The wavelengths which contribute the most towards defining the stability constants correspond to peak maxima in the spectra of the individual species, both the free reagents and the complexes. However, the spectra of the species are initially unknown so this criterion is difficult to apply in practice. Instead, wavelengths should be chosen which show the most variation in intensity values.
In the example above wavelengths longer than 344 nm have been deselected. Also wavelengths with very high intensity and consequent high noise level should not be selected. The option to use every second wavelength etc. is available to speed up the stability constant refinement by reducing the amount of data to be processed. Also, Increasing the number of wavelengths selected does not necessarily give better results. The reason for this is that the molar intensity values at each wavelength are quantities that have to be determined, so increasing the number of wavelengths selected also increases the number of unknown quantities.
If there is a large region where intensity is negligible, those wavelengths should be removed. This will improve the appearance of the plots.
In the case of spectrophotometric data the plot of intensity is essentially 3-dimensional with two independent variables, the measurement number and wavelength. HypSpec shows two planes: in one plane the data are shown for a single wavelength as a function of data point and in the other plane the data are shown for a single data point as a function of wavelength. The two plots are linked so that changing the cursor in one plot will cause the other to be re-drawn.
The objective of manual fitting is to reduce the systematic trends in the residuals by adjusting the stability constant values. A residual is the difference between an observed and calculated data value. In this case there are residuals on both the point and wavelength plots. The point plot (optionally) shows the speciation. The wavelength plot shows the contribution of each species to the calculated intensity.
To illustrate this phenomenon consider absorbance data for which Beer's law applies.
A=Seici
where ei is a molar absorbance and ci is the concentration of a species. If the species is a complex of formula MpLq, the concentration is given by
cpq = bpq[M]p[L]q
Thus in this example
A = eM[M] + eL[L] + S(epqbpq)[M]p[L]q
epq and bpq are highly correlated because they are multiplied together in the Beer's law expression so that an increase in one can be compensated by a decrease in the other with little effect on the calculated absorbance. This is an inherent property of spectrophotometric data.
Contents > Computing: > General | Potentiometric | Spectrophotometric | NMR
