Cause(s) is/are the stimuli that make(s) something happen.
cause and effect chain
The cause and effect chain begins with the root cause and continues through to the appearance of the failure mode.
A corrective action is the change made to eliminate the cause of failure.
From this definition, it is clear that the cause of failure must be known before an effective corrective action can be devised. Lacking accurate knowledge of cause, any action taken risks missing the point, wasting time and money, and establishing false practices, which often confound the problem, rather than correcting it.
failure / defect mechanism
A failure mechanism is the physical or chemical process that results in failure. A defect mechanism is the physical or chemical process that prevents an IC from working in the first place. Corrosion and ionic charge phenomena are common causes of failure mechanisms. Defect mechanisms include those forms of corrosion that occur on the fabrication line and trapped ionic charge or oxide damage associated with plasma processes. Defect mechanisms also include such things as etch time too short to clear a contact hole and re-deposition of insulating material on the bottom of a contact hole.
The failure mode for an individual device is the symptom or test result which causes the device to be labeled a failure.
The failure mode for a wafer or lot of wafers can be expressed as the predominant shared failure mode(s) for structures in the lot. The fact that a lot of wafers failed to meet a specified yield minimum is not a failure mode since yield is not measured on any structure.
The root cause is the event or combination of events that initiate a failure.
Although root cause is the ultimate goal of failure analysis, the root cause may not be obvious even though the defect mechanism is known. For example, analysis may show the failure resulted from distorted deposition of layers around a foreign particle. The particle is the cause. However, when the fabrication process is fully understood, the particle and many like it may occur normally in the process. The root cause might be the cleaning process that should have removed the particle. Even at this level of understanding, possibilities are many. The clean might be inherently inadequate, the chemicals may be overused, the temperature may be too low, or the content of the particles may have changed.
Great progress has been made in increasing wafer sort yields by eliminating known causes of failure. Particles, critical dimensions, alignment, doping levels are all statistically related to yield. Defect reduction and statistical process control are key ingredients to yield improvement. Occasionally wafer yield is low despite control of known process and defect parameters. Statistical techniques can not determine the cause of failure.
Wafer failure analysis, on the other hand, begins with symptoms of a failure occurrence and works back to identification of the specific cause of failure.