Identical Particles and Exchange Symmetry

Classical mechanics tracks particles by their trajectories. Two billiard balls of identical make are still distinguishable in principle: follow each continuous path and the label attached at the start stays attached. Quantum mechanics removes that thread. A wavefunction spread over space carries no trajectory, and when two electrons occupy overlapping regions there is no measurement, even in principle, that reports which one is the first. The states of the theory must be built so that this ignorance is exact rather than merely practical. The consequence is not a small correction; it fixes the structure of atoms, the stability of matter, and the two great families into which every particle falls.

The two-particle state space

A single spinless particle in one dimension has states in the Hilbert space $\mathcal{H}$ of square-integrable functions $\psi (x)$. Two such particles live in the tensor product $\mathcal{H}\otimes \mathcal{H}$, spanned by products ${\psi }_a(x_1){\psi }_b(x_2)$, and a general two-particle state is a wavefunction of both coordinates,

The Born rule reads $|\Psi (x_1,x_2){|}^2\mathbf{d}{x}_1\mathbf{d}{x}_2$ as the probability of finding one particle in $\mathbf{d}{x}_1$ about $x_1$ and the other in $\mathbf{d}{x}_2$ about $x_2$. Written that way the interpretation already anticipates the problem: it speaks of one particle and the other, not of particle 1 and particle 2. If the two particles are identical, the labels $1$ and $2$ are a fiction of the notation, and any physical prediction must be independent of how they are assigned.¹

For a Hamiltonian this means invariance under the interchange of the two sets of coordinates. Every physical two-particle Hamiltonian is symmetric,

because the kinetic terms enter identically and any interaction depends only on the relative configuration, $V(|x_1-x_2|)$. Nothing in the dynamics can tell the two particles apart.

The exchange operator

Give the interchange of labels an operator. The exchange operator $\hat{P}$ acts on a two-particle wavefunction by swapping its arguments,

Applying it twice returns the original state, so

Its eigenvalues therefore satisfy ${\lambda }^2=1$, giving $\lambda =\pm 1$ and nothing else. $\hat{P}$ is Hermitian, since exchanging labels inside an inner product moves the swap from one factor to the other without conjugation, and it is unitary because ${\hat{P}}^{†}\hat{P}={\hat{P}}^2=1$. A single operator that is at once Hermitian and its own inverse is a reflection: it splits the space into a $+1$ eigenspace and a $-1$ eigenspace.²

The decisive property is that $\hat{P}$ commutes with any identical-particle Hamiltonian. Because $\hat{H}$ is symmetric under the interchange of coordinates,

so $[\hat{P},\hat{H}]=0$. The exchange eigenvalue is a conserved quantity. A state that starts totally symmetric stays totally symmetric under evolution, and a totally antisymmetric state stays antisymmetric. Symmetry type is not a convention imposed once; it is protected by the dynamics for all time.

The symmetrization postulate

Commuting with $\hat{H}$ lets symmetric and antisymmetric states exist as stationary symmetry types, but it does not by itself forbid states of mixed symmetry, superpositions with both $+1$ and $-1$ components. The restriction to the two extremes is a separate physical law, confirmed without exception.

Which family a species belongs to is fixed by its spin. Particles of integer spin ($0,1,2,\dots$: photons, pions, the $W$ and $Z$, helium-4 atoms) are bosons; particles of half-integer spin ($\frac{1}{2},\frac{3}{2},\dots$: electrons, protons, neutrons, neutrinos, helium-3 atoms) are fermions.

The theorem is stated here and used, not derived. Its non-relativistic content is the empirical rule: attach symmetric to integer spin and antisymmetric to half-integer spin, and every atomic and statistical consequence follows.³

Constructing symmetric and antisymmetric states

Start from two orthonormal single-particle states, ${\psi }_a$ and ${\psi }_b$ with $a\ne b$. The bare product ${\psi }_a(x_1){\psi }_b(x_2)$ is neither symmetric nor antisymmetric: it assigns state $a$ to particle $1$, a statement the labels are not entitled to make. Project it onto the two allowed symmetry types by adding or subtracting the exchanged product,

with $+$ for bosons and $-$ for fermions. The prefactor $1/\sqrt{2}$ normalizes the state when ${\psi }_a$ and ${\psi }_b$ are orthonormal, because the two cross terms integrate to zero and each square integrates to one. Under $\hat{P}$ the symmetric combination returns itself and the antisymmetric one changes sign, as required.

The antisymmetric state carries an immediate consequence. Set $a=b$: the two terms cancel, and ${\Psi }_{-}\equiv 0$. There is no antisymmetric state built from two copies of the same single-particle state.

The symmetric combination has no such restriction; setting $a=b$ there gives the perfectly good state ${\psi }_a(x_1){\psi }_a(x_2)$, and any number of bosons may pile into one mode. The exclusion principle is not an added rule for fermions. It is the single visible corollary of antisymmetry.

Normalization when states overlap

The clean factor $1/\sqrt{2}$ assumed $⟨{\psi }_a|{\psi }_b⟩=0$. When the two single-particle states are not orthogonal, with overlap $S=⟨{\psi }_a|{\psi }_b⟩$, the norm of ${\Psi }_{\pm }$ picks up the cross terms,

This normalization matters in molecular bonding, where atomic orbitals on different nuclei overlap substantially, and it reappears in the helium exchange integrals of the next lesson.

The exchange force

Symmetrization correlates the positions of the two particles even when the Hamiltonian contains no interaction between them. Compute the mean-square separation $⟨(x_1-x_2{)}^2⟩$ in the three cases: distinguishable particles in the product state, and identical particles in ${\Psi }_{\pm }$.

Expand the square,

For the distinguishable product ${\psi }_a(x_1){\psi }_b(x_2)$ the coordinates are independent, so $⟨x_1{x}_2⟩=⟨x{⟩}_a⟨x{⟩}_b$ and

For ${\Psi }_{\pm }$ the single-particle averages $⟨x_1^2⟩$ and $⟨x_2^2⟩$ come out to $\frac{1}{2}(⟨x^2{⟩}_a+⟨x^2{⟩}_b)$ from each term, summing to the same $⟨x^2{⟩}_a+⟨x^2{⟩}_b$. The cross term is where the symmetry enters. Carrying the interference terms through,

Assembling the pieces gives the central result,

The symmetric (boson) state has the smaller mean-square separation, the antisymmetric (fermion) state the larger. Identical bosons behave as though a weak attraction draws them together; identical fermions as though a weak repulsion holds them apart.⁴

Two features deserve emphasis. First, the exchange force is not a force in the mechanical sense; no potential generates it and it does no work. It is a correlation built into the state by antisymmetrization, a kinematic effect of the statistics. Second, it requires spatial overlap. Two electrons in atoms on opposite sides of a room have $⟨x{⟩}_{ab}\approx 0$, and their antisymmetrization is physically irrelevant. The Pauli principle bites only where wavefunctions overlap.

Spin and the total wavefunction

Electrons carry spin, so the exchange that matters is of the full state, position and spin together. The complete two-electron wavefunction is a product of a spatial part and a spin part,

and exclusion applies to the whole object: $\Psi$ must change sign under the simultaneous exchange of positions and spins. That single antisymmetry can be met in two ways, since a product is odd when exactly one factor is odd.

• Symmetric spatial, antisymmetric spin. The two spins combine into the singlet ${\chi }_{00}=\frac{1}{\sqrt{2}}(\uparrow \downarrow -\downarrow \uparrow )$, total spin $S=0$, which is odd under exchange. The spatial part is then even.
• Antisymmetric spatial, symmetric spin. The two spins combine into a triplet ($S=1$, the three states $\uparrow \uparrow$, $\frac{1}{\sqrt{2}}(\uparrow \downarrow +\downarrow \uparrow )$, $\downarrow \downarrow$), which is even; the spatial part is then odd.

The addition of two spin-½ angular momenta supplies exactly one antisymmetric combination (the singlet) and three symmetric ones (the triplet), so the two options above exhaust the possibilities.

The pairing ties spatial correlation to total spin. Electrons in the spin triplet have an antisymmetric spatial state, so they avoid one another (fermionic exchange), lower their mutual Coulomb repulsion, and sit at lower energy whenever a repulsive interaction is present. Electrons in the singlet share a symmetric spatial state, crowd closer, and pay more repulsion. This is the mechanism behind Hund's first rule and the ortho/para splitting of helium, both worked out in the next lesson.

Identical-particle interference

The exchange term is not only a statistical bookkeeping device; it produces observable interference. Consider two identical particles emitted from sources $1$ and $2$ and detected at $a$ and $b$. Let $⟨a|1⟩$ be the amplitude for the particle from source $1$ to arrive at detector $a$, and likewise for the other combinations. Two indistinguishable histories end in the same event, one at each detector, and their amplitudes add:

The coincidence probability is $|A_{\pm }{|}^2$, which contains the interference cross term

The classical, distinguishable result is the sum of the first two terms. The third term is the quantum signature of identity: constructive for bosons, destructive for fermions. When the two detectors coincide, $a=b$, the two amplitudes become equal, and the fermionic combination vanishes identically ($A_{-}=0$) while the bosonic one doubles. Two fermions are never detected in the same state; two bosons arrive together more often than chance allows. This is the content of the Hanbury Brown–Twiss correlation: photons (bosons) bunch, and the same apparatus run with electrons shows the complementary antibunching.⁵

The N-particle generalization

For $N$ identical particles the exchange operators ${\hat{P}}_{ij}$ swap each pair, and the physical states are the ones left invariant (bosons) or sign-flipped under every transposition (fermions). Symmetrizing a product of $N$ single-particle states ${\psi }_{{\alpha }_1},\dots ,{\psi }_{{\alpha }_N}$ over all $N!$ permutations builds the totally symmetric state (a permanent) or the totally antisymmetric one, written as a determinant,

This Slater determinant is antisymmetric automatically, because swapping two particles exchanges two rows and a determinant changes sign under row exchange, and it vanishes when two single-particle states coincide, because a determinant with two equal columns is zero. The exclusion principle is the vanishing determinant, now for any number of particles. The systematic use of Slater determinants to build atoms, with the exchange integrals that split helium and order the periodic table, is the subject of the next lesson.

The split runs through the whole of physics. Fermionic avoidance is why matter occupies volume and why the periodic table has the shape it does; bosonic bunching is why lasers and Bose–Einstein condensates exist. Both descend from the single algebraic fact that ${\hat{P}}^2=1$ and nature selects its two eigenvalues.