Negative Ions Extraction by 3D Particle-Based Models

Transcription

1 Negative Ions Extraction by 3D Particle-Based Models Francesco Taccogna Pierpaolo Minelli Istituto di Metodologie Inorganiche e dei Plasmi (IMIP) Consiglio Nazionale delle Ricerche (CNR) Bari francesco.taccogna@cnr.it Keio University 2013

2 Outline o o o o o o o Importance of a detailed extraction region model Description of EXTRA_3D Electrostatic Structure Shape / Size Effects Beam Optics Negative Ion Statistic Conclusions

3 H - extraction: open questions The crucial region is around an extraction aperture 1 cm from PG!! What is the electric field structure in the source-extraction transition region? - It is one of the most complex plasma-surface transition region: magnetic field, electronegative, surface emission; aperture; - Make EG electric field penetrates the production surface. Is it possible to increase the extraction probability χ s of surface-produced H-? Is the H- transport purely electrostatic or collision- or magnetic-induced? What is the best shape and extraction/production area ratio? Meniscus -> Consequences for beam quality

4 3D PIC-MCC Model: extra_3d Simulation domain single element of a multiaperture flat grid: - 1cm diameter; 1 cm spacing between apertures - φ PG = 25 V - φ EG = 9 kv Assumptions-Limitations 3D cartesian geometry [1] : PG + part of EG Magnetostatic fields: a) filter: homogeneous b) e-suppression: realistic map Plasma injection condition (from Fubiani source model) Fixed gas H/H 2 (v) background - n H =1x10 19 m -3 ; n H2 =4x10 19 m -3 - T H =0.8 ev; T H2 =0.1 ev Surface-produced H- by ion/neutral conversion: - J H-,0 =q<γ 0 Y 0 >=660 Am -2 - MC probabilistic methodology [2] from expansion region Simulation parameters Δt = 1x10-11 s Δx ~ local λ D -> N g =N x xn y xn z =161x161x257 N part = 4x10 7 (w=4.8x10 4 ) Monte Carlo null collision method: plasma-gas (TPMC) & plasma-plasma (DSMC) T tot = 0.5 µs in 7 days 12 s per PIC 8 Quad Core Intel Xenon X5570 (2.93 GHz, 96 GB RAM) - MPI paradigm [1] Poisson equation solver: [2] M. Seidl, H.L. Cui, J.D. Isenberg, H.J. Know, B.S. Lee, S.T. Melnychuk, J. Appl. Phys. 79, 2896 (1996).

5 Particle Injection Condition Extraction region models allow resolving gradient lenght scale typical of space-charged regions; Necessity to have fine mesh: for example meniscus shape Extraction region models need detailed data from the expansion region: - plasma particle parameters: flux / temperature / distribution function - plasma potential and results (plasma parameters structure/extraction currents) are sensitive to injection conditions Data from Fubiani [1] expansion region model: - Electron flux: J e =200 A/m 2 - Positive ions flux: J H+ =80 A/m 2 J H2+ =43 A/m 2 J H3+ =40 A/m 2 - Negative ion flux: J H- =20 A/m 2 J tot,hx+ =163 A/m 2 v th =13786 m/s from expansion region [1] From G. Fubiani (today afternoon)

6 Surface Conversion Fixed number of atoms are launched with a prescribed EDF against PG: Y & E = = $ ' th / R ) Rn) Rn( 1 % Ein ' ( H 0 e #! " J H-,0 =660 A/m 2 - H- from PG by neutral conversion are uniformely launched over y - Possibility with Cs dynamic model results to launch a disomogeneous flow [1] M. Seidl, H.L. Cui, J.D. Isenberg, H.J. Know, B.S. Lee, S.T. Melnychuk, J. Appl. Phys. 79, 2896 (1996) [2] N. Kohen, private comunication (2012).

7 Electron Suppression Field Map At this scale the filter field is considered homogeneous B es >B mf considered homogeneous in y-direction From G. Fubiani (private communication; 2012)

8 Flat vs Chamfered: Penetration of EG field vs virtual cathode Flat LAG PG 45 Chamfered LAG PG - In the flat case the EG field penetrates the full collar but no H- are produced there; - In the 45 chamfered case, the penetration of EG field is limited to the last 0.8 mm inside the collar and the potential well is shrunk to the first part of the collar and deeper (~30 V).

9 LAG vs CEA: aperture size effect 45 Chamfered CEA PG 45 Chamfered LAG PG - In the CEA case the EG field penetrates deeper in the source region creating a bigger meniscus; - This leads to an increases of the extraction probability in the oblique part (80% against 37% of LAG) nevertheless the larger CEA extraction area and extraction to production area ratio gives a lower current density; - The larger meniscus allows also a larger volume-produced H - and co-extracted electron current densities.

10 Statistics on Negative Ions (Production / Extraction) Note that the volume production is limited to the domain simulated Production Population Extraction Population Extraction probability of surface-produced H - by ion conversion 16 % Extraction probability of surface-produced H - by neutral conversion 1.7 % H- produced by ion-conversion have energy sufficient to overcome the barrier of the virtual cathode

11 CX Influence on Extraction of Surface-Produced H - CX frequency computed: ν CX =2.54x10 5 s -1 28% of the surface-produced H - ions extracted have suffered a CX collision with H

12 Flow Field electrons The (y,z) plane is interesting because it shows the ExB drift effect: typical fountain between PG and EG. It starts to play some role already in the source region: flow to PG in the y-direction. molecular ions negative ions Total H x + current on PG 150 A/m 2 H - produced 2/3 mm around the aperture deviate towards the PG aperture; it is an electrostatic effect!! (no gyro-magnetic rotation is visible)

13 Influence on H - beam structure H - density map in (y,z)

14 Influence on H- beam structure a b H- density map in (y,z) a bcd a) illumination asymmetry due to electron deflection; b) virtual cathode; c) EG field penetration; d) H- beam quality (halo and divergency); c d

15 Future Directions Present Model Gas (H/H2) Kinetics/Dynamics - H recombinative desorption on PG Acceleration Region - Self-consistent coupling with extraction (Meniscus) Future work: - application to ISCRA-CINECA HPC project: Fermi-BlueGene/Q: - Cores: Theoretical Peak: TFlop/s