The WaferWoRx analysis software uses data collected from inspected probe marks on wafers (PMI – probe mark inspection) to identify signatures not readily available from a single set of scrub data.  This PMI data can be provided from either a WaferWoRx (WWX) 300 or from an inspection tool capable of measuring PMI, i.e NSX or AXi System.

With the WWX on NSX, the PMI data is provided by the NSX or AXi tool. During the PMI inspection several parameters are measured including – 

• scrub area
• scrub placement-within the bond pad i.e. edge distance
• center of scrub
• scrub angle or rotation
• length and width of scrub
• percent of bond pad damaged 

From this data, the WWX software is able to break the error components into a number of parameters which are associated with various components of the probing process i.e. prober or probe card. Table 1 lists the associated parameters.

Table 1

By understanding the various error components and where they may originate the WWX software can help the end user rapidly troubleshoot and correct process limiting sort issues. WWX provides several different reports or views for the error components (Table 1). The data can be viewed in a Pareto chart, in a color scaled wafer map, or a vector view wafer map. Examples of each are show in Figure 1.

Figure 1

An application study - applying the software

A recent applications study done on an NSX 320 provided the following data sets. Looking at Y scrub size in the wafer map view (Figure 2), we can see an increase in error as you move from left to right. Drilling down further and looking at a single die, we can see greater error between the top and bottom bond pad locations. This type of error can be caused by a fixture translation, in this case in the Y direction.

Figure 2

Further analysis also revealed additional errors associated with prober stage performance as seen in Figure 3. Looking at the overall scrub area, again in wafer view, we see the area increasing from left to right across the wafer. This can occur when the chuck is not planar to the probe card, this causes the amount of overtravel applied to the card to increase as you move across the wafer. The increased overtravel causes a longer scrub on the bond pad.

Figure 3

Multi-pass test

While it is true that having a single scrub mark per pad will provide the cleanest data and most readily help to establish root cause, it is not true that only single pass scrubs can be analyzed. In the PMI applications, most often, multi-pass scrubs are combined and measured as a single scrub. This will lead to larger than expect scrub sizes and greater errors, but the errors may still reveal signatures. To help understand this concept better let us look at a recent applications study.

This study was done on wafers that had been probed at cold and hot temperatures. At cold temperatures a 64 site square probe card was used and at hot temperatures a 256 sight 4 touch rainbow card was used.

Looking at Figure 4 & Figure 5, we can clearly see the 64 site card pattern across the wafer. This pattern is evident because the majority of the error occurred during this probe application. This shows that in the overall process, the end user will gain the largest improvement by addressing the cold probe process first. They could then pull wafer from cold sort after a single insertion for more detailed analysis and assistance in finding the root cause.

Figure 4 and 5