Chromatography Forum

Apologies if this is the wrong forum to post this under, I've been thinking about this question since I was an undergraduate and up till now, I can't seem seem to find an explanation for the matter.

When we have a carrier gas we also know its optimal linear velocity e.g. hydrogen has an optimal velocity near 35cm/s. Theoretically, would this be calculated using the square root of B/C in the van deemter equation?

As for the analyte, I believe it also has its own optimal velocity by taking the square root of B/C. And this value would be different from the carrier gas' optimum velocity. So why is it that we tend to say that hydrogen or helium are preferred over nitrogen as it has a higher optimum velocity?

Essentially, why should we even be bothered by the carrier gas' optimum velocity? Shouldn't we just set the velocity based on the analyte's optimum velocity? I don't quite understand the purpose of having to consider the optimum velocity for the carrier gas.

Thanks in advance for the help!

And this value would be different from the carrier gas' optimum velocity.

No, it would not because the B and C terms cannot be estimated for the mobile phase without assuming (or measuring) the properties of the analyte.

And this value would be different from the carrier gas' optimum velocity.

No, it would not because the B and C terms cannot be estimated for the mobile phase without assuming (or measuring) the properties of the analyte.

Hi Tom, thanks for the reply. So in a practical case, we wouldn't know the actual analyte's optimal velocity. If that's the case, how would the carrier gas' (mobile phase) optimal velocity be of any significance to that of the analyte?

There is no such thing as "carrier gas' (mobile phase) optimal velocity" in the absence of an analyte (think about it: no analyte = no peak to measure).

As it happens, both the C and B terms in GLC are dominated by diffusivity (along the column or into/out of the liquid phase) so if the molecular weights are comparable, then the optimum linear velocity will be similar for all your analytes.

On a practical level, other than academic/theoretical studies, I have never seen a case where anyone wants to run at the optimum velocity. In the real world, the goal is not to obtain optimum resolution, but to obtain adequate resolution in minimum time.

If I don't have adequate resolution, increasing the plate number is not the best way to improve it. Resolution is roughly proportional to the square root of N so that doubling the plate count improves resolution by roughly 40%. On the other hand, resolution is proportional to [(alpha -1)/alpha] so that a 10% increase in alpha roughly doubles the resolution (I'm assuming an alpha of 1.1 here for simplicity). In other words, if I'm "adequatizing" a separation, I will pay more attention to selectivity than to efficiency and so I run at the highest practical flow rate. Only if the resolution is borderline (i.e., I only need a small improvement) will I look at decreasing the flow to eke out an improvement in N.

Thank you Tom for the answer it really cleared up the misconceptions I had about this. May I check then, for the typical reported values for the optimum velocity for the various carrier gas, is there a particular analyte they measured it with? Since as you mentioned if the analyte has to have comparable molar weights, realistically wouldn't that result in a large range of velocities since the molar weight range also be quite large?

Thanks again for the help

First of all, the last time I ran a GC was in 1977 (when I started doing LC), and the last (and only) time I ever actually measured the optimum flow was in 1968. We actually used lighter fluid -- basically butane/pentane/hexane. If someone were publishing a comparison today, I imagine they would do something similar. Lighter fluid is easily obtained and the alkanes generally have very few secondary interactions that would cause broad or tailing peaks. But TBH, I've never really looked.

I see, thanks a lot for the clarifications Tom! It really helped clear up the misconceptions I have.

I see, thanks a lot for the clarifications Tom! It really helped clear up the misconceptions I have.

In practical use, I take the "optimum" linear velocity as a starting point then adjust until I get the results I need for the test. As Tom mentioned if you are separating analytes that are very similar then they will optimize at near the same flow rate. However if you are doing something like I do every day where you are separating 175 analytes ranging in boiling point from about 60C to 300C and having functional groups of alkanes, alcohols, amines, aromatics, and all of these both halogenated and non-halogenated, well, there is no one "optimum" and what you end up with is a compromise to get the best separation in the best time, with the best peak shapes possible.

When I first began doing this 30 years ago I found out quickly that theory and classroom knowledge is only the beginning when you start doing real world work