The injection, the unknown and complex being

  • Fig. 1: Maximum injection volume (based on 10% Vcol,eff) as a function of the column inner diameter.Fig. 1: Maximum injection volume (based on 10% Vcol,eff) as a function of the column inner diameter.
  • Fig. 1: Maximum injection volume (based on 10% Vcol,eff) as a function of the column inner diameter.
  • Fig. 2: Comparison of an a) fixed-loop and b) flow-through autosampler.
  • Fig. 3: Illustration of the full-loop injection using a fixed-loop autosampler for a) sample loop loading, b) switching of the sample loop into the flow path for analysis and c) switching the sample loop out of the flow path.
  • Fig. 4: Illustration of the metered injection using a fixed-loop autosampler for a) sample loop loading, b) time-dependent switching of the sample loop into the flow path, c) switching the sample loop out of the flow path and start of the gradient.
  • Tab. 1: Comparison of the relative standard deviation (RSD) for the full-loop and metered injection using micro-LC-MS/MS for three different cytotoxic drugs at a concentration of 10 ng mL-1.

Miniaturization in chromatography also requires a corresponding reduction of the injection volume. Whereas several microliters can be injected on columns with an inner diameter (i.d.) between 2.1 and 4.6 mm, such high volumes can lead to a dramatic loss of the separation efficiency for micro- or nano-LC columns. The following article therefore presents the technical concepts that exist for miniaturized HPLC systems to keep the injection volume as low as needed.

The smaller the column inner diameter, the smaller the injection volume should be to minimize the band broadening caused by the injection plug. According to a “rule of thumb”, the maximum injection volume should not be higher than 10% of the effective column volume (Vcol, eff), which can be calculated from the following equation:

The effective column volume describes the volume fraction of the column which is actually occupied by the mobile phase. In addition to the column inner diameter (dc) and length (L), this volume depends on the porosity (ε) of the packing. Therefore, the effective column volume is always higher for fully porous particles compared to core shell particles of identical diameter, resulting in higher loadability. Figure 1 depicts the “maximum” injection volume as a function of the column i.d. For this calculation, a column length of 5 cm and a porosity of 70% were assumed.

The maximum injection volume thus decreases from 58 µL for a 4.6 mm i.d. column to 12 µL when a 2.1 mm i.d. column is used. A further reduction of the column i.d. to 300 µm, which is a typical i.d. for micro-LC, necessitates that only 250 nanoliters should be injected. The question now arises as to which technical concepts exist to transfer such a low volume reproducibly onto the column.

In general, two different autosampler types that allow for different injection modes are typically used. A distinction is made between the so-called fixed-loop and flow-through autosampler.

Both variants are shown in Figure 2.

Both autosampler types enable the so-called “partial-loop“ or “full-loop” injection”. Whereas for the partial-loop injection only a portion of the sample loop is filled, the sample loop is completely filled when full-loop injection is applied. Afterwards, the whole sample loop volume is transferred onto the column. This is demonstrated in Figure 3 for the fixed-loop autosampler. Independently of the autosampler type and injection mode, the sample loop usually remains in the flow path during the complete analysis and therefore contributes to the gradient delay volume. This applies to the fixed-loop as well as flow-through autosampler.

Typically, the sample loop of an analytical HPLC system equipped with a flow-through autosampler has a volume between 20 and 100 µL. However, such high volume sample loops are not suitable for micro-LC. For example, if a sample loop with a volume of 100 µL is used and the flow rate is 20 µL min-1, it would take 5 minutes until the gradient is completely flushed through the sample loop. To minimize the gradient delay time when using a flow-through autosampler, the sample loop can be switched out of the flow path after the sample has been transferred to the column. Nevertheless, a decisive advantage of the fixed-loop autosampler compared to the flow-through autosampler is its lower contribution to band broadening because the sample plug has not to be flushed through a needle-seat and needle seat capillary as is shown in Figure 2 b). Therefore, it is recommended to use a fixed-loop autosampler for micro-LC due to the higher flexibility in terms of the sample loop volume (≤ 10 µL). Thereby, the gradient delay volume can be reduced and a decreased contribution to band broadening can be achieved. However, if a volume of 250 nL should be injected using the full-loop technique, a sample loop of 50 µm diameter and 12.7 cm length has to be used. This is technically feasible but no flexibility with regard to higher injection volumes is given.

The rule of thumb for the “maximum” injection volume derives from the fact that polar compounds show only weak or no retention on conventional reversed phase stationary phases and are thus being transported through the column with the injection plug. In contrast, non-polar substances can be enriched or focussed on a reversed phase column if the injection plug contains no or only a small portion of organic solvents. Therefore, it is useful to increase the sample loop volume up to 5 or 10 µL for micro-LC systems to obtain higher flexibility for a large volume direct injection (LVI) and to avoid a frequent change of the sample loop. The higher the sample loop volume for a full-loop injection, the higher is the contribution to the gradient delay volume as was discussed above. However, to inject low sample volumes onto the column while keeping the gradient delay volume low, dedicated micro-LC systems enable the so-called metered injection (see Figure 4).

Here, the sample loop is completely filled as is the case for the full-loop injection but only a portion of the sample is transferred from the sample loop onto the column. Therefore, the sample loop is only switched into the flow path for a defined time, e.g. 1 s. After this transfer time, the sample loop is switched out of the flow path. Using this technique, very low volumes can be injected. The minimum transfer time for our micro-LC system (Eksigent ExpressLC ultra) is 150 ms. At a flow rate of 25 µL min-1, the resulting minimum injection volume is 62.5 nL which is clearly below the recommended injection volume for a micro-LC column with an i.d. of 300 µm. When the flow rate is increased to 50 µL min-1, the injection volume doubles to 125 nL since the minimum switching time is constant. The principle of the metered injection allows for a very flexible system design. The injection can always be adjusted according to the analytical requirements without changing the sample loop. When for example nonpolar compounds at very low concentrations should be analysed, a large volume direct injection of several microliters can be performed using the full-loop injection. Of course this is only possible if the solvent composition of the sample allows enrichment on the column head. On the other hand, the metered injection enables to reduce the injection to the nanoliter range when the compounds of interest are quantified with a very sensitive detector like e.g. a mass spectrometer. Consequently, a dilution of the sample prior to injection is not necessary because the signal intensity can be adjusted by the injection volume.

Now the question arises whether this technique is robust and can provide reproducible results. Table 1 shows the standard deviations for a ten-fold injection of three different cytotoxic drugs for the full-loop (Vinj = 8 µL) and metered injection (Vinj = 4 µL) using micro-LC-MS/MS. As can be seen, the relative standard deviations are independent of the applied injection technique. At this point it should be mentioned that these results are based on unsmoothed mass spectrometric data and a low analyte concentration and cannot be compared to UV detection.


From the theoretical point of view, the reduction of the injection volume is always striving for miniaturization. That this is possible with today’s commercially available systems was demonstrated. However, from the practical perspective, it is often necessary to choose an injection volume much higher as was specified by the rule of thumb (VInj,max equals 10% Vcol,eff) due to the requirements depending on the detection technique and limits of detection. The smaller the injection volume or amount of substance which is introduced into the system, the more sensitive the detection technique has to be. In the field of bioanalytics, the sample volume is often limited. Using a column with a smaller i.d. will always lead to an increased signal compared to an increased column i.d. at a constant injection volume. In the field of environmental analysis, the available sample volume is usually not a limiting factor. In addition, the required limits of detection are often very low (1 ng L-1 or less). For such applications, a large volume direct injection must be applied even when highly sensitive detectors are used.

Thorsten Teutenberg1, Terence Hetzel2, Juri Leonhardt1

1 Institute for Energy and Environmental Technology e. V. (IUTA), Duisburg, Germany
2 Bayer AG, Wuppertal, Germany

Dr. Thorsten Teutenberg
Department Head Research Analysis and
Institute for Energy and Environmental
Technology e. V. (IUTA)
Duisburg, Germany

Project Advanced Liquid Chromatograph

More on Chromatography


Microsite Advanced Liquid Chromatography

Register now!

The latest information directly via newsletter.

To prevent automated spam submissions leave this field empty.