Metal halide lamps typically have cold fills of tens to a few hundred Torr of a rare gas and the vapour from the dosing of a metal halide solid and mercury. Breakdown and starting of the lamp occurs following application of multi-kV pulses across electrodes separated by a few centimetres. Restarting of warm lamps is often problematic as the available voltage is insufficient to break down the higher pressure (>many atm) of metal halide vapour. In this paper, fundamental processes during breakdown in cold and warm, idealized metal halide lamps in mixtures of Ar and Hg are investigated using a two-dimensional fluid model for plasma transport. We find that the capacitances of the walls of the discharge tube and adjacent ground planes are important in determining the breakdown voltage and avalanche characteristics. The prompt capacitance represented by, for example, external trigger wires provides a larger E/N to sustain ionization early in the avalanche. This effect is lost as the walls charge and shield the plasma from the ground planes. More rapid breakdown occurs in slightly warm lamps having small vapour pressures of Hg due to the resulting Penning mixture. Warmer lamps, having larger mole fractions of Hg, have less efficient breakdown as the increase in momentum transfer of the electrons is not offset by the additional ionization sources of the Penning mixture.

Electrical breakdown of cold (room temperature) metal-halide arc lamps typically occurs through the fill of a rare gas (at a pressure of tens of Torrs) and the vapour produced by the metal donor. Restarting a warm lamp is often made difficult by the high pressure of the metal and metal-halide vapours. To reliably start cold lamps with a minimum voltage and a minimum sputtering of the electrodes, and to restart warm lamps that have a high pressure of the metal and metal-halide vapours, auxiliary sources of ionization are often used. As a point of departure for the study of these processes, measurements of formative breakdown times were made in a cylindrical discharge tube resembling a compact polycrystalline alumina envelope metal-halide lamp. Breakdown times were measured for Ar/Xe gas mixtures at total pressures of 10-90 Torr and biases up to 2 kV applied to a 1.6 cm gap. The data provide a knowledge base for a companion computational investigation. We found that breakdown times generally decreased with small admixtures of Xe in Ar (5-15%) and increased with larger admixtures. We attribute these trends to the changing shape of the tail of the electron energy distribution.

No abstract