[Zurück]

W. Linert, A.B. Koudriavtsev:
"Do experimental errors really cause isoequilibrium and isokinetic relationships?";
Bericht für Jahresbericht des ZID - phoenix-cluster; 2013; 20 S.

Numerical experiments simulating spectroscopic determination of equilibrium constants with normally distributed random noise in spectral amplitudes and temperatures have shown that the `statistical´ correlation between estimates of ΔH0 and ΔS0 only appears when measurements are performed on systems with ΔH0 and ΔS0 varying within an interval smaller than two standard deviations (σ ) of their estimates, i.e. when the precision of an instrument or a method is insufficient for studies of such small variations of chemical reactivity. Random variations of ΔH0 ±σΔH and ΔS0 ±σΔS completely destroy statistical correlation. This means that in properly designed experiments the experimental errors cannot be the origin of the observed correlations between ΔH0 and ΔS0 . This has been proven by numerical experiments on reaction series exhibiting compensation effect. When analysing experimental data of unknown precision the Exner- Linert method can be advantageously used to distinguish the `statistical´ compensation effect.