semi合集-English.pdf - 第4947页

SEMI M11-0704 © SEMI 1988, 2004 4 Figure 2 This is a m ound that does not h ave the sharp edges of an Epitaxial St acking Fault , but is n ot as high as a Bump. They are typically s ized much smaller in a SSIS th an thei…

SEMI M11-0704 © SEMI 1988, 2004 3

• Device characteristics that may be affected —

leakage and gate oxide integrity, from

crystallographic changes.

• Detection characteristics used for characterization

— side length and depth, orientation parallel to a

<110> direction.

• Discrimination characteristics used for

characterization — shape, length ~ epi layer

thickness: 1–10 µm. Stacking faults may cluster

(see Figure 7) and scatter more than a single

stacking fault.

• Specification characteristics used for wafer

qualification — Number per wafer. Since certain

types of epi stacking faults have no observed

impact on device performance while others are

killer defects no number can be assigned without a

clear identification of the type of epi stacking fault

that is involved.

Figure 1

A Mound. Magnification 1000 times using Nomarksi Interference Microscopy. Approximate SSIS scattering

event size 0.22 µm.

23 µm

Figure 2

This is a mound that does not have the sharp edges of an Epitaxial Stacking Fault, but is not as high as a

Bump. They are typically sized much smaller in a SSIS than their actual size, but some have facets that

scatter enough light to be more easily detected. Some people refer to this feature as a hillock which is a type

of mound. The shape can be either circular or square. Magnification 500 times using Nomarski Interference

Microscopy. Approximate SSIS event size: 0.15 µm

Figure 3

The Epi Stacking Fault (ESF) defect is a grown-in defect with 1 to 4 sides of a square visible. Magnification

1500 times using Nomarski Interference Microscopy. Approximate SSIS event size: 0.165 µm