Peak Tailing and Asymmetry: What a Skewed Peak Means
A symmetrical peak is an assumption built into most chromatographic arithmetic. When a peak departs from it, the departure is measurable, it has identifiable causes, and it changes how reliable an area-based purity figure is.
A skewed peak is one whose two halves are unequal. A tailing peak rises sharply and then falls away slowly. A fronting peak rises slowly and falls sharply. Both are measurable, both have identifiable causes, and both matter for purity, because a purity figure is calculated from area and a skewed peak makes its own area harder to define. This article sets out how asymmetry is quantified, what produces it, and where it enters the purity calculation.
The symmetry assumption
In an ideal separation, a compound travels through the column as a narrow band with a Gaussian profile. The detector records that band as a symmetrical peak. Many of the calculations applied to chromatograms assume this shape, including plate count and several of the conventions used to locate where a peak begins and ends.
Real peaks depart from the ideal. Some departure is tolerable and is simply recorded. Larger departure signals that something in the system, the sample or the column is behaving differently from the way the method assumes. The measurement of asymmetry exists to put a number on that departure so it can be judged against a limit.
Tailing factor and asymmetry factor
Two related quantities are in common use. They are often confused.
- USP tailing factor (T). Measured at 5% of peak height. It is the full peak width at that height, W0.05, divided by twice f, where f is the horizontal distance, measured at 5% of peak height, from the leading edge of the peak to the vertical line through the peak maximum. In short, T = W0.05 / 2f.
- Asymmetry factor (As). Usually measured at 10% of peak height. It is the back half-width divided by the front half-width, often written B/A.
For both, a value of 1.0 is a symmetrical peak. Values above 1.0 describe tailing. For both factors, values below 1.0 describe fronting. Because the two factors are taken at different heights, the same peak gives two different numbers. A peak with a long, low tail often scores a higher tailing factor at 5% than an asymmetry factor at 10%, since the tail is more visible closer to the baseline.
Acceptance limits vary by method. A tailing factor of 2.0 or lower is a widely used ceiling in pharmacopoeial system suitability, and many monographs set a symmetry range of roughly 0.8 to 1.8 unless the individual test states otherwise. An asymmetry factor between about 0.9 and 1.2 is generally regarded as good, with values up to roughly 1.5 often accepted. These are conventions, not universal laws. The limit that applies is the one in the validated method.
A worked example makes the tailing factor concrete. Suppose a peak is measured at 5% of its height. The leading edge sits 0.10 minutes before the vertical line through the peak maximum, so f is 0.10 minutes, and the full width at that height is 0.26 minutes. The tailing factor is 0.26 divided by twice 0.10, which is 1.3. A symmetrical peak measured the same way would give a full width of 0.20 minutes and a factor of 1.0. The arithmetic is simple. What it depends on is reading the width and the half-width at the correct height, which is why the measurement height is always stated.
Common causes of tailing
Tailing usually has a chemical cause, a physical cause, or both.
- Secondary interactions. Silica-based reversed-phase columns carry residual silanol groups that the bonded phase has not covered. Basic analytes can interact with these sites in addition to the intended hydrophobic retention. The second retention mechanism releases molecules more slowly and produces a tail. Many peptides contain basic residues, so they are exposed to this effect.
- Mass overload. Too much analyte saturates the available retention sites, and the peak shape changes.
- Extra-column dead volume. Excess tubing length, wide-bore tubing, or a poorly seated fitting lets the band spread outside the column. The effect is greatest on early, narrow peaks.
- A void at the column head. A settled or damaged bed disturbs the flow path and distorts the band as it enters the column.
- Sample solvent stronger than the mobile phase. The sample band is carried forward before it focuses at the head of the column, which distorts the shape.
- Detector settings. A data rate that is too low, or a long detector time constant, can sample a narrow peak too coarsely and make it appear to lean.
These causes can be told apart by pattern. If every peak in the run tails, the cause is more likely physical, such as a connection or a column void. If only basic compounds tail, the cause is more likely chemical, such as silanol interaction.
Common causes of fronting
Fronting is less common than tailing. Its usual causes include column overload, poor sample solubility in the injection solvent, a sample solvent that is mismatched with the mobile phase, and column temperature effects. Overload is the most frequent. A useful check is to dilute the sample and inject again. If the peak becomes more symmetrical at lower load, overload was contributing.
A peak can also be distorted by a collapsed or damaged column bed. In that case, dilution does not restore the shape.
How skew affects integration
Integration software must decide where a peak begins and where it ends. On a symmetrical peak, the decision is easy because the signal returns to baseline quickly. On a tailing peak, the signal returns slowly, and the point at which it has returned is a judgement. Two reasonable settings can give two different end points, and therefore two different areas.
The baseline choice adds a second variable. A drop to the baseline at the peak end and a valley-to-valley construction assign area differently, especially when a small peak sits on the tail of a large one. The same raw data can yield different areas depending on which construction is applied.
Integration parameters such as slope sensitivity and minimum area also interact with shape. A long, shallow tail may fall below the slope threshold before the signal has truly returned to baseline, so the software ends the peak early and leaves part of the tail uncounted.
Effect on an area-based purity figure
Purity by area normalisation is the main-peak area divided by the total peak area. Any change to how the main peak is bounded changes the figure. Skew produces two opposite errors.
- The tail absorbs an adjacent impurity. If an impurity elutes on the back edge of the main peak, a long tail can carry the integration past it. Its area is then counted as part of the main peak. The reported purity rises, and the impurity is not reported separately.
- The tail is split off. If the integration ends the main peak early, or a valley is drawn into the tail, part of the main compound is counted as a separate small peak. The reported purity falls, and an apparent impurity appears that is really the same compound.
Neither error is visible in the percentage alone. Both are visible in the chromatogram. This is why a purity value should be read with the trace behind it, and why the shape of the main peak is part of the evidence. The relationship between area and purity is covered in more detail in the article on peak area and purity. Where an impurity sits directly under the main peak, no integration choice resolves it, a problem addressed in the article on peak purity and coelution.
Effect on plate count and resolution
Theoretical plate count is a measure of column efficiency. The common formulae calculate it from retention time and peak width, and they assume a Gaussian peak. A tailing peak is wider than a symmetrical peak of the same retention, so the calculated plate count falls. Part of that fall reflects real band broadening. Part reflects the formula applying a Gaussian model to a shape that is not Gaussian.
The practical consequence is for resolution. A tail extends into the space between peaks. Two peaks that would be separated if symmetrical may overlap once one of them tails, which makes accurate area assignment harder for the smaller peak. A tailing factor within the method limit does not guarantee that resolution is adequate, and adequate resolution does not guarantee that the tailing factor is acceptable. Both are checked.
Limits of the measurement
A tailing or asymmetry value describes the shape of one peak in one run. It does not describe the sample as a whole. It says nothing about impurities that do not absorb at the detection wavelength or that elute inside the main peak, and it cannot confirm identity. It indicates whether the conditions of the measurement were behaving as the method expects, which allows the associated purity figure to be read with appropriate caution.
Novagen is a third-party analytical laboratory. A certificate reports what was measured in the sample submitted and nothing beyond it. Purity describes composition, not safety. Quoted turnaround is 21 days from receipt of samples. Certificates can be checked against the public register at novagenlaboratories.com/verify, and questions can be sent to admin@novagenlaboratories.com. For related reading, see why purity is measured by peak area and peak purity and coelution.
Common questions
- What is a good tailing factor in HPLC?
A value of 1.0 describes a perfectly symmetrical peak. In practice, a tailing factor up to about 2.0 is a common upper limit in pharmacopoeial system suitability, and many methods aim for 1.5 or lower. The acceptable value depends on the method, the analyte and the resolution required from neighbouring peaks, so the limit stated in the validated method takes precedence.
- What causes peak tailing in HPLC?
The most common causes are secondary interactions between basic analytes and residual silanol groups on silica-based columns, column overload, extra-column dead volume from poor connections or excess tubing, a void at the head of the column, a sample solvent stronger than the mobile phase, and detector data-rate settings that are too slow. Peptides with basic residues or a free N-terminus are particularly exposed to silanol effects.
- What is the difference between tailing factor and asymmetry factor?
Both compare the back of a peak with the front, but they are measured at different heights. The USP tailing factor uses the peak width at 5% of peak height divided by twice the front half-width (f) at that height. The asymmetry factor is usually the back half-width divided by the front half-width at 10% of peak height. For the same peak, the two numbers are not identical, so they should not be interchanged.
- Does peak tailing affect purity results?
It can. Purity by area normalisation depends on where each peak starts and ends. A long tail makes the end of the peak ambiguous, and the integration choice can either absorb a closely eluting impurity into the main peak or split it off. The first raises apparent purity. The second lowers it. The size of the effect depends on how close the impurity elutes and how the baseline is drawn.
- How do you fix a tailing peak?
The fix depends on the cause. Common steps include reducing the injected mass or volume, dissolving the sample in a solvent no stronger than the starting mobile phase, shortening and properly seating connections, adjusting mobile phase pH or ion-pairing conditions, using a column with fewer exposed silanols, and replacing a column that has developed a void. Changing one variable at a time shows which cause was responsible.