System suitability limits for "fronting" peaks

[danko]

I'd hate to have to justify it to an auditor, though.

I realize that this is more of an art than science (music and art are two of my additional passions ) but it works OK never the less.
Addressing the issue of normalization: Really, the absolute value is not of any importance, so 1000 times more or less won’t change the ratio, as long as the normalization is done utilizing the equal sided triangle.
I hope one ore more of you guys will try the test one day and maybe come up with comments, ideas or just rejection.

HW Mueller, unfortunately I can’t see the problem with the different flow rates. You only need the time as an X ax, in terms of distance. Could you explain your point?

[tom jupille]

Victor, you already gave my reply (thanks! ). If the peak shape changes as a function of loading, then you are outside the linear range of the analysis. If the peak shape changes as a function of any other parameter, then it would be wise to exploit that to fix the shape "problem" (I will agree that "funny" peak shapes are suggestive of problems with the chromatography).

All that said, tailing (or peak shape generally) is not the only system suitability parameter. In much of the discussion, we have been postulating problems that could occur with non-ideal-but-symmetrical peaks. Most of these would be caught by one of the other parameters (resolution, linearity, etc.).

[HW Mueller]

danko, probably I don´t understand what you are doing, but your hight/width ratio minimum would depend on flow rate. X depends on flow rate, Y does not.

[Victor]

Tom,

"If the peak changes as a function of loading, then you are already outside the linear range of the analysis".

If you are measuring peak area, it is perfectly possible to have a linear relationship between the peak area and the concentration of the analyte even if the peak shape is changing? Were you referring to measurements using peak height? I think most people are using peak area. Maybe I misunderstood what you mean.

[tom jupille]

Victor,

Yes, I need to clarify what I meant. If you are within the linear range, only the amplitude of the peak will change, but the distribution will remain qualitatively the same (i.e., all the statistical moments, which we interpret as width, TF, tR, etc. remain constant). One of the hallmarks of overload is that σ (width) and skewness (tailing) increase and mean (retention time) decreases.

In some cases, fronting can increase and retention time increase on overload, which is what started this whole thread.

[Victor]

Tom-thanks for that. Can I ask you to explain your rationale for your statement that fronting peaks may be caused by inadequate buffering?
I agree with this, but I am not sure if you are making a general claim or describing a specific case.

[danko]

danko, probably I don´t understand what you are doing

Hi HW Mueller,

It may be because of my English you are in doubt – It’s a little bit rusty. I’ll try to explain how I calculate the test-results and it’ll make it plainer, I hope: In this sharpness test, I see the peak as a triangle with 3 equal sides. The ratio between the height and the base (or any other of the 3 sides) of such triangle equals 0.87 (h/c = 0.87).
I know from my experience with chromatography that this peak shape (equal sided triangle) is just acceptable – especially when you’re working with very small peaks (around the detection limit). So I say: If the above mentioned ratio results in the value of 0.88 or higher, then I am working just at the edge or better, as the value gets larger. Now as several of you gentlemen noticed I have two different units to deal with. That’s why I normalize the time to the signal as follows: 1 min ~ 1 mV/mAU.
So, no need to think about the flow rate (in “myâ€

[tom jupille]

Danko, thanks for the explanation; I think I see how your test works. If it is useful for you, by all means go ahead and use it. However, it is arbitrary (depends on the units chosen), and it tells nothing about peak shape. I think any value of the "sharpness test" is more as a surrogate for signal/noise ratio than as a test of peak shape.

Visualize the following situation: you are monitoring a separation with a photodiode array detector. There is only one peak, which has an absorbance maximum at 250 nm. You plot the chromatogram three times: at 280, 250, and 220 nm. It's the same peak each time, but you will get different peak heights and hence different "sharpness parameters".

[tom jupille]

Can I ask you to explain your rationale for your statement that fronting peaks may be caused by inadequate buffering?
I agree with this, but I am not sure if you are making a general claim or describing a specific case.

It's a general claim, but occurs infrequently and under specific circumstances.

First of all, the most common result of insufficient buffer is tailing of basic compounds (assuming silica-based packings and a low enough pH to suppress silanol ionization).

However, if "active-silanol" tailing is not an issue (non-silica column, or very pure, fully covered silica, etc.) and the analyte is basic, you can get fronting as illustrated in the sketch below (as above, I apologize for the poor artwork, but it's easier to sketch than to try and explain entirely in words).

The upper graph shows the k' vs. pH curve for a basic compound. Below its pKa the molecule is ionized and relatively hydrophilic. Above its pKa, it is unionized and relatively hydrophobic. Imagine that you are using a low-pH buffer, with a pH at point "A" on that curve. Now visualize what happens if you inject your basic sample at a higher concentration than the buffer can handle (lower drawings). The low-concentration parts of the peak (leading and trailing edge) is still within the effective range of the buffer, and "see" a pH value of "A". If the high-concentration part of the peak (apex) exceeds the buffer capacity, it will "see" a higher pH of "B", because the analyte will in effect act as a buffer and pull the pH toward its pKa.

Per the upper graph, the "A" parts of the peak will have a low k' (move fast) while the "B" parts of the peak will have a high k; (move slowly) so that the apex of the peak lags behind the leading and trailing edges, resulting in a "fronting" shape.

[HW Mueller]

Danko, so for a flowrate of, lets say 1 mL/min, you use the factor 0.88 and for 0.5 mL/min you use a factor of 1.8??

[HW Mueller]

Ups, that would be a factor of ~0.44 for 1/2 the flow rate?

[danko]

Hi HW Mueller,

As mentioned earlier I don’t use any volume units in my calculation. It’s pure time (in minutes) I use in the denominator. You see, if you convert the time and the signal/absorbance into distance, you can “buildâ€

[HW Mueller]

Sorry, can´t follow that.

[Victor]

Tom,

Thanks for your reply. I assume you are talking about RP chromatography and that you have assumed that the hydrophobic retention of the neutral base in the example that you have given is greater than the ionized base. I guess this would also hold for an acidic compound if it lowered the pH of the mobile phase and thus reduced its own ionization, dependent of course on it pKa and other variables.

[tom jupille]

Victor, yes on all counts.

The deeper point is that chemical tailing and fronting are closely related:

• - If the distribution coefficient (and hence k') of the analyte remains constant as concentration increases, you will get a symmetrical, approximately Gaussian peak.
  
  - If the distribution coefficient of the analyte decreases as concentration increases, you will get tailing (the apex moves faster than the leading edge).
  
  - If the distribution coefficient of the analyte increases as the concentration increases (as in my earlier post), then you will get fronting (the apex moves more slowly than the leading edge). In practice, this situation does not arise very often, so fronting is relatively uncommon.

Physical problems, such as partially-plugged frits, head spaces, mixing issues, temperature variations along the column, etc. by and large do not give predictable peak shape patterns; anything is possible, including tailing, fronting, flattened, split, shoulders (have I missed anything?).