Hi,

I'm searching for the best strategy for method development. I realize that this can't be answered with a simple yes or no, but I'm trying to get as manny stategies as possible to choose the most suitable strategy for our company.

I hope also that others can learn from this too.

The key element in efficient method development is an early screening of the influence of chromatographic variables. This is then followed by finetuning.
We have used the following generic protocol, applicable to all compound classes (small molecules that are not too polar):
1. Use 3 or 4 columns of maximum selectivity difference. We have used a standard C18, XTerra MS C18, a packing with an embedded polar group, XTerra RP18, and a phenyl packing, XTerra Phenyl.
2. Run screening gradients on all columns using at least methanol and acetonitrile as organic solvents at low pH (formic acid or ammonium formate) and high pH (ammonia or ammonium carbonate). (=4 gradients)
3. Examine the data and select the most promising separation for further finetuning.
From this blind study, you get a ton of information for further finetuning of the separation. You know if the retention properties of all compounds of interest are close enough to go for isocratic separations or if only a gradient separation is feasible. The most generic tool for both isocratic and gradient separations is now the solvent selectivity. Sometimes, the fine-adjustment of pH is useful as well. In rare cases, the first column chemistry does not get you to a separation, but now you know which other column chemistry might be useful as well.
If you contact me, I send you a paper on this.

Uwe's approach is universal and useful in most cases. Before you start you should answer yourself some questions:
1. Detection: are the analytes suited for UV (fluorescence, MS,...) detection?
2. Nature of analyte: Can it be ionised (acids, bases)? If not, then screening with buffers at different pH doesn't make sense. Try water.
3. Nature of analyte/matrix: is it soluble in mobile phase (and the solutions stable)?
4. If your lab has a "standard column" use that one first.
When selecting columns it does make sense also to use "similar" columns (eg. C18 from different vendors) as they might show different selectivities. Column switching valves can save much time.
The organic solvent has a strong influence on the separation (often more than using different columns). Beside acetonitrile and methanol also mixtures (1:1, 1:2 and 2:1) of both can be useful in screening.
For evaluation of data from scouting (=screening) runs programs like DryLab or Chromsword can help.
However, method development includes a lot of other issues beside chromatographic resolution: sample preparation, solution stability, quantitation (especially when you work at different wavelengths), system suitability parameters and so on. Here very much depends on the actual system (analyte(s), purpose of analysis, cost issues....) and no general advise can be given.

Uwe and Alex made some important statements. I'd like to add some thoughts:

- If possible, check the chemical structure of the compounds before choosing a column. Bases could react irreversible with acidic silanols from the stationary phase.

- Check out the stability in the choosen mobile phase of your compound. Don't forget main impurities. Sometimes the main compound and/or the main impurities tends to react in some very unpleasant ways.
Changing the pH of the mobile phase could prevent this, even when your main compound/impurities seems to be neutral.

- It's (normally) the best way to solute your sample in mobile phase. But sometimes it is nessecary to solute your sample in a different solution than mobile phase. When using THF or a buffer, injecting more than 2 or 5 µl could lead to misshapen peaks, caused by strong interactions of THF or buffer with your sample.

- Articles about buffered mobile phases:
http://www.lcgcmag.com/lcgc/data/articlestandard/lcgc/462002/38298/article.pdf
http://www.lcgcmag.com/lcgc/data/articlestandard/lcgc/502002/40823/article.pdf
http://www.lcgcmag.com/lcgc/data/articlestandard/lcgc/052003/45059/article.pdf

- If your compound tends to build complexes be carefully when unsing acids as modifier in the mobile phase. While sulfuric acid could lead to nearly no retention at all, trifluoracetic acid could increase the retention in some fascinating manner - for example when you have to analyse a small, polar and basic compound and use a fluorinated stationary phase (like Fluofix or Fluophase).

- Analysing (even weak) acidic compounds on polar embedded phases showes sometimes no peaks at all. I'll never use a embedded polar phase without any modifier in the mobile phase again... ;-)

- Organic acids were not only acids, they were organic solvents too. Sometimes that's the difference between seperation and co-eluation. Keep this in mind when replacing an anorganic acid with an organic acid in order to use lc-ms.

- Use only fresh tetrahydrofuran. Store THF under nitrogen and in the refrigerator. Running analyses at 220 nm with TFH in the mobile phase is not impossible. When using mobile phases with THF at low wavelenght always check absorbance against pure acetonitril to make sure you're not out of linearity.

- Never be afraid of giving unusual ideas a try.

Greetings, Andreas

ANdreas,
Good suggestions!
I would like to contradict you for the case of packings with embedded polar phases. Many of the packings with embedded polar phases behave like you described, because they were synthesized by amateurs in a two-step bonding process, with an amino-phase bonded in the first step. This leaves residual amino groups on the surface, and you can wave good-bye to the acids that you are injecting. Other phases with embedded polar groups have been designed to eliminate this problem...

Uwe,
I think I know who is the amateur and who is the pro...at least in your eyes. And I think we all know what "other phases with embedded polar groups" you are adressing ;)

(Just kiddin' a bit)

This approach like Uwe has described sounds universal, but is this the only universal approach? Are there other approaches suitable?

Oh, many experienced method development chemists have developed their own approaches. There are methods that step through the pH scale in blind overnight runs. There is the old three-solvent method by Snyder. We used to use a simpler method as well, on a single column.
The main reason why I like this particular method is because it gives me a ton of information up front, and it gets me halfway to the final method.

Is it possible to mix for example KH2PO4 and K2HPO4 solutions with the HPLC system so you can explore the whole pH scale overnight?

If it's possible what's the best approach for this?

It is possible. Keep in mind that pH is not linear with K2H / KH2 PO4 buffer ratios. Best approach depends on your special problem.
To Andreas: I would try to avoid THF, at least if you are heading to a routine analysis.

It is possible, but you need to remember that with these two salts you will cover only a limited pH range. You may consider using H3PO4/KH2PO4 for the acidic pH range as well. If you do not mind going blind in the low UV, you may add acetic acid for the pH range around 4.75.

Dear Alex,

sometimes there's no real alternative to THF. Using DCE together with water (and lots of MeOH) could work but it's no pleasure. Acetic acid has the disadvantage of being an acid and could damage column and system when used in higher concetrations (together with small amounts of water).

To H3PO4:
in low pH systems ammonium sulfat is a good replacement for phosphoric buffers. One of the great advantages is that only small amounts of low concentrated sulfuric acid is needed to adjust pH to 3 or lower. Low UV detection is possible but not as good as with KH2PO4 or NaH2PO4.
In some special cases sulfat buffers are recommended to detect all peaks.