June 9, 1943

Dr. M. A. Peacock

Department of Mineralogy

University of Toronto

Toronto 5, Canada

Dear Dr. Peacock:

Your letter has tempted me to think some more about your crystal Ni3Bi2S2, even though I should be devoting my time to other things. I have finally decided that this crystal, like many others containing bismuth and sulfur, must contain Bi-S-Bi-S-Bi-S- strings, connected with covalent bonds. The Bi-S distance would he 2.50 Å or slightly larger, perhaps as large as 2.60 Å, corresponding to a covalent bond. The bond angle for sulfur would be expected to be about 105°; that for bismuth would be uncertain, because of the fact that the electronic structure must be such as to correspond to an increase of the shell beyond an octet for bismuth. Presumably the bismuth atom has three unshared pairs of electrons and forms two covalent bonds, a total of ten pairs of electrons. It is probable, however, that with this electron configuration the bond angle for bismuth would also be around 100°. The Bi-S distance 2.50 Å corresponds, with the lattice constant a0 equal to 4.02 Å to a bond angle for both sulfur and bismuth of 105°. This is, then, completely reasonable.

I have studied thoroughly the possible ways of introducing this chain, and I have not found any satisfactory way except by having the sulfur atoms also form a covalent bond with one another. This leads to what seems to be a reasonable structure. If we divide the unit cell up into four quadrants, as shown on the attached sketch, then we can bring the sulfur atoms together, to within 2.1 Å of one another, in one of the quadrants. The sulfur atoms are at (½,y,z¯) and (½,½y¯,½z+) with y = O.063 and z = 0.26. Then nickel atoms would be put in the centers of the other three quadrants, as you have done. This would surround the first nickel atom with four bismuth atoms at 2.82 Å, two nickel atoms at 2.86 Å, and two sulfur atoms at 1.75 Å. The last of these distances is too small, and I think that perhaps this nickel atom may be at the positions occupied by sulfur in your structure instead. The other two nickel atoms are in the positions you assign them. Why don't you try calculating intensities for structures of this sort, shifting the nickel atoms around among the possible positions.

Ten or twelve years ago I was planning to make a study of the sulfide minerals, but I abandoned the plan. I am interested to see the work which you and Mary have done on cell dimensions of some of these minerals. Have you noticed that those containing bismuth and sulfur usually have a dimension along a fiber axis of 4.05 ± O.05 Å? I think that this indicates the presence in them of the same Bi-S-Bi-S-Bi- chain described above.

The compound PdBi2 presents an interesting problem. I suggest that it consists of layers 2Bi-2Pd-2Bi----2Bi-2Pd-2Bi----, each layer 2Bi consisting of a puckered Bi-Bi-Bi-Bi- chain similar to that described above, with Bi-Bi equal to 3.0 Å and bond angle about 100 or 105°. These chains would stretch along the 4.28 Å axis, in the plane 100. Two such bismuth layers would be held together by a palladium layer midway between them, each palladium atom presumably having six bismuth neighbors at 2.9 Å. A structure of this sort accounts for the good cleavage and the subsidiary cleavage, and for the intensities of reflection from the basal plane 100. I think that the β modification probably contains essentially the same layers.

I shall be interested to know how you get along with your structure determinations.

Sincerely yours,

Linus Pauling

LP:jr

Enclosure