254 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS tent between the Baddiel and DRIFTS results is that as the SiMe band intensity in- creases, indicating more fiber surface response, the Amide I and Amide II bands shift to lower wavenumbers, indicating a cuticle-rich phase. Again, these observations are con- sistent with surface-vs-bulk analysis of the fibers. It also demonstrates the importance of maintaining constant, repeatable sample preparation techniques to generate consistent spectra and data. Additional experiments were conducted to define the optimum KBr/hair ratio. The spectral definition of the bands (sharpness) increased as the KBr/hair ratio was varied from 1 to 4. Above 4, the total signal output decreased due to dilution of the hair sample with KBr. Therefore, a KBR/hair ratio of 4 was used. The quantity of sample (WIG-L-BUG vial loading) also influences spectral definition, as illustrated in Figure 5. Both spectra were collected at a 4:1 KBr/hair ratio and two-minute grinding time, but spectrum A was collected with a 0.50 g vial loading and spectrum B with a 0.25 g loading. The level of agitation in the vial is greater for the 0.25 g loading, as shown by the decreased intensity of the SiMe band relative to the Amide III band, indicating greater bulk analysis of the fiber rather than surface anal- ysis. The 0.25 g loading was selected even though the SiMe band intensity appears greater with the 0.5 g loading for two-minute grinding times. The reasoning for selec- tion of the 0.25 g of loading concerns the heterogeneous character of the KBr/hair mixture. The 'DRIFTS sample cup capacity for KBr/hair mixtures is 0.2-0.3 g. Thus, 75.5 69.0 62.5 uJ o 56.0 z 49.5 43.0 36.5 30.0 SpectrumHa••GAKBr/0.10 0.40 G ide III 0.20 G KBr/0.05 G Hair Spectrum B i 1850 1750 1650 1550 1450 1350 1250 1150 1050 WAVENUMBERS Figure 5. Influence of sample size in grinding vial on spectra of treated hair. i 950

SILOXANES ON KERATINS BY FTIR 255 81.9 • 1 Min. Gnndlng T!me • /• ß / 2 Min. Gnndlng Time 1850 1750 1650 1550 1450 1350 1250 1150 1050 950 WAVENUMBERS Figure 6. Optimization of sample preparation using 0.20 g KBr/0.05 g hair. all of the prepared sample can be used from the 0.25 g loading, which contributes to increased method repeatability. Spectral definition of the 0.25 g loading was improved by reducing the grinding time to one minute, as illustrated in Figure 6. Siloxane detection and quantitation. The first technique evaluated for siloxane detection and quantitation was spectral subtraction. An FTIR spectrometer provides greater signal/noise (S/N) response and increased wavenumber accuracy over dispersive units. Both of these advantages are required to eliminate artifacts in the subtracted results. Spectral subtraction, as the name implies, involves subtracting the spectrum of one species present in a sample from the total spectrum of that sample. This procedure is used frequently to identify unknown constituents in a sample whose primary ingre- dients are known. Spectral subtraction is based on the assumption that the total absor- bance in a sample is the sum of the absorbances from all the species in the sample. This absorbance additivity rule applies when intermolecular interactions are negligible. Ab- sorbance rather than % transmittance units are required for spectral subtraction because absorbance is proportional to concentration. Three examples of spectral subtraction are presented in Figure 7. The AA results (mg/ kg Si), determined independently, are listed for each sample. The spectrum of un- treated hair was subtracted from spectra of siloxane-treated hair. The absorbance scale was held constant and the spectra were shifted along the y-axis to allow band height comparisons for the SiMe bands. The overall intensity of the SiMe band increases with