|
As part of the manufacturing process of electronic assemblies, the wire bonding technology represents an established friction welding process to generate an electrical contact between a chip and the respective substrate surface. It is used as a thin wire method with gold, for example in the COB (Chip On Board) technology, or as a thick wire technique with aluminium wire in the production of high power modules.Within the low-cost range, copper wires can also be used. Wire bonding issues The different methods have a common weakness: impurities on the surface, such as flux residues from soldering or oxide layers, impair the bonding quality and its long-term reliability. Reduced pull and shear forces as a direct result of such contaminant films lead to an insufficient wire connection if the bonding energy is not corrected. This is manifested by so-called lift-offs (figure 1). For example, modern wire bonding devices can generate a friction weld connection with sufficient adhesion on contaminated surfaces by automatically increasing the bonding time and energy. This however causes greaterdeformation of the bond wire. Increased wire breakage from heel cracks is then a frequent result (figure 2). 
Overall, the different bonding times and energies negatively affect the reproducibility shown by widely fluctuating results (large standard deviation). In addition, flux residues may cause adhesion problems when subsequently applying casting compounds such as glob top. Furthermore, wire and chip corrosion (figure 3) has been observed, which was traced back to these residues. Such problems impair the long-term reliability of theassembly or component.  Improving bonding quality Due to the previously mentioned situation, cleaning usually precedes the wire bonding process to improve the bonding quality. This circumvents the balancing act between good wire adhesion on one side and minimum wire deformation on the other. However, it is important to use a cleaning application tailored to the requirements of the wire bonding process. The cleaning application must be capable of removing the described impurities from the bond pad and should not leave any residues on the surface. To determine the influence of different cleaning applications on the wire bonding process, extensive studies were carried out at Zestron. Wires were bonded onto test substrates with different surface finishes using a manual wire bonder by F&K Delvotec (figure 4). A device byDelvotec was also used for the pull and shear tests.  For a representative comparison of the different cleaning methods used in the field, the parameters most frequently used were altered (Table 1). Therefore, the most conventional metallization were used for the tests. The investigated contaminations represent the most significant contamination issues of the most important flux and soldering systems. In addition, the two tested cleaning methods represent 95 percent of the cleaning applicationsin the market.  Effects of the cleaning process The investigations revealed that selecting a suitable cleaning process yields the following improvements: Less bonding energy required > reduced risk of heel cracks
Greater pull/shear force > higher bond reliability
Less standard deviation > unchanging quality/higher reproducibility After optimization of the cleaning process, the bonding energy and time can be substantially reduced to create a reliable wire bond. The danger of excessive wire deformation from overbonding and the associated heel cracks are significantly reduced. The ultrasonic output required for bonding parts, which were cleaned with the surfactant-free Microphase (MPC) Technology was found to be even lower than that of an untreated, newly-delivered part (figure 5). As a positive side-effect, quality fluctuations in outsourced parts can be compensated to a certain degree.  Additionally, with a cleaning process based on the MPC Technology, much greater pull and shear forces canbe achieved and the quality of the bond is correspondingly improved. The danger of lift-offs decreases. Furthermore,the pull and shear forces statistically show less standarddeviation. The overall bonding process is accordinglymore stable and reproducible. The number of exceptionsin the bonding quality is minimized. Figure 6 showsa comparison of the pull and shear forces betweensubstrates cleaned with traditional surfactant cleaners andMicrophase cleaning agents. In an equivalent number oftested bond connections, the pull-off forces of parts cleanedwith surfactant products are lower than those of the partscleaned with MPC cleaners. This is due to the rinseabilitylimitations of traditional surfactants, which remain in athin layer on the surfaces. Surfaces cleaned with thesesurfactant systems therefore also have a greater spread ofpull and shear forces, which sometimes even lie outside ofthe established tolerances. Traditional surfactants thereforetend to be unsuitable for cleaning before bonding.  In addition to the mentioned benefits, metallic substrate surfaces are activated by optimal cleaning. This is manifested by an increase in surface energy which inturn positively influences subsequent processing. The adhesiveness of the surface is increased for subsequentembedding and coating processes. Analytical methods for the qualification of surfaces In addition to performing the described cleaning and bonding experiments, Zestron developed a series of analytical methods over the course of the study to meet the need for simple and fast qualification alternatives. These can be directly used during production for processmonitoring. The bonding process is usually monitored by recording the bonding time and energy, observing the wire deformation, and a statistical comparison of these values with prior empirical results. In addition, destructive pull and shear tests are also used to evaluate the bond quality. Non-destructive methods to test the bondability of surfaces before the bonding process have been neglected until now. Based on Zestron΄s studies, the following methods for the qualification of surfaces before wire bonding appear to be the most recommendable: Demonstrating intact metallization of bond pads by interference contrast
Demonstrating absence of resin/rosin with the Zestron Resin Test
Measurement of surface tension
Demonstrating reactive surfaces with the Zestron Organic Layer Test Investigating the surfaces by means of interference contrast provides information on whether the metallization of the bond pad is intact or not. The test reveals when oxide layers (figure 7) or structural errors exist, which make wire bonding more difficult.  The Zestron Resin Test demonstrates resin-based flux residues which negatively affect the bondability of the surface and can also impair the adhesion of embeddingmaterials (figure 8).  Other organic impurities on the surface, which could influence the bonding and embedding processes, can be determined by measuring the surface tension. The Zestron Organic Layer Test can be used to analyze the reactivity of the metallic surfaces. A highly-active surface is particularly important for the further processing of high power modules bonded to copper ceramic substrates, as well as leadframe-based high power components. When users combine the quoted methods, they can check the condition of the substrate surface and hencecontrol the process in an easy and efficient way. Summary Experiences from day-to-day production show that wire bonding on contaminated surfaces can become a serious issue. Process adjustments are required, which in turn have a negative impact on product quality and reliability. Tests at Zestron showed that cleaning after soldering and before wire bonding ensures reliable processes. However, traditional surfactant or hydrocarbon systems should not be used as cleaning media. They are sufficient for basic applications, but tests have demonstrated that they are clearly limited when used to clean before wire bonding. Modern systems such as surfactant-free MPC cleaners achieved much better results. With these cleaning agents, the results even exceed those of newly delivered base substrates, which are conventionally used as references. Suitable cleaning processes can therefore even compensate the fluctuations in quality arising from the supplier. At the same time, the error rate can be reduced and added value increased by using non-destructive qualification methods in a process control system to evaluate the surfaces prior towire bonding. EM | 
