Paper 8.2

Chemical Transport of Alumina in metal halide lamps - Thermodynamic Modelling and experimental data


Markus T., Hilpert K.


Institute for Materials and Processes in Energy Systems (IWV-2) Research Centre Juelich, D-52425 Juelich, Germany


Chemical transport of alumina is observed as the main phenomenon relating to the corrosion of metal halide lamps with ceramic discharge vessels. These containers which are made of poly crystalline alumina (PCA) are filled with salt mixtures partly evaporating under operating conditions. Typical constituents of the melt are alkali halides, such as NaI and rare earth metal halides, for example DyI3, TmI3, HoI3. The radiation of the gaseous metal halides contribute significantly to the visible spectrum of high intensity discharge lamps. Depending on the melt composition, dissolution of Al2O3 at hot parts of the vessel, chemical transport and deposition of PCA at cold parts is observed [1].


To get a detailed insight into the chemical interactions between the wall material consisting of alumina at the inner surface and the salt mixture, annealing experiments with sealed alumina ampouls containing different salt mixture fills were carried out under isothermal conditions and in a temperature gradient. The annealing experiments show that Al2O3 is transported from the hot part of the ampouls to the cold side.


The determination of reliable values for the thermodynamics of corrosion reactions between the salt mixture and solid alumina are acomplished using Knudsen Effusion Mass Spectrometry. The formation of hetero complexes and oxihalides between salt components and additionally between NaX und AlX3 (X=I, Br) (the latter arising from reaction between TmX3 and alumina) could be monitored and the enthalpies and entropies of formation were determined. Thermodynamic data (Kp, DH, DS) describing the stability of the complexes NaAlX4(g), NaLnX4(g) and Na2LnX5(g) (Ln= Tm, Dy) and the oxihalides AlOX were determined from the measured partial pressures.


The obtained thermodynamic data were assessed and implemented into a FactSage[2] salt database. Model calculations were carried out using this database. As results for example the gas phase composition over complex condensed phases are calculated.


Using the FactSage Equilib algorithm the observed chemical transport inside the annealing ampouls is modeled in dependence of the applied temperature gradient and the composition of the salt melt and the corresponding gas phase, respectively. The transport model was developed by dividing the ampoul into two parts. One is called source side; this is the hot part of the vessel, where the alumina is depleated. The other is the sink side where the Al2O3 is beeing deposited after it has been transported from the hot to the cold part of the container. For the calculations local thermodynamic equilibrium is assumed in each of the two parts. The input parameters for the calculations are the temperatures at the source– and the sink side, the amount and composition of the metal halide salt mixture as well as a certain amount of alumina at the source side. The result of the calculations deal with the gas phase compositions and with the amount of alumina which is condensed at the sink side of the vessel.


Using this transport model under concideration of our determined thermodynamic data we could explain why Al2O3 is beeing transported when a condensed salt melt phase is present and when the salt mixture contains alkali metal halides as well as rare earth metal halides.










Combining the experimental thermochemical methods with model calculations  results in a detailed analysis and characterization of real lamps. In this way, modern light sources are optimised with respect to light technical properties and life time performance.



[1]       T. Markus, Ber. Forschungszent. Juelich, Juel-3955, ISSN 0944-2952, 1-167 (2002)

[2]       FactSage Thermochemical Software and Databases, C.W. Bale, P. Chartrand, S.A. Degterov, G. Eriksson, K. Hack, R. Ben Mahfoud, J. Melançon, A.D. Pelton, S. Petersen, CALPHAD 26(2), 189-228 (2002)



Corresponding author: