Plasma Medicine in Vorpal Alexandre Likhanskii Tech-X Corporation Tech-X Workshop / ICOPS 2012, Edinburgh, UK 8-12 July, 2012 Motivation Fluid plasma models need artificial seed electrons to launch streamers J. Phys. D: Appl. Phys. 43 (2010) Can fluid model accurately resolve streamers? • Charged species number density – from 1016 m-3 to 1022 m-3 • Typical sheath size – 10 microns • Typical grid size for accurate resolution – 1 micron •Validity of Fluid approach – Maxwellian EEDF • Consider one 3D cell with 1 micron grid size • One electron per one cell -> 1018 m-3 • Is fluid approach valid for description of low density plasma phenomena at micron scales? • Is it possible to resolve 3D structure using fluid code? Kinetic effects can be captured using PIC approach: • Poisson or full Maxwell equations for electric field • Track motion of macroparticles (groups of charged particles) instead of considering number densities • MC collision model for all relevant plasma processes Advantage More accurate physics Disadvantage Slow speed for many particles VORPAL has a comprehensive PIC-DSMC plasma model • Poisson equation is solved using biconjugate gradient method with algebraic multigrid preconditioner (in Trilinos package) • Plasma model includes kinetic electrons, kinetic nitrogen and oxygen molecular ions, fluid neutral molecular nitrogen and oxygen • Several types of collisions: inelastic collisions, ionization, excitation, charge exchange, recombination, attachment • Serial/Parallel 2D/3D simulations Particles are pushed using standard FDTD algorithm e e e e e Area weighting preserves charge exactly Why are atmospheric pressure discharges so challenging for PIC codes? Exponential growth of number of particles due to avalanche ionization -> significant increase in computational time for PIC e e i e 1 2 e i e e i e 4 8 …… 1000 Why are atmospheric pressure discharges so challenging for PIC codes? e Small ND Need PIC e i e e i e e i e Consider different stages of discharge for one cell Moderate ND PIC -> Fluid transition Large ND PIC is not feasible Need Fluid code How does VORPAL handle the problem of exponential particle growth? • PIC code -> particles are represented via macroparticles • 1 macroparticle = N (nominal number) regular particles • Introduce weight W of macroparticle -> one macroparticle contains W*N regular particles How does VORPAL handle the problem of exponential particle growth? • PIC code -> particles are represented via macroparticles • 1 macroparticle = N (nominal number) regular particles • Introduce weight W of macroparticle -> one macroparticle contains W*N regular particles How does it work? How does VORPAL handle the problem of exponential particle growth? • PIC code -> particles are represented via macroparticles • 1 macroparticle = N (nominal number) regular particles • Introduce weight W of macroparticle -> one macroparticle contains W*N regular particles How does it work? Step 1: Have 6 macroparticles with W=1 each How does VORPAL handle the problem of exponential particle growth? • PIC code -> particles are represented via macroparticles • 1 macroparticle = N (nominal number) regular particles • Introduce weight W of macroparticle -> one macroparticle contains W*N regular particles How does it work? Step 1: Have 6 macroparticles with W=1 each Step 2: Combine pairs of particles How does VORPAL handle the problem of exponential particle growth? • PIC code -> particles are represented via macroparticles • 1 macroparticle = N (nominal number) regular particles • Introduce weight W of macroparticle -> one macroparticle contains W*N regular particles How does it work? Step 1: Have 6 macroparticles with W=1 each Step 2: Combine pairs of particles Step 3: End up with 3 macroparticles with W=2 each How does VORPAL handle the problem of exponential particle growth? • PIC code -> particles are represented via macroparticles • 1 macroparticle = N (nominal number) regular particles • Introduce weight W of macroparticle -> one macroparticle contains W*N regular particles What can be assigned? • Different sorting algorithms • Threshold number of macroparticles per cell for the combining • Maximum weight of macroparticles What happens during plasma decay stage? • PIC code -> particles are represented via macroparticles • 1 macroparticle = N (nominal number) regular particles • Introduce weight W of macroparticle -> one macroparticle contains W*N regular particles How does it work? Step 1: Start with 3 macroparticles with W=2 each What happens during plasma decay stage? • PIC code -> particles are represented via macroparticles • 1 macroparticle = N (nominal number) regular particles • Introduce weight W of macroparticle -> one macroparticle contains W*N regular particles How does it work? Step 1: Start with 3 macroparticles with W=2 each Step 2: Split particles Into pairs What happens during plasma decay stage? • PIC code -> particles are represented via macroparticles • 1 macroparticle = N (nominal number) regular particles • Introduce weight W of macroparticle -> one macroparticle contains W*N regular particles How does it work? Step 1: Start with 3 macroparticles with W=2 each Step 2: Split particles Into pairs Step 3: End up with 6 macroparticles with W=1 each Does it really work? 1D Ex, V/m 2D Ex, V/m Set 1 Set 2 3.3 ns 3.3 ns We performed studies of surface discharge propagation with different combination parameters and observed no visible difference Back to plasma medicine: Simulation parameters • Simulation Domain – 1cm x 1cm • Grid – 5000 x 5000 (2µm grid size) • Time step = 75 fs • 1 macroparticle = 4*104 particles/m • Threshold number of particles in cell for combining is 5 • Gas – atmospheric air (Oxygen/Nitrogen mixture) • Collisions – ionizations, excitation, elastic • Boundary conditions – bottom electrode is grounded, Negative voltage of -30kV (with 0.1ns rise time) is applied to top electrode • Relative dielectric permittivity of tissue is 20 • Initial electrons are randomly seeded near top electrode • Tissue surface acts as an absorber for charged species Evolution of electron number density Evolution of electron number density • Streamers are independently generated, but start to overlap during propagation • If streamer is close to the tissue, it propagates faster and tends to shield/deviate neighboring streamer • Once one streamer touches the surface, surface discharge starts to propagate Evolution of electric potential Evolution of electric potential • Electric potential is quasi-uniform within the streamer body • When plasma touches the tissue, the electric potential of the tissue is mainly defined by the thickness and permittivity of top dielectric Evolution of vertical component of electric field • Electric field inside streamer body is 1-2 orders of magnitude lower than outside the streamer • The is an enhancement of electric field near the tissue when streamer approaches the tissue • When streamer touches the tissue surface, strong electric field penetrates into the tissue