ISSN 10637842, Technical Physics, 2013, Vol. 58, No. 10, pp. 1426–1431. © Pleiades Publishing, Ltd., 2013. Original Russian Text © N.V. Gavrilov, A.S. Kamenetskikh, 2013, published in Zhurnal Tekhnicheskoi Fiziki, 2013, Vol. 83, No. 10, pp. 32–37.

PLASMA

SelfOscillating Mode of Electron Beam Generation in a Source with a Grid Plasma Emitter N. V. Gavrilov* and A. S. Kamenetskikh Institute of Electrophysics, Ural Branch, Russian Academy of Sciences, ul. Amundsena 106, Yekaterinburg, 620016 Russia *email: [email protected] Received January 9, 2013

Abstract—Plasma processes and electron beam generation in an electron source with a grid plasma cathode are investigated. Experiments are conducted under the conditions of efficient electron extraction and an intense counter ion flux, which break grid stabilization. It is shown that a rise in the gas pressure and in the emitting plasma potential leads to the plasma potential modulation in the frequency range 104–105 Hz. Under the selfoscillation conditions, an electron beam is obtained with a constant current of up to 16 A and an electron energy modulated up to 100% of the accelerating voltage level (100–300 V). An explanation is given for relaxation selfoscillations arising when the plasma potential grows and for the system’s inertial non linearity arising when the plasma potential induced by the space charge of the counter ion flux lags behind the current of electronbeamgenerated ions. DOI: 10.1134/S1063784213100101

INTRODUCTION Plasmaemittertype electron sources are used for generation of both broad and focused electron beams with an energy of up to 150 keV and a pulsed current from several fractions of an ampere to a kiloampere [1, 2]. The performance of such sources is efficient if grid stabili zation conditions, under which the extraction of elec trons does not influence the plasma parameters, are provided [3]. Langmuir layers forming in the meshes produce a potential barrier or bound the open surface area of the plasma. As a result, the electron emission current stabilizes at a level not exceeding the discharge current in the plasma cathode. The grid stabilization conditions are met at low pressures of the gas and suf ficiently high accelerating voltages [4]. When the gas pressure rises and the accelerating voltage declines, the rate of gas ionization by the electron beam increases and so does the counter ion flux into the emitting plasma. This raises the plasma density, elec tron extraction efficiency, and emission current. A fur ther increase in the rate of gas ionization in the accel erating gap and in the drift space may cause, under certain conditions, the process of discharge switchover from the anode to the extractor may go in an avalanche manner [5].

efficiency of electron extraction from the plasma and at a constant emission current, which is due to stabili zation provided by the circuitry. Efficient extraction of electrons from the emitting plasma makes the com pensation for the counter ion flux positive charge dif ficult; therefore, when the gas pressure and counter ion flux grow, the emitting plasma potential may rise to a value comparable to the accelerating voltage [7]. It may be expected that the growth of the emitting plasma potential and its respective decline of the accelerating voltage will cause switchover from the electron beam generation mode to the mode of dis charge at the collector. However, high ion losses at the grid (which bridges the discharge gap and is under a high negative potential relative to the beam plasma potential) in combination with decreased ionizing capacity of electrons (due to the decrease in the poten tial difference between the beam and emitting plasma) may hinder switchover of the discharge electron cur rent to the remote collector. In this work, we study plasma processes in an elec tron source with a grid plasma cathode and the gener ation of lowenergy electron beams under the condi tions of grid stabilization disturbance due to efficient electron extraction and formation of an intense counter ion flux.

Plasma sources of electrons used to generate a plasma by an electron beam usually require a low accelerating voltage (about several hundred electron volts) and an increased gas pressure (0.1–1.0 Pa). To raise the life time of the grid exposed to intense ion bombardment, coarse grids with a mesh as wide as sev eral millimeters are usually applied [6]. In such condi tions, a plasma cathode operates with an almost 100%

EXPERIMENTAL In experiments, a hollowcathode glow discharge (0.5–2.5 A) and a selfheating hollowcathode dis charge (3–16 A) were used for electron plasma sources. In the first case, the electrode system of the source includes hollow cathode 1 (200 mm in height

1426

SELFOSCILLATING MODE OF ELECTRON BEAM GENERATION

and in diameter) with an exit aperture diameter of 8 mm and hollow anode 2 (100 mm in height and in diameter). Grid electrode 3 (mesh grid with 1.2 × 1.2mm meshes or a 2mmthick plate with holes 4 mm in diameter arranged within an area 80 mm in diameter) is placed on the end face of the hollow anode. The grid was electrically insulated from the cylindrical part of the anode, so that currents toward these electrodes could be measured separately. In the anode cavity, Langmuir probe 4 was mounted at a dis tance of 40 mm from the wall and 50 mm from the grid. The probe had 5 mm in length and was made of a tungsten wire 0.3 mm in diameter. Electron beam col lector 5 was mounted at a distance of 70 mm from the grid electrode. A stabilized dc current source allowed variation of discharge current I in the range 0.5–2.5 A. Stabilized accelerating voltage U was varied from 50 to 300 V using a power supply connected with the grounded grid and collector. Pressure p of argon (working gas), which was delivered through the cath ode area, was varied in the beam drift space within 0.1–0.8 Pa when the electrode with holes 4 mm in diameter was used and within 0.2–3.0 Pa when the grid with 1.2 × 1.2mm meshes was applied. The elec tron source generated a beam plasma in between the grid and collector. The electrons of the glow discharge anode plasma were extracted through the holes in the grid and accelerated in a space charge layer between the emitting and beam plasmas. In the source based on a selfheatingcathode discharge (for details, see [8]), a titanium nitride hollow cathode in the form of a 70mmlong tube with an inner diameter of 8 mm was applied. The floating potential of the probe and the current in the electron beam collector’s circuit were measured using a TPS 2024 Tektronix digital oscilloscope with a bandwidth of 200 MHz. The energy spectrum of elec trons was taken with the retarding potential technique using threegrid electrostatic energy analyzer 6 with collimator 7 placed at the entrance to the analyzer. This method measures only the longitudinal compo nent of the velocity; therefore, a collimator is neces sary to cut off electrons having gained a considerable transverse velocity because of the curved plasma boundary or interaction with the beam plasma. The diametertolength ratio of the collimator’s channels was selected so that the angle of entry of electrons into the collimator was within ≤3°. Electron delay curves were taken in real time with a Hioki 8835 digital mul tichannel recorder. The energy spectra of electrons were determined by differentiating the delay curves. The analyzer was calibrated using a thermionic cath ode at a residual gas pressure in the vacuum chamber of 2 × 10–3 Pa. The energy resolution of the analyzer at the full width at half maximum was about 5% of the accelerating voltage. In a number of experiments, the plasma density in the electron beam drift space was increased by initiat ing an auxiliary discharge between thermionic cathode TECHNICAL PHYSICS

Vol. 58

No. 10

2013

1427

Ar

1 I 0.5−2.5 A 2

4

3

ϕ

11 8 9

U 50−300 V

10 5 7 6

Fig. 1. Electrode system of the gasdischarge system: (1) hollow cathode, (2) hollow anode, (3) grid, (4) probe, (5) collector, (6) energy analyzer, (7) collimator, (8) ther mionic cathode, (9) rodlike anode, (10) walls of beam drift space, and (11) vacuum chamber. Dashed and dotted lines are joints used when a thermioniccathode auxiliary dis charge is applied.

8 and rodlike anode 9. In this case, the thermionic cathode, beam collector, and walls 10 of the beam drift space were under the potential of the plasma cathode grid. The accelerating voltage was applied between the rodlike anode and grid. EXPERIMENTAL DATA Figure 2 shows the waveforms of the probe’s float ing potential in the emitting plasma, current in the beam collector’s circuit, and currents in the circuits of the hollow anode and grid electrode. As is known, floating potential φ of the probe in the plasma with the Maxwellian distribution of electrons is found from the relationship eφ/kTe ≈ ln(0.77 (M/m)1/2). In the argon plasma, it roughly equals 5.6kTe [9]. Under our exper imental conditions, the electron temperature was equal to 3–5 eV and the difference between the float ing potential of the probe and the plasma potential is 20–30 V. As follows from Fig. 2, the floating potential (curve 1) has constant and variable components. The plasma potential modulation frequency grows with gas pressure. The anode current (curve 2) and grid current (curve 3) are mostly ionic. At the beginning of plasma potential growth, the ionic grid current rises stepwise and the anode current and the floating potential vary

1428

GAVRILOV, KAMENETSKIKH

0(4)

1.6 120 1.2

2

3

60

0.4

0(3)

4

1 0(2) 2, 3:0.1 A/div;

4: 2 A/div 10 μs/div

Fig. 2. Waveforms of the (1) probe potential, (2) current in the hollow anode circuit, (3) current in the grid circuit, and (4) current in the collector circuit. p = 0.25 Pa, I = 2 A, and U = 200 V. The diameter of the grid meshes is 4 mm.

in a similar manner. The electron current in the beam collector circuit (curve 4) is weakly modulated (the modulation rate is 0.2–0.3). The components of the floating potential and electrode currents are plotted versus the gas pressure in Fig. 3. The current in the circuit of the collector (curve 1), on which the cur rent of secondary ions (due to electron beam gas ion ization) also closes, increases sharply when the pres sure rises to 0.2 Pa and then saturates. At the instant the current sharply grows, when the plasma cathode passes to the stimulated emission mode with 100% electron extraction efficiency, the modulation of the plasma potential occurs with a modulation frequency linearly growing with gas pressure (curve 2). The con stant component of the plasma potential (curve 3) also grows with pressure, whereas the variable component varies insignificantly (curve 4). Figure 4 plots the floating potential’s components in the anode plasma and the floating potential oscilla tion frequency versus the thermionic current for the case when the auxiliary discharge with the thermionic cathode is initiated in the beam drift space. With an increase in the discharge current (curve 1), the fre quency and the variable component of the floating potential (curve 2) vary nonmonotonically and finally vanish. The constant component of the potential (curve 3) grows to a value close to the accelerating voltage. The variations of the currents in the circuits of the plasma cathode and rodlike anode of the auxiliary discharge with the thermionic current and interelec trode voltage are shown in Fig. 5. When the thermi onic current and voltage grow, so do the ionic anode current (curves 1, 2), ionic current of the plasma cath ode grid (curves 3, 4), and electronic current of the auxiliary anode current (curves 5, 6).

0

40 20

0 0.1

0.2

0.3 p, Pa

0.4

0.5

0 0.6

Fig. 3. (1) Current in the collector circuit, (2) oscillation frequency, relative (3) constant and (4) variable compo nents of the probe’s floating potential vs. the gas pressure. I = 1 A and U = 200 V. The diameter of the grid meshes is 4 mm.

The electron energy spectra taken with the electro static analyzer and their variation with pressure, dis charge current, and accelerating voltage are detailed in [10]. The shifts of the peaks and their broadening are in good agreement with probe measurement data (Fig. 6). An increase in the discharge current, with the gas pressure kept constant, leads to the plasma potential modulation (Fig. 7). As the gas pressure grows, the plasma cathode switches to the selfoscillating emis sion mode at a lower discharge current. The oscillation 0.8

60 50

1

0.6 ϕ−/U, ϕ~/U

0(1) 1:20 V/div;

80

2

0.8

40

2 0.4 3

30

0.2

0

f, kHz

3

100

1

f, kHz

Ik/I, ϕ−/U, ϕ~/U

4

20

0.4

0.8 1.2 Ie, A

1.6

10 2.0

Fig. 4. (1) Oscillation frequency and relative (2) constant and (3) variable components of the probe’s floating poten tial vs. the emission current from the thermionic cathode. p = 0.25 Pa, I = 1 A, and U = 100 V. The diameter of the grid meshes is 4 mm. TECHNICAL PHYSICS

Vol. 58

No. 10

2013

SELFOSCILLATING MODE OF ELECTRON BEAM GENERATION 6 0.8

8

1429

I, arb. units 1.0

1

4 6

4 0.4

3

2

3 2

0.2

2 0.6

1

4 0.4 5

6 0.5

1.0

Ie, A

1.5

2.0

0 2.5

Fig. 5. Current in the (1, 2) anode circuit, (3, 4) grid cir cuit, and (5, 6) circuit of the auxiliary discharge anode vs. the emission current from the thermionic cathode. U = (1, 3, 5) 100 and (2, 4, 6) 200 V.

frequency, as well as the constant and variable compo nents of the floating potential, varies weakly with dis charge current. DISCUSSION OF RESULTS The variation of the plasma potential influences the thickness of the Langmuir layers in the meshes. This may result in oscillation buildup; however, the transi tion time constant estimated for the ion layer in the grid apertures (on the order of 10–7 s) disagrees with the potential oscillation frequency. The selfoscilla tion process with a cycle duration of about 10 μs or more is induced by factors varying with a longer time constant. From the experiments with the auxiliary dis charge, which were conducted at constant values of the gas pressure and electron emission current, it fol lows that when the counter ion flux grows, the plasma potential rises, selfoscillations cease, and the system switches from the beam generation mode to the dis charge current commutation to the collector. This indicates that the rise in the plasma potential is due to insufficient compensation for the counter ion flux. The moderate electron temperature of the plasma (2– 5 eV) cannot maintain the plasma potential at a high level (several hundred volts). At low accelerating voltages (100–300 V) used in the experiments, a rise in the emitting plasma poten tial significantly decreases the potential difference between the plasmas and also the ionizing capacity of beam electrons. As a result, the rise in the plasma potential due to a rise in the counter ion flux becomes a factor causing the ion flux to diminish. However, the existing negative feedback does not steady the state, since it is delayed. The lifetime of an ion from the time TECHNICAL PHYSICS

Vol. 58

No. 10

2013

0.2

0 40

80

120

160

200 E, eV

Fig. 6. Energy spectra of beam electrons. p = (1) 0.12, (2) 2.2, (3) 2.8, (4) 3.6, (5) 5.2, and (6) 0.76 Pa; I = 2 A and U = 200 V.

of its appearance in the beam plasma to the escape from the emitting plasma toward the anode or grid equals several microseconds in the multiple charge exchange mode. Therefore, the instant of time when the ion gener ation rate in the beam plasma is maximal does not coin cide with the instant when the plasma potential reaches a maximum. As a result, the system with delayed feedback evolves into the selfoscillating mode [11]. In the highvoltage mode of generation of an elec tron beam several tens of kiloelectronvolts in energy, 8

0.8 6 0.6

ϕ−/U, ϕ~/U

0

5

6

0.4 4 2 3

0.2

f, kHz

Ia, Ig, mA

5

Ia2, mA

0.8

0.6

2 4 0

1 4

8

I, A

12

0 16

Fig. 7. Relative (1, 4) constant and (2, 5) variable compo nents of the probe’s floating potential and (3, 6) oscillation frequency vs. the discharge current. p = (1–3) 0.2 and (4–6) 2.6 Pa. U = 100 V.

1430

GAVRILOV, KAMENETSKIKH

“plasma” breakdown with growing pressure and emis sion current was observed [12]. This breakdown is explained by penetration of the plasma from the dis charge region into the accelerating gap. This, in turn, is the result of the growth of (i) the counter ion flux intensity, (ii) emitting plasma concentration, (iii) emis sion current, and (iv) electron beam gas ionization rate. All these factors cause avalanchelike growth of the emission current and breakdown of the gap [13]. In contrast to the lowvoltage electron beam gener ation considered in this work, a rise in the plasma potential in the highvoltage generation mode influ ences the accelerating voltage and counter ion flux insignificantly. However, in a gasdischarge system with electron stabilization of the discharge current, a rise in the emitting plasma concentration under the action of the counter ion flux may raise the electron extraction efficiency and also the emission current but only up to a value comparable to the discharge current. For the beam current to grow in an avalanchelike manner, additional electron emission from the elec trode surface is necessary. This may be provided by ini tiating a glow discharge in the anode cavity, through a rise in the plasma potential, with its subsequent trans formation into an arc. In the selfoscillating mode of generation, the elec tron beam current remains almost constant and the electron energy varies in a wide range comparable to the range of accelerating voltage with a frequency of several tens of kilohertz (depending mostly on pres sure). Such an effect was observed in [14], where the behavior of the collecting Langmuir probe under a positive potential was studied. When its potential was increased to a certain value depending on the pressure, the probe began to operate as an anode and oscillations at the same frequencies arose in the descending branch of the I–V characteristic provided that the current in the probe’s circuit was limited by a series resistor. The selfoscillating mode of electron beam genera tion in plasmacathode sources can be successfully used for plasma generation in applications requiring a pressure higher than that usually used in conventional thermioniccathode plasma generators. Examples are ionassisted plasma deposition of coatings, plasma chemical deposition of coatings (including diamond like and nanostructured ones), and nitridation in the electronbeam plasma. CONCLUSIONS The waveforms of the emitting plasma potential and beam current in a source with a grid plasma cath ode operating under the conditions of efficient electron extraction and intense counter ion flux generation show that when the accelerating voltage is not too high (100– 300 V), a rise in the gas pressure above 0.2 Pa causes the electron source to switch to the selfoscillating mode with highfrequency (104–105 Hz) electron energy

modulation (up to 100%) in the beam. The modulation depth of the beam varies within 0.2–0.3. The shift of the peak of the electron energy spec trum and its broadening after switching into the self oscillating mode, both measured with the retarding potential technique, support oscilloscopic measure ment data for the emitting plasma potential. The experiment aimed at enhancing the counter ion flux by increasing the plasma density in the beam drift space using a thermioniccathode nonself maintained discharge demonstrated that an increase in the counter ion flux, with the gas pressure and elec tron emission current kept constant, results in a pro portional rise in the plasma potential. This indicates that if the electron extraction is efficient, the plasma potential grows because the positive charge of the counter ion flux is difficult to compensate for. The plasma potential growth ends with switching the sys tem from the beam generation mode to the mode of discharge ignition at the collector. A controlled increase in the counter ion flux using an additional discharge decreases the frequency and amplitude of the plasma potential oscillations in the transient process. The selfoscillating mode arises because the plasma potential changes in opposition to the accelerating voltage. As a result, the growth of the plasma potential is attended by the decrease in the gas ionization rate and the counter ion flux. Feedback present in the sys tem cannot stabilize its parameters because the varia tion of the plasma potential (created by the space charge of the counter ion flux) lags behind the varia tion of the electronbeaminduced ion current. The selfoscillation frequency depends on the time it takes for the plasma potential to increase to a value at which the counter ion flux becomes insufficient to maintain the potential growth. The rate of change of the ion space charge in the plasma depends on the ion dynamics: it is not high, because the lifetime of ions from the instant they originate in the beam plasma to their escape toward the anode or the grid of the plasma emitter under the multiple charge exchange condi tions is long. ACKNOWLEDGMENTS This work was partially supported by the Program of Basic Research at the Presidium of the Russian Academy of Sciences (project no. 12P21046) and the Russian Foundation for Basic Research (project no. 120831142amol). REFERENCES 1. S. P. Bugaev, Yu. E. Kreindel’, and P. M. Shchanin, Electron Beams of Large Cross Section (Energoatomiz dat, Moscow, 1984). 2. N. N. Koval’, Yu. E. Kreindel’, and P. M. Shchanin, Sov. Phys. Tech. Phys. 28, 1133 (1983). TECHNICAL PHYSICS

Vol. 58

No. 10

2013

SELFOSCILLATING MODE OF ELECTRON BEAM GENERATION 3. A. V. Zharinov, Yu. A. Kovalenko, I. S. Roganov, and P. M. Tyuryukanov, Sov. Phys. Tech. Phys. 31, 39 (1986). 4. A. V. Zharinov, Yu. A. Kovalenko, I. S. Roganov, and P. M. Tyuryukanov, Sov. Phys. Tech. Phys. 31, 413 (1986). 5. Yu. A. Burachevskii, V. A. Burdovitsin, A. V. Mytnikov, and E. M. Oks, Tech. Phys. 46, 179 (2001). 6. N. V. Gavrilov and A. S. Kamenetskikh, Tech. Phys. 52, 301 (2007). 7. N. V. Gavrilov, D. R. Emlin, and A. S. Kamenetskikh, Tech. Phys. 53, 1308 (2008). 8. N. V. Gavrilov and A. I. Men’shakov, Prib. Tekh. Eksp., No. 5, 140 (2011). 9. Yu. P. Raizer, Gas Discharge Physics (Springer, Berlin, 1991).

TECHNICAL PHYSICS

Vol. 58

No. 10

2013

1431

10. N. V. Gavrilov, A. S. Kamenetskikh, and A. I. Men’sha kov, Plasma Emission Electronics (Buryatsk. Nauchn. Tsentr Sib. Otd. RAN, 2012), pp. 7–13. 11. V. V. Migulin, V. I. Medvedev, E. R. Mustel’, and V. N. Parygin, Basic Theory of Oscillations (Mir, Mos cow, 1983). 12. V. A. Burdovitsin, M. N. Kuzemchenko, and E. M. Oks, Tech. Phys. 47, 926 (2002). 13. V. A. Burdovitsin, A. K. Gorev, A. S. Klimov, A. A. Zenin, and E. M. Oks, Tech. Phys. 57, 1101 (2012). 14. B. N. Klyarfel’d, A. A. Timofeev, N. A. Neretina, and L. G. Guseva, Zh. Tekh. Fiz. 25, 1581 (1955).

Translated by V. Isaakyan