white dwarf stars in mass transferring binaries and their
TRANSCRIPT
White Dwarf Stars in MassTransferring Binaries and their
OutburstsDean Townsley
Joint Institute for Nuclear Astrophysics
The University of Chicago
Townsley - SUNY Stonybrook 2008 – p.1/22
Binary Evolution
Time
Orbital PeriodIncreasing
WideNo interaction
Ejection
CommonEnvelope
Secondary Mass
Supernova?
MS
WD
CE2
WD
WD
Merger to Supernova?
WD
Giant
Stable hydrogenburning"Supersoft" X−ray sources
WD
MS
Stable mass transfer
Recurrent surface H runaways
Stable H mass transfer
long interval surface H runaways
or collapse?
Post Common Envelope Binaries
Townsley - SUNY Stonybrook 2008 – p.2/22
Binary Evolution
Time
Orbital PeriodIncreasing
WideNo interaction
Secondary Mass
Supernova?
Merger to Supernova?
WD
Giant
Stable hydrogenburning"Supersoft" X−ray sources
WD
MS
Stable mass transfer
Recurrent surface H runaways
Stable H mass transfer
long interval surface H runaways
or collapse?
Townsley - SUNY Stonybrook 2008 – p.3/22
CV Angular Momentum LossJ determines evolution of compact binary
Companion B field
WINDWIND
WINDWIND
gravity
waves waves
gravity
Magnetic Brakinghigh J , Porb & 3 hours
Magnetically attached wind from compan-ion star
Jmb = −9.4 × 1038
erg
0
@
M2
M⊙
1
A
0
@
R2
R⊙
1
A
3 Porb
hr
!
−3
Gravitational Radiationlow J
Jgr = −32GQ2ω5
5c5
= −2.7 × 1037
erg
0
@
a
R⊙
1
A
4 0
@
MWDM2
MtM⊙
1
A
2 Porb
hr
!
−5
Townsley - SUNY Stonybrook 2008 – p.4/22
Interrupted Magnetic (Wind) Braking?
0 1 2 3 4 5 6P
orb (hr)
10-11
10-10
10-9
10-8
<⋅ M
> (
MO· y
r-1)
0 1 2 3 4 5 6
0.65
Mcompanion
=0.22 MO·
0.08
∆t=108 yr10
9 yr3·10
9yr
pre CV
~109yr
MagneticBraking
Grav.Radiation
Evolved fromprescriptions whichreproduced thecompanion contractionnecessary for the periodgap.
Predicts a strongcontrast in both 〈M〉and evolution time – andtherefore space density– of period bins
Difficult to test due toCV variability andcomplexity of disks, butprogress can be madeby other means such asWD Teff . (Townsley & Bildsten 2003,
ApJ, 596, L227)
MWD = 0.7M⊙ , Howell, Nelson, & Rappaport 2001, ApJ 550, 897
Systems evolve from long to short orbital periodsdue to angular momentum losses causing the or-bit to decay.Period gap caused by sudden drop in angularmomentum loss rate.
Townsley - SUNY Stonybrook 2008 – p.5/22
Available Parameter Space
0.6 0.8 1 1.2M
WD (M
O·)
10-11
10-10
10-9
10-8
10-7
10-6
⋅ M (
MO· y
r-1) M
ign=10
-5M
O·10
-4M
O·
Stable H burning
Contours spaced by ∆ log(Mign/M⊙) = 0.2Townsley & Bildsten 2005, ApJ, 628, 395
Strong contrast in Mign at aroundfew×10−10M⊙ yr−1 created bychange in ignition mode due to dif-ferent Tc as determined by 〈M〉(more on this later).
CVs generally are thought to haveaccretion rates that are low or high,but not much in between.
A system at a given mass canhave a factor of 10 range in Mign
depending on what evolutionarystage it is in.
Townsley - SUNY Stonybrook 2008 – p.6/22
Heat Sources
COMPRESSION
ADVECTION
HEAT TRANSPORT
5 10 15 20 25log P
104
105
106
107
T (
K)
C-O core
Adi
abat
material moves inwarddue to accretion
H-He Envelope
(very) leaky entropy advection
Heat liberated by compression is transferred out to surfaceand in to core. Often called “compressional heating”.
Heat sources:
Accretion light: only very near surface while activelyaccreting
Compression: throughout star, mostly in light-elementlayer (really gravitational potential energy)
Nuclear “simmering”: fusion near base of accreted layer(eventually becomes fast and triggers classical nova)
Core heat capacityTownsley - SUNY Stonybrook 2008 – p.7/22
Quasi-static Profile
-8 -7 -6 -5 -4 -3log (∆M/M
O·)
1
2
3
4
5
6
7
8
log
(t/y
r)
tacc
tth
Macc
Mign
Local thermal time short compared to
accretion
tth ≡cP T
“
4acT4
3κy2
” < tacc ≡∆M
〈M〉
where y = ∆M/4πR2 is the column depth.
Thermal state set by flux from deeperlayers rather than from fluid element’s
history.
0 staticHeat equation near surface:
−dL
dMr+ǫN = T
„
∂
∂t+vr
∂
∂r
«
s = T∂s
∂t+Tvr
∂s
∂r
where vr = −〈M〉/4πr2ρ. Solve with structure equations. Gives excellentrepresentation of envelope sructure.
L ≃kTc
µmp〈M〉
Energy release related to heat content of compressed material.Townsley - SUNY Stonybrook 2008 – p.8/22
Tc and Classical Nova IgnitionPhysical Conditions at base of H/HeEnvelope determine runaway
102
103
104
ρ (g cm-3
)
0.5
1
2
T (
107 K
)
Tc=?
Stable
Unstable
Evaluating envelope stability:
∂ǫN
∂T=
∂ǫcool
∂T
One-zone approximation,ǫcool ∝ 4acT 4/κy2, only worksin upport portion.
Lower part of curved bettermodeled byǫcool = L(Tc)/Macc, were L(Tc)is given by that of a coolingWD: radiative envelopeoverlying a conductive region.
Thermal state (Tc) has animportant influence on whenthe instability line is crossed.
Composition has significantinfluence on position of upperportion.
Townsley - SUNY Stonybrook 2008 – p.9/22
Two Kinds of Ignition
m
〈M〉 = 3 × 10−9M⊙ yr−1
Tc = 107
Direct to p + C or 3He +3 He
Most novae by number
〈M〉 = 5 × 10−11M⊙ yr−1
Tc = 5 × 107
p + p (partial chain) envelope heatingeventually leads to p + CLarge accumulated mass
Townsley - SUNY Stonybrook 2008 – p.10/22
Cooling-Heating Cycle
-1.5
-1
-0.5
0
0.5
Lco
re (
10-3
LO·)
0 0.5 1 1.5 2 2.5 3Accumulated Mass ( 10
-4M
O·)
12
13
14
15
Tef
f
WD Cooling WD Heating
1014
1015
1016
1017
1018
1019
1020
P (erg cm-3
)
106
107
T (
K)
Base of accreted layer
Core will be Reheated until equilibrium is reached.Core thermal time ∼ 108 yr
Townsley - SUNY Stonybrook 2008 – p.11/22
Cooling-Heating Cycle
-1.5
-1
-0.5
0
0.5
Lco
re (
10-3
LO·)
0 0.5 1 1.5 2 2.5 3Accumulated Mass ( 10
-4M
O·)
12
13
14
15
Tef
f
WD Cooling WD Heating
1014
1015
1016
1017
1018
1019
1020
P (erg cm-3
)
106
107
T (
K)
Base of accreted layer
Core will be Reheated until equilibrium is reached.Core thermal time ∼ 108 yr
Townsley - SUNY Stonybrook 2008 – p.11/22
Cooling-Heating Cycle
-1.5
-1
-0.5
0
0.5
Lco
re (
10-3
LO·)
0 0.5 1 1.5 2 2.5 3Accumulated Mass ( 10
-4M
O·)
12
13
14
15
Tef
f
WD Cooling WD Heating
1014
1015
1016
1017
1018
1019
1020
P (erg cm-3
)
106
107
T (
K)
Base of accreted layer
Core will be Reheated until equilibrium is reached.Core thermal time ∼ 108 yr
Townsley - SUNY Stonybrook 2008 – p.11/22
Cooling-Heating Cycle
-1.5
-1
-0.5
0
0.5
Lco
re (
10-3
LO·)
0 0.5 1 1.5 2 2.5 3Accumulated Mass ( 10
-4M
O·)
12
13
14
15
Tef
f
WD Cooling WD Heating
1014
1015
1016
1017
1018
1019
1020
P (erg cm-3
)
106
107
T (
K)
Base of accreted layer
Core will be Reheated until equilibrium is reached.Core thermal time ∼ 108 yr
〈Lcore〉 =1
tCN
Z tCN
0
Lcore dt
Townsley - SUNY Stonybrook 2008 – p.11/22
〈Lcore〉 and the equilibrium Tcore
〈Lcore〉 =1
tCN
Z tCN
0
Lcore dt
When Mej = Mign, 〈Lcore〉 = 0 defines anEquilibrium Tcore which is set by M and 〈M〉
Can approximate evolution:
〈Lcore〉(Tc, M) = CWDdTc
dt
where CWD is the total heat capacity of theWD – proportional to mass (have to becareful with latent heat at crystallization) 2 4 6 8 10 12
Tc (10
6 K)
-3
-2
-1
0
Ave
rage
Lco
re (
10-3
LO·)
10-11
10-10
10-9
10-8
10-10.4
Townsley - SUNY Stonybrook 2008 – p.12/22
WD Thermal State Evolution
10-12
10-11
10-10
10-9
10-8
10-7
<⋅ M
> (
MO· y
r-1)
2
3
4
5
Tc/1
06 K
5
10
50
Tef
f/103 K
0 2 4 6 8 10t (Gyr)
0
1
2
3
4
5
P orb (
hour
s)
Phases of accretion
1. Magnetic Braking 〈M〉 ∼ 5 × 10−9M⊙ yr−1
2. Period gap 〈M〉 = 0
3. Gravitational radiation 〈M〉 ≃ 5 × 10−11M⊙ yr−1
4. Post-period minimum 〈M〉 < 10−11M⊙ yr−1
Phases of WD evolution
1. Reheating – Teff set by 〈M〉
2. Equilibrium – Teff set by 〈M〉
3. Cooling – Teff set by core cooling
Accretion resets the clock for WD cooling
Townsley - SUNY Stonybrook 2008 – p.13/22
Tc EvolutionEpelstain, Yaron, Kovetz, Prialnik 2007, MNRAS, 374, 1449Full, multi-cycle nova simulations
M = 1.0M⊙, 〈M〉 = 10−11M⊙/yr M = 0.65M⊙, 〈M〉 = 10−9M⊙/yr
time (100 Myr)time (Gyr)
core
T (
10 K
)7
cooling WD
nova WD
Demonstrates equilibrium and evolution times. Unlikely to come fully into equilibriumabove gap, but plenty of time below gap, especially with the “boost” from above-gapevolution.
Also demonstrates that nova WDs in CVs generally will not stay very hot (& 2× 107) formore than a few 100 Myr. (Note being "caught" in this state would be exceedingly rarein any case due to post-CE cooling.)
Townsley - SUNY Stonybrook 2008 – p.14/22
Teff vs. PorbTownsley & Bildsten 2003, ApJ, 596, L227Townsley & Gänsicke, submitted
1 2 3 4 5Orbital Period (hours)
10
20
30
40
50
Tef
f/103 K
0.6-1.0 MO·
Low disk state systems (DN, SW Sex)⋄ Magnetics
Theory range shown: 0.6-1.0M⊙
Factor of ∼ 10 〈M〉 contrast across
period gap confirmed
Current Mag. Braking prescription
matches well with DN at 4-5 hours
Separate population of high 〈M〉 at
3 hours?
Magnetic CVs above gap nearGrav. Radiation prediction
– WD magnetic field preventingmagnetic braking?!(Li, Wu, & Wickramasinghe 1994, MNRAS, 268, 61)
Townsley - SUNY Stonybrook 2008 – p.15/22
Classical NovaPorb Distribution
0 1 2 3 4 5 6P
orb (hr)
0
2
4
6
8
10
Num
ber
of N
ovae
0 1 2 3 4 5 60
2
4
6
8
10
Relative number of CVs
Theory curve uses Interrupted MagneticBraking for Porb(〈M〉) and population nP
(Howell, Nelson, Rappaport 2001, ApJ 550, 897)
νCNP = nP〈M〉
Mign
But since nP ∝ M2/〈M〉 this gives
νCNP ∝1
Mign
Thus the dominant contribution is fromthe variation in the ignition mass acrossthe period gap (2-3 hours)
(Townsley & Bildsten 2005, ApJ, 628, 395)
Supports a factor of > 10 drop in 〈M〉 across gap
Consistent with idea that CVs evolve across the gap
Possible population of magnetic systems filling in gap
Ignores selection effects – hard to quantify
Townsley - SUNY Stonybrook 2008 – p.16/22
Classical Nova〈M〉 Distribution
0
0.5
1
Rel
ativ
e N
ova
rate
0
0.5
1
-11 -10 -9 -8log (< ⋅M>/ (M
O· yr
-1) )
0
1
2
3
Eje
ctio
n ra
te p
er N
ova
(10-5
MO·)
Φ(〈M〉)
Most observed Novae have “high”
〈M〉 ∼ 10−9M⊙ yr−1
Similar amount of matter is ejected fromNovae with 〈M〉 ∼ 10−9M⊙ yr−1 and
∼ 10−10M⊙ yr−1.
Character of ignition very different for
these two
direct Carbon or 3He trigger
p-p heated deep envelope trigger
Features of Novae which depend on 〈M〉
are expected to have a bimodal character.
The Porb distribution below 6 hours shows
initial indications of this.Townsley - SUNY Stonybrook 2008 – p.17/22
Luminosity Function of Old CVs
10-12
10-11
10-10
<⋅ M
> (
MO· y
r-1)
M = 0.6 MO·
0 2 4 6 8 10 12t (Gyr)
5
10
Tef
f/103 K
WD Cooling
11.5 12 12.5 13 13.5 14 14.5WD M
V
Rel
ativ
e N
11.5 12 12.5 13 13.5 14 14.5
M = 0.6 MO·
Gra
v. R
ad.
Near Porb
min.WD Cooling
< ⋅M> dropping
Low 〈M〉 leads to infrequent disk outbursts
CV V magnitude dominated by WD
Most old CVs appear as cooling WDs until inspected carefully
Townsley - SUNY Stonybrook 2008 – p.18/22
Broadband CV Spectral EvolutionTownsley & Bildsten 2002, ApJ, 565, L35
0 1 2 3 4V-I
20
21
22
23
24
25
26
I
0.2 MO·
0.3 MO·
0.15 MO·
Predict
ed C
V Evolution
MWD
=0.6 MO·
MWD
=0.4 MO·
= 1 Gyr
M4 Predicted CV Color Evolution0.6 M
O· WD No accretion disk included
Proper-motion selectedmembers of M4 at 4 coreradii (Richer et al. 2002, ApJ, 574L, 151)
Color selection criteria forold CVs
CVs Mixed with WDpopulation used to datecluster
Townsley - SUNY Stonybrook 2008 – p.19/22
Evolution of He Accretors (AM CVns)Bildsten, Townsley, Deloye, & Nelemans 2006, ApJ, 640, 466
10-12
10-11
10-10
10-9
10-8
10-7
10-6
<⋅ M
> (
MO· y
r-1)
M=0.65 MO·
M=1.05 MO·
104
105
Tef
f (K
)
DBV strip
0 10 20 30 40 50 60 70P
orb (hr)
6
8
10
12
14
16
18
MV
WDs which accrete helium from a
companion lower mass heilum WD
〈M〉 monotonically decreases with time as
Porb increases
Curves show 2 WD masses and 2 possible
donor thermal states(Deloye & Bildsten 2003, ApJ, 598, 1217)
Similar evolution: reheating, equilibrium(short!), WD cooling
Accretion disk phenomenology not wellunderstood, two-state (DN) accretion
expected with increasing time spent inquiescence
Both measured MV agree well with theory!
Townsley - SUNY Stonybrook 2008 – p.20/22
Accreting WD SeismologyCan greatly change characterof mode spectrum
Naturally gives sets of manyclosely spaced modes
Surface eigenfunctions havedistinctive shape (squeezed toequator) that can be confirmedby multi-band observations
Can give modes at frequenciesmuch higher and much lowerthan the driving frequency inthe rest rest frame
0 0.001 0.002 0.003 0.004 0.005observer frame mode frequency |ω-mΩ| (Hz)
0
0.002
0.004
0.006
0.008
0.01
spin
fre
quen
cy Ω
(H
z)0 0.001 0.002 0.003 0.004 0.005
0
0.002
0.004
0.006
0.008
0.01
Shown are the first ∼ 20 modes from the same physical model of GW Lib.The horizontal line represents a moderate spin hypothesis, in which Ω ∼ ω for the lowradial order modes.Heavy blue indicates a single mode triplet (m = −1, 0, +1).Plotted are frequencies in the observer’s frame |ω − mΩ|.Modes that "bounce" off of zero are modes travelling opposite to the rotational sense
Townsley - SUNY Stonybrook 2008 – p.21/22
SummaryGaining understanding about the accreting WD binary population constrains the
properties of outbursts
Accreting WDs are reheated by “compressional heating” and Hydrogen “simmering”
Equilibrium Tcore allows relation of observables to M, 〈M〉
Consistent with quiescent Teff , indicating variation in 〈M〉 across period gap
Reproduces classical nova Porb distribution
Evolution of broadband colors in quiescence
Late time magnitudes and Teff for both CVs and Helium accretors
Seismology can determine spin, M , Macc
Townsley - SUNY Stonybrook 2008 – p.22/22