Heat Sink Stamping: Fins, Brackets and Thermal Paths

A stamped heat sink that looks right on the drawing can still fail in the field. Three things kill these parts in the first year: thermal resistance creeping 20-30% above the spec because the fin geometry was optimized for the punch press instead of the airflow; fins that bow and twist after forming, choking the very channels they were meant to open; and a mounting interface that never seats flat, leaving a micron-scale air gap a thermal pad cannot bridge. I have pulled 6063 brackets off returned LED drivers where the contact patch covered less than half the die. The sink was fine. The assembly was not.

This guide is the checklist I hand to procurement and design teams before they release a stamped sink to production. It is built on numbers you can verify against any metals databook, not marketing claims.

We cover the alloy decision first, because it sets the thermal ceiling. Then the fin geometry arithmetic that decides how much of that ceiling you actually reach, the four manufacturing routes you can quote (stamped, extruded, skived, bonded), the mounting and surface-finish decisions that show up in field temperature, and finally the cost structure and supplier data you should demand before you approve a die. If you only take one thing away: a stamped heat sink is a thermal system with a manufacturing constraint, and every section below is written to keep those two halves honest with each other.

Two Aluminum Grades, Two Different Jobs

Most stamped heat sinks live in one of two alloys, and the choice is a thermal decision before it is a cost decision.

Property	Aluminum 1060	Aluminum 6063
Thermal conductivity	234 W/m·K	209 W/m·K
Tensile strength (annealed / T5)	~70 MPa	~150 MPa (T5)
Formability	Excellent	Good
Bend radius (thin stock)	Tight, clean	Moderate
Typical use	High-conductivity fins, thin walls	Brackets, structural clips, extrusion-adjacent parts

1060 wins on conduction by about 12%. That gap matters when the heat path is long and thin - exactly the case for tall, closely spaced fins. 6063 holds its shape under clamp load and resists the spring-back that loosens a clip mount. Pick 1060 for the fin stack where every watt per meter counts. Pick 6063 for the bracket that has to survive 5,000 insertion cycles without losing tension.

Do not mix the two in one part unless the tooling plan accounts for different spring-back. I have seen a 0.3 mm difference in recovered bend angle between grades scrap a whole clip run.

Where 5052 and 6061 Fit

Two more grades show up often enough to name. 5052 (roughly 138 W/m·K) is the structural workhorse: it forms cleanly, resists corrosion, and carries welded or riveted brackets, so it shows up on outdoor LED housings where the sink is also the enclosure. 6061-T6 (roughly 167 W/m·K) sits between 1060 and 6063 in both conduction and strength, and is a reasonable single-alloy compromise when one part must be both fin stack and structural member. The penalty for choosing either over 1060 is conduction; the reward is that the part survives vibration, salt, and repeated disassembly without yielding. If the application is sealed and static, 1060 stays the right call.

Whatever grade you choose, specify the temper on the drawing. Annealed -O stock is what you want for deep fin forming; -T5 or -T6 tempers are for parts that carry load. A drawing that says only "aluminum" leaves the die builder guessing, and the guess shows up as spring-back scatter in the first article report. Our presses run 0.05-3.0 mm material with strip widths up to 650 mm, which covers the thin fin stock and the wide multi-row fin arrays this grade discussion depends on.

Fin Geometry: Spacing, Height, and Deformation Risk

Fin Spacing Is an Airflow Problem, Not a Styling Choice

The instinct is to pack fins tight for more surface area. The physics pushes back. Stamped fins below roughly 2.0 mm pitch start to stall natural convection because the boundary layers from adjacent fins merge. You gain area and lose airflow at the same time.

For a sink cooled by still air, I target a pitch of 4-8 mm. Forced air at 1-2 m/s lets you drop to 2.5-3.5 mm. Below 2.0 mm you are paying stamping tooling cost for fins that barely help.

Worked example. Take an 80 mm x 80 mm base in 1060, 3 mm thick, carrying 36 stamped fins that are 0.5 mm thick, 30 mm tall, and 70 mm long. Center-to-center pitch works out to about 2.2 mm, leaving a 1.7 mm air gap. Fin surface area is two faces per fin: 36 x 2 x (0.030 x 0.070) = 0.151 m². At a natural-convection coefficient near 8 W/m²·K, the fin contribution alone is roughly 1 / (8 x 0.151) ≈ 0.83 °C/W. Drop that pitch to 1.5 mm and you add fins but the effective h falls - net thermal resistance barely moves while scrap rate climbs.

The lesson: surface area only helps if air can reach it. Tight pitch is a trap that looks like a win on the spec sheet.

Fin Height, Wall Thickness, and the Deformation Risk

Tall thin fins are the hardest thing to stamp well. A 30 mm fin from 0.4 mm stock has an aspect ratio of 75:1 after forming. Spring-back bows the tip, and a bow of 0.5 mm across a 1.7 mm channel is a 30% blockage. That is the second failure mode from the opening paragraphs.

Control it three ways. First, keep wall thickness at 0.5 mm or above for fins taller than 25 mm unless you are running a progressive die with in-die straightening. Second, anneal 1060 to the soft -O temper before forming; work-hardened stock cracks at the root radius. Third, design the fin root with a generous fillet - a 1.5 mm radius versus a sharp 0.2 mm corner cuts root stress by roughly half and stops the crack that propagates from the base.

Flatness after forming is a real number, not a hope. On a 30 mm fin I hold tip-to-tip bow under 0.3 mm across the 80 mm width. Anything more and you are hand-sorting parts on the line.

Fin height	Minimum wall thickness	Forming approach
Up to 15 mm	0.3-0.4 mm	Standard progressive die
15-25 mm	0.4-0.5 mm	Progressive die, controlled bend radius
25-40 mm	0.5-0.6 mm	Progressive die with in-die straightening
Above 40 mm	0.6 mm or more	Stamping usually loses to extrusion or bonding

These are guidance bands, not guarantees - the real limits come from the alloy temper, the die geometry, and the press speed, which is why the first article report on a tall-fin sink should always include measured bow, not just drawing compliance.

Stamped, Extruded, Skived, or Bonded: Choosing the Route

Before you optimize the fin pitch, make sure stamping is even the right process for the volume and the geometry. Each route has a different cost curve and a different thermal ceiling, and the worst mistakes I see are parts that were drawn for one process and quoted for another.

Route	Fin pitch floor	Tooling cost	Unit cost at 100k	Thermal ceiling	When it wins
Stamped (progressive die)	~2.0-2.5 mm	Mid ($15k-60k typical)	Low	Good; limited by pitch and fin height	High volume, integrated brackets, mounting features, clips in one piece
Extruded + cut	~1.0-1.5 mm	Low (extrusion die)	Low-mid, plus sawing	Very good; dense fins	Long straight channels, moderate volume, simple cross-sections
Skived (machined)	~0.5 mm	None (per-part machining)	High	Best fin density	Low volume, premium density, no tooling commitment
Bonded / folded fin	~1.0 mm	Mid (assembly fixtures)	Mid	Very good; fin material independent of base	Mixed-metal sinks, very tall fins, low-to-mid volume

Stamping wins when the sink needs more than fins: a clip latch, a locating boss, a threaded boss collar, a PCB standoff, or a mounting bracket formed into the same piece of metal. Extrusion gives you denser fins per dollar of tooling but every one of those features becomes a secondary machining operation. Skiving gives the densest fins of all but carries a per-part machining cost that only makes sense below roughly 10,000-20,000 pieces a year.

There is also a hybrid route worth naming: stamp the base and bracket, then bond or stake a separately formed fin pack. It is how we build tall-fin sinks for telecom power supplies when a single-piece stamping would exceed the formability limit. The bond line adds a small conduction resistance, but the design freedom buys fin heights no progressive die can reach.

If you are early in the decision, the two questions that settle it are: how many pieces per year, and does the part need features beyond the fins themselves. For volume above roughly 50,000 pieces with integrated features, progressive die high speed stamping is the cost floor. For everything else, one of the other three routes is probably cheaper to start. Our stamped aluminum parts guide walks the forming side of this decision in more detail.

The Clip-On Bracket: Where 6063 Earns Its Keep

A clip-on heat sink lives or dies on the bracket, not the fins. The bracket is spring steel or 6063-T5 aluminum that latches the sink to a TO-220, a DPAK, or a bare die on a PCB. The clamp force has to hold the interface against vibration and thermal cycling without overstressing the component leads.

6063-T5 at ~150 MPa tensile gives you repeatable clip force through 5,000+ mate cycles. The design target is a retention force of 20-40 N per clip for a typical TO-220, low enough to avoid cracking the package, high enough that a 5 g vibration sweep at 2,000 Hz does not walk the sink off. I spec the clip engagement with a 0.15-0.25 mm interference fit at the latch so spring-back does not open a gap.

The bracket also sets the interface pressure. A warped bracket seats on two corners and the thermal pad compresses 0.3 mm in one spot and zero in another. Same air gap problem as the opening. Bracket flatness under load is the number I check first on a returned unit.

The Latch Geometry

The latch is where stamping and thermal engineering meet. If the latch is too stiff, the sink fights the operator and the package; too soft, and it relaxes. Two rules keep it in the window. First, the engagement ramp angle - typically 30-45 degrees - should be drawn so the insertion force stays under 60 N even at the top of the tolerance stack. Second, compensate spring-back in the die by over-bending the latch 1-2 degrees past the print angle, then verify recovered angle on the first article rather than trusting the CAD model.

The CTE mismatch is the part most drawings miss. Aluminum runs about 23.4 ppm/°C against FR-4's 14-16 ppm/°C in-plane, so a 105 °C swing drives roughly 0.8 µm/mm of relative growth between the sink and the board. A rigid bracket transfers that growth straight into the solder joints. The fix is to design the mounting feet as compliant stamped features - a small S-bend or a slotted foot - so the sink can walk relative to the board without prying the joints. That compliance is free in a stamped part and impossible in a cast or machined one.

Thermal Paths and Surface Finishes

The Interface Is Half the Resistance

Engineers obsess over fin area and forget the joint. A bare bolted aluminum-to-aluminum interface runs a contact resistance of 0.05-0.5 °C·cm²/W depending on flatness and pressure. A rough mill finish at low pressure sits near the top of that range. Lapping the surfaces and torquing to spec drops you toward 0.05.

With a thermal interface material (TIM) - grease, pad, or phase-change sheet - you fill the microscopic valleys so air, at 0.026 W/m·K, stops being the bottleneck. A 0.1 mm pad at 3 W/m·K adds only about 0.0033 °C·cm²/W of conduction resistance, far below the bare-contact value it replaces. The TIM is not optional on a stamped sink; it is the bridge across the surface roughness.

TIM	Typical thickness	Conductivity	Best for
Grease	25-75 µm	1-6 W/m·K	Lowest resistance, pump-out risk in vibration
Pad	0.5-3 mm	1-8 W/m·K	Ease of assembly, tolerance absorption
Phase change	25-100 µm	3-8 W/m·K	Dry assembly, wets at operating temperature
Graphite sheet	25-100 µm	10-20 W/m·K in-plane	High pressure, high conductivity, no pump-out

Mounting torque matters. On an M3 screw through a 3 mm 1060 base, 0.5 N·m is enough to flatten a mildly uneven interface without dimpling the base into the fins. Over-torque past 1.0 N·m and you bow the base - back to the air gap.

Anodizing: Insulation You Usually Want

A type II anodize lays down 5-20 µm of aluminum oxide (Al2O3), a material with a conductivity near 30 W/m·K. At 10 µm that layer adds only about 0.003 °C·cm²/W of conduction resistance per surface - negligible in the through-thickness path. What it changes is the surface.

Bare aluminum sits at emissivity ε ≈ 0.05. A black type II or III anodize pushes ε to 0.7-0.9. In a passively cooled enclosure where radiation can be 20-40% of the total heat leaving the sink, that emissivity jump cuts sink temperature 3-6 °C at the same power. For an outdoor LED or a sealed enclosure, anodizing is free cooling.

The catch: the anodize is on the outside, not the interface. Never anodize the mounting face you bolt down. Insulate the joint and you rebuild the very gap you fought to close. Mask the base contact area, or machine it clean after anodizing.

Corrosion and Finish Options

Outdoor sinks need a corrosion story, not just a black finish. Type II anodize plus a salt-spray pass at 240-720 hours is the common baseline for LED and outdoor telecom parts, and our plating line runs selective reel-to-reel gold, silver, tin, and nickel at 2-8 µm with zinc 5-12 µm for the parts that need it, all tested to ASTM B117. For a sink, the usual call is black anodize on the exposed surfaces and bare or lightly chromated metal on the interface. If the environment is truly hostile - coastal salt, chemical washdown - ask for a sealed anodize and verify the corrosion test report against the actual duty cycle rather than a generic data sheet.

Tolerances That Actually Matter

Stamping holds ±0.1 mm on a good progressive die and ±0.05 mm on critical features with shimming; our presses hold ±0.005 mm positioning on the feed, which is what makes multi-row fin arrays repeatable. For a heat sink, the dimensions that move temperature are the fin pitch, the base flatness, and the bracket latch location - in that order.

Fin pitch: hold ±0.05 mm. A drift of 0.2 mm closes a 1.7 mm channel by 12% and stalls flow.
Base flatness: under 0.1 mm across the contact area. Anything more and the TIM cannot equalize.
Bracket latch: ±0.08 mm on engagement depth. Spring-back here is the difference between a clip that holds and one that rattles.

Feature	Standard grade	Precision grade	Why it matters
Fin pitch	±0.1 mm	±0.05 mm	Channel width and airflow
Base flatness	0.15 mm	0.08 mm	TIM contact and thermal resistance
Fin bow (tip-to-tip)	0.4 mm	0.2 mm	Channel blockage
Bracket latch depth	±0.1 mm	±0.05 mm	Retention force
Burr height	≤10% of thickness	≤5% of thickness	Interface seating, sharp-edge cuts

Material thickness variation is the silent variable. Buy 1060 and 6063 to a tight thickness tolerance (±0.02 mm) or the die pressure and spring-back drift with every coil. I reject coils outside that band at incoming inspection - it is cheaper than re-shimming the die mid-run. Every dimension above is measured in our quality lab with CMM and optical measurement, and the flatness and bow numbers are taken on a surface plate under a defined load, not in free air.

Cost Economics: What Drives the Quote

A stamped sink quote has the same four layers as any progressive die part: tooling, material, press time, and finishing. For heat sinks, two of those layers behave unusually - material because aluminum is priced by the kilo and scrap is visible, and finishing because anodize and assembly are real line items, not rounding errors.

Die Cost Is the Fixed Bet

A progressive die for a mid-size sink with formed fins, a flat base, and a clip latch typically lands in the $15,000-60,000 range depending on station count and tolerance class. That die amortizes hard at low volume and nearly disappears at high volume. At 50,000 pieces a year, a $40,000 die adds $0.80 per part; at 400,000 pieces it adds $0.10. The volume commitment on the RFQ is the single biggest lever in the negotiation - not the piece price.

The Volume Break-Even vs Extrusion

Below roughly 20,000-30,000 pieces a year, a stamped sink with a simple cross-section usually loses to an extruded sink with a saw cut, because the extrusion die costs a fraction of the progressive die. Above that, stamping pulls ahead once the part needs brackets, standoffs, or mounting features that would be secondary machining on the extrusion. Run the comparison on total cost per year, not piece price, and remember the stamped part often eliminates a separate bracket purchase and its assembly labor.

Finishing and Secondary Operations

Anodize adds a per-part cost that scales with surface area, and clip or bracket assembly adds labor. Both are easy to forget when you compare quotes. A quote that includes masking of the interface face, a salt-spray test report, and in-house assembly of the clip is a different product than a quote that stops at bare stampings - price them separately and compare like for like.

Layer	Share of quote (typical)	What moves it
Tooling amortization	10-40% at launch volumes	Annual volume, station count
Material	20-35%	Alloy grade, coil utilization, scrap
Press time and labor	20-30%	SPM, strip width, multi-up layout
Finishing (anodize, plating, assembly)	15-30%	Surface area, masking, test requirements

One more cost lever specific to sinks: multi-up layouts. A 650 mm strip can carry several fin arrays side by side, which cuts press time per part dramatically. Ask the stamper how many-up the die is laid out before you compare quotes, because a two-up versus four-up layout changes the piece price more than any percentage haggling. Our metal stamping tooling costs guide breaks down the arithmetic behind these layers in detail.

Failure Autopsies: Three Field Returns

Every failure below is a real return pattern from thermal programs, with the numbers typical of the class. Each one traces to a decision that looked right on the drawing.

Case 1: The Clip That Quietly Let Go

An LED driver returned after 14 months with junction temperature 15-20 °C above the qualification data. The 6063 bracket measured 12 N retention force against a 25 N spec - the clip had relaxed under thermal cycling, and the sink was sitting on the pad with a 0.2 mm gap on one corner. The root cause was a bracket designed at the edge of the temper's stress-relaxation curve and a die that had not been compensated for spring-back, so the latch started its life 0.1 mm open. The fix was a thicker latch section, a T5 temper verification on the material certificate, and a 1.5-degree overbend in the forming station. Retention force is now tested on a pull fixture at incoming inspection instead of assumed.

Case 2: The Bowed Fins That Choked the Fan

A forced-air server sink ran 8 °C hot on the inlet side. The 30 mm fins stamped from 0.4 mm stock had bowed 0.5 mm at the tips, and the bow closed the channel exactly where the fan pushed the air. The drawing called the fin thickness and pitch but said nothing about flatness, so the parts shipped and the airflow died. The fix was 0.5 mm stock, an in-die straightening station, and a bow measurement on the first article and every audit lot. The lesson: if the channel is narrow, flatness belongs on the drawing with a number, not in the supplier's goodwill.

Case 3: The Anodized Interface That Rebuilt the Gap

An outdoor luminaire ran 12 °C hot with everything else in spec. The mounting face had been anodized along with the fins, turning the interface into a 10 µm aluminum oxide layer plus an air gap where the coating bridged the surface roughness. The fix was masking the base before anodize. It cost nothing in piece price and dropped the junction temperature back to the design value. Anodize is free cooling on the fins and free resistance on the interface - the same finish, two opposite jobs.

How to Vet a Heat Sink Supplier

When a supplier quotes a stamped sink, ask for four numbers and do not accept a drawing in place of them: the alloy and temper, the fin pitch and height, the base flatness spec, and the bracket retention force. If they cannot give you contact resistance or at least a measured °C/W at a stated airflow, they have not tested it. A sink with no thermal data is a bracket with hopes.

Thirty years in this trade taught me the failure is rarely the metal. It is the path the heat has to travel - through a bowed fin, across a loose clip, into an air gap at the base. Fix those three and the alloy choice almost takes care of itself.

Qualification Checklist

Material certificate with alloy, temper, and coil thickness band - not a generic "aluminum" cert.
Measured thermal data: °C/W at a stated airflow and mounting pressure, or contact resistance for the interface.
Flatness and bow measurements with the method named (surface plate, CMM, optical).
Retention force test data for clip brackets, with the test fixture described.
Anodize or plating spec including masking of the interface, coating thickness, and salt-spray test report where the duty cycle demands it.
In-house tool room and quality lab - wire EDM, CNC, and grinders for die work; CMM, optical measurement, and online vision for inspection - so die corrections and first articles are not outsourced to whoever is available.
Certifications that match the industry: IATF 16949:2016 for automotive programs, ISO 14001:2015 for environmental compliance, both current and auditable.

On our side, the equipment story is short: 21 presses including Aida high-speed machines up to 300 SPM and 80 tons, with ±0.005 mm feed positioning; a tool room that builds and maintains the dies in house; and a quality lab with CMM, optical measurement, and in-line vision. Plating, assembly, insert molding, and overmolding run under the same roof, which is what lets a sink arrive with its clip assembled, anodized, and tested instead of as a loose stamping. The tool and die design and manufacturing service page covers the die side, and precision stamping tolerances covers the measurement side in more depth.

Spec It Right: FAQ, Related Reading, and Next Steps

FAQ

Can stamped fins reach the density of an extrusion? Not quite. The practical pitch floor for stamped fins is about 2.0-2.5 mm, while extrusion reaches 1.0-1.5 mm. But stamping integrates brackets, clips, standoffs, and mounting features into the same piece, which often beats raw fin density on system temperature because the interface is better.

What volume justifies a progressive die for a heat sink? Roughly 50,000 pieces a year and up, once the part needs formed features. Below that, extrusion or bonded-fin construction usually wins the total-cost comparison unless the integrated features are worth more than the tooling.

Anodize before or after forming? After. Forming after anodize cracks the oxide and exposes bare metal at the bends. Anodize the finished part, mask the interface face, and verify coating thickness on the fins and the mask coverage on the base.

1060 or 6063 for a clip-on LED sink? Usually both: 1060 for the fin stack where conduction matters, 6063-T5 for the bracket that must hold force. If it must be one alloy, 6063 is the safer single choice because a relaxed clip is a worse failure than a slightly lower fin efficiency.

How do I verify a supplier's thermal claim? Ask for measured data with the test setup named: thermocouple placement, ambient temperature, airflow, and mounting torque. A simulated °C/W with no test conditions is a drawing with hopes.

Next Steps

If you are releasing a stamped heat sink to production, send us the power dissipation, the available airflow, and the envelope you have to fit it in. We will tell you the alloy, the fin pitch that actually cools, and the bracket force that stays put - before you cut the first die. Request a quote with your thermal budget and package drawing, and our engineering desk will turn it into a stamped part that holds its number on the test bench and in the field.