Nova Eruption

What a nova actually is

The word "nova" — Latin for "new" — was coined long before astronomers knew what one was. Tycho Brahe used "stella nova" for the supernova of 1572 and the word stuck for any star that appeared where none had been visible before. Modern usage is narrower. A classical nova is a specific binary phenomenon: a thermonuclear flash on the surface of a white dwarf that has been slowly pulling hydrogen-rich gas off a Roche-lobe-filling companion. The white dwarf never had nuclear fuel of its own — it long ago exhausted hydrogen and helium and settled into electron-degenerate retirement. But by stealing fresh hydrogen from a partner, it acquires a thin envelope that can burn, under conditions that turn ordinary nuclear burning into an explosion.

"The white dwarf survives" is the one fact that most cleanly separates novae from supernovae. After a few weeks the ejected envelope thins, the pseudo-photosphere recedes back to the white-dwarf surface, residual burning on the surface keeps the star bright in soft X-rays for months to years, accretion resumes, and the entire cycle restarts. The same white dwarf can go nova many times — and recurrent novae such as T Coronae Borealis and RS Ophiuchi let us watch it happen on human timescales.

How the fuel arrives — Roche-lobe overflow

A classical nova progenitor is a close binary star, called a cataclysmic variable when in the inter-eruption phase. The components are a white dwarf (the primary, typically 0.6 – 1.3 M☉) and a low-mass donor — usually a K or M dwarf on or near the main sequence, less commonly a red giant in symbiotic systems. The donor fills its Roche lobe, the teardrop-shaped equipotential surface in the rotating reference frame of the binary. Any gas pushed beyond the inner Lagrangian point L₁ falls down the gravitational gradient toward the white dwarf.

That gas does not crash directly onto the surface. Because the donor is orbiting, the transferred material carries angular momentum and forms an accretion disk around the white dwarf. Viscous stresses in the disk transport angular momentum outward and matter inward; the inner edge of the disk meets the white dwarf surface in a "boundary layer" where the orbital velocity (~3000 km/s at a typical WD radius) shears to zero. Steady accretion rates in cataclysmic variables span Ṁ ≈ 10⁻¹¹ to 10⁻⁷ M☉/yr — the wide range reflects the difference between dwarf novae in quiescence and high-state systems on the verge of stable hydrogen burning.

Why the layer ignites explosively — degeneracy

Accumulated hydrogen settles into a thin layer on the white dwarf, compressed by ferocious surface gravity (g ≈ 10⁸ – 10⁹ cm/s², a hundred million Earths). At those pressures the electrons at the base of the layer become degenerate. In an electron-degenerate gas, the pressure is set by Pauli exclusion almost independently of temperature:

P ≈ K ρ^(5/3)   (non-relativistic degenerate)

This is the defining problem. In an ordinary thermally supported gas, a small temperature rise causes expansion, which cools the gas, which damps the perturbation — burning is stable. In a degenerate layer there is no expansion: ρ is fixed by gravity, P barely cares about T, and the layer cannot lift itself off the surface to vent the heat.

As more hydrogen piles on, the base of the layer is squeezed denser and hotter. When the temperature reaches roughly 10⁷ K and the column density is about 10⁴ g/cm², the CNO cycle kicks in. Burning rate goes as T^16 in this regime; the base heats up; degeneracy means it cannot expand; the rate accelerates exponentially. Within seconds to minutes the temperature crosses 10⁸ K, the electrons become non-degenerate (thermal pressure overtakes Fermi pressure), and only then is the layer free to lift off — but by that point the burning is supersonic over much of the envelope and a shell flash propagates outward.

The CNO runaway in detail

Hydrogen burning at 10⁸ K does not run through the proton-proton chain — it runs through the CNO cycle, in which protons are captured one at a time onto pre-existing ¹²C, ¹³C, ¹⁴N, and ¹⁵N nuclei, with β⁺ decays returning each cycle to ¹²C. The catalytic CNO nuclei come partly from the white dwarf's own carbon/oxygen surface, dredged up by convective mixing during the early seconds of the runaway. This is why nova ejecta show CNO abundance enhancements of factors 10 – 100 over solar.

The bottleneck reaction is ¹⁴N(p,γ)¹⁵O. Its decay half-life is 178 seconds, so at temperatures above 10⁸ K the CNO cycle becomes "β-limited" — locked to the half-lives of ¹³N, ¹⁴O, ¹⁵O — and the energy generation rate ε saturates near 10¹³ erg/g/s. This hot CNO mode is what powers the explosion. About 10⁴⁵ erg (≈ 10³⁸ J) is liberated in the burst, the bulk of it during the few seconds of peak burning before mass-loss and expansion shut the engine down.

The light curve: rise, peak, decline

From the observer's perspective the eruption goes like this. In quiescence the system is invisible to the naked eye — a 12th-magnitude (or fainter) cataclysmic variable, typically catalogued only because of an earlier eruption. Then over hours to a day or two, the visual brightness rises by 8 to 15 magnitudes — a factor of 10⁴ to 10⁶. The peak corresponds to a luminosity of order 10⁴ – 10⁵ L☉, reaching the Eddington limit for the white dwarf mass. The optical photosphere at peak is the expanding hydrogen envelope; it can swell to several AU before the run-up ends.

After peak, the light curve declines. The classical observational parameter is t₂, the time to fade by 2 magnitudes from peak. Speed class follows:

Speed class	t₂	M_V at peak	Example
Very fast	< 10 d	−9.0 ± 0.4	V1500 Cyg 1975, V838 Her 1991
Fast	11 – 25 d	−8.3 ± 0.4	GK Per 1901, V603 Aql 1918
Moderate	26 – 80 d	−7.6 ± 0.4	DQ Her 1934, V723 Cas 1995
Slow	81 – 150 d	−7.0 ± 0.4	RR Pic 1925, HR Del 1967
Very slow	> 150 d	−6.5 ± 0.4	RT Ser 1909, V723 Cas tail

The Maximum Magnitude – Rate of Decline (MMRD) relation that emerges — fast = bright, slow = faint — is the basis for using novae as secondary distance indicators in the local universe. The physics is straightforward: faster declines come from more massive white dwarfs, which require less accreted mass to ignite, eject material at higher velocity, and recede to the white-dwarf surface sooner.

The ejecta and the nebular phase

The shell expelled by a classical nova is small by stellar standards — typically 10⁻⁵ to 10⁻⁴ M☉, comparable in mass to a planet's worth of hydrogen and helium. Velocities range from a few hundred km/s in slow novae to 3000 km/s in fast ones, and 5000 km/s in some recurrent eruptions. The shell expands roughly homologously, swept up against the slow donor wind in symbiotic systems and against the ISM in field systems.

Early spectra are dominated by P-Cygni absorption from the optically thick pseudo-photosphere. As the ejecta thin, forbidden lines such as [O III] λ5007, [N II] λ6584, [Ne III] λ3869 brighten — diagnostic of the low-density (n_e ~ 10⁶ cm⁻³) regime where collisional excitation is fast but radiative decay outruns collisional de-excitation. Forbidden-line strengths give electron density and temperature, and the ratios of carbon, nitrogen, oxygen, neon, and magnesium relative to hydrogen pin down the abundance pattern of the burned envelope.

Decades after the eruption, deep imaging often reveals a resolved nova shell — small (≲ 1 pc), structured, sometimes bipolar. GK Persei (1901) has a famous expanding shell now several arcminutes across; DQ Herculis (1934) shows a clear equatorial-ring + polar-blob morphology that hints at non-spherical mass loss shaped by the binary and the disk.

Famous novae

• GK Persei (Nova Persei 1901). One of the brightest classical novae of the 20th century, peaking at V = 0.2. Now a dwarf-nova / X-ray-binary system showing recurring outbursts. Resolved expanding shell visible at the eyepiece.
• V603 Aquilae (Nova Aquilae 1918). Brightest classical nova on record, V = −1.4 at peak — outshone Altair. A textbook fast nova; the underlying CV is still active.
• DQ Herculis (Nova Herculis 1934). Slow nova; the post-nova system established the "DQ Her" subclass of intermediate polars — magnetic white dwarfs whose spin period differs from the orbital period.
• RS Ophiuchi. Recurrent nova, eruptions in 1898, 1933, 1958, 1967, 1985, 2006, 2021. Symbiotic system: red giant donor, massive white dwarf, recurrence ~ 15 – 20 yr. The 2021 outburst was detected at TeV γ-rays by H.E.S.S. and MAGIC — direct evidence that nova shocks accelerate hadrons.
• T Coronae Borealis ("the Blaze Star"). Recurrent nova, eruptions documented in 1866 and 1946. Quiescent magnitude 10; at peak it reached V ≈ 2 — visible to the unaided eye for several nights. Pre-eruption photometric and spectroscopic indicators suggest the next eruption around 2026.
• T Pyxidis. Recurrent nova with eruptions in 1890, 1902, 1920, 1944, 1966, 2011. Hosts a small ring nebula of clumpy ejecta resolved by HST. Recurrence ~ 22 yr (most cycles) lengthened to 45 yr after 1966 — a possible clue that recurrence times themselves drift.
• M31N 2008-12a. Andromeda recurrent nova with the shortest known recurrence: about 1 year. Inferred white-dwarf mass ≳ 1.38 M☉, very near Chandrasekhar — a leading candidate progenitor of a Type Ia supernova within ~ 10⁵ years.

Novae versus supernovae — a direct comparison

Property	Classical nova	Type Ia supernova	Core-collapse SN (II / Ib / Ic)
Site of explosion	WD surface	Entire WD interior	Massive-star core
Mechanism	Surface H thermonuclear runaway	Whole-star C/O detonation	Iron-core gravitational collapse
Trigger condition	Degenerate H ignition at base	Approach to M_Ch ≈ 1.4 M☉	Iron core > M_Ch, no fuel
Energy released	~ 10⁴⁵ erg (10³⁸ J)	~ 10⁵¹ erg (10⁴⁴ J)	~ 10⁵¹ erg (10⁴⁴ J)
Peak absolute mag	−6 to −9	−19.3 ± 0.3	−16 to −19 (wide spread)
Ejected mass	10⁻⁵ – 10⁻⁴ M☉	~ 1.4 M☉	1 – 20 M☉
Ejecta velocity	10² – 3 × 10³ km/s	~ 10⁴ km/s	5 × 10³ – 3 × 10⁴ km/s
Survivor	WD intact, accretion resumes	WD unbound — nothing left	Neutron star or black hole
Recurrence	10 – 10⁵ yr (same system)	Never (system destroyed)	Never
Galactic rate	~ 30 – 50 per yr	~ 0.5 per century	~ 2 per century

The cleanest mental model: a nova is a thin-shell surface event; a Type Ia supernova is the entire star detonating. Both involve white dwarfs and thermonuclear burning, both involve a binary partner — but the energy scales differ by a million, the ejecta mass scales by ten thousand, and the star itself either dies or is reset.

What sets the recurrence time?

Time between eruptions is approximately the time to accumulate the ignition mass:

t_rec ≈ M_ign / Ṁ_acc
M_ign ≈ 2 × 10⁻⁴ M☉ × (M_WD / 0.6 M☉)^(−2)   (rough scaling)

Two trends fall out. First, the ignition mass drops steeply with white-dwarf mass, because heavier white dwarfs have higher surface gravity and so compress a given column to ignition conditions sooner. Second, the recurrence time falls with accretion rate. For a 0.6 M☉ white dwarf accreting at 10⁻⁹ M☉/yr — a typical classical-nova progenitor — t_rec ≈ 2 × 10⁵ yr. For a 1.35 M☉ white dwarf accreting at 2 × 10⁻⁷ M☉/yr, M_ign drops to ~ 10⁻⁷ M☉ and t_rec to a year or two. M31N 2008-12a sits at the extreme end of that scaling.

Modern multi-wavelength view

• Radio. Synchrotron emission from external shocks and thermal free-free from ionised ejecta. VLBI imaging of RS Oph 2006 and 2021 resolved expanding bipolar shocks within weeks of eruption.
• Infrared. Dust formation in cooling ejecta — the "dust dip" 30 – 100 days after peak in some novae produces extra extinction at optical bands.
• UV. The pseudo-photosphere is hottest in the UV during the rise; SWIFT/UVOT and HST have detailed photometric coverage of recent novae.
• X-ray. Two distinct phases: hard X-rays from internal shocks during the first weeks (10⁻¹³ – 10⁻¹¹ erg/cm²/s), then the supersoft source phase when the photosphere recedes, exposing the still-burning hydrogen layer at T ≈ 3 × 10⁵ K.
• γ-ray. Fermi-LAT detected ≳ 100 MeV emission in V407 Cyg 2010 and a dozen subsequent novae — direct evidence of relativistic particle acceleration in nova shocks. RS Oph 2021 was the first TeV-detected nova.

The open question: do white dwarfs net-gain or net-lose mass?

This is the single most consequential question in nova physics, and it is unsettled. The argument for net gain: many recurrent novae host white dwarfs already near the Chandrasekhar mass, and they would not be there without long-term growth. If the ignition mass is M_ign and the ejected mass is M_ej < M_ign, the WD gains M_ign − M_ej per cycle. Integrated over a Hubble time of accretion this is enough to reach Chandrasekhar and trigger a Type Ia supernova — the "single-degenerate" channel.

The argument for net loss: high-resolution spectra often show enhanced abundances of C, O, Ne, Mg, characteristic of the underlying WD interior. That implies the eruption excavates white-dwarf material along with the burned envelope — so M_ej > M_acc. Detailed hydrodynamic models (MESA + KEPLER calculations) give different verdicts depending on convection treatment, opacity, mixing prescription, and Ṁ; recent work tends to find net gain for high accretion rates and very high WD masses, net loss for lower-mass white dwarfs.

If novae do feed Type Ia supernovae, we can quantitatively check: the Galactic nova rate (~ 30 – 50/yr) and the Galactic Ia rate (~ 5/century) imply that of order one nova per million eventually leads to a Type Ia, which is consistent with only the most massive white dwarfs being progenitors. The next decade — with LSST detecting thousands of extragalactic novae per year and continued multi-wavelength monitoring of nearby systems — should narrow the answer.

The next event in the sky: T CrB ~ 2026

T Coronae Borealis is forecast to erupt again any time between now and the late 2020s, with several lines of evidence pointing to the central few years of that window. The system's high-state-then-pre-eruption-dip photometric pattern preceding its 1946 outburst is currently being re-played. At peak T CrB should reach about magnitude 2 — comparable to Polaris — visible to the unaided eye in the constellation Corona Borealis for several nights, fading over weeks to its 10th-magnitude quiescence.

It will be the first naked-eye recurrent nova of the multi-wavelength, robotic-survey, social-media era. Continuous coverage from radio to TeV is already pre-positioned, and the eruption — whenever it occurs — will be the highest-quality nova dataset ever taken. Watching it happen, and folding the result into the recurrence and Chandrasekhar-growth questions, is a community-scale experiment running in real time.

Common pitfalls

• Calling everything that brightens a "nova". Modern usage reserves the term for the thermonuclear-runaway phenomenon on a white dwarf. Dwarf novae are disk-instability outbursts of cataclysmic variables, not actual H-burning. "Nova-like variables" are CVs in their high accretion state. None of these are the surface H runaway proper.
• Confusing nova brightness with Type Ia. A classical nova peaks at M_V ≈ −7 to −9; a Type Ia at M_V ≈ −19.3. The brightness ratio is ~ 10⁴ – 10⁶, the same as the energy ratio. They are not on the same scale.
• Assuming the ignition mass is universal. M_ign scales steeply with WD mass and weakly with accretion rate. Massive (~ 1.35 M☉) WDs accreting fast can ignite on 10⁻⁷ M☉; light (~ 0.6 M☉) WDs accreting slowly need 10⁻⁴ M☉. A single number is misleading.
• Treating CNO enhancement as proof of net mass gain. Mixing and dredge-up bring core material into the burned layer, but they do not automatically imply that more underlying core mass is lost than envelope mass gained — the accounting depends on integrated cycle bookkeeping, not single-cycle composition.
• Forgetting the disk during the eruption. The accretion disk is typically destroyed or severely disrupted by the outburst. It reforms over weeks to months as accretion resumes. Quiescence-phase models that assume continuous accretion straight through the outburst over-predict ejecta interaction signatures.