DescriptionMost food materials are composed of polymeric molecules with different chemistry and properties. Increasing demand for food products with new/improved functionalities require formulations to include/exclude ingredients. Processability, texture, stability and palatability of the food products are greatly influenced by interactions and miscibility/immiscibility between these polymeric components. Molecular and thermodynamic basis of these phenomena are still not well understood and are the focus of scientific debate. A set of quantitative predictive rules are needed to be developed.
The major objective of this dissertation was to fundamentally understand the molecular and thermodynamic basis of miscibility and to develop quantitative miscibility predictions for carbohydrate mixtures. Dextrans (glucose polymers) with different molecular weights (Mw) and chemically derivatized forms were used as model carbohydrate polymers. Thermal analysis on individual and mixtures of standard dextrans showed that physical blend of dextrans was immiscible due to the diffusion barrier, whereas freeze-dried solution of these dextrans was miscible. In the mixtures of chemically derivatized dextrans, thermal analysis showed miscible or immiscible systems depending on the concentration and ionic strength through addition of salt. Systematic differences in the FTIR spectra of miscible systems with different component ratios were assigned to the change in hydrogen-bonding distribution resulting from changes in intra- and inter-molecular interactions, whereas FTIR spectra of immiscible systems didn't show such systematic changes, indicating insufficient hydrogen-bonding to form miscible systems.
In order to quantitatively predict miscibility, first, a methodology was developed to determine solubility parameters of dextrans with different Mw using their Tg. Using these solubility parameters, thermodynamic ideas based on the number of configurational arrangements and quantitative measures of dispersive interactions (Flory-Huggins theory) were demonstrated to be insufficient to quantitatively explain the miscibility in carbohydrate systems. This failure was due to the limitation of these ideas in underestimating the effect of specific bonding interactions, including hydrogen-bonds. A more advanced framework, Painter-Coleman association model, quantitatively demonstrated that hydrogen-bonding significantly contributed to predictive miscibility in carbohydrate systems. It was shown that with appropriate approximations it was possible to successfully predict miscibility in dextran systems. The quantitative understanding gained with dextrans was validated on real carbohydrate systems by testing miscibility/immiscibility in inulin-amylopectin systems.