Gradient Pump Validation Problem

[hexa]

Hi all,

I'm doing the PM's/calibrations of about 20 HPLC's this summer. One of the tests is to run a gradient mixing to check the mixing of a binary pump system of shimadzu 10vp's.

The test is water, and .3% aqueous acetone. This is a corporate procedure so I can't change it, but I'm wondering if you guys see anything wrong with it.

What i'm finding is that EVERY single test so far (on about 5 different machines) is getting the same results: the acetone pump pumps about 1.5% too much at 50%, and .75% too much at 10% and 90%. The 50% fails the instruments and the only way I can make them pass is to change the alpha values on one of the pumps. Flow rates are perfect. Any ideas? Column is either a Nova-Pak C8 or C18

[JJG]

If your flow rates are perfect, I would not touch the alpha values of the pumps. One suggestion is to replace the column with 2-3ft. of PEEK tubing, such that your backpressure is at least 35 bar at your flow rate. What flow rate are you running this test at?

[Bruce Hamilton]

Are you sure that your phase compositions are correct?.
Is the second phase 0.3% acetone in water, or 0.3% water in acetone?

What happens when you transpose the mixtures, does the problem move with the acetone-containing phase?.

If you're feeling bored and are confident you followed the provided protocol, just fail all the instruments and watch the ensuing action - should be fun.

Bruce Hamilton

[hexa]

Hi all,

I just substituted for PEEK tubing and I get the exact same results. Moreover, I also tried switching pumps, AND GOT THE SAME DATA.

For instance: Run 1- Pump A = aqueous acetone, Pump B=water
A10/A100= .103 (Tolerance .990-.110)
A50/A100= .514 (.490-.510)
A90/A100= .906 (.890-.910)


Run 2 - Pump A= water, Pump B= aqeous acetone
B10/B100=.103
B50/B100=.513
B90/B100=.905

So frustrating. Doesn't seem to be a pump issue, but possibly a control/mixing issue. I don't even know what else to try at this point.

Some final points: The compositions are correct, and my readings are of an appropriate magnitude. Looking back the machines have always been somewhere in the high range. Seems like a faulty method to me, if I had any logic and developed an alternative then I might have a way to present my case. (I'm a summer intern btw)

edit: 1ml/min flow rate

[Noser222]

What is your 100% absorbance value?

[Bruce Hamilton]

OK - I think we are going to need far more information...

When I encounter such a problem, I generally assume that the previous user was able to make it work, so they either haven't detailed the procedure completely, or I have misunderstood something.

You need to go back to the previous year's data, and see how your results compare, looking at all aspects of instrument setup and solution preparation, and raw data.

We need confirmation that you are using UV detector. If so, detection wavelength, spectral band width, and whether any reference wavelength is turned on. If possible, you should try to be reading on the top of a peak, if not the choice of spectral bandwidth could affect linearity as concentration changes.

Ensure that the water is pure, and has no absorbance at the measured wavelength. Composition of the aqueous acetone, and the absorbance at 100%. If your mobile phase B absorbance is too high, you will see non-linear detector response.

If it is a mixing issue, choosing a lower wavelength with narrow spectral bandwidth may show a perturbed baseline, compared to that obtained with a premixed 50:50 A;B mix just going through one pump.

Please keep having fun,

Bruce Hamilton

[lmh]

I can't swear to this, but if you plot your actual ODs versus concentration of acetone, you'll find they don't lie in a straight line. They are on a slight downward-curving curve, and a best-fit straight line passes just above the origin.

The reason is probably that the acetone concentration is a little too high, and is revealing detector non-linearity. You have to remember that detector linearity is often not quite as good as a manufacturer would like you to think!

Many test procedures for hplcs seem to be based on older systems with less sensitive detectors, and many use far too much acetone. Modern detectors can see smaller ODs, but aren't at their best dealing with larger ODs. If (big if!) this is the problem, and you really can't reduce the acetone concentration because it's enshrined in corporate QA, then you are probably wasting your time working for this company. The only way their pumps will ever pass spec is by cheating or widening the tolerances!

[Fiz]

I agree to imh that the aceton concentration is far to high. Reaching the non-linear range of the detector at higher absorption value results in too high values for the 50% step.

I usually used aceton concentrations of 0.05% for such tests, so that the highest absorption is about 100-200 µAU.

In case you can't change the concentration I suggest to change the wavelength where the detector is less sensitive to aceton.

Fiz

[HW Mueller]

Just mentioned that elsewhere; One does not merely have to make sure that the detector doesn´t blackout, but also that the Beer-Lambert Law holds. Remember that the law holds only for relatively low concentrations, at higher concentrations there can be molecular interactions which cause deviation from linearity. Going to a lower absorbing wavelength only works if the analyte is not too concentrated for obeying the B.-L. law.

[hexa]

Thanks guys, this is most likely the issue.

Looking back at past data (even 15 years) it seems that the same trend always holds. Some fictional but illistrative examples:

A10/A100= .1013
A50/A100= .5061
A90/A100= .9037


A10/A100=.1007
A50/A100=.5032
A90/A100=.9018


As you can see, it really looks like there is a trend in the way these things look. I'll step the concentration down and see what kind of results I get.

I think I should also let you know what kind of run this is- there are no real 'peaks' but more just looking at the peak height. This is probably not even a good way to do this procedure it's just what has been in the books since at least '94. If anybody has a similar, but better way of doing this I'm open to suggestions.

An example chromatogram (with A10, A50, A90, and A100 being the respective heights of the steps):

[Uwe Neue]

Sorry for not having read this post in the past...

There are a lot of things that I do not like about your procedure. First, please specify the absorbance units that you get at the end. I also assume that you calculate the individual values based on this final value. For example, if you are outside the linear range, your final absorbance will be too low, which then inflates the other values.

Another element is that acetone adsorbs on the surface of a C18 in water. However, this should not be an issue if you get indeed a flat response with time (what goes in must come out).

I also would prefer to run the test in methanol, but this probably not an important issue.

[Alfred88]

To verify the Gradient of binary pumps, we use A= water; B = 0.1% Acetone in water. WL = 254nm.

0.3% is probably too high. (hexa's post)

Our current specs: gradient accuracy +/- 2% absolute, and gradient reproducibility <= 0.5% RSD. (I prefer to widen this to 1% RSD).

Alfred

[lmh]

Further to Uwe's point; he's right, don't use a column. At best it adds a lot of dead volume which means it takes longer to reach a good plateau for measuring the OD. Use a back-pressure regulator set to a sensible pressure.

The back-pressure regulator is most relevant when testing low-pressure mixing pumps, as the gradient proportioning valve typically has valves closed by the back-pressure. If there isn't enough back pressure, the inlet valves to the gradient proportioning mixy thing don't close promptly and you don't get the right gradient.

Some high-pressure mixing binary pumps will pass the test quite happily at low or high pressure, but it still makes sense to test them under realistic conditions. They have no idea whether the pressure is generated by a column or by a restriction of some sort, so there's no danger in using a back-pressure regulator.

I bought one from Alltech as was, Grace now I think, relatively cheaply. If you don't want to buy one, you might get away with a very long thin bit of tubing.

[hexa]

Just for anyone who is wondering-

This was resolved by using a back pressure regulator (it brought it to passing) and degassing for a LONG time. We don't have an inline, so I've been sonicating for ~45 mins and using asap afterwards and I get nearly perfect results

[Peter Apps]

45 min sonication could well have achieved the suggestions made about too much acetone

Peter