SSW_S2013_09

Uh, any questions from before? We were talking about diffusion bonding and whatnot. Um, I want to finish up a couple of things. I mentioned that diffusion bonding would allow you to uh join dissimilar materials, and in fact that's true, but when we join dissimilar materials one of the first things we always have to worry about is thermal expansion. And it turns out this thermal expansion mismatch between the materials — it turns out sometimes the materials are used at elevated temperatures and there's a difference between room temperature and elevated temperatures, and then you have thermal cycling, so you get thermal fatigue because you have cyclical stresses from thermal cycling. Or you form the joint at an elevated temperature and cool it down. Even if it's formed at room temperature you have residual stresses because of the thermal expansion.

Does anyone have an idea of how much temperature differential it takes to create yield-level residual stresses? Have you ever — probably never thought about that question, have you? Well, it turns out a typical value, let's say for steel, is 10⁻⁵ per degree C, okay, that's inches per inch, centimeters per cm, you know. That's the thermal expansion of steel. Now it turns out most ceramics, and that's not unusual for nickel or other things — aluminum's maybe 50% or 100% higher than that, could be double that, I don't remember, but it's on the order of 10⁻⁵, okay. So if we take steel and we say okay, what's the yield strain of steel approximately? What's the yield point? 5%? It's point — 2%, 0.002 strain. If you draw a stress-strain curve, this is the yield point, and they usually define the yield point at 0.02, 2% offset, or a strain of 002, okay, for a metal, for your stress-strain curve. Stress-strain, okay, they match the slopes, right? Yeah, they match the slope, so you could argue that it's 0.003 or something, right? I mean, you have to look at the modulus and stuff, but let's just take our 0.002.

Well that means that um 200° centigrade should — so if it's alpha delta T to give you this much strain, 0.002 strain, if 0.002 strain is the strain at yield, 200° centigrade. And if you want to say okay, let's take the fact that this is a secant, you take the secant modulus and everything, .003, so I usually say 200 to 300° temperature differential is going to cause yield-level residual stresses approximately. And most people don't realize — even in soldering, except in soldering you have this material that's going to creep, and so you don't get a lot of residual stresses because even at room temperature the solder creeps. But if you're doing brazing or diffusion bonding and you're up at 1,000° centigrade for your bonding temperature, this thing's going to go through yielding on its cooling for a dissimilar metal joint, and it's just the way it is. Most people haven't stopped to think how little the delta T has to be.

Now I think about it all the time, because I spent my career talking about welding, and welding is nothing more than dealing with big thermal gradients all the time, okay. I make an arc weld, I got thermal gradients of a 1,000, 2,000° centigrade per centimeter. Sometimes I got 1,000° per millimeter if I'm doing laser or electron beam — shoot, that means in, you know, 10,000ths of an inch I can have yield levels, I can have temperature gradients that would cause yield-level residual stresses, okay.

Now when we do dissimilar material bonding — and here's a turbine blade which they put a coating, a ceramic coating, on here — the problem in general with metals and ceramics is they vary by about an order of magnitude in the thermal expansion coefficient. A ceramic has something on the order of maybe 2 × 10⁻⁶, five times less, okay. So maybe it's 400 degrees centigrade or 600 degrees centigrade, you still have some pretty serious problems. And if I take this — shoot, I got temperature gradients on this thing. If you do the modeling, you've got 500° between here and here. This is on the — this is attached to the stator, and this big massive thing, and the hot gases are hitting right here in the center of this. You look at the thermal model, you got a big hot spot in the middle and then you got cool spots down here. How do you keep the ceramic from just forming a dried riverbed pattern and flaking off?

Well, you do it several ways. First of all, the ceramic they always use is zirconia. And why zirconia? You ought to know this answer. Well, it's got a phase transformation, but the coefficient of thermal expansion of zirconia matches that of metals very closely. It's not a factor of five or ten off, okay, it's almost the same, if you — particularly elevated temperatures. So zirconia has excellent thermal expansion matching. I mean, they'd love to use aluminum oxide, but the only way you can use aluminum oxide is in very thin layers, okay. And in fact they do — that's what the natural oxide coating is, either chromium oxide or aluminum oxide on most of the super alloys. But that's an extremely thin layer, that's 100 nanometers thick or something, okay. But thicker coatings like this — but they also do one other thing, they plasma spray on a bond coat first, which is a nickel-base alloy, okay. So a nickel-base alloy on nickel-base alloy. But in fact the bond coat is porous.

And remember, I think I told you the other day, that I — when I was showing you the Farberware pot, and I realized they didn't really want 100% bonding in that cold bonding of the aluminum base to the stainless steel pan, if it was 100% the difference between aluminum and steel when you go up 6 or 700° would cause warping of that pan. But in fact if you actually get only a 10% bondage joint, and let's face it, what's the stress on that aluminum at the bottom of a cooking pot? Okay, there's not a lot of stress compared to the area of bonding, right? You actually want porosity at that interface. They actually want porosity in the bond coat of these things to a certain extent, because what has a better ability to absorb strain than empty space? Voids are wonderful.

And I mentioned that I actually had a student do a doctoral thesis 10 years ago, and she showed through modeling and some experiments that if you looked at the amount of porosity at an interface in a ceramic-metal joint, you actually would get something like this, where 10% porosity gave you the strongest joint. So this is porosity and this is strength. Actually, you had — the residual stresses would kill you here because of the thermal mismatch. If you had some porosity, sure you're losing bonded area and you got stress concentrators and everything else, but you actually first increase, and somewhere around 10% you start to decrease, 'cause now the amount of bonded area is so small, or you're losing bonded area, but initially you're reducing residual stresses, and then you got this competing effect of losing bonded area, right, with porosity. So some porosity actually can be good in a dissimilar material joint in terms of combating residual stresses.

But another principle of dissimilar metal joints — that it's the reason I put this up, the only reason I keep this blue thing is because it's the only way I can find my pointer when I throw it down everywhere. I have this little green laser pointer but it doesn't have a flag on it, okay. Um, so I use my blue laser pointer, and by the way this is not just your simple Blu-ray laser pointer, those are actually sort of purple. This one actually is blue. This is like 30 nanometers difference or something in its wavelength from — which means the price is about, uh, instead of about a $5 laser it's about a $400 laser, but nonetheless, hey, you know, nothing's too good for you guys.

Okay, so if I look at plane stress, just a flat slab, and look at the strain energy, you can write down a formula which is — there's the difference in modulus of the two materials, so this is the elastic effect, delta alpha squared, delta T squared, width squared, error function, it's red, it's being relieved as you go further away from the interface. Um, but in any case the strain energy is just the — you can think of as a spring, dx, and you got ½ kx² — that's the strain energy. The bond energy just goes linearly with the bond area, so it's proportional to the length or the width of this whole thing. Well, if you look at this, you notice the strain energy increases as the square of the length of the bonded area, and the bond energy or strength in force, it goes as the distance to the first power. And if you do it axisymmetric, you find that the strain energy goes as the third power of the radius, but the bond area goes as the square of the radius in a circle, duh, okay.

The point is, strain energy always goes one dimension higher than the bonding strength, which means you can always bond small things but you can't bond big areas, because strain energy will kill you, okay. Just think it is — you're just stretching the spring bigger and bigger, stretching the spring bigger and bigger, and there's more and more energy there, and eventually it's going to fall apart. And in fact the student showed that strain energy was, uh, one of the best ways to look at the problem of dissimilar metal bonding.

You have a problem if you want to do stress or strain. Stress and strain are not scalar properties, they're not vector properties — what are they? They're tensors. And the strain tensor, um, well, there's nine values of strain and nine values of stress, and the tensor that, you know, matches the two has 81 coefficients, right? Now you can prove that only — can't remember, anybody remembers? — 31 or something or independent, okay, because there's some symmetry to this, and then you can prove by, if you assume that there's no volume change, you can get it down a little bit, uh, by another one or two, but you still end up with all these things. What strain do you use? Well, people have been worrying about this for years, and not just in dissimilar metal joints, but just in a simple tensile test where you have b — or not simple tensile test, but a biaxial tensile test, what do you use for the criteria for failure?

People have been using von Mises, which is a strain energy criteria, or a von Mises not — it can be a von Mises strain, which is a strain energy, or it can be a von Mises stress, which is basically a square of your stress, sort of an algebraic square, uh, of your stress. Um, or do you use — there was originally the Tresca condition, which we don't have to go through, but that was basically just say, oh, you say where's the maximum principal stress, okay. Well, you'd like to have a single number in the real world. I mean, you talked about how does it work in the real world. So as an undergraduate you'll learn about the 81 strain — stress-strain tensor coefficients, and then you learn that you can reduce that to some smaller number, but it's still a huge number. And what number would you use? Well, when you get to how it's actually done in practice, what we found is there is a single scalar number, it's basically the strain energy of the joint, and that goes one dimension larger.

And so it turns out that I won't say that, um, computer chips haven't been made larger than about a centimeter on a side, 'cause they had, they have, um. But in general actually integration sizes are such that you can put — what, 10 million transistors on a single chip a centimeter on a side? 10, 12 years ago they couldn't get that kind of packing density, and when they wanted, I think it was the Pentium 6 or whatever they called that, they actually had two chips, because they couldn't go larger than about a centimeter. If they tried to go to a centimeter and a half on a side, they couldn't bond it to its substrate without it cracking, just from — there's about a 30° centigrade temperature differential that you have to carry the heat out of the chip. I mean, if you don't have a temperature gradient, Fourier doesn't help you get the heat out, right? In fact Fourier says the heat won't flow out if you don't have a temperature gradient. So with a 30° — if you just put in a 30 or 40° temperature gradient on the dimensions that you have here, you really can't bond chips more than about 1 cm on a side, and that's just dissimilar metal bonding.

That's not to say that back in the mid-90s the Defense Department didn't make chips that were 2 inches or an inch on a side, or 2 inches on a side, but this was all super-classified military stuff, and where they had to have the interconnect distance had to be small enough to get the speeds they wanted. But now as they've downsized the chips, that technology has sort of gone away, they don't need to do that. But these were some of the things that Motorola's defense division was doing for uh the Department of Defense, were uh, um, they were really pushing well beyond the limits in terms of the dimensions that you can bond.

Now think about — a couple of you are mechanical engineers or civil structural engineers — what size brazing tool bit do you typically see? If you braze a carbide bit to a piece of steel, it's about a centimeter on a side, isn't it, okay. You don't see two inch layers of carbide bonded to by a brazing joint to a piece of steel, or diffusion bonded or anything else. This is just dissimilar metal bonding, and it's not necessarily diffusion bonding I'm talking about right now, and it's because of the strain energy, okay. Same thing that I just had up here on the board. So that's not to say there aren't ways around it. You can add porosity to the interface, you can pattern in the interface, you can actually separate this thing so you essentially put a tile on there with spaces in between. It's like looking at the Stata — Status Center [Stata Center] first floor lobby, you know, they got — they got concrete on the floor and they got these little gaps in between, so that allows for the thermal expansion, okay.

So uh, I did want to talk — so I've covered that. Now there is another type of diffusion bonding that is not used very often, um, but let's say I've got some surface that has — why? — well, it has poor surface qualities for diffusion bonding, uh, there's something that we call sometimes called activated diffusion bonding. Um, and the example of this comes out of what used to be known as Rocky Flats, where they used to do things to make nuclear warheads, and they had to do things that were sort of offbeat and they could afford to do things no one else could afford to do. But there's a research paper by O'Brien, Rice and Olson, and it's in your notes somewhere, uh, not necessarily the paper but something talking about this.

They had very high strength maraging steel, so 200 ksi steel, and they wanted to bond it. But the problem is you can't form a diffusion bond at low temperatures. If you go up to very high in temperature, these maraging steels lose their properties. In fact that's one of the problems of diffusion bonding — I said you get a near-perfect joint but you can't always get the microstructure you want, okay. I go and I heat treat a steel and I make it martensitic, and then if I go up to 1,200° to diffusion bond it, it is no longer martensitic and it lost its strength, okay. So you can't always keep your properties. And then there are other things in here, rather than just iron, but you have oxides on the surface that won't let you diffusion bond at low temperature.

So what O'Brien, Rice and Olson did is they basically deposited a layer of silver on both of these. And so now they call it activated diffusion bonding, but now you're not bonding iron to iron or high strength iron to high strength iron, you're bonding silver to silver. Silver is a relatively soft material, and silver doesn't form oxides above 400° F. If you look at the thermodynamics of silver oxide, it decomposes at about 400° F. So you can heat up silver surfaces in a vacuum and drive off by vaporization the surface contamination of the oxide and also a lot of the oils and other stuff, and now you have a very clean surface. And it's got its peaks and valleys, but silver is so soft, it can be really pure silver, is 10 ksi yield strength. So now the amount of force I need to deform those asperities and essentially do a diffusion bond at 4, 500, 600° temperatures below the temperatures at which I would destroy the properties of my high strength base materials.

Essentially you do this, and if you have a thin enough joint — and this could be on the order of a thousandth of an inch thick — you can actually get a joint that has a strength in tension of 100 ksi. Well, how do you get that? Well, it's called contact strengthening, and I'll talk about that later. But very thin joints, the material enters a state of triaxiality when you pull on it, okay, and it won't yield. Now it may not be very good in fatigue and it's terrible in shear, okay, because you got this soft interlayer in between here. But in pure tension, when you pull this in tension, you set up a triaxial stress state in that very thin joint, which can take something that has a strength of 10 or 20 ksi and it can appear to have a strength of 100 ksi, okay. So if you have the proper stress state and you don't want to get high temperatures, there's an example of a technique where you can essentially change the surface, okay.

Now that's one example I can give you, but another is not necessarily as esoteric as this. But let's say I've got a material that's hard to diffusion bond, take aluminum with its aluminum oxide. I can put tin on the surface, and tin will diffuse into aluminum and hopefully tin might break up some of that aluminum oxide. Zinc sort of works — what can work on some of these things is not necessarily on aluminum, but titanium works on a lot of your higher temperature alloys. You change your surface to a titanium surface, remember titanium is the perfect material for diffusion bonding at above 900° centigrade. So just change your surface, plate something on the surface that's easier to diffusion bond, whether it's silver or titanium or whatever, and you often can get around a problem, um, so far as that goes.

Student: [inaudible — question about thickness of plated layer]

They could be a half a thousandth, you're electroplating them on, so they can be very thin. They have to be very thin, or you actually have to use enough force that you actually flatten them significantly. But you'll see — actually, it's in some of the notes — I've got some graphs later. It's actually in the brazing manual, and so when I get to brazing is when I usually talk about contact strengthening. But in fact braze joints can have a fivefold increase in tensile strength, uh, if they're 1,000ths thick as opposed to 5,000ths thick, okay. And it all has to do with this triaxial state of stress you can set up in very thin joints.

And basically, when I say triaxial state of stress, just to give you a little bit, uh, 'cause I didn't prepare to go through the whole thing uh today, but this is an unyielding surface. The iron is unyielding. This stuff wants to suck in, if you want to think of Poisson's ratio or whatever, it wants — as you're pulling on it, it wants to pull in. It can't pull in because it's connected to an unyielding surface, and so that means it has an effective stress pulling it out this way. And you set — you basically go with tension in this direction and biaxial tension in the plane, and that's a state of triaxial tension. Does not lead to shear, okay. If you get into Mohr's circles or uh von Mises cylinders and things like that, you can prove to yourself that state of triaxial tension or triaxial compression does not contribute to deformation, okay. The material will appear to be much stronger, um, because it's not going to shear. Metals deform by dislocation motion, and you don't move dislocations with triaxial stresses, okay. So you can play little games like that, okay. That's all I wanted to say about that stuff.

I want to now go into soldering and brazing. And Nikolai, you brought up soldering and brazing — what is it? Well, this is standard terms and definitions of the American Welding Society, mostly written by a guy H. Cary [Cary], I think I mentioned H. once before. And brazing down here — I'm not going to blow it up — a group of joining processes that produces coalescence of materials, blah, in presence of a filler metal having a liquidus above 450° centigrade or 800° F. If you look up soldering, you will see the exact same definition except is below 450° F. Okay, where's soldering — uh, soldering, okay here's soldering, not exceeding 450° F, okay, or C, grade.

So turns out soldering and brazing are the exact same process. Ones — one occurs at high temperatures, one occurs at low temperatures. How did we get to this state of two different sets of terms? Well, we got there because soldering came first. In ancient — I mean, soldering has been done for thousands of years. I showed you these little solder beads on the Etruscan earrings and King Tut's dagger, we think, where they basically were soldering with tin. And tin has a melting point below 450° C, it's 212° C, uh, so far as that goes, so it's soldering. But they actually ended up forming a transient liquid phase diffusion solder joint. Um, soldering has been around for years, brazing came a little bit later and people gave it a different name.

And there are other problems with some of these things, but um, this comes out of an old soldering manual and shows the complexity of a solder joint. You got, in this case a soldering iron which has this intermetallic compound of iron tin, if you're talking about lead tin solder, you have molten solder, you have solid solder, you have a flux, you have a contaminated surface which is little different down here, and you have your uh base material that you want to solder to, and you have temperature gradients all through here. So turns out soldering is a fairly complex process, and it's controlled by the inherent wettability of the flux, the solder, the base material, and the liquid — the solid solder and the liquid solder. And that, as many of you know, is governed by Young's equation.

Okay, Young's equation, you probably studied in some sophomore thermodynamics course or something. But basically you have a vapor, liquid, and a solid, and a little drop of liquid on top of the solid, and you want to know if the liquid will wet the solid. And there's some contact angle of the liquid, so you can think of raindrops on a fresh-waxed car hood. And you actually can draw force vectors coming off this contact point right here, and if you do that — someday I'll make a better overhead but anyway — you have some surface tension between the liquid and the vapor, the solid and the liquid, and the solid and the vapor, and the magnitude of that has different um magnitudes. And you can resolve the forces in the horizontal direction and end up with Young's equation, that the solid liquid surface tension is equal to this — the solid vapor surface tension is equal to solid liquid plus the horizontal component of the liquid vapor, which is just the cosine of this thing, okay. And there's your contact angle is right in here, that's Young's equation.

Now I'll tell you what I've learned about Young's equation over my career. The more you think about it — it's a very simple equation, but the more you think about it, the less you understand it. This — there's actually a couple things I've learned about — that one's a tensile test. The more you think about a tensile test, the more complex you realize it really is, okay. You got things shrinking this way, pulling that way, you got triaxial states of stress when it starts to neck, you got gradients in all kinds of directions. One of the things about Young's equation — it only applies to equilibrium, and you will see all the time in the literature people trying to apply Young's equation to non-equilibrium situations, okay.

Um, one time we were trying — one time 20 years ago I was actually providing funding for Professor Sonin over in mechanical engineering, who's now retired I think. But Professor Sonin is excellent mathematician, analy — you know, fluid flow analyst. And he was trying — we were trying to be able to do sort of 3D printed — pre-3D-printed — printing of circuit boards. So, you know, this would be great, you could make one-of-a-kind circuit boards by just shooting out little drops of lead tin solder, and people have been doing this with polymers, why couldn't we do it with the liquid metal? Well, it turns out tens and hundreds of millions of dollars have been spent trying to do this with liquid metals, and it just — it has not yet been perfected after 20 years.

But anyway, so Professor Sonin called me up and said, well Tom, I've been — I'm getting all confused, and I thought okay, here's one of the better theoreticians in mechanical engineering department, he's asking me a question. Um, anyway so he was thinking of a solid that had a liquid drop on it, and he said what's the contact angle? Well, I said you just use Young's equation. Well, no you don't. It turns out the contact angle here is not a function of the surface tensions, it's a function of the mass of this drop and the weight of the drop and everything else, okay. If it wets you can put larger amounts of liquid on top of the material here. But in addition, what he was really thinking about — you could think of it as if it was at the melting temperature and so it was isothermal, but in the real life of the real experiment it wasn't even at equilibrium. I'm putting a hot drop on top of a cold post when I'm trying to build these things up, okay, by putting little drops of solder on top of something else. But whether you assumed it was isothermal or not, you try to work out the math for that and you find it's an indeterminate problem, okay. By just changing the geometry — it works great in a flat plane, I can explain it to any sophomore in the Institute, okay, but you change the geometry a little bit and oops, doesn't work anymore, okay.

That's just an example, uh, that, and the fact that it only applies to equilibrium, which I could always figure out, um, shows kind of how Young's equation is a lot more subtle than, uh — or is a lot more — is, well, it's oversimplified. Young's equation is oversimplified.

So let's kind of look at different solder alloy systems and say, well why do we have this 450° break? The — how did nature — well, how did this come about? Was it just a historical thing that people started talking about soldering and then sometime a couple hundred years ago they started talking about brazing? Well, it turns out you can look at different families of solders. So here are gallium solders with tin, indium and tin, bismuth, indium tin lead, gold with silicon, germanium. And tin forms very deep eutectics, and gold tin is often used on some of these bonding of um semiconductors to their substrates, okay. Here's your lead tin — let's see, where's lead tin — uh, hey, this must be a new book that's environmentally sound, I don't know — where is lead tin? I don't have it here, but anyway, oh, tin lead probably — ah, here we go, 37, okay, sorry.

U, anyway, um, so there are temperatures that go up to 400° and by the way this 450° C and 840° F — uh, older versions of the American Welding Society it was 800° F, 'cause they didn't know about the centigrade scale back in the old days, was like 1990. Um, the Welding Society has learned to use the metric system. But then you can look at some of your brazing alloys, and here we got braze alloy systems: aluminum, silver, gold, copper, nickel, and palladium. And here's actually your silver solders, are actually silver brazes — we call them silver solders but they are brazes if we want to be precise.

Now there's an interesting thing here. This one starts at 500° centigrade, the other one stops at 400. What happened to 400 to 500° centigrade? What happened is there's nothing on the periodic table that melts in that range. Think of an element that melts in that range. You can even go through the uh — well I don't know if you can go through the transuranic — you can go through the lanthanide series and stuff, you can't find anything with a melting point in that range. Nature didn't give it to us. So it turns out there is a natural breaking point between soldering and brazing. It's the fact that there are a certain class of alloys that melt below 400 and there's a certain class that melt above 500, and that's — 450 is the happy hunting ground between 400 and 500, and that's where the International Institute of Welding defines the difference between soldering and brazing. So there is actually a scientific foundation for it, and that's because nature forgot to give us something that melts in that range.

Reason I've thought about it, and uh, a little bit, meaning I probably spent 60 seconds of my life thinking about it, I haven't thought of a good reason for it. I'm sure that some chemist who understands bonding kinetics and stuff might be able to come up with something, but it just — there's a gap there, and that's a pretty large gap. Out of a 100 elements roughly, you know, 90-some elements, whatever the number is, to find that you just miss that range — now sure, we missed the range between helium and hydrogen and whatever the next thing is in terms of melting points, right, or vaporization points, but among the metals, we just don't have anything, okay.

Student: Um, uh, I don't know if I need to show you that — is there any temperature dependence in there? Just because, for example, you know, when you're — when uh soldering something, if you just try to drop a cold, or you know, solder wire kind of —

That just — right. Um, there is a temperature dependence, but once we start talking temperature dependence, we're no longer at equilibrium, right, okay. So, and people have tried to take the Young's equation and develop a kinetic equation of wetting from it, okay, but it turns out I've never seen a model that really works or gives you new insight that you didn't have before. Um, I guess my insight to you is, um, don't spend too much time trying to explain what you're thinking about in terms of Young's equation, because you will often just go around in circles, okay. Um, and sometimes it's because you're not at equilibrium, you just impose the temperature gradient. The other problem is, you don't know what the cleanliness of that surface is. If I go back to my little model of what soldering is, well, look at what I got going on here at the interface. What is my interfacial tension that I should be talking about? Is it solder against the base material? Is it solid solder, liquid solder, or solid solder? Is it flux? Is it a contaminated surface?

When you're talking about a wetting phenomena spreading across the surface, I have never seen a really good model that can predict the solder flow or the braze flow across the surface in the presence of a flux and things like that. I've never even seen one other than Young's equation, which is very simple, that tends to explain just on a clean surface — get, take contamination out of it. Um, yeah, there's Young's equation, and Young's equation gives us sort of a warm fuzzy feeling, but that's about all it does, okay. And I'm not saying that Young's equation doesn't have some utility on a simple scale, but that's all it does, it's just on the most simplistic scale does Young's equation help us, okay.

In the real world of soldering though, it turns out that the kinetics of cleaning are such that it controls most of our soldering operations. And this comes out of an old soldering manual from the 60s, where — we don't have to blow it up so you can see a little bit better, we don't have to see everything. This is what they're looking at: here is 6 months old, 21 days old, 24 hours in steam, and a freshly clean copper surface, okay. And they're looking at different coatings, um. They could have had um a resin lacquer — and we'll talk about what that is — tin plated copper, silver plated copper, 6440 — this is the eutectic solder, so you pre — pre-tin it with the low melting solder. Um, this is oh, this was put on by roller, immersion, roller, electroplating, so this is the same thing, roller and electroplated, how they put it on. So they have different coating methods of the copper, um, you've got different thicknesses, uh.

So they tried a number of different experiments, and what you see here, this is the wetting time in seconds. So there's a couple of different wetting tests — I may talk about the different ones later, because wetting time becomes important particularly when you're trying to do a printed circuit board, and you want — you put the little solder paste on top of some circuit board, might look like this, and you've screen-printed on a solder paste. This one is pre-tinned, it's copper with electroplate of tin on the surface, and you're going to melt the solder to the surface. And this is old, it's lead tin solders. Nowadays the last 15 years people spent a lot of money trying to get the lead out of the solder for environmental reasons, okay, which has been a big problem because nothing wets as well or as fast as lead tin solder. No one has found, been able to develop anything, and hundreds of millions of dollars have been spent by industry and government.

But in any case, the wetting time — how long it takes on a freshly clean copper surface once you get to the melting point to allow the wetting to occur and the flow to start being pulled across by Young's equation, except it's not even Young's equation because it's not equilibrium, right? That's why I said how are we using Young's equation? We're using it for kind of the general idea, but we never use Young's equation in an equilibrium situation, okay. Anyway, it's less than a second, something on the order of a half second. And they actually were doing these things — I'll show you the dip solder test or the uh later, how they measure this. But um, if you let it 24 hours in steam, unplated, no coating, fresh copper, boy you go from half a second to more than 10 seconds to get the basic solder to take off that copper oxide tarnish that formed over 24 hours.

And so if you go to a really good electroplating shop, they go from the cleaning bath directly to the soldering, and they may have — they may have to uh put some components on in between, okay, you know, the capacitors and the chips and stuff. But typically there's not much more than an hour, and there are military specs that say you must go from cleaning to soldering, melting your solder, within an hour, and this is the reason why. And if you get other surfaces 20, 21 days long-term damp heat, you know, you get similar types of problems. Let's face it, if it doesn't start wetting within 10 seconds, it's not much good, okay. Um, 6 months at normal storage, it's sort of interesting, some things actually clean up over time, um, with — after 6 months.

Um, this handout is in your handouts. Um, well, we might as well go to the solder test now, so you can see how this is done. Uh, there's various solder tests, some of which are just very practical. One is um a globule test, and you basically have your mandrel, you have a wire, and you put a piece of solder on there, and you heat it up, and afterwards you see if it wet it. You stick it in a furnace and oh, did it wet it or not, well duh, okay. That's pretty simple test.

Now there is the automated dip test schematic, where you basically have your whole thing on a uh rotary dip. You take your circuit board, and you basically have a pendulum, and you swing it spring-loaded so it goes swinging across the surface of this liquid solder bath, and then you look at it and see did all the joints get made in the amount of time. And depending on what you have for damper, dampers or springs, you can change the amount of time on your solder bath that it actually is in contact with the solder.

The best thing — and for merely $50,000 you can have one of your own — um, you can buy a machine that measures a wire. So this is going to be a copper wire going into a solder bath as a function of time. And I might as well blow this up a little bit — oops, um, so you can see it. I blew it up too much. Okay, blow up way too much, okay. So I've got, I've got this little wire um that I'm going to drop into the solder bath, and when it first touches the solder bath, that's time T is equal to zero, and initially the wetting angle is such that it makes the wire feel like it weighs less. The wire is actually on a balance, so you're weighing the wire. And it's floating — surface tension is such that it's floating, and therefore the weight, or the force on the balance, decreases. When the wetting actually increases — so you got a 90° contact angle — it comes back through zero when it actually wets, and now it's actually trying to suck it in, surface tension is sucking it in, it goes up to a positive force. And it stays there, and if you now try to pull it — this is the buoyancy force — um, if you try to pull it out, it'll actually should show this with a little residual force because you have a little bit of solder stuck to the surface, okay.

Uh, you can do it in a system where everything is hot, or you can do it where it's cold. Um, typically the wire is cold, because this is a practical test. You're going to be putting your circuit board into the system cold and the bath is going to be hot, so the whole thing is in temperature gradients. But these $50,000 machines can do various things. But you're trying to simulate typically what's going to happen in what we call wave soldering, and wave soldering — just happen to have a picture of wave soldering.

Um, wave soldering is the way most circuit boards are joined today, or soldered today. You basically have a circuit board with the components on it, uh — this is sort of old uh thing. They either put a foam flux on the surface, or they spray a flux, a liquid flux, on the surface, or it could be a gaseous flux in the system, to do the cleaning. These are your components on the other side of the mounted components, but actually they have many components that will survive the plating bath. And you have a bath, and the bath might be — well, some of these things are three or four times as long as from here to the window, and the actual bath of the solder might be two of these tables long, okay. And you got 5 or 10 tons of solder being pumped, liquid solder being pumped at a couple hundred degrees centigrade, pumped up, and it goes over a little wear, and so you've got a waterfall or a solder fall as it may be, going over this. And you have a high spot over the wear, and you bring this in. Now typically bring it in a slight angle, okay, but this shows everything flat. But anyway, you end up scouring off the oxides and the contaminants and making your wave soldering — they call it wave soldering. And just pass — your solder becomes the heat source to heat up the copper wires and everything else, could be preheated, okay.

But in a good shop, this occurs within 1 hour of cleaning the copper for circuit board and then maybe plating it and everything else, okay. Now back 24 years ago, um, in the manufacturing program at MIT I had three of the four students I was supervising that year working in a partnership between um, uh, well, was a partnership between Motorola and Intel [Digital] to build the VAX 9000 computer. This was Digital — no wasn't Intel, it was Digital Equipment, and Motorola I think, or maybe it Rogers Corporation, anyway whoever it was, two firms down in Arizona. And they wanted to make — they wanted to bond Digital Equipment's latest and greatest type of chip, like this little Intel chip, and it was on a TAB bonding tape, okay. And we haven't really talked about this, we'll probably talk about it Monday.

Um, by the way, you know, I'm around Monday, Tuesday, Thursday, Friday next week, no class on Wednesday, and Dr. Belmar's gone next week, so four classes next week of me. Um, in any case, it was a tape — TAB, tape automated bonding, a tape like this that had a computer chip on it. And Digital Equipment was making the tapes, and Rogers Corporation was supposed to be bonding them down here in Arizona. And they would be made in Cupertino, California, and the inner lead bonds would be made up there, the outer lead bonds were made about 3 or 4 weeks later in Arizona.

Well, I told you cleanliness is next to godliness, and you should bond within an hour. 3 or 4 weeks is more than an hour, okay. They were having terrible problems with bond quality, and there's 400 inputs and outputs, and one bad joint means you got a piece of junk. And so their yield was essentially near zero. And this was back in the days when in the late 80s IBM was the world's market leader in uh computer technology, or had been, because they had this $40 billion business which today would be equivalent of a $100, $120 billion business on mainframe computers. PCs were not yet powerful enough to do a lot of the number crunching.

And so it turns out that uh Digital Equipment decided that they were going to go after this $40 billion business instead of mini-computers, which were before PCs and which are microcomputers, um, they were going to go after the mainframe computer business. And they invested most of their several — well they invested several billion dollars. They never really got the VAX 9000 to work very well, they didn't sell very many. They gave one to MIT, okay. Um, but they uh, they essentially killed Digital Equipment, they went bankrupt by the mid-90s because of that VAX 9000 project primarily, and a few other poor choices, like they didn't believe in RISC, uh, Ken Olsen, who had founded Digital Equipment.

And by the way, early on MIT owned 7% of Digital Equipment, but they decided it wasn't appropriate for a university to own that much of a company, so they divested of themselves well before they could have made a few hundred million dollars off selling their stock. But nonetheless, Ken Olsen thought RISC, reduced instruction set computing — are you familiar with the difference? A PC runs on what's called complex instruction set computing, but most of your Hewlett Packard printers run on reduced instruction set computing, where you don't have a whole lot of — it's in your, what do they call the uh, the machine language, not your software that you know Java or something that you're used to, but if you go to the machine language, which someone has to still program that chip in terms of binary code. Their complex instruction set was what everyone had used for 35, 40 years.

Um, and around 1990 people came along, and computer scientists said you don't need to have so many um mechanisms or instructions in your binary code, you can have a reduced set of 10% of those things, and you — it's more lines of code, but you can — it's going to be simpler, okay. And Ken Olsen, president and chairman of Digital Equipment, said RISC is a fad. And I remember guys from Hewlett Packard saying great, he doesn't understand where the world is going. And it turns out they were absolutely right. That was what allowed Hewlett Packard to grow. They were not a computer company in 1990. They were sort of making some printers, and their printer business took off as over half their business, and it was because they were doing RISC, uh, programming of these things, and it turns out it was just the right niche. Meanwhile Ken Olsen was trying to make a VAX 9000 thing that didn't really work, and Digital Equipment no longer exists.