## Abstract

Single crystal X-ray analyses of the structure of potassium ferricyanide at ca. 300 and 95$^\circ$ K are reported. Least-squares refinements within the monoclinic space group, P2$_1$/c, have converged with the residuals R$^{300^{\circ K}}$ = 0$\cdot$092 and R$^{95^{\circ K}}$ = 0$\cdot$077. The unit cell dimensions at 300 $^\circ$K are: a = 7$\cdot$06, b = 10$\cdot$38, c = 8$\cdot$40 $\overset{\circ}{\mathrm A}$, $\beta$ = 107$\cdot$0$^\circ$, and at 95 $^\circ$K are: a = 7$\cdot$03, b = 10$\cdot$31, c = 8$\cdot$35 $\overset{\circ}{\mathrm A}$, $\beta$ = 107$\cdot$2$^\circ$. The coordination octahedra are only slightly distorted, both analyses independently revealing a small tetragonal elongation of the molecule along Fe-CN bonds lying nearly perpendicular to the crystallographic c axis. The principal crystal paramagnetic susceptibilities are characterized by an essentially unique, and lower, moment parallel to the c axis. These two observations are to be reconciled only in relation to the role of second-nearest neighbour effects of either antiferromagnetic or crystal-field character. Exchange is ruled out by earlier e.s.r. work and the present susceptibility dilution experiments. The crystal anisotropy appears to be dominated by the coulombic field set up by adjacent potassium ions in the lattice. While all potassium ions in the structure will play some role in this second-order perturbation, those forming 'chains' with the octahedra parallel to the c axis are likely to be the more important. A simple crystal-field approach, using symmetry-adapted, zero-order wavefunctions on the iron atoms, has shown how the correct direction and relative magnitudes of the crystal magnetic anisotropies may be predicted by this model. The splitting of the ground $^2$T$_{2g}$ term of the complex ion is estimated as 150 to 300 cm$^{-1}$.