Effects of Water Activity and Temperature on the Caking Properties of Amorphous Carbohydrate Powders

Authors: Sukritta Anantawittayanon, Takumi Mochizuki, Kiyoshi Kawai

Abstract

Water sorption reduces the glass transition temperature (Tg) of amorphous carbohydrate powders due to water plasticization. Caking of amorphous powder occurs when Tg decreases below the storage temperature (T), that is, when the glass-to-rubber transition occurs. Although glass-to-rubber transition also occurs when *T *is greater than Tg, knowledge regarding the caking of amorphous powders induced by T elevation is limited. Thus, caking properties were investigated using amorphous carbohydrate powders with varying water activity (aw) values prepared at 25 °C, stored at a higher temperature, and then returned to 25 °C (T-cycled samples) for storage. Maltodextrin and glucose mixtures at weight ratios of 0, 0.1, and 0.2 glucose were employed. The caking behavior of T-cycled powders with high aw values was similar to that of aw-cycled samples (dried powders were stored under various aw conditions and then returned to the dry condition via vacuum-drying) reported previously. T-cycled powders with a low aw value, by contrast, were resistant to caking even in the rubbery state. This suggests that water molecules support the progression of caking as the binder under high-aw conditions. To analyze the hydration level at which water molecules begin to act as a binder for caking, determination of the multilayer adsorbed water content and multilayer adsorbed aw values is proposed. The fracture stress increased with increases in T − Tg, depending on the sample. The binding effect of water also contributed to the formation of a harder cake.

Abbreviations

Tg, glass transition temperature; W, water content; aw, water activity; T, storage temperature; MD, maltodextrin; DM, dry matter; Wm, monolayer water content; awm, monolayer water activity; Wc, critical water content; awc, critical water activity.

INTRODUCTION

Caking of powders is an important issue in the food, feed, fertilizer, and pharmaceutical industries [1, 2, 3]. The caking process is classified into four bridging, agglomeration, compaction, and liquefaction [1]. From stable “free-flowing” particles, surface deformation at contact points initiates bridge formation. These bridges then irreversibly consolidate during the agglomeration stage, although the overall porosity of the system is preserved. The compaction stage is marked by a significant decline in system integrity due to thickening of the bridges resulting from viscous flow. Finally, the interparticle space decreases and eventually vanishes in the liquefaction stage.

In amorphous powders, caking behavior is characterized based on the sticky point temperature, collapse temperature, and glass transition temperature (Tg) [4]. The sticky point temperature was originally defined as the temperature at which the energy required to stir the powder with a propeller increases dramatically [5]. The collapse temperature is defined as the temperature at which structural weakening leads to changes such as volume shrinkage and sticking of dry powders [6, 7]. The Tg represents the temperature at which the amorphous substance changes from a “solid-like” glassy stage to a more “liquid-like” rubbery stage, which can be identified based on the change in heat capacity determined using calorimetric methods [2, 8]. The sticky point temperature of the sucrose/fructose model was found to be close to the end point of the glass transition [9].

Because Tg decreases with increasing water content (W) and water activity (aw), amorphous materials can undergo glass-to-rubber transition even under isothermal conditions when Tg decreases below the storage temperature (T) due to water sorption. The caking of amorphous carbohydrate-rich food powders induced by water sorption has been extensively studied for maca [10], skim milk [11], mango [12], lactose-milk protein isolate [13], and tomato [14]. In the previous studies [10, 12], fully dried amorphous powders were stored under a higher-aw condition at constant temperature (25 °C) and then returned to the dry condition; these samples are hereafter referred to as aw-cycled. The results of the previous studies indicated that the degree of caking for aw-cycled powders can be characterized based on the temperature difference between T and Tg (i.e., T − Tg). No caking occurs at T − Tg < 0 (glassy state), but the degree of caking linearly increases as T − Tg increases at T − Tg > 0 (rubbery state). By contrast, limited data are available regarding the caking of amorphous powders induced by T elevation. For example, it was reported that the flowability of maltodextrin (MD) powder decreases when T is set above Tg under the condition of consolidation pressure [15]. The data, however, are insufficient to facilitate a better understanding of powder caking induced by temperature elevation as a function of T − Tg.

In this study, amorphous carbohydrate powders with varying aw values were prepared at 25 °C, stored at a higher temperature, and then returned to 25 °C; these samples are hereafter referred to as T-cycled. The caking properties (microscopic appearance of powder, degree of caking, and fracture stress of the cake) of the T-cycled powders were investigated, and the results were compared with those for aw-cycled samples reported previously [10, 16].

MATERIALS AND METHODS

Sample preparation

MD with a dextrose equivalent of 17–21 was obtained from San-ei Sucrochemical Co., Ltd. (Aichi, Japan). Crystalline glucose was obtained from Nacalai Tesque, Inc. (Kyoto, Japan). The amorphous carbohydrate powders were prepared by freeze-drying. Carbohydrate solutions with a 0.1 solute mass fraction were prepared by blending MD and glucose at dry weight ratios of 1.0:0.0, 0.9:0.1, and 0.8:0.2; these samples are hereafter referred to as MD, G0.1, and G0.2, respectively. The carbohydrate solutions were frozen at −30 °C for 16 h before loading into the freeze-dryer (Lyovapor L-200; Büchi Labortechnik AG, Flawil, Switzerland). The initial shelf temperature, chamber pressure, and cold-trap temperature were set to −15 °C, 38 Pa, and approximately −50 °C, respectively. Freeze-drying was carried out for 24 h, and then the freeze-dried solids were powdered by kneading in a mortar. Finally, the freeze-dried powders were sieved through a 1.4-mm mesh using a mechanical shaker (MVS-1; As One Instruments Co., Tokyo, Japan) operated at a vibration amplitude of 4.5 mm.

Scanning electron microscopy (SEM)

Sieved MD powder was spread on carbon double-sided tape and then placed on the sample stage. The sample was equilibrated under aw conditions of 0.529, 0.576, and 0.688 at 25 °C, as described above, and then the effect of aw on the microscopic appearance of the powder was observed at 25 °C by SEM using a TM4000Plus scanning electron microscope (Hitachi High-Technologies Corporation, Tokyo, Japan).

Sieved MD powder equilibrated at an aw condition of 0.576 at 25 °C was transitioned to an aw condition of 0.688 for 30, 60, and 90 min, after which the effect of aw elevation on the microscopic appearance of the powder was observed using SEM at 25 °C. In addition, sieved MD powder equilibrated at an aw condition of 0.576 at 25 °C was enclosed in a glass container, stored at 30 °C, 40 °C, and 50 °C for 7 days, and then the effect of T elevation on the microscopic appearance of the powder was observed using SEM at 25 °C.

Degree of caking of T-cycled powders

The degree of caking of amorphous carbohydrate powders was investigated according to the method employed in previous studies [16, 17, 18]. Sieved powder (approximately 0.5 g) was added to a 30-mm diameter aluminum dish and then equilibrated under various aw conditions using saturated salts at 25 °C for 7 CH3COOK (aw 0.225), MgCl2 (aw 0.328), K2CO3 (aw 0.432), Mg(NO3)2 (aw 0.529), and NaBr (aw 0.576) [19]. The effect of aw on the Tg (°C) and W (g/100 g dry matter, DM) of MD, G0.1, and G0.2 was reported previously [16], and the values are shown in Supplemental File (Table S1; see J. Appl. Glycosci. Web site). The aw-equilibrated samples were enclosed in a glass container without saturated salts and then stored at 30 °C, 40 °C, 50 °C, 60 °C, and 70 °C for 7 days, a holding period sufficient to reach equilibrium. The samples were then stored at 25 °C for 1 day and sieved as described above. The degree of caking was evaluated as a percentage of the retained weight versus the total weight. All measurements were performed in triplicate, and the results were averaged.

Fracture stress of T-cycled cakes

The fracture stress of T-cycled cakes was evaluated using a texture meter (CR-500DX; Sun Scientific Co. Ltd., Tokyo, Japan) according to a previous study [10]. The cake sample was placed on the sample stage and then compressed with a 3-mm diameter plunger at 0.5 mm/s. Fracture stress was evaluated from the first fracture point observed in the force-deformation curve. All measurements were performed in triplicate, and the results were averaged.

RESULTS AND DISCUSSION

Microscopic appearance of MD powder

SEM images of amorphous MD powders equilibrated at 25 °C for 7 days under aw conditions of 0.529, 0.576, and 0.688 are shown in Fig. 1. The Tg of MD powders with aw values of 0.529, 0.576, and 0.688 were 32.1 °C, 27.6 °C, and 9.4 °C, respectively (Table S1). The MD powders with aw values of 0.529 and 0.576 were in the glassy state (T − Tg < 0); thus, angular fragments with a shard-like appearance of broken window glass were observed. The MD sample with an aw value of 0.688 was in the rubbery state (T − Tg = 15.6 °C); thus, liquefaction-the final stage of caking-was observed.

SEM images of amorphous MD powder in which aw was elevated from 0.529 to 0.688 at 25 °C are shown in Fig. 2. It is thought that the Tg decreased from 32.1 °C (glassy state) to 9.4 °C (rubbery state) during this process (Table S1). Similar to MD powder equilibrated at an aw value of 0.529 (Fig. 1), many angular fragments were observed for the 30-min aw-elevated MD sample. The edges of the fragments became smoother after 60 min, and some of the fragments combined. These results reflect the bridging, agglomeration, and compaction stages of the caking process. The smoothened and combined fragments increased in size after 90 min, and partial liquefaction was observed.

SEM images of T-cycled amorphous MD powder with an aw value of 0.576 (from 25 °C to 30 °C, 40 °C, and 50 °C, then back to 25 °C) are shown in Fig. 3. Strictly speaking, aw is a function of temperature. Thus, an aw value obtained at 25 °C changes when the temperature is increased. However, the aw value obtained at 25 °C determines a constant W, and the W is assumed to remain constant during the development of caking due to temperature rise in the closed system; the W value would be expected to affect the caking behavior of T-cycled amorphous powders. For simplification, the aw obtained at 25 °C is referred to here as the typical aw value for T-cycled amorphous powders. The Tg of the MD powder was 27.6 °C (Table S1). Many angular fragments were observed in SEM images of the MD sample held at 30 °C, similar to glassy MD (Fig. 1) because the sample was in the rubbery state, but near to a glassy state (T − Tg = 2.4 °C). For MD samples held at 40 °C (T − Tg = 12.4 °C) and 50 °C (T − Tg = 22.4 °C), the angular fragments disappeared, and round and smooth lumps were observed. These results are indicative of liquefaction. In comparing the liquefaction characteristics of aw-elevated and T-cycled MD powders, it was noted that T-cycled MD exhibited a more droplet-like appearance than aw-elevated MD. In the rubbery state, the amorphous materials became mobile (liquid-like) and therefore became shaped like a droplet because of the interfacial tension. Thus, T-cycled MD forms a droplet-like cake. By contrast, aw-elevated MD forms a spreading cake because the interfacial tension is reduced due to the increase in W.

Degree of caking of T-cycled powders

The effect of T on the degree of caking of T-cycled MD, G0.1, and G0.2 is shown in Figs 4-a, 5-a, and 6-a, respectively. The degree of caking increased with increasing aw value and T due to the increase in molecular mobility. The variations in the degree of caking observed among the samples stored at differing T values likely resulted from the differences in Tg.

It is thought that the T − Tg parameter reflects the molecular mobility of amorphous materials. As explained earlier, the degree of caking for aw-cycled amorphous carbohydrate powders linearly increased with increasing T − Tg value in the rubbery state (T − Tg > 0) [16]. The effect of T − Tg on the degree of caking for T-cycled MD, G0.1, and G0.2 is shown in Figs 4-b, 5-b, and 6-b, respectively. The dotted lines indicate the tendency for the aw-cycled MD, G0.1, and G0.2 powders. Some powders (MD with aw values of 0.529, 0.576, and 0.688, G0.1 with aw values of 0.529 and 0.576, and G0.2 with aw values of 0.432 and 0.529) showed similar caking behavior to aw-cycled powders (dotted lines), but the others were resistant to caking even in the rubbery state (T − Tg > 0). T-cycled powders that exhibited similar caking behavior to aw-cycled powders were under high-aw conditions. This suggests that water molecules support the progression of caking as a binder under high-aw conditions.

It is known that monolayer water adsorption occurs in the low-aw region [20]. In this state, water molecules are ineffective as a binder for caking. The monolayer water content (Wm) and monolayer water activity (awm) can be determined from the water sorption behavior. Wm and awm are regarded as important parameters because biomaterials, including foods, stabilize around this hydrated state [21]. It is believed that water does not exert a binding effect in T-cycled amorphous powders when W is less than Wm and/or aw is less than awm, which prevents caking.

At hydration levels exceeding monolayer adsorption, multilayer water adsorption occurs, causing the water content to become highly sensitive to changes in aw; a small change in water content results in a large change in aw. In addition, water sorption–induced glass-to-rubber transition occurs. The W and aw values at which glass transition occurs at 25 °C (Tg becomes 25 °C) are commonly referred to as the critical W (Wc) and critical aw (awc) values, respectively [22]. When W is greater than Wc and/or aw is greater than awc, water molecules will begin to dissolve in the solute because the solute enters a mobile (i.e., rubbery) state. Thus, the hydration level at which water molecules begin to act as a binder for caking is between Wm and Wc and/or between awm and awc.

Taking the above data into account, the following degrees of multilayer adsorbed W and aw are proposed.

The higher these values, the greater is the number of water molecules involved in the multilayer water adsorption process. The Wm, awm, Wc, awc, degree of multilayer W, and degree of multilayer aw values for MD, G0.1, and G0.2 are listed in Table S1 [16]. As discussed above, MD with an aw value of 0.432, G0.1 with an aw value of 0.328, and G0.2 with aw values of 0.225 and 0.328 were resistant to caking induced by T-cycling. The degrees of multilayer adsorbed W and aw were less than 0.44 and 0.56, respectively; they are denoted as low-hydrated T-cycled samples hereafter. By contrast, similar caking behavior was observed between T-cycle and aw-cycle for MD with an aw value ≥ 0.529, G0.1 with an aw value ≥ 0.432, and G0.2 with an aw value ≥ 0.432. The degrees of multilayer adsorbed W and aw were greater than 0.71 and 0.82, respectively; they are denoted as high-hydrated T-cycled samples hereafter. These data suggest that water molecules begin to act as a binder for T-cycle-induced caking at a degree of multilayer W between 0.44 and 0.71 and at a degree of multilayer aw between 0.56 and 0.82.

In the previous studies [18, 23], it was demonstrated that the aw-cycled caking was described by a stretched exponential function (Eq. 3),

where α and n are constants. In the case of T-cycled caking, Eq. 3 can be rearranged as follows,

where β and m are constants (see Appendix). Since aw/awc can be converted to T/Tg through the relationship between Tg and aw (see Appendix), the degree of caking of aw-cycled samples can be also described by Eq. 4. The effect of T − Tg on the degree of caking for the aw-cycled [16], high-hydrated T-cycled, and low-hydrated T-cycled samples (MD, G0.1, and G0.2) is shown in Fig. 7. The solid, dash, and dotted curves were obtained by the fitting Eq. 4 to the data for aw-cycled, high-hydrated T-cycled, and low-hydrated T-cycled samples, respectively. As expected, there was a minor difference in the degree of caking between aw-cycled (β 0.01 and m 110) and high-hydrated T-cycled (β 0.04 and m 87) samples. The low-hydrated T-cycled sample (β 2×10^−12^ and m 367), on the other hand, showed a largely different behavior from the aw-cycled and high-hydrated T-cycled samples. These results will provide a quantitative understanding and prediction for the caking behavior of amorphous powders in terms of both aw-cycle and T-cycle.

Fracture stress of T-cycled cakes

No fracture peaks were observed in T-cycled cakes at a degree of caking < 50 %, as the cakes were not fully consolidated. The initially rubbery samples (equilibrated Tg significantly below 25 °C) also did not show fracture peaks due to their highly viscous state. In addition, some T-cycled cakes held at 70 °C turned into fragile lamina. As a result, only samples of MD with aw values of 0.529 and 0.572, G0.1 with aw values of 0.432 and 0.529, and G0.2 with aw values of 0.328 and 0.432, all held at temperatures ranging from 40 °C to 60 °C, were used in the fracture stress analysis of T-cycled cakes.

The effect of T − Tg on the fracture stress of T-cycled MD, G0.1, and G0.2 cakes is shown in Fig. 8. The fracture stress increased with increasing T − Tg depending on the sample. MD samples with aw values of 0.529 and 0.572 showed much higher fracture stresses than G0.1 and G0.2 at each T − Tg value. As discussed above, water molecules act as a binder for caking; thus, the caked samples were strengthened physically. For comparison, the fracture stress of aw-cycled MD cake (from a fully dried state to aw 0.753, then back to a fully dried state at 25 °C) was taken from a previous study [10]. The sample was caked at T − Tg = 41.7 °C, and the fracture stress was 27.7 MPa; the fracture stress was in good agreement with the extrapolated value obtained from the tendency (dotted lines) for T-cycled MD cakes.

G0.1 and G0.2 exhibited a lower fracture stress than MD at each T − Tg. This observation was attributed to the lack of water to function as a binder for caking in G0.1 with an aw value of 0.432 and G0.2 with an aw value of 0.328. Because G0.1 with an aw value of 0.529 (Tg = 15.3 °C) and G0.2 with an aw value of 0.432 (Tg = 16.9 °C) were in the rubbery state at 25 °C (Table S1), it is unlikely that their cakes would show a high fracture stress. From these results, it is concluded that amorphous carbohydrate powders that can form harder cakes need to be in the glassy state despite a high aw; meaning that the anhydrous Tg must be as high as possible.

CONCLUSION

The caking properties of T-cycled amorphous carbohydrate powders were investigated, and the results were compared with those of aw-cycled samples based on the T − Tg values. Although T-cycled powders with high aw values showed a similar caking behavior to aw-cycled powders, T-cycled powders with low aw values were resistant to caking even in the rubbery state. These results suggest that water molecules begin to act as a binder for T-cycle-induced caking at a degree of multilayer W between 0.44 and 0.71 and at a degree of multilayer aw between 0.56 and 0.82. The fracture stress increased with increasing T − Tg depending on the sample. The binding effect of water also contributes to the formation of harder cakes. Amorphous carbohydrate powders that can form harder cakes exist in a glassy state despite having a high aw. These results enhance current understanding of the caking behavior of amorphous carbohydrate powders based on the value of T − Tg. The caking properties of amorphous carbohydrate powders were investigated under the equilibrium and/or equilibrium-like conditions, but they are intrinsically time-dependent phenomenon. Kinetic analysis of the caking is one of the future issues.

APPENDEX

In the previous studies [18, 23], it was demonstrated that the aw-cycled caking was described by a stretched exponential function (Eq. 3). It is noted that Eq. 3 is mathematically equivalent to the Avrami equation, which describes time-dependency of crystallization at a constant temperature. That is, the Avrami equation was rearranged to aw-dependency of caking according to the analogy of the agglomeration of particles.

The aw/awc in Eq. 3 is a parameter reflecting the molecular mobility of rubbery materials. The higher aw/awc, the higher molecular mobility at a constant temperature. Similarly, the T/Tg in Eq. 4 is a parameter reflecting the molecular mobility of rubbery materials at a constant aw. Thus, the aw/awc in Eq. 3 can be replaced with the T/Tg as shown in Eq. 4.

Effect of aw on the equilibrium water content (W) of amorphous materials can be described by the Guggenheim–Anderson–de Boer (GAB) equation (Eq. a1),

where Wm, C, and K are constants. In addition, effect of water content on the Tg of amorphous materials can be described by the Gordon–Taylor (GT) equation (Eq. a2),

where M is the mass fraction of water (dimensionless), Tgs is the anhydrous Tg of solute (K), Tgw is the Tg of water (136 K), and k is a constant (dimensionless). Since W and M are interchangeable, the aw can correspond to Tg through the GAB and GT equations. The GT and GAB parameters (Tgs, k, Wm, C, and K) are listed in Table S1.

CONFLICTS OF INTEREST

The authors declare that they have no competing interests.

Supplementary Material