# On Mechanical Properties of Sago Starch/Poly([varepsilon]-caprolactone) Composites.

H. W. KAMMER [*]Four types of saga starch were incorporated into a poly([varepsilon]-caprolactone) (PCL) matrix, native, predried, thermoplastic starch (TPS) granules and TPS. All systems were found to be phase-separated. Tensile properties were obtained after formulation of various mixtures and processing of suitable test specimens. It was found that elongation at break of composites comprising native starch and thermoplastic starch decreases almost linearly with volume fraction of starch whereas tendencies to nonlinear dependencies were observed for predried and thermoplastic starch granules. Except for composites containing native starch, tensile strength was found to decrease linearly with volume fraction of starch. However, statistical analysis of the corresponding regression coefficients shows that the coefficients ruling the compostion dependence of tensile properties are not significantly different for the four starch types. One may conclude that in all cases, tensile properties decrease almost linearly with volume fraction and maximum volume fraction of starch loading is around 0.6. Scanning electron micrographs of fracture surfaces revealed weak interfacial adhesion between sago starch and the polymer matrix.

INTRODUCTION

Contemporary methods aimed at reducing the overall cost of biodegradable plastics have led to the use of biologically inert materials, such as starches, as fillers for commercially available synthetic biodegradable plastics, e.g. poly([varepsilon]-caprolactone) (PCL) [1-3]. The method involves dispersion of starch particles into a molten polymer matrix, within the barrel of an extruder, an equipment that offers the advantages of intimate mixing and fine dispersion of the starch particles. The resulting two-phase structure is referred to as starch/thermoplastic composites [2-4].

Starch/PCL systems offer an interesting pair due to the ability of PCL to accommodate up to 60% by weight of starch. The possibility of formulating a transparent, resilient and mechanically stable biodegradable blend of thermoplastic starch and PCL is also a challenging task. Sago starch was adopted for this study due to the fact that it is obtained from a palm that gives a higher yield than the traditional starch sources and it also has a slower biodegradation rate [5, 6]. Previous attempts aimed at incorporating starch into PCL matrix include the use of granular starch [2], thermoplastic starch [7] as well as crosslinking between starch and PCL [8]. It was found that crosslinking of the composites improves their thermal stability but reduces modulus of elasticity and tensile strength. Moreover, graft copolymers of starch and poly([varepsilon]-caprolactone) (starch-g-PCL) were successfully used as compatibilizing agents in starch/PCL blends [9, 10]. Addition of starch-g-PCL reduced again modulus and tensile strength whereas an increase of elongation and toughness was observed in the blends. However, there are only a few comparative studies on the effect of the physical condition of starch on the resultant mechanical properties of the materials. In this study, four types of starches were incorporated into the PCL matrix: native starch, predried starch, thermoplastic starch (TPS) granules and thermoplastic starch. Especially promising are the thermoplastic granules, since one expects that TPS undergoes deformation in a manner similar to plastics when subjected to stresses, unlike the brittle granules which are incapable of plastic deformation under imposed stresses.

The composition dependence of mechanical properties is discussed in terms of a slightly generalized version of Nielsen's micromechanical model [11] that gives the following dependence of relative elongation of the composite, [[varepsilon].sub.c]/[[varepsilon].sub.o] on volume fraction [Phi]

[[varepsilon].sub.c] = [[varepsilon].sub.o] (1 - [[Phi].sup.1/3]) (1)

Validity of this equation has been tested by several authors for various filler/plastic systems [3, 12-14]. However, applicability of this model to the systems under consideration suffers from two defects. First, the model was proposed for spherically shaped fillers (glass beads) with strong adhesion to the matrix. Second, the fillers are assumed to be uniform in size. These conditions are not met in starch/PCL composites. Sago starch like most starches has a broad size range with an average particle diameter of 35 [mu]m [6].

MATERIALS AND METHODS

Materials

Poly([varepsilon]-caprolactone), Tone P-787 ([M.sub.w] = 80,000 g/mol), was supplied by Union Carbide, Asia Pacific Inc. Sago starch, containing 29% amylose (determined by GPC after debranching the amylopectin), was provided by the Government of Sarawak State, Malaysia. Glycerol was obtained from BDH chemicals. Native and predried starch were used in powdered form. All other reagents and chemicals were analaR grade.

Preparation of Thermoplastic Sago Starch

A standard mixture of sago starch and glycerol (ratio 4/1) was prepared by use of a kitchen aid mixer (15 min on setting No 2). The starch-glycerol mixture was then blended by hand with urea (ratio 5/1). Water was present at 15% content of overall solids. Fifty grams of this mixture were charged into the mixing chamber of the Brabender plasticorder model PLE 331, set at 80[degrees]C and rotor speed of 40 rpm for 20 min. Care was taken to minimize moisture loss during the compounding process by keeping the vents tightly locked (air tight). Thermoplastic starch is obtained at the end of the compounding process. After production, the TPS was cryogenically ground using liquid nitrogen and employing a Retsch hammer cutter mill. The powder was then filtered with a 100 [mu]m sieve, to obtain a fairly uniform aspect ratio of TPS granules (average granule diameter of 40 [mu]m), and allowed to equilibrate at 60% RH to regain its initial moisture level. Sago starch was dried to [less than] 1% moisture level using a vac uum oven at 60[degrees]C, 30 kPa, for 72 h. The dried starch was stored over phosphorus pentoxide in a desiccator until required. Moisture content of the materials was determined by weight difference before and after vacuum drying. Thermoplastic starch granules differ from thermoplastic starch in that TPS granules form a powder whereas TPS is non-powderized starch.

Preparation of Composites and Test Specimens

Compounding was done by means of a Brabender Plasticorder model PLE 331 internal mixer equipped with a double mixing cam. It was carried out at 90[degrees]C and 40 rpm rotor speed. PCL was charged into the mixing chamber and allowed to melt, followed by gradual addition of starch granules, allowing for sufficient time for the starch particles to be worked into the matrix, and mixing was allowed to continue for 20 min. PCL/starch composites as well as the thermoplastic sago starch/PCL blend were mixed to give a mixing ratio of 0% to 60% by weight of starch including the water content of starch. Effective mixing was determined by a stable torque-time curve, at which stage, blending was stopped. The mixtures were sheeted out on a two-roll mill after which they were compression molded into 1.0 mm sheets at 90[degrees]C and 10 MPa, using a Kao GO Tech compression-molding machine. Dumbbell samples were cut from the prepared sheets according to American Standard Test Methods ASTM D 412-68. Tensile measurements were carried out at room temperature using a Monsanto T10 tensiometer, at initial grip distance of 50 mm and a test speed of 10 mm/mm. The results are an average of 6 determinations. Morphological studies were conducted using a Leica Cambridge S-360 model Scanning Electron Microscope (SEM).

RESULTS AND DISCUSSION

Figure 1 shows typical stress-strain tensile curves of PCL and selected PCL/Starch composites. The initial part can be seen as linear. The yield is always present; however, it decreases with increasing starch content. Plastic behavior is observed beyond the yield point. Moreover, tensile strength, strain as well as toughness decrease with increasing starch content. Quantitatively, results of tensile measurements are listed in Table 1. Mechanical properties of TPS/PCL blends were included up to 80% TPS content. Reported are the means and standard deviations of six determinations. For the starch composites, it became physically impossible to exceed 60% starch loading. It was only possible to exceed this limit of starch inclusion for the deformable TPS blends. Fritz et al., who demonstrated the presence of dispersed PCL droplets Within a TPS matrix, had reported this level of TPS blended with PCL [1].

As can be seen, elongation at break and tensile strength of PCL are higher than that of the composites. The effect of fillers should depend chiefly on the surface area of the fillers. If [N.sup.3] spherical particles of radius r are dispersed in the unit cube, the cross section of the continuous phase reads [15-17]

1 - A [cong] [(Nr).sup.2] (2)

and the volume fraction of fillers is [Phi] [cong] [(Nr).sup.3]. Substituting this in Eq 2, one arrives at

A = 1 - [[lgroup][frac{[Phi]}{[[Phi].sub.m]}][rgroup].sup.2/3] (3)

where [[Phi].sub.m] denotes the maximum packing factor. It is well known that the maximum packing volume fraction for uniform spheres in a close-packed hexagonal lattice is [[Phi].sub.m] = 0.74 and that random close packing corresponds to [[Phi].sub.m] = 0.637. Since in our PCL/starch composites the fillers are neither of uniform size nor of uniform shape, we replace the exponent in Eq 3 by a general exponent x that has to be determined experimentally. It follows

[[varepsilon].sub.c] = [[varepsilon].sub.o] [1 - [[lgroup][frac{[Phi]}{[[Phi].sub.m])}][rgroup].sup.x]] (4)

where the symbols have the following meaning:

[[varepsilon].sub.c]-elongation at break of the composite, [[varepsilon].sub.o]-elongation at break of PCL,

[[Phi].sub.m]-maximum packing volume fraction of starch, [Phi]-volume fraction of starch.

The decrease of elongation at break and tensile strength with ascending starch content are discussed in terms of Eq 4. The exponent accounts for the peculiarities of sago starch granules, i.e., granule shape, size and its distribution, adhesion between starch and PCL and interaction between starch particles. This exponent should be equal to 2/3 for uniform spherical fillers with strong adhesion to the matrix. However, for the irregular starch fillers, one expects different exponents. Volume fraction [[Phi].sub.m] reflects the random close-packed arrangement of the starch fillers. Weight fractions were converted to volume fractions by employing the relation

[[Phi].sub.j] = [frac{[w.sub.j]/[d.sub.j]}{[[Sigma].sub.i] [w.sub.i]/[d.sub.i]}]

where [w.sub.i] and [d.sub.i] stand for weight and the density of each component in the composite. The average densities used were 1.4 g/[cm.sup.3] for sago starch, and 1.14 g/[cm.sup.3] for PCL.

Double logarithmic plots of (1-[[varepsilon].sub.c]/[[varepsilon].sub.o]) versus volume fraction [Phi] of starch are shown in Fig. 2 for two composites: native sago starch/PCL and granular thermoplastic sago starch/PCL. As can be seen, the experimental results fairly fit Eq 4. The curves are least squares fits obtained by averaging all six sets of data at each starch content. Errors of quantity (1 - [[varepsilon].sub.c]/[[varepsilon].sub.o]) were calulated using the standard deviations given in Table 1. It turned out that within the limits of experimental accuracy, the curves superimpose for predried sago starch/PCL and for granular thermoplastic sago starch/PCL as well as those for native sago starch/PCL and thermoplastic sago starch/PCL composites. The parameters x and [[Phi].sub.m], as estimated from Fig. 2, for the starch/PCL mixtures are summarized in Table 2a. The regression parameters were subjected to a variance analysis at a 95% confidence interval. Results are also indicated in Table 2. It becomes o bvious that parameters x and [[Phi].sub.m] are not significantly different for the starch types studied. Values for maximum volume fraction [[Phi].sub.m] are around 0.6 and exponent x changes between 1 and 2. It is interesting to note that native and thermoplastic starch blends display almost a linear decrease of elongation at break with volume fraction [Phi]. It means that elonga tion at break is proportional to the volume of the dispersed phase in the composite. Values of exponent x greater than unit reflect agglomeration of the filler particles. Figure 3 illustrates the linear decrease of elongation with increasing volume fraction for native starch. It can be seen that significant decrease in elongation at break occurs already at 20 wt% of starch content. With respect to elongational behavior, native starch and TPS in blends with PCL display the same almost linear decrease with increasing volume fraction of starch, whereas predried starch and TPS granules tend to nonlinear decrease. However, differences are not significant.

In an analogous way, one may discuss the tensile strength, [sigma], by replacing [varepsilon] by [sigma] in Eq 4. The exponent corresponding to x in Eq 4 is here denoted by [alpha]. The results for parameters [alpha] and [[Phi].sub.m] are listed in Table 2b, as derived from double logarithmic plots of the relative tensile strength (l-[[sigma].sub.c]/[[sigma].sub.o]) versus volume fraction [Phi] (Fig. 4). Errors and the least squares fits were determined as in Fig. 2. Here, only the composite of native starch and PCL apparently displays a different behavior whereas the curves for the other mixtures superimpose in the limits of experimental error. The variance analysis of the regression coefficients, given in Table 2b, reveals that there are no significant differences between the four starch blends. Exponent [alpha] is close to unit and maximum volume fraction [[Phi].sub.m] [approx] 0.7. It is recognized that qualitatively, the tendencies for quantities [alpha] and [[Phi].sub.m] are as in Table 2a for x and [[Phi].sub.m]. The values for [[Phi].sub.m] are in close agreement. These results suggest that PCL can be loaded higher with native starch and TPS than with pre-dried and TPS granules. Changes in values of exponent [alpha], however, are less pronounced than changes of exponent x. One may say, tensile strength decreases linearly with volume fraction of starch to a good approximation for mixtures of PCL and starch.

According to Nielsen [11], the linear composite the ory implies a linear relationship between composite modulus [E.sub.c] and the volume fraction of filler [[Phi].sub.st]. The modulus of elasticity of the composite is given by

[E.sub.comp] = [E.sub.PCL](1 - [[Phi].sub.starch]) + [E.sub.starch] [[Phi].sub.starch] (5)

where [E.sub.comp], [E.sub.starch] and [E.sub.PCL] are the moduli of the composite, starch and PCL, respectively, and [[Phi].sub.starch] denotes volume fraction of starch. Moduli of elasticity were determined for native and predried starch. It turned out that there are no significant differences between the two starch types. Figure 5 shows results for predried starch. Experimental data points are in good agreement with Eq 5. Extrapolation to [[Phi].sub.starch] = 1 allows estimation of the modulus of elasticity for starch. It follows for predried starch

E = (890 [pm] 70)Mpa (correlation r = 0.998).

The uncertainty indicated results again from a variance analysis of the regression parameter at 95% confidence interval.

Schroeter and Hobelsberger [12] calculated the modulus of starch by analyzing the properties of blends of potato starch and epoxy resin and poly([varepsilon]caprolactone) at a strain rate of 1%/min. Although the initial moisture content of the starch was reported, the starch was not dried before use, and the moisture content after the composite preparation was not reported. They found also a linear relationship between modulus and volume fraction which served to estimate the modulus of potato starch under these conditions to 2.7 GPa. Compared to the results obtained in this work, the value of 2.7 GPa is considerably higher. On the other hand, if one extrapolates values for moduli reported by Koenig and Huang [8] for corn starch, it results in a modulus for this starch of 1.4 GPa. These differences in the modulus may indicate that a linear relation between modulus and volume fraction exists only in a certain range of composition rather than over the entire composition range. Then, extrapolation to [Phi] = 1 is not justified. The different results obtained for the moduli of sago and potato starches are not attributable to different amylose/amylopectin ratios since these are very similar [6, 17].

SEM micrographs of test specimens after deformation are shown in Figs. 6 and 7. The Figures demonstrate weak interaction between sago starch and PCL. PCL does not wet perfectly the starch fillers. Figures 6a and b show the effect of increasing starch loading on the composite morphology. One recognizes that even in composites containing 50 wt% of starch, starch does not form a continuous phase. Occurrence of void formation, perpendicular to the principal stress direction, is clearly evident. This voiding, however, is less pronounced at 500% starch content (Fig. 6b). Nevertheless, an indication is given for the failure mechanism of the material. The voids are clearly seen on the longitudinal section through the test pieces in Fig. 7. Failure starts perferably at the starch-PCL interface. Figure 8 shows tensile test pieces of neat PCL (a) and a PCL/starch composite (b). It illustrates the deformation of samples occurring at 45[degrees] angle to the tensile axis. The composite demonstrates evidence of crazing di stributed along the entire gauge length of the test piece.

CONCLUSIONS

PCL has a remarkable ability to incorporate large quantities of starch. Deformable thermoplastic starch could be successfully blended with PCL up to 80 wt% content, whereas intact granular starch (native and predried) could only be incorporated up to a maximum of 60 wt% content. With respect to the maximum volume fraction of starch before phase inversion occurs, it could be concluded that PCL can be loaded higher with native starch than with predried starch and less with TPS granules compared to TPS. The inclusion of starch into PCL causes a reduction in tensile stress and strain above 10 wt% of starch. An almost linear decrease of tensile strength with volume fraction of starch was observed. Elongation at break displays essentially the same behavior although in some cases, one may also recognice tendencies to nonlinear variation with volume fraction of starch. SEM studies revealed incomplete wetting of the starch by PCL. This weak adhesion causes reduced tensile strength of the composites.

ACKNOWLEDGMENT

This research was sponsored by CRAUN Research Sdn. Bhd. Sarawak (Malaysia). The support is gratefully acknowledged.

(*.) To whom correspondence should be addressed.

REFERENCES

(1.) H. G. Fritz, W. Aichholzer, T. Seidenstuker, and B. Widmann, Starch/St[ddot{a}]rke, 47. 475 (1995).

(2.) H. Pranamuda, Y. Toikiwa, and H. Tanaka, J. Envr. Polymr Degrad., 4, 1 (1996).

(3.) J. L. Willet, B. K. Jasberg, and C. L Swanson, Polym. Eng. Sci., 35, 202 (1995).

(4.) R. L. Shogren, G. F. Fanta, and W. M. Doane, Starch/St[ddot{a}]rke, 45, 276 (1993).

(5.) H. Okamoto, Imagawa, Y. Aral, and T. Ozawa, Agric. Biol. Chem., 52, 2215 (1988).

(6.) 0. S. Odusanya, PhD thesis, Universiti Sains Malaysia, Penang, Malaysia (1999).

(7.) C. V. Benedict, W. J. Cook, P. Jarret, J. A. Huang, and J. P. Bell, J. Appl. Polym. Sci., 28, 327 (1983).

(8.) M. F. Koenig and S. J. Huang, Polymer, 36, 1877 (1995).

(9.) R. Mani, J. Tang, and M. Bhattacharya, Macromol. Rapid Comm., 19, 283 (1998).

(10.) E. J. Choi, C. H. Kim, and J. K. Park, J. Polym, Sci., B Polym. Phys., 37, 2430 (1999).

(11.) L. E. Nielsen, Mechanical Properties of Polymers and Composites, 386-414, Marcel Dekker, New York (1974).

(12.) J. Schroeter and M. Hobelsberger, Starch/St[ddot{a}]rke, 44, 247 (1992).

(13.) J. C. Halpin and J. L. Kardos, Polym. Eng. Sci., 16, 344 (1976).

(14.) A. Arvanitoyannis, I. E. Psomiadou, and C. G. Biliaderis, Starch/St[ddot{a}]rke, 49, 306 (1997).

(15.) L. Nicolais and M. Narkis, Polym. Eng. Sci., 11, 194 (1971).

(16.) Z. L. Nikolov, R. L. Evangeslista, W. Wung, J. L. Jane, and R. J. Geilna, Ind. Eng. Chew. Res., 30, 1841 (1991).

(17.) J. L. Willett, J. Appl. Polym. Sci. 54, 1685 (1994).

Printer friendly Cite/link Email Feedback | |

Author: | ODUSANYA, O. S.; ISHIAKU, U. S.; AZEMI, B.M.N.; MANAN, B.D.M.; KAMMER, H.W. |
---|---|

Publication: | Polymer Engineering and Science |

Geographic Code: | 1USA |

Date: | Jun 1, 2000 |

Words: | 3321 |

Previous Article: | Nonisothermal Crystallization Kinetics of Polytetrafluoroethylene. |

Next Article: | High Temperature Slow Crack Growth in Polyoxymethylene. |

Topics: |