ABSTRACT
INTRODUCTION Investigations of fluorescent lamps (FL) are often focused on electrodes because the lifetime of lamps is limited by the electrode durability. In general, a commercial electrode system consists of a tungsten coil coated with a work function reducing emitter mix of alkali oxides, such as BaO, SrO and CaO. The spot temperatures, during steady state, typically are in the range of 1200 - 1400 K. Until now investigations of electrodes have been directed mainly toward preheated ignition [1-4]. Driven by the desire of saving costs, electronic control gear (ECG) without preheating circuits are becoming more common. A FL driven with such a 'cold start ECG' starts in a glow mode. In this mode, which last typically for tens of ms, the discharge current is relatively low (few mA), but the lamp voltage may exceed 500 V because of the very high cathode fall. This causes high energy ion bombardment of the electrode which heats the electrode, and induces a transition from glow to arc mode. In this mode the electrode emits thermionically and the cathode fall drops to the 12 - 15 V range. Unfortunately, the high energy ion bombardment during the glow mode leads also to intense sputtering of electrode material, including tungsten as well as emitter. Thus, cold started FL often suffer from early failures due to coil fracture. Haverlag et al. [5] have shown that coil breakage is caused by tungsten sputtering at one of the emitter-free ends mainly during the glow-to-arc transition. Our investigations on both linear and compact fluorescent lamps by fast emission spectroscopy and high-speed video observation support that behaviour [6]. Recently, we additionally have employed the ultra-sensitive method of laser-induced fluorescence (LIF). This technique is particularly well-suited to determining absolute population densities of neutral and singly ionised atoms of liberated electrode material (e.g. of Ba [3, 4]) and, in certain cases, excitation temperatures [7]. Its experimental complexity is offset by the multitude of advantages it offers, such as point-wise spatial resolution instead of integrations over lines of sight, and independency of plasma parameters. Since ne and Te typically have steep gradients near the electrode, this is an especially useful feature. The main goal of this work is to investigate the W erosion and, thus, to improve the basic understanding of the underlying sputtering process. In addition to FL, our investigations have been performed also on hollow cathode lamps (HCL). These are useful because they provide a variable source of sputtered W atoms, and can serve as tuning tools for precise adjustment of the laser radiation.
