Table of contents

The principle of laser-induced fluorescence (LIF) is first reviewed and the necessary hardware for the diagnostic is described. Following this is a review of the present state of the art, including what has so far been measured and discussed, and what is currently being developed. LIF is the technique in which atoms, molecules and ions are excited from low-lying energy levels to higher levels, and the resulting fluorescence observed when they decay back to the lower levels. Because spectral lines of different species very rarely overlap and the cross sections for the resonant excitations are large, the technique is very powerful for observing a particular species with very high sensitivity. This characteristic offers a unique feature for the understanding of particle behaviour in plasmas, and if Stark and Zeeman effects are large enough to be detectable, for measurements of electric and magnetic fields. The key hardware for the technique is a tunable laser, whose wavelength is tuned to transitions of the species to be detected. Various tunable lasers are described, combined with techniques for narrowing and scanning of the laser output spectrum for spectral profile measurements.

The nonlinear behaviour of tearing modes in a plasma with density and temperature gradients is reviewed. The effects of inhomogeneities can essentially modify the evolution of small scale islands from that predicted by Rutherford theory. Plasma gradient effects provide the mechanism for island excitation even in situations when the linear tearing mode stability parameter Delta is negative. The magnetic islands are sustained by the differential response of electron and ion components of a plasma in a fluctuating electric field. Such nonlinear magnetic islands are not related to the linear instability of drift-tearing modes. The nonlinear equations describing the evolution of the width and frequency of the rotating islands are derived. In the framework of one-fluid MHD, the general equation for a neighbouring equilibrium in a finite pressure plasma is considered. The dynamics of unstable m=2 and m=1 magnetic islands based on this equation is described. The quasilinear saturation of island growth in a finite pressure plasma leads to the bifurcation of the island type equilibrium into states without islands. A new evolution equation of m=1 islands is derived. For monotonic safety factor and temperature profiles this equation predicts saturation of the m=1 island growth.

Deterioration of neutron yield in a low-energy plasma focus operated by a single 32 mu F, 15 V, (3.6 kJ) capacitor is observed. When the cumulative discharge energy over successive shots across an insulator sleeve approaches 1.6 MJ, the neutron yield from the device starts deteriorating. The insulator sleeve, when examined, is found to have a approximately 3 mu m thick layer of copper deposit. The contaminated sleeve surface appears rough with a grain-type structure. It is therefore concluded that the degradation of neutron yield in the low-energy device occurs due to Cu deposit on the insulator sleeve surface. The situation may improve if low-sputtering-rate conductors are employed for the electrodes of the device.

Soft X-ray spectra have been measured in the RTP tokamak both during the ohmic phase for densities 2*10¹⁹<or=n_e(0)<or=7*10¹⁹ m^-3, and during the ECRH phase at a density n_e(0)=2*10¹⁹ m^-3 for several power levels. Large deformations of the soft X-ray spectra have been observed down to energies as low as hv approximately=4 keV making the determination of the electron temperature difficult. The soft X-ray spectrum of the Ohmic phase has an almost flat tail, while the spectrum during ECRH shows a strong deformation at energies several times thermal and has a moderate tail. Simulations of the experimental data have been performed using a (2+1)-dimensional bounce-average Fokker-Planck code and good agreement has been found. Using the simulated electron velocity distribution, some important quantities like the energy of the suprathermal population and the change in conductivity with respect to the neoclassical value have been determined.

The current rampdown phase followed by non-disruptive current termination is studied for future alternating current (AC) operations in the STOR-M tokamak. The plasma current is smoothly reduced to zero without disruptions with the rampdown rate of around 1.5 kA ms^-1. The plasma density in the current rampdown phase is controlled by gas puffing. The plasma current decreases, maintaining the relationship 1/q_a approximately 0.01 nR/B_t (10^-18 m²T). At current termination, I_p=0, a residual plasma density (0.58-2.9)*10¹⁸ m^-3 is observed with a corresponding Murakami parameter nR/B_t=(0.41-2.1)*10¹⁸ m^-2 T^-1. After current termination, the low-density plasma remains for 5 to 15 ms, which would allow AC operation in the STOR-M tokamak if the radiation barrier does not affect the subsequent current evolution.

Evolutions of the local pressure gradient and the amplitude of the global resistive interchange instability are studied in torsatron/heliotron plasmas. (Torsatron/heliotron: the high-aspect-ratio toroidal helical plasma with magnetic shear and magnetic hill.) Coupled equations for the background pressure profile and a single helical mode are derived by keeping the effects of the thermal transport and ion viscosity. Three characteristic dynamic evolutions are identified; monotonic saturation relaxation oscillation and periodic limit cycle solution. As the heating power is increased, the dynamic evolution changes from monotonic saturation, to relaxation oscillation and finally to the limit cycle. The time average of the pressure gradient is limited by the critical gradient against the linear stability. Dependences of the amplitude and period of the sawtooth oscillation on the heating power are also analysed.

Based on a recent theory for fast-ion redistribution during sawtooth crashes, an approximate analytical model is given for describing the sawtooth induced changes in the emission profile of 14 MeV neutrons generated by triton burn-up. Comparison with measurements of sawtooth-induced neutron emission profile broadening on JET shows good agreement, which indicates that the theory for fast-ion redistribution is consistent with the experimental results.

A multichannel microwave reflectometer system is used to measure the electron density profile at JET. A new measurement technique is described which gives continuous information on the electron profile in time. Profile measurements using swept-frequency measurements are combined with accurate measurements of the evolution of the density using fixed-frequency operation. The analysis of the data can be carried out routinely for each discharge at JET. The method includes estimates for the uncertainties on the measurements, corrections for finite sweep on the data and relativistic effects. The results compare well with other measurements of the electron density in JET.

The scale invariance technique is employed to discuss pressure gradient driven turbulent transport when an Ohm's law with electron inertia, rather than resistivity, is relevant. An expression for thermal diffusivity obtained by Itoh et al. (1992), which has many features appropriate to L-mode transport in tokamaks, is seen to have greater generality than indicated by their particular calculation. The results of applying the technique to a more appropriate collisionless Ohm's law are discussed.

The analysis of charge exchange recombination spectra represents a very challenging problem due to the presence of many overlapping spectral lines. Conventional approaches are based on iterative least-squares optimization and suffer from the two difficulties of low speed and the need for a good initial approximation to the solution. This latter problem necessitates considerable human supervision of the analysis procedure. The authors show how neural network techniques allow charge exchange data to be analysed very rapidly, to give an approximate solution without the need for supervision. The network approach is well suited to the fast intershot analysis of large volumes of data, and can readily be implemented in dedicated hardware for real-time applications. The neural network can also be used to provide the initial guess for the standard least-squares algorithm when high accuracy is required.