§1. Alclad and the invention of aluminum refining [00:05]
Student: Why do they use Alclad for things like that?
We've made Alclad alloys since the 1930s. Alclad composites have a very thin layer — two to four thousandths of pure aluminum on the surface. Pure aluminum has excellent corrosion resistance and excellent pitting resistance. So if I have an aluminum-copper alloy — this is a piece of something called Novelis Skooter. Novelis is the sheet metal skin off of Alcan, when Alcan merged with Péchiney of France. Two guys invented the refining process in 1888 — and actually last spring in my class one of the students did their presentation on Paul Charles Martin Hall, the founder of Alcoa, and the other was Paul Héroult. Héroult invented the exact same process within two months of Hall, thousands of miles apart, with no communication between them. Refining aluminum, making it cheap.
Before that, aluminum was more expensive than gold. The crown jewels of the French royal family included aluminum dinnerware because it was so unique and the more valuable material. The top of the Washington Monument has a little aluminum pyramid as a lightning rod — at the time it was more valuable than gold. So in 1888 Hall and Héroult developed this process. Pure aluminum is very soft, so you have to strengthen it — six or seven ksi otherwise.
What they did is roll-bond. They'd take a two-inch-thick pure aluminum slab and a twenty- or thirty-inch aluminum alloy casting, and roll-bond them together in the rolling mill. I just read something recently — they had about a fifteen percent success rate, with a very poor yield. But Novelis Skooter came along, and it turns out they can co-cast pure aluminum with the alloy. The alloy has lousy corrosion resistance but high strength, so your Alclad sheet metal ends up with about two thousandths of pure aluminum on the surface — excellent corrosion resistance. The outside skin of all Airbus aircraft is Alclad: high-strength alloy in the center. It's a peanut-butter sandwich — the bread is pure aluminum. It's a composite. And this process has essentially one hundred percent yield because it's bonded in the casting operation. So it sort of took over the market about seven or ten years ago.
§2. The aluminum alloy series [03:48]
So these are aluminum alloys. The 1100 series is ninety-nine percent aluminum — architectural and decorative applications, furniture, deep-drawing parts, hollowware, things that don't need much strength. Bottoms of paint cans, pots and pans for the kitchen. 1060 is 99.6 percent aluminum — so you can see, one percent impurities, point-six percent impurities. There are other alloys in between; the ASM book will follow that pattern. 1350 is 99.5 percent aluminum, an electrical conductor. You see these 350,000-volt transmission towers, and the cable strung between them is aluminum — but it has a steel core for restraint. Aluminum is known for electrical conductivity. Electrical and thermal conductivity go together in metals — there's actually a formula from a hundred years ago that relates the two. They're not equated terms but they are so correlated that there's a formula relating electrical conductivity, thermal conductivity, and reflectivity of the metal, all related to free electron concentration. The better the electrical conductivity, the better the thermal conductivity, in virtually anything with metallic bonding.
Then we get to the 3000 alloys. They've got one percent of manganese, maybe some other small additions. This one's up a little bit with magnesium for extra strength — sheet metal requiring higher strength, general-purpose applications where strength and formability are required, process and food handling equipment. Soda cans. 5005 here is a structural application. The 5454 and 5456 are marine components. These are 20 ksi, 25 ksi strength levels — not tremendous strength, not as strong as steel, but a lot lighter than steel. So if you need to move something, and you can weld it up to full strength without great difficulty.
Here's the heat-treaters' guide for aluminum alloys. The 2000 series alloys are aluminum-copper. With magnesium-silicon, you get precipitation hardening. And here's the aluminum-copper-lithium alloys — an 8000 series extruded bar. Lithium has an interesting property: it's lightweight and reduces the density of aluminum by ten percent. It also increases the Young's modulus by ten percent. And E over rho is the important parameter for structural stiffness. So if I want something slender and stiff, I want a good E over rho. With a little bit of math, lithium alloys give me twenty percent more stiffness performance. So they started making part of the Space Shuttle main tank out of aluminum-lithium alloys at the end of the lifetime of the shuttle. A lot of Boeing 787 aluminum components are 8000 series.
The Soviets first discovered this in the 1970s, and we worked on it for thirty years. Alcoa worked on it for thirty years. There were lots of problems with hot cracking and weldability. Finally, after thirty years of development, it's still somewhat of a pain in the neck and still pricey, because aluminum is very active, like titanium. So how you weld aluminum-lithium is a little bit different. But we're now using a little bit of lithium alloys. The odds of using it in a corrosion-critical application aren't great — it's not that weldable, and the lithium actually makes the corrosion worse. But when you're maintaining an aircraft worth $200 million and you've got mechanics on it — not like helicopters with two hours of maintenance for every hour of flight, but a commercial aircraft at probably one hour of maintenance for every hour of flight — that's per person, out of play. Helicopters are the worst.
§3. Precipitation hardening and the temper designations [09:23]
So here's a phase diagram — this might be copper-aluminum. You solution-ize by bringing the material up while still solid in this range. Then you quench it down to low temperature, and you can precipitation-harden in this range. If you're going to anneal it, you use these ranges and temperatures. So the dissolve-and-heat-treat process looks like this.
Alloys that are not as weldable — we can weld them, but we don't get full strength because we don't heat treat. You go up to the solid solution, quench it down, and then your heat-treat cycle looks like this: held at 940°F for an hour, then quench it, cold water. Then hold it for ten hours at 300 or 400 degrees in cool down. And you'll double or even triple the strength. Here are the tempering curves.
Here's your hardness number, Vickers hardness — we've talked about hardness testing. So it's the easy way to test what we're doing. The as-quenched state is down here at solid solution. This is an annealed product. If you just leave it at room temperature, after an hour it starts to harden just naturally, as the precipitation takes place even at room temperature. After ten hours you actually reach a significant increase in strength. And that will go out to hundreds of thousands of hours — twenty years. But if you heat it up a little bit — this is 230 degrees Fahrenheit — you can actually get, depending how long you hold it there, a hundred hours or ten hours of treatment. Let's take ten hours. If I want a ten-hour heat treatment at 320 degrees Fahrenheit, I can double the strength, 110 Vickers. And that's what we call a T6 treatment, to optimize strength. We have problems with that strength and I'll talk about that in a second. There's a whole series. If you heat at too high a temperature, you reach the peak and soften before you get the optimum. Usually you wait a hundred hours to get the maximum strength, where you can get two-and-a-half times the strength.
They once did a study of the Wright brothers' engine, the one in the Smithsonian. Someone took a little sample off, ran it under the scanning electron microscope, and looked at how the aluminum-copper precipitates had grown over a hundred years. They put it on a log-scale graph, and those things are still precipitating at room temperature sitting in the Smithsonian. Everything falls on a straight line. They're getting coarser over time.
There are a number of different designations. Let's say I have 7075. So 7075 is the aluminum-magnesium-silicon heat-treatable alloy, and it might have a T6 designation. This is the alloy composition, defined in the tables. T6 means it's heat treated. The aluminum alloys typically have a dash designation. Dash F means as-fabricated. H-O means annealed — the softest condition. So if I have a 4504 aluminum plate, not heat-treatable, and I get it in the annealed condition, it's the softest, the lowest-strength condition. H is cold-worked. Say I get it annealed at two inches thick, and I put it through the mill and go down to one inch thick. It's now harder — the grains have gotten smaller, they've turned into pancakes from equiaxed grains. And you get greater strength. So what you're going to build a hull out of is going to be an H-something. H2 is quarter hard, H4 is half hard, H6 is three-quarter hard, H8 is full hard — it's just how much reduction did I have, how much cold work did I put in. I can more than double the strength with H8. So I can go from an annealed alloy at maybe 10 ksi up to 20 ksi, or even 25 in some cases.
There's also solution heat-treated — we're heating it up, quenching it down — but I haven't done the tempering, because I'm going to do that later, after I've used it for forming and other things. Because once I heat-treat it and make it hard, it's difficult to form. So heat-treated in a stable condition means it's not going to keep on precipitating. There are lots of different versions of this, depending on the aluminum alloy. T6 is solution-heat-treated and artificially aged to the optimum condition. Well, it turns out the optimum condition doesn't have the best stress-corrosion-cracking resistance. And here is an example.
§4. The America's Cup hydraulic cylinder failure [15:38]
I got this from a guy who graduated in the department, Jerry — I might be empty-headed on his last name. He was working for a company called Navtech. Navtech used to be here in Massachusetts; they've moved to Connecticut. They make specialty marine hardware for fancy yachts and America's Cup boats — pretty expensive stuff. [Tom holds up the fractured piston cylinder.] This is a piston cylinder that operated part of a winch system that fractured. In 2002 he came to me. It was 7075-T6 alloy. They'd been using it at a yield strength of 73 ksi. K-1c — fracture toughness — was 26 ksi √inch. Wall thickness was about an inch. And it operated at a pressure of 7,500 psi. I wouldn't want to stand next to that and let go of a hydraulic line. Well, neither did the guys working the winches for the sails on the America's Cup, and this one let go. They were concerned. The problem was stress corrosion cracking, which I always tend to find in saltwater environments.
[Tom adjusts the document camera focus on the fractured part.] Right in here there's a thumbnail crack — you can kind of see it; you can imagine what it was before, twelve years ago. If you look at the texture here — I could blow it up some more — there was actually a color differentiation when I looked at it twelve years ago. You see a little texture, a thumbnail crack right there, and then lines radiating from it. Maybe you have to look at these things for thirty years. I agree, at least for somebody who hasn't looked at this for years it may not be as obvious. It went about a third of the way through. The flaw went about a millimeter deep before it fractured all the way through. I'm looking, from experience, at some of the other features on here. See lines radiating, coming out at an angle, straight-down lines — you can kind of see those lines. So maybe I integrate things with my experience, but I can select the fracture in it. It was more visible at the time. I've handled it since then and it's not as easy to see.
So that's at 7,500 psi. It turns out, Steve explained to me, in the America's Cup business, if something doesn't fail after a hundred hours or two hundred hours, you may be too heavy — you're going for lightweight. This is a hundred-million-dollar rich man's venture, or pride, or whatever you want to call it, right? To tell some other billionaire I beat you. And so you'll go hire some MIT professor in marine engineering, ocean engineering, to design your hull. One year — I think they won the Cup — Jerry Milgram was one of the designers, and another professor was on the other side. Then three years later they combined into one side to beat some other country. There was an active competition in the department among the faculty over who would do the best ship design. You guys have to do ship design in your third year here, right, the second-summer term or whatever? Well, these guys did ship design only, and they were being paid pretty well — far better than all of you put together, probably.
So anyway, if they don't fail fairly often, they're too heavy. You should make them thinner and weaker. But you want them to fail safe, not unsafe. Obviously breaking like that, you could kill somebody. So they want them to leak before they break. I've talked about leak-before-break before. So Steve came to me and said, what do we do? And I said, well, you've got a T6 treatment, and T6 is well known to be very susceptible to stress corrosion cracking. If on those aging curves I showed you, we went into overage, just past the peak — we could sacrifice a little strength and get a lot more toughness and corrosion resistance. That's called a 7075-T73 treatment. T7 means slightly over-aged. T6 means optimal aging, maximum strength. T7 is optimal aging for toughness, high strength, and corrosion resistance.
And the properties — you go look them up in the handbook. The properties: 63 ksi yield, 30 ksi √inch fracture toughness. My calculation to make everything work out in terms of wall size to hold 7,500 psi working pressure — it was operating at sixty-three percent of yield as the design criterion. Fracture toughness equals sigma times the square root of pi times C, the crack length. We want the crack length to be greater than the wall thickness, so it will leak before break. So K is 26 for the T6, sigma is 46. Solving that, with a 1.12 multiplier — you should know, M times the square root of pi C — you find the critical crack length is 0.0308 inches, which is three-quarters of a millimeter. So I told them the critical flaw size was about one millimeter. Well, that's the half crack length for a crack at the surface — an internally embedded crack would be a 2C crack, but on the surface is a mirror plane, so practically half is correct. So we proved the critical flaw size was one millimeter. We looked at the sample, and I could see it was about a millimeter deep before brittle fracture. That explained why I had a brittle fracture and the whole thing split.
Then we worked out: switch to T73, fifty-nine percent of 63 ksi yield, which is 37 ksi. Let's make the wall 0.140. So now you've got 30 ksi √inch over 37 ksi equals 1.12 times the square root of pi C. If you work it all out, C is equal to 1.16 inches, which is greater than the 0.140 inches. You should have a leak before break, if you just switch over to a T73 to get better corrosion resistance and better toughness. You lose yield strength — increase the wall a little bit. You want to be failsafe; sometimes you have to trade off. So I can have leak before break. We solved the problem together. And I told them my consulting fee was, I want the part to show the class. [Tom holds up the fractured cylinder.] It shows you an example of how you can work out a problem. They were going for lightweight, maximum strength. They had a failure, and they went back to re-engineer it, and you end up with a solution that says, I can solve this.
Okay, we'll see you tomorrow.