Separating a small peak from the tail end of the main peak

[southmelon2003]

Hi All,

I am working on improving the resolution of the main drug peak and a peak (~0.4% by area) that is at the tail end of the main peak. I can only get baseline separation of these two peaks if I run isocratic condition. The compound is an acid with Pka of about 4, my mobile phases are 0.05% TFA/Water and 0.05% TFA in Acetonitrile. I can get the main peak to elute at 12 mins using Zorbax SB-CN, 15cm column, isocratic 35%B, with the desired the separation. However, my peak height is only 250mV, I tried to inject more compound( right now is at 100ug/mL, 20uL), but the peak shape is distored. I can of course get a higher response if I elute the main peak in the gradient portion, but I can not get baseline separation of the peaks of interest. Does anyone has any suggestions?

Thanks!

[DR]

More TFA?

[tom jupille]

More TFA?

. . . or less TFA.

Other possibilities include changing the temperature and tweaking the steepness of the gradient (usually, but not always, shallower = better). However, any of these things may also change the spacing between other peaks.

[Daren]

Or you can use no TFA at all and raise the pH of your mobile phase to around 6 and use TEA. Increasing the TFA to increase retention/separation works well as an ion pairing with basic compounds. For your acidic analyte you should try going above your pka, create an anion, and use TEA to act as the ion pairing agent.

[Srinivas SK]

Hi

Here's an idea that I've tried before, with success:

Assuming you're using a UV detector, with a programmable senstivity feature, try programming your detector to increase the sensitivity at the retention time of interest. Say from 0.1 aufs to 0.001 aufs, for example.

Peak height will increase significantly. Of course, there's a price to pay - baseline noise may also increase and there may be a drift in the baseline, but it's usually worth the trade-off.

There's another approach you can look at :

Try increasing the concentration of your sample and reducing the injection volume - try 200 ug/ml and an injection volume of 10 ul or 5 ul. You should get sharper peaks and better peak heights.

Hope this helps. If you do try these ideas out, do let me know if they succeeded - or not !

Warm rgds,

S.K. Srinivas

[southmelon2003]

Thank you all for the suggestions, I will give them a try and let you know how it goes.

[HW Mueller]

Hm, are there still any detectors and data system arrangements (integration or whatever you want to call it) out there which respond to a sensitivity setting on the detector? Normally one would just jack up the display of the crude chromatogram to get what Srinivas suggested.
Also, injecting a lower volume with higher concentration may only give an improvement if your sample diluent has similar or higher elution power than your mobile phase, also of course, if there is an incompatibility of pH, some additives like ion pairing agents. Missed anything?

Furthermore, I wonder whether southmelon had his isocratic method at its optimum flow rate. I wouldn´t give up an isocratic separation so easily.

[Kostas Petritis]

I would actually try to change completely the selectivity of my compounds, with the minor peak coming first, followed by the major peak...

[Alex Buske]

Hans,
There are still systems where you can change the "sensitivity" or "gain" or "attenuation". That happens almost always when you have a centralised CDS and other Vendors HPLCs, that are connected via A/D converter.
Of course, changing "sensitivity"during the run doesn't change the S/N ratio and not the ratio between big and small peak, but you wouldn't be able to draw a baseline below the big peak.

Southmelon, if you can't get a suitable integration (I am sure there are some options in the software), try to change selectivity (playing with TFA, % organic, temp....) or use a longer column..

Alex

[rc_12321]

South change your method from isocratic to gradient one to have sufficient resolution.

[mbulentcakar]

South change your method from isocratic to gradient one to have sufficient resolution.

is there any possibility for a gradient run to provide better resolution than an isocratic one?

[rc_12321]

Certainly you can get better resolution with gradient runs.Try to study the nature of twopeaks of interest and logically try to find out the strengt of the mobile phase required to elute the peaks and programme your gradient composition accordingly.

[Rob Burgess]

This is the kind of example where I have used the Drylab modelling software to great effect... assuming you have access to it?

[tom jupille]

If you have only two peaks, then a gradient will provide approximately the same resolution as an isocratic run. In most cases, the isocratic will be slightly better; occasionally the gradient will be slightly better.

A gradient can resolve more peaks, but only if they have a wide range of retention.

At the risk of sounding pedantic, for a given pair of peaks, there are only three ways to improve resolution:
- increase retention (k')
- increase efficiency (N)
- change (improve) selectivity (α)

Increasing retention can only take you so far, because resolution is proportional to [k'/(1+k')]. If your peaks have k' values above about 4, you have already obtained 80% of your theoretical maximum resolution. If your peaks have k' values below 2, then this is the place to begin. Note that, in practice, changing k' may also affect selectivity (α), as discussed below.

Increasing N is very expensive, because resolution depends on the square root of N. Assuming you are starting with a reasonably state-of-the art column, something like doubling the column length (and hence doubling N) would only improve resolution by 40%. If your resolution is close to what you need, then this is often the most straightforward approach, especially for complex samples, because the improvement applies to the entire chromatogram (i.e., no need to worry that improving one part will degrade another).

Changing selectivity provides the greatest potential for improving resolution because resolution is approximately proportional to (α-1) [if you want to nit-pick, it's proportional to (α-1)/α]. The potential problem with complex samples is that a change which improves α for one pair of peaks may well degrade α for another pair, so that selectivity control becomes a balancing act.

While the preceding paragraphs apply to isocratic separations, the same arguments can be made for linear gradients if you substitute the average k' value (k*) for k'.