Continously Decreasing Retention Times from Run to Run

[Hampow]

Hello all,

I am relatively new to RP HPLC, and very new to this forum. In fact, found the forum while searching for answers to my problem. I did find a post close, but not quite the same and from a couple of years ago, so decided might be best to start a fresh post.

I am transferring a tried and true method from a colleague's system to a new system (less than year old). Same C18 column used for both: Zorbax Extend 18, 100 x 2.1mm. 35 minute gradient program is A: methanol; B: water... time table as follows: 0 min 92/8, 15 min 100/0, 20 min 100/0, 21 min 92/8, 35 min 92/8.

On the old system, this was good. On the new system, running the same standard over and over, sequential runs drop in retention time each time from one run to the next. By the end of about 10 runs, the times of a 16 minute peak have dropped down to 10 minutes. Purging the lines with fresh solvents resets all to good, but then repeated runs do the same thing. Switching to different channels gives same result; freshly prepared solvents, same results.

Thanks for any thoughts on this.

Regards,
Mike

[Blazer]

What is your flow rate? What sort of HPLC system did your colleague use? What is your HPLC system?

[Hampow]

0.2 ml/min; colleague system Agilent 1100; new system Thermo Fisher Dionex UltiMate 3000.

[Hampow]

For the first round of runs, the Thermo run retention times are nearly spot on to what the Agilent times are; and area response on the detector remains constant as retention times decrease.

[Hollow]

What about the pressure?
Are there any changes too?

if it would be my case, I would check if the retention times drop correlates with number of inj or volume of solvent re-(equilibration step)
See troubleshooting wizard for guidance: http://www.lcresources.com/resources/TSWiz/ (general); http://www.lcresources.com/resources/TSWiz/hs30.htm (topic)

Maybe you can add some uv active substance to the methanol (like aceton 0.1% or thiourea) to the methanol, so you can monitor the gradient between several runs (blank injections of course). Are there comparable? Maybe your colleague can do that too.
In similar way, determine the gradient dwell volumes of your systems (methanol and methanol with aceton 0.1%).

And maybe it's a controller thing: try to run your gradient not to 100/0 but to 99.9/0.1

[bunnahabhain]

Try to prepare eluent B as MeOH/H2O 92/8 and run a gradient from 100% A/0% B to 0% A/100% B. Maybe the pump (is it binary with high pressure mixing ?) does not work that accurate at 0.08*0.2 = 0.016 mL/min.

[Hampow]

What about the pressure?
Are there any changes too?

Yes, the pressure does change. The initial pressure at the start of the gradient program is 60 bar (same is on the Agilent system), drops to the mid 40s during the run, and returns to 6 following the equilibration as expected. However, on the Thermo system presently, sequential runs do not return to 60 bar, in a step-wise fashion, they decrease with each run. Purging lines brings it all back to normal, only to do the same thing with the net sequence of runs.

I switched channels for the solvents; same results.

Also, following a sequence in which the initial pressure continuous decreases, the pressure stays down even if left just to pump initial solvent proportions for an extended period of time.

[Hampow]

Try to prepare eluent B as MeOH/H2O 92/8 and run a gradient from 100% A/0% B to 0% A/100% B. Maybe the pump (is it binary with high pressure mixing ?)

It is a quat pump with low pressure mixing.

I did try switch the B mix from 100% water to 75:25 MeOH:water; and of course alter the gradient from 92/8 -> 100/0 to 68/32 -> 100/0. Pressure drop less, and stable. Was able to run all samples with consistent results. Upon switching back to 100% water, retention times decreased by a couple of minutes, but held for several test runs.

[Hollow]

Hm, very interesting...
Seems your system has problem to pump the water part again.
Did you change the channels only from a-b or also checked on c/d?

Next would be to check the performance of the proportioning valve ->troubleshooting wizard for guidance
then call thermo. As it's in the first year you possibly still has warranty on it.

and have a look at this article found by google:
http://globalresearchonline.net/journal ... 7-2/32.pdf

it mentions something about pressure/viscosity sensor on ultimate 3000 systems when having retention time/gpv issues... don't know if it applies to you

[Hampow]

Yes, I switched from C/D to A/B. I plan to try reversing which bottle is which next C/D to D/C.

I performed the GPV test today. The first run looked a little off, but then next 5 looked fine. I also ran a linear gradient test. That looked mostly good except for one small portion at about 30%. Will repeat tomorrow. Was running late today and was rushing a bit, may have been a good on my part.

Thanks for the link! Yes, I will call Themo on the warranty soon. I just like to have some input on what I have tried and seen before I call. And who knows, there is a fair chance it is something silly that I am doing incorrectly.

[carlo.annaratone]

in my experience even very small defects in the GPV test will lead to huge RT changes. Which tracing agent did you use for the GPV test? I found acetone too volatile, with changes from the start of the test to the end, and settled with reserpine

[Hollow]

for curiosity, it would be interesting to know, if the flow rate is decreasing or the mixture is changing.

Maybe you can add some UV absorbing compound like acetone or some less volatile one, as Carlo suggested, to see if the UV trace comes back to initial after the gradient. (use a restriction capillary for this)

My thinking is, if UV trace comes back but the pressure is lower, then you have to have a lower flow rate. If pressure and UV trace are different, then you should have a different Water/MeOH ratio so pressure change can be due to lower viscosity. In that case you should measure the real flow by volumetric flask method, to make sure you don't have both effects...

Anyway, guess it is something Thermo should fix.

[Hampow]

in my experience even very small defects in the GPV test will lead to huge RT changes. Which tracing agent did you use for the GPV test? I found acetone too volatile, with changes from the start of the test to the end, and settled with reserpine

I did use acetone; 0.1% in water.

[Hampow]

for curiosity, it would be interesting to know, if the flow rate is decreasing or the mixture is changing.

Flow rates from each channel with 1m of 0.005" id tubing in place of the column are good. However, that was an isocratic check for each channel. I will try checking the flow as it runs though some of the tests, and then again as you suggest with the method run though the capillary.

[Hollow]

My thinking is, if UV trace comes back but the pressure is lower, then you have to have a lower flow rate.

sorry have to correct me:
You're observing decreasing retention times, so lower flow rate is out of the question. Would give increasing rt...