shifting retention time

[c_gryniewicz]

We are running an OQ/PQ with a mobile phase of 50:50 water:methanol. The problem that was observed is that after multiple injections the retention time of the peak shifts to a lower retention time by 0.5 minutes. We checked the flow stability and pressure of the pump and it is ok. We also checked the column oven temperature and that is stable also. After the shift in retention time, multiple injections will be reproducible with a consistent retention time, but then it may shift again, with out warning....Any suggestions or ideas would be appreciated???

Thanks

[tom jupille]

First question: " does t0 change by the same percentage as tR?"

If t0 is shifting by the same % as tR, then you have a flow problem.

If t0 remains constant, then you have a column/mobile phase problem. In that case, the second question is: "does your plate number (N) remain constant, or does it decrease when tR changes?"

If the plate number decreases, then you have a column problem (channeling? void space at the column head?). If it remains constant, then you have a chemistry problem. Are you using an in-line degasser (which may be cycling)? Are you controlling temperature? Is this happening with only standards, or do you have possible buildup of sample matrix contamination on the column?

[Peps]

Have you premixed your mobile phases or do you use the system to mix the two phases? If you use the system for mixing you can try with a pre-mixed mobile phase. We have sometimes experienced problem with our Waters Alliance systems when we do not premix the mobile phases...

/Peps

[c_gryniewicz]

t0 does not change by the same percentage as tR and the plate number (N) remains constant. I am not sure that a chemistry problem with the method exsists. 6 months ago the same method was performed on the instrument and it worked well, it has been tested on other systems with no problems observed. We are controlling the temperature and I have monitored it for an extended period of time, there was no change in the temperature. The mobile phase has always been premixed and we have seen this problem with not only standards but in the performance of another method as well. We thought then that it was a column problem, we added a column wash, with column equilibration time, to the method and that seemed to correct the problem then. However, we are using a different column for the OQ/PQ and after an extended period of time during the analysis the retention time will shift. We are using an in-line vacuum degasser, but it does not seem to be cycling??? Any other suggestions???

[Steve]

Another thought

I did not see any buffering of the mobile phase water mixture. Ultra pure water will start immeditately to absorb CO2 from the air forming carbonic acid lowering the pH from the starting point. Depending on how long the mobile phase is sitting there may have an effect on reproducitibility. Just another thing to check out.

[tom jupille]

I am not sure that a chemistry problem with the method exsists.

Problem-solving rules:
1. If if wiggles, it's biology
2. If it stinks, it's chemistry
3. If it doesn't work, it's physics.

Once you eliminate biology, all HPLC problems are either chemistry or physics

You've done a thorough job of eliminating possibilities. Absorption of CO2 is a remote possibility, but unlikely given that the same method has run successfully before. I'm still inclined to suspect a flow rate problem. What is the the tR of the shifting peak? If it's long, then a 0.5 min shift may be a small enough % that you wouldn't notice t0 shifting (?)

[c_gryniewicz]

Thank you for all your input.

The tr of the peak is 10.5 min which then shifts to 9.91 minutes.

We have discovered that an intermittent pump problem exists, which may be related to a faulty proportioning valve. Our previous runs had been isocratic, with only a noticeable shift in retention time. A step gradient performance test led to an unusual step pattern for the first 25 minutes and then a normal step pattern for the remaining 35 minutes. Do you have any suggestions on how to test the proportioning valve???

[Alex Buske]

proportioning valve testing:
Version 1: practical approach, not compliant with vendor: line A: Water, Line B MeOH: Replace Column by some capillary just to provide backpressure, run a step gradient (i.e. 0, 1, 3, 5, 10, 25, 49, 51, 75, 90, 95, 97, 99 100 %B), detect at 200nm. I have seen that working on several systems, but i know no vendor that does it that way.
Version 2: as above, use 0.1% acetone in water for line B, detection 260 nm and skip the 1, 3, 97 and 99% levels. Usually 1% deviation is specified.
Additionally you can run a linear gradient and check that it is linear.

[Jason Hendricks]

What manufacturer/model is the LC?

[unmgvar]

follow the procedure told by John W. Dolan in his article
"Performance Qualification of LC Systems"
LCGC NORTH AMERICA VOLUME 20 NUMBER 9 SEPTEMBER 2002.

or

"Gradient Performance Checks"
LCGC NORTH AMERICA VOLUME 22 NUMBER 10 OCTOBER 2004

if your pump is very old go for the +- 2% limit, but most vendors now go for +-1 %.
don't skip the 48-52 steps, the reason vendors don't do the test around those % is because that is where it is harder to mix most of the time, and where you will have the most deviation between the result and set values so those % will show you more about the mixing then about the proportion valve which in your case is important as well.
also take a closer at the baseline at those % if it makes a lot of S shapes you might have a serious mixing problem.
the 10:90 ratio are there to tell you which proportion valves don't work properly.

BTW is your system able to sync. between pump and autosampler? if yes then use the option you will dramatically reduce you RT shifts.

if after that you don't find the the problem, then redo the first PQ test you told us about with a column that won't create more then 40-50 bars, do it until you have an injection where the RT shifts, at the same time monitor your pressure signal on all of the injections.
if your pressure drops you could easily have a check valve problem, or your pump might be "missing" a step. you most probably get a tiny air buble in the check valve and the pump or an electrical problem with the pump, start looking at the "cards".
the tiny air bubles could come from the filters on the eluent lines, so clean them or change them.

[Uwe Neue]

Hi unmgvar,
I do not agree with you that mixing at the 50/50 level is a big problem. Please explain why you think that this might be the case!

[c_gryniewicz]

Thank you for all the advice. I am going to try the suggested test, if we can not figure out the problem, we may just have to send it in for repair...as the last resort...

The HPLC system is a Thermo Separations with pump model P4000.

[westlake]

Is the pressure shifting? I had a problem like this before. I used 85% ACN run a set. In the first 20 samples, the RT was no change. But later the RT start shifting. In may case, the seal of pump B was bad. After we changed the seal, the pressure was stable and RT was back. So may something wrong with you pump A. Can you switch Pump A and Pump B, see how RT changes?

[unmgvar]

Uwe,
i did not say that it was a big probem, i said that it was the hardest composition to mix, and in relation to all other possible composition for 2 liquids it's correct.
most vendor's SOP's do not cover this area in a gradient test because it is the "noisest and biggest drift" (no point in showing your equipment at it's "weakest"), and will the least contribute to spotting a problem with the proportion valves anyway.
at that composition you will many times see mixing problems that will not be associated with the proportion valves but with other components of the pump.