Modelling the Formation of Planets
Transcrição
Modelling the Formation of Planets
Modelling the Formation of Planets Wilhelm Kley Institut für Astronomie & Astrophysik Abtlg. Computational Physics 20. Mai, 2010 W. Kley Planet Formation Organisation • Context - Solar and extrasolar Planets - Planet Formation • Accretion Disks • Planet-Disk Modelling - Numerics - Results • Summary/Outlook (A. Crida) W. Kley Stuttgart: 20. Mai, 2010 1 Solar System The Planets Sun Mercury Venus Earth Mars Jupiter Saturn Uranus Neptun (Pluto) W. Kley Stuttgart: 20. Mai, 2010 2 Solar System Summary 8 Planets: Mercury to Neptune 3 Dwarf Planets: Ceres, Pluto, Eris, Makemake, Haumea Minor bodies: TNO, asteroids, comets Tiny bodies: meteorites, dust • • • • • • • coplanar, circular, uniform orbits prograde rotation (with exceptions) 99% of mass in Sun 99% of angular momentum in planets Age: about 4.5 billion years Solid and gaseous planets Titius-Bode rule W. Kley Stuttgart: 20. Mai, 2010 3 Exoplanets Main Characteristics Planets orbiting other solar-like stars • Total detected: ≈ 450 (May 2010) (by RV method, Transits, direct imaging, microlensing) • Transiting Planets: ≈ 80 • Systems with 2 or more planets: ≈ 40 • Planetary Systems in Binaries: 50 • Planetary Systems in Mean Motion Resonance: 8 • Fraction of stars with planets: ≈ 15% List: http://exoplanet.eu/ W. Kley Stuttgart: 20. Mai, 2010 Source: http://www.oklo.org/ (by Greg Laughlin) 4 Exoplanets First Object: 51 Peg K = 55 m/s P = 4.23 days e = 0.00 M = 0.45 MJup/ sin i (Mayor & Queloz, 1995) Kepler III: Amplitude (K), Period (P ) and Stellar mass (M∗) ⇒ planet mass (Mp) & distance (a) Sinusoidal radial velocity curve ⇒ circular orbit W. Kley Stuttgart: 20. Mai, 2010 5 Exoplanets Direct Imaging (Nov. 2008) 17. brighest star, 25 Light yrs, Age: 200 Mio. yrs, Debris disk W. Kley Stuttgart: 20. Mai, 2010 6 Exoplanets Direct Imaging II Star: Fomalhaut: (Mag: 1.16) 2.59MSun Planet: Distance 115 AU Orbit 900 Yrs Mp ≈ 3MJup at inner edge of Debris Ring W. Kley Stuttgart: 20. Mai, 2010 7 Exoplanets Mass vs Distance Sonnensystem Hot Jupiters at small distances M sin i [MJup] 10 MEarth = aM ercury (Data: exoplanet.eu) 1 .1 W. Kley 1 MJup 300 = 0.4AU .1 1 Abstand [AE] Stuttgart: 20. Mai, 2010 10 8 Exoplanets .8 Eccentricity Distribution Sonnensystem Extrasolar OGLE Exzentrizität .6 Large eccentricity spread - similar to binary stars .4 Gravitational scattering (Data: exoplanet.eu) .2 0 W. Kley .1 1 Abstand (AE) Stuttgart: 20. Mai, 2010 10 9 Planet Formation Grand Picture Planets form in protoplanetary disks ≡ Accretion Disks W. Kley Stuttgart: 20. Mai, 2010 10 Accretion disk Overview Circumstellar disk around young stars (angular momentum conservation) Matter spirals inwards & angular momentum outwards Require: Turbulent Transport Interaction with magnetic field of star Stellar activity → ionisation of disk ? W. Kley Stuttgart: 20. Mai, 2010 11 Accretion disk Magnetized Disks For a Keplerian Disk: ΩK (r) = GM r3 1/2 Stability Criterion for Purely Hydrodynamic Disks (Rayleigh) dL > 0, dr (LK = ΩK r2 ∝ r1/2) −→ stable Stability Criterion for magnetized Disks (Chrandrasekhar, et al.) dΩ > 0, dr (ΩK ∝ r−3/2) −→ unstable Magneto-Rotational-Instability - MRI (Balbus & Hawley, 1993) W. Kley Stuttgart: 20. Mai, 2010 12 Accretion disk MRI-Principle Lower Orbit: Larger Ω, smaller L Spring: Transferral of Angular Momentum from lower to higher orbit =⇒ Instability Study the MRI in accretion disks numerically (saturation) Calculate transport coefficients for angular momentum W. Kley Stuttgart: 20. Mai, 2010 13 Accretion disk ∂% + ∇ · (%v) ∂t ∂(%v) + ∇ · (%vv) ∂t ∂e + ∇ · (ev) ∂t ∂B − ∇ × (v × B) ∂t ∇·B Equation of state Ideal RMHD-Equations = 0 = −∇p − ρ∇Ψ = −p∇ · v 1 + (∇ × B) × B 4π +(Q : ∇)v +∇ · Q −∇ · F = 0 = 0 p = (γ − 1)e Ideal Hydrodynamics & Viscosity (Q) ⇒ Viscous Disks Magnetic Terms (B) ⇒ Magnetic disks radiative Diffusion (F) W. Kley Stuttgart: 20. Mai, 2010 14 Accretion disk The local Shearing Box Effect of Keplerian shear flow Numerical implementation Small section of accretion disk in equatorial plane (scale free) Easy: Periodic boundaries in y ≡ ϕ and z Tricky: shearperiodic in x ≡ r (conservation properties) W. Kley Stuttgart: 20. Mai, 2010 15 Accretion disk No vertical stratification Shearing box with dimension: x = 1.0 x y = 4.0 x z = 1.0 Numerical resolution 64x128x64 ~ No vertical gravity, vertical B-field, plasma-β=400 (Pgas/Pmag ) Animation of magnetic field: y −z slice, covers 16 Orbital periods Initial random perturbation of velocity and temperature (M. Flaig) W. Kley Stuttgart: 20. Mai, 2010 16 Accretion disk vertically stratified disks 3D Magneto-hydrodynamical Turbulence with Radiation Transport in Accretion disks, stratified Local Shearing Box, Movie: 6 Orbits Finite Volume Methode (Grid-Resolution 64×128×512) Resources on hpc-bw (BW-GRID): * 128 Cores * 5 MB Memory per Core * 100 GB Storage sapce * 125.000 CPU hours * Runtime: 40 days (M. Flaig) W. Kley Stuttgart: 20. Mai, 2010 17 Turbulent state Accretion disk Here non-zero vertical magnetic field, Bz Magnetic field strength (Emag /p0 vs. Time) α = hTT + TM i/p0 TR = ρδvr δvφ TM = −δBr δBφ W. Kley Angular momentum transport (α vs. Time) (∝ Trφ efficiency of angular momentum transport) (Reynolds Stress) (Magnetic Stress) Stuttgart: 20. Mai, 2010 18 Planet Formation Two main scenarios Gravitational-Instability (top-down) Sequential Accretion (bottom-up) (L. Mayer) (NASA, U2) Density-Fluctuations grow Fast Formation (103 years) Require: Fast Cooling large disk mass From small to large particles Slow Formation (106 Years) Need: High Sticking probability (small timescales, distant planets) W. Kley (Comets, asteroids, solid planets, cores of planets) Stuttgart: 20. Mai, 2010 19 Planet Formation I. Gravitational-Instability Consider local density perturbation in disk Analytical Numerical Stability-Criterion (Toomre) Evolution of an isothermal Disk • Finite-Difference Hydrodynamics • Viscous Disk csκ0 >1 Q≡ πGΣ0 with cs = sound velocity κ0 = Epicyclic-Frequency (ΩK ) Σ0 = surface density • Pressure & Rotation stabilize • Density destabilizes (Tobias Müller) Disk heats up upon compression, need fast cooling Require: coolingtime ≈ period Only in large distances from star W. Kley ⇒ fast formation (for ca. 30-50 AU, no cores) Stuttgart: 20. Mai, 2010 20 Particle dynamics in disk Planet Formation ( Jürgen Blum (Braunschweig)) Particles have relative velocity with respect to the gas ⇒ Forces Problem I: Fast radial drift towards star (for 1m Size: 1 AU / 100 Years) Problem II: Destructive Collisions W. Kley Stuttgart: 20. Mai, 2010 21 Planet Formation Planetesimals I Growth of planetesimals by collisions/accretion SPH (Smoothed-Particle-Hydrodynamics) Here: elastic-plastic strength model, formation and evolution of cracks 2 Basalt Spheres: ρ = 3g/cm3, R1=9m, R2=7.5m, Vrel=25m/s ∆ t = 30ms, tmax = 4.5 sec (Christoph Schäfer, PHD-Thesis, Univ. of Tübingen 2005) W. Kley Stuttgart: 20. Mai, 2010 22 Planet Formation Planetesimals II Solution: study porous particles Particle method Smoothed Particle Hydrodynamics (250,000 Particles) Calibration with real experiments (TU Braunschweig) Fragmentation (Test-Case) Vrel = 10m/s (R. Geretshauser) Vrel = 1m/s Computer time on hpc-bw: 1-3 days on 24-32 cores (240k-480k Particles) =⇒ DFG-Forschergruppe: The Formation of Planets (2007-2012) W. Kley Stuttgart: 20. Mai, 2010 23 Migration Planet Formation Evolution of planet in Disk (here viscous disk parametrization by α ) Require: Torques (disk pulls on planet) =⇒ Migration =⇒ motion in disk Accretion (planet attracts material) =⇒ Mass growth rate On the right: Saturn mass planet in disk - Spiral arms - Gap formation (A. Crida) W. Kley Stuttgart: 20. Mai, 2010 24 Modelling A test project Actors Star (1 M), Disk (0.01 M), Planet (Mp = MJup) Disk: Hydrodynamical Evolution 20 - In potential of Star & Planet - 2D (r, ϕ or x − y) - rmin = 2.08, rmax = 10.4 AU - locally isothermal: T ∝ r−1 - const. viscosity: ν = const. Planet: At 5.2 AU - here on fixed orbit - point mass (smoothed potential) Compare 17 Codes in De Val-Borro & 22 Co-authors, MN (2006) (Grid-Sample: 128x128) 10 0 -10 -20 -20 -10 0 10 20 EU comparison project: http://www.astro.su.se/groups/planets/comparison/ W. Kley Stuttgart: 20. Mai, 2010 25 Modelling Equations-2D: Conservative Mass Conservation ∂Σ + ∇ · (Σu) = 0 ∂t u = (ur , uϕ) = (v, rω) Radial Momentum ∂p ∂Φ ∂(Σv) 2 + ∇ · (Σvu) = Σ r(ω + Ω) − −Σ + fr ∂t ∂r ∂r Angular Momentum ∂Φ ∂[Σr2(ω + Ω)] ∂p 2 + ∇ · [Σr (ω + Ω)u] = − −Σ + fϕ ∂t ∂ϕ ∂ϕ Energy H/r = const. =⇒ T ∝ 1/r W. Kley Stuttgart: 20. Mai, 2010 p = Σ T /µ 26 Modelling Typical Result Viscous disk: Mp = 1 MJup, ap = 5.2 AU (WK, 2002) W. Kley Stuttgart: 20. Mai, 2010 27 Modelling EU comparison: Results Types: solid - Upwind, HighOrder, short-dashed - Riemann, long-dashed SPH; (Cyl. vs. Cartesian) W. Kley Stuttgart: 20. Mai, 2010 28 Modelling Coriolis Terms I Angular Momentum Equation Conserved (as before) ∂[Σr2(ω + Ω)] + ∇ · [Σr2(ω + Ω)u] = ∂t ∂p ∂Φ − −Σ + fϕ ∂ϕ ∂ϕ Not-Conserved (Explicit) ∂[Σr2ω] + ∇ · [Σr2ωu] = −2 Σ vΩ r ∂t ∂p ∂Φ − −Σ + fϕ ∂ϕ ∂ϕ W. Kley Stuttgart: 20. Mai, 2010 29 Modelling Coriolis Terms II Conservative Non-Conservative Conservative results: identical to inertial frame =⇒ If rotating frame: Use conservative formulation W. Kley Stuttgart: 20. Mai, 2010 30 Modelling FARGO - Basics A fast eulerian transport algorithm for differentially rotating disks (Masset, Astronomy & Astrophysics 2000) Code-page: http://fargo.in2p3.fr/ The problem: - Cylindrical coord. system - explicit code → Courant condition: ∆ϕ ∆t < ω with ω ∝ r−3/2 → ∆t ∝ r3/2 The solution: Calculate average velocity ω̄i of ring i, and perform Shift with ni = Nint(ω̄i∆t/∆ϕ) Transport on with remaining velocity =⇒ much larger ∆t Take care: no ring separation W. Kley Stuttgart: 20. Mai, 2010 31 Modelling FARGO - Results Azimuthally averaged surface density 1.4 Fargo vs. Standard inertial vs. rotating 1.2 Jupiter mass planet after 25 Orbits Used Timesteps – 51000 – 39000 – 5800 – 5800 Σ 1 t=25 orbits r−phi (128x384) inertial, std. rotating, std. inertial, Fargo rotating, Fargo .8 .6 .5 W. Kley 1 1.5 r 2 Stuttgart: 20. Mai, 2010 =⇒ Identical results ! =⇒ Huge Savings ! larger savings for log-grid 2.5 http://fargo.in2p3.fr/ 32 Modelling The Grid Azimuthally averaged surface density 1.4 Cartesian vs. Cylindrical (inertial frame) Jupiter mass planet after 25 Orbits 1.2 Models – Cyl.: 128×384 – Cart.: 320×320 – Cart.: 640×640 Σ 1 .8 t=25 orbits q01, cyl, std q04, cart, 320 q05, cart, double .6 .5 W. Kley 1 1.5 r 2 Stuttgart: 20. Mai, 2010 =⇒ Cartesian problematic ! – no operator-splitting – 2nd order RK-integrator [only PENCIL code reasonable (HiOrder)] 2.5 33 Modelling Nested Grid I Nested-Grid System: Centered on Planet - corotating frame (D’Angelo et al. 2002/03) l=1 l=3 W. Kley l=2 Stuttgart: 20. Mai, 2010 34 Modelling Nested Grid II 1 MJ (here 2D, ϕ − r plots) (D’Angelo et al., 2002, PhD Thesis Tübingen, 2004) W. Kley Stuttgart: 20. Mai, 2010 35 Torque calculation Modelling Sum over all grid-cells, P Tz = cells (xp Fy Which parts belong to planet ? − ypFx) Migration of Saturn mass planet Mp = 3.10-4M* ; Nr = 460 ; Ns = 1572 SEMI - MAJOR - AXIS 1 Use Truncation Radius Rtrunc ≈ 0.5 − 1.0 Rroche Smooth tapering-function fb fb(s) 1 b=0.2 0.5 b=0.4 b=0.6 0.95 0.9 0.85 Method : (A) (B) b = 1. (B) b = 0.9 (B) b = 0.8 (B) b = 0.7 (B) b = 0.6 (B) b = 0.5 (B) b = 0.4 (B) b = 0.3 (B) b=0. <=> none 0.8 300 b=0.8 b = 1. 305 310 315 TIME [orbit] 320 325 (A. Crida) 0 0 W. Kley 0.2 0.4 0.6 0.8 s / rHill 1 1.2 1.4 – (A) fully self-gravitating =⇒ need b ≈ 0.5 − 0.6 Stuttgart: 20. Mai, 2010 36 Disk physics Planet Formation ∂Σcv T + ∇ · (Σcv T u) = −p∇ · u +D − Q −2H∇ · F~ ∂t 4 Adiabatic: Pressure Work, Viscous heating (D), radiative cooling (Q) (∝ Tef f) radiative Diffusion: in disk plane (realistic opacities) 4e-05 local cool/heat Specific Torque [(a2 Ω2)Jup] 3e-05 2e-05 fully radiative Mp = 20Mearth (Kley&Crida ’08) 1e-05 0 adiabatic -1e-05 Disk Physics determines Direction of motion -2e-05 isothermal -3e-05 -4e-05 0 100 200 300 Time [Orbits] W. Kley Stuttgart: 20. Mai, 2010 37 Planet Formation Migration (Type I) Isothermal and radiative Models: Migration inward or outward possible Vary Mp (Kley&Crida ’08) W. Kley Stuttgart: 20. Mai, 2010 38 Planet Formation Summary/Outlook New properties of extrasolar Planets • High Masses • Large eccentricities • Close distances Planet-disk simulations • Coord. System • Angular momentum conservation • FARGO-Accelaration • Grid-Refinement • Thermodynamics Accurate accretion & migration history of planets requires • Full 3D simulations • High resolution (locally) • Radiation transport • Self-Gravity • Magnetic Fields W. Kley Stuttgart: 20. Mai, 2010 39 Planet Formation The End Thank you for your attention ! (A. Crida) W. Kley Stuttgart: 20. Mai, 2010 40