Comptonization and the Sunyaev-Zel'dovich Effect

Compton scattering between one photon and one electron conserves energy exactly, but the average outcome over many scatterings tilts toward whichever partner is hotter. When electrons are far more energetic than the photons bathing them, energy flows uphill from matter to radiation — inverse Compton scattering — and a soft photon field is slowly hardened, its spectrum pushed to higher frequencies. Comptonization is the theory of this diffusive redistribution of photons in frequency, valid where scattering, not absorption, dominates, so photon number is conserved while photon energy is not. Its master equation, written by Kompaneets in 1957, is a Fokker-Planck equation in frequency governing the occupation number n(ν). The most celebrated arena for this physics is the Sunyaev-Zel'dovich effect. CMB photons streaming through the hot, tenuous plasma trapped in the gravitational wells of galaxy clusters are inverse-Compton scattered by intracluster electrons at Tₑ ∼ 10⁷–10⁸ K and nₑ ∼ 10⁻³ cm⁻³. The optical depth is tiny (τ ∼ 10⁻²), so only about a percent of the photons scatter — but each gains a fractional energy of order k_BTₑ/mₑc². The net imprint shifts photons from the Rayleigh-Jeans side of the Planck spectrum toward the Wien side, a distortion with a fixed spectral signature: a decrement below 217 GHz, a null at 217 GHz, an increment above. Its amplitude, set by the Compton y-parameter, is independent of the cluster's redshift — the property that makes the SZ effect a nearly ideal instrument for finding and weighing the most massive structures in the universe out to the epoch when they first assembled.

Key takeaways

Comptonization is the cumulative energy exchange between photons and free electrons through repeated (inverse) Compton scattering, described by the Kompaneets equation for the occupation number n(x).
Off electrons much hotter than the radiation, photons gain energy on average — the spectrum drifts up by ⟨Δν/ν⟩ ≈ 4k_BTₑ/mₑc² per scatter.
The Compton y-parameter, y = ∫ (k_BTₑ/mₑc²) nₑσ_T dℓ = ⟨k_BTₑ/mₑc²⟩ τ, factorizes energy-per-scatter times number-of-scatters and sets the distortion amplitude.
The thermal Sunyaev-Zel'dovich (tSZ) effect imprints this on the CMB through hot intracluster gas: ΔT/T = y g(x) with g(x) = x coth(x/2) − 4 — a decrement below 217 GHz, an increment above, redshift-independent amplitude.
Because SZ surface brightness is redshift-independent while X-ray brightness dims as (1+z)⁻⁴, the SZ effect uniquely probes distant clusters; with X-rays it yields distances and H₀.
The kinematic SZ effect, ΔT/T = −τ v_r/c, is sourced by a cluster's bulk peculiar velocity and has a blackbody spectrum, separating it from the tSZ null.
Key idea: A single Compton scatter off a moving electron shifts a photon's frequency by a small random amount; averaged over a thermal electron gas the net drift is upward by ⟨Δν/ν⟩ ∼ 4k_BTₑ/mₑc² per scatter, and the accumulated distortion of the whole spectrum is governed by the single parameter y = ∫ (k_BTₑ/mₑc²) nₑσ_T dℓ.

The core insight

A single Compton scatter off a moving electron shifts a photon's frequency by a small random amount; averaged over a thermal electron gas the net drift is upward by ⟨Δν/ν⟩ ∼ 4k_BTₑ/mₑc² per scatter, and the accumulated distortion of the whole spectrum is governed by the single dimensionless parameter y = ∫ (k_BTₑ/mₑc²) nₑσ_T dℓ.

Where this lives today

The SZ effect has graduated from a hard-won single-cluster detection to an industrial cosmological probe. Because the integrated signal Y = ∫ y dΩ ∝ ∫ nₑTₑ dV is essentially the total thermal energy of the gas — a low-scatter, redshift-independent mass proxy — SZ surveys with Planck, the South Pole Telescope, and the Atacama Cosmology Telescope now catalog thousands of clusters selected almost purely by mass out to z ≳ 1. Counting the most massive clusters versus redshift constrains σ₈, Ω_m, and the dark-energy equation of state through the growth of structure, in mild and closely watched tension with primary-CMB values. The kinematic effect, long a theorist's footnote, is now detected statistically: stacking cluster pairs (the pairwise kSZ) reveals the coherent infall of the cosmic web and weighs the missing baryons in halos independently of X-rays. And the same physics sets the target for a future CMB spectrometer: a PIXIE- or Voyage-2050-class mission measuring the sky-averaged μ- and y-distortions — a few parts in 10⁸ predicted from Silk damping and reionization — would open a spectral window on energy release back toward inflation.

The physical picture

Picture a slow, soft photon meeting a fast electron. In the electron's rest frame the photon has a higher energy but scatters essentially elastically — the Thomson limit, hν′ ≪ mₑc². Transform back to the lab, though, and the outgoing photon is Doppler-boosted by the electron's motion: on average it emerges with slightly more energy, because the geometry favours the head-on encounters that up-shift it. Repeat over a hot gas and the entire photon population creeps upward in frequency. The first figure freezes one such encounter and attaches the number that controls it — the mean fractional energy gain per scatter.

Figure

Inverse Compton scattering of a soft photon off a fast thermal electron: the outgoing photon has a higher frequency than the incoming one, and averaged over a Maxwellian electron gas the mean fractional gain per scatter is ⟨Δν/ν⟩ = 4k_BTₑ/mₑc² ≈ 0.06 for k_BTₑ = 8 keV

For the CMB passing through a cluster the number of scatters matters as much as the gain per scatter. Most photons pass straight through (τ is small), but the roughly one percent that scatter are lifted from low frequencies to high. An observer below 217 GHz therefore sees fewer photons than the surrounding sky — a decrement, or "cold spot" — while above 217 GHz sees more. Total photon number and total CMB energy are unchanged; energy is merely borrowed from the hot electrons. The y-parameter collapses all of this into one number, y = (energy per scatter) × (number of scatters) = (k_BTₑ/mₑc²) × τ along the line of sight; only this product fixes the amplitude, while the spectral shape is universal. The second figure is that universal shape — the fingerprint every SZ experiment is built to detect.

Figure

The universal thermal SZ spectral distortion of the CMB intensity: a decrement peaking near 128 GHz, a null exactly at ν₀ = 217 GHz where x₀ = 3.83 solves x coth(x/2) = 4, and an increment peaking near 370 GHz — the shape is fixed while its amplitude scales with the single parameter y

Redshift independence — the last piece of the picture — follows because both the incoming CMB intensity and the distortion imposed on it scale with the CMB temperature at scattering, and both redshift by the identical factor on the way to us. Two clusters of equal gas content at z = 0.1 and z = 1 produce the same SZ brightness — a startling contrast with X-ray or optical images, which fade steeply with distance, and the property that turns the SZ effect into a cosmological survey tool.

The core equations

The whole subject descends from one evolution equation. The Kompaneets equation is the Fokker-Planck limit of the Compton collision integral: repeated small energy exchanges diffuse the occupation number n across frequency, with a drift toward equilibrium with the electrons. In the dimensionless energy x = hν/k_BTₑ it reads:

\frac{1}{c}\frac{\partial n}{\partial t}=\frac{k_BT_e}{m_ec^2}\,\frac{n_e\sigma_T}{x^2}\,\frac{\partial}{\partial x}\left[x^4\left(\frac{\partial n}{\partial x}+n+n^2\right)\right] \qquad (1)

where: n — photon occupation number (dimensionless); x = hν/k_BTₑ — photon energy in units of the electron thermal energy; Tₑ — electron temperature; nₑ — electron number density; σ_T — Thomson cross section, 6.65×10⁻²⁵ cm²; mₑc² — electron rest energy, 511 keV. Reading it: k_BTₑ/mₑc² is the energy exchanged per scatter, nₑσ_T sets the rate, and the bracket holds diffusion (∂n/∂x), recoil drift (+n), and stimulated scattering (+n²). For a thin, hot cloud the full equation is overkill; what survives is the single amplitude multiplying its linearized action. The Compton y-parameter factorizes into energy-per-scatter times number-of-scatters along the line of sight:

y=\int \frac{k_BT_e}{m_ec^2}\,n_e\sigma_T\,\mathrm{d}\ell = \left\langle\frac{k_BT_e}{m_ec^2}\right\rangle\tau \qquad (2)

where: y — dimensionless Comptonization parameter (amplitude of the distortion); dℓ — path length along the line of sight; τ = ∫ nₑσ_T dℓ — Thomson optical depth. Reading it: y is energy-gain-per-scatter times number-of-scatters, and it alone controls how strong the distortion is. Linearizing the Kompaneets operator on the CMB Planck spectrum (valid because y ≪ 1) gives the observable brightness-temperature change, whose frequency dependence is the universal tSZ spectral function g(x):

\frac{\Delta T}{T_{\rm CMB}}=y\,g(x),\qquad g(x)=x\coth\!\left(\frac{x}{2}\right)-4 \qquad (3)

where: x = hν/k_BT_CMB — dimensionless frequency relative to the CMB temperature (note this uses T_CMB, not Tₑ); g(x) — tSZ spectral function, with x coth(x/2) = x(eˣ+1)/(eˣ−1); ΔT — brightness-temperature change. Reading it: the shape g(x) is fixed once and for all; the cluster's physics enters only through the scalar y. The frequency where g(x) changes sign is the observational anchor of the whole effect. Setting g(x) = 0 gives the crossover, or null:

g(x)=0\ \Longrightarrow\ x_0\simeq3.830\ \Longrightarrow\ \nu_0\simeq217\ \mathrm{GHz} \qquad (4)

where: x₀ — the root of x coth(x/2) = 4; ν₀ = x₀ k_BT_CMB/h — the crossover frequency where decrement turns into increment. Reading it: g < 0 below 217 GHz (decrement), g > 0 above (increment), and exactly at 217 GHz the thermal effect vanishes — where the kinematic effect is best measured. Experiments observe intensity, not temperature, so the distortion is pushed through the Planck-to-intensity Jacobian. This gives the bounded intensity distortion plotted in the second figure:

\frac{\Delta I_\nu}{I_0}=\frac{x^4e^x}{(e^x-1)^2}\,y\,g(x),\qquad I_0=\frac{2(k_BT_{\rm CMB})^3}{(hc)^2} \qquad (5)

where: ΔI_ν — change in specific intensity; I₀ — characteristic Planck intensity scale. Reading it: the Jacobian tames the divergence of g(x) at large x, so ΔI_ν dips to a decrement peak near 128 GHz, crosses zero at 217 GHz, peaks in increment near 370 GHz, then returns to zero. A second, distinct distortion arises not from the electrons' random thermal motion but from the cluster's coherent bulk motion through the CMB rest frame — a plain Doppler shift:

\frac{\Delta T_{\rm kSZ}}{T_{\rm CMB}}=-\tau\,\frac{v_r}{c} \qquad (6)

where: v_r — line-of-sight peculiar velocity (positive for a receding cluster); τ — Thomson optical depth. Reading it: unlike the thermal effect this has a blackbody spectrum — a pure temperature shift with no null — so it survives at 217 GHz where the thermal signal vanishes. Finally, the two microphysical numbers that seed everything above — the mean fractional energy gain per scatter, and the net energy the radiation gains overall:

\left\langle\frac{\Delta\nu}{\nu}\right\rangle=\frac{4k_BT_e-h\nu}{m_ec^2}\ \xrightarrow{h\nu\ll k_BT_e}\ \frac{4k_BT_e}{m_ec^2},\qquad \frac{\Delta E}{E}\approx 4y\quad(y\ll1) \qquad (7)

where: 4k_BTₑ — the Doppler up-scattering (gain) term; hν — the recoil (loss) term; ΔE/E — fractional increase in photon energy density. Reading it: gain beats recoil whenever the electrons are hotter than the photons, and over the cloud the radiation's energy rises by 4y.

Derivation AMean energy gain per Compton scatter off a thermal electron gas

The goal of this derivation is to obtain the single most important microphysical number in the subject — the mean fractional frequency shift ⟨Δν/ν⟩ = 4k_BTₑ/mₑc² per scatter — from two Lorentz boosts and the statistics of a Maxwellian electron gas. It seeds the drift term of the Kompaneets equation and the whole y-parameter, so its factor of 4 and the competing recoil term underpin everything downstream. We take a photon of energy hν ≪ mₑc² off an electron of speed β = v/c, follow it into the electron frame, scatter it elastically, and boost it back.

Step 1. Boost the photon into the electron's rest frame. With θ the angle between photon and electron velocity, the relativistic Doppler formula gives ν′ = νγ(1 − β cos θ), γ = (1 − β²)⁻¹ᐟ². Why: the scattering rule is simplest where the electron is at rest, so we transport the photon there first.

Step 2. Scatter elastically in that frame. While hν′ ≪ mₑc² the scattering is Thomson and elastic: ν″ ≈ ν′, with only an angular redistribution. Why: the electron recoil is negligible when the photon energy in its rest frame is far below 511 keV — exactly the CMB-in-a-cluster regime.

Step 3. Boost back to the lab. The inverse Doppler transformation with the aberrated outgoing angle θ″ gives ν_out = ν′γ(1 + β cos θ″); composing the two boosts leaves two factors of γ and a pair of angle-dependent Doppler factors. Why: the net shift is a competition between the boost in and the boost out, and only the angular average reveals which way it tips.

Step 4. Average over scattering angles and electron directions. Expanding to second order in β and averaging over an isotropic electron gas and the Thomson phase function yields ⟨Δν/ν⟩ = (4/3)⟨β²⟩ − hν/mₑc²: a Doppler gain and a recoil loss. Why: to first order in β the shifts average to zero, so the physics lives entirely in the β² term, which is why the electron temperature enters through ⟨β²⟩.

Step 5. Insert the thermal average. Equipartition gives ½mₑ⟨v²⟩ = (3/2)k_BTₑ, so ⟨β²⟩ = 3k_BTₑ/mₑc²; then (4/3)⟨β²⟩ = 4k_BTₑ/mₑc², and dropping recoil for soft photons leaves the boxed result. Why: the factor 4 is 4/3 from kinematics times 3 from the three translational degrees of freedom.

\boxed{\ \left\langle\frac{\Delta\nu}{\nu}\right\rangle=\frac{4k_BT_e}{m_ec^2}\ }\qquad(\text{non-relativistic thermal electrons})

Note: the coefficient 4 is standard, and the recoil term −hν/mₑc² correctly makes the net shift vanish when the photon energy equals the electron thermal energy — the equilibrium condition.

Go deeper · Derivation B · 2 minThe thermal SZ spectral function from the Kompaneets equation

Derivation BThe thermal SZ spectral function from the Kompaneets equation

The aim here is to extract the observable tSZ shape g(x) = x coth(x/2) − 4 and its 217 GHz null directly from the Kompaneets equation, showing the universal spectrum is not an empirical fit but a first-order consequence of the collision operator acting on a Planck spectrum. Because a cluster is optically thin (y ≪ 1) we need only the equation's action to first order in y on the unperturbed CMB occupation n₀ = (eˣ − 1)⁻¹. The one subtlety is a change of variable: the Kompaneets x = hν/k_BTₑ, but the observed distortion is expressed in x = hν/k_BT_CMB, with the temperature ratio absorbed into y.

Step 1. Integrate the Kompaneets operator once over the thin cloud. Since the distortion is small, Δn = y·(1/x²) ∂/∂x [ x⁴ (∂n₀/∂x + n₀ + n₀²) ], with the electron temperature pulled into y and x now denoting hν/k_BT_CMB. Why: over a thin cluster the operator acts just once, and all the temperature and density information collapses into the single prefactor y.

Step 2. Use the Planck identity. Differentiating n₀ gives ∂n₀/∂x = −eˣ/(eˣ − 1)² = −(n₀ + n₀²), so the bracket would vanish for a Planck spectrum at Tₑ — the fixed point — leaving only the temperature-mismatch piece. Why: only the deviation from a Planck spectrum at Tₑ drives a distortion.

Step 3. Carry out the derivative. Differentiating x⁴(∂n₀/∂x + n₀ + n₀²) with the identity gives Δn/n₀ = y·(x eˣ/(eˣ − 1))·[x coth(x/2) − 4]. Why: the derivative brings down the x⁴ weighting of high-frequency photons, tipping the balance toward net up-scattering and the sign change.

Step 4. Convert to brightness temperature. Using ΔI/I = Δn/n₀ and the Planck-to-temperature Jacobian, the occupation factor cancels: ΔT/T = (Δn/n₀)(eˣ − 1)/(x eˣ) = y [x coth(x/2) − 4]. Why: in brightness temperature the spectral function stands alone as g(x).

Step 5. Locate the null. Setting g(x) = 0 means x coth(x/2) = 4, root x₀ ≈ 3.830; with T_CMB = 2.725 K, ν₀ = x₀ k_BT_CMB/h ≈ 217 GHz, g < 0 below (decrement) and g > 0 above (increment). Why: the null is guaranteed zero regardless of the cluster — the ideal window for the kinematic effect.

\boxed{\ \frac{\Delta T}{T_{\rm CMB}}=y\left[x\coth\!\left(\frac{x}{2}\right)-4\right],\qquad x_0\simeq3.830,\ \ \nu_0\simeq217\ \mathrm{GHz}\ }

Note: the root x₀ = 3.8300 and ν₀ = 217.5 GHz follow directly from solving x coth(x/2) = 4 with T_CMB = 2.725 K.

Go deeper · Derivation C · 2 minRedshift independence of the SZ surface brightness

Derivation CRedshift independence of the SZ surface brightness

The purpose of this argument is to show cleanly why the SZ brightness ΔT/T carries no explicit redshift dimming while the X-ray surface brightness falls as (1+z)⁻⁴ — the single fact that makes SZ surveys mass-limited to arbitrarily high redshift and turns the SZ-plus-X-ray combination into a distance ladder. We compare, term by term, how each observable scales with redshift, distinguishing quantities intrinsic to the gas from those imposed by cosmological geometry.

Step 1. Identify what is scattered. The photons are the local CMB at the cluster's redshift, T(z) = T₀(1+z), and the scattering imposes a fractional distortion ΔT/T = y g(x) depending only on y and dimensionless frequency, not on absolute temperature. Why: the effect reshapes a bath present at every redshift, so the brightness it modifies comes for free with the universe.

Step 2. Note that y is intrinsic to the gas. In y = ∫ (k_BTₑ/mₑc²) nₑσ_T dℓ every factor is a local plasma property along the line of sight; no distance or luminosity factor appears. Why: an integral of local properties cannot know how far away the cluster is.

Step 3. Follow the ratio to the observer. Both the CMB and its imprinted distortion redshift by the identical factor (1+z)⁻¹, so their ratio ΔT/T is preserved exactly. Why: dividing two quantities that dim by the same factor cancels the dimming — the reason the SZ decrement is redshift-independent.

Step 4. Contrast with X-ray emission. Bremsstrahlung from the same gas has S_X ∝ ∫ nₑ² Λ(Tₑ) dℓ × (1+z)⁻⁴. Why: X-rays are a genuine luminosity, so they suffer the full Tolman surface-brightness dimming a fractional distortion escapes.

Step 5. Read off the consequences. SZ surveys are mass-limited to high z, and the different density scalings — SZ ∝ nₑTₑ versus X-ray ∝ nₑ²Tₑ^(1/2) — break the density-length degeneracy, yielding the physical depth, the distance d_A, and hence H₀. Why: two observables with different powers of nₑ solve for both density and size, converting an angle into a distance.

\boxed{\ \frac{\Delta T}{T}\ \text{independent of }z,\qquad S_X\propto(1+z)^{-4}\ }

Note: the (1+z)⁻⁴ dimming is the Tolman surface-brightness law; the SZ fractional distortion evades it.

Limiting cases

The behaviour is organized by two ratios: how photon and electron energies compare (k_BTₑ versus hν), and how many scatterings occur (y, or the optical depth τ). The four regimes below span the important cases from a single cluster scatter to full thermalization in the early universe.

Case	Result	Meaning
Single-scatter, non-relativistic (k_BTₑ ≪ mₑc², τ ≪ 1)	Δν/ν ≃ 4k_BTₑ/mₑc²	Mean fractional energy gain per Thomson scatter off a thermal gas; the regime of the cluster SZ effect.
Weak Comptonization (y ≪ 1, thermal SZ)	ΔT/T_CMB = y[x coth(x/2) − 4]	Linear distortion; the spectral function vanishes at x ≃ 3.83, i.e. ν ≃ 217 GHz, with a decrement below and increment above.
Saturated Comptonization (y ≫ 1)	n(x) → 1/(eˣ⁺ᵘ − 1), x = hν/k_BTₑ	Many scatters drive photons to a Bose-Einstein spectrum with chemical potential μ, the equilibrium fixed point of the Kompaneets equation.
Kinematic limit (bulk velocity v_r)	ΔT/T_CMB = −τ v_r/c	A pure Doppler shift with a blackbody spectrum; receding (v_r > 0) gives a decrement, and the distinct spectrum separates it from the tSZ signal.

Common trapThe SZ decrement is a cluster cooling the CMB by absorption

It is neither. The SZ decrement is not absorption and does not lower the CMB temperature: Compton scattering conserves photon number, so no photons are destroyed and no energy is removed — the CMB actually gains energy (ΔE/E ≈ 4y). What happens is a redistribution in frequency: photons move from below 217 GHz to above it, so low-frequency intensity falls (the decrement) while high-frequency intensity rises (the increment), total conserved. Two corollaries follow. First, never model the SZ signal as an extinction e⁻ᵗ — it has the wrong sign at high frequency and the wrong physics everywhere. Second, the tSZ shape g(x) is universal and carries no cluster information; only the amplitude y does. Confusing the fixed shape with the variable amplitude is the most common analysis error.

Worked examples

1 · A rich cluster: y-parameter, decrement, and increment (fully worked)

A massive cluster has an electron temperature k_BTₑ = 8 keV and a Thomson optical depth τ = 0.01 through its centre. Find the mean energy gain per scatter, the Compton y-parameter, the central Rayleigh-Jeans temperature decrement in mK, and the fractional energy it adds to the CMB.

Step 1. Energy gain per scatter. From equation (7), ⟨Δν/ν⟩ = 4k_BTₑ/mₑc² = 4 × (8 keV)/(511 keV) = 0.0626, about a 6% up-shift per scatter — small, which is exactly why the linearized (y ≪ 1) treatment holds.

Step 2. The y-parameter. By equation (2), y = (k_BTₑ/mₑc²) τ = (8/511)(0.01) = 1.57×10⁻⁴ — one quarter of the energy gain per scatter times the number of scatters.

Step 3. Rayleigh-Jeans decrement. As x → 0, g(x) → 2 − 4 = −2, so ΔT/T_CMB → −2y = −3.13×10⁻⁴. Against the T_CMB = 2.725 K background, ΔT = −2y × 2.725 K = −0.85 mK — a textbook rich-cluster decrement, the depth of the "cold spot" a low-frequency radio telescope sees toward the cluster.

Step 4. Energy added to the CMB. The boost to the radiation energy density is ΔE/E ≈ 4y = 6.3×10⁻⁴ — minuscule per cluster, but summed over all the hot gas in the universe it sets the cosmic y-distortion budget FIRAS constrained. Summary: ⟨Δν/ν⟩ = 0.063, y = 1.6×10⁻⁴, ΔT_RJ = −0.85 mK, ΔE/E = 6.3×10⁻⁴.

2 · Kinematic versus thermal SZ at the null

The same cluster (τ = 0.01) moves toward us with a line-of-sight peculiar velocity v_r = −300 km/s. Estimate its kinematic SZ temperature shift, and compare its size to the thermal signal at the 217 GHz null.

Step 1. Kinematic amplitude. From equation (6), ΔT_kSZ/T_CMB = −τ v_r/c = −(0.01)(−300/299792) = +1.0×10⁻⁵, so ΔT_kSZ = +27 μK. The positive sign (an increment) reflects an approaching cluster blueshifting the scattered CMB.

Step 2. Thermal signal at 217 GHz. By construction g(x₀) = 0 there, so the thermal contribution is exactly zero to first order in y, while the kinematic effect, having a blackbody spectrum with no null, is undiminished.

Step 3. Interpretation. At a generic frequency the thermal decrement (∼0.85 mK) dwarfs the ∼27 μK kinematic signal by a factor of ∼30, which is why isolating kSZ demands observations at or near 217 GHz where the thermal effect switches off. That frequency separation — thermal null exactly where kinematic survives — is the entire observational strategy behind kSZ. kSZ ≈ +27 μK, thermal = 0 at 217 GHz, so the null is the place to measure peculiar velocities.

Physics of the portrait

The effect carries two names because it was born from a collaboration, but the theory of how hot electrons reshape a photon spectrum belongs to a Moscow lineage running from Kompaneets through Zel'dovich to his young collaborator Rashid Sunyaev.

Origins · 1 minRashid Sunyaev and how this was discovered

Rashid Sunyaev

b. 1943, Tashkent · relativistic astrophysics and the cosmic microwave background

Rashid Sunyaev came to Moscow from Uzbekistan and joined the extraordinary group that Yakov Zel'dovich — an architect of the Soviet nuclear programme turned cosmologist — had assembled at the Institute of Applied Mathematics. In a pair of papers around 1970–1972, the two turned Kompaneets's 1957 equation for photon-electron energy exchange onto the newly discovered CMB. They realized that the hot gas filling galaxy clusters must inverse-Compton scatter passing CMB photons, and computed the exact spectral fingerprint: a decrement on the Rayleigh-Jeans side, a null near 217 GHz, an increment beyond — the thermal Sunyaev-Zel'dovich effect. In the same work they identified the kinematic effect from a cluster's bulk motion and, looking backward in time, showed how early-universe energy release would leave μ-type and y-type distortions on the CMB spectrum itself. For two decades these were predictions awaiting instruments able to see a millikelvin dip against a 2.7 K sky; the first secure cluster detections came in the 1980s, and the harvest of thousands of clusters from Planck, SPT, and ACT only in the 21st century. Sunyaev went on to lead high-energy astrophysics with the Granat and INTEGRAL missions and the Max Planck Institute for Astrophysics, sharing the 2008 Crafoord Prize for this work. The through-line of his career is the single idea in this article: scattering off hot electrons writes a legible, quantitative signature onto radiation, and reading it lets us weigh clusters, clock cosmic flows, and audit the thermal history of the universe.

History

1957

— Kompaneets derives the Fokker-Planck equation for photon-electron energy exchange

1970

— Sunyaev and Zel'dovich apply Comptonization to spectral distortions of the CMB

1972

— the thermal and kinematic Sunyaev-Zel'dovich effects are predicted in full

1984

— first secure detection of the SZ decrement toward a galaxy cluster

1996

— COBE/FIRAS bounds the sky-averaged distortions to y < 1.5×10⁻⁵ and μ < 9×10⁻⁵

2011

— Planck, SPT, and ACT begin delivering large SZ-selected cluster catalogs

Going further

Texts. Rybicki & Lightman, Radiative Processes in Astrophysics, ch. 7 (Compton scattering, the Kompaneets equation, the 4k_BTₑ/mₑc² gain) · Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., ch. 9 and 27 (radiative processes; clusters) · Dodelson, Modern Cosmology, 2nd ed., ch. 4 and 9 (Compton/Boltzmann, secondary anisotropies) · Mo, van den Bosch & White, Galaxy Formation and Evolution, §8.9 (intracluster medium and the SZ effect) · Ryden, Introduction to Cosmology, 2nd ed., ch. 8–9 · Birkinshaw (1999), Physics Reports 310, 97 (the definitive SZ review). Key numerical anchors: x coth(x/2) = 4 gives x₀ = 3.8300, and ν₀ = x₀ k_BT_CMB/h = 217.5 GHz with T_CMB = 2.725 K, σ_T = 6.65×10⁻²⁵ cm², mₑc² = 511 keV; the worked-example decrement −0.85 mK follows from ΔT = −2yT_CMB with y = 1.57×10⁻⁴. FIRAS bounds |y| < 1.5×10⁻⁵ and |μ| < 9×10⁻⁵.

Prerequisites. Blackbody Radiation and the Planck Law · Specific Intensity and Radiative Flux · The Equation of Radiative Transfer · Optical Depth · Stefan-Boltzmann and Wien Displacement Laws. Leads to. The Boltzmann Excitation Equation · Local Thermodynamic Equilibrium.

Ask the tutor: Ask me to walk through the Kompaneets bracket term-by-term, recompute the 217 GHz null from scratch, or work a full SZ-plus-X-ray distance estimate for a cluster of your choosing.

Is the SZ decrement the cluster casting a shadow on the CMB?

No — a shadow would mean the cluster blocks or absorbs CMB photons, and it does neither. The gas is optically thin (τ ∼ 0.01), so almost every photon passes through untouched, and those that interact are scattered, not absorbed: photon number is conserved. The decrement is a deficit only at low frequencies, created because scattering nudges photons from below 217 GHz to above it; at high frequencies you see a surplus. So it is a frequency redistribution, not a shadow — and unlike a shadow it changes sign at 217 GHz.

Why is the tSZ effect independent of redshift when almost nothing else in astronomy is?

Because it is a fractional distortion of a bath that exists at every redshift. The SZ effect modifies the CMB, and ΔT/T = y g(x) depends only on the local gas (through y) and dimensionless frequency — no distance factor appears. Both the CMB and the distortion imprinted on it dim by the identical (1+z) factor on the way to us, so their ratio is preserved. Contrast X-ray or optical emission, a genuine luminosity subject to the (1+z)⁻⁴ Tolman dimming. So a fixed-mass cluster gives the same SZ brightness at z = 0.1 or z = 1.5, making SZ surveys mass-limited to high redshift.

What is the difference between the y-parameter and the optical depth τ?

The optical depth τ = ∫ nₑσ_T dℓ counts how many times a photon scatters on average — pure geometry and density. The y-parameter multiplies that by the energy exchanged per scatter: y = ∫ (k_BTₑ/mₑc²) nₑσ_T dℓ = ⟨k_BTₑ/mₑc²⟩ τ. So τ is how often photons interact, y is how much energy they gain overall. For a typical cluster τ ∼ 0.01 and k_BTₑ/mₑc² ∼ 0.016, so y ∼ 10⁻⁴. The thermal SZ amplitude is set by y; the kinematic amplitude, being a single Doppler shift, is set by τ alone (times v_r/c).

Why does the null sit exactly at 217 GHz, and why does that number matter?

The null is where g(x) = x coth(x/2) − 4 crosses zero, at x₀ = 3.83; with x = hν/k_BT_CMB and T_CMB = 2.725 K this is ν₀ = 217 GHz — the frequency where up-scattering removes as many photons as it adds, so the thermal signal vanishes. It matters twice over. First, it is a fixed prediction independent of any cluster property, so a signal that changes sign at 217 GHz is a decisive confirmation of tSZ over a foreground. Second, because the thermal effect is zero there, 217 GHz is the cleanest window for the kinematic effect, which has no null.

Can the SZ effect ever heat or cool the CMB overall?

It always adds energy to the radiation field — never removes it — because cluster electrons are far hotter than the CMB, so inverse Compton scattering transfers energy from matter to photons, with net gain ΔE/E ≈ 4y > 0. It never cools the CMB; the low-frequency decrement is a local deficit in frequency space, more than offset by the high-frequency increment. Cooling the radiation would require photons hotter than the electrons — the opposite regime, relevant to the recoil term in the early universe. For clusters the sign is fixed: energy flows uphill into the CMB.

What sets the tiny μ- and y-distortions of the early-universe CMB, and why do we care?

Any energy released into the primordial plasma after the thermalizing processes (double Compton, bremsstrahlung) freeze out leaves a permanent spectral scar. Injection at high redshift (z ∼ 10⁵–10⁶), when Comptonization is efficient but photon-number-changing processes are not, gives a Bose-Einstein μ-distortion; injection at lower z gives a y-distortion like the cluster effect. The standard model predicts small nonzero values from Silk damping and reionization — of order 10⁻⁸ in μ and a few 10⁻⁶ in y. Measuring them would test inflation on otherwise inaccessible scales and constrain decaying dark matter, which is why proposed spectrometers aim far below the FIRAS ceiling.