Controlled-Expansion Nickel-Iron Alloys
Controlled-expansion alloys are iron-nickel (and iron-nickel-cobalt) compositions whose thermal expansion is set by the nickel content: at 36 % nickel the coefficient collapses to about 1.3 µm/m per °C (Invar 36 — a tenth of steel), at 29 % nickel and 17 % cobalt it matches borosilicate glass and alumina at about 5 µm/m per °C (Kovar), and at 42–52 % nickel it tracks soft glasses and ceramics for sealed leads (Alloy 42, Alloy 48, Alloy 52). They are soft (240–350 MPa yield annealed), magnetic, weldable and machinable, and they are specified for dimensional stability, glass sealing, magnetic shielding or constant elastic modulus — never for strength or corrosion resistance.
Popular controlled-expansion alloy grades
Compare in the explorer| Grade | Yield, MPa | Tensile, MPa | Density, g/cm³ | Machinability | Weldability | Best for |
|---|---|---|---|---|---|---|
| Invar 36 | 280 | 490 | 8.1 | Poor | Very good | About 1.3 µm/m per °C; metrology, optics, layup tooling |
| Kovar | 345 | 520 | 8.36 | Poor | Very good | Matches borosilicate glass and alumina; hermetic packages |
| Alloy 42 | 270 | 520 | 8.12 | Poor | Very good | Lead frames, soft-glass seals, bimetal low side |
| Alloy 48 | 200 | 500 | 8.25 | Poor | Good | Seals to soda-lime glass; lamp and relay leads |
| Alloy 52 | 280 | 520 | 8.3 | Poor | Very good | Seals to soft glass; reed-switch and telecom leads |
| Ni-Span-C 902 | 310 | 690 | 8.05 | Poor | Good | Constant-modulus springs, Bourdon tubes, diaphragms |
| Mu-metal | 250 | 550 | 8.74 | Poor | Good | Magnetic shielding after a final hydrogen anneal |
Choose controlled-expansion alloy for
Optical and laser benches, interferometer and lithography frames, telescope and satellite structure, metrology gauges, carbon-fiber layup molds, hermetic packages and feedthroughs, glass-to-metal and ceramic-to-metal seals, lead frames, thermostat bimetals, precision springs and pressure elements, and magnetic shields — parts whose size, seal or spring rate must not change with temperature.
Look elsewhere when
The part carries real load or sees corrosion (Invar rusts like a low-alloy steel — 17-4 PH or 316L for anything wet), the service temperature runs past about 200 °C where Invar 36 loses its low expansion (a ceramic or a titanium alloy), stiffness per weight matters more than expansion (carbon-fiber composite or beryllium), or thermal growth only has to be predictable, in which case 304 or 6061 aluminum with a calculated allowance is cheaper. For a seal to soda-lime glass Kovar is wrong; that is the job of Alloy 48 and Alloy 52.
Loading the library…
Controlled-expansion alloy by application
- Optics and metrology
- Invar 36 for benches, mirror mounts, laser resonator spacers, gauge blocks and CMM fixtures, machined then stabilised; the coefficient stays within about 1.5 µm/m per °C between −50 and 100 °C.
- Hermetic packages and seals
- Kovar for headers, lids, pins and housings glass-sealed with 7052-type borosilicate or brazed to alumina; Alloy 42 for lead frames and matched seals to soft glass; Alloy 48 and Alloy 52 for leads through soda-lime and lead glass.
- Composite layup tooling
- Invar 36 molds for autoclave-cured carbon-fiber parts, because it expands like the laminate (1–3 µm/m per °C) rather than the 11–12 of steel, so the cured part comes off to size at room temperature.
- Springs and shields
- Ni-Span-C 902 for Bourdon tubes, diaphragms, hairsprings and resonators whose rate must not drift with temperature; Mu-metal for shields, cans and shielded cable braid, formed first and hydrogen-annealed last.
Working with controlled-expansion alloy
- Machining
- Soft, gummy and work-hardening — Invar 36 and Kovar cut like a slow 316 with a tendency to tear and burr; sharp positive carbide, low speed, heavy feed, sulfurised oil and rigid setups. Machining stress shifts the expansion coefficient of Invar, so rough, stabilise, finish, and end with a low-temperature stabilisation for parts measured in micrometres.
- Heat treatment
- Invar 36 is stabilised after machining — about 830 °C with a water quench, 315 °C for 1 h, then 95 °C for 48 h — to lock in the coefficient and dimensions. Kovar is hydrogen-annealed at 900–1,100 °C to decarburise before glass sealing (ASTM F15). Mu-metal is hydrogen-annealed at 1,100–1,180 °C after all forming and must not be bent or struck afterwards. Ni-Span-C 902 is aged at about 650 °C after cold work.
- Welding
- Invar 36 welds by TIG with matching filler or by laser and electron beam, with hot cracking a risk in constrained joints — low heat input and low sulfur matter. Kovar welds by TIG, laser and resistance methods and brazes to glass and ceramic with silver-copper or gold-tin alloys; Alloy 42 and Alloy 52 are usually brazed or resistance-welded to leads.
- Forming
- The whole group forms, spins and deep-draws easily annealed, with less springback than austenitic stainless. Kovar draws into cans and lids; Mu-metal folds and rolls into shields; Invar 36 bends to tight radii but is stabilised afterwards. Alloy 42 and Alloy 52 strip is etched and stamped for lead frames.
Other nickel alloy series
Controlled-Expansion Nickel-Iron Alloys — frequently asked
What is controlled-expansion alloy?
Controlled-expansion alloys are iron-nickel (and iron-nickel-cobalt) compositions whose thermal expansion is set by the nickel content: at 36 % nickel the coefficient collapses to about 1.3 µm/m per °C (Invar 36 — a tenth of steel), at 29 % nickel and 17 % cobalt it matches borosilicate glass and alumina at about 5 µm/m per °C (Kovar), and at 42–52 % nickel it tracks soft glasses and ceramics for sealed leads (Alloy 42, Alloy 48, Alloy 52). They are soft (240–350 MPa yield annealed), magnetic, weldable and machinable, and they are specified for dimensional stability, glass sealing, magnetic shielding or constant elastic modulus — never for strength or corrosion resistance.
Which grade of controlled-expansion alloy should I use?
Invar 36 for anything that must not move with temperature between about −50 and 150 °C: optical benches, laser and lithography frames, metrology fixtures, composite layup tooling and waveguides. Kovar for hermetic electronic packages, headers, lids and pins sealed to borosilicate (Corning 7052) glass or alumina — its expansion curve follows the glass up to the 450 °C set point, and it is the ASTM F15 grade. Alloy 42 is the lead-frame and soft-glass sealing alloy and the low-expansion side of thermostat bimetals; Alloy 48 and Alloy 52 seal to soda-lime and lead glasses for lamp, reed-switch and relay leads. Ni-Span-C 902 is the constant-modulus alloy — its stiffness barely changes from −45 to 65 °C and it age-hardens to about 1,000 MPa — for Bourdon tubes, diaphragms and hairsprings. Mu-metal (about 80 % nickel with molybdenum) is the magnetic-shielding alloy, useful only after a final hydrogen anneal. Chromel and Alumel (type K thermocouple legs) and NiCr80/20 heating wire sit here by chemistry rather than by job.
Which grade of controlled-expansion alloy is the strongest?
ERNiFeCr-2 has the highest yield strength in the series at 550 MPa; ERNiFeCr-1 follows at 540 MPa. Values are for each grade's default condition.
Which grade of controlled-expansion alloy is easiest to machine?
Kovar (machinability poor, 32/100), then Alloy 48 (30/100).
Invar or Kovar?
Invar 36 for the lowest expansion — about 1.3 µm/m per °C up to 100 °C, so a metre of it grows a few micrometres over a lab day — and therefore for metrology, optics and tooling. Kovar expands about four times as much, roughly 5 µm/m per °C, but that curve tracks borosilicate glass and alumina from room temperature to the 450 °C glass set point, which is what a crack-free hermetic seal needs; Invar sealed to the same glass would shatter it on cooling. Kovar is also the better machining and plating base.
Why did my Invar part change size after machining?
Three usual causes. Residual machining stress relaxes over weeks and moves the part by micrometres — stabilise after roughing and again after finishing. Invar 36 keeps its low coefficient only to about 100–150 °C; by 230 °C (the Curie point) it expands like ordinary steel, so a part heated in a coating or bonding step may not have returned to size. And the coefficient depends on cold-work and thermal history, so a bar that skipped the 830 / 315 / 95 °C stabilising cycle can sit at 2 µm/m per °C rather than 1.3.