J O U R N A L O F C H E M I S T R Y Materials On sorption and thermal properties of the zirconium phosphate fluoride [(H2en)0.5][Zr2(PO4)2(HPO4)F] H2O Michael Feist,a Martin Wloka,a Matthias Epple,b Ekkehard Postc and Erhard Kemnitz*a aInstitut fu� r Chemie der Humboldt Universita� t zu Berlin, Hessische Str. 1–2, D-10115 Berlin, Germany bInstitut fu� r Anorganische und Angewandte Chemie der Universita� t Hamburg, Germany cNetzsch Gera�tebau GmbH, Selb, Germany The thermal behavior of the title compound (ZrPO-1) has been investigated using conventional thermal analysis techniques as well as coupled TG–MS, TPD and XRD. Owing to the channel structure, ZrPO-1 reveals zeolitic properties in the dehydration range (up to 200 °C) whereas the removal of the organic template above 400 °C results in a complete loss of the sorptive activity.Water may be substituted by ammonia which is then released in two well separated steps at ca. 150 and 300 °C, respectively. The stronger diVerentiation of the water sites by the ammonia sorption allows attribution of the desorption ranges to the channels and the cavities, respectively, that exist in the structure.X-Ray characterization of the product after removal of the template above 400 °C did not yield any known zirconium phosphate or oxide. EXAFS measurements show that no significant changes occur in the short-range order around zirconium neither by dehydration nor by annealing. However, the environment of zirconium becomes increasingly disordered at higher temperatures. Recently, we reported the crystal structure of matically shown in Fig. 2, the cation is disordered between two positions around an inversion center, which results in an [(H2en)0.5][Zr2(PO4)2(HPO4)F] H2O which was designated ZrPO-1.1 It represents the first zirconium homologue to the empty space remaining in the channel, whereas the nitrogen atoms adopt identical positions. They are linked to the fluorine well known microporous aluminium phosphate zeolites, (AlPO4-n).2 The structures of further representatives of the atoms of the ZrO5F octahedra via hydrogen bridges.Fig. 1 further demonstrates that the water molecules occupy two ZrPO-n type which have been obtained with diVerent templates will be described elsewhere.3 crystallographically diVerent sites. For each orientation of the template cation, the remaining empty space is occupied by one The structure of ZrPO-1 contains a three-dimensional arrangement of zirconium octahedra (one ZrO6 and one water molecule, which is disordered as well (occupancy factor 0.5).Another water molecule being situated on the twofold ZrO5F), and phosphate tetrahedra [one PO3(OH) and two PO4] being connected via common oxygen atoms.Both fluor- axis is not disordered (occupancy factor 0.5 as well). Within a small cavity, it is tetrahedrally linked to the inorganic frame ine atoms and OH groups are terminal and, therefore, do not participate in the polyhedra connectivity. The eight-membered via four hydrogen bridges: two times with a donor, and two times with an acceptor function. cyclic arrangement of alternating octahedra and tetrahedra forms holes with a diameter of approximately 6.5 A° .Channels Our study of the thermal behavior of ZrPO-1, assumed to exhibit zeolitic properties, has been directed by the following in the y-direction exist in the structure, that form bottlenecks owing to the eight-membered polyhedra arrangement (Fig. 1). aspects. First, the dehydration and rehydration processes and their reversibility with special regard to the two diVerent The organic template, i.e.the ethylenediammonium cation, occupies these channels between two bottlenecks. As sche- bonding situations of the water; secondly, the possible substitution of water in the channel by other molecules of comparable size, and, thirdly, the chemical and structural changes that are caused by the thermally induced removal of the template.Fig. 1 Crystal structure of [(H2en)0.5][Zr2(PO4)2(HPO4)F] H2O (ZrPO-1). Half of the water molecules are situated in the tetragonal cavities (shadowed circles), the other half in the eight-membered ring Fig. 2 Schematic view of the disordering of the ethylenediammonium channels. Owing to the disordering of the template cations, which are not shown here for better legibility (see Fig. 2), the water molecules in cation in the channels of ZrPO-1 (view in y direction; dotted circles: oxygen positions of the disordered water molecule) the channels are disordered as well (empty circles, both positions shown). J. Mater. Chem., 1998, 8(2), 433–438 433Experimental Synthesis The preparation of ZrPO-1, which is described in more detail elsewhere,1 has been performed by hydrothermal synthesis (180 °C) from an aqueous solution of ZrOCl2·8H2O, ethylenediamine monohydrate, 40% HF and H3PO4.Thermal investigations We have used a conventional DTA–TG apparatus (Netzsch STA 429) as well as a TG–MS coupling (Netzsch STA 409/skimmer system).4 The macro sample holder system with platinum crucibles was used (Pt/PtRh10 thermocouples; a- Al2O3 as reference; 100 ml Ar min-1 for the DTA–TG, 100 ml He min-1 for the TG–MS measurements; b=5 K min-1; sample mass ca. 100 mg). For the enthalpimetric evaluation of the DTA curves (maximal precision of 10–12%) the measuring system was calibrated according to literature recommendations. 5,6 The TPD studies were performed with a Perkin Elmer FTIR spectrometer PE 2000 (TGS detector, resolution 4 cm-1, 70ml N2 min-1).A mixture of N2 : NH3 (ca. 351) was used for the NH3 sorption. In order to avoid additional surface NH3 sorption, every loading was followed by one hour flushing in pure N2. For the X-ray characterization, a powder diVractometer XRD 7 (SeiVert PM, Freiberg, Germany) was used. Fig. 3 STA curves for the dehydration and the template step: (a) of EXAFS measurements non-treated ZrPO-1; (b) of a dehydrated (200 °C) sample followed by NH3 sorption (cf. Table 1); (c) NH3 TPD curve of sample (b) X-Ray absorption fine-structure spectroscopy (EXAFS) was carried out at the Hamburger Synchrotronstrahlungslabor (HASYLAB) at Deutsches Elektronen-Synchrotron (DESY), In accordance with the idea of diVerent water sites (Fig. 1), Hamburg, at beamline X (ROEMO II). The DORIS III the ion current (I.C.) curves for m/z 17 and 18 reveal a storage ring was operated at 4.5 GeV positron energy and discontinuous shape (Fig. 4) whereas the DTA curves do not currents of 50–120 mA. The incoming synchrotron beam was show any shoulder or double peak [Fig. 3(a)]. Additional monochromatized by a Si(311) double-crystal monochromainformation on the dehydration step is obtained from the DTA tor.Experiments were performed at the Zr K-edge (ca. and DTG traces in Fig. 5, which correspond to the manifold 17 999 eV) in transmission mode at room temperature. The sample treatment described in Table 1. The reversibility of the ground samples were thoroughly mixed with boron nitride dehydration as well as the ammonia sorption properties of the and pressed to a pellet.For quantitative data evaluation we dehydrated phase clearly reveal the zeolitic character of ZrPO- used the programs AUTOBK and FEFFIT of the University 1. Upon rehydration, the original bonding situation of water ofWashington package.7 Theoretical standards were computed is recovered. This may be deduced from the X-ray powder with the program FEFF 5.048 using the crystal structure of pattern (see next section) as well as from the identity of signal [(H2en)0.5][Zr2(PO4)2(HPO4)F] H2O.1 All fits were carried shape and area (DdehydH ca. 51 kJ mol-1) for the first to third out with k3-weighted data using a k-range of 2–14 A ° -1 and an dehydration steps (cf.Table 1 for details of the experimental R-range of 0.6–6.6 A° .The amplitude reduction factor S02 was procedure). The slightly enhanced mass gain upon rehydration fixed to 1.22 as this value was obtained from fits of the pure may be explained by the higher sorption activity of the educt compound with constant coordination number. One primarily dehydrated phase. When it is reheated without prezero- rection [E0 =+12.1(6) eV] was chosen for the flushing in argon, this surface adsorbed water is released, first four shells (6 O/F, 5.5 P, 6.125 O, 2.75 O) and another causing a small separate eVect [Fig. 5(b)], whereas the true [E0=-2.8(5) eV] for the zirconium shell at 6.62 A ° . In order dehydration eVect has the same shape and area as for the first to reduce the number of fit parameters, the same Debye– heating.On the contrary, if a rehydrated sample is exposed to Waller factor was chosen for the three oxygen shells. a dry argon flow (20 h, 25 °C), the adsorbed water is released, and the pre-eVect does not occur upon heating [Fig. 5(c)]. Results and Discussion The interpretation of the second endothermal reaction range turned out to be more complicated even with the use of Thermoanalytical and absorption studies TG–MS measurements.The curves in Fig. 4 show that the template removal step is primarily caused by the release of Two well separated endothermal eVects, correlated to mass loss steps, characterize the thermal behavior of ZrPO-1 under carbon-containing species from the ethylenediammonium cation. Taking into account earlier findings on the thermal normal pressure in argon [Fig. 3(a)]. With respect to the above described structural features, the first mass loss is attributed behavior of the related triethylenediammonium compounds, e.g. of chlorometalates like (H2trien)[CoCl4],9 several mass to dehydration (up to 200 °C) where the observed mass loss of 3.5% corresponds well to the value calculated for 1 mol H2O spectral peaks may be readily attributed to fragments of the template cation.However, the whole decomposition process, (3.36%). Consequently, the second step (350–450 °C) is due to the removal of the template but, looking at the low mass loss unfortunately, is rather unspecific because several processes are overlaying each other in this temperature range. This of 3.5% (Dmcalc 9.35% for 0.5 en·2HF, or 5.62% for 0.5 en) and the curve shape (no plateau and not complete at 450 °C), produces the quasi-continuum of higher mass peaks (cf.Fig. 4) and does not allow postulation of any main reaction for it is rather incomplete or, at least, more complicated than a simple evaporation process. the decay. 434 J. Mater. Chem., 1998, 8(2), 433–438Fig. 5 DTA and DTG curves for the dehydration range of ZrPO-1 for repeated dehydration/rehydration cycles corresponding to the stepwise sample treatment described in Table 1: (a) first dehydration (step 1); (b) second dehydration (step 4); (c) third dehydration (step 8) The disappearance of the zeolitic properties of ZrPO-1 is expressed by the fact that neither water nor ammonia may be absorbed by the product phase which is formed during the template removal step (Table 1).The small amount of water taken up at 25 °C subsequent to the template removal step is only surface adsorbed. Therefore, the small mass loss of ca. 1% that appears on reheating has no corresponding DTA signal. An identically treated sample, cooled in ammonia, does Fig. 4 TG–MS curves of ZrPO-1 (ion current curves for m/z not show any mass loss upon reheating under argon. 17,18,19,20 and mass number scan at the beginning of the template step) A completely diVerent situation is found when ZrPO-1 is only dehydrated (200 °C) and is then cooled to room temperature in an ammonia flow. Upon reheating, two well separated The template removal step consists of a first partial step with bigger substance flow (340–450 °C) and a remarkably TG steps [Fig. 3(b),(c)] indicate diVerently bound NH3 molecules which have been absorbed by the host structure retarded following step (up to 700 °C), and is not completely finished even above 900 °C. Under these dynamic conditions, (DmNH3#DmH2O !), possibly onto the positions that were originally occupied by the two diVerent types of water molecules.either the organic template is partially retained by the solid phase or it reacts secondarily with the zirconium phosphate With NH3, the second desorption maximum is observed at a much higher temperature (DT ca. 150 K) than the first one matrix to yield further volatile products. Obviously, no massconstant phase, e.g. related to a zirconium oxide, is formed which is in the dehydration range.This shift is too large for the assumption that only variations in the hydrogen bonding here as might have been expected. Nevertheless, some further qualitative details may be would diVerentiate the water sites, now being substituted by ammonia molecules. The changes caused by the ammonia deduced from the TG–MS data. First, it may be seen that water is released not only in the dehydration step but is also sorption seem to be more profound to rationalize the considerable shift of the second desorption peak.One possible reason formed during the template decomposition. This may be explained by condensation reactions of two neighbouring is the formation of NH4+ ions, which could explain a stronger fixation of ammonia in the solid via additional coulomb PMOH groups.Secondly, the formation of ammonia in the second TG step is observed when considering the I.C. intensity interactions rather than just hydrogen bonding. The formation of ammonium ions is possible if one takes into account the ratio for m/z 17518. The mass loss corresponds to that expected only for water in the first TG step but, owing to a certain terminal OH groups existing in the structure of ZrPO-1.Owing to the greater basicity of ammonia relative to water the OH contribution by ammonia, it is much higher in the second step. Thirdly, HF liberation (m/z 19, 20) takes place predominantly groups can act as proton donors to form NH4+ ions. Thus, the two desorption maxima could be interpreted in in the template removal step.That means that not only the organic component is removed from the channels but the terms of, first, hydrogen bound NH3 molecules, which would be quite analogous to the situation of water in the structure zirconium phosphate matrix is attacked as well and the sorptive activity of the Zr–P–O phase(s) disappears. The small amounts of ZrPO-1. Secondly, NH4+ ions, which are formed via reaction between OH groups and NH3, would be more strongly retained of HF detected in the first TG step were attributed to surface adsorbed HF which is rationalized by the conditions of the in the structure and explain the liberation of NH3 at higher temperatures.That means that we are able to attribute the hydrothermal synthesis. J. Mater. Chem., 1998, 8(2), 433–438 435Table 1 Procedure of sample treatment for three diVerent samples of ZrPO-1a step process temperature programb Dm (%) 1 dehydration (200 °C in Ar -3.6 2 cooling 325 °C in Ar — 3 rehydration 12 h at 25 °C in moist air +4.0 4 dehydration (200 °C in Ar -0.6 and -3.3 5 cooling 325 °C in Ar — 6 rehydration 43 h at 25 °C in moist air +4.0 7 dry gas flow 20 h at 25 °C 0 (!) 8 dehydration and further heating (460 °C in Ar -3.5 (H2O) and -3.5 (template) 9 cooling 325 °C in Ar — 10 rehydration 20 h at 25 °C in moist air +1.0 11 heating (460 °C in Ar -1.0 1 dehydration (200 °C in Ar -3.6 2 NH3 absorption 325 °C in NH3 +2.9 3 desorption (370 °C in Ar (TPD) -1.2 and -1.6 1 dehydration and further heating ( 460 °C in Ar -3.5 (H2O) and -3.5 (template) 2 NH3 absorption 3180 °C in Ar followed by 325 °C in NH3 — 3 heating (200 °C in Ar 0 (!) aSteps 1–11 performed in a single experimental run on the thermobalance, the ammonia sorption in a preparative scale followed by the STA measurement.b(=heating, 3=cooling. first desorption peak to channel sites whereas the second peak (O, N, C) are present. Furthermore, there are zirconium neighbors at 4.85 A ° (N=1), 5.1–5.25 A ° (N=3), 5.45 A ° (N= corresponds to cavity sites. 1.5) and 6.62 A ° (N=4). It was not possible to separate fluorine and oxygen in the Structural investigations first neighboring shell owing to almost identical backscattering The structural changes caused by the dehydration yield powder behavior and distance. Therefore we combined the fluorine diagrams (Fig. 6) which are quite similar to the educt phase, shell and the first oxygen shell into one oxygen shell with N= but not so readily assigned. However it can be deduced that 6 at 1.99–2.10 A ° .The phosphorus atoms were combined into the rehydrated phase is identical with the educt state. a single shell. The next oxygen shells (N=6.125 and 2.75) were Meanwhile, we succeeded in preparing a single crystal of the also fitted.However, it should be kept in mind that higher dehydrated phase for a crystal structure analysis, the evaluation shells are influencing the third oxygen shell. Additionally, of which is underway. The following template step, as already multiple scattering paths come into play above R=3.5 A °. As discussed above, eVects deeper chemical and structural changes an indicator for long-range order we used a distinct shell of of ZrPO-1 resulting in a completely diVerent powder pattern.four zirconium atoms at 6.62 A ° . No other shells were included Our attempts for an indexation or an interpretation in terms into the fit. of known crystalline phases existing in the Zr–O–P system As expected, the main structural framework did not change failed, which led us to investigate the structural changes by upon dehydration.Only small changes are visible even upon EXAFS measurements. annealing at 800 °C. This is seen from the close similarity of Two diVerent crystallographic positions exist for zirconium. EXAFS Fourier transforms as shown by visual inspection Surrounding shells were computed for both positions to derive (Fig. 7 and 8). Consequently, the coordination numbers were the average zirconium environment. Zirconium is surrounded fixed to their theoretical (crystallographic) values during the by 0.5 fluorine atom (1.99 A ° ), 5.5 oxygen atoms (2.02–2.10 A ° ), fits. The results of the fits (average of two EXAFS scans at 5.5 phosphorus atoms (3.34–3.55 A ° ), 6.125 oxygen atoms room temperature) are listed in Table 2.(3.81–4.02 A ° ) and 2.75 oxygen atoms (4.08–4.14 A ° ). Above Two main results can be summarized as follows: all five 4.14 A ° , a number of overlapping shells of light backscatterers shells are present in the four samples and the interatomic distances remain the same within the experimental scatter. Fig. 6 Calculated (a) and experimental powder diVractogram of ZrPO- Fig. 7 k3-weighted Zr K-edge EXAFS functions x(k) for ZrPO-1: (a) untreated, (b) dehydrated at 200 °C, (c) heated to 400 °C, and (d) heated 1 (b) compared with those of a dehydrated (200 °C) (c) and a rehydrated sample (d) to 800 °C, template removed 436 J. Mater. Chem., 1998, 8(2), 433–438thermal behavior of ZrPO-1 diVers from that of typical representatives of the neighbouring AlPO family, e.g.that of the VPI-5 type which represents a wide-pore 18-ring aluminium phosphate.10 Upon dehydration (120 °C !) it rapidly rearranges to another intact AlPO structure, AlPO-8, if the compound is prepared in the presence of dipropylamine. On the other hand, the VPI-5 structure remained stable if the hydrothermal synthesis was performed in the presence of dipentylamine.11 For the known ZrPOs, a template eVect governing the thermal stability of the metal phosphate fluoride framework is not observed with the main structural motif and reversible hydration/dehydration behavior remaining unchanged.Conclusions ZrPO-1 shows zeolite-like properties at low temperatures. This Fig. 8 Fourier-transform magnitudes of ZrPO-1 after diVerent thermal is evidenced by the conservation of the channel structure of treatments (cf.Fig. 6). The coordination of zirconium remains the zirconium phosphate framework upon dehydration, by its unchanged up to R#4 A° ; however, the disorder increases as visible reversibility, and by the possibility of substituting water by from the decreasing height of the Zr–O/F peak at 2.1 A ° . The Zr–Zr ammonia.On the other hand, a complete loss of the sorptive shell at 6.6 A ° almost vanishes upon heating to 800 °C, indicating a activity is observed when the template is thermally removed structural change. The data are not corrected for phase shifts. from the ZrPO-1 structure. The channel structure collapses but the short-range order of the Zr atoms remains nearly This illustrates that the short-range order around zirconium is unchanged which might indicate that the formed structure is not changed by dehydration and annealing.However, the very similar. However, this is not necessarily the case since the structural disorder, as represented by the Debye–Waller factors, short-range arrangement of layered zirconium phosphates is, increases, indicating a slow breakdown of the zeolitic structure. to some degree, similar to the short-range constitution of 3D This change is most obvious in the ZrMZr shell at 6.62 A ° .If ZrPOs.3 It is not clear as yet whether the product after removal the coordination number is fixed to four as in the original of the template represents a single phase system or a mixture structure, the Debye–Waller factor almost quadruples in the of various zirconium phosphates and/or oxides.Therefore, sample heated to 800 °C. If the Debye–Waller factor is kept further structural investigations of the detemplated phase constant at 2.6×10-3 A ° 2 and the coordination number is are required. varied, we obtain N=2.3 (400 °C) and N=0.6 (800 °C). The strong correlation between N and s2 prevents the exact determination of N, but it is quite clear that the environment of The support of Dr.W.-D. Emmerich (Netzsch Gera�tebau GmbH, Germany) who enabled the TG–MS investigations to zirconium becomes increasingly disordered at higher temperature. It was very diYcult to fit the phosphorus shell. The be carried out is gratefully acknowledged. We thank Larc Tro� ger (Hamburg) for experimental assistance, and we are ZrMP distance was found to be too large, and also an unreasonably high Debye–Waller factor was found. We ascribe grateful to HASYLAB at DESY for generous allocation of beamtime.The financial support by the Deutsche these problems to a strong correlation between the ZrMP shell and the neighboring ZrMO shell. Forschungsgemeinschaft (DFG) and the Fonds der Chemischen Industrie (FCI) is also gratefully acknowledged. Despite the structural similarity, the template eVect and the Table 2 Results of EXAFS experiments carried out at the zirconium K-edge for ZrPO-1 after diVerent thermal treatments (coordination numbers fixed to their crystallographic values for hydrated and dehydrated phases) sample treatment 400 °C 800 °C 25 °C 200 °C (beginning of (template (untreated) (dehydrated) template step) removed) 1st shell: 6 O/F Na 6 6 6 6 R/A ° 2.06(2) 2.08(2) 2.08(2) 2.07(2) s2/10-3A ° 2 4.1 5.2 5.4 7.2 2nd shell: 5.5 P Na 5.5 5.5 5.5 5.5 R/A ° 3.72(2) 3.85(2) 3.72(2) 3.66(2) s2/10-3 A ° 2 7.8 95.5 8.9 6.2 3rd shell: 6.125 O Na 6.125 6 6 6 R/A ° 3.74(2) 3.86(2) 3.74(2) 3.66(2) 4th shell: 2.75 O Na 2.75 2.5 2.5 2.5 R/A ° 4.06(2) 4.04(2) 4.05(2) 4.03(2) higher shell: 4 Zr Na 4 4 4 4 R/A ° 6.63(2) 6.62(2) 6.62(2) 6.63(2) s2/10-3A ° 2 2.6 2.9 4.8 10.5 aFixed.J. Mater. Chem., 1998, 8(2), 433–438 4377 E. A. Stern, M. Newville, B. Ravel, Y. Yacoby and D. Haskel, References Physica B, 1995, 208–209, 117. 1 E. Kemnitz, M. Wloka, S. Trojanov and A. Stiewe, Angew. Chem., 8 J. J. Rehr, R. C. Albers and S. I. Zabinsky, Phys. Rev. L ett., 1992, 1996, 108, 2809. 69, 3397. 2 S. T. Wilson, B. M. Lok, C. A. Messina, T. R. Cannan and 9 M. Feist, R. Kunze, D. Neubert, K. Witke and E. Kemnitz, E. M. Flanigan, J. Am. Chem. Soc., 1982, 104, 1146. J. T herm. Anal., 1997, 49, 635. 3 M. Wloka, S. I. Troyanov and E. Kemnitz, J. Solid State Chem., 10 M. E. Davis, C. Saldarriaga, C. Montes, J. M. Garces and in press. C. Crowder, Nature (L ondon), 1988, 331, 698. 4 T. Rampke, W. D. Emmerich, E. Post and L. Giersig, J. T herm. 11 J. O. Perez, P. Jen Chu and A. Clearfield, J. Phyhem., 1991, Anal., 1996, 47, 633. 95, 9994. 5 K. Heide, Dynamische T hermische Analysenmethoden, VEB Deutscher Verlag fu� r GrundstoYndustrie, Leipzig, 1982. 6 H. K. Cammenga, W. Eysel, E. Gmelin, W. Hemminger, G. W. Paper 7/05459I; Received 28th July, 1997 H. Ho�hne and S. M. Sarge, T hermochim. Acta , 1993, 219, 333. 438 J. Mater. Chem., 1998, 8(2), 433&ndash