Analyzing adsorption data of erythrosine dye using principal component analysis
| Autor: | Yahya S. Al-Degs, Samer S. Abu-Alrub, Rajab Abu-El-Halawa |
|---|---|
| Rok vydání: | 2012 |
| Předmět: |
Chemistry
General Chemical Engineering Analytical chemistry Fraction (chemistry) General Chemistry Erythrosine Endothermic process Industrial and Manufacturing Engineering chemistry.chemical_compound Adsorption Ionic strength Principal component analysis medicine Environmental Chemistry D-value Activated carbon medicine.drug |
| Zdroj: | Chemical Engineering Journal. 191:185-194 |
| ISSN: | 1385-8947 |
| Popis: | Erythrosine or Acid Red 51 (C.I. 45430) is a common food dye used for coloring many food products including luncheon meat. Few adsorption studies conducted on erythrosine removal from solution and the tested adsorbents showed modest adsorption values for this dye. Pittsburgh commercial activated carbon showed a reasonable adsorption performance for erythrosine from solution with a maximum capacity of 89.3 mg/g at 40 °C and pH 7. The formation of a complete mono-layer was not established where the fraction of surface coverage was only 0.21. The equilibrium distribution value K d was 6.51 L/g at 40 °C indicating the high affinity of erythrosine toward activated carbon. Adsorption of erythrosine was an endothermic process (Δ H ads 31.6 kJ/mol) and spontaneous over the studied temperatures (293–313 K). By conducting 27 adsorption tests, the influence of six experimental variables (shaking time, pH, mass of adsorbent, initial dye content, ionic strength/NaCl concentration, and solution temperature) on dye adsorption was investigated. Using principal component analysis PCA, an empirical relationship was created for correlating K d values with the studied variables and the relationship contains six linear terms, six non-linear terms and one interaction term of the variables. Analysis of adsorption data by PCA revealed that linear and non-linear terms of variables were more significant than interaction term for data modeling. Two sets were created for model building and validation. The developed model was effective for predicting K d , the sums of square errors squared SSE values were 1.6 and 2.1 for calibration and validation sets, respectively. PCA predicted the best combination of the experimental variables that would give the highest possible K d value, 20.3 L/g. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect