Chen Tiehong^*,#, Wouters B H^#,  Grobet P J ^#
(*Department of Chemistry, Nankai Univerisity, Tianjin, 300071;  ^#Center for Surface Chemistry and Catalysis, Katholieke Universiteit Leuven, Kardinaal Mercierlaan 92, B-3001 Heverlee (Leuven), Belgium )

Received  Aug.14, 2000; Part of this work was supported by the National Natural Science Foundation of China (Grant No. 29803004)

Abstract  Molecular sieve SAPO-37 has been studied by ²⁷Al MQMAS and multi-nuclear solid state NMR methods. It is proved that in the dehydrated state, the aluminum signal of the Bronsted sites in SAPO-37 is visible, and tri-coordinated Al may cause intensity loss due to the quadrupolar broadening. When calcined at 1173 K, Si atoms become mobile and detach from the framework to aggregate to form polymorph silica, while an Al-P dense phase was formed, corresponding to a new Al signal which is distinguished by the ²⁷Al MQMAS spectrum for the first time. The move of atoms in SAPO-37 at 1173 K is the start of the collapse of the framework.
Keywords molecular sieve SAPO-37, ²⁷Al MQMAS NMR, polymorph silica

Silicoaluminophosphate molecular sieve SAPO-37, an isotype of faujasite, has been the subject of numerous researches on the study of mechanism of silicon substitution, the silicon environments and their catalytic properties ^[1-6]. A drastic change of the ²⁹Si MAS NMR spectrum in SAPO-37 after heating at 1173K was observed and a model of "solid state transformation" was proposed to explain the environment change of Si atoms calcined at the temperature between 1073-1173K in SAPO-37 ^[7,8]; this model involves the floating of atoms in the framework lattice to form the siliceous islands and silicoaluminate domains, while at the same time, the faujasite crystal structure is maintained. An increase of the catalytic activity of SAPO-37 calcined at 1173 K was explained by the newly formed silicoaluminate phase during the "solid state transformation".
    Although the ²⁹Si NMR spectrum shows different Si environments (Si with different numbers of Al as next nearest neighbors and Si islands), the 1D ²⁷Al MAS NMR spectra could not be well resolved due to the quadrupolar broadening of aluminum nucleus. ²⁷Al Multiple Quantum (MQ) MAS NMR method ^[9,10], which can refocus the second-order quadrupolar broadening, has been used to study the SAPO-37 and an enhanced resolution was obtained, and the Al environments with and without Si as the nearest neighbours has been distinguished ^[11]. Here by means of   ²⁹Si, ²⁷Al, ³¹P, ¹H MAS NMR, together with ²⁷Al MQ MAS NMR methods, we examined the thermal stability and structure change of the SAPO-37 after high temperature calcination.

1 EXPERIMENTAL   
A SAPO-37 sample was synthesized hydrothermally ^[12] with the composition of [Si₂₃Al₉₈P₇₈O₃₈₄]:TMA₈:TPA₁₂:64H₂O as determined by electron probe microanalysis and thermal analysis. The XRD spectrum demonstrates the sample is highly crystalline. The ²⁹Si MAS NMR spectrum shows isolated Si (pure SAPO phase) as well as a small amount of Si islands in this sample.
    The template-free sample was obtained by heating the as-synthesized sample at 873 K in a quartz tube in a flow of O₂ for 20 hrs, with temperature increment of 1 K/minute. And the samples were calcined at 1073 or 1173 K for 10 hrs in a quartz tube in vacuum respectively. Because H-SAPO-37 is very sensitive to water, after the thermal treatment the quartz tube was closed and transferred into a glove box with dry N₂ where a 4 mm rotor was packed.
    The ²⁷Al and ³¹P MAS NMR experiments were performed on a Bruker MSL-400 spectrometer at 104.26 MHz and 161.5 MHz, respectively, with a 4 mm ZrO₂ rotor spinning at 12.5 kHz. The ³¹P chemical shift is referenced to 85% H₃PO₄. The 1D ²⁷Al MAS NMR were performed with a pulse length of 0.6 m s corresponding to a p/18 pulse length and a recycle delay of 0.1 s. A two-pulse 3Q MAS method [10] was employed with pulse lengths of 2.7 and 0.8 m s respectively, a radio frequency field of ca. 120 kHz and a 0.5 s recycle time; 512 points were acquired in the t₁ dimension with increments of 6 m s; 480 scans were recorded per FID. Chemical shift of F2 dimension is referenced to an Al(NO₃)₃ solution.
    ¹H NMR was performed on a Bruker AMX-300 spectrometer at 300.13 MHz, with a spin rate of 10 kHz and 10s recycle delay with a p/2 pulse length of 3.5 ms. ²⁹Si NMR was performed also on a Bruker AMX-300 spectrometer at 79.5 MHz, with a spin rate of 4 kHz and 10 s recycle delay. The ¹H and ²⁹Si chemical shift was referenced to tetramethylsilane (TMS).

Fig. 1 ²⁹Si MAS NMR spectra of the (a) as-synthesized, (b) 873, (c) 1073, (d) 1173 K calcined SAPO-37

2 RESULTS AND DISCUSSION
The ²⁹Si MAS NMR spectra of as-synthesised, 873, 1073, 1173 K calcined SAPO-37 are displayed in Fig. 1. The -89 ppm in the as-synthesised sample is the isolated Si(4Al), and a small up-field hump corresponding to small amount of Si islands [2]. In the spectrum of the 873 K calcined samples (Fig 1b), the signal of isolated Si shifts to -92 ppm and become broader due to the burnt-out of the templates. There is no much difference displayed in the ²⁹Si NMR spectra of the 873 and 1073 K calcined samples (Fig.1b,c), though some dehydroxylation takes place in the later sample (as observed by the ¹H NMR measurement). The ²⁹Si MAS NMR spectrum of the 1173K calcined (Fig.1d) looks quite different, a broad hump emerges: a -115 ppm signal can be distinguished and some signals in between -115 and the -92 ppm, but could not be resolved due to their wide chemical shift dispersion. According to the literature ^{[8, 13]}, the -115 ppm signal can be assigned to the polymorph silica structure, for instance, the tridymite. This single can also be seen in the Peltre's results ^[7], in which the change of the ²⁹Si MAS NMR spectrum of the SAPO-37 calcined at 1173 K was reported. This newly formed structure can be identified by the XRD spectrum that an additional peak of tridymite, supporting the assignment. It is a clear evidence that when calcined at 1173 K, some topology phase transformation takes place in SAPO-37.

    The ¹H MAS NMR spectra were shown in Figure 2. The 1.8 ppm was assigned to the silanol group, and 3.7 and 4.2 ppm were assigned to the bridging hydroxyls in the supercages and sodalite cages, respectively ^[14]. By the comparison of the ¹H NMR signal intensities, it can be concluded that dehydroxylation takes place when calcined at 1073 K, and the hydroxyls in the supercage are detached firstly. When the sample is calcined at 1173 K, most of the hydroxyls are removed, and a signal at 2.6 ppm emerges which could be assigned to the extraframework Al-OH, indicating that some dealumination takes place at this high temperature.

    27Al MAS NMR spectra of the as-synthesised, 873, 1073, 1173 K calcined samples are displayed in Figure 3. In the study of zeolites, ²⁷Al NMR measurement is difficult to perform on the dehydrated samples due to the very strong quadrupolar interaction of the aluminum, and there is always intensity loss ^[15]. This it is not the case in the SAPO-37 system. By quantitative intensity comparison it is shown that there is no ²⁷Al NMR intensity loss in the spectrum of the 873 K calcined sample compared with the as-synthesized sample, i.e., in the dehydrated H-SAPO-37, the Al (nSi) environment is completely detectable. It is obviously due to the chemical composition of the SAPO system with the presence of the phosphorus in the balanced framework.

Fig. 4 ²⁷Al 2D 3Q MAS NMR spectra of the (a) 1073, (b) 1173 K calcined SAPO-37; ** indicate the spinning side bands, A,B,C see text.

    The ²⁷Al intensity of 1073 K calcined sample shows an intensity loss of 15%, due to the formation of the tri-coordinated Al by dehydroxylation. Although further dehydroxylation continues, it is very interesting to note that in the 1173 K calcined sample, the Al intensity is recovered to 95%, i. e., some invisible Al become visible again. This is hardly due to the recovery of the tri-coordinated Al to its original configuration because the dehydroxylation continues at higher temperature, but to some further configuration change of the"invisible" part of the Al. The peak position of ²⁷Al NMR line of the 873 and 1073 K (Fig 3b,c) calcined samples moves gradually up-field, probably due to the increasement of the quadrupolar interaction. For the 1173 K calcined sample (Fig 3d), the peak goes back downfield to 37 ppm, in agreement with Derewinski et al. ^[8], and it is postulated that some new Al species are emerging, whose environment is more symmetric (the quadrupolar effect is smaller ) to give rise to a downfield contribution beside the framework Al(4P) signal.
    To enhance the resolution, ²⁷Al triple quantum NMR method was employed here. It has been reported that in the as-synthesised SAPO-37, there are different Al coordinatons with different templates, and in the calcined sample, only two Al (Al(4P) and Al(Si, 3P)) were identified ^[11]. In Figure 4a, the ²⁷Al 3Q MAS of 1073 K calcined displayed only two parts, similar to the spectrum of the 873 K calcined one^[11], which could be assigned to the Al(4P) phase and Al(Si, 3P) environments ( A and B as indicated in the Figure 4, respectively). In the ²⁷Al 3Q MAS NMR spectrum of the 1173 K calcined sample (Fig 4b) it shows that a new Al phase emerging (indicated as C in the Figure) at 35.9 ppm on F2 and 35.6 ppm on F1, its quadrupolar constant could be calculated to be about 1 MHz by the formula in reference 16.

    This newly formed new Al phase could result in the chemical shift of the Al(4P) peak to lowfield and the intensity re-gain due to the reduction of the quadrupolar interaction. By its position this newly formed Al phase should also be tetrahedral coordinated. Due to its rather symmetric environment, it is either due to the Al with 4P or 4Si as the nearest neighbors. By our experiments, the 3Q MAS NMR spectra of the higher Si content SAPO-37, which contains the Al(4Si) structure units, did not resolve the Al(4Si) environment as a third signal. Thus we reasonably assign the newly formed Al phase to the Al with 4P environments; because it can be distinguished from the framework Al(4P) units of the SAPO-37, its configuration should be different, i.e., it can not be regarded as in the topology of the faujasite, but due to some dense phase.
    The ³¹P NMR spectra of the as-synthesized, 873, 1073, 1173 K calcined samples are showed in Figure 5. For the as-synthesised sample, a peak at around -26.2 ppm represents the P(4Al) environment, and small hump at -33 ppm can be assigned to the some defect sites in the framework. The spectra of the 873 and 1073 K calcined samples display narrower line width (300 Hz) of the -26.5 ppm P(4Al) signal than that of the as-synthesized sample (510 Hz), due to the removing of the templates. But the spectrum of the 1173 K calcined sample became broader again (460 Hz), and there appeared a broader hump; a wide dispersion of environment of the ³¹P NMR line of the 1173 K calcined sample, gives evidence that part of the P were involved in the framework change at high temperature, together with the change of the Al and the Si environment, as indicated by the NMR results.
    Combining the multi-nuclear NMR results above, it could be postulated that when calcined at 1173 K, some part of the Al-Si connection in the SAPO-37 framework break and new Al-P phase form, distinguished as a new separate signal by the ²⁷Al MQMAS spectrum; the broadening of the ³¹P NMR signal corresponds to this framework structure modification. Some Si may be detached from the lattice, as Al does in the zeolites framework, to recrystallize at high temperature as to form polymorphs.
    By our experiments, the SAPO-37 is not thermally stable as it was regarded. When calcined at 1173K, the framework atoms are in high mobility, and it could be the start stage of the collapse of the topology. The defect sites during the process of the framework change may be responsible for the catalytic properties at high temperature. More experiments on other SAPO systems are needed to see whether the "de-siliceous" effect is universal in the SAPO systems.

3 CONCLUSION  
Silicoaluminophosphate molecular sieve SAPO-37 has been studied with ²⁷Al MQMAS and multi-nuclear solid state NMR methods. It is proved that the Al of the Bronsted sites in SAPO-37 is visible, and tri-coordinated Al may cause intensity loss due to the quadrupolar interaction. When calcined at 1173 K, Si atoms become mobile and detach from the framework to aggregate to form polymorph silica, while Al-P dense phase was formed, corresponding to a new Al signal which can be distinguished by the ²⁷Al MQMAS spectrum. At 1173 K the new dense phase of Al-P form and the new Si signals can be explained as extraframework silica phases. The move of atoms in SAPO-37 is the start of the collapse of the framework.

REFERENCES    
[1] Lok B M, Messina C A, Patton R L et al. J.Am. Chem. Soc., 1984, 160: 6092.
[2] Martens J A, Janssens C, Grobet P J et al. Stud. Surf. Sci. Catal., 1989, 49A: 215.
[3] Blackwell C S, Patton R L. J. Phys. Chem., 1988, 92: 3965.
[4] Derewinski M, Briend M, Peltre M J et al. J.Phys.Chem., 1993, 97: 13370.
[5] Su B L, Barthomeuf D. J. Catal., 1993, 139: 81.
[6] Su B L, Barthomeuf D. Zeolites, 1993, 13: 626.
[7] Peltre M J, Man P P, Briend M et al. Catal. Lett., 1992,16: 123.
[8] Derewinski M, Peltre M J, Briend M et al. J.Chem.Soc., Faraday Trans., 1993, 89: 1823.
[9] Frydman L, Harwood J S. J. Am. Chem. Soc., 1995, 117: 5317.
[10] Fernandez C, Amoureux J.  Chem. Phys. Lett., 1995, 449: 242.
[11] Chen T H, Wouters B H, Grobet P J. Colloids and Surfaces, A, 1999, 158:145.
[12] Kovalakova M. Ph.D. thesis, Katholieke Universiteit Leuven, 1996.
[13] Briend M, Peltre M J, Massiani P et al. Stud. Surf. Sci. Catal., 1994, 84: 613.
[14] Hunger M, Anderson M W, Ojo A et al. Microporous Mater., 1993, 1: 17.
[15] Freude D, Ernst H, Wolf I, Solid State Nucl. Magn. Reson., 1994, 3: 271.
[16] Rocha J, Esculcas A, Fernandez C et al. J. Phys. Chem., 1996, 100: 17889.

[ Back ] [ Up ] [ Next ] Mirror Site in USA  Europe  China  CSTNet ChinaNet