Selective Laser Sintering

How it works, layer by layer

An SLS build starts with a sealed chamber filled with inert gas (nitrogen for polymers, argon for reactive metals like titanium) and a thin layer of powder — typically 50–150 μm thick — spread evenly across a horizontal build plate. The chamber is held just below the powder's melt temperature, hot enough that only a small extra dose of energy from the laser is needed to fuse particles together.

A pair of galvanometer mirrors steers the laser beam (CO₂ at 10.6 μm for polymers, fiber at 1.06 μm for metals) across the powder bed, tracing out the cross-section of the part for that layer. Where the beam hits, the powder absorbs the photon energy and either partially melts at the contact points between grains (sintering) or fully melts into a continuous pool that solidifies into a dense solid (melting). The unscanned powder around the part stays loose.

When the layer is done, the build plate lowers by one layer thickness. A recoater blade or roller sweeps a fresh blanket of powder across the bed from a feed cartridge, levelling it to the new top. The galvo scans the next cross-section. The cycle repeats — for a typical part, several thousand times.

The whole build sits inside a powder block called the cake. When the build finishes, the cake cools slowly (sometimes for as long as the build itself, to prevent thermal-shock cracks). Then an operator opens the chamber, brushes and vacuums away the unsintered powder, and pulls the finished part out of what looks like a sandbox.

Why the unmelted powder is the secret weapon

Compare this to FDM. An FDM nozzle deposits molten plastic into open air; an overhang past about 45 degrees has no foundation under it, so the molten material sags. To print overhangs, FDM has to lay down sacrificial support structures — pillars of extra plastic that get broken or dissolved off after printing. Designers spend half their time orienting parts to minimise supports.

SLS has no overhang problem. The part is buried in powder; every overhanging feature, however dramatic, is sitting on a bed of unmelted grains that hold it in place. A horizontal cantilever, a hollow internal channel, a fully enclosed lattice, a printed-in-place hinge with moving parts — all of these emerge from an SLS build with no special preparation. Designers describe this as "complexity-free" geometry: the part doesn't get more expensive to print just because it's complex.

That single advantage is what lets engineers think differently about design for additive manufacturing (DfAM). Topology-optimised brackets that look like organic bone structures, fluid manifolds with conformal cooling channels routed exactly where the heat is, integrated lattice infill for lightweight stiffness — all of it is enabled by SLS's powder-bed self-support.

Metal DMLS is a partial exception: even though the powder bed holds the part up against gravity, the laser melt pool creates large thermal gradients that warp the part if it's not anchored to the build plate by sacrificial supports. So metal SLS still uses supports — but only for thermal/structural anchoring, not for overhang support. The supports are smaller and fewer than FDM equivalents.

SLS, DMLS, SLM — what's the difference?

The three acronyms describe variations of the same process. The distinction is material and degree of melting, not the geometry of the machine.

Name	Material	Melt state	Laser	Typical use
SLS	Polymer powder	Partial sintering	CO₂ (10.6 μm)	Nylon parts, prototypes, end-use plastic
DMLS	Metal powder (EOS term)	Partial-to-full melt	Fiber (1.06 μm)	Aerospace, dental, tooling
SLM	Metal powder (generic)	Full melt pool	Fiber (1.06 μm)	High-density metal parts
EBM	Metal powder	Full melt	Electron beam	Ti-6Al-4V medical, in vacuum
HSS / MJF	Polymer powder + ink	IR-fused via absorber	Halogen lamp	HP MJF, faster than laser SLS

ASTM/ISO 52900 standardises the neutral umbrella term Powder Bed Fusion (PBF), with subcategories PBF-LB (laser beam) and PBF-EB (electron beam). In informal usage "SLS" has become the generic name for any laser-powder-bed process, regardless of whether the material is polymer or metal.

Materials and their quirks

Polymer SLS lives and dies by Nylon-12 (PA-12). Nylon-12 has a wide supercooling window — about 26 °C between melt and recrystallisation — which gives the process the temperature headroom it needs. Most other polymers have tiny supercooling windows that make them impractical for SLS. Variants include glass-filled (PA-12 GF, +35% stiffness), carbon-filled (PA-12 CF, lightweight and stiff), and flame-retardant grades for aerospace. PA-11, polypropylene, TPU elastomers, and high-temperature PEKK round out the polymer catalogue.

Metal DMLS / SLM covers an industrial alloy set:

• 316L stainless steel — corrosion-resistant, the easiest metal to print. Pumps, manifolds, food/medical hardware.
• 17-4PH stainless — precipitation-hardenable, used after solution + ageing for high-strength tooling.
• Ti-6Al-4V (Ti-64) — the workhorse aerospace and medical titanium. Bone implants (its modulus is closer to bone than stainless), spinal cages, aircraft brackets.
• Inconel 625 / 718 — nickel-based superalloys for hot sections of gas turbines; very hard to machine, ideal for printing.
• CoCr (cobalt-chrome) — dental crowns, knee implants, the original GE LEAP fuel nozzle alloy.
• AlSi10Mg aluminium — lightweight, used for housings, brackets, and Bugatti brake calipers.
• Copper alloys — heat exchangers, induction coils, rocket-engine combustion chambers.

Each alloy has its own qualified parameter set (laser power, scan speed, hatch spacing, layer thickness) that machine vendors lock down. Switching alloys without re-qualifying the parameters is how you get unfused porosity or hot cracks.

The GE LEAP fuel nozzle — and why aerospace cared

The reason DMLS broke out of the prototyping bubble in the 2010s is a single part: the fuel nozzle inside the CFM LEAP engine, used on the Boeing 737 MAX and the Airbus A320neo.

The previous generation of fuel nozzles was an assembly of about twenty separately machined and brazed Co-Cr components, with internal cooling channels limited to what could be cast or drilled. GE Aviation redesigned the nozzle as a single DMLS-printed part with internal cooling passages routed in three-dimensional curves no subtractive process could reproduce. The benefits:

• 25% lighter than the brazed assembly.
• Five times more durable in engine testing.
• Twenty brazed joints reduced to zero — every weld was a potential leak path.
• Internal cooling routed exactly where coke buildup needed it, not where a drill could reach.

GE has delivered tens of thousands of LEAP fuel nozzles in production since 2015, making them the first mass-produced AM part flying in a commercial jet engine. The same LEAP engine uses additively printed turbine shrouds and other ceramic-matrix components. The certification path GE forced open — proving that an additively manufactured part can be qualified to FAR Part 33 standards — is the door every other aerospace AM program has walked through since.

Medical implants — titanium that matches bone

Ti-6Al-4V has a Young's modulus of about 110 GPa — closer to bone's 10–30 GPa than the 200 GPa of stainless steel. That matters: when a titanium implant takes too much load, the surrounding bone stress-shields (un-stresses) and atrophies. SLS / EBM-printed titanium implants are routinely designed with porous lattice structures that further reduce the effective stiffness, and the rough printed surface promotes osseointegration (bone growth into the implant).

Production cases:

• Spinal interbody cages. 4WEB Medical, Stryker Tritanium — porous Ti-6Al-4V cages printed with engineered trabecular lattices that fuse to the vertebrae.
• Acetabular cups in hip replacements — Stryker, Smith & Nephew, Zimmer Biomet, all in printed Ti-6Al-4V with porous outer surfaces.
• Custom cranial / mandibular implants. Patient-specific from a CT scan; the same week as the diagnosis.
• Dental crowns and bridges in CoCr or zirconia — millions per year through chairside scanners and centralised printing labs.

The economic argument is unique to medical: a single patient's anatomy is a batch size of one. Subtractive manufacturing has fixed setup cost per part; SLS has near-zero, so the cost per implant doesn't rise when each is unique.

Build rate and the trade-offs that constrain SLS

SLS isn't cheap or fast. Build rates are 10–100 cm³/hour for polymer SLS and 5–80 cm³/hour for metal DMLS — order-of-magnitude slower than CNC milling and far slower than injection moulding. A single small part might take twenty hours in the printer plus another twenty in post-processing.

The economic crossover with CNC milling or moulding is well understood:

Production volume	Best process	Why
1–10 parts	SLS / DMLS	No tooling cost; complex geometry is free
10–1,000 parts	SLS or CNC (case-by-case)	Crossover depends on geometry complexity
1,000–10,000	CNC milling, casting	Tooling amortises; faster cycle times
10,000+	Injection moulding, die casting	Pennies per part once tooling is built

Then there's the part itself. SLS justifies its cost when one or more of these holds:

• The geometry is impossible (or absurdly expensive) any other way — internal channels, lattices, topology-optimised organic shapes.
• Volume is low (under a few thousand) and tooling cost dominates.
• Each part is custom — patient-specific medical, customer-fit consumer goods.
• Lead time matters more than unit cost — spare parts on demand, military forward-deployed printers.
• Consolidating an assembly saves more than the per-part premium — the LEAP nozzle, where eliminating 20 weld joints transformed reliability.

Multi-laser machines (EOS M 400-4, SLM Solutions NXG XII 600) parallelise the scanning with up to twelve lasers operating simultaneously, lifting metal build rates above 100 cm³/hour and bringing per-part costs down by 30–50% on suitable geometries.

Post-processing — most of the cost is after the printer

An SLS part doesn't leave the machine ready to use. The post-process chain depends on whether the part is polymer or metal.

Polymer SLS:

1. Cooldown. The cake cools slowly in the build chamber or a separate cooling station — typically 8–24 hours to avoid thermal stress.
2. Depowdering. Brush, blow, and vacuum the unfused powder away. Recovered powder is mixed with virgin powder (typically 30–50% recycled) for the next build.
3. Bead-blasting. Removes residual surface powder, smooths visible roughness slightly.
4. Optional dyeing, vapor-smoothing, painting. SLS surfaces are matte and slightly porous; dye soaks in well. Vapor-smoothing with solvent dramatically smooths the surface for cosmetic applications.

Metal DMLS / SLM:

1. Stress relief in furnace — typically before separating from the build plate, to prevent the part from warping as anchors are cut.
2. Wire EDM or saw separates the part from the build plate.
3. Support-structure removal by mill, EDM, or hand.
4. Hot isostatic pressing (HIP) — 100–200 MPa argon, ~1000 °C — closes residual sub-surface porosity, lifting density from ~99.5% to >99.9%. Essential for fatigue-critical aerospace parts.
5. Heat treatment / ageing sets the final microstructure (e.g. solutionising + ageing Inconel 718).
6. Finish machining of critical surfaces — bearing fits, sealing surfaces, threads. As-built surface roughness is Ra 6–20 μm, too rough for most engineered fits.
7. Inspection — CT scan for internal voids, dye-pen / mag-particle for surface cracks.

For aerospace, the post-process chain can take longer than the print itself. The printer is one machine in a flow that includes furnaces, HIP, mills, and metrology — none of which "additive manufacturing" marketing slides usually show.

Machines and vendors

• EOS (Germany). Polymer (P series — P 396, P 770) and metal (M series — M 290, M 400-4) industry leader. Founded 1989, the same year Deckard's patent issued.
• 3D Systems (USA). Acquired DTM (Deckard's company) in 2001; polymer SLS (sPro, ProX) and metal (DMP) lines.
• Stratasys (US/Israel). Polymer-focused; recent acquisitions added SLS (Stratasys H350 via HSS).
• GE Additive (Concept Laser). Metal DMLS, used internally for GE Aviation LEAP nozzles and Catalyst engine.
• SLM Solutions (Germany). Multi-laser metal SLM specialist, acquired by Nikon in 2023.
• HP / Multi Jet Fusion (MJF). Not laser-based, but adjacent — uses inkjet-deposited absorber + IR lamps to fuse polymer powder, often grouped with SLS. Faster build rates for nylon-12 production.
• Formlabs Fuse 1. Benchtop polymer SLS that brought powder-bed printing to small shops at ~$30k/machine.

Common pitfalls

• Assuming SLS prints any metal alloy. Only alloys with qualified parameters print well. Pure copper, magnesium, and some aluminium grades are notoriously difficult; refractories require electron-beam or laser-wire variants.
• Ignoring the supercooling window for polymer SLS. If chamber temperature drifts by 5 °C, parts warp, or the powder cake itself sinters into a brick. Process control is everything.
• Treating "complexity-free" as cost-free. Complex geometry is free in the printer, but downstream depowdering, support removal, and inspection costs scale with geometry — internal channels need flushing, lattices need CT scanning.
• Forgetting the build orientation. Anisotropy is real: Z-direction tensile strength is 15–30% lower than XY, and surface roughness is worst on downward-facing skin. Orient the part so loads run XY and critical surfaces face up.
• Skipping HIP on metal fatigue parts. Sub-surface porosity is the dominant fatigue crack initiation site in as-built DMLS parts. HIP is non-negotiable for aerospace structural use.
• Treating recycled polymer powder as identical to virgin. Powder degrades thermally over multiple builds; the refresh ratio (% virgin in each new build) directly controls part properties. Vendor-qualified ratios exist for a reason.