Superplastic Forming: Stretching Metal to 1000% Elongation Without Tearing

How it works: gas-pressure forming at half the melting point

A superplastic forming cell is deceptively simple. A single fine-grained sheet blank — typically 0.8 to 3 mm thick — is clamped like a diaphragm over the mouth of a one-sided die cavity, inside a heated platen press. The whole assembly sits at the superplastic temperature: about 900 °C for Ti-6Al-4V (≈ 0.6 of its melting point in kelvin) or 480–530 °C for aluminum 5083. Inert argon gas is then admitted above the sheet at a controlled, ramping pressure of roughly 0.3 to 4 MPa (45–600 psi), pushing the metal down into the cavity where it stretches and drapes onto the tool surface.

Because the metal is superplastic, it does not simply stamp — it flows quasi-statically, thinning as it goes, and eventually conforms to every feature of the die. The forming is deliberately slow: strain rates are held between 10⁻⁴ and 10⁻² s⁻¹, so a deep part can take 20 to 120 minutes. In exchange you get shapes impossible by pressing — hemispheres, sharp-cornered boxes, and deep multi-cavity structures — from a single blank, with forming forces small enough that the press only needs to hold the die halves shut, not deform the metal itself.

The governing equation: strain-rate sensitivity m

Superplastic flow obeys a simple power law relating flow stress σ to strain rate ε̇:

σ = K · ε̇ᵐ

where K is a temperature-dependent strength coefficient (Pa·sᵐ) and m is the strain-rate sensitivity exponent. The value of m is everything. For a normal metal at room temperature m ≈ 0.01–0.03; a superplastic material has m ≈ 0.4 to 0.7, and the theoretical ideal is m = 1 (a Newtonian fluid, where σ ∝ ε̇ exactly).

Why does high m suppress tearing? Consider a slight local thinning. The reduced cross-section must carry the same load, so its true stress rises. In a low-m material that hotspot deforms faster, thins further, and runs away into a neck — the classic necking instability. But when m is high, the faster local strain rate raises the local flow stress steeply (σ climbs as ε̇ᵐ), which resists further thinning and diverts deformation to thicker neighboring regions. The consequence, from the Hart/Considère analysis, is that the maximum uniform elongation scales strongly with m — which is exactly why m ≈ 0.5 buys hundreds of percent of stable stretch:

• m = 0.02: uniform elongation ~ 20% — ordinary sheet steel.
• m = 0.5: uniform elongation ~ 300–500% — good SPF alloy.
• m → 1: essentially unbounded, tearing suppressed like taffy.

The mechanism: grain-boundary sliding, not dislocation slip

Conventional plasticity is dislocation glide — atoms shearing along slip planes through the grains. Superplasticity is different: it is dominated by grain-boundary sliding (GBS), in which whole grains slide and roll past one another like ball bearings in a slurry, swapping neighbors so the aggregate elongates while individual grains stay roughly equiaxed. This is why an SPF part can stretch 1000% yet show almost no grain elongation under the microscope — the grains rearrange rather than distort.

Pure sliding would open voids at triple junctions, so it must be accommodated by diffusional and dislocation creep at the boundaries — atoms shuffling by grain-boundary (Coble) and lattice (Nabarro–Herring) diffusion to heal the gaps. That accommodation step is rate-limiting, which is why superplasticity needs three things at once:

• A fine, equiaxed grain size, typically ≤ 10 µm (ideally 2–5 µm). GBS rate scales roughly as ε̇ ∝ 1/dᵖ with grain-size exponent p ≈ 2–3, so halving d can multiply the achievable rate several-fold.
• A high homologous temperature, ≥ 0.5 Tₘ, so grain-boundary diffusion is fast enough to accommodate sliding.
• Grain stability at temperature — the microstructure must resist coarsening, usually via a two-phase structure (α/β titanium, or Al₃Zr dispersoids in aluminum) that pins boundaries during the long hot cycle.

Lose any one — grains coarsen past ~15 µm, temperature drops, or you form too fast — and m collapses, the flow stress spikes, and the metal necks and tears like any hot-worked alloy.

Controlling variables and the central trade-off

The engineer's dials are temperature, pressure history, and time, and they all funnel into keeping the strain rate inside the superplastic window. There is a distinct optimum strain rate — usually 10⁻⁴ to 10⁻³ s⁻¹ for titanium, up to ~10⁻² s⁻¹ for high-rate aluminum grades — where m peaks. Form faster and m falls into the dislocation-creep regime (necking); form slower and you waste hours while grains coarsen.

The key control problem is that as the sheet stretches and thins, the stress needed to keep it at constant strain rate changes, so the applied gas pressure cannot be constant. Modern SPF presses run a pressure-vs-time schedule computed from the target strain rate, typically ramping pressure up as the bubble grows and thins, then holding to "iron out" the final radii. This is the core trade-off table for SPF:

• Rate vs. quality: higher pressure forms faster and cuts cycle time, but overshoot the optimum ε̇ and you lose m, get non-uniform thinning, and risk rupture.
• Temperature vs. tool life & grain growth: hotter lowers flow stress and pressure needs, but accelerates grain coarsening (killing m for the next part) and destroys tooling — dies are pricey Inconel or ceramic-coated steel run at 900 °C in an argon or vacuum atmosphere to avoid α-case oxidation on titanium.
• Thinning uniformity: unlike drawing, there is no blank-holder to feed material in — SPF is nearly pure stretch forming, so wall thickness only ever decreases. Deep or sharp features can thin the wall to 30–50% of the starting gauge; designers pre-thicken blanks or use back-pressure/preforming to steer the thinning.

Sizing the process: pressure, time, and wall thinning

A first-order design uses the free-bulging of a circular diaphragm. For a thin sheet blown into a hemisphere of instantaneous radius R and current thickness t, a membrane (thin-wall pressure-vessel) balance gives the biaxial flow stress:

σ = p·R / (2t)

Combine that with the superplastic constitutive law σ = K·ε̇ᵐ and a target strain rate ε̇, and you can solve for the gas pressure p(t) needed at each instant — which is exactly what the press controller integrates. A worked feel for Ti-6Al-4V at 900 °C, where the superplastic flow stress is only about 5–20 MPa: to hold ε̇ ≈ 5×10⁻⁴ s⁻¹ while forming a part with R ≈ 100 mm and t ≈ 1 mm, p ≈ 2σt/R ≈ (2 × 12 MPa × 1 mm)/100 mm ≈ 0.24 MPa early on, ramping as R and t evolve. The full stroke lands in the 0.3–4 MPa band — orders of magnitude below the 100–400 MPa of hydroforming.

Two sizing rules dominate practice:

• Cycle time scales inversely with strain rate: total true strain ε ≈ ln(final area / initial area); at ε̇ = 5×10⁻⁴ s⁻¹, reaching ε ≈ 1.6 (≈ 400% engineering) takes ~ ε/ε̇ ≈ 3200 s ≈ 53 min. This slowness is the economic ceiling on SPF.
• Thinning follows geometry: for a hemisphere, apex thickness ≈ t₀·(t/R-driven stretch); pole thinning of a full hemisphere leaves ~40–50% of original gauge. Corner-radius zones can hit 25–30%, so minimum-wall and burst are checked against the local biaxial stress, with safety factors of ~1.5–2 on burst pressure.

Real hardware: aerospace shells and SPF/DB titanium

SPF is overwhelmingly an aerospace and defense process, because it wins where geometry is complex, volumes are low, and part-count reduction pays for the long cycle. Classic parts include Airbus and Boeing wing access panels and leading-edge skins, engine nacelle and fan-cowl frames, F-15/F-16 titanium formers, and single-piece exhaust and duct assemblies. Aluminum 5083 SPF (and Al-Li grades) makes complex interior panels, train nose fairings, and architectural cladding at lower temperature (~500 °C) with cheaper tooling.

The signature high-value variant is SPF/DB — superplastic forming combined with diffusion bonding. Titanium sheets are selectively coated with a "stop-off" (yttria ink) where bonding is not wanted, stacked, and pressed at temperature so the uncoated regions diffusion-bond into a single solid via atomic interdiffusion across the clean, oxide-free interfaces. Argon is then blown between the sheets, superplastically inflating the un-bonded areas into a hollow internal truss or sandwich. In one hot cycle this produces stiff, hollow, near-net titanium structures — fan blades, aircraft floor panels, missile fins — that would otherwise need hundreds of fasteners. The payoff is dramatic: SPF/DB assemblies routinely replace 20–50 machined-and-riveted parts with one, cutting weight 20–40% and eliminating fastener fatigue sites.

Failure modes, limits, and best practice

Superplastic forming has real limits, and most defects trace back to violating one of the three enabling conditions. The dominant failure modes:

• Cavitation (internal void growth). Where grain-boundary sliding outruns diffusional accommodation — especially in aluminum alloys — microvoids nucleate and grow, degrading the part's fatigue strength and ductility even if it doesn't burst. It is suppressed by applying a superimposed hydrostatic back-pressure (a few MPa on the die side), which closes voids and is standard practice for SPF aluminum.
• Rupture / burst. Forming faster than the optimum rate drops m into the necking regime; a corner or apex thins locally and blows out. Prevented by strain-rate-controlled pressure schedules and by not exceeding the burst safety factor.
• Excessive/ non-uniform thinning. Because SPF is pure stretch with no material feed-in, deep sharp features can thin below spec. Mitigations: tailored (differential-gauge) blanks, movable-tool preforming, or reversing the bubble against a back-pressure to redistribute strain.
• Grain coarsening and α-case. Long hot dwell coarsens grains and kills m; on titanium, any oxygen forms a brittle oxygen-enriched α-case surface layer that must be chemically milled off. Countered with inert-gas or vacuum atmospheres and by minimizing time at temperature.

Best practice: keep the starting grain size ≤ 10 µm and thermally stable; run pressure from a live strain-rate model, not constant pressure; use back-pressure for aluminum; design for the thinnest wall, not the nominal; and reserve SPF for genuinely complex, part-consolidating, low-to-medium-volume parts — for high-volume shallow panels, conventional stamping is cheaper by two orders of magnitude in cycle time.

Superplastic forming vs. conventional press forming of sheet metal

Attribute	Superplastic forming (SPF)	Conventional deep drawing / stamping
Max useful elongation	200–1000%+ (single stroke)	20–50% before necking
Strain rate	10⁻⁴–10⁻² s⁻¹ (slow)	1–100 s⁻¹ (fast)
Cycle time	5–120 min per part	seconds per stroke
Tooling	One heated die + gas; no matching punch	Matched punch + die + blank-holder
Forming force	Low: gas at 0.3–4 MPa	High: presses of 1–20 MN
Best for	Complex, deep, low-volume aero shells	High-volume shallow automotive panels