Noisy baseline

<p>The attached picture is the baseline from an injection of blank ACN into 2.1x100 mm BEH column. The original mobile phase A in 0.05% TFA, but I found the baseline is very noisy. To make sure it is not the TFA problem, I used pure water instead.</p><p></p><p>Mobile phase A: pure water, mobile phase B: ACN; UV at 272nm @ 20Hz. </p><p></p><p>The gradient is initial 98%A, 1.8min 98%A, 6min 35% A, 11min 10%A, 14.5min 10% A, 15min 98%A, 18min 98%A.</p><p></p><p>It is very hard to believe that water and ACN will produce such a noisy baseline, especially for the "peak" at 9.7 minutes, which has a 0.02 UV absorbance. If scaled up the same gradient and run on a regular HPLC with a regular C18 column, it is fine. </p><p></p><p>The column is clean and in good condition. I also try the same gradient on another BEH column, the same noisy baseline obtained.</p><p></p><p>Anyone see this problem before?</p>


  • Couple of questions -

    What kind of detector are you using, UV or PDA?

    When was the last time the detector cell cleaned?

  • lizh

    Noisy with multiple solvents is a clue. What is your reference energy, if there is too little light energy you will see noise. However, I have also forwarded your data to other chemists at Waters, but I believe the reference energy will provide a clue. Also just for completeness what is the flow rate. Did you actually run the HPLC separation on the ACQUITY UPLC?

    I will add further comments.


  • The detector is an extended PDA. The UPLC was installed less than two months ago, and I am the first user. So I believe the flow cell is ok. Actually I try the gradient with two different columns on two different UPLC systems, the baseline is the same.

    The attached pictures are the energy reading. The #1 is direct energy reading and the #2 is the reading with dark current subtracted (in both case, the flow cell is filled with ACN).

    The UPLC method I showed before was converted from a HPLC method. I didn't see the noisy baseline in the original HPLC method, that's why I said that the baseline is different.

  • I'm guessing this is a dirty flow cell issue that I've experienced a few times now. Basically the internal coating (Teflon AG) of the flow cell gets contaminated somehow and needs cleaning sometimes.

    As a precaution from this occuring in the future ALWAYS TURN YOUR LAMP OFF after your analysis has finished. Leaving the lamp on with no flow going through the cell exacerbates the problem apparantly.

    I have heard Waters is working on a retro fit to ensure a shutter comes down after an analysis is complete to stop light getting to the flow cell - any news on this from Waters reps on here?

    I have attached an official Waters flow cell cleaning regime. Try this and see if it improves - and feedback to the community would be helpful please. Fingers crossed it works for you, it has for me

    Edited by: Rob Burgess on Jan 16, 2009

  • lizh

    I have re-read the email, is it in fact the additional peaks on the baseline that is concerning you? It may simply be bad water?

  • I have tried different water, our in-house milli-Q water and bottle HPLC water as well as different ACN. The reason I am concering about the baseline is because one of the impurities eluted during that time. It seems that big "additional peak" only shows up in this method, and I don't see such a big peak or noisy baseline in other methods.

  • Thank you very much! Got to try it later!

  • Hello,

    Your energy is good (near 100,000 counts on both plots), so I think your flowcell is okay, but I also see that your noise is about 150 to 200 microAU, which is well above expectation at 10 or 20 pps with reasonable filtering. Could we please see an image of the method editor, or a report of the method? What wavelength are we looking at on the plot?



  • AJA

    Just a couple of things. What you are looking at here is not a noisy baseline, noise problems are something different. I tell you this only because sometimes the terminolgy used can hinder the troubleshooting process (especially on the phone or in a forum like this). It looks like to me that you are suffering from contamination or carryover. There are two expeiment you need to do. Run a series of blanks. With each blank increase the amount of re-equilibration time. If the size of the peaks in your blanks increases with increasing re-equil time you likely have contamination from your mobile phase. This kind of contamination often comes from the water, buffer components or dirty glassware. If the peaks from the blank injections stay the same size you may have contamination of the blank itself or the vials themselves. If the the peaks are gradually getting smaller, you may have a carry over issue. If this may be the case you should then run several gradients without an injection (if using empower choose inject immediate samples to do this). If the carryover is from the loop or needle, peaks should get smaller. Hopefully, you can troubleshoot the problem, and work on a solution. As well, I dont really think this is a flow cell issue. Also current detector firmware does close the shutter after (guessing here) 10 minutes of inactivity. AA

  • Attached is a before and after chromatogram resulting from acid flow cell cleaning.

    This seems to be quite a common occurence and I starting to recognise the symptons. I'm wondering if there is alternative internal flow cell coating to Teflong AG if it causes contamination problems with such regularity.

  • Hello,

    We have explored coating protection possibilities, but the principle of total internal reflection, which is leveraged by the refractive index property of the Teflon AF, is reliant upon the fact that the mobile phase is in intimate contact with the Teflon AF. A coating would obviously present an interstitial barrier.

    In this case, it appears that TCheng's flowcell is okay, based on favorable intensity plots, so I think the concern is the peaks on the blanks, probably due to dirty water/glassware/etc. It appeared to me that the high frequency baselie noise was also high, although the plot does not ideally present it, but this could be an 80 pps run with little or no filtering, or, I'm not reading the plot properly, so I'm not ready to be concerned about the detector yet!


  • Hello TCheng,

    Here is an excellent article for anyone battling with "ghost peaks". It talks about the potential sources and ways to pinpoint the source. Williams, Stuart. “Ghost peaks in reversed-phase gradient HPLC: a review and update”. Journal of Chromatorgraphy A, 2004, 1052, 1-11.

    If only this method gives you this "additional peak", then I would suggest that you start with the unique aspects of this method and narrow it down from there.

    Can you let us know the brand/grade of TFA, ACN, and water used?

    Do you have any mass spec data for this "additional peak"? This might help us to get an idea of the chemical identity of the peak, i.e., is it phthalate?



  • Thank you very much for your information. The ACN (HPLC grade) in my lab is from EMD, as well as the bottle water (HPLC grade). I also try our in-house MilliQ water as well.

    I did try several injections with increasing equilibrate time. This "additional peak" does increase - clearly indicates itcomes from mobile phase(s) / bottles. One interesting thing this "additional peak" only shows up in this particular gradient, and we don't see it in other gradients as well as on regular HPLC.

    Thanks for everyone's help!

  • Hello TCheng,

    It is interesting to note that Section 3.3 in Stuart Williams' paper talks about "Living with ghost peaks". But if you would like to get rid of them, please try the suggestions documented in the attached white paper from Waters.

    Document title: Controlling Contamination in Ultra Performance LC/MS and HPLC/MS Systems

    Document location:

    Specifically try the brands/grades of solvents recommended there. They are readily available in North America, to a less degree in Europe and Asia.

    The sensitivity improvement of UPLC vs. HPLC is a fact. It applies to both the analytes (peaks we would like to see or wanted peaks) and the contaminants (peaks we would not like to see or unwanted peaks), unfortunately.

    I did take a closer look at the gradient you are using. It seems to have a relatively long time sitting on 98% A, which will allow more contaminants in the mobile phases to be focussed on the head of the column.





    I agree with the suspicion of the method using too high a data rate (80Hz). For this method, I recommend using a data rate closer to 20Hz. On a similar note from some personal experience, I also have seen noise in the baseline when my BSM seal wash runs dry and I continue to make injections for a period of time before noticing it was dry. Take a look at your delta psi for a few minutes at an isocratic condition. If the delta is >70psi, then you might need a PM of your BSM or a little more flushing of your pump seals, if they were in fact dry.

    The mystery peaks are a whole other issue....water? reagent quality? extremely hydrophobic peaks finally eluting from a previous injection? Separate tests must be performed for each suspected cause.