Klein-Gordon equation in external EM field: Coulomb and uniform-magnetic cases

Intuition Beginner

Special relativity ties energy and momentum together as $E^2=p^2{c}^2+m^2{c}^4$. The simplest way to make this rule into a quantum wave equation is to promote energy and momentum to differential operators — one time derivative for energy, three space derivatives for momentum — and write down the resulting second-order wave equation. The result is the Klein-Gordon equation, the simplest relativistic quantum wave equation. Schrödinger tried it first, before he settled on the non-relativistic equation that bears his name. It works for spinless particles: the pions, kaons, and Higgs boson all satisfy a Klein-Gordon equation in their free-particle theories.

To include an external electromagnetic field, use the standard rule from electromagnetism called minimal coupling: replace ordinary momentum with the canonical momentum $\mathbf{p}-e\mathbf{A}/c$ and ordinary energy with $E-e{A}_0$. The Klein-Gordon equation then describes a charged spin-0 particle moving in the external potentials $A_0$ (electric) and $\mathbf{A}$ (magnetic).

Two cases let you solve this equation exactly. The first is a Coulomb potential $A_0=-Ze/r$, the field around a point nucleus. This describes a negatively charged pion orbiting a proton — a pionic atom. The spectrum looks like hydrogen's but with relativistic corrections, matching the formula Sommerfeld guessed in 1916 by adding relativistic kinematics to Bohr's orbits.

The angular-momentum label in the formula is $\ell$ (the orbital one), not $j$ (the total one) as in the Dirac case — because spin-0 particles have no spin to contribute. This is exactly why Schrödinger's first attempt at relativistic quantum mechanics did not match experiment for hydrogen: the electron has spin, and the spin contribution to fine structure was missing.

The second case is a uniform magnetic field $\mathbf{B}=B\hat{z}$. Charged particles in a magnetic field move in circles classically (cyclotron motion). Quantum mechanically — even non-relativistically — the energy quantises into evenly spaced Landau levels $E_n=\hslash {\omega }_c(n+1/2)$ with cyclotron frequency ${\omega }_c=eB/(mc)$. The relativistic Klein-Gordon version keeps the same $(n+1/2)$ level structure but puts it inside a square root: $E_n^2=m^2{c}^4+p_z^2{c}^2+2eB\hslash c(n+1/2)$. At large $n$ or strong $B$ the relativistic and non-relativistic predictions differ noticeably; at weak field and small $n$ they agree to leading order.

The Klein-Gordon equation has known issues. Its conserved current can be negative (which a probability current cannot), and at strong fields the wave function can "leak" through barriers in unphysical ways (the Klein paradox). These problems are resolved by treating the field as a quantised object — the route taken in quantum field theory. But the bound-state spectra derived here remain physically correct: they describe the energy levels of a charged spin-0 particle in an external field, with the right experimental predictions for pionic atoms.

Visual Beginner

The left panel shows the spin-0 fine-structure: $\ell$-dependence within each principal shell, with the $2s$ above the $2p$ (opposite to the Dirac case, where the analogue $2s_{1/2}$ and $2p_{1/2}$ are degenerate). The right panel shows the relativistic and non-relativistic Landau spectra: at small $n$ they coincide (the square root expansion gives $E_n\approx m+eB(n+1/2)/m$); at large $n$ the relativistic levels saturate to $\sqrt{2eBn}$ while the non-relativistic levels grow linearly without bound — an artefact of the non-relativistic approximation in strong fields.

Worked example Beginner

Compute the ground-state binding energy of pionic hydrogen (${\pi }^{-}$ bound to a proton) using the spin-0 Sommerfeld formula.

Pionic hydrogen is a ${\pi }^{-}$ meson (mass $m_{\pi }{c}^2\approx 139.57$ MeV) orbiting a proton (mass $m_p{c}^2\approx 938.27$ MeV). The pion is much lighter than the proton, so the proton acts as a fixed point charge of $+e$ at the origin, and the pion sees a Coulomb potential $V=-e^2/(4\pi {ϵ}_{0}r)$. Use natural units with $Z=1$ and the fine-structure constant $\alpha =e^2/(4\pi \hslash c)\approx 1/137.036$.

The Klein-Gordon Sommerfeld formula for the ground state $n=1,\ell =0$ is

$$E_{1,0}=m_{\pi }{c}^2{\left[1+\frac{{\alpha }^2}{{\left(1-\frac{1}{2}+\sqrt{(1/2{)}^2-{\alpha }^2}\right)}^2}\right]}^{-1/2}.$$

Numerically, with ${\alpha }^2\approx 5.33\times 10^{-5}$:

$$\sqrt{1/4-{\alpha }^2}\approx 0.5-{\alpha }^2\approx 0.5-5.33\times 10^{-5}\approx 0.4999467,$$ $${\left(1/2+\sqrt{1/4-{\alpha }^2}\right)}^2\approx (0.9999467{)}^2\approx \mathrm{0.9998934.}$$

So

$$E_{1,0}\approx m_{\pi }{c}^2\cdot {\left[1+\frac{{\alpha }^2}{0.9999}\right]}^{-1/2}\approx m_{\pi }{c}^2\left[1-\frac{{\alpha }^2}{2}(1.0001)+\frac{3{\alpha }^4}{8}+\cdots \right].$$

Plugging in $m_{\pi }{c}^2=139.57$ MeV:

$$E_{1,0}\approx 139.57\text{MeV}-\frac{139.57\times 5.33\times 10^{-5}}{2}\text{MeV}\approx 139.57\text{MeV}-3.72\text{keV}.$$

The binding energy is $\sim 3.72$ keV, much larger than electronic hydrogen's 13.6 eV because the pion is $m_{\pi }/m_e\approx 273$ times heavier. The Bohr radius of pionic hydrogen scales inversely: $a_{\pi }=\hslash /(m_{\pi }c\alpha )\approx a_0\cdot m_e/m_{\pi }\approx 5.29\times 10^{-11}/273\approx 222$ fm, less than the typical electronic orbit size by the same factor.

What this tells us: pionic atoms are much smaller and more tightly bound than ordinary atoms, with the inner-shell orbits comparable in size to the nuclear radius itself. This makes pionic spectroscopy a probe of nuclear structure — the inner-shell levels shift noticeably depending on the finite extent and the strong-interaction sensitivity of the nucleus, an effect that bare electronic hydrogen does not see. The 3.72 keV figure is in the X-ray regime and is the basis for pionic-atom spectroscopy at meson factories.

Check your understanding Beginner

Formal definition Intermediate+

Work in natural units $\hslash =c=1$ with mostly-minus metric ${\eta }^{\mu \nu }=\mathrm{diag}(+1,-1,-1,-1)$. The Klein-Gordon equation for a complex scalar field $\psi$ of mass $m$ and charge $e$ in an external electromagnetic four-potential $A_{\mu }$ is obtained by minimal coupling from the free equation $({\partial }^{\mu }{\partial }_{\mu }+m^2)\psi =0$:

$$\boxed{[(i{\partial }_{\mu }-e{A}_{\mu })(i{\partial }^{\mu }-e{A}^{\mu })-m^2]\psi (x)=0.}$$

Expanding the square and using ${\eta }^{\mu \nu }$ to raise indices,

$$[-{\partial }_{\mu }{\partial }^{\mu }-2ie{A}^{\mu }{\partial }_{\mu }-ie({\partial }_{\mu }{A}^{\mu })+e^2{A}_{\mu }{A}^{\mu }-m^2]\psi =0.$$

In Lorenz gauge ${\partial }_{\mu }{A}^{\mu }=0$ the middle term simplifies. The equation is second-order in time, second-order in space, and gauge-covariant: under $A_{\mu }\to A_{\mu }+{\partial }_{\mu }\chi$, $\psi \to e^{ie\chi }\psi$ the equation is invariant.

The conserved current associated with the global $U(1)$ symmetry is

$$j^{\mu }=i[{\psi }^{\ast }(D^{\mu }\psi )-(D^{\mu }\psi {)}^{\ast }\psi ],D^{\mu }:={\partial }^{\mu }+ie{A}^{\mu },$$

satisfying ${\partial }_{\mu }{j}^{\mu }=0$ on solutions. Unlike the Schrödinger probability current, $j^0$ is not positive-definite: it can be negative for certain solutions, a feature traditionally interpreted as evidence that the single-particle Klein-Gordon equation breaks down at the strong-field or pair-creation threshold and must be replaced by quantum field theory. Within the single-particle theory, $e{j}^{\mu }$ is a conserved electric four-current and the bound-state spectra below remain physically meaningful as eigenvalues of the relevant stationary problem.

Coulomb case: stationary-state ansatz

For a static central Coulomb potential created by a point nucleus of charge $+Ze$,

$$A_0=-\frac{Ze}{r},\mathbf{A}=0,e{A}_0=-\frac{Z\alpha }{r},$$

where $\alpha =e^2/(4\pi )$ in natural Heaviside-Lorentz units. For a stationary state $\psi (\mathbf{r},t)=e^{-iEt}\psi (\mathbf{r})$, the spatial equation becomes

$$\left[(E+Z\alpha /r{)}^2+{\nabla }^2-m^2\right]\psi (\mathbf{r})=0,$$

i.e.,

$$\left[{\nabla }^2+E^2-m^2+\frac{2EZ\alpha }{r}+\frac{(Z\alpha {)}^2}{r^2}\right]\psi (\mathbf{r})=0.$$

This is a Schrödinger-like equation with effective energy $\mathcal{E}=(E^2-m^2)/2m$ (bound states have $|E|<m$ so $\mathcal{E}<0$), an effective Coulomb potential ${\tilde{V}}_{\text{eff}}(r)=-EZ\alpha /(mr)$, and an additional inverse-square term $-(Z\alpha {)}^2/(2m{r}^2)$ that effectively shifts the angular momentum from $\ell$ to a non-integer value.

Separation of variables in spherical coordinates with $\psi (\mathbf{r})=R(r)Y_{\ell ,m}(\theta ,\varphi )$ reduces the equation to the radial form

$$\boxed{\left[\frac{d^2}{d{r}^2}+\frac{2}{r}\frac{d}{dr}+E^2-m^2+\frac{2EZ\alpha }{r}-\frac{\ell (\ell +1)-(Z\alpha {)}^2}{r^2}\right]R(r)=0.}$$

The combination $\ell (\ell +1)-(Z\alpha {)}^2$ in the centrifugal-barrier coefficient is the structural difference from the non-relativistic Schrödinger-Coulomb problem and is what shifts the spin-0 spectrum away from the Schrödinger result.

Magnetic case: gauge-fixed harmonic-oscillator form

For a uniform magnetic field $\mathbf{B}=B\hat{z}$, choose the symmetric gauge

$$\mathbf{A}=\frac{B}{2}(-y,x,0),A_0=0,$$

so that $\nabla \times \mathbf{A}=B\hat{z}$. The Klein-Gordon equation for a stationary state $\psi (\mathbf{r},t)=e^{-iEt}\psi (\mathbf{r})$ becomes

$$\left[E^2-(\mathbf{p}-e\mathbf{A}{)}^2-m^2\right]\psi (\mathbf{r})=0.$$

Translation invariance in $z$ gives $\psi (\mathbf{r})=e^{i{p}_{z}z}\varphi (x,y)$, so the equation reduces to a two-dimensional problem in the $(x,y)$ plane:

$$\left[E^2-p_z^2-m^2\right]\varphi =({\mathbf{p}}_{\perp }-e\mathbf{A}{)}^2\varphi .$$

The operator $H_{\perp }:=({\mathbf{p}}_{\perp }-e\mathbf{A}{)}^2$ is identical to the non-relativistic charged-particle-in-magnetic-field Hamiltonian (up to a factor of $2m$), with the well-known harmonic-oscillator algebraic structure. Its eigenvalues are $2eB(n+1/2)$ for $n=0,1,2,\dots$, each with an infinite degeneracy per unit transverse area $eB/(2\pi )$.

Key derivation Intermediate+

Theorem 1 (spin-0 Sommerfeld fine-structure formula). The discrete bound-state spectrum of the Klein-Gordon-Coulomb problem for $0<Z\alpha <1/2$ (when $\ell =0$) or $0<Z\alpha <\ell +1/2$ (for general $\ell$) is

$$\boxed{E_{n,\ell }=m{\left[1+\frac{(Z\alpha {)}^2}{(n-\ell -\frac{1}{2}+\sqrt{(\ell +\frac{1}{2}{)}^2-(Z\alpha {)}^2}{)}^2}\right]}^{-1/2},}$$

where $n=1,2,3,\dots$ is the principal quantum number and $\ell =0,1,\dots ,n-1$ runs over the allowed orbital angular momenta.

Proof. Solve the radial equation by the standard confluent-hypergeometric ansatz. Introduce the scaled coordinate $\rho :=2\lambda r$ where $\lambda :=\sqrt{m^2-E^2}$ (real and positive for bound states $|E|<m$), and the dimensionless parameter $\nu :=EZ\alpha /\lambda$. The radial equation becomes

$$\frac{d^{2}R}{d{\rho }^2}+\frac{2}{\rho }\frac{dR}{d\rho }+\left[-\frac{1}{4}+\frac{\nu }{\rho }-\frac{\ell (\ell +1)-(Z\alpha {)}^2}{{\rho }^2}\right]R=0.$$

Look for asymptotic behaviour. As $\rho \to \infty$ the equation reduces to $R^{\prime \prime }-R/4\sim 0$, with bounded solution $R\sim e^{-\rho /2}$. As $\rho \to 0$, the centrifugal term dominates: trying $R\sim {\rho }^s$ gives the indicial equation $s(s+1)=\ell (\ell +1)-(Z\alpha {)}^2$, with two roots

$$s_{\pm }=-\frac{1}{2}\pm \sqrt{(\ell +\frac{1}{2}{)}^2-(Z\alpha {)}^2}.$$

The regular root is

$$s_{+}=-\frac{1}{2}+\sqrt{(\ell +\frac{1}{2}{)}^2-(Z\alpha {)}^2}=:\gamma -\frac{1}{2},$$

where $\gamma :=\sqrt{(\ell +1/2{)}^2-(Z\alpha {)}^2}$. Normalisability ${\int }_0^{\infty }{R}^2{r}^{2}dr<\infty$ requires $s_{+}>-3/2$, comfortably satisfied for $0<Z\alpha <\ell +1/2$.

Write $R(\rho )={\rho }^{s_{+}}{e}^{-\rho /2}F(\rho )$ and substitute. The function $F$ satisfies Kummer's confluent hypergeometric equation

$$\rho F^{\prime \prime }+(2\gamma +1-\rho )F^{\prime }-(\gamma +1/2-\nu )F=0,$$

whose regular-at-origin solution is the Kummer function $F(\rho )={}_1{F}_1(\gamma +1/2-\nu ,2\gamma +1;\rho )$. For the wavefunction to remain bounded at $\rho \to \infty$, the Kummer series must terminate (otherwise $F\sim e^{\rho }$ at infinity, cancelling the $e^{-\rho /2}$ factor and producing $R\sim e^{+\rho /2}$, contradicting normalisability). Termination requires

$$\gamma +\frac{1}{2}-\nu =-n_r,n_r=0,1,2,\dots ,$$

i.e., $\nu =n_r+\gamma +1/2$. Recall $\nu =EZ\alpha /\lambda =EZ\alpha /\sqrt{m^2-E^2}$, so

$$EZ\alpha =\sqrt{m^2-E^2}\cdot (n_r+\gamma +1/2).$$

Squaring:

$$E^2(Z\alpha {)}^2=(m^2-E^2)(n_r+\gamma +1/2{)}^2,$$ $$E^2[(Z\alpha {)}^2+(n_r+\gamma +1/2{)}^2]=m^2(n_r+\gamma +1/2{)}^2,$$ $$E=m{\left[1+\frac{(Z\alpha {)}^2}{(n_r+\gamma +1/2{)}^2}\right]}^{-1/2}.$$

Identify the principal quantum number $n:=n_r+\ell +1$, so $n_r+1/2=n-\ell -1/2$. The denominator becomes

$$n_r+\gamma +1/2=n-\ell -1/2+\gamma =n-\ell -1/2+\sqrt{(\ell +1/2{)}^2-(Z\alpha {)}^2},$$

which reproduces the boxed spin-0 Sommerfeld formula. $\square$

Corollary 1 (non-relativistic limit). Expanding to leading order in $Z\alpha$, $E_{n,\ell }\to m-m(Z\alpha {)}^2/(2n^2)$, the Schrödinger-Coulomb spectrum 12.06.01 depending on $n$ alone.

Corollary 2 (fine-structure expansion). $$ E_{n,\ell} = m\left[1 - \frac{(Z\alpha)^2}{2 n^2} - \frac{(Z\alpha)^4}{2 n^3}\left(\frac{1}{\ell + 1/2} - \frac{3}{4 n}\right) + O((Z\alpha)^6)\right]. $$

The structural form is identical to the Dirac case 12.11.03 except $j\to \ell$. For $\ell =0$ states the spin-0 correction has $1/(\ell +1/2)=2$, larger than the Dirac $1/(j+1/2)=2$ (for $j=1/2$); for higher $\ell$ states the spin-0 and Dirac fine-structure coefficients differ noticeably.

Theorem 2 (relativistic Landau levels). The discrete bound-state spectrum of the Klein-Gordon equation in a uniform magnetic field $\mathbf{B}=B\hat{z}$ at fixed longitudinal momentum $p_z$ is

$$\boxed{E_n^2=m^2+p_z^2+2eB(n+1/2),n=0,1,2,\dots ,}$$

with infinite degeneracy per unit transverse area equal to $eB/(2\pi )$.

Proof. The reduced two-dimensional equation $({\mathbf{p}}_{\perp }-e\mathbf{A}{)}^2\varphi =(E^2-p_z^2-m^2)\varphi$ involves the operator $H_{\perp }=({\mathbf{p}}_{\perp }-e\mathbf{A}{)}^2$. Introduce the magnetic length ${\ell }_B:=1/\sqrt{eB}$ and the complex coordinate $w:=(x+iy)/(\sqrt{2}{\ell }_B)$. In the symmetric gauge, $H_{\perp }$ can be written in terms of ladder operators

$$a:=\frac{1}{\sqrt{2}}(\bar{\partial }w+\frac{1}{2}w),a^{†}:=\frac{1}{\sqrt{2}}(-{\partial }_w+\frac{1}{2}\bar{w}),$$

satisfying $[a,a^{†}]=1$, and

$$H_{\perp }=\frac{2}{{\ell }_B^2}(a^{†}a+1/2)=2eB(a^{†}a+1/2).$$

This is a quantum harmonic oscillator 12.04.02 with frequency $2eB$, with spectrum $2eB(n+1/2)$ for $n=0,1,2,\dots$. The Klein-Gordon eigenvalue condition $E^2-p_z^2-m^2=2eB(n+1/2)$ produces the boxed Landau spectrum.

The infinite degeneracy is the standard Landau degeneracy: a second set of ladder operators $b,b^{†}$ (built from the orthogonal linear combinations of $\bar{\partial }$ and $w$) commutes with $H_{\perp }$ and labels degenerate states. In a finite-area sample of area $A$, the number of degenerate states in each Landau level is $A\cdot eB/(2\pi )$, the magnetic flux through $A$ divided by the flux quantum $h/e=2\pi$ (in $\hslash =1$ units). $\square$

Corollary 3 (non-relativistic limit, magnetic case). Expanding $E_n=\sqrt{m^2+p_z^2+2eB(n+1/2)}\approx m+p_z^2/(2m)+(eB/m)(n+1/2)+O(1/m^3)$ at $eB\ll m^2$, recovers the non-relativistic Landau spectrum with cyclotron frequency ${\omega }_c=eB/m$. Higher-order corrections are $O(B^2/m^3)$, becoming significant only at extreme field strengths $B\gtrsim B_c=m^2/e\approx 4.4\times 10^9$ T for electrons (the Schwinger critical field).

Comparison with the Dirac case. A Dirac particle in a uniform magnetic field has spectrum $E_n^2=m^2+p_z^2+2eB(n+1/2\pm 1/2)$, where the $\pm 1/2$ is the spin Zeeman splitting $-g{\mu }_B\mathbf{S}\cdot \mathbf{B}$ with $g=2$. The result is

$$E_{n,s}^{\text{Dirac}}=\sqrt{m^2+p_z^2+2eB(n+1/2\pm 1/2)},s=\pm 1/2.$$

For $s=+1/2$ and $n\ge 0$, the energies are $\sqrt{m^2+p_z^2+2eB(n+1)}$; for $s=-1/2$ and $n\ge 1$, the same. The $n=0,s=-1/2$ "lowest Landau level" has $E_0=\sqrt{m^2+p_z^2}$ — the spin Zeeman energy exactly cancels the zero-point oscillator energy. This zero-mode is a structural feature of the Dirac magnetic spectrum absent from the Klein-Gordon case.

Worked example: spin-0 boson in $B=1$ T at $p_z=0$

Consider a spin-0 boson with mass equal to the pion mass $m=139.57$ MeV in a $B=1$ T magnetic field, restricted to longitudinal momentum $p_z=0$. Convert to natural units: $eB$ in MeV² is

$$eB=(1.602\times 10^{-19}\text{C})\times (1\text{T})\times (\hslash c{)}^2/[(\hslash c)\times (eV{)}^2\text{ conversion}].$$

More cleanly, $eB\hslash c^2=eB\cdot (\hslash c{)}^2/(\hslash c)=$ (in SI) $eB\cdot \hslash c$; numerically $eB\hslash c\approx 1.602\times 10^{-19}\times 1\times 1.973\times 10^{-7}\text{eV}\cdot \text{m}\times c$, giving $\sqrt{eB\hslash c^2}\approx 5.55\times 10^{-4}{\text{eV}}^{1/2}$, or $eB\hslash c^2\approx 3.08\times 10^{-7}{\text{eV}}^2$.

Compute the lowest two relativistic Landau levels:

$$E_0=\sqrt{m^2+eB}\approx m\sqrt{1+3.08\times 10^{-7}{\text{eV}}^2/(1.396\times 10^8\text{eV}{)}^2}\approx m\sqrt{1+1.58\times 10^{-23}}.$$

Taylor-expanding, $E_0-m\approx 0.79\times 10^{-23}m\approx 1.10\times 10^{-15}\text{eV}$. The non-relativistic Landau spacing is $eB/m\approx (3.08\times 10^{-7}{)}^{1/2}/(1.396\times 10^8)\cdot \text{(units adjusted)}$. In SI, $\hslash {\omega }_c/m=e\hslash B/(m^{2}c)\approx 1.602\times 10^{-19}\times 1.054\times 10^{-34}\times 1/[(2.49\times 10^{-28}{)}^2\times 3\times 10^8]\text{J}$, which gives $\hslash {\omega }_c\approx 1.10\times 10^{-15}$ eV for the lowest excitation. At $B=1$ T and $m=m_{\pi }$ the relativistic and non-relativistic predictions agree to 23 decimal places; the relative correction is $\sim (eB/m^2{)}^2\sim 10^{-46}$. The relativistic Landau formula only matters for elementary particles in extreme fields — neutron-star magnetospheres ($B\sim 10^{12}$ T) or laboratory pulsed-field setups ($B\sim 100$ T for light charged particles like electrons).

Exercises Intermediate+

Lean formalization Intermediate+

Mathlib's relevant infrastructure covers parts of the foundation but does not assemble the Klein-Gordon-external-field result. The components are:

• Mathlib.Analysis.SpecialFunctions.Gamma and a yet-unwritten Kummer/confluent-hypergeometric library provide the special-function basis needed for the radial Coulomb solution.
• Mathlib.LinearAlgebra.QuadraticForm and Mathlib.Analysis.NormedSpace.OperatorNorm provide the linear-algebraic and operator-theoretic infrastructure for the indefinite-norm Klein-Gordon Hilbert space, but the specific Krein-space formalism (Bognar 1974; Azizov-Iokhvidov 1989) needed for the indefinite-inner-product self-adjointness of the Klein-Gordon Hamiltonian has not been formalised.
• The gauge-fixed magnetic-field problem reduces to a quantum harmonic oscillator 12.04.02, where Mathlib has the algebraic ladder construction in Mathlib.Analysis.InnerProductSpace.Spectrum; the physics-layer step of identifying the magnetic-translation operators ${\pi }_{\pm }$ as the relevant ladder generators is missing.
• The Pauli-Weisskopf second-quantisation resolution of the negative-norm problem requires the charged-scalar Fock space, which builds on Mathlib.Analysis.InnerProductSpace.l2Space and bosonic creation-annihilation operators but is not assembled.

lean_status: none. The unit ships without a Lean module path; the Mathlib gap analysis frontmatter field records the specific gap. Aggregated with the other relativistic-QM none units, this feeds the upstream Mathlib physics-formalisation roadmap.

Advanced results Master

The Klein-Gordon Coulomb problem and its angular-momentum reorganisation

The spin-0 Sommerfeld formula derived in Theorem 1 exhibits an angular-momentum dependence that differs structurally from both the non-relativistic Schrödinger-Coulomb spectrum (which depends only on $n$) and the Dirac-Coulomb spectrum (which depends on $n$ and $j$). The Klein-Gordon dependence is on $n$ and $\ell$ separately, with the centrifugal term in the radial equation reorganised from $\ell (\ell +1)$ to $\ell (\ell +1)-(Z\alpha {)}^2$. This effective non-integer angular momentum is the structural fingerprint of the relativistic kinematic corrections in the absence of spin.

The Schrödinger-Coulomb $\mathrm{SO}(4)$ accidental degeneracy generated by the Runge-Lenz vector is broken by the relativistic corrections. In the Dirac case 12.11.03, a relativistic analogue (the Johnson-Lippmann operator) survives and produces the residual $(n,j)$-degeneracy. The Klein-Gordon case has no analogous symmetry: the spectrum depends fully on both $n$ and $\ell$, and no two-state degeneracy persists. The structural reason is that the Johnson-Lippmann operator involves the Pauli matrices and the Dirac $\kappa$-operator, both of which require the spinor structure that the Klein-Gordon equation lacks.

The radial Klein-Gordon-Coulomb wavefunctions can be written explicitly in terms of confluent hypergeometric functions:

$$R_{n,\ell }(r)=\mathcal{N}(2\lambda r{)}^{\gamma -1/2}{e}^{-\lambda r}{}_1{F}_1(-n_r,2\gamma +1;2\lambda r),$$

with $\gamma =\sqrt{(\ell +1/2{)}^2-(Z\alpha {)}^2}$, $\lambda =\sqrt{m^2-E_{n,\ell }^2}$, $n_r=n-\ell -1$, and a normalisation $\mathcal{N}$ involving Gamma functions of $\gamma$ and $n_r$. The asymptotic behaviour near the origin $R\sim r^{\gamma -1/2}$ shows the relativistic deviation from the Schrödinger $R\sim r^{\ell }$: the wavefunction has a mild integrable singularity at $r=0$ for $\ell =0$, and a slightly modified power-law dependence for $\ell \ge 1$. The integral ${\int }_0^{\infty }{R}^2{r}^{2}dr<\infty$ provided $\gamma >-1/2$, i.e., $(\ell +1/2{)}^2>(Z\alpha {)}^2$, which is the condition for the bound state to exist.

The spin-0 catastrophe and self-adjoint extensions

The Klein-Gordon radial Coulomb operator $-d^2/d{r}^2-(2/r)d/dr+[\ell (\ell +1)-(Z\alpha {)}^2]/r^2-2EZ\alpha /r$ on $C_0^{\infty }(0,\infty )$ has, for $(\ell +1/2{)}^2<(Z\alpha {)}^2$ — i.e., the supercritical regime — multiple inequivalent self-adjoint extensions. The deficiency-index analysis (Weidmann 1980; Reed-Simon Vol. II §X.1) gives deficiency indices $(1,1)$ in the supercritical regime, parameterised by a $U(1)$ phase that specifies the boundary condition at $r=0$. Each choice of self-adjoint extension corresponds to a different physical regularisation of the point-charge limit, and the choice is fixed by the finite-nucleus form factor.

For $\ell =0$, the critical $Z\alpha$ is $1/2$: at $Z\alpha =0.5$ the regular solution at $r=0$ becomes $R\sim r^{-1/2}$, oscillating logarithmically at the origin and failing the normalisability criterion. The Pomeranchuk-Smorodinsky catastrophe in its spin-0 form sets in at half the Dirac threshold, making the supercritical KG-Coulomb problem accessible at lower $Z$ than the Dirac case. Experimentally, pionic atoms with $Z\gtrsim 70$ probe this regime, with measurable shifts of the inner-shell levels from the point-nucleus prediction.

The mathematical question — which self-adjoint extension is physical — is resolved by matching to the finite-nucleus solution. As the nuclear radius $R_N\to 0$, the finite-nucleus eigenvalue converges to a specific point-nucleus eigenvalue, and the corresponding extension is the distinguished extension. The construction parallels Greiner-Reinhardt 1985 for the Dirac case.

Relativistic Landau levels and the indefinite-metric Hilbert space

The relativistic Landau spectrum $E_n^2=m^2+p_z^2+2eB(n+1/2)$ derived in Theorem 2 admits two physically distinct branches: positive-energy $E_n=+\sqrt{\cdots }$ and negative-energy $E_n=-\sqrt{\cdots }$. In the single-particle Klein-Gordon theory, only the positive branch is interpreted as a physical particle state; the negative branch corresponds to an antiparticle in the Pauli-Weisskopf second-quantised theory. The conserved current $j^0$ is positive on the positive-energy branch and negative on the negative-energy branch, formalising the particle / antiparticle assignment.

The infinite degeneracy of each Landau level — $\mathcal{N}/A=eB/(2\pi )$ states per unit transverse area — is the same as the non-relativistic count, because it derives from the gauge structure (the magnetic-translation algebra $[T_x,T_y]=ieB$ generates a Heisenberg algebra whose representations have intrinsic dimension equal to the symplectic volume / Planck area) and not from any relativistic kinematics. The Landau-level degeneracy is the prototypical example of a magnetic flux quantum: each Landau orbit encloses a flux quantum ${\Phi }_0=2\pi /e$ (in natural units).

The lowest Landau level $n=0$ has the special structure of being annihilated by the $a$ operator: $a|0{⟩}_{LLL}=0$, with wavefunctions ${\psi }_0(\bar{z})\propto {\bar{z}}^k{e}^{-|z{|}^2/4{\ell }_B^2}$ for $k=0,1,2,\dots$ labelling the intra-level degeneracy. In the relativistic case, $E_0=\sqrt{m^2+p_z^2+eB}$ — the zero-point energy $eB$ does not vanish, unlike the Dirac case where the spin Zeeman term cancels it for $s=-1/2$. This relativistic-Landau zero-point energy is the seed for vacuum-polarisation effects in strong magnetic fields and contributes to the Heisenberg-Euler effective Lagrangian.

Comparison with the Dirac case: where spin matters

The Dirac equation in a uniform magnetic field [12.11.03 (Dirac in Coulomb is the published analogue; Dirac in magnetic field is the natural successor unit)] gives the spectrum $E_{n,s}^2=m^2+p_z^2+2eB(n+1/2+s)$ for $s=\pm 1/2$ and $n\ge 0$ (with the constraint $n+1/2+s\ge 0$). The $s$-dependence is the Zeeman splitting due to the electron's intrinsic magnetic moment $g{\mu }_B=2\cdot e/(2m)$. The Klein-Gordon equation, lacking spin, has no analogous $s$-quantum number and no Zeeman splitting: it gives only the single tower $E_n^2=m^2+p_z^2+2eB(n+1/2)$.

The same structural pattern repeats in the Coulomb case: KG fine structure depends on $\ell$, Dirac fine structure on $j$. The replacement $\ell \to j$ between the two formulas is the rule: every spin-0 result becomes the spin-1/2 result by promoting orbital angular momentum to total angular momentum. The Dirac case adds a residual $(n,j)$ degeneracy via the Johnson-Lippmann symmetry; the KG case has no such residual symmetry.

Pionic atoms: experimental realisation

Pionic atoms — bound states of a negatively charged pion ${\pi }^{-}$ and an atomic nucleus — are the principal experimental laboratory for spin-0 Sommerfeld physics. The pion's mass $m_{\pi }{c}^2=139.57$ MeV produces Bohr radii $a_{\pi }=a_0(m_e/m_{\pi })\approx 222$ fm for pionic hydrogen, comparable to nuclear radii for medium-heavy nuclei. As a consequence, pionic-atom spectra exhibit (a) the spin-0 Sommerfeld fine structure derived in Theorem 1, (b) substantial finite-nucleus corrections from the breakdown of the point-charge approximation, and (c) measurable strong-interaction effects from the pion-nucleon coupling, especially in the inner-shell levels.

The pionic-hydrogen $2p\to 1s$ transition has been measured to better than 1 eV precision (Backenstoss 1970; Hennebach et al. 2014 Eur. Phys. J. A 50 — the PSI pionic-hydrogen experiment), providing the most precise determination of the $\pi N$ scattering length. The spin-0 Sommerfeld formula provides the QED baseline against which the strong-interaction shifts are extracted; agreement with the formula at the calculated fine-structure precision validates the relativistic-QED treatment of the bound-state problem.

For heavier pionic atoms ($Z=50-80$), the strong-interaction shifts of the inner-shell levels are of order MeV — much larger than the QED fine-structure corrections — and provide a direct probe of meson-nucleus dynamics. The pionic-atom programme at meson factories (PSI, J-PARC) has been the source of pion-nucleus optical-potential parametrisations for forty years.

Klein paradox and pair-creation pointer

Strong external fields can drive the Klein-Gordon equation out of the single-particle regime. The Klein paradox — a charged scalar plane wave incident on a step potential $V_0\Theta (x)$ shows reflection coefficient $R>1$ when $V_0>2m$ — is the symptom: the "transmitted" wave in the strong-potential region is a negative-energy KG solution carrying opposite current sign, and the apparent excess reflection corresponds to spontaneous pair creation at the boundary. In second-quantised QFT (Pauli-Weisskopf 1934), the resolution is that the strong-potential boundary acts as a source for ${\pi }^{-}{\pi }^{+}$ pair production, and the missing flux is accounted for by the outgoing antiparticle current.

This is the same physics that drives the supercritical-field decay of the QED vacuum at $Z\alpha >|\kappa |$ in the Dirac case, and at $Z\alpha >\ell +1/2$ in the KG case. The bound-state catastrophe and the Klein paradox are two faces of the same underlying instability: at field strengths comparable to $m^2/e$, the vacuum of a charged scalar field becomes unstable, and the single-particle KG description must be replaced by the second-quantised QFT. The Schwinger effect — pair production from a uniform electric field at $E\gtrsim m^2/e$ — is the canonical realisation of this physics and the natural successor topic in the QED chapter.

Full proof set Master

The derivations of Theorems 1 and 2 in the Key derivation section are complete in outline. The technical step that warrants further justification is the indicial calculation at the origin for the Klein-Gordon-Coulomb radial equation, and the construction of the magnetic ladder operators.

Indicial analysis at $r=0$ (Theorem 1). Near $r=0$ the radial equation has leading-order form

$$R^{\prime \prime }+\frac{2}{r}{R}^{\prime }+\left[\frac{(Z\alpha {)}^2-\ell (\ell +1)}{r^2}+\frac{2EZ\alpha }{r}+(E^2-m^2)\right]R\approx 0.$$

The $1/r^2$ term dominates as $r\to 0$. Try $R\sim r^s$:

$$s(s-1)r^{s-2}+2s{r}^{s-2}+[(Z\alpha {)}^2-\ell (\ell +1)]r^{s-2}=0,$$

i.e., $s(s+1)+(Z\alpha {)}^2-\ell (\ell +1)=0$, with roots

$$s_{\pm }=-\frac{1}{2}\pm \sqrt{\frac{1}{4}+\ell (\ell +1)-(Z\alpha {)}^2}=-\frac{1}{2}\pm \sqrt{(\ell +1/2{)}^2-(Z\alpha {)}^2}.$$

For normalisability ${\int }_0^{ϵ}{R}^2{r}^{2}dr<\infty$ — i.e., $R\cdot r$ in $L^2$ at the origin — we need $s>-3/2$. The regular root $s_{+}=-1/2+\gamma$ with $\gamma =\sqrt{(\ell +1/2{)}^2-(Z\alpha {)}^2}$ satisfies this for any $\gamma >-1$, i.e., for any $(Z\alpha {)}^2<(\ell +1/2{)}^2+1$, well within the bound-state regime $0<Z\alpha <\ell +1/2$. The singular root $s_{-}$ gives $R\sim r^{-1/2-\gamma }$, which fails normalisability for $\gamma >1/2$, i.e., for $(\ell +1/2{)}^2-(Z\alpha {)}^2>1/4$; the singular root is discarded.

Asymptotic analysis at $r\to \infty$. The leading behaviour at large $r$ is governed by $R^{\prime \prime }+(E^2-m^2)R\approx 0$. For bound states $E^2<m^2$, set ${\lambda }^2:=m^2-E^2>0$; the bounded solution is $R\sim e^{-\lambda r}$.

Kummer reduction. Substituting $R(r)=r^{s_{+}}{e}^{-\lambda r}F(r)$ into the full radial equation and changing variable to $\rho =2\lambda r$, the function $F(\rho )$ satisfies Kummer's confluent hypergeometric equation

$$\rho F^{\prime \prime }(\rho )+(b-\rho )F^{\prime }(\rho )-aF(\rho )=0,$$

with $b=2\gamma +1$ and $a=\gamma +1/2-\nu$ where $\nu =EZ\alpha /\lambda$. The regular-at-origin solution is $F={}_1{F}_1(a,b;\rho )$. For $F$ to be a polynomial — necessary for $R$ to be bounded at infinity — we need $a=-n_r$ for non-negative integer $n_r$, giving the quantisation $\nu =n_r+\gamma +1/2$ and the Sommerfeld formula as derived in the main text.

Magnetic ladder operators (Theorem 2). In the symmetric gauge $\mathbf{A}=(B/2)(-y,x,0)$, define the kinetic momentum operators ${\pi }_x:=p_x-e{A}_x=p_x+eBy/2$ and ${\pi }_y:=p_y-e{A}_y=p_y-eBx/2$. Direct computation:

$$[{\pi }_x,{\pi }_y]=[p_x+eBy/2,p_y-eBx/2]=-\frac{eB}{2}[p_x,x]+\frac{eB}{2}[y,p_y]=-\frac{eB}{2}(-i)+\frac{eB}{2}(i)=ieB.$$

(In conventions with $e>0$ for the boson charge.) Define ${\pi }_{\pm }:={\pi }_x\pm i{\pi }_y$. Then $[{\pi }_{-},{\pi }_{+}]=-2i[{\pi }_x,{\pi }_y]=2eB$. Define $a:={\pi }_{-}/\sqrt{2eB}$ and $a^{†}:={\pi }_{+}/\sqrt{2eB}$; then $[a,a^{†}]=1$. The transverse kinetic energy is

$$H_{\perp }={\pi }_x^2+{\pi }_y^2={\pi }_{+}{\pi }_{-}+i[{\pi }_x,{\pi }_y]=2eB\cdot a^{†}a+eB=2eB(a^{†}a+1/2),$$

with spectrum $H_{\perp }|n⟩=2eB(n+1/2)|n⟩$ for $n=0,1,2,\dots$. Substituting into $E^2-p_z^2-m^2=H_{\perp }$ gives the relativistic Landau spectrum.

Intra-level degeneracy. A second pair of operators commuting with $H_{\perp }$ generates the infinite intra-level degeneracy. Define the canonical position operators $X:=x+{\pi }_y/(eB)$ and $Y:=y-{\pi }_x/(eB)$. These satisfy $[X,Y]=i/(eB)\ne 0$ but commute with ${\pi }_x$ and ${\pi }_y$ (the kinetic momenta). Define $b:=\sqrt{eB/2}(X-iY)$ and $b^{†}:=\sqrt{eB/2}(X+iY)$; then $[b,b^{†}]=1$ and $[b,a]=[b,a^{†}]=0$. States $|n,m⟩:=(b^{†}{)}^m(a^{†}{)}^n|0,0⟩/\sqrt{n!m!}$ are simultaneous eigenstates of $H_{\perp }$ with eigenvalue $2eB(n+1/2)$ and of the centre-of-orbit displacement $b^{†}b$ with eigenvalue $m$. The integer $m=0,1,2,\dots$ labels the infinite degeneracy at each Landau level $n$.

In a finite-area sample of area $A$, the requirement that the orbit centre lie within the sample bounds the degeneracy: $m\le A\cdot (eB)/(2\pi )$, recovering the standard flux-quantum-per-state count. Each Landau level holds $A\cdot eB/(2\pi )$ states, equal to the number of flux quanta ${\Phi }_0=2\pi /e$ piercing the sample.

Connections Master

• Dirac equation and relativistic spin 12.11.01 is the prerequisite framework: minimal coupling, gamma matrices, and the relativistic kinematics that the Klein-Gordon equation inherits without the spinor structure. The present unit specialises the abstract minimal-coupling rule to two integrable external-field cases for the spin-0 partner of the Dirac equation.

• Dirac equation in a Coulomb field 12.11.03 is the spin-1/2 analogue. The Dirac Sommerfeld formula is obtained from the KG formula derived here by the substitution $\ell +1/2\to j+1/2$; comparing the two isolates the contribution of spin to relativistic fine structure. Schrödinger's 1926 attempt to apply the relativistic KG equation to hydrogen gave incorrect fine-structure splittings precisely because the electron has spin 1/2, not 0.

• Hydrogen atom bound states 12.06.01 is the non-relativistic limit. Both the KG and the Dirac Sommerfeld formulas reduce to the Bohr spectrum $E_n=-m(Z\alpha {)}^2/(2n^2)$ at leading order in $Z\alpha$; the relativistic corrections then split the levels by $\ell$ (KG) or $j$ (Dirac). The Schrödinger-Coulomb $\mathrm{SO}(4)$ accidental degeneracy is broken differently in the two relativistic theories.

• Quantum harmonic oscillator 12.04.02 supplies the algebraic ladder construction used in the magnetic-field problem. The relativistic Landau spectrum is the spectrum of a quantum harmonic oscillator with frequency $2eB$, sat inside a square root with the rest mass and longitudinal momentum.

• Klein paradox (forthcoming 12.11.04) examines the strong-field transmission anomalies of the KG and Dirac equations. The KG bound-state catastrophe at $Z\alpha =\ell +1/2$ and the Klein paradox at $V_0>2m$ are two faces of the same supercritical-field instability.

• Bosonic Fock space and second quantisation 12.13.01 resolves the indefinite-norm problem of the single-particle KG theory. The charged scalar quantum field (Pauli-Weisskopf 1934) provides the operator-level interpretation: positive- and negative-energy KG modes are creation operators for particles and antiparticles respectively. The bound-state spectra derived here are eigenvalues of the relevant single-particle operators, and they remain physically correct as the spectrum of single-quantum excitations of the second-quantised theory.

• Angular momentum and $\mathrm{SU}(2)$ 12.05.01 supplies orbital angular momentum $\mathbf{L}$, spherical harmonics $Y_{\ell ,m}$, and the separation-of-variables technique used in the Coulomb case. The KG case uses only orbital angular momentum (no spin), simplifying the angular-momentum analysis compared to the Dirac case.

• Pionic-atom spectroscopy and meson-nucleus physics is the principal experimental application. Pionic hydrogen and pionic deuterium provide the most precise determinations of the $\pi N$ scattering length; heavier pionic atoms probe the pion-nucleus optical potential. The QED baseline is the spin-0 Sommerfeld formula derived in this unit; strong-interaction shifts are measured against this baseline.

Historical & philosophical context Master

The Klein-Gordon equation was the first written-down relativistic quantum wave equation. Schrödinger derived it in 1926 ^{[Klein 1926] [Gordon 1926]} before he settled on the non-relativistic equation that bears his name, and applied it to hydrogen — getting a fine-structure formula that disagreed with Sommerfeld's 1916 empirical result. Schrödinger published the non-relativistic equation instead, and the relativistic equation was independently rediscovered by Klein, Gordon, and Fock within months, with all three names now attached to it (the "Klein-Gordon-Fock equation" in older Russian literature).

The disagreement with Sommerfeld's 1916 formula ^{[Sommerfeld 1916]} was a puzzle that Schrödinger could not resolve. Sommerfeld's formula matched experiment for hydrogen, and the relativistic KG equation gave a different formula. The resolution came two years later with Dirac's 1928 equation, which has spin built in and reproduces the Sommerfeld formula exactly (with the substitution $k\to j+1/2$). With hindsight, the KG-Dirac comparison shows that the electron's spin contributes a measurable fine-structure correction; the spin-0 KG result was simply wrong for hydrogen, but correct for spin-0 particles. This was vindicated experimentally by pionic atoms: when the pion was discovered in 1947 (Lattes et al.) and pionic atoms were first studied (Camac et al. 1952), the spectra agreed with the KG Sommerfeld formula derived here.

The second canonical KG-external-field problem — uniform magnetic field — has a slightly different history. The non-relativistic Landau quantisation was derived by Landau in 1930 ^{[Landau Lifshitz 1977 §35]} in the course of explaining the diamagnetism of conduction electrons; the relativistic generalisation was implicit in the Klein-Gordon and Dirac formalisms of the late 1920s but received less attention until the development of QED in the late 1940s. The relativistic Landau spectrum is now standard background for any treatment of charged-particle motion in strong magnetic fields, including pulsar magnetospheres (where surface fields reach $10^{12}$ T) and quark-gluon plasmas in heavy-ion collisions (transient fields up to $10^{15}$ T at LHC).

Klein's 1929 paradox ^{[Klein 1929]} — the apparent excess transmission of a charged wave through a strong barrier — sparked a decade of debate about the consistency of single-particle relativistic quantum mechanics. The eventual resolution by Pauli and Weisskopf in 1934 ^{[Pauli-Weisskopf 1934]} — that the KG equation must be quantised as a field operator, not interpreted as a single-particle wave function — was a milestone in the development of QFT. The charged scalar field they constructed was the prototype for every subsequent QFT, and the negative-norm states of the single-particle theory were reinterpreted as antiparticles in the second-quantised theory. The Pauli-Weisskopf construction made the KG equation respectable a decade before the Dirac equation was finally given an analogous second-quantised treatment.

The mathematical analysis of the KG-Coulomb bound-state problem, including the rigorous justification of the Sommerfeld formula and the analysis of self-adjoint extensions in the supercritical regime, was carried out by Bargmann in 1932 ^{[Bargmann 1932]} in a work that anticipated by decades the modern operator-theoretic treatment of singular Schrödinger operators. The same machinery later proved essential for the Dirac-Coulomb analysis at $Z\alpha >1$ (the Pomeranchuk-Smorodinsky regime). The Klein-Gordon case, with its lower critical $Z\alpha =1/2$ (for $\ell =0$), is experimentally accessible at smaller atomic numbers than the Dirac case and provides a controlled laboratory for the operator-theoretic phenomena that the Dirac case exhibits only at extreme $Z$.

Bibliography Master

Primary literature:

• Klein, O., "Quantentheorie und fünfdimensionale Relativitätstheorie", Z. Phys. 37 (1926), 895–906. [Original Klein-Gordon equation in the context of 5D Kaluza-Klein theory.]
• Gordon, W., "Der Comptoneffekt nach der Schrödingerschen Theorie", Z. Phys. 40 (1926), 117–133. [Independent derivation; first application to Compton scattering.]
• Fock, V., "Über die invariante Form der Wellen- und der Bewegungsgleichungen für einen geladenen Massenpunkt", Z. Phys. 39 (1926), 226–232. [Independent derivation with minimal coupling to external EM field.]
• Sommerfeld, A., "Zur Quantentheorie der Spektrallinien", Ann. Phys. (Leipzig) 51 (1916), 1–94. [Pre-quantum fine-structure formula; the spin-0 KG equation reproduces this for $\ell$ replacing the old "azimuthal quantum number" $k$.]
• Klein, O., "Die Reflexion von Elektronen an einem Potentialsprung nach der relativistischen Dynamik von Dirac", Z. Phys. 53 (1929), 157–165. [The Klein paradox; applies to both KG and Dirac.]
• Bargmann, V., "Bemerkungen zur allgemein-relativistischen Fassung der Quantentheorie", Z. Phys. 75 (1932), 539–576. [Rigorous bound-state analysis of the relativistic Coulomb problem.]
• Pauli, W. & Weisskopf, V., "Über die Quantisierung der skalaren relativistischen Wellengleichung", Helv. Phys. Acta 7 (1934), 709–731. [Second quantisation of the charged scalar field; resolution of negative-norm and Klein-paradox issues.]
• Landau, L. D., "Diamagnetismus der Metalle", Z. Phys. 64 (1930), 629–637. [Original Landau-level quantisation in the non-relativistic case.]
• Hennebach, M. et al. (PSI Pionic Hydrogen Collaboration), "Hadronic shift in pionic hydrogen", Eur. Phys. J. A 50 (2014), 190. [Modern precision measurement of pionic-hydrogen $1s$ level for $\pi N$ scattering-length extraction.]

Textbooks and monographs:

• Greiner, W., Relativistic Quantum Mechanics: Wave Equations, 3rd ed. (Springer, 2000), Ch. 9. [Canonical detailed treatment of KG in external fields; pionic atoms; magnetic-field solutions.]
• Bjorken, J. D. & Drell, S. D., Relativistic Quantum Mechanics (McGraw-Hill, 1964), Ch. 9. [Standard reference for the single-particle KG theory; Klein paradox treatment.]
• Landau, L. D. & Lifshitz, E. M., Quantum Mechanics: Non-Relativistic Theory, 3rd ed. (Pergamon, 1977), §§35-36. [Landau-level analysis in the non-relativistic case; relativistic generalisation in Vol. 4.]
• Berestetskii, V. B., Lifshitz, E. M., Pitaevskii, L. P., Quantum Electrodynamics, 2nd ed. (Pergamon, 1982), §§32-33. [Landau-Lifshitz Vol. 4 treatment of KG in external fields.]
• Itzykson, C. & Zuber, J.-B., Quantum Field Theory (McGraw-Hill, 1980), §2-1. [KG equation, conserved current, Klein paradox, and pointer to second quantisation.]
• Weinberg, S., The Quantum Theory of Fields, Vol. I (Cambridge UP, 1995), Ch. 5. [Modern Wigner-classification derivation of the KG equation as the spin-0 case of a general field-construction theorem.]
• Backenstoss, G., "Pionic atoms", Annu. Rev. Nucl. Sci. 20 (1970), 467–508. [Experimental review of pionic-atom spectroscopy; the principal physical realisation of spin-0 Sommerfeld physics.]
• Greiner, W. & Reinhardt, J., Quantum Electrodynamics, 4th ed. (Springer, 2009), Ch. 9. [Supercritical-field analysis; finite-nucleus regularisation; relevant to both KG and Dirac catastrophe regimes.]
• Reed, M. & Simon, B., Methods of Modern Mathematical Physics, Vol. II (Academic Press, 1975), §X.1. [Self-adjoint extension theory for singular Schrödinger operators; foundation for the KG-Coulomb supercritical analysis.]