Interpreting Light from the Atmosphere of the Sun A primer on how spectra make pretty pictures, and how we can use them to get physical quantities Interpreting Light from the Atmosphere of the Sun “What do you think you’re looking at??” A primer on how spectra make pretty pictures, and how we can use them to get physical quantities Light Light • Almost everything we know about the Sun comes from light Light • Almost everything we know about the Sun comes from light • So why do we spend so much time getting it wrong Light • Almost everything we know about the Sun comes from light • So why do we spend so much time getting it wrong • I’m projecting: why have *I* spent so much time getting it wrong? Solar Spectrum Solar Dynamics Observatory: images in the UV to EUV SolarDynamicsObservatory: imagesintheUVtoEUV Where did the light come from? • What we’re looking at (measuring) is where the photon came from • This might seem philosophical, but it’s a crucial question • The photon was created or scattered by the conditions at that point. • So what are the chances that it can make it to our eye or telescope? Journey of a photon through gas Photons can be absorbed by ions with bound electrons Journey of a photon through gas incident beam Photons can be absorbed by ions with bound electrons emergent beam And they can be scattered (mostly by free charges like electrons) Journey of a photon through gas incident beam Photons can be absorbed by ions with bound electrons emergent beam And they can be scattered (mostly by free charges like electrons) So how far can they get, on average, before one of these processes removes them from the beam, Iν ? Mean-free-path of a photon So how far can they get, on average, before one of these processes removes them from the beam, Iν ? • This isn’t such a hard question to make a stab at, once we know two basic quantities • • What’s the number density n of the particles? • We have to treat all the different absorbers and scatterers, separately • e.g., e–, H I, He II, Si VII, Fe XIII, … What’s the probability that a photon will interact with any individual particle? • This will depend on wavelength or frequency, ν Probability of removing photons from the beam • The probability that a single particle will absorb a photon can be expressed in terms of how large that particle “looks” to the photon. • This is given as a “cross-section” σν, given in units of area • The more particles there are, the greater the total “cross sectional area” there is to intercept it. side view view along beam each particle has an apparent cross-sectional area for interaction σν at frequency ν • The probability for a photon to be removed from the beam is going to be proportional to • The probability of interaction (apparent area) for each type of particle, σν • the number density of each type of particle, np • • more particles will occupy more area perpendicular to the beam the distance the beam travels along the column, s • The probability for a photon to be removed from the beam is going to be proportional to • The probability of interaction (apparent area) for each type of particle, σν • the number density of each type of particle, np • • more particles will occupy more area perpendicular to the beam the distance the beam travels along the column, s the view along the beam is going to start looking very crowded! ! Changing n or σν column density (n ✕ s) cm-2ij (⌫ ⌫ // Same, but with more absorbers 1 2 3 ⌫0/ ) ⌫ / ij (⌫ -⌫ 1 if σ depends on ν (or λ) then the effective size of the absorbers can change as we tune ν: 2 if σ depends on ν (or λ) then the effective size of the absorbers can change as we tune ν. If the scatterers 0are free electrons, for example, then this crosssection won’t change 3 ij (⌫ - ⌫ ) We’re only changing the frequency/wavelength of the light between these three situations: n remains the same, only σν decreases for the absorbers - ⌫0 ) projecting whole column onto its face area For a given particle p (e.g., Fe IX ion, e–) and a given frequency ν, this will produce a probability Premν,p of a photon being absorbed or scattered: [cm2 cm-3 rem P⌫,p (s) = cm] = dimensionless ⌫ np (s)s The fraction of the beam δIν/Iν removed by particles of type p will be equal to this probability: I⌫ = -I⌫ ⇥ ⌫ np (s)s where δIν has a negative sign because we reduce the intensity of the beam. • In reality, though, np is not necessarily constant along our line of sight, s we need to integrate its effects along the beam direction • dI⌫ =I⌫ • ⌫ np (s)ds = -⌫ ds The quantity κν is called the linear extinction coefficient, since it represents the removal (“extinction”) probability per unit distance along the beam • κν is sometimes called the “opacity” of the plasma, but watch out for other defitions: e.g., per unit mass Mean-free-path at last…! • The inverse of κ… number of photons removed per unit length • • – …is the mean free path, sν length between photon removals! • • You can also think of it like this: • • distance travelled s⌫ = = number of particles encountered s = ⌫s · n 1 ⌫n where σνs sweeps out a volume, which you multiply by the number density n to get a number of particles encountered. κν & τν dI⌫ = -⌫ ds I⌫ Z ) Z dI⌫ = - ⌫ ds I⌫ The integral on the right hand is one of the most fundamental concepts in all astronomy: the optical Z s0 depth, τν • ⌫ ds 0 ⌧⌫ (s ) = - 0 What’s the τ? • • So what’s the physical meaning of τ? Let’s use a meaningful length scale for s: • • • in this case, we’ll use the mean-free-path, s–ν We can see that τ=1 corresponds to one photon mean-free-path into the plasma! Meaning that on average, photons tend to come from τ=1 -⌧⌫ (s) = = Zs Zs ⌫ ds 0 ⌫ n ds 0 = s s ⌫n 0 = s⌫ ⌫ n 1 = ⌫n ⌫n -⌧⌫ (s⌫ ) = 1 Optically thin or thick? • • Typically, if a block of plasma has τν >> 1 • it would take a photon several mean free paths to make it to the front • the back of that plasma can therefore rarely be seen • This case is called optically thick and if if τν << 1 • then the almost all photons initially headed towards the observer just carry on along the same direction • The mean free path is much longer than the physical depth of the plasma • This case is called optically thin Effect of τν on a beam of light The form of our simple differential equation means that a beam of intensity Iν is attenuated by a factor that is exponential with distance s along the beam: • Z h ln • This does assume only extinction and not emission within the plasma column, of course. ln I⌫ ✓ iI(s 0 ) = -⌧⌫ (s 0 ) ◆ = -⌧⌫ (s 0 ) I(0) 0 I⌫ (s ) I⌫ (0) 0 dI⌫ = -⌧⌫ I⌫ -⌧⌫ (s 0 ) I⌫ (s ) = I⌫ (0)e Dependence of τν on ν ⌧⌫ (s 0 ) = Because Z0 n(s) s0 ⌫ ds the optical depth, and therefore the mean free path, can depend strongly on the wavelength we look at. If we consider bound-bound absorption ij (⌫ - ⌫0 ) where ψ is the profile of the probability for absorbing the photon around ν = ν0, where hν0 = Ej – Ei . ij (⌫ / / ⌫ ⌫ ⌫ then - ⌫0 ) photon is absorbed and excites an electron from state i to state j / ij (⌫ -⌫⌫0/ ) ij (⌫ -⌫ 0 ⌧⌫ (s ) = ⌫ Z0 • • Column density N(s) is the number of absorbing particles we can see along a column of unit crosssectional area A and length s s0 = • n(s) ds 0 N(s ) ⌫ The integral here N(s’) is the column density (number of absorbers per unit area along the line of sight) between s=0 and s=s’ σν As a consequence, an τν = 1 requires a far smaller number of particles along the line of sight at ν = ν0 than it does at frequencies (wavelengths) away from the centre of the absorption line. We therefore tend to see photons from higher in the atmosphere at the centre of the line because the mean free path is a shorter distance at the central frequency/wavelength • we see deeper into the atmosphere for wavelengths/frequencies away from line centre. Α = 1 cm2 ⌫ / ⌫ / ij (⌫ ij (⌫ - ⌫0 ) -⌫⌫0/ ) ij (⌫ - Varying visible depth with λ or ν Varying visible depth with λ or ν Varying visible depth with λ or ν Varying visible depth with λ or ν https://svs.gsfc.nasa.gov/cgi-bin/details.cgi?aid=11708 Varying visible depth with λ or ν Credit: NASA/IRIS/T. Pereira At the centre of the line, we see the plasma closest to us (highest) ν0 https://svs.gsfc.nasa.gov/cgi-bin/details.cgi?aid=11708 Varying visible depth with λ or ν Credit: NASA/IRIS/T. Pereira At the centre of the line, we see the plasma closest to us (highest) ν0 https://svs.gsfc.nasa.gov/cgi-bin/details.cgi?aid=11708 Varying visible depth with λ or ν Credit: NASA/IRIS/T. Pereira At the centre of the line, we see the plasma closest to us (highest) ν0 Different equilibria in the Sun Thermodynamic equilibrium is a complete interaction of material and radiation – a black body. The radiation intensity is given by the Planck function, Bν(T). It holds in the solar interior. Different equilibria in the Sun Thermodynamic equilibrium is a complete interaction of material and radiation – a black body. The radiation intensity is given by the Planck function, Bν(T). It holds in the solar interior. Local thermodynamic equilibrium (LTE) is where matter and radiation almost completely interact, but with a small escape of radiation. The equilibrium is defined in terms of the local temperature. It holds in the solar photosphere. Iν = Bν(T) is still a good approximation Photon mean free path is very short in both cases with respect to any scales of change in temperature, pressure, etc. So the radiation field is a black body one, characterised by the same temperature as the particles’ Maxwellian distribution Different equilibria in the Sun Non-local thermodynamic equilibrium (NLTE) is where there is incomplete interaction of matter with radiation, with radiation freely escaping from the region. It holds in the solar chromosphere. Mean free path starts to be larger than size of the system. Radiation can therefore escape the chromosphere and is not returned, so Iν starts to depart from Bν(T) Different equilibria in the Sun Non-local thermodynamic equilibrium (NLTE) is where there is incomplete interaction of matter with radiation, with radiation freely escaping from the region. It holds in the solar chromosphere. Mean free path starts to be larger than size of the system. Radiation can therefore escape the chromosphere and is not returned, so Iν starts to depart from Bν(T) Coronal equilibrium is an equilibrium between the numbers of ionisation and recombination processes per unit volume. The radiation from the photosphere passes through the material without any appreciable effect. It holds in the solar corona. Mean free path is now typically several solar radii. As a result, there is basically no connection between the radiation field and the populations of ions or excited states within those ions. Population balance almost completely determined by collisions with electrons. Radiative transfer equation This is a way of expressing how a beam’s intensity is modified by both subtraction of photons from (absorption + scattering) = κν and addition by emission = jν into a line-of-sight along s. Radiative transfer equation This is a way of expressing how a beam’s intensity is modified by both subtraction of photons from (absorption + scattering) = κν and addition by emission = jν into a line-of-sight along s. ⌫ = X np ( (⌫)bb p + p ff Th (⌫)bf + (⌫) + (⌫) p p p + . . .) Radiative transfer equation This is a way of expressing how a beam’s intensity is modified by both subtraction of photons from (absorption + scattering) = κν ⌫ = and addition by emission = jν into a line-of-sight along s. X np ( (⌫)bb p + p ff Th (⌫)bf + (⌫) + (⌫) p p p + . . .) here the superscripts indicate: • bb – bound-bound absorption of a photon • bf – bound-free absorption of a photon • ff – free-free absorption • Th – Thomson scattering of a photon from a free electron Specific Intensity • We can define the intensity of the beam as a specific intensity E⌫ = I⌫ d! d⌫ dA dt • • …which takes account of the energy emitted: • at frequency ν • in frequency range dν • through area dA • in time dt • into a solid angle dω Typically this takes units of erg cm-2 s-1 sr-1 (Å-1 | Hz-1) Changes to Iν along the beam direction, s dE⌫ + dE⌫ = -⌫ E⌫ ds = ⌫ I⌫ ds d! d⌫ dA dt = j⌫ ds d! d⌫ dA dt + dE⌫ = dE+ dE ⌫ ⌫ Changes to Iν along the beam direction, s dE⌫ + dE⌫ = -⌫ E⌫ ds = ⌫ I⌫ ds d! d⌫ dA dt = j⌫ ds d! d⌫ dA dt + dE⌫ = dE+ dE ⌫ ⌫ dI⌫ d! d⌫ dA dt = -⌫ I⌫ ds d! d⌫ dA dt + j⌫ ds d! d⌫ dA dt 1 ds : ⇠ ( ( ⇠ ( ( ( ( ⇠ ( ( ⇠dA dt = -⌫ I⌫ ( (d⌫ dA dt + j⌫ ( (d⌫ dA dt dI⌫ ⇠ d! ds( d! ds( d! ⇠d⌫ Changes to Iν along the beam direction, s dE⌫ + dE⌫ = -⌫ E⌫ ds = ⌫ I⌫ ds d! d⌫ dA dt = j⌫ ds d! d⌫ dA dt + dE⌫ = dE+ dE ⌫ ⌫ dI⌫ d! d⌫ dA dt = -⌫ I⌫ ds d! d⌫ dA dt + j⌫ ds d! d⌫ dA dt 1 ds : ⇠ ( ( ⇠ ( ( ( ( ⇠ ( ( ⇠dA dt = -⌫ I⌫ ( (d⌫ dA dt + j⌫ ( (d⌫ dA dt dI⌫ ⇠ d! ds( d! ds( d! ⇠d⌫ dI⌫ = -⌫ I⌫ + j⌫ ds Changes to Iν along the beam direction, s dE⌫ + dE⌫ = -⌫ E⌫ ds = ⌫ I⌫ ds d! d⌫ dA dt = j⌫ ds d! d⌫ dA dt + dE⌫ = dE+ dE ⌫ ⌫ dI⌫ d! d⌫ dA dt = -⌫ I⌫ ds d! d⌫ dA dt + j⌫ ds d! d⌫ dA dt 1 ds : ⇠ ( ( ⇠ ( ( ( ( ⇠ ( ( ⇠dA dt = -⌫ I⌫ ( (d⌫ dA dt + j⌫ ( (d⌫ dA dt dI⌫ ⇠ d! ds( d! ds( d! ⇠d⌫ dI⌫ = -⌫ I⌫ + j⌫ ds This is the equation of radiative transfer Radiative transfer equation in s dI⌫ = -⌫ I⌫ + j⌫ ds j⌫ = S⌫ ⌫ • The ratio of emission to absorption is called the source function, Sν • In the s co-ordinate, along the direction of the beam, we would write dI⌫ = ⌫ (-I⌫ + S⌫ ) ds • This says that intensity will decrease along s if Sν < Iν, and vice versa. (Note that Iν has a negative sign in this arrangement.) Radiative transfer equation in τ dI⌫ = -⌫ I⌫ + j⌫ ds ÷ d⌧⌫ = -⌫ ds dI⌫ j⌫ = I⌫ d⌧⌫ ⌫ • So, if increase in intensity is positive as we go deeper into the Sun… • the source function is not as large as the incident radiation field from beneath it. What about the direction of s? • If we have a particular reference direction, r • • e.g. a direction specified because we have gravity then we have to think about Iν as a function of r and θ • θ = cos-1μ is the angle between our beam direction s, and r. E⌫ = I⌫ d! d⌫ dA dt What about the direction of s? We can think of the optical depth with respect to an imposed direction, r • • dI⌫ µ = I ⌫ - S⌫ d⌧⌫ for which a general solution is Z1 - ⌧µ⌫ ⌧⌫ S⌫ (⌧⌫ )e d I⌫ (µ, 0) = µ 0 • (not for today) E⌫ = I⌫ d! d⌫ dA dt Radiative transfer equation in τ • But, what falls out of that general solution is more specific solutions: I⌫ (µ = 1, 0) = S⌫ Z ⌧2 0 • e.g., if Sν is constant then it comes outside the integral • If the plasma is optically thick, then the back of the plasma is at τ2 >> 1 • so radiation emerging from the surface at the normal is Iν(1,0) = Sν - ⌧⌫ 1 e ✓d µ ⌧⌫ 1 µ 7 ◆ ◆ ⌧2 = S⌫ -e-⌧⌫ 0 = S⌫ -e-⌧2 - (-1) I⌫ (µ = 1, 0) = S⌫ 1 - e-⌧2 Approximation can be used for the photosphere where Iν = Sν = Bν(T) i.e., giving temperature Radiative transfer equation in τ • But, what falls out of that general solution is more specific solutions: I⌫ (µ = 1, 0) = S⌫ Z ⌧2 0 • e.g., if Sν is constant and plasma is optically thin, then the back of the plasma is at τ2 << 1 series representation of exponential function for small τ: 1-e -⌧ =1- z }| { 1 X (-⌧)k 0 k! ⌧2 ⌧3 =1-1+⌧+ + ... 2 6 I⌫ (µ = 1, 0) = S⌫ ⌧2 - ⌧⌫ 1 e ✓d µ ⌧⌫ 1 µ 7 ◆ ◆ ⌧2 = S⌫ -e-⌧⌫ 0 = S⌫ -e-⌧2 - (-1) I⌫ (µ = 1, 0) = S⌫ 1 - e-⌧2 This approximation is useful in the corona and transition region where total emission can be used to quantify emitting material Optically thin line emission • So what does it mean to see an optically thin emission line? • Why does it form? • What does its brightness tell us about the plasma there? • Conditions: ne, T • Dynamics: vflow Example spectrum of the Sun in the extreme UV (100 < λ < 912 Å), showing spectral lines from different Fe ions Optically thin line emission • • What does its brightness tell us about the plasma there? • Conditions: ne, T • Dynamics: vflow How do we turn the pretty pictures into quantitative information about the plasma? Sample image of the Sun, made from adding up all the light in an emission line of Fe X How does an emission line happen? • An ion doesn’t have all its electrons in the lowest-energy (ground) configuration • could be there for a number of reasons: collisional or radiative excitation; dropped down from a higher level or recombined from free state… • One of those excited electrons can spontaneously drop to the lowest available energy state • in doing so, it gives up a photon • spontaneous emission • also called bound-bound emission because it transitions between two bound orbits within the atom/ion A radiating volume of plasma • • How to interpret the brightness of a spectral line from transition j→i ? Energy emitted from unit volume containing nj ions in upper state j is the radiance εji ions in state j electrons ions not in state j A radiating volume of plasma • If measure the brightness of a volume of plasma, ΔV • Power radiated in that spectral line is integral of radiance Pji = • Z h⌫ji Aji nj dV V Problem is… we can’t directly measure the population density, nj ! ions in state j electrons ions not in state j • However, we can get nj via a bunch of other ratios that we can calculate / estimate Z = h⌫ji Aji Pji = h⌫ji Aji nj dV V Z nj n+q nX nH n e +q nX nH ne V n fraction of ions of charge +q that have electrons in excited state j fraction of element X with ion charge +q number density of electrons (multiple ways to estimate this) 0.83 relative abundance of an element X compared to hydrogen Coronal Temperatures Fractions of Fe ions as a function of T (plotted logarithmically) n+q nX Ion notation in these graphs: Mazzottaetal.(1998) 9 = Fe+9 = Fe X 19 = Fe+19 = Fe XX, etc. Note: for corona, Fe+8 to Fe+14 are the ionisation stages we usually come across A more exhaustive list of ionisation and Recombination and Excitation and De-excitation processes Aschwanden, ‘Physics of the Solar Corona’ A more exhaustive list of ionisation and Recombination and Excitation and De-excitation processes Aschwanden, ‘Physics of the Solar Corona’ Coronal equilibrium: de/excitation Aschwanden, ‘Physics of the Solar Corona’ • We assume that only collisions with electrons are important for excitations • …and that almost all de-excitation happens spontaneously, i.e., by radiation (and not by collisions) • -1 typically trad = Aji << time between electron collisions Simple level diagram for 2-level atom Level j Collisional excitation Radiative de-excitation Rate = ni ne Cij cm-3 s-1 Rate = nj Aji cm-3 s-1 Level i ne ni Cij = Aji nj * 2-level atom: assumptions • no radiative excitations Ni Ne Cij = Nj Aji j i • radiation field is mostly photons of too low an energy (only several eV) compared to energy gap between i and j • no population of j from higher levels • there is no higher level • no stimulated emission (lasing) • no recombination of electrons from outside into j Coronal approximation 1. Equilibrium: number of collisional excitations and radiative de-excitations per unit time and volume is the same • volume doesn’t get brighter or dimmer over time • no other processes matter 2. Optically thin emission j 3. Two-level atom i • see previous slide for assumptions 4. Maxwellian distribution of particle energies -1 5. If trad = Aji << tcoll then almost all ions have the lowest +q energy configuration – i.e., nj / n ~ 1 Coronal equilibrium: excitation of ion with 2 levels Collisional excitation (rate coefficient Cij ): X i+q + e– → X j+q + e– is balanced by radiative de-excitation (rate Aji): X j+q → X i+q + hν ne ni Cij = Aji nj where Aji is the radiative transition probability (in s-1) from j to i ne, ni, nj = number densities of electrons and of ions in states i and j (cm-3). (Unlike LTE, there is in general no collisional de-excitation Cji.) So photon emission rate = collisional excitation rate in a simple 2-level ion. This is very useful… Cij nj ne = <1 Aji ni in coronal equilibrium (usually much less) “Simplifying”… Z nj n+q nX nH Pji = h⌫ji Aji +q ne dV n nX nH ne V Z 1 n+q nX nH = h⌫ji Aji nj +q ne dV n nX nH ne V AX 0.83 1 Z +q nX◆ ni✓n 7 nH◆ 7 = h⌫ji ◆ ◆ Cij ne +q ne dV nH ◆ne n nX V ◆ 1 Fji = h⌫ji 0.83 AX 2 4⇡d Z +q n 2 Cij ne dV nX V Dependence of the line brightness on T collision rate coefficient Cij(T) Cij = r 8⇡ me kTe a20 ✓ ⌦ij Eij exp wi kTe ◆ Saha equation is also f(T) q n+(q+1) Z+(q+1) 2 3/2 kT = (2⇡me kT ) e +q +q 3 n Z ne h 1 Fji = h⌫ji 0.83 AX 2 4⇡d • Z +q n 2 Cij ne dV nX V G(T ) The temperature-dependent terms can be combined into a contribution function, G(T) • strictly, G(T, ne), but ne can usually be independently estimated Contribution functions G(T, ne) Contribution functions for two of those coronal-temperature EUV resonance lines Temperature(K) Some coronal-temperature EUV lines • Each line lights up under different temperature conditions • there is therefore a mix of plasma at different temperatures in this solar active region Courtesy: Hinode/EUV Imaging Spectrometer team Contributionfunctionsforsomehigh-temperaturecoronal X-rayresonancelines Temperature(106K) 58 Emission Measure (EM) Z 1 n+q 2 Fji = h⌫ 0.83 A C n dV ji X ij e 4⇡d2 nX V Z 2 EM = ne dV V 1 Fji = h⌫ 0.83 A ji X 2 4⇡d Z V Gji (T ) n2e dV The brightness of a line is the integral over the emitting volume of the contribution function multiplied by the emission measure Emission Measure (EM) Z 1 Fji = h⌫ 0.83 A ji X 2 4⇡d Z 1 = h⌫ 0.83 A a ji X pix 2 4⇡d V V 2 Gji (T ) ne dV 2 Gji (T ) ne dh If we know how large a pixel area apix is, we can separate the volume into an area and a distance along a column dh The second term in theZ integrand is now the column emission measure 2 V ne dh Differential emission measure φ(T) • We know that in the Sun, not all the plasma is going to be at the same temperature: • and we have optically thin emission so our line-of-sight will pass through these multiple temperatures Z 1 2 Fji = h⌫ji 0.83 AX apix Gji (T ) ne dh 2 4⇡d V Z 2 dh ... Gji (T ) ne dT dT V Z φ(T) is known as the differential emission ... Gji (T ) (T )dT V measure Differential emission measure φ(T) • The shape of this φ(T) is useful for understanding the temperature distribution of plasma from a set of observable fluxes {Fji} • Its shape can tell us about heating and cooling processes In summary • Be careful to find out where the light you’re measuring really comes from • Spectroscopy is incredibly powerful for understanding the Sun • Once you know what the energy output truly is, you can start thinking about the energy input • …and do some quantitative physics! "