Wafer Breakage

A broken wafer is a loss and an opportunity. When a wafer breaks during fab processing, the first impulse is to clean up the mess and trash the broken pieces. But the fracture markings on the broken edges actually represent a detailed recording of the rapid sequence of events that took place during the breakage event. By examining the fracture markings with a microscope, it is often possible to discover how and why the wafer broke so that appropriate corrective and preventive actions can be taken.

The Fracture Analysis Procedure below outlines the analysis procedure and gives two references where instructive descriptions and sample photos and drawings can be found. The photos on this web page show samples of the most commonly encountered fracture markings, rib lines and tear marks. The photo captions tell how to interpret rib lines and tear marks and describe four cases where fracture analysis was useful in solving a wafer fab breakage problem.

A wafer begins to break apart when the silicon at some surface location is subjected to a sufficiently large tensile stress. The tensile stress can arise due to wafer bending, uneven cooling, penetration, or during an impact when the silicon is compressed in one direction but pulled apart in other directions. The stress required to break a wafer is greatly reduced if the wafer surface has a stress concentrator such as a preexisting crack.

The goal of fracture analysis is to discover how and why a wafer broke so that future occurrences can be prevented. If a wafer broke because of a weakness, such as a scratch that occurred at a prior processing step, then that weakness can be eliminated. If a wafer broke because it was bent or penetrated by a piece of wafer handling equipment, then that equipment can be adjusted. A broken wafer is not only a loss but an opportunity for improvement.

Fracture Analysis Procedure

  1. Collect samples without destroying the evidence.
  2. Use a scale drawing to record and report the results.
  3. Examine the fracture markings on the broken edges to determine the direction of travel for each crack that split the wafer apart.
  4. Trace the crack(s) back to the approximate origin of the breakage.
  5. Examine the fracture markings to determine the exact origin.
  6. Examine the silicon at and around the origin for evidence of why the wafer broke.

Edge chip. The curved ripples on the fracture surface are rib lines. Rib lines are produced when the leading edge of the craft that is splitting the silicon apart wanders upward and downward as it travels along. Rib lines show where the leading edge of the crack was at successive instants in time. These rib lines show that the crack that separated the flake of silicon from the wafer originated at "O" and spread to the left and toward the front surface of the wafer. This means that the force that created the chip was directed right-to-left toward the edge of the wafer.

Fracture markings on a broken edge. The curved rib lines show that the leading edge of the crack that split the silicon apart traveled from the back surface of the wafer upward. The more or less vertical lines are called tear marks. Tear markes are produced when the leading edge of the crack travels on more than one plane at the same time. Tear marks are generally parallel to the direction of crack travel and they radiate away from the origin of the crack. The fracture surface is rough along the back surface of the wafer near the origin "O." An examinatin of the back surface of the wafer at "O" showed that the back side of the wafer had been scratched during fab processing. Later, a bending stress put the back surface of the wafer into tension and this caused some of the scratch cracks to propagate, breaking the wafer apart.

Fracture markings on a broken edge. The tear marks and other fracture markings showed that the breakage originated at the front surface of the wafer and spread downward. No wafer-weakening defect was found at the origin. In this case, the wafer was being held against a vacuum chuck by the atmospheric pressure when a wafer lifter pin came up under the back side of the wafer. The lifter pin pushed up a small part of the wafer, bending it over the lifter pin and putting the front surface into tension. The wafer broke at the front surface where the tensile stress was greatest.

Wafer backside after backgrinding and caustic etching. Fracture analysis revealed that the origin of the breakage was at this row of caustic etched grooves. The backgrinding process created rows of extra deep cracks in the wafer backside. Caustic etching produced the grooves by etching away part of the crack damage. However, the remaining crack damage weakened the wafer and it broke apart during subsequent handling.

Descriptions and photos of fracture markings are given in the following references:

  1. Lawrence Dyer, Jeffrey Seaman, and Dennis Olmstead, "Hertzian Contact and Thermal Fracture of Silicon Wafers During Processing", in Semiconductor Silicon 1990, Proceedings Volume 90-7, edited by H. R. Huff, K. G. Barraclough, and Jun-ichi Chikawa, (The Electrochemical Society, Pennington, NJ, 1990), pp. 156-167.
  2. Lawrence D. Dyer, "Fracture Tracing in Semiconductor Wafers", in Semiconductor Processing, ASTM Special Technical Publication 850, edited by D. C. Gupta, (American Society for Testing and Materials, Philadelphia, PA, 1984), pp. 297-308.