TEA substitute for LCMS

[Jasmine]

Hi,

I have a method that I used to run with HPLC-UV and I have used 0.1%TEA as an additive to obtain sharp peaks. Without TEA, my peaks were short and slightly fat. I now intend to transfer the method into our LCMS equipment. Can anyone suggest any good substitute for TEA and what concentration should I use?

Thanks

[MG]

What compound are you analyzing? What column? What ionization mode do you expect to use (if known)?

[Jasmine]

Hi MG,

I am analyzing a few highly polar compounds. I have used a C18 column before. With this mass spec method, I have done preliminary runs on several reversed phase MS columns like Extend C18, Xterra C18, Xterra Phenyl, but a couple of them come out very fast and some coelute with each other. I also tried Atlantis columns, but results were very much similar. When I reduce the solvent strength, I was able to achieve better separation, but the peaks ended up broad and short. I am using positive mode of detection. Can you suggest any additive that I can use to make the peaks sharper like the way I was able to achieve in my old method using TEA?

Thanks for replying

[Uwe Neue]

Jasmine

Give us a few more details on how you run your chromatography! Column dimensions, flow rate, gradient, pH, what additive (e.g. formic acid) etc. At this moment, I do not understand what you are doing, so I can't give you advice.

[Jasmine]

Hi Uwe,

The dimensions of the columns that I have tried so far are either 2.1x50mm or 2.1x100mm. The gradient solvents that I used were (A) 10mM ammonium acetate with 0.1%formic acid (pH about 3) (B) 70% acetonitrile. The gradient condition that I have tried is 100%A initially increasing to 80%B at 2.5mins, and then the gradient moved back to 100%A again later. The flow rate is at 0.3ml/min. Some peaks come out well separated and sharp too but a couple stayed near the solvent front and the shape of the peaks are not sharp/symmetrical. When I tried to use isocratic with 75A/25B proportion at same flow rate, all the peaks are well separated but have poor shape, hence I need an additive which hopefully can make them better.

Thank you

[Uwe Neue]

We have several issues with the method. The gradient method uses a very steep gradient. For the long column, the ratio of gradient volume to column dead volume is as little as 3.3. For the shorter column, it is about 6.6. Gradients that are that steep do not separate very well. My recommendation is to at least double the flow rate. If your MS system lets you do that, I would even go to 1 mL/min. A good chromatographic method over a solvent composition range as wide as yours should span at least 10 column volumes.

The source of the broad peaks early in the chromatogram is most likely the injector or the injection solvent. Going to a higher flow rate could help with this as well. If this does not help, dilute your sample with more water (I assume that the sample is dissolved with the help of some organic solvent). This is the only way to get rid of the ugly peaks early in the chromatogram. Since the source of the ugly peaks is mechanical, the addition of TEA to the mobile phase will not help.

The ugly peak shape in isocratic mode is most likely related to the same issue, plus the added influence of pre-column bandspreading. Pre-column bandspreading tends to go away in gradients, except for the early peaks if the sample is dissolved in a solvent composition with more organic solvent than the start of the gradient.

Bottom line:
1. Increase the flow rate
2. Dilute the sample with water

[JA]

For the long column, the ratio of gradient volume to column dead volume is as little as 3.3. For the shorter column, it is about 6.6. Gradients that are that steep do not separate very well. My recommendation is to at least double the flow rate. If your MS system lets you do that, I would even go to 1 mL/min. A good chromatographic method over a solvent composition range as wide as yours should span at least 10 column volumes.

Uwe, could you please go through that ratio determination for me? I'm sure I'll be like when I read this at work tomorrow..

Also, you use column dead volume and later column volume in suggesting a gradient spans 10 or more column volumes. I just wanted to make sure these are separate?

Thanks.

[Uwe Neue]

OK, I had thrown my spreadsheet away, but it is not a problem.

The column volume (of the short column) is 0.173 mL. The column dead volume is about 66% of this: 0.114 mL. The gradient volume was 2.5 minutes x 0.3 mL/min = 0.75 mL. The ratio of the gradient volume to the column dead volume is about 6.6.
The same calculation would result in a value of 3.3 for the longer 10 cm column.
In my mind, any gradient with a ratio of gradient volume to column dead volume of less than 10 is a very steep gradient, especially if the gradient covers a wide range of organic solvent composition. My rule of thumb is that the gradient just becomes useful, when it ranges over 10 column volumes with such a large difference in solvent compoistion from the beginning to the end of the gradient. Anything steep than that especially if it is as little as 3 column volumes, does not result in a useful separation.

References: Anal Bioanal Chem 377 (2003), 788-802 or Rapid Commun Mass Spec 15 (2001), 141-151

[tom jupille]

To put another perspective on Uwe's post, a gradient volume of 3 times the column volume is equivalent to an isocratic separation with k' values around 0.5 (not conducive to good results!).

[Jasmine]

Dear Uwe and Tom,

Thank you Uwe. I will try your advice. I want to tell you too that the sample solvent that I used was 0.05% Formic acid, not a strong solvent. I am glad you explained about the column/dead volume and ratio and stuff because I was a little bit lost there. I am a bit of an amateur LC person you see

Tom, can you please tell me how you arrived at the equivalent value for isocratic?

Thank you

[tom jupille]

Jasmine, one way of relating (linear) gradient and isocratic separations is to use the "average" k' of peaks during the gradient (this is the k' that peaks have at the midpoint of the column); that average is symbolized as k*.

Steeper gradients are equivalent to smaller k* values, which makes intuitive sense (shorter times, sharper peaks, poorer resolution).

If you want to get deeper into it, check the thread on HPLC Methods development for Basic Compounds:
http://www.sepsci.com/chromforum/viewto ... highlight=

Uwe and I have gone around on this topic a couple of times, but if you look at the math, you see that his estimate that the gradient volume should be at least 10x the column volume for a wide range gradient is equivalent to saying the k* value should be at least 2. A k' value of at least 2 is, not coincidentally, the FDA recommended guidance for a validated HPLC method.

[Jasmine]

Thanks everyone for your help, I will try all your advice and suggestions.

[adam]

If I'm not too late, can I ask a question and also make a comment regarding this discussion:

Uwe: for what reason did you suggest a higher flow rate?

Uwe and Tom: I wonder if there is such a thing as over-intelectualizing a problem. Personally, I've been doing HPLC for some time now, but never have I found it necessary to think in terms of "ratio of gradient volume to dwell volume" and as for "linear solvent strength theory" yikes! What a headache, and - in my humble opinion - of zero practical utility.

My answet to Jasmine would have been "the gradient is way too fast (nice and simple - end of discussion). The rule of thumb I've always heard is that the '% increase in B solvent / min' should be no more than 3. In her original method it was 35. Way too fast.

Uwe: I've heard you advocate the KISS principle in other posts. Why go the other way here?

Holler back. Curious to read the responses.

[tom jupille]

Adam, I understand where you're coming from. I'm sort of flattered to be thought of as "everintellectuallizing" because in the past, I've been accused of doing "boy scount merit badge"-level" chromatography .

For better or for worse, I like to try to understand what makes things work. If someone tells me the rule of thumb is "3%/minute", my first question is "why?" The trick (which I don't succeed at often enough) is to explain why in terms that are meaningful to the listener/reader.

[MG]

Adam, my methods rarely follow that rule of thumb. In general, I think my gradient methods usually increase 5% to 9% per minute A to B. But then again, I am usually running at 0.35 to 0.45 mL/min on a 2.1mm id column, higher than the "usual" linear velocity, so there gradient volume would not be too small.