Skip to main content

ChemistryLab.jl

A Julia numerical laboratory for computational chemistry, from formulas to a certified equilibrium: thermodynamic databases, activity models up to Pitzer, solid solutions, surfaces, kinetics and heat. Initially aimed at cement chemistry.

Julia
Chemistry
Materials Science

ChemistryLab.jl carries a chemical problem from its written formulas to its equilibrium state. Formulas, species, reactions and stoichiometric matrices are first-class objects, thermodynamic databases are imported in the formats in common use, and equilibrium is computed by Gibbs energy minimization and returned with a certificate stating what it proves. Every quantity carries its unit, and exact ForwardDiff derivatives pass through equilibria, kinetic runs and recipes. Initially focused on cement chemistry — CEM I to CEM V entered from an oxide analysis — its scope is that of aqueous and mineral systems at large. Developed with Anthony Soive and released under LGPL v2.1+. Part of the MicroPoroChemoMechanics ecosystem.

Key features

  • Formulas, species and reactions — chemical formulas with charge, Unicode and PHREEQC notations; Species and CemSpecies for solution and solid phases; stoichiometric matrices built automatically
  • Thermodynamic databases — databases of reactions (PHREEQC, The Geochemist’s Workbench, EQ3/6) and of formation properties (JSON and YAML formats) imported with the activity model they were fitted with; none is stored in the package: the freely redistributable ones are downloaded from their publisher on first use and checked against a SHA-256, the others installed by hand under their publisher’s terms
  • Certified equilibrium — the certificate says what it proves: global minimum, KKT point or self-consistent speciation, each element balance judged against its own total
  • Activity models — B-dot, Davies, Truesdell–Jones, SIT and Pitzer; SIT and Pitzer solutes solved by Newton, halite solubility within 0.3 % of its measurement
  • Solid solutions — ideal, Redlich–Kister, regular, sublattice (C–N–A–S–H) and compound-energy (CASH+) models, miscibility gaps detected and located
  • Surfaces and interfaces — surface complexation, constant-capacitance and diffuse-layer electrostatics, cation exchange and a Donnan layer, checked against PHREEQC
  • Recipes, kinetics, heat — binders blended, hydrated, carbonated or leached as recipes; Parrott–Killoh and Waller rate laws; isothermal and semi-adiabatic calorimetry; for a host code, an implicit kinetic step at each node, certified or refused with its cause

Installation

ChemistryLab.jl is registered in Julia’s General registry. Solving an equilibrium needs an optimization backend, loaded through a package extension; OptimaSolver.jl is the default one, and the one the certificate needs:

using Pkg
Pkg.add(["ChemistryLab", "OptimaSolver"])
using ChemistryLab, OptimaSolver         # OptimaSolver: the default equilibrium backend

Optimization with OptimizationIpopt remains available as a backend. A database is downloaded once, on first use; fetch_databases() prepares a machine for offline work.