First Experiments with / Plasmas: Enhanced Centrifugal Separation from Diocotron Mode Damping

Transcription

1 First Experiments with / Plasmas: Enhanced Centriugal Separation rom Diocotron Mode Damping A.A. Kabantsev a), K.A. Thompson, and C.F. Driscoll Department o Physics, University o Caliornia at San Diego, La Jolla, CA , USA a) Corresponding author: akabantsev@ucsd.edu Abstract. Negative hydrogen ions are produced and contained within a room-temperature electron plasma, by dissociative electron attachment onto exited neutrals. We observe a strongly enhanced centriugal separation o electrons and ions when a diocotron mode is present. The outward ion transport rate is proportional to the diocotron mode amplitude, with concurrent diocotron mode damping. This is not yet understood theoretically. INTODUCTION Two-species non-neutral plasmas, such as, have the desirable coninement properties expected rom conservation o angular momentum; and they also exhibit novel waves, damping, and transport eects due to the disparate masses. Here, we describe the apparent enhancement o centriugal mass separation driven by -species diocotron mode dynamics, inviting theory analysis. In the case o multispecies plasmas conined in the Penning-Malmberg trap, an important eature is that the plasma (as the electrons cool down radiatively with e-olding time 386 ) relaxes towards a thermal equilibrium characterized by a centriugal separation between the components o dierent masses [1] with heavy species pushed to the outside o the plasma. The degree o separation depends on the ratio between the dierence in centriugal potentials or each species and the plasma thermal energy. Due to the mass dierence, the rotation requencies are never the same (beore reaching thermal equilibrium), so there is a collisional drag between the species as the basic separation mechanism. This drag orce produces a drit in the radial direction that drives heavier particles out and the other in. The characteristic time describing the mobility drit toward the plasma edge is much longer than the eective interspecies collisional time by the ratio / []. The mass separation was clearly observed in / plasma experiments [3, 4], apparently with a rate set by the cyclotron cooling time scale [3], which is too ast (by several orders o magnitude) comparing it to predictions o the separation rate based on the collisional drag diusion/mobility theory [5] or the test particle calculations []. Thus, it appears that some other processes than collisional drag are responsible or the accelerated centriugal mass separation. Here we study the eects o omnipresent diocotron modes on the dynamics o the plasma. These experiments were done on the CamV electron plasma apparatus at magnetic ield 1, plasma column length 34 and radius 1., electron density ~ 10. The detail apparatus description and typical electron plasma and trap parameters may be ound elsewhere [6 8]. PLASMA PODUCTION In our experiment negative hydrogen ions are produced and conined in a room-temperature electron plasma [9]. The ions are ormed in the plasma volume by dissociative electron attachment (DEA)

2 , (1) o low energy electrons onto vibrationally and rotationally excited hydrogen molecules [10]. Wide spectrum o ro-vibrationally excited molecules, is generated in the recombinative desorption process o atomic hydrogen on the trap electrode and wall suraces [11]. The yield strongly depends on the electrodes temperature, and some conditioning o the electrodes by heating and/or positive hydrogen ion bombardment leads to an enhanced recombinative desorption, resulting in a signiicantly increased in-volume production rate. Since the total yield is proportional to the number o conined electrons, we characterize the ion production (electron conversion) rate as / /. In our best vacuum conditions ( 10 Torr) the production rate saturates at the level ~ 10 ater 48 hours o the trap s electrodes heating (conditioning) by thermal radiation rom the hot tungsten ilament. This rate allows one to accumulate the charge raction o ions up to 0% in about 00 sec plasma coninement time. It is instructive to note here that the DEA process conserves the total plasma charge, i.e., 0. DIAGNOSTIC TECHNIQUES Measurements o the diocotron and plasma requencies is an attractive way to monitor the dierent mass ractions, since the requencies can be measured with high precision, and the entire evolution can be reconstructed rom one experimental cycle. Neglecting or a moment the temperature dependence, one can state that requency o the primary 1 diocotron mode, ~.3, represents the total (net) charge line density o the plasma,, as the both mass ractions evolve in the slow ( drit) diocotron motion, and 0 0. On the other side, requency o the primary 0, 1 plasma (Trivelpiece-Gould) mode, ~.9, represents the electron charge line density o the plasma,, as only the electrons participate in the ast plasma oscillations, and 0 0. Thus, ater an initial plasma cooling by cyclotron radiation o electrons down to the wall temperature, the diocotron requency remains nearly constant and independent o the production (accumulation) rate (as ), while the TG-mode requency ollows the continuously decreasing electron raction density as ln 0.5. The 1 diocotron mode in electron plasma is neutrally stable (neglecting its interaction with the resistive wall), so in an ideal case a single mode can persist during the entire plasma evolution. The inite length diocotron mode requency has a magnetron shit arising due to the electrostatic orce and the thermal pressure o plasma column balanced by the end coninement voltages [1]: cen 1 T ( t) = ln ql W W π BW L 4 p e N ql Here, i we assume that during an experimental evolution the total plasma charge is conserved, then temporal requency variations are inlicted only by changes in plasma temperature and/or characteristic plasma radius (the bulk transport). It is instructive to keep in mind that these changes also vary and, accordingly. For a typical plasma electrostatic energy, ~ 10, and the trap aspect ratio, 0.1, a 1 change in plasma temperature results in about 0.5% change in the diocotron requency. The TG mode requency is also temperature dependent through the thermal pressure term as ( ) ln 1, 1 e NeL W 3 T TG1 t = + L me p 4ln( W p) e NeL and the thermal correction here is much stronger ( 0) than the one in since it is not multiplied by the small aspect ratio. In the end o a particular evolution, by dumping the plasma column axially onto the phosphor screen with a CCD camera sensitive only to the ast electron component, the remaining amount o electrons is veriied and calibrated against the plasma mode requency. I the electron dumping process is done in quick enough manner, the ions have no time to escape the trap, and the diocotron mode requency ater the dump relects the charge. () (3)

3 In addition, at any given time during the plasma evolution we can rapidly clear the plasma column o the accumulated ions, using a unique shaking technique briely described below. This brings down the current total charge line density to the electron line density, and measurement o determines the lost as well. There is no signiicant change in during the clearing process, i.e., there are no noticeable losses or heating o electrons. EXPEIMENTAL ESULTS AND DISCUSSIONS Figure 1 shows the measured decrease in as electrons are converted to ions, or two dierent apparatus conditions. In the irst case, the electron source ilament was ON only or a couple o hours (beore the experiment), and the ions production rate was as low as.19/. In the second case, the ilament was ON or more than 48 hours, which results in 1./. Note that this second rate would give complete conversion o to in about 1000 sec, i no loss process occurred. 900 TG 1 [ khz] ph.19 ksec 800 ph 1. ksec [ T t> T = const] ( 0s) wall FIGUE 1. Measured requency evolutions o thermally exited Trivelpiece-Gould modes or two extreme (low/high) ion production rates. Only the late time parts o the evolutions ( 0, when ) are shown here. For clearing the trapped ions, the shaking technique was developed. The longitudinal motion o the ions is caused by a small electric ield internally generated by the mobile electron plasma as a result o variations in the wall potentials (the sloshing column motion) close to the ion plasma requency ~. Properly tuned (to keep the plasma length ) oscillations o the end coninement voltages shake plasma column back and orth axially ( 1 ) at a resonant clearing requency 4 ~. The ull clearing cycle o accumulated ions requires about 500 rounds (10 ) o the plasma column shakings. Figure shows the diocotron mode amplitudes and requencies as an injected electron plasma cools down to room temperature, exhibiting diocotron growth and damping. The weak diocotron growth is due to ~10 o transiting ions [13], incidental to the trapped ions eects considered here. The diocotron damping is due to accumulation and outward transport o ions; and the damping ceases with application o " clearing" steps. Figure (a) shows examples o the 0 diocotron requency evolution or small amplitude modes 10 excited right upon the electron plasma injection. Ater initial exponential cooling down as, the diocotron requency shows a very small decline rate as ln 10, coming rom the slow radial expansion (transport) o the electron plasma. Note that this decline rate is at least 0 times smaller than concurrent rate ln 0.5 (see Fig. 1). However, when we apply a train o clearing cycles, the diocotron requency drops down by the present raction o ions, matching the concurrent total charge line density to the electron line density.

4 Figure (b) shows evolutions o diocotron mode amplitudes corresponding to Fig. (a). With no ongoing ion accumulation the diocotron mode growth irst shows a slow exponential growth with an increment ~. 05 deined by the single pass ion-induced instability [13]. As the electron temperature cools down and the DEA process starts to convert electrons into ions, the diocotron mode growth irst shows a slowing down, and then ultimately turns into a damping. Even so, the (repetitive) plasma clearing o the ions recovers the original exponential growth t c ւ e τ, τ c.7 sec 1 d ( t) i 1 d ( t ) 1 d (a) ւ p H 0.15 k sec 0.01 < k sec D( t) H Clears H W Clear t 0 d e epetitive H Clears (b) Clear H FIGUE (a, b). Measured evolutions o (a) requency / and (b) amplitude.01 o diocotron modes or the slow ion production rates, ~ 0.1/ sec, with a variety o clearing cycles. The bold (red) lines represent the case o seven consecutive clearing cycles, and the thin (black) lines represent the case with only a single clearing cycle at 80. Each clearing cycle adjusts the concurrent total charge line density to the electron line density. Figures 3(a) and 3(b) illustrate diocotron mode evolutions observed in the case o high ion production rates, ~ 1/. Here the modes are excited to moderate amplitudes 0.01 to mitigate the stronger late stage damping. At this high ion production rates the concurrent damping o diocotron modes becomes very potent ( ) and it clearly shows (see Fig. 3(b)) its algebraic character as, where ~. The observed algebraic damping o diocotron mode is coincident with the accelerated downturn in that relects an increase o the plasma column mean-square radius relative to the center-o-mass (com) rame. When, and, the only change in the diocotron requency is set by (the net charge) plasma expansion as ln 0.6 (see the magnetron term in Eq. ()). We credit this change to a diocotron-induced centriugal mass separation that by drits moves uniormly produced in plasma volume ions to the (radial) edge o the electron column. This outward lux o heavy charges pushes back on the diocotron displacement, thus conserving the mean-square radius o the total charge ensemble in the center-o-cylinder rame (canonical angular momentum). Figure 3(a) demonstrates that this accelerated radial transport in the total charge density (as the two pronounced downturns in the late time evolution) is active only during the diocotron mode damping stages ( 30, and 70 ), i.e., when (1) the diocotron mode and () its algebraic damping are both present. There is no comparable all in when the diocotron mode is growing, and in the time span when the diocotron mode is damped away. This decline o the ln magnetron term in Eq. () then continues again ollowing the external re-excitation o the diocotron mode. It is instructive to note that all three rates here (the ion production rate ~ 1/, the diocotron damping rate 1.9/, and the radial transport rate 1.6/ ) are meaningully close to each other. Heating o the electron plasma above a ew tenths o an ev seizes up both the diocotron damping and the ion production/separation processes.

5 The algebraic nature o this damping is more clearly shown in Fig. 3(b). At this time, the original ( 0) diocotron mode was already completely damped (ater its initial exponential growth) by a ast ion production (and separation) rate, so at 60 the diocotron mode was re-excited. In addition to the linear slope,, the itting curve in Fig. 3(b) also includes a exponentially vanishing surplus lux term,, to account or a surplus o ions accumulated in the plasma volume during the preceding absence o the diocotron mode, i.e., beore the diocotron-induced mass separation kicks in again. Derived rom this it the surplus loss (separation) time, 0.3, is extremely ast compare to the collisional drag separation time, / ~ 10, as well as to the top damping and ion production rates ( t) 1 d ( t ) d( t) ( ) d t D W ph 1/ k sec d-excitation 0.0 D( t) W D( t) W 0.01 p γ H H 1 / k sec 1.9 / k sec d = γ H σ e tτs ɺ ɺ ρ ρ (a) (b) FIGUE 3 (a, b). Measured evolutions o the diocotron mode in the case o high ion production rates, ~ 1/. (a) Frequency / and amplitude with a strong late time damping, 1.9/. The damping is coincident with the accelerated downturn in that relects plasma expansion as ln 0.6. (b) Amplitude o the diocotron mode re-excited at 60 on a ine time scale emphasizing the algebraic (nonexponential) character o the damping process, and also showing the eect o exponentially vanishing non-separated surplus accumulated in the plasma volume during the absence o the diocotron-induced mass separation. SUMMAY The ability to accumulate up to 0% charge raction o ions (in about 00 seconds) in room temperature cold electron plasma has been demonstrated in our experiments. The ion accumulation process causes a novel algebraic damping o diocotron modes with the damping rates set by the ion production rate, and this damping is driven by the abnormally ast centriugal mass separation (outward transport o the newly produced ions in the wobbling electron plasma). Whereas all these rates are meaningully close to each other, thus ar no theoretical explanation has been given or the phenomenon. While the collisional drag centriugal mass separation rates are admittedly slow in many applications, these rates can be greatly enhanced by the ubiquitous presence o diocotron modes discussed here. ACKNOWLEDGMENTS This research was supported by NSF/DoE Partnership grants PHY and DE SC T. M. O Neil, Phys. Fluids 4, (1981). EFEENCES

6 . M. Amoretti, C. Canali, C. Carraro, V. Lagomarsino, A. Odino, G. Testera, and S. Zavatarelli, Phys. Plasmas 13, (006). 3. G. B. Andresen et al. (ALPHA Collaboration), Phys. ev. Letters 106, (011). 4. G. Gabrielse et al. (ATAP Collabortaion), Phys. ev. Letters 105, 1300 (010). 5. D. H. E. Dubin, Equilibrium and dynamics o multispecies nonneutral plasmas with a single sign o charge, in Non-Neutral Plasma Physics VIII, AIP Conerence Proceedings 151, edited by X. Sarasola et al. (American Institute o Physics, Melville, NY, 013), pp A. A. Kabantsev, Yu. A. Tsidulko, and C. F. Driscoll, Phys. ev. Letters 11, (014). 7. A. A. Kabantsev, C. Y. Chim, T. M. O'Neil, and C. F. Driscoll, Phys. ev. Letters 11, (014). 8. A. A. Kabantsev, D. H. E. Dubin, C. F. Driscoll, and Yu. A. Tsidulko, Phys. ev. Letters 105, (010). 9. A. A. Kabantsev and C. F. Driscoll, TG wave autoresonant control o plasma temperature, in Non-Neutral Plasma Physics IX, AIP Conerence Proceedings 1668, edited by H. Himura et al. (American Institute o Physics, Melville, NY, 015), p J. Horáček, M. Čížek, K. Houek, P. Kolorenč, and W. Domcke, Phys. eview A 70, 0571 (004). 11. S. Markelj and I. Čadež, The Journal o Chemical Physics 134, (011). 1. K. S. Fine and C. F. Driscoll, Phys. Plasmas 5, (1998). 13. A. A. Kabantsev and C. F. Driscoll, eview o scientiic instruments 75(10), (004).