Quantum Transport Simulation with Non-Equilibrium Green's Functions

82 concepts75 papers1 proposal

When electronic devices shrink to the nanometer scale, the classical laws of physics lose their grip. In a macroscale copper wire, electrons move like a fluid experiencing friction, governed by drift-diffusion equations. But in a sub-5 nm transistor channel or a single-layer two-dimensional (2D) material, electrons behave as waves. They tunnel through energy barriers, confine themselves into discrete quantum states, and shoot across channels ballistically without scattering.

To engineer next-generation nanoelectronics, we need a simulation framework that embraces this quantum reality. As highlighted by Samuel Poncé in a review of first-principles modeling, traditional semiclassical transport equations fail to capture these non-local quantum phenomena. The modern gold standard for solving this problem is the Non-Equilibrium Green's Function (NEGF) formalism.

NEGF provides a rigorous mathematical bridge between the microscopic quantum structure of a material and the macroscopic currents measured in a lab. Here is the intuition behind how it works, why it is necessary, and how it handles the complex physics of open, driven quantum systems.

The Challenge of the Open System

Standard quantum mechanics is typically taught using closed systems. You take a particle in an isolated box, define a Hamiltonian $H$, and solve the time-independent Schrödinger equation:

$$H\psi = E\psi$$

This yields discrete, localized energy eigenvalues. But an electronic device—like a field-effect transistor (FET)—is fundamentally an open system. It consists of a central channel connected to two macroscopic, semi-infinite metallic reservoirs: a source and a drain. Electrons are continuously injected from one lead, travel through the channel, and escape into the other under an applied voltage bias.

Trying to simulate an infinite number of atoms in the leads is computationally impossible. Conversely, treating the channel as a closed box misses the very essence of conduction: states must broaden, and electrons must flow out.

NEGF elegantly solves this dilemma by partitioning the universe into three parts: the Left Lead ($L$), the Central Channel ($C$), and the Right Lead ($R$). Through a mathematical technique called partitioning, the infinite degrees of freedom of the leads are projected out and folded directly onto the boundaries of the central channel.

Instead of solving an infinite Schrödinger equation, we solve for the Green's function $G(E)$ of the central channel:

$$G(E) = \left[ E \cdot I - H_C - \Sigma_L(E) - \Sigma_R(E) \right]^{-1}$$

Here, $H_C$ is the Hamiltonian of the isolated channel, and $I$ is the identity matrix. The magic lies in $\Sigma_L$ and $\Sigma_R$, which are known as self-energies.

Understanding Self-Energies and Broadening

A self-energy is not a physical energy value; it is a complex matrix that acts as an effective boundary condition. It captures two distinct physical effects that the leads exert on the channel:

  1. Energy Shifting (Real Part): The real part of $\Sigma$ shifts the energy levels of the channel because the channel electrons feel the electrostatic presence of the leads.
  2. State Broadening (Imaginary Part): The imaginary part of $\Sigma$ represents the lifetime of an electron inside the channel. Because an electron can escape into the leads, its quantum state no longer has an infinite lifetime. According to the Heisenberg uncertainty principle, a finite lifetime ($\Delta t$) causes a broadening of the state's energy ($\Delta E$).

We quantify this broadening by defining broadening matrices, $\Gamma_{L,R} = i \left[ \Sigma_{L,R} - \Sigma_{L,R}^\dagger \right]$.

Once we have the Green's function $G(E)$ and the broadening matrices, we can calculate how easily an electron at a specific energy $E$ can traverse the channel. This is described by the transmission function $T(E)$:

$$T(E) = \text{Tr} \left[ \Gamma_L(E) G(E) \Gamma_R(E) G^\dagger(E) \right]$$

Using the Landauer-Büttiker formalism, the total current is found by integrating this transmission function across the energy window opened by the bias voltage, weighted by the probability that the source has an electron available and the drain has an empty state. The utility of this approach was demonstrated by Shijie Hu et al., who derived exact analytical solutions for electron transmission to reveal how lattice orientation dictates conductivity and noise in graphene sheets.

The Self-Consistent Electrostatic Loop

An electron moving through a transistor does not travel through a static landscape. As electrons crowd into the channel, their negative charges repel one another, altering the local electrostatic potential. Furthermore, the voltage applied to an external gate electrode is designed to push the channel's energy levels up or down to turn the device ON or OFF.

Therefore, quantum transport cannot be calculated in a single step. It requires a self-consistent loop between quantum mechanics (NEGF) and electromagnetism (Poisson's equation).

               +------------------------+
               |  Poisson's Equation    |
               |  Finds Potential V(r)  |
               +-----------+------------+
                           |
                           | Electrostatic Potential V(r)
                           v
               +------------------------+
               |     NEGF Solver        |
               |  Finds Charge Density  |
               +-----------+------------+
                           |
                           | Quantum Charge Density rho(r)
                           +------------------------+
  1. Start with a Guess: We assume an initial electrostatic potential $V(r)$ across the device.
  2. Build the Hamiltonian: We add this potential to the material's intrinsic Hamiltonian: $H = H_0 + V(r)$.
  3. Solve NEGF: We compute the Green's function $G(E)$ and integrate it to find the spatial quantum charge density $\rho(r)$ inside the channel.
  4. Update Electrostatics: We plug $\rho(r)$ into Poisson's equation, $\nabla \cdot (\epsilon \nabla V) = -\rho$, along with the boundary conditions set by the gate, source, and drain voltages, to calculate a new potential $V(r)$.
  5. Iterate: We repeat this process until the potential and charge density stop changing.

As shown by Anuja Chanana et al., this self-consistent Poisson-Schrödinger/NEGF coupling is essential for accurately predicting device performance metrics like ON/OFF current ratios in hybrid nanoribbons. Similarly, Madhuchhanda Brahma et al. relied on this self-consistent loop to evaluate whether monolayer chalcogenide tunnel FETs could scale down to 8 nm while maintaining electrostatic control. To make these iterations computationally feasible for complex geometries, Hamilton Carrillo-Nuñez et al. demonstrated the utility of solving the NEGF equations in a mode-space representation rather than a full real-space grid, drastically reducing the matrix sizes without sacrificing quantum accuracy.

Bridging Chemistry and Transport: Ab Initio NEGF

To simulate a real device, where do we get the initial, unperturbed Hamiltonian $H_0$?

Historically, researchers used simplified models like the effective mass approximation, which treats electrons as simple waves moving through a uniform background. However, as Mathieu Luisier et al. demonstrated, simplified approximations miss critical multi-band interactions and atomic-scale variations. To capture the true atomistic skeleton of a device, NEGF is combined with Density Functional Theory (DFT) or Density Functional Tight-Binding (DFTB).

This ab initio (from first principles) NEGF approach allows researchers to simulate how real-world chemical imperfections alter transport. For example, Jón Steinar Garðarsson Mýrdal et al. combined DFT and NEGF to model electron movement across the interfaces of lithium-air batteries, revealing how the growth of insulating films triggers performance limits.

When applied to electronics, this pairing allows for the precise mapping of defects and structural variations:

To scale these computationally demanding simulations to screen hundreds of materials, Cedric Klinkert et al. developed a multi-scale pipeline. Their approach uses DFT to calculate the bulk electronic structure, maps those states onto localized Wannier functions to generate highly efficient tight-binding Hamiltonians, and then feeds them into an NEGF solver to simulate large-scale quantum transport.

Beyond Ballistic Flow: Scattering and Heat

In a perfectly pristine, ultra-short channel, electrons travel ballistically like light down an optical fiber. But in reality, electrons collide with phonons (lattice vibrations). This scattering destroys the phase coherence of the electron waves and generates heat, a major bottleneck in modern nanoelectronics.

NEGF handles scattering by introducing an additional self-energy term: the scattering self-energy ($\Sigma_{scat}$). Just as the lead self-energies represent the coupling to an external environment of states, the scattering self-energy represents an electron's coupling to an environment of phonons.

                  +----------------------------------+
                  |    Lead Self-Energies (\Sigma)   |
                  |  Models escape into reservoirs   |
                  +----------------+-----------------+
                                   |
                                   v
  +--------------------------------+--------------------------------+
  |                      Central Channel Green's Function           |
  |                                 G(E)                            |
  +--------------------------------+--------------------------------+
                                   ^
                                   |
                  +----------------+-----------------+
                  | Scattering Self-Energy (\Sigma_s) |
                  |   Models electron-phonon collisions  |
                  +----------------------------------+

Accounting for these interactions requires immense computational care. Jonathan Backman et al. demonstrated that by incorporating non-diagonal entries in these scattering self-energies, the NEGF framework can perfectly reconcile with the classical Linearized Boltzmann Transport Equation (LBTE) for bulk materials, while maintaining the ability to simulate out-of-equilibrium nanoscale architectures.

Furthermore, this allows for the simultaneous modeling of electrical and thermal dynamics. Christian Stieger et al. and Sara Fiore et al. implemented frameworks that self-consistently solve the Schrödinger and Poisson equations for electrons alongside a dynamical equation for phonons. This enables researchers to observe atomic-level self-heating and phonon generation directly as current flows through a device.

Accelerating the Formalism

The primary drawback of the NEGF formalism is its severe computational cost; calculating large matrix inversions at every energy point and every step of a self-consistent loop is incredibly expensive.

To bypass this bottleneck, Yike Huang introduced the DeePTB-NEGF framework. By training deep learning models to predict the self-consistent Hamiltonians directly, this approach replaces expensive iterative field calculations with neural network inferences. This achieves a 100x to 1000x speedup, opening the door to simulating realistic, large-scale nanoelectronic devices without sacrificing the foundational quantum accuracy that makes NEGF work.

History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_quantum_transport_simulation_16bc1305,
  title = {quantum transport simulation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/quantum-transport-simulation/rdr_16bc1305ec1d4e139d67fef41229cc7f},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-22},
}

This exact version (version 1, frozen text):

@misc{lacuna_quantum_transport_simulation_16bc1305_v1,
  title = {quantum transport simulation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/quantum-transport-simulation/rdr_16bc1305ec1d4e139d67fef41229cc7f/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-22},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake