Styrene Purification by Guest-Induced Restructuring of Pillar[6]arene

The separation of styrene (St) and ethylbenzene (EB) mixtures is important in the chemical industry. Here, we explore the St and EB adsorption selectivity of two pillar-shaped macrocyclic pillar[n]arenes (EtP5 and EtP6; n = 5 and 6). Both crystalline and amorphous EtP6 can capture St from a St-EB mixture with remarkably high selectivity. We show that EtP6 can be used to separate St from a 50:50 v/v St:EB mixture, yielding in a single adsorption cycle St with a purity of >99%. Single-crystal structures, powder X-ray diffraction patterns, and molecular simulations all suggest that this selectivity is due to a guest-induced structural change in EtP6 rather than a simple cavity/pore size effect. This restructuring means that the material “self-heals” upon each recrystallization, and St separation can be carried out over multiple cycles with no loss of performance.

S tyrene (St) is an important aromatic feedstock in the chemical industry. 1 More than 80% of St production is used for polymerization or copolymerization to produce thermoplastics, synthetic rubbers, and resins. St is mainly produced by dehydrogenation of ethylbenzene (EB): 2 after dehydrogenation, the product stream still contains a large fraction (20−40%) of unreacted EB that must be removed. 3 Separation by conventional distillation on an industrial scale is not practical because of the small difference in boiling point (9 K) between St (bp 418.3 K) and EB (bp 409.3 K). Currently, the preferred technology is extractive distillation and vacuum distillation, performed in the presence of polymerization inhibitors like phenylene-diamines or dinitrophenols. 5 However, this is energetically intensive and most of the energy used in the production of St can be accounted for by the separation process. Also, small amounts of impurities with similar boiling points such as toluene (bp 393 K) and oxylene (bp 418 K) must also be removed from the St-EB product stream, further complicating the procedure. 3,4 An alternative and potentially more energy-efficient separation strategy is to exploit the molecular, chemical, and geometrical differences of St and EB in an adsorptive separation, for example by using nanoporous materials such as metal−organic frameworks (MOFs) or zeolites. 5−10 Adsorptive technologies do not require elevated temperatures, and hence the problem of side reactions can be circumvented, providing that the adsorbent itself is not reactive. However, the molecular sizes of St and EB are very similar (Figure 1a), making it difficult to find a suitable porous material for their separation. In principle, crystallizationbased separations can also be energy efficient. Unlike adsorptive separation, which uses materials with pre-existing pores, the "pores" in crystallization separation are created during assembly of the crystalline inclusion compound. These "pores" are generally only stable when occupied by guest molecules. 11 To our knowledge, no such separation method has been reported for styrene purification. Pillar[n]arenes (n = 5−15) were first reported in 2008 as a novel class of supramolecular host. 12−22 The host−guest properties of pillar[n]arenes have been investigated intensively and they have been applied in the fabrication of various supramolecular systems, such as interlocked structures, molecular machines, supramolecular polymers, and supramolecular amphiphiles. 12−20 There are only a few occasions where pillar[n]arenes have been used as solid materials. For example, Yang et al. reported a pillar [5]arene-based supramolecular organic framework which has selective adsorption of CO 2 over N 2 at room temperature. 21 Ogoshi et al. reported that a pillar [5]arene functionalized with ethyl groups could encapsulate n-alkanes in its cavity in the solid state. 22 Here, we investigate two pillararenes with different cavity sizes, perethylated pillar [5]arene (EtP5) and pillar [6]arene (EtP6) (Figure 1b), as adsorptive separation materials to separate St and EB. We show that both crystalline and amorphous EtP6 can   St molecules are located in the extrinsic pores between distorted EtP6 molecules in a crystal structure that transforms to accommodate the St guest. The separation process for EtP6 is a solid-state recrystallization separation, rather than an adsorptive separation. This process is quite different from separations that use porous adsorptive materials such as MOFs or zeolites, where intrinsic pore size and shape plays the determining role.
EtP5 and EtP6 were synthesized by a previously reported method. 19 Single crystals of EtP5 and EtP6 loaded with St and EB were obtained by slow evaporation of a St or EB solution of the pillararene, respectively. Despite numerous attempts, no diffractable single crystals could be grown of St-loaded EtP5. By contrast, EB forms a 1:1 host−guest complex with EtP5 (EB@ EtP5; Figure 2a, left). In EB@EtP5, an EB molecule is threaded into the cavity of EtP5 with its ethyl group in the cavity center. This gives rise to CH···π interactions between the ethyl group on EB and the benzene rings on EtP5. The pentagonal structure of EtP5 forms infinite 1D channels with EB in the channels ( Figure  2a, right). In the analogous solvated crystal structure of EtP6 (EB@EtP6), an EB molecule is located in the center of the cavity. This is stabilized by π−π stacking between EB and benzene rings on EtP6, also forming a 1:1 host−guest complex (Figure 2b, left). The hexagonal pillar structure of EtP6 contributes to the formation of infinite intrinsic diagonal 1D channels with EB in the channels (Figure 2b, right). The solvated crystal structure of EtP6 with St (St@EtP6) is markedly different. Two opposite repeating units of EtP6 are turned perpendicular to their adjacent units, giving EtP6 a more deformed hexagonal structure with a cavity that is too small to accommodate a St guest molecule (Figure 2c, left). However, the EtP6 molecule does form infinite extrinsic 2D channels between the macrocycles along the a and b axes, with St molecules accommodated in these channels (Figure 2c, middle and right).
From these three crystal structures alone, it might be expected that EtP6 has potential to selectively capture EB from a St-EB mixture due to its good fit in the EtP6 intrinsic cavity. Solid− vapor adsorption experiments were performed to test this. Desolvated crystalline EtP5 (EtP5α) and EtP6 (EtP6α) were first prepared as adsorptive materials (for details, see the Supporting Information). Neither EtP5α nor EtP6α showed significantly different adsorption rates for St or EB vapor ( Figures S7 and S9). For EtP5α, EB was adsorbed slightly faster than St whereas EtP6α adsorbed both St and EB at similar rates. The major difference is the amount of St and EB adsorbed by EtP6α, which is much higher than EtP5α ( Figure S7). To monitor the adsorption of neat St and neat EB by EtP5α and EtP6α, we carried out in situ PXRD studies. The PXRD patterns of EtP5α did not change after adsorption of St or EB, indicating that the structure of EtP5α did not change after exposure to St or EB ( Figure S8). We concluded that EB or St was not, in fact, adsorbed in the bulk by EtP5α. Instead, we ascribe the small, substoichiometric uptake of St (and EB) by EtP5α to surface adsorption on the crystals. By contrast, the PXRD patterns of EtP6α after adsorption of St or EB were different from EtP6α and in good agreement with the simulated PXRD patterns for St@EtP6 and EB@EtP6, respectively ( Figure S9b). Hence, the adsorption of St or EB by EtP6α triggers a crystal transformation from EtP6α to St@EtP6 or EB@EtP6, respectively ( Figure  S9c). EtP6 is thus a more promising material than crystalline EtP5 for the capture of St or EB.
To investigate whether EtP6α could discriminate between a mixture of EB and St, we carried out time-dependent EtP6α solid−vapor sorption experiments for a St-EB mixture (the St-EB mixture used is always 50:50 v/v). The uptake rates of St and EB in EtP6α were essentially the same over the first hour ( Figure  3a). Surprisingly, however, the uptake of St increased dramatically after the first hour and quickly increased to approximately 1 mol/EtP6 after 4 h. The uptake of EB slowly decreased after the first hour, presumably because it was being displaced by St molecules and forced out of the intrinsic cavity as the structural transformation occurred. Gas chromatography was used to deduce the ratio of St and EB adsorbed by EtP6α over the course of the experiment (Figures S10 and S11). This showed that EtP6α could selectively capture St from a St-EB mixture, contrary to our initial hypothesis that EB might be selectively adsorbed. A time-dependent PXRD experiment was performed to monitor the transformation of EtP6α after exposure to a St-EB mixture (Figure 3b). Comparison of the PXRD pattern recorded after adsorption of the St-EB mixture

Journal of the American Chemical Society
Communication vapor for 1 h with the simulated profiles of St@EtP6 and EB@ EtP6 suggests the coexistence of St@EtP6 and EB@EtP6 phases after the initial exposure period ( Figure S12). With increasing adsorption time, the PXRD pattern becomes more like St@EtP6 until it becomes fully consistent with the St@EtP6 structure. A putative mechanism, consistent with these observations is as follows: initially, the uptake rates of both St or EB in EtP6α are similar, and both St and EB can occupy empty EtP6α crystals, transforming them into two new polymorphs, St@EtP6 and EB@EtP6. However, St@EtP6 is more stable than EB@EtP6, and over time, St molecules diffuse into EB@EtP6, displacing EB molecules and transforming EB@EtP6 into St@EtP6.
We then carried out additional time-dependent solid−vapor sorption experiments using EB-loaded EB@EtP6 and St-loaded St@EtP6 to further investigate this guest-induced structural transformation. The uptake of St increased rapidly within 1 h when a St-EB vapor mixture was diffused into preloaded EB@EtP6 (Figure 4a). A saturation point was reached after about 7 h. EB, which was originally located in the 1D channels of EB@EtP6, was displaced by St and its loading decreased as the St content increased. Time-dependent PXRD patterns (Figure 4b) indicate the guest-induced crystal-to-crystal transformation from EB@EtP6 to St@EtP6 over time. Hence, even when EB molecules were fully preloaded into crystals of EtP6, St molecules in the St-EB mixture can penetrate these crystals and replace EB. By contrast, when fully solvated St@EtP6 is exposed to the mixture of St-EB vapor, these crystals maintain a high loading of St and the EB uptake is negligible (Figure 4c). As expected, the PXRD pattern did not change at all, indicating that EB molecules in the St-EB mixture cannot replace St molecules in St@EtP6 (Figure 4d,e). This suggests that the St@EtP6 phase is more thermodynamically stable than the EB@EtP6 phase.
To rationalize this, we performed electronic density functional theory calculations to understand the structural stability of three EtP6 structures loaded with styrene molecules, i.e., (1) the experimental St@EtP6 phase; (2) the experimental EB@EtP6 phase with EB replaced by St, and; (3) the experimental EtP6α phase with St artificially inserted (the molar ratio of St and EtP6 = 1:1; St molecules were placed at the center of each EtP6). Table S4 summarizes the lattice energies determined for these three styrene-loaded structures. The experimentally observed St@EtP6 structure has the lowest lattice energy (−529 kJ/mol) among the three polymorphs, explaining its preferred formation.
To show that EtP6 has practical potential for styrene purification, we developed a procedure to obtain St with high purity from a St-EB mixture using crystalline EtP6 as the adsorbent. After adsorption of a St-EB vapor mixture, crystalline EtP6 was heated at 40°C for 30 min to remove any unbound St or EB molecules adsorbed on the crystal surface. St could then be released from the selectively formed St@EtP6 phase with a purity of over 99% (Figure 5b).
One bane of adsorbent technology is decreased performance over time, because of either instability of the porous framework or fouling. To be practically useful, an adsorbent must perform well over multiple cycles without any degradation. Thus, an important question was whether the resultant desolvated EtP6 crystals can still selectively capture St from a St-EB mixture in a second cycle, after St is completely removed from the St@EtP6 crystals. Indeed, the desolvated St@EtP6 was shown by PXRD experiments to be a new polymorph, EtP6β ( Figure S17). Nonetheless, time-dependent solid−vapor experiments using the same St-EB vapor mixture showed that EtP6β has adsorption properties for St and EB that are just like EtP6α, indicating that EtP6β can also selectively capture St from a St-EB mixture ( Figure S20). PXRD experiments suggest the same selectivity mechanism as EtP6α and EB@EtP6 ( Figure S21). Since the selectivity seems to be unaffected by this polymorphism in EtP6, we further investigated whether amorphous EtP6 might also

Journal of the American Chemical Society
Communication have the same performance. Adsorption and PXRD experiments showed that amorphous EtP6 could also selectively adsorb St (Figures S24 and S25). As such, it is not necessary to use crystalline EtP6 for styrene purification and EtP6 can be recycled multiple times without losing its styrene selectivity or adsorption capacity (Figure 5c). Effectively, this crystallization route is a "self-healing" system: the separation works irrespective of the starting EtP6 structure. This is different from processes involving extended porous frameworks where loss of porosity or phase changes are often irreversible and catastrophic.
In summary, we have investigated the adsorptive properties of two easily obtained pillar[n]arenes, EtP5 and EtP6, toward St and EB. EtP6 was found to be a much better absorbent for both St and EB, and either crystalline or amorphous EtP6 can selectively capture St from a St-EB mixture. This selectivity arises from the guest-induced selective structural change of EtP6 rather than the suitable cavity size. Compared with other small molecule organic separation materials, such as intrinsically porous cage compounds, 23 the separation process for EtP6 is closer to a crystallization separation, rather than an adsorptive separation. While the separation of St and EB has been achieved in porous extended frameworks, such as MOFs, this new molecular approach offers potential advantages. For example, EtP6 is soluble, is easy to synthesize, and has better chemical stability than many crystalline MOFs and COFs. While the overall uptake capacity in EtP6 is relatively low compared with porous extended frameworks, and the uptake kinetics are relatively slow, St can be separated with high purity in just one cycle, which is highly desirable. Future work will attempt to increase the uptake capacity and adsorption kinetics without losing the remarkable selectivity, for example by cocrystallization of two or three different pillar[n]arenes. Other hydrocarbon separations, such as the separation of xylene isomers, are also under investigation.

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.6b13300.