Ultra-High-Energy Cosmic Rays and the GZK Cutoff

Cosmic rays span more than eleven decades in energy, and their differential flux falls as a steep power law — roughly $\mathrm{d}N/\mathrm{d}E\propto E^{-2.7}$ , steepening to $E^{-3.1}$ above the "knee" near $3\times10^{15}\,\mathrm{eV}$ and flattening again at the "ankle" near $5\times10^{18}\,\mathrm{eV}$ . The particles above the ankle — the ultra-high-energy cosmic rays (UHECRs) — are almost certainly extragalactic, because their Larmor radii in Galactic magnetic fields exceed the size of the Milky Way and no known Galactic engine can confine and accelerate them. At $10^{20}\,\mathrm{eV}$ the flux is a mere one particle per square kilometre per century, so they are seen only through the giant extensive air showers they seed in the atmosphere. This article is a synthesis of high-energy astrophysics and special relativity. In 1966, just after the cosmic microwave background was discovered, Kenneth Greisen and — independently — Georgiy Zatsepin and Vadim Kuzmin realised that the CMB is an opaque medium for the most energetic protons. Above a threshold energy a proton streaming through the $2.725\,\mathrm{K}$ photon bath resonantly makes pions, $p+\gamma_{\mathrm{CMB}}\to\Delta^{+}(1232)\to p+\pi^{0}$ or $n+\pi^{+}$ , losing about a fifth of its energy each time over a mean free path of only a few megaparsecs. The cumulative effect is a horizon: protons injected above $\sim10^{20}\,\mathrm{eV}$ cannot reach us from beyond $\sim100\,\mathrm{Mpc}$ . This is the Greisen–Zatsepin–Kuzmin (GZK) cutoff. The chapter takes the threshold kinematics from Special relativity kinematics, the photon-bath number density from aq-astro-cosmic-microwave-background, and the source-confinement bound from the Hillas criterion, and asks what the resulting suppression of the highest-energy sky actually tells us about where these particles are born.

Key takeaways

Ultra-high-energy cosmic rays are atomic nuclei detected above $10^{18}\,\mathrm{eV}$ and past $10^{20}\,\mathrm{eV}$ — a macroscopic $\sim16\,\mathrm{J}$ of kinetic energy carried by one subatomic particle.
Above the GZK threshold $E_{\mathrm{th}}\approx5\times10^{19}\,\mathrm{eV}$ , protons make pions off CMB photons via the $\Delta(1232)$ resonance, losing energy over a horizon of only $\sim100\,\mathrm{Mpc}$ .
The threshold is pure special relativity: a $\sim10^{-3}\,\mathrm{eV}$ CMB photon, blueshifted into the proton rest frame above the $\Delta$ mass, produces a pion and steals a large fraction of the proton's energy.
The Pierre Auger Observatory and Telescope Array measure a spectral suppression near $4\times10^{19}\,\mathrm{eV}$ consistent with GZK, though whether it is pure attenuation or a source-injection cutoff is still debated.
The GZK horizon confines sources to the local universe; combined with the Hillas rigidity bound this points to nearby AGN, radio-galaxy lobes, and starburst galaxies.
Key idea: The GZK cutoff is relativistic kinematics applied to a cosmological photon bath — a proton fast enough to see a CMB photon blueshifted above the $\Delta(1232)$ threshold loses energy to pion production, capping the energy of protons that traverse cosmological distances at $\sim5\times10^{19}\,\mathrm{eV}$ .

The core insight

The GZK cutoff is relativistic kinematics applied to a cosmological photon bath: a proton fast enough to see a CMB photon (energy $\sim10^{-3}\,\mathrm{eV}$ ) blueshifted above the $\Delta(1232)$ resonance threshold loses energy to pion production, capping the energy of protons that can traverse cosmological distances at $\sim5\times10^{19}\,\mathrm{eV}$ .

Where this lives today

UHECR physics is now a precision, multi-messenger enterprise. The Pierre Auger Observatory in Argentina and the Telescope Array in Utah between them cover thousands of km² with hybrid detectors — surface arrays of water-Cherenkov tanks or scintillators that sample the shower footprint, cross-calibrated by fluorescence telescopes that image the ultraviolet glow of nitrogen excited along the shower track. Both see the spectral suppression above $\sim4\times10^{19}\,\mathrm{eV}$ , and the live debate is its origin: Auger's shower-maximum ( $X_{\max}$ ) data favour an increasingly heavy composition toward the top of the spectrum, which would make the suppression partly a source-rigidity limit ( $E_{\max}\propto Z$ ) rather than pure proton GZK. Auger has also reported a large-scale dipole anisotropy above $8\times10^{18}\,\mathrm{eV}$ pointing away from the Galactic centre — direct evidence of extragalactic origin — plus intriguing correlations with nearby starburst galaxies. The decisive test is cosmogenic neutrinos from the $\pi^+$ channel: neutral and undeflected, they would confirm the photopion interaction and fix the proton fraction independent of the spectrum, and are prime targets for IceCube-Gen2, GRAND, and radio arrays such as the planned in-ice and balloon-borne detectors. An upgraded AugerPrime and next-generation concepts like POEMMA aim to close the composition-versus-attenuation question outright.

The physical picture

Imagine sprinting through a gentle mist. Walk slowly and the droplets are harmless; sprint near light-speed and, in your frame, the mist becomes a hail of hard projectiles slamming into your face. A UHECR proton moving through the CMB experiences exactly this: a $10^{-3}\,\mathrm{eV}$ microwave photon is Doppler-boosted into the proton's rest frame to hundreds of MeV — enough to excite the proton into a $\Delta$ resonance that promptly decays by spitting out a pion, carrying away a chunk of the proton's energy. The first figure is that collision drawn frame-by-frame, with the threshold and the per-collision energy loss written on it.

Figure

A UHECR proton meets a CMB photon: boosted above the Δ(1232) threshold the pair makes a resonance that decays to p + π⁰ or n + π⁺, each hit removing κ ≈ 20% of the energy over a mean free path of a few Mpc — threshold E_th ≈ 5×10¹⁹ eV

The result is a self-cleaning universe at extreme energies. Because each interaction costs $\sim20\%$ of the energy and the mean free path is only a few Mpc, a proton born at $10^{21}\,\mathrm{eV}$ is ground down toward $\sim10^{20}\,\mathrm{eV}$ within $\sim100\,\mathrm{Mpc}$ regardless of its starting energy. Earth therefore sees a pile-up just below threshold and a sharp fall-off above it — a spectral fingerprint of the CMB written across the highest-energy sky. That fingerprint is what the observatories actually measure, and the second figure is the quantity they report: the all-particle spectrum with its knee, its ankle, and the GZK suppression at the top.

Figure

The all-particle cosmic-ray spectrum: a steep power law with the knee at 3×10¹⁵ eV, the extragalactic ankle at 5×10¹⁸ eV, and the GZK suppression above 4×10¹⁹ eV where the horizon collapses to about 100 Mpc

Because these particles are charged, they do not point straight back to their sources — Galactic and intergalactic magnetic fields bend their trajectories. But the same GZK horizon that limits their reach also guarantees the sources are cosmologically nearby, so with enough events the arrival directions should trace the local matter distribution rather than a smooth isotropic sky.

The core equations

The spectrum is the raw observable: a steep, featured power law that no single accelerator population produces alone. The two breaks — knee and ankle — mark transitions between source classes and between Galactic and extragalactic origin.

\frac{\mathrm{d}N}{\mathrm{d}E}\propto E^{-\gamma},\quad \gamma\simeq\begin{cases}2.7 & E<E_{\mathrm{knee}}\\ 3.1 & E_{\mathrm{knee}}<E<E_{\mathrm{ankle}}\\ 2.6 & E>E_{\mathrm{ankle}}\end{cases} \qquad (1)

where: $E$ — cosmic-ray energy; $\gamma$ — spectral index (power-law slope); $E_{\mathrm{knee}}\approx3\times10^{15}\,\mathrm{eV}$ , where the spectrum steepens; $E_{\mathrm{ankle}}\approx5\times10^{18}\,\mathrm{eV}$ , where it flattens at the extragalactic onset. Reading: a slope steeper than a few means the flux plummets — the decade of energy from ankle to cutoff costs three to four decades of flux, which is why exposure of thousands of km² is required. The GZK physics starts with a kinematic threshold. A proton and a photon can make a pion only if their invariant mass reaches the proton-plus-pion rest energy; below that the reaction is simply forbidden.

E_{\mathrm{th}}=\frac{(m_p+m_\pi)^2-m_p^2}{4\,\epsilon\,(1-\cos\theta)}\,c^4 \qquad (2)

where: $E_{\mathrm{th}}$ — minimum proton energy for pion production; $m_p=938.3\,\mathrm{MeV}/c^2$ ; $m_\pi=135.0\,\mathrm{MeV}/c^2$ ( $\pi^0$ ); $\epsilon$ — target-photon energy in the lab (CMB) frame; $\theta$ — collision angle, with head-on ( $\theta=\pi$ , so $1-\cos\theta=2$ ) giving the lowest threshold. Reading: the threshold scales as $1/\epsilon$ , so the abundant high-energy tail of the CMB sets the effective onset well below the naive mean-photon estimate. Above threshold the reaction runs through a single dominant resonance, which is why the cross section is large and the mean free path short.

p+\gamma_{\mathrm{CMB}}\to\Delta^{+}(1232)\to\begin{cases}p+\pi^{0}\\ n+\pi^{+}\end{cases} \qquad (3)

where: $\Delta^{+}(1232)$ — the baryon resonance of mass $1232\,\mathrm{MeV}/c^2$ that dominates the cross section near threshold; $\pi^0\to2\gamma$ feeds the cosmogenic gamma-ray background; $\pi^+\to\mu^+\nu_\mu$ feeds the cosmogenic neutrino flux. Reading: the two branches carry roughly equal weight, so about half of GZK interactions convert a proton to a neutron and back, producing neutrinos as an unavoidable by-product. How far a super-threshold proton travels between collisions is fixed by the target density and the resonant cross section.

\lambda_{p\gamma}=\left(n_\gamma\,\sigma_{p\gamma}\right)^{-1}\sim\text{few}\,\mathrm{Mpc} \qquad (4)

where: $\lambda_{p\gamma}$ — photopion mean free path above threshold; $n_\gamma\approx411\,\mathrm{cm}^{-3}$ — CMB photon number density; $\sigma_{p\gamma}\sim5\times10^{-28}\,\mathrm{cm}^2$ (0.5 mb) — cross section at the $\Delta$ peak. Reading: only the above-threshold fraction of the photon bath counts, so the effective density is a fraction of $411\,\mathrm{cm}^{-3}$ and $\lambda$ lands in the few-Mpc range. That photon density is a blackbody result, computed once from the CMB temperature and carried into every GZK estimate.

n_\gamma=\frac{2\zeta(3)}{\pi^2}\left(\frac{k_BT}{\hbar c}\right)^3\approx411\,\mathrm{cm}^{-3} \qquad (5)

where: $\zeta(3)\approx1.202$ ; $T=2.725\,\mathrm{K}$ is the CMB temperature; $k_B,\hbar,c$ are the Boltzmann, reduced-Planck, and light constants. Reading: this is the same number density used for photon-baryon ratios in cosmology — recalled from aq-astro-cosmic-microwave-background, not re-derived here. Turning the mean free path into a horizon requires the fractional energy lost per collision.

L_{\mathrm{loss}}=\frac{1}{\kappa\,n_\gamma\,\sigma_{p\gamma}},\qquad \kappa=\frac{\Delta E}{E}\approx0.2 \qquad (6)

where: $L_{\mathrm{loss}}$ — energy-loss (attenuation) length; $\kappa$ — inelasticity, the fraction of energy lost per photopion event. Reading: because each hit costs only $\sim20\%$ , several are needed to degrade the proton, so the loss length is a few times $\lambda$ — the origin of the $\sim100\,\mathrm{Mpc}$ GZK horizon. Finally, whether a source can reach these energies at all is bounded by confinement: it must hold the particle long enough to accelerate it.

E_{\max}\simeq Z\,e\,\beta\,B\,R\,c \qquad (7)

where: $E_{\max}$ — maximum energy a source can impart; $Z$ — charge number; $\beta$ — characteristic speed of the scattering centres in units of $c$ ; $B$ — magnetic field; $R$ — size of the acceleration region. Reading: this is the Hillas criterion — a source reaches $10^{20}\,\mathrm{eV}$ only if the product $BR$ is large enough, which eliminates most astrophysical objects at a glance.

Derivation AThreshold energy for GZK photopion production

The goal is to compute, from special relativity alone, the minimum proton energy at which the reaction $p+\gamma\to\Delta\to N+\pi$ can proceed on a CMB photon, and to see why the answer sits near $5\times10^{19}\,\mathrm{eV}$ rather than higher. The clean route is the Lorentz-invariant Mandelstam variable $s$ : because $s$ is the same in every frame, we can evaluate it in the lab (where the CMB photon has a tiny energy and the proton is ultra-relativistic) and set it equal to its value in the centre-of-mass frame at threshold (where the final $N+\pi$ system is at rest). Equating the two gives the threshold with no reference to any particular frame, and inserting the CMB photon energy fixes the scale.

Step 1. Form the invariant mass of the proton-photon system, $s=(p_p+p_\gamma)^2=m_p^2c^4+2\,p_p\!\cdot\!p_\gamma$ , using $c=1$ bookkeeping and the fact that the photon is massless. Why: $s$ is frame-independent, so this single quantity carries the whole threshold condition and lets us dodge any explicit Lorentz boost into the proton frame.

Step 2. Evaluate the four-vector dot product for an ultra-relativistic proton ( $E\gg m_pc^2$ , so $|\vec p_p|c\approx E$ ) meeting a photon of energy $\epsilon$ at angle $\theta$ : $p_p\!\cdot\!p_\gamma=E\epsilon(1-\cos\theta)/c^2$ , giving $s=m_p^2c^4+2E\epsilon(1-\cos\theta)$ . Why: head-on collisions ( $\cos\theta=-1$ ) maximise the momentum transfer and therefore minimise the required proton energy, so we take $1-\cos\theta=2$ for the lowest threshold.

Step 3. Impose the threshold: the final state must have at least the rest energy of a proton plus a pion, $s\ge(m_p+m_\pi)^2c^4$ , with equality when the products are produced at rest in the centre-of-mass frame. Why: this is the minimum invariant mass that can materialise a pion — below it the reaction is kinematically impossible no matter how the energy is arranged.

Step 4. Solve for the proton energy: $E_{\mathrm{th}}=\dfrac{(m_p+m_\pi)^2c^4-m_p^2c^4}{4\epsilon}=\dfrac{2m_pm_\pi+m_\pi^2}{4\epsilon}c^4$ . The numerator is $2(938.3)(135.0)+(135.0)^2\approx2.72\times10^{5}\,\mathrm{MeV}^2$ . With $\epsilon\sim10^{-3}\,\mathrm{eV}$ this gives $E_{\mathrm{th}}\sim7\times10^{19}\,\mathrm{eV}$ ; using the mean photon energy $\langle\epsilon\rangle\approx2.7k_BT\approx6\times10^{-4}\,\mathrm{eV}$ gives $\sim10^{20}\,\mathrm{eV}$ , and the abundant high-energy tail of the CMB pulls the effective onset down to $\sim5\times10^{19}\,\mathrm{eV}$ . Why: the threshold scales as $1/\epsilon$ , so the reaction switches on at the energy where enough photons lie above the local threshold — a tail effect, which is why the observed suppression is gradual rather than a razor edge. *As a check: the numerator $(1073.3)^2-(938.3)^2=2.716\times10^{5}\,\mathrm{MeV}^2$ ; dividing by $4\epsilon$ with $\epsilon\sim10^{-3}\,\mathrm{eV}$ recovers $\sim10^{20}\,\mathrm{eV}$ , the quoted GZK scale.*

Go deeper · Derivation B · 2 minEstimating the GZK horizon length

Derivation BEstimating the GZK horizon length

The goal is to turn microscopic collision data — the CMB photon density, the $\Delta$ -resonance cross section, and the per-collision inelasticity — into the macroscopic distance a super-threshold proton can travel before it is degraded below the cutoff. The strategy is two-stage: first the mean free path $\lambda=1/(n_\gamma\sigma)$ tells us how far the proton goes between hits, then dividing by the inelasticity $\kappa$ accounts for the fact that each hit removes only a fraction of the energy, so several are needed to matter. The product is the attenuation length, and comparing it to cosmological distances shows why the highest-energy sky is a local sky.

Step 1. Write the mean free path $\lambda=1/(n_\gamma\,\sigma_{p\gamma})$ , but note that only photons above the kinematic threshold are viable targets, so the effective density is a fraction of the full $411\,\mathrm{cm}^{-3}$ . Why: a proton at a given energy sees only the high-frequency tail of the bath as "hard enough" to make a pion, so using the full density would badly overestimate the collision rate.

Step 2. Insert numbers with an effective above-threshold density $\sim50\,\mathrm{cm}^{-3}$ and $\sigma_{p\gamma}\sim5\times10^{-28}\,\mathrm{cm}^2$ : $\lambda\sim1/[(50)(5\times10^{-28})]\approx4\times10^{25}\,\mathrm{cm}\approx13\,\mathrm{Mpc}$ , using $1\,\mathrm{Mpc}=3.086\times10^{24}\,\mathrm{cm}$ . Why: this is the actual few-Mpc mean free path near the peak of the cross section, and it already shows the proton collides many times within a supercluster.

Step 3. Divide by the inelasticity to get the energy-loss length: $L_{\mathrm{loss}}=\lambda/\kappa\approx13/0.2\approx65\,\mathrm{Mpc}$ , of order the canonical $\sim100\,\mathrm{Mpc}$ GZK horizon. Why: because each collision drains only $\sim20\%$ , it takes several collisions to knock a proton below threshold, so the horizon is a few times the bare mean free path.

Step 4. Note the energy dependence: as $E$ rises, more of the CMB lies above threshold, $\lambda$ shrinks, and $L_{\mathrm{loss}}(E)$ falls toward $\sim10\,\mathrm{Mpc}$ near $10^{21}\,\mathrm{eV}$ . Why: the horizon is not a fixed wall but an energy-dependent surface, which is exactly why the suppression is a curve and not a step, and why the very highest events must come from the very nearest sources. *As a check: $\lambda\approx4\times10^{25}\,\mathrm{cm}=12.96\,\mathrm{Mpc}$ ; $\lambda/\kappa\approx65\,\mathrm{Mpc}$ , inside the standard $50$ – $100\,\mathrm{Mpc}$ horizon.*

Go deeper · Derivation C · 2 minWhich sources can reach 10²⁰ eV — the Hillas criterion

Derivation CWhich sources can reach 10²⁰ eV — the Hillas criterion

The goal is to convert the abstract statement "a source must confine a particle to accelerate it" into a concrete filter on real astrophysical objects, and then run candidate accelerators through it. The physics is a single inequality: the particle's Larmor radius in the source's magnetic field must not exceed the size of the acceleration region, or the particle escapes before gaining more energy. Rearranged, this demands a minimum product of magnetic field and size, $BR$ , for a target energy — a criterion so restrictive that it eliminates almost every object in the sky and leaves a short list that must also lie within the GZK horizon.

Step 1. Write the relativistic Larmor radius $r_L=E/(ZeBc)$ for a particle of energy $E$ and charge $Ze$ gyrating in field $B$ . Why: this is the natural length scale of a charged particle's orbit, and acceleration mechanisms that scatter the particle back and forth (diffusive shock acceleration) fail once the orbit no longer fits inside the accelerator.

Step 2. Impose confinement, $r_L\lesssim R$ , and solve for the maximum energy: $E_{\max}\sim ZeBRc$ , refined by a shock-efficiency factor $\beta$ to $E_{\max}\simeq Ze\beta BRc$ . Why: $\beta<1$ because scattering centres move slower than light, so realistic engines fall short of the geometric bound and the true requirement on $BR$ is even more demanding.

Step 3. Compute the requirement for a $10^{20}\,\mathrm{eV}=16\,\mathrm{J}$ proton ( $Z=1$ ). In Gaussian units $E=eBR$ , so $BR=E/e=(1.6\times10^{8}\,\mathrm{erg})/(4.8\times10^{-10}\,\mathrm{esu})\approx3.3\times10^{17}\,\mathrm{G\,cm}$ . Why: this single number is the whole Hillas plot in one line — any accelerator with a smaller $BR$ simply cannot reach the highest observed energies, whatever its luminosity.

Step 4. Test candidates. Radio-galaxy lobes: $B\sim10^{-5}\,\mathrm{G}$ , $R\sim100\,\mathrm{kpc}\sim3\times10^{23}\,\mathrm{cm}$ , so $BR\sim3\times10^{18}\,\mathrm{G\,cm}$ — comfortably above the bar. Neutron-star magnetospheres: $B\sim10^{12}\,\mathrm{G}$ , $R\sim10\,\mathrm{km}=10^{6}\,\mathrm{cm}$ , so $BR\sim10^{18}\,\mathrm{G\,cm}$ — also sufficient in principle. Why: the criterion is necessary but not sufficient — it admits AGN lobes, jets, magnetars, and gamma-ray bursts while excluding ordinary stars and supernova remnants, giving a physically motivated short list of accelerators to look for inside the GZK volume. *As a check: $16\,\mathrm{J}=1.6\times10^{8}\,\mathrm{erg}$ divided by $e=4.803\times10^{-10}\,\mathrm{esu}$ gives $BR=3.33\times10^{17}\,\mathrm{G\,cm}$ , the standard Hillas value for $Z=1$ .*

Limiting cases

The physics separates into regimes set by the proton (or nucleus) energy relative to the photopion and photodisintegration thresholds:

Case	Result	Meaning
Below GZK ( $E\ll E_{\mathrm{th}}$ )	$E\lesssim10^{19}\,\mathrm{eV}$	Photopion production is kinematically forbidden on typical CMB photons; the universe is transparent and cosmic rays propagate over Gpc scales, losing energy only adiabatically to expansion.
Pair-production dip	$E\sim10^{18}\text{–}10^{19}\,\mathrm{eV}$	$p+\gamma_{\mathrm{CMB}}\to p+e^{+}+e^{-}$ has a lower threshold but long loss length ( $\sim\mathrm{Gpc}$ ); it carves a shallow dip read by some as the ankle for a pure-proton spectrum.
GZK suppression	$E\gtrsim5\times10^{19}\,\mathrm{eV}$	Photopion production switches on at the $\Delta(1232)$ resonance; the horizon collapses to $\sim100\,\mathrm{Mpc}$ and shrinks further with energy — the observed flux is sharply suppressed.
Heavy nuclei	$A\gg1$	Photodisintegration $A+\gamma\to(A-1)+N$ via the giant dipole resonance dominates; iron survives to somewhat larger distances than protons, so a mixed or heavy composition shifts and softens the effective cutoff.

Common trapThe GZK cutoff is a hard wall at a single fixed energy

It is not. The GZK "cutoff" is a suppression of the observed flux from distant sources, set by an energy-dependent horizon — not a forbidden zone. Particles above the threshold certainly exist locally; the Oh-My-God particle at $3\times10^{20}\,\mathrm{eV}$ was recorded in 1991. What the horizon forbids is their arrival from far away: a $10^{20}\,\mathrm{eV}$ proton born a Gpc distant is degraded below threshold long before it reaches us, while one born within $\sim100\,\mathrm{Mpc}$ survives. The deeper trap is interpretive: a pure-proton GZK attenuation and an intrinsic source-injection cutoff ( $E_{\max}\propto Z$ from the Hillas bound) produce nearly identical spectral shapes. You cannot tell them apart from the spectrum alone — only composition ( $X_{\max}$ ) and cosmogenic-neutrino data can break the degeneracy.

Worked examples

1 · The GZK threshold on a real CMB photon (fully worked)

A proton propagates through the CMB. Taking the mean photon energy $\langle\epsilon\rangle\approx6.3\times10^{-4}\,\mathrm{eV}$ and a head-on collision, find the threshold proton energy for photopion production, and comment on why the observed suppression sits lower, near $5\times10^{19}\,\mathrm{eV}$ .

Step 1. Assemble the invariant-mass numerator. With $m_p=938.3\,\mathrm{MeV}/c^2$ and $m_\pi=135.0\,\mathrm{MeV}/c^2$ , $(m_p+m_\pi)^2-m_p^2=2m_pm_\pi+m_\pi^2=2(938.3)(135.0)+(135.0)^2=253{,}341+18{,}225=\mathbf{2.716\times10^{5}\,\mathrm{MeV}^2}$ . Converting to eV², this is $2.716\times10^{5}\times(10^{6})^2=2.716\times10^{17}\,\mathrm{eV}^2$ .

Step 2. Apply the head-on threshold formula ( $1-\cos\theta=2$ , so the denominator is $4\epsilon$ ): $E_{\mathrm{th}}=\dfrac{2.716\times10^{17}\,\mathrm{eV}^2}{4(6.3\times10^{-4}\,\mathrm{eV})}=\dfrac{2.716\times10^{17}}{2.52\times10^{-3}}\approx\mathbf{1.1\times10^{20}\,\mathrm{eV}}$ .

Step 3. Interpret. This mean-photon estimate says a proton needs $\sim10^{20}\,\mathrm{eV}$ to make pions on a typical CMB photon. But the CMB is a blackbody with a high-energy tail: photons several times $\langle\epsilon\rangle$ are plentiful, and because $E_{\mathrm{th}}\propto1/\epsilon$ , those harder photons are viable targets for protons of only $\sim5\times10^{19}\,\mathrm{eV}$ . The effective onset is therefore lower than the mean-photon number. The mean-photon threshold is $\approx1.1\times10^{20}\,\mathrm{eV}$ ; the tail of the CMB drops the effective GZK onset to $\approx5\times10^{19}\,\mathrm{eV}$ .

2 · The horizon from photon density and cross section

Estimate the photopion mean free path and the GZK attenuation length using $n_\gamma\approx411\,\mathrm{cm}^{-3}$ , an effective above-threshold density $\sim50\,\mathrm{cm}^{-3}$ , $\sigma_{p\gamma}\sim5\times10^{-28}\,\mathrm{cm}^2$ , and $\kappa\approx0.2$ .

Step 1. Mean free path: $\lambda=1/(n_{\mathrm{eff}}\sigma_{p\gamma})=1/[(50)(5\times10^{-28})]=1/(2.5\times10^{-26})=4.0\times10^{25}\,\mathrm{cm}$ . Converting with $1\,\mathrm{Mpc}=3.086\times10^{24}\,\mathrm{cm}$ gives $\lambda=\mathbf{13\,\mathrm{Mpc}}$ .

Step 2. Attenuation length: because each collision removes only $\kappa\approx0.2$ of the energy, $L_{\mathrm{loss}}=\lambda/\kappa=13/0.2=\mathbf{65\,\mathrm{Mpc}}$ , of order the canonical $\sim100\,\mathrm{Mpc}$ GZK horizon.

Step 3. Sanity check against the sky. The Virgo cluster is $\sim16\,\mathrm{Mpc}$ away and the nearest powerful radio galaxy, Centaurus A, is $\sim3.8\,\mathrm{Mpc}$ away — both comfortably inside the horizon, whereas a source at cosmological distance ( $\gtrsim1\,\mathrm{Gpc}$ ) is $\sim15$ horizons away and contributes nothing above the cutoff. The horizon is $\sim60$ – $100\,\mathrm{Mpc}$ , so the highest-energy sky samples only the local supercluster.

3 · Does a galaxy-cluster shock qualify as an accelerator?

A cluster accretion shock has $B\sim1\,\mu\mathrm{G}=10^{-6}\,\mathrm{G}$ over $R\sim1\,\mathrm{Mpc}=3.1\times10^{24}\,\mathrm{cm}$ . Can it reach $10^{20}\,\mathrm{eV}$ for protons?

Step 1. Form the Hillas product: $BR=(10^{-6}\,\mathrm{G})(3.1\times10^{24}\,\mathrm{cm})=3.1\times10^{18}\,\mathrm{G\,cm}$ .

Step 2. Compare to the requirement $BR\gtrsim3.3\times10^{17}\,\mathrm{G\,cm}$ for a $10^{20}\,\mathrm{eV}$ proton. The cluster exceeds it by about a factor of ten geometrically. Geometrically yes — but with a realistic shock efficiency $\beta\sim10^{-2}$ the achievable energy drops to $E_{\max}\sim\beta\times10^{21}\sim10^{19}\,\mathrm{eV}$ , so slow cluster shocks fall an order of magnitude short in practice — a reminder that the Hillas bound is necessary, not sufficient.

Physics of the portrait

The GZK cutoff is where high-energy astrophysics and special relativity collide — literally. Its story runs from balloon flights over Austria to a resonance discovered in a particle accelerator, tied together by three people who, within months of the CMB's discovery, saw that the newest cosmological measurement had capped the oldest cosmic-ray puzzle.

Origins · 1 minKenneth Greisen and how this was discovered

Kenneth Greisen

1918–2007, Cornell cosmic-ray physicist · air showers and the CMB energy cutoff

Kenneth Greisen spent his career on extensive air showers — the cascades of secondary particles that a single high-energy cosmic ray triggers in the atmosphere — co-authoring with Bruno Rossi the standard shower theory used to reconstruct primary energies. When Penzias and Wilson reported the $2.7\,\mathrm{K}$ microwave background in 1965, Greisen immediately grasped its consequence for the most energetic protons: the newly measured photon bath would be opaque to them. In a terse 1966 Physical Review Letters he showed that above $\sim5\times10^{19}\,\mathrm{eV}$ a proton must lose energy to pion production on the CMB, so the spectrum should steepen sharply — "the cosmic-ray spectrum should be cut off," he wrote. The same conclusion was reached independently and almost simultaneously by Georgiy Zatsepin and Vadim Kuzmin in the Soviet Union, and the effect has carried all three names ever since. The prediction sat unconfirmed for four decades because the flux is so low that no detector was large enough; the paradox deepened when the University of Utah's Fly's Eye recorded the Oh-My-God particle at $3\times10^{20}\,\mathrm{eV}$ in 1991, apparently above the cutoff and therefore necessarily from a nearby source. It was only with the Pierre Auger Observatory — a $3000\,\mathrm{km}^2$ hybrid array named for the discoverer of air showers — that the suppression was measured directly in 2008, vindicating a calculation Greisen had done on the back of the CMB's discovery.

History

1912

— Hess discovers cosmic rays on high-altitude balloon flights

1966

— Greisen and, independently, Zatsepin and Kuzmin predict the CMB cutoff

1984

— Hillas formalizes the confinement criterion for cosmic accelerators

1991

— the Fly's Eye detector records the 3×10²⁰ eV Oh-My-God particle

2008

— the Pierre Auger Observatory reports the flux suppression above 4×10¹⁹ eV

Going further

Texts. Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., §12.2 and Ch. 27 (cosmic rays, high-energy astrophysics) · Ryden, Introduction to Cosmology, 2nd ed. (CMB properties and the blackbody photon density) · Dodelson, Modern Cosmology (CMB spectrum and number density) · Mo, van den Bosch & White, Galaxy Formation and Evolution (intergalactic medium, magnetic fields) · Longair, High Energy Astrophysics, 3rd ed. (acceleration and propagation). The threshold formula follows from Mandelstam- $s$ invariance: the numerator $(m_p+m_\pi)^2-m_p^2=2.716\times10^{5}\,\mathrm{MeV}^2$ with $m_p=938.3\,\mathrm{MeV}/c^2$ and $m_{\pi^0}=135.0\,\mathrm{MeV}/c^2$ , divided by $4\epsilon$ with $\epsilon\sim6\times10^{-4}\,\mathrm{eV}$ , recovers $E_{\mathrm{th}}\sim10^{20}\,\mathrm{eV}$ . The CMB density $n_\gamma\approx411\,\mathrm{cm}^{-3}$ follows from $\tfrac{2\zeta(3)}{\pi^2}(k_BT/\hbar c)^3$ at $T=2.725\,\mathrm{K}$ ; the horizon from $\lambda/\kappa$ with $\lambda=1/(n_{\mathrm{eff}}\sigma)$ ; the Hillas bound $E_{\max}\simeq Ze\beta BRc$ , giving $BR\gtrsim3.3\times10^{17}\,\mathrm{G\,cm}$ for $Z=1$ . The observed suppression near $4\times10^{19}\,\mathrm{eV}$ and the dipole anisotropy are from the Pierre Auger Collaboration.

Prerequisites. Special Relativity Kinematics · The Cosmic Microwave Background · Relativistic Jets and Launching Mechanisms · Active Galactic Nuclei: Unified Model. Leads to. Blazars and Relativistic Beaming · Supermassive Black Holes · Black Hole Spin and the Ergosphere.

Ask the tutor: Ask me to boost a specific CMB photon into a $10^{20}\,\mathrm{eV}$ proton's rest frame and show it clearing the $\Delta(1232)$ threshold step by step.

How can the CMB, at only 2.7 K, stop the most energetic particles in the universe?

Temperature is the wrong yardstick; what matters is the photon energy in the proton's rest frame. A CMB photon carries only $\sim10^{-3}\,\mathrm{eV}$ in the lab, but a proton at $10^{20}\,\mathrm{eV}$ has a Lorentz factor $\gamma\sim10^{11}$ , and the head-on collision blueshifts the photon by roughly $2\gamma$ into the proton frame — lifting a milli-eV microwave photon to several hundred MeV, above the $\Delta(1232)$ resonance. So the "cold" bath looks like a beam of gamma-ray-energy photons to an ultra-relativistic proton. The bath is also dense ( $\sim411$ photons per cm³) and fills all of space, so even a small cross section adds up to a short mean free path. Coldness is irrelevant; extreme relativistic boosting is everything.

Is the GZK suppression the same thing as the sources simply running out of energy?

Not necessarily, and telling them apart is the field's central problem. Pure GZK attenuation degrades protons above $5\times10^{19}\,\mathrm{eV}$ over $\sim100\,\mathrm{Mpc}$ , producing a spectral steepening. But sources also have a maximum energy set by the Hillas bound, $E_{\max}\propto Z$ , and if the accelerators simply cannot push particles past a few $\times10^{19}\,\mathrm{eV}$ , the spectrum steepens at the same place for an entirely different reason. The two mechanisms yield nearly degenerate spectral shapes. What breaks the tie is composition — a source-limited cutoff scales with charge $Z$ , so it predicts a specific mass trend across the cutoff — and cosmogenic neutrinos, which are copious only if protons are actually undergoing GZK interactions.

How can one proton carry 16 joules — the energy of a served tennis ball?

The number is real: $10^{20}\,\mathrm{eV}\times1.602\times10^{-19}\,\mathrm{J/eV}=16\,\mathrm{J}$ , and the Oh-My-God particle at $3\times10^{20}\,\mathrm{eV}$ carried $\sim48\,\mathrm{J}$ . It is not that the proton is heavy — its rest energy is a fixed $938\,\mathrm{MeV}$ — but that its Lorentz factor is staggering, $\gamma\sim10^{11}$ , so it moves at all but a part in $10^{22}$ of light speed. Over a light-year it would lose a race to a photon by less than the width of an atom. The kinetic energy is macroscopic while the momentum is carried by a single particle, which is precisely why the air shower it produces contains billions of secondaries and can be detected from kilometres away.

Why don't ultra-high-energy cosmic rays point straight back to their sources?

Because they are charged, and every magnetic field they cross bends them. A proton's Larmor radius even at $10^{20}\,\mathrm{eV}$ in a microgauss intergalactic field is tens of Mpc, so over a $100\,\mathrm{Mpc}$ path it can be deflected by several degrees; heavier nuclei with charge $Z$ deflect $Z$ times more, smearing iron arrival directions across tens of degrees. The Galactic field adds a further, poorly-mapped deflection near the end of the journey. The saving grace is the GZK horizon: because the sources must lie within $\sim100\,\mathrm{Mpc}$ , the arrival directions — though blurred — should still cluster toward the local large-scale structure rather than being isotropic, which is exactly the dipole anisotropy Auger reports.

Do heavy nuclei like iron have their own GZK-type cutoff?

Yes, but through a different reaction. Instead of photopion production, a nucleus of mass number $A$ loses nucleons by photodisintegration, $A+\gamma\to(A-1)+N$ , driven by the giant dipole resonance when a CMB or infrared-background photon is boosted to $\sim10$ – $20\,\mathrm{MeV}$ in the nuclear rest frame. Iron is more robust than a proton of the same energy because its energy per nucleon is lower and the disintegration threshold is correspondingly higher, so iron survives to somewhat larger distances. But a nucleus that does disintegrate cascades into lighter fragments that are then themselves attenuated. The net effect is a cutoff whose position and sharpness depend on composition — one reason the observed suppression is softer than a pure-proton GZK prediction.

Has the GZK cutoff actually been confirmed, or is it still theoretical?

The spectral suppression is firmly observed. Both the Pierre Auger Observatory and the Telescope Array independently measure a sharp steepening of the flux above $\sim4\times10^{19}\,\mathrm{eV}$ , at high statistical significance, exactly where GZK predicts. What is not yet settled is whether that suppression is pure GZK attenuation on the CMB or a source-imposed maximum energy — the two are spectrally degenerate. Confirming the mechanism rather than just the feature requires either a clean proton composition at the cutoff (favouring GZK) or the detection of cosmogenic neutrinos, the unavoidable by-product of proton photopion interactions. So the cutoff is real and measured; its interpretation as the Greisen–Zatsepin–Kuzmin process specifically is strongly supported but awaits the neutrino smoking gun.