1: Early Work at Shell (1952-1956) and CIT (1956-1964)
Molecular electronic structure plays a central role in chemistry and in the thinking of chemists. As a graduate student at the California Institute of Technology (CIT, 1947-1950) and a postdoc in Mulliken’s group in Chicago (1950-1952), I found ongoing discussions of the electronic structures of molecules to be pretty unconvincing and discouraging. There were lots of theories, some derived from valence bond theory, and others derived from molecular orbital theory, but there were very few experimental methods to test these theories critically. In fact there was a significant group of students and mature scientists who were openly skeptical of much of the work in this field. (See below for George Pake’s comment about driving the last nail in the coffin of theoretical chemistry.) Ab initio calculations of electronic structures of molecules and their properties were tedious, slow, and not accurate enough for most chemical problems. Even so, my interest in this subject remained, since it was at the center of chemistry. Also I was encouraged to continue in this field by some of my own unpublished work on fine structure splitting in diatomic molecules and solvent effects on the spectra of cyanine dyes as well as my graduate work on “optical interaction absorption.” However, I had no grand plan for a research program and the prospects for an academic job were not good. In 1952 I was most fortunate to obtain a job in the spectroscopy department of Shell Development Company, in Emeryville California.
The job at Shell was fortuitous in several respects. First, the Chairman of the Department, R. Robert Brattain, was an enlightened supervisor. His brother was a co-inventor of the transistor, and Brattain himself had some involvement in the determination of the structure of penicillin. Secondly, Shell had bought one of the first Varian NMR spectrometers.
My first job assignments at Shell were relatively routine (47). However the operator of the NMR spectrometer soon ran into trouble in interpreting some NMR spectra; I was asked to interpret the NMR spectrum of CH2=CF2. At the time I knew virtually nothing about NMR, but was able to quickly familiarize myself with the subject and provided an interpretation of the spectrum (and developed a useful but non-elegant version of group theory for NMR spectra (17). See also Carrington and McLachlan, pp 47-50. Except for some brief work on the spectroscopy of Kuwaiti crude oil (47), I now had time to work full time on magnetic resonance problems, doubtless much more time for research than if I were in an academic institution. My research included an attempt to relate molecular electronic structure (molecular orbital theory) to the nuclear spin-spin splitting seen in high-resolution NMR spectra (16).
A second fortuitous event occurred due to a brief visit to Shell by Professor George Pake, then of Stanford University. He wanted to see the new Varian NMR instrument, and I just happened to be near the instrument at the time of his visit. During the course of his short visit Pake made the off hand remark, “the last nail is being driven into the coffin of theoretical chemistry.”
This statement of course greatly sparked my interest and I pressed the issue with him. Proton nuclear hyperfine splitting was being observed in the paramagnetic resonance spectra of aromatic free radicals. In these radicals the “odd electron” and its spin were generally believed to be confined to pi molecular orbitals. These pi orbitals have a node in the molecular plane. The protons attached to the aromatic carbon atoms also lie in the molecular plane of the aromatic molecule. The observed isotropic proton hyperfine splitting is due to the Fermi contact hyperfine interaction, which requires a finite electron spin density at the position of the proton. Therefore there appeared to be a fundamental discrepancy between the experimental results and the expectations of “theoretical chemistry.” I saw the flaw in this argument that neglects electron spin correlations. I went home that evening and wrote a paper accounting for the proton hyperfine splitting in these aromatic free radicals (20). See also Carrington and McLachlan, 1967, pp 81-83, 103-110.
Two things struck me about my calculations of indirect proton hyperfine splitting in aromatic free radicals. First, the mathematical integrals involved were largely local, and were linear in the theory. This implied to me that there should be an approximate, simple, linear relationship between the splitting constant and the pi electron spin density on the carbon atom to which the in-plane proton is bound (21). One of my first graduate students at CIT and I worked out the theoretical details of this conjecture (40). There is now ample experimental evidence in support of this proportionality. (See (478) and Carrington and McLachlan, p83 for leading references.) One of the surprising consequences found in applying this relationship to a wide variety of aromatic free radicals is the accuracy of elementary molecular orbital theory for calculating spin densities on unsaturated carbon atoms. One peculiarity is found in the case of odd-alternate aromatic radicals where molecular orbital theory yields a zero spin density on certain carbon atoms, whereas valence bond theory generally yields a small negative spin density on these atoms.
The second unique theoretical result was the predicted negative spin density at the proton. I was very anxious to prove this experimentally, which I thought might be done by creating an oriented malonic acid free radical in crystalline malonic acid by radiation damage.
As I recall, one of my students thought that was a dumb idea, so that sent me to the stockroom to get the malonic acid myself, and to begin making crystals. The experiment worked beautifully. The well-oriented malonic acid free radical was the dominant free radical produced by radiation damage. The paper finally proving the negative spin density at the proton is (61). See also Carrington and McLachlan, pp 103-106. The work takes advantage of the known sign and magnitude of the dipolar interaction between an electron and a proton, and the fixed orientation of the radical in the crystal. This was the first time that a negative spin density in a free radical was demonstrated. All the students involved with the experimental instrumentation were great, including Terry Cole making microwave cavities, and Dick Fessenden who managed the K-band experiments after much trouble with the vendor of the K-band instrumentation. Chonon Heller was involved with the crystallography.
Much of the early experimental work on the paramagnetic resonance spectra of aromatic radicals in solution came from the laboratory of Sam Weissman and his collaborators at Washington University in St. Louis, to whom I am most indebted.