SIMPLE RADIATION TRANSFER FOR SPHERICAL STARS
Under LTE (Local Thermodynamic Equilibrium) condition radiation has a Planck (black body) distribution. Radiation energy density is given as
8πh ν3dν U dν = , (LTE), (tr.1) r,ν c3 ehν/kT 1 − and the intensity of radiation (measured in ergs per unit area per second per unit solid angle, i.e. per steradian) is c I = B (T )= U , (LTE). (tr.2) ν ν 4π r,ν
The integrals of Ur,ν and Bν (T ) over all frequencies are given as
∞ 4 Ur = Ur,ν dν = aT , (LTE), (tr.3a) Z 0 ∞ ac 4 σ 4 c 3c B (T )= Bν (T ) dν = T = T = Ur = Pr, (LTE), (tr.3b) Z 4π π 4π 4π 0 4 where Pr = aT /3 is the radiation pressure. Inside a star conditions are very close to LTE, but there must be some anisotropy of the radiation field if there is a net flow of radiation from the deep interior towards the surface. We shall consider intensity of radiation as a function of radiation frequency, position inside a star, and a direction in which the photons are moving. We shall consider a spherical star only, so the dependence on the position is just a dependence on the radius r, i.e. the distance from the center. The angular dependence is reduced to the dependence on the angle between the light ray and the outward radial direction, which we shall call the polar angle θ. The intensity becomes Iν (r, θ). The specific intensity is the fundamental quantity in classical radiative transfer. The radiation energy density is its zeroth angular moment, the flux is its first angular moment, and the radiation pressure is its second angular moment.
tr — 1 Let us consider a change in the intensity of radiation in the direction θ at the radial distance r when we move along the beam by a small distance dl = dr/ cos θ. The intensity will be reduced by the amount proportional to the opacity per unit volume, κν ρ multiplied by dl, where ρ is density of 2 −1 matter and κν is monochromatic total opacity in units cm g . κν is comprised of an absorptive a s (κν ) and a scattering (κν ) part. The former is connected to the emissivity by Kirchhoff’s law and the intensity will increase by the amount proportional to the emissivity of gas. Under nearly a LTE condition the emissivity per unit volume is given as a product κν ρBν (T ). The equation of monochromatic radiation transfer is written as
s ∂Iν (θ, r) ∂Iν (θ, r) a κ ′ ′ ′ = cos θ = κν ρIν (θ, r)+ κ ρBν (T )+ I (θ, r)Φ(Ω , Ω)dΩ , (tr.4) ∂l ∂r − ν 4π Z ν where the last term represents the redistribution of radiation by scattering back into the beam with angular indicatrix (coupling) Φ(Ω′, Ω). Note that the dependence of the various opacities on photon frequency, as well as on the density, temperature, and chemical composition of the gas, is implied. Monochromatic radiation energy density may be calculated as 1 Ur,ν = Iν (θ, r) dΩ, (tr.5) c Z 4π where the integration is extended over the whole 4π solid angle. This is the zeroth angular moment of the radiation field. Because of azimuthal symmetry this integral may be written as
π 2π Ur,ν = Iν (θ, r) sin θdθ. (tr.6) c Z 0 The total radiation energy density is given as
∞
Ur = Ur,ν dν. (tr.7) Z 0
tr — 2 Monochromatic flux of radiation in the direction r is the first angular moment of the specific intensity and may be calculated as
π
Fν = Iν (θ, r)cos θdω =2π Iν (θ, r)cos θ sin θdθ, (tr.8) Z Z 4π 0 and the total flux of radiation, measured in erg cm −2 s −1, is given as
∞
F = Fν dν. (tr.9) Z 0
The radiation pressure is the second angular moment of Iν , integrated over frequency. In the Eddington/diffusive approximation, the second moment is one-third of the zeroth moment. It can then be shown that the flux is driven by the spatial gradient of the Planck function:
4π ∂Bν (T ) 4π ∂Bν (T ) dT Fν = = . −3κνρ ∂r −3κνρ ∂T dr
The total radiative energy flux is an integral of Fν over all frequencies, i.e
∞ 4π dT 1 ∂B (T ) 4π dB (T ) c dP F = ν dν = = r , (tr.10) − 3ρ dr Z κν ∂T −3κRρ dr −κRρ dr 0 where the Rosseland mean opacity, κR, is defined as
∞ ∞ 1 ∂Bν (T ) dν 1 ∂Bν (T ) dν 1 κν ∂T κν ∂T R0 R0 ∞ = ac 3 . (tr.11) κR ≡ ∂Bν (T ) π T ∂T dν R0 Of course, the Rosseland mean opacity is a function of density, temperature and chemical composi- tion, κR(ρ,T,X). We shall at times write the Rosseland opacity as κ.
tr — 3 We now have
L c dU 4acT 3 dT dT r = F = r = = λ , (tr.12) 4πr2 −3κρ dr − 3κρ dr − dr
2 where Lr = 4πr F is stellar luminosity at a radius r, i.e. the total amount of radiation energy flowing across a spherical surface with a radius r, and Ur is the radiation energy density. The last equation looks just like the equation for heat diffusion, with the coefficient of thermal conductivity λ related to the coefficient of opacity with a relation
4acT 3 λ = . (tr.13) 3κρ As heat may be transferred not only by photons, but also by electrons, it may be safer to write the last equation as
4acT 3 λrad = , (tr.14) 3κradρ
where λrad and κrad are explicitly related to radiation. We may write a similar relation for electrons:
4acT 3 λel = , (tr.15) 3κelρ
where λel and κel are the coefficients of thermal conductivity and ”opacity” for the electrons. In general, we may have some heat transferred by photons, and some by electrons. As the two means of heat transport are additive, the combined coefficient of thermal conductivity may by calculated as
λ = λrad + λel, (tr.16) or equivalently, we may write a formula for the combined coefficient of opacity as 1 1 1 = + . (tr.17) κ κrad κel
tr — 4 Near the stellar surface the luminosity and radius can be taken as L and R, and the radiation energy flux is F = L/4πR2. Let us define optical depth τ as
dτ κρdr, τ =0 at r = R. (tr.18) ≡− Now, we may write the equation as
dT 4 3F = const, (tr.19) dτ ac ≈ and therefore 3F T 4 = T 4 + τ, (tr.20) 0 ac
where T0 is the temperature at the stellar surface. Consider now a surface radiating as a black body with a temperature T. At a point just above the surface the radiation comes from one hemisphere only, and we may use equations (tr.7), (tr.6), and (tr.2) to calculate
∞ π 2π Ur = Iν (θ) sin θdθ dν = (tr.21) Z c Z 0 0 ∞ π/2 2π 2π 1 4 Bν (T ) sin θdθ dν = B (T )= aT . Z c Z c 2 0 0 We obtained only one half of the radiation energy density expected under LTE conditions for the temperature T , because radiation was coming from one hemisphere only. The radiative energy flux may be calculated for our case:
∞ π ∞ π/2
F = 2π Iν (θ) sin θ cos θdθ dν = 2π Bν (T ) sin θ cos θdθ dν = (tr.22) Z Z Z Z 0 0 0 0 ac πB (T )= T 4 = σT 4. 4 We shall define the effective temperature of a star with a relation L = F σT 4 . (tr.23) 4πR2 ≡ eff This is a temperature that a black body would have if it radiated just as much energy per unit area as the star does.
tr — 5 The radiation energy density at the surface of a black body is half of the LTE energy density corresponding to the temperature T . We shall adopt an approximation that at the stellar surface, 4 i.e. at τ = 0 the radiation energy density is aTeff /2, by analogy with a black body case. We find that 1 T 4 = T 4 , (tr.24) 0 2 eff and the temperature distribution close to the stellar surface is given as
1 3F 1 3 T 4 = T 4 + τ = T 4 + τ . (tr.25) 2 eff ac eff 2 4
Therefore, we have T = Teff at τ = 2/3. The optical depth 2/3 corresponds to a photosphere. This formula is the origin of the classic τ =2/3 condition for photospheres. Note that in the diffusion/Eddington approximation the surface temperature is 21/4 times lower than the effective temperature. The exact value is (4/√3)1/4, when the opacity is independent of frequency. The radiation diffusion equation may be written as dP κρ κρ L r = F = r . (tr.26) dr − c − c 4πr2 This may be combined with the equation of hydrostatic equilibrium to obtain dP dP dP GM κρ L g = r = r ρ + r = (tr.27) dr dr − dr − r2 4πc r2
GMr κLr 2 ρ 1 . − r − 4πcGMr
The emergence of the local Eddington luminosity, 4πcGMr/κ is not a coincidence and is gratifying.
tr — 6 Near the stellar surface we have Lr = L, and Mr = M. We shall find later on that when luminosity is very high then density in a stellar atmosphere is very low, and the opacity is dominated by scattering of photons on free electrons. For a fully ionized gas the electron scattering opacity is given as n κ = e σ =0.2(1+ X) , [ cm 2 g −1], (tr.28) e ρ e where X is hydrogen content by mass fraction, ne is number of electrons per cubic centimeter, and σe is equal to the Thomson scattering cross-section for scattering photons on electrons
8π 8π e2 2 σ = r2 = =0.665 10−24 cm 2. (tr.29) e 3 e 3 mc2 × Putting the electron scattering opacity into the equation we obtain near the stellar surface
dP GM κ L g = ρ 1 e , (tr.30) dr − r2 − 4πcGM while the gradient of the total pressure P is given as dP GM = ρ. (tr.31) dr − r2 Dividing the last two equations side by side we obtain dP κ L g =1 e = const. (tr.32) dP − 4πcGM This may be integrated to obtain
κ L P = (P P ) 1 e , (tr.33) g − 0 − 4πcGM where P0 = Pr,(τ=0) =2F/3c is a very small radiation pressure at the stellar surface. It is clear, that at a modest depth below the stellar surface the pressure is very much larger than at the surface, and therefore, the equation gives
P κ L β g = 1 e . (tr.34) ≡ P − 4πcGM
It is obvious that 0 <β< 1, and, therefore,
38 −1 4πcGM 4πcG 2.50 10 erg s M 65300L⊙ M 0
tr — 7