Sandeep
Sharma
,
Konstantinos
Ntetsikas
,
Viko
Ladelta
,
Saibal
Bhaumik
and
Nikos
Hadjichristidis
*
Physical Sciences and Engineering Division, KAUST Catalysis Center, Polymer Synthesis Laboratory, King Abdullah University of Science and Technology (KAUST), Thuwal 23955, Saudi Arabia. E-mail: nikolaos.hadjichristidis@kaust.edu.sa
First published on 28th October 2021
The synthesis of cyclic polymers on a large scale is a challenging task for polymer scientists due to the requirement of ultra-high dilution conditions. In this paper, we demonstrate an alternative method to prepare cyclic polymers with moderate dilution and up to 1 gram scale. We employed a simple Williamson etherification reaction to prepare cyclic polymers with a good solvent/non-solvent combination. In this way, various polystyrene (PS) and polyethylene glycol (PEG) cyclic homopolymers were synthesized. Anionic polymerization using high vacuum techniques combined with the postpolymerization reaction was used to generate linear dihydroxy PS precursors. The synthesized linear and cyclic homopolymers were fully characterized using various spectroscopic and analytical techniques, such as size exclusion chromatography (SEC), matrix-assisted laser desorption/ionization time-of-flight mass spectroscopy (MALDI-TOF-MS), and differential scanning calorimetry (DSC). Detailed nuclear magnetic resonance (NMR) spectroscopic studies were also performed to obtain the complete structural information of the synthesized polymers.
On the other hand, the ring-closure method comprises the coupling reaction of an α,ω-functionalized linear polymer, generally performed under high dilution conditions on a small scale and using a catalyst.8 This strategy can further be divided into unimolecular and bimolecular coupling reactions. The unimolecular ring-closure technique involves the cyclization of homo-difunctional9,10 or hetero-difunctional linear polymers.11–13 Unimolecular cyclization is not affected by the stoichiometric ratio; however, it requires high dilution to suppress the oligomerization. The bimolecular ring-closure strategy involves the reaction between an α,ω-homodifunctional linear polymer and a difunctional coupling agent in a dilute solution. The exact stoichiometric ratio of all reagents is required to generate pure cyclic polymers.
Initially, Rempp et al. used the bimolecular coupling method to synthesize cyclic polymers.14 They reported the synthesis of cyclic polystyrene (PS) by the coupling reaction of a living difunctional PS, prepared via anionic polymerization, with a difunctional electrophilic compound (dibromo-p-xylene) in a stoichiometric ratio under high dilution. In the same year, Höcker et al. reported the synthesis of cyclic PS by the slow addition of living difunctional PS to α,α′-dichloro-p-xylene solution in tetrahydropyran (THP), but still the yield of cyclization was below 50%.15 Roovers et al.16 reported the synthesis of high molecular weight ring PS (molecular weights in the range of 5000 to 450000 g mol−1) using sodium naphthalene as a difunctional initiator and dichlorodimethylsilane as the linking agent. Similarly, various cyclic homopolymers and block copolymers were synthesized using anionic polymerization and coupling agents with different functionalities.17–21
Apart from anionic polymerization, other polymerization methods have also been used to synthesize cyclic polymers combined with different bimolecular and unimolecular coupling methods, such as the self-accelerating click reaction22 and the electrostatic self-assembly and covalent fixation (ESA-CF) method.23 However, the upgrading of these processes to a larger scale (grams) is still a very challenging task. Some efforts have been made to synthesize cyclic polymers by the continuous-flow technique24 and the self-accelerating double strain-promoted azide–alkyne click reaction (DSPAAC) for cyclization.25 Nonetheless, these reactions require a large amount of solvent to prepare cyclic polymers on a large scale.
The Williamson etherification is a well-known and straightforward reaction in organic chemistry. The synthesis of cyclic polyethers via the Williamson etherification reaction has been reported, where a homodifunctional polyethylene glycol (PEG) bearing hydroxyl end groups was cyclized by reacting with dichloromethane (DCM) in the presence of potassium hydroxide.26 Recently, the scale-up of cyclic PEG polymers was also attempted via the Williamson etherification.27 In the latter, cyclic PEG polymers were generated through the reaction of dihydroxyl PEG with tosyl chloride (Ts-Cl) and purified through multi-step procedures. A combination of a good solvent and non-solvent was used for the synthesis of cyclic PEG to minimize the chain distance. The influence of good solvent/non-solvent mixtures on the preparation of cyclic PS was also studied in radical trap assisted atom transfer radical coupling (RTA-ATRC).28
Cyclic polymers, which are synthesized by click and other coupling reactions, are generally prepared by atom transfer radical polymerization (ATRP)29 and reversible addition–fragmentation chain-transfer (RAFT) polymerization.30 These techniques do not always lead to well-defined polymers with controlled molecular characteristics and a narrow polydispersity index (PDI < 1.1). On the other hand, living anionic polymerization leads to well-defined polymers having the highest molecular weight, structural and compositional homogeneity. Hence, we designed a bimolecular ring-closing reaction between telechelic polymers and coupling agents [1,4-bis(bromomethyl)benzene and/or 2,6-bis(bromomethyl)pyridine] to generate cyclic polymers in a good solvent/non-solvent mixture. Benzylic bromo compounds as linking agents were chosen due to their higher reactivity compared to the aliphatic ones.
Herein, we report a traditional Williamson etherification-based efficient cyclization reaction to prepare cyclic PS and PEG homopolymers up to 1 gram scale in moderate dilution. The linear α,ω-dihydroxy PS precursors (HO-PS-OH) were synthesized via anionic polymerization using high vacuum techniques and the postpolymerization reaction (deprotection of the α-tert-butyldimethylsilyl group). Overall, seven different cyclic polymers (four PS and three PEG) have been prepared with different molecular weights ranging from 4000 to 23000 g mol−1 for the PS and 2000 and 6000 g mol−1 for the PEG samples, respectively.
Scheme 1 General synthetic strategy for the preparation of HO-PS-OH homopolymers via anionic polymerization and the postpolymerization reaction. |
The telechelic polymers were cyclized through a bimolecular ring-closure strategy by reacting an equimolar quantity of 2,6-bis(bromomethyl)pyridine or 1,4-bis(bromomethyl) benzene with a mixture of solvents (THF and n-hexane) (Schemes 2A and B). The purpose of using a non-solvent (n-hexane) was to improve the ring-closure reaction by reducing (on average) the end-to-end distance of the polymer chains. This approach helped to perform the cyclization process in higher concentrations due to the presence of the non-solvent. Thus, a series of cyclic polymers up to 1 g scale were synthesized in 80–85% yield.
In the case of cyclic PEG, commercial linear dihydroxyl l-PEG-2k and l-PEG-6k were used for the bimolecular coupling reaction with 2,6-bis(bromomethyl)pyridine (Scheme 2C) and 1,4-bis(bromomethyl) benzene as linking agents (Scheme 2D).
To confirm the effect of the non-solvent (hexane) on the cyclization of l-PEG-2k with 1,4-bis(bromomethyl) benzene, we performed the reaction in the presence and the absence of a non-solvent and the synthesized polymers were characterized using SEC and 1H-NMR spectroscopy.
The molecular characteristics along with the corresponding glass transition temperatures (Tg) of all synthesized linear and cyclic homopolymers are given in Table 1. As a representative example, SEC chromatographs of the linear and cyclic PS-4k polymers are given in Fig. 1. The SEC chromatograph of the linear PS-4k (l-PS-4k) showed a peak molecular weight of 4800 g mol−1 (Mp), while the cyclic PS-4k (c-PS-4k) showed a lower Mp value (3700 g mol−1), leading to a higher retention time compared to the linear precursor. It is well known that cyclic polymers exhibit a lower hydrodynamic volume than linear ones due to the absence of free chain ends.39 Therefore, the SEC analysis indicated the first evidence for the successful synthesis of the cyclic polymer. The SEC chromatograph of the crude cyclic polymer revealed some additional peaks (mainly the condensation product), which were removed by fractionation in a toluene/methanol (good solvent/non-solvent) system. The SEC chromatographs of the crude and the purified c-PS-4k are shown in Fig. S1.† The ratio of the cyclic and the high molecular weight condensation products (linear) was calculated from each SEC chromatograph (Table S1†). For the c-PS-4k-crude sample, the ratio of the cyclic and linear products is 79:21 (Fig. S1†). The SEC chromatographs of all cyclic crude samples, and the respective linear and purified cyclic polymers are shown in the ESI (Fig. S2–S7†). The PDI of the cyclic polymer and its linear counterpart was found to be nearly the same. The number-average molecular weight (Mn) of all linear PS was determined by SEC since the instrument was calibrated with PS standards, and in all cases, the PDI was found in the range of 1.02 to 1.05.
Sample | M na (g mol−1) | M pa (g mol−1) | PDIa (Đ) | T gb (°C) |
---|---|---|---|---|
a M n, Mp, and Đ were determined by SEC in THF at 35 °C. b T g was measured by DSC. | ||||
l-PS-4k | 4100 | 4800 | 1.04 | 79.5 |
c-PS-4k | 3500 | 3700 | 1.05 | 86.9 |
l-PS-7.5k | 7600 | 7900 | 1.02 | 89.7 |
c-PS-7.5k | 6500 | 6800 | 1.03 | 94.2 |
l-PS-13k | 13200 | 14200 | 1.02 | 92.8 |
c-PS-13k | 10200 | 10900 | 1.03 | 96.6 |
l-PS-23k | 23100 | 25200 | 1.02 | 94.2 |
c-PS-23k | 19100 | 19800 | 1.03 | 103.2 |
The molecular characteristics along with the corresponding melting and crystallization transition temperatures (Tm and Tc) of l-PEG and c-PEG are given in Table 2. The corresponding SEC chromatographs of l-PEG-2k and c-PEG-2k are shown in Fig. 2. The l-PEG-2k polymer showed a peak molecular weight (Mp) of 2400 g mol−1, while c-PEG-2k showed a lower value (1800 g mol−1), leading to a higher retention time than the linear polymer. The ratio of the cyclic and linear products was calculated from the SEC chromatographs of the crude cyclic polymers (Fig. S8 and S9†), given in Table S2.† The SEC chromatographs of l-PEG-2k and the purified c-PEG-2k-b are presented in Fig. S10.† Similarly, c-PEG-2k-b showed a lower value of peak molecular weight (Mp) at 1750 g mol−1, compared to l-PEG-2k. Furthermore, c-PEG-6k also followed a similar trend confirming the successful cyclization (Fig. S11†).
Fig. 2 SEC chromatographs of l-PEG-2k and the corresponding purified c-PEG-2k (THF, 35 °C, PS standards). |
To further investigate the role of the good solvent/non-solvent system, cyclization reactions were carried out in a good solvent, i.e. THF and in a mixture of good and non-solvents (THF/hexane) for the case of l-PEG-2k using 1,4-bis(bromomethyl) benzene as the linking agent. As is evident from the SEC chromatographs (Fig. S12†), a mixture of good and non-solvents is an effective medium to promote cyclization (only 13% of the condensation product); however when only THF was used, the proportion of high molecular condensation products significantly increased (13% to 49%).
The synthesized (linear and cyclic) polymers were further characterized by one-dimensional (1D) and two-dimensional (2D) NMR spectroscopy, and the complete structural information is reported. In all cases, the successful deprotection reaction for PS and subsequently the cyclization reaction were confirmed by 1H-NMR spectroscopy. Out of all the synthesized polymers, the PS-4k case was selected as an example for detailed NMR studies.
The 1H-NMR spectra of Pr-l-PS-4k and l-PS-4k (after the deprotection) are given in Fig. 3A and B, respectively. The quantitative deprotection of the t-butyldimethyl-silyl groups was accomplished using TBAF solution. The progress of the reaction was monitored by 1H-NMR spectroscopy and specifically by the observation of the characteristic peak of the t-butylsilyl moiety at 0.9 ppm (protons d in Fig. 3A). Treatment of the protected PS homopolymer with TBAF led to the disappearance of the chemical shift at 0.9 ppm (Fig. 3B), indicating the complete removal of the t-butyldimethyl-silyl groups and resulting in the formation of an α,ω-dihydroxy-PS homopolymer. Interestingly, the –OCH2 protons, corresponding to both chain ends, presented two different peaks at 3.51 (proton a in Fig. 3B corresponds to the propyl chain end) and 3.34 ppm (proton b in Fig. 3B corresponds to the ethyl chain end). To further confirm this assumption, a monofunctionalized-OH PS of similar molecular weight was synthesized via anionic polymerization (details of the synthetic procedure in the ESI†) for comparison. The corresponding 1H-NMR spectrum was compared with that of the l-PS-4k polymer (Fig. S13†). The peak position and shape confirmed that the 3.34 ppm peak corresponds to the –OCH2 protons for the chain end group and is located near the phenyl ring of PS. Due to the effect of the phenyl ring of PS, these protons were shielded and showed a peak around 3.34 ppm, while the protons of the –OCH2 group of the α-chain end are located far from the phenyl ring and appeared at 3.51 ppm. However, the integration of the area of both peaks was calculated to be almost equal (Fig. S13†).
Fig. 3 1H-NMR spectra of (A) Pr-PS-4k, (B) l-PS-4k, and (C) the corresponding purified c-PS-4k (CDCl3, 25 °C, 500 MHz). |
Additionally, the quantitative 13C-NMR spectrum of the l-PS-4k polymer also indicated two different peaks at 62.86 and 61.33 ppm (Fig. S14†). The peak at 62.86 ppm corresponds to the propyl carbon denoted as b in Fig. S14,† while the peak at 61.33 ppm corresponds to the ethyl carbon (represented as a in Fig. S14†). A 1:1 integration ratio in the quantitative 13C-NMR analysis also verified that these signals belong to the protons of the –OCH2 end groups. The complete structural information of the l-PS-4k polymer was established by 1H–1H correlation spectroscopy (COSY) and total correlation spectroscopy (TOCSY), as presented in Fig. S15† and Fig. 4 respectively.
The COSY spectrum of the l-PS-4k polymer revealed the correlation between neighboring protons separated by one C–C bond and provided useful information regarding the end-group analysis (Fig. S15†). Protons (b) at 3.33 ppm correlate with protons (i) at 1.67 ppm and protons (a) at 3.51 ppm correlate with protons (g) at 1.38 ppm. Furthermore, long-distance 1H–1H correlation analysis by TOCSY (Fig. 3) revealed that protons (a) correlate with the protons (g), (h), and (j) at 1.10, 1.47, and 2.05 ppm, respectively. These protons [(g) and (h)] can be assigned as the –CH2–CH2– protons of the propyl end-group (α-chain end group) and (j) protons from the main chain. On the other hand, protons (b) correlate with protons (i), indicating that they correspond to the ω-chain end group. Thus it was confirmed that the two distinct peaks are assigned to the end groups: the propyl end group (–OCH2) signal indicated at 3.51 ppm and the ethyl end group (–OCH2) signal at 3.34 ppm.
The cyclization reaction of l-PS-4k was carried out using 2,6-bis(bromomethyl)pyridine as a linking agent in the presence of KOH. The 1H-NMR spectrum of the c-PS-4k polymer revealed three new peaks at 4.46, 4.53, and 7.63 ppm (Fig. 3C). The peaks at 4.46 and 4.53 ppm (d and e) correspond to the –OCH2 groups near to the pyridine ring and the signal at 7.63 ppm (protons f in Fig. 3C) to the proton of the pyridine ring. Along with this, the upfield shift of protons at 3.51 (protons a in Fig. 3B) to 3.42 (protons b in Fig. 3B) and 3.34 (protons a in Fig. 3C) to 3.27 (protons b in Fig. 3C) ppm confirmed the successful cyclization (transformation of the –OH group to the –OCH2 group). Unfortunately, the corresponding carbons could not be detected in the quantitative 13C-NMR spectrum (Fig. S16†) after the cyclization reaction. In the cyclic polymer, the –CH2 group near the pyridine ring (protons d and e, at 4.46 and 4.53 ppm) is surrounded by a tertiary carbon and oxygen as heteroatoms. Therefore, these protons did not show any short- and long-distance correlation as confirmed by the TOCSY spectrum of the c-PS-4k polymer in Fig. 5. Other long-distance correlations from propyl and ethyl end-groups remained intact.
Likewise, all the other linear and cyclic PS samples (l-PS-7.5k, l-PS-13k, l-PS-23k, and the corresponding cyclic ones) were also characterized by 1H-NMR spectroscopy, and the cyclization was confirmed in all cases (Fig. S17–S19†). Furthermore, l-PEG and c-PEG samples were also characterized by 1H-NMR spectroscopy (Fig. S20–S22†). The 1H-NMR spectra of l-PEG-2k showed peaks around 3.65 ppm (protons a, Fig. S20†), referring to the ethylene glycol repeating unit (–OCH2CH2–). The protons of the end groups (–OCH2) of l-PEG-2k were merged with the –OCH2CH2– protons of the main chain at 3.65 ppm. The molecular weight of the commercial PEG polymers was confirmed by the end-group analysis technique using trifluoroacetic anhydride (TFA) (Fig. S23 and S24†).40 To distinguish the chain end group, a few drops of TFA were added into the NMR sample tubes of the commercial l-PEG samples. TFA can interact with the hydroxyl end-group (–CH2–OH), resulting in the downfield shift of the adjacent methylene protons from 3.56 ppm to 4.49 ppm. Thus, the obtained molecular weights were in good agreement with the reported ones; i.e., l-PEG-2k showed 2000 g mol−1 and l-PEG-6k showed 6200 g mol−1. The 1H-NMR spectrum of c-PEG-2k was compared with that of l-PEG-2k, and new peaks were observed at 4.67, 7.38, and 7.70 ppm (protons c, e, and d, respectively) in c-PEG-2k (Fig. S20†). Also, in the case where 1,4-bis(bromomethyl)benzene was used as the linking agent for the cyclization, the 1H-NMR spectrum of c-PEG-2k-b (Fig. S21†) showed new peaks at 7.29, 4.53 and 3.93 ppm (protons a, b and c, respectively) compared to the linear precursor.
Similarly, in c-PEG-6k also new peaks were observed, which confirm the successful cyclization of the linear precursor (Fig. S22†).
The synthesized linear and cyclic polymers (l-PS-4k and c-PS-4k) were also analyzed by MALDI-TOF MS. The MALDI-TOF MS spectra of l-PS-4k and c-PS-4k are depicted in Fig. 6. According to the structure of l-PS-4k, the spectrum should show mass distribution: M + Na+ = [104.08 (chain ends) + 104.15n + 23 (Na+)] and the corresponding cyclic: M + Na+ = [207.13 + 104.15n + 23 (Na+)]. The peak distribution of l-PS-4k could be accurately assigned to the linear l-PS-4k ionized with Na+. Similarly, for c-PS-4k, the peak distribution was precisely ascribed to the cyclic structure ionized with Na+. A regular m/z difference of 104.1 was observed between the neighboring peaks in the distribution of both the linear and cyclic polymers, which corresponds to the molar mass of the monomeric styrene unit. The presence of linear and higher molecular weight impurities such as macrocyclic dimers was not observed. Thus, MALDI-TOF MS confirmed the high purity of the final c-PS-4k.
The MALDI-TOF MS spectra for l-PEG-2k and c-PEG-2k are presented in Fig. 7. According to the structure of l-PEG-2k, the spectrum should show mass distribution: M + Na+ = [18.02 (chain ends) + 44.05n + 23 (Na+)] and the corresponding cyclic: M + Na+ = [121.05 + 44.05n + 23 (Na+)]. The peak distribution of l-PEG-2k and c-PEG-2k could be accurately assigned to the corresponding polymer ionized with Na+. The m/z difference of 44.05 was observed between the neighboring peaks in the distribution of both linear and cyclic polymers, which corresponds to the molar mass of the ethylene glycol repeating unit (–OCH2CH2–). The presence of linear and higher molecular weight products like macrocyclic dimers and condensation products was also not observed in this case.
In the DSC traces of the l-PEG and c-PEG samples, melting and crystallization transitions were observed, as presented in Fig. 9 and S28.† The thermal behavior of the cyclic and linear crystalline polymers is still debatable.41,42 In some cases, cyclic polymers show lower melting point (Tm) and crystallization temperature (Tc) values compared to linear analogs.43 The DSC trace of l-PEG-2k revealed transitions at 42.3 °C and 16.3 °C (attributed to Tm and Tc, respectively), while the DSC trace of c-PEG-2k showed transitions at 38.5 °C and 9.6 °C (Fig. 9), respectively. The c-PEG-6k polymer also followed similar behavior regarding Tm and Tc (Table 2 and Fig. S28†).
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1py01337h |
This journal is © The Royal Society of Chemistry 2021 |