Density mapping and chemical component calibration development of four-component compacts via terahertz pulsed imaging

The purpose of this research was to investigate suitable procedures for generating multivariate prediction vectors for quantitative composition and density analysis of intact solid oral dosage forms using terahertz pulsed imaging (TPI) spectroscopy. Both frequency- (absorbance and refractive index) and time-domain data are presented. A set of calibration and prediction samples were created according to a quaternary mixture design with five levels of compaction at each concentration design point. Calibration models were generated by partial least-squares, type II (PLS-2) regression of the TPI spectra against nominal composition and relative density reference measurements. Quantitative frequency-domain composition calibration models were created for all crystalline components (R2 > 0.90), but the calibration models for individual amorphous components (R2 < 0.76) did not perform as well in testing. Combining both amorphous components into a single component variable for regression resulted in lower error statistics and equally good predictions of crystalline components. A non-linear attenuation of time-domain spectra was observed as a function of compaction force, which corresponded to compact density predictions (R2 = 0.948). While refractive index spectra were sensitive to density (R2 = 0.937), the absorbance spectra were not. Surface density maps were prepared based on refractive index calibrations.