RT variation between Quaternary and Binary Pumps

[deepankargupta143]

An RS method was developed on Agilent HPLC 1260 series with quaternary pump. The chromatographic conditions are as follows:
M.P-A: 25mM Ammonium Formate Buffer, pH 4.0
M.P-B: Acetonitrile
Column: C8, 150 x 4.6 mm, 2.7 µ
Detection wavelength : Excitation at 360 nm, Emission at 460 nm
Detector: FLD
Gain: 10
Response factor: 0.1
Injection volume: 20 µL
Sample temperature: 5 °C
Column temperature : 30 ˚C
Run time: 40 minutes
Flow rate: 1 mL/min
Gradient Program
Time (minutes) %A %B
0.0 20 80
10.0 20 80
25.0 15 85
30.0 13 87
31.0 00 100
35.0 00 100
35.1 20 80
40.0 20 80
The RT of the main peak was observed at 26-29 min.The same method when transferred to Agilent HPLC 1200 series with binary pump, the RT of the main peak was observed at 22-24 min but the RRT of all the impurities matches with that while using the quaternary pump.

Can anyone suggest what can be the reason for this variation and what can be done to resolve this issue?

[tom jupille]

What are the dwell volumes on the two systems? And have you done PQ to verify the proportioning accuracy?

[Consumer Products Guy]

Quaternary pump uses low pressure mixing before the single pump.
Binary pump uses high pressure mixing after the two pumps.

For an isocratic analysis, should be similar if dwell volumes are similar.
For a gradient analysis, differences such as what you've observed are not unexpected, are typical.

Welcome to real life.

[lmh]

no, it's not how everything should be. Not at all. The original poster has specified that their target peak has moved but all the contaminant peaks are at the same retention time as before. This is not typical behaviour if they simply moved the column from one pumping system to another. If it were a matter of dead volumes, everything would just move a bit earlier or a bit later. If it were a gradient inaccuracy, then the spacing of the peaks might change, as well as their retention time, but you'd have to try really hard to get a peak to move 4 minutes while another doesn't move at all (I wish I could do that, every time I get co-eluting peaks!). The peaks should come out in more-or-less the same order on the two systems, even if their actual retention times change!

My concern is that the windows for the peaks are already being quoted as very wide 26-29 or 22-24min. Are the peaks 2-3min wide, or are they moving around? The gradient is very flat, so it's possible the peaks are broad, and flat gradients will also give much greater variation in retention time with small changes in solvent composition. The method also uses a buffer that can be hard to make up completely reproducibly (ammonium formate is a bit hygroscopic so you can never be totally sure how much water you're weighing! Different analysts also go about making ammonium formate buffers in different ways). Looking at what's happened, I worry that this may be a method that isn't very robust, and it's revealing its lack of robustness when using a different system with differently-prepared buffers.

[Consumer Products Guy]

Still seems like an overly complex gradient to me. We typically have way shorter run times than that for 98% of our assays.

[HPLCaddict]

no, it's not how everything should be. Not at all. The original poster has specified that their target peak has moved but all the contaminant peaks are at the same retention time as before.

No . Look at that post again. The original poster wrote that the RRT (= relative retention time) of the contaminants did not change, which means that their retention time changed in the same way relative to the main peak. Just as expected from a change in dwell time...
Generally, I would not worry too much about that shift in absolute retention time, although I share the opinion that this gradient doesn't look too optimal, just from my guts. Very long initial isocratic hold, very flat linear gradient segments and 80% acetonitrile is quite a high level to start a gradient.

[lmh]

sorry, yes, I was having a bad day and not reading properly! And yes, I agree, long flat bits and very shallow gradients aren't very efficient - my limited experience is that people put them in because they've got poor selectivity between peaks (things eluting too close together) and what happens is the peaks get twice as far apart and twice as wide....