Hacker News new | past | comments | ask | show | jobs | submit login
Some alloys don't change size when heated – recent work on why (caltech.edu)
146 points by _Microft on July 27, 2023 | hide | past | favorite | 60 comments



A summary to motivate reading the paper:

Invar, a nickle-iron alloy, was commercially highly relevant for accuracy of mechanical watch balance springs in the 19th century. Investigations of that presumably lead to the 1920 Nobel in physics.

The article claims to produce the first equation to model this effect accurately, together with an experimental technique to validate the main components. This would support in-silico material exploration, esp. predictions for high temperatures that induce expansion.

But because this demonstrates phase shifts in how electrons interact, the significance could be broader that just the use of constant-size invar (iron/nickel alloy).

Paper excerpts:

----

Here we use a thermodynamic Maxwell relation to explicitly separate the contributions to thermal expansion from phonons and spins. [...] These two contributions were measured by nuclear resonant X-ray scattering on Invar under pressure. We find that a competition with phonons is necessary to complete the explanation of the near-zero thermal expansion of Invar.

An advantage to [our] equation is that the two main components of thermal expansion—phonon and magnetic—can be experimentally obtained by nuclear resonant X-ray scattering

Excellent agreement between experiment and theory is found. There is a remarkable spin–lattice coupling, and a precise cancellation of the phonon and spin contributions that causes the anomalously low thermal expansion in Invar near ambient conditions of T and P. Furthermore, the transition to a more typical thermal expansion at higher pressures is shown to arise from the magnetic transition to the paramagnetic state that quenches the negative contribution from the spin system. Finally, the electronic contribution is found to have only a small effect on thermal expansion.


Is 'invar' a class of materials, a specific material, or both? From the OP:

> There is, however, a class of metal alloys called Invars (think invariable), that stubbornly refuse to change in size and density over a large range of temperatures.


It's a small family of the nickel alloys. Lots of iron, as the primary alloying element, then other things. Different invar alloys exist, with different goals of 'invariant'. None are truly invariant, but some will behave better in a thermal band than others. Changing the alloying can get you into kovars (covariant, typically to glass or ceramic) another family. Inconel is yet another family of nickel alloy, favoured for its strength under some conditions. Small percentage alloying changes can have a sizeable impact.


its a small set of nickel iron alloys that are available (to machinists at least) in a few configurations dictated by their iron/nickel content.

with the addition of cobalt it becomes Kovar, a metal that is common for telescope use or joints where metal bonds to glass.

they're all fairly proprietary and expensive. I've used Invar a lot in the production of heated press platens.


Inconel is a name here in the US


Yes, it is a name, but no, is not THE name. Invar and Inconel are completely different alloys with completely different use cases. I have specified them both. In aerospace, Inconel 625 and 718 are used in high temperature structural and transport element applications in aircraft. Invar is never used in aircraft structure, but it is used to make tooling that needs to be invariant in size, especially composite layup mandrels that are cured in autoclave ovens.


My takeaway was that it might allow for “magnetic refrigeration”; two words I don’t think I’ve ever seen together so elegantly summarize the concept.


Totally naive layman's question: If the phonon and spin contributions cancel out, would it be possible to build a material in which the "negative" contributor is larger and which therefore shrinks when heated?


There are some materials like this, it's called negative thermal expansion in materials science, the wikipedia page [0] on it has some examples of materials where this is the case.

[0] https://en.wikipedia.org/wiki/Negative_thermal_expansion


The article somewhat downplays the historical understanding of this. (It's understandable! This is a dense topic!) The impact of spin states has been known for a looong time. My first introduction to the topic was Zener's 1955 paper The Impact of Magnetism on Metallurgy.


It's a very interesting effect, magnetism 'perfectly cancelling' out thermal expansion.

Are there any other cases where magnetism is responsible for something this subtle?


How about the Zeeman effect, in which strong magnetic fields in locations where light is emitted, will cause the spectral lines associated with emitting material to split?

The strength of the magnetic field is encoded in how broadly the line is split, allowing us to make spatially-resolved maps of the magnetic field of the Sun ("magnetograms").

Like getting the chemical composition of the emitting surface of the Sun, it's the kind of thing you'd think sounds impossible until some clever physicist figures out how to exploit it.

See the little animation at the top of the page: https://en.wikipedia.org/wiki/Zeeman_effect


I dont remember what its called but I always liked the magnetic braking thing where if you drop a ferrous cylinder through a copper tube it falls slowly with constant speed, because the magnetism induces current which induces a braking force


Eddy currents


Cool. What would applications be here? Could we for example make basic alloys of vehicles, railroad tracks, and refrigeration systems more resilient when they undergo temperature variations, and therefore last longer?


Well, they knew the effect existed, so applications are already in use; this just explains the how, which I'm sure makes the effect more predictable - and allows for researchers to find more alloys with this effect in a focused manner, instead of via trial and error.

As for applications, it probably won't be garden variety appliances, thermal expansion isn't much of an issue there and designs for all of the things you mentioned have been tweaked a hundred years ago to deal with thermal expansion (although railroad tracks are still an issue sometimes). And of course there's other parameters, like wear resistance; nickel is a pretty soft metal I believe.

But, things like precision industry or space will find a use for this. Sattelites have to deal with hundreds of degrees of temperature variation.


Invar is actually a very bad material for most uses, because having zero CTE usually won't match the rest of your system's nonzero CTE. So things will fail pretty quickly.

Matching CTE is a big deal in careful engineering. "Alloy 42" is a great example of that in action: it's an invar-like alloy with its CTE matched to silicon, so chip lead frames expand with silicon dies as they heat up during operation. Not that many things use lead frames anymore....


What do you mean? MOST ICs use lead frames still. Although there is a large push in consumer electronics to use flip chip or CSP devices, I would suggest that most devices are still using lead frames. Any plastic packaged device is lead frame based.


Sorry, you're right, I was thinking of the newer stuff (flip-chip, wafer-scale, nearly-bare-die, etc). It's also interesting just how few leadframes are magnetic these days, even for something ancient like a SOIC-8. I don't know what they use; I guess it's not very important if there isn't much power dissipation.


I disagree, even modern cars have a lot of unsolved problems with thermal expansion and cracking causing failures of cylinder heads, gaskets, exhaust manifolds, turbochargers, etc.

It may be solved in principle, but certainly not in practice.


There are problems that are game stoppers and there are problems that are unavoidable. Cars work. Yes, if you drive 1 million miles you will have cracks in the engine due to thermal expansion, but who cares? Make a new engine and get over it.

Now consider a space elevator. Material problems are a game stopper.


These problems are already solved for very high end aftermarket exhaust systems, which use Inconel.


"Modern" cars run on electricity and don't have any of those parts or their corresponding problems.


Your "modern" cars have cooling loops for their battery, therefore cyclic thermal fatigue.


Plus the bearings, heat from (regenerative) braking, and I'm sure a few other parts.


I haven't looked it up, but I'm sure the magnitude of thermal cycling in a battery is way less than in a combustion engine.


I think the important bit is that this isnt 'a lot' of systems, it is comparatively few.


I didn't even know that they didn't have a good explanation related to magnets, as invar for example has a curie point, above which it loses it's neutral thermal expansion properties.

I've seen it used for low temp things or tooling but for that reason it can't be used for anything really high temperature.


No need - you ever cross a bridge and note a mesh of metal teeth? Thats for thermal expansion. Its already designed around and isnt generally noticeable - for steel it is 0.0000065 / °F [0].

[0] https://www.metalsales.us.com/thermal-expansion


Would be nice to not have the holes though.


Why is that? I've already put all my knowledge on the topic on display so I can't tell if this is a joke about aesthetics or if there is actually something neat having to do with the gaps


I know from tales of professional cycling that those metal bridges are vastly disproportionately responsible for broken bones.

They're also squirrely when wet, and if you've ever ridden across one on a bike or a motorcycle, they are fucking terrifying because you can see the river below you, and the railings for some reason tend to be very low.

And while I understand that many bridges don't really prevent runoff into the water flowing beneath them, metal mesh bridges really can't.


I'm not sure you're talking about the same thing though, I think they meant these thermal expansion joints, not necessarily a bridge made from metal mesh:

https://i.pinimg.com/originals/4d/82/91/4d8291e022bd14c87985...

I know that picture is a mesh, I thought it had the best detail, but they're on paved bridges too:

https://siamagazin.com/wp-content/uploads/2018/03/23h32h-min...


I think they meant the former, because otherwise having expansion joints that don't expand doesn't really accomplish anything. The concrete still expands, hence the need for the joints.


My reading is that they were saying this expansionless material isn't likely to be useful for bridges, because thermal expansion is well managed by these traditional techniques.

I agree though that both the thermal coefficient of the teeth probably don't impact their performance (the expansion in the teeth relative to the width of the gap is negligible, so we can probably pretend we're using expansionless teeth already), and I've never had an accident on one of these bridges, but I'm sure you're right and that the mesh concentrates the force on your bones like a golf club.


And because this material is 35% nickel. Nickel costs $21/kg.


The whole idea with reinforced concrete is that concrete has thermal expansion very similar to that of steel, thus both steel rebars and the concrete that encloses them move in the same manner.

A rule of the thumb with bridges is that the movements (thermal expansion/contraction + elongation or contraction due to loads + in case of pre-stressed concrete fluage/creep) is in the range of 6-9 mm every 10 m.

Expansion joints on (long) continuous beam bridges can thus need to have very large displacement, up to 1,200-1,400 mm are relatively common.


> and the railings for some reason tend to be very low

This isn't just a problem on bridges. Guardrails in shopping malls seem to come up to thigh height, well below a normal person's center of gravity. I hate it and I can't understand who thought that would be a good idea.

I want guardrails that -- if I should happen to be propelled into them -- will stop me from falling over the side. Not rails that will tip me over headfirst.


And then on a bike it hits you below the knee.


I'm actually just talking about the expansion joints! Like one or two places that are metallic, the rest can be concrete. I havent biked over one but I expect that they are one of the least hazardous of the hazardous terrains - I'm comparing them mentally to pot holes and rail lines (not sure the right for them, street car lines that as it turns out are perfectly shaped to get a bicycle tire stuck in while at speed).

Just between you and me, fuuuuuck those metal grates on bridges. I'm grateful that I have almost always had the opportunity to get onto a good sidewalk/path instead of navigating a bridge with an 'interesting' surface.


Only precision applications can afford the high cost of specialized materials. All the things you mentioned are unrelated, as they are cost driven and steel will always win.


Blackbird SR-72? SR-71 leaks before take off in anticipation of the thermal expansion plugging everything up.

https://nodum.org/was-sr-71-blackbird-leaking-fuel/


The heat generated by drag at such high speeds means that titanium was, at the time, the only suitable material due to is high melting temperature.


Not high melting point but its ability to remain strong at high temperature.


An annealed aircraft is not a particularly safe aircraft, yes.


From the understanding of the mechanisms that cause this invariance, many other applications could be derived, that even don't need the invariance property. For example materials of which cooling systems (e.g. Peltier-like) can be made. Also their measurement methods seem pretty advanced and could be used within other research like in the area of nano technology.


Yes, the main value would probably be in better understanding the temperature / pressure / volume / magnetism relationship within invar-like materials. Perhaps the result will be new piezomagnetic applications, or magnetocooling applications, and so on.


Traditionally, holding optics or some other delicate sensors that we want to stay aligned, or at least point in the same direction consistently.


Tangent: reminded me for some reason of the lengths (ha ha) clock makers went to to account for the expansion and contractions of the clock pendulum. To keep consistent time the length of the pendulum too needed to remain constant. Enter the brilliant John Harrison:

https://en.wikipedia.org/wiki/Gridiron_pendulum


Probably not. You typically cannot change one variable in isolation with an alloy, so while you can gain in one area other things change as well. Strength - both compression and tension, hardness, resistance to bending, springiness, melting temperature, are just a few of the properties (note that the properties have engineering names and common names - I mixed with no concern so there is duplication)


> That anomalous behavior makes these alloys useful in applications where extreme precision is required, such as in the manufacture of parts for clocks, telescopes, and other fine instruments.



Forms and tooling for composites are a big one in the aerospace world. Keeps dimensional stability better through autoclave cycles.


I was going to say something similar but then I started having doubts.

For precision instruments you probably want devices that have exactly the same modulus of expansion as what you are cutting. So that 0.15m is always 0.15m no matter the temperature of the factory.

For molds you would want the outer mold to shrink slower than the molded material, but would you perhaps not want an inner mold to shrink faster? So that the material pulls away from both as it cures/cools (I'm asking, I don't know)?


automatic weapons


who needs to read the article?

> That anomalous behavior makes these alloys useful in applications where extreme precision is required, such as in the manufacture of parts for clocks, telescopes, and other fine instruments.


You can do this with composites too. You can even create materials with slightly negative CTEs


Hot take: If the Earth's core is Iron-Nickel, is it an Invar, and therefore, there's an effect too reduce earthquakes from expansion from heat movement being lower, or is the pressure alone enough to counteract that?


The core of the earth changes temperature very slowly. So any effect is probably pretty minimal.

I would guess that comparatively huge thermal characteristics from the churning and moving of the crust and mantle due to plate tectonics probably overshadows this.


The core of the Earth is also far beyond the temperature/pressure at which this invar effect occurs.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: