J. Phys. Chem. 1996, 100, 16641-16646
16641
UiO-7: A New Aluminophosphate Phase Solved by Simulated Annealing and High-Resolution Powder Diffraction D. E. Akporiaye,*,† H. Fjellva˚ g,‡ E. N. Halvorsen,‡ J. Hustveit,‡ A. Karlsson,† and K. P. Lillerud*,‡ SINTEF Applied Chemistry, P.O. Box 124, Blindern, N-0314 Oslo, Norway, and Department of Chemistry, UniVersity of Oslo, P.O. Box 1033, Blindern, N-0315 Oslo, Norway ReceiVed: April 8, 1996; In Final Form: July 8, 1996X
A new aluminophosphate material UiO-7 has been prepared and the structure solved. The solution of the structure is the first known example applying the recently developed simulated annealing algorithm to identify the framework topology of an unknown structure. High-resolution synchrotron X-ray powder diffraction was then utilized to carry out a full Rietveld structure refinement. The importance of fluoride, used as a means of modifying a standard synthetic gel, is indicated by its presence as part of the structure, bridging two aluminum atoms within the cagelike building block.
Introduction Crystalline microporous molecular sieve structures like zeolites and related metalloaluminophosphates find many important uses in the oil, petrochemical, and chemical industries, typically in the fields of catalysis and separation. An important factor in defining potential areas for their application is the availability of the widest variety of framework types covering a range of pore diameters, shapes, and geometries. Thus, one of the goals in the hydrothermal synthesis of these materials is the identification of new synthetic parameters that result in the appearance of novel phases which may be then isolated in a pure form. The introduction of fluoride into known synthetic recipes1 has opened up new possibilities which in our case has resulted in the formation of a new aluminophosphate UiO-7, the structure of which has been solved through a combination of recent methods based on simulated annealing and high-resolution synchrotron diffraction data. Experimental Section Synthesis. UiO-7 was synthesized from a fluoride-modified aluminophosphate gel containing the organic amine tetramethylammonium hydroxide (TMAOH). This system, in the absence of the fluoride, usually forms AlPO4-20,2 an aluminophosphate variant of the mineral sodalite. The synthesis was carried out from a gel having a molar oxide ratio of 1 Al2O3:1 P2O5:0.2 HF:1 TMAOH:50 H2O, using Catapal B pseudobohemite (Vista Chemical Co.), phosphoric acid (85 wt % H3PO4 (Kebo Lab)), hydrofluoric acid (48 wt % HF extra pure (Merck)), and tetramethylammonium hydroxide (TMAOH‚5H2O (Fluka)). The gel was prepared in Teflon liners in which pseudobohemite was first mixed with water and phosphoric acid. The amine was then added followed by HF after which the gel was well stirred. The liners were put in stainless steel autoclaves and heated with no agitation in an oven at 150 °C for 21 h, after which they were quenched in cold water and the microcrystalline product was separated, washed with water, and dried. The presence of the novel phase was identified by powder X-ray diffraction, and XRD analysis after calcination in air at 600 °C indicated that it was stable at relatively high temperatures. The calcined †
SINTEF Applied Chemistry. University of Oslo. X Abstract published in AdVance ACS Abstracts, September 15, 1996. ‡
S0022-3654(96)01046-5 CCC: $12.00
sample was subsequently brought into contact with normal atmospheric humidity. Characterization. The X-ray powder diffractogram was routinely obtained using a Siemens D500 using Cu KR radiation and a Ge primary monochromator. High-quality X-ray diffraction powder data was obtained at the Swiss Norwegian beamline (D1) at the European Synchrotron Radiation Facility (ESRF). The sample was held in a rotating, sealed glass capillary and photons of wavelength 0.999 34 Å (from measurements on a Si powder standard) were obtained from a Si(111) channel cut monochromator. Intensity data were collected between 5 and 65° in 2θ using a step length of 0.005°. 31P MAS NMR spectra were obtained using a Varian VXR 300 S spectrometer operating at 300 MHz proton resonance frequency. The instrument was equipped with both Jakobsen and Doty MAS probes. The conditions for acquisition of the spectra were as follows: frequency 121.421 MHz, sweep width 16 000 Hz, pulse width 8 ms, acquisition time 0.5 s, number of scans 10 000, MAS spinning speed 4.5 kHz, and reference 85% H3PO4 solution. The water adsorption isotherm was measured using a homeconstructed fully automated system consisting of a CAHN D200 microbalance connected to a high-vacuum system and a highresolution dosing valve. The sample, from which the template had been removed by calcination at 600 °C in air, was dehydrated overnight at 400 °C after which the isotherm was measured. Scanning electron micrographs were taken using a JEOL JSM-840 instrument at a magnification of 10 000 to 20 000. The thermogravimetric analysis was obtained using a StantonRedcroft STA 785 combined TG/DSC system. 30 mg of the fresh sample was heated in an oxygen gas flow of 20 cm3/min, using a heating rate of 10 K/min. The reference sample was alumina. Results and Discussion The presence of the novel aluminophosphate phase was initially identified by powder X-ray diffraction analysis, shown in Figure 1, and its high thermal stability was confirmed by removal of the organic amine by calcination at 600 °C in air. However, due to the platelate crystal morphology of this material, as evident in the scanning electron micrographs presented in Figure 2, further careful optimization of the synthetic conditions © 1996 American Chemical Society
16642 J. Phys. Chem., Vol. 100, No. 41, 1996
Akporiaye et al.
Figure 3. Water adsorption isotherm of calcined UiO-7. Inset: Correlation of framework density and micropore volume VT.
Figure 1. Powder X-ray diffraction patterns of UiO-7 in calcined and as-synthesized form. The synchrotron data have been converted to a wavelength of 1.5406 Å.
Figure 2. Scanning electron micrographs of UiO-7.
was necessary to obtain the final sample in which the effects of line broadening were minimized, allowing more detailed structural studies to be carried out. The high-quality of the sample is evident in the diffraction data obtained from the synchrotron source, Figure 1. A set of 20 nonoverlapping reflections were chosen from the synchrotron data and individually fitted to obtain an accurate list of d values for input into
the indexing program TREOR90.3 The results of this analysis gave an orthorhombic unit cell (a ) 7.266 Å, b ) 15.333 Å, c ) 16.592 Å) with a very high figure of merit. A careful analysis of systematic absences based on this unit cell indicated two possible primitive space groups, Pbcm or Pbc21. In the initial determination of the structure, two approaches were selected (i) model building based on intuitative knowledge from known topologies, and (ii) solution based on simulated annealing. Trials using the first strategy were only successful in identifying a set of known structures with unit cell dimensions similar to the three edges of the suggested orthorhombic cell. In particular, the a axis was rather unusual for aluminophosphate materials, with only the three structures AlPO4-12, SAPO-40, and AlPO4-254 having similar repeat distances. A variety of proposed systems based on these structures were obtained but were not consistent with the observed data. The second approach, structure determination through simulated annealing (SA), is based on the wealth of prior structural knowledge available for these four-connected frameworks, which can be combined with the symmetry restrictions imposed by the space group.5 Thus, applying a penalty factor based on deviations from prescribed connectivities, bond lengths, bond angles, and site occupancy, it is possible to generate a series of framework topologies ranked in terms of a figure of merit. Most importantly, the match with the observed powder diffraction pattern can also be included, greatly improving the success rate. This method has been extensively validated using the known zeolite and aluminophosphate topologies and has been shown to have a high success rate. To our knowledge, this is the first published use of the SA strategy in the solution of a completely unknown framework topology. Important input information for the SA algorithm includes the total number of T atoms (NT, T ) Al or P) in the unit cell, the number of T atoms in the asymmetric unit (NU), the space group, and the relative intensities of reflections from the decomposed diffraction pattern. Further characterization of the sample was undertaken in order to obtain values for these input parameters. Assignment of NT. The total number of T atoms in the unit cell was determined through the known relationship of the framework density (FD) and the total micropore volume VT. The latter property was estimated by measuring the complete water adsorption isotherm of the calcined material at 298 K. As shown in Figure 3, the adsorption isotherm for UiO-7 is of type V under the standard IUPAC classification.6 This appears to be typical for aluminophosphate materials,7 which show a low affinity for water as indicated by the small water uptake at low pressures. Following the procedures of Davis et al.,8 an
UiO-7: A New Aluminophosphate Phase
J. Phys. Chem., Vol. 100, No. 41, 1996 16643
estimate of the micropore volume was taken as the point where the low-pressure branch intersects with the high-pressure branch, giving a value of 0.21 cm3/g. This value of VT was correlated with three other aluminosphosphates8 in a plot of FD against VT. As shown in Figure 3, the estimated value of the framework density (FD) was 17, giving a value of 32 for NT according to the relation:
NT ) FD(T/1000 Å3)Vuc (Å3)/1000 where Vuc is the volume of the unit cell. Assignment of NU. The determination of NU was based on the analysis of the 31P and 27Al solid state MAS NMR spectra of the calcined and as-synthesized materials (Figure 4). Considering first the 31P spectra, three distinct peaks could be identified in both samples. Deconvolution of the peaks, assuming a Gaussian profile, resulted in a relative area of, in the order of decreasing chemical shift, 1:2:1 for the assynthesized material and 2:1:1 for the calcined sample. The difference in the spectra before and after calcination, in particular the disappearance of the peak at -16.5 ppm after calcination followed by the appearance of the peak at -29.1 ppm, was considered to be due to the removal of fluorine. Earlier studies of aluminophosphates prepared using HF have already shown that the presence of fluoride when coordinated with the framework can result in a framework distortion that manifests itself in the 31P spectra.9 The effect of fluorine is to some extent evident in the 27Al NMR spectra (Figure 4b), which exhibits a major peak at 30.9 ppm, usually associated with aluminum in a tetrahedral environment, and additional intensity in the range of 0.2 to -15 ppm, which could be associated with aluminum within a distorted environment with higher coordination. This assumption was further supported by mass spectroscopic analysis of desorbed products from a thermogravimetric analysis of the as-synthesized sample (Figure 5), from which it was evident that the weight changes associated with the decomposition of the template also involved the simultaneous release of fluorine. Considering the 27Al spectra of the calcined material, three distinct peaks are observed at 39.7, 10.7, and -17.1 ppm. These can be associated respectively with tetrahedral, pentahedral, and octahedral aluminum environments, giving an indication of some degree of interaction of the framework with water after the removal of the organic amine. In previous 31P NMR studies of aluminophosphates such as VPI-5,10 it has been evident that individual peaks in the spectra could be associated with distinct crystallographic sites in the structure. Thus, in the example of VPI-5, even small distortions in the framework symmetry due to the presence of water were reflected in changes in the spectra. Therefore, in considering the assignment of a value to NU, the three peaks of both the as-synthesized and calcined form could be taken as indicating the presence of three distinct P enivronments in the assymetric unit, equivalent to a NU value of 6 (Al + P). However, in addition to this value, there are at least two other possible values which may be derived from alternative analyses of the spectra. Firstly, it is feasible that the peak at -19.9 and -23.8 ppm in the as-synthesized and calcined material, respectively, could result from the coincidence of two independent peaks, resulting in four distinct P environments and an NU value of 8. Alternatively, the close approach of the two high-field peaks at -29.1 and -30.4 ppm in the calcined material may indicate that these two P sites are so similar that they represent the same topological site and the overall value of 4 should be given to NU. The differences of the chemical shifts of these two peaks in the as-synthesized material, as previously discussed, could come from the effect of fluorine.
Figure 4. Solid state MAS NMR spectra of UiO-7 in the as-synthesized and calcined form: (a, top) 31P spectra (observed solid line, deconvoluted dashed line); (b, bottom) 27Al spectra.
Simulated Annealing. Following the assignment of possible values to the relevant input parameters, a series of SA runs were then carried out for each of the three values of NU in combination with the two proposed space groups. Tests of the NU value of 6 and 8 in over a hundred SA trials within space groups Pbc21 and Pbcm did not result in generated structures with a viable figure of merit. In contrast, for NU ) 4, SA runs with the diffraction intensities from the as-synthesized repeatedly generated a single distinct framework topology in the Pbcm space group, always with a very high figure of merit. This topology appeared very promising and was further optimized using the DLS program,11 assuming all T atoms to be of the same type.
16644 J. Phys. Chem., Vol. 100, No. 41, 1996
Akporiaye et al.
Figure 5. Thermal gravimetric analysis (oxygen atmosphere) of assynthesized UiO-7: (s) % weight change, (‚‚‚) differential weight change. Figure 7. Detailed view of the structural subunit, indicating the assignments of the various T atoms and the location of fluorine within the unit.
Figure 6. Framework of UiO-7 showing the [010] projection of the framework along the main 8-ring channel. Sections of the upper and lower layers are deemphasized to highlight the alternating nature of the subunit chains and the position of two of the TMA+ molecules are indicated. Aluminum T atoms are shown in white and phosphorus in black.
The powder diffractogram simulated from the optimized coordinates is shown in Figure 8. As is evident, the close similarity of the observed pattern to the simulated results gave a very strong indication that the correct topology had been identified. Slight differences in intensities were assumed to be due to the effects of partial rehydration of the calcined material. Further examination of the framework then showed that, in order to take account of the Al/P ordering, the mirror plane of the Pbcm space group had to be removed, resulting in Pbca symmetry, a doubling of the unit cell along the a axis, and a doubling of NU from 4 to 8. As is clear from the [010] projection of the framework (Figure 6), UiO-7 has a two-dimensional channel system which is accessed through 8-ring windows. This topology has also been very recently identified in an independent single-crystal study of a zincoaluminophosphate12 prepared using a combination of TMA and tetrapropylammonium hydroxides in the absence of fluorine. The framework is built up of a rather unusual cagelike unit which we have also identified in the recently discovered structure of SAPO-40.13 In UiO-7, these units are connected together in chains which run parallel to the a axis. Adjacent
Figure 8. Simulated XRD diffraction patterns of alternative polymorphs from DLS optimized frameworks. Observed pattern of calcined UiO-7 is included for comparison.
chains are tilted, having the same orientation along the c axis, and opposite orientations along the b axis. In SAPO-40 these chains are not tilted with respect to each other, with the resulting interconnections forming a 12-ring channel system. The simple means by which these chains interconnect also suggest alternative hypothetical polytypes of the SAPO-40 and UiO-7 framework topologies, for example, that in which all chains in UiO-7 have the same orientation along both b and c directions (polytype A) or where they alternate in orientation along the c instead of b direction (polytype B). The simulated diffraction patterns for these two other polytypes have been calculated and are shown in Figure 8.
UiO-7: A New Aluminophosphate Phase
J. Phys. Chem., Vol. 100, No. 41, 1996 16645 TABLE 2: Fractional Atomic Coordinates of As-Synthesized UiO-7 from Refinement of the Synchrotron X-ray Diffraction Data
Figure 9. Observed (upper solid line), calculated (middle solid line), and difference (lower solid line) synchrotron powder diffraction patterns from the Rietveld refinement of UiO-7.
TABLE 1: UiO-7 Crystallographic Data unit cell contents framework density (T atoms/1000 Å3) refined pattern range in 2θ (deg) step scan increament in 2θ (deg) wavelength (Å) space group a (Å) b (Å) c (Å) unit cell volume (Å3) no. of profile data steps no. of contributing reflections no. of soft constraints no. of structural variables no. of profile parameters Rp Rwp Rexp
Al32P32O128(TMAF)8 17.3 6.0-65.0 0.005 0.99934 Pbca (No. 61) 14.533(3) 15.334(6) 16.601(4) 3699.8 11919 2483 86 93 6 0.025 0.036 0.014
Structure Refinement. As the significant differences in the diffraction patterns and NMR spectra of the calcined and assynthesized materials were considered to be caused by the influence of fluorine and amine on the structure, a more detailed structural refinement of the synchrotron data was carried out, in order to fully confirm the structural model and to locate all nonframework atoms. The geometric optimized coordinates within the new enlarged unit cell were used as input into the Rietveld refinement of the high-resolution XRD data.14 The initial stages of the refinement gave a quite good agreement which allowed subsequent cycles of refinements and difference Fourier syntheses to locate the position of the TMA molecule. As shown in Figure 6, the organic amine cation fully occupies all of the cages formed from the intersection of the two 8-ring channels. No significant residual electron density was evident in the rest of the channel. This was in good agreement with the TGA results, which exhibited a 20% weight loss in the range of 450-550 °C, this being equivalent to eight TMA-F ion pairs per unit cell. Since one of the most likely locations of the fluorine anions was considered to be within the cagelike subunit, difference Fourier maps were analyzed carefully within this region, eventually showing that the fluorine was accommodated between pairs of Al atoms, giving them 5-fold coordination. The fluoride thus bridges one of the 4-rings and points into the subunit (Figure 7). The final results of the refinement are shown in Figure 9, the details are given in Table 1, the atomic
atom
x
y
z
P1 P2 P3 P4 Al1 Al2 Al3 Al4 O1 O2 O3 O4 O5 O6 O7 O8 O9 O10 O11 O12 O13 O14 O15 O16 F N C1 C2 C3 C4
0.442(5) 0.065(2) 0.085(6) 0.272(9) 0.062(2) 0.442(9) 0.428(4) 0.217(1) 0.442(9) 0.523(3) 0.356(3) 0.459(1) 0.152(5) 0.072(3) -0.012(9) 0.060(2) 0.070(2) 0.044(4) 0.039(9) 0.186(7) 0.324(4) 0.325(1) 0.245(5) 0.191(5) 0.553(8) 0.321(0) 0.408(7) 0.313(9) 0.233(6) 0.313(8)
0.595(5) 0.273(0) 0.578(5) 0.838(0) 0.118(1) 0.782(1) 0.096(8) 0.339(0) 0.509(6) 0.596(2) 0.600(3) 0.665(6) 0.310(6) 0.282(2) 0.328(3) 0.181(6) 0.487(9) 0.586(8) 0.643(6) 0.602(0) 0.816(8) 0.894(2) 0.751(5) 0.887(1) 0.281(8) 0.432(9) 0.471(3) 0.347(6) 0.468(5) 0.396(29
0.928(7) 0.668(3) 0.076(5) 0.948(1) 0.554(6) 0.864(9) 0.421(1) 0.545(0) 0.881(4) 0.980(9) 0.978(4) 0.865(5) 0.629(7) 0.756(8) 0.639(6) 0.643(6) 0.108(1) -0.005(8) 0.132(1) 0.068(5) 0.874(9) 0.007(1) -0.013(8) 0.923(3) 0.512(1) 0.258(2) 0.265(4) 0.289(3) 0.256(4) 0.181(0)
TABLE 3: Selected Geometric Parameters P-O (mean)/Å P-O (range)/Å Al-O (mean)/Å Al-O (range)/Å O-P-O (mean)/deg O-P-O (range)/deg O-Al-O (range)/deg P-O-Al (range)/deg AL1-F1/Å AL2-F1/Å AL1-F1-AL2/deg O10-AL1-F1/deg O14-AL1-F1/deg O9-AL1-F1/deg O8-AL1-F1/deg O4-AL2-F1/deg O6-AL2-F1/deg O11-AL2-F1/deg O13-AL2-F1/deg
1.50 1.46-1.53 1.79 1.72-1.85 109.43 104.93-118.67 89.10-123.46 126.69-168.25 1.90 2.06 124.56 84.45 84.14 173.83 92.03 88.78 175.53 86.61 86.77
TABLE 4: Assignments of the 31P NMR Spectra of As-Synthesized and Calcined Forms as-synthesized
calcined
obsd
calcd
assignment
obsd
calcd
assignment
-16.5 -19.9 -19.9 -31.9
-25.8 -24.5 -24.2 -31.6
P3 P1 P4 P2
-23.8 -23.8 -29.1 -30.4
-23.3 -29.2 -31.5 -33.5
P1 P4 P3 P2
parameters listed in Table 2, and selected calculated geometric parameters presented in Table 3. After obtaining the full structural details from the Rietveld refinement, a full assignment of the 31P spectra was then considered, based on trends predicted by the empirical relationship between the average P-O-Al bond angle and the chemical shift.15 A comparison of the calculated and observed chemical shifts for the four P sites is given in Table 4. Site P2, with the largest bond angle can be safely assigned to the peak at -31.9
16646 J. Phys. Chem., Vol. 100, No. 41, 1996 ppm in the as-synthesized material. The P3 site is assigned to the -16.5 peak, even though it has a predicted chemical shift which is somewhat different at -25.8 ppm. Support for this assignment comes from the fact that when the effect of fluorine removal is simulated by optimizing the structure using DLS, the predicted chemical shift for the P3 site moves to -31 ppm, a change that is consistent with the changes observed after calcination. The peak at -19.9 ppm, which results from the overlap of two of the phosphorus sites, must then be due to the P1 and P4 sites. This is also consistent with the similarity of the environment around these two sites, both being centered within three 4-ring units. Conclusions A new aluminophosphate molecular sieve has been prepared from a fluoride modified gel. The solution of the structure is the first example of the use of a combination of simulated annealing and high-resolution synchrotron X-ray powder diffraction. The structure exhibits a two-dimensional 8-ring channel system containing a TMA-F ion pair, with the fluorine coordinated between two Al atoms. The discovery of UiO-7 shows the continued potential for the preparation of novel phases through modification of known gel systems. Acknowledgment. The authors acknowledge the financial assistance of the Research Council of Norway and the opportunity to collect the high-resolution diffraction data at the Swiss Norwegian beamline at ESRF, contribution x.96. We also thank Norsk Hydro and Statoil for their cooperation in the use of the Biosym Technologies Software.
Akporiaye et al. References and Notes (1) Guth, J. L.; Kessler, H.; Caullet, P.; Hazm, J.; Merrouche A.; Patarin, J. In Proceedings from the 9th International Zeolite Conference; von Ballmoos, R., Higgins J. B., Treacy, M. M., Eds. ButterworthHeinemann: Stoneham, MA, 1993; Vol. 1, p 215. (2) Wilson, S. T.; Lok, B. M.; Messina, C. A.; Cannan, T. R.; Flanigen, E. M. J. Am. Chem. Soc. 1982, 104, 1146. (3) Werner, P. E.; Eriksson, L.; Westdahl, M. J. Appl. Crystallogr. 1985, 18, 367. (4) Meier, W. M.; Olson,. D. H. Atlas of Zeolite Structure Types; Butterworths: London, 1993). (5) (a) Deem, M. W.; Newsam, J. M. J. Am. Chem. Soc. 1992, 114, 7189. (b) Structure Solve, Biosym Technologies, Inc. (6) Gregg, S. J.; Sing, K. S. W. Adsorption, Surface Area and Porosity; Academic Press: New York, 1991. (7) Carrott, P. J. M.; Kenny, M. B.; Roberts, R. A.; Sing, K. S. W.; Theocharis, C. R. In Characterisation of Porous Solids II; Proceedings of the IUPAC Symposium, Alicante (1990); Rodrigues-Reinoso, F., Rouguerol, J., Sing, K. S. W., Unger, K. K. Eds.; Studies in Surface Science; 1991, Vol. 62, p. 685. (8) Davis, M. E.; Montes, C.; Hathaway, P. E.; Arhancet, J. P.; Hasha, D. L.; Garces, J. M. J. Am. Chem. Soc. 1989, 111, 3919. (9) Guth, F. Thesis, Mulhouse, 1989. (10) van Eck, E. R. H.; Veeman, W. S. J. Am. Chem. Soc. 1993, 115, 1168. (11) Baerlocher, Ch.; Hepp, A.; Meier, W. M. DLS-76: A Program for the Simulation of Crystal Structures by Geometric Refinement; Institute of Crystallography and Petrography, ETH, Zurich, Switzerland, 1978). (12) Marler, B.; Patarin, J.; Sierra, L. Microporous Mater. 1995, 5, 151. (13) Estermann, M. A.; McCusker, L. B.; Baerlocher, Ch. J. Appl. Crystallogr. 1992, 25, 539. (14) Larson, A. C.; Von Dreele, R. B. GSAS, Generalized Structure Analysis System; MS-H805, Los Alamos Neutron Scattering Center, Los Alamos National Laboratory, Los Alamos, NM, 1988. (15) Mu¨ller, D.; Jahn, E.; Ladwig, G.; Haubenreisser, U. Chem. Phys. Lett. 1984, 109, 332.
JP961046A
