Development of Adaptive Kinetic-Fluid Computational Tools for Low-Temperature Plasmas Vladimir Kolobov Quantemol Workshop April 4, 2014 Agenda • Introduction: Plasma Technologies & Physical Models • Commercial Software for Simulations of Low Temperature Plasma Devices and Processes • Adaptive Mesh and Algorithm Refinement (AMAR) and Unified Flow Solver • Towards Multi-Physics AMAR: Fluid Plasma Models with Adaptive Cartesian Mesh • Kinetic Solvers with Adaptive Mesh in Phase Space • Conclusions • Acknowledgements • References Plasma Technologies •Thrusters "The XXI century will be not only the century of informatics, biology and space exploration, but also the century of plasma technologies" A.I.Morozov 2006 Propulsion Lighting •Spray Coatings Materials Processing Displays Kinetic and Hydrodynamic Approach Kinetic Hydrodynamic Particles are described by five characteristics: 1. Density n ( r, t ) 2. Mean directed velocity, 3. Temperature, T ( r, t ) v ( r,t ) They depend on 4 scalar arguments – 3 spatial coordinates and time. The only characteristic is the velocity distribution function (VDF) – the particle density in phase space f ( ξ, r, t ) It depends on 7 scalar arguments – 3 spatial coordinates, 3 velocity components and time. • Kinetic description is much more detailed and much more expensive • Many problems in gas discharge physics can only be understood at kinetic level Commercial Software for Simulations of Plasma Devices and Processes Commercial Software Development CFDRC developed the first commercial tool for simulation of industrial plasma reactors and processes In 2004, CFD-ACE+ software package was sold to ESI group for further commercialization CFDRC retains rights for software customization and consulting CFD-ACE+ Software Architecture Plasma Models Spatial gradients can be so steep that a continuum description may break down even at pressures of many atmospheres High Pressure Thermal Plasmas • Non – Equilibrium Chemistry • Two Temperature Te≥ T • Te = T, LTE • Gas Flow (turbulence) • Heat Transfer • Radiation (HF EMAG) Key Features • Radiation Transport • Gas Heating T > Troom Low Pressure Non-Equilibrium Plasmas • Magnetic Field Effects • Kinetics of Ions & Fast Neutra • Kinetics of Electrons • Non-Local Electrodynamics The need to develop new modeling techniques that address vastly different scales having different physics at both low and high pressures speaks to the convergence of the discipline Plasma Simulations with CFD-ACE+ 2D Cartesian or Cylindrical 3D Cartesian 1D Positive column 0D Well stirred Global Plasma Model Non - Equilibrium Flow Heat Chemistry Magnetic Local Thermodynamic Equilibrium (LTE) Plasma Electric ICP Flow Kinetic CCP PC Heat Electric Global Radiation Magnetic Turbulence User Subroutine Reaction Manager Arrh. Rate Cross-Sections Maxw EEDF Non-Maxw EEDF Dashed lines indicate an optional or alternative choice Plasma Aided Manufacturing Electron Kinetics in ICP Calculated electron temperature (left) and electrostatic potential (right) for Ar ICP at 10 mT, 6.8 MHz, 100W. ∂f 0 v 1 ∂ 2 eE + div(f1 ) + 2 ⋅ f1 = S0 v ∂t 3 3v ∂v m EEPFs at different points along the discharge axis at radial position r=4 cm: a) measured, b) calculated. ∂f1 eE ∂f 0 +ν f1 =− v∇f 0 − m ∂v ∂t Two-term SHE reduces 6D Boltzmann equation to a 4D Fokker-Planck equation for electrons in v f (r= , v.t ) f 0 (r, v,t ) + ⋅ f1 (r, v, t ) collisional plasmas ∂f 0 ∂φ ∂f 0 v − − ∇ ⋅ Dr ∇f 0 − ∂t ∂t ∂ε Using total energy (kinetic + potential) simplifies analysis of 1 ∂ ∂f χ Dε (r, ε ) 0 + Vε (r, ε ) f 0 = S electron kinetics for many gas discharge problems χ ∂ε ∂ε V. Kolobov & R. Arslanbekov, Simulation of Electron Kinetics in Gas Discharges, IEEE TPS 34 (2006), 895 Ionization Waves - Striations Argon 2 Torr 100 mA L = 20 cm R = 1 cm Argon 1 Torr Atmospheric Pressure Discharges Photo by M. Laroussi CFD-ACE+ simulations of atmospheric pressure Nitrogen discharge between parallel plates covered by dielectrics at driving frequency 10 KHz. Electrostatic potential Pattern Formation in DBD CFD-ACE+ simulations: fluid model with LUTs Bhoj & Kolobov, IEEE TPS 39,(2011) 2152 Experimental measurements of pattern formation dynamics in He DBD 200kHz, 30kPa, 690 V. Stollenwerk et al. Phys. Rev. Letters 96 (2006) 255001 • Base case conditions:1000 V, 200 kHz, perfect dielectric with permittivity ε =7.6. • Patterns disappear when driving frequency increased to 300 kHz • Increasing ε increased peak electron density and pattern density Plasma Solvers with “Traditional” Mesh • Hybrid models have been widely used for simulations of weakly ionized plasmas using traditional body-fitted mesh techniques • We have performed simulations of numerous plasma reactors and processes for semiconductor manufacturing, lighting and other applications, and accumulated considerable experience in developing chemical reaction mechanisms: • rare gases (lighting, PDP) • O2, N2, Cl2, H2 • Fluorocarbon plasmas (SF6, C2F6, ..) • SiH4 for deposition of SiO2 • Hydrocarbon plasmas (DLC films, CNT, etc) An example of 3D fluid simulations of an industrial ICP source performed with CFDACE+ software using body-fitted mesh techniques • Traditional methods lack efficiency for simulations of dynamically evolving features requiring high resolution in selected areas of computational domain (shock waves, ionization fronts, streamers, etc). Adaptive Mesh & Algorithm Refinement and Unified Flow Solver Third Wave of CFD Evolution of Mesh Types Structured body-fitted Hybrid body-fitted Unstructured body-fitted Cartesian with embedded boundary Embedding CFD inside an MCAD tool, Mesh generation becomes a part of the solution process (Mentor Graphics) Unified Flow Solver (UFS) Developed by CFDRC for U.S. Air Force Supported by NASA, MDA, and U.S. Navy Direct Numerical Solution Boltzmann Solvers Domain Decomposition Direct Simulation Monte-Carlo (DSMC) Next Generation Computational Tool for High-Altitude Aerothermodynamics and Low-Speed Micro-Flows Adaptive Mesh & Algorithm Refinement Coupling Algorithm Continuum Flow Solvers • Direct Boltzmann Solver by Discrete Velocity Method • Dynamic adaptation of computational (Cartesian) mesh to solution and geometry • Auto switching between kinetic and fluid models based on continuum breakdown criteria V. Kolobov et al, J. Comp. Phys 223, (2007) 589 Illustration of UFS methodology Steady Gas Flow Around Cylinder for different Kn numbers: M=3 Kn=0.01 Kn=0.1 Kn=0.3 Kn=1 Kinetic (Red) and Fluid (Blue) Domains Mesh Refinement Criterion: ∇ρ < ε = 0.1 Density Profile Continuum Breakdown Criterion 2 2 2 ∇p 1 ∂u ∂υ + + Kn 0.1 <δ = p T ∂x ∂y UFS Simulations of Transient Flows Computational grid Shock Wave Interaction with Boundary Layer in Micro Channel Kinetic & Fluid Domains Ma = 3 Kn = 0.2 Gas Density V. I. Kolobov, et al, Unified Flow Solver for Transient Rarefied-Continuum Flows, AIP Conf. Proc.1501 (2012) 414; V. V. Aristov, et al, Vacuum 86 (2012) Space Filling Curves & Forest of Octrees The procedure of domain decomposition is performed using SFC. Using SFC, 2D or 3D spatial data can be stored in a single dimensional array and domain decomposition is easily achieved by chopping this array into even pieces. • During sequential traversing of cells by natural order, the physical space is filled with curves in N-order (Morton ordering). After this ordering of cells, all cells can be considered as a onedimensional array. • A weight is assigned to each cell, proportional to CPU time required for computations in this cell. The array modified with corresponding weights, is subdivided into sub-arrays equal to the number of processors. • This method allows most efficient domain decomposition between different processors. Towards Multi-Physics AMAR Kinetic Module: • DVM Boltzmann & Vlasov Solvers • Fokker Planck solver for electrons • Lagrangian Particle Solvers (DSMC, Photon Monte Carlo) Mesoscopic: • Lattice Boltzmann Method Fluid: • Plasma & Chemistry Hybrid Model of Radiation Transport: selection of PMC or diffusion solver base on local photon MFP Fluid Plasma Models with Adaptive Cartesian Mesh Minimal Plasma Model A simplest set of equations for the electron and ion densities, and Poisson equation for electrostatic potential. Electron diffusion and ion transport are neglected ∂ne + div Γ e = ν ne ∂t ( ) Γ e =µ Ene ∂ni = ν ne E ∂t = ν ν 0 exp − 0 ∆ = ϕ 4π ( ne − ni ) Mesh Adaptation criteria ne ne = + α 20 + 3 ( log10 (ne ) + log10 (ni ) + log10 (ν i ) + log10 ( E ) ) max( ) max( ) n n e e When gradient of α exceeded a fixed value this particular cell was refined ∇α > 1 in a cell, When ∇α < 1/ 4 in a cell, this cell was coarsened The procedure of mesh adaptation performed every 10 time steps E Streamer development in positive corona 2D axisymmetric simulations 3D parallel simulations on 8 cores with ~1.5 million cells: axi-symmetric channels break into separate streamers. electron density Computational grid, electron density contours at 107 cm -3, and the electric field strength (color, maximum value 7 106 V/m ) Extended Plasma Model • A set of equations for the electron and ion densities, and Poisson equation for electrostatic potential • Ion transport is included • Multiple species ions are allowed by introducing species specific mass, diffusion coefficients, and ionization rates • Local field approximation for ionization rate • Electron Energy Transport with account for electron thermal conductivity • CANTERA link implementation is underway for complex reaction mechanisms ∂ne + div Γ e = ν ne ∂t ( ) Γ e =− µe Ene − De∇ne ∆ = ϕ 4π ( ne − ni ) ∂ni + div(Γi ) = ν ne ∂t = Γi µi Eni − Di ∇ni E = ν ν 0 exp − 0 E ∂nε + ∇ ⋅ Γ ε = Sε ∂t nε = neε 5 5 Γε =− µe Enε − De∇nε 3 3 Effect of Electron Energy Transport 2D axi-symmetric simulations The model reproduces basic structure of classical DC glow discharges Spatial distribution of electron density Axial distributions of electron and ion densities (left), electrostatic potential and electron temperature (right) with account of electron thermal conductivity Kinetic Solvers with Adaptive Mesh in Phase Space Adaptive Mesh in Phase Space spatial mesh VDF Tree-of-Trees Structure: Tree-based velocity mesh created in each spatial cell. Hypersonic flow over a square at M=30,Kn=0.1: VDFs at different locations velocity mesh Electron Kinetics electron diffusion in energy thermal electron runaway • Kinetic solvers for electrons in different regimes of EDF formation: • Fokker Planck (collisional, slow) • Vlasov (nearly collisionless regimes) • Boltzmann (fast runaway, anisotropic) genuine electron runaway Different regimes of EDF formation: Kolobov, Phys. Plasmas 20, 101610 (2013) Computational mesh & EDF contours in velocity space a : electron streaming; b : isotropic “body”, anisotropic “tail” Adapted mesh in velocity space Linking Collision Theory & Physical Kinetics Computational mesh and VDF contours for different values of electric field strength, elastic and inelastic collisions. White circle corresponds to inelastic threshold. In semiconductors, phased motion of hot electrons in configuration and momentum spaces causes spatial modulation of electron concentration and mean electron velocity with spatial period λE = ε 0 / eE In gas discharges, similar phenomena result in plasma stratification and propagation of ionization waves (moving striations). Precise calculations of EDF require detailed knowledge of differential collision cross sections for elastic and inelastic processes Conclusions • Commercial software for non-equilibrium low temperature plasmas (CFD-ACE+, 10+ years old) • Third Wave of CFD: embedding CFD inside MCAD tools, mesh generation is part of the solution process, AMR • Unified Flow Solver (4th Wave of CFD ?): • AMAR methodology (AMR + Adaptive Physics) • massively parallel, GPU • AMAR extensions to Multi-Physics: • Hybrid Model of Radiation transport • Hybrid Plasma Models • Electron Kinetics • Kinetic Solvers with Adaptive Mesh in Phase Space • Linking Scattering Theory & Physical Kinetics Acknowledgements Collaborators and co-authors: • Dr. Robert Arslanbekov (CFDRC) • Dr. Anna Frolova (RAS, Moscow, Russia) • Dr. Vladimir Aristov (RAS, Moscow, Russia) • Dr. Sergey Zabelok (RAS, Moscow, Russia) • Dr. Felix Tcheremissine (RAS, Moscow, Russia) • Mr. Eswar Josyula (AFRL) • Dr. Jonathan Burt (AFRL) • Dr. Ranjan Mehta (CFDRC) Financial support provided by DoD and NASA through SBIR/STTR Projects References • V.I. Kolobov, "Fokker Planck modeling of electron kinetics in plasmas and v semiconductors", Computational Materials Science 28, 302 (2003). • V.I. Kolobov and R.R. Arslanbekov, "Simulation of Electron Kinetics in Gas Discharges", IEEE Trans. Plasma. Sci. 34, 895 (2006) • V.I.Kolobov, R.R.Arslanbekov, V.V.Aristov, A.A.Frolova, S.A.Zabelok, “Unified solver for rarefied and continuum flows with adaptive mesh and algorithm refinement”, J. Comput. Phys. 223 (2007) 589 • V.I.Kolobov and R.R.Arslanbekov, “Towards Adaptive Kinetic-Fluid Simulations of Weakly Ionized Plasmas”, J. Comput. Phys. 231 (2012) 839 • R.R.Arslanbekov, V.I.Kolobov, and A.A.Frolova, “Kinetic Solvers with Adaptive Mesh in Phase Space”, Phys. Rev. E 88 (2013) 063301 • V. I. Kolobov, “Advances in Electron Kinetics and Theory of Gas Discharges”, Phys. Plasmas 20, 101610 (2013)