научная статья по теме WHETHER ABNORMAL ENERGY ELECTRONS ARE BEING PRODUCED IN ELECTRIC DISCHARGES IN DENSE GASES? Физика

Текст научной статьи на тему «WHETHER ABNORMAL ENERGY ELECTRONS ARE BEING PRODUCED IN ELECTRIC DISCHARGES IN DENSE GASES?»

Pis'ma v ZhETF, vol. 101, iss. 11, pp. 830-835

© 2015 June 10

Whether abnormal energy electrons are being produced in electric

discharges in dense gases?

L. P. Babich1\ T. V. Loiko

Russian Federal Nuclear Center, 607188 Sarov, Russia Submitted 27 February 2015

Reviewing results of experimental research of picosecond pulses of runaway electrons (REs) generated by discharges in dense gases at multiple overvoltages, including, along with routine measurements of voltage pulses and RE current, direct measurements of RE energy distributions, pressure dependence of RE numbers and experiment with retarding voltage similar to the accelerating voltage, a reality of the effect of "abnormal energy" REs is being substantiated. With this goal we emphasize non-conventional qualitative RE characteristics rather than quantitative.

DOI: 10.7868/S0370274X1511003X

I. Introduction. High-energy runaway electrons (REs) produced by discharges in dense gases at high overvoltages relative to the static (direct current) self-breakdown voltage and observed for the first time in the end of 1960s, are being studied further during the next decades in various configurations (cf. for instance [1-8] and citations therein). In the first experiments discharges in air at P = 1 atm were discovered to generate REs with the energy e exceeding a magnitude eUmax corresponding to the maximum value of the voltage pulse Umax [1]; obviously, this is impossible in the framework of the ordinary linear acceleration. Initially the energy e was estimated using electron attenuation curves measured by wedges technique at the anode outlet. In air at P = 1 atm rather frequently the curves were as those typical for monoenergetic electrons: a small initial section with a weak inclination, rather long decreasing linear section and final straggling are inherent for the curves [1-3]. The linear section allows estimating the initial electron energy e using the extrapolated range i?ex(t) [9-11], the energy being somewhat underestimated in the case of wide electron beams, that is the case for REs. One of such curves fitted to ten measured points is illustrated in Fig. 1 [2,3] along with attenuation curves of monoenergetic electrons by Seliger [11]. The RE energy given by this curve is e « 270 keV > eUmax. Note Umax is significantly less than a maximum voltage magnitude [/¡die in the idle-running mode, which can be doubled relative to amplitude of a self-breakdown voltage ?7seif of a shaping switch in a used generator. Thus, quantitatively (e > eUmax), but, what it is more important, qualitatively (almost monoenergetic) the REs of

-'-'e-mail: ba.bich@elph.vniief.ru

Fig. 1. Attenuation curves (aluminum foils) [2, 3]: 1 - curve of REs generated by the discharge (Unie = 270 kV, air, 1 atm, d = 15 mm, rcath = 6 mm); 2 and 3- monoenergetic electron curves by Seliger [11]

"abnormal energy" revealed themselves. However, a reality of this effect is denied [5]. In this paper we review available experimental data arguing in favor of its reality. To prove the effect we emphasize, first of all, qualitative RE characteristics, not quantitative. Such approach allows making the term "abnormal energy" less misleading.

II. Abnormal energy and voltage pulses. In

connection with a negation of the "abnormal energy" REs generation [5], based on that measurements of highvoltage pulses with subnanosecond front are unreliable, it is pertinent to note that voltage pulses were being measured close to the cathode working surface [3]. We

cannot agree with "that the idle-running mode is a typical regime at the cathode in the pre-breakdown stage in the gap" [5]. The amplitude [/¡die is achieved during a discharge in sufficiently long gaps, but in gaps with rather small interelectrode spacing d [1-3] (cf., for example, Table) Umax < ?7;die or even <C tAdie, because

Dependences on the interelectrode spacing d of the discharge voltage amplitude I/mcharacteristic energy Sm, and energy excess Ae = em — el/max of "abnormal energy" REs in air at 1 atm. Amplitude of the voltage pulse in the idle-running mode I/jdie = 270 kV, d = 2 cm, conic cathode with rcath = 200 |im, grid anode

d, cm 3.5 2 1 0.5

CU, kV 210 190 150 130

£m, keV 320 290 260 180

Ae, keV 110 100 110 50

the conductivity current in the primary channel in the near-cathode domain, which is closed to the anode by the eddy current in front of the channel along with a current of the secondary avalanches initiated by the RE pulse, increases so fast that the [/¡die is not achieved.

The term "abnormal energy" has been introduced comparing the RE energy t, given by i?ex(t), to the Umax magnitude, the accuracy of which raises well-founded doubts from the very beginning of the research of discharges with REs. More intriguing and significant are discovered later qualitative characteristics of these REs, which can not be understood in the framework of the ordinary linear acceleration. We must forewarn against interpreting the straggling [9-11] as high-energy "tails" of electron energy distribution. The straggling "tail" appears even in the attenuation curves of initially monoenergetic electrons, such as measured by Seliger [11] (cf. Fig. 1). Obviously electrons with the energy above the initial magnitude can not appear in such experiments. The authors of the paper [8] interpret a long "tail" of attenuation curves, which they observed in the range of the most thick absorber layers, as the "abnormal energy" REs. Most likely, for the "tail" the straggling is responsible.

III. RE energy distribution (magnetic spectrometry). We measured RE energy distributions using the magnetic spectrometry technique with a resolution better than 10% [2,3]. The attenuation in substance layers outside the evacuated spectrometer chamber limited measurements from below to the energy of s « 50 keV. In Fig. 2 the RE energy distributions for one of gas-discharge configurations are illustrated. The energy losses reduce the highest electron energy tmax. With the pressure decrease the tmax decreases (cf. upper

100 200 p

n n max

e (keV)

Fig. 2. RE energy distributions (arbitrary units) [2,3]: Uidie = 270 kV, air, 1 atm, d = 2 cm, sharp conic cathode with Teat.h = 200 /L/.m, grid anode

and middle panels Fig. 2), but then increases (Fig. 2c) according to the Umax(P) variation with the pressure. Above approximately 200Torr the distributions have expressed maximum, a position of which em increases with the pressure rise. Under conditions of the experiment upper panel in Fig. 2 at P = 1 atm. em « 270 keV. Under the same conditions the attenuation curve gives s fa 300 keV. The measured RE distribution width is of 2Aemeas « 60keV (upper panel in Fig. 2). The intrinsic width of the distribution 2At;ntr in each separate discharge is much narrower because of the widening due to the scatter of the self-breakdown voltage of the shaping switch Use\{ and the voltage Umax at the studied gap, electron scattering in the spectrometer window and due to non-ideal collimation. Both em > eUmax and extremely narrow energy spectrum (2Ae;ntr £m) agree with the attenuation curves (Fig. 1). With the discharges in sufficiently long gaps (d > 1 cm, volumetric glow) the excess of the RE energy At above the "allowed" magnitude eUmax is of At « 100—llOkeV; at d < 1 cm (contracted channel) the At is much less (cf. Table).

In the experiment in [5] with "the acceleration pulse at the cathode with amplitude of - 570 kV" and using "the cut-off threshold" of the attenuation curve in Fig. 17 [5] the electron energy was estimated to be of t « (160—170) keV. Estimating e magnitude by "the cut-off threshold" is not reliable because in the straggling domain fluctuations are comparable to electron numbers themselves and the straggling and, hence, "the

832

L. P. Babich, T. V. Loiko

cut-off threshold" are dependent on the initial numbers of electrons at the outlet of gas-discharge gap; nevertheless, we estimated approximately the same e magnitude using the i?ex(t) for the curve in Fig. 17 [5]. These estimations do not agree with the data in Figs. 1 and 2, according to which the e magnitude is much bigger, though voltages were significantly less. Taking into account that a dependence of the RE current on the voltage after approximately 250 kV achieves a plateau of ~ 1A (Fig. 15

[5]), it is doubtful if the "accelerating" voltage above 250 kV was being achieved with a sharp cathode edge. According to Fig. f 5 in [5] the RE current vanished completely at the voltage of ~70kV, whereas in our work

[6], in which microsecond voltage pulses 44-66 kV were used, the RE numbers of Ne « fO7 — fO8 per pulse were detected. Note, to the RE current of ~ 1A with a duration of ~ 50 ps in [5] the RE number of Ne « 3 • f 08 per pulse corresponds; this magnitude is not essentially different from Ne « 9 • fO8 f/pulse in Fig. f.

IV. The RE pulse is generated at the conductivity current front. It is commonly believed that the REs are generated during rise-time of a voltage pulse with extremely steep front. The REs can, though not obligatory, be generated at the voltage front, but, being charge carriers and initiators of secondary electron avalanches, the REs are generated self-consistently with the conductivity current rise, which can sharply grow at the front of the applied voltage pulse, thus limiting amplitude of a really achieved voltage amplitude. To demonstrate this rather obvious statement we detained the ionization development such that the delay time idei of the conductivity current to significantly exceed a duration tu of the front of the voltage pulse U(t). This was reached using the barrier discharge. Oscilloscope traces of voltage U(t), total gas-discharge current I(t) and RE current are illustrated in Fig. 3 [2,3]. The peak in the left part of I(t) in Fig. 3b is the eddy current due to the interelectrode capacity charging. The tdel value is of 4 ns. In view of that RE pulse onset coincides with that of I(t) and

Для дальнейшего прочтения статьи необходимо приобрести полный текст. Статьи высылаются в формате PDF на указанную при оплате почту. Время доставки составляет менее 10 минут. Стоимость одной статьи — 150 рублей.

Показать целиком