Chemical Research Institute of St. Petersburg State University
Universitetsky pr., 2., St. Petersburg 198904, Russia
e-mail: gc_lab@nonel.chem.lgu.spb.su

Typical design on numerous contemporary algorithms for pre-calculation of different analytical parameters (including GC retention indices, RI) from other constants (those are determined more detail and/or less available) implies obtaining of single estimations:

Unfortunately, the evaluation of these single results correctness (quality control) often is impossible in the framework of this classical concept. Its rejection permits us to establish a new class of algorithm for the pre-calculation of RIs of various organic compounds on standard non-polar stationary phases (polydimethyl siloxanes).

Initial data for RI pre-calculation:

large and well systematized collections of physico-chemical constants: boiling points (T_b), refractive indices (n²⁰_D), densities (d²⁰₄) and molecular weights (MW). The values of n²⁰_D, d²⁰₄ and MW are necessary for the calculation of molar refractions (MR_D) those are proportional to molecular polarizabilities and reflect the van der Waals' contribution of sorbent interactions.

By analogy with linear interpolation formulae for calculation of GC retention indices, every compound with known MW, T_b and MR_D may be characterized by three "like-IR" parameters:

(P.G. Robinson, A.L. Odell, 1971)

(M.B. Evans, J.K. Haken, 1986)

(I.G. Zenkevich, L.M. Kuznetsova, 1990)

Moreover, in addition to the mentioned parameters the best linearization of function RI(T_b,MR_D) must be taken into account: log RI(T_b-MR_D) = a log T_b + b MR_D + c (reference homologous series – n-alkanes or others). The comparison of all four pre-calculated pseudo-RI values on the base of set of logical criteria is the most important part of RI_precalcd evaluation:

1. The case RI(MW) >RI(T_b) > RI(MR_D) is typical for non-polar compounds (without active hydrogen atoms and/or functional groups CO, CN, NO₂, S O, SO₂, etc.). Observed RI_experim values are in accordance with following inequality: RI(T_b-MR_D) < Ri_experim < RI(T_b). So far as RI(T_b-MR_D) RI(T_b), the high correctness of results is guaranteed.
2. Other inequality RI(T_b) > RI(MW) > RI(MR_D) is typical for polar compounds. In this case the window for RI_experim values is very large, but no exceptions have been found after the testing of data for about 2,000 examples: RI(MW) < Ri_experim < RI(T_b). There are no correct rules to determine more precisely the "position" of RI_experim between low and high limits of this window and only partial recommendation may be proposed.

Important examples of GC Ris precalculation with new algorithm are discussed.