NEAR IR SURFACTANT ANALYSIS 449 i.800. i.600' i.400' u i.aOO- m i.000- ß 800- ß 600 ß 400 ß •oo NATER ALS • ' | ' | ' | ' • ' I ' I ' i•00 t400 t600 t800 •000 •00 •400 NAVELEN6TH Figure 2. Typical nearin•ared reflectancespectra •rammonium ]aurylsulGte and water. RESULTS AND DISCUSSION RAW MATERIALS Since the NIRA technique is strictly correlative, the precision and accuracy of the primary analyses affect the quality of the calibrations. Standard procedure is to calculate the pooled standard deviation (PSD) on blind duplicates on ten different samples. This quantity is defined as = •/i•d,2/2n .= PSD (2) Table III Compositional Data for Anionic Surfactants Ammonium Lauryl Sulfate Sodium Lauryl Sulfate Constituent Mean % SD Range Mean % SD Range Active Detergent 28.99 0.68 27.45-29.97 29.36 0.88 27.91-30.81 Solids 30.61 0.85 27.84-32.16 30.98 1.20 29.01-33.24 Benzoic Acid 0.35 0.04 0.28-0.45 0.35 0.07 0.25-0.57 Viscosity* 21.58 6.11 11.00-32.00 ** pH 5.01 0.11 4.81-5.25 5.38 0.17 5.06-5.83 * All terms must be multiplied by 100. ** Viscosity not measured.

450 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Table IV Linear Regression Data for NlRA Prediction of ALS Constituent nm Ko.. Kn r SEC F Active Detergent 323.05 0.90 0.30 96 1722 -637.60 2083 - 1084.12 2139 460.34 Solids Benzoic Acid Viscosity 93.05 0.87 0.42 83 2130 583.94 2230 - 1323.70 2270 581.39 - 20.26 0.86 0.02 51 1600 - 182.61 1668 280.41 1344 -237.77 1356 150.97 - 1286.12 0.66 4.7 20* 1580 -56315.59 2332 15451.09 2456 -3312.81 2300 - 9048.39 1596 56052.38 pH 127.50 0.60 0.11 20' 1710 564.37 1722 -480.97 2139 109.29 2250 -329.76 Not statistically significant. where di is the difference between duplicate analyses and n is the number of duplicate analyses. An example of this procedure is given in Table I. The calculation of residual standard deviation for five repetitive analyses was also made according to: RSD = (x i -- •)2/(n -- l) (3) where x i is the individual value of an analysis, i is the sum of all analyses, and n is the number of analyses. The rule of thumb is that the standard error of calibration in NlRA should be slightly higher than the pooled standard deviation by primary analyses. The targeted standard deviations for the primary analyses used are listed in Table II. The near infrared reflectance spectra of raw material surfactants do not differ very much from that of pure water. These products are at least 70 percent water and the near infrared region is especially absorbent to water. Figure 2 is a typical example of a surfactant spectrum. The two major peaks can be readily identified as the first overtone for water at 1440 nm and the combination stretch/bend at 1940 nm (11). Carbon-hy- drogen doublets at 2300-2400 nm and 1700-1800 nm are the only features distin- guishing surfactants from the water spectrum also seen in Figure 2. Means, standard deviations (SD), and ranges for percentage active detergent, solids, benzoic acid, vis-