In this work, diffusion and precipitation of Si in Al films and growth of Si epitaxial structures in solid Al were studied. Diffusivity and solubility of Si in Al films of an integrated circuit structure were investigated by means of electron microprobe analysis at temperatures between 360°C and 560°C. The thin Al films were also analyzed for the final location of Si precipitation with the electron microprobe.
According to the results of precipitation experiments, the investigation of the growth of epitaxial structures of Si in solid Al per cryst. The morphology of these structures depended on the amount of Si deposited in the Al film, the annealing conditions, the stress in the metal film, and the crystal orientation of the substrate surface. A study of the Si-Au-Cu ternary solid solution. was performed considering the more extensive solubility of Si in Cu compared to that in Au.
Investigations on Si-Au-Cu phase diagram 72 2. INTERFACE EFFECTS IN THE SOLUTION OF Si IN THIN 74. DIFFUSION OF Si IN Al AND SOLID PHASE GROWTH OF EPITAXIAL Si STRUCTURES IN Al.
INTRODUCTION A. Genera 1
A problematic aspect of the above facts was that the dissolution pattern at the contact intersections was uneven and such that it produced depressions at the contact periphery, i.e. the Si-SiO2 boundary (Fig. l)o The pit led to short circuits between thin planar diffuse layers, or at best caused excess leakage currents and soft breakdown characteristic of transitions. This problem is eliminated in the current technology by incorporating Si into the Al films to suppress the Si dissolution. SEM image of the Si surface after extensive Si dissolution in Al has taken place. Note the faceting of some pits, reflecting crystalline symmetry of the substrate, and different rates of dissolution of different crystalline planes.
This is done by irradiating the sample with an energetic electron beam, focused to lµ diameter or less, to generate the characteristic X-rays of the elements. The deposition sites of Si dissolved in Al films were investigated by means of the electron microprobe analysis in structures similar to those used for diffusion studies. Si in solution, away from crystalline Si substrate, precipitated clumps near the free surface of the Al film.
Most of the Si remained in solution upon quenching, which corresponded to cooling of the sample. Electrical characterization of the above growth structures indicated that they were p-type and that they formed a rectifying bond with the underlying n-type substrate.
The beam voltage and Al film thickness were such that the electron beam sampled the top half of the film and therefore did not excite the substrate Si. However, our values show more scatter because of the uncertainty in locating the oxide excision boundary and because Al on the oxide near the boundary · had to be sampled. Si in solution in Al over bare Si tended to precipitate [8] and thus diffuse out of range of the electron beam.
The faster diffusion of Si in Al films is due to the disordered structure of the evaporated matrix. The error was possible due to uncertainty in the determination of the origin for reasons mentioned in the "solubility II section. Preparation of the samples was identical except for annealing to that for diffusion studies as explained in the previous section.
Some of the specimens were annealed at 540°C for 5 minutes and then allowed to cool in the oven at an average rate of 3°C to 1so0c. The presence of vaporized Si, which is known to be amorphous [67] in the Al film prevented any noticeable dissolution from the wafer substrate. The evaporated layer dispersed and saturated the Al film due to the higher free energy of the amorphous state.
This may reflect effects of deviation of the substrate from (111) orientation [68] and interfacial energy of Si-Al crystals. Often, vertical membrane-like growths developed in the center of the clefts~ which is a growth feature also observed by other researchers [43]. The increase is necessary to prevent charging of the surrounding oxide by the electron beam.
The driving force behind this mechanism is the higher free energy of amorphous Si compared to crystalline Si. The morphology of the outcrop is probably affected by surface cleanliness and defects. In fact, in the growth of epitaxial layers with liquid or vapor phase epi, limited dissolution of the substrate is allowed to obtain a clean surface at the atomic scale.
Fine-dimensional control is also facilitated due to the absence of the surface tension of the liquid phase. A variety of growth structures, including mesas, faceted structures, large flat plateaus, were observed depending on the Si content of the Al film and annealing conditions.
The voltage contrast mode of SEM is based on the fact that applying a small bias to the specimen affects the efficiency of the secondary electron collection, since the collector is only a few volts positive with respect to the specimen. Microscopic observation of samples revealed that the topology of the printing plate was replicated on the sample. SEM analysis indicated that raised structures had formed on the sample that corresponded to grooves on the printing plate.
Square-shaped pits were seen between them, corresponding to the parts of the plate that were in contact with the film. from the figures, the surface is not smooth; however, there was no ambiguity in the relative height of the structures. The difference in contrast upon application of reverse bias is an indication of an n-p junction between the substrate and the raised growth structure. The negative potential of the latter structures results in greater collection of secondary electrons in the SEM, which gives them greater brightness. The position of the p-type growth structures relative to the notches in the pressure plate indicates that the growth must have been dominant.. a) Schematic of the waffle-iron compression plate and plain specimen.
Growth structure (cruciform) on the sample after pressure-temperature treatment, as shown by SEM after removal of Al film. On the other hand, dissolution of substrate Si must have been dominant in high-pressure regions. These results can be explained by considering the atomic volume of Si in its crystalline form (20.2i3) and its partial molar volume in Al.
The compression of the sample promotes dissolution as it results in a decrease in the Al lattice parameter and a decrease in the volume occupied by Si atoms. Alternatively, strain should promote the precipitation of Si because it results in the formation of a more voluminous phase and relaxation of the lattice parameter of Al to the higher value in its pure state. The mechanical stress as a driving force for the transport and growth of epitaxial Si structures in Al films was investigated.
It was shown that mechanical stresses are important and effective in the growth of these structures. The stresses needed to effect growth may be only a fraction of what is needed to cause plastic deformation of Al.
APPENDIX 1
PART II
INTRODUCTION A. General
The present investigation showed that the dissolution is non-uniform across the plane of the Si surface. In the present investigation, we also studied the solubility and diffusivity of Si in Au films using electron microprobe analysis. The solubility of Si in Au appears to be below the detection range of some currently available techniques.
In the present study, the ternary solid solution at the Cu-Au-rich end of the phase diagram was studied. The Si substrates were cleaned in organic solvents, rinsed, and briefly immersed in HF before the Au film was evaporated. Microprobe analysis of the silicide growth gave an indication of the above results: Fig.
SEM analysis of the Si surface showed that pits (Fig. 8); channels in the form of gaps 0. The raised ·-structures, as observed on the side of the Au f"ilm in contact with. SEM image of the Au surface of a sample that was annealed in an oxidizing atmosphere.
This treatment should produce significant recrystallization of the Au film [33]; However, the Si dissolution behavior did not change. The microprobe analysis of the growth rate of the silicide on the Au film did not result in quantitative data on the diffusion of Si in thin Au films. This suggests that texture of the Au film or thermal stresses in it may affect the diffusion of Si.
These provide pathways for rapid diffusion due to lowering the activation energy for diffusion. The localized dissolution was attributed to the effect of the surface adsorbed impurity or native oxide layer. Analysis of the alloys with low Si concentrations in the two-phase region indicated that micrographic analysis was more sensitive than powder pattern analysis.
However, careful examination of the powder pattern did not reveal the formation of the ordered solid solution. The phase boundary of the binary solid solution in the Cu-Au-Si system is determined to be 349+1°C.
