Minimizing cost and maximizing thermal performance are the obvious goals of cooling a power semiconductor. Thermal performance is related to the power semiconductor's junction temperature; the lower, the better and the higher the reliability. Overall cost depends primarily on the thermal management device that absorbs and dissipates heat from the power semiconductor either directly or by radiation. Currently, heat sinks are the lowest cost, most widely-used thermal management device for power semiconductors.
One of the newest heat sink improvements is ultra-thin aluminum fins spaced closer together than traditional heat sinks. Ultra-thin fin extrusions are just an iteration of conventional technology. Their cost penalty is only an increased tooling cost, which is about 150% of traditional tooling; however, it is only a one-time cost. The benefit of ultra-thin fins is lower assembly labor costs compared with other heat sink types. For example, it lowers cost by eliminating the fin-to-bond joint, of the frequently-used bonded fin type.
To understand how the newer heat sinks provide better performance at lower cost, we have to look at a typical heat sink employed with a power semiconductor. Fig. 1 shows the typical configuration of a heat sink cooling a power semiconductor. The junction temperature reduction occurs in three steps. First, heat moves from the junction to the case via θj-c, the thermal resistance from the junction-to-case. Then, the heat moves from the case to the heat sink via the thermal interface material whose thermal resistance is θc-s. In the final step, the heat moves from the heat sink to the air through thermal resistance θs-a. In most cases, we can assume a thermal interface material with a very low thermal resistance, so the heat sink is the major contributor to removing heat from the power semiconductor's die. The heat sink's ability to reduce the temperature of the power semiconductor's die, or junction, is a measure of its thermal performance.
Fig. 2 is an example of determining the thermal resistance required by a heat sink based on a power semiconductor's characteristics. This power semiconductor has a 10W power input, a maximum junction temperature of 100 °C, and an internal thermal resistance of 1.0 °C/W (θj-c). The thermal interface resistance (θc-s) is 1.5 °C/W. Therefore, this system requires a heat sink with θs-a = 3.5 °C/W to achieve a maximum ambient temperature of 40 °C. In this case, the maximum temperature rise in the heat sink will only be 10W × 3.5°C/W, or ≤ 35 °C.
Basically, the heat sink removes heat from the power semiconductor by providing a cooler temperature direction for the heat to move toward. The overall low resistance path provides a place for the waste energy to drain off. Therefore, the heat sink must be maintained at a lower temperature than the heat source. Ultimately, all heat removed from the power semiconductor will be exhausted into the air.
EXTRUDED HEAT SINKS
Today, many heat sinks are produced by an extrusion process. Thermal extrusions, as with most aluminum extrusions, are formed as two-dimensional parts along a length. Most extrusions are made in large hydraulic presses that use a specifically shaped tool to form parts. The size of the press is designated by the number of tons of force it can apply and the maximum circle size of the tooling. The circle size of the tool limits the overall size of the maximum extrusion profile that can be produced by a given press. This largest allowable size is the combination of width and height that will fit within the given circumscribed diameter. Press circle sizes range from 2.0 inches to as large as 30.0 inches. Generally, there are fewer presses available for the larger the tool sizes.
Continue to next page
The process of extrusion makes parts that come from the extrusion press at upwards of 100 feet in length. Typically stocked in 8 foot or 3 meter stick lengths, they are raw material used as the basis of custom machined heat sinks. The process of transforming them into usable heat sinks requires that the parts be cut to length, assembly holes be added to attach power dissipating devices, and mounting holes added for mounting. All raw material shapes have characteristics used for comparison. For example:
- Weight per unit length - typically pounds per foot in the U.S.
- Surface area per unit length - square inches per inch.
- Thermal resistance for natural convection at 3.0 inch (76 mm) length; specified in terms of °C/Watt, including a black anodize finish.
The key to cooling with convective heat transfer is the surface area exposed to the ambient air flow. The greater this surface area, the greater its cooling capability. Combining this dissipation surface with a high heat transfer number for each square centimeter reduces the heat sink's overall thermal resistance. This increases heat dissipated at a given temperature rise. It is true for all heat sinks, especially those that are cooled by air. The objective of different manufacturing styles is to offer the greatest cooling area at the smallest volume at the lowest cost. Optimization of all three of these parameters results in a best fit design for any given application.
ULTRA-HIGH RATIO EXTRUSIONS
The innovation of ultra-high heat sink extrusion technology has resulted from improved manufacturing techniques that produce very thin fins, including:
- In-depth computer analysis of the forces in the extrusion pressing process dedicated to each extrusion shape.
- Improved tool steel for the extrusion dies.
- Improved controls of the hydraulic ram used in the extrusion “push,” which took some of “black magic” out of the pressing process.
Typically, ultra-high ratio extrusions provide:
- Up to 23:1 in 5.0-inch diameter tool
- Up to 20:1 in 8.0-inch diameter tool
- Fin thicknesses down to 0.040-in. at 1.0-in. height.
Advantages of ultra-high ratio heat sink extrusions include:
- Single-piece construction
- Fin thickness' down to 0.50 mm depending on fin height
- Typically 6063-T5 (high conductivity) aluminum alloy
- More cooling surface area in equal volume
- Lighter weight - lower cost to buy and ship
- Increased thermal performance in forced air
- Their lower cost allows replacement of bonded-fin heat sinks
Compared with traditional extrusions, ultra-high ratio types increase the number of fins per inch of width, providing more cooling surface in a smaller volume. By definition a heat sink extrusion ratio is the height of the open air gap between fins divided by the width of that gap. For example: 40 mm tall fins with 3 mm average air gap ratio of 13.3:1. This ratio applies to all extruded aluminum heat sinks and is limited by the tooling using to make extrusion dies. Typically made of very strong tool steel, these circular dies must hold back tons of force at elevated temperature during the extrusion process. Many extrusion presses place more than 1000 tons of force against the die during processing. The key importance of ultra-high fin ratios is that the technology used to increase cooling is done at extremely reasonable costs, not much more than conventional extrusions.
Continue to next page
FORCED AIR CONVECTION REQUIRED
Previously, extrusion ratios were limited to roughly 10:1 within an 8-inch diameter tool and roughly 8:1 in 12-inch diameter tool. Fig. 3 shows the cross-section of an older traditional extruded heat sink with a Y/X ratio of 8:1. Compared to newer heat sink, these heat sinks are relatively inefficient in their usage of material, and may lack the cooling capability to sufficiently reduce thermal junction temperatures to the safe operating levels required for today's hotter-running power semiconductors.
Thanks to manufacturing process refinements, companies such as Thermshield can now extrude heat sinks offering ultra-high ratios up to 23:1. Fig. 4. shows the cross-section of an ultra-high extrusion heat sink.
Most of these high ratio shapes provide a higher level of cooling capability in a smaller volume of space than their predecessors. Such heat sinks are more efficient in their material usage and achieve lighter weight than traditional extruded heat sinks. Fig. 5 shows some of the current ultra-thin heat sinks available.
The closely spaced fins are not without their drawbacks, however. To achieve maximum heat dissipation and thermal performance, ultra-high extrusion heat sinks may still require forced convection cooling, often using fans that add space, mechanical complexity, and cost to the design. Natural convection will not be used very often with ultra-high extrusion heat sinks.
COPPER VS. ALUMINUM HEAT SINKS
Sometimes, designers want an alternative heat sink material, like copper, to improve performance when an aluminum sink may not provide enough cooling. In some cases this switch is justified, in other cases it may not be. Following are a few rules of thumb for when the extra cost and weight of copper makes sense.
Pure copper has about 2X the conductivity of extruded aluminum. This added conductivity helps reduce semiconductor temperature in heat spreading and fin efficiency. However, this is only useful when air flow speeds are very high (over 800 linear feet per minute). The heat input area, the hot spot on the mounting surface, is small (25% or less) compared to the heat sink base size. When airflow speed is below 400 linear feet per minute and/or the heat input area is a high percentage of the heat sink base area, the extra cost of copper may not be justified. The cost of an all copper heat sink is typically 3X the cost of an equivalent size (fin count, base thickness, fin height, etc.) aluminum part.
This cost increase is based on several factors. Copper cannot be extruded in the same manner as aluminum; therefore it must be machined from flat plate with fins or other features brazed in place.
Due to copper's density and its abrasive nature, machining holes and other details in copper takes significantly longer and wears out tooling at a higher rate.
Cost of the base copper material is about the same as aluminum per pound but at 3X the density per cubic inch, resulting in a raw material cost of 3X that of aluminum.
Copper heat sinks may have an advantage in increasing heat removal and lowering semiconductor temperatures. However, the part's added cost and custom nature adds system complexity from both the thermal and the economic sides of a design.