Extra-column dispersion is the peak broadening that the injector, connecting tubing, fittings and detector add outside the column; their variances, not their widths, add to the column variance.1,2 Small, early-eluting peaks from short, narrow, small-particle columns lose the most efficiency.3,2 Measure it with a zero-dead-volume union, then change contributors one at a time.1,4
AI Method Builder
Draft a starting method for your separation and refine it — save your work to your account.
What is extra-column dispersion and how do the variances add?
Extra-column dispersion covers every band-broadening effect other than the column: injection volume, detector characteristics, time constants and plumbing.2 The variances of system and column peaks are additive; the widths are not.1,2
σtotal2 = σcol2 + σec2, with σec2 = σinj2 + σtube2 + σfit2 + σdet2 + σtc2
Here σtotal2 is the observed peak variance, σcol2 the variance from the column alone and σec2 the extra-column variance, the sum of the injection, tubing, fitting, detector-cell and time-constant terms.5,2 The sum assumes each element disperses independently, but it is usually accurate enough to guide method development.5 All terms must be in the same domain, time or volume, which are related through the flow rate.6 The same addition gives the observed plate number:4
Nobs = Ncol × σcol2 / (σcol2 + σec2)
Ncol is the intrinsic, column-only plate number and Nobs the plate number seen at the detector.4 The plate number is calculated from retention time and peak width at half-height.7 See the theoretical plate number calculation for the formula.
Why do early-eluting peaks on narrow columns lose the most efficiency?
Shorter, narrower columns packed with smaller particles are more sensitive to extra-column dispersion than larger columns, because they generate narrower peaks.3,2 Peaks with longer retention times are broader and less influenced by tubing effects.2 In a study of high-speed narrow-bore columns, widening the post-column tubing usually affected only the first-eluting peak in isocratic mode or shallow gradients.4
Stoll and Broeckhoven modeled instrument variances of 30 µL² (a good 400-bar-era instrument), 10 µL² (an off-the-shelf ultrahigh-pressure liquid chromatography, UHPLC, configuration) and 2 µL² (a highly optimized UHPLC configuration).1 A 150 mm × 4.6 mm, 5 µm column lost at most about 5% efficiency at 30 µL² for any compound with k > 1.1 A 150 mm × 2.1 mm column of 2.7 µm superficially porous particles lost more than 30% on the same instrument for compounds with k < 5.1 With a 30 mm × 2.1 mm sub-2 µm column, even an optimized system reached 80% of column efficiency only for k > 4.1
USP General Chapter ⟨621⟩ calls for caution when an adjustment gives smaller peak volumes through smaller particles or a smaller internal diameter, citing instrument connections, detector cell volume and sampling.7 See scaling an HPLC method to new column dimensions. The wider ⟨621⟩ rules for column changes are covered in USP column substitution and revalidation.
Which parts of the HPLC flow path add extra-column variance?
Figure 1 maps each contributor along the flow path.
Table 1 gives each contributor’s mechanism and reported remedy.
| Contributor | Mechanism | Reported remedy |
|---|---|---|
| Injector | A flow-through injector adds a needle seat capillary between sample and column; observed injection variances are usually much larger than the rectangular-plug prediction.3 | A small fixed loop can lower injection variance; a shorter, narrower needle seat capillary reduces it.3 |
| Connecting tubing | Parabolic laminar flow; dispersion rises with capillary length and internal diameter.5 | Shorter, narrower capillaries.5 |
| Fittings and dead volume | Poorly swept volumes at unions and connectors act as mixing chambers that broaden peaks and cause tailing.6 | A timed pinch injection mode could be used to reduce the tailing-type contribution in that study.6 |
| Detector flow cell | A longer or broader flow path gives more dispersion; longer cells give higher sensitivity.5 | A smaller cell gives narrower peaks at the cost of lower peak height and higher noise.5 |
| Time constant and data rate | A large time constant averages too many points and can decrease peak height.5,8 | Working rules: time constant ≈1/10 of the narrowest peak width; ≈20 points across the peak.8 |
In the Taylor–Aris model, tubing variance is linear in length and has a fourth-power dependence on tube radius.6 Narrow bores cost pressure: changing from a 120 µm to a 75 µm internal diameter capillary raises the pressure drop 6.5-fold.5 The time constant exists to reduce baseline noise and degrades the signal only when it is large relative to peak width.8 As a working rule, up to about 15% of the peak volume can be injected without a noticeable loss of resolution.2 Injection estimates assume a sample solvent matched to the mobile phase; a very different solvent can make the injection contribution larger or smaller; see how diluent strength and injection volume distort peaks.3
How much plate number does extra-column variance cost in a worked example?
Take a 150 mm × 2.1 mm column of 2.7 µm superficially porous particles with the reduced plate height of 1.5 and total porosity of 0.5 that Stoll and Broeckhoven assumed.1
H = h × dp = 1.5 × 2.7 µm = 4.05 µm; Ncol = L/H = 150,000 µm / 4.05 µm = 37,037
The plate height H is column length L divided by plate number, and the reduced plate height h is H divided by particle diameter dp.7
VM = εT × π × (dc/2)2 × L = 0.5 × π × (0.105 cm)2 × 15 cm = 259.8 µL; VR = VM(1 + k); σcol2 = VR2/Ncol
VM is the hold-up volume, εT the total porosity, dc the column internal diameter and VR the retention volume at retention factor k; this follows Stoll and Broeckhoven’s isocratic approach.1 For k = 1, VR = 519.5 µL and σcol2 = 7.29 µL²; for k = 5, VR = 1,558.6 µL and σcol2 = 65.59 µL². Table 2 applies the Nobs equation at the three instrument variances.
| σec2 (µL²) | Nobs, k = 1 | Loss, k = 1 | Nobs, k = 5 | Loss, k = 5 |
|---|---|---|---|---|
| 2 | 29,062 | 21.5% | 35,941 | 3.0% |
| 10 | 15,613 | 57.8% | 32,137 | 13.2% |
| 30 | 7,239 | 80.5% | 25,413 | 31.4% |
Figure 2 plots the model from 0 to 30 µL². At 30 µL² the k = 1 and k = 5 peaks keep 19.5% and 68.6% of their plates, the pattern reported for this column format.1 At 2 µL² they keep 78.5% and 97.0%.
One design target in the literature is extra-column variance of 10–20% or less of the column variance for early-eluting compounds on narrow-bore, small-particle columns.4 No source gives a universal acceptance limit. For validated or compendial methods, the applicable procedure and regulatory framework take precedence over the general rules of thumb given here.
How do you measure system dispersion with a zero-dead-volume union?
Replace the column with a zero-dead-volume union and inject a sample; the width of this system peak estimates the total extra-column dispersion.1 Keep mobile phase composition, flow rate, temperature and injection volume identical, then subtract the system variance from the variance measured with the column.6
Union peaks are difficult to integrate, which leads to significant underestimation of the system variance.4 Because extra-column peaks tail, half-width methods do not describe their total variance, as Grinias et al. showed for capillary LC.6 For an asymmetric peak, the width at 50% height underestimates variance, the width at 4.4% height is more conservative and statistical moments are more rigorous.1 Plotting observed plate numbers against k for a homologous series is an alternative, but it is also prone to errors.4
How do you reduce extra-column band broadening step by step?
Table 3 sets out the order of work, adapted from the one-at-a-time approach of Vanderlinden et al.4
| Step | Action | What it shows |
|---|---|---|
| 1. Measure system variance | Zero-dead-volume union under identical conditions; integrate by moments or 4.4% width.6,1 | The total σec2 to compare with σcol2.1 |
| 2. Find the largest contributor | Separate pre-column (injector, inlet tubing) from post-column (outlet tubing, cell, electronics) terms.4 | Which term in Table 1 dominates. |
| 3. Change one part | Shorter or narrower capillary, smaller loop, smaller cell or shorter time constant; hold the others constant.4,5 | The effect of that part alone.4 |
| 4. Re-measure | Repeat the union test and the column run; return to step 1. | The new Nobs; compendial methods must still meet system suitability.7 |
Figure 3 draws the same four steps.
Vanderlinden et al. found that pre-column broadening can be avoided by focusing analytes at the column head, but post-column broadening is inescapable in isocratic and gradient elution alike.4 Outlet tubing and the detector therefore deserve attention first in gradient work.2 If early peaks also tail, follow the diagnostic path for tailing HPLC peaks. The guide to poor HPLC peak shapes sorts other symptoms.
Frequently asked questions
Does gradient elution protect a method from extra-column dispersion?
Only partly: post-column broadening in the outlet tubing and detector cannot be avoided.4 Pre-column broadening can be avoided by focusing analytes at the column head.4 In a study of narrow-bore columns packed with 1.3 µm particles, the steeper the gradient, the more pronounced the extra-column losses; widening the post-column tubing from 65 to 250 µm increased relative peak width by a factor of about 1.9 for all peaks under steep gradients.4
Is narrower connecting tubing always better?
Not without limits. Dolan recommends 0.007-in. tubing as a working rule because 0.005-in. tubing is more prone to blockage; his calculations showed little resolution loss from 25 cm of 0.007-in. tubing with 150 mm columns of 4.6 or 2.1 mm internal diameter.2 With a 150 mm × 1.0 mm column, the same tubing cost about 25% of the resolution.2
Do 4.6 mm columns need a low-dispersion instrument?
Usually not for retained peaks: the 150 mm × 4.6 mm, 5 µm column in Stoll and Broeckhoven’s model lost at most about 5% for k > 1, even at 30 µL².1 A lower-dispersion instrument helped this column mainly for compounds with k < 1, which matters in size-exclusion chromatography.1 Dolan found that even 100 cm of 0.010-in. tubing only degraded resolution from 2 to about 1.9 for a 150 mm × 4.6 mm column of 3 µm particles.2
Does a mass spectrometer add extra-column dispersion?
Far fewer studies cover mass spectrometric than UV detectors; in those, the ionization source and the mass spectrometer itself had little impact on peak widths when settings were optimized, and the tubing connecting the LC to the mass spectrometer was the most critical contributor.5 For the 50 mm × 2.1 mm LC-MS format, Dolan calculated that 1 m of 0.005-in. tubing reduces resolution by about 20%, so tubing length should be kept to a minimum.2
Can the plumbing be changed on a compendial HPLC method?
USP ⟨621⟩ requires compliance with the system suitability criteria throughout the chromatographic procedure, so any instrument change still has to meet the method’s system suitability.7 The chapter also cautions about broadening from connections, detector cell volume and sampling when an adjustment shrinks peak volumes.7 For validated or compendial methods, the applicable procedure and regulatory framework take precedence over the general rules of thumb given here.
The takeaway
Extra-column dispersion is a variance budget: the injector, tubing, fittings, detector cell and electronics each add variance to the column’s own, and the sum sets the observed plate number.1,5 Early peaks from short, narrow, small-particle columns suffer most.1 Measure the system with a zero-dead-volume union, integrate by moments rather than half-height width, change one contributor at a time and re-measure.1,6,4
References
- Stoll DR, Broeckhoven K. Where has my efficiency gone? Impacts of extracolumn peak broadening on performance, part I: basic concepts. LCGC North America. 2021;39(4):159–166. https://www.chromatographyonline.com/view/where-has-my-efficiency-gone-impacts-of-extracolumn-peak-broadening-on-performance-part-i-basic-concepts — variances (not widths) add; zero-dead-volume union test; efficiency loss versus retention factor, column format and instrument variance (30, 10 and 2 µL²); porosity and reduced-plate-height assumptions for the worked example.
- Dolan JW. Extracolumn effects. LCGC North America. 2005;23(2). https://www.chromatographyonline.com/view/extracolumn-effects-2 — list of extra-column contributors; variances add; narrow-peak columns more sensitive; later peaks less affected; post-column tubing matters in gradients; injection-volume heuristic.
- Stoll DR, Broeckhoven K. Where has my efficiency gone? Impacts of extracolumn peak broadening on performance, part 2: sample injection. LCGC Europe. 2021;34(6). https://www.chromatographyonline.com/view/where-has-my-efficiency-gone-impacts-of-extracolumn-peak-broadening-on-performance-part-2-sample-injection — injector contribution (fixed-loop versus flow-through, needle seat capillary); small columns more sensitive.
- Vanderlinden K, Broeckhoven K, Vanderheyden Y, Desmet G. Effect of pre- and post-column band broadening on the performance of high-speed chromatography columns under isocratic and gradient conditions. J Chromatogr A. 2016;1442:73–82. https://doi.org/10.1016/j.chroma.2016.03.016 — Nobs from Ncol and the variance ratio; union-test underestimation; first-eluting peaks most affected isocratically; post-column broadening unavoidable; varying contributors one at a time.
- Stoll DR, Broeckhoven K. Where has my efficiency gone? Impacts of extracolumn peak broadening on performance, part III: tubing and detectors. LCGC North America. 2021;39(6):252–257. https://www.chromatographyonline.com/view/where-has-my-efficiency-gone-impacts-of-extracolumn-peak-broadening-on-performance-part-iii-tubing-and-detectors — component variances sum (independence caveat); tubing length and internal diameter; flow-cell volume versus sensitivity trade-off; detector time constant and sampling frequency.
- Grinias JP, Bunner B, Gilar M, Jorgenson JW. Measurement and modeling of extra-column effects due to injection and connections in capillary liquid chromatography. Chromatography. 2015;2(4):669–690. https://doi.org/10.3390/chromatography2040669 — poorly swept volumes at unions/connectors cause broadening and tailing; zero-dead-volume connector method; second-moment variance needed for tailing system peaks; fourth-power radius dependence of tubing variance (capillary LC study).
- United States Pharmacopeial Convention. USP General Chapter ⟨621⟩ Chromatography (Stage 4 harmonization, official 1 December 2022). https://www.usp.org/sites/default/files/usp/document/harmonization/gen-chapter/harmonization-november-2021-m99380.pdf (accessed 2026-10-19) — caution on extra-column band broadening when adjustments reduce peak volume; system suitability required throughout; plate number definition.
- Dolan JW. Enhancing signal-to-noise. LCGC North America. 2010;28(3). https://www.chromatographyonline.com/view/enhancing-signal-noise — detector time constant and data rate heuristics (≈1/10 peak width; ≈20 points across the peak).
Further reading
- Snyder LR, Kirkland JJ, Dolan JW. Introduction to Modern Liquid Chromatography. 3rd ed. Wiley.
- Neue UD. HPLC Columns: Theory, Technology, and Practice. Wiley-VCH.
Reviewed against cited sources. The variance-addition model and its independence caveat follow Stoll and Broeckhoven (parts I and III), Dolan (2005) and Vanderlinden et al. (2016); the column-format comparisons, instrument variances and worked-example assumptions follow Stoll and Broeckhoven part I; injector, tubing and detector mechanisms follow Stoll and Broeckhoven parts 2 and III; fittings and union-test integration follow Grinias et al. (capillary LC scope); time-constant and data-rate working rules follow Dolan (2010); compendial statements follow USP ⟨621⟩. The worked-example numbers are LabVeda calculations from the stated model and assumptions. For validated or compendial methods, the applicable procedure and regulatory framework take precedence over the general rules of thumb given here. Evidence review: October 2026.