The homopolymerization and block/statistical copolymerization of 2-hydroxyethyl methacrylate (HEMA) using atom transfer radical polymerization (ATRP) in methanol at 20 °C has been investigated. For the homopolymerizations, both high conversions and low polydispersities (M w/M n < 1.25) were obtained over a wide range of target degrees of polymerization. According to the literature, HEMA homopolymer is usually described as only water-swellable, but in this work low molecular weight HEMA oligomers (target degrees of polymerization, DPn, less than 20) exhibited water solubility over a wide temperature range (no cloud point behavior). Furthermore, for actual DPn's between 20 and 45, HEMA homopolymers exhibited inverse temperature solubility in dilute aqueous solution at pH 6.5, and their cloud points increased systematically as the DPn was reduced. Gravimetric studies indicated that “water-insoluble” HEMA homopolymers with DPn's higher than 50 were actually partially soluble:  GPC studies confirmed that fractionation occurred due to preferential dissolution of the shorter chains. Furthermore, HEMA homopolymers with DPn's up to 50 are water-soluble at pH 2.2 and do not exhibit cloud points. This is attributed to protonation of the terminal morpholine groups derived from the ATRP initiator. Thus, depending on the mean DPn and the solution pH, water can be a good solvent, a marginal solvent or a nonsolvent for HEMA homopolymer. Chain extension (self-blocking) experiments conducted for the ATRP of HEMA in methanol at 20 °C using a Cu(I)Cl catalyst and bpy ligand indicated reasonable living character. Statistical copolymerizations of HEMA with other comonomers such as glycerol monomethacrylate (GMA) and 2-hydroxypropyl methacrylate (HPMA) allowed the cloud point behavior to be manipulated. Finally, a range of novel HEMA-based block copolymers were synthesized in which the HEMA block was either thermoresponsive or permanently hydrophilic, depending on its DPn and the nature of the second block. Thus, diblock copolymer micelles with either hydroxylated cores or coronas could be prepared.

The branching copolymerization of 2-hydroxypropyl methacrylate (HPMA) with either ethylene glycol dimethacrylate (EGDMA) or bisphenol A dimethacrylate (BPDMA) as the branching agent has been carried out using atom transfer radical polymerization (ATRP) in methanol at 20 °C. With EGDMA, soluble branched copolymers were obtained provided that less than one branching agent was incorporated per primary chain:  higher levels of EGDMA led to gelation, as expected. Analysis of the changes in the molecular weight and polydispersity of the polymers shows that the formation of highly branched chains occurs only at high (>90%) conversions. Chain coupling is close to the ideal behavior predicted by the Flory−Stockmayer theory, suggesting that all double bonds are equally reactive and that there is no significant cyclization, in contrast to conventional free radical polymerization. This analysis is confirmed by comparison of the consumption of the EGDMA branching agent with predictions from both theory and simulation. With BPDMA as the branching agent, similar results are obtained although branching is slightly less efficient.

Branching copolymerisation of a monovinyl monomer with a small amount of a divinyl monomer under pseudo-living conditions has been modelled using a Monte Carlo program written in C# code. Molecular weights, polydispersities and the distribution of primary linear chains amongst the branched copolymer molecules can be modelled as a function of conversion. The effect of varying reactivities can be modelled by assigning different probabilities for the reaction of the double bonds of the monovinyl monomer, the unreacted divinyl monomer and the half-reacted divinyl monomer. The predictions of this Monte Carlo model have been compared with experimental data previously obtained for the ATRP of 2-hydroxypropyl methacrylate with ethylene glycol dimethacrylate (I. Bannister et al., Macromolecules, 2006, 39, 7483-7492). The model predicts higher molecular weights and polydispersities than are actually observed experimentally, probably because branching becomes diffusion controlled in real reactions at high conversion and because the model allows couplings between chains which would be spatially constrained in the real system. The model predicts (i) formation of a proportion of residual linear primary chains if the proportion of divinyl monomer is low and (ii) only very low levels of intramolecular (primary) cyclisation to form loops. Comparison of experimental and predicted molecular weight vs. conversion data confirms that the experimentally determined onset of rapid molecular weight rise can be modelled assuming random addition of double bonds of equal reactivity, leading to statistical branching and more homogeneous structures than are obtained by conventional free-radical polymerisation.