Chem. Mater. 1993,5, 1805-1813
1805
Thermochemical Study of the Stability of Frameworks in High Silica Zeolites Ivan Petrovic,*ftAlexandra Navrotsky,+ M a r k E. Davis,$ and Stacy I. Zones5 Princeton Materials Institute and Department of Geological and Geophysical Sciences, Princeton University, Princeton, New Jersey 08544;Department of Chemical Engineering, California Institute of Technology, Pasadena, California 91125;and Chevron Research and Technology, 100 Chevron Way, Richmond, California 94802 Received May 12,1993. Revised Manuscript Received October 4,199P The series of high-silicazeolites ZSM-5,ZSM-ll,ZSM-l2,SSZ-24,cubic and hexagonal faujasite has been studied to understand the relation between crystal structure and stability in open silicate frameworks. High-temperature solution calorimetry using lead borate (2Pb02-BzOS) solvent a t 977 K measured enthalpies of solution and transposed temperature drop calorimetry obtained heat contents at 977 K. Data have been compared with those for quartz and other “dense”, crystalline polymorphs of silica. The enthalpies of formation a t 298 K are as follows: ZSM-12, -(902.0 f 1.3); ZSM-5, -(902.5 f 1.3); ZSM-11, (902.5 f 1.4); SSZ-24, -(903.5 f 1.3); hexagonal faujasite, -(900.2 f 1.3); and cubic faujasite, -(897.1 f 1.2) kJ/mol. The value for quartz is -(910.70 f 1.00) kJ/mol. All zeolitic silicas are only 7-14 kJ/mol less stable in enthalpy than quartz. This implies an entropic or kinetic rather than a large stabilizing energetic role of the template in zeolite synthesis. The small variations in energy among these structures cannot be directly related either to the degree of ”openness” (framework density or molar volume) or to the mean Si-0-Si angle. Rather, the overall distribution of bond angles seems to dictate the energy of these structures, with Si-0-Si angles below 140° being the major destabilizing including those for “dense” crystalline factor. On the other hand, heat contents, (&77-&38), polymorphs, show a linear dependence on the framework density or molar volume.
Introduction Since the synthesis of the first high-silica zeolites--8 and ZSM-5-proprietary zeolites from Mobil Oil Co.,lv2 high-silica zeolites and molecular sieves have attracted attention and triggered major research efforts both in industry and academia. Numerous new, high- and puresilica structures have been reported3 and their number has been growingcontinuoualy.4 These materialstypically have increased hydrothermal stabilities, specificcatalytic propertiea including increased individual acid site strengths relative to aluminosilicate zeolites, and their sorptive characteristicsare hydrophobic or organophilicrather than hydrophilic.6 However, despite growing importance of zeolites, information on their stability and methods of synthesis is largely empirical, and fundamental relations between crystal chemical features and thermodynamics are understood rather vaguely. Further insight into the factorsthat govern relative stability of zeolites as a function of both lattice topology and composition is of relevance
* To whom correspondence should be addressed.
f Princeton University. t California Institute of Technology. Chevron Research and Technology. 8 Abstract published in Advance ACS Abstracts, November 15,1993. (1) Wadlinger, R. L.; Kerr, G . T.; Rosinski, E. J. U.S. Patent 3,308, 069, 1967. (2) Argauer, R. J.; Landolt, G. R. U.S. Patent 3,702,886, 1972. (3) Jacobs, P. A,; Martens, J. A. Synthesis of High Silica Aluminosilicate Zeolites; Elsevier: Amsterdam, 1987. (4) Higgins, J. B., von B a l l ” , R., Treacy, M. M. J., Eds.; Book of Abstracts ofthe9thZnt.Zeol. Conf.;Butterworth-Heinemann: Montreal, 1992. (5) Cheetham, A. K.,Day, P.,Eds.;Solid State Chemistry;University Press: Oxford, 1990; Vol. 2: Compounds.
to the physical chemistry of zeolite synthesis, to mineralogy, and to technological appli~ations.~J Over the past several years two types of theoretical approaches to the relative stability of zeolite structures have emerged. Quantum mechanical ab initioa10 and semiempirical techniquesl1J2have been used to calculate relative stabilities of small structural subunits, e.g., rings and cages. Another approach employs infinite lattice techniques using electrostatic (two-body) potentials and energy minimization by relaxing the structure. Both rigid ion13J4and shell models6J4have been used. Valence forcefield calculations have also been reported,16J6 and a combination of valence force fields and electrostaticforces has recently been used to simulate an isolated sodalite cage.17 Akporiaye and Price18 presented an analysis of (6) Ooms, G.; van Santan, R. A.; den Ouden, C. J. J.; Jackson, R. A.; Catlow, C. R. A. J. Phys. Chem. 1988,92,4462. (7) La Iglesia, A.; Aznar, A. J. Zeolites 1986, 6, 26. (8) van Beest, B. W. H.; Verbeek, J.; van Santen, R. A. Catal. Lett. 1988, 1, 147. (9) Fripiat, J. G.;Berger-Andr6, F.; Andrb, J.; Derouane,E. G.Zeolite8 1983,3, 306. (10) Sauer, J. J. Phys. Chem. 1987,91, 2315. (11) Ooms, G.; van Santen, R. A. Red. Trav. Chim. Pays-Bas. 1987, 106,69. (12) Mortier, W.J.; Geerlings, P.; van Alsendy, C.; Figeys, H. P. J. Phys. Chem. 1979,83,3257. (13) Ooms, G.; van Santen, R. A.; Jackson, R. A,; Catlow, C. R. A. Stud. Surf. Sci. Catal. 1988, 37, 317. (14) de Man, A. J. M.; van Beest, B. W. H.; Leslie, M.; van Santen, R. A. J. Phys. Chem. 1990,94,2524. (15) Blackwell, C. S. J. Phys. Chem. 1979,83,3251. (16) Blackwell, C. S.J. Phys. Chem. 1979,83,3267. (17) Mabilia, M.; Pearlstein, R. A.; Hopfinger, A. J. J.Am. Chem. SOC. 1987,109,7960. (18) Akporiaye, D , E.; Price, G . D. Zeolites 1989, 9, 321.
0897-4756/93/2805-1805$04.00/00 1993 American Chemical Society
1806 Chem. Mater., Vol. 5, No. 12, 1993 zeolite frameworks on the basis of their calculated lattice energies and coordination sequences (CSQ). An empirical method for estimating the standard Gibbs free energies of formation of zeolites has been r e p ~ r t e d . ~ It is based on the assumption that the free energy of formation can be written as a s u m of contributions of the individual oxide components. Thus, it neglects structure and would give all frameworks of a given composition the same free energy. Vieillardlg presented a computational study of the relations between the crystal structure and thermodynamic properties of 11 silica phases including silicalite (puresilica ZSM-5) and derived their enthalpies and Gibbs free energies of formation. Only a few experimental thermochemical studies on pure-silica zeolites have been reported. Silicalite has been studied, and a comprehensive thermodynamic data set, including relative stability with respect to other silica polymorphs, has been presented.20 Standard enthalpies of formation of the MFI type zeolites and their precursors with different templates were investigated by low temperature HF calorimetry.21 Zeolite energetics are of interest from a fundamental point of view. It is clear that the small rings in silicates such as quartz or cordierite are an important and integral part of the crystal structure, and that the energy of that structure reflects the distribution of bond lengths and angles within it. It is equally obvious that for a macroporous material (e.g., a ceramic with large pores and less than theoretical density), the pores form a second phase which affects the energy only through surface or interface effects. At what, generally small, size scale does the transition from the former to the latter regime take place? Microporous crystalline materials, with bulk densities between 80 and 50% those of their nonporous counterparts, offer and ideal testing ground for this question. Furthermore, what we learn from crystalline materials, whose pore distribution is well constrained by their periodic structures, will be useful in understanding the behavior of amorphous microporous materials such as gels and the synthesis of both crystalline and amorphous framework materials from gels and by templating. For these reasons, and because of the technical importance of high-silica zeolitesto catalysis, we have undertaken a calorimetric study of the zeolitic silica polymorphs as a first step in a systematic investigation of the relation of energetics and structure in microporous materials. Using high-temperature solution calorimetry with molten lead borate as solvent, we have determined the enthalpies of formation of a series of six high-silica zeolite specimens-high-silica samples of ZSM-12,ZSM-5,ZSM-l1, and 882-24, and dealuminated samples of cubic (FAU) and hexagonal faujasite (EMT) with Si/Al ratios of 360 and 11.4, respectively. This paper reports these results and discusses them in terms of structural variation, porosity, and implications for the stability of various zeolitic cages and the role of templates in zeolite synthesis.
Petrouic et al. TPABr.4Si02.4NW-8H20, where TPABr is tetrapropylammonium bromide. The mixture was heated at 368 K and autogenouspressure in a Teflon-lied autoclave for 8 days. ZSM11 was synthesized from a reaction mixture of composition 5.75TBAOH.2.8Na~0.96SiOp880Hz0, where TBAOH is tetrabutylammonium hydroxide. This mixture was aged at room temperature with stirring for 1day and subsequentlyheated at 423 K and autogenous pressure in a Teflon-lined autoclave for 1day. A specimen of ZSM-12 was crystallized from a reaction mixture of composition O.~~R(OH)Z.~.~KF’.S~OZ.~OH~~, where R represents 4,4’-trimethylene-l,l-dimethyldipiperidine.The mixture was first aged at room temperature with stirring for 1 day and then heated at 423K and autogenous pressure in a Teflonlined autoclave for 5 days. In all cases the product was recovered by filtration and subsequentlycalcined at 823 K for 4 h. In the synthesis of all-silica SSZ-24 the NJVJV-trimethyl-1-adamantammonium hydroxide was first made by permethylation of 1-aminoadamantane (Aldrich Chemical) using the synthetic procedure described in ref 22, which produced the iodide salt. The hydroxide saltwas prepared by anion exchangeusingBioRad AGl-X8 hydroxide resin. The zeolite was then prepared by the synthesis described by van Nordstrand et al.m The recovered sample was calcined at 723 K for 4 h. Hexagonal faujasite was synthesized using procedure described previously by Annen et The as-synthesized materialhadaSi/Alratioof -3.5, which was subsequently elevated by dealumination following the procedure used by Li et al.s The material was then calcined at 773 K for 4 h. The high-silica faujasite, Tosoh 390 HUA, is a commercial material from the Toeoh Chemical Co., Japan, and has been kindly provided by Dr. John Cook of Tosoh USA. Prior to the calorimetric measurements, all samples were characterized in termsof both chemical compoeitionand physical properties. The following techniques have been employed. Chemical analysis for alkali metals (Na, K), alkaline earths (Ca, Mg), and aluminum was carried out by atomic absorption spectroscopy (AAS) using a Perkin-Elmer AAS-603 and inductively coupled plasma emission spectroscopy (ICPES) using a Perkin-Elmer ICP-6OOO instrument. Electron microprobe analysis using the JEOL JXA-8600 Superprobe at Rutgers University was performed to determine the Si02 content. The specimen of hexagonal faujasite was also analyzed by Galbraith Laboratories, Inc., Knoxville, TN. X-ray powder diffractionpatterns were collected on a Scintag PAD V automated diffractometer equipped with a solid state detector and utilizing monochromatic Cu Ka radiation. For qualitativeidentificationof the phases present, the patterns were taken from 5 to 65’ 28. Thermogravimetricanalysis (TGA)instaticair anddifferential scanning calorimetry(DSC) in static air, flowing argon, and dry air were performed using Setaram TG-DSC 111 apparatus. Typically, approximately 40-50 mg of the sample was used at a heating rate of 10K/min for TGA and 5 K/min for DSC. In both cases the samples were run from 298 to 978 K. Calorimetry. The apparent thermal stability of all samples at the operating temperature of the calorimeter (977 K) was examined before calorimetric measurements as follows. The X-ray powder diffraction pattern was taken from the sample aa received. Then a portion of the zeolite was loaded into the calorimeter and equilibrated (approximately 8-10 h) under calorimetric conditions (including the presence of lead borate solvent). After the specimen waa pulled out, the X-ray pattern was collected again. Comparison of the X-ray patterns showed whether the structure persisted at a given temperature. All thermochemical measurements were performed using a Tian-Calvet twin microcalorimeter described in detail elsewhere,%in a staticair atmosphere. High-temperatureoxide melt solution calorimetry with lead borate (2PbO.BzOs) solvent at
Experimental Section
Sample Preparation and Characterization. A sample of ZSMd was prepared using a reaction mixture of composition (19) Vieillard, P. Bull. Mineral. 1986, 109, 219. (20) Johnson, G. K.; Taker, I. R.; Howell,D. A.; Smith, J. V. J.Chem. Thermodyn. 1987,19,617. (21) Patarin, J.; Soulard, M.; Kessler, H.;Guth, J. L.; Diot, M. Thermochim. Acta 1989,146,21.
(22) Chen, F. C. M.; Benoiton, N. L. Con. J. Chem. 1976,54,3310. (23) van Nordstrand, R. A.; Santilli, D. S.; Zones, 5.I. Perspectiues
in Molecular Sieve Science; Flank, W . H.,Whyte Jr., T. E.,Eds.; ACS Symp. Ser. 368; American Chemical Society: Waehington, DC, 1988; p 236. (24) Annen,M. J.;Young,D.;Arhancet, J.P.;Davis,M.E.;Schramm, S. Zeolites 1991, 11, 98. (25) Li, H.X.;Annen, M. J.; Chen, C. Y.; Arhaucet, J. P.; Davis, M. E. J. Mater. Chem. 1991,1, 79. ( 2 6 ) Navrotsky, A . P h y s . C h e m . M i n e r . 1977, 2 , 89.
Chem. Mater., Vol. 5, No. 12, 1993 1807
Frameworks in High Silica Zeolites
Table I. Physicochemical Characterization of ZSM-12, ZSM-6, ZSM-11, 882-24, Hexagonal Faujasite, and Cubic Faujasite zeolite ZSM-12 ZSM-5 ZSM-11 982-24 EMT FAU
SiOz(wt%) 95.42 97.48 94.09 97.56 86.75 95.73
Al(wt%) 0.043 0.011 3.542 0.119
Na(wt%) 0.157 0.071 0.247 0.234 2.152 0.013
K(wt%)
0.229
Ca(wt%) 0.058 0.068 0.076 0.013 0.239 0.005
Mg(wt%) 0.016 0.012 0.024 0.005 0.032 0.001
TGAwtloss(%) 3.50 2.32 4.61 2.07 2.58 3.97
Table 11. Framework Densities (FD), Molar Volumes, and Results of the Solution,. Drop Solution,b and Transposed Temperature Drop. Calorimetry Measurements, Together with Calculated Relative Instabilitiesb of Zeolitic Silicase zeolite ZSM-12 ZSM-5 ZSM-11 SSZ-24
EMT
FD (I"/ lo00 A3) 19.4 17.9 17.7 17.5 12.9 12.7
mol vol (cms/mol) 31.05 33.65 34.02 34.42 46.51 47.43
H977
-H z ~ e
(kJ/mol) 42.27 f 0.31 41.77 f 0.21 41.80 f 0.11 41.63 f 0.34 42.52 f 0.54 40.66 f 0.22
AHW1,977d (kJ/mol) -10.43 f 0.53 -9.47 f 0.38 N/A N/A -12.49 f 0.77 -14.07 f 0.26
~ ~ 1 , 9 7 1 ~ ~trm,Zs$
(kJ/mol) -10.64 f 1.14 -9.40 f 1.68 -9.54 f 1.01 -8.30 f 0.78 -12.67 f 0.55 -13.51 f 0.68
(kJ/mol) 8.66 f 0.64 8.19 f 0.47 N/A N/A 10.41 f 0.96 13.90 f 0.39
AHtraM,d
AHtrmd
(kJ/mol) 8.86 f 1.19 8.12 f 1.70 8.24 f 1.03 7.16 f 0.87 10.59 f 0.79 13.34 f 0.74
(kJ/mol) 8.7 f 0.8 8.2 f 0.8 8.2 f 1.0 7.2 f 0.9 10.5 f 0.9 13.6 f 0.7
FAU a Uncertainties are reported as two standard deviations of the mean. Uncertainities are calculated by propagation of errors. Note that for all samples including FAU and EMT the molecular weight has been taken as 60.0843 (pure SiOz) in all calculations. Obtained by solution calorimetry. e Measured by combination of drop solution and transposed temperature drop calorimetry of small pellets. f Enthalpies calculated using data.d 8 Enthalpies calculated using data.a I/ Recommended value, consistent with both f and g.
*
977 K was employed to measure the enthalpies of solution, and transposed temperature drop calorimetry was used to obtain the heat contents ( H 8 7 i - H ~ ) . Enthalpies of solution at 977 K have been measured by two independent paths: the first one involves solution calorimetry, which measures directly the enthalpy of solution at the calorimeter's temperature; the second one uses a combination of drop solution and transposed temperature drop techniques. In the solution experiments the sample size was approximately 1015 mg; for the drop cycle pellets of about 5 mg were used. Heat contents at 977 K were obtained using the transposed temperature drop technique. In this experiment, the sample either as a loose powder or as a pellet, wrapped in a platinum capsule, was dropped from room temperature to the calorimeter's temperature. The zeolite specimen typically contributed about 75% to the total measured heat effect, the platinum about 25%.
Results Characterization. Chemical analysis (see Table I) shows that in the first three samples-ZSM-5, ZSM-11 and ZSM-12-which were prepared from aluminum-free reaction mixtures, the Si02 content determined by the electron microprobe analysis is 94-97.5 % , and silicon accounts for almost all the metal content. The contents of alkali metals, alkaline-earth metals, and aluminum determined by AAS and ICPES are less than 0.25% or below detection limits. This confirms that these samples are essentially pure silica, with (see below) some H2O present. The SSZ-24specimen, also synthesized from Alfree reaction mixture, has been analyzed only for alkali metals, alkaline earth metals, and aluminum. The Si02 content has been back-calculated. This material is also essentially pure-silica in composition. In the synthesis of high-silica cubic and hexagonal faujasite, an aluminosilicate material has to be crystallized first and the product dealuminated. Analysis of cubic faujasite (Tosoh) shows very low content of impurities, and the Si/Al ratio, using calculated Si02 content, is about 360. For the purposes of this study, then, this specimen will be considered as pure-silica material. For EMT, the Si/A1 ratio of 11.4 was determined by Galbraith Laboratories. X-ray powder diffraction patterns of all specimens exhibited sharp peaks with no broadening, implying wellcrystallized and relatively large particles (>1pm). This is important from the calorimetric point of view, because
for particle sizes larger than approximately 1pm, heats of solution are generally independent of grain size. Thermogravimetric analysis showed weight losses of 1.54.6% (see Table I), with most of the weight loss occurring by 473 K. The exception were ZSM-11 and the Tosoh sample, for which the weight loss occurred to approximately 823 and 978 K, respectively. This weight loss is attributed to HzO and suggests that the water content in these samples may be higher than in the other four samples. However, the water contents in all samples are far below the values expected from micropore filling indicating that these materials are hydrophobic. DSC analysis in air showed in all cases large endothermic peaks at about 353 K, but when analyses were repeated in dried air and nitrogen, no peak was present. The endothermic peak had essentially the same size and position when the sample was cooled in air and the next run done after at least 12 h. This strongly suggests that the TGA weight loss is mostly due to the reversible adsorption of atmospheric moisture. Since pure-silica zeolites have hydrophobic character (see above), the effect is presumably due to physically adsorbed moisture. All structures except for ZSM-11 and SSZ-24 persist at 977 K for a time period needed for calorimetric equilibration prior to dissolution. The change in X-ray patterns of both ZSM-11and SSZ-24samples kept overnight in the calorimeter is obvious: the original pattern is degraded and peaks matching cristobalite developed. Thus, these two samples were studied only by drop-solution and transposed temperature drop calorimetry. Enthalpies of Transformation (Quartz Zeolite) and Enthalpies of Formation. Calorimetric data and zeolite at calculated enthalpies of transition, quartz 298 K obtained via two thermochemical cycles, are summarized in Table 11. In both cycles, measured enthalpies of solution have been corrected for the weight loss determined by TGA. The following two cycles have been used: For the transition quartz silica zeolite at 298 K the reaction scheme can be written as
-
-
SiO, (quartz, 298 K)
-
Si02 (zeolite, 298 K)
The first cycle then can be written as
(1)
Petrouic et al.
1808 Chem. Mater., Vol. 5, No. 12, 1993
SiO, (quartz, 298 K)
-
SiO, (quartz, 977 K) q 9 7 7
-
SiO, (zeolite, 977 K)
SiO, (soln,977 K)
(quartz) (2)
-
SiO, (zeolite, 298 K) -AHc,977 (zeolite) (4)
SiO, (zeolite, 977 K)
SiO, (quartz, 298 K)
-AHml 077 (zeolite) (5) SiO, (zeolite, 298 K) ~ t r a n s , 2 9 8(6)
where
AHtrans,m8 = A&,l,977(quartz) - LVl,01,g77(zeolite) + (MH7,(quartz) - ~c,g77(zeolite)) (7) Symbols AHc,977and “1,977 represent heat content of the dried sample and enthalpy of solution at 977 K, respectively. In this cycle, enthalpies of solution (eq 5) were measured directly by solution calorimetry. For the second cycle, enthalpies of solution of zeolites at 977 K were obtained by the combination of drop solution and transposed temperature drop calorimetry of small pellets. The rest of the cycle is essentially the same, so instead of eq 7 we can write SiO, (zeolite, 977 K)
SiO, (zeolite, 298 K)
-
Energy released in eliminating lo00 m2 of surface area measured by low temperature calorimetry is 0.26 kJ/mol,m so energetic differences between individual samples due to the specific surface area variation can be neglected. Johnson et aL20reported a value of 5.50 f 1.31 kJ/mol for the reaction quartz silicalite at 298 K. Though the difference between this and value obtained in our study, 8.2 f 0.8 kJ/mol for high-silica ZSM-5, is not very large, the source of this discrepancy needs to be identified, especially since the differences between different structures observed in this study are similarly small. Both materials should be the same both compositionally and structurally, and both techniques have proved reliable in other cases. One possible source of this discrepancy might be uncertainty in energy of combustion in fluorine measurements for silicalite. As noted by the authors, despite using silicon powder as an ignition aid, the reaction did not go to completion. de Man et al.14 reported the following calculated data on the energy difference between fully dealuminated cubic faujasite and a-quartz: the rigid ion model gave 44 kJ/ mol SiOZ, and the shell model yielded 19.5 kJ/mol SiOz. While the former value seems to significantly overestimate the real energetic difference between these two structures, the latter is of similar magnitude as our experimental value of 13.6 f 0.7 kJ/mol SO,. From the measured enthalpies of transformation and the known enthalpy of formation of quartz:O standard enthalpies of formation of all zeolitic silicas can be evaluated. The formation of zeolite can be written as
SiO, (soln, 977 K) hHd.&,1,977
Si@,298 K)+ O,(g, 298 K)
(zeolite) (9)
-
Si(s, 298 K) + O,(g, 298 K)
from which AHao1,977
SiO,(quartz, 298 K)
(zeolite) = m d . ~ 1 , 9 7 7(zeolite)
-
SiO,(zeolite, 298 K) (11) and the thermochemical cycle then takes the following form:
SiO, (zeolite, 298 K)
-
SiO,(zeolite, 298 K)
- AHt.Lt.td,977 (zeolite)
(10) is then cal-
and enthalpy of transformation, AHtrana,298, culated using eq 7. In this second cycle AHt.t.d,gp’]denotes enthalpy of transposed temperature drop to 977 K of the initial undried sample, which is different from the heat content of the dried sample, and A.Hd.w1,977refers to the enthalpy of drop solution at 977 K. For quartz, the following values have been used in all calculations: -3.51 f 0.18 kJ/mol for the enthalpy of solution,27 measured previously in this laboratory, and 44.00 kJ/mol for the heat content,28both at 977 K. Table I1 shows that the enthalpy of transformation obtained by the two cycles is the same, within experimental error, for each structure. This is important in two respects. First, it shows that the transposed temperature drop cycle can be used reliably for samples such as ZSM-11 and SSZ24, which do not persist metastably at calorimetric temperature. Second, it implies that the volatiles (mainly HzO) responsible for weight loss have been properly accounted for. (27) Akaogi, M.; Navrotaky, A. Phys. Earth Plan. Int. 1984,36,124. (28) Richet, P.; Bottinga, Y.; Denielou, L.; Petitet, J. P.; Tequi, C. Geochim. Cosmochim. Acta 1982,46,2639.
SiO,(quartz, 298 K) AHOf(WartZ)(12)
Si@,298 K) + O,(g, 298 K)
-
w”
(6)
SiO,(zeolite, 298 K) AHof(zeolite) (13)
from which AHof(zeolite) = AHof(quartz) + AHt,m,zs8 (14) For all samples recommended enthalpies of transformation, in the case of ZSM-12, ZSM-5, EMT,and FAU consistent with both cycles (see Table 11),have been used in these calculations. Calculated enthalpies of formation are summarized in Table 111. For the hexagonal faujasite structure, the value reported in this study represents an initial estimate because of the nonnegligible aluminum content of the sample. We did not attempt to correct the calorimetric data for the effect of aluminum, since there is no unique way of doing so because the A1 may be charge balanced by H,Na, or other cations, and the effect of these charge-coupled substitutions on energetics is not known. Work on a purer sample is needed. (29) Brunauer,S.; Kantro, D. L.; Weise, C. H. Can. J. Chem. 1966,34, 1483. (30) CODATA Task Group, CODATA recommended key values for thermodynamics, 1976; J . C h e m . T h e r m o d y n . 1976, 8 , 603.
Frameworks in High Silica Zeolites
Chem. Mater., Vol. 5, No. 12,1993 1809
Table 111. Comparison of Standard Molar Enthalpies of Formation of Zeolitic Silicas Obtained in This Study and Reported Elsewhere, with Those of Dense Silica Polymorphs enthalpy of formation (kJ/mol) dense SiOdzeolite quartz'
cristobalite2 tridymitea ZSM-12 ZSM-5/silicalite ZSM-11 982-24 EMT
FAU
'[301, 2[281,9[311
this work
Johnson et d.20
Pattarin et al.2l
Vieillard's
-905.20 k 0.84
-908.5 k 1.7
-900.49 k 0.28
-910.70 f 1.00 -907.86 -907.49 -902.0 -902.5 -902.5 -903.5 -900.2 -897.1
&
1.3
k 1.3 f 1.4 & 1.3 & 1.3 & 1.2
The value obtained for ZSM-5in this study is less negative or more destabilizing than values reported previously for silicalite.20*21 The first discrepancy has been already discussed above; the other value of -908.5 kJ/mol obtained by low-temperature acid calorimetry seems to be too negative, especially with respect to corresponding values of cristobalite and tridymite (see Table 111). Vieillardlg obtained AHofof silicalite of -900.49 kJ/mol from parametrized calculations based on the relations between crystallographic parameters (i.e., mean Si-0 bond length, Si-0-Si bond angle), refractive index, and molar volumes. Though in the right general range, it corresponds more closely with our value for hexagonal faujasite. Despite the uncertainties described above, it is clear that all the zeolitic silicas studied are very similar in enthalpy (and energy), namely, 7-14 kJ/mol less stable than quartz. The difference between the densest and least dense framework is only 4.9 kJ/mol, and the small differences in enthalpies of the silica zeolites show, at most, a weak dependence on molar volume, framework density or the average Si-0-Si angle (see Figure 1). Heat Contents. Figure 2 and Table I1 show the heat content, H w -Hm, ~ for all samples measured by transposed temperature drop calorimetry on dried specimens. Data for quartz and cristobalite are taken from Richet et a1.28 and for tridymite from Robie et al.31 A linear decrease in relative enthalpies, Hw, -H m , with decreassingframework density is seen (see Figure 2a). Hexagonal faujasite, which is not a pure-silica material, deviates slightly from this trend. Its heat content lies approximately 1.8 kJ/mol higher than the line for pure-silica materials. This increase can be explained by the presence of aluminum and nonframework cations in the structure, leading to more atoms per unit cell and a larger number of vibrational degrees of freedom. When heat contents are plottedversus molar volume, the opposite trend, i.e., a decrease in heat contents with increasing molar volume, is seen (see Figure 2b). However, a small deviation from linearity caused by the inverse relation between the molar volume and framework density (mol vol = F ( l / F D ) )can be observed. The solid lines in both figures represent linear least squares fit to the data (for the reasons mentioned above, the hexagonal faujasite was not included).
E w
-30
40
45
50
ZSM-12 ZSM-11
\.
I
Framework denelty (T/1000A3 )
1
N c .
a
U
2
10
40
Relation of Structure and Energetics
(31) Robie, R. A.; Hemingway, B. S.; Fisher, J. R. Geol. Survey BulZ. 1452,1979.
35
Molar volume (cc/mol)
l
The energetics of the pure silica zeolites may be viewed in comparison to those of dense silica polymorphs, which include stishovite with Si in octahedral coordination, and the *dense" tetrahedral frameworks of coesite (a highpressure polymorph), quartz, tridymite and cristobalite
4
EMT
_
.
.
_
,
145
_
.
_
_
,
_
150
.
.
.
.
155
Average SI-0-SI angle (")
Figure 1. Enthalpies of transition of zeolitic silicas as a function of (a) molar volume, (b) framework density, and (c) average Si&Si angle.
(high-temperature forms), and silica glass. Quartz is the thermodynamically stable form at 298 K and must lie at a minimum of any curve relating energy (enthalpy) and molar volume (see Figure 3). This curve rises steeply to
Petrouic et al.
1810 Chem. Mater., Vol. 5, No. 12, 1993
J
c
3
411
For molar volumes greater than that of quartz (decreasing density) the trend seen is quite different. The enthalpy increases only slowly with increasing molar volume (see Figure 3). Indeed, comparing FAU with a molar volume of 47.43 cm3/mol to quartz, with a molar volume of 22.69 cm3/mol, the difference is only 13.6 kJ/ mol; thus, a more than 2-fold expansion of the structure raises its energy by an amount only 1.5 times the enthalpy of vitrification. It is clear that framework silica structures are energetically very tolerant of changes in the topological linkages of tetrahedra to form rings and cages of different sizes. Returning to the question of the role of pores posed in the introduction, we suggest that in zeolitic silicas the large cavities are beginning to act as “external” pores, with increase in their volume fraction (decrease in framework density) affecting the energy in only a minor way. Thus, the region 5-8-A diameter pores may mark the transition, at least in terms of energetics, from a uniform structure to a heterogeneous one in which the interior of the pores is not strongly interacting with the framework, and is beginning to act like a second phase. Whether this is true in microporous materials in general, remains to be determined. The existence of nearly 90 distinct framework topologies33is in general accounted for by the flexibility of shared oxygen linkages or T-0-T angles (T = tetrahedrally coordinated framework cation). These can accommodate angles from -130’ to -180°, whereas the individual tetrahedra (0-T-0 angles and T-0 bond lengths) are deformed to a lesser extent.34 Several structural studies of pure “dense” crystalline silica polymorphs at high pressures36 and at low (10 K) and elevated t e m p e r a t u r e ~ have ~ l ~ ~shown virtually no variation of 0-Si-0 angles or Si-0 bond lengths, but changes in Si-0-Si angles with both pressure and temperature were more pronounced. Thus, it is reasonable that more open zeoliticsilica structures would differ mainly in Si-0-Si angles, neglecting any structural differences due to the presence of hydroxyl groups. For almost all synthetic zeolites studied to date, only powder diffraction data have been available for structure determinations.38 The only silica material with a singlecrystal structure determination appears to be ZSM-5.39 For the purposes of this study, the following (to our knowledge most recent and reliable) structure determinations have been considered: ZSM-5,39ZSM-11, and ZSM-12,4°>41SSZ-24,42FAU,43 and EMT.44 The data for monoclinic (calcined) ZSM-5 support the premise stated above. The Si-0 bond length range 1.582-
4 15 20 25 30
40 10
Framework density (TI1000A’ )
tt
-
2
44-1
0
‘ t -1
-
40 4 20
30
50
40
Molar volume (cclmol)
-
Figure 2. Heat contents (Hw7 - Hm) versus (a) framework density (FD) and (b) molar volume of zeolitic silicas and “dense” silica polymorphs (q = quartz, cr = cristobalite, tr = tridymite).
Y,
5
60 50
30
10
20
30
Molar Volume
40
50
(cclmol)
Figure 3. Enthalpy of dense (stishovite, coesite, quartz, tridymite, cristobalite, and silica glass) and zeolitic silicas relative to quartz as a function of molar volume. The curve is a suggestion of a smooth trend consistent with the data and with the thermodynamically required minimum at quartz. stishovite, with octahedral Si and a molar volume of 14.01 cm3/mol. Any hypothetical tetrahedrally coordinated structure with equally high density would be metastable with respect to stishovite and even higher in energy. Thus it is clear that energy and enthalpy rise sharply with decreasing molar volume. A similar trend is seen for amorphous silicas densified by pressure or radiation damage.32 (32) Navrotaky, A,; Wang, Y.; Liebermann, R. C. EOS tram.; AGU fall mtg. abst. suppl.; 1992, 73, 581.
(33) Meier, W. M.; Olson, D. H. Atlas of Zeolite Structure Types, 3rd ed.; Butterworth-Heinemann: London, 1992. (34) Liebau, F. Structural Chemistry of Silicates; Springer-Verlag: Berlin, 1985; p 16. (35) Levien,L.;Prewitt, Ch. T.;Weidner, D. J. Am. Mineral. 1980,66, 920. (36) Wright, A. F.; Lehman, M. S. J. Solid State Chem. 1981,36,371. (37) Pluth, J. J.; Smith, J. V.; Faber, J., Jr. J. Appl. Phys. 1985,57, 1045. (38) Baerlocher, Ch. Zeolites 1986,6, 325. (39) vanKoningsveld,H.;Jansen, J. C.;vanBekkum,H. Zeolites 1990, 10, 235. (40) Fyfe, C. A.; Gies, H.; Kokotailo, G . T.; Paeztor, C.; Strobl, H.; Cox, D. E. J. Am. Chem. SOC.1989,111,2470. (41) Fyfe, C. A.; Gies, H.; Kokotailo, G. T.; Marler, B.; Cox, D. E. J. Phys. Chem. 1990,94,3718. (42) Bialek, R.; Meier, W. M.; Davis, M.; Annen, M. J. Zeolites 1991, 11,438. (43) Parise, J. B.; Corbin, D. R.; Abrams, L. Acta Crystallogr. 1984, C40,1493. (44) Baerlocher, Ch. 15th Congr. IUCr.; Bordeaux, 1990; pp C177178.
Chem. Mater., Vol. 5, No. 12, 1993 1811
Frameworks in High Silica Zeolites
Table IV. Selected Structural Parameters of Zeolites zeolite/structure type code ZSM-lSIMTW ZSM-WMFI ZSM-ll/MEL SSZ24b/AFI cubic faujaaite/FAU hexagonal faujasite/EMT
S i 4 bond length (A) 1.53-1.70 1.58-1.61 1.49-1.68 1.58-1.62 1.60-1.63 1.60-1.67
Si-0-Si angle (deg) 134-159 141-169 144-165 142-164 136-151 132-148
av Si-0-Si angle (deg) 150.1 153.0 153.8 149.1 143.5 141.4
0 - S i 4 angle (deg) 103-114 107-112 92-133' 97-119 107-113 103-117
ref 41 39 40 42 43 44
*
0 Calculated from data in ref 40 using GSAS program. All parameters calculated from fractional coordinates given in ref 42 ueing GSAS program.
1.607 A is in excellent agreement with the same data for quartz. The presence of slightly shortened Si-0 bonds suggests that Si-0-Si angles larger than the quartz Si0-Si angle of 143.6'3s are present in the framework as predicted by molecular orbital calculation^.^^ In fact, the angles range from 141.3' to 169'. The average 0-Si-0 angle has been reported to be 109.5', ranging from 107.1' to 111.5', which is barely different from the angles in quartz. Powder X-ray diffraction data on FAU ((Si/ &mework = 10.8) give T-0 bond lengths that range from 1.60 to 1.63A, 0-T-0 angles from 107' to 113' and T-0-T angles between 136' and 151'. These data support the above trend. The slightly longer T-0 bond lengths of 1.63Aare consistent with T-0-T angles of 136°46but also may reflect the aluminum content. However, all other powder X-ray diffraction data show deviations from the above trends. Detailed comparison of all structures is made in Table IV, and histograms of T-0-T angles are shown in Figure 4. Note the presence of T-0-T angles significantly smaller than 143.6' (the angle in quartz) in the frameworks of ZSM-12, cubic faujasite, and hexagonal faujasite. Noteworthy are the very short Si-0 bond lengths (1.49 and 1.53 A) reported for ZSM-11 and ZSM-12, the very long Si-0 bond lengths (1.67 to 1.70 A) reported for ZSM11, ZSM-12, and EMT, and the rather large range of 0-Si-0 angles (92-133') encountered in several of these refinements. These ranges are wider than those generally seen in most reliable single-crystalstudies of silicates.They may reflect, not so much real structural variation, but the difficulty in getting stable and reasonable convergence of Rietveld refinements of such complexstructures. Whether these anomalous bond lengths and angles within silicate tetrahedra are confirmed by future single-crystal studies, powder work on better samples, or silicon NMR remains to be seen. The above discussion suggests that the small differences in energy among the silica zeolites might be related to the distribution of Si-0-Si angles which connect relgtively undistorted Si04tetrahedra. Ab initio molecular orbital calculations4 show that the potential energy curve for the Si-0-Si angle shows a small barrier to linearity and does not rise sharply with decreasing intertetrahedral angle until that angle is less than about 135' (see Figure 5). The distribution of Si-0-Si angles observed in silicates in general, and in "densen Si02 polymorphs in particular>' follows a parallel trend, with angles between 140' and 180' very common, but angles below 135' infrequent, or absent in case of dense silica polymorphs (see Figure 6). Histograms for individual dense and zeolitic silicas (see Figure 4) show several interesting features. In quartz there is a single Si-0-Si angle of 143.6', near the optimum angle from the potential energy curve, cristobalite also has a (46) Hill, R. J.; Gibbs, G. V. Acta Crystallogr. 1979, 835,25. (46) Newton, M.D.; Gibb, G. V. Phys. Chem. Miner. 1980,6,221. (47) Baur, W. H. Acta Crystallogr. 1980, 836, 2198.
single angle, with a value of 146.7°.37 All other, including tridymite,a show a distribution of angles. The average angle ranges from 141' to 154' and does not appear to correlate with energetics (see Figure IC). Coesite, a highpressure polymorph and the densest tetrahedral Si02 structure, has been reported to have a 180' angle149 but there is still some debate about ita structure. The silica zeolites all have some Si-0-Si angles between 145' and 170'; however, ZSM-12,EMT,and FAU have a population of Si-0-Si angles 1140'. Though these are very likely in small four-membered rings, such rings in a structure do not automatically imply the presence of small angles, as is evident in the case of ZSM-5 and ZSM-11. Faujasite is an extreme case, with two distinct populations of rings, one represented by four- and six-membered, and the other by twelve-membered. Indeed, as has been argued for amorphous s i l i ~ a s the , ~ ~presence ~~ of small rings in a structure does not imply a higher density, because the region between three- or four-membered rings contains larger rings and is less dense. The shape of the potential energy curve from quantum calculations suggests that small Si-0-Si angles are more destabilizing than large ones (see Figure 5). Figure 7 shows the enthalpy of silica zeolites relative to quartz plotted against the fraction of Si-0-Si angles 1140'. A reasonable correlation is seen, which is more pronounced than that for energy versus average Si-0-Si angle, framework density, or molar volume (compare Figure 1). Thus, Si-0-Si angles between 130' and 140' indeed appear to destablize silica zeolites slightly. This observation provides the basis for a hypothesis as to why very large pore frameworks seem harder to make in silicate and aluminosilicate zeolites than in aluminophosphates or zeolites containing boron, beryllium, or zinc. The hypothesis is that the pure-silica materials having large pores are destabilized not by their large cages, but by the small rings (three or four silicate tetrahedra) with angles between 120' and 135'. This is consistent with the observed large pore materials being those with P, Be, B, or Zn in the framework (i.e., in 4-fold coordination), since the optimum T-0-T, T-0-Si, or T-0-A1 angles for these cations are indeed shifted to lower values, as seen both from observed angles in other materials5254 and from molecular orbital calculations." Another interesting consequence of the similarity in energetics of all the observed silica zeolites has to do with the role of the template in zeolite synthesis. The energetics (48) Konnert, J. H.; Appleman, D. E. Acta Crystallogr. 1978, 834, 391. (49) Gibbs, G. V.; Prewitt, Ch. T.; Baldwin, K. J. 2 . Kristallogr. 1977, 145,108. (50) mi,P. D.; Navrotaky, A.; Rabinovich, E. M.;Ying, J. Y.; Benziger, J. B. J.Non-Crystal. Solrds 1990, 124, 101. (51) Gerber, Th.; Himmel, B. J. Non-Crystal. Solids 1987, 92, 407. (52) Heese, K. F.; Liebau, F.; Bchm, H. Acta Crystallogr. 1977,833, 1333. (53) Rudolf, P. R.; Crowder, C. E. Zeolites 1990,10, 163. (54) Toesell, J. A.; Gibbs, G. V. Acta Crystallogr. 1978, A34, 463.
1812 Chem. Mater., Vol. 5, No. 12, 1993
Petrovic et al. ZSM-5 I "clinic
t
Quart!
9
ZSM-11
25
f
1
Av. W "#a:153.8 d.0 MdW34,KW
-"'is0
S i 0 3 angle (deg)
Hexaaonal fauiasite
I
I
25
ZSM-12
e A". S0-9 mde: 150.1 d.0 Mdwbn~:ll.DSEdml
Figure 4. Histograms of Si-0-Si angles for individual Si02 polymorphs, and zeolitic silicas in order of increasing molar volume: (a) coesite, (b) quartz, (c) cristobalite, (d) tridymite, (e) ZSM-12, (0ZSM-5, (g) ZSM-11, (h) SSZ-24, (i) EMT, and (j)FAU.
here imply that, with no template present, the various polymorphs are all energetically quite similar, indeed largely within 2RT, twice the thermal energy (6.2 kJ/mol
at 373 K, a common synthesis temperature) of each other. Thus, the role of the template is to select, on geometric, entropic, and kinetic grounds, aparticularstructureamong
Chem. Mater., Vol. 5. No. 12, 1993 1813
Frameworks in High Silica Zeolites
-70
-
100
120
140
160
%Z-ZO
180
Si-OSi angle (9
Figure 5. Relative total energy (ET)for €bSilOl calculated aa a function of the Si-O-Si angle for bridging bond lengths of 1.65 A (upper curve), 1.62 A (middlecurve),and 1.59 A (lower curve). Arrows indicate theminimumenergySi-O-Siangle(afterNewton and Gibbs).‘B
20 Y
60
40
% SI-0-Si angles 5 140
degrees
Figure ‘7. Enthalpy of silica zeolites relative to quartz va the fraction of Si-0-Si angles 5140’.
the cages, hut it does not have to overcome the harrier of stabilizing a structure which would he very unstahle in the absence of template. We are presently investigating the energetics of zeolite-template interaction hy direct calorimetric measurements.
Silicates
’ r i
Conclusions The crystal structure/stahility relations in the series of pure- and high-silica zeolites have been studied by hightemperature solution calorimetry in lead borate solvent and transposed temperature drop calorimetry a t 977 K. The enthalpies of formation have been evaluated.
Silica polymorphs
b
All zeolitic silica structures studied are similar in enthalpytoquartz,destabhd hy7-14 !d/mol, andwithin 7 kJ/mol of each other. This implies that the role of the template in zeolite synthesis is not to stabilize a structure, which without the template would he very unstahle, hut to fulfill a kinetic or entropic role hy directing the path of the reaction. The small variations in enthalpies relative to quartz do not correlate systematically either with the degree of “openness” of the structure (framework density or molar volume) or with the average Si-0-Si angle. However, the presence of a significant number of Si-0-Si angles lower than 140° in ZSM-12, EMT, and FAU correlateswith their somewhat greater energetic destabilization. Heat contents (Hg7,- Hm) including those for “dense” silica polymorphs show a linear decrease with decreasing framework density or increasing molar volume.
Figure 6. Histograms of Si-0-Si angles observed in (a)silicates (254 angles)and (b)silicapolymorphs(39 angles). The histograms were prepared using ‘uncorrected” values from Baur.“
many of similar energy. This is in agreement with empirical OhSeNatiOIIS in zeolite synthesis.”aM The template may or may not interact strongly to stabilize that particular structure relative to others with template in (55) Zones,S.I.;vanNordstrand,RA.;~tiUi,D.S.;Wilson,D.M.; Yuen, L.;Scampavia, L. D. Zeolites: Facts, Figures, Future;Jambs, P. A., van Santen. R. A., Eds.; Elsevier: Amsterdam, 1989: p 299.
Acknowledgment. TheauthorsthankDr. J.B.Higgins of Mohil Research and Development Co. for many useful discussions, especially of crystallography,and Dr. J. Cook of Tosoh USA for providing the sample of high-silica cubic faujasite. We also thank M. Borcsik from Princeton University for ICPES and AAS analyses and Dr. J. s. Delaney from Rutgers University for help with electron microprobe work. Comments from Dr. E. M. Flanigen of UOP Research and Development are alsoappreciated. This work was supported hy the U. S. Department of Energy (Grant DE-FG 02-85ER13437).