DP_S2012_11

Deformation Processing Spring 2012 Session · 11 sections 8 cases · Watch on YouTube ↗ all files
Layer 3 — readable edition

§1. Single-crystal tensile tests and texture [00:03]

§1.p1

We talked about slipping on different crystallographic planes, and how that leads to anisotropy. I came across this in one of the books I was looking at. If you take a bar that is a single crystal and you pull on it, without any bending, it would just start shearing on certain crystallographically preferred planes, and you would actually get a translation off to the side. This is just a demonstration of that. In fact, when you actually pull on a single-crystal bar, you pull in the same tensile direction because of the stiffness of the machine pulling it, and that requires that you also get a bending as well as a shear. You can see that here — it's just a nice drawing to show that even a simple tensile test is not so simple.

§1.p2

I remember Backofen saying once when I was a student — and he'd been looking at tensile tests all his life — he'd concluded that the tensile test, even though it seemingly was the most simplistic test you could imagine, was actually extremely complex when you got into the details. We're going to go through some of that a little bit later. Here's another picture from Hertzberg's book: a Charpy bar that fractured in a steel. The texture we develop because of anisotropic behavior gives you a fiber orientation, just like wood is anisotropic. You can see the fibers in this particular piece of steel.

§2. Superplasticity and SPF/DB [01:57]

§2.p1

We've been talking about sheet metal properties, and we're starting to talk about superplasticity. We have n values, R values, and M values. n is the work hardening coefficient, R is the anisotropy ratio, and M is the strain rate sensitivity. We looked at the forming limit diagrams for certain types of metals. When the biaxial strain in the transverse direction E2 is zero, you might have 30 or 40 or 50% elongation in a regular tensile bar. But in a superplastic bar, you can have 100%, 250%, 300%. When you start stretching or compressing in the transverse direction, you actually can get even more strain. That allows you to do all kinds of cheap metal forming operations when it's superplastic — and it also helps you even when the material is not superplastic.

§2.p2

So to review a little bit of superplasticity: superplastic forming, diffusion bonding. They call it SPF/DB — superplastic form, diffusion bonded. The Air Force put a lot of money into this after they realized they could form titanium this way. You basically form two sheets and weld them together in a hermetic seal. You put them in a die cavity, and since it's hermetically sealed, you put pressure in between the sheets as you heat the whole thing up. It might take ten or twenty minutes, it might take an hour or two depending on the temperature and the alloy you're using. Your dies might be operating at 800 or 900° Centigrade. And you can form a very complex part.

§2.p3

The bond — this is one of the diffusion bonds between those areas — you end up getting an excellent bond. Titanium is one of the easiest materials to diffusion bond, and it also happens to be superplastic. You can see a finer grain structure in the top — the stiffening ribs — and in the outside shell you can see a coarser structure. This is a Rolls-Royce alloy.

§3. Strain rate, grain size, and dislocation speeds [04:46]

§3.p1

Superplastic forming is actually a creep phenomenon in the metal. This is just for aluminum alloys, but these are different strain rates from 10⁻⁶ to 10⁵. So 10⁵ is 10 million percent deformation per second — these are ballistic velocities over here. Over here at 10⁻⁶ to 10⁻⁸ per second, this is pretty slow. Forming and tensile testing would be in the 10⁻³ to 10⁻⁴ range. Your grain size can vary by orders of magnitude. Basically aluminum, zinc, magnesium alloy, some of your 7000 series aerospace aluminum alloys, may have grain sizes on the order of 20 microns.

§3.p2

As you get finer and finer grain, creep becomes easier and easier. In fact, when we want the best high-temperature creep properties, we try to get rid of grains — we go to single crystals to completely eliminate grain boundaries. But if you want to get creep and make things superplastic, you want fine grain. This shows the fine grain size as a function of strain rate for aluminum alloys — powder metallurgy aluminum alloys. You're getting into superplastic forming strain rates of a tenth per second, which means you'll get 100% strain in 100 seconds. So in a couple of minutes you're pulling this thing — it's almost like my taking the clay and pulling it before your very eyes.

§3.p3

If you get very fine grains by physical vapor deposition, or if you go to mechanically alloyed aluminum — which is beating it in a ball mill and then consolidating it by extrusion — you get even finer grain. If you go to amorphous or nanocrystalline and you get grain sizes down in the thousand-angstrom range, you get superplasticity and you get high strain rates. In this region you have negative exponents of the strain rate; here we have positive exponents; and then we have impact superplasticity. You can have superplasticity in all these alloys at different strain rates, but it's basically a creep phenomenon.

§3.p4

Now, a question I get about once a year from someplace: someone will claim that the metal was deformed so rapidly that it was very highly strengthened. Superplasticity basically says you can deform it very fast and you're not necessarily going to be stronger — you're going to be weaker at higher temperatures. But the question is, at what strain rates do metals start to become stronger? It's at the strain rates where you start to compete with dislocation motion speeds. Dislocations go at about a third of the speed of sound in the material. When you're at movements of the metal on the order of a thousand meters per second — which you can be at ballistic velocities — things get stronger because the dislocations don't have time to move. But in general, in an automobile crash or a building collapse, the metal is not stronger. That's on a very slow time scale compared to dislocation movement. You hear this old wives' tale all the time: "oh, it was so fast the metal was much stronger." No — not unless you're talking ballistic velocities.

§4. Conform and dynamic recrystallization [09:27]

§4.p1

I mentioned Conform before. [Tom passes around a section of trolley wire.] This is a commercial Conform machine. Conform is the way they made this trolley wire. It's an extrusion process developed in the late 60s, early 70s at a British national laboratory — an energy laboratory doing work for the nuclear business. Some guys were doing basic research, and you basically pour some powder into a rotating channel, and at the end of the channel, just by friction, all this powder is being consolidated and churned together. The working end is this big circular thing right here. You've got powder filling in here, this is the motor to run it — a big motor. Conform stands for continuous forming. As long as you pour the powder in, you're mechanically working it, and you're getting tremendous redundancy. If you wanted to figure out some delta for this operation, it'd be like a thousand. You're kneading the powders, cold welding them together, and you end up with a tremendous amount of mechanical work in the alloy.

§4.p2

For these alloys, I determined, you end up getting what's called dynamic recrystallization. It comes out hot — you don't want to touch this piece of copper, because all the work that's gone into it brings it out at 400 or 500°. You'll burn your hand. But at that 400 or 500° for this particular alloy, which has a little bit of silver in it, you get to the recrystallization range. So you're mechanically working it, you're also nucleating fresh grains at the same time, but you're continuing to mechanically work it while you're nucleating the grains. That's dynamic recrystallization. You can get fantastic properties: super fine grain size, which you need for superplasticity, and fantastic material strength and toughness. Fine grain size gives you good toughness — the Hall-Petch equation, which hopefully you may have learned somewhere else.

§4.p3

Dynamic recrystallization is used in the steel industry to make sheet and plate, where they have rolling mills now with enough force to roll the steel at lower temperatures. Instead of rolling at 1100° Centigrade, they can roll at 1300 or 1400° Fahrenheit, which is kind of like forging temperatures. You're squeezing the stuff, deforming the grains, and it's recrystallizing while you're deforming, so you get fine grain size and much better properties. That's the basis for most of our high-strength low-alloy steels — being able to get dynamic recrystallization.

§4.p4

When I looked up Conform recently — I actually tried to do Conform to make some superconductors 30 years ago and it didn't work very well — if you look at Conform now on the web, it's used for the production of copper flat strip, copper bus bar, commutator connectors for motors, trolley contact wire, and other copper conductors. You can do it with aluminum, but it's easier to extrude aluminum the old-fashioned way; aluminum is at a low enough temperature. Obviously this is good for copper alloys. I tried to do it for a mixture of copper and niobium alloys, and I just broke the dies. The niobium strengthened things too much; I couldn't come up with a steel die that wouldn't break. When you've got something relatively soft that undergoes dynamic recrystallization, you end up with very good electrical conductors.

§4.p5

The only two countries I could find that have Conform machines now are India and China. Why? Because they still build railroads, and use trolley wire. Most of the rest of the world, we just buy it from them if we need it. So it's a specialized process, but it is a very high redundancy process. The North American company that was trying to sell Conform in the early 70s was located in Rhode Island — close enough that I could go do some research down there.

§5. The 30% rule: why two-phase alloys are superplastic [14:22]

§5.p1

If I'm going to try to get something superplastic, I need something that's going to help me maintain the fine grain size. One of the problems with very fine grain size is that the grains want to grow, and the smaller they are, the bigger the driving force for growth, because it goes as the inverse of the radius. This is a plot for the alpha-beta titanium alloys that are superplastic, showing elongation in percent. We're getting up to 1000%, certainly superplastic for a metal. But you get a peak somewhere around 30 volume percent.

§5.p2

Does anyone know what's significant about 30 volume percent? It comes out of geometry, but it shows up in metals. If you studied solidification of composites — anybody have an idea? It turns out, if a structure is controlled by surface tension, you can either get a bunch of little spheres embedded in a matrix, or you can get it to form plates or fibers embedded in the matrix. [Tom sketches at the board.] Actually cylinders — plates are cylinders, but just think of these little cylinders. The question is, will the alloy form spheres, or will it form cylinders, if you heat treat it, recrystallize it, and let the grains grow?

§5.p3

It turns out that the minimum surface energy for a mixture, if you have less than 30 volume percent of your second phase, is spheres. Above 30% you want to form fibers. You can do a minimization. I had to do this as a student — the method of Lagrange multipliers, variational calculus. You're MIT students, you can go look it up. If you write down the differential equation and do a minimization, you can prove that at 31 volume percent, exactly, the system prefers to have cylinders above 31%; below 31% it prefers a bunch of spheres for the second phase. These can coarsen into something else, but the point is, this is basically where we're getting structures stabilized by two interpenetrating phases in the alpha-beta mixture.

§5.p4

If I look at titanium and the microstructure for a superplastic alpha-beta titanium alloy — they're not exactly spheres in this case, for other reasons; one's hexagonal close packed (the alpha phase) and the beta phase is body-centered cubic. But you get something where the grains can't grow rapidly and coarsen because the beta grains, or the alpha grains, are completely surrounded by the other phase at around just under 30%. It's harder to get superplasticity in a single-phase alloy. It's easiest to get superplasticity when you have about 30% of one phase inside another, because that 30% phase is completely encapsulated. Superplasticity is really a creep phenomenon of grain boundaries sliding against each other. You want as much grain boundary area as possible to get the most superplasticity.

§5.p5

Student: Why isn't there a peak at 70%?

In theory there should be. In this case, in titanium, it has to do with the diffusion of the different elements in one phase versus the other phase. If it were just simple geometry, yes, there would be one at 70. If you include the coarsening phenomena and the diffusivity in the different phases, that's probably why we get rid of the 70% peak. If you look at the structures of these alloys, you'll find they're about 30% one phase and another. For example, the aluminum-zinc alloy — this first superplastic thing that Backofen did — I can't remember if it's 70% zinc or 70% aluminum, but it's right around 30%. If you look at the phase diagram, it's right around 30% of one phase in another. That's not happenstance — it comes out of trying to introduce the most grain boundary area per unit volume and have one phase stabilize the other.

§5.p6

Student: Solubility also.

Solubility and diffusivity — well, they're interrelated. You have to look at the whole phase diagram. Usually it's one phase in another, and nature chooses which wants to be the minor phase and which the major phase. If solubility and everything else were all the same, you should get two peaks. Or maybe it just becomes one big thing in the middle — maybe the 50% in between isn't much of a trough. There's a lot of scatter in that data. But the point is, at 30%, from pure geometry, you get the most grain boundary area for two phases interlocking. It's hard to get superplastic alloys that are single phase. You need the stabilization of the microstructure by having the two phases intermixed.

§6. Superplastic alloy systems and the M-value plateau [21:44]

§6.p1

Here are some compositions of superplastic alpha-beta titanium alloys. 6-4 titanium is the workhorse titanium alloy, invented over here at Watertown Mall. Actually it was Watertown Arsenal until about 20 years ago, when it died in the first round of BRAC. It was built before the Civil War to make weapons for the Army. They made cannonballs there during the Civil War. In the 1940s they invented titanium-6-aluminum-4-vanadium, which is superplastic in the 900° Centigrade range. There are a number of different titanium alloys, but they're all in this 800 to 1000° Centigrade range, with elongations at strain rates of 10⁻⁴ per second in the 300 to 1000% range.

§6.p2

There are lots of different alloys that behave superplastically, not just titanium. Aluminum-copper alloys — notice 33% copper. It's actually the copper intermetallics, but if you started looking at the phases. Zinc-aluminum — here's zinc 22 aluminum, so this alloy is 70% zinc and about a third aluminum (22 by weight, but aluminum is lighter than zinc, so volume percent works out). That's superplastic at 200 to 250. It's one of the reasons Backofen did it — he liked to use low-temperature systems, it's easier. That occurs over about two orders of magnitude in strain rates.

§6.p3

These two are duplex stainless steels. They are both FCC and BCC, and they have a duplex structure. They're in the 30% range again, because you need that to stabilize. You can make these stainless steels superplastic. If we could make 304 stainless superplastic, believe me we would — it's the workhorse alloy. But you can't do it. You have to use a duplex stainless, because it's best to have that 30% concentration. IN-100 is a nickel-based superalloy. 738, 629, this one that actually is a copper base. IN just means International Nickel. Then nickel aluminum bronze, which is basically a copper-base alloy. So you've got copper, aluminum, iron base, and titanium base superplastic alloys. But in general they're going to be things that occur at about 30% one phase in another — plus or minus 10 or 15%, because you have to stabilize the small grain size and not have it just take off on you.

§6.p4

The superplastic temperature — this is the strain rate exponent M, which controls superplasticity. Typical M values for metals are down in the 0.02 range. We're down here at 0.1 and 0.2. This is T over T_M, the melting temperature — the homologous temperature. As you get up to the creep regime — creep is usually considered above six-tenths of the melting point — nickel-based alloy, titanium, 304 stainless. Even 304 stainless will get a higher M value, but it's not superplastic. 0.2 is not really superplastic. You get higher elongation, things start to stretch more, but it may only be 50 or 60 or 70%, as opposed to 30 or 40%. In general, the M value goes up rapidly once things get into the creep regime, even if they're not superplastic.

§6.p5

This is for a couple of aluminum alloys. If you do everything in Kelvin, you find this is the same homologous temperature — strain rate sensitivity M versus the homologous temperature. It is a creep phenomenon. You can have diffusional creep, superplasticity, or power law creep. This is the log of the stress. This is going to be a lower temperature, this an intermediate temperature, this a high temperature. Here you have low M values — 0.1 or 0.2. When you get a high M value, like a half or greater, you get superplasticity. When you get into regular creep at high temperatures, you have lower M values. So there are different types of creep.

§7. Ashby deformation maps and creep mechanisms [27:23]

§7.p1

There's a book in the library — I looked to see if I could buy it — that Mike Ashby wrote, called Deformation Maps for creep. They look like this. He ended up at Cambridge University and retired a couple of years ago. He's very prolific, has written many books. He has a material selector program you can buy for $50,000. It works pretty well. He's one of the top metallurgists in the world for the last 30 or 40 years — top half-dozen metallurgists in the world. He really is among the very best. He came up with these broad-scale ways of looking at things. This is the normalized tensile stress over orders of magnitude versus the homologous temperature — a creep diagram.

§7.p2

He had a student when he was at Harvard do a doctoral thesis at General Electric — Coble, who graduated from MIT, went to General Electric research, invented Coble creep, and came back as a faculty member when I was a student. And Nabarro creep — Nabarro was a South African metallurgist who came up with an explanation of grain boundary creep. So Nabarro creep is grain boundary creep, Coble creep is a variation on Nabarro creep with a different type of diffusional mechanism. And dislocation creep is movement of dislocations within the crystal rather than at the grain boundary. So there are different ways to look at the dislocations.

§7.p3

If you look at this creep phenomenon as a function of stress and strain rate, you had superplasticity in the middle, diffusional creep, and power law creep over here. Power law is Nabarro-Herring and Coble creep. If you look at the M value plotted in the same region, you see you get a peak in your M values somewhere around 10⁻³ at intermediate stresses. It's a form of creep, but it gives you a very high M value. Here's that same type of plot for what looks like aluminum-copper alloys.

§7.p4

This work was done by David Holt and Backofen. Holt was hired as an assistant professor to work with Backofen, and he taught me part of Backofen's course with Backofen. He didn't get tenure at MIT, went off to become an Episcopalian minister as I remember. He used to do metallurgical consulting on the side. I don't know what happened to him since then. I think there's some relationship — Carl Thompson's wife's sister is married to Dave Holt or something. Anyway, here are your M values, stress, and these are M values as a function of strain rate around 10⁻³. You start getting things above five, and when you get above five on your M value, you're getting useful superplasticity. Looking at the grain sizes here, you're seeing 1.7 to 7.7 microns. Any grain size less than 10 micron is a very fine grain material. Most commercial alloys are in the 25-micron range, just because of the heating values and the way they're formed.

§7.p5

Here's the correlation between activation energy for creep and self-diffusion. So it is a diffusional process — atoms are moving around at the grain boundaries and changing the shape of the grains. All you've got is atoms moving at the grain boundaries. If you started out with all equiaxed grains and you put a stress on it, the atoms move to the tensile stress direction and away from the other direction, and you get elongated grains. It's just a diffusion phenomenon.

§8. Backofen on grain growth and Schuh's nanocrystals [34:03]

§8.p1

There are lots of problems with trying to get fine grain size. This comes out of Backofen, sort of at the end of his book, which he published in 1972 — about ten years after he discovered superplasticity. You can see in the back of his book, this is sort of an afterthought; he was still thinking about how to maintain fine grain size. You can talk about D in millimeters, diameter of grains. D to the minus one-half is the Hall-Petch equation. If you're a physical metallurgist, you know finer grain sizes give you higher strength and higher toughness. This is millimeters, this is inches to the minus one-half, here's D in inches. Then there's the ASTM grain size number — the steel guys came up with how many grains per square millimeter at 100x on a microscope. A typical grain size in a forging or a casting might be around zero or two — the grains are big, you can see them, centimeter in size. A typical structural steel might be a grain size of eight — that might be 25 microns or so. 10 or 12 is a high-strength low-alloy steel, where we get this dynamic recrystallization to try to maintain fine grain size, same as Conform and the trolley wire. You can get 10 to 12 sometimes; if you put in some precipitates too, you might get to 13.

§8.p2

But if you really want superplasticity in steels, you've got to be down at ASTM 16, where you're getting down to one to five microns grain size — one or two microns. I've never seen a steel with ASTM 16. About the finest I've ever seen is ASTM 13. But it would start to act like a superplastic material. The problem is, if you heat it up to superplastic temperatures, you'll find the grains grow. That's Backofen's next plot — he's trying to think globally about this whole thing. Lifespan of succeeding generations of grains at different homologous temperatures.

§8.p3

If you cycle around a phase transformation — like alpha-beta in titanium, or the eutectoid transformation in aluminum-zinc — you get recrystallization as you go from one phase to the other. You have it in titanium, in steels, in the aluminum-zinc alloys. You don't have it in every metal system. If you want a 50% reduction in grain size every time you cycle through the temperature — just thermal cycling to get finer grain size — this is what happens. Backofen plotted this for a tenth of a millimeter grain size, which is ASTM 4, pretty coarse grain. Depending on whether you do 2, 4, 6, or 8 cycles, you'd have to cycle between temperatures within a second or two. That's not practical when you've got a great big thing that takes tens of seconds or minutes for the heat to penetrate.

§8.p4

So Backofen in the late 60s is sitting there thinking, how can I get fine grain size? Since then, people have realized that the 30% one-phase-in-another helps stabilize the grain size. You get people like Professor Schuh, who made part of his tenure basically working with ultra-fine-grain metals. His company is based on a nickel-tungsten alloy, where the tungsten — it gets into not only volume fraction but solubility and diffusivity. Nickel and tungsten have lousy solubility in each other, and tungsten has lousy diffusivity because of its high melting point. So Professor Schuh has been able to stabilize very fine grains. It took 30 years between Backofen and Professor Schuh for people to get to ultra-fine-grain materials. Some of those alloys should have excellent forming characteristics in certain temperature ranges. But you have to do your mechanical shearing before the grains get so big that they lose their elongation capability.

§9. Production quantity and the SPF niche [39:47]

§9.p1

I guess I bothered a few people when I asked you, for your presentation, to think about production quantity. To me, when someone comes to me about a production process to make some material, one of my first questions is: how many do you want to make? Are you going to make one? A hundred? A hundred thousand? Because what process you choose depends on how many parts you make. If I want to make some complex egg-crate construction out of titanium for some fancy aerospace application, and I'm going to make less than 10 — let's say the space shuttle — unit production cost versus production quantity for space shuttle parts, I'm going to make them by hand. I'm going to get a great big forging, get a milling machine, and hog it out, with a buy-to-fly ratio of 200 to 1. It's going to cost a fortune. The space shuttle cost about $5 billion last time we bought one, 30 years ago. Right now you can't even buy them, and they fight over them.

§9.p2

I was in a meeting in Houston on Thursday, and the people from Houston were very upset that New York got one of the leftover space shuttles and they didn't get one in Houston. Well, that's where the control was. Houston, if you had done a better job, we wouldn't have lost two of them anyway. The superplastic methods are competitive in a production quantity somewhere between 100 and 4 or 5,000. Now we're talking aircraft, right? How many F-22 Raptors are we supposed to make? 183. It's right in there — superplastic forming is great. It only cost us $250 billion, or whatever the number is.

§9.p3

If you start getting into high production volumes, you're not going to use superplastic forming. It's a slow process; it might take an hour or two in the furnace to get one part. If you're going to make 200 parts a year, no problem. One part a week, no problem. But if you're going to make a million parts a year, you're going to need an awful lot of furnaces. Superplasticity has an important niche. It just happens to have that niche in the aerospace market, which is a very important market. Any questions? That's superplasticity. We spent more time on it than maybe its inherent importance, but it illustrates lots of microstructure and principles of mechanics of forming.

§10. Farberware Millennium cookware and hydrogen cracking [42:51]

§10.p1

Let's talk about plain old deep drawing. [Tom produces a set of progressively drawn cups.] Here's my set of deep draws, where you go from a round circular blank and you do the first draw, the second draw, third draw, fourth draw. These have all been cut off the end. Backofen's dead, I can't find out where those came from. Farberware is the company — they came to me 15 years ago through the MIT Industrial Liaison Program. They said, "We make these very expensive pots called Millennium Cookware, and we've shipped them to Japan, and lo and behold, when they get to Japan they're cracked."

§10.p2

I didn't know at the time that this was in Hosford's book: stainless steel cups with stress corrosion cracking in a laboratory atmosphere within 24 hours after drawing — the cracks perpendicular to the residual hoop tension stress. The Farberware pots had cracks like this. They put them on a boat, shipped them to Japan, and when they came off the boat they had cracks running down here. [Tom passes a cup around.] If I pass it around — there are a few cracks here, more than that. Two here, two here, one. Three right there. They asked me, why do we have these cracks?

§10.p3

[Tom produces a magnet.] I went and got my magnet and said: because you've changed the structure of your steel. It's very magnetic up here, and it's not magnetic down here. I said, you've got hydrogen cracking. They said, what's that? I said, well, there are two types of steel. There's 18-8 stainless, which is 304 annealed, and if you strain 304 stainless, you get work hardening. This is actually a 301 stainless, this is out of Backofen. The stress, 50 ksi yield stress, and then look at this work hardening rate — it goes up above 100, 200 ksi. This percent M is the percent martensite. 301 stainless is a TRIP steel — transformation-induced plasticity. It goes from face-centered cubic non-magnetic to body-centered cubic ferromagnetic. Martensite is ferromagnetic. Right in front of them in my office — this is probably pre-1995 — I got my magnet out and showed them the different magnetism going up the wall, different amounts of strain.

§10.p4

When they were drawing that, the material at the bottom doesn't get strained at all, the material going up the side is hardly strained, and the material up at the top got drawn in and had lots of deformation, so you have a variation of mechanical properties. Hosford calls it stress corrosion cracking; it's actually hydrogen embrittlement. 70% of the time metallurgists don't make a distinction between stress corrosion cracking and hydrogen embrittlement. There is a difference — one occurs at the anode, one occurs at the cathode. I learned this the hard way. For two weeks I was trying to crack a piece of steel in a little test, applying a battery to try to reproduce the cracking. I couldn't get it to crack, and the literature said it was stress corrosion cracking. Corrosion occurs at the anode — so I had the piece as the positive electrode and I couldn't crack it. Finally I thought, I wonder if this paper's wrong and it's really hydrogen cracking. So I reversed the polarity, made it negative, and within one hour I had a crack. I wasted two weeks of effort because someone just called it stress corrosion cracking and it was actually hydrogen cracking. There is a difference in polarity.

§10.p5

70% of the metallurgical literature makes no distinction, but there is one. One is a corrosion phenomenon at the anode; the other is hydrogen embrittlement at the cathode. Here's an example: Hosford doesn't know what he's talking about — he calls it stress corrosion cracking, it's actually hydrogen embrittlement. What's the corrosion for stainless steel just sitting on a laboratory bench? So I said to the guys from Farberware: I don't know where you're getting your hydrogen. How are you getting hydrogen into your steel to cause this cracking? They said, well, would it have anything to do with the fact that to put the Teflon coating on, we spray stainless steel powders on the inside surface to make a porous surface to impregnate the Teflon, and we do that in an argon 5% hydrogen atmosphere? I said, oh, that could be a source. Duh. There we had it. It's hydrogen cracking. It's delayed cracking — it didn't exist when they put it on the boat; by the time it got off the boat in Japan, the hydrogen had done its dirty work. Even Hosford said, well, it cracked in 24 hours, well, that could be corrosion — but 24 hours is pretty fast for corrosion. It's not fast at all for hydrogen. So there's a story about deep drawing changing the properties fairly dramatically.

§11. Course logistics [49:35]

§11.p1

Our next class will be on Friday. Jeremy, are they going to watch the rest of the videos on their own? Let's take a vote. The videos are online for the casting class, and Jeremy you've shown them too. How many people have gone ahead and watched the other videos? Okay, one of you. Good. Do you want to have Jeremy walk you through it at 11:00 tomorrow morning? I suggest you get online to eagar.mit.edu and find the casting things. You've already seen the first two, and there are about another 10 after that. That's part of the course — you're going to watch one-third of the course by video. For the next three days, frankly, you ought to be able to finish all of that this week or next.

§11.p2

I'm going to be down in Florida on Monday and Tuesday, and I'm tied up Wednesday, so we may have class on Thursday or Friday of next week. Right now we'll have class this Friday, then probably a class Thursday or Friday, and then it's spring break. Sometime this week or next I'll send everybody the schedule. If you haven't signed up for which week you're going to make your presentation, some of you will be making it the week we get back from spring break, the first week in April. That's all you have to do for the course: watch the other videos, make your presentation. If all goes well, we'll be done by the end of April. We're not going to videotape the presentations unless someone really wants to see themselves on video. In general, I'm not going to put you on the web.

Cases referenced

  • Faberware Millennium cookware hydrogen cracking §10.p1

    Tom's full forensic walkthrough of the case. Farberware came to him through the MIT Industrial Liaison Program after Millennium Cookware pots shipped to Japan developed cracks running parallel to the drawing axis during the voyage. Tom used a magnet to demonstrate that 301 stainless deep-drawing had transformed the FCC austenite to BCC martensite (TRIP behavior), creating the cathode for hydrogen embrittlement. Source of hydrogen identified as the argon-5% hydrogen atmosphere used to spray-deposit the stainless steel powder substrate for the Teflon coating. Tom uses the case to teach: (1) that 70% of metallurgists conflate stress corrosion cracking with hydrogen embrittlement, (2) that the distinction is anode (SCC) vs. cathode (HE), and (3) that delayed cracking with stainless on a bench is the diagnostic giveaway for hydrogen.

  • Watertown Arsenal titanium development (Ti-6Al-4V invention) §6.p1

    Tom locates the invention of Ti-6Al-4V at the Watertown Arsenal in the 1940s while introducing the workhorse superplastic titanium alloy. Includes the BRAC closure aside and the Civil War cannonball history. Used as biographical color for the alloy, not a developed case.

  • Conform process trolley wire manufacture §4.p1

    Tom passes a section of trolley wire around the class while explaining the Conform extrusion process. The wire is the artifact; the case is the process. Includes Tom's own failed attempt to use Conform for copper-niobium superconductor wire 30 years earlier (broken dies).

  • Conform process for superconductor wire (Tom's failed attempt) §4.p4

    "I tried to do it for a mixture of copper and niobium alloys, and I just broke the dies." Brief mention as a counter-example — Conform doesn't work when the second phase is too hard. Same teaching unit as the trolley wire case but a distinct cluster in the canon.

  • Backofen's discovery of superplasticity §5.p6

    Tom references Backofen's foundational aluminum-zinc superplasticity work (around 30% zinc or 30% aluminum) and his 1972 book where he was still wrestling with how to maintain fine grain size. Not a developed case but a recurring methodological anchor — Backofen was Tom's teacher.

  • Christopher Schuh ultra-fine-grain nickel-tungsten alloys §8.p4

    Tom characterizes his MIT colleague Christopher Schuh's tenure work and commercial venture (nickel-tungsten alloy with stabilized ultra-fine grain due to poor mutual solubility and tungsten's low diffusivity). Used as the modern follow-on to Backofen's unsolved grain-stabilization problem from §8.p1-3.

  • Coble creep / Nabarro creep mechanism naming §7.p2

    Tom recounts that Coble was Ashby's student at Harvard who did his doctoral thesis at General Electric, invented Coble creep, and came back to MIT as a faculty member during Tom's student years. Nabarro identified as a South African metallurgist. Not a developed case — purely biographical anchor for the creep mechanisms.

  • David Holt's MIT tenure denial §7.p4

    Tom mentions that David Holt — Backofen's collaborator on the M-value plots in this section — was hired as an assistant professor at MIT, didn't get tenure, and "went off to become an Episcopalian minister." Used as biographical color for the figure he's discussing.

Layer 2 — cleanup edit
p1 00:03

Actually talk mechanical behavior, deformation, things like that. Okay, couple of things I wanted to go over. Um, this is, uh, we talked about slipping uh on different crystallographic planes and that leads to anisotropy. But I came across this in one of the books I was looking at. Um, if you just take a bar that would be a single crystal and you were to pull on it, if it didn't have any bending and it would just start shearing on certain crystallographically preferred planes, you would actually get a translation off to the side. And this is just a demonstration of that. In fact, when you actually pull on a bar, a single crystal, you actually will pull in the same tensile direction because of the stiffness of the machine that's pulling it, and that requires that you also get a bending as well as a shear. Okay, and which you can see here, that's just a nice drawing um to show that even when you're a simple tensile test is not so simple.

p2 01:06

I remember Backofen saying that once when I was a student, that uh, and he'd been looking at tensile tests all his life, and he'd concluded that the tensile test, even though it seemingly was the most simplistic test you could imagine, but actually when you get into the details of it, it's extremely complex. And we're going to go through some of that a little bit later. This is another picture. When I was going through Hertzberg's book, he had a Charpy bar that fractured, but it's in a steel. I told you that the texture that we develop because of this anisotropic behavior gives you a fiber orientation, just like wood is anisotropic. Okay, and here's, you can see the fibers um in this particular piece of steel, so far as that goes. So that's just sort of enrichment or whatever you want to call it.

p3 01:57

We've been talking about sheet metal properties and we're starting to talk about super plasticity. We have n values, R values, and M values. n is the work hardening coefficient, R is the anisotropy ratio, and M is the strain rate sensitivity of the material. And super plasticity, we looked at the forming limit diagrams for certain types of metals. Instead of being having a, when the biaxial strain in the transverse direction E2 is zero, you might have 30 or 40 or 50% elongation in a regular old tensile bar. But in a super plastic bar, you can have 100%, 250%, 300%. And in fact, when you start stretching or compressing in the transverse direction, you actually can get even more strain. And that allows you to do all kinds of cheap metal forming operations when it's super plastic. But it also helps you even when the material is not super plastic.

p4 02:59

So then to review a little bit of super plasticity, super plastic forming, diffusion bonding. In fact, they call it SPFDB for super plastic diffusion bonded. Okay, SPF super plastic form diffusion bonded um was something that um the Air Force put a lot of money into after they realized that they could form titanium this way. Let's just look at this one. You basically would form two sheets and you'd weld them together in a hermetic seal. You'd put them in a die cavity, and you would, since it's hermetically sealed, you actually would put pressure in between the sheets as you heat the whole thing up. And it might take um 10, 20 minutes, it might take an hour or 2 hours depending on the temperature and the alloy that you're using, and your dies are going to have to be operating, I'll give you some temperatures in a little bit, they could be operating 800, 900° Centigrade. But you actually can form a very complex part, okay.

p5 04:09

Um, the bond, um, this is one of the diffusion bonds between one of those uh areas, and you end up getting an excellent bond. Titanium is one of the easiest materials to diffusion bond, and it also happens to be super plastic. And so you can see a finer grain structure in the top, the stiffening ribs, and in the outside shell you can see a coarser structure. Uh this is a Rolls-Royce alloy or sample.

p6 04:46

Now, um, super plastic forming is actually a creep phenomena in the metal. Um, this is just for aluminum alloys, but these are different strain rates from 10⁻⁶, okay, to 10⁵. So this is a 100, uh well it's a 100,000 per second, which is, if you want, 10 million percent deformation per second. This is pretty fast — this is sort of ballistic velocities over here. Uh and over here at 1 millionth or 1 100 millionth per second strain rate, this is pretty slow. Forming, tensile testing and stuff would be in the 10⁻³, 10⁻⁴. And your grain size um can vary — they've got orders of magnitude here, but basically aluminum, zinc, magnesium alloy, some of your 7000 series aluminum alloys used in aerospace, they may have grain sizes on the order of 20 microns.

p7 05:57

As you get finer and finer grain, creep becomes easier and easier. In fact, we try to get rid of grains when we want to get the best high temperature creep properties — we actually go to single crystals to completely get rid of grain boundaries. Well, if you want to get creep and make things super plastic, you want fine grain. And this shows the fine grain size as a function of strain rate for aluminum alloys. This is powder metallurgy aluminum alloys. Now you're getting into super plastic forming strain rates of a tenth of um a tenth per second, which means that you're going to get 100% strain in 100 seconds. So in a couple minutes you're actually going to pull this thing — you can, it's almost like my taking the clay and pulling it before your very eyes. That's the type of 10⁻¹, 10⁰ type strain rates, okay, to give you some idea.

p8 06:51

Uh, if you get very fine grains, physical vapor deposition, if you go to mechanically alloyed aluminum, which is a way of beating it in a ball mill and then consolidating it by extrusion, even finer. And then if you go to amorphous or nanocrystalline and you get grain sizes down here in the thousand um angstrom range, um, you get super plasticity and you get high strain rates. This, in this region you have negative exponents of the strain rate, here we have positive exponents, and then we have impact super plasticity. So you can have super plasticity in all these alloys at different strain rates, but it's basically a creep phenomena. This is just looking at aluminum alloys.

p9 07:39

Now, a question that I get, I would say about once a year from someplace, is someone will claim that the metal was deformed so rapidly that it was very highly strengthened. Okay, well super plasticity basically says, well, you can deform it very fast but you're not necessarily going to be stronger. You're going to be weaker, okay, at higher temperatures. But the question is, and I don't know if I've said this, um we do impact extrusion, okay, and we do um, we do ballistic, um well the Army and other people do ballistics where they're doing things at very high strain rates. And the question is at what strain rates does the metals start to become stronger. And I think I may have mentioned this — it's at the strain rates at which you start to compete with dislocation motion speeds. And the dislocations go at about a third of the speed of sound in the material. And so when you're at movements of the metal on the order of a thousand meters per second, which you can be in ballistic velocities, things get stronger because the dislocations don't have time to move. But in general, you know, in an automobile crash or um a building collapse or something like that, the metal's not stronger. I mean, that's on a very slow time scale compared to dislocation movement, okay. So, but you hear this little wives' tale, old wives' tale all the time about oh it was so fast the metal was much stronger. No — not unless you're talking ballistic velocities, okay.

p10 09:27

Um, this I mentioned before, Conform. And this is actually a commercial Conform machine. Conform is the way they made — and I pass this around — this trolley wire, which I mentioned before. Conform is an extrusion process that was developed in the late 60s, early 70s at a British national laboratory, an energy laboratory doing work for the nuclear business. And some guys are sort of doing basic research and you basically just kind of pour some powder into a channel and you rotate, it's a rotating channel, and at the end of the channel just by friction all this powder is being consolidated and churned together. Um, the actual working end of this is this end right here — it's just a big circular thing. You got powder filling in here, this is the motor to run it, it's a big motor, okay. Um, and you'll just, it stands "contin—" stands for Conform stands for continuous forming. As long as you pour the powder in, you're just mechanically working it, and you're getting tremendous redundancy. I mean, if you want to try to figure out some delta for this operation, it'd be like a thousand. I mean, it's just, you're kneading the powders and cold welding them together, and you end up ending up with a tremendous amount of mechanical work in the alloy.

p11 10:51

And that, it turns out for these alloys, I determined, you end up getting what's called dynamic recrystallization. It comes out hot, okay, you don't want to touch it when it comes out, this piece of copper, uh, because it's been all this work that's going into it and it comes out 400 or 500°. You'll burn your hand. But at that 400 or 500° for this particular alloy, which has got a little bit of silver in it, you get to the recrystallization range. And so you're mechanically working it, you're also nucleating fresh grains at the same time, but then you're continuing to mechanically work it while you're nucleating the grains. That's called dynamic recrystallization [recrystallization]. And you can get fantastic properties, super fine grain size, which you need for super plasticity, and fantastic material strength and toughness. 'Cause fine grain size gives you good toughness, if you know the Hall-Petch equation, um, which hopefully you may have learned somewhere else.

p12 11:52

But anyway, dynamic recrystallization is used in the steel industry to make sheet and plate material where they have rolling mills now that have enough force, and they roll the steel at lower temperatures. Instead of rolling them at 1100° Centigrade, they roll, which is kind of like forging temperatures, they can roll them at um 1300, 1400° Fahrenheit. And so you're squeezing the stuff, deforming the grains, and it's recrystallizing even while you're deforming, and so you can get fine grain size and you get much better properties. And that's what is the basis for most of our high strength low alloy steels, is being able to get dynamic recrystallization.

p13 12:33

Now, as I looked up Conform — I actually tried to do Conform to make some superconductors back 30 years ago and it didn't work very well — but if you look at Conform now on the web for the production of copper flat strip, copper bus bar, commutator connectors for motors, trolley contact wire, which is what this is, okay, and other copper conductor. You can do it with aluminum but it's easier to just extrude aluminum the old-fashioned way. Aluminum is at a low enough temperature. Obviously this is good for copper alloys. People have tried to do it — I tried to do it for mixture of copper and niobium alloys, and I just broke the dies. It couldn't, the niobium just strengthened things too much and I couldn't, uh, couldn't come up with a steel that, a steel die that wouldn't break. But when you got something relatively soft that undergoes dynamic [re]crystallization, um, you end up with some very good electrical conductors.

p14 13:37

However, the only two countries I could find that have Conform machines — I did it because they had a Conform machine, the North American in the early 70s, the North American uh company that was trying to sell Conform was located down here in P— Rhode Island, okay, so it was close enough that I could go do some research down there. Um, but the only countries now that have these great big machines are India and China. Why? Cuz they still build railroads, okay, uh and use trolley wire. Uh most of the rest of the world, well, we just buy it from them um if we need it. So it's sort of a specialized process, but it is a very high redundancy process.

p15 14:22

Now, if I'm going to try to get something super plastic, I need something that's going to help me maintain the fine grain size. And one of the problems with very fine grain size is the grains want to grow, and the smaller they are the bigger the driving force for growth, because it goes as the inverse of the radius. The smaller the radius the bigger the driving force, um, to grow larger grains. And this is a plot out of for the titanium alpha-beta titanium alloys that are super plastic, and it shows you the elongation in percent. Well, we're getting up to 1000%, so that's certainly super plastic for a metal. But you get a peak somewhere around 30 volume percent.

p16 15:07

Does anyone know what's significant about 30 volume percent? Just out of geometry, but it comes out of metals. If you studied solidification of composites or things, anybody have an idea? If you, I don't know if you run across this, but it turns out, if something is controlled — if a structure is controlled by surface tension, you can either get a bunch of little spheres embedded in a matrix — it shouldn't be looking like a happy face, anyway — or you can get it so it wants to form plates or fibers embedded in the matrix. [Tom is sketching at the board.] And this is in, or actually cylinders, okay, plates are cylinders, but just think of these little cylinders. And the question is, will the alloy want to form um spheres, or will it form cylinders, if you heat treat it, recrystallize it and let the grains grow?

p17 16:15

It turns out that the minimum surface energy for a mixture of spheres, if you have less than 30 volume percent of your second phase, is spheres. Above 30% you're going to want to form fibers. You can do a minimization. Actually if you go through a, I had to do this as a student, um, what do they call the method, of Lagrange multipliers, variational calculus, okay, you're MIT students, you can go look it up. But basically, if you can write down the differential equation and then do a minimization on that, you basically use the method of Lagrange multipliers, or you can use variational calculus. And you can prove at 31 volume percent, okay, exactly the minim— the system would prefer to have cylinders above 31%; below 31% it would rather form a bunch of spheres for the second phase. And then of course these things can coarse[n] into something else, but the point here, when I looked at this, I said, oh, this is basically where we're getting structures that can be stabilized by two interpenetrating phases in the alpha-beta mixture.

p18 17:39

So if I look at titanium and the microstructure for super plastic alpha-beta titanium alloy, I basically — they're not exactly spheres in this case for other reasons; one's hexagonal close packed, the alpha phase, and the beta phase is body-centered cubic. But in any case, you get something where the grains can't grow rapidly and coarsen because the beta grains, or the alpha grains, are completely surrounded by the beta grains at around something just under 30%, okay. So it's harder to get super plasticity in a single-phase alloy, okay. And a-, well, a single-phase alloy on either side. But it's easiest to get super plasticity when you have about 30% of one phase inside another, because that 30% phase is completely encapsulated in the uh in the second phase, as not exactly spheres necessarily but something close to spheres. If you had a bunch of fibers like this, it wouldn't have a lot of surface energy. This is the way to get the maximum surface energy. Because super plasticity is really a creep phenomena of the grain boundaries sliding against each other. So you want as much surface energy as possible, you want as much surface grain boundary, as many grain boundaries as possible, to get the most super plasticity.

p19 19:07

Student: Why is there not a [peak] at 70%?

Pardon me?

Student: Why isn't there one at 70%?

Well, I guess there would be, okay, but in this case, uh, I'm sure that in the titanium it has to do with the diffusion of the different elements in one phase versus the other phase. So you're right, in theory you could say there should be one at 30 and 70. I'm sure the reason there isn't one at 70 is probably because, uh, well I don't know exactly, it's a good point. If it's just a simple geometry, yes, there would be one at 70. If you also include the coarsening phenomena, the diffusivity in the different phases, then I'm sure that's one of the reasons we get rid of the 70% peak, okay. But if you look at the structures of these alloys, you'll find that they're about 30% on one phase and another. For example, if I went back to the aluminum zinc um, this first super plastic thing that Backofen did, it's about, uh, I can't remember if it's 70% zinc or 70% aluminum, but it's right around 30%. If you look at the phase diagram, it's right around 30% of one phase in another phase. And that's not a happenstance — that basically comes out about the fact you're trying to introduce the most grain boundary area per unit volume and have one phase stabilize the second phase. But you're right, you should have two peaks if you didn't have diffusion effects also.

p20 20:34

Student: Solubility also.

Well, solubility and diffusivity — well the solubility and the diffusivity are interrelated, right. You have to look at the whole phase diagram. Usually it's one phase in another phase, and nature chooses which phase wants to be the minor phase and which one wants to be the major phase. If the solubility and everything else were all the same, I think he's right, you should get two peaks, okay. Or maybe you just, it kind of becomes one big thing in the middle, okay, maybe the 50% in between isn't much of a trough. A lot of scatter in that data. But the point is, at 30, from a pure geometry, all nothing else considered, you get the most grain boundary area for two phases interlocking in one another. It's hard to get super plastic alloys that are single phase, okay. You kind of need the stabilization of the microstructure by having the two phases intermixed. That's what I'm trying to get across, okay.

p21 21:44

Um, now here are some compositions of super plastic alpha-beta titanium alloys. 6-4 titanium, that is the workhorse titanium alloy, invented over here at Watertown, uh, Mall, okay. Actually was the Watertown Arsenal until about 20 years ago when they had the BRAC. It was, it died in the first round of BRAC. But it was built before the Civil War to make weapons for the Army. They made cannonballs there during the Civil War. In any case, in 1940s they invented titanium 6 aluminum 4 vanadium, uh, which is super plastic in the 900° Centigrade range. There's a number of different uh titanium alloys, but they're all in this 800 to 1000° Centigrade range, and they have elongations at strain rates of 10⁻⁴ per second in the 300 to 1000% range, okay. So, lots of different alloys, but they're all going to be processed in a way to give you some sort of microstructure that looks like a bunch of 30% one phase in another phase, okay, approximately.

p22 22:59

Now there are lots of different alloys that behave super plastically, not just titanium. Aluminum copper alloys — notice 33% copper, I mean it actually is the copper intermetallics, but nonetheless, if you started looking at the phases. Zinc aluminum — here's the zinc aluminum, zinc 22 aluminum, so this alloy is 70% zinc and about a third aluminum, um, 22 by weight but volume percent aluminum is lighter than zinc and stuff. Um, and that's super plastic at 200 to 250. It's one of the reasons Backofen did it, he liked to use low temperature systems, it's easier. Um, and it, that occurs over about two orders of magnitude in strain rates.

p23 23:42

Uh, these two are what we call duplex stainless steels. They are um both uh FCC and BCC, and they have a duplex structure. They are in the 30% range again, because you need that to stabilize. You can't just — because, you can make these stainless steel super plastic. If we could make 304 stainless super plastic, believe me we would, it's the workhorse alloy, okay. But you can't do it. You have to use a duplex stainless because you, it's best to have this kind of 30% concentration. IN-100 is nickel-based super alloy. 738, 629, um, this one that actually is a copper base. IN just means International Nickel. Um, and then nickel aluminum bronze, which is basically a copper base alloy. So you got copper, aluminum, iron base, um, and of course we have titanium base as well, um, super plastic alloys. But in general they're going to be things that occur at about 30% one phase and another phase. It can be 30% plus or minus 10 or 15%, but it's going to be in that range, because you have to stabilize the small grain size and not have it just take off on you.

p24 25:05

The super plastic temperature, this is the strain rate exponent. Remember M is what controls super plasticity. Typical M values for metals are down in the 0.02 range — I've showed you that data before. And so we're down here, this is 0.1 and 0.2. And as you get above, this is T over T_M, that's the melting temperature, so this is what's called the homologous temperature. As you get up to the creep regime — creep usually is considered above six-tenths of the melting point for homologous temperature — nickel-based alloy, titanium, 304 stainless. Now even 304 stainless, you'll get a bit higher M value, but it's not super plastic. Point 2 is not really super plastic. You get higher elongation, things start to stretch more, but it may only be a 50 or 60 or 70%, as opposed to a 30 or 40%, um. But just in general the M value goes up rapidly once things get into the creep regime, even if they're not super plastic, okay.

p25 26:14

This is just for a couple of aluminum alloys, again showing this. If you do everything in Kelvin, you would find this is the same homologous temperature, strain rate sensitivity M versus the homologous temperature. It is a creep phenomena. Um, if you look at, this may be the best version I found. Um, you can have diffusional creep, super plasticity, or power law creep. And this is the log of the stress, and so actually this is going to be a lower temperature, this is going to be an intermediate temperature, this is going to be a high temperature. Here you have um, that should be 0.1 or 0.2, okay, low M values. When you get a high M value, like a half or greater, you get super plasticity. And when you get into just regular creep at high temperatures, you have lower M values. And so, uh, there's different types of creep.

p26 27:23

Uh there's a book that I look to see if I could buy it, it's in the library, that Mike Ashby wrote, where he called deformation maps for creep, and they actually they look like this. Comes out of Hertzberg [?] places. He ended up at Cambridge University and retired a couple years ago. He's very prolific, he's written many books. He has a material selector program that you can buy for $50,000. It works pretty well, okay, but any case, he's one of the top metallurgists in the world for the last 30 or 40 years. Top half dozen metallurgists, let me say, okay. He really is among the very best. Um, he came up with these kind of broad scale ways of looking at things. This is the normalized tensile stress over orders of magnitude versus the homologous temperature, and this is a creep diagram. He started, he had a student when he was at Harvard do a doctoral thesis at General Electric — he was, he graduated from MIT, went to General Electric research, invented Coble creep, and came back as a faculty member when I was a student. Um, and Nabarro creep — Nabarro is a South American metallurgist who, South African metallurgist, who um came up with an explanation of grain boundary creep. So Nabarro creep is a grain boundary creep, Coble creep is a variation on Nabarro creep with a different type of diffusional mechanism. And then dislocation creep is movement of dislocations within the crystal rather than at the grain boundary. So there's different ways to look at the dislocations.

p27 30:05

But in any case, if you look at, again, this creep phenomena, as a function of temperature and strain rate — not, well, stress and strain rate — you had super plasticity in the middle, you had diffusional creep, and you had power law creep over here. Power law is kind of Nabarro-Herring and Coble creep, um. And you have, if you look at the M value plotted in the same region, you see you get a peak in your M values somewhere around 10⁻³ at intermediate stresses. But it is a form of creep, but it gives you a very high M value. If you want to look at that same type of plot for, this is a bunch of aluminum copper [?] alloys I guess, okay.

p28 30:59

Uh, this work was done by David Holt and Backofen. And Holt was hired as an assistant professor to work with Backofen, and he taught me part of Backofen's course with Backofen. And um, he didn't get tenure at MIT, went off to become an Episcopalian minister as I remember, okay. And he used to do metallurgical consulting on the side anyway. And I don't know what happened to him since then. I think he may have, I think there's some relationship — Carl Thompson's wife's sister is married to Dave Holt or something, I don't remember. Anyway, um, but anyway, here are your M values, stress, and these are your M values as a function of strain rate around 10⁻³. You start getting these things that are above five, and that's where you're getting, when you get above five on your M value, you're getting to useful super plasticity, okay. But exactly which alloy, um, what grain size — looking at the grain sizes here, you're seeing grain sizes 1.7 to 7.7 microns. If you know anything about grain size, any grain size less than 10 micron is a very fine grain material. Uh, most commercial alloys will be in the 25 micron grain size, okay, just because of the heating values and the way they're formed and everything.

p29 32:23

Um, so here's um, correlation between activation energy for creep and self-diffusion. So it is a diffusional process, as atoms are moving around at the grain boundaries and changing the shape of the grains. Basically, all you've got is you've got atoms moving at the grain boundaries. So if you started out with all equiaxed grains and you put a stress on it, now the atoms are going to move to the tensile stress direction and away from the other direction, and now you're going to get elongated grains, okay. So it's just a diffusion phenomenon.

p30 34:03

Um, there are lots of problems with trying to get fine grain size. This comes out of Backofen, sort of at the end of his book, uh, which he wrote, he published in 1972, which is not all that many years after he discovered super plasticity. In the about 10 years after he discovered super plasticity. And you can sort of see in the back of his book, this is sort of an afterthought, he was still thinking about this whole problem of how do you maintain fine grain size. So you can talk about D in millimeters, diameter of grains in millimeters. You can think of in D the minus one-half, that's the Hall-Petch equation. If you're a physical metallurgist you know that finer grain sizes give you higher strength and higher toughness. This is in millimeters, this is inches to the minus one-half, here's D in inches. And then there's the ASTM grain size number. ASTM — really the steel guys years ago came up with how many grains per square millimeter at 100x on a microscope or something, okay. If you want, I could bring in the plates of the ASTM grain size, but typical a grain size in a forging or a casting might be around zero or two. I mean, the grains are big, you can see them, okay. Centimeter in size. Um, a typical structural steel might be a grain size of eight — that might be your 25 microns or so. 10 or 12 is a high strength low alloy steel, where we get this dynamic recrystallization to try to maintain fine grain size, same type of thing they did with Conform and the uh trolley wire, okay. So you can get 10 to 12 sometimes, if you put in some precipitates too you might get to 13.

p31 35:56

But if you really want to get superplasticity in steels, you got to be down here in the 16 ASTM, 16 grain size, where you're getting down to in microns, um, you're going to get down to one to five micron, or one or two microns grain size, okay. That's the 16 range. I've never seen a steel with ASTM 16, okay. About the finest I've ever seen is ASTM 13. But it would start to act like a super plastic material. The problem is, if you heat it up to the super plastic temperatures, you're going to find that the grains grow. And that's Backofen's next plot, where he's trying to think — you can kind of see him trying to think globally about this whole thing. Lifespan of succeeding generations of grains at different homologous temperatures, to keep um the uh, what you want each time you try to, well let me back up what he was doing here.

p32 37:00

If you cycle around a phase transformation, like the alpha-beta phase transformation in titanium, or there is a eutectoid phase transformation in aluminum zinc, if you cycle from alpha phase to beta phase and back down, you can get, you get recrystallization as you go from one phase to the other phase, as you go up and above and below this phase, uh, temperature transition. You have it in titanium, you have it in steels, you have it in the aluminum zinc alloys. You don't have it in every metal system. But if you want to get a 50% reduction in grain size every time you cycle through the temperature, to just do thermal cycling to get finer grain size, this is what happens. And Backofen basically plotted all this for a tenth of a millimeter grain size, which is ASTM 4, which is pretty coarse grains. And he shows that depending on the number of cycles you do, 2, 4, 6, or 8, you're going to have to be doing cycling between temperatures within a second or two. Well, that's not very practical, not when you got a great big thing that takes tens of seconds or minutes to get the heat to penetrate in, right?

p33 38:16

So basically Backofen's in the early u— in the late 60s is sitting there thinking, how can I get fine grain size? Now since then, people have realized that this kind of 30% one phase and another phase helped stabilize the grain size. You get people like Professor Schuh, who made part of his tenure basically on working with ultra fine grain metals, where he used, actually he used titani—, well, a lot of his stuff, his company's based on a nickel tungsten alloy, where the tungsten — it gets into not only volume fraction but solubility and diffusivity. Nickel and tungsten have lousy solubility in each other, and lousy, tungsten's got lousy diffusivity um because of its high melting point. And so Professor Schuh has been able to stabilize very fine grains. Um, but it took 30 years between Backofen and Professor Schuh for people to be able to get to ultra fine grain materials. But some of those alloys that Professor Schuh has should have some excellent forming characteristics in certain temperature ranges, as long as — but it's sort of, eh, well, the grains recrystallize and grow bigger before you get to form it, okay. So you know, you got to do your mechanical shearing before the grains get so big that they lose their elongation capability, okay.

p34 39:47

Um, I guess I bothered a few people when I asked you to talk about, uh, for your presentation, to think about production quantity. But, and I never discussed it, um, but to me I just automatically, when someone comes to me about a production process to make some material, one of my first questions is, how many do you want to make? Are you going to make one? You going to make a 100? You're going to make 100,000? Okay, because what process you choose is going to be dependent on how many parts you make. So if I want to make some complex egg crate construction out of titanium for some fancy aerospace application, and I'm going to make less than 10 — so let's say it's the space shuttle, the unit production cost versus the production quantity for space shuttle parts — I'm going to make them by hand. I'm just going to get a great big forging, I'm going to get a milling machine, and I'm going to hog it out and have a buy-to-fly ratio of 200 to 1. It's going to cost a fortune. Space shuttle cost about $5 billion last time we bought one 30 years ago. Right now you can't even buy them, um, and they fight over them.

p35 41:00

I was talking, this in a meeting on, in Houston on Thursday, and the people from Houston were very upset that New York got one of the leftover space shuttles and they didn't get one in Houston, right. Well, that's where the control was. Well Houston, if you had done a better job we wouldn't have lost two of them anyway. Um, the super plastic methods are competitive in a production quantity somewhere between 100 and 4 or 5000. Well, now we're talking about aircraft, right? How many uh F-22s, were Raptors, are we supposed to make? 183? Well gee, we're, it's right in there, right, super plastic forming is great, right. It only cost us $250 billion, right, or whatever the number is.

p36 41:53

Um, if you start getting into high production volumes, you're not going to use super plastic forming. It's a slow process, it might take an hour or two in the furnace to get one part. If you're going to make 200 parts in a year, no problem, okay. Or one part a week, no problem. But if you're going to make a million parts a year, you're going to have an awful lot of furnaces, okay. So super plasticity has sort of an important niche. It just happens to have an important niche in the aerospace market, which is a very important market. Any questions? That's super plasticity. Now we're going to go on to something else. So, um, we spent more time on super plasticity than maybe its, um, its inherent um importance, but it illustrates lots of microstructure and uh principles of mechanics of forming.

p37 42:51

Let's talk now about just plain old deep drawing again. [Tom produces a set of progressively drawn cups.] And you know, here's my set of deep draws, where you go from a round circular blank and you do the first draw, the second draw, third draw, and the fourth draw. And these have all been cut off the end, um. I don't know — Backofen's dead, I can't find out where those came from. But in any case, this is something that um, someone came to me — actually Farber—, Farberware is the company — came to me and, I looks like I forgot to move a picture up to somewhere else. Um, but in any case, Farberware came to me 15 years ago through the MIT Industrial Liaison Program, and they said, we make these very expensive pots called Millennium Cookware, and we've shipped them to Japan, and lo and behold, um, when they get to Japan they're cracked.

p38 44:00

And in fact, I didn't know that this was in Hosford's book at the time, but this stainless steel cups with stress corrosion cracking um in a laboratory atmosphere within 24 hours after drawing — the cracks are perpendicular to the residual hoop tension stress. So they had that — the Farberware pots, um, had cracks like this. And they had, they didn't form well, I don't know if they formed within 24 hours, but they put them on a boat, they shipped them to Japan, and when they came off the boat they had cracks running down here. [Tom passes a cup around.] This, if I pass it around, there's a few cracks around here, more than that. Well, there's two here, there's two here, and there's one. Maybe it's only got, no, it's got three right there, okay. So they asked me, um, why do we have these cracks?

p39 44:53

And so I went and got my magnet and I said, because you've changed the structure of your steel. [Tom demonstrates with a magnet.] It's very magnetic up here, and it's not magnetic down here, right. You want to pass that around so everybody can. And I said, you've got hydrogen cracking. And they said, what's that? And I said, well, if you get hydrogen into your steel, there's two types of steel, there's 18-8 stainless, which is 304 and anneal, and if you strain 304 stainless steel, you will get um work hardening. Maybe I messed this up again, yeah, okay. You'll get work hardening that causes the steel to transform. This is actually a 301 stainless, this is out of Backofen, this is the stress, 50 ksi yield stress, and then boy, look at this work hardening rate, it will go up above 200, 100 ksi. And this percent M is the percent martensite. 301 stainless steel is a TRIP steel, transformation-induced plasticity. So it goes from face-centered cubic non-magnetic to body-centered cubic ferromagnetic. Martensite is ferromagnetic. And I just right in front of them — my office, this is probably pre-1995 — I just got my magnet out and I bang bang showed them how the different magnetism going up the wall, why, different magnetism different amounts of strain.

p40 46:34

The material when they were drawing that, the material down at the bottom doesn't get strained at all, the material right here going up the side is hardly strained very much, the material up here at the top got drawn in and had lots of deformation, and so you have a variation of mechanical properties. Hosford shows, he calls it stress corrosion cracking, it's actually hydrogen embrittlement, but that's another thing. 70% of the time metallurgists don't make a distinction between stress corrosion cracking and hydrogen embrittlement. There is a difference — one occurs at the anode, one occurs at the cathode. I learned that in a very difficult way. For 2 weeks I was trying to crack a piece of steel in a little test, and we were applying a battery to try to reproduce the cracking. I couldn't get it to crack, and the literature said it was stress corrosion cracking. Corrosion occurs at the anode, right? So I had the piece as the positive piece and I couldn't crack it, I couldn't crack it. And finally I thought, I wonder if this paper's wrong and it's really hydrogen cracking. So I reversed its polarity, made it negative, and within one hour I had a crack. I wasted two weeks effort because someone, some metall— just called it stress corrosion cracking and it was already hydrogen cracking. There is a difference in polarity.

p41 47:53

70% of the metallurgical literature makes no distinction between the two, but there is. One is a corrosion phenomena occurring at the anode, the other is a hydrogen embrittlement phenomena occurring at the cathode, okay. I knew that was true, but I hadn't thought about it. I just assumed the people in the literature knew what they were talking about. And here's an example: Hosford doesn't know what he's talking about, he calls it stress corrosion cracking, it's actually hydrogen embrittl[ement]. Um, cuz what's the corrosion for stainless steel just sitting on a laboratory bench? And so I said to the guys from uh Farberware, I said, uh, I don't know where you're getting your hydrogen. How are you getting hydrogen into your steel to cause this cracking? They said, well, would it have anything to do with the fact that to put the Teflon coating on we spray stainless steel powders on the inside surface to make a porous surface to impregnate the uh the Teflon, and we do that in an argon 5% hydrogen atmosphere? I said, oh, that could be a source, okay. Duh, okay. So there we had it. That is hydrogen cracking. It's delayed cracking. It didn't exist when they put it on the boat, by the time it got off the boat in Japan the hydrogen had done its dirty work, okay. And even Hosford said, well it cracked in 24 hours, well that could be corrosion, but 24 hours is pretty fast for corrosion. But it's not fast at all for hydrogen, okay. So there's a story about deep drawing changed the properties fairly dramatically, okay.

p42 49:35

Uh, we will, I'm not, well I'm in and out uh for the next three days. Our next class will be on Friday. Um, Jeremy, are they going to watch the rest of the videos on their own, or — um, so let's take a vote. I mean, the videos are online for the casting class, and Jeremy you've shown them too. How many people have gone ahead and watched the other videos? Okay, one of you. Good. Well, do you want to have, and Jeremy walk you through it at 11:00 tomorrow morning? And I would suggest that you just get online to eagar.mit.edu and find the casting things. You've already seen the first two, and there's about another 10 after that, and that's part of the course, okay. You're going to watch one-third of the course by video. And so for the next three days, frankly, you ought to be able to finish up all of that this week or next week.

p43 50:31

I'm going to be down in Florida on Monday and Tuesday, and I'm tied up Wednesday, so um we will, we may have class on Thursday or Friday of next week. So right now we'll have class this Friday, and then we'll probably have a class on Thursday or Friday, and then it's spring break. And sometime this week or next I'll send everybody the schedule. In fact, if you haven't signed up for when you're going to make your, which week you're going to make your presentation, it's going to be, some of you are going to be making it the week we get back from spring break, the first week in April. So um, but that's all you have to do for the course: watch the other videos, make your presentation. And we should, if all goes well, be done by the end of April. Even with your presentations — we're not going to videotape the presentations unless someone really wants to see themselves on video. Um, but in general, I'm not going to put you on the web.