Template-aluminosilicate structures at the early stages of zeolite ZSM-5 formation. A combined preparative, solid-state NMR, and computational study

Species at three stages in the self-assembly of zeolite ZSM-5 have been studied with one- and two-dimensional magic-angle-spinning 13C, 27Al, 29Si, and 1H NMR spectroscopy and compared with the earlier proposed structures: (1) precursor species containing 33-36 T sites around a tetrapropylammonium (TPA) cation, (2) nanoslabs consisting of a flat 4 x 3 array of such precursors, and (3) the final TPA-ZSM-5 zeolite. Synthesis was carried out in D2O to suppress the water and silanol protons. Under such conditions, the effective Si-H and Al-H distances measured with 29Si-{1H} and 27Al-{1H} rotational echo double resonance (REDOR) reflect the interactions between TPA cations and the surrounding aluminosilica. The 29Si-{1H} REDOR curves for Q4-type silicon atoms at the three mentioned stages are closely similar, as well as the observed 27Al-1H REDOR curve for the precursor species compared to that for the TPA-ZSM-5. This indicates that in addition to externally attached TPA, there is also internal TPA already incorporated at an early stage into the aluminosilicate in a similar way as in the final zeolite, in accordance with the earlier proposed MFI self-assembly pathway (Kirschhock et al. Angew. Chem. Int. Ed. 2001, 40, 2637). However, the effective distances extracted from the initial REDOR curvatures are significantly (10-15%) larger than those computed for the model. Since there is no temperature effect, we tentatively assign this difference to a reduction of the 29Si-1H and 27Al-1H interactions by multispin decoherence effects or self-decoupling caused by proton spin diffusion. By assuming the computed model distances and fitting Anderson-Weiss curves to the observed REDOR data, we obtain similar "decoherence times" in the order of 0.1 ms. The observed 29Si-{1H} REDOR dephasing for the Q3 sites in the precursors is significantly faster than that for the Q4 sites. This is tentatively ascribed to a partial deuteron-proton back exchange at the silanol positions.