Please give suggestion: high polar impurity purification

[blackdrum]

Recently we are trying to identify a high polar impurity (1% by UV) in a API, after several months’ hard work, we get some progress, but the result isn’t very good, so I need expert’s suggestion, below are background and our procedure:

API purity method: ion pair chromatography MP: acetonitrile/water(SDS) column: phenomenex Gemini 150*4.6 3u

Hilic methods: waters atlantis hilic silica column and phenomenex luna hilic column don’t work, target impurity and API can’t be washed out even we run gradient to 60% water, with or without TFA, finally we find supelco discovery HS F5 column works, but because of the poor peak shape, resolution and loadability, we give up the idea to purify target impurity by this kind prep-column directly (actually we tried, but the impurity came out together with API even we injected about 10mg API on a 250*21.2mm column)

Our procedure: 1. run prep-hplc to get target impurity with SDS by phenomenex Gemini column(250*21.2mm, 5u), 2. freeze-dry the fractions, 3. get rid of SDS by ion exchange SPE (waters oasis, first by cation exchange to take out most SDS, then by anion to get rid of left), 4. run hilic prep-hplc to get rid of a UV transparent impurity by supelco discovery HS F5 column(250*21.2mm,5u), 5. run TOF ms and ion trap ms to get ms tree, 6. run high sensitivity NMR to get structure info(1D and 2D)

But the problem is: after step 4, we can’t get even 1mg target impurity though we have run 10g API, we can’t gain any signal form 1C NMR or 2D, and our target is at least 20mg, such a procedure obviously doesn’t work. I know some guys here have strong chemistry and chromatography background, maybe you can give me a totally different way to do the purification, please help me! And, please, don’t use ad to confuse me.

[Kostas Petritis]

Substitude your SDS with a high chain perfluorocarboxylic acid such as Tridecafluroheptanoic acid or Pentadacafluorooctanoic acid. You should be able to get rid of it by speed vac after snap freezing your samples. In this way you might be able to avoid step three where in my opinion you must be loosing a lot of your impurity.

Even if you are left with some of it in your sample it won't interfere with the MS, especially if you do flow injection analysis (FIA) experiments by using a C18 pre-column (or a small analytical column, i.e. the PDFOA will stick in the column while your compound will wash through, alternatively you can do C18 clean up... now that I am thinking about it, C18 SPE clean up would have worked better with the SDS also... SDS is retained your compound is washed through...).

You may want to use an ELSD in order to track down your UV transparent impurity (how are you monitoring that HILIC separates your impurity of interest from the undesired impurity?...

[Patrik Appelblad]

If the target impurity and API can’t be washed from your HILIC column, it probably is an indication that you have a very strong interaction.

The experimental conditions are not given, only that you have tried gradients to 60% water, with or without TFA. My guess is that if you alter your experimental conditions, i.e. ionic strength of the mobile phase, pH and buffer system, both the impurity and the API should be possible to elute, and probably also separated from each other.

Additionally, TFA will act as an ion-pairing reagent, affecting the HILIC mechanism.

In our "Practical Guide to HILIC" we strongly recommend users to stay away from IP reagents, as of this reason. If you have not received a copy you can order it free of charge (http://www.sequant.com/hilicguide).

If you could describe your HILIC setup in more detail, we can provide you with more input.

[Uwe Neue]

It is unusual for an analyte not to elute from a silica HILIC column. Have you considered the opposite, i.e. that the analyte is not retained with the TFA mobile phase?

[SIELC_Tech]

This is comlicated approach. Here is my 2 cents:
Your compound is basic in nature with possibility of few basic groups. with hilic approach you have a very strong interaction due to polar mechanisms and due to multiple ionic interactions. You need more buffer/salt in your mobile phase. Another possibility is that your compound co-elutes in HILIC mode.
I would suggest to go with SIELC mixed-mode HPLC. We have several columns with ion-pairing reagent attached to the surface of silica. mIxed-mode will allow you to reduce your 5 steps to one or two, it will give you alternative selectivity and provide you with increased capacity (up to 50 times compare to RP or HILIC approach).
Your compounds will retain based on reverse phase and ion-exchange mechanisms, it will provide you with 2D separation on one column:
http://www.sielc.com/Technology_2D_Properties.html


It is very good approach for prep separation: you are not limited by solubility (like in HILIC) or overloading (RP)
http://www.sielc.com/Technology_Prepara ... raphy.html

http://www.sielc.com/Products_Obelisc.html (comparison of traditional C18 and mixed-mode approach)

Also here is an article for loading capacities of SIELC columns compare to RP:
"A study of retention and overloading of basic compounds with mixed-mode reversed-phase/cation-exchange columns in high performance liquid chromatography" by Nicola H. Davies, Melvin R. Euerby and David V. McCalley (Journal of Chromatography A, Volume 1138, Issues 1-2, 5 January 2007, Pages 65-72).

Please contact me by phone or email and I will try to help you resolving your problem. Alternatively we can develop method for you free of charge.

http://www.sielc.com/AboutUs_ContactUs.html

If you are in US - Have a great Memorial Day weekend

[blackdrum]

Thank you all, I'll try every possible directions.

some additional info about API: two secondary amines, one boric acid.

we tracked down UV transparent impurity by ms, thank god, it has strong ms signal, actually, we almost finish the identification of this impurity because we got enough amount.

we ran a lot of hilic methods: isocratic between 95% and 40% acetonitrile, gradient from 95% to 40% acetonitrile and several methods within this area. we can't find API's mass and there isn't any peak around void time; there should be a strong interaction. Though we can run hilic mode on PFP column, the result isn't good.

I'm reading the provided references, then I'll make some changes, please give more suggestions.

[Uwe Neue]

If there was no peak in HILIC, it it probably due to ion-exchange interactions. I would use ammonium formate around pH 3.5 instead of TFA.

[Carl Sanchez]

You could also try HILIC at pH 8.5 (aqueous pH using ammonium formate) to change the ion exchange and hydrogen bonding properties of the molecule. The Luna HILIC column is stable at this pH. Maintain the buffer concentration constant at 5mM (wrt ammonium) in both mobile phase A (95% ACN, 5% 100mM buffer) and mobile phase B (50% ACN, 5% 100mM buffer, 45% H2O). You can make the buffer by making 100mM aqueous ammonia then titrating to pH 8.5 with formic acid.

[vickig]

We have found that methanol is better than acetonitrile for chromatography of Boron-containing compounds

[Carl Sanchez]

Just a clarification on your last post: methanol (instead of acetonitrile) works better in HILIC mode or RPLC? Which phase was used in which mode.
Thanks

[vickig]

RPHPLC

[Nalizer]

While ago we faced similar problem with isolation of impurity during degradation studies. Our compound was a partially acylated amino sugar. We dedicated a lot of time optimizing/screening columns (IP, IE, RP, etc.) and narrowed it down to HILIC (Atlantis from Waters) and Mixed-Mode (Primesep from SIELC). After loadability study ended up with 10x150 mm Primesep column....by the time we isolated 100 mg, our project was transferred to India….but we learned a lot!

[blackdrum]

After half a year’s hard work, we come to close this project; we use a total different procedure to do the purification, it’s more convenient and efficient than before: 1.use n-butanol/water(ammonia) to extract target compound, after that, target compound’s content increase from 1% to 25%, the extraction efficiency is about 40%. 2. isolate target compound by waters xbridge c18 column with water( 0.1% formic acid ) as MP, IPA is eluted at dead time, and target compound is eluted at 5min, so we can get pure target compound.
The most important part is step 1, because of the high content of target compound, the overlap in step 2 is minimized. This project gives me two lessons: 1. try all possible ways to do the purification, not only restricted by chromatography, though it’s not easy for a chromatographer, sometimes, chromatography is a low efficiency way. 2. avoid using non-volatile additive in MP when you do preparative HPLC, it’s low efficient and time consuming to clean it up, I know almost everybody here know that, but it’s difficult to do it.
Actually I enjoy the process of this project, it’s very interesting to do the interpretation of unknown compound, you do the purification and get info from MS, NMR and put them together, finally, you get the only structure, feel like a detective.

[mbicking]

Could you give me some information on your PFP results? This phase does some unusual things if you know how to adjust the mobile phase.

You can send privately if you don't want to clutter up the board with details.

[blackdrum]

For PFP column, it surely can give different retention and selectivity, we just use it in hilic mode, isocratic or gradient, TFA or FA as additive. But its biggest drawback is loadability, for this project, it is about 1/5 of a C18 column, so we can't scale up. Btw, I feel if someone do chromatography for a long time, because it's so powerful, he'll rely it very much and forget other means, it's very harmful.