david cohen swarthmore college

X-ray Spectroscopy of the X-ray Spectroscopy of the Radiation-Driven Winds of Massive Stars:Radiation-Driven Winds of Massive Stars:

Line Profile and Line Ratio DiagnosticsLine Profile and Line Ratio Diagnostics

David CohenDavid CohenSwarthmore CollegeSwarthmore College

Carina/Keyhole Nebula (HST)Carina/Keyhole Nebula (HST)

massive stars: massive stars:

20 to 100 M20 to 100 Msunsun

101066 L Lsunsun

T ~ 50,000 KT ~ 50,000 K

young, massive star: young, massive star: 11 Ori C Ori C

Chandra X-ray Telescope image of the Orion Nebula ClusterChandra X-ray Telescope image of the Orion Nebula Cluster

Color coded according to photon Color coded according to photon energy (red: <1keV; green 1 to 2 energy (red: <1keV; green 1 to 2 keV; blue > 2 keV)keV; blue > 2 keV)

Whirlpool/M51 (HST)Whirlpool/M51 (HST)

Crab Nebula (WIYN)Crab Nebula (WIYN)

1000 yr old supernova remnant1000 yr old supernova remnant

NGC 6888 Crescent Nebula (Tony Hallas)NGC 6888 Crescent Nebula (Tony Hallas)

wind-blown bubble: stellar wind impact on its environmentwind-blown bubble: stellar wind impact on its environment

Radiation-driven massive star Radiation-driven massive star windswinds

MM ~ 10 ~ 10-6-6 M Msunsun/yr/yr

Prinja et al. 1992, ApJ, 390, 266Prinja et al. 1992, ApJ, 390, 266

Velocity (km/s)Velocity (km/s)

v∞=2350 km/s

UV spectrum: C IV 1548, 1551 ÅUV spectrum: C IV 1548, 1551 Å

Power in these winds:Power in these winds:

€

12 M

•

v∞2 ≈ 3×1036

≈ .001L∗

erg serg s-1-1

while the x-ray luminositywhile the x-ray luminosity

€

LX ≈10−7L∗

To account for the x-rays, only To account for the x-rays, only one part in 10one part in 10-4-4

of the wind’s mechanical power is needed to of the wind’s mechanical power is needed to heat the windheat the wind

LLsunsun = 4 x10 = 4 x103333 erg s erg s-1-1

LLmassivemassive ≈ 4 x10 ≈ 4 x103939

Three models for massive star x-ray emissionThree models for massive star x-ray emission

1. Instability driven shocks1. Instability driven shocks

2. Magnetically channeled 2. Magnetically channeled wind shockswind shocks

3. Wind-wind interaction in 3. Wind-wind interaction in close binariesclose binaries

Three models for massive star x-ray emissionThree models for massive star x-ray emission

1. Instability driven shocks1. Instability driven shocks

2. Magnetically channeled 2. Magnetically channeled wind shockswind shocks

€

Γedd ≡FradFgrav

=κL

4πcGM

Winds of massive stars are driven by Winds of massive stars are driven by radiation pressureradiation pressure

luminosityluminosityopacityopacity

massmass

For the Sun, For the Sun, ΓΓedd edd ~ 10~ 10-5-5

For massive stars, For massive stars, ΓΓedd edd approaches unity approaches unity

The Line-Driven Instability (LDI; Milne 1926)The Line-Driven Instability (LDI; Milne 1926)

€

v = v∞(1− R∗ /r)β

Consider a positive velocity perturbationConsider a positive velocity perturbation

Positive feedback: ion moves out of the Doppler shadow, Positive feedback: ion moves out of the Doppler shadow, sees more radiation, gets accelerated…sees more radiation, gets accelerated…

1-D rad-hydro simulation of a massive star wind1-D rad-hydro simulation of a massive star wind

Radiation line driving is inherently unstable: Radiation line driving is inherently unstable: shock-heating and X-ray emissionshock-heating and X-ray emissionOwocki, Castor, & Rybicki 1988Owocki, Castor, & Rybicki 1988

Predictions of the rad-hydro wind simulations:Predictions of the rad-hydro wind simulations:1.1. Significant Doppler broadening of x-ray emission Significant Doppler broadening of x-ray emission

lines due to bulk motion of the wind flow (1a. Shock lines due to bulk motion of the wind flow (1a. Shock onset several tenths Ronset several tenths R** above the surface) above the surface)

2.2. Bulk of the wind is cold and unshocked – source of Bulk of the wind is cold and unshocked – source of attenuation of the X-rays.attenuation of the X-rays.

Puppis in the Gum NebulaPuppis in the Gum Nebula

Puppis: 50 MPuppis: 50 Msunsun, 10, 106 6 LLsunsun

PupPup

Chandra Chandra HETGS/MEG spectrum HETGS/MEG spectrum ((RR ~ 1000 ~ 300 km s ~ 1000 ~ 300 km s-1-1))

SiSi MgMg NeNe OOFeFe

H-likeH-likeHe-likeHe-like

PupPup

Low-mass star (Capella) for comparisonLow-mass star (Capella) for comparison

PupPup

CapellaCapella

Ne XNe X Ne IXNe IX Fe XVIIFe XVII

PupPupmassivemassive

CapellaCapellalow masslow mass

PupThe x-ray emission The x-ray emission lines are broad: lines are broad: agreement with rad agreement with rad hydro simulationshydro simulations

But… they’re also blue shifted and asymmetricBut… they’re also blue shifted and asymmetricIs this predicted by the wind shock scenario? Is this predicted by the wind shock scenario?

€

τ∗≡κM

⋅

4πR∗v∞

opacity opacity of the cold of the cold wind componentwind component wind mass-loss ratewind mass-loss rate

wind terminal velocitywind terminal velocityradius of the starradius of the star

€

M•

= 4πr2vρ

The basic wind-profile modelThe basic wind-profile modelτ=1,2,8

key parameters: key parameters: RRoo & & ττ**

€

τ∗≡κM

⋅

4πR∗v∞

j ~ 2 for r/R* > Ro,

= 0 otherwise

€

τ =τ∗R∗dz'

r'2 (1− R∗r ')β

z

∞

∫

Ro=1.5

Ro=3

Ro=10

ττ=1 contours=1 contours

We fit these x-ray line profile models to each line in the We fit these x-ray line profile models to each line in the Chandra Chandra datadata

And find a best-fit And find a best-fit ττ** and and RRoo & place confidence limits on & place confidence limits on

these fitted parameter valuesthese fitted parameter values

Fe XVIIFe XVII 68, 90, 95% confidence limits68, 90, 95% confidence limits

Wind opacity: photoelectric absorptionWind opacity: photoelectric absorption

Abundances; ionization balance; atomic cross sectionsAbundances; ionization balance; atomic cross sectionsVerner & Yakovlev 1996Verner & Yakovlev 1996

Pup: three emission lines Pup: three emission lines

Mg LyMg Ly: 8.42 Å: 8.42 Å Ne LyNe Ly: 12.13 Å: 12.13 Å O LyO Ly: 18.97 Å: 18.97 Å

ττ** = 1 = 1 ττ** = 2 = 2 ττ** = 3 = 3

Recall: Recall:

€

τ∗≡κM

⋅

4πR∗v∞

Fits to 16 lines in the Fits to 16 lines in the Chandra Chandra spectrum of spectrum of PupPup

Traditional mass-loss rate: Traditional mass-loss rate: 8.3 X 108.3 X 10-6-6 Msun/yr Msun/yr

Our best fit: Our best fit: 3.5 X 103.5 X 10-6-6 Msun/yr Msun/yr

Part 2: Magnetically Channeled WindsPart 2: Magnetically Channeled Winds

The central star in the Orion Nebula Cluster - The central star in the Orion Nebula Cluster - 11 Ori C Ori C - is a - is a source of strong and relatively hard x-rayssource of strong and relatively hard x-rays

Chandra Chandra image: color coded by photon image: color coded by photon energy (red: <1keV; green 1 to 2 keV; energy (red: <1keV; green 1 to 2 keV; blue > 2 keV)blue > 2 keV)

Recently discovered dipole magnetic field Recently discovered dipole magnetic field of > 1 kG : of > 1 kG : Zeeman Doppler spectroscopy (Wade et al. 2006)Zeeman Doppler spectroscopy (Wade et al. 2006)

Simulation/visualization courtesy R. TownsendSimulation/visualization courtesy R. TownsendMovie available at astro.swarthmore.edu/~cohen/presentations/apip09/Movie available at astro.swarthmore.edu/~cohen/presentations/apip09/rrm-o25-i75-b60-redt.avirrm-o25-i75-b60-redt.avi

This field confines and channels the stellar wind out to This field confines and channels the stellar wind out to several stellar radii – several stellar radii –

what is the effect? what is the effect?

……MHD simulations (ZEUS, 2-D)MHD simulations (ZEUS, 2-D)

Simulation/visualization courtesy A. ud-DoulaSimulation/visualization courtesy A. ud-DoulaMovie available at astro.swarthmore.edu/~cohen/presentations/apip09/Movie available at astro.swarthmore.edu/~cohen/presentations/apip09/ t1oc-lowvinf-logd.avit1oc-lowvinf-logd.avi

Simulation/visualization courtesy A. ud-DoulaSimulation/visualization courtesy A. ud-DoulaMovie available at astro.swarthmore.edu/~cohen/presentations/apip09/Movie available at astro.swarthmore.edu/~cohen/presentations/apip09/ t1oc-lowvinf-logT.avit1oc-lowvinf-logT.avi

Simulation/visualization courtesy A. ud-DoulaSimulation/visualization courtesy A. ud-DoulaMovie available at astro.swarthmore.edu/~cohen/presentations/apip09/Movie available at astro.swarthmore.edu/~cohen/presentations/apip09/ t1oc-lowvinf-speed.avit1oc-lowvinf-speed.avi

temperaturetemperature emission measureemission measure

MHD simulations of magnetically channeled windMHD simulations of magnetically channeled wind

Channeled collision is close to head-on – Channeled collision is close to head-on – >1000 km s>1000 km s-1-1 : T > 10 : T > 1077 K K

simulations by A. ud-Doula; Gagné et al. (2005)simulations by A. ud-Doula; Gagné et al. (2005)

Predictions: Predictions:

1.1.Shocks are strong – head-on – and so Shocks are strong – head-on – and so plasma is hotter;plasma is hotter;2.2.Hot plasma is moving much slower Hot plasma is moving much slower (confinement);(confinement);3.3.Rotational modulation of X-ray flux;Rotational modulation of X-ray flux;4.4.Hot plasma is ~1 RHot plasma is ~1 R** above the surface. above the surface.

Pup

1 Ori C

ChandraChandra grating spectra grating spectra

11 Ori C: hotter plasma, narrower emission lines Ori C: hotter plasma, narrower emission lines

Pup: cooler plasma, broad emission linesPup: cooler plasma, broad emission lines

Pup

1 Ori C

Si XIIISi XIVSi XIVMg XIMg XIMg XIIMg XII

H-likeH-like//He-like He-like ratio is temperature sensitiveratio is temperature sensitive

Pup

1 Ori C

Si XIIISi XIVSi XIVMg XIMg XIMg XIIMg XII

The magnetic O star – The magnetic O star – 11 Ori C – is hotter Ori C – is hotter

Differential emission measure Differential emission measure (temperature distribution)(temperature distribution)

MHD simulation of MHD simulation of 11 Ori C Ori C reproduces the observed reproduces the observed

differential emission measuredifferential emission measureWojdowski & Schulz (2005)Wojdowski & Schulz (2005)

1000 km s-1

Emission lines Emission lines are are significantly narrower in significantly narrower in the magnetic massive the magnetic massive star’s x-ray spectrumstar’s x-ray spectrum

1 Ori C(O7 V)

Pup(O4 If)

0.0

0.5

1.0

1.5

Sim

ulat

ion

EM

(10

56 c

m-3)

0.0

0.1

0.2

0.3

0.4

θ1 Ori

C A

CIS

-I c

ount

rat

e (s

-1)

0.0 0.2 0.4 0.6 0.8 1.0Rotational phase (P=15.422 days)

ChandraChandra broadband count rate vs. rotational phase broadband count rate vs. rotational phase

Model from MHD simulationModel from MHD simulation

0.0

0.5

1.0

1.5

Sim

ulat

ion

EM

(10

56 c

m-3)

0.0

0.1

0.2

0.3

0.4

θ1 Ori

C A

CIS

-I c

ount

rat

e (s

-1)

0.0 0.2 0.4 0.6 0.8

1.0 Rotational phase (P=15.422 days)

The star itself occults the hot plasma torusThe star itself occults the hot plasma torus

The closer the The closer the hot plasma is to hot plasma is to

the star, the the star, the deeper the dip deeper the dip

in the x-ray light in the x-ray light curvecurve

Emission measureEmission measure

contour encloses T > 10contour encloses T > 1066 K K

Helium-like species’ forbidden-to-intercombination Helium-like species’ forbidden-to-intercombination line ratios – z/line ratios – z/(x+y) (x+y) – provide information about – provide information about the the location location of the hot plasma of the hot plasma

……not the not the densitydensity, as is usually the case. , as is usually the case.

g.s. 1s2 1S

1s2s 3S1s2p 3P

1s2p 1P

resonance (w)

intercombination (x+y)forbidden (z)

10-20 eV

1-2 keV

Helium-like ions (e.g. OHelium-like ions (e.g. O+6+6, Ne, Ne+8+8, Mg, Mg+10+10, Si, Si+12+12, S, S+14+14) – ) – schematic energy level diagramschematic energy level diagram

1s2s 3S

1s2p 3P1s2p 1P

resonance (w)

intercombination (x+y)forbidden (z)

g.s. 1s2 1S

Ultraviolet light from the star’s photosphere drives Ultraviolet light from the star’s photosphere drives photoexcitation out of the photoexcitation out of the 33S level S level

UV

1s2s 3S

1s2p 3P1s2p 1P

resonance (w)

intercombination (x+y)forbidden (z)

g.s. 1s2 1S

The f/i ratio is thus a diagnostic of the local UV mean The f/i ratio is thus a diagnostic of the local UV mean intensity…intensity…

UV

1s2s 3S

1s2p 3P1s2p 1P

resonance (w)

intercombination (x+y)forbidden (z)

g.s. 1s2 1S

……and thus the distance of the x-ray emitting plasma and thus the distance of the x-ray emitting plasma from the photospherefrom the photosphere

UV

Model of f/i ratio dependence on dilution factor (radius)Model of f/i ratio dependence on dilution factor (radius)

Rfir=1.2 R*

Rfir=4.0 R*

Rfir=2.1 R*

He-like f/i ratios and the x-ray light curve both He-like f/i ratios and the x-ray light curve both indicate that the hot plasma is somewhat closer to indicate that the hot plasma is somewhat closer to

the photosphere than the MHD models predict. the photosphere than the MHD models predict.

Conclusions

Normal massive stars have x-ray line profiles Normal massive stars have x-ray line profiles consistent with the predictions of the wind instability consistent with the predictions of the wind instability model.model.

Photoelectric absorptionPhotoelectric absorption’s effect on the profile shapes ’s effect on the profile shapes can be used as a mass-loss rate diagnostic: can be used as a mass-loss rate diagnostic: mass-mass-loss rates are lower than previously thoughtloss rates are lower than previously thought. .

Magnetic massive stars have harder spectra with Magnetic massive stars have harder spectra with narrower lines and rotationally modulated variability, in narrower lines and rotationally modulated variability, in general agreement with MHD simulations. general agreement with MHD simulations.

Line ratio Line ratio diagnostics are useful for localizing the hot diagnostics are useful for localizing the hot plasma, and indicate that the plasma, and indicate that the MHD simulations predict MHD simulations predict a location that is too far from the photospherea location that is too far from the photosphere. .