Heat Sink
A heat sink presents a thermal resistance Rθsa (case-to-ambient, °C/W) that, in series with the device's Rθjc (junction-to-case), determines the steady-state junction temperature for a given dissipated power. Common forms: extruded aluminium black-anodised fin stacks, skived copper heat sinks, bonded-fin heat sinks, vapour chambers, and pin-fin coldplates.
Heat flows by conduction from the silicon die through the package case, through a thermal interface material, and into the heat sink's base; the base spreads heat across the fin array. The fins enlarge the surface area so air can carry heat away by natural or forced convection at a rate proportional to surface area, fin emissivity, and air velocity. Modern heat sinks combine extruded aluminium for low cost, embedded copper heat pipes that move heat by phase change (evaporation/condensation) to remote fin stacks, and vapour chambers for ultra-low spreading resistance.
In plain terms
A pot's pasta strainer used backwards: the metal of the strainer has lots of surface area, so a hot pasta resting on it cools faster than it would in a sealed bowl.
Why designers use it
- Keep semiconductor junctions below their Tjmax to prevent thermal runaway and lifetime collapse.
- Move heat from a small die package to a large air-cooled fin stack passively, no fan required.
- Stabilise device parameters that drift with temperature (oscillator frequency, op-amp offset).
- Comply with regulatory case-temperature limits on touch-accessible enclosures (IEC 60950).
Best for
- Power devices
- CPUs
- LED drivers
Key specifications
- Operating limits: Check the exact manufacturer's datasheet (A family name does not establish voltage, current, temperature or timing limits.)
- Pin assignment: Match the complete part and package code (A similar name or function does not guarantee the same wiring.)
- Mechanical fit: Use the exact package drawing (Check pad layout, dimensions and viewing direction before building.)
When not to use it
- For pulsed loads where transient thermal impedance matters more than steady-state — a heat sink with too high heat capacity may help, but in milliseconds the die heats up before any heat reaches the fins.
- In sealed enclosures with no airflow path — the heat sink simply heats the trapped air; a heat pipe to a chassis-mounted radiator works better.
Common mistakes
- Tightening the heat-sink screws too aggressively over a TO-220 — case warpage opens an air gap under the die, raising thermal resistance instead of lowering it.
- Mounting fins horizontally instead of vertically in a still-air application — natural convection chimneys only work when fins line up with gravity-driven airflow.
Where you will find it
- An Intel Core i9-13900K processor uses a Noctua NH-D15 tower heat sink: six 6 mm copper heat pipes wick heat from the IHS to two large aluminium fin stacks, achieving an Rθsa of about 0.13 °C/W with a 140 mm fan — without it, the 250 W TDP CPU would cross 100 °C in under 30 seconds and throttle.
- A Cree XHP70.2 high-power LED is bonded to a 50 mm-diameter pin-fin aluminium heat sink in a flashlight torch body: the LED dissipates 30 W at 25 A; the heat sink + thermal grease + machined aluminium body achieve under 2 °C/W from junction to ambient, holding the LED below its 105 °C Tjmax during the high-output 'turbo' mode.
- A Tesla Model 3 SiC inverter uses a finned aluminium baseplate with embedded ethylene-glycol coolant channels (a 'cold plate' rather than air cooled): the SiC MOSFETs in the power module reject 30 kW of waste heat through a 0.04 °C/W liquid path, allowing the inverter to fit under the floor while sustaining 250 kW peak output.
A short history
Heat sinks provide an additional path from a hot device to the surrounding environment. Thermal-resistance models help estimate temperature rise, but their values depend on mounting, airflow and the actual assembly. A datasheet's junction-to-ambient result on a test board is not a universal heat-sink sizing rule.