The role of Cu-Al IMC coverage and aluminum splash in Pd-copper wire HAST performance

Abstract

Copper wire bonding process has become popular interconnection process over gold because of its obvious cost advantages as well as lower electrical resistivity and good thermal conductivity. Recently, Pd-coated Cu wire is emerging as an alternative to bonding with bare Cu wire to prevent copper oxidation during the bonding and improve manufacturability [1]. Compared to bare copper, Pd copper wire shows robust bonding process, especially offers better 2nd bond performance and better reliability. Although copper wire bonding technology has been already at the phase of mass production, the discussion about reliability failures is still ongoing. The occurrence of the related open circuit failures are often reported as a consequence of humidity related stress test, Biased-HAST and PCT[2]. It is a specially challenge for the device with thicker bond pad metal. Most of open failure is caused by galvanic corrosion of one or more of intermetallic layers formed at the Cu-Al interface post HAST or Auto clave test. In this paper, a CMOS65 nm low k device with 47um fine pitch and 2.8um bond pad metal thickness was selected to establish indicator to pass HAST for the device with thicker bond pad metal. First, the distribution of the Pd on the surface was investigated under EFO parameter combination by HPM, SEM and EDX. Next, Capillary feature design was studied to shrink aluminum splash. Capillary with different MTA and chamfer angle were screened to know how these factors affect aluminum splash. Pro-stitch function of K&S ICONN also was studied to improve 2nd bond performance, bare copper as control. Third, the 1st bond parameters were optimized using selected capillary aimed to improve aluminum splash. Critical responses such as Ball size, Ball height, wire pull strength, ball shear strength, and stitch pull strength, cratering, IMC coverage and aluminum splash and remnant were studied to understand effect of Pd copper. DOE (Design of Experiment) and RSM (response surface methodology) was used to optimize the wire bond process. Thermal aging test coupled with wire pull and ball shear test with recording failure mode were studied. Last, four cells were built to be subject to HAST. The four cells were determined based on previous pass and failure, it covers different bonder platform, difference 1st bonding concept and different IMC coverage and aluminum splash. HAST& UHST pass rate, Ball shear, IMC coverage and aluminum as key wire bond response were studied. Studied shows Ball shear and IMC coverage are effective indicator to pass HAST and UHAST stress test. IMC coverage play main role in addressing moisture induced corrosion issue. Regarding IMC coverage, big void in ball center is not preferred and should be avoided. The gap between ball periphery and pad should be considered seriously.