Episode 68: Peak Fitting, Deconvolution & AI in Chromatography

Chromatography software hasn't fundamentally changed in decades — but the peaks it's trying to measure have gotten a lot harder to solve. In this episode of  @ChromatographyTalk  I sit down with Ron Brown, creator of PeakFit and PeakLab, and returning guest Dr. M. Farooq Wahab, analytical chemist and research engineering scientist at the University of Texas at Arlington, to dig into the real science (and art) of peak modeling.We cover why baseline subtraction and denoising matter before you ever touch a peak model, the difference between true mathematical deconvolution and the "deconvolution" most chromatographers mean when they say it, and why mass balance is the line between a valid peak model and bad science. Ron walks through the historical roots of peak modeling going back over a century, and explains how his HVL and generalized moment models build on pioneers whose work rarely gets the recognition it deserves.We also get into resolving completely co-eluting peaks using 3D PDA/DAD data, multivariate curve resolution, and derivative-based zero-crossing methods — techniques that go well beyond what a single chromatographic dimension can tell you. And we close with a candid conversation about generative AI's growing role in analytical chemistry: where it genuinely accelerates good science, where it's eroding foundational skills in newer chemists, and why Farooq argues AI is only as useful as the prepared mind using it.In this episode:- What peak modeling actually solves that chromatography alone can't- Why baseline correction and denoising come before any modeling step- The historical origins of peak fitting, from Carl Clarison to Van der Linden- True deconvolution vs. what the field commonly (and incorrectly) calls deconvolution- Why mass balance is non-negotiable in a valid peak model- Using the F-statistic and information criteria to choose the right model complexity- Resolving fully co-eluting peaks with 3D spectral data and MCR-ALS- Generative AI's expanding role in chromatographic data processing — and its risks for early-career scientistsAbout our guests:Ronald E. Brown is a chemical engineer trained at Purdue University with 36 years of experience developing chromatography and spectroscopy peak modeling software. He is the sole developer of PeakFit and PeakLab, used by spectroscopists, chromatographers, and defense researchers worldwide. His paper on detecting outliers in nonlinear regression using the false discovery rate has been cited more than 1,950 times.Dr. M. Farooq Wahab is an analytical chemist and research engineering scientist at the University of Texas at Arlington. He earned his PhD from the University of Alberta and focuses his research on data processing, non-linear denoising, deconvolution, and generative AI applications in analytical chemistry.🎧 Listen to more episodes of Concentrating on Chromatography: https://open.spotify.com/show/1wZg67rCjoGju4yhqxI3tL?si=n9GLz9zjSCOrVfXPxFPhng🔗 Learn more about Organomation's nitrogen blowdown evaporators and sample prep instruments: https://www.organomation.com/#Chromatography #PeakFitting #AnalyticalChemistry #MassSpectrometry #Deconvolution #ChemometricsAI #HPLC #LabScience