Dr. Anders: Professor Parr, thank you very much for giving me this opportunity to talk to you. I have seen in the notes that you were borne in 1922. Is that right?

Professor Parr: No, you have in a new hand-out on the web 1923, it's also wrong. It's 1921.

A: And you were born in?

P: Chicago.

A: For me it's important to learn how you got to chemistry, how an American specifically got to chemistry and got to physical chemistry. So in Chicago you went through normal high school?

P: No, I lived in Chicago only two years and then my father took a job at American University in Beirut where he taught in the medical school. When we came back to the States I lived in succession in Iowa, back in Chicago for a short time, in Alabama, and then in Washington D. C. where I arrived at the age of eleven. I graduated from high school in Washington, then I went to Brown University and the University of Minnesota for a Ph.D.

A: Why did you choose chemistry?

P: I was always interested in sciences generally and mathematics in particular. And my father, who was an university professor, understood that. He knew that I liked chemistry and mathematics and teaching at George Washington University at that time he knew two important physicists who were there, Teller and Gamov who were there a short while before they went elsewhere. This is in the late 1930s. He knew them and had, I think, the good sense to put the question of can you combine mathematics with chemistry. I remember an interview with Professor Gibson, who was another one there - about what university to go to, where you could study mathematics and chemistry and/or physics. So I went to Brown University where I fell under the good atmosphere in the sciences - in physics, chemistry and mathematics.

A: And then you sort of - of course with your mathematical background - you leaned towards physical chemistry then?

P: Yes, the infrared spectroscopist Paul Cross was there and this being 1941/1942 and I was a senior in molecular spectroscopy. In physical chemistry in those days in the U.S., molecular spectroscopy was a major interest and just everybody that seemed to be important was a physical chemist. My undergraduate supervisor Paul Cross had as a graduate student Bryce Crawford, whom he sent me to do graduate work with at the University of Minnesota. Crawford had postdoced with Bright Wilson.

A: Was that the one of Wilson, Decius and Cross?

P: Yes. And so I spent my senior year getting into spectroscopy although I didn't do any. Then I went off to graduate school in Minnesota and started studying there. After a couple of years knowing that I was interested in theoretical work he, having noticed the important paper of Goeppert-Mayer and Sklar and being interested in vibrations of molecules like benzene, he said: why do you not proceed to compute the out of plane vibrational force constants for the benzene molecule. Here is Goeppert-Mayer and Sklar. And that's how I got into quantum chemistry as we now call it, theory of electronic structure.

A: And there were no computers at those days, there were just electromechanical calculators?

P: Yes, and ourselves.

A: That's very interesting. So your first approach was, like you said Goeppert-Mayer and Sklar. And your thesis you did on what?

P: Force constants for benzene.

A: And then you were postdocing?

P: Yes, for a short time only. In our work on benzene we discovered errors in integrals. This got us in contact with Robert Mulliken, Chicago, who was working on similar things and in particular with his student Roothaan. We had difficulties reaching Sklar. Goeppert-Mayer was available but she didn't have any first hand interest in such work now. We did get in touch with the student of Sklar, name of Lyddane, who was at a naval research lab. And I remember the fun of finding something like 20 errors in his published paper on integrals which was essentially a very, very long formula for a two-center hybrid integral involving two pi-orbitals. And what did we hear him say when se sent him the list of 20 errors, that something like 16 or 17 of them were typographical errors, but two or three were real errors and 'thank you very much'. And one of the amusing side light is that if Goeppert-Mayer and Sklar had used corrected integrals, the correct integrals, they would not have obtained a good agreement with the benzene electronic spectrum. It took incorrect integrals to get the correct experimental tendencies!

But this put me in touch with Mulliken. So soon after I was at my first academic job at Carnegie Institute of Technology (now Carnegie-Mellon University), I arranged to spend a summer working with Mulliken. This was very useful. Mulliken and I produced a couple of papers together, one on butadiene and one on resonance energy. We had thought that we had written a definitive theoretical paper on resonance energy, it happens that it was almost the last paper ever written on this subject. At Chicago that summer there were among others Roothaan to publish a paper with. So, Longuet-Higgins was there, also Platt, and then there was Rudenberg. So I was then in a contact with the whole American group working on electronic structure problems.

A: What were the specifics of work with Mulliken? You would go and do a part of the theory of the work concerned on the question, and then how would the interaction be with Mulliken? Because he is always said to be a sensible man but he was also withdrawn. So in scientific work how would you describe him?

P: Perhaps I got along with him well because I also perhaps have a tendency to be withdrawn. Our personalities went well together. I can't remember who initiated the studies - the butadiene with a very natural follow-up of our work on ethylene. I got into ethylene because doing benzene you had to do ethylene first. It is very natural for him to loosely supervise that, and also resonance energy was something he had been interested in. I did not talk to him very much but he certainly was very helpful, in one way or another. One problem I had with him was, when I was alone there in the summer time and he had his wife and family I couldn't make myself work longer in the office than he did. He was there Sunday mornings and all day, day after day after day.

A: Could that be the reason for one of his wife committed suicide, didn't she?

P: No, one of his daughters. She was a nice but mentally ill young girl. I think, there was a small affect on her from his not being willing to move away from the University of Chicago. The physical atmosphere, it was not then in a very good place in Chicago. And Mrs. Mulliken wanted him to move out, but he didn't want to do that and it had some affect, perhaps. Mrs. Mulliken was in her day known as a beauty, at the university campus as a young woman. She had a stroke when she was perhaps roughly 70, 75 years old. And he tended her with much love for several years after she had the stroke. Mrs. Mulliken used to give dinner parties for the whole group. Of the people I've known, they were as completely hospitable as anybody. Very friendly.

A: Coming back to the work around Mulliken that must have been tremendous fulminant and inspiring?

P: Yes, certainly. But his papers are hard to read. They are not written very well. They are not written in simple a,b,c ... so much detail. I can never match such detail.

A: He was extremely careful.

P: Yes. And he was a very good spectroscopist which he was in the first instance. He wouldn't slosh over small details. I never had the stomach for all the details. One summer when I was in college I worked in the National Bureau of Standards, atomic spectroscopy section, developing photographic plates all summer. Too much detail for my taste.

A: With the famous Charlotte Moore?

P: I don't know. I was in the group of a fellow named Meggers, a great spectroscopist at the back of all of that at the Bureau of Standards. But all of that detail! Electronic structure was easier in those days - - -

P: Of course getting to know all these people was very valuable I mentioned. Rudenberg (Rüdenberg) was a student, Roothaan was a student. I also met the people from Washington, the great German physicist Karl Herzfeld was I think quite important. But he was also into the theory of color. And the various meetings everybody attended there was and still is a Annual Ohio Spectroscopy Symposium. And so on.

A: So, after Mulliken how did you proceed then?

P: I worked with students and postdocs at Carnegie-Mellon University. My first student there was Russ Taylor, still a very good friend. He did beautiful work on configuration interaction; Then there were other good students there. I was more or less sensitive to what was going on in this narrow but wide open field of electronic structure. Probably associated with it color for instance. But I was well aware of all you need to compute these things correctly and we helped start calculating things better. Configuration interaction method, compute all integrals accurately. And having gotten to Carnegie-Mellon in 1948, already in 1949 or 1950 I noticed that David Craig with Coulson in London was beginning work on the spectra of aromatic substances and started talking about configuration interaction. And meanwhile Roothaan and I had had part of the integrals that had been left out by Goeppert-Mayer and Sklar and got the benzene case correctly, the correct integrals and good predictions. And Craig and I published this paper on "configuration interaction in benzene", where we took the pi-system and did essentially a complete configuration interaction calculation on it.

That has some trivial historical interest in that it was there that the word, the term ab initio was introduced. Craig and Ross had computed everything from the start in London and I had personally computed everything from start in Pittsburgh. Then we compared our answers when we were finished- This involved computing of all the integrals as best as they could be done and selecting the configurations to mix for the ground and exited states because there were electronic states that were of experimental interest and we checked our answers one against each other when we were finished. And what the paper says is, that these calculations were done ab initio by Craig and Ross and by me, independently. And Mulliken later said that this was the introduction of the term ab initio into quantum chemistry. In the short review that you have, I talk about this and reproduce a picture of a letter from Craig to me where he uses the term ab initio in a different context. So ab initio was introduced in the quantum chemistry by Craig in a letter to me and I put it into the manuscript. That's where ab initio came from. {pause}.

You have asked where the term configuration interaction came from. The answer is, it is carefully explicitly introduced by Hartree. Hartree or Hartree-Fock, that's the same. He did a calculation, I believe, on the Beryllium atom where there is obviously an important configuration interaction 1s²2p², 1s²2s² and carried through the selfconsistent calculation. That's really, I think, the first configuration interaction calculation.

I used the term in the paper of my first student, Taylor. But in this context one should mention the English theoretician Frank Boys. Frank Boys immediately in 1950 was declaring already in the Proceedings of the Royal Society in 1950 that if one does enough configuration interaction one will get the correct answer. I suspect that he would have referenced Hartree there and this definitely affected us in the selection of this problem here {Parr is looking through some of the reprints he had brought along. I.'s n.}.

The Parr, Craig and Ross paper involved a lot of integrals as I mentioned, but it was noticeable that the integrals fell into groups, of integrals that were very close to numerically the same each other. And in our paper there is a footnote, "one might for example assume constant values throughout each set and proceed semiempirically or one might seek atomic orbitals which render the approximate constancies exact". That is really the impetus for the discovery of the zero-differential overlap approximation which dates from the fall of 1951 in a little paper of mine how to estimate these things. The fall of 1951 was important also for the fact that there was a small conference in the U.S. organized by Mulliken at Shelter Island, to which I was kindly invited and at which I met there many important people: Slater, probably the most important, but also Kotani and many others, Lennard-Jones was very important for me, I have met him in 1951. The conference emphasized this problem of integrals. It was a nice mix of older and younger people. I had met Löwdin briefly before that but he was at this meeting, and many others who have remained in the field for many years.

There was Kimball, of Eyring, Walter and Kimball, who was talking about using Gaussians. He published very little but there are theses of his students where Gaussians were used. Boys was not present but it was very interesting and appealing to have a chance to meet all these others and many who have been friends for many years.

Then in the fall of - - -, well, in 1951 and 1952 then, came the Pariser-Parr method, which later became into the Pariser-Parr-Pople method.

Pariser had been a graduate student at Minnesota when I was a student. And I had the pleasure of having him in a class, I stayed at Minnesota for one year after I got my degree, while the chairman took off for Hawaii and I taught the graduate thermodynamics class. I would say, that experience certainly helped me a lot in my career because good training in classical thermodynamics greatly helps one enter density functional theory. Pariser was a student of mine in thermodynamics. He was an experimental spectroscopist, a student working in fluorescence and phosphorescence and these things, a very good experimentalist.

After receiving his Ph.D. Pariser had gone the work for the DuPont Company. And DuPont was interested in the color, at the time they had a connection with I.G. Farben, a German company, which produced dyes and so on. They asked him to work on dyes and do anything he wanted, I guess. He decided that experiment was not a very quick way to get much information about color and called me up, remembering that I had been a student in quantum chemistry at Minnesota and asked me if I would help him out. And we got together and hit it off. By the spring in 1952 we presented two papers at this Ohio State Spectroscopy Symposium and published these works in 1952 and 1953. Pariser later did some wonderful actual predictions of colors of organic molecules. He put off publication until 1956 probably because the DuPont wanted him to put it off. The calculations were on all the polyacenes, the work became known everywhere and that gave much pleasure to us.

I described all this in detail in a summary ten years ago in the International Journal of Quantum Chemistry, 37, 327 (1990). And I prefer to refer to that summary itself for my detailed description, which includes documentation of that whole story. It was really a wonderful experience working with Pariser, he is a fantastically good scientist who has a very good feel for theory and how to use it. We are very good friends and I think I have seen him many times over the year.

P: Back to the ?? story in 1952 the meanwhile, at Carnegie, I was working on making (??) the next theory better. A student, Peter Lykos, produced a nice thesis on various pi-electron approximations in the mathematical terms - it got us going in other directions. We were always aware that we ought to compute things better where we could.

Then in 1952 I decided I would take the benefits of fellowship opportunities and take off for England. I got the Guggenheim and Fulbright support to go to Cambridge (England), to Lennard-Jones's group. I had in mind not to work with Lennard-Jones, although he was certainly a fine scientist doing good work but with Frank Boys who was his lecturer and doing this beautiful work on ab initio theory. With my family (??) I arrived in England in August of 1953.

It seemed that Lennard-Jones had decided to become the chancellor of a new university in Keele, Staffordshire. Where incidentally he took McWeeny, McWeeny's first academic job was there. But Lennard-Jones was leaving Cambridge. He wrote me a nice letter saying, sorry I'm leaving, but you should be happy, you wanted to work with Boys anyway. So the amusing thing was I got to the group which was now a little bit shocked because the professor had gone and there was not a new professor. And I was assigned to an office and I found myself occupying Lennard-Jones's chair because Lennard-Jones had his big office and had his desk and chair and I got Lennard-Jones's chair, at the age of 32, pretty good (I shared it with John Pople).

And Boys was there, George Hall was there, Robert Nesbet was there. Andrew Hurley had just been there and his presence was in the place because of the beautiful work he had done. Griffith was there, Buckingham arrived in January. All of these people were wonderful to get to know, I talked with Frank Boys almost every day at lunch but never took to really doing anything with him. He was busy with the Cambridge EDSAC. Everything was done by tape, everything was done twice, everything was machine programmed and I was not made for that. So I really didn't to do anything with Boys except continue to appreciate him.

And I got to know John Pople well, we have been life long friends. I remember well that when he showed me a manuscript of a paper he had just sent off to the Transactions of the Faraday Society. I looked at this and I thought, my goodness, that's about all there is to it; we are not alone.

Of course it is the best thing in science if you can do something, which becomes common knowledge, that's the best thing you should want. Anyway I had this and I wrote to my professor, Bryce Crawford (who has been a live long a friend of mine). I remember saying to him, I'm a little sorry to discover that this young fellow over here is doing things that seem to be very similar to what Pariser and I did. They were doing the self consistent field while we were doing configuration interaction from the beginning with. configuration interactions. And we didn't care so much about the ground state. And Crawford wrote me back saying: "Remember Bob, science is made for men, not men for science."

Very good advice actually. Anyway, have never let that bother me all that much. I went on to the understanding of the pi-electron approximation and we did a quick UHF study going beyond the pi-electrons. In the work of Joe Parks on the formaldehyde molecule where we discovered that when you go to more sigma electrons with similar ?? you have errors that seem to be a little more annoying than if you don't do that.

And Pople soon came out with a simple all-electron version of the theory. I have never myself done a calculation with the Pariser-Parr-method. I've never been interested in applications for the sake of application but for understanding the methods and going to the next thing that needs to be understood.

A: So you continued to be interested in force constants?

P: Having come out of spectroscopy of course we are always interested in force constants and things like understanding force constants. Understanding forces automatically leads to rather basic principles of quantum mechanics, leads to understanding electron density, since electron density determines all the forces by the Hellmann-Feynman Theorem. So we were working along directions which always brought those things in mind. The work of Platt especially other people, such as Kuhn in Europe, indicated that there is a lot of true essence to free electron models for instance, something is right about them. I have always been interested in electron density.

The 1960s brought the birth of density-functional theory essentially in two stages. Firstly J.C. Slater got much involved developing what came to be known as the X_a-method. Which from the point of view of the strict theoretical, theoreticians, this molecular school oversimplified on the one hand - and not so understood by Slater which he would admit he didn't understanding exactly what he was doing - but because of the natural difficulties in understanding the X_a-exchange. But also it annoyed me, I'm sure others as well, because he presented the X_a-method as the answer to everything. Writing about it as if there did not exist a proper quantum chemistry of the Hartree-Fock method, he probably never gave a reference to Roothaan in his life which was ridiculous. It was annoying to listen to him because he was essentially insulting especially to Roothaan. So I had a negative a reaction to that but also there was the importance of forces. The work of the Roluff (??) on density and the structure of the water molecule of Richard Bader - needs to be some density in the middle to give you the nonlinear configuration.

I was always sensitive to that. But then in 1964 and 1965 came the announcement of exact density-functional theory by Kohn and Hohenberg and a method, called the Kohn-Sham-method, analogous to the Hartree-Fock method, from which all became more or less clear, gradually. This was a theory which had roots in the Thomas-Fermi-Dirac method of the 1920s.

And then I had a stroke of good luck, in about 1972 I was at Johns Hopkins at the time, and there was a student, a graduate student named Ho Jing Kim, who was working with a physical mechanics expert named Everett Thiele who unfortunately did not get tenure in the university and the students needed a professor. And I had a really good idea. I said why don't you start by studying the whole literature of the Thomas-Fermi method. So we decided to do something in that direction. First, Kim did a beautiful job of studying the whole Thomas - Fermi literature, I still have his bibliography, that I refer to often. The Thomas-Fermi theory is a first approximation in a density-functional theory. All you need is a density, if you know how to handle it properly. You don't need a wave function. That's how we got into density functional theory.

I was fortunate in 1974 already to have had a postdoctoral who understood this kind of stuff pretty well and he transferred from Johns Hopkins University to North Carolina with me - - this is Mel Levy, a fantastic theoretician. He helped a student and me at Hopkins uncover, prove - whatever you want to call it - an important theorem in wave function theory which is the generalized Koopmans theorem.

Everybody knows that you can compute an ionization potential approximately from the Hartree-Fock highest occupied eigenvalue. Not everybody knew then and maybe everybody doesn't know now - but they should - that if you have an accurate wave function you can compute by a generalization of the Hartree-Fock method the exact ionization potential. This is called the extended Koopmans theorem which later became the foundation for a very important theorem in density-functional theory that we discovered together, several years later, after he had already moved to Tulane University. This is that the highest eigenvalue of the Kohn-Sham-method (which like the Hartree-Fock method gives a single determinental wave function) is the exact ionization potential. At the time this was discovered it was not known, nobody knew what Khon-Sham eigenvalues meant.

We got more and more into density-functional theory. Firstly it was essentially obvious, but a lot of fun working out, that this highest eigenvalue was also the properly - - - the following is a slight simplification. It turned out that the highest eigenvalue is the exact chemical potential for the system, which is the tendency to gain or loose an electron in presence of another species. This is just what is called the electronegativity in the subject of chemistry.

And a lot was known about that. Mulliken's work in the early 1930s or Pariser's work. Pariser's brilliant contribution to the PPP theory was: we should take the one-center integral as the ionization potential minus the electron affinity. The I and A are terribly important. And the chemical potential turns out to be the average so it is the same thing as the electronegativity. So this is pulling into the density-functional theory the realization that the electronegativity concept here and chemical potential are the same thing.

Subsequently the chemical hardness concept which had been made famous by Ralph Pearson in inorganic chemistry, was easily identified as the second derivative in the density-functional theory and so is rightly then called the hardness. And perhaps the most interesting of all of these things is the site reactivity index, the quantity identified by Yang and called the Fukui function.. In the usual orbital theory you talk about highest occupied orbital, lowest unoccupied orbital and the electron densities in different places. It's obviously approximate in the Hartree-Fock theory that the density of the highest occupied orbital and lowest unoccupied orbital identify to and from where charge will flow. But these are each approximations to the exact quantity in density-functional theory which is the derivative with respect to the particle number of the electron density. So this is brought into the density-functional theory which is an exact theory. These are old, approximate concepts, they are systematized and improved, and there is a lot of literature in recent years. That is the cleverness of Ralph Pearson, then Levy and also Palke and Donnelly in the chemical potential case. Ralph Pearson and then Wei Tao Yang, the able young student of mine who for several years now has been a professor at Duke University. Yang authored with me a book on density-functional theory.

Now, this theory for how to work with the electron density is really quite hard to work out correctly because of the difficulty of density-functional analysis exists on some functional derivatives in the theory. We have been into many theoretical aspects in the theory. I think, Walter Kohn who, I'm fortunate, has been my friend for a long time, has been surprised how accurate this all has become, essentially in the hands of chemists. He is very appreciative.

Important for my personal development was the fortunate fact that I was a Carnegie-Mellon at the time when the physics department was very strong in solid state physics, and first of all the great physicist Frederick Seitz, Seitz's great book on solid state theory and then that two or three years after that Kohn arrived and one could not be in that environment with opened eyes and ears but to realize that the electronic structure as viewed by solid state physics and as viewed by chemists really are the same subject. Advances on the one side are apt to help advances on the other. I have taken advantage of that. We are still working in DFT.

A: After your life long interest, how do you view electron density now?

P: You have asked, what's my picture of electron density. I can talk about my picture and wouldn't want to talk about other people's picture, that's their business what ... {laughing}. Mine is enough to get a hold of. I like to think that perhaps I have a little more feeling for the physical or chemical side of some of these things that some of the computational chemists who - - - . Electron density is electron density. It sticks there through all levels of orbital theory. I mean, you could include as complicated a wave function as you could imagine to get a very accurate answer. And you wouldn't be able to understand that in any helpful terms.

But the electron density is electron density which changes very little from approximation to approximation. A remark that should be made, and the case is strong, that the Kohn-Sham method is closer to all of the correct physics than the Hartree-Fock method is. I have already mentioned one aspect, the highest orbital energy in Kohn-Sham is the exact {pronounced with emphasis} ionization potential, not the approximate ionization potential. The Hartree-Fock operator is non-local, the Kohn-Sham is local, like the Hamiltonian operator. Another one is - believe it or not - every orbital in Hartree-Fock theory has got the same long range tail. Even a one s-orbital has the same tail as a 6f-orbital in some heavy atom. In Kohn-Sham theory the orbitals have their own tails. The difference of excitation energies is more cleanly, or maybe not more cleanly, but more accurately connected to the excitation energy in Kohn-Sham than that in Hartree-Fock. Hartree-Fock is really quite non-physical compared to Kohn-Sham and it can be argued that the Hartree-Fock determinant should be thrown away in favor of the Kohn-Sham determinant as the prime tool of the description in chemistry (as it already is in physics). The quantum mechanics of electrons in ?? is a fascinating subject.

A: Professor Parr, thank you very much for this interview in this highly pleasant surrounding of Menton.