S UBMITTED FOR PUBLICATION IN T HE A STROPHYSICAL J OURNAL ON J UNE 12, 2009.

Preprint typeset using LATEX style emulateapj v. 08/22/09

SHOCK BREAKOUT FROM TYPE IA SUPERNOVA

A NTHONY L. P IRO , P HILIP C HANG 1 , AND N EVIN N. W EINBERG
Astronomy Department and Theoretical Astrophysics Center, University of California, Berkeley, CA 94720;
[email protected], [email protected], [email protected]
Submitted for publication in The Astrophysical Journal on June 12, 2009.

ABSTRACT
arXiv:0909.2643v1 [astro-ph.HE] 14 Sep 2009

The mode of explosive burning in Type Ia SNe remains an outstanding problem. It is generally thought to
begin as a subsonic deflagration, but this may transition into a supersonic detonation (the DDT). We argue that
this transition leads to a breakout shock, which would provide the first unambiguous evidence that DDTs occur.
Its main features are a hard X-ray flash (∼ 20 keV) lasting ∼ 10−2 s with a total radiated energy of ∼ 1040 ergs,
followed by a cooling tail. This creates a distinct feature in the visual light curve, which is separate from the
nickel decay. This cooling tail has a maximum absolute visual magnitude of MV ≈ −9 to −10 at ≈ 1 day, which
depends most sensitively on the white dwarf radius at the time of the DDT. As the thermal diffusion wave
moves in, the composition of these surface layers may be imprinted as spectral features, which would help
to discern between SN Ia progenitor models. Since this feature should accompany every SNe Ia, future deep
surveys (e.g., m = 24) will see it out to a distance of ≈ 80 Mpc, giving a maximum rate of ∼ 60 yr−1 . Archival
data sets can also be used to study the early rise dictated by the shock heating (at ≈ 20 days before maximum
B-band light). A similar and slightly brighter event may also accompany core bounce during the accretion
induced collapse to a neutron star, but with a lower occurrence rate.
Subject headings: hydrodynamics — shock waves — supernovae: general — white dwarfs

1. INTRODUCTION such an event and how long does it last

The use of Type Ia supernovae (SNe Ia) as cosmological In this paper, we characterize the light curve expected from
distance indicators has brought attention to the theoretical an SN Ia shock breakout. In §2, we present the initial WD
uncertainties that remain about these events. Generally it is profile before shock passage and investigate when the shock
thought that they result from the unstable thermonuclear igni- runs away form the detonation using both a numerical, hydro-
tion of a C/O white dwarf (WD). One key question is how dynamic simulation and analytic arguments. We then present
the burning front propagates during the incineration. The equations that describe how the shock steepens as it propa-
consensus is that the flame begins as a subsonic deflagration gates through the surface layers ahead of the detonation. In
(Nomoto et al. 1976, 1984) to match the observed nucleosyn- §3, we follow the expansion and cooling of the shock-headed
thesis and light curves (Filippenko 1997), but the later propa- envelope using a semi-analytic analysis. We demonstrate that
gation is more uncertain. Motivated by terrestrial combustion, the entire light curve can be modeled with self-similar solu-
many have argued for a delayed detonation transition (DDT, tions. The breakout produces a sharp, hard X-ray flash with
Khokhlov 1991; Woosley & Weaver 1994). Although a DDT a duration of ≈ 10−2 s (also see Imshennik et al. 1981). As
may be needed to better replicate observations (Plewa et al. the shocked envelope expands and cools, the lightcurve shifts
2004; Livne et al. 2005), how and if it happens is uncertain from the X-ray to the UV and eventually to the visual wave-
(Niemeyer & Woosley 1997; Niemeyer 1999; Woosley 2007). bands. In §4 we show that it reaches a peak absolute mag-
Any observational feature that unambiguously demon- nitude of MV = −9 to −10 at ≈ 1 day. We also discuss what
strates that a detonation occurs would be helpful for resolving depths of the WD are probed by the receding thermal diffu-
this uncertainty. One consequence of a detonation is that it sion wave as a function of time. In §5 we summarize our
drives a shock through the WD surface layers, the breakout results and discuss the possibility of shock breakouts from ac-
of which is a clear signature of the DDT. For Type II super- cretion induced collapse (AIC) to a neutron star.
novae (SNe II), it was long expected that shock breakout from
core-bounce would produce an X-ray and/or ultra-violet flash 2. THE DYNAMICS AND THERMODYNAMIC
(Colgate 1974; Klein & Chevalier 1978; Falk 1978). As the PROPERTIES OF THE SHOCKED ATMOSPHERE
earliest electromagnetic emission available, it is an important We first summarize the properties of the unshocked WD
probe of the progenitor star and its circumstellar environment envelope (§2.1) and then model the detonation to determine
(Matzner & McKee 1999). These X-ray flashes have been ob- how the outward propagating shock steepens as it runs down
served in the cases of GRB 060218 (Campana et al. 2006). the background density gradient (§2.2). From these we derive
which was associated with the SN Ic 2006aj (Mazzali et al. the equations that describe how the envelope responds to the
2006) and the X-ray transient XRT 080109, associated with shock’s passage (§2.3).
SN 2008D (Soderberg et al. 2008). Indeed, since the SN
Ia rate in star forming regions is ∼ 1/3 that of SNe II 2.1. The Pre-shocked Atmosphere
(Mannucci et al. 2005), there may be hope that a SN Ia shock
breakout may soon be seen. The question is how bright is The pre-shocked envelope, which we denote with the sub-
script 0, is estimated as a constant flux atmosphere, F =
4
1 Canadian Institute for Theoretical Astrophysics, 60 St. George Street, σSB Teff , and is in radiative equilibrium. We assume a Chan-
University of Toronto, ON M5S 3H8, Canada; [email protected] drasekhar mass (M = 1.4M⊙ ) WD with a radius R∗ ≈ (3 −
2 PIRO, CHANG, & WEINBERG

6) × 108 cm, which is larger than the pre-ignition radius due

to expansion during the deflagration phase (in contrast, the
pre-ignition radius is ≈ 1.6 × 108 cm). Radiative diffusion
gives
4 16T03 g dT0
Teff = , (1)
3κ dP0
where we take a constant opacity, κ = 0.2 cm2 g−1 , corre-
sponding to electron scattering in hydrogen deficient mate-
rial. We take the gravitational acceleration g = GM/R2∗ to be
constant in these shallow, surface layers. Integrating equation
(1), assuming an ideal gas equation of state, P0 = ρ0 kB T0 /µm p ,
where µ is the mean molecular weight, we get the pressure as
a function of density,
−1/3
Teff,5 ρ0 ergs cm−3 ,
4/3 4/3
P0 (ρ0 ) = 6.1 × 1013 g9 (2)
9 −2 5
where g9 = g/10 cm s , Teff,5 = Teff /10 K, ρ0 is in cgs units,
and µ = 4/3 (for a helium dominated composition).
At sufficiently large depths, the pressure becomes domi-
nated by non-relativistic electrons, in which case
P0 (ρ0 ) = 9.91 × 1012(ρ0 /µe )5/3 ergs cm−3 , (3)
where µe is the mean molecular weight per electron. This
switch occurs around ρ0 ≈ 9 × 103 g cm−3 . These two power
laws (eqs. [2] and [3]) motivate our use of a polytropic back-
ground model in the following sections. F IG . 1.— The composition (top panel) and fluid velocity (bottom panel) as
a function of initial WD density. The detonation is initiated at 107 g cm−3 ,
2.2. Detonation and Shock Runaway sending a burning wave into the flammable lower-density region, and a shock
wave down into the WD. At ≈ 2 × 106 g cm−3 , the steepening shock begins
The energy budget available to the shock is set by where it to race ahead of the burning, as can be seen by the marked decline in in-
runs ahead of the detonation, giving a characteristic velocity termediate mass element synthesis. At this depth, the shock has a speed of
and density for the shock, Vrun and ρrun . The strong shock sub- ≈ 6 × 108 cm s−1 , which sets the initial conditions for the shock breakout
sequently steepens in the declining density gradient according calculation.
to
−β
Similar results are obtained by using simple analytic scal-

ρ0
V0 = Vrun , (4) ings. As the detonation propagates, it causes a change of pres-
ρrun
sure
with β = 0.1858 for a radiation pressure dominated shock pdet Enuc
(Sakurai 1960, γ = 4/3). ≈ , (5)
We use a one-dimensional, hydrodynamics code (described p0 Eint
previously in Weinberg & Bildsten 2007) to follow the det- where Enuc ≈ 0.8 MeV nucl−1 is the specific energy release
onation’s propagation and transition to a shock, which we
from burning two 12 C and two 16 O to 56 Ni, Eint = 3EF /4µe is
show in Figure 1. The background is initialized with a con-
the internal energy, and
stant flux radiative atmosphere, which smoothly transitions
to a nearly isothermal degenerate star. A detonation is initi- EF = 0.41ρ6 (2/µe )1/3 MeV nucl−1 ,
1/3
(6)
ated at 107 g cm−3 (near where the DDT is expected, Woosley
2007) by artificially raising the temperature. Above this depth is the Fermi energy. The speed and temperature of the det-
the composition is a flammable mixture of equal parts 12 C onation are then Tdet = (3pdet /a)1/4 and Vdet = (pdet /p0 )1/2 cs ,
and 16 O, and below it is inert material to prevent the deto- respectively, where cs is the sound speed. At the depths where
nation from propagating inward (in accord with the preced- the electrons are degenerate and relativistic, this gives a den-
ing deflagration). The simulation shows an outward prop- sity independent detonation speed of Vdet ≈ 6 × 108 cm s−1 ,
agating detonation, which is accompanied by a weak shock consistent with our numerical results.
sent downward into the WD core. The detonation propagates The burning in the detonation occurs over an “induction
until ≈ 2 × 106 g cm−3 , at which point the shock runs away. length” given by
This is most clearly seen in the top panel of Figure 1 where
the synthesis of intermediate mass elements falls off in com- λ = (Enuc /ǫ)Vdet (7)
parison to 16 O (the depletion of 12 C at these depths is due to where ǫ is the energy generation rate. As the density de-
residual burning behind the shock). In the bottom panel, we creases, the burning rates does as well, increasing λ. Once
see that the velocity falls off as a V0 ∝ ρ−0.186
0 power law as λ is greater than a pressure scaleheight H = P/ρg, the burning
expected (eq. [4]). From this plot we determine that the char- can no longer keep up and the shock races ahead of the burn-
acteristic numbers at runaway are ρrun ≈ 2 × 106 g cm−3 and ing front. Using the energy generation rate, ǫ, for burning 12 C
Vrun ≈ 6 × 108 cm s−1 . to 24 Mg from Woosley at al. (2004, see their eq. [4]), this
 SHOCK BREAKOUT FROM SNe IA 3

occurs at a critical density of We assume that the pressure decreases adiabatically as the
−3
shell expands, i.e., P ∝ ργ , so that
ρrun ≈ 2 × 106 g0.11
9 g cm . (8)  γ
ρ(ρ0 ,t)
The prefactor is robust because of the strong temperature and p(ρ0 ,t) = psh (ρ0 )
ρ0
density scalings of the energy generation rate. This compares 1−2β  γ
well with our numerical calculation (see Fig. 1).

2 2 ρ0 ρ(ρ0 ,t)
= ρrunVrun . (17)
γ+1 ρrun ρ0
2.3. Shock Heating and Adiabatic Expansion
The jump conditions for a strong shock (Whitham 1958, Although the adiabatic approximation is good, it is in general
his eq.[32]) give the pressure psh , density ρsh , and velocity V not exactly true, an issue that we discuss in §3.3. By combin-
behind the shock in terms of the shock velocity V0 , ing equations (15) and (17), we solve for the pressure of the
expanded shell with initial density ρ0 at any time t.
2 γ+1 2
psh = ρ0V02 , ρsh = ρ0 , V = V0 . (9)
γ+1 γ−1 γ+1 3. THE LIGHT CURVE FROM SHOCK BREAKOUT
AND SHOCK-HEATED COOLING
We next need to understand how these properties change with
3.1. The Prompt Breakout Flash
expansion.
Consider a shell of mass ∆M, density ρ0 , and thickness Photons in the radiative shock stream out once τ . c/V ,
∆r0 ≪ r0 , where r0 ≈ R∗ is the local radius in the WD enve- where τ is the optical depth. Substituting V → ΓV , where Γ =
lope. The shell’s radius as a function of time is [1 − (V /c)2 ]−1/2 (the shock is mildly relativistic at breakout),
we find a breakout depth of
r(ρ0 ,t) = r0 + V (ρ0 )t ≈ R∗ + V (ρ0 )t. (10)
" #1/(1+1/n−β)
Pressure gradients may increase the velocity V (ρ0 ) by as much γ + 1 Γc g
ρ0,br = ρrun
as a factor of 2 (Matzner & McKee 1999), but we find this 2 Vrun κKρrun1+1/n
difference has little effect on our solutions. By continuity of
mass, the shock compresses the shell to a thickness ∆rsh = ≈ 10−3 (g9 /K13.8 )0.87V9−0.87 ρ−0.16
6 (Γ/1.7)0.87 g cm−3 ,
(γ − 1)/(γ + 1)∆r0 . Subsequently, the thickness increases due (18)
to the difference in velocity between the top and bottom of the
shell, giving a thickness as a function of time where K13.8 = K/6 × 1013 cgs, V9 = Vrun /109 cm s−1 , and ρ6 =
ρrun /106 g cm−3 (using typical values from Fig. 1).
∆r(ρ0 ,t) = ∆rsh + ∆V (ρ0 )t, (11) This initial streaming of photons gives rise to the prompt
where ∆V (ρ0 ) is breakout flash. The energy density available is E = 3psh , and
using equation (16) for psh , we estimate a total energy budget
∂V ∂ρ0 at a depth ρ0,br of
∆V (ρ0 ) ≈ ∆r0 . (12)
∂ρ0 ∂r0
ρ0 Eflash = 4πR2∗ H0 × 3psh
We take the equation of hydrostatic balance, dP0 /dr0 = −ρ0 g, ≈ 4 × 1040 (g9 /K13.8 )−0.16V91.0 ρ0.22 2
6 R8.5 ergs, (19)
(n+1)/n
and assume a polytropic equation of state, P0 = Kρ0
(where K and n are set by either eq. [2] or [3]). Setting where R8.5 = R∗ /3 × 108 cm, and associated temperature
1/n
∆r0 ≈ H0 = P0 /ρ0 g = Kρ0 /g, where H0 is the pressure scale Tflash = (3psh/a)1/4
height, we find
≈ 2 × 108 (g9 /K13.8 )0.14V90.32 ρ0.068
6 K. (20)
 −β
nβ 2 ρ0 This temperature is a lower limit, since non-LTE effects
∆V (ρ0 ) = Vrun , (13)
n+1 γ+1 ρrun (which we ignore to simplify our analysis) will only harden
the spectrum (Katz et al. 2009). We further discuss the limita-
which is smaller than V by a constant factor of nβ/(n + 1) ≈ tions of this assumption in §3.6. These numbers are consistent
0.14 (for n = 3 and β = 0.186). with the findings of Imshennik et al. (1981), who predicted a
By mass conservation hard (≈ 20 keV) X-ray flash associated with WD detonation.
The main difference is that we are quantifying this flash in
4πρ0 R2∗ ∆r0 = 4πρ(ρ0 ,t)r(ρ0 ,t)2 ∆r(ρ0 ,t), (14) terms of the DDT picture. The timescale for the energy re-
where ρ(ρ0 ,t) is the expanded density as a function of time. lease will be dominated light travel effects (R∗ /c ≈ 10−2 s),
This gives us a relation for the fractional change in density which we include in our calculations in §3.2. Also, the DDT
will not occur everywhere at once (for example, as found in
ρ(ρ0 ,t) R2∗ ∆r0 the simulations by Röpke et al. 2007), which further smears
= , (15) out the breakout flash.
ρ0 r(ρ0 ,t)2 ∆r(ρ0 ,t)
which is a function of only ρ0 and t. 3.2. Cooling of the Ejected Shock-Heated Envelope
From the shock jump conditions (eq.[9]), we find the pres-
sure in the shocked and expanded shell, Following breakout, a thermal diffusion wave begins prop-
agating back into the shock-heated envelope, releasing pho-
2 2

ρ0
1−2β tons that will be seen as the shock breakout flash. At any
psh (ρ0 ) = ρ0V02 = 2
ρrunVrun . (16) given depth in the shock-heated envelope, the energy density
γ+1 γ+1 ρrun
4 PIRO, CHANG, & WEINBERG

F IG . 3.— The early time light curve, including time travel effects across
the WD surface (thick lines) in comparison to the one-dimensional solutions
plotted in Figure 2 (thin lines). Light travel effects decrease the observed lu-
F IG . 2.— Solutions for the cooling, shock-heated WD envelope with Vrun = minosity at early times when light from the entire surface has not yet reached
109 cm s−1 and ρrun = 106 g cm−3 . The top and middle panels show the time- the observer.
dependent luminosity and effective temperature from the moment of shock
breakout until ≈ 1 day. The bottom panel shows the thermal diffusion depth
∆rdiff ≡ ∆r(t = tdiff ). The solid lines (dashed lines) are for an initial WD flash, light travel effects and not simultaneous DDT ignition
radius of R∗ = 3 × 108 cm (6 × 108 cm). In the bottom panel, the horizontal will smear out the light curve over R∗ /c ≈ 10−2 s or more. To
dotted line denotes R∗ = 3 × 108 cm. better quantify this effect, we plot the light curves altered by
light travel effects in Figure 3. These are calculated according
is E = 3p and leaks out of the envelope on a thermal diffusion to
timescale, which at a depth ρ0 and a time t, is Z π/2
κ L(t) = L[tr (t, cos θ)]d cosθ, (25)
tdiff (ρ0 ,t) = ρ(ρ0 ,t) [∆r(ρ0 ,t)]2 . (21) 0
c
Since tdiff sets the timescale when an observer sees down to a where tr = t − r(tr )(1 − cosθ)/c is the retarded time of photons
certain depth in the shock heated envelope, we set tdiff (ρ0 ,t) = emitted at a angle θ with respect to the observer. Note that
t to solve for ρ0 (t), using the prescription for ρ(ρ0 ,t) and r(tr ) must be evaluated at the retarded time because the ra-
∆r(ρ0 ,t) from §2.3. Once we have ρ0 (t), we can solve for dius was smaller in the past The robust feature to take away is
any other property of the envelope as a function of time. In that ∼ 1040 ergs of energy is released during this initial peak
particular, the luminosity of the expanding and cooling enve- (consistent with eq. [19]). In the second stage, the luminos-
lope is ity decreases more slowly, from 1042 ergs s−1 to 1040 ergs s−1
over ≈ 1 day. The effective temperature evolution covers a
4πr(t)2 E(t)c wide range of wavelengths, going from X-rays to ultraviolet
L(t) = , (22)
τ (t) to visual over the course of ≈ 1 day.
The power law behavior of L(t) can be understood ana-
where lytically by considering the expansion and cooling in cer-
τ (t) = κ∆r(t)ρ(t) (23) tain limiting cases. To emphasize this, we have plotted R∗ =
3 × 108 cm in the bottom panel of Figure 2 (dotted line). A
is the optical depth of the diffusion layer. comparison with L(t) clearly shows that the luminosity break
The general solution of L(t) is non-algebraic, but easily occurs when ∆rdiff has expanded to a thickness ≈ R∗ . Moti-
solved numerically. In Figure 2 we plot L(t) (top panel), the vated by this, we consider the time-dependent behavior in two
effective temperature (middle panel) different limits:
 1/4
L(t) 1. At early times, t ≪ R∗ /V , the envelope has barely
Teff (t) = , (24)
4πr(t)2 σSB moved and r(t) ≈ R∗ . By equation (15), we see that
ρ/ρ0 ∝ t −1 since the radial position of the shell remains
and the thermal diffusion depth ∆rdiff ≡ ∆r(t = tdiff ) (bottom constant and the thickness increases as ∆r ∝ t.
panel). The luminosity decreases as a broken power law in
time. The initial decline is steeper, with a L ∝ t −4/3 power 2. At late times, t ≫ R∗ /V , both r and ∆r increase like t,
law. In practice, this power law and the very high initial lu- so that ρ/ρ0 ∝ t −3 .
minosities are an artifact of our one-dimensional treatment,
and will not be seen in observations. Just as for the breakout We next derive the analytic self-similar solutions in each limit.
 SHOCK BREAKOUT FROM SNe IA 5

3.3. Case 1: t ≪ R∗ /V which has a power law time dependence, L ∝ t −0.34 , consistent
In this limit the expansion is plane-parallel so that the ra- with what Chevalier (1992) found in the same regime. The
dius is basically fixed at r(t) ≈ R∗ , whereas the thickness has effective temperature is
expanded by ∆r(t) ≈ ∆V t. Since ρ(t) ∝ t −1 , we find that the
Teff (t) ≈ 1 × 106 (g9 /K13.8 )−0.065V90.019 ρ0.0035
6 R0.13
8.5 t
−0.46
K.
thermal diffusion depth is independent of time when we set
the condition tdiff = t, and we find (34)
c
ρ0,diff = (κH0 )−1 Since the surface area is increasing and the diffusion wave
∆V is moving into higher temperature material, the evolution is
≈ 4 × 10−3 (g9 /K13.8 )0.87V9−0.87 ρ−0.16
6 g cm−3 . (26) shallower than in the plane-parallel case.
The physics expressed by equation (26) is that the thermal dif-
fusion wave sits at a fixed depth because the column of mate- 3.5. The Exposure of Originally Degenerate Material
rial sitting above it is fixed. The optical depth is proportional At sufficiently late times, the diffusion wave moves into
to column, and thus also fixed, material that was originally degenerate before being hit by
τdiff = κ∆Vtρ(t) = κH0 ρ0 ≈ 7(g9 /K13 )0.87V9−0.87 ρ−0.16 . the shock. Setting ρ0,diff in equation (31) equal to ≈ 9 ×
103 g cm−3 (§2.1), this occurs at ≈ 700 s. The dynamics are
6
(27) essentially the same as that of §3.2, but now n = 3/2 instead of
This gives a luminosity of n = 3. The diffusion wave now moves through the layer more
slowly, with ρdiff,0 ∝ t 1.1 , which gives a luminosity,
4πR2∗ E(t)c
L=
τ L(t) ≈ 2 × 1040 (g9 /K13 )−0.41V91.9 ρ0.36 0.83 −0.16
6 R8.5 t4 ergs s−1 ,
2 −4/3
≈ 9 × 1041 (g9 /K13.8 )−0.34V90.72 ρ0.13 −1
6 R8.5 t−2 ergs s , (35)
(28)
and effective temperature,
where t−2 = t/10−2 s and
−1/3 Teff (t) = 2 × 104 (g9 /K13 )−0.058V90.030 ρ0.0058 R0.11 −0.44
8.5 t4 K,
Teff ≈ 1 × 107 (g9 /K13.8 )−0.086V90.18 ρ0.033
6 t−2 K (29) 6
(36)
is the effective temperature.
Since in this limit, the same mass shell is always doing the where K13 = K/1013 cgs and t4 = t/104 s. The initially de-
radiating (as indicated by a fixed ρ0,diff in eq. [26]), it can- generate material causes the light curve to flatten and the lu-
not be evolving adiabatically (thus our assumption in eq. [17] minosity remains at ∼ 1040 ergs s−1 well until a day after the
is not strictly valid). In a more detailed study we solve the initial shock breakout (as is shown in Fig. 2).
entropy equation in the plane parallel limit, including the ra-
diative loss term (Chang, Matzner, & Piro, in preparation). 3.6. Assumption of Local Thermodynamic Equilibrium
We find that the cooling is only marginally faster when we
include radiative losses, with a logarithmic time dependence. In our derivations we have assumed that the photons and
The plane-parallel limit applies until Vt ≈ R∗ . Setting electrons are thermally equilibrated throughout the shock pas-
ρ0,diff = ρ0 into equations (4) and (9), we find this occurs at sage and subsequent expansion. This was done for the sake of
attaining concrete results and simple analytic expressions, but
t ≈ 1 × 10−2 (g9 /K13.8 )0.16V9−1.2 ρ−0.22
6 R8.5 s (30) it begs the question of how accurate this is.
After the initial shock passage, the envelope is heated
at which point we must begin to account for the radial expan-
and the electron and photons can reach the same tem-
sion.
perature via Comptonization on a timescale tComp ∼
3.4. Case 2: t ≫ R∗ /V (me c2 /kB T (ρ0 ))/(κρ0c), where me is the electron mass and
The layer has now moved an appreciable distance from the T (ρ0 ) is solved from equation (16). This timescale must be
star, so that r(t) ≈ V t (as addressed in Chevalier 1992). Solv- less than the local expansion timescale, texp ∼ H0 /∆V (ρ0 ),
ing for the diffusion depth in this limit otherwise significant expansion occurs before equilibrium is
reached. For the condition tComp . texp , we find that the den-
"  2 #1/(1+β+1/n)
n+1 2 g Vrun ct sity must be greater than
ρ0,diff = ρrun ,
nβ γ + 1 κKρrun
1+1/n c R∗ ρ0 & 3 × 10−3 (g9 /K13 )0.59V90.030 ρ0.0055 g cm−2 . (37)
6
0.66 −1.3 1.3 −3
≈ 2 (g9 /K13.8 ) V90.66 ρ0.12
6 R8.5 t g cm . (31) This is not much greater than the density of the shock breakout
The optical depth is (eq. [18]), so we conclude that at least initially all but the very
β outer envelope reaches equilibrium.
n+1 γ+1 c

c ρ0 Subsequently, expansion and adiabatic cooling can drive
τ= =
∆V nβ 2 Vrun ρrun the photons and electron back out of thermal equilibrium.
And in fact, the timescale for thermalization via Comptoniza-
= 21 (g9 /K13.8 )0.12V9−0.88 ρ−0.16
6 R−0.25
8.5 t
0.25
. (32) tion is always shorter than the thermal diffusion time at the
This gives a luminosity of diffusion depth. Since the photons and electrons were cou-
pled in the past, we don’t expect the spectrum to be altered
L(t) ≈ 3 × 1041 (g9 /K13.8 )−0.50V91.8 ρ0.42 1.0 −0.34
6 R8.5 t ergs s−1 , too greatly, but this should be quantified by more detailed cal-
(33) culations in the future.
6 PIRO, CHANG, & WEINBERG

F IG . 5.— The depth of the thermal diffusion wave during cooling following
shock-heating of the WD envelope as a function of time. Also marked by dot-
ted lines are estimates for the maximum mass helium shell (≈ 10−3 M⊙ , Iben
& Tutukov 1989; Shen & Bildsten 2009b) and the maximum mass hydrogen
F IG . 4.— The top panel plots the absolute magnitude of the expanding and shell (≈ 10−5 M⊙ ; Shen & Bildsten 2009a) that can survive on a 1.4 M⊙
cooling shock heated envelope. The upper curves show the bolometric lu- WD without igniting. This shows what layers of the WD surface may be
minosity, and the lower ones show the absolute magnitude in the wavelength probed by the thermal diffusion wave to help discern between possible SN Ia
range of 300 − 700 nm. The solid lines (dashed lines) are for R∗ = 3 × 108 cm progenitors.
(6 × 108 cm). The bottom panel shows the distance out to which an m = 24
limited optical survey could see such an event. year.
Detailed spectral modeling of the shock-heated cooling
4. OPTICAL LIGHT CURVE may also help discern between SN Ia progenitor models. In
Figure 5, we plot the thermal diffusion depth in units of mass.
In Figure 4 we show the bolometric and optical (300 − 700
This shows which mass shells of the progenitor WD are being
nm) absolute magnitude for two different progenitor radii, as-
suming that the WD emits as a blackbody. In the bottom panel probed by the cooling wave as a function of time. Depending
on the progenitor model in question, different compositions
we plot the distance out to which such an optical event can be
are expected in these surface layers, and once the thermal dif-
seen for an m = 24 limited exposure. The optical peaks at
around ≈ 1 day following shock breakout. The observabil- fusion wave is below them, these elements may imprint their
presence as spectral features. As a comparison, we plot dot-
ity depends sensitively on the WD radius at the time of shock
ted lines denoting the characteristic maximum thickness of a
breakout.
Although we have assumed a blackbody emitter for these hydrogen shell (≈ 10−5 M⊙ ; Shen & Bildsten 2009a) or he-
light curve estimates, the atmosphere will in fact be scatter- lium shell (≈ 10−3 M⊙ , Iben & Tutukov 1989; Shen & Bild-
ing dominated. This hardens the spectrum, making the cool- sten 2009b), as expected to be present in the single degenerate
ing tail dimmer in the optical. Also, for the wavelengths that scenario for SNe Ia.
dominate at ≈ 1 day, metals may play an important role in
setting the opacity, especially since they will experience some 5. SUMMARY AND CONCLUSIONS
recombination for the temperatures at this time. More sophis- We have calculated the cooling of a shock heated WD enve-
ticated spectral modeling is needed in order to predict the pre- lope due to a DDT during a SN Ia. The general features of the
cise shape of the light curve. resulting flash are a short (∼ 10−2 s), hard X-ray flash with a
In Figure 4 we have focused on the light curve due to total energy of ∼ 1040 ergs. The flash is followed by a cooling
from the cooling of the shock-heated envelope, but eventu- tail as the thermal diffusion wave travels back into the ex-
ally this will be overtaken by the nickel decay. Whether panding envelope. The light curve transitions from the X-rays
or not the shock-heated cooling can be seen above the ris- to the ultraviolet, and eventually, visual wavebands, reaching
ing light curve of nickel decay depends on the power of the a peak absolute magnitude of MV = −9 to −10 at ≈ 1 day.
early time nickel decay and the WD radius at the time of Whether or not this emission is observable as a distinct com-
the DDT. If we extrapolate the typical L ∝ t 2 law found em- ponent in the SNe Ia optical light curve depends on the WD
pirically (Conley et al. 2006), or the exponential luminosity radius at the time of DDT and the nature of the early-time
function of Arnett (1982), back to early times, they would go nickel decay. Nevertheless, the detection (or lack thereof) of
right through our optical light curves, indicating that nickel emission during the first ∼ 20 days before peak would be an
decay will be comparable to our optical cooling luminosity important constraint on the occurrence of DDTs and on the
at ≈ 1 day. Since such extrapolations to early times are un- amount of expansion during the deflagration phase.
certain, early sampling of the SN Ia light curves is needed A similar shock breakout event may also be associated with
to better address this issue. As our calculations show, de- the AIC of a WD to a neutron star. In fact, Tan et al. (2001)
tecting the signature of a SN Ia at such early times is pos- and Dessart et al. (2007) considered whether the breakout
sible with current and future surveys. Using a SN Ia rate of shock from core bounce in this case would be a gamma-
2.93 × 10−5 yr−1 Mpc−3 (Dilday et al. 2008), we estimate that ray burst progenitor, concluding the answer is no. But this
≈ 10 − 60 SNe Ia can be observed at these early times each does not preclude a less powerful, but nevertheless interest-
 SHOCK BREAKOUT FROM SNe IA 7

ing, shock breakout as we have described here. The energy 1998), and the shock breakout may be confined along the
is greater in the AIC case than the SN Ia case, because it is poles (Dessart et al. 2006, 2007), both of which will make it
powered by the gravitational binding energy of the WD (me- difficult to catch an AIC shock breakout.
diated by neutrinos). Using a typical energy of 1050 ergs de-
posited into 0.1 M⊙ (as found by Fryer et al. 1999), this gives
Vrun ≈ 109 cm s−1 , but with a much larger ρrun ≈ 107 g cm−3 .
This provides a late time light curve that is brighter by a factor We thank Lars Bildsten, Chris Matzner, and Eliot Quataert
of ≈ 2.5, making it easier to observe. However, the AIC rate for helpful discussions. This work was supported by the The-
is much lower than the SNe Ia rate (∼ 1%, Yungelson & Livio oretical Astrophysics Center at UC Berkeley.
REFERENCES
Arnett, W. D. 1982, ApJ, 253, 785 Niemeyer, J. C. 1999, ApJ, 523, L57
Campana, S., et al. 2006, Nature, 442, 1008 Niemeyer, J. C., & Woosley, S. E. 1997, ApJ, 475, 740
Chevalier, R. A. 1992, ApJ, 394, 599 Nomoto, K., Sugimoto, D., & Neo, S. 1976, Ap&SS, 39, L37
Colgate, S. A. 1974, ApJ, 187, 333 Nomoto, K., Thielemann, F.-K., & Yokoi, K. 1984, ApJ, 286, 644
Conley, A., et al. 2006, AJ, 132, 1707 Plewa, T., Calder, A. C., & Lamb, D. Q. 2004, ApJ, 612, L37
Dessart, L., Burrows, A., Livne, E., & Ott, C. D. 2007, ApJ, 669, 585 Röpke, F. K., Hillebrandt, W., Schmidt, W., Niemeyer, J. C., Blinnikov, S. I.,
Dessart, L., Burrows, A., Ott, C. D., Livne, E., Yoon, S.-C., & Langer, N. & Mazzali, P. A. 2007, ApJ, 668, 1132
2006, ApJ, 644, 1063 Shen, K. J., & Bildsten, L. 2009a, ApJ, 692, 324
Dilday, B., et al. 2008, ApJ, 682, 262 Shen, K. J., & Bildsten, L. 2009b, ApJ, 699, 1365
Falk, S. W. 1978, ApJ, 225, L133 Sakurai, A. 1960, Commun. Pure Appl. Math., 13, 353
Filippenko, A. V. 1997, ARA&A, 35, 309 Soderberg, A. M., et al. 2008, Nature, 453, 469
Fryer, C., Benz, W., Herant, M., & Colgate, S. A. 1999, ApJ, 516, 892 Tan, J. C., Matzner, C. D., & McKee, C. F. 2001, ApJ, 551, 946
Iben, I. J., & Tutukov, A. V. 1989, ApJ, 342, 430 Weinberg, N. N., & Bildsten, L. 2007, ApJ, 670, 1291
Imshennik, V. S., Nadezhin, D. K., & Utrobin, V. P. 1981, Ap&SS, 78, 105 Whitham, G. B. 1958, J. Fluid Mech. 4, 337
Katz, B., Budnik, R., & Waxman, E. 2009, arXiv:0902.4708 Woosley, S. E. 2007, ApJ, 668, 1109
Khokhlov, A. M. 1991, A&A, 245, 114 Woosley, S. E., & Weaver, T. A. 1994, in Les Houches Session LIV,
Klein, R. I., & Chevalier, R. A. 1978, ApJ, 223, L109 Supernovae, ed. S. Bludman, R. Mochovitch, & J. Zinn-Justin
Livne, E., Asida, S. M., & Höflich, P. 2005, ApJ, 632, 443 (Amsterdam: North Holland), 63
Mannucci, F., Della Valle, M., Panagia, N., Cappellaro, E., Cresci, G., Woosley, S. E., Wunsch, S., & Kuhlen, M. 2004, ApJ, 607, 921
Maiolino, R., Petrosian, A., & Turatto, M. 2005, A&A, 433, 807 Yungelson, L., & Livio, M. 1998, ApJ, 497, 168
Matzner, C. D., & McKee, C. F. 1999, ApJ, 510, 379
Mazzali, P. A., et al. 2006, Nature, 442, 1018