SSW_S2013_04

Class on Monday — it's MIT holiday, there's no class on Tuesday because I just, I got to go back to Cal[ifornia], Southern California. Now I just got back on the red eye, I got to go back. I'll be back on Wednesday on the red eye, but Dr. Belmar will lecture on Wednesday. You can just have the day off — this would have been a Monday classes. Um, everybody's watching different videos, so you can watch your videos or just skip things however you want to do it. I should be around to lecture Thursday and Friday, I actually will be around, and who knows, I might even come to class um on Wednesday, but I won't be lecturing, uh, Dr. Belmar will. And then I don't know what's going to happen the next week, cuz something else.

I wanted to show a couple of things here just to catch up on some things. I have to apologize — I actually was using my old lecture notes, and I found this morning my newer lecture notes. Um, but I talked about the Alexander Kielland disaster, and here's, uh, I open that up a little bit. This is the Alexander Kielland offshore platform. Um, and they might — I don't — oh, 50 m across. So that's, uh, 50 m is, uh, half a football field, right? 81 meters in that direction, 80%, you know, almost 90% of a football field. And I don't remember these things — these legs were pretty big. But in any case, if you looked at the legs a little closer, um, this was the hydrophone that they — you have these big columns of legs and then you have these cross braces, and then they put this piece on here. And it was 325 mm, so what's that, 10 in diameter. So that's 10 in diameter, and it looks like a little thing on this thing, so you can tell these things are pretty good size. Um, at the time it was one of the largest offshore oil rigs — this was probably early '70s. So, but it's not the largest anymore.

But in fact, there was this piece of carbon steel — well, the bracket was, they had — for the critical structures, what they thought were the critical structures, they had all kinds of quality control things in place to make sure they didn't screw up in the welding, and they used the right welding electrodes and the right welding procedure. But this hydrophone was just an attachment, and so the maintenance department was charged with doing this in this French shipyard, and they just got, went and got a piece of pipe. Well, this piece of pipe was like 1060 steel, it was very high carbon, which is difficult to weld. And they welded it, and afterwards — this is the hydrophone and this is the bigger pipe that it was welded onto — and they had fatigue cracks. And this is a little fatigue crack map going around. The crack ran almost all the way around before it finally let go. Um, and you lost that cross brace between the leg, then you lose one of the five legs. This was a tremendous storm that it was in, and the whole thing goes down in the storm, and 220 people I think lost their lives.

Um, the Norwegian commission that, uh, looked at it, uh, spent most of their time talking, mostly not about the failure, but about the lack of safety equipment. I mean, they did have a few boats — uh, you know, that these are boats that have covers on them, it's like sort of like getting in a clamshell, and who cares if it flips over, because, you know, um, it's watertight, and you ride out the storm, no matter how big the storm. But they didn't have enough of them. This platform had been designed as a drilling platform, but they'd converted it to a hotel, but they didn't increase the number of life rafts, okay, and things like that. Um, anyway, there's another one of my lectures I go through, not just that one but, uh, there's one in Brazil that was very similar a few years later, with some of the same safety problems.

Another one is the Comet aircraft, um. In this case, um, this was actually in the middle of the, uh, the 1950s. Anybody hear about the Comet? These things were —

Student: Yeah, what was it, you remember — low cycle fatigue.

It was low cycle fatigue, but the problem was, they were falling apart in the middle of the Atlantic, and so when — no one could get the evidence to find out what was causing them, they just break up in the middle of the Atlantic Ocean, and you know, there's no evidence. And finally this guy Alan Wells, who was head of the British Welding Institute and a fracture mechanics guy, kind of looked at it and realized the windows were square. Well, they had a big tank, they put him in —

Student: A big tank, right? A water-filled tank, they pressurized him and depressurized him for months, yeah, and they actually grew the fatigue cracks from there in one day.

Yeah, and it's — well, you basically got into brittle fracture and stuff, I mean, but they had cracks. This is something they, when they finally started doing it in the pressure cycles, and is low cycle fatigue. It's not all that different than Aloha Airlines, except Aloha was, uh, 37,000 cycles, cuz you were in Hawaii and it was all 15, 20, 30 minute flights from island to island. And so the plane had 40,000 hours, which typical Boeing aircraft would go 100,000 hours. But this one in terms of ground-air-ground cycles, you know, takeoffs and landings, it was 250,000 hours equivalent compared to everything else in terms of low cycle fatigue.

So in any case, the Comet, it was a stress concentration at the corners. So you don't — if you fly an airplane now and you look out, you'll never look out the window the same, um, because they have rounded corners, okay. They learned about stress concentrations, even though they had known about stress concentrations going back to, you know, um, 1880 was the first calculation of, uh, the stress concentration of a hole in semi-infinite plate. Anybody know what the stress concentration around a hole in a semi-infinite plate, if you pull it in tension? This was solved by some guy in 1883. If this is Sigma at the tip of that hole, just circular hole, it's 3 Sigma, okay. So you get a three-fold stress concentration from a round hole and semi-infinite plate. Uh, I guess it's actually an infinite plate, anyway, it's not just semi-infinite.

Um, some other things I thought I'd show you. When I talk about improving the reliability of joints — so I had a case, there's, it doesn't even — Boston Gas doesn't even exist anymore, but when they did exist, I think they're NSTAR or National Grid now or something. But the way they used to — if you go down to their training facility down here in, um, the South part of Boston, not South Boston, but the South part of Boston, you know, the South Boston is just east of Boston, right? Have you ever heard that one? You know, the North End is, uh — anyway, there's something that goes, you know, the South End is east, and the, uh, the southeast is west of Boston, the North End is east, and they go through all these things with the directions and they're all mixed up.

Anyway, but what it turns out is, there used to be a little — Boston was out on this peninsula and the Back Bay didn't exist. Well, it existed as a bay full of water, and there's little isthmuses of land. You know why Beacon Street is called Beacon Street? Because you had a right to graze your cows — even today, you can take your cow to Boston Common and graze your cow, it's in the law from back, you know, 400 years ago. But Beacon Street was a causeway across Back Bay so the farmers in Allston could take their cows to graze them, but it was a very narrow causeway and you had to have a beacon on your cow when you're bringing him back at night, so you wouldn't knock someone else walking the causeway off into the bay. Yeah, that's why it's called Beacon Street. Anyway.

Okay, so here are some pictures of some of the things that they dig up.

Student: Are they the ones that used to have that big tank that had the colors?

Yeah, that's the one with the blue silhouette of Ho Chi Minh. I told you that story, right? Anyway, I think I told you that story. Anyway, in any case, these are pictures of things they dug up out of the Boston street. These are old gas pipes that were abandoned in the ground in Boston. Okay, um, this is just a cast iron sleeve with two pipes, and the whole thing is filled with oakum. And this is a bell-and-spigot joint. And again, it's filled with oakum, just kind of stuff some old rope and, uh, tar in there. Um, there's another one.

Student: [inaudible question about lead-sealed joints]

Okay, well, that didn't come along till around 1900, 1910. I've worked on pipes that failed — water pipes that are lead-sealed, okay, that failed, you know, 90 years later. Um, these were kind of 1880s type of stuff. But I wish I had taken a picture when I was down there — they had one that was even older, it was a wooden pipe. They just take a log. The original street lights — in order to pipe the gas down the street, they would take a log and they would gun-drill it and put a hole in it. And it might be 6 in diameter log and it might have a 2 in hole in it. And you just hope that, you know, it wouldn't rot away before — anyway, they finally went to cast iron, and the cast iron lasts for a long time. Not that it doesn't leak at the joints, but that's another story.

But you know what gun drilling is? If you ever try to drill a really deep hole, the drill will walk on you and it will go off to the side. Gun drilling — if you may need to make a long hole in a gun barrel, you basically rotate both, so you're average out any asymmetries in the force and cutting action, sharpness of the teeth and everything else. And you can drill a very straight hole with a gun drill. So why on a lathe don't the end stocks rotate?

Student: On a gun-drill lathe they do.

Okay, that's the only difference between a regular lathe and a gun-drill lathe, is the tail stock rotates on a gun-drill lathe. Not a lot of gun drills around anymore, but anyway, we do extrusions and other things.

Um, I wanted to — I'm just, I'm catching up on a couple things. By the way, does anybody have any questions? Okay. Uh, I wish I had saved it. I remember reading it in some magazine, but the nano guys with the nanofibers and stuff, uh, there was one group that were claiming that they had strengths of 10 million PSI. Well, that would be a bond energy of about 10 electron volts, if you remember my little scale. What two elements have a bond energy of 10 electron volts? None. Okay, there are physical limits, okay, and that is an absolutely absurd statement. But, you know, people, some scientists will say, oh well, we've, we're working on nanofibers and these things, we predicted they have a strength of 10 million PSI. I haven't seen your prediction, folks, but your prediction defies all of known chemical thermodynamics, okay. But aside from that, it's okay, you know.

Um, actually, was one of you working on little micro-catheters? Yeah, is that you? Okay, one of the pro — micro — I should have brought my micro-catheter stainless steel piece, but micro-catheters. Does anyone have a parent or a grandparent that has a stent in their heart, in the artery? And then, how do they get in? Did they tell you the story too, Young? Okay, well, they take one of the arteries in your leg, and they go in there with this long stainless steel Teflon-coated tube or wire, and they're taking flash x-rays while they're doing this so they can see. In fact, they put a little bit of barium on the tip of that thing so they can see exactly where it is in your arteries. And they basically are taking images, a movie, x-ray movies, while they're doing this. This has really been a great boon for radiation, you know, x-ray people, medical doctor x-ray folks, they putting stents in. It used to be the surgeon get did open heart surgery. Uh, now these guys do this, much less invasive stuff. But they go in, um, with this thing and they end up — then they call that, the first one they put in is a guide wire, and then they could, they use that kind of little trolley wire to bring other things in, and they eventually put the stent in there.

Where was I going with that? I don't remember where I was going. Oh, stiffness. One of the problems with these things is not just that, but also the arthroscopic knee surgery or something — you're going in with very slender tools. But what happens is, something gets slender and slender, it wants to bend. You know, the stiffness goes as the cube of the diameter, you remember that, right? No, you don't, do you? But anyway, you now know — I remember a lot of things, okay. But in beam theory, a simple beam, the stiffness — you know that, right? Okay, see, he's a structural guy, you should have known it too. Okay, anyway, goes as the cube. So you cut it down, you make it slender by a factor of two, you just lost a factor of eight in stiffness. And so you want a nice flexible guide wire, but when you're doing arthroscopic surgery and you're trying to go in there and cut out the cartilage in someone's knee or something, you need something that's stiff.

Well, steel's got a stiffness of 30 million, tungsten has a stiffness of about 60 million, and molybdenum's about 50 million I think, sapphire is about 60 million. But 60 million is about the top — there might be something like iridium might be at 70 million, but it's a little, it's pricier than platinum. But in any case, there's only about a factor of two. And, you know, medical company came to me once and said, well, how do we get a stiffer part? And I said, why don't you use sapphire? And I wanted to — I still want to do this — you can buy single crystal sapphire and it won't wear out very much, it's very okay, very strong, you know, it'll hold its edge. And you're going in there with this little thing that's away the cartilage, but you want it to be thin so you don't have to make a very big slit in the person's knee, and you want it to be stiff so it's strong enough to do this nibbling. You can get a factor of two in stiffness if you went to sapphire. But do you know what happens if you add just a little bit of chromium to sapphire? It's sort of milky white, and it becomes ruby. Can you imagine selling surgeons sapphire and ruby instruments? They're color-coded, which is a wonderful thing, right? They like color-coded instruments. So anyway, I never could convince them, I never could convince them they should make, use sapphire for making instruments.

Um, so catching up on other things. Oh, um, oh, first of all, if you weren't here, we don't have class on Tuesday, but we will have class the other three days next week. So we've talked about — there are two fundamental reasons why something won't bond. And I talked about friction welding as a sort of a form of cold welding. It's not really cold welding. I actually found my notes, which is just out of the welding handbook. Uh, there is a video that should be online — I don't know if it is on my website, but there's supposed to be five or six videos, and one of them is a commercial video on friction welding. You actually get to see a friction welding machine run. But this is the tooling, and it's just a simple circular friction weld, and it heats up really quickly as you spin it together. It's basically just a fancy lathe, okay.

See, and here's the kind of force-displacement. The welding starts — one of's rotating, one of's not, the completed weld. You have all this flash that we passed it around once. Where'd I put it? [Tom searches for the sample.] If we didn't, p— if you haven't seen it before, you can look at it. There's the flash. But the whole process is, you accelerate — you have speed, force, and upset all being plotted on different graphs. So you have accelerate to some friction speed, you start pushing the two things together, okay, with some force, and that increases, and you actually end up finally jumping the force to forge it together. And you're making a hot forge weld, but in a sense it starts out as a cold weld, and you're using friction to get enough heat to melt down these asperities, cuz you got to break down the asperities and you extrude out the contamination.

Student: What kind of tolerance do you [need] on the flatness of that surface, just so it doesn't, like — you mean when you start it, cuz I suppose if it's off by a little, kind of kicks and that can be really —

Right, well, a lot of people actually put a taper on it, okay. They actually make one of them conically shaped, and I'll show you the reason for that.

Student: Yeah, but I mean, like, instead of, um, like, what happens if it's, if the end is cut off at an angle so it's not —

Oh, you don't — you actually have a machined end on these things. You don't just do a sheared end, if that's what you're thinking of, okay. A sheared end would be terrible, you'd be very inconsistent results, cuz you're only going to be heating up one area. But in fact, just trying to put two flat surfaces together doesn't really work that well. Most people put a very shallow conical machine on it as far as that goes. Does, in fact, what you can see here — this is a bunch of them, and you can see right in the center, well, I have no shear in the center. There's, as I go, as R goes to zero, the speed goes to zero, and you actually can see the heat patterns. You're not really getting a good joint right there unless you start with this cone, okay. And then now you have to vary your forging force and things like that, but a good machine will do all of that stuff. And some of these machines can cost millions of dollars, okay. Um, but they make very — they make good parts very quickly, or they make scrap very quickly, okay.

This is actually out of the welding handbook and kind of shows what you're doing. You've got a very rough surface, sort of like my little foam things there, and you're touching them together, and you're getting better contact. Turns out — I'm going to show you, maybe even later today — that if you just do a straight normal force, you'll never get more than about one-third bonded area on these things. You have to have shear. You have to plow these things, these mountain peaks, away. If you just squeeze down, by the time the base material starts to yield, you still have about two-thirds, two-thirds voids, okay. So just a normal squeezing of things doesn't work. And we'll get into that in a little bit more detail.

Um, I wanted to mention two other types — well, three other types of friction welding. Probably one is linear friction welding. I mean, this is circular friction welding. It's nice and easy to build a machine that goes in circles, but they would love to do linear friction welding. And if you want to do some friction welding, you can do it with your hands — you feel your hand is warming up, okay. They should do this — they would love to be able to make turbine discs for jet engines, where they weld the turbine blade to the substrate. Um, a regular turbine blade — here's a regular turbine blade — and the jointing is just a mechanical joint, and they call this the tree, cuz as you know, turn it upside down, it looks kind of like a tree. Um, but this is ground very precisely. I mean, this is the most expensive part of the turbine, I think — $10 million grinding machine to grind these flats and stuff. And you mount the whole thing in Wood's metal, which is a low-melting, uh, metal, to fixture it, because you can't clamp this precisely enough to get the tolerances that they want on this. Has to fit into the mechanical, you know, sleeve on the turbine disc within a few tenths of a thousandths of an inch, because otherwise at the speed it's going, you'll get vibration, you get fretting wear. So it has to fit very precisely. Um, and you have to be careful at what temperature you're trying to assemble these things, because a few degrees temperature, you basically have to, most cases, assemble these things in a controlled temperature environment.

But in any case, over half the weight of this thing is the mechanical attachment. And then, if you look at the turbine disc, most of the weight is, or the most of the bad weight, out on the outer edges of this rotating thing, is the big heavy section. Looks like a railroad wheel or something, with the outer edge being large. You can — if the value of a pound saved in an aircraft structure is $2,000 — $200 a pound — the value of a pound saved on the engine is $2,000 a pound, because that turbine engine is going very fast. One of my rules of thumb is, the faster something moves, the more valuable the weight savings, right? It's just F times velocity, or force times velocity, or mass times velocity, uh, to get your energy that requires to move it. So the Air Force spent tens of millions of dollars trying to make bladed discs, which are called blisks.

Okay, so this was back in the '80s and the '90s, blisk technology. You can probably look it up and Google it. Um, it's a bladed disc, and one of the ways they wanted to do it, cuz arc welding, well, it's hard to get in there. You got a hundred of these blades around a rim this big, and it's kind of hard to get in there and weld. But they tried electron beam and laser and arc welding, they tried linear friction welding. And the great thing about linear friction is, if everything's working well, it works great. You have to have a fairly expensive machine to oscillate back and forth, okay, uh, it's even more expensive than a simple circular machine. But the problem is, if you make one bad weld out of 100, you now have a piece of scrap. So the economics never worked out. If you can figure out how to get, uh, Six Sigma quality welds on linear friction welding, you could make a lot of savings for the airline industry and engine technology. But they never solved it.

There's only one engine that I know of in commercial production that uses a blisk, and it's a cast blisk. And the — I wish I could get one, it's about this big around. I — anyway, it's made by Rolls-Royce, it's the M250 engine. It's used on helicopters and small aircraft, so far as that goes. But it's actually probably the highest production jet aircraft jet engine in the world. But it's a small helicopter or, you know, small plane like a Lear jet or something like that. I don't know if it's on a Lear, but anyway, um, but that is one example of blisk, but that's a cast, uh, um, part.

Well, why else is friction welding important? One of the reasons is, well, or cold welding. I mean, you say, why is he spending two lectures on cold welding? First of all, it's not just cold welding, we're talking about surface roughness and surface contamination. And cold welding, I don't have to worry about the heat and, you know, all this other stuff. In fact, I probably should have gone through that with you at some point — what is a weld? The American Welding Society actually has the definition of a weld. I gave a whole keynote lecture at a welding conference on this definition. Uh, weld is a — you know, if they ask you to give a talk and you don't have anything to say, make something up, right.

So I actually — I gave the paper that I wrote was called "In Search of the Perfect Weld." But I started out by taking the American Welding Society definition from their glossary of a weld: "a localized coalescence of metals or non-metals produced by either heating the materials to the welding temperature with or without the application of pressure, or by the application of pressure alone, and with or without the use of filler metal." So if you analyze that, this is what you get. A weld is something that can be metal or non-metal, can have temperature or not, can have pressure or not, and can have filler metal or not. Isn't that a great definition?

Anyway, you also have to understand the psychology of this. The guy who wrote most of the AWS definitions was the guy who's passed away now, Hallock Campbell. And Hallock was a Harvard grad, and uh, he was a real stickler for words, okay. And this is back in 1985, I was on sabbatical in Japan, and I had written a paper in the Welding Journal talking, I used the word "parameter" in the title. And, uh, I get this letter in Japan from Hallock telling me that I'd used the word "parameter" incorrectly. And, uh, so I read his thing, and I went to the office I was in, had an unabridged, you know, American English Merriam-Webster dictionary that was about this thick back those days. And, uh, so I looked up the word "parameter," and there it had four or five meanings, but it gives examples, and it had an example from Percy Bridgman. Anybody know who Percy Bridgman was? A professor at Harvard who won the Nobel Prize. He was the guy to use the tetrahedral anvil press to develop huge pressures to do geological studies of minerals and phase transformations and stuff. But he won the Nobel Prize for his studies at ultra-high pressure. These are like gigapascals of pressure, okay, millions of PSI. What happens to — and that's how people make artificial diamonds now, using Bridgman's tetrahedral anvil press. Well, Bridgman was a professor at Harvard in the '30s and maybe into the '40s, and Hallock Campbell probably was one of his students. Um, but I wrote back to Hallock, and said, with a copy of that, the quote, the definition from the book, and said, this is how I used it and this is how P.W. Bridgman used it. And he wrote, he just handwrote on the back, and sent me back a note, says, "Percy, how could you?" Okay.

But, so this guy, I mean, this is a H. Campbell type of, um, definition of a weld. So I actually use that as the beginning opening of my talk, that they needed — actually, you know, this is sort of irrelevant to — well, some of the things I talk about are not totally irrelevant. It's sort of, um — and I sometimes talk about, what can you say about something when you know nothing at all? And there's a number of things — you know, they wanted me to give a keynote talk, I'd given keynote talks before and it's kind of running out of topics, so I decided to talk about "In Search of the Perfect Weld." Well, first you have to define the word "weld," and so I went and I looked it up and I saw this is a silly definition. And I actually built the talk around — the problem with his definition is, he was trying to define a weld in terms of how you do it, not what you're trying to accomplish, okay. He was trying to define things in terms of how.

And, um, I came up in my talk with a different definition, which so far as I know the American Welding Society has not bothered to adopt. A weld is the joining of two similar materials, uh, for which the properties of the weld are substantially similar to the properties of the base materials, okay. Because solder and brazing, you have weaker material in between, adhesive bonding your joint is weaker, okay. A weld, you have substantially similar strength, okay. And we talk about diffusion bonding — and sometimes we call diffusion bonding — this is a diffusion bond in platinum for a turbine blade for the F22, I think I showed it to you once before — we sometimes call it diffusion welding because this is actually a near-perfect weld in that the properties across that joint are identical to the base material if you do it properly. So a weld, I define it in terms of properties not process, if you will. So you know, even though I knew nothing about it, I could cobble something together, okay.

Um, the last thing I want to talk about on cold welding specifically — and see if I can find — somewhere in here I have a diagram of this. But we actually do use cold welding in a few cases. Well, actually two things I want to talk about. One I haven't talked about, um, one is we use it to make wire bonds. Well, I'm not seeing my wire bonding drawing right now. But if I want to, if I'm trying to draw wire — and if you go over here to Newton, Massachusetts, there's a company H.C. Starck that is one of the world's largest producers of tantalum, okay. And this is, uh, H.C. Starck is part of Bayer Chemical in Germany, one of the world's largest chemical companies, rivals DuPont in size. Um, but it turns out, this factory over here in the middle of Newton, the residential area of Newton, was started after World War II, probably because of the MIT mechanical engineering department, okay. The mechanical engineering department developed a way to develop, to generate very good vacuum on an industrial scale. And as a result, Norton Research Company started looking at being able to melt refractory materials. And I don't know if this work came out of the Manhattan Project or not, it might have, because they were trying to melt things like uranium, which is very reactive, you need to melt it in a vacuum. But in any case, they started working on refractory materials. Watertown Arsenal over here, which is now Watertown Mall, um, developed, uh, titanium alloys. And the workhorse titanium alloy — titanium six aluminum four vanadium — so about 1950 they developed titanium alloys, all because of this technology just to generate a vacuum on an industrial scale, when you're melting metals that are giving off all kinds of gas. You had a big enough vacuum pump to suck this stuff off. And Norton Research started right down here in the end campus, and the building is actually the building that's between the Wang Center and the, uh, um, the river. And that was the Norton Research building. Now it's part of the Sloan School.

Student: [Was] that Norton —?

No, different Norton.

Student: [inaudible]

Um, no. There were two Norton brothers who were faculty members in this department, but it's not either one of them. I don't know which Norton that is, but it's a different Norton.

Student: [inaudible — something about when it started]

Yeah, yeah, yeah, but in any case, this is — I'm pretty sure this is — I'm almost positive this is a different Norton. But anyway, my thesis adviser in the 1950s did his doctoral thesis on single crystal tungsten, in deformation of single crystal tungsten, because we all of a sudden had vacuums that we could grow single crystal tungsten and niobium and all the refractory metals, and titanium light metals and stuff. And tantalum is one of those such metals. And so H.C. Starck over here in Newton — I mean, these — this is kind of going back 70, 80 years, but how one little development out of mechanical engineering at MIT has spawned off a number of different companies.

There's another company started out in Concord, which is pretty much gone belly up now, but it was called Nuclear Metals, and they used to extrude, uh, depleted uranium for all the bullets in the world. But after the first Kuwait — World War, uh, the first Gulf War, and Kuwait got contaminated with depleted uranium, and they weren't real happy. Um, uh, the Army has been spending lots of money trying to come up with tungsten alloys that will replace depleted uranium as bullets. Anyway, um, they make tantalum wire, uh, they make all kinds of tantalum products. But I got this from them, and basically they just take in a die that you can actually — if you look closely on here, you actually see some of the die marks where it grabs this and just keeps on pushing it together as a cold weld. And you can see, just, they kept pushing it together about 10 times to exclude the contamination. And it's just cold-forged. Tantalum can be deformed probably 99% without cracking, even though it's a refractory metal. It's very ductile. Tantalum and niobium are both extremely ductile.

Um, let me see if I — [Tom searches for a figure.] there's, I wonder if I — oh, here it is, here's the, uh, this is the — this comes out of the welding handbook, but it's the technique, it's the tooling that's used. You grab a wire and clamp it in a vice, and you — they actually kind of push them at an angle and just squeeze them together and just forge it over and over. And what you end up with is, that then you shear off the, uh, the flash. And you just draw the wire. So they have a continuous drawing operation that can run for weeks without stopping. So you're just drawing wire continually, but you have to make these welds, um, in between. Um, you don't have to stop it, because they actually have a little accumulator that just runs a coil, and the coil keeps on feeding while this thing is making the weld, and hopefully you get the weld made before you run out a coil in your accumulator. There's a number of metal working processes that use accumulators to be able to do a batch-type operation following into a continuous flow operation. So as far as that goes.

Um, but in any case, the last type of cold welding that I want to talk about is actually probably the most common welding in the world, at least in terms of number of units. Anybody have any idea? I estimated once, 20 years or so ago, they make 50 trillion of these welds a year.

Student: [inaudible guess]

Yeah, no, that's not a cold weld. This is actually wire bonding in microelectronics. So if I have my little tab-bonded tape — this is the Intel thing that I passed around before — and there are 400 input and output lines on there, 100 on each of the four sides. And we'll talk about that later. This one's soldered. But what they will do, after they've soldered the chip to this little composite polyimide-copper tape, they will then, to join it to the rest of the package, they will come in with a one-thousandth-of-an-inch diameter gold wire, and they will squeeze that wire onto the surface and push it to the side. They often add a little bit of heat, like 2 or 300°, and it's called thermocompression welding, okay.

And here's a picture of it from the welding handbook. This is actually one that they started to break apart. But you basically — the tooling looks sort of like a sewing machine, except you're using a thin gold wire rather than a thread. And uh, you feed the stuff through, the head comes down, squeezes the gold into the surface, and actually spreads it out. So here's the 1-mil diameter wire. See this non-bonded region in the center? That's called the dead zone. That's where you have a straight normal force and no shear. The real bond is really out here in the wings. All we're looking for is to carry nanoamps or microamps. So this is how they do it. And we'll talk about it later about how they do these. But there's a company called Kulicke and Soffa in northwestern Pennsylvania, and they have about 90% of the world's market for making these machines. And you can buy one for $50 or $100,000, and it will, it's like a sewing machine feeding a gold wire. And it automatically indexes this thing, and it makes about seven bonds a second, okay, and it's really kicking along. They might be up to 10. But you actually run into the problems same type of stiffness and vibration problems are the limiting how fast you can go. I mean, you can try to run it faster, but now just the mechanical inertia of all these parts moving up and down — and so you have to try to make the heads, the sewing machine heads, lightweight and things like that, because of the inertia.

I guess I have a couple of other things here, but — oh, here's some other — here's a book on wire bonding and microelectronics. So wire bonds are, I say 50 trillion. Here's the — a gold wire bond, an ultrasonic wedge bond. I mean, we talked about ultrasonics — you can improve this problem if you do it with ultrasonics. You can't make seven a second if you're trying to do ultrasonics along same time. You might make one a second, okay, or two a second.

Student: Process — par[allel], me, parallel process it, like will they do a four next?

Not so much because the size of the heads. In fact, we'll get into, when we get into designing microelectronics, and how do you get more input and output bonds, you're limited to about four mils between each bond pad center-to-center because of the size of the head, the sewing machine head. You go any smaller, you get the same stiffness problem we talked about on the tools that go into your knee for, you know, surgery. You got the thing doesn't have enough stiffness, it's going to bend, you know. You lose your mechanical, your geometric stiffness, okay. And they do things like sapphire tips, okay, because there's a lot of wear when you're doing, you know, seven of them a second. Uh, even though it's gold and stuff, but um, so there's a lot of technology, because this is the bread and butter of Intel and other people.

And, that you might have to make 400 bonds, but if you make one bad one, you got scrap. So you can take a $500 chip — remember I said one of the problems of joining is, if you don't do it well, it comes at the end of the manufacturing process where the cost of scrap is high. You can make a $500 chip and has to have 400 good bonds, and if it has 399 good bonds, it's a $0 chip, okay. Actually it's not zero — they actually take the bad chips, they put an ink spot on them, and they sell them as keychains, don't they?

Student: Also I heard that they have some that, when it was — it's something like when they have a Pentium 4 and it gets part of it goes bad, that it's a Pentium 3 basically.

That's what they've started to do. This was a Pentium 1, this was 1995 version. In fact, the reason I had these — in fact, if you were a little older, you could have had your own. The, uh, when I first became department head in like '95 or '96 — some of you were still in elementary school or whatever — but they had a — they used to bring groups of high school students through MIT, and like 500 of these students from around New England and stuff, top science students in their high schools, and try to give them a Saturday at MIT. And they chose materials when I was department head, and so we had to put on a dog and pony show for these high school science stars. And so I actually, I got some Pratt & Whitney to give me about, uh — maybe it wasn't 500 of them, maybe it was 200 students, but um, I got enough turbine blades, I gave every one of them a turbine blade, okay. These were defective turbine blades — not defective, but worn-out turbine blades. Took a little problem, because you have to be careful about turbine blades. If this blade is worth $6,000 — now this one's cut in two, but even this one, it's got a corner cut off of it, and that tells a mechanic, don't put it into an engine, there's something wrong with it. And so they actually do — this one's got a notch right here, don't put it in an engine. This one did not pass quality control, okay. But when these things are worth four or five or $6,000 a piece, you know, there's a great incentive for people to steal bad ones and put them in engines, which could cause the engine to blow up, okay. So that's a little quality control concern. That's not a problem with a Pentium chip.

Uh, but getting back to your question, starting around Pentium 2, I remember I was at Intel when they were designing the Pentium 3, this was probably around 2000. And actually, we'll get into this — the Pentium 3 was a chip this size plus another chip right next to it at about half its size. Now Pentium 4, they were able to put everything back into one chip as they miniaturized it. But as they got to that density, they started making things redundant, so that if, when you made the chip, if this area of the chip didn't work, you could just, you could re— you could reprogram it so that you had — and you had some redundancy on the chip, because otherwise you're trying to make 10 million transistors and every one of them's got to be good, and one of them being bad, and you haven't got a good chip. I — this isn't a bonding problem, this is just a processing problem. And now they're up above what, half a billion chips or something like that, or transistors on a thing. And so yes, it was, sort of about 10 years ago, the big thing was to design redundancy in, so that you can basically have something that — it's like, I got two kidneys, right? If one of them dies, then I can still live, right? Well, they design chips like that. You know, one kidney goes, then you can still go to the bathroom or whatever. Chips don't go to the bathroom — anyway, they might produce the same stuff. But um, okay, that's another story.

Um, there are other types of cold welding, uh, I forgotten I even had all these things. One is explosive bonding, okay. Um, hold — time, okay, oops, time to finish up here. Uh, but let me just quickly tell you, and we'll talk about more — this actually was discovered during World War II, and they had bombs go off on ships and, or, you know, blow a tank apart, and they would find two sheets of material bonded together perfectly by the explosion. And so they did a bunch of research after World War II. DuPont did a lot of it, because DuPont started in the Revolutionary War making gunpowder, and they are still a big — most of the explosive plants in the United States are in the Pennsylvania-Delaware area, okay, because that's the home of DuPont, that's where explosives manufacturing kind of got its start in the 18th century, late 18th century.

But anyway, an explosive bond — they found, if you're at the right angle, if you take a plate of steel — or actually, let's say you want to take a piece of a plate of tantalum. That tantalum has a value of approximately the same as silver, okay. Tantalum is only one of two materials that can tolerate concentrated sulfuric acid without corrosion. Every ounce of gasoline in the world is made in either a vat of tantalum, okay, because they use concentrated sulfuric acid to get rid of the water that's in the gas from the cracking of the, uh, when you break down the heavy hydrocarbons to make octane or such things, you end up generating a bunch of water, and you put it in sulfuric acid, will suck the water out of that. And so sulfuric acid, they call it a catalyst, okay, but you have to use tantalum tanks. Well, you can't build a great big pressure vessel out of tantalum 3 or 4 inches thick — I mean, no one could afford gasoline. So they take a great big steel tank, and they clad it with like a quarter inch of tantalum, and they do it by explosive bonding. And what they do is, they put a layer — they have the steel plate maybe 4 in thick, and they go to this place called Explosive Fabricators outside of Boulder, Colorado, and they have their own little valley between the mountains where there's not a lot of neighbors, and they can set off explosions without a lot of people complaining. And the top sheet of tantalum will have a layer of explosive on it, and you set off the explosive. You actually put it at a slight angle — the tantalum plate to the steel plate, just laying on the ground, you have the tantalum plate at a slight angle, you set off the explosive, and the explosive actually burns in one direction. So you're just — you're forging these things together, okay. That's called explosive bond, and I'll tell you more about the technology of how it's used, um, next time. Okay, so no — [recording ends]