Mathematical models that describe the outcomes of surface initiated polymerization (SIP) accurately can aid in the precise fulfillment of design objectives for polymer brushes. If they possess the ability to predict polymer brush properties, such models can be used to zero in on the best synthesis conditions to achieve target attributes such as thickness, wettability, swelling ratio, friction coefficient, roughness etc. We solve this problem by employing a Design of Experiments (DOE) approach to describe brush properties as a mathematical function of SIP parameters. Specifically, we have investigated the surface initiated-atom transfer radical polymerization (SI-ATRP) of a model zwitterionic hydrogel, Poly[2-(methacryloyloxy) ethyl dimethyl-(3-sulfopropyl) ammonium hydroxide] (PMEDSAH).
PMEDSAH is of special interest to the stem cell engineering community as it possesses the property of sustaining feeder-free and xeno-free stem cell proliferation. We will begin by examining results that reveal how the rate of stem cell propagation is sensitive to modifications of the gel architecture of PMEDSAH. We hypothesize that a predictive model will help us access polymer brushes possessing the optimal gel architecture for facilitating rapid expansion of hESC populations. Then we will turn our attention to the development and validation of a predictive model that simultaneously captures both phenomena that determine the gel architecture of PMEDSAH – the rate of the SI-ATRP reaction and the ensuing degree of zwitterionic self-association. Using a systematic factorial design of experiments, we have quantified the impact of experimental variables on the chain length as well as the self-association behavior of the zwitterionic polymer chains. After identifying four regimes of gel architecture, we obtained predictions for the operating boundaries within which each regime can be accessed experimentally. To verify whether the model can "dial up" the desired regime of gel architecture, validation experiments were performed and the results were found to lie within the 95% confidence interval of model predictions In the conclusion, we will see how DOE offers an interesting quantitative framework within which the design and synthesis of polymer brushes for biomedical applications can be explored, with its advantages of multiple property prediction and ease of implementation and interpretation.