Rheology of thermoplastic starch: effects of temperature, moisture content, and additives on melt viscosity.
The use of starch in plastics has been an area of active study for some time. Among the factors driving this interest are the inherent biodegradability of starch, the potential of starch as a low cost raw material, the fact that starch is an annually renewable resource, which can replace nonrenewable petrochemical feedstocks, and the existence of a multibillion bushel surplus of corn, the major source of starch in the U. S.
Starch is constituted of (1-[greater than]4) linked [Alpha]-D-glucopyranosyl units. Two forms are found in natural starches, amylose and amylopectin. Amylose is a highly linear, lightly branched polymer with molecular weights typically in the range of several hundred thousand. Amylopectin is a highly branched polymer, with molecular weights typically in the millions. The branches are due to infrequent (1-[greater than]6) bonds in the polymer structure. Corn starch typically contains about 27% amylose and 73% amylopectin, although genotypes of nearly pure amylopectin or high amylose content exist. Starch exists in the form of granules; corn starch granules range in size from about 3 microns to 30 microns in diameter, with an average size of about 10 microns in diameter. Starch can be readily dispersed in water at temperatures of 60 [degrees] C to 80 [degrees] C (1).
Granular starch has been incorporated into polyolefins at levels of 6 to 15 wt% (2-6). Composites of starch and water-soluble polymers or water-dispersible polymers have been developed with starch contents on the order of 40% to 60% (7-9). Melt processable graft copolymers of starch and acrylate esters have been extensively studied (10-12). The starch content in these technologies has been limited primarily because of the decrease in mechanical properties to unacceptable levels as the starch content increases.
Recent developments in starch extrusion have led to the development of thermoplastic starch materials with starch contents of up to 95% to 98% (13-15). Under the temperature and pressure conditions found in extruders, the granule structure of starch is destroyed when sufficient moisture is present. The resulting material behaves as a thermoplastic, and extrudates are clear and glassy with properties similar to polystyrene and poly(methyl methacrylate).
Starch extrusion has been widely studied in the food processing industry because of the extensive use of extruders to cook or otherwise process starch (16). Several studies of starch viscosity and the effects of variables such as screw speed, moisture content, extrusion temperature, and additives have been reported (17-30). There is considerable variation in the results, reflecting the effects of different types of starches and varying degrees of starch modification during extrusion. Despite the variance in results, several trends are clear from these studies. First, considerable molecular weight degradation occurs during extrusion of starch. It has been shown by several groups that chain scission is primarily restricted to the amylopectin molecules owing to their large size relative to amylose (19, 20, 27). Amylose molecules apparently suffer little molecular weight degradation. Second, the physical structure of the starch is modified during extrusion. Disruption of the hydrogen bonding network in the granules leads to melting of the starch, and the shear forces lead to destruction of the granule structure. Finally, extrusion processing of starch results in a high degree of mixing of the amylose and amylopectin. Gels prepared from extruded starch exhibit behavior more similar to pure amylopectin than gels prepared from starch that has been gelatinized by other means (21, 28).
The effects of temperature, moisture content, and additives on the rheological behavior of thermoplastic starch are reported in this work. A single screw extruder with a capillary die was used for viscosity measurements. The shear rates were controlled by varying the screw speed. As a result, materials extruded at low shear rates experienced less specific mechanical energy (SME) and longer residence times than materials extruded at high shear rates. Accordingly, there may be material differences over the shear rate range because of the effects discussed above. Despite these differences, this technique allows the theological behavior of thermoplastic starch melts to be characterized under conditions similar to those experienced during industrial extrusion processes.
The corn starch used in this study was Buffalo 3401 from CPC International, an unmodified pearl starch. Urea (EM Science), lecithin (Fisher Scientific), triethylene glycol (TEG, Fisher Scientific), glycerol monostearate (GMS, Kemester 5500, Witco), and polyoxyethylene stearate (POES, ICI Americas) were all used as received without further purification.
The starch was first plasticized and pelletized. Rheology measurements were taken during a second extrusion step using the pellets as feed. The pelletizing step eliminates the problems associated with feeding moist powders, which bridge easily and can lead to surging. This procedure also allows for starch to be extruded at one moisture content and subsequently equilibrated to a different moisture content for rheology measurements. Since the equilibrium moisture content of starch increases with increasing relative humidity, the moisture content of the starch pellets was increased or decreased by conditioning in an environmental chamber with the appropriate relative humidity.
A Brabender PL 2000 torque rheometer was used for pelletizing and rheological measurements. Starch formulations were pelletized using a 19 mm diameter, 40/1L/D double mixing zone screw at 60 rpm. For all formulations except those with 30% moisture content, the temperature profile during pelletizing was 140/160/160/130 [degrees] C, measured from the feed section to the die. Strands from a die with 17 holes (1.6 mm diameter) were air cooled and pelletized. The die zone temperature was reduced to 100 [degrees] C when extruding formulations with 30% moisture to prevent bubble formation by steam expansion during cooling.
Rheological measurements were performed using a 19 mm diameter 25/1 L/D extruder with a 3/1 compression ratio screw. Three dies were used; all were 2 mm in diameter. Length/radius (L/R) ratios were 20/1, 30/1, and 40/1. All additive formulations were tested with a temperature profile of 150/155/160/160 [degrees] C; starch/water samples were also tested at temperatures from 110 [degrees] C to 180 [degrees] C. Screw speeds ranged from 1 to 60 rpm, and outputs ranged from 0.8 to 40 g/min. Die temperatures were measured with a thermocouple in melt contact prior to the capillary entrance; pressures were monitored with a Dynisco pressure transducer at the same position. Approximately ten output/pressure readings were taken for each die length.
Apparent shear rates were calculated from the relation
[Mathematical Expression Omitted]
where Q is the output in [cm.sup.3]/s and R is the capillary radius. Shear stress values were calculated using the following equation
[Tau] = [Delta] P/2(L/R) (2)
where [Delta] P is the pressure measured at the capillary entrance, L is the capillary length, and R is the radius, as in Eq 1. Shear rate values were corrected using the Rabinowitsch method:
[Mathematical Expression Omitted]
[Mathematical Expression Omitted]
Most thermoplastics exhibit power law behavior:
[Mathematical Expression Omitted]
where K is the consistency and m is the power law index. Substituting Eq 4 into the Newtonian relationship between [Tau] and [Mathematical Expression Omitted] yields the expression
[Mathematical Expression Omitted]
Values of m are between 0 and 1 for thermoplastics, When m [less than] 1, the viscosity decreases with increasing shear rate: such materials are called shear thinning. Values for the power law index m and the consistency K were obtained from the linear regression analysis of a double logarithmic plot of Eq 5.
The effect of moisture content (MC) on starch viscosity is illustrated in Figs. 1 through 3. In Fig. 1, increasing the MC from 15% to 30% at 160 [degrees] C leads to a reduction in viscosity by an order of magnitude for starch pelletized with 15% MC. A threefold reduction is seen for starch pelletized at 20% MC and tested with MC values of 20% and 30% at 120 [degrees] C. Figure 3 shows a similar magnitude of reduction in viscosity for starch pelletized at 30% MC and tested at 20% MC and 30% MC at 130 [degrees] C. The conditions were selected to provide a wide range of temperature and moisture content conditions during the viscosity measurement step.
A slight decrease in the power law index m is observed for each of these sets of curves. For the material pelletized at 20% MC, m decreases from 0.37 at a test MC of 20% to 0.34 at 30% MC. Likewise, a decrease from 0.42 to 0.37 is seen for the starch pelletized with 30% MC. These decreases are on the order of the standard error in the estimation of m by linear regression. A larger reduction in m, from 0.49 to 0.32, is observed for the starch pelletized at 15% MC when the test MC changes from 15% to 30% at 160 [degrees] C. It therefore appears that while increasing the MC at constant temperature significantly reduces the viscosity of the starch melt, there is little change in the non-Newtonian behavior until the MC is greater than 20%. The consistencies reflect the decrease in viscosity by increased MC.
The effect of increasing temperature on starch melt viscosity is shown in Figs. 4 through 6. The viscosity is significantly decreased as the test temperature is increased.
In general, m values show a greater dependence on temperature than moisture content. For samples pelletized at 20% MC and tested at 20% MC, m increased from 0.37 to 0.52 as the temperature increased from 120 [degrees] C to 150 [degrees] C. The effect of temperature on m diminished as the MC during pelletizing decreased or the measurement temperature increased to 160 [degrees] C or greater. Similar results were obtained for temperature/MC combinations not shown. These results are in general agreement with those of Vergnes and Villemaire (29), as well as Senouci and Smith (30), despite differences in processing methods. The values of Senouci and Smith were obtained using corn grits on a twin-screw extruder with a slit-die viscometer. The m values they report are significantly higher than those reported here and those of Ref. 29. This difference may be due to the use of corn grits instead of starch and the differences in processing. It is clear, however, that increasing the temperature results in an increase in the power law index m. This increase indicates that the starch melts become more Newtonian as the temperature is increased. No significant temperature effect on m was observed for the starch pelletized at 15% MC, regardless of test conditions.
Moisture content during the pelletizing step had a significant impact on melt viscosity, as shown in Figs. 7, 8, and 9. When tested with 30% MC at 110 [degrees] C, the starch pelletized with 15% MC had the lowest viscosity, starch pelletized with 20% MC had the highest, while that pelletized at 30% MC had an intermediate value. This same trend is observed at a test temperature of 130 [degrees] C, as shown in Fig. 8. At temperatures of 160 [degrees] C and 180 [degrees] C, moisture content during pelletizing had little effect on measured viscosities. As shown in Fig. 9, the viscosities at either temperature are independent of pelletizing conditions, within experimental error.
The effects of various additives on the melt viscosity of starch are shown in Figs. 10-13, At a level of 1 wt%, based on dry starch, lecithin and POES significantly reduce the viscosity. Both additives reduce the consistency by a factor of 3 to 4, while increasing the power law index m. The net result is that while the viscosities are reduced over the entire shear range studied here, POES and lecithin result in the melts being less shear thinning. Therefore, the viscosity differences are reduced at higher shear rates relative to the starch/water system under the same conditions. Formulations with higher levels of these two additives would not flow in the extruder owing to excessive slip caused by overlubrication. The scatter in the data for the 1% lecithin formulation is an indication that this formulation is close to overlubrication.
TEG and urea also reduce starch viscosity, although they are not as effective as lecithin and POES. When Figs. 10b and 11 are compared, it is seen that 5% of TEG is needed to reduce the viscosity by the same degree as 1% of POES. In addition, the lines for TEG are parallel to the starch line, indicating little change in m. Note also that the data scatter increases as the TEG content increases. This scatter reflects the fact that this system is approaching the point of overlubrication, and formulations with higher TEG levels could not be continuously extruded.
Urea behaves much like TEG, except that the consistency decrease is less as the level of urea is increased. It is expected that some of the urea is hydrolyzed to C[O.sub.2] and ammonia during processing and rheology measurements, lowering its effectiveness as a plasticizer. Samples of pellets with urea that were stored in sealed plastic bags had an odor of ammonia when opened, indicating hydrolysis. No significant change in m is observed in the formulations with urea compared to the starch extruded with an equal moisture content.
GMS produces a different type of behavior when added to starch, as shown in Fig. 13. The addition of 2% GMS reduces the consistency by less than 10%, while 5% GMS has essentially the same viscosity-shear rate curve as the starch. It is clear that GMS has a different type of effect on starch viscosity than the other additives studied here. As shown in Fig. 13, additive levels that cause large viscosity changes with lecithin, POES, and TEG cause little reduction and even a slight increase with GMS.
The consistency and power law index values for the starch/additives systems are compiled in Table 1.
Because of its thermally sensitive nature, additives such as water are necessary to extrude starch. Even in the presence of water and other low molecular weight additives, considerable physical and chemical changes may occur during extrusion. An understanding of how various additives and processing conditions affect the rheology of thermoplastic starch is needed in order to convert starch into materials with useful properties.
Table 1. Effects of Additives on Thermoplastic Starch Melt Viscosity.
Additive Wt % Consistency (Pa * [s.sup.m]) m
Control - 38,900 0.43 TEG 2 21,400 0.48 TEG 5 14,100 0.48 Urea 2 25,700 0.45 Urea 5 26,900 0.41 GMS 2 36,300 0.40 GMS 5 50,100 0.39 Lecithin 1 8300 0.67 POES 1 11,220 0.56
Moisture content exerts a strong influence on starch rheology. As shown in Figs. 1 through 3, melt viscosity is decreased as moisture content is increased. This behavior is observed regardless of the processing conditions or test conditions. It has been shown that the viscosities of starch melts (15, 29, 30) and flour doughs (31, 32) exhibit an exponential dependence on moisture content:
[Mathematical Expression Omitted]
A semilogarithmic plot of consistency (viscosity at [Mathematical Expression Omitted]) versus MC at 160 [degrees] C for starch pelletized at 15% MC is shown in Fig. 14. Linear regression yields a value for [Alpha] of 12.6, which is in good agreement with the values reported in other studies (15, 29, 30) but is somewhat higher than that of Ref. 30. Analysis of the temperature dependence of the melt viscosity (see below) also allows calculation of values. For starch pelletized at 20% MC and tested at 120 [degrees] C, 140 [degrees] C, and 160 [degrees] C, a value of 11.5 is calculated. This value agrees well with the value determined above.
The reduction in viscosity follows the Arrhenius-type dependence of the consistency K on temperature, which has been observed in other studies of starch rheology (29):
K(T) = K([T.sub.0])exp([E.sub.a]/RT) (7)
Figure 15 is a semilogarithmic plot of the data for starch pelletized at 20% MC. The slopes of the lines yield reduced activation energies ([E.sub.a]/R) of 9950, 8500, and 8500 Kelvins for test MCs of 15%, 20%, and 30%, respectively. These values are about twice as high as those measured by other workers (29-31) but are lower than that reported by Tomka for potato starch (15).
Processing conditions also affect the melt viscosity of thermoplastic starch. As shown in Figs. 7 and 8, the consistency factors for melts tested at 30% MC show a minimum for starch pelletized at 15% MC and a maximum when pelletized at 20% MC. The consistency for the starch pelletized at 30% MC lies between the other two. It has been shown for a wide variety of starches and extrusion conditions that increasing the moisture content at constant extrusion temperature and screw speed reduces the amount of molecular weight degradation of starch (19-21, 25, 33). Since increasing moisture content reduces the melt viscosity, it is expected that the starch pelletized at 15% MC would undergo higher shear stresses owing to its higher viscosity. These larger stresses would induce more chain scission, resulting in a material with a lower viscosity, which is in fact observed. The fact that the starch pelletized with 30% MC has a lower melt viscosity than that pelletized with 20% MC may reflect water catalyzed hydrolysis during pelletizing. It appears that an optimal moisture level exists somewhere between 20% and 30%, which would result in maximum melt viscosity, an important consideration when trying to extrude starch for use as a thermoplastic material.
Further evidence of the effect of processing on melt viscosity is illustrated by Fig. 9. The melt viscosity is independent of processing moisture content when measured at 160 [degrees] C and 180 [degrees] C. These results suggest that significant molecular weight degradation can occur during the testing phase at these temperatures, to the point that any differences in starting materials are eliminated, The temperature effect on melt viscosity may be a combination of thermal activation of melt flow and molecular weight degradation, depending on extrusion conditions.
Low molecular weight additives generally lower the melt viscosity of thermoplastic starch. Lecithin and POES are the most effective at lowering the viscosity of the additives used in this study. At a 1% level, both of these additives had the lowest consistencies measured in this study. In addition, melts with these additives had the highest values of the power law index m, indicating more Newtonian behavior. A TEG level of 5% was needed to reduce the viscosity as much as 1% of POES. Urea was also effective at lowering the viscosity, but its effects are diminished with increasing levels, probably because of hydrolysis to water and C[O.sub.2].
The additives discussed above were all more efficient at reducing melt viscosity than additional increased in MC. Linear regression analysis of the data for TEG yields an [Alpha] value of 19.8, which is significantly higher than that measured for water (see above). Values for lecithin and POES are expected to be even greater since they reduce viscosity at a 1% level as effectively as 2% to 5% of TEG.
GMS affected starch viscosity in a different manner than the other additives studied here. A level of 2% resulted in only a slight lowering of viscosity, while 5% GMS had little effect relative to the pure starch/water system. Similar behavior has been observed by Cervone and Harper with sodium stearoyl lactate in doughs with moisture contents from 22% to 31% at 90 [degrees] C (31). This temperature is below the expected melting range of amylose/lipid complexes at these MCs (34), and Cervone and Harper suggest complex formation may be the cause of the viscosity increase they observed.
DSC analysis of extrudates with lecithin, POES, and GMS at 15% MC (data not shown) exhibit weak, broad amylose/lipid complex melting endotherms followed by amylose recrystallization exotherms. The peak temperature of the exotherms increases in the order lecithin [less than] POES [less than] GMS. The peak temperature for the 1% lecithin complex is approximately 150 [degrees] C, that of the 1% POES complex is about 175 [degrees] C, while the peak for the 2% GMS is above 200 [degrees] C. If one assumes the amylose cannot recrystallize until the amylose/lipid complex melts, the exotherm peaks can be taken to be an indication of the complex stability. If the amylose/GMS complexes do not melt at the test temperature of 160 [degrees] C, one would expect the rigid nature of the helical complex to contribute to the high viscosity measured relative to the other additives.
It has been shown that the addition of additives such as lecithin and mixtures of stearic and palmitic acid monoglycerides reduce the amount of molecular weight degradation during the twin-screw extrusion of starch (19). Intrinsic viscosities of starch extruded with 2% of either monoglycerides or lecithin were approximately double the intrinsic viscosity of starch extruded with only water. Because of this fact, one would expect starch/GMS extrudates to exhibit melt viscosities similar to those of starch/lecithin blends; the starch/lecithin blend has in fact a much lower viscosity than starch/GMS. This fact supports the interpretation that unmelted amylose/GMS complexes are the source of the high viscosities seen with this system.
The melt viscosity of thermoplastic starch is compared to a low density polyethylene at 160 [degrees] C in Fig. 16. The LDPE is Petrothene 3404B from Quantum Chemical, and has a melt index of 1.8 (g/10 min) at 190 [degrees] C. Its viscosity is intermediate between that of starch with 15% MC and 20% MC. The LDPE is more shear thinning (non-Newtonian) than starch over the shear rates illustrated, evidenced by its greater slope and lower m value. This figure illustrates the ability to formulate thermoplastic starches with viscosities and shear thinning characteristics similar to those of commercially available thermoplastics.
The melt viscosity of thermoplastic starch has been studied as a function of moisture content, processing temperature, and various additives. Starch melt viscosity exhibits power law behavior. Temperature had a greater effect on the power law index m than moisture content; m increased with increasing temperature, indicative of more Newtonian behavior. Moisture content during pelletizing affected melt viscosity, possibly by modifying the extent of molecular weight degradation during extrusion. Of the three moisture contents examined, starch extruded with 20% MC had the highest melt viscosity, while pelletizing at 15% MC gave the lowest viscosity. These differences were diminished at higher test temperatures. Melt viscosity exhibited an exponential dependence on moisture content. All the additives except GMS were effective in lowering the melt viscosity, with lecithin and POES exhibiting the greatest efficiency. Relative to melts with 15% MC, all the additives except GMS were more effective than additional water at reducing viscosity, as measured by [Alpha] values. Melt viscosities with GMS were essentially the same as, or slightly higher than, those of starch/water. This behavior may be due to the presence of unmelted amylose/GMS helical complexes in the melt. Thermoplastic starch formulations exhibit melt viscosity characteristics similar to LDPE with a melt index of 1.8.
The authors wish to express their gratitude to R. P. Westhoff, R. Haig, and G. D. Grose for assistance in extrusion and rheology measurements, and to Dr. R. L. Shogren for assistance with DCS analysis and helpful discussions of thermal stability of starch/lipid complexes.
[E.sub.a] = Activation energy for melt flow (Joules/mole).
K = Consistency (Pa * [s.sup.m]).
L = Capillary length (cm).
MC = Moisture content (wt%).
m = Power law index.
[delta difference]P = Pressure at capillary entrance (Pa).
Q = Volumetric output ([cm.sup.3]/s).
R = Gas constant (Joules/mole * [degrees] C).
T = Absolute temperature (Kelvin).
[T.sub.0] = Reference temperature (Kelvin).
[Alpha] = Moisture content proportionality constant (1/percent).
[Mathematical Expression Omitted] = Shear rate ([s.sup.-1])
[Eta] = Melt viscosity (Pa * s).
[Tau] = Shear stress (Pa).
1. Starch Chemistry and Technology, R. L. Whistler, J. N. BeMiller, and E.G. Paschall, eds., Academic Press, Orlando, Fla. (1984).
2. G. J. L. Griffin, U.S. Patent 4,021,388 (1977).
3. G. J. L. Griffin, U.S. Patent 4,125,495 (1978).
4. G. J. L. Griffin, U.S. Patent 4,983,651 (1991).
5. J. L. Willett, U.S. Patent 5,087,650 (1992).
6. R. L. Evangelista, Z. L. Nikolov, W. Sung, J.-L. Jane, and R. J. Gelina, Ind. Eng. Chem. Res., 30, 1841 (1991).
7. R. P. Westhoff, W. F. Kwolek, and F. H. Otey, Starch, 31, 163 (1979).
8. F. H. Otey, R. P. Westhoff, and C. R. Russell, Ind. Eng. Chem. Prod. Res. Dev., 16, 305 (1977).
9. F. H. Otey, R. P. Westhoff, and W. M, Doane, Ind. Eng. Chem. Prod. Res. Dev., 19, 592 (1980).
10. E. B. Bagley and G. F. Fanta, "Starch Graft Copolymers," in Encyclopedia of Polymer Science and Technology, H. F. Mark and N. M. Bikales, eds., John Wiley and Sons, New York (1977).
11. G. F. Fanta, "Synthesis of Graft and Block Copolymers of Starch," in Block and Graft Copolymerization, R. J. Ceresa, ed., Wiley Interscience, London (1976).
12. E. B. Bagley, G. F. Fanta, R. C. Burr, W. M. Doane, and C. R. Russell, Polym. Eng. Sci., 17, 311 (1977).
13. W. Wiedmann and E. Strobel, Starch 43, 138 (1991).
14. G. Lay et al., U.S. Patent 5,095,054 (1992).
15. I. Tomka, "Thermoplastic Starch," in Water Relationships in Food, H. Levine and L. Slade, eds., Plenum Press, New York (1991).
16. Extrusion Cooking, C. Mercier, P. Linko, and J. M. Harper, eds., Am. Assoc. of Cereal Chem., St. Paul, Minn. (1989).
17. R. A. Anderson, H. F. Conway, V. F. Pfeifer, and E. L. Griffin, Cereal Sci. Today, 14, 4 (1969).
18. C. Mercier, R. Charbonniere, J. Grebaut, and F. Gueriviere, Cereal Chem., 57, 4 (1980).
19. P. Colonna and C. Mercier, Carbohy, Polym., 3, 87 (1983).
20. V. J. Davidson, D. Paton, L. L. Diosady, and G. Larocque, J. Food Sci., 49, 453 (1984).
21. P. Colonna, J. L. Doublier, J. P. Melcion, F. Monredon, and C. Mercier, Cereal Chem., 61, 538 (1984).
22. J. Owusu-Ansah, F. R. Voort, and D. W. Stanley, Cereal Chem., 60, 319 (1983).
23. M. H. Gomez and J. M. Aguilera, J. Food Sci., 48, 378 (1983).
24. P. Colonna, J. P. Melcion, B. Vergnes, and C. Mercier, J. Cereal Sci., 1, 115 (1983).
25. R. Chinnaswamy and M. A. Hanna, Cereal Chem., 67(5), 490 (1990).
26. M. M. Erdemir, R. H. Edwards, and K. L. McCarthy, Food Sci. Technol. 25, 502 (1992).
27. V. J. Davidson, D. Paton, L. L. Diosady, and L. J. Rubin, J. Food Sci., 49, 1154 (1984).
28. C. Metres, P. Colonna, and A. Buleon, J. Cereal Sci., 7, 123 (1988).
29. B. Vergnes and J. P. Villemaire, Rheol. Acta, 26, 570 (1987).
30. A. Senouci and A. C. Smith, Rheol. Acta, 27, 546 (1988).
31. N. W. Cervone and J. M. Harper, J. Food Proc. Eng., 2, 83 (1978).
32. N. Sharma, M. A. Hanna, and Y. R. Chen, Cereal Chem., 70(1), 59 (1993).
33. L. L. Diosady, D. Paton, N. Rosen. L. J. Rubin, and C. Athanassoulias, J. Food Sci., 50, 1697 (1985).
34. C. G. Biliaderis, C. M. Page, L. Slade, and R. R. Sirett, Carbohy. Polym., 5. 367 (1985).