One of the most important problems in the design of high power generators of coherent radiation is the competition of many eigenmodes of the resonator for the beam energy. A simple one dimensional, multi-mode model for a low gain FEL oscillator is used to study numerically the effect of mode competition on the efficiency to produce a coherent radiation. To operate with the maximum efficiency at a single mode equilibria the current should be about three times start-oscillation current. As beam current is increased beyond four times the start oscillation value single mode operation is no longer possible. For relatively large values of the slippage parameter multimode stable equilibria are found for a broad range of the beam current values. These states have higher efficiency than the maximum efficiency of the single mode equilibrium states. With further increase of the current the radiation spectrum becomes chaotic and the behavior of efficiency is irregular.

Experimental and theoretical studies of high power microwave generation from three different plasma-filled backward-wave oscillator (BWO) configurations are reported. In the first, a neutral fill gas inside the BWO slow wave structure is ionized by the electron beam and a resonant enhancement of microwave output is observed at optimum values of fill gas pressure and applied axial magnetic field. In the second, a low energy electron beam is used to preionize the fill gas in advance of electron beam injection. In the third configuration, plasma is injected into the slow wave structure in advance of beam injection using a Marshall plasma gun. A fivefold increase in microwave output power and efficiency is observed in this case. Possible theoretical explanations for the observed interaction enhancement are also discussed.

Several high power X band traveling wave amplifiers (TWA) have been fabricated and tested. The tubes have gains ranging from 13dB to 29dB at 8.76 GHz, and output powers ranging from 3 to 100 MW. The amplifiers are driven by the interaction of a slow space charge wave on an electron beam with an electromagnetic wave supported by the structure. The electron beam has an energy of 850 keV, and currents in the 1 kA range. The amplifiers operate over a narrow range of frequencies which increases slowly with the length of the structure. The gain of the amplifiers increases rapidly with the beam current and with increases in the structure length. No evidence of oscillation in higher order modes has been found. We summarize in this paper the main features of the amplifiers.

A series of experiments is underway to utilize the virtual-cathode-driven oscillator as a source of high-power microwaves on Aurora. The most recent experiment launched 270 J per pulse to resistive terminations through eight radial waveguides, with the frequency in a band around 750 MHz. The highest power, ~10 GW in a short burst, was obtained with a second foil placed 0.5 m downstream of the anode. The peak power in a single waveguide exceeded 1 GW. This report compares the most recent measurements with those from previous experiments at Aurora and discusses the microwave extraction efficiency.

This paper addresses the new development of high power RF Klystron - like amplifiers using modulated intense relativistic electron beams. This research followed an earlier work in which the interaction between a high impedance ( 120 ohm) intense relativistic electron beam and a low power RF pulse resulted in the generation of coherent bunches of electrons with excellent amplitude and phase stabilities. In the present experiment a low impedance (30 Ohm) large diameter ( 14 cm ) annular electron beam of power ~ 8 gigawatts was modulated using an external RF source (Magnetron) of 0.5 megawatt power. The sequential interaction of the modulated electron beam with a structure generated a 3-gigawatt RF pulse that was radiated into the atmosphere.

This paper addresses several issues which are relevant to the high power klystron amplifier experiments at the Naval Research Laboratory. To a large extent, these issues originate from the DC self field of the intense beam and therefore are not treated in the conventional klystron theory. Among them include electrostatic insulation, limiting current which can flow through a gap in the presence of a biased gap voltage, and the modification of the transit time effects by the beam's space charge. We examine the proper interpretation of magnetic probe measurements of current modulation on an intense beam. Particle simulation of the large diameter beam yields excellent agreement with experiments and theory. Magnetic insulation in the coaxial rf extraction structure is shown to be adequate, in agreement with experimental observations.

The Proton Maser (PM) introduced by D. Ensely falls within the general class of devices known as transit-time oscillators (TM). A theory for both the small and large amplitude behavior of the intense-beam-driven TTO is summarized. It includes the effects of the beam self-fields and space-charge effects and it has been used to develop expressions which are compared with the results of particle simulations, which are presented in a companion paper.

Self-consistent, 2D, electromagnetic particle-in-cell computer simulations agree with the linear and nonlin-ear theoretical model presented in a companion paper and confirm the importance of space-charge effects. Simulations under idealized conditions demonstrate 18% conversion efficiency of ion beam power into 100 GW of extracted microwaves. Internal fields in the cavity reach 13 MV/cm. Typical constraints on the beam parameters are Vb > 3 MeV to avoid resonance shifts due to gap closure, Ib > 100 kA to maximize the growth rate and minimize the interaction Q, and Tb > 30 ns to keep the cavity gap d > 1 cm. This excludes small scale experiments from any proof-of-principal demonstration. Further theoretical research is needed for the necessary external magnetic field in cylindrical cavities and for plasma formation.

The average power spectral density per steradian, radiated perpendicular to the turbulent electron beam in a turbutron, is expressed analytically in terms of computed parameters of a one-dimensional time-dependent relativistic sheet simulation of the electron dynamics of the turbulent electron plasma.

Theory and simulations predict that an intense relativistic electron beam can generate high power microwaves simply by propagating axially through a pillbox cavity of appropriate dimensions. Basically, the beam is deflected sideways by the magnetic field of a TMmn1 cavity mode and then exchanges energy with the mode in proportion to J⊥•E⊥ and to JzΔ⊥Ez•X⊥ The microwave energy grows or damps depending on the beam transit time, T, across the cavity. Maximum growth generally occurs for Wmn1T ≈ (ℓ + 1/2)π for an unmagnetized beam, ℓ an odd integer. Results are qualitatively similar but more complicated when an axial magnetic field is present. Beam bunching plays a minor role, at least linearly, and particle reflections do not occur. Thus, this new device is not related closely to the vircator or the monotron. Several distinct realizations of this basic mechanism can be imagined, depending on the beam cross section, the cavity mode excited, and the magnetic guide field strength. The case of an unmagnetized annular electron beam propagating near the cavity wall and interacting with a TM011 mode is discussed extensively in a companion paper. A solid beam near the cavity axis interacting with a TM111 mode in a moderate applied magnetic field is an interesting alternative; it resembles a compact Gyrocon, a device known for high efficiency. Configurations analogous to single cavity Scantrons also are feasible, although they appear to have lower gain than the previous examples. These results also may be relevant to optimization of other microwave sources and to the beam breakup instability in high current accelerators.

The transvertron oscillator or amplifier is a new and efficient type of intense relativistic electron-beam-driven microwave radiation source. In the m = 0 axisymmetric version, it consists of single or multiple cylindrical cavities driven at one of the TMonp resonances by a high-voltage, low-impedance electron beam. There is no applied magnetic field and the oscillatory transverse motion acquired by the axially-injected electron beam is an essential part of the drive mechanism. The transvertron theory was systematically tested for a wide range of parameters and two possible applications. The simulations were designed to verify the theoretical predictions, assess the transvertron as a possible source of intense microwave radiation, and study its potential as a microwave amplifier. Numerical results agree well in all regards with the analytical theory. Simulations were carried out in two dimensions using CCUBE with the exception of radial loading cases, where the 3-D code SOS was required.

We have discovered in our two-dimensional particle-in-cell simulations that the oscillation of virtual cathodes in reditrons can produce a highly modulated electron beam. Full (100%) current modulation of the leakage electron beam was observed in our simulations. The modulation is at the frequency of the oscillating virtual cathode and the transverse magnetic mode generated by the reditron. We had further incorporated an inverse diode with a line impedance of 50 ohms in the reditron and showed that 28% of the kinetic energy of the modulated electron beam was converted into transverse electromagnetic waves with peak power of 1 GW.

The magnetically insulated transmission line oscillator (MILO) is a high power microwave device that combines the technologies of magnetically insulated electron flow and slow-wave tubes. This combination makes the MILO a unique and robust device capable of operating over a wide range of voltages. Microwaves are generated in the MILO via an unstable interaction between slow electromagnetic modes and space-charge waves. We present a linear analysis of this microwave instability that includes the physics of magnetically insulated electron flow. A dispersion relation is derived for transverse magnetic (TM) modes in a cylindrical MILO. The dispersion relation is solved numerically to obtain the MILO frequency spectrum and instability growth rates. We compare the linear theory results with numerical simulation, and obtain excellent agreement.

The gigatron is a new design concept for microwave power devices. A gated field-emitter array is employed to produce microwave-modulated electron beam directly from the cathode. A ribbon beam configuration is used to mitigate space-change effects and provide efficient output coupling. A traveling wave output coupler is used to obtain optimum coupling to a wide beam. RF conversion efficiency is estimated at ~75%. Gigatron-family devices have been designed for applications from 3 GHz to 60 GHz frequency, from 30 W microtubes for phased-array antenna drivers to 500 MW drivers for linac colliders.

The cusptron microwave tube concept is under development at NSWC as a compact high power device using the harmonic frequency generating scheme by an axis-rotating beam and a multivane circuit. Sixth and fourth harmonic frequencies have been independently generated with electronic efficiencies of approximately 10%. With this successful oscillator mode of operation, the cusptron is now being converted to an amplifier. The input and output couplers are one-turn loops located in different vanes. A TWT operating 4.0-8.0 GHz will be used as an RF driver. In this paper, we will present theoretical and experimental studies on the cusptron amplifier with a 30-50 keV, axis-rotating electron beam and a six-vane circuit.

The need for new, very intense microwave sources has been stimulated by the requirements for high-power phased arrays, vulnerability and effects tests sources, and the next generation of particle accelerators including the proposed linear colliders. All of these applications place stringent demands on the microwave source in addition to high power, including frequency and phase stability, efficiency, and long pulse ( > 1 μs) repetitive operation. Los Alamos is working on the resonant-cavity virtual cathode source and the large-orbit gyrotron in an attempt to experimentally address some of these issues. The resonant-cavity virtual cathode research has succeeded in demonstrating single-mode, narrow-band operation by surrounding the oscillating virtual cathode with a resonant-cavity structure. An experimental effort to injection lock the oscillating virtual cathode to a low-power injected signal is currently underway. The large-orbit gyrotron research is directed toward developing a technique for efficiently coupling the microwave output power into rectangular wave-guide. A 1-MV, 20-kA modulator with a 1-μs pulse length and a 5-Hz pulse repetition frequency (prf) has been designed and is currently in development; it will be used to drive a repetitively pulsed large-orbit gyrotron. The overall experimental effort is described and current experimental results are discussed.

Phase-locking of two relativistic magnetrons has been achieved at power levels of ≈ 3 GW at 2.8 GHz by connecting a short waveguide coupler of length LC ≈ n λ/2 between vanes. Phase-locking is demonstrated by both phase-sensitive diagnostics and power density enhancement due to source coherence in the radiation field. Phase-locking occurs in ≈ 5 ns and is reproducible. Extension to 10-100 GW appears feasible with magnetron arrays. Issues in operating arrays of coupled oscillators are discussed; an experimental approach to resolving these in a 4-magnetron module (expandable to a 7-magnetron module) is described.

We report results of the initial operation of a 1μs relativistic magnetron system to be used in the phase locking experiment. Full voltage pulse length with no evidence of diode shorting due to plasma closure is achieved in smooth bore geometry. The operating characteristics of the 700 kV, 700 A modulator together with the computerized control/acquisition system for the phase locking experiment is also described.

Nonlinear frequency-shift effects in a regenerative oscillator driven by an external locking signal can be modeled in canonical fashion by inclusion in the Van der Pol equation of an additional cubic restoring force term. This can be seen from an analysis based on the complex Poynting's theorem and from the presence of an additional frequency shift term in the first-order phase evolution equation. This Van der Pol/Duffing equation was proposed by Lashinsky as a model of the generic regenerative oscillator. Both Van der Pol and Duffing nonlinearities are important in conventional high-power sources. In the case of microwave sources the linear and nonlinear behavior of the system is governed primarily by the interaction term which appears in the complex Poynting's theorem. This term is itself complex and in the small-signal limit the real and imaginary components are related by the Kramers-Kriinig relations. Extension into the large-signal regime thus naturally results in the appearance of both Van der Pol and Duffing contributions to the nonlinear coupled-mode equations. Approximate analytic and numerical studies of the Van der Pol/Duffing equation, with emphasis on phase-locking behavior, will be presented. Finally the connection between the general theory and intense microwave source behavior will be illustrated by an analysis of the elementary monotron.

Linear stability properties of the extraordinary mode perturbations in a relativistic electron flow generated inside a planar magnetron are investigated. The stability analysis is carried out within the framework of a linearized macroscopic fluid model and the eigenvalue equation is obtained. In a tenuous density limit, an algebraic dispersion relation for the diocotron instability is obtained from this eigenvalue equation. Results of numerical investigation of this dispersion relation are presented. Particle simulation is also carried out for the diocotron instability and it is shown that the simulation results agree extremely well with the analytical results. Making use of finite element methods, the eigenvalue equation of the extraordinary mode is numerically solved and the results are also presented.

A macroscopic cold-fluid model is used to determine the influence of cylindrical effects on the operating range and properties of the electron flow in relativistic smooth-bore magnetrons. Assuming operation at Brillouin flow, it is found that cylindrical effects (such as the centrifugal force on an electron fluid element) can significantly modify several features of the equilibrium flow and diode operating range relative to the case of planar flow.

In cooperation with the SDIO/IST goals of generating phase-locked microwave and millimeter-wave power at the hundreds of megawatts to gigawatt power level, a program is under way at the Naval Research Laboratory to generate such sources by enhancing the performance of gyrotron oscillators. The NRL program has emphasized the higher frequency portion of the microwave spectrum, since low frequency phase-locked sources already exist, such as SLAC klystrons. The principal experimental effort is at 35 GHz.

Physics International (PI) has completed the first two phases of a three-phase program to explore the feasibility of constructing and operating a large array of high-power phase locked vircators. The approach has been to study multiple high-power sources in all the relative configurations that might be encountered in an array architecture. In Phase 1, dual source operation in the "master-slave" relationship was demonstrated at a power level of ~ 300 MW [Fig. 1(a)]. Power from the "master", relativistic magnetron selected the resonant mode, and locked the phase of the "slave" cavity vircator. Results from the experimental portion of this phase have been previously reported. Interpretations of the results from a theoretical model of the locking process is summarized in Section 3 of this report. In Phase II, two cavity vircators are operated as mutually coupled "peers" [Fig. 1(b)]. Phase locking in this arrangement was demonstrated at the gigawatt level. This is the principal subject of this report. Discussion is contained in Section 2. In the final Phase III, an array of multiple cavity vircators will be locked together in phase both in the "peer coupled" configuration and by a common high power driver (Fig. 10). A brief description of this proposed experiment concludes this report in Section 4.

A particle-in-cell model has been employed to investigate the phase-locking phenomenon of multi-cavity relativistic gyrotron oscillators. Simulation results show that a prebunched beam causes the output wave to overshoot, which in turn prolongs the time for establishing phase locking. The beam axial velocity spread is observed to reduce the locking bandwidth. The phenomenon of priming or injection seeding is simulated. The phase locked time depends on the growth rate of the oscillator and the amount of inject frequency deviation from the locking boundary.

Recent theory showed that phase locking of two oscillators can occur only when the connector contributes zero or r phase change. Here we will present results on the phase locking of multiple oscillators. Two configurations are considered: (1) Centrally connected oscillators - peripheral units are not connected, but connected individually to a central unit; (2) Ring-connected oscillators - all units are connected in a ring configuration. The preliminary result is that the centrally connected oscillators phase lock better than the ring-connected oscillators. This is because it takes a longer time for all the oscillators to communicate to each other in the second case. Temporal behaviors were solved numerically, which found the phase locking depends strongly upon the connector lengths, the connecting ports and the modes of operation of the oscillators.

Experiments are underway to investigate the generation of MW to GW microwave power levels utilizing intense (1-10 kA), relativistic (0.4-1 MV), electron beams with pulselengths from 0.4 to 1 μs. Significant research issues concern the effects of beam voltage fluctuations and space charge on long-pulse microwave emission power and frequency stability. Two prototype masers have been developed utilizing a moderate energy pulseline generator with peak parameters: 0.4 MV, 1.2 kA, and 0.4 μs. Prototype I utilized a nonadiabatic step in the diode magnetic field to impart a large α (V⊥/V11≈1-2) to a relatively low transported current of 30-60A. With a uniform waveguide resonator this maser generated peak microwave power levels of 0.6 MW and pulselengths of 0.3 μs over 85% voltage droop. The emission frequency was close to the relativistic cyclotron frequency. The Prototype II maser employed an improved magnetic field profile which transported 500-1000 A of beam current, but at a lower value of α(0.2 - 0.3). A series of permanent magnet wiggler/kickers were used in preliminary tests. Power levels were achieved in the range of 2 MW in the X-band and at MW levels in the K-band. These emission frequencies apparently correspond to the CRM forward-wave and backward-wave interaction with the TE11n cavity modes. By lowering the magnetic field below a critical value the microwave emission was cut off. Design studies have been performed for a long-pulse MW-GW cyclotron maser on the MELBA accelerator with parameters: 0.8 - 1 MV, 1 - 10 kA, and 1 - 1.5 μs.

Studies of a cyclotron autoresonance maser (CARM) are presented. The measurements are carried out at a frequency of 35 GHz using a mildly relativistic electron beam (1.5 MeV, 260 A) generated by a field emission electron gun followed by an emittance selector that removes the outer, hot electrons. Perpendicular energy is imparted to the electrons by means of a bifilar helical wiggler. Amplification measurements give a small signal gain of approximately 90 dB/m. Computer simulations are also presented.

A high power, 140 GHz, cyclotron autoresonance maser (or CARM) amplifier is under development at M.I.T. Theory, simulations, and the experimental design are discussed. The experiment will employ a high voltage (450 - 700 kV) Pierce-type beam and a helical wiggler to produce the required elec-tron beam. The initial experiment will be carried out using a 450 kV electron gun; the second phase of the experimental program will employ a 700 kV electron gun. The sensitivity of the CARM to velocity spread is discussed. The stability of the CARM amplifier to absolute instabilities has also been investigated, and the criteria for stable CARM amplifier operation are presented.

The CARM's advantages of high efficiency and high frequency can be obtained with a low voltage electron beam by incorporating a slow wave structure. A wave with βph ≈ 1 has been excited by a 100 keV electron beam generated by a Pierce/Wiggler gun.

Theory and experimental designs of a 3 MW, 560 GHz CARM as well as a 10 MW, 33 GHz, 400 kV proof of principle experiment will be presented.

A prebunched beam has been used to further increase the efficiency of our axis-encircling high harmonic gyrotron. The experiment was performed at the third harmonic with a TE312 mode at 27.7 GHz. The conversion power was 6.7 kW which was significantly greater than that of a non-prebunched comparison experiment. Also, mode competition was substantially reduced. The dependence of the output power on the square of current has been verified. The start of oscillation current is effectively zero.

A gyro-TWT is often plagued by absolute instabilities in lower order modes and even in the operating mode itself. These problems will be avoided by performing an experiment in the lowest order mode in cylindrical waveguide ( TE11 ) and using a low α beam to excite a wave with large kn. A gyroresonant wiggler will partially transform the axial velocity of a high quality Pierce beam into transverse velocity. A 100 keV, 5 A, α = 0.5 beam will generate 9 GHz rf in a 3 kG magnetic field. Magnetic tapering will be used to enchance the efficiency. The theoretical gain is predicted to be 60 dB with a bandwidth over 20%.

The use of a small period wiggler (ℓw < 1 cm) together with a sheet electron beam has been proposed as a low cost source of power for electron cyclotron resonance heating (ECRH) in magnetic fusion plas-mas and for space-based radar systems. We have experimentally demonstrated stable propagation of a sheet beam (18 A, 1 mm x 20 mm) through a ten-period wiggler electromagnet with peak field of 1.2 kG. Calculation of microwave wall heating and pressurized water cooling have also been carried out, and indicate the feasibility of operating a near-millimeter, sheet beam FEL with an output power of 1 MW CW (corresponding to power density into the walls of 2 kW/cm2). Based on these encouraging results, a proof-of-principle experiment is being assembled, and is aimed at demonstrating FEL operating at 120 GHz with 300 kW output power in 1 μs pulses; electron energy would be 410 keV. Preliminary design of a 300 GHz, 1 MW FEL with an untapered wiggler is also presented. Finally, a method of modulating high power CW signals for radar applications is suggested.

We present the design, construction, and test results of a novel micro-wiggler structure with a periodicity of 2.4 mm for free-electron laser (FEL) applications. The experimentally demonstrated tunability of field amplitudes provides versatile means for field tapering, optical klystron configurations, improving field uniformity, and electron beam matching at the wiggler entrance.

The planar orotron structure is a versatile resonator which can be used as the basis for a cm-wavelength microwave generator. In its most elementary form, it consists of a narrow planar grating which is embedded in a conductor and opposed by a second conductor. Cylindrical section mirrors at the ends of the grating complete the resonator. The width of the grating is determined by the width of the driving electron beam. Thus, within limits set by beam stability, transverse scaling of the interaction volume is possible. In addition, the planar geometry allows easy stacking of additional resonators, and this provides a further scaling of the interaction volume. The aperture area of a complete multi-module resonator will increase with the total number of resonators, and this scaling can be used to maintain field strengths in an intense-beam-driven high-power source at levels below the breakdown threshold. A summary of the results needed in order to optimize the performance of a single orotron module will be presented and scaling of multi-resonator structures will be discussed.

A high power (GW level), high efficiency (⪆ 10%), high gain (40 db), long pulse (1 μsec) Plasma Cerenkov Maser (PCM) is under development. The preparations for preliminary experiments at reduced power (.1 GW) and pulse length (.1 μsec) are near completion. The experimental design has been modeled using linear and particle simulation codes for both reduced and full power operation. Several designs have been modeled in an effort to find a method to reduce the AC space charge effect of high current beams. Optimum efficiency is obtained with a thin beam close to a thin plasma. Calculations with a full 2-1/2 D relativistic particle code are presented for the first reported experimental device. The results generally agree with our previous 1 D calculations. Stability analysis indicates self-oscillation due to imperfect coupling can limit the gain of the amplifier. The design of the UCD PCM is described.

Optical diagnostics using laser fluorescence techniques measure the RMS electric field in a superstrongly turbulent, relativistic beam-plasma system. This yields detailed mapping of <E2> as a function of axial position, z, radial location, r and time, t. This <E2(r,z,t)> allows studies of growth and evolution of turbulent fields, their diffusion and decay. Fluctuating electric fields occur when a 700 keV, 4 kA, 2 µs electron beam propagates into 20 cm diameter, 1.5 m long drift tube filled with 10 mTorr of Helium plasma. Stark effect shifts appear in suitable forbidden and allowed transitions, originating from the same upper energy level for the measurement: HeI 6632 Å and HeI 5015.7 Å. The spectral bandwidth includes the forbidden line and its satellites. Using the ratio of the intensity of the forbidden plus satellite lines, to the allowed line intensity, yields the R.M.S. field as the combined field of oscillation near the plasma frequency. Fields up to 28 kV/cm result. An analytical model of production of strong electric fields by beam-plasma instability, including modulational transfer in k-space, plasma heating, radiation and wave convection explains these results. Comparison between the experiment and the numerically integrated model is good.

A non-nuclear energy conversion concept-MTL (Microwave Power to Low Earth Orbits) was previously presented in order to supply SDI platforms power in the housekeeping, alert and burst power modes. In this paper the major issues addressed are:, system design, integration and analysis. Parametric design of the major subsystems of the MTL bus, which includes the rectenna, the monolithic solid oxide fuel cell etc, is presented. The results of the parametric design,and of computer simulation are used as inputs to construct a comprehensive systems design code. A reference MTL system design which meets the requirements of duty cycles spelled out in open literature is presented. A comparison of mass and power is made between the MTh system and the SP-100 and burst power systems, which demonstrates the competitiveness of the proposed MTh design.

An inportant problem in strategic missile defense is discrimination between decoys and reentry vehicles. During midcourse flight, a ground-based HPM directed energy system could distinguishably modify (by a physical damage mechanism) the radar signature of decoys. A very large ground-based transmitting station imposes unique requirements on its phased array antenna and on the RPM sources which will power the array.

The electromagnetic field from the current in a circular disk and its Poynting vector off the axis of propagation have been investigated both analytically and experimentally. These studies show that the electromagnetic (EM) missile has the following properties: (1) The waveform of the EM field when propagating remains similar in shape but diminishes in size. (2) The transverse distribution of energy around the axis is stable; i.e., when the longitudinal distance increases, the transverse pattern of the energy remains the same. This "plane-wave beam-like" property is different from continuous-wave (CW) radiation, in which the energy pattern is like that of a spherical wave. (3) The transverse energy pattern has a cusp on the axis. This property is also different from CW radiation. With CW, the radiation has a flat top on the axis of the main lobe.

Transient electromagnetic fields can exhibit interesting behavior in the limit of great distances from their sources. In situations of finite total radiated energy, the energy reaching a distant receiver can decrease with distance much more slowly than the usual r-2. Cases of such slow decrease have been referred to as electromagnetic missiles. All of the wide variety of known missiles propagate in essentially straight lines. A sketch is presented here of a missile that can follow a path that is strongly curved. An example of a curved electromagnetic missile is explicitly constructed and some of its simpler properties are discussed. References to details available elsewhere are given.

Theoretical results are reviewed pertaining to the behavior of transient electromagnetic fields in the limit of great distances from their sources. In 1985 it was shown that pulses of finite total radiated energy could propagate to a distant receiver, delivering energy that decreases much more slowly than the usual r-2. Such pulses have been referred to as electromagnetic (EM) missiles. The types first discovered propagate along a straight line with a monotonically (though slowly) decreasing time-integrated flux. Other types are now known. One type can be made to rise and fall with increasing distance; another is the curved EM missile. Early efforts to classify EM missiles are reviewed. References to recent papers reporting on the experimental studies of EM missiles at Harvard are given.

Development of new concepts for time-harmonic and transient electromagnetic radiation in general, and for directed energy transfer in particular, requires detailed understanding of the spectral as well as the configurational behavior of the fundamental wave constituents employed in field synthesis. The simultaneous treatment of these aspects can be effected in a phase space covering configurational and spatial wavenumber variables for the time-harmonic regime, with the addition of time and frequency variables for the transient regime. The full phase space constructed in this manner is generally not the most convenient for tracking physical observables. Instead, it is desirable to employ reduced or partial phase space representations based on the spectral decomposition of only some of the space-time variables. Various choices in the reduction lead to alternative representations with different physical content, and to corresponding alternative parametrizations of the radiation process. The relation between these alternative forms, their evolution in the phase space, and their manifestation in real space-time, is explored in general terms. Special attention is given to incorporating spatial or spectral windows in the formulation, thereby effecting localization in the form of beams. In this brief summary, only the formulation is presented. Full versions of the theory will be published separately.

In previous publications [L.B. Felsen and E. Heyman, Proc. SPIE Vol 873 pp. 320-328 1988; E. Heyman, Wave Motion, in press], it has been shown how highly focused pulsed fields in vacuum can be generated analytically by assigning complex values to the space-time source coordinates of the conventional transient free-space Green's function. These new wave objects have been called complex source pulsed beams (CSPB). Their utility can be extended to generating new solutions for pulsed beam propagation and diffraction in a perturbed environment by making the space-time source coordinates in the corresponding Green's function complex. The analytic extension required in this process is performed systematically via the spectral theory of transients (STT) [E. Heyman and L.B. Felsen, IEEE Trans. Antennas Propagat. AP-35 (1987), 80-86, 574-580]. A canonical test case for reflection, including critical angle and lateral (head) wave eflEaTis provided by a dielectric half space. The exact solution for CSPB scattering is derived in spectral integral form, evaluated in terms of the time-dependent and time-independent spatial spectrum singularities in the complex plane, and interpreted physically. Numerical evaluation reveals the detailed space-time behavior of these physical constituents and their role in establishing the total scattered field.

Exact solutions of the scalar wave and Maxwell's equations that describe localized transmission of wave energy and their representations will be reviewed briefly. These acoustic (ADEPT) and electromagnetic (EDEPT) directed energy pulse train solutions can be optimized so that they are localized near the direction of propagation and their original amplitude is recovered out to extremely large distances from their initial location. Pulses with these very desirable localized transmission characteristics have a number of potential applications in the areas of directed energy weapons, secure communications, and remote sensing. The feasibility of launching an ADEPT from an array of acoustic transducers has been tested experimentally. As will shown, excellent agreement between theoretical and experimental results was obtained.

We show that all finite energy, causal solutions of the time-dependent three-dimensional acoustic equation and Maxwell's equations have forms for large radius which, except for a factor 1/r represent one-dimensional wave motions along straight lines through the origin. The asymptotic region is called the wave zone.We also show, conversely, how the exact finite energy causal solutions can be obtained from the asymptotic solutions in the wave zone through the use of a refined Radon transform. We have thus obtained a way of finding exact, causal three-dimensional solutions from the essentially one-dimensional solutions in the wave zone. When the asymptotic solutions are confined to a finite radial interval within a cone, the exact solutions are termed "bullets" and asymptotically represent packets of acoustic and electromagnetic energy "shot" through the cone. No reflectors are needed. Explicit examples will be given.

The following paper reports on a series of experiments that were carried out to verify the possibility of using a plasma, as the nonlinear medium, for generating phase conjugate reflection of electromagnetic waves in the microwave region, by almost degenerate four wave mixing. In a theoretical work done at UCI we have shown that the phase conjugate reflection is of significant magnitude if the frequency and the wave vector difference of the signal wave, with respect to the pump waves, resonate with the frequency and the wave vector of the ion acoustic mode of the plasma. The reflected beam can be highly amplified by the plasma. The main results of the theory developed are described, as well as the series of experiments we have carried out to verify the phase conjugation. Possible applications of the emergent technology are discussed. The experiment was terminated without observation of the PCR because of lack of funding. Possible ways to continue the experiment are discussed.

In the Plasma Wakefield Accelerator short electron bunches accelerate by surfing on ultra-high gradient plasma waves excited by a lower voltage, higher current electron beam. This paper reviews the concept of and recent progress on the Plasma Wakefield Accelerator. Proof-of-principle experiments have recently been carried out at Argonne National Laboratory and modelled at UCLA. Computer simulations of the experimental conditions confirm the importance of non-linear beam self-pinching and wave steepening in the experiment. The results of the experiment are encouraging for a next phase of experiments to demonstrate a 100 MeV compact accelerator ( < 1 to 2 meters) based on a plasma wakefield device driven by a laser photo-injector linac.

Lanthanum hexaboride has been tested as a photoemitter when irradiated by unpolarized UV lasers. For photon energies of 5 eV or less, the material, in a multicrystalline form, is measured to have a quantum efficiency at least an order of magnitude greater than that of simple metals. Maximum currents, from a 1.27 cm2 sample, limited by the available laser power, were recorded to be 52, 36, and 0.9 A at irradiating wavelengths of 193, 248, and 308 nm, respectively. At 193 and 248 nm the corresponding normalized rms brightnesses were 6.7 x 106 and 2.6 x 106 A/cm2-rad2. The results appear insensitive to chamber pressure in, at least, the range of 10-5 to 10-8 toff.

Conventional dc sources of H- are limited to current densities of the order of 50 mA/cm2 for sources with area larger than a few cm2. Early work at UCI and more recent work at the Lebedev Institute have shown that pulsed magnetically insulated ion diodes can produce current densities larger by factors of the order 102 - 104. We studied the production of negative ion beams in the coaxial, racetrack and annular diode geometries. The experiments showed that when using a passive dielectric cathode negative ions are emitted mainly from a few "hot spots" located on the cathode surface. The angular divergence of the beam was about 300 milliradians. We have developed a TiH2 plasma source that consists of 120 discharges on a flashboard. The resultant cathode plasma is quite homogeneous compared to typical polyethylene passive cathodes. H- ions with a current density of 6 A/cm2 and a divergence of 10 milliradians were observed with this source. The total current of H- ions was about 1 kA. The current density is many orders of magnitude larger than would be expected from a thermal plasma source. A non-equilibrium model is advanced to explain this fact.

The spiral line recirculating induction accelerator (SLIA) is an open-ended spiral configuration in which the electron beam recirculates in independent transport lines (< 20) passing through a common induction accelerating section with high gain/ pass (~ 10 MeV). A toroidal (axial in the straight sections) field threads the transport lines for space charge confinement and suppression of instabilities. A strong focussing ℓ = 2 stellarator field is used in the bends to provide improved tolerance to field errors and energy mismatches at lower energies. Experimental and theoretical programs are underway to investigate key physics and design issues for SLIA. A beam transport and 180° beam bending experiment which uses a 1 MeV, 1 kA, 100 ns beam is investigating control of emittance growth and energy bandwidth of the bend. A full scale prototype accelerating test cell is(being built to measure the effective impedance of the cell, magnetization current for the ferrite, effective impedance of the ferrite and the interaction of the cell, beam, and power supply. The programs are intended to provide the information required for scaling to a multi-pass induction acceleration experiment to study the beam breakup instability and complete a proof-of-concept demonstration.

Image charge electric fields can produce large cumulative erosion of the brightness of a high current electron beam during transport through successive high-gradient accelerating gaps. However, this erosion in beam brightness may be significantly reduced through the use of smoothly shaped, impedance-matched accelerating gaps with optimized gap-to-gap spacing. Moreover, for beams transported with solenoidal magnetic focusing, it is possible to spatially tune the solenoidal field within the accelerating gaps to compensate the image charge perturbations. Calculations suggest.that with such techniques it is possible to deliver from a multi-gap accelerator, high current beams whose diode brightness is essentially preserved during transport and acceleration. If disruptive, high current beam kinking instabilities are also controlled, such high-brightness beams are uniquely capable of driving high performance FEL operation at wavelengths in the infrared and below.

The high field-gradient accelerator employs two beams of electrons which interact through an RF structure. The first, a large-diameter modulated intense relativistic electron beam (MIREB), generates an electromagnetic field which in turn accelerates a second beam. This effect is large, only, when the second beam is much weaker and shorter in duration than the first. The accelerating structure is designed in a way to produce a resonance interaction with the MIREB, leading to the energy transfer from the beam to the structure. The electromagnetic wave induced in the structure, converges radially, generating an average electric field of the order of 100 MV/m on the axis. The short duration of the MIREB pulse (140 nsec) allows us to establish these high electric fields in the accelerating region and complete the acceleration before breakdown can occur. A secondary particle beam is injected on the axis and accelerated by the high electric field. This particle beam can be initiated and controlled by a modulated laser light, phase-locked to the RF source used to modulate the MIREB. The high power RF established in the structure by the MIREB allows acceleration of up to 100 Amps of electrons, in the present experimental configuration. Preliminary results show electron acceleration of at least 3MeV with a beam current of about 100 Amps.

The generation of intense, high-brightness H- beams requires solutions of several physics and technology problems in the low-energy beam transport (LEBT) section of the accelerator to avoid particle losses and emittance growth. Practical constraints limit the current and intrinsic emittance of the beam. On the other hand, the focusing capability of magnetic lenses is limited, and most existing accelerators use space-charge neutralization in a background gas such as Xenon, often in combination with magnetic lenses, to transport the beam. However, the demands for higher beam brightness and intensity require a better understanding of the gas focusing mechanism. Previous one-dimensional fluid studies have given insight into the approach to steady state from the charge-neutralized state when considering a frozen beam profile. In this paper, we present the results of including the response of the beam to the plasma.

Beam stability in high-current recirculating accelerators using magnetic beam transport is investigated. We focus on three instabilities which have the potential for causing beam disruption: the three-wave, negative-mass and beam breakup instabilities. The three-wave instability is a parametric instability caused by strong-focusing fields like the helical quadrupole in the SLIA (Spiral Line Induction Accelerator). We have obtained approximate analytic expressions for the growth rate and the convection velocity of this instability. A simple numerical simulation code E3WAVE has been written to model the linear behavior of the instability. Its results confirm the analytic results in the appropriate limits. We have used E3WAVE to make predictions for the SLIA experiments presently under way. We also have investigated the short-wavelength negative-mass instability modes, which interact resonantly with cavity modes of the drift-tube. Particle simulations show that the main nonlinear effect of the instability is to create an energy spread on the beam, rather than leading directly to current loss. We have made progress in minimizing the beam breakup instability without resorting to ion focusing. By adding ferrite in the induction gaps to increase damping, and by using gaps that are considerably wider than usual, with correspondingly higher voltage, growth can be reduced to several e-foldings during a typical acceleration cycle.

MA focused discharges are very intense, pulsed sources of ion cluster and superciuster beams with energy > 1 MeV/nucleon. Our experimental data refer to: (i) The mechanism of cluster formation and related plasma cooling processes. (ii) The mass/charge spectrum (1 ≤ m/z ≈ 1012 in A.U.). (iii) The anisotropy of the energy spectrum and of the emission intensity. (iv) The space and time structure of the source. (v) Beam focalization and neutralization. The experimental data have been obtained by operating the source at different levels of energy Wo = 6-25 kJ (Vo = 15-30 kV) of the fast capacitors bank which feeds the discharge. The inductive fields in the discharge pinch accelerate deuterium ions to energy values Ei ≈ 10 MeV. Beam focalization and neutralization is achieved with a pulsed-magnetic plasma lens on a 0.1 μs time scale. A Thomson spectrometer with nanosec periodic-or ramped-electric field is used for a time-resolved analysis of the beams.

This paper reviews the theory of emittance growth of a space-charge-dominated particle beam in a solenoidal focusing field and proposes an experiment to measure the emittance of an electron beam through a magnetic optic. This electron beam can have two different radial charge distributions. The experiment will attempt to show a relationship between the nonlinear field energy and emittance growth.

We performed experiments on the injection of a high-power relativistic electron beam (130 A, 300 keV) into the UNM Betatron. The beam entered at an angle of 20° with respect to the main axis of the accelerating ring. An electrostatic inflector provided a transverse electric field which bent the beam onto the main axis of the machine. An electron beam current of 30 A was contained in a well-defined core, as measured at the exit of the first 180° curved section of the ring.

A brief overview of the physical mechanisms of negative ion production in a hydrogen/deuterium plasma is given. An analytical model is presented which describes many of the features observed in experiments on the production of negative ions in these plasma sources. This model leads to a prediction of the performance capabilities of these sources. Comparison of experimental results for the production of H- and D- ions is also presented.

Formation of beams of negative ions by the process of dissociative electron attachment to a molecule AB (e- + AB → A + B-) is investigated. It is demonstrated that the initial rovibrational excitation of the molecule AB plays a significant role in enhancing the rate of production of negative ions. A simple physical picture, involving the formation of a transient resonant state of the molecular anion, is used to explain this enhancement.

The causes of observed saturation of hydrogen negative ion current at high power operation of tandem volume sources are investigated using a zero-dimensional time-dependent model, which assumes that dissociative electron attachment to highly vibrationally excited molecules is the main formation mechanism of hydrogen negative ions. It was found that the reason for saturation was the reduction of the density of molecular hydrogen, as a result of dissociation. The mean free path of the vibrationally excited molecules for various destruction processes will be discussed in relation with the source high power operation.

Merged Beam studies of the branching ratio for the dissociative recombination (DR) of H3+ (v = 0) indicate that this process is dominated by the indirect mechanism yielding +a rate of ~10-8cm3s-1 at 300K. Measurements for the DR for HeHyield a non-negligible cross section containing deep resonances. The role of dielectronic resonances in the electron impact excitation of H3+ is discussed.

A four-step model for recombination and dissociative recombination of H2+ and H3+ ions on metal surfaces is discussed. Vibrationally excited molecules, H2(v''), from H3+ recombination are produced in a broad spectrum that enhances the excited level distribution. The application of this latter process to hydrogen negative ion discharges is discussed.

Negative hydrogen ions are produced by electron tunnelling from a solid surface (converter) to hydrogen atoms moving away from the surface. A simple theoretical model of this process indicates reasonable ionization probability (>0.1) for low initial energy of the H atoms (order of 1 eV) when the work function of the converter is smaller than 1.5 eV. Low initial energy guarantees low H- ion energy spread. It also makes it possible to use a variety of compounds for reducing the converter work function. Experimental studies of Cs oxide films (work function 0.9 to 1.2 eV) are presented. Production of H- ions by backscattering thermally produced H atoms from a thick film of Cs oxide is demonstrated.

Cesium, hydrogen and oxygen adsorption on beryllium clusters are studied using restricted Hartree-Fock (RHF) calculations and ab initio relativistic effective core potentials. The clusters are taken as cylindrical plugs from Be wafers. Cs-, H- and 0-to-substrate internuclear distances are optimized. For each system numerous low-lying electronic states are investigated, and the Mulliken electron populations are analyzed. The calculations are carried out in the context of an experimental study to determine the effects of various adsorbates on the work function of the substrate. Auger electron spectroscopy and experimental work function measurements indicate that H2 does not adsorb on polycrystalline Be, while photoemission and thick Cs overlayer measurements show a 2.3 eV lowering in the work function of Be metal upon Cs adsorption. The continuous oxidation of Cs has been studied using ultraviolet photoelectron spectroscopy and electron spectroscopy by deexcitation of metastable noble gas atoms. Results indicate that the work function of Cs is lowered upon exposure of the surface to small doses of oxygen. RHF calculations show that a 19 atom Be cluster, with three layers of atoms, is too small to adequately model the Be surface, while the 33 atom cluster, a five-layered system, and the 45 atom cluster, a seven-layered system, are more accurate representations of the bulk metal. The emitted electron is clearly seen as vacating a molecular orbital which is localized in the surface layer of the cluster, thereby giving further credence to the model.

Experiments on the production of negative hydrogen ions via surface conversion for incident positive hydrogen ion currents in the range of 0.2 to 1 A/cm2 are reported. The conversion efficiency, i.e. the ratio of the produced negative ion current and the positive ion current on the converter shows a strong temperature dependence and has a maximum for a positive ion current density of about 0.5 A/cm2. Due to the limited magnetic confinement of the intense hydrogen plasma, the base pressure of these experiments was of the order of 50 mTorr which resulted in very large losses of negative ions due to stripping. With correction for these losses, we obtained a conversion efficiency, of the order of 8 percent which corresponds to a surface produced negative ion current density of the order of 60 mA/cmh.

The operation of a Dudnikov-type H- ion source using lanthanum hexaboride as the cathode material has been achieved and has been previously reported. The cesium-free operation achieved by using this thermionic emitter is of benefit to the overall neutral particle beam system performance. The source requirement for a neutral particle beam will require higher output from this source. The experimental apparatus is a Penning-type discharge ion source with a variable magnetic field. A brief discussion of the parametric optimization will be covered. In an effort to further increase performance of the Dudnikov source, xenon and argon have been added to the hydrogen fuel used in the source. The inert gas has been added in small amounts (0.2-2.0%) as the H- and e- levels were monitored. Apparatus, diagnostics and results will be discussed.

Measurements of the rovibrational population of H2 products following dissociative charge transfer of D3+ in Cs vapor are reported in the context of the recent observation that the addition of Cs to an ion source discharge can lead to a significant increase in the H- ions extracted from the source. Consideration is also given to wall and plasma effects. It is concluded that dissociative charge transfer produces a hot H2(v) distribution, centered around v=6,with all levels populated into the vibrational continuum. It is also found that the discharge is an efficient producer of H atoms. Through wall collisions these H atoms can contribute both directly and indirectly to H- production in a Cs seeded discharge. Comparison with results from Xe seeded H2 discharges suggest that in Cs seeded discharges the effectiveness of the magnetic filter may be reduced, the electron density is significantly increased and the plasma potential will be increased. However, with Xe seeding, there is no conclusive evidence for electron cooling of the bulk electrons.

H- and Li- have many useful applications in accelerator and fusion research. It has been demonstrated that these negative ions can be generated either by surface conversion or by volume processes. Since the H- or Li- ions produced by volume processes have lower transverse energy than those formed by surface conversion processes, the emittance of the H- or Li- beam extracted from volume sources will be much lower. Experimental investigations have been conducted to enhance the H- and Li- yield in the volume multicusp source by optimizing the source geometry and by mixing various elements with the background hydrogen or lithium. The largest improvement is achieved by adding cesium vapor to the source, resulting in a factor of sixteen increase in H- output. Source operation with an admixture of lithium and cesium has also been tested for the production of Li- ions.

Calculations are presented on continuous plasma generation by the optical pumping of a supersonic lithium beam. The plasma density is computed as a function of pump intensity on the Li(2s-2p) and Li(2p-3d) transitions. The factors affecting Li- generation are considered.. It is found that modest optical pump powers allow the generation of Li currents at up to 10 mA cm-2, at approximately unity ratio of electrons to Li- ions.

Theoretical studies of collisional ionization and electron attachment mechanisms involving light elements (H, He, Li) are being carried out in support of particle beam sources and beam neutralization concepts. In particular, the electronic structure of stable anion; their resonance and metastable states, and radiative and collisional processes describing their formation and/or destruction are being investigated. The volume production of atomic anions via electron-molecule dissociative attachment has previously been examined, both experimentally and theoretically, for e + H2 and e + Li2 collisions. An alternate production route, involving ion-pair formation from Rydberg-Rydberg atom collisions, is currently being investigated. The possible mechanisms include: ionic-Rydberg molecule curve crossings, collisional radiative stabilization, formation of metastable or resonance anion states followed by radiative stabilization, and associative ionization followed by secondary electron-neutral attachment processes. Recent experiments by McGeoch, et al indicate that associative ionization is highly favored over direct ion-pair production in the Li system. However, both Ciocca, et al and Cheret and Barbier report efficient ion-pair production in collisions between heavier (Na, Rb) alkali Rydberg atoms. Several possible reaction mechanisms have been examined for these systems.

One of the methods of neutralization of a negatively charged particle beam is to photo-detach the extra electron on the particle by a laser beam. At high beam energies this method is more efficient than the charge exchange technique used for a positively charged particle beam. Therefore, development efforts have been in progress in the field of negatively charged particle beams. Among several techniques which are used for the production of negative ion beams the technique of surface conversion of molecules into negative ion fragments has been least explored. Two recent experiments where dissociation of a molecule gave rise to negative ions enhanced our interest in the study of negative ion production by surface conversion. First one was conducted by Palmer who observed that when a beam of LiH was made to impinge upon a heated tungsten surface (heated between 600K and 1000K) H- was produced. They attributed this to surface chemi-ionization. The second was reported by Wang et al. They found that thermal SF6 molecules decomposed into various negative ion fragments upon impact with room temperature alkali surfaces and, thus, these surfaces became copious sources of negative ions. Therefore, we undertook the study of H- formation by LiH. In the following the experimental arrangement is described and the results are presented.

A new technique, vacuum-ultraviolet laser-absorption spectroscopy, has been developed to quantitatively determine the absolute density of H and H2 within a plasma. The technique is particularly well suited to measurement in a plasma, where high charged particle and photon backgrounds complicate other methods of detection. The high selectivity and sensitivity of the technique allows for the measurement of the rotational-vibrational state distribution of H2 as well as the translational temperature of the atoms and molecules. The technique has been used to study both pulsed and continuous IC ion-source plasma discharges. H2 state distributions in a multicusp "volume" If ion-source plasma show a high degree of internal excitation, with levels up to v = 5 and J = 8 being observed. The method is applicable for a very wide range of plasma conditions. Emission measurements from excited states of H are also reported.

It has been suggested that the main source of IV in hydrogen plasmas is dissociative attachment to H2(v≥4). In order to test models that include this process, it is necessary to measure the populations of the relevant vibrational levels inside the plasma. We have constructed a source of vibrationally excited H2 using a recently discovered hot-wire effect. Using 2+1 resonantly enhanced multiphoton ionization (REMPI) through the EF state, we have detected H2 in 27 bands from X(v"=4-11) to EF(v'=0-14). Rotational temperatures appear to be 300-800 K, well below that required for thermal excitation of the observed vibrational levels. The population of vibrational levels appears to fall off dramatically between v"=9 and 10.

An H- source for high brightness H° beams should have the lowest possible internal temperature T, because the beam-power density at a distance scales as 1/T. We determine local T inside a discharge with the use of a tunable ArF excimer laser that measures the populations of the lowest four rotational states of H2 by means of (2+1) resonance enhanced multiphoton ionization (REMPI). The number of laser-created ions causes a proportional change of the discharge's electrical impedance. A calibration in room temperature H2 (without a discharge) showed that the REMPI works well. Another objective is to develop a monitor for the vibrationally excited molecules, H2(v), that are believed to be the primary origin of H- within volume-type ion sources. We use the same apparatus and approach, but add a Raman shifter to change the laser wavelengths to those appropriate for (2+1) REMPI of H2(v).

A method for electron suppression in plasma volume sources of H- ions is descibed. A theoretical model of the suppressor effect is proposed and the results are compared with experiment.

The basic features of low energy (≤ 200 keV) electrostatic accelerators for negative ion beams are described. It is shown how transmitted current and divergence depends on various parameters. Features which are specific to negative ions, such as the extraction and handling of electrons together with the problem of collisional neutralisation of the ions due to gas flow from the source, are emphasised.

Results are presented from a pulsed volume ion source used to produce high current beams of H-. The data presented here concentrate on the dependence of beam current on extraction voltage as one of the limiting factors in achieving high current beams. Peak currents reported here range from 50 to 60 mA at energies of 80 kV. Analytical models developed for positive ion extraction have been used to interpret the data with moderate success.

Attention is directed towards instabilities induced in Hydrogen plasmas due to the presence of the desired H- ions. Preliminary investigations indicate that the low density H- beam causes collective instabilities even at very low beam velocities compared to ion acoustic and ion thermal velocities, provided that the non-ideal properties of the background plasma are taken into consideration. The most dangerous instabilities involve oscillations transverse to the direction of beam extraction, and may thereby degrade the beam emittance.

The Oak Ridge National Laboratory (ORNL) Fusion Energy Division (FED) negative ion development program has concentrated on the use of Penning discharges in a strong magnetic field. Extraction from the discharge has taken place both perpendicular and parallel to the confining magnetic field. The former has received by far the greater development effort. The ion source produced is called VITEX for Volume Ionization with Transverse EXtraction. Attempts to extract parallel to themagnetic field have recently begun and the ion source principle has been labeled Ringatron. A plan view of a Penning discharge as used in VITEX is shown in Fig. 1. An elevation view is shown in Fig. 2. A thin sheet plasma of thickness D(1) is formed with energetic electrons of density on the order of 1013 cm-3. A secondary cold plasma of thickness D(2) is formed in the region immediately surrounding the energetic electron sheet due to diffusion. Negative ions formed in the discharge may be extracted transverse to the magnetic field with suitable electrodes as shown in Fig. 2. Probe studies coupled with geometrical variations on the thickness of D(1) and D(2) indicate that the two chamber model of Hiskesl is a valid production mechanism. A variation to the extraction geometry of VITEX is to take the sheet plasma of energetic electrons and wrap it around into a cylinderical annulus as shown in Fig. 3. Thus, excited molecular neutrals are fed into the core of the discharge from the entire annulus. The core is similar to the region D(2) of the VITEX geometry. Negative ions produced in the core are free to drift along the magnetic field to the extraction electrodes at which point they are accelerated parallel to the magnetic field. This source is called Ringatron due to the ring or annular nature of the discharge.

Compared to conventional low intensity ion sources, high intensity particle beams, derived from pulsed, positive- or negative-ion sources, may offer distinct benefits for applications in beam heating and current drive in fusion reactors, inertial confinement fusion, and space defense. In recent work at the University of California, Irvine we have shownthat pulsed, magnetically-insulated diodes can produce intense beams of positive or negati i ve ions with current densities in the range of (at 0.5 MeV): 10 - 100 A/cm for negative beams and 100 - 1000 A.cm2 for positive beams. The normalized emittance of these intense sources is approximately, ε = 0.02 cm-mrad, which is comparable to existing volume-or surface-ionization negative-ion sources; the beam divergence angle is also comparable, Δθ = 10 mrad. We have also used these same sources to produce intense beams of high atomic number, such as carbon.