The OP was measured with the oxygen-specific carrier gas method according to DIN 53380-3. The measurements were performed with an Ox-Tran Twin Oxygen-Permeation Measuring Machine (Mocon Inc., Minneapolis, MN, USA) at a temperature of 23°C and 50% r.h. The measurement stopped when a value was constant for at least 10 h. A twofold determination for all coatings and cast film specimens was carried out. The OP is given in the unit cm³ (STP) m⁻² day⁻¹ bar⁻¹ (DIN, 1998).

The WVTR was measured by the gravimetric screw-cup method according to the standard DIN 53122-1. According to DIN 53122-1, the film samples were fixed on a glass or metal cup containing dried silica-gel, which generates an atmosphere of 1.3% r.h. (referred to 0% r.h.) and was sealed with vacuum grease and screws to avoid leaks. The samples were weighted and placed in a climate controlled cabinet (23°C/50% r.h.). This generates a relative humidity gradient of 50% → 0% r.h. The cups were reweighted periodically. The gained weight indicates the amount of permeating water vapor. A fourfold determination was carried out unless indicated otherwise. Arithmetic mean and SDs were calculated. The WVTR is the change in weight of the whole cup and can be calculated using Eq. 1, whereby t is the period of time between two weight measurements [h], Δm is the weight difference between two weight measurements [g], A is the test area [cm²], and WVTR is the water vapor transmission rate [g m⁻² day⁻¹] (DIN, 2001; Robertson, 2006): (1)WVTR=24t∗ΔmA∗104[g m−2 day−1].

The permeability, Q, of a polymer can be calculated using Eq. 2 as the quotient of the permeation coefficient, P, and the thickness, d, of a polymer (Langowski, 2008): (2)Q=Pd[cm3 (STP) m−2 day−1 bar−1].

To adapt permeability properties of a packaging material, multilayers can be used instead of raising the thickness of a monolayer. The total permeability of a multilayer structure, Q_total, can be calculated using Eq. 3, whereby d_i represents the thickness of the individual layers (d_i, d = ∑d_i), and P_i represents the permeation coefficient of the individual layers (P_i, P = ∑P_i). Alternatively, the total permeability of a multilayer structure can be calculated from the respective permeabilities of the individual layers, Q_n, whereby the superscripts (1 and 2) represent the respective layers (Schmid et al., 2012): (3)1Qtotal=∑i=1i=ndiPi=1Q1+1Q2+…1Qn.

For a better comparability between different specimens independently of the respective layer thicknesses, the permeability, Q, can be standardized to the Q₁₀₀—value relating to a thickness, d, of 100 µm (Menges, 2002): (4)Q100=Q∗d100.

Oxygen permeability measurements were performed for cast films and coatings (Table 1). For better compatibility, only the permeability of the coating (monolayer) is considered as the OP of PET is much higher. Permeability of the coating monolayer can be calculated according to Eq. 3. The measurements were normalized to a layer thickness of 100 µm (Q₁₀₀) using Eq. 2. The following figures demonstrate the Q₁₀₀ values of the cast films and coatings with a protein content of 10 and 15% protein (w/w) and either 66.70% glycerol/sorbitol or 100.00% glycerol/sorbitol (w/w) as plasticizer. The percentages of MMT always refer to the weight of the protein. Figure 3 and Table 2 show the OP of WPI-based nanocomposite cast films and coatings. OP measurements were performed according to DIN 53380-3 at 23°C and 50% r.h.

The results show that the addition of MMT nanoplatelets led to a continuously decreasing OP of the coatings monolayer. In comparison to the reference, the addition of 3% MMT (w/w protein) into glycerol-plasticized coatings already improved the oxygen repellency by a factor about 2. The lowest OP, as expected, could be observed for coatings with the highest nanofiller loadings, representing a reduction of the OP compared to the reference (without nanofiller) by a factor about 10. For sorbitol-plasticized nanocomposite coatings, with exception of coatings with 9% (w/w protein) nanofillers, the OP also decreased but, however, the effect of the increasing nanofiller loading toward the OP was not as clear as for glycerol-plasticized coatings. The renewed increase in OP at 9% (w/w protein) nanofillers could be explained by the orientation of the nanoplatelets within the protein network as with further addition of nanofillers the OP started to decline again until it reached its least permeability of approximately 0.58 cm³ m⁻² day⁻¹ bar⁻¹ at the highest filler loading which is in the range of the permeability values of EVOH (Schmid et al., 2012). These results underline that WPI-based nanocomposite coatings are capable of replacing commercial petroleum-based barrier polymers, e.g., ethylene–vinyl alcohol copolymer (EVOH 44%) which is used in multilayer flexible packaging applications as an effective barrier against gases and particularly against oxygen permeation. An improvement of the OP with increasing nanofiller-content was also observed by Oguzlu and Tihminlioglu (2010) for chitosan-based nanocomposite films. Decreases in OP could be attributed to the presence of the dispersed nanoplatelets which, depending on their alignment and state of exfoliation, exhibit very large aspect ratios. This consequently led to increased tortuosity and longer pathways for permeating gas molecules (Ray et al., 2006; Yoo et al., 2014; Visakh and Sigrid, 2016). However, based on a more uniform dispersion at low filler loadings and thus better exfoliation of nanoparticles, other researchers reported lower OP with decreasing nanofiller content (Weizman et al., 2017). For coatings with 15% (w/w) WPI and 100.00% (w/w protein) plasticizer (Table 2), similar results as for the nanocomposite coatings with 10% (w/w) WPI have been observed. The addition of 9% (w/w) MMT led to a reduction of the OP compared to the reference sample (without nanofiller) by about 90% in case of glycerol, and about 70% in case of sorbitol.

Similar observations were also made for the effect of different nanofiller loadings on the OP of nanocomposite cast films with 10% (w/w) WPI and 66.70% (w/w protein) plasticizer (Table 2). In case of glycerol-plasticized nanocomposite cast films, an addition of 3% (w/w protein) MMT reduces the OP by about one third compared to the reference sample. Between 3 and 6% (w/w protein), nanofiller addition did not result in considerable differences regarding OP. However, further addition of nanofillers reduced the OP. At the highest filler content, the addition of MMT led to a reduction of the OP by a factor of about 8.45 compared to the reference sample. In case of the cast films with 15% (w/w protein) nanofillers and, especially for sorbitol-plasticized films, strong deviations in permeation data could be attributed to the orientation of the nanofillers in the WPI-matrix as an increased diffusion path, among other factors, is highly dependent on the lengthwise orientation of the nanofiller particles (Ray et al., 2006). To prove this hypothesis, the morphology of the cast films has to be examined by microscopy methods. Compared to the permeation data of the coatings, the cast films showed more deviations in OP. This can most likely be attributed to a more uniform application of the coating monolayer as the cast films showed large differences in their layer thicknesses due to uneven alignment during convection drying or the curing in a climate cabinet. Nevertheless, the reduction of the OP with increasing nanofiller content could be demonstrated.

As mentioned earlier for sorbitol-plasticized cast films and coatings, reduction of the OP was not as clear as for glycerol-plasticized nanocomposite cast films. However, it should be noted that sorbitol was much more effective as cast films and coatings plasticized with sorbitol generally yielded much lower oxygen permeabilities compared to glycerol-plasticized cast films and coatings. Generally, protein-based films and coatings which are plasticized with sorbitol were reported to exhibit lower oxygen permeabilities than glycerol-plasticized films (McHugh and Krochta, 1994; Chick and Ustunol, 1998). Protein-based films in general are good barriers against the diffusion of small gas molecules like oxygen or carbon dioxide due to their hydrophilic nature (Miller and Krochta, 1997). However, the incorporation of plasticizers, e.g., polyols, increases the permeation coefficient (Lieberman and Gilbert, 1973; Schmid et al., 2012) due to increased absorption of moisture into the hydrophilic biopolymer network which consequently leads to increased intermolecular spacing as supported by free-volume theory (Lieberman and Gilbert, 1973). As glycerol, compared to sorbitol, shows a higher hygroscopicity (Shaw et al., 2002), its effectiveness in increasing the free-volume and thus increased permeation of oxygen molecules could be attributed to the structure of polyol plasticizers as glycerol is much smaller than sorbitol and thus, based on their molecular weight, more abundant at the same weight-% bases. Counteracting to the contributions of the plasticizers in increasing the free-volume, however, extensive interfacial interactions between the hydrophilic polymer, plasticizer, and the MMT via hydrogen bonding could significantly lower the free-volume and thus lower the permeability (Huang et al., 2015).

The WVTR, as already mentioned earlier (see Table 1), was only determined for sorbitol-plasticized WPI-based nanocomposite cast films. Evaluation of glycerol-plasticized cast films with 10 and 15% protein (w/w) and 66.70% or 100.00% glycerol (w/w protein), respectively, was not possible due to strong thickness fluctuations during the WVTR measurement. Therefore, it was not possible to obtain normalized transmission rates (Q₁₀₀-values) for the different cast film formulations. This observation can most likely be attributed to the high-relative humidity gradient during the WVTR measurements (85 to 0% r.h.) which led to increased sorption of water molecules (Lieberman and Gilbert, 1973) into the hydrophilic protein matrix due to the highly hygroscopic properties of glycerol (Shaw et al., 2002). Table 3 shows the WVTRs (Q₁₀₀) for sorbitol-plasticized cast films with 10 and 15% (w/w) protein. All measured samples showed weak barrier properties against water vapor. The reference sample did not contain nanofillers and was measured to evaluate the effect of the nanofiller. It can be seen in Table 3 that with increasing nanofiller loading the WVTR for cast films containing 10% (w/w) WPI significantly (p ≤ 0.05) declines by about 60%. The lowest transmission rates, as expected, were achieved for the highest nanofiller loadings [12 and 15% (w/w, relative to protein) MMT] even though these differences were not significantly different from each other. The addition of nanofillers reduces the WVTR because inorganic MMT nanoplatelets are impermeable to water molecules. Consequently, the water molecules must diffuse past these layers, which leads to an extension of the diffusion path (Giannelis, 1996).

Improved barrier properties against water vapor were also achieved for nanocomposite cast films containing 15% (w/w) WPI (Table 3). Although the reduction of the WVTR is in part not significantly different, WVTR values could be significantly (p ≤ 0.05) reduced by approximately 37% for compositions containing the highest nanofiller loadings. With increasing MMT concentration this effect, however, reverses as the influence of an extended tortuous pathway for the diffusion of water molecules due to the impermeable nanoplatelets gains more influence which leads to a significant reduction of the WVTR as the water molecules must diffuse past the layered MMT (Giannelis, 1996; Bharadwaj, 2001; Bharadwaj et al., 2002). Another possible explanation for the reduced transmission of water molecules through the biopolymer matrix could be attributed to the ability of the WPI-matrix and the highly polar plasticizer molecules to build a more cohesive network with hydroxyl groups of the MMT via hydrogen bonding and thus lowering the absorption and diffusion of water molecules (Almasi et al., 2010). However, given the applied measuring conditions of 85 to 0% r.h. during WVTR measurements revealed that WPI-based nanocomposite cast films, regardless of the nanofiller-content, are still poor moisture barriers. This can clearly be attributed to the high relative humidity which most likely caused the hydrophilic protein matrix to swell. At high relative humidity water molecules accumulate between polar amino acid side chains consequently increasing the intermolecular spacing which results in an increased WVTR due to the increased free-volume between the polymer chains (Lieberman and Gilbert, 1973). This effect is further intensified by the presence of sorbitol as highly hygroscopic polyol plasticizers compete with water molecules for binding sites between polar amino acid sidechains (Lieberman and Gilbert, 1973). As a result, theoretical assumptions, e.g., constant diffusion and layer thicknesses, to calculate WVTR values are no longer valid which leads to inaccurate data (Oymaci and Altinkaya, 2016). However, when comparing the barrier improvements between WVTR and OP, it is apparent that generally much better barrier improvements were achieved in terms of OP compared to water vapor transmission even though the measurement conditions (high relative humidity, swollen biopolymer matrix) were the same. This observation is consistent with observations from Choi et al. (2006) using a non-swelling PET/clay-nanocomposite matrix. A possible explanation for the enhanced oxygen barrier properties could be referred to extensive interactions between the hydrophilic protein/plasticizer matrix, the permeating water molecules, and the hydrophilic surface of the nanofiller. Rhim et al. (2009) reported that the relative surface hydrophilicity of the used MMT nanoplatelets compared to other clays with relatively lower surface hydrophilicity was responsible for increased water vapor permeability’s in a clay/PLA nanocomposite film. By taking into account the highly non-polar character of oxygen, this could be a suitable explanation for the clearly better barrier efficiency against the permeation of oxygen molecules than against water vapor.

The effects of an increasing nanofiller concentration onto YM, elongation at break (E%), and TS of WPI-based nanocomposite cast films are illustrated in Figures 4 and 5. Mechanical characteristics can be useful in identifying structural relationships of composite materials as they relate to their chemical structure (Ninnemann, 1968). The YM defines the relation between applied stress and the associated strain during the elastic deformation of a material. Thus, a high YM indicates high resistance against plastic deformation. The TS is the maximum tensile stress that a film can withstand without failure. Elongation at break (E%) is defined as the ultimate relative length change a material can withstand before breaking and thus a measure for a materials ability to lengthen in relation to its original length. Mechanical measurements were performed at 23°C and 50% r.h. All films were prepared by solvent-casting method thus there is no specific testing direction. Due to extreme brittleness of sorbitol-plasticized nanocomposite cast films, only films containing glycerol as a plasticizer were used to evaluate the mechanical performance. The percentage of the nanofiller MMT refers to the weight of the protein in the film compositions.

For cast films with a protein content of 10% (w/w) WPI and 66.70% (w/w protein) glycerol, it can be observed from Figure 4, that increasing the nanofiller, even at low filler loadings, has significant effects (p ≤ 0.05) on the mechanical properties. A significant improvement of the YM could already be observed between the reference sample and 3% (w/w protein) MMT. The best results regarding the improvement of the YM were obtained at the highest nanofiller loadings (12 and 15% w/w MMT). Although the differences between the highest nanofiller loadings (12 and 15%) were not significantly (p > 0.05) different from each other, they were significantly different from the reference sample by about a factor of 4. Similar improvements were also achieved in the evaluation of the TS of these films. The addition of the nanofiller at its lowest concentration (3% w/w protein) led to improved TS values, even though these differences were not significantly (p > 0.05) different. However, further addition to a filler concentration of 6% (w/w protein) led to significant (p ≤ 0.05) changes in TS. Again, as for the YM measurement, it can be stated that the best results for the TS measurement could be achieved at the highest MMT concentrations. In case of elongation properties, in contrast to YM and TS, it can be observed from Figure 4 that the elongation at break (E%) generally decreases with increasing nanofiller content, even though the reduction of film extensibility was not as clear as the changes in the other tensile properties (YM and TS). This can be attributed to strongly fluctuating film thicknesses which led to extensive standard variations for E% as the values which were obtained by the evaluation of the relative length change of the films are not standardized to the thickness of the film specimens. Respective film thicknesses regarding elongation at break measurement are shown in Figure 4.

For cast films with a protein loading of 15% (w/w) WPI and 100.00% (w/w protein) glycerol, similar results were observed (Figure 5). In this case, as discussed earlier for the viscosity measurement, an addition of MMT was only possible up to a concentration of 9% (w/w protein) as the viscosity of the nanodispersion increases with increasing nanofiller content. Again, as for the films with 10% (w/w) WPI and 66.70% (w/w protein) glycerol, a significant improvement of the YM could already be observed between the reference sample and 3% (w/w protein) MMT. The best results regarding the improvement of the YM, again, were obtained at the highest nanofiller concentrations (6 and 9% w/w) MMT, although these differences were not significantly (p > 0.05) different from each other. Similar results were also obtained for TS. The addition of the nanofiller starting from 6% (w/w protein) led to significant differences in TS. Regarding E% values, the addition of 3% (w/w protein) MMT led to a significant (p ≤ 0.05) reduction of the E% values by a factor of about 2. A further addition of the nanofiller did not lead to significant changes of the elongation properties. Respective film thicknesses regarding the elongation at break measurement are shown in Figure 5.

The addition of nanoscale filler materials, e.g., MMT-nanoplatelets, is a common method to increase the stiffness (YM) of polymeric matrices as widely described by composite theories (Fornes and Paul, 2003; Lee and Paul, 2005). As expected, YM increased with increasing volume fraction of the nanofiller. Generally, the stiffness of polymeric materials compared to their respective reference matrix (without nanofiller) increases with increasing filler concentration as long as the nanofiller is sufficiently dispersed and a certain degree of exfoliation within the matrix is achieved (Utracki and Limited, 2004; Aït Hocine et al., 2008). Therefore, continuously increasing YM values can be understood as good indicators for sufficiently-dispersed and well-exfoliated nanofiller particles within the WPI matrix. These observations also suggest that a critical nanofiller concentration is not yet reached as reinforcing effects for nanocomposites with high filler loadings are reported to be in the range of the neat matrix or even lower due to decreased dispersion/exfoliation and aggregated or stacked nanofiller (Alexandre and Dubois, 2000; Sur et al., 2001; Mittal, 2009).

The addition of nanofillers into polymeric matrices can also increase film strength (TS) (Fornes et al., 2001) but, however, in contrast to YM, TS is a non-linear mechanical property which will decrease to values even lower than those of the neat matrix after a critical filler loading is exceeded (Zhou et al., 2009). The improvement of film strength (TS), in contrast to material stiffness (YM), is mainly governed by the effectiveness of the load transfer within the composite film, as film strength is determined at higher strains compared to YM values where the load transfer between matrix and reinforcing particle plays a crucial role (Mittal, 2009). As reported by Almasi et al. (2010), this improvement in the mechanical properties is mainly due to strong interfacial interactions between the matrix and the dispersed nanofiller via hydrogen bonding. Another explanation for the improvement of the TS values can be given by means of the free-volume theory. In addition to the reinforcement of the material due to sufficient interfacial interactions between matrix and filler, the nanofiller also acts as an anti-plasticizer in the WPI-based nanocomposite films. Due to this anti-plasticizer effect, a reduced free-volume and consequently more entangled polymer chains are caused whereby more stress can be supported by the films (Oymaci and Altinkaya, 2016). This may also explain the lowered extensibility of the nanocomposite films observed during the E% measurement as more entangled polymer chains are less flexible. Consequently, contrary to YM and TS, material ductility of WPI-based nanocomposites, as shown by E% measurement, generally decreases with increasing nanofiller content (Paul and Robeson, 2008). These findings are concurring with the findings of Zhou et al. (2009). They reported increased YM and TS values for WPI-TiO₂ nanocomposite films compared to WPI-films without nanofiller addition. However, they also found that these properties (YM and TS) started to decline again when a critical TiO₂ nanoparticle concentration (>1% w/w) was exceeded. Regarding elongation at break (E%) they generally observed decreasing values regardless of the nanofiller concentration. As a possible reason for the reduction of tensile properties they clearly identified inhomogeneous distribution and aggregation of TiO₂ within the WPI-matrix due to relative polarity differences between the matrix and the nanoparticles (Zhou et al., 2009). On the basis of these findings, and, as mentioned earlier, the continuously increasing tensile values (YM and TS) found during this study, we assume that the nanofiller particles used in this study are sufficiently dispersed and well exfoliated. Comparable results were also found by Azevedo et al. (2015) and Hassannia-Kolaee et al. (2016b) for denatured WPI/MMT films, and Zolfi et al. (2014) for kefiran-WPI/MMT films.

The effect of the plasticizer type could only be evaluated for glycerol-plasticized films as sorbitol-plasticized samples, caused by MMT addition, were too brittle to prepare samples for the tensile test. Therefore, only the reference samples, containing either glycerol or sorbitol, were evaluated. Table 4 shows the results of the plasticizer comparison. Generally, films plasticized with sorbitol were significantly (p ≤ 0.05) stiffer and stronger compared to glycerol-plasticized films. This observations are consistent with the findings of Shaw et al. (2002) who reported significantly lower YM and TS for WPI-films plasticized with glycerol compared to sorbitol at each plasticizer:protein ratio as well as with the findings of Chick and Ustunol (1998) who reported significantly lower TS for glycerol-plasticized films in lactic acid and rennet precipitated casein-based films. On the contrary, glycerol-plasticized films generally yielded significantly (p ≤ 0.05) higher values regarding elongation at break (E%) which is also consistent with the literature (McHugh and Krochta, 1994; Chick and Ustunol, 1998; Shaw et al., 2002).

These observations can be explained by the ability of plasticizers to improve mechanical performance in protein-based films and coatings by increasing protein chain mobility. This is achieved by reducing protein–protein interactions as highly polar plasticizer molecules interfere with other polar-binding sites, e.g., polar amino acids or water molecules (Cuq et al., 1997). In addition, due to their relative high hygroscopicity (and relative differences in hygroscopicity between glycerol and sorbitol) (Shaw et al., 2002; Verbeek and van den Berg, 2010), polyol plasticizers cause increased sorption of water molecules from the surrounding environment which leads to an increased free-volume and consequently further increased protein chain mobility (Lieberman and Gilbert, 1973). Thus, the mechanical properties of WPI-based nanocomposite films may be further improved by the choice of a more compatible plasticizer type (or mixture) and concentration (McHugh and Krochta, 1994).