physics2u
Tier
⌕ Search ⌘K
physics2u.com/vault/probability-current.html
Derivation · the atom

The continuity equation & probability current

D-007 probability-current Home PU-202 Threads chance · waves Depends on verified
Statement

The Schrödinger equation conserves total probability locally, and the flow is described by a current whose form follows without any additional assumption.

Why it matters

Normalisation is not an extra postulate bolted onto quantum mechanics — it is preserved automatically by the dynamics, and this derivation proves it. The current is then the tool that makes transmission and reflection coefficients meaningful in D-011.

Assumptions
The potential is real.The single assumption on which everything rests. A complex potential breaks conservation deliberately — that is exactly how absorption is modelled in optical-potential scattering.
ψ and ∂ψ/∂x are continuous and ψ → 0 at infinity.Needed to discard the surface term when integrating. Fails for plane waves, which is why scattering states need box normalisation or a flux argument instead.
Derivation
1
∂ψ/∂t = − (ħ2/2m) ∂2ψ/∂x2 + Vψ
The Schrödinger equation, taken as given. A
2
∂ψ*/∂t = − (ħ2/2m) ∂2ψ*/∂x2 + Vψ*
Complex conjugate of step 1. The potential survives unchanged because it is real — the assumption is spent here and nowhere else. A
3
∂/∂t |ψ|2 = ψ* ∂ψ/∂t + ψ ∂ψ*/∂t
Product rule on the probability density ρ = ψ*ψ. A
4
∂ρ/∂t = (/2m) ( ψ*2ψ/∂x2 − ψ ∂2ψ*/∂x2 )
Insert steps 1 and 2 into step 3. The two potential terms cancel exactly — Vψ*ψ appears once with each sign. The potential has vanished from the problem entirely. B
5
ψ*ψ″ − ψψ*″ = ∂/∂x ( ψ*ψ′ − ψψ*′ )
Recognise the bracket as an exact x-derivative. Verify by expanding: the cross terms ψ*′ψ′ cancel. This is the step that turns a statement about rates into a statement about flow. C
Result
∂ρ/∂t + ∂j/∂x = 0,    j = (ħ/m) Im( ψ* ∂ψ/∂x )

Reading. Identical in form to charge conservation in electromagnetism and to mass conservation in fluids. Probability does not appear or vanish; it flows. Integrating over all space and using the boundary condition gives d/dt ∫|ψ|2dx = 0 — normalisation, once imposed, is preserved forever.

Units check. [ħ/m] = m2 s−1, times |ψ|2∂ψ/∂x giving m−1 · m−1, yields s−1 — a probability per unit time crossing a point. ✓

Limiting cases
  • Plane wave ψ = Aeikx: j = ħk|A|2/m = v|A|2 — density times velocity, exactly as a classical flux would be.
  • Real wavefunction: j = 0 — bound stationary states carry no current, which is why they are stationary.
Breaks when
  • The potential is complex — probability is deliberately not conserved, modelling absorption or particle decay.
  • Magnetic fields are present — the current acquires a term −qA|ψ|2/m and the naive expression is gauge-dependent and therefore wrong.
  • Relativistic treatment — the Klein–Gordon density is not positive definite, which is one of the failures that forces field theory in PU-402.
Failure modes
  • Writing j = ħ/m · ψ*∂ψ/∂x without taking the imaginary part — the result is then complex and not a current.
  • Assuming a non-zero j means the particle is definitely moving. It is a probability flux, not a trajectory.
Worked number

For a free electron plane wave with k fixed, the current reduces to j = v|A|2: evaluate the group velocity v = ħk/m and multiply by the density |A|2 to confirm the flux units come out as s−1.

Run the check →
Discussion

The result is a continuity equation, and its power lies in being local. A statement like d/dt∫|ψ|² dx = 0 only fixes the grand total; on its own it would even permit probability to vanish here and reappear elsewhere. The differential form ∂ρ/∂t + ∂j/∂x = 0 forbids exactly that: whatever probability leaves a small interval must physically cross its endpoints, carried by the current j. Probability behaves like an incompressible, indestructible fluid — never created, never annihilated, only transported. The global conservation law is then a corollary, recovered by integrating over all space and discarding the surface term.

Notice where the current actually comes from. Writing ψ = √ρ e in amplitude–phase form, one line gives j = (ℏ/m) ρ ∂φ/∂x. The current is the density times a velocity field v = (ℏ/m) ∂φ/∂x set entirely by the gradient of the phase. This is why the imaginary part in the boxed result is not a technicality: a real wavefunction has constant phase, zero phase gradient, and hence zero current — bound stationary states carry no flux, which is precisely what “stationary” means. All flow in quantum mechanics is stored in how fast the phase winds through space, not in the shape of the amplitude.

This phase-gradient reading — the Madelung, or hydrodynamic, form of quantum mechanics — exposes the deep origin of the law. Probability conservation is Noether’s theorem applied to the global U(1) phase symmetry ψ → eψ of the Schrödinger Lagrangian: a symmetry that costs nothing yields a conserved charge (total probability) and its current j. This is also why the assumption that decides everything is V real: only then is the Hamiltonian Hermitian, and Hermiticity is what guarantees unitary, norm-preserving evolution. Promote α to a local, position-dependent phase and consistency forces a gauge field — the electromagnetic potential — correcting the current to j = (ℏ/m) Im(ψ* ∇ψ) − (q/m) A|ψ|². Only this gauge-covariant combination is gauge invariant; the naive expression is not, which is the “breaks when” magnetic warning made precise.

The same skeleton recurs everywhere: charge conservation ∂ρ/∂t + ∇·J = 0 in electromagnetism, mass conservation in fluids, energy–momentum conservation via the stress tensor. Each is the local shadow of a global symmetry. What is special to the Schrödinger case is that ρ = |ψ|² is automatically non-negative, so it can be read as a probability. That positivity fails relativistically: the Klein–Gordon density is not positive definite, and rescuing a sensible interpretation is one of the pressures that forces the move from a single-particle wavefunction to quantum field theory in PU-402.

Common misconceptions. (i) A non-zero j does not mean the particle is definitely moving along a path — it is a probability flux, not a trajectory. (ii) Dropping the Im and writing j = (ℏ/m) ψ*∂ψ/∂x leaves a complex number, which cannot be a physical current. (iii) Probability conservation is not an extra postulate; it is a theorem whose only input is a real potential.

Worked examples

Example 1 — Incident and reflected flux of a stationary scattering state. To the left of a barrier the energy eigenstate is a superposition of a right-moving incident wave and a left-moving reflected wave, ψ = A eikx + B e−ikx. Find the current and the reflection coefficient for a 5 eV electron with |A|² = 1, |B|² = 0.36.

1
ψ = A eikx + B e−ikx,    ∂ψ/∂x = ik(A eikxB e−ikx)
Same |k| both ways: one momentum magnitude, opposite signs.
2
ψ* ∂ψ/∂x = ik[ (|A|² − |B|²) + 2i Im(AB* e2ikx) ]
Multiply out; the two x-dependent cross terms combine into a single purely imaginary piece.
3
j = (ℏ/m) Im(ψ* ∂ψ/∂x) = (ℏk/m)(|A|² − |B|²) = v(|A|² − |B|²)
Taking Im annihilates the cross term (it lives in the real slot). Flux splits cleanly: incident minus reflected. It is x-independent, as a stationary state demands (∂ρ/∂t = 0 ⇒ ∂j/∂x = 0).
4
v = ℏk/m = √(2E/m) = √(2 · 8.01×10−19 / 9.109×10−31) = 1.33×106 m s−1
E = 5 eV = 8.01×10−19 J; group velocity of the incident wave.
5
j = 1.33×106 × (1 − 0.36) = 8.5×105 |A|²,    R = |B|²/|A|² = 0.36
Net flux is 64% of the incident flux; the rest is reflected.
j = v(|A|² − |B|²) = 8.5×105 |A|² m s−1,    R = 0.36

Answer. The current is the incident flux minus the reflected flux, with no interference cross term, and is uniform in x. For the 5 eV electron the net probability flux is 8.5×105|A|² m s−1 and the reflection coefficient is R = 0.36 — the origin of the transmission/reflection bookkeeping formalised in D-011.

Example 2 — Current of a moving Gaussian wave packet. A normalised electron packet is ψ(x) = (2πσ²)−1/4 e−x²/(4σ²) eik0x with width σ = 1 nm and mean wavenumber that of a 5 eV electron, k0 = 1.14×1010 m−1. Find the probability current at the peak, x = 0.

1
ρ = |ψ|² = (2πσ²)−1/2 e−x²/(2σ²)
The complex phase eik0x cancels in |ψ|²; the density is a real Gaussian.
2
ψ/∂x = ( −x/(2σ²) + ik0 ) ψ  ⇒  ψ* ∂ψ/∂x = ( −x/(2σ²) + ik0 ) ρ
The envelope term −x/(2σ²) is real; only the phase term ik0 is imaginary.
3
j = (ℏ/m) Im(ψ* ∂ψ/∂x) = (ℏk0/m) ρ(x) = v0 ρ(x)
Current = local density × group velocity. The real amplitude envelope carries no flux; only the phase winding does.
4
ρ(0) = (2πσ²)−1/2 = [2π(10−9)²]−1/2 = 3.99×108 m−1
Peak density; σ = 1 nm.
5
v0 = ℏk0/m = 1.33×106 m s−1,    j(0) = 1.33×106 × 3.99×108 = 5.3×1014 s−1
Multiply velocity by peak density.
j(0) = v0 ρ(0) = 5.3×1014 s−1

Answer. At the packet centre the probability flux is 5.3×1014 s−1 — the probability per second crossing the origin. Because j = v0ρ everywhere, the whole packet drifts rigidly at the group velocity v0, exactly as a classical density-times-velocity flux would (spreading is a higher-order effect from the spread in k).

Problems
  1. Plane-wave current. For a free-particle plane wave ψ = A eikx show that j = ℏk|A|²/m = v|A|². Evaluate for an electron with k = 5×109 m−1 and |A|² = 1.
    Solution

    ψ/∂x = i, so ψ*∂ψ/∂x = ik|A|², whose imaginary part is k|A|². Hence j = (ℏ/m)k|A|² = v|A|² with v = ℏk/m. Numerically v = (1.055×10−34 × 5×109)/(9.109×10−31) = 5.79×105 m s−1, so j = 5.79×105|A|² (in units of |A|² m s−1). Density × velocity, as expected.

  2. Standing wave vs. partial reflection. (a) Show that the standing wave ψ = A(eikx + e−ikx) carries zero net current. (b) Now take ψ = A eikx + B e−ikx with |A| = 1, |B| = 0.5; find the reflection coefficient and the net flux as a fraction of the incident flux.
    Solution

    (a) ψ = 2A cos kx is real (up to the constant phase of A), so Im(ψ*∂ψ/∂x) = 0 and j = 0: total reflection, equal counter-propagating fluxes cancel. (b) From the boxed result / Example 1, j = v(|A|² − |B|²) = v(1 − 0.25) = 0.75v. The reflection coefficient is R = |B|²/|A|² = 0.25; the net transmitted flux is 75% of the incident flux.

  3. Three dimensions. Starting from the 3D Schrödinger equation with real V, derive ∂ρ/∂t + ∇·j = 0 with j = (ℏ/m) Im(ψ* ∇ψ).
    Solution

    ρ/∂t = ψ*∂ψ/∂t + ψψ*/∂t. Insert iℏ∂ψ/∂t = −(ℏ²/2m)∇²ψ + and its conjugate. The V|ψ|² terms cancel (V real), leaving ∂ρ/∂t = (iℏ/2m)(ψ*∇²ψψ∇²ψ*). Use the identity ψ*∇²ψψ∇²ψ* = ∇·(ψ*∇ψψψ*) (the ∇ψ*·∇ψ cross terms cancel). Thus ∂ρ/∂t = −∇·j with j = −(iℏ/2m)(ψ*∇ψψψ*) = (ℏ/m) Im(ψ*∇ψ), using zz* = 2i Im z.

  4. Complex (optical) potential. Let V = V0 − iΓ/2 with Γ > 0 constant. Show that d/dt∫|ψ|² dx = −(Γ/ℏ)∫|ψ|² dx, so the norm decays as e−Γt/ℏ with lifetime τ = ℏ/Γ. Evaluate τ for Γ = 0.1 eV.
    Solution

    Repeating the density calculation but keeping V complex, the potential terms no longer cancel: ∂ρ/∂t = −∂j/∂x + (i/ℏ)(V* − V)|ψ|². Here V* − V = 2i Im(V)·(−1)... explicitly V* − V = (V0 + iΓ/2) − (V0 − iΓ/2) = iΓ, so (i/ℏ)(iΓ)|ψ|² = −(Γ/ℏ)|ψ|². Integrate over all space; the current’s divergence integrates to the vanishing surface term, giving d/dt∫|ψ|² dx = −(Γ/ℏ)∫|ψ|² dx. Hence N(t) = N(0) e−Γt/ℏ and τ = ℏ/Γ = (1.055×10−34)/(0.1×1.602×10−19) = 6.6×10−15 s. A negative imaginary potential removes probability — the standard model of absorption and decay.

  5. Phase-gradient (Madelung) form and the magnetic current. (a) Writing ψ = √ρ eiS/ℏ, show j = ρS/m, i.e. a velocity field v = ∇S/m. (b) With minimal coupling (ppqA) the conserved current becomes j = (ℏ/m) Im(ψ*∇ψ) − (q/m)A|ψ|². Show it is invariant under the gauge transformation AA + ∇χ, ψ → eiqχ/ℏψ, whereas the first term alone is not.
    Solution

    (a) ∇ψ = (∇√ρ + i√ρS/ℏ) eiS/ℏ, so ψ*∇ψ = √ρ ∇√ρ + iρS/ℏ = ½∇ρ + iρS/ℏ. The first term is real; Im(ψ*∇ψ) = ρS/ℏ, giving j = (ℏ/m)(ρS/ℏ) = ρS/m. So v = ∇S/m: the local velocity is the phase gradient. (b) In this notation the gauge current is j = (ρ/m)(∇SqA). Under the transformation SS + qχ (the phase shift) and AA + ∇χ, so ∇SqA → ∇S + q∇χ − q(A + ∇χ) = ∇SqA, unchanged, and ρ = |ψ|² is manifestly invariant. Thus j is gauge invariant. The bare term (ℏ/m)Im(ψ*∇ψ) = ρS/m alone picks up the spurious piece ρq∇χ/m and is therefore unphysical in a magnetic field.