Catechol-based macrocyclic aromatic ether-sulfones: Synthesis, characterization and ring-opening polymerization

Cyclocondensation between 4,4'-bis(4"-chlorobenzenesulfonyl)biphenyl and catechol, with subsequent chromatographic separation of the reaction products, led to the isolation of four novel ether-sulfone macrocycles (cyclic dimer, -trimer, -tetramer and -pentamer). Similarly, cyclocondensation of catechol with a novel seven-ring diketone/disulfone monomer allowed the isolation of the two new aromatic ether-ketone-sulfone macrocycles, a cyclic monomer and a cyclic dimer. Transannular shielding and deshielding effects in the cyclic monomer produce substantial chemical shift differences for chemically equivalent protons in the 1 H NMR spectra of the cyclic monomer and -dimer. Fluoride-initiated ring-opening polymerization of the ether-sulfone cyclic trimer affords a novel, high-molecular weight poly(ether-sulfone).


Introduction
The molecular structures of high-temperature engineering thermoplastics generally comprise linear chains of aromatic rings, linked together by thermo-oxidatively stable units such as direct arene-arene bonds, ether, ketone, sulfone, amide or imide groups. 1,2Aromatic poly(ether-sulfone)s, exemplified in Chart 1, comprise an industrially-significant class of such polymers: these materials may in principle be accessed either by electrophilic chemistry (polysulfonylation) 3 or by activated nucleophilic substitution (polyetherification) 4 at the aromatic rings.In practice the nucleophilic route, involving the displacement of sulfone-activated chloride by phenoxide ion, has generally proved more selective and versatile both in the laboratory and in commercial production. 5,6A more recently-discovered approach to polymers of this type involves the ring-opening polymerization (ROP) of macrocyclic aromatic (ether-sulfone)s. 7,8Macrocyclic oligomers are present in small, equilibrium quantities (typically 1-3 wt%) in many linear step-growth polymers, including the poly(ethersulfone)s, 9,10,11 but they may also be obtained in high yield either by cyclo-condensation of monomers under pseudo-high-dilution conditions, [12][13][14][15][16] or by ring-closing depolymerization of high molecular weight (MW) polymer at low concentration in solution. 7,17,18art 1. Industrially-significant aromatic poly(ether sulfone)s, showing current (2020) trade-names and manufacturers.
0][21] An important feature of such a system is that at high concentrations the equilibrium lies heavily in the direction of polymer, whilst at low concentrations it shifts to favor macrocylic oligomers.For example, under very concentrated or even solvent-free reaction conditions an equilibrated system would typically contain ca.98 wt% polymer and only ca. 2% macrocycles, whilst under dilute conditions (say 1 wt% of solutes) the system can easily comprise >90% of macrocycles, and may even consist exclusively of these. 22Thus, if a neat macrocyclic oligomer, or a mixture of homologous macrocylic oligomers, is allowed to undergo reversible cleavage, for example at high temperature in the presence of a catalyst, then ring-opening polymerization will occur.Such polymerizations involve only a shuffling of the linkages between repeat units and have several potentially valuable features.For example, no volatiles or other by-products are generated and, as the macrocycles are generally large enough to be strainless, little or no heat is evolved.The latter process may therefore be described as an entropy-driven ring-opening polymerization (ED-ROP). 22Conversely, when a dilute solution of high molar mass polymer is allowed to equilibrate by reversible chain-cleavage, then macrocyclic oligomers are formed in high yield by ring-closing depolymerization (RCDP). 7,17,18It has been proposed that a combination of RCDP and ED-ROP could form the basis of a technique for recycling high-value step-growth polymers, and such processes have been investigated for several different polymers of this type, 18,[23][24][25] including the poly(ether-sulfones). 24 The vast majority of industrially-significant aromatic thermoplastics are based on 1,4-linked phenylene units, 1,2 although 1,3-linkages are found in a number of cases, as for example in the high-temperature polyamide poly(1,3-phenylene isophthalamide), 27 trademarked as Nomex®.Relatively little work has been reported on analogous systems based on the 1,2-phenylene unit, although a benzene-1,2-dioxy-based analogue of the all-para engineering poly(ether-ketone) known as PEEK has been obtained by Hodge and coworkers from polycondensation of benzene-1,2-diol (catechol) with 4,4'-difluorobenzophenone, and an equilibrium between this polymer and its homologous family of macrocylic oligomers (Chart 2; a) was shown to be established at high temperature in the presence of fluoride ions. 28More recently, o-PEEK macrocycles have been employed to fabricate carbon fibre composites via ring opening polymerization reactions. 29The present study might also lead the way to novel in-situ fabrication of high performance thermoplastic composites.
Other investigations on polymers containing catechol as co-monomer include work on poly(esterimide)s, PEIs, synthesized by polycondensation of N-(4-carboxyphenyl) trimellitimide or N-(3carboxyphenyl)trimellitimide with catechols or 5-methylresorcinol (Chart 2; b). 30It is also noteworthy that catechol residues are present in the highly complex aromatic backbone of the biopolymer lignin, which is attracting increasing interest as a component of bio-thermoplastics as potential alternatives to petroleumderived materials. 31art 2. (a) Catechol-based macrocyclic aromatic ether-ketones 28,29 ; (b) Catechol-based aromatic poly(esterimide)s (R = H, Me, or t-Bu). 30 the present work, we report the synthesis, characterization, and entropy-driven ring-opening polymerization of a novel series of macrocyclic aromatic (ether-sulfone)s derived from catechol by nucleophilic cyclo-condensation with extended aromatic dichloro-compounds, activated towards nucleophilic (SNAr) substitution by the presence of sulfone groups para to the chloro-substituents.
Fractionation of the resulting mixture of cyclic oligomers by gradient elution chromatography afforded a series of macrocycles as pure compounds including the cyclo-dimer 2 (7.8%), -trimer 3 (3.4%),-tetramer 4 (2.0%) and -pentamer 5 (1.5%).The thermal characteristics and solubility properties of the four isolated macrocycles are summarized in Table 1.The cyclic oligomers with an even number of repeat units (dimer and tetramer) are crystalline, high-melting, and soluble only in strongly acidic solvent mixtures (e.g.chloroform/trifluoroacetic acid), while the cyclic trimer and pentamer are essentially amorphous compounds with good solubility in simple chlorinated solvents.Very similar odd/even behaviour has previously been reported for the corresponding cyclic oligomers of the commercial polysulfone known as Radel  (Chart 1). 7The 1 H NMR spectra of macrocycles 2, 3, 4 and 5 are very similar, with only small chemical shift differences as a function of ring size as reported in the Supplementary Material (SM).The 1 H NMR spectrum of cyclic trimer 3 is shown in Figure 1.The symmetrical pair of multiplets at 7.25 and 7.38 ppm is characteristic of the catechol residue, and the highest field doublet, at 6.87 ppm, is assigned to the protons ortho to the ether linkage and meta to sulfone.Detailed assignments of all 1 H NMR spectra are given in the Experimental Section and the SM.The intermediate 4-(4'-chlorobenzenesulphonyl)biphenyl, 6, was obtained in 50% yield from a Friedel-Crafts reaction between biphenyl and 4-chlorobenzenesulphonyl chloride using anhydrous ferric chloride as catalyst. 4Subsequent reaction of 6 with isophthaloyl chloride afforded the seven-ring monomer 7 in 42% yield after recrystallization from DMF (Scheme 2).Its DSC thermogram showed a sharp melting peak at 284 C.The 1 H NMR spectrum observed was fairly complex (see ESI) but was fully consistent with the proposed structure, as indeed were the MALDI-TOF mass spectrum, COSY NMR analysis and 13 C NMR spectrum.
Condensation of monomer 7 with catechol under pseudo-high dilution conditions (Scheme 3) afforded a range of oligomeric molecules, and analysis of the crude product by MALDI-TOF MS demonstrated that these oligomers comprised exclusively macrocyclic species -specifically the cyclic monomer 8, cyclic dimer 9, and trace amounts of a cyclic trimer which was not subsequently isolated.The cyclic monomer (27%) and cyclic dimer (2%) were recovered as pure compounds by gradient elution chromatography.Their structures were confirmed by MALDI-TOF MS, and by 1 H and 13 C NMR spectroscopy (see Experimental Section and SM).In the 1 H NMR spectrum of cyclic monomer 8, the resonance assigned to the isophthaloyl proton Hc, ortho to both carbonyl groups, (identified by 1 H-1 H COSY analysis; see SM) lies at significantly higher field than the resonance for the corresponding protons in cyclic dimer 9, the upfield shift being ca.0.3 ppm (Figure 2).This difference is however readily explained in terms of the allowed conformations at the isophthaloyl group.Energy-minimization of computational models for 8 and 9 (molecular mechanics with charge-equilibration, Cerius2) showed that the diarylisophthaloyl unit in cyclic monomer 8 is constrained to adopt a syn conformation in which the proton of interest (Hc) is "sandwiched" between two adjacent aromatic rings and is thus subject to significant ring-current shielding (Figure 3).
In contrast, the minimum-energy conformation of cyclic dimer 9 is much more open, with the diarylisophthaloyl units adopting an anti conformation in which there can be no intramolecular ring-current shielding of proton Hc.Moreover, in this model for cyclic dimer 9, it is the isophthaloyl protons Hb, meta to Hc, that now lie in the ring-current shielding zones of the adjacent aromatic rings (Figure 3), accounting for their resonance position upfield (by ca.0.2 ppm) relative to the corresponding signal in cyclic monomer 8 (Figure 2).Finally in comparing the 1 H NMR spectra of macrocycles 8 and 9, the difference in chemical shift of ca.0.2 ppm, seen for the protons Hi (Figure 2) may be accounted for in terms of mutual transannular deshielding effects between the adjacent ether-sulfone rings in cyclic monomer 8. Further details of this modelling study, together with atomic coordinates for the final models of macrocycles 8 and 9, are given in the SM.Significant deshielding of the protons Hi is seen in the 1 H NMR spectrum of 8 relative to 9, and this is ascribed to mutual, transannular, ring-current effects between the adjacent "ether-sulfone" rings in cyclic monomer 8.
Entropy-driven ring-opening polymerization of macrocycle 3 was successfully achieved at high temperature (320 °C) in the melt phase, with anhydrous cesium fluoride as catalyst.In this type of process the fluoride ion acts as a nucleophilic initiator, cleaving the activated ether linkage of one macrocycle and generating a free phenoxide end-group which then attacks a second macrocycle, leading to chain-growth polymerization as shown in Scheme 3. The resulting linear polymer 10 was tough and flexible, and had an inherent viscosity of 0.45 dL g -1 , indicative of high MW and comparable with values for several commercial polyethersulfones including Radel  (also 0.45 dL g -1 ; see Chart 1 and ref. 7).The new polymer was fully soluble in 96% sulfuric acid and in mixtures of dichloromethane and trifluoroacetic acid, indicating there was no significant degree of cross-linking between the chains. 6In view of the extremely high melting points of 2, 4, 8 and 9 (> 400 °C), and the very low yield of 5 (1.5%), no comparable polymerizations were possible with these macrocycles.Scheme 3. Nucleophilic ring-opening polymerization of macrocycle 3 leading to polymer 10 (note: the fluorophenyl and phenoxide end-groups of 10 are not shown).

Conclusions
This work reports two new families of aromatic ether-sulfone macrocycles.Cyclocondensation under pseudohigh dilution conditions between 4,4'-bis(4"-chlorobenzenesulfonyl)biphenyl and catechol, with subsequent chromatographic separation of the reaction products, led to the isolation of four novel ether-sulfone macrocycles (cyclic dimer, -trimer, -tetramer and -pentamer) that have all been fully characterized.Similarly, reaction between catechol and a novel seven-ring diketone/disulfone monomer allowed the isolation of two new aromatic ether-ketone-sulfone macrocycles, a cyclic monomer and a cyclic dimer, with the cyclic monomer being obtained in unexpectedly high yield (27%).Interestingly, proton NMR spectrum of the cyclic monomer showed a significant high field resonance-shift of the isophthaloyl proton ortho to both carbonyl groups relative to the corresponding resonance in the cyclic dimer.Potential reasons for this difference were explored by computational modelling, which revealed a constrained syn conformation of the diarylisophthaloyl unit in the cyclic monomer, contrasting with a more open conformation in the cyclic dimer where the diarylisophthaloyl units adopts an anti conformation.As a result, in the cyclic dimer there is no intramolecular ring-current shielding of the corresponding isophthaloyl proton.Finally, given the industrial relevance of poly(ether-sulfone)s, entropy-driven ring-opening polymerization of one of the new macrocycles was studied using a catalytic amount of cesium fluoride as initiator.The resulting poly(ether-sulfone) showed good solubility, indicating an absence of cross-linking, and an inherent viscosity comparable to values for commercially available poly(ether-sulfone)s.

Experimental Section
General.Starting materials were obtained from Sigma-Aldrich.Potassium carbonate was used after drying under vacuum.Anhydrous aluminum chloride was sublimed under a nitrogen atmosphere before use.N,N-Dimethylacetamide (DMAc) was distilled from calcium hydride, and catechol was recrystallized from toluene.All other materials were used as received.NMR data were obtained on Bruker AC250 and JEOL EX400 spectrometers, with chemical shifts recorded in δ (ppm) and referenced to residual solvent resonances.The abbreviations s, d, dd, t, m and br represent singlet, doublet, doublet of doublets, triplet, multiplet and broad respectively.Labelled structures for 1 H NMR assignments are given in the SM.Infra-red spectra were obtained from mulls in mineral oil (Nujol) and were recorded on a Perkin-Elmer FT1700 instrument.Electron ionization mass spectra were run on a VG-BioQ triple quadrupole instrument in positive ion mode.MALDI-TOF mass spectra were obtained on an SAI LT3 Lasertof instrument using 1,8,9-trihydroxyanthracene as matrix.A typical sample preparation was as follows: 0.1 mL of a solution of the compound in THF (1 mg/mL), 0.1 mL of a solution of sodium trifluoroacetate in THF (1 mg/ml) and 0.1 mL of a solution of the matrix in THF (20 mg/ml) were combined, and an aliquot of the mixture was then carefully transferred to a sample plate and left to dry in a vacuum oven (40 C) for 30 minutes prior to the analysis.Differential scanning calorimetry (DSC) was performed using a Mettler DSC20 system (nitrogen atmosphere, scan rate 10 C/min).Solution viscosimetry was carried out using a Schott-Gerate CT 150 semi-automated viscometer.The resulting time measurements are applied to the following equation: Where inh is the inherent viscosity, (ts) is the flow time for the polymer solution at concentration c (g/100 mL), and t0 is the flow time for the pure solvent.
Synthesis of macrocycles 8 and 9. Potassium carbonate (0.76 g, 5.5 mmol), DMAc (150 mL) and toluene (70 mL) were placed in a 500 cm 3 three-necked flask equipped with a magnetic stirrer, nitrogen gas inlet and a Dean-Stark trap with condenser and gas outlet.The reaction mixture was then heated until the toluene began to reflux (ca.160 C).Water was removed by azeotropic distillation with toluene over 2 hours.A solution of compound 7 (2.00 g, 2.54 mmol) and catechol (0.28 g, 2.54 mmol) in DMF (200 mL) was added with a syringe pump at a rate of 10 cm 3 per hour.After the end of the addition, the mixture was refluxed overnight.The resulting brown solution was filtered while hot to remove insoluble salts (K2CO3, KHCO3, KCl).Distilled water (300 ml) containing HCl (10ml) was added slowly to the filtered solution until the desired cyclic oligomers precipitated as a brown solid.The mixture of oligomers was collected by vacuum filtration, washed with methanol, filtered again and dried in a vacuum overnight at 70C to give a brown solid.Gradient elution chromatography with DCM/EtOAc (100/0 to 97/3 v/v) on silica gel allowed only two of the macrocyclic oligomers to be isolated as pure compounds, specifically the cyclic monomer 8 (  Ring-opening polymerization of macrocycle 3. Polymerization of 3 was carried out in an aluminium DSC pan.The macrocycle was ground with 3 mol% of anhydrous cesium fluoride and a pelletized 20 mg sample of the mixture was heated at 320 °C for 1 hour under nitrogen in the DSC instrument and then cooled to room temperature.A subsequent DSC scan to 400 °C showed the resulting polymer to be amorphous (no melting transition) and to have a glass transition temperature (Tg) of 207 °C.On cooling, the polymer was recovered in the form of a tough, yellow, transparent pellet that dissolved completely in 96% sulfuric acid, in which solvent it showed an inherent viscosity (inh ) of 0.46 dL g -1 .

Figure 3 .
Figure 3. Energy-minimized models for the cyclic monomer 8 and cylic dimer 9.The isophthaloyl proton Hc in 8 is constrained to lie within the ring-current shielding zones of the two adjacent aromatic rings by the syn conformation of the diarylisophthaloyl unit.Conversely, in 9, the diarylisophthaloyl units can adopt the more open anti conformation, with no ring-current shielding of Hc but now with shielding of the protons Hb.Significant deshielding of the protons Hi is seen in the 1 H NMR spectrum of 8 relative to 9, and this is ascribed to mutual, transannular, ring-current effects between the adjacent "ether-sulfone" rings in cyclic monomer 8.