| US 7,395,168 B2 | ||
| Method for evaluating semiconductor device error and system for supporting the same | ||
| Hidefumi Ibe, Yokohama (Japan); Yasuo Yahagi, Tokyo (Japan); and Hideaki Kameyama, Uenohara (Japan) | ||
| Assigned to Renesas Technology Corp., Tokyo (Japan) | ||
| Filed on Jan. 19, 2005, as Appl. No. 11/37,105. | ||
| Claims priority of application No. 2004-090352 (JP), filed on Mar. 25, 2004. | ||
| Prior Publication US 2005/0211890 A1, Sep. 29, 2005 | ||
| Int. Cl. G01V 5/10 (2006.01) | ||
| U.S. Cl. 702—115 | 2 Claims |

| 1. A semiconductor device evaluation support apparatus, comprising,
a spectrum storing means which stores multiple spectrum data of white neutrons having spectrum shapes different from one another,
a testing result storing means which stores multiple SEE (Single Event Effect) counts obtained by a white neutron method conducted
utilizing each of said multiple spectrum data, and
an SEE cross section calculating means which calculates SEE cross section, wherein,
said SEE cross section calculating means carries out,
a first processing including,
a processing for reading out from said spectrum storing means, predetermined spectrum data having a large flux contribution
in a lower energy portion, inputting the predetermined spectrum data into the Weibull Fit function representing SEE cross
section, thereby calculating a first relating value of the SEE counts, and simultaneously reading out from said spectrum storing
means arbitrary spectrum data other than said predetermined spectrum data, inputting said arbitrary spectrum data into said
Weibull Fit function, thereby calculating a second relating value of the SEE counts, and calculating a computed ratio as a
ratio between the first and the second relating values,
a processing for reading out from said testing result storing means the SEE counts as to said predetermined spectrum data
and the SEE counts as to said arbitrary spectrum data, and calculates an actually measured ratio, as a ratio therebetween,
and
a processing for calculating a difference between said computed ratio and said actually measured ratio, and
the first processing then repeating the above processing multiple number of times to obtain a sum of said differences,
a second processing for changing a parameter value of said Weibull Fit function and executing said first processing repeatedly,
so as to detect a parameter value to make said sum of said differences equal to or less than a predetermined value, and
a third step for generating the SEE cross section represented by the Weibull Fit function by use of said parameter value thus
detected.
|