Models of Stellar Atmospheres

Department of Astronomy
Vienna University

21.11.2007

Models of Stellar Atmospheres

Prerequisites...

In 21th century we still get most of the information via

analysis of electromagnetic radiation
80% of electromagnetic radiation is coming from stars

Analyzing electromagnetic radiation we should understand that:

We directly analyze only the region where the radiation
becomes the one we register via our telescopes. This region is
transparent for the radiation to leave the star and is called
stellar atmosphere.
Everything that is taking place deeper the stellar atmosphere
we can analyze only indirectly

Models of Stellar Atmospheres

Definition: Stellar Atmosphere

General
Stellar atmosphere: it is a part
of the star which does not have
its own energy sources. Only
redistribution of radiative en-
ergy takes place.

Practical
Stellar atmosphere: it is a part
of the star where the line spec-
tra is formed (outcoming radia-
tion which we observe)

Models of Stellar Atmospheres

Definition: Model of Stellar Atmosphere

Since the radiation we see interacts with the plasma of stellar

atmosphere, we need a tool to describe this process and to
interpret the observed data we get.

We will call model of stellar atmosphere a self-consistent

mathematical model of stellar atmosphere which contains the
information about the distribution of main physical quantities
(T,P,...) with geometrical depth counting from some zero-level.

Self-consistency means that having a set of free parameters the

obtained solution is unique.

Models of Stellar Atmospheres

Importance of Model Atmospheres

1. Model atmospheres provide a link between theoretical and

observational astrophysics
2. Model atmospheres are the upper boundary conditions for
modeling of stellar structure and evolution
3. Model atmospheres are the intermediate region between
stellar envelopes and interstellar medium
4. Model atmosphere is the only tool now for abundance analysis
in wide range of physical conditions that is important for
studying the chemical evolution of our Galaxy and Universe as
a whole

Models of Stellar Atmospheres

Model Parameters

Energy Conservation
Since there are no energy sources in stellar atmosphere, the total
amount of radiative energy falling at the bottom of the atmosphere
from the inside should be the same at each point throughout the
atmosphere:
div (Frad ) = 0

A black-body Concept
A useful concept is that of black-body, as defined as a perfect ab-
sorber and emitter of light. It radiates energy at the same rate as
it is being absorbed. The ratio of emitted and absorbed energy at
each frequency depends only on temperature and is known as Plank
function

Models of Stellar Atmospheres

Model Parameters
emission 2hν 3 1
= Bν (T ) = 2 hν/kT
absorption c e −1
The total energy radiated by a black-body goes as a fourth power
Z temperature known as Stefan-Boltzmann Law:
of
∞
Bν (T )dν = const × T 4
0

If the radiation at the bottom of the atmosphere is black-body,

then we can introduce a quantity called effective temperature Teff
so
Z that at each atmospheric layer we have:
∞
Fν dν = const × Teff 4
0

Teff is the temperature of black-body which radiates (from the unit

surface area) the same amount of energy as the star of interest and
thus is an energetic characteristic of the star

Models of Stellar Atmospheres

Model Parameters

Most of the stars do not change their radii on time scales

compared with their life-times. Thus, we assume that the
atmosphere of the star is in hydrostatic equilibrium, i.e. the
gravitational force is balance by kinetic properties of plasma.
GMstar
ratm  Rstar , g=
Rstar

Stars with extensive atmospheres

Giants, supergiants, hot stars with strong stellar wind, etc.: R, M

Stars with extended atmospheres

Pulsating stars, supernovas, etc.: should be considered in hydrody-
namic (t-dependance)

Models of Stellar Atmospheres

Model Parameters

Properties of radiation field and plasma in stellar atmosphere

depend upon chemical composition (abundances) used {εi }
1. abundance of each element is given relative to the total
number of atoms of all chemical elements (Ni /Ntotal )
2. abundance of each element is given relative to the number of
atoms of all chemical elements in a volume contains 1012
atoms of hydrogen
3. abundance of each element is given relative to the total mass
of stellar plasma in unit volume

Self-consistency in spectroscopy
Model atmosphere → abundances → model atmosphere...

Models of Stellar Atmospheres

“Classical” Assumptions

Plane-parallel geometry (ratm  Rstar ). All the physical

quantities depend only on geometrical depth h
Homogeneous abundances
Hydrostatic equilibrium (no large-scale motions)
The atmosphere is time-independent (statistical equilibrium)
There are no sources or sinks of energy
Energy transport takes place only by radiation and convection
(no heat conduction, acoustic waves, MHD waves, etc.)
The free electrons as well as the free heavier particles obey
the Maxwell distribution with local kinetic temperature Te .
Local Thermodynamical Equilibrium (LTE) is assumed

Models of Stellar Atmospheres

Hydrostatic Equation

dPtotal
= g, dM = −ρdx
dM
Ptotal contains:
Gas pressure Pgas = nkT
4π
R R
Radiation pressure Prad = grad (M)dM, grad = c κν Hν dν
1
Micro-turbulent pressure Pturb = 2 ρξturb
Convective pressure Pconv
υ2
R
Pressure due to rotation Prot = R dM
Macro-turbulent pressure Pmacro
Magnetic pressure Pmag

Models of Stellar Atmospheres

State Equation

Boltzmann formula for the ratio of occupation numbers:

ni gi Ej − Ei
= exp(− )
nj gj kT

Saha equation for the ratio of ion numbers:

nI 2 gI 2πmkT 3/2 EJ − EI
= ( 2
) exp(− )
nJ ne gJ h kT

Note
ne = f (T , P, i ) and is unknown. However, we know Ntotal and thus
should solve the system of non-linear equations to find ne . Iterative
solution is needed.

Models of Stellar Atmospheres

Equation of Radiative Transfer

Going through gas of the atmosphere, the radiation is absorbed,

re-emitted and scattered many time. These define the way how the
radiation at each frequency is transferred through the atmospheric
region.

dEν
Iν =
cos θdσdνdωdt
[Iν ] = erg cm−2 s −1 Hz −1 sterad −1

Number of photons times hν per dνdωdt

and per cos θdσ

Models of Stellar Atmospheres

Equation of Radiative Transfer

dIν
µ = Iν − Sν
dτν
ην
Source function: Sν =
κν
Optical depth: dτν = −κν ρds
µ = cos θ

Formal solution of RTE is:

Z τν
dt
Iν (µ) = −e τν /µ Sν e −t/µ
c µ

Models of Stellar Atmospheres

Moments of RTE

In model atmospheres we are meanly interested in integral

characteristics of radiation field
Z Z Z
1 1 1
Jν = Iν dω, Hν = µIν dω, Kν = µ2 Iν dω
4π 4π 4π
Jν – mean intensity is an energy emitted by unit surface area in
any direction

Hν – radiative flux is an energy emitted by unit surface area in all

directions

Kν – is called K-integral and does not have physical meaning

(rad) 4π
However: Pν = Kν – radiative pressure acting on unit
c
surface area

Models of Stellar Atmospheres

Moments of RTE

Source Function
It is seen that to solve RTE one needs to know source function
Sν which also depends upon the solution of RTE due to additional
scattering term σν Jν

dIν dHν dKν

µ = Iν − Sν , = Jν − Sν , = Hν
dτν dτν dτν
κν Bν + σν Jν
Sν =
κν + σ ν
∞
X 1 d n Bν
Diffusive approximation (τν  1): Sν (tν ) = (tν − τν )n
n! dτνn
n=0
1 dBν (τν ) 1
Jν (τν ) ≈ Bν (τν ), Hν (τν ) ≈ , Kν (τν ) ≈ Bν (τν )
3 dτν 3

Models of Stellar Atmospheres

Radiative Equilibrium

Z Z
dH
= (κν + σν )Jν dν − (κν + σν )Sν dν = Q(M)
dM

Q(M)–function describing the losses of radiative energy via its

transformation to other types

Our experience shows that in most cases Q(M) ≈ 0 so that the

amount of absorbed energy is equal to that of emitted:
Z Z
(κν + σν )Jν dν = (κν + σν )Sν dν

dH
=0
dM

Models of Stellar Atmospheres

Convective Energy Transport
Schwarzschild instability criteria
   
γ−1 dlnP dlnT Cp
− < − , γ=
γ dr R dr R
Cv

Convection take place in the regions with high opacity due to fast
ionization of main plasma components (H, He)
Early type of stars: H is completely ionized so we can expect
only weak convective zones due to He, He+ ionizations
A-type: hydrogen convective zone close to photosphere
F-type: more strong and deep hydrogen convective zone
G-type: convection transfers most of the energy in deep layers
and penetrate surface layers
M-type: convection rules the whole structure of the
atmosphere

Models of Stellar Atmospheres

Convective Energy Transport
Introducing convection in model atmosphere calculations we are
mainly interested in which per sent of the total energy is carried
out by convection

Models of Stellar Atmospheres

Convection Treatment

Mixing-Length Theory (MLT)

Based on local pressure scale height:
Ptotal 1
Hp = =
ρg α
α has to be tuned for different stars
Application: cool stars atmosphere with strong convection regime

CM convection
Canuto & Mazzitelli (1991, 1992) – further improvement of convec-
tion model: no need for such free parameter as α any more
Application: atmospheres of A–F stars with weak convection regime

Models of Stellar Atmospheres

Temperature Correction

Problem
T (M) which satisfies the conditions of RE is initially unknown

T (M) = T0 (M) + ∆T (M)

Z Z
(κν + σν )Sν (T0 + ∆T )dν = (κν + σν )Jν dν

∂Sν
Sν (T0 + ∆T ) ≈ Sν (T0 ) + ∆T
∂T
Z
(κν + σν )(Jν + Sν )dν
∆T ≈ Z
∂Sν
(κν + σν ) dν
∂T

Models of Stellar Atmospheres

Convergence Criteria

Optically thick layers

(Hrad + Hconv ) − σTeff4 =0

Optically
Z thinZlayers
κν Jν dν − κν Sν dν = 0

Convergence criteria:
1. ε(flux)6 1%
2. ε(rad. equilibrium)6 1%
3. ∆Ti 6 1 K at each layer

Models of Stellar Atmospheres

Definition of the Problem

1. All equations are bound

Sν , Jν , Hν , Kν → f (αν , ην )
αν , ην → f (T , P)
T , P → f (Sν , Jν , Hν , Kν )
2. The system of equations is non-linear
Jν → Sν → Jν . . .
3. The system of equations is globally dependent throughout the
atmosphere. Scattering processes force the properties of
radiation field and plasma at one given point to be dependent
from all other points.

Models of Stellar Atmospheres

Simple Model Atmosphere Calculation Flow Block
Teff , log g (or R, M), {i }, ξturb

First approximation: T (τstd )

Hydrostatic: P(τstd )

State Equation: ne , ni , n0

Model Iteration
ν–integration
Opacity: ανcont , ανline

Radiation Field: Jν , Hν , Kν , Sν

Convection: Hconv
Tnew = Told + ∆T
T-correction: ∆T

Models of Stellar Atmospheres

Interaction Between Matter and Radiation

Absorption
True Absorption
Atom gets photon, goes to exited state and hits another atom
transmitting its energy to the later one.
Plasma heating
Negative Absorption (or Stimulated Emission)
Atom is exited by particle collision and emits the photon of
energy hν in direction ω
Plasma cooling

Models of Stellar Atmospheres

Interaction Between Matter and Radiation

Emission
Atom spontaneously emits a photon due to finite life-time. The
atom should be in exited state (whatever was the reason).

Scattering
Direct Scattering
Atom gets photon from the beam, goes to exited state and
then re-emits its energy in arbitrary direction. Depending
upon the probability of direction of the re-emitted photon we
have isotropic and anisotropic scattering. If the energy of
re-emitted photon is equal to that of absorbed one, we have
coherent scattering. Otherwise incoherent.
Reverse Scattering
The photon is added to the beam due to direct scattering of
another beam on a given atom

Models of Stellar Atmospheres

Interaction Between Matter and Radiation

Absorption and Emission processes can take place between the

following energy levels:
discrete-discrete (bound-bound) ⇒ spectral lines
discrete-continuous (bound-free) ⇒ ionization/recombination
continua
continuous-continuous (free-free) ⇒ ion required

Models of Stellar Atmospheres

Line Blanketing and Model Atmospheres
Opacity due to atomic lines absorption is one of the most
important effects in stellar atmospheres
Deep layers – backwarming
Surface layers – cooling

Models of Stellar Atmospheres

Line Opacity Treatment

Problem
Hundred of thousands of lines which have to be taken into account

Solution
Several statistical methods have been developed

Opacity Distribution Function (ODF)

Opacity Sampling (OS)
Direct Opacity Sampling

Models of Stellar Atmospheres

Line Opacity Treatment

Models of Stellar Atmospheres

Line Opacity Treatment

Models of Stellar Atmospheres

Energy Redistribution

Models of Stellar Atmospheres

Model Structure

Models of Stellar Atmospheres

Model Atmospheres in Spectroscopy

Fundamental parameters → model atmosphere → synthetic

spectra → physical output

For most spectroscopists model atmosphere is a black box.

Learn how to use existing codes by yourself

Use already calculated public model grids
If you are lazy, ask other people to calculate models for you
(not always work perfectly)

Models of Stellar Atmospheres

Self-consistency in Spectroscopy

Model atmosphere is consistent with observations if the following

observables are fitted simultaneously:
Energy distribution
Contains information about total energy balance
Hydrogen line profiles
Cool stars: most sensitive to temperature changes
Hot stars: most sensitive to pressure changes
Metallic line spectra
Abundances (stratification, spots), micro-turbulence, rotation,
magnetic fields, etc.

Models of Stellar Atmospheres

Self-consistency in Spectroscopy

Once abundances and other physical properties have been derived

it is necessary to recalculate the model atmosphere to consistency
consistency
Iterative solution is needed

Models of Stellar Atmospheres

Self-consistency in Spectroscopy

Models of Stellar Atmospheres

Self-consistency in Spectroscopy

Models of Stellar Atmospheres

Self-consistency in Spectroscopy

Models of Stellar Atmospheres

Self-consistency in Spectroscopy

Models of Stellar Atmospheres

non-LTE

In almost all plasmas (except very thin one with magnetic fields),
the Maxwell distribution is valid. The validity of Saha and
Boltzmann formulas on the other hand depend on the ratio of
photon number density εphotons to the particle number density
εparticles Z
4π 4 3
εphotons = Sν dν = σB T 4 , εparticles = nkT
c c 2
εphotons T3
f = = 36.5
εparticles n
f  1: LTE
f ∼ 1: LTE questionable
f  1: non-LTE

Models of Stellar Atmospheres

non-LTE

dni X X
= −n(i) [R(i→j) + C(i→j) ] + n(j)[R(j→i) + C(j→i) ] = 0
dt
j j

Upward rates, Zbound-bound and bound-free:

Jν
Rup (i → j) = 4παnu dν
hν
Downward rates, bound-bound, free-bound
8πν 2
Z  
nj −hν/kT 4π
Rdown (i → j) = αν e Jν + 2
ni hν c

ni /niLTE
biZwaan ≡ ni /niLTE , biMenzel ≡
nc /ncLTE

nc –concentration of next ionization stage

Models of Stellar Atmospheres

Validity of LTE

In stellar model atmospheres we are mainly interested in non-trace

elements like H, He, Fe, etc. (most abundant species having many
strong lines in the spectra).

Models of Stellar Atmospheres

Additional Physics

Additional physics have to be included to improve models and our

understanding about outer regions of the stars and to model their
spectra

Spherical symmetry: stars with extended atmospheres

Stellar wind: hot stars (multi-component stellar wind)
Diffusion of chemical elements: CP stars with quite
atmospheres (horizontal and vertical distribution)
Magnetic fields: CP2 stars with strong magnetic fields
Rotation: fast rotating hot stars, meridional circulations
Binarity effects: dust disks, x-ray emission, tidal effects, etc.
...

Models of Stellar Atmospheres

Model Grids

ATLAS9, ATLAS12 (R.L. Kurucz)

[Link]
TLUSTY (Lanz T., Hubeny I.)
[Link]
MARCS (Gustafsson B. et al.)
[Link]
NEMO grid based on ATLAS9 with different convection
treatment
[Link]

Models of Stellar Atmospheres

Bibliography

Bender, R. 2002, Astrophysics introductory course

Kurucz R.L. 1970, ATLAS: a computer program for
calculating model stellar atmospheres, SAO Special Report
Mihalas, D. 1978, Stellar atmospheres (2nd edition)
Rutten, R.J. 2003, Radiative transfer in stellar atmospheres

Models of Stellar Atmospheres