Phase diagrams of bond and site frustrated magnetic materials: Experiments and theory

We have carried out large scale Monte Carlo simulations of site and bond frustrated Heisenberg models in order to gain insight into experimental phase diagrams on bond and site frustrated magnetic materials. In the case of site frustrated models, we find that spin glass ordering does not take place on either the simple cubic or body-centered cubic lattices which are bipartite. Instead, the model decouples into either one or two percolating clusters of ferromagnetic (antiferromagnetic) sites which order ferromagnetically (antiferromagnetically). The ordering is similar to the situation which exists when ferromagnetic and antiferromagnetic sites do not interact at all. Including an interaction between ferromagnetic and antiferromagnetic sites, to create frustration, causes the ferromagnetic (FM) and antiferromagnetic (AF) order to become mutually perpendicular while increasing both TC and TN. This behaviour is similar to what occurs in our experiments on site frustrated a-(Fe100-xMnx)78Si8B 14; for this alloy ferromagnetic and spin glass order co-exist transverse to one another for 0.17 < x < 0.31. Geometrical frustration likely plays a significant role in the material since the glass structure cannot support antiferromagnetic order, unlike bipartite lattices. We conjecture that the site frustrated model on the non-bipartite face-centered cubic lattice, which is geometrically frustrated with respect to AF order, produces a spin glass with sufficient randomness and would make a superior model.;In the case of bond frustrated models, we have found that the phase diagram includes all of the bulk phases predicted in mean field theory by Gabay and Toulouse over 25 years ago, despite the long held view that Heisenberg spin glasses do not order in three dimensions. In particular we find that the model orders ferromagnetically at small concentrations of frustration, and in all cases where ferromagnetic order occurs (0 < x < 0.21) there is a second low temperature transition where spin glass ordering occurs transverse to the magnetization at Txy, in full agreement with our experiments on the bond frustrated alloy a-FexZr 100-x. In addition, we find that the spin glass phase exists at finite TSG for 0.21 < x ≤ 0.5 (the phase diagram is symmetrical about x = 0.5). TSG possesses a small clearly detectable concentration dependence, and we find that at x c = 0.21(1), TC, Txy, and TSG merge at a multicritical point. We have followed the evolution of Txy with magnetic field B for both the model and the material and we find full agreement between the two; Txy is found to behave as Txy(B) = Txy(0)(1 - BB+J ) where J is a constant. That this particular equation, first suggested to describe experimental results, also accounts for the field dependance of Txy in our bond frustrated model demonstrates that the only ingredients needed to describe the material is the presence of random competing bonds.