§1. Friction stir welding and the history of welding processes [00:01]
Student: Did they use friction stir welding and that sort of technique back then?
No, friction stir welding had not been developed. Actually it had been developed as a research thing about 1975 at the British Welding Institute. It was still sort of a curiosity. For aluminum it wasn't until about 1990 that Boeing built some parts of the space shuttle with it — not the main space shuttle but some of the external tanks or something. Boeing built some aerospace components with friction stir welding of aluminum. People have tried steel. The problem with steel for friction stir welding is you just don't have a stirrer rod with high temperature strength. There's really no good material that will go to the types of temperatures you need without just wearing away very rapidly. They have done it with titanium, but titanium is very reactive with a lot of metals and you can't tolerate the impurities that you get in there. Aluminum is the sweet spot for friction stir welding.
Aluminum melts at a low temperature, so you can just use regular old steel or stainless steel or whatever you want and not have a big problem, and aluminum doesn't react a lot with iron. But when you get to higher temperatures of titanium and steel, friction stir welding doesn't really work very well. Plus it's very expensive for the tooling because you have to have big rigid tools.
I did a paper once, a keynote lecture at a welding conference. I looked in the back of the welding handbook — there's a list of 199 different welding and cutting processes. That includes soldering and brazing and plasma cutting and oxy-acetylene cutting, but lots of arc welding processes. They list 199. I looked at how many were developed between 1900 and 1950, and it was seventy-five percent of them. Between 1950 and 1975 it was 24.5 percent of them. And there was one new welding process that had been developed in the previous twenty-five years between 1975 and 2000 when I was given this talk, and that was friction stir.
So welding — there was forge welding and brazing and soldering before 1900, and there was arc welding before 1900, but the age of widespread welding fabrication really didn't start until around 1900. In that first century you had just an explosion of welding processes. In those first fifty years gas tungsten arc was developed in France in the 1930s or so. Gas metal arc was invented in the United States in the late 40s, early 50s. Submerged arc in the United States in 1936. Electroslag about the same time in the Soviet Union. Most of them were developed in the first half century of the 20th century. Then it started to taper off a little bit. Not too much, but a little bit. Gas metal arc was invented basically in the second half of the 20th century, and then there just haven't been a whole lot more welding processes. When you take the welding course I'll spend a little bit of time talking about some of that.
§2. Titanium microstructure and the moisture problem [03:56]
As far as titanium, the microstructures you get — this is the furnace cooling structure of plate-like alpha (these plates) and intergranular beta, which is the dark phase. This actually looks sort of like a Widmanstätten steel. It's actually called Widmanstätten. Widmanstätten is a plate-like structure. So titanium and the base material look sort of like a steel weld metal in some cases.
If you go to air cooling of the same thing, you'll actually get some very fine acicular structure. In titanium we call it a basket weave. This becomes very fine — it becomes basket weave with plates going in a bunch of different directions. So sometimes the weld metal in titanium has better mechanical properties because of the finer structure. It's not too hard to get a good microstructure in titanium alloys if you adjust your alloy content properly.
This is moisture content in the argon shielding gas if you're welding titanium, either gas tungsten arc or gas metal arc. You look at the minimum bend radius — this is titanium six-four — and you can see that if you don't have a very good dew point, like minus 20 or minus 30. Typically if you buy a tank of pure argon, it might be around minus 20 or minus 30. You could order it to minus 60, but they have to do special precautions when they fill the tank from the liquid argon. If you start getting very much oxygen in there, the hardness is going to go way up by a factor of three. The minimum bend radius goes up, which means the minimum you can tolerate is larger. It makes it easier to crack. And the hydrogen content increases.
What's the problem with hydrogen? It's not a big problem in titanium. With aluminum you get porosity, and in titanium you get similar types of spherical porosity. With titanium, you have to keep the oxygen and nitrogen out to begin with, and if you keep the oxygen and nitrogen out you'll usually keep the hydrogen out. Here obviously the hydrogen and oxygen are coming from the moisture, the dew point, which is the moisture in the shielding gas.
§3. Shielding requirements and acceptable weld colors [06:50]
One of the big problems and one of the big cost drivers for welding titanium — and we do weld titanium; the aerospace industry welds in thin sections all the time — is the fact that you have to have excellent shielding. You can't tolerate just a regular gas tungsten arc welding torch. You actually have to build a box around the whole thing with sliding seals. If you're going to weld the plate down here, you may end up welding a box around this whole thing. This is a heavy section weld, sort of like this deep TIG I talked about in titanium. You can't just weld on the top, you've got to weld on the bottom, you've got to put a backing bar on the bottom, and you've got to have backing gas on this side. For corner fillet welds you have to have big complex shielding dams for your welding.
Here's an example of someone just trying to make a simple gas tungsten arc weld. Here's your gas tungsten arc electrode, the welder's glove, here's the electrode right here, there's the arc. This little thing here is basically a manifold blowing argon gas on top of the weld. So you've got one up here, you've got one down there, you've got the shielding gas coming in at the top. But you can't just let the stuff solidify and cool down and expect your torch gas to protect the hot weld behind you. If you do, you'll end up with all kinds of contamination. By the way, you also have to cut back your electrode after you finish welding. You've got to cut it back a lot further — this thing's contaminated all the way back. You have to cut a half an inch off, not an eighth of an inch off, your electrode.
They have acceptability of colors, which is what you'll find in the welding handbook — if you Google titanium welding. They have bright silver — you can see the ones I passed around, nice silvery weld metal with proper argon shielding. Silver, light straw, violet, blue, and white. This is titanium oxide — you just get a powder on the surface. They say all the way up through brown is acceptable. There are welding standards that basically show you silver, straw, brown as acceptable. Blue is not acceptable. Here you can see the colors on the side. A lot of those colors — it's shiny here, they call it green-blue, but it's shiny here because the thing was so hot it dissolved away the oxygen, and nitrogen is actually contaminated.
The US Navy's standard, which they've had for forty years, basically says it's got to be silver everywhere to be an acceptable weld. They wouldn't accept something like this. American Welding Society doesn't accept something like this either. You might say it's just as silvery as that in the weld metal, maybe even more so, but you see this contamination on the side. Well, if you've got contamination on the side, you've got it in the center too. So the Navy doesn't accept colors for acceptability of titanium welds. Here's another thing from some welding spec — that's an acceptable weld, that's an unacceptable weld; this is the white thing, you're starting to form a cloud of titanium dioxide snow that's coming down.
Cleanliness of your shielding gas is critical. Here's a situation where the welder is not in the glove box, but they put the part into a glove bag. This is what we used when I was nineteen years old at the Naval Air Rework Facility in Norfolk, Virginia. They had bags like this, and the welder would weld titanium parts for jet engines inside a pure argon atmosphere.
§4. Semiconductor piping, hose permeability, and the Alpha submarine shipyard problem [11:23]
People also see these same types of colors not just on titanium but on stainless steels. About ten or fifteen years ago I used to get people coming through on a regular basis welding stainless steel piping for high-purity semiconductor crystal growth piping. For the gases — if you're going to make gallium arsenide or something, you actually use arsine, which is arsenic hydride gas. IBM would purchase some gas system of stainless steel piping and they might pay $15,000 for this little gas train, which is not very big. They would just cannibalize it and cut it open if they saw anything like those colors on the side on the stainless steel. They would reject everything you built, because they didn't want the contamination to get into their semiconductor. In their opinion it would destroy their process. Well, some of that contamination is oxide, but some of it is not. Some of it is just stainless steel vapor deposited on the side, and you get what we call interference colors, or Newton's rings — just different layers of vapor-deposited material.
When you're welding, if you want to be ultra clean — you don't have to be this ultra clean for stainless steel in general, but you do for titanium. I got this out of some welding metallurgy book recently. Different types of rubber hoses have different permeability for moisture, and if you just use natural rubber or neoprene or PVC, you're liable to get enough permeation of moisture right through your hose. You could have six nines of argon, which is 99.9999 percent argon, which is a dew point about minus 60 or minus 100, plenty good for welding. That's what's coming out of the tank — they filled that argon tank at a gas supply house from a liquid argon source, and there's no liquid oxygen in there, except what they might have gotten from things that weren't particularly clean in their piping.
The other thing — if you just use regular old gas piping that we use for stainless steel and nickel and all these other alloys, you're going to get bad welds in titanium. You have to use metal piping for your gases. You could use some Teflon and some of these things are pretty good, but really you need to be as absolutely clean as you can. Your gas shielding for your welding process has to be brought in as if you were in a semiconductor manufacturing plant. That was one of the things back in the days of the Alpha sub, when we were talking down at David Taylor — you weren't going to build that in the shipyards you have today. No one's going to build a full-scale titanium submarine in that type of shipyard. They're going to have to build it in something that looks more like an aerospace facility in terms of cleanliness. You've seen pictures of Boeing — that doesn't look like the shipyard does, does it. If you're going to do a lot of titanium welding, that's what you're going to have to do. It's going to have to be a facility that you can sort of eat off the floor whether you want to or not.
§5. Engelhard, the 747-400 catalytic converter, and reagent-grade acetone [15:07]
Titanium is a different animal in terms of the cleanliness that's required to get good properties. You do have to clean it well. I guess I'll tell a Boeing story. Actually it's not a Boeing story, it's an Engelhard story. I'd just come back from Japan, maybe it was late 80s, and I get a phone call. They were about to roll out the 747-400. So if we went back and found out when the 747-400 — the dash series, this dash 100, 200; I think we're up to 747-800 now. They improve them every now and then. They were going to come up with a longer-distance 747.
Boeing had a spec that all the welds had to be radiographed on the titanium, and you could not have a flaw smaller than ten thousandths of an inch. There was no good reason for this, because what we were talking about was the air ducting to bring in the air. The air outside has got lots of ozone in it because you're at very high altitudes. But you couldn't let people just breathe the air coming in from the outside without getting rid of the ozone. So they put it through a catalytic converter. Engelhard built the catalytic converters. You had thin-walled — it was like a millimeter thick titanium tubing. They'd made it out of stainless steel before, but this might be a four-inch diameter, two-millimeter-thick — it's not very thick. Then they would have a catalytic converter in the line which just looked like an anaconda that had just eaten a pig — a big bulbous thing for the catalytic converter. When you're flying on an aircraft you're breathing air that's been through a catalytic converter to get rid of the ozone, because if you're up there for a number of hours, you get a headache from as much ozone as there is at higher levels.
Engelhard had taken the contract to make the catalytic converters, to weld the titanium piping. They had been welding stainless steel piping for some time, but in order to get the longer range they were cutting weight everywhere, and they were going to titanium piping for the inlet ducting. Even if you had a crack in this, what's the big deal? You'd be leaking a little air through a little hole. You've already got something four inches in diameter — there's going to be plenty of air for people to breathe. But nonetheless, Boeing had a spec written that you could not have on an x-ray a flaw larger than a quarter millimeter, which is ten thousandths of an inch. From a fracture mechanic's point of view it makes no sense whatsoever. From a practical point of view, who cares? I could have used gasketed joints that leaked. There's no environmental concern here — it just has a little less ozone in it. As long as I still get enough in the passenger compartment for them to breathe — and you would.
They were welding this stuff, and the anaconda where the catalytic converter was looked like this. It was a larger diameter because you had to have constant flow through here. So you had a catalytic converter which restricted flow, a four-inch pipe, and they had to make four circumferential welds to make this. They were finding they could make these gas tungsten arc welds, but they would fail one out of two welds. So on average, if you made four welds, you'd do x-rays and you'd have two bad ones. If you had two bad ones you could repair them once, according to Boeing spec. You only get one chance of repair. What are the odds of failing the next two? Pretty good, right. They had not been able to produce a single catalytic converter. This was going to hold up the whole aircraft for this lousy little spec.
I'd just got back from two weeks in Japan, and they called up and said, can you come down and help us. I said, well, how are you cleaning your titanium? They told me, and I said, they're cleaning in acetone or something to degrease it. I said, look, if you're getting a little porosity — they had little ten-thousandths or fifteen-thousandths pores, and this is an eighth of an inch thick — if you're getting hydrogen in titanium, a fingerprint can do this. Enough grease in a fingerprint to end up with pores like that in titanium. They said, well, can you come down? I said, I can't come down this next week. Get some reagent grade acetone — don't just use regular commercial grade acetone as your degreaser, or trichloroethylene or whatever they were using. You've got to get some reagent grade and degrease it with the reagent grade degreaser.
They kept on begging me to come down. It turns out I had to be in New Jersey for something else, and I said, I'll come by, but I can't get by until six o'clock in the evening, then I've got to get out that night. So I went by to where they were doing this and they'd kept people around that evening to talk to me. I told them the same thing I'd said before — you've got to use your reagent grade degreaser. I walked them through their process and said this is what you've got to do. And once you do it, people have to be wearing white gloves. You can't handle this with your fingers after you've cleaned it. Everybody's got to be wearing white gloves. It's like a semiconductor manufacturing facility.
They tried it, and a week later they were delighted. They had a ninety percent success rate on their welds. At ninety percent they actually could get some of these through the door, so they were happy. About a month later they called me up and said, it's come back. I said, well, how are you cleaning it? Well, we're doing exactly what you said. So I went down there again. Guess what — they weren't doing what we told them. They had slipped back to the old process. So I told them again how to clean it. That was the last I heard from them. The 747 so far as I know is flying. At one point I said, well, why don't you consider asking Boeing to come up with a reasonable spec? Oh, we could never get that through. I had to believe that was probably true. Getting a reasonable specification out of the designer was probably a more difficult task than just figuring out a way to solve the problem.
§6. GE Lynn spot welding adaptive control [22:28]
Have I told you the titanium electron beam welding at General Electric? No, okay. This is a similar type of problem. At the time, this was mid 80s, General Electric Lynn up here. They sort of abused me, actually. General Electric Lynn Aircraft Engines had wanted to improve the spot welding. This is where you use two big copper electrodes and you put two pieces of sheet metal between the two electrodes, and you pass 10,000 amps for a third of a second, and you can weld a little button between those two electrodes. It's called a resistance spot weld, invented by Elihu Thomson, who had been president of MIT. He was the co-founder of General Electric with Thomas Edison back around 1900. It's the way they weld all kinds of automobiles. I'm known for having said you need 3,000 spot welds in the average automobile because you need 2,000 good ones. When it's working well it's a very reliable, very fast process. It costs about a nickel to ten cents for every weld that goes in an automobile, because you're just bang — you're making 60 welds a minute on an assembly line, which is incredible. But that's why they get a thousand bad ones out of three thousand.
They were trying to weld nickel-based superalloys for the combustor cans of the jet engines, and you had to have a good weld. They had contracted with General Electric Schenectady Research to help them come up with something called spot welding adaptive control. They'd been giving them a million dollars a year — this is 1980 money, a lot of money. They didn't tell me any of this. They told me they had contracted and they wanted me to take a contract. Now I was heading off to Tokyo, Japan to work for the US Navy Office of Naval Research about four months later. I go up with one of my postdocs who was going to manage my lab while I was gone, we go up and we listened to them. They said we'll give you $150,000 to work on this for the next year. I said, well, that was pretty good money in the mid 80s for a project. I didn't know they were giving General Electric Research a million dollars a year. So I was sort of small potatoes here to help them with the project.
I go off to Japan, and Gordon and Carl, my students, are working on this, and I get back from Japan thirteen months later. It turns out that Gordon and Carl had figured out — they had sent me notes while I was over in Japan — that General Electric Research would not tell us anything about the system. We found out very quickly that General Electric Lynn had cut the million dollars a year of funding from General Electric Research for having spent seven million dollars and not having a product yet. They hired us as the whipping boys to punish General Electric Research. So Gordon and Carl had to go into this seven-million-dollar system and figure out the software on their own. They actually didn't even have access to it. All they could do was run experiments. They ran some experiments and did some tests, and they concluded that — even though this is 60-cycle current and you're supposed to control it in four cycles of current, one fifteenth of a second — the adaptive controller wasn't even kicking in until like six cycles after the weld was over. It was trying to adapt the weld.
I get back, and Carl and Gordon tell me about this, and the engineer at General Electric said the big boss, who was a graduate of mechanical engineering at MIT, had wanted a presentation. I was freshly back from a year in Tokyo. So I said, we can, but their adaptive controller is not adapting, and you've spent seven million dollars for a piece of junk. Oh well, come out anyway and explain this. So I go out — we schedule for two weeks later, I go out, and it turns out that they called me up the week before and said, oh by the way, General Electric Research, we decided to fund them again, and they're going to be here at your presentation where I'm going to trash General Electric Research.
I get up to show the data in front of the big boss at GE Lynn, and I said, see, the weld only lasts four cycles. Your adaptive controller doesn't kick in and start responding until six cycles later, and therefore it's not adapting. The guy from General Electric Research said, "we fixed that last week." He didn't deny that it was a problem. Anyway, they gave me another year's funding. They brought me in anyway.
§7. GE Lynn titanium electron beam welding and the acetone barrel [28:12]
While I'm doing this next year's funding, I get a call from the engineer, and he says, Tom, we have some electron beam welds on titanium that we're making, and we'd like you to help us with it. We've been making these welds for ten years on this engine, and we always used to pass, but now we're flunking, we're not passing the x-ray. I said, well, Lyle, what are you using to clean this? He says, I don't know. I said, well, I bet they're using acetone. The answer is always acetone, right. I said, I bet you it's just commercial grade. Go get some reagent grade acetone and clean the welds with acetone before you put them in the vacuum system for the electron beam welds.
He calls me up two weeks later, he says, Tom, we still have this problem and we really need you to come out. I said, did you go get the acetone? Well, no, I haven't yet. Okay, well, could you come out? Okay, I'll come. And by the way — this is an old engine project, we don't have any money to charge people's time. So you don't want to pay me. He says, well, we figured we're giving you this other resistance welding contract and you do it as part of that, and that's goodwill. Okay, I'll come out. Oh, and can you get here at seven o'clock in the morning and we'll walk you through the whole cleaning process that we use.
I got there at seven o'clock in the morning. They took me over to the cleaning shop — this place where you don't want to breathe the air; you can just feel it cleaning your nostrils. We finally get there about ten o'clock in the morning, and they've got these two titanium rings that are both probably worth $25,000 a piece. They're going to put them together and make a $70,000 part out of two twenty-five-thousand-dollar parts if they can get a good weld. While we're there, I counted seventeen General Electric engineers and other people who came to watch this weld. This was getting to be very high profile. I thought, well, they can't afford to pay me as a consultant, but I know these other seventeen people are charging their time to some account number, right.
I'm sort of the star because I'm the unpaid consultant. The technician goes through his process — he takes a white rag, he's got white gloves on, he takes some acetone out of a little squeeze bottle, squeezes it on, and wipes it. I get down with my 10x magnifier, and I'm looking at the surface to see how clean it is. I look at another area, and I see there's some white specks that were not there when I looked at it a few seconds before. I realized the room is raining dust. This is not exactly a clean room, this is a welding shop. I point this out, and all these other seventeen people start making excuses. The technician says, well, you know, I would have put the two parts together before, but you delayed me, and so they all start paying attention to that.
He puts the two parts together, and in the meantime everybody's ignoring me, so I go over to this bench and I see a glass slide. I clean it off with my handkerchief as well as I can, look at it — it's nice and clean, just like cleaning your eyeglasses. I take some of that acetone, squeeze it on, blow it dry, and I look at it and you can see Newton's rings, a film of grease. I said, Lyle, where are they getting their acetone? He says, oh, some 55-gallon drum out back. That's eleven o'clock; now we're waiting for it to pump down, we go off to lunch. We come back, they have a leak in the vacuum system, probably won't get the weld made until later in the afternoon. So I'm leaving. I said, I'm not going to wait while they fix the leak on their vacuum system. Let me know what happens. In the meantime, will you go get some reagent grade acetone? So we say goodbye. Turns out it's a bad weld, just like all the others.
About six weeks later Lyle calls me up and he says, Tom, I got the reagent grade acetone you asked for, and we've made four of these since then, and not a single one of them has had a defect. I said, well, that's good. He says, but the problem is all I got was a letter of commendation for having saved a quarter million dollars a year. I said, Lyle, I didn't even get a letter.
There is a moral to that story. I know it's a little bit of a long story, but the moral is, this is the way industry works. You give them advice, whether it's Engelhard or General Electric, you give them an answer, they ignore you, they keep telling you they have the problem, they keep having the problem, and they keep ignoring you until you beat them over the head, get them to do what — and maybe you're wrong, but maybe you're right. Why are they asking you to consult if they — anyway. If you can get them to actually do it, it might work. There's no chance of it working if they just keep doing the same thing they've always done, which is what they do.
§8. Alcoa Davenport plate thickness and the water-pooling discovery [33:41]
I guess we have time, I can tell you one more story that goes along with that. In 1989 or 90 or so, I had a student in our manufacturing program. She was getting a master's degree in what you'd call Leaders for Global Operations around here now — it was Leaders for Manufacturing back then. She was working in the Davenport, Iowa works for Alcoa. This is where they roll the great big heavy plates — I saw the stretcher leveler for the four-inch or six-inch thick wing plates and stuff. What Julie was given for her master's thesis was to figure out why their biggest — if you did a Pareto on their plate, the thickness, the variations in thickness — they rejected more plate for being off thickness than any other reason. These are big heavy plates; this is the world's largest rolling mill.
Julie and I tour the plant, and she comes back midway through her seven-month internship. I told her to go out and get an ultrasonic thickness gauge and plot the thickness of these plates. It turns out the thickness variation of the plates was typically thick on one end or the other. They measured the thickness by a guy bringing up a little wooden step stool with a micrometer, getting up there on the hot mill, and measuring the thickness with a micrometer while it's still hot. Now if you've got a thick plate and you're using a micrometer, are you taking into account anything about thermal expansion? What temperature are you measuring this at? And it's at one spot.
Julie does a survey across the plate along the whole length — this might be 40 feet long, and she's got a whole pattern. You can see — she was from Denver; I used to tease her that it looked like the Rocky Mountains, looked like Colorado. Nice and flat, and you get to Denver and you get the Rocky Mountains. You get here and you get the mountain on the end. So it depends on where they measured when they got the thickness of the plate. Then I remembered that when we were out there, they have water cooling on this great big steel rolling mill. The mill is almost as wide as this room. The rolls are not as tall as this room but almost — probably ten feet tall and probably twenty feet wide or something, world's largest rolling mill. They pour water on it, and the water comes down this hot aluminum plate. Well, hot aluminum plate is nice and thick, and it would just come off as steam while they're rolling, but at the end I remember I would see water pooled.
This variation here is only like a millimeter or two variation out of one inch, so it might be seven percent variation, which is off gauge, and they might scrap the plate or send it to some other application. But I remembered that I would see water boiling right on the end of the plate. Some of the water would pool there and you'd see it boiling — like if you ever threw liquid nitrogen on the floor, you'd see a little gas layer of nitrogen, and the liquid would be floating on top of this gas layer as it's cooling everything down. So Julie's showing me her data, and I said, I'll bet you there's a reason why that water's pooling. Oh yeah — they run the top roll a little slower than the bottom roll.