23 Inline and Online Process Analytical Technology with an Outlook …
567
23.3.3 Selectivity: Classification of NIRS Within
the Spectroscopic Toolbox
The selectivity of a method is defined by how the method can determine certain
analytes in a complex mixture without interference from other components. Therefore, if the signal of the analyte can be separated from signal of the interfering
substance, the sensitivity in complex mixtures increases significantly. Several authors
define a so-called net analyte signal (NAS), which describes the part of a sample spectrum that is orthogonal to the spectra of all other components in the sample. This
is also called “the unique spectral signature for the analyte of interest,” i.e., the part
of the instrument signal that is not lost due to spectral overlap. This underlines the
importance of finding peaks of the analyte at wavelengths that are clearly separated
from the residual matrix [1, 33, 41, 42].
Diode array or Fourier transform spectrometers show a fixed optical resolution
over the entire wavelength range. Dispersive scanning spectrometers often compensate the intensity of the illumination (reference) by changing the slit width. This
means that the optical resolution and thus the selectivity changes over the wavelength and depends very much on the absorption range. It is important to stress that
the optical resolution is not equal to the pixel resolution.
23.3.4 Robustness, Detection Limit (DL),
and Signal-To-Noise Ratio (SNR)
23.3.4.1 Definition of Robustness and SNR
The ICH quality guideline (International Council for Harmonization, ICH Q2) on the
robustness of an analytical method states [10]: “Robustness of an analytical procedure
is a measure of its capacity to remain unaffected by small, but deliberate variations
in method parameters and provides an indication of its reliability during normal
usage.” This statement therefore accepts errors due to fluctuations of the method
parameters, but they must be within given limits of the methodology. Furthermore, the
statement for a system qualification also includes the optical setup, the wavelengthdependent detectivity of the sensors and the wavelength-dependent illumination,
which determine the overall signal-to-noise ratio [33, 43, 44].
The signal-to-noise ratio (SNR) is defined as
SNR = (S−B)/N SD
With S = average value of the signal, B = average value of the background,
N SD = standard deviation (SD) of the noise amplitude.
567
23.3.3 Selectivity: Classification of NIRS Within
the Spectroscopic Toolbox
The selectivity of a method is defined by how the method can determine certain
analytes in a complex mixture without interference from other components. Therefore, if the signal of the analyte can be separated from signal of the interfering
substance, the sensitivity in complex mixtures increases significantly. Several authors
define a so-called net analyte signal (NAS), which describes the part of a sample spectrum that is orthogonal to the spectra of all other components in the sample. This
is also called “the unique spectral signature for the analyte of interest,” i.e., the part
of the instrument signal that is not lost due to spectral overlap. This underlines the
importance of finding peaks of the analyte at wavelengths that are clearly separated
from the residual matrix [1, 33, 41, 42].
Diode array or Fourier transform spectrometers show a fixed optical resolution
over the entire wavelength range. Dispersive scanning spectrometers often compensate the intensity of the illumination (reference) by changing the slit width. This
means that the optical resolution and thus the selectivity changes over the wavelength and depends very much on the absorption range. It is important to stress that
the optical resolution is not equal to the pixel resolution.
23.3.4 Robustness, Detection Limit (DL),
and Signal-To-Noise Ratio (SNR)
23.3.4.1 Definition of Robustness and SNR
The ICH quality guideline (International Council for Harmonization, ICH Q2) on the
robustness of an analytical method states [10]: “Robustness of an analytical procedure
is a measure of its capacity to remain unaffected by small, but deliberate variations
in method parameters and provides an indication of its reliability during normal
usage.” This statement therefore accepts errors due to fluctuations of the method
parameters, but they must be within given limits of the methodology. Furthermore, the
statement for a system qualification also includes the optical setup, the wavelengthdependent detectivity of the sensors and the wavelength-dependent illumination,
which determine the overall signal-to-noise ratio [33, 43, 44].
The signal-to-noise ratio (SNR) is defined as
SNR = (S−B)/N SD
With S = average value of the signal, B = average value of the background,
N SD = standard deviation (SD) of the noise amplitude.