Astrophysical Combustion Elaine S. Oran Laboratory for Computational Physics US Naval Research Laboratory 30th International Combustion Symposium Chicago, July, 2004 SN 1994D in NGC 452626 Acknowledgments Vadim N. Gamezo Alexei M. Khokhlov J. Craig Wheeler Geraint Thomas Naval Research Laboratory University of Chicago University of Texas, Austin University of Wales, Aberystwyth Michael Frenklach Rowena Ball K. Kailasanath University of California, Berkeley Australian National University Naval Research Laboratory Sponsors: NASA and ONR D. Huw Edwards University of Wales Aberystwyth "... among the giants of gasdynamics worldwide..." "... an innovative experimentalist with a deep understanding of theory..." (1925 ?? - 2003) Combustion Since the founding of the Combustion Institute, what is considered "legitimate" combustion has become more and more inclusive: * Oxidation with energy release (emphasis on chemistry) * Gases with exothermic chemical reactions * Gas dynamics with exothermic chemical reactions * Physics of flames - laminar and then turbulent * Detonations, gaseous and condensed; propellants * Often involving complex chemical reactions, multiphase flow and reactions, that may produce pollutants, soot, diamonds, fullerenes, nanoparticles ... Now .... Combustion ... has come to mean: The Result of Fluid Dynamics Combined with Exothermic Reactions, and Everything This Implies ... A typical Symposium has representation from Physics Chemistry Biology Computer Science Mechanical Engineering Aerospace & Propulsion Chemical Engineering Electrical Engineering Bioengineering Environmental Sciences Math, Applied Math Materials Science Atmospheric Sciences Other (stray astrophysicist?) The field has broadened, become more inclusive and dynamic ! Outline Evolving definition of combustion Astrophysical combustion - Taking it to the limit Thermonuclear combustion Nucleosynthesis, Nuclear kinetics Birth and death of stars - Stellar evolution White dwarfs Type Ia supernovae Turbulent flames, detonations, and DDT Astrophysical-terrestrial combustion analogies Where is Exothermicity in Astrophysics? Thermonuclear Reactions - Big Bang - Nuclear reactions in an expanding background Forms H, He, Li Stars (Sun) - Nuclear fusion reactions Burns H to He, He to C, O, etc. Shell burning in supergiants; novae in white dwarfs Supernovae - nuclear reactions to form higher Z elements Chemical Reactions - Interstellar Combustion - Dilute chemical reactions Particle, soot formation. Multiphase flow Relatively slow, exothermic Quark-Baryon Transition - Exothermic? Elementary particle reactions In very early universe. An explosive transition? Interstellar Dust and Particulates (Table courtesy of M. Frenklach and A. Tielens) Example Observed Location Red Giants1 ISM2 Meteorites (June, 2004) Formation Mechanism Mineral Carbides 4 8 4 Nucleation from gas phase, T ~ 1000 - 3000 K Aliphatics 4 4 4 Surface and gas-phase reactions Kerogens 8 ?? 4 Neutral reactions at T ~ 1000 K Aromatics 8 4 4 Cyclization by acetylene additions at T ~ 1000 K Amorphous Carbon ?? ?? ?? Deposition onto seed particles at T > 1000 K Diamonds 8 4 4 Excited or ionized reactions (SiC, TiC) (polyyne chains) (coals) ( PAHs,5 fullerenes soot, graphite) (DLC,3 HAC,4 glassy C 1. Cicumstellar envelopes around carbon-rich giants 2. Interstellar Medium (ISM) 3. Diamond-like carbon (DLC) 4. Hydrogenated amorphous carbon (HAC) 5. Polycyclic aromatic hydrocarbon (PAH) Interstellar Dust and Particulates (June, 2004) Where is Exothermicity in Astrophysics? Thermonuclear Reactions - Big Bang - Nuclear reactions in an expanding background Forms H, He, Li Stars (Sun) - Nuclear fusion reactions Burns H to He, He to C, O, etc. Shell burning in supergiants; novae in white dwarfs Supernovae - nuclear reactions to form higher Z elements Chemical Reactions - Interstellar Combustion - Dilute chemical reactions Particle, soot formation. Multiphase flow Relatively slow, exothermic Quark-Baryon Transition - Exothermic? Elementary particle reactions In very early universe. An explosive transition? Nucleosynthesis ... the reactions by which atomic nuclei are transformed to produce the observed abundances of elements Thermonuclear Supernovae Stars, Core-Collapse SN H, He, some Li made in Big Bang Heavier than Li, made in stars All stars burn H to He Massive stars burn He to C, O, Si, S, Fe Big Bang Above Fe, elements made in the outer envelopes of supergiants and in supernovae Nuclear Kinetics Over 100 species, Over 1000 reactions Cross sections found from: (1) Nuclear-reactor measurements or ion-beam experiments, sometimes extrapolating to the right temperature range (2) Theoretical estimates based on (1) Number of Protons (Z) 100 Z=N 80 60 40 Very Unstable 20 0 0 20 40 60 80 100 120 140 Number of Neutrons (N = A - Z) 160 Birth and Death of Stars (1) Molecular Clouds Eagle Nebula -M16 HST 1995 (Hester and Scowen) > Main Sequence Stars Hertzsprung-Russell Diagram Birth and Death of Stars Molecular Clouds < 0.08 M Clouds ~ > Main Sequence Stars with mass M0 > Brown Dwarf (these never ignite) <8 M M0 ~ > Red Giant >8M M0 ~ > Supergiant { > > Planetary Nebula > White Dwarf > Thermonuclear Supernova > Core-Collapse Supernova > Neutron Star or Black Hole > 200 M Clouds ~ > ??? pulsate and blow apart ? Birth and Death of Stars (3) Molecular Clouds <8M M0 ~ > Main Sequence Stars > Supergiants > > Core-Collapse Supernovae Betelguese Neutron Star or Black Hole 1987A in Large Magellanic Cloud (Progenitor ~ 20 solar masses) Eta Carina NGC 3372 Birth and Death of Stars (2) Molecular Clouds > Main Sequence Stars Arcturus <8M M0 ~ > Red Giants > > White Dwarfs Novae > Planetary Nebulae V2487 Oph (1998) > HST, in M4 Thermonulcear Supernovae HST, NGC 6543, Cat's Eye (~1000 years old) SN 1994D in NGC 4536 White Dwarfs f f f f f Burned out stars Mostly C and O Radiate heat, eventually die Mass approximately that of Sun Size approximately that of Earth f Many (>50%) are part of binary systems f Accrete matter from their neighbor f If they accrete just a small amount of H or He from neighbor, this can burn on the outside of the star and create flashes called novae. f If enough matter is accreted, so M0 approaches 1.4 M , density in the center increases, and star ignites. Type Ia Supernovae These do not just cook quietly, but burn explosively. Before ...... NGC 4536 After ..... 1981B Lifetime of WD: 108-1010 yrs Explodes in 1 - 2 seconds Bright as an entire galaxy .... Releases 1051 ergs (1027 Mton) Forms heavy elements ... 12C + 16O heavier elements Y SN 1994D in NGC 4526 Type Ia Supernovae Virgo Cluster SN 1994D Significance of SNIae * Create and disperse higher Z elements important to our way of life * Provide "standard candles" for cosmological measurements * Laboratory for studying basic combustion processes Nuclear Kinetics Leading Reaction (~50% energy release) Over 100 species Over 1000 reactions ∆ Unstable nuclides 40 Z + Very unstable = N 30 Ni Co Fe 20 Si 10 0 Triple alpha reaction (<10%, NSE relax) 12C 3 4He 3 O C 0 10 20 .... Stable nuclides .... Atomic Number (Z) 50 ] 20Ne + 4He 12C + 12C ] 23Na + p 12C + 12C ] 23Mg + n Pair Reactions (20-40%,quasiequil) p + (Z,N) ] (Z+1, N) n + (Z,N) ] (Z, N+1) 3He + (Z,N) ] (Z+2, N+2) 4He + (Z,N) ] (Z+1, N+2) + p 12C + 12C 30 Neutrons (N) 40 50 ] Electron capture, beta decay 56Ni 56Co 56Fe Y Y Significance of SNIae * Create and disperse higher Z elements important to our way of life * Provide "standard candles" for cosmological measurements * Laboratory for studying basic combustion processes Cosmological Significance of SNIae The Best Standard Candles Relative Brightness (magnitude) -20 Shape of light curve (brightness vs time) is correlated to the peak luminosity. -19 -18 -17 Bright SNIa decline more slowly. -16 -15 Scaling determined by Phillips or luminosity width relation (LWR). Scaled Magnitude -20 -19 -18 -17 -16 -15 -20 0 20 Time (days) 40 60 Calan/Tololo Survey of SNIa (Hamuy et al., 1996, Kim et al., 1997) Candidate Standard Candles Measure Velocity of Galaxies - not so hard ... Calculate from the observed spectral redshift Measuring Distances of Galaxies - so hard ... Trigonometric parallax for nearby stars - Use earth orbit as baseline to triangulate a star's distance Cepheid variables - Good to ~100 Mparsecs Stars with regularly pulsating atmospheres Relation between luminosity and pulsation period Elliptical and spiral galaxies - Good to ~200 Mparsecs Spiral - rotational velocity correlated to luminosity Eliptical - orbital velocity correlated to luminosity Type II Supernovae - Good to ~300 Mparsecs Not really a standard candle, but use (e.g.) spectra of their expanding atmospheres Type Ia Supernovae - Good to ~Gparsecs* Peak luminosity correlated to shape of light curve * 1 Gparsec ~ 3.2 x 109 Lt yrs ~ 3 x 1022 km Cosmological Constraints Friedmann Equation: Hubble parameter mass density curvature of the universe vacuum energy density Implies a flat universe, the current idea Use SNIae to determine H and 4% is regular matter; rest is dark matter Significance of SNIae * Create and disperse higher Z elements important to our way of life * Provide "standard candles" for cosmological measurements * Laboratory for studying basic combustion processes Combustion in SNIa - What do we know? (1) Initial Conditions T ~ 105 - 108 K M ~ 2.8 x 1033 g (1.4 M ) ρc ~ 2 x 109 g/cc, radius ~2 x 108 cm = 2000 km Binding energy ~ - 5 x 1050 ergs cs~ 5 x 108 cm/s ~ 1/60 c Explosion by a Thermal Instability Normal stars are thermally stable and do not explodethey expand as their temperature rises A degenerate WD does not expand when its temperature rises -> Instability Explosion takes 1-2 seconds Energy released ~1.5 x 1051 ergs Temperature reaches (8 - 9) x 109 K Produces ~ M of Fe, ~ 0.6 M of radioactive 56Ni, rest is Si-group Velocities up to ~20,000-30,000 km/s in outer layers > Combustion in SNIa - What do we know? (2) The Physical Model Highly exothermic nuclear process Binary reactions, no chain branching, typical rate is R(T) ~ ρ2 exp(-Q/T1/3) or R ~ ρ2 exp(-Q/T), Q/T ~ 10 - 50 (similar to hydrocarbon fuels) T ~ 109 - 1010 K, ρ ~ 106 - 1010 g/cc Among all nuclei from C to Ni, protons, neutrons, He Heat conduction- very high: degenerate e-'s, photons Viscosity - small - due to scattering e-'s, photons Species diffusion - none (Lewis number = 0) Equation of state - known to very high accuracy Ideal gas of Fermi-Dirac e-'s and e+'s, Boltzmann gas of ions, Planck radiation P = E / (γ - 1), γ ~ 4/3 to 5/3 These are the "ingredients" of flames and detonations! Thermonuclear Detonation Detonation speed: 1 - 1.4 x 109 cm/s Detonation front thickness: 10-1 - 103 cm Three fairly distinct burning regimes: Burn C (50% of energy released), then form Si (10-40% energy released), then form Fe Structure is extremely sensitive to density! The regimes are compressed at high density. Detonation leaves behind a highly inhomogeneous mixture of intermediate products of reaction and a complex multilevel detonation structure (Gamezo et al., 1999) Detonation cells form over a wide range of scales, depending on the density and the burning regime. Important Scales in White-Dwarf Deflagrations Laminar flame-front thickness: 10-4 - 100 cm Laminar flame speed: 107 - 104 cm/s (<< sound speed) Flame is Rayleigh-Taylor unstable in gravitational field Require a subgrid model to account for flame stretching and acceleration in cascade regime and below. Calculations show that St does not depend on Sl, but is determined primarily by the Rayleigh-Taylor instability: St = α (gLA)1/2, α = 0.5, g - graviational acceleration L - length scale A - Atwood number Numerical Approach to Supernova Simulations Solve unsteady, compressible fluid dynamics equations with: Equation of state for a degenerate electron gas Reduced kinetic mechanism for C+C burning, formation of Si and Fe. Tracking method for nuclear flame Fully Threaded Tree for adaptive mesh refinement Effective resolution 20483, 108 cells, ∆x = 2.6x105 cm Subgrid burning rate determined by gravity-related RayleighTaylor instability Model one octant of the star. Initial conditions: Hydrostatic equilibrium Chandresekhar-mass C-O white dwarf Spherical initial flame of radius 3x106 cm 3D SNIa Calculations .... Deflagration Stage of the Explosion Time (s) Mass Fraction Burned Flame Surface Gamezo, Khokhlov, Oran, Chtchelkanova, Rosenberg, Science, 2003 Converged Simulation of SNIa Explosion 0.576 s 1.257 1.393 1.487 1.531 s 1.573 1.613 1.652 1.724 s 1.760 1.883 1.902 The Hard Data on SNIa Spectra: From spectral evolution, we can determine velocities of material ejected, composition, and densities of the species, as a function of depth in the ejected matter. Light curves: From the rise and fall of these curves, we can constrain the amount of Ni 56 produced by the explosion, the values of density, and the expansion rate. WD environment: From the type of the host galaxy, location of the SNIa within the galaxy, and local environment in that galaxy, we can determine something about the age of the WDs. Nuclear Reaction Mechanism: Reasonable knowledge of nuclear reactions and nucleosynthesis. There are three main stages related to the burning of carbon, oxygen, and silicon. Roles and scales of each of these differ by orders of magnitude. Observations on Converged SNIa Simulations (Science, January 2003) This simulation is converged numerically ... Varying resolution does not change the global features of the results, e.g., the rate of expansion, final flame-surface area, ... The size of the lobes formed is roughly the same at all scales computed and for all converged resolutions. Varied initial condition, e.g., different C/O ratios. But the calculations do not produce the observed energy release and spectra. The results do not (cannot?) show DDT. Temperatures, densities in unreacted funnels are not high enough to allow detonation ignition. Approximations in model: Expanding flame - not computed from a complete enough model as we did for laboratory DDT computations. Solution ?? - Microphysics of this flame? Does Transition to Detonation Solve the Problem? If so, What is the Mechanism of DDT in SNIae? How does a high-speed turbulent flame suddenly become a supersonic reaction wave, a detonation? This is a difficult problem in terrestrial combustion!! Difficulty for experiments: Diagnostics of very high-speed, 3D reactive flows Difficulty for numerical simulations: Disparity of scales ... Must resolve the flame front structure and turbulent flame, in a large enough system to see the interaction and watch evolution. We can test whether DDT would solve the problem !!!!!!! Simulation of a Type Ia Supernova Delayed Detonation Simulation of a Type Ia Supernova Delayed Detonation Deflagration x1050 Etot (ergs) b Delayed Detonation a Deflagration Time (s) Delayed Detonation Where and how can DDT occur? How can detonation ignition centers (hot spots) form in a rapidly expanding flame ? Is the flame quenched before DDT occurs? The disparities between the time and space scales of the SN event and hot-spot formation are daunting. Gamezo, Khokhlov, & Oran, 2001, 2002 Can we use laboratory experiments and numerical simulations to study how this transition might occur? High-resolution Computation of a Hot-Spot Explosion µs Density 868.4 867.1 866.5 Explosion transitions to a detonation. 866.3 866.1 Hot spot explodes. 865.9 864.0 Time 860.7 859.0 857.6 Train of shocks runs through unreacted material. Flame brush contains regions of unreacted material. This is a Big Calculation ... Current calculation: 19,000 cpu hours SGI 3000; 48 or 128 processors Initial grid evenly spaced, with 2x107 smaller cells at center Use AMR, evolves to ~108 cells ∆xmin = 2.6x105 cm To calculate DDT, need to resolve the flame thickness, δf , with at least 10 computational cells. At ρ = 5 x 107 g/cm3, δf ~ 3.6x10-1cm Therefore, need ∆xmin = 2.6x10-2 cm This is 107 times smaller than current. Using AMR, Ncells increases as N2, not N3 ; ∆t 107 times smaller. Therefore, computer times increases by 107*3 = 1021~ 270. Moore's Law: computer power doubles in 1.5 yrs 70x1.5 = 105 years until we can do this calculation in equivalent time!! Astrophysical and Combustion Analogies Astrophysical Object Big Bang Combustion Analogue Explosion, He Formation Weak ignition? Supersonic combustion? Origin in space-time fluctuations? Deflagration, detonation or SHS Forms ordinary matter Supersonic combustion Main-Sequence Stars Furnace, thermonuclear reactor Red Giants Supergiants Multicomponent droplet burning, with microexplosions Multicomponent droplet burning White Dwarfs Decaying embers Novae Throwing a little fuel on embers Type Ia Supernovae Chemical plant explosion Type II Supernovae Structural collapse, shock propagation, ignition, explosion Smoldering combustion, soot formation Quark-Hadron Transition Interstellar Combustion (To be taken with a very large grain of salt) Several Concluding Thoughts ... Astrophysics as a subfield of Combustion? The entire universe and many of its most important astronomical components are (very difficult) combustion systems. An equivalently difficult combustion problem to many common astrophysical problems: Try to explain what has happened in a chemical-plant explosion from spectroscopic observations taken a hundred miles away, as observed through the properties of the observer's atmosphere. An astrophysical laboratory... There are naturally occurring combustion laboratories for testing models and theories over a very wide range of time and space scales. for testing combustion concepts ... Typical combustion phenomena, such as ignition, laminar and turbulent flames, detonations, multiphase flows, and particle and soot formation, are also fundamental processes occurring in astrophysics. under extreme conditions. The combustion concepts apply, but the conditions under which they apply are usually quite different and usually much more extreme in astrophysics.