Emerald Group Publishing Limited
Copyright © 2000, MCB UP Limited
1st International Workshop on Electronics Materials and Packaging (EMAP'99)
Keywords Conferences, Electronics, Materials, Packaging
EMAP '99 was held in Singapore at the end of September and attracted 110 participants, mostly from Singapore and the surrounding region, but with a significant number from the USA and Europe. It was organised by two Singapore institutions, Nanyang Technological University (NTU) and the Institute of Materials Research (IMRE). Most of the papers were from academic or R&D institutions with relatively few from industry.
Among the papers on flip-chip technology was an interesting reliability study by Z.W. Zhong of NTU and K.S. Goh of SPT Asia comparing flip chip on flex, flipchip on ceramic, both using eutectic solder bumps, and flip chip on FR4 using anisotropic conducting adhesive. As expected, the flip-chip on ceramic was more reliable than flip chip on FR4 and eutectic solder gave better performance than conducting adhesive. But the comparison values of cycles to failure and resistance change for temperature cycles -40°C to 125°C gives an indication of the relative performance of the options so an informed judgement can be made of their areas of application. Another comparison between solder and adhesive joints, this time for conventional SMT applications, was described by P. Gratz of Dresden University. The conclusion was that for AgPd coated components the adhesive joints provided similar electrical performance and reliability to the solder joints, even after 1,000 cycles -40°C to +125°C. The finish of the components is critical; pre-tinned components showed high and unstable electrical resistance.
Several papers described performance and characterisation of the polymer materials used in packaging. From NTU came a paper by C.P. Foo et al. describing electron beam curing of a resin coating based on epoxy resins and bismaleimide. The advantage of this process is that curing can take place in seconds at low temperatures compared with hours at elevated temperature for conventional thermal cure. For epoxies the electron beam and thermally cured materials have similar properties but the degree of cure of bismaleimide was limited to 50 percent. Yang Ju et al. of Tohuko University described an interesting method of measuring the moisture content of encapsulant resin in IC packages by determining the phase shift and insertion loss of a co-axial microwave sensor operating at 20Hz. The technique is more convenient and faster than the traditional approach of weighing the package before and after drying; agreement between the two techniques was good. In a paper presented by Judy Cheng of Delphi, several techniques, DSC, TGA and rheometry, were used to characterise the curing behaviour of fluorosilicone gels used for passivating pressure sensors mounted on the air intake manifolds of car engines. After cure the rheology of the gel must be stable if the calibration of the sensor is to remain valid. An interesting aspect was that a gel which performed well in laboratory tests failed to cure properly in the real product due to an adverse reaction with another polymer in the plastic housing. There is no substitute for real life! In another paper on characterising polymers, Ah-Chin Tan of Micron Semiconductor Asia described the use of dynamical mechanical analysis to determine the modulus of epoxy mould compounds as a function of post mould cure duration and moisture preconditioning.
Among other papers on characterisation techniques was one by Abdul Jaleel et al. from NTU in which scanning acoustic microscopy was used to study degradation of the interface between a silicon die adhesively bonded to a copper plate. The reflection amplitude at 25 MHz increased as the interface strength degraded due to moisture uptake by the adhesive. In my own paper I described several applications of infrared microscopy to flip-chip on board structures. Because silicon is transparent in the infrared, IR microscopy can give valuable information on defects such as cracks and delamination of the underfill encapsulant.
There were few papers on wire bonding but J. Seuntjens of American Fine Wire gave an interesting overview of the manufacturing process of something we all tend to take for granted. The twin drives to reduce cost and reduce size both demand finer wire; the gold content of 15µm wire is only 36 percent that of the traditional 25µm wire and the finest pad pitch is 35µm compared with 70µm. The down side is the greatly reduced stiffness, leading to demands for higher modulus alloys.
Perhaps because of the academic background of many of the participants, there were many papers on mechanical and thermomechanical modelling of packages and materials. E. Suhir of Lucent gave an excellent overview of the different approaches and their fields of application together with a brief historical introduction. Of particular interest were Suhir's comments on the different requirements for modelling of photonic modules, in which very precise dimensional tolerances between components are essential, and more conventional modules, in which stress and reliability tend to be the dominant issues. Other modelling papers included applications to vapour pressure during reflow on ball grid arrays (Institute of Microelectronics Singapore), steam driven delamination of plastic encapsulated packages (Hong Kong University of Science & Technology) and thermomechanical stress in proximity sensors (NTU & Pepperl & Fuchs). There was only one paper on purely thermal, as distinct from thermomechanical, aspects of packaging, by Whalley and Sarvar of Loughborough University, UK. That described the packaging requirements of power modules for the avionics industry and thermal modelling techniques to forecast their performance. Cooling was by forced air flow across finned heat sinks and wind tunnel experiments were used to verify the thermal modelling experimentally.
The last day of the workshop was given over to lead free solders. Although the initial drive was based on environmental considerations, the move towards lead free soldering has now taken on a dynamic of its own, driven more by business and legislative demands than concern for a greener planet. Lead in electronic solder comprises only 0.49 percent of the world lead use and no one presented any evidence that the use of lead/tin eutectic solder for electronic packaging had any measurable impact on the real environment. Both Japan and Europe expect to be essentially lead free by 2004. As one speaker put it, the issue is not "whether" but "when" and "how".
The alternatives being proposed are nearly all high tin alloys such as Sn/Ag3.5, Sn/Cu0.7 and a range of Sn/Ag/Cu alloys with Ag from 3.0-4.7 percent and Cu from 0.5-1.7 percent. All of these alloys have melting points in the range 217°C to 221°C, considerably higher than Pb/Sn eutectic. The addition of bismuth reduces the melting point and improves wettability, but Bi52/Pb32/Sn16 eutectic with a melting point of 96°C can form if the solder is used on components having surface finishes containing lead. Once the electronics industry is lead free this may not be an issue, but for the next few years both lead based and lead free technologies need to co-exist. Ironically bismuth is produced as a by-product of lead extraction, so if the world really goes lead free where will the bismuth come from?
The lead free solders have poorer wettability than Pb/Sn and do not generally have the smooth shiny surface traditionally associated with a "good" solder joint, but they are coming whether we like it or not and are already being used in consumer products built in Japan. Issues that arise from lead free solders include their higher melting points which require reflow temperatures of up to 260°C. These temperatures may challenge the use of FR4 for complex multilayer boards and may damage some components, especially if multiple reflows occur. At present most components have lead based finishes and considerable activity is in hand to replace these. Rework and repair will be challenging and problems may arise if lead free and traditional eutectic solder are mixed. For contract manufacturers the fact that there is no consensus on a single solder to replace Pb/Sn is a concern. For a period of several years they may have to run multiple lines with different solders. Some of the solder compositions are covered by patents so licences for their use may have to be obtained. At present there does not appear to be any substitute for the high melting point Pb/Sn10% solder used for flip-chip bonding high performance devices such as microprocessors. It is far from clear whether a lead free solution exists for some high-end products. We live in interesting and challenging times!