Page 130 - Wire Bonding in Microelectronics
P. 130

108    Cha pte r  F o u r



                                                       Blade
                       Aluminum
                        bond pad
                                                             Shear
                                        Gold ball bond       force



                              Die
                                         Au-Al
                                       intermetallic


              FIGURE 4-19  Schematic of a ball bond with isolated intermetallic growths.
              These growths can offer considerable resistance to a ball-shear probe, but can
              often be “pried” up with a scalpel or will lift with low force in a pull test [4-32].






              A fine scalpel blade is used to pry or “flip” the bond up, leaving the
              intermetallic spikes on the pad and on the lifted ball for examina-
              tion. This was called the “pluck test.” Weak, “as-made,” bonds
              could more easily have been revealed by a shear test during pro-
              duction (before any spikes were thermally generated), and the
              problems of poor bonding machine setup or contamination could
              have been solved at that time. But when it does occur, failure anal-
              ysis may be required. Thermal stress experiments on intentionally
              weak bonds (shear force <50% of optimum) made on clean pads
              by Harman (unpublished) indicate that, when the shear force has
              decreased to about half its original value (<25% of normal), the
              described mechanism becomes significant. In some cases, bonds
              with shear forces of 10 to 15 gf will lift in a pull test at 3 to 5 gf.
              This happened frequently enough to require application of the
              nondestructive pull test at various stages of the ball-shear, ther-
              mal-stress experiments to remove bonds with this failure mecha-
              nism. It should be noted that strongly welded Au ball bonds that
              are thermally stressed result in relatively uniform intermetallic
              formation and have not been observed to fail by this mechanism.
              An example of this phenomenon would be similar to that shown
              in Fig. 5A-1 in Chap. 5, App. 5A.
                 Such failure analysis methods have not been applied to fine pitch
              balls, because of the difficulty of inserting a probe under them. Almost
              as much information can be obtained from etching the Al pad out
              from the ball with 20% potassium hydroxide and examining the
              “turned over” ball. However since the pull test is often applied to
              these bonds, one occasionally sees spikes remaining in the pad after
              thermal stress test lift-offs.
   125   126   127   128   129   130   131   132   133   134   135