RRT: relative retention time

[ym3142]

I did try to search the archives but found it is not easy.

Recently, I try to copy a HPLC testing performed in oversee lab. I found all peaks were shifting, which was very common(about 1 min). We are using the identical columns and all the same related conditions: C18, 250x4.6x5, 1ml/min, 0.1TFA in water (A) and in ACN(B), gradient, column 35 C, etc.

We also found some peaks had very similar RRT such 0.84 to 0.85; but other could have 0.14 difference: meaning 1.45 vs 1.59.

Does any one summarize, in theory, when RRT should be comparable and when not? How close is considered to be "similar' or " comparable"?

Thanks
Jim

[HPLCaddict]

What you're observing is most probably the effect of different dwell volumes (aka gradient delay volumes) on the HPLCs involved. This will lead to retention times shifting and, occasionally, also shifts in selectivity. So it's absolutely possible that relative retention times are also different between different HPLCs when using gradient elution. As long as the elution profile is unchanged (i.e. peak order is the same and peak spacing comparable) and "critical" parameters are OK (e.g. the resolution of a critical peak pair) I wouldn't mind. That's why a decent system suitability test is important. Just include "your" RRTs in your SOP.

[James_Ball]

I would agree, no two HPLC systems or columns are exactly the same. I usually see small differences even on the same system with different columns. If running a gradient it can also be how the gradient is mixed. With Agilent Quat pumps a tiny difference in the solenoids can cause small percent differences in the actual gradient. You can either vary the conditions of one system until it matches the other, or note the differences as mentioned above and show you still have acceptable performance. Making note of the differences is a much easier route to take to be honest.

[danko]

Depending on which peak is used for reference when calculate the RRT (I’ve seen some people use the dead time which isn’t correct) the dwell volume should not affect the RRT significantly. And not at all the selectivity.
A different column on the other hand – even if it’s the same brand and sometimes the same lot – could cause RTT change and even small selectivity differences.
Finally a brand new column could separate differently than a column that has been use for some time.

Best Regards

[HPLCaddict]

Depending on which peak is used for reference when calculate the RRT (I’ve seen some people use the dead time which isn’t correct) the dwell volume should not affect the RRT significantly. And not at all the selectivity.
A different column on the other hand – even if it’s the same brand and sometimes the same lot – could cause RTT change and even small selectivity differences.
Finally a brand new column could separate differently than a column that has been use for some time.

Best Regards

Depending on the shape of the gradient and, as you already wrote, the reference peak, the dwell time CAN have an influence on RRT. As an artificial example, imagine a gradient with an isocratic step at the beginning and later on a linear increase in %B. If the reference peak elutes during the isocratic step and the peak in question during the linear gradient (or vice versa), RRT will heavily depend upon the dwell volume. This is quite artificial, I admit. On the other hand, suppose a more complex gradient with multiple isocratic/linear segments - here the dwell volume will have an even higher influence. But, you're correct, for the general linear gradient used in probably >80% of gradient methods, dwell volume differences will have little to no influence on RRT.
Concerning the selectivity, my argumentation probably was a bit unclear. With "selectivity" in THIS case I refer to the peak spacing - which if RRTs differ a little will also differ slightly.
Generally, you're right. Stationary phase variability might be the underlying problem here. Could be interesting to see the validation data of this procedure. Hopefully, they've done robustness studies...

[danko]

Agree