Effect of Freezing Rate and Starch Granular Morphology on Ice Formation and Non-Freezable Water Content of Flour and Starch Gels

Abstract

Ice formation and non-freezable water (W_NFW) of rice flour and tapioca starch gels were studied at two different freezing rates (–10 and –100°C/min) using differential scanning calorimetry. Ice crystal growth was observed in the slow freezing but not in the fast one. Ice melting enthalpies, however, were similar since more ice formed in holding and reheating steps. Melting enthalpy of fully gelatinized systems with water contents ~ 0.50–0.66 was associated to starch composition and granule morphology. Highly swollen tapioca starch gave the lowest enthalpy and the highest W_NFW (0.40 g/g dry starch versus 0.32 and 0.38 g/g dry starch of normal and waxy rice flours, respectively). The further studies revealed that the W_NFW values were associated to swelling power, solubility, and granule morphology.

Keywords:

• Ice formation
• Non-freezable water
• Rice flour
• Tapioca starch
• Granule morphology of starch

Introduction

In the past decades, demands of frozen ready-meals have risen substantially in food markets. The current market in the United States is very large and well-established. In Asia, especially South East Asia, the market is rather small but growing rapidly. Frozen starch-based meals are commonly available in the South East Asia. However, these types of products still not meet most of the consumer’s preference since they typically have poor texture upon storage at low temperature as a result of starch retrogradation^[1] and phase separation during the process of ice formation.^[2] Understanding the state of water and how it forms ice during processing and storage could help improving textural quality of starch-based frozen foods.

Often constituted in food products, water is recognized as the most important dispersing medium which contributes to their sensory, processing, and storage properties.^[3,4] Moreover, its interaction with other food constituents, such as starch and other polysaccharides, controls the structure-property relationships of the food products. Generally, the presence of water in food matrix is defined into three categories: non-freezable water, freezable water, and free water.^[5] Non-freezable water is strongly bound to the hydrophilic groups which are normally present in the food matrix. This type of water immobilizes and remains unfrozen under subfreezing temperatures^[6] and it is also responsible for the stability of foods, i.e., texture and microbiological stability.^[7] Freezable water is loosely bound to molecules in food matrix. It can be frozen and melted at a temperature lower than pure bulk water (0°C).^[8] On the other hand, free water is mobile and unbound to the food matrix. It can be crystallized to ice crystal with melting and freezing points at the temperatures identical to those of pure bulk water. The amount of these three types of water in any particular systems has been investigated as functions of water content,^[9,10] hydrophilicity of food components,^[11] freezing temperature,^[9] freezing rate,^[12] and the presence of other small molecules such as sucrose.^[13]

Formation of ice crystals is associated to freezable water and free water present in food matrix. Since many solute molecules (i.e., polysaccharide, protein, salt, sugar, lipid, etc.) usually exist in the food matrix, during freezing ice formation simultaneously takes place with the freeze-concentration process of those solutes. Thus, freezing temperature is very much reduced compared to that of pure water.^[14] Similar to a typical crystallization process, the amount and size of ice crystal formed are dependent on freezing rate.^[15] Slow freezing usually leads to the process of ice crystal growth. On the other hand, rapid freezing results in less ice formation (since more water is in glassy state) and shows low glass transition temperature (T_g); yet, upon reheating above the T_g of the freeze-concentrated matrix, more ice may possibly form, which is also known as the devitrification. This is mainly due to the decreased viscosity of the unfrozen matrix.^[14] When the temperature is further increased, all ice crystals melt and the freezable water is segregated from the food matrix, which can consequently deteriorate the food texture and its stability.^[16]

As ice formation and distribution affects texture, quality, and stability of food,^[17] molecules with strong water-binding capacity, that yield large amount of non-freezable water and thus reducing the formation of ice crystal, are desirable in frozen products. Interaction of the food product with water can be investigated through the determination of non-freezable water content using a differential scanning calorimetry (DSC) technique based on freezing and thawing approach. The non-freezable water can be utilized to select suitable processing and storage conditions for relevant dehydrated and frozen foods.^[18] In the DSC technique, the amount of non-freezable water can be determined by measuring the melting enthalpy of ice (freezable bound water and free water) with respect to the water content in the investigated system.^[8]

For gelatinized starch-based systems, their freezable and non-freezable water have been extensively characterized as functions of water content,^[19,20] botanical origin of starch,^[21] amylose-lipid complex,^[19] starch modification,^[10] and heating temperature.^[21] Limited studies are available regarding the influence of starch granular morphology and freezing rate on the formation of ice and water state in starch samples.^[19,21,22] Information on the effect of structure or morphology of starch granule on water-binding capacity (or the amount of non-freezable water) can be very useful in predicting phase behavior and properties of food systems under freezing and thawing conditions. As rice flours (RFs) and tapioca starch (TS) are generally used as food ingredients in Thailand and other countries in Asia, the purpose of this study was to investigate the effect of freezing rate on the formation of ice (from freezable bound water and free water) and the amount of non-freezable water on RFs (non-waxy and waxy samples) and TS using the DSC technique. Furthermore, the influence of the different starch granular morphology on the amount of non-freezable water was examined in connection with the observation via optical microscopy. To further elucidate the effect, pregelatinized tapioca starch (PG) was studied and the results were then compared to those of TS. The effect of chemical composition and granular morphology on water-binding capacity, ice formation, and amount of non-freezable water was discussed.

Materials and Methods

Materials

Rice flour, non-waxy (RF), and waxy (WRF), and tapioca starch (TS) (Dragonfish®, Thailand) were purchased from the local supermarket. PG (Pretap®) was kindly supplied by Tapioca Development Corp., Ltd., Thailand.

Material Composition

Moisture content was measured by drying the sample (3 g) in an oven at 105°C for 3 h, according to the method of the Association of Official Agricultural Chemists (AOAC),^[23] and was calculated as the percentage of the ratio of the weight difference of sample before and after drying to the weight of sample before drying. Protein content was determined using the modified Kjeldahl method based on AOAC^[23] with a protein-nitrogen conversion factor of 5.95 for RF.^[24] Amylose content was evaluated using a Megazyme amylose/amylopectin assay kit (Megazyme International, Ireland). All measurements were carried out in triplicate.

Swelling Power and Solubility

Swelling power and solubility were determined using a modified procedure of Lin and coworkers.^[25] Dry sample (0.1 g) in 40 mL of deionized water was heated at 120°C in an oil bath while constant stirring for 15 min. The mixture was then centrifuged at 5000 × g for 30 min at 25°C. The sediment was immediately weighed, and then the supernatant was evaporated at 100°C until a constant weight was obtained. Swelling power is calculated from the weight ratio of sediment to the initial weight of dry starch. Solubility is obtained from the weight ratio of dried supernatant to the initial weight of dry starch.^[25]

DSC

Ice melting enthalpy was determined from freezing and thawing experiments in a DSC (822e, Mettler Toledo, Schwerzenbach, Switzerland). Appropriate amounts of dry sample and deionized water were mixed directly in a 120 μL stainless steel medium pressure pan to give the systems with the desired water content. In case of low water content, excess water was initially added to the powder to provide the water absorption of all starch particles, and then excess water was subsequently allowed to evaporate until the desired water content was achieved. Water content (W_C) was defined as the ratio of mass of water to that of the dry sample.

The pan was sealed with a lid using an O-ring made of polytrifluorochloroethylene (Kel F, 3M) in order to prevent the leakage of water vapor during heating scan. To allow the complete water absorption of starch granules, the sealed pan was kept at 25°C overnight prior to DSC analysis. For DSC measurement, all samples and sealed empty pans as references were first heated from 25°C to the investigated temperatures (i.e., 60, 70, and 120°C) at heating rate of 10°C/min with a nitrogen flow rate of 60 cm³/min, then held at that temperature for 15 min before freezing to –80°C using either freezing rates of –10 or –100°C/min. Such large different freezing rates were used to elucidate an influence of number and size of ice crystals on non-freezable water content. Rapid freezing typically gives glassy ice or very tiny ice crystals. After that, the sample was equilibrated at –80°C for 60 min to ensure the complete formation of ice crystals and finally reheated to 120°C at a constant rate of 10°C/min.

During the first heating scan, all starch samples were completely gelatinized. In the freezing step, or cooling scan, exothermic peak of water crystallization was recorded. Finally, upon reheating (the second heating scan), exothermic peak of cold crystallization and endothermic ice melting peak were analyzed. The amount of non-freezable water (W_NFW, g/g dry sample) was determined from an extrapolation of a plot of the normalized ice melting enthalpy (ΔH_{m, ice}) by weight of dry sample against its corresponding W_C to zero water content.^[8,26]

Optical Microscopy

Flour or starch paste at 7% w/w dry basis was used as a representative of each system. The paste was prepared by heating aqueous suspension of dry sample to the desired temperatures of 60, 70, and 120°C using an oil bath with constant stirring. It was hold at that temperature for 15 min prior to cooling to room temperature (25°C). Then, the cooled paste was loaded onto a glass slide for microscopic examination (AXIOSKOP-POL, Ziess, Germany). Unstained samples were examined under cross polarized light to determine birefringence. Samples stained with Lugol’s iodine solution were examined to distinguish the regions of amylose and amylopectin via enhancing the contrast of a complex between iodine and helical structure of starch.

Statistical Analysis

Experimental results were subjected to statistical analyses using the commercial SPSS 11.5 (SPSS Inc., Chicago, IL) computer program. Data were averaged and mean comparisons were evaluated using the least significant difference (LSD) technique at 95% confidence. A statistical difference at p < 0.05 was considered to be significant.

Results and discussion

Composition of Flours and Starch

Table 1 shows physicochemical properties, i.e., moisture, amylose, and protein contents, of materials used. Moisture contents of RF, waxy rice flour (WRF) and TS were similar. The amylose content of RF (28.13 ± 0.6%) was much greater than those of TS (17.48 ± 0.55%) and WRF (4.12 ± 0.32%). The protein contents of RF and WRF were 6.30 ± 0.04% and 6.16 ± 0.03%, respectively, while none was detected in TS. In general, a small amount of protein (0.7–1.2%) in TS has been reported since it was removed during starch extraction.^[27] When comparing TS and PG, PG had a comparable amylose content (19.00% for PG and 17.50% for TS) to that of TS and its moisture content was lower than that of TS. The protein content of PG was not detected, consistent to that of TS.

TABLE 1 Physicochemical properties and swelling and solubility at 120°C of all studied flour and starch samples

Parameter	RF	WRF	TS	PG
Moisture (%)	7.08 ± 0.63^a	6.79 ± 0.55^a	6.58 ± 0.32^a	4.26 ± 0.20^b
Amylose (%)	28.13 ± 0.60^a	4.12 ± 0.32^d	17.48 ± 0.55^c	19.00 ± 0.35^b
Protein (%)	6.30 ± 0.04^a	6.16 ± 0.03^b	N/D	N/D
Swelling (g/g)	14.25 ± 0.36^b	23.16 ± 0.78^a	23.90 ± 3.33^a	14.71 ± 1.94^b,*
Solubility (%)	11.55 ± 0.26^d	25.01 ± 3.29^c	48.70 ± 4.24^b	63.47 ± 1.73^a

Amylose was measured using an amylose/amylopectin assay kit, Megazyme, Ireland. Protein was measured using Kjeldahl method. N/D: non-detectable.

^a,b,c,dmeans values followed by the same superscript letter within the same row are not significantly different (p ≥ 0.05).

*means the remaining starch granules in PG that could absorb large amount of water and then sediment upon centrifugation.

Effect of Freezing Rate and Starch Types on Ice Formation

Samples with total amount of water (W_C) from 0.3 to 2.3 g/g dry sample were initially heated from 25 to 120°C. DSC profiles revealed that the conclusion temperature of starch gelatinization peak was below 100°C, suggesting that all samples were fully gelatinized (data not shown). Influence of freezing rate and sample type on ice formation and melting was shown as follows.

Freezing rate of –100°C/min

DSC profile (Figs. 1a, 1c, and 1e) shows that during a very rapid freezing process (–100°C/min), exothermic peak of water freezing was not observed in all studied samples. This suggests that the water may undergo vitrification (or glassy ice formation) without any ice crystal formation during supercooling process.^[28] On the second heating, DSC profiles (Fig. 1b, 1d, and 1f) showed an exothermic peak (shown the onset by arrows in the figures) in all systems at low water contents (W_C ~ 0.40–0.67), and followed by a broad endothermic peak. The appearance of an exothermic peak during reheating indicated that glassy water could devitrify and subsequently crystallize, i.e., cold crystallization, from nucleation sites created in cooling stage. It has been suggested that devitrification is likely due to the increased polymer mobility when the temperature is suitable to enhance its molecular motion. This causes the decrease in viscosity of the rubbery matrix, thus reducing physical constraints and inducing motion of glassy water to form ice.^[21] Zhang and coworkers performed studies using infrared spectroscopic analysis and reported that this cold crystallization was attributed to the transition of glassy water to cubic ice.^[29] As W_C increased, the cold crystallization peak (T_{cc, water}) was unnoticeable which was similarly found in the previously investigated starch and flour systems.^[5,30,31] Hatakeyma and coworkers^[32] investigated cold crystallization of poly(ethylene glycol)-water systems and explained that with increasing W_C, molecules of the polymer which associated with glassy water could rearrange more freely, thus T_{cc, water} shifted to the low temperature side. Moreover, it should be noted that in these studied systems, there was no observation of a baseline change in the second heating scan which corresponded to a heat capacity change due to glass transition temperature.

FIGURE 1. DSC cooling (a, c, e) and the second heating (b, d, f) thermograms of rice flour (RF), waxy rice flour (WRF), and tapioca starch (TS) at various water contents.

The endothermic peak after cold crystallization represented the ice melting, with the onset temperature ≤0°C. Theoretically, the melting of freezable water (T_{m, ice}) was found below 0°C, whereas T_{m, ice} of free water was at about 0°C and slightly affected by the polymer matrix.^[5,30] In this study, the ice melting peak was observed around 0°C (~3°C at W_C = 0.40) and shifted to the higher temperature as W_C increased (~20°C at W_C = 2.33), where the onset temperature of all systems was below 0°C (see Figs. 1b, 1d, and 1f). The shift of the peak and the end temperatures of ice melting event to high temperature with increasing W_C indicated the increase in size of ice crystal.^[10]

As seen in Figs. 1b, 1d, and 1f, in addition to the main peak, the ice melting peak with a shoulder on the low temperature side was clearly observed in samples with W_C ~ 0.6–1.0, indicating the existence of different states of freezable water (freezable bound water and free water).^[5,30] However, when W_C increased, i.e., W_C > 1, ice melting peak became broader, resulting in the disappearance of the peak shoulder. Enthalpy of ice melting (ΔH_{m, ice}), determined by taking into account of the cold crystallization,^[8] increased as well with increasing W_C, especially at W_C ≥ 1, as observed from an increase in size of the ice melting peak. The progressive increase of ΔH_{m, ice} suggests that ice formation of excess (unbound) water existed during freezing period at –80°C (for 60 min). Hatakeyama and coworkers point out that peak temperature and the width of the ice melting peak depended on the matrix of biopolymer and the amount of co-existing water.^[30]

Our results indicated that although rapid freezing rate (–100°C/min) exhibited no ice formation during freezing, ice could be formed during frozen storage (60 min at –80°C) and thawing (due to cold crystallization). It has been reported that thawing could potentially deteriorate textural quality of frozen food.^[33] To clarify this matter, the experiment on devitrification and ice melting of TS system at W_C ~ 2.33 was performed with a shorter frozen storage (2 min) at the final freezing temperature. Similar to our previous observation, no ice crystal (water freezing exotherm) was produced during rapid freezing scan. Unlike the 60-min freezing period, the cold crystallization peak appeared and was noticeably large. Nonetheless, the amount of normalized ice melting enthalpy in both studies was in the same range (ΔH_{m, ice} ~ 280 and 270 J/g water for 2-min and 60-min holding periods, respectively). This means that although ice crystallization did not occur during the process of fast freezing, the presence of excess water could allow ice formation during freezing storage or during reheating (crystallization of glassy water). Therefore, the result herein confirms that the total amount of “frozen” ice was rather constant in all the studied systems and was independent on freezing condition.

Freezing rate of –10°C/min

The investigated samples were thermally scanned using the same heating procedures (for both 1st and 2nd scan) as previous experiment, except that freezing scan was carried out at a slower freezing rate of –10°C/min. At this freezing rate, the distinct exothermic peak of water freezing (T_{f, water}) was observed (Figs. 2a, 2c, and 2e), which indicated the process of ice crystal growth. However for TS samples at W_C ≤ 0.55, the T_{f, water} was not clearly revealed (Fig. 2e). The disappearance of T_{f, water} might be due to the amount of freezable water in the samples was small and remained in glassy state. Figures 2a, 2c, and 2e indicated that T_{f, water} was in the similar range (–20 to –30°C) irrespective to sample type and W_C, whereas water freezing enthalpy (ΔH_{f, water}) increased with increasing W_C. Upon reheating, T_{cc, water} and T_{m, ice} were detected (Figs. 2b, 2d, and 2f). The appearance of T_{cc, water} (its onset shown by an arrow in Fig. 2) in systems with cooling rate of –10°C/min implied that some parts of water still remained in a glassy state^[5] and dependent on W_C.^[34] When comparing ΔH_{f, water} + ΔH_{cc, water} (enthalpy of cold crystallization) and ΔH_{m, ice} at the corresponding W_C, it was found that ΔH_{m, ice} was always larger than ΔH_{f, water} + ΔH_{cc, water}, especially at high water content. This indicated an additional ice formation during frozen storage at –80°C with a 60-min holding period. Particularly in the excessive water systems, large amount of water was available to form ice crystals on the existing ones produced during slow freezing. Similar to the faster freezing rate experiment, T_{m, ice} shifted toward higher temperature and ΔH_{m, ice} increased with W_C.

FIGURE 2. DSC cooling (a, c, e) and the second heating (b, d, f) thermograms of rice flour (RF), waxy rice flour (WRF), and tapioca starch (TS) at various water contents.

Figure 3 shows a plot of normalized ice melting enthalpy by water weight against water content in samples frozen through both fast and slow freezing rate. No clear difference on the normalized ΔH_{m, ice} was observed among the examined samples for both freezing rates. However, the result showed that at W_C ~ 0.5–0.66, the difference in ice melting enthalpy among sample type was apparently noticeable. In this range, RF had the highest value. Since ice melting enthalpy contributed to the amount of frozen water (freezable and free water), the type of sample seemed not to obviously influence on the amount of freezable water when W_C > 0.66 due to the presence of excess water and when W_C < 0.4 due to the limited water content. Our finding was similar to the previous observation by Tran et al.^[10] in the sense that a transition between limited and excessive water content in both native and modified cassava starch systems was detected at 40% moisture content (wwb) or at W_C ~ 0.66. Consequently, in excessive water content, most water was available to form ice, whereas in the limited water content, water was mainly bound to starch molecules with a small frozen fraction.^[10]

FIGURE 3. Normailized ice melting enthalpy by weight of water as functions of Wc of RF- (square), WRF- (triangle), and TS (diamond)-water systems at the studied freezing rate of –100°C/min (filled symbol) and –10°C/min (unfilled symbol).

Types of starch

Types of starch on ice crystal formation was analyzed using the results from freezing step. In a fast freezing (–100°C/min) process, an exothermic peak of ice crystal growth was not observed in all studied systems since the rate was too fast to allow the water to form ice crystals. On the other hand, the peak appeared in the slow freezing process (–10°C/min) of almost all samples, except TS samples having limited water content (W_C ≤ 0.55, see Fig. 2e). No exothermic peak upon freezing of TS with limited water content, i.e., having W_C of 0.45 and 0.55, led to a noticeable exothermic cold crystallization peak upon reheating (see Fig. 2f). This observation indicated that TS had a strong water binding ability to bind water, thus prohibiting the formation of ice crystal when the water content was limited. This water binding ability might be related to chemical composition, molecular structure, and starch granule rigidity. Consequently, effect of those factors was identified and discussed in correlation to the formation of ice crystal in the latter section.

In this experiment, we can conclude that first, subjected to a slow freezing process of –10°C/min, TS systems having limited water content can impede the growth of ice crystal. Second, at W_C ~ 0.5–0.66 which was a transition between limited and excess water, RF apparently had higher ice melting enthalpy, and thus larger amount of frozen water than WRF and TS. And finally, after carefully freezing the food product using a fast freezing rate, there might still be a risk of the additional ice formation during prolonged-freezing time if the freezing temperature was above the glass transition temperature of the frozen product, and a risk of slow thawing process that cold crystallization of water can take place, resulting in deterioration of food texture quality.

Determination of Non-Freezable Water

The amount of non-freezable water W_NFW (g/g dry sample) was determined as the X-intercept through the extrapolation of a plot between the normalized ice melting enthalpy by weight of dry sample and the corresponding W_C to ΔH_{m, ice} = 0.^[8,26] This value represents the water binding ability of the studied samples. The W_NFW from all studied systems and freezing rates are shown in Table 2. It was found that W_NFW were insignificantly changed for the same sample, regardless of freezing rate condition (0.315 for RF, 0.375 for WRF, and 0.395 for TS and g/g dry sample).

TABLE 2 Non-freezable water weight content (W_NFW) of starch-water systems determined by the method described in Lui and Cowman^[26]

Sample	1st heating temperature (°C)	Freezing rate (°C/min)	WNFW (g/g dry starch)	R^2
RF	120	–100	0.32	0.9906
WRF	120	–100	0.37	0.9924
TS	120	–100	0.40	0.9958
RF	120	–10	0.31	0.9944
WRF	120	–10	0.38	0.9979
TS	120	–10	0.39	0.9978
TS	70	–100	0.28	0.9983
TS	60	–100	0.27	0.9999
PG	70	–100	0.39	0.9969

When comparing between rice samples, RF had lower W_NFW value than WRF (Table 2), since the non-freezable water is strongly bound to hydrophilic molecules. This observation indicated that RF had the lowest water binding ability. The measurement of swelling ability also revealed the smaller swelling power and solubility values of RF (Table 1). Tsai and coworkers^[35] suggested that higher swelling power and solubility of starch were the consequences of weaker granule rigidity. This indicated that although RF was heated above the gelatinization temperature, its swelling was rather restricted comparing to the WRF, thus implying the stronger and more rigid granule of RF. The availability of the binding sites of starch toward water molecules was proposed to depend on its structure (molecular arrangement, amorphous, and crystalline regions) and composition (amylose and amylopectin molecules).^[36] A previous study by Tananuwong and Reid^[21] showed that waxy starch sample had a higher value of W_NFW than that of the non-waxy one. Furthermore, as the starch granule is progressively disrupted, more starch components, especially amylopectin, become disentangled and expose their hydroxyl groups to water, thus increasing the starch–water interaction. Their study also pointed out a significant role of amylopectin in controlling the interactions between water and gelatinized starch. Moreover, the study of Hagenimana and Ding concludes that the presence of amylose restricts granular swelling.^[37] Therefore, it can be stated here that the lower amylose content in WRF as compared to RF lead to higher swelling power and solubility, which in turn results in higher amount of W_NFW, i.e., lower ice formation.

Fu et al.^[19] studied starch systems of different degrees of gelatinization and reported that the number of available water binding sites was different among differently treated samples, thus resulting in the varying amounts of non-freezable water content. Full gelatinization, i.e., with more severe disruption of starch granules, could expose more hydroxyl groups of starch to water, thus giving rise to higher value of W_NFW when compare to partially gelatinized or ungelatinized starch.^[19] In case of TS, the larger swelling and solubility of the starch can be attributed by less compact structure and weaker granular rigidity, allowing more solubilizing and leaching of polymers form the granule,^[35,38] thus the highest value of W_NFW, i.e., 0.395 g/g dry sample, was achieved.

As W_NFW content was strongly influenced by water binding capacity of starch systems which can bind and entrap water within its structure, the effect of granule characteristics was subsequently examined to evaluate the impact of different starch gel morphology, i.e., granular characteristics, on W_NFW content. Besides rice samples, TS samples heated to the different temperatures were investigated to determine the influence of different granular morphology on the non-freezable water content which results are shown and discussed in the following section.

Evidence of an Influence of Granular Morphology on Ice Formation and Non-Freezable Water

To investigate the effect of granular characteristics on ice formation and non-freezable water content, TSs at various water contents (W_C ~ 0.35 – 2.35) were heated to two different heating temperatures in the first heating scan; i.e., 60°C (below pasting temperature, i.e., representing ungelatinized starch sample) and 70°C (a pasting temperature according to Rapid Visco Analysis [data not shown], i.e., representing partially gelatinized starch sample). All subsequent procedures were identical to the process described as in the fast freezing rate experiment. Similarly, it was found that water crystallization was not observed during fast freezing (–100°C/min). Unlike the fully gelatinized TS sample (heating to 120°C), no cold crystallization peak was detected in partially gelatinized and ungelatinized TS samples upon reheating. Consistently, Tananuwong and Reid did not observe cold crystallization exotherm in partially gelatinized gels from their studied starches.^[21] This may be related to a low amount of freezable water that is bound to partially gelatinized and ungelatinized starch molecules frozen in glassy state. However, the ice melting peak was observed and its corresponding enthalpy was used to determine the amount of non-freezable water content (Table 2). The determined W_NFW contents of the systems heating below and at the pasting temperature were 0.27 and 0.28 g/g dry sample, respectively, which values were lower than that of the fully gelatinized TS system (W_NFW ~ 0.40 g/g dry sample). The greater fraction of W_NFW in the fully gelatinized TS system was attributed to the complete disruption of starch granules. Accordingly, weak associative bonds in the amorphous region of starch granule were disrupted, enabling increased hydration of starch molecules and exposure of more starch molecules to water.^[20,36]

Morphology of starch granules in all studied systems was examined using an optical microscope. With iodine staining, TS granules after heated to 60°C displayed blue color due to the complex formation of iodine and amylose (Fig. 4a). A large amount of TS granules remained intact, but some showed slight swelling. Only a few granules were found in the form of swollen granules. For the unstained TS granules, birefringence (Maltese cross) under cross polarized light was observed which suggests that parts of semicrystalline structure were still retained in the granules (shown as inset in Fig. 4a). Upon heating to 70°C, i.e., onset of gelatinization, Fig. 4b showed that all TS granules swelled extensively but were still intact and contained disrupted internal ordered structure resulting in the loss of birefringence (not shown). Furthermore, when TS was heated to 120°C (Fig. 4c), the granules mostly disrupted and only a few starch ghosts or fragments were observed. The micrograph showed a complete dispersion of leached starch molecules corresponding to large amount of starch solubility (see Table 1). In case of RF sample heated to 120°C, Fig. 4d demonstrates the aggregations of wrinkle granules. WRF exhibited completed granule disruption after heated to 120°C (Fig. 4e); and few granule fragments were observed. Atkin et al.^[39] reported that granule surface, which changed into envelopes and ghost afterward upon heating, comprised of amylopectin as a major component. The degradation of ghost at high temperature created ghost fragments of amylopectin which joined together and were stained purple brown color with iodine solution as observed in Fig. 4e. The findings shown in this study suggest that high value of W_NFW (or lower ice formation) appears to be connected with the complete dissolution morphology of starch molecule in water, whereas samples exhibiting incomplete gelatinization and the presence of intact swollen granule generally have low W_NFW value.

FIGURE 4. Optical micrographs of iodine-stained granules; tapioca starch (TS) granules after heated to 60, 70, and 120°C (a–c, birefringence shown in inset in a), rice flour (RF) and waxy rice flour (WRF) after heated to 120°C (d–e), and pregelatinized tapioca starch (PG) after heated to 70°C (f).

In order to confirm the influence of morphological state of starch dispersion on the amount of non-freezable water content, an additional experiment was performed on PG sample to compare with TS heated to 120°C. Typically, the process of pre-gelatinization causes substantial changes of granular starch resulting in the disruption of molecular orders within the starch granule. Therefore, a high value of granule solubility, i.e., 63.47 ± 1.73%, was obtained when heating PG at 120°C (see Table 1). Nevertheless, heating PG at 70°C gave starch suspension with similar morphology to TS heated to 120°C. (cf. Fig. 4f and Fig. 4c). The similar DSC experiment was thus performed on PG dispersion by adjusting the final temperature of the first heating step and the holding heating temperature to 70°C. The W_NFW was calculated to be 0.39 for PG heated to 70°C which was much larger than that of partially gelatinized TS heated to 70°C, and was similar to the amount of W_NFW in fully gelatinized TS heated to 120°C. The obtained result thus confirms the strong influence of granule morphology on water binding ability of starch molecules or the amount of non-freezable water.

Conclusions

This study examined the formation of ice and determined the amount of non-freezable water (W_NFW) of different systems, i.e., RF, WRF, and TS, as influenced by freezing rate (–100 and –10°C/min) and granular morphology based on freezing and thawing experiment by DSC technique. Through slow freezing scan (–10°C/min), ice crystal growth was detected as the presence of exothermic peak, but no ice formation was observed during fast freezing one (–100°C/min). Comparing both freezing rate experiments, the amount of ice melting enthalpy was relatively close in each sample owing to the additional formation of ice during holding period of freezing and the cold crystallization process during reheating. This led to the similar determined W_NFW content, irrespective to rate of freezing, in each sample. It is very interesting that systems having water content of ~ 0.50 – 0.66, i.e., in the regime between limited and excessive water, RF had the highest ice melting enthalpy, indicating high amount of freezable water. However, in limited and excess water conditions, type of sample did not clearly provide the difference in amount of freezable water. For systems with complete gelatinization (heated to 120°C), RF had the lowest W_NFW since it had the lowest swelling power and solubility upon heating, as a result of strong and rigid granule. Optical micrograph of RF (120°C) demonstrated the presence of many wrinkle RF granules which was believed to be responsible for its low W_NFW. On the other hand, TS and WRF showed the comparatively high value of W_NFW over RF due to the complete disruption of the granules. Moreover, optical micrograph exhibited the correlation between W_NFW and morphology of TS granules when heated to different temperatures. Ungelatinized and partially gelatinized TS samples with intact (swollen) granules yield low value of W_NFW. The obtained information can be beneficial in designing frozen foods based on RF and TS to reduce deterioration of food texture under freezing process, freezing storage, and reheating conditions, thus improving their overall quality.

Acknowledgments

The authors are grateful to the Tapioca Development Corp. Ltd. for the provision of pre-gelatinized TS.

Funding

This study was granted by the National Metal and Materials Technology Center (MT-B-55-POL-07-519-I).

Additional information

Funding

This study was granted by the National Metal and Materials Technology Center (MT-B-55-POL-07-519-I).