core33 min studyLesson 371 in Physics

Matching a finite square well

Determine bound energies by joining oscillatory interior solutions to normalizable exterior tails.

On this page

What you’ll understand

  • Derive the even and odd finite-well matching equations.
  • Normalize a state including its exterior probability.
  • Distinguish a finite wall from an imposed Dirichlet boundary.

An infinite well enforces a wavefunction that vanishes at its walls. A finite well instead allows a tail outside the classically allowed region, and its energy is selected by joining that tail smoothly to an interior oscillation. The matching problem illustrates a general spectral principle: a differential equation gives many local solutions, but the domain, boundary behaviour, and normalizability select the physical eigenstates.

Choose one energy reference throughout

Take V(x)=−V0V(x)=-V_0 for ∣x∣<a|x|<a and zero outside, with V0>0V_0>0 and half-width aa. Bound energies satisfy −V0<E<0-V_0<E<0. Inside, kinetic energy E+V0E+V_0 is positive and solutions oscillate. Outside, negative total energy gives exponentials, of which only the decaying branches are square-integrable. A positive exterior energy instead permits scattering waves; it is not another member of the discrete bound spectrum.

At a finite jump of a potential with constant particle mass, both ψ\psi and ψ′\psi' are continuous. Integrating Schrödinger's equation across the step shows that a finite potential contributes no finite derivative jump. A discontinuity in the wavefunction would also generate an unsupported singular second derivative. These conditions differ from a delta interaction and from an infinite-wall boundary. Writing the same second-order equation in each region is insufficient until the correct interface conditions are imposed.

k=2m(E+V0)ℏ,κ=−2mEℏ,k2+κ2=2mV0ℏ2k=\frac{\sqrt{2m(E+V_0)}}{\hbar},\qquad \kappa=\frac{\sqrt{-2mE}}{\hbar},\qquad k^2+\kappa^2=\frac{2mV_0}{\hbar^2}
The constant mass m is in kg, a is a length, V₀ and E are energies in consistent units, and both k and κ are inverse lengths. These expressions apply to real bound eigenvalues between the well floor and the exterior threshold; κ must be positive.↔ Scroll wide equations horizontally.

Parity reduces two interfaces to one

The potential is reflection symmetric. Its ordinary one-dimensional bound levels are nondegenerate, so eigenfunctions can be chosen even or odd. Use Acos⁡(kx)A\cos(kx) for an even interior state and Asin⁡(kx)A\sin(kx) for an odd one. At the positive edge, continuity fixes the tail amplitude. Dividing the derivative condition by the value then removes the normalization constant and leaves a transcendental equation for energy. Reflection supplies the negative-edge condition automatically.

Dimensionless variables organize the roots without hiding the physical threshold. Set z=kaz=ka, y=κay=\kappa a, and z0=a2mV0/ℏz_0=a\sqrt{2mV_0}/\hbar. Allowed roots lie on the circle z2+y2=z02z^2+y^2=z_0^2 with y>0y>0. Even and odd matching curves intersect different branches. A pole of tangent is not an eigenvalue just because a numerical residual changes sign there; the continuous branch and the original matching conditions must both be checked.

ztan⁡z=y(even),−zcot⁡z=y(odd),z2+y2=z02,z,y>0z\tan z=y\quad\text{(even)},\qquad -z\cot z=y\quad\text{(odd)},\qquad z^2+y^2=z_0^2,\quad z,y>0
All three variables are dimensionless. These exact equations refer to the symmetric finite well of full width 2a. An exterior threshold solution with y = 0 is excluded from the square-integrable bound-state set, even if it solves a limiting matching equation.↔ Scroll wide equations horizontally.

Normalize both interior and exterior

For an even state the positive-side tail is Acos⁡(ka)e−κ(x−a)A\cos(ka)e^{-\kappa(x-a)}. Integrating its square on both exterior half-lines gives ∣A∣2cos⁡2(ka)/κ|A|^2\cos^2(ka)/\kappa. The interior contribution is ∣A∣2[a+sin⁡(2ka)/(2k)]|A|^2[a+\sin(2ka)/(2k)]. Their sum, not the interior alone, must equal one. A shallow state can put substantial probability outside its nominal well; being bound means normalizable localization, not absolute exclusion from every region above the energy.

Deepening the well approaches infinite-well energies when measured above the floor, while exterior tails shrink. Near zero binding, the decay length 1/κ1/\kappa grows and a spatially large integration box is needed to normalize the state numerically. Matching and normalization answer different questions: a guessed energy may produce an easily normalized piecewise function that still fails derivative continuity. Conversely, an eigenfunction's overall amplitude is fixed only after its energy and matching ratios have been found.

An exactly checkable matching root

Let Ca=ℏ2/(2ma2)=1 eVC_a=\hbar^2/(2ma^2)=1\,\mathrm{eV} and V0=(π2/8) eVV_0=(\pi^2/8)\,\mathrm{eV}. Test the even candidate z=π/4z=\pi/4. Find its bound energy and the probability outside the well, retaining the exterior tails in normalization.

  1. The depth gives z02=V0/Ca=π2/8z_0^2=V_0/C_a=\pi^2/8. Therefore y2=z02−z2=π2/16y^2=z_0^2-z^2=\pi^2/16 and y=π/4y=\pi/4. Since tan⁡(π/4)=1\tan(\pi/4)=1, the even equation ztan⁡z=yz\tan z=y is exactly satisfied.

  2. The energy relative to the exterior is E=−Cay2=−π2/16 eV≃−0.617 eVE=-C_ay^2=-\pi^2/16\,\mathrm{eV}\simeq-0.617\,\mathrm{eV}. It lies above the floor −1.234 eV-1.234\,\mathrm{eV} and below zero, checking the bound-state regime independently of matching.

  3. The interior norm factor is a(1+2/π)a(1+2/\pi) and the exterior factor is 2a/π2a/\pi. Thus ∣A∣2=[a(1+4/π)]−1|A|^2=[a(1+4/\pi)]^{-1} and Pout=(2/π)/(1+4/π)=2/(π+4)≃0.280P_{\rm out}=(2/\pi)/(1+4/\pi)=2/(\pi+4)\simeq0.280. Both sides of the well are included.

The result

The candidate is an even bound state at approximately −0.617 eV, with 28.0% exterior probability. Replacing the finite well by zero wavefunction at its edges would discard a substantial part of this state.

Explore the model

Interactive lab

Interactive lab · root-finding

Practise a continuous root bracket

Reset and set both controls to target c = 1 and upper bound U = 3; predict that bisection retains the root 1. Reset and set both controls to c = 4 and U = 3; predict root 2, then Step. This lab bisects x² − c on [0,U]. It illustrates bracketing, but has no tangent poles, finite-well eigenvalues, parity branches, or normalization calculation.

Practise a continuous root bracketBisection of x² − c on [0,U], capped at 50 iterations for floating-point resolution. The error bound is the maximum distance from the displayed midpoint to either bracket endpoint (half the width in exact arithmetic). Compensated square evaluation preserves the bracket when ordinary multiplication rounds a residual to zero. A truly exact endpoint or midpoint collapses the bracket. This is one monotone quadratic, not a general root solver. Current readings: Midpoint after 0 refinements: 1.5; Bracket error bound: 1.5; Computed residual · x² − c: 0.25. All parameters are adjustable with the labeled controls below.Bisection on x squared minus 2: bracket from 0 to 3 after 0 refinements.00.751.52.253−2.360.192.745.297.84x · bracket stays nonnegativef(x) = x² − cn = 0 · [ 0, 3 ]
Quadratic / midpointCurrent bracketPositive root
Midpoint after 0 refinements
1.5
Bracket error bound
1.5
Computed residual · x² − c
0.25
t = 0 s

Bisection of x² − c on [0,U], capped at 50 iterations for floating-point resolution. The error bound is the maximum distance from the displayed midpoint to either bracket endpoint (half the width in exact arithmetic). Compensated square evaluation preserves the bracket when ordinary multiplication rounds a residual to zero. A truly exact endpoint or midpoint collapses the bracket. This is one monotone quadratic, not a general root solver. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.

Look forA valid bracket belongs to a continuous residual branch; the quadratic lab does not certify brackets across a finite-well tangent pole.

Check your understanding

What must remain continuous at an ordinary finite potential step for the constant-mass Schrödinger Hamiltonian?

Try it yourself

A finite-well bound state has binding energy ∣E∣=Ca/4|E|=C_a/4, with Ca=ℏ2/(2ma2)C_a=\hbar^2/(2ma^2). Find its exterior amplitude decay length. Conditional on already being beyond the positive edge, what fraction lies farther than another distance 2a2a? How does the decay length change if binding falls to Ca/16C_a/16?

Take a moment to make a prediction or work it out on paper. Open a hint when you need one.

Give me a hint

The amplitude decays with κ, while probability density decays with twice that exponent.

Compare with the solution

Initially κa=∣E∣/Ca=1/2\kappa a=\sqrt{|E|/C_a}=1/2, so the amplitude decay length is 2a2a. Conditional exterior probability beyond an additional distance dd is e−2κde^{-2\kappa d}. For d=2ad=2a this is e−2≃0.135e^{-2}\simeq0.135. This conditional fraction is not the state's total exterior population, which also depends on matching and normalization. At binding Ca/16C_a/16, κa=1/4\kappa a=1/4 and the amplitude decay length doubles to 4a4a. We have used a stated bound energy rather than claiming these numbers independently determine the depth.

Take it with you

  • Finite steps require continuous wavefunction and derivative, not zero edge values.
  • Parity matching selects discrete energies from local differential-equation solutions.
  • Exterior tails are part of normalization and grow long near threshold.

Read further

Original sources and deeper explanations to follow at your own pace.

Optional progress, saved in this browser when available. Revisit any lesson at any time.

Lesson not marked complete.