(155b) Composition Broadening in Polymerization Systems

Because of their stochastic nature, polymerization processes generate mixtures of chains that differ from each other in their molecular architecture. One should not expect chains with the same number of repeating units, for example, even when process conditions remain the same across both time and space. These differences are mostly determined by the interaction among the various steps of the polymerization mechanism. Other factors play important roles too, with two of the most important additional factors being reactor configuration and mixing. Our work considers the impact all of these effects on polymer composition.

 Polymer composition is a key quality characteristic. The production of Linear Low Density Polyethylene (LLDPE) illustrates its importance. In this system, a-olefins like butene, hexene, and octene introduce small side chains into the ethylene based polymer backbone. These side chains reduce the density and the crystallinity of the final polymer, a desirable characteristic for certain applications. Increasing levels of a-olefins can also lead to products with lower molecular weights, since a-olefins are typically not as reactive as ethylene towards chain growth. The strong dependency of key product properties on the level of incorporation of a-olefins comes at a cost, however, since missing a target by adding too much or too little of these comonomers can lead to products with unexpected properties.

Tadmor and Biesenberger[1] already looked at differences in molecular weight distributions using continuous tank reactors with both perfect micromixing and complete segregation. Their intention was to illustrate how incomplete mixing affects the spread of those distributions in reactors that would be homogeneous otherwise. Mecklenburgh[2] used a Free Radical polymerization system to study the impact of mixing in various reactor configurations. The study concluded by stating that copolymerization distribution curves should provide insight into the mixing patterns inside a reactor and lamented on the lack of analytical techniques to validate this hypothesis. In a more recent attempt at measuring micromixing, Atiqullah and Nauman[3] applied gradient elution thin-layer chromatography to the styrene - methyl methacrylate system. They developed an extension of previous lamellar stretch models that included backmixing effects, but concluded that their reactor was too small for them to observe the impact of micromixing on copolymer composition distributions. The impact of imperfect mixing on polymer properties has been addresses by various authors by relying on compartmental models, a good example of this being the work of Marini and Georgakis[4]. In this work, polydispersity index was used as a measure of quality of Low Density Polyethylene (LDPE), and the effect on this property was studied for different reactor configurations and process conditions. This work was later reviewed and analyzed in more detail, with particular emphasis on compartmental models, by Zhang and Ray[5].

Using examples in Free Radical polymerization, we rely on computational tools to study the spread in the polymer composition distribution under the influence of the factors mentioned earlier. Our work shows how critical comonomer reactivity is and how important it is to control all factors in order to generate a product with the desired composition distribution. Comonomer reactivity is critical because wide residence time distributions lead to differences in conversion and, as a result of this, differences in composition of the unreacted monomer mixture. These differences in turn lead to wider polymer composition distribution. In extreme cases, multimodal distributions are even possible. Our work uses simple geometries to illustrate these problems and demonstrates that product quality is highly sensitive to what most would consider simple details in polymerization reactor design.