Experimental study of a plasma-filled backward wave oscillator

We present experimental studies of a plasma-filled X-band backward wave oscillator (BWO). Depending on the background gas pressure, microwave frequency upshifts of up to 1 GHz appeared along with an enhancement by a factor of 7 in the total microwave power emission. The bandwidth of the microwave emission increased from ≤0.5 GHz to 2 GHz when the BWO was working at the RF power enhancement pressure region. The RF power enhancement appeared over a much wider pressure range in a high beam current case (10-100 mT for 3 kA) as compared to a lower beam case (80-115 mT for 1.6 kA). The plasma-filled BWO has higher power output compared to the vacuum BWO over a broader region of magnetic guide field strength. Trivelpiece-Gould modes (T-G modes) are observed with frequencies up to the background plasma frequency in a plasma-filled BWO. Mode competition between the Trivelpiece-Gould modes and the X-band TM01 mode prevailed when the background plasma density was below 6×1011 cm-3. At a critical background plasma density of ncr≅8×1011 cm-3 power enhancement appeared in both X-band and the T-G modes, with mode collaboration. Power enhancement of the S-band in this mode collaboration region reached up to 8 dB. Electric fields measured by Stark-effect method were as high as 34 kV/cm while BWO power level was 80 MW. These electric fields lasted throughout the high power microwave pulse.


Introduction
Intense relativistic 'electron beam excitation of slow wave structures has been an active subject since the possibility was frrst conftrmed by Nation l in 1972. Many vacuum slow wave devices have been studied 2-6 since, with conversion efficiency of beam kinetic energy into microwaves as high as 30%. Using current pulse power technology, vacuum backward wave oscillators (BWOs) can emit microwave power as high as 1 GW 7 -9 . Injecting plasma into the slow wave structure10,11 enhances power emission by factors of 3 to 8. Introducing plasma into the BWO increased power output and efficiency. However, some of the basic mechanisms of the system are still not well understood. All previous plasma-filled BWO experiments concentrated on' producing higher power microwaves, higher efficiency and longer pulse. The goals of this work were to study the effect background plasma had on the microwave frequency and bandwidth of the output radiation in a high power plasma filled BWO; to study the plasma modes (Trivelpiece-Gould modes) in the plasma-filled BWO and their effect on the BWO waveguide mode (TMOI); and to measure the electric field in the plasma-filled BWO using light emission from the backgroun~plasma via the Stark shift.

Experimental Setup
In our experiment (Fig. 1), a Marx capacitor bank generates a 650 kV, 2 kA voltage pulse with 500 ns pulse duration. The electron beam was produced by field emission from a graphite cathode. The beam is annular with 1.8 em diameter and 2 mm thickness, and is injected into the BWO along a guiding magnetic field of 6-16 kG. Our plasma was produced by background helium ionization by the beam. The BWO is a cylindrical waveguide with a periodically varying wall radius, R(z), sinusoidally rippled about the mean radius, Ro, such that R(z)=Ro+ hcos<ko z), ko=2p/zO, where h=O.45 em is the ripple amplitude, zo=l.67 cm is the period and Ro=I.45 cm. An Xband hom placed 2 m away received the RF generated by the BWO. The RF was guided by waveguide to a screen room, where the RF was split and detected by an X-band crystal detector and an X-band 8-channel spectrometer which covered frequencies 8.2<f<12.4 GHz. Each channel covered about 500 MHz. There was an observation window on the 9th ripple of the BWO for the electric field measurement. Two optical lenses collected photons from the center of the BWO then focused the photons onto two optical fibers, which guided the light to two spectrometers in the screen room 12. One spectrometer "looked" at the helium allowed line A=501.56 nm and the other one cable, then were measured by a crystal detector, or by an 8 channel S-band (2.6 GHz<f<3.9 GHz) and J-band (5.85 GHz<f<8.2 GHz) microwave spectrometer. Each channel of the S-band spectrometer covered approximately 160 MHz bandwidth with the J-band channels covering approximately 294 MHz bandwidth. We measured frequencies between 3.9 GHz and 5.85 GHz with high pass filters, since a C-band spectrometer was not available to us.
Frequency resolved measurements below 2.6 GHz were not made due to a lack of diagnostic equipment. In the X, J and S-band spectrometers the ftIters bad 50 dB stopband insertion loss with 0.9 dB passband insertion loss. The plasma density was measured by a heterodyne microwave interferometer. In order to avoid strong X-band TM01 radiation and to not modify the oscillator, we replaced our BWe with a stainless steel tube of 10 mm radius, keeping anode and cathode geometry the same. Since the X-band radiation consistently arrived 140 ns (± 10 ns) into the beam pulse, we could correlate the microwave signal with the plasma density measurement at the turn on of the microwaves, this method likely underestimates the plasma density during the RF pulse, because of plasma production in the background helium due to the high RF field.

Experimental Results
Our vacuum BWO efficiency of converting beam kinetic energy into RF radiation was about 5%. Depending on the beam current, the RF power output was between 30 MW to 100 MW in vacuum with a frequency of 8.3 GHz.
The RF pulse duration was 50-150 ns depending on the RF power output level, As the RF power output increased the RF pulse duration decreased. As we added helium to the BWO, we observed a power increase, frequency upshift and bandwidth increase in the RF GHz, measured by X-band crystal detector) and to 8 dB. The I-band RF signal amplitude increased with the S-band the T-G modes (measured by a crystal detector) RF but no more than 3 dB. The X-band TMOI mode emission microwave signals (2 GHz<f<5.7 GHz).
increased for 6x10 11 cm-3 <Dp<Sx10 11 cm-3 and peaked at ncr' Power enhancement was typically a factor of 3 at nCf' but up to a factor of 6 in some shots. Given tlle error in the plasma density measurement, the background plasma density could be as high as n p -9.5x10 11 cm-3 (f p -S.8 GHz) when the TMOI mode and T-G modes power emission are enhanced. This could indicate the pOSSibility that part of the enhanced X-band signal was a result of T-G mode radiation. Although the absolute T-G modes power emission was not calibrated, we found that the power carried by the T-G modes emission (in I-band) was at least 27 dB less than the X-band lMOI mode. For plasma density n p >8xIO ll cm-3 , power emission in the 1MOI mode gradually decreased, and the T-G modes pulse became much shorter (-50 ns) but their amplitude kept increasing (both S and J-band). When the pressure reached 170 mT the J and S-band power emission was 2-3 times larger than that for 120 mT. (1) Using a spectroscopic method to measure the electric field distribution in a relatively bigb noise level system is very efficient and convenient. We choose the four energy-level system ( 3 1 P, 3 1 D, 2 1 P, 2 1 S) of belium I for the spectroscopic measurement. Transitions from 3 1 P to 2 1 S (AA=501.56 nm) and from 3 1 D to 2 1 P (A=667.80 nm) are allowed, and the transition from 3 I p to 2 I p is forbidden (lp663.20 nm) in the electric dipole approximation. In a perturbing electric field, energy levels 3 from shot to shot). This 34 kVfcm electric field is lower than we expected from the direct RF power measurement, since it gives a power flux lower than the measured X-band power of 80 MW± 10 MW. The electric field of the electron beam charge is not important here, since the plasma density was 8xl0 11 cm-3 and the beam density was 5xl0 11 cm-3 , so the beam was charge neutralized.
However, this calculation assumes a smooth tube with a 1MOI mode propagating axially with average electric field We increased the diode A-K gap reducing the beam current to -1 kA to get a longer microwave pulse. We counted photon numbers in each time interval for both the forbidden and allowed lines in each shot, then averaged over -100 shots. The ratio of the average photon numbers in the forbidden and allowed lines was used to calculate the electric field using equation (1). 5 shows the results of the electric field measurement in the plasma-filled BWO when the RF was enhanced by the background plasma by a factor of 2 over its the same as our measured fields. The real situation differs greatly from this, since the BWO RF electric field is strong near the wall. Our optical system focused in the BWO center when we measure E. Since we collect most of our photons from the center of the BWO our measured E field is much lower than the peak electric field. This may explain why the estimated microwave power is much lower than the measured RF power. To do an accurate power calculation requires knowing E(f) in the plasma ftlled BWO, a very complex calculation. (see Ref. 14) In conclusion, we measured the average electric field strength as a function of time on our BWO axis, where the relativistic electron beam, high power microwaves and plasma interact. While the microwave power output was enhanced by the background plasma, the electric field peaked at 34 kV/cm and lasted only as long as the high power RF pulse (-80 MW), about 50 ns. We measured the T-G modes with frequencies up to the background plasma frequency in our plasma-filled BWO. This could be the "dense spectrum" of plasma waves as discussed in Ref. 15. At a critical plasma density the T-G modes and the BWO TMOI mode output power were sinlultalleously enhanced.
The T-G mode measurement and the electric field measurement, to the best of our knowledge, have not been done previously for a device of the kind.
We thank Dr. K. Kato, David Sac and General Dynamics, Pomona Division, for the use of their equipment. This work was supported by AFOSR under contract # 90-0255.