Prediction of Surface Component Composition, Migration, and Packing
The Need for a Multiscale Model for Surface Composition Prediction
The spray drying process, together with the feed material, determines the micro- to macroscale structures of spray dried particles, and hence their properties and functionalities (see Figure 2.4.1). The surface structure and composition are especially important, because they significantly influence particles’ properties (such as
FIGURE 2.4.1 Multiscale view of the spray drying system.
wettability and stickiness) as well as their dissolution or rewetting behavior (Vehring et al., 2007; Wu et al., 2014). Consequently, reliable prediction of surface composition becomes crucial for powder quality control.
By resorting to electron spectroscopy for chemical analysis (ESCA, also known as X-ray photoelectron spectroscopy (XPS)), many experimental investigations have reported a big difference between the surface composition and the bulk composition of spray dried particles. This observation suggests that solute segregation must occur during spray drying (Kim et al., 2003; Faldt and Bergenstahl, 1994, 1996a, 1996b).
The distributed parameter model developed by Langrish and his group (Wang and Langrish, 2009; Wang et al., 2013) is one representative effort to characterize solute segregation. In order to obtain localized composition, a particle was divided into a number of shells. The diffusive mass flux of each solute component was calculated individually for each layer before being combined to estimate the total mass flux. However, the predicted surface concentration was always lower than that calculated from the XPS measurements. Another representative approach was proposed by Chen et al. (2011). In their work, the governing equations of convective diffusion were simplified into a set of algebraic equations by using the idea of the characteristic length. The solute composition on the particle surface was then obtained by solving those algebraic equations. For a two-component system (i.e., the protein-lactose system), the theoretical estimates, however, were always greater than the XPS results.
The methods listed above are both based on the continuum theory. It is understandable that their predictions cannot match XPS results on surface composition. As shown in Figure 2.4.2, the surface layer in a continuum model is the layer a (not drawn to scale), whose thickness corresponds to the mesh grid size used in solving the governing equations of mass transfer. The XPS detectable surface layer, /, has a thickness of -10 nm only, which is even smaller than the size of many solute molecules (such as protein and fat). The continuum theory is not capable of characterizing solute segregation and powder surface formation at nano-scale, and hence this theory alone cannot predict surface composition that needs to be validated by the XPS data.