The polymeric materials employed in this study were commercially available cellulose esters such as cellulose triacetate (CTA), cellulose diacetate (CDA) and cellulose acetate propionate (CAP). CTA was produced by Acros Organics, while both CDA and CAP were produced by Eastman Chemical. The molecular characteristics are summarized in Table 1. As seen in the table, the nominal values of degree of substitution (DS) of CDA and CTA are 2.41 and 2.96, respectively. The molecular weights were evaluated using a gel permeation chromatograph (Tosoh, HLC-8020) with TSK-GEL^® GMHXL.

The polymeric films used in this study were prepared using a solution casting method since melt processing is not applicable for CTA due to the severe thermal degradation [2,21]. Initially, CTA and CAP were dissolved into chloroform, while CDA in tetrahydrofuran, and stirred for 24-hours at room temperature. The solution was then poured onto a metal plate and left at room temperature for 48-hours to allow the solvent to evaporate slowly in order to obtain a smooth polymeric film. It was then cut into rectangular specimens of 10 mm in width and 20 mm in length. The thickness of the films obtained was about 150 μm.

Subsequently, the films were hot-drawn using a uniaxial tensile machine with temperature controller (UBM, DVE-4) to prepare uniaxial oriented films. All samples were drawn at temperatures where the tensile storage modulus E′ is 500 MPa at 10 Hz, upon considering their drawability. The initial distance between the clamps was 10 mm and the stretching rate used was 0.5 mm/s. All samples were stretched at a draw ratio of 1.5. Then, the drawn samples were quenched by cold air blowing to avoid relaxation of the molecular orientation. They were kept in a humidic chamber (Yamato, IG420) at 25 °C and 50%RH for one day. After that, the samples were removed from the humidic chamber and put in a vacuum oven at 70 °C for a day before finally being immersed in water at 25 °C for one day. The orientation birefringence and moisture content of the samples were measured and recorded after each process.

The temperature dependence of oscillatory tensile moduli in the solid state was measured from −50 °C to 200 °C by a dynamic mechanical analyzer (UBM, E-4000) using rectangular specimens of 5 mm in width and 20 mm in length. The frequency and heating rate used were 10 Hz and 2 °C/min, respectively.

Moisture content was measured using a moisturemeter (Mitsubishi Chemical Corporation, CA-200) using the Karl Fischer method in which the Karl Fischer reagent, which consists of iodine, sulfur dioxide, a base and a solvent, reacts quantitatively and selectively with water [22].

The birefringence of the drawn films was measured at room temperature by an optical birefringence analyzer (Oji Scientific Instruments Inc., KOBRA-WPR) as a function of wavelength by changing color filters.

Attenuated Total Reflection (ATR) measurements using an infrared absorption spectrometer (Perkin Elmer, Spectrum 100) were performed to estimate the interaction of cellulose esters with water molecules upon exposure to three ambiences with different humidities.

Oscillatory tensile moduli such as storage modulus E′ and loss modulus E″ of CTA, CDA and CAP have been measured to determine the stretching temperature. Figure 1 shows the temperature dependence of the oscillatory tensile modulus at 10 Hz for CTA, CDA and CAP. Both CTA and CDA show higher storage modulus than CAP in the glassy region, suggesting that chain packing density of CAP is lower than those of the others. Furthermore, the storage modulus in CTA shows a gradual decrease compared to a sharp drop observed in CDA and CAP. The gradual decrease observed in CTA is due to the presence of crystallites, which affects the mobility of CTA chains. Consequently, it leads to a broad distribution of relaxation time as demonstrated in Figure 1. On the other hand, both CDA and CAP are not affected by crystallites as they are fully amorphous [8,12].

The sharp drop in the E′ of CAP corresponds to the narrow peak of E″, around 144 °C. In contrast, both CTA and CDA show a broad peak, located at almost 40 ° C higher temperature than that of CAP. According to Ohno et al. [23], T_g of cellulose ester derivatives becomes lower with increasing the side-chain length, and with increasing degree of substitution (DS). However, the T_g which can be defined by the peak temperature of E″, of CTA in this study is found to be located almost at the same temperature with that of CDA. This is presumably due to the presence of crystallites that have contributed to the appearance of the loss modulus peak at a somewhat higher temperature position owing to the restriction of molecular motion. Moreover, the results also indicate that higher acyl groups such as propionyl and butyryl contribute toward the low T_g observed, instead of acetyl.

The wavelength dispersion of orientation birefringence of stretched CTA is shown in Figure 2. Numerals in the figure represent the ambient humidity. CTA shows negative birefringence that decreases with increasing wavelength, i.e., ordinary dispersion which is a typical behavior for most polymers [5,6]. As demonstrated in our previous papers [8,9], the orientation birefringence in cellulose esters is determined by polarizability anisotropy and orientation of the side groups, e.g., acetyl, propionyl and hydroxyl groups. The result in Figure 2 indicates that the acetyl group in stretched cellulose esters contributes to negative birefringence with ordinary dispersion. Moreover, it has been found that the amount of crystallites is not an important factor in determining the value of orientation birefringence in cellulose esters, as discussed in our previous paper [12].

On the other hand, CDA produces positive orientation birefringence that increases with increasing wavelength, as shown in Figure 3. Considering that CDA contains the hydroxyl group as well as the acetyl group, the hydroxyl contribution to positive birefringence must be significantly large enough to offset the negative birefringence associated with the acetyl group [12]. Combination of (i) positive birefringence with ordinary dispersion and weak wavelength dependence from hydroxyl group with (ii) negative birefringence with ordinary dispersion and strong wavelength dependence from the acetyl group, results in positive birefringence with extraordinary wavelength dispersion.

The orientation birefringence of CAP is shown in Figure 4. Similar to CDA, CAP shows positive birefringence with extraordinary dispersion. However, the magnitude of birefringence in CAP is small compared to that of CDA. The result suggests that the positive birefringence contribution associated with the propionyl group is smaller than that with the hydroxyl one.

The moisture contents for CTA, CDA and CAP after exposure to three different levels of humidity, i.e., 0%, 50% and 100%, are shown in Figure 5. The samples were prepared by (i) drying in a vacuum oven at 70 °C for 24 hours, (ii) storing in a humid chamber at 50%RH, 25 ° C for 24 hours, and (iii) immersion in distilled water at 25 °C for 24 hours respectively.

As demonstrated in Figure 5, the value of moisture content increases with increasing ambient humidity. The increase is due to the interactions between water molecules and the carbonyl groups or the hydroxyl groups in the cellulose esters via weak hydrogen bonds. CDA shows the most significant water regain for 24 hours water immersion, more than twofold increase from the value obtained in 50% humidity. On the contrary, CAP shows the lowest moisture content at any conditions. The high water regain of CDA suggests that the unsubstituted hydroxyl groups are hydrophilic compared to the acetyl or propionyl groups, allowing the formation of more hydrogen bonds in cellulose esters.

Upon drying in the vacuum oven (0% humidity) for 24 hours, the orientation birefringence of CDA and CAP increases. Meanwhile, under the same condition, the negative birefringence of CTA decreases in magnitude. However, it should be noted that, in terms of the direction of polarizability anisotropy, the change observed in CTA corresponds to what is observed in the CDA and CAP. In other words, the orientation birefringence increases to the positive direction. Using the relation given in Equation (2), it can be concluded that exposure to low humidity ambience increases the polarizability anisotropy parallel to the stretching direction in cellulose esters.

After 24 hours of immersion in water, the orientation birefringence of CDA and CAP decreases while that of CTA increases in magnitude. The results show that the negativity of the birefringence, which is associated to the polarizability anisotropy perpendicular to the stretching direction, increases with increasing water content.

The relation between orientation birefringence at 589 nm and moisture content for CTA, CDA and CAP is shown in Figure 6. In more hygroscopic cellulose esters such as CTA and CDA, the orientation birefringence decreases linearly with moisture content. However, the more hydrophobic CAP does not show the similar characteristic. As the amount of moisture regain and the change in birefringence in CAP are miniscule even after 24 hours water immersion, the result could have been affected by experimental errors. In any case, the results suggest that moisture content has a greater effect on the polarizability anisotropy of the acetyl than propionyl groups.

IR spectrum of CDA exposed to 0% humidity is exemplified in Figure 7. The anhydrous CDA shows an absorption band ascribed to OH stretching vibrations, centered at 3,492 cm⁻¹. The band is more obvious in CDA than in CTA because of a large amount of hydroxyls available in CDA. Furthermore, an absorption peak ascribed to the stretching vibrations of the carbonyl bond is detected at 1,736 cm⁻¹. Moreover, bands with strong intensity are also found at 1,214 cm⁻¹ and 1,031 cm⁻¹, both correspond to the stretching vibrations of C–O single bonds. A weak band caused by the vibrations of methyl groups can be observed at 1,367 cm⁻¹.

ATR-FTIR spectra (3,000–4,000 cm⁻¹) for CDA and CTA exposed to 0% and 100% humidity are shown in Figure 8. Upon hydration, the band in the 3,200–3,600 cm⁻¹ range which is associated with the OH stretching vibrations shows an increase in intensity accompanied with a widening of the band. This is due to the overlapping of the band associated with the hydroxyl groups in cellulose esters with the bands associated with OH stretching vibrations of the sorbed water molecules. The increase of the absorption peak is more prominent in CDA than that in CTA, while CAP is found to show the smallest increase, corresponding to the amount of moisture regain shown in Figure 5. For all ATR-FTIR measurements, the peak intensity for 50% humidity is almost at an intermediate value between those of 0% and 100% humidity.

In cellulose esters, water molecules could form hydrogen bonds with the hydroxyl and the carbonyl groups [20,23]. Considering that CAP and CTA contain only a small amount of the hydroxyl group, hydrogen bonding between carbonyl groups and water molecules would mainly be responsible for the decrease in orientation birefringence upon hydration. Therefore, the formation of hydrogen bonds between cellulose esters and water molecules is determined by evaluating the band correspond to the stretching vibration of the carbonyl groups (1,660–1,760 cm⁻¹). The formation of hydrogen bonds is demonstrated by the shift of the carbonyl band to lower wavenumbers as has been reported in previous studies [16,17].

The IR spectra in the 1,600–1,900 cm⁻¹ range for CDA and CTA exposed to 0% and 100% humidity are shown in Figure 9. In the hydrated CDA, the absorption peak ascribed to the carbonyl groups shifts from 1,736 to 1,732 cm⁻¹, while for the hydrated CTA, a similar displacement of the particular peak, from 1,737 to 1,734 cm⁻¹ is observed. CAP shows the smallest displacement of the peak, from 1,739 to 1,737 cm⁻¹, reflecting the hydrophobic nature of the propionyl groups. These results correspond to the amount of moisture regain observed in Figure 5.

As evident in Figure 6, the signs of the orientation birefringence of cellulose esters show a consistent change with moisture content, while the same does not hold true for the absolute values. The absolute value of orientation birefringence in CDA and CAP decreases with the increase of water content, while in CTA it increases. In particular, the increase in the absolute value of birefringence in CTA indicates that the change in orientation birefringence is not due to the disruption of the orientation of the side groups by absorbed water molecules, but rather by the increase of the polarizability anisotropy perpendicular to the stretching direction in the hydrated cellulose esters due to the formation of hydrogen bonding.

Moreover, a weak absorption band appears at 1,639 and 1,637 cm⁻¹ in CDA and CTA spectra respectively. These bands are typical of bending vibrations of OH. The band is more obvious in CDA than in CTA, which corresponds to the higher moisture regain of CDA.