magnetic superexchange couplings between the tig cores which make
CaMnO3 the antiferromagnet it is. It is a difficult many-parameter problem and the outcome seems to be that LaMnO3 misses becoming a ferromagnet, but it is not wide off the mark [362]. The orbital order is nearly of the d3z2-r2/d3y2-r2 kind (Fig. 5.2, right). One should add to it a small uniform d322-,.2 component because it is helpful for the interplane antiferromagnetic coupling. All in all, the actual magnetic structure of LaMn03 is understood. Given that LaMn03 is an antiferrornagnet (AFM), and (for exam- ple’s sake) Lao.7Sro.3Mn03 is definitely a ferromagnetic metal (FM), we should ask how does the system get from one state to the other. The simplest possibility is to use a pure spin model of Lal-zSrzMn03, which contains the AFM t2g-t2g superexchange, and the double exchange term with the amplitude (8.41). It is understood that the latter is really the spin-orientation-dependent hopping energy of the eg electrons. We en- visage that the sublattice magnetizations gradually tilt towards each other as x increases. In other words, the [loo] Mn planes remain uni- formly polarized, but the angle subtended by the total spins of nearest- neighbour planes is an 2-dependent 19. I9 = 7r at 2 = 0, and I9 = 0 at some critical zcr. The expression of the mean field energy density (valid for infinite core spin42and JH/t -+ co) is [82, 2791
E(I9) = J cos I9 - x[4t + 2t cos (I9/2)]. (8.51)
Here we assumed that z is sufficiently small so that the band bottom
energy can be ascribed to all eg electrons. Minimizing E(19)with respect to 6 , we get tX cos(I9/2) = -. (8.52) 25 The notable features of the mean field solution are: a) the canting of the sublattice magnetizations at arbitrarily small 2 (at x (< 1, 19 M n - ( t z / J ) ) , and b) the existence of a critical xcr = 2 J / t . We predict that the transition from AFM to FM is gradual, with a canted phase for 0 < 2 < zcr,and full ferromagnetic alignment for x > z, 1821. stood purely from superexchange, and that coupling to the lattice influences only the numerics [113]. There is continuing dispute concerning such details. 42Thefinite-S mean field theory is developed in [227].