Damage Analysis

Authentic information about the origin and the development of fracture is recorded in the appearance of the fracture. The fractographic and microscopic investigation of original wafers broken during manu­facturing, therefore, can reveal the history of the damage. Examples are given in Figures 4 to 9.

Figure 4 shows micrographs with typical markings on the fracture surface in the vicinity of the crack origin.

The crack started at a micro-crack located in the surface of the wafer (left picture) and in the edge of the wafer (right picture), respectively. The dashed lines represent crack propagation fronts.

comprises a heat treatment which obviously contributes to a mitigation of micro-damages and thus causes a further increase in strength. Step 4 (a coating procedure) does not substantially influence the damage status. Step 5 (black bar in the figure) however leads to an essential decrease in strength revealing a severe generation and growth of micro­damages. This step represents the printing station. The fast firing process 6 can only partially reverse the loss of strength. Inspections of the printing procedures showed that an inappropriate temperature control effectuated high thermally induced stresses in the wafers. After an improved temperature adjustment the loss rate diminished noticeably.

In the sample shown in Figure 5 the crack initiated in the edge of the wafer at a micro­crack which is due to imperfect laser-cutting. The micro-crack is widened by etching processes.

О) 1 ^

1,0

0,8

0,6

о ’

0,0

Figure 3: Strength values of wafers subjected to different etching procedures

A B C D

Etching Procedure

Figure 4: Fracture surfaces at crack origin (arrow). The crack was initiated at a micro-damage in the surface of the wafer (left) and in the edge of the wafer (right)

In Figure 6 the crack was released at a boundary within the crystalline structure of the silicon. This boundary presumably originates from a former internal fissure in the ingot which was later on filled by still liquid silicon during the solidification process.

Figure 7 shows the surface of a wafer along the edge. The damages at the edge are due to a knocking of the wafers against guide pins during fluttering in the air flow of a fan. After constructional changes at the guiding device these damages did not occur any more.

Figure 8 exhibits the fracture surface of a wafer which was broken because of a small crack within the wafer surface. The problem was to detect at which step in the manufacturing chain this initial crack was originated. Because the Figure 5: Crackinitiation at a micro-crack area of the initial crack (marked by the dotted in the edge of a laser-cut wafer
line) shows typical patterns associated to etch treatment it was evident that the initial crack was already existent prior to the etching process. After this limitation the cause for the cracks, a handling problem, could be identified and eliminated.

Figure 6: Crack release at a crystalline boundary. The surface area marked in the left micrograph by a rectangle is shown to the right in larger magnification. (View onto the wafer surface)

Figure 7: Edge damaging due to inadequate handling

Figure 8: The area surrounded by the dotted line exhibits structures which are associated to etch treatment. This evidences that the crack was already existent previous to the etching process

In Figure 9 a wafer fragment is shown from which a complex evolution of a breakage could be revealed. The primary crack initiated at the position which is indicated by the black arrow (right in the picture) and propagated some millimetres in both directions and then stopped. The fracture surface on the right hand side (along the grey dotted line) exhibits traces of the blue AR-coating, thus indicating that this crack segment originated before the

AR-coating process. On nearly the entire fracture surface (along the black dotted line) traces of the backside-metallization can be found. This part of the crack, therefore, must be developed after AR-coating and prior to the backside-metallization process. The final break through (on the very left on Figure 9) occurred due to the thermo-mechanical loading during the fast firing process, because the breakage was discovered after the fast firing furnace.

As these examples show, the real sources of loss cannot be discovered generally by considering only those processes at which major breakage rates appear. The breakage of wafers during manufacturing can have a complex history: In any of the process steps a micro-damage is generated, in (one or more) other processes the damage is extended and only in a later process applying higher stress the wafer breaks completely and evidently.