When a plasma interacts with a material surface subject to an applied voltage, a sheath results. A one-dimensional model, specific to the magnetized ion sheath, is developed and applied to the radiofrequency (RF) sheath which forms near the Faraday screen of an ion cyclotron heating antenna. The RF sheath rectification of the applied voltage is shown to provide a large DC potential drop which can accelerate tons and cause sputtering. Numerical estimates of the high-Z impurity influx are compared with fast wave experiments, and it is concluded that the DC acceleration mechanism is a plausible explanation for the observed high-Z impurity release. Means of controlling the sputtering are examined, including operating with low densities at the Faraday screen.

It is known that tokamaks display a second region of stability to ideal magnetohydrodynamic (MHD) internal modes. An important determining factor for MHD properties is the radial profile of toroidal current. Here it is shown that in a low aspect ratio tokamak with high on-axis safety factor (qo ~ 2) and high shear a path to high beta can be obtained that remains completely stable against ideal MHD modes. By maintaining high shear this scenario avoids fixed boundary instabilities for both high and low toroidal mode numbers for beta values well above the Troyon limit (stability was tested up to c = 1.4, Q = 10.8% ). For a close fitting wall (awaul/aplasma ~ 1.2) this configuration is also stable to low toroidal mode number balloon-kink modes. I

Particle simulations have been made of an infinite plasma slab, bounded by absorbing conducting walls, with a magnetic field parallel to the walls. The simulations have been either one dimensional or two dimensional, with the magnetic field normal to the simulation plane. Initially, the plasma has a uniform density between the walls and there is a uniform source of ions and electrons to replace particles lost to the walls. In the one-dimensional (1-D) case, there is no diffusion of the particle guiding centers and the plasma remains uniform in density and potential over most of the slab, with sheaths about a Debye length wide where the potential rises to the wall potential. In the two-dimensional (2-D) case, the density profile becomes parabolic, going almost to zero at the walls, and there is a quasineutral presheath in the bulk of the plasma, in addition to sheaths near the walls. Analytic expressions are found for the density and potential profiles in both cases, including, in the 2-D case, the magnetic presheath resulting from a finite ion Larmor radius, and the effects of the guiding center diffusion rate being either much less than or much greater than the energy diffusion rate. These analytic expressions are shown to agree with the simulations. A 1-D simulation with Monte Carlo guiding center diffusion included gives results that are in good agreement with the much more expensive 2-D simulation.

The effect of the addition of a cooler ion component on the drift cyclotron loss cone instability was investigated using the local approximation linear dispersion relation. The parameters considered were mirror ratio, density gradient, density (including finite β), ratio of cool to hot ion temperature, and type of cool component (Maxwellian or loss cone). Several different regimes of the parameter space were defined and mapped, according to the characteristics of the instability (e.g., source of free energy, critical density gradient, range of unstable wavenumbers, maximum growth rate). The most effective stabilization was found to occur for addition of a small Maxwellian cool component with density ratio (nc/nh) ≳3.4[(ah/n)(dn/dx)]3/2 (R−1/2−R−1)1/2 and with temperature ratio satisfying 1≳ (Tc/Th)3/2(nh/nc), (R−1/2−R−1)−1≳0.43; ah is the hot ion Larmor radius, R is the mirror ratio, nc and nh are the cool and hot ion densities, and Tc and Th are the cool and hot ion temperatures. The stabilization allows a larger critical gradient, (ah/n) (dn/dx) ≈R1/4 (ωci2/ωpi2+me/mi)1/2 , and a larger minimum unstable wavenumber kah≈R1/4(ωci2/ωpi2 +me/mi)−1/2, than that with no cool component. The instability was also studied numerically for (ah/n) (dn/dx) ∼1, and the minimum nc/nh required for this stabilization was consistent with the density ratio used in the 2XIIB experiment at Livermore. For β≳0, even smaller amount of cool plasma is needed for this stabilization.

A magnetostrictive water pump using Terfenol-D has been designed and built, achieving a flow rate of 15mi/sec at 5 psi, using 41 watts. This is a higher flow rate and lower pressure than previous magnetostrictive pumps. The pump is 6" long and 3.6' in diameter. A model ofpump performance has been developed, including valve inertia which limits the drive frequency, and trapped air in the chamber, which can reduce the flow rate and make the pump noisy. Methods have been developed to eliminate trapped air. The pump uses a hydraulic stroke amplifier, which turned out much stiffer axially than it was designed to be. This has adversely affected pump performance, because of fmite Terfenol compliance and fmite housing compliance. With a stroke amplifier of optimal stiffness, and with better quality Terfenol, the pump should be able to achieve a flow rate of 30 mllsec at 5 psi, consuming 25 to 35 watts. Although the power is more than would be needed by a piezoelectric pump of the same performance, a Terfenol-D actuator offers important advantages, including low voltage and no known fatigue mechanism. Furthermore, much ofthe modeling would be relevant to a piezoelectric pump as well.

A uniform plasma, with a thermal core and a warm ring, in velocity space, is examined for instabilities. The parameters are the ratio of ring to core densities, the ratio of core thermal velocity to ring peak velocity, and the ring thermal velocity. The boundary between stable and unstable distributions is given, as dependent on these parameters, as obtained from the Penrose criterion. In addition, the upper and lower bounds on the wavenumbers (kmax, kmin) are found for typical distributions, as well as growth rates, ωi(k). For either dense ring or dense core distributions, the distributions are unstable for (vt/vr) ≲0.1, independent of the thermal spread of the ring, where vt is the core thermal spread, and vr is the ring peak speed. For more equal ring and core densities, the stability boundary moves to larger values, dependent on the thermal spread of the ring. Growth rates of ωi≈0.01 to 0.1 ωp are typical. Application of these results to magnetized plasmas is given.

Plasma production and heating in the central cell of the Tara tandem mirror [Nucl. Fusion 22, 549 (1982); Plasma Physics and Controlled Nuclear Fusion Research, 1986, Proceedings of the 11th International Conference, Kyoto, Japan (IAEA, Vienna, 1987), Vol. 2, p. 251] have been studied. Using radio-frequency excitation by a slot antenna in the ion cyclotron frequency range (ICRF), plasmas with a peak β⊥ of 3%, density of 4×1012 cm−3, ion temperature of 800 eV, and electron temperature of 75–100 eV were routinely produced. The plasma radius decreased with increasing ICRF power, causing reduced ICRF coupling and saturation of the plasma beta. About 70% of the applied ICRF power can be accounted for in direct heating of both ions and electrons. Wave field measurements have identified the applied ICRF to be the slow, ion cyclotron wave. In operation without end plugging, the plasma parameters were limited by poor axial confinement and the requirements for maintenance of magnetohydrodynamic stability and microstability.

The stability of plasmas produced by radio-frequency heating in the ion cyclotron frequency range (ICRF) has been studied in the central cell of the Tara tandem mirror [Nucl. Fusion 22, 549 (1982); Plasma Physics and Controlled Nuclear Fusion Research 1986, Proceedings of the 11th International Conference, Kyoto (IAEA, Vienna, 1987), Vol. II, p. 251]. Ion cyclotron wave excitation by a slot antenna provided stability against macroscopic plasma motions in an axisymmetric configuration. The maintenance of macroscopic stability depended on the ICRF power, gas fueling rate, ion cyclotron resonance location, and ω/ωci at the antenna location. The ICRF ponderomotive force model is consistent with many of the observed stability features and predicts that the E+ component of the ion cyclotron wave was responsible for the stabilization. The Alfvén ion cyclotron microinstability was observed when the plasma β⊥ and anisotropy were sufficiently high. Magnetic probe measurements of the unstable mode identified it as an ion cyclotron wave and the instability threshold was within a factor of 2 of the theoretical value.