720 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 39. NO. 3, MARCH 1992

Design and Testing of an Electron Gun Producing a Segmented Sheet Beam for a Quasi-Optical

Gyrotron Michael E. Read, Member, IEEE, Alan J. Dudas, John J. Petillo, and M . Q. Tran

Abstract-A high-power sheet beam electron gun for a quasi- optical gyrotron has been developed. The gun produces two beams with rectangular cross sections, with one beam seg- mented into 11 beamlets. These beamlets are spaced so as to be in line with the maxima of the RF field profile in the resonator of a 120-GHz quasi-optical gyrotron. The maximum current in the segmented beam is nominally 34 A, while the that for the unsegmented beam is 68 A. The gun has been tested, and has operated essentially as predicted by 2D and 3D theory. In par- ticular, the overall beam shape is largely undistorted, and the beamlets are, for most cases, well aligned.

I. INTRODUCTION HE QUASI-OPTICAL (QO) gyrotron is a promising T source of high-power millimeter-wave radiation [I]-

[4]. With optimized conditions, efficiencies of over 40% are predicted, and stable, single-mode operation should be possible over a wide parameter range [2].

While the physics of the QO gyrotron is essentially the same as that of the conventional (microwave-cavity) gy- rotron, the QO gyrotron has several advantages, most of which are by virtue of the type of resonator. In the QO gyrotron, the resonator is formed by two mirrors, similar to those used in high-power lasers. The geometry is shown in Fig. 1 . The area of the mirror can be much larger than that of the cavity walls in a microwave-cavity gyrotron, and thus the ohmically dissipated power density can be much lower for a given power. Output powers of several megawatts CW at 300 GHz are thus expected to be pos- sible.

An example design of a 1-MW, 300-GHz QO gyrotron is given in Table I.

To date, QO gyrotrons have been constructed using cy- lindrically symmetric magnetron injection electron guns (MIG’s) [3]-[5]. This type of electron gun is a holdover

Manuscript received September 4, 1990; revised June 11, 1991. This work was sponsored by the U.S. Department of Energy under Contract DE- AC05-87ER80523 and by the Ecble Polytechnique Federal de Lausanne. The review of this paper was arranged by Associate Editor R. J . Temkin.

M. E. Read and A. J . Dudas are with Physical Sciences, Inc., Alexan- dria, VA 22124.

J . J . Petillo is with Science Applications International Corporation, McLean, VA.

M. Q. Tran is with the Centre de Recherches en Physique des Plasmas, Association Euratom-Confederation Suisse,Ecole Polytechnique Federal de Lausanne, Lausanne, Switzerland.

IEEE I.og Number 9104695.

OUTPUT ‘I MIRROR

i I

Fig. 1. Basic geometry of the quasi-optical gyrotron

TABLE I EXAMPLE DESIGN FOR A I-MW, 300-GHz

QO GYROTRON

Beam Voltage 80 kV Current 34 A

Resonator Minimum waist 0.5 cm Mirrors

separation 90 cm diameter 3 cm

ohmic loss (peak) ohmic loss (total) 20 kW

output coupling 2 % 1 .5 kW/cm*

from more traditional cylindrical-cavity gyrotrons. While this geometry is well-suited to a microwave-cavity gyro- tron, it is not at all optimum for the QO gyrotron, where the cross section of the resonator, as viewed by the beam, is roughly rectangular. For good efficiency the height of this cross section (i.e., twice the radiation beam waist) must be on the order of 10 wavelengths. In order for the RF field to be essentially uniform (within 90% of the peak value) over the electron beam, the diameter of an annular or circular beam must be less than 3.2 wavelengths. This limitation is significant for megawatt operation at wave-

0018-9383/92$03.00 0 1992 IEEE

READ er al. : ELECTRON GUN PRODUCING A SEGMENTED SHEET BEAM 72 1

lengths approaching 1 mm. For example, for a 1-MW, 300-GHz gyrotron, even a solid circular beam would have a current density of approximately 500 A/cm2, a value substantially larger than the 300 A/cm2 achieved in even high-current-density microwave-cav-ity gyrotrons, and one for which the cathode current density and/or the magnetic compression would be prohibitively high. In addition, a solid circular beam cannot be generated using the usual magnetron injection gun (MIG). A sheet beam can be much wider than high, and thus can have a much larger area than a circular beam. Further, as is shown below, it can be generated by a MIG.

Additionally, a sheet beam can be segmented into “beamlets,” such that each beamlet will pass through the resonator at the position of a peak of the standing electric field pattern. This will increase the efficiency over that achieved with an annular or uniform sheet beam because little or no beam current will pass through the standing wave in the resonator at a position where the electric field is zero. It also reduces the mode density by one-half, since the n f 1 modes will have nulls at the center of the beam- lets.

In this paper, we describe the design and testing of an electron gun to produce a sheet electron beam. The gun was designed for use on a 120-GHz QO gyrotron, but the principles of design are relevant to other frequencies. We discuss 2D and 3D simulations as well as measurements of the cross section. From the measurements, we con- clude that the gun produces a beam which should nearly optimize the efficiency of a gyrotron. With the segmented beam, which should produce the optimum efficiency, the beam power will be 2.7 MW. With somewhat reduced efficiency, both segmented and unsegmented beams can be combined for an electron beam power of 8.2 MW.

11. ELECTRON GUN DESIGN The basic geometry of the gun is shown in Fig. 2. The

parameters are given in Table 11. The gun was designed to produce two beams, one from each side of the cathode. One of the beams was segmented into 11 beamlets, while the other was unsegmented. This should allow for a clear test of the efficiency enhancement, in that the beam cur- rents are individually controllable. After full compres- sion, each beam was 1.3 cm wide and 0.2 cm high, with a maximum distance from the axis of 0.42 cm. This is one-half of the 3.2 wavelengths (discussed above) for a 120-GHz device. The current for this gun (using both beams) would be 102 A, and the power, 8.2 MW.

In the design, a single anode was used. This is in con- trast to most other gyrotron MIG’s, in which there are two anodes. The use of only one anode significantly reduced the mechanical complexity of the gun, and did not com- promise the design.

The gun was modeled using two computer codes. For calculation of the velocity distribution, where high reso- lution was required, and edge effects were of lesser im- portance, a 2D modeling with a conventional trajectory code. was found to be optimum. However, prior to this

0

TOP V I E W

Fig. 2 . Geometry of the sheet beam electron gun.

TABLE I1 PARAMETERS OF THE PSI SHEET BEAM ELECTRON GUN

Voltage 80 kV

Current segmented unsegmented both

34 A 68 A 102 A

Magnetic field compression 38

Cathode width 78 .8 mm current density

width/spacing 0.63/1.25 mm height 2 mm current density 270 A/cm2 (Y 1.5

3.6 A/cm2

Beamlet (fully compressed)

A V 1 /U1 5 %

study, there was concern that the beam would be substan- tially distorted due to E X B drifts. These drifts were modeled using a 3D code.

The beam CY (E zll /U,,) and velocity spread were cal- culated using the 2D simulation code EGN [6 ] . This was run in the Cartesian geometry mode, where the beam was modeled in the x-z plane and was infinitely “wide” in the y direction. The cylindrical magnetic field information was input as an axial array. The code determines the x com- ponent of the magnetic field over all space by an expan- sion. This produces a good modeling in the critical gun region, where the magnetic field is essentially constant along the axis. However, in the drift region, where the magnetic field increases, the modeling is not strictly cor- rect. Conservation of canonical angular momentum re- quires that, in the Cartesian geometry, the guiding centers of the beam electrons compress as 1 /Bz, while in a cylin- drical geometry, the compression goes as l / B p . This clearly gives an incorrect modeling of the beam trajecto- ries for our problem. However, the beam current density is correctly modeled, and thus space-charge effects can be observed. In addition, based on earlier experience with cylindrical beams, the trajectories of the guiding center radii can be quite accurately determined using analytic calculations based on adiabatic compression. The incor-

722 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 39, NO. 3, MARCH 1992

_ _ _ _ _ - - - - -

rect modeling will have a very small effect on the velocity spread, since the compression is adiabatic. Finally, the 2D modeling cannot take into account changes in the magnetic field across the cathode (in y ) . However, this variation in B is less than 0.5%. This will have a negli- gible effect on the velocity spread.

Some of the results of the EGN simulations are shown in Fig. 3. The figure shows the normalized spread AuL / U , as a function of axial position along the beam. (As is usually the practice, the spread is a “half-width.”) Due to the relatively low current density, the 38: 1 area compression of the beam was adiabatic, i.e., the perpen- dicular velocity spread remained essentially constant. (In earlier designs, with a cathode current density of 10 A/cm2, an increase in the perpendicular velocity spread with magnetic field was observed, consistent with other reported experience [7].) The initial cooling of the beam is not well understood, but was observed in most of the designs modeled.

The predicted velocity spread is expected to be low enough to avoid either degradation in the gyrotron effi- ciency or significant beam mirroring, and is similar to the velocity spread predicted for the “MIT” megawatt gun (Varian number VUW-8144) [8].

The results of the velocity spread calculations were gratifying, although not surprising, since the current den- sity and compression are similar to existing MIG guns with circular symmetry [ 7 ] . Of more concern was the de- gree of distortion that would be caused by E x B drifts. For ease of modeling, these were separated into two types. Drifts in they direction due to E, and Ez in the diode could be calculated using EGN. Those involving E,,, and drifts in the compression region had to be modeled using a 3D trajectory code.

The drifts in the y direction caused by E, X B, and Ez X B,, as calculated using EGN, are shown in Fig. 4. There are two features of the motion of interest. The beam had an average drift of approximately 1.8 cm. Of course, the beams from the two sides of the cathode moved in opposite directions. This had to be accounted for in the mechanical design of the tube, which meant making the tube adequately wide. In addition, the portion of the beam originating from the back of the cathode drifted signifi- cantly farther than that from the front, resulting in a skew- ing of the beam. This is important if the beam is seg- mented into beamlets for efficiency enhancement, as the beamlets could be tilted such that they overlap and thus become ineffective. As can be seen from Fig 4, the EGN simulations indicated a 1.2-cm skewing. This was an un- acceptable amount because the beamlet separation at the cathode was only 0.75 cm. However, this problem was corrected by slanting the emitters in the opposite direc- tion, as shown in Fig. 5 .

True 3D modeling [9] has been accomplished using ARGUS, a 3D Poisson solver/particle trajectory code. [IO]. As with the 2D simulation, symmetry is assumed with respect to the y-z plane. Modeling simply the gun region was not a difficult task for the code (on a

22 -

-20 - E E ’8- Y

3 1 6 -

3 14-

_ - ,’

_ - - - -a .e

/’

, I

c

a

Percent spread in V,, Axial magnetic field Percent spread in V,,

field allowing gate

I

, for ’ valve ,’ , ,

I

, i.

Fig. 4 . E X B drifts in the y direction, as calculated using EGN. Drifts for the rays originating from the top and bottom of the cathode are shown. The both rays start at y = 0.

Fig. 5 . Cathode with emission surface skewed to counter unequal drifts.

CRAY-2 computer), however, modeling the beam through compression proved to be quite formidable. The difficulty was due to the severe restriction on the size of the time step imposed by the magnetic field, so that cyclotron or- bits would be properly resolved temporally. For proper field resolution, 46 cells, 351 cells, and 65 cells were used in the x, z, and y directions, respectively. This simulation could only fit on a CRAY-2 without resorting to domain decompression. Even then, it required approximately 28

READ et al.: ELECTRON GUN PRODUCING A SEGMENTED SHEET BEAM

t . 1 4

t . . . . l . . . . i ~ 1 . 0 0 0. 1 . 0 0

- Y . 0 0

- 3 . 0 0

- 2 . 0 0

: 1 . 0 0

m~ 0 . 1 . 1 . . . . 1 . . . . 1 . . . . 1 . i

0 . 2 . 0 0 q . 0 0 6 . 0 0

Fig. 6. (a) Top view b - z plane) o f the ARGUS simulation. The beam current was 17 A . (b) Side view of (x-z plane) of the ARGUS simulation.

million words, which was 37% of the available memory of the F machine CRAY-2 at NERSC.

For obvious reasons, the E X B drifts were the focus of the simulations. For this purpose, the beamlets were each represented by three rays. There were 11 beamlets, as in the actual device.

Results of simulation of the full system are shown in Fig. 6. The simulation covered the region from the diode to the point at which the beam was nearly full com- pressed, a distance of approximately 34 cm. The current for this simulation was 17 A. (A planned simulation at full current was not possible due to limitations on com- puter time.)

Fig. 6(a) is a top view of the beam. Fig. 6(b) shows a side (x-z plane) view. The beam compression is clearly shown.

As can be seen in Fig. 6(a), the beamlets appear to re- main distinct (nonoverlapping) over the entire region. This is shown more clearly in Fig. 7, which shows a cross sec- tion of the beam, at z = 34 cm. Here, some distortion of the beam is evident, but the beamlets are distinct enough to selectively interact with the standing wave in the res- onator.

Through trial and error, it was found that the slant an- gle of the emission surfaces that was required to properly align the beamlets at full compression was 18". This was

123

Fig. 7 . View o f the beam cross section at full compression. Each cross indicates the position o f a ray, with the extent of the cross indicating the uncertainty in a representative particle orbit's gyro-center.

only 70% of the tilt indicated by EGN. This was puz- zling, given that additional skewing was expected to oc- cur in the compression section due to unequal' electric fields on the top and bottom surfaces of the beam. (A rough calculation of those drifts indicates that, at a cur- rent of 34 A, the skewing from the drift section would be approximately equal to that from the diode. At 17 A, it would be 50%.) Thus EGN calculations, which did not include drifts in the compression section, should have in- dicated a smaller correction than did ARGUS. We found no way to resolve this discrepancy, other than by fabri- cating and testing the gun. The number finally chosen for the cathode was equal to that indicated by ARGUS, plus 50% to account for the additional drifts due to operation at the full, 34-A, current.

As can be seen in Fig. 7 , the drift of the sides of the beam in the x direction are very small. At full current, rough calculations indicate that the movement of the edges would be approximately 0.9 mm, which would not be a serious distortion of the beam.

The velocity spread can be calculated from the results of the 3D calculations. However, the results will be highly inaccurate, due to the fact that only three rays per beamlet were used. Since two of the three rays are (by definition) at the edges, and will experience relatively large forces from the self-electric field, the velocity spread calculated by the 3D code is expected to be larger than in the actual beam. The calculated spread in U I is 9 % .

111. RESULTS OF TESTING The electron gun has been tested, using the apparatus

shown in Fig. 8. Power was supplied to the gun using a modulator with a 2-ps pulse, while the magnetic field was produced by a superconducting solenoid. In the tests, the beam cross section was measured by observing the X-rays produced by the beam striking a thin tantalum witness plate, as shown in Fig. 9. Tantalum was chosen because of its relatively high efficiency in generating X-rays and its high melting temperature. The high melting tempera- ture is important because of the high power density ( - 7 MW/cm2) in the beam when fully compressed. Despite the use of this material, the measurement could be made at full compression with currents only up to 4 A. Higher currents produced strong optical radiation from the target,

124 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 39, NO. 3. MARCH 1992

CRYOCENIC nRCNET OEURR r--l

Fig. 8 . Apparatus for measurement of the cross section of the sheet beam

BERM 10 cn. SCRRPPER

BERM EDGES

COPPER RNO C E R R n l C D R I F T S E C T I O N

V - S I O E

Fig. 9. Details of the beam drift section. showing the beam cross section at various positions

1 C t l . H

Fig. 10. X-ray photograph of both segmented and unsegmented beams, 37 cm from the cathode, with a current of 10 A

indicating that the temperature was at or above = 1000°C. Examination of the witness plates after exposure to the beam showed substantial melting.

The X-rays were observed using photographic film, with imaging accomplished by use of a pinhole camera. A 0.3-mm-diameter by 3.2-mm-thick lead pinhole was used, giving a resolution of approximately 0.6 mm with a 1 : 1 magnification. Polaroid black and white film, with an ASA speed of 3000 was used in conjunction with a rare-earth image intensifying screen.

Cross sections were taken at 27, 37, and 47 cm from the cathode, where the axial magnetic field was 3.75, 26, and 50 kG, respectively. (The correlation with axial po-

sition and magnetic field for the experiment was different from that in the 3D simulations due to the addition of a gate valve in the actual apparatus.) A photograph of the cross section with both (segmented and unsegmented) beams on is shown in Fig. 10. As can be seen, the seg- mentation of the upper beam is quite distinct. Part of the beam was intercepted by the analyzer, which was limited in diameter by the relatively small bore of the supercon- ducting magnet. Thus only 5-6 beamlets of the segmented beam and approximately one-half of the unsegmented beam can be seen. This is shown schematically in Fig. 9.

It was found that the cross section of the segmented beam was essentially independent of the presence of the

READ et al.: ELECTRON GUN PRODUCING A SEGMENTED SHEET BEAM 125

1 C r l . H

(C) (f )

Fig. 11. X-ray photos o f the beam at 27 cm from the cathode. (a) 4 A. (b) 10 A. (c) 30 A; 37 cm. (d) 4 A. (e) 10 A. ( f ) 25 A; 47 cm. (8) 4 A.

unsegmented beam. Operation with only one beam was significantly simpler than with two, and thus most of the parameter scans were done with the unsegmented beam turned off. Cross sections of the segmented beam, for var- ious currents and positions are shown in Fig. 11. At 27 cm, cross sections were observed with currents up to the design limit. At 37 and 47 cm the current was limited to 25 and 4 A, respectively, due to heating of the target.

IV. DISCUSSION AND CONCLUSION As can be seen, the beamlets are vertical and well sep-

arated for a current of 10 A. Similar cross sections were found with currents of 15 A. At 4 A, the beamlets are canted in one direction, while, at 25-30 A, they are canted in the opposite direction. At the highest currents, the cant- ing is most pronounced in the inner beamlets, which are on the right side of the photographs, while the outer ones are still fairly well aligned. This may have been due to the strong electric fields produced by the beam being closer to the wall than modeled. The beam as a whole is also distorted near the edge of the drift tube. This is also consistent with a strong radial electric field near the wall. Considering that the simulations did not include the beam interception and the close proximity of the drift tube wall, the results are consistent with the predictions of the AR- GUS simulation. Further tests, with an unintercepted

beam are required to check the consistency more care- fully. These are in progress.

Of course, even with the given geometry, the beamlet pattern could be optimized by use of a cathode with a different emission pattern. The changes in the pattern would be determined by mapping the deviations from the desired beamlet shapes back to the cathode. Similarly, the cathode could be modified so that the beam would be op- timized for a higher (or lower) current, or for a longer or shorter drift region.

These results show good stability for a high-current sheet beam over a significant distance. We note that this is in contrast to experience with sheet beams that have been examined for use in linear beam devices. The reason for the difference in stability is, however, quite simple: the magnetic field in which the sheet beam for the gyro- tron propagates is far higher than those which are com- monly used for linear beam devices. Since the magnitude of the drift is given by E / B , the rate of distortion will vary inversely with the magnitude of the magnetic field. Typical fields for a linear beam device are 0.1-0.2 T, while the average magnetic field in the gyrotron modeled here is approximately 2.5 T . It is clear that if the distor- tion found was magnified by the ratio of the fields, the beam would be substantially disfigured. Thus our results are not inconsistent with previous experience.

In summary, we have realized a gun which produces

126 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 39. NO. 3. MARCH 1992

both segmented and unsegmented high-current sheet beams for a quasi-optical gyrotron. The current in each beam has exceeded 30 A. The beams have been shown to be adequately stable. In the segmented one, the beamlets were, for most conditions, well aligned and separated. Simulations indicate that the beam has a low velocity spread ( 5 % ) . This combination should enable the effi- ciency of a QO gyrotron with this type of gun to be nearly optimum. The gun is expected to be tested in a quasi- optical gyrotron in the near future.

ACKNOWLEDGMENT The fabrication of the gun was very capably overseen

by R. Bosenberg. We are also happy to acknowledge help by Dr. W. Herrmannsfeldt on EGN.

REFERENCES [ I ] P. Sprangle er U / . , “Theory of quasi-optical electron cyclotron

maser,” Phys. Rev. A . vol. 23, p. 3127, 1981. [2] B. Levush and T. Antonsen. data transmitted personally, 1989. [3] M. E. Read, M. Q . Tran, J . McAdoo, and M. Barsanti, Int. J . Elec-

tro>i., vol. 65, no. 3, pp. 309-325. 1988. [4] A. W . Fliflet et al., “Operation of a quasi-optical gyrotron with

variable output coupling,” in Conf: Dig. 13th Int . Cunj on Infrured und Millimeter Waves ( S P f E . vol. 1039), 1988, p. 273.

[SI M. Q . Tran, “Operation and prospects of high power, high frequency quasi-optical gyrotrons,” in Cotij: Proc. 13th f r i t . Conf. on ftifrured and Millimeter Waves, 1988, paper W8.1.

[6] W. B. Henmannsfeldt, “Electron trajectory program.” SLAC Rep. 166, Stanford Linear Accelerator Center, 1973.

[7] K. Felch, Varian Associates, personal communication. 1988. [8] H. Huey. N. Lopez, R. Garcia, and K . Kreischer, “A magnetron

injection gun for the MIT megawatt gyrotron,” in ConJ Proc. /Or/? I n / . Con8 on lnfrured und Millinleter Wuves. 1985, p. 223.

[9] J . J . Petillo, M. E. Read. and A. J . Dudas, “Sheet beam gun design using ARGUS,“ in Conj Dig. Accelrrutor Technology Conf., 1989.

[ I O ] A. Mankofsky et al . , in Proc. 12rh Cotij of1 the Nurnericul Simula- tion of Plasmus (San Francisco, CA). 1987. paper PM 18.

Michael E. Read (M‘87) received the Ph.D. de- gree in 1975 from Comell University. Ithaca. NY, in electrical engineering and plasma physics. His graduate work was on the generation and propa- gation of intense relativistic electron beams.

During 18 months of postdoctoral work at Cor- nell. he utilized IREB‘s for the generation of mi- crowaves via space charge and cyclotron modes. Following his postdoctoral work, he worked at the Naval Research Laboratory, Washington, DC, and from 1983 to 1986 was Head of the Gyrotron Os-

cillators and Plasma Interactions Section in the Plasma Physics Division. While at NRL. he Derformed and managed fundamental research on elec-

journals, and a contributor to two books. He also is the author of a patent on a cavity commonly used in gyrotrons. He has been a reviewer for DOE, AFOSR, and ONR. and has served as guest editor for the fntrrnutionul Journul of Electronics.

Alan J . Dudas received the Master’s degree in electrical engineering in 1978 from Cornell Uni- versity, Ithaca. NY, and the B.S.E.E. degree in 1974 from Case Institute of Technology, Cleve- land, OH. His master’s thesis concerned high- power niicrowave generation using intense rela- tivistic electron beams.

He was employed as a klystron engineer at the Stanford Linear Accelerator Center for 2: years, where he designed a new electron gun and input coupler. He was in charge of the microwave cold

test for S-band production klystron cavities. In 1979, he was employed by Jaycor and worked at the Naval Research Laboratory (NRL) in the High Power Electromagnetic Radiation Branch. where he worked on gyrotron oscillators and MIG electron gun design. In 1982, while still employed at Jaycor. he began work for the Laser Physics Branch at NRL. He performed design and experimentation on X-ray pre-ionized, discharge-pumped ex- cimer lasers. While at Jaycor he was Principal Investigator on eight con- tracts. He has authored several papers concerning trirotrons, gyrotrons, and X-ray pre-ionized, discharge-pumped excimer lasers.

John J. Patillo received the B S degree in elec- trical engineering from Northea\tem Univervty , Boston. MA, in 1980. and the Ph.D degree in applied plasma physics trom the Massachusetts Institute of Technology. Cambridge, in 1986

Prior to 1986, he was a Graduate Research As- sociate at the Los A h n o s National Laboratory, where he worhed on betatrons and imploding liner driven electron accelerators He also worked as an electrooptical engineer at Raytheon Corporation on missile tracking sy\tem\ and at GTE Sylvania

on radio propagation phenomena He joined the Science Applications In ternational Corporation, McLean, VA, in 1986 He is a theoretical physi- cist and an electrical engineer who has broad experience in the application of modern numerical and dnalyticdl technique\ to the design and modeling of electron and ion gun\. spent-beam collectors, negative-ion wurces, high- current accelerators. electrmtatic quadrupole focusing acceleratodtrans- port systems, and magnetostatic quadrupole transport systems He is the primary developer of the gun modeling capability in the ARGUS aimula- tion code, a three-dimencional numerical modeling tool developed by SAIC He has written a Fpecialized multigrid electron gun code for use in the atomic Lapor la\er isotope separation (AVLIS) program at Lawrence Liv- ermore National Laboratory This code will utilize a user interface which operate\ on a Macintmh computer to facilitate the setup of structures and to allow interactive post-processed graphics In addition to his experience with the ARGUS code, he has implemented an implicit particle algorithm in the MASK code. a tuo-dimenwnal \imulation code. and he ha\ en- hanced the EGUN code. developed by B Herrmannsfeldt at the Stanford Linear Accelerator Center. to include secondary electrons He has also worked on the physics and design of compact electron accelerators tor the DARPA charged-particle-bcdm weapon\ program -

tron cyclotron masers (gyrotrons), electron cyclotron heating, and micro- wave-induced air breakdown. In October, 1986, he became a principal re- search engineer with Physical Sciences Inc., Alexandria. VA, where he is responsible for research programs on microwave and electron beam gen. eration and their applications. He is the author of over 30 papers in refereed

M. Q. Tran, photograph and biography not available at the time of pub- lication.