Some liquid chromatography (LC) troubleshooting topics never get old because there are some problems that persist in the practice of LC, even as instrument technology improves over time. There are many ways for things to go wrong in an LC system that ultimately manifests as detection sensitivity that is lower than expected. Developing a short list of the likely causes of these results can help streamline our troubleshooting experience when sensitivity-related problems occur.

Writing this “LC Troubleshooting” column and thinking about topics each month is interesting in the sense that there are some topics that just never get old. Whereas in the chromatography research world certain topics or ideas become obsolete as they are displaced by newer and better ideas, in the troubleshooting world there are certain topics that have remained relevant since the very first troubleshooting article appeared in this magazine (

 at that time) in 1983 (1). Over the last few years, I’ve focused several “LC Troubleshooting” installments on contemporary trends (for example, the relatively recent advances in our understanding of the effects of pressure on retention [2]) in liquid chromatography (LC) that are affecting the way we approach our interpretation of LC results, and approach troubleshooting with modern LC instruments. With this month’s installment, I am continuing a series I started in December of 2021 (3) that is focused on some of the “bread and butter” topics of LC troubleshooting—those elements that are essential for any troubleshooter, no matter the vintage of the system we are working with. The topics at the heart of this series are highly related to 

’s well-known LC Troubleshooting wall chart (4) that hangs in many laboratories. For the fifth installment in this series, I have chosen to focus on problems related to detection sensitivity in LC separations. As with all of the “essential” troubleshooting topics, problems with sensitivity have been discussed in “LC Troubleshooting” articles in the past (5); however, both the causes of and solutions to these problems have evolved over time. The causes of low sensitivity can have many origins, some of which depend on the type of detector being used. Discussing some of the most commonly observed causes will empower users to spot these problems when they occur and get a start on the troubleshooting effort by narrowing the possibilities to investigate and the range of possible solutions. I hope LC users young and old find some useful tips and reminders related to this important topic.

More often than ever, I find myself responding to troubleshooting questions with "everything is possible," which might seem like an easy out when considering observations that are hard to explain, but I find that this response is appropriate more often than not. With so many possible causes of poor detection sensitivity, it is important to keep an open mind when considering what might be the problem. Even aspects of LC systems and methods that are normally well controlled (for example, surfaces exposed to the mobile phase) can change; therefore, it is important to consider a wide range of potential causes of low or different sensitivity, even if most of them are normally not a concern.

A critical step in any troubleshooting exercise—but one that I think is underappreciated—is recognizing that there is a problem to be solved. Recognizing that there is a problem usually amounts to recognizing that what is happening with the instrument is different from what is expected to happen, and our expectations are formed from theories, empirical knowledge, and experience (6). What constitutes “normal” sensitivity depends strongly on the type of detector in use, the molecules we are trying to detect, and the conditions of the experiment. However, there are some trends that we can expect, even if real experiments don’t perfectly align with these ideals.

First, though, a few words about the definition of 

 are in order. In the context of chromatography, the International Union of Pure and Applied Chemistry (IUPAC) defines 

 as “the signal output per unit concentration or unit mass of a substance in the mobile phase entering the detector” (7). In other words, it is the slope of a plot of detector signal as a function of the concentration of the injected analyte (or mass). In practice, most people talk about the signal part without speaking explicitly about the concentration part. In the rest of this column, I use the term 

 in this way, which is focused on the detector signal.

Now, back to our expectations. In the best-case scenario, we can expect:

that the signal observed for a particular analyte on a particular instrument should be consistent from one sample to the next, and across days of operation, and

that the signal observed for a particular analyte should be similar on different instruments, in different laboratories, provided the specifications of the different instruments are similar.

Of course, there are many possible reasons why these expectations may not be met. The following discussion is focused on some common reasons, most of which are also highlighted on the 

The chromatographic efficiency, or plate number (

), of the column in use affects detection sensitivity. Equation 1 shows the relationship between the analyte concentration at the detector (

) and the peak width as measured by the peak standard deviation in volume units (

), in the case of a Gaussian peak. Here, 

 is the time at any point in the chromatogram; 

 is the standard deviation of the peak in time units.

Because the plate number and peak width are inversely related, as shown in equation 2, analyte concentration at the detector is proportional to the square root of the plate number (equation 3). Figure 1 shows a series of hypothetical chromatograms where only the plate number is varied; all other major variables, including retention time, flow rate, injection volume, and column dimensions, are held constant, which is relevant to a situation where the plate number for a brand-new column is 10,000, but slowly decreases over time because of column aging. At the point where the plate number has decreased by a factor of four, the peak height has decreased by a factor of two. This effect is relevant for both isocratic and gradient elution separations. In this case, the apparent decrease in sensitivity has nothing to do with the detector. Rather, it is because of a decrease in column performance.

Column diameter can also affect the observed detection sensitivity because—when all other variables are held constant—the volume of mobile phase that the analyte is eluted in increases as the column diameter is increased. This volume, which is proportional to 

, is inversely related to the detected analyte concentration, as shown in equation 1. A decrease in detection sensitivity as a result of an increase in column diameter could appear as an unintended consequence of making a change in the column used for other reasons. For example, increasing the column diameter from 2.1 mm to 4.6 mm generally makes the method less susceptible to peaks broadening because of extracolumn effects (8), which is a good example of why we need to consider methods holistically when making changes: Many variables in the chromatographic process are interconnected.

Chemical Causes of Decreased Sensitivity 

Unfortunately, some analytes adsorb strongly (that is, they stick) to materials and surfaces used in LC systems. The mechanisms behind this stickiness are analyte dependent and too complex to discuss in detail here. However, this issue has gained significant attention in the research community and from instrument manufacturers in recent years because of increasing interest in the analysis of biomolecules including nucleotides and proteins, and we (9–11) and others (12) have written about it in some detail.

The way I talk about this issue with my students is to say that the instrument likes to “eat” certain types of molecules under certain conditions, which is particularly problematic whenever a new instrument component is used in the flow path between the injector and detector. This is true for connecting capillaries, filters, columns, and even detector flow cells. A more technical description is that the first molecules injected into the system after installing a new component (or even used components that have not seen a particular molecule type for a while—for example, a column that has been used for protein separations is suddenly repurposed for oligonucleotide separations—adsorb to the material and then desorb very slowly (over minutes or hours), such that they are not observed at the detector as an eluted peak. Eventually, if enough of the sticky analyte is injected, all of the available adsorption sites in the flow path will be saturated with the bound analyte, and subsequent injections of the analyte will actually make it all of the way to the detector and be observed as an eluted peak. This process of saturating the adsorption sites with the analyte by sacrificing some initial sample injections before doing any separations that are important is sometimes referred to as “priming” the system (13). For example, in the protein and peptide separation area, it is common to prime a new column by repeatedly injecting a low-cost protein sample (such as bovine serum albumin [BSA]) until the peak area stabilizes, which is an indicator that adsorption sites have been saturated, and the fraction of the analyte that makes it through the system to the detector becomes more consistent. Clearly, one of the biggest mistakes the analyst could make here would be to use data from the first injections for quantitative purposes. Doing so would result in an underestimation of the quantity of the analyte in the sample because some of the analyte would have been stuck in the system and not accounted for at the detector—sometimes seriously so.

As discussed above, the growing importance of biomolecule analysis in recent years (that is, peptides, proteins, and nucleotides) has led to more detailed studies of the adsorption issue, and some instrument manufacturers have introduced new columns and instrumentation with surfaces that are intended to mitigate strong adsorption effects under chromatographically relevant conditions.

Low Sensitivity When Using Spectroscopic (UV Absorbance and Fluorescence) Detectors No Chromophore

Although ultraviolet–visible (UV-vis) absorbance detectors are the most commonly used type of detector for LC, they do not provide enough detection sensitivity to be universally useful for all types of analytes. An analyst who has successfully used UV-vis detection for many applications involving analytes containing aromatic functional groups may be surprised at the apparent loss of sensitivity of the system when switching to develop a method for a different type of molecule. However, if the new group of molecules does not have a chromophore (that is, part of a molecule that absorbs light; in the case of UV detection, these are usually functional groups that contain pi electrons) that absorbs strongly, then very little UV absorption signal will be observed, even at high analyte concentrations. A classic example of this process is the analysis of sugars. Simple sugars do not have any pi electrons and are weak absorbers of UV light above 200 nm where most LC work is done. The sensitivity is so poor that other detector types are almost always used for sugar analysis.

Occasionally, an apparent loss of sensitivity is because of the inadvertent use of a data acquisition rate that is too low and causes severe broadening (and thus also a severe decrease in apparent peak height) of peaks. If the spacing between data points registered by the data system is on the order of the peak width, it is possible to nearly entirely miss the peak apex. Default methods supplied by instrument manufacturers sometimes specify a data acquisition rate that is rather low by default. Users developing very fast separations with peak widths less than 1 s should be especially careful when choosing an appropriate data acquisition rate.

Loss of Sensitivity When Using Mass Spectrometric Detectors 

Mass spectrometers are powerful, yet complex instruments, and there are many ways for things to go wrong that ultimately manifest as a loss of signal relative to expectations. Two common problems that actually have nothing to do with the inner workings of the instrument are ionization suppression and adduct formation. In the case of electrospray ionization (ESI), which is the most commonly used ionization method when coupling LC and mass spectrometry (MS), analytes and other molecules present in the shrinking droplets produced by the ionization source compete for a fixed amount of available charge in the droplet. Molecules with higher affinities for cations (for example, amines readily associate with protons; esters and similar functional groups can associate strongly with ammonium ions) present in the droplets will associate with the cations more readily than molecules with lower affinities (like hydrocarbons), thus producing the ions that are eventually separated by the mass analyzer and detected to produce the observed signal. If two samples that contain identical concentrations of a target analyte contain different concentrations of other molecules (that is, the sample matrix), very different signals may be observed for the target analyte if some of the matrix molecules compete with the analyte for available charge, and thus “suppress” the ionization of the target analyte. This ionization suppression is most often because of sample matrix components. However, in my laboratory, we have also observed several instances where mobile-phase impurities have caused significant ionization suppression and loss of MS signal. When the problem is because of sample matrix components, improving the LC separation can reduce the degree of ionization suppression (14).

Although it is most common in the use of ESI–MS to detect analytes in their protonated forms (for example, association of a proton with an amine or an alcohol to produce a positively charged cation), it is also possible, and sometimes even advantageous, to detect analytes associated with other cations, including ammonium, sodium, and potassium ions. When carefully controlled, detection of these adducts can provide sensitivity enhancements (15); however, in many cases, the non-proton adducts are viewed as a nuisance; in the worst case, they can lead to a significant apparent loss of sensitivity. For example, consider the detection of the molecule benzophenone (C

O). The monoisotopic mass of the neutral molecule is 182.07 Da.

When the molecule associates with a proton to produce a +1 ion during the electrospray process, the monoisotopic mass-to-charge (

) ratio of the ion is 183.08. However, if the same molecule associates with a sodium ion to also produce a +1 ion, the observed 

 ratio will be 205.06. Obviously, if a method is developed such that only the signal for the 183.08 is recorded (for example, an extracted ion chromatogram (EIC) for the [M + H

] ion) but the target analyte enters the MS as the sodiated adduct with an 

 of 205.06, it will not be “detected,” and its concentration would inaccurately be estimated at a value that is too low.

You might ask, “Where do these sodium ions come from to produce these adducts given that I don’t have sodium ions in my mobile phase?” Unfortunately, many reagents contain impurities including sodium, potassium, and ammonium salts, and glass bottles in common use for mobile-phase containers also leach these cations slowly over time. These effects can be minimized by choosing appropriate reagents and switching to plastic containers, along with using specific mobile-phase additives (16), but analysts should at least be aware that adduct formation like this is a distinct possibility and can suddenly and unexpectedly lead to apparently low sensitivity.

In this fifth installment on essential topics in LC troubleshooting, I discussed some potential causes of apparent losses in detection sensitivity. These problems can have physical origins, chemical origins, or both, and they can manifest differently depending on the type of detector being used. As with other essential troubleshooting topics, effectively troubleshooting sensitivity problems starts with a clear expectation of what the typical detector response looks like for a given target analyte with a given detector, operating under a particular set of conditions. Understanding the details associated with the potential causes of sensitivity loss discussed here provides a good place to start troubleshooting, but it does not capture all possibilities. Readers interested in learning about a deeper list of causes and solutions for problems related to detector baselines are referred to the 

(4) LCGC North America, LCGC Troubleshooting Wallchart. https://www. chromatographyonline.com/view/ troubleshooting-wallchart (accessed September 2022).

(7) IUPAC, Compendium of Analytical Nomenclature (2022). 

https://media.iupac.org/publications/analytical_compendium/

(9) S. Fekete, D. Guillarme, and D.R. Stoll, 

(10) J.J. Hsiao, G.O. Staples, and D.R. Stoll, 

(11) J.J. Hsiao, T.-W. Chu, O.G. Potter, G.O. Staples, and D.R. Stoll, 

(12) H. Lardeux, A. Goyon, K. Zhang, J.M. Nguyen, M.A. Lauber, D. Guil- larme, and V. D’Atri, 

https://doi.org/10.1016/j.chroma.2022.463324

https://doi.org/10.1016/j.jpba.2020.113105

Articles

Column: LC Troubleshooting

There are various physical and chemical causes of low detection sensitivity. Here, we address some of these causes, and how to troubleshoot them.

Essentials of LC Troubleshooting, Part V: What Happened to My Sensitivity?

Porous layer open tubular (PLOT) columns are traditionally built with particles that are adhered to the tubing walls. These columns have unique selectivity and provide a great alternative when gaseous samples need to be separated, but these columns also have been used to separate higher boiling point analytes. There are many different commercially available stationary phases of PLOT columns, including alumina-based columns, molecular sieves, and porous polymers. Alumina-based columns have an aluminum oxide stationary phase that is then deactivated with different salts. These columns have high capacity, superior loading ability, and produce symmetrical peaks. Molecular sieve columns are designed specifically for permanent gas separations because the columns have high retention. Porous polymer columns are highly hydrophobic, making them more applicable to analyzing a wider range of samples.

There are a wide variety of commercially available stationary phases of porous layer open tubular (PLOT) columns (Table I), such as alumina, molecular sieve, carbon, silica, and a variety of porous polymers. In general, PLOT columns are traditionally built with particles that are adhered to the tubing walls.

Alumina columns have an aluminum oxide stationary phase that is then deactivated with various salts. These columns have high capacity and
produce symmetrical peaks. Molecular sieve columns are designed specifically for permanent gas separations. Carbon-layer columns provide
selectivity for inorganic and organic gases, like separating carbon monoxide and carbon dioxide. Silica columns are a great alternative to packed
columns because they are made to be selective for C1–C4 compounds. Porous polymer columns are hydrophobic, which makes them more applicable to analyzing a wider range of samples. DVB/vinylpyridine polymers are moderately polar and DVB/ethylene glycol/dimethacrylate are the most polar polymer columns. These commercially available PLOT columns (Table I) are extensively tested to produce superior separations and are made to not shed particles, which was previously a concern with PLOT columns (1). Although PLOT columns have been primarily applied in gas chromatography (GC) with the main focus on separating low boiling point analytes (2,3), there is evidence of PLOT column use beyond light samples (4–6). The versatility of PLOT columns has been exploited even further by their use in dynamic vapor microextraction (DVME) (7–9), and even in liquid chromatography (LC) (10–12). Here, we discuss the application of PLOT columns beyond their original primary use for permanent gaseous sample separations.

The first laboratory-made PLOT columns were introduced in 1963 (13,14). Since that time, these columns have been mostly used for separating light mixtures like permanent gases (2,15,16) or light hydrocarbons (3,17,18). It has been shown that PLOT columns are very retentive for light compounds. For example, a 10-m molecular sieve column has been used to separate permanent gases in the presence of hydrocarbons and nitriles (2). With the exception of helium and neon (which were coeluted), all the other permanent gases (argon, dinitrogen, krypton, xenon), along with methane and carbon monoxide, were baseline separated (2). Alumina columns have been evaluated by Ji and others, where they showed that these columns are highly retentive for C1–C6 compounds (3). The baseline separation of methane, ethane, ethylene, propane, and higher carbon content compounds was achieved using a 30-m alumina column (3). Retentiveness of PLOT columns that was obtained with light compounds could potentially be applied to higher boiling point analytes and there has been some work done throughout the years to broaden the use of PLOT columns.

In 1974, fast, high temperature separations were performed using a molecular sieve column by Soulages and associates where the paraffin and naphthalene content was investigated in saturated hydrocarbon distillates (19). Arpino and colleagues in 1977 presented a new PLOT column that was made with a graphitized thermal carbon-black polar stationary phase that could be used for GC and LC (20). This column was used to investigate two oxidation reactions, where citral (boiling point: 225 °C) was produced from geraniol (boiling point: 230 °C) (20). Reaction products were separated, and more than 30 by-products were identified (20). Another application, reported in 1980 by Uden and others, involved separating volatile metal chelates and organometallic compounds (like chromium trifluoroacetylacetonate, chromium hexatluoroacetylacetonate-trifluoroacetylacetonate che- lates, copper (II) tetradentate p-ketoamine chelates, and cyclopentadienylmanganese tricarbonyl compounds) using carborane-silicone Dextil PLOT stationary phases in a stainless-steel column (21). Following that, in 1983, Nijs and associates implemented alumina-coated PLOT columns to separate hydrocarbons up to C10 (4). In the same study, the authors demonstrated a beautiful separation of naphtha, with p-xylene (boiling point: 138 °C) being the heaviest compound present (4). Another report, where some higher boiling point analytes were separated in conjunction with permanent gases, was published by Greally and colleagues in 1998 (5). In this study, a novel PLOT column with a hydrophobic silica layer coated on a fused-silica capillary was used (5). Various halogenated hydrocarbons were separated in this study; although most analytes had a boiling point <0 °C, they also separated heavier halogenated hydrocarbons, like chloroform (boiling point: 61 °C) (5).

Another more recent work involving hydrocarbon mixture separations was reported by Wawrzyniak and others while evaluating a novel stationary phase of 4,4,4-trifluoro-1-phenyl-1,3-ketoiminepropyl silane used as a porous layer bonded on to 5-μm nucleosil silica (22). Halogenated hydrocarbons up to C6 were analyzed, which facilitated the separation of the heaviest compound in this study of 1,2-dichlorobenzene (BP: 180 °C) (22). Another unique application of PLOT columns includes isotopic separations where a commercially available porous polymer PLOT column has been used to separate acetaldehyde and methanol from their deuterated isotopologues, performed using headspace sampling (6).

Implementation of PLOT columns has also emerged in the field of comprehensive two-dimensional gas chromatography (GC×GC), where the complementary separation mechanisms provided by two different PLOT columns can be leveraged to produce intriguing 2D separations. Light hydrocarbons and some solvents were separated using different commercially available PLOT columns by Patrushev and associates in 2018, with the heaviest compound separated being methylcyclohexane (boiling point: 101 °C) (23). GC×GC has also been shown to provide informative 2D separations for gasoline, which we described in our recent report, where a distinct commercially available PLOT column was applied for each dimension. On the first dimension (

D), a silica PLOT column was applied, while on the second dimension (

D), a porous polymer PLOT column was used (24).

PLOT columns have unique selectivity and provide a great alternative to wall-coated open tubular (WCOT) columns or packed columns for separations of light samples (25–27). Environmental Protection Agency (EPA) Appendix IX volatiles calibration megamix was separated by GC using the same temperature program and chromatographic conditions on a WCOT column (Rtx-200) in Figure 1a, and on a PLOT-Q column in Figure 1b. Most of the mix eluted at the beginning of the chromatogram using a WCOT column; while using the PLOT-Q column, the entire mixture was well retained and separated, everything from carbon disulfide (boiling point: 46 °C) to naphthalene (boiling point: 218 °C). In this example, even a very short length PLOT column provides unique selectivity for low boiling point analytes relative to the selectivity that is provided using a WCOT column. A heavier sample of gasoline was used to further show the unique separation abilities of PLOT columns. Gasoline is one of the heaviest “common” samples that can be run on PLOT columns. Figure 2 presents optimized gasoline separations on a porous polymer (PLOT-S) column (Figure 2a) and an alumina oxide (Rt-Alumina BOND/MAPD) column (Figure 2b). The heaviest compound that elutes from both columns is 1-methyl naphthalene, which has a boiling point of 244 °C. Although both columns provide intriguing separations of gasoline with relatively short run times, Rt-Alumina BOND/MAPD provides a better separation of very similar molecules. As shown in Figures 2c and 2d, neopentane, 1-pentene, isopentane, and pentane are separated into four peaks on Rt-Alumina BOND/ MAPD (Figure 2d), whereas on PLOT-S column (Figure 2c), these compounds overlap into two peaks. Rt-Alumina BOND/ MAPD also provides a mix of sharper and broader peaks. These separations show the ability to apply PLOT columns for samples containing compounds with a wide variety of boiling points—for example, isobutane (boiling point: -11 °C) to 1-methyl naphthalene (boiling point: 244 °C).

Dynamic Vapor Microextraction (Cryoadsorption)

More recently, PLOT columns have also been used for cryoadsorption, which was developed at the National Institute of Standards and Technology (NIST) in 2009 by Bruno and colleagues (7). This technique is also referred to as dynamic vapor microextraction (DVME). More specifically, it is an ultrasensitive, low volume purge-and-trap sampling method where vapor-phase analytes are concentrated on a section of the PLOT column (9). DVME is a dynamic headspace sampling technique that is especially useful when low volatility compounds are of interest.

A cryoadsorption apparatus for DVME adapted from the pioneering report by Bruno and colleagues is shown in Figure 3a (7). In this work, the apparatus was equipped with a short alumina PLOT column as a purge trap (Figure 3b) operating at a low temperature. The apparatus was evaluated using a pure explosive compound 2,3,4-trinitrotoluene as well as a practical explosive C-4 (7). Later, the same group published a broader study using alumina-coated PLOT columns for cryoadsorption where industrial and military plastic bonded explosives (PBX) tagged C-4, Semtex-1A, Semtex-H, detonating cord (detcord), and sheet explosive (Detaflex), were investigated (28). DVME was adapted later by Nichols and others. In this study, short PLOT columns were used in cryoadsorption on arson fire debris samples (29).

Most recently, cryoadsorption was adapted to develop a portable vapor sampling device. The first account of such work was published in 2016 by Bruno and colleagues, where the instrument was laboratory tested using naphthalene, kerosene, and other hydrocarbons, producing accurate detection and quantitation after a short collection time of 3 s with a concentration limit of detection (LOD) of less than 1 ppm (8,30). Then, this device was in-field tested by Harries and associates in 2019 on an old U.S. Army communication bunker as a simulated shipping container (31). This device was able to detect naphthalene, two-explosive-related compounds, four protein decomposition markers, and gasoline (31). This work was further expanded on by another study that examined how long after in-field collection the samples stay preserved when collected on the PLOT column. It was determined that the integrity of the samples was fully preserved for 24 h, with losses occurring by the seventh day (9).

DVME has been shown to be a very sensitive technique. Burger and others examined the difference between natural gas sampling using multi-PLOT-cryo (Figure 3c) and a direct injection of the same amount of natural gas using a gas-tight syringe (Figure 3d). The study showed how much more sensitive and suitable for the analysis of trace components the PLOT-cryo technique is over traditional sampling methods (32). Because of the superior sensitivity of DVME, it has been used in recent years to analyze a broad range of samples: fuels (33), cannabinoids (34), and forensics (28,29,35,36).

Traditionally, LC is performed using packed columns with bonded phases, but it has been shown that open tubular columns can provide higher column efficiency (37). There have been a few studies into using open tubular columns in LC in the 1970s and 1980s (38,39), and it has been shown that to reach column efficiencies close to what is achieved in GC, the columns must be not more than 15 μm in inner diameter (40). Open tubular column use was revitalized in LC after coupling to electrospray (ES) mass spectrometry (MS) (41).

PLOT columns have become increasingly popular in LC in recent years. Most papers have focused on using PLOT columns for LC separations of various biological samples. In 2010, Rogeberg and associates separated intact proteins in skim milk using polystyrene divinylbenzene PLOT columns (10); narrow peaks (0.144–0.33 min) and good within-column repeatability in retention times (0.6% RSD) were achieved, with less than 1.1% carryover injection-to-injection. In 2011, Wang and colleagues analyzed haptoglobin from lung cancer patients with a 10 μm i.d. polystyrene-divinylbenzene PLOT column by LC coupled to collision-induced dissociation/electron transfer dissociation–MS (LC−CID/ETD-MS) (11). Separation of three glycopeptides (T1, T2, and T3) was performed, whereby the intensity of a 10-fmol sample separation using the PLOT column was similar to that of a 200-fmol sample separation when a standard packed column was used for LC–MS (11). A new zwitterionic PLOT column was introduced by Peng and others in 2016 that was used to separate flavonoids from rootstock of licorice (42). Good RSD for differences in retention time run-to-run (≤1.1%), day-to-day (≤0.78%), and column-to-column (≤1.1%), were obtained.

There have also been nano-LC and pico-LC systems that have been introduced in the recent years, where a PLOT column was used for the separations. In 2009, a nano-LC system using an amine-bonded poly(vinylbenzyl chloride-divinylbenzene) hydrophilic interaction LC (HILIC) PLOT column was used for the separation of a glycan mixture, and only 50 fmol of a dextran ladder was needed to identify dextrans up to G22 (43). In 2011, Thakur and associates presented a workflow where breast cancer cells were analyzed using PLOT-LC tandem MS (MS/MS), demonstrating good quantitative reproducibility between the same patient samples (44). In 2017, Li and colleagues presented a new pico-PLOT HPLC system that was used to separate amino acid enantiomers (45). A very recent report from Vargas Medina and others in 2022 describes the development of nano-LC coupled to electron ionization (EI)–MS. Polyaromatic hydrocarbons and thermolabile pesticides were separated, and the LOD of such a system was determined to be 25 μg/L (12).

PLOT columns provide unique retention and selectivity for low boiling point analytes relative to WCOT columns (25–27). Although most of their use has been shown with light mixtures (2,3), there has been some work done to broaden their use throughout the years (4–6). We demonstrated their unique abilities to separate light compounds as well as heavier boiling point analytes (Figures 1 and 2), which shows a broad analyte range that can be analyzed using PLOT columns. As in one-dimensional (1D) chromatography, GC×GC using PLOT columns has mostly been employed to separate samples containing light boiling point compounds and continues to be an underutilized application area for PLOT columns. Further applications of PLOT columns have been shown by their use in a new sampling technique of dynamic vapor microextraction (DVME), and this technique has been applied to analyze various samples, including explosives, fuels, and cannabinoids (33–35). PLOT columns have also gained attention in the LC community because of their higher efficiency, where they have been adapted to be used in nano-LC systems to analyze mostly biological samples (10–12).

It is important to note that because there are no commercially available PLOT-LC columns, this is a research area under development with ample opportunities. Broader use of PLOT columns is a growing research area, especially with their use for heavier samples in GC. Applying these columns for sampling using dynamic vapor microextraction and utilizing their better column efficiency in liquid chromatography also demonstrate expanding utility of the technology.

(1) “Restek’s PLOT Column Family,” Restek Corporation (Bellefonte, PA, 2020).

(2) E. de Vanssay, P. Capilla, D. Cos- cia, L. Do, R. Sternberg, and F. Raulin, 

https://doi.org/10.1016/00219673(93)80261-6

https://doi.org/10.1016/S0021-9673(01)93598-8

(5) B.R. Greally, G. Nickless, P.G. Sim- monds, M. Woodward, and J. de Zeeuw, 

https://doi.org/10.1016/S00219673(98)00240-4

(6) H.-G. Schmarr, M. Wacker, and M. Mathes, 

https://doi.org/10.1016/j.chroma.2016.12.023

https://doi.org/10.1093/chromsci/47.7.569

https://doi.org/10.1016/j.chroma.2015.12.013

https://doi.org/10.1016/j.chroma.2021.462670

(10) M. Rogeberg, S.R. Wilson, T. Greibrokk and E. Lundanes, 

https://doi.org/10.1016/j.chroma.2010.02.025

(11) D. Wang, M. Hincapie, T. Rejtar, and B.L. Karger, 

(12) D.A. Vargas Medina, J.S. da S. Burato, J.V.B. Borsatto, and F.M. Lanças, 

https://doi.org/10.1016/j.chroma.2022.463143

https://pubs.acs.org/doi/pdf/10.1021/ac60202a038

(14) D.L. Petitjean and C.J. Leftault, Jr., 

https://doi.org/10.1002/(SICI)1520-6521(1996)1:1<60::AID-FACT8>3.0.CO;2-I

(16) H. Yun, M.L. Lee, and K.E. Markides, 

(17) W.-C. Liao, C.-F. Ou-Yang, C.-H. Wang, C.-C. Chang and J.-L. Wang, 

https://doi.org/10.1016/j.chroma.2013.04.008

https://doi.org/10.1016/S0021-9673(00)82853-8

(20) P. Arpino, C. Vidal-Madjar, G. Guiochon and S. Békássy, 

https://doi.org/10.1016/S0021-9673(00)98008-7

(21) P.C. Uden, D.E. Henderson, F.P. DiSanzo, R.J. Lloyd, and T. Tetu, 

https://doi.org/10.1016/S0021-9673(00)84743-3

https://doi.org/10.1016/j.chroma.2012.06.025

(23) Y.V. Patrushev and V.N. Sidelnikov, 

https://doi.org/10.1016/j.chroma.2018.10.015

(24) L. Mikaliunaite, P.E. Sudol, C.N. Cain, and R.E. Synovec, 

https://doi.org/10.1016/j.chroma.2021.462358

(25) C.-H. Wang, C.-Y. Huang, H.-K. Yak, H.-C. Hsieh, and J.-L. Wang, 

https://doi.org/10.1016/j.chemosphere.2020.128504

(26) P. Kotnik, M. Čolnik, M. Finšgar, Ž. Knez, and M. Škerget, 

https://doi.org/10.1016/j.polymdegradstab.2020.109386

https://doi.org/10.1007/s10337-021-04092-1

(29) J.E. Nichols, M.E. Harries, T.M. Lovestead, and T.J. Bruno, 

https://doi.org/10.1016/j.chroma.2014.01.080

(30) M. Harries, S. Bukovsky-Reyes, and T.J. Bruno, 

https://doi.org/10.1016/j.chroma.2015.12.014

https://doi.org/10.1016/j.forc.2019.100182

(32) J.L. Burger, T.M. Lovestead, and T.J. Bruno, 

https://doi.org/10.1021/acs.energyfuels.6b00043

(33) M.E. Harries, S.S. Wasserman, J.L. Berry, and K.M. Jeerage, 

https://doi.org/10.1016/j.forc.2020.100301

https://doi.org/10.1016/j.forc.2017.06.003

https://doi.org/10.1016/j.foodchem.2010.01.044

https://doi.org/10.1016/j.forsciint.2010.05.024

(37) K. Göhlin, A. Buskhe, and M. Larsson, 

(39)T. Tsuda, K. Nomura, and G. Nakagawa, 

https://doi.org/10.1016/S0021-9673(00)87275-1

, C. Horváth, Ed. (Academic Press, Cambridge, Massachusetts, 1980), pp. 1–56.

(41) G. Hulthe, M.A. Petersson, and E. Fogelqvist, 

(42) L. Peng, M. Zhu, L. Zhang, H. Liu, and W. Zhang, 

(43) Q. Luo, T. Rejtar, S.-L. Wu, and B.L. Karger, 

https://doi.org/10.1016/j.chroma.2008.09.105

(44)D. Thakur, T. Rejtar, D. Wang, J. Bones, S. Cha, B. Clodfelder-Miller, E. Richardson, et al, 

https://doi.org/10.1016/j.chroma.2011.09.022

(45) R. Li, Y. Shao, Y. Yu, X. Wang, and G. Guo, 

Column: Column Watch

The application of porous layer open tubular (PLOT) columns can be expanded beyond light compounds to separations of analytes with heavier boiling points.

Application of Porous Layer Open Tubular Columns: Beyond Permanent Gases

Biotechnology products such as biopharmaceuticals, including biosimilars, are produced in living organisms or cells. A common phrase that we hear is “the process is the product.” Thus, this process of producing these products in living organisms presents several unique challenges to consider. One of those key challenges is the characterization of not only the product itself, but other impurities in the form of host cell proteins (HCPs) that may be present. To ensure that these HCPs do not have clinical implications, it is important to understand what they are. Therefore, in this column, we take a closer look at specific analytical technologies and tools to analyze these host cell proteins, with a specific focus on liquid chromatography (LC) and mass spectrometry (MS) approaches.

Biopharmaceuticals, unlike small molecules, are manufactured using living cells, such as Chinese hamster ovary (CHO) cells. The active pharmaceutical ingredient (API) is produced in cells such as CHO cells using recombinant DNA technology, which allows for the production of large amounts or quantities of the respective APIs. As the desired API is being manufactured by living cells, these cells are going through their normal cellular life cycle and processes; that is to say, they are still producing their own macromolecules, such as nucleotides, lipids, carbohydrates, and proteins, to stay alive. These host cell macromolecules, including host cell proteins (HCPs), are considered impurities, and should be removed to ensure product quality, safety, and efficacy from the drug product (1–4).

Nearly two-thirds of biopharmaceutical drug shortages are because of quality issues during the manufacturing process, including the presence of host cell proteins (5). HCPs and other impurities can affect product quality, safety, and efficacy in numerous ways, which is why they need to be reduced or removed from drug products. One of the most prominent concerns with HCPs is immunogenicity. Residual HCPs are foreign objects to our immune systems; thus, they could cause an immune response (immunogenicity). HCPs also may be proteases or proteins that have proteolytic activity, which can cause biologics such as monoclonal antibodies (mAbs) to be cleaved into fragments, thus reducing product yield and activity. HCPs with catalytic activity can also cause leaching of protein A from protein A columns (for example, cleaving protein A and releasing fragments into the purified product). Protein A causes a detrimental immune response. Also, HCPs could cause protein aggregation of the biologically active pharmaceutical ingredient, thus reducing its yield and activity (2).

Host Cell Proteins As a Critical Quality Attribute

An important part of quality by design (QbD) is defining critical process parameters (CPPs) and critical quality attributes (CQAs). A 

 is a parameter that can affect the quality of the product. Critical quality attributes are physical, biological, chemical, or microbiological characteristics that should be monitored, and they fall within a specific limit, range, or distribution (6). For drug products, CQAs may include degradation products, post-translational modifications, and host cell proteins.

Often times, CQAs are described in the context of the drug substance (API), and changes that may occur to it. For example, changes to post-translational modifications such as glycosylation and charge state variants of a mAb are monitored as CQAs. However, impurities such as HCPs are also important CQAs to monitor to ensure product quality (3).

Host cell proteins can be monitored in both the upstream process and the downstream process. In the upstream process, the characterization of HCPs can give an indication of what host cell proteins may be present throughout the process, such as an abundance of a specific protease or other cellular protein. It can also give an indication of host cell proteins that may be hitchhiking with the APIs through specific or nonspecific binding events. The presence of certain HCPs may inform the downstream purification scheme that is implemented. Monitoring the presence in the downstream purification process can also ensure if adjustments need to be made to the purification process, or suggest alternative approaches to remove these HCPs. Thus, monitoring HCPs is another characterization of the drug substance and product that ensures product quality (3).

Enzyme-Linked Immunosorbent Assay (ELISA) for Characterization of Host Cell Proteins

The enzyme-linked immunosorbent assay (ELISA) has been used as the primary tool for detecting HCPs, largely because of its high throughput, selectivity, and sensitivity. However, there are several limitations to the ELISA method. The antibodies raised against the cell line supernatant or lysate do not always recognize all HCPs present, especially weak and non-immunoreactive HCPs. The development of a suitable ELISA can take several months, if not years, to develop. Limited antibodies to a specific HCP or excess antigen can cause a dilution-depended nonlinearity that prevents proper identification. Also, ELISAs are not able to identify the specific HCP present, but instead just the presence or absence of HCPs and the reduction of their presence through process-related purification (4).

LC–MS/MS for Characterization of Host Cell Proteins

These limitations and others to the use of an ELISA to monitor HCPs are partially the reason for the development of orthogonal techniques to characterize host cell proteins. The key to the development of an orthogonal technique to the ELISA includes, but is not limited to, being able to identify the specific HCP, as well as the need not to have to create a new assay to measure the presence of new HCPs. In addition, as an impurity assay, the method should be able to detect low levels of any given HCP (3).

Over the last several years, regulatory authorities such as the U.S. Food and Drug Administration (FDA) have asked for improved monitoring methods, such as liquid chromatography–mass spectrometry (LC–MS), to characterize HCPs. Interestingly, the FDA and other regulatory authorities (such as the European Medicines Agency), as well as international standard-setting bodies (such as the International Council for Harmonization), do not provide specific numerical limits of the HCP level, as the risk associated with HCP levels are dependent on several factors, including dose and patient population, along with several others. These and other factors are indications that HCP analysis by LC–MS will only increase in the coming years (3).

 was first used by Marc Wilkins in 1996, and, for decades, LC has been used for protein separation and purification, and mass spectrometry used for the characterization of proteins within a proteome (7). Thus, proteomic techniques can be used to characterize and monitor HCPs, as it is a “simple” proteomics experiment. Perhaps the greatest advantage to LC–tandem MS (LC–MS/MS) analysis of HCPs is the ability to positively identify and quantify specific HCPs, as well as characterize the total amount of HCPs present, something that an ELISA cannot do. Each specific HCP can be monitored in a high-resolution mass spectrometer without the need for anti-HCP antibodies (3), which allows for the identification of new or previously undetected HCPs to be characterized without the need for developing a new assay. The same LC–MS/MS assay can be used to characterize HCPs across products produced in various cell lines, allowing for faster characterization of the product, which becomes particularly important when developing new biologics to deploy during something like a global pandemic (such as Covid-19). In addition, by knowing the specific identity of the HCP, specific approaches can be developed and employed to remove it from the process, or to remove it as an impurity (3).

As for every experiment, reproducibility is of prime importance, and LC–MS/MS analysis of HCPs is no different, especially as the analysis is comprised of a complex workflow. Briefly, sample preparation includes denaturation, reduction alkylation, and (tryptic) digest of the cell lysate or cell supernatant, creating thousands of peptides (8). The peptides created from the digest are then analyzed using LC–MS/MS to identify the HCP and quantify it. In addition to these steps, proper quality control standards are required to ensure reproducibility, and more sensitive mass spectrometers are likely to give a deeper understanding of the HCP profile (3).

Both data-independent acquisition (DIA) and data-dependent acquisition (DDA) can be used to characterize HCPs. In a DIA experiment, all the peptides present are fragmented by MS/MS analysis, which increases the probability of identification of low abundant peptides or proteins. In a DDA experiment, the instrument selects the peptides for MS/MS fragmentation based on certain thresholds set by the operator (3), usually characterizing 4–10 of the most abundant peaks. DDA experiments may miss peptides or proteins of low abundance; however, by selecting the most intense peaks, the quality of the data, and thus the confidence in the protein identification, is increased. The data produced is then searched in a database to make positive protein identifications. To ensure reproducibility, it is recommended that at least two peptides per protein be identified to state the presence of a specific HCP (3).

A Few Thoughts on Multi-Attribute Method (MAM) for HCP Characterization

The multi-attribute method (MAM) is a method used to monitor critical quality attributes of biological products. It includes the digestion of the drug substance and monitoring sequence coverage of the product, as well as characterization of post-translational modifications (PTMs) 

on the product. The other aspect of MAM, which is often disregarded, is new peak detection (NPD) (9).

NPD can be used to detect new peaks that are specific to the product, such as new peptides from missed cleav- ages, or the addition of new PTMs. At its core, however, NPD is used to detect impurities in the product. Thus, in theory, NPD could be used to detect new HCPs that are present in the drug product or substance.

Ensuring that safe, effective, and quality products get to patients is of utmost importance. Over the last several years, there has been great momentum gained in supporting a quality by design approach to drug development; that is, ensuring quality is considered from the very beginning of the drug development process (6). This has led to the process of understanding critical process parameters and characterizing critical quality attributes.

Critical quality attributes often are considered attributes that change on the active pharmaceutical ingredient; however, they can also include monitoring changes to the drug product, specifically to impurity profiles. One of those impurity profiles that has become increasingly important to monitor are host cell proteins, as they can lead to adverse effects to patients (3).

Historically, multi-analyte ELISAs have been used to monitor HCPs because of their speed, sensitivity, and high throughput. However, over the last several years, regulatory authorities such as the FDA have asked that orthogonal techniques, such as LC–MS/MS, be included in HCP analysis. LC–MS/MS analysis can allow for the specific identification and quantification of a given HCP, which can inform downstream purification strategies, thus potentially increasing the purity of the drug product (3).

LC–MS/MS analysis of HCPs has only been used since 2005 (3); therefore, there are likely to be significant advances over the coming years, especially with the increased regulatory requirements for LC–MS/MS analysis of HCPs (3). Improvements will likely follow, from new sample preparation methods for proteomic experiments to the development of more advanced high throughput mass spectrometers. In addition, new workflows (such as MAM and others) may play a significant role in HCP analysis is in the future.

(1) J. Conner, D. Wechterl, M. Lopez, B. Minshall, R. Prusti, D. Boclair, J. Peterson, and C. Allen, in 

Biotechnology Entrepreneurship: Starting, Managing, and Leading Biotech Companies

, C. Shimasaki, Ed. (Academic Press, London, United Kingdom, 2014), pp. 351–385.

(2) C.H. Goey, S. Alhuthali, and C. Kontoravdi, 

(3) K. Pilely, M.R. Johansen, R.R. Lund, T. Kofoed, T.K. Jorgensen, L. Skriver, and E. Mortz, 

(4) J. Zhu-Shimoni, C. Yu, J. Nishihara, R.M. Wong, F. Gunawan, M. Lin, D. Krawitz, P. Liu, W. Sandoval, and M. Vanderlaan, 

(5) A.C. Fisher, M.H. Kamga, C. Agarabi, K. Brorson, S.L. Lee, and S. Yoon, 

(6) L.X. Yu, G. Amidon, M.A. Khan, S.W. Hoag, J. Polli, G.K. Raju, and J. Woodcock, 

(7) B. Aslam, M. Basit, M.A. Nisar, M. Khurshid, and M.H. Rasool, 

(8) A. Shevchenko, H. Tomas, J. Havlis, J.V. Olsen, and M. Mann, 

Column: Focus on Biopharmaceutical Analysis

In response to regulatory concerns, host cell protein (HCP) analysis is now often conducted using LC–MS/MS. Unlike ELISA, LC–MS/MS can positively identify and quantify specific HCPs and characterize the total amount of HCPs present.

Analytical Characterization of Host Cell Proteins (HCPs)

Two-dimensional gas chromatography (GC×GC) offers improved separation power for complex samples containing hundreds to thousands of analytes. However, several considerations must be made to determine whether multidimensional gas chromatography (MDGC) is the logical instrument choice to answer a particular scientific question, including, but not limited to, whether the analysis is targeted or non-targeted, the number of analytes of interest, and the presence of interferences that are coeluted, as well as any potential regulatory or industrial constraints. Currently, MDGC remains daunting for many users because of data complexity and the limited tools commercially available, which are critical for improving the accessibility of MDGC. Herein, we discuss considerations that may assist analysts, laboratory managers, regulatory agents, instrument and software vendors, and those interested in understanding the applicability of 2D-GC for the scientific question being investigated.

Since the introduction of modern gas chromatography (GC) in the 1950s, GC has been established as the premier technology for the analysis of volatile and semi-volatile organic compounds because of its ease of use, rugged design, and separation power (1). GC is employed across research, regulatory, and industrial applications, requiring the development of more powerful instrumental and data analysis approaches to keep pace with the wealth of complex samples that require analysis. One such instrumental improvement aims to increase peak capacity by employing a second GC separation dimension, referred to as multidimensional gas chromatography (MDGC). The most popular form of MDGC, and the type discussed herein, is comprehensive two-dimensional GC (GC×GC), a technique pioneered approximately 30 years ago by Liu and Phillips (2) that offers an order of magnitude higher peak capacity when method run times are held constant (3–7). For samples such as the diesel fuel as seen in Figure 1, the alkane band dominates the one-dimensional (1D) separation in Figure 1a. The addition of a second column separates alkanes in the 2nd (

) dimension, increasing peak capacity, and allowing improved separation of more trace level analytes, such as aromatics, as demonstrated in Figure 1b.

With the invention of more powerful instrumentation comes the increased responsibility of clearly defining the problem to be addressed, and then carefully selecting the best analytical approach, whether it be GC×GC or 1D-GC with an appropriate detector, be it a mass spectrometer or otherwise. In general, there are several considerations that should be made when performing GC separations, including sample preparation and introduction (8–12), hardware and software selection (5,13,14), and more (15). This article will specifically focus on the chromatographic questions and considerations that need to be made regarding whether to perform one- or two-dimensional GC.

The best technique for any analytical problem depends on multiple factors, such as: 1) whether the study is targeted or non-targeted; 2) if targeted, the number of target analytes to be separated; 3) whether there are interferents in the sample from the extraction process, stationary phase background, or matrix that are co-eluted with the analytes of interest; 4) the advantages or limitations of available detectors; and 5) resources, such as analyst training and availability, time, money, and other regulatory or industrial constraints. For example, as can be observed in Figure 2, quantification of preservatives in cosmetic products would likely fall in the bottom left corner of the diagram because of the relatively low sample complexity and targeted quantification (16). In contrast, a differential comparison between similar matrix samples would likely benefit from the increased peak capacity of GC×GC (3,17,18).

Comprehensive Multidimensional Gas Chromatography: Strengths and Benefits

In the case of highly complex samples (those that contain more than 150–250 relevant compounds, or have complex matrices not removed via sample preparation), the separation power of traditional 1D-GC decreases because of limited peak capacity (7). As observed in Figure 1b, the addition of a second separation dimension, when shown in a contour plot, enables both 

-direction time separations. Consistently co-eluted compounds, especially low-boiling point compounds, can now be separated by the second dimension (generally polarity differences) in addition to the primary separation (generally mobile-to-stationary phase affinity differences). When complementary stationary phases are selected, the second dimension enables a large increase in available space for analytes to occupy. In terms of raw peak capacity, multidimensional chromatography can provide a ~10-fold increase in peak capacity (17).

Complementary stationary phase selection can not only dramatically increase peak capacity of the entire separation, but it also allows chromatographers to independently determine the stationary phase of the primary or secondary column, which is selective for the analytes of interest. For example, a 100% dimethylpolysiloxane primary column, coupled with a secondary column containing 8% phenyl polycarborane-siloxane, will allow for polychlorinated biphenyl (PCB) compounds to be separated according to ortho ring substitution (19). Additionally, GC×GC chromatograms are highly structured according to compound type, as evidenced in Figure 1b of a diesel sample. Alkanes are eluted in a discrete band and are separate from the aromatic region, for example (20–22). Ultimately, one of the advantages of GC×GC is its ability to provide valuable information about probable compound structure and isomer identity, even when an analytical standard is unavailable or library match is poor (19,23–27).

The heart of a GC×GC instrument is the modulator, which, in essence, acts as a second injector onto the shorter, second dimension. Regardless of the type, the modulator traps and compresses analytes prior to the detector, which increases signal to noise (S/N) and can improve limits of detection (LOD) (3–5,28–30). In the case of cryogenic jet-based modulators, peak compression in the modulator can impart a three- to nine-fold increase in analyte sensitivity compared to 1D-GC (17,31). This is of particular interest for environmental analyses, where scientists are asked to detect ever-decreasing levels of analytes (such as perfluoroalkyls and other persistent organic pollutants [POPs]), or when the target presents a health concern, as in the cases of some carcinogens and mutagens (19,23,26,32–34). Furthermore, the oven ramp programs for many chromatographic separations often reach a plateau, causing late-eluting analytes to broaden and lose S/N. The modulator can assist detection by increasing S/N for analytes at the edge of detectability for 1D-GC (35,36).

The evidence is clear that GC×GC is an effective tool for complex analyses, non-targeted analyses, and low detection limits, and could be deployed more universally in laboratories. However, cases exist where, despite its clear advantages, GC×GC is unnecessary for accurate and timely achievement of the analytical goal at hand.

The Flip Side: When 1D-GC is Sufficient or Preferred

The largest subset of analyses that are not suitable for GC×GC consist of small, well-defined problem sets. These include targeted analyses that already have well-established methods, analyses occurring in simple sample matrices that are easily separated from the compounds of interest, or analyses in which a suitably selective detector already exists. For example, quantitative methods confirming ingredient-level concentrations of safe targets, such as preservatives in cosmetic products or caffeine in cocoa, generally have well-established methods using GC or LC coupled with either an univariate or multivariate detector (37,38). Even some cases of increasing complexity can be easily analyzed by chromatography coupled with an appropriate mass spectrometer when a unique 

 or appropriate multiple reaction monitoring (MRM) transition can increase the selectivity and sensitivity of the target over the matrix, such as the analysis of pesticides by GC-triple quadrupole mass spectrometry (TQMS) (37). Detectors that are selective for certain compound classes, such as sulfur chemiluminescence, can provide the selectivity needed for a method to differentiate between the sample and target (39–41). When such techniques are available, and additional information about the sample is not needed, they should be employed over utilizing GC×GC.

There has been increased interest within the GC×GC field regarding the development of GC×GC instruments with other mass spectrometer types (18,42–44). However, a limiting factor that may exclude GC×GC from use with MS is the scan rate required for effective sampling of modulated peaks. These peak widths can be as low as 50 ms (29). To effectively sample such narrow peaks for quantification, a scan rate of at least 100 Hz is generally required. Older detectors and higher resolution instruments do not have this fast, broad-range scan rate (45–47). For those instruments that do have the necessary scan rate, higher resolution data can impart upon GC×GC data a level of complexity beyond which most analysts can manipulate and interpret. Whether it is a hardware or software limitation, detector selection must be suitable for GC×GC analysis—from separation to data interpretation.

Raw GC×GC data can be complex and difficult to interpret, especially in non-targeted work. Consequently, analysts must choose between the following: 1) manual interpretation; 2) vendor software; and 3) in-house data analysis tools. Depending on the complexity and volume of data, manual interpretation may be tedious and impractical for industrial or regulatory laboratories. Although there has been a large amount of growth with vendor data analysis software in recent years, there exists a need to expand the abilities of each software platform and make these tools accessible to a broad audience (48–50). Currently, many GC×GC-focused analysts have turned to the development of in-house chemometric tools to bridge the gap between vendor software (5,13,14,51–54). However, this generally requires the knowledge of at least one scripting language (such as Python or Matlab) and a strong understanding of the statistics-based and chemometric algorithms that aid in the manipulation of GC×GC data and the elucidation of important features within the data. Ultimately, chromatographers now need not only a deep understanding of analytical principles, but they also need a working knowledge of data science.

Beyond the chiefly analytical reasons not to employ GC×GC, there are more practical reasons some avoid the technique, namely, time and resources. GC×GC requires time and expertise to set up, maintain, optimize, and validate GC×GC systems and methods. For urgent industrial problems, there may not be time to employ GC×GC when same-day answers are needed. For non-targeted discovery, GC×GC can be an invaluable tool, but many will move from non-targeted GC×GC to targeted heart-cutting MDGC or 1D as soon as targets are identified for simplicity and allocation of resources (47,55). For example, a GC×GC system can take up valuable resources, whether it be bench space for a dedicated GCxGC system in the laboratory, liquid nitrogen for a cryogenic modulator, or an experienced analyst to maintain the system and analyze the data. Although R&D teams may be able to simply "turn off" the modulator and run a GC×GC system as 1D, such modifications of validated methods may not be permitted in regulatory or industrial environments (56,57).

Ultimately, the principles and considerations when selecting a GC platform can be influenced by a researcher’s own implicit bias related to GC×GC. Experienced members of the GC×GC community have seen firsthand the incredible separation power of these instrument platforms, especially when higher order chemometric analysis is desired. The danger in this mentality is the reality that not all scientific questions require an additional separation dimension. However, although the use of GC×GC can provide valuable information regarding complex samples, it has not gained widespread use thus far. Perhaps a major obstacle for the implementation of GC×GC in many laboratories involves the inherent complexity of the data produced (14). Ultimately, MDGC provides multiple dimensions of data—primary and second dimension separation times, and if a mass spectrometer is used as a detector, mass spectral information. Data analysis may seem daunting, especially in the cases of large data sets, and when coupled with the relative infancy of many chemometric software platforms, may delay the desire to fully integrate GC×GC into regulatory, industrial, and forensic laboratories (18). Undoubtedly, GC×GC would benefit from continued investment in GC×GC hardware and software tools, as well as effective communication, to ultimately bridge the gap between chemistry and data science, and increase accessibility of these powerful separations.

https://doi.org/10.1093/chromsci/29.6.227

(3) L.M. Blumberg, F. David, M.S. Klee, and P. Sandra, 

https://doi.org/10.1016/j.chroma.2008.02.044

(5) S.E. Prebihalo, K.L. Berrier, C.E. Freye, H.D. Bahaghighat, N.R. Moore, D.K. Pinkerton, and R.E. Synovec, 

https://doi.org/10.1021/acs.analchem.7b04226

(6) M.S.S. Amaral, Y. Nolvachai, and P.J. Marriott, 

https://doi.org/10.1021/acs.analchem.9b05412

https://doi.org/10.1016/S0021-9673(99)00815-8

(10) N. Reyes-Garcés, E. Gionfriddo, G.A. Gómez-Ríos, Md.N. Alam, E. Boyacı, B. Bojko, V. Singh, J. Grandy, and J. Pawliszyn, 

When should multidimensional GC approaches be used for analyzing complex samples and analytes?

October 2022 | LCGC North America

Essentials of LC Troubleshooting, Part V: What Happened to My Sensitivity?

Application of Porous Layer Open Tubular Columns: Beyond Permanent Gases

Analytical Characterization of Host Cell Proteins (HCPs)

Multidimensional Gas Chromatography: Benefits and Considerations for Current and Prospective Users

Drug Development Process: Nonclinical Development of Small-Molecule Drugs

The Return of the HPLC Conference – A Viewpoints Column

Determination of Polychlorinated Biphenyls in Water Samples Using a Needle Trap Device Combined with Gas Chromatography