US 20050093115 A1
A joint structure and method for bonding together two components, such as when attaching an electrical circuit element to a conductor on a substrate. The joint structure comprises a mesh infiltrated by a solder material, in which the mesh is preferably formed of a material having a higher thermal conductivity than the solder material. The joint structure is able to offer improvements in thermal conductivity, electrical conductivity, reflow processing, and stress distribution between the structures it connects. Each of these attributes of the joint structure can be tailored to some degree by the choices of materials for the mesh and the solder material.
1. A semiconductor assembly comprising a heat-generating integrated circuit chip that is attached to a heatsink on a substrate with an electrically-conductive joint structure located between and contacting a surface of the integrated circuit chip and a surface of the heatsink, the joint structure comprising an electrically-conductive flexible mesh infiltrated by a solder material that bonds together the integrated circuit chip, the heat sink and the mesh, a solder-free portion of the mesh extending outside of the joint structure and from between the integrated circuit chip and the heatsink to define a flexible jumper to the integrated circuit chip, the mesh being formed of woven strands of copper or copper alloy wire, the mesh substantially establishing the thickness of the joint structure.
5. The semiconductor assembly according to
6. The semiconductor assembly according to
(1) Field of the Invention
The present invention generally relates to bonding methods, such as when mounting silicon devices and packaged electronic circuit devices and packages to copper, ceramic thick film and other substrate materials, and mounting interconnect straps to silicon power devices as a replacement for conventional wire bonding. More particularly, this invention relates to a method for soldering materials having different coefficients of thermal expansion (CTE) by forming a joint structure that improves stress distribution so as to improve the thermal fatigue resistance of the joint, while also promoting and improving heat transfer through the joint and maintaining or promoting the electrical conductivity of the joint, in part by inhibiting void formation within the joint.
(2) Description of the Related Art
A variety of approaches are known for dissipating heat generated by high power semiconductor surface-mount (SM) devices. For example, heat-generating integrated circuit (IC) chips are often mounted to ceramic substrates, which have higher thermal conductivities than laminate substrates such as printed circuit boards (PCB). Alternatively or in addition, a device may be soldered to a thermal conductor or heatsink on a substrate to increase heat transfer away from the chip, yielding a structure referred to as a “thermal stack.” Because semiconductor devices (e.g., silicon), conductors (e.g., copper) and circuit boards (e.g., ceramics and laminates) have widely varying coefficients of thermal expansions (CTE), the solder joint of a thermal stack is subject to thermal fatigue if the assembly is exposed to extreme temperature cycles, as occurs in many automotive applications. The solder joint may shear or crack due to thermal fatigue, thereby reducing heat transfer from the device and leading to device overheating.
Thin solder joints with minimum voiding are required to achieve adequate heat transfer through thermal stacks used with high power semiconductors. A thin solder joint (e.g., about 0.003 inch, about 80 micrometers) is preferred because of the relatively low thermal conductivity of solders as compared to the other materials typically used in high power semiconductor thermal stacks. For example, the coefficients of thermal conductivity (k) for 60 Sn/40 Pb and 25 Sn/75 Pb solders are about 46 and 38 W/mK, respectively, as compared to copper and silicon with coefficients of about 399 and 83 W/mK, respectively. However, a thin solder joint is more susceptible to thermal fatigue if the materials it joins have widely varying CTE's, such as a silicon power IC and a copper heatsink. If a soft solder is used, a fatigue fracture is most likely to initiate at the outer periphery of the solder joint and propagate toward the center, reducing the area through which heat transfer occurs until the device eventually overheats. In contrast, if a hard solder is used in the thermal stack, a fatigue crack may originate in one of the adjacent materials. For example, in an application in which a silicon chip is soldered to copper, a fatigue crack may initiate in the silicon chip.
As noted above, minimum voiding is also required to achieve adequate heat transfer through the solder joint of a thermal stack. Solder voiding occurs in solder joints in part as a result of the capillary action that draws the surfaces being jointed together. In a thin solder joint, this capillary action can inhibit the ability for flux volatiles and gases to escape the solder joint, with the result that gas bubbles remain trapped in the solder joint. Solder joints containing many voids or a few large voids have significantly reduced heat transfer capability because the voids are barriers to heat flow. Voids can also provide a path for crack propagation through the solder joint. Large voids present at the bottom surface of a die are particularly detrimental to a silicon die.
As a solution to the above problems, solder joints of greater thicknesses have been employed to better distribute the thermally-induced stresses through the thickness of the solder joint, utilizing the ductility of the solder to buffer the CTE mismatch. While doing so has the capability of improving thermal fatigue resistance, there is a point at which the solder thickness is such that the CTE mismatch between the solder and either of the adjacent materials may become a primary source of crack initiation. This thickness is the minimum spacing desired to achieve between materials with a CTE mismatch, and will depend in part on joint area (e.g., die size) and the particular materials present in the thermal stack. In addition to the difficulty of determining this minimum solder thickness, there is the difficulty of producing a thick solder joint to a minimum thickness, such as in the range of about 0.010 to 0.025 inch (about 250 to 635 micrometers). Furthermore, and as noted above, increasing the thickness of the solder joint has the effect of increasing the thermal and electrical resistances of the circuit because of the relatively low thermal and electrical conductivity of solders. Finally, it is thought that thicker solder joints may increase the probability of voiding.
In view of the above, it would be desirable if the thermal stresses of a thermal stack could be reduced without incurring the performance and processing shortcomings associated with the use of thicker solder joints.
The present invention is directed to a joint structure and bonding method, such as when attaching an electrical circuit element to a substrate or a conductor on a substrate. The joint structure comprises a mesh infiltrated by a solder material. The mesh is preferably formed of a material having a higher thermal conductivity than the solder material. In a preferred embodiment, the mesh is formed of woven strands formed of copper or a copper alloy, such that the mesh is able to significantly improve the thermal conductivity of the joint structure beyond that possible with conventional solder materials.
In view of the above, several benefits can be realized with the present invention. One important benefit is that the mesh can be used to positively establish the thickness of the joint structure, such that the thickness of the joint structure can be tailored to improve the thermal fatigue resistance of the assembly by better distributing thermal stresses arising from a CTE mismatch between the solder and the materials of the circuit element and substrate. However, the joint structure of the present invention does not suffer from the shortcomings associated with thick solder joints of the prior art. For example, the improved fatigue resistance made possible with the greater thicknesses of the joint structure is not achieved at the expense of thermal conductivity, because the mesh fills the solder joint structure and acts as a composite material. If formed of copper or another highly thermally conductive material, the mesh is able to improve the thermal and electrical conductivity of the joint structure beyond that possible with a thin (e.g., 0.003 inch (80 micrometers)) unfilled solder joint. Furthermore, the mesh appears to further improve the resistance of the joint structure to fatigue cracking because it inhibits the creation of a thin shear plane within the joint structure. The individual columns of woven strands in the mesh are able to expand and contract independently of each other, thereby minimizing the total effect of any CTE mismatch between the mesh (e.g., copper) and the adjacent materials (e.g., silicon). Finally, the mesh provides multiple paths by which flux and other gases can escape the joint structure, such that the formation of relatively large voids in the joint structure is reduced. Any large voids (e.g., contaminants that are not able to escape) are broken up by the mesh and reduced to smaller voids that are distributed and confined to openings in the mesh, so that uniform contact is obtained between the joint structure and the elements it is bonding (e.g., a die) to provide good thermal and electrical conduction.
Other objects and advantages of this invention will be better appreciated from the following detailed description.
The mesh 14 is shown in
The individual columns of wire in the mesh 14 are believed to act substantially independently of each other, and therefore inhibit the creation of a thin shear plane through the joint structure 12 that would promote fracturing of the joint structure 12 from thermal fatigue. As seen from
A suitable method of forming the joint structure 12 shown in
On cooling, the solidified solder 16 has a structure similar to many individual columns of solder through the openings in the mesh 14. The solder columns are connected to each other by fillets at the surfaces of the device 18 and conductor 22, but are otherwise generally independent of each other such that the solder 16 is able to further improve the thermal cycle life of the joint structure 12. In addition, solder voiding that would result from the gases formed during reflow, including those produced when the flux melts, is controlled and minimized because the mesh 14 prevents the joint structure 12 from collapsing and trapping the gases during reflow, and provides numerous routs for the gases to escape.
Stacked capacitors of the type represented by the component 78 are sometimes desirable when limited board space is available. In conventional practice, the opposing contact rails of each stacked capacitor assembly are bonded with a high temperature solder to a pair of stainless steel strain-relieved terminals. The terminals are typically thin to provide strain relief for the CTE mismatch between the dielectrics 76 (e.g., ceramic) and the substrate 80 (e.g., epoxy laminate), but have limited thermal and electrically conductivity as a result of being formed of stainless steel. The terminals are also known to promote void formation in the solder joining the terminals to the rails and conductors. The joint structures 72 and 73 of this invention overcome the disadvantages of the prior art because the mesh can be formed of copper or another suitable material capable of improving the electrical and thermal conductivity of the structures 72 and 73, as well as better distribute shear stresses between the contact rails 74 and the substrate 80, and provide a path for gases to escape from the structures 72 and 73 during soldering.
With each of the applications described above, joint structures formed in accordance with this invention are able to offer improvements in thermal conductivity, electrical conductivity, reflow processing, and stress distribution. These attributes of the joint structures can be tailored to some degree by the choices of materials for the mesh and the solder. For example, some applications may require greater emphasis on thermal and electrical conductivity as compared to stress distribution. Still other applications may impose a limitation on the thickness of the joint structure. Therefore, while the invention has been described in terms of a preferred embodiment, it is apparent that other forms could be adopted by one skilled in the art. Accordingly, the scope of the invention is to be limited only by the following claims.