RT variations

[FredMSD]

Has anyone ever had a RT that fluctuates in both directions from one injection to another (up to 0.1 min)? What could cause this ? The autosampler has been checked and the synchronization with the MS acquisition is fine. It's related to one compound and has happened using 2 different columns, no pb with other compounds... Any advice will be VERY appreciated !

[danko]

First: 0.1 min is not that bad at all. Second: The fact that you only see the variation for one compound and not for another tells me that these 2 compounds elute under different conditions (e.g. the first under isocratic and the second under gradient conditions).
If you really want to do something about it, you should focus your efforts on the pump/s, gradient valve (GPV) etc.

Best Regards

[FredMSD]

+/- 0.1 min is not acceptable in terms of GLP validation
The other compound was analyzed under isocratic condition as well so I'm 100% sure the pump and valves are working fine

[tom jupille]

A number of questions here:

first of all, was this a gradient or isocratic separation? (From context, I'll assume that the separation in question was a gradient.)

second, when you say that only one compound shows the RT variation, are you running several compounds in the same run?

next, (again assuming a gradient) are you equilibrating for exactly the same time before every run?

(again assuming a gradient) are you using a high-pressure ("two-pump") or low-pressure ("one-pump") mixing system?

finally, approximately what are the retention time and peak width of the peak in question?

Here's what I'm getting at with all of these question:

1. If you have a relatively large column, with long retention and wide peaks, a 6-second variation may, in fact, not be a significant problem.

2. No gradient system generates a perfectly smooth gradient; there are always fluctuations around the average composition. Low-pressure mixing systems (on average) seem to be more prone to this problem, but it's always there so some extent.

3. Some compounds are more sensitive to mobile phase composition than others. That means that small fluctuations in composition may make some peaks move quite a bit while others are almost unaffected. In general, higher molecular weight compounds are more likely to move. The same thing applies to stationary phase equilibration.

All that said, the first thing I would do is to run a "stair-step" test. There is a brief write-up in the Troubleshooting Wizard on our web site (http://www.lcresources.com/resources/TSWiz/hs450.htm).

If you're running a narrow-range gradient (e.g., 20 - 40%), you may be able to sidestep the problem by putting your gradient start and end solvents in the A and B reservoirs and running from 1 - 100%.

[Tomasz]

Tom

Fred wrote
"The other compound was analyzed under isocratic condition as well "
I conclude from this that it was rather isocratic analysis.

Fred

Are you using buffered mobile phase? Some compounds give fluctuation in retention time in unbuffered conditions.

[tom jupille]

I conclude from this that it was rather isocratic analysis.

Oops! That's what I get for being in a hurry.

That said, if you are set up for "on-line mixing" (i.e., having the pump control the mobile phase composition), the same comments and suggestions would apply.

[Bruce Hamilton]

I'd agree about the buffered mobile phase, and also suggest ...

- does the variation affect a pure standard of the compound as well?.
- is the sample in a solvent similar to the mobile phase, including any buffer?
- is the sample clean?, as extraneous junk can affect RT.

Please keep having fun,

Bruce Hamilton

[FredMSD]

Thanks to all for your replies. I solved the problem, it was the binary pump (Agilent 1100) that was not delivering a constant composition (!). I found that by using only one channel and one bottle of solvent containing A & B. Anyone having met this problem before ?

[tom jupille]

Anyone having met this problem before ?

It's not that uncommon. The same kind of problem was discussed in one of John Dolan's LC Troubleshooting columns in LC-GC:
LC-GC 24(7) 662-668 (2006)