What causes very spiky baseline in UPLC that gradually gets better?


I put on a new column, a 1.7 micron CSH phenyl hexyl, 2.1 x100 mm column with an inline filter on UPLC and my mobile phase A is potassium phosphate 20nM and B is 100% Acetonitrile. I run a gradient over 25 minutes cycling through 20% B 80% A back to initial conditions again. When I put on the column, I put 100% ACN through it for 10 mins, then 50:50 ACN:Water for 10 minutes and then put on about 6 condition columns.

When  I tried some trial injections of a standard solution, the baseline was very erratic at first but after a few injections it calmed down and gave normal baselines. However, when I put on a sample set last week, with only 2 condition columns in the sample set, the first injection (blank) was horrendous, spikes all over the baseline from start to end, the only way I could get it to look acceptable was massive minimum heights and thresholds but its obvious I'm just covering a problem. The 2nd sample, a low concentration sample, was just as bad, with no difference between the spikes and the component peaks. I work at 220nm so quite low.

As the run went on with higher concentration samples, the baseline settled down more and looked ok near the end. I used the same mobile phase on a second machine and that was fine so im pretty sure its not the mobile phase. I'm open minded to any suggestions as to what may have caused this- was the column not conditioned enough, poor fittings, detector issue etc. Any suggestions would be welcome. Thanks for reading!


  • 1) make sure the potassium phosphate is completely dissolved before using it.
    2) flush with 80% ACN to ensure that everything stays in solution
    3) It appears that it just takes this column a while to equilibrate
    4) If the noise has anything to do with the gradient forming process, you can reduce it a little by premixing your A and B phases so your gradient actually goes from 0 to 100% B
  • Thanks very much for that Dan. I suspect your third point is close to home and I'm going to condition this column longer and do a few more trial injections before I use it in a full sample set again. Might be worth dissolving the buffer salt a bit longer as you say too. Thanks.