Providing new formulation for white compound chocolate based on mixture of soy flour, sesame paste, and emulsifier: An optimization study using response surface methodology

Abstract This study was aimed at evaluating the suitability of sesame paste as an ingredient in white compound chocolate using response surface methodology. A D‐optimal combined mixture‐process design with three mixture components, sesame paste (15%–30% w/w), soy flour (0%–15% w/w), and milk powder (0%–15% w/w) under variable amount of emulsifier was used to optimize textural (hardness, cohesiveness, and adhesive force) and thermal (T onset and T m) properties of white compound chocolate containing sesame paste. The results showed that the linear effect of all the mixture components was significant (p < .05) on the responses. Applying a desirability function method, the optimum proportion of mixture components, and emulsifier level were as follows: sesame paste 15.5% w/w, milk powder 7.5% w/w, soy four 7% w/w, and emulsifier 0% w/w, respectively. At optimum point, hardness, cohesiveness, adhesive force, T onset, and T m were 367.07 (N), 0.63, 8.46 (N), 28.1 (°C), and 33.7 (°C), respectively. The predicted values were confirmed through validation experiment.

Problem of low protein content in chocolate compound can be solved by incorporating soy flour. Therefore, formulation of white compound chocolate containing sesame paste and enriched with soy flour leads to a nutritional natural novel product. Several different types of vegetable fat can be used in place of cocoa butter to produce a compound chocolate, and it is more usual to find a cocoa butter substitute (CBS) in the manufacture of compound chocolates. CBS fats have a sharp melting point, giving similar texture properties to those of chocolate (Dale, 2017). Recently, many studies have reported the rheological properties of sesame paste and blends containing sesame paste. The details of the afore-mentioned studies can be found in Alpaslan and Hayta (2002), Abu-Jdayil (2004). Studies on oil separation problem in halva (a sesame paste product) are very limited (Elleuch et al., 2014;Ereifej et al., 2005;Guneser & Zorba, 2014). To our knowledge, no data have been published on the presence of the sesame paste and soy flour in white compound chocolate formulation. The objective of this study was optimizing the formulation of a white compound chocolate containing sesame paste, soy flour, milk powder, and emulsifier based on textural and thermal properties using a combined mixture-process experimental design.

| Chocolate production
All the white compound chocolate containing sesame paste (500 g batch for each formulation) was produced in laboratory ball mill. A weighed amount of CBS was melted in a microwave oven at a temperature of approximately 45-50°C. Suspension was prepared at ambient temperature (20-22°C) by adding soy flour, sesame paste, and milk powder to CBS while mixing evenly with a spatula. The suspension then was transferred to the ball mill (ball diameter 9.1 mm; mixer rotation speed 100 rpm). The applied refining time in the mill was 70 min at 65°C. During the first 10 min of the experiment, the icing sugar was gradually added. To obtain the desired flow characteristics, monoglyceride was added 15 min prior to the end of the process. After the established refining time, the refined suspension white compound chocolate was discharged into molds shaken gently to remove air bubbles and then placed in refrigerator at temperature of 10°C. After cooling and demolding, samples were wrapped in aluminum foil for packaging and conditioned at ambient temperature (20-22°C) for 24 hr prior to analysis.

| Texture analysis
The texture profile analysis (TPA) with 30% compression was conducted on prepared samples using a texture analyzer (TA-XT plus,

| Thermal properties
The thermal properties of white compound chocolate containing sesame paste were determined utilizing a SPICO-DSC-100.
Approximately, 17-20 mg of the sample was placed in hermetically sealed aluminum pans with an empty pan as a reference. Melting profile was recorded by heating at 10°C/min from 10 to 80°C.
Following parameters were extracted from the melting curve: Onset temperature (T onset ) Maximum temperature (T m ) Each sample was analyzed in triplicate and mean values reported (Kiumarsi et al., 2020).

| Experimental design
D-optimal combined mixture-process design was used to investigate the effect of two factors; the first one is mixture components, including A: sesame paste (15%-30% w/w), B: milk powder (0%-15% w/w), C: soy flour (0%-15% w/w) with A + B + C = 30% w/w, and the second is D: emulsifier (0%-2% w/w) on five responses (hardness, cohesiveness, adhesive force, T onset , and T m ). To independently investigate the effect of the monoglyceride emulsifier, we considered it as a process variable. Other components of white compound chocolate containing sesame paste were sugar and CBS. Twenty combinations of the variables were selected by experimental design as shown in Table 1.
Another control sample was included to compare the results to a commercial sample. All experiments were conducted at three replicates.

| Statistical analysis
Design-Expert version 10.0.1 statistical software (Stat-Ease Inc.) was used to analyze the data. For each response, different models and their suitability were evaluated, and significant terms in generated mathematical models were determined by analysis of variance (ANOVA). The significance was judged at 5% probability level. The validity of the selected models was assessed by the coefficient of determination, R 2 , adjusted-R 2 (adj-R 2 ), and coefficient of variation (CV). R 2 is defined according to its magnitude which is the ratio of the explained variation to the total variation. A good mathematical fitting model should have a large R 2 (larger than 80%) and adj-R 2 . CV expresses standard deviation as a percentage of the mean.
Generally, CV should not be greater than 10% and small values of CV represent a better precision and reliability of the conducted experiments (Karazhiyan et al., 2011).

| Optimization and validation
The numerical optimization process using the Design-Expert software was carried out to optimize multiresponses based on desired function methodology. To describe the desirability of responses, they were either maximized or minimized (to target the control sample characteristics) while the mixture components and process variable were kept in range. To test the adequacy of the models, additional experiments at the optimum levels obtained by RSM optimization were conducted and the experimental data were compared with the predicted ones.

| Model fitting
According to data analysis, one of the two polynomial models, linear × linear, and linear × quadratic was selected to be the most appropriate model to analyze the responses. Table 2 illustrates analysis of variance for the fitting models. A significant lack of fit reflects the failure of the models to represent the data in the experimental design in which points were not included in the regression (Karazhiyan et al., 2011). The ANOVA depicted that lack of fit was not significant for all output responses at 95% confidence level, meaning that the models represented the data appropriately. To check the model validity, R 2 , adj-R 2 , and CV were also calculated. According to Table 2, R 2 of the models are all higher than 80%, indicating a TA B L E 1 Mixture optimal design matrix with uncoded values of the factors and observed textural and thermal responses

| Response surface plots
The relationship between dependent and independent variables can be illustrated using three-dimensional response surface graphs generated by the model. The data were generated through keeping one variable at its respective level and varying the other three within the experimental range.

| Textural and thermal properties
Texture is one of the key attributes for consumers' acceptance The TPA hardness is defined as the peak force during the first compression cycle (Kahyaoglu et al., 2005). The hardness value had a linear (mixture components) × quadratic (emulsifier) model as illustrated in Table 3 where the linear effect of mixture components (A, B, and C) and the interaction between C and D (emulsifier) were significant (p < .05). The effect of mixture components and emulsifier on hardness are presented in Figure 1a,b. Increasing milk powder has a TA B L E 2 Model statistics and adequacy of the models for hardness, cohesiveness, adhesive force, T onset , and T m responses  The effect of emulsifier on hardness at a constant amount of milk powder (5% w/w) is given in Figure 1b. At high proportion of sesame paste, emulsifier did not have large effect on the hardness compared with lower proportion of sesame paste. It seemed that at high level of sesame paste due to the high amount of sesame oil, the effect of emulsifier was not observed well. At low proportion of sesame paste (15% w/w), increasing emulsifier up to a certain value (1% w/w) decreased the hardness but which then increased at increased value of emulsifier (2% w/w).
The TPA cohesiveness is defined as the ratio of the positive force area during the second compression (A 2 ) to the force area of the first compression (A 1 ; Messens et al., 2000). The linear (mixture components) × quadratic (emulsifier) model in Table 3  The TPA adhesive force is defined as the necessary force to overcome attractive forces between surface of chocolate and contacting material (Bryant et al., 1995). Adhesive force had linear (mixture components) × linear (emulsifier) model (Table 2). Table 3 showed that for the model of adhesive force, the linear effect of all the mixture components (A, B, and C) and interactions of B and C with emulsifier (D) were significant (p < .05). The effect of mixture components and emulsifier on adhesive force is given in Figure 3a,b.
The replacement of milk powder by sesame paste had no substantial effect on the white compound chocolate adhesive force (Figure 3a).
Increasing sesame oil did not lead to significant increases in the adhesive force. Decrease in sesame paste and milk powder proportion with simultaneous increase in soy flour proportion leads to a decrease in adhesive force. The lowest and highest values of adhesive force were recorded for samples containing 15% soy flour and 15% milk powder, respectively.
According to Figure 3b, the adhesive force was emulsifier dependent and decreased from 8.1 to 5.2 N as the emulsifier concentration increased from 0% to 2% at low sesame paste proportion (15%w/w).
The T onset temperature in DSC thermogram was determined by the intersection of the baseline with the absolute highest tangent of the melting curve at which a specific crystal form started to melt. T onset had a linear (mixture components) × linear (emulsifier) model where the linear effect of the mixture components (A, B, and C) and the interactions of C with D (emulsifier) were significant (p < .05) to the response ( Table 3). The effect of mixture components and emulsifier on T onset is given in Figure 4a As shown by Figure 4b at higher sesame paste proportions (25% w/w), T onset did not change with emulsifier, but at lower proportion of sesame paste (15% w/w) T onset decreased along with the increase in emulsifier.
The maximum melting temperature (T m ) is the most important parameter of the DSC thermogram, and at this temperature, melting curve reaches its peak and melting rate is greatest (Aidoo et al., 2015). Regarding T m , Table 2 showed that the model was linear (mixture components) × linear (emulsifier). Table 3 indicated that linear effect of all mixture components and interaction of CD on T m were significant (p < .05). Effect of mixture components and emulsifier on T m are shown in Figure 5a,b. Increasing soy flour and sesame paste proportions from 0% to 15% w/w and 15% to 30% w/w, respectively, led to a decrease in T m (Figure 5a). Effect of soy flour was greater than sesame paste. The reduction in T m indicated that white compound chocolate containing soybean oil and sesame oil had a lower thermal resistance.
As shown in Figure

| Optimization and validation of response surface methodology results
To target the commercial sample's characteristics, our optimization experiments were designed to maximize some of the textural and thermal properties such as hardness, cohesiveness, T onset , T m , and minimize adhesive force. Sesame paste, soy flour, milk powder, and emulsifier were selected in the range of 15%-30%, 0%-15%, 0%-15%, and 0%-2% w/w, respectively.

ACK N OWLED G M ENTS
This project was supported by the Department of Food Science and Technology, Ferdowsi University of Mashhad, Iran. Support from the Research Institute of Food Science and Technology of Iran is gratefully acknowledged, as well.

CO N FLI C T O F I NTE R E S T
The authors declare that they do not have any conflict of interest. Characterization of melting properties in dark chocolates from varying particle size distribution and composition using differential scanning calorimetry. Note: T onset = onset temperature; T m = maximum temperature.