The crystal structure of Cd-LEV at RT (Table S1) is in agreement with that determined in our previous study (Cametti et al. 2021). Minor differences are found in the distribution of the EF cations. C1 is split over two positions, C1 (Occ. = 0.32(8)) and C1A (Occ. = 0.48(8)), with total occupancy equal to that determined for C1 in Cametti et al. 2021. Ca atoms are found at C3 site and, in addition, at C4 site (not present in Cametti et al. 2021). This tiny rearrangement of the Cd and Ca atoms also lead to a shifted position of the H2O at the W3 site, whereas no significant differences are found for W1, W2 and W4. Their atomic coordinates are in agreement with those previously reported for Cd-LEV (Cametti et al. 2021).
Unit-cell parameter variation as a function of temperature
The change of the unit-cell parameters as a function of temperature is reported in Fig. 1. Corresponding values of the unit-cell volume trend of levyne-Ca (Cametti 2018) and Ag-LEV (Cametti et al. 2020) are also shown for comparison (Fig. 1a). Upon heating, the simultaneous expansion and contraction along the c- and a-axis (Fig. 1b), respectively, causes an overall decrease of the unit-cell volume. At first, a tiny contraction occurs from 25 to 125 °C, when the volume decreases by approximately 1% if compared to that measured at RT. A more pronounced change is observed form 175 to 250 °C, when the reduction amounts to 7%. Such value is significantly higher compared to that of both levyne-Ca and Ag-LEV at corresponding temperatures. As a reminder, the data reported in this study were collected under the same experimental conditions used for the other cationic forms. At 400 °C, the unit-cell volume further decreases, reaching a total contraction of 8% with respect to the RT structure.
Structural modifications upon heating
The dehydration of Cd-exchanged levyne starts at 50 °C, when part of the H2O at W2, W3 and W4 sites is released. In contrast, the W1 site remains fully occupied (Table 2). The water loss is accompanied by a change in the Ca distribution and Ca completely migrates to the C3 site (Occ. = 0.46(3)), emptying C4. The H2O at W2, W3 and W4 also readjust their positions, which result in a more disordered configuration (Fig. 2a).
The dehydration proceeds with the gradual depletion of the W1, W2, W3 and W4 sites and at 100 °C, the content of structural H2O decreases to 11 H2O pfu (Table 3). This decrease is accompanied by an increase of the disorder of the extra framework species. Ca and Cd start migrating toward the wall of the lev cavity at the CW2 and C8 sites, respectively (Fig. 2b). However, the high structural disorder makes complex the unequivocal assignment of Ca and H2O, which spread over low occupied sites close to each other (Table 3). From this temperature on, electron density was detected at the centre of the double six-membered ring cages refined with Cd atomic scattering factor. The position of this site corresponds to that of C6 and Ag1 in the high-temperature structures of levyne-Ca (Cametti 2018) and Ag-LEV (Cametti et al. 2020), respectively. Upon heating, the occupancy at C6 gradually increases, indicating that, similar to levyne-Ca and Ag-LEV, part of the EF cations approaches the D6R due to dehydration.
At 150 °C, a significant relocation of the EF cations takes place. Ca at C3 completely moves to the middle of the eight-membered ring windows of the lev cavity, at the CW2, CW2A and CW2B sites (Table 4, Fig. 2c). In addition, the migration of Cd inside the D6R cage continue and the occupancy of C6 reaches 30%. Residual H2O content (approximately 5 H2O pfu) is retained at W1, W2, W2B and W3 sites. At 175 °C, the increasing structural disorder hampers a clear determination of the EF-cations/H2O distribution. The refined content of structural H2O amounts to 0.3 H2O pfu that is found exclusively at W2 site. However, it cannot be excluded that some H2O might be present also at the CW2 and CW2A sites (Table 5), as suggested by the slightly higher value of refined Ca atoms (i.e. 0.9 vs 0.6 apfu based on the chemical composition of this sample).
Between 200 and 250 °C, in correspondence of the pronounced drop of the unit-cell volume (Fig. 1a), the interpretation of the diffraction data was complicated by the strong smearing and broadening of the reflections (Fig. S1), the intensity of which significantly decreased. Thus, a meaningful solution of these structures was not possible and only the cell parameters were determined. At 275 °C, although a high Rint (8.84%), the structure could be solved and refined in space group R-3. This high- temperature structure, described in the next paragraph, is assumed to be completely dehydrated.
The dehydrated structure of Cd-exchanged levyne at 275 °C
The contracted phase of Cd-LEV has a structural topology similar to that of levyne-Ca between 200 and 250 °C. This structure is characterized by the statistical rupture of the T–O–T connections (T1–O2–T11 and T1–O3–T11) of the D6R tetrahedra (Fig. 3a, Table 6). This mechanism involves the partial migration of T1 and T11 sites toward new position (T1B and T11B), with corresponding formation of two new tetrahedra with oxygen apices OB1 and OB2. Since this is a statistical process, two topologies coexist at high temperature: the original levyne (LEV) and the levyne B (Arletti et al.2013). The latter, which originates if only the new tetrahedral sites (i.e. T1B and T11B) and T2 are considered, has ABABCBCAC stacking sequence (Fig. 3b). According to the structural refinement of Cd-LEV, 50% of the original tetrahedral sites (T1 and T11) migrated to the new positions T1B and T11B. The EF cations spread mainly along the threefold axis and residual Cd is found close to the wall of the lev cavity at C2, C7 and C8 sites (Table 6). Since it was not possible to monitor the structural changes between 200 and 250 °C, it remains enigmatic whether the rupture, similar to levyne-Ca, already began at lower temperatures and gradually increased up to 275 °C, or it was a more abrupt change. A plausible hypothesis is that the strong smearing and broadening of the reflections along the c-axis (Fig. S1) observed at 200 °C, might be associated with the onset of the breaking process.
To check if this structural configuration is maintained at higher temperature, tentative refinements were performed on the structures at 350 and 375 °C. Although the low quality of the data, characterized by low resolution, no additional changes with respect the structure at 275 °C were noticed. As an example, the refined parameters of the structure at 350 °C are reported in Table S2a, b.
Diffraction data at 400 °C (Fig. 4a) consisted of very broad reflections with a maximum resolution of 1.0 Å, pointing out the onset of the structural collapse. The estimated unit-cell volume determined at this temperature was 3226(1) Å3. The determination of the correct unit-cell parameters was complicated not only because of the additional damping of the atomic scattering-factors at high temperature but also because of the strong smearing and splitting of the reflections (Fig. 4a). Such a contracted structure, as expected from the extent of the structural changes, is not able to reabsorb water. Diffraction data collected on the crystal used for the HT measurements and subsequently exposed to high humidity conditions at room temperature were limited to 2θ = 41.7° and are characterized by smeared and weak reflections (Fig. 4b). The unit-cell volume (3198.8(9) Å3) was comparable with that obtained for the structure at 400 °C, confirming that it could not re-hydrate.
Differences with levyne- Ca and Ag-exchanged levyne
The structural evolution of Cd-LEV upon continuous heating can be summarized by the following two steps: (i) initial dehydration with corresponding diffusion of the EF cations within the zeolitic cavities; (ii) pronounced contraction of the unit-cell volume associated with the rupture of 50% of the T–O–T bonds constituting the D6R cage. This step leads the so-called levyne B topology (Arletti et al. 2013). Thus, differently from the pristine material levyne-Ca (Cametti, 2018), no additional transition to the levyne B’phase occurs in Cd-exchanged levyne.
In levyne-Ca, this transformation is associated with the breaking of an additional T–O–T bridge of the framework (T2–O5–T2), which starts at 275 °C in correspondence of the loss of residual H2O. It may be argued that, in levyne-Cd, the onset of the breaking of an additional T–O–T was not detectable due to the quality of the diffraction data (Table 1). However, two arguments support a different transformation path:
in levyne-Ca, at 300 °C, the migration of (Si,Al) atoms toward the new T1B site amounted to 37%, differently from Cd-LEV where at 275 °C the population of the new T sites (T1B and T11B) is 50%;
the pronounced difference (from 200 °C on) between the unit-cell volume curves (Fig. 1a) indicates that in Cd-LEV the dehydration process is not only shifted toward lower temperatures (e.g. the unit-cell volume of the structure at 225 °C corresponds to that of levyne-Ca at 275 °C), but also more drastic in terms of structural contraction. Thus, it is plausible that such behaviour is associated with different phase transformations.
It has to be pointed out, at this point, that the drastic contraction of the unit cell of Cd-LEV is also related to a reduced thermal stability. Because of the smaller size (and higher ionic potential) of Cd2+ with respect to both Ca2+ and Ag+, it is reasonable that the thermal stability of Cd-LEV is lower compared to Ca- and Ag-LEV (Cruciani 2006). In this sense, Cd2+ exerts more strain on the framework that causes an earlier collapse with respect to Ca-LEV.
Interestingly, similar to Cd-LEV, upon dehydration levyne B’ structure does not form in Ag-LEV too (Cametti et al. 2020). In that case, the absence of the B’ topology was explained based on the H2O content of the RT structure. Ag-LEV at ambient conditions contains less H2O compared to the pristine material; therefore, a second dehydration step, which would trigger the B-B’ transition, does not take place. In Cd-LEV, the number of H2O pfu determined by the structural refinement (Table 1) is even slightly higher than the corresponding one in levyne-Ca. This is not surprising, considering that, in natural levyne-Ca, Na+ and K+ are also present as EF cations. In Cd-LEV, only divalent cations occupy the zeolitic cavities, where, consequently more H2O can be accommodated. Nevertheless, the structure does not experience a second T–O–T rupture. Thus, the missing transformation to B’ must be related to specific properties of Cd2+.
In the SR-XRPD study by Arletti et al. (2013), this additional modification was not reported as well. In this case, the kinetic of the experiments (faster and far from the equilibrium) was responsible for the rapid water loss and the structure had not enough time to complete the transformations. In our SC-XRD experiments, it seems like a similar mechanism is driving the dehydration of Cd-LEV. That is, a faster (compared to levyne-Ca) release of water may lead to a more abrupt dehydration and, therefore, to the absence of the B- > B’ transformation. The main difference between the trend of the unit-cell volume of levyne-Ca and Cd-LEV in Fig. 1a is certainly the stronger contraction of Cd-LEV. However, Cd-LEV is also characterized by a steeper unit-cell volume curve, pointing to a sudden water loss similar to that observed by Arletti et al. (2013). Hence, Cd2+ may play a double role in such process as follows: (i) it tends to coordinate less H2O than Ca2+ due to the smaller radius; therefore, the H2O are released earlier at lower temperature; (ii) it exerts more strain on the framework, which starts collapsing already at 400 °C.