Home Examples Screenshots User manual Bluesky logo YouTube
OghmaNano Multiphysics simulation platform for optoelectronic devices and photonic systems DOWNLOAD Quick Start guide

The Einstein Relations in Semiconductor Device Physics

The Einstein relations connect the mobility and diffusion coefficients appearing in the drift–diffusion equations. They are fundamentally thermodynamic consistency conditions ensuring that the drift and diffusion currents exactly cancel at equilibrium. In semiconductor device simulation, the Einstein relations are essential for obtaining physically correct zero-current equilibrium solutions, stable heterojunction transport, and numerically robust drift–diffusion solvers.

This page introduces the Einstein relations directly from the drift–diffusion equations, then derives the generalized form from carrier pressure and Fermi–Dirac statistics. Particular emphasis is placed on the role of quasi-Fermi levels, heterojunction transport, and the thermodynamic origin of diffusion.

1. Drift–diffusion transport

In semiconductor transport theory, the carrier current is typically written as the sum of a drift term and a diffusion term. For electrons:

\[ \mathbf{J}_n = q\mu_n n\,\nabla E_c + qD_n\nabla n. \]

The two terms have distinct physical origins:

\[ \underbrace{ q\mu_n n\,\nabla E_c }_{\text{drift current}} \qquad \underbrace{ qD_n\nabla n }_{\text{diffusion current}} \]

The full derivation of the drift–diffusion equations from the Boltzmann Transport Equation is presented in:

Drift–Diffusion Equations Derivation from the Boltzmann Transport Equation

At equilibrium, these two current components must exactly balance. If they do not, the transport equations would predict finite current flow at zero applied voltage, violating thermodynamic equilibrium.

2. Classical Einstein relation

At equilibrium:

\[ \mathbf{J}_n = 0. \]

Therefore,

\[ q\mu_n n\,\nabla E_c + qD_n\nabla n = 0. \]

Rearranging:

\[ \nabla n = -\frac{\mu_n}{D_n} n\,\nabla E_c. \]

In the non-degenerate Maxwell–Boltzmann limit:

\[ n = N_c \exp\left( \frac{E_{Fn}-E_c}{k_B T} \right). \]

At equilibrium, the quasi-Fermi level is spatially constant:

\[ \nabla E_{Fn} = 0. \]

Taking the gradient gives:

\[ \nabla n = -\frac{n}{k_B T}\nabla E_c. \]

Substituting into the equilibrium current condition:

\[ q\mu_n n\,\nabla E_c - qD_n \frac{n}{k_B T} \nabla E_c = 0. \]

Since this must hold for arbitrary \(\nabla E_c\),

\[ D_n = \frac{\mu_n k_B T}{q}. \]

This is the classical Einstein relation.

An identical derivation for holes gives:

\[ D_p = \frac{\mu_p k_B T}{q}. \]

👉 Key point: The Einstein relation is the condition required for zero net current at equilibrium. If the diffusion coefficient and mobility are inconsistent, drift and diffusion will not cancel correctly.

3. Carrier pressure and the generalized Einstein relation

The derivation above assumes Maxwell–Boltzmann statistics. More fundamentally, the Einstein relation emerges from the carrier pressure term obtained from the momentum-balance equation derived from the Boltzmann Transport Equation.

As shown in the drift–diffusion derivation , the electron current may be written in the more general form:

\[ \mathbf{J}_n = q\mu_n n\,\nabla E_c + \mu_n \nabla p. \]

Here \(p\) is the carrier pressure:

\[ p = \frac{2}{3} \int (E-E_c)\, g(E)\, f(E)\, \mathrm{d}E. \]

In the non-degenerate limit:

\[ p = nk_B T, \]

recovering the classical diffusion term.

However, under degenerate conditions the carrier pressure no longer scales linearly with carrier density, and the Einstein relation must be generalized.

4. Generalized Einstein relation

For Fermi–Dirac statistics:

\[ n = N_c F_{1/2}(\eta), \]

where

\[ \eta = \frac{E_{Fn}-E_c}{k_B T}, \]

and \(F_j(\eta)\) is the complete Fermi–Dirac integral of order \(j\).

The carrier pressure becomes:

\[ p = nk_B T \frac{F_{3/2}(\eta)}{F_{1/2}(\eta)}. \]

Substituting this pressure into the momentum-balance-derived current equation yields:

\[ D_n = \frac{\mu_n k_B T}{q} \frac{F_{3/2}(\eta)}{F_{1/2}(\eta)}. \]

This is the generalized Einstein relation.

In the non-degenerate limit:

\[ \frac{F_{3/2}(\eta)}{F_{1/2}(\eta)} \rightarrow 1, \]

recovering the classical Einstein relation:

\[ D_n = \frac{\mu_n k_B T}{q}. \]

5. Thermodynamic form of the Einstein relation

A particularly important and fully general form of the Einstein relation is:

\[ D_n = \frac{\mu_n}{q} \frac{n}{\partial n / \partial E_{Fn}}. \]

This form directly exposes the thermodynamic origin of diffusion. The quantity

\[ \frac{\partial n}{\partial E_{Fn}} \]

is the electronic compressibility of the carrier gas.

Using the Fermi–Dirac density expression:

\[ n = N_c F_{1/2}(\eta), \]

one obtains:

\[ \frac{\partial n}{\partial E_{Fn}} = \frac{N_c}{k_B T} F_{-1/2}(\eta). \]

Substitution into the thermodynamic Einstein relation yields:

\[ D_n = \frac{\mu_n k_B T}{q} \frac{F_{1/2}(\eta)}{F_{-1/2}(\eta)}. \]

For parabolic bands, this expression is equivalent to the pressure-based form:

\[ D_n = \frac{\mu_n k_B T}{q} \frac{F_{3/2}(\eta)}{F_{1/2}(\eta)}. \]

The Einstein relation therefore fundamentally connects:

6. Importance in numerical device simulation

From a numerical perspective, the Einstein relations are essential for stable and physically correct semiconductor device simulation because they guarantee that the drift and diffusion currents exactly cancel at equilibrium, giving zero net current at zero applied voltage in the dark.

If the mobility and diffusion coefficients are inconsistent, the solver can produce spurious equilibrium currents, incorrect built-in potentials, and poor convergence near equilibrium.

In many practical drift–diffusion solvers, the free carrier states are represented using parabolic bands or Maxwell–Boltzmann-like statistics because this leads to analytically simple Einstein relations with excellent numerical behaviour. However, if a non-parabolic density of states is used for the free states, the generalized Einstein relation must instead be evaluated from

\[ D_n = \frac{\mu_n}{q} \frac{n}{\partial n / \partial E_{Fn}}, \]

where both the carrier density and its derivative may contain numerical noise. Since the drift and diffusion currents are obtained from the difference between two large terms, even small numerical errors in the generalized Einstein relation can prevent exact equilibrium cancellation and lead to finite current at zero bias. This is one of the main numerical reasons why many semiconductor transport models prefer analytically well-behaved carrier statistics wherever possible.

If full Fermi–Dirac statistics are enabled for the free carriers, the full generalized Einstein relation must be used. In OghmaNano, this is handled automatically by the transport solver.

👉 Practical interpretation: The Einstein relation is fundamentally a thermodynamic consistency condition. Its purpose is to ensure that the transport equations relax toward the correct equilibrium state with zero net current.

7. Summary

The Einstein relations connect mobility and diffusion in semiconductor transport and are fundamentally the conditions required for thermodynamic equilibrium in drift–diffusion theory. In the Maxwell–Boltzmann limit, the diffusion coefficient reduces to the familiar classical form proportional to mobility and temperature. More generally, under Fermi–Dirac statistics, the diffusion coefficient becomes dependent on the carrier statistics and electronic compressibility of the carrier gas. The Einstein relations therefore link carrier mobility, carrier statistics, carrier pressure, diffusion, and quasi-Fermi-level transport within a single, self-consistent framework. In practical semiconductor device simulation, they are essential for obtaining zero-current equilibrium, stable numerical convergence, physically correct heterojunction transport, and accurate quasi-Fermi-level transport.

👉 Key equations:

Classical Einstein relation:

\[ D = \frac{\mu k_B T}{q} \]

Generalized Einstein relation:

\[ D_n = \frac{\mu_n k_B T}{q} \frac{F_{3/2}(\eta)}{F_{1/2}(\eta)} = \frac{\mu_n}{q} \frac{n}{\partial n / \partial E_{Fn}} \]