CAFFEINE IN HAIR-CARE AND ANTICELLULITE COSMETICS 255 separated and good shaped peaks, with a run time less than 5 min. Thus, the retention time of caffeine was about 3.3 min. The low retention time enabled a fast chromato- graphic analysis but also very good selectivity of the method. As can be seen in Figure 2, caffeine-enriched shampoos had a compound that appeared at 4.1 min with a low resolved peak, which gave an absorbance in a similar UV region as caffeine (260–280 nm) (Figure 3). The resolution between peaks of caffeine and this ingredient was 0.98 recorded at 260 nm, whereas at 274 nm, this potentially interfering peak gave a very low signal. Ade- quate selection of working detector wavelength eliminated the signal of an unknown substance. Thus, the unknown substance at 4.1 min does not interfere in the caffeine analysis when a detector was set at 274 nm because at this wavelength, this substance has very low absorption. On the other hand, the chromatograms of gel after SPE (Figure 4) were very clean when the detector was set at 260 nm, 270 nm and 274 nm. For these reasons, isocratic elution was adopted, and by selecting the detection at 274 nm, good peak area, peak width, and selectivity for caffeine were achieved (Table I). The examination of t he system suitability was conducted in terms of retention time re- peatability, peak symmetry, peak width, number of theoretical plates, capacity factor (k′), resolution, and selectivity (Table I). The capacity factor (k′) is optimal as reference values are between 1 and 5. The values of k′ larger than 5 lead to longer retention and analysis time, whereas values less than 1 are unreliable and cause low resolution. The number of theoretical plates higher than 5,000 indicates a good separation of caffeine. Also, the se- Figure 2. Chromatogram of shampoo 1 extract on the HR-X cartridge at 260, 270, and 274 nm c affeine retention time 3.332 min.

JOURNAL OF COSMETIC SCIENCE 256 lectivity factor and resolution had excellent values demonstrating satisfactory overall col- umn effi ciency in the caffeine analysis. OPTIMIZATION OF SAMPLE PREPARATION PROCEDURE In cosmetic products, c affeine is dispersed in complex emulsion matrices. Before the analysis, cosmetic emulsions are usually dissolved in water, methanol, or buffers by heat- ing in a water bath. As the hair-care products and anticellulite gels are highly viscous, the fi rst step in the optimization of the preparation procedure was the selection of the dissolu- tion solvent. The optimal solvent should fulfi ll two requirements: to provide a solution that is not viscous and that can easily fl ow through SPE cartridges, and to ensure good solubility and release of the caffeine from the sample matrix. The latter is actually a clas- sic single-stage solid–liquid extraction that is desirable to perform before the application of SPE purifi cation. Considering that caffeine is sparingly soluble in water at room tem- perature, but freely soluble in boiling water (21), we selected the sample preparation which included dissolution in deionized water and heating at 40°–50°C in a water bath. This avoids the use of solvents such as methanol or acetonitrile which are harmful to health and the environment. In addition, phosphate buffer (pH 7) was also tried, but the samples dissolved in phosphate buffer clogged the SPE cartridges as some insoluble phos- phate compounds were precipitated. The additional procedure was applied to anticellulite gels. They were fi rst dissolved in warm deionized water, but viscous samples that could not pass through the cartridge were obtained. One of the main problems in these samples was the presence of carbomer. Carbomer is a synthetic high–molecular weight polymer of acrylic acid which is soluble in water. It is widely used in anticellulite gels as a vehicle for incorporation of liposomes Figure 3. UV spectrum of the caffeine peak and peak of unknown compound from shampoo 1 extra c t on the HR-X cartridge.