Final passivation films

Silicon dioxide films have been an essential factor in the manufacture of integrated circuits from the earliest days of the industry. They have been used as a final passivation film to protect against scratches and to getter mobile ion impurities (when doped with phosphorus). Another application has been as an interlayer dielectric between the gate polysilicon and the aluminum metal-ization. Initially, most such films were deposited in atmospheric pressure systems. In recent years, low pressure processes have assumed greater importance. We will begin by examining the atmospheric process. 

The low-temperature depositions described in the present section can be used for either interlayer dielectrics or final passivation films. Their primary disadvantage is one of film quality, because the process is susceptible to gas-phase nucleation and incorporation of particles into the film. 

When phosphorus is added to Si02, in addition to gettering mobile alkali ions, it tends to reduce the intrinsic tensile stress in such films, thereby reducing their tendency to crack. Both functions are important when the film is used as a final passivation film for integrated circuits encapsulated in plastic. Phosphorus additions of 7 weight percent seem to be optimum in order to produce the above desirable film characteristics. 

PECVD of silicon nitride has been of commercial importance since 1976.1 The original motivation was to find a final passivation layer for an integrated circuit that would replace the doped silicon dioxide films then in use. The latter were not reliable enough to permit packaging of integrated circuits in plastic. Silicon nitride was recognized as a better final passivation film, but the only available technique for its deposition was the high-temperature thermal process. Since it had to cover an aluminum final metallization layer that would melt at 600°C, this clearly could not work. The solution was to use PECVD at 350° to 400°C. 

Earlier, we reviewed silicon dioxide (thermal) films deposited with added phosphorus to serve as a getter for mobile ion impurities, as a final passivation film. Plasma-enhanced silicon nitride can also be doped with phosphorus.6 Some of the film characteristics have been reviewed, and it was found that the films with 2 to 3% P had the best electrical quality. No measurements of stress or H2 content were reported, so it is not clear that these would be use-able films. 

Since one major motivation behind the use of plasma BPSG was to provide an improved passivation barrier, the better crack resistance is an advantage, but the greater sodium penetration is a negative. Therefore, it is not clear if it would be advantageous to make this replacement11 for a final passivation film. Its use as an inter-metallic dielectric may be more useful. 

Attention was then turned to a higher temperature, low-pressure process. Although such films would not be useful for final passivation, they could provide valuable interlayer dielectrics. At first. SiH4 with N20 was tried, where the reaction would proceed as follows ... 

Finally, we will consider PECVD silicon oxynitrides, and their unique characteristics. When oxygen is added to a PECVD nitride film, there are indications that it may improve its crack resistance as a final passivation layer.13 Also, there may be advantages in terms of its electrical characteristics as an interlayer dielectric. Therefore, the nature of films grown when N20 is added to a SiH4, NH3 and He gas mixture in a high frequency (13.56 MHz), cold-wall, parallel-plate reactor have been studied. 

In class, we discussed metastable pitting, and you may have seen some indications of it in some of the previous tests. The easiest way to see metastable pits is to perform potentiostatic holds at potentials just below the pitting potential. In order to reduce our background current, as well as provide a baseline, we will first grow the passive film at a low potential before moving it to a potential where metastable pitting can be observed. Finally, we will polarize just above the pitting potential to see the induction time and the similarity between metastable and stable pits. 

In contrast to the Ce(N03)3/HN03 slurry, the HjOj slurry does not passivate the tungsten surface. Passivation in the H2O2 slurry is not expected because the slurry pH of 5.1 is above the pH where tungsten passivation occurs. The lack of a passivation film is suggested by the relatively high etch rates in the H2O2 slurry. In addition, the potentiodynamic scans in the static solution do not indicate passivation. Finally, the fact that abrasion does not... 

Collision ionization model The common collision ionization model of the electric breakdown as summarized by O Dwyer [156] requires a sufficiently high probabihty for the ionization process. Later this model was extended by Ikonopisov [150]. If the ionization cross section, film thickness and energy dissipation are large enough, an ionic avalanching will finally destroy the passive film. 

---