230

Nuclear Instruments and Methods in Physics Research A285 (1989) 230-232North-Holland, Amsterdam

A 35 GHz CYCLOTRON AUTORESONANCE MASER AMPLIFIER

G. BEKEFI, A. DIRIENZO, C. LEIBOVITCH and B.G . DANLYDepartment of Physics, Research Laboratory of Electronics and Plasma Fusion Center, Massachusetts Institute of Technology, Cambridge,Massachusetts 02139, USA

Studies of a cyclotron autoresonance maser (CARM) are presented. The measurements are carried out at a frequency of 35 GHzusing a rruldly relativistic electron beam (1 .5 MeV, 260 A) generated by a field emission electron gun followed by an emittanceselector that removes the outer, hot electrons Perpendicular energy is imparted to the electrons by means of a bifilar helical wiggler.Superradrant measurements give a small signal gain of 90 dB/m . Computer simulations are also presented.

1. Introduction

The cyclotron autoresonance maser (CARM) hasbeen subjected to extensive theoretical [1,2] studies andnumerical simulations [3-5]. However, unlike thegyrotron and the free electron laser, its capabilities as asource of coherent millimeter wavelength radiation re-main virtually untested in the laboratory . To the best ofour knowledge, only CARM oscillator experiments [6,7]have been reported in the literature . We present herewhat we believe to be the first, albeit preliminary,studies of a CARM amplifier.

The emission from a CARM results through aninteraction between the Doppler upshifted cyclotronwave on the electron beam

ca = Qo/Y + k l l o l l

and an electromagnetic waveguide mode

ca t = k2C2 + tae .II

C'

co and kll are the frequency and axial wavenumberrespectively; Qo= eB ll/m is the nonrelativistic electroncyclotron frequency associated with an axial guide mag-netic field BII ; y = [1 -

(oil/c)2- (V1/c)2]-1/2 is the

relativistic energy factor ; and w. is the cutoff frequencyof the waveguide mode in question . Maximum gain ofthe CARM instability occurs near phase velocity syn-chronism of the two waves. This yields the radiationfrequency:

) 211/2~

12 07 11

Here ßll = oil/c, YII =(1 -

'8112 ) -1/2, and the positive sign

refers to the sought after Doppler upshifted CARMmode of operation.

0168-9002/89/$03 .50 © Elsevier Science Publishers B.V.(North-Holland Physics Publishing Division)

2. Experiment

A schematic of the CARM amplifier is shown to fig .1. The accelerator potential is supplied by a Marxgenerator (Physics International Pulserad 110 A) with amaximum capability of 1.5 MV and 20 kA . The electronbeam is generated by a field emission gun composed ofa hemispherical graphite cathode and conical anode,which also acts as an emittance selector . The entire 2 mlong system is immersed in a uniform magnetic field of7 kG .

The 260 A, 1.5 MV beam that issues from theemittance selector has a radius of 0.254 cm and a

TO

WAVE LAUNCHER

BIFILAR HELICAL

CYLINDRICALWIGGLER

DRIFT TUBE

GRAPHITECATHODE

GRAPHITE ANODEAND

EMITTANCESELECTOR

_J RFTUNER

(b)

RF INPUTPORTI

"-RF POWER

DRIFT TUBE

RF INPUTPORT 2

Fig 1. Schematic of the CARM amplifier experiment.

measured [81 normalized beam brightness equal to 2.4 x104 A cm -2 rad-z . This corresponds to a normalizedrms emittance of 4.9 x 10-2 cm rad and an RMS energyspread Dyll/yll = 0019. We note that considerably higherbeam brightness is achieved by sacrificing current. Whenthe radius of the emittance selector is reduced to 0.076cm, the current drops to 8.4 A, but the brightnessincreases to 9.5 x 104 A cm-2 rad -z . The correspond-ing emittance is now 4.5 x 10-3 and the energy spread

Ayll/yl1 = 00017.The aforementioned 260 A electron beam is injected

into a bifilar helical wiggler which imparts perpendicu-lar energy to the electrons. The wiggler has a periodicityof 7 cm and is six periods long . Within the first fourperiods, the wiggler magnetic field increases slowly andthereby provides an adiabatic input for the electronbeam; the last two periods provide a uniform wigglerfield with an amplitude on axis equal to 525 G. Theresulting transverse electron velocity (v_,_ =0.3v,1) isestimated from witness plate observations of the beamdimensions .

The downstream end of the wiggler is terminatedabruptly by means of a metal shorting ring and thespinning electrons are allowed to drift into the 86cm-long CARM interaction region where they are sub-jected to the uniform axial magnetic field only. We notethat as a result of the wiggler excitation and the abruptwiggler termination, the energy spread Ay11/yll of theelectrons entering the CARM region can be consider-

0 0

~ 250zw

Sa

Z

z0aâ 'L77j-

0 40 80 120 160 200TIME Ins)

Fig. 2 . Oscilloscope traces of the electron beam voltage, currentand the radiation intensity .

G Bekefi et al / A 35 GHz CARM amplifier

F--z

W

âW

r

zrzz0

âa

ici,

162

IO-3

10"4

10 -5

231

lo ft0 20 40 60 80 100

INTERACTION LENGTH Z (cm)

Fig. 3 Radiation intensity at a frequency of 35 GHz, as afunction of the CARM interaction length .

ably worse than the energy spreads quoted above inreference to electrons leaving the emittance selector .

The - 2 m long, 0.787 cm radius evacuated drifttube acts as a cylindrical waveguide whose fundamentalTEtt mode has a cutoff frequency w,/2,Tr = 11 .16 GHz.Fig. 2 illustrates the time history of the voltage, currentand radiation characteristics of the device . The CARMhas been operated in the superradiant mode, in whichthe signal is allowed to grow out of background rf noise.

At the output end of the CARM, a mica windowtransmits the circularly polarized radiation generated inthe drift tube, where it is measured by means of stan-dard calibrated crystal detectors, and its spectrum isanalyzed by means of a 98-m-long dispersive line. Twospectral features are observed having differing tuningcharacteristics with axial magnetic field BII . Thefrequency of the stronger spectral line increases linearlywith Bl1 as predicted by eq . (1) . The weaker line israther insensitive to B11 . We believe that this line repre-sents the second harmonic of the Doppler downshiftedbranch (see eq . (3)) .

In order to determine the growth rate of the wave,the output intensity is measured as a function of thelength of the interaction region . This is accomplished bymeans of an axially movable horse-shoe "kicker" mag-net that deflects the electron beam into the waveguidewall at any desired position z, thereby terminating theinteraction at that point. Fig. 3 shows how the rf poweroutput measured at the far downstream end varies withthe "kicker" magnet position z. The slope of the curveyields a single pass gain of 0.9 dB/cm, a value that isconsistent with computer simulations . In fig. 4 the pre-dicted linear growth rate F and the efficiency at satura-

VII. UNCONVENTIONAL FEL SCHEMES

23 2

ö

UZWU

ww

0

G. Bekefi et al / A 35 GHz CARM amplifier

20Em

15 ,,w

0 t30

05ô

a0a

0005 010 015

ENERGY SPREAD (or,/y )Fig . 4 . Computed saturation efficiency and linear growth rate

as a function of the electron beam energy spread .

tion are plotted as a function of the relative parallelenergy spread . The results were obtained from a two-dimensional, self-consistent CARM amplifier code [5] .The experimentally measured growth rate is consistentwith theory for a beam energy spread Ayl1/yl1 = 0.1 . Inthe superradiant mode the amplifier does not reachsaturation, so that the observed rf output power is low(- 100 kW).

In conclusion, we have reported what we believe arethe first CARM amplifier measurements . The super-radiant measurements described here are being followedup by injection of a high power (60 kW) monochro-matic, wave from a 35 GHz magnetron driver (see fig.lb). Initial measurements yield a saturated power out-put of 10 MW, corresponding to an electronic efficiencyof approximately 3% . We note that the Doppler down-shifted mode can become absolutely unstable [9] andthereby cause serious deterioration of the CARM

amplifier. However, calculations show [9] that this is ofno concern in our parameter regime, since we operatewell below the critical current (4 .2 kA) and the criticalmagnetic field (B 1l = 11 .4 kG) for onset of the absoluteinstability .

Acknowledgements

This work was supported by the Air Force Office ofScientific Research and the Innovative Science andTechnology Office of the Strategic Defence InitiativeOrganization. It was aided greatly by equipment onloan from the Lawrence Livermore National Labora-tory .

References

[31

[41[51

[61

[71

[81

[91

A.W . Fliflet, Int . J . Electron . 61 (1986) 1049 .V.L . Bratman, N S. Ginzburg, G.S . Nusmovich, M.I . Pete-lin and P S. Strelkov, Int J. Electron. 51 (1981) 541.K.D . Pendergast, B.G . Danly, R.J Temkin and J.S . Wurtele,IEEE Trans Plasma Sci., 16 (2) (1988) 122-128.A.T. Lin, Int . J Electron . 57 (1984) 1097 .B.G . Danly, K.D. Pendergast, R.J . Temkin and J.A . Davies,Proc . SPIE 873 (1988) 143-147.I.E. Botvmnik, V.L . Bratman, A.B . Volkov, N.S . Ginzburg,C.G . Demsov, B.D . Kol'chugn, M.M . Ofitserov and M.I .Petelin, Pis'ma Zh . Eksp . Teor . Fiz. 35 (1982) 418.I.E . Botvinnik, V.L . Bratman, A.B. Volkov, G.G . Demsov,B.D . Kol'chugm and M.M. Ofitserov, Pis'ma Zh. Eksp .Teor. Fiz . 8 (1982) 1376 .The technique is described by D. Prosnitz and E.T . Scharle-mann, Lawrence Livemore National Laboratory . ATA NoteNo . 229 (1984) .J.A. Davies, MIT Plasma Fusion Report. No . JA-88-29(1988) .