GPC- Peaks suddenly elute early, poor resolution

[whf4]

Hi,

sorry if this is an obvious question, but I am not experiencedin chromatography (yet).

I am using a Tosoh TSK-GEL column for GPC of dextrans, with phosphate-buffered saline as the solvent. We (finally) got our refurb waters system stable and were getting repeatable chromatograms of our dextran mixtures. We had been in the habit of leaving the column with a trickle of flow when not in use- like 0.1 ml/min.

After an interval of several days, the chromatograms suddenly elute off early, and are bunched together-- ie peak resolution is poor. This suggests channeing in column-- an unhappy observation in a fairly new column!

Anything one would recommend short of a new column to trouble shoot this furtehr? We verified pump flow rates,a nd pressures are not different than previosuly.

thankds!

Bill

[HW Mueller]

Do small molecules come earlier also?

[whf4]

yes, the whole spectrum is shifted to earlier elution-- at least for the 3kD dextrans. As I am running isocratic for GPC, I can't tell if for example a solvent peak moved.

[Uwe Neue]

The simplest thing is to do a platecount on the column. The certificate of analysis should have an example on what to use.

If the plate count peak is terribley distorted, it could eb a column void.

What was the absolute elution time of your standards at what flow rate, and what is it now? And give the column dinesions too.

[whf4]

Hi,

I am running a 30cm 7.8mm TSK-Gel column at 1.0ml/min.

I am not sure how easily I can generate a plate count as we are using very broad size distribution dextrans-- even the size standards are a bit broad.

The spectrum used to span 400kD - 3kD and woudl elute off startinjg at 4 min and ending at about 12 min. It now takes off at 3 min and is over by 6 min.

If you have a suggestion regarding how to estimate a plate count, that would be very helpful-- I realize this is basic thing I haven't done

thx!

[Uwe Neue]

The original numbers are not unreasonable, but you have some elution of peaks earlier than what I would anticipate.

The new data are difficult to interpret.

Look for the certificate of analysis. It should have a plate count method.

If you don't: use a sugar as your plate count marker. Make up a solution that your detector sees in the mobile phase that you are using, and report back what retention time you are getting, and if the peak looks like a normal narrow symmetrical peak, and what the plate count is...

[Uwe Neue]

The software that you are using should have ways to calculate the plate count. If it doesn't:

Make a printout of the sugar peak. Draw tangents on the sides of the peak. Then count the plates by the following formula:

Plates = 16 * (tr/w)^2
tr is the retention time, and w is the width of your peak at the bottom of the tangents.

I appologize, if I have been too basic.

[HW Mueller]

What I meant with small molecule is akin to what you had in mind when you mentioned solvent retention. It shouldn´t be difficult to find a material in the lab which you could detect and which is very soluble in your mobile phase. If this comes where it should your problem is not channeling.

[whf4]

Hello,

Drs Mueller and Neue, thank you for your sugegstions. We'll do that today.

[whf4]

Sooooo.....


at the risk of irritating forum participants with the epiphany of the obvious, we did as suggested and tried to calculate a plate number-- and got something like six. Yup. six-point-oh-times-ten-to-the-zero, with a couple of different solutes. Even 20uL of distilled water came off at 5 minutes and lasted 10 minutes.

MFR quotes 1.2x10^4. plates

The chromatogram was between absurd and obscene.

Acting on a hunch, we tore down both pump heads and found a whole pile of seal fragments inside one of the heads.

We replaced both seals, and for the first time ever in our history of the instrument, we have a flat pressure tracing-- the guy who did the install told us that the comb with 60% variation from peak to trough was 'normal'. This may explain why we were trashing columns as fast as we were.

we replaced the precolumn filter, the guard column, and the column, and we'll try to repeat our plate count.

OTOH, if we had just done that when we receioved the instrument, we would ahve saved 6 months in figuring out that somethign was REALLY wrong.

thanks for your help, and if it's okay, I'll report back on the new plate count.


Will pressure spikes as I describe collapse/trash a column?

[Uwe Neue]

Polymer-based columns, especially GPC columns with their large pore volume, do not like it very much when the pump constantly hits them on the head. Most column manuals ask for SLOW changes in the flow rate for these columns.

[HW Mueller]

Nice demonstration for the routine measurement of flow rate and determination of tm (here the small molecule stuff).
Incidentally, there is no excuse for not measuring the flow rate regularily and often, just did one in 18.4 sec with a "home built" device: Chopped off pointed end of a 200µL graduated pipet to put teflon reducing (to 1/16") fittings on it, a T on the lower end so that a syringe can be attached. The syringe is used to push in an air bubble the traverse of which is followed via a stop whatch. (Idea comes from the soap bubble GC devices, of course).

[whf4]

Welll.


did measure flow rates very frequently-- whaty we did NOT do was heed ther pressure tracings.

still rebuilding system.

thanks

[HW Mueller]

You probably measured the flow rate via collecting a given volume.....? Too much averageing there. Or did your column average the flow? No waves in the baseline? (I am just trying to elucidate what sort of detection and flow measurement would not show this pumping deficiency immediatly)

[whf4]

thank you for your help--- please see followup post abouit pressure problems....