BY ROBERT SYKES
As the age of miniaturization advances, the need to check the integrity of interconnects becomes ever more demanding. Traditionally, this integrity has been determined using standard techniques such as pull and shear testing; however, these techniques have been refined to allow for new technology such as bumped packages and devices and ultra-fine-pitch bonding.
Pull Testing Solder Balls
Four possible reasons for considering cold bump pull testing (CBP) in place of the more traditional shear test are:
- Shear testing is not appropriate for balls mounted in a solder resist or similar cavity. When the bond site is recessed below the substrate surface, it gains support from the sidewalls of the cavity. The support given by the sidewalls can be substantial, to such an extent that a bad bond could pass if tested by the shear method. In a CBP test, the cavity adds little or no support to the bond.
- During a shear test when test load is at maximum, the ball becomes deformed above the bond site. With correctly designed tooling, ball deformation above the bond is reduced significantly using a CBP test.
- In most cases, a bond is tested to obtain an indication of its electrical and metallurgical strength. Surface irregularities at the bond site often couple together even though there may be little or no chemical adhesion between them. In shear testing, the locking together of surfaces may produce higher test loads than in a CBP test where the surfaces are separated.
- CBP tests the bond with an application of a simple, tensile load as opposed to shear test that applies a complex load consisting of tensile, shear and compressive forces. This complex mixture of loads may present problems when analyzing test data and trying to interpret failure modes. With the exception of the slight shear force produced when reforming the ball, the CBP test produces simple tensile loads (Figure 1).
CBP Testing in Practice
The test force that is exerted on samples with a good pad design and with high bond strength is limited by solder strength. This type of sample typically exhibits a single failure mode when the solder fails. If this happens, no data on the pad or bond strength can be collected. Data collected from several test samples indicates that the limit imposed by solder strength is more common during shear tests than during CBP testing. CBP tests typically produce more and a greater variety of out-of-spec pad and bond failures. The tendency for the pad and bond to resist shear testing more than force applied in CBP tests is explained by the mechanical support the pad-to-bond site receives from the substrate.
In both the shear and CBP test, the test tool distorts the solder ball under test. However, the distortion in a CBP test is controlled and minimizes the effect on the portion of the ball used during the test. Distortion resulting from shear testing can result in the shear tool cutting into the ball during the test.
A simple force analysis of both the shear and CBP test suggests that the stress on the bond under shear is more complex than that of a bond being pulled. It is considered that the simpler loading of tensile force only could be an advantage when trying to analyze failure modes and the relative strengths of the solder ball bond, its mating pad and the substrate. CBP test result comparisons from samples with different ball geometries also is considered more meaningful because the simple tensile load per unit area will not be as affected by any changes in size.
 Figure 1. Shear test load vs. CBP test load |
null
CBP testing requires a gripping tool. This tool should be designed to hold the ball as firmly as possible without significantly changing the characteristics of the bond and can be connected to a conventional pull load cell.
The CBP gripping tool consists of a pair of precision tweezers, with a special cavity machined into each jaw of the tweezer (Figure 2). Each jaw contains one half of a cylindrical cavity that reshapes the ball into a “rivet head” or “mushroom” as they close around the ball. The jaw cavity is designed so the shaft of the rivet or stem of the mushroom is the same diameter as the bond interface. The underside of the rivet/mushroom head transmits the pull force to the bond face. The jaws normally are closed while in contact with the surface of the substrate to ensure the maximum amount of solder forms the head. This produces an optimal grip on the ball. Once the ball is gripped, it is reformed by the jaw cavities, creating a maximum grip and allowing the test force to reach its potential.
 Figure 2. CBP gripping tool |
null
The goal is to produce the maximum pull force possible for this type of test. As is the case with other tests, like ball shear or wire pull, the sample material and test tool design limit the maximum possible test load. In the case of wire pull, the ultimate tensile strength of the wire sets the limit of the maximum load. However, if the pull hook test fixture is poorly designed, it could affect the maximum load. A system capable of testing at maximum force provides more information on the strength of the bond(s).
After several pull test fixture design iterations, the cavity shape that reforms the ball into a rivet/mushroom head proved optimum—it gripped the ball and effectively produced a pull force equivalent to the tensile strength of solder with an area equal to that of the bond site (root diameter).
Shear Testing Advances
Historically, applying a shear force to the ball has been used to test the integrity of the bond between gold balls and their mating pads. A new method of shear testing and tooling has been developed that significantly improves the quality of data obtained for a range of samples tested.
Principally, the new tool design results in these advantages:
- An increase in the accuracy and quality of the bond strength test data.
- The cavity tool mimics the shape of the ball, allowing the load to be applied to a larger surface area.
- A reduction in the sensitivity to step-back height.
- A substantial increase in the test force applied to the bond.
- The tested ball can be cleared more easily, so bond failure can be viewed clearly for grading or analysis.
Using Cavity Shear Tools
The aim of bond test equipment is to produce the maximum test force possible. As previously mentioned, wire pull testing can return poor results if incorrect or badly designed tools or fixtures are used. In a similar manner, a standard shear tool (chisel tool) (Figure 3a) can deform the ball before a significant force is applied to the bond. In many cases, the ball fails before the bond interface, resulting in test results with incorrect data. By using a cavity shear tool (Figure 3b) with a curved cavity, the new shear tool reduces ball deformation and increases the maximum possible force applied to the bond.
When a chisel tool is used, ball deformation affects the failure mode—the amount of gold left on the pad. Also, the initial deformation and subsequent failure mode is affected significantly by step-back height. As the step-back height is increased, the amount of gold residue increases and test reliability decreases.
The cavity tool's curved shape supports the ball, resulting in less deformation. As a result, the test is more effective at testing bond failure and the negative effects of step-back are minimized. This is particularly evident when observing the percentage of gold left on the pad. Gold residue is far less and the test is more consistent with the cavity tool.
 Figure 3a. Schematic views of standard shear test. |
null
 Figure 3b. Schematic views of “cavity” shear test. |
null
For optimum test performance, maintaining a low step-back is important for both chisel and cavity tools. The step-back height from the bond results from the tool resting on the substrate and then lifting a controlled distance from the surface. In most tests, the tool rests on the passivation layer, resulting in a minimum step-back equal to the height of the passivation layer above the pad. For ultra-fine-pitch applications, this is typically 0.5 to 1.0 µm. At this height, the data from a chisel tool will be degraded, resulting in a high percentage of gold shearing. If a cavity tool is used, step-back can be as little as 0.5 µm and still produce little gold shearing.
 Figure 4a. Standard shear tool at 2 µm stepback |
null
 Figure 4b. Cavity shear tool at 2 µm stepback |
null
The performance of the new cavity shear tool was tested by a series of scenarios. In one example, it was used to test 35 µm diameter ¥ 50 µm pitch gold balls over a range of step-back heights. In all cases, the failure mode was either separation at the inter-metallic zone, a shear through the gold ball or a combination of both. For each test, the area of gold left on the substrate was estimated visually as a percentage of the total bond area.
When there is no gold left on the pad, the bond has failed. This is considered a “good” test, because the resulting test data relates solely to the strength of the bond (Figure 4a).
Conversely, if gold is left on the pad, the strength of the full bond was not tested and the data collected on the bond strength is corrupted. In cases when the bond site is totally covered with gold, the resulting test data relates only to the strength of the gold and not that of the bond (Figure 4b). Both examples are considered to be “bad” tests. As a load is applied to the bond through the gold ball, the strength of the ball ultimately will limit the maximum bond strength that can be measured successfully. A new cavity tool design provides improved bond strength data over a wider range of step-back heights.
The Future of Bond Testing
So what does the future hold for pull and shear testing? As functionality and density increase, bond size and pitch will decrease. Associated with this decease are four technical obstacles to overcome:
The manufacture of CBP tools with smaller jaw cavities. Below 100 µm diameter, the manufacture of jaw cavities with sufficient precision is not possible using current manufacturing methods. Methods must be refined or alternatives found.
Jaw actuator design. The jaws must be able to move within their limits of motion (closed) with improved dimensional and force control. This requires a new generation of jaw actuators and control.
Cavity shear tools. Cavity tools can be made for ball diameters to 30 µm. As pitch decreases, a process to manufacture smaller cavities will be required.
Viewing optics for pull and shear applications. Magnification must be increased to 180¥ of above, necessitating improvement of the microscope, its mount and illumination.
With the increasing use of devices with recessed balls, there is a need to depart from traditional shear testing and embrace the newer and more reliable pull technology—the CBP test. With this new test technique, the performance of the metallurgic bond is tested rather than just the solder's shear strength.
For several years, the standard in ball bond testing has been a shear test using chisel tools. So why is the standard test unsuitable for today's testing requirements? It is suspected that in the early days, when the standards were developed, the chisel was an effective tool because bond strengths were lower and ball deformation was not significant. The current technology of ball attachment allows component manufacturers to significantly increase bond strength, resulting in the need for more effective test apparatus. The chisel tool remains a useful gauge today, but if maximum precision is required, the use of a cavity tool for both shear and CBP is far superior. In ultra-fine-pitch wire bonding, the effect of the passivation layer on minimum step back height can present problems with a standard chisel tool.
Robert Sykes, mechanical design manager, may be contacted at Dage Precision Industries Ltd., Rabans Lane Aylesbury, Buckinghamshire HP I9 8RG, England; +44 (0) 1296 317800; Fax: +44 (0) 1296 435408; E-mail: [email protected].