Jacky S.
Bouanga Boudiombo
,
Hong
Su
,
Neil
Ravenscroft
,
Susan A.
Bourne
and
Luigi R.
Nassimbeni
*
Centre for Supramolecular Chemistry Research, Department of Chemistry, University of Cape Town, Rondebosch 7701, South Africa. E-mail: luigi.nassimbeni@uct.ac.za; Tel: +27 21 650 5893
First published on 22nd June 2020
The six xylenol (XYL) isomers can be separated by selective enclathration with the host 4,4-isopropylidene bisphenol, H1. Crystal structures were elucidated for the following single and mixed guest inclusion compounds with H1: H1·34XYL (I), H1·35XYL (II), H1·23XYL·26XYL (III), H1·23XYL·35XYL (IV), where the xylenol isomers are abbreviated as, for example 34XYL for 3,4-xylenol. The crystal structures of selected H1·xylenols showed that there is extensive host⋯host and host⋯guest hydrogen bonding. Competition experiments with equimolar mixtures of pairs of xylenols (XYL) showed that the preference for inclusion was in the sequence 34XYL > 35XYL > 26XYL > 23XYL > 25XYL > 24XYL. By analogy to the Dutch resolution method (in which families of resolving agents are used to achieve chiral separations), two host compounds similar to H1 were used in pairs with H1 to improve the selectivity of the xylenols. 4,4′-(9-Fluorenylidene)bisphenol, H2, and 4,4′(cyclohexylidene)bisphenol, H3, were used in pairs with H1 and were shown to enhance the selectivity of a given xylenol which had been poorly separated by H1 alone. The crystal structure was elucidated for an unusual mixed host–mixed guest inclusion compound, H1·H2·26XYL/35XYL (V).
Host–guest (or inclusion) chemistry has proved to be a useful methodology for the separation of closely related molecular species.2 When a host compound, H, is exposed to a mixture of guest molecules, A, B, C… this may result in an inclusion compound
H + n1A + n2B + n3C… → H·Am |
The above represents an ideal case, in which A is exclusively selected. In practice this occurs seldom, and where there are many components a common strategy is to take the guests in pairs and analyse the final product by a suitable technique which yields its stoichiometry. By analysing the combination of all possible pairs of guests, it is possible to obtain the preferential affinity sequence of the host H for the complete series of guests A, B, C…
This methodology is driven by the phenomenon of molecular recognition, which is central to host–guest chemistry and crystal engineering. The special factors that lead to a suitable fit between host and guest molecules arise from the sum of multiple secondary interactions that impinge upon the molecular system. These secondary bonds are often directional, resulting in specificity and allowing a host molecule to discriminate between a given guest in a mixture of guests. This makes the host H selective, which is crucial to separation processes.
An important example of the separation of isomers is the petrochemical industry in which isomers of the C8 hydrocarbons (ortho-, meta-, para-, xylenes and ethyl benzene), the cresols and the xylenols are produced in large quantities from the catalytic reforming of crude oil. A well-known case is the separation of the isomers of xylene, which have similar normal boiling points ranging from 138.1 °C to 144.4 °C, rendering fractional distillation inefficient. This topic is the subject of a comprehensive review,3 which discusses various materials employed in the separation techniques that include metal–organic frameworks, zeolites and organic host molecules. Other contributions describe the optimal synthesis of p-xylene separation4 and the use of Werner clathrates for the separation of xylenes from in the vapour state and the concomitant kinetics of adsorption.5
Hydrocarbons are not the only compounds that have been separated by enclathration and there are several recent examples where mixtures of both aliphatic and aromatic compounds have been selectively included into host–guest complexes.6–10
A particularly challenging problem for the separation of isomers is that of enantiomer resolution of a racemic modification. Here, the two enantiomeric components have identical physical properties such as melting point, boiling points, refractive index, density, dipole moment, and only differ in their response to polarized light. One successful strategy for separating enantiomers combines a chiral resolving agent which forms a compound preferentially with either one or the other enantiomer. This separation process is not always routine and may yield incomplete resolution. However, a significant advancement was made by the discovery by T. Vries et al.11 who used a combinational approach of related “families” of resolving agents to improve the resolution of racemates. This has been summarized in the Handbook of Optical Resolutions edited by D. Kozma12 and is known as the “Dutch Resolution Method”. We were inspired by this idea to extend the usual host–guest method of separation of isomers to an analogy of the Dutch resolution method, in which we employed pairs of similar host compounds for the separation of isomers from binary mixtures, with the aim of obtaining enhanced selectivity of the guest species.
In this work, we aimed to achieve separation of the six xylenol isomers, by selective inclusion using three organic host compounds which contain the common bisphenol moiety. The structures of the host compounds are shown in Scheme 1, which also presents the six xylenol isomers with their normal boiling points, as well as the abbreviations used for these compounds.
Xylenol (XYL) | 23XYL | 24XYL | 25XYL | 26XYL | 34XYL | 35XYL |
---|---|---|---|---|---|---|
24XYL | [1] | |||||
23XYL-90 | ||||||
24XYL-10 | ||||||
25XYL | [2] | [3] | ||||
23XYL-55 | Host | |||||
25XYL-45 | ||||||
26XYL | [4] | [5] | [6] | |||
23XYL-40 | Host | 25XYL-6 | ||||
26XYL-60 | 26XYL-94 | |||||
34XYL | [7] | [8] | [9] | [10] | ||
23XYL-10 | 24XYL-16 | 25XYL-6 | 26XYL-3 | |||
34XYL-90 | 34XYL-84 | 34XYL-94 | 34XYL-97 | |||
35XYL | [11] | [12] | [13] | [14] | [15] | |
23XYL-46 | 24XYL-9 | 25XYL-7 | 26XYL-12 | 34XYL-94 | ||
35XYL-54 | 35XYL-91 | 35XYL-93 | 35XYL-88 | 35XYL-6 |
Structure | I | II | III | IV |
---|---|---|---|---|
Compound | H1·34XYL | H1·35XYL | H1·23XYL/26XYL | H1·23XYL/35XYL |
Formula asymm. unit | (C15H16O2)·C8H10O | (C15H16O2)·C8H10O | 2(C15H16O2)·2C8H10O | 4(C15H16O2)·4C8H10O |
M [g mol−1] | 350.4 | 350.4 | 700.9 | 1402 |
Data collection temp T [K] | 173(2) | 173(2) | 173(2) | 173(2) |
Crystal shape and size [mm] | Orange block, 0.23 × 0.28 × 0.30 | Orange block, 0.05 × 0.06 × 0.10 | Orange block, 0.18 × 0.25 × 0.28 | Colourless needle, 0.05 × 0.09 × 0.48 |
Crystal system | Monoclinic | Monoclinic | Monoclinic | Monoclinic |
Space group | P21/c | P21/c | C2/c | Pn |
a [Å] | 6.4218(6) | 11.783(3) | 30.650(2) | 6.3007(3) |
b [Å] | 14.8289(14) | 11.175(2) | 6.3048(5) | 20.7627(11) |
c [Å] | 20.2173(19) | 15.163(3) | 39.852(3) | 29.9858(14) |
β [°] | 96.132(2) | 93.080(4) | 91.359(2) | 95.787(2) |
Volume [Å3] | 1914.2(3) | 1993.8(8) | 7699(1) | 3902.7(3) |
Z | 4 | 4 | 8 | 2 |
D c, calc. density [g cm−3] | 1.216 | 1.209 | 1.193 | |
Absorption coefficient [mm−1] | 0.079 | 0.079 | 0.078 | |
F(000) | 752 | 3008 | 1504 | |
θ range | 1.706–28.339 | 1.329–27.910 | 1.962–26.385 | |
Reflections collected | 30151 | 101365 | 69771 | |
No. independent reflections | 4783 | 9208 | 14770 | |
No. reflections with l > 2sigma(I) | 3382 | 8537 | 14075 | |
R int | 0.0654 | 0.0255 | 0.0395 | |
Final R indices, R1, wR2 [I > 2sigma(I)] | 0.0486, 0.1095 | 0.0387, 0.0996 | 0.0412, 0.1009 | |
R indices (all data), R1, wR2 | 0.0753, 0.1226 | 0.0417, 0.1021 | 0.0440, 0.1025 | |
Max, min residual electron density (e Å−3) | 0.220, −0.225 | 0.309, −0.175 | 0.263, −0.188 |
The crystals were harvested and blotted dry and subjected to NMR and single crystal X-ray diffraction analysis. The crystals were not washed with a different solvent for fear of partial dissolution and loss of the included xylenol guests.
Integration of the peaks of the methyl substituents of the isomers were used to determine their relative proportions of guests in the different samples. Fig. 1 shows an overlay of the expansion of the methyl region of the 1H NMR spectra of the guests (23XYL and 26XYL) and the host and guests. Fig. 1a shows the methyl groups of the host compound at 1.15 ppm. Fig. 1b has CH3-2 at 2.03 ppm and CH3-3 at 2.18 ppm for 23XYL and Fig. 1c has CH3-2 and CH3-6 at 2.15 ppm. These diagnostic peaks were used to determine the relative ratio of 60/40% for the 23XYL/26XYL guests in the host–guests mixture (Fig. 1d). The same procedure was applied to all other mixtures and gave the results shown in Table 1. Representative NMR spectra are included in the ESI.† When two isomers had overlapping methyl peaks in the 1H spectrum, but the 13C NMR signals were resolved, then quantification was performed by integration of the relevant HSQC cross peaks. Thus, NMR analysis elucidated the relative ratio of the guests, whereas the percentage host–host and host–guest was also obtained when mixed hosts method was used.
Fig. 1 Overlay of the expansion of the 1H NMR methyl region for (a) H1, (b) 23XYL, (c) 26XYL, (d) complex of H1 + 23XYL + 26XYL obtained in experiment [4]. |
Experiments [3] and [5] resulted in the recrystallisation of the empty host structure (sometimes referred to as the apohost structure). Although experiments [1] and [6] indicated high selectivity for 23XYL and 26XYL respectively, they yielded poor quality crystals which could not be used for data collection. Experiments [7], [8], [9] [10] and [15] showed a high selectivity for 34XYL, and a crystal was selected from [15] for X-ray analysis (crystal structure I). 35XYL appeared to be selected in experiments [12], [13] and [14] and inclusion compound II was selected for crystal structure analysis from the latter. Experiments [4] and [11] resulted in the crystallization of mixed guest inclusion compounds III and IV respectively. Experiment [2] also showed evidence of a mixed-guest compound, but the crystals did not diffract adequately for single crystal analysis. In total, we elucidated four crystal structures from the competition experiments given in Table 1. Their crystal data and refinement parameters are listed in Table 2. The identity of the guests included can be unequivocally confirmed for those cases where single crystal diffraction was performed, but we note that similar confirmation is not possible for the experiments where single crystal structures were not obtained. However, we saw no evidence, on inspection under polarized light, of two or more crystalline phases being produced in any of these or subsequent experiments.
Structure I, H1·34XYL, crystallizes in P21/c with Z = 4. The packing is characterized by chains of H1 which are stabilized by O–H⋯O(H) H-bonds, and in addition, form H-bonded rings with the 34XYL guest. The packing is shown in Fig. 2, and may be described by graph-set analysis22,23 as C11(12)[R33(6)]. The H1 chains propagate in the [010] direction and the 34XYL guests are in the loops of the twisted chains.
A search of the Cambridge Structural Database24 found 11 structures in which H1 had formed a co-crystal or solvate. One of these, refcode SIXDOS, is the co-crystal of H1 with p-cresol.25 This compound is isostructural with compound I, having almost identical unit cell parameters and packing arrangements. Although I has an extra methyl group at the meta-position, the same hydrogen-bonding motifs can be formed in both structures, which accounts for this similarity. The CSD search revealed 20 solvate or co-crystal structures with H2 and 47 with H3, though none are similar to the structures reported here.
Structure II, H1·35XYL could not be refined satisfactorily and we only report the cell parameters in Table 2. However, the structure could be solved sufficiently to show that it bears certain packing similarities with I, in that the host forms a series of chains running along [010], with heavily disordered 35XYL guests forming hydrogen bonded bridges between chains (Fig. 3).
Fig. 3 Packing of compound II viewed along [100]. Colour coding as in Fig. 2. |
Structure III crystallizes in the space group C2/c. The asymmetric unit consists of two H1 hosts, a disordered 23XYL (86% site occupancy) and 26XYL (14% site occupancy) and one 26XYL with full site occupancy. The ratio of H1:23XYL:26XYL is thus 1:0.43:0.57 which is in good agreement with the NMR data in Table 1 for experiment [4]. The packing, shown in Fig. 4 shows the H1 hosts packed in double layers running along [001]. The interior of the double layer is hydrophobic, containing the gem-dimethyl groups, while the outer sides of the double layer may be deemed hydrophilic in that it contains the hydroxyl moieties which hydrogen bond with the 23XYL and 26XYL guests in a series of R33(6) hydrogen bonded rings.
Fig. 4 Packing of III (H1·23XYL/26XYL) viewed along [100] showing the two layers formed within the structure. |
The refinement of structure IV (H1·23XYL/35XYL) proved difficult. Although the synthesis was repeated several times, the resultant crystals were of poor quality. The best preparation yielded crystals that did not extinguish completely under polarized light and the structure was initially solved in the space group P1. The resultant structure was checked for higher symmetry by the program Platon26 which strongly suggested the space group Pn, duly adopted. Platon further identified twinning which was resolved by application of the appropriate twin law.
The asymmetric unit contains four H1 hosts, two 23XYL and two 35XYL guest, all crystallographically independent. The ratio of H1:23XYL:35XYL is thus 1:0.50:0.50, which is in good agreement with the ratio 1:0.46:0.54 determined by NMR on the bulk sample (Table 1, experiment [11]). The packing, shown in Fig. 5, bears strong resemblances to that shown for structure III, in that one notes a host double layer with hydrophobic and hydrophilic surfaces. The latter features H-bonded 23XYL and 35XYL guests. The hosts and guests form chains running along [010].
Fig. 5 Packing of IV showing (a) hydrogen bonding between host and host, and host and guest, (b) host forms H-bonded chains (in orange) parallel to the b-axis. |
The conformation of the host H1 in all the structures elucidated is reported in Table S1† which lists the torsion angles τ1 and τ2 are defined as τ1 = (a − b − c − d) and τ2 = (b − c − d − e), Scheme 2. The variations of the torsion angles τ1 and τ2 are relatively small, showing that the host conformation is fairly constant and is thus does not play a key role in the selectivity of xylenol isomers.
Experiments [16], [17], and [18] tested the selectivity H2 alone in competition experiments of 23XYL/26XYL, 23XYL/35XYL, and 26XYL/35XYL. These were chosen because H1 had shown poor selectivity in the first two combinations while the third combination yielded a strong preference for 35XYL (Table 1). The results testing the efficacy of H2 are reported in Table 3, and show a similar lack of selectivity for 23XYL/26XYL, a preference for 35XYL in 26XYL/35XYL and an inversion in selectivity, towards 23XYL in 23XYL/35XYL.
Composition | H1 | H2 | 23XYL | 26XYL | 35XYL |
---|---|---|---|---|---|
Expt [16] start | — | 100 | 50 | 50 | — |
End | — | 100 | 50 | 50 | — |
Expt [17] start | — | 100 | 50 | — | 50 |
End | — | 100 | 81 | — | 19 |
Expt [18] start | — | 100 | — | 50 | 50 |
End | — | 100 | — | 28 | 72 |
Expt [19] start | 50 | 50 | 50 | 50 | — |
End | 98 | 2 | 55 | 45 | — |
Expt [20] start | 50 | 50 | 50 | — | 50 |
End | 48 | 52 | 88 | — | 12 |
Expt [21] start | 50 | 50 | — | 50 | 50 |
End | 51 | 49 | — | 38 | 62 |
Experiments [19]–[21] detail the results of the mixed host/mixed guest competition experiments with H1 and H2. An interesting aspect of the Dutch resolution method is that the final crystalline product may have host–host and host–guest stoichiometries that differ from those of the starting solution. This was particularly evident in experiment [19] where an equimolar mixture of H1 and H2 resulted in crystals containing almost entirely H1, while the guest selectivity was poor in this case. Experiments [20] and [21] retained the H1:H2 ratio and showed altered selectivity towards the xylenols compared to single host experiments. In experiment [20], the competition with 23XYL/35XYL, the selectivity changes from 54% 35XYL (with H1), or 81% 23XYL (with H2) to 88% 23XYL in the mixed host system. In experiment [21], the competition with 26XYL/35XYL, the selectivity changes from 88% 35XYL (with H1), or 72% 35XYL (with H2) to 62% 35XYL in the mixed host system. Understanding these results would be enhanced if one were able to obtain single crystals of each of these outcomes. Unfortunately, this is often not possible, but we were able to obtain a crystal structure from the product of experiment [21].
Structure V is unusual in that it was derived from a solution containing an equimolar mixture of H1 and H2, and an equimolar mixture of 26XYL and 35XYL, and retained all four species in the crystalline product. The NMR results showed that the bulk sample contained an almost equimolar amount of H1 and H2, which had enclathrated an unequal mixture of 26XYL and 35XYL. The resulting product crystallized in P21/c and the asymmetric unit contains two H1 molecules, two H2 molecules and a disordered 26XYL/35XYL sharing the same site in unequal proportions (26XYL:35XYL = 0.34:0.66, in good agreement with NMR data). The important feature of this structure is that the disordered 26XYL/35XYL guest is surrounded by four host molecules (one pair of H1 and one pair of H2) as shown in Fig. 6. The crystallographic data parameters of V are given in Table 4. Mixed-host, mixed-guest crystal structures are still relatively rarely reported in the literature. One of us has previously reported a similar example, in which a family of four similar diol hosts were used singly and in pairs in an attempt to enhance the resolution of 2-butylamine.27 Although there was no improvement in the enantiomeric resolution, a structure of two hosts with two guests was reported.
Fig. 6 Asymmetric unit of structure V showing 26XYL/35XYL enclathrated by a mixture of the hosts H1 and H2 (atom colours as per Fig. 2. Hydrogen atoms have been omitted for clarity). |
V | |
---|---|
Compound | H1·H2·26XYL/35XYL |
Formula asymm. unit | 2(C15H16O2)2(C25H18O2)·C8H10O |
M [g mol−1] | 1279.5 |
Data collection temp T [K] | 173(2) |
Crystal shape and size [mm] | Orange block, 0.18 × 0.20 × 0.32 |
Crystal system | Monoclinic |
Space group | P21/c |
a [Å] | 21.0580(9) |
b [Å] | 15.5380(6) |
c [Å] | 21.249(1) |
β [°] | 98.481(1) |
Volume [Å3] | 6876.6(5) |
Z | 4 |
D c, calc. density [g cm−3] | 1.236 |
Absorption coefficient [mm−1] | 0.079 |
F(000) | 2712 |
θ range | 0.978–28.370 |
Reflections collected | 78563 |
No. independent reflections | 17204 |
No. reflections with l > 2sigma(I) | 11140 |
R int | 0.0707 |
Final R indices, R1, wR2 [I > 2sigma(I)] | 0.0500, 0.1110 |
R indices (all data), R1, wR2 | 0.0894, 0.1300 |
Max, min residual electron density (e Å−3) | 0.239, −0.208 |
Composition | H1 | H3 | 23XYL | 26XYL |
---|---|---|---|---|
Expt [22] start | 90 | 10 | 50 | 50 |
End | 100 | 0 | 55 | 45 |
Expt [23] start | 80 | 20 | 50 | 50 |
End | 73 | 27 | 65 | 35 |
Expt [24] start | 70 | 30 | 50 | 50 |
End | 38 | 62 | 72 | 28 |
Expt [25] start | 60 | 40 | 50 | 50 |
End | 9 | 91 | 90 | 10 |
Expt [26] start | 50 | 50 | 50 | 50 |
End | 6 | 94 | 92 | 8 |
The observation that the relative amounts of both host and guest vary in this way led us to explore the phenomenon further by varying the starting ratio of H1:H3 systematically for each pairwise combination of xylenol isomers. The results for the H1/H3 and the guest pair 23XYL/26XYL are reported in Table 5 and shown graphically in Fig. 7. The results for the other isomer pairs are given in the ESI† as Tables S4 and S5.
Fig. 7 Trends observed in the competition experiments varying both host and guest ratios. (a) Host H1 decreases while H3 is selected; (b) guest 23XYL is consistently selected across all compositions. |
Fig. 7a shows H1 starting at 90% and decreasing linearly to 50%, while H3 starts at 10% and increases to 50%. The trend of the experiments shows H1 has decreased at the expense of H3 in the products obtained. The concomitant selectivity of the 23XYL/26XYL (Fig. 7b) shows a steady increase in 23XYL and decrease in 26XYL. There is thus a direct correlation between H1/H3 and the corresponding selectivity of 23XYL/26XYL. Similar results are obtained for H1/H3 and the guests 23XYL/35XYL (Table S4, Fig. S1†). However, Table S5† shows that changes in the H1/H3 composition had no effect in the 26XYL/35XYL selectivity, with 35XYL being exclusively enclathrated.
The crystal structures I, II, III, and IV obtained with H1 are all stabilized by extensive hydrogen bonded networks, linking adjacent host molecules into chains which also hydrogen bonded the captured guests. The strength of the hydrogen bonds, as estimated by the O(donor)⋯O(acceptor) distances vary from 2.64 Å to 2.95 Å, and may be considered to change from strong to weak.32 These may be regarded as a constant feature throughout the structures elucidated. The synergistic selectivity effects which occur with the addition of a second host molecule may be attributed to the packing effects brought about by the different moieties (fluorenylidene in H2 and cyclohexylidene in H3). Confirmation of this effect will be sought in further work aiming to isolate crystals of these mixed-host, mixed-guest compounds.
Footnote |
† Electronic supplementary information (ESI) available: Tables S1–S5, Fig. S1–S37. Crystallographic data for this paper have been deposited with the CCDC, accession numbers 1994170 and 1994172–1994174. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d0ce00510j |
This journal is © The Royal Society of Chemistry 2020 |