US20090255660A1

US20090255660A1 - High Thermal Conductivity Heat Sinks With Z-Axis Inserts

Info

Publication number: US20090255660A1
Application number: US12/421,494
Authority: US
Inventors: James A. Cornie; Himanshu Pokharna
Original assignee: Metal Matrix Cast Composites LLC
Current assignee: Metal Matrix Cast Composites LLC
Priority date: 2008-04-10
Filing date: 2009-04-09
Publication date: 2009-10-15

Abstract

A heat sink including a composite material base plate that defines at least one through hole. At least one composite material insert is position into the at least one through hole prior to pressure infiltration. The composite material insert is oriented to increase thermal conductivity in the through-plan direction.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

The present application is a non-provisional application of copending U.S. Provisional Patent Application Ser. No. 61/044,030, filed on Apr. 10, 2008. The entire contents U.S. Patent Application Ser. No. 61/044,030 is herein incorporated by reference.

FEDERAL RESEARCH STATEMENT

This teaching was made with Government support under National Science Foundation Grant Number 0638035. The Government has certain rights in this teaching.
The section headings used herein are for organizational purposes only and should not be construed as limiting the subject matter described in the present application.

INTRODUCTION

Modern optical and electronic devices and systems, such as lasers, LEDs, imaging systems, cellular phones, radar systems, high power RF devices, and high power microwave devices have ever increasing power requirements and operating speeds. In addition, modern electronic devices are continually increase semiconductor die sizes and device densities in order to provide more functions and higher performance in smaller system dimensions.
Modern electronic devices must dissipate large amounts of heat during normal operation. For example, wide band gap semiconductors, such as GaN and SiC operate at relatively high temperatures and can generate heat energy densities that are greater than 100 W/cm². Such devices generally require heat spreader/heat sinks to dissipate the heat energy. It is expected that the heat generated by future electronic devices will continue to increase.
Electronic and optical devices can be directly attached to a heat spreader/heat sink. The coefficient of thermal expansion (CTE) of the optical and electronic devices and the CTE of the heat spreader/heat sink are usually matched as closely as possible to avoid thermal cycling induced mechanical stress failures. Thermal cycling arises during power up and power down cycles in combination with resistive heating caused by current flowing in the device. Electronic and optical devices can also be encased in a ceramic package that protects the device and provides electrical connections and optical windows. Common ceramic packages include alumina, aluminum nitride, beryllium oxide, and silicon.
In addition, many other industries require materials that match their coefficient of thermal expansion to other materials. Some of these materials must also be lightweight, stiff, and capable of damping undesirable vibrations. For example, materials used for precise motion control often must have a particular CTE. Also, some materials used in the optics industry for mirrors, optical benches, metering devices, as well as other kinds of mechanical hardware, must have a particular CTE.

BRIEF DESCRIPTION OF THE DRAWINGS

The aspects of this teaching may be better understood by referring to the following description in conjunction with the accompanying drawings. Identical or similar elements in these figures may be designated by the same reference numerals. Detailed description about these similar elements may not be repeated. The drawings are not necessarily to scale. The skilled artisan will understand that the drawings, described below, are for illustration purposes only. The drawings are not intended to limit the scope of the present teachings in any way.

FIG. 1 illustrates calibration data for AlGr_p™ material that present thermal conductivity (TC) data and coefficient of thermal expansion (CTE) data of natural flake graphite reinforced A1413 as a function of volume fraction of flake graphite.

FIG. 2 is a diagram that illustrates the fabrication of a high thermal conductivity heat sink according to one embodiment of the present teaching that can have nearly radial heat spreading and a uniform thermal expansion coefficient.

FIG. 3 shows steps performed to machine and insert AlGr_p™ z-axis preform into an AlGr_p™ preform base plate to provide high thermal conductivity heat sinking and orthogonal spreading from the inserts according to the present teaching.

FIG. 4A shows a completed z-axis insert preform assembly.

FIG. 4B illustrates post pressure infiltration casting base plates before plating.

FIG. 4C illustrates post pressure infiltration casting base plates after Ni and Au plating.

FIG. 5A shows pressure infiltration casting base plates plated with a Ni plated base layer and then a top gold plated layer as shown in FIG. 4C being inserted into a finished fin cooled heat sink.

FIG. 5B shows the fin array soldered on the casting base plate.

FIG. 6 illustrates a diagram of a cutting scheme for three sets of orthogonal mitered preforms with z-axis inserts according to the present teaching.

FIG. 7 illustrates diagrams of hexagonal cross-section z-axis insets according to the present teaching that provide relatively uniform coefficients of thermal expansion and that can be matched to the device being cooled.

FIG. 8 shows a diagram of one embodiment of a geometry according to the present teaching for z-axis inserts into AlGr_p™ or another composite and a metallic base plate material.

FIG. 9 shows a diagram that illustrates the use of a skin of composite material to minimize local variations in the coefficient of thermal expansion on heat sink materials with z-axis inserts according to the present teaching.

FIG. 10A shows two base plates with z inserts mitered into an AlGr_p™ base plate.

FIG. 10B shows a diagram of the base plate shown in FIG. 10A with z inserts in a configuration that was used to experimentally determine the CTE above the pads.

DETAILED DESCRIPTION

Reference in the specification to “one embodiment” or “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the teaching. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment.
While the present teachings are described in conjunction with various embodiments and examples, it is not intended that the present teachings be limited to such embodiments. On the contrary, the present teachings encompass various alternatives, modifications and equivalents, as will be appreciated by those of skill in the art. In particular, while some aspects of the present teaching are described in connection with a particular aluminum graphite material, it should be understood that the present teaching can be used with an almost unlimited number of different composite materials.
It should be understood that the individual steps of the methods of the present teaching may be performed in any order and/or simultaneously as long as the teaching remains operable. Furthermore, it should be understood that the apparatus and methods of the present teaching can include any number or all of the described embodiments as long as the teaching remains operable.
Many known heat sinks are commonly fabricated from metals, such as copper, molybdenum, tungsten and aluminum. A metal heat sink is often plated with nickel and gold prior to attachment to a ceramic package at an elevated temperature. Alternatively, silver-filled adhesives, or other conductive metal powder-filled adhesives, are sometimes used for bonding.
Choosing a metal or other material for a heat sink often involves an engineering trade-off between desirable and undesirable properties. Some metals, such as aluminum and copper have high thermal conductivity, but have coefficient of thermal expansion values (CTEs) that are several times greater than that of the ceramic package or semiconductor die. During power cycling of an electronic component attached to a heat sink, the temperature of the component and the attached heat sink fluctuate significantly. Consequently, the metal heat sinks cause mechanical stress to the heat sink bonding material during power cycling. The differential expansion of the heat sink relative to the ceramic package or semiconductor die can cause failure of the bond material or cracking of the package or die.
Other metals, such as tungsten and molybdenum, have relatively small coefficient of thermal expansion values. Although such metals can permit a reliable bond, they have lower thermal conductivity than aluminum or copper substrates and they are difficult to electroplate. Furthermore, tungsten and molybdenum are undesirable for applications that require relatively light weight.
Composites of copper and tungsten or of copper and molybdenum have certain advantages over elemental materials. These composites can be made by various methods of powder metallurgy, such as, for example, infiltrating copper into a sintered body of tungsten or molybdenum, or sintering a mixed powder of the two metals. However, sintered ingots of tungsten and molybdenum are difficult to roll into elongated plates. Alternatively, metal layers can be joined by cladding or lamination. Cladded and laminated products, however, require precise machining, which is labor-intensive, error-prone, and expensive.
Some heat sinks combine a sintered ceramic with a metal matrix. The fabrication process involves the formation of a ceramic preform, which can be made by, for example, sintering silicon carbide powder. The ceramic preform microstructure typically has a predetermined void volume fraction that is subsequently filled with a molten metal, which is typically aluminum. The thermal conductivity of aluminum ceramic heat sink can be improved by using copper-based inserts. Such heat sinks, however, can be difficult to manufacture and can have a relatively narrow range of possible coefficient of thermal expansion values.
Other heat sinks are formed of metal matrix composites that include infiltrated inorganic fiber material. Infiltration of fibers is sometimes difficult because of problems with fiber wetting and non-uniform fiber distribution. In addition, molten metal infiltration of fibers under pressure can displace the fibers due to the fiber breakthrough pressure threshold. Furthermore, it is often difficult to control fiber volume fraction, and thus difficult to obtain desired properties of the composite. These factors have limited the use of metal matrix fiber composites as heat sinks.
Metal matrix composite (“MMC”) materials that include discontinuous high-modulus graphite fibers that are randomly arranged in-plane at desired volume fractions have significant advantages over many known material used for heat spreaders and heat sinks. Such composites are disclosed in U.S. patent application Ser. No. 10/379,044, filed Mar. 4, 2003, entitled “Discontinuous Carbon Fiber Reinforced Metal Matrix Composite,” which is assigned to the present assignee. The entire application of U.S. patent application Ser. No. 11/163,486 is incorporated herein by reference. One advantage is that MMC materials can be used to fabricate heat sink base plates with relatively high thermal conductivity and with coefficient of thermal expansion values that match the coefficient of thermal expansion values of common ceramic package materials.
One method of manufacturing a MMC with randomly distributed graphite fibers is disclosed in U.S. patent application Ser. No. 10/379,044. This method includes mixing dry PEG binder material with dry-milled graphite fibers having average length of about 300 microns. The mix is then poured into a mold, pressed, and heated to liquefy the binder. The mix is then chilled to set the binder prior to removal from the die.
The resulting preform is inserted into a pressure infiltration casting mold vessel for metal infiltration and solidification. This process is relatively simple and inexpensive. The fiber distribution obtained with this process is relatively non-uniform and may results in a standard deviation on order of 2 ppm at a volume fraction that results in a coefficient of thermal expansion value of 7 ppm/K. Thus, this method may not be suitable for applications requiring particularly close coefficient of thermal expansion value matching to a material. In addition, non-uniform and largely unpredictable fiber distribution may result in warping of plates machined from the casting while processing through the various machining steps or through soldering operations.
Another method of manufacturing a MMC with randomly distributed graphite fibers is disclosed in U.S. Pat. No. 5,437,921. This method includes dispersing milled fibers in an aqueous slurry, which is then poured into a filter vessel. The aqueous slurry is formed into a filter cake under vacuum and then pressed to a desired volume fraction. The filter cake is then dried and pressed to the desired volume fraction. This process, and other processes that use milled fibers, tend to develop preferred fiber orientations when the fibers experience flow alignment. Flow alignment occurs when dry milled fibers are poured into a mold where they exhibit aligned flow that cannot be re-randomized. Also, this process, and other processes that use milled fibers, is prone to forming localized non-uniform distributions due to localized flow alignment of milled fibers during pouring and vacuum filtration steps. Other problems with this process include a variation of packing density with thickness and significant variations of coefficient of thermal expansion values. For example, the standard deviation can be 1.25 ppm/K at the average level of coefficient of thermal expansion (7 ppm/K).
Another known method of manufacturing a MMC with randomly distributed graphite fibers includes incorporating chopped CKD graphite fibers with an average chop length of 25 mm into a paper product. This method is used commercially by Technical Fibre Products of Cumbria in the United Kingdom. The method requires adding a co-polyester fiber, which serves as a binder. The paper product is laid out into a die and then heated to soften the binder fiber. The preform is then pressed to the desired volume fraction. Each ply is rotated through a sequence of orientations to produce a substantially planar isotropic preform. This process results in a lower standard deviation that is about 0.9 ppm/K. However, the process is relatively expensive and the through-plane thermal conductivity is relatively low. In addition, the polyester binder is typically difficult to remove and has relatively a high char yield during the outgasing and preheating operation.
There currently is a significant need in the electronic thermal management and packaging industry to fabricate MMC base plates with high thermal conductivity in various directions in order to achieve both heat spreading and through-plane thermal conductivity. Heat spreading requires high in-plane thermal conductivity. Heat sinking requires high z-axis or through-plane thermal conductivity. The term “in-plane,” as used herein, refers to the plane parallel to a bonded surface of a heat sink. The term “through-plane,” as used herein, refers to a direction that is orthogonal to an in-plane surface. The “through-plane” direction is also referred to herein as the z-axis.
U.S. patent application Ser. No. 11/306,343, entitled “Hybrid Metal Matrix Composite Packages with High Thermal Conductivity Inserts,” describes the use of composite material structures that provides both high through-plane thermal conductivity and high structural strength and that can be engineered to CTE match common ceramic package materials. U.S. patent application Ser. No. 11/306,343 is assigned to the assignee of the present application. This patent application describes the use of certain materials, such as highly-oriented pyrolytic graphite (“HOPG”) materials in foil or sheet form, that can be machined with cavities for rivets and rivet-vias which, after infiltration with molten Al or Cu alloys, serve to produce a cladding material with cast-in rivets that add significant strength and robustness to a package assembly.
Such rivet-vias can be used to enhance through-plane thermal conductivity in order to improve the transfer of heat into the highly conductive graphite planes. These rivet-vias can also be used to modify local expansion coefficients to improve the CTE match of an electronic package or semiconductor device. In addition, these rivet-vias can also be used to both enhance through-plane conductivity and to modify local expansion coefficients to improve the CTE matching.
One problem with many prior art z-axis insert heat sinks is that they are expensive to manufacture. Another problem with many prior art z-axis insert heat sinks is that they have relatively low through thickness thermal conductivity. For example, the through thickness thermal conductivity of standard aluminum-graphite mixtures material in the “z” direction, is on order of 30 to 40 W/mK. This makes normally oriented aluminum-graphite inadequate for heat sinking in the Z direction.
Another material called Al MetGraf™ has also been used for z-axis insert heat sinks. The Al MetGraf™ material has only moderate through thickness thermal conductivity, which is on order of about 120 to 140 W/mK. The through thickness thermal conductivity depends upon the volume fraction of milled graphite fiber present. While this material has a moderate ability to sink heat, it is insufficient for many demanding applications.
The term “MetGraf™” is a trade name for a metal matrix composite material commercially available from Metal Matrix Cast Composites of Waltham, Mass., the assignee of the present application. The Al MetGraf™ material is a metal matrix composite material that includes a matrix of random discontinuous high-modulus graphite fibers that is pressure infiltrated with molten Al. The Al MetGraf™ preform material can have a volume fraction that is chosen to provide precise CTE matching to a particular material. For example, MetGraf™ 4 and 7 refer to discontinuously reinforced Al and Cu alloys in which the matrix alloy is pressure infiltrated into a compressed preform to produce an in-plane CTE of 4 or 7. Al MetGraf™preforms can be produced using one of the processes described in U.S. patent application Ser. No. 11/163,486.
In one embodiment, z-axis inserts according to the present teaching are made from oriented aluminum-graphite material, which is referred to herein as AlGr_p™ material and is trademarked by Metal Matrix Cast Composites of Waltham, Mass., the assignee of the present application. Aluminum-graphite material preforms are made of aluminum-graphite that is rotated 90 degrees to form thermal vias during preform fabrication. The AlGr_p™ preforms can also be mitered in the orthogonal direction or in the hexagonal direction to improve the thermal transport properties of the insert. The rotation of the AlGr_p™ preform stock 90 degrees and orthogonal or hexagonal mitering creates an extremely high thermal conductivity base plate that is thermally isotropic. Such properties are highly desirable if they can be achieved at relatively low costs. In some embodiment of the present teaching, the “z” inserts are assembled to direct heat downwards and radially into a heat sink baseplate. In some embodiments, the base plate can be formed of Metgraf™ or AlGr_p™ material. In other embodiments, any type of base plate material can be used.
Thus, z-axis inserts according to the present teaching have outstanding thermal transport properties compared with known z-axis inserts. In addition, the AlGr_p™ material is much less expensive than many other heat sink materials because natural flake graphite can be used to make the aluminum-graphite material. Natural flake graphite is very inexpensive, especially compared with pyrolytic graphite. Such preforms are a very inexpensive way to insert high thermal conductivity materials without creating a high thermal impedance interface from any post fabrication soldering procedures.
However, one skilled in the art will appreciate that highly oriented pyrolytic graphite (HOPG) products, such as APG™ and TPG™, can also be used to construct z-axis inserts according to the present teaching. Such z-axis inserts can be oriented and mitered in the orthogonal direction or in the hexagonal direction to improve the thermal transport properties of the insert to achieve a 1500 W/mK sinking path that is orthogonally or quasi-radially dispersed into a base plate material.
Although the examples presented herein use either AlGr_p™ or MetGraf™ materials prior to pressure co-infiltration, one skilled in the art will appreciate that many other types of materials can be used. For example, particulate reinforced composites, such as SiCp reinforced Al alloys can be used to manufacture the z-axis inserts according to the present teaching.
The AlGr_p™ inserts according to the present teaching can be manufactured according to processes described in U.S. patent application Ser. No. 11/900,727, filed Sep. 13, 2007, and entitled “Thermally Conductive Graphite Reinforced Alloys.” These processes results in a preform composed of natural graphite flakes arranged in a planar-isotropic array with de-booking/micro-gating agents, such as fine grained silica or alumina inserted between the flakes. The de-booking/micro-gating agents are useful to assure complete access to all flake surfaces by the infiltrating Al Si alloy. For example, U.S. patent application Ser. No. 11/900,727 describes a process where micro-gating and de-booking agents, such as colloidal silica, and/or fine grained alumina, are impacted between individual crystal (flakes) of natural graphite to permit complete infiltration and contact with the matrix alloy.
In the AlGr_p™ preform process according to the present teaching, multiple layers of preforms are formed with a binder solution which can be a thermal plastic material, such as a wax like polyethylene glycol (PEG). These thermal plastic materials soften and melt during heating. The binder solution can also be an organic binder, such as PVA (polyvinyl acetate), PMMA (polymethylmethacrylate), or polystyrene. The preforms with binder are formed into mats that can be pressed above the melting point of the binder and then chilled to form a rigid preform of a desired volume fraction. The mats are then vacuum out-gassed to remove the binder. Pressure infiltration is then performed with an Al-Si alloy. The result is a metal matrix composite that has high thermal conductivity and low thermal expansion. Although the examples are given for a material that is subsequently infiltrated with an Al alloy, similar approaches can be made for copper and magnesium matrix composites.
FIG. 1 illustrates calibration data 100 for AlGr_p™ material that present thermal conductivity (TC) data and coefficient of thermal expansion (CTE) data of natural flake graphite reinforced A1413 as a function of volume fraction of flake graphite. The calibration data 100 indicate that the thermal conductivity and thermal expansion of the natural flake graphite reinforced A1413 composite material can be precisely engineered simply by controlling the volume fraction natural graphite. During fabrication, the volume fraction of natural graphite is controlled in the preform stage by the compression that is applied to the preform assembly prior to thermally setting of the binder material.
The data 100 indicate that the through thickness thermal conductivity (or z-axis thermal conductivity) of AlGr_p™ composites is on the order of 30 to 40 W/mK, which is relatively low as noted above. However, materials, such as AlGr_p™ composite materials, are excellent heat spreader. Separate high thermal conductivity vias would be required to make such materials useful as heat sinks. While it may be possible to solder copper or other high conductivity inserts into these materials after processing, such added steps would be expensive and there would be poor thermal expansion matching between these material, the inserts, and the device attached to the heat sink.
The data 100 also indicate that the AlGr_p™ material has in-plane thermal conductivity that is approximately 640 to 800 W/mK over the useful CTE matching range for most device packages. One aspect of the present teaching is the realization that the AlGr_p™ material can be rotated 90 degrees and used as an insert into a preform in order to achieve a 700-to 800 W/mK heat sinking. Such a preform can also have good local heat spreading depending on the volume fraction of the natural graphite in the preform as well as the intrinsic high thermal conductivity of the AlGr_p™ or MetGraf™ base plate. Such a heat sink would result in an isotropic heat sink with 700-800 W/mK heat sinking from selective hot spots as well as high thermal conductivity spreading with the same thermal conductivity of the particular AlGr_p™ or MetGraf™ base plate material.
Thus, high conductivity inserts can be fabricated according to the present teaching by first fabricating materials, such as AlGr_p™ through the binder stage where the set PEG, PMA, PMMA, or polystyrene binder produces a block of material. Then the blocks of material are machined, sawed, or extruded into shapes that are used as z-axis inserts. The z-axis inserts are then reinserted into the heat sink with controlled orientation so that they provide high through-plane as well as high in-plane thermal conductivity. The manufactured preforms can be AlGr_p™ or MetGraf™ preforms that are engineered with very high conductivity and controlled thermal expansion after subsequent metal infiltration in selected areas. Numerous other types of performs can also be used.
Is some embodiment, the insert material is taken from the same material as the material used for the heat sink. When the insert material is the same material as the heat sink material or is formed in a substantially identically process as the heat sink material, the coefficient of thermal expansion of the insert material is identical or nearly identical to the coefficient of thermal expansion of the heat sink material. Such heat sinks can have nearly radial heat spreading with high thermal conductivity and a uniform thermal expansion. Such heat sinks are particularly useful for heat sinking fragile semiconductor devices, such as GaAs and GaN devices.
FIG. 2 is a diagram 200 that illustrates the fabrication of a high thermal conductivity heat sink according to one embodiment of the present teaching that can have nearly radial heat spreading and a uniform thermal expansion coefficient. In most applications, it is desirable to generate a map of high thermal output mounting sites on the heat sink. Typically, these high thermal output mounting sites are areas where devices that generate large amount of heat are located. Insert sites or vias can then be machined into a preform at these high thermal output mounting sites while the preform is being fabricated. Any type of composite preform can be used. For example, the composite preform can be an AlGr_p™ or a Metgraf™ preform. One skilled in the art will appreciate there are an almost unlimited number of other possible preforms.
The machined AlGr_p™ preforms are rotated 90 degree and then re-inserted into the machined insert or via sites prior to infiltration. The inserts shown in FIG. 2 can be inexpensively inserted into a preform prior to pressure infiltration casting. The assembly is then pressure infiltrated with molten Al-Si alloy and then solidified. During pressure infiltration, the inserts as well as the remainder of the preform are co-infiltrated. The co-infiltration reduces or eliminating high impedance interfaces associated with plating and soldering insert post castings.
The casting is then removed from the mold after cooling. After cooling, the casting is sliced into sections. Numerous slices can be obtained from a single ingot, resulting in multiple parts with z-axis inserts for each casting. The resulting sliced castings are then machined for mounting bases. The mounting bases can be surface lapped to reduce surface roughness. In some embodiments, the mounting base is then Ni plated for bonding.
Alternatively, the preforms can be pre-sliced and placed into a net-shape mold which provides an engineer with a skin and CTE matching pad over the z-axis inserts. Such a CTE matching pad would reduce local inhomogenities in the coefficient of thermal expansion due to the machining process. Post casting processing cost should be identical to post casting processing cost for preforms with non-z-axis inserted parts, and should be substantially less than components with high thermal conductivity vias formed by various prior art types of post casting/fabrication processes that are currently used in the industry.
The performance of these z-axis inserted materials compared with prior art materials is excellent. For example, heat sinks fabricated with AlGr_p™ preforms and AlGr_p™ z-axis inserts can have a thermal conductivity equal to about 700-800 W/mK after infiltration with Al. Very few state-of-the art z-axis insert materials are capable of achieving a thermal conductivity equal to almost 800 W/mK. In addition, such heat sinks can provide near radial heat spreading and substantially constant thermal expansion. The heat spreading can be about 640-760 W/mK depending upon the design of the overall coefficient of thermal expansion of the base plate.
FIG. 3 shows diagrams 300 illustrating steps performed to machine and insert AlGr_p™ z-axis preform into an AlGr_p™ preform base plate to provide high thermal conductivity heat sinking and orthogonal heat spreading from the inserts according to the present teaching. The z-axis insert is mitered by pre-machining the preform and then reassembling the parts. In this particular embodiment, a square z-axis insert is inserted into a square via so that the perform base plate has high thermal conductivity in the z-axis direction and also high thermal conductivity in one of the two in-plane directions.
The diagrams 300 in FIG. 3 illustrate that if the electronic package is mounted precisely on the junction of the four wedge shaped parts that make up the mitered preform z-axis insert after infiltration with the matrix alloy, the package would be exposed to an average coefficient of thermal expansion that is equivalent to the coefficient of thermal expansion of the body material.
FIG. 4A shows a completed z-axis insert preform assembly 400. The z-axis inserts were fabricated and then re-inserted into the preform as shown in FIG. 3. FIG. 4B illustrates post pressure infiltration casting base plates 420 before plating. In particular, FIG. 4B shows the as-pressure infiltration cast parts. The post pressure infiltration casting base plates 420 were fabricated according to the steps described in connection with FIG. 3. FIG. 4C illustrates post pressure infiltration casting base plates 440 after Ni and Au plating. FIG. 4C shows the same part as FIG. 4B with a nickel plated base layer and then a top gold plated layer.
FIG. 5A shows the pressure infiltration casting base plates 500 plated with the nickel plated base layer and then plated with the top gold plated layer as shown in FIG. 4C. The pressure infiltration casting base plates are then inserted into a finished fin cooled heat sink. FIG. 5A shows the z-axis inserts positioned in the base plate. FIG. 5B shows the fin array 550 soldered on the casting base plate.
FIG. 6 illustrates a diagram 600 of a cutting scheme for three sets of orthogonal mitered preforms with z-axis inserts according to the present teaching. The z-axis inserts are rotated “quarter” pieces that maintain orthogonal spreading. The inserts shown in FIG. 6 have triangular cross sections that provide orthogonal heat distribution for heat spreading as well as heat sinking. These inserts are machined or sawed from an appropriate thickness preform of the desired volume fraction flake graphite. The half sections can be spliced so as to eliminate waste other than that consumed by the sawing process.
One advantage of the shape of the half sections shown in FIG. 6 is that using such half sections minimizes wasted material. FIG. 6 illustrates how the various insert sections can be rotated and fitted to provide for high thermal conductivity in orthogonal in-plane directions as well as in the through-plane (“z”) thickness direction without wasting any preform material other than the sawing kerf loss.
FIG. 7 illustrates diagrams 700 of hexagonal cross-section z-axis insets according to the present teaching that provide relatively uniform coefficients of thermal expansion and that can be matched to the device being cooled. Thus, the present teaching provides several high performance and cost effective solutions to z-axis heat sinking. The choice of the particular type of z-axis heat sinking depends upon the performance and cost requirements. Both high thermal conductivity heat sinking and unidirectional heat spreading can be achieved. In various embodiments, bars of z-axis insert material are machined and placed into round, square, rectangular or hexagonal through holes or via with no attempt to achieve radial or orthogonal orientation of the high conductivity planes in the material. For example, FIGS. 3 and 4 show one embodiment of the present teaching that uses square mitering. For many applications, it is not important to organize the x-y orientation of the inserts. Very low cost inserts can be manufactured by extrusion of round, square or hexagonal cross section preform material at appropriate temperatures above the liquidus of the binder.
In other embodiments, bars of z-axis insert material are machined and placed into round, square, rectangular or hexagonal through holes or vias in particular orientations, such as 90 degree rotations as described herein. For example, z-axis inserts can be mitered to have hexagonal symmetry with simple rotation and joining of half-⅙th pieces to maintain hex-radial heat spreading.
FIG. 8 shows a diagram 800 of one embodiment of a geometry according to the present teaching for z-axis inserts into AlGr_p™ or another composite and a metallic base plate material. The AlGr_p™ or other composite preform materials are machined or extruded, mitered and then inserted into a base plate so that the high thermal conductivity directions are orthogonal for square mitering (and at 0, 60°, 120°, 180°and 240° for hexagonally mitered inserts).
One advantage of the geometry shown in FIG. 8 is that the average thermal conductivity of the z-axis inserts after the Al infiltration can be calculated using the Schapery equation as shown in FIG. 8. In addition, the average thermal expansion of the material can be estimated by a variant of the Schapery equation. We can estimate the thermal expansion properties of AlGr_p™ materials as a function of volume fraction using the Schapery equation given below and assume values of AlGr_p™ material modulus and coefficient of thermal expansion.
$α_{ave} = \frac{0.5 (E_{x} α_{t} + E_{y} α_{t})}{0.5 (E_{x} + E_{y})}$
The Schapery equation gives an estimate of the average thermal expansion across the mounting surface of an AlGr_p™ z-axis insert that is positioned in a through hole or via. One advantage of using the AlGr_p™ material is that the volume fraction can be continuously varied from about 0.4 to 0.9. The thermal conductivity and the coefficient of thermal expansion can be calibrated as a function of the volume fraction of natural graphite using the data 100 shown in FIG. 1. The average coefficient of thermal expansion can be engineered to match the coefficient of thermal expansion of the electronic package or die.
The coefficient of thermal expansion of the insert can be engineered so that the average coefficient of thermal expansion of the insert matches or approximately matches the coefficient of thermal expansion of the baseplate or of the device being cooled. Such inserts will have a relatively high volume fraction of natural graphite. For example, the coefficient of thermal expansion of an AlGr_p™ preform with a volume fraction of natural graphite that is in the range of about 0.75 to 0.8 approximately matches the coefficient of thermal expansion of a GaAs chip or of a co-fired ceramic base electronic package. The coefficient of thermal expansion of an AlGr_p™ preform with a volume fraction of natural graphite that is about 0.9 approximately matches the coefficient of thermal expansion of commonly used semiconductor devices, such as GaN, Si or Si devices. In these examples, the z-axis thermal conductivity will be higher than the in-plane thermal conductivity of the planar isotropic AlGr_p™ base plate material.
Processing variants require the development of a separate calibration curve for each specific process modification. For example, if a base plate with an average coefficient of thermal expansion of 7 ppm/K is desired, a natural graphite flake volume fraction of 0.58 can be selected according to the calibration of a low ash, high purity graphite flake material, such as Asbury grade 3061 material, and the vacuum de-booking/microgating process used for that batch of material. According to the calibration curve, this material would have a thermal conductivity of approximately 640 W/mK.
The coefficient of thermal expansion can be engineered to result in the desired average coefficient of thermal expansion over the area of the surface of the mitered insert that is substantially equal to the dimensions for direct die attachment. However, the local coefficient of thermal expansion depends strongly upon the orientation of the mitered regions under the die. Thus, the stress on the die may be non-uniform and, in some circumstances, could cause damage to the device being cooled.
FIG. 9 shows a diagram 900 that illustrates the use of a skin of composite material to minimize local variations in the coefficient of thermal expansion on heat sink materials with z-axis inserts according to the present teaching. The use of a skin of composite material reduces or levels the stress in a localized area on the die attached to the device being cooled by the heat sinks with z-axis inserts. A skin of composite material according to the present teaching is engineered to have a volume fraction of composite material that produces a coefficient of thermal expansion that is matched to the device being cooled.
In one embodiment, a MetGraf™ skin is used to reduce or level the stress in a localized area on the die attached to the device being cooled by a heat sink fabricated according to the present teaching. The MetGraf™ composite material has a reasonably high thermal conductivity in the z-direction. The MetGraf™ composite material is particularly suitable for this application because a relatively thin skin of MetGraf™composite does not offer appreciable thermal impedance to the assembly.
In some embodiments, local variations in the coefficient of thermal expansion on heat sink materials with z-axis inserts according to the present teaching are minimized by encapsulating the composite material with a thin MetGraf™ skin or other composite material skin. In one particular embodiment of the present teaching, AlGr_p™ core material is encapsulated with a thin MetGraf™ skin material.
In some embodiments, a better quality surface and a more precise coefficient of thermal expansion can be achieved if the entire heat sink is covered with a composite skin. For example, covering the entire heat sink with a MetGraf™skin will seal the coarser flakes into the core region and permit standard plating of metals that are commonly used with un-hybridized MetGraf™ heat sinks.
In other embodiments, composite material, such as AlGr_p™, are skinned with other materials, such as silica scrim or felt or alumina felt. For example, skinning with silica results in the silica decomposing in-situ into alumina and Si in solution. The decomposed silica is harmless and will permit a metallic skin that is suitable for many different surface treatment processes and machining.
In one embodiment, the skin region is formed by laminating a thin layer of MetGraf™ milled carbon fiber preform, carbon felt, refractory felt, or a dusting of refractory powder. Laminating the thin layer of material excludes the graphite flakes from the AlGr_p™ material from contact with the surface. The resulting structure is a surface that can be precisely finished and plated for soldering or hard bonding semiconductor devices.
Using a composite skin material is particularly desirable when producing net shape castings. For example, an AlGr_p™ preform can be sandwiched with a skin preform. The preform can be a refractory felt or fiber glass scrim which prevents the coarse flake graphite material from contact with the surface of the composite material. After pressure infiltration, the net-shape composite material will consist of a high thermal conductivity core with or without z-axis inserts.
In some embodiments, local CTE control over z-axis insert sites is achieved using CTE matching surface pads. Inserts, such as AlGr_p™ z-axis inserts, sink heat from the semiconductor package to the base plate where it can be spread and removed. However, it is also necessary to match the CTE of the semiconductor or other type of device so as to permit a direct low thermal impedance solder attachment between the device and the heat sink. AlGr_p™ material oriented in the “z” direction has a CTE of approximately 4 ppm/K in the in-plane direction and 22 ppm/K in the transverse (through-plane) direction. Hence a semiconductor device hard bonded to such a heat sink would be subjected to complex stresses. Stresses between the device and the heat sink can be mitigated and even eliminated by inserting a thin “pad” of a controlled CTE material. MetGraf™ preform inserts can be applied as a “skin” to the base plate heat sink preform and co-infiltrated to produce panels.
FIG. 10A shows two base plates 1000, 1002 with z inserts mitered into an AlGr_p™ base plate. The first base plate 1000 has mitered z inserts into an AlGr_p™ base plate. The second base plate 1002 has mitered z inserts and a thin Al rich skin. Below the skin are MetGraf™ pads designed to level the CTE to a value of 7 ppm/K at the surface of the insert after infiltration. The CTE of the MetGraf™ skins were determined by the volume fraction of graphite fiber in the preform. At a volume fraction of 0.48, the skins have a CTE of 4 ppm/K.
FIG. 10B shows a diagram of the base plate 1050 shown in FIG. 10A with z inserts in a configuration that was used to experimentally determine the CTE above the pads. Pad A was fabricated without any MetGraf™ skin. Pad B was fabricated with a 0.010 MetGraf™ skin. Pad C was fabricated with a 0.020 MetGraf™ skin. Pad D was fabricated with a 0.040 MetGraf™ skin. The actual CTE was determined by the material of the base plate, the volume fraction of the AlGr_p™ inserts, and the thickness of the pads. For example, in the configuration shown in FIG. 10B, a Metgraf™ base plate with an average CTE of 9 ppm and a 0.010 inch thick pad of Metgraf™ 9 resulted in a CTE of 9 ppm/K above the inserts in both the x and y directions. In devices where local CTE values of 7 ppm are required, a lower CTE MetGraf™ pad can be used. Experimentally, local values of CTE have been produced from 4 ppm/K to 9 ppm/K in a base plate where local heat sinking using “z” inserts is 750 to 800 W/mK and local heat spreading is 200 W/mK for MetGraf™ baseplates to 650 to 750 W/mK for AlGr_p™ base plates.

EQUIVALENTS

While the present teachings are described in conjunction with various embodiments and examples, it is not intended that the present teachings be limited to such embodiments. On the contrary, the present teachings encompass various alternatives, modifications and equivalents, as will be appreciated by those of skill in the art, may be made therein without departing from the spirit and scope of the teaching.

Claims

1. A heat sink comprising:

a) a composite material base plate that defines at least one through hole; and

b) at least one composite material insert that is position into the at least one through hole prior to pressure infiltration, wherein the composite material insert is oriented to increase thermal conductivity in a through-plane direction.

2. The heat sink of claim 1 wherein the composite material base plate comprises a graphite material.

3. The heat sink of claim 1 wherein the composite material comprises a natural graphite flake material.

4. The heat sink of claim 1 wherein the composite material comprises chips of pyrolitic graphite.

5. The heat sink of claim 1 wherein the at least one composite material insert is formed of substantially the same composite material that defines the at least one through hole.

6. The heat sink of claim 1 further comprising a skin material that is positioned over the heat sink prior to pressure infiltration

7. The heat sink of claim 6 wherein the skin material comprises at least one of silica scrim, felt, alumina felt, metal matrix composite material, and a thinly applied ceramic powder.

8. The heat sink of claim 1 wherein the at least one composite material insert is machined from natural graphite.

9. The heat sink of claim 1 wherein the composite material base plate comprises a metal matrix composite material.

10. The heat sink of claim 1 wherein the at least one composite material insert is mechanically restrained and cast into at least one of molten Al, Mg, and Cu.

11. The heat sink of claim 1 wherein the composite material base plate and the at least one composite material insert are infiltrated with an Al-Si alloy.

12. The heat sink of claim 1 wherein the composite material base plate and the at least one composite material insert are infiltrated with an Mg alloy.

13. The heat sink of claim 1 wherein the composite material base plate and the at least one composite material insert are infiltrated with a Cu alloy.

14. The heat sink of claim 1 wherein an average coefficient of thermal expansion of the at least one composite material insert is varied from 4 ppm/K to 12 ppm/K to provide an approximate coefficient of thermal expansion match.

15. The heat sink of claim 1 wherein the composite material base plates comprise at least one of AlSiCp, and a metal matrix composite material.

16. The heat sink of claim 1 wherein the composite material inserts are pre-positioned into a metal matrix composite material preforms prior to infiltration.

17. The heat sink of claim 1 wherein the composite material insert are formed using hexagonal mitering to provide high thermal conductivity in regions around the center about 60 degrees apart so as to approach high thermal conductivity in radial directions.

18. The heat sink of claim 1 wherein the at least one composite material insert is formed using natural flake graphite and the volume fraction of natural flake graphite is adjusted to provide a predetermined average coefficient of thermal expansion.

19. The heat sink of claim 1 wherein the at least one composite material insert is mitered to provide high thermal conductivity in the through-plane direction.

20. The heat sink of claim 1 wherein the at least one composite material insert is mitered to provide high orthogonal heat spreading.

21. The heat sink of claim 1 wherein the at least one composite material insert comprises highly oriented pyrolytic graphite material mitered so as to direct heat in both the through-plane direction and radially outward.

22. A heat sink comprising:

a) a base plate that defines at least one through hole; and

23. The heat sink of claim 22 wherein the base plate comprises at least one of aluminum, copper and magnesium.

24. The heat sink of claim 22 wherein the composite material insert comprises at least one of aluminum graphite, copper graphite, and magnesium graphite.

25. A heat sink comprising:

a) a composite material base plate that defines at least one through hole;

b) at least one composite material insert that is position into the at least one through hole prior to pressure infiltration, wherein the composite material insert is oriented to increase thermal conductivity in a through-plane direction; and

c) at least one metal matrix composite material preform skin pad, wherein the metal matrix composite material comprising the preform skin pad is designed to achieve a predetermined local thermal expansion coefficient.

26. The heat sink of claim 25 wherein the composite material base plate comprises AlGr_p.

27. The heat sink of claim 25 wherein the composite material base plate comprises MetGraf™.

28. The heat sink of claim 25 wherein the at least one metal matrix composite material preform skin pad is co-infiltrated with the at least one composite material insert.

29. The heat sink of claim 28 wherein the co-infiltration is performed with at least one of Al, Mg, and Cu matrix alloys.