Distinctive, relatively small, analyte extinction coefficients in the near-
infrared (NIR) spectral region, have led to NIR methodologies to probe the
composition of concentrated solutions. However, as exemplified with alkalin
e hydroxide solutions, it is shown that the analyte strongly affects the co
nventional aqueous baseline of these measurements, and this baseline behavi
or must be addressed for effective NIR analysis. In the domain of molal and
higher analyte concentration, the water baseline changes drastically as so
lute interacts and confines bulk solvent. At 967 nm an (hydroxide) absorpti
on peak increases smoothly with increasing hydroxide activity in LiOH and N
aOH electrolytes, Yet at this same wavelength, there is an unexpected absor
ption decrease in concentrated cesium (>5 m CsOH) and potassium (>13 m KOH)
electrolytes, evidently due to solvent bridged anion-cation association. A
peak for bulk water occurs nearby at 976 nm, with a unit pathlength absorb
ance of 0.24. This absorbance diminishes by an order of magnitude at high a
nalyte concentration, as water activity decreases. A broad strong (0.53 cm(
-1)) series of absorbances centered near 1200 nm diminish smoothly with dec
reasing bulk water activity. However, a broad strong absorption (14 cm(-1))
centered near 1450 nm, is resolved into two peaks, one disappearing, the o
ther increasing at high hydroxide concentrations. Recognition of these vari
ations permits effective NIR determination of analyte (in this case hydroxi
de). Second-order differentiation, partial-derivative(2)A/partial-derivativ
e-lambda(2), at either 967 or 1421 nm, minimizes the effects of the near-ly
ing bulk water absorption, and of baseline shift, and provides a Linear var
iation with hydroxide concentration. (C) 1999 Elsevier Science B.V. All rig
hts reserved.