The Peculiar Physics of Line-Driving

Stan Owocki

Bartol Research Institute

University of Delaware

Outline:• Radiative force from free electron scattering• Resonant amplification of line-scattering• Doppler sweeping of thick lines• CAK theory for steady, spherical wind• Line-driven instability

• Multi-D winds with vector line-force:• Winds from rotating stars

•Wind Compressed Disks (WCDs)• WCD inhibition by nonradial line-force• Spindown of wind rotation

• Colliding wind binaries• Radiative braking

• Line-driven ablation

• Summary

Colloborators: Ken Gayley, U. Iowa Joachim Puls, U. Munich Steve Cranmer, CfA

Winds that Sail on Starlight

Outline■ What is a Stellar Wind?■ Intercepting Light's Momentum■ Doppler Sweeping by Spectral Lines■ CAK Model for Steady, Line-Driven Wind ■ Instability of Line-Driving■ Simulating Wind Structure■ Summary

Collaborators

K. Gayley, U. Iowa

J. Puls, U. Munich

D. Cohen, Bartol/UDel.

Stan Owocki

Bartol Research Institute

University of Delaware

What are stellar winds?

Solar Wind* Sun has a very hot (10^6 K) corona* High Pressure => expansion* Supersonic, v ~400-700 km/s ~ v_esc >> v_sound* But mass loss rate is tiny, 10^-14 Msun/yr* Implies sun will lose only 0.01% of mass in whole 10^10

yr life

"A continuous outflow of mass from a star"

Hot-Star Winds* Massive stars (M~10-50 Msun) are hot (T~few 10^4 K) and

luminous (L~10^5-10^6 Lsun)

* Wind outflow diagnosed by asymmetric "P-Cygni" lines in UV

* Show v ~ 1000-3000 km/s!

* Much higher mass loss rates, up to 10^-4 Msun/yr

* Affects:

Stellar Evolution

ISM energy and mass balance

Bubbles may even trigger star formation – "starbursts"

* Driven by radiation pressure, scattered in spectral lines

Castor, Abbott, Klein (1975; "CAK") developed basic formalism

Formation of P-Cygni Line Profile

Intercepting Light's Momentum

* Light transports energy (& information)

* But it also has momentum, p=E/c

* Usually negligible, because speed c is so high.

* But becomes significant for very bright objects, e.g. Lasers, Luminous stars, Quasars/AGNs

* Key question: how big is force vs., e.g. gravity?

* Expressed through electron scattering Eddington factor

* For sun,

* But for hot stars with

Γ <~

1

Γ ≡g

el

ggrav

≡

L

4 π r2

c

σTh

μe

GM

r2

=κ

eL

4 π GM c

ΓO•=2.7×10

−5

L =105

−106

LO•

M =10 −50 MO•

Free Electron Scattering

Thompson Cross Section

σ = 2/3 barn = 0.66e-24 cm2Th

Th

Line Scattering

For High Quality Line Resonance:

Cross Section >> Electron Scattering

Q ~ Z ×Q ~10−4

×107

~103

σlines

~ Q × σTh

glines

~ 103

× gel

F = Fthin

Q ~ν × t ~1015

Hz ×10−8

s ~107

} iffΓ

lines~ 10

3

×Γel

>> 1

Doppler Shifting of Line-Absorptionin an Accelerating Stellar Wind

Line Scattering in an Expanding Wind

Optically thick line-force

Sobolevapproximation

independent of κ!& !like inertia

CAK model of steady-state wind

+ =>

inertia gravity line-force

Line-Driven Instability fromPerturbed Profile Doppler Shift

Time snapshot of a wind instability simulation

0.0 0.5 1.0

0

500

1000

1500

-15

-14

-13

-12

-11

-10

Height (R*

)

Velocity & Density vs. HeightCAK

Steady-State

Snapshot of Velocity and DensityPlotted vs. Height and Mass

in a Time-Dependent Wind Simulation

Flow "Structure" on the Autobahn

Back Scattering from Multiple Line Resonancesin a Non-montonic Velocity Field

Formation of Black Troughs in Saturated P-Cygni Line-Profiles

from Structured Stellar Winds

P-Cygni Profile Synthesized for aSmooth (---) and Structured ( ___ )

Stellar Wind Models

Black Troughfrom Structured Wind

Profile for smooth,CAK Wind

Instability Models with Energy Equation

Feldmeier 1995

Computational Requirements forStellar Wind Simulations

CAK/Sobolev Models

* line-force computed from local density and velocitygradient

* modest timing requirements, comparable to standard hydro

* allows for 2D (in principle even 3D) models, e.g. withrotation, disks, even B-field

Instability Simulations

* Line-force requires nonlocal solution of radiation transfer,in principle in hundreds of spectral lines of varying strength

* Current approximations use integral escape probabilities

* Requires computation of line optical depth

* Analytically averaged over power-law line ensemble

* But still requires nested integrations over

- angle (or ray)

- frequency

- depth

* Thus far most models artificially restricted to 1D

* Efforts toward 2D instability simulations

- 3-ray aligned grid

- Short characteristics

- 2nd order Sobolev (A. Feldmeier)

Ray Integration Grids for2D Radiation Hydro Models

Co-Rotating Interaction Region Models

localCAK

model

nonlocalsmoothmodel

nonlocalstructured

model

c. Δ ( )log Density. b. a

Ongoing Projects in Stellar Winds

QuickTime™ and aGIF decompressor

are needed to see this picture.

Wind Compressed Disks

O star

O star*WR

star

*WR star

Radiative Braking in Colliding Wind Binaries

Wind Rotation Spindown fromAzimuthal Line-Torque

gφ

(10 3 /cm s2)

[Vφ( )-nrf V φ( )] wcd

* (sin θ)* /r R eq

( / )km s

.a .b

-10

-30

-50

-70

-90

-0.1

-0.3

-0.5

-0.7

-0.9

Summary

* Massive, hot, luminous stars have strong stellar winds

* Driven by line-scattering of stellar radiation

* Highly unstable, leading to:

- high speed rarefactions

- slower dense clumps

- separated by Reverse Shocks

* Non-monotonic velocity evident in UV line Black Troughs

* But reverse shocks produce few X-rays

* Ongoing problems

- 2D (& 3D) models of compressible turbulence

- explain X-ray scaling laws

- how small-scale instability affects global windstructure, e.g. wind collisions, disks, etc.

- Role of line-driving in other luminous systems, e.g.CV disks; AGNs/QSOs