An unresolved pair of peaks. A convincing mass spectral library match that does not quite fit the sample. An overlapping chromatogram that looks much cleaner after processing.
These situations raise different questions, but they share one starting point: diagnosis. Before changing a column, concentrating an extract, accepting an identification, or trusting a fitted peak, determine which part of the analytical workflow needs attention.
Through Concentrating on Chromatography, we have had the opportunity to speak with Lee Polite, James Little, and Ronald Brown. For this article, I looked beyond those conversations to their public teaching materials, publications, and technical resources. Their work offers three complementary perspectives: understanding the separation, building a defensible identification, and interpreting what the recorded signal can actually tell us.
Taken together, these perspectives provide a practical framework for better chromatography and mass spectrometry—and for recognizing where sample preparation supports the result without asking it to solve the wrong problem.
What should you check before changing a chromatography method?
Start by defining the failure. Poor resolution, unexpected signal response, uncertain identification, and difficult integration are not interchangeable problems. They require different evidence and different interventions.
• Two analytes remain poorly resolved: ask whether retention, selectivity, or efficiency limits the separation.
• LC–MS response behaves unexpectedly: investigate whether coeluting matrix constituents affect ionization.
• An unknown produces several plausible library hits: determine which candidate agrees with the full body of analytical evidence.
• Overlapping peaks complicate integration: ask whether a supported mathematical model can estimate useful parameters—and where its reliability ends.
The point is not to make troubleshooting more complicated. It is to avoid spending time optimizing a variable that does not address the limitation.
Lee Polite: improve the separation by understanding what limits it
Lee N. Polite, president and founder of Axion Analytical Labs, has spent more than three decades teaching chromatography to professional scientists. His background includes a chemistry PhD earned under Harold McNair at Virginia Tech and industrial experience at Amoco. His educational materials consistently connect chromatographic theory with decisions analysts make at the bench.
Read the resolution equation as a troubleshooting map
Chromatographic resolution depends on three major factors: efficiency, selectivity, and retention. The conventional resolution equation expresses their relationship:
Rs ≈ (√N / 4) × [(α − 1) / α] × [k / (1 + k)]
Here, N describes column efficiency, α describes selectivity between adjacent analytes, and k describes retention. This relationship is a useful conceptual guide, rather than a universal prediction for every gradient separation or asymmetric peak.
Think of the terms as three separate questions:
• Retention: are the analytes sufficiently retained?
• Selectivity: does the chromatographic system differentiate between them?
• Efficiency: how narrow are their bands?
That distinction matters because narrower peaks do not necessarily solve inadequate chemical discrimination. If two compounds interact almost identically with the chromatographic system, increasing efficiency may help less than changing the chemistry that controls their relative retention.
Polite’s GC teaching makes this tangible: temperature influences retention, stationary-phase chemistry influences selectivity, and column dimensions and operating conditions influence efficiency. The practical lesson is to connect an adjustment to the factor it is intended to improve.
Choose columns for more than particle size
Column selection also involves trade-offs. In his column-selection teaching, Polite considers particle size, surface area, carbon load, diffusion path length, pressure limitations, speed, and loadability. “Use smaller particles” is therefore not a complete method-development strategy.
A column must provide appropriate chemistry while remaining compatible with the instrument and the sample load. A choice that looks attractive on an efficiency specification may not address the critical pair—or may create an operational constraint elsewhere.
A practical exercise for your next difficult separation
Before changing the method, write down:
1. Which pair of peaks limits the analysis?
2. Is the problem insufficient retention, inadequate selectivity, or excessive width?
3. Which variable will you change, and which limitation should that change address?
This exercise applies Polite’s resolution framework to experimental planning. It gives each trial a purpose and makes an unsuccessful experiment informative rather than merely disappointing.
James Little: identify unknowns with converging evidence
James Little brings decades of industrial mass spectrometry experience, including his work at Eastman Chemical Company, to his educational resources and Mass Spec Interpretation Services. His public materials emphasize systematic unknown identification using complementary databases, experimental evidence, and chemical reasoning.
Understand the “known unknown”
An unknown compound is not necessarily a new compound.
Little uses “known unknown” to describe a substance that is unknown to the analyst but already reported in the chemical literature or represented in a reference database. This distinction changes the investigation: the task may be to locate and evaluate existing information rather than establish an entirely novel structure.
His workflow combines spectral searching with structure databases such as CAS Registry/SciFinder and ChemSpider. These “spectra-less” databases can provide candidate structures even when a suitable reference spectrum is unavailable. Accurate mass, molecular formula, sample history, and other information then help narrow the possibilities.
Treat a library match as evidence, not the entire case
A strong library match is valuable. It is not a substitute for checking whether the proposed identity agrees with the rest of the data.
Little’s identification framework considers:
• EI or MS/MS spectra.
• Accurate mass and molecular formula.
• Isotope patterns.
• Fragmentation pathways.
• Adduct chemistry.
• Structural plausibility.
• Sample history and analytical context.
The important relationship is agreement. A candidate should explain the observations, not simply produce the most attractive search score.
For example, when a candidate looks promising, ask whether the observed ion has been assigned correctly before interpreting its accurate mass. Then consider whether its isotope pattern and fragments support the proposed chemistry. Finally, ask whether the compound makes sense in that particular sample. Those checks apply the evidence categories in Little’s workflow without assuming that one measurement establishes identity.
Where practical, comparison with an authentic standard provides stronger confirmation. Little and colleagues describe purchasing a standard for comparison as the best means of confirming a known unknown in their published commercial-product identification workflow.
Check the matrix before assuming the instrument lacks sensitivity
Little’s research also connects identification and method development to sample preparation.
In a 2006 Journal of Chromatography B paper, Little, Michael F. Wempe, and Charles M. Buchanan examined glycerophosphocholine-related matrix ionization effects in plasma LC–MS/MS. Their work used an in-source MRM approach involving an m/z 184 marker to investigate relevant lipid matrix constituents during method development.
This is a specific plasma application, not a universal diagnostic test. Its broader lesson is nevertheless useful: the observed response depends on more than how much analyte reaches the instrument.
When matrix constituents affect ionization, concentrating the extract is not automatically the right next step. Increasing the amount of sample introduced does not, by itself, establish that the measurement will become more reliable. Investigating the matrix and its chromatographic relationship to the analyte may be more informative than simply pursuing a larger signal.
A useful identification record therefore includes four entries: the proposed compound, supporting evidence, conflicting evidence, and the additional measurement needed to distinguish alternatives. That structure makes uncertainty visible instead of hiding it behind a library score.
Ronald Brown: examine the residuals, not just the fitted curve
Ronald E. Brown is the scientific founder and chief architect of R²N Software. His work includes nonlinear curve fitting, chromatographic and spectroscopic peak fitting, digital signal analysis, and chemometric modeling. His technical contributions also appear in published work on regularized deconvolution and extra-column effects in fast separations, discussed in an independent LCGC article.
His perspective addresses a different analytical question: how much useful information can be extracted from the signal that was actually recorded?
What peak fitting can tell you
Peak fitting uses mathematical functions to describe measured signals and estimate properties such as peak position, shape, and area. More advanced approaches can account for asymmetry and instrument-response contributions rather than assuming every peak is Gaussian.
That distinction is important. An observed chromatographic peak contains information about both the separation and the measurement system. Models that consider those contributions can offer a more informative description than a simple symmetric peak approximation.
R²N’s PeakLab documentation describes chromatographic peak models and procedures for locating possible hidden components using residuals and derivatives. These are analytical tools, however—not independent proof that every fitted component represents a distinct chemical compound.
Why a high R² is not enough
A fitted curve can look excellent while still leaving important questions unanswered.
Residuals are the differences between the observed data and the model’s predictions. Their pattern matters: systematic structure can indicate that the model has not adequately captured some feature of the signal.
An independent 2024 LCGC International article by M. Farooq Wahab, Troy T. Handlovic, and Daniel W. Armstrong explains applications of advanced peak fitting and emphasizes residual inspection. The authors note that neither R² nor an F-statistic alone establishes satisfactory goodness of fit.
For a difficult chromatogram, review the raw trace, baseline, fitted components, and residuals together. Do not let a polished overlay replace examination of the underlying signal.
What software cannot recover reliably
Mathematical separation is not physical chromatographic separation.
A fitted model estimates component contributions from recorded data. It does not change what occurred in the column, and it cannot automatically correct every interaction that occurred during detection.
The LCGC article cautions against fitting entirely overlapping peaks and notes that confidence decreases as overlap increases. It also warns that coelution-related ion suppression can produce erroneous LC–MS results. A model that describes the measured profile cannot simply restore an analyte response that was suppressed during ionization.
Before accepting a fit, ask:
1. Is the model appropriate for the observed peak shape?
2. Is the baseline treatment defensible?
3. Do the residuals show unresolved structure?
4. Is there enough information to support the proposed components?
5. Could coelution have affected detector response?
These questions turn software output into something that can be evaluated, rather than merely accepted.
Where sample preparation connects all three lessons
Separation chemistry, compound identification, and mathematical modeling are different tasks. Sample preparation influences the conditions under which each operates.
Organomation’s chromatography sample-preparation guide discusses solvent compatibility, sample concentration, filtration, and quality-control samples. These steps help define what enters the analytical system; they should be considered alongside the chromatographic and detection conditions, not after them.
Match the preparation step to the analytical objective
Solvent exchange and controlled concentration can be useful when the method requires a different solvent or final volume. But those objectives should remain distinct from cleanup and particulate removal. Removing solvent is not equivalent to removing dissolved matrix constituents, and filtration serves a different purpose from evaporation.
That distinction connects directly to the three expert perspectives:
• Polite’s framework asks whether the chromatographic system provides appropriate retention, selectivity, and efficiency.
• Little’s work asks whether the sample and its matrix support a defensible interpretation of the MS response.
• Brown-related modeling asks whether the recorded signal contains sufficient information for reliable parameter estimation.
A useful workflow review should therefore document the extraction or cleanup procedure, final solvent, final volume, and any concentration or reconstitution step alongside the instrumental method. Consistency in those details supports interpretation when chromatograms or responses change.
Give concentration a defined role
At Organomation, we manufacture evaporation equipment for analytical sample preparation. The relevant question is not simply whether an extract can be concentrated, but whether concentration or solvent exchange serves the method’s requirements.
Do not use a larger signal as the only success criterion. Consider whether the preparation supports the intended separation, acceptable matrix behavior, and defensible analysis.
Before changing the method, identify the limiting step. Then choose the preparation, chromatographic adjustment, identification measurement, or data-analysis tool that addresses it.
For workflow guidance, explore Organomation’s chromatography sample-preparation guide. When controlled evaporation is part of the method, review our nitrogen evaporators for analytical sample preparation.
Frequently asked questions
How do selectivity and efficiency differ in chromatography?
Selectivity describes how differently a chromatographic system retains two analytes. Efficiency describes band broadening and therefore peak width. Improving efficiency can narrow peaks, but changing selectivity addresses how differently the analytes interact with the separation system.
Is a mass spectral library match enough to identify a compound?
A library match is an important piece of evidence, but reliable identification should consider other available information, including accurate mass, isotope patterns, fragmentation, and sample context. Comparison with an authentic standard can provide stronger confirmation.
Why can LC–MS signal change with the sample matrix?
Coeluting matrix constituents can affect ionization, changing the observed analyte response. Little and colleagues’ plasma research investigated glycerophosphocholine-related effects, illustrating why matrix behavior deserves attention during method development.
Can peak fitting replace chromatographic separation?
Not generally. Peak fitting can estimate parameters from some overlapping signals, but reliability depends on the information present and the suitability of the model. Entirely overlapping peaks and coelution-related ion suppression are important limitations.
When should an extract be concentrated or solvent-exchanged?
Use these steps when the analytical method requires a defined final volume or compatible solvent. Evaluate them separately from cleanup, and investigate matrix behavior when response is unexpected rather than assuming further concentration will solve the problem.