Understanding the Complexities of Solder Ball Pull Testing on BGAs
03/01/2005
BALL TESTING ALTERNATIVES
One of today’s needs is obtaining reliable, meaningful data about the adhesion of solder balls to the substrate of a ball grid array (BGA) package. Should we approach it as we do wire bonds, and perform a combination of shear and pull testing? Or should we perform shear testing alone? The answer to these questions lies in understanding why adhesion testing of solder balls on BGAs is more difficult than testing gold wire bonds.
Initial reasoning suggests that we use shear testing, which is easy to set up and quick to perform. Shear testing is a straightforward approach that applies a load to a ball with a flat tool until it breaks away at the bond interface. The first concern when shearing BGA solder balls, however, is that errors can result if the passivation layer around the solder ball joint is too high. A high passivation layer can prevent the shear tool from applying the shear force in the plane of the bond intermetallics. You may have tried this method and observed a nice, shiny, sheared surface that looks like a valid test. However, it is possible that what you observed is a cross section of the solder ball, and the only thing that you achieved is the evaluation of the shear strength of the solder itself, not the targeted bond interface. While obvious, the caveat is that in any bond testing scenario, it is fundamentally important that you are testing for the right failure mode.
The seemingly obvious solution is to use ball pulling to determine the strength of the bond. With wire bonds, we are presented with a convenient loop under which to insert a hook, which is then pulled upward until the bond fails. But, can we just grab a solder ball and pull it until the bond breaks at its interface? Unfortunately, solder balls present a number of problems to this approach. Foremost, traditional lead-tin solder is relatively soft, and after reflow presents less than a full hemisphere-an inconsistent shape to match the curved shape of the test tool gripper. In addition, the curved shape of the gripper, which is matched to a given ball size, leaves little room for alignment error.
Ball Testing Alternatives
There are two ball testing methods that currently rely on manual implementation: the “hot ball test” and the “cold ball test.”
The hot ball test inserts a hot pin into the solder ball being tested. The pin partially reflows the solder, and when cooled solders the pin into the ball. After cooling, the well-attached pin and ball are pulled until the bond breaks.
The cold ball test uses a mechanical gripper to connect the ball to the pulling load cell (Figure 1).
 Figure 1. Cold ball pull test tooling, consisting of two grippers, is clamped onto the solder ball and then pulled upward. |
Reservations with hot ball testing involve how much the pin-insertion reflow modifies the metallurgy of the ball interface, which can lead to poor repeatability. Finally, because the pin and ball need to be heated and cooled, this test is quite slow. Yet, despite the disadvantages of hot ball testing, this method has been adopted (along with cold ball pull testing) as the Japan Electronics and Information Technology Industries Association (JEITA) standard, and is widely used in Japan and by some contract assemblers.
The cold ball test is faster, although it requires a significant degree of sophistication in the gripping process to achieve repeatable test results. When grabbing the ball, the gripper does deform it slightly. Since the point of maximum material flow (or bond deformation) is close to the plane of the intermetallics, some disruption to this region is unavoidable. If the ball grippers are placed too high on the ball or poorly aligned laterally, the grippers can distort the ball to the point where it looks like a Hershey’s Kiss, in which case the strength of the bond is not being evaluated. The greater the distortion to the ball, the more questionable the test results become. It is important to inspect the ball site after the test to verify that a failure at the intermetallics occurred.
Cold Ball Test Alignment
For cold ball pull testing, tooling alignment and adjustment can be a difficult and time-consuming operation. The current manual alignment techniques are ripe for a testing system that provides a higher degree of precision and methods that take guesswork and individual skills out of the procedure.
Proper testing requires accurate alignment of the gripping tools during setup (aligning the two gripper halves together) and before performing each test aligning the grippers to the solder ball. Misalignments in either case, as small as a few mils, can result in invalid test data. The key objective while gripping a sample ball during a pull test is to minimize any structural disturbances at the intermetallic region. This can be better accomplished via simultaneous self-centering and self-alignment with symmetrical clamping.
Once the test has been properly set up, the next concern is controlling the displacement of the grippers (the distance each finger travels) must involve a mechanical adjustment that is easily accessible to the user and offers the precision required for minute adjustments. Too little force causes the grippers to slip on the sample. Too much force results in breaking the sample or reducing the cross-sectional area to the point of premature failure at the site of tool contact.
 Figure 2. Two grippers clamping a solder ball with unequal forces (F1>F2). The bond experiences Fnet in the horizontal plane before the pull test is initiated. |
If the clamping of the grippers is not symmetrical, the forces on the ball will be out of balance (Figure 2). Simple vector addition shows a net force on the bond in the horizontal plane. This means that the bond under test is experiencing a shear force, which means that the subsequent pull test results will not be valid.
Conclusion
Pull testing of solder balls should be done with minimum disruption to the bond intermetallics before the test takes place. It is difficult to achieve this, whether using the hot ball pull method or the cold ball pull method. For the cold ball pull test to be attractive to manufacturers, it must be easy to set up and align. The risk of damaging the product while aligning the tool must be removed, and the instances of undesirable failure modes must be reduced.
The uncertainties in cold ball pull testing can be reduced by using thoughtful clamping and alignment methods such as self-aligning, symmetrical gripper mechanisms; quick-change tooling; and easy-to-adjust displacement control. If these are achieved, the cold ball pull test will play a more significant role in process monitoring of BGA solder joints, instead of being used in rare instances as a diagnostic tool.
LISA GERBRACHT, head of applications engineering, MALCOM COX, president and CEO, and SCOTT WHITING, mechanical engineer, may be contacted at Royce Instruments Inc., 500 Gateway Drive, Napa, CA 94558; (707) 255-9078; e-mail: lgerbracht@royceinstruments. com; [email protected]; swhiting@ royceinstruments.com.