3D Printed Titanium Metal: How the Process Actually Works

• 1
  What 3D printed titanium metal really is
• 2
  Why titanium AM parts warp, and what stress relief does
• 3
  Tolerances and surface finish as-built vs machined
• 4
  Where titanium 3D printing wins on geometry
• 5
  When 3D printed titanium metal is the wrong choice
• 6
  Material quality and the hydrogen question
• 7
  Support removal, machining, and finishing
• 8
  Titanium route comparison by part type
• 9
  Ti-6Al-4V properties: printed, HIPed, and wrought

Most titanium 3D printing is laser powder bed fusion, often sold under the name DMLS or SLM. A recoater arm spreads a layer of gas-atomized Ti-6Al-4V powder, usually 20–63 μm in diameter, across a build plate inside an argon-filled chamber. A fiber laser then traces the part cross-section and melts that layer into the one below. The plate drops by one layer thickness, typically 30–60 μm, and the cycle repeats.

Oxygen is the enemy. Titanium at melt temperature pulls oxygen out of any air it can find, and the result is a brittle alpha case and a part that cracks at the first load. The chamber stays below roughly 100 ppm oxygen during the build, which is why these machines cost what they cost.

The part you pull out is not finished metal. It is near-net shape: attached to the plate, coated in semi-sintered powder, and carrying residual stress from thousands of rapid heating and cooling cycles. Everything after the build decides whether the part is usable.

• 1
  Layer thickness30–60 μm typical for Ti-6Al-4V
• 2
  Powder size20–63 μm gas-atomized, spherical
• 3
  AtmosphereArgon, oxygen kept low to avoid alpha case

Each melt pass cools at something like 10⁴ to 10⁶ K/s. That is fast enough to freeze a metastable martensitic phase, and it leaves a stress gradient through the wall thickness. Thin ribs curl, long flat sections lift at the edges, and tall builds sometimes separate from the support structure entirely.

The fix is thermal, not mechanical. Parts are stress-relieved in vacuum or argon, commonly held in the 600–800 °C range for a couple of hours and then cooled slowly. That relaxes the gradient and converts some of the martensite to alpha plus beta. Skip this step and you will watch a part move 0.2 mm during wire EDM cut-off.

For anything that carries load or sees fatigue cycles, hot isostatic pressing follows. HIP closes internal porosity, which is where crack initiation starts. A HIPed Ti-6Al-4V part behaves much closer to wrought material in fatigue, and it is the usual requirement in aerospace and implant work.

• 1
  WarpingCaused by fast cooling, not by bad powder
• 2
  Stress relief600–800 °C in vacuum or argon
• 3
  HIPRequired for fatigue-critical parts

As-built titanium holds roughly ±0.1 to ±0.2 mm on well-supported features. That is fine for brackets that bolt to slotted holes and useless for a bearing bore or a sealing face. Down-facing surfaces are the worst offenders because they sit on support material and often come out rough and slightly oversized.

Surface finish as-built lands around Ra 8–15 μm. That is a sandpaper-like texture, not a mating surface. Bead blasting drops it to roughly Ra 4–6 μm and removes the loose powder, but it does not make the surface dimensionally accurate.

This is where post-machining earns its place. Add 0.3–0.5 mm of stock on critical faces, then machine them on a 5-axis center. We hold ±0.005 mm on those faces and reach Ra 0.8–1.6 μm, and Ra 0.2–0.8 μm when the drawing calls for it. The printed body keeps its internal channels; only the interfaces get cut.

• 1
  As-built±0.1–0.2 mm, Ra 8–15 μm
• 2
  Bead blastedRa 4–6 μm, cleaner but not accurate
• 3
  Post-machined±0.005 mm, Ra 0.8–1.6 μm

The process earns its cost when the shape cannot be cut. Conformal cooling channels that follow a curved mold surface, lattice blocks that cut weight without losing stiffness, and organic brackets with load paths that a mill simply cannot reach are the classic cases. Titanium makes these attractive because the material is strong and light.

Internal features are the strongest argument. A 4 mm internal channel with a 90° bend inside a solid block is trivial to print and close to impossible to drill. If your part is a manifold, a heat exchanger, or a hydraulic block, that geometry alone can justify the route.

Consolidation matters too. A printed titanium assembly that replaces twelve bolted pieces removes eleven joints, eleven sets of fasteners, and eleven leak paths. On low-volume programs the part count drop often beats the per-part price increase.

• 1
  Conformal channelsFollow the surface instead of straight drilling
• 2
  LatticesStiffness with less mass
• 3
  Part consolidationOne printed body, fewer joints

If the part is a simple prismatic block, a plate with holes, or a turned shaft, printing it wastes money. Titanium bar stock is available, and 3-axis or mill-turn work on TA1, TA2, or TC4 will be cheaper, tighter, and faster. A printed version of a simple bracket can cost several times the machined one with no functional gain.

Size is another boundary. Powder bed machines have a build envelope, and a part that exceeds it must be printed in sections and joined, which adds a weld or a bolted joint and a new failure location. Very large titanium parts usually go to machining or forging instead.

Quantity changes the math as well. At one to fifty pieces, printing avoids tooling and wins on lead time. Past a few thousand pieces, casting or forging plus finish machining usually undercuts it. The crossover depends on geometry, not on a fixed number.

• 1
  Simple shapesMachine them from bar or plate
• 2
  Oversized partsBuild envelope forces splitting and joining
• 3
  High volumeCasting or forging plus machining wins

Powder is not inert. Every build cycle exposes it to heat and moisture, and reused powder picks up oxygen and changes its flow behavior. Most shops screen and blend used powder back in at a controlled ratio, then check chemistry on the finished part rather than trusting the powder certificate alone.

Hydrogen is the quiet problem with titanium. Picked up during processing, it forms hydrides that embrittle the metal, and the effect shows up as delayed cracking rather than an obvious defect. Vacuum annealing pulls hydrogen back down, which is one more reason stress relief is not optional.

For implant and aerospace work, chemistry is verified per lot. We check oxygen, nitrogen, hydrogen, and iron against the Ti-6Al-4V specification, and we keep the powder lot traceable to the parts it produced. Without that chain, a printed titanium part is a claim rather than a documented component.

• 1
  Reused powderScreened and blended at a controlled ratio
• 2
  HydrogenVacuum anneal to avoid hydride embrittlement
• 3
  TraceabilityPowder lot tied to finished parts

Supports come off first, by hand tools, a band saw, or wire EDM for the plate connection. Do not skip stress relief before this step. Cutting a stressed part free releases the stored energy all at once and the part bends in a way no drawing can describe.

Then the critical faces get machined. We leave 0.3–0.5 mm on interfaces, bores, and sealing surfaces, and set up the printed body on a 5-axis center with the same datums used in the CAD model. Holes that need a tolerance class get drilled and reamed rather than printed to size.

Finishing depends on the application. Bead blasting is standard for a clean matte surface. Anodizing in clear, color, or hardcoat works on titanium, though the color range is narrower than on aluminum. Laser marking is available with a minimum character height of 1.5 mm for part numbers and lot codes.

• 1
  Support removalAfter stress relief, never before
• 2
  Machining stock0.3–0.5 mm on critical surfaces
• 3
  MarkingLaser engraving, 1.5 mm minimum character height