Issue



Bonding Wire: Is Scalability the Wave of the Future?


07/20/2012







DOMINIK STEPHAN, RED Micro Wire Pte., Ltd., Singapore


A new technology enables bonding wire to be cast instead of drawn. The micro wire can be used like traditional copper bonding wire but offers several advantages that make it more cost-effective.


Miniaturization and scalability continue to be major trends in the semiconductor industry, and they necessitate that the entire manufacturing infrastructure adapt and evolve in order to grow. In order to facilitate such growth, and greater miniaturization, it is important that the entire industry works together to attain new development targets, including those of wire bonding.


Miniaturization of bonding is critically correlated with the number of bond pads that must be placed per unit area. In the case of ball bonding, each bond pad must fully accommodate the ball bond within the bond pad, without encroaching onto adjacent bond pads. The ball/pad reduction factor applied in the industry is typically between 70-80%.


Next in line of correlation is the FAB (free air ball) during the process of bonded ball formation. Typically, the FAB is about 10-20% smaller than the bonded ball, due to the squashing of the latter during bonding. At the moment, the limiting factor for this FAB diameter is the wire diameter, assuming a typical BSR (the ratio between the FAB diameter over the wire diameter) is about 1.6-2.0. If we assume the smallest wires currently in mass production are about 0.6mil (15??m), this means that the FAB is about 24-30??m, bonded ball about 29-33??m, bond pad about 33-36??m, and bond pad pitch, at best, is 35-38??m.



Table 1 shows the correlation between process stage/bond pad pitch (BPP) and required bonding wire diameter to facilitate FAB formation.


The International Technology Roadmap for Semiconductors (ITRS) suggests the bond pitch should be down to 20??m in 2013. This will require a large leap in wire manufacturing know-how. Back-calculating indicates the need for wire of 8-10??m in diameter. All major wire bonding plants today are running 0.7mil copper wire (18??m) and have plans to qualify 0.6mil (15??m) within the next year or two. One of the critical factors for not seeking more aggressive targets is the lack of availability of a bonding wire of lower diameter and an infrastructure that supports such a process, including bonding machines and capillaries.





Figure 1. Fusing current (based on calculation) for wires of different wire diameter and testing length.



Current standard materials and tools make it possible to scale down to a certain level, but attempting to go beyond that leads to the figurative brick wall. Capillary manufacturers are producing caps for 0.6mil wire in mass production having a hole diameter of about 21??m. There have been attempts to go as low as 15??m, accommodating 0.5mil wires, mainly for Au wire bonding. For copper wire bonding designs, the sizes to date have been typically slightly larger.


Wire bonding infrastructure is so extensive that no other chip-interconnection technology can displace wire bonding in the foreseeable future, although other technologies, particularly flip chip, will experience increasing utilization. Increasing miniaturization of electronic circuits has put relentless pressure on wire bonding technology to (1) increase yields (<5 ppm defects); (2) decrease pitch (<30??m for ball bonds) and (3) provide the lowest possible and ever decreasing cost.


Creating smaller wires ??? the challenges


Wire makers, wire drawing equipment manufacturers and tool makers are all approaching a boundary with existing technologies, although there are efforts to make smaller wires. When wire diameter is successfully reduced, more issues correlated with wire uniformity and production control of the low tension required during the drawing process appear. These may result in lower yield and higher cost. Another limiting point is the drawing die, which can experience high erosion and cause lower yields at very small diameters. Furthermore, once the wire is drawn to the final stage, it has dramatically "work-hardened" and must be recrystallized by an annealing process. This is typically done at about 30-60% of a metal's melting point at which, in return, it loses significant strength. This introduces further breaks, lower yield and higher cost. In order not to break the wire or stretch it (which would weaken its mechanical performance), annealing tension must be controlled at <0.5g. This is much less than the tension control capability of the typical dancer arms and pulleys supplying the tension. Here new technologies would be needed to facilitate wire drawing and annealing processes for cost effectiveness.


Even if the wire can be successfully drawn, the properties of the wire itself can be a challenge. As diameter decreases, strength decreases as a square function (0.5 mil is 25% the strength of 1 mil).


Assuming an average copper wire, having a tensile strength of about 200MPa, the force needed to break the wire is only about 10g, making a 0.5mil wire only about 2.5g. Such a low material strength is very difficult to manually handle, and poses issues to testing strength during the application. The stiffness is a fourth power (0.5 mil is 1/16 the stiffness of 1 mil ??? stiffness = deflection under a load.) This poses many issues to the handling and the application of the wire. In general, elastic modulus is not the same as stiffness. Elastic modulus is a property of the constituent material; stiffness is a property of a structure.


Another inherent issue with lower wire diameters is the electrical performance. Naturally, the current carrying capacity is lower with lower diameter. This needs to be taken into consideration by packaging designers.


Recently there has been a new wire introduced that encases traditional copper wire in glass coating. Following, we'll take a look at how that wire scales versus traditional copper wire.


If wire properties are not sufficient, the wire will not hold its shape and will sag under its own weight. Here again a stiffer solution might help. Thinking outside the box, the ability to ignore the issue of adjacent wire shorting would ease this pain. However, increased strength is usually correlated to increased hardness, which is detrimental to the bond. The first bond poses a larger risk of cratering or bond pad deformation. The second bond poses the risk of lower strength values based on lower deformation and, respectively, a lower bond area. However, in glass coated wire, some of the strength comes from the glass layer, providing a geometric support. The hardness test shows that the copper core is even softer then a typically highly annealed copper wire, as seen in Table 3.



Table 2 is a comparison of breakload values (grams) for soft, normal and glass coated wire assuming 200, 300 and 400Mpa respectively (RMW wire is referred to as composite, because the glass is a structural member of the wire).



Table 3 wire hardness values of selected technologically available materials (*refers to the copper core).


In normal bonding wire, we need high elongation in copper to ensure sufficient ductility to make a strong second bond. However, in the case of glass coated wire, the EL/BL of the wire includes the glass, but the glass is not part of the first or second bond. So the wire EL will not be high, however the copper core is still very ductile. From the tensile test chart in Fig. 2, one can see that the glass plays an initial part of providing strength and limiting ductility. After further elongation, the glass will eventually crack and give way to the extension of the copper.





Figure 2. Tensile test chart of a RED Micro Wire glass-coated wire.



Using smaller diameter wires


Let's look at the next step: How wires are used in the wire bonder. Users are having difficulty threading the wires and only highly experienced users are able to thread a 0.6mil wire though the wire path and into the capillary, without breaking it.


Further reduction in wire diameter will decrease visibility and, more dramatically, decrease the stiffness of the wire, which makes threading more challenging. A wire that is in its geometry stiffer than usual wire would make this process easier.


The next issue is related to the loop formation. A2 wire is supposed to keep its loop shape, which was imparted from the bonder trajectory during the loop formation (which is, most of the time, highly advanced with various forward and backwards motions).


The next step in the process is typically the molding process, where the wires are heavily exposed to mechanical stress/sweep. Mold compound viscosity and melt front velocities require scrutiny.


Another consideration on the infrastructure is the use of a wire bonding capillary as a tool for the bonding machine. Typically made from ceramic, they are currently only available in mass production, down to a diameter of about 15??m. They are getting increasingly difficult to make, but since there was no wire to drive the dimension, not enough effort has been applied to open these boundaries.


Glass-insulated wire ??? a solution to the problem


It is clear that there is a need for smaller wire diameter to support advances in miniaturization. There are, however, intrinsic challenges of physics that might limit the ability to manufacture and use such wire. Most of the issues are related the mechanical strength, and concerns about wire stability and shorting.


It seems natural to think about an insulated wire eliminating worry about shorting. Glass coated wire provides such an insulation and slightly inaccurate looping can be accommodated.


The issue of strength (or stiffness in this case) can also be addressed with the glass-coated wire. Glass that has an inherently higher strength compared to copper acts not only as a surface layer, but as an active element providing mechanical support to the wire. This leads to much higher strength and stiffness values compared to a bare copper wire (or any kind of coated wires, be it conductive or insulating).


Looking at the wire manufacturing process from a simplistic view, one could ask: How logical is it to cast the wire at a very large diameter, just to draw it down to a small diameter?


A solution whereby wire can be manufactured directly out from the melt covered by a glass coating is currently being tested and optimized for bonding wire applications.


Glass can greatly increase wire strength and stiffness, yet still provide a smaller, effective wire diameter on smaller bonds. Based on the manufacturing method, a full coverage of glass can be ensured, which in return ensures insulation. In addition, floor and shelf life are no longer limiting factors since there is no exposure of copper.


Acknowledgement


The author would like to thank Dr. Jeffrey Seuntjens and Mr. Steven Creswick for their technical feedback, discussion and review, as well as the industry partners SPT, ASM, ITE and TPT for their experimental support.


DOMINIK STEPHAN is the Director for Application and Product Marketing forRED Micro Wire (RMW) in Singapore.


Solid State Technology, Volume 55, Issue 6, July 2012


More Solid State Technology Current Issue Articles

More Solid State Technology Archives Issue Articles