Chemometrics role on optimizing industrial processes or product characteristics: The comparison of D-optimal and I-optimal mixture designs on the prediction of dye mixture composition

• Main Authors: Heryanto, R., Syafitri, U. D., Anggraini, C., Septaningsih, D. A., Izzati, M. N.
• Format: Conference Proceeding
• Language: English
• Subjects: Azo dyes; Chemometrics; Color; Coloring agents; Composition; Design of experiments; Indigo; Model testing; Optimization; Quadratic equations; Sunset
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0002670

Mixing materials to get a certain characteristic can be done through an experimental design approach. This study aims to compare D-optimal and I-optimal mixture design based on experimental efficiency and accuracy of predictions. The study case was carried out on three food coloring agents (XR =Ponceau 4R, XY= sunset yellow, and XB = indigo carmine) and the response was measured as the L* a* b* color scale. The quadratic equation model was made from the relationship between the mixed composition and the color response. The assumption model for the two optimal design was a quadratic model. The quadratic equation model of the two designs shows that the coefficient values are not much different. The XY variable increases the response of L* and b*, while the XR variable increases the response of a*. The value of R2 is around 98% to 100%. The model obtained was tested further for its ability to predict ten colors. Both designs can predict 8 of 10 colors well. Based on the results obtained, the optimal I-ability is almost the same as D-optimal.