Agilent vs. Waters - peak discrepancy

[dkollmorgen]

We are running a derviatization method using an Inertsil C18 column with the following:
Buffer solution: sodium citrate dihydrate and anhydrous dibasic sodium phosphate, pH 8.

Mobile phase A
(85:15 v/v) Buffer solution : acetonitrile

Mobile phase B
(30:70 v/v) Buffer solution : acetonitrile

Diluent
Dissolve 29.4 g of sodium citrate dihydrate in 1 liter water.

Gradient conditions:

Time
(minutes) MP A MP B
0 100 0
15 50 50
25 0 100
27 100 0
32 100 0

Here is the problem: The active peak on both the Agilent and Alliance systems exhibits the same response, but one of the related compound peaks, which elutes around 28 minutes on both systems, responds differently on the two systems. The Alliance systems give a larger response (2 -10 X) for the related compound peak than the Agilent systems. We have tried this method on multiple systems, columns and with the same sample preps.
Has anyone had a similar experience? We are having trouble figuring out what is going on. Could it be anything with injection system differences? The way the injectors rinse? Dwell volume differences of the two systems?? It's strange that it is only occuring with the one peak and not the others in the chromatogram.

Thanks in advance for any assistance!

Diane

[juddc]

In order to determine whether it might be a difference in dead volume, it might be instructive to measure it on both systems. Also, more information as to exactly what the systems are (i.e Waters components vs an Agilent 1200 or a Waters UPLC vs an Agilent 1050) would help.

I suspect differences in dead volume are causing your gradient to shift a bit and your peak is sharper on the Waters system due to those differences.

Compensate for the differences in dead volume by modifying your gradient times accordingly and see what happens.

[Mark Tracy]

I think that your peak at 28 minutes is riding a gradient artifact. The different gradient delay volumes of the two instruments will change the time of the artifact. The artifact may not be directly visible at your working wavelength. Right now, you have a very steep gradient starting at 25 min, and it probably is appearing at the detector at around 28 min. So, try changing the gradient to have an isocratic hold at 100%B from 25-28 minutes before returning to initial conditions. It is not a good idea to have a peak that elutes during the re-equilibration stage of the gradient.

[dkollmorgen]

We're using Agilent 1100s and Alliance 2695s with 2487 detectors. I was thinking the same as you, but the peak shapes are beautiful on both systems and the retention is the same also.

[Bruce Hamilton]

I'm with Mark. Change your gradient to have the peaks coming off before the re-equilibrium time, as that will be much less affected by system differences, and will reduce the possibility of baseline effects.

Another possibility is that the detector spectral bandwidths or reference wavelengths are slightly different from each manufacturer, and your impurity's absorption just happens to fall on that boundary. Check the spectrum of the impurity versus the other compounds.

Bruce Hamilton

[tom jupille]

Three key pieces of information are missing here:

1. What column dimensions?
2. What flow rate?
3. What instrument dwell volume(s) ?

I'll throw some "typical" numbers in:

150 x 4.6 mm column
2 mL/min
Vd = 1 mL

That column will have an internal volume of about 1.5 mL. Add the dwell volume (gradient delay volume), and the end of the gradient is reaching your detector somewhere around 26-27 minutes. That's what judd, Mark, and Bruce were aiming at; it would take only a small difference in dwell volume or mixing characteristics to put that "28 minute" peak on the terminal isocratic hold or just at the edge of the reequilibration.

Extending the terminal hold time would improve the latter problem, but probably not do too much for the former. A shallower second segment on your gradient might help, but if it were my problem, I would increase the acetonitrile concentration in your B solvent by maybe 5%, then re-adjust the gradient profile as necessary to compensate for any selectivity changes.

[vofelyp]

Check the detectors wavelenght and the spectra of impurity. I think your measurement is run on the steep part of the spectra. The wavelength differencies can cause this phenomena. I found similar symptom in Dinoprost thrometamine USP method

[HW Mueller]

It´s not completely clear, is the same type of or one and the same detector used on both systems? If the detectors have the same performance on both systems you can simply rule out detection as having fault.

[wanda50]

I had a similiar problem, but with 2 different Agilent 1100 series PDA (analyst 1) detectors and a shimadzu SPD10Avp (Analyst 2). Analyst 1 analyzed Analyst 2 samples, both simultaneously. Analyst 1 got 0.06% for an impurity while Analyst 2 got 0.16% for the impurity.

Yes, I know about low vs high pressure and PDA vs UV detectors, bandwidth and all that. However, this was a significant difference. I tried 2 Agilent detectors just to see if it was system dependent. (No) The other variable is that I was borrowing unfamiliar machines while learning ChemStation.

We can't figure this one out.

Prodigy ODA-2, 5 um, 4.6 x 150 mm
1.0 mL/min
280 nm
30C
MPA - 0.5 ACN/95.5 water/0.01 TFA
MPB - 50 ACN/50 water/0.01 TFA

Gradient
0 100 A
20 0 A
20.1 100 A
25 End

[Mark Tracy]

One thing you might try is to run a linearity check on each instrument for the impurity peak. See if the intercepts are zero for both instruments (don't use forced-zero calibration). I suspect one will not be zero.

[wanda50]

Good point, Mark. Funny thing, though. Analyst 2 used several Shimadzu uv/vis and I used 2 Agilent PDAs. the only thing I can come up with is my bandwidth was too small - I used default settings ass-u-me-ing they were fine.

Boy, my credibility went into the toilet on this one.

[Uwe Neue]

Retention the same, detector the same, column the same... If all of this is true and you still get a difference, I would suspect adsorption to injector components.