TY - JOUR
T1 - Influence of grinding process on semiconductor chip strength
AU - Wu, Enboa
AU - Shih, I. G.
AU - Chen, Y. N.
AU - Chen, S. C.
AU - Tsai, C. Z.
AU - Shao, C. A.
PY - 2002
Y1 - 2002
N2 - The objective of this paper is to study the strength distribution of semiconductor chips on a wafer, and the influence of the back-side grinding process on the chip strength. The three-point bending test, complying with the ASTM standard E855, was adopted to measure the chip strength. The first set of test vehicles is from three 8-inch wafers. One is of 28 mils thick without backside grinding, and the other two are backside ground to 18 mils and 11 mils thick. Then, four 6-inch wafers were used as the second set of test vehicles. The first two were 22 mils thick which were backside ground and the other two wafers were 27 mils in thickness without grinding. The third set of test vehicles was formed by three 8-inch wafers of identical thickness (11-mil) and size, but they were backside ground by different factories. It is found that, whereas the chip strength distributed randomly on a wafer which did not experience any backside grinding, any wafers that were subjected to backside grinding always resulted in weak regions. The averaged strength for the chips in the weak region was approximately 30% lower than the averaged strength calculated from the whole wafer, regardless of the chip dimension. This phenomenon is independent of wafer thickness to be ground, wafer diameter, backside grinding equipment and process, existence of metal and passivation layers, etc. Meanwhile, this weak region, in the shape of a pair of sectors, is axisymmetric to the wafer center.
AB - The objective of this paper is to study the strength distribution of semiconductor chips on a wafer, and the influence of the back-side grinding process on the chip strength. The three-point bending test, complying with the ASTM standard E855, was adopted to measure the chip strength. The first set of test vehicles is from three 8-inch wafers. One is of 28 mils thick without backside grinding, and the other two are backside ground to 18 mils and 11 mils thick. Then, four 6-inch wafers were used as the second set of test vehicles. The first two were 22 mils thick which were backside ground and the other two wafers were 27 mils in thickness without grinding. The third set of test vehicles was formed by three 8-inch wafers of identical thickness (11-mil) and size, but they were backside ground by different factories. It is found that, whereas the chip strength distributed randomly on a wafer which did not experience any backside grinding, any wafers that were subjected to backside grinding always resulted in weak regions. The averaged strength for the chips in the weak region was approximately 30% lower than the averaged strength calculated from the whole wafer, regardless of the chip dimension. This phenomenon is independent of wafer thickness to be ground, wafer diameter, backside grinding equipment and process, existence of metal and passivation layers, etc. Meanwhile, this weak region, in the shape of a pair of sectors, is axisymmetric to the wafer center.
UR - http://www.scopus.com/inward/record.url?scp=0036292862&partnerID=8YFLogxK
U2 - 10.1109/ECTC.2002.1008323
DO - 10.1109/ECTC.2002.1008323
M3 - Journal Article
AN - SCOPUS:0036292862
SN - 0569-5503
SP - 1617
EP - 1621
JO - Proceedings - Electronic Components and Technology Conference
JF - Proceedings - Electronic Components and Technology Conference
ER -