Isotherm Fitter

Paste pure-component adsorption equilibrium data, fit a library of isotherm models by Levenberg–Marquardt with multi-start, and rank them by AICc (which penalizes parameter count — lowest SSE alone would always pick the model with the most parameters). Multi-temperature data can be fitted per temperature or globally with van ’t Hoff affinities b(T) = b₀e−ΔH/RgT, giving −ΔH and the model-based isosteric heat. Fitted parameters export in the format used across this suite (IAST, Uptake Curves, chromatography apps). notation ↗
Paste data or load an example to begin.
Objective: weighted least squares on q(p) (weighting selectable); Levenberg–Marquardt on logit-bounded parameters with heuristic + randomized multi-start. AICc = N ln(SSEw/N) + 2k + 2k(k+1)/(N−k−1) ranks models with a parameter-count penalty; confidence intervals from the Jacobian at the optimum (±1.96σ), with a warning when parameters are >98% correlated (over-parametrized fit). Global multi-T mode: qs temperature-independent, affinities b = b₀e−ΔH/RgT with −ΔH positive for exothermic adsorption; isosteric heat from the fitted model via Clausius–Clapeyron. Validated against scipy.optimize.least_squares on seeded synthetic data (see README).
Works offline — single file, share freely. Built by Arvind Rajendran with Claude Code · v1.0 · August 2026 · educational & research use · no warranty — use at your own risk
Runs entirely in your browser, no data collected · © 2026 · CC BY 4.0 · Part of the Research Apps suite →