PART II OF A II-PART SERIES ON LEAD-FREE
BY JOHN H. LAU AND KATRINA LIU
Part I of this II-part series provides a discussion about global lead-free regulations, lead-free solder pastes and products, and lead-free components. Part II includes PCBs, SMT assembly processes, solder joint reliability, transition issues and substitutes.
Lead-free PCBs
There are many different surface finishes for lead-free PCBs. Included are: organic solderability preservative (OSP) Entek; NiAu (electroless Ni and immersion Au, ENIG); ImAg (immersion or galvanic Ag); ImBi (immersion Bi); Pd (electroplate or electroless Pd); NiPd (electroless Ni and immersion Pd); NiPdAu (electroless NiPd and immersion Au); ImSn (immersion Sn); NiSn (electroplate Ni and Sn); SnAg (electroplate Sn and Ag); and HASL (hot-air solder leveling) SnCu. The leading lead-free PCBs are OSP with Entek, ENIG, ImAg and HASL SnCu.
As a result of much higher lead-free SnAgCu solder reflow temperatures, and requirements such as glass transition temperatures, Young's module and moisture absorption resistances of PCB materials are higher. Thus, costs and reliability pose great challenges.
Lead-free SMT Assembly Processes
A key difference between SnPb and lead-free SnAgCu SMT assembly processes is solder reflow, because the melting point of SnAgCu is 34°C higher than that of SnPb. Higher reflow temperatures may damage components. Recently, extensive research efforts have focused on redesigning the reflow machine for SnAgCu solder pastes to keep peak component temperature to a minimum. This enables manufacturers to automate soldering of low-heat-tolerance components using SnAgCu lead-free solder alloys, takes pressure off component suppliers and reduces costs.
Oki Electric and Furukawa Electric jointly developed a heating technology for lead-free soldering, called Component Temperature Control Reflow Technology, that combines far-infrared radiation (IR) with heated air (convection) to heat lead-free solders in a reflow furnace. By adding a cooling mechanism to the convection heating area, thermal effects of IR heating can be controlled, making large temperature differential settings possible. For example, topside reflow components can be kept at moderate temperatures while melting the SnAgCu solder at high temperatures. This is achieved with a top/bottom temperature differential heating system and high-heating capacity IR heating on the bottom side, with convection heating on the top.
Lead-free Solder Joint Reliability
Solder joint reliability is defined as the probability that joints will perform their intended function for a specified period of time, under a given operating condition, without failure. Are lead-free solder joints more or less reliable than SnPb solder joints? Are lead-free solder joints reliable? The industry does not have enough data to answer these equations now. More lead-free reliability tests are necessary, and field data must be collected. However, based on material properties of SnPb and SnAgCu solders, some insights of thermal-mechanical reliability may be obtained.
 Table 1. Creep material constants for SnPb and SnAgCu solders. |
The material properties of 95.5 wt percent Sn-3.9 wt percent Ag-0.6 wt percent Cu solder alloys such as the Young's modulus (E), coefficient of thermal expansion (CTE), and normal creep strain rate (dε/dt) are given by:
 |
null
In these equations, E is the Young's modulus (GPa), R is the gas's constant, ε is the normal creep strain, and σ is the normal stress (MPa). The last equation can be rewritten for the input of ANSYS as:
 |
null
The constants C1, C2, C3 and C4 are given in Table 1, where the material constants for the SnPb solder are also provided.
Figure 1 shows Young's modulus vs. temperature plots of lead-free and SnPb solders. The Young's modulus of the lead-free solder is larger than that of the SnPb solder. When the assembly is subjected to thermal cycling loadings, stress in the lead-free solder joints is expected to be higher than in SnPb solder joints. Fortunately, the low-cycle, thermal-fatigue life of solder joints is dominated by strains, not stresses.
 Figure 1. Young’s modulus of SnPb and SnAgCu solders. |
The creep strain rate difference between lead-free solder and SnPb solder is temperature dependent. The higher the temperatures, the higher the creep strain rate difference. Creep strain rates of lead-free solder are lower than SnPb solder at a given stress, for most temperatures. For example, at temperatures greater than or equal to 50°C, the 95.5Sn-3.9Ag-0.6Cu solder has a creep rate that is half to two orders of magnitude lower than the SnPb solder for the given stress level. Hence, significantly lower creep strains are expected with the 95.5Sn-3.9Ag-0.6Cu solder in comparison with the SnPb solder.
A 3-D finite element model captured the construction along a diagonal strip from the 256-pin PBGA lead-free assembly's geometric center to a corner. Because of the mid-plane symmetry, the mesh actually models a one-half strip (with one-half of a solder joint) using the appropriate in-plane constraints placed on one symmetry plane. Coupled in-plane translations are applied to the other symmetry plane to produce a state of generalized plane strain. Using exclusively hexahedral solid elements, the model can capture the precise shape of the packages' solder joints and potential DNP (distant to neutral point) effects while retaining significant computational efficiency over full octant models. ANSYS is the code selected for modeling and analyses.
Temperature profiles imposed on the PBGA assembly consisted of three temperature loading conditions: (1) 25 ↔ 75°C; (2) 0 ↔ 100°C; and (3) –25 ↔ 125°C. For all cases, the cycle time was 1 hour and the ramp-up, ramp-down, dwell-at-hot, and dwell-at-cold were each 15 minutes. In each model simulation, the solder joint assembly was subjected to five cycles of one of the three temperature regimes.
Creep responses for multiple cycles were observed when the hysteresis loops stabilized. Creep shear strain vs. shear stress loop became stabilized after the third temperature cycle (Figure 2).
 Figure 2. Shear stress and creep shear strain hysteresis loops. |
For the PBGA assembly, lead-free solder assemblies produce larger shear stress ranges with increasing applied temperature ranges. Also, for ΔT = 100°C, the shear stress ranges for lead-free solder is larger than that of SnPb solder. This is because the Young's modulus of the lead-free solder is larger than that of SnPb solder.
For the PBGA assembly, lead-free solder assemblies also exhibit larger creep shear strain ranges with larger temperature spans. As illustrated by ΔT = 100°C, creep shear strain ranges in the lead-free solder joint are smaller than those in the SnPb solder joint. This difference in creep deformation is a direct result of the lower creep strain rates of lead-free solder vs. those of the SnPb solder.
Creep strain energy density per cycle (ΔW) is determined by the area within one of the hysteresis loops after the third temperature cycle. Thermal fatigue life of solder joints (dominated by creep responses) may be predicted by the following equation:10
Nf = Ψ(ΔW)Ψ
where Nf is the number of cycle to failure, ΔW is the creep strain energy density per cycle, and Ψ (always positive) and φ (always negative) are constants for solder joints. In this equation, ΔW can be determined by creep analysis of the structure subjected to the specified loading conditions. However, Ψ and φ are usually determined by isothermal fatigue tests of the real solder joint. Since constants for lead-free PBGA solder joints are unavailable at this time, thermal fatigue life prediction of 256-pin PBGA lead-free solder joints is impossible. Thus, fatigue crack-growth constants (determined by isothermal fatigue tests) for lead-free PBGA solder joints are needed to make quantitative solder-joint thermal-fatigue life predictions.
Lead-free Transition Issues
People want safe, 'green' products that pose no risk to human health or the environment. Figure 3 shows two of the most ideal packaging technologies (PQFP and PBGA) for making green products. Lead-free plays an important role. However, the electronics industry cannot be converted from SnPb soldering to lead-free soldering in one day. Figure 4 illustrates possible routes to lead-free soldering, i.e., using lead-free soldering pastes with lead-free plated leads or solder balls/columns. Transition A is the most direct route; however, because of the existing infrastructure and inventories, it only applies to very specific and simple products.
 Figure 3. 'Green' products with flip chip, WLCSP and PBGA. |
Depending on the specific product, the electronics industry may take both route B (using lead-free solder pastes with SnPb-plated leads or solder balls/columns) and route C (using SnPb solder pastes with lead-free-plated leads or solder balls/columns) to totally lead-free soldering. For most consumer, computer and computer-related products, route B is being applied, since these products have a marketshare advantage when labelled 'green.' However, during the transition period, these products may not have all the necessary (lead-free) components available for their PCBs. Thus, OEMs and EMs will continue pushing their component suppliers to go lead-free so they can do 100 percent lead-free soldering. Concerns of forward incompatibility are greater voiding in the SnPb solder balls and potential reliability concerns. These concerns are related to the higher lead-free reflow temperature on molten SnPb solders, conventional components and PCBs, and the potential for lead-rich areas forming in the solder joints.
 Figure 4. Lead-free transition issues. |
For most servers, storage, storage array systems, monitoring and control medical equipment, network infrastructure equipment for switching, signalling, transmission as well as network management for telecommunication products, route C is being applied. These products are exempt from lead-free status until 2010, under the RoHS directive. Thus, there is no immediate urgency for these products to be converted from SnPb to lead-free soldering. However, since the industry is transitioning, some of the SnPb components necessary for these products will convert to lead-free components before 2006. There are two primary reasons for this. First, due to cost, component suppliers can only carry dual-line production of SnPb and lead-free components for a short period of time. Second, the volume of components (e.g., memories) for these high-end products is only a fraction of that needed by the consumer- and computer-related products, which are rushing toward lead-free production through route B.
There are many concerns over route C. A number of these are summarized in the following:
- Cross-contamination of lead-free and SnPb solders during repair or rework in the field (this can also apply to route B).
- New wave solder pots are needed, adding equipment cost for the EMS manufacturer.
- Component and board reliability targets may not be met in certain applications (also a potential concern with route B).
- Overheating of PCBs or components during rework or assembly.
- The criteria for acceptable solder joints may get mixed up by the contract manufacturer, i.e., the voiding and wetting criteria (allowing more voiding and poorer wetting) acceptable for lead-free may inadvertently be applied to SnPb solder joints.
- Incompatibility of fluxes with SnPb solder; e.g., certain fluxes developed for lead-free are more reactive with Pb and can cause corrosion if used as a “no-clean” on SnPb-containing assemblies.
- Incompatibility of SnPb solder pastes and lead-free components.
- If Bi is used in lead-free components as a surface finish, the SnPbBi ternary eutectic phase (melting point temperature = 96°C) can cause early failure, fillet lifting and other problems.
- When lead-free manufacturing becomes mainstream, SnPb manufacturing may have added costs.
- Challenging supply chain logistics of supplying lead-free and SnPb during a potentially long transition period, such as part numbering and inventory for all companies, will be difficult when coordinating a mix of lead-free and SnPb components.
- Backward incompatibility reliability concerns (such as cold solder joints and not fully reflowed solder joints, mixed metals concerns, etc.) when forced to use lead-free (with >183°C melting temperature) components with SnPb paste.
Based on the above, routes B and C both have a number of transition issues. Which route is better greatly depends upon the product and application.
Lead-free Substitutes
Because of lead-free regulations, flip chip with various solderless bumps (e.g., Au bump, Cu bump, Ni-Au bump and Au-stud bump) have been studied.
These solderless bumps are usually applied with nonconductive adhesives (NCA), isotropic conductive adhesives (ICA), and anisotropic conductive adhesives (ACA).
NCA is not electrically conducted and is a conventional underfill.15 It consists of thermosetting adhesive and nonconductive fillers. In general, these are used for Au-Au diffusion bonding and flip chip on board or substrate. When the NCA cures, it shrinks and brings the bumps on chip, pads and on PCB into a state of compression.
ICA electrically conducts in all directions. Usually, ICA is made of epoxy with Ag-Pd filler particles. It can be applied to the pads of PCBs or substrates by stencil or screen printing. The most commonly used method is dipping the Au, Cu, Ni-Au bumps, or the Au-stud bumps, on the chip into the ICA. A doctor blade in a rotating disk controls the amount of ICA.13
ACA electrically conducts only in the vertical direction, also called Z-axis conductive materials. There are two groups of ACA: anisotropic conductive film (ACF) and anisotropic conductive paste (ACP). ACF consists of thermosetting adhesive, conductive particles (solids or plated plastic spheres) and release film, and looks like a paper. ACP consists of thermosetting adhesive and conductive particles and looks like a paste.
A drawback of this technology is the waste of conductive particles. Concerns with this technology include: no self-alignment; no SMT compatibility; low manufacturing yield; bonding reliability; and low current capacity.
More work needs to be done in metal-bumped flip chip with adhesive technologies to improve the structural geometry, material properties and density of the conductive particles so that higher current capacities can be achieved; to improve the manufacturing throughput and assembly yield; and to obtain reliability data of the bonding assemblies and establish corresponding acceleration models and factors for normal operating conditions.
Conclusion
Global trends of lead-free soldering and technologies have been briefly discussed, with an emphasis on costs, regulations, definitions, designs, materials, process incompatibilities, lead-free substitutes and reliability of lead-free components and solder joints.
Acknowledgement
The first author would like to thank Ted Lancaster and Pete Woodhouse of Agilent Technologies Inc. for their strong support of the Lead-free Program.
References
Please contact the authors for a complete list of references.
JOHN H. LAU, Lead-free technical program manager, may be contacted at Agilent Technologies Inc., 5301 Stevens Creel Blvd., Santa Clara, CA 95052; (408) 553-2358; [email protected]. KATRINA LIU, senior manager, may be contacted at Top Touch Consultants Ltd., A-17C, Block A, YueHai Bldg., Nanyou Rd., Nanshan Shenshen, Guandong, 518054 China.