Chem. Mater. 2007, 19, 2261-2268
2261
Synthesis and Structure Determination of a Novel Layered Aluminophosphate Material Templated with 1-Phenylethylamine: [AlPO4(OH)](NH3C2H4C6H5) Robert W. Dorner,* Malek Deifallah, David S. Coombes, C. R. A. Catlow, and Furio Cora`* DaVy Faraday Research Laboratory, Royal Institution of Great Britain, 21 Albemarle Street, London W1S 4BS, United Kingdom ReceiVed January 11, 2007. ReVised Manuscript ReceiVed March 7, 2007
The synthesis and crystal structure of a novel layered aluminophosphate is described. The structure was solved from single-crystal X-ray diffraction data and confirmed by high-resolution powder diffraction and computational studies. The as-synthesized layered material, with composition [AlPO4(OH)](NH3C2H4C6H5), crystallizes in the monoclinic space group C2/c with a ) 39.06(10) Å, b ) 5.31(13) Å, c ) 9.67(2) Å, R ) 90°, β ) 94.6(4)°, and γ ) 90°. In the aluminophosphate layers, the six-coordinated aluminum polyhedra form infinite chains that are cross-linked by phosphate groups. These inorganic layers are stabilized via strong hydrogen bonding to the protonated organic templates, involving the terminal oxygen atoms of the phosphate groups. Using quantum mechanical (QM) and interatomic potential (IP) techniques, we established the location of the protons in the layer and the structure’s stability.
Introduction Since the work of Wilson et al. in the early 1980s, there has been a large increase in the number of known framework topologies of aluminophosphates1,2 many with interesting catalytic and adsorption properties.3 Structure direction in the hydrothermal synthesis of these materials occurs generally through derivatives of organic amines, with the organic groups usually being alkanes. It is still not clearly understood how certain organic molecules act as templates or structuredirecting agents (SDAs). Some frameworks, e.g., AlPO-18, can be synthesized using several different organic molecules, examples being N,N-diisopropylethylamine4 or tetraethylammonium hydroxide,5 and some templates can form several different frameworks, e.g., triethylamine can form the CHA6 and the AFI topology.7 In this paper, we describe the synthesis and crystal structure of a two-dimensional aluminophosphate, [AlPO4(OH)](NH3C2H4C6H5), obtained using 1-phenylethylamine as the organic SDA. Similar layered structures have been reported previously8,9 and can be found in nature as the * Corresponding author. E-mail:
[email protected] (F.C.);
[email protected] (R.W.D.). Fax: 44 (0)20 7670 2958. Tel: 44 (0)20 7409 2992.
(1) Wilson, S. T.; Lok, B. M.; Flaningen, E. M. U.S. Patent 4310440, 1982. (2) Wilson, S. T.; Lok, B. M.; Messina, C. A.; Cannon, T. R.; Flaningen, E. M. J. Am. Chem. Soc. 1982, 104, 1146. (3) Szostack, R. Molecular SieVes; Blackie Academic and Professional: London, 1992. (4) Concepcion, P.; Blasco, T.; Nieto, J. M. L.; Vidal-Moya, A.; MartinezArias, A. Microporous Mesoporous Mater. 2004, 67, 215. (5) Wendelbo, R.; Akporiaye, D.; Andersen, A.; Dahl, I. M.; Mostad, H. B. Appl. Catal., A 1996, 142, L197. (6) Denavarro, C. U.; Machado, F.; Lopez, M.; Maspero, D.; Perezpariente, J. Zeolites 1995, 15, 157. (7) Liu, Y.; Withers, R. L.; Noren, L. Solid State Sci. 2003, 5, 427. (8) Massa, W.; Yakubovich, O. V.; Karimova, O. V.; Dem’yanets, L. N. Acta Crystallogr., Sect. C 1995, 51, 1246. (9) Simon, N.; Guillou, N.; Loiseau, T.; Taulelle, F.; Ferey, G. J. Solid State Chem. 1999, 147, 92.
mineral tancoite with octahedral aluminophosphate chains.10 However, the arrangement of the linking phosphate groups present in our new structure and the phenyl-based anchors have not been seen before. Layered aluminophosphate materials are usually synthesized with diaminoalkanes as SDAs, with the diamine generally doubly protonated and bridging adjacent aluminophosphate layers. The focus on the choice of SDAs has been on alkane-based amines, because aromatic amines (e.g., aniline) do not tend to mix easily with the gel during synthesis. To the best of our knowledge, no new topology within the aluminophosphate materials has been synthesized using a phenyl- or benzyl-based template. Phenyl- and benzyl-based compounds have so far received little attention in the synthesis of aluminophosphates because of difficulties mixing the template with the gel composition. Aniline, the most readily available aromatic amine, forms AlPO4-C, although this material normally does not need an organic SDA to be synthesized. Aniline tends not to mix with the gel normally; it instead forms a layer on top of the gel, with the pH of the solution remaining at +2. Aniline is a relatively weak base (pKb ) 9.7)11 and is only weakly hydrated. The hydrophilicity of the compound, and hence its solubility in the synthesis gel, can be modified via the introduction of inductive groups (e.g., alkane groups) to the phenyl ring11 or by replacing aniline by benzylamine. Compounds such as 1-phenylethylamine (pKb ) 4.22),12 N-methylaniline (pKb ) 4.42), benzylamine (pKb ) 3.66), (10) Ramik, R. A.; Sturman, B. D.; Dunn, P. J.; Poverennykh, A. S. Can. Mineral. 1980, 18, 185. (11) Elliott, J. J.; Mason, S. F. J. Chem. Soc. 1959, 2352. (12) Tuckerman, M. M.; Mayer, J. R.; Nachod, F. C. J. Am. Chem. Soc. 1959, 81, 92.
10.1021/cm070106u CCC: $37.00 © 2007 American Chemical Society Published on Web 04/07/2007
2262 Chem. Mater., Vol. 19, No. 9, 2007
and phenylpropylamine (pKb ) 3.80)13 with lower pKb values than aniline form a homogeneous mixture after addition to the gel, as the nitrogen is more hydrophilic. Our synthetic strategy in the present work has been to alter the features of the aromatic amine in the SDA molecules by the addition of inductive groups to aniline and to employ a benzyl-based SDA, which we find leads to new materials. Our experimental study is supported by a computational investigation of the resulting layered material, which helps in deriving detailed structural information, particularly on the proton distributions. Methodology and Characterization Synthesis. The synthesis of the layered material was carried out hydrothermally in a Teflon-lined autoclave under high pressure. Aluminum hydroxide hydrate (Al(OH)3xH2O, Aldrich), phosphoric acid (H3PO4, Aldrich, 85%), 1-phenylethylamine (H2N(C2H4)C6H5, Aldrich, 99%), and distilled water were introduced while being stirred continuously in a Teflon container, with the molar ratio of 1:1:1.3:30. A liner containing the gel was then placed in an autoclave and heated at 150 °C for 24 h. The pH of the solution (+8) was the same at the beginning and end of the reaction. The material was also synthesized in the presence of 10% zinc as well as cobalt acetate tetrahydrate (Aldrich, 99%). Other amines used in the synthesis gels were N-methylaniline (Synchemica), benzylamine (Aldrich), and phenylpropylamine (Aldrich), which did not yield single crystals of sufficient size nor crystallinity to be analyzed using single-crystal X-ray diffraction techniques. Our subsequent discussion concentrates therefore on the material synthesized using the 1-phenylethylamine template. X-ray Structure Collection and Structure Determination. Crystals of sufficient size and crystallinity could be collected to record single-crystal data at a wavelength of 0.6869 Å (120 K) on Station 9.8 at the SRS in Daresbury.14 The crystal was mounted on a glass fiber using Fomblin polyether. Station 9.8 is a highflux, tuneable, monochromatic, single-crystal diffraction station employing a D8 diffractometer and utilizing a Bruker-Nonius APEXII CCD area detector. The monochromator employed was a Silicon (111) crystal with an asymmetric cut of 2.01°, cooled by a GaInSn alloy and mounted on a Huber rotatory table, which can be positioned to a precision of 0.001°. The crystal selected was synthesized from a gel mixture with a 1.0:0.9:0.1:1.3:30 P:Al:Co:1-phenylethylamine:H2O molar ratio and had a size of 0.05 × 0.05 × 0.03 mm3. The space group was assigned on the basis of systematic absences and intensity statistics, leading to a satisfactory refinement. The structure was solved using the single-crystal software Crystals.15 SHELXS was used to locate the heavy atoms of Al and P, with the remaining atoms located through the difference electron density map. The phenyl H atoms were also found via a difference Fourier synthesis, and their positions were refined. The final refinement of the structure was performed on the data having I > 2σ(I) and included anisotropic thermal parameters for all non-hydrogen atoms. A summary of the single-crystal X-ray experiment is given in Table 1. Final positional and equivalent isotropic thermal parameters are listed in Table 2. The asymmetric unit and 3D crystal structure are shown in Figures 1 and 2, clearly showing a 2D layered structure. (13) Hall, H. J. J. Am. Chem. Soc. 1957, 79, 5697. (14) Cernik, R. J.; Clegg, W.; Catlow, C. R. A.; Bushnell-Wye, G.; Flaherty, J. V.; Greaves, G. N.; Burrows, I.; Taylor, D. J.; Teat, S. J.; Hamichi, M. J. Synchrotron Radiat. 1997, 4, 279. (15) Watkin, D. J.; Prout, C. K.; Lilley, P. M. Crystals; Chemical Crystallography Laboratory, University of Oxford: Oxford, U.K.
Dorner et al. Table 1. Crystallographic Data and Structure Refinement for Layered Material identification empirical formula fw T wavelength cryst syst, space group unit cell dimensions
V F(000) cryst size no. of independent reflns (I > 0) no. of obsd data [I > 2σ(I)] final R indices
layered material [AlPO4(OH)](NH3C2H4C6H5) 261 g mol-1 120 K 0.68690 Å monoclinic, C2/c a ) 39.06(10) Å b ) 5.31(13) Å c ) 9.67(2) Å R ) 90° β ) 94.6(4)° γ ) 90° 2002 Å3 594.0 0.05 × 0.05 × 0.03 mm3 3337 2388 R1 ) 0.0639 wR2 ) 0.073
Table 2. Atomic Coordinates and Equivalent Isotropic Displacement Parameters for Layered Material atom
x
y
z
Ueq (Å2)
P1 Al1 Al2 O1 O2 O3 O4 O5 H12 N1 H1A H1B H1C C1 H1D H1E H1F C2 H2 C3 C4 H4 C5 H5 C6 H6 C7 H7 C8 H8
0.04550(4) 0.00000 0.00000 0.02704(10) 0.03500(9) 0.03282(10) 0.08348(10) 0.01925(10) 0.023906 0.09246(13) 0.08906 0.09458 0.07401 0.12075(18) 0.09802 0.13845 0.12383 0.12429(15) 0.12482 0.15654(15) 0.18367(17) 0.18183 0.21374(18) 0.23240 0.21706(18) 0.23735 0.19034(17) 0.19239 0.16031(16) 0.14201
0.2022(3) -0.3168(4) 0.00000 -0.0177(7) 0.1897(7) 0.4544(7) 0.1779(8) -0.2802(7) -0.431250 -0.2652(10) -0.40020 -0.12358 -0.24902 -0.5594(13) -0.57128 -0.57295 -0.69318 -0.3042(12) -0.17165 -0.2745(12) -0.4415(13) -0.58155 -0.4067(15) -0.52023 -0.2047(16) -0.18446 -0.0350(13) 0.10353 -0.0685(12) 0.04921
0.26295(11) 0.25000 0.00000 0.3379(3) 0.1050(3) 0.3178(3) 0.2936(3) 0.0835(3) 0.047812 0.4398(4) 0.38142 0.38782 0.49068 0.6024(6) 0.63831 0.68122 0.53673 0.5321(5) 0.60652 0.4566(5) 0.4784(6) 0.53847 0.4135(8) 0.43151 0.3246(8) 0.27853 0.3025(6) 0.24010 0.3678(6) 0.35343
0.01704 0.01837 0.01684 0.00086 0.01858 0.01982 0.02439 0.01953 0.04972 0.02869 0.04303 0.04303 0.04303 0.03589 0.05383 0.05383 0.05383 0.02611 0.03134 0.02720 0.03372 0.04047 0.04364 0.05236 0.0438 0.05266 0.03538 0.04245 0.02873 0.03448
In addition, powder XRD measurements were performed. The data collected from Daresbury on Station 2.3 using a wavelength of 1.306 Å were analyzed by Rietveld methodology using the GSAS EXPGUI software.16,17 The data were collected with a step increment of 0.01° and the time for each step was 2 s. Two sets of patterns between 2 and 25° and five sets of patterns between 25 and 55° were recorded and the data summed. Scanning Electron Microscopy. The morphology of the crystals was characterized by scanning electron microscopy employing a Jeol JSM-630 IF scanning microscope. The material was obtained in a plate-like layered structure (Figure 3). Thermogravimetric Analysis. Thermogravimetric analysis (TGA) and differential thermal analysis (DTA) were performed on a PU (16) Larson, A. C.; Von Dreele, R. B. Report LAUR 86-748; Los Alamos National Laboratory: Los Alamos, NM, 2000. (17) Toby, B. H. J. Appl. Crystallogr. 2001, 34, 210.
[AlPO4(OH)](NH3C2H4C6H5)
Chem. Mater., Vol. 19, No. 9, 2007 2263
Figure 4. Thermogravimetric analysis of the layered material under a 90% argon and 10% oxygen atmosphere (heating rate 5 °C/min). Figure 1. Asymmetric unit of the layered aluminophosphate.
Figure 2. Structure of the layered aluminophosphate, showing two template molecules hydrogen bonded to layers of the material.
Figure 3. SEM image of the layered material. The layering of the structure is also clearly visible in the crystal morphology.
4K (Rigaku) equipped with a quadrupole mass spectrometer (MS, Anelva M-QA200TS), under a mixture of argon (90%) and oxygen (10%) atmospheres with a heating rate of 5 K/min between 22 and 800 °C. The thermogram revealed a weight loss of 51.8% up to 650 °C (Figure 4). The residue collected after heating the material
to 800 °C was examined using X-ray diffraction employing a Siemens D4 diffractometer. The material has lost all long-range order and is completely amorphous; the 51.8% weight loss can be explained by the loss of the organic anchors (46.7%), which hold the inorganic planes together as well as water molecules implying dehydroxylation of the layers. Decoupled Solid-State Magic Angle Spinning. Decoupled solid-state magic angle spinning (MAS) NMR was acquired for 27Al to gain further insight into the coordination chemistry of the aluminum. 27Al MAS NMR spectra were recorded on a Bruker MSL-300 spectrometer at 78.2 MHz. All measurements were carried out at room temperature, with Al(H2O)6+3 being used as the external standard. Computational. The computational work was performed first to determine the location of the protons in the layers, which is difficult to characterize via XRD techniques, and second to investigate the structural stability of the layers. A two-step approach was adopted using quantum mechanical (QM) and interatomic potential (IP) techniques. Computationally, the study of the AlPO structure itself is most suited to quantum mechanical techniques because of its unusual structure incorporating 6-coordinated Al and 1-, 2-, and 3-coordinated O ions. To begin with, we exploited the layering of the structure and cleaved the solid along the [100] direction, forming a single-layer model of composition [AlPO5H]-; the symmetry (C2/ c) and cell parameters were taken from the experimentally derived values (a ) 39.06 Å, b ) 5.31 Å, c ) 9.67 Å, R ) 90.0°, β ) 94.6°, γ ) 90°). The experimentally derived coordinates for Al, P, and O were used in this initial structure. Each primitive unit cell contains five symmetry-unique O ions, labeled 1-5 (Figure 1); in principle, each can act as a protonation site. We therefore generated five protonated single-layer models, each corresponding to binding of the proton to a different, symmetry unique oxygen. The lattice parameters of these two-dimensional computational models were not varied during our QM calculations, but internal optimizations of all the fractional coordinates were performed. Quantum mechanical calculations have been performed within the density functional theory (DFT). The QM code CRYSTAL18 was employed for these calculations, using the hybrid DFT functional B3LYP. The CRYSTAL code has been successfully used to model AlPOs in previous work.19-22 The Gaussian basis sets (18) Saunders, V. R.; Dovesi, R.; Roetti, C.; Orlando, R.; Zicovich-Wilson, C. M.; Harrison, N. M.; Doll, K.; Civalleri, B.; Bush, I. J.; D’Arco, P.; Llunell, M. CRYSTAL 2003 User’s Manual; University of Torino: Torino, Italy, 2004. (19) Cora, F.; Alfredsson, M.; Barker, C. M.; Bell, R. G.; Foster, M. D.; Saadoune, I.; Simperler, A.; Catlow, C. R. A. J. Solid State Chem. 2003, 176, 496. (20) Cora, F.; Catlow, C. R. A. J. Phys. Chem. B 2001, 105, 10278. (21) Cora, F.; Saadoune, I.; Catlow, C. R. A. Angew. Chem., Int. Ed. 2002, 41, 4677.
2264 Chem. Mater., Vol. 19, No. 9, 2007 used to describe the electrons in the system were the same as those described previously19-22 and can be obtained from the online library of the code.23 Ammonium ions (NH4+) were used in our single-layer model to represent the protonated 1-phenylethylamine ions. Our rationale for using protonated SDA ions arises from the observed change in pH from 2 to 8 of the homogeneous gel mixture, indicating that protons are removed from solution upon addition of the SDA. Each ammonium ion used here mimics the ammonium end of one organic template, which is the only part of the SDA close enough to the layer to have any significant chemical interaction with it. The remaining part of the organic template is likely to be involved in template-template interactions and/or is necessary for space filling. To examine the AlPO layer when no SDAs are present, we generated a neutral single-layer system in which the ammonium ion and the OH- ion in the layer are removed. A full optimization including the lattice parameters was conducted on this neutral layer.
Dorner et al. the system’s energy, Ea, as a function of the interlayer separation. The interaction energy between the layers, Elay, was then evaluated according to ∆Elay )
Ea - E∞ 2
(1)
where ∆Elay is the interaction energy per unit cell between two layers (which are separated by 4 SDA ions per unit cell); Ea is the total energy at lattice spacing a; E∞ is the total energy at infinite layer separation (measured at a ) 150 Å).
Results and Discussion
The stable protonated layer identified in the QM work was employed in the second stage of our computational study, performed using IP techniques. Energy minimization and Molecular Dynamics (MD) calculations were carried out on the system with the experimentally derived composition, [AlPO4(OH)](NH3C2H4C6H5). The goal of this approach was to study the organic templates residing in the interlayer region and in particular to examine the forces responsible for holding adjacent layers together. The cvff forcefield24 available in the Open Forcefield (OFF) suite of methods under Cerius2 was selected for this function because it yields good results for such organic-organic intermolecular interactions.25 The process adopted for this part of the computational study initially involved replacing the NH4+ ions present in the QM model by an equal number of protonated 1-phenylethylamine ions. The structure of the AlPO layers present in this system was fixed to that obtained in the QM work. The charges assigned to atoms in the layers are +1.4 to Al, +3.4 to P, and -1.2 to O ions, as customary in AlPO calculations performed with cvff.26 The FF type and charge of the SDA atoms were obtained using the direct atom typing and the charge equilibration features available in Cerius2. The Ewald method was selected to describe the nonbonded dispersive and Coulombic interactions. The smart minimizer function and an N, V, T ensemble set at 423 K with a time step of 0.001 ps for a total simulation time of 100 ps were used for the energy minimization and MD calculations. To examine the forces responsible for holding adjacent layers together, we calculated the energy of the system corresponding to different values of the a lattice parameter in the range of 35 Å e a e 150 Å. The values selected allowed us to examine the system at small interlayer separation (i.e., compression of the bulk system) and large interlayer separation (i.e., expansion of the bulk system). For the experimentally derived value of the a lattice parameter, we performed a sequence of energy minimization, molecular dynamics, energy minimization steps. No disorder or reconstruction was observed after the MD calculations and in fact the internal energy calculated by energy minimization of the structure before and after the MD step differed by only 1 meV. Therefore, for all other values of the a lattice parameter, only energy minimizations of the organic molecules were performed, to calculate
Experimental Structure Determination and Characterization of the Layered Material. The crystal structure of the newly synthesized layered material revealed on the basis of single-crystal X-ray diffraction data consists of aluminophosphate layers, which are bound by electrostatic interactions and hydrogen-bonding to protonated 1-phenylethylamine molecules (Figure 2). Usually, only one organic molecule can be found, connecting the sheets in the layered aluminophosphates.9,27,28 However, in this layered aluminophosphate material, two 1-phenylethylamine molecules are located between the inorganic sheets, leading to a large d-spacing between the inorganic layers of approximately 14 Å. Indeed, this is the largest known value to date; the largest previously determined spacing between layers in other 2D aluminophosphate materials is approximately 11 Å,29 arising from the diaminooctane template’s relatively long alkane chain, rather than from two smaller template molecules as in this case. Although two organic templates are not known to be found in-between aluminophosphate layers, this has been observed in layered gallophosphates.30 Another interesting characteristic of our layered aluminophosphate is the fact that the two template molecules between the two sheets appear to be mostly held together by van der Waals forces, a feature that will be discussed in greater detail in the computational study reported later. We note that the SEM image (Figure 3) shows the material’s crystals to be of a plate-like nature. The inorganic layer of the material is made up of octahedral aluminum atoms, which are connected to each other through bridging oxygens, forming cis-chains. All octahedra are interconnected through edge-sharing, giving rise to a zigzag arrangement lying along the (010) axis. This AlPO material thus contains Al-O-Al linkages. However, the aluminum is in an octahedral arrangement and thus does not violate Lowenstein’s rule, which forbids Al-O-Al bonds systems with tetrahedrally coordinated aluminum. Other layered materials are known with this type of octahedral Al linkage,9,27,29,31,32 which is, however, an unusual
(22) Saadoune, I.; Catlow, C. R. A.; Doll, K.; Cora, F. Mol. Simul. 2004, 30, 607. (23) Ramaswamy, V.; McCusker, L. B.; Baerlocher, C. Microporous Mesoporous Mater. 1999, 31, 1. (24) Dauber-Osguthorpe, P.; Roberts, V. A.; Dauber-Osguthorpe, J.; Wolff, J.; Genest, M.; Hagler, A. T. Proteins 1988, 4, 31. (25) Beale, A. M.; Sankar, G.; Catlow, C. R. A.; Anderson, P. A.; Green, T. L. Phys. Chem. Chem. Phys. 2005, 7, 1856. (26) Gomez-Hortiguela, L.; Cora, F.; Catlow, C. R. A.; Perez-Pariente, J. J. Am. Chem. Soc. 2004, 126, 12097.
(27) Kongshaug, K. O.; Fjellvag, H.; Lillerud, K. P. Microporous Mesoporous Mater. 2000, 38, 311. (28) Tuel, A.; Lorentz, C.; Gramlich, V.; Baerlocher, C. J. Solid State Chem. 2005, 178, 2322. (29) Tuel, A.; Gramlich, V.; Baerlocher, C. Microporous Mesoporous Mater. 2001, 47, 217. (30) Lakiss, L. S.-M. A. G. V. P. J. Solid State Sci. 2005. (31) Huang, Q.; Hwu, S. J. Chem. Commun. 1999, 2343. (32) Mali, G.; Meden, A.; Ristic, A.; Tusar, N. N.; Kaucic, V. J. Phys. Chem. B 2002, 106, 63.
[AlPO4(OH)](NH3C2H4C6H5)
Chem. Mater., Vol. 19, No. 9, 2007 2265
Figure 5. Layers of the layered aluminophosphate material viewed (a) along the [100] axis showing octahedral aluminum chains interlinked by tetrahedral phosphate groups and (b) along the [010] axis showing the tetrahedral phosphate groups above and below the aluminum chains.
feature for AlPO materials that generally have alternating aluminum and phosphorus tetrahedra and avoid Al-O-Al bridges. Within the layers, the Al-O-Al chains are interconnected to each other by tetrahedral phosphate groups above and below the (100) plane, which in turn are linked to three different aluminum octahedra. The fourth oxygen on the phosphate tetrahedron points outward from the layer and is hydrogen bonded to the protonated amine group of the SDA. The phosphate groups running along the c-axis are in alternating arrangements, giving rise to a criss-cross pattern. These arrangements differ from those in the mineral tancoite, where the aluminum octahedra of the chains are also surrounded by four phosphate groups, which are not exactly above each other but staggered; here, the phosphate groups lie exactly in the same position above and below the alumina chains (i.e., along the a-axis), but adopt a staggered arrangement (Figure 5). Because of this arrangement of terminal phosphate groups, it may not have been possible to form three-dimensional structures, unlike the case with MIL-12, which is an intermediate in the formation of ULM-4.33 In the aluminum octahedron, distances range from 1.835(7) to 2.057(9) Å for Al(1) and from 1.826(3) to 1.958(2) Å for Al(2). The mean values (1.914(9) and 1.902(6) Å for Al(1) and Al(2), respectively) are in good agreement with previously reported bond lengths for octahedral Al-O bonds.27 The P-O distances range from 1.502(0) to 1.580(3) Å with an average distance of 1.540(7) Å, which is close to the bond lengths reported in similar compounds.9 The presence of octahedral aluminum was confirmed by decoupled 27Al solid-state magic angle spinning NMR, giving rise to a peak at -7.93 ppm. The arrangement of aluminum octahedra and tetrahedral phosphate groups gives rise to an unusual arrangement of three-, four-, and six-membered rings (MR) (Figure 5) in the inorganic layer. Pseudo 2 MRs are made of two aluminum octahedra linked together by two oxygen atoms, i.e., giving rise to the edge-sharing between the octahedra. Aluminum octahedra with a tetrahedral phosphate group located just above the pseudo 2 MRs then give rise to the 3 MRs. The 4 MRs are made up of 2 octahedral aluminum atoms in two different chains connected trans to tetrahedral phosphate groups, which bind the aluminum chains together. Four aluminum octahedra and two phosphate tetrahedra make up the 6 MRs, which lie along the z-axis. (33) Gerardin, C.; Loiseau, T.; Ferey, G.; Taulelle, F.; Navrotsky, A. Chem. Mater. 2002, 14, 3181.
Figure 6. XRD pattern of layered material solved using the crystal data obtained through single-crystal X-ray diffraction (wRp ) 9.7% and Rp ) 7.7%).
Figure 7. XRD pattern of materials synthesized with (a) N-methylaniline and (b) phenylpropylamine.
As noted earlier, the powder-XRD pattern of the sample collected on Station 2.3 at the SRS, was analyzed using GSAS Rietveld software to show that the crystal structure observed is representative of the bulk material synthesized rather than an impurity. There was a very good match between the single-crystal data and the powder data (Figure 6). It is interesting to note that the other templates (Nmethylaniline, benzylamine, and phenylpropylamine) did not yield crystals of sufficient size to collect single-crystal data (Figure 7). Moreover, N-methylaniline did not yield a phasepure sample but produced large crystal impurities of the Gismondine34 and the SBS structure35 that were analyzed using single-crystal X-ray diffraction. The powder XRD
2266 Chem. Mater., Vol. 19, No. 9, 2007
Dorner et al.
Figure 9. Optimized structure shown as polyhedra with proton residing on the (a) O1 site and (b) O5 site.
Figure 8. XRD patterns of layered compound synthesized with (a) 1-phenylethylamine and (b) benzylamine.
pattern of the material synthesized with benzylamine shows that it is very similar to the layered material formed by 1-phenylethylamine (Figure 8). Computational Analysis. The computational study aimed to establish the position of the proton within the layer and the stability of the overall material. Out of the five possible protonation sites available in the layer, i.e., the five symmetry-unique O ions, only two are found to be local minima in the potential energy surface (PES). These are the O ions labeled as O1 and O5 in Figure 1 and are the two oxygen atoms bridging the Al ions in the Al-Al edge-sharing units. The oxygen ion labeled as O5 is coordinated to only two aluminum ions; O1 however, is coordinated to 3 ions, two Al, and one P ion in the layer. The oxygens labeled 2, 3, and 4, which are 1- and 2-coordinated, belong to Al-O-P units and the terminal PdO group, and are not involved in the Al-Al edge-sharing linkage. A proton initially located on one of the latter oxygen types moves to O5 upon geometry optimization, indicating that there is no energy barrier for these proton jumps. Examination of the relative energies of the protonated layers, after optimization, reveals that protonation of sites O1 and O5 incurs at a noticeable energetic difference, calculated as 1.52 eV per proton in favor of O5. Although O1 is a local minimum in the PES, protonation of this oxygen is unlikely to occur because of its relatively high energy. O5, that is the oxygen bonded only to the two aluminum ions, is therefore expected to be protonated on the basis of the QM results. The geometry-optimized structures (panels a and b of Figure 9) of the AlPO layer obtained upon protonation of O1 and O5 show major differences. It is important to first notice that the phosphorus atoms of the structure remained tetrahedrally coordinated to four oxygens throughout our computational study. This result is consistent with earlier work25 indicating the molecular ionic nature of AlPO materials. All major differences observed in the optimized structures refer to the local environment around the Al ions. The structure presented in Figure 10a corresponds to the optimized layer when O1 is protonated. It bears no resemblance to the experimentally determined one. In fact a structural reconstruction of the AlPO layer occurs, with half (34) Baerlocher, C.; Meier, W. M. HelV. Chim. Acta 1970, 53, 2080. (35) Bu, X. H.; Feng, P. Y.; Stucky, G. D. Science 1997, 278, 2080.
Figure 10. Comparison of optimized structures with proton bonded to the (top) O1 site, (middle) O5 site, and (bottom) experimentally obtained picture.
of the Al ions being 4-coordinated in a tetrahedral geometry. Generally, this is the stable environment for Al in 3D AlPO frameworks, but not in the case of this layered material. The other half of the Al ions in the system are 6-coordinated, forming AlO6 octahedra each with two long and four short Al-O bonds. The structure obtained when O5 is protonated yields a very good match with experimental results in both coordination numbers and bond distances (strcutures b and c in Figure 10). Upon closer examination, we find that protonation of O5 gives uniquely octahedrally coordinated Al where pro-
[AlPO4(OH)](NH3C2H4C6H5)
Chem. Mater., Vol. 19, No. 9, 2007 2267
Figure 11. Interaction energy between adjacent layers as a function of interlayer separation.
tonation of the O ions in each of the AlO6 units occurs along vertices situated trans with respect to each other. The optimized values of all bond lengths in the layer are within 1% difference to the experimentally deduced values. As these differences are extremely small, we can be confident in the reliability of the experimental structural refinement, the calculated structure itself, and the suitability of the B3LYP functional to model AlPOs. The combination of results discussed above clearly indicates that only one protonation site is present in the layer, i.e., the oxygen ion labeled O5. In practice, because this oxygen is always protonated, it should be described as part of a hydroxide ion. Anchoring of the hydroxide ion to the layer is necessary to charge-balance the positive charge of the SDAs residing in the interlayer region and to yield the experimentally observed octahedral coordination of Al. The material can be considered to be an AlPO4 structure with a hydroxyl group that acts as a thermodynamic stabilizer incorporated into the framework. In the past, hydroxyl and fluoride groups have been shown to act as stabilizers in the synthesis of zeolitic materials.36-38 In some cases (i.e., the NON and STT structures), the fluoride anions are directly bonded to one silicon T atom and are thus integrated into the framework.39 Within the group of layered aluminophosphates, some structures such as MIL-12 can only be synthesized in the presence of fluorine, which is often incorporated into the structure and stabilizes the material in the process.9 We therefore consider that in the case of this layered AlPO, the hydroxide group anchored to the layers acts as a thermodynamic stabilizer for the structure and therefore plays a vital role in the construction of the layers. The protonated SDA (1-phenylethylamine) acts not only as a space filler within the structure but also as a hydrogenbonding donor to the framework oxygens. The inorganic sheets are being held together by H-bonding between oxygen atoms of the framework and hydrogen atoms attached to the nitrogen from the SDA, which stabilizes the structure. The presence of the hydroxyl group, O5H, is necessary to chargebalance the protonated SDA. (36) Millini, R.; Perego, G.; Berti, D.; Parker, W. O.; Carati, A.; Bellussi, G. Microporous Mesoporous Mater. 2000, 35-6, 387. (37) Salehirad, F.; Aghabozorg, H. R.; Manoochehri, M.; Aghabozorg, H. Catal. Commun. 2004, 5, 359. (38) Liu, Z. Q.; Xu, W. G.; Yang, G. D.; Xu, R. R. Microporous Mesoporous Mater. 1998, 22, 33. (39) Camblor, M. A.; Villaescusa, L. A.; Diaz-Cabanas, M. J. Top. Catal. 1999, 9, 59.
Figure 12. (a) Optimized charge neutral AlPO layer when symmetry is present; (b) the optimized structure obtained removing all symmetry constraints.
The second part of our computational work focused on the role that the SDA molecules play in holding the AlPO layers together. Our MD calculations, performed on the protonated layers at full loading of SDAs and at the experimental value of the interlayer separation, indicate that the templates are stable within the channels. No disorder and/ or reconstruction of the SDA molecules is observed after 100 ps of MD simulation at 423 K. This result indicates that the SDAs are tightly packed within the interlayer region of the AlPO. It also suggests that the removal of the SDAs during calcination results in the creation of voids, leading to the collapse of the AlPO framework. Figure 11 shows the relative energy (calculated using eq 1) as a function of the interlayer separation. Our results indicate that the optimal interaction energy between two layers in the bulk material amounts to 26.1 kcal/mol per unit cell (comprised of 2 layers) and corresponds to the lattice parameter a ) 38.3 Å, which is in reasonable agreement with the experimentally derived value of 39.06 Å. The calculated interlayer binding corresponding to this value of a, is -0.353 J/m2, a small, yet significant value. For all values of a, the SDAs remain close to the layers. Specifically, the ammonium end of each 1-phenylethylamine SDA is strongly bound to the negatively charged inorganic layers, even at large layer separations (Figure 13). A combination of nonbonded coulombic and dispersive interactions is therefore responsible for holding the SDAs and AlPO-layers together. Finally, we attempted to model some of the effects of calcination on the structure using the same quantum mechanical techniques described earlier. One of the effects of calcination on the AlPO-SDA adduct is the removal of the SDA and other nonframework ions, except those needed for
2268 Chem. Mater., Vol. 19, No. 9, 2007
Figure 13. Structure of the layers at large separation, showing 1-phenylethylamine’s affinity to remain bonded to the inorganic layers.
charge-balance. The QM calculations designed to examine the effects of calcination focused on hypothetical systems of one layer thickness; in our initial examination, the O5Hgroup and all extraframework ions were removed, leaving a slab of composition AlPO4. Full geometry optimizations were performed on the slab (1) retaining all symmetry operators of the original AlPO5H-SDA system, and (2) with all symmetry constraints removed. The structures obtained after full geometry optimization are shown in Figure 12. In the resulting calculated AlPO structures, we see that the one-dimensional chains of AlO6 octahedra present in the as-synthesized material are no longer present and are replaced by a chess-board-like arrangement of the Al ions in which the coordination polyhedra around Al are connected by corner-sharing only. The PO4 tetrahedra are located above and below the plane but cover only half of the empty squares of the chess-board pattern. In practice, this geometry defines dense one-dimensional chains in the structure, of composition [AlP2O8]3-, bonded via AlOx units (x ) 4 or 6) in the interchain region. Al-O2-P edge sharing, which is an unstable feature in AlPOs is found for calculations both with and without symmetry constraints. For the case, in which the geometry optimization was performed with full symmetry constraints, shown in Figure 12a, the one-dimensional [AlP2O4]3- chains consist of 6-coordinated AlO6 octahedra and 4-coordinated PO4 tetrahedra. These chains are held together by 4-coordinated square planar AlO4 units in the interchain region. The composition ratio of Al in 4- and 6-coordination is 1. The presence of AlO4 units in square planar coordination suggests that this structure is unstable, and that Al in this geometry can be easily solvated and possibly removed from the interchain region; therefore, we do not expect this phase to be stable under harsh experimental conditions, such as those present during calcination. Performing a full geometry optimization on the layer with all symmetry constraints removed has instead predicted a more stable structure, which is shown in Figure 12b, in which the [AlP2O8]3- one-dimensional chains consist solely of AlO4 and PO4 tetrahedra and are held together by AlO6 octahedra in the interchain region. Akin to the previous calculation, the ratio of 4- and 6-coordinated Al ions is found to be 1, but now the Al ions linking the AlP2O8 chains are in a stable 6-coordinated octahedral environment. The differences in the geometry of the Al ions in the structures examined explains the energetic difference, calculated as 0.13 eV/unit cell in favor of the structure obtained in the absence of symmetry. This result confirms the instability of the square planar AlO4 units in comparison to the tetrahedral AlO4 geometry. The more energy favorable conformation of the charge-neutral
Dorner et al.
AlPO layer, predicted in the absence of any symmetry constraints, shows a major change in the bonding between ions of the framework. Specifically, we find that the AlO-Al zigzag linkage reconstructs and forms a linear chain of corner sharing AlO4 and PO4 tetrahedra that are held by AlO6 octahedra in the interchain space. The latter AlPO4 structure is very unstable; its energy difference with respect to the stable AlPO4 polymorph, i.e., Berlinite, is calculated as 2.07 eV per AlPO4 formula unit. This value should be compared with the relative energy with respect to Berlinite of other stable microporous modifications, which at the B3LYP level is calculated in a range of 0.1-0.2 eV per formula unit.20 The energetic instability of the calcined layered structure, as well as the reconstruction of the atoms in the layers discussed above, explain why the layered AlPO collapses upon calcination; the one-dimensional chains formed are unable to maintain the integrity of the layers. These results indicate that the presence of charged species, the O5H- ions in the layer and the ammonium templates (i.e., the protonated 1-phenylethylammonium ions during synthesis) in the interlayer region are essential to the integrity and structural stability of the new layered AlPO described here. Summary and Conclusion The utilization of phenyl- and benzyl-based templates has led to the formation of a layered aluminophosphate. The structure consists of octahedral aluminum chains, which are interlinked by phosphorus tetrahedra into a 2D layered structure. Two template molecules separate subsequent inorganic layers, leading to the largest known d-spacing observed between layers in lamellar aluminophosphate materials. The two template molecules are held together mainly by van der Waals interactions, resulting in a stable material. Our calculations reveal that the synthesis of this AlPO is only possible because of the presence of the hydroxyl (O5H)- ions. These ions are primarily needed first to charge-balance the positive SDAs in the interlayer region and second to coordinate the otherwise 4-coordinated Al forming octahedral Al, thus acting as a thermodynamic stabilizer to the formation of the layered structure. Upon modeling the effects of calcination, our calculations predict that removal of charged extraframework species, namely the O5H- group and the SDA, leads to structural reconstructions of the AlPO layer, in which the zigzag AlO6 chains transform into a 1-dimensional [AlP2O8]3- conformation, which is thermodynamically unstable. Acknowledgment. We acknowledge Saudi Aramco and EPSRC for financial support. The authors thank Professor Gopinathan Sankar, Dr. Simon Teat, Dr. Scott Woodley, and Dr. John Warren for useful contributions and discussions. Supporting Information Available: Crystallographic information in CIF format; table of bond lengths and angles in PDF format. This material is available free of charge via the Internet at http://pubs.acs.org. CM070106U
