SSW_S2013_08

Okay, so we were talking about diffusion bonding, and I mentioned that it's actually one of the only ways we can really make essentially a near perfect joint. Okay, if you do a good diffusion bond, it is virtually impossible to do metallography and see where the interface is, do any type of mechanical testing and not tell the difference between the joint and the base material. So it's essentially a nearly perfect weld.

Now, we have to, it's also used in many cases for joining of dissimilar materials, because you can, as long as there's no brittle intermetallics that form. For example, one of my examples here is tantalum to vanadium. Tantalum and vanadium are very close to each other on the periodic table but they form a brittle intermetallic, and you can diffusion bond them just fine, they dissolve away their oxide at higher temperatures, but then you form a brittle intermetallic and you tap that joint and it shatters because of the brittle intermetallic.

Now it turns out if you put a sheet of niobium in between, which is also close to these — in fact tantalum and niobium are called the twins because they're found in the same ores. Don Saturday's [Satterthwaite's?] doctoral thesis was separation of tantalum from niobium, okay. Um, tantalum is named after Tantalus, some Greek god, from which tantalizing and other stuff came, and niobium was the, uh, a twin of Tantalus, and so niobium was named because it was a twin of tantalum. Very difficult to separate.

But it turns out niobium does not form an intermetallic with vanadium, but tantalum does. So if you put a sheet of niobium in between and now do a diffusion bond, everything's fine. Okay, you get a good bond. Well, that's really useful because how many people are trying to make tantalum to vanadium joints? I know of no commercial application for that. But it does illustrate the brittle intermetallic, and I've got another example of that in a little bit.

Um, and another example that I want to talk about is the problem when you're bonding dissimilar materials, the problem of thermal expansion mismatch. Okay, because that's inherent in lots of things other than just diffusion bonding. Lots of your electronic packaging you've got dissimilar materials. Now, but before I get there I want to talk about some of the problems of diffusion bonding. Your fixturing pressures are typically about 5000 psi, okay, you're trying to break down these disparity contacts with pressure, okay. You're going to do it at elevated temperatures so things are softer, but you really have to operate somewhere between six tenths and eight tenths of the absolute melting point.

[Tom writes on the board.] Should be greater than six tenths of the melting point in Kelvin, and less than about eight tenths of the melting point in Kelvin. So, why do you usually do it above six tenths of the absolute melting point?

Yeah, exactly. Much less than six tenths of the absolute melting point, diffusion takes weeks if not years, and so it's too slow a process. In fact a typical diffusion bond, pure diffusion bond will take somewhere between 10 and 20 hours in a furnace, and typically these furnaces, which are often vacuum furnaces or inert gas furnaces, are costing you something like a thousand dollars an hour to operate. So these are not exactly automotive parts. In fact I can't think of any diffusion bonded automotive part. There might be some titanium fuel injectors in some high-end racing car or something, but in general you can't afford it, because the number of parts — it's a batch operation and you're paying $20,000 for one batch of parts in a furnace, okay.

I think I did know of some fuel injectors. Chrysler had a sports car and the fuel injectors were made in Italy, and the real problem was — happened to be an O-ring — but I think they may have had some diffusion bonding in them, and they were actually flying them over from Italy to Detroit in the Learjet because they were having so many of them fail because of these O-rings that were, when they were sticking them in, the O-rings would get cut and then you had a leak in your fuel injector, and that wasn't good. But they were, it was holding up the production of the automobiles, okay. But in any case that wasn't really a diffusion bonding problem.

But below six tenths of the absolute melting temperature, diffusion is just plain too slow. You don't want to be in that furnace for two days or three weeks. Above eight tenths of the absolute melting point — anybody have any idea why you can't go too high? Obviously you get lower forces if you went higher. It actually gets back to this diagram. This is stage one, this is stage two, this is stage three. At stage three everything is over. You have to stay below far enough below, or close enough to — let me put it the recrystallization temperature of the metal — so that you don't start getting lots of grain growth. You want to have diffusion but you don't want these grain boundaries to move too fast, because if they do and the grains get too big you get to stage three too early and you're full of porosity at the interface.

So a really good joint you're going to have to do somewhere in this range, and you can start figuring it out. For something like titanium or something, you're probably in the 1200 degree Fahrenheit range or something. You can calculate it yourself so far as that goes. But the point I really want to make is, diffusion bonding is with 5000 psi fixturing pressures, which may mean you have to have a big press, or they use other things for fixturing. A lot of times diffusion bonding is used in the aerospace industry as opposed to the automotive industry. Why? I told you the relative cost of things. The value of a pound saved in an automobile is two dollars over the life of the vehicle, in the aircraft it's two hundred dollars, in a spacecraft it's twenty thousand dollars per pound.

Well, of course you can afford diffusion bonding in aircraft engines where it's probably more like two thousand dollars a pound, okay. The reason the engine is more than the frame of the aircraft is because those engines sit out on those wings, and every pound you take off the engine means you can take extra weight off the wings. So there's kind of a follow-on. If you can take the weight out of the engine, you can take it out of the wings too.

And remember I talked about linear friction welding and blisk, where they were trying to attach the turbine blades directly to the disc, to get rid of this big heavy mechanical structure, which has got extremely tight machining tolerances. Some of the tightest tolerances in commercial manufacturing in general are how these little curved Christmas tree, they call it, inserts fit together.

Student: [inaudible question]

No, well, you actually have to slide them in very carefully, but the tolerances are on the order of one ten-thousandths to ten-thousandths on that ground surface. They fit and they are ground. And if you go through Pratt Whitney's manufacturing facility — well, this is 20 years ago — they actually had a bunch of little cells that would do different turbine blades, and they had gone from a plant half the size of a football field, and they used to have all the milling machines over here and the laser drillers over here — and we haven't talked about laser drilling and stuff — but different machines in different parts, and then they put them into cells, so they could, instead of a part having to go like two miles from the incoming casting to the outgoing part, they got it down to 100 feet or 100 yards or something, from a couple of miles, okay. So they simplified their production, streamlined their production.

Except they only had one grinder. It's like a 10 million dollar grinder for grinding these things. 10 million dollars, you can only buy so many. Plus it was a machine that was the size of this room, and it wasn't just clamping the parts, they actually cast them in Wood's metal. They held them in Wood's metal, which is a bismuth — 50 bismuth alloy that actually expands when it freezes, so there's no looseness whatsoever, because when you're trying to hold one ten-thousandths, okay — one ten-thousandths is about two microns, okay, two and a half microns — you've gotta have very good clamping, okay.

Um, actually I got another story on that, which went with — in 1990 I bought a Ford Taurus, okay, and I bought it in February. In May I turned on the air, tried to turn on the air conditioner, no air conditioning. So I took it back to the Ford dealership and they said, uh, we don't have the part. And I said, well, then give me a new car, it's under warranty. And they said, well, we can't do that. I said, well then you better figure out what to do, because you got a problem. I mean I'm not going to go through the summer while you tell me I don't have an air conditioner. I'll take a new car. You want when you go lemon law? Okay, I'll bring it in three times in a week, and if you don't have it fixed by the end of that week I can go lemon law and you'll give me a new car, okay. So they decided that maybe they could do something. Plus I told them I knew a few vice presidents at Ford, which I did actually.

Well, I didn't learn the story from them, I actually learned it from Kim Clark, who later was the dean of the Harvard Business School. But apparently when they designed this compressor, they had a single shuttle for the compressor, and it was a very simple air conditioning compressor, new design, but it had extremely tight tolerances for the piston inside this thing, and it had to be machined just so. And so the guys in the development at Ford who had designed some of this, they figured out the fixturing — just like you had to be careful how you fixture the turbine blades to get these tolerances, you can't just put them in a vise and clamp them, it doesn't work, the vises wear out and the tolerances change and everything. So they came with this very special fixturing. They sent it off to the machine shop, the part came back, they tested the prototype, everything worked fine. Decided to spend 100 million dollars to put in the production line based on their fixturing.

Starts coming off the production line and none of them work. And who buys one of these but Tom Eagar, right? And of course they go back to try to figure out why aren't they working, and it turns out they went and they talked to the machinist, and they said, well, did you fixture it the way we told you? He said, oh no, I knew that wouldn't work, I figured out my own fixturing. Okay, and you never bothered to tell anybody. Okay, so they designed all the fixturing in this hundred million dollar plant — not that it was 100 million dollars worth of fixturing — based on what the engineer thought was proper fixturing, when in fact this guy had been a machinist for 30 years, knew a lot more about how to fixture something. So they had to redo the plant. In the meantime summer was coming. I finally did get a new air conditioner within about three weeks, so it wasn't too bad for me.

But fixturing for very precise machining is not a trivial problem, okay. Ford learned the hard way, and Pratt Whitney certainly in making these blades, I'm sure they learned the hard way over time too. But there are different ways to do it. But in any case, I'm supposed to be talking about diffusion bonding.

But I want to talk about — there are high pressures, 5000 psi, long times, and expensive furnaces, and so you're really talking aerospace type of components rather than automotive, because of the value added to something. Oh, I know how I got on that.

It turns out the blisk technology, the linear friction welding, the reason you want to get rid of that big heavy rim on the end of the disc — aside from the fact it's out towards the edge and lots of mechanical energy needed to spin that thing around, and high stresses out near as it's spinning around at 20,000 rpm or whatever — you'd like to get rid of that weight. And it was considered that every pound you could take off the disc would take 20 pounds out of the engine and would take 200 pounds off the, uh — let's see — 20 pounds out of the engine and 200 pounds off the aircraft.

If you're talking about a commercial aircraft, you say, well, 200 times 200 is only 40,000. But if you're talking military aircraft, you're really talking about a thousand dollars a pound. So you're talking two hundred thousand dollars per aircraft, and if they have multiple engines, you're talking a half a million dollars. Now, yeah, nowadays the F-22 and the F-35 are costing 200 million dollars apiece, so you may say that's not that much, but it's still worth something in terms of these things. So the numbers add up pretty quick when you start getting into some of these aerospace components in terms of the cost and whatnot.

Well, it turns out, we like diffusion bonding, but it's slow. But there is a technique that was patented by Pratt Whitney in 1972, and it's called transient liquid phase diffusion bonding. And instead of just having a dry interface between two pieces of titanium, or a piece of titanium and a piece of steel or stainless steel or whatever you're going to diffusion bond, you have this oxygen — actually I have a nice little book here somewhere as a picture.

[Tom locates the book.] Here it is. So you got the oxide layer — this is actually out of the chapters of this Welding Institute publication on diffusion bonding. You got the oxygen interface, you squeeze it together, you break down some of the mountain points, you still have some vacancies, you diffuse away, you see the oxide layer getting thinner, and eventually you end up with your diffusion bond. So that's that as far as that goes.

In TLP, what we're going to do is we're going to introduce a liquid in between, so you don't have to have all this heavy pressure to break down the mountain peaks. Okay, if you interpose a liquid, you may have a thicker joint, but if you're smart about what liquid you interpose, you can introduce a brazing alloy in between. And typically on nickel-based superalloys, you can use nickel boron interlayers, which is exactly what Pratt and Whitney — they actually patented the process. This is like 1974 paper, you should have it in your handouts, but it's in the Welding Journal, April '74. It was presented at a conference in 1973, I think. The patent is 1972. Okay, you can't publish something and then patent it; you have to apply for the patent and then publish it.

But Scott Duvall, Bill Ozarski [Owczarski], and Dia Polonis — um, Duvall was a metallurgist just as Owczarski was a welding engineer out of RPI. TLP bonding: a new method for joining heat-resistant alloys. And they're talking about nickel-based alloys, and in particular they're using nickel boron, which I should — there are a lot of brazing alloys based on various nickel compounds or nickel alloys. This is nickel, this is boron, and this is nickel and boron really lowers the melting point from 1455 C to 1035 C or something like that. You form a eutectic with boron. You also form a eutectic with carbon, with phosphorus, and there's a whole company in Detroit called Wall Colmonoy [Wall Colmonoy] which sells nickel-based brazing alloys.

And what they found is that if you make a braze using a nickel boron-based braze — boron is nice because it's an interstitial diffuser, it diffuses quickly, it may have three times the diffusion constant of a substitutional atom diffuser. So they like nickel boron. And this is not the original obviously, but you can start out with the joint, you can still see — but once you finish making this there's the joint there and you don't see, even at high magnification, anything. If you went through and did a chemistry scan across here you actually can sort of see different grain size maybe. If you did a chemistry scan you would find a concentration of boron — or higher value of boron here than here — because it's diffusing away.

And in fact in this paper they go through and describe the process known today as isothermal solidification. There is a thermal solidification, which is what you're familiar with — you heat something up, you pour it in a mold, and you let the temperature cool down and it solidifies. If you're doing an alloy, that's known as coming down the phase diagram, okay. You drop the temperature and it solidifies, you go from liquid down to solid. Isothermal solidification, you start out here in the liquid and you go across the phase diagram at one temperature. So big deal, you're just going in a different direction to get to the solid phase, right?

So you start out with a liquid in between two solids. This would be a nickel boron layer potentially between two nickel superalloys. It has a high concentration of boron — this is just schematic — a high concentration of boron, you heat it up, you'll melt some of the parent metal and so this joint actually gets thicker, which is sort of going the wrong way. But as it gets thicker, you let it diffuse, and eventually it starts becoming thinner, and eventually you'll have a little peak, and if you wait long enough you'll end up with a joint, the boron's diffused completely away, and you end up with a near perfect joint.

So they actually started using this. It's used widely in the superalloy business because nickel boron is such a wonderful system. And back 20 years ago I used to say it's a wonderful process, the problem is it only works in a few limited systems. And then I started doing a little more work and I realized it's a lot more common than we thought.

For example, the Etruscans — I think I told you this before — this is a gold earring made by the Etruscans around 600 BC, and they used this to put these little gold beads on what is a copper or brass substrate. You can kind of see the color, oops, if I get this thing right, the exposure right, the color of the gold beads — this exposure thing on this — the beads are sort of gold. There, you can see sort of a different color. Here's a bad joint, see the defect — that obviously they threw this earring out. There. Anyway, people have actually sectioned some of this Etruscan jewelry and shown that actually they used tin with copper and gold to make a transient liquid phase diffusion bond. So in fact you could argue that the Pratt Whitney patent was invalidated, could have been invalidated, but the Etruscans didn't file an infringement.

Now, it turns out about 10 or 15 years ago, Pratt and Whitney decided they wanted to destroy Chromalloy Corporation, which was the largest remanufacturer of jet engine parts. This was because peace had broken out — it was probably 15 years ago now — peace had broken out because of the fall of the former Soviet Union, and so they weren't selling as many engines anymore. And so they wanted to do something to boost their business, and there was this five or 10 billion dollar repair industry out there to remanufacture jet engine parts. And Chromalloy was the largest remanufacturer of jet engine parts in the world, essentially remanufacturing Pratt Whitney parts, General Electric parts, and Rolls-Royce parts.

And I remember going down to the Pratt Whitney plant in the mid-90s in North Haven, Connecticut, and the general manager of the plant taking us to dinner the night before, and he said, we've been giving away this five billion dollar industry to all these other people and we're not going to let that happen in the future. So they turned around and sued Chromalloy for a bunch of patents, and one of them was this paper and the patent on TLP bonding, except I found a 1956 paper in the Welding Journal, just a little paragraph, showing that Rohr Corporation had described the exact same thing. Every element of that patent disclosure was in that paragraph in the Welding Journal in 1956. So that patent got thrown out, but they still had a lawsuit about all this other stuff. In the end Chromalloy won. The judge awarded them no damages, and it cost them 30 million dollars in legal fees. So you can say did you win or did you lose, okay, but anyway, they're still in business.

In any case, now you can kind of see the gold color a little bit more. So people have sectioned these and essentially they had tin diffusing into copper and gold. Copper and gold will readily diffuse tin. Now, the Etruscans — we know the Etruscans essentially did it, because you can go in there and you can take a microprobe and you can find the tin gradient across there.

Student: [inaudible — apparently asks whether this would be called soldering]

This would be soldering. You could call this diffusion soldering, okay, because tin is within the melting point. We're going to get to soldering, but that's soldering. Well, I'm going to talk about what's the difference between soldering and brazing maybe today, but if not today than tomorrow, okay.

Here is King Tut's dagger, which was 2500 BC, and it's got these little beads on it. This gold sheath is what everybody thinks is so wonderful, but nonetheless they had little gold beads attached to brass or gold or something, and they think that the Egyptians used TLP to join the beads. But no one's willing to cut up the sword to find out, okay. You got to cut it up, okay. No interest in science, okay.

So there's other stories as I looked into this. Um, one story: Varian, a company up here in Beverly, called me in once as a consultant, and it was actually a fairly interesting — the dynamics of this are interesting. So they say, can you come up, we got a problem, we got a brazing problem. And uh, I go up, this meeting starts about one o'clock in the afternoon, I go in this room, there's about 10 engineers and managers and stuff there — and this is not actually all that uncommon — and they start telling me about their problem, and it's the crossed-field amplifier.

Okay, a crossed-field amplifier for the phased array radar system. This in this particular case it was the Aegis missile system, but essentially it's the same type of radar system that allows jets to fly very low over the mountains because they actually have a phased array that kind of tells them, oh, there's a mountain coming up, you better have something automatically to go over it rather than try to go through it. So it's basically a transistor for microwaves, okay. At microwave frequencies regular old silicon doesn't work, but you can have fingers. They basically take a piece of solid copper and machine a bunch of radial fingers — so these are actually slots, they've done electrical discharge machining to making this — the whole thing is probably that big around, okay.

[Tom draws on the board.] And they have a piece of molybdenum right here, which they originally braze in here, and then they basically then they cut the slots, but they braze it in here. And the problem was they were getting these intermetallics between copper and molybdenum — oh, actually I'm sorry, it wasn't copper and molybdenum, it was gold and moly. Gold — they were using a gold nickel brazing alloy. Here's the gold nickel phase diagram. Um, diffusion bonding is sort of fun for a metallurgist just because you actually get a look at phase diagrams to understand what's going on.

So here's the gold nickel phase diagram. You've got nickel at 1455, you get gold at — should be 1060, says 1064 — I think it's 1063 but yeah, it's 1064.43. Gold is one of the standards for the international temperature, so probably a defined temperature. But in any case you have a eutectic — or actually, yeah, it is eutectic — at 955 C at 18 nickel, 82 gold. That is a brazing alloy and that's what they were using to braze this crossed-field amplifier.

Well, the problem was that you would form a gold nickel intermetallic — no, a molybdenum nickel intermetallic. So you got 18 percent nickel, and there's a nickel three moly — I didn't bring that phase diagram — but there's a very high temperature, very stable nickel three molybdenum. And every now and then one of these little tips would break off, just like that little bead broke off the nice Etruscan earring.

But this screws up the whole electric field. What you're doing is you're modulating the electric field on this thing, and you're putting a pulse of high power electron beam right down through the center, and the modulating field modulates the high power pulsed electron beam and gives you a big magnification in microwave energy. Okay, so this is what you're using for your pulsed microwaves to do your phased array radar system. This is all 20 years ago, so maybe they've done some other things. And they said, if we lose one of these tips — and there must have been 50 of them going around this thing — you lose one because of this brittle intermetallic, now your electron beam gets distorted, it comes over, it destroys the whole thing.

First I thought, well, why are you using molybdenum on the tip? Other than molybdenum melts at a very high temperature — when the electron beam goes through there it's very high heat intensity, if you put copper on there you actually would melt the copper, but you have copper behind it because copper has good thermal conductivity. Turns out molybdenum has a thermal diffusivity which is 60 percent that of pure copper. So molybdenum would take the high heat flux for a fraction of the second while you have the pulse, and then the copper would suck the heat out over the dead time, okay, in between. So they had a system that worked, but there's a reliability problem.

And so, they're explaining all this to me, it takes about 40 minutes to kind of go through everything, and after the introductions and stuff, and I started looking at — I said, well, you know, I talked about some different ways you might solve some of the problems and redesign and stuff, and I said, you know what you'd really like to have is a TLP joint. They said, well what's that? And so I explained what a TLP joint is. If you could come up with a system with copper and molybdenum that wouldn't give you the brittle intermetallic but would give you a diffusion bond, you wouldn't have to have the high pressures for fixturing, you could do it faster, you know — in TLP joint you only need five psi, not 5000, for fixturing.

So you save a lot on your big pressure and stuff. You know, these big turbine components — in an aerospace plant you go down to Pratt Whitney where they're doing diffusion bonding, not TLP bonding, but diffusion bonding — they'll have a big ring of molybdenum, and that's how they fixture it in the furnace. These are all circular parts and they put this big donut of molybdenum around the part, and it doesn't expand as much because it's molybdenum. And so you can slip it on at room temperature, and you get up to temperature, and all of a sudden this big ring of molybdenum is holding everything together in the vacuum furnace, okay.

But beyond that, if you got flat parts, gee, you gotta have a huge press. And the world's largest diffusion bonding press is one in Japan. It's in one of the homework problems, I don't remember from 25 years ago, but the data is in the thing. But it's a plate that's probably as long as this room, it'll diffusion bond two plates together probably as long as this room, and eight feet wide or something. And it's got — I don't remember — 10,000 tons in the press. It's a pretty good sized press, and a big vacuum furnace. Instead of explosive bonding to make a clad metal, they basically — the Japanese were doing diffusion bonding, okay.

So anyway, fixturing's easier, I told them. And you actually can make the joints in an hour rather than 10 hours, typically. And they say, well, that sounds pretty good. You think you could do something to find out about that? I said yeah, we could look into it. And they finally told me, well, they'd like me to take some of these back to the lab and do some metallography on them and see what we had at the interface. And we kind of knew that maybe we were getting this nickel 3 molybdenum intermetallic. And they said, oh, by the way — this is at the end of almost two hours — they said, we've tried something else, that instead of 82 nickel — or 82 gold and 18 nickel — as the braze alloy, we've tried an alloy that's 37 gold, 3 nickel, and 60 copper. Turns out this is a standard gold braze alloy. And they said, we tried this and we can bend these things 180 degrees without their breaking.

I said, you can? And so they gave me one of these too. And it turns out lower nickel meant you didn't form the intermetallic, but the Navy didn't want to buy it because they thought it was just a silly cost savings, okay. They said, you're just trying to make it cheaper, okay. We want the 82 gold, we don't want 37 gold, we want 82 gold.

Well, it turns out we did the metallography. My graduate student found out that if — they had stumbled on a transient liquid phase diffusion bond, okay. What it was is, the gold was diffusing into the copper. Remember this is copper base material. The gold wasn't diffusing into the molybdenum, but it was diffusing into the copper as a one-way diffusion. But they were getting a transient liquid phase diffusion bond. And I'm sure we probably threw out the samples, but I wish I still had them, because you could bend those 180 degrees and you just see all the copper stretching and everything. It was essentially a perfect diffusion bond between two dissimilar materials.

And what I did is I wrote a letter to the Navy and explained this was not just a cost savings, this was transient liquid phase diffusion bonding. And by giving it a title, okay, the Navy bought off on it. Wasn't a cost savings, it actually was an improvement because it had a name, okay. I'm not kidding when I tell you these stories, this is real, okay.

Another story of TLP bonding, which is actually TLP soldering. I get a call from Raytheon. And this was in a plant right out here in Waltham, which is now — the one side of the road is now a BJ's Warehouse, the other side of the road is the building with the World Club Gym and SGH, okay. But there used to be a Raytheon facility where they made defense components, and they've since torn down both those buildings. They actually had a crosswalk on the second floor across the road in front. So you come in on the east side building, you go upstairs, you walk over, and you come into the manufacturing facility where the SGH building is now.

And so I get this call from them. Oh, actually I'm mixing up some of my Raytheon stories, but it actually was at the same facility, so that's why I'm getting mixed up. Anyway, I get this call. They were making transformers, in this case, for the U.S. Navy nuclear subs, for the engine room. This was the transformer that controlled the control rods for the nuclear reactor. So a million dollar transformer, it was sort of critical application, you didn't want it to fail. And in the transformer you got all these copper windings, right, and then outside of that for electrical shielding they had a copper sheet that had to be grounded to get rid of any electrical interference from anything else that was around.

And so they were supposed to solder the ground wire to this copper sheet inside this transformer with a 95 lead, five percent — I'm not copper — 5 tin. 10 [tin]. Okay, so lead tin, 63/37 is the eutectic for lead tin. This is almost pure lead. So the eutectic for lead tin is 183 C. Lead melts at, what, 420 or something like that, 300 — I can't remember, 300 something C. This is what they were supposed to make.

And this whole transformer, after it was put together and they'd pot it up, they'd put it into an oven at 220, or something, maybe 212 degrees Fahrenheit — um, I don't know, 200 degrees centigrade I guess, anyway, because it must have been centigrade anyway. They put it in this oven for — it had to be there for 24 hours. And the whole thing was potted in this plastic compound, and that's what part of this heat treatment was. But also you kind of like to give a thermal excursion to all your stuff before you certify it for service. That's common in the electronics business.

Well, when they did this, out of one of these little feed-throughs they have some liquid metal come out — like, that's not supposed to happen. And they had four of these they put in the furnace, so four million dollars worth of transformers sitting there. So they analyze it, and they find that it's — um, I guess it's, oh, the eutectic I guess is 67 tin, no, 63 tin and 37 lead, I think that's right anyway, I should have looked it up, it doesn't really matter for this — but this is the eutectic, 183, and this is something north of 300 degrees centigrade.

So they were supposed to have something never should have melted at 200 degrees centigrade in their furnace, but in fact when they analyzed this liquid that was coming out through their feed-through, it was lead tin eutectic, which is the most common solder alloy for electronics at the time. This is 20 years ago, 25 years ago. And obviously someone had used the wrong solder when they made this joint, and now it's potted up, there's no way to unpot it. They have four million dollars worth of scrap, okay. And they weren't happy.

And so they wanted to know if there was anything they could do. And in fact it turns out, just to try it out, they had taken one of these solder joints and soldered it with the wrong alloy and stuck it in the furnace — this wasn't another transformer, it's just a copper sheet to a copper wire, soldered with this — stuck it in the furnace at 200 degrees C, and then when they took it out they did a pencil test on it as a function of temperature, they just hung weights on it, and they found it didn't fail until 300 degrees centigrade. And they said, we don't understand what's going on. I said, well, the transient liquid phase diffusion solder joint, they diffused the tin into the copper, leaving behind a very rich lead, which is what they had intended to get anyway.

And so, oh, that's great. And again, because I could provide that paper, I go to my notes and I pull out Ozarski's [Owczarski's] paper, and I write a one-page letter explaining that what had happened is they had formed a transient liquid phase diffusion solder joint, and they now had a better joint than if they had just soldered it, okay. And the Navy bought it and bought the four million dollars worth of transformers. And I think I made $700 on the deal, but anyway.

Here's a book on soldering that actually goes through and shows base material, base material, a preform layer that may have some interlayer material, and you end up with a joint after it homogenizes. It's not like your traditional diffusion bond where you don't have anything in between. You still have your solder in between, but in fact it can be a higher melting solder. And here's a picture of an actual joint that has not been completely — copper using a tin interlayer, okay — but it hasn't, it's still got some other junk in the center, they maybe haven't done it long enough. Here's from the same book, some of the types of interfaces that you can get. You'll still find something at the interface.

So I started thinking about this. It was around 1990, and they basically when they would build integrated circuits and things like that, circuit boards, they would put them in these things at 200 degrees Fahrenheit typically, and they would hold them there for 24 hours. I can't remember what they called it, aging or something, okay. It was basically to try to stress all the joints so that the circuit board still worked after it came out of this thermal cycling test. It wasn't really cycling, it was just putting in a furnace at a couple hundred degrees for 24 hours, and if it still works it's probably good, okay. It's just a stress test on each circuit board.

And I started thinking about it, and realized that a lot of the joints they were making — they might have been making solder joints, but it's the same old lead tin solders with copper or gold, and in fact what they're really doing is transient liquid phase diffusion soldering. So I had a student who — it's time to finish — I had a student from Raytheon in the early 90s, and he came in and we actually worked on transient liquid phase diffusion bonding, or diffusion soldering, to try to join silicon, or in his case gallium arsenide substrates, gallium arsenide chips to the substrate. Big thermal expansion problems. We'll talk about that stuff tomorrow.

Just to tell you, next week Dr. Belmar [Belmar?] is going to go off to Africa to shoot wild game. Yes, actually, um, with his brother. And so I will be lecturing on Monday, Tuesday, Thursday and Friday, and that should finish me up, okay. And then he will be lecturing the week after that. But anyway, we probably won't have class on Wednesday unless something really strange happens, but I will.