Introduction

Supramolecular chemistry is a branch of chemistry that, in part, involves the formation of supramolecules consisting of an organized and complex unit that is formed from two or more different species that are held together by intermolecular forces [1]. These supramolecules consist of an equilibrated structure that involves a comparable balance between both enthalpy and entropy [2]. Numerous processes are involved in the formation of supramolecular complexes, such as self-assembly, molecular recognition and crystal engineering [3]. Self-assembly is the independent organization, by means of non-covalent interactions, of molecular components that produce structures that can exist at any scale. The structure of the assembly is dependent upon the molecular structures, as molecules will tend to form molecular assemblies in conformations that have enhanced thermodynamic stabilities [4]. Crystal engineering has been defined by Desiraju as “the understanding of intermolecular interactions in the context of crystal packing and in the utilization of such understanding in the design of new solids with desired physical and chemical properties” [5]. If one considers crystalline complexes as the supramolecular equivalent to molecules, then crystal engineering is the supramolecular equivalent to organic synthesis [6]. What, in essence, all this equates to is simply how important are the features, characteristics and structures of the building blocks that form the crystal phase [7].

The kinetics of guest desolvation from inclusion compounds have been examined over the years in order to build an understanding of the mechanism thereof, as well as to evaluate such factors as rate constants and activation energies [8]. This process may be monitored by measuring the change in mass loss as a function of time [9]. Early work by Coats and Redfern investigated the use of thermogravimetric experiments for the determination of kinetic data [10]. This work was then expanded upon by Ozawa [11] and Flynn [12], who developed a method for calculating the activation energy of the guest desolvation process from an inclusion complex using different heating rates and recording the temperatures at which specific fixed mass losses were attained across these various heating rates. Numerous examples of activation energies of guest loss from host-guest inclusion compounds have subsequently been reported in literature [13,14,15].

The wheel-and-axle compounds 9,9′-(1,4-phenylene)bis(fluoren-9-ol) (H1), 9,9′-(ethyne-1,2-diyl)bis(fluoren-9-ol) (H2) and 9,9′-(biphenyl-4,4′-diyl)bis(fluoren-9-ol) (H3) (Scheme1) have been demonstrated to be efficient host species for the organic solvents tetrahydrofuran, diethyl ether, 1,4-dioxane and morpholine; these host compounds were designed and synthesized by Weber and co-workers [16,17,18]. In the present investigation, H1‒H3 were crystallized from the polar aprotic solvent tetramethylurea (TMU, Scheme 1) and complexes were formed in each instance. We note that the TMU-containing inclusion compound with H1 has been reported previously [19] and, in that work, was described the crystal structure of this complex together with the results obtained from Hirshfeld surface analyses, thermal experiments and lattice energy considerations. In the present work, these characteristics of the TMU-containing complex with H1 obtained in our laboratories are compared with the reported results and, furthermore, these features are also revealed for the novel complexes of TMU with H2 and H3; our findings are now reported.

Scheme 1
scheme 1

Structures of host compounds H1‒H3 and the guest molecule TMU

Full size image

Experimental

Materials

All chemicals/solvents were purchased from Sigma Aldrich, South Africa, and were used without further purification.

Synthesis of host compounds H1‒H3

Host compounds H1‒H3 were synthesized according to a previous report [16].

Preparation of the single solvent host-guest inclusion complexes

The complexes of TMU with H1, H2 and H3 were prepared by adding 15 mmol of the guest solvent to 0.04 g (0.09 mmol for H1, 0.10 mmol for H2 and 0.078 mmol for H3) of the host compound in glass vials. To ensure complete dissolution of the host species within TMU, 15 drops of chloroform were added to the vials with H1 and H2, while 30 drops of acetonitrile were required for those with H3, whereafter mild heat was applied. The vials were stored in a refrigerator set to 5 °C, whereafter crystals formed in the solutions. These were isolated by vacuum filtration, washed with petroleum (PET) ether (b.p. 40–60 °C) and analysed by means of 1H NMR spectroscopy. Successful complexation was confirmed when both the host and guest resonance signals were observed on the proton NMR spectra. The relative integrals of appropriate host and guest resonance signals were then compared in order to determine the host: guest (H: G) ratios. Relevant 1H NMR spectra are provided in the Supporting Information (SI, Figures S1a‒c).

Analytical instrumentation

NMR spectroscopy

The 1H NMR experiments were carried out by means of a Bruker Ultrashield Plus 400 MHz spectrometer with CDCl3 (for experiments involving H1 and H2) and DMSO-d6 (H3) as the deuterated solvents. The spectra were analysed using TopSpin 4.3.0 software.

Single crystal X-ray diffraction analyses

The complexes H1·2(TMU), H2·2(TMU) and H3·2(TMU) were analysed using a Bruker D8 VENTURE single crystal diffractometer and graphite-monochromated MoKα-radiation (λ = 0.71073 Å). The crystals were maintained at 173(2) K with nitrogen vapour supplied by an Oxford Cryostream-800 (Oxford Cryosystems, U.K.). The data-collections were controlled using APEX3/v2019.1-0 (Bruker) software; ω- and ϕ-scans of width 1.0° were employed. Data reductions were carried out by means of SAINT v8.40 A (Bruker) [20]. SADABS (2016/2) [21] was implemented as the multi-scan method for the absorption corrections. The structures of H2·2(TMU) and H3·2(TMU) were solved using direct methods, with refinement by full-matrix least-squares using programs in the SHELX suite [22]. Version 4.0 [23] of X-Seed (for supramolecular crystallography) was employed as the graphical user interface [24]. Generally, non-hydrogen atoms were refined anisotropically, with all hydrogen atoms located in difference Fourier syntheses and placed in idealised positions in a riding model with isotropic displacement parameters in the range 1.2‒1.5 times those of their parent atoms. The structure of H1·2(TMU) was also solved using direct methods and refined by least-squares procedures using SHELXL-2019/3 [25] with SHELXLE [26] as a graphical interface. Diagrams were drawn with ORTEP-3 for Windows version 2023.1 [27]. All non-hydrogen atoms were refined anisotropically, while the carbon- and oxygen-bound hydrogen atoms were placed in idealised geometrical positions in a riding model. Nitrogen-bound hydrogens were located on a difference Fourier map and refined using a riding model. There was disorder of the TMU guest component in H1·2(TMU) by a 180° rotation around the carbonyl moiety with a slight wobble.

Thermal analysis

Thermogravimetry (TG) and differential scanning calorimetry (DSC) experiments were carried out on each of the three inclusion compounds. The solids were first recovered from their solutions by means of vacuum filtration, washed with PET ether and patted dry in folded filter paper prior to these analyses. The instrument was a Perkin Elmer Simultaneous Thermal Analyzer (STA) 6000, and data were analysed by means of Pyris Series data software. Samples were placed in ceramic pans with an empty pan serving as the reference. The purge gas was high purity nitrogen, and samples were heated from approximately 40 to 350 °C at a heating rate of 10 °C·min‒1.

Kinetics of guest release

The activation energy of the guest release process was subsequently investigated for each of the three complexes. This was achieved by means of the method of Ozawa [11], and Flynn and Wall [12], which employed thermogravimetric analysis to measure the mass loss experienced by each complex as a function of different heating rates [13]. In more detail, the method involved heating each of the inclusion complexes at different heating rates (β), and the temperatures at which certain fixed mass loss percentages (α) (of the overall mass loss) were achieved were then recorded. Thereafter, ln(β/βo) was plotted against 1000 K/T, according to the equation

\(\text{ln}(\beta/{\beta}_{\circ})=\frac{E{}_{a}}{RT}\,+\,\text{lnA},\)

where βo is the standard state used to remove units from the logarithmic calculation (1 °C·min‒1), Ea is activation energy, T is the absolute temperature (K) and A is a pre-exponential factor of the Arrhenius equation. The negative gradient from the resultant plots was then multiplied by ‒R (where R = 8.314 J·mol‒1·K‒1), which furnished Ea at each mass loss percentage. The heating rates (β) employed for each of the complexes were 2, 4, 10, 16 and 32 °C·min‒1, while the fixed mass losses that were selected were 10, 20, 30, 40, 50, 60, 70, 80 and 90% of the total mass loss. The samples were prepared for these analyses in the same fashion as for the usual thermal analysis experiments.

Software

The interactions from SCXRD analyses between host and host, host and guest, and guest and guest molecules, were investigated using program Mercury 2022 1.0 [28]. This program, in addition, enabled the construction of unit cell, host-guest packing and void diagrams. In the latter instance, the guest molecules were deleted from the packing calculation and the spaces which resulted in this way were analysed by means of a probe with a radius of 1.2 Å. These void diagrams served to determine the guest accommodation type, whether in infinite channels or in discrete cages. Moreover, the non-covalent host‧‧‧guest interactions that were present in the complexes were examined and quantified using program CrystalExplorer 21 [29, 30]. A three-dimensional Hirshfeld surface was generated around the guest molecule and the surroundings outside of this surface were then explored. These surfaces were translated into two-dimensional fingerprint plots in which de and di are the distances to the nearest atom outside and inside the guest surfaces, respectively.

Results and discussion

Formation of the inclusion compounds

Host compounds H1‒H3 were each crystallized from TMU, and after 1H NMR analyses on the so-formed crystals, it was revealed that complexes were successfully formed in each instance. The host: guest (H: G) ratio of each of the complexes was consistently 1:2. This ratio was also obtained by the research group of de Vries et al. for the TMU-containing complex with H1 [19].

Single crystal X-ray diffraction (SCXRD) analyses

The crystal structures of the three complexes produced in this work were analysed by SCXRD. Table 1 contains the relevant crystallographic parameters pertaining to these analyses. These inclusion compounds crystallized in the monoclinic space group P21/c, and their analyses were performed in P21/c for H1⋅2(TMU) and in the alternative setting P21/n for both H2·2(TMU) and H3·2(TMU). The structure of H1⋅2(TMU) reported by de Vries et al. [19] crystallized in the same crystal system and space group as that in our own laboratory. However, their SCXRD analysis was carried out at 183 K and, in an attempt to improve upon their data (they reported a high wR2 factor, 0.2020), we conducted this experiment for H1⋅2(TMU) at a significantly lower temperature of 100 K (Table 1). These results are therefore significantly improved upon compared with those previously published (here, wR2 is much lower, viz. 0.0939), and the crystal structure shared characteristics with respect to the disorder in the methyl groups of the TMU molecule over two positions.

Figure 1a‒c illustrate the unit cell with the host-guest packing motif in each complex (left) and the guest accommodation type (right). From Fig. 1a and b, the guest molecules in H1·2(TMU) and H2·2(TMU) are located in infinite, unidirectional channels which are parallel to the a-axis, while in H3·2(TMU), four guest molecules are observed in an X-shaped cavity, best visualized along the c-axis (Fig. 1c, bottom).

Table 1 Relevant crystallographic data for H1·2(TMU), H2·2(TMU) and H3·2(TMU)
Full size table
Fig. 1
figure 1

Unit cells and host-guest packing motifs (left) and void diagrams (right) for complexes (a) H1·2(TMU), (b) H2·2(TMU) and (c) H3·2(TMU). The host compounds are in capped stick representation and the guest molecules in spacefill form. A clear representation of the cavities was required for H3·2(TMU), hence the c-axis visual is also provided here (Fig. 1c, bottom)

Full size image

In each complex, the guest molecules are retained in the crystals by means of classical (host)O‒H···O(guest) hydrogen bonding interactions. These are illustrated in Fig. 2a‒c (for H1·2(TMU), H2·2(TMU) and H3·2(TMU)). More specifically, Fig. 2a and c are stereoviews of these interactions for clarity; in the complex H1·2(TMU) (Fig. 2a) the host molecule is located on a centre of inversion and interacts with two inversion-related TMU molecules through identical hydrogen bonds. Each methyl group of the guest species is disordered over two positions, as discussed earlier (two views are provided, one showing only the major disorder guest component with s.o.f. 0.513(3) (left) and the other, both guest disorder components (right)), while in the complex H3·2(TMU) (Fig. 2c), no disorder is present and the two guest molecules are not inversion related. However, once more, the two TMU molecules are related by inversion in H2·2(TMU) (Fig. 2b) (the host molecule is, again, positioned on a centre of inversion), and no disorder is evident here either.

Fig. 2
figure 2

Classical (host)O‒H···O(guest) hydrogen bonding interactions in (a) H1·2(TMU) (stereoview of the complex showing only the major methyl group disorder guest components (left) and with both major and minor disorder guest components (right)), (b) H2·2(TMU) and (c) H3·2(TMU) (also a stereoview)

Full size image

Table 2 summarises these classical hydrogen bonding parameters.

Table 2 Host···guest classical hydrogen bonding parameters in the three complexes
Full size table

Owing to the two guest molecules being related by inversion in their respective unit cells in the H1 and H2 complexes, these hydrogen bond parameters are identical (1.88, 2.72(4) Å, 177° and 1.82, 2.661(1) Å, 174°, respectively), while these parameters are unique to each of the two crystallographically independent TMU molecules in H3·2(TMU) (1.91, 2.709(1) Å, 160° and 1.95, 2.672(2) Å, 144°). de Vries and coworkers also reported this host···guest classical hydrogen bond in the H1·2(TMU) complex [19], and with very similar bond parameters (1.81, 2.72(1) Å, 178°) as reported in the present work, but with a somewhat lower precision in the O···O distance).

Not only are classical bonds of this type present in these complexes, but so too are non-classical (host)C‒H···O(guest) and (guest)C‒H···O(host) interactions, one in H1·2(TMU) and two and three in H2·2(TMU) and H3·2(TMU), respectively. These interactions, their parameters being summarised in Table 3, range from 2.43 to 2.67 Å (H···A) and 3.260(2) to 3.478(2) Å (C···O), with the associated ∠DHA angles between 133 and 157°. It is interesting to note that three of the non-classical hydrogen bonds are of the (guest)C‒H···O(host) type. Figure 3a‒c are depictions of these interactions (note that the classical hydrogen bonds between the host and guest molecules have been repeated in these figures).

Table 3 Host···guest and guest···host non-classical hydrogen bond parameters in the three complexes
Full size table

In addition to intermolecular hydrogen bonding, intramolecular non-classical (guest)C‒H···N(guest) hydrogen bonds are also present but their angles are, understandably, low (2.34‒2.53 Å (H···N), 2.722(7)‒2.907(2) Å (C···N), 102‒105°).

Fig. 3
figure 3

Hydrogen bonding (both classical and non-classical) present in (a) H1·2(TMU), (b) H2·2(TMU) and (c) H3·2(TMU)

Full size image

Additionally, the H1·2(TMU) and H3·2(TMU) complexes have two C‒H(guest)···π(host) interactions (Fig. 4a and c) through the guest methyl hydrogen atoms, which further facilitate guest retention in their complexes. Applicable distances in H1·2(TMU) measure 2.96 and 2.88 Å (H···π), 3.922(3) and 3.746(5) Å (C‒H···π), with associated angles of 167 and 148°, respectively, while these parameters in H3·2(TMU) are 2.75 and 2.78 Å, 3.7233(17) and 3.7016(18) Å, and 170 and 158°. The complex H2·2(TMU) has only one contact of this type (2.75, 3.4246(13) Å and 127°) (Fig. 4b).

Fig. 4
figure 4

The unique C‒H(guest)···π(host) interactions in the complexes (a) H1·2(TMU), (b) H2·2(TMU) and (c) H3·2(TMU)

Full size image

Finally, a single (guest)C‒H···CAr(host) interaction was also identified in H2·2(TMU) (Fig. 5), measuring 2.88 Å (127°).

Fig. 5
figure 5

The (guest)C‒H···CAr(host) interaction present in H2·2(TMU)

Full size image

Hirshfeld surface analyses were subsequently undertaken [29, 30] by means of program CrystalExplorer 17.5 in an attempt to quantify the non-covalent interactions present between the host and guest atoms in each of these complexes. More specifically, the interactions of interest were those of the (host)H···O(guest) type. Figure 6a‒c display the two-dimensional fingerprint plots (left: the (host)H···O(guest) contacts, the sharp spikes that are present; right: all interactions) and the three-dimensional surfaces that were generated around the guest molecules (centre). Each of the two guest disorder components of TMU in H1·2(TMU) was considered independently (Fig. 6a, top and bottom).

Fig. 6
figure 6

Two-dimensional fingerprint plots (left: (host)H···O(guest), the sharp spikes; right: all interactions) and three-dimensional surfaces (centre) for (a) H1·2(TMU) (top and bottom represent the two disorder guest components), (b) H2·2(TMU) and (c) H3·2(TMU)

Full size image

After quantification of the (host)H···O(guest) interatomic interactions, the percentages that were obtained were, in H1·2(TMU), 7.8% (disorder guest component 1 (major)) and 8.1% (disorder guest component 2 (minor)), in H2·2(TMU), 8.4%, and in H3·2(TMU), 7.9%, respectively. These were all therefore reasonably comparable. In the Hirshfeld surface investigation for H1·2(TMU) reported previously [19] was calculated 9% for these (host)H···O(guest) interactions, which is slightly different to the average of the two values found in the present work (7.8 and 8.1% for each disorder guest component).

Thermal analyses

Thermal analyses were carried out on each of H1·2(TMU), H2·2(TMU) and H3·2(TMU) in order to investigate their relative thermal stabilities. The resultant curves (overlaid TG (red) and DTG (olive green)) are provided in Fig. 7a‒c, while Table 4 summarises the more relevant data from these curves. These plots indicate that for all three complexes, their respective enclathrated TMU contents escaped from the crystals in a single step. The Ton values (the onset temperature of the guest release event which also serves as a measure of the relative thermal stability of the complex) were extremely similar (83.1 and 81.1 °C) for the TMU complexes with H1 and H2 (Fig. 7a and b), while this temperature for H3·2(TMU) was significantly higher, 90.3 °C (Fig. 7c). Tend, the temperature at which the last of the guest has been released, which is unrelated to Ton, decreased in the order H1·2(TMU) > H3·2(TMU) > H2·2(TMU). In the report of de Vries et al. [19], the Ton was 99 °C for H1·2(TMU), which is significantly different to the 83.1 °C that was measured in our laboratories; the reason for this anomaly may be that their heating rate differed from 10 °C·min‒1 (this rate was not unambiguously stated in that work).

Table 4 Thermal data for the three complexes
Full size table

The expected mass losses correlated closely with those that were determined experimentally, thus confirming the 1:2 H: G ratios as obtained from 1H NMR spectroscopy. Interestingly, the percentage TMU released upon heating H1·2(TMU) at 10 °C·min‒1 was significantly closer (35.7%) to the theoretical loss calculated (34.7%) compared with that previously reported (30.0%) [19].

Fig. 7
figure 7

Thermogravimetric curves for (a) H1·2(TMU), (b) H2·2(TMU) and (c) H3·2(TMU)

Full size image

Kinetics of the guest release event

The TG curves of the complexes of H1, H2 and H3 with TMU at the various heating rates are provided in Fig. 8a‒c. The final experiment with the H2·2(TMU) complex, at 32 °C, was carried out in triplicate since the curve appeared to be an outlier; however, this was not the case since the same result was obtained each time (the curves for these repeated TGs and the plots used to determine Ea have been deposited in the SI, Figures S2 and S3, with associated values in Tables S4 and 5). The average total mass losses calculated for the complexes with H1, H2 and H3 were 35.8, 36.7 and 31.2%, respectively. The plots of ln(β/βo) against 1000/T for each of the complexes H1·2(TMU), H2·2(TMU) and H3·2(TMU) are illustrated in Fig. 9a‒c. These were obtained by considering mass losses of 10 to 90% (α) of the total mass loss, in increments of 10% each time. The activation energies for each of the complexes were then calculated from the slopes of these plots (SI, Tables S1‒S3).

Fig. 8
figure 8

TG curves for the TMU-containing complexes when heated at 2, 4, 10, 16 and 32 °C·min‒1 for (a) H1·2(TMU), (b) H2·2(TMU) and (c) H3·2(TMU)

Full size image
Fig. 9
figure 9

Plots of ln(β/βo) versus 1000 K/T at 10, 20, 30, 40, 50, 60, 70, 80 and 90% of the total mass loss observed for (a) H1·2(TMU), (b) H2·2(TMU) and (c) H3·2(TMU)

Full size image

The average activation energy (Ea) of desolvation was calculated to be 148.7 ± 5.4, 128.6 ± 10.8 and 149.4 ± 0.8 kJ·mol‒1 for H1·2(TMU), H2·2(TMU) and H3·2(TMU), respectively. In Fig. 9a and c, the lines are all parallel, which indicate that there is a single mechanism at work for the desolvation of the guest species from the complex [8]. The lines obtained for H2·2(TMU) (Fig. 9b), however, do not all appear to be parallel (largely as a result of the line at 90% mass loss, as discussed earlier) and, therefore, the mechanism of guest release, quite plausibly, varies as a function of the extent of desolvation. Since the host compounds are not identical, comparisons of these energies are not valid.

Conclusions

The three fluorenone-derived host compounds H1, H2 and H3 were all found to be effective at enclathrating TMU, and H: G ratios were consistently 1:2. SCXRD showed that these inclusion compounds crystallize in the monoclinic space group P21/c (their analyses were performed in P21/c for H1⋅2(TMU) and P21/n for both H2·2(TMU) and H3·2(TMU)). The TMU molecules in H1·2(TMU) and H2·2(TMU) are situated in unidirectional channels, while in H3·2(TMU), these are located in an X-shaped cavity, with four guest molecules in each one. In each complex are, additionally, observed (guest)C‒H···π(host) as well as classical and non-classical hydrogen bonding interactions, which assist in the stabilization of these complexes. Hirshfeld surfaces were employed to quantify the (host)H···O(guest) interatomic interactions in the three complexes, and percentages calculated range between 7.8 and 10.3%. All of the complexes have high thermal stabilities (Ton > 80 °C), and the expected mass losses correlate closely with those measured upon releasing the guests from the crystals by the application of heat. A consideration of the kinetics of the guest release event shows that the activation energies for the desolvation of H1·2(TMU), H2·2(TMU) and H3·2(TMU) range from 128.6 to 149.4 kJ·mol‒1. The complexes H2·2(TMU) and H3·2(TMU) are novel, and no data as furnished here for these clathrates may be found in the literature. However, H1·2(TMU) has previously been reported, and while the X-ray structure of H1·2(TMU) from that worked agreed closely with our findings, distinct differences were observed in both the Ton and the percentage of guest release in thermal experiments. Overall, this investigation has provided insights into the various supramolecular aspects of these TMU-containing inclusion compounds, as well as their relative thermal stabilities and the novel investigation into the kinetics of their guest desolvation events, while applicable comparisons of the results obtained here were made with that of a previous report on the H1·2(TMU) inclusion compound.