S. Toth and B. Lake
Laboratory for Neutron Scattering, PSI; Helmholtz-Zentrum Berlin; EPFL; TU Berlin
(Dated: March 4, 2014)
Abstract
This paper introduces a generalized numerical framework for calculating the magnetic excitations of complex materials. It specifically tackles the challenge of "incommensurate" magnetic structures—systems where the magnetic order does not repeat in sync with the underlying crystal lattice. The authors present a general algorithm and open-source software (SpinW) that calculates spin wave spectra for virtually any magnetic structure, including complex incommensurate ones, by employing a rotating coordinate transformation.
Introduction: The Challenge of Incommensurability
Linear Spin Wave Theory (LSWT) is the standard tool for modeling how spins in a solid "ripple" or excite, creating quasiparticles called magnons. While LSWT is well-established for simple magnets, it struggles with incommensurate structures. These occur frequently in frustrated magnets and multiferroics, where the magnetic pattern (such as a spiral) has a wavelength that is not a simple integer multiple of the atomic spacing.
Previously, researchers addressed this by creating massive "supercells" to approximate the spiral. However, this approach is computationally expensive and produces a "spaghetti" of folded energy bands that are difficult to interpret physically. The authors sought a general, efficient algorithm capable of handling arbitrary exchange interactions, anisotropies, and magnetic field effects without these workarounds.
Methodology: The Rotating Frame Transformation
The core innovation of this work is a two-step coordinate transformation:
- Local Transformation: Every spin in the chemical unit cell is assigned its own local coordinate system so that its equilibrium direction points along the $z$-axis.
- Rotating Frame: To handle incommensurability, the authors introduce a rotating frame that follows the magnetic ordering wave vector $\mathbf{Q}$. This mathematically "unwinds" the spiral, transforming an incommensurate structure into a commensurate one within the rotating frame.
Following this transformation, the standard Holstein-Primakoff transformation is applied to convert spin operators into bosonic operators. To solve the resulting quadratic Hamiltonian, the authors utilize Colpa’s method—a specialized numerical technique for diagonalizing bosonic matrices that ensures physical commutation relations are preserved.
Key Insights and Implementation
- Mode Reduction: By using the rotating frame, the number of spin wave modes is reduced to $3N$ (where $N$ is the number of atoms in the chemical unit cell), regardless of the magnetic period's length. This makes the physical origin of each mode significantly easier to identify.
- Phasons vs. Canting: The formalism naturally distinguishes between "phasons" (fluctuations shifting the phase of the magnetic spiral) and "canting" modes (fluctuations pushing spins out of their ordering plane).
- Open Science: The authors implemented this theory into the SpinW Matlab toolbox, which has since become a standard tool in the neutron scattering community for modeling inelastic neutron scattering data.
Validation and Results
The algorithm was validated using the distorted triangular lattice antiferromagnet $\alpha$-CaCr$_2$O$_4$. The authors successfully modeled the 8 distinct spin wave modes of this material, demonstrating that the spectrum is characterized by dispersions at both $\mathbf{k}$ and $\mathbf{k} \pm \mathbf{Q}$, a hallmark of incommensurate systems. The method allows for the calculation of dynamical structure factors in a fraction of the time required by traditional supercell methods and can account for isotropic exchange, Dzyaloshinskii-Moriya (DM) interactions, and arbitrary single-ion anisotropy simultaneously.
Limitations
As a linear theory, this framework assumes small fluctuations and does not account for magnon-magnon interactions or "magnon decay," which can be significant in $S=1/2$ systems or at high temperatures. Furthermore, the rotating frame method is most powerful when the Hamiltonian is invariant under the rotation defined by $\mathbf{Q}$; systems that strongly break this symmetry may still require more complex treatments.
Summary updated