CS_Su2012_02

One, I think we got most people here. Um, so if you had to pick a theme for yesterday's lecture, what would it be?

Well, this is actually more important than you think. And uh, a large part — it was a real revelation to me when I was a junior here. And for whatever reason I decided to take introduction to quantum mechanics in the physics department. And uh, I was taking this course — you have to understand, I came here, I was in the bottom third of my entering class, okay. And they put us in Kresge Auditorium, they said, oh, you know, fifteen years ago we'd say look to your right, look to your left, one of the three of you won't be here next year, okay. And they said we don't do that anymore. And they flash up the, you know, the high school SAT scores and stuff. And I was smart enough to know that I was in the bottom third, okay, and so one of the three of us wouldn't be here next year.

Um, actually MIT doesn't treat its freshmen like that anymore. They actually have a fairly high graduation rate, like ninety-six percent of the freshmen end up graduating. Um, some schools are higher, like Princeton is like ninety-eight, and in fact Princeton, they really pushed the students to graduate on time, okay. Uh, and in fact MIT's graduation rate on time is only about ninety-six. So we don't really flunk people out, but we do make it difficult for them while they're here.

But it was a great revelation to me in my junior year, to take this course, introduction to quantum mechanics, and I didn't have a clue what was going on in this course, okay. I was getting fifteens out of a hundred on the homework sets and the averages were eighty-five. And I was just — I didn't know what's going on. Um, so the night before the three-hour final — and I already told you what I thought of three-out-of-five three-hour finals — I decided, okay, I'm just going to take the book and I'll try to pick out the highlights, okay. And I walked into the final, three-hour final, finished in an hour and twenty minutes, checked it over another twenty minutes, walked out, and got an A in the course.

That's odd. What? And then all of a sudden it dawned on me. It's not all this detail the professor goes through, like cow magnets, or passing out these little half-inch balls of different materials. Um, that's all fluff. They're only trying to get a couple of points across in each lecture, and they're trying to hide it from you, okay. If you can figure out one or the two or three points in a one-hour lecture and focus on those, then you actually can do quite well around here, okay. And so that's what I did for the rest of my career here. I quit taking notes, I just sat there and tried to see if I could guess the professor's lecture notes, okay, uh, what his outline was. Uh, and anyway, it's worked for me. I've tried to explain it to students over the years, I don't know if anybody, um, knows what it is.

Anyway, I gave you a little quick introduction that we're going to go over codes and standards. You might — so getting back to what was my lecture yesterday, and I started talking about measurement leads to science, okay. There's the Lord Kelvin quote. And then we start talking about observables, and the real theme on observables is — I actually, I wanted to take you from the sublime to the ridiculous. That was what my goal was yesterday. The sublime is LIGO, one part in 10 to the 22nd, okay. And the ridiculous is I get up here with a grinding wheel and start showing you sparks, right.

So the point actually is that if you're going to do science you have to measure and get data and work on facts, okay. But you don't have to spend a half a billion dollars to measure gravity waves in space, in one part in 10 to the 22nd. Although there are people out there who do that, most of the time it's not necessary. A lot of times a cow magnet up against a piece of metal will tell you a lot. Or taking a grinding wheel and grinding the surface of something will tell you a lot. So there's an old Yiddish saying that Einstein used to quote, "as simple as possible but no simpler," okay. And a large part of engineering is to get things simple.

And so that was yesterday, the sublime to the ridiculous in measurement, because measurement is fundamental to science. So today I want to talk — I'm going to tell you today what my theme is, okay, you don't have to guess it. I'm going to talk about what's the practical limits on some of these things that we try to measure, okay.

There are many properties of materials. We did some of them when we took the little half-inch spheres, a lot of which are ball bearings materials. There's a brass one, there's a bunch of polymer ones, there were some ceramic ones. I was actually a little surprised nobody could figure out the aluminum oxide. You're clearly not ceramists, okay. But I thought some of you might get it based on — actually, you were doing quite well when you were bouncing things. That's a hardness test. Did you know that, that's a hardness test on a material? There are many portable hardness testers, there are thousands of hardness testers. But one of them is the scleroscope, and it basically takes a little steel ball and bounces it off the surface. The problem you had is your table wasn't stiff enough, okay.

That's right. Well, the scleroscope has a little tube, and so you actually bounce — you measure the rebound height, okay. It's called the coefficient of restitution in physics. So I mean, I call it rebound height, but you can call it — if you want to sound scientific, you call it the coefficient of restitution, okay. Did you know the ping pong ball has a coefficient of restitution of 0.97? Pretty good.

Right, so they put limits on the coefficient of restitution so it's a fair game. The rich can play — a fair game, a fair game the rich can play golf. I mean, you know, I mean, you'll hear me saying some of the videotape lectures that you can sell anything to a golfer. Um, anyway.

So there are many properties of materials, and we're going to today get to some of those, but first I want to talk about the practical limits of some of these measurements. Now, we talked about mass, length, and time, and I want to talk to you a little bit about what are the practical limits, because I actually have something I want to get to which might surprise you, where I've been heading with a lot of this.

So mass. If I want to weigh something, how many parts per thousand or parts per million — and when we were talking about the NIST people doing the seven fundamental, you know, speed of light and the mass of the kilogram and Planck's constant and these things, are all about one part in 10 to the seventh or 10 to the eighth. We know those to seven or eight significant figures, right. If you go out there — if I sent you over to the lab and said weigh this piece of chalk, how many parts per thousand or a million do you think you might be able to weigh this to? I mean, anyone ever work in a chemistry lab, you had to weigh things before you did an analysis or something? Yeah.

Okay, that I can do about a part per thousand with an electronic balance in my kitchen right now, okay. Maybe one part in 500. So with electronic balances I actually can do pretty good. Um, and when you in the chemistry lab, did you see this marble table with a balance on it that had doors on it? So you put the sample in, you close the doors, and weigh it on this big massive table with doors to isolate it. Why were you doing that? It turns out you can get about 10 micrograms out of a gram, okay, which is 10 to the minus 5, so one part in 10 to the fifth.

Now typically it's about one part in ten thousand — you said one part in a thousand, you weren't far off — for everyday weighing. But if I wanted to be precise, what limits that? And think about why am I doing it on a big marble table, why do I have to have these little doors that close around the thing. Did you have to wear gloves? Well, if you did, but you had — or you've lifted things with tweezers, okay. I'll just give you a hint — the three most common errors in weighing something, okay, if — now you're thinking too scientifically. Fingerprints. That's why you wear gloves or use tweezers. Wind. Or what was the other one? Oh, vibration.

Temperature — because at MIT we're going to teach you to think fundamental physics, you know, what are the fundamental thermodynamic parameters, temperature, pressure, you know, and whatnot. But multiple sense — if you're trying to weigh something, if you blow on it — I mean I'm talking about something that's on the gram quantities, right. If I blow on the scale, I will be measuring one part — the wind will create one part in 10 to the fifth error. My fingerprint will weigh one part in 10 to the fifth, okay. Not if I'm weighing something that weighs a thousand pounds, a fingerprint won't, but other pieces of contamination might do, okay. So it's difficult to get much better than one part in 10 to the fifth. You can get better than that, but you have to kind of go to this type of, you know, very expensive things. One part in 10 to the fifth is pretty good.

And for that reason, when you do a chemical analysis of a material, the most accurate chemical analyses have historically been dissolving the thing in acid and analyzing that way. Because you start out by weighing the sample that you dissolve in the acid, and then measure the concentration. So it's actually a weight basis, and one part in 10 to the fifth is pretty good, okay.

So that's mass, and the big problems are things like wind, fingerprints, and vibration, and little things like that when I start getting into the fifth decimal place, we start getting into those types of things. Length. What type of — if I'm going into — anybody done any machining in a machine shop, or seen anybody do it? Okay, so what's it — if I'm going to machine a one-inch diameter bar, what's a typical tolerance I might get if I sent you into the laboratory? What would you come up with if I told you to measure it to a hundred? What would you actually get, in parts per thousand?

Oh, less than what? One? Less than one part per thousand. How accurate can I get it? It's going to be one inch plus or minus how many thousands or millionths of an inch? Okay, I'm just doing a lathe. 100 micro inch? What, 100 micro inch? I like 100 micro inch. 1/100? Oh, one — one — oh, one, okay, 1/100th is what I expect most of you to get, okay. Uh, that's ten thousandths of an inch. Okay, that's not very good. A good machinist will do — can easily do one part in a thousand, plus or minus one mil, one thousandths of an inch. And the machinist who's been around the block a couple of days can get half a thousandths. So typically one part in two thousand. Um, half a thousandths per inch, or if it's two inches long or two inches in diameter, come up with one thousandths, okay.

What's the limitation in coming up with that? Relative stiffness of your tool and your work piece, okay, good. But there's something in the environment, okay. It's called temperature, okay. I mean, um, you know, that's no reason you should be able to guess these, but temperature is an important problem in measuring length. The coefficient of thermal expansion of steel is about 10 to the minus 5 per degree C. So if I want to measure something one inch long, a one degree change in temperature of the room will change it by 10 micro inches, and a 10 degree change will change it by 100 micro inches, okay.

So 10 degrees centigrade is not that big a change. I mean, if I measure it outdoors in the winter, okay, as opposed to indoors at ambient temperature, I can be off by a tenth of a thousandths. And there are many bearings and such things that require a tenth of a thousandths over one inch or half an inch. Anybody ever seen a ball bearing micrometer? Yeah, what's it look like? I mean, you've seen a micrometer, this little thing I hold in my hand and I kind of twirl it and it measures the length of something. And most of those are accurate to about a thousandths or a half a thousandths, one part in two thousand, okay. That's a simple little thirty-dollar, hundred-dollar tool.

But a ball bearing mic — you've actually seen one? They're about the size of a lathe, okay. And they can measure to 50 micro inches. And when you are working at Pratt and Whitney or General Electric aircraft engines, they have a bunch of these ball bearing lathes, because they make ball bearings just like those little steel balls that I gave you. And they make them — they try to make them uniform, but they can't make them all uniform within 50 micro inches. So they have an air-conditioned room for temperature control, and they have these — what you know of is a little ball mic, you know, what they call a ball mic, but a micrometer, they cost a hundred bucks, which the machinist uses to measure something while he's working on the lathe or something. But they have something that looks like a lathe, and they put that ball bearing in there, and this thing has got all kinds of compensation, and it can measure down to 10 micro inches on an inch, one part in 10 to the fifth, okay. Which is about — oh, about the same as on the mass scale. But it takes, you know, a ten thousand dollar measuring machine to do it.

So what do they do? They'll have a hundred thousand balls, and they will grade them, because they're all half inch plus or minus less than a tenth of an inch. But they will then sort them into "these are within 50 micro inches of this value," this reference value, and these will be within this, so they match them. And then they make their high speed precision bearings up from a matched set of these balls, okay.

So we're kind of pushing the limits of practicality when we get into one part in 10 to the fifth on measuring length. Now, I can buy things — anybody ever seen Johansson blocks? Jo blocks. These are actually Webber blocks because they're made by a different company. But they're precision length scales. These are actually made out of chromium carbide. You can buy them made out of steel. But if you're careful, they have lap surfaces, and I can ring them together if I'm careful.

[Tom rings two precision blocks together.] The little bit of moisture or grease that remains after I wiped them on my shirt will hold them together as the adhesive, okay. So if you want to pass these around, you can't ring these surfaces — the ground, these ground surfaces — together. It's only the end surfaces, okay. And you have to — it's called ringing them together, you slide them together and stuff. These have a certificate of standard that comes with them, and it says these are accurate to, um, plus or minus — well, it's grade one, two, and three, and grade three is plus or minus three micro inches per inch. You can get a laboratory master which is federal grade 0.05, which is 0.8 micro inches, so that's less than one part in 10 to the sixth. But I can get these in one part, or three parts in — three parts per million, okay. I can get those — are laboratory standards. Well, what good are they?

Okay, well let me tell you the story. This is a Navy story, sort of a Navy story. There is a helicopter company, a company in Connecticut that builds helicopters, that remain nameless, but they're in Stratford. If anyone knows, there's not a lot of helicopter companies in Stratford, so you can probably guess. So the Navy buys uh, MH-53 helicopters. Anybody know the MH-53? Yeah. Okay, big helicopter, right? They use it to pull minesweeping sleds and things like that, right.

So these are like twenty or thirty million dollar helicopters, they're not cheap. So, um, they were manufacturing one of these things for the Navy, and the Navy had not yet taken custody. There is a standard — I can't remember the number, it's a DOD 501 form or something. There's actually a form that has to be signed off for when the equipment, whether it's an aircraft carrier, whether it's a helicopter, or when someone's manufacturing something for the military, there's a DOD form that when it's signed is when the ownership transfers from the manufacturing company, whether it be Pratt and Whitney or Northrop Grumman or whatever, to the Navy. And when this one single page piece of paper is signed by the company and by the Navy, the ownership transfers. Why do you have to worry about that? Well, let's say there was a fire and the whole thing got destroyed. Well, who takes the loss, the Navy or the company that was building it, right? So it has to be a point in time. There's a standard time, so this is another type of standard, if you will.

So they had not yet signed this thing. The helicopter was in a — it had part of the spec, it had to hover for, I think it was an hour. So it had two pilots, and they were just hovering about 50 or 100 feet above the tarmac at the airstrip there at the manufacturing facility. And all of a sudden while they're hovering, the swash plate bearing — people know where the swash plate is on a helicopter? Anybody know what swash plate is? It's the control surface, this part of the rotor. Yeah, it gives you cyclic and stuff, but basically it's a plate, a circular plate that you can rotate in two dimensions, right, X and Y directions, and that will allow you to tilt your rotor blade. If you want to turn left, you tilt your rotor blade to the left. If you want to turn right, you tilt it to the right — I was doing that in your coordinate system, not mine. Or if you want to go forward, you tilt it forward so it pulls the air in like a fan blade. Or if you want to slow down, you can tilt it back. So you can tilt this swash plate, and it's what gives you motion control, turning right, left, forward, backwards. And to go up or down, you basically just increase the speed or the power to lift, or whatever.

So all of a sudden the bearing started to seize, and it seized, and the helicopter crashed and the two pilots died. And the Navy said, oh well, we had not signed the DOD form, and so therefore you, helicopter company, can run your own investigation of why you lost this helicopter. You still have to build us another one, okay, because we ordered one and we're not going to accept this one, it's not in acceptable condition after the crash.

So, because if it actually had been — if it occurred after they'd signed the DOD form, when some Navy pilot was trying to fly it away, they would have formed a joint commission to figure out what went wrong, okay, and there would have been some Navy officers who had been put on that commission. And they would be just like you, they didn't know squat about how to do a failure investigation. But they would have been, okay, you're now the failure analysis expert, and you're going to figure out why this thing crashed. And you'll have all kinds of people coming to you and say, oh, I think it was the phase of the moon, and you have to figure it all out, okay. But you're smart, you graduated from MIT at this point, and you'll be able to figure it out, good luck. They're actually fairly complex structures.

But in any case, so this thing crashed, and the Navy said, ah, we hadn't signed off on it, it's yours, you can do your own failure investigation, no Navy officers on this. Well, this is sort of like the fox watching the henhouse. The Navy wanted to know what happened, because they don't want it to happen anywhere else. And in fact over the next few months they actually did have — actually the Japanese Navy had one go down, and they fished it out of like 5,000 feet deep, because ocean's pretty deep right off the coast of Japan. Um, and then they had a couple go down at other places, and so Cherry Point was just collecting old, hulled MH-53 helicopters. And so the Navy did start to get interested in this, okay, because it was happening in service.

But what happened — it was a — there was a ball bearing in there, and that ball bearing — I've been to the plant where they manufactured them — it had — it was 42 inches in diameter and it had a tolerance of plus or minus two thousandths. Well, plus or minus two thousandths over 42 inches, you can figure it out, is about one part in, uh, ten thousand, okay. That's about five times more accurate than if I sent most of you into the laboratory with the machinist of what he could accomplish.

What they had — when they would machine these things, and they were called Ready Slim bearings, so they were only a half inch in diameter — 42 inches, half inch diameter, this thing's flexible, I mean, you pick it up, it'll bow, okay. So it doesn't even have its own shape, it gets its shape from the aluminum swash plate housing which is massive, I mean, it's, you know, several inches of aluminum.

And uh, so the helicopter company — I have to be careful not to use their name here on tape, but they are in Stratford, Connecticut, so someone could figure it up. Um, but they decided to form their own investigation committee, which they did, and they put their attorney on it. And the attorney was smart enough that whenever they started discussing anything substantive — it was all sort of open meetings that they taped — he would go off the record, and so there were no tapes whenever they started talking about anything of importance other than what we're going to have for lunch. They would tape that, you know, but if they'd start talking, why did this thing — why did the bearing seize suddenly catastrophically — uh, that was always sort of, oh, we need to go off the record, the attorney for the company would say, good time to go off the record. This is why it would have been good to have a Navy person there, okay. You can say, no, I think we ought to stay on the record, okay. This is how you could have contributed to this.

Anyway, they come to the conclusion that — because there had been a bearing produced by this company that makes these bearings, and there's only one company in the world that could make these bearings — um, I'll just use this board. They had once produced one of these bearings with a closed conformity. So if the bearing race looks something like this [Tom draws on the board], a closed conformity means the ball is a little too big for the race, and it touches on the endpoints. It's not a good design, okay. A lot of wear very quickly, the balls get scored, they get where they start to gall, you have a catastrophic failure. They didn't have a catastrophic failure on that, but the company actually had produced one and sold one, it got out of their quality control, and they still don't know how it got out of their quality control. But it had a closed conformity.

Here's an open conformity. You want — an open conformity, you want these little balls to be — that's not a very good circle — but smaller diameter, so the contact point is right there, so it can roll around the race, that's good. So this is closed conformity and this is open. This is bad, that's good. They made six bearings at one time, and one of them had a closed conformity. Someone noticed it before anything happened. And so what do they do when this helicopter crashes in Stratford? They conclude, oh, it must have had a closed conformity. And they say, ah, too bad, can't prove it, there's no observable here, because it crashed, the whole thing is just torn to pieces. But that's the reason it is, and it's the fault of this bearing manufacturer. So FBI, come in, okay.

So the FBI comes in, they quarantine all the bearings, okay. In the meantime, the Navy is stopping flights of all MH-53s while they've tried to figure out what's going on, particularly after they start investigating the Japan accident. And they had a couple of close calls, I think one of them was landing at Cherry Point, and as he's landing, and he's a couple of feet off the ground, all of a sudden his bearing starts to seize. Fortunately he was only a couple feet off the ground, he lands, no one gets hurt, but they couldn't start it back up again because the bearing was seized, okay. So they had a couple of close calls.

Student: [question about open vs closed conformity terminology]

Yeah, uh, maybe it is. I thought that was based on what you said. Oh, because — I think you said closed was when you were heading — oh, it may have — I said what this is — preferential wear, and that's bad. So I know that's bad. I may have my open and closed wrong, okay. It's been several years. Yeah. Okay, which one is called open and closed — don't worry, you won't be quizzed on it, okay. But anyways, one was open, one was closed, and they decided, oh, it must have been a closed conformity — you must have produced an open conformity that was bad just like you did once before. And this is, you know, they didn't have any data to say this, but they started criminal charges against the bearing manufacturer.

Um, well, I was brought in — not for the criminal charges, but then they said, oh, bearing company, you owe us twenty million dollars for our crashed aircraft, but it was only — we make about five million profit, so you only owe us fifteen million. So all of a sudden there's about thirty million dollars worth of lawsuits. I come in for the bearing company along with some other people who are experts in bearings, and we go out and we look at these bearings out in Michigan. And whenever we looked at them, we had to have an FBI agent standing next to us, okay, because there was a criminal investigation. We weren't involved in that, but there was one going on, so you couldn't look at it without a FBI agent present.

And um, on about the second day all of a sudden it hit me. I said, this can't be, um, the wrong conformity. Because if I had this type of conformity, we'll call it Type A and Type B, how about that — they're open and close — and this one, it turns out, when you get the high speed and everything and it's starting to gall, it heats up. And what happens is it squeezes everything together, and the balls get squeezed into the race.

What would be the shape of a spherical ball being squeezed into a circular race? If you had this type of contact, it would be an hourglass, right. The wear mark of the ball being squeezed into the race — this is, looking — it's like the ball here being squeezed into a channel — it would make an hourglass shape, right. And if I have something like this, it would be a barrel shape, okay. So if you squeeze the ball in here, you get an hourglass, here you get a barrel shape. And I had 192 balls, and I could actually, after that incident, I could measure 188 of them, and they were all hourglass. So what's the chance I had the wrong conformity? Zero, folks, okay.

So I just, all of a sudden, and everybody, you know, a hundred engineers at the helicopter company and in the FAA — not the FAA — but a lot of people have been looking at this, and everybody was trying to point their finger at this manufacturer of the bearing, saying you screwed up, you killed two pilots, their widows are asking for fifteen million dollars, and the helicopter company wants fifteen million dollars for the loss of the hull. Yeah.

Student: [question about whether the wear marks were hourglass or barrel]

No, there were no — the ones we saw were barrel-shaped, okay, makes sense. They couldn't have been an open — they couldn't have been this type of conformity, they had to be this type of conformity, they were claiming, because they had once made a set of six bearings, and one of them had been the wrong conformity, didn't cause a crash.

Anyway, so now we had to figure out why the thing had failed. But I learned a lot about the bearings — to make something that was 42 inches in diameter, which is this whole — this thing going around, right — 42 plus or minus 0.002 on the diameter. They had a micrometer that would measure 42 inches — a big long pipe with a little regular micrometer like you use, you're used to using the machine shop. But it was on a pipe, and they would put it in the machine with the oil bath that was doing the machining of this 42 inch diameter half inch square race. So it was at the same temperature as the oil that was bathing the whole thing. So I wasn't trying to get a precise measurement of 42 inches plus or minus two thousandths, I was doing a reference where everything was at the same temperature, okay.

It turns out, when they said it had to be this, they had to define the temperature at which — on the drawing — at which you would make the measurement. But in order to machine it, they didn't care if you were five degrees or 10 degrees above or below, just as long as your micrometer was at the same temperature, and they're both made out of steel, they would have the same coefficient of thermal expansion, right. So that's how they did it. That had nothing to do with the conformity, but nonetheless we're looking at this and I say, it can't be the wrong conformity, we had the proper conformity on this bearing, why did it fail? You got to come up with an alternate hypothesis.

So we started looking a little bit more. They weren't all equal barrels. Some of them were skinny barrels and some of them were fat barrels. And they actually went in quadrants, okay. And you could look at the quadrants, you had — at first and third quadrant you had fat barrels, and in the second and fourth you had skinny barrels. Well, why would that be?

Use my photos. Going around the whole 42 inch bearing, okay, you got 192 bearings, so it's 48 in each quadrant. And you kind of look at this and you see the barrels change shape, okay, from fat in the center of one quadrant to skinny in the next quadrant to fat in the next quadrant. Well, how can that be? This outer race is only one half inch by one half inch, but it's 42 inches in diameter. It doesn't have enough stiffness on its own. It gets its stiffness from the aluminum housing it's in, right. And if I'm getting different shaped barrels in each quadrant, that means my aluminum housing is potato-chipped. You know, a potato chip is, it's a saddle shape, okay. It's not circular.

And so we went back, and we said to this — it turns out the helicopter company made the aluminum housing — we said, what, uh, horizontal or vertical boring machine did you machine the aluminum housing on? And said, oh, we scrapped that, it was three years old, we sent it to the scrap yard, that's been destroyed. Oh, three-year-old, thirty thousand — you know, three hundred or three million dollar machine, and it's time to scrap it, it's three years old. It's called hiding the evidence, okay.

And so the upshot of the whole thing is — we actually ended up having like five or six, uh, helicopters — I had to go down to Cherry Point a couple of times and look at destroyed bearings — five or six helicopters, only one of them killed anybody, and that was the one in Stratford. But they had a milling machine that wasn't square itself, it was making distorted aluminum housings. They knew it, they hid it, and they passed the buck to the manufacturer of the bearing. And, you know, basically the bearing manufacturer ended up — they got out pretty cheap on the criminal case, um, but they uh, ended up paying off the helicopter company for the hull.

And the reason was, they had a new manager or president of the bearing company, and all of these problems — the improper conformity, the failure in Stratford — didn't occur on his watch. And so when he came in and he had this thirty million dollars worth of lawsuits in his first, you know, hundred days or whatever, the honeymoon period, he could say, let's pay it off and get rid of it, and he wouldn't have to explain to the board of directors of his company that he had been the problem, right. So he paid it off. The helicopter company got their blackmail.

And uh, fortunately, a few of us were able to kind of tell some of the people in the Navy, it's not the bearing, it's the swash plate housing. We had to say this in secret, okay, because the official story was it was an improper conformity bearing — whether it's open or closed I can't remember, but anyway. Um, that's the politics of it all too, okay.

But in fact the precision machining of these things is pretty interesting, I mean it's — that was one part in ten thousand, which is five times better than the accuracy. But it had nothing to do with this, the steel ball on the steel race. It really had to do with the steel race inside the aluminum housing. But fortunately the helicopter company had been smart enough to scrap the three-year-old milling machine. I mean, you know, three years old, you get rid of milling machines, right.

Oh, okay, so that's the story that goes with that. There's a lot of things in these stories, such as the politics of this whole — why they paid off fifteen million dollars, because the new manager didn't want to — you know, he could do it without taking the blame. He could pass the blame to somebody else. You should understand these things because you could be on either side of that equation someday in your career, okay.

Anyway, so mass, length, and time. We want to talk about tolerances on time. Uh, yesterday someone mentioned, uh, quartz clocks, okay. One of you said quartz, and you were right, in the old days. Well actually, um, you're in the Navy — what was driving the need for precise time measurement? Yeah, yeah, navigation. They could do just fine with latitude, right, looking at the angle of the sun, right. But what do you do for longitude, right, there at certain points, right. Yeah, so you need accurate clocks. So the most accurate clocks were being developed in the 1600s and 1700s for navigation, for marine navigation.

Um, anyway, so they had mechanical, and it turns out mechanical clocks — the question is how many pieces can you break a second into? And if a mechanical clock works at 10 Hertz or 100 Hertz in terms of that little mechanical movement, or even less, you're going to end up with a fair error, because if you're off by one percent, uh, in that mechanical movement at 10 Hertz, then that's going to add up over the day. So they got quartz clocks. And so I'll just say 10 Hertz, um, so that's one part in — but seconds, and then you can decide how long you're going to go for a day or whatever. Then we have quartz. Anybody have any idea what the frequency of a quartz oscillator is? On the order of 10 to the fifth, okay, Hertz. So a hundred thousand Hertz, that's a significant improvement, good enough for most of us. That, uh, you know, I think this is a quartz clock, okay.

Um, but the atomic clocks — and that's at about 10 to the 10th Hertz. So it was Isidor Rabi, one of the physicists in the 1930s, who suggested that we could use the vibrations of atoms at very high frequencies as a time standard. And then, before the laser, they had masers, from basically — came out of radar in World War II, which were microwave. Remember, laser is light amplification by stimulated emission of radiation, or something like that. Uh, that before that they had masers, microwave-assisted stimulated emission of radiation. And in fact, the guy working on masers who invented the laser — there was a dispute about it, a patent dispute, just got resolved recently in the last 10 years. But the guy who won the Nobel Prize for the laser was Charles Townes, and he at one time was Provost here at MIT. He was at Bell Labs when he did his laser work, but anyway, he won the Nobel Prize for light amplification by stimulated emission radiation.

But masers — and I was wrong, they did some of the work up here at, uh, Harvard Smithsonian, but the first maser — I looked it up this morning — the first maser for atomic clock was done at the United Kingdom in their measurement laboratory, okay. Today they actually can get about one part in 10 to the 14th, if you're talking about averaging over a day. Uh, most recently these folks in the UK have come up with improved systems. How do you get something better? You go from microwaves to light. And now they have Bose-Einstein condensates, where they cool individual atoms down to within a fraction of a degree of absolute zero, and they go into a quantum state where the vibrate — they can measure the vibrations very precisely, and now they're at 10 to the 16th, which is good for one second in 138 million years, okay.

So any of you who need to be on time and you've got a long time scale for this next meeting 138 million years from now, okay. Um, so you actually get to the point of, who cares, okay. But someone out there does. Um, because — and some of them care just because it's science, okay. But now how do we use some of this, all this — you know, these are the types of numbers, one part in 10 to the fifth for mass, one part in 10 to the fifth for length, on average — you can get better than that — and time, one part in 10 to the 10th, 10 to the 12th, or something like that. And that's why we do a lot of things based on time. The meter is defined in terms of time, or the frequency standard and the speed of light, because we can measure it, it's at such high frequencies.

In any case, so now the real upshot of all of this is — I want to talk to, you know, a manufacturing person. What is the most precise manufacturing that we do on a regular basis? Is it making one of these little ball bearings to 50 micro inches on an inch or a half inch? Well, that's about one part — that's in parts per thousand, that's one part in ten thousand, okay.

Microprocessors, it turns out you're right. But 10 years ago you would have been wrong. But in most examples or questions of that sort, you need to go to the extremes. And one of the extremes is, what's the smallest thing that we manufacture right now? It's a computer chip, right. And so here, if you want to see — typical dimension of a computer chip is about one square centimeter. And there they have features — that's an old Pentium 5 or something, that's what I got, what can I say — um, is now about 10 nanometers. I don't remember, it might be something a little bit larger like 11 or 12, but order of magnitude 10 nanometers. And 10 to the minus 2 meters, or one centimeter, which is one in 10 to the sixth, okay.

But I said you go to an extreme. One extreme is extremely small. What's the other end of the extreme range of manufacturing? Large, thank you, okay. That's an MIT answer, that's a 40,000 foot view answer. Yes, large. What is the largest manufactured object you can think of? Yeah. Okay, they actually have larger ships, longer ships. I'm going to say 300 meters long. What is a carrier in length? It's about 300 meters in — they have some ultra large crude carriers that I think are longer than aircraft carriers, but yes, for the Navy it's an aircraft carrier, 300 meters. And when you're putting that hull together, you better bring those plates together within about a millimeter, okay, side to side, right. So 10 to the minus 3, and that's 3 times 10 to the sixth. So ships are almost as precise as computer chips in parts per million, okay. So you guys didn't know you worked in precision manufacturing, did you.

Okay, so I was down at NIST in March, and this person, Catherine Gabby [Gebbie], who heads up their measurement system in there, which is now part of their manufacturing program — Catherine is well into her 80s, but she looks like a very prim and proper Bostonian, okay, woman. And she actually graduated from Bryn Mawr, came to MIT in her last two years at undergraduate, got her PhD in phys— or got her bachelors in physics at MIT, but because MIT physics department will not allow their undergraduates to go on to graduate school at MIT — they say you should go somewhere else. Same thing happened to, uh, Richard Feynman. He was an MIT undergraduate, and he couldn't get into graduate school at MIT, not because he wasn't smart enough, but because the physics department here thinks you should go somewhere else, you don't spend your entire career at MIT. And that's true, there are other good physics departments. We don't say that in the materials department, because there is no better materials department than the MIT, why should we penalize our students by making them go somewhere else.

Anyway, so Feynman didn't get his PhD from MIT, he went to visit — to Princeton. And um, Catherine had to go up to the school up the road at called Harvard, and she got her PhD there. And she's been working on atomic clocks and heading up this whole area. She's actually quite a gracious lady, and she had talked to — to tell — that was the first time I had — I might have read once about the rice-size atomic clock that they developed in 2004, which as of last year you now can purchase commercially from a company that makes them.

But anyway, I went — she had said that their big problem is they're trying to, um, come up with a uniform standard for mass. The problem is mass — I told you yesterday — is there's still this weight of a platinum-iridium mass held in the vault in Sèvres, France, and is only taken out of the vault about every 10 years. And so far as they can tell, it's losing weight at 60 micrograms per century, or something like that. And so I went up to her at the break, I said, why is this thing losing weight, and how do you know if it's the reference standard? She says, oh well, we weighed it against a bunch of other reference standards, and they're either — they're gaining — they're all either all gaining weight at the same rate, or this one's losing weight. I said, well, how can it be losing weight? And she said, oh, well, platinum oxidizes. Excuse me, I'm a material scientist, I didn't know that platinum oxidized a whole lot. But apparently before they weigh it — and it's been in a vault under vacuum — they wipe it down or something. So they're wearing it away. You know, every 10 years they wipe it with some cloth before they weigh it. And that's probably — but it's all statistics that they're dealing with, okay. But they think it's losing weight, so they want to define it in terms of some, um, non-variable standard.

And so they're going to try to relate it to Planck's constant. So this was part of her talk, and I was sitting there — of course I'm sitting there worrying for the next 45 minutes, how do you get mass out of Planck's constant? And the only thing I could think of was, just getting my, you know — it took me 45 minutes to think of it, but Planck's constant is, E is equal to h nu, right, where that's energy and that's Planck's constant, and that's the frequency of whatever it is. This is the formula that Einstein won the Nobel Prize for. He didn't win it for relativity, he won it for the photoelectric effect, and this was the formula. But there's also energy is equal to m c squared. And if you put these together, you can get E is equal to hc over lambda. And so here's mass, here's h, you equate these two, and you got the speed of light and you got a wavelength, and so you're off to the races, you can now define mass.

And I said, so is it just using E is equal to mc and E is equal to h nu? She said, oh no, it's more complicated than that. She says, I'll get you a paper. Um, and so the paper that I gave out to you is the same paper that she gave me. And it turns out, I read the paper that night, and it turns out it is just these, but it's a little more complex than that, she's right, okay. So we're both right, uh, it's a little more complex.

So in any case, I go up the next day, and I'm just kind of chatting with her, and saying, well, it is sort of just these two, but you're right, it's actually — I found the paper very intriguing, to see what these guys — give them an unlimited budget to go and see if you can come up with another decimal place in these constants, sort of interesting, okay. Um, and I said, but do you know what the most precise manufacturing is? She says — I said, in parts per thousand. I mean, you know, you can machine a little one-inch bar to one inch in diameter and, you know, how precise is that? She said, well no. And I said, it's shipbuilding, okay. And she says, it is? I said, yeah, the bigger it gets, the more — oh, that makes sense. So she accepted that. And then she says, well, but how do they do it, because the problem with shipbuilding is, if the sun goes out and comes out, or goes behind the clouds, the length of the ship changes significantly, it can grow a half an inch depending on the time of the day, right. And you've got to be more precise than that. I said, well, they use big hydraulic rams, that's how they do it, right.

So that's sort of the sublime to the ridiculous. You know, how do you do it? You just go in there and do it by brute force, right. If you've been to a shipyard, you've seen the rams, you've seen the strongbacks, okay, they just pull it together, right. And you hope that the sun doesn't get too hot in the middle of the day, you change all your dimensions, right.

Anyway, so in a sense, I mean, there was a few other things you might have hopefully learned on the way, but one of my themes in this whole thing was to teach you that you were working in one of the most precise industries in terms of parts per thousand, okay. The guy who first pointed this out to me was Dan Whitney over in mechanical engineering. Um, so far as that goes, okay. Any questions?

Now I want to — oh, first I'll introduce Dr. Simone Belmar. Hi, Simone will be lecturing tomorrow on safety factors, okay, which is sort of a standalone thing. So I won't be here tomorrow, but I will be here on Thursday and Friday of this week. And I should be here next Monday, and anyway, whoop, we made — I don't know if we'll finish next Monday and Tuesday probably. I won't be here Wednesday, Thursday, and Friday, I have a son who's graduating. But anyway, another thing we did yesterday, um, well, let me back up. You don't have any questions, huh? If you ask me a question, I can tell another story. I had to tell you my own story about Stratford, Connecticut, and helicopters and DOD forms.

Student: [question about the bearing specification]

And yeah, the bearing manufacturer, when they make these, they're called Ready Slim or Really Slim or something bearings. If you needed a stiff bearing, you'd have to make it several inches in cross-section, not a half an inch by half an inch, and that would be too heavy to fly for a helicopter. So they put it in an aluminum housing which is lightweight, which gives you the stiffness. And the thing is — was there anything wrong with the specification for the bearing, okay? And let me tell you, the FBI was looking at the drawings, okay, and going through all the specifications for that bearing. That drawing for that bearing is a piece of military hardware, when they sold it for a military helicopter. If it was a commercial helicopter, there'd be a separate set of drawings and it wouldn't be military. But if it's a military piece of hardware, the DOD owns that drawing. And the question for the FBI was, did they not meet specification, and why was the criminal intent, okay. And the FBI was ready to crucify this company, okay.

Another problem in this whole thing, the FBI actually came in and started taking over the quality control section of this bearing manufacturer in Michigan. Why? The whole fleet was grounded, okay. This had risen all the way up to the Pentagon, and people were saying, if, you know, if we have some problem, you know, someone attacks us, we don't have any MH-53s available. So whenever something like that happens, it gets to very high levels, and people start paying. And they start calling across the river, and they call the FBI. And the FBI came in because the government has the right to take over a manufacturing facility for national security interests, okay. And they basically did that. They came in, and because the helicopter manufacturer said, oh, it's not our fault, our poor pilots died because this company made a bad bearing once before, and they must have done it again, okay. We don't have any proof, but it's impossible to prove, but we know that's [the cause] because we never made a mistake, not ABC helicopter company, we've never made a mistake, and it must be these people.

So the FBI, being a bunch of law enforcement officials with all their engineering expertise, says oh — so they went in, they took over the QC of this company, and they made sure that the company produced — an absolute first priority — new bearings that met the specification, okay. And until they got the fleet back up and running with brand new bearings that everyone could agree — and we're talking about tolerances that are one part in ten thousand. Not everybody can even measure those, you actually have to know how to measure that, okay. Remember old, good old Lord Kelvin, you don't have any science unless you can measure it. It's not easy to measure that, okay. They knew it met spec when they machined it, because they machined everything in the same bath of oil and stuff. Now you take it out of the bath of oil, and you have to measure it in a clean room at 72 degrees plus or minus one degree in the room, in order to make sure it meets that, so it will fit in the aluminum housing.

A lot of times failures occur at the interfaces, so we have the interface between the steel bearing made by company A and the helicopter swash plate made by company B. And if they don't fit together, whose fault is it? That's — that was your question, sort of, right, okay. Whose fault is it? Well, that's a good question, whose fault is it if they don't fit. Well, huh, yeah, specifications, right. So we were brought in, and first of all we couldn't even look at these other bearings unless the FBI would open the padlock to let us look at them. And everything anyone measured met specification at the bearing company. But it doesn't matter, the helicopter company — oh no, it had to be an improper conformity bearing, okay.

And then I'm sitting there looking at it and said, they're all barrel-shaped, I don't see any hourglasses. I mean, everybody had been looking at this, and all of a sudden I said, you know, I've been looking at it for a day and a half, and all of a sudden it just sort of hit me, and gee, if this is what I had, when the whole thing heats up during the frictional heating of the crash and everything, the heat that caused the thing to seize soaks back and pushes the balls into the race. If it's an improper conformity, I have this. But what I had was this, which is a proper conformity. So the whole hypothesis of the helicopter company, I could prove was wrong.

Unfortunately, the third day was when the new president came in, and we went to dinner. The attorneys came in and the new president, and he explained at dinner, well, this wasn't his problem, it wasn't on his watch, and he was going to settle the case. Even though I had found the key that was going to unlock this, but then we also were trying to find out what happened to this milling machine that machined the swash plate. If we'd gotten that milling machine and found that the bed of the milling machine was potato-chipped, and that's how you machine a potato-chip housing on the milling machine. But they scrapped a three-year-old machine. Everybody throws out three million dollar machines after three years. I mean, I bet you, your refrigerator, you don't keep it for more than six weeks, do you, right.

Student: [question about design]

Uh, you hope that someone designed it well in the beginning, okay. And how do you know that? It's mostly, if you — it's mostly by experience, if you want to know the truth, okay. We like to think that we do fancy computer calculations today, and we do very fancy computer calculations today, much more sophisticated than we did 25 years ago, because computers are much more sophisticated. And you can actually similarly simulate things on the computer, and design them, and play what-ifs, and come up with a good design by computer today, in the last 10 to 20 years.

But in the past, it was, we made one like this — sort of like this — 20 years before, and we're going to improve it and make it 10 times better than we did 20 years before, and we're going to increase the size, and it was always sort of incremental, you know, increases in size and stuff. And you would kind of go until you had a failure. In fact, there's a guy, Henry Petroski at Duke University, who's a civil engineer, and he wrote a book To Engineer is Human. I mean, playing off the phrase, "to err is human," to engineer is human. And his whole thesis is that we only progress in engineering by making mistakes and having failures. And then we go in, and we analyze the failure, figure out why it occurred, and then we fix it. And so that's how we progress. And that's — he goes back through hundreds of years of bridge building and explains that's how we learn to build bridges, that's how we learned to build ships. We made mistakes, okay.

We're actually starting to transition out of that in the last 20 years. I think we can now say the information age — we can now put enough into a computer program, not just the geometry and the stresses — see originally, the first CAD program was actually developed in mechanical engineering here at MIT in like 1969 or something, okay, Sketchpad or something, it was called. But it was very crude, but you could input geometry, and you could, you know, lengthen something or shrink something, you know, change — if you had a rectangle, you could make it square or longer or whatever. It was a pretty crude thing, but computers were pretty crude. And then that led to, um, computer-aided drafting, basically. I mean, you know, if you were mechanical engineering 50 years ago, you would have to take drafting courses here, okay, you know, pen and pencil, little rulers and stuff. Well, Sketchpad came along, and we could put geometry into the computer. And then in the 70s, there were programs that were finite element analysis, and so you could put stresses into that geometry. Of course we had to learn to be compatible with the CAD program and the finite element analysis.

And now in the 80s, people started saying, well, can we put properties or performance in with the stresses? Now that we know the stresses in the geometry, can we calculate fatigue cracking, or stress corrosion cracking, or, you know, performance? And so now the national labs — actually a lot of this is driven by the weapons labs, the nuclear guys, because you can't go test nuclear weapons anymore, okay. So you have to do it all in computers. So who are driving the super computers — they simulate nuclear explosions in, you know, 100 million dollar or billion dollar computer programs and computers, because you can't go and spend a billion dollars blowing up Nevada, okay. Um, it's just part of the treaties, right. And all those little sheep in Utah don't die from radiation poisoning anymore, right. You know that story, right, okay? How the winds from the nuclear test sites blew over St. George, Utah, and they had a bunch of sheep die, and some of the people in St. George said, why are sheep dying? The people at DOE are going, no. Okay, unfortunately some of the humans in St. George were also being rained on too, but anyway.

Um, anyway, uh, you know, well, we can get into other stories. Anyway, so now we actually are getting to the point where — and it's partly a really driven by the nuclear guys, because they can't do full-scale tests, they can't even do small-scale tests, because a small-scale nuclear test is still a pretty big test, okay. Uh, and we actually are getting — is that fair to say, that we're getting to the point you can design something you've never — you have no experience with — and come up with a pretty good idea of structure, of properties and performance?

Student: Yeah, my comment would be, when it comes in terms of putting it together — manufacturing, and I'm also service maintenance, right, operators — we're not there yet, we all predict all of those, right?

Um, but that's — that could be the next step for the next 10 years, where people are going to try to say, what's our manufacturing capability, you know, what tolerances do we think we can really control, and what type of QC programs will we put in. But if you actually look at the last 40 years, we go from Sketchpad in the 60s, which was geometry in the computer, to finite element in the 70s, which is stresses, to the 80s and 90s was properties and performance in the computer. But now we have to get back, as Simone says, to manufacturability, and we're not there yet. You can't simulate everything, okay. Yeah, but someday we'll have computers that are smart enough to do everything, right.

So anybody seen 2001: A Space Odyssey? They don't show that movie anymore. You have seen it. Okay, here's a trivia question for you. Why was the computer named HAL? You know? Exactly. See how many people knew that. So if we want to talk about what was the theme for today, the theme is, why was the computer named HAL? See, there's some things that really are in the noise in terms of whether you need to know them, but this is — these are the important things to me, okay, like why was the computer named HAL, okay. Not everybody knows that, okay.

In any case, um, I've got a little bit of time here, so I could start something else, but let's go back to what was I trying to do today. I actually was trying to get to the point of giving you some idea what are some of the practical limits on manufacturing, and what are some of the limitations, whether it's fingerprints, temperature, frequency of your measurement for time. And we could have gone through, you know, the ampere or other things, but I don't know anything about those. Well, that wouldn't have kept me from going through it, just because I didn't know anything about it.

But nonetheless, I really would — what I really wanted to do was get to the point that — oops, it must be under here — ships, shipbuilding is actually on one level of, is a precision manufacturing process. But we don't spend anywhere near the time or money to accomplish in shipbuilding what they do in semiconductor manufacturing and things like that. You know, a new semiconductor fab is going to cost 10 billion dollars. Who's invested 10 billion dollars in a shipyard lately? Well, there's a difference, they're going to sell how many millions of these chips and how many dozens of these ships, right.

But they are bigger. When you sell fewer of them and you're getting sort of specialized — one of the problems, one of the big problems in manufacturing is that a lot of our technology is built around military needs. In fact, I taught a freshman seminar — Mike reminded me he was a freshman in the seminar — and Professor Hosler, who teaches archeology, and I was a department head at the time, and I decided — she asked me if I would do a freshman seminar with her, and we did one on what drives technology. And my theory was, military needs have driven technology through the ages, okay, catapults and things like that.

The whole term engineering — does anyone know what the word engineer comes from? It comes from the French word ingenieur, which means maker of work — well, sorry, begins with an i though — but it means maker of war machines. And what was the first engineering school in the world? They call it École Polytechnique in Paris, uh, in the 1750 or something like that. The first one in the United States — West Point, 1797 or whatever it was, okay. And um, where did we get the term civil engineering? We got it in 1823, when they founded Rensselaer Polytechnic in Troy, New York. And they started a curriculum of study and called it civil engineering — there was only one curriculum, it was called civil engineering to distinguish it from military engineering, okay. And that's why they call it civil engineering. Why did they need civil engineers in New York in 1823? Erie Canal, exactly. See, it all makes sense, okay, if you put it in context.

Well, in any case, engineering is a relatively new profession, so far as that goes. My theory was that we have advanced technology over the ages because of security, military needs. Professor Hosler's theory — that we did it all for art and culture, that people in the Andes and, you know, South America, where she does — actually Central America, where she does her archeology — made these little bells for religious purposes. You know, they're going to kill somebody and, you know, offer a sacrifice, or something, or they need a special knife to dig out their heart, you know, and something, but they would do this for religious and cultural things.

So that was the conundrum, that was our theme for the students to try to figure out, why they did these things. And the students had to do a presentation. And I do remember what the students chose — chocolate, okay. They were going to do presentations on chocolate. This was in the late 1990s. And I remember as I'm listening to these five groups do their presentations, about the third group I said, did anyone here go to the library? Did everyone try to do all their research on the World Wide Web? And that's when, for me, that was the transition of realizing students don't even know what a library is anymore. They do everything — if you can't Google it, doesn't exist, right.

So anyway, these are the stories that — but the theme of today is, there are some practical limits to the precision with which we do manufacturing. Now in the next — any questions? See if you can get me off, and I can tell stories about — I haven't told you a story about how MIT fits in the engineering equation, or how Harvard does, but anyway, we can save that for another day.

Last time I also took you through a little bit of an introduction of chemistry. This all, this stuff of physical measurements, this has to do with physics, but chemistry — I handed out these different — a dozen or 13 or whatever different materials of half inch spheres. Your scope McMaster-Carr and order every sphere you can think of. What I didn't tell you is the tungsten carbide sphere costs about 50 bucks. That's why only one of them — I only own one of them. And the silicon nitride also costs about 50 bucks, and I only own one of those. The others, you can buy a hundred, uh, stainless steel spheres for two bucks, you know. So anybody wants some stainless steel steelies, you know, for your marble collection, I got them.

Um, but in any case, so far as chemistry, it turns out you need to have an appreciation — and this does kind of go back to the big broad MIT answers, like "large," you know, that was a good answer. It's kind of — we kind of go back to the basics, okay. But so far as material properties, and what I was asking you to do with those — that little exercise with the balls — was, what material properties do you know about to separate out these unknown materials? Well, magnetism, uh, could help you do some. You didn't know enough to take out your Bic lighter and start burning the polyethylene to, you know, and stuff. But there are many material properties.

And if I take a simple material like carbon, okay, this is the structure of graphite. And you can tell it's graphite because the atoms are black, okay. This is diamond cubic, and you can tell it's carbon because the atoms are black, okay, they're black here, they're black here. In any case, two different structures. This is hexagonal close pack, this is cubic — it's called diamond cubic, actually. And they have very different properties. Same composition, but very different properties.

And that's what this thing is for. I don't think I handed this out, but anyway, so I put this together years ago and said, okay, what are the properties of diamond and graphite? Um, turns out, what's going on here — oh, not that in the way — um, well, it gets back to the chemical bonding. The diamond is what we call sp3 hybrid bonding. It forms a tetrahedron shape, okay. And so all the atoms, every carbon atom has four nearest neighbors arranged like the sides of a tetrahedron, which is a pyramid of four equilateral triangles, okay. Graphite, on the other hand, has sp2 hybrid bonding, which is 120-degree angles in a plane. And you can see that if you look on the basal plane end-on, you can see 120-degree angles, and they're all planar, okay. Whereas this forms a three-dimensional structure.

[Diamond] is very hard, very strong bonds. Planar structure — these things slip, slide over each other. And when you get to, um, cold welding — I'll tell you about, you'll hear me tell the story about the first initial space shots and things like that, and how graphite didn't work as a lubricant. So the crystal structures, diamond cubic, hexagonal. Density is different, significantly different. Optically, diamonds are transparent to visible light, graphite's black. Why is that? Well, it turns out this structure leaves me some free electrons which absorb light. This structure leaves me no free electrons. And because of that, this one's an insulator, this one's a conductor. I mean, these are orders of magnitude between insulating properties and conducting properties. I probably got 12, 15 orders of magnitude difference in conductivity of diamond versus graphite.

Thermal properties — because it's cubic, the properties are uniform in all three directions, very uniform and isotropic. Um, it turns out that certain forms of diamond — or not diamond, but graphite — have the highest thermal conductivity in one direction. And so if I really wanted good thermal conductivity, to suck the heat out of something, if I could do it I would make single crystal graphite. But that's not an easy thing to do. But you can have, uh, anisotropic properties, where in one direction the graphite has about a hundred times the thermal conductivity of any other material, okay.

So nowadays people are getting very excited because they've discovered — the physicists have discovered graphene. Anybody know what graphene is? A single layer sheet of carbon atoms, okay. You heard of Buckyballs, okay? Buckyballs is a little hexagonal 60-cycle — 60-atom soccer ball of carbon atoms that have basically sp3 hybrid type bonding. And then they have Bucky tubes, carbon nanotubes, which are little cylinders, cylindrical tubes. And now some guys just won the Nobel Prize a year or so ago because they have discovered graphene, and they found a way to make single layer sheets of this material.

Why are they so interested in it? Because the carbon-carbon bond is the second strongest bond we know. The only stronger bond is silicon-oxygen, okay. So very strong, and that's why it's very hard. And these graphene sheets, oh, they're predicting all kinds of things. They predict that the strength of these carbon nanotubes is 2 million PSI. Well, when you watch some of the other videos, I'll go through and explain that we had iron whiskers with 2 million PSI 60 years ago. The physicists don't know about it though, don't tell them.

But the thermal properties are dramatically different. Mechanical wear — this is an abrasive, this is a lubricant most of the time. Hardness — extremely hard, diamond is the hardest material because it's got the strongest bonds in a three-dimensional network. And this is relatively soft. The abundance — diamond is rare, carbon is abundant, that's all the coal we've burned. And the cost, very high — anywhere from low to high depending on how uniform the properties. And I was thinking about machinability for some other reasons this morning, and so I added machinability, which you can't read — see if you can read it there, okay. You can kind of read at the top, diamond is very difficult to machine, can be done with diamond powder, you can abrade it away. And graphite's very easy to machine.

And you can think of all kinds of properties and say, you know what, um, what properties do I have? It turns out it's not just the atom, it's really the bonding configuration of that atom. And so the whole field of materials science basically works with understanding the bonding between the atoms and what properties result from that. And the physics folks have to worry about, in a manufacturing sense, how do you put those things together? If you're going to build a ship, you have to worry about all these things.

So that's probably enough for today. You'll get safety factors tomorrow, okay. And I'll see you Thursday. Thank you.