Heatsinks Shape Up To Face New Thermal Challenges

Rising power dissipations at the device and board levels continue to create more challenging thermal management problems in many applications. While heatsources are becoming more concentrated, pc boards are growing more densely populated. Naturally, these trends have put the onus on heatsinks to do a better job of transferring heat away from the semiconductors that they cool. Consequently, heatsink designs have evolved from the low-aspect-ratio, extruded aluminum heatsinks of yesterday to a wider array of options to deal with higher power levels.

Comparing different styles of heatsinks is a tricky business. Designers may look for a certain level of heatsink performance in their applications based on their power dissipation levels, ambient temperature, and allowable junction temperature. Those factors help the designer to determine the value of thermal resistance in degrees Celsius per watt that will be needed in their application. That requirement alone leaves them with a variety of options.

However, heatsink requirements become more focused, and more complex, as the designer factors in other constraints. For example, the space available in the application determines the possible heatsink volume. Another consideration is the designer's options for forced-convection versus natural-convection cooling. Even within a single heatsink style, performance can vary widely with different rates of airflow and depending on whether the airflow will be ducted (channeled) or nonducted.

The problem expands further when the same assembly contains multiple heat-generating or heat-sensitive components. Heatsink requirements will need to reflect the power being dissipated at both the component and board levels. Add to these requirements the many mechanical concerns, such as heatsink weight and method of attachment—two considerations that must take into account shock and vibration requirements—and other packaging details. Additional environmental factors like altitude also make an impact on heatsink performance.

Then there's reliability, which contributes to almost every aspect of heatsink selection. The temperature of the components affects their reliability in accordance with Arrhenius' Law: For every 10ºC drop in junction temperature, device reliability doubles. The reliability of the attachment method and thermal interface materials (how the heatsink gets assembled in production) also affects reliability. The reliance on forced-air cooling, which is of course influenced by heatsink design, adds the fan to the list of potential risk factors.

As with all engineering decisions, cost looms large. That differs from one heatsink style to another, from low to high volume, and from standard product type to custom. Tooling charges vary, depending on the type of heatsink selected. Keep in mind that the process in which the heatsink is fabricated, such as die casting, extruding, bonding, brazing, or machining, determines not just the heatsink's cost, but also the vendor's ability to optimize the heatsink to the user's specifications.

The designer must weigh all of these concerns when evaluating different types of heatsinks for the application at hand. Unfortunately, making an apples-to-apples comparison from one heatsink style to another, on paper, is bound to be difficult because heatsink vendors don't obtain their datasheet specifications under the same test conditions.

Moreover, these conditions may not resemble the conditions present in the application. Ultimately, some testing of the heatsink in the application will be in order, a task that may be mitigated by intelligent thermal modeling of the design. So choosing a heatsink isn't a trivial task. It must be considered as simply one step within a well thought out thermal management scheme.

In fact, thermal management starts well before the heatsink design. Hardware and software optimization can minimize power dissipation. Careful pc-board layout can reduce cooling re- quirements. Designers with experience in thermal management can address these concerns early in a project, perhaps sparing the need for an expensive high-performance heatsink in lieu of a smaller, lower-cost component.

For customers lacking this experience within their own companies, there are thermal management experts within the heatsink industry who can help engineers optimize their thermal designs, rather than simply selling them a thermal fix. So, a customer's experience in thermal management may influence the choice of a heatsink technology and the selection of a supplier.

Extrusions And Beyond: For many years, extruded aluminum heatsinks provided adequate cooling through convection (Fig. 1). However, increases in power dissipation caused many applications to switch to forced-convection (forced-air) cooling, which lowers the thermal resistance of the heatsink. But for forced convection to be effective, the air must move in the direction of the fins, so the positioning of the heatsink and the fan must be coordinated.

Nevertheless, those specifying the heatsink may not always know how it will be oriented with respect to airflow. To accommodate those cases, heatsink vendors introduced cross-cut extrusions—extrusions that have been machined using a high-speed saw to make perpendicular cuts in the fins. Cutting the fins produces square pins that allow omnidirectional cooling.

The omnidirectional concept has been extended to develop a variety of fin and pin shapes. For example, pin-fin heatsinks feature round rather than square pins (Fig. 2). These have a few advantages over square pins. When air is blown across them, more turbulence is created, improving cooling. This effect is accentuated at higher air speeds. Another advantage is that the round edges of the pins reduce the pressure drop over the heatsink in comparison to squared edges of cross-cut extrusions. A lower pressure drop over the heatsink translates to faster airflow over components that are downstream from the heatsink.

Cross-cut extrusions and pin-fin heatsinks are assembled in very different processes. The extrusion process may be compared to squeezing a tube of toothpaste. The extrusion manufacturer presses heated aluminum through a profiled orifice. That process only allows symmetrical profiles, so it can't produce pins.

Pin fins are produced in a process called cold forging, also known as impact extrusion. In this process, which is popular in the canning industry, a slab of the raw heatsink material is placed under an open die. The material is forced through the holes in the die to form the pins. Whereas the extrusion process requires a separate die for each heatsink height, cold forging permits the use of one die to fabricate heatsinks of varying heights.

Therefore, pin-fin customers can optimize the heatsink height for maximum cooling in their applications without incurring additional tooling costs. According to Barry Dagan, technical director of pin-fin heatsink manufacturer Cool Innovations, pricing for pin-fin heatsinks is in the same range as cross-cut extrusions. As volumes in-crease, the falloff in heatsink price between low- and high-volume becomes even more dramatic for pin-fin heatsinks than cross-cut extrusions.

A variation on the pin-fin heatsink is one that uses elliptical pins. These are produced in a process known as microforging that differs from impact extrusion. While both processes use high temperature to shape the metal, impact extrusion is carried out at a lower temperature than microforging. Nevertheless, microforging never completely melts the metal as does the casting process.

The elliptical pin, shaped like an ocean liner, may produce even less of a pressure drop than the round pin fins. According to Chris Soule of Therm-shield, manufacturer of elliptical pin heatsinks, these devices can outperform pin fins, even when the pin fin is made from the more thermally conductive copper, and when the elliptical pin is made from aluminum. Soules says that an aluminum elliptical pin heatsink can achieve 10% less pressure drop and 20% better thermal resistance, weigh less, and cost less than the copper part. The aluminum performance, however, is better than copper at lower air speeds, below 400 lfm. The advantage will diminish as the air speed increases.

Radian, another source for elliptical pin-fin heatsinks, advises that their thermal performance at airflows of under 300 lfm is generally lower than that of the round pin fins, but greater than the plate fin, when fins of equivalent thickness, spacing, and material are compared.

While efforts to develop pin-style heatsinks are ongoing, heatsink vendors also are working to improve their extrusions. This primarily occurs by increasing fin density to achieve greater surface area because the ability to transfer heat away from the heatsink is proportional to its surface area.

The pursuit of greater surface area has also produced variations on the standard extrusion (Fig. 1, again). For instance, the semi-hollow extrusion increases surface area by forming a "T" at the tip of the fin. While this does aid heat removal, it also increases the static pressure drop of the air moving over the fins.

Another variation, the full-hollow style, ensures proper airflow by keeping the airflow within the heatsink. It has other advantages, too, such as the ability to draw heat from both sides of the heatsink. In addition, the structure has use in liquid-cooled cold plates. However, the full-hollow shape is more difficult to tool than the conventional extrusion.

Those variations aside, conventional extrusions continue to develop greater surface area by packing fins more closely together. Fin density is typically expressed as an aspect ratio (also referred to as an extrusion ratio) of the fin height to the distance between fins. In the past, that ratio was normally around 3:1, but today some vendors claim extrusion ratios of as high as 15:1 or 16:1. One vendor, Therm-shield, currently boasts the ability to fabricate extrusions with an aspect ratio of 24:1.

But fin density is a topic that sparks some controversy. According to Cliff Schroeder of Tran-tec, a company that builds aluminum extrusions to order, an aspect ratio of 10:1 or 12:1 is pushing the state-of-the-art. Above that fin density, Schroeder says, extrusion manufacturers run the risk of breaking the die. Thermshield's Soule concurs, noting that the higher the extrusion ratio, the taller and thinner the die sections, and the more prone these sections are to movement and possible failure during the pressing process.

To obtain its high aspect ratio, Thermshield reinforces the die through the use of stronger die materials. But even with those steps taken, when it develops a high-aspect ratio profile, it builds multiple dies as one will either fracture or be ground down too far to run straight.

Nevertheless, for high fin density, heatsink manufacturers have another option in plate-fin heatsinks. These can be either bonded or brazed fin constructions. In the former style, the vendor machines grooves into a base plate and glues the fins into the base plate using a thermally conductive epoxy. In the brazed fin approach, the fins are essentially soldered into the base plate.

Customers pay a premium for the plate-fin heatsink approach, which could cost three to four times that of an extruded aluminum heatsink, says Schroeder. The payoff, though, is an aspect ratio up to approximately 25:1. The limit on aspect ratio isn't so much a limit on heatsink manufacturing, but on application concerns.

At higher aspect ratios, which could be achieved using tighter fin pitch, thinner fins, or taller fins, the user may begin to experience heat entrapment or hot spots within the heatsink. That could force the user to apply greater airflow in the application. Increasing fin density also raises the pressure drop over the heatsink, which may require designers to select a more powerful fan to compensate.

In terms of fin height alone, the issue of heat entrapment is a greater consideration than heatsink manufacturability. With plate fins, it would be possible to fabricate a 2-ft high fin in a bonded-fin heatsink. (That could result in an extraordinarily high aspect ratio.) With an extrusion, the limit on fin height would probably be closer to 6 in. But even at that height, heat entrapment can become an issue.

Testing performed at Tran-tec indicates that fins of up to about 4 in. still provide good heat transfer from base plate to fins, and many customers prefer this height. Naturally, for applications demanding low-profile heatsinking, space restrictions can establish lower limits for heatsink heights.

In general, space constraints weigh heavily in the specification of a heatsink. As Gerry McIntyre, president of Radian, explains, the customer wants to know, "Given the amount of space, how much cooling can he get? Each fin type is something you can pull out of your bag of tricks to help your customer." There are tradeoffs with each heatsink style.

Another alternative is the folded- or rolled-fin heatsink. As with some of the other structures, these heatsink styles have existed for years. But only in the past few years have they begun to appear in mass-produced consumer applications like desktop PCs. Their performance is similar to the plate-fin designs, which offer similar surface area. However, the processes for manufacturing rolled and folded fins are said to be simpler and less costly than those used to build plate fins. Compared to extrusions, folded- or rolled-fin heatsinks provide thinner fins at a higher density but naturally cost more than the extrusions (Fig. 3).

The difference between a folded fin and a rolled fin is primarily a matter of the different manufacturing techniques used to bend the aluminum stock to obtain the fin shape. The rolled-fin method permits some enhancements to the fin structures, such as the inclusion of louvers, offsets, and splits. But these enhancements don't offer dramatic improvements in performance at the low levels of airflow normally seen in consumer electronics. The folded- and rolled-fin constructions can be manufactured in fin heights ranging from about 1 in. (easy to fabricate) to 2.5 in. (difficult).

Thermacore has developed a variation of the folded-fin construction that improves performance of this heatsink to meet the cooling requirements of Intel's P4 processor. In a typical design, the folded fin is mounted through brazing to a flat base plate. That assembly orients the fins vertically. If a blower is placed above the heatsink, as in a straight impingement design, air is blown down into the folds and exits through the sides of the heatsink.

Thermacore's new design employs a vertical spline conduction plate as opposed to a more typical horizontally mounted base plate. The vertical spline permits the fins to be rotated such that a blower mounted above the heatsink blows air through the fins. As a result, the air passes over more of the fins' surface area, lowering thermal resistance (Fig. 4).

Yet another option, first introduced by Advanced Thermal Solutions (ATS) about five years ago, is the fan-tail heatsink. These heatsinks resemble extrusions and plate-fin heat-sinks, except for the way in which the fins fan out toward the edge of the base plate and actually hang over the edge. The height of the fins with respect to the base plate is constant. While the center fins are mounted at 90° to the base plate, the outer fins are mounted at lower and lower angles such that the outer fins are longer than the center fins. Several advantages accrue from this design, like lower pressure drop and airflow over a greater length of the fins.

Consider airflow through a typical extrusion. As air flows down the channel, pressure increases against the airflow and diverts the air out the top of the heatsink. In a 30-mm long heatsink, the air starts to leave somewhere between 25% and 50% of the length of the fins. Fanning out the fins can extend that point to somewhere between 50% and 75% the length of the fins.

The outer fins are longer and thicker than center fins, so they have greater surface area and lower thermal resistance. The low resistance path at the edges of the heatsink produces greater heat spreading in the base.

For many applications, the base plate footprint and heatsink height are the key parameters. Within these given dimensions, the fan-tail heatsink effectively allows for a wider heatsink than a standard extrusion or plate-fin heatsink. That's be-cause the outer fins of the fan tail provide some clearance off the board where using a larger extrusion would increase the heatsink's base plate dimensions and interfere with other components on the board.

Originally, these heatsinks were manufactured by brazing the fins to the base plate. Yet in the last 18 months, ATS has developed extruded versions, which offer lower cost at the expense of lower fin density. But more recently, the company has developed techniques for machining fan-tail heatsinks, making the higher fin densities possible at lower cost than the brazed versions. An extruded model might employ a fin thickness of 0.060 in. with a 0.090- or 0.120-in. fin spacing. Machined versions of the heatsink could have 0.020-to 0.040-in. fin thicknesses with fins spaced at 0.060- to 0.090-in. intervals.

The Bases For Change: While different fin structures affect the heatsink's ability to transfer heat through natural or forced-convection cooling, heatsink manufacturers have another tool in their arsenal—copper. Aluminum has long been the standard material for heatsink design, but copper offers higher thermal conductivity and lower spreading resistance. Compared to extruded aluminum, pure copper exhibits twice the conductivity, which translates into better heatspreading and greater fin efficiency. But there are several caveats.

The higher fin efficiency is only apparent at airflow rates in excess of 800 lfm. For spreading to be effective in the base, there should be a hot spot on the mounting surface that's no greater than 25% of the heatsink base. Moreover, a copper heatsink may cost three times that of a similarly sized aluminum part, reflecting the higher material and fabrication costs. Because copper can't be extruded, copper heatsinks must be machined, and that process is more demanding than the machining of aluminum. See Copper Heat-sinks Versus Aluminum Heatsinks—Advantages and Disadvantages at www.thermshield.com.

Copper does lend itself to fabrication of pin-fin heatsinks. One vendor, Cool Innovations, indicates that the forging process allows it to run the same production lines using either aluminum or copper. Some vendors fabricate plate-fin heatsinks using a copper base with aluminum fins. This approach improves heatspreading in the base over an all-aluminum heat-sink, yet it weighs less than an all-copper heatsink.

But copper is just one way to improve heatspreading. Heatpipes and vapor chambers offer other, sometimes complementary, alternatives. A heatpipe encloses a working fluid, such as water or methanol, within a vacuum-tight copper envelope that also contains a wick structure.

When heat is applied at the evaporator end of the heatpipe, the working fluid within the pipe changes to vapor and travels through the pipe to the condenser end where its heat is transferred to the attached heatsink. As that heat is released, the vapor changes back to liquid, which the wick then absorbs. Next, through capillary action, the working fluid travels back to the condenser end. A vapor chamber is similar to a heatpipe but has a flat planar envelope that spreads heat in two dimensions.

According to Jon Zuo of Thermacore, using heatpipes within an aluminum base provides a lighter alternative to the solid copper base while offering similar performance and cost. For applications that require even better heatspreading, heat pipes can be deployed in a copper base. Beyond the heat pipe and copper combination, designers can obtain still greater heatspreading via vapor chambers. In those cases, the vapor chamber typically replaces the base as the means for heatspreading.

While aluminum and copper account for the bulk of heatsink development for electronics, thermally conductive thermoplastics are beginning to offer some interesting alternatives. One company that develops plastic heat-sinks is Cool Shield. According to Jeff Panek, the company's thermoplastics offer conductivity on par with cast aluminum, which goes up to about 100 W/(m•k). (Extruded aluminum and copper exhibit conductivities of 200 and 300 W/(m•k), respectively.) Most heatsinks developed by Cool Shield have conductivity values in the 20- to 45-W/(m•k) range because their customers' applications are convection- rather than conduction-limited.

Use of a plastic heatsink provides two main advantages—weight and high-volume manufacturability (Fig. 5). Plastic heatsinks weigh half as much as aluminum heatsinks. In addition, they're produced in injection molds, which provide an easier and more flexible manufacturing process than the common stamping, casting, and machining techniques used to build standard heatsinks. Moreover, the thermoplastic materials maintain tight three-dimensional tolerances better than other plastics because the thermoplastics cool more uniformly in the molds. As a result, these composite materials also find uses outside of thermal management.

But for building heatsinks, the process also has economic advantages. Tooling costs for the molds is half that of cast aluminum, which typically requires secondary operations. According to Panek, the plastic heat-sinks become cost effective with orders of "at least a couple thousand units per run."

By varying the composition of the plastic, Cool Shield can control its conductivity. That enables designers to build a heatsink that doubles as an electromagnetic interference (EMI) shield. In addition, the ability to mold different shapes and features with each heatsink means that the final design can address other mechanical requirements, such as the need for heatsink attachment and board stiffening. Going beyond these features, the heatsink may even be integrated within the design of the product enclosure, creating new opportunities to merge packaging and heatsink design.

For further reading on heatsinks and other thermal management topics see:
  • "How to Select a Heat Sink," at www.aavidthermalloy.com/technical/index.shtml along with many related papers; "The Basics of Aluminum Extrusions" at www.thermshield.com; "Heat Pipes: A Practical and Cost Effective Method For Maximizing Heat Sink Effectiveness" at www.thermacore.com/papers.htm; "Heat Sink Design and Fan Selection Optimization" at www.enertron-inc.com/library.htm; and for a broad list of resources see the "Reference Library" of the Electronics Cooling Web site at www.electronics-cooling.com.