<<

PHYS 633: Introduction to Stellar Astrophysics Spring Semester 2006 Rich Townsend ([email protected])

Opacity The equation of radiative diffusion, which is one of the governing equations of stellar structure, relates the radiative flux of energy,FR, to the local temperature gradient, 4ac dT F = . (1) R 3κρ dr Playing a very important role in this equation is the Rosseland mean opacity κ (also known as the mass absorption coefficient; and note that some authors use the term ‘opacity’ actually to refer to the product κρ). The opacity charac- terizes the degree to which the passage of is retarded by the presence of matter, due to and absorption of photons. For a sample of stel- lar material with mass dm, the total cross section presented by the particles constituting the sample is given by κdm. If the volume of the sample is dV , then we have κdm = κρdV , and we can see that the term κρ can be recognized as the cross section per unit volume. Since κρ has dimensions of inverse length, we can see that we can multiply it by a distance dz and obtain a dimensionless number. What does this number represent? Well, if dz is infinitesimal, then κρdz is the probability that a photon will be absorbed when traveling through dz, and 1−κρdz is the complementary probability that the photon won’t be absorbed. To work out the probability of an ‘average’ photon (in the Rosseland mean sense) traveling a macroscopic distance, let P (z) be the probability that the photon traverses distance z without being absorbed. Then, the probability that the photon travels a distance z + dz unabsorbed is given by

P (z + dz) = (1 − κρdz)P (z), (2) which is the product of the individual probabilities for each leg of the journey. Thus, we have P (z + dz) − P (z) = −κρP (z), (3) dz and in the limit of vanishing dz, dP = −κρP (z). (4) dz Integrating, with the boundary condition of P (0) = 1, we therefore obtain

P (z) = e−τ , (5)

1 where we have introduced the (dimensionless) optical depth τ as Z z τ(z) = κρ dz. (6) 0 This latter equation can be used to define the mean free path ` of an average photon, such that τ(`) = 1. So far, we have been dealing with the Rosseland mean opacity κ. We recall from out treatment of the diffusion equation that for material in thermodynamic equilibrium with temperature T (by which we mean the radiation field has a black-body spectrum with temperature T ), the Rosseland mean is defined by

R ∞ 1 ∂Bν 1 0 κ ∂T dν ≡ ∞ , (7) κ R ∂Bν 0 ∂T dν where 2π ν3 B = . (8) ν c2 ehν/kT − 1 is the Planck function. Clearly, we need to know the frequency-dependent opac- ity κν to calculate κ. Unfortunately, there is no simple recipe for this quantity; calculation of κν — and thus κ — is one of the most complex tasks of stellar structure and evolution, and in reality any physically-realistic opacity model is far too complex to incorporate directly in a stellar evolution code. In fact, the standard approach nowadays is to store the results from de- tailed opacity calculations in table form, so that the evolution code can simply look up/interpolate appropriate values. Typically, the Rosseland mean opac- ity κ is tabulated as a function of temperature T and density ρ, or sometimes 3 6 temperature and R ≡ ρ/T6 (here, T6 is the temperature in units of 10 K). The present-day state-of-the-art tabulations are by two main groups: OPAL, based at Lawrence Livermore National Laboratory, and the international Opac- ity Project (OP). Of course, even though these tabulations are available for download by any- one who wants to use them, we still should have an idea about where the values come from. So, let’s briefly discuss the physical processes contributing toward stellar opacity.

Electron Scattering When an electromagnetic wave is impingant on a free electron, it causes the electron to oscillate. The electron itself then emits a new electromagnetic wave, with the net effect being that the original wave has been scattered through some 2 angle. If the original wave has a frequency ν such that hν  mec (which occurs typically for blackbody radiation at temperatures cooler than ∼ 108 K), then the scattering is coherent: the outgoing wave (photon) has the same frequency (energy) as the incoming wave (photon). Such coherent electron scattering is known as Thomson scattering, and has a cross-section per electron of 8π σ = a2 ≈ 6.652 × 10−25 cm2 (9) T 3 0

2 where e2 a0 = 2 (10) mec is the classical radius of the electron (in cgs units; SI units require an extra factor of 4π0 in the denominator). To calculate the opacity due to Thomson scattering, we multiply σT by the number of free electrons per unit mass, n κ = σ e . (11) ν T ρ Since the mean molecular weight per electron is defined by ρ µemu = , (12) ne we also have σT κν = . (13) µemu Note that, although we indicate that the opacity κν can in principle depend on the frequency, in the case of Thomson scattering it is frequency independent. In the limit where the stellar material is completely ionized, µe is given by 2 µ = (14) e 1 + X (see eqn. 20 of the notes on the Equation of State), where X is the hydrogen mass fraction. Hence, the Thomson scattering opacity becomes

σT 2 −1 κν = (1 + X) ≈ 0.20(1 + X) cm g (15) 2mu For typical hydrogen abundances X = 0.7, we therefore obtain a value 0.34 cm2 g−1, independent of the temperature, density or frequency. 2 If the scattered radiation has a frequency such that hν > mec (typically occurring for blackbody radiation at temperatures hotter than ∼ 108 K), then the electron recoils during the scattering by a non-negligible amount. To con- serve energy, the recoil means that the scattered radiation (photon) must have a frequency (energy) less than that of the incoming radiation (photon); the scat- tering is no longer coherent. In this regime, we speak of Compton scattering. The cross section for Compton scattering is obviously frequency dependent, and is always smaller than the Thomson cross section.

Free-Free Absorption When a free electron passes an ion, they temporarily form an (unbound) sys- tem capable of absorbing electromagnetic radiation of any frequency. The opac- ity associated with such free-free absorption1 was originally characterized by Kramers, and has the form −3 1/2 κν ∝ ν neT . (16) 1The reverse process of free-free emission occurs when a charged particle decelerates in the electrostatic field of another particle, and is known as Bremsstrahlung (radiation braking).

3 −3 The ν term comes from classical electrodynamics. The ne term appears be- cause, for a given amount of mass, the number of possible free-free absorption systems depends on the number of available (free) electrons. Finally, the T 1/2 term appears because the time that electrons and ions spend together scales inversely with their relative velocity due to thermal motions, which itself scales as ∼ T 1/2. The constant of proportionality in Kramers’ expression above depends on a number of things, including the charge on the ion, and a factor g, known as the Gaunt factor, that corrects for quantum mechanical effects (remember, Kramers’ expression is derived from classical electrodynamics). One of the most important corrections is for degeneracy; if the density is very large or the temperature very low, then there is a scarcity of quantum states for free electrons to move into, and the free-free opacity is found to be smaller than in the corresponding non- degenerate limit. If free-free opacity is the only (or dominant) opacity source in the , then we can go ahead and calculate the Rosseland mean opacity as

R ∞ 3 ∂Bν 1 1/2 −1 0 ν ∂T dν ∝ T ne ∞ (17) κ R ∂Bν 0 ∂T dν. With a little bit of algebra, we find that

−7/2 κ ∝ neT , (18) a characteristic temperature and density dependence often termed ‘Kramers opacity’ (irrespective of whether the opacity source is due to free-free absorption or not). Note that there may be an additional temperature dependence hidden in this equation, because the electron number density ne itself depends on the (temperature-sensitive) ionization state of the stellar material. Clearly, free-free absorption is most effective at high densities and low temperatures — typically, the sorts of conditions found in lower-mass .

Bound-Free Absorption Bound-free absorption is photoionization from the point of view of the radiation field: a photon is removed from circulation, being used to eject an electron from a bound state of an atom or ion. Bound-free absorption only works for photons with frequencies satisfying hν > χ, where χ is the ionization potential – the amount of energy required to remove the electron from the bound state and place it at rest at infinity. The excess energy of the photon over χ goes into the electron kinetic energy. For sufficiently energetic photons, the Rosseland mean opacity for bound- free absorption exhibits the same basic temperature and density dependence as free-free absorption (see eqn. 18), and thus is also often referred to as ‘Kramers opacity’. However, as with the free-free case, there are additional tempera- ture dependencies hidden away, relating in this case to the number of bound (un-ionized) atomic systems. If the temperature is too high (typically, above

4 106 K), then all elements will be completely ionized, and there is no bound-free absorption.

Bound-Bound Absorption If a photon has insufficient energy to ionize an atom (i.e., if it is too low-energy for a bound-free absorption), it can still be absorbed by the atom — if its energy hν matches the difference hν0 between two of the atom’s energy levels. During such a bound-bound absorption, one of the atom’s electrons makes a transition from the lower energy level to the upper energy level, with the energy for the transition being provided by the absorbed electron. Bound-bound absorption only works within a narrow range ∆ν of frequen- cies centered on the frequency ν0 of the transition, and therefore is often termed ‘(spectral) line opacity’. The line width ∆ν is set by the type of transition involved, but also depends on the density and temperature of the stellar mate- rial, which contribute respectively to pressure (natural) and thermal (Doppler) broadening. The width-to-frequency ratio ∆ν/ν0 is never much above a percent or so, and so it would seem at first glance that bound-bound opacity can never make a significant contribution to the Rosseland mean opacity κ. However, it should be remembered that although spectral lines are narrow, they contain a huge amount of opacity, because the bound electrons in an atom behave like resonant systems that are incredibly efficient at absorbing radia- tion. Furthermore, there are situations where atoms can exhibit huge numbers of closely-spaced lines that overlap and effectively ‘blanket’ whole regions of the electromagnetic spectrum. Perhaps the most famous example of this line blanketing is the so-called ‘iron bump’, discovered in the early 1990s with the ad- vent of the new OPAL and OP opacities. Around a temperature T ∼ 200, 000 K, transitions between the M-shell (n = 3) energy levels of ionized iron and nickel, which are split into many sublevels by the spin-orbit effect, lead to millions of overlapping spectral lines. The combined effect of these lines is to introduce a pronounced ‘bump’ (peak) in the opacity. Shortly after its discovery, the iron bump was found to be the origin of the oscillations of the massive β Cephei stars and slowly-pulsating B-type stars. These objects oscillate due to a ‘κ mechanism’: a small increase in the temper- ature causes a corresponding increase in the opacity. This retards the escape of energy from the stellar core (since the radiative flux goes down; see eqn. 1), leading to a further increase in temperature until the star expands and lets the radiation escape. The whole cycle then repeats, leading to the establishment of a periodic oscillation that can be observed through brightness and spectroscopic variations.

The Negative Hydrogen Ion One special case of bound-free absorption is that associated with the single energy level of negative hydrogen ions. If an electron passes close to a (neutral) hydrogen atom, it can be captured to form a negatively-charged H− ion. The

5 binding energy of the ion is quite small (0.75 eV, so H− bound-free opacity is able to absorb photons in the near infrared as well as in the optical and UV. To create H− ions, there must be a sufficient supply of free electrons. In pure hydrogen, this rather ironically requires that some of the hydrogen atoms be ion- ized, so we have a mixture of H− ions, neutral hydrogen, and H+ ions. Typically, however, the free electrons required to create the H− ions come from the small quantity (Z ∼ 0.02) of metals found in solar-abundance stellar material. These metals can be ionized even at low temperatures (T ∼ 3, 000 − 5, 0000 K, pro- viding the necessary source of electrons. At significantly higher temperatures, there are free electrons in abundance, but it becomes difficult for the H− ions to stay together (think of the Saha equation). Hence, bound-free opacity due to H− ions is usually only important in cooler stars.

Molecular Opacity An important opacity source in very cool stars comes from the formation of molecules — hydrogen (H2), water (H2O), titanium oxide (TiO), monox- ide (CO), etc. In addition to electron energy levels, these molecules exhibit a range of energy levels associated with rotation and vibration. These levels are typically so close together that they merge to form continuous energy bands, and transitions between differing bands lead to bound-bound opacity over wide ranges of frequency (this is similar to the line blanketing effect discussed above). This opacity is typically situated in the infrared parts of the spectrum.

High-Energy Processes In addition to the Compton scattering discussed above, there are a variety of processes that are able to absorb high-energy photons. The most straightforward is pair production: an energetic photon, in the electromagnetic field of a nucleus, decays into an electron and a positron. The threshold energy for the photon is given by the combined rest-mass energy of the electron and positron. This comes out as 1.022 MeV, so the photons undergoing pair production are gamma rays. In principle, other types of pair production can arise, so long as they obey the relevant conservation laws. Neutrino pair production can be quite an im- portant opacity source in supernova explosions. Also important in supernovae is photodissociation, where photons are absorbed by nuclei, causing them to break apart.

6