Binder Jetting: Printing Metal and Sand With Ink-Jet Glue

How It Works: Two Machines Pretending to Be One

Binder jetting is really two processes stapled together. First a printer makes a fragile green part; then a furnace turns it into metal (or ceramic). Standardized as BJT under ASTM/ISO 52900, the print cycle is:

• Recoat: a counter-rotating roller or blade spreads a fresh powder layer of thickness t = 35–200 µm across the build box. Layer height must be ≥ 3× the mean particle diameter d₅₀ to spread cleanly — for 15 µm powder, t ≈ 50 µm.
• Print: a piezo or thermal ink-jet head (often a rebadged HP or Xaar array with 1,000–70,000 nozzles) fires binder droplets of 5–80 pL wherever the slice is solid. Firing frequency is 10–50 kHz; carriage speed ~0.3–1 m/s.
• Cure/dry: an IR lamp or heater (~150–200 °C) partially cures the binder in that layer so it survives the next recoat.
• Lower & repeat until the box is full. Unbound powder stays in place and self-supports every overhang.

The green part is then depowdered, debound (thermal or catalytic burn-out of the polymer, typically ramping to 400–600 °C), and sintered near 0.9–1.0·T_m (T_m in kelvin) — about 1,250–1,400 °C for stainless steels. Because nothing melts during printing, there is no melt pool, no residual thermal stress, and no support structure — the three things that dominate laser powder-bed fusion.

The Physics: Capillarity, Sintering, and Shrinkage

Two physics regimes govern the outcome: fluid infiltration during printing, and diffusion during sintering.

Binder infiltration. When a droplet lands, it wicks into the pores by capillary action. The equilibrium is set by the Young–Laplace pressure ΔP = 2γcosθ / r, where γ is the binder surface tension (~30–70 mN/m), θ the contact angle on the powder, and r the effective pore radius (~a few µm). Penetration depth follows the Washburn equation, L = √(γ·r·cosθ·t / 2µ), so lower binder viscosity µ (kept at 5–20 mPa·s for jetting) and finer pores drive deeper, faster wicking. Over-saturation bleeds binder sideways and blurs edges; under-saturation leaves the green part too weak to handle. Saturation S — the fraction of pore volume filled — is the master print knob, typically tuned to 60–90 %.

Sintering. In the furnace, atoms diffuse across particle contacts to reduce total surface energy, growing necks between particles and pulling the whole part together. The driving pressure of a neck scales as γ_sv/r, so finer powder sinters faster and denser — the reason binder jetting favors 5–25 µm feedstock. Densification is thermally activated (Arrhenius, rate ∝ exp(−Q/RT)), which is why a 20–50 °C change in sinter temperature can swing final density by several percent. Isotropic shrinkage follows directly from mass conservation: linear shrinkage ε ≈ 1 − (ρ_green/ρ_sinter)^(1/3). Going from 60 % green to 97 % sintered gives ε ≈ 1 − (0.60/0.97)^(1/3) ≈ 0.15, i.e. ~15 % linear, ~40 % volumetric.

Controlling Variables and the Core Trade-offs

Every binder-jetting knob trades resolution against strength or density:

• Powder size (d₅₀): fine powder (5–15 µm) → higher green + sintered density and finer features, but poor flowability and dangerous handling. Coarse powder (30–50 µm) flows and spreads easily but sinters incompletely. A bimodal blend (fine + coarse) maximizes packing fraction toward the ~74 % geometric limit.
• Binder saturation S: more binder → stronger green part but more edge bleed, more binder to burn out, and more debind porosity/soot. Too little → cracking on depowder.
• Layer thickness t: thin layers → better Z-resolution and density; thick layers → faster builds. Print time scales roughly as build height / t.
• Sinter schedule (T, time, atmosphere): hotter/longer → denser but risks slumping, warping, and grain growth (Hall–Petch says coarser grains lower yield strength). Reducing atmospheres (H₂, vacuum, or N₂) prevent oxidation of steels and remove surface oxides so necks can form.

The signature failure of the sinter step is distortion under gravity: a part 50–65 % dense at 1,300 °C behaves like warm clay, and long unsupported spans sag. Practitioners counter it with setter plates, ceramic sinter-support powder, and CAE tools that pre-compensate geometry — you deliberately print the part distorted the opposite way so it sinters into tolerance.

Sizing and Quantitative Scale

Shrinkage compensation is the design math that matters most. If you need a 50.0 mm sintered bore and expect 17 % linear shrinkage, you scale the CAD by 1/(1−0.17) = 1.205, printing it at 60.24 mm. A 1 % error in predicted shrinkage on a 100 mm part is a 1 mm miss — which is why shrinkage is characterized per powder lot and orientation (parts often shrink 1–3 % more in Z than in XY because of layer-wise binder gradients).

Throughput. Volumetric build rate ≈ (carriage speed × print width × layer thickness × duty). A production head printing at 0.7 m/s over a 0.4 m width at 100 µm layers deposits on the order of 10³–10⁴ cm³/h of occupied bed — one to two orders of magnitude above the 5–100 cm³/h of laser PBF, because ink-jet parallelism beats a single scanning spot. HP quotes its Metal Jet at building tens of thousands of small steel parts per week from one system.

Typical achievable numbers:

• Final density: 96–99.5 % (sinter + optional HIP), or ~90 % as-sintered without HIP.
• Tolerance: ±0.1–0.2 mm or ±0.5–2 % of dimension after sintering — looser than CNC, comparable to MIM.
• Surface roughness Ra: 4–15 µm as-sintered, driven by powder size.
• Minimum feature/wall: ~0.3–1 mm; minimum drainable hole for depowder ~1 mm.

Mechanical properties after full densification approach wrought/MIM values — e.g. sintered 316L at ~500 MPa UTS, 17-4PH heat-treated toward 1,000+ MPa — provided porosity is closed, since residual pores act as stress concentrators and fatigue-crack initiation sites.

Real Applications and Hardware

Binder jetting's economics — cheap powder, no lasers, huge parallel throughput — make it the natural choice wherever you need hundreds to millions of a metal or sand part.

• Sand casting molds & cores (the largest use by volume): ExOne and voxeljet print silica or ceramic sand bonded with furan or phenolic resin at meter-scale (voxeljet's VX4000 has a ~4 × 2 × 1 m box). Foundries print complex cores no core-box can make, feeding conventional aluminum/iron casting. This was binder jetting's first mass application and still dominates the tonnage.
• Metal production parts: HP Metal Jet and Desktop Metal's Production System (single-pass ink-jet) target auto and consumer volumes — 316L and 17-4PH brackets, gears, and hydraulic manifolds. Volkswagen and GKN have run structural steel parts this way.
• Cemented carbide & refractory metals: WC-Co cutting tools and tungsten parts, which are hard to machine and to laser-melt, sinter beautifully from binder-jetted green bodies.
• Full-color polymer/gypsum models: the original 3D Systems/Z-Corp process jets colored binder into gypsum powder for architectural and anatomical models — no metallurgy, no furnace.

Key vendors: ExOne (now Desktop Metal), Desktop Metal, HP (Metal Jet), voxeljet, Digital Metal/Markforged, and GE Additive. Binders range from thermoset polymers to catalyst-triggered systems and inorganic silica sols.

Failure Modes, Limits, and Best Practice

The green-to-sintered journey is where parts die. Watch for:

• Green cracking / crumbling: under-saturated binder or a too-fast depowder. Cure fully and depowder gently (soft brushes, low-pressure air).
• Debind soot & blistering: ramping burn-out too fast traps decomposition gas and blows the part apart. Debind is slow and stepwise (hours), with controlled atmosphere and adequate porosity for gas escape.
• Sinter distortion & slumping: the dominant metal defect. Use setters, ceramic support, wider bases, and simulation-driven pre-distortion. Thick-thin section mismatches shrink at different rates and warp.
• Residual porosity: if pores don't close, fatigue life collapses (pores are stress raisers) and pressure-tightness fails. HIP (hot isostatic pressing, ~100–200 MPa at temperature) or bronze infiltration of steels closes them; infiltration also lets you hit near-full density at lower temperature but yields a composite, not pure alloy.
• Anisotropic shrinkage & carbon pickup: Z ≠ XY shrinkage needs per-axis compensation; leftover binder carbon can alter steel chemistry (a problem, or a deliberate carburizing lever).

Best-practice checklist: characterize shrinkage per powder lot and orientation; design for uniform wall thickness and self-draining internal channels ≥1 mm; add depowder access holes; keep a stable base for sintering; verify density and closed porosity by Archimedes or CT before load-bearing use; specify HIP for fatigue-critical parts. The hard limit to internalize: binder jetting cannot make a finished metal part by itself — the furnace is not optional, and its shrinkage physics, not the printer's resolution, sets your final tolerances.

Binder jetting vs. laser powder-bed fusion (L-PBF) for metal parts

Attribute	Binder Jetting	Laser PBF (SLM/DMLS)
Bonding mechanism	Polymer glue, then furnace sinter	Full melt via 200–1000 W laser
As-printed density	50–65 % (green)	99.5 %+ (net)
Post-process	Debind + sinter (+ HIP/infiltration)	Stress relieve, support removal
Residual stress	None (cold print)	High — needs stress relief
Build rate	1,000–12,000 cm³/h	5–100 cm³/h
Cost per part (volume)	$ (lowest for 100s–1000s)	$$$ (best for 1s–100s)