IN718 Lives or Dies at 650 °C: Heat Treatment and Grain Size, Not Chemistry, Set the Real Service Limit
By Harris, Technical Engineer, FUSHUN METAL
There is a persistent belief among buyers that once IN718 (the nickel-iron superalloy Inconel 718, UNS N07718) meets its chemistry specification, the material is essentially defined. I want to push back on that, hard. Two IN718 bars with identical chemistry can differ by hundreds of megapascals in useful strength and by orders of magnitude in fatigue life, depending on two things chemistry certificates never show: the precision of the solution-and-age heat treatment, and the grain structure the thermomechanical process left behind. Chemistry is the recipe; heat treatment and grain size are the cooking. Nobody praises a restaurant for its shopping list.
The evidence comes from a 2025 process capability study of aircraft-engine IN718 bar qualified to AMS 5662N. Beyond the headline capability indices, its most instructive content is the heat-treatment discipline and the batch-by-batch tensile data — because 650 °C is where this alloy either earns its place in an engine or fails to.
The Aging Cycle Is a Precision Instrument
IN718’s strength does not come from its matrix. It comes from nanometer-scale precipitates — principally the metastable gamma double-prime phase, a nickel-niobium compound that coherently strains the lattice and blocks dislocation motion. Those precipitates exist only if the heat treatment creates them at the right size, in the right quantity, at the right locations. The studied material followed the classic AMS route, under AMS 2750 pyrometry control:
| Step | Temperature | Hold / rate | Cooling |
|---|---|---|---|
| Solution treatment | 960 °C ± 14 °C | 1 h | Air cool to room temperature |
| First aging step | 718 °C ± 8 °C | 8 h ± 0.25 h | Furnace cool at 56 °C ± 8 °C per hour |
| Second aging step | 621 °C ± 8 °C | 8 h ± 0.25 h | Air cool |
Look at those tolerances. Fourteen degrees on solution. Eight degrees on aging. A furnace-cooling rate specified to within eight degrees per hour. This is why AMS 2750 pyrometry compliance is not bureaucratic decoration: an aging furnace with a lazy 25-degree hot spot will over-age one end of a load and under-age the other, producing certified-chemistry material with quietly divergent properties. My contention is that furnace pyrometry discipline is a stronger predictor of superalloy quality than most line items buyers actually audit.
Solution temperature choice embeds a genuine engineering trade-off. Solutioning near 960 °C preserves a fine grain structure and retains some delta phase to pin grain boundaries — the fatigue-optimized condition used for turbine disks. Solutioning higher dissolves the delta phase and coarsens grain, favoring creep resistance at the expense of fatigue strength. Same chemistry, different metallurgy, different part life. The certificate will not tell you which one you received; the process specification will.
What Correct Cooking Produces: Three Heats, Head and Tail
The study tested three production heats at both head and tail of each Φ200 mm bar — ten room-temperature specimens per heat, sixty tensile tests in total across both temperatures. The batch-by-batch ranges are the real story, because they show not just strength but repeatability:
| Heat | UTS (MPa) | Yield (MPa) | Elongation (%) | Reduction of area (%) |
|---|---|---|---|---|
| 1 | 1,430–1,450 | 1,190–1,210 | 22–23.5 | 41–43 |
| 2 | 1,430–1,450 | 1,189–1,200 | 23–24.5 | 41–43 |
| 3 | 1,433–1,458 | 1,216–1,246 | 23–25 | 41–44 |
| Heat | UTS (MPa) | Yield (MPa) | Elongation (%) | Reduction of area (%) |
|---|---|---|---|---|
| 1 | 1,144.5–1,165.5 | 962.5–1,092 | 31–37.5 | 34–64 |
| 2 | 1,127–1,165.5 | 959–1,050 | 31–37.5 | 36.5–60.5 |
| 3 | 1,151.5–1,183 | 1,022–1,071 | 31.5–40.5 | 41–67 |
Read the room-temperature table again: sixty test values from three separate melting campaigns, and ultimate strength never leaves a 28 MPa band. At 650 °C, every single specimen cleared the 965 MPa ultimate and 841 MPa yield minimums with the weakest result — 1,127 MPa — still carrying 162 MPa of margin. That is not luck three times in a row. That is a furnace and a forge under control.
Grain Size 7.5: A Number Worth More Than Most Certificate Lines
The studied bar showed a uniform ASTM grain size of 7.5, with no segregation banding, no fine-grain streaks, evenly distributed carbides and carbonitrides, and controlled Ni3Nb delta-phase precipitation at grain boundaries. Fine, uniform grain is not an aesthetic preference. Grain boundaries obstruct dislocation motion, so finer grain directly raises yield and tensile strength — the Hall-Petch effect. Finer grain also shortens the effective crack initiation length in fatigue, which is why disk-grade IN718 is deliberately processed for fine structure while blade alloys chase the opposite.
Uniformity matters as much as fineness. A bar with average grain size 7.5 but embedded coarse-grain patches behaves, in fatigue, like a bar of the coarse patch. Averages flatter; extremes kill. The forging sequence in the study — fast-forge cogging followed by radial forging — exists precisely to drive uniform recrystallization through a Φ200 mm cross-section, one of the harder thermomechanical problems in bar production because the center of a thick section sees less deformation than the surface.
The Cliff at 650 °C — and Why Margin Below It Matters
Published property data across the industry shows IN718 holding roughly 1,100 MPa ultimate strength at 600 °C, then falling off a cliff: by 700 °C, strength roughly halves as gamma double-prime coarsens and transforms into the incoherent, non-strengthening delta phase. This is not gradual degradation; it is a phase-stability boundary. The alloy’s entire application envelope — turbine disks, shafts, casings, compressor hardware — is drawn just below that boundary.
This is why I argue that 650 °C tensile capability, not room-temperature results, should anchor supplier evaluation for engine hardware. Room-temperature testing is cheaper and universally reported, but it interrogates the material far from its failure regime. A supplier whose 650 °C properties sit barely above minimum is operating close to the alloy’s physical edge, where minor aging deviations or slight grain coarsening convert directly into scrapped hardware — or worse, into hardware that passes and flies.
There is a broader industry implication. As engine cycles push temperatures upward, some designers abandon IN718 for costlier alloys like 718Plus or Waspaloy. Often that is necessary. But a portion of those substitutions, in my view, compensate for mediocre 718 processing rather than genuine alloy limits. Material with a demonstrated 155 MPa average margin at 650 °C — the kind documented above — extends the confident application envelope of the most economical superalloy in the catalog. Better processing is cheaper than a better alloy.
What This Means When You Qualify a Source
Ask for the heat-treatment specification actually applied, not merely “per AMS 5662,” and confirm AMS 2750 furnace class and instrumentation type. Ask for grain size with its uniformity assessment — a single averaged number across positions conceals exactly what you need revealed. Ask for elevated-temperature tensile data at 650 °C with statistics, not a lone passing coupon. And treat any reluctance to provide these as data in itself.
Chemistry defines what IN718 could be. Heat treatment and grain structure define what your particular bar actually is. The distance between those two things is where engine programs succeed or stumble, and it is entirely within the supplier’s control. Judge suppliers on the cooking, not the shopping list.
