Copper wire bonding process in leaded packages with zero loss in quality, capacity, scrap machine efficiency

摘要

For the past few years, acceptance and implementation of copper wire has grown from optional to mandatory, especially after gold price increased triple for past 5 years. However, there are many companies still not able to achieve desirable performance, especially without trading off other performance indices such as quality, capacity, scrap or machine efficiency. The copper wire has two fundamental properties that can cause complication in wire bonding, it is harder than bond pad metal and it gets oxidized easily. The bonding of copper wire will require tedious optimization on all elements that described in SIPOC chart (Supplier, Input, Process, Output and Customer). Base on many comprehensive studies performed in conversion to copper wire, there were many breakthrough findings and solutions obtained. The first observation was surfacing out of many reject modes in production mode, which required a lot of improvement efforts. Normally, execution of several Six Sigma DMAIC (Define, Measure, Analyze, Improve and Control) projects could improve process defect level to very minimum, as good as gold wire bonding process. On potential quality risk, new process control method would need to be introduced in both development and production mode. New innovative control method would address all the potential risks due to bond pad metal displacement and copper oxidization. Despite that, additional bonding time was required for copper wire bonding to compensate pad damage risk. That caused increment in bonding cycle time, sometime reaching 30% loss. In regards to this, multiple approaches might be required to study every millisecond contained in the entire cycle time. At the end of study, combination of all the optimized settings could result in same throughput performance as gold wire bonding process. As for wire bonder, there were many occasions that same machine ran well with gold wire just could not produce good results after converted to copper wire. This repeated many time- until achieving breakthrough in removing mental block, copper wire bonding process required more stringent machine conditions. Every single wire bonder required upgrade to achieve new bonding conditions. Furthermore, optimization of several machine hardware designs could really make significant differences in copper wire bonding performance. Improper hardware design could result in ball shear performance differed by 50% among units on the same leadframe strip. The other observation was process parameters optimization through normal DOE (Design of Experiment) and RSM (Response Surface Methodology) might not lead to robust and stable performance across all devices. That could create a lot of doubts about the process robustness, and affecting overall progress of copper wire conversion. With multiple studies, combining leadframe design, material selection (capillary & wire), machine hardware and parameters optimization, a new process window was created. This window could be used commonly across all devices with very minimum adjustments. It also turned out to be the most robust and stable process ever found. In most of the improvement activities, when 1st and 2nd bond reached good performance, it could lead to good machine efficiency. Additional works on machine was on oxidation control, especially when BSOB (Bond Stitch on Ball) process was required. As for bond pad factor, assembly team need to work closely with wafer fabrication plants to develop robust structure and reasonable probe mark condition. A new set of design rules was required to enable implementation of copper wire bonding process in current and future products. Throughout all the improvement activities to convert from gold to copper wire bonding, the most significant factor found was mindset change. The copper wire bonding process does not only require conversion activities, but actually it requires full development efforts, just like what we have done for gold wire since 50 years ago.