Optimization of acetylated starch films from purple sweet potato: effect of glycerol, carboxymethylcellulose, and stearic acid

Purple sweet potato tubers were harvested in Charo, Michoacán, Mexico (19°44'40.8'N 101°03'37.9'W). The pieces were washed with running water and immersed in 1 l of distilled water for 5 min. Thereafter, they were peeled and scratched with a stainless-steel knife.

Native sweet potato starch was isolated according to the method of Jayakody et al [20], with some modifications taken from Ojeda [21]. One hundred grams of peeled and minced purple tuber samples were weighed and soaked in 500 ml of potassium metabisulphite solution (50 mg l⁻¹) for 1 h. The samples were ground twice with a blender, and the homogenate was vacuum-filtered on a blanket and washed with 300 ml of distilled water. The filtrate was soaked with distilled water for 24 h. The supernatant was discarded, and the sediment was washed thrice in 500 ml of 4.99 mM l⁻¹ NaOH aqueous solution until the solution became clear. The starch precipitate was neutralized with 10 ml of a 2 N HCl solution followed by filtration, the aqueous phases were discharged, and the final precipitate was filtered and washed with 500 ml of distilled water. The precipitated starch was dried in an oven at 50 °C for 24 h, ground into powder, and passed through a 100-mesh sieve. The yield was calculated through the quotient between the quantity of product obtained and the quantity of the starting sample (expressed in units of weight) and multiplied by 100.

The modified method of Vargas et al [22] was used to prepare chemical acetylated starch. Twenty grams of starch and 3.5 g of sodium sulfate were dissolved in 40 ml of distilled water at 25 °C and stirred at 800 rpm. Three drops of a sodium hydroxide solution (50% w w⁻¹) were used to adjust the pH to 8–8.4. Acetic anhydride was added to the solution using a concentration of 10% based on the concentration of starch. The solution was continuously stirred (800 rpm) for 5 h, maintaining the pH value at 8. The slurry was adjusted to pH 7.0 with hydrochloric acid solution (1 M) and centrifuged (CRM globe, Centrificient) at 2000 rpm for 15 min. The precipitate was washed four times with distilled water, oven-dried (ECOSHEL, 9053 A) at 50 °C for 16 h, ground into powder and passed through a 100-mesh sieve. The percentage of acetylation (% acetyl) and the degree of substitution were determined by back titration according to the JECFA (2001) [23] methodology. 1 g of modified starch was placed in 50 ml of distilled water and 3 drops of phenolphthalein, the solution being neutralized to a slightly pink color with 3 drops of 0.1 N NaOH. Then 25 ml of 0.45 N NaOH was added, and the solution was vigorously stirred (600 rpm) for 30 min. After time, the samples were titrated with 0.2 N HCl. Likewise, a blank was prepared using native starch instead of acetylated starch.

The percentage of acetyl represents the percent by weight of acetyl groups in the starch on a dry basis and was calculated as follows:

$$\begin{eqnarray*}&&Acetyl\,\left( \% \right)=\displaystyle \frac{\left[\left(Blank\,\left(ml\right)-Sample\,\left(ml\right)\,\times \,Normality\,of\,HCl\right)\,\times \,0.043\,\times \,100\right]}{Sample\,weight\,\left(g\right)}\end{eqnarray*}$$

The degree of substitution is defined as the average number of sites per glucose unit that has a substituent group and was calculated using the following equation:

$$\begin{eqnarray*}&&Degree\,of\,substitution\,\left(DS\right)=\displaystyle \frac{\left(162\,\times \,Acetyl\,\left( \% \right.\right)}{\left[4300-\left(42\,\times \,Acetyl\,\right.\left( \% \right)\right]}\end{eqnarray*}$$

where 0.043 are the milliequivalents of the acetyl group (CH₃-C = 0), 162 is the molecular weight of anhydrous glucose, 4300 is the multiplication of the molecular weight of the acetyl group (CH₃-C = O) by 100, and 42 is the molecular weight of the acetyl group minus 1.

The following procedure was used to produce the films: Fifty five mL deionized water was heated in a beaker up to 70 °C and the corresponding amount of stearic acid (AE) was added after melting the fatty acid by sonication (SONICS vibracell, VC505) for 5 min at 40% amplitude to generate an emulsion. Subsequently, it was again placed in agitation and heating while 5 g of acetylated purple sweet potato starch was added, and after the temperature reached 85 °C, it was held for 20 min. Separately, fifty-five ml of deionized water was placed in a beaker on a heating plate and heated up to 70 °C with agitation at 800 rpm and the corresponding amount of carboxymethylcellulose (CMC) was added immediately. Once dissolved, it was sonicated in an ultrasonic bath for 5 min to remove bubbles. Once the two solutions were prepared, they were mixed in a beaker on a grill with stirring and heating for 10 min. The corresponding amount of glycerol was added and held for 5 min more, and finally the solution was filtered through gauze. The film-forming solution was poured into 20 × 20 × 0.3 cm acrylic plates; the plates were placed in an oscillating shaker for 20 min to homogenize the gel and finally dried in a drying oven with air circulation for 24 h at 35 °C. The obtained films were detached and characterized.

Statgraphics Centurion XVI software was used to design a response surface with central composite experimental design star points to optimize the combination of factors affecting the film formation. The composite central design star points generated 18 runs, with four repetitions at the central point, and used process factors to evaluate the effect of independent variables (glycerol concentration (G): 0.3, 0.35, and 0.40 g g⁻¹ starch; CMC: 0.5, 0.75, and 1 g/5 g starch; stearic acid (SA) 0.025, 0.05, and 0.075 g/5 g starch) on response variables (solubility, swelling, opacity, luminosity, tensile strength, elongation, water vapor permeability, and water activity). The second-degree polynomial equation generated by the design was as follows:

$$\begin{eqnarray*}&&\begin{array}{l}Y={\beta }_{0}+{\beta }_{1}A+{\beta }_{2}B+{\beta }_{3}C+{\beta }_{12}AB+{\beta }_{13}AC\\ \,\,+\,{\beta }_{23}BC+{\beta }_{11}{A}^{2}+{\beta }_{22}{B}^{2}+{\beta }_{33}{C}^{2}\end{array}\end{eqnarray*}$$

where Y is the predicted response; ${\beta }_{0}$ is a constant; ${\beta }_{1},{\beta }_{2}\,and\,{\beta }_{3}$ are the linear coefficients; ${\beta }_{12,\,}{\beta }_{13}\,and\,{\beta }_{23}$ are the coefficients of interaction among the three factors under investigation; ${\beta }_{11},\,{\beta }_{22}\,and\,{\beta }_{33}$ are the quadratic coefficients; and A, B, and C are the process codes for glycerol, carboxymethylcellulose, and stearic acid, respectively. Analysis of variance (ANOVA) of the regression at a level of significance of $\alpha =0.05$ was performed to verify the validity of the model.

2.6.1. Thickness

2.6.2. Luminosity and opacity

The luminosity was determined by averaging four measurements, two in the center and two at the edges, using a colorimeter (BYK-Gardner GmbH). The films were placed on a standard board and CIE-Lab scales were used to measure their luminosity. The opacity of the films was calculated by dividing the absorbance value at 600 nm by the film thickness (mm) [24]. The sample was cut into a rectangular piece (3 × 1 cm) and placed in the spectrophotometer cell. The measurements were made with air as a reference. All determinations were made in triplicate and the opacity was calculated according to the following equation:

$$\begin{eqnarray*}&&Opacity\left({A}_{600}m{m}^{-1}\right)=\displaystyle \frac{{A}_{600}}{d}\end{eqnarray*}$$

where A₆₀₀ is the absorbance of the film measured at 600 nm, and d is the film thickness in mm.

2.6.3. Solubility and degree of film swelling

Water solubility (WS) was defined as the proportion of water-soluble dry matter in the film that dissolved after immersion in distilled water [25]. Samples were prepared in triplicate by cutting the sample into 2 cm² square pieces, followed by drying in a hot air oven at 105 °C for 24 h. After the first dry film was accurately weighed to record the initial weight, the samples were immersed in 50 ml of distilled water and kept in slow and constant stirring at 90 rpm in an orbital shaker (Scientific & Instruments, Inc. CVP2000P) at 25 °C for 24 h. The remaining pieces of the film were then removed and dried at 105 °C to obtain a constant weight (final dry weight). The percentage of total solubility of the films was calculated using the following equation:

$$\begin{eqnarray*} &&\% \,Solubility=\displaystyle \frac{{\rm{wi}}-{\rm{wf}}}{{\rm{wi}}}\times 100\end{eqnarray*}$$

where wi and wf represent the initial and final dry weights, respectively.

The swelling of the films was measured by placing 2 cm² pieces at constant weight in containers with 50 ml of distilled water for 24 h at 25 °C after weighing the films (W1). The films were then superficially dried with filter papers and the wet film (W2) was weighed. The degree of swelling was calculated through the following equation:

$$\begin{eqnarray*}&&Degree\,of\,swelling\,\left( \% \right)=\displaystyle \frac{W2-W1}{W1}\,\times \,100\end{eqnarray*}$$

2.6.4. Permeability to water vapor

The water vapor permeability of film samples was determined according to the official method ASTM D885–02 [26], with some modifications. Cylindrical plastic bottles with an internal diameter of 25.4 mm, containing 5 g of calcium chloride, were covered with the films. The test vessels were placed in a desiccator maintained at 25 °C and relative humidity of 55%. The changes in the weight of the cups were controlled for 24 h until the stable state (straight line) was reached. The linear regression graph of weight loss versus time determined the slope. Water vapor transmission rate (WVTR) was defined as the slope (g h⁻¹) divided by the transfer area (m²). The WVP (g m⁻¹ s⁻¹ MPa) was calculated as follows:

$$\begin{eqnarray*}&&{\rm{WVP}}=\displaystyle \frac{{\rm{WVPRT}}-{\rm{tm}}}{{\rm{\Delta }}p}\,\times \,100\end{eqnarray*}$$

where WVTR is the rate of transmission of water vapor (g m⁻² s⁻¹) through the film, tm the average thickness of the film, and Δp the difference of partial water vapor pressure (MPa) across the two sides of the film. All measurements were made on three replicates of each film.

2.6.5. Physical and mechanical properties of films

The texture analyzer (Brookfield, CT3 25 K) was used to determine the tensile strength, elongation at break, and maximum force, according to the ASTM D 882-02 standard method [27]. Film strips of 10 mm width and 50 mm length were cut; the initial gripping distance was 30 mm, and the speed of the crosshead 0.50 mm sec⁻¹. Six replicas of each film were tested. Tensile strength was calculated by dividing the maximum breaking force of the film by its cross-section area. Elongation at rupture was determined by dividing the difference in the final distance travelled to rupture and the initial separation distance by the initial separation distance multiplied by 100.

$$\begin{eqnarray*}&&TS=\displaystyle \frac{Fm}{A}\,\times \,100\end{eqnarray*}$$

where TS = tensile strength (MPa); Fm = maximum force at break (N), and A = area of film cross-section (m²).

$$\begin{eqnarray*} &&\% \,Elongation=\displaystyle \frac{\left(do-df\right)}{do}\,\times \,100\end{eqnarray*}$$

where E = elongation (%); do = initial distance of separation between claws (cm); and df = distance to rupture (separation between the claws at the time of rupture) (cm).

The results of the different response variables used to establish the optimal conditions to produce films from acetylated purple sweet potato starch are shown in table 1.

The second-order polynomial models in table 2 show the regression coefficients and the influence of the variables on the physical and mechanical characteristics of the produced films. All the films were smooth, clear, and transparent (figure 1).

Zoom In Zoom Out Reset image size

The linear terms of the concentrations of G (A), CMC (B), and stearic acid (SA) (C) had significant effects (p ≤ 0.05) and the Fisher F values were 26.49, 6.69, and 21.31, respectively, implying that the model was significant. The coefficient of determination (R²) was 0.89 and the adjusted R² was 0.77 for G response, indicating a good alignment between the observed and predicted values in the quadratic equation. The Durbin-Watson statistic (DW) is another value that shows the presence of autocorrelation between errors [31]. A DW value below 2 indicates a positive correlation and above 2 indicates a negative correlation [32]. If the DW value is close to 2, it implies a good fit of the model. The DW for elongation was 1.56686. The three-dimensional response surface graph (figure 2(a)) indicates an increase in the percentage of elongation of the films by increasing the glycerol content and decreasing the concentration of CMC. It was also observed that by increasing the concentration of SA, the elongation decreased.

Zoom In Zoom Out Reset image size

The methodology developed to optimize the response surface led to the location of the stationary point, which represents the maximum point. The values of factors in which the highest percentage of elongation was presented were glycerol 0.43 g g⁻¹ starch, CMC 0.45 g/5 g starch, and SA 0.007 g/5 g starch, which led to an elongation of 118% (table 3).

The ANOVA indicated that the linear term glycerol concentration (A), the terms of the G-CMC (AB) interactions, G-SA (AC), and the quadratic term CMC (BB) had a significant effect (p $\leqslant \,$0.05). Fisher's F values of 15.58, 7.67, 7.22, and 5.69 respectively, implied that the model was significant. R² was 0.84 and the adjusted R² of 0.67, the statistical term DW (1.73) was close to 2, which indicated the model to be a good fit. The three-dimensional surface response diagram in figure 2(b) shows that high concentrations of G and SA, and low CMC, generate films with higher opacity values. The stationary point corresponding to the minimum opacity value of 0 (table 3) was reached with the amounts of G 0.26 g g⁻¹ starch, CMC 0.49 g/5 g, and SA 0.09 g/5 g starch.

The glycerol concentration (linear term A) was found to be significant (p ≤ 0.05) with reference to the permeability factor and showed a value of 13.32 in Fisher's F. The regression model had a moderate correlation value (R²) of 0.76 and adjusted R² of 0.50. High G and low CMC concentrations generated films with higher water vapor permeability values, as can be seen in the three-dimensional response surface diagram in figure 2(c). The optimal conditions that minimized water vapor permeability were G 0.26 g g⁻¹ starch, CMC 0.33 g/5 g, and SA 0.007 g/5 g starch. These process conditions were necessary to reach the stationary or minimum permeability point, which corresponded to a value of 0.0031 g msMPa⁻¹ (table 3).

In the production of films, the activity of water, luminosity, swelling, stress force, and solubility indicated that none of the concentrations of these variables had a significant effect ($p\leqslant 0.05$), except for the quadratic term of the concentration of CMC (BB) in the water activity and the linear term of the concentration of SA (C) in the luminosity, with values of Fisher F 8.36 and 11.38, respectively. R² values were 71.75, 69.20, 67.50, 44.91, and 58.78%, respectively, and adjusted R² values were 39.98, 34.55, 30.94, 26.77, and 12.47%, respectively, in which the probability of a good alignment between observed and predicted values in the quadratic equation was lower. The optimal values and targets to achieve the optimization are shown in table 3. However, only trends were observed in the factors analyzed (figures 2(d)–(h)). As the G concentration increased the luminosity, the swelling and solubility of the films also increased, and the tensile strength decreased. At high concentrations of CMC, the tensile strength increased and reduced the luminosity, swelling, and solubility. The presence of SA decreased the water activity and swelling, while increasing luminosity. Tables 4 and 5 summarize the experimental results (X) and those predicted by the equation for each of the answers using the complete equation (Y). Table 6 shows the correlation coefficients (R) between the response variables analyzed. A moderate correlation was found between the elongation (%) and the swelling of the films and between the elongation (%) and the strength of stress, indicating that the ease with which a film tends to stretch depends to some extent on its water content. An inverse behavior between elongation and stress force was also observed, which coincided with the results of another study [33] which evaluated these factors in films based on cassava starch.

After analyzing the surface response model, a good correlation was observed for the prediction relating to the factors of opacity, elongation, and permeability to water vapor. However, as the minimum and maximum opacity values obtained did not greatly affect the appearance of the films, the desirability function in MSR was tested to demonstrate simultaneously that the best responses (elongation and permeability) were affected by the combination of experimental parameters, which led to the acquisition of optimal conditions based on the established criteria of minimum and maximum levels of the selected response variables. The general approach of the desirability function ranges from 0 (completely undesirable response) to 1 (fully desired response) [34]. In figure 3, the three-dimensional response surface implies that concentrations of G and CMC near the upper limit generate a general desirability value of 0.92 for simultaneous optimization.

Zoom In Zoom Out Reset image size

From the response surface graphs and the data, a simultaneous optimization was achieved using a concentration of glycerol 0.30 g g⁻¹ starch, CMC 0.32 g/5 g starch, and SA 0.007 g/5 g starch. This desirability model predicted 99.62% elongation and 0.005 g msMPa⁻¹ (table 7) permeability to water vapor. Additionally, the predicted optimal values for response variables were compared with the predicted values for simultaneous optimization. For the validation of the results predicted by the simultaneous optimization, verification of the experiments was performed in triplicate, which resulted in 91.846% elongation and 0.005 g msMPa⁻¹. The good correlation between these results confirmed the suitability of the model to reflect the expected simultaneous optimization since the difference between the predicted and experimental values was less than 7.80% for elongation and 1.81% for WVP. The results were related to the data from the optimization analysis using desirability functions, which indicated that the composite core design star points with the desirability function for the two selected responses can effectively optimize the characterization parameters.

Table 8 shows the complete characterization of the optimized film, and that the modification by acetylation and incorporation of CMC, SA, and G produce films with a higher percentage of elongation than films obtained with only other sources of starch such as potato with 57.4% [35], and higher than those presented with sweet potato starch and G with 11.24% [36]. This increase in elongation is because the modification of starch allows the masking of some of the OH groups of glucose with acetyl groups, thus increasing the flexibility between polysaccharide chains. There was also an increase in the tensile force, attributable to the formation of intermolecular interactions between the OH groups still present in the starch and the carboxyl group of the CMC, during the processing and drying of the films. The original hydrogen bonds could be replaced by new ones formed between the OH groups in the starch molecules and the hydroxyl and carboxyl groups in the CMC. This intermolecular interaction between starch and CMC results in a more compact molecular structure, thus increasing tensile strength [33].

Optimization has been reported through the response surface of films focused on food coatings from various starch sources. However, there is little literature found to date evaluating the properties obtained with acetylated starch of purple sweet potato. Indrianti et al [37] produced edible films prepared from native sweet potato starch (SPS) or modified sweet potato starch by heat-moisture treatment (HMT). The results obtained were higher in the elongation percentages, similar in tensile strength and water vapor permeability, and lower in solubility. Mohd and Muhamad [38] used the central design composed of 2 factors with a factorial of 10, as well as 2 central points, studying purple sweet potato starch to which were added glycerol as plasticizer and carrageenan kappa as gelling agent. Thakur et al [39] reported the optimization of a film made from pea starch and chitosan. Paramanantham and Thottiam [40] used tapioca starch with glycerol and acetic acid in the development of a biodegradable polymer film and applied the Box-Behnken design with four factors on three levels to study the individual and interactive effects of the film composition and the Barrier property of the developed film. Monteiro et al [41] applied the statistical design of the experiments, combined with regression techniques, to optimize the conditions for improving the mechanical properties of starch films. These studies concluded that the incorporation of materials in polymer matrices and starch source improves the physical, mechanical, and optical properties of films. These basic data on the influence of variables on the properties of starch-based films are important for assessing the applicability of films in the food and pharmaceutical industries. Therefore, the formulation used in our study can obtain a low permeability to water vapor (0.55 g msMPa⁻¹), high elongation percentage, and low solubility, which allows the film to be considered for food packaging, in addition to being converted into a functional edible film incorporating compounds such as antioxidants, reinforcements, and sensors with favorable results, reflecting a positive perspective on the marketing of starch food packaging in the food industry.

3.7.1. Concentration of glycerol

3.7.2. Concentration of CMC

3.7.3. Concentration of stearic acid