Experiments on collective ion acceleration by means of the formation of a virtual cathode have been carried out for a number of years in the Soviet Union and in the United States. Recently, there has been renewed interest in the subject as a possible means of accelerating ions to very high energies. By understanding the physics underlying the acceleration process it may be possible to determine the feasibility of virtual cathode staging for very high energy ion production.

A virtual cathode collective acceleration configuration is examined for use as a pulsed, intense ion injector. Numerical calculations of the coll ective acceleration have shown peak ion energies of several times that of the electron beam energy. This is consistent with a variety of experiments. The late time behavior is to accelerate ions to lower energies but at a higher current. This is characteristic of a reflex ion triode, which produces an ion beam largely charge but not current neutralized. Dynamic neutralization must be achieved if accelerator applications, requiring transport and focusing, are to be realized. Representative applications have been simulated and are discussed.

A new high-voltage scaling based on Kilpatrick's criterion is presented that suggests that voltages more than twice the Kilpatrick limit can be obtained with identical initial conditions of vacuum and surface cleanliness. The calculations are based on the experimentally observed decrease in secondary electron emission with increasing ion impact energy above 100 keV. A generalized secondary-emission package has been developed to simulate actual cavity dynamics in conjunction with our 2½-dimensional fully electromagnetic particle-in-cell code CEMIT. The results are discussed with application to the suppression of vacuum breakdown in rf accelerator devices.

Circular accelerators based upon increasing magnetic flux comprise one of the oldest classes of machines, the Betatron being a prime example. We have found that intense, charged neutralized ion rings can be accelerated to interesting energies by imploding a conducting liner on the field/ring configuration. Because the ring radius is not fixed, we find that the energy scaling varies from linear in the field (p 2 θ ∝ B Z ) at low energies to square root scaling (γ ∝ B &3x000BD; Z ) at relataivistic energies. Simple modeling has confirmed many of the characteristics we have seen in 2½-D fully electromagnetic particle simulations.

Simulations demonstrating effective growth of slow cyclotron waves on a beam exhibiting a scattered distribution of particle velocities are described. No dramatic changes from the cold beam results for the dispersive properties are observed, but significant modifications of radial eigenmode structure appear.

The PHERMEX facility consists of a 50-MHz standing- wave linear accelerator. Electrons with up to 550-keV kinetic energy are injected into the accelerating column by a hot-cathode electron gun. The present lO0-mm-diam cathode has been in operation since 1963.1 The gun has performed extremely well since that time, and to date there have been no substantial design changes. Early gun experiments involved measurement of current, voltage, and beam distribution; however, no emittance measurements had been made. As part of the PHERMEX upgrade program, it was felt that this gun design could be extended to higher perveance. Early experiments indicated that it was necessary to maintain beam quality to transport the beam through the three PHERMEX accelerator cavities.

Growth of large amplitude coherent cyclotron waves on unneutralized relativistic electron beams has been studied analytically and numerically. The mechanism for growth is the unstable coupling of helical waveguide modes with relativistic electron beam cyclotron waves. Approximate analytic growth rates are found to be in good agreement with the exact numerical solution of the linearized plasma equations on a self-consistent cylindrical beam equilibrium. Particle simulations performed in cylindrical geometry quantitatively confirm the theory. The calculations were conducted in both an infinite medium and realistically terminated configuration. In the latter, matched impedances on the helix were required to reduce transients and unwanted reflections to tolerable levels. Saturation in both cases is due either to convection or, in sufficiently long waveguides, growth of the wave until it physically intersected the helix.

Slow cyclotron beam waves provide an attractive means for accelerating ions in intense relativistic electron beams. 1-3 There are several characteristics of this mode which lend themselves directly to collective acceleration. First and most important is the field magnitude which can be generated in current electron beams (I = 10-50 kA; γ = 6 - 10) with cyclotron waves. Fields between 10 5 - 10 6 V/cm seem achievable in the near term. These can never reach the levels of beam front mechanisms, such as virtual cathodes, but easily exceed conventional ion acceleration sources. Second, the cyclotron phase velocity can be controlled externally and can be reduced far below that of the other collective mode, the slow space-charge beam wave. This facilitates ion loading and permits operation in low velocity regimes that are difficult even for conventional accelerators. An extreme limit of this is the commonly observed zero-frequency cyclotron wave, which possesses a phase velocity of zero in the laboratory frame. Use of this mode has been explored by us and will be discussed later. Finally, the slow cyclotron wave is moderately insensitive to transverse beam spread, or scatter. This relaxes a constraint on beam quality.

The possibility of constructing compact, highcurrent traveling wave ion accelerators through employment of relativistic electron beam collective modes has been suggested often in recent years. A new mechanism based upon temporal modulation of the beam kinetic energy has been studied by us, using both analytic and numerical tools. Preliminary linear studies for the slow cyclotron beam mode indicate that such a mechanism can yield efficient acceleration to ion velocities of roughly 0.5 c in a simple, straight-walled waveguide. Numerical simulations have been performed to verify the theory and investigate nonlinear wave/particle interactions.

Excitation and suppression of zero-frequency cyclotron waves on a relativistic electron beam in a parameter regime appropriate to autoresonant collective ion acceleration are examined through numerical simulation. The excitation mechanism considered is that of anode shorting, while suppression is achieved through a combination of nonadiabatic change in the guide magnetic field and the introduction of a finite radial velocity divergence in the beam. Some brief remarks are included concerning the effect of a scattered beam distribution.

Generator and diode problems have reduced electron beam parameters for the proof-of-principle autoresonant collective ion acceleration experiment from 3 MeV, 30 kA, and 200 ns to 2.25 MeV, 15 kA, and 100 ns. This reduction limits acceleration of hydrogen ions to about 4 MeV, if present experimental plans are followed. The output ion energy can be increased to a more acceptable level by reducing the electron beam radius at the diode and by accelerating ions in a weaker guide magnetic field. Analysis further suggests that preacceleration of ions is necessary for efficient trapping in the slow cyclotron waves. The maximum ion current is of order 15 A.

A plan to address physical questions of interest for exoatmospheric military applications of intense neutralized plasma beams is described. After a brief review of earlier work relevant to this matter and a detailed explanation of why such work cannot answer questions of present interest, a plan employing interactive application of several numerical and analytic techniques to treat relevant phenomena occurring on the various rather disparate time and length scales involved is suggested. The first part of the study would determine the macroscopic features of beam propagation through calculations effected with a magnetohydrodynamical numerical code. Classical transport coefficients would be employed in this initial phase. Using information thus gained concerning gross charge and current distributions, particle-in-cell simulations would be initialized to study those microscopic, phase-space-dependent phenomena which can alter the phenomonological transport coefficients appearing in the fluid description. Insight thereby gained concerning anomaous, collectively induced transport effects would then be applied to yield a refined, accurate description of the macroscopic aspects of neutral plasma beam propagation. Personnel and computational resources available at the Los Alamos Scientific Laboratory are described. Results of a very preliminary particle-in-cell simulation of a neutral plasma beam propagating across a magnetic field are presented.

The Autoresonant Accelerator (ARA) offers great promise for collective ion acceleration provided large amplitude cyclotron waves can be generated with long coherence scales and controllable propagation characteristics. Numerical simulations have been performed to examine cyclotron wave growth in a helical slow-wave structure. No inhibition of growth was observed, short of an intrinsic space charge limitation. Extraction of such waves from the amplifying section through realistic terminations has been performed. The radial structure and propagation of these large, extracted cyclotron waves has been studied and comparisons with linearized waves have been drawn. The effect of nonlinear wave properties on ARA designs are presented.