2001-02-19 :
HERVE ARRIBART (HA) : I studied at the École Polytechnique in Paris. The selection to the school is done mainly on mathematics. But during my studies I learnt to appreciate physics in particular. I decided to pursue research in solid-state physics. It was a good place to study physics. While in my last year as an undergraduate I did a Diplome d’Étude Approfondie in parallel (an intermediary between an M.Sc. and a PhD typically done for a year before starting one’s PhD studies). In Orsay, near the École Polytechnique, there is a very famous place in solid state physics, a lab started by Jacques Friedel - a great name in solid-state physics. I followed this course and afterwards I did the PhD at the École Polytechnique in the field of condensed matter physics. In principle I ought to have started with a topic distant from materials science. I extracted spin-polarized electrons from semiconductors. This was in 1974.
ARNE HESSENBRUCH (AH) : How did one extract spin-polarized electrons in 1974 ?
HA : It is true of all solids, but in semiconductors it is especially interesting that when light falls upon a surface there is a coupling between the spin of photons (in classical physics : the polarization of light) and the spin of electrons. Electrons in the upper layer absorb light photons depending upon the spin. If by some technique you can extract electrons from the conduction band of the semiconductor, you can find ways to select electrons of specific spins. This was quite important at the time because at big-science institutions such as LEP [Large Electron Positron collider] or SLAC [Stanford Linear Accelerator Center], there was a need for spin-polarized electrons. And of course you then needed solid-state physics to do it. But the man who in principle was my supervisor decided to do something else. His name is Claude Weisbuch, and he is now a good friend of mine. For a few years he was the scientific director of the French Department of Defense. He is still working in solid-state physics, in the optics of semiconductors. But he decided to do something else and Bernard Sapoval, another professor at the lab, proposed that I work on new materials. At that time there was little contact between solid-state physics and solid-state chemistry. The idea was to link up with chemists. This is why very early on in my career I had contact with chemists. We worked with Parisian solid state chemists on a new material. We found a new way to draw single crystals of an already existing material. It was very nice because we could examine transport and NMR phenomena. And the material, a copper vanadium sulfide exhibited astonishing properties : a large spread of conductivity that one can measure in a standard experiment. We suspected that this was due to mixed conduction properties. Mixed conductivity refers to conductivity by both electrons and ions. The experiment appeared to verify our a priori suspicion. This gave me the possibility to present a model for mixed conduction in this material and to understand the influence of ion transport and electron transport. I also used NMR in order to understand which ions moved. It turned out that the copper ions moved. So, this was the subject of my first thesis. At the time, in France, there were two theses. The first one was called "thèse de troisième cycle". The second was the "Docteur es sciences". This degree does not exist any longer. The thesis that is done now is shorter.
I decided to continue to work with chemists. I decided to combine NMR and transport measurements. I changed my collaborators, turning to two different groups. In my PhD there had been two chapters on NMR. But I wanted to study proton transport. I had two reasons. One was that protons give a strong NMR signal. The second reason was that two reasons had been given for proton transport. In one, protons move in individual jumps. In the other the proton is a part of a more complex molecule such as the ammonium ion (NH4+) or hydroxonium (H30+). In the former case we can see the transport phenomenon as a result of molecule rotation and proton jump. The molecule would turn and the proton jumps to the neighboring molecule, which again turns and so on. This was called the rotation-jump model. The second model was for the whole complex ion to jump. This was called the vehicle model because the whole molecule acts as a vehicle. So I worked with one group of chemists in Nantes, at the Institut de Matériaux de l’Université de Nantes. It had just been created by Jean Rouxel, a chemist. With them I worked on a substance called antimony acid - a solid. I was able to show using NMR that transport occurred in this case with rotation-jump. Protons used H30+ as a complex rotator. I was also able to show that the jump was due to quantum mechanics within a certain temperature range. It was not the usual ion transport of classical mechanics.
AH : A tunneling effect ?
HA : Yes, this is one aspect of protons, because protons are very light ions allowing for this quantum effect. The other material I studied was ammonium beta alumina. This was the standard beta aluminium in which the sodium had been ion exchanged with ammonium. This material was very interesting from the perspective of NMR. All kinds of ionic motion took place at different temperatures. At the lowest temperatures, that of liquid helium (1-4K), there was rotational quantum motion. As the temperature increases the motion becomes classical.
AH : If I may make a comparison with Stanley Whittingham here. You were working on some of the same materials, you were using some of the same tools (NMR), but you were asking very different questions, right ?
HA : Yes, that is true. I was not at all involved in the application. For two reasons : French chemists were interested in materials and did not look to the application. And chemists were between me and the application, so I had no contact with attitudes such as Whittingham’s. I was very happy working on the solid-state physics problems.
AH : And we are talking about the late 1970s now ?
HA : Yes, I began the proton transport research, I think, in 1976.
AH : And it went on for how long ?
HA : For five or six years.
AH : And you lived in Nantes ?
HA : No, I remained in Paris while collaborating with the Nantes group.
AH : Were you employed in Nantes ?
HA : No, not at all. At the beginning I was employed at the Ecole Polytechnique as a research assistant, and then I was hired by the CNRS - in 1977.
AH : So, the CNRS paid your salary, you were able to do basically whatever you wanted, and you collaborated with Jean Rouxel and coworkers because you found it interesting ?
HA : Yes. It was a chance to work with an outstanding chemist. French chemists were really very good. The problem, as we just said, was that there was little interest in application.
AH : What did you do next ?
HA : After my PhD thesis, I found it interesting to go to industry.
AH : I imagine that there were many advantages and disadvantages to leaving academia for industry. For instance, where was status greater, what paid better, where were working conditions better ?
HA : First of all, it was rare, even more so than today, for CNRS people, or people within the public system, to go to industry. I cannot give you a clear answer about my motivation - it was not even clear to myself at the time. I did get a higher salary in industry. I also had personal reasons for leaving Paris and going to the Elf company. I went to an Elf research lab in the Southwest of France, in a very nice place in the Pyrénées. I had small children at the time and it was much better for them to grow up in the countryside and in a very nice climate. I was also curious. So the decision involved many elements. And anyway, it was not irreversible. The CNRS allowed me to take a three-year leave after which I could have gone back. With regard to the working conditions : I was of course less free than I had been at the CNRS, but I found it more stimulating because there were a lot of different problems on the horizon, arriving almost every day. We could easily get the necessary equipment at the CNRS and at Elf, so there were no differences there.
AH : The restrictions at Elf had to do with what you were allowed to study ?
HA : Yes.
AH : What did Elf want you to do ?
HA : In principle I was hired to work on solid-state sensors. Because it was not in the direct line of my previous work I proposed that I work on solid-state sensors and ion conduction. We developed a family of sensors named ISFETs (Ion Sensitive Field Effect Transistors). It was a new kind of transistor at the time but now it is very common. You control the electrode using field effects, opening and closing the circuit between the two other electrodes. This is the way the transistor works. My idea - not an original one - was to replace this way of controlling the electrode, the gate, to replace it with a membrane, selective to such and such an ion. If you put the device in a solution containing the ion for which you have designed the system, the membrane will be charged. This charge will change the state of the solid-state transistor. It worked all right for protons. We could use the device to measure pH and afterwards we just had to change the nature of the membrane, choosing a different solid electrolyte, such as calcium fluoride.
AH : Your toolkit remained the same and you still used NMR ?
HA : Not NMR, but yes. You need large samples in order to do NMR. So it was mainly electrochemistry and surface analysis. This was between 1982 and 1985. But as I told you, in industry new projects can arrive almost every day. I had developed some skills in electronics using instrumentation at the École Polytechnique. Elf applied for a patent for a medical analysis system, a small instrument to be sold to private practitioners, as opposed to hospitals. This had nothing to do with solid-state ionics. But the people working on this project needed someone who knew about electronics, and so I got progressively more involved. After one or two years it had become my main project. This worked very well. I was very proud to design an electronic system that required no manual setting. It was set in the factory forever. This was a critical issue, because we thought that doctors could not be expected to deal with electronics - and I am sure that we were right in this. So there was nothing to check or calibrate - the system was self-calibrating. So it worked very well, and after only two or three years Elf built a plant and people were hired. But in 1984 and 1985 there were big changes in chemistry. And there was a great redistribution of all chemical industries. And Elf, that had been an oil company, in this period expanded to become also a chemical company. As a result a lot of the more diversified lines of business lost in importance. Many projects like ours were discontinued. But because we were already quite advanced we found a way to keep going. In fact it was Dupont de Nemours that found that our system was complimentary to some of theirs. The result was that Elf shipped the patent and everything else to Dupont. For a few months I considered following the project to Dupont and to the United States. In the end I decided against. I still wanted to work in solid-state physics and not to work completely in the instrument making business. But for one or two years I continued as a consultant.
2000-05-29 :
HA : For both personal and professional reasons, I decided to stay in Paris, and then Saint-Gobain offered me a position, working in a new research field : polymer adhesion on glass and other materials. It was a new topic for me too. At the time adhesion was not even considered a science. It was before Pierre-Gilles de Gennes’s Nobel Prize in polymer adhesion [1991]. It was rather considered an art. Even though I had no background in the field I was interested. What interested me in the Saint-Gobain proposal was that real breakthroughs were to be expected in the science of adhesion when two materials are brought into contact. In fact this was my first real industrial experience. Of course CNRS had not been an industrial experience at all, and even at Elf I was always in the research lab. As I explained, my work at Elf had nothing to do with the industrial activities. I never visited factories. At Saint-Gobain I had to do this, at least in the beginning.
AH : Did you not say that your development of the medical analysis system resulted in the setting up of a plant ?
HA : Yes. I did participate in the design of the plan, in order to make it efficient. But I had no role in the plant itself after construction. It was also a small plant for high-tech activity.
AH : You had nothing to do with the fabrication side of it, situated in the plant – merely the R&D before the plant became functional ?
HA : Yes, exactly. Saint-Gobain of course has many plants all over France and Europe, and even the United States. But at that time, the company was still franco-français [French through and through] in its general spirit and culture, despite the many factories in other countries. There were only French directors and the system was based on the French system of education. There is a hierarchy from the École Polytechnique through the École des Mines and the École Centrale to lesser schools, and you carry the status of your school within you for the rest of your life. I remember that I strongly felt the weight of tradition when I first joined the company. It is true that winds of change were already blowing then, but they were barely noticeable and needed a couple of years before really expressing themselves. But eventually the company changed its culture, and now the company considers itself an international one. I think a deep change has taken place during my 15 years with the company.
So, anyway, this was the first time I gained experience of the industrial aspect of research. My first task was to examine and synthesize different kinds of adhesion in Saint-Gobain’s products and processes. I decided to simultaneously pursue fundamental reflection and a practical approach, helping the factories improve their processes. This was a very instructive experience. I learnt many things although I am not sure that I helped the factories all that much. I certainly learnt for myself that I preferred to stay within R&D and not to progress into production. On the fundamental side, I developed a network of contacts in public labs in France and the US. This became useful later on. After three years in the field, and having created a small research lab, I decided to gain some distance from the practical aspect of my work. It was also obvious to me that fundamental research was required first. Progressively the idea came to me to propose the creation a special laboratory dedicated to the basic aspects of polymer adhesion - and of course also to related issues such as the surface science of glass. But I knew that Saint-Gobain was not ready to have a laboratory for basic science by itself, so my idea was to set up a lab jointly with the CNRS. This was in 1988. From the administrative point of view this was feasible : a number of such joint ventures already existed, an example of which is Rhône-Poulenc. Of course I had to convince both Saint-Gobain and the CNRS of the utility of the project which was not straightforward. Although I managed to convince Saint-Gobain in a manner of hours, CNRS needed more prompting.
AH : Would you explain the nature of Saint-Gobain’s research before your proposed laboratory ?
HA : It was a quite common kind of R&D geared towards problem solving. Helping the development of new products and solving problems within production.
AH : So the research agenda was driven by questions arising out of production ?
HA : Yes, and my idea was to get a more fundamental understanding of the questions which would enable us to help with such questions in a much better way.
Figure 1. Saint-Gobain Recherche, Paris
At Saint-Gobain I had to sell the idea primarily to the Vice-President of R&D. He took the decision just before retiring. The CNRS process was more complex. It has a democratic organization where decisions are taken by committees. The members consist of both elected researchers and individuals named by the Ministry of Research. They are divided up into different scientific sections. So here I had to convince a diverse group of people, and not just one person, as at Saint-Gobain. As I mentioned, I had developed a network of relationships in the fields of adhesion and surface science and now this turned out to be useful. I knew that many people approved of my research agenda. My project was accepted without much fanfare, but it still took a while because of the administrational hoops that a proposal has to jump through within the CNRS. They meet only twice a year, and every decision has to be validated by the CNRS directors and so on. It took maybe 12 months. The laboratory started on January 1, 1990. But there was only a building and neither instruments nor people.
Figure 2. Joint lab : Saint-Gobain Recherche & CNRS
The three yellow arrows point to the units of the joint lab within the Saint-Gobain Recherche building.
In the meantime I conferred with scientists in many other labs trying to recruit people. Of course the CNRS could not order people to go, so I had to entice scientists away. I estimated that I needed three scientists from the CNRS in addition to three scientists from Saint-Gobain. Two of the latter had already worked with me, and they followed me to the new project. A further researcher came from somewhere else – it was a young Chinese woman. We also had two or three technicians and some PhD students. Altogether, it took a year or so to gather everyone together. We also had to buy instruments and the process of getting the laboratory shipshape lasted altogether something like 2 years. We began to actually do some research in late 1990. And from then on the activities progressed rapidly. In two to three years we reached a plateau of 20 people, a level that had been stipulated by the CNRS. A third of the people had come from the CNRS, a third from Saint-Gobain, with PhD students constituting the last third.
AH : Where did they come from ?
HA : The latter were doing industrial PhDs (Contrat à Durée Déterminée) with Saint-Gobain, and their salary came jointly from the French Ministry of Research and from Saint-Gobain. Of the entire staff, about half each came from chemistry and physics. It was crucial that we develop knowledge and expertise in both these fields. Later we also developed an interest in mechanical problems.
Figure 3. The SPM from Park Scientif Instrument
AH : What was the instrumentation ?
HA : There was a conjunction of the beginning of our lab with the very early days of scanning probe microscopy. This new kind of instrumentation offered a very exciting opportunity. There was a risk in this. We purchased the first Atomic Force Microscope (AFM) ever in France. We bought it from Park Scientific Instruments. Later we built the first AFM for UHV purposes. There were many people then who thought the instrument had no future, so it was a risk to invest time and money in it.
AH : Why did people think it had no future ?
HA : The objection was that it was not clear that atomic resolution could actually be achieved with the AFM. It was not until 1993 that Binnig showed true atomic resolution.
AH : Well, yes, but before he had claimed to achieve atomic resolution. In 1993 he only claimed that so far he had been mistaken and only in 1993 did he achieve true resolution. Is that not right ?
HA : Yes. But in 1993 the community was convinced. The reason I did not hesitate was that atomic resolution was not actually the big issue for our purposes. Even a resolution of 1 nanometer amounted to a great deal. Much could be done with such a resolution in the field of adhesion, and also in fracture mechanics and surface chemistry.
AH : I have the impression that since 1995 or so many people argue that atomic resolution is not really that important, and that in the early 1990s it was still considered the holy grail. So you were unusual in that you had this attitude so early ?
HA : You are right that atomic resolution had a special ring to it in those days.
AH : Did you emphasize the issue of atomic resolution in your application to the CNRS ?
HA : I am not sure. Even today, nobody has achieved atomic resolution in glass. So it would have been a hard sell, also then. The same goes for polymers. And those two were our substances under investigation.
There is a difference between STM and AFM. They obey two different logics. The STM has remained a tool of basic research, in surface science. The AFM, even early on (and this would be interesting to discuss with Calvin Quate or Gerd Binnig), there was a hope that it could be useful, for example in other fields of science, such as mine, or in technology, such as process control, microelectronics, semiconductors, and so on. Generally speaking, in early phases there are always many people who think that a novelty will never become common. We have to remember that in 1987 or 1988 solid probe microscopes were still big and unwieldy instruments. Of course miniaturization had set in by 1990, but it was a novelty. Only very few people were convinced that the AFM would become so common. Calvin Quate is one of the few. The STM has revolutionized basic research on metals and semiconductors. There was a reaction against it, because surface science was done using diffraction techniques working in reciprocal space. Surface scientists were formed in this mode of research. They resisted the change, feeling that newcomers would enter their field without the kind of abstraction that had hitherto been key to access to the field. Working in ordinary space was too easy ! Of course it has not actually become easy because the instrument has brought its own problems, and there are still people working with diffraction and in reciprocal space. The two complement each other.
AH : So this is the background against which your decision has to be seen. You went out on a limb.
HA : Yes. The beginning of my lab coincided with the first commercial scanning probe microscopes (SPMs). We had to grasp the opportunity.
AH : How did you know about the AFM ? Was it a very visible instrument at the time ?
HA : No. I knew about it from publications, but in order to actually see an instrument, I had to travel to California - although I guess I could have seen one at IBM Zurich. There was an STM at Marseille, because two physicists there (Salvan and Humbert) had worked at IBM Zurich, and they had brought one back with them. But they had no experience with the AFM. So I went to the US and visited the very few labs with AFM, both academic labs and the start-up companies of PSI [Park Scientific Instruments] and DI [Digital Instruments]. At Stanford University I met Calvin Quate and at UC Santa Barbara I met Paul Hansma.
AH : Was there a relationship between Paul Hansma and DI ?
HA : I don’t remember. But at any rate it was not as close as the one between Quate and Park. I think Park was a former student of Quate’s.
AH : So you purchased an AFM from Park. What about the other kinds of instrumentation you purchased for your lab ?
HA : Yes, we had to get other instruments, partly because it took a long time for the AFM to arrive. I had to go to the US to compare the DI and the Park instruments, and I discussed it with the physicists and chemists in our lab before ordering, and then we had to wait for the delivery – maybe 4 months or so. We got a 40% discount, because we were the first French customers, and they hoped that we would open the French market for them. I had very good discussions with Quate, and I think he trusted me to be a good advertisement for him in France. I think we paid 400,000 French Francs, so that the catalogue price was in the order of 800,000 French Francs [approximately US$100,000].
We bought also an infra-red spectrometer, in order to study molecular grafting on oxides. This we used as a complement to the AFM. And as I said in a previous part of the interview, our approach was to combine the traditional surface science (very clean surfaces) with “true surfaces” interacting with the environment. The infra-red spectrometer, XPS (X-ray Photoelectron Spectroscopy), and LEED (Low-Energy Electron Diffraction) were good tools for the traditional surface science approach working in UHV Ultra-High Vacuum). And also HR-EELS (High-Resolution Electron Energy Loss Spectroscopy). Our choice was risky, but it turned out to be correct. Our decision to build bridges between the two approaches was taken in 1992 or 1993. Quite early on in our project we built a surface force apparatus (SFA). It is not at all an AFM – there is no concept of high resolution, but it is similar in that you can get a direct measurement of the force interacting between two objects only a few Ångstroms apart. The idea is to make the measurement quantitative in order to study whether the interaction is due to van der Waals or electrostatic forces. In fact this project took six years – not for technical reasons but simply because we had to get the right people.
AH : Each instrument had its strengths and weaknesses in terms of resolution and the scale of the surface analyzed. And each instrument required special skills. The AFM, for example, requires quite some expertise to disentangle signal from instrumental artifacts, right ?
HA : Yes, artefacts were a real concern at the beginning, when we all had very limited experience. We had to pay much attention in order to ascertain the results.
AH : Can you explain how one separates signal from artefact ?
HA : There are different kinds of artefact. One that now seems quite natural but was hard to understand then is the tip effect. If the surface under examination has sharper topographic features than the tip, then the tip will be imaged rather than the surface. We had trouble with this kind of artefact. In fact, when studying tin oxide deposits on some substrate we got very nice images that we at first interpreted as small crystals having the similar orientation. We were very excited to find a growth mechanism of specific orientations on isotropic surfaces such as glass. I decided to present this result at a small meeting in Davos, Switzerland. The topic there was in fact “The AFM for Technological Applications”. Famous scientists attended, including Calvin Quate, Jim Gimzewski, and Heinrich Rohrer. There were only some 10 people there, because this was very early, maybe 1991. The night before my presentation, I began to wonder that the result was really too beautiful to be true. I telephoned my lab and asked people there to turn the sample by some angle and do the experiment again. That way, the features should have changed if they belonged to the surface. But they did not, and so we knew that the features belonged to the tip. So I did not present that particular slide in my talk.
AH : So rotating the sample by some degree is one way of identifying artefacts.
HA : Yes, that will eliminate this kind of artefact, the tip effect. There are also adhesion artefacts, some of which have been solved in the meantime thanks to new recording techniques such as the tapping mode.
AH : Digital Instruments has a patent for the tapping mode, right ?
HA : Yes.
AH : So the Park instrument that you bought did not have the tapping mode ?
HA : No it did not. The tapping mode did not become available until 1993 or so. Later on, Park Scientific Instruments did do something similar, but they may not call it tapping mode. The DI patent covers the name. And in the straightforward contact mode many artefacts were possible ; for example when looking at soft materials and polymers surface scratches easily occur. If you do that you image the substrate only. One way to identify this effect is to scan again with a smaller tip-surface interaction. In some cases you will find miniature small squares where the surface had been damaged in the course of the first scan. Some artefacts are very common, others are quite specific and harder to identify.
AH : In what you have explained, the identification of artefacts is internal to the instrument itself. It is not that you can go and compare the results of an AFM scan with those from a different instrument ?
HA : You can change the tip, and you should identify artefacts unless you are very unlucky to get the same tip. Everybody understood that the AFM has great potential not just as an imaging instrument but also to measure adhesion, hardness and so on.
AH : Using force-distance curves ?
HA : Yes, force-distance curves. This turned out to be very useful for us. For instance in order to understand the electrostatic interaction between oxide and a silicon nitride tip under water. This was original work. For example, in polymer adhesion we checked if it stayed on the substrate and what scratching would do. Of course such ideas were floating around at the time.
AH : Were you important to the subsequent spread of the AFM in France ?
HA : Yes, people came to our lab. Another lab, at the Institut Curie, that got an AFM at almost the same time. For a while we were a small community but then gradually we grew larger and larger. Yes, we were the pioneers. It was exciting.
2001-02-20 :
HA : I went with one of my sons who was 11 years old at the time to see Park Scientific Instruments. There were no more than 10 people working there, in fact I think it was more like three. It was very small and familial. We discussed and had tea. I enjoyed discussing with these people. It was nothing like an established company.
AH : What did it look like ? Did they work out of a garage ?
HA : Something like between a home and a garage. It was a small house. Even Digital Instruments started out like this. Already in those days DI, and especially Virgil Elings, was much more commercially aggressive, but they were very small too at the time.
AH : Did you stay in touch with some of these guys ?
HA : I stayed in touch with Calvin Quate for five or six years, until 1996. After that I lost the contact but he will probably remember me because we had many discussions. It was curious to see his impact upon materials science. In fact it was very difficult for him to get the first paper on the AFM accepted in Physical Review Letters. Some of it was considered just a pure mechanical profilometer. It had good resolution but it was not really anything new. His project now is very interesting from what I can tell reading his articles in the scientific journals. And he really is a very nice person. Maybe the last time I saw him is when I invited him to give a talk at Saint-Gobain Recherche.
AH : So you stayed in touch with him in the early 1990s, while you were developing your own AFM. I guess the use of the AFM changed the project from what you had originally envisaged ? Did you continue using all the other tools or did you focus exclusively on the AFM ?
HA : We used the other tools.
AH : What did you buy for your laboratory ?
HA : Infrared spectrometer, XPS, HR-EELS (High Resolution Electron Energy Loss Spectrometer), LEED. Quite quickly we had three AFMs. I wanted to develop a PSTM working in the infrared but unfortunately that particular project died because the physicist we had working on it left.
AH : What journals show the history of these instruments best ?
HA : In the beginning it was mainly in the general physics journals such as Applied Physics, Applied Physics Letters, Physical Review Letters, Surface Science. Now there are specialized journals. A journal like Journal of Scientific Instruments is not so important in this respect. Langmuir is also important for soft matter.
AH : Do any of these journals have review articles ?
HA : I am almost sure that all of them do.
AH : We were talking about the various instruments you had in your lab. How did you apply them to your research project ?
HA : The idea was to have two parallel approaches. We were mainly interested in adhesion, molecular grafting and so on. One approach is the classical view of surface science, the ideal surface approach. The other is the REAL surface approach, taking the environment as a part of the system.
AH : Not working in Ultra-High Vacuum (UHV) ?
HA : Yes. But we were trying to make the two approaches meet.
AH : So when you started working with the AFM in UHV, the point was to simplify the experiment ?
HA : Yes.
Figure 4. UHV Chamber et AFM in UHV Chamber
If a probe were to be introduced directly into the UHV chamber, it would take days of pumping to achieve UHV. Instead, it is first introduced into an antechamber, whereupon a vacuum is produced there. Only then can walls be opened without reducing the UHV too much. By pushing the rods labelled 1 and 2, the sample is transported in successive stages into the central chamber. Several instruments are attached to the chamber, including an XPS. On the right, an AFM can be discerned in the UHV chamber.
AH : How did the various instruments complement each other ?
HA : The spectrometers provided structural information. They give a chemical signature. One point of interest was silver on magnesium oxide. In order to have a simple model of glass we chose to study this problem within pure single crystal. We had the probe in situ in the same UHV chamber where we had the instruments to add the deposition techniques. In the case of silver it was just thermal evaporation. We wanted in situ real-time studies of the atoms arriving upon the substrate, the oxide surface. There were two models in this problem. One was that the atoms remain isolated or form small islands, so that the growth process is two-dimensional, so that you first get a perfect monolayer before a second layer is started upon. The other is that growth is three-dimensional with occasional collapses into flatness. To study this it is of course useful both to look directly and to use diffraction techniques. But in order to understand the process you need to grasp the interaction between the silver and the oxide. And only spectroscopic techniques will help here. We always tried to look at a problem from two differing points of view – in this case geometrical and chemical.
AH : You make it sound easy. You just use one tool and you get the topography, and then you use another and you get the chemical composition.
HA : Well of course it is not at all easy. It was very difficult because for instance, the STM works very well when you have a smooth surface but when you have corrugation it becomes much more difficult, because this corrugation interferes with the instrument. In spectroscopy you integrate over the size of the beam which is much larger than the surface scanned by the AFM. So you have to do many different experiments to see what effect the temperature has and so on. You also have to model the interaction. This was a little known problem. What is the mechanism of very small silver clusters on magnesium oxide with other silver clusters in the neighborhood ? It was a new problem. So it took time to understand the system.
AH : What is the measure of success ? It was partly CNRS, so you were under pressure to publish ?
HA : Yes.
AH : And since it was partly Saint-Gobain you had to get patents ?
HA : We had to do both. It was an interesting exercise in communication. In my position as head of the lab, I could not use the same words, the same way of presenting things when addressing different audiences. From time to time it was necessary to gather the scientific and the industrial people together under one roof.
AH : And what language did you speak then ?
HA : Fortunately everyone was happy with this lab, so it was not quite so difficult. The conditions were good. Nonetheless your question is quite correct. It was interesting.
AH : How did you convince Saint-Gobain that this would have a pay-off ? And how did you negotiate long- and short-term goals ?
HA : The short term was a problem. It was not straightforward to plan a new product for the company. The pay-off was very diffuse and difficult to identify. One way of motivating the directors was the argument that we trained very good PhD researchers for Saint-Gobain. And this was not expensive for Saint-Gobain, because they shared all the expenses with CNRS. Up until now this has not been a problem.
AH : Stanley Whittingham told me that in the last 15 years or so there has been a tremendous shift in company planning towards the short term in industry. Partly this was due to the MBA education and the fanning out of this new generation of business administrators into all nooks and crannies of industry. As a result the long-term disappeared, because everything had to fit into the financial year so that you have something to show to your shareholders.
HA : It is true that this has taken hold in industry. We had the good fortune that it was not very developed in Saint-Gobain. But also, the time required for the development of new glass materials just is acknowledged to be greater than that in electronics or informatics. When we start new projects, we are simply not able to show a product six months later. So we are less exposed than people in other fields, but the general development that you alluded to certainly has taken place. Maybe our situation will also change in the future. We may be excessive.
AH : Has the accountancy changed for you ? Did you have to write annual reports ? And has it changed over the last ten years ?
HA : In general ?
AH : Well, for the CNRS I can sort of imagine it. In academia you would specify the number of publications that you have produced and that is the measure. End of story. And that is very simple accountancy. But if you account to a company, keeping in mind the increasing influence of MBAs : did you have to account for your expenses in ever greater detail ?
HA : I do not think there has been such a change in the last decade.
AH : And do you write annual reports ?
HA : Bi-annual. But I am not in this lab anymore ; I left two years ago.
AH : Okay, so during the 1990s up until two years ago you wrote biannual reports to the company and in that period the structure of the reports did not change.
HA : That is correct.
AH : Did you have to specify just how much money you spent ?
HA : Yes, but also there, no change took place. And I always reported to the same person within Saint-Gobain. He was basically content with what we did, so it was never critical. It is true, it might have changed with a different person in charge.
AH : So, how did the instrumentation change throughout the 1990s ? The AFM became commercially available to an ever greater extent, you were able to buy many more things off the shelf. Is that true also of all the other instruments ?
HA : Yes, there are different aspects to your question. We used to build many instruments ourselves, and this was of great use for training. And this has changed. A reason the French PhD has been shortened is that equipment is being bought and not made in-house. That is a general trend. Science is changing as a result, because using a commercial instrument is not the same. When you develop an instrument yourself you know exactly how to get the result. In the specific case of AFM/STM : probably the AFM has been developed much more than the STM. In the STM the major breakthrough was with the driver and that was quite early. I think it was possible to purchase an STM driver already by 1992. Variable temperature was a little more difficult, but it was certainly available by 1994. Different ways of scanning and acquiring information were developed. Otherwise the evolution was purely technical : cheaper, and more diverse (such as an STM expressly for electrochemical research). By contrast the AFM has developed rapidly. Tapping mode and other modes where you measure not only the distance but also hardness, conductivity, adhesion, chemistry. It has become possible to map all these parameters. This explains why more and more people use the AFM.
AH : It has also become cheaper, right ?
HA : Yes.
AH : It has certainly become more user-friendly, adaptable to different circumstances.
HA : Yes. For the STM : there have very beautiful studies made of the coupling between tunneling and modulation. You might modulate the tunneling current with light for instance. You can even leverage the spin of the tunneling electrons. So you can do beautiful physics. But this contributes little to the democratization of the technique.
AH : I have the sense that Calvin Quate, by contrast, is working hard to increase throughput.
HA : Yes, that is right. There can be two reasons for doing that. To make the investigated part of the surface larger – of use in the semiconductor industry. And to shorten the time required for a scan. He is trying to use the system technologically.
AH : Okay. Two years ago you left your lab. Your own lab. Why ?
HA : I wanted to try something new and I was lucky to find someone who was well capable of taking over and for whom I have a lot of respect. He is from a different background. So now it is a different group. I became the Scientific Director of Saint-Gobain Recherche. There are two parts to the job ; one is to be the scientific manager, the other is to establish contacts in the outside world, and to promote innovations within the company, for instance with the marketing people.
AH : You were promoted ?
HA : Yes.
AH : And you have become slightly removed from lab work ?
HA : Yes, completely, I am now involved in organizational work.
AH : In fact, our project resembles your job in the sense that we stand back and look at the scientific research and try to gain a perspective ?
HA : Yes, you could say that.
Fin de l’enregistrement
Pour citer l’entretien :
« Entretien avec Hervé Arribart », par Arne Hessenbruch, 19 février, 29 mai et 20 février 2001, Sciences : histoire orale, /spip.php ?article47.