Molecular Sieves - Science and Technology - Vol. 6 - Characterization II / 06-Isomorphous Substitution in Zeolites

Scheme 1 Schematic representation of the deformed aluminum site

synthesized by the methanol procedure. Figure 42 shows the 27Al MAS NMR spectra of the three different samples of TON zeolite confirming the absence of aluminum in octahedral coordination. A very low amount of octahedral aluminum is detected for the Cu-TON prepared by the solid-state reaction. We suppose that this small quantity of aluminum is due to the procedure of copper introduction.

There is a correlation between the amount of the copper introduced into the zeolite framework and the deformation of Al tetrahedral sites. This is also confirmed by chemical analysis. When Cu is over-exchanged by the solidstate reaction procedure, the low intensity NMR line at 0 ppm assigned to extra-framework aluminum is observed. In the case of the two solutionmediated ionic-exchange procedures, only slight deformation of the AlT sites is observed.

The nature of the deformed tetrahedral aluminum site is still a question of controversy. In many dealumination processes, this line (at ca. 30 ppm) has been attributed to the tetrahedral framework atoms where at least one Al–O– Si bond is broken. The 2D 3QMAS NMR data unambiguously show a chemical shift at ca. 70 ppm confirming its attribution. In the present case, we can hypothesize that one Al–O–Si bond is broken, which leads to the deformed tetrahedral aluminum sites (see Scheme 1).

3.12 [Zn]-MFI

As indicated in Fig. 43, zincosilicate samples with zinc contents show high crystallinity up to a Zn/Si ratio of 0.07. There are no reflections that could be attributed to a ZnO admixture. The higher zinc loading results in a noticeable contribution of quartz and amorphous phase in the products. The scanning electron microscopic photographs (Fig. 44) show regular crystallites of the samples. The particle size depends on zinc loading decreasing with growing Zn content (3–10 µm). The IR spectra show two bands in the range 1000–1100 cm–1 for the samples with low zinc loading, whereas the zinc-rich sample exhibits only one band at 1050 cm–1 (Fig. 45a). The reported spectra of other metallosilicates [19] indicate bands attributed to the Si – O – Me

Further temperature rise does not result in very distinct peaks, although the consumption of hydrogen is considerable. This indicates that lattice Zn is involved in the reduction. The peak attributed to the Zn cations is still noticeable, although shifted above 600 ◦C. The TPR curve declines above 900 ◦C, which can suggest that the zinc reduction is complete at this temperature. The shape of TPR profiles differs very much from those for silicalite impregnated with ZnO described by Wan [277], where the reduction of zinc oxide already started at 200 ◦C, as well as from zeolite MFI modified with Zn cations (introduced either by ion-exchange or impregnation), where the main TPR peak appeared at 550 ◦C [276].

Zinc can be introduced into the MFI structure by conventional hydrothermal synthesis. The presence of phosphate anions in the initial mixtures is advantageous. The resulting MFI zincosilicates show high crystallinity for Zn/Si ratios below 0.07. Further increase in the Zn content results in formation of amorphous products or non-porous crystalline phases (quartz). The crystallization time depends on the zinc content. Higher Zn loading requires longer crystallization. The properties of the resulting samples are distinctively affected by the introduced “heteroatoms”. There are considerable differences in properties of the samples with low zinc contents (up to Zn/Si = 0.03) and those exceeding this ratio. Thermal analyses of the as-synthesized products show similar curves for the samples with low Zn contents (up to Zn/Si = 0.03). The introduction of Zn into the lattice results in enhanced temperature of template removal, compared with zinc-free silicalite. The negative lattice charge of zincosilicates results in stronger interaction with the TPA cations and increases the thermal stability of the latter. Thermal analyses do not indicate structure degradation up to 600 ◦C, which is consistent with XRD data.

The 29Si MAS NMR spectra of zincosilicates under study exhibit a signal at ca. – 102 ppm, which results from silicon surrounded by three Si and one Zn tetrahedra. The intensity of the above signal declines on heating because of the removal of some Zn from the lattice. The Raman spectra also exhibit a band at ca. 300 cm–1, which could be assigned to tetrahedral lattice Zn. The intensity of this band decreases after thermal treatment, because some of the Zn atoms are released from the lattice. Regardless of that, the calcined samples can be modified with various cations. TPR measurements indicate quite high resistance of zincosilicate to reduction with hydrogen (up to ca. 600 ◦C). The possibility of modifying the obtained zincosilicates by ion exchange and their noticeable catalytic activity (contrary to the zinc-free silicalite-1) in cumene cracking and in propan-2-ol decomposition strongly supports the localization of the introduced zinc in the lattice. The acid-site strength in H-modifications of zincosilicates is much lower than that in H-ZSM-5, but the modification of these materials with transition metal cations (e.g., Cu) could be promising in the red-ox reactions.

3.13

Dealumination of Levyne –

Characterization of Framework and Extra-Framework Species

For the synthesis of levyne, the global composition of the initial gel was 4Na2O – 2K2O – 6MeQI – Al2O3 – 30SiO2 – 500H2O where MeQ is the methylquinuclidinium ion. The reaction gel was prepared by mixing of 30 wt % aqueous solution of NaOH (pellets EPR, Carlo Erba), MeQI, Al(OH)3 (dry gel, Pfaltz and Bauer), distilled water and SiO2 (fumed silica, Serva). MeQI was prepared by mixing quinuclidine (1-azabicyclo[2,2,2]octane, Aldrich) and iodomethane. The reaction mixture was heated at 150 ◦C under autogenous pressure in static conditions for programmed times, using modified Morey-type autoclaves (8 cm3). The calcination of the sample was carried out in air, heating the sample from 30 to 700 ◦C at a rate of 10 ◦C min–1 under a 15 ml min–1 air flow. The 1D NMR spectra were recorded either on a Bruker MSL-400 or a Bruker CXP-200 spectrometer. For 29Si (39.7 MHz), a 6.0 µs (θ = π/2) pulse was used with a repetition time of 6.0 s. For 27Al (104.3 MHz), a 1.0 µs (θ = π/12) was used with a repetition time of 0.2 s. The 3QMAS experiments at 9.4 T were performed on a Bruker ASX-400 using a recently developed 4 mm MQMAS probehead (Bruker) spinning at 15 kHz, with a radio-frequency field, νRF, estimated at ca. 280 kHz. The pulse sequence was composed of three pulses corresponding to the Z-filter MQMAS method [278], which yields pure absorption spectra. The pulse lengths were experimentally adjusted to 1.75 µs, 0.6 µs (νRF = 280 kHz) and 8 µs (νRF = 10 kHz). The recycle delay was the same as for 1D 27Al MAS experiments. The delay, t1, between the first and second pulse was regularly incremented by 67 µs, according to the method of rotor synchronization [279]. This allows the removal of the spinning sidebands, generally appearing along the isotropic axis, and to reduce significantly the acquisition time. A total of 576 and 2304 scans per increment were used for the as-made and calcined levyne samples, respectively.

The MQMAS method has been previously described [280–283]. We may recall that in addition to the total elimination of the second-order quadrupolar interactions, this technique yields a separation of the different species by both their isotropic chemical shift, δCS, and their second-order quadrupolar interaction PQ = CQ(1 + η2/3)1/2. Nevertheless, it remains difficult to deduce quantitative information from the MQMAS spectra, since the efficiency for the excitation of multiple quantum transitions strongly depends on the quadrupolar coupling constant of each species [283]. Indeed, it is shown that the intensity detected in MQMAS is likely to be underestimated for sites experiencing very weak and very strong quadrupolar interactions, whereas for sites with similar quadrupolar parameters, the direct comparison of the isotropic projection may give a good approximation of their relative population. However, using PULSAR [284], a home made simulation software, which

calculates the response of a nucleus to a series of pulses, taking into account both the quadrupolar parameter and the experimental radio-frequency field, one can predict the actual spectral intensity for each site. Thus, a method to recover the correct population of aluminum on each site is to compare the experimental MQMAS spectra with the theoretically calculated data. This is easy to perform when each site is well-characterized by a pure quadrupolar lineshape and, thus, a unique set of (PQ, δCS) parameters [282]. But in the case of a distribution of parameters, such a comparison becomes extremely difficult and inaccurate, and it is advantageous to use the method recently developed by Zwanziger [285] for the analysis of DAS spectra. A detailed description of the inverse theory and regularization method used for this analysis and their effective application to MQMAS is outside the scope of this paper. This method has already been successfully applied to provide precise knowledge of the relative population of aluminum on five sites in a wellcrystallized AlPO-11 aluminophosphate [286].

The Si/Al ratio of the as-synthesized sample is close to 15, the ratio of the initial gel [180]. This equality means that the incorporation of aluminum into the levyne structure is quite effective [180, 220]. Indeed, the 27Al MAS NMR spectrum of the as-made sample clearly shows that only one NMR line at 53.8 ppm is detected, which is characteristic of tetrahedral coordination (Fig. 49a). During the calcination at 700 ◦C, some of the tetrahedral framework aluminum leaves the structure, and this extra-framework aluminum becomes octahedral, giving rise to a signal at ca. 0 ppm (Fig. 49b).

Previous 13C NMR measurements of the occluded MeQ+ ions have shown that they are incorporated intact in the levyne channels [220]. However, the thermal analysis of the precursor samples still containing the MeQ+ ions has demonstrated that two different MeQ+ ions were present in the channels [220]. The ones that are decomposed at lower temperature (460 ◦C) are neutralizing some of the SiO– defect groups (2.7/u.c.), while the ones decomposed at higher temperature (590 ◦C) are neutralizing the negative charges linked to the presence of the tetrahedral aluminum in the structure, the (SiOAl)– groups (3.4/u.c.). The high-resolution solid-state 29Si NMR spectra of the as-made and calcined levyne samples are shown in Fig. 50. As the levyne structure contains two crystallographically different tetrahedral sites, 36 T1 and 18 T2 sites [221], care has to be taken in interpreting the NMR spectra. The relative intensities of the various lines of both the as-made and the calcined samples are reported in Table 35. The NMR line at – 115 ppm is assigned to a Si(0Al) configuration of the T2 sites [218, 220]. The other NMR lines are tentatively assigned as follows. The – 108 ppm line could stem from a Si(1Al) configuration on T2 sites and Si(0Al) configuration of T1 sites. The

– 103 ppm line could correspond to the sum of Si(2Al)T2 and Si(1Al)T1 configurations. Finally, the – 97 ppm line could stem from the sum of Si(3Al)T2 and Si(2Al)T1 configurations [220]. However, this hypothesis cannot lead to a quantitative interpretation of the NMR spectra.

Fig. 49 27Al MAS NMR spectra of the as-made levyne sample (a) and of the sample calcined at 700 ◦C (b)

The discrepancy very probably stems from the presence of a high amount of defect groups SiOM (M = Na, K, H and/or MeQ). It can be seen from Fig. 50 that the intensity of the – 103 ppm line decreases during calcination and that the intensity of the – 97 ppm line is drastically reduced. It is rather well known that the former could include the -SiOM defect groups, while the latter includes either = Si(OM)2 or = Si(OAl)(OM) defect groups [214].

The presence of two crystallographically different sites raises interesting questions concerning the siting of aluminum on the T1 and T2 sites. Is the distribution of aluminum random or specific [230, 240]? Is the dealumination random or specific? The high-resolution 27Al NMR spectra give valuable in-

Fig. 50 29Si MAS NMR spectra of the as-made levyne sample (a) and of the sample calcined at 700 ◦C (b)

Fig. 51 27Al 3QMAS NMR spectrum of the as-made levyne sample (tetrahedral region)

formation and helps in answering the above-mentioned questions. The 1D 27Al MAS NMR spectra (Fig. 49) already suggest the presence of two different tetrahedral aluminum atoms in the structure. In order to better characterize the levyne samples, 2D multiple-quantum 27Al NMR experiments have been carried out. Indeed, it has already been shown that the use of threequantum transitions greatly increases the resolution. The 3QMAS 27Al NMR spectra of the as-made and calcined samples are shown in Figs. 51 and 52, respectively.

Fig. 52 27Al 3QMAS NMR spectrum of the levyne sample calcined at 700 ◦ C: a tetrahedral region; b octahedral region

Table 36 3QMAS 27Al NMR data of the as-made and levyne samples calcined at 700 ◦ C

The 3QMAS NMR spectrum of the as-made sample clearly shows two different species at δ = 62 ppm (Al1) and δ = 56 ppm (Al2) (Fig. 51, and Table 36). Moreover, the relative intensities of the two lines are, respectively 2 : 1, suggesting that the aluminum distribution is random in the structure. Indeed, this ratio corresponds to the ratio of the crystallographically different tetrahedral sites in the levyne structure, T1/T2= 2. The attribution of the two different tetrahedral aluminum species is hence obvious. The 62 ppm line (Al1) characterizes the tetrahedral atoms on sites T1, while the 56 ppm line (Al2) stems from the tetrahedral A1 atoms on sites T2. The determined quadrupolar coupling constants PQ being similar (1.9 MHz and 2.7 MHz, respectively), the relative intensities on the isotropic projection are regarded as correct. This assumption is confirmed by the computation of the 3QMAS spectrum (Fig. 53), which gives an accurate result for the relative populations (Table 36).

During calcination, a small amount of octahedral extra-framework aluminum species (O site) appears around 4 ppm (Fig. 52). In addition, a new tetrahedral species (Al3) is also detected at δ = 64 ppm, which could be due to deformed tetrahedral aluminum atoms. The PQ of the Al species is much larger than for the other two tetrahedral sites and is equal to 4.8 MHz.

The qualitative analysis of the 3QMAS spectrum (Fig. 52) yields a better understanding of the dealumination process. Table 36 reports all the data for