Skip to content
Back to search results

Finite-Difference Time-Domain

Introduced by
Kane Yee
Year
1966
Type
Finite-difference algorithm
Family
Computational
Font size

What does it do?

The Finite-Difference Time-Domain method is a computational technique for solving Maxwell's equations by discretizing space and time on a grid.

Introduced by Kane Yee in 1966, FDTD is a foundational approach in computational electrodynamics and optical simulation, enabling direct modeling of electromagnetic wave propagation through complex media.

The intuition behind the method

FDTD works by replacing continuous electromagnetic fields with discrete values at grid points, and time with discrete steps. At each time step, the electric and magnetic fields are updated using finite-difference approximations of Maxwell's equations. The staggered Yee grid ensures numerical stability and physical consistency. This explicit time-stepping approach naturally captures transient phenomena and nonlinear interactions, making it ideal for simulating optical devices, antennas, and metamaterials.

History

Kane Yee introduced FDTD in 1966 as a direct numerical scheme for Maxwell's equations. In the 1980s–1990s, Allen Taflove popularized FDTD with Perfectly Matched Layer (PML) absorbing boundaries, enabling efficient open-domain simulations. FDTD became the standard in computational photonics, metamaterial design, and nanophotonics. Modern variants include subgrid methods, nonlinear FDTD, and GPU-accelerated implementations.

When to use it

FDTD is preferred for time-dependent, broadband, or nonlinear optical problems, especially with complex geometries or dispersive media. It is less efficient for very high frequencies or high-Q cavities where frequency-domain methods excel. Choose FDTD when you need transient dynamics, harmonic generation, or detailed near-field information.

Limitations and considerations

  • Computational cost scales as O(N^4) for a 3D grid of size N on a side, with many time steps needed for narrow-band analysis
  • Memory requirements can be prohibitive for large-scale 3D simulations
  • Stability criterion requires small time steps, limiting efficiency for certain problems
  • Grid dispersion and numerical anisotropy require fine mesh resolution to minimize errors

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Direct simulation of Maxwell's equations without approximation in the time domain
  • Natural handling of nonlinear materials and transient phenomena
  • Versatile application to arbitrary geometries and inhomogeneous media
  • Straightforward parallelization and GPU acceleration
  • Stable and explicit integration scheme with low per-timestep cost

Applications

  • Microwave and millimeter-wave antenna design and electromagnetic compatibility analysis
  • Optical waveguide and photonic crystal device simulation
  • Metamaterial and plasmonic resonator design and characterization
  • Bioelectromagnetics and specific absorption rate (SAR) assessment in medical devices
  • Terahertz spectroscopy and imaging system design

How it works

How it works

1

Divide the computational domain into a uniform or non-uniform grid with spacing Δx, Δy, Δz, and choose a time step Δt satisfying the Courant stability condition.

2

Set initial conditions for the electric field E and magnetic field H at grid points, typically zero or a specified source profile.

3

Update E and H fields using finite-difference equations derived from the curl form of Maxwell's equations, leapfrogging in time to maintain accuracy.

4

Apply boundary conditions (Dirichlet, Neumann, or absorbing such as PML) to prevent unphysical reflections and model open domains.

5

Extract field snapshots or integral quantities such as power flow, mode profiles, or transmission/reflection spectra from the transient solution.

Common pitfalls

  • Violating the Courant stability condition (Δt > Δx/c) leads to exponential growth and blow-up
  • Underresolved features smaller than the grid spacing will be smeared or ignored
  • Improper boundary conditions can introduce artificial reflections or edge modes
  • Insufficient domain size causes wrap-around errors or early field saturation

Frequently asked questions

Why is the Yee grid staggered, and why does it matter?

The Yee grid staggers E and H fields by half a cell, so E components are evaluated at cell edges and H at cell centers. This arrangement naturally satisfies the divergence-free conditions of Maxwell's equations and ensures second-order accuracy in both space and time, avoiding spurious modes and maintaining energy conservation.

What is the Courant criterion, and how do I choose a safe time step?

The Courant criterion states Δt ≤ Δx / (c √3) for 3D (or Δx / (c √2) for 2D, Δx / c for 1D) where Δx is the smallest grid spacing and c is the light speed in the medium. A safe choice is Δt = 0.99 × Δx / (c √3). Smaller time steps increase accuracy but slow the simulation; larger steps risk instability.

When should I use absorbing boundaries (PML) instead of metal walls?

Use PML to model open or semi-infinite domains where you want outgoing waves to exit without reflection. Use metal walls (perfect electric conductor) to model closed cavities or shielding. PML adds computational cost but is essential for modeling antennas, free-space radiation, and scattering from isolated objects.

How do I handle dispersive materials like glasses or metals?

Dispersive materials require frequency-dependent permittivity ε(ω). In FDTD, use auxiliary differential equations (ADE) to track polarization or magnetization currents, or use frequency-dependent conductivity (Drude models). These approaches add complexity but enable accurate simulation of chromatic dispersion and plasmonic resonances.

Related methods

Sources

  1. Yee, K. S. (1966). Numerical solution of initial boundary value problems involving Maxwell's equations in isotropic media. IEEE Transactions on Antennas and Propagation, 14(3), 302-307.

    Open DOI reference
    Copied.
  2. Taflove, A., & Hagness, S. C. (2005). Computational Electrodynamics: The Finite-Difference Time-Domain Method (3rd ed.). Artech House.

    Open publication reference
    Copied.
  3. Sullivan, D. M. (2000). Electromagnetic simulation using the FDTD method. IEEE Press.

    Open publication reference
    Copied.