Ph · Physics

How Do Boundary Conditions Affect the Density of States?

Published By the 9to5Science editorial team

At a glance

  1. 1. Quantize allowed states

    Boundary conditions restrict which wave numbers and energies are allowed in a finite system.

  2. 2. Count states below energy

    Find the number of eigenstates with energy at most E, including any degeneracies.

  3. 3. Differentiate the count

    The density of states is the derivative of the count, or a set of delta peaks for an exact finite spectrum.

  4. 4. Separate bulk and boundary

    Compare systems at the same volume; boundary-dependent corrections usually become small per unit volume as the system grows.

Boundary conditions specify how a wavefunction behaves at the edge of a system. They therefore affect which energy eigenstates are allowed. Since the density of states (DOS) counts those energy levels, it generally depends on the boundary conditions in a finite system.

The important distinction is between the finite-system spectrum and the bulk DOS per unit volume. The former can change noticeably when you change the boundary conditions. For many large systems, the leading bulk DOS is the same, while boundary-dependent corrections become relatively small.

What the density of states counts

For discrete energy levels EnE_n, the exact DOS can be written as a sum of delta functions:

g(E)=∑nδ(E−En).g(E)=\sum_n \delta(E-E_n).

If an energy level is degenerate, it appears once for each independent state, or equivalently its delta peak has a weight equal to the degeneracy. In practice, one often smooths or averages this spectrum over a small energy interval. The cumulative number of states at or below energy EE is

N(E)=∑nΘ(E−En),N(E)=\sum_n \Theta(E-E_n),

where Θ\Theta is the step function. For a smoothed spectrum, g(E)=dN/dEg(E)=dN/dE.

Boundary conditions enter through the eigenvalue problem: they determine the allowed wavefunctions and hence the values of EnE_n. Examples include periodic boundaries, where the wavefunction repeats across opposite faces, and hard-wall (Dirichlet) boundaries, where it vanishes at the walls.

Worked example: a free particle in one dimension

Take a particle of mass mm in a segment of length LL. Its energy is E=ℏ2k2/(2m)E=\hbar^2 k^2/(2m).

With periodic boundary conditions, the allowed wave numbers are

kn=2πnL,n=0,±1,±2,…k_n=\frac{2\pi n}{L},\qquad n=0,\pm1,\pm2,\ldots

With hard walls at both ends, the wavefunctions are standing waves and

kn=πnL,n=1,2,3,…k_n=\frac{\pi n}{L},\qquad n=1,2,3,\ldots

These are different lists of levels. For example, the periodic system has a zero-momentum state, while the hard-wall system does not. At low energies, that distinction can be significant.

To compare the number of states below a large wave number kk, count the allowed values. For periodic boundaries, the positive and negative values of nn both count as states; for hard walls, the positive integers count the standing-wave states. In either case, the leading count is

N(k)≃Lkπ.N(k)\simeq \frac{Lk}{\pi}.

Using k=2mE/ℏk=\sqrt{2mE}/\hbar gives

N(E)≃Lπℏ2mE,g(E)≃Lπℏm2E.N(E)\simeq \frac{L}{\pi\hbar}\sqrt{2mE}, \qquad g(E)\simeq \frac{L}{\pi\hbar}\sqrt{\frac{m}{2E}}.

This is the leading, smoothed result for one spinless particle species. The exact finite-size counts are step functions with details that depend on the boundary condition. The approximate DOS above is the same for both choices because those details contribute only a small correction to the large-LL count.

Why the bulk DOS often becomes boundary-independent

For a free particle in a large dd-dimensional region of volume VV, the leading number of states below wave number kk is the volume of the allowed region in wave-number space, divided by (2π)d(2\pi)^d and multiplied by the physical volume:

N(k)≃V(2π)d πd/2Γ(d/2+1)kd.N(k)\simeq \frac{V}{(2\pi)^d}\,\frac{\pi^{d/2}}{\Gamma(d/2+1)}k^d.

This leading term depends on the volume, not on whether the walls are periodic, hard, or another standard type. For a smooth boundary, further terms in the count scale with the surface area and lower-dimensional geometric features. They depend on the boundary condition. Since surface area grows more slowly than volume for a large three-dimensional sample, these terms usually vanish relative to the leading term when divided by volume.

For example, the leading three-dimensional DOS per unit volume for a free, spinless particle is

g(E)V=14π2(2mℏ2)3/2E.\frac{g(E)}{V}=\frac{1}{4\pi^2}\left(\frac{2m}{\hbar^2}\right)^{3/2}\sqrt{E}.

An internal degeneracy, such as spin, multiplies this result by the number of states per spatial level. The formula describes the bulk continuum approximation, not the exact discrete spectrum of a small box.

When boundary conditions matter substantially

Boundary effects can be important when the system is small, when you care about its lowest few levels, or when energy resolution is fine enough to distinguish individual levels. They can also create states localized near an edge or surface. In such cases, the boundary may contribute distinct peaks to the DOS rather than merely a small correction.

In statistical mechanics, observables are calculated by summing over energy levels—for example, the canonical partition function is Z=∑ne−βEnZ=\sum_n e^{-\beta E_n}, where β=1/(kBT)\beta=1/(k_{\mathrm B}T). Replacing the sum by an integral over a smooth DOS is useful when many levels lie within the energy range relevant to the temperature. If only a few levels contribute, the boundary-dependent discrete spectrum should be used directly.

Common mistakes

  • Assuming the DOS is always a smooth curve. A finite isolated system has discrete levels; its exact DOS consists of delta peaks.
  • Comparing different boundary conditions without matching the size. The leading bulk comparison assumes the same volume and particle model.
  • Treating “same in the bulk” as “identical.” Finite-size corrections, edge states, and low-energy levels can depend strongly on the boundary.
  • Forgetting degeneracy. Spin or other internal degrees of freedom multiply the state count when they are included in the system being described.

References

  1. Statistical Mechanics, 3rd ed. — R. K. Pathria and Paul D. Beale
  2. Introduction to Quantum Mechanics, 3rd ed. — David J. Griffiths and Darrell F. Schroeter

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.