(262a) Insights Into Crosslinking Nitroxide-Mediated Radical Copolymerization of Styrene and Divinylbenzene with a Unimolecular Initiator

Crosslinked polymer networks are very important in technology, medicine, biotechnology, agriculture and other areas. They find uses over a wide variety of applications, including construction materials, paints and coatings, polymer glasses with high mechanical strength and thermal stability, rubbers, ion-exchange resins and sorbents, insoluble polymer supported reagents, controlled drug-release matrices, electronics and cables, food packaging, sensors, “smart” materials, artificial organs, implants, superabsorbent materials, etc. Most of these applications of polymer networks require a homogeneous structure (morphology) to obtain optimal performance. However, polymer networks obtained by regular free radical copolymerization are rather heterogeneous in nature. The non-uniformity of network morphology within crosslinked polymers makes consistent production challenging and limits their marketability. Hence, it would be desirable to have a synthetic route to produce homogeneous polymer networks. Copolymerizing vinyl and divinyl monomers in the presence of controlled radical polymerization (CRP) controllers might result in a more homogeneous network, therefore, synthesis, characterization and modelling of polymer networks by CRP processes have received considerable attention in the last few years.

The crosslinked copolymer of styrene (STY) and divinyl benzene (DVB) is used for chromatographic applications and as a precursor for ion-exchange resins. It is also a system relatively well studied under regular radical polymerization conditions, hence a good choice for fundamental comparisons between CRP and regular crosslinking polymerizations. In regular crosslinked copolymerization of STY/ DVB, the rate of polymerization (and the corresponding evolution of conversion versus time) is very fast and high (theoretically, infinite) molecular weight crosslinked network material forms very early in the reaction. The (complex, multi-radical) gel portion grows very quickly like a ‘sponge’, consuming rapidly the adjacent sol molecules, and hence the polymerizing mixture becomes viscous from the very beginning. This greatly affects the nature of the crosslinked polymer network, resulting in the formation of a heterogeneous one.   

Recently, the claim has been made that several types of controlled radical polymerization may lead to the formation of much more (almost, ideally) homogeneous crosslinked networks (Ide and Fukuda, Macromol., vol 32, 95 (1999)). The ultimate target for characterizing a polymer network is its crosslink density distribution, albeit a very difficult distribution to obtain (and even predict via a mathematical model, due to many unknowns in the involved parameters). Hence, several other indicators are used that give some idea about the degree and kinetics of crosslinking. Essentially, researchers use these indirect macro-indicators in order to make claims about the homogeneity of the polymer network, i.e., the fact that the crosslink density distribution is spatially the same within the network.  However, this jump from, essentially, kinetic data and gel average characteristics is based on very speculative arguments. These are at best vague and often based only on indirect, theoretical (and usually not well founded modelling) studies. Perhaps, a more formal, direct, and reliable way (if such a way exists) of characterizing the crosslinked polymer network with respect to its crosslink density distribution could clarify many existing contradictory statements encountered in the literature. In fact, there is not even a widely accepted definition for the concept of homogeneity in a polymer network. These observations motivated our study on polymer networks in controlled radical polymerization, given our prior experience with both regular and CRP-type free radical crosslinking copolymerizations, both experimentally and via mathematical modelling approaches (Vivaldo-Lima et al., Polym. React. Eng. J., vol 2, 97 (1994); Tuinman et al., J. Macromol. Sci., Part A: Pure & Appl. Chem., vol 43, 995 (2006); Hernandez-Ortiz et al., Macromol. React. Eng., vol 3, 288 (2009)).

Our current work examines crosslinking nitroxide-mediated radical polymerization (NMRP) of STY in the presence of a small amount of DVB as the crosslinker, at elevated temperatures (above 100 °C). N-tert-butyl-N-(2-methyl)-1-phenylpropyl)-O-(1-phenylethyl) hydroxylamine (TIPNO) is used as the unimolecular initiator, which acts as both initiator and nitroxide-type controller. The reasons for choosing a unimolecular initiator (instead of a bimolecular system with a peroxide) were simplicity and prior reliable kinetic information on the novel TIPNO (Drache et al., Polymer, vol 48, 1875 (2007)). Results are contrasted with regular free radical copolymerization of STY/DVB and nitroxide-mediated radical homopolymerization of STY in the presence of TIPNO, as reference systems. The effects of TIPNO concentration and DVB level are investigated on polymerization rate (conversion versus time profiles), average molecular weights (number- and weight-average), gel content and swelling index.

Our investigations show that the presence of TIPNO as a mediator in the copolymerization of STY/ DVB slows down the rate of polymerization, delays the onset of gelation, yields lower average molecular weights and polydispersities, and produces a ‘looser’ polymer network compared to the regular free radical copolymerization of the two monomers. The copolymerization reaction exhibits controlled behaviour up to the vicinity of the gelation point (i.e., linear increase of average molecular weights with conversion, and low polydispersity values (well below the typical PDI values in regular free radical polymerization, i.e., much less than 2). However, after the gelation point, and due to the presence of the crosslinker (DVB), the polymerization mixture becomes very viscous, and this leads to the loss of ‘livingness’ of the NMRP process at higher conversion levels.

In our designed experiments (with careful independent replication, in order to build more confidence in the observed data from such complex, noisy and experimentally uncertain systems), two concentration levels were used for TIPNO and DVB, which resulted in four different concentration ([DVB]/ [TIPNO]) ratios. Results show that TIPNO and DVB concentrations do not affect the rate of polymerization (at least at the DVB levels employed; the non-significant effect of TIPNO concentration was as expected). On the other hand, their effect on average molecular weights, gel content and network morphology are noticeable. Using a higher TIPNO concentration, while keeping the same DVB concentration, will delay the gel point. In contrast, using a higher DVB concentration, while keeping the same TIPNO concentration, will accelerate the formation of gel. All in all, our studies reveal that combinations of high DVB concentration and low TIPNO concentration (i.e., maximizing the [DVB]/ [TIPNO] ratio) result in the earliest gelation point and hence, fastest loss of ‘livingness’. As the [DVB]/ [TIPNO] ratio decreased, the gel point was observed at higher conversion levels. These observations were confirmed both indirectly, based on the jump in average molecular weight values versus conversion data, and also through more direct steps measuring the gel content and swelling index using a Soxhlet extraction set up. Furthermore, our Soxhlet results indicated that the lower the [DVB]/ [TIPNO] ratio, the ‘looser’ the polymer network was at higher conversions (for instance, at 85% conversion, highest swelling index and lowest gel content were obtained for the lowest [DVB]/ [TIPNO] ratio).

In parallel to our experimental investigations, a detailed mathematical model for the copolymerization kinetics of STY/ DVB in the presence of different nitrodixe-type controllers has been developed. The predicted profiles for polymerization rate, molecular weight averages and gel content (or swelling index) have been validated with the respective experimental data for NMRP copolymerization of STY/ DVB in the presence of TIPNO. These validations took place independently (between two different research groups in two universities in two countries), without additional parameter fitting (thus indirectly confirming the validity and generality of the developed mathematical model and accompanying database of kinetic and physical/ chemical characteristics and parameters). Not only did model predictions follow the general experimental data trends but also were in good agreement with experimental observations. As our investigations for a more reliable and comprehensive indicator for network homogeneity continue, more detailed discussions on the results of our studies, both modelling and experimental, will be presented at the time of the conference.