HPLC injection volume: how much can you inject on your column?

For a sample dissolved in mobile phase and an isocratic method, a safe starting point is about 1-2% of the column's void volume, or 15% of the volume of the first peak you care about: roughly 17-34 µL on a 4.6 × 150 mm column and 1-2.5 µL on a 2.1 × 50 mm column. Gradient methods tolerate more, because analytes injected in a diluent weaker than the starting mobile phase focus at the column inlet. A diluent stronger than the mobile phase cuts the limit sharply, so check the diluent before raising the volume, and confirm any choice by injecting half and double the volume.

Key points

• Column volume is π × (d/2)² × L; for fully porous particles the void volume is about 0.68 of that.
• Isocratic, sample in mobile phase: inject 1-2% of the void volume, or up to 15% of the first peak's volume (Dolan's 15% rule).
• Gradient: Waters' general guidance is 1-10% of the column volume, closer to 1-3% where peak shape matters. The upper end needs a diluent weaker than the starting mobile phase.
• A strong diluent makes early peaks front or split; a weak diluent focuses them at the column inlet.
• The first peak of interest sets the limit. Later peaks are wider and always tolerate more.

How do you calculate column volume and void volume?

V_column = π × (d / 2)² × L        d and L in mm gives µL (1 mm³ = 1 µL)
V_M      ≈ 0.68 × V_column         void volume, fully porous particles
t0       = V_M / F                 F = flow rate

Worked example for a 4.6 × 150 mm column:

V_column = 3.1416 × 2.3² × 150 = 2493 µL
V_M      ≈ 0.68 × 2493       = 1695 µL
t0       = 1.695 mL / 1.0 mL/min ≈ 1.7 min

The factor 0.68 is a rule of thumb for the total porosity of fully porous packings; Merck uses 0.7 (Merck), and superficially porous particles have less pore volume, so their factor is lower. If you can measure t0 with an unretained marker, t0 × F is the best void volume you will get.

Suggested maximum injection volumes for common columns

The table applies three published rules of thumb:

• Isocratic, sample in mobile phase: 1-2% of the void volume. Merck states the rule as 1-2% of column volume, but its worked numbers (1.2-2.4 µL for a 50 × 2.1 mm column with about 120 µL void volume) apply it to the void volume, so that is what the table uses.
• Gradient, diluent no stronger than the starting mobile phase: 1-3% of the column volume. Waters gives 1-10% of total column volume as the general range and 1-3% to avoid volume overload and strong-solvent effects (Waters WKB48961).
• Gradient ceiling: 10% of the column volume. The top of Waters' range. Only realistic with a diluent clearly weaker than the starting mobile phase, and only after testing.

Restek's guideline values for samples in mobile phase sit in the same range: 1-3 µL for 2.1 mm columns of 30-100 mm, 2-12 µL for 3.0-3.2 mm columns of 50-150 mm and 8-40 µL for 4.6 mm columns of 50-250 mm (Restek FAQ). Note how small the numbers get for 2.1 mm columns: on a 2.1 × 50 mm column, 1-5 µL covers both the isocratic and the gradient starting range, and the loop and syringe have to deliver those volumes precisely (see full-loop vs partial-loop injection).

The 15% rule: a better estimate for isocratic methods

Column-volume rules ignore retention and efficiency. John Dolan's approach does not: the injection may be up to about 15% of the volume of the first peak of interest for a 1% loss in resolution, 30% for a 5% loss and 40% for a 10% loss, assuming the sample is in mobile phase (Dolan, Part I).

V_R  = V_M × (1 + k)          retention volume
V_p0 = 4 × V_R / √N           baseline peak volume for a small injection
V_max ≈ 0.15 × V_p0           1% loss of resolution

Worked example from column data. A 4.6 × 150 mm column with an assumed 12,000 plates and a first peak at k = 1:

V_R   = 1695 × 2           = 3390 µL
V_p0  = 4 × 3390 / √12000  = 124 µL
V_max = 0.15 × 124         ≈ 19 µL

Worked example from your chromatogram. No k or N needed: measure the baseline width of the first peak of interest from a small injection and multiply by the flow rate. A width of 0.08 min at 1.5 mL/min gives V_p0 = 120 µL and a 15% limit of 18 µL.

Dolan's follow-up shows why the first peak decides. On a 100 × 4.6 mm, 3 µm column with about 10,000 plates, the k = 1 peak (80 µL wide) tolerates 12 µL and the k = 10 peak (440 µL) tolerates 66 µL, but injecting 66 µL broadens the k = 1 peak by 38% (Dolan, Part II).

Volume overload or mass overload?

Volume overload is too much liquid. With the sample in mobile phase, Dolan's simulations show the front edge of the peak behaving normally while the peak becomes flat-topped and its back edge moves later, raising the valley to the next peak. Merck describes what you typically see in practice: peaks that front more (symmetry factor below 1) and retention that may decrease, with falling efficiency and resolution.

Mass overload is too much analyte for the stationary phase, whatever the volume. The textbook signature for the common case is a right-triangle peak, steep at the front with a long tail, whose retention time moves earlier as the amount rises. With some analytes the triangle points the other way and the peak fronts. Detector overload clips the peak top and breaks the area-volume linearity even when the separation is fine.

Separate them with one change at a time:

Injection solvent strength: fronting, splitting and focusing

Stronger diluent. While analyte molecules are inside the injected plug, the plug is their mobile phase. If it holds more organic solvent than the eluent, the front of the band moves ahead until the plug is diluted, while the back of the band moves at the normal rate (Dolan, Part II). Dolan's rule of three puts numbers on it: for a typical 400 Da molecule, 10% more organic in the plug makes the analyte move about three times faster, and 20% more about ten times faster.

Waters shows the effect in a gradient (Waters primer). Six analytes in 50/50 acetonitrile/water on a 2.1 × 50 mm column, with a gradient starting at 5% acetonitrile: 2 µL (1.1% of the column volume) gave good peaks; 4 µL broadened the first two peaks with pronounced fronting (USP tailing 0.71 and 0.72). The same 4 µL with the sample in 10/90 acetonitrile/water gave tailing factors of 1.07 and 1.29 and the expected doubling of peak height.

Split peaks. When a large, strong plug carries part of an early analyte ahead before it mixes, the rest stays behind and the peak shows a shoulder or splits in two. A void at the column inlet splits every peak whatever you inject; a diluent effect hits mainly early peaks and disappears when you inject the sample in mobile phase. IDEX suggests an extra blank injection of the sample solvent itself to show whether artefacts come from the solvent difference (IDEX injectors troubleshooting guide).

Weaker diluent. This is the useful direction. With the diluent 10% or 20% weaker than the mobile phase, the analyte slows to about a third or a tenth of its normal speed during loading, which compresses the band ("on-column concentration"). A 100% aqueous sample injected into 50% organic or more effectively stops at the column inlet, which is why Dolan suggests that 10 mL or more of river water might be injected for trace work. As a rough worked example, 100 µL loaded in a diluent 20% weaker than the eluent forms a band about as long as a 10 µL injection in mobile phase.

Practical fixes, in order of preference:

1. Dilute the sample with water or the weak solvent instead of injecting less. This is Waters' fix above: more volume, same mass, better peaks.
2. In gradients, match the diluent to the starting mobile phase or make it weaker.
3. If solubility forces a strong solvent, keep the volume small: Dolan reports a 1 µL injection in toluene working in reversed phase because it dispersed before it could migrate.

On-column focusing in gradient methods

Gradients usually start weak, often 5-10% B. Dolan notes this restricts how much organic solvent the diluent may contain, but also that a weak starting mobile phase dilutes a strong plug more effectively than a 50-60% B isocratic eluent. Merck sums up the consequence: isocratic runs are much more prone to volume overload than gradient methods.

The mechanism is retention at the starting composition. Analytes that elute late in the gradient have very large k at 5% B, so they stop at the column inlet however much liquid carries them there, and the gradient releases them as sharp bands. Early eluters have low k at the start and behave almost isocratically, which is why they are the first to front or broaden. Until the gradient arrives, the column also runs isocratically at the starting composition for the dwell volume divided by the flow rate; a large loop left in the flow path adds to that time (see how an HPLC autosampler works).

How do you find the maximum injection volume for your method?

1. Pick a starting volume from the table or the 15% rule and round it (12 µL becomes 10 or 15 µL).
2. Inject it, half of it and twice it, from the same vial.
3. Compare the resolution of the critical pair, the tailing factor of the first peaks and the peak area per microlitre.
4. Adjust. If the half injection is clearly better, step down and repeat. If the double injection still looks good, try more.
5. Leave a margin. Dolan's example: if 20 µL is fine and 40 µL is not, try 30 µL. If 25 µL already shows problems, 20 µL is too close to the edge (Dolan, Part II).
6. Check linearity. Peak area should rise in proportion to volume. Merck suggests starting at the smallest volume your injector delivers reproducibly and doubling until you reach 3% of the column volume.

If area per microlitre drifts or peaks distort below the expected limit, the injector may be part of the problem; see autosampler troubleshooting, and for formal linearity and precision tests autosampler qualification.