Gradient performance checks

[Gallen]

I am confused by results of water:0.1% acetone gradient performance checks (Gilroy and Dolan, 2004, LCGC North America, Vol 22 Number 10 pp 982-988), and have a question re the step test calculations, and would appreciate any help you can offer. Reservoirs A & B contained water, and C & D contained 0.1% acetone for all tests. It’s a low pressure mixing system.

The GPV test passed the 5% criteria, but was not ideal (1% or less) at 2.29%. The plateau for the A-D valve combination was a little taller than those for the other 3 combinations, which agreed very well between themselves, so I suspected valve D might be a little off.

There were step tests for A-C, B-D, A-D, and B-C combinations. The article said to compare the step test to a theoretical programmed value. How do you get a theoretical value without running an acetone calibration? I predicted theoretical values by measuring the height of the first (10%) step and then assuming the 2nd (20%) step would be twice as tall, etc. Is this what was meant? I also tried calculating an average corrected delta step size and using that for the theoretical value.

Most of the steps agreed more or less within the 1% criteria using the former approach, but error at the highest step (100% 0.1% acetone solution) was always a bit higher whenever valve D was used. I suspected valve D already, so that made some sense. However, for all of the combinations, once I got above about 50% of the 0.1% acetone solution, delta step size decreased as proportion of the acetone solution increased.

Linearity tests for A-C and B-D combinations were run, and also one for A+B and C+D (i.e. a 50:50:0:0 to 0:0:50:50 gradient). These results looked nearly perfect.

I don’t understand how the linear gradients could appear so perfect when the step increments were shrinking as the UV absorber concentration increased. Any insights?

[JJG]

For the step tests, the theoretical values are calculated by measuring the absorbance at 0%C/D and 100%C/D. Your steps are just X%*(absorbance at 100%C/D - absorbance at 0%C/D).

Having a trend in the step tests can be common but so long as the differences for each step are less than 1% from theoretical then there is no need to worry.

Try using the above equation and let us know if your results are any different. These results may correlate with your observation that the steps were shrinking at higher %C/D values.

[Gallen]

I can't thank you enough, JJG. The problem at the top end was apparently due to overly high absorbance, so I diluted the acetone solution in half and repeated some tests. I got some excellent results for certain valve combinations and some not so good for others. Now I'm sure how to interpret the results, they do make sense, and I have learned a lot about gradients produced by the particular instrument.

[JJG]

I usually run the test with a higher acetone concentration than you used (0.2% v/v) and I usually target an absorbance at 100%B of ~0.2-0.5 AU at 265 nm. I'm not sure what brand HPLC you are using but I have also noticed that for the Waters low pressure mixing HPLC (model 2695) that using methanol in MPA and methanol with methylparaben (~10 ug/mL) really improves step test and linearity test performance. I'm not certain why, but I have seen several steps with >1% difference from theoretical using water/acetone fall to <0.5% difference or better using methanol/methylparaben. I have only observed the need to use methanol with low pressure mixing systems.

I'm glad you're able to understand your results better now.

[Gallen]

Wow JJG, that is very useful information, because I happen to be testing a Waters 2695. Thanks again.

[Klaus I.]

I usually run the test with a higher acetone concentration than you used (0.2% v/v) and I usually target an absorbance at 100%B of ~0.2-0.5 AU at 265 nm. I'm not sure what brand HPLC you are using but I have also noticed that for the Waters low pressure mixing HPLC (model 2695) that using methanol in MPA and methanol with methylparaben (~10 ug/mL) really improves step test and linearity test performance. I'm not certain why, but I have seen several steps with >1% difference from theoretical using water/acetone fall to <0.5% difference or better using methanol/methylparaben. I have only observed the need to use methanol with low pressure mixing systems.

I'm glad you're able to understand your results better now.

Have you ever observed differences between a PDA and a 2WD Detector by doing these tests?

[Gallen]

Sorry, I can't comment on the PDA question because we don't have one.

I don't think I mentioned that we found the original test was off-scale due to a prep error. (Actually more of a communication error resulting in 10x the intended strength being tested.) The tests worked as designed after switching to the correct concentration using a methanol solution in our low pressure mixing system. Again, thanks so much for sharing your insight and experience regarding these test methods.

[JJG]

No, I have not observed differences between PDA and dual-wavelength detectors. I have had success with both.

[Peter Skelton]

If your degasser is in great condition you can sometimes trace error in gradient steps back to the degasser pulling acetone out of solution as well, which is why using a non-volatile UV absorber tends to give better results from these tests