BTY

The New TI Method for Comprehensive Analysis of Wafer Sort BIN Variance -YBIN

Nick Atchison
Texas Instruments
Santa Cruz, California, USA / ??????
Texas Instruments
Tustin, California, USA / Ron Ross
Texas Instruments
Santa Cruz, California, USA

Abstract

The continual effort to shrink design rules of existing circuit designs and the continual introduction of larger format wafers creates a distinct need for automated bin yield analysis. In the next 10 years the number of process steps will double making the automated analysis of device to process equiment interactions by geometric wafer region a requirement.

Elevation of the bin 1 yield above 75% bin requires the ability to differentiate between BIN yield means of less than 1% with 95% confidence. For a sample of 2,250 wafers, normal process analysis techniques utilizing wafer averages can typically detect a minimum of 3% bin 1 yield differences with a 95% confidence level. The automated analysis techniques covered in this paper extend the minimum 95% confidencece level to be extended to defferences between means of less than .3%. The allocation of BIN losses to particualr process euiiment is even more difficult because the magnetude o ht bin loss is much less than the magnetude of the BIN 1 yield. When sort volumes of geven BINs are low, the one order of magnetude of increased analytical resolution provided by this method is vital if bin 1 yield is it be ramped to above 90%.

This paper is one of several follow on papers to the series of papers “A Comprehensive Sequential Bin yield Analysis Methodology: By Ross and Atchison in the Texas Instruments Technical Journal Vol.15 No. 4. This paper is an extension of the concepts presented in the paper “Numerical Analysis of Equipment Yield Variance YE“ in the same issue of the TI Journal.

In this paper the analysis of the bin yields of geometrically defined areas of wafers processed in each equipment unit at each process stage is described. This analysis provides the basis for the allocation of BIN losses to particular equipment.process/recibes. It not only isolates the source of bin yield problems, it is part of an overall IYM methodology that attempts to rationalize all bin yield losses associated with a wafer fab

This analysis is part of a group of programs and methods called the TI SCZ Sequential Bin yield Enhancement Method. The analysis described in this paper can be accomplished with commercial software from companies such as “Heuristic Physics Laboratories (HPL)”.