The positions of the magnetic peaks observed in both the powder and single-crystal diffraction measurements put severe constrain on the possible magnetic structures. Firstly the magnetic cell is the same as the crystallographic one hence the magnetic propagation vector k=(0, 0, 0,). Next the presence of the 111 reflection indicates that the glide plane perpendicular to [1 1 0] reverses the spin direction. Finally the absence of the 001 and 003 reflections suggests that the spin direction is along the c-axis. The only structure which retains hexagonal symmetry and which is consistent with these observations is that in which the moments of Fe2 atoms in the same layer (z=¼) are ferromagnetically aligned while the magnetic coupling between adjacent layers is antiferromagnetic (shown below).
click on figure for more details
Magnetic structure of TiFe2. Only Fe atoms are shown.
The resultant molecular field at the Fe1 atoms is zero as it lies midway between two antiferromagnetically coupled planes on a centre of symmetry which is associated with time reversal. Under these circumstances the Fe1 atoms can have no ordered moment.
On the other hand, the magnetic scattering by the magnetization induced by the applied field field in the polarized neutron diffraction experiment was compared with that calculated for a simple model consisting of superposition of spherical magnetization distributions as calculated for the Fe²+ free ion centred at the iron sites. The magnitudes of the moments associated with Fe1 and Fe2 sites in this model were obtained from least-squares fits of the calculated structure factors to the observations. For the two orientations i.e. applied field parallel and perpendicular to the c-axis, the susceptibilities of the two sites were found to be approximately equal, but the absolute value of the moment induced, by the applied field of 4.6T for the c-axis orientation (0.008(2) µB/atom) was ten times smaller than that with the c-axis perpendicular to the field (0.095(5) µB/atom).
The lack of an ordered moment on the Fe1 atoms, as in some RMn2 Laves phases compounds, is a common feature in frustrated magnetic structures in which there are sites at which the magnetic symmetry imposes zero molecular field. In TiFe2 the fact that a site with no ordered moment persists down to low temperature suggests that the Fe moment also is near to instability in this compound. The Fe1 atom would then be non-magnetic rather than magnetically disordered. Such an assumption leads to a consistent interpretation of the induced moments obtained from the polarized neutron measurements.