LEAD-FREE SOLDERING IMPLEMENTATION IN THE SEMI-AUTOMATIC IMMERSION TINNING PROCESS FOR THE TO-92 PACKAGE INTEGRATED CIRCUITS.

Seventeenth Annual Conference of POMS. April 28 - May 1, 2006, Boston, MA. OM in the New World Uncertainties" Abstract code: 004-0423 LEAD-FREE SOLDERING IMPLEMENTATION IN THE SEMI-AUTOMATIC IMMERSION TINNING PROCESS FOR THE TO-92 PACKAGE INTEGRATED CIRCUITS. Elsa M. Benavides & Rafael A. Delgado draelsa@benavidesphd.com, rafael.delgado@us.bosch.com División de Estudios de Posgrado e Investigación Instituto Tecnológico de Ciudad Juárez, Ave Tecnológico 1340, Cd. Juárez, Chih. 32500 México. http//www.itcj.edu.mx 1

ABSTRACT This paper describes the work developed in the lead-free semiautomatic immersion tinning through a wave soldering process of electronic terminals for Hall-Effect technology components. Currently, new pending environmental legislation of the European Union Waste Electrical and Electronic Equipment and the Restriction of Hazardous Substances will require electronics manufacturers to eliminate lead from soldering. Lead is the major constituent in tin-lead solders alloys currently being used in the electronics industry. A process DOE was developed with five input variables detailed as solder bath temperature, conveyor speed, percentage of copper content, pre-heating temperature and base material pre-cleaners concentration, and nitrogen pressure. Although the type of flux and its preheating temperature are considered critical in this immersion tinning process, they were clearly defined at the initial stages of the research transforming the initial flux type variable into a constant to reduce the DOE scope. The validation tests depend of the selected alloys compatibility with the different lead frame underplatings. The results indicate that the output variables of wetting and brightness acceptance criteria, plating thickness, peeling and solderability tests were all critical. These high impacted variables in conjunction with some others can be used to predict the process behavior. Keywords: tin-lead solders, lead-free alloys, wave soldering. INTRODUCTION The semiautomatic tinning process where this research work was developed has some particularities that make it unique in some way. It s an immersion tinning performed basically through a wave soldering process. This special characteristic forced the researcher to combine the information, experiences and technical skills from two wellstudied processes into a solely different application where qualitative and quantitative variables had to be simultaneously considered and finally analyzed during this research. Empirical investigation in conjunction with Six Sigma Methodology scrutiny concluded that the high impact input factors were mainly the bath soldering temperature, percentage of copper contained in the lead-free alloy (directly related to the solder temperature variable) the conveyor speed and the base metal pre-cleaners composition. Although the flux type and its pre-heating temperature are considered critical in this immersion tinning process, they were clearly defined at the initial stages of the research transforming the flux type initial variables into a constant in order to reduce the DOE scope. The output key variables that supported the success of the work implementation were the wetting and brightness acceptance criteria, plating thickness and peeling test. Solderability is also clearly a critical output for this process application. This last test mostly depends on the selected alloys compatibility with the different lead frame underplatings. During the 2nd Annual Convention for Lead-Free that took place in November 2002, an initiative was established which finally defined that the maximum percentage of lead allowed in any electronic device would be 2

0.1%. Several associations were involved during the discussions like the European SOLDERTEC and Japanese JEITA (Japan Electronics and Information Technology Association). It was initially created as a recommendation for the electronics industry, but later it became an international legislation when the EU WEEE (European Union Waste Electrical and Electronic Equipment) and RoHS (Restriction of Hazardous Substances) Directives decided to adopt July 1 st, 2006 as the deadline to implement this requirement. This determination made obligatory that every supplier that would want to continue or start trading of his electronic products within the European and Japanese markets, should prove that his soldering and similar processes would be lead-free among other substance restrictions. Besides lead, the European RoHS Directive also restricted the use of another five substances, which were considered of high risk for the human health and the environment. Those substances were the hexavalent chromium, cadmium, mercury and some flame retardants utilized in molding better known as polybrominated biphenyls (PBB) y polyibrominated diphenyls-eters-oxides (PBDO). Lead was defined as a substance which nature will have carcinogen repercussions and genetic affectation. Because of these reasons, its use was restricted exclusively in those applications where there is not a similar reliable substitute yet (i.e. very high temperature processes). This change entails an immediate impact for those markets vendors specifically concerning the electronics industry and more recently, in the automotive supply chain. China (the future world greatest market) also adopted equivalent legislations to the EU WEEE/RoHS which target date has been set in some cases, also July 1 st. The obligation is to be ready, the soonest possible, to allow every vendor in the supply chain to perform its required products re-qualifications. This research by itself represents a high impact project and strategic planning for all of us who worked in the electronics industry. This industry is probably the one that has experimented the fastest evolution in the last 50 years since the invention of the transistor.. It represents for us the engineers a true challenge in technology, processes and materials science applications because of the high numbers of lead-free alloys available in the market. Since the end of 2001, there has been a constant development of lead-free components by the big manufacturers, however every process is in some way unique and there aren t two processes entirely equal. The definition of the critical variables in each case depends of many factors. Some of these factors have to do with the product specific application, the equipment costs associated, the specific process under study and the cost-benefit analysis usually requested by the organization. All this should be managed very carefully to maintain the same level of product quality and reliability when using the conventional tin-lead alloy. The projects of the evaluation of new products and processes within the organization, specifically in the Hall Effect IC manufacturing area, represented the kick-off for the qualification of the following soldering and tinning processes: immersion tinning and manual dipping; wave, reflow, immersion and hand soldering. Also and in accordance to the global initiative to convert all soldering and tinning processes to lead-free within the 3

organization, the starting point was to be focused in the identification and implementation of this initiative in the key high volume process of semiautomatic immersion tinning of the Hall Effect IC sensors terminals. The compilation and organization of the information generated during the project development at Honeywell was put together as an initiative to establish this research as an exploratory study since it covered a topic that had not been widely studied even after looking for supporting data from different sources (seminars, webinars, lead-free forums, etc.). However as the topic where became deeply researched, and data was generated trail after trail, the study became more correlation-explicative. This, because one of the main objectives is to establish the relationships among the different processes input variables and their impact or effect on the output key variables. Even that this would suggest a mere correlative study, it does not end at this point since it pretends to decode the causality dilemma. This means that the main purpose is to obtain the explanation about the impact of the critical input variables in order to predict the results of the output variables through a mathematical model. This was only possible through a series of systematic observations that allowed the repeatability of the process under analysis. Even that a process like this process implies the detailed description of the different variables under study; it does not pretend to be independent. The goal is to achieve an analysis by establishing relationships between the variables in order to predict the behavior of this process, considering that it is very specific. In terms of knowledge, the objective is to reach an explicative investigation because of the nature and properties of the new materials utilized. Due to the types of variables to relate, and to the initial goal is to determine which ones are critical as process inputs based in their impact in the following key outputs: quality of the solder bath of the components terminals (variables correlation). The new alloys characteristics and the critical of some process parameters (i.e. temperature window, conveyor speed, and some other will have a high impact in the outputs soldering wetting solder composition and acceptance criteria. This in addition to the flux self-properties that are required to guarantee the correct cleaning and contamination removal from the terminals base material. Bixenman et al. (2003) provided evidence that cleaning will be more difficult for lead-free assemblies. All this involves knowledge in deep about the different materials behavior in order to meet the electronics industry product standards. In its correlative nature, a series of key input and output variables are identified for being DOE analyzed once the process is in control. By this means, there will be a big opportunity to establish a high confidence prediction level. Using the explicative approach, we ll be able to set the bases for similar lead-free processes (i.e. wave and immersion soldering, re-flow, hand soldering and manual tinning) even if the they differentiate quite a bit from the one studied here. This is a great advantage of this research, the methodology and knowledge that will be obtained from it will allow its application to processes where similar materials will be utilized even if the variables may be substantially different. This will only be possible through a clear and systematic structure of the techniques of data management. 4

The greatest advantage of this investigation is based in the methodology and knowledge that will be obtained from it and its application to processes where similar materials will be utilized even if the variables may be substantially different. This will only be possible through a clear and systematic structure of the techniques of data management. In the same manner, the planning and problem definition have been determined through their respective investigation questionings generating the correspondent hypothesis and variables. The theoretical framework is presented and supported by studies mainly performed for wave soldering developed by researchers, since that type of soldering is the most approximate and contribute to this study. This research work is considered a mixed approach because of the involvement of qualitative and quantitative variables. A process DOE was developed with five input variables detailed as solder bath temperature, conveyor speed, percentage of copper content, pre-heating temperature and base material pre-cleaners concentration, and nitrogen pressure. The validation tests depend of the selected alloys compatibility with the different lead frame underplatings. The results indicate that the output variables of wetting and brightness acceptance criteria, plating thickness, peeling and solderability tests were all critical. The goal was to achieve the response to the existing dilemma about the best method to evaluate, qualify and implement a lead-free solution for the semiautomatic immersion tinning process for the Hall Effect sensors in TO-92 package. THEORETICAL FRAMEWORK Ecologic Awareness in the Electronics Assembly Industry and Environmental philosophies and policies Recent terms as Green Manufacturing or Industrial Ecology (Munie et al. 2002) have become, more and more often, part of the industry standards especially for the electronics assembly operations, where the overall environmental impact of any decision made has been taken into consideration. The intent is to optimize the use of limited resources such as energy and materials, which should yield in the most environmental benefit possible. Environmental issues have affected the methods used in electronics manufacturing. Take the 1990 Clean Air Act and the Montreal Protocol for the elimination of the chlorinated fluorocarbons (CFC solvents) as two of the most recent examples of these policies (CFCs were traditionally used to clean electronics hardware). New cleaning (aqueous and semi-aqueous cleaning) and soldering technologies (low residue/no clean) were developed and implemented as a direct result of this legislation. The impact of lead-free solders is much wider than the effects caused by the CFCs elimination since lead is a major constituent in tin-lead solder alloys used everywhere in electronics manufacturing from individual components finishes and boards (through hand, wave and reflow soldering) (Whiteman, 2000) With the WEEE Directive in Europe outlawing lead from some electronics devices produced and imported in the EU by 2006 and foreign competition driving the implementation of lead-free electronics assembly around the world, additional questions concerning the integrity and reliability of the various alloy compositions continue to arise. A tremendous amount of interest exists in lead-free soldering. Most of this is derived from a fear of 5

legislation and marketing activities. This has spurred a great deal of committee and consortia activity, some of which has been very valuable to the industry. The total impact of lead-free implementation depends not just on the characteristics of lead and the alternative alloys but also on such factors as energy use, recyclability, reliability of products and support from an infrastructure for the chosen materials. The environmental metrics required to evaluate the alternative options to lead alloys should rest on: Materials used in the manufacturing of the product (including the environmental consequences of obtaining these materials). Materials consumption during the operation of the product. Energy used in manufacturing the lead-free product. Energy used in product operation. Recyclability and/or re-usability at end of product life (including the new End of Life of Vehicle ELVcurrently under implementation definition). Emissions through the life cycle, i.e. in materials extraction, manufacturing, use and disposal/recycle. Recyclability of manufacturing waste streams. It is this approach, treating the manufacturing process as part of the total environment and not in isolation that characterizes today s industrial ecology. The concept of industrial ecology is one in which economic systems are viewed not in isolation from their surrounding systems but in concert with them. As applied to industrial operations, it requires a system view in which one seeks to optimize the total industry materials cycle from virgin material to finished good, to component, to product, to waste product and to ultimate disposal. Factors to be optimized include resources, energy and capital [Gilbert, 2004]. Once this concept is accepted, the question becomes one of application. The tool-set putting these concepts into practice is called Design for the Environment which is the next step. The move toward lead-free electronics now underway in Europe and Japan is in addition motivated by the requirements of end-of-life management and marketing strategy. Most of the low cost special tailored alloys, mainly tin-copper (both metals with enough reserves, known to support the conversion to lead-free) are patented by different vendors (Cookson, Nihon, Sony, etc.). These alloys can present manufacturing problems because it will be difficult to maintain the exact composition in wave soldering as an example. Also most of the proposed alloys require higher processing temperatures giving rise to a significant increase in energy usage. Even so, the main concern regarding lead-solder has been the leaching of this heavy metal from electronics in landfill. Although at present, several lead-free alternatives have been studied even for years in hope of finding a drop-in replacement for Sn-Pb solder, none have been found. Each one depends of the specific process application and requires a particular study in order to guarantee the same level of quality used to have with eutectic tin-lead. In conjunction to the elimination of lead approach, it has been the improvement of the infrastructure and methods for metal recovery and reuse in electronics. 6

Several proactive actions have to be taken by both, the industry and governments to address the environmental issues from the holistic perspective that represents the conversion to lead-free soldering and similar processes. These can be mentioned as: Develop an industry wide educational program to spread the word of Industrial Ecology and Design for the Environment. Add these attributes (of Industrial Ecology and Design for the Environment) to industry roadmaps. Work to showcase examples of recycling and environmental excellence and improvement of many of them now in place. Foster environmentally Green Design by the government, tax incentives, foster education at the state level to aid industry in making improvements in process (efficiency, materials and power), encourage economically viable recycling for an infrastructure for returned electronic equipment (efficient recycling centers), encourage research in the area of recycling and green design. Raise the level of public awareness and education in the areas of conservation and recycling. Stress the challenges of energy and materials consumption. And finally foster the environmental and business improvement through cooperation Existing Soldering Processes and Technologies The American Competitiveness Institute s Electronics Manufacturing Productivity Facility (EMPF) initiated a lead-free soldering program. The program s objective was to become familiar with lead-free soldering by determining the process variables associated with lead-free soldering concentrating on surface mount technology (SMT) applications. The program then will determine the differences between lead-free soldering, tin-lead and tinlead-silver soldering; with respect to solder joint appearances, board finishes, solderability and process residues. The plan was to compare results from using lead-free solders with tin-lead and SnPbAg solders serving as a baseline. The issues covered in the program consisted of: Screen-printing/components placement. Reflow soldering thermal profiling/equipment. Boards finish solderability. Components finish availability. Lead-free solder joints. Lead-free solder residues. a) Wave Soldering From prior investigations, it s understood that lead-free solders do not wet as well as their Sn-Pb counterpart. Solder paste vendors have indicated that their lead-free solders are not affected by board surface finishes. 7

Consequently, this asseveration can be also applied to our case of study since the surface is bare copper just like the boards mentioned here. By experimentation, this claim will be verified. For the different processes discussed in this paper, we should consider that almost every component is already available in a lead-free finish. Vendors started introducing lead-free finishes almost from the beginning to the market although it s being mainly determined by market demand and specific customer requirements. In addition, there is not enough product reliability data with respect to material compatibility, solder joint (which in our case we will associate with the term of wetting in order to avoid confusions) and electrical reliability (see. Clech, 2004; Seeling and Suraski, 2005). Lead-free finishes under consideration have been Sn, Sn-Cu (and its family series), Pd-Ni, Sn-Ag and Ag although this last one in much less percentage. While it has been possible to manufacture modules with lead-free solders with current soldering equipment, the lead-free solder joints will look different than their Sn-Pb counterparts. The lead-free solder joints will not be as bright and shinny as Sn-Pb and will have a grainy appearance. Havia et al. (2005) reported the solder quality achieved in a lead-free wave soldering process using Sn-Ag-Cu (SAC) alloy. On the basis of their results, it cannot be shown that SAC process could be optimized as well as tin lead process with respect to solder defects. Arra et al. (2002) found in their research that with optimum flux and process parameters, it is possible to achieve acceptable process performance with the SAC alloy using a dual wave system. b) Hand Soldering: iron and pot. EMPF performed brief trials with lead-free hand soldering with the objective to determine the difference between lead-free and Sn-Pb solders. It was determined that the solder tip temperatures were higher. Generally speaking, the solder temperature required was above 650ºF (343ºC) for lead-free versus 600ºF (315ºC) for tin lead. The soldering iron must remain on the solder joint longer, prolonging the dwell time to promote adequate heat transfer to the hardware. However the soldering iron must be removed quickly to avoid causing icicles on the solder joint. The phenomenon is dependent upon de alloy utilized. Operators must be more diligent in assuring that their soldering irons are clean when using lead-free solders as opposed to Sn-Pb solder. Lead-free solders are more sensitive to dirty solder tips than their Sn-Pb counterparts, probably due to the higher solder temperature employed, which accelerates the tips oxidation process. The lead-free solder joints will appear grainy, with a duller finish than Sn-Pb s. It is felt that in order to achieve the same quality of solder joints; a more active flux may be required to overcome any solderability issues. c) Reflow Soldering There is not drop-in replacement for Sn-Pb pastes used in surface mount technology soldering; but when considering melt point, toxicology, cost, availability and chemical resistance, the Sn-Ag-Cu or SAC series has emerged as the most acceptable compromise, both for solderability and reliability. For SMT, the most serious issue is backward compatibility related to insufficient reflow temperatures or time above liquidus for SAC balls- (217ºC 8

eutectic temperature) with Sn-Pb paste combined solder to melt. For the combined alloy in the joint to completely melt, the joint temperature must be greater than 207-210ºC and held long enough for mixing to occur. Low peak reflow temperatures (208ºC) with short time above Sn-Pb liquidus (30-60 sec) resulted in unacceptable solder joints. The Sn-Pb solder paste melted, but large portions of the SAC solder balls were intact. This lack of melting in the SAC alloy originated reliability issues. A solution to this issue is based on longer times above liquidus (90-120 sec) at a higher temperature (222 ºC). Analysis, identification and justification of the new materials This paper takes an interesting depth view of lead-free Sn-Ag-Cu and Sn-Cu-Ni alloys and compares the reliability testing results and process considerations. Although the Sn99/Ag0.3/Cu0.7 and Sn99/Cu0.7/Ni0.3 alloys have a relatively short history in the hybrid circuit and electronics assembly industry, the fact that they re low cost alloys (since they derive from the Sn-Cu family) has made them very attractive for manufacturing operations. Some in the industry feel comfortable utilizing these alloys as lead-free alternatives. Several potential issues have to be considered when working with these alloys. First, the high melting temperature of the alloys, 221-228ºC for Sn99/Ag0.3/Cu0.7 better known as SACX0307; and 227ºC for Sn99/Cu0.7/Ni0.3 known as SN100C in the industry. Second their peak reflow temperatures of 240-260ºC, which in some cases is too high for many surfacemount parts and processes. However, their low silver content condition has made them accessible or the some of the best choices for high volume low cost operations and consequently for the general consumer electronics applications. More important, is that these alloys have shown good reliability tests very similar to the standard tin lead and better results if compared with the original Sn-Cu (Sn99.3/Cu0.7). This has been an advantage over the high-silver content alloys. In addition to cost; the high silver alloys have experimented fatigue-testing fails which further investigation has revealed the silver phase change as the root cause. This is provoked by the various cooling rates at the different areas of the high-silver content alloy. A structural weakness observed during this test could occur in a solder interconnect and potentially lead to a field failure. Despite the concern regarding patent legislation, in general most of the world is setting on the Sn-Ag-Cu (SAC) family of alloys. But which exact alloy formulation should one select? The paper will focus on the low-silver content alloys and their product, process and equipment implications of the selection. Before we go any further, it s useful to compare these two alloys empirically. In general both alloys are very similar since both are considered from the Sn-Cu family. Both offer very good fatigue characteristics and good overall joint strength and sufficient supply of base materials (Nihon Superior (2003). However some minor differences do exist which worth further discussion. 9

Melting points The melting points of these two alloys are very similar (218-228 ºC for SACX0307 and its balance SACX0300 Sn99.7/Ag0.3; and 227 ºC for SN100C and its balance SN100Ce Sn99.7/Ni0.3). Although it is debatable as to whether this will have an impact in real world applications. However if one can control the wave-immersion process strictly, the temperature reduction will have a positive effect initially in terms of reducing the components exposure to high temperatures avoiding reliability risks; secondly in extending the life of the solder alloy because of the reduction of the copper dissolution rate in the deposit, and finally in prolonging also the life of the equipment since even 5-10 ºC reduction when continuous use can contribute to this purpose. Wetting When comparing these two alloys, it is necessary to question why one would select an alloy with silver content instead of nickel, as this will increase costs (silver and nickel better known as stabilizers in these alloys). Some have theorized that the silver content, even if it s low, will aid in wetting. However, as the wetting tests have demonstrated, the alloy with low nickel content actually wets stronger and faster than the one with the silver content, although this better condition is obtained at a higher temperature because of the nickel material intrinsic properties (typically 10ºC higher). While it is logical to contain costs, there are some issues with Sn-Cu series (where the SnCuNi SN100C belongs to) alloys that must be considered. First, the high melting point of the leadfree alloys selected (227 C typically) makes them in some cases, prohibited in many temperature sensitive applications. In addition, and as widely proven, these SACX0307 and SN100C lead-free are poorer wetting alloy compared to the tin-lead solder. However the cost tendency drove the efforts of the experimentation through the low cost alloys mentioned above. Initially and during the preliminary evaluation runs, the process itself forced us basically to increase the immersion contact time at the wave in order to try to reach above 2 sec if possible. Scimeca et al. (2003) point out this is the typical value for lead-free alloys. However it has a direct impact in the process output, which by the way, was very closely monitored by production management. The contact time increase was accomplished by reducing the conveyor speed from the typical tin-lead 1.1 ft/sec to 0.6 ft/sec. We kept in mind that insufficient contact time would result in incompletely tinned leads. We noticed also that wetting time could be shortened by raising solder temperature, although this will have a direct impact in the solder life time and also might affect the critical internal gold wire bonding of the IC since raised temperature would be too close to the absolute maximum operational. An important difference between the SACX0307/SACX0300 and SN100C/SN100Ce is their good fatigue characteristics (similar to tin-lead) if compared with the pure Sn-Cu alloy. This poorer wetting results in the mandatory use of a nitrogen atmosphere to guarantee the optimum cleaning of the solder. More active fluxes are also required to minimize the wetting related defects. An important consideration when deciding about the flux 10

selection has to be with the intrinsic nature of the friendly-environment lead-free process itself. In our case of study, the corporate implemented an environmental policy, which restricted the use of alcohol-based fluxes. Tests information available (supported by the market main flux vendors) has indicated that the best options to start are the volatile organic compounds free or VOC-free fluxes. After working very close with the flux manufacturer, the Superior 30DS (double-strength) VOC-free watersoluble became the right choice because of its environmental-health safe and lead-free specific application design. This flux was defined after several short trials where specific fluxes from different suppliers were evaluated and once reviewing the technical specifications of the samples received. The conclusion was that in order to meet both the environmental and lead-free requirements, the initial selection had to be a VOC-free water soluble type because the immersion tinning requires a post cleaning operation to remove all undesired residues from the process The Superior 30DS which met above two requirements and with a very good pre-heating activation temperature range (90-135 C) instantly became the number one selection for our evaluation. The fact that this flux will have a low activation temperature will avoid the necessity of changing this section of the equipment. In addition, the flux reaches is maximum activation at 260 C, which is the temperature the new lead-free solder alloy suppose to be working at. The selected flux was entirely compatible with both lead-free alloys and as a background, it had been successfully utilized by a couple of tinning or immersion soldering manufacturers although both at a much higher soldering temperature (330 ºC typ). This was going to be the first time the Superior 30DS flux would be used at 275 ºC. Although the soldering quality achieved with fluxes designed for traditional tin-lead soldering was also reasonable; the quality could not be sustained for more than 2-3 days which agreed to what vendor had indicated before initiating the trial runs. This difference may originate from the fact that lead-free fluxes are usually higher solids content and higher acid values. In order to compensate the residues left by these high acid fluxes, a watersoluble type was chosen to match the post washing section of the immersion tinning process. Fig. 1 shows the Semi-automatic Immersion Tinning Process for TO-92 IC. Similar to a traditional wave soldering process, this lead-free immersion tinning-wave soldering was recommended to have a slightly higher temperature in order to mitigate the thermal shock. When raising the preheating temperature, caution was taken against deactivating flux prematurely. Since a rise in temperature subject components to greater thermal stress, it was weighted that for critical characteristics purposes, the more adverse effect to the ICs was a higher preheating temperature since this would result in a poorer wetting condition. After some analysis, it was determined that the component could withstand the thermal shock originated by the switching from 135 to 275 ºC in a matter of 1 sec typ. The raising should be in such a way that soldering intermetallic will occur. 11

Patent situation It is desirable for the industry to find an alloy that is widely available. Therefore, patented alloys have been viewed as undesirable. Unfortunately, the alloys discussed throughout this paper belong to this last classification: SACX0307 patented by Cookson Electronics and SN100C by Nihon Superior. This is because there isn t still in the market similar low silver or nickel equivalents that can perform as well as the patented ones. Qualitek has initiated with its low-silver own alloy Sn99/Ag0.2/Cu0.8 own alloy although a little late we believe. A more circumspect view needs to be taken to understand the impact of patents and the true number of sources available for these alloys. For example, while SN100C started being manufactured and distributed by Nihon Japan exclusively; it s being recently released for being manufactured and distributed by several other vendors who have merged in this lead-free enterprise. This joint has created a wider availability of the alloy within the US where this material was obtained for this research work. Cost of metals The high silver content alloys results in a costly bulk solder form alloy (typically in the range of $18+ per pound). To fill a wave soldering pot of 250 lbs (like the one used for this paper experimentation), every lb used represents a difference of $10-13/lb. To combat this expense, the alternative patented alloys studied here have emerged as the best options which standard cost per lb is in the range of $9-11. The risk of this dual alloy process remains since we should remember that lead-free technology would be in combination with tin-lead for some time until the transition is fully implemented worldwide. Dual Alloy Assembly It should be noted that, in addition to the problems associated with the use of a new lead-free alloy, the utilization of two solder alloys (i.e. Sn-Pb and Sn-Ag-Cu or Sn-Cu-Ni for wave soldering and similar applications) could result in application incompatibility as well. Although it s is undesirable to intermix two alloys because this could result in non-uniformly alloyed solder joints, it s also known by experience that this intermix is inevitable due to the electronics world won t switch to lead-free simultaneously. The whole process could take years before being entirely lead-free and even then, we will end facing the issue that there will be several approved lead-free alloys combined in the different soldering processes. The bottom line is that a dual alloy assembly process could potentially result in reliability problems. This is another reason to be very careful when deciding about the leadfree alloy to be implemented. There isn t a drop-in solution especially if there are several soldering processes that need to be converted to lead-free in a very short period of time (i.e. wave, reflow, immersion and hand soldering and even immersion tinning). 12

Fig. 1 Fig. 1 Semi-automatic Immersion Tinning Process for TO-92 IC Although the materials standardization should be the main goal, it s very possible that we will end with different fluxes for the same alloy depending of the specific application. For example: while a water soluble flux may be the best option for a wave soldering or immersion tinning process, a no-clean flux would be the optimum for hand or reflow soldering where no post cleaning is possible. As pointed out earlier, silver is the cost element in the Sn-Ag-Cu alloys. Since the low-silver content alloys eliminate the potential for silver phased change problems and offer improved wetting and a slightly lower melting temperature (in addition that they are available from several solder manufacturers), they have been recommended for widespread use in Japan by the JEIDA industry organization. The low-silver alloys provide users with the advantages of the Sn-Ag-Cu family of solders at a lesser cost, which makes them suitable for all soldering applications contributing to eliminate the problems associated with a dual alloy process. Equipment Considerations Lead-free solders may cause severe corrosion to materials used in wave and similar soldering machines like our Corfin solder-dip immersion tinning pot and pump. Different solutions to prevent material degradation are for example titanium construction, nitrided stainless steel treatment (which is the option we selected for our new leadfree pots), melonite QPQ coating, ceramic-coated stainless steel (option utilized for our manual dipping pots matter of an ulterior study) and cast gray iron. The two 250 lbs capacity solder pots used during the research 13

project were nitrided stainless steel made in order to withstand the eroding impact of the lead-free alloys. Pots were manufactured to have a 10-year lifetime. Solder and Dross Concentration During the research, solder and its dross were analyzed regularly. Dross was analyzed by solder vendor laboratory to observe if solder and dross compositions differ from each other. Any copper percentage above 0.7% would indicate a solder degradation by copper dissolution from the IC lead-frames. It was supposed, based in other investigations and technical information and seminars from lead-free specialists, that the intermetallics between tin and copper like Sn 6 Cu 5 are heavier than SAC/SN solders meaning that they will not float on the solder surface like with tin-lead. It was assumed that intermetallics heavier than solder will sink to the bottom of solder pot and on that way copper would not be removed with dross removal in the same proportion as there is copper in the solder. On that way, copper concentration will increase. Copper will dissolute from the component lead frames and this process characteristic is even more critical in lead-free alloys than in tin-lead even after a few hundreds of lots. If by any cause, the lead-free tinning process (pot) has been contaminated by lead over solder substitution or solder transition, lead dissolved into the solder pot from that contamination (even if it s just the components) might entail problems. If the process has been contaminated and the applied solder contains 0.1% of lead, the lead concentration will not be considerably lessen in the solder pot even if new solder is introduced into it. Almost the whole solder volume ought to be changed, which would be a remarkable expensive action. An important note to mention at this point is that during the different trials of the research, it was noticed that needles formations tended to appear when utilizing the SACX0307 alloy every time the solder was reaching those few hundreds of lots just mentioned in above paragraph. This needles formation behavior did not happen with SN100C alloy. At the beginning the needles were associated with iron or another heavy metal contamination, however they seemed to appear almost every time wetting problems started even on the surface of the solder.. This was contradictory with the expectations of this lead-free alloy. Laboratory analysis identified a high copper concentration as the root cause of those needles determining that copper was dissolving at a much higher rate than initially expected. The copper concentration raising was in direct relation with the fact that every certain dozens of lots, the tinning process forced itself to increase its solder temperature in 5-7 ºC because of the wetting problems experienced each time. When reaching 310ºC (which is the absolute maximum operational for the Hall devices) a high concentration of tin-copper whiskers on the surface is noticed and they start to stick to the IC leads. A cleaning process basically consisting in removing the dross and whiskers then is performed. The tin-copper whiskers are found at the bottom of the pot just below where the wave is produced. Even after this cleaning process, the wetting problems did not go away and it was necessary to start fresh with new solder. The root cause was determined to be an accelerated copper dissolution rate due to the continuous increasing of 14

solder temperature, which was necessary to correct the wetting problems. Previous studies had reported less large primary Ag 3 Sn precipitates in SAC alloys depending of the cooling rate of the specific alloy composition. Large primary Ag 3 Sn would have an influence on the ductility of the alloy. However in our particular case of study the Sn 6 Cu 5 intermetallics were the ones primarily found as whiskers when composition was analyzed at the laboratory. Evaluation, Analysis and Qualification Procedures (Product Reliability And Process Repeatability) Since copper dissolution can not be eliminated but just reduced in acceleration, we can only hope it can be controlled to maximize the number of processed lots and equalize it to tin-lead behavior. This is achievable by following the next steps: Copper dissolution rate is much higher in this immersion-tinning-by- wave process than in any standard PCB soldering process. This is because the lead frames of the ICs are bare copper C19400 without any under coating. Then copper frame immediately starts to dissolve as soon as it makes contact with the wave. As the copper concentration increases, a higher temperature is required to maintain the same initial wetting characteristics. This makes a no-return cycle until reaching the maximum allowed temperature of the device. The start point was the replacement of the SACX0307 (Sn99Ag0.3Cu0.7) alloy for the SACX0300 (Sn99.7Ag0.3) as the preferred alloy. By eliminating the initial copper concentration, the life of the solder in the deposit can be prolonged. Although solder manufacturers theory had indicated that the less copper in the alloy, the higher the rate of dissolution until reaching its natural stabilization. This means that the initial 0.7% copper concentration also represents the rate dissolution stabilizer. Not having copper in the starting alloy should increase the dissolution rate until reaching stabilization from which both alloys will start behaving the same. This characteristic would represent that both alloys could end having the same lifetime. However the only alternative was the SACX0300 since the SN100C required a higher operational temperature (290-295ºC), which was close to the device allowed maximum absolute (310ºC). Any process variation could raise the solder temperatures and in order to avoid this potential situation, the SN100C and its SN100Ce balance were placed as second alternatives in case the SACX0300 would have not worked. The initial temperature of the fresh solder was maintained as low as possible at 262ºC instead of going to 275ºC or higher from the beginning. The initial optimum was set by visual examination of the wetting at the process output until it was considered completely acceptable and repeatable. This 262ºC starting temperature was very close to the 260 C used for tinlead. Cleaning by dross elimination was performed once a week minimum, or even more often depending of the wetting inspection feedback. Although more solder was consumed (probably an additional 15-20%), this process 15

helped in achieving the 1000 lots goal set by engineering management. This amount represented the same quantity of lots regularly processed when using tin-lead. By the time this research was considered complete, a total of +1300 lots had been processed with the same leadfree solder change showing no signs of remarkable copper dissolution (corroborated by the weekly laboratory analysis performed to the solder sample showing less than 0.4% copper) where solder temperature had just been increased to 272 ºC leaving still a 30ºC gap before reaching the maximum allowed. The dross removal was possible by reducing the solder temperature to 240 C. At that specific temperature, the dross emerges to the surface and it can be easily removed assuring the periodic elimination of tin-copper whiskers The refill or balance was done by using only the SACX0300 which became the sole approved alloy for this immersion tinning process. Conveyor speed could be increased although not as fast as with tin-lead. End conveyor speed was 0.85 ft/sec with opportunity of continue being improved. Acceptance Criteria and Specifications Both tin-lead and lead-free alloys are going to be used in the same factory and even in the same process during the transition period. This is because automotive and some other customers will continue ordering tin-lead products while the rest has already requested them in the lead-free versions. Over this time, caution must be taken to prevent solders from mixing with one another. Each of the two processes requires separate and well-marked tools so that no tin-lead solder will be transferred to the pot for lead-free and vice versa. On adding solder, the operator must make sure that no tin-lead bars are accidentally put in the lead-free pot and vice versa. Every solder bar has its composition marked over it. Solder bars of different materials should therefore be stored well apart or have their composition clearly marked over the packaging (ProTechnik, 2000). Regarding wetting in general, solder tinning of both lead-free alloys fulfilled the IPC-A-610 workmanship standard specification. However, the wetting result could fall closer to the acceptance limit depending of the frame under-plating, the direction of immersion wave solder tinning (parallel preferred to perpendicular) and the frame design itself. Another major difference was the rougher and duller surface appearance exhibited by the lead-free alloys (especially the SACX0307), which is only cosmetic if voids and other porosity imperfections are not present. Galarza et al. (2002) reported the impact of soldering atmosphere, solder bath temperature and conveyor speed concerning the soldering defects needs to be analyzed. For our study, tinning is the soldering equivalent and the average bridging and wetting results were considered. As could be seen, nitrogen clearly reduced bridging due to better conservation of flux activity and improved wetability in a non-oxidizing atmosphere. 16

Fig 2 Copper Frames before and after the Immersion Tinning. METHODOLOGY The correct utilization of all the resources during this research allowed us to obtain the answers for the key objective of this research. Which match alloy-flux should be selected based in the product and process principles? And, which technique should be applied to allow us the establishment of a reliable prediction method for the process behavior? The specific piece of equipment under study is the semiautomatic immersion tinning solder-dip machine where the Hall Effect sensors TO-92 package are processed. The Standards from the Association Connecting Electronics Industry (2005) were used. Data collecting and processing Using Design of Experiments (DOE) the approach for the different evaluation runs will be presented. From here, the results of the critical factors will be analyzed. The DOE has been planned to be a 2 levels and central points, complete factorial and with at least one replica, since the tinning process it allows us to develop the whole experimentation. To approach the DOE and also to analyze the data generated from it, we were assisted by the statistical software Minitab a) Critical factors: To define the factors that should be considered for the DOE (screening phase) we used the Methodology Six Sigma in conjunction with the existing studies about wave soldering. Also we considered the process experience from the key personnel and the new alloy-flux suppliers expertise about these new materials, (all together) The formats were focused in collecting the most valuable information for the research. b) Acceptance criteria: KAPPA studies supported by the new annexes for lead-free soldering contained in the IPC Standard IPC-A-610, will assure the product reliability and best available training of the key people that will eventually be inspecting the product and monitoring the process. This is in order to standardize the 17

acceptance criteria within the organization. We should not forget that qualitative (proportions) and quantitative (averages) data would be handled simultaneously. c) Sample size: A preliminary sample size of 6000 peaces has been set which equals one production lot to guarantee the repeatability of the process. This is based in the current experience and knowledge we have about the immersion tinning process. The final sample size will be according to the results obtained from the evaluation runs and in accordance to the MIL-STD sampling tables for equivalent populations contained in a production shift of 24-26 lots. d) Supporting tooling: Data analysis tooling like: Pareto analysis (solder bath defects), cause-effect diagram (critical factors screening), SPC (for the critical factors identified through the PFMEA and Control Plan), t- tests for difference of means and possibly correlation-regression analysis would be used during this study. e) Primary and secondary sources: The data collection was taken by direct observation when the runs were performed.. The equipment was available so most of the information came from the primary source, which is the immersion tinning process. The rest of the primary sources were the alloy supplier technical support and the materials analysis laboratories and also from the specifications of the new materials. The secondary sources have been mainly seminars, webinars, scientific magazines, studies and papers from other organizations that utilized similar processes and specialized web sites and forums. As the deadline comes closer, day-to-day, there are more and more research papers published mainly through scientific bulletins in the web showing the reliability results about the different processes, materials and applications web showing especially the latest reliability results about the different processes, materials and applications. The whole research was performed as a direct field study since it was performed at the moment the process is running with the modified equipment dedicated to it once it is been converted for lead-free. RESULTS Solder joint reliability testing In order to analyze the reliability of SACX0307 and SN100C alloys, both were subjected to various thermal and mechanical fatigue tests. The description and results of these tests are detailed in below paragraphs. Thermal cycling: a 30 pc TO-92 Hall Effect sensors sample from three different lots were tinned using SACX0307 and another three samples from same lots were tinned using SN100C lead-free alloys. All these samples were thermal shocked from -40 o C to +125 o C for 1000 cycles @ 30 min cycle. Tinned terminals were then examined under magnification and cross-sectioned inspecting for cracks. Post-test inspection showed that neither of SACX0307 nor SN100C exhibited any solder crack issue as a result of poor wetting (peeling). In addition, well-formed solder layer between base metal (copper C19400) and the lead- 18