Page 138 - Wire Bonding in Microelectronics
P. 138

116    Cha pte r  F o u r


                 Thus, adequate standards are  available to implement the ball-
              bond shear test. The recommended values (below) do need to be
              eventually modified in the future. The top curve in Fig. 4-23 [5.5 g/
                 2
              mil (84 MPa)] should be chosen as the minimum individual shear
              reading for Au ball bonds on Al pads, when long-term reliability
              (1000 h at 175°C or equivalent) is required [4-48]. This is typically a
              required test for some important industries.


         4.4  Evaluating Both the Ball and the Wedge Bond
               on a Single Wire
              The wedge (or crescent) bond of a ball-wedge bonded wire, as with a
              wedge-wedge bonded wire, is best evaluated with a pull test. One
              study found that there was only negligible degradation of the ball-
              shear force after bond pulling and concluded that the pull test can
              therefore be done first [4-39]. Similar statements are made in ASTM F
              1269-06. Thus, the ball can be sheared after the loop is pull tested,
              allowing data to be obtained from both bonds of a single wire and
              minimizing the required sample size.



         4.5  Thermal Stress Test for Au-Al Wire Bond Reliability
              Gold bonds on Al pads (or the reverse) have long been observed to
              fail beyond some level of thermal stress (see Chap. 5, Intermetallics).
              However, Horsting [4-49] (see Chap. 6 for Al bonds on Au) found
              that, if the bond is well made and there are no impurities present in
              the bond interface, the bond will remain strong even after long times
              at high temperatures. If impurities are in the interface or the bond is
              poorly welded, then the bond-strength may degrade rapidly during
              such stress. To reveal potential problems in a new gold-plated pack-
              age lot, Horsting applied a stress test that consisted of a 390°C bake
              for 1 h, followed by a pull test. If the bonds lifted (interface separa-
              tion) in a pull test, the entire package lot was rejected. Ebel [4-50]
              introduced an entire bake schedule as a screening procedure to
              reveal similar potential bond failures for hybrids. Later, MIL-STD-
              883, Method 5008 for hybrids, specified a similar, though less severe,
              test of 30°C for 1 h and a pull-test value specified as greater than or
              equal to 1.5 gf (14.7 mN)—post seal. Currently (2008), this is in MIL-
              PRF-38534. The time, temperature, and other conditions of these
              various stress tests are given in Table 4-2. It should be emphasized
              that this stress test is only useful for bimetallic bonds that diffuse
              and react readily (e.g., Au-Au and noble metal bonds improve with
              temperature, and Al-Al bonds stay about the same—see Chap. 5,
              Sec. 5.3.7).
   133   134   135   136   137   138   139   140   141   142   143