Computational Methodology
How Qchemvyx implements DFT calculations: basis set handling, SCF convergence, integration grids, dispersion corrections, geometry optimization, and thermochemical corrections.
1. Electronic Structure Solver
Qchemvyx uses a custom implementation of the Kohn-Sham DFT equations with real-space integration. The Hamiltonian is solved iteratively using the DIIS (Direct Inversion in the Iterative Subspace) algorithm, with convergence threshold set to ΔE < 10⁻⁸ Hartree per cycle by default (tightening to 10⁻¹⁰ available).
Exchange-correlation functionals are evaluated using a numerical integration grid. Default: Becke pruning scheme, (75,302) grid points (75 radial, 302 angular per atom). High accuracy mode: (99,590).
2. Basis Set Handling
All contracted Gaussian basis sets are implemented from the EMSL Basis Set Exchange library. Linear dependencies are removed via canonical orthogonalization with threshold 10⁻⁶. For elements heavier than Ar, effective core potentials (ECPs) from Stuttgart/Cologne are applied automatically for the core electrons when the basis supports them (e.g., def2 family).
Default basis set for screening runs: 6-31G*. Default for geometry optimizations and energy calculations: 6-311G++. For high-accuracy single-point energies: aug-cc-pVTZ.
3. Dispersion Corrections
Grimme D4 atom-pairwise dispersion correction is applied by default for all GGA and hybrid functionals. D4 charges are computed from Mulliken population analysis. The three-body Axilrod–Teller–Muto term is included for systems ≥ 30 atoms. D3 is available as an alternative.
4. Geometry Optimization
L-BFGS geometry optimizer in redundant internal coordinates (Pulay-Fogarasi). Convergence criteria: max force < 4.5×10⁻⁴ au, RMS force < 3.0×10⁻⁴ au, max displacement < 1.8×10⁻³ au. Analytical gradients from the DFT energy expression. MMFF94 pre-optimization provides starting geometry for all 3D-generated structures.
5. Thermochemical Corrections
Analytical Hessian calculation at the optimized geometry. Vibrational frequencies from mass-weighted Hessian diagonalization. Thermal corrections under rigid-rotor / harmonic oscillator (RRHO) approximation. Zero-point energy, enthalpy (H), and Gibbs free energy (G) at 298.15 K and 1 atm standard state.
Quasi-RRHO treatment (Grimme) for low-frequency modes ≤ 100 cm⁻¹ to correct for anharmonic contributions.
6. Solvation
Implicit solvation via SMD (Solvation Model D, Marenich–Cramer–Truhlar) or PCM (IEFPCM). Cavity construction by UAKS radii with solvent-dependent scaling. Electrostatic and non-electrostatic contributions (cavitation, dispersion, repulsion) computed per the SMD parameterization.
A note on accuracy and interpretation
DFT accuracy depends on functional and basis set selection; no single method is universally optimal across all chemical systems. Qchemvyx provides functional recommendations based on the reaction class and metal center involved, but the chemist makes the final methodology decision. Calculated activation energies and reaction energies are useful for relative ranking and down-selection — they are not substitutes for experimentally measured rate constants, yields, or selectivity data. Qchemvyx accelerates the computational screening phase. Experimental validation remains the authority on whether a candidate catalyst performs as predicted.
Users running calculations on open-shell transition metal systems should verify spin state assignments independently. Broken-symmetry DFT calculations (for antiferromagnetically coupled systems or biradicals) require additional care; see our guide to broken-symmetry DFT setup for detailed guidance.