The Reverse Shock: A Blast Wave That Turns Back Inward

Why a blast wave turns around

Within days of a core collapse or a thermonuclear detonation, a star's debris settles into free expansion. Roughly 10⁵¹ erg of kinetic energy is shared among a few solar masses of gas, so the bulk moves at v ≈ √(2E/M) ≈ 4,000 km/s while the thin outermost layers reach ~10⁴ km/s. Nothing outside is massive enough to matter, so every element coasts and the ejecta settle into homologous flow, v = r/t.

That free ride ends when the shell has swept up an ambient mass comparable to its own. Setting (4/3)πR³ρ₀ = M_ej for M_ej = 5 M_☉ and a typical interstellar density n₀ ≈ 1 cm⁻³ (ρ₀ ≈ 2×10⁻²⁴ g cm⁻³) gives R ≈ 3.4 pc — reached in roughly 600–700 years at 5,000 km/s. From then on the leading edge genuinely decelerates.

Deceleration is the whole story. Gas piling up behind the forward shock is compressed to a pressure of roughly (3/4)ρ₀v², while the ejecta right behind it are still cold, fast and almost pressureless. That pressure cannot simply sit there: it drives a second shock down the density gradient into the ejecta. In the frame comoving with the expanding debris this shock moves inward — hence reverse. In our frame it still moves outward at first, just slower than the ejecta streaming past it; only later does its radius actually shrink. Christopher McKee set this out quantitatively in 1974.

The double-shock structure and how fast it grows

What emerges is a sandwich. From outside in: undisturbed interstellar gas, the forward shock, hot shocked ambient gas, the contact discontinuity, hot shocked ejecta, the reverse shock, and finally the cold freely expanding ejecta it has not yet reached. Pressure is continuous across the contact discontinuity; density, temperature and composition are not.

Roger Chevalier's 1982 self-similar solution gives the growth law. If the outer ejecta follow ρ_ej ∝ r⁻ⁿ and the ambient medium ρ ∝ r^−s, the shell radius obeys

• R ∝ t^{(n−3)/(n−s)}

Type Ia ejecta have n ≈ 7 and usually expand into uniform gas (s = 0), giving R ∝ t^4/7. Core-collapse ejecta are steeper, n ≈ 9–12, and often run into the progenitor's r⁻² wind (s = 2), giving R ∝ t^6/7. Measured expansion parameters bracket this: Cassiopeia A gives m = d ln R/d ln t ≈ 0.6–0.7, Tycho (SN 1572) about 0.5.

The reverse shock is not permanent. Truelove & McKee's 1999 unified solutions have it converging on the centre after roughly twice the characteristic time t_ch ≈ 420 E₅₁^−1/2(M_ej/M_☉)^5/6n₀^−1/3 yr — about 1,000 years for a 1.4 M_☉ Type Ia in ordinary interstellar gas, several thousand for a massive core-collapse remnant, and far less if a dense circumstellar wind is present. Once the ejecta are fully processed the remnant forgets its initial mass and enters the Sedov-Taylor phase, R = 1.15(Et²/ρ₀)^1/5 ∝ t^2/5: about 12 pc and 490 km/s at 10,000 years for E = 10⁵¹ erg and n₀ = 1 cm⁻³. Radiative cooling takes over near 200 km/s, after some tens of thousands of years.

How hot it gets — and why the electrons lag behind

The Rankine-Hugoniot jump conditions for a strong adiabatic shock (γ = 5/3) compress the gas by exactly a factor of 4 and set the mean post-shock temperature to kT = (3/16)μm_Hv_s². For cosmic-abundance gas (μ ≈ 0.6) that is T ≈ 1.4×10⁷ K per (1,000 km/s)², so Cas A's forward shock at ~4,800 km/s should reach ~3×10⁸ K.

Behind the reverse shock two things change. The relevant speed is the relative speed between ejecta and shock, typically 1,500–3,000 km/s rather than 10⁴. And the gas is not hydrogen: each species is heated in proportion to its own mass, so a silicon nucleus (A = 28) crossing a 2,000 km/s shock picks up ~220 keV, about 2.5×10⁹ K, and iron more still.

Yet X-ray spectra of shocked ejecta indicate electron temperatures of only ~1–3 keV, i.e. ~10⁷ K. No contradiction: collisionless shocks dump most of their energy into ions, and Coulomb collisions equilibrate the electrons only over a time longer than the remnant's age. The same slowness afflicts the ionization state, set by the product n_et — the ionization age — and equilibrium needs n_et ≳ 10¹² cm⁻³ s. Young remnants sit near 10¹¹, so they are strongly underionized: the plasma is far hotter than its charge states suggest. Every temperature and abundance quoted for a young remnant therefore comes from a non-equilibrium ionization model, not a simple thermal fit.

Reading the star's onion shells

This is the reverse shock's scientific payload. A massive star before it dies is layered like an onion — iron core, then silicon and sulfur, then oxygen, neon and magnesium, then carbon, helium and any remaining hydrogen. Those layers are cold, transparent and invisible for centuries; the reverse shock is the only agent that heats them until they radiate.

What it lights up are the He-like and H-like lines of the burning products: Si XIII at 1.86 keV, S XV at 2.45 keV, Ar, Ca, and the Fe-K complex near 6.4–6.7 keV. A million-second Chandra exposure of Cassiopeia A resolves them spatially, and because the shell is expanding, Doppler shifts of a few thousand km/s turn the image into a three-dimensional map (DeLaney et al. 2010; Milisavljevic & Fesen 2013) — an exploded star's chemical anatomy, ring by ring.

It also shows that anatomy scrambled: in parts of Cas A, iron-rich ejecta lie outside the silicon and sulfur they were born beneath, overturned by the neutrino-heated convection that revived the explosion. NuSTAR's 2014 map of ⁴⁴Ti at 68 and 78 keV (Grefenstette et al., Nature) supplied the check: the ~1.5×10⁻⁴ M_☉ of freshly forged titanium is not where the X-ray iron is, because X-ray iron traces only the shocked fraction. Several tenths of a solar mass of Cas A's ejecta remain cold and unshocked inside, seen instead in infrared [Si II] and [O IV] with Spitzer and, since 2023, JWST. XRISM's Resolve microcalorimeter (launched September 2023, ~5 eV) now measures those line widths and bulk motions directly.

SN 1987A gives the close-up. Its reverse shock, at a radius of a few ×10¹⁷ cm — well inside the equatorial ring — is visible because neutral hydrogen atoms in the coasting ejecta cross it at ~12,000 km/s and are collisionally excited before being ionized, producing the broad Lyman-α and Hα emission mapped with HST/STIS from 1997 onwards.

Fingers: the unstable seam between the shocks

The contact discontinuity separates dense, decelerating shocked ejecta from lighter shocked ambient gas, with the pressure gradient pointing the wrong way. That is exactly the Rayleigh-Taylor configuration — heavy fluid supported by light — and simulations by Chevalier, Blondin & Emmering (1992) showed the interface growing fingers that push outward toward the forward shock while ambient gas sinks between them.

Those fingers are not theoretical. They are the knotted, filamentary structure of Cassiopeia A and Tycho, resolved by Chandra and, at higher contrast, in JWST's 2023 infrared views of Cas A. They mix ejecta with ambient gas, blur the compositional layering, and can carry dense clumps clear through the shocked shell.

The gap between forward shock and contact discontinuity has become a diagnostic in its own right. Pure hydrodynamics predicts a certain separation; Chandra measurements of Tycho by Warren et al. (2005) found the shocked shell pressed unexpectedly close to the blast wave. The favoured explanation is efficient cosmic-ray acceleration at the forward shock: relativistic particles have an effective adiabatic index near 4/3 rather than 5/3, so the shock compresses by more than 4, the shell thins, and the two surfaces crowd together. The reverse shock's position, in effect, weighs how much of the explosion is spent on cosmic rays rather than heat.

The dust the reverse shock destroys

Supernovae are prolific dust factories. Herschel and ALMA found ~0.4–0.7 M_☉ of cold dust in SN 1987A (Matsuura et al. 2011, 2015) and ~0.3–0.5 M_☉ in Cassiopeia A (De Looze et al. 2017). If most of it survived, supernovae would comfortably explain the ~10⁸ M_☉ of dust seen in quasar hosts less than a billion years after the Big Bang, where there has been no time for the slow AGB-star route.

The reverse shock is the obstacle. It sweeps through the very ejecta where the grains condensed and destroys them two ways: thermal sputtering by ~10⁷–10⁸ K ions, and kinetic sputtering as grains, decoupled by their inertia from the shocked gas, plough through it at high relative velocity until gyration in the compressed post-shock magnetic field and gas drag finally couple them — betatron acceleration in that field can raise their speed, and so their erosion, first. Models (Bianchi & Schneider 2007; Nozawa et al. 2007; Micelotta et al. 2016; Slavin et al. 2020) span a wide range, but for n₀ ≈ 1 cm⁻³ typically only ~10–20% of the dust mass survives, and denser surroundings can push that to a few per cent. Large grains (≳0.1 μm) and grains locked in dense clumps do best.

Every measured dust mass is taken before the reverse shock has finished its pass, so what enters a galaxy's dust budget is the survival fraction, not the formation yield — and that fraction is uncertain by an order of magnitude.

Look-alikes, misconceptions and open questions

It is not a reflection. The reverse shock is not the blast wave bouncing off a wall; it is a new shock continuously launched by the pressure of the decelerating shell. Nor does it move inward at first — for centuries it travels outward in our frame and is reverse only relative to the ejecta streaming past it. When it finally converges at the centre it reflects and rings back outward, reheating the interior.

Several near neighbours get confused with it:

• Shock breakout is the explosion shock erupting through the stellar surface hours after collapse — outward, and long over before the reverse shock exists.
• A stellar-wind termination shock also decelerates outflowing material, but it is a steady structure in a continuous wind, not a transient in a ballistic ejecta shell.
• A pulsar wind nebula (the Crab) is driven from inside by a neutron star's relativistic wind and has its own termination shock; it is not the supernova's reverse shock.
• Gamma-ray burst reverse shocks are the same physics at Lorentz factors of hundreds — the bright optical flash of GRB 990123 is the textbook case — but they last seconds, not centuries.

Still contested: how much unshocked ejecta hides in remnant interiors, and therefore what the true nucleosynthetic yields are; how collisionless shocks partition energy among electrons, ions and cosmic rays, which fixes the electron-to-ion temperature ratio every abundance measurement rests on; whether clumpy, aspherical ejecta shield enough dust to rescue supernovae as the early universe's dust source; and how badly three-dimensional turbulence undermines the one-dimensional solutions the field still leans on.

The three surfaces inside a young supernova remnant, and how to tell them apart

Property	Forward shock	Reverse shock	Contact discontinuity
Gas it processes	Ambient ISM or the progenitor's wind (H/He, cosmic abundances)	The star's own ejecta (O, Ne, Mg, Si, S, Ar, Ca, Fe)	Neither — it is the seam between the two shocked gases
Motion in the observer's frame	Always outward; ~4,800 km/s in Cas A	Outward at first, then inward; ~2,000 km/s inward in Cas A's west today	Outward, sandwiched between the two shocks
Motion in the ejecta frame	Far ahead and receding	Always inward — hence 'reverse'	Comoving with the shocked gas
Post-shock temperature	Ions ~10^8-10^9 K; electrons far cooler	Ions ~10^9-10^10 K for heavy nuclei if heating is mass-proportional; electrons ~1-3 keV	Pressure is continuous; density and temperature jump
Observational signature	Thin synchrotron rims (SN 1006, Cas A), faint thermal continuum	Bright He-like Si, S, Ar, Ca and Fe-K emission lines	Rayleigh-Taylor fingers and knots
Fate	Survives into the Sedov and radiative phases, for ~10^5 yr	Ends when it converges on the centre, after ~10^3 yr	Shredded by Rayleigh-Taylor mixing