​‌​​‌​

Kovar / (1.3981) / UNS K94610 / ASTM F15

​‌​​‌​

Kovar is a nickel-iron-cobalt based alloy with a specifically controlled coefficient of thermal expansion. Within the nickel alloy group it forms, together with Invar 36, the controlled expansion sub-class. Its UNS designation is K94610.

The alloy is formulated to give an expansion coefficient very close to that of borosilicate glass over the 20-400 °C range. That match makes it possible to form a hermetic, gas-tight joint between glass or ceramic and metal that does not crack under thermal cycling. Invar 36 aims at the lowest possible expansion, whereas Kovar aims to match a particular glass — that is the difference between them.​‌​​‌​

This property makes the material indispensable in electronic packaging and vacuum technology. Although the name Kovar is used commercially for most Fe-Ni-Co low expansion alloys, technically only material conforming to ASTM F15 is regarded as true Kovar.

It is used in electronics for transistor, diode and integrated circuit packages, TO-3 / TO-5 metal housings and microwave sealing components; in aerospace for hermetic sub-system enclosures and for the glass-to-metal seals of vacuum and laser tubes; in medical devices for pacemaker and implant bodies; and in optoelectronics for fibre optic connector housings.​‌​​‌​

​‌​​‌​

What Kovar Is — and Why It Is a MATCHED-Expansion Alloy, Not a Low-Expansion One

Kovar (UNS K94610 · W.Nr. 1.3981 · DIN designation FeNi29Co17 — some publishers write FeNi29Co18 · ASTM F15 · Chinese 4J29) is a ternary iron-nickel-cobalt alloy, nominally 29 % nickel – 17 % cobalt – balance iron. This cobalt-bearing Fe-Ni-Co family was historically developed under the name Fernico. Kovar contains no chromium and no molybdenum, is never bought for strength, and is not a corrosion alloy.​‌​​‌​

The one distinguishing sentence is this: Kovar exists not because its thermal expansion is low, but because the SHAPE of its expansion curve is tailored to that of borosilicate glass. Alloys that expand less than Kovar exist — Invar 36 heads the list and expands less than a quarter as much over 20–100 °C. But Invar does not match glass; Kovar does. What a hermetic glass-to-metal transition needs is not low expansion but a metal curve that FOLLOWS the glass curve throughout cooldown.

The second distinguishing fact is cobalt. Cobalt is there not only to shape the expansion curve but to make the oxide interface between glass and metal work: the bond forms through a deliberately grown nickel(II) oxide – cobalt(II) oxide film on the metal surface, and cobalt makes that oxide easier to melt and to dissolve in the molten glass. The proportion of iron oxide in such a seal is low, because cobalt reduces it. In a cobalt-free Fe-Ni alloy this mechanism does not work — the second reason you cannot substitute Invar for Kovar.​‌​​‌​

Trade names — not all produced to the same specification​‌​​‌​

Kovar® is a registered trademark (CRS Holdings / Carpenter Technology). Other trade names for the same alloy: Nilo® K (Special Metals), Pernifer® 2918 (VDM Metals), Dilver® P, Rodar®, Nicoseal®, Telcoseal®, Fenicoloy®, Ulbravar®, Nicosel®. The Chinese equivalent is 4J29. All belong to the K94610 family, but carbon and impurity ceilings — and, more importantly, the as-delivered decarburisation condition — vary from mill to mill. If you are going to seal to glass, write “decarburised in wet hydrogen” separately into the order; the trade name does not guarantee it.

Standards by Product Form — and What Happened in 2024​‌​​‌​

Kovar’s ASTM basis disappeared in 2024 and most distributor pages still do not publish this. Read this section before you quote.​‌​​‌​

What the withdrawal of F15 means in practice​‌​​‌​

The alloy is not banned; the document the certificate rested on is gone. In the same 2024 round ASTM F1684 (low-expansion alloys including Invar) and ASTM F30 (Fe-Ni sealing alloys) were also withdrawn — the entire controlled-expansion family lost its ASTM basis at once. For Kovar this carries particular weight, because F15 was not only a chemistry table: it also carried the expansion acceptance requirements and the hardness/temper requirements.
Three practical consequences. (1) Cite the standard with its edition year in the purchase text: “ASTM F15-04(2022), withdrawn edition”. (2) AMS 7726 / 7727 / 7728 remain in force and are the strongest published specification route left for Kovar — cite the one that matches the form (wire / bar-forging / sheet-strip-plate). (3) Move the work the standard used to do into the order text: chemical ranges, expansion measurement range and acceptance criterion, temper and hardness, and the as-delivered heat treatment condition (decarburised in wet hydrogen or not?).

Temperature Limits — Kovar’s Real Ceiling Is 450 °C, for Two Reasons​‌​​‌​

Kovar’s service limit is not a code limit but the coincidence of two separate physical events. Both occur at roughly the same temperature, and that coincidence defines the whole application window of the alloy.​‌​​‌​

Product Forms With NO Standard — the Commercially Valuable Section​‌​​‌​

After the withdrawal of F15 this list grew. These are the sentences a sales engineer should have memorised.

Chemical Composition — and How C, Mn, Si and S Limits Affect Glass Sealing

Chemistry in Kovar does two separate jobs: it fixes the expansion curve, and it keeps the glass interface clean. The major elements do the first, the impurity ceilings the second.​‌​​‌​

C, Mn, Si and S — why glass sealing hangs on these​‌​​‌​

Carbon: enemy number one. Carbon dissolved in Kovar reacts in the glass-sealing furnace with the surface oxide and with the glass to produce carbon monoxide and carbon dioxide. That gas becomes trapped in the molten glass and at the glass-metal interface as bubbles and blisters. The consequences are: the bonded area shrinks and the joint weakens mechanically; and when blisters coalesce they open a leak path. Leaks seen in hermeticity testing usually originate here. This is why Kovar is DECARBURISED in wet hydrogen before glass sealing — it is not merely cleaned, carbon is physically removed from the alloy. The decarburising anneal is not a cleaning step but a metallurgical operation, and it is MANDATORY for glass-to-metal sealing.

Silicon and aluminium: they corrupt the quality of the oxide film. What the seal bonds to is a deliberately grown NiO–CoO film on the metal surface; the glass wets that film and dissolves it. Silicon and aluminium form stable, stubborn oxides that do not dissolve in glass; those oxides leave locally non-wetting islands within the film and weaken the bond. That is why the Si and Al ceilings are held so low — not for strength.
Manganese: the ceiling is 0.50 %; more than that changes the composition and colour of the oxide film and makes the process harder to control.
Sulphur and phosphorus: they segregate to grain boundaries. Sulphur governs both hot cracking in welds and tearing in deep drawing and stamping; since hermetic package bodies are usually deep drawn, this is directly a yield issue.​‌​​‌​

The one sentence that reaches purchasing: on a Kovar order, write the carbon ceiling and the decarburisation condition separately. “Kovar to ASTM F15” — even if the standard were still in force — does not guarantee that material arrives ready for glass sealing.

Mechanical Properties​‌​​‌​

Kovar is a soft, ductile, deep-drawable alloy and is never bought for strength. Do not mix the two source groups below.​‌​​‌​

Physical Properties​‌​​‌​

Heat Treatment — the Wet Hydrogen Decarburising Anneal and Controlled Oxidation

Heat treatment in Kovar is done for the GLASS INTERFACE, not for strength. This is the section anyone making glass-to-metal seals must read.​‌​​‌​

Welding, Soldering and Plating​‌​​‌​

Machining​‌​​‌​

Kovar is not as difficult as Invar 36, but it is not easy either. It comes from the same family and behaves the same way: it work-hardens like austenitic stainless, produces stringy, gummy chips and wraps them around the tool. The one thing in its favour is a thermal conductivity about 1.7 times that of Invar (17 W/m·K versus 10 W/m·K), so heat leaves the tool tip a little better. Do not raise speeds on the strength of that.

Corrosion — Kovar Is NOT Stainless

Kovar is not a corrosion alloy. It rusts. The fact that it is used in hermetic packages does not change that — those packages are plated.​‌​​‌​

Why it rusts

In Kovar chromium exists only as an impurity ceiling (≤0.20 %), which is nowhere near enough to form a passive oxide film. The bulk of the alloy is iron (~53 %). Nickel and cobalt add some resistance but do not passivate. For corrosion purposes, think of Kovar as a low-alloy steel. That is not what the alloy was designed for anyway: its chemistry is chosen entirely for the expansion curve and the glass interface.​‌​​‌​

Where it is GOOD

It is good when plated and enclosed. The field reality is that almost all Kovar components are used nickel + gold plated, and corrosion resistance comes from the plating, not from the alloy. In vacuum and inside a dry hermetic cavity there is no corrosion problem — keeping the inside dry and inert is the whole point of a hermetic package. Bare Kovar stored in a dry, climate-controlled environment is fine for reasonable periods.​‌​​‌​

WHERE IT FAILS — the list that must be published

1. Bare Kovar plus humid air. Unplated Kovar rusts in a humid environment. Condensation in storage, moisture in transit, sweating in the shop all produce surface rust — and you cannot seal glass to a rusted surface.
2. Seawater, chlorides and salt spray. No resistance whatsoever. Electronic packages in marine environments are protected by plating and enclosure, not by the alloy.
3. Acids. With no chromium and no molybdenum it is unprotected in both reducing and oxidising acids. Kovar is not a process equipment material.
4. Plating pinholes. This is the most insidious failure mode. Micro-pores in the nickel plate create galvanically accelerated local corrosion beneath the gold: gold is the cathode and Kovar the anode, and that couple accelerates corrosion at the bottom of the pore. This is why nickel underplate thickness and freedom from porosity are critical — gold alone is not enough.
5. Fingerprints. Kovar parts destined for glass sealing must not be handled with bare hands. Hand perspiration contains chloride and organic acids; it both pits the surface and prevents the oxide film from forming evenly.
6. Furnace atmosphere upsets. If atmosphere control is lost in the decarburising or oxidising furnace, the surface ends up with an oxide of the wrong thickness or wrong composition; a black (over-oxidised) surface gives a weak bond to glass. That is not corrosion, but it leads to the same outcome — a leak.
7. Long open storage. For unplated Kovar strip and bar, VCI packaging, dry storage and a protective film are mandatory.​‌​​‌​

Be honest about numerical corrosion data: no published corrosion rate table, PREN value or critical pitting temperature for Kovar could be found — and none of those would be meaningful for this alloy anyway. Do not publish a corrosion rate figure.

Frequently Asked Questions​‌​​‌​

Can we skip the decarburising anneal and go straight to glass sealing? Furnace time is expensive.

No. This is the most expensive shortcut in glass-to-metal sealing, and its failures often show up not in the hermeticity test but in the field.
Here is why. Carbon dissolved in Kovar reacts in the sealing furnace with the surface oxide and with the glass to produce carbon monoxide and carbon dioxide. That gas becomes trapped inside the molten glass and precisely at the glass-metal interface: bubbles and blisters form. They have two consequences — the bonded area shrinks and the joint weakens mechanically; and when blisters coalesce they open a leak path along the interface. A part that barely passes a helium leak test drifts outside the limit after a few thermal cycles.
The correct sequence is: finish fabrication → complete degreasing (cutting and drawing oils leave carbon on the surface) → wet hydrogen decarburising anneal (900–1050 °C for 1 hour; another mill says 838–1099 °C, about 2 hours at the low end and 20 minutes at the high end, then cool below 300 °C) → controlled oxidation (650–700 °C until a dark grey to slightly brown film forms) → glass sealing → plating.
The difference between wet and dry hydrogen lies exactly here: dry hydrogen degasses but does not remove carbon. What removes carbon is the water vapour in wet hydrogen oxidising it to CO and CO₂.
Commercial advice: if you want to save furnace time, this is not the step to save it on. Write “delivered decarburised in wet hydrogen” into the order and buy the material in that condition — but remember it must be repeated after fabrication.​‌​​‌​

Our package goes through a furnace that reaches 500 °C. Will Kovar be a problem?

Yes. 450 °C is a genuine red line in this alloy, and for two reasons — both of which occur at about the same temperature.
The first is in the metal: the inflection point of Kovar’s expansion curve is 450 °C and the Curie temperature is 435 °C. Above these points expansion climbs rapidly: the 25–450 °C mean is 5.26, but at 25–500 °C it is 6.14, at 25–600 °C 7.81 and at 25–700 °C 9.11 × 10⁻⁶/K. In a cycle reaching 500 °C the metal expands more than the glass expects.
The second is in the glass: the strain point of 7052 borosilicate glass is 435–440 °C. Above that the glass is no longer elastic and relieves stress permanently. When the part cools, the joint is left with a residual stress distribution different from the one that was designed. The design holds the glass in slight compression; when that balance is disturbed, the glass can be driven into tension — and glass is weak in tension.
Practical outcome: a 500 °C process cycle may not break a hermetic Kovar–borosilicate transition in one pass, but it raises the probability of a leak with cycle count. What to do: bring the process temperature below 450 °C; or replace the transition with a ceramic-to-metal solution (alumina with an active metal braze); or change the assembly sequence so the glass feedthrough stays out of the furnace cycle. What not to do is reason that “Kovar melts at 1450 °C” and treat 500 °C as safe. The melting point is completely irrelevant here.​‌​​‌​

Our datasheet says Corning 7052. Our supplier says they cannot get 7052. Now what?

Your supplier is right. Corning 7052 has been discontinued since 1991. The published succession is: 7052 → Corning 7056 → Corning 7720 → today Schott 8250. When you ask for “Kovar sealing glass” today, what you actually get is Schott 8250 or an equivalent.
This does not make old specifications worthless — but you must rewrite the specification against the glass PROPERTY, not the glass brand. The numbers to hold on to, for 7052: 46–47 × 10⁻⁷/°C over 0–300 °C (i.e. 4.6–4.7 × 10⁻⁶/K), 53 × 10⁻⁷/°C from 25 °C to the setting point, strain point 435–440 °C, annealing point 480–484 °C, softening point 710–712 °C, density 2.27 g/cm³. Ask how close the new glass is to those values.
Why it matters: Kovar over the same ranges gives 5.13 and 5.26 × 10⁻⁶/K. The metal contracts a few percent more than the glass, and that difference keeps the glass in compression — which is the design principle of a good glass-to-metal seal. If the new glass expands appreciably differently, that balance is disturbed and the transition must be requalified.
What to do: write the specification as “borosilicate sealing glass matched to Kovar (K94610), expansion 4.6–4.7 × 10⁻⁶/K over 0–300 °C, strain point ≥435 °C” and give the brand name as an example. Any document that makes a brand name the specification eventually goes out of supply.​‌​​‌​

Invar 36 expands less and costs less. Can we use it instead of Kovar?

No — and “expands less” is precisely the wrong part of that sentence.
What a hermetic glass-to-metal transition needs is not low expansion but MATCHED expansion. Invar 36 gives 1.5 × 10⁻⁶/K over 20–100 °C; borosilicate glass over the same region gives 4.6–4.7. The difference is more than threefold. In a glass-to-metal seal that means a mismatch accumulating through cooldown that puts the glass into tension — and glass is weak in tension and cracks.
The second and less known difference is the SHAPE of the curve. Kovar’s coefficient is 5.85 over 25–100 °C, falls to 4.90 over 25–350 °C, then rises again to 5.26 over 25–450 °C. That dip is not an accident; it is engineered to follow the glass’s own curve. Invar’s curve, by contrast, climbs continuously above 220 °C (5.5 over 20–300 °C, 10.1 over 20–500 °C). The two curves are nowhere parallel.
The third difference is cobalt. The seal bonds through a nickel-cobalt oxide film grown on the metal surface, and cobalt makes that oxide easier to melt and to dissolve in the glass. Invar has no cobalt; its oxide is largely iron oxide and does not behave the same way in glass.
In short: buy Invar 36 for dimensional stability — optical benches, metrology frames, composite tooling, LNG membranes. Buy Kovar for a hermetic glass-to-metal transition. The price difference is trivial next to the cost of a single leaking package.​‌​​‌​

Common Datasheet Errors — Check Before You Order

1. Missing the withdrawal of ASTM F15. F15 was withdrawn in 2024 with no replacement — and F1684 and F30 went with it. Most pages that still say “conforms to ASTM F15” do not know this. Cite the edition year, and know that AMS 7726/7727/7728 are still in force.
2. Not separating the AMS numbers by form. 7726 = WIRE, 7727 = BAR and FORGINGS, 7728 = SHEET, STRIP and PLATE. Listing all three as “the Kovar specification” leads to a wire specification being written onto a plate order.
3. Skipping the decarburising anneal, or mistaking it for a cleaning step. Wet hydrogen decarburisation is a MANDATORY metallurgical operation for glass-to-metal sealing, not degreasing. Skip it and carbon produces CO/CO₂, leaves blisters at the interface and hermeticity is lost. Most distributor pages never mention it at all.
4. Confusing dry hydrogen with wet hydrogen. Dry hydrogen degasses; it does NOT remove carbon. What decarburises is the water vapour in wet hydrogen.
5. Writing the operation sequence backwards. The correct order is fabricate → degrease → decarburise → controlled oxidation → glass seal → plate. Glass is not sealed to a plated surface; what the glass bonds to is the nickel-cobalt oxide film, not the nickel plate.
6. Confusing the Curie temperature with the inflection point. Curie 435 °C, inflection 450 °C. One distributor writes 425 °C for Curie. The design ceiling is 450 °C, and that temperature also sits just above the strain point of 7052 glass (435–440 °C) — two separate reasons, one line.
7. Publishing a single expansion number. In Kovar expansion is a curve that falls, then rises: 25–100 °C 5.85 · 25–350 °C 4.90 (minimum) · 25–450 °C 5.26 · 25–500 °C 6.14 · 25–900 °C 11.25. The entire value of the alloy is in the SHAPE of that curve; a single number destroys it.
8. Assuming Corning 7052 is still made. 7052 has been discontinued since 1991. The succession is 7056 → 7720 → Schott 8250. Tie the specification to expansion and strain point values, not to a brand name.
9. Density conflict. 8.36 g/cm³ is the majority and mill value. One manufacturer’s bulletin gives 8.16, one distributor 8.25. If you price by weight, state which density you used.
10. Resistivity conflict. 49 µΩ·cm is confirmed by two independent sources (294 ohm-cir-mil/ft is the same number). One manufacturer’s bulletin gives 43 µΩ·cm. State which you used in RF and power calculations.
11. DIN name confusion. FeNi29Co17 and FeNi29Co18 are the same alloy; the difference is which end of the cobalt band (17.0–18.0 %) went into the name. The material number is 1.3981 in both cases.
12. Material number confusion. 1.3981 = Kovar, 1.3912 = Invar 36, 1.3917 = alloy 42. All three are called “low-expansion Fe-Ni” and all three do entirely different jobs.
13. Confusion with “Invar 42”. Alloy 42 (K94100) is sometimes sold as “Invar 42”. It is neither Invar 36 nor Kovar: Fe–42 Ni, inflection 370 °C, matched to silicon and alumina. Read the UNS number on the certificate, not the trade name.
14. Treating the chemistry table as the only acceptance criterion. The specification’s own note reads: the iron, nickel and cobalt percentages are NOMINAL and are adjusted by the manufacturer so that the alloy meets the expansion requirements. So the governing acceptance criterion is EXPANSION. With F15 withdrawn, you must write the expansion acceptance band into the order yourself; otherwise the only property that matters is never measured.
15. Omitting the carbon ceiling, or hiding the conflict in it. Published ceilings range over ≤0.02 · ≤0.03 · ≤0.04 · ≤0.05. If you are sealing to glass this difference matters; demand the lowest ceiling and write it into the order.
16. Assuming Kovar is “stainless”. Chromium is only a ≤0.20 % impurity ceiling; no passive film forms and bare Kovar rusts in a humid environment. Field corrosion resistance comes from nickel + gold plating, not from the alloy.
17. Treating nickel plating as cosmetic. The nickel underplate does three jobs at once: corrosion barrier, solderability, and a pore-free base for the gold top layer. Gold alone is not enough: micro-pores in the nickel accelerate local corrosion under the gold–Kovar galvanic couple.
18. Omitting the magnetic behaviour. Kovar is ferromagnetic below 435 °C and is a constraint for magnetically sensitive circuits. Do not publish a relative permeability (µr) figure — no verifiable value could be found.
19. Mistaking the melting point for a safety margin. 1450 °C is correct and completely irrelevant. Kovar’s service ceiling is 450 °C, set by the glass match, not by melting.
20. Specifying elongation without a gauge length. Published elongation values range over 25 %, 30 % and 42 %; most of the spread comes from gauge length and product form. Do not impose an elongation requirement without stating the gauge length.​‌​​‌​

​‌​​‌​

​‌​​‌​