Key Points:
- Unlike conventional LC×LC approaches, LC×SFC offers significantly higher orthogonality between separation dimensions. This allows for more complete retention space coverage and higher effective peak capacities.
- When combined with MS/MS, LC×SFC enables comprehensive structural characterization of isomers. The four-dimensional dataset (1D & 2D retention times, molecular ions, and fragments) supports precise identification of closely related compounds, even in highly complex matrices like lignin derivatives, microalgae sterols, or synthetic polymers.
- Recent technical developments—such as optimized interface configurations, modulation valve control, and flow-splitting strategies—have overcome previous limitations in online LC×SFC. These innovations enable faster analysis without compromising separation quality, making LC×SFC a reliable, high-throughput tool.
Combining liquid chromatography (LC) with supercritical fluid chromatography (SFC) in a two-dimensional (2D) setup offers high orthogonality and improved isomer separation—ideal for complex, neutral analytes. Although online LC×SFC remains rare because of chemical incompatibilities between dimensions, offline formats already show strong results in non-targeted analysis. LC×SFC–MS/MS enables isomer distinction even with similar fragmentation. Recent instrumental developments have shown that the modulation interface in online LC×SFC plays a major role in injection effects, and its simple modification can lead to a significant improvement of the separation quality. Performance has been enhanced to achieve a high level of coupling reliability, resulting in a promising future for new applications.
Two-dimensional liquid chromatography (2D-LC) is commonly used for the analysis of complex samples. When performed in comprehensive mode, it offers huge peak capacity, theoretically equal to the product of peak capacities in each dimension. However, insufficient orthogonality between retention mechanisms in the two separation dimensions can lead to incomplete retention space coverage, which can limit 2D-LC, particularly when dealing with samples containing a large number of neutral molecules for which pH shifting between dimensions is ineffective (1). In recent years, liquid chromatography coupled with supercritical fluid chromatography (LC×SFC) has emerged as a promising approach for the analysis of neutral compounds (2,3). SFC as a second dimension (2D) provides a wide range of stationary and mobile phases, offering opportunities for a strong orthogonality with the first dimension (1D) retention model. This powerful combination has demonstrated significant gain in peak capacity compared to conventional LC×LC (4,5) and huge coverage of the separation space across various applications (4–9). Over the past two years, this technique has demonstrated strong performance across a broader range of applications and has proven particularly effective at separating isomers in highly complex samples. The latest instrumental advances have also confirmed the versatility and robustness of the online setup.
Offline LC×SFC: Isomer Separation and Identification Across Various Applications
In numerous applications featuring complex samples with many molecules, whether structurally close or including isomers, a 2D separation method hyphenated with mass spectrometry is compulsory to achieve both peak resolution and ion information. SFC is well known for its ability to separate isomers, making it an interesting approach when employed as the second separation dimension in such a 2D method. In addition, isomers can sometimes be distinguished by tandem mass spectrometry (MS/MS) fragmentation. The LC×SFC–MS/MS technique therefore offers an attractive strategy for the analysis of complex samples where isomers are suspected to be present in large numbers. In such an approach, the time scale is limited by the acquisition frequency of tandem mass spectrometry. Decoupling the two separation dimensions in an offline mode, in which the fractions are physically collected and reinjected, allows such a slower detection system. It also allows for a longer gradient time in the second dimension and thus increased peak capacity compared to the online process. This increases resolution and favors the separation of isomers.
Despite the inevitably longer analysis time, the power of the offline LC×SFC–MS/MS approach at providing a large amount of information in a single injection has been demonstrated in three different applications, including the valorization of natural products (depolymerized lignin, microalgae) and the monitoring of industrial processes (synthetic lubricants) (6–8). It generates a four-dimensional data set with two retention times, a molecular ion, and fragment ions for each peak, helping the characterization of each isomer (6,7). This data set facilitates the identification of potential structural candidates from databases and enables the construction of molecular networks to further improve confidence in the annotation and identification process (6).
In a first application, offline LC×SFC has proven to be an effective valorization method by characterizing lignin compounds derived from the fractionation of lignocellulosic biomass (6). The 2D method was developed on more than 30 standards and then applied to a catalytically depolymerized lignin sample. A high effective peak capacity value of 3218 was achieved and the significant orthogonality enabled the classification of lignin compounds based on their double bond equivalent value (DBE). Detected compounds with DBE ≥ 4 had at least one isomer, as is the case for the six presented in Figure 1a, with DBE = 11. Using LC only, compounds 2 and 5 would coelute, as would compounds 3, 4, and 6. Similarly, if only the SFC dimension was used, compounds 1, 2, and 3 would appear as a single peak, as would compounds 5 and 6. These isomers share the same mass-to-charge ratio (m/z) value, making it impossible to differentiate them using one-dimensional analysis. Moreover, the coelution would render the MS/MS fragmentation pattern unreadable. In addition, it was observed that these isomers (aside from compound 1) exhibit very similar fragmentation patterns, sharing more than three MS2 ions as indicated by their correlation in the molecular network (Figure 1a), further complicating the understanding of the MS/MS spectra. Due to the high peak resolution achieved by LC×SFC, the isomers have been successfully separated. The clean MS/MS data obtained for each compound and the correlation between fragmentation spectra and structural databases have allowed us to attribute a structure to each isomer with good confidence.
A second study on a microalgae sample focused on the separation of phytosterols (7), which are valued for their bioactive properties. Method development was carried out using a mixture of sterol standards, varying in either mass or isomeric forms, to optimize LC and SFC separations while considering the orthogonality between dimensions. The LC×SFC analysis of the microalgae sample covered 67.5% of the separation space, with the sterol retention zone delimited from non-sterols compounds, preventing coelution and ionization suppression in the MS source. The standards used during method development were successfully identified in the actual sample through four-dimensional data validation: retention time in 1D, retention time in 2D, m/z molecular ion in MS1, and m/z fragments in MS2. Moreover, the study revealed that isomers with distinct MS/MS fragments were coeluted in SFC but separated by LC (horizontal lines in Figure 1b), while isomers exhibiting the same fragmentation pattern were separated in the SFC dimension (vertical lines in Figure 1b). These differentiations are based on the nature of the lateral chain and on the position of the double bond within the sterol core, respectively. Expanding this isomer correlation to new compounds detected in non-targeted analysis led to the identification of new sterol suspects. Their standards were purchased, injected into each dimension, and their presence in the sample was confirmed using 4D data. One-dimensional chromatography would result in coelution of isomers across both dimensions, while MS/MS alone would be unable to distinguish those with identical fragmentation patterns.
To sum up, the combination of LC×SFC and MS/MS data enabled the characterization of the microalgae sample, confirming the presence of known sterols and precisely annotating new ones.
The last case involves the separation of synthetic polymers. Bio-lubricants can incorporate synthetic polyester polyols as either a base oil or a viscosity modifier. The polymer chain is formed by reacting ricinoleic acid with itself and/or with fatty acids of varying lengths (structure on Figure 1c). The process is stopped by adding a polyalcohol, which provides different hydroxyl (OH) sites to perform esterification. However, this reaction may remain incomplete, leaving free OH sites. Since standards for method development are not commercially available, the method was directly developed on the industrial sample (8). Once again, the resulting LC×SFC–high-resolution (HR)MS method allowed isomer separation. Five distinct isomers were assigned to the same raw formula and regrouped into different families based on the progress of the reaction (Figure 1c). One compound corresponded to a fully esterified polyalcohol, while the other four resulted from incomplete esterification, with structure variations depending on the position of polyester and free OH groups. Due to their structural differences, these isomers were successfully separated in both the LC and SFC dimensions. Compounds containing free-OH groups exhibited stronger interactions with the 1-AA polar SFC stationary phase, leading to increased retention in the second dimension. In contrast, the completely esterified compound showed higher affinity for the C8 column as a result of hydrophobic interactions, leading to a higher retention time in LC. Because of their fragile nature, polyester polyols undergo in-source fragmentation, which, rather than being a limitation, proved advantageous by providing fragmentation data without the need for MS/MS analysis. By correlating these distinct fragmentation patterns with chromatographic retention times, structural attribution was successfully achieved for each of the five isomers.
In each of the presented studies, LC×SFC–MS(/MS) enabled the separation of isomers thanks to the orthogonality between both dimensions and thus to the different retention mechanisms based on chain nature, chain length, double bond position, or functional group position. The fragmentation process facilitated differentiation between these isomers, and, when feasible, validation through standard analysis ensured highly reliable sample characterization.
Online LC×SFC: The Latest Technical Advances
Switching to online coupling is challenging because of instrumental non-availability of LC×SFC, and the many incompatibilities between dimensions. First, the use of large concentrations of water or strong solvents such as methanol in the 1D gradient can lead to injection effects in SFC, causing peak distortion. Trapping columns can be set up at the valve interface between dimensions (10,11), replacing empty loops. This allows solvent removal before injection but requires further method optimization. The presence of CO2 in the transfer loops limits modulation possibilities, such as splitting upstream of the 2D LC valve. To balance between low 2D injected volumes while ensuring modulation periods long enough to accommodate the 2D gradient, very low 1D flow rates have been employed in the past (4,9). This strategy, while technically effective, was at the detriment of 1D pump stability and method repeatability.
Our recent work has shown that modifying the interface between LC and SFC can significantly enhance the quality of 2D separations. Two preferential configurations were identified (Figure 2a and 2b), depending on chromatographic conditions in each dimension (12). These configurations involve transferring the fraction using empty loops, either into the CO2 stream when both dimensions use protic solvents, or into the modifier stream when aprotic solvents are employed. As a result, peak widths were reduced and peak shapes improved, even at high injection volumes; this was in contrast to the conventional mixed stream configuration, where CO2 and modifier streams remained combined. To further gain in analysis time and effective peak capacity, the CO2 stream configuration benefited from the addition of a split upfront the modulation valve. A second valve controlling the depressurization step before fraction loading in the transfer loop was added to prevent sample loss from CO2 uncontrolled depressurization through the split (Figure 2c) (13). With this modification, the 1D effluent entered an empty modulation valve. The successful implementation of a flow split in both preferential configurations enabled an increase in 1D flow rate and 2D peak capacity while ensuring perfect repeatability and achieving a tenfold improvement in retention time precision.
These technological advancements hold great potential for integrating LC×SFC coupling into other research laboratories and expanding its application to the industrial sector.
Online LC×SFC: Advantages Compared to Other Separation Techniques: An Illustration through the Analysis of a Synthetic Lubricant
Characterizing complex samples with more than a hundred molecules is challenging or even impossible using one-dimensional liquid chromatography (1D-LC) because of numerous coelutions and ion suppression in the MS source. When the sample contains only non-ionizable compounds, as is the case in the selected example here, three options involving 2D separation upfront of a MS detection system can be considered: i) using reversed-phase LC (RPLC)×RPLC with the stationary phase/mobile phase combination able to generate the most varied interactions, ii) using RPLC×normal-phase LC (NPLC), or iii) using RPLC×SFC with a polar stationary phase in the second dimension. Using the example of the industrial lubricant made of ricinoleic acid and esterified fatty acids, we wanted to illustrate here the advantages and disadvantages of each 2D strategy.
When using a RPLC×RPLC combination, it is usually of interest to have different pH values in the mobile phases of the two dimensions to change the ionization state of the analytes and thus the selectivity. However, when dealing with non-ionizable molecules, this is ineffective. Maximizing the differences in interaction between the two dimensions can be achieved by the selection of i) very different kind of stationary phases, such as in this example a C18 stationary phase in 1D vs. a fluoro phenyl column in 2D, and ii) the use of different mobile phases, here an acetonitrile–MTBE gradient in 1D, and a 20:80 (v/v) H2O–acetonitrile/EtOH gradient 2D (Table I). Nonetheless, the retentions remain highly correlated, resulting in a strong diagonalization and a low effective peak capacity (Figure 3a).
To enhance orthogonality, the 2D separation can include interactions based on the polarity of the analyte using NPLC. To allow a fair comparison in our example in Figure 3, the first dimension remained unchanged, while the second dimension was replaced by a 1-AA column (the best option among polar stationary phases) and a heptane–isopropanol gradient (Table I). The developed method significantly improved the 2D plot (Figure 3b), with the appearance of three distinct parallel lines, demonstrating an increase in orthogonality. The coverage of the separation space expanded from 18% in RPLC×RPLC to 47% for RPLC×NPLC. However, sample characterization was still complex because of broad peak widths, limiting the effective peak capacity to only 142.
Using the very same 1-AA stationary phase combination, SFC with a CO₂–acetonitrile gradient was investigated to replace the NPLC dimension (Table I). Due to the unavailability of an online system, offline LC×SFC was conducted. A visual assessment clearly demonstrates improved separation compared to previous LC×LC methods, with three distinct parallel groups and a well-defined distribution of homolog compounds within each family (Figure 3c). With an exceptional effective peak capacity of 3211 and high orthogonality covering 65% of the available space, our previous research successfully achieved a full structural characterization of the sample (8). This clearly showed the superiority of the interactions set with this combination for the analysis of neutral molecules. Unfortunately, the offline analysis took more than 12 h for a single sample.
While achieving an online coupling of the two chromatographic dimensions drastically reduces analysis time, incompatibilities between LC and SFC remain a key challenge. Strong solvents used in 1D lead to injection effects such as peak fronting when using a straightforward coupling as performed in LC×LC, that is, when the fraction is transferred in the SFC mixed stream of CO2 and modifier (Figure 3d). The presence of CO2 in the transfer loops forbids 1D flow splitting and so the 1D flow rate was limited to 5 µL/min (Table I).
Our recent work on the study of the interface between the two dimensions has shown effective ways to enhance peak shapes and minimize injection effects (12). A decision tree can guide the selection of the preferential configuration to set up for specific chromatographic conditions, such as the modifier stream that offers a huge improvement in separation quality in the present case (Figure 3e). The interface was optimized by implementing a split, thus allowing a 1D flow rate increase to 24 µL/min that reduced the total analysis time down to 110 min. The resulting decrease in normalized slope led to an almost twofold increase in effective peak capacity, reaching a huge value of 2708. The average standard deviations in retention time were 0.58 min in 1D, nearly half of the modulation period of 1.14 min, and 0.11 s in 2D, approximately six times smaller than the average peak width of 0.68 s. These values highlight the high reliability of the interface.
The in-depth study of the online LC×SFC coupling and its latest improvements have proven to be a robust, high-performance tool for the analysis of complex samples, offering superior performances compared to conventional LC×LC for the analysis of neutral compounds.
Online LC×SFC Industrial Process: Sample Sourcing
The industrial sourcing strategy relies on the careful selection of samples from different suppliers or different synthetic routes that share the same properties, usually linked to their molecular composition. In the previous example, we have shown that online LC×SFC can offer strong reliability on retention times and a peak capacity rate close to 1500 peaks per hour, a performance that allows rapid sample comparison, and which can provide a rapid update on the continuous quality of a source or the composition of a new sample.
In the first step, it was crucial to check that speeding up the analysis by working with online modulation was not detrimental to the molecular information that was attainable in a longer offline process. In previous work (8), sample 1 was fully characterized using offline LC×SFC (Figure 3c). With the online modulation and a much shorter analysis time, the same profile was observed (Figure 3e). The three main families were attributed to incomplete esterification of the polyester polyol, corresponding to the number of free OH groups in the molecule. The distribution of homologs within each family was associated with polymer length, influenced by the number of ricinoleic acid (R) units and by the chain length of fatty acids (CX) (structure on Figure 1c).
The comparison with a sample from another supplier (sample 2) was straightforward and simply based on visual observation (Figure 3e, sample 1 and 2). First, both samples exhibited an identical distribution between homologs within each family, suggesting that the same type of polyester had been used during the reaction. Second, sample 2 revealed only two main families instead of three, and both were shifted toward higher retention zones in SFC. These observations suggest a reduction in the number of free OH groups and thus in the number of families associated with less ramified molecules, increasing polarity and retention in SFC. Supporting these hypotheses, the polyalcohol backbone may contain a reduced number of hydroxyl groups.
To validate this theory, we selected the first peak of the homologous series for each sample—assumed to contain the shortest alkyl chain and a single ricinoleic group (highlighted by black dotted rectangles in Figure 3e). The extracted mass spectra (Figure 4) revealed the same fragmentation pattern for both samples, with neutral losses corresponding to ricinoleic acid or fatty acids at the chain end. This confirms that the same type of polyester was used for the reaction. Combining the fragments and HRMS-assigned raw formula allowed the molecular structure to be reconstructed and confirmed that the polyol used in the synthesis differed between the samples. Sample 2 incorporated a di-alcohol, which could be dimethylolpropane (DMP) or its more common isomer, neopentyl glycol (NPG) (attributed structure on Figure 4b), while sample 1 used a tri-alcohol, trimethylolpropane (TMP) (attributed structure on Figure 4a).
This confirms that LC×SFC offers a promising approach for industrial analyses requiring rapid comparison of several synthesis or supplier samples, enabling sample fingerprinting through simple visual inspection and allowing MS confirmation if needed.
Conclusion
Coupling liquid chromatography with supercritical fluid chromatography has emerged as the most effective approach for analyzing neutral molecules in complex samples. This two-dimensional combination provides significantly higher orthogonality than conventional LC×LC coupling, enhancing peak capacity and expanding the covered space.
Given SFC’s renowned ability to separate isomers, its role in the second dimension helps resolve multiple coelutions and allows for clear differentiation of isomeric peaks. Hyphenated with MS/MS detection further enables structural characterization of each isomer through the identification of specific fragment ions.
When performing online coupling, preferential configurations have demonstrated better separation qualities than those obtained with a conventional interface. The technique has proven to be a robust solution for industries that need rapid and reliable results.
Finally, the versatility of LC×SFC creates opportunities for future applications, opening the door to innovative approaches and advancements in a wide range of fields.
References
(1) Lazzari, E.; Arena, K.; Caramão, E. B.; Herrero, M. Quantitative Analysis of Aqueous Phases of Bio-Oils Resulting from Pyrolysis of Different Biomasses by Two-Dimensional Comprehensive Liquid Chromatography. J. Chrom. A 2019, 1602, 359–367. DOI: 10.1016/j.chroma.2019.06.016
(2) Burlet-Parendel, M.; Faure, K. Opportunities and Challenges of Liquid Chromatography Coupled to Supercritical Fluid Chromatography. TrAC Trend. Anal. Chem. 2021, 144, 116422. DOI: 10.1016/j.trac.2021.116422
(3) Kaplitz, A. S.; Mostafa, M. E.; Calvez, S. A.; Edwards, J. L.; Grinias, J. P. Two‐dimensional Separation Techniques Using Supercritical Fluid Chromatography. J. Sep. Sci. 2021, 44 (1), 426–437. DOI: 10.1002/jssc.202000823
(4) Sarrut, M.; Corgier, A.; Crétier, G.; et al. Potential and Limitations of On-Line Comprehensive Reversed Phase Liquid Chromatography×Supercritical Fluid Chromatography for the Separation of Neutral Compounds: An Approach to Separate an Aqueous Extract of Bio-Oil. J. Chrom. A 2015, 1402, 124–133. DOI: 10.1016/j.chroma.2015.05.005
(5) Crépier, J.; Corbel, E.; Lerestif, J.-M.; Berthod, A.; Heinisch, S. Characterization of Positional Isomers of Drug Intermediates by Off-Line RPLC x SFC Hyphenated to High Resolution MS. J. Pharm. Biomed. Anal. 2021, 202, 114142. DOI: 10.1016/j.jpba.2021.114142
(6) Tammekivi, E.; Batteau, M.; Laurenti, D.; Lilti, H.; Faure, K. A Powerful Two-Dimensional Chromatography Method for the Non-Target Analysis of Depolymerised Lignin. Anal. Chim. Acta 2024, 1288, 342157. DOI: 10.1016/j.aca.2023.342157
(7) Batteau, M.; Bouju, E.; Ramirez-Romero, A.; et al. Resolving Phytosterols in Microalgae Using Offline Two-Dimensional Reversed Phase Liquid Chromatography-Supercritical Fluid Chromatography Coupled with Quadrupole Time-of-Flight Mass Spectrometry. Anal. Methods 2024, 16 (15), 2278–2285. DOI: 10.1039/D3AY02261G
(8) Sanchez, M.; Lacroix-Andrivet, O.; Crozet, D.; Crepier, J.; Faure, K. Structural Elucidation of Complex Polyesters Polyols from Bio-Lubricant Using off-Line Liquid Chromatography x Supercritical Fluid Chromatography Coupled with Orbitrap Mass Spectrometry. Talanta 2024, 276, 126295. DOI: 10.1016/j.talanta.2024.126295
(9) Devaux, J.; Mignot, M.; Rouvière, F.; et al. On-Line Reversed-Phase Liquid Chromatography x Supercritical Fluid Chromatography Coupled to High-Resolution Mass Spectrometry: A Powerful Tool for the Characterization of Advanced Biofuels. J. Chrom. A 2023, 1697, 463964. DOI: 10.1016/j.chroma.2023.463964
(10) Sun, M.; Sandahl, M.; Turner, C. Comprehensive On-Line Two-Dimensional Liquid Chromatography × Supercritical Fluid Chromatography with Trapping Column-Assisted Modulation for Depolymerised Lignin Analysis. J. Chrom. A 2018, 1541, 21–30. DOI: 10.1016/j.chroma.2018.02.008
(11) Venkatramani, C. J.; Al-Sayah, M.; Li, G.; et al. Simultaneous Achiral-Chiral Analysis of Pharmaceutical Compounds Using Two-Dimensional Reversed Phase Liquid Chromatography-Supercritical Fluid Chromatography. Talanta 2016, 148, 548–555. DOI: 10.1016/j.talanta.2015.10.054
(12) Sanchez, M.; Rouvière, F.; Crepier, J.; Faure, K. LCxSFC Valve Technologies: Guidelines Toward a Successful Online Modulation. Anal. Chem. 2024, 96 (39), 15692–15701. DOI: 10.1021/acs.analchem.4c0326
(13) Sanchez, M.; Crepier, J.; Faure, K. Introducing a Depressurization Step in Online LCxSFC Modulation. Anal. Chem. 2025, 97 (7), 3822–3826. DOI: 10.1021/acs.analchem.4c06767
Margaux Sanchez is a PhD candidate in the Institute of Analytical Sciences, France. She holds a MSc in analytical chemistry from University Lyon. Her research focuses on the development of online LC×SFC instrumentation. The potential of this coupling was recently confirmed through the characterization of complex samples with high valorization impact.
Julien Crepier completed his PhD in 2017 on the interest of supercritical fluid chromatography for the characterization of complex samples and especially fast pyrolysis bio-oil at IFPEN. He is currently in charge of the liquid chromatography team in the research center of TotalEnergies (CRES). Multidimensional chromatography and hyphenation with high-resolution mass spectrometry represents the main development axes of the laboratory to achieve characterization of complex samples with high valorization impact for TotalEnergies company.
Karine Faure is a CNRS director of research at Institut des Sciences Analytiques, France. She holds a PhD in analytical chemistry from the University of Cork, Ireland, and an habilitation from University of Lyon. She has co-authored more than 50 articles and five book chapters. Her research group is dedicated to the development of two-dimensional LC and SFC with a focus on the comprehensive analysis of natural products, biowaste, and recycled products. Strongly interested in new technologies, she enjoys working with large and SME companies to transfer research into the economical world.
