JOURNAL OF COSMETIC SCIENCE 240 temperature and pH, were involved in the RSM regression model. The response value was the peak area ratio of 3-O-ethyl ascorbic acid and methylparaben. Based on preliminary studies, the central point was set to 45°C and pH 6.00. After running the regression analy- sis, the coeffi cients of the second-order regression model were determined and substituted in equation 1. 2 2 1 1 2 2 1 2 = –3.8093+0.1063 – 0.0004 +1.8378 – 0.1222 +0.0139X(6)X Y X X X X This regression RSM model (R2 = 0.822) was used to predict the system’s response. Table II shows the ANOVA results of independent factors from the 3-O-ethyl ascorbic acid stabil- ity study. The p values and regression coeffi cients of the model are shown in Table III. Because the linear term of pH was a signifi cant factor (p 0.05), the peak area ratio was signifi cantly affected by changes in the pH. Thus, 3-O-ethyl ascorbic acid was very sensitive to pH. However, temperature and pH were both included when running the regression analysis. Because the analysis included nonsignifi cant terms, the regression results might be affected by these nonsignifi cant terms and resulted in a reduction of R2 value. Figure 5 Table II Analysis of Variance for the Overall Effect of Factors on 3-O-Ethyl Ascorbic Acid Stability (X1: Temperature, °C and X2: pH) SS df MS F p value X1 (linear) 0.3984 1 0.3984 3.1981 0.1482 X1 (quadratic) 0.0408 1 0.0408 0.3273 0.5978 X2 (linear) 1.1618 1 1.1618 9.3266 0.0379* X2 (quadratic) 0.3454 1 0.3454 2.7730 0.1712 X1 × X2 (interactive) 0.3913 1 0.3913 3.1409 0.1510 Error 0.4983 4 0.1246 Total SS 2.7981 9 *Signifi cant at p 0.05. Table I Experimental Design and Observed Response Values (Area/IS) with Two Independent Factors, X1 (Temperature, °C) and X2 (pH) Run no. X1 X2 Response values Temperature pH Area/IS 1 45.0 (0)a 8.12 (+1.414) 2.176 ± 0.018 2 60.0 (+1) 7.50 (+1) 1.297 ± 0.009 3 66.2 (+1.414) 6.00 (0) 2.887 ± 0.036 4 60.0 (+1) 4.50 (-1) 2.891 ± 0.005 5 45.0 (0) 3.88 (-1.414) 2.962 ± 0.053 6 30.0 (-1) 4.50 (-1) 3.097 ± 0.005 7 23.8 (-1.414) 6.00 (0) 2.973 ± 0.024 8 30.0 (-1) 7.50 (+1) 2.754 ± 0.021 9 45.0 (0) 6.00 (0) 3.021 ± 0.013 10 45.0 (0) 6.00 (0) 2.977 ± 0.030 a (-1.414), (-1), (0), (+1), and (+1.414) are coded symbols for levels of independent factors. Response values are means of three replicates (means ± SD), and they represent the area ratio of 3-O-ethyl ascorbic acid and methylparaben (internal standard, IS).

ANTIOXIDANT ABILITY AND STABILITY STUDIES OF 3-O-ETHYL ASCORBIC ACID 241 shows the 3-D mesh plot of the RSM model. The surface curve of the fi gure represented the HPLC peak area ratio of 3-O-ethyl ascorbic acid and methylparaben. A larger peak area ratio indicated that more 3-O-ethyl ascorbic acid was detected. The highest calcu- lated point of the surface curve in Figure 5 was 3.133, which occurred at 36.3°C and pH 5.46. When cosmetics were stored at the optimal conditions, 3-O-ethyl ascorbic acid had the best stability. Because this compound was very sensitive to changes in pH, buffers would be required for cosmetics to maintain the optimal pH (5.46). On the other hand, the stability of 3-O-ethyl ascorbic acid was not signifi cantly affected by temperature. These results would be useful for cosmetic manufacturers who could use 3-O-ethyl ascor- bic acid in their products. CONCLUSIONS Based on the reducing and DPPH radical scavenging ability analysis results, 3-O-ethyl ascorbic acid is a good antioxidant. Moreover, this compound inhibited the activity of tyrosinase to prevent the formation of melanin. Therefore, it could be used in cosmetics Figur e 5. Response surface plot showing the effect of temperature (°C) and pH on the stability of 3-O-ethyl ascorbic acid. Table III Regression Coeffi cients of the 3-O-Ethyl Ascorbic Acid Stability RSM Model Regression coeffi cient Estimate Intercept, β0 -3.8093 β1 0.1063 β11 -0.0004 β2 1.8378 β22 -0.1222 β12 0.0139