Page 1 Hot topics on Galaxy Formation and Evolution 2. The IMF of galaxies Roberto Saglia Max-Planck Institut für extraterrestrische Physik Garching, Germany Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 2 2. The varying IMF of Galaxies • Introduction • What is the IMF and how to measure it • The integrated galaxy IMF • High-mass IMF constraints from emission: Ha • Low-Mass IMF constraints from absorption: NIR indices • IMF constraints from mass measurements: stellar dynamics and gravitational lensing. The problem of dark matter. Padova, March 2012 Hot topics on galaxy formation and evolution 2 References Page 3 Weidner 2004 & Kroupa, 2004, MNRAS, 350, 1503 (WK) Weidner & Kroupa 2005, ApJ, 625, 754 Howersten & Glazebrook 2008, ApJ, 75, 163 (HG) Van Dokkum & Conroy, 2010, Nature, 468, 940 Van Dokkum & Conroy, 2011 ApJ, 735, L13 Thomas et al. 2011, MNRAS, 415, 545 Cappellari et al., 2012, Nature, in press Kroupa et al. 2012, Stellar Systems and Galactic Structure, Vol. V (K12) Padova, March 2012 Hot topics on galaxy formation and evolution 2 A more modern view of the Hubble classification Page 4 Kormendy & Bender, 2012, ApJS, 198, 2 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 5 Milky Way (Sbc-galaxy) in different wavebands Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 6 The color-magnitude diagram of all Hipparcos stars with relative distance error < 0.1. Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 7 The HRD of a Globular Cluster The stars of a stellar cluster have the same age and the same chemical composition dependence only on mass HST Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 8 The HRDs of Open Clusters of Different Ages Zero Age Main Sequence (ZAMS) high masses Plejades small masses Turn-off point Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 9 The Initial mass function (IMF) IMF = distribution of stellar masses at the time of birth. Definition: Φ(m) = number of stars formed per mass-interval [m,m+dm] and per total mass of formed stars (Unit = 1/mass2) mΦ(m) = mass of stars per mass-interval and per total mass mΦ(m)dm = mass of stars within mass-interval and per total mass resulting normalization: ∫ mΦ(m)dm = 1 Observational determination of IMF: " from the solar neighbourhood using star counts and taking into account the lifetime of stars; requires assumptions concerning the star formation history in the solar neighbourhood: low-mass stars (M < 1MΘ) have not disappeared since their creation, while O- and B-stars disappear within 107 yrs " from star counts in star forming regions with infrared photometry (dust extinction smaller in IR) Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 10 Note: sometimes it is claimed that the IMF is bimodal, i.e. high-mass and low-mass stars form in separate regions, or, in the same region at different times. Common analytical approximations: Salpeter-IMF: 0.1 ≤ (APJ 121, 161 (1955)) m ≤ 100 : Φ(m) 0.17m−2.35 M Miller-Scalo-IMF: (see also figure on next page) mass-fraction 0.1 ≤ m ≤ 0.5 M 0.93m−0.85 0.31 0.5 ≤ m ≤ 1.0 M 0.46m−1.85 0.31 1.0 ≤ m ≤ 3.16 M 0.46m−3.4 0.26 m ≤ 100 0.21m−0.27 M 0.12 3.16 ≤ Padova, March 2012 Φ(m) = Hot topics on galaxy formation and evolution 2 Page 11 Pagel: Nucleosynthesis and Chemical Evolution, Cambridge University Press 1997 Padova, March 2012 Hot topics on galaxy formation and evolution 2 The IMF of Star Clusters ξ (m) ∝ m−α Page 12 α = 2.35 : Salpeter Kroupa IMF 0.5M e WK Padova, March 2012 0.08M e Hot topics on galaxy formation and evolution 2 Page 13 A collection of IMFs Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 14 IMFs and cumulative functions Padova, March 2012 Hot topics on galaxy formation and evolution 2 ... has a constant slope (to first approximation) Padova, March 2012 Page 15 K12 Hot topics on galaxy formation and evolution 2 Metallicity and density Page 16 High mass slope α 3 : flatter at high densities and low metallicities? Marks et al. 2012, MNRAS, in press Padova, March 2012 (modeling involved) Hot topics on galaxy formation and evolution 2 Page 17 Observed ranges of the IMF? Marks et al. 2012, MNRAS, in press Padova, March 2012 Hot topics on galaxy formation and evolution 2 Gas ejection rate of all stars: Basics of Galaxy Evolution M = Ms + Mg E (t ) = t −τ m ( m ) Φ (m)dm mass of stars dying at time t m - wm : ejected mass dM s =Ψ−E dt ⎧Ψ = star formation rate ⎨ ⎩ E = gas ejection rate of all stars Ψ t −τ m ( m ) Φ (m): birth rate at t-τ m = death rate at time t, stars of different generations are involved τ m : main-sequence lifetime at mass m = −Ψ + E + f − e Remnant mass: Spectrum of a galaxy f (λ , t ) = ∫ f m,Z (λ , t − t ')Ψ(t − t ')Φ(m, t ')dt ' dm Luminosity : L(t ) = ∫ f (λ, t )dt m mt : turn-off mass at time t = lowest ⎧ f = rate on infalling gas ⎨ ⎩e = rate on ejected gas dt ∫ [m − w ] Ψ mt ⎧ M = total mass ⎪ ⎨ M s = mass in stars ⎪ M = mass in gas ⎩ g dM = f −e dt dM g ∞ Page 18 ⎧⎪ wm = 0.11m + 0.45M (m < 6.8M ) ⎨ (m ≥ 6.8M ) ⎪⎩ wm = 1.5M Mass-to-light ratio M / L(t ) see: Tinsley 1980 and corrections by Maeder 1992 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 19 IMF Variations at low and high m Steep IMF at low masses: Flat IMF at high masses: Lots of M-dwarfs Lots of 'black' remnants, i.e. neutron stars and black holes, X-ray binaries Old stellar populations have large M/L Extrapolated 'evidence' for high metallicities?! Marks et al. 2012 MNRAS Padova, March 2012 Old stellar populations have large M/L. Lots of SNII: More a elements Hot topics on galaxy formation and evolution 2 Page 20 Basics of star formation We first discuss under which conditions a gas cloud will collapse. We consider an infinite uniform gas cloud in hydrostatic equilibrium with initial conditions: ρ = ρ0 = const., T = T0 = const and v 0 = 0. The basic equations of hydrodynamics (without magnetic fields!) which govern the dynamics of the cloud are: Continuity equation: Equation of motion: Poisson Equation: Equation of state for an ideal gas: ∂ρ + ∇ ⋅ ( ρ v) = 0 ∂t ∂v ∇P + ( v ⋅ ∇)v = − ∇Φ ∂t ρ ∇ 2Φ = 4π G ρ P= Padova, March 2012 k ρT = Cs2 ρ µ Hot topics on galaxy formation and evolution 2 Page 21 where Ф is the gravitational potential and Cs is the speed of sound. We assume that we know an equilibrium solution to which we add a small perturbation: ρ = ρ0 + ρ1 P = P0 + P1 Φ = Φ 0 + Φ1 v = v0 + v1 We furthermore assume a constant speed of sound (and temperature!, the gas is assumed to be isothermal). Inserting the perturbed solution into the equation and neglecting all second order terms, we obtain differential equations for the perturbations. These can be combined to give a wave equation: which is solved by: ⎛ 2 1 ∂ 2 4π G ρ0 ⎞ ⎜ ∇ − C 2 ∂t 2 + C 2 ⎟ ρ1 = 0 ⎝ ⎠ s s ρ1 = Aexpi( k ⋅ r − ω t) This represents a planar density wave with wave vector k and angular velocity ω fulfilling the dispersion relation 2 2 2 ω = k Cs − 4π G ρ0 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 22 This relation shows that ω is real only when k is larger than a critical value kj = (4πGρ0)1/2/ Cs, or when the wavelength is smaller than a critical wavelength 1/ 2 λj = 2π ⎛ π ⎞ =⎜ ⎟ k j ⎝ G ρ0 ⎠ Cs In that case the density perturbation is oscillating, but the amplitude will not grow. However, if: λ > λj the frequency ω becomes imaginary, so that density perturbations will grow exponentially with time. Thus the region of the density perturbation has to be large enough (and massive enough) to start collapsing. This is known as the Jeans criterion. The collapse time can be estimated via: τ collapse ≈ 1/ 2 λj ⎛ π ⎞ =⎜ ⎟ Cs ⎝ G ρ 0 ⎠ We finally define the critical or Jeans mass as: or, explicitely Padova, March 2012 4 M j = volume ⋅ density = πλ 3j ρ 3 ⎛ πk ⎞ Mj =⎜ ⎟ ⎝ Gµ ⎠ 3/ 2 T 3/ 2 ρ −1/ 2 Hot topics on galaxy formation and evolution 2 Page 23 ⎛ T ⎞ M j = 1.2 ⋅105 M e ⎜ 2 ⎟ ⎝ 10 K ⎠ 3/ 2 ⎛ ⎞ ρ ⎜ −24 3 ⎟ 10 g / cm ⎝ ⎠ −1/ 2 µ −3/ 2 Therefore, for typical properties of the diffuse ISM, Mj is much larger than the mass of a star. Giant molecular clouds, on the other hand, are well above their Jeans mass. However, the temperature of the gas in molecular clouds unterestimates the internal energy significantly. Most of the internal energy is in the relative motions of the cloudlets which is of the order of up to 10km/s (see above). Therefore, giant molecular clouds do not seem to be in a collapse phase yet but rather appear as virialized quasi-stationary systems. Magnetic field pressure can further stabilize a cloud. (Side remark: if all molecular clouds in the Milky Way would collapse in the time scale estimated 1/(Gρ)1/2, the star formation rate in the Milky Way would have to be 200MΘ/yr, observed is only ≈ 1MΘ/yr). Collisions between clouds, or tidal forces (e.g. from the Galactic spiral density wave) can however finally trigger the collapse. Once the cloud is compressed and the density increases, the collapse can start. Also instabilities in the cloud itself may finally cause it to collapse without external influence. At lower metallicities, cooling is less efficient, therefore higher temperatures are present in the clouds --> larger masses are needed for collapse --> a flatter IMF at high masses is expected. At high densities, the crossing time is so short that protostars will 'coagulate' tending to create again a flatter IMF at high masses. Padova, March 2012 Hot topics on galaxy formation and evolution 2 Fragmentation of collapsing clouds Page 24 From the observations, we know that molecular clouds have complex substructure indicating that fragmentation takes place already in the pre-collapse phase. We now discuss the conditions which determine the size of the fragments. For fragmentation to proceed, Mj should not be constant during collapse but should decrease with increasing density. For an isothermal collapse we found M j ∝ ρ −1/ 2 (isothermal collapse) which means Mj indeed decreases with rising density. The situation is opposite for an adiabatic collapse, where T ~ P2/5, so from P ~ ρT for an ideal gas, or T ~ ρ2/3, we have M j ∝ T 3/ 2 ρ −1/ 2 ∝ ρ 1/ 2 (adiabatic collapse) In other words, Mj increases with density in an adiabatic collapse. To ensure that the collapse is isothermal, the gravitational energy released during the collapse of the cloud should be radiated away in order not to increase the internal energy of the gas. Therefore the collapse time τcollapse should be much larger than the thermal adjustment timescale τadj: τ collapse τ adj Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 25 As we have seen, the collapse time for a cloud(let) only depends on its density and is: τ collapse ≈ ( G ρ ) −1/ 2 ⎛ ⎞ ρ ≈ 108 y ⎜ −24 3 ⎟ ⎝ 10 g / cm ⎠ −1/ 2 On the other hand, the thermal adjustment timescale for interstellar clouds is found to be of the order τadj ≈ 100y, which obviously fulfills the criterion for isothermal collapse. Thus Mj ~ ρ−1/2 which in turn leads to continued fragmentation during collapse. We now want to estimate the lower mass limit for fragmentation. For this purpose we have to know when the collapse will change from isothermal to adiabatic. We first calculate the energy radiated per second during the isothermal collapse. The total gravitational energy released in the collapse is Eg = GM2/R during the timescale τ ≈ (Gρ)−1/2, so that the luminosity is given by Eg 1/ 2 ⎛ M ⎞ GM 2 1/ 2 3 Lg ≈ = (G ρ ) = (3G / 4π ) ⎜ ⎟ τ R ⎝ R⎠ 5/ 2 This luminosity should be much smaller than that of a black body with the same temperature and volume. Because, if the cloud does emit black body radiation, it means that it is optically thick, has reached an equilibrium configuration and does not loose energy Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 26 efficiently anymore. The luminosity emitted in the black body case would be: Lc = 4π R 2σ T 4 When Lc ≥ Lg the collapse becomes adiabatic because the released gravitational energy cannot be radiated away. Obvioulsy, the condition Lc = Lg corresponds to a lower mass limit for fragmentation: ( Mj = π /9 9 ) ( 1/4 σG 3 ) ( −1/2 R/µ ) 9/4 T 1/4 = 10−2 M T 1/4 where again R = (3/4π)1/3(M/ρ)1/3 and the equation for the Jeans mass given above have been used. Because of the exponent 1/4 of the temperature, the temperature dependence of this mass is small. For temperatures of 1000 K we obtain 0.05MΘ. This shows why during the transition from diffuse HI clouds to molecular clouds fragmentation already takes place. It is this fragmentation which may actually slow down the further collapse of the molecular cloud because the fragments can start to move as individual almost ’collisionless particles’. This permits relative velocities of the cloudlets which are larger than what corresponds to their internal thermal motions and helps to stabilize the molecular cloud. Interestingly, the lower mass limit which we obtain for the cloudlets is similar the lowest masses which are observed for stars. Padova, March 2012 Hot topics on galaxy formation and evolution 2 The maximum star mass Stars are born in Star Clusters of a Star Cluster and only there! Page 27 Largest star mass in a star cluster: m max 1= mmax* ∫ ξ (m) dm mmax Mass of the star cluster: M ecl = mmax ∫ mξ (m) dm mmin Maximum stellar mass: m max* (150M ?) Padova, March 2012 K12 Hot topics on galaxy formation and evolution 2 The mass distribution of Star Clusters Page 28 −β P ( M ecl ) ∝ M ecl β ≈ 2 − 2.35 Hunter et al. 2003, AJ, 126, 1836 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 29 The SFR- M ecl ,max relation Mass of a star formation event: M ecl ,max ∫ M TOT = M ecl P ( M ecl )dM ecl M ecl ,min SFR = M TOT / δ t 100 10 dt=1 Myr → M ecl ,max = f ( SFR ) Dotted: b=2 Dashed: b=2.4 M ecl = mmax ∫ mξ (m)dm mmin → mmax = g ( SFR ) WK Padova, March 2012 Hot topics on galaxy formation and evolution 2 The composite IMF of galaxies Page 30 All stars in galaxies are made in star clusters: ξ IGIMF (m, t ) = ∫ t ∫ M ecl ,max ( SFR ( t ')) 0 M ecl ,min ξ (m ≤ mmax ( M ecl ))P( M ecl )dM ecl dt ' / t 100 Myr burst 14 low level bursts Low constant SF Padova, March 2012 Hot topics on galaxy formation and evolution 2 The slope(s) of the IGIMF WK Padova, March 2012 Salpeter Page 31 K12 Hot topics on galaxy formation and evolution 2 IMF constraints from emission Page 32 Star forming galaxies have ionized gas that produces emission lines: The Orion Nebula in optical light, from the Hubble Space Telescope. The bar at the lower left is a wall of gas viewed edge-on. This gas glows because it is ionized by the four hot ”Trapezium stars” just to the left of center. Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 33 Star formation rate indicators: Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 34 Diagnostic plots Ionization photons can be produced by young stars in regions of star formation or by central non-thermal sources, i.e. AGNs. Diagnostic plots allow to disentagle the two cases: q: ionization parameter z: metallicity Kewley et al. 2001,ApJ, 556, 12 The lines show the line ratios predicted when the ionization source is a star bursting population for different metallicities. Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 35 The photoionizing continuum of AGNs has a significant fraction of its energy in the X-rays. These photons can penetrate deeply into the neutral regions and produce large partly ionized zones with H+/H~0.2-0.4 that do not exist in HII regions. Ions of atoms having ionization potentials similar to H (O, S, N…) will also be present. The hot free electrons produced by X-ray photoionization are available to excite and increase the strenght of lines produced by collisional excitation, like OIII, SII, NII. Therefore we expect stronger intensities of the [OIII] 5007 line relative to the Hb, and stronger intensities of the [SII] or [NII] lines with respect to Ha than in star forming regions. Padova, March 2012 Hot topics on galaxy formation and evolution 2 The equivalent width Page 36 I cont − I λ w = ∫ Rλ d λ = ∫ dλ line line I cont The equivalent width w evidently has the unit of a wavelength. Geometrically, the product of continuum flux times equivalent width covers the same area in a spectrum as the absorption or emission line does. The equivalent width increases with increasing number of ions in the corresponding quantum state. In the optically thin case, the increase is linear. For higher densities of ions, the absorption line starts to saturate and the equivalent width almost stops increasing (at which density this happens depends on the Doppler broadening). Padova, March 2012 Hot topics on galaxy formation and evolution 2 The Kennicutt method Page 37 The equivalent width of Hα vs B − V color can be used to constrain the initial mass function of stars in spirals (figure from Kennicutt et al. ApJ No 272, p54, 1983) The three evolutionary models shown in the figure use different IMFs (from top to bottom): Ф ~ m−2 Ф ~ m−2.35 Ф ~ Miller-Scalo IMF Wλ ; or: Wλ ; Padova, March 2012 Hα flux continuum flux present SFR luminosity of old stars Hot topics on galaxy formation and evolution 2 How to measure the IMF slope: the Kennicutt method Page 38 The Ha equivalent width is sensitive to the ratio of the number of massive stars (driving the emission line strenght) to the number of giant stars (driving the continuum). When combined with a color it can constrain the slope of the IMF. log Hα EW Age Metallicity α = Γ +1 t=1.1 decl. Z=0.025 t=1.5 inc. 0.005 HG08 Padova, March 2012 Hot topics on galaxy formation and evolution 2 g-r Page 39 SDSS Padova, March 2012 Hot topics on galaxy formation and evolution 2 SDSS Data Products Page 40 • Images of 1/4 of the full sky (10000 square-degrees) in 5 wavelength bands • photometry of 100 million objects • spectra of > 1 million objects Padova, March 2012 Hot topics on galaxy formation and evolution 2 Hoversten & Glazebrook 2008 The SDSS Sample Page 41 130602 galaxies with good lines S/N, Ha representing SF and not AGN. Colors corrected for dust using Balmer decrement and K-corrected to z=0.1. χi2 (Γ, Z , t , SFH ) = (ci − c(Γ, Z , t , SFH )) / σ c2i ) 2 + ( wi − w(Γ, Z , t , SFH )) / σ w2i ) 2 Dust vector : Balmer decrement f = Hα / H β Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 42 A variable IMF slope? α −1 SMC MW Salpeter Many alternative explanations tested and rejected: Malquist bias, redshift, aperture, extiction, f ratio, bursty star formation histories. HG08 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Constraints on the IMF from absorption lines Page 43 S ⎛ ⎞ I ⎜ A ⎟ = Δ − ∫ dλ C ⎝ ⎠ Δ 0 C S Δ GC NGC 6626 Worthey et al. 1994 ApJS 94, 687 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Simple Stellar Population models Page 44 1 Age, 1 Metallicity, Initial Mass Function IMF ∝ M −α ,α ≈ −2.35 E-AGB 5-6 HB 4-5 SGB 2-3 RGB 3-4 dMS 0-1 MS-TO 1-2 f c*, j = Fc*, j ∑F L. Greggio Padova, March 2012 * c, j j Hot topics on galaxy formation and evolution 2 Simple Stellar Population models Page 45 1 Age, 1 Metallicity, Initial Mass Function E-AGB 5-6 HB 4-5 SGB 2-3 RGB 3-4 dMS 0-1 MS-TO 1-2 f c*, j * ⎛ ⎞ I j * * = ; Fl , j = Fc , j ⎜1 − ⎟ * ⎜ Δ⎟ ∑ Fc, j ⎝ ⎠ Fc*, j j IMF ∝ M −α ,α ≈ −2.35 * j Fitting Function: I (Te , g , Z ) ⎛ ∑F ⎞ ⎛ F⎞ ⎜ ⎟ * l, j I SSP = Δ ⎜1 − l ⎟ = Δ ⎜1 − j * ⎟ = ⎝ Fi ⎠ ⎜ ∑ Fc , j ⎟ j ⎝ ⎠ * * ⎛ ⎞ F c, j I j * * ⎜ ∑ Fc , j − Fc , j + ⎟ Δ ⎟ j ⎜ =Δ * ⎜ ⎟ F ∑ c, j ⎜⎜ ⎟⎟ j ⎝ ⎠ = ∑ I *j × f c*, j j Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 46 In the optical: M-dwarfs not important Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 47 NIR indices in stars Strong features present only in M dwarfs 9900 10000 o A M dwarfs flux (more) important in the NIR A steep IMF will have lots of M dwarfs 8300 Padova, March 2012 8500 Hot topics on galaxy formation and evolution 2 Page 48 NIR indices Van Dokkum & Conroy 2010, Nature, 468, 940 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 49 Constraining the IMF Padova, March 2012 Hot topics on galaxy formation and evolution 2 NaI and FeH Wing-Ford indices in GCs Page 50 M31 globular clusters have 'normal' IMF Van Dokkum & Conroy, 2011 ApJ 375, L13 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 51 Comparing GCs and Es Padova, March 2012 Hot topics on galaxy formation and evolution 2 IMF constraints from mass measurements Page 52 • Compute the luminosity of a galaxy L • Compute the M/L of the stellar population that best reproduces the colors and/or spectrum/line indices of a galaxy for a given IMF • Measure the mass of a galaxy using: stellar dynamics or gravitational lensing M • Compute the ratio M/L • Compare the two values, taking into account the amount of dark matter and the mass of central black holes Padova, March 2012 Hot topics on galaxy formation and evolution 2 Gas ejection rate of all stars: Basics of Galaxy Evolution M = Ms + Mg E (t ) = t −τ m ( m ) Φ (m)dm mass of stars dying at time t m - wm : ejected mass dM s =Ψ−E dt ⎧Ψ = star formation rate ⎨ ⎩ E = gas ejection rate of all stars Ψ t −τ m ( m ) Φ (m): birth rate at t-τ m = death rate at time t, stars of different generations are involved τ m : main-sequence lifetime at mass m = −Ψ + E + f − e Remnant mass: Spectrum of a galaxy f (λ , t ) = ∫ f m,Z (λ , t − t ')Ψ(t − t ')Φ(m, t ')dt ' dm Luminosity : L(t ) = ∫ f (λ, t )dt m mt : turn-off mass at time t = lowest ⎧ f = rate on infalling gas ⎨ ⎩e = rate on ejected gas dt ∫ [m − w ] Ψ mt ⎧ M = total mass ⎪ ⎨ M s = mass in stars ⎪ M = mass in gas ⎩ g dM = f −e dt dM g ∞ Page 53 ⎧⎪ wm = 0.11m + 0.45M (m < 6.8M ) ⎨ (m ≥ 6.8M ) ⎪⎩ wm = 1.5M Mass-to-light ratio M / L(t ) see: Tinsley 1980 and corrections by Maeder 1992 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 54 The M/L ratio of stellar populations For main sequence stars: (0.1M < M < 100M ) v Mass–luminosity relation L∝M4 for M > 0.6M L∝M2 for M < 0.6M : : Padova, March 2012 Hot topics on galaxy formation and evolution 2 M/L from SSPs Page 55 If you now the (mean) age and metallicity of a galaxy, Simple Stellar Population Models can predict their mass-to-light ratio for a given IMF. M / LKroupa = 0.56 M / LSalpeter The problem is then to compute the mean age and metallicity of a galaxy. Maraston 1998, 300, 872 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 56 The age-metallicity degeneracy problem: colors are hopelessly degenerate in age and metallicity ! absorption lines in the blue can help: hotter stars at MSTO have stronger Balmer lines, this allows to break the degeneracy (see further below) (Worthey 1993) Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 57 Illustration of effects of age and metallicity on isochrones (Worthey 1993), both higher age and higher metallicity make the integrated colors redder and cause stronger metal absorption lines (spectra get more dominated by K, M giants) Padova, March 2012 Hot topics on galaxy formation and evolution 2 Ages and metallicities Page 58 Lick indices Mgb , < Fe >, H β to break the age-metallicity [Z / H ] = log(Z / H ) − log(Z / H )e degeneracy Ze = 0.02 Trager et al. 2000 AJ, 120, 165 Padova, March 2012 Hot topics on galaxy formation and evolution 2 The Bulge of M31 Padova, March 2012 Page 59 Saglia et al. 2010, A&A, 509, 61 Hot topics on galaxy formation and evolution 2 The M/L of Coma galaxies Page 60 M/L from dynamics Average of Salpeter SSP M/L M / LKroupa = 0.56M / LSalpeter Padova, March 2012 Hot topics on galaxy formation and evolution 2 The IMF changes with sigma Page 61 Thomas et al. 2011, Wegner et al. 2012 Padova, March 2012 Hot topics on galaxy formation and evolution 2 The density profiles Page 62 If all 'stellar' mass in excess of Kroupa is 'moved' to the DM profiles, they all become isothermal... Padova, March 2012 Hot topics on galaxy formation and evolution 2 Jeans Modeling Padova, March 2012 Page 63 Hot topics on galaxy formation and evolution 2 Highest density ellipticals Page 64 4000 densest ellipticals from SDSS: dark matter not important Salpeter IMF needed to explain the dynamical masses Dutton et al. 2012, MNRAS in press Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 65 Masses of galaxies from gravitational lensing Padova, March 2012 Hot topics on galaxy formation and evolution 2 Basics of Gravitational Lensing Page 66 The light path from the source to the observer can then be broken up into three distinct zones: 1. Light travels from the source to a point close to the lens through unperturbed spacetime, since b « Dd. 2. Near the lens the light is deflected. 3. Light travels to the observer through unperturbed spacetime, since b « Dds. Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 67 In a naive Newtonian approximation one would derive: α= v z 1 dΦ 1 dΦ = ∫ dt = 2 ∫ dl c c dz c dz ∗ *: acceleration in z direction; because the acceleration doesn’t depend on the energy of the photons, gravitational lenses are achromatic. This result differs only by a factor of two from the correct general relativistic result: 2 α = 2 ∫ ∇ ⊥ Φ dl G.R. c where the deflection angle α , written as vector α perpendicular to the light propagation l, is the integral of the potential gradient perpendicular to the light propagation. For a point mass the potential can be written as: Φ(l, z) = Therefore: −GM (l 2 + z 2 )1/2 dΦ +GMz = 2 dz (l + z 2 )3/2 Padova, March 2012 = (∇ ⊥ Φ) Hot topics on galaxy formation and evolution 2 Page 68 After integration: 2 α= 2 c ∞ ∞ ∞ ⎤ GMz 4GMz dl 4GMz ⎡ l dl = = ∫−∞ (l 2 + z 2 )3/ 2 c 2 ∫0 (l 2 + z 2 )3/ 2 c 2 ⎢⎣ z 2 (l 2 + z 2 )1/ 2 ⎥⎦ 0 Thus the deflection angle α for a light ray with impact parameter b = z near the point mass M becomes: α= 4GM 2 RS = c 2b b where RS = 2GM/c2 is the Schwarzschild radius of the mass M, i.e. the radius of the black hole belonging to the mass M. Therefore for the sun (MΘ ≈ 2 · 1033 g → RS ≈ 3.0 km) we get a deflection angle α at the Radius of the sun (≈ 700000km) of: α ,R 1.7'' In order to calculate the deflection angle α caused by an arbitrary mass distribution (e.g. a galaxy cluster) we use the fact that the extent of the mass distribution is very small compared to the distances between source, lens and observer: Δl « Dds and Δl « Dd Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 69 Therefore, the mass distribution of the lens can be treated as if it were an infinitely thin mass sheet perpendicular to the line-of-sight. The surface mass density is simply obtained by projection. The plane of the mass sheet is called the lens plane. The mass sheet is characterized by its surface mass density ∑ (ξ ) = ∫ ρ (ξ , l ) dl Δl The deflection of a light ray passing the lens plane at ξ by a mass element 2 dm = ∑ (ξ ')d ξ ' at ξ ' is: Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 70 dα = 4Gdm 2 c |ξ −ξ ' | To get the deflection caused by all mass elements, we have to integrate over the whole surface. Doing this we must take into account that, e.g., the deflection caused by mass elements lying on opposite sides of the light ray may cancel out. Therefore we must add the deflection angles as vectors: 4G (ξ − ξ ')∑ (ξ ') 2 α (ξ ) = 2 ∫ d ξ' 2 c |ξ −ξ ' | Special case: For a spherical mass distribution the lensing problem can be reduced to one dimension. The deflection angle then points toward the center of symmetry and we get: α (ξ ) = 4GM (< ξ ) c 2ξ where ξ is the distance from the lens center and M(< ξ) is the mass enclosed within radius ξ, ξ M (< ξ ) = 2π ∫ ∑ (ξ ')ξ ' dξ ' 0 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 71 Lensing Geometry and Lens Equation deflection angle just computed Important relations: αˆ ⋅ Dds = α ⋅ Ds θ ⋅ Ds = β ⋅ Ds + αˆ ⋅ Dds Note: The distances D are angular diameter distances. Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 72 Using the previous two equations one obtains the so called lens equation: β = θ −α = θ − Dds αˆ Ds The lens equation relates the real position (angle) of the source (without a lens) with the position of the lensed image. Important note: only angular distances are needed for deriving the lens equation. In general, i.e. over cosmological distances: Dds ≠ Ds − Dd. Consider now a circularly symmetric lens with an arbitrary mass profile. Due to the rotational symmetry of the lens system, a source, which lies exactly on the optical axis (θ = α ↔ β = 0 ) is imaged as a ring. This ring is the so called Einstein ring: β =0 a θ =α = Dds D 4G M (< ξ ) ⋅ αˆ = ds ⋅ 2 ⋅ Ds Ds c ξ The radius of the Einstein ring can be calculated using the previous equation and ξ = Ddθ: Dds 4G θ = ⋅ 2 ⋅ M <θE Ds Dd c 2 E Padova, March 2012 Einstein radius Hot topics on galaxy formation and evolution 2 The SLAC survey Page 73 Bolton et al. 2006, ApJ, 638, 703 Sloan Lens ACS Survey: search for spectra having multiple nebular emission lines at redshift higher than the target galaxy. Measure mass inside the Einstein radius z=0.5812 Padova, March 2012 @0.32 Hot topics on galaxy formation and evolution 2 Page 74 Comparison Lensing-Dynamics rEin ≈ 0.5re rEin ≈ 0.75re Dynamical models with dark matter match lensing results well the dynamical M/L ratios are good. Dynamical models without dark matter underestimate the lensing mass Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 75 The SWELLS survey The Sloan WFC Edge-on Late-type Lens Survey Brewer et al, astro-ph/1201.1677 Padova, March 2012 Spirals have a Kroupa/ Chabrier IMF Stellar mass fraction >1 if Salpeter Hot topics on galaxy formation and evolution 2 Constraints on the IMF from the IGM metallities The light of the (old, Page 76 massive) ellipticals of galaxy clusters comes from ~solar stars. The oxigen and silicon in the IGM came from massive stars that exploded as SNII. → the M O / LB or the M Si / LB ratios probe the IMF slope. They are compatible with a Salpeter IMF. Renzini 2005 Padova, March 2012 Hot topics on galaxy formation and evolution 2 Conclusions 1. Page 77 • The IMF observed in our MW is Salpeter power-law at M>Msol and turns to Kroupa/Chabrier at smaller stellar masses. • GCs at low metallicities and/or Ultra Compact galaxies at high densities might have a steeper than Salpeter IMF at high stellar masses • Speculations that larger numbers of low mass stars than Kroupa/Chabrier might be present in high metallicity systems Padova, March 2012 Hot topics on galaxy formation and evolution 2 Conclusions 2. Page 78 • If stars are only produced in star clusters, then the IMF of a galaxy is the sum of the IMFs of all star clusters produced there • A star cluster will be populated up to a maximal star mass, that depends on the total mass of the star cluster • The maximal total mass of a star cluster is set by the current star formation rate • Galaxies with low star formation rate histories will have steeper IMFs, because they will seldon manage to build high mass stars. Padova, March 2012 Hot topics on galaxy formation and evolution 2 Conclusions 3. Page 79 • The EW of Ha carries information on the IMF slope (at high mass) • The modeling of the Ha EW and colors of SLOAN galaxies suggests that the IMF slope is a function of the galaxy luminosity: dwarf galaxies have steeper IMFs than giant spirals (at high mass). • Modeling of the stellar dynamics of elliptical galaxies allows to constrain the luminous and dark matter content • Stellar M/L ratio from age and metallicity derived from line indices ( M / L )dyn / ( M / L )* indicates a Kroupa IMF at low σ and a Salpeter (or even steeper IMF at low masses) at high σ • NIR indices suggest steeper than Salpeter IMF (at low mass) in high s Es Padova, March 2012 Hot topics on galaxy formation and evolution 2 Page 80 Yet to come Constraints on the evolution of the IMF with time might come from the analysis of the evolution of elliptical galaxies --> see final lecture on size evolution. Padova, March 2012 Hot topics on galaxy formation and evolution 2