Where do all those extra peaks come from?
This is the final installment in the recent series of "LC Troubleshooting" discussions on gradient elution (1–6). In the last
two columns, we've considered problems related to gradients when we looked at dwell volume (5) and drift (6). This month,
we'll look at what many users consider the nemesis of gradient elution — ghost peaks. Also referred to as artifact peaks,
ghost peaks occur in gradients even when no sample is injected. That is, if you run the gradient program without cycling the
sample injection valve, peaks often appear in the chromatogram. We'll look at two examples of such problems that have been
resurrected from the archives of "LC Troubleshooting."
A Long Time Coming
I often refer to the "Rule of Three" as a guide to help us estimate what happens to retention time when we change the mobile-phase
concentration in reversed-phase liquid chromatography (LC). The rule of three states that the retention factor, k, changes by approximately threefold for each 10% change in the organic solvent concentration (%B, usually acetonitrile or
methanol). So, for example, if we had an analyte in our sample that normally was retained with k = 5 under isocratic conditions of 50% B, at 40% B, we would expect k ≈ 3 × 5 = 15. Another 10% change to 30% B would yield k ≈ 3 × 15 = 45. Similarly, 20% B would give k ≈ 135. We can convert this to retention time from the formula for k:
R is the retention time and t
0 is the column dead time (retention time of the solvent front). Rearrange equation 1 to solve for the retention time:
For discussion purposes, let's assume we're using a 150 mm × 4.6 mm column operated at 2 mL/min. This will generate t
0 ≈ 0.75 min. Now we can calculate retention times for each of our conditions. For k = 5, t
R = 4.5 min; for k = 15, t
R = 12 min; for k = 45, t
R ≈ 35 min; and for k = 135, t
R ≈ 102 min. So, you can see that by the time we reduce the mobile-phase concentration to 30% less than the starting concentration,
R = 102 min the peak is so strongly retained, it is unlikely that we'll wait around for it to come off the column.
We've been looking at isocratic retention, but the same process will occur during gradient elution. Because many gradients
start at very weak mobile-phase conditions (small %B), we would expect that many analytes would be very strongly retained
under such conditions. This is the basis of on-column compression, a process whereby the sample becomes concentrated or held
in a very narrow band at the inlet of the column when it is injected under very weak mobile phase conditions or is injected
in a very weak injection solvent. Sometimes it is possible to inject a large-volume sample — for example, 5–10 mL — in a very
weak injection solvent, such as water, and have the same chromatographic result as if we injected a much higher concentration
of the sample in a few microliters. This can be a practical way to preconcentrate water-based samples when gradient elution
So, we can see that on-column concentration can be a very useful analytical tool if we have very dilute samples. However,
the LC column is not very clever — it can't tell the difference between peaks that originated from an intentionally injected
sample and those that result from preconcentrating impurities from the starting, weak mobile phase. Peaks originating from
contaminants in the weak component of the mobile phase (the A-solvent, generally water or buffer) are the most common source
of ghost peaks in gradient elution.