Swan

Official Resources

• Repository: https://github.com/wushidonguc/swan
• License: MIT License

Overview

Swan is an open-source C++ software package designed for nanoscale quantum electron transport simulations. It employs the Non-Equilibrium Green's Function (NEGF) formalism coupled with a self-consistent Poisson solver to calculate the electrical characteristics of realistic nanodevices. Swan distinguishes itself by using Wannier functions as the basis set, which allows for accurate material representation while maintaining computational efficiency for large-scale atomistic simulations.

Scientific domain: Nanoelectronics, Device Physics, Quantum Transport Target user community: Device engineers and physicists modeling transistors and quantum structures

Theoretical Methods

• NEGF Formalism: Solves the Keldysh Green's functions for open quantum systems.
• Poisson Equation: Self-consistently solves for the electrostatic potential using a finite difference or finite element scheme.
• Wannier Basis: Uses Maximally Localized Wannier Functions (MLWFs) from Wannier90 to construct the device Hamiltonian.
• Schrödinger-Poisson: Iterative solution loop until convergence of charge density and potential.

Capabilities

• Device Simulation:
  • I-V characteristics of FETs (FinFET, Nanowire FET, TFET).
  • Charge density profiles.
  • Band diagrams under bias.
• Geometries:
  • 1D Nanowires.
  • 2D Ultrathin bodies / Ribbons.
• Materials:
  • Silicon, Germanium, III-V semiconductors.
  • Transition metal dichalcogenides (TMDs).

Key Strengths

• Efficiency: The use of Wannier functions provides a minimal basis set compared to plane waves or large Gaussian bases, enabling the simulation of larger devices (thousands of atoms).
• Parallelism: Efficient MPI parallelization for energy integration and bias points.
• Modularity: Object-oriented C++ design facilitates extension.

Inputs & Outputs

• Inputs:
  • Hamiltonian Files (_hr.dat, _xyz.dat from Wannier90).
  • Device configuration file (JSON/Input script).
• Outputs:
  • .dat files for current, density, and potential.
  • Visualization files for ParaView (VTK).

Interfaces & Ecosystem

• Upstream:
  • Wannier90: Essential for generating the material parameters.
  • DFT Codes: VASP, QE, etc., via Wannier90.
• Downstream:
  • ParaView: For 3D visualization of scalar fields (potential, density).

Performance Characteristics

• Speed: Optimized dense/sparse matrix operations.
• Scaling: Scales up to hundreds of cores for energy points.

Comparisons with Other Codes

• vs. NanoTCAD ViDES: ViDES is more comprehensive (includes drift-diffusion, Python interface); Swan is a focused C++ NEGF solver.
• vs. OMEN: OMEN is a high-performance HPC code; Swan is lighter and easier to deploy for smaller clusters.

Community and Support

• Development: Maintained by researchers (check GitHub contributors).
• Source: GitHub.

Verification & Sources

• Repository: https://github.com/wushidonguc/swan
• Verification status: ✅ VERIFIED
  • Active codebase.
  • Implements standard NEGF-Poisson loop.