(345c) Experimental and Modeling Analysis of Lactic Acid Polycondensation

Polylactic acid (PLA) is a biodegradable polyester exhibiting attractive mechanical and chemical properties at high molecular weight. As such, it is a promising candidate to replace non-degradable petroleum-based polymers of large use. High molecular weight PLA is industrially produced by a three step process: synthesis of a low molecular weight pre-polymer by polycondensation of lactic acid (LA), separation and recovery of the cyclic dimer, lactide, and production of the final PLA by ring opening polymerization (ROP) of lactide. The polycondensation step, an intrinsically low reaction, is usually the rate determining step of the entire process. Thus, a good understanding of its mechanism is important in order to design and optimize the whole industrial process.

Polycondensation kinetic models are available in the literature since many years. Most of the papers refer to the probabilistic approach by Flory, a simplified model involving assumptions such as equal reactivity of all functional groups and ideal thermodynamics. On the other hand, the same assumptions were proved to be not fulfilled for several systems of large interest, such as PET and Nylon 6,6. Thus, step growth polymerizations are reactions exhibiting remarkable thermodynamic and kinetic nonidealities. While thermodynamic nonidealities refers to the non ideal behaviour of the reacting components in the polymer melt mixture, kinetic nonidealities usually reflect reactivity dependences upon polymer chain length.

In this work, a large set of equilibrium data for PLA condensation has been produced. Batch reactions were carried out at different temperatures, ranging from 130 to 170 oC, and at long enough reaction time to ensure that chemical equilibrium is established. Different equilibrium compositions were explored, in a range of water contents of industrial interest (from 0.05 to 0.25 water molar fraction). In all cases, a detailed characterization of the equilibrium composition has been carried out: the polymer water content was determined by Karl Fischer titration and the entire distribution of the oligomer concentrations was determined by a recently published method based on reverse phase HPLC.

Such a detailed set of equilibrium data, were analyzed to estimate the impact of the nonidealities mentioned above. To overcome the major difficulty of a reliable evaluation of the activity coefficients of all components, the semi-empirical strategy proposed by Doherty et al. for polycondensation reactions has been applied: ?apparent? equilibrium and rate coefficients are introduced which are function of temperature and composition of a reference species, typically water. Moreover, different reactivity of chains with different length has been considered, thus accounting for kinetic nonidealities. A satisfactory description of average polymer properties is achieved even in the simplest, ideal case. On the other hand, a detailed description of system composition has been obtained only considering rate coefficients function of chain length and accounting for the dependence of apparent equilibrium constants upon the system composition. In particular, the system can be accurately simulated by reducing it to a ternary system composed by water, monomer and polymer, i.e. lumping together all oligomers.