A reaction model developed by Hitchman and Kane for semi‐insulating polysilicon (SIPOS) deposition is adapted to the case of silicon dioxide growth on silicon wafers at low temperature and pressure. The model is used to explain the radial nonuniformities associated with the oxide deposition process and the function of a cage surrounding the wafers. The effect of providing openings in the cage for introduction of reactants is examined. The proportionality of oxide growth rate and silane flow rate, as well as a near‐linear dependence of growth rate on wafer spacing, are also interpretable in terms of the reaction model.

Fig. 5. Model of anisotropic etching in evacuated wet system. Effective diffusion barrier was formed by small hydrogen bubbles on aluminum side wall surface. Here, open and filled circles indicated hydrogen and phosphoric acid, respectively, and arrows show the flux of each molecule.shown in Fig. 4 as a function of pressure. This shows that lateral etching uniformity was improved remarkedly with decreasing pressure. That is, the deviation of 0.2 ~m at atmospheric pressure decreased to 0.03 ~m (5%) at 30-100 torr and increased again at still lower pressures. This indicates clearly that evacuated wet etching system offers excellent lateral etching uniformity. Thus, one is able to achieve fine pattern electrode etching with 1.0-1.2 ~m linewidth, by the evacuated etching, if the end point detection technique employing electrochemical potential is used.In wet aluminum etching, hydrogen bubbles are generated on the aluminum film by the chemical reaction with phosphoric acid. Therefore, control of hydrogen bubble removal is important to the improvement of etching uniformity and reproducibility. The fact that the etch rate of aluminum films normal to the wafer surface was constant with time while the lateral (undercutting) etching rate decreased with decreasing pressure suggests the model shown in Fig. 5, which shows the flux of hydrogen bubbles (open circles) and ions of phosphoric acid (filled circles) during the etching. It is clear from pressure and time variations of the etch rate that the removal rate of hydrogen bubbles through the very narrow channel region (1 ~m thick) beneath the photoresist can apparently be restrained with decreasing pressure. Detailed features of this model are reported elsewhere (3). ConclusionsThe consideration of the evacuated wet etching of A1/Si (2%) electrodes resulted in the following conclusions.1. The etch rate of aluminum films normal to the wafer surface was nearly constant at a value of approximately 0.18 ~m/min over the pressure range studied.2. The amount of lateral etching in the grooves decreased with decreasing pressure, resulting in increased linewidth. Typical undercutting of the conventional (isotropic) wet etching was 0.8 ~m. This was reduced to 0.42 ~m at 9 torr. Thus, the etching has changed from isotropic to quasi-anisotropic.3. Uniformity of lateral etching within a wafer was also improved with decreasing pressure. The standard deviation for side etching within a wafer reaches a minimum value of 0.03 ~m (5%) at 30-100 torr. This is greater uniformity than conventional wet and dry etchings.4. Overetching for fine pattern aluminum electrodes can be reduced markedly by precisely monitoring the end point. This leads to 1.0-1.2 ~m linewidth etching including the deviation. ABSTRACTThe reaction of silane and nitrous oxide at high temperatures and low pressures was investigated for a large range of mole ratio of the reactants. For a nitrous oxide/silane mole ratio of approximately 100, silicon dioxide having the stoichiometric refractive index of 1.45 is obtained for growth...

The low-pressure chemical vapor deposition of phosphorus-doped silicon films on oxidized silicon wafers was investigated as a function of phosphine/silane mole ratio, silane partial pressure, temperature, and wafer spacing. The deposition rate decreases, concomitant with increased phosphorus incorporation, as the mole ratio increases. The deposition rate tends to saturate or vary linearly with silane partial pressure for undoped and heavily doped films, respectively. This, together with differing deposition-rate activation energies of 1.5 and 2.0 eV for undoped and doped films, respectively, is indicative of different reaction mechanisms in the two regimes. As the mole ratio increases, the deposition rate becomes increasingly dependent on the wafer spacing and the radial position on a wafer. Because phosphorus incorporation varies inversely with deposition rate, it develops similar dependencies on wafer spacing and radial position. The majority of these observations are interpretable in terms of a model that has been defined for the growth of oxide from the silane-oxygen reaction where the phosphorus and oxygen play analogous roles. The resistivity of annealed films decreases with increased phosphorus incorporation (mole ratio) and for O.5-,um-thick films reaches a minimum value of approximately 440 f.lH em at about 10 21 phosphorus atoms cm --3. The resistivity decreases with increasing deposition temperature which may be attributable to one or a combination of the decreased phosphorus incorporation at higher temperature or deposition rate, or decreased grain size at higher temperature. The decrease in resistivity with increased thickness is attributed to increased grain size with increasing thickness. At least for thicknesses less than O,S/-lm, lower resistivity is achieved by in situ doping than by doping of films subsequent to deposition.

No abstract

The effect of barrier and porous anodizations, either singly or sequentially, on the electromigration resistance of aluminum films was characterized. Samples were tested for current densities in the 4×105−2×106-A/cm2 range and temperatures between 150 and 280°C with the median failure time being used as a basis for comparison of lifetimes. An initial barrier anodization was found to be necessary in order to consistently enhance lifetime. The addition of a porous layer as well led to the best result—an increase in lifetime by a factor of 23 at 227°C. All lifetime increases were attributable to increased activation energy for the electromigration-failure process. An increase of 0.14 eV was obtained for the double anodic structure. Results for structures having a barrier layer adjacent to aluminum were considered in terms of surface sealing of the aluminum and/or response to mechanical constraints of the overlayer. Direct porous anodization resulted in preferential oxidation at grain boundaries, thereby widening these regions, in pitting at grain boundaries, and in general cratering characteristic of this type of anodization. It was concluded that the widening phenomenon was most instrumental in reducing activation energy and lifetime.