Ab InitioConstruction of Magnetic Phase Diagrams in Alloys: The Case ofFe1−xMnxPt

Abstract

A first-principles approach to the construction of concentration-temperature magnetic phase diagrams of metallic alloys is presented. The method employs self-consistent total energy calculations based on the coherent potential approximation for partially ordered and noncollinear magnetic states and is able to account for competing interactions and multiple magnetic phases. Application to the Fe 1−x Mn x Pt "magnetic chameleon" system yields the sequence of magnetic phases at T = 0 and the c-T magnetic phase diagram in good agreement with experiment, and a new low-temperature phase is predicted at the Mn-rich end. The importance of non-Heisenberg interactions for the description of the magnetic phase diagram is demonstrated.Magnetic substitutional alloys are often found to excel in applications [1,2], because alloying broadens the parameter space for tuning the desired properties. However, wide tunability, combined with the need to target certain operating temperature ranges, presents a challenge for empirical materials design. Competing magnetic interactions in alloys can produce complicated magnetic phase diagrams (MPD) with multiple magnetic phases [3][4][5][6]. Understanding of the c-T MPD's is therefore essential for the development of advanced magnetic materials. Some MPD's can be computed using the Heisenberg model with empirical or calculated exchange parameters combined with the mean-field approximation (MFA) [7,8] Monte-Carlo simulations [9][10][11][12], or spin-fluctuation theory [13]. However, many systems are not adequately described by the Heisenberg model. In metallic alloys the interaction parameters are sensitive to the electronic structure and population and thereby to the content of the alloy [8,9,11] and to the degree of spin disorder [14]. To avoid the limitations of the Heisenberg model, one can use first-principles spin-dynamics simulations [15] or construct a generalized spin Hamiltonian to map the adiabatic energy surface [16,17] for use in thermodynamic calculations. The energies of disordered spin configurations can also be obtained using the disordered local moment (DLM) model [18,19], where the spin-rotational averaging is done in the coherent potential approximation (CPA). While all these approaches fail in strongly itinerant magnets, they are applicable when the spin moments do not vary by more than 10-20% in different spin configurations. We restrict ourselves to such systems here.First-principles spin-dynamics and the construction of a microscopic generalized spin Hamiltonian are computationally very demanding and unfeasible for most systems of practical interest. We have developed [20] an alternative approach, in which self-consistent DLM and noncollinear CPA calculations are used to construct a Ginzburg-Landau-type totalenergy functional expressed through a small number of relevant magnetic order parameters. Combined with the MFA expression for the magnetic entropy, this method provides the variational free energy. A similar scheme was used to describe the phase transitions in F...

Spin-Fluctuation Mechanism of Anomalous Temperature Dependence of Magnetocrystalline Anisotropy in Itinerant Magnets

“…Our implementation of the DLM method is described in Ref. 23. (See Supplemental Material [24] for additional details.…”

First-Principles Calculation of Curie Temperature Tuning in <i>L</i>1₀-Type FePt by Element Substitution of Mn, Cu, Ru, and Rh

“…magnetic structure is assumed in the present study, we consider that the obtained result is related to previous studies in which the canted ferromagnetic order was observed in (Fe 1−σ Mn σ )Pt (0.1 ≲ σ ≲ 0.2) by neutron diffraction 7) and first-principles calculations. 31) If the substitution amount of Mn is increased further, an antiferromagnetic order appears due to complex magnetic interactions. In Fig.…”

Effects of intrinsic defects and alloying with Fe on the half-metallicity of Co2MnSi

“…The effects of thermal spin disorder are introduced using the vector disordered local moments (DLM) model [29][30][31][32], which assumes that the orientation of each local magnetic moment i has an independent distribution function, p i (θ i ) ∝ exp(α i cos θ i ), where θ i is the polar angle made by that local moment with the global magnetization. The parameters α i are different for inequivalent lattice sites and for different components on the same site; they are obtained from the solution of the coupled equations of the mean-field approximation (MFA) at the given temperature with the magnetic exchange parameters calculated using the linear response technique [33].…”