Solvent and Reagent Pure Component Spectra

The arrays of preliminary, pre-catalytic spectral data A x, can be processed in order to get pure component spectra which are valid in the vicinity of the initial reaction conditions. 

For infrared spectroscopy, this requires proper subtraction of the background moisture and carbon dioxide to get a good approximation of solvent (in the cell). These can then be used to obtain a good approximation of solute A, followed by solute B etc. This process can be done manually by trial and error,or it can be automated for optimal subtraction. The output is a set of n reference spectra. 

An automated and optimal subtraction procedure was previously developed for Eq. (8). The algorithm takes one spectrum after another and optimally subtracts previous signals. The criterion used is minimization of the signal entropy (see Section 4.5.2). The recursive equation used appears in Eq. (9), where is an experimental spectrum, is a set of reference spectra, and x are scalar coefficients. The values of x are easily determined. 

The automatic procedure for reference spectra generation was first demonstrated for the start-up of a homogeneous catalyzed rhodium hydroformylation of cyclo-octene using Rh4(CO)i2 as precursor, n-hexane as solvent and FTIR as the in situ spectroscopy at 298 K [63]. The first n spectra were (i) empty spectrometer compartment (background), (ii) n-hexane at 0.2 MPa in a high pressure thermostatically controlled cell fitted with Cap2 windows (iii) system equilibrated with 2.0 MPa CO, (iv) system upon addition of cyclo-octene, and (v) system upon addition of Rh4(CO)i2. The n=l reference spectrum, which contained atmospheric 

1) Many commercial spectrometers, particularly IR spectrometers, have software capability for automated subtraction of a pair of ejq)eri- 

---