Transcript Slide 1
Helicons: A Physics Dilemma Solved Francis F. Chen, UCLA Tsing Hsua University, Taiwan, March 2008 How helicons started: 1962 3kW 17 MHz 500G UCLA How helicons started: 1970 - 85 1 kW, 1 kG, argon n = 1013 cm-3 10X higher than normal UCLA B (a) In helicon sources, an antenna launches waves -+ in a dc magnetic field -- + -- + + -- (b) (c) The RF field of these helical waves ionizes the gas. The ionization efficiency is much higher than in ICPs. A large number of problems arose • Absorption mechanism and efficiency • Weak m = –1 mode and the Big Blue Mode • Downstream density peak, axial ion flow • Non-monotonic axial decay • Triangular radial profile • Mass-dependent density limit • Low-field density peak (~30G) • Density jumps with increasing B0, Prf • Half-wave antenna better than full-wave • Endplate charging with small diameters • High ion temperatures • Parametric instabilities UCLA The Landau damping hypothesis In Landau damping, electrons surf on the wave + -- The helicon’s phase velocity is close to that of an electron near the peak of the ionization cross section (~100eV) UCLA Landau damping disproved A fast (RF) energy analyzer was built and calibrated vacuum feedt hrough col lect or fi lament heater -0- +12V anode ceramic cov ered wires g rid bias -0- -200V BN s pacers RF modulated electron gun for calibration g ro un d ed h o us in g 2-electrode gridded analyzer with RF response D.D. Blackwell and F.F. Chen, Time-resolved measurements of the EEDF in a helicon plasma, Plasma Sources Sci. Technol. 10, 226 (2001) UCLA Time-resolved EEDFs show no fast electrons above a threshold of 10 -4 60 1 0 0 0 .0 3 .3 5 e V 40 1 0 .0 Ic (mA) Electron current (mA) 1 0 0 .0 1 .0 20 0 .1 3.00 eV 0 0 .0 0 50 100 150 200 - 15 time (nsec) -5 I-V swept by oscillating Vs 5 Vo lt s 15 25 I-V at two RF phases 3 1.2 Evolution of beam-created plasma 30 20 0.8 D is tanc e I (mA) 2 fro m gu n I (mA) R p ( ) 10 1 5 cm 10 cm 0 0 25 50 t (nsec) 75 100 20 cm 0.4 n o p la s m a 0.0 0 -180 -90 0 90 A n te n n a p h a s in g (d e g r e e s ) Loading resistance agrees with calculations w/o L.D. 180 -150 -100 -50 Volts 0 50 Injecting a current causes a beam-plasma instability The Trivelpiece-Gould mode absorption mechanism • Helicon waves are whistler waves confined to a cylinder. • Their frequencies are << c, so that normally me 0 is OK. • However, if me 0, the dispersion relation has another root. • The new root is an electron cyclotron wave in a cylinder. It is called a Trivelpiece-Gould (TG) mode. • The TG mode exists in a thin layer near the surface and is damped rapidly in space, since it is slow. The helicon wave has weak damping. • This mechanism was suggested by Shamrai and Taranov of Kiev, Ukraine, in 1995. UCLA Why are helicon discharges such efficient ionizers? Trivelpiece-Gould mode The helicon wave couples to an edge cyclotron mode, which is rapidly absorbed. Helicon mode 2 P(r) ( / cm ) 1.2 1.0 Parabolic Profile: R = 1.47 Ohms 0.8 Square Profile: R = 1.71 Ohms 0.6 0.4 0.2 0.0 0 1 2 r (cm) 3 4 The H and TG waves differ in k 1.0 n = 4x1011 cm-3 B(G) 20 30 60 300 -1 k (cm ) 0.8 k|| 0.6 TG 0.4 H 0.2 0.0 0 1 2 -1 b (cm ) 3 4 This axis is essentially k UCLA Detection of TG mode was difficult 14 RF PROBES 35G, 4E11 MAGNET COILS ANTENNA & SHIELD AXIAL PROBES Jz, Bz (arb. units) 12 10 8 6 Jz calc Bz calc 4 2 MATCHING CIRCUIT RF 0 -6 Faraday shield -4 -2 0 r (cm) 2 4 6 12 30G, 2E11 Jz, Bz (arb. units) 10 8 6 Jz data Bz data 4 2 Return loop UCLA 0 -6 -4 -2 0 r (cm) 2 4 6 An RF current probe had to be developed A large number of problems arose • Absorption mechanism and efficiency • Weak m = –1 mode and the Big Blue Mode • Downstream density peak, axial ion flow • Non-monotonic axial decay • Triangular radial profile • Mass-dependent density limit • Low-field density peak (~30G) • Density jumps with increasing B0, Prf • Half-wave antenna better than full-wave • Effect of endplates and endplate charging • High ion temperatures • Parametric instabilities UCLA Types of antennas B RH and LH helical Nagoya Type III Boswell double saddle coil 3-turn m = 0 UCLA The m = +1 (RH) mode gives much higher density Comparison of time with space rotation 4 L = 10 cm (+1,+1) (+1,-1) (-1,+1) (-1,-1) B=0 RH mode n (10 13 -3 cm ) 3 2 1 LH mode 0 -40 -20 0 20 40 z (cm) 60 80 100 120 UCLA m = +1 mode much stronger than m = –1 D.D. Blackwell and F.F. Chen, 2D imaging of a helicon discharge,Plasma Sources Sci. Technol. 6, 569 (1997) m = –1 m = +1 UCLA Reason m = -1 mode is not easily excited m = +1 m = -1 The m = -1 mode has a narrower wave pattern; hence, it couples weakly to the TG mode at the boundary. The Big Blue Mode The dense core (n = 1013-14 cm-3) is due to neutral depletion, allowing Te to increase No Faraday shield With shield A large number of problems arose • Absorption mechanism and efficiency • Weak m = –1 mode and the Big Blue Mode • Downstream density peak, axial ion flow • Non-monotonic axial decay • Triangular radial profile • Mass-dependent density limit • Low-field density peak (~30G) • Density jumps with increasing B0, Prf • Half-wave antenna better than full-wave • Endplate charging with small diameters • High ion temperatures • Parametric instabilities UCLA Machine used for basic studies r ~ 5 cm L ~ 160 cm B 1kG, Prf 2kW @ 13-27 MHz, 1-10 MTorr Ar UCLA Symmetric and asymmetric antennas The maximum density occurs DOWNSTREAM, while Te decays. This is due to pressure balance: nKTe = constant. UCLA Line radiation is main loss in Te decay UCLA Non-monotonic decay of wave downstream Oscillations are due to beating of radial modes with different k||. Theory fails as density changes further out. Average decay rate agrees with collisional damping. UCLA Triangular density profiles 1.0 2.0 0.8 1.5 0.6 n / no n (1013cm-3 ) 2.5 1.0 0.4 0.5 0.2 0.0 heat source 0.0 -5 -4 -3 -2 -1 0 r (cm) 1 2 3 4 0.0 0.2 0.4 0.6 r/a 0.8 1.0 Nonlinear diffusion, coupled with a bimodal ionization source, can explain "triangular" density profiles. UCLA 1.2 A large number of problems arose • Absorption mechanism and efficiency • Weak m = –1 mode and the Big Blue Mode • Downstream density peak, axial ion flow • Non-monotonic axial decay • Triangular radial profile • Mass-dependent density limit • Low-field density peak (~30G) • Density jumps with increasing B0, Prf • Endplate charging with small diameters • Half-wave antenna better than full-wave • High ion temperatures • Parametric instabilities UCLA Mass-dependent density limit As B0 is increased, n rises but saturates at a value depending on the ion mass. This effect was first observed by T. Shoji. UCLA A drift-type instability occurs H 20 L f (KHz) M. Light (Ph.D. thesis) found that an instability occurs at a critical field and causes the density to saturate. This is the oscillation spectrum for neon. He identified the instability as a drift-Kelvin Helmholtz instability and worked out the theory for it. 0 2.0 ne (1013/cm3) M. Light, F.F. Chen, and P.L. Colestock, Plasma Phys. 8, 4675 (2001), Plasma Sources Sci. Technol. 11, 273 (2003) 0 0 0.5 B (KG) 1.0 1.5 0 UCLA Anomalous diffusion results 2 Calculated -2 -1 (m s ) 10 21 Measured 0 -0.5 0 B (KG) 0 1600 Outward particle flux was measured with n – f correlations, agreeing with that calculated quasilinearly from the growth rate. UCLA Density limit due to neutral depletion Axial density profile with two 2-kW antennas 1m apart UCLA A large number of problems arose • Absorption mechanism and efficiency • Weak m = –1 mode and the Big Blue Mode • Downstream density peak, axial ion flow • Non-monotonic axial decay • Triangular radial profile • Mass-dependent density limit • Low-field density peak (~30G) • Density jumps with increasing B0, Prf • Endplate charging with small diameters • Half-wave antenna better than full-wave • High ion temperatures • Parametric instabilities UCLA A density peak occurs at low B-fields The cause is the constructive interference of the reflected wave from a bidirectional antenna UCLA HELIC computations of plasma resistance 3.0 d 2E+11 4E+11 6E+11 8E+11 1E+12 2.0 5 cm 10 cm No bdy 5 4 R (ohms) 2.5 R (ohms) 6 n (cm-3) 1.5 3 1.0 2 0.5 1 0 0.0 0 50 100 150 200 B (G) Vary the B-field Vary the with endplate conductivity 0 50 100 150 200 250 300 B (G) Vary the endplate distance Uni- and b-directional antennas The end coils can also be turned off or reversed to form a cusped B-field to pump END COILS The field lines then end on the glass tube, which forms an insulting endplate. An aperture limiter can also be added. A cusp field or and end block can greatly increase the density G. Chevalier and F.F. Chen, Experimental modeling of inductive discharges, J. Vac. Sci. Technol. A 11, 1165 (1993) A large number of problems arose • Absorption mechanism and efficiency • Weak m = –1 mode and the Big Blue Mode • Downstream density peak, axial ion flow • Non-monotonic axial decay • Triangular radial profile • Mass-dependent density limit • Low-field density peak (~30G) • Density jumps with increasing B0, Prf • Endplate charging with small diameters • Half-wave antenna better than full-wave • High ion temperatures • Parametric instabilities UCLA Discharge jumps into helicon modes n vs. RF power n vs. B-field R.W. Boswell, Plasma Phys. Control. Fusion 26, 1147 (1984) UCLA Transition to helicon mode W: helicon H: inductive E: capacitive A.W. Degeling and R.W. Boswell, Phys. Plasmas 4, 2748 (1997) UCLA UCLA A new interpretation of the jumps 1000 P in (W per tube) Prf (W) 100 B = 80G Rc = 0.5 400W, ICP 400 300 200 100 50 20 Loss Pin Prf R p Rc The power into the plasma depends on the plasma loading (Rp) and the circuit losses (Rc) 10 1 0.01 Rp 0.1 1 n (1011 cm-3) 10 100 If Rp is too small, the input power is less than the losses. The jump into helicon mode can be computed from theoretical Rp’s. The critical power agrees with experiment. F.F. Chen and H. Torreblanca, Plasma Sources Sci. Technol. 16, 593 (2007) A large number of problems arose • Absorption mechanism and efficiency • Weak m = –1 mode and the Big Blue Mode • Downstream density peak, axial ion flow • Non-monotonic axial decay • Triangular radial profile • Mass-dependent density limit • Low-field density peak (~30G) • Density jumps with increasing B0, Prf • Endplate charging with small diameters • Half-wave antenna better than full-wave • High ion temperatures • Parametric instabilities UCLA A 1-inch diam helicon discharge UCLA Critical field is where rLe ~ a UCLA A large number of problems arose • Absorption mechanism and efficiency • Weak m = –1 mode and the Big Blue Mode • Downstream density peak, axial ion flow • Non-monotonic axial decay • Triangular radial profile • Mass-dependent density limit • Low-field density peak (~30G) • Density jumps with increasing B0, Prf • Endplate charging with small diameters • Half-wave antenna better than full-wave • High ion temperatures • Parametric instabilities UCLA Half wavelength helical antennas are better than full wavelength antennas 4 13 -3 n (10 cm ) 3 2 10cm half wavelength 15cm full wavelength 20cm full wavelength 1 0 0 20 z (cm) 40 60 L. Porte, S.M. Yun, F.F. Chen, and D. Arnush, Superiority of half-wavelength helicon antennas, LTP-110 (Oct. 2001) 80 A large number of problems arose • Absorption mechanism and efficiency • Weak m = –1 mode and the Big Blue Mode • Downstream density peak, axial ion flow • Non-monotonic axial decay • Triangular radial profile • Mass-dependent density limit • Low-field density peak (~30G) • Density jumps with increasing B0, Prf • Endplate charging with small diameters • Half-wave antenna better than full-wave • High ion temperatures • Parametric instabilities UCLA Anomalously high ion temperatures Unusually high Ti’s are observed by laser induced fluorescence. This happens near lower hybrid resonance, but no special heating is expected there. J.L. Kline, E.E. Scime, R.F. Boivin, A.M. Keesee, and X. Sun, Phys. Rev. Lett. 88, 195002 (2002). A large number of problems arose • Absorption mechanism and efficiency • Weak m = –1 mode and the Big Blue Mode • Downstream density peak, axial ion flow • Non-monotonic axial decay • Triangular radial profile • Mass-dependent density limit • Low-field density peak (~30G) • Density jumps with increasing B0, Prf • Endplate charging with small diameters • Half-wave antenna better than full-wave • High ion temperatures • Parametric instabilities UCLA The energy absorption mechanism near the antenna may be nonlinear, involving parametric decay of the TG wave into ion acoustic waves Lorenz, Krämer, Selenin, and Aliev* used: 1. Test waves in a pre-formed plasma 2. An electrostatic probe array for ion oscillations 3. Capacitive probes for potential oscillations 4. Microwave backscatter on fluctuations 5. Correlation techniques to bring data out of noise *B. Lorenz, M. Krämer, V.L. Selenin, and Yu.M. Aliev, Plasma Sources Sci.Technol. 14, 623 (2005). A helicon wave at one instant of time Note that the scales are very different! UCLA Damping rate in the helicon afterglow The damping rate increases with Prf, showing the existence of a nonlinear damping mechanism. UCLA Excitation of a low-frequency wave The LF wave is larger with the e.s. probe than with the capacitive probe, showing that the wave is electrostatic. As Prf is raised, the sidebands get larger due to the growth of the LF wave. UCLA Oscillations are localized in radius and B-field The fluctuation power and the helicon damping rate both increase nonlinearly with rf power. UCLA Proposed parametric matching conditions k2 k1 0 1 2 k 0 k1 k 2 k0 k 0 k 1 k 2 0, k 1 k 2 k0 = helicon wave, k1 = ion acoustic wave k2 = Trivelpiece-Gould mode This was verified experimentally. UCLA Evidence for m = 1 ion acoustic wave The cross phase between two azimuthal probes reverses on opposite sides of the plasma. kq is larger than kr, and both increase linearly with frequency. From the slope one can calculate the ion acoustic velocity, which yields Te = 2.8 eV, agreeing with 3 eV from probe measurements. UCLA With a test pulse, the growth rate can be seen directly From probe data Growth rate vs. power From mwave backscatter Growth rate vs. power Conclusion on parametric instabilities Kramer et al. showed definitively that damping of helicon waves by parametric decay occurs near the axis. They identified the decay waves, checked the energy balance, and even checked the calculated instability threshold and growth rate. However, this process is too small to be the major source of energy transfer from the antenna to the plasma. It is still unknown what happens under the antenna, where it is difficult to measure. It could be that the waves observed were actually created under the antenna but measured downstream. UCLA Title here