Rheology and Expansion of Starch-Water-CO₂ Mixtures with Controlled Gelatinization by Supercritical Fluid Extrusion

Abstract

A slit die rheometer fitted to a twin-screw extruder was used to measure the viscosity and expansion characteristics of starch-based melts plasticized with water and supercritical CO₂. Experiments were conducted for four intermediate moisture starch-water mixtures with varying degree of starch gelatinization (DG), achieved by blends of pregelatinized and native common corn starch. SC-CO₂ was shown to be an effective plasticizer for starch-water mixture at 0.45 g SC-CO₂/100 g sample, lowering the viscosity of the melt by an average of 14%. Apparent viscosity was measured for shear rate ranging from 100–200 s⁻¹. The viscosity reduction factors based on DG and SC-CO₂ indicated that the free volume added to starch-water mixtures with SC-CO₂ is an effective mechanism for viscosity reduction. In addition, the higher water affinity of pregelatinized starch increased the water induced plasticization effect in the mixtures. SC-CO₂ was a good blowing agent producing distinct expanded microcellular morphology for the starch-water mixtures with DG at or above 80%. The expansion and cellular characterization of these starch-based extrudates produced by SC-CO₂ injection may be governed by the viscosity-dependent parameters including gas retention capability, CO₂ diffusivity and the pressure drop rate.

Keywords:

• Extrusion
• Expansion
• Viscosity
• Starch melt
• Supercritical CO₂
• Gelatinization

INTRODUCTION

Generation of extrusion-based foams from starch melt has received considerable attention in recent years for both food and non-food applications. It typically involves low-moisture (13–20 g water/100 g sample) and high-temperature (130–170°C), short-time (HTST) cooking of starch to produce steam-induced expanded products.[1–3] The ability of starch polymer to form thermoplastic melt upon water plasticization as well as its abundant supply, cheap cost, and high biodegradability makes it a competitive alternative to polystyrene (PS) and other popular foaming polymers. Steam expanded starch foams of conventional process, however, have its limitations in that they have a coarse and very non-uniform cellular structure with cell sizes in the range of 1–3 mm.[4] Moreover, there is little or no control over cell size, density, and therefore the resultant mechanical properties of the end product.

Conventional plastic foam processing by extrusion involves melting of polymer resins and incorporation of blowing agents ranging from hydrocarbon, chlorofluorocarbon to environmental friendly nitrogen and carbon dioxide. Recognizing the potential dissolution and expansion capability of supercritical carbon dioxide (SC-CO₂) in starch melt, Rizvi and Mulvaney[5] proposed a novel patented extrusion technology combined with injection of SC-CO₂ as a blowing agent and/or an active carrier for production of highly expanded starch foams.[6,7] In this high pressure (> 7.38 MPa), low temperature (< 100°C) process termed supercritical fluid extrusion (SCFX), the conventional roles of water as a plasticizer as well as a blowing agent are decoupled. It obviates the need for steam expansion and high mechanical energy inputs by increasing the moisture content of the feed to reduce the viscous dissipation of heat. Hence, low product temperature becomes attainable, which offers major advantages over steam puffing such as incorporation of heat sensitive ingredients and actives.[8] A key criterion to achieve the good cellular expansion during the SCFX processing of starch-water-CO₂ mixtures is to maintain both low temperature (< 100°C) and high moisture content of the mixtures (35–45 g water/100 g sample), which are conditions necessary to attain high CO₂ solubility limit in the starch-water mixtures as the main driving force of expansion. In addition, the highly gelatinized matrix is required to achieve good gas retention capacity. It was found from our previous study,[9] by differential scanning calorimeter (DSC), that complete starch gelatinization was not needed to achieve gas retention capacity for cellular expansion during SCFX processing. This has brought forth the hypothesis that there exists a minimum required degree of gelatinization in the starch-based mixtures to produce good microcellular expansion. Such design of partial starch gelatinization not only has the ability to achieve various final product textures, but it also has its application in resolving product texture loss problem typically among the quick-cooking and microwavable food novelty such as noodles, pastas, snacks, and bread products.[10]

Numerous studies on the rheology of polymer/CO₂ systems have been previously evaluated. Gerhardt et al.[11] investigated the viscosity-shear rate relationships for polydimethylsiloxane (PDMS)/CO₂ mixtures up to 20 g CO₂/100 g sample at 50 and 80°C. Kwag et al.[12] studied the viscosity of PS melts containing CO₂ from 0 to 5 g/100 g sample up to 16 MPa and 150–175°C. Areerat et al.[13] studied the melt viscosity of low density polyethylene (LDPE)/CO₂ solutions up to 5 g CO₂/100 g sample and temperature in the range of 150–175°C. Maio et al.[14] examined the viscosity of polycaprolactone(PCL)/CO₂ system up to 10 g CO₂/100 g sample in the range of 70–85°C. In all, the rheology results concluded that the free volume added to the melt by dissolved CO₂ is the principal mechanism for viscosity reduction. Furthermore, the rheology of CO₂/polymer melt system has been modeled successfully in terms of classical viscoelastic scaling theory using a shift factor approach, reflecting both dilution effect and free volume increase effect contributed by the dissolved CO₂. The extent of viscosity reduction has been attributed to the degree of molecular mobility influenced by the additional free volume, reflected upon the proximity of melt temperature from the glass transition of the CO₂/polymer system at operating pressure and temperature condition.[11]

Many SCFX applications ranging from bread dough[15] to masa-based chips,[9] and protein-fortified snacks[6] have been investigated. In this article, we used blends of pregelatinized and native corn starches as the model system to determine the minimum degree of gelatinization (DG) required for good cellular expansion, and to characterize the rheology of starch-water-CO₂ mixture at SCFX operating pressure and temperature using an online slit-die rheometer system. The expansion characteristics were measured and correlated with the melt viscosity of varying concentrations of SC-CO₂ and the DG in the starch-water mixtures.

MATERIALS AND METHODS

Preparation of Materials

Native (C∗Gel) and pregelatinized (C∗Instant) corn starch (∼23% amylose) were obtained from Cerestar (Hammond, IN). Four starch-water mixtures containing a 60:40, 80:20, 90:10, and 100:0 ratio of pregelatinized to native starch were prepared by dry blending. Water was added during SCFX processing to achieve 42 g water/100 g extrudate. High purity (∼ 99%) CO₂ from Air Gas (Ithaca, New York) was used in all SCFX trials.

Extrusion and Rheometer System

The viscosities of the SC-CO₂/starch-water mixtures were measured with an online slit die rheometer attached to a co-rotating twin screw extruder (Wenger TX-57 Magnum, Sabetha, KS). This extruder with a L/D ratio of 28.5 was specially configured for the SCFX process and SC-CO₂ was injected into the barrel at L/D of 24. The screw configuration, pressure profile and temperature control zones are shown in Fig. 1. The slit die rheometer used in this study had length (L) = 325 mm, width (W) = 50.8 mm, and height (H) = 3.175 mm as illustrated in Fig. 2. This die dimension with (W/H) ratio of 16 and (L/H) ratio of 102 had the ability to produce fully developed flow and to minimize the wall slippage and entrance effects. Two precalibrated pressure transducers were inserted at two ports along the slit die rheometer at a distance of 170 and 300 mm from the die entrance. Immediately downstream 38 mm from the die entrance, a bypass line with a valve was included to vary the rate of main flow traveling across the slit die rheometer.

Figure 1 Typical screw configuration and corresponding pressure profile developed along the extruder barrel during SCFX processing.

Figure 2 Schematic of the slit die rheometer system attached to extruder outlet.

A SC-CO₂ generation and metering system with a PID control loop was used to stabilize and maintain the mass flow to the set point at high operating pressure range between 10 and 20 MPa and target temperature. By activating a back-pressure type regulator, this system can rapidly and smoothly correct for any changes in the extruder conditions without becoming unstable. Hence, the concentration of injected SC-CO₂ into the starch-water mixture can be ensured during viscosity measurement.

For each set of extrusion treatments, three to four different flow rates of extrudate were used by manipulating the bypass valve. Flow rates of the extrudate were measured by weighing material outflow during a 30 second time interval at steady extrusion conditions. Moisture of the extrudates was determined by oven drying method. Simultaneously, corresponding pressure drop data from the pressure transducers were recorded automatically every 10 seconds.

Experimental Procedures

Starch melt viscosity was measured for the starch-water mixtures with four different blends of pregelatinized and native corn starches with and without injected SC-CO₂. The temperature of the extruder zone was set respectively at 35, 45, 45, 30, and 25°C from zone 1 to 5 (Fig. 1). The product temperature at the exit was at 55°C. Screw speed was set at 180 rpm. Dry feed rate was set at 9.7 × 10⁻³ kg/s. The moisture contents of the extrudate melt were adjusted to 42% by adding water to the barrel at 5.82 × 10⁻³ kg/s (60% dry basis). Specific mechanical energy (SME) input of the extruder was 240 kJ/kg. SC-CO₂ injection was set at 6.79 × 10⁻⁵ kg/s (0.7% dry basis) for expanded samples. At this injection rate, the SC-CO₂ concentration is at 0.45 g SC-CO₂/100 g CO₂-starch-water mixture.

Experiments started with maximum flow rate of the extrudate while keeping the bypass flow rate to minimum. Next, the flow rate of extrudate was decreased stepwise by adjustment to the bypass valve. Total of 4 flow rates through the die were obtained for each starch blend formulation tested. The raw data of pressure readings and measured mass flow rates were collected in triplicates and used to calculate shear stress and shear rate. Data reported were the results of extrusion runs for these 4 starch blends with and without SC-CO₂ injection.

Extrudate Expansion and Microstructure Analysis

The expansion ratio (ER) was calculated as the cross-section area of the extrudate divided by the cross-section area of the circular shaping orifice attached at the end of the slit die. From each formulation, extrudate samples were dried at 75°C convection oven for 8 hours. Each ER value was an average of 30 measurements. A scanning electron microscope (Leica 440 SEM) was used for examining the cellular structure of extrudates and measuring the average cell size. Samples were cut into 5mm-long cylinders and mounted on aluminum stubs with double side conductive carbon tape, and sputter coated with palladium gold before mounting in the vacuum chamber of SEM for imaging. Image processing software was used to measure the average cell diameters from the SEM analysis. The cell density for each sample was calculated by:

(1)

where N_b = number of cells/cm³ of expanded extrudate; ρ_s = piece density of unexpanded extrudate (g/cm³); ρ_b = piece density of expanded extrudate (g/cm³); and, D_b = average cell diameter (cm). Polydispersity index (PDI) is a measure of the uniformity of the cell size distribution and was calculated by:

(2)

where is the weight average of cell sizes, and is the number average of cell sizes.

Apparent Viscosity Calculations

The shear stress at the walls of the slit die was determined from the slit height (H) and the pressure drop (ΔP) data across the first and second transducers separated by a distance of ΔL:

(3)

The apparent shear rate at the walls was calculated from the measured mass flow rate (kg/s) and density ρ (kg/m³) of the mixture estimated by their weight fraction contribution from the density of starch-water mixtures and of CO₂ at measured pressure range and 55°C using the following equation:

(4)

where W is the slit width. The rheological parameters, K’(consistency index) and n (flow behavior index) were calculated from linear regression of the log-log plot of shear stress versus apparent shear rate using the power law equation to obtain the best fitting values:

(5)

The actual shear rate at the wall was then calculated using n as follows:

(6)

Apparent viscosity of the mixture (η_app) was calculated as a function of the wall shear rate:

(7)

where the consistency index, K (Pa-s) was given as:

(8)

Viscosity Reduction Factors Calculation

To quantitatively determine the effects of added SC-CO₂ and native starch on the viscosity of a molten polymer-like starch-water mixture, a fundamental understanding of the plasticization is required. Several research groups have attempted to address the relationship of free volume and plasticization on the molten polymer rheology. The SC-CO₂/polymer mixtures are generally regarded as polymer diluted by a solvent, where entangled polymer chain dynamics prevail and where the polymer contribution to viscosity predominates over solvent contribution. Thus, the viscosities of the SC-CO₂/polymer mixtures are expected to be reduced, relative to pure molten polymer, by the dilution of the concentration of entangled polymer chains and by the increase in mixture free volume due to dissolved SC-CO₂. Han and Ma[16,17] reported that the viscosity curves for mixtures of polystyrene and polyethylene with fluorocarbon blowing agents can be reduced to a master curve by applying a shift factor to only the viscosity values, defined as the viscosity reduction factor (VRF):

(9)

Their data were taken in the shear-thinning power-law portions of the viscosity curves. For this work, two viscosity reduction factors: VRF_c and VRF_g to describe for the viscosity change due to SC-CO₂ concentration and DG were determined respectively. Hence, the viscosity reduction contributions would facilitate us in better understanding the high pressure rheology behavior of SC-CO₂ plasticized starch-water mixtures during SCFX process.

RESULTS AND DISCUSSION

Rheological Parameters

Table 1 shows the consistency index (K), and the flow behavior index (n) of starch-water mixtures determined in the shear rate range of 100 to 200 s⁻¹ using Eqs. (5) and (8). The data for all starch-water and starch-water-CO₂ mixtures show the consistency index, K, increased as DG increased from 60 to 100% level. In addition, the K values of starch-water mixtures were always higher than their respective SC-CO₂ plasticized mixtures.

Table 1 Consistency index and flow behavior index as a function of degree of gelatinization and SC-CO₂ addition for starch-water mixtures with 42% water content

	Consistency index, K(kPa.s^n)		Flow behavior index, n		R^2
Degree of gelatinization	0 %	0.45% SC-CO2	0 %	0.45% SC-CO2	0 %	0.45% SC-CO2
100%	6383	3347	0.21	0.31	0.93	0.94
90%	2409	1809	0.37	0.39	0.80	0.81
80%	2191	711	0.37	0.57	0.78	0.95
60%	961	250	0.46	0.71	0.82	0.78

This trend of K value is well supported by the work of Willet et al.,[18] in which the K values of thermoplastic starch melts were shown to decrease with increasing moisture content and addition of other plasticizer-like low molecular weight additives. The flow behavior index, n, increased with the addition of SC-CO₂ but decreased as DG increased from 60 to 100% level. For the starch-water mixtures, n decreases from 0.46 to 0.21, and for starch-water-CO₂ mixtures from 0.71 to 0.31. These results correspond to the works of Singh and Smith[[19] and of Willet et al.,20] in which the flow behavior index of thermoplastic wheat starch and pre-extruded corn starch increased with moisture content, temperature, and with the addition of plasticizer such as wheat germ oil.

These results are in line with the relationship between the flow behavior index and the polymer entanglement density in the mixture, in which a smaller n value reflects an increased polymer entanglement density.[19,21] In their work, Della Valle et al.[22] observed a decrease in the power law index with increasing amylose content and came to conclude that linear amylose would more likely contribute to entanglements whereas the highly branched amylopectin would not form much entanglements. In our study, the lowest n value was observed for the starch-water mixtures with 100% DG with no CO₂ added. Such result would suggest the mixtures with 100% DG to have the highest polymer entanglement density due to the least dilution effect contributed by either CO₂ molecules or native starch concentration in the mixtures.

Apparent Viscosity of Starch-Water Mixtures

Apparent viscosity of the starch-water mixture was measured with K and n values respective of each starch-water mixture and the corresponding shear rate using Eq. (7). The viscosity versus shear rate relationship for starch-water mixtures only and with 0.45 g SC-CO₂ /100 g sample added at 55°C are shown respectively, in Figs. 3 and 4. For the starch-water only mixtures, the viscosity shows a decrease as DG lowered from 100 to 60%. Such viscosity dilution induced by the lowering of DG can be attributed to the higher water affinity of pregelatinized starch over native starch fraction. At the same water content, the starch-water mixture with lower DG or lower concentration of pregelatinized starch would take more water away from the native starch, inducing a stronger water plasticization effect to produce a lower viscosity. Moreover, the viscosity of starch-water mixture in Fig. 4 shows to decrease consistently with addition of 0.45 g SC-CO₂/100 g sample, suggesting that the free volume added to starch-water mixtures with SC-CO₂ is an effective mechanism for viscosity reduction.

Figure 3 Viscosity of starch-water mixtures with starch of various degrees of gelatinization as a function of shear rate at 55°C.

Figure 4 Viscosity of starch-water-CO₂ mixtures with 0.45% wt CO₂ and starch with various degrees of gelatinization as a function of shear rate at 55°C.

From the viscosity data measured between 100–200 s⁻¹ in Figs. 3 and 4, the gelatinization and concentration dependent viscosity reduction factors, VRF_g and VRF_c, were calculated using Eq (9) and determined for the various combinations of starch blends and SC-CO₂ concentration at 55°C. VRF's showed to be slightly dependent on shear rate, yielding about 10% more reduction from 100 s⁻¹ to 200 s⁻¹.

Fig. 5 illustrates the VRF results with respect to DG levels and SC-CO₂ addition at the shear rate of 150 s⁻¹. For the starch-water mixtures with 0.45 g SC-CO₂ /100 g sample addition, the VRF_c has a mean value of 0.86, indicating that mixture with 0.45 g SC-CO₂/100 g sample added consistently achieves a 14% viscosity reduction. Moreover, the VRF_g without the effect of SC-CO₂ addition yields 17%, 24%, and 46% viscosity reduction for 90%, 80%, and 60% DG mixtures, respectively.

Figure 5 Variation of viscosity reduction factors with respect to SC-CO₂ and DG levels for starch-water mixtures at the shear rate of 150/s.

In comparison of the two diluents, SC-CO₂ appears to be a more effective diluent on a weight basis to reduce melt viscosity. In their rheology evaluation of starch-water mixtures containing native and pregelatinized wheat starch blends, Alavi et al.[23] indicated that native starch granules act as an inert diluent by increasing the free volume of the starch-water mixtures and decreasing the volume fraction of starch polymers in the starch-water system. Moreover, the higher water affinity of pregelatinized starch increased the water induced plasticization effect in the mixtures with higher native starch concentration. It is important to note that the mixture viscosity would play a key process control for the extrudate expansion and microstructure development during SCFX processing.

Effect of Viscosity on Extrudate Expansion

Fig. 6 illustrated the expansion ratio (ER) for the starch-water mixtures with respect to their DG level and SC-CO₂ injection. With 0.45 g SC-CO₂/100 g sample injection, ER of the extrudates increased with DG from 2.7 to 7.1 times the cross sectional area of the circular shaped die. The extrudates without SC-CO₂ injection only produced ER of 1.1 to 1.7 times, reflecting the sole contribution effect of die swelling based on the material property of the starch-water mixtures with 60 to 100% DG levels. Based on the ER results and expansion observation, the starch-water mixture with the lowest DG of 60% did not possess sufficient gelatinized gas holding capability and could not produce distinct expanded cellular morphology. Since all tests with SC-CO₂ addition had 0.45 g SC-CO₂/100 g sample, it would indicate that considerable amount of CO₂ had escaped from the extrudate for the starch-water mixtures with lower viscosity corresponding to the 60% DG samples. Moreover, the lower viscosity produced by the starch-water mixtures with 60% DG would also more likely promote product collapse due to its weaker structural integrity.

Figure 6 Expansion of starch-water mixtures with 0.45 wt% SC-CO₂ injection using SCFX system with die diameter of 4.50 mm.

In general, SC-CO₂ showed to be a good blowing agent at 0.45 g/100 g sample level where the starch-water mixtures with highest viscosity corresponding to 100% DG samples produced the maximum expansion. The extent of extrudate expansion is likely governed by the gas retention capability which correlates well with the measured apparent viscosity in the DG range examined. As expected, the higher pressure observed in the slit die rheometer was obtained by more viscous starch-water mixtures with higher DG levels. The formation of a less expanded product by the less viscous mixture was likely due to larger loss of CO₂ gas since the pressure near die may fall below the saturated solubility of injected SC-CO₂ to prematurely nucleate and more likely to escape from the extrudate surface. Hence, with prudent manipulation of formulation, SC-CO₂ injection rate, and other processing conditions, we can achieve the target expansion and desired texture for novel food product applications using SCFX process.

Effect of Viscosity on Extrudate Microstructure

Table 2 showed the effect of DG and SC-CO₂ injection on the expanded microstructure in terms of their average cell size, average cell density, and cell size distribution for the various starch-water mixtures with 60–100% DG levels. Based on the scanning electron microscopy (SEM) at 32 times magnification, the average cell size decreased from 410 to 310 μm as DG increased from 80 to 100% while average cell density increased an order of magnitude to 1.4 × 10⁵ cells/cm³. These SCFX extrudates demonstrated its uniformity in cell size distribution with the polydispersity index (PDI) values calculated between 0.81–0.84, much higher compared to the PDI value of 0.30 of typical steam puffed extrudates.[4]

Table 2 Microstructure characteristics of starch-water mixtures with 0.45 wt% SC-CO₂ injection

Degree of gelatinization (DG)	Average cell size (μm)	Average cell density (cells/cm^3)	Polydispersity index (PDI)
100%	310	1.40E + 05	0.84
90%	340	6.87E + 04	0.82
80%	410	2.93E + 04	0.81
60%	NA	NA	NA

Fig. 7 showed the expanded starch-based extrudate with 100% DG to produce microstructure with more smaller bubbles and higher cell density. This result was attributed to the lower gas diffusivity found in the more viscous starch-water mixtures corresponding to the higher DG level.[24] In addition, cell coalescence and ruptures were more frequently observed for the starch-water mixtures with lower DG levels, indicating that the lower mixture viscosity generated a weaker structural integrity unable to withstand the dynamics of adjacent growing nuclei. As expected, the smaller pressure drop rate observed by the starch-water mixtures with lower DG would experience a smaller thermodynamic instability to induce fewer nuclei formation. In all, the final expanded cellular microstructure would be governed by the viscosity-dependent CO₂ diffusivity into the nuclei and the competition between nuclei formation and bubble growth affected by the pressure drop rate experienced at the die. CO₂/polymer system with higher pressure drop rate and lower diffusion rate would favor more nucleation activity, while CO₂/polymer system with lower pressure drop rate and higher diffusion rate would favor more bubble growth activity. These interpretations are similar to those proposed by Goel and Beckman[25] about CO₂-induced microcellular formation.

Figure 7 Characteristic SEM of expanded starch-based extrudates with 0.45 wt% SC-CO₂ injection using SCFX system with die diameter of 4.50mm. SEM were taken at 32× magnification.

CONCLUSIONS

Apparent viscosity and expansion characteristics of starch-based melts plasticized with water and supercritical CO₂ were evaluated using a slit die rheometer fitted to a twin-screw extruder for four intermediate moisture starch-water mixtures with varying DG levels, achieved by blends of pregelatinized and native common corn starch. Apparent viscosity of the starch-water mixtures measured between 100–200 s⁻¹ was found to reduce by an average of 14% with the addition of 0.45 g SC-CO₂/100 g sample independent of the DG level in the mixtures. The viscosity reduction factors (VRF) determined based on SC-CO₂ and DG levels suggested that the free volume added to starch-water mixtures with SC-CO₂ effectively reduced viscosity, and that the higher water affinity of pregelatinized starch increased the water induced plasticization effect in the mixtures. Observing from the extrudate expansion, the results indicated that the starch-water mixtures would require DG at or above 80% to achieve good gas-holding rheology to produce nonporous skin and distinct cellular interior morphology. Moreover, expansion and cellular characterization of these starch-based extrudates can be well justified by the viscosity-dependent gas retention capability, CO₂ gas diffusivity in the starch-water mixtures, and the pressure drop rate experienced at the die. Using this knowledge of high pressure CO₂-plasticized rheology of the starch-water mixtures, we may then achieve better controlled expansion and greater flexibility of innovation by supercritical fluid extrusion (SCFX).

ACKNOWLEDGMENTS

The authors would like to acknowledge the technical assistance support from Esref Dogan, and the pilot scale extruder system provided by Wenger Manufacturing, Inc.