Figure 3. Sketch of the ultrasonic bonding cycle. The first bond is formed, as shown in (a), by the combination of a tool force down on the wire and an ultrasonic motion parallel to the wire. A loop is then made, as shown in (b) and the second bond is made in the same manner as the first. The wire is then cut off by a pull of the wire by the wire clamp so that the machine is ready to repeat the process. a b wire. The appearance of the heel of the bond depends on the amount of deformation. For example, Figure 4 shows bonds typical of those examined from one manufacturer compared with those of another manufacturer shown in Figures 5a and 5b. The fact that there is a great difference in bond deformation results from the different combinations of force, ultrasonic power and time that are used in bonding on these production lines. These adjustments of the bonding machine (the bonding schedule) involve a compromise. Setting combinations that yield a small deformation result in a small crack at the heel leaving the wire itself as strong as possible. However, if the deformation is very small, the bond is frequently weak due to lack of welding at the interface. This is called underbonding. In the extreme, the wire does not stick at all; it lifts off. More welding at the interface can be attained by increasing one of the settings in the bonding schedule. This will also result in increased deformation so that the heel will be thinner and weaker. In extreme cases the wire will be nearly severed at the heel. This is called overbonding. The optimum bonding schedule would be a compromise between underbonding and overbonding. Typical examples of the variation in heel cracks with different deformations are shown in Figure 6 which shows two first bonds from devices from two different production lines. The bond in Figure 6a is less deformed than the one in Figure 6b so that it has a considerably smaller crack in the heel. It should be noted that the bonding machine which made the bond in Figure 6a had a malfunction in its cycle (the tool was allowed to bounce on initial impact) which caused the indentation in the wire above the heel. The indentation in the bond surface below the crack is probably due to build-up of aluminum in a small region of the bonding surface of the tool. There can also exist a variation in the amount of deformation from bond to bond on a single device. The amount of this variation is usually (but not always) less than the variation described as existing from production line to production line. Two examples are shown in Figure 7. In Figure 7a, which shows eight of the bonds on a device, there is seen a large difference in deformation particularly visible in the two bonds marked A and B. These variations are easily observable with ordinary optical microscopes. Figures 7b and 7c show two consecutively made first bonds on another device. There is a significant difference in deformation between the two bonds. These bond-to-bond deformation differences are characteristic of unwanted motion between the tool and the work stage during bonding. This motion can be caused by many different factors such as movement caused by the operator, building vibrations, or just the normal motion or vibration of the bonder itself as it goes through the bonding cycle. This kind of motion is significant because it can result in the type of occasional bond lift-off which has not otherwise been explained. Relative motions of 0.00025 in. (about 6 μm) or smaller between the tool and the work stage are large enough to be significant. In addition to deformation differences there are other aspects of the bond appearance which give indication of unwanted motion during bonding. Figure 8 shows a bond which suffered from gross motion of the tool as 1 |