X-Ray Crystallography and the Nature of the Chemical Bond

Linus Pauling    Linus Pauling Institute of Science and Medicine

One evening in February 1912, P. P. Ewald, a student in Arnold Sommerfeld's Institute of Theoretical Physics in the University of Munich, came to Max von Laue to ask for advice about how to attack a problem given to him by Sommerfeld, that of interaction of electromagnetic waves with a space lattice of scatterers. While Max von Laue was thinking about this problem, he had the idea that x-rays might be electromagnetic waves with wavelength of the order of magnitude of the distance between layers of atoms in crystals. He got two young experimental physicists, Friedrich and Knipping, to carry out an experiment in which a beam of x-rays passed through a crystal of copper sulfate pentahydrate, and observed spots, diffraction maxima, on the photographic plates set up around the crystal. Laue then developed the Laue equations, relating the direction in which diffraction maxima occur to the dimensions of the unit of structure of the crystal. A few months later W. Lawrence Bragg, a student in Cambridge University, while preparing a talk on the Laue discovery, developed a simple equation, the Bragg equation, to describe the diffraction maxima. His father, William H. Bragg, set up an apparatus to test the Bragg equation, and in 1913 Lawrence Bragg and his father published several papers reporting not only the wavelengths of x-ray lines but also the structures of a number of crystals, diamond, sodium chloride, fluorite, and others. This was the start of x-ray diffraction and the determination of the structure of crystals.

In 1913 I was 12 years old. I was spending part of my spare time reading books on mineralogy and trying to understand the properties of minerals in relation to their chemical composition. In 1914 my interest became broader, to include the whole field of chemistry. Later, around 1920, I became interested in metals and intermetallic compounds, and I attempted, without success, to make single crystals of iron by electrolytic deposition in a magnetic field.

I consider my entry into the field of x-ray crystallography, nine years after it had been developed, to be just about the most fortunate accident that I have experienced in my life. I had become interested in the question of the nature of the chemical bond, after having read the 1916 paper on the shared-electron-pair chemical bond by G. N. Lewis and the several 1919 and 1920 papers by Irving Langmuir on this subject. Then, in the spring of 1922, when I was appointed a teaching fellow in the California Institute of Technology and had accepted the job, Arthur Amos Noyes wrote to me to suggest that when I came to Pasadena in the fall I should work with Roscoe Gilkey Dickinson, who was then a National Research Council fellow, on the determination of the structure of crystals by x-ray diffraction. He suggested that I get a copy of the book X-Rays and Crystal Structure by W. H. and W. L. Bragg and read it, which I did during the summer, and when I arrived in Pasadena toward the end of September 1922, Roscoe Dickinson began teaching me the techniques of x-ray diffraction.

X-ray diffraction had been begun in Pasadena in 1917 by Lalor Burdick, who had spent some time with W. H. Bragg after having got his Ph.D. in chemistry in Switzerland. Dickinson came with Noyes from M.I.T. in 1918, continuing the work, and receiving his Ph.D. in 1920 as the first Caltech Ph.D.

There were several new graduate studfor physiology is greater than ents in chemistry at C.I.T. in 1922, and I can only surmise why Noyes decided that I should work with Dickinson on x-ray diffraction. The assignment may have been the result of a statement of my interest that I made in my letter of application for the teaching fellowship. It was a fortunate decision for me, in that even in those early days the technique was very powerful and, through the determination of interatomic distances, bond angles, ligancies, and other structural features it had important consequences for the question of the nature of the chemical bond.

In those early days it was difficult to determine the structure of a crystal with more than two or three parameters locating the atoms, in addition to the dimensions of the unit cell. It was soon recognized that in a crystal with high symmetry the number of parameters may be small, because atoms often were located on symmetry elements or at their intersections. Cubic crystals, because of their large number of symmetry operations, were the ones that were usually most easily susceptible to attack. In fact, Dickinson and a senior student, Albert Raymond, had just completed the determination of the first organic substance to have its structure determined. This was hexamethylenetetramine, C6H12N4, which forms cubic crystals, such that the carbon atom is fixed by one parameter and the nitrogen atom by one parameter.

At Dickinson's suggestion, I began searching the literature, especially the several volumes of Groth's Chemische Kristallographie, to find cubic crystals that might be worth study. During the first two months I prepared several inorganic compounds and made crystals of them, in total number 14, and made Laue photographs and rotation photographs of each of them. They all turned out to be so complicated as to make it unwise to try to determine their structures. Dickinson then gave me a specimen of molybdenite, a hexagonal crystal, and helped me to solve the problems that arose in the determination of its structure. It turned out to be quite interesting; not only was the value of the molybdenum-sulfur distance of interest to me, but also the fact that the molybdenum atom, surrounded by six sulfur atoms, had these atoms at the corners of a triangular prism rather than a regular octahedron.

As I recall, I was interested in both the interatomic distance and the unusual coordination polyhedron, but I had no hope of developing an understanding of these structural features. I had begun collecting experimental values of interatomic distances in crystals, stimulated by a paper by W. L. Bragg, published in 1922. Bragg had formulated a set of atomic radii, starting with 1.04 ? for sulfur, half the distance between the two sulfur atoms in pyrite, with other values obtained by the assumption of additivity from observed interatomic distances. The trouble with the Bragg radius was clear from the molybdenite structure, where there are close-packed layers of sulfur atoms in juxtaposition to one another, with the sulfur-sulfur distance 3.49 ?, rather than 2.08 ?. The idea that an atom of an element could have several radii, a van der Waals radius, an ionic radius, a single-bond covalent radius, a double-bond covalent radius, a metallic radius, and so on, had not yet been developed. During the next few years, however, Wasastjerna developed a set of ionic radii and V. M. Goldschmidt developed a set of radii that was a sort of combination of covalent radii and metallic radii. It was clear that much additional information about the nature of the chemical bond was being provided and was going to be provided in the future by the determination of the structure of crystals by the x-ray diffraction method.

I shall not say much about the next decade, during which many interesting crystal structures were determined and some general principles about these structures were developed, especially in relation to the silicates, which in the course of a decade changed from being about the most poorly understood minerals to the best understood minerals.

In 1934 the transition from early x-ray crystallography to modern x-ray crystallography was begun by the discovery of the Patterson diagram by A. L. Patterson. Use of the Patterson diagram permitted a straightforward attack on the determination of the structure to be made for many crystals. I remember that in 1937 Robert B. Corey and I decided together that structures should be determined of crystals of amino acids and simple peptides, as a method of attack on the protein problem. At that time no correct structure for any of these substances had been reported. Structure determinations had been made for some related substances, simple amides, and the theory of the chemical bond had been developed to such an extent as to permit the conclusion to be reached with confidence that the peptide group (the amide group) in polypeptides should be very closely planar, and there was also reliable knowledge about the expected interatomic distances, including the N-H-O bond length. I had tried to use these structural elements in predicting ways in which polypeptide chains might fold, with the formation of hydrogen bonds, and in 1937 had decided that my efforts had failed, and that probably there was some structural feature yet to be discovered about proteins.

Eleven years later, when the alpha helix was discovered, about a dozen of these structures had been determined, all of them in the Gates and Crellin Laboratory of Chemistry. Verner Schomaker has pointed out to me that every one of these structure determinations made valuable use of Patterson diagrams. In 1948 I recognized, however, that my 1937 idea that there might be something new and surprising about amino acids and peptides was wrong. The structural features were the same in 1948 as in 1937.

During the period of 20 years, beginning in 1934, a number of investigators, including Patterson himself, made further contributions to the solution of...