Full-loop vs partial-loop injection (and how to size a sample loop)

Full-loop injection fills the sample loop completely with an excess of sample, typically two to five loop volumes, so the loop sets the injected volume. It gives the best precision (around 0.1% RSD) but wastes sample and is only as accurate as the loop's real volume. Partial-loop injection lets the syringe meter a smaller volume into the loop, no more than about half of it, so the volume is flexible and little sample is wasted, but precision and accuracy depend on the syringe. Size the loop so your largest partial-fill volume is at most 50% of it, or, for full-loop work, so that its calibrated volume is the volume your method needs.

Key points

• Laminar flow sets both rules: the centre of the stream moves at twice the average speed, so partial fills are linear only up to about half the loop and complete fills need two or more loop volumes.
• Loading between 50% and 200% of the loop volume is a problem zone where precision is often worse than the syringe itself.
• Nominal loop volumes are approximate: about 5% for 2 mL loops, 10% for 20 µL and 30% for 5 µL, according to IDEX. Calibrate the loop if you need the absolute volume.
• A syringe with a constant absolute error has a relative error that grows as the volume shrinks: ±0.1 µL is ±10% at 1 µL but ±0.1% at 100 µL.
• Backflushing the loop sends a partial-fill plug straight to the column instead of dragging it through the rest of the loop.

What happens inside the loop when you fill it?

When sample is pushed into a loop full of mobile phase, the boundary between the two takes a parabolic shape. IDEX describes it precisely: the velocity at the centre of the tube is twice the average and almost zero at the wall, so sample in the centre travels 2 µL along the loop for every 1 µL dispensed (IDEX injectors troubleshooting guide, Appendix F). Plot the amount injected against the volume dispensed and you get three regions:

Data from the IDEX guide and Rheodyne Technical Note 5. In Rheodyne's own test, a nominal 20 µL loop left the linear region at about 15 µL, roughly 60% of its actual volume. Their advice is to assume linearity only up to 50%.

Full-loop injection: precision, overfill and cost

How much to overfill. Published guidance agrees on the principle and differs on the multiple:

Small loops need proportionally more: three loop volumes for 5 µL against 1.6 for 50 µL in the Rheodyne data.

Precision. Once the loop holds pure sample, the syringe drops out of the equation and only the loop and the liquid matter. Rheodyne points out that the mass injected then follows the liquid's density, which changes by about 0.02% per °C for water and 0.1% per °C for most organic solvents, far more than the steel loop expands (TN5, footnote 3).

Accuracy. Loops are cut to length, not calibrated. VICI sells tubing with a typical bore tolerance of ±0.001 inch (VICI tubing chart). Volume scales with the square of the bore, so:

0.020 in bore ± 0.001 in  ->  (0.021/0.020)² = 1.10  ->  about ±10% in volume
0.010 in bore ± 0.001 in  ->  (0.011/0.010)² = 1.21  ->  about ±20% in volume

IDEX attributes its figures of about 10% for 20 µL loops and 30% for 5 µL loops to exactly this bore tolerance. If you need the true volume, calibrate the loop while it is attached to the valve, since the valve passages are part of it.

Sample consumption. A 20 µL loop filled with five loop volumes uses 100 µL per injection, plus whatever the autosampler needs to fill its needle and lines. Triplicate injections use 300 µL or more, a fifth of a well-filled 2 mL vial. A 20 µL partial fill from a 100 µL loop uses 20 µL plus the line fill.

Partial-loop injection: the 50% rule and syringe accuracy

In partial fill the syringe sets the volume, so everything depends on how well it delivers. Rheodyne measured this with a 25 µL syringe of 50 divisions, assuming the plunger can be set to 0.1 division (TN5, Table III):

The pattern is general: a constant absolute error becomes a larger relative error at smaller volumes. A motor-driven syringe removes the operator, but the same arithmetic applies to its volume resolution. Spark Holland makes the same point for autosamplers: partial loopfill performance relies directly on the accuracy and precision of the syringe or metering device (Spark Holland white paper).

Worked example: what ±0.1 µL means. The design target* for the Elutrolab AS-48 is ±0.1 µL volume accuracy after solvent-specific calibration (* marks values still to be confirmed by testing):

Solvent-specific calibration matters because the syringe displaces a volume, but what arrives in the loop depends on the liquid: its compressibility in the line between syringe and loop, its viscosity at the chosen draw speed, how it wets the needle tip and its density. Calibrating per solvent removes the systematic part of that error.

Check your own syringe. Spark Holland suggests injecting one standard at increasing volumes and looking at the offset of the line as well as r²: a high r² with a non-zero intercept points to a bias in aspiration. More tests of this kind are in autosampler qualification.

Full loop or partial loop: side by side

How to size a sample loop: four worked examples

Loop volume follows from its bore and length:

V_loop = π × (ID / 2)² × L
L = V_loop / (π × (ID / 2)²)

VICI's chart gives 0.51 µL/cm for 0.010 in bore, 2.03 µL/cm for 0.020 in and 4.56 µL/cm for 0.030 in (VICI).

1. Routine assay at a fixed 10 µL, plenty of sample, best precision. Use full loop with a 10 µL loop. Load at least 25 µL (95% of the maximum mass in Rheodyne's data), or 50 µL (five loop volumes) for about 0.1% RSD. In 0.010 in tubing the loop is 19.6 cm long, and with ±20% bore-related uncertainty you should calibrate it if you report absolute amounts.
2. Method development, 2-50 µL. Use partial fill. The largest volume must be at most half the loop, so choose at least 100 µL. In 0.020 in tubing that is 49 cm.
3. Small injections on a 2.1 mm column, 1-5 µL. Use a small loop. On the AS-48 the optional 25 µL loop covers 1-5 µL, 20% of the loop at most. A small loop adds less volume to the flow path while the valve is in INJECT, which matters for gradient delay on narrow-bore columns (see how an HPLC autosampler works).
4. Large-volume injection for trace work, 500 µL in a weak diluent. Partial fill needs a loop of at least 1 mL. IDEX recommends pumping at least ten sample volumes through the loop before returning to LOAD: 5 mL, or 5 min at 1 mL/min. Check first how much your column can take in how much can you inject.

Loop material and internal diameter

VICI offers loops in stainless steel, PEEK or titanium for HPLC and UHPLC, and FEP, PTFE or PEEK for low pressure, and notes that loop volume is controlled closely but not calibrated (VICI sample loops). Stainless steel is inert to most samples. PEEK is attacked by concentrated sulfuric and nitric acid, and methylene chloride, DMSO and THF swell PEEK tubing (IDEX fluidics FAQ).

The bore matters more than most people expect. For a fixed volume, the length scales with 1/ID², and the pressure drop while filling scales with length/ID⁴, so with 1/ID⁶ overall. A worked example with the Hagen-Poiseuille equation, water (0.89 mPa·s) and a fill rate of 50 µL/s:

Halving the bore raises the filling pressure 64-fold. A syringe pushing sample can cope with a few bar; a pull-to-fill system cannot exceed about 1 bar of suction and will form bubbles first. Wider bores fill easily but, by textbook Taylor-Aris theory, spread the sample plug more for the same volume. For syringe-filled loops of 100 µL and more, 0.5 mm (0.020 in) bore is a sensible middle ground.

Forward flush or backflush: which way should the loop empty?

Take VICI's standard 6-port injector numbering (VICI TN808). In LOAD, sample enters at port 1, passes to port 6, fills the loop from the port 6 end and leaves through port 3 to waste at port 2. In INJECT, the pump at port 4 connects to port 3, so mobile phase enters the loop at the port 3 end and pushes the sample back out of the port 6 end towards the column at port 5. The sample leaves through the end it came in: that is backflush. Swap the pump and column connections and the sample is pushed forward through the rest of the loop instead.

For partial fill, backflush is clearly better. A 10 µL plug in a 100 µL loop goes straight to the column instead of travelling through 90 µL of tubing where laminar flow would stretch it further. Spark Holland's measurements suggest most autosampler dispersion happens when the sample is flushed out of the loop, so the shorter that path, the better. For full-loop injections the direction matters little, because the whole loop is sample. Port layouts and valve internals are covered in the 6-port injection valve explained.