Unknown UV peak in blanks.

[excelsioranalytical]

We have a persistent issue with our Thermo Accela HPLC with PDA (UV) detector. Buffer A is 50%/50% water/methanol with 0.1% formic acid and buffer B is 100% methanol with 0.1% formic acid. We are interested in neutral and acidic compounds so the formic acid keeps our retention times consistent. This is a well known, validated method for the analysis that we are doing. The problem is that we get a UV peak at about 236 or 238nm with a smaller noisier secondary peak at about 218 or 220nm when buffer B gets above a certain level. The absorption continues to grow as the concentration of buffer B grows all the way until 100% buffer B which is the end of the gradient. The signal is very consistent run to run. I'm not sure how to post pictures here or I would post some screenshots but I hope this description is good enough.

My understanding was that methanol and formic acid should not absorb in the region where we see this peak either alone or together.

Things we have done to troubleshoot that eliminated the peak:

-use only buffer A isocratic
-pure water
-pure methanol

Things that have not eliminated the peak:

-replace lamps
-calibrate dark current and wavelength
-clean system (including flow cell) with 20% nitric acid
-replace buffer bottle
-replace bottle filter


We are stumped now. Any help would be appreciated.

[tom jupille]

Check out the quick tutorial on gradient baseline problems here:
viewtopic.php?f=31&t=19085

That said, you have already eliminated some of the suspects. At this point I'd say the most likely culprit is your formic acid. Try a new bottle and/or a better grade. Other things to look at:
viewtopic.php?f=1&t=10655 (I know you're using methanol, not acetonitrile, but still relevant)

[Gerhard Kratz]

In addition to Tom: did you check flow rates of your channel A and B?

[kubowicz.tomasz]

Hello

If you have this problem only when you're running gradient it looks like something is coming out of column when elution strength is going higher ("The problem is that we get a UV peak at about 236 or 238nm with a smaller noisier secondary peak at about 218 or 220nm when buffer B gets above a certain level. The absorption continues to grow as the concentration of buffer B grows all the way until 100% buffer B which is the end of the gradient")
So first I would check column.
Are you using reference wavelength for your method?

Regards

Tomasz Kubowicz

[HPLCaddict]

The problem is that we get a UV peak at about 236 or 238nm with a smaller noisier secondary peak at about 218 or 220nm when buffer B gets above a certain level. The absorption continues to grow as the concentration of buffer B grows all the way until 100% buffer B which is the end of the gradient.

From this description it sounds as if your problem is rather an upward drifting baseline and not a particular ghost peak? Could you please clarify? An example chromatogram would be helpful...

[dr-statz]

You can use an Ghost-Guard or Ghost-Trap column to avoid ghost peaks. Such Guard-colums are installed before the auto sampler. The absorber material in the column absorb all kind of impurities in the mobile phase (caued by solvents, buffer, HPLC-pump, mixer, fittings ...).

We use the ghost-guard column all kind of gradients and get good results! You should try it!

[excelsioranalytical]

Thank you all for the responses I'll go through them one by one:

The formic acid is Fisher LC/MS grade. Not even running the MS right now, just the PDA, but I suppose there could be a contaminant. I'll look into a new source for the formic acid. I was starting to suspect this as well.

I have checked the flow rate by collecting the output and the flow rates are dead on.

The column is brand new and this problem has been exhibited with different types of columns as well as another one of the same column. FYI it's a Phenomenex 2.6um particle 150mm length C18 with a 2cm guard column with the same funtional group.

HPLC addict, yes it's an upward drifting baseline not a particular ghost peak. When I say peak I meant in the wavelength axis not the time axis.