Change buffer or change column? Or else?

[pserodio]

Hi forum,

We are facing some challenges in redeveloping an original HPLC method to a new UPLC method for a intermediate drug product.

Some details below.

API: C26H20FN5O2S2, MM: 517.6 g/mol (2 benzene rings, pyridine, etc...)
Intermediate drug product: API + 1 single polymer

UPLC method
Mobile Phase A: 10mM K2HPO4 Buffer, pH 7
Mobile Phase B: MeOH : ACN (50:50)
Column: BEH Phenyl 100 mm x 2.1 mm x 1.7 µm
Column temp: 30 ºC
Autosampler temp: 5 ºC
Injection volume: 5 µL
Time (min) Flow (mL/min) %Mobile Phase A %Mobile Phase B
0.00 0.3 90 10
2.00 0.3 60 40
20.00 0.3 45 55
30.00 0.3 45 55
50.00 0.3 25 75
55.00 0.3 25 75
55.10 0.3 90 10
60.00 0.3 90 10
Diluent: [DCM/EtOH 80:20, 10% + H2O/MeOH 70:30, 90%] !!!
Nominal concentration: 0.1 mg/mL API

60 min in UPLC, I know it's to long but we really need to separate the impurity before main peak. And until now....this was the "best".
Some typical and consistent profiles are presented below (Blank and Reference Standard)

UPLC profile (full scale)
[img][IMG]http://i347.photobucket.com/albums/p475/pserodio/UPLC_profile_full%20scale_zpsru4bigf3.png[/img][/img]

UPLC profile (zoomed)
[img][IMG]http://i347.photobucket.com/albums/p475/pserodio/UPLC_profile_zoomed_zpsgoh2yhhq.png[/img][/img]

UPLC pressure plot
[img][IMG]http://i347.photobucket.com/albums/p475/pserodio/UPLC_pressure%20plot_zps52vebob9.png[/img][/img]

This UPLC method provides good separation/resolution for the pair (impurity imeediately before main peak/main peak). This was not achieved with the original HPLC although this was well balanced in terms of MPA (0.1% FA in H2O) and MPB (0.1% FA in MeOH/ACN 1:1). Diluent was similar Diluent: [DCM/EtOH 80:20, 10% + 0.1% in H2O/MeOH 70:30, 90%].

Issues in UPLC:
1) To optimize the overall profile is it critical to use 10 mM phosphate buffer in MPB? To avoid dilution of buffer concentration along gradient?
2) Highest limit of 25:75 MPA:MPB ensure no buffer precipitation? Pressure plot seems "strange". Small but cyclic vertical oscillations (buffer precipitation?)
3) Unexpected(?) low absorbance at 0.1 mg/mL. Higher injection volumes tend to broaden the main peak.
4) Blank is not "well clean". There are some interfering peaks. Our COP allows only up to 10% of LOQ!!!! Starting the gradient with less MPA can eventually help but will impact on separation power.
5) Baseline depletion tend to systematically occur after the main peak. How to eliminate this?

Should we try different column? Different buffer? Any other suggestion?

Can you please take a look on this topic in detail and give back some thoughts?

Many thanks,
Pedro

[Mattias]

Hi,

It is an "unusual" gradient profile going from 10 to 40% of mobile phase B in the first 2 minutes... But why not if it gives you the separation you need.

The problem with impurities in blank is commonly due to impurities in the mobile phase. My first suspect is usually the acetonitrile, but it can be anything that is in contact with the solution. Another quality of acetonitrile or methanol may help.
The baseline depletion is also present in the blank injection, so this is a baseline disturbance (most probably also due to impurities).

75% organic and phosphate buffer sounds risky, but the buffer conc is low. I would use a premixed mobile phase B (and go to 100% B instead)

[pserodio]

Hi,

It is an "unusual" gradient profile going from 10 to 40% of mobile phase B in the first 2 minutes... But why not if it gives you the separation you need.

The problem with impurities in blank is commonly due to impurities in the mobile phase. My first suspect is usually the acetonitrile, but it can be anything that is in contact with the solution. Another quality of acetonitrile or methanol may help.
The baseline depletion is also present in the blank injection, so this is a baseline disturbance (most probably also due to impurities).

75% organic and phosphate buffer sounds risky, but the buffer conc is low. I would use a premixed mobile phase B (and go to 100% B instead)

Thanks Matthias.

In fact we have tried premix MPA and MPB to have always the same buffer concentration (10 mM) along gradient. No special improvement.

Starting with 40% MPB was also tested and it seemed to impact on the first eluting impurity (start splitting into two peaks) and slightly worse the separation of post main peak impurities. But we are still exploring this "route".

We are also facing poor sensitivity (~ 0.40 ABS when we inject 5 uL of 0.1 mg/mL nominal concentration). When we try to inject 10 uL or increase nominal concentration the separation of critical pair become unacceptable.

Any thoughts on how we can increase sensitivity without sacrificing separation? Any other ideas to get optimized conditions?
Thanks

[mattmullaney]

Hi Pedro,

A possible solution: I didn't see the diluent for the intermediate product, but you could add water to the sample and inject a proportionally higher volume onto the column (if the diluent isn't 100% water). For example, add 2 volumes of water to the prep and inject 15 microliters.

[mattmullaney]

Apologies, I re-read the original post...still seems worth trying to me. Just dilute the standard or sample solution with the diluent, doesn't matter how complex the diluent solution is, and inject proportionately more onto the column and see what happens. Also, if the autosampler permits, pick up 1 or 2 microliters of air prior to the solution you're injecting, this will help focus the solvent plug as it is placed onto the column.

Best Wishes!

[pserodio]

Apologies, I re-read the original post...still seems worth trying to me. Just dilute the standard or sample solution with the diluent, doesn't matter how complex the diluent solution is, and inject proportionately more onto the column and see what happens. Also, if the autosampler permits, pick up 1 or 2 microliters of air prior to the solution you're injecting, this will help focus the solvent plug as it is placed onto the column.

Best Wishes!

Thank you very much mattmullaney for you suggestions. In fact diluent is a very complex mixture and we have noticed that, when the gradient starts at 90% A, injection volumes 2, 3, 4x 5 uL will impact on peak splitting of the first eluting impurity. But the proportional injection (the same amount on the column) of 5 uL using 4x nominal concentration, does not slip the first peak. However, critical pair separation becomes worst. We will try to start at 50% A to see if we can increase sensitivity.

In the meanwhile, any other ideas would be for sure very valuable.

Thanks again!