Issues with wire bonds and wire bonding can be easily solved with good observation and attention to processing details.

The EMPF recently performed a failure analysis on a wire bonding issue. The manufacturer’s assembly and final packaging is outsourced to two locations. At the first location the product is assembled onto a flex based System in Package (SiP).  Fine gage aluminum wires are attached from the device I/Os (input/output) to the flex board with wedge bonds.  The wirebonded assemblies are then encapsulated with a silicone gel, cured at room temperature for 24 hours, and then shipped half way around the world for further packaging.  Once there, the assemblies are electrically tested. The customer discovered that a high percentage of assemblies were failing for broken wire bonds.  The assembly failure rate was above 50% for a period of time and the failures occurred within the same local area.  Some batches showed significant yield issues while others did not.

First, the EMPF looked into any recent machine adjustments.  The contract manufacturer (CM) determined that while there were no apparent changes to the wirebond parameters, the process yield returned to the 96% pass rate that was traditionally encountered for this assembly. However, tooling changes may have been implemented, or a wire spool replaced, eliminating a possible source of the failure.  It was also determined that the CM previously performed preventive maintenance on the wirebonder on a quarterly basis, but suspended the activity due to lower business demands.  The general maintenance requirements for the bonder include replacement of the wire wedge every 400,000 bonds.  This is recorded in a log book.  After replacing the wire spool, the machine should be requalified at the beginning and end of shifts.  To further quarantine product when bonding issues exist, an even shorter interval should be used between verification of bond strength. 

The customer uses an endcap material to protect the wirebonds.  This material is typically a silicone and it protects the bonds from exposure to the environment and minor mechanical damage.  During the curing process some shrinkage occurs.  Often, tunnel voids form under long rows of wirebonds.  If the wire loop is too long, the flow of the coating material can cause the wires to flex and bend, a phenomenon known as wire sweep.  This can cause shorts if they are pushed together or fractures if the wirebonds are weak.  In order to minimize the number of temperature excursions during the assembly process, a room temperature curing, solvent based system was selected.  However, this system also exhibits a large amount of shrinkage. 

Damaged Bonds

The failure analysis begins by chemically removing the protective coating material in order to examine the bonds.  This process left behind a residue containing sulfur, which was identified by analysis of energy dispersive spectroscopy spectra.  Typical wire bonds should have low deformation.  Examination of the sample wirebonds by scanning electron microscopy showed that the bonds were overly deformed and had some tool impression (Figure 3-1).  The bonds have large smooth ears (the edges of the deformed area), and the bond area was completely flattened.  This type of defect can indicate excessive bond force, mechanical binding of the bond head during overtravel, loose or sloppy Z-direction drive assembly, low search heights, or high elongation wire.  Figure 3-1 shows some cutting marks at the foot of the bond as expected.  There is some impression of the tool that is evident, indicating excessive pressure being applied.  There is also micro-cracking at the top of the heel, indicating damage from excessive deformation. 

Pull testing of the same wirebonds (Figure 3-2) showed values in the range of 8-12g.  A majority of the failures occurred at the first bond site while fewer failed at the second bond site.  The expected range for aluminum wire products is 19-25g, but this depends on both the elongation percentage and temper conditions.  In this case, the location of the failure was at the heel of the bond and not at the pull testing location, as expected.  This indicates a poorly formed bond, due to excessive wire deformation.  Of the three main well known variables that affect the wirebond strength (namely, power, force, and time), excessive force was apparent.

The weaker wirebonds were also stressed by the application and curing of the end cap material.  The presence of voids in the encapsulant may not be initially detrimental.  But the presence of the worm or tunnel void alters the amount of encapsulant and changes the stress field along the area where the tunnel voids end or coalesce to form a single void.  The exact mechanism is not entirely understood, but evidence suggests that enough stress occurred during shrinkage of the material to ultimately fracture the poorly formed wirebond.  The EMPF and customer are identifying materials that will be more suitable for this application. 

Examination of the wire bond tool on a regular basis, inspection and monitoring of the tool change interval and a decrease of this value to less than 400,000 bonds is recommended.

In conclusion, the CM had poor maintenance of the wire bonding process,  indicated by the presence of  foreign objects and the lack of preventive maintenance.  Foreign objects on the bonds and marks on the wires also suggested poor maintenance.  Poor machine setup and improper qualification led to a high failure rate.

Improper bonds are typified by any or all of the following observations:

     •    Excessive deformation noted by tool marks
     •    The presence of micro-cracks on the heel of the bond
     •    Lower pull testing values than those expected from wire
          property data sheets
     •    The pull testing fracture location not occurring in the
          central length of the wire (in this particular case the
          failure was at the foot). 

These four significant observations can be corrected by proper wirebond machine setup and qualification.