HPLC 2025 Preview: Fundamentally Speaking (Part 1)

Michael Lämmerhofer: Your work is focused on fundamentals in separation science. You received the JFK Huber lecture award specifically for your “highly recognized contributions to the fundamental understanding of adsorption characteristics". I first came across your work when you described the heterogeneous adsorption of chiral solutes on chiral stationary phases, particularly protein phases. All models on chiral recognition described previously treated the surface of a chiral stationary phase as a homogenous surface, neglecting the role of non-specific adsorption sites. You overcame this simplistic paradigm. Can you tell us briefly the implication of the surface heterogeneity of chiral stationary phases on the adsorption process?

Torgny Fornstedt: I have a strong connection to this subject, as I began working on it with Georges Guiochon. Our key insight was that chiral stationary phases are not uniform—especially those based on proteins. They consist of a large number of weak, non-selective sites and only a few strong, chiral-discriminating ones. This heterogeneity explains why enantioselectivity can vanish at higher concentrations—the selective sites simply become saturated. To describe this behavior, we introduced the bi-Langmuir isotherm, which models adsorption as the interaction with two distinct types of sites:

• Type I sites: non-selective, high-capacity, responsible for general retention.

• Type II sites: Selective, low-capacity, essential for enantio-recognition.

This model accounts for key behaviors such as:

a) Peak tailing and distorted elution profiles under overload conditions.

b) Loss of resolution as a predictable outcome of site competition—not a method failure.

Although originally developed for protein-based selectors, the model also fits many synthetic and polysaccharide-based chiral phases, making it broadly applicable to preparative chiral chromatography. We have also observed that the true chiral contribution is often masked by dominant non-selective retention, emphasizing the need to separate and quantify both mechanisms through nonlinear isotherm modeling (1).

The bi-Langmuir model enables this distinction, and experimental validation has shown excellent agreement between predicted and observed elution profiles across a wide range of chiral selectors. Recognizing this heterogeneity has allowed us to design more robust and predictive chiral separations—not only in analytical applications but also under the demanding conditions of preparative-scale chromatography.

ML: What is the effect on thermodynamics and kinetics? What can we learn from it in other chromatographic modes?

TF: In chromatography, peak broadening originates from different sources depending on the operating conditions. Under linear (analytical) conditions, it is primarily due to kinetics—how fast molecules interact with the stationary phase. In nonlinear (preparative) conditions, broadening is governed by thermodynamics—mainly the strength and saturation behavior of the adsorption process.

Peak tailing can result from heterogeneous interactions, which may be thermodynamic (variation in adsorption strength) or kinetic (variation in interaction speed). Both types can cause tailing.

In thermodynamic heterogeneity, tailing appears when strong binding sites become saturated. In kinetic heterogeneity, tailing arises when some sites have slower exchange rates. This effect is especially noticeable when most sites are fast because the peak becomes narrower overall, making the contribution from the few slow sites more pronounced in the tail.

To distinguish the underlying cause of peak tailing, simple tests can be used. If tailing decreases at lower flow rates, the origin is kinetic. If tailing decreases at lower sample concentrations, the cause is thermodynamic.

These principles apply across all chromatographic modes. Understanding whether peak distortion stems from kinetics or thermodynamics allows for more effective method development, optimization, and troubleshooting.

ML: In your work you introduced the concept of adsorption energy distribution. Can you briefly explain this concept? How can it enhance our understanding of adsorption processes?

TF: Adsorption energy distribution (AED) is a generalized tool that reveals how adsorption energies are distributed across a chromatographic surface. Rather than assuming one or two distinct types of adsorption sites—as traditional models such as Langmuir or bi-Langmuir do—AED shows the full spectrum of binding strengths, giving a detailed energetic “fingerprint” of the surface. This makes it especially useful for identifying heterogeneity in adsorption behavior.

AED was first introduced to the chromatography field by Brett Stanley and Georges Guiochon in the early 1990s, using mathematical inversion techniques to extract energy distributions from experimental isotherms (2).

In our group , together with Dr. Jörgen Samuelsson (3), we developed a structured four-step workflow to identify the correct physical adsorption model (4):

• Visual classification of the isotherm shape (linear, convex, or concave).

• Scatchard analysis to explore interaction patterns—linear plots suggest Langmuir, curved ones suggest heterogeneity.

• AED calculation, which distinguishes unimodal from bimodal energy distributions and narrows down candidate models.

• Model fitting and statistical testing, including parameter estimation and Fisher analysis to confirm the best fit.

A good example of AED’s power comes from our study (5) of glycine peptide adsorption (GG and GGG) on a polar, 12% crosslinked agarose gel (Superose 12), using water–acetonitrile mobile phases. Scatchard plots and model fits pointed to heterogeneity, but couldn’t clearly distinguish between Tóth and bi-Langmuir models. AED resolved the question: it showed a unimodal, slightly tailed distribution, clearly supporting the Tóth model as the most realistic.

Similarly, in our study (6) on alkaline-stable C18 columns, AED revealed how surface heterogeneity changes with pH, explaining why basic solutes like metoprolol tail at low pH but not at high pH. AED showed a strongly bimodal distribution at low pH and a more uniform one at high pH, allowing us to select the bi-Langmuir model where appropriate and to quantify site differences—something standard isotherm fits could not resolve.

In short, AED is far more than a diagnostic tool—it’s a strategic asset in chromatography. When integrated into a stepwise workflow, it provides mechanistic clarity and improves our ability to select the right physical model, enhancing both scientific understanding and practical separation performance.

ML: Your name shows up also in the field of biosensor research. This is a surprise at first glance. Can you tell us how you became acquainted with this non-chromatographic analytical technique?

TF: Yes, it might appear surprising at first, but in reality, my engagement in biosensor research grew quite naturally out of my work in fundamental separation science.

The common thread is this: both chromatography and biosensors are governed by the same fundamental principles of molecular interaction, particularly adsorption kinetics and thermodynamics. What really sparked my interest was realizing that biosensors—especially modern platforms like quartz crystal microbalance (QCM) and surface plasmon resonance (SPR)—can generate high-resolution, time-resolved data on binding events that are extremely difficult to isolate in chromatographic systems.

In chromatography, we’re often interpreting indirect readouts such as retention times, peak shapes, or overload behavior. In contrast, biosensors allow us to observe binding and dissociation in real time, at a single surface, without flow dispersion or interference from mobile phase components. This makes it possible to directly study association and dissociation rates, building a mechanistic understanding that complements chromatographic data.

At some point, I realized that by combining these two approaches, we could gain a much deeper picture of adsorption processes. That led us to develop advanced tools like the rate constant distribution (RCD) and later our own adaptive interaction distribution algorithm (AIDA)—designed to analyze complex, multi-site binding kinetics, particularly on heterogeneous surfaces. AIDA is conceptually similar to the AED tool used in chromatography, but it focuses on kinetic rather than thermodynamic distributions.

A clear example comes from our 2020 Analytical Chemistry paper, where we reanalyzed published biosensor data describing the interaction between human ACE2 and the SARS-CoV-2 receptor binding domain (RBD) (7). The original studies (8–9), based on standard one-to-one kinetic models, assumed a single homogeneous interaction and reported affinity constants accordingly. When we applied AIDA, we uncovered a broad, heterogeneous distribution of rate constants—strong evidence that the interaction involved multiple concurrent modes, not just one. In one case, the affinity constant (K_D) differed by more than 300% from the originally reported value. This highlights a major pitfall of traditional fitting approaches: they can oversimplify complex biological interactions and lead to misleading mechanistic conclusions.

So, in short, biosensors didn’t take me away from chromatography—they allowed me to look deeper into the same phenomena from another angle. Today, I see biosensors and chromatography as two sides of the same coin in molecular interaction analysis, and our work continues to bridge the two.

ML: What can we learn in chromatography from research on biosensors?

TF: While chromatography and biosensors are different in format, both are rooted in studying molecular interactions at surfaces. Biosensors—especially techniques like SPR and QCM—offer powerful, real-time insight into binding events that are otherwise difficult to isolate in chromatography. This makes them highly valuable for advancing separation science.

Here’s how biosensor research supports and strengthens chromatography:

Direct kinetic insight:Biosensors provide real-time measurements of how quickly molecules bind and unbind (association and dissociation rates). In chromatography, these rates are often hidden by flow and dispersion effects. Biosensor data help identify true kinetic limitations—such as slow mass transfer—that contribute to peak tailing and asymmetry.

Validation of surface heterogeneity: Chromatographic peak behavior often suggests heterogeneous binding sites, but biosensors can directly confirm this. Tools such as RCD and AIDA allow us to visualize and quantify multiple site populations with distinct kinetic profiles, giving solid evidence of surface complexity.

Improved mechanistic modeling: Kinetic and thermodynamic parameters from biosensors can be used in chromatographic simulations. This enhances model accuracy and reduces the need for trial-and-error fitting, making predictions more reliable—especially under nonlinear or overloaded conditions.

In summary, biosensors bring precise, real-time data to the table, filling gaps left by chromatographic measurements. Used together, the two techniques offer a more complete, mechanistic understanding of molecular interactions, moving us from empirical methods toward predictive separation science.

ML: From your scientific articles I understand that you are an expert in simulation of additive effects in liquid chromatography (LC). Every chromatographer understands that the mobile phase composition has an important effect on the separation. Effects are however often interpreted phenomenologically. In your work you develop fundamental frameworks that allow the simulation of additive effects. Where do you see the potential of this work from a practical viewpoint?

TF: To understand additive effects in chromatography, it’s important to first distinguish between modifiers and additives. A modifier is a major mobile phase component—such as acetonitrile or methanol—used to adjust the overall polarity of the eluent, influencing retention and elution strength across the board. An additive, by contrast, is a minor component—typically in low millimolar concentrations—that works by competing with the solute for adsorption sites or by forming complexes, for example, as counter-ions in ion-pairing. This allows precise control over selectivity and peak shape.

However, additives also introduce dynamic behavior that’s often overlooked. When a sample is injected, any difference between the injected plug and the bulk mobile phase creates a concentration front of the additive—what we call a system peak. These travel through the column and interact with solutes in subtle but powerful ways. Though usually undetected directly, system peaks can distort solute profiles, particularly when the additive is strongly adsorbed yet elutes ahead of the solute—a situation that defies intuition but can be explained by wave propagation theory, similar to how ripples on water move faster than the current beneath.

In analytical LC, such effects might cause moderate peak tailing or fronting. But in preparative or overloaded LC, they can produce severely asymmetric or even bizarre chromatographic profiles. We’ve shown that this can actually be turned into an advantage: in chiral prep LC, carefully tuning the additive system can lead to enantiomer peaks with complementary shapes—the first eluted peak has a diffuse front and a sharp rear whereas the second one is “normally” shaped with a sharp front and diffuse rear—boosting both resolution and productivity.

Our simulation framework makes it possible to predict and manage these effects. It moves us beyond phenomenological interpretation to a mechanistic, design-oriented approach, helping chromatographers anticipate problems and even exploit additive behavior for better separations. Notably, this applies not just to classical ion-pair systems. We’ve shown that even low levels of methanol or water in SFC can act as additives, producing system peaks and altering separations; these effects are often missed without a fundamental understanding.

In short, every system with additives has system peaks, but only strongly adsorbed additives cause serious distortions. With the right models, we can now simulate, understand, and control these effects for smarter and more robust chromatographic design.

References

(1) Fornstedt, T.; Sajonz, P.; Guiochon. G. Thermodynamic Study of an Unusual Chiral Separation. PropranololEnantiomers on an Immobilized Cellulase. JACS1997, 119 (6), 254–1264. DOI: 10.1021/ja9631458

(2) Stanley, B. J.; Guiochon, G. Numerical Estimation of Adsorption Energy Distributions from Adsorption Isotherm Data with the Expectation-Maximization Method. J. Phys. Chem. 1993,97 (30), 8098–8104. DOI: 10.1021/j100132a046

(3) Samuelsson, J. Development of Methods for Phase System Characterization in Liquid Chromatography. Ph.D. Dissertation, Uppsala University, 2008. https://www.diva-portal.org/smash/get/diva2:171707/FULLTEXT01.pdf

(4) Fornstedt, T. Characterization of Adsorption Processes in Analytical Liquid–Solid Chromatography. J. Chrom. A 2010, 1217, 792–812. DOI: 10.1016/j.chroma.2009.12.044

(5) Zhang, X.; Samuelsson, J.; Janson, J-C.; et al. Investigation of the Adsorption Behavior of Glycine Peptides on 12% Cross-linked Agarose Gel Media. J. Chrom. A 2010, 1217, 1916–1925. DOI: 10.1016/j.chroma.2010.01.058

(6) Samuelsson, J.; Franz, A.; Stanley, B. J.; Fornstedt, T. Thermodynamic Characterization of Separations on Alkaline-stable Silica-based C18 Columns: Why Basic Solutes May Have Better Capacity and Peak Performance at Higher pH. J. Chrom. A2007, 1163, 177–189. DOI: 10.1016/j.chroma.2007.06.026

(7) Forssén, P.; Samuelsson, J.; Lacki, K.; Fornstedt, T. Advanced Analysis of Biosensor Data for SARS-CoV-2 RBD and ACE2 Interactions. Anal. Chem. 2020, 92, 11520–11524. DOI: 10.1021/acs.analchem.0c02475

(8) Tian, X.; Li, C.; Huang, A.; et al. Potent Binding of 2019 Novel Coronavirus Spike Protein by a SARS Coronavirus-specific Human Monoclonal Antibody. Emerg. Microbes Infect. 2020, 9, 382–385. DOI: 10.1080/22221751.2020.1729069

(9) Lan, J.; Ge, J.; Yu, J.; et al. Structure of the SARS-CoV-2 Spike Receptor-binding Domain Bound to the ACE2 Receptor. Nature2020, 581, 215–220. DOI: 10.1038/s41586-020-2180-5