursor chemistry, the product has favorable properties as evidenced by the thermal stability of its microstructure. Table I is a compilation of the calculated specific surface area, total pore volume, and average pore radius for the heated materials. The microstructure becomes more coarse and the smaller pores are vanquished with increasing temperature. This is manifested as a decrease in total surface area and an increase in the average pore radius at higher temperatures due to sintering. There is no value for the average pore radius of the 800°C BP04 sample since it was already densified. The relatively constant total pore volume for the AlP04 reflects the stability

213 orthis material's microstructure. The porous solid s were exposed to ambient ai r pnor 10 surl'ace analysis The lower surface area or the 200 O( AlP04 aerogel renec1 S adsorbed moist ure in lhe micropores nOI removed by outgassing.

Table I: Effects or heat treatment 011 the specific surface 3rea, lotal pore volume, and average pore radius or the phosphate compositions (aI/ temperatures given are in CeLsiusl.

AIP04 BP04 S i JI POJ)4

r.kinillll " !\IN'" ... , .. ... 201,...... " Tt-nl .. .. Surf.... ·" Aru 1m2,!!.) 41tt J2t/ ", " " " " " 1'0", Vutllmt- ((n,l /J() ,., ,.. '.l 0. ' 1.0 H.M 11.41 11.11 1I. t? A .... u gt I',,", lA) " " 77 1411 - '"

Figure I: Scanning eleclron photom icrograph or SOO oC aerogel.

'" Scanning electron micrographs of BP04, AlP04 and Sh(P04)4 indicate a similar morphoio!:,'Y of spherical submicron aggregates. Figure I is a photomicrograph of BP04 c.alcined at 500 QC The calculated surface area of the visible structure, assuming an average particle diameter of 0.3 3 2 Ilm and a density of2.52 glcm was -20 m !g, indicating that some microporosity remains after 500 QC heating. Now that we have some basic understanding of these materials we are currently working on developing applications for these phosphates, including micro filtration membranes and cataly1ic membranes. In addition we hope to develop new types of crystalline catalysts by processing the solvent containing gels under similar conditions as those used to manufacture zeolites.

ACKNOWLEDGMENT

This work was funded by the Arkansas Science and Technology Authority and a faculty research grant from UALR. We would like to thank Mr. Alan Toland for his assistance in obtaining the NMR spectra.

References

I. 1.R Van Wazer, Pho,\phorlls and its Compound,' (Jnterscience Publications, New York, 1958).

2. S.N. Borisov, M.G. Voronkov, and E. Ya. Lukevits, Orgal1osilicoll Derivatives

3 C.l Brinker and G.W. Scherer, Sol-Gel Science (Academic Press, Inc., San Diego, 1990)

4. William Gerrard and P.F. Griffey, Chem. & Ind. 1959, 55; 1. Chem. Soc. ] 960, 3170; ihid. 1961,4095.

5. K.K. Kearby, U.S. Patent 3 342 750 (September 1967).

6. R Glernza, Y.O. Parent, and W.A.Walsh in ACS Symp. SeT. (Div. Petr. Chem.) 36,469- 477 (1991).

7. B. Rebenstorf, T. Lindblad, and S.L.T. Andersson, 1. Catal. 128,293-302 (1991).

8. NMR Characterization Data for (CH3CH20hP=O:BCI3 adduct in chlorobenzene solution with CDCI3 lock solvent. IH 0 4.57 multiplet (CH2 intensity 1.5), 1.58 multiplet (CH3 intensity] .0). lIB (BF3"Et20) 05.4 (h/2-B Hz). BC 0 63.6 (CH2), 0 15.0 (CH3). 31 P (85% H3P04) 0 -11.6.

9. NMR Characterization Data for aluminum phosphate oligomer in CDCl3. IH 04.05 multiplet [CH2-(P) intensity 1.0),0 1.35 triplet [CH3-(A1) intensity 1.8],00.96 triplet [CH3-(P) intensity 1.3],0 -0.23 quartet [CH2-(AI) intensity 1.3].

215