3D Printing on Copper: How Each Process Actually Works

• 1
  Why copper is hard to print: absorptivity and heat flow
• 2
  Laser powder bed fusion: green lasers, IR lasers, and alloys
• 3
  Binder jetting, DED, and material extrusion compared
• 4
  What copper printing cannot fix: tolerance, finish, and cost
• 5
  Print then machine: the practical production path
• 6
  Copper additive processes at a glance

Pure copper reflects roughly 95% of near-infrared laser light at room temperature. A 1,070 nm fiber laser, the standard source on most metal printers, therefore delivers only a small fraction of its energy into the powder bed. Without enough absorbed energy the powder does not fully melt, and the track breaks into balls instead of a continuous line.

The second problem is thermal. Copper conducts heat at about 400 W/m·K, roughly twenty times better than titanium or stainless steel. Heat leaves the melt pool almost as fast as the laser adds it. The pool stays small, viscosity stays high, and the molten track tends to pull apart. Thin walls and fine features suffer most.

Oxidation compounds both effects. Copper forms a surface oxide above roughly 200 °C, and that oxide layer absorbs light differently from the clean metal underneath. In an argon build chamber with residual oxygen above a few hundred parts per million, layer-to-layer wetting gets inconsistent, and you see porosity clustered near the part surface.

The usual fix is not one change but several together: a shorter wavelength laser, a tighter spot, a hotter build plate, and a scan strategy that keeps the pool alive long enough to wet the previous layer. Every one of those choices has a cost, which is why copper printing remains a specialty process rather than a shop default.

Laser powder bed fusion (LPBF) spreads a layer of gas-atomized copper powder 20–50 μm thick, melts a cross-section with a scanning laser, then repeats. With a green laser at 515 nm, copper absorptivity rises by roughly an order of magnitude compared with 1,070 nm. That single change makes dense pure copper practical, with relative density above 99% on well-tuned machines.

Most installed LPBF capacity still runs infrared fiber lasers. On those machines, pure copper is difficult and alloying is the workaround. CuCrZr, CuNi2SiCr, and similar precipitation-hardening grades absorb better, flow better, and hold dimensional accuracy. Conductivity drops to roughly 80% IACS in the as-built state and climbs after aging, which is often acceptable for mold inserts and heat sinks.

Typical as-built LPBF copper lands at Ra 8–15 μm and needs support structures under overhangs beyond about 45°. Support removal is manual work on internal channels. Stress relief is usually required before the part is cut from the plate, and the plate itself is often the same alloy to keep the thermal match.

Where LPBF earns its place: conformal cooling channels inside injection mold cores, cold plates with dense pin arrays, induction coil geometries, and RF cavities where the internal surface cannot be reached by a cutter. Where it does not: flat plates, simple bushings, and anything that fits a 3-axis mill in one setup.

Binder jetting deposits a liquid binder into a copper powder bed with an inkjet head, then sinters the green part in a furnace. There is no melt pool, so laser absorptivity is irrelevant and the process scales to large batches. The trade-off is shrinkage: copper sinters with 15–20% linear contraction, and you must compensate the CAD model before the build.

Sintered binder-jetted copper reaches 90–95% density in a typical cycle. That is fine for thermal spreaders and busbars but leaves closed porosity that hurts fatigue life and vacuum sealing. A second step, such as bronze infiltration, fills the pores and raises density, at the cost of lowering conductivity to roughly 40–60% IACS.

Directed energy deposition (DED) feeds copper wire or powder into a laser or electron beam and builds near-net shapes at high deposition rates. It suits large, simple forms: rocket nozzle liners, busbar transitions, and repair of worn copper tooling. Surface finish is coarse, usually Ra 15–30 μm before machining, so DED parts almost always need a finishing cut.

Material extrusion of copper filament is a different animal. The filament is copper particles in a polymer binder. After printing you debind and sinter, and the final part shrinks by around 20%. It works for small, low-stress parts and for teaching, but porosity and dimensional spread are far wider than any metal powder route. Do not specify it for a production thermal part.

Powder bed processes hold roughly ±0.1 mm on small features and worse as the part grows. That is an order of magnitude looser than what a mill achieves. If a drawing calls for ±0.005 mm on a bore, the printed blank is a starting point, not the finished part. You print oversize and cut the critical surfaces.

As-built surfaces are rough. LPBF copper comes out around Ra 8–15 μm, and internal channels are rougher still because supports and partially melted powder cling to downward-facing surfaces. A channel that carries coolant loses pressure to that roughness. Machining, abrasive flow, or electropolishing is often needed before the part performs as designed.

Cost follows machine time and atmosphere control. Green-laser LPBF runs on a small installed base, so capacity is scarce and hourly rates are high. Binder jetting is cheaper per part at volume but needs sintering furnaces and shrinkage compensation work up front. For a run of 50 simple bushings, none of this competes with bar stock on a lathe.

The honest rule: choose additive when the geometry is impossible to cut, not when it is merely inconvenient. Internal conformal channels, lattice heat exchangers, and topology-optimized brackets qualify. A copper electrode with two drilled holes does not.

Most production copper parts that use additive are hybrid. The printer creates the near-net shape with the internal features, and CNC machining brings the functional surfaces into tolerance. This split plays to what each process does well and avoids paying additive rates for features a cutter handles in seconds.

A typical sequence: print with 0.5–1.0 mm stock on all machined faces, stress relieve, cut from the plate, then face and bore on a 3-axis or 4-axis mill. Datum features are established in the first machining setup, so the printed geometry is located from real cut surfaces rather than from the print bed. That removes the bed-to-part variation from the stack.

For CuCrZr mold inserts, we usually print the core with conformal cooling, then machine the parting line, the core face, and the ejector bores. The thermal benefit comes from the printed channels; the dimensional accuracy comes from the mill. Neither process alone delivers both.

Copper also machines well, which makes the hybrid route cheap on the subtractive side. C101 and C110 cut freely at high surface speed, and C36000 brass is easier still. If your part is mostly prismatic with one internal channel, starting from solid bar and gun-drilling may beat printing outright. We compare both routes in the DFM review before quoting.