Evaluation of starch-PE multilayers: processing and properties.

INTRODUCTION

Multilayer systems are used more and more in packaging. In the last two decades, the technology of complex barrier films has undergone considerable development. The association of different materials with complementary properties has led to high performance films, increasing the lifetime of fresh products. These films generally require good oxygen barrier properties. In addition to protection from oxidation, a high permeation to C[O.sub.2] is also highly desirable. A high C[O.sub.2]/[O.sub.2] selectivity is therefore necessary to allow fresh food respiration. Multilayer barriers are usually made up of a polar polymer in the internal layer covered by an apolar polymer, the former acting as a gas barrier and the latter as an hydrophobic skin layer that prevents fast water absorption in the internal layer. Water plasticizes polar polymers. An increase in water content leads to an increase of the molecular mobility and a decrease of the barrier properties. Typical structures are made of polyolefin skin layers covering a polar polymer, e.g., polyethylenevinyl alcohol copolymer (PE-EVOH).

Transport properties of multilayers are mainly governed by the intrinsic barrier properties of each layer, and it is generally considered that the apparent permeability coefficient can be directly deduced from the permeability of each elementary layer [1]. Since outer layers generally have low gas barrier properties, the permeability of a multilayer is mainly governed by the permeability of the inner layer. Since the inner layer's water content is time dependant (the equilibrium being quickly reached only in thin materials), the difficulty consists in being able to predict the change in permeability behavior in a given environment over time.

The originality of this work consists in replacing the classical rather expensive EVOH inner layer material, by a much cheaper material of similar properties. Starch-based materials are good low cost potential substitutes because of their interesting oxygen barrier properties. Plasticized starch, called "thermoplastic starch" (TPS), is obtained by extrusion of native starch in presence of a plasticizer (water, glycerol, sorbitol,...) which controls its glass transition [2, 3]. Recently, TPS has been shown to possess good oxygen barrier properties. The behavior of this system is complex due to the simultaneous evolution of [alpha] and [beta] relaxation temperatures with water and plasticizer contents [4]. Good barrier properties are retained for a variety of conditions [4]: in the glassy state the barrier properties of plasticized TPS are comparable to those of EVOH. The starch local molecular mobility is decreased by an antiplasticization effect at an optimum value of 8% sorbitol. This illustrates that TPS and EVOH could be used in similar applications. In the rubbery state, due to high plasticizer/starch interactions, samples with low plasticizer content retain good barrier properties. For high plasticizer content, the permeability increases exponentially with the water content but TPS maintains good barrier properties up to 65% relative humidity. For gas barrier multi-layer applications, films with a very low plasticizer content are too brittle and cannot be used. In this paper, we investigate systems with intermediate plasticizer contents. The TPS formulation must also fulfill good layer adhesion properties to prevent both delamination and stable coextrusion process.

The problem of adhesion between TPS and polyethylene in blend applications can be solved by adding ethylene maleic anhydride (EMA) copolymers as shown by abundant literature on PE/starch compatibilization in blends [5]. Optimization of this solution will be dealt with in this paper first by studying the use of EVOH in TPS formulations, since commercial maleic acid grafted copolymers are generally designed for EVOH contact, and second by grafting alkyl chains onto starch.

Multilayer coextrusion has been widely used in the past decades to combine the properties of two or more polymers into one single multilayered structure. However, some problems inherent to the multiphasic nature of the flow (such as non-uniform layer distribution, encapsulation, and interfacial instabilities) are likely to occur during coextrusion operations. These problems are critical since they directly affect the quality and functionality of the multilayer products. Past research suggests that these phenomena are ruled by the shear viscosity of each material, as well as the layer thickness ratio [6, 7]. Measurements of shear viscosity of plasticized starches, either by in-line rheometry or specific rheometers, have shown that TPS behavior follows a power law, the parameters of which depend on temperature, plasticizer content, and mechanical energy [8-12]. These results have allowed us to find a fair agreement on the numerical values found for these coefficients, depending mainly on the botanical origin of starch. The ratio of highly branched macromolecules (amylopectin) to linear ones (amylose) and the molecular weight of the latter are the most important structural features. Although this knowledge has rarely been extended to elongational and elastic properties [13], the practical importance of such results being the ability to predict the behavior of TPS in shaping processes [14]. The aim of this paper is to analyze the effects of the formulation of the starch-based layer (i.e., the starch botanical origin, the plasticizer's nature and content, the chemical modification, and the blending with other polymers) on the coextrusion process and the adhesion between layers. A good compromise must be found between the process conditions, interlayer adherence, and the gas barrier behaviors. Water sorption and oxygen permeation properties are tested on optimized multilayer systems.

MATERIALS

Starches and Plasticizers

Unless mentioned otherwise, for the coextrusion section, the starch used in the study was potato starch, provided by Roquette company (Lestrem, France). Potato starch contains 20% amylose. The residual protein content is lower than 0.1% and no lipids can be detected. Native wheat starch was purchased from Chamtor (Pomacle, France). It contained 74% amylopectin and 26% amylose. According to the supplier, the residual protein and lipids content is less than 0.2% and 0.7%, respectively. Initial starch moisture content was around 12%. Plasticizers, glycerol and sorbitol, were purchased, respectively, from Aldrich and Roquette company (Lestrem, France).

Extruded Plasticized Starch (TPS)

A dry blend of glycerol-plasticized starch was prepared according to the following steps. Native starch was weighed and introduced in a Turbo-Mixer (SCAMIA, France). Glycerol was then slowly added to starch under continuous gentle stirring (300 rpm) for 2 min. After glycerol addition, the mixture was dispersed at high speed (3000 rpm) to obtain a homogeneous dispersion. Sorbitol formulations were obtained by the direct blend of powders in the turbomixer. The mixture was then placed in a ventilated oven at 170[degrees]C for 45 min and stirred occasionally, allowing complete evaporation of water and diffusion of glycerol into the starch granules. After cooling, the proper amount of water was added to the mixture that was once more dispersed at high speed in the mixer (3000 rpm). This dry blend was then co-extruded or blended.

Casted Starch

Casting was used to obtain thinner films than the ones obtained by extrusion. These thin monolayer or multilayer structures were designed to display barrier properties low enough to be measured, in order to reduce testing time.

Native starch aqueous suspensions, 4% weight (dry basis), were heated at 120[degrees]C for 20 min in a commercial pressure cooker (SEB, France). The solutions were spread on an anti-adhesive coated hot mold maintained at 60[degrees]C to accelerate water evaporation and avoid extensive starch retrogradation. The films obtained had a 35-[micro]m average thickness and were transparent. All films were stored in a 57% relative humidity (RH) atmosphere, controlled by a saturated NaBr solution at 25[degrees]C, for one week before testing.

Other Polymeric Materials

Starch octanoates were obtained from potato starch by reaction with octanoyl chloride at ENSIACET (Toulouse, France) [15]. The degree of substitution (DS) was determined by nuclear magnetic resonance (NMR) [16]. Polyethylene films, low density polyethylene (LDPE), polyethylene grafted maleic anhydride (PEg, two grades: Orevac OE330 and Orevac OE322), and ethylene vinylic alcohol (EVOH) were provided by ATOFINA. The films thickness used was 20 [micro]m for permeation and sorption tests, and 100 [micro]m for peeling test. EVOH/starch blends were obtained by extrusion at 150[degrees]C.

"Model" Multilayers

"Model" films were obtained by a thermomolding process. Film composition and characteristics are reported in Table 1.

Peel Test on Bilayer Films. Several systems were first tested: thermoplastic starch (containing 18% sorbitol), starch octanoate with different degrees of substitution in contact with polyethylene or maleic anhydride grafted polyethylene (PEg), and starch/EVOH blends in contact with PEg.

Plasticized starch was preconditioned for 2 hr at 70[degrees]C in a ventilated oven. Some samples were re-hydrated (when specified) to obtain the same final water content. Sorbitol and glycerol plasticized starch films had to be equilibrated in different conditions: 57% RH (NaBr saturated solution in water at 20[degrees]C) and 33% RH (Mg[Cl.sub.2] saturated solution in water at 20[degrees]C), respectively, to obtain 10% water content. The 4% water content was obtained using 15% RH (LiCl saturated solution in water at 20[degrees]C) with both plasticizers.

PEg films were preconditioned for 2 hr at 90[degrees]C for water removal and anhydrides reformation. Bilayer samples were obtained by compression molding between hot plates (130[degrees]C) of 200-[micro]m thick starch samples and an LDPE or PEg film (100 [micro]m) just after preconditioning. The final thickness of LDPE and PEg layers were about 80 [micro]m. Peel tests were performed immediately after molding.

Water Sorption and Gas Permeation on Multilayers. Multilayers films (five layers: LDPE/PEg/casted starch/PEg/LDPE) were obtained by thermomolding at 120[degrees]C under pressure with a laboratory press equipped with heated plates and a constant thickness film maker (Specac, UK) having a diameter of 3 cm. Edges were cut to fit the dimensions for permeation properties and water sorption testing. The inner starch layer was weighed before molding and equilibrated under 11% RH (LiCl saturated solution) to limit water content. The final thickness was constant and equal to 64 [micro]m (16-[micro]m plasticized starch layer coated by two 24-[micro]m LDPE + PEg layers).

Reference monolayer samples (barrier layer only) were processed under the same conditions.

Coextruded Multilayers

Coextrusion feasibility of PEg/Starch/PEg structures for multilayer films was tested on an experimental set-up consisting of two single-screw extruders, a feedblock attached to a wide flat die, and a three-roll calendering system. A specially designed torpedo shaped screw was used for the extrusion of starch, and a 30-mm diameter, 26:1 L/D single screw extruder was used for the cap layer. The die was constituted of a rectangular entry channel (L X W X h = 50 X 30 X 4 mm), a coat-hanger of decreasing cross section, a die land area of adjustable height, a relaxation area of decreasing thickness (from 4.5 to 1 mm), and finally the die lips (L = 350 mm).

METHODS

Peel Testing

Peel tests were carried out on a Twin-Albert peel tester (model 225-100) at a rate of 50 mm/min. The test specimens were cut from the bilayers into 100 X 20 mm strips. The outer layer (PEg or LDPE) was delaminated manually and attached to the load cell, while the film was secured on a sliding plate, so that a constant 90[degrees] angle between the outer layer and inner layer was maintained during the test. Load data collection started after 3 s of pre-peel. A mean value of peeling force was determined from each test, and each sample was tested five times.

Water Sorption and Oxygen Permeability

Samples were conditioned over saturated salt solutions at 20[degrees]C for various times depending on the test (2 weeks for multilayer films, 1 week for monolayer starch films). The saturated salt solutions used were: NaHS[O.sub.4] (50% RH) for oxygen permeability measurement, and NaCl (76%), [Na.sub.2]C[O.sub.3] (92%), and CuS[O.sub.4] (98%) for water sorption measurements. The weight of the inner starch membrane was measured at 15% RH (LiCl saturated salt solution) before thermomolding and the formed films were not cut in order to avoid weight loss on edges. The multilayer film was then rolled up carefully and placed for 2 weeks in four different environments of desired water activities, as mentioned above, in order to establish the corresponding isotherms. Rolled up sample weights were then measured with a thermal gravimetric analyzer (TGA 2950, TA instrument). The film was then heated at 70[degrees]C until constant weight was reached, i.e., after about 4 hr. A blank test was performed on a multilayer film made without starch in the inner layer. The measured weight loss was almost completely due to water desorption. Oxygen permeability was analyzed with a Mocon OxTran 200 H (Lippke, Germany) following ASTM standard 3985. Film samples were double masked with two aluminum foils leaving a circular uncovered film area of 5 [cm.sup.2]. The circumference was glued with an epoxy glue thick circle in order to avoid any oxygen leak during measurement. On one side of the film, 100% oxygen gas was flowed, and on the other side was a 100% nitrogen gas [N.sub.2] flow. Nitrogen was directed to a coulometric sensor. Measurements were obtained when the steady state was reached. The RH of both gases was controlled by a humidifier and was set to 50% RH. Measurements were conducted at 23[degrees]C. Prior to the experiments, the films were conditioned for 2 weeks at 50% RH and 25[degrees]C, and in the measuring cells at the desired RH for 48 hr. The oxygen fluxes through the multilayer or monolayer films were obtained from the instrument as [cm.sup.3]/[m.sup.2]/24 hr. It was then multiplied by the thickness of the barrier membrane (the inner layer of plasticized membrane measured after delamination) and divided by the pressure difference of oxygen (101 kPa) to obtain the oxygen permeability value.

RESULTS AND DISCUSSION

Adhesion Between Layers

This test first allowed us to choose the system to be further studied and to select the potential way to improve adherence between PE and TPS. EVOH can be blended by extrusion with TPS. Adherence between the two polymers was sufficient and no comonomer was needed to improve compatibility. Because of the lower hydrophilicity of EVOH, EVOH/TPS blends were thought to have a better compatibility with polyethylene or modified polyethylene. Results of peel tests are shown in Fig. 1. A commercial system generally leads to a minimum of 10 N peel strength. EVOH had no positive effect on PEg adhesion; EVOH-starch blends even led to a slightly higher peel force than pure EVOH (denoted 100/0).

[FIGURE 1 OMITTED]

As expected, a considerable drop, nearly by a factor of 10, was obtained when PE was used instead of PEg with TPS. This drop was only slightly reduced by replacing TPS by a 57% substituted starch octanoate. No measurable force was obtained for a 93% substituted starch octanoate: even though this sample can be considered as apolar. In this case, the low polarity is not a key factor since less substituted samples led to better results, but still not satisfactory enough for adequate adhesion (results not shown). The use of a grafted PE instead of polyethylene did not improve the system; the reactivity of starch octanoate towards maleic anhydride was probably too low since most reactive sites were likely to have already been consumed. Consequently, starch esterification did not improve adhesion. PEg provided the only compatibilization of the two layers.

The PEg/Plasticized Starch/PEg System

Temperature Effects. Increasing the compression molding temperature did not produce higher peel forces but the appearance of bubbles at the interface caused by the rapid volatilization of residual water led to a larger dispersion of the experimental data. Another attempt with completely dried samples (60[degrees]C 1 day under vacuum), heated at temperatures between 130 and 160[degrees]C, also led to low scattered results. Moreover, degradation, leading to brown coloration, was observed on samples heated at 150 and 160[degrees]C. In the end, TPS in contact with grafted PE, heated for 1 min at 120[degrees]C, was the most efficient system.

Water and Plasticizer Content. Previous experiments were performed with the same starch formulation on dried sample but this system was too brittle to have any practical use. To overcome this problem, plasticizer and water contents were increased. However, large plasticizer contents lead to phase separation and exudation on the surface, which can perturb adhesion, whereas water inhibits reaction between anhydrides and starch. Consequently, the composition of the starch/grafted PE system was adapted by modifying both the water content and the plasticizer's nature and content. Glycerol and sorbitol contents were adjusted to obtain similar glass transition temperatures at 50% RH in order to obtain comparable physicochemical and mechanical properties (for example, a 14% glycerol sample is considered equivalent to a 18% sorbitol sample [3]). When the original water content was 10%, all systems showed weak adhesion between layers, as illustrated by the low peeling strength values ([less than or equal to] 5N) (Fig. 2). Dry films generally led to good adhesion (strength values [greater than or equal to] 8N) except in the case of films with high glycerol contents ([greater than or equal to] 20%). Finally, films made with a starch layer containing 4% water and high plasticizer content (>18% sorbitol/>14% glycerol) led to the best adhesion (peel strength ~10N). This low water content was obtained by moderate drying of plasticized starch granules (1 hr at 70[degrees]C in a ventilated oven).

[FIGURE 2 OMITTED]

The behavior of glycerol and sorbitol formulations was identical, except in the case of highly plasticized dry samples, in which glycerol exudation may have inhibited interfacial reactions. With the exception of this former example, the increase of plasticizer content had a positive influence on adherence.

Suitable adherence was obtained when a good balance between a low local viscosity and chemical inhibition mechanisms at the surface was achieved. The presence of water decreased the quantity of anhydrides, but induced lower viscosity. The presence of plasticizer improved the reactivity by lowering viscosity (even at high contents in the case of sorbitol). On the contrary, when the glycerol content was too large, adherence decreased, possibly due to an exudation phenomenon, which led to a weaker boundary layer.

Viscous Behavior of Low Hydrated Plasticized Starch and Coextrusion Optimization

Based on the adhesion results, the determination of TPS viscosity focused on the formulations with 5% water and 16-20% glycerol. Two glycerol contents were selected, namely formulations S75G20W5 (20% glycerol, 5% water) and S79G16W5 (16% glycerol, 5% water). An important factor for reducing interfacial instabilities generated during coextrusion is the rheologic compatibility between the molten layers in contact. Both polymers should have similar viscosities under the same flow conditions in order to minimize encapsulation and interfacial instabilities. Former experimental results have shown that plasticized starch viscous behavior can be classically described by a power law [11, 12]:

[eta] = K[dot.[gamma].sup.m-1] (1)

K = [K.sub.0] exp(E/R[T.sub.a] - [alpha]MC - [alpha]'Gl - [beta]SME) (2)

where [T.sub.a] is the melt temperature (K), MC and Gl are moisture and glycerol contents (total basis), and SME the specific mechanical energy (J*[m.sup.3]) delivered to the starch melt during extrusion. The values of the coefficient obtained by regression of the experimental results on molten starches in the above cited studies are summarized in Table 2 for starch from two different botanical origins [11, 12]. Results obtained during preliminary experiments for TPS potato starch at 160[degrees]C, containing 25% glycerol and 10% water, led to K = 78,000 Pa*[s.sup.m] and for wheat starch at 125[degrees]C, containing 23% glycerol and 14% water, to K = 11,000Pa*[s.sup.m] [14-16]. These values were extrapolated to other compositions or operating conditions using Eq. 2, assuming that the flow index "m" did not change in this interval in agreement with former results and assuming that the SME value is not significantly modified by the change of TPS composition. The resulting flow curves are shown in Fig. 3. Potato starch has often been selected over wheat starch because of the larger mechanical resistance of the resulting material, which has been attributed to longer amylose chains. This is also reflected by the larger shear viscosity at T = 150[degrees]C (typical temperature for standard processing conditions). In the examined range, there is little difference between flow curves corresponding to different glycerol contents, the botanical origin being the most important factor.

As mentioned before, a graft polyethylene had to be used to obtain suitable adherence with the skin layer. For comparison purpose, the flow curves of the selected grafted polyethylene are also reported in Fig. 3 in a shear rate range of 10 to [10.sup.3] [s.sup.-1]. For conventional shaping processes, the average die shear rate is close to [10.sup.3] [s.sup.-1]. Taking into account these constraints, we can see that for these target compositions, wheat starch has a viscosity closer to PEg's than that of potato starch. Finally, we selected the 16% glycerol TPS/PEg couple, which presented, at 150[degrees]C and at [10.sup.3] [s.sup.-1], the closest viscosity values (near 550 Pa.s). The co-extrusion tests have confirmed that the system OE 330/16% glycerol wheat starch/0E 330 presents less instability than other tested systems (using potato starch, even for high glycerol contents).

[FIGURE 3 OMITTED]

However, delaminated systems observed by optical microscopy had holes (roughly 50 [micro]m in diameter) revealing the porosity of the PEg layer. Even when the layer thickness ratio was decreased, the porosity of the PEg layer could not be totally suppressed, although low skin thickness tended to reduce interfacial instabilities [7]. Interfacial instabilities are controlled by viscosity, thickness ratio, and compatibility (wettability) between layers. In conclusion, the suitability of the polyethylene system for a coextrusion process is questionable. Interfacial instabilities should be controlled by using less apolar skin layers [17].

Water Sorption of Multilayers

For water sorption and permeation tests, thermomolding was preferred to coextrusion, since coextrusion led to samples with defects (adhesion defects and layer thickness heterogeneities). Complex multilayer systems (LDPE/PEg/TPS/PEg/LDPE) were thus prepared [5]. Based on the adherence results, the water and glycerol contents were maintained at 5% and 18%, respectively. The aim of this part of the work was to investigate the "skin effect" on water uptake. Multilayer samples and TPS samples were stored for 2 weeks under the desired RH. Multilayer samples had a lower water content than monolayer ones for RH [less than or equal to] 75% (Fig. 4). The higher content at 92% RH could be attributed to the experimental error in this range of water activity, for which the samples were partly delaminated and showed visible opaque areas. Samples stored at RH [less than or equal to] 75% remained cohesive and transparent.

The multilayers thickness was fixed to a low value of 65 [micro]m, in order to favor a fast equilibrium. For this range of thickness the typical equilibration time before permeation testing was 1 day at room temperature [1]. However, after 15 days, samples were not yet at equilibrium, except for those stored at high RH. Two main reasons may be evoked in order to explain the very slow equilibration compared to classical multilayers:

1) the water content of starch at equilibrium is high, around 20% at 70% RH. This high value is detrimental to barrier properties, but favorable to the increase of the transition period.

2) TPS water sorption isotherm has a specific shape with a "plateau" between 20 and 60% RH and a sharp exponential increase above 70% RH. This shape is much less pronounced in the case of classical EVOH barrier layer. During water sorption, LDPE/gas barrier/LDPE systems are equivalent to an LDPE membrane submitted to a decreasing water pressure gradient ([DELTA]P = external [constant] water pressure - time dependant water pressure) in equilibrium with hydrated starch or EVOH. In the case of our experimental systems, because of the isotherm's specific shape (with a "plateau"; Fig. 5), a slight increase of TPS water content results in a large increase of the water activity in the TPS layer. As a consequence, [DELTA]P applied to polyethylene layers decreases drastically after a short water sorption time. Consequently the majority of the transition period of water equilibration proceeds at low [DELTA]P. This [DELTA]P decrease results in a lower water permeation rate through the polyethylene layers.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

Taking into account all the possible resistances to mass transfer (PEg's and LDPE's water permeabilities, concentration dependant water diffusivity in starch layer, effect of water activity gradient decrease) several attempts at numerical simulation of water sorption were performed to illustrate these specific properties. All the simulations led to water equilibration times (around 3 days) lower than experimental ones (more than 15 days). This discrepancy suggests that other phenomena tend to increase the calculated equilibration rate.

Oxygen Permeability of Multilayers

The same model multilayers, as in the previous section, were considered for this property study. The effect of TPS glycerol content was studied in the 12-25% range. As for the sorption, the behavior of the multilayer system was compared to a TPS monolayer.

In the limit of the studied range, the oxygen permeability of the monolayer sample increased exponentially with glycerol content. This is in good agreement with the results from the literature [18] at higher glycerol contents (Fig. 6). In this section, permeation properties were determined at 50% RH. The use of a multilayer structure at this water activity had little benefits since the properties in the equilibrium conditions are suitable for most applications.

For the highest glycerol contents, the presence of polyethylene layers had no influence on the permeation properties. In terms of mechanical criteria, the adhesion value obtained with the sample containing 25% glycerol was acceptable. The disappearance of the "skin effect" on the observed permeability values may be attributed to a lack of adherence.

In the case of lower glycerol contents, the multilayer system displayed a much lower permeability coefficient than the monolayer system, up to 10 times lower for 12% glycerol. This difference was attributed to the slower water uptake as discussed before. These results confirm the specific properties of starch based multilayers, showing a long transition period for water equilibration. It must be emphasized that the measured properties were obtained on very thin samples, specifically designed to reduce the transition period. Consequently, the use of thicker layers should fulfill a variety of packaging barrier applications.

[FIGURE 6 OMITTED]

CONCLUSION

This study focused on starch-polyethylene based multilayers and involved two steps: feasibility and performance of the designed materials. The first step was based on an adherence criteria. Grafted starches, chosen as possible candidates for their possible compatibility with polyethylene, had to be discarded since they exhibited adhesion values 10 times lower than the other materials. A system based on five layers was selected as the optimum design: low density polyethylene/poly(ethylene-co-maleic anhydride) PEg/plasticized starch/poly(ethylene-co-maleic anhydride)/low density polyethylene, noted PE/PEg/TPS/PEg/PE. The composition of TPS (plasticizer and water contents) was optimized on the basis of this system.

In a second step, the optimized TPS formulations were used in order to study coextrusion feasibility. TPS shear viscosity, calculated for the target composition, showed that wheat starch was the more suitable botanical type and allowed one to optimize co-extrusion processing conditions of TPS and PEg in order to reduce interfacial instabilities. However, the extreme incompatibility of the phases led to interfacial instabilities, inducing a partial porosity of the skin layer.

The gas barrier properties of model (thermoformed) multilayer materials were improved at high RH when compared to a monolayer film. Due to the high quantity of water sorbed by thermoplastic starch on the one hand, and the specific plateau shape of starch water sorption isotherm on the other, the water equilibration time in the inner layer is very long compared to classical EVOH containing systems, thus enhancing the gas barrier properties. This specific "water-buffering property" of the starch inner layer should prove useful for applications requiring gas barrier properties in environments of variable RH such as in the packaging of perishables with extended shelf life.

Since the coextrusion process does not seem to be appropriate to the polyethylene-starch system, and given the very attractive gas barrier properties, other processing routes, e.g., lamination with adhesives, should be explored.

TABLE 1. Tested films main characteristics.

                    Composition
Film      TPS                          Synthetic polymer

1 layer   TPS plasticized by glycerol
2 layers  TPS (glycerol, sorbitol,     LDPE or PEg
            various contents) Starch
            Octanoates EVOH/starch
            (18% sorbitol) blends
5 layers  TPS plasticized by glycerol  LDPE and PEg

Film      Manufacturing    Thickness           Test
                           [micro]m

1 layer   Casting          35                  Water sorption and
                                                 gas permeation
2 layers  Compression T =  Synthetic polymer:  Peeling
          130[degrees]C    100 [micro]m
                           Starch materials:
                             200 [micro]m
5 layers  Compression T =  65 [micro]m         Water sorption and
            120[degrees]C                         gas permeation

TABLE 2. Power law coefficient values for TPS viscosity from Eqs. 1 and
2.

Starch    E/R                                  [beta]
origin    (K)   [alpha]  [alpha]'  [10.sup.-9](J*[m.sup.3])[.sup.-1]

Wheat     4250   10.6      3.5                  0.88
Potato    5700    9.5      3.5                  1.55

ACKNOWLEDGMENTS

The authors thank Isabelle Alric and Elisabeth Borredon (ENCIACET, Toulouse, France) for supplying octanoated starch, and G. Stockton for revision of the manuscript.

Contract grant sponsor: AGRICE (ADEME).

REFERENCES

1. Y. Germain, Les emballages actifs, N. Gontard, editor, Editions Lavoisier Paris, 65 (2000).

2. C.L. Swanson, R.L. Shogren, G.F. Fanta, and S.H. Imam, J. Env. Polym. Deg., 1, 155 (1993).

3. D. Lourdin, L. Coignard, H. Bizot, and P. Colonna, Polymer, 38, 5401 (1997).

4. S. Gaudin, D. Lourdin, P.M. Forsell, and P. Colonna, Carbohyd. Polym., 43, 33 (2000).

5. C. Bastioli, Polym. Degrad. Stab., 59, 263 (1998).

6. W.J. Schrenck, N.L. Bradley, T. Alfrey Jr., and H. Maack, Polym. Eng. Sci., 18, 620 (1978).

7. C.D. Han and R. Shetty, Polym. Eng. Sci., 16, 697 (1978).

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

9. G. Della Valle, Y. Boche, P. Colonna, and B. Vergnes, Carbohyd. Polym., 28, 255 (1995).

10. G. Della Valle, P. Colonna, A. Patria, and B. Vergnes, J. Rheology, 40, 347 (1996).

11. W. Aicholzer and H-G. Fritz, Starch, 50, 77 (1998).

12. L. Averous, L. Moro, P. Dole, and C. Fringant, Polymer, 41, 4157 (2000).

13. G. Della Valle, P.J. Carreau, P. Colonna, and B. Vergnes, Vth European Rheology Conference Proceedings, Slovenia, I. Emri and R. Cvelbar, editors, Steinkopf Darmstadt, 260 (1998).

14. O. Martin and L. Averous, J. Appl. Polym. Sci., 86(10), 2586 (2002).

15. J. Aburto, S. Thiebaud, I. Alric, E. Borredon, D. Bikiaris, J. Prinos, and C. Panayiotou, Carbohyd. Polym., 34(1-2), 101 (1997).

16. C. Fringant, J. Desbrieres, and M. Rinaudo, Polymer, 37(13). 2663 (1996).

17. L. Averous, C. Fringant, L. Moro, and J.C. Prudhomme, French Patent 16/10/2000. N FR 2791603.

18. J.C. Rankin, I.A. Wolf, H.A. Davis, and C.E. Rist, Ind. Eng. Chem., 3, 120, (1958).

Patrice Dole, Luc Averous, Catherine Joly

UMR URCA/INRA FARE--CPCB, Moulin de la Housse, BP 1039, 51687 Reims Cedex 2, France

Guy Della Valle

INRA, rue de la Geraudiere, BP 71627, 44316 Nantes Cedex, 3 France

Christophe Bliard

FRE URCA/CNRS 2715--Bat 18, Moulin de la Housse, BP 1039, 51687 Reims Cedex 2, France

Correspondence to: P. Dole; e-mail: [e-mail]