SiO2 LAYERS GROWN UNDER B IAS 141

The experiments with positive bias during drift are still in a prel iminary stage and more work has to be done. It can be stated, however, that the drift effect is not inherent in SiOe (1). Neither does it seem to depend on the contact metal. Some influence of the metal was found here, but most of it could be corre-lated with the difference of evaporation conditions. Conclusions The conclusions are: 1. N-type inversion in oxidized surfaces xceeding a density of about 4 x 10 n cm-2 is caused by positive impurity ions. It is not directly caused by the pres-ence of water during oxidation. Most of the ions are highly mobile at elevated temperatures and c~: be removed by an electric field.: 2. Of the different models proposed for MOS drift effects, alkali ion drift (1, 2) is the likeliest. Removal of these ions from SiO ~ layers wil l probably result in dr i ft- free structures. 3. The bias oxidation technique described here is capable of reducing the density of surface donors after steam oxidation reproducibly to the range of 3.5-6 x 1011 cm-2. Oxides produced this way exhibit greatly reduced drift effects. Acknowledgment