Baseline problems Waters UPLC

[Mattias]

Hi,

We have severe problems with the baseline on three different Waters Acquity, when running with UV-absorbing buffers (such as ammonium acetate and formate). We have the instruments coupled to MS instruments, and we typically only use 5 mM of these buffers.

The noise has a high frequency (up-down-up-down etc...), and is at least 10 times higher than on a Alliance or Agilent systems. Since all systems behave the same way, I suspect that it has something to do with the design of the instrument. We have had technicians "living" here now for a while, but with no improvement. Officially Waters claim that there is no problem with organic buffers and Acquity.

I wonder if any of you have the same experience, and if you have found ways to solve it. As it is now, we cannot use the UV trace for anything but to see the main peak.

[Victor]

Matthias- please could you clarify your message. You talk about UV absobing buffers, but then you go on to say that you are using these with MS detection. Is the problem you are seeing showing up with MS detection, UV detection or both? If you are just using MS detection, I am not sure why you have thrown in the phrase "UV absorbing buffers" because MS and UV detection are totally unconnected.

[Victor]

Mattias,

Re-reading your message, it seems that you are having these problems with UV detection? If so, please could you tell us:

a) How do you make up the buffer-the exact description eg for ammonium formate-what pH you are using and how you adjust the pH. There are many ways of doing this.

b) What wavelength you are using.

[Mattias]

We are using 220 nm "by default", since we are analysing peptides with no other chromophores.

The buffers are typically prepared by dissolving ammonium acetate/formate in water and then pH adjusting with concentrated acetic/formic acid. However, we see the same problem when running with only acid (e.g. 0.1% formic acid in water).

The MS baseline (TIC) is very nice, it is only in UV that there is a problem.

[Bruce Hamilton]

What is the the UV absorbance of your mobile phase at 220 nm,compared to milli-Q water, using the same spectral bandwidth as your detector settings?. What is the noise of your system when you use the sample mobile phase without the buffer?.

My guess is that your mobile phase is absorbing more than you think at the settings you use, and the detector is seeing very little energy.

Assuming a high sample rate of a UPLC detector, noise may be magnified. I assume the instrument passes the manufacturer's tests, so I'd be looking closely at the mobile phase, and possibly the spectral bandwidth settings.

Please keep having fun,

Bruce Hamilton

[danko]

Hi Mattias,

Are you running gradients or mixing 2 different eluents inline? If so, here is a tip, I got from a guy I know:
Install a larger mixing chamber or an extra one identical to the one you already have installed.
I am aware of the fact that this will increase the system volume a bit, but it might be a reasonable price to pay for a flat baseline.
Unfortunately I haven’t tried it yet (not that I haven’t observed the problem) due to lack of time. But in my department we only have one Acquity system and it’s only used once in a while for tests. We’ve seen a lot of noise with both buffer/salt containing mobile phases and with Water/TFA/Acetonitrile
I have reported the observation to Waters but they don’t seem to have an adequate answer.

If the tip is applicable to your observation/problem, please get back with your thoughts.

Best Regards

[AA]

Danko,

You say you have reported the observation to Waters but they don’t seem to have an adequate answer. I assume you mean baseline noise at low wavelengths using UV absorbing mobile phases. I have read dozens of your posts and i know that you know the answer as well as anyone. It is probably inadequate mixing. Here is the issue in a nutshell. Customers (good chromatographers that is) want very low system volumes so they can run rapid (ie Steep) gradients using short norrow bore columns so they can minimize re-equilibration time and still get all the benifits of gradient chromatograpghy. Further to that they want to take advantage of smaller particle columns, which really preform best with low system volumes. Then, surprise, surprise, they seem puzzled by baseline noise at 214 nm using ACN/TFA gradients. Low system volumes mean low mixing volumes (its the physics of mixing folks). For ACQUITY systems, there is a 100 uL mixer available, it helps. For peptide (or any large molecule work) ther is a 400 uL mixer available. It has a 10000 psi pressure limit, but if you have small particle columns and large molecules, the optimal flow rate is ~ 0.3 mL/min anyway, so the lower pressure limit is not an issue. On the other hand, there are a number of things that can look like mixing issues, partially working check valves, partially blocked pre-check valve filters, small particulate matter partially blocking the tee/filter/mixer assembly and I have even seen almost dead lamps produce this kind of noise as well as leaking detector flow cells. And, as was pointed out by Bruce poor selection of detector parameters (sample rate and filter constant) can contribute as well.

I think thats it.

AA

[danko]

Hi AA,

And thank you for response. As I wrote I believe in the "mixer solution" and that’s the reason I passed the tip over to Mattias and the rest of the board. The solution is only applicable if one mixes at least 2 eluents (gradient or whatever) and that’s the reason I asked Mattias if he did.

My lamp is almost completely new (50 – 100 hours) and it would be something of a co accident if everybody that has experienced this noise had “almost deadâ€

[Mattias]

To put in a larger mixer was the first thing that we did (we had to sacrifice the dvell volume a bit). The noise did not change. Even with premixed mobile phase the noise is the same, which indicates that it is not a mixing problem. I also have to tell you that the noise is very low with only water/acetonitrile.

Bruce: You may be right that the amount of light reaching the detector is the problem. But will it help to increase the bandwidth here? Won't it just be more diodes that will record too low energy? And why does it work with the "old" HPLC's? We don't want to increase the time constant, since the peaks are so narrow.

This is a very severe problem for us (and in the end also Waters), and we have actually stopped buying Acquity systems until this is sorted out. The new systems are now the "not quite UPLC" Agilent 1200. Organic buffers pH 3-5 is present in about 80% of our separations...

[Alex Buske]

Hi.

This might not be relevant to the UPLC but I had recently some noise problems with an alliance.
After maintenance, we washed the system with IPA, water, nitric acid, water, run a series of tests with methanol and checked the gradient performance using water ad water with acetone. everything was fine at 250 nm.
At 220nm we then had extensive noise around 1mAU (running a water / methanol/acetonitrile gradient). that noise had exactly the same frequency as the signal from the system pressure. The noise was to be observed with and without column, running mixtures or pure solvents. the 2996 could be used as a pressure monitor!
Finally I used a second detector and found out that the noise was real UV-noise and would get stronger when the solvent passed the column switchting valve twice instead of once. Dissembling, IPA washing, sonicating and reassembling the column switching valve parts solved the problem.
I assume that during maintenance some movement was loosening the valve assembly an a thin liquid film between rotor a stator came up. Upon pressure increase on the precolumn side some solvent leaked in the post columnm side.

Alex

[Alex Buske]

Hi.

This might not be relevant to the UPLC but I had recently some noise problems with an alliance.
After maintenance, we washed the system with IPA, water, nitric acid, water, run a series of tests with methanol and checked the gradient performance using water ad water with acetone. everything was fine at 250 nm.
At 220nm we then had extensive noise around 1mAU (running a water / methanol/acetonitrile gradient). that noise had exactly the same frequency as the signal from the system pressure. The noise was to be observed with and without column, running mixtures or pure solvents. the 2996 could be used as a pressure monitor!
Finally I used a second detector and found out that the noise was real UV-noise and would get stronger when the solvent passed the column switchting valve twice instead of once. Dissembling, IPA washing, sonicating and reassembling the column switching valve parts solved the problem.
I assume that during maintenance some movement was loosening the valve assembly an a thin liquid film between rotor a stator came up. Upon pressure increase on the precolumn side some solvent leaked in the post columnm side.

Alex

[Mark Tracy]

Does the noise follow the pump cycle? If not you have a detector problem. If so you have a fluidic problem, but it could be almost anywhere. Have you overlaid a trace of the pump pressure?

[Bruce Hamilton]

Bruce: You may be right that the amount of light reaching the detector is the problem. But will it help to increase the bandwidth here? Won't it just be more diodes that will record too low energy? And why does it work with the "old" HPLC's? We don't want to increase the time constant, since the peaks are so narrow.

This is a very severe problem for us (and in the end also Waters), and we have actually stopped buying Acquity systems until this is sorted out. The new systems are now the "not quite UPLC" Agilent 1200. Organic buffers pH 3-5 is present in about 80% of our separations...

Firstly, the problem is insoluble, so please write off all the UPLCs from your books, package up them up, and donate them to me, I promise to give them an excellent, caring, home....

Seriously, I assume they work fine for MS, so we just have address the UV concerns.

Have you measured the UV absorbance profiles of your mobile phases?. I was guessing that at 220nm, you might be on the side of the buffer absorbance, and widening the bandwidth ( or moving to 230 ), might demonstrate what happens if more energy reaches the detector.

I suspect you need to have a play with a few parameters - starting with a mobile phase and detector wavelength that has low noise, then evaluating the effects of wavelength, flowrate, bandwidth, mobile phase absorbance and buffer composition/concentration, etc. on noise.

Ensure your components are fully miscible, especially the buffers, and low wavelength noise can sometimes be attributed to post column precipitation as pressure drops/volume expands. Trace precipitation has a characteristic UV spectrum - a long downhill run ( rather than a peak ), high at 220 going all the way to 300+nm. Obviously, it's not the main volatile buffer component, but any slightly-soluble impurities present in the buffer or solvents.

I'm suspecting that, if the problem is not your mobile phase absorbing/precipitating, then it's the detector cell's sensitivity to flowrate ( diffraction ? ), and lower flowrates may lower the noise, but more data is needed....

Please keep having fun,

Bruce Hamilton

[Mattias]

This is embarrassing, but we have found the solution now.

The flowcell that we have installed in our Acquity systems is a "high sensitivity flow cell" with a pathlength of 25 mm. I did autozero with water in the flowcell and changed to the buffer. The detector went almost ded at 220 nm. No wonder that we have a noise problem.

I know that this is basic knowledge, but we were so convinced that it would work after the information that we got from Waters. I wonder how many service hours (on guarantee) that Waters has put into this...

Anyway, I have ordered the standard 10 mm flowcells now (which we have on the standard LC's).

[Victor]

Mattias,

I do not understand your enthusiasm for this solution. I am not an expert on detector noise, but I would not think there would be a huge difference between a 10mm path length cell and a 25 mm path length. Has someone verefied that the detector noise will go away with the shorter path length cell? Do you know what the cell volumes of these two detector cells are? Will the change to a smaller path length cell improve the peak sharpness by reducing extra column band broadening?