Glass transition and physical properties of polyol-plasticised pullulan-starch blends at low moisture

The effects of water and polyols, at low weight fractions, on water sorption behaviour, thermal and mechanical properties, and gas permeability of blends of pullulan and gelatinised corn starch (ps), prepared by either hot pressing or casting aqueous solutions, were studied. Incorporation of sorbitol or xylose in the ps blends resulted in lower equilibrium moisture contents in the low to medium a(w) range, and much higher moisture contents at a(w) > 0.75; the Guggenheim-Anderson-DeBoer isotherm model adequately described the sorption data up to the a(w) of 0.9. Water and polyols exerted a strong plasticising action, lowering the T(g) of the blends and allowing enthalpy relaxation events to occur during aging of the amorphous specimens. At the low polyol levels examined (10 and 20% dry basis), a single glass transition temperature for the polymeric constituents was identified in all samples by DSC and DMTA; apparent activation energies of 226-296 kJ mol-1 for the alpha-relaxation were estimated from multifrequency mechanical measurements. Large deformation mechanical tests demonstrated sharp decreases in Young's moduli with increasing levels of polyol and water, typical of the glass-rubber transition of amorphous polymers; the relationship of flexural modulus and moisture content was quantified using the Fermi's model. The relationship between maximum stress (sigma(max)) and water content showed an increase in stiffness of the blends from 7 to 11% moisture, and a strong softening effect when the water content exceeded this range. The Arthenius plots of O2 and CO2 permeability data showed distinct changes in slope in the glass transition region of the blends. Analysis of viscoelastic data with the time-temperature superposition principle and treatment of gas permeability data with the Williams-Landel-Ferry equation suggested that application of this model is meaningful if the coefficients, C1 and C2, are allowed to vary instead of assuming their 'universal' values.