Modelling additive transport in metal halide lamps

In 1912 Charles Steinmetz was granted a patent for a new light source. By adding small amounts of sodium, lithium, rubidium and potassium to a mercury lamp he was able to modify the light output from "an extremely disagreeable colour" to "a soft, brilliant, white light". Much later, at the New York world trade fair in 1964 General Electric was the first to introduced a commercial lamp based on the same principle. The light emitting metallic elements are introduced as components of halide salts. Hence, they are called metal halide lamps. The physics behind discharge lamps of this type, however, is still a matter of active investigation. One well-known phenomenon is that, when operated vertically, the metal halides in the lamp tend to demix; the concentration of metal halides in the gas phase is much greater at the bottom of the lamp. Demixing, or segregation as it is also called, has a negative impact on the lamp’s efficacy. It is currently avoided by using lamp designs with very small or very large aspect ratios. Gaining more insight into the process of demixing would allow a broader range of lamp designs with still better luminous efficacies. The demixing is caused by a competition between convection and diffusion. The centre of the lamp must be hot to produce as much light as possible. The walls must stay relatively cool to avoid them weakening and releasing the mercury vapour. Thus, large temperature gradients are present in the lamp, driving convective flows. In the hot centre the molecules are dissociated into atoms. The atoms are smaller and more mobile than the molecules. The atoms are dragged up by the convective currents while diffusing outward. Because of their larger mobility, however, the atoms do not reach the top of the lamp. The result is a larger concentration of metal additives at the walls and at the bottom of the lamp than at the centre and the top of the lamp. This thesis describes the process of demixing in a self consistent and quantitative manner using state-of-the-art computational methods. The competition between convection and diffusion is studied using a variety of models built with the plasma modelling toolkit Plasimo. Using Plasimo allows for the construction of models in a modular fashion. Partial models are used to study the conveci vtive flow as a result of the temperature gradients, the chemical composition as a function of temperature and pressure, and the radiation transport on the lamp. A grand model is formed by combining modules for ray tracing, elemental diffusion, convective flow and the temperature equation. The model result is validated against experiments done by colleagues: Experiments which have been carried out in Eindhoven, at the Argonne National Laboratories in the USA, and in the International Space Station. Cross validation with theoretical work has also been performed. Axial demixing is shown to be the result of the competition between axial convection and radial diffusion. This competition is best expressed by the dimensionless Peclet number. When the Peclet number is approximately equal to unity, axial segregation is strongest. The degree of axial segregation is best expressed by the dimensionless segregation depth t . The largest value of t depends on the element under study and on the position in the discharge where the molecules dissociate to form ions.