An interesting thing to note from the results in Table 6 is the significant increase in viscosity of the coating solution of more than ten times from experiment γ5C to γ5F and similar from γ5B to γ5E as only the amount of Na2SO4 in the coating solution was changed. It seems that the increased Na2SO4 concentration has a salting-out effect on the CMC causing the significant increase in viscosity [16]. Visually there was a difference in solution appearance as well, as the 10 wt./wt.% solutions were slightly dim whereas the 30 wt./wt.% solutions were completely opaque.
From microscopy analysis of the coated granules in Fig. 12 it was seen that the coating surface morphology changes significantly with droplet viscosity. Whereas the coating layer is smooth and homogenous in batch γ5B with a low droplet viscosity, the surface layer becomes increasingly rough and raspberry-like when the droplet viscosity increases from batch γ5B to γ5C and further from batch γ5C to γ5F. These results are in full accordance with experiments by Link and Schlünder [9] indicating that the viscosity-affecting dry matter contents of the coating solution have a primary importance for the resulting roughness of the coating layer, independently of the porosity of the core material. This was investigated further by SEM analysis.
Fig. 13 shows SEM pictures of granules and cut-through edge profiles of coated granules from batch γ5B and γ5F. It can be seen by comparing Fig. 13A with Fig. 13B that although both coating layers are fluffy and covered with small dried droplets, the increase in coating liquid viscosity from Fig. 13A to B makes the coating layer flaky, stratified and inhomogeneous. There is a significant shrinkage in the coating layer in Fig. 13B as the coating liquid layer has dried first on the outer surface leaving the inner layer still wet. As the inner wet layer afterwards dries by the evaporation of remaining coating liquid solvent, the inner coating layer is left as a highly porous cave-like structure, which eventually shrinks and cracks. This is further verified in Fig. 13D where a cut-through edge profile of the coating layer of a granule from batch γ5F is seen to be highly porous with a cave-like structure. These observations agree with experiments by Link and Schlünder [9] and results by Keningley et al. [17]. The sharp distinction between core and coating material complies fully with the results in Table 6 stating that for experiment γ5F, the time of droplet penetration is somewhat larger than the droplet drying time. Hence as expected, the majority of the coating material remains at the core surface in Fig. 13D.