More Tools for Sample Injection When Faced with Mobile Phase/Sample Solvent Mismatch

Dealing with the so-called sample/mobile phase solvent mismatch problem has become a bit of a hobby of mine – some might even call it an obsession. For me, this all started in the early days of my work on two-dimensional liquid chromatography (2D-LC) when I realized that the influence of the first dimension mobile phase on the performance of the second-dimension separation strongly determines the overall performance of the 2D method. In this context, the first dimension mobile phase solvent gets injected into the second dimension column and thus becomes the sample solvent. At that time, one of the tricks we used was to use a highly retentive stationary phase in the second dimension separation. This is generally a good idea, but the approach has limitations, particularly when the selectivity of that stationary phase is insufficient for the separation goals at hand. We need other solutions. And, unfortunately, 25 years after I started thinking about this, there still is no single approach that addresses this problem in all situations. What has evolved is a set of approaches that vary in their complexity and ability to address the mismatch problem in specific situations. In this installment of “LC Troubleshooting,” I discuss two approaches that have gained attention in the last couple of years. The solvent mismatch problem has also been addressed in past installments of “LC Troubleshooting.”^1-3 Interested readers might consider revisiting them if they would like to develop a fuller appreciation for the options available.

Mobile Phase/Sample Solvent Mismatch: The Fundamental Problem

Figure 1 illustrates the fundamental problem we face in situations where there is a significant mismatch between the composition of the sample and mobile phase solvents. With the word “composition” here, we are typically referring to the volume fraction of organic solvent, particularly in the case of reversed-phase (RP) and hydrophilic interaction (HILIC) separations, although other types of mismatch related to the pH and ionic strength of the solutions can also be important.^4,5 When using conventional injection methods, including “fixed loop” and “flow-through needle” designs, it is instructive to imagine that the sample is “spliced,” or inserted into the mobile phase stream, bracketed on the upstream and downstream sides by mobile phase (Figure 1a). As the sample moves toward and enters the column, it can occupy a significant fraction of the column volume (Figure 1b), with a solvent composition that is not very different from what was in the sample vial. The accuracy of this illustration improves as the sample volume approaches the column volume. For example, the dead volume of a typical 50 mm x 2.1 mm i.d. column is about 100 µL. If we inject 60 µL of sample, we will have a scenario like that depicted in Figure 1b, particularly if we do not take any deliberate steps to blend the sample with the surrounding mobile phase between the point of injection and the column.

Examples of the impact of mobile phase/sample solvent mismatch are shown in Figure 2. In Panel A, we see that when a small volume of sample is injected, and the composition of the sample solvent matches that of the mobile phase used as the starting point of the solvent gradient used for elution, the peak shapes are nice – symmetrical and narrow. Panel B shows that when the compositions of the sample and mobile phase solvent are matched, increasing the injection volume can have a negative impact on peak shape. Here we see tailing and broadening, particularly for the early peaks. In this case, the 40 µL injection volume is about 70% of the dead volume of the column. However, in Panel C, we see how things can get ugly quickly. In this case, the sample solvent only has 20% more acetonitrile (ACN) than the mobile phase used to start the gradient, but even this degree of mismatch causes all resolution of the injected mixture to be lost. On the other hand, Panel D shows that, if the fraction of ACN in the liquid entering the column with the sample can be reduced to 30% (20% lower than the mobile phase in this case), resolution can be restored. In other words, the combination of the injected volume and the mismatch between sample and mobile phase solvents can strongly influence separation performance.

At this point, a reasonable reaction to the discussion would be to say, “Just don’t inject such large volumes.” Indeed, this is exactly why we generally advise that the injected volume should be held to less than about 1% of the column dead volume. When detection sensitivity is not a problem, this is great advice. However, as analysts, we are not always in this luxurious position. In many contexts, we would like to improve detection sensitivity, and one attractive means of doing so is to get more moles of analyte into the column by injecting a larger volume of the sample. This then brings us to the central question: How can we beat this mobile phase/sample solvent mismatch problem and enable large volume injections without sacrificing resolution of important analytes? In this installment of “LC Troubleshooting,” I will briefly address two approaches: inline mixing and feed injection.

Approach #1: Inline Mixing

The idea of mixing the sample with the surrounding mobile phase in-line between the point of injection and the column is not at all new. In a “LC Troubleshooting” article authored in 2019 with Zach Breitbach and colleagues, we showed how this strategy could be used to address the mismatch problem in the context of RP separations of both relatively hydrophilic and hydrophobic compounds, by injecting samples rich in organic solvents.³ However, as separation challenges evolve, we see the approach applied in new ways, and sometimes to great effect. In 2025, Verduin, Somsen, and coworkers demonstrated the utility of inline mixing to address the mismatch problem in the context of HILIC separations of therapeutic oligonucleotides (ONs).⁷ Figure 3 illustrates the experimental setup in this case. A mixer is simply installed between the sample injection point and the column (Mixer-2). Figure 4 shows the dramatic improvement in separation performance that resulted in this case. Solvent gradient elution was used, with the gradient starting at 64% ACN and running down to 36% ACN over 14 min. The ON sample solvent was water, and either 20 or 40 µL of the sample was injected into a 150 mm x 2.1 mm i.d. column (dead volume is about 300 µL). Under HILIC conditions, water is a very “strong” solvent, meaning that if we inject a lot of water without doing anything to mitigate the solvent mismatch, there is high potential for severe distortion of peak shapes. Indeed, this is what we see in this case. The bottom chromatogram in each set of three in Figure 4 shows that, when no mixer is used, all the ON analytes break through at the column dead time with no retention. When Mixer A is added (35 µL mixer volume) the dilution of the water in the sample is sufficient to reduce the breakthrough and improve the peak shape of the ONs; however, moving to the larger Mixer B (100 µL mixer volume) is required to fully resolve the artifacts observed in the bottom two traces. When the injection volume is increased to 40 µL, the smaller Mixer A is totally inadequate, but a very nice separation is obtained with Mixer B. These effects are impressive, and inline mixing is clearly a very powerful approach to mitigate the solvent mismatch problem in this case.

A cautionary note is warranted here, however. The extent to which inline mixing is a viable solution depends on how “sticky” the analytes are—that is, how strongly retained are they in the solvent mixture after the sample is diluted with surrounding mobile phase? If the analytes are only moderately retained, they will migrate some as the sample moves into the column, and the benefits of the decreased solvent strength will be outweighed by the negative effect of increasing the injected volume by mixing the sample with the mobile phase. A clear example of this is shown in the same paper by Verduin and coworkers, where their Figure 2 shows that adding inline mixing can make the problem worse in the context of HILIC separations of the nucleobases adenosine, uracil, and cytosine. Readers interested in learning more about applications that have been published using the inline mixing approach are referred to the references.^3,8

Approach #2: Feed Injection

A few years ago, a novel approach to sample introduction for liquid chromatography was introduced commercially – this approach has become known as “feed injection.”⁹ Rather than “splicing” the sample into the mobile phase stream as it happens in conventional injection approaches for liquid chromatography, Figure 5 shows that, in feed injection, the sample is fed into the mobile phase stream at a rate controlled by the chromatography data system, with the required force provided by a high-pressure syringe. In this illustration, the sample feed rate is about 40% of the nominal mobile phase flow rate, which is reduced during the feeding step so that the total flow delivered to the column is not changed during the analysis. The beauty of this approach is that the mobile phase, which is already being supplied by the pump, can be used to dilute the sample inline during the injection. For example, if the sample solvent is 100% methanol, we set the pump to 2/98 ACN/water during the feeding step, with a feed rate that is 40% of the pump flow rate, then the methanol content of the sample that arrives at the column will be reduced to about 40%. For analytes that are sufficiently “sticky” to be very strongly retained under these conditions, this dilution of the sample solvent can be used to great effect.

Figure 6 shows representative results for the analysis of triclosan in a sample matrix of 100% methanol. Triclosan is an antibacterial compound commonly found in hand soaps and has been studied extensively for its persistence in the environment. A typical analysis would involve the extraction of triclosan from water by solid phase extraction (SPE), followed by the elution from the SPE material in 100% organic solvent, such as methanol, and then injection of 1 µL of this extract into an HPLC system. Much of the actual extract is wasted because injecting more of it using conventional means results in severe peak distortion, as shown in Figure 6. Panel A shows that a nice peak shape is obtained when 1 µL of a sample of triclosan in 100% methanol is injected into an RP column. However, in Panel B, we see that, when the injection volume is increased to 45 µL, the peak splits and a large fraction of the analyte mass is eluted at the column dead time. On the other hand, when we analyze the same sample using the feed-injection approach, Panel C shows that we can inject 45 µL without compromising the peak shape. Because we are increasing the mass of analyte introduced into the column, the peak height and detection sensitivity improve dramatically (note the different scales in Panels A and C).

As with most other approaches used to manage the solvent mismatch problem in liquid chromatography, feed injection also works best for “sticky” analytes. Nevertheless, in our own work, we are finding that the range of analyte chemistries and conditions over which the feed injection approach is useful is quite broad. Readers interested in learning more about this emerging technique are referred to the references.^10-12

Summary

In this installment of “LC Troubleshooting,” I have revisited the topic of mismatch between mobile phase and sample solvents. There are many applications in which the sample solvent is “strong” relative to the mobile phase solvent, which can lead to severe peak distortion, particularly when we want to inject large sample volumes to improve detection sensitivity. Two experimental approaches that have been gaining attention recently to address this challenge are inline mixing and feed injection. Both techniques work best when working with “sticky” analytes that are highly retained in weak solvents, such that they can be efficiently focused at the column inlet after a sample composed of strong solvent is diluted with weak solvent. As with other techniques used to mitigate the solvent mismatch problem, neither inline mixing nor feed injection is a one-size-fits-all solution. However, they can be very effective in a variety of situations, and they are valuable additions to the liquid chromatographer’s tool kit.

Acknowledgments

I’d like to thank Molly Stein for her ideas related to the illustration shown in Figure 5, and Tina Dahlseid for acquiring the data used to produce Figure 6.

References

1. Dolan, J. Separation Artifacts I: Sample overload and injection-solvent problems, LCGC Magazine 1986, 4, 16–20.

2. Stoll, D. R. What’s trending in LC troubleshooting? LCGC North Am. 2019, 37, 18–23.

3. Breitbach, Z.; Randstrom, C.; Chang, J.; et al. Mixing and mixers in liquid chromatography, part III: solutions for problems with sample diluents, LCGC North Am. 2019, 37, 368–373.

4. Lauer, T. J.; D. Stoll, D. R. Effects of buffer capacity in reversed-phase liquid chromatography, part II: visualization of pH changes inside the column, LCGC North Am. 2020, 38, 70–74.

5. Lauer, T. J.; Stoll, D. R. Effects of buffer capacity in reversed-phase liquid chromatography, Part I: relationship between the sample and mobile-phase buffers, LCGC North Am. 2020, 38, 10–15.

6. Stoll D. R.; Leme, G. M. Instrumentation for Two-Dimensional Liquid Chromatography. In Multi-Dimensional Liquid Chromatography: Principles, Practice, and Applications; CRC Press: Boca Raton, FL, 2022; Chapter 4.

7. Verduin, J.; Tutiš, L.; Kritsima, A.; et al. Enabling large-volume injections in hydrophilic interaction chromatography of oligonucleotides through in-line mixing. J. Sep. Sci. 2026, 49 (2) e70372. DOI: 10.1002/jssc.70372

8. Tang, S.; Venkatramani, C. J. Resolving solvent incompatibility in two-dimensional liquid chromatography with in-line mixing modulation. Anal. Chem. 2022, 94 (46), 16142–16150. DOI: 10.1021/acs.analchem.2c03572

9. Graf, H. G.; Ortmann, T.; Yang, P.; et al. Overcoming the strong sample solvent effect for sustainability measurement challenges in liquid chromatography. Example for bisphenol-A. Anal. Chem. 2024, 96 (42), 16854–16860. DOI: 10.1021/acs.analchem.4c03624

10. Foshag, D.; Buckenmaier, S.; Petersson, P. Feed injection-enabled reversed phase liquid chromatography for simplified analysis of lipophilic drugs and formulations. J. Chromatogr. A 2026, 1774, 466887. DOI: 10.1016/j.chroma.2026.466887

11. Böth, A.; Foshag, D.; Schulz, C.; et al. Feed injection in liquid chromatography: reducing the effect of large-volume injections from purely organic diluents in reversed-phase liquid chromatography. J. Chromatogr. A 2024, 1730, 465165. DOI: 10.1016/j.chroma.2024.465165

12. Buckenmaier, S.; Riemenschneider, C.; Schächtele, A.; et al. Chromatographic techniques for improving the LC/MS quantification of PFAS. J. Sep. Sci. 2025, 48 (5), e70155. DOI: 10.1002/jssc.70155