All these capture experiments demonstrate that the entire family of MOFs shows more than satisfactory capture results, being particularly remarkable for 3. In this context, and with the aim of elucidating the host–guest interactions that account for this capture efficiency, the next step was to attempt to resolve the crystalline structures of two different host–guest aggregates, with amphetamine molecules embedded within the channels of thioether-decorated MOFs 2 and 3. Thus, insertion experiments were carried out on single crystals of 3 and also crystals of 2′ (with the formula {Sr^IICu^II₆[(S,S)-mecysmox]₃(OH)₂(H₂O)}·15H₂O⁵⁶ (2′)), which is an isoreticular MOF to 2, where Ca^II cations are replaced by Sr^II and the resulting crystals better resisted the single-crystal to single-crystal process²² than 2 (see ESI†, experimental section). Thus, samples of host–guest aggregates of 2′ and 3 with amphetamine –suitable for single-crystal X-ray diffraction (SCXRD) and named amphetamine@2′ and amphetamine@3 – were obtained, and their crystal structures could be determined (Table S1†). In this sense, the N₂ adsorption isotherms of amphetamine@2′ and amphetamine@3 were measured (Fig. S9†) and show a decrease of ca. 43 and 51% compared to those of 2 and 3, respectively, which must be attributed to the partial occupation of the pores of both MOFs with amphetamine molecules in amphetamine@2′ and amphetamine@3.

Although amphetamine molecules were disordered in the pores, we succeeded in getting their possible configurations and locations (see the ESI† for structural details), as well as details on their main interaction sites with the hosting matrices 2′ and 3 (Fig. 4, 5 and S10–S12†). Compounds amphetamine@2′ and amphetamine@3 are isomorphs to 2, 2′ and 3, respectively and crystallize in the P6₃ chiral space group of the hexagonal system, confirming the preservation of the 3D network of the hosting matrices 2′ and 3 even after guest capture. The crystal structures clearly show that amphetamine guest molecules are encapsulated in the nanopores of 2′ and 3, where they are simultaneously recognized by the methyl-cysteine and methionine residues, respectively. In the amphetamine@2′ crystal structure, the most stabilizing forces are assured by strong S/π interactions,⁵⁷ (known also as π-hole interactions) where sulfur atoms interact with the amphetamine phenyl ring (APh) [S⋯APh_centroid and S⋯APh_{mean plane} of 3.55 and 2.40 Å, respectively] (Fig. 5a). On the other hand, in the amphetamine@3 crystal structure, the prominent host–guest interaction is of the type σ-hole, which are available for interaction with electron donors such as nitrogen atoms exhibiting a S_mecys⋯N_amphetamine distance of 3.8 Å, shorter than the sum of van der Waals radii (Fig. 5b). Hence, despite the identical nature of guest molecules, methyl-cysteine and methionine arm conformations make the difference. The longer thioether chains in 3, distend their conformation and move the guest towards the centre of the channels with amphetamine molecules orienting in such a way to confine the aromatic side towards the hidden centre of the pores (Fig. 4b, 5, S11 and S13b†). In contrast, the shorter methyl-cysteine arms leave enough space for the larger amphetamine aromatic rings located in the nest formed near the walls of the hosting matrix 2′ (Fig. 4a, 5b, S10 and S12a†). In doing so, in both 2′ and 3 only one of the two crystallographically distinct methyl-cysteine (2′) and methionine (3) moieties show a distended conformation, confining the other one in the small voids, generated along the a crystallographic axis, in a more bent conformation (Fig. 5a).

The slightly higher performance of 3versus2′ for amphetamine capture could be hypothesized with the help of X-ray crystallography. Interactions found in amphetamine@2′ and amphetamine@3 have been discussed both in the context of sulphur-base interactions, involving low-lying sulphur σ* orbitals (known as σ-hole or π-hole in amphetamine@2′ and amphetamine@3, respectively). σ- and π-holes are a particular class of less explored non-covalent interactions, whose relevance has been underestimated despite their relevance in nature, and they have been identified as key agents in the molecular recognition and capture of emergent contaminants in our family of MOFs.^29–31 Even though the strength of these interactions can largely vary depending on the specific molecular context, their overall energy of interactions for similar distances to the ones observed in the obtained host–guest aggregates was found to lie in the range of 20–40 kcal mol⁻¹. Thus, they are relatively weak and reversible, which is advantageous for recycling purposes. Lower interaction energies facilitate the release of captured guests under mild conditions, such as washing with a polar solvent, thereby enhancing the material's recyclability. However, when facing such dangerous contaminants, we consider it worth mentioning that the use of biocompatible MOFs that allow the safe disposal of the drugs of abuse-loaded material represent a safer and preferred option compared to desorption/recycling of the MOF adsorbent, thus mitigating any potential risks associated with residual contamination. In this scenario, in terms of energy involved for host–guest stabilizing effects 2′ and 3 might be considered very similar. However, the differentiation point is related to the static and dynamic features of each pore, and the higher size of the residue and flexibility of 3versus2′, which ensures the most efficient close packing observed, could definitively be at the origin of the better performance in capture of drugs.

Finally, aiming at completing this work by elucidating the interactions enabling such extraordinary capture of fentanyl – a molecule of significant current interest – we also attempted to resolve the structure of a new host–guest adsorbate by encapsulating this molecule (fentanyl) in the most efficient MOF (3). For that, as previously done with MOF 2 and amphetamine molecules, we used crystals of 3′ (an isoreticular SrCu₆ analogue to 3 with the formula {Sr^IICu^II₆[(S,S)-methox]₃(OH)₂(H₂O)}·16H₂O²⁹), which better resisted fentanyl insertion, retaining crystallinity after insertion (see ESI,† experimental section). Thus, after completing the insertion process of these molecules within the MOF by immersing crystals of MOF 3′ in a saturated aqueous fentanyl solution for one week, a new adsorbate structure, named fentanyl′@3′ was obtained. However, this structure shows only a fragment of the fentanyl molecule (termed fentanyl′) within the MOF pores, suggesting that the molecule undergoes fragmentation, as previously reported.⁵⁸ This could be due to the combination of two propitious factors that trigger the known fragmentation: (i) the conditions required to carry out the insertion of the target molecules into the MOF involve prolonged exposure of a concentrated aqueous solution of the guest molecules (amphetamine or fentanyl) in the presence of the corresponding oxamidato-based MOF, which is known to provide a basic medium,³⁸ and (ii) the oxamidato-based MOF can catalyse the degradation of this molecule upon long exposure as also observed before.^59,60 This may well explain the degradation phenomenon.⁵⁸ Please note the different time-scale for the capture experiments with respect to the crystallographic studies (30 s vs. days), which rule out the fragmentation in capture experiments. Nevertheless, despite containing only a fragment of the molecule, the fentanyl′@3′ structure allows us to get some insights into the host–guest interactions between the MOF and the part of the molecule within its pores, which might be also operative in the efficient capture of fentanyl molecules (see Table 1, S2† and Fig. 3) helping us to understand its exceptional effectiveness.

As similar adsorbates of the family, the crystal structure of fentanyl′@3′ compound is isomorphous to 3 and crystallizes in the P6₃ chiral space group of the hexagonal system. Details of the structure (Fig. 6) show guest molecules statistically and severely thermally disordered on three configuration sets (see Fig. S13 and S14†) residing in the pores. They represent the fragment of fentanyl molecules after a break on the 4-piperidinyl position, thus rejecting the piperidine moiety with the linked phenethyl group, and incorporating only the benzene ring and propionyl group connected to the central nitrogen atom.

They are packed via straight S⋯N_fentanyl′ and S⋯C <svg xmlns="http://www.w3.org/2000/svg" version="1.0" width="13.200000pt" height="16.000000pt" viewBox="0 0 13.200000 16.000000" preserveAspectRatio="xMidYMid meet"><metadata> Created by potrace 1.16, written by Peter Selinger 2001-2019 </metadata><g transform="translate(1.000000,15.000000) scale(0.017500,-0.017500)" fill="currentColor" stroke="none"><path d="M0 440 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z M0 280 l0 -40 320 0 320 0 0 40 0 40 -320 0 -320 0 0 -40z"/></g></svg> O_fentanyl′ bonds involving methionine residues [S⋯N and S⋯CO_Centroid distances of 1.84(1) and 2.20(1) Å], which block guest moieties embracing both kind of arms, in a lock and key fashion (Fig. 6c). Indeed, the pores are decorated with methionine residues exhibiting a distended conformation (Fig. 6a). Both S⋯N_fentanyl′ and S⋯CO_fentanyl′ distances fall in the range of typical bonds. It is quite surprising, but it is worth considering 3′ as a nanoreactor. A fentanyl degradation occurs followed by a coordination reaction within pores. Such a situation has been observed in this family of MOFs³⁵ and, again, it might be ascribed to the extraordinary flexibility confined only in pores, for MOF 3, where highly bendable arms are intrinsically prone to adopt different conformations of the thioether chains, depending on the different chemical environments determined by the guests' nature. In particular, the crystal structure of fentanyl′@3′ shows methionine arms to be more distended than methyl-cysteine ones in amphetamine@2′ and methionine in amphetamine@3 (Fig. 4a and b), featuring available sulphur groups for interaction or even chemical bonds with electron donors – including oxygen and nitrogen atoms. Thus, they encapsulate the targeted guest molecules assuming the favourite conformation and interactions or bonds, in each case, to maximize their receptor assets. Indeed, 3 exhibits a quite unique dynamism and flexibility. It is the one related to the different conformations thioalkyl residues pointing toward the hexagonal channels adopt depending on the nature of the molecules hosted within its channels. These nice molecular recognition properties, together with the high crystallinity of 3, allow us to have beautiful snapshots of the methionine residue conformations, which are shown in Fig. S15†. Overall, after analysing the presented crystal structures, we can conclude that the specific sulphur-based chemical functionalities pointing into the pores of our MOFs play a crucial role in dictating the observed interactions, particularly in relation to the molecular structures of amphetamine and fentanyl. So, our crystallographic analysis provides key insights into how these host–guest interactions govern the capture of such substances, which is instrumental in designing future porous materials with optimized adsorptive properties.

In order to confirm the purity of the bulk samples of amphetamine@2′, amphetamine@3 and fentanyl′@3′, powder X-ray diffraction (PXRD) studies were carried out for polycrystalline samples of these compounds, which confirm both that they maintain their structural integrity and also that all crystals are identical to those selected for SCXRD (Fig. S16–S18†). Moreover, aiming at verifying the robustness of MOF 3, PXRD studies were carried out before and after the 16 reusability experiments (Fig. S19†), confirming that the material remains crystalline and unaltered after these experiments. Moreover, no leaching of metal atoms was observed after reusability experiments (as verified with ICM-MS experiments), which is not surprising considering that the robustness of these MOFs had been already confirmed.³⁸ Finally, the solvent contents were ultimately established with the help of thermogravimetric analyses (Fig. S20†).