UV spectrum background problem

[BurntChem]

Also thanks for the pdf Matt.

[BurntChem]

Since we are on the topic I am wondering what exactly is the relationship between: filter rise time (sec), sample rate (Hz), filter bandwidth (nm), and wavelength step size (nm) in the Thermo Accela PDA. I am having a hard time understanding how to optimally set these parameters and if they might be source of my background problem.

Originally I had the system set at:

Rise time- 1 second
Wavelength step- 1 nm on a 200-400nm scan
Filter bandwidth- 1 nm
Sample rate- 20 Hz

The run time of the method is 11 minutes flow rate 0.7ml/min and the pressure typically starts out at about 450 bar and decreases into 300's.

The width of the largest peak which is one of the analyates of interest is 0.3 min (18 seconds) while the width at half peak height is 0.12 min (7.2 seconds).

Any help on this would be appreciated as I find the Thermo help file to be rather cryptic.

[sga]

hmm !

[BurntChem]

Tough problem right? heh

[BurntChem]

Hello I am bumping up this topic because we are still experiencing this UV background problem and I want to try and tackle this issue before a formal method validation is undergone.

I've concluded based on what I've posted above that problem is likely a contaminated degasser or flow cell. Since replacing either of these parts is expensive I want to be sure whether problem is degasser or flow cell. The degasser was not cleaned with nitric acid and this is not recommended by manufacturer. Does anyone know any strategies for cleaning Thermo Accela 600 pump degasser without strong acids? I was thinking to rinse the degasser lines with dichloromethane or some other non-polar solvent.

[mattmullaney]

Hi Again, BurntChem,

Okay then, normally one considers setting data rates (sample rates) using the narrowest peak of interest. You're using 20 points per sec (Hz), so for your largest peak you're acquiring 18 sec x 20 points/sec or 360 data "points" across the peak...but really this is multiplied by 200 as well (200 wavelengths are being acquired simultaneously), so I'd guess that you're over-acquiring data. If you have peaks of interest that are, say 1 sec wide at the base, then you'd have little choice other than doing what you already are doing here. You may want to lessen the data rate here...depends on the other peaks you're interested in.

The time constant (filter rise time) refers to the time the detector takes to transduce ca. 66% of the electronic signal gathered from the detector...generally the higher the value, the "smoother" the peaks will appear at the expense of "peak broadening" due to the slower response of the electronics. If your largest peak of interest is only 18 sec wide at the base and you have narrower peaks you're interested in...lower this value (maybe 0.2). This may not be the source of the entire issue(s) you're observing, but it's surely not helping you out, either. A time constant of 1 sec is typical for peaks having widths at their base of 60 sec or so. You may have more "noise," but your peaks will also narrow a fair bit as well.

Wavelength Step...well, in your case, you're scanning from 200 - 400 nm and gathering data at every single whole-numbered wavelength over that range...200 data channels out.

The filter bandwidth refers to the "slit width" of the detector...typically if you're concerned with keeping high UV spectral fidelity (narrow, high peaks out) you want this value to be SMALL...on the order of 1 nm or so. If you are interested in the ability to see very small peaks, you want this value to be LARGE, on the order of say 8-10 nm. In the old days...the filter bandwidth was a physical slit width...and the width of the slit after the light source was physically changed when this setting was changed...now it may be an adjustment of the voltage/currents inside the detector, instead (attenuation).

Here's a couple of websites for you to look at.

http://www.chromacademy.com/chromatogra ... tings.html

http://www.chromatographyonline.com/lcg ... ail/838057

https://www.chem.agilent.com/Library/sl ... zation.pdf

Wow...if the flow cell windows are visibly clean, I'd try flushing the degasser with lots of Isopropanol instead of methylene chloride...make sure the degasser can handle whatever solvent you try out. Hot deionized water is also an option...say 60 - 70 degrees Celsius.

Please see what you think, and thank you.

[BurntChem]

Thanks a lot for the advice about relationships between bandwidth and rise time etc. I'll post more on this issue soon.

Edit: Interesting twist in this story. I tried using a different HPLC system that uses ammonium formate (50 mM) instead of formic acid to control pH for separation of my analyates. So solvent A is 50 mM ammonium formate in water and solvent B is pure methanol. The gradient increases concentration of methanol throughout the run.

Interestingly it appears that the UV background problem has gone away. I find this very weird. Either our formic acid is dirty or the formic acid is helping some contaminant to leech out of HPLC system. Even if the formic acid is dirty it doesn't make sense that the background absorbance increases as methanol concentration increases because in that system formic acid is 0.1% in both mobile phase solutions.

Although I am tempted to just forget about the problem I am still skeptical as to what is going on.

[mattmullaney]

Hi Again, BurntChem,

Well, I'm not as sure as you are...about the behavior of additive (e.g. formate) in different solvents...back from that reference I posted earlier in the this thread...and yes, I know this ISN'T your case here...just an example to consider, TFA is trifluoroacetic acid:

Abs. at Wavelength, nm
Solvent 205 210 215 220 230 240 250
Water with 0.1% TFA 0.78 0.54 0.34 0.2 0.06 0.02 0
Methanol with 0.1% TFA 0.33 0.37 0.38 0.37 0.25 0.12 0.04

there CAN be a dependence of the observed absorbance of a solvent mixture when the solvent is changed. Seems that these effects for the "commonly-used" HPLC eluents are most pronounced at lower wavelengths. In the case of the new method you're using, the formate concentration lessens as methanol increases...and also as formate conc. decreases...agreed here. Sadly, no data in this source for formate in multiple solvents, but if you're so inclined and have the time, they could be generated for MeOH and water if you've a spectrophotometer. Heck, maybe I'll see if I can do this on my own for fun next week and post it, if we've formate at work.

Anyhow, I'm sort of concerned that we didn't apparently find the source of what you observed...with the formic acid. That is weird and sort of disturbing, since your troubleshooting seems quite solid to me.

[Sorry that the columns in the table aren't lining up properly...I tried.

[Hollow]

how do you take the formic acid or methanol out of the bottle?

did you (or others) use plastic pipettes or other plastic consumables? Did you use membrane filters e.g. PET?
could be some accumulated leachables/plastizisers/slip agents from plastic materials

[HPLCaddict]

It could also be possible that with 50mM formate you're creating such a high background signal that you simply don't see a "trace" impurity anymore

[BurntChem]

The plot thickens...

So I tried to isolate the problem to the formic acid. Basically I switched back to the original system with solvent A containing 50% H2O / 50% MeOH plus 0.1% formic acid and solvent B containing just MeOH (whereas it would normally contain MeOH + 0.1% formic acid).

Now as the gradient increases in MeOH concentration the baseline goes down and the UV impurity is not visible. So this pushes me back towards the possibility that this is a result of both batches of formic acid being contaminated. I know users have used plastic pipette tips before so its possible it got contaminated from that. Although the formic acid (LC/MS grade) is stored in plastic I understand that different kinds of plastic have different impurities.

I'll try buying new formic acid and see if the problem goes away. Its cheaper then replacing system hardware which might not be cause of problem. I'll post the results. Thanks for all the feedback and help.

HPLCaddict: Yes I agree the formate could just be blocking out the impurity. Which is why I try to further isolate problem.

[dr-statz]

you can use an "inline" cleaning column like the Ghost-Guard-LC or the Ghost Trap DS. This cleaning columns removes traces of impurities in the mobile phase and inhibit the appearance of ghost peaks in gradient HPLC.
This inline-cleaners are easy to use! Try it and result will satisfy you!

Ask for test column and try it!

We use then for many years with good success in our quality control lab.

[BurntChem]

Further update. New batch of formic acid did not solve problem

Now on to more expensive hardware maintenance / parts replacement...

[Arne]

Hi,

maybe try the following:

1) you can "substract spectra" in Xcalibur from neighboring regions that are very similar in mobile phase composition but outside of the elution window of the peaks of interest. This functionality is a background substraction and can really clean spectra up!

2) Could you not add FA to mobile phase B (MeOH). That might attenuate the problem. WE have stopped adding FA to our mobile phase B and have seen no ill effects.


3) Run the gradient without the formic acid additive and track the low wavelength region of the UV spectra.

What's puzzling is that this only happens in a Thermo PDA and not an Agilent PDA...that's quite strange. I would think that the methanol environment shifts FA absorption.

There might be surfaces or metal ions in the PDA that catalyze some sort of arrangement or maybe even formation of methyl formate.

[BurntChem]

Did routine pump maintenance to see if the issue was pump related. Although the pistons were dirty and the pump seals did need to be replaced this did not solve the problem either. I did speak with a Thermo technician recently. Since I've ruled out most issues they said the problem could be the beam combiner or the flow cell in the PDA. The beam combiner can get dirty and an optics type of problem might make sense given I've ruled out so many other issues. Replacing it might be a cheaper option then the flow cell but then again it might not work. I still find it crazy that the flow cell could be dirty after nitric acid cleaning. Maybe I need to clean with a very non-polar solvent.

Arne- Yes I've tried removing formic acid both from mobile phase B and both mobile phases. It does make the UV background problem go away which is why I thought it was an issue with formic acid contamination. But apparently its not the formic acid (the cut off is way off) or any impurity in the bottle so its a weird problem. Plus I really do need the pH control. I should also add that the UV background problem was never seen on agilent PDA in a different lab using the same methods.

However, your suggestion about methyl formate forming might be a cause I'm just having trouble finding info on the UV properties of methyl formate.