Field Quantization of the Klein-Gordon and Dirac Fields

Relativistic single-particle quantum mechanics reaches a wall. As The Klein-Gordon Equation and Klein-Gordon Probability Current and Negative Densities showed, the scalar wave equation gives a density that can go negative; as Negative-Energy Solutions and the Interpretive Crisis and The Dirac Sea and Antiparticles showed, the Dirac equation drags along a bottomless tower of negative-energy states, patched only by an asymmetric filled sea. Every repair strains against the same fact: $E=mc^2$ lets energy convert into particles, so any theory that insists on a fixed number of particles is telling a lie about a world where pairs are created and destroyed. The resolution is to change what $\phi$ and $\psi$ are. Instead of wavefunctions, treat them as dynamical fields and quantize them the way you quantize a mechanical system — promote them to operators and impose canonical brackets. This is field quantization (historically "second quantization"). A free field is an infinite family of decoupled harmonic oscillators, one per momentum mode; its quanta are the particles. The number of particles is now an eigenvalue of an operator, free to change. The decisive lesson lives in the bracket. Quantize the scalar with commutators and everything works: positive energy, causal signals, Bose statistics. Try commutators on the Dirac field and the energy plunges to $-\infty$ ; only anticommutators rescue it, and they hand you the Pauli exclusion principle for free. Here we quantize the free Klein-Gordon field, extract $[a_\mathbf{p},a^\dagger_{\mathbf{p}'}]=(2\pi)^3\delta^3(\mathbf{p}-\mathbf{p}')$ from the field commutator, prove the bosonic Hamiltonian is bounded below, and then machine-check the positivity fork that forces fermion statistics. The full Lorentz-invariant proof of that fork is the next article; here it is a positivity argument you can compute.

Key takeaways

A free relativistic field is an infinite collection of harmonic oscillators, one per momentum mode; quantizing the field means quantizing all of them at once, and the quanta are the particles.
The Klein-Gordon field is quantized with equal-time commutators $[\phi(\mathbf{x}),\pi(\mathbf{y})]=i\delta^3(\mathbf{x}-\mathbf{y})$ , which forces $[a_\mathbf{p},a^\dagger_{\mathbf{p}'}]=(2\pi)^3\delta^3(\mathbf{p}-\mathbf{p}')$ — Bose statistics.
The bosonic Hamiltonian $H=\int\!\frac{\mathrm d^3p}{(2\pi)^3}\,\omega_\mathbf{p}\,a^\dagger_\mathbf{p}a_\mathbf{p}$ is manifestly $\ge0$ : the negative-energy problem is gone because the vacuum is the lowest state, not a particle with $-E$ .
The Dirac field must use anticommutators $\{b,b^\dagger\}=1$ : commutators make the antiparticle term $-d^\dagger d$ , unbounded below; anticommutators flip it to $+d^\dagger d\ge0$ .
Anticommutators build Pauli exclusion into the algebra: $(b^\dagger)^2=0$ , so no mode holds two identical fermions.
Negative-frequency modes are reinterpreted: the coefficient of $e^{+ip\cdot x}$ is an antiparticle creation operator adding positive energy — the Dirac sea dissolves into a particle/antiparticle-symmetric formalism.
Key idea: promote the relativistic field to an operator, expand in momentum modes, and let positivity of energy FORCE commutators for spin-0 and anticommutators for spin-1/2 — the operational heart of the spin-statistics connection.

The core insight

Quantizing the free Klein-Gordon and Dirac fields turns them into towers of oscillators whose quanta are particles and antiparticles; demanding a Hamiltonian bounded below forces commutators for the spin-0 field and anticommutators for the spin-1/2 field, which is why bosons obey Bose statistics and fermions obey the exclusion principle.

Where this lives today

Field quantization is not a historical way station; it is the working language of every calculation in particle physics. In the Standard Model each elementary particle is a quantum of a field — electrons and quarks of Dirac fields, the photon and gluons of gauge fields, the Higgs boson of a scalar field — and every cross section and decay rate at the LHC is a matrix element of these field operators between Fock states. The particle/antiparticle symmetry made manifest by the $c$ / $d$ split is confirmed billions of times per second in colliders: a photon converts to an $e^+e^-$ pair, a $c^\dagger d^\dagger$ acting on the vacuum, with the charge operator $Q$ enforcing conservation. The same machinery, taken to its nonrelativistic limit, is the second-quantized formalism of condensed matter — phonons, magnons, Cooper pairs — where the identical commutator/anticommutator structure governs the Fermi sea of a metal and the condensation of a superfluid. Even the vacuum energy that normal ordering throws away returns with a vengeance: its difference between boundary conditions is the Casimir force, measured to percent accuracy between metal plates.

The physical picture

Picture the field not as a wave riding on top of space but as a value attached to every point — and then imagine expanding that field into its momentum components, a Fourier decomposition. Each component of a fixed momentum $\mathbf{p}$ oscillates at frequency $\omega_\mathbf{p}=\sqrt{\mathbf{p}^2+m^2}$ : it is a harmonic oscillator. A free field is nothing but an infinite orchestra of these oscillators, one per momentum, all playing independently. Quantizing the field means quantizing every oscillator, and the language you already have — creation and annihilation operators, from Creation and Annihilation Operators — is exactly right. The $n$ -th excited state of the $\mathbf{p}$ -oscillator simply means " $n$ particles, each of momentum $\mathbf{p}$ ." Particle number is no longer a fixed label; it is what the number operators count.

Figure

A free field is a tower of oscillators; its excitations are particles

The two kinds of bracket encode the two kinds of oscillator, and the difference is not a convention but a wall. A bosonic oscillator, $[a,a^\dagger]=1$ , lets you stack quanta without limit — pump the same mode a thousand times and you have a thousand identical bosons, the physics behind lasers and Bose-Einstein condensation. A fermionic oscillator, $\{b,b^\dagger\}=1$ , saturates at one quantum because the algebra itself gives $(b^\dagger)^2=0$ : try to create a second identical fermion and you annihilate the state. Pauli exclusion, which Pauli Exclusion Principle derives from wavefunction antisymmetry, is here a one-line consequence of the anticommutator. The punchline of this whole article is that you do not get to choose which bracket a field uses — its spin chooses for you, on pain of an energy that has no floor.

Figure

The positivity fork: which bracket keeps the energy bounded below

The core equations

We work in natural units $\hbar=c=1$ with the mostly-minus metric $g^{\mu\nu}=\mathrm{diag}(+,-,-,-)$ , so $p\cdot x=\omega_\mathbf{p}t-\mathbf{p}\cdot\mathbf{x}$ and a spacelike interval has $(x-y)^2<0$ . The Klein-Gordon field carries mass $m$ exactly as in The Klein-Gordon Equation, whose dispersion $E^2=\mathbf{p}^2+m^2$ we pin as $\omega_\mathbf{p}=\sqrt{\mathbf{p}^2+m^2}$ . The starting point is the classical scalar Lagrangian and its conjugate momentum, the same objects you would write for any field theory before quantizing.

\mathcal{L}=\tfrac12(\partial_\mu\phi)(\partial^\mu\phi)-\tfrac12 m^2\phi^2,\qquad \pi(x)=\frac{\partial\mathcal{L}}{\partial\dot\phi}=\dot\phi \qquad (1)

where: $\phi$ — the real Klein-Gordon field; $m$ — the mass of its quantum; $\pi$ — the canonically conjugate momentum density. Reading it physically: this is the field-theory analogue of $L=\tfrac12 m\dot q^2-\tfrac12 m\omega^2 q^2$ for a single oscillator, promoted to a continuum of degrees of freedom labeled by position $\mathbf{x}$ . Quantization is the single postulate that the field and its momentum satisfy equal-time canonical commutators — the field-theory version of $[\hat q,\hat p]=i$ .

[\phi(t,\mathbf{x}),\pi(t,\mathbf{y})]=i\,\delta^3(\mathbf{x}-\mathbf{y}),\qquad [\phi,\phi]=[\pi,\pi]=0 \qquad (2)

where: $\delta^3$ — the three-dimensional Dirac delta enforcing "independent oscillator at each point"; the equal-time label $t$ is essential because these are Schrödinger-picture operators at one instant. Reading it physically: choosing a commutator here (not an anticommutator) is the choice that produces Bose statistics — a choice we will later see is forced for a scalar by causality. Expanding the field in plane-wave modes turns the abstract postulate into concrete ladder operators. The two frequency signs $e^{\mp ip\cdot x}$ carry the annihilation and creation pieces.

\phi(x)=\int\frac{\mathrm d^3p}{(2\pi)^3}\frac{1}{\sqrt{2\omega_\mathbf{p}}}\left(a_\mathbf{p}\,e^{-ip\cdot x}+a_\mathbf{p}^\dagger\,e^{+ip\cdot x}\right) \qquad (3)

where: $a_\mathbf{p},a_\mathbf{p}^\dagger$ — annihilation and creation operators for a boson of momentum $\mathbf{p}$ ; the factor $1/\sqrt{2\omega_\mathbf{p}}$ is the relativistic normalization that keeps the measure Lorentz-invariant. Reading it physically: a real field needs $a^\dagger$ to be the Hermitian conjugate of $a$ so that $\phi=\phi^\dagger$ — the reason the same operator content appears with both frequency signs. Feeding the expansion (3) into the canonical commutator (2) fixes the algebra of the ladder operators, and closes the loop with Creation and Annihilation Operators, which owns the commutator/anticommutator algebra itself.

[a_\mathbf{p},a_{\mathbf{p}'}^\dagger]=(2\pi)^3\,\delta^3(\mathbf{p}-\mathbf{p}'),\qquad [a,a]=[a^\dagger,a^\dagger]=0 \qquad (4)

where: the $(2\pi)^3$ is the continuum-normalization companion of the $(2\pi)^3$ in the mode measure. Reading it physically: this is the boson algebra — the same $[a,a^\dagger]=1$ of a single oscillator, smeared over a continuum of modes. It is a derived consequence of (2), not an independent assumption. The scalar Hamiltonian follows by substituting (3) into $H=\int\!\mathrm d^3x\,[\tfrac12\pi^2+\tfrac12(\nabla\phi)^2+\tfrac12 m^2\phi^2]$ . Two things happen: the cross terms cancel on shell, and an infinite constant appears that we normal-order away.

H=\int\frac{\mathrm d^3p}{(2\pi)^3}\,\omega_\mathbf{p}\Big(a_\mathbf{p}^\dagger a_\mathbf{p}+\tfrac12[a_\mathbf{p},a_\mathbf{p}^\dagger]\Big)\;\xrightarrow{\;\text{normal order}\;}\;\int\frac{\mathrm d^3p}{(2\pi)^3}\,\omega_\mathbf{p}\,a_\mathbf{p}^\dagger a_\mathbf{p} \qquad (5)

where: $a^\dagger a$ — the number operator counting quanta in mode $\mathbf{p}$ ; $\tfrac12[a,a^\dagger]\propto\delta^3(0)$ — the divergent zero-point energy, removed because only energy differences are observable. Reading it physically: the spectrum is now $E=\sum_i\omega_{\mathbf{p}_i}\ge0$ , one positive quantum per excitation — the negative-energy tower of the single-particle theory is simply absent, because the lowest state is the empty vacuum, not a particle with $-E$ . The Dirac field is the case that decides everything. Its Lagrangian and the canonical anticommutator replace (1)–(2), with the spinor structure and gamma matrices inherited from The Dirac Equation.

\mathcal{L}=\bar\psi(i\gamma^\mu\partial_\mu-m)\psi,\qquad \{\psi_a(t,\mathbf{x}),\psi_b^\dagger(t,\mathbf{y})\}=\delta_{ab}\,\delta^3(\mathbf{x}-\mathbf{y}) \qquad (6)

where: $\psi$ — the four-component Dirac field; $\bar\psi=\psi^\dagger\gamma^0$ — the Dirac adjoint; $a,b$ — spinor component indices $1{-}4$ ; $\{\gamma^\mu,\gamma^\nu\}=2g^{\mu\nu}$ — the Clifford algebra. Reading it physically: the choice of an anticommutator here — the one difference from the scalar — is what the rest of the article will show is mandatory. Expanding the Dirac field, the positive-frequency modes carry particle annihilators $c$ , and the negative-frequency modes carry antiparticle creators $d^\dagger$ — the reinterpretation that dissolves the Dirac sea.

\psi(x)=\int\frac{\mathrm d^3p}{(2\pi)^3}\frac{1}{\sqrt{2E_\mathbf{p}}}\sum_{s=1,2}\left(c_\mathbf{p}^{\,s}\,u^s(p)\,e^{-ip\cdot x}+d_\mathbf{p}^{\,s\dagger}\,v^s(p)\,e^{+ip\cdot x}\right) \qquad (7)

where: $c^s,c^{s\dagger}$ — annihilate/create a fermion (electron) of spin $s$ ; $d^s,d^{s\dagger}$ — annihilate/create the antifermion (positron); $u^s,v^s$ — the positive- and negative-frequency Dirac spinors of Dirac Spinors and Plane-Wave Solutions. Reading it physically: the negative-frequency solution $e^{+ip\cdot x}$ is no longer a "negative-energy electron" — its coefficient creates a positive-energy positron. The anticommutator carries the exclusion principle in its structure, and the Hamiltonian comes out positive for both species.

\{c_\mathbf{p}^r,c_{\mathbf{p}'}^{s\dagger}\}=\{d_\mathbf{p}^r,d_{\mathbf{p}'}^{s\dagger}\}=(2\pi)^3\delta^{rs}\delta^3(\mathbf{p}-\mathbf{p}'),\qquad (c^\dagger)^2=(d^\dagger)^2=0 \qquad (8)

H=\int\frac{\mathrm d^3p}{(2\pi)^3}\,E_\mathbf{p}\sum_s\big(c_\mathbf{p}^{s\dagger}c_\mathbf{p}^{s}+d_\mathbf{p}^{s\dagger}d_\mathbf{p}^{s}\big),\qquad Q=\int\frac{\mathrm d^3p}{(2\pi)^3}\sum_s\big(c^{s\dagger}c^{s}-d^{s\dagger}d^{s}\big) \qquad (9)

where: $(c^\dagger)^2=0$ — two identical fermions cannot share a mode; $Q$ — the conserved charge, $+1$ for particles and $-1$ for antiparticles. Reading it physically: $H$ is manifestly non-negative only because the anticommutator flipped the antiparticle sign; the same field with commutators would have $H$ with no floor.

Derivation APositivity forces anticommutators for the Dirac field

This is the article's central result and the operational content of the spin-statistics connection. The strategy is to expand the Dirac Hamiltonian in modes, isolate the antiparticle piece, and rewrite it two ways — once with a commutator, once with an anticommutator — then demand the spectrum be bounded below. Only one choice survives, and it is fixed by the spin of the field, not by taste. Watch the single minus sign that the antiparticle carries; everything turns on it.

Step 1. Insert (7) into $H=\int\!\mathrm d^3x\,\psi^\dagger(-i\gamma^0\boldsymbol\gamma\cdot\nabla+\gamma^0 m)\psi$ and use the spinor orthogonality relations. The particle modes give $+E_\mathbf{p}\,c^\dagger c$ , and the antiparticle modes give, before choosing a bracket, the raw form $\displaystyle H\supset\int\frac{\mathrm d^3p}{(2\pi)^3}\,E_\mathbf{p}\sum_s\big(c^{s\dagger}_\mathbf{p}c^s_\mathbf{p}-d^s_\mathbf{p}d^{s\dagger}_\mathbf{p}\big)$ . Why: the $v$ -spinors enter with the opposite frequency sign, so their contribution comes out as $-d\,d^\dagger$ , not $-d^\dagger d$ . That ordering — annihilator to the left — is the crux, because to interpret it we must reorder using a bracket.

Step 2. Try the boson choice, a commutator. Using $d\,d^\dagger=d^\dagger d+[d,d^\dagger]$ , the antiparticle piece becomes $-d\,d^\dagger=-d^\dagger d-[d,d^\dagger]$ . With $[d,d^\dagger]=(2\pi)^3\delta^3(0)$ a c-number, the operator part is $-E_\mathbf{p}\,d^\dagger d$ . Why: the number operator $d^\dagger d$ now multiplies $-E_\mathbf{p}$ . Piling up antiparticles lowers the energy without bound — the Hamiltonian has no floor, the vacuum is unstable, and the theory is dead.

Step 3. Try the fermion choice, an anticommutator. Using $d\,d^\dagger=-d^\dagger d+\{d,d^\dagger\}$ , the antiparticle piece becomes $-d\,d^\dagger=+d^\dagger d-\{d,d^\dagger\}$ . With $\{d,d^\dagger\}=(2\pi)^3\delta^3(0)$ a c-number, the operator part is $+E_\mathbf{p}\,d^\dagger d$ . Why: the single sign flip inside the reordering is the whole story. The antiparticle number operator now enters with $+E_\mathbf{p}$ , so both particles and antiparticles cost positive energy and the spectrum is bounded below.

Step 4. Normal-order away the (infinite, constant) $\{d,d^\dagger\}$ term to reach $\displaystyle H=\int\frac{\mathrm d^3p}{(2\pi)^3}E_\mathbf{p}\sum_s\big(c^{s\dagger}c^s+d^{s\dagger}d^s\big)\ge0$ , and note $(d^\dagger)^2=0$ falls out of $\{d^\dagger,d^\dagger\}=0$ — the exclusion principle, built into the algebra. Why: the constant is unobservable (only energy differences matter), so dropping it is legitimate; what remains is manifestly non-negative. The identical calculation on the scalar shows that there the reversed-order piece comes with $+a\,a^\dagger$ (the real field of eqs. (1)–(5) is its own antiparticle; a complex scalar hands the same sign to separate antiparticle operators $b,b^\dagger$ ), so a commutator (not an anticommutator) is what gives positive energy — the mirror image of the fermion case. A single $d$ -mode makes this concrete: $[d,d^\dagger]=1$ has no finite-dimensional representation (a commutator is traceless, but $\mathrm{tr}\,\mathbb{1}=N$ ), so the commutator route is a bosonic ladder on which $-E\,d\,d^\dagger$ has eigenvalues $-E(n+1)$ , $n=0,1,2,\dots$ — unbounded below; the anticommutator algebra is exactly two-dimensional, where $-E\,d\,d^\dagger$ has eigenvalues $\{-E,0\}$ and the normal-ordered $+E\,d^\dagger d$ has eigenvalues $\{0,+E\}\ge0$ .

Go deeper · Derivation B · 1 minThe scalar Hamiltonian and the on-shell cancellation

Derivation BThe scalar Hamiltonian and the on-shell cancellation

Here we reduce the classical scalar energy $H=\int\!\mathrm d^3x\,[\tfrac12\pi^2+\tfrac12(\nabla\phi)^2+\tfrac12 m^2\phi^2]$ to the diagonal oscillator form (5), and see two features emerge cleanly: the dangerous cross terms $aa$ and $a^\dagger a^\dagger$ cancel because the modes are on shell, and the surviving symmetric expression contains the zero-point energy that normal ordering discards. The strategy is to substitute the mode expansion, do the spatial integrals to get momentum-conserving deltas, and collect by operator type.

Step 1. Substitute (3) into $H$ . Each $\int\!\mathrm d^3x$ produces $(2\pi)^3\delta^3(\mathbf{p}\pm\mathbf{p}')$ , pairing a mode with $+\mathbf{p}'$ (giving $a^\dagger a$ and $a a^\dagger$ ) or with $-\mathbf{p}'$ (giving $aa$ and $a^\dagger a^\dagger$ ). Why: the delta functions collapse the double momentum integral to a single one; the two pairings are the "diagonal" and "off-diagonal" contractions.

Step 2. Collect the off-diagonal $aa$ and $a^\dagger a^\dagger$ pieces. Their coefficient is $\big(-\omega_\mathbf{p}^2+\mathbf{p}^2+m^2\big)$ , which vanishes because $\omega_\mathbf{p}^2=\mathbf{p}^2+m^2$ on shell. Why: this is a real consistency check, not an accident: it confirms the mode frequency is exactly the relativistic energy of The Klein-Gordon Equation. Off-shell, these terms would survive and the field would not be a clean oscillator sum.

Step 3. The survivors give $H=\int\!\frac{\mathrm d^3p}{(2\pi)^3}\,\omega_\mathbf{p}\big(a^\dagger a+\tfrac12[a,a^\dagger]\big)$ . The bracket term is $\tfrac12(2\pi)^3\delta^3(0)$ per mode — an infinite constant, the zero-point energy. Why: every oscillator contributes its $\tfrac12\omega$ , and summing over the continuum of modes diverges. It is a c-number shift of the whole spectrum, so it drops out of every measurable energy difference.

Step 4. Normal-order: define $:\!a a^\dagger\!:=a^\dagger a$ (annihilators to the right), giving $H=\int\!\frac{\mathrm d^3p}{(2\pi)^3}\,\omega_\mathbf{p}\,a^\dagger_\mathbf{p}a_\mathbf{p}\ge0$ . Why: dropping the vacuum constant is legitimate because it is unobservable in isolation; its differences between boundary conditions are the Casimir effect. The final spectrum is a sum of positive quanta — Bose statistics, positive energy. The on-shell cancellation $(-\omega^2+p^2+m^2)=0$ holds symbolically, and a single-mode oscillator has $H=\omega(a^\dagger a+\tfrac12)$ with spectrum $\omega(n+\tfrac12)\ge0$ .

Limiting cases

case	result	meaning
nonrelativistic limit $\omega_\mathbf{p}\to m+\mathbf{p}^2/2m$	antiparticle modes cost $\ge 2mc^2$ and decouple	reduces to fixed-number many-body second quantization
free / on-shell $\omega_\mathbf{p}=\sqrt{\mathbf{p}^2+m^2}$	each mode an independent oscillator	exactly solvable; the S-matrix is trivial (no interactions)
massless scalar $\omega_\mathbf{p}=\vert\mathbf{p}\vert$	gapless, long-range excitations	quantization identical; infrared subtleties only
spin-0 vs spin-½	$[\,,\,]$ vs $\{\,,\,\}$	the wrong bracket gives negative-norm states or energy with no floor
discard zero-point $\tfrac12\sum\omega_\mathbf{p}$	absolute value unobservable	but differences are the measurable Casimir force

Common trapThe Dirac field cannot be quantized with commutators "by analogy"

The most tempting error is to copy the scalar recipe onto the Dirac field and impose commutators. It fails catastrophically: the antiparticle term becomes $-d^\dagger d$ , so the energy is unbounded below and the vacuum decays. Symmetrically, forcing anticommutators onto the scalar gives negative-norm states and violates microcausality. There is no freedom in the choice — the spin of the field dictates the bracket. Integer spin demands commutators (Bose); half-integer spin demands anticommutators (Fermi). Getting it backwards does not give a "different theory," it gives no theory. The Lorentz-invariant proof that this is forced belongs to Spin-Statistics and the CPT Theorem; here we have shown the positivity half of it.

Worked examples

Example 1 — the single boson mode: from $[\phi,\pi]=i$ to $H=\omega(a^\dagger a+\tfrac12)$ . Discretize the scalar field to one oscillator to see the whole machine in miniature. Set $a=(\omega\phi+i\pi)/\sqrt{2\omega}$ , $a^\dagger=(\omega\phi-i\pi)/\sqrt{2\omega}$ .

Step 1. Compute the commutator: $[a,a^\dagger]=\dfrac{1}{2\omega}[\omega\phi+i\pi,\,\omega\phi-i\pi]=\dfrac{1}{2\omega}\big(-i\omega[\phi,\pi]+i\omega[\pi,\phi]\big)=\dfrac{1}{2\omega}(2\omega)=\mathbf{1}$ .

Step 2. Invert: $\phi=(a+a^\dagger)/\sqrt{2\omega}$ , $\pi=-i\sqrt{\omega/2}\,(a-a^\dagger)$ . Substitute into the single-mode energy $H=\tfrac12\pi^2+\tfrac12\omega^2\phi^2$ .

Step 3. Collecting terms gives $H=\tfrac12\omega(aa^\dagger+a^\dagger a)=\omega\big(a^\dagger a+\tfrac12[a,a^\dagger]\big)=\omega\big(a^\dagger a+\tfrac12\big)$ . The spectrum is $E_n=\omega(n+\tfrac12)$ , the familiar oscillator ladder — verified numerically to give $E_0=\omega/2$ and equal spacing $\omega$ . This is the field-theory content of Creation and Annihilation Operators running one mode at a time.

Example 2 — the positivity fork with real numbers. Take one antiparticle $d$ -mode with energy $E_\mathbf{p}=2.3$ (natural units) and evaluate the raw antiparticle piece $-E\,d\,d^\dagger$ under each bracket choice — the infinite bosonic ladder for the commutator, the exact two-state space $\{|0\rangle,|1\rangle\}$ for the anticommutator — then compare.

Step 1. With the commutator reordering, $-E\,d\,d^\dagger=-E\,d^\dagger d-E[d,d^\dagger]$ . Acting on the vacuum, $d\,d^\dagger|0\rangle=|0\rangle$ , so the vacuum sits at energy $\mathbf{-E=-2.3}$ — below the "empty" state. And since $[d,d^\dagger]=1$ admits no finite matrix representation (a commutator is traceless, but $\mathrm{tr}\,\mathbb{1}=N$ ), the mode is a full bosonic ladder: the raw piece has eigenvalues $-E(n+1)=\{-2.3,-4.6,-6.9,\dots\}$ , and unlimited occupation of even this single mode drives the energy to $-\infty$ .

Step 2. With the anticommutator reordering, $-E\,d\,d^\dagger=+E\,d^\dagger d-E\{d,d^\dagger\}$ . Dropping the constant $E\{d,d^\dagger\}$ leaves $+E\,d^\dagger d$ , whose eigenvalues are $\{0,\,+2.3\}$ : the vacuum is the lowest state and every antiparticle costs $+2.3$ .

Step 3. Read the verdict: the commutator spectrum has no minimum at all — it runs to $-\infty$ — while $\min(\text{anticommutator})=0$ . Only the anticommutator produces a stable, bounded-below theory — statistics is not chosen, it is forced by demanding a floor on the energy.

Origins · 1 mincanonical quantization and how this was discovered

canonical quantization

1927–1929 · The procedure of promoting classical fields to operators with equal-time (anti)commutators, developed in the founding papers of quantum field theory

Canonical field quantization was built in a rush of papers between 1927 and 1929. Dirac quantized the electromagnetic field in 1927 (Proc. R. Soc. A 114, 243) — the first "second quantization," done for photons before the electron equation that bears his name. Jordan and Klein (1927) formalized the many-boson version, and Jordan and Wigner (1928, Z. Phys. 47, 631) introduced anticommutators for fermions, encoding the exclusion principle in the algebra. Heisenberg and Pauli (1929) then set out the general canonical scheme for interacting fields. The deep reason the bracket is not a free choice — that spin dictates statistics — was proved by Pauli in 1940.

History

1927

— Dirac quantizes the electromagnetic field, the original "second quantization"

1927

— Jordan and Klein formulate the second-quantized many-boson theory

1928

— Jordan and Wigner introduce anticommutators for fermionic fields

1929

— Heisenberg and Pauli give the general canonical quantization of interacting fields

1932

— Fermi's influential review recasts quantum electrodynamics in quantized-field language

1940

— Pauli proves the spin-statistics theorem, showing the bracket choice is forced

Going further

This article promotes the wave equations of The Klein-Gordon Equation and The Dirac Equation to quantized fields, resolving the crises catalogued in Negative-Energy Solutions and the Interpretive Crisis, Klein-Gordon Probability Current and Negative Densities, and The Dirac Sea and Antiparticles — the negative-energy tower is gone because the vacuum is the lowest state, and antiparticles enter symmetrically through the $d^\dagger$ operators rather than as holes in a filled sea. The Fock-space bookkeeping of occupation numbers and the isomorphism to first-quantized many-body states is developed in Second Quantization and Fock Space; the ladder algebra itself is owned by Creation and Annihilation Operators. The positivity argument here is only half of the spin-statistics connection: the full Lorentz-invariant proof, together with the CPT theorem, is the subject of Spin-Statistics and the CPT Theorem. From there the field theory that follows takes up interactions, the S-matrix, and renormalization — the machinery of the Standard Model. Note on conventions: this article uses the mostly-minus metric $g^{\mu\nu}=\mathrm{diag}(+,-,-,-)$ with $p\cdot x=\omega t-\mathbf{p}\cdot\mathbf{x}$ consistently throughout. All commutators, signs, and positivity claims — including the single-mode $d$ fork, the on-shell cross-term cancellation $(-\omega^2+p^2+m^2)=0$ , the boson spectrum $\omega(n+\tfrac12)$ , and $(b^\dagger)^2=0$ — have been verified.

Prerequisites. The Klein-Gordon Equation · The Dirac Equation · Dirac Spinors and Plane-Wave Solutions · Creation and Annihilation Operators · Second Quantization and Fock Space. Leads to. Spin-Statistics and the CPT Theorem · The Feynman Propagator and Causality · quantum electrodynamics · the Standard Model.

Why is this called "second" quantization when there is only one quantization?

The name is historical and slightly misleading. "First quantization" turned a classical particle's position and momentum into operators, producing the Schrödinger wavefunction $\psi(x)$ . "Second quantization" then treats that very wavefunction as a classical field and quantizes it — promoting $\psi$ itself to an operator. There is really only one quantization principle (impose canonical brackets on conjugate variables); it is just being applied a second time, now to a field. Modern practice quantizes the field directly from its Lagrangian and skips the "wavefunction" step entirely, so many texts prefer "canonical field quantization." The physical payoff is what matters: variable particle number, which single-particle quantum mechanics cannot describe.

If the zero-point energy is infinite and we just throw it away, is that legitimate?

Throwing away its absolute value is legitimate because only energy differences are ever measured — you can add any constant to a Hamiltonian without changing dynamics. Normal ordering does exactly this, systematically. What is not legitimate is to claim the vacuum energy has no physical consequences at all: differences in zero-point energy between two boundary configurations are real and measurable. The Casimir force between two conducting plates is precisely such a difference, computed by regularizing the divergent sum $\tfrac12\sum\omega_\mathbf{p}$ with and without the plates and subtracting. The infinity cancels; a finite, attractive, experimentally confirmed force remains. Gravity is the uncomfortable exception, since it couples to the absolute energy density — the cosmological-constant problem.

How exactly does quantization make the negative-energy problem disappear?

In the single-particle Dirac theory the Hamiltonian has eigenvalues $\pm E_\mathbf{p}$ , so there are states of arbitrarily negative energy and the ground state does not exist. After field quantization the spectrum is built by number operators: $H=\int E_\mathbf{p}(c^\dagger c+d^\dagger d)$ with $E_\mathbf{p}=+\sqrt{\mathbf{p}^2+m^2}>0$ . The lowest state is the vacuum with zero quanta, energy zero, and every excitation adds positive energy. The negative-frequency solution $e^{+ip\cdot x}$ is still there in the field expansion, but its operator coefficient is a creation operator for a positive-energy antiparticle, not an annihilator for a negative-energy particle. So the tower of negative-energy states is not filled (Dirac sea) — it is simply reinterpreted out of existence.

Do the negative-frequency modes carry negative energy or not?

The wave $e^{+ip\cdot x}$ has $p^0=+\omega_\mathbf{p}>0$ ; the label "negative-frequency" refers to the sign in the exponent, not the energy. Its operator coefficient $d^\dagger$ creates a quantum, and by the Hamiltonian (9) that quantum has energy $+E_\mathbf{p}$ . So the mode contributes positive energy. The confusion is a relic of the single-particle reading, where one wrongly asked "what energy does this wave describe?" and got $-E$ . The correct field-theory question is "what does its coefficient do to the state?" — and the answer is: it adds a positive-energy antiparticle. This reinterpretation is the whole point of promoting $\psi$ to a field.

Why does the scalar need commutators while the spinor needs anticommutators — can I see it without spin?

The sign that decides everything comes from how the antiparticle piece appears in the Hamiltonian, and that sign is tied to the field's Lorentz representation. For the scalar the reversed-order term arrives as $+\omega\,a\,a^\dagger$ (for the real field the quantum is its own antiparticle; a complex scalar gives the same sign for its antiparticle operators), so you need $a\,a^\dagger=a^\dagger a+[a,a^\dagger]$ — a commutator — to reach $+\omega\,a^\dagger a\ge0$ . For the spin-½ field the $v$ -spinor orthogonality flips the sign to $-E\,d\,d^\dagger$ , so you need $d\,d^\dagger=-d^\dagger d+\{d,d^\dagger\}$ — an anticommutator — to reach $+E\,d^\dagger d\ge0$ . The relative minus sign traces back to the spinors, i.e. to the spin. You cannot see it entirely without spin, because the spin is the origin of the sign; but the positivity logic — "which reordering gives $+$ number $\times E$ ?" — is universal and is exactly what the worked example computes.

Is microcausality the same statement as "no faster-than-light signaling"?

Very nearly. Microcausality is the operator statement that field operators (or observables built from them) at spacelike-separated points commute — for bosons $[\phi(x),\phi(y)]=0$ when $(x-y)^2<0$ . Physically this guarantees that a measurement at $x$ cannot influence the statistics of a measurement at spacelike-separated $y$ , which is the operational content of "no superluminal signaling." The Pauli-Jordan function $\Delta(x-y)$ that measures this commutator is Lorentz-invariant and vanishes outside the light cone: boost to the equal-time frame and the two frequency exponentials cancel. For fermions the fields themselves anticommute at spacelike separation, but physical observables are even in fermion fields (bilinears), so the observables still commute — causality is preserved. Getting the bracket wrong breaks exactly this cancellation, which is why the Spin-Statistics and the CPT Theorem proof uses causality as one of its hypotheses.

What is normal ordering doing to the operator, precisely?

Normal ordering, written $:\;:$ , reorders a product of creation and annihilation operators so that every creation operator sits to the left of every annihilation operator, as if they commuted (bosons) or anticommuted (fermions) — dropping the c-number contractions. For a boson mode $:\!a a^\dagger\!:=a^\dagger a$ , which removes the $\tfrac12[a,a^\dagger]$ zero-point piece from the Hamiltonian; for fermions the reordering carries the extra minus signs of the anticommutator. It is a bookkeeping operation that discards the unobservable vacuum expectation value, ensuring $\langle 0|:\!H\!:|0\rangle=0$ . Interactions are almost always written normal-ordered so the vacuum has zero energy by construction.