The magnetic behaviour of the RMn2Ge2 sub-series (R = rare earth) of naturally layered ternary compounds depends critically on the Mn next-neighbour distance. For values greater than 2.87 Å ferromagnetism is observed, whereas below this antiferromagnetic ordering is favoured. Within this sub-series, SmMn2Ge2 is a unique case because the Mn-Mn planar distance is approximately equal to this critical value, and numerous magnetisation measurements have shown a complex temperature dependence of the magnetic ordering in the compound. SmMn2Ge2 has three magnetically ordered phases. At 345 K (above which the compound is paramagnetic), the material becomes ferromagnetic. Below 155 K, antiferromagnetic ordering occurs, and remains for temperatures down to 105 K, then the compound becomes ferromagnetic once more, i.e., it exhibits re-entrant ferromagnetism. The ordering is strongly anisotropic: in the high temperature ferromagnetic phase, the easy axis lies along the [001] (c-axis) direction, whereas in the low temperature ferromagnetic phase it lies along the [110] (basal plane) direction. As a naturally-layered material, the properties of SmMn2Ge2 provide an important complement to studies of artificial multilayer materials, and therefore a full understanding of its magnetisation is essential.

Magnetic Compton scattering, which is uniquely sensitive to the spin component of magnetisation, was used to study the spin density distribution in SmMn2Ge2. The basal plane magnetic Compton profile (MCP) for SmMn2Ge2 was measured on the high energy X-ray beamline ID15A, with an incident beam energy of 296 keV, and at temperatures of 40 K and 230 K. The MCP measured at 40 K is shown in Figure 75, together with model profiles for Sm 4f and Mn 3d electrons based on relativistic Hartree Fock free atom wavefunctions, which have been fitted to the data for the momentum range pz > 2 a.u. The profiles for Mn 3d and Sm 4f electrons are significantly different, the latter being 50% broader, and therefore fitting at high momenta can be used to separate the moments. This difference in the characteristic widths of the Compton lines can be thought of as simply arising from the fact that the Sm 4f electrons are more tightly bound than the Mn 3d electrons. The position space and momentum space wavefunctions constitute a Fourier pair and thus this difference manifests itself in higher momentum components for the 4f against the 3d electrons: a result which is also evident from simple consideration of the uncertainty principle.

The results presented in Figure 75 clearly show that there is a large negative 4f spin moment, opposed to the positive Mn 3d moment. For 4f electrons, deviations from the atomic behaviour are small. The area under the fitted 4f curve therefore gives a reliable estimate of the Sm spin moment which we deduce to be 3.2 ± 0.5 µB. The total spin moment, calculated simply by integrating the MCP, is zero. Therefore the spin moment associated with the Mn 3d electrons together with the delocalised electrons, also amounts to the same value, 3.2 µB, but with an antiparallel alignment. The fitted Mn 3d profile (dotted line in Figure 75) is inappropriate at pz < 2 a.u. because the 3d electrons are sensitive to the solid state environment, and their contribution will differ from free atom behaviour. In addition there are likely to be small delocalised low momentum contributions from both Mn (4sp-like) and Sm (5d- and 6sp-like). From these results, we may simply interpret that there is a Sm 4f spin moment of 3.2 µB, which is aligned antiparallel to the Mn 3d, and total magnetisations. Since the total spin moment is zero, in order to account for the macroscopic ferromagnetic moment of 4.1 µB, measured by SQUID magnetometry, there must be an orbital moment of this size. This orbital moment is aligned with the total magnetisation, i.e. parallel to the Mn spin direction. The total Sm moment is therefore 0.9 ± 0.5 µB and is aligned parallel to the bulk magnetisation direction. These results show clearly that, despite the small size of the total Sm magnetisation as measured by neutrons, the spin and orbital contributions are both large, and the magnetic Compton scattering experiment provides a unique method to access this information. The magnetic Compton profile measured at T = 230 K and presented in Figure 76, shows a clear difference in shape. The Sm 4f moment is very much reduced, but the high momentum region is not purely Mn 3d-like. In order to fit to these data, there must still be a small Sm 4f moment antiparallel to the Mn moment. As far as the authors are aware, this is the first time the high temperature Sm moment has been observed conclusively in this material.

Authors
J.E. McCarthy (a), J.A. Duffy (b), C. Detlefs (a), M.J. Cooper (b), P. Canfield (c).

(a) ESRF
(b) University of Warwick (UK)
(c) Ames Laboratory, Iowa State University (USA)