64 GENERAL DISCUSSION GENERAL DISCUSSION Dr. B. Bleaney (Oxford University) said : It is important to emphasize the distinction between the amount of " s-character " introduced by hybridization in molecules in the formation of directive bonds, and that introduced in atoms through " configurational interaction ". In an atom with only one electron, such as hydrogen, there is a central field with a pure Coulomb potential (V=Ze/r) and the wave equation can be solved exactly. In a many-electron atom the potential variation is more complex and no exact solution is possible. An accurate representation of the electronic state requires the superposition of several combinations of one-electron wave functions (configurations) each of which corresponds to the same term. Thus the ground state of the Mn2+ ion, normally written as (3~)2(3p)6(345, 6S5,2, may be more correctly represented by a small admixture of the state (%)(3~)6(3d>5(4~), 6S5,2.Such an admixture has been postulated by Abragam and Pryce 1 to explain the hyperfine structure observed in the paramagnetic resonance spectrum of this ion, which would otherwise be expected to be zero, since there is no resultant orbital momentum, and a spherically symmetrical distribution of electron spin magnetism (not containing any unpaired s-electrons) would give zero magnetic field at the nucleus. Since the wave-function of an s-electron does not fall to zero at the origin, the magnetic moment of its spin 1 Abragam and Pryce, Proc. Roy. Sac. A, 1951,205, 135.GENERAL DISCUSSION 65 produces a finite magnetic field at the origin.(The difference in this respect be- tween s-electrons, and other S-terms whose wave functions vanish at the origin, corresponds to the classical result that there is a finite magnetic field at the centre of a solid pzrmanently magnetized sphere, but not at the centre of a hollow one.) Although the 3s and 4s electrons in the admixed configuration must have their spin moments parallel, the 3s electron has a much greater density at the nucleus than the 4s electron and produces most of the resultant magnetic field, giving a non- zero hyperfine structure. It should be noted that configurational interaction may give admixtures with states with promoted p or d electrops, as well as ones with promoted s-electrons, but only the latter are important as regards hyperfine structure.The effect is not confined to configurations with half-filled shells, but it is most noticeable there since it gives a finite hyperfine structure where none would be expected. In other configurations it gives a change in the size of the structure. A similar effect has been found by Heald and Beringer 1 in the nitrogen atom, which has also a half-filled shell, with the configuration (ls)2(2S)2(2p)3, 4S3/2. This atom is found to have a small hyperfine structure, presumably because of a small admixture of a (ls)2(2s)(2p)3(3s), 4S3p state. The observed size of the hyperfine structure is only about 0.6 % of that which a 2s electron would give, so that the admixture required is small. In the NO molecule, the h.f.s attributed to this admixture is some four times bigger, and thus is distinctly greater than in the nitrogen atam.A similar effect is to be expected in the oxygen atom, but no measurements have been made; it is probably again rather smaller than that ob- served in the oxygen molecule. Prof. C. A. Coulson (Oxford University) said : The work of Pryce and Eisenstein, described by Dr. Bleaney, provides an answer to a question of much interest to 0 FIG. 1 .-Co-ordination of NO3 groups around the uranyl axis U89+ in uranyl nitrates. All three NO3 groups lie in the equa- torial plane shown, and the dots indicate possible partial covalency. I 0 chemists : do f electrons take part in bonding? On the basis of orbital bond strength calculations by Pauling, they would be expected to do so strongly, since Pauling’s strengths of s, p , d and f orbitals are 1, 4 3 , 4 5 and 2/7 respectively, for o-type bonds.Yet many of the chemical and spectroscopic properties of the rare-earths and the transuranic elements suggest that the f-electrons are really “ inner ” electrons unsuited to the formation of normal chemical bonds. Now, however, Dr. Bleaney shows that 5f electrons must be involved in bonds. It is interesting to point out that Dr. G. R. Lester and the present writer have obtained quite alternative (unpublished) evidence pointing to a similar con- clusion. Chemical measurements by Glueckauf and others show plainly that in solution the uranyl group U022+, which is linear, possesses a marked ability to co-ordinate six atoms (usually oxygen atoms from surrounding water or nitrate groups), and these six atoms lie either exactly or nearly in the equatorial plane defined by the U02 axis (fig.1). This regular geometrical arrangement argues strongly for directional properties such as those to be associated with chemical 1 Heald and Beringer, Physic. Rev., 1954, 96, 645. C66 GENERAL DISCUSSION bonds. But no central atom can form six simultaneous bonds of this kind without using f atomic orbitals, and to the extent, therefore, that these may be called ’‘ bonds ”, we are led to suppose that f electrons participate in forming them. But it seems as if the conventional language of bonds is not really adequate, and we have a situation in which the old-fashioned electrovalence and the more modern covalence are merging into one.As Griffiths and Owen have shown, many molecular complexes may be regarded as essentially ionic though the polarization of the migrating electrons is sufficiently great as to correspond to a partial covalence. But as Lester and the present writer have shown in their work on the uranyl nitrate complexes, the hybrids from the uranium atom which may be used in forming the a-bonds to the equatorial ligands have exceedingly strongly directed density patterns, such as that whose polar diagram is shown in fig. 2. This is even more directed than the familiar sp3 tetrahedral or the sp3d2 octahedral hybrids of Pauling. Now an elec- tron in an orbital of this kind has its mean centre of position at some place such as P in the figure. In this it differs from a pure unhybridized s, p , d, .. . orbital whose centre is at the nucleus. One way of describing this is to say that an electron in a hybrid orbital such as fig. 2 already confers a partial ionic character on the ‘‘ bond ” to the appropriate ligand. Thus starting with electrons on the central atom we find their orbits moving towards the outer atoms: alternatively starting with electrons on the outer atoms (ionic model) we find their orbits creeping in towards the central atom. We have a situation in which electrovalence and covalence each have character- istics conventionally associated with the other. Dr. D. J. Millen and Mr. K. M. Sinnott (University College, London) (cam- municated) : In the interpretation of the microwave spectrum of nitryl chloride, NOzCI, we have used an extreme case of one of thc methods described by Prof.Gwinn for establishing planarity. By making use of the nuclear quadrupole fine structure, a number of lines were assigned to low J transitions. The rotational constants of the molecule were evaluated and the positions of further lines cal- culated. In no case could a line be found corresponding to a transition involving states whose rotational eigen-functions would be antisymmetric with respect to the operation C f . This is consistent with these states having a statistical weight of zero. The only structure which satisfies this requirement is planar \N-CI with C2v symmetry. For this structure the operation C2a interchanges the oxygen nuclei, which have zero nuclear spin. Prof. C. H.Townes (Columbia University) said : It seems to me rather difficult to be sure from theoretical arguments whether deuterium or hydrogen should have the higher barrier in nitromethane. The effect of the several normal vjbra- tional modes on the barrier height gives some of difficulty. In addition, the barrier height measured, which depends on cos 66, represents a small difference in the main features of the barrier for one hydrogen. On the other hand, Prof. Gwinn’s experimental measurements of barrier height for the two cases are, of course, quite definite. Dr. W. J. Orville Thomas (Aberystwyth) said : In answer to Prof. H. C. Longuet- Higgins, centrifugal distortion constants, obtained from microwave data, can be combined with infra-red measurements to yield information about the structure FIG.2.-polar diagram of Possible SY P, d, f hybrid orbital. 0 O/GENERAL DISCUSSION 67 of valence bonds. In principle, information concerning force constants (including interaction constants) is given by the displacement of rotational lines due to centrifugal distort ion. It is hoped in the near future to obtain the infra-red spectrum of methyl mercury chloride and to assign frequencies to the stretching modes of the linear 6-Hg-C1 grouping. If the methyl group is treated as a single particle it will then be possible to obtain the allowed solutions for the two bond-stretching force constants f(C-Hg), f(Hg-Cl) and the bond-bond interaction constant f(CHg/HgCl), occurring in the potential function. The allowed solutions for the force constants represent an i w t e number of sets for the three force constants. Using the known value for the centrifugal distortion constant Dij it might be possible to determine which set of force constants reproduces the infra-red data and the microwave data best thus yielding an explicit solution for the force constants. It is clearly important, then, to obtain accurate values for the centrifugal distortion constants of molecules since although at present the calculations are difficult a certain amount of progress is being made in utilizing this comparatively new source of information about the behaviour of valence bonds.