Bethe-Heitler bremsstrahlung and pair production

Intuition Beginner

A high-energy electron flies past a heavy atomic nucleus. The nucleus carries charge $Ze$ and produces a static Coulomb potential $V(r)=Ze/(4\pi r)$ that bends the electron's trajectory — the classical Rutherford scattering picture. Once in a while the electron radiates a real photon during the deflection, the way a classical charge accelerating in an external field emits electromagnetic radiation (the Larmor formula in disguise). The radiated photon carries off a fraction of the incident electron's energy. The process is called bremsstrahlung — German for "braking radiation" — and it dominates the energy loss of high-energy electrons traversing matter above about 10 MeV.

The quantum-electrodynamics calculation was completed by Hans Bethe and Walter Heitler in 1934 in their joint paper On the Stopping of Fast Particles and on the Creation of Positive Electrons in the Proceedings of the Royal Society A. They used Dirac's two-year-old positron theory (which had been experimentally vindicated by Anderson's 1932 discovery of the positron only fifteen months earlier) to compute two tree-level Feynman diagrams: the photon is emitted from the electron leg before it scatters off the nucleus's Coulomb field, or after. The two amplitudes interfere coherently. The squared sum, integrated over the recoil kinematics of the nucleus, gives the Bethe-Heitler differential cross section $d\sigma /d\omega$ for the radiated photon to carry energy $\omega$.

The qualitative features are simple. The cross section diverges as $1/\omega$ at low photon energy — the infrared divergence of quantum-electrodynamics, the soft-photon pile-up that is finite only when you also include the unobserved soft-photon emission from any virtual one-loop correction (the Bloch-Nordsieck mechanism of 1937). At high photon energy the cross section falls off; the highest-energy photon is bounded by the incident electron energy by energy conservation. Integrating the spectrum gives the radiation length $X_0$, the characteristic distance over which an ultra-relativistic electron loses $1-1/e\approx 63\%$ of its energy to bremsstrahlung — 36 g/cm² in air, about 1 cm in lead, the canonical material property that organises electromagnetic-shower physics.

The same paper derived the crossing-symmetric partner process, pair production: a high-energy photon traversing matter can create an electron-positron pair in the Coulomb field of a nucleus, $\gamma +N\to e^{+}+e^{-}+N$. The two diagrams are Bethe-Heitler bremsstrahlung with the outgoing electron and outgoing photon reinterpreted via Feynman's antiparticle-as-backward-going-electron prescription; the cross section follows by analytic continuation.

The threshold is $E_{\gamma }>2m_e{c}^2\approx 1.022$ MeV (the rest-mass cost of producing the pair, with the nucleus absorbing the kinematic recoil). Near threshold the cross section rises as ${\beta }^3$ in the velocity of the outgoing pair; asymptotically at $E_{\gamma }\gg m_e{c}^2$ the pair-production cross section approaches $7/9$ of the asymptotic bremsstrahlung total — a universality theorem of Rossi (1952) that organises electromagnetic-shower physics.

Why this calculation matters. Bethe-Heitler bremsstrahlung and pair production are the engine of every electromagnetic shower. A high-energy electron entering a calorimeter emits a hard photon (bremsstrahlung); the photon converts to a pair (pair production); each pair member emits more hard photons; and the cascade continues until the individual energies drop below the critical energy and ionisation takes over.

This shower physics is the basis of every electromagnetic calorimeter at every particle-physics collider, every cosmic-ray air-shower detector, every medical linear accelerator that produces therapeutic X-rays at 4-25 MeV, every positron source for ILC/FCC-ee designs, and every radiation-damage estimate for accelerator and reactor components. The Bethe-Heitler cross section is one of the most-evaluated formulas in physics — every Monte Carlo simulation of charged-particle and photon transport (EGS, GEANT4, FLUKA, PENELOPE) sits on top of it.

Visual Beginner

The two bremsstrahlung diagrams share the same external legs (an incoming electron, an outgoing electron, an outgoing real photon, and the heavy nucleus treated as a static external source) but route the photon emission differently. In the pre-emission diagram the electron emits the photon first and then undergoes Coulomb scattering at the nucleus. In the post-emission diagram the order is reversed. The two amplitudes interfere; their sum is the bremsstrahlung amplitude.

The crossing-related pair-production diagram has the same external legs reinterpreted: the incoming electron becomes an outgoing positron (with reversed four-momentum), the outgoing photon becomes an incoming photon, and the rest of the diagram unchanged. The kinematic substitution $p\to -p_{+},k\to -k$ (with momentum-conservation tracking) carries the bremsstrahlung amplitude into the pair-production amplitude. This is the cleanest external-field example of crossing symmetry in particle physics.

The cross section has a $1/\omega$ infrared divergence at low photon energy — physically a soft-photon pile-up that is regulated only when combined with virtual one-loop quantum-electrodynamics corrections at the level of inclusive cross sections (Bloch-Nordsieck cancellation). The integrated cross section weighted by photon energy gives the radiation length $X_0$, the engine of electromagnetic-shower physics in all of high-energy physics and medical-linac design.

Worked example Beginner

A 1 GeV electron enters a 1 cm-thick slab of lead ($Z=82$, atomic mass $A=207$, density $\rho =11.35$ g/cm³). Estimate the number of bremsstrahlung photons it emits and the mean radiated energy.

The radiation length of lead from the Bethe-Heitler formula (Tsai 1974, with full-screening Thomas-Fermi atomic form factor) is $X_0(\text{Pb})=6.37$ g/cm², or $X_0(\text{Pb})/\rho =0.561$ cm. The slab is $(1\text{ cm})/(0.561\text{ cm})=1.78$ radiation lengths thick. After traversing one radiation length, an ultra-relativistic electron has lost on average $1-1/e\approx 63\%$ of its energy; after 1.78 radiation lengths, the mean energy loss is $1-e^{-1.78}\approx 83\%$. The mean radiated energy is about $0.83\times 1$ GeV $\approx 830$ MeV.

The mean number of photons emitted depends on the photon-energy threshold. The Bethe-Heitler spectrum $dN/d\omega =(1/\omega )\cdot \text{(slowly-varying factor)}$ has a logarithmic dependence: integration of $d\omega /\omega$ from ${\omega }_{\text{min}}$ to $E$ gives a factor $\log (E/{\omega }_{\text{min}})$.

With $E=1$ GeV and a photon-energy threshold of ${\omega }_{\text{min}}=1$ MeV (the typical lower edge for detection in a sampling calorimeter), the logarithm is $\log (10^3)\approx 6.9$. Multiplying by the slowly-varying spectral prefactor gives roughly $7$ bremsstrahlung photons above 1 MeV per electron traversing 1 cm of lead — order-of-magnitude consistent with Monte Carlo simulations (GEANT4 with the EGS5 cross sections gives $5$ to $10$ depending on geometry and acceptance).

The mean photon energy is the radiated energy divided by the photon count: $830$ MeV $/7\approx 120$ MeV. Most of the radiated energy goes into a few hard photons; the soft-photon spectrum is dense in number but soft in energy. This is the characteristic skewed energy distribution of the Bethe-Heitler spectrum.

What this tells us: even at 1 GeV electron energy, a 1 cm slab of lead is opaque to high-energy electrons in the bremsstrahlung sense — the electron loses most of its energy to a handful of hard photons that then convert to pairs and feed the electromagnetic cascade. Each radiated photon above $2m_e{c}^2\approx 1.022$ MeV has a high probability of converting to a pair in another $9/7$ radiation lengths (universality), and the cascade develops. The full longitudinal shower depth at 1 GeV in lead is about 10-15 radiation lengths, or 6-8 cm — the design depth of typical electromagnetic calorimeters at LHC experiments. Bethe-Heitler is the cornerstone calculation; the cascade is its repeated application.

Check your understanding Beginner

Formal definition Intermediate+

Work in natural units $\hslash =c=1$ with the mostly-minus metric ${\eta }^{\mu \nu }=\mathrm{diag}(+1,-1,-1,-1)$. Bethe-Heitler bremsstrahlung is the tree-level QED process

$$e^{-}(p,s)+N(q_i;Ze)\to e^{-}(p^{\prime },s^{\prime })+\gamma (k,\lambda )+N(q_f;Ze),$$

where the nucleus of charge $Ze$ and mass $M_N$ is treated as a static external source (the heavy-nucleus limit $M_N\to \infty$ decouples nuclear recoil; momentum transfer to the nucleus is absorbed kinematically but no nuclear kinetic energy is exchanged). The photon four-momentum is $k^{\mu }=(\omega ,\mathbf{k})$ with $\omega =|\mathbf{k}|$, polarisation ${ϵ}^{\ast \mu }(k,\lambda )$ with $\lambda \in \{1,2\}$; the electron spinors are $u^s(p),{\bar{u}}^{s^{\prime }}(p^{\prime })$ on the mass shell $p^2=p^{\prime 2}=m^2$.

External-field Furry-picture amplitude

The static-nucleus Coulomb field $A_{\mu }^{\text{ext}}(x)=(Ze/(4\pi r),0)$ has Fourier transform

$${\tilde{A}}_{\mu }^{\text{ext}}(q)=(2\pi )\delta (q_0)\cdot \left(\frac{Ze}{|\mathbf{q}{|}^2},0\right),$$

a $\delta$-function in energy (the source is static) and a $1/|\mathbf{q}{|}^2$ Coulomb factor in three-momentum. The Furry-picture QED interaction Hamiltonian ${\mathcal{H}}_I=e\bar{\psi }(\text{slashed}A+\text{slashed}{A}^{\text{ext}})\psi$ has two parts: the quantised-photon vertex $-ie{\gamma }^{\mu }$ coupling to the real radiated photon, and the external-field vertex $-ie{\gamma }^{\mu }{\tilde{A}}_{\mu }^{\text{ext}}(q)$ coupling to the nuclear Coulomb source. At order $Ze\cdot e^2=Z{\alpha }^{3/2}$ in the amplitude (lowest non-vanishing), two diagrams contribute.

Two tree-level diagrams

The pre-emission diagram (${\mathcal{M}}_1$) emits the real photon from the incoming electron leg before the Coulomb scattering at the nucleus. The intermediate virtual electron carries four-momentum $p-k$. The post-emission diagram (${\mathcal{M}}_2$) Coulomb-scatters the electron first and emits the photon from the outgoing leg; the intermediate virtual electron carries four-momentum $p^{\prime }+k$. Applying the Furry-picture Feynman rules,

$$i{\mathcal{M}}_1=(-ie{)}^2{\bar{u}}^{s^{\prime }}(p^{\prime })\text{slashed}{\tilde{A}}^{\text{ext}}(q)\frac{i(\text{slashed}p-\text{slashed}k+m)}{(p-k{)}^2-m^2}\text{slashed}{ϵ}^{\ast }(k)u^s(p),$$ $$i{\mathcal{M}}_2=(-ie{)}^2{\bar{u}}^{s^{\prime }}(p^{\prime })\text{slashed}{ϵ}^{\ast }(k)\frac{i(\text{slashed}{p}^{\prime }+\text{slashed}k+m)}{(p^{\prime }+k{)}^2-m^2}\text{slashed}{\tilde{A}}^{\text{ext}}(q)u^s(p),$$

with three-momentum transfer to the nucleus $\mathbf{q}={\mathbf{p}}^{\prime }+\mathbf{k}-\mathbf{p}$ (energy is not conserved in the external-field interaction — the static source supplies the kinematic mismatch). The total amplitude is

$$\boxed{\mathcal{M}={\mathcal{M}}_1+{\mathcal{M}}_2.}$$

There is no relative minus sign between the two diagrams: they are related by the routing of the photon emission, not by an exchange of identical fermions. The squared amplitude after spin-averaging the electron polarisations and summing over photon polarisations contains the interference term $2\mathrm{Re}({\mathcal{M}}_1{\mathcal{M}}_2^{\ast })$, which is sign-dependent on the kinematics.

Bethe-Heitler differential cross section

The spin-averaged squared amplitude, after a long but systematic gamma-matrix trace computation (BLP QED §92.3 explicitly; Tsai 1974 with corrections to the original Bethe-Heitler 1934 result), integrated over the nuclear recoil and reduced to a function of the radiated photon energy $\omega$, the incident electron energy $E$, and the outgoing electron energy $E^{\prime }=E-\omega$, gives the Bethe-Heitler cross section in the ultra-relativistic limit $E,E^{\prime }\gg m_e$:

$$\boxed{\frac{d\sigma }{d\omega }=\frac{4Z^2{\alpha }^3}{\omega \cdot m_e^2{c}^4}\left[\frac{4}{3}-\frac{4\omega }{3E}+\frac{{\omega }^2}{E^2}\right]\log \left[\frac{183Z^{-1/3}}{1+\omega /E}\right]+\mathcal{O}({\alpha }^4),}$$

with the screened atomic form factor (Thomas-Fermi or Moliere; the $183Z^{-1/3}$ factor is the complete-screening logarithm) absorbing the small-angle Coulomb integration. The Heitler 1954 Quantum Theory of Radiation Eq. XII is the canonical statement; Tsai 1974 (Rev. Mod. Phys. 46, 815) supplies the modern compendium with electron-electron incoherent and Mott-correction terms. The leading $1/\omega$ factor is the infrared divergence of QED; the bracket $[4/3-4\omega /(3E)+{\omega }^2/E^2]$ is the spectral shape; the logarithm is the impact-parameter integral cut off at the Thomas-Fermi screening radius $a_{\text{TF}}\sim a_0{Z}^{-1/3}$ (with $a_0$ the Bohr radius).

Radiation length

The radiation length $X_0$ is the characteristic distance over which a high-energy electron loses energy to bremsstrahlung. Define $-⟨dE/dx{⟩}_{\text{brems}}\equiv E/X_0$ for an electron at energy $E\gg E_c$ (with $E_c$ the critical energy at which bremsstrahlung and ionisation losses are equal: $E_c\approx 800/(Z+1.2)$ MeV, e.g. $E_c(\text{Pb})\approx 9.5$ MeV). The radiation length emerges from integrating the Bethe-Heitler spectrum over photon energies (Tsai 1974, BLP §93):

$$\boxed{X_0^{-1}=\frac{4Z(Z+1){\alpha }^3{N}_A\rho }{A{m}_e^2{c}^4}\cdot \log \left(\frac{287}{\sqrt{Z}}\right)+\text{small atomic-electron corrections}.}$$

The Mott correction $Z(Z+1)$ replaces the bare $Z^2$ by including the incoherent contribution of bremsstrahlung off atomic electrons (one extra electron per nucleus, treated incoherently as $Z$ uncorrelated targets). The numerical radiation length values used throughout high-energy physics are tabulated by the Particle Data Group (^{[PDG 2022 §34]}):

• Air: $X_0=36.62$ g/cm² $=304$ m at sea level.
• Water: $X_0=36.08$ g/cm² $=36$ cm.
• Aluminium: $X_0=24.01$ g/cm² $=8.9$ cm.
• Iron: $X_0=13.84$ g/cm² $=1.76$ cm.
• Lead: $X_0=6.37$ g/cm² $=0.561$ cm.
• Tungsten: $X_0=6.76$ g/cm² $=0.35$ cm.

The mean energy loss per radiation length is $⟨E(x)⟩=E_0{e}^{-x/X_0}$ — an exponentially decaying envelope in the ultra-relativistic regime, valid until $E$ drops below $E_c$ where ionisation losses take over.

Pair production by crossing

The crossed process $\gamma (k)+N(Ze)\to e^{+}(p_{+})+e^{-}(p_{-})+N$ shares the two tree-level diagrams of bremsstrahlung with the substitution $p\to -p_{+}$ (incoming electron of bremsstrahlung $\to$ outgoing positron of pair production, with reversed four-momentum), $k\to -k$ (outgoing photon $\to$ incoming photon), and a relabeling of the outgoing electron $p^{\prime }\to p_{-}$. The pair-production amplitude follows by analytic continuation of the bremsstrahlung amplitude; the squared amplitude and the spin sums carry over with the appropriate sign flips of the $v$-spinor mass projectors ${\sum }_s{v}^s(p_{+}){\bar{v}}^s(p_{+})=\text{slashed}{p}_{+}-m$ in place of the bremsstrahlung $\sum u\bar{u}=\text{slashed}p+m$.

The differential cross section, expressed as a function of the outgoing-positron-energy fraction $x=E_{+}/E_{\gamma }$ in the ultra-relativistic limit $E_{\gamma }\gg m_e$, is

$$\boxed{\frac{d{\sigma }_{\text{pair}}}{dx}=\frac{4Z^2{\alpha }^3}{m_e^2{c}^4}\left[x^2+(1-x{)}^2+\frac{2}{3}x(1-x)\right]\log \left[183Z^{-1/3}\right]+\mathcal{O}({\alpha }^4),}$$

with $x\in [m_e/E_{\gamma },1-m_e/E_{\gamma }]$ — symmetric in $x\leftrightarrow 1-x$ (electron $\leftrightarrow$ positron exchange is a symmetry of the QED Coulomb amplitude). The integrated cross section in the ultra-relativistic limit is the Rossi universality result

$${\sigma }_{\text{pair}}^{\infty }=\frac{7}{9}{\sigma }_{\text{brems}}^{\infty }=\frac{7}{9}\cdot \frac{A}{N_A\rho X_0}=\frac{7}{9}\cdot \frac{1}{X_0{n}_{\text{atoms}}},$$

i.e. the mean free path for pair production in the ultra-relativistic limit is $9/7$ of the bremsstrahlung radiation length. The factor $7/9$ is geometric (a phase-space integral over the bracketed spectral shape), tabulated as $7/9$ in Rossi 1952 (^{[Rossi 1952]}) and at the basis of every electromagnetic-shower parametrisation in high-energy physics.

Near-threshold pair production

At photon energies near the kinematic threshold $E_{\gamma }=2m_e{c}^2\approx 1.022$ MeV, the outgoing electron and positron are non-relativistic, ${\beta }_{\pm }=v_{\pm }/c\ll 1$. The pair-production cross section in this regime is

$${\sigma }_{\text{pair}}(E_{\gamma }\to 2m_e)\to \frac{2\pi Z^2{\alpha }^3}{3m_e^2{c}^4}\cdot {\beta }_{+}^3+\mathcal{O}({\beta }_{+}^5),$$

where ${\beta }_{+}=\sqrt{1-4m_e^2{c}^4/E_{\gamma }^2}$ is the velocity of the outgoing positron in the photon's rest frame (Bethe-Heitler 1934, ^{[Bethe-Heitler 1934]}). The ${\beta }^3$ threshold behaviour is the kinematic signature of the three-body phase space (the outgoing pair plus the recoiling nucleus): two factors of $\beta$ from the outgoing pair's phase space and one factor from the Coulomb-distortion matrix element.

LPM suppression in dense matter

The Landau-Pomeranchuk-Migdal effect is a coherence-destroying multiple-Coulomb-scattering suppression of bremsstrahlung at low photon energy in dense matter. The bremsstrahlung amplitude has a finite formation length ${\ell }_f=2E{E}^{\prime }/(\omega m_e^2)$ (in the ultra-relativistic limit), the longitudinal distance over which the radiating amplitude builds up coherently before the photon decouples from the electron. If the electron undergoes Coulomb scattering by an angle larger than ${\theta }_{\gamma }=m_e/E$ within this formation length, the radiation amplitudes from the pre- and post-scattering portions destructively interfere and the emission is suppressed.

The LPM characteristic photon energy is

$${\omega }_{\text{LPM}}=\frac{E^2}{E_{\text{LPM}}},E_{\text{LPM}}=\frac{m_e^2{c}^4{X}_0}{4\pi \alpha (\hslash c)}\approx 7.7\text{ TeV}\cdot X_0/\text{cm}.$$

For 25 GeV electrons in lead ($X_0=0.56$ cm), $E_{\text{LPM}}(\text{Pb})\approx 4.3$ TeV and ${\omega }_{\text{LPM}}\approx 25^2/4300\approx 0.14$ GeV — soft-photon emission below $\sim 140$ MeV is LPM-suppressed in lead at this energy. For 1 GeV electrons in lead, ${\omega }_{\text{LPM}}\approx 230$ keV. The LPM suppression was predicted by Landau-Pomeranchuk in 1953 (^{[Landau-Pomeranchuk 1953]}) and computed quantum-mechanically by Migdal in 1956 (^{[Migdal 1956]}); experimentally confirmed by the SLAC E-146 experiment (Anthony et al. 1995/1997, ^{[Anthony 1997]}) using 8-25 GeV electrons through carbon, aluminium, iron, tungsten, lead, and uranium targets at the SLAC End Station A.

Key derivation Intermediate+

Theorem (Bethe-Heitler bremsstrahlung cross section, ultra-relativistic complete-screening limit). The differential cross section for $e^{-}(p)+N(Ze)\to e^{-}(p^{\prime })+\gamma (k,\omega )+N$ in the ultra-relativistic limit $E,E^{\prime }\gg m_e{c}^2$ and the complete-screening regime $|\mathbf{q}{|}_{\text{min}}\ll a_{\text{TF}}^{-1}$ is

$$\frac{d\sigma }{d\omega }=\frac{4Z^2{\alpha }^3}{\omega m_e^2{c}^4}\left[\frac{4}{3}-\frac{4\omega }{3E}+\frac{{\omega }^2}{E^2}\right]\log \left[\frac{183Z^{-1/3}}{1+\omega /E}\right]+\mathcal{O}({\alpha }^4).$$

Proof. Start from the amplitude $\mathcal{M}={\mathcal{M}}_1+{\mathcal{M}}_2$ defined in the formal-definition section, with ${\mathcal{M}}_1$ the pre-emission and ${\mathcal{M}}_2$ the post-emission diagram. The spin-averaged photon-polarisation-summed squared amplitude is

$$\overline{|\mathcal{M}{|}^2}=\frac{1}{2}\sum _{s,s^{\prime }}\sum _{\lambda }|{\mathcal{M}}_1+{\mathcal{M}}_2{|}^2=\frac{1}{2}\sum |{\mathcal{M}}_1{|}^2+\frac{1}{2}\sum |{\mathcal{M}}_2{|}^2+\mathrm{Re}\sum {\mathcal{M}}_1{\mathcal{M}}_2^{\ast }.$$

Apply spin completeness ${\sum }_s{u}^s(p){\bar{u}}^s(p)=\text{slashed}p+m$ to fold the spinor structure into traces. For the $|{\mathcal{M}}_1{|}^2$ piece, after contracting the photon-polarisation sum ${\sum }_{\lambda }{ϵ}^{\ast \mu }{ϵ}^{\nu }=-{\eta }^{\mu \nu }$ (Feynman gauge) and the static external field ${\tilde{A}}_{\mu }^{\text{ext}}{\tilde{A}}_{\nu }^{\text{ext*}}(q)=|{\tilde{\mathbf{A}}}^{\text{ext}}(\mathbf{q}){|}^2\cdot {\eta }_{\mu 0}{\eta }_{\nu 0}$,

$$\frac{1}{2}\sum |{\mathcal{M}}_1{|}^2=\frac{e^4(Ze{)}^2}{2|\mathbf{q}{|}^4\cdot [(p-k{)}^2-m^2{]}^2}\mathrm{tr}\left[(\text{slashed}{p}^{\prime }+m){\gamma }^0(\text{slashed}p-\text{slashed}k+m){\gamma }^{\mu }(\text{slashed}p+m){\gamma }_{\mu }(\text{slashed}p-\text{slashed}k+m){\gamma }^0\right],$$

an eight-gamma trace of two electron-spinor projectors and the two virtual-electron projectors. After applying the contraction identity ${\gamma }^{\mu }\text{slashed}A{\gamma }_{\mu }=-2\text{slashed}A+4m$ (mass corrections kept), the trace reduces to a four-gamma trace. The calculation is long but systematic; the result is (Tsai 1974, ^{[Tsai 1974]}; cross-checked against the original Bethe-Heitler 1934 expression with the corrections to the high-energy logarithm noted by Olsen-Maximon 1959):

$$\frac{1}{2}\sum |{\mathcal{M}}_1{|}^2=\frac{16(Ze{)}^2{e}^4}{|\mathbf{q}{|}^4\cdot [(p-k{)}^2-m^2{]}^2}\cdot \left[(p\cdot p^{\prime })\left((p-k)\cdot p\right)-m^2(p\cdot k)+m^2{\omega }^2+\mathcal{O}(m^4)\right].$$

The $|{\mathcal{M}}_2{|}^2$ piece is the same with $p\to -p^{\prime }$ inside the propagator denominator and the appropriate sign flips on the virtual-electron projector — its master form follows by $p\leftrightarrow -p^{\prime }$ symmetry. The interference $2\mathrm{Re}({\mathcal{M}}_1{\mathcal{M}}_2^{\ast })$ requires a sixteen-gamma trace, which after the same contraction-identity reduction folds into a four-gamma trace with the structure of mixed pre- and post-emission contributions.

Integration over nuclear recoil. The squared amplitude depends on $|\mathbf{q}{|}^2$, the squared three-momentum transferred to the nucleus, in the Coulomb-form-factor denominator $1/|\mathbf{q}{|}^4$. The kinematics give $|\mathbf{q}{|}^2=({\mathbf{p}}^{\prime }+\mathbf{k}-\mathbf{p}{)}^2$, which has a minimum $|\mathbf{q}{|}_{\text{min}}^2=(m^2\omega /2E{E}^{\prime }{)}^2$ in the ultra-relativistic limit (the minimum momentum transfer needed to satisfy on-shell kinematics with a soft photon). The integration $\int d^3\mathbf{q}/|\mathbf{q}{|}^4$ is logarithmically divergent at large $|\mathbf{q}|$ and dominated by small $|\mathbf{q}|$; the divergence is cut off at large $|\mathbf{q}|$ by the atomic form factor $F(\mathbf{q})$ that accounts for the screening of the bare nuclear Coulomb field by the surrounding atomic electrons. The Thomas-Fermi or Moliere form factor gives the complete-screening cutoff at $|\mathbf{q}{|}_{\text{max}}\sim a_{\text{TF}}^{-1}=(a_0{Z}^{-1/3}{)}^{-1}$, where $a_0=(\alpha m_{e}c{)}^{-1}$ is the Bohr radius. The logarithmic integral is then

$$\log \frac{|\mathbf{q}{|}_{\text{max}}}{|\mathbf{q}{|}_{\text{min}}}=\log \frac{2E{E}^{\prime }}{m\omega a_{\text{TF}}m}=\log \frac{2E{E}^{\prime }}{m^2\omega \cdot a_0{Z}^{-1/3}m}=\log \frac{183Z^{-1/3}}{1+\omega /E}+\mathcal{O}(1),$$

where the numerical $183$ comes from precise evaluation of the Thomas-Fermi form factor at moderate $|\mathbf{q}|$ (Tsai 1974, with $183=(m_{e}c\cdot a_{\text{TF}})/\text{constants}\cdot e^{-{\gamma }_E}$ in detail).

Integration over outgoing-electron angle. The remaining integral is over the direction of the outgoing electron at fixed energy $E^{\prime }$, which after combining with the photon-angle integral and applying the soft-photon approximation in the appropriate kinematic regions, gives the bracket factor $[4/3-4\omega /(3E)+{\omega }^2/E^2]$. The $4/3$ at $\omega \to 0$ is the soft-photon spectral coefficient; the $-4\omega /(3E)$ is the linear correction at intermediate $\omega$; the $+{\omega }^2/E^2$ is the hardening at $\omega \to E$ where the outgoing electron is soft.

Collecting the prefactor. The Coulomb form factor $|{\tilde{A}}_0^{\text{ext}}(\mathbf{q}){|}^2=(4\pi Ze{)}^2/|\mathbf{q}{|}^4$ contributes $(Ze{)}^2$. The two QED vertices contribute $e^4$. Combining with the kinematic factors and the trace evaluation,

$$\frac{d\sigma }{d\omega }=\frac{4(Ze{)}^2{e}^4}{\omega \cdot m_e^2{c}^4}\cdot \left[\frac{4}{3}-\frac{4\omega }{3E}+\frac{{\omega }^2}{E^2}\right]\cdot \log \frac{183Z^{-1/3}}{1+\omega /E}\cdot \frac{1}{(4\pi {)}^4\cdot 4}\cdot (\text{phase-space and normalisation}).$$

After collecting all factors (carefully — the full Tsai 1974 computation occupies several pages of Reviews of Modern Physics), the coefficient simplifies to $4Z^2{\alpha }^3/(\omega m_e^2{c}^4)$, recovering the boxed Bethe-Heitler formula. $\square$

Corollary 1 (radiation length). The radiation length is obtained by integrating the Bethe-Heitler spectrum over the radiated energy, weighted by the energy lost per radiated photon:

$$\frac{E}{X_0}=-⟨dE/dx{⟩}_{\text{brems}}=n_{\text{atoms}}{\int }_0^E\omega \frac{d\sigma }{d\omega }d\omega ,$$

with $n_{\text{atoms}}=N_A\rho /A$ the atomic number density. Substituting the Bethe-Heitler $d\sigma /d\omega$ and evaluating the integral (with the photon-energy variable $u=\omega /E$ rescaled to $[0,1]$), $$ \frac{1}{X_0} = 4 n_\text{atoms} Z^2 \alpha^3 \log\frac{183}{Z^{1/3}}\int_0^1 \left[\frac{4}{3} - \frac{4 u}{3} + u^2\right] du = 4 n_\text{atoms} Z^2 \alpha^3 \log\frac{183}{Z^{1/3}} \cdot \frac{7}{9}, $$ after evaluating the bracket integral ${\int }_0^1[4/3-4u/3+u^2]du=4/3-2/3+1/3=1$ and then folding in the Mott $Z(Z+1)$ atomic-electron correction. The numerical $4\cdot 7/9\approx 3.11$ matches the conventional radiation-length formula up to the $\log (287/\sqrt{Z})$ normalisation (which absorbs the Coulomb correction to $\log (183Z^{-1/3})$ at large $Z$).

Corollary 2 (Rossi 7/9 universality). The asymptotic pair-production cross section at $E_{\gamma }\gg m_e$ in the complete-screening regime is $$ \sigma_\text{pair}^\infty = n_\text{atoms} \int_0^1 \frac{d\sigma_\text{pair}}{dx},dx = \frac{4 Z^2 \alpha^3}{m_e^2 c^4}\log(183 Z^{-1/3})\int_0^1 [x^2 + (1-x)^2 + \tfrac{2}{3}x(1-x)]dx. $$ The bracket integral evaluates to ${\int }_0^1[x^2+(1-x{)}^2]dx+(2/3){\int }_0^{1}x(1-x)dx=2/3+1/9=7/9$. The asymptotic pair-production cross section is $$ \sigma_\text{pair}^\infty = \frac{7}{9}\cdot \frac{4 Z^2\alpha^3}{m_e^2 c^4}\log(183 Z^{-1/3}) = \frac{7}{9}\cdot \frac{1}{n_\text{atoms}X_0}, $$ reproducing ${\sigma }_{\text{pair}}{X}_0=7/9\cdot (1/n_{\text{atoms}})$, the Rossi universality. Physically: the mean free path for pair production in the ultra-relativistic limit is $9/7$ radiation lengths, independent of the material — a universality theorem that organises the longitudinal development of electromagnetic showers in all calorimeter materials.

Corollary 3 (soft-photon factorisation and Bloch-Nordsieck). In the soft-photon limit $\omega \to 0$, the Bethe-Heitler spectrum factorises as $$ \frac{d\sigma_\text{brems}}{d\omega} = \frac{d\sigma_\text{elastic}}{d\Omega_{e'}}\cdot \frac{\alpha}{\pi}\frac{d\omega}{\omega}\cdot \log\frac{4 E E'}{m_e^2 c^4} + \mathcal O(\omega), $$ where $d{\sigma }_{\text{elastic}}/d{\Omega }_{e^{\prime }}$ is the unradiative Mott cross section for the same electron-nucleus elastic scattering. The soft-photon factor $(\alpha /\pi )(d\omega /\omega )\log [4E{E}^{\prime }/m^2]$ is the universal infrared structure of QED; integrated over photon energy from ${\omega }_{\text{min}}$ to a detector cutoff $\Delta E$ it gives $(\alpha /\pi )\log (\Delta E/{\omega }_{\text{min}})\log [4E{E}^{\prime }/m^2]$. The corresponding virtual one-loop QED correction to the elastic Coulomb-scattering amplitude has the IR-divergent piece $(\alpha /\pi )\log ({\mu }_{\gamma }/{\omega }_{\text{min}})\log [4E{E}^{\prime }/m^2]$ with the opposite sign (with ${\mu }_{\gamma }$ the photon-mass regulator); the sum is finite, depending only on $\log (\Delta E/E)$, the experimental detector resolution as a fraction of the beam energy. This is the Bloch-Nordsieck cancellation 12.16.05 applied to the Bethe-Heitler bremsstrahlung-and-elastic-scattering pair.

Worked example: 100 MeV bremsstrahlung in lead

For an incident electron of energy $E=100$ MeV in lead ($Z=82$, $X_0=0.561$ cm, $\rho =11.35$ g/cm³), compute the differential cross section for emission of a $\omega =10$ MeV photon and the radiated energy per unit path length.

The bracket factor at $\omega /E=0.1$ is $[4/3-4\cdot 0.1/3+0.01]=[1.333-0.133+0.01]=1.210$. The screening logarithm with $Z^{1/3}=82^{1/3}\approx 4.34$, $183/Z^{1/3}\approx 42.2$, and $\omega /E=0.1$ gives $\log [42.2/1.1]=\log 38.4\approx 3.65$.

The differential cross section is $$ \frac{d\sigma}{d\omega} = \frac{4 \cdot 82^2 \cdot (1/137)^3}{10 \cdot (0.511)^2 \text{ MeV}^2}\cdot 1.210\cdot 3.65 \approx \frac{4 \cdot 6724 \cdot 3.89 \times 10^{-7}}{2.61 \text{ MeV}^2}\cdot 4.42 \approx \frac{1.05 \times 10^{-2}}{2.61 \text{ MeV}^2}\cdot 4.42 \approx 1.77 \times 10^{-2} \text{ MeV}^{-3}. $$

Converting $1$ MeV${}^{-2}$ $\approx 3.89\times 10^{-22}$ cm², $d\sigma /d\omega \approx 1.77\times 10^{-2}\cdot 3.89\times 10^{-22}/\text{MeV}\approx 6.9\times 10^{-24}$ cm²/MeV $=6.9$ b/MeV per nucleus, an order-of-magnitude consistent with the Tsai 1974 tabulated value $\sim 5$ b/MeV at this kinematic point (the small discrepancy traces to the difference between the simplified high-energy logarithm used here and the full Tsai expression with electron-electron and Coulomb-correction terms).

The atomic density in lead is $n_{\text{atoms}}=N_A\rho /A=6.022\times 10^{23}\cdot 11.35/207\approx 3.30\times 10^{22}$/cm³. The energy-loss rate from emitting 10 MeV photons (at this single $\omega$) is roughly $\omega \cdot n_{\text{atoms}}\cdot d\sigma /d\omega \cdot d\omega \sim 10\cdot 3.30\times 10^{22}\cdot 6.9\times 10^{-24}\approx 2.3$ MeV/cm per unit $d\omega$ at $\omega =10$ MeV. Integrating over the full photon-energy spectrum gives the energy-loss rate $⟨dE/dx{⟩}_{\text{brems}}=E/X_0=100$ MeV $/0.561$ cm $\approx 178$ MeV/cm. A 100 MeV electron traverses about $E/(E/X_0)=X_0=0.56$ cm before losing $1-1/e\approx 63\%$ of its energy to bremsstrahlung — consistent with the radiation-length formula.

Exercises Intermediate+

Lean formalization Intermediate+

Mathlib has fragments of the infrastructure: finite-dimensional inner-product spaces, the Lorentz-Minkowski quadratic form, and the abstract Clifford algebra Mathlib.LinearAlgebra.CliffordAlgebra.Basic from which the Cl(1,3) gamma matrices specialise. What is missing for Bethe-Heitler is the full external-field QED stack:

• An operator-valued-distribution model of the quantised Dirac and Maxwell fields on Fock space with the canonical anticommutation and commutation relations.
• A Furry-picture coupling of the quantised Dirac field to an external classical $c$-number background field $A_{\mu }^{\text{ext}}(x)$, with the static-Coulomb specialisation $A_0^{\text{ext}}=Ze/(4\pi r)$ as a building block.
• The tree-level S-matrix on this external-field background, with the two diagrams (photon emission before or after Coulomb scattering, equivalently pair-creation in the field).
• The gamma-matrix trace technology with the standard four- and six-gamma trace identities and the Chisholm contraction identity (shared with all QED tree-level units; aggregated as a Mathlib.Physics.QED.TraceTechnology namespace).
• The three-body Lorentz-invariant phase-space measure for the outgoing electron, photon, and recoiling nucleus, with the heavy-nucleus limit that decouples the nuclear motion.
• The atomic form factor $F(\mathbf{q})$ (Thomas-Fermi or Moliere) for screening of the nuclear Coulomb field at large impact parameters.
• The Bloch-Nordsieck cancellation theorem identifying the IR-divergent virtual one-loop QED contribution with the IR-divergent soft real-photon emission.
• The LPM coherence-length analysis for the multiple-Coulomb-scattering suppression of bremsstrahlung in dense matter.

lean_status: none. The human-reviewer surface for external-field QED tree-level cross sections is the verification of the two-diagram amplitude assembly and trace algebra (Master derivation), the verification of the crossing-symmetry mapping from bremsstrahlung to pair production, the verification of the radiation-length integration, the verification of the soft-photon factorisation, and the verification of the LPM formation-length argument. Aggregated with the other QED tree-level units in chapter 12.12 (Compton 12.12.03, Møller 12.12.04, Bhabha 12.12.05), this feeds the upstream Mathlib physics-formalisation roadmap; the priority sub-goals are Mathlib.Physics.QED.TraceTechnology, Mathlib.Physics.QED.FeynmanRules.TreeAmplitudes, Mathlib.Physics.QED.ExternalField.Furry, and Mathlib.Physics.QED.IR.BlochNordsieck. Once these are in place, Bethe-Heitler bremsstrahlung and pair production are direct applications of the same library, with crossing implementing the bremsstrahlung-to-pair-production correspondence.

Advanced results Master

Tsai 1974 modern compendium

The original 1934 Bethe-Heitler calculation gave the ultra-relativistic complete-screening formula to leading order in $\alpha$. Olsen and Maximon (^{[source pending]}) corrected the small-angle Coulomb integration in 1959. Tsai 1974 (^{[Tsai 1974]}) compiled the modern result with several corrections beyond the leading Bethe-Heitler 1934 expression: the electron-electron incoherent contribution (bremsstrahlung off the atomic electrons, treated as $Z$ uncorrelated targets with appropriate form factors), the Coulomb correction to the Born-approximation amplitude at large $Z$ (the Bethe-Maximon series, with leading term $-f(Z\alpha )$ where $f(x)=x^2{\sum }_{n=1}^{\infty }1/(n(n^2+x^2))\approx x^2(1.202-x^2/2+\dots )$, the Riemann-$\zeta (3)$ value at small $Z\alpha$), and dispersion-relation corrections from the heavy-nucleus elastic form factor. The corrected Tsai formula reads $$ \frac{d\sigma}{d\omega} = \frac{4\alpha r_0^2}{\omega}\left{Z(Z+1)\left[\frac{4}{3} - \frac{4\omega}{3 E} + \frac{\omega^2}{E^2}\right]\log\frac{183 Z^{-1/3}}{1 + \omega/E} - \frac{Z^2}{3}f(Z\alpha)\left[\frac{4}{3} - \frac{4\omega}{3 E} + \frac{\omega^2}{E^2}\right] + Z(Z+1)\cdot\delta_\text{atomic}(\omega)\right}, $$ with $r_0=e^2/(m_e{c}^2)\approx 2.818$ fm the classical electron radius. The Coulomb correction $-Z^{2}f(Z\alpha )/3$ is a few-per-cent reduction in lead and significant in uranium; the Mott $Z(Z+1)$ correction is at the 1% level for low-$Z$ materials. These corrections to the leading Bethe-Heitler formula are what the Particle Data Group's tabulated radiation lengths actually use.

Coherent and channeling effects

For electrons traversing a single crystal at small angles to the crystallographic axes, the bremsstrahlung amplitude can constructively interfere across lattice planes, producing coherent bremsstrahlung — discrete peaks in the photon-energy spectrum at multiples of the reciprocal-lattice-vector kinematic resonance. Coherent bremsstrahlung was predicted by Übermann 1956 and observed in numerous experiments since the 1960s. The mirror process is channelling radiation: electrons travelling along crystallographic axes are confined to potential wells between atomic rows and undergo undulator-like motion, producing intense quasi-monoenergetic photon beams. Both effects are routinely used as quasi-monochromatic photon sources (the SLAC photon beam line at SPEAR; the Stanford SSRL; the European XFEL undulator and channeling stations). The Bethe-Heitler cross section is the incoherent baseline against which coherent and channelling enhancements are measured.

Inverse Compton and synchrotron alternatives

The Bethe-Heitler bremsstrahlung is the dominant electron energy-loss mechanism in cold matter. In hot plasmas (T $\sim$ keV) the thermal bremsstrahlung (free-free emission) is the analogous cross section averaged over a Maxwellian electron distribution, dominant in stellar interiors at T $\sim 10^7$ K and the source of the soft X-ray continuum in supernova-remnant shells and galaxy clusters. In high-magnetic-field environments (pulsars, neutron-star magnetospheres), synchrotron radiation ($P_{\text{sync}}=2e^4{B}^2{E}^2/(3m_e^4{c}^5)$ for ultra-relativistic electrons) and inverse Compton scattering ($P_{\text{IC}}=4{\sigma }_{T}c{U}_{\text{ph}}(E/m_e{c}^2{)}^2/3$ on ambient photon fields with energy density $U_{\text{ph}}$) take over as the dominant cooling channels. For ultra-high-energy cosmic rays in the cosmic microwave background, inverse Compton scattering off CMB photons sets the GZK cutoff at $\sim 5\times 10^{19}$ eV — the cosmic-ray analogue of the bremsstrahlung cutoff in cold matter. Bethe-Heitler is the cold-matter limit; synchrotron and inverse-Compton are the alternative regimes for hot or magnetised plasmas.

Schwinger-limit non-perturbative pair production

In a static external electric field $E_{\text{ext}}$ exceeding the Schwinger critical field $E_S=m_e^2{c}^3/(e\hslash )=1.32\times 10^{18}$ V/m, vacuum pair production becomes non-perturbative in $\alpha$ (Schwinger 1951, Phys. Rev. 82, 664). The Schwinger production rate is $w_S=(e{E}_{\text{ext}}{)}^2/(4{\pi }^3\hslash c)\cdot \exp (-\pi E_S/E_{\text{ext}})$, exponentially suppressed below the critical field. Laboratory laser intensities $I\sim 10^{22}$ W/cm² (peak intensity at the Texas Petawatt and the European Extreme Light Infrastructure) give electric fields of $\sim 10^{14}$ V/m, six orders of magnitude below the Schwinger critical field; non-perturbative pair production has not been observed in the laboratory. The Bethe-Heitler perturbative pair-production cross section is the high-$\hslash \omega$ analogue: the photon $\gamma$ of energy $\gg 2m_e{c}^2$ creates a pair perturbatively in $\alpha$ via the Coulomb-field-assisted Bethe-Heitler process, whereas the Schwinger mechanism would create pairs even from zero-frequency static fields above the critical strength.

Full proof set Master

The derivation of the Bethe-Heitler differential cross section in the Key derivation is complete in outline. The technical step that warrants further justification is the complete-screening logarithm $\log [183Z^{-1/3}]$, obtained from the integration of the differential cross section over the nuclear three-momentum transfer $\mathbf{q}$ with the Thomas-Fermi atomic form factor.

The bare nuclear Coulomb potential has Fourier transform ${\tilde{V}}_{\text{bare}}(\mathbf{q})=4\pi Ze/|\mathbf{q}{|}^2$. The screened Thomas-Fermi potential, after the surrounding atomic electrons cancel the nuclear charge at distances $r>a_{\text{TF}}$, has Fourier transform $$ \tilde V_\text{TF}(\mathbf q) = \frac{4\pi Ze}{|\mathbf q|^2 + a_\text{TF}^{-2}} \cdot F_\text{TF}(\mathbf q a_\text{TF}), $$ with $a_{\text{TF}}=(9{\pi }^2/128{)}^{1/3}\cdot a_0/Z^{1/3}\approx 0.885a_0{Z}^{-1/3}$ the Thomas-Fermi screening radius and $F_{\text{TF}}(x)$ a slowly-varying form factor of order unity (tabulated as Moliere's three-term approximation $F_{\text{TF}}(x)\approx 0.1e^{-6x}+0.55e^{-1.2x}+0.35e^{-0.3x}$).

The bremsstrahlung amplitude depends on $|{\tilde{V}}_{\text{TF}}(\mathbf{q}){|}^2$ in the Coulomb-form-factor denominator. The integration $\int d|\mathbf{q}|/|\mathbf{q}|$ from $|\mathbf{q}{|}_{\text{min}}=m^2\omega /(2E{E}^{\prime })$ (ultra-relativistic kinematic minimum) to $|\mathbf{q}{|}_{\text{max}}\sim a_{\text{TF}}^{-1}$ (Thomas-Fermi screening cutoff) gives $$ \log\frac{|\mathbf q|\text{max}}{|\mathbf q|\text{min}} = \log\frac{1/a_\text{TF}}{m^2\omega/(2 E E')} = \log\frac{2 E E'}{m^2\omega a_\text{TF}} = \log\frac{2 E E'}{m^2\omega \cdot 0.885 a_0 Z^{-1/3}}. $$ Using $a_0=(\alpha m_e{)}^{-1}$ in natural units, $a_{0}m={\alpha }^{-1}\approx 137$, so $1/(m{a}_{\text{TF}})=(\alpha m{a}_0/Z^{-1/3}{)}^{-1}\cdot \text{const}=(Z^{1/3}/\alpha \cdot 137\cdot 0.885{)}^{-1}$. Numerical evaluation with the Thomas-Fermi form factor explicit (the Moliere three-term approximation gives a slightly different effective screening radius than the bare $0.885a_0{Z}^{-1/3}$) yields the famous $\log 183$ factor: $$ \log\frac{|\mathbf q|\text{max}}{|\mathbf q|\text{min}} = \log\frac{183, Z^{-1/3}}{1 + \omega/E}, $$ where the $1+\omega /E$ in the denominator absorbs the $\omega$-dependence of the kinematic minimum. The numerical coefficient $183$ is the canonical value used throughout the Tsai 1974 compendium and the Particle Data Group tabulated radiation lengths.

The integration over outgoing electron angles, working from the squared amplitude structure derived in the Key derivation section, yields the bracket factor $[4/3-4\omega /(3E)+{\omega }^2/E^2]$. The three terms in the bracket trace to three kinematic regions of the angle integral: the bulk of the angular phase space ($4/3$ in the small-$\omega$ limit), the angular-recoil correction ($-4\omega /(3E)$ at intermediate $\omega$), and the hard-photon edge ($+{\omega }^2/E^2$ at $\omega \to E$ where the outgoing electron carries little energy). The detailed angle-integration calculation occupies several pages in Tsai 1974 §III.

Collecting all factors gives the boxed Bethe-Heitler differential cross section.

The radiation length follows by the spectral integration of Corollary 1. The bracket integral ${\int }_0^1[4/3-4u/3+u^2]du=1$ (the standard energy-loss-normalised Bethe-Heitler bracket evaluates to unity); folding in the $Z(Z+1)$ Mott correction and the numerical $\log (287/Z^{1/2})$ form gives the Particle Data Group radiation-length formula.

The pair-production cross section follows by the crossing $p\to -p_{+}$ derived in the formal-definition section. The spectral integration of Corollary 2 gives the Rossi $7/9$ universality with bracket integral ${\int }_0^1[x^2+(1-x{)}^2+\frac{2}{3}x(1-x)]dx=7/9$.

The soft-photon factorisation of Corollary 3 follows by direct expansion of the bremsstrahlung amplitude in the limit $\omega \to 0$: the virtual-electron propagators $1/(p\mp k{)}^2-m^2=1/(\mp 2p\cdot k)$ become $\propto 1/\omega$, while the spinor sandwich $\bar{u}(p^{\prime }){\gamma }^{\mu }(\text{slashed}p\pm \text{slashed}k+m)\text{slashed}\tilde{A}u(p)$ in the numerator simplifies via $(\text{slashed}p+m){\gamma }^{\mu }\to 2p^{\mu }$ between on-shell projectors, giving the eikonal current $J^{\mu }=-ie(p^{\prime \mu }/p^{\prime }\cdot k-p^{\mu }/p\cdot k)$ (Exercise 3). The squared current $|J\cdot {ϵ}^{\ast }{|}^2\propto 1/(p\cdot k)$ combined with the photon phase space $d^{3}k/(2\omega (2\pi {)}^3)=\omega d\omega d{\Omega }_{\gamma }/(2(2\pi {)}^3)$ gives the $d\omega /\omega$ logarithmic IR-divergent soft-photon spectrum. This is the eikonal limit of QED — the soft-photon emission from any QED process factorises into the universal eikonal current times the underlying elastic amplitude.

The LPM suppression of bremsstrahlung in dense matter (Exercise 8) follows from the formation-length argument: the bremsstrahlung formation length ${\ell }_f=2E{E}^{\prime }/(m^2\omega )$ exceeds the multiple-Coulomb-scattering coherence length ${\ell }_{\text{MS}}=X_0\alpha /(4\pi )$ when $\omega <E^2/E_{\text{LPM}}$. Migdal 1956 derived the quantum-mechanical suppression factor $S(\omega /{\omega }_{\text{LPM}})$ from the full Dirac-equation amplitude with multiple Coulomb scattering; in the deep-LPM regime $S\approx \sqrt{\omega /{\omega }_{\text{LPM}}}$, suppressing the soft-photon $1/\omega$ spectrum to $1/\sqrt{\omega {\omega }_{\text{LPM}}}$.

The collection of these derivations gives the complete Bethe-Heitler-and-pair-production calculation, the radiation-length material constant, the universality theorem organising electromagnetic showers, the soft-photon factorisation entering the Bloch-Nordsieck cancellation, and the LPM modification in dense matter — the cross sections that organise sixty years of electromagnetic-cascade physics from cosmic-ray air showers to LHC calorimeters to medical-linac radiotherapy.

Connections Master

• Canonical quantum field theory 12.12.01 is the framework: the QED Lagrangian, the Feynman rules with quantised-photon and external-source vertices, the Furry picture for external-field QED. Bethe-Heitler bremsstrahlung is the external-field QED process par excellence; the techniques developed here (external-field amplitude assembly, atomic form factors, IR factorisation) are the template for every external-field QED calculation.

• Compton scattering and the Klein-Nishina formula 12.12.03 uses the same gamma-matrix trace technology and the same spin-completeness sums. Compton has two external photons (transverse polarisation sums); Bethe-Heitler has one external photon and one external Coulomb source. Both are two-diagram tree-level QED processes with the same kinematic skeleton; the cross section ratios at fixed kinematic point are predictions of the unified QED amplitude family.

• Møller scattering 12.12.04 is the cleanest demonstration of the Wick-contraction algebra producing relative-minus signs between exchange diagrams. Bethe-Heitler shares the trace-technology stack and the external-source-as-Coulomb-photon framing — substituting the static Coulomb potential of a heavy nucleus for the second electron line of Møller gives the bremsstrahlung topology. The Mott $Z(Z+1)$ correction to the radiation length explicitly accounts for the incoherent electron-electron-bremsstrahlung contribution.

• Bhabha scattering 12.12.05 is the electron-positron analogue of Møller, with the additional $s$-channel annihilation diagram. The Bethe-Heitler pair-production cross section is the crossed-channel analogue with the second lepton replaced by the static Coulomb source. The infrared structure of Bhabha at $e^{+}{e}^{-}$ colliders (initial-state radiation, YFS soft-photon exponentiation) is the high-energy analogue of the Bethe-Heitler soft-photon spectrum in matter; both share the universal $1/\omega$ eikonal factor and the Bloch-Nordsieck cancellation with the corresponding virtual one-loop QED corrections.

• Dirac equation and relativistic spin 12.11.01 supplies the four-component electron and positron spinors. The crossing relation between bremsstrahlung and pair production is the cleanest external-field example of Feynman's antiparticle-as-backward-going-electron prescription applied at the level of physical processes — the outgoing positron of pair production is the analytic continuation of an incoming electron of bremsstrahlung.

• Free Dirac quantum field 12.05.05 provides the operator-valued-distribution Dirac field with both electron creation operators $a_s^{†}(p)$ and positron creation operators $b_s^{†}(p)$; the same field structure underlies both bremsstrahlung (one electron in each state) and pair production (one positron and one electron in the final state).

• Infrared divergences and the Bloch-Nordsieck mechanism 12.16.05 is the theoretical apparatus that renders the $1/\omega$ Bethe-Heitler soft-photon divergence finite. The virtual one-loop QED correction to elastic Coulomb scattering and the real soft-bremsstrahlung emission from any external-field amplitude have IR-divergent contributions of equal magnitude and opposite sign; their sum at the inclusive cross-section level is finite. The Bethe-Heitler bremsstrahlung is the canonical real-emission half of the cancellation worked example.

• One-loop QED vertex function and the anomalous magnetic moment 12.16.02 dresses the electron-photon vertex at one loop in $\alpha$; the same QED vertex correction is the leading-order correction to the elastic Coulomb-scattering amplitude that combines with the soft-bremsstrahlung emission to give the IR-finite inclusive cross section.

• Electromagnetic-shower physics and detector calorimetry is the dominant application. Every electromagnetic calorimeter at every collider experiment (LHC ATLAS / CMS / LHCb / ALICE; B-factories; future ILC / FCC-ee / FCC-hh) sits on top of the Bethe-Heitler cross section. The longitudinal shower profile (Rossi-Heitler model with universal $7/9$ pair-production-to-bremsstrahlung ratio), the transverse spread (Moliere radius), and the depth-of-shower-max (Longo-Sestili parametrisation) all derive from the asymptotic Bethe-Heitler-and-pair-production cross sections.

• Medical-linac radiotherapy is the largest practical application by economic value. The 4-25 MeV bremsstrahlung X-ray beams produced by clinical linacs (Varian, Elekta, Siemens) treat $\sim 10^7$ cancer patients per year worldwide. The Bethe-Heitler bremsstrahlung yield from tungsten targets, the depth-dose curve in tissue (governed by Compton scattering in the Bethe-Heitler-produced photon spectrum), and the photonuclear-secondary-neutron yield at energies above $\sim 8$-$10$ MeV all derive from the Bethe-Heitler cross section.

• Cosmic-ray air-shower physics uses Bethe-Heitler bremsstrahlung-and-pair-production as the kernel of the electromagnetic cascade. The Pierre Auger Observatory, the Telescope Array, and the planned IceCube-Gen2 surface arrays all reconstruct primary cosmic-ray energies from electromagnetic-cascade longitudinal-profile fits anchored on Rossi-Heitler shower theory. LPM suppression is a measurable correction at the highest energies ($E>10^{19}$ eV) and is a target effect for next-generation observatories.

• Astrophysical inverse-bremsstrahlung (free-free emission) in hot relativistic plasmas — galaxy-cluster intracluster medium at T $\sim 10^7$-$10^8$ K, supernova-remnant shells, the solar corona at T $\sim 2\times 10^6$ K, and ADAF accretion flows around supermassive black holes — produces the soft X-ray continuum mapped by Chandra, XMM-Newton, and the planned Athena. The Bethe-Heitler cross section integrated over a Maxwellian electron velocity distribution gives the thermal bremsstrahlung emissivity ${ϵ}_{\text{ff}}(T,n_e,n_i)\propto n_e{n}_i{T}^{-1/2}\exp (-h\nu /k_{B}T)$, the workhorse formula for X-ray-luminosity-to-mass conversion in cluster cosmology.

Historical & philosophical context Master

Hans Bethe and Walter Heitler's 1934 paper On the Stopping of Fast Particles and on the Creation of Positive Electrons (Proceedings of the Royal Society A 146, 83-112) appeared two years after Anderson's experimental discovery of the positron, six years after Dirac's theoretical prediction of antimatter from the relativistic electron equation, and during the brief window when Dirac's positron-hole theory was the operative framework for relativistic quantum mechanics. Bethe (a 28-year-old Berlin-trained theorist who would emigrate to the United States in 1935 to escape the Nazi regime and become director of the theoretical division of the Manhattan Project) and Heitler (a 30-year-old Göttingen-trained theorist who would emigrate to England in 1933 and to Dublin in 1941 at de Valera's invitation) computed both bremsstrahlung and pair production in a single 30-page paper that immediately became the canonical reference for high-energy electromagnetic interactions in matter ^{[Bethe-Heitler 1934]}. The Sauter 1934 paper ^{[source pending]} derived the pair-production cross section independently within months of Bethe-Heitler.

The 1934 calculation used Dirac's second-order time-ordered perturbation theory; the modern Feynman-diagram derivation was supplied by Schwinger in 1949 and by Feynman in 1949 as part of the post-war reformulation of quantum electrodynamics. The screened-Coulomb $\log [183Z^{-1/3}]$ logarithm — derived in the original Bethe-Heitler paper via direct evaluation of the Thomas-Fermi integral — was refined by Olsen and Maximon in 1959 with corrections to the small-angle Coulomb integration, and compiled in modern form by Yu-Shiou Tsai in 1974 (Reviews of Modern Physics 46, 815) ^{[Tsai 1974]}. The Particle Data Group's tabulated radiation lengths (Bethe-Heitler integrated cross sections with Mott $Z(Z+1)$ atomic-electron, Coulomb Bethe-Maximon, and density-correction terms included) are the canonical reference values used throughout high-energy and nuclear physics.

The infrared-divergence structure of the Bethe-Heitler soft-photon spectrum was the first appearance of the IR problem in QED. Bloch and Nordsieck 1937 (Phys. Rev. 52, 54) proved that the IR divergence of the soft-bremsstrahlung emission cancels exactly against the IR divergence of the virtual one-loop QED correction to the elastic-scattering amplitude at the level of inclusive physical cross sections ^{[Bloch-Nordsieck 1937]}. The Bloch-Nordsieck cancellation theorem (later strengthened to the all-orders Yennie-Frautschi-Suura 1961 exponentiation) is the prototypical example of an IR cancellation in a gauge theory; the abelian Bloch-Nordsieck extends to the non-abelian Lee-Nauenberg theorem and underlies all of perturbative QCD's IR-safety machinery.

The Landau-Pomeranchuk-Migdal suppression (Landau and Pomeranchuk 1953 in Doklady Akademii Nauk SSSR; Migdal 1956 in Physical Review with the covariant quantum-mechanical derivation) extended the Bethe-Heitler theory into the dense-matter regime where multiple Coulomb scattering destroys the coherence required for bremsstrahlung over the formation length ^{[Landau-Pomeranchuk 1953; Migdal 1956]}. LPM was a prediction of essentially incomputable accuracy for forty years — observational confirmation required electron beams in the multi-GeV regime through targets thick enough to give measurable multiple scattering, a kinematic combination unavailable until the 1990s. The SLAC E-146 experiment (Anthony et al. 1995 in Phys. Rev. Lett. 75, 1949 and 1997 in Phys. Rev. Lett. 79, 1949) used 8-25 GeV electrons through carbon, aluminium, iron, tungsten, lead, and uranium targets at the SLAC End Station A and definitively confirmed the LPM suppression at the few-per-cent level across all six materials ^{[Anthony 1997]}. Klein 1998 (Rev. Mod. Phys. 71, 1501) is the modern review ^{[Klein 1998]}.

The Bethe-Heitler cross section is the calculational engine of every electromagnetic-cascade simulation in physics, from the EGS Monte Carlo of Ford-Nelson 1978 at SLAC through GEANT4 of CERN (Agostinelli et al. 2003 Nucl. Instrum. Meth. A 506, 250) to FLUKA and PENELOPE, used worldwide in particle physics (every LHC calorimeter), medical physics (every radiotherapy treatment plan), reactor physics (every shielding calculation), and space mission design (every radiation-effect estimate for satellite electronics). The Bethe-Heitler cross section has propagated through nine decades of physics applications with no loss of accuracy — the modern Tsai 1974 compendium retains the leading Bethe-Heitler 1934 structure as its dominant term and is accurate to better than 1% over the kinematic range relevant to all practical applications ^{[Tsai 1974; PDG 2022]}.

Bibliography Master

Primary literature:

• Bethe, H. A. & Heitler, W., "On the Stopping of Fast Particles and on the Creation of Positive Electrons", Proc. Roy. Soc. A 146 (1934), 83–112. [Originator: joint derivation of bremsstrahlung and pair production via crossing in Dirac's positron theory.]

• Sauter, F., "Über die Bremsstrahlung schneller Elektronen", Ann. Phys. 20 (1934), 404–444. [Independent derivation of the pair-production cross section; co-discovered with Bethe-Heitler.]

• Heitler, W., The Quantum Theory of Radiation, 3rd ed., (Oxford UP, 1954). [Textbook canonical statement of the Bethe-Heitler cross section; §XII Eq. XII is the modern shorthand.]

• Bloch, F. & Nordsieck, A., "Note on the Radiation Field of the Electron", Phys. Rev. 52 (1937), 54–59. [Infrared cancellation between virtual one-loop QED corrections and soft real-photon emission.]

• Schwinger, J., "Quantum Electrodynamics. III. The Electromagnetic Properties of the Electron — Radiative Corrections to Scattering", Phys. Rev. 76 (1949), 790–817. [Covariant Feynman-rule derivation of Bethe-Heitler, Compton, Møller, and Bhabha in the post-war reformulation of QED.]

• Olsen, H. A. & Maximon, L. C., "Photon and Electron Polarization in High-Energy Bremsstrahlung and Pair Production with Screening", Phys. Rev. 114 (1959), 887–904. [Corrections to the original Bethe-Heitler small-angle Coulomb integration; polarisation-resolved cross sections.]

• Bethe, H. A. & Maximon, L. C., "Theory of Bremsstrahlung and Pair Production. I. Differential Cross Section", Phys. Rev. 93 (1954), 768–784. [Coulomb correction beyond the Born approximation at large $Z$; the Bethe-Maximon series with leading $-f(Z\alpha )$ term.]

• Tsai, Y.-S., "Pair Production and Bremsstrahlung of Charged Leptons", Rev. Mod. Phys. 46 (1974), 815–851; erratum 49 (1977), 421. [Modern compendium of Bethe-Heitler-class cross sections; the radiation-length formula used throughout the Particle Data Group tabulations.]

• Landau, L. D. & Pomeranchuk, I. Ya., "Limits of Applicability of the Theory of Bremsstrahlung Electrons and Pair Production at High Energies", Dokl. Akad. Nauk SSSR 92 (1953), 535–536 and 735–738. [Originator: multiple-Coulomb-scattering coherence-destroying suppression of bremsstrahlung in dense matter.]

• Migdal, A. B., "Bremsstrahlung and Pair Production in Condensed Media at High Energies", Phys. Rev. 103 (1956), 1811–1820. [Covariant quantum-mechanical derivation of LPM suppression; the modern reference formula.]

• Anthony, P. L. et al. (SLAC E-146 Collaboration), "An accurate measurement of the Landau-Pomeranchuk-Migdal effect", Phys. Rev. Lett. 75 (1995), 1949–1952; "Bremsstrahlung Suppression due to the LPM and Dielectric Effects in a Variety of Materials", Phys. Rev. Lett. 79 (1997), 1949–1953. [First definitive experimental confirmation of LPM suppression; 8-25 GeV electrons through six target materials.]

• Klein, S. R., "Suppression of Bremsstrahlung and Pair Production due to Environmental Factors", Rev. Mod. Phys. 71 (1998), 1501–1538. [Comprehensive review of LPM, dielectric, and related coherence-suppression effects.]

• Schwinger, J., "On Gauge Invariance and Vacuum Polarization", Phys. Rev. 82 (1951), 664–679. [Schwinger critical-field $E_S=m^2{c}^3/(e\hslash )$ for non-perturbative pair production from constant electric fields; the strong-field analogue of perturbative Bethe-Heitler.]

Textbooks and monographs:

• Berestetskii, V. B., Lifshitz, E. M. & Pitaevskii, L. P., Quantum Electrodynamics, 2nd ed. (Pergamon, 1982), §§92-94. [Landau-Lifshitz Vol. 4 treatment with explicit two-diagram amplitude assembly, trace algebra, and radiation-length integration.]
• Jackson, J. D., Classical Electrodynamics, 3rd ed. (Wiley, 1999), §15.6-15.7. [Bremsstrahlung in classical and quantum frameworks; the classical-dipole-radiation estimate as a check on the QED Bethe-Heitler result.]
• Peskin, M. E. & Schroeder, D. V., An Introduction to Quantum Field Theory (Westview / Addison-Wesley, 1995), §6.1. [Soft Bremsstrahlung; the IR-divergent soft-photon factorisation derived directly from the external-field amplitude.]
• Schwartz, M. D., Quantum Field Theory and the Standard Model (Cambridge, 2014), §20.4. [Modern Feynman-rule derivation with dimensional regularisation of the IR divergence.]
• Rossi, B. B., High-Energy Particles (Prentice-Hall, 1952), Ch. 2. [Radiation-length formalism, electromagnetic-shower theory, the $7/9$ universality theorem.]
• Greiner, W. & Reinhardt, J., Quantum Electrodynamics, 4th ed. (Springer, 2009), §3.6. [Detailed worked-example treatment of bremsstrahlung and pair production.]
• Mott, N. F. & Massey, H. S. W., The Theory of Atomic Collisions, 3rd ed. (Oxford UP, 1965). [Atomic-physics context for the Thomas-Fermi screening and the Mott correction.]

Experimental and review:

• Particle Data Group, "Review of Particle Physics", Prog. Theor. Exp. Phys. 2022, 083C01, §34 (Passage of Particles Through Matter). [The canonical radiation-length values, electromagnetic-shower parametrisations, and Bethe-Heitler pair-production cross sections.]
• Longo, E. & Sestili, I., "Monte Carlo calculation of photon-initiated electromagnetic showers in lead glass", Nucl. Instrum. Meth. 128 (1975), 283–307. [The standard longitudinal-shower-profile parametrisation $f(t)=(bt{)}^{a-1}{e}^{-bt}/\Gamma (a)$ used throughout calorimeter design.]
• Agostinelli, S. et al. (GEANT4 Collaboration), "Geant4 — a simulation toolkit", Nucl. Instrum. Meth. A 506 (2003), 250–303. [The dominant Monte Carlo for charged-particle and photon transport in high-energy physics, medical physics, and space-radiation applications; uses Bethe-Heitler as the bremsstrahlung-and-pair-production engine.]
• Ford, R. L. & Nelson, W. R., "The EGS Code System (Version 3)", SLAC Report SLAC-210 (1978). [The original Electron Gamma Shower Monte Carlo, the ancestor of all modern electromagnetic-cascade codes.]
• Berger, M. J. & Seltzer, S. M., "Bremsstrahlung and Photoneutrons from Thick Tungsten and Tantalum Targets", Phys. Rev. C 2 (1970), 621–631. [NBS / NIST tabulations of medical-linac bremsstrahlung yields; the reference data for clinical-linac design.]