Ionisation energy loss (Bethe-Bloch)

Every charged particle that crosses a detector announces itself the same way: it strips electrons off the atoms in its path and leaves a thin trail of ionisation behind. The rate at which it bleeds energy doing so — the stopping power $-dE/dx$ — is one of the most useful numbers in all of experimental particle physics, because it is at once nearly universal and finely diagnostic. Universal, because once you divide out the material's density the answer depends only on the ratio $Z/A$ , which is close to $1/2$ for almost everything; a minimum-ionising muon deposits about the same $1$ – $2$ MeV per gram-per-square-centimetre in helium, carbon, iron, or lead. Diagnostic, because the shape of the curve — a steep $1/\beta^2$ fall, a broad minimum near $\beta\gamma\approx 3$ , and a gentle logarithmic relativistic rise — depends on the particle's velocity but not its mass, so measuring $dE/dx$ and momentum together tells you what the particle is. The formula that captures all of this is the Bethe–Bloch equation. Its heart is a piece of nineteenth-century mechanics — a fast charge kicks each electron it passes — dressed in relativistic kinematics and quantum cutoffs. The $1/\beta^2$ prefactor is why slow particles dump their energy in a sharp Bragg peak at the end of their range, the physics that lets a proton beam kill a tumour while sparing the tissue in front of it. The logarithmic rise is why a $10$ GeV muon still punches through several metres of iron. This article rebuilds the equation from the collision up, verifies every factor, and shows where each term earns its keep in a modern experiment.

Key takeaways

A charged particle traversing matter loses energy by ionising atomic electrons at a mean rate given by the Bethe–Bloch equation, $-dE/dx = Kz^2(Z/A)\beta^{-2}[\tfrac12\ln(\cdots)-\beta^2-\delta/2]$ .
The dominant feature is the $1/\beta^2$ prefactor: slow particles lose energy fastest, producing the Bragg peak at the end of a range.
The rate bottoms out near $\beta\gamma\approx 3$ at $1$ – $2$ MeV cm² g⁻¹ — the minimum-ionising particle, the standard yardstick of a detector's response.
Above the minimum, $dE/dx$ climbs logarithmically with $(\beta\gamma)^2$ (the relativistic rise) until the medium's polarisation caps it at the Fermi plateau.
Normalised to density the rate scales only as $Z/A\approx\tfrac12$ , so ionisation loss in MeV cm² g⁻¹ is nearly the same in every material — it is set by the particle's velocity, not its energy or mass.
Key idea: a fast charge sheds energy by knocking electrons out of atoms at a rate that falls as $1/\beta^2$ , bottoms out near $\beta\gamma\approx 3$ , then rises slowly with $\ln(\beta\gamma)^2$ — and once divided by density that rate barely depends on what the material is.

The core insight

A fast charged particle bleeds energy by knocking electrons out of atoms at a rate that falls off as $1/\beta^2$ , bottoms out near $\beta\gamma\approx 3$ , then rises slowly with the logarithm of $(\beta\gamma)^2$ — and once normalised to density this rate barely depends on what the material is.

Where this lives today

The Bethe–Bloch curve is not a historical relic; it is read out millions of times a second at every collider. Silicon and gaseous tracking detectors are calibrated in units of the minimum-ionising particle, and the same $dE/dx$ measurement doubles as particle identification: the ALICE and STAR time-projection chambers separate pions, kaons, and protons by binning each track's average ionisation against its momentum, exactly as in worked example 2. The relativistic rise and Fermi plateau are measured, not just assumed — they set the ceiling on how much identification power survives at high momentum. Muon spectrometers in ATLAS and CMS exploit the flip side: because muons only ionise while electrons and hadrons shower, the particle that penetrates metres of iron or the calorimeter steel is tagged a muon on that basis alone. And the steep low-velocity $1/\beta^2$ divergence — the Bragg peak — is the entire premise of hadron therapy: a proton or carbon-ion beam tuned to stop inside a tumour deposits the bulk of its dose in the final millimetres of range, sparing the healthy tissue upstream, a precision impossible with photons. Radiation dosimetry, cosmic-ray detectors, and the LET (linear energy transfer) used in radiobiology are all the same equation wearing different clothes.

The physical picture

Zoom into the track and the machinery is almost embarrassingly simple. The projectile is a point charge $ze$ moving at velocity $v=\beta c$ . As it flies past an atomic electron sitting at perpendicular distance $b$ — the impact parameter — its Coulomb field yanks the electron sideways. The pull lasts only while the particle is within a distance $\sim b$ , a time $\sim b/v$ , so the sideways impulse the electron receives is stronger the closer the pass and the slower the particle. That single collision, summed over the billions of electrons the particle sweeps past per centimetre, is the whole of ionisation energy loss. The first figure freezes one such encounter and names the quantities that control it.

Figure

A fast charge ze sweeps past atomic electrons: at impact parameter b it delivers a transverse impulse Δp⊥ ≈ 2ze²/(4πε₀ bv), so each electron absorbs ΔE ≈ Δp²/2mₑ ∝ z²/(β²b²) — summed over every electron in the path this is the 1/β² heart of Bethe-Bloch

Because the energy handed to one electron goes as $1/(\beta^2 b^2)$ , the total loss is an integral over impact parameter — and $\int db/b$ is a logarithm. That logarithm is bounded at both ends: distant, gentle passes ( $b$ large) stop contributing once they are slower than the electron's own orbital motion (the binding energy, encoded in the mean excitation energy $I$ ); violent close passes ( $b$ small) are capped by the maximum energy a single collision can transfer, $T_{\max}$ . Everything quantitative in the Bethe–Bloch formula is the value of that bounded logarithm times the $1/\beta^2$ prefactor. Plot the result against velocity and you get the single most recognisable curve in detector physics — steep fall, broad minimum, slow rise — shown next with real numbers for copper.

Figure

Mean ionisation loss for a heavy singly-charged particle in copper (Z/A = 0.456, I = 322 eV): the 1/β² fall bottoms at the minimum-ionising point −dE/dx ≈ 1.4 MeV cm² g⁻¹ near βγ ≈ 3, then the relativistic rise climbs as ln(βγ)² before the density effect flattens it into the Fermi plateau

The core equations

The mean rate of ionisation energy loss for a heavy charged particle is the Bethe–Bloch equation. It is written per unit mass thickness (grams per square centimetre) rather than per centimetre, because in that form the trivial dependence on how tightly packed the atoms are has been divided out and what remains is nearly material-independent. Every symbol below is either a property of the medium ( $Z,A,I,\delta$ ) or of the projectile's velocity ( $\beta,\gamma$ ) — strikingly, the projectile's mass does not appear except through the weak dependence buried in $T_{\max}$ .

\left\langle-\frac{dE}{dx}\right\rangle = K\,z^2\,\frac{Z}{A}\,\frac{1}{\beta^2}\left[\frac{1}{2}\ln\frac{2 m_e c^2\,\beta^2\gamma^2\,T_{\max}}{I^2}-\beta^2-\frac{\delta(\beta\gamma)}{2}\right] \qquad (1)

where: $\langle-dE/dx\rangle$ — mean mass stopping power, MeV cm² g⁻¹; $z$ — charge of the projectile in units of $e$ ; $Z,A$ — atomic number and mass of the medium; $\beta\gamma$ — projectile velocity factor; $m_ec^2=0.511$ MeV; $I$ — mean excitation energy of the medium; $T_{\max}$ — maximum single-collision energy transfer; $\delta(\beta\gamma)$ — density-effect correction; $K=0.307$ MeV mol⁻¹ cm². Reading it: a $1/\beta^2$ prefactor times a slowly growing logarithm, minus two corrections that matter only at the extremes. The logarithm's ceiling is $T_{\max}$ , the most energy the projectile can dump on one free electron in a head-on collision. Relativistic two-body kinematics fixes it, and for any projectile much heavier than the electron it reduces to a clean form that grows as $(\beta\gamma)^2$ .

T_{\max}=\frac{2 m_e c^2\,\beta^2\gamma^2}{1+2\gamma\,m_e/M+(m_e/M)^2}\;\xrightarrow{\;M\gg m_e\;}\;2 m_e c^2\,\beta^2\gamma^2 \qquad (2)

where: $M$ — projectile mass; the denominator is the only place $M$ enters the whole calculation, and it is negligible for protons, pions, and muons below the TeV scale. A $T_{\max}$ large enough to eject a visibly tracked electron produces a δ-ray. The constant $K$ collects the fundamental couplings, and writing the electron density as $n_e=N_A\,\rho\,Z/A$ exposes the origin of the near material-independence: divide by density and only the ratio $Z/A$ survives, and that ratio sits close to one-half for every nucleus with roughly equal protons and neutrons.

K=4\pi N_A r_e^2 m_e c^2=0.307\ \mathrm{MeV\,mol^{-1}\,cm^2},\qquad \frac{1}{\rho}\left(-\frac{dE}{dx}\right)\propto \frac{Z}{A}\approx\frac{1}{2} \qquad (3)

where: $N_A$ — Avogadro's number; $r_e=2.818$ fm — classical electron radius; $\rho$ — mass density. Hydrogen ( $Z/A=1$ ) loses about twice as much per g cm⁻² as everything else; heavy elements dip toward $Z/A\approx 0.4$ , which is why the copper and lead curves sit slightly below the light ones. Finally, two anchor facts pin the curve numerically. The mean excitation energy scales roughly linearly with $Z$ , and the whole family of curves shares a minimum at nearly the same velocity and the same height — the minimum-ionising particle, the working unit of detector calibration.

I\approx 10\,Z\ \mathrm{eV}\quad(Z\gtrsim 20),\qquad \left\langle-\frac{dE}{dx}\right\rangle_{\min}\approx 1\text{–}2\ \mathrm{MeV\,cm^2\,g^{-1}}\ \text{ at }\ \beta\gamma\approx 3\text{–}3.5 \qquad (4)

where: $I$ is an effective ionisation potential averaged over all atomic shells (water $78$ eV, silicon $173$ eV, copper $322$ eV); the minimum height creeps down slowly with $Z$ (water $\approx 2.0$ , copper $\approx 1.4$ MeV cm² g⁻¹). A "MIP" is any particle sitting in this broad minimum.

Derivation AThe classical impulse picture and the origin of the 1 over beta-squared law

The goal of this derivation is to show that the entire structure of the Bethe–Bloch formula — the $1/\beta^2$ prefactor multiplying a logarithm — falls straight out of one classical collision followed by an integral over impact parameter, before any quantum mechanics is invoked. We compute the sideways impulse a passing charge gives a single electron, turn it into an energy transfer, sum over all the electrons in a cylindrical shell of the material, and watch the integral $\int db/b$ generate the logarithm whose two cutoffs are later supplied by quantum kinematics. This is Bohr's 1913 argument, and it gets every scaling right.

Step 1. Send the projectile $ze$ past a stationary electron at impact parameter $b$ with velocity $v$ , and integrate the transverse Coulomb force over the flyby. The longitudinal kick cancels by symmetry (the field pushes forward on approach and back on recession), leaving a net transverse impulse $\Delta p_\perp=\int F_\perp\,dt = 2ze^2/(4\pi\varepsilon_0\,b\,v)$ . Why: the interaction lasts a time $\sim b/v$ during which the transverse field is $\sim ze/(4\pi\varepsilon_0 b^2)$ , and their product carries the $1/(bv)$ that controls everything.

Step 2. The electron, treated as free during the fast collision, absorbs kinetic energy $\Delta E(b)=\Delta p_\perp^2/2m_e = 2z^2e^4/[(4\pi\varepsilon_0)^2 m_e v^2 b^2]$ . Why: this single line already contains the two headline dependences — energy loss $\propto z^2$ (charge squared) and $\propto 1/v^2$ (the slow-particle enhancement), and the $1/b^2$ that will be integrated next.

Step 3. Count the electrons. In thickness $dx$ the projectile passes $n_e\,(2\pi b\,db)\,dx$ electrons in the shell between $b$ and $b+db$ , where $n_e$ is the electron number density. Multiply by $\Delta E(b)$ and integrate over $b$ : $-\dfrac{dE}{dx}=\dfrac{4\pi z^2 e^4 n_e}{(4\pi\varepsilon_0)^2 m_e v^2}\displaystyle\int_{b_{\min}}^{b_{\max}}\frac{db}{b}=\dfrac{4\pi z^2 e^4 n_e}{(4\pi\varepsilon_0)^2 m_e v^2}\,\ln\frac{b_{\max}}{b_{\min}}$ . Why: the $2\pi b\,db$ geometry exactly cancels one power of $b$ , so the remaining $\int db/b$ is a logarithm — the log in Bethe–Bloch is a phase-space count over impact parameters, nothing more.

Step 4. Fix the cutoffs physically. The closest useful approach $b_{\min}$ is set by the largest possible momentum transfer, i.e. by $T_{\max}$ ; the farthest $b_{\max}$ is set by the fact that a pass slower than the electron's orbital frequency $\sim I/\hbar$ is adiabatic and transfers nothing, giving $b_{\max}\sim \gamma v\hbar/I$ . Substituting turns $\ln(b_{\max}/b_{\min})$ into $\tfrac12\ln(2m_ec^2\beta^2\gamma^2 T_{\max}/I^2)$ once the collisions are treated quantum-mechanically in the Born approximation. Why: Bethe's 1930 quantum calculation replaces Bohr's ad-hoc cutoffs with the mean excitation energy $I$ and the kinematic ceiling $T_{\max}$ , promoting the classical estimate to the exact leading-order result.

Go deeper · Derivation B · 1 minThe kinematic ceiling — maximum energy transfer to an electron

Derivation BThe kinematic ceiling — maximum energy transfer to an electron

The aim here is to derive $T_{\max}$ , the most kinetic energy a heavy projectile of mass $M$ and velocity $\beta\gamma$ can give a single free electron, because it is $T_{\max}$ that caps the Bethe–Bloch logarithm and that sets the energy of the δ-rays which blur a track. We treat the encounter as a relativistic elastic collision, work in the frame where the electron is initially at rest, and use invariance of the momentum transfer to get the closed form, then take the heavy-projectile limit that applies to every particle-physics projectile up to the TeV scale.

Step 1. Model the hard collision as elastic scattering of the projectile off a free electron at rest — legitimate because the transferred energy $T_{\max}$ (MeV scale) dwarfs the eV-scale binding, so the electron may be treated as free and stationary. Why: hard, close collisions are exactly the ones that set the upper cutoff, and for those the atomic binding is a negligible perturbation.

Step 2. Maximum energy transfer occurs in a head-on collision, where the electron recoils straight forward. Conservation of relativistic energy and momentum for $M$ striking $m_e$ gives $T_{\max}=\dfrac{2 m_e c^2\,\beta^2\gamma^2}{1+2\gamma m_e/M+(m_e/M)^2}$ . Why: the same two-body algebra that gives the non-relativistic $4Mm/(M+m)^2$ fractional transfer generalises, with $\beta^2\gamma^2$ carrying the relativistic growth of the available momentum.

Step 3. Take $M\gg m_e$ . Both correction terms in the denominator vanish and $T_{\max}\to 2 m_e c^2\,\beta^2\gamma^2$ . Numerically, a minimum-ionising particle at $\beta\gamma\approx 3.5$ can transfer up to $2(0.511)(12.25)\approx 12.5$ MeV to one electron. Why: this shows δ-rays can carry many MeV — enough to leave their own visible tracks — while the denominator's $2\gamma m_e/M$ term only bites for very light or ultra-relativistic projectiles, the sole place the projectile mass enters the physics.

Step 4. Feed $T_{\max}\propto\beta^2\gamma^2$ back into equation (1). The log's argument then carries a full $\beta^4\gamma^4$ from the product $\beta^2\gamma^2\cdot T_{\max}$ , so $\tfrac12\ln(\cdots)$ grows as $\ln(\beta\gamma)^2$ at high energy. Why: this is precisely the relativistic rise — two powers of $\beta\gamma$ from the transverse field extension and two from $T_{\max}$ — and it is what the density effect must later tame.

Go deeper · Derivation C · 2 minFrom stopping power to material independence and the minimum

Derivation CFrom stopping power to material independence and the minimum

The purpose of this derivation is to explain the two experimental facts that make ionisation loss so useful: that the mass stopping power is nearly the same in every material, and that every material's curve has a minimum at almost the same velocity. We substitute the electron density in terms of $Z/A$ , divide out the density explicitly, then analyse the competition inside equation (1) between the falling $1/\beta^2$ prefactor and the rising logarithm to locate the minimum and characterise the rise and its density-limited plateau.

Step 1. Write the electron number density as $n_e=N_A\,\rho\,Z/A$ and insert it into the stopping power. The density $\rho$ factors out cleanly, leaving the mass stopping power $\tfrac{1}{\rho}(-dE/dx)=K\,(Z/A)\,\beta^{-2}[\cdots]$ with no reference to $\rho$ at all. Why: this is why detector physicists always quote MeV cm² g⁻¹ — it strips away the one hugely material-dependent factor, packing density, and isolates the near-universal remainder.

Step 2. Note that $Z/A$ is $0.5$ for light nuclei with equal proton and neutron counts, falling only to $\approx 0.40$ for lead as the neutron excess grows. Hence helium, carbon, iron, and lead curves overlap to within tens of percent, all normalised the same way, with hydrogen the lone outlier at $Z/A=1$ . Why: the near-constancy of $Z/A$ across the periodic table is the entire reason ionisation loss is "universal"; the residual $Z$ -dependence hides only inside $I$ and $\delta$ .

Step 3. Locate the minimum. At low $\beta\gamma$ the $1/\beta^2$ prefactor plunges while the logarithm still grows, so $-dE/dx$ falls steeply; at high $\beta\gamma$ the prefactor saturates ( $\beta\to1$ ) while the log keeps climbing, so it rises. Setting the derivative to zero puts the minimum near $\beta\gamma\approx 3$ – $3.5$ , where the falling prefactor and rising log balance. Why: the minimum is not a special piece of physics but the turning point of a product of one decreasing and one increasing function, which is why its location barely shifts between materials.

Step 4. Cap the rise. A naïve reading of equation (2) would let $-dE/dx$ grow forever as $\ln(\beta\gamma)^2$ , but the projectile's own field polarises the medium, screening the long-range collisions that supply the highest $b$ contributions. Fermi's density correction $\delta\to 2\ln(\beta\gamma)+C$ at high energy subtracts exactly the logarithmic growth of the distant-collision term, flattening the curve into the Fermi plateau. Why: without the density effect a cosmic-ray muon would appear to lose runaway energy; with it, the plateau sits only $\sim 10$ – $50\%$ above the minimum, and $dE/dx$ becomes an almost velocity-independent tag at high energy.

Limiting cases

Case	Result	Meaning
Non-relativistic, $\beta\ll 1$	$-dE/dx\propto 1/\beta^2$	slow particles lose energy fastest; loss diverges as the particle stops → the Bragg peak
Minimum ionising, $\beta\gamma\approx 3$	$-dE/dx\approx 1$ – $2$ MeV cm² g⁻¹	the near-universal MIP yardstick; height creeps down with $Z$
Relativistic rise, $\beta\gamma\gtrsim 4$	$-dE/dx\propto\tfrac12\ln(\beta\gamma)^2$	slow logarithmic climb above the minimum; useful for particle ID
Ultra-relativistic in a dense medium	$\delta\to 2\ln(\beta\gamma)+C$ → Fermi plateau	polarisation screens distant collisions; the rise flattens off
Very low velocity, $v\sim v_\text{orbital}$	Bethe–Bloch fails; charge-exchange, Barkas & Bloch terms	the projectile picks up and sheds electrons; $I$ -based logarithm breaks down
Electrons/positrons as projectiles	modified formula; radiative loss dominates above a few tens of MeV	identical-particle kinematics and bremsstrahlung change the physics entirely

Common trapThinking dE/dx tells you the particle's energy, or that it depends strongly on the material

Two linked misconceptions. First, ionisation loss is a function of the particle's velocity $\beta\gamma$ , not of its energy or its mass — a $1$ GeV proton and a $150$ MeV pion have very different energies but if they share the same $\beta\gamma$ they lose energy at nearly the same rate. This is precisely what makes $dE/dx$ a particle-ID tool: measure the loss and the momentum $p=\beta\gamma Mc$ , and the combination fixes $M$ , revealing the identity. A single $dE/dx$ number is ambiguous on its own. Second, the loss is not strongly material-dependent once you normalise correctly. In MeV per centimetre it obviously tracks density (lead stops particles in far less length than air), but that is the boring geometric factor. Divide by density and the residue scales only as $Z/A\approx\tfrac12$ , so the intrinsic stopping power in MeV cm² g⁻¹ is nearly the same for helium and for lead. Quoting "iron stops particles better than carbon" without saying "per centimetre" versus "per g cm⁻²" is where the confusion always starts.

Worked examples

1 · A minimum-ionising particle in a silicon tracker (fully worked)

A relativistic muon ( $\beta\gamma\approx 3.5$ , sitting in the minimum) crosses a $300\ \mu\mathrm{m}$ silicon strip detector. Estimate the mean energy it deposits, and the resulting number of electron–hole pairs.

Step 1. Get the mass stopping power at the minimum. For silicon ( $Z=14$ , $A=28.09$ , $Z/A=0.4985$ , $I=173$ eV, $\rho=2.33$ g cm⁻³) the Bethe–Bloch minimum evaluates to $\langle-dE/dx\rangle_{\min}\approx 1.66$ MeV cm² g⁻¹ (equation 1 with $\beta^2=0.925$ and a small density term; this is the tabulated PDG value). Mass stopping power $\approx 1.66$ MeV cm² g⁻¹.

Step 2. Convert to loss per unit length by multiplying by density: $-dE/dx=1.66\times 2.33=3.87$ MeV cm⁻¹. Over the strip thickness $t=300\ \mu\mathrm{m}=0.0300$ cm, the mean deposited energy is $\Delta E=3.87\times 0.0300=0.116$ MeV. Mean deposit $\approx 116$ keV.

Step 3. Fluctuations matter here. Thin-absorber energy loss follows the highly asymmetric Landau distribution, whose long high-energy tail (rare hard δ-rays) drags the mean well above the most probable value; for $300\ \mu\mathrm{m}$ of silicon the most-probable deposit is only $\approx 80$ keV, and this is what a real detector reads out per hit. This is why detector notes distinguish "mean" from "most probable" loss.

Step 4. Turn energy into signal. Silicon takes $3.6$ eV to make one electron–hole pair, so the most-probable $80$ keV liberates $N\approx 8.0\times 10^4/3.6\approx 2.2\times 10^4$ pairs. A MIP makes $\sim 22\,000$ e–h pairs in $300\ \mu\mathrm{m}$ of silicon — a few femtocoulombs, comfortably above the noise of a modern front-end amplifier, which is exactly why $300\ \mu\mathrm{m}$ is the standard sensor thickness.

2 · Particle identification from dE/dx (pion versus proton at 1 GeV/c)

A time-projection chamber filled with argon ( $Z/A=0.451$ , $I=188$ eV) records two tracks, each of momentum $p=1$ GeV/c. One is a pion ( $M=0.140$ GeV), one a proton ( $M=0.938$ GeV). Show that their ionisation rates differ enough to tell them apart.

Step 1. Convert momentum to velocity. Since $p=\beta\gamma Mc$ , the pion has $\beta\gamma=1.000/0.140=7.16$ and the proton $\beta\gamma=1.000/0.938=1.07$ . Same momentum, very different velocity.

Step 2. The pion, at $\beta\gamma\approx 7$ , sits on the relativistic rise: equation (1) gives $\approx 1.6$ MeV cm² g⁻¹. The proton, at $\beta\gamma\approx 1$ , sits on the steep $1/\beta^2$ shoulder well above the minimum: $\approx 2.1$ MeV cm² g⁻¹. The proton ionises about $35\%$ more heavily than the pion at the same momentum.

Step 3. That $35\%$ gap is large compared with the few-percent resolution a TPC achieves by averaging hundreds of samples along each track, so the two bands separate cleanly on a $dE/dx$ -versus- $p$ plot. This is the everyday particle-ID method of ALICE, STAR, and every collider TPC: bands for $e,\mu,\pi,K,p$ fan out below $\sim 1$ GeV/c and merge only where the curves cross.

3 · Why a 10 GeV muon punches through iron

Estimate how far a $10$ GeV muon travels in iron before ionisation stops it, and contrast it with an electron of the same energy.

Step 1. A muon of energy $10$ GeV has $\beta\gamma=p/Mc\approx 10\,000/105.7\approx 95$ , high on the relativistic rise but still ionisation-dominated (iron's muon critical energy is $\sim 350$ GeV). At this $\beta\gamma$ the stopping power is $\approx 1.6$ MeV cm² g⁻¹; with $\rho=7.87$ g cm⁻³ that is $-dE/dx\approx 12.6$ MeV cm⁻¹. About $13$ MeV per centimetre of iron.

Step 2. A crude range estimate divides energy by loss rate, $R\sim E/(dE/dx)$ , but $dE/dx$ climbs as the muon slows, so the true continuous-slowing-down range is shorter; the tabulated CSDA range of a $10$ GeV muon in iron is several metres (order $5$ – $6$ m). A muon is the most penetrating charged particle in a detector. Muons cross the whole calorimeter and reach the outer muon chambers.

Step 3. Contrast an electron. At $10$ GeV an electron loses energy overwhelmingly by bremsstrahlung, not ionisation, with a radiation length in iron of only $1.76$ cm; it showers away almost all its energy in $\sim 20$ cm. This mass dependence — light electrons radiate, heavy muons only ionise — is the physical basis of muon identification: the particle that survives the iron is the muon.

Physics of the portrait

Ionisation energy loss is a rare corner of physics where a single formula, refined by four people across three decades, still carries all their names in its terms — Bohr's classical skeleton, Bethe's quantum flesh, Bloch's stitching, and Fermi's density skin.

Origins · 1 minHans Bethe and how this was discovered

Hans Bethe

1906–2005 · the quantum theory of stopping power and the mean excitation energy

The stopping-power problem was opened by Niels Bohr in 1913, who used the classical impulse picture — the very argument reconstructed in this article — to get the $1/v^2$ law and the logarithm, with cutoffs he had to guess. The decisive step was Hans Bethe's in 1930: applying the brand-new quantum mechanics in the Born approximation, he replaced Bohr's guessed limits with a rigorously defined mean excitation energy $I$ , an average over all the atomic transitions weighted by their oscillator strengths, and obtained the logarithmic term that still bears his name. Felix Bloch in 1933 showed how Bohr's classical and Bethe's quantum results are two limits of one formula and supplied the interpolating correction between them. Enrico Fermi completed the picture in 1940 by accounting for the polarisation of the medium by the passing particle's own field — the density effect that flattens the relativistic rise into a plateau. Bethe himself is better remembered for other things: the theory of how stars shine (the CNO and proton–proton cycles, Nobel Prize 1967), the Bethe ansatz in many-body physics, and his wartime leadership of the theory division at Los Alamos. But the equation a detector physicist reaches for first is his, and the $I$ tabulated for every element in the data books is the quantity he defined.

History

1913

— Bohr derives the classical stopping-power formula from the impulse picture

1930

— Bethe gives the quantum stopping power and defines the mean excitation energy I

1933

— Bloch interpolates between the Bohr and Bethe regimes with the Bloch correction

1940

— Fermi accounts for the medium's polarisation, predicting the density-effect plateau

1944

— Landau derives the asymmetric distribution of energy-loss fluctuations in thin absorbers

Going further

Texts. Thomson, Modern Particle Physics, §1.2.1 (interactions and detection of charged particles), Eqs. (1.1)–(1.2) · Particle Data Group review, Passage of Particles Through Matter (the definitive tabulation of $K$ , $I$ , $\delta$ , and the curves) · Leo, Techniques for Nuclear and Particle Physics Experiments, Ch. 2 · Grupen & Shwartz, Particle Detectors, 2nd ed., Ch. 1 · Jackson, Classical Electrodynamics, §13.1–13.3 (the classical and Bethe derivations). Key numerical anchors: the classical energy transfer per collision is $\Delta E(b)\propto z^2/(\beta^2 b^2)$ , and integrating over impact parameter yields the $\int db/b$ logarithm at the heart of the formula. In the heavy-projectile limit, $T_{\max}\to 2m_ec^2\beta^2\gamma^2$ . The overall constant evaluates to $K=4\pi N_A r_e^2 m_ec^2=0.307$ MeV mol⁻¹ cm². Minimum stopping powers are: copper $\approx 1.4$ , water $\approx 2.0$ , silicon $\approx 1.66$ MeV cm² g⁻¹; the pion/proton $dE/dx$ ratio at $1$ GeV/c in argon is $\approx 1.35$ . Mean excitation energies are standard tabulated values: $I_\text{water}=78$ , $I_\text{Si}=173$ , $I_\text{Cu}=322$ eV.

Prerequisites. Special-relativistic kinematics · Natural units · Interactions of particles with matter. Leads to. Tracking detectors · Bremsstrahlung radiation · Čerenkov radiation.

Ask the tutor: Ask me to walk through the Bethe–Bloch curve for any particle and material — muon in iron, proton in water, or your own detector — and we will locate its minimum, its rise, and the signal it leaves.

Why does energy loss go as 1/β² — what is physically slower about a slow particle?

It comes entirely from how long the electron feels the passing charge. A slow projectile lingers near each electron for a longer time $\sim b/v$ , so the transverse Coulomb impulse it delivers is larger, $\Delta p_\perp\propto 1/v$ . The energy transfer is that impulse squared over twice the electron mass, so $\Delta E\propto 1/v^2$ . There is no quantum mechanics in this — it is the same reason a car rolling slowly past a magnet feels its pull for longer. The consequence is dramatic near the end of a track: as the particle slows almost to a stop, $1/\beta^2$ blows up and it dumps a huge spike of energy in the final fraction of a millimetre, the Bragg peak that hadron-therapy beams are aimed to place inside a tumour.

What exactly is a minimum-ionising particle, and why is its loss so nearly universal?

A minimum-ionising particle, or MIP, is any charged particle whose velocity puts it in the broad minimum of the Bethe–Bloch curve, around $\beta\gamma\approx 3$ – $4$ . There the falling $1/\beta^2$ prefactor and the rising logarithm balance, so $-dE/dx$ is at its lowest, about $1$ – $2$ MeV cm² g⁻¹. It is nearly universal because the mass stopping power scales only as $Z/A$ , which is close to $1/2$ for every nucleus with roughly equal protons and neutrons, and because the minimum's position depends on velocity, not on the material. Detector builders love the MIP precisely because it is a fixed, material-independent reference: "this sensor gives $22\,000$ electrons for a MIP" is a calibration you can carry from one experiment to another.

Does the relativistic rise go on forever?

No — the medium stops it. Left to the bare formula, the logarithm would keep climbing as $\ln(\beta\gamma)^2$ without bound, because an ever-faster particle's transverse electric field reaches ever-farther electrons. But those distant electrons are shielded: the particle's own field polarises the intervening atoms, screening the long-range collisions that supply the growth. Fermi's density-effect term $\delta(\beta\gamma)$ subtracts exactly the runaway part, so above a few hundred in $\beta\gamma$ the curve flattens into the Fermi plateau, sitting only $10$ – $50\%$ above the minimum (higher in gases, lower in dense solids). This is why $dE/dx$ becomes a nearly velocity-independent tag for very fast particles, and why the plateau height is itself a measured material property.

What are δ-rays, and why is the "most probable" energy loss less than the mean?

A δ-ray is a knock-on electron given so much energy in a single hard collision — up to $T_{\max}$ , which can be many MeV — that it flies off and makes its own short track. Because these hard collisions are rare but carry large energy, the distribution of energy lost in a thin absorber is strongly asymmetric: most crossings lose a modest, typical amount, but a long tail of occasional big δ-ray losses pulls the mean well above the peak. This is the Landau distribution. A real detector reading out one hit sees a value near the peak (the most probable loss), which is why detector documents quote the most-probable value, often $20$ – $30\%$ below the Bethe–Bloch mean, rather than the mean itself.

Why doesn't the answer depend on the projectile's mass?

Because ionisation loss is governed by the projectile's electric field and velocity, not its inertia. The impulse it gives each electron depends on its charge $ze$ and how long it lingers ( $\propto 1/v$ ), and neither knows the projectile's mass. Mass enters only through the tiny denominator of $T_{\max}$ , which matters only for very light or ultra-relativistic projectiles. So a proton and a muon at the same $\beta\gamma$ ionise almost identically — but at the same momentum they differ, because a heavier particle at fixed momentum is slower (smaller $\beta\gamma$ ) and therefore ionises more. That momentum-versus-velocity distinction is exactly what makes combined $dE/dx$ and momentum measurements identify the particle.

When does Bethe–Bloch stop being valid?

At both ends of the velocity range. At very low velocity — when the projectile moves no faster than the atomic electrons it is trying to ionise — the assumption of a fast, impulsive collision fails; the projectile starts capturing and losing electrons, its effective charge changes, and the Barkas and Bloch corrections (and eventually nuclear stopping) take over. At very high energy for light projectiles, radiative losses (bremsstrahlung) overtake ionisation entirely: for electrons this happens at a few tens of MeV, for muons only above a few hundred GeV. Bethe–Bloch is the correct description across the wide middle band — roughly $0.1\lesssim\beta\gamma\lesssim 1000$ for heavy particles — which happens to be exactly the regime of most particles crossing a collider detector.