Toward a Universal Metric for Chromatographic Separation Quality: A Unified Separation Quality Factor (SQF)

Traditional chromatographic quality metrics—such as plate number, resolution, and peak capacity—capture only limited aspects of separation performance and fail to generalize across modes or complex samples. To address this, the Separation Quality Factor (SQF) integrates five normalized sub-metrics into a single 0–1 score, penalizing peak asymmetry, co-elution, uneven spacing, the utilization of the elution window, and poor elution order. Unlike earlier composite metrics, SQF is adaptable across size exclusion chromatography (SEC), reversed phase (RP), hydrophilic interaction liquid chromatography (HILIC), and ion exchange chromatography (IEX), making it more universal. It supports both manual decision-making and automated optimization workflows, with optional extensions for baseline stability, offering a holistic and practical framework for evaluating and improving separations. LCGC International spoke to Szabolcs Fekete of Waters Corporation, corresponding author of a recent paper published in the Journal of Chromatography A (1) which introduces and validates a novel SQF metric, designed to overcome the limitations of existing performance descriptors by integrating multiple normalized and weighted criteria into a single score.

Why has it historically been so difficult to define a single, universal metric that captures chromatographic separation quality across different modes and conditions?

The quality of chromatographic separations depends on several aspects, such as kinetic efficiency, selectivity, peak shapes, and analysis time. However, most traditional metrics only capture one of these aspects. For instance, plate number (N) provides information about column efficiency, but not about peak spacing or symmetry (S). Resolution (Rs), on the other hand, focuses only on the separation between two peaks. Peak capacity (n_c) provides a broader view, but it assumes ideal peak spacing and uniform shapes, which rarely occur in real samples.

Consequently, one metric may perform well in one context but fail to accurately describe quality in another. Additionally, metrics developed for isocratic systems, such as SEC, do not translate well to gradient-based modes like RP, HILIC, or IEX chromatography. This lack of universality has historically prevented the adoption of a single normalized measure. Another complicating factor is that chromatographic goals differ between applications (for example, impurity profiling versus metabolomics) which makes it even harder for one descriptor to serve all needs equally well.

Classical metrics like plate number and resolution remain widely used—what are their greatest strengths, and what makes them insufficient for modern, complex separations?

N and Rs are appealing because they are straightforward, intuitive, and grounded in chromatography theory. They provide clear physical meaning: N indicates efficiency while Rs distinguishes peak pairs. One measures the kinetic performance, while the other indicates the thermodynamics of the separation. Additionally, they are essential for regulatory compliance and method validation. However, their limitations emerge when analyzing complex samples with many peaks, as local Rs values may not reflect the overall quality of the chromatogram. For example, Rs only considers the critical pair and ignores the broader peak distribution. Plate number does not account for selectivity (α) or S. These metrics also fail to address practical issues, such as asymmetrical peaks, or uneven peak spacing, which can affect quantitation. In modern workflows, especially in pharmaceutical industry and omics, we need descriptors that can evaluate global separation quality rather than isolated attributes. This is especially true as separations increasingly involve large peak counts and subtle selectivity changes, conditions under which simple pairwise descriptors quickly become misleading.

How does the proposed SQF noted in your paper (1) differ from earlier composite metrics such as chromatographic response functions (CRFs) or peak capacity models?

Earlier composite metrics, such as CRFs or n_c, attempted to integrate multiple descriptors, but they often emphasized trade-offs between speed and efficiency or assumed ideal conditions. These metrics usually lack penalties for S, co-elution, or poor spacing and were often not normalized across modes, which made comparisons challenging. For instance, n_cprovides only a best-case scenario and cannot account for real-world deviations, such as tailing or clustering of peaks. By contrast, the SQF integrates five normalized, dimensionless sub-metrics into a single score between 0 and 1, ensuring that poor performance in any one aspect meaningfully reduces the overall score and reflecting a holistic view of separation quality. Importantly, the SQF is generalizable; it was designed to work across SEC, RP, HILIC, and IEX, making it more universal than previous models. Furthermore, SQF can be weighted or adapted for different analytical priorities, allowing it to serve not just as a diagnostic but as an optimization target in automated workflows.

Can you explain how the SQF incorporates penalties for peak asymmetry, coelution, or uneven peak spacing—factors often overlooked in traditional descriptors?

One of the strengths of SQF is that it was designed to explicitly account for practical imperfections that analysts deal with every day. Peak asymmetry is captured through a normalized symmetry factor that penalizes both tailing and fronting, ensuring that even moderate deviations from ideal peak shape reduce the overall score. This is important because asymmetry can compromise integration accuracy, quantitation, and reproducibility, even if nominal resolution values appear acceptable. Uneven peak spacing is addressed by the peak distribution uniformity metric, which measures how closely the actual peak pattern matches an ideal equidistant distribution across the elution window. This prevents scenarios where a chromatogram has many peaks, but they cluster in only part of the gradient, leaving large gaps unused.

Co-elution is tackled through the critical peak order factor, which evaluates not only the Rs between critical pairs but also whether the more symmetrical peak elutes first. This matters because a tailing peak eluting before a smaller neighbor peak can severely distort or obscure it, even if their measured Rs is technically above the acceptance threshold. By penalizing such unfavorable elution orders, SQF reflects what practitioners intuitively recognize as problematic, but which traditional descriptors often ignore. All these sub-metrics are normalized, so they can be combined fairly without one dominating the others. Together, they create a quality score that is sensitive to subtle but meaningful features of chromatographic performance. This holistic treatment ensures that SQF provides a realistic assessment of separation quality, closer to how practicing chromatographers actually judge chromatograms. Additionally, the normalization makes SQF amenable to automation, as the penalties are intuitive yet quantifiable, bridging the gap between visual inspection and objective evaluation.

What challenges did you face in generalizing your earlier SEC-focused quality factor to gradient-based, retentive separations such as RP, HILIC, or IEX?

The original SEC-focused model was effective because SEC separations are usually isocratic with symmetrical and relatively small number of peaks (typically 2–5 peaks). However, gradient retentive modes, such as RP, HILIC, and IEX, are more complex. They involve broader retention ranges and a larger number of peaks, and they allow for more frequent occurrences of peak asymmetries or coelutions. A major challenge was redefining the peak distribution metric so that it would remain normalized and meaningful across these varied conditions. For example, peak capacity and spacing are relatively straightforward in SEC, but in gradient separations, peaks often cluster or spread unevenly. This requires new peak distribution uniformity and elution window utilization metrics.

Another challenge was incorporating elution order penalties, which are crucial when overloaded or tailing peaks interfere with minor components, as is typical in impurity profiling assays. These adaptations were essential to making the separation quality factor universal rather than mode-specific. A further complication was ensuring the geometric mean formulation of SQF remained balanced, so that no single sub-metric could artificially dominate the overall score, while still being sensitive enough to detect subtle performance differences.

How might the SQF framework be applied practically in method development—for example, in column screening or automated optimization workflows?

SQF can act as a powerful decision-making tool during method development because it condenses multiple aspects of separation performance into one easy-to-interpret score. In column screening, for instance, analysts often rely on visual inspection or simple metrics such as peak count, which can be subjective or incomplete. By calculating SQF for each candidate column, it becomes possible to rank them objectively and immediately identify the one providing the best overall separation. This saves time and reduces bias, especially when many columns or chemistries are under consideration.

In automated optimization workflows, SQF can serve as the objective function that software algorithms seek to maximize. Because it integrates efficiency, symmetry, peak spacing, elution window utilization and critical peak-pair order, optimization no longer needs to focus on one narrow metric such as critical resolution. SQF is also normalized, which makes comparisons meaningful across different gradient programs, flow rates, or mobile phase conditions. This means that the metric is versatile enough for both early-stage scouting and fine-tuning of robust methods. Importantly, SQF can also be customized by applying weights to emphasize the aspects most relevant for a given application. For example, prioritizing critical peak order in impurity profiling or peak distribution uniformity in omics. Overall, SQF offers a practical, scalable way to bring more objectivity, reproducibility, and automation into method development. Beyond method development, it could also serve in system suitability testing (SST) or column replacement decisions.

In your paper (1), in addition to SQF, you also mention the baseline drift as an additional (optional) quality measure. Can you elaborate?

Baseline drift is often overlooked in separation quality assessments, but it can significantly impact the accuracy of peak integration and therefore the reliability of derived metrics. Even when peaks appear well resolved, a drifting or unstable baseline may obscure small impurities or cause errors in quantitation. In the SQF framework, baseline stability can be quantified using normalized measures such as net drift, maximum slope, or root mean square deviation, each scaled relative to peak height. This optional baseline stability factor ensures that chromatograms with noisy or unstable baselines receive appropriate penalties, creating a more realistic quality score. While not always necessary for routine evaluations, it becomes especially useful in automated workflows, where visual inspection is not feasible. Incorporating baseline drift extends SQF beyond peak-centric metrics, addressing another practical aspect of real-world chromatography.

References

1. Fekete, S.; Kormány, R.; Imiołek, M. A Universal Separation Quality Factor (SQF) for Evaluating Liquid Chromatographic Separations. J. Chromatogr A 2025,DOI: 10.1016/j.chroma.2025.466182