The rate of reaction between Si (100) surfaces and tungsten films deposited by rf diode sputtering depends on the preparation of the silicon surface. If rf substrate bias is used to clean the silicon, then the rate of reaction in the temperature range 700–850 °C is independent of time, with an activation energy of 3 eV/mole W. The native oxide layer between the silicon and tungsten, that exists when sputter cleaning is not used, can act as a barrier to WSi2 formation. In this case, the time-independent region is preceded by a period when the reaction rate increases with time. The rate is then controlled by two-dimensional spreading of discontinuous WSi2 regions that originate at sites where the reaction barrier can be penetrated. After a continuous WSi2 layer is formed, additional growth can produce a stage where the increased path length for silicon diffusion causes the transport step to control the over-all rate of the reaction. Quantitative models are presented for each of the three stages in the reaction. The models explain some of the macroscopic observations made on reacted layers.

ANODIC FILMS ON ZINC 1659 microprobe x-ray analyzer, while David Fisher and John Bartlett did the atomic absorption analysis. The identification of the anodic films by x-ray and electron diffraction was the work of William Dorfeld and Louis Osika.

Oxide films on aluminum can be sputtered off in oxygen or in argon glow discharges, when a positive bias is applied to the specimen. It is shown that this phenomenon limits the thickness of films that can be grown by plasma anodization. The sputtering does not necessarily require negative ions, since these are absent in the argon discharge.

Thin films of hafnium nitride and tantalum nitride with resistivities of 1010 and 104 ohm‐cm, respectively, have been deposited by rf + d‐c reactive sputtering. By sputtering hafnium‐tantalum mixtures in pure nitrogen at 7.0×10−2 normalTorr , the complete range of resistivities is obtained, with a linear increase between 105–09 ohm‐cm, corresponding to a relative hafnium content of 40–70%. For a fixed Hf:Ta ratio, the resistivity, deposition rate, and nitrogen content of the films depend on the pressure and substrate temperature. For pure tantalum a dinitride false(TaN2false) is deposited, but for pure hafnium the dinitride has not been obtained. An alloy nitride containing 60% Hf can be represented by the formula Hf0.6Ta0.4Nx , where x increases from 1.2 to 1.4 as the substrate temperature increases from 225° to 400°C. A sequence of reaction steps on the substrate surface is proposed to explain the deposition rates observed in reactive sputtering.

Manganese oxide thin films deposited over the Ta205 layer in tantalum capacitors are known to improve yield and reliability. In this investigation reactive sputtering is used to prepare manganese oxides. The results show that films with resistivities below 1 ohm-cm can be produced, but the interdependence of oxygen partial pressure, power input, and substrate temperature limits the deposition rates that can be obtained without producing oxide phases with higher resistivities.With d-c sputtering at 4.0 kV, Mn203 is produced as the oxygen pressure increases from 3 X 10 -2 to 4 X 10 -2 Torr. This corresponds to a substrate temperature increase from 275 ~ to 360~ and a 50% increase in deposition rate. The complexity of the interactions in both d-c and rf reactive sputtering is further illustrated by the maximum in substrate temperature obtained with argonoxygen mixtures containing 0.5% O2. For reactive sputtering of materials similar to Mn-O, where several phases and variable stoichiometry are possible, it is therefore necessary to determine experimentally the conditions of power density, pressure, and substrate temperature that give the maximum deposition rate for each phase.