GC Retention Index

Quick Start

The most popular way to predict a GC retention time is to do it via the Kovat's Retention Index which is normalized to values of standard alkanes. The app predicts this from the HSP values of a molecule.


BPt °C

It would be useful if one could predict the retention time of a molecule within a GC system. But each system is so different that it would need many different parameters and complex parameterisation to be able to specify an actual retention time. Much more practical is to predict a relative index and a good set of standard materials to act as references are the normal alkanes. If you have a standard set of conditions where you know that, say, octane to hexadecane come out in reasonable times and if you can predict that a molecule will elute around, say, decane that is useful information.

A good way of doing this is via Kovat's GC Retention Index, GCRI:

GCRI = 100 * ( log(X) - log(A))/(log(B) - log(A)) + n * 100

This means that you have an alkane with n carbons which elutes in time A then the alkane with n+1 carbons elutes in time B and your molecule elutes in time X between A and B. If n=9 then n+1=10 and so the GCRI of your molecule will be somewhere between 900 and 1000.

In practice what this means that if you have a predicted GCRI of, say, 950 you will set up your GC so that Heptane to Dodecane are eluted quickly and with good resolution, so that when you inject your molecule there is a good chance of finding it comfortably within the conditions you have set up.

There are a number of schemes for predicting GCRI. Because of my interest in Hansen Solubility Parameters which capture the essence of a molecule in the three parameters: δD = Dispersion (van der Waals); δP = Polar; δH = Hydrogen bonding, I have used an HSP-based scheme. My colleague Dr Hiroshi Yamamoto has devised a scheme for estimating the GCRI depending only on δP, δH, BPt and MVol (Molar Volume). Because the MVol effect is relatively small, in this app only the first three parameters are needed. Of course this raises the issue of where to get estimates for the three parameters. You can find tables of known HSP values at the Practical Solubility site and various schemes for estimating them are known, and BPt estimation schemes are also available.