Interview: Daniel Pierce and Robert Celotta
1994 Gaede-Langmuir Award Recipients
Interviewed by Ted Madey, October 25, 1994
MADEY: Hello. I'm Ted Madey, a member of the American Vacuum Society. Today's interview is part of the American Vacuum Society historical archives series. I'll be talking with two scientists: Dr. Daniel Pierce and Dr. Robert Celotta of the National Institute of Standards and Technology. Today is October 25, 1994. We're at the American Vacuum Society Symposium in Denver, Colorado. I'll be talking with Dr. Celotta and Pierce, as I mentioned, who will be receiving the Society's Gaede-Langmuir Award. They've been cited for innovative development of advanced spin-polarized electron beam technology and their contributions to atomic, surface, and microstructural physics.
I'd like to begin by asking Bob and Dan some of the background about the development of the spin-polarized electron source and detector, and the use of spin-polarized electrons in science and technology. Bob, how did you come to develop an interest in spin-polarized electrons?
CELOTTA: Actually, it goes pretty far back to when I was in graduate school. The group that I worked with had an interest in polarized electrons. They were an atomic physics group, and they were interested in figuring out what happened when an electron that had, say, spin up, collided with an atom that had a spin in either the up or down direction. But they were frustrated by the fact that there were no good sources of polarized electrons at the time, so the experiments were very, very difficult, and they could only do the most rudimentary experiments. I was not involved in any of those, but I kept that interest, and it seemed like a real challenge to me for the future that somebody would develop a way of doing that, and then one could go back and do those important experiments.
MADEY: How about you, Dan?
PIERCE: I was first introduced to spin-polarized electrons when I went to the ETH in Zurich. I went there as a post-doc, which actually extended into three and a half years. And there, with Hans-Christoph Siegmann and others in his group, we did spin-polarized photoemission measurements. That was the first application or the first use of spin-polarized electrons to study surface magnetism.
MADEY: While you were with Hans-Christoph Siegmann, you worked on the development of a spin-polarized source, isn't that correct?
PIERCE: That's right. Actually, we were motivated to do it when Wolfgang Panofsky came over from SLAC. They wanted one for the high-energy experiments. One of the materials that had given very high spin polarization at that time was europium oxide. Ed Garwin from SLAC came over with a pulse source, and we did make measurements; we got nice, high-polarization pulses of electrons, but not really a very high current. The gallium arsenide source came somewhat later. Initially, the idea that you might get the polarized electrons from a semiconductor was stimulated by Siegmann after he went to a meeting and heard that people had optically pumped electrons in silicon. Then at some point, through various discussions, I think even Bill Spicer was involved one time, and, when we walked through the Alps, we talked about gallium arsenide and realized gallium arsenide was a natural material to do it. And Siegmann and I and Felix Meier and Peter Zurcher, who was a student at the time, did the experiments that showed you can actually get polarized electrons out of negative electron affinity treated gallium arsenide.
MADEY: And meanwhile, Bob, back at the Bureau of Standards, I guess before you even knew of Dan and his work and the work in Zurich, you yourself were involved in some beginning studies involving spin-polarized electrons.
CELOTTA: That's right, Ted. At that time, Ward Plummer was at the Bureau of Standards and in the same group as I. We were planning to do field emission to produce a source of spin-polarized electrons. The idea was that we would produce the spin-polarized electrons. Ward wanted to use them for surface experiments, and I was interested in doing some of those electron-atom scattering experiments. Soon after we had set these plans in motion, but before we actually got started, Ward left what was then NBS to go to the University of Pennsylvania. In replacing Ward and in finding someone to take Ward's slot, we were fortunate to find that Dan Pierce was available. He was coming back to the United States from Switzerland. The combination of my interest in polarized electrons and his interest in polarized electrons made a natural combination.
MADEY: I see. But when you came to the Bureau of Standards, what year was that, Dan?
PIERCE: It was early 1975.
MADEY: What were the first things that you did together involving spin-polarized electrons?
PIERCE: I think the first thing we set out to do-- We didn't have the funds to build up a gallium arsenide polarized electron source at the start. We started with the idea of, well, we'll make some trials by scattering an electron beam from a tungsten single-crystal, a high atomic number element that would polarize the beam, and we would be able to test and see if it was polarized by a second scattering from a tungsten crystal. So, a double scattering experiment. Fortunately, sometime in a year or so - maybe even less than that - after we ran into that, we got some funding both from within NBS, when Art McCoubrey gave us some money - he was the Director of the Institute of Basic Standards at the time - and then we got some Office of Naval of Research money.
MADEY: But those first experiments were sufficiently promising that they encouraged you to go on and develop a source?
PIERCE: Actually, we never really got into them that far because we got the money to build a source, and we went directly to that because that's clearly so much better.
CELOTTA: Just changed directions.
MADEY: I see. Well, what are the essential elements of the source?
CELOTTA: Basically, the idea is to do photoemission from the gallium arsenide semiconductor and to use circularly polarized light so that the angular momentum of the light gets translated into a spin orientation in the electrons inside the source when you excite them to the conduction band. But if you didn't do anything else, the electrons would merely be polarized in the conduction band, and they'd re-radiate the light out, and that would be the end of it. But it turns out that it's possible with gallium arsenide, by putting cesium and oxygen on the surface, to have those electrons higher up in energy than the vacuum level so they can actually diffuse toward the surface and come out of the surface with great efficiency. So you end up with a source of electrons where the incident light produces the electrons; the intensity of the light produces the number of electrons, and the polarization of the light determines their spin direction. The key factor there is that you can use this source to make an electron gun, which has an absolutely constant current just as high as a conventional electron gun, but with the spin direction of the electrons modulated in time according to the modulation of the light you apply. And this is very important because, when you use those electrons in an experiment, the spin affects the fact that something different will happen if the spin is up or down; might be a very small part of the interaction. The ability to modulate that spin direction without changing anything else is key to being able to measure those small effects. So a gallium arsenide source was a real breakthrough in that it both produced copious amounts of electrons and also allowed you to modulate the spin direction without affecting anything else.
MADEY: And there was nothing else in the world that could do this at that time.
PIERCE: It's one of those rare times, I think, when nature works for you. Gallium arsenide treated with cesium and oxygen to get negative electron affinity is the world's best photo-emitter. It's used with photomultiplier tubes and these other applications. And not only that, it gives you a high degree of spin polarization when you excite the electrons with circularly polarized light. So it's just perfect.
MADEY: Once you had built the source and had this fine source of spin-polarized electrons, what were some of the first things that you did with them?
PIERCE: We started off looking at scattering electrons in a LEED experiment. Basically, we replaced a Varian electron gun in a Varian LEED system with a Faraday cup and put in the polarized source, as Bob just mentioned. You know, modulating the spin polarization. We looked at scattering from tungsten, and we compared, in fact, our scattering measurements, where you look at the asymmetry in the scattering of polarized electrons, with measurements that had been made by the group at Rice University, where they came in with an unpolarized beam and measured the spin polarization of scattered electrons. We showed that those are equivalent ways to get the information. However, the information that you get in spin-polarized LEED, while it's an additional bit of information - in principle, it can help you determine structure and all that - you still have to go back to compare with calculations. These are not trivial calculations; you have to do a full relativistic calculation in order to get the spin-orbit term in there that is causing this spin dependence. It became clear that the real application was in scattering from magnetic surfaces, or in measurements on magnet surfaces, and using it as a probe of surface magnetism. That's kind of where we went.
MADEY: You moved in that direction.
PIERCE: After maybe a year or so.
MADEY: What was the first thing that you did in the area of surface magnetism using the spin-polarized electrons there?
CELOTTA: Well, the first thing was to see if we could actually see an effect, which we could characterize as surface magnetism. We did an experiment with Sam Bader and Gian Felcher from Argonne National Laboratory where we basically made a magnetic surface. It was a nickel surface, a crystal surface that was magnetized at the end of a horseshoe magnet, and that produced a magnetization lying inside the surface. We scattered polarized electrons from that surface, with the electrons either being parallel or antiparallel to the surface magnetization. We knew there would be some difference in the scattering when the spins were parallel and antiparallel. We weren't quite sure how big and how well we would be able to see it, but we could see it readily in the scattered current, and that was a great success. We were very excited because we could reverse the magnetization of the material and see a hysteresis curve in the modulated signal and the scattering. So we saw magnetism, and we knew it was surface magnetism because the mean free path of the electrons were such that we were really only sensing the surface layer. Then, after we knew it was surface magnetism, the next thing we did was-- the key thing in magnetism is changing the temperature, so we measured the temperature dependence of this magnetization we were measuring and found that, indeed, it varied, and it varied differently from what you would see in the bulk. So we then knew we were really measuring a surface magnetic effect, and it was different from the bulk. This opened up a whole range of things to look at, and basically to see how surface magnetism differed from bulk magnetism.
There's an interesting parallel. We learned so much about bulk magnetism by using neutrons, and neutrons are very weakly interacting. They go right through things, and they interact with the bulk of the material. Electrons are strongly interacting, and they interact right at the surface. We were going to use polarized electrons to study surface magnetism, an analogy to the way people use neutrons to study bulk magnetism. So this was the start of that.
MADEY: I see. Now, you had told us how you developed the source for spin-polarized electrons. Could you tell us a little bit about how you detect spin-polarized electrons? How can you tell what the spins are?
CELOTTA: Can I interject before we do?
MADEY: Yes, sure.
CELOTTA: There's one other kind of neat experiment that was nice using the polarized electron source. Somewhere in there, people started doing inverse photoemission to look at the electronic structure of unfilled states. In fact, Neville Smith and Peter Johnson - Peter was a post-doc and Neville was at Bell Labs at the time - did some of the first inverse photoemission experiments. These were angle-resolved, or k-resolved, so there was k-resolved inverse photoemission spectroscopy. That was Neville Smith's KRIPES right there. So we talked with them and said, "Well, look, we can do that spin resolved too because we have just as good an electron gun as you're using, and our electron gun has spin-polarized electrons." They brought down one of the little Dose type of Geiger-Mueller detectors, and we put that in our apparatus, and we did spin-resolved inverse photoemission for nickel. Clearly, you could see the large peak just above the Fermi level of the minority-spin electrons and absence of that peak for majority-spin electrons. So it's a really nice way now to get at spin-dependent electronic structure.
MADEY: As I recall, that resulted in the Phys Rev Letter as well.
PIERCE: Another one, right.
MADEY: Yeah, another one. That's right. [Chuckles]
PIERCE: So you asked about detectors?
MADEY: About detectors, yeah. Maybe you could tell a little bit about the Mott detector and then what you've done in order to simplify and move to lower energy and smaller size.
PIERCE: Let me start telling about Mott detectors, since that's what I started out working with, and then maybe Bob can tell about getting smaller. The traditional Mott detector operates at, say, 100 kilovolts or so. You accelerate an electron beam into a thin gold foil. There's a slight asymmetry in the scattering to the left or to the right of these electrons due to the spin-orbit interaction. There's a couple things about this. First, if you're accelerating something to 100 kilovolts, typically these things were mounted on big insulators inside a cage maybe ten feet by ten feet by the height of the room, so they weren't real convenient. Mott detectors are also rather inefficient, and most of the electrons go right through the foil. The scattering asymmetry is quite small. Since then, there have been some improvements. Maybe Bob could say a little bit about what we did.
CELOTTA: Just to go back one further step. Detectors, as well as sources, were very inefficient, and it was our idea that if we could make an improvement in both sources and detectors, there were a lot of experiments that polarized electrons could be used for. So we set about to try to improve detectors as well, and one of the problems, as Dan points out, was the size. We discovered that you could use a very low-energy scattering, from still a high Z target, to get a slightly smaller effect. The degree of filtering the electrons to the right and to the left was slightly smaller, but you could collect a larger solid angle. When the day was over, you got a much more efficient polarization detection. Plus, you ended up with a very small detector, which was critical. The reason we had developed it was we wanted to do polarized electron microscopy. We wanted to be able to look at magnetic domains in polarized electron scattering in a scanning electron microscope, and it was just unthinkable to put this huge Mott detector on the system. So we set about with John Unguris to develop a detector that could be capable of being mounted on an electron microscope, and then further to use that technique for observing magnetic domains.
MADEY: And that technique, what is the acronym or the name?
CELOTTA: Well, we called it SEMPA, or scanning electron microscopy with polarization analysis. We didn't call it polarized electron microscopy because it might be conceivable that you could take a polarized electron beam and scatter it from a magnetic surface and get the same sort of results. It's just that at very high energies, which were used in the microscopes then, that interaction wouldn't be very large, so we did it the other way.
MADEY: What are the sorts of developments that have occurred since you completed your work with the detectors? And maybe you can tell us a little bit about some of the things that have been done with SEMPA as well as other experiments that you've done since.
PIERCE: Our particular detector, this one that Bob described, where we're scattering low-energy, diffused scattering from evaporated gold film - a small, compact detector. We built one of those for the U5 beam line at NSLS (that's in Brookhaven) as part of an MRG to establish a spin-polarized photoemission beam line, and that's been used extensively by Peter Johnson to do spin-polarized photoemission studies, ultraviolet photoemission studies of the spin-dependent band structure. He's mapped out the dispersion of surface states. He's looked at absorbates, seeing how absorbates hybridize with magnetic states. Actually, you can get some spin moment on things like oxygen or sulfur because of its hybridization. They've also looked at core levels, outer 3p levels, of iron, for example. When they oxidize it, there are chemical shifts in those levels, and he looked at the different spin dependence. The different chemically shifted lines were like a spin-polarized marker, if you will. So they can actually begin to tell the difference between a gamma-iron oxide, which is ferromagnetic, and alpha-Fe2O3, which is antiferromagnetic. It's been a very rich source of experiments there in Brookhaven. Our involvement has only been reading the Phys Rev Letters, [chuckles]. They have the analyzer; yes, and then when something goes wrong with the analyzer, we get a question. Bob got a call the other day. "Hey, Bob, where did you get those capacitors?"
MADEY: Seven or eight years ago?
PIERCE: Seven or eight years ago. That's right.
CELOTTA: Actually, the original idea, that if you developed much more efficient sources and detectors of polarized electrons, there were numerous experiments to do, really has turned out to be correct. All over the world, there are many groups using polarized electron technology now to study surface magnetism, although I shouldn't leave out Charlie Sinclair and Ed Garwin, who developed the high-energy analog of this source and did a very important experiment to confirm the standard model in high-energy physics at SLAC a number of years ago. And in atomic physics, it's used for polarized electron, polarized atom interactions. In surface physics, it's not only used for polarized LEED to do either spin-orbit interactions or magnetic interactions, there is spin-polarized electron energy loss spectroscopy, with polarized electrons going in and polarization detection coming out, and polarized inverse photoemission; a large number of people are doing spin-polarized photoemission. There's a growing number of SEMPA apparatuses around the world studying magnetic surfaces, either two-dimensional magnetism or domain structure on surfaces, or trying to use these techniques to develop better magnetic recording materials or better magnetic materials in general. The SEMPA allowed one to go maybe two orders of magnitude past optical techniques, and since everybody's trying to make smaller and smaller magnetic devices, and magnetic recording devices, and magnetic storage devices, this turns out to be fairly important to people. So there's a lot of activity there, too.
In general, we're very pleased that this road we started on 20 years ago of developing the sources and detectors, and then doing the follow-on experiments, has really branched out. There's a large community of people working in this area. I want to be sure to say that there have been other people developing detectors and modifying sources as well all over the world, and they're too numerous to mention altogether. But for us, it's been just a great adventure. It's turned out to be constantly growing.
MADEY: It certainly sounds as if it has. Do you want to add to that?
PIERCE: I don't want you to get the wrong impression. There are still a lot of challenges left.
MADEY: Yes?
PIERCE: Whereas the gallium arsenide source was a breakthrough that was orders of magnitude better than any source up to that time, about the best we've been able to do on spin analyzers is factors of two or three or so better. We've gotten a lot smaller, but the efficiency is still very limited. Compared to making an intensity measurement, you have to count 103 to 104 times longer. So there's a lot to be done. Still, you'd like to have a spin polarization analyzer that analyzed the polarization of electrons like you analyze the polarization of light, and we're a long ways from that.
MADEY: Could you comment a little bit more about the future? About, for instance, the impact on atomic physics and nanometer technology of some of the developments that you've been discussing today?
CELOTTA: Well, that's sort of interesting because we were using the spin-polarized electrons to study electron-atom collisions, and there was a very curious spin-off that occurred. We used the polarized electrons to scatter from polarized atoms. Our prototypical scattering situation was a sodium atom with a one valence-electron spin, and an incident electron either spin up or spin down, and then we'd study all those interactions-- elastic collisions, inelastic collisions, superelastic collisions, where the electron gained energy. Because we could control everything - and even the energy, the angles, the spins, even the nuclear spin - instead of cross-sections, we could actually measure quantum phases and amplitudes. So it was quite a challenge to theory. We decided to challenge theory even more by going to a very high spin-density target. We chose chromium, where you could actually align six spins simultaneously. Sort of a large moment. And we were doing spin-polarized electron-chromium experiments. It then became apparent to us, and to others - people at Bell Labs and Harvard, Mara Prentiss at Harvard and Greg Timp at Bell Labs, as well as my colleague, Jabez McClelland - that you could use laser fields to focus atoms. It further turned out that chromium was the ideal atom for this. So having perfected all the technology and preparing atoms of chromium in selected states, we could now focus these atoms to make nanostructures on a target on a silicon surface, for example. As a consequence, there's a new area of research for nanostructure fabrication and precise wavelength measurements or length measurements on silicon wafers that might be used for lithography calibrations or scan-probe calibrations. The techniques we had to develop along the way are branching out in other interesting directions as well.
MADEY: So in effect, you have a laser beam shining on a target in some sort of a patterned array, and you simply evaporate the chromium atoms through this laser field and you get this focusing effect and patterned deposition?
CELOTTA: Correct. The laser beam skims in front of the target. It has a certain modulated light structure, and that allows us to focus atoms to the surface. We even have plans of maybe somehow making magnetic structures that way, which then we'll look at with some of these other techniques - the interaction of little magnetic dots. So there's a continuing flavor of magnetism and spin polarization in our lab.
PIERCE: If they get those little chromium dots and oxidize them so they're a magnetic oxide, a chromium, or… One of the nice ways to look at them, if we can perfect it, would be some spin-polarized, scanned-probe microscopy. There have been some preliminary measurements of spin-polarized scanning tunneling microscopy, but we'd like to be able to develop that - we can try a number of different ways - into a routine technique. You can think about ways where you might have a magnetic tip emitting polarized electrons into your sample. One thing you might think about doing, and we've written about and talked about before, is actually having a gallium arsenide tip be the inverse of the polarized electron source, where you have a magnetic surface here, and you tunnel polarized electrons from a magnetic surface into the gallium arsenide tip. Then it emits circularly polarized light. Just the opposite of the source, where you inject circularly polarized light and emit polarized electrons. That'd be a very nice way, if it works out and if it can be efficient enough, to have a tip that you could use on any magnetic surface. There are other optical methods. There has been some indication that by tunneling and exciting photons from a magnetic material, the photons might have helicity to them. There are a number of ways that we're pursuing. This may be one of the things of the future, if we can get down towards atomic resolution and getting spin-dependent or magnetic contrast.
MADEY: That really sounds exciting. Well, I think we've covered quite a lot of ground in our discussion today. Is there anything else that you wanted to add?
PIERCE: Maybe just to mention that, of course, Bob and I have been very fortunate, both to work at a place like NIST that supported us and encouraged us. We've managed to be able to do a lot of these things. We've also had a lot of really fine collaborators at NIST. Some of them are still are there, and some people were post-docs and have left, as well as these collaborators we mentioned from other laboratories. Of course, these are very hard measurements, and they don't get done by one or two people; they get done by a lot of people. We really benefit and enjoy, just on a personal basis, these interactions.
CELOTTA: Yes. I highly support that.
MADEY: Okay. Well, thank you very much, Bob and Dan, for taking the time to be with us today and to contribute to the AVS historical archives. It's been a very interesting discussion. Thank you.