Richard S. Harris

Optimization of memory redundancy laser link processing

Memory repair through the use of laser processing of redundant elements is an industry standard procedure for memory chip manufacturing. But, shrinking memory feature sizes and the industrys tendency to use metals as link materials rather than polysilicon imposes new challenges for laser processing. So far, the majority of the research on memory link laser processing has concentrated on: The vertical structure of a link (such as the multiple layers of passivation, link, field oxidation and silicon substrate); the laser beam absorption; and, the different temperature distribution within the structure as the result of laser beam heating. Until now, the emphasis in laser link processing optimization has been aimed at creating uniform temperature distribution while severing the link before exploding the passivation layer. Our study has shown that the link width plays an important roll in the processing as well. Analysis of the mechanical stress beneath the passivation layer using finite element modeling has been carried out. Different link width and passivation layer thicknesses vary the stress dramatically. The results of this simulation will be presented and their implication on link processing optimization will be discussed. To optimize the laser processing further, we have proven that absorption contrast of laser energy between the link material and the silicon substrate beneath the link must be maximized. Based upon the fact that while the absorption of most metal materials in the 1.3- to 2-micron range remains the same as that at 1 micron, it drops dramatically for silicon. By using laser wavelengths within the 1.3- to 2-micron range, a much wider laser processing window can be realized. Comparison analysis of link processing by different laser wavelengths will be discussed.