236 JOURNAL OF COSMETIC SCIENCE Sample solutions of dye intermediates at a concentration of 0.025 g/100 g in matrix media (at the different concentrations given above) were prepared by dilution from the respective stock solutions at various proportions of MeOH and Soerensen. It has to be noted that all samples contained NaAsc as an antioxidant agent added to the Soerensen buffer to ensure the stability of the dye samples. PROCEDURES Reversed-phase HPLC conditions. A non-linear MeOH (A)/aqueous phase (B) gradient was used as follows: 0-25% A for 19 rain, 25-80% A for 10 rain, 80% A for 5 rain, 80-95% A for 5 rain, 95% A for 10 rain, and 95-0% A for 3 min. The total flow was ! ml/min. Between the injections, the column was equilibrated by a 25-ml mobile phase. Each analysis was repeated five times. The column temperature was kept at 48øC. The data acquisition was carried out at two or three selected wavelengths: 220 nm, 235 nm, and 290 nm, in parallel with the spectra acquisition. Isolation of the matrix components from the final analyte solution. First, experiments using anion exchange, solid phase extraction, liquid cation exchange, and liquid-liquid ex- traction were carried out. The results led to the selection and the optimization of a liquid-liquid extraction procedure of the matrix components from the sample solution by n-heptane. Two milliliters of sample solution (single-matrix product solution or mixture of dye intermediates and a matrix compound, at the concentrations given in the Experimental section) were treated with n-heptane. Depending on the matrix product, the extraction involved one to three steps. In the first step, the extraction was performed on the 2-ml sample using 20 ml n-heptane, the two phases were separated, and the resulting aqueous phase 1 was thereafter submitted to HPLC or analyzed by UV-Vis spectrophotometry. For additional extraction steps, 20 ml n-heptane were added to the resulting aqueous phase n-l, separation of the two phases was performed, and the resulting aqueous phase n was submitted to HPLC or analyzed by UV-Vis spectrophotometry. RESULTS AND DISCUSSION A systematic study was carried out on each of the eighteen selected matrix compounds and forty-seven selected hair dyes (Tables I and II). First, their individual retentive behavior was determined and their UV spectrum recorded, using the DAD or the UV-Vis spectrophotometer. Second, the efficiency of the liquid-liquid extraction pro- tocol on the selected matrix compounds was checked. When a matrix compound could not be extracted from the sample solution, its influence on the retentive behavior of the dye intermediates was investigated. CHROMATOGRAPHIC STUDY Single solutions of dye-forming compounds. Forty-seven dye-forming compounds were sub- jected to chromatography and their retention times and UV spectra were recorded. Results are presented in Table II. It comes out that 91% (43 dye intermediates) of the dye intermediates tested have their

MATRIX COMPOUNDS AND OXIDATIVE HAIR DYES 237 retention time between 5 min and 30 min. Some of the tested dyes have very close retention times--4-n-l,2-pd, 2-a-4-np, and 4-a-2-ht for instance, their retention times being 15.10 min, 15.50 min, and 15.66 rain, respectively. However, since their respec- tive UV spectra show remarkable differences (8), their identification can still be carried out easily. In order to test the repeatability, five injections of each sample were carried out. The relative standard deviation (RSD) for the retention time was between 0.2% and 1.1%, while that of the peak corrected areas (defined as the peak area divided by the retention time), recorded at 290 nm, was between 1.1% and 6.0% (9). The chromatographic behavior showed good repeatability. The injections were repeated on three different columns in order to test the reproduc- ibility (Table II). For thirty-six hair dye intermediates, the relative standard deviation (RSD) on the retention time was less than 5% for nine dye intermediates, this RSD was between 6% and 10% and more than 10% for only two dyes. The reproducibility between columns was considered satisfactory. Single solutions of matrix compounds. Eight different compounds were subjected to chro- matographic separation, and their retention times and UV spectra were recorded. Results are presented in Table III. It appears that some of the tested compounds have very close retention times--BHT, OA, and ORA, for instance. However, with their respective UV spectra also showing remarkable differences (Figure 1), their identification can still be carried out easily. It must also be noted that most of the matrix compounds, with the exception of DMDM, PVP and MP, have retention times greater than 30 minutes, meaning that confusing dye intermediates with matrix products should not occur since the retention times of the hair dyes generally range from 5 to 30 minutes (1). Concerning the four dye intermediates for which retention times are out of the range of 5 to 30 min, the discrimination from the matrix compounds can also be easily made according to both retention time and UV spectrum. In order to test the repeatability, five injections of each sample were carried out. The relative standard deviation (RSD) for the retention time was between 0.02% and 0.7%, while that of the peak corrected areas was between 0.3% and 3.7%, except for DMDM. The chromatographic behavior of the matrix compounds showed good repeatability. Table III Retention Times and Corrected Peak Areas for Eight Matrix Compounds, Each in a Single Solution (n = 5) Retention time (min) Corrected peak area (AU) Matrix compound Mean RSD (%) Mean RSD (%) k of measurement (rim) DMDM 6.80 0.21 4.3 12.4 220 PVP 7.10 0.24 1.1 0.2 220 MP 26.87 0.06 1.0 1.8 290 NNO 33.80 0.70 5.2 3.4 235 EOP 35.85 0.08 5.8 0.3 290 OA 37.97 0.30 20.6 2.9 235 BHT 38.11 0.02 2.5 3.7 290 ORA 38.35 0.09 12.2 0.5 235 AU: arbitrary unit.