Official Resources
- Homepage: https://tb2j.readthedocs.io/
- Documentation: https://tb2j.readthedocs.io/en/latest/
- Source Repository: https://github.com/mailhexu/TB2J
- License: BSD-2-Clause
Overview
TB2J is an open-source Python package designed to calculate magnetic interaction parameters—specifically the isotropic Heisenberg exchange ($J$), anisotropic exchange, and Dzyaloshinskii-Moriya interaction (DMI)—from first-principles density functional theory (DFT) data. It utilizes the Green's function method combined with the Magnetic Force Theorem, allowing for the extraction of these parameters from a single unit-cell calculation without the need for computationally expensive supercell approaches. TB2J operates within a localized basis set framework, supporting both Maximally Localized Wannier Functions (MLWFs) and Linear Combination of Atomic Orbitals (LCAO).
Scientific domain: Magnetism, Spintronics, Topological Magnets, 2D Materials
Target user community: Researchers studying magnetic materials, magnon transport, and spin dynamics
Theoretical Methods
- Magnetic Force Theorem: Treats small rotation of spins as a perturbation to the total energy.
- Green's Function Approach: Calculates inter-site magnetic interactions by integrating the Green's function, avoiding supercell total energy differences.
- Localized Basis Sets:
- Wannier Functions: Via Wannier90 (for plane-wave codes like VASP, QE).
- LCAO: Native support for numerical atomic orbitals (SIESTA, OpenMX, ABACUS).
- Heisenberg Hamiltonian: Maps DFT energetics to:
$$H = -\sum_{i<j} J_{ij} \mathbf{S}i \cdot \mathbf{S}j + \sum{i<j} \mathbf{D}{ij} \cdot (\mathbf{S}_i \times \mathbf{S}_j) + \dots$$
Capabilities
- Interaction Parameters:
- Isotropic exchange ($J$).
- Antisymmetric exchange / Dzyaloshinskii-Moriya Interaction (DMI, $\mathbf{D}$).
- Anisotropic exchange tensor ($\Gamma$).
- Biquadratic exchange (experimental).
- Post-Processing:
- Magnon band structure calculation.
- Spin-wave stiffness.
- Curie temperature estimation (Mean Field Approx).
- Data Generation:
- Input files for spin dynamics codes (Vampire, Multibinit, UppASD).
- Visualization of magnetic interactions.
Key Strengths
- Efficiency: Computes full distance-dependent interactions from a single unit-cell ground state calculation.
- Versatility: Works with almost all major DFT codes via Wannier90 or native interfaces.
- Spin-Orbit Coupling: Fully supports non-collinear calculations for extracting DMI and anisotropy.
- Automation: User-friendly Python interface simplifies the workflow from DFT to magnetic modeling.
Inputs & Outputs
- Inputs:
- Wannier90 outputs (
*_hr.dat, *.win, eig files).
- DFT Hamiltonian/Overlap matrices (for SIESTA/OpenMX/ABACUS).
- Atomic structure and magnetic moments.
- Outputs:
- Text files with $J_{ij}$ and $\mathbf{D}_{ij}$ lists.
exchange.xml (standard format).- Input files for Multibinit, Vampire, UppASD.
- Plotting scripts for interactions.
Interfaces & Ecosystem
- Native Interfaces:
- SIESTA: Reads
.HS and .DE files directly (siesta2J.py).
- OpenMX: Reads
.scfout and Hamiltonian files (openmx2J.py).
- ABACUS: Uses LCAO output files (
abacus2J.py).
- Wannier90 Interface:
- Supports VASP, Quantum ESPRESSO, Abinit, HK and any code compatible with Wannier90 (
wannier90J.py).
- Downstream Integration:
- Vampire: Atomistic spin dynamics.
- Multibinit: (Abinit project) Lattice dynamics and spin dynamics.
- UppASD: Atomistic spin dynamics.
Performance Characteristics
- Speed: The Green's function integration is lightweight; the dominant cost is the preceding DFT/Wannierization step.
- Scaling: Efficient for standard unit cells; cost scales with the number of basis functions (orbitals).
- Parallelization: Python multiprocessing supported for k-point integration.
Limitations & Known Constraints
- Mott Insulators: Requires a valid DFT ground state; for strongly correlated systems (DFT+U), accurate U values are critical.
- Metals: RKKY interactions are captured, but convergence with k-points must be carefully checked.
- Basis Quality: Accuracy depends entirely on the quality of the Wannierization or the LCAO basis set.
Comparison with Other Codes
- vs. Total Energy Mapping: Standard approach requires many large supercells to fit J values; TB2J uses a single cell and is much faster/less error-prone.
- vs. Lichtenstein formula codes: TB2J implements a variant of the Lichtenstein formula but generalizes to non-orthogonal/Wannier bases and DMI.
- vs. SPR-KKR: KKR methods also use Green's functions but are all-electron and muffin-tin based; TB2J works with standard pseudopotential codes.
Application Areas
- 2D Magnets: CrI3, Fe3GeTe2, etc. - studying distance-dependent exchange and thickness dependence.
- Topological Magnets: Skyrmion-hosting materials (DMI calculation).
- Spintronics: Antiferromagnets and complex magnetic textures.
Community and Support
- Source: Hosted on GitHub.
- Documentation: ReadTheDocs.
- Forum: Issues and discussions on GitHub.
- Tutorials: Examples provided for all supported interfaces.
Verification & Sources