Chromatography Forum

No problem Mr HPLCConsult, I was not trying to step on your toes – Just thought this was interesting and merely trying to suggest how a pulsating flow can cause the detector response to give symmetrical variations to the peak in the chromatogram whilst leaving the baseline flat. Not what I expected and possibly unusual to others too?

The more explanations I see the more puzzled I am. Carls pointed out his experience of seeing noise in the baseline with checkvalve problems. I can second that, as a matter of fact, I have never seen a change of flow rate that did not show in a shift in the baseline, there seems to be a complicated relationship to absorbance. It could be that the scale chosen by atram is too insensitive to show the shifts (here pulsation) in the baseline.
What about the symmetry? Note especially the apex of the two peaks. I would like to see the individual data points of this chromatogram. Also an expansion of peaks which were obtained after the pump fix might be interesting.

Perhaps this problem is puzzling to most since its rare that an HPLC pump still "works" while delivering half the flow rate specified. My understanding is the Agilent 1200 quaternary pump has identical dual pistons. One piston is the "low pressure" side and runs at twice the speed of the second "high pressure" side. In order to get half the flow rate specified both pistons would have to be leaking. To get the saw tooth pattern due to flow variations alone the flow would have to vary substantially (and thus pressure). These dramatic flow variations will cause the UV signal to vary significantly since the signal has a time component AU*s, i.e. depends on the time in the detection path. The signal variations are not visible in the baseline at the scale shown since the phosphate mobile phase does not absorb at the detection wavelength used. The flow variations become apparent when an absorbing component is present in the detection path. When the flow is low the signal increases due to longer residence time in the detection path and when the flow is high the signal decreases due to reduced residence time in the detector.

Just thinking out loud.

That´s exactly what I don´t see, the residence time influencing the absorbance intensity if the setting of time constant of the detector and data point rate of the computer are not completely inadequate. The baseline fluctuations with flow rate that I have seen appeared to be primarily a refractive index or scattering phenomenon that varies with the absorbance in a complicated way (not sure on that, maybe it just has to do with the difference in refractive index when an analyte is present).
Could it be that the chromatogram shows two different problems at once?

I tried to provide a simplistic explanation to the problem posed as there is insufficient data for me to speculate as to all of the reasons why the chromatogram appeared as shown. We have not been provided with enough information as to what the chromatogram looked like before the problem appeared; what the 'Y' axis looks like where the noise is; what the details are regarding the type of sample used; details about the analysis conditions or what the chromatogram looked like after the "fix". Based on this, I simply related that the solution to the problem is to fix the pump first and then move on from there to address any other issues (UV plot).

Yes, the Agilent/HP pumps use dual speed pistons and when you couple this with a computer controlled valve running the show some very interesting things can happen. As I mentioned before, you can actually have one piston BROKEN and the system will report normal flow (not actual flow) because the computer tries to compensate for the problem using the one functioning piston and AIV. Many years ago, I actually visited a customer's site whom reported that they were having trouble reproducing an established HPLC method with exactly double the normal retention times resulting (again, this one fact should have tipped the user's off that their pump was malfunctioning). They claimed to be very experienced chromatographers (at a major pharma company) so imagine the shock when I looked at their Pump and found solvent leaking from one of the pistons. White ground up powder appeared just below the opening in the pump head where the broken piston resided. The end of the broken piston was stuck in the pump seal. A quick check with a graduated cylinder found the flow rate was exactly half of what the pump/computer reported. So why did the system not detect the leak ? Answer: The clients had known about the leak for weeks and solved the problem by stuffing paper towels underneath the pump head to absorb the liquid and evaporate it quickly before the leak detector would sense it (I have also seen customers simply reorient the leak sensor so it could not detect a leak too). The pump was still able to deliver some of the solvent to the column and because the sample was concentrated, you could not easily notice the extra noise on the baseline axis unless you zoomed in.

This is a great example of how some basic knowledge of how things work and an always reliable glass graduated cylinder will always be valuable tools.

"carls": I just read your earlier post so here are some quick comments.

Regarding: "Since you have a grip on this problem could you explain how a bad check valve would only shown noise when a peak is eluting, i.e. noise on baseline much lower (non-existent) than that observed on a peak."

<Great point, but please remember that this is not just a check valve issue as this system has a computer controlled one. An old school mechanical check valve problem would indeed make everything show the pattern to the same degree. This system is different. As mentioned before, 'real' baseline noise is not visible on the poster's chromatogram so the noise is there, we just can not see it at the scale presented. As the peak elutes the time delay and slope might change the shape of the line seen (?). What does the sample look like w/o the spikes when the pump works fine ? Are their oligomers under there ? Clearly something is wrong with the solvent delivery system so I choose to ignore the other issues for now. I agree it is odd, but without more info I can not speculate.>

"I have another idea. Is it possible the reference wavelength is set to a wavelength range where the analyte also absorbs and thus exaggerates the pump noise when a peak elutes?"

<This is certainly possible and I am glad to see someone bring it up! Using the reference wavelength feature (for those using it, please turn this feature OFF for 99% plus of your samples) you can indeed get some very odd chromatograms as you will be adding or subtracting other potential peaks to your chromatogram and the resulting plot can come in all forms ! I have seen it many times. Again, we do not have any other info from the poster as to what their settings are so I ignored this point initially because the initial problem was 100% pump related, not detector. I like to fix one problem at a time. I would make an educated guess that the initial poster has multiple problems/issues with their system. Not just the pump.>