separation problem

[mememe]

Hi there!

I'm having a problem with a method I'm trying to develope. I'm analizing DNPH-derivatized carbonyls and it seems i can't fully separate two compounds, derivatized crotonaldehide and methyl ethyl ketone.

As mobile fases, I'm using for A a mixture of water, ACN, THF and IPA, and for B a 65/35 mixture of ACN and water.

The column is the same as in the already developed methods i found. I tried many radients, tried to change the composition of mobile fase A (less ACN), changing the column temperature, decreasing the flow but nothin works.

Any sugestion? Could increasing the pecentage of THF in the mobile fase A solve the problem?

[tom jupille]

Could increasing the pecentage of THF in the mobile fase A solve the problem?

It might. So might decreasing it, or increasing/decreasing the percentage of IPA. The best situation would be to contact the developers of the original method (or get hold of the validation package) to find out *why* they used the ratios they did. Failing that, set up a series of experiments where you change (or even eliminate) *one* of the components at a time. If there is no effect on selectivity, then you can go back to the original level (or, if you have eliminated the component and the selectivity didn't change, just leave it out entirely. If there *is* an effect on selectivity, then spend the time to optimize that component.

Finally, it may be that you need a different column (you have already looked at gradient steepness and temperature).

[Consumer Products Guy]

THF definitely can affect selectivity; we use it mainly when we WANT to affect the selectivity to help us; otherwise, why bother.

This is the fun stuff. I typically set up a series of experiments to learn what happens. Like 10% higher THF and 10% lower THF. Like higher and lower IPA. Like 10 degrees higher column temperature and 10 degrees lower. Like earlier hitting gradient and later hitting gradient. Flow rate won't affect this much unless you're pretty far off optimum linear velocity.

Also, make sure that you are not overloading your column by injecting less volume. And maybe try a different type of column, like there's tons of C18 phases.

For each, the question is: did my change make the separation better, worse, or no effect. That should point you in the correct direction.

"I'm using for A a mixture of water, ACN, THF and IPA, and for B a 65/35 mixture of ACN and water."

That's a pretty complex mixture of solvents, we never use anything that complex. Is your matrix so complex that you can't just start without the IPA (and maybe the THF)? When we've made DNPH derivatives of aldehydes, we've used wavelengths like 336nm so the matrix wasn't much of an issue.

[mememe]

the samples are complex because it is cigarette smoke I'm analising, a lot of peaks from impurities that coelute with my target peaks. The wavelenght in the method is 360 nm. What realy bothers me is that with both the columns (not the same manufacturer) I used I didn't get the results from the methods I tryed, very different RT and lower separation. Is it possible that the selectivity of the columns isn't just high enough? i don't have the experiences to judge.

[tom jupille]

Let's back up a bit.

If you are trying to run a published or previously validated method, you need to be aware that methods are validated for a specific column from a specific manufacturer, and that columns are *not* interchangeable (in other words, a diol column from vendor A will not give the same chromatography as one from vendors B, C, or D). For simple samples with only a few peaks, those differences may not matter, but the more complex the sample, the conditions need to match closely in order to resolve all the peaks of interest.

In many (but not all) cases, column selectivity differences can be compensated for by adjusting the mobile phase or the temperature, but this involves, in effect, re-developing the method. CPG outlined the approach: start with the "nominal" method and then do "offset" runs on each parameter to see which ones have the biggest effect.