3.1.3. Microscopy

Macro- and microscopically, no phase separation was detected in the concentrated nanoparticle dispersions after 16 weeks at 23 ◦C. The morphology of the raw materials is shown in Figure 1a,b. Cotton linters have a fibrillar structure, whereas corn starch has a granular shape. Sulfuric acid primarily degrades amorphous regions of the polysaccharides. Upon freeze-drying, agglomeration of nanoparticles promoted by the presence of residual salt occurs (Figure 1c,d). Additionally, microscale hydrolyzed residues with a high aspect ratio can be found in both products.

**Figure 1.** SEM images of (**a**) cotton linters, and (**b**) corn starch and the freeze-dried hydrolyzed products from (**c**) cotton linters, and (**d**) corn starch after hydrolysis and neutralization.

#### *3.2. Properties of Nanoparticle Coatings*

#### 3.2.1. Surface Tension

Bare and untreated PLA had a surface tension of <34 mN m−1, which is slightly below reported values (36–38 mN m−<sup>1</sup> [50]). Corona-treating the substrate elevated the surface tension to 42 mN m<sup>−</sup>1.

#### 3.2.2. Dry Coating Thickness and Nanoparticle Loading

PLA and paper had thicknesses of 26.8 ± 0.1 μm and 62.5 ± 2.9 μm, respectively. Considering the concentration and the density of both the nanoparticles of 1.5 g cm−<sup>3</sup> and Na2SO4 of 2.7 g cm−3, the thickness of a single coated layer for a wet film thickness of 51 μm was estimated to be 2.5 μm for CNC and 2.7 μm for SNP. The actual coating thicknesses on PLA were 2.6 ± 0.8 μm for CNC and 5.9 ± 0.9 μm for SNP. Paper substrates were double-coated with both CNC and SNP, resulting in thicknesses of 9.5 ± 0.4 μm and 12.0 ± 1.0 μm, respectively.

The nanoparticle loading was calculated from the ratio of the mass of nanoparticles in the product *m*np and the dry mass *m*dry and was 75.6 ± 1.7 wt % in the CNC coating and 92.2 ± 2.6 wt % in the SNP coating.

#### 3.2.3. Surface and Optical Properties

Coating PLA with CNC did not yield a uniform film (Figure 2a). Hydrolyzed residues and agglomerates were randomly distributed over the substrate surface. Hydrolyzed residues and agglomerates were found for SNP coatings as well, accompanied by fine fissures in the coating layer (Figure 2b).

**Figure 2.** PLA coated with (**a**) CNC and (**b**) SNP in reflected bright-field microscopy.

The top side of the paper substrate was microscopically uniform (Figure 3a). The subjacent fibrous structure was visible via reflected light microscopy. While the top side appeared microscopically dense, the back side of the paper substrate showed pores with diameters in the micrometer range. Coating paper with CNC (Figure 3b) and SNP (Figure 3c) involved the deposition of hydrolyzed residues and agglomerates onto the surface, similar to the coatings on PLA. SEM imaging of the coated paper substrate shows the irregular surface topography caused by these residues (Figure 3e,f). Double-coating paper with SNP caused a more distinct topography.

**Figure 3.** Paper (**a**,**d**) coated with CNC (**b**,**e**) or SNP (**c**,**f**) via reflected bright-field microscopy (**a**–**c**) and by secondary electron imaging via SEM (**d**–**f**).

The light transmittance of bare PLA of 91.4% ± 1.3% at a wavelength of 550 nm was reduced by the application of nanoparticle coatings. The CNC coating reduced the absolute light transmittance by 10.3% ± 1.6%, whereas SNP reduced the absolute light transmittance by 34.2% ± 1.7%.

#### 3.2.4. Barrier Properties

Coating PLA with a single layer of CNC decreased the OP from 514.6 ± 3.8 cm<sup>3</sup> (STP) m−<sup>2</sup> <sup>d</sup>−<sup>1</sup> bar−<sup>1</sup> to 129.7 ± 8.7 cm3 (STP) m−<sup>2</sup> <sup>d</sup>−<sup>1</sup> bar−<sup>1</sup> at 50% r.h. (74.8%) (Figure 4a). A decrease to an OP of 110.1 ± 14.2 cm3 (STP) m−<sup>2</sup> <sup>d</sup>−<sup>1</sup> bar−<sup>1</sup> (78.6%) was observed for a single coating layer of SNP. Considering the coating thickness, the *OP Q*<sup>100</sup> of 4.7 ± 0.4 cm3 (STP) 100 <sup>μ</sup>m m−<sup>2</sup> <sup>d</sup>−<sup>1</sup> bar−<sup>1</sup> and

8.5 ± 1.4 cm3 (STP) 100 <sup>μ</sup>m m−<sup>2</sup> <sup>d</sup>−<sup>1</sup> bar−<sup>1</sup> for CNC and SNP resulted, emphasizing the noticeable barrier performance of CNC against oxygen compared to SNP (Figure 4b).

**Figure 4.** (**a**) The measured OTR of bare PLA and PLA substrate coated with CNC and SNP; (**b**) the normalized OTR to a layer thickness of 100 μm (*OTR Q*100).

The nanoparticle coatings did not improve the water vapor barrier of coated PLA. The *WVP Q*<sup>100</sup> of bare PLA substrate of 76.1 ± 3.1 g (STP) 100 <sup>μ</sup>m m−<sup>2</sup> <sup>d</sup>−<sup>1</sup> remained almost constant for a coating with CNC (80.3 ± 4.6 g (STP) 100 <sup>μ</sup>m m−<sup>2</sup> <sup>d</sup><sup>−</sup>1; 85→0% r.h.) and SNP (81.4 ± 1.8 g (STP) 100 <sup>μ</sup>m m−<sup>2</sup> <sup>d</sup><sup>−</sup>1; 85→0% r.h.). The slight increase is explicable by water adsorption due to the hygroscopic character of both the coatings and the substrate in combination with the gravimetric measurement method.

Paper substrates double-coated with either CNC or SNP did not pass the pinhole test and were therefore excluded from the determination of barrier properties.

#### *3.3. Properties of Cast Films*

CNC and SNP were incorporated in hydrophilic starch matrices at different concentrations *c*filler by solution casting. The target thickness was 50 μm. All experiments were repeated at least five times.

### 3.3.1. Microscopy and Optical Properties

Plastification at 70 ◦C for 1 h did not completely degrade the granular structure of corn starch. Swelling induced an increase of the grain size in the pure starch-glycerol film and ghost remnants were recognizable [51] (Figure 5a). Accordingly, the film surface displays the topography of the shells of the native starch granules (Figure 5d). Adding CNC (Figure 5b,e) or SNP (Figure 5c,f) in concentrations of 0 ≤ *c* ≤ 9 wt % did not alter the microstructure of the film. Agglomerates or microscale residues from the hydrolyzed cellulose product were visible in both reflected bright-field microscopy and via SEM imaging. The visible accumulation suggests their segregation from the starch matrix during drying. Due to the similar appearance of the SNP and the starch matrix, no hydrolyzed starch residues were recognizable in these nanocomposites.

**Figure 5.** Pure starch-glycerol film (**a**,**d**), starch-CNC nanocomposite (**b**,**e**), and starch-SNP nanocomposite (**c**,**f**). The shown nanocomposites had a filler content of 3 wt %. (**a**–**c**) Reflected bright-field microscopy and (**d**–**f**) secondary electron imaging (SEM).

The light transmittance of a starch-glycerol film *T*starch with a thickness *d* of 50 μm was 83.5 ± 2.1% (13.38 × <sup>10</sup>−<sup>3</sup> ± 2.3 × <sup>10</sup>−<sup>3</sup> <sup>μ</sup>m−1) at 550 nm. For better comparability, the light transmittance *<sup>T</sup>* was normalized with respect to *d*. The addition of CNC and SNP reduced the light transmittance with increasing nanoparticle concentration *c*nanoparticles (Figure 6a). This effect was more pronounced for the addition of SNP. Accordingly, the extinction coefficient ε decreased with increasing filler content and was overall higher for SNP nanocomposites (Figure 6b). The high uncertainties prevalent at low *c*nanoparticles arose from the strong relative weighting of variable film thicknesses.

**Figure 6.** (**a**) The normalized transmittance of CNC and SNP nanocomposites and (**b**) the extinction coefficient at different nanoparticle concentrations.

#### 3.3.2. Barrier Properties

A slight improvement of the *WVP Q*<sup>100</sup> of cast starch-glycerol films was achieved by adding CNC or SNP (Table 3). However, all measured values lie within the 95% confidence interval of the *WVP Q*<sup>100</sup> of 47.3 ± 20.6 g (STP) 100 <sup>μ</sup>m m−<sup>2</sup> <sup>d</sup>−<sup>1</sup> (85→0% r.h.) of the pure starch-glycerol film.


**Table 3.** Normalized water vapor permeability of starch nanocomposites with different amounts of CNC or SNP.
