| Abstract
| - High-throughput experimentation and screening methodsare changing work flows and creating new possibilities inbiochemistry, organometallic chemistry, and catalysis.However, many high-throughput systems rely on off-linechromatography methods that shift the bottleneck to theanalysis stage. On-line or at-line spectroscopic analysisis an attractive alternative. It is fast, noninvasive, andnondestructive and requires no sample handling. Thedisadvantage is that spectroscopic calibration is time-consuming and complex. Ideally, the calibration modelshould give reliable predictions while keeping the numberof calibration samples to a minimum. In this paper, weemploy the net analyte signal approach to build a calibration model for Fourier transform near-infrared measurements, using a minimum number of calibration samplesbased on blank samples. This approach fits very well tohigh-throughput setups. With this approach, we canreduce the number of calibration samples to the numberof chemical components in the system. Thus, the questionis no longer how many but which type of calibrationsamples should one include in the model to obtain reliablepredictions. Various calibration models are tested usingMonte Carlo simulations, and the results are comparedwith experimental data for palladium-catalyzed Heckcross-coupling.
|
