Transcript Thermonuclear supernovae and cosmology

Thermonuclear supernovae and
cosmology
Andrey Zhiglo
Kharkov Institute of Physics and
Technology
Dark matter, dark energy and their
detection 2013
NSU, Novosibirsk
23 July 2013
1
Talk outline
• Supernovae: overview
• Classification, observations,
phenomenology
• Stellar evolution. Stellar models
• SNIa: Thermonuclear supernovae. Theory
• Core-collapse supernovae: theory
2
Overview: history
SN = Supernova = Super[bright] nova [ = “new star”: fast
drastic increase in brightness: by ~20m in ~3 weeks].
Brightest (optical band) stellar-mass objects in the Universe.
In our Galaxy: in 185AD; 1054 (visible in daylight for 23 days,
653 days at night with naked eye. Remnant: M1, = “Crab”
nebula); 1572 (“Tycho”. Brahe recorded light curve, book “De
Stella Nova”); 1604 (Kepler) – last one seen in our Galaxy.
S Andromedae 1885 – first reliable distance to host M31
estimate by Baade&Zwicky, term “Supernova” (not a typical
Nova; brighter) by Zwicky in 1931. ~6m peak brightness.
Before 1990: <~30 SNe found each year. 2012: >5000
SNe known, surveys find hundreds SNe a year; new
dedicated surveys under way.
Most distant SNIa: z=1.914 (10Gyr. 04/13: CANDELS, HST).
3
~3 SNe/century should occur in the Galaxy.Names: SN2006jc
Overview: history; brightness
Historic SNe were noticed by naked eye: anomaly on the sky.
SN1006 ~ –7.5m (visible in daylight), SN1987A (in LMC) +2.9m
peak magnitude. S And ~6m.
Absolute magnitude: average SNIa M= –19.3 at maximum
light, ~20÷25m (factor of 108÷10) brighter than its progenitor.
Milky Way: M= –20.5. Average classical nova: –8.8M . Max
brightness variation in Mira-type variable stars is 11m (o Ceti:
8m: 2m to 10m – 1st variable star discovered, 1638).
SN classification: spectral: Minkowski’1941: Type II and I (H
lines, no H). Elias+’1985: SNIa (SiII lines). Further subdivision
progresses: SNIa classified into Branch-normal (BFN’93;
SN1991bg-like), 90% of total; SN1991t-like (~9%: fast
decline, low amount of Si-group elements), SN2000cn-like,
SN2001ay-like, Hyper-Chandra. Similar 1-parameter
properties in 1st&2nd groups; → way to guess their absolute
4
brightness- best to date cosmological distance determination.
Normal SN Ia: average rise
time ~17.5 days (B-band),
peak absolute magnitude
M= –19.3 , decline by ~ 3m
within ~ 20 days, then ~ 1m
/month.
5
From Filippenko ARAA 1997
[From
Trimble
(1982)]
a
B
6
From [Turatto 2003] 7
P Cygni profile
8
Zwicky (1938) model:
collapse into Neutron Star.
First 35 SNe studied were Ia.
Minkowski (1940): types I &
II – after discovery of
SN1940c. Also Type II:
SN1054 & SN1181.
Hoyle&Fowler (1960):
thermonuclear explosion of
a White Dwarf (WD) star.
Colgate&McKee (1969),
Truran: SNIa light curve
powered by beta-decay
6d
d
56
56
Ni 
Co 77


 56 Fe
Khokhlov (1995): First 3D
simulations of SNIa
From [Weiler 2003]
9
SN Ia, in NGC
4526, 108x106
LY from Earth.
Peak magnitude
+15.2m
10
SN Ib (corecollapse) in
UGC 4904,
77x106 LY
From Earth.
Progenitor:
Bright blue
giant, spectral
class O,
luminosity
class II.
11
Remnant of
SN1006, Ia, 7000
LY away.
SN1006 was
observed by Arab,
Chinese,
Japanese
astronomers.
~–7.5m peak
brightness.
Image 55’ across.
Nearly spherical,
radial stratification
in chemistry,
density, velocity.
12
Remnant of
SN1572
“Tycho”, Ia,
13000 LY
from Earth.
SN1572 was
studied by
Tycho Brahe,
among
others. He
recorded its
light curve.
Book “De
Stella Nova”.
Image 10.5’
across. 13
Crab nebula,
M1.
Remnant of
SN1054,
type II, 6500
LY from
Earth.
Pronounced
filamentary
structure,
intricate small
scale
features.
SN1054 peak
magnitude
14
~–6m
Overview: energetics
Characteristic energy of SN explosion E~(1÷2)x1051 erg.
Mostly kinetic energy of expanding ejecta. SNIa: completely
disrupted: average expansion v = (2E/M)1/2 =~8000 km/s
Energy of EM emission in SNIa ~ 6x1049 erg, mostly optical
and IR; ~1049 erg in SNII.
Progenitor mass ~ a few solar masses, M☼=1.989x1033 g.
M☼c2=1.8x1054 erg. Energy of 4p → 4He+26.72MeV burning
of M☼~12.8x1051 erg. Burning 1M☼ 12C → 56Ni: ~6.5x1051 erg.
Gravitational binding energy of C+O WD ~0.6x1051 erg.
The Sun energy output: 1.2x1041 erg/yr.
Energy carried away by neutrinos and antineutrinos in corecollapse supernova (ccSN): ~(3÷5)x1053 erg =~10÷15%
MFec2.
15
SN Ia emission: double beta
decay ~0.6 M☼ 56Ni → 56Fe.
Emitted e+, γ excite and
heat ejecta, that thus emits
light from its photosphere.
Exponential tail: rad.decay
law when the ejecta
becomes optically thin.
In ccSN: shock heating, H
recombination,
~0.02-0.2 M☼ 56Ni → 56Fe.
16
From Filippenko ARAA 1997
From [Nadyozhin & Imshennik (2005)]
17
SNIa: standard candles for cosmology
Phillips relation (1993)
SNIa – standard candles for
cosmology. 10%-accurate
brightness. 1% accuracy
needed for constraining Dark
Energy Equation of state. 18
SNIa: thermonuclear supernovae
Mechanism: thermonuclear explosion of carbon-oxygen
White Dwarf (WD). 3 channels:
1) Steadily approaches Chandrasekhar mass by accretion,
M ~ 1.38MSun , nuclear runaway near center, explosion.
C+O → 56Ni, intermediate mass elements (IM), starts as
deflagration front, likely transitions to detonation.
Similar initial conditions => uniform outcome, favored model
for most spectroscopically-normal SNIa.
3D simulations show correct energetics and species
distribution in ejecta when detonation is triggered after WD
preexpansion in initial deflagration phase, or when many
ignition points are used. But too few soft supersoft XR sources.
2) Explosion after collision of 2 low-mass WDs.
3) Detonation of degenerate surface layer of accreted He on
19
<MCh WD. Known progenitor systems produce atypical SNe.
Core collapse supernovae
Engine: released gravitational energy of collapsing iron core
of a massive star, MMS>8 M☼, after the inert core reaching
critical mass. The energy is transferred to outer layers by a
shock wave, born at infalling matter bounce from stiffened
nuclear density core, supported via neutrino heating.
Remnant: expanding nebula of IME and iron-group
elements, interacting with circumstellar material
(CSM)/extended envelope (esp. SNII: rich in H); collapsed
core – neutron star (NS) or black hole (BH).
Source of heavy elements, cosmic rays, NS and BH,
gravitational waves and ~20MeV neutrinos.
20
SN EM emission. Light curves
Expanding envelope is heated by
SNIa: (e+, γ) of 56Ni beta decay into 56Fe.
ccSN: shock heating, H recombination, 56Ni → 56Fe.
Envelope becomes optically thinner with time, photosphere
recedes to its center. It becomes optically thin eventually.
SNIa: total optical depth τ≈30 at max light (t=17 days). Light
diffusion time becomes less than expansion time at t=60÷90d,
→ luminosity ~ instantaneous energy release then, end of fast
decline stage. Same t – envelope becomes transparent to γ’s,
e+ heating dominates. After t~400d e+ start escaping too; late
time LC sensitive to magnetic field (that aids in trapping e+).
LC declines slower when more energy is stored in the
envelope (more 56Ni): hotter envelope → higher opacity,
longer light diffusion and cooling time; both peak brightness
and LC width are proportional to m(56Ni) → Phillips relation.
1) Robust within limits; 2) hard to refine underlying physics. 21
Stars with 0.5 M☼
<M<~7-8M☼ burn He
→ C after H is
exhausted in the core,
form degenerate C+O
core. Stars M>~910M☼ burn nondegenerately C,O, the
products to Fe. 22
Evolution tracks. Low mass stars
Isolated stars
with Mainsequence
mass of <~8
M☼ shed H-rich
envelope after
He ignition, and
turn into WDs.
[Wheeler et al.
1990]
23
Evolution tracks. Massive stars
[Wheeler et al. 1990]
24
[Woosley&Janka (2005)]
25
26
27
Talk outline
•
•
•
•
•
Supernovae: overview
Phenomenology and classification
Stellar evolution. Stellar models
SNIa: Thermonuclear supernovae. Theory
Core-collapse supernovae theory
28
SNIa always
occurs in a
tight binary
system, to e–degenerate
WD.
Favored singledegenerate
scenario:
ignition near
the center of a
WD that
approached
MCh through
mass transfer
from the
companion 29
Instability, to collapse in self grav. field
Chandrasekhar limit: degenerate electrons can’t support mass above MCh:
P~ρ5/3, R~1/M for nonrelativistic electrons. P~ρ4/3 ,R(MCh)=0 for ultrarelat. e
Tolman–Oppenheimer–Volkoff limit for NS mass.
30
Massive C+O WD explosion: basics
WD mass grows due to accretion from companion, RG, He or
MS. Stable accretion rate between ~5x10-8 and 2x10-6 M☼/yr.
Smoldering phase: starts when MWD≈0.99 MCh, RWD≈2200
km: strong convection, fluctuations in T, ρ. Lasts ~ 1000 yr, then
nuclear runaway occurs at some ignition point, stable flame.
Pressure wave may trigger ignition at several points with
suitable conditions almost simultaneously. Multipoint ignition
Explosive burning in degenerate matter: sharp increase in T
from ~105 to >109 K leads to little expansion (by 17% in
0.5C+0.5O WD at ρ=2x109 g cm–3).
Explosion: fast nuclear transformation of original C+O
(+traces of Ne, Mg, ...) into heavier elements, in a thin front
sweeping through WD. Most energy released within 3 s after
the near-center ignition. Disruption, homologous expansion.
31
Deflagration and detonation
 Nuclear network in the front:
→20Ne+4He; reaction rates
12C+ 12C→23Na+p→24Mg,
or
.
Q = 8.4x1010. Products, O → local NSE (Nuclear Statistical
Equilibrium) mixture, at
. Mostly Fe, Ni at
density >~108 g cm–3. Incomplete burning at lower densities.
 Detonation: fuel heated by compression in the shock front.
Pure detonation model of explosion produces too little intermediate
mass elements (Si, S, ...), produces mostly Fe, Ni, ...; ruled out.
 Deflagration: almost isobaric. Fuel preheated by heat
conduction from thin reaction zone. Wider preheating zone, still
total W < 1 cm. Premixed flame.
 Near WD center (density ~ 2109 g cm–3) Slam ~ 100 km/s,
slower at smaller densities. Deflagration accelerated immensely by
flame instabilities, Rayleigh-Taylor dominating at largest scales.
2
6S lam
. Pure deflagration 3D model underproduces Ni.
cr , RT 
Ag
32
Laminar flame structure
Heat conduction: by degenerate e–.
Species diffusion unimportant, Le=~107
[Gamezo et al. 2005]
[Calder et al. 2007]
33
Deflagration instabilities
 Landau-Darrieus instability (LDI): more pronounced at larger expansion
1/ = ρfuel/ρash –1. Critical wavelength of the flame width scale. Nonlinear
regime: stabilizes in cellular front shape.
[Zhiglo 2009]
 Rayleigh-Taylor instability (buoyancy) and Kelvin-Helmholtz (shear flow
34
instability) on large scales. Less sensitive to flame structure.
SN Ia: Modeling deflagrations
Important hydrodynamic scales: from <10–3 cm (flame width near
WD center: large ρ, u, etc. across. Kolmogorov scale: Re=1014) to
>108 cm (gravity scale; box size). Impossible to disentangle scales:
instabilities on lowest scales determine integral burning rate → star
expansion → local conditions at the flame, governing instabilities, and
burning rate. Gibson scale~ 104 cm, numerical resolution 5x104 cm.
Early simulations of 70s-early 90s: 1D spherical, DDT by hand.
Current mainstream 3D simulations: simplified physics (no rotation,
magnetic fields). Attempts to track flame zone without resolving actual
flame, to reproduce on large resolved scales average flame zone
propagation, and correct heat release rate. Method:
1) Physics: Detailed nuclear network computed once with no
hydro, yields Dlam. Renormalized for unresolved scales effective flame
speed, Df (Δ).-To account for the fact that real flame area cannot be
directly estimated in simulations with crude grid scale.
2) Numerics: technique to propagate the flame zone with such
renormalized speed.
35
Steady RT-enhanced burning in 2D; quasiperiodic
Density distribution approximately linear in the flame brush (on average)
36
RT-enhanced burning in 2D, various Df
[Zhang et al (2007)]
37
RT-enhanced burning in 3D, various Df
[Zhang et al (2007)]
38
39
Flame models. Numerical artifacts
( f )
 (uf )  ( Kf )   ( f , T )
t
T
1
q
 uT  ( T )   ( f , T )
t

cp
Reaction-diffusion equations
governing reaction progress
variable f, coupled to Euler
equations for matter velocity,
stellar EOS (degenerate electronpositron gas, photons, ideal ion
gas, with electron screening,
Coulomb corrections).
Flame model exhibits its own
Landau-Darrieus instability,
generates noise. The artifacts
must not obscure real physics.
40
Detonation triggered
by hand at certain
point, at t=1.62 s
(left) and t=1.51 s
(right). The
computational
domain size
xmax = 5.35x108 cm.
Pure deflagration
model leads to weak
explosion
(E=0.8x1051 erg),
sunk fingers of nonprocessed C and O
near WD center, in
conflict with
observations.
[Gamezo et al 2005]
41
[Jordan
IV,
et al,
2008]
42
43
Gravitationally confined detonation model
[Meakin+, 2009]
When ignited off-center (40 km in the simulation shown) the bubble of burnt
material is buoyantly driven to the surface, while the WD expands due to
nuclear heat released. After breakout a surface wave is generated, that
collides at the opposite pole, creating suitable conditions for detonation. 44
~1.1 M☼ of 56Ni produced, 0.08-0.14 M☼ of IME. Mostly IME at v>14000 km/s.
SNIa: conclusions
SNIa = thermonuclear explosion of a C+O WD. Only happens in tight
binary systems, influx of matter from the companion is crucial for all
explosion mechanisms.
Favored mechanism for spectrally-normal SNIa: single-degenerate
scenario. WD mass steadily increases by accretion from the companion,
accreted matter steadily burns into C and O on WD surface. When WD
mass reaches ~99% MCh a nuclear runaway occurs near WD center, the
flame sweeps through the WD in ~2 s. Most fuel burnt into iron-group
elements in these 2 s. ~0.1-0.4 M☼ IME formed in outer less dense regions.
Burning and WD hydro expansion rate are about the same. This leads to
significant dependence of the 56Ni produced on ignition conditions. Both peak
luminosity and the width of light curve (powered by 56Ni decay into 56Fe) are
about proportional to m(56Ni) → Phillips relation. This rates coincidence, as
well as disparate spatial scales of important physics make SNIa simulations
extremely challenging. Used simplified models are questionable.
Pure deflagration models produce under-energetic explosions, with
spectral inconsistencies with observations. Detonation happening at ~1.5 s
after ignition make the outcome consistent with observed ensemble of
SNIa. Universal robust mechanism for DDT is missing.
45
SNIa: cosmology
• Involved procedures for estimating absolute brightness from LC (best
stretch), spectra (Δm correction based on multicolor LC shapes – for ISM
dust and found systematic low brightness of redder SNIa).Mostly empirical.
• Attempts to pinpoint and learn to correct for secondary parameters
influencing SNIa brightness. Known: SNIa at higher z were brighter, and
residual bias remains after standard Δm correction. Corrections from stellar
population in the galaxy (whenever it can be estimated).
Theory: evolutionary path and metallicity affect central WD density at
ignition and metallicity throughout. At ρcentr >~109 g cm-3 e–-capture rate
grows fast, yieding stable 56Fe and 58Ni, less energetic explosion.
Metallicity likely affects DDT. Simulations are still controversial.
Circumstellar material affects the spectra (e.g. introducing high-velocity
features [Gerardy+ 2005]) and could be required to be taken into account;
although classical CSM seems insufficient in mass for this [Zhiglo 2011].
Empirics: analyzing multicolor difference LC between pairs of SNIa and
statistically finding best functions of the set of the secondary parameters to
fit these differences
[Hoeflich+, 2013].
46
Inner part of
M1, remnant
of Type II
SN1054.
Wisps of
plasma
accelerated in
gravitational
and magnetic
fields (~1012 G)
of a pulsar.
Composite
image, Xray+optical;
1.6’ across.
47
Interaction
of ejecta
of
SN1987A
(Large
Magellanic
Cloud)
with CSM.
48
SN Type
Ib
Ic
Characteristics
Guess at progenitor
No hydrogen in spectrum
•Absorption near 5700 A, due to He
(plus other He lines)
•Late-time emissions from O-I, Ca-II
Massive stars which has
been stripped of H before
core collapse?
•Wolf-Rayet stars?
No hydrogen in spectrum
•No helium in spectrum
Massive stars stripped of H
before core collapse?
•WR stars? Late disk galaxy
•Late-time spec. emission from O-I, Ca-II
H in spectrum, with P-Cygni profile
•Light curve has a plateau for 30-90
(plateau)
days soon after maxium
II-P
Massive red supergiant
II-L
(linear)
Hydrogen is spectrum weak or no PCygni profile
•Light cure falls linearly after maximum
Less massive supergiant?
•Lost some of envelope?
IIb
Hydrogen in spectrum but not much
•Helium in spectrum
•Late-time spectrum emissions from OI, Ca-II, H
Massive stars which has lost
MOST of its H envelope (in
binary?)
II-n
H in spectrum, with narrow emission
lines on top of broad emission features
•Slow decline in light curve at late times
Massive star which sits in
the middle of massive stellar
49
outflow?
[M. Richmond]
50
[Janka (2007)]
Rs: Shock radius. Rν: radius of neutrino sphere (where “optical depth” τ=1).
Rg: gain radius, above which neutrino heating exceeds neutrino cooling.
Neutrinos effectively trapped at ρ>1012 g cm-3. S hock stalls due to energy
loss on nuclei dissociation. Revived,
51
52
53
54
55
56