



S1: in uh, your interest i'd like to mention that we are in fact uh uh, quite a bit ahead of our schedule, uh that means one or the other either you could talk more, and uh slow us down, <SS LAUGH>or or by your silence you might go home a little earlier tomorrow and i'm not offering anything particularly.<SS LAUGH> um, sometime before you go though, here are the um, certificates. uh some companies like to know that you actually were here <SS LAUGH> and uh so there's an_ there's one for everybody, and uh uh i signed them as uh the verification indeed that you were good boys and all that sort of thing so, there it is pick one, it's not quite like you know we have uh you know up the road there's a place called Michigan State University, and uh every now and then around campus here you find uh, you know rolls of toilet paper you know a little sign above uh M-S-U, diplomas take one. <SS LAUGH> 
S2: you you mean you're gonna pass those out before Friday? <SS LAUGH>
S1: yeah, yeah.
SU-M: (i see)
S1: well one of the things i um uh, did not cover or uh zipped by on Monday morning, is the matter of the literature. uh, i postponed that uh, to the point, uh where um, i i've heard questions uh now from quite a number of people, where do you find information on this or that? i've got a product, that's not working so well, how do i know, uh what materials work better than the ones we've got right now. well, um, we don't know. and uh, there again you kinda wonder if uh you know if uh if if i've really been uh diligent in these forty years, why don't i know? uh it's because the subject i say isn't so complicated, but, you know, when you're looking for a w- uh uh a solution to a problem and you're looking perhaps for a different material, or a different lubricant, um, you you you expect that there may be one out there now, can you trace, can you go back in your mind as to why you expect, that there's a solution laying out there somewhere. hm...? well, isn't it because, uh, very often, we have things that break, and the answer to things that break, is to find a stronger material. simple, hm a good connection there? or, we've got a material that, um uh that fatigues, or buy a material that doesn't fatigue as as as readily you know that sort of thing. uh or you might get into certain nuances you know relating to residual stresses so forth and so forth, you know that, the majority of us, are are are well steeped in mechanics, aren't we? and in mechanics or the mechanics ethic, leads us to to suppose, that uh there are materials out there that satisfy, a need, but now, <BACKGROUND NOISE> how do we, measure, how well a need can be satisfied by a different material, well we've got equations huh...? <DOOR CLOSED> and uh, so perhaps you hope thereby that in wear, there is a direct equivalent. that there are equations on wear. now you asked earlier about what material does which to. uh, i haven't the foggiest idea. because i don't know whether your present product, is tweaking, a fatigue mechanism, or an oxidation mechanism you see, or a corrosion mechanism, or what, uh um, what the present problem is and so, i have no idea which way to turn. and uh, that's disappointing, that's discouraging, that's disconcerting, uh, disconcerting f- to uh uh for managers, you know, uh, managers have um, i think well an urgency to get something done, and uh there comes this wear crowd they never have the answers you know, or they're likely to say well give us two hundred thousand dollars we'll research it for you, and uh, uh we you know we'll we'll call on you don't call on us, in three years and uh so on. uh, along that line, uh, at this Gordon conference i was just uh uh uh mentioned i mentioned a couple times, the Gordon conference out in New Hampshire, uh, the first paper was delivered by uh, a uh a man, uh a savvy designer, who works for a diesel engine manufacturer in Columbus Indiana and i won't reveal the name of the company, well Cumm- you know Cummins. <SS LAUGH> and uh, um, he, wrote of his frustrations as a designer of diesel engines in a successful industry, and among other things, um, he had to say or he did say, that he doesn't know which way to turn. uh, he can't of course go down the street to Peoria and ask certain questions of those fine folks down there <LAUGH> or the people of generous motors or any of the other manufacturers uh, of uh diesel engines, for good reason, then where do they go? he says in this in this talk, he doesn't trust universities, uh because they uh you know, just don't deliver, uh he doesn't trust the literature, they don't deliver, or the literature the articles just don't describe anything close to his problem. and uh this is uh very true i must say that at the at that Gordon conference about a hundred and ten people attended, thirty-three were from industry, uh the rest were from, uh universities and government... there was quite a stir, after this paper, one, professor of another university said, to me, i've never been so insulted in all my life. <SS LAUGH> well, now why was he insulted? because uh this man at Cummins, uh did not, state his problem to that professor. well it is a bit of an arrogant sort of attitude you know, and i mean the idea there then if you tell me your problem i'll be able to solve it, hm? and um, i'm, one to tell you that uh that doesn't work. uh, i cannot solve problems immediately, at all, really, um, your problems are_ have already been picked over and all the easy solutions have been done, right? and what there's left is uh uh several paths, several alternatives. now, how do we proceed down, uh to pick an alternative? well that requires a lot of discussion. now, you can call in a consultant, but that consultant, doesn't know the problem. so you've got to educate the consultant, and that was, the Cummins' man complaint. i first educate the consultant, then he goes off and, helps Caterpillar make better products. <SS LAUGH> uh okay, kay
S3: what's wrong with that? <SS LAUGH>
SU-M: from his point of view yeah.<SS LAUGH>
S1: uh, and uh, so you know one aga- one uh uh again um, uh expects a relationship with a consultant just like you would with uh with uh might i say simpler, problems? you know if you've got a corrosion problem, uh, a corrosion consultant can point you, the way fairly quickly. and likewise in the metallurgy problem but w- wear, uh y- y- it's very hard to point the way, and you've gotta work a while, and where i have succeeded uh in in consulting, and uh where as you see it uh Vern Wedeven has succeeded, it's when you're taken inside of the problem for a while, now my inclination generally, is i'm inside the problem for a day or two, then i say, you know, there's a guy at Penn State, that knows this topic better than i. why don't you call him. and then i can help describe the problem and so forth so on, and expand uh you know the, kind of thi- the, the the database, uh, or, uh let's do some tests. but not at the university. the tests at the university will be on a pin-on-disc machine, which doesn't tell you, very much you know? let's do some tests in the company, with the people, in the company, and when that is done the intelligence is is it is embedded in the people in the company where it belongs, it doesn't get farmed out to the university, the interface between universities and external research groups and the sponsoring industry, is uh uh well it's a it's a rather high resistance interface. it's hard to transmit information from outside to inside, of a problem that has so many, uh almost, um, uh subjective nuances, that people in the company, have got to be deeply involved in the in the in the uh investigation. well alright that doesn't mean you, don't, look in the literature, now, early on or in the in the first part of the book, there's a mention of, the literature, there are many journals, and i get most of them, uh, and there are many books and i've got uh probably nine-tenths of the available books, of substance now there's a little su- a little you know, uh, uh subjective sort of evaluation, but uh, what are the what are these books well i i you know i can mention here, Elsevier, has a tribology series and here's a very gripping title, Materials, for Research, written by a very competent guy Bill Glaeser at Battelle, now, Battelle, or Glaeser is a, uh, metallurgist, and he describes all the metallurgical types of failure that happen, and uh, this is a helpful thing to read. it's a lengthy thing to read, whether you can derive something in one reading, uh, i- i- is uh you know something for you to judge i wanna pass these around, um, there are, uh, as i say many books there's this tribology series there's about thirty of them by now, to look into, uh there are, the professional society journals, the A-S-L-E now called Society of Tribologists and Lubrication Engineers, p- turns out the Tribology Transactions. here's a journal of, very high quality, that is to say they, are quite careful, to accept only, good papers... they rejected mine three years ago, <SS LAUGH> and uh, uh, they're we- very carefully edited, and uh very well done. uh many of them are uh you know, rather esoteric and very, um uh narrow in scope, but, now and then you can uh, uh find something of use to you in that journal. the A-S-M-E, turns out the Journal of Tribology, another, very high quality journal, uh, really top in the field in hydrodynamics elastohydrodynamics and that sort of thing. Tribology Transactions will have more articles by the uh the chemical industry the auto- the uh additive people, uh and grease people and so on solid lubricants, uh journal of tribology is mostly, hydrodynamics elastohydrodynamics, uh so-called limits of lubrication, so-called boundary lubrication and so forth, physical, boundary lubrication not so much chemical, and both articles both will have uh articles a smattering of articles in wear of materials as well, then there's the journal called Wear, which is uh, very much oriented to wear, W-E-A-R, and uh, this is an Elsevier journal, fewer people get the Elsevier, uh journals, for one fairly good reason, you know you can, get the Journal of Tribology, if you're a member of A-S-M-E for a hundred and ten bucks. hm? Elsevier journals cost four thousand bucks... <GASP SS> a year. now you get twice as much paper, <SS LAUGH> uh, but uh, and and they're good articles, they're, you know there's a there's a there are, more, uh articles for your four thousand but, uh, that's very expensive. virtually, all libraries around the country, you know are are trying to save money, and the first things they go after are these Elsevier journals, cuz all of them are expensive. but at the same time Elsevier, is just growing l- by leaps and bounds and absorbing, many many other independent journals. before too long they'll have a, you know corner on the, on the commercial market, uh, and all the prices are gonna go up. <SS LAUGH> the strange thing is that uh, i'm involved with that journal, and uh that society that group, the uh upper administration of A-S-M-E just simply will not hear any proposals to raise the price of that journal. and i think they can go a long ways but uh, mm, so it goes. anyway the uh, uh Wear journal, i, got two of them, i think they're different ones, yeah they're different ones, and uh, this is edited by Duncan Dawson, the author of that book, you saw the other day on the history of tribology, uh Duncan has just retired, as editor there, and Ian Hutchings has now uh, uh taken his place. well, um, how does one, pick through those articles and uh, uh, know whether there's something valuable, it just takes time. um, i suppose, a quarter of my time, uh is spent, going through papers and journals. you know? uh, and i don't think that's enough, to stay on top of the field i- i- in the manner i want to stay on top of the field you know? uh, uh, i, i i i think that's far too much time for most everybody that's gotta work on real problems, so, well uh so we're um, moving on now to chapter nine, a topic called chemical, boundary lubrication. now... this is a uh uh an an appropriate time for that, uh, being, directly adjacent to, uh the presentation of Doctor Wedeven, um, Vern, i'll use the um, uh informal term, um, started out in the fluid mechanics, side, of the house, and um, uh bit by bit, he had uh moved toward, uh, uh thinner thinner and thinner films, uh in uh the lubrication problem, uh down to elasto, hydro, dynamics and then, micro, elasto, hydro, dynamics, and now they talk about nano, elasto, hydro, dynamics, and next will be tera, elasto, hydro, dynamics, and so on. uh that's the natural progression of things, um, but i he he says, to me anyway did he say this to you? uh that his techniques have run about to the limit of uh, of uh what shall we say uh, uh, of um treatment of hydrodynamics or fluid films as a load carrying element? did he say it qui- quite that way, at all? well, uh we described the problem uh this way, that, um, you know, hydrodynamic lubrication is the case where, uh there's a very thick film between two surfaces, and uh we take the, fluid film thickness as being, something here, a value called H, we take the, surface roughness as a value called sigma, and uh we take the ratio, of these, as um, uh equal to lambda, and that's the magic number, now uh, uh, formally we had the luxury of expecting that lambda would be greater than one. and for that there is an adequate mathematics, in simple fluid mechanics, then, the next thing we get to is, um, elastohydrodynamics, and uh uh the um, what what's flipped in along with elastohydrodynamics, uh is the um, uh, uh the g- the uh the fact, really, that in so very many cases of sliding, lambda is less than one. and uh maybe more like say, point-one, and, often smaller. well now that says something about, how each individual asperity will collide with somebody else, uh they feel each other's presence, whereas these do not. and so, the study of hydrodynamics is getting on toward, uh considering what happens on a microscale, and here you've got to admit, that um, when you've got asperities colliding there goes an asperity that way and here stands one, and um at some point this one has to glide up over that one that one has to scrunch down, uh they are separated by, a very, thin film, and if we look at it from the side you see we can envision that, but, uh even this treatment you know has its difficulties, this contact pressure is high, and so we've got to consider the, pressure- viscosity index, that is to say how does that oil, uh thicken up or harden up uh with contact pressure, and uh if it becomes, hard uh rigid as uh Vern mentioned, how does it shear, there's uh plenty to study in that area, but uh now, uh uh the other side or the other half of the picture, is that where we've got, large or a contact nominal contact area, um, they uh, uh microelastohydrodynamics, happens in, isolated little patches, and they're highly transient. at one instant, they're there, and an instant later, we've got those uh microcontact areas in different locations, and that just keeps changing with time, now, how do you calculate what happens, in these isolated regions, uh, when you, uh, you know have a large surface like this, what is the meaning, of, uh well, uh you see downstairs in the elastohydrodynamic equation, is W, over L, E, R, to the point-one-three power, well W and L, um, uh uh well if you got, two rollers, rolling together, you've got implied load, and you've got a, length dimension L like this, and from, contact mechanics you can work out this dimension which is two-A, we have an average, area or average contact pressure, through those numbers, that's not where the action is. the real action is, on these individual asperities, um, which you have to, calculate by a different way. so the, equations of elastohydrodynamics, are moving over to, a very different form than we had seen, to accommodate the situation where we've got, uh, uh very isolated contact all over, and i'm not equal to that task i'm not a mathematician at all, uh, but, eventually though, though you, uh uh uh somebody, well there's some fairly good work going on in that area, some of it is by uh, finite element methods, um, but eventually we all know, that um, surfaces, uh will give up, somehow or other, we can uh, you know, with the elastohydrodynamics we've got alpha and E, to the point-seven power and then here we've got uh velocity and viscosity, divided by E-R, to the point-seven power, and a couple of other little things that are constant out front and so on, H over R, uh, that is a uh, we can calculate a film thickness. uh, as i say, working with lambda, we kinda get a sense of, you know the comparative, comparisons there, but none of these equations say much about, what do the materials think of it all. uh, will these materials uh you know uh, go on forever with smaller and smaller lambdas, how do we start to get into this equation, the stress state on each asperity or little points of contact, and then work outward from there, uh, well, by by what sort of mathematics? uh you could say i suppose that an asperity comes by, hits this one and tears it away a single contact causing adhesion that tears away, or, could you say that many asperities pass by here and eventually this thing falls off by fatigue, we don't know the bridge, from hydrodynamics, to material failure, yet. well uh i'm not going to help all that much in that area, but i'd like to uh uh report, some of the uh uh some of the chemical effects as we had found them, this chapter really is a kind of a summary of uh the work in our lab for the last twenty years, and uh boundary, lubrication, and uh well let's go into the chapter one uh chapter nine page one-fifty-seven and and -eight, um how do surfaces fail? well we've got, general terms. we, don't speak specifically about the experience of individual asperities or regions, surfaces, fail, by uh a number of um, well mechanisms that we'd we've come up with, we say there's galling, on the surface, or we say there's scuffing or there's scoring, and uh other such words that we use, um, i think it doesn't help us understand how it all started, um, but we do call that, uh uh the situation that leads to those kind of failures, we call that marginal lubrication which i've written before on the board, uh boundary lubrication, and um, inadequate lubrication, or catastrophic lubrication, or, words like that, uh but let's go back in this this word boundary, lubrication. uh Vern used the term. Vern used the term in the sense of, fluid mechanics. usually, uh boundary lubrication is when, uh lambda is too small to be uh very helpful. uh, it's it's, the the word boundary in that sense is uh boundary, is uh in the sense of, between, good and bad hm? <P :06> uh, that's one sense uh, there's another sense in uh, uh that uh in which people use this sorta thing, whenever, whenever the coefficient of friction is greater than some number, say point-one. or maybe point-one-five or maybe point-two or point-one-eight. uh that is functionally, one defines boundary lubrication, uh as um well there's that this is this is evidence that things are not, as you as you like them. a different definition. um, well another, uh evidence for, or, definition of boundary lubrication is you know as we see- as we've seen before, the lambda ratio, as being less than one, uh well there's no kind of limit to where the boundary lubrication regime is, uh on the Stribeck curve, some put boundary lubrication here, some put boundary lubrication there, i think what you see is, that boundary lubrication is just everything to the left of where things work right. <SS LAUGH> (xx) which isn't all that bad i mean say, if you're working out here in the lubrication uh area, you can't spend too much time helping other people define their terms you know? uh, so you see these terms in the literature i i i say this without value judgment particularly, well marginal or boundary lubrication's been studied um, in a number of ways, and by a number of people. you remember i mentioned um, that uh Sir William B Hardy, in nineteen twenty-five or thereabout, uh did an experiment dropping fatty acid on water, and then, picking up that film on a glass plate and then sliding doing s- uh friction experiment, and uh, uh that was the experiment that dispelled the the uh the uh um, interlocking idea, uh, but Hardy went on to do more work, this boundary film or this fatty acid film, attached to the surface by some, method, uh different from um, uh surface tension. that is to say you put a drop of water on a surface, and tilt the surface, there's surface tension indeed but that drop will just flow right off, and um uh, leaving scarcely any mark behind, now, these fatty acids, didn't, just flow right off, they would, uh uh, form um uh a residue where they parked. there was some chemical reaction. boundary film um, real chemical boundary film work started there, but then um, uh there were several other kinds of uh boundary films there are, uh on page uh one-fifty-nine, uh one you see in the literature three classifications of uh, of boundary lubrication, uh A, an unreactive lubricant metal pair in which lubricant viscosity decreases, as uh mu rises as temperature increases. uh, well that's just simply the straight mechanical boundary lubrication. friction goes up as the temperature goes up. how about B? well there's a fatty acid or, the metal soap of a fatty acid, which melts at a particular temperature and becomes ineffective as a lubricant, at higher temperature. but it sure worked at lower temperature. then we've got the third kind of boundary lubricant, it's a liquid. containing a reactive constituent that forms low shear strength compounds, on the sliding surface at even at room temperature, uh a little bit, these reactive constituents are referred to as E-P, extreme pressure additives these days, but what happens with this kind you see, there's some reactivity, in the view of the fact that the friction is lower than say case A, uh but then as the temperature rises there's greater reactivity and makes the temperature go dow- uh the friction go down, but eventually even this, gives up, and the friction goes up. well the um, the uh uh effective, constituents, are uh phosphates, which work up to about two hundred and fifty C, chlorides which work up to about four hundred, and sulfides which work up to about four hundred and thirty. well so the uh, uh_ one could ask then why not just dump a lot of this stuff in your engines, and or transmissions gearboxes or whatever because, um uh they will work, but one has to be quite careful because they work, ultimately, by corroding, the metal, and the product of corrosion, lies on the surface, and if it's, the right, has the right properties it's a lubricant. that's the, that's the mechanism of chemical boundary lubrication. it's that simple, well relatively simple. um, a chemical reaction occurs with the substrate metal, and uh, that, product of corrosion, has a low shear strength relative to the substrate if you do it right, and that produces low friction, lower than if you didn't have it there, and whenever you have lower friction, do you have less wear? well, 
S4: not necessarily 
S1: sometimes. certainly you have less tendency to scuffing... so that's the friction in marginal lubrication, next we have the wear, in marginal lubrication... uh how fast do these materials wear, uh when um in various uh types of lubrication? in figure nine two we have three curves. three lines, line A is for, inert substances line B is for mineral oil and line C is for commercial grade, engine oil, now we use these three, uh oils or, environments, in our work, as uh uh easy ways to get certain effects. but first of all let's say, that line A, has been predicted by um, uh considerations of elasto- hydrodyn- cont- uh elastohydrodynamics and contact, mechanics. um, uh Tallian of S-K-F Bearing Company, uh uh very, uh sharp contact mechanics guy, says that, uh you've got rough surfaces well everybody says that, uh, some of these bumps are higher than others you know statistically, uh now if you've got a uh, uh an adequate film thickness there, none of the asperities are in a state of high stress. but if you increase the load or decrease the viscosity of the lubricant in here, um, your calculations will tell you that you could get a thinner film. and uh, now if that thinner film is thinner, than the height of these asperities then you could say well, some of these then must be making contact with each other, and i'll you know you just have to imagine these are touching, uh they're making contact with each other, and uh, uh, more of these will make contact with each other as that film gets thinner and thinner. so, uh Tallian comes up with a load sharing idea, that the thinner the film, the less the load will be carried on a fluid film and the more it will be carried on asperities, then he says that, any loads carried on asperities, is subject to the considerations of you know of dry wear or something like that the Archard type of wear. hm? that if you've got, minute, areas of contact, you're gonna have wee little bits of wear, if you've got several regions of of contact, you're gonna have more wear, or what happened is uh uh c- uh uh Tallian's final calculation is curve A, uh that is the amount of wear one would get, that's on a log scale you notice, that's the amount of wear one would get, uh uh as a function of lambda. when the lambda is three and over, ah i should have mentioned that earlier too you know lambda is the ratio, of uh H over sigma, sigma you recall is a statistical quantity, and uh, when you deal with a statistical quantity, let's say you've got an R-A of ten, likely there will be, uh a good number of them also at height fifteen, some at twenty fewer at twenty-five but some also at thirty, and this is kind of the uh the the range that people take, uh that um a lambda of one may have asperities that are three times as high, as your average. and so you see the Tallian curve starts at about three, well that's carefully crafted to come out that way i did it on my own Macintosh, uh but if you look in uh Tallian's article it does come out that way, so then as lambda gets smaller and smaller, the, load carried by asperities increases, and the wear then should increase, uh in the same uh uh uh in the same way. is it true? does it happen? well in order to, in order to check this out, you've gotta test this, uh you know hypothesis in a neutral atmosphere. why a neutral atmosphere? the assumption is that when there's contact you've got adhesion, and adhesion can come about if you don't have, uh oxides. alright? so, our way of doing that, is to use, vacuum pump oil, um, uh diffusion pump oil which is a silicone, and it has no gas in it. uh uh unless you you know leave the bottle open and shake it or something like that but as you buy it there's no gas in vacuum pump oil it's a silicone fluid, then you, you know on our machine that we had here, you pour some of that on the specimen, and then we put a plastic bag around it and blowed nitrogen through there, and we've got an iner- iner- inert atmosphere as inert as uh as you needed you don't have to have a vacuum. in my younger days i always pumped up a vacuum but now we can get it much more, cheaply, with just a bag of nitrogen. alright, then using that inert atmosphere, we got a wear rate that followed curve A. uh Tallian is vindicated, i believe in that, uh load sharing concept, well, we did another experiment, then with laboratory grade mineral oil, and that always has some entrained air, uh in it and uh, uh something different happens. uh at the uh lambda of three or thereabout there's very little wear, and this curve kind of go- goes up along with A, but then finally peels off at some point. uh well why does it peel off at that point, or what is the meaning of peeling off in the first place? this, surface will function at a, lower lambda ratio, without severe wear. well, that is the case of, um some kind of coating forming on these, uh these these points of uh collision, and uh, those coatings um, well i'll give you the bottom line, those coatings will prevent high, shear stress on that asperity. and we'll mention this two or three times, the the the_ my best explanation of that mechanism, comes out to something this, we know that uh, uh let's just kind of make a couple of uh of asperities here, um, when one slides over the other, uh there is a uh uh likelihood of uh you know some kind of traction being transmitted from one to the other, and we can use our familiar terms when there's a load on that asperity, and there's a uh traction, uh which is uh we we can call F or, F is the force required to overcome that traction, and we're familiar that uh with these terms that that is F over W, and um, um, now from mechanics, we know a few things that, backing up to a simple normal load, the first plastic flow happens there, and then with increased normal load there's gonna be uh a larger and larger plastic field, now if we instead of or in addition to W have a, finite value of F applied, the point of first plastic flow moves out and finally appears at that surface, when mu is about point-three. on this asperity. well uh on that asperity that's different i want to reiterate, that's different from a uh coefficient of friction that one measures over a large surface. because we're talking here about very remote little spots, each of these might have a coefficient of friction, of, point-five point-six point-seven or something like that, and it might not it probably would not be detectable by, the measurement of the overall friction of the system, because the rest of this area is so large compared to those little dots, and we've got fluid, shearing, over every place else, and we might have solid adhesion, in those minute spots, but there are too few to measure. and we, have uh we've done this many many times, we know that in scuffing tests, in our machines like the one i had here, uh that the coefficient of friction bobbles along like this, so on and so on, eventually that friction goes up as the surface really goes to pot, but if we had in- and we do investigate this look at that surface in detail all along the way, you often find that about two-thirds of the way through final failure, is_ you see the first evidence of, scuffing. scuffing starts, way back there, and you can never detect it by friction measurements, by sound, you know ultrasonic devices, by, temperature by nothing, uh it just is an insidious little force, i suppose like a cancer or something like that, we, how do we detect it? well by, diligent microscopy, we don't of course, take a specimen, and run it at, uh this le- this amount of time, examine it put it back in so on and so on because, whenever you take it out of the uh of the system, it's oxidizing while it's outside of the system, and the situation has changed, uh and you dare not, start from there. so rather, we usually uh put in a specimen one for five minutes ten minutes fifteen minutes and so on down the line, and do this often enough until they were reasonably sure, that out here is where the action is, uh in scuff initiation, and uh out here is where the, where the where you get the real evidence. and this, th- this is the point of a lot of discussion at conferences. uh, we, um well you can probably reconstruct the arguments (uh) well enough anyway, when the coefficient of friction on an asperity exceeds three-tenths or some number like that, the plastic flow field comes out, to the surface, and uh begins to, uh, push materials ahead, of the um uh of the upper slider, scuff it forward like that, and uh, that bump is higher than the original bump. hm, on average. how do we know that? make surfaces rub together and just keep rubbing and keep rubbing and keep rubbing, uh and do, surface roughness tracing, uh during those experiments, and you'll see that the surfaces get rougher. and uh, that has to be due to, this sort of mechanism. there's nothing else and there are no other interactions, on the surface, that could make the roughness increase like that, so there it is, alright, uh, so now we got uh uh, uh, oh yes and, and obviously, when little chunks pull out that coefficient of friction must be very high on that surface, uh but somehow or other, when a soft substance forms, and that soft_ if that soft substance produces an asperity coefficient of friction, less than three-tenths, we have uh uh shall we say, arrested the uh uh uh roughening of the surface, and uh in fact you'd probably even get a smoothing of the surface with the coefficient of friction is low enough, and, what was a danger point here, uh has uh now been, uh uh smoothed down, and then here's another danger point and that gets smoothed down so forth and, eventually the sur- the the the surface kind of uh, uh uh those two surfaces are accommodated to each other, well we'll go over this a couple more times, but curve B, uh, follows curve A for a while, and uh then uh peels off, and uh uh looks like it's a protected surface, we don't get an awful lot of_ that's a log scale on the left side, we get a great reduction of, wear, we_ because of the protective action of a film, which in this case is, oxide. a chemical analysis doesn't show anything other than iron and oxygen, mineral oil, uh we get laboratory grade, mineral oil with very low sulfur content, and what we see on the surface uh well, uh you see a sulfur zipping by every now and then but it's predominantly oxide. F-E-three-O-four oxide, which is nice and soft and protective. 
SU-M: and this is still done under uh oxygen-free atmosphere?
S1: uh no this is done in air.
SU-M: oh that's in air.
S1: yeah, done in air and we've worked with dry air and moist air and you get the same results... 
S5: what happens if you do it in nitrogen?
S1: if you do it in nitrogen, after time, the surface uh uh fails. uh, the oxygen is used up on the surface, but you still have uh uh contact and uh, uh oxide gets knocked off, and it doesn't get replenished after some time, and the surface moves toward failure. yes sir,
S5: um, one question i have is, as that plastic field moves up and forward, [S1: yes ] how did the shear planes change? do- does it rotate, with it, or does it stay in the same, same orientation? 
S1: yeah you mean to say, does this then, kind of go this way?
S5: well yeah right now our_ with, direct normal loading it's at the forty-five-degree angle. [S1: yeah ] um, when it moves up toward the front does it rotate, or does it uh, [S1: well ] stay in the same orientation and move forward?
S1: yeah there are, a couple of views on that are you acquainted with slip-line fields... in plasticity?
S5: very little.
S1: very little, yeah you'll_ you know there are there are slip-lines, uh, not always oriented at a forty-five, uh with a large normal force they'll be oriented a little flatter than that, uh or in other words with a lower coefficient of friction they'll be flatter with a higher coefficient of friction they're higher, uh so on, slip-line fields um, uh_ Peter Oxley of Australia and uh Ken Johnson of Cambridge are trying to work this sort of thing out, uh just exactly how does that material flow. hm, i ought to know slip-line fields, i uh took a course in the subject uh, back in eighteen seventy i think it was, <SS LAUGH> uh, and i've still got some of those notes in my file and uh i look at those things and i say jeez, wonder who wrote this i mean i just don't understand it anymore. 
S5: does that have to do with the dislocations in a particular material? 
S1: no. [S5: no ] they just they treat the material as homogeneous isotropic and, rigid and plastic perfectly plastic. you know? it's really, as, us chickens, know it's really due to dislocations but the continuum mechanics people uh, uh treat it differently. alright so uh, now the next experiment is done with commercial-grade engine oil, uh a uh, uh we bought a couple of cans of uh, ordinary Mobil ten-W-thirty one day, and uh used that, and i got Mobil ten-W-thirty because uh Uncle Ed where i have my car oil changed that's what they sell. and uh, so that's our choice. <SS LAUGH> scientifically uh uh determined you see? um, well there again uh you see that the the wear rate, is uh orders of ten, couple of orders of ten less than you get with straight mineral oil, uh with uh and goes to even a smaller_ you_ surface goes to a smaller, lambda, ratio before, it fails, uh and the end of those lines incidentally are points at which, wear is so severe the surface gets rough and you'd rather not run it anymore... alright what, forms on the surface when you use engine oil? um, um well um a whole lotta stuff. now um, when doing this kind of research um, you are driven to uh looking for some shorthand way of doing things, because, scuffing, tests or, uh lubricated wear tests can be quite lengthy, and um uh there again your, conclusions are so strongly connected with the geometry that you use, and um, you had seen and i had mentioned before, that in our work we prefer to use the disc with a cylinder laying on it, and we had found over the years that that configuration connects with or correlates, significantly better with practical problems, than the pin, on disc. and, i i mentioned just a little bit and i'll just mention some more here, that you know whenever there's uh whenever this disc turns, uh with high-contact pressures, something gets, loosened from the surface, and those are usually oxides, and uh those oxides get loosened, and come up from under the slider, and move around, out in front again, and um uh if you've got a pin, a contact, you've got uh uh geometry of contact such that the, oxide and debris other debris particles, are washed aside. but if you've got a cylinder laying there, the uh debris are constrained to to pass through the contact area again, and that is what happens in, most practical things. the debris somehow hangs around and gets, shoved through contact again, and does its own uh peculiar type of damage. so we'd used the uh cylinder on flat, uh geometry, other people use other, geometries, um, their conclusions come out a little different in time and wear rates and so forth, uh but i just want to mention the kinds of tests we do because the conclusions are, are uh are uh given in those kind of terms, how do you do a scuffing test? or, um any test that's going to last for a long time? uh you you know you don't try to hasten them with high loads, uh that's, you know brings on a different wear mechanism <P :05> you can, do a test, at very low load a very safe load, and uh, run the test for, a year, <LAUGH> you know? before you get failure, or you can increase the load, and get failure in say a month this is a log scale, or you can increase the load and get a failure in a week, or a day, or an hour, or, whatever, and we've done some of those not the year-long ones, uh to see (xx) if there's some regularity, you know? if you get_ if you don't get regularity in other words if you don't get the same surface appearance, in every one of these uh, at the end of every one of those experiments, uh you're not, invoking the same, failure mechanisms, hm? well we had found uh that you know there's a limit above which you dare not go, uh, and um, uh so we kind of restricted ourselves to, uh some kind of lower area of load, uh but then we were_ we didn't want to uh, um, run these tests for a month and a week and all that sort of thing, uh what we tried then is to is to see once if it's possible, to use a step-load test. here is the load, and uh here is the duration or time at load <P :06> <WRITING ON BLACKBOARD> and so we put a load on like that leave it on for ten minutes and increase the load (xx) for ten minutes and so on down the line, and now we're looking for, the step at which failure occurs. a step-load test. and the first question one has to ask is whether or not that is a fair, uh way to test, uh or fair way to simulate this, this we feel is probably a val- more valid way of testing, but it takes a long time, does this short test, connect with this? and in um our work it does if you do it right again, uh, and uh what is doing it right well you could do a whole lotta wee little steps you see, and uh or you could do uh several big ones or, you could do a step that lasts twenty minutes and so on down the line, uh we had found that um, if you, if if you have uh uh step lasting about ten minutes or so, uh it works okay it gives you time to get your instruments going and take a sip of coffee and so forth there's no point in rushing the thing, and if the, uh you you do find uh that say in this uh sort of graph, that there is a load at which the thing only lasts you know a little, a second or so, uh so now there is a critical, load, if we divide our steps, up to that, particular point in say about, seven eight or ten units, and that's how we pick our, step height, or so, uh we end up with a test that's pretty good. and uh uh this i do recommend if you're, into the, uh topic of of uh scuff, re- scuffing research, the step-load test is pretty good, um, it's certainly short and you've gotta know what uh what you're seeing, when you do that.
S2: is there any correlation between uh the failure mode there and the compressive failure, that you might do in, one of these, pencil machines or opposite (xx) where you actually get a number, for the pencil failure, and then there's another number that's higher for the compressive failure. [S1: ah ] is there any relation between those? 
S1: i don't think so. i don't think so. no <P :05>
S2: i mean (xx) for the particular material.
S1: yeah. no, i don't think so. (xx) uh, um, the uh_ you see we're looking for the failure of asperities, and that involves both normal load and shear load on that asperity, and furthermore we're looking at, repeat, uh events. [S2: yeah ] we're more into plastic fatigue, than any other property of the material, i think. i say then, right up front i do believe that galling scuffing and scoring, are basically plastic fatigue mechanisms <P :05> plastic fatigue mechanisms... but of course one has to take account of, what is the, coefficient of friction on the asperities, you know. uh, X number of uh collisions, um, uh, uh is not the answer i mean you gotta know how mu- what the severity of stress was. and then the other imponderable of course is with a whole lot of asperities (xx) how do you know who's got bang how many times you know sort of thing?
S5: does that coefficient of point-three, change with uh, different materials? i-
S1: ductile materials.
S5: yeah is that just uh, steel, per se? 
S1: that's all ductile materials. 
S5: all ductile materials okay 
S1: all ductile materials. yeah. mhm. 
S6: at at what point, in the wear, then does any noise begin to be produced? 
S1: at what point in the in the in the test?
S6: yeah if you, started out, with good lubrication and you, slowly, 
S1: and at what time does noise at what point does noise change? [S6: mhm, increase ] after well after the first uh initiation or the_ yeah the first initiation is first isn't it. well after the, <SS LAUGH> well after the initiation of scuffing, the microscopic uh evidence appeared you know.
S6: is that at the knee of the curve then?
S1: uh knee of which curve?
S6: the_ your uh wear rates.
S1: oh no. no the_ that's not connected. uh, let's say, those curves there [S6: mhm ] um, let's just put one of them up <WRITING ON BLACKBOARD> uh let's take the mineral oil one and go like that something like that. yeah. uh, these_ this is this is wear rate that is a steady progression of wear and where does the noise start? [SU-M: yes ] at the kickoff. [SU-M: so ] you know you know at at the very start. the kind of noise, uh doesn't get particularly worse. uh the the noise that uh throughout the whole test doesn't get particularly worse until we get to that point then we deem that to be a a failure. but all the while that this wearing is taking place the noise is about the same. [S6: mkay ] just kind of a low level of noise, somewhat like you know when you run a lathe or a cutting tool there's a kind of a rumble or a low noise all over the place? uh but then when the when the cutting tool fails you know the_ what kind of noise you get? that's about uh what we, what we have here. you know? in terms of in te- you know uh uh noise pressure.
S7: so when you're doing the step test, um, noise is not a good indicator of when you're [S1: no ] starting to get [S1: no ] (xx) it always happens well after 
S1: yeah. uh when you're doing the step load test, typically um, you you measure friction uh during during during these uh uh load applications, and when you're measuring friction uh the friction will generally go up you know very much in parallel with the load. so you put on a load and if let's say you put this load on, and if the friction jumps up a tad then settles down, the next time you put on the load you're gonna get failure. usually or something like that you see? if that_ that works if your steps are not too big. uh and_ so then you see when you put this load on you'll get a little bit of a, quivering of the machine. it wants to fail, but it thinks better of it you see, <SS LAUGH> and then you put this on and then you get a, a significant noise. and that's failure. uh now that that is a that's a little different from uh, uh what i described earlier that um, uh, w- uh uh friction v- versus time, if you run one of these long-term long-term tests, <WRITING ON BLACKBOARD> friction versus time eventually goes up and the first evidence of failure is there, that's with a steady load on it, and you've got to kind of guess that that's happening somewhere along this line too every time you dump a new load on you're changing conditions. every time you dump a new load on is the equivalent to setting yourself up with a curve that's gonna fail earlier. so you jump from that curve to that curve. seems unfair but it sure is a lot shorter test. you know you can_ this this kind of test if you're doing things right, that kinda test is out of the way in an hour's time. maybe two hours, something like that.
S3: i- is that uh, is that cumulative damage or are you just trying to find the point that_ that two-thirds point where the scuffing actually initiates? 
S1: uh, it is it is s- s- somewhat of an accumulated failure, uh because you see the length of time that you run these at these steps, matters. uh if you have very short steps you'll probably get failure at a lower load. okay? uh so that accumulated time means something in two ways, that's either an accumulated damage or an accumulated curing. you know healing of damage. in other words the chemistry has time to, to work, to protect uh_ produce a protective film. hm? and if your steps are too short, you don't give it e- ti- enough time to build up a film.
S3: none of the area under the, under the curve then, really is uh, uh_ defines failure?
S1: <SIGH> well um i i r- i doubt that if you see if you have very short steps, shorts steps like that, you'll get failure quicker than if you have longer steps like this. [SU-M: (xx) ] 
S4: of the same height so
S1: at the same height yeah, [SU-M: okay ] yeah. <P :05> well so we have um, uh step_ a step load test, and um, uh i suppose we, did these kind of things for four or five years with about uh five or six uh, uh PhD projects, trying to work out macroscopically what we're seeing and connecting with practical parts. uh we had gotten satisfied with our uh, uh our uh three ways of testing, uh in a neutral atmosphere, a uh uh, uh mineral oil only, and a uh uh Mobil ten-W-thirty oil, uh and so we we tried a good number of things, but we always did see that um and i showed you yesterday, on one of those disks that had a dark streak on it. and uh, bit by bit we became aware that uh, uh that dark streak um, was uh uh involved in uh um um surface protection, and so what's in that dark streak? uh there have been several claims as to what could be in that dark streak, we um uh, though did our own analysis, and uh found that it was F-E-three-O-four largely, almost exclusively F-E-three-O-four, ah but there was something else in it. uh we had always found a relatively high amount of carbon, in that film, an amount of carbon that was, uh much greater than you could attribute to the uh carbon connected with the iron in the form of the substrate steel. you know we worked with four-tenths percent carbon steel forty-three-forty steel, and uh the amount of carbon we saw now, uh relative to the amount of iron, was uh several times greater, than you could connect with uh the carbon in the steel. well furthermore we had found that um, the um uh coefficient of friction, of dry steel on dry steel was uh about point-two-five, and uh this is uh before lubrication dry, uh before <WRITING ON BLACKBOARD><P :04> that coefficient of friction at about point-two-five, well then the next thing was to um_ we formed that film under th- the step-load kind of test, and when the film was nice and dark and uniform, uh we then dried off that surface with solvent solvent dried and the coefficient of friction was point-one-two... um, that's (on) a new_ other uh mating slider and so on. well could we safely infer from this that the uh uh that whatever that film is uh it uh provi- produces low friction uh reduced friction, there's something in that film, uh what is that carbon? uh that's_ remains a mystery. now uh we did one other thing, uh, since now we are_ had come to uh believe that there's uh, uh uh_ that oxygen, uh is important in the system to form F-E-three-O-four, uh we began to uh uh uh work with uh uh different uh uh oxygen states. we deaerated... is that the right way to spell that? oil, and uh, uh we got earlier failure. uh we aerated, that_ i don't know_ i forget_ is that the way to spell that? somebody must know how to spell that but uh, oil, that is uh we bubbled oil into the thing and uh uh you know cooled down the oil and bubbled oil into it and shook it and all that sort of thing got more oxygen in, and uh we got higher load-carrying capacity. um, well then the next thing we did_ i was uh at the same time as this project was going on i was working with uh, um a uh maker of uh, um um of um hydraulic fluid, uh water-base hydraulic fluid, and the water-base hydraulic fluid uh would uh chew up the pumps rather badly. uh but it was still worth doing because uh you could re- re- replace a pump every week rather than replacing a steel mill every year, uh, you know when when hydraulic lines break in the presence of uh red-hot steel, you have fireworks. <SS LAUGH> well um, we then took uh_ used water in our experiments and uh, uh, found a uh, uh... a coefficient of friction after sliding with water, a coefficient of friction of point-one-eight rather than the point-one-two. now that doesn't have to be solvent-dried just dry in the atmosphere, uh then we um um deaerated water, took the air out of water by the freeze-thaw cycle with a vacuum pump attached, and uh oh excuse me this, produced F-E-two-O-three. uh we deaerated water, and uh um got a lower coefficient of friction, about point-one-five, and then we found on the surface there was F-E-three-O-four. well uh we took another shot at uh pumping air into the uh oil and uh we did that long enough so that uh we ended up getting F-E-two-O-three, on the surface, rather than F-E-three-O-four. uh then if you_ well there are obviously, uh uh uh, uh combinations that one can get, but uh in every case, when we formed F-E-three-O-four, we got low wear, and whenever you had F-E-two-O-three you had high wear. <WRITING ON BLACKBOARD> <P :06> F-E-two-O-three is harder than F-E-three-O-four, and it's a lot thicker. and we found that the F-E-two-O-three was approximately eighteen hundred angstroms thick, and the F-E um three-oh-four was something of the order of three hundred or four hundred or so angstroms thick or thin, and it was of course uh well well very protective, low wear. well now uh, so much for identifying what occurs_ what happens on the surface, [S4: um ] sir 
S4: where did the carbon come, or go? i mean is the carbon in all these other tests (xx) 
S1: oh the carbon is not_ where you have water you don't have carbon. 
S4: okay.
S1: yeah that's a good point (xx) uh now, then what does that lead you to to guess, that perhaps that high carbon content is um [SU-M: breaking down the oil? ] some oil. in what form? now if you uh, if you think of F-E-th- or any ceramic material, uh furthermore let's see at this point we started looking at these surfaces with quite high magnification, and found that the uh F-E-three-O-four, was present as flakes. and uh like this, and these flakes might be five ten maybe fifty layers high. and uh, where do these flakes come from? uh well i can believe that these are the native oxides that reside on t- tops of asperities, and they get knocked loose and they're carried downstream somehow or other maybe piling up in front of the next asperity well we kind of uh are reasonably sure of that, piled up in front of the next asperity in various layers and then when a real serious, asperity comes along, it uh finds a uh a film_ multiple films of F-E-three-O-four with what in-between? perhaps there is, uh a hydrocarbon in-between, but uh, that hydrocarbon cannot be uh very thick. uh because of the coefficient of friction point-one-two, uh at very uh uh um_ when the surface is the so- the top surface is dried, there are these um uh layers possibly of uh of oil between, we are are_ we have some specimens in a an expensive analysis lab this at as we speak, trying to figure out what's there. uh is it uh oil so there's_ th- th- the oil has been tagged we've got some, specific elements in that oil for tracing, uh uh that's uh you know uh the the oil between the film_ the flakes. but if you_ if this is oil lubricated, then we would expect to find that the higher the sliding speed the higher the friction as in the case of viscous lubrication, and we don't find that. we get a uh solvent-dried surface uh as a coefficient of friction about point-one-two no matter how you slide it and we covered two orders of ten in sliding speed. a constant friction suggests something closer to a solid than to a liquid. so we don't know the state of that stuff what it is, but we could i suppose uh uh begin to believe that it's uh uh these are films that are molecularly thin, and what about films that are molecularly thin? this is only uh um uh vain speculation maybe i don't know, but you know the people that uh uh do molecular dynamics uh do some calculating on this line but, there_ at the same time there are people that take mica, sheets, and um well they actually take mica sheets and make uh c- uh uh cylinders of them, and uh then they uh cross these cylinders like this such that there's an a_ put a load on, and there's a uh kind of a a a flat region of contact between them, and then they uh, uh_ well (we) gotta have a little separation there_ put solvents in there uh, not uh y- your simple ones but uh solvents very low viscosity substances, and then uh by uh uh multiple-beam interferometry, work on how thin those films are, and they are working with films that are uh, w- uh two three and four molecules thick, and they measure effective viscosities that are a hundred to a thousand times greater than bulk viscosity... now they also pur- uh purport to see waves passing through here just as Vern mentioned, uh the work of Winer and (Bear) uh, fluid between two uh hard steels, are solidified and then when this is sheared you see shear lines or shear plains. well are we at the uh, boundary between liquids and solids here? and is this stuff, uh really uh molecularly thin, oil? uh which is_ well what about that oil? why is this stuff ha- so viscous? well it's simply because you see this mica or the steel or the oxide or whatever, you know it's composed of molecul- of atoms rather, and uh surface atoms have a much higher energy state than any other row of atoms in the substrate by virtue of having them deprived of neighbors, and uh their bitterness shows, <SS LAUGH> uh and any liquids that form, uh that uh wander in there uh take on the orientation of the substrate and if we've got another surface here not too far away, perhaps the entire liquid in here becomes oriented or settles in some, uh some order. and an ordered structure is a solid or a solid is an ordered structure, so very very thin films do have viscosities very different from the conventional, uh uh kinematic or uh uh dynamic viscosities. i don't know if we've got that here or not. um, but uh the fact that uh this coefficient of friction is independent of shear rate suggests that we are not looking at bulk viscosities there <P :05> well um, let's see that was your question are you_ did you have any follow up on that at all or, do you_? you know your people are working on this kind of thing too aren't they? uh yeah. yeah. you have anything you can reveal at all? <SS LAUGH>
S2: you mentioned several things which thing are you asking about? 
S1: oh i see. <SS LAUGH> okay...
S4: we probably don't know as much as you would like to know about it. <SS LAUGH> it's not really our (xx) we're trying to catch up with them.
S1: mean to say we ought to exchange our ignorance sometime or other? <SS LAUGH>
S2: we don't always know what's going on.
S1: boy i don't either i tell you. alright now, now that we kind of know what forms on the surface, and we're quite sure that it's F-E-three-O-four i mean uh, um, people in uh uh, another car company and uh and one of the suppliers to the other car company hm have favored us with some very thorough analysis, uh and it's uh, uh uh by Auger and other, spectrographic methods are quite sure that it's F-E-three-O-four. and uh in the case of l- liquid_ uh uh oil lubrication and quite sure it's F-E-two-O-three. but we've got a verification of that anyway we've done ellipsometry and we uh see certainly the F-E-two-O-three it stands out like a sore thumb but uh, so obvious in uh ellipsometry. but let's uh take a little look at uh uh something of the dynamics, of this uh sequence of events. it's not enough to know what the chemistry is, what we wanted to then find out was, what surface roughness, what hardness of substrate, what sliding conditions and so on and so on and so on will form this film? because you know there is a com- competition in practical machines. you know that uh a new, uh uh steel part, has only an oxide uh layer on it, and that oxide layer is simply not equal to the task of uh carrying a signif- a significant load. uh, and i think by now we see on the basis of the uh those those curves those three wear rates that uh, with no oxide or oxygen present we're on the way to uh uh a short life or early failure, and so any surfaces that survive you gotta say well there must be some oxygen getting in there to form new oxide. uh and then if we add um uh other things to the oil this stuff must get in there too. so we're faced i think with a um chemical kinetics issue, um new surfaces are on the way to failure from uh the kickoff of the game. how do they survive? or in other words what is the mechanism of break-in of sliding surfaces? most surfaces we deal with are inadequate to the task to start with, how do they end up surviving, if uh they don't have the proper antibodies in them to prevent failure? well i i i portray that in a, certain fashion on the front of the textbook, on the front of the textbook is a graph. and the uh graph is uh_ on the abscissa is time, and uh on the ordinate is something else. well uh, on the ordinate are uh two curves, and one of them is a curve that s- uh uh that shows uh something like this that, a surface, has a a uh uh a tendency to fail as new, and th- when you start rubbing it you are consid- you are already doing some damage to it so a tendency to fail increases with time... it's_ and if you don't do the right thing it's gonna fail. but there's another curve on there that says the capability of the lubricant present, is say down there when you pour that oil into the uh engine or transmission, the capability of that lubricant is not adequate to the task as it starts. you've gotta do some rubbing to get the right thing done. now then you've got to rub, with that oil present so that the capability of the oil catches up with and maybe exceeds the need of the part... this is the break-in time <P :08> so we've got two competing events taking place. the surface is on its way to failure, but the Lone Ranger comes in and rescues the part before it fails. hm? <P :13> well, the study of that surface dynamic, is uh uh_ requires a certain amount of faith i suppose in uh what may be happening on the surface but i think we all believe don't we? that surfaces are composed of little asperities little bumps, i have now put this kind of picture on the blackboard some eighty times <SS LAUGH> uh, <LAUGH> and (the name) you know uh the preponderance alone ought to uh be convincing i suppose but um, uh two surfaces touch each other on little regions of contact and we're gonna see oxides floating around so forth. now i ask you how you know that's our picture that uh these are bumps uh that, uh knock each other's oxides loose, how else do we find, oxides piles up in flakes like this? and we think those flakes are of the same order of thickness as the, native oxide on the surface. twenty-eight or so nanometers or whatever. uh, that's kind of hard to verify you see unless you get to a mic- uh scanning microscope uh up to about ten thousand X. uh, but until, until we see evidence to the contrary (xx) we're gonna say that, uh what happens is that uh contact between asperities knocks the as- knocks the oxides loose, and these oxides get shoved downstream and uh pile up, well there's gonna be some over here that pile up over here and so on eventually, as time goes by, uh you'll have more and more of these oxides all ready to uh uh to prevent severe contact by some passing marauding, asperity. that's the hypothesis now how do you verify or prove this? well, the common metallurgical tools don't work and uh, uh this was the uh uh a low point in my, my time being a metallurgist, uh you can do a hardness test, and what does that tell you hm? what does the hardness test tell you on this surface right here? well you know the hardness test penetrates that deep it doesn't tell you anything about what goes on on the surface. <SS LAUGH> uh you can look at that with a microscope an optical microscope and uh you don't see those you can't, tell what's in those flakes huh? you can clean everything off, um with solvents and so forth, and uh use a scanning electron microscope, but by the time you clean everything off with the solvents you've already taken the oxide off and so you don't see the evidence there either hm? in the nineteen late sixties and seventies uh several people were working on this problem including myself, and we would get to these conferences and everybody would sa- see different things. uh nobody reported oxides on the surface, but we all reported uh uh uh the different residues of the different solvents we were using. <SS LAUGH> ultimately, and uh uh that's because you know back in that day, scanning microscopes were nice and new and you dare not introduce a single atom of chlorine into a microscope in that day. half an atom maybe, but certainly not a complete atom you know? <SS LAUGH> and uh uh_ but then as time went by the miscrosco- microscopes got better and we were less careful about cleaning and all (that) we started working with dirty surfaces but by that time the cat was out of the bag. uh so there there was no good way to observe this, uh phenomenon, and uh at the same time i was carrying on some work on corrosion, and i was intrigued by an instrument called an ellipsometer. and uh, that corrosion people use... and any chemist worth his salt, will say don't use an ellipsometer, it produces ambiguous results, you get more confused than enlightened with it, don't even look at an ellipsometer. is that uh roughly the attitude of the chemists? roughly? roughly huh? hm, and it is true. i mean the uh_ it's a pretty cruddy instrument to use, but i must say for this kind of work the rest of them are worse. <SS LAUGH> uh, now the ellipsometer is a device that um, uh it's been long time used in in corrosion work, uh and i'll just describe roughly how it works, and um, what you do is you you uh you send a polarized light at a surface, and if you know it's polar state, uh then you observe what has changed upon reflection, and that change is due to the properties of this surface the properties of index of refraction and the index of absorption kappa. well now this requires of course some uh instrumentation because you can't observe these changes of polar state with uh the naked eye, uh but then this gets a little more complicated too this is now_ these are the, properties of the substrate, if you shine your ellipsometer on here you get an answer, for a clean surface in a vacuum hm? good enough. and now we put a coating on here and uh that coating will have a uh, an index of refraction of the film plus the index of absorption of that film, um and then course there's this other quantity uh the thickness of that film. now, uh, this_ these numbers being different from this, will alter this beam, from what you get from just reflection from that solid, and of course the thickness matters because there's more material that you pa- that the light passes through. well if you if you if you already know what that stuff is on the surface, a single shot with the ellipsometer will tell you what the thickness is. now if you don't know these quantities, uh but the surface_ the stuff looks kind of transparent you say well at the moment we'll set that to be zero, but then we'll do the measurements at two different angles, and from that you can get the thickness and the index of refraction of that film. if you don't know this do three measurements either angle or colors of light. and s- by_ that means that the corrosion people have s- slain many dragons, have uh worked out the rate of corrosion of surfaces, uh now the beauty of all this is this can be done under liquid. and uh this finally is uh, i- it makes more sense than trying to do an experiment, clean off the surface and put the part in a scanning microscope and so on down the line, watch this experiment in, actual environments. you can uh put a slider on that surface, and uh i- immerse this oil, uh and you put a tube uh uh uh a glass t- or any old tube with a glass window here like that so the light can uh enter the fluid and leave the fluid uh through a lens that's perpendicular to the path the optical 
<BREAK IN RECORDING> 
S1: but uh, apparently uh, you know better get at it uh, tis breaktime
{END OF TRANSCRIPT}

