Debunking Myths and Half-Truths in HPLC and Pharmaceutical Analysis

Despite major advances in instrumentation and column technology, many long-standing, high-performance liquid chromatography (HPLC) practices persist as rules that may no longer apply in modern laboratories. This article reviews several widely circulated myths and half-truths in reversed-phase liquid chromatography (RPLC) and evaluates their relevance to today’s small-molecule drug testing. Using experimental evidence and case studies, this discussion clarifies when these practices add value and when they should be questioned or abandoned.

1. Mobile Phase (MP) Filtration Is Not Always Mandatory

It is commonly believed that MP filtration is required to remove particulates that may clog column inlet frits, pump check valves, or pistons. However, several practical reasons explain why many laboratories no longer filter mobile phases,^1-4

• HPLC-grade solvents (mobile phase B [MPB]) are prefiltered by the manufacturer.
• HPLC-grade solvents (MPB) are prefiltered by the manufacturer.
• Water from purification systems is already deionized and filtered.
• High-purity buffers (such as ammonium formate 99.995%) eliminate the need for filtration.
• Modern pumps contain internal filters that are replaced during annual preventive maintenance.

If MP filtration is still preferred as additional insurance, ensure the vacuum flask is intact, use nylon membranes for both aqueous and organic solvents, protect the vacuum pump with a solvent trap, and keep filtration glassware clean and protected from contamination.

2. Buffers Are Useful but Often Overused

Most small-molecule drugs are weak bases and are typically analyzed using acidified MPA at a pH of 2 to 4, with or without buffers.⁵ Buffers do provide accurate pH control, especially in complex separations, and improve peak shape by shielding ionic interactions between protonated analytes and residual silanol groups.⁶

However, preparing buffered MP such as phosphate buffers (weighing, pH adjustment, filtration) adds workflow complexity and increases the time analysts spend preparing solutions, thereby increasing turnaround time. As a result, high-throughput screening and in-process control methods often rely on simple acidic or basic MP (e.g., 0.1% formic acid or 0.1% ammonia),⁷ which are easier and quicker to prepare, have longer shelf lives, and still suppress silanol ionization.¹

These low-ionic-strength MP at low pH often produce severe peak tailing (also known as shark finning) for highly basic drugs ⁵ (e.g., drugs with amines, piperidines, or multiply charged groups), as shown in Figure 1 for the 4 labeled new chemical entities (NCEs).^8,9 The lower chromatogram in Figure 1 illustrates how this problem can be mitigated using a positively charged–surface bonded phase.⁹ This stationary phase fix is similar to the use of amine MP additives, such as triethylamine (TEA). In the 1980s, it was common to improve peak shape and batch-to-batch reproducibility¹⁰ by adding TEA to the MP. In comparison, permanently bonded charged-surface phases enable simpler HPLC or LC–MS workflows without the need for buffered MP.

In summary, buffers are essential for complex separations and for achieving good peak shapes on most silica-based RPLC columns (Figure 2),¹¹ but acidified MP paired with charged-surface phases can simplify workflows, as shown in Figure 3.

3. Avoiding MP pH Near an Analyte’s pK_a Is Overstated

A common teaching is to avoid MPA pH values near an analyte’s pK_a to prevent peak splitting from partial ionization. However, experimental evidence shows this concern is overstated.

Figure 4 presents isocratic separations of amitriptyline and nortriptyline across different pH levels while keeping mobile phase B constant. The capacity factors for amitriptyline are k = 1, 16, and 34 at pH 2, 8, and 10, respectively, reflecting the large hydrophobicity differences between ionized, partially ionized, and nonionized forms. No split peaks were observed at pH 8 because proton transfer in water occurs on a subpicosecond timescale via the Grotthuss mechanism.¹²

Figure 5 shows a combination drug product method at a high pH of 9.10 with 2 opioid active pharmaceutical ingredients (APIs). The optimum MPA pH was close to the APIs’ pKa values (8.8 and 9.3). Again, no split peaks were observed, and only method-robustness concerns were noted, as small pH changes affected retention times. The method is robust if MPA pH is maintained at 9.10 ± 0.05 units using a pH meter with automatic temperature compensation.¹³

4. Defaulting to 4.6-mm i.d. Columns Is a Historical Habit

Current usage of HPLC column diameters¹:

• 4.6-mm i.d.: 60%-70%
• 3.0-mm i.d.: 10%-15%
• 2.1-mm i.d.: 15%-20%

The 1/4” outer diameter (o.d.), 4.6-mm i.d. format originated in the 1960s from standard tubing stock of stainless-steel precision-bore tubing, and it persisted despite the absence of regulatory requirements. Modern USP monographs allow changing column parameters, such as the internal diameter. Today, 3-mm i.d. columns offer several advantages for quality control (QC):

• Lower solvent consumption
• Optimal flow ~1 mL/min (vs. ~2 mL/min for 4.6 mm)
• Higher peak capacity and resolution due to smaller void volume
• Compatibility with both HPLC and UHPLC systems.

For most QC applications, a 3 mm column packed with sub-3 µm superficially porous particles provides better chromatography with less solvent and time.^1,14,15

5. AI/Automation Helps More in Late-Stage Optimization Than in Early NCE Method Development

Column/MP screening systems, simulation software, and method development platforms are powerful tools for late-stage optimization and complex separations. However, they are less useful for early-phase NCE method development for stability-indicating assays¹⁶ because they require a test mixture containing the API, impurities, and degradants—materials that are rarely available early on. Forced degradation can generate degradants, but an initial method is needed to evaluate them, creating a circular dependency.^1,16

Artificial intelligence (AI) can assist with literature mining and structure-based retention prediction, but its impact is limited for NCE method development because:

• Ab initio retention predictions require empirical training data, which are unavailable for NCEs.
• Degradation products often fall outside current AI training domains.

Thus, AI complements rather than replaces established tools.¹⁷ Early-phase development often benefits more from a traditional 5-step selectivity-tuning approach¹⁸ or from using a modernized generic gradient method as a shortcut strategy.⁹

Summary and Conclusions

Figure 6 summarizes the 5 HPLC myths or half-truths and provides guidance on when to adopt, abandon, or modify these practices:

• MP filtration can often be skipped when using high-purity solvents and buffers.
• Buffers are valuable but are frequently overused; simpler acidic MP often suffices when paired with charged-surface phases.
• pH values near an analyte’s pKa do not cause peak splitting, though method robustness must be managed.
• Three-mm columns offer clear advantages over 4.6-mm columns for QC.
• AI and automation excel in late-stage optimization but have a limited impact in early NCE method development.

Most of these practices are not simply wrong. They are conditional. Sometimes they are useful, sometimes unnecessary, and sometimes counterproductive.

Acknowledgments

I gratefully acknowledge the valuable technical and editorial contributions from Matt Mullaney, Alice Krumenaker, Mike Shifflet of Kenvue, Leon Doneski of Arcutis Biotherapeutics, He Meng of Sanofi, Jacob Fairchild of Kao Specialties Americas, Manzi Malhotra of Elanco, Matthew Morse, Abraham Samuel Finny of Waters, and Sumit Mukherjee.

This work was presented in part at HPLC 2026 in Indianapolis, Indiana, on June 11.

References

1. Dong, M. W. HPLC and UHPLC for Practicing Scientists, 2nd Ed; Wiley, 2019. Chapters 2, 3,7, 9, and 10.
2. Dong, M. W. Seven Common Faux Pas in Modern HPLC. LCGC North Am. 2014, 32 (8), 552–557.
3. Dong, M. W.; Boyes, B. E. Modern Trends and Best Practices in Mobile-Phase Selection in RPC. LCGC North Am. 2018, 36 (10), 752–767.
4. YouTube Channel: HPLC in Pharmaceutical Analysis Series, EP-9. Why is Mobile Phase Filtration NOT Needed? https://youtu.be/auu150zGZ18
5. Dong, M. W. Why Are Most Drugs Basic: Implications for Pharmaceutical Testing by HPLC. LCGC Int. 2026, 3 (3), 30–33. DOI: 10.56530/lcgc.int.yi4180w5
6. YouTube Channel: HPLC in Pharmaceutical Analysis Series, EP-4: Why Don’t You “Always” Need Buffers in HPLC? https://youtu.be/d9RORcZskVc
7. Wong, M.; Murphy, B.; Pease, J. H.; Dong, M. W. High-Throughput Purification. LCGC North Am. 2015, 33 (6), 402–413.
8. Fountain, K. J., et al. Practical Applications of Charged Surface Hybrid (CSH). Waters Application Note 720003720, 2010.
9. Dong, M. W. A Universal Reversed-Phase HPLC Method for Pharmaceutical Analysis. LCGC North Am. 2016, 34 (6), 408–419.
10. Hill, D. W.; Langner, K. J. HPLC Photodiode Array UV Detection for Toxicological Drug Analysis. J. Liq. Chromatogr. 1987, 10 (2), 377–409. DOI: 10.1080/01483918708066724
11. Dong, M. W.; Prudhomme, D. et al. High-Resolution Separations of Complex Pharmaceuticals by UHPLC: Case Studies and Quality Control Implications. LCGC North Am. 2014, 32 (11), 868–76.
12. Agmon, N. The Grotthuss Mechanism. Chem. Phys. Lett. 1995, 244, 456–462. DOI: 10.1016/0009-2614(95)00905-J
13. Dong, M. W.; Miller, G.; Paul, R. MS-Compatible ICH Impurity Analysis with a High-pH Mobile Phase. J. Chromatogr. A, 2003, 987, 283-290. DOI: 10.1016/S0021-9673(02)01961-1
14. YouTube Channel: HPLC in Pharmaceutical Analysis Series, EP-7: Why Should You Use a 3-mm i.d. Column? https://youtu.be/J77WVUYx-Yw
15. YouTube Channel: HPLC in Pharmaceutical Analysis Series, EP-5: The “Best” HPLC Columns for Pharmaceutical Analysis. https://youtu.be/F8lJ5_Thdbg
16. Dong, M. W.; Huynh-Ba, K.; Ayers, J. T. Development of Stability-Indicating Analytical Procedures by HPLC: An Overview and Best Practices. LCGC North Am. 2020, 38 (8), 440–455.
17. Marchetto, A.; Besenhard, M. O. et al. In Silico HPLC Method Development via Machine Learning. Anal. Chem. 2025, 97 (13), 6991–7001. DOI: 10.1021/acs.analchem.4c03466
18. Snyder, L. R.; Kirkland, J. J.; Glajch, J. L. Practical HPLC Method Development, Wiley, 1997.