Post-polymerization modification reactions of poly(glycidyl methacrylate)s

Post-polymerisationmodification of poly(glycidylmethacrylate) (PGMA) through the nucleophilic ring opening reactions of the pendent epoxide groups allows for the installation of a variety of functionalities onto the reactive scaffold. The primary modification processes involve amine-epoxy, thiol–epoxy, azide-epoxy, acid-epoxy, and hydrolysis reactions. In all cases, sequential post-synthesis modification reactions can also be carried out if multiply-functionalised polymers are required. This, in particular, includes reactions of the hydroxyl group(s) that come into being through the initial oxirane ring-opening reaction. The overall flexibility of these functionalisations, coupled with the commercial availability of glycidyl methacrylate monomer, its controlled polymerisation behaviour through free radical polymerisation methods and high shelf life of the resulting polymers makes PGMA one of the most adaptable reactive scaffolds in polymer chemistry. In this review article, our aim is to discuss the fundamental aspects of the epoxy ring-opening reactions and highlight the utilitarian nature of PGMA by addressing the range of chemistry that has been used to transform this simple structure into a plethora of customised functional polymers.


Introduction
Poly(glycidyl methacrylate) (PGMA) is an interesting polymer. At rst glance, its highly strained three-membered epoxide side chains would suggest instability resulting in a short shelf-life.
One could also imagine that such a reactive polymer might be difficult to access synthetically especially as a homopolymer. However, reality is far from such assumptions. In fact, PGMA can be easily synthesised from its commercially available and inexpensive monomer, glycidyl methacrylate, by free radical polymerization processes (Scheme 1). [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16] Stability-wise, the puried homopolymer can be stored at room temperature and under ambient conditions for years without compromising its structural integrity. In this way, it differs from other reactive scaffolds like alkene and alkynecarrying which can crosslink during polymerisation if the Here, he is leading a research group with interests ranging from synthesis of multifunctional polymers to fabrication of stimuli-responsive nanostructured so materials.
concentration of the active-site-carrying monomer is high and the active site is unprotected.
PGMA, therefore, is a unique structure that can be accessed with a considerable degree of freedom over the synthesis method, puried, and stored in its homopolymer structure. Of course, co-polymerization is possible with any other acrylate or methacrylate monomer to further diversify the range of structures and applications of the nal materials. [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16] When required, the epoxy side chains of PGMA can be subjected to a nucleophilic ring-opening reaction under a variety of conditions to afford functionalised polymers (Scheme 2). The wide range of chemistry available allows one to freely choose the nature of the functional group and the manner of the ring-opening reaction. It is this versatility that establishes PGMA as a general and powerful reactive scaffold in polymer chemistry.

General considerations
How stable is PGMA?
Although epoxides are inherently strained three-membered rings, an acid catalyst or a strong nucleophiletypically created through use of a base catalyst, is essential for the ringopening reaction. In the absence of either, epoxy derivatives remain perfectly stable under ambient conditions. It is for this reason that polymers of epoxides such as PGMA can be stored for years without any detection of crosslinking or degradation.
What will be the regio-chemistry of functionalisation?
Let us rst consider the case of acid-catalysed ring-opening reaction of epoxides (Scheme 3). Here, the rst step is the protonation of the oxygen atom which is stabilised by inductive effect from the most substituted (i.e. most electron rich) carbon atom. This renders the tertiary carbon atom electrophilic and susceptible to a nucleophilic attack by a mild nucleophile such as water or methanol that would otherwise not be able to force the ring-opening reaction if acting alone.
In the absence of an acid activating the molecule, the ringopening reaction can only proceed if a strong nucleophile is available to eject the oxygen atom so that the oxygen atom may form a favourable leaving group (Scheme 4). In this case, since a positive charge is absent, steric rather than electronic factors determine the outcome of the reaction and the second possible regio-isomer forms exclusively in which the nucleophile is attached to the least substituted and least hindered carbon atom. A model-compound-based systematic study has recently conrmed this notion. 42 Can alkoxide anion not initiate oxirane ring-opening polymerization?
Under acidic conditions, the oxygen atom is already protonated before ring-opening and poses no threat for further reaction. Under basic conditions, an alkoxide anion is formed (Scheme 5). Polymer chemists would envisage that such a structure would quickly bring about anionic ring-opening polymerisation of an epoxide monomer. However, no such polymerisation occurs because the alkoxide anion (pK a ¼ 17) is Here, her group is investigating utility of the corannulene motif for the preparation of functional so materials. a very strong base and is rapidly quenched by a proton transfer reaction. The proton can be abstracted from multiple sources within a reaction mixture. For instance, weakly acidic protons associated with the nucleophile, oen employed aqueous/ alcohol medium, or intentionally added proton sources such as ammonium chloride to the reaction, all possess pK a values below 17 (see Scheme 5). It is a simple acid-base reaction that causes the quenching of the alkoxide ion by the mildly acidic protons present in the system, therefore impeding a ringopening polymerisation.

Post-polymerisation modifications
Post-synthesis modication of reactive polymeric scaffolds is an important route to functional polymers. [43][44][45][46][47][48][49][50][51][52][53] An advantage of this route is that a general scaffold can lead to various structures that differ in the nature of the attached functionality but the molecular weight attributes remain the same within a functionalised polymer family. Such a system allows for an unambiguous development of structure-property relationships. PGMA is such a general reactive scaffold which can be turned into a functionalised polymer family through a range of chemical reactions. The following is a discussion of its chemistry arranged based on the nature of the functionalisation reaction.

Functionalisation with amines
In 2010, Leroux and co-workers described the synthesis of linear and branched GMA polymers and their post-polymerization modications with butylmethylamine, propylamine, methylpropylamine, and trimethylethylenediamine (Scheme 6). 54 The ring-opening reactions were carried out at 90 C, overnight in an argon atmosphere. Ratios of 2 : 2, 1 : 1, 1 : 0.5 of the amine to the epoxy group were used in separate experiments. Interestingly, water was added to the reaction mixture and a second reux was carried out at the same temperature for the same period of time. Elemental analysis indicated that sterically demanding trimethylethylenediamine led to a lower degree of functionalisation (72-81%) whereas smaller amines exhibited better efficiencies (88-98%). Most of the functionalised Scheme 2 Post-polymerisation modification reactions of polyglycidyl methacrylate. Sequential multiple functionalisations are possible in most cases as the hydroxyl group(s) formed through the ring-opening reaction can be utilised to install a second functionality. Alternatively, the first functionality can create a new reactive site for subsequent modifications.
Scheme 3 Acid-catalysed oxirane ring-opening reaction. The positive charge builds on the most substituted carbon atom that is most suitable for stabilising it through inductive effect based on the substituents. structures were, however, found to be insoluble in water. It is likely that the post-polymerization reaction did not stop at the formation of secondary/tertiary amines but proceeded to the higher levels of tertiary/quaternary structures especially under the employed reaction conditions. Such intermolecular crosslinks can explain the reduced solubility of the functionalised polymers. Earlier, Huck's group had described such crosslinking processes in polymer brushes grown onto silicon substrates. 55 Furthermore, Li and co-workers, while working with oligoethylene amines at 70 C noticed the same issue of intermolecular crosslinking and decreased functionalisation efficiency with increasing bulk of the amine group. 56 The issue of intermolecular crosslinking could be solved by reducing the reaction temperature (55-37 C) and/or by using a large (at least 10 fold) excess of the amine. [57][58][59][60][61] Remarkably, Tang and coworkers have shown that the functionalisation reaction can be carried out at room temperature for 24 h with near quantitative efficiencies. 62 Such mild conditions are certainly useful to avoid unwanted further reaction of the secondary/tertiary amines and crosslinking.
The amine functionalisation strategy has been heavily used in the preparation of amphiphilic copolymers, star polymers, and gra copolymers that have found applications mainly in the arena of polyplexes formation and gene delivery. [63][64][65][66][67] In some cases, the tertiary amines were further quaternised through the use of an alkyl halide to give access to side-chain cationic polymers. 68,69 In most cases so far, atom transfer radical polymerization (ATRP) was used for the preparation of the PGMA scaffold. Haddleton and co-workers showed that synthesis by a cobaltcatalysed chain transfer polymerisation could lead to the polymer chains displaying a methacrylate chain-end. In such cases, therefore, difference in reactivity of the chain-end (methacrylate group) as compared to the side-chain (oxirane group) can allow for sequential modications and preparation of doublyfunctionalised structures (Scheme 7). 70 This was demonstrated in two different fashions. In case of primary amines being the nucleophiles designed to open the side chains, the acrylate group was consumed rst through a thiol nucleophile. This is because primary amines can add to the C]C bond via a Michael-type of addition reaction. 71 In case of secondary amines, the post-polymerisation modication of the side chain epoxides could be achieved rst and the methacrylate was modied in a subsequent step. Both reaction sequences led to the formation of dual-functional polymers.

Scheme 5
The alkoxide anion is a strong base and is quenched by weak acidic protons present in the system. Therefore, it cannot act as an initiator for ring-opening polymerisation of the pendent epoxide groups. The bottom shows pK a of various functional groups, solvents, or additives that are typically present in the PGMA functionalisation reactions. Nu stands for nucleophile. published in 2012 (Scheme 8). 81 It utilised conventional free radical polymerisation process for the synthesis of the reactive scaffold. Therefore, molecular weights were very high and the degree of polymerization was estimated to be about 12 000 based on the number average molecular weight (P n ). The study showed that even at room temperature, such high molecular weight polymers could be completely functionalised with a number of thiol nucleophiles. Furthermore, the generated secondary hydroxyl groups could be modied through an esterication reaction with a variety of acid chlorides. In a follow up study, copolymerization of GMA monomer with polyethylene glycol-based macro-initiator or a macromonomer through ATRP was shown to give water soluble reactive scaffolds that can be transformed into bifunctionalised structures (Scheme 9). 82 A later study investigated the optimised reaction conditions and quantied the results of the bifunctionalisation strategy using thiol-epoxy and esterication reactions (Scheme 10). 42 This study suggested that the choice and amount of a catalyst heavily inuenced the outcome of the thiol-epoxy reaction. Tetrabutylammonium uoride (TBAF) in tetrahydrofuran (THF), and lithium hydroxide in aqueous THF (10% water) were found to be good catalyst choices for reactions involving aliphatic and aromatic thiols. In the case of TBAF, >20 mol% catalyst loading was required to achieve complete conversion of the epoxy groups in a reaction time of 3 h. Lithium hydroxide proved to be a better catalyst as full conversions could be obtained at a catalyst loading of 1-4 mol% and in a reaction time of 1-3 hours. Triethylamine (TEA), when used in THF, failed to provide any ring-opening reaction. However, it could be used in the case of aromatic thiols. Nonetheless, high catalyst loading (>34%) and long reaction times (12 h) were required for quantitative conversions at room temperature. Interestingly, a change of reaction medium to dimethylsulfoxide (DMSO) could lead to the ring-opening reaction while using TEA as a base and an aliphatic thiol as a reactant. Similarly in this case, longer reaction times, high catalyst loading, higher temperature, and higher thiol content were required to observe >85% epoxy group conversion.
Scheme 8 First application of the thiol-based ring-opening reaction for modification of the PGMA scaffold. In this study, bi-functional structures were also synthesised through esterification of the generated secondary hydroxyl groups. The top shows the general synthetic scheme and the bottom shows various singly and doubly functionalised structures.

Scheme 9
Application of the thiol-epoxy and esterification reactionsbased sequential modifications strategy to prepare poly(ethylene glycol)-based doubly modified water soluble copolymers. This polymer-polymer coupling reaction resulted in the formation of water-soluble bottlebrush copolymers with molecular weights ranging from 50-426 kDa. When low molecular weight PEG polymers with relatively low steric demand were used as the side-chain precursor, high graing densities were observed irrespective of the length of the PGMA backbone (96-97%). The graing density, however, decreased with an increase in the length and steric demand of the PEG side-chain (88-95%). In all cases discussed above, a second modication through an esterication reaction was shown to give doubly-functionalised structures. [84][85][86][87] Finally, this strategy was applied in the preparation of an amphipathic homopolymer library and shown to have siRNA delivery capacity (Fig. 1). 88 Benagali and co-workers showed that the secondary hydroxyl group could also be modied through urethane bond formation with small molecular isocyanates. 89 Stepping away from linear PGMA structures, Gao and coworker showed that segmented hyperbranched polymers with high density of functional groups could be prepared by coupling with a number of thiol molecules. In some cases, the thiol molecule carried another functionality such as an olen bond. Therefore, a second functionalisation reaction using the thiolene 'click' reaction could be carried out to install a different functionality onto the hyperbranched scaffold in a sequential manner. 90 In another impressive example of creating complex and sophisticated segmented yet linear structures, Haddleton and co-workers utilized the orthogonal reactivity of the epoxy group Scheme 11 Preparation of bivalent molecular bottlebrushes through the use of a sequential thiol-epoxy and esterification reactions.
Scheme 10 Examination of the optimum reaction conditions and quantification of the PGMA bifunctionalisation strategy was done using the shown synthetic scheme. Fig. 1 Amphipathic homopolymer family synthesized for siRNA delivery applications. and the alkyne functionality to gain access to multiply-reactive and multiply-functional sequence regulated polymers (Scheme 12). 91

Functionalization with azide
Matyjaszewski and co-workers have established this concept through use of copolymers carrying GMA and methyl methacrylate monomers (Scheme 13). 92 The epoxide groups were opened with sodium azide at 50 C and in the presence of ammonium chloride. As mentioned earlier, ammonium chloride is a proton source to quench the alkoxide anion and stop an anionic ring-opening polymerisation from proceeding. NMR and IR spectroscopies were employed for the characterisation of the produced materials. It was shown that the ring-opening reaction was quantitative. The prepared polymer was further subjected to a copper-catalysed click reaction with polyethylene glycol (2000 g per mol) terminated with an acetylene functionality to nally afford gra copolymers.
In an alternative fashion, Zhao and co-workers showed that a small molecular acetylene appended with a RAFT agent can be attached to the azide group of the polymer and a subsequent RAFT polymerisation of NIPAM monomer could then lead to the gra copolymer structures (Scheme 14). 93 The formed secondary hydroxyl group could be subjected to an esterication reaction with a pyrene-based acid molecule. In aqueous solution, these polymers assembled into vesicular structures in which the hydrophilic NIPAM chains protected the hydrophobic pyrene groups. Replacing pyrene with a polyethylene glycol chain and carrying out the RAFT polymerization with styrene and NIPAM monomers in a sequential manner afforded even more complex amphiphilic structures also capable of assembling into vesicles in water. 94 This theme of preparing molecular bottlebrushes through the azidation strategy of PGMA is further seen in the works of Chen and co-workers who prepared a large family of gra copolymers based upon the ethylene-oxide, ester, acrylate, methacrylate, acrylamide and styrenic polymers (Fig. 2). 95,96 In a completely different manner, Gao and co-workers employed reversible addition-fragmentation chain transfer to Scheme 12 Multi-block orthogonal reactivity concept established by Haddleton and co-workers for the preparation of sequence-regulated multifunctional polymers.
Scheme 13 Epoxy-azide reaction for the functionalisation of the PGMA scaffold as established by Matyazweski and co-workers.

Scheme 14
Grafting-from approach through use of RAFT polymerization for the preparation of bivalent graft copolymers by Zhao and co-workers. synthesise PGMA-based highly branched architectures. The authors then utilised the azide-epoxy reaction to generate polymeric azides that could be modied in a subsequent copper-catalysed 'click' chemistry step with small molecular alkynes to access a myriad of functionalised polymers. In some cases, the hydroxyl group was also modied either prior or aer the azide-alkyne reaction. 97 Finally, glycopolymers could be prepared through azidealkyne coupling reaction between the polymer and freeacetylene carrying sugar molecules for use in synthesis of bioimaging probes (Fig. 3). 98

Functionalisation with water
Hydrolysis of PGMA can be achieved using acid or base catalysis. Some of the early examples however rely only on thermal activation at temperatures of 120-140 C for long periods of time (24 hours) (Scheme 15). 99,100 The high temperatures can be avoided by using sulfuric acid as a catalyst that allows for the hydrolysis to occur smoothly at ambient conditions. 101,103 In terms of regio-chemistry, irrespective of the method, the nal hydrolysis product is the same diol structure that can further be functionalized. For this, esterication of the hydroxyl groups with fatty acid is shown to allow the polymer chain to form reverse micelles that could trap and release peptides and proteins.
Gao and co-workers developed a different strategy for utilizing the diol functionalities. Exploiting higher reactivity of the primary hydroxyl group, they attached acid groups through ring-opening reaction with succinic anhydride (Scheme 16). 102 The secondary hydroxyl groups were then used to form intermolecular polyurethane crosslinks to furnish polymer nanoparticles that could trap and release the anticancer drug doxorubicin. 103 Functionalization with carboxylic acids Minko and Luzinov described the use of PGMA as a means to prepare uorescent core-shell type of nanoparticles while still having the reactive epoxy groups on the particle surface (Scheme 17). 104 To achieve this, PGMA was rst functionalised with rhodamine-B through an acid-epoxy reaction and later deposited on silica particles. This functionalisation required a relatively high temperature of 80 C. Through elemental analysis, it was conrmed that the extent of reaction was such that one epoxy group out of 311 reacted to yield functionalised polymer chains (0.3% efficiency). This synthetic strategy was used by Liu and co-workers to prepare polymers capable of emissive or self-healing properties. 105,106

Conclusions
Following are the conclusions that can be drawn from the aforementioned discussion.
1. PGMA scaffold can be modied through the use of amines, thiols, azides, carboxylic acids, and water as nucleophiles. In all cases, the epoxide ring-opening reaction generates a second reactive site, a secondary hydroxyl group. This hydroxyl group can also be modied in a second post-polymerisation modication step. In case of hydrolysis, a diol forms and can be modied either selectively at the primary hydroxyl group due to its higher reactivity or at both hydroxyl groups. Alternatively, the initially installed functionality may be able to react further as in the case of amines. In many cases, multiply functionalised structures have been obtained through sequential postpolymerization modication reactions.
2. While using amines for modication purposes, at least a 10-fold excess of the amine reactant must be used and the reaction temperature must be lowered to 35 C or room temperature. These two factors can avoid inter and intramolecular crosslinking reactions from occurring. These crosslinking reactions originate from the desire of the primary amine to become secondary, secondary to tertiary, and tertiary to quaternary amine. This is due to the fact that amine reactivity increases as a result of electron-donating nature of the alkyl groups to the nitrogen atom making each product more nucleophilic than its predecessor.
3. Amine functionalisations require long reaction times (24 hours) for completion at room temperature.
4. Functionalisation with thiols is regio-selective, quantitative, and fast. The functionalisation reactions can be performed at room temperature and requires 1-3 hours for completion even if very high molecular weights (P n ¼ 12 000) of the PGMA scaffold is used. Sterically demanding thiols can also be used. However, the degree of graing decreases with increasing molecular weight (>2 kDa) of the thiol used.
5. Reaction with sodium azide (50 C) requires a mild acid such as ammonium chloride as the proton source for quenching the alkoxide anion. The reaction generates polymeric azides in high efficiency. A subsequent modication of the azide groups through copper-catalysed 'click' reaction then gives access to functionalised polymers. The utility and efficiency of this second post-polymerisation modication is shown through preparation of a number of gra copolymers with high graing densities.
6. Carboxylic acids represent the least active and least efficient system towards functionalisation of the PGMA scaffold even when relatively high temperatures (80 C) and relatively long reaction times (12 h) are employed.
In general, the PGMA scaffold can be transformed into various functionalised structures through the opening of sidechain epoxide rings with a number of nucleophiles. This particular synthetic strategy to functional polymers bears the hallmark of simplicity and adaptability. This is evident from a commercial access to the monomer, polymerisability with robust and tolerant free radical polymerisation protocols and functionalizations with a number of nucleophiles in the presence of air and moisture and sometimes, under ambient conditions. This strategy has been put to use by many researchers and the nal materials have oen found biorelevant applications. In most cases, a single nucleophile is used for the rst functionalisation of the PGMA scaffold.
If we examine the modication reactions in terms of the reaction speed, efficiency of functionalisation and reaction conditions, it becomes clear that distinct differences in reactivity exists and relate to the nature of the nucleophile. This presents an opportunity to modify one scaffold in a one-pot or sequential manner with different nucleophiles to generate multiply functionalised polymers. Overall, it seems that the single-nucleophile modications will still dominate the research scene but new materials based on the (orthogonal) reactivity concepts are likely to add new dimensions to this historic chemistry.

Conflicts of interest
There are no conicts to declare.