Supernova Remnant

Why the remnant outlives the explosion

A supernova lasts seconds. The optical fireball is bright for months, the radioactive afterglow for years. But the remnant — the part the galaxy actually feels — keeps growing for a hundred millennia. The reason is simple: the explosion deposits roughly 10⁵¹ erg of mechanical energy into the surrounding gas, and that energy cannot be radiated away quickly. It takes the form of bulk motion of expanding ejecta and, later, the swept-up shell of interstellar gas behind a shock front. Until the shock slows enough for radiative cooling to win, that energy stays as kinetic and thermal, and the structure keeps plowing outward.

The numbers are blunt. A typical core-collapse SN ejects ~10 M☉ at v_ej ~ 10⁴ km/s, carrying KE = (1/2)(10 M☉)(10⁴ km/s)² ≈ 10⁵¹ erg. The Milky Way's interstellar medium has a mean density of n ~ 1 cm⁻³, so a remnant has to sweep through about 200 pc³ before it gathers enough mass for the shock to slow significantly. That alone takes ~10,000 years. The full sequence — from explosion to merger — spans ~10⁵ years and ~30 pc, comparable in volume to a small molecular cloud.

Mechanism — the shock structure of a young SNR

A young supernova remnant has a layered structure. Outermost is a forward shock, where the supersonic ejecta drive into the ambient ISM. The forward shock heats the swept-up gas to 10⁷ – 10⁸ K and compresses it by a factor of four (the strong-shock Rankine-Hugoniot jump). Just inside lies a contact discontinuity separating ejecta from shocked ISM. Inside that, a reverse shock propagates backward (in the ejecta's rest frame) into the ejecta, heating them and lighting up their characteristic emission lines — silicon, sulfur, iron — in X-rays. Inside the reverse shock is the as-yet-uncompressed, freely-coasting ejecta.

    [ free ejecta ] [ reverse-shocked ejecta ] | [ shocked ISM ] [ undisturbed ISM ]
                                              CD            FS
        cool                         hot                hot                cool
                                  10⁷–10⁸ K           10⁷ K

The forward and reverse shocks are both collisionless — the gas is so tenuous that direct particle-particle collisions are negligible on the relevant length scale. Instead, plasma instabilities mediate the shock jump on the ion gyroradius, and a fraction of the incoming kinetic energy is diverted into accelerating a non-thermal tail of particles — the cosmic rays.

The four canonical phases

Stitched together, the SNR life cycle has four standard phases, each governed by a different balance of forces.

Phase	Duration	Radius scaling	Physics	Observable signature
1. Free expansion	0 – ~10² yr	R ∝ t	M_swept ≪ M_ej; ejecta cruise ballistically	Strong reverse shock, ejecta-dominated X-ray lines (e.g. Cas A)
2. Sedov-Taylor / adiabatic	~10² – 10⁴ yr	R ∝ t²ᐟ⁵	M_swept ≫ M_ej; energy conserved; radiative losses negligible	Sedov self-similar profile; thermal X-rays (e.g. SN 1006)
3. Snowplow / pressure-driven radiative	~10⁴ – 10⁵ yr	R ∝ t²ᐟ⁷	Cooling time < expansion time; thin dense shell forms; momentum conserved	Bright optical filaments (e.g. Cygnus Loop)
4. Merger	≳ 10⁵ yr	v_shock → c_s,ISM	Remnant subsonic; blends into turbulent ISM	No longer identifiable as a discrete object

Phase 1 — free expansion. The ejecta have swept up less than their own mass in ISM. Their velocity is roughly the original ejection velocity. The forward shock leads, the reverse shock lags. Radius grows linearly with time. This phase lasts until M_swept ≈ M_ej; for ejecta mass ~10 M☉ in n = 1 cm⁻³ medium, that happens after about 200 years and a radius of ~2 pc. Cassiopeia A (~340 yr, 2.5 pc radius) is the textbook young remnant just transitioning out of free expansion.

Phase 2 — Sedov-Taylor. Once swept-up mass dominates ejecta mass, the blast wave forgets its origin. The only remembered parameters are total energy E and ambient density ρ₀ — and dimensional analysis on (E, ρ₀, t) gives one combination with units of length: R ~ (E/ρ₀)^(1/5) t^(2/5). The full self-similar solution (Sedov 1946, Taylor 1950, also von Neumann 1941) gives

R_ST(t) ≈ 1.15 (E/ρ₀)^(1/5) t^(2/5)
v_ST(t) = dR/dt = (2/5) R/t   ∝ t^(-3/5)
T_ST    ∝ v²  ∝ t^(-6/5)

Plugging E = 10⁵¹ erg, n = 1 cm⁻³, t = 10³ yr gives R ≈ 4 pc, v ≈ 2000 km/s, T ≈ 10⁷ K — accurately describing remnants like Tycho (450 yr, 3 pc) and Kepler.

Phase 3 — snowplow / pressure-driven radiative. The post-shock temperature has dropped below ~10⁶ K, where the cooling function of an astrophysical plasma peaks. Radiative cooling time becomes shorter than the expansion time. A thin, dense, radiative shell forms behind the forward shock; the interior remains hot and supplies pressure that pushes the shell outward. Conservation of momentum (not energy — energy is now being radiated away) gives R ∝ t^(2/7) in the pressure-driven case, slowing further to R ∝ t^(1/4) when interior pressure has equilibrated with the ISM. The Cygnus Loop is in this phase; its delicate optical filaments are the cooling shell glowing in [O III] and Hα.

Phase 4 — merger. The shock velocity falls below the sound speed in the warm-ionised ISM (~10 km/s). The remnant ceases to be supersonic, blends with ambient turbulence, and ceases to be identifiable. Its mass, energy, and freshly-synthesised elements are now part of the diffuse ISM.

Morphologies — shell, plerion, composite, mixed-morphology

Despite this clean phase sequence, no two SNRs look identical. The dominant classification is morphological:

• Shell-type. A limb-brightened ring with little or no central emission. The radio synchrotron and X-ray emission come from the shocked region just behind the forward shock. Cassiopeia A, SN 1006, and Tycho are canonical shells. About 80 % of Galactic SNRs are shell-type.
• Filled-center (plerion / pulsar wind nebula). Centrally peaked, filled with synchrotron emission powered by a central pulsar's spin-down. The Crab Nebula is the prototype; a 33-millisecond pulsar (PSR B0531+21) supplies the 5 × 10³⁸ erg/s wind that lights the entire nebula across the EM spectrum.
• Composite. A shell with an interior plerion. The pulsar wind nebula sits inside the SNR forward shock. About 5–10 % of Galactic SNRs are composite — e.g. G11.2-0.3, MSH 15-52, 3C 58.
• Mixed-morphology / thermal composite. Radio shell plus centrally-peaked thermal X-ray emission with no obvious pulsar. The standard picture is an evolved SNR that has encountered dense interstellar clouds; thermal evaporation off shocked clouds fills the interior with hot gas. W28, W44, IC 443 are examples.

Cosmic-ray acceleration at the shock

Galactic cosmic rays carry an energy density of ~1 eV/cm³ — comparable to starlight, the Galactic magnetic field, and ISM turbulence. The Milky Way needs to supply roughly 10⁴¹ erg/s to keep that population steady against escape. Three core-collapse supernovae per century at 10⁵¹ erg apiece yield about 10⁴² erg/s; converting ~10 % into accelerated particles gives exactly the required luminosity. This dimensional argument has been the standard reason for thinking SNRs are the main Galactic cosmic-ray accelerators since the 1960s.

The mechanism is diffusive shock acceleration (DSA), a relativistic generalisation of Fermi's 1949 second-order proposal. A charged particle scatters off magnetic irregularities frozen into the plasma on both sides of the shock. Each time it crosses the shock, the converging flow gives it a small fractional energy boost:

⟨ΔE/E⟩ ~ v_shock / c     (first-order Fermi)
P_escape ~ 4 v_shock / c   (probability of leaving per cycle)
Power-law index: dN/dE ∝ E^(-(r+2)/(r-1)) = E^(-2)  for r=4 strong shock

The predicted E⁻² spectrum at injection, modified by energy-dependent escape to E⁻²·⁷ at Earth, matches observation. The mechanism's smoking gun has been the discovery of non-thermal X-ray synchrotron rims on young SNRs — SN 1006, RX J1713.7-3946, Cas A — which require electrons accelerated to 10–100 TeV. TeV gamma-ray detections by HESS, MAGIC, VERITAS, and HAWC at the same SNRs confirm that ions are being accelerated too (via π⁰ decay from p-p collisions), closing the case for SNRs as PeV-scale "PeVatrons".

ISM enrichment — how the galaxy gets heavy

Core-collapse supernovae from massive (≳8 M☉) progenitors eject the layered onion of nucleosynthesis products — H and He envelopes outside, O and Ne and Mg in middle, Si and S and Ar near the core, Fe-peak iron in the innermost shells that survive infall. Type Ia supernovae from a white-dwarf detonation are pure thermonuclear: they fuse roughly 1.4 M☉ of degenerate C/O fuel into Fe-peak elements, ejecting ~0.6 M☉ of ⁵⁶Ni (which decays to Fe) per event.

The remnant phase is the mixing phase. The forward shock sweeps and stirs interstellar gas. The reverse shock heats the ejecta to X-ray emitting temperatures and lights up the spatial distribution of the synthesised elements — Cassiopeia A's Chandra images show iron blobs offset from silicon and oxygen blobs, evidence for asymmetries imprinted by the explosion itself. Over a Hubble time, this combined population is the dominant source of α-elements (O, Mg, Si, Ca) from core-collapse and of Fe-peak elements from Type Ia. Roughly 70 % of every iron atom in your blood was forged in a Type Ia SN and dispersed through its remnant.

Observed examples

Remnant	Age	Distance	Type	Notable features
Crab Nebula (SN 1054)	972 yr	2.0 kpc	Plerion + filaments	33-ms pulsar; 5 × 10³⁸ erg/s spin-down wind; broadband synchrotron
Cassiopeia A	~340 yr	3.4 kpc	Shell (CC, IIb)	Reverse-shocked ejecta knots; central neutron star with carbon atmosphere
SN 1006	1020 yr	2.2 kpc	Shell (Ia)	Non-thermal X-ray synchrotron rims; gold-standard DSA laboratory
Tycho's SNR (SN 1572)	454 yr	3.5 kpc	Shell (Ia)	Probable subluminous Ia; X-ray ejecta knots from Chandra
Kepler's SNR (SN 1604)	422 yr	5 kpc	Shell (Ia)	Asymmetric due to interaction with progenitor wind
Vela SNR	~11,000 yr	0.29 kpc	Composite	Closest known SNR; central 89-ms pulsar; bright soft X-ray
Cygnus Loop / Veil	~10,000 yr	0.7 kpc	Shell, radiative	Snowplow phase; intricate optical filaments
RX J1713.7-3946	~1,600 yr	1 kpc	Shell (CC?)	TeV gamma-ray PeVatron candidate; thin synchrotron rims
3C 58	~840 yr (possibly SN 1181)	2 kpc	Plerion	Slow pulsar; very cool central NS challenges cooling models
SN 1987A	39 yr	50 kpc (LMC)	Young CC	Only modern naked-eye SN; collision with progenitor ring; embryonic remnant

Why we only know about 300 in the Galaxy

The Galactic core-collapse rate is ~2 / century; the Ia rate ~0.5 / century. Over the ~10⁵ year remnant lifetime that predicts roughly 2,500 active SNRs in the Milky Way at any moment. Catalogues currently list ~300 (Green's catalogue, 2022). The shortfall is observational selection — dust extinction hides most of the Galactic plane in optical, the radio surveys lose faint extended emission against the Galactic background, and old radiative remnants are simply faint. Deep wide-field surveys (LOFAR, GLOSTAR, SKA) are projected to roughly triple the catalogue over the next decade.

Worked example — when does Sedov-Taylor begin?

The Sedov-Taylor phase begins when the swept-up mass first exceeds the ejecta mass. Assume an isotropic blast in uniform density:

M_swept = (4/3) π R³ ρ₀

set M_swept = M_ej:
(4/3) π R³ ρ₀ = M_ej
⇒ R_ST = (3 M_ej / 4π ρ₀)^(1/3)

Plug in M_ej = 5 M☉ = 10³⁴ g, n₀ = 1 cm⁻³ (ρ₀ = 2.34 × 10⁻²⁴ g/cm³):

R_ST = (3 × 10³⁴ / (4π × 2.34 × 10⁻²⁴))^(1/3)
     = (1.02 × 10⁵⁷)^(1/3)
     ≈ 1.0 × 10¹⁹ cm
     ≈ 3.3 pc

At free-expansion velocity v ~ 5000 km/s, that radius is reached in

t_ST = R_ST / v ≈ 10¹⁹ cm / (5 × 10⁸ cm/s) ≈ 2 × 10¹⁰ s ≈ 650 yr

So a typical core-collapse SN enters the Sedov-Taylor phase after roughly 600 – 1000 years and at a radius of a few parsecs — consistent with the observed transitions in remnants like Cas A and Tycho. The exact numbers depend on M_ej and ambient density: a SN exploding in a low-density cavity carved by its progenitor wind has a much longer free-expansion phase, and SN 1987A is the textbook case.

Common pitfalls

• Confusing the supernova with the remnant. The SN is the explosion event (seconds of mechanical energy injection, months of optical luminosity). The remnant is the gas-dynamical aftermath that lasts ~10⁵ yr. Saying "the supernova SN 1006" and "the SNR SN 1006" refer to two different temporal stages of the same event.
• Assuming Sedov-Taylor starts at t = 0. The self-similar Sedov solution is the asymptotic form once M_swept ≫ M_ej. Applying it at t < 100 yr (free-expansion regime) overestimates the radius and underestimates the velocity.
• Treating all SNR luminosity as thermal. Young shells emit a non-thermal X-ray synchrotron continuum from accelerated electrons; plerions are nearly pure non-thermal across the EM spectrum. Fitting only thermal models to a synchrotron source gives nonsense temperatures.
• Equating shock compression with strong-shock factor 4. For a strong adiabatic shock with γ = 5/3, ρ₂/ρ₁ = 4. But cosmic-ray pressure on the shock precursor effectively reduces γ toward 4/3, raising compression to 7 or more — observable in some young remnants' density jumps.
• Forgetting the ambient medium. A SN exploding in a pre-existing wind-blown cavity, molecular cloud, or H II region evolves very differently from the textbook uniform-medium case. SN 1987A's interaction with its progenitor's circumstellar ring, and Kepler's interaction with its progenitor's outflow, are the well-studied cases.
• Conflating plerion energy with SN energy. A plerion is powered by the central pulsar's rotational energy (~few × 10⁴⁹ erg), not the SN explosion energy (~10⁵¹ erg). The Crab Nebula has dissipated about 1 % of the Crab pulsar's initial spin energy in 970 years; it is not powered by leftover SN ejecta kinetic energy.

Frontiers and open problems

• The PeVatron problem. The cosmic-ray spectrum has a "knee" at 3 × 10¹⁵ eV — historically interpreted as the maximum energy of Galactic accelerators. Whether SNRs alone can reach 10¹⁵ eV (PeV) is unresolved; magnetic-field amplification at the shock precursor (Bell 2004, non-resonant streaming instability) raises the limit, but observational confirmation needs deep TeV-PeV gamma-ray imaging (CTA).
• Reverse shock asymmetries. Chandra-resolved iron, silicon, and oxygen blobs in Cas A do not match the spherically symmetric onion model. They imply explosion-imprinted asymmetries — clumps, jets, mixing — that 3D core-collapse simulations are still working to reproduce.
• Type Ia SNR diagnostics. Distinguishing the single-degenerate (WD + companion) from double-degenerate (WD + WD) Ia progenitor channel from remnant morphology and spectra is an active field. Tycho's ejecta structure and Kepler's CSM interaction both bear on this.
• The missing 90 %. About 2,000 Galactic SNRs predicted by the SN rate remain unseen. SKA-era radio surveys, with sub-mJy sensitivity over the full Galactic plane, should resolve this within a decade.