CAFFEINE IN HAIR-CARE AND ANTICELLULITE COSMETICS 253 up are important for the high-performance liquid chromatography (HPLC) analysis, where the sample solutions are directly injected into an HPLC column. The complexity of some cosmetics may obstruct their direct analysis. Thus, in many cases, cosmetic samples cannot be analyzed only by “dilute and shoot” approach because direct injec- tion of untreated samples may cause a large increase in column pressure. Also, injecting excipient molecules of higher molecular weights can lead to a decrease in column qual- ity, shortening its lifetime. In addition, retention of excipients may require a long column rinse after measurements. Therefore, a sample pretreatment step may be re- quired to eliminate proteins, emulsifying agents, and many other viscous constituents. Also, anticellulite gels and creams contain a wide variety of synthetic and natural (plant-derived) substances which may interfere in the caffeine determination. Solid- phase extraction (SPE) is useful for sample preparation and clean-up in the analysis of pharmaceutical creams. It can be more rapid and effi cient than liquid–liquid extraction and yields quantitative extractions without using large volumes of harmful organic solvents. In this work, a solid-phase extraction method followed by HPLC with ultra- violet detection was developed and applied to caffeine determination in anticellulite gels and hair-care products. EXP ERIMENTAL CHE MICALS AND REAGENTS Caf feine (1,3,7-trimethylxanthine) standard solution of LC-MS grade was obtained from Thermo Scientifi c (Waltham, MA), (Lot ER01081601). Caffeine powder of analytical grade (serial number 0030810) was supplied from NRK Inženjering (Beograd, Serbia). Methanol was of HPLC grade (Sigma-Aldrich, Eschenstrasse, Germany, Lot STBG4373V). For the sample preparation, 96% ethanol Ph. Eur. grade (Zorka-Pharma, Šabac, Serbia, serial number 57/226) was used. STOC K SOLUTIONS AND CALIBRATION CURVE The stock solution of caffeine with a concentration of 1.0 mg/mL was prepared in deion- ized water. This solution was stored for up to 1 week at 4°C. Working standard solutions were prepared by diluting the stock with deionized water. COSME TIC PRODUCTS Two c affeine-enriched shampoos, one caffeine-enriched hair balsam, one caffeine-free sham- poo, and two caffeine-enriched cellulite reduction products (in the form of gels) were obtained from the local cosmetic shops. According to the manufacturers’ declaration, both cellulite reduction products and hair-care products contained caffeine, plant extracts, and a large number of other substances. In all the examined cosmetics, caffeine was one of the major components as it was one of the fi rst seven ingredients in the ingredient list. The exact concentration of caffeine in any of the examined products was not declared.

JOURNAL OF COSMETIC SCIENCE 254 SAMPLE PR EPARATION After opt imization of the sample preparation procedure (see discussion), the following proto- col was adopted: 0.5 g of the anticellulite gel was accurately weighed and dissolved in 10 mL of 96% ethanol. After that, NH4OH was added to obtain pH 7–8. Insoluble excipients were separated by centrifugation for 10 min at 4,000 rpm at 20°C. Then, 1 mL of the supernatant was loaded on the cartridge, and the SPE procedure was performed. 0.5 g of the hair-care product was dissolved in 10 mL of distilled water and heated at 50°C in a water bath for 45 min. The obtained pH of the samples was 5–6. After cooling and centrifugation for 10 min at 4,000 rpm at 20°C, 1 mL of the supernatant was loaded on the SPE cartridge. SOLID-PHASE EXTRACTION VisiprepTM S PE vacuum manifold Supelco 57030-U (Sigma-Aldrich Chemie GmbH, Darmstadt, Germany) was used for the SPE. Five different SPE sorbents (C18, C18ec, C18 Hydra, C8, and HR-X) of the same volume and capacity (1 mL/100 mg) from the same producer (Macherey-Nagel, GmbH, Düren, Germany, REF 730 207, LOT 60.007) were tested. The cartridge C18ec was octadecyl modifi ed endcapped silica, C18 Hydra was a special octadecyl phase for polar analytes, and HR-X was a hydrophobic polystyrene- divinylbenzene copolymer. SPE cartridge s were conditioned with 3 mL of methanol and 3 mL of deionized water. One milliliter of the prepared sample was loaded on the cartridge at a fl ow rate of 1 mL/ min. The elution of caffeine was achieved with 4 × 1 mL of methanol. HPLC CONDITIO NS HPLC analyses w ere performed on Agilent Technologies 1200 Series apparatus (Santa Clara, CA) with an diode array detector (DAD) and a fl uorescence detector. The separa- tion was carried out on the Restek Ultra IBD C18 column (150 × 3 mm, 3 μm, Lot 13122OP, Ser 14020153J) (Bellefonte, PA) at 30°C. The mobile phase consisting of a methanol–water mixture (40:60, v/v) was pumped in an isocratic mode at a fl ow rate of 0.4 mL/min. Two microlitre of the fi nal eluate was injected into the HPLC column. UV detection set at 274 nm was used as the optimal wavelength for caffeine determination. RESULTS AND DISCUSSI ON OPTIMIZATION OF THE CHROMATOGRAPHIC CONDITIONS Chromatographic cond itions for the determination of caffeine in food matrices are well established and have been published in many articles (16–20). Because the C18 analytical column has been used for caffeine separation in most of the articles, this column was also selected in our work. To optimize HPLC conditions for the caffeine analysis of cosmetic products, the methanol-to-water volume ratio was varied from 20:80 (v/v) to 50:50 (v/v). The most suitable mobile phase composed of 40% methanol and 60% water, giving well-