4.10 Extended Tight-Binding Methods (xTB)4.10 Extended Tight-Binding Methods (xTB)4.10.2 Job Control

4.10.1 Introduction

(July 4, 2026)

The xTB family of semiempirical tight-binding methods offers an efficient route to low-cost electronic-structure calculations for large molecular systems. The present Q-Chem implementation is based on the GFN2-xTB Hamiltonian of Bannwarth, Ehlert, and Grimme, ⁷³ Bannwarth C., Ehlert S., Grimme S.
J. Chem. Theory Comput.
(2019), 15, pp. 1652. Link which combines an extended Hückel-like zeroth-order Hamiltonian with self-consistent isotropic and anisotropic electrostatic and exchange-correlation contributions. For open-shell systems, Q-Chem also supports an unrestricted spin-polarized variant in the spirit of the spGFN2-xTB developments described in Refs. ⁹⁸⁶ Neugebauer H. et al.
J. Comput. Chem.
(2023), 44, pp. 2120. Link , ⁹⁵⁹ Moradi S. et al.
J. Comput. Chem.
(2024), 45, pp. 2786. Link , and ⁹⁵⁸ Moradi S. et al.
J. Comput. Chem.
(2026), 47, pp. e70346. Link .

4.10.1.1 Theoretical Background

GFN2-xTB belongs to the density-functional tight-binding (DFTB) family, in which the Kohn–Sham DFT energy is expanded around a superposition of neutral atomic reference densities,

\rho_{0}(\mathbf{r})=\sum_{A}\rho_{A}^{0}(\mathbf{r}),

(4.77)

with the density fluctuation defined as $\delta\rho=\rho-\rho_{0}$ . In schematic form, the energy expansion is

E[\rho]=E^{(0)}[\rho_{0}]+E^{(1)}[\rho_{0},\delta\rho]+E^{(2)}[\rho_{0},(% \delta\rho)^{2}]+E^{(3)}[\rho_{0},(\delta\rho)^{3}]+\cdots.

(4.78)

The zeroth-order term defines the reference atomic energies together with a short-range repulsion model, the first-order term yields the extended Hückel-like valence Hamiltonian, and the higher-order terms account for the self-consistent response to charge-density fluctuations. ⁷³ Bannwarth C., Ehlert S., Grimme S.
J. Chem. Theory Comput.
(2019), 15, pp. 1652. Link In GFN2-xTB, the higher-order response includes isotropic shell-charge terms as well as anisotropic multipole corrections.

The working GFN2-xTB energy expression is ⁷³ Bannwarth C., Ehlert S., Grimme S.
J. Chem. Theory Comput.
(2019), 15, pp. 1652. Link

E_{\mathrm{GFN2-xTB}}=E_{\mathrm{rep}}+E_{\mathrm{disp}}+E_{\mathrm{EHT}}+E_{% \mathrm{IES}}+E_{\mathrm{IXC}}+E_{\mathrm{AES}}+E_{\mathrm{AXC}}+G_{\mathrm{% Fermi}}.

(4.79)

Here, $E_{\mathrm{rep}}$ is a classical short-range repulsion term, $E_{\mathrm{EHT}}$ is the extended Hückel covalent-bonding contribution, $E_{\mathrm{IES}}$ and $E_{\mathrm{IXC}}$ are the isotropic electrostatic and exchange-correlation terms, $E_{\mathrm{AES}}$ and $E_{\mathrm{AXC}}$ are the anisotropic multipole electrostatic and exchange-correlation corrections, and $G_{\mathrm{Fermi}}$ is the finite-temperature entropy term that allows fractional occupations. In the original standalone GFN2-xTB model, $E_{\mathrm{disp}}$ is the self-consistent D4 dispersion contribution. For open-shell systems, the spin-polarized extension adds a shell-wise spin-polarization energy based on Mulliken spin populations and atomic spin constants. ⁹⁸⁶ Neugebauer H. et al.
J. Comput. Chem.
(2023), 44, pp. 2120. Link

This implementation is intended for rapid screening, structure refinement, and inexpensive single-point calculations. In practice it is most useful as a fast preoptimization or exploratory tool before higher-level calculations are carried out.

Note: The present Q-Chem implementation does not include the D4 dispersion contribution that is part of the original standalone GFN2-xTB model. Consequently, total energies and optimized structures can differ from those obtained with the reference xtb program, especially for systems where dispersion interactions are important.

Search Results

4.10.1 Introduction

4.10.1.1 Theoretical Background