Magnetoelastic Effects in KMnF3

Measurements on KMn${\mathrm{F}}_3$ have revealed several anomalies in the magnetic susceptibility near $T_N=87.9\mathrm{}$ °K. When a crystal is cooled below $T_N$ a small field-dependent susceptibility component, saturating at ∼ 100 Oe, is seen; near $T_N-1\mathrm{}°$K this component disappears; it is much weaker upon subsequent warming, but recovers when the sample is again cooled from slightly above $T_N$; a negative-field hysteresis and a low-frequency oscillatory behavior are also observed. Since this anomaly is much smaller in powdered samples, it is assumed to be associated with a weak moment induced by residual local stresses, caused by the crystallographic distortion at $T_N$. A more basic property of the bulk material, seen in all samples, is an increase of ∼8% in the antiferromagnetic susceptibility below $T_N$; the data are inconsistent with an exchange-magnetostriction mechanism. Many of the experimental results, including those of Heeger, Beckman, and Portis can be explained in terms of a magneto-elastic coupling mechanism. When the effective elastic constant, for strains which result in magnetic canting, is very low, then a large enhancement in the transverse antiferromagnetic susceptibility is expected, and a field-induced canting transition can occur. This transition and the first-order canting transition at $T_c\approx 82°$K will occur in the present model only if a stable crystallographic state with a spontaneous strain exists independently of magnetic interactions, at low temperatures. In samples of decreasing particle size, the width and thermal hysteresis of the canting transition increase until in particles ∼ 25 μm and smaller, the canted state can persist up to $T_N$. The measured canted moment at 77 °K is 9.6 emu/mole, approximately half the low-temperature value, which suggests that the canting angle of the antiferromagnetic sublattices is practically constant below $T_c$. It is proposed that an observed abrupt decrease in the ultrasonic attenuation below $T_c$ is due to magnetoelastic propagation of sound waves across the crystallographic domain walls, which above $T_c$ cause a large amount of scattering and attenuation.