Random loss of retention

[Mattias]

Hi,

I am running a standard reversed-phase separation on some extraxted tablet samples. The problem is that I get completely loss of retention is some injections. It seems to occur randomly - and the performance of the column in the other injections is not showing any problems (constant and normal plate counts).

When I reanalyse the samples - I get perfectly normal chromatograms. Could this be a problem with the autosampler - or with the column?

The sample is dissolved in 0.9% NaCl and the mobile phase is a mix of phosphate buffer and acetonitrile (about 18%).

[danko]

Premixed mobile phase (no mixing online) – right?

[Mattias]

No, this is mixed by the instrument (Waters Alliance 2690)

[Hfranz]

I would question the proportioning valve of the pump. Replace it and see if the situation remains.

Cheers,

Holger

[juddc]

Holger, I disagree with your approach. First one must track down the source of the problem, then replace parts only if necessary. Don't replace bits as a first line diagnostic tool - that's far too expensive! It's easy enough to test the proportioning valve. I'd recommend testing it then replacing it only if necessary.

When having such problems, one should systematically go through your instrument and method to determine where the problem lies. If your GPV is changing the MP composition on you, you should see some abnormal change in pressure on your column. Make sure your instrument is working as it should, then make sure that chromatography is consistent on clean standards - or another method even - then check samples again.

An alternate possibility is that some excipient is extracting from your tablets, coating your column, and causing irregular retention. Maybe there's a sample or mobile phase pH issue. With the information as given, it's almost impossible to tell.

[unmgvar]

is your loss of retention completly random?
when you look at those bad chromatograms, are the pictures giving you a trend or each one looks different from the other?
based on the first description i would go for 2 possibilities:
the column get coated with something
the porportion valves are bad.

because the issue comes and goes then you will need to check each issue over a long period of time.

my guess for the proportionning is that is is not really faulty but clogging.
first look at your solvents. shake the bottles. see any weird stuff floating around? if yes then change the slovents, purge your system.
if not then, using your method mobile phase run a gradient from buffer to ACN. from 0 to 100% say in 5 min or so . no samples. run it several times. if the proportioning is faulty you will see it in the baseline of the chromatograms.
because it occurs every few injections it generally means;
bad miscibility between your 2 solvents. salts plugs the valves until it is released.

if the column gets coated then try this:
do a 1% acetone in mobile phase sample.
inject about 10-20 ul also change the wavelenght to 254 nm. peak will generally elute in the first 5 minutes, within 10 minutes max.
keep everything else constant.
inject that many times for a long time, as long as it take the problem to occur.
nothing happen. then it is your sample coating the column.
the problem still occurs look somewhere else.

[YMC-Kune]

Just to be sure it is not something simple as
- your injection needle blocking from an exipient present in the sample (I expect you filtrate your samples before injection)
- a faulty entry for the injection volume (e.g. 1 instead of 10 µl injection) in the autosampler table.

I am just asking, as both scenarios have happened before!

I would look at the HPLC column as a consumarable and install a new one pretty early in the troubleshooting process. Unless you are using extremely expensive columns, one day looking for an error will cost you nearly as much as a new column (operator cost, instrument down-time, delay in generating results, maybe rework of samples because you exeeded sample stability). If, after changing the column it works bin the old one. If it does not work the column is not the problem and both columns are very likely still good. In this case you lost about 1 to 2 hours but nothing more and are sure to have eliminated one parameter.
(And I am not saying this because I am now working for a column manufacturer)

[Mattias]

Thank you for all your replies!

When I took a closer look at the flawed chromatograms, I realised that the peak is not eluting in the front - but after 3-4 minutes (normal retention time is 17 minutes). The peak is also much bigger in area than expected (about then times!).

I am starting to wonder if the system has injected a large airbubble rather than my sample in those injections. There was plenty of sample in the vial - so it was not related to the vial.

[HW Mueller]

If you think it´s air, just inject different amounts of air, or strongly airated samples, and see what happens.
It would be quite instructive for the rest of us if you told us how ~ a ten? times larger peak escaped your immediate attention.
It probably covers your analyte?

[Mattias]

It sounds strange, I must admit. The software autoscales the chromatograms - and the early peaks were not integrated (due to intergration events). The peaks looked quite identical by a glance, the air peak narrower (as expected by the retention time).

It has now been confirmed as air. Some questions raised by this:

Has air retention on C18 column?
How can air elute as a symmetrical positive peak?
Where did all this air come from?

[tom jupille]

Air is indeed retained on a C18 column. See the "LC Troubleshooting" article by: Yoshihiro Egi and Atsuro Ueyanagi
LC-GC 16(2) 112-118 (1998)

Here's a figure from that article:

[Mattias]

That is quite amazing! Makes you wonder if you even can analyse air in samples using fully degassed MF...

I verified the air peak by injecting from an empty vial. The peak was huge but recognisable.

In my case, the main peak is not there at all in the "air-injections" so I assume that all 10 µl injected was air in those injections. I have no idea where this has entered the injection loop.

[HW Mueller]

Seems to happen quite often: One is temporarily fooled by the elctronics which put the largest peak on scale of the screen.
A quantitative analysis of air in the sample might be difficult if not impossible. It appears to be caused by light scattering and who knows what else. Attempts to repeat "standard" injections did not yield good results (put mildly) in this lab. Strangely, we have gotten very similar, interfering air peaks in sample injections over some periods of time, but were unable to repeat it some other session (in other words: We didn´t really have the air problem under control).

[tom jupille]

Hans, I'm not sure about the light scattering. That article had plots of absorbance spectra for degassed vs. air-saturated solvents that showed a shift to longer wavelength for the air-saturated. This is for MeOH:


Would light-scattering behave this way (this is getting outside my area of competence!)?

[danko]

A genuine “air solutionâ€