Meth. Dev. Query / Problem - DryLab Accuracy

[Rob Burgess]

This is going to sound "long winded" to anyone reading this so bear with me if you please.

Background:

I'm trying to come up with a CLND compatible related impurity method for a basic pharmaceutical blend drug product (2° amine functionality amongst other moieties). It also has a large aromatic acidic counter ion.

I performed 4 modelling runs, as per usual with DryLab, (short grad - high and low temp. & conversley long grad high and low temp.). Gradient was 20 mins and 60 mins 5 - 100 % B (A = water + 0.1% formic acid; B = MeOH + 0.1% formic acid) - temp. were 30 and 70°C. This seemed to be OK although I had to switch wavelength to 254 nm as the slope @ our usuual wavelength of interest (220 nm) was just too severe to be able to detect any small impurity peaks! the column I used was a std. Luna C18 (2) (150 x 4.6 mm) 3µ. Flow rate set to 1.0 mL/min. The diluent for the drug product blend was 20% aq. IPA (there was a danger of using MeOH forming methylated imps apparently!). Samples were filtered before injection.

I eventually got to a point where I was able to model between 9 -11 peaks in total (including the main basic peak and its acidic counter-ion). The DryLab modelling seemed to be OK and the temp. gradient mode appeared to predict a variety of conditions that could have been promising.

The Problem:

I've used DryLab in the past and it has been very useful, but for some reason the accuracy of tR prediction seems to rather poor for the early eluting peaks (basic compound) +/- > 20% of actual run. However, for the later eluting peaks (acidic counter ion etc.) the comparative run was as predicted from DryLab (within +/- 1-2%) tR. Is there be any possible reason for these observations? Have I modelled the runs properly I wonder? Should I repeat the modelling runs again? Are strong solvent affects causing a problem I wonder (the peak shapes are poor for several gradients I've tried from predicted by DryLab although the modelling runs appeared OK generally).

Any ideas on this problem would be much appreciated... thanks in advance... Rob.

[DR]

wag(s):
1)How large in your injection volume? If it is significant, the IPA used for sample dilution could be throwing a wrench into the works.

2)Also - How well do you know what the void volume of the system? If you're off a bit on that, it would tend to cause more inaccuracies for earlier peaks than for later ones.

[Alex Buske]

With the Luna C18(2) 3µm, 150x4.6 mm I was at 1ml/min allways close to the system pressure limit (especially with methanol and at 25°). Did you see the same thing.
As DR already said, the void volume is quite important, especially for for early eluting peaks.
Can you reduce the gradient range? I allways feel, that less steep gradients are "more chromatographic) whereas steep gradients (in 20 min at 1ml/min for 5-100% on a 150x4.6 column IS steep) are more or less expensive SPE.
Is starting at 20% MeOH a option?
Alex

[Rob Burgess]

Thanks for the comments... they are much appreciated!

The injection volume I'm using is 10µL and the sample diluent is 20% IPA.
It appears there are solubility issues with our drug product / impurities. I went back to using MeOH (@ 20 & 40% v/v diluent strength) and I got significantly more from the 40% solution. However it appears there is a fine balance between 1) sensitvity of the method, 2) peak shape compatibility and 3) extraction efficiency of all components.

I've been using different gradient starting compositions: 5, 10, 20, and 40% MeOH (as modelled in DryLab). The pressure appears to be OK within this range - we are operating @ 60 odd degrees to resolve a critical pair of impurities. Strangely enough the 5% starting point appears to have the smallest impact on peak shape! Any suggestions to why this is so. We either see what looks like a "shark fin" tailing (@ 40% starting composition) or a "shark fin" fronting (@ 20% starting composition). Any ideas on what starting composition to try and what solvent diluent strength should be?

I'm not 100% sure about the dwell volume of this particular system. It is a high pressure binary Agilent 1100. We have measured the dwell volume of 3-4 other similar systems and they all came out at about 1.1 - 1.2 mL. Does this sound correct for this system anyone? When I changed this value in DryLab (varied by + or - 0.3 mL) it only seemed to affect tR by less than 5% for the early eluting problem peaks.. are confirmationary run peaks are off by 20%!

[adam]

Just one correction: from the standpoint of retention time, the peaks that elute in the middle or end of the run will be more effected by dwell volume than the peaks that elute very early.

With respect to peak spacing, it is the peaks in the early to middle part of the gradient ramp that will be most effected.

[Alex Buske]

Did you change the void volume in the data emtry or in the result page? That makes a difference.
I would expect the best peak shapes at 5% starting: The diluent mixes with mobile phase at the start of column and analytes stick to stationary phase.

Alex

[tom jupille]

Rob, the biggest problems with modeling early peaks have to do with "pre-elution" and mixing volume effects.

If the early peaks are weakly enough retained to start moving through the column isocratically before the gradient arrives, then the gradient model won't give accurate predictions (for obvious reasons). It's been a couple of years, but my recollection is that if there's more than about 30% pre-elution, the DryLab program will show a warning dialog, flag the peaks as being "estimated", and they will be gray in the simulated chromatogram. This used to cause me all kinds of grief when outraged users would call and say "whaddaya mean, my data isn't any good ?"

The other problem that will affect both very early and very late peaks is rounding of the actual gradient profile because of mixing effects.

There is a thorough discussion of these effects in a paper by Ghrist, Cooperman, and Snyder, J. Chromatog, 459 (1988) 1-23.
If you can't find a copy, let me know; I think I still have a couple of spare reprints in my file cabinet.