Spectral stripping

Multivariate calibrations are powerful tools, but the number and type of calibration samples required often is prohibitive. To overcome this problem, Pelletier employed a powerful but relatively uncommon tool, spectral stripping. This technique takes advantage of existing system knowledge to use spectra of fewer, more easily generated samples. More applications of this approach can be expected. 

The figures of merit of primary interest to analytical chemist that are discussed are noise characteristics and powers of detection, linearity and dynamic range, spectral stripping and interpolative correction for variable underlying background, compromise between resolution and spectral coverage and application to nonideal "real" samples. 

Figure 4. Net spectra after spectral stripping with 1% HN03 blank spectrum. Spectra obtained with Spectrometer 2, Table I. Key a, 15 pgjmL 77 with 5000 pg/mL Ca and Mg, and 500 pgjmL Al and b, 15 pg/mL H.

Figure6. Analytical calibration for 77190.8 nm after spectral stripping with l%HNO blank. Key , high concomitants level = Ca and Mg 5000 yg/mL, A1500 yg/mL a, medium concomitants level= Ca and Mg 750 yg/mL, AI75 yg/mL o, low concomitants level = Ca and Mg 100 yg/mL, A110 yg/mL and , no concomitants.

Figure 7. Analytical calibration for As 193.7 nm after spectral stripping with 1% HNO blank. Concomitant concentrations are the same as in Figure 6.

A simplified version of this spectral stripping technique was demonstrated with the SPD for the V doublet (309.27, 309.31 nm), which is unresolved from the A1 doublet (309.27, 209.28 nm). The spectra of pure A1, pure V (as shown in Figure 13), and various mixture solutions of the two were first measured and stored. 

Finally, the effects of spectral interferences could be at least partially alleviated by more effectively exploiting the multichannel nature of the SPD. Spectral interferences could be easily recognized via line intensity ratios and alternative, non-interfered analyte lines could be chosen or spectral stripping procedures may be used, as previously discussed. 

Spectral subtraction is an extremely useful method to remove solvent absorption features from the spectrum of a solution, thereby revealing more detail about the spectral features of the solute(s). Spectral subtraction has proved very effective in other applications, including successive spectral stripping of pure components from a complex mixture and monitoring starting material depletion and product formation (Figure 10). Spectral subtraction is performed on absorbance spectra as the absorbance scale is linear. The difference spectrum is calculated by equation (12) ... 

Figure 9 (A) Spectral stripping of components from the tunable diode laser infrared spectrum of cigarette smoke. From bottom to top ethylene, acrolein, methanol, and comparison with hydrazine. The reference spectrum is the lower of the two traces in each case. (B) Expanded view of the residual spectrum compared to that of hydrazine. (Reprinted with permission from Plunkett S, Parrish ME, Shafer KH, Nelson D, Shorter J, and Zahniser M (2001) Time-resolved analysis of cigarette combustion gases using a dual infrared tunable diode infrared laser system. Vibrational Spectroscopy 27 53-63 Elsevier.)...

The use of chemometrics will continue to increase and is almost mandatory for online applications. The spectral stripping of cigarette smoke could be done using a chemometrics approach rather than by sequential subtraction. The advantage is that all the components would be determined simultaneously without the laborious and somewhat subjective subtractions. The disadvantage is that the work to setup the method is somewhat greater however, for a large number of samples this would be recovered in reduced time per analysis and the use of less skilled personnel. The trade-off between time to setup the method and time to carry out the analysis is the usual case. 

Many attempts were made historically to solve the problems of interferences affecting rare earth element (REE) ion signals [10,11]. Spectral stripping or deconvolution of the mass spectrum, combined with moderate energy filtering, have been so far the most common SIMS procedure for REE analysis in geological materials (e.g.. References [12,13]). In recent years at CNR-... 

The processing, so-called spectral stripping (see, for example, Ellis, 1987 Hearst and Nelson, 1985), solves a matrix algorithm for the concentration of... 

Multicomponent analysis is used to find the concentrations of various species in a mixture, in a sense it is automated spectral stripping although you never see successively substracted spectra. Standard spectra must be available. See ref. 9 for details, manufacturers usually supply software. 

---