Phase separation in amorphous hydrophobically modified starch–sucrose blends: Glass transition, matrix dynamics and phase behavior

Highlights

• Amorphous–amorphous phase separation is observed for modified starch–sucrose blends.

• Double glass transition (T_g) and mobile compounds induce dynamic phase behavior.

• T_g of sucrose-rich phase impacts T_g of the starch-rich phase via water migration.

• Quantitative phase model based on a glass transition temperature (T_g) master curve.

• The current findings explain many stability phenomena in food and pharmaceutics.

Abstract

The phase behavior and matrix dynamics of amorphous blends of octenyl succinic anhydride (OSA) modified starch and sucrose was studied as function of blend composition and water content. Phase separation into two amorphous phases, one enriched in OSA starch and the other in sucrose, was confirmed by differential scanning calorimetry (DSC). DSC and ¹H solid-state NMR show that the phase separation is only partial. The glass transition temperature (T_g) of the OSA starch-rich phase was found to be ∼30–100 K higher than the T_g of the sucrose-rich phase, depending on blend composition and water content. A novel type of coupling between changes in physical state of the sucrose-rich phase and plasticizer redistribution is proposed, leading to an unexpected increase of the glass transition temperature of the modified starch-rich phase at higher matrix water contents. A quantitative model for the phase separation of the anhydrous blends into two amorphous phases is presented. The model predicts that, with increasing blend sucrose content, the weight fraction of the sucrose-rich phase decreases, while the sucrose content of both the OSA starch-rich phase and the sucrose-rich phase increases. This novel phenomenon is relevant in the understanding of the stability and performance of multiphase food and pharmaceutical components.

Introduction

One of the principal applications of microencapsulation is to protect sensitive actives during storage against the detrimental effects of water and atmospheric oxygen. In most cases such protective matrices consist of amorphous carbohydrates in the glassy state, so-called glass encapsulation systems (Ubbink, 2016). Non-optimal barrier properties of the encapsulation matrix lead to increased rates of chemical degradation, mainly by oxidation, of the encapsulated actives (Karel, 1990). In addition, the active may prematurely diffuse out of the matrix into the environment. It is therefore of significant importance to optimize the formulation of carbohydrate-based encapsulation matrices in order to maximize the barrier properties and thereby improve the stability and shelf life of microencapsulated actives (Reineccius & Yan, 2016; Ubbink, Burbidge, & Mezzenga, 2008).

At the molecular level, barrier properties are governed by two parameters: the solubility and the mobility (diffusivity) of the permeating compounds in the barrier material. The mobility of small molecules, such as water and oxygen, in glassy carbohydrates was initially believed to be governed primarily by the proximity of the glass transition temperature (T_g) of the matrix to the temperature of the storage environment (Levine, 2002). It has since been recognized that this interpretation of molecular mobility is inadequate (Cicerone & Douglas, 2012; Cicerone, Pikal, & Qian, 2015; Ubbink, 2016; Ubbink & Krüger, 2006). While the mobility of the larger carbohydrate molecules that constitute the matrix is governed by the α-relaxation and therefore effectively cease at temperatures below T_g, the mobility of smaller molecules in the matrix is thought to be related to the β-relaxation and remains appreciable even in the glassy state (Cicerone & Douglas, 2012). The β-relaxation, which is in turn hypothesized to be related to the molecular packing of the matrix in the glassy state, can be directly probed by positron annihilation lifetime spectroscopy (PALS) (Ubbink, 2016; Ubbink et al., 2008).

The principal variables affecting the barrier properties of glassy carbohydrate matrices are temperature, water content and matrix formulation. Of particular importance is water, first and foremost as it is a strong plasticizer of amorphous carbohydrates, reducing the matrix T_g significantly (Roos, 1995). Secondly, in the glassy state, water impacts the molecular packing and the dynamics of amorphous carbohydrates via a complex mechanism where, depending on the concentration present, water may act as an antiplasticizer or plasticizer of the carbohydrate matrix (Ubbink, 2016). At low concentrations (≲5 wt.% Roussenova, Murith, Alam, & Ubbink, 2010), water acts as an antiplasticizer, lowering the matrix T_g but also reducing the average size of the free volume holes in the glassy state (Roussenova, 2011; Townrow, Kilburn, Alam, & Ubbink, 2007). At higher concentrations, water acts as a plasticizer, continuing to reduce T_g of the matrix but increasing the average size of the free volume holes in the glassy state (Townrow et al., 2007). In this regime, water not only enhances its own molecular mobility (Tromp, Parker, & Ring, 1997), but also of other small molecules (Gunning, Parker, & Ring, 2000; Schoonman, Ubbink, Bisperink, Meste, & Karel, 2002). The water content of carbohydrate-based glass encapsulation systems should therefore be as close to the so-called “antiplasticization threshold” as possible (Seow, 2010), as this minimizes the local free volume.

Low molecular weight matrix additives other than water have been shown to impact the molecular packing of glassy carbohydrates as well (Ubbink, 2016, Ubbink et al., 2008). Specifically, low molecular weight polyols, such as glycerol and sorbitol (Roussenova et al., 2010), and mono- and disaccharides such as glucose and maltose (Kilburn, Claude, Schweizer, Alam, & Ubbink, 2005; Kilburn et al., 2004; Townrow et al., 2007; Townrow, Roussenova, Giardello, Alam, & Ubbink, 2010) act as antiplasticizers. These molecules reduce the average molecular hole size of glassy matrices consisting of intermediate and high molecular weight carbohydrates, such as starches and maltodextrins. The decrease in molecular hole size with increasing additive content corroborates findings from shelf-life testing (Kasapis, Norton, & Ubbink, 2009; Ubbink, 2016).

Blends of starches or maltodextrins with low molecular weight polyols or mono- and disaccharides usually mix well at the molecular level and therefore consist of a single phase. Recently, however, it was observed that blends of flour and sucrose may show a limited degree of phase inhomogeneity, as witnessed by detailed modeling of the glass transitions as determined by differential scanning calorimetry (DSC) (Roudaut & Wallecan, 2015). To assume molecular miscibility of blend components may thus be incorrect. A more pronounced separation into two distinct phases was observed for blends of octenyl succinic anhydride-modified (OSA) starch and sucrose (Tedeschi, Leuenberger, & Ubbink, 2016).

The partial incompatibility of matrix components leads to a number of phenomena not observed for homogeneous systems. For some applications, such as the targeted release of pharmaceuticals, matrix incompatibility may be sought in order to produce two distinct phases with specific properties, one acting to impart stability during storage and the other to act as the “delivery vessel” (Tedeschi et al., 2016). Phase separation of the constituents of encapsulation matrices could however also lead to reduced barrier properties of glassy carbohydrate blends and is thus a vital property to consider and control in the development of matrix formulations (Hughes et al., 2016, Tedeschi et al., 2016).

In this study we first present the analysis of the glass transition behavior of the blends, as measured by DSC, as a function of composition and water content. Then, using ¹H low-resolution solid-state NMR, aspects of the matrix dynamics that are related to the phase behavior of the matrices. Finally, we introduce a model to quantify both the relative abundance of the phases present within the anhydrous blends and the composition of these phases. Our overarching aim is to provide a quantitative description of the phase behavior, matrix structure and component dynamics in relation to the blend composition and thermodynamic parameters.