An efficient, fast and cost-effective method for detecting adulteration in perfumes by UV spectroscopy and multivariate analysis is proposed. Classification of perfumes, either as original or fake, was accomplished with the spectroscopic data using chemometric techniques such as principal component analysis (PCA), soft independent modeling of class analogy (SIMCA) and linear discriminant analysis (LDA).
Keywords
perfume; falsification; UV spectrophotometry; chemometrics
| Perfumea a Fake perfume (F) and original (O). | Mark | Origin |
|---|---|---|
| F1-F10 | A | unauthorized retailer |
| F11-F25 | B | Ceará State Forensic Police Department |
| 01-010 | A | authorized retailer |
| 011-025 | B | authorized retailer |
| Class | Training' | Validationa a One original and one fake perfume samples were classified as outliers |
|---|---|---|
| Fake | 12 | 12 |
| Original | 12 | 12 |
| Total | 24 | 24 |
| Class | Sample | Variance / % | Correctly classified | Misclassified | Not classified |
|---|---|---|---|---|---|
| Origin | 24 | 80.81 | 23 | 0 | 1 |
| Fake | 24 | 86.71 | 24 | 0 | 0 |
| Variable | Discriminant function | Classification function | |
|---|---|---|---|
| Non-standard | Group 1 (fake) | Group 2 (original) | |
| V203 | 3.661 | 19.002 | 3.859 |
| Constant | -3.270 | -14.092 | -1.323 |
