Stresses in semiconductors: Ab initio calculations on Si, Ge, and GaAs

Spectral-partitioned Kohn-Sham density functional theory

We introduce a general, variational scheme for systematic approximation of a given Kohn-Sham free-energy functional by partitioning the density matrix into distinct spectral domains, each of which may be spanned by an independent diagonal representation without requirement of mutual orthogonality. It is shown that by generalizing the entropic contribution to the free energy to allow for independent representations in each spectral domain, the free energy becomes an upper bound to the exact (unpartitioned) Kohn-Sham free energy, attaining this limit as the representations approach Kohn-Sham eigenfunctions. A numerical procedure is devised for calculation of the generalized entropy associated with spectral partitioning of the density matrix. The result is a powerful framework for Kohn-Sham calculations of systems whose occupied subspaces span multiple energy regimes. As a case in point, we apply the proposed framework to warm- and hot-dense matter described by finite-temperature density functional theory, where at high energies the density matrix is represented by that of the free-electron gas, while at low energies it is variationally optimized. We derive expressions for the spectral-partitioned Kohn-Sham Hamiltonian, atomic forces, and macroscopic stresses within the projector-augmented wave (PAW) and the norm-conserving pseudopotential methods. It is demonstrated that at high temperatures, spectral partitioning facilitates accurate calculations at dramatically reduced computational cost. Moreover, as temperature is increased, fewer exact Kohn-Sham states are required for a given accuracy, leading to further reductions in computational cost. Finally, it is shown that standard multiprojector expansions of electronic orbitals within atomic spheres in the PAW method lack sufficient completeness at high temperatures. Spectral partitioning provides a systematic solution for this fundamental problem.

Adhesion and friction behaviors of a γ-Fe/diamond heterogeneous contact interface: a density functional theory study

Diamond tools play a vital role in precision machining. However, the adhesive wear restricts their application when Fe-based workpieces are cut by diamond tools. Thus, it is significant to theoretically explain the interface binding mechanism between the diamond and Fe alloy matrix. In this study, the adhesion and friction behaviors of a γ-Fe/diamond (denoted as Fe/C) heterogeneous contact interface were investigated employing density functional theory (DFT). The results show that the transfer of the Fe atom to C atom occurs when the interaction energy for a given configuration is larger than the separation energy of the corresponding Fe surface layers. The energy barriers of the Fe/C(100), (110) and (111) sliding interfaces along the minimum energy path are 1.45, 0.48 and 0.42 J m-2, respectively, indicating that the Fe/C(111) interface is the easiest to slide. Furthermore, the friction potential barrier increases with an increase in the load (1-5 nN) according to the potential energy curves. Moreover, the friction coefficient of the Fe/C interface is larger than 0.2 and provides a theoretical minimum friction coefficient for the Fe/C sliding interface.

Unexpected low thermal expansion coefficients of pentadiamond

A new carbon allotrope, pentadiamond, was recently reported in the literature. Herein, we investigate its thermal expansion and thermoelastic properties by first principles. It is observed that the bulk modulus and hardness of pentadiamond are far less than those of diamond, but the thermal expansion of pentadiamond is lower than that of diamond in the range of 0 K to 2000 K, and even negative in the temperature range of 0-190 K. The negative thermal expansion at low temperature originates from the transverse vibrations of the edge-shared atoms in the coplanar double-pentagon. The low thermal expansion at high temperature is contributed by the strong bonds in pentadiamond. Benefiting from the low thermal expansion, the elastic constants of pentadiamond decrease very slowly with respect to temperature compared with those of diamond. The low sensitivity of thermodynamic and thermoelastic properties to temperature makes pentadiamond a promising material for high anti-thermal-shock and accurate electronic device applications.

Finite-temperature atomic relaxations: Effect on the temperature-dependent C44 elastic constants of Si and BAs

The effect of atomic relaxations on the temperature-dependent elastic constants (TDECs) is usually taken into account at zero temperature by the minimization of the total energy at each strain. In this paper, we investigate the order of magnitude of this approximation on a paradigmatic example: the C₄₄ elastic constant of diamond and zincblende materials. We estimate the effect of finite-temperature atomic relaxations within the quasi-harmonic approximation by computing ab initio the internal strain tensor from the second derivatives of the Helmholtz free-energy with respect to strain and atomic displacements. We apply our approach to Si and BAs and find a visible difference between the softening of the TDECs computed with the zero-temperature and finite-temperature atomic relaxations. In Si, the softening of C₄₄ passes from 8.6% to 4.5%, between T = 0 K and T = 1200 K. In BAs, it passes from 8% to 7%, in the same range of temperatures. Finally, from the computed elastic constant corrections, we derive the temperature-dependent Kleinman parameter, which is usually measured in experiments.

Revisiting the Rate-Dependent Mechanical Response of Typical Silicon Structures via Molecular Dynamics

Strain rate is a critical parameter in the mechanical application of nano-devices. A comparative atomistic study on both perfect monocrystalline silicon crystal and silicon nanowire was performed to investigate how the strain rate affects the mechanical response of these silicon structures. Using a rate response model, the strain rate sensitivity and the critical strain rate of two structures were given. The rate-dependent dislocation activities in the fracture process were also discussed, from which the dislocation nucleation and motion were found to play an important role in the low strain rate deformations. Finally, through the comparison of five equivalent stresses, the von Mises stress was verified as a robust yield criterion of the two silicon structures under the strain rate effects.

Insight into the inclusion of heteroatom impurities in Silicon structures

The bonding properties of tilt boundary in poly-silicon and the effect of interstitial impurities are investigated by first-principles. In order to obtain thorough information on the nature of chemical bondings...

The role of Si vacancies in the segregation of O, C, and N at silicon grain boundaries: An ab initio study

Grain boundaries (GBs) are defects originating in multi-crystalline silicon during crystal growth for device Si solar cell fabrication. The presence of GBs changes the coordination of Si, making it advantageous for charge carriers to recombine, which brings a significant reduction of carrier lifetimes. Therefore, GBs can be highly detrimental for device performances. Furthermore, GBs easily form vacancies with deep defect electronic states and are also preferential segregation sites for various impurity species, such as C, N, and O. We studied from first principles the correlation between structural, energetics, and electronic properties of the Σ3{111} Si GB with and without vacancies, and the segregation of C, N, and O atoms. C and O atoms strongly increase their ability to segregate when vacancies are present. However, the electronic properties of the Σ3{111} Si GB are not affected by the presence of O, while they can strongly change in the case of C. For N atoms, it is not possible to find a clear trend in the energetics and electronic properties both with and without vacancies in the GB. In fact, as N is not isovalent with Si, as C and O, it is more flexible in finding new chemical arrangements in the GB structure. This implies a stronger difficulty in controlling the properties of the material in the presence of N impurity atoms compared to C and O impurities.

General Molecular Dynamics Approach to Understand the Mechanical Anisotropy of Monocrystalline Silicon under the Nanoscale Effects of Point Defect

Mechanical anisotropy and point defects would greatly affect the product quality while producing silicon wafers via diamond-wire cutting. For three major orientations concerned in wafer production, their mechanical performances under the nanoscale effects of a point defect were systematically investigated through molecular dynamics methods. The results indicated anisotropic mechanical performance with fracture phenomena in the uniaxial deformation process of monocrystalline silicon. Exponential reduction caused by the point defect has been demonstrated for some properties like yield strength and elastic strain energy release. Dislocation analysis suggested that the slip of dislocations appeared and created hexagonal diamond structures with stacking faults in the [100] orientation. Meanwhile, no dislocation was observed in [110] and [111] orientations. Visualization of atomic stress proved that the extreme stress regions of the simulation models exhibited different geometric and numerical characteristics due to the mechanical anisotropy. Moreover, the regional evolution of stress concentration and crystal fracture were interrelated and mutually promoted. This article contributes to the research towards the mechanical and fracture anisotropy of monocrystalline silicon.

Real-space formulation of the stress tensor for O(N) density functional theory: Application to high temperature calculations

We present an accurate and efficient real-space formulation of the Hellmann-Feynman stress tensor for O(N) Kohn-Sham density functional theory (DFT). While applicable at any temperature, the formulation is most efficient at high temperature where the Fermi-Dirac distribution becomes smoother and the density matrix becomes correspondingly more localized. We first rewrite the orbital-dependent stress tensor for real-space DFT in terms of the density matrix, thereby making it amenable to O(N) methods. We then describe its evaluation within the O(N) infinite-cell Clenshaw-Curtis Spectral Quadrature (SQ) method, a technique that is applicable to metallic and insulating systems, is highly parallelizable, becomes increasingly efficient with increasing temperature, and provides results corresponding to the infinite crystal without the need of Brillouin zone integration. We demonstrate systematic convergence of the resulting formulation with respect to SQ parameters to exact diagonalization results and show convergence with respect to mesh size to the established plane wave results. We employ the new formulation to compute the viscosity of hydrogen at 10⁶ K from Kohn-Sham quantum molecular dynamics, where we find agreement with previous more approximate orbital-free density functional methods.

Quasi-harmonic temperature dependent elastic constants: applications to silicon, aluminum, and silver

We presentab-initiocalculations of the quasi-harmonic temperature dependent elastic constants. The isothermal elastic constants are calculated at each temperature as second derivatives of the Helmholtz free energy with respect to strain and corrected for finite pressure effects. This calculation is repeated for a grid of geometries and the results interpolated at the minimum of the Helmholtz free energy. The results are compared with the quasi-static elastic constants. Thermodynamic relationships are used to derive the adiabatic elastic constants that are compared with the experimental measurements. These approaches are implemented for cubic solids in thethermo_pwcode and are validated by applications to silicon, aluminum, and silver.

Graphene thermal break-down induced by anharmonic bending mode

A novel approach is proposed allowing us to prove, self-consistently, that free-standing graphene reaching a certain temperature loses its mechanical stability resulting in abrupt breakdown, which can be interpreted as melting. Our study is based on the idea of the crucial role of the anomalously soft bending 'sound' mode in the jump transition of graphene from the state with relatively small bending fluctuations to a state with fluctuations close in amplitude to the graphene lattice constant. The acme of the developed theory is in establishing a quantitative relationship connecting the graphene elastic moduli of second, third (negative!), and fourth orders at the melting temperature T _m that permits us to calculate T _m. The results obtained lay a theoretical foundation for an analog of Lindemann criterion for graphene.

First-Principles Calculation of Third-Order Elastic Constants via Numerical Differentiation of the Second Piola-Kirchhoff Stress Tensor

A general method is presented to calculate from first principles the full set of third-order elastic constants of a material of arbitrary symmetry. The method here illustrated relies on a plane-wave density functional theory scheme to calculate the Cauchy stress and the numerical differentiation of the second Piola-Kirchhoff stress tensor to evaluate the elastic constants. It is shown that finite difference formulas lead to a cancellation of the finite basis set errors, whereas simple solutions are proposed to eliminate numerical errors arising from the use of Fourier interpolation techniques. Applications to diamond, silicon, aluminum, magnesium, graphene, and a graphane conformer give results in excellent agreement with both experiments and previous calculations based on fitting energy density curves, demonstrating both the accuracy and generality of our new methodology to investigate nonlinear elastic behaviors of materials.

On the calculation of the stress tensor in real-space Kohn-Sham density functional theory

We present an accurate and efficient formulation of the stress tensor for real-space Kohn-Sham density functional theory calculations. Specifically, while employing a local formulation of the electrostatics, we derive a linear-scaling expression for the stress tensor that is applicable to simulations with unit cells of arbitrary symmetry, semilocal exchange-correlation functionals, and Brillouin zone integration. In particular, we rewrite the contributions arising from the self-energy and the nonlocal pseudopotential energy to make them amenable to the real-space finite-difference discretization, achieving up to three orders of magnitude improvement in the accuracy of the computed stresses. Using examples representative of static and dynamic calculations, we verify the accuracy and efficiency of the proposed formulation. In particular, we demonstrate high rates of convergence with spatial discretization, consistency between the computed energy and the stress tensor, and very good agreement with reference planewave results.

Equation of state of warm dense deuterium and its isotopes from density-functional theory molecular dynamics

Of the two approaches of density-functional theory molecular dynamics, quantum molecular dynamics is limited at high temperature by computational cost whereas orbital-free molecular dynamics, based on an approximation of the kinetic electronic free energy, can be implemented in this domain. In the case of deuterium, it is shown how orbital-free molecular dynamics can be regarded as the limit of quantum molecular dynamics at high temperature for the calculation of the equation of state. To this end, accurate quantum molecular dynamics calculations are performed up to 20 eV at mass densities as low as 0.5g/cm^{3} and up to 10 eV at mass densities as low as 0.2g/cm^{3}. As a result, the limitation in temperature so far attributed to quantum molecular dynamics is overcome and an approach combining quantum and orbital-free molecular dynamics is used to construct an equation of state of deuterium. The thermodynamic domain addressed is that of the fluid phase above 1 eV and 0.2g/cm^{3}. Both pressure and internal energy are calculated as functions of temperature and mass density, and various exchange-correlation contributions are compared. The generalized gradient approximation of the exchange-correlation functional, corrected to approximately include the influence of temperature, is retained and the results obtained are compared to other approaches and to experimental shock data; in parts of the thermodynamic domain addressed, these results significantly differ from those obtained in other first-principles investigations which themselves disagree. The equations of state of hydrogen and tritium above 1 eV and above, respectively, 0.1g/cm^{3} and 0.3g/cm^{3}, can be simply obtained by mass density scaling from the results found for deuterium. This ab initio approach allows one to consistently cover a very large domain of temperature on the domain of mass density outlined above.

Equation of state of a dense plasma by orbital-free and quantum molecular dynamics: examination of two isothermal-isobaric mixing rules

We test two isothermal-isobaric mixing rules, respectively based on excess-pressure and total-pressure equilibration, applied to the equation of state of a dense plasma. While the equation of state is generally known for pure species, that of arbitrary mixtures is not available so that the validation of accurate mixing rules, that implies resorting to first-principles simulations, is very useful. Here we consider the case of a plastic with composition C(2)H(3) and we implement two complementary ab initio approaches adapted to the dense plasma domain: quantum molecular dynamics, limited to low temperature by its computational cost, and orbital-free molecular dynamics, that can be implemented at high temperature. The temperature and density range considered is 1-10 eV and 0.6-10 g/cm(3) for quantum molecular dynamics, and 5-1000 eV and 1-10 g/cm(3) for orbital-free molecular dynamics. Simulations for the full C(2)H(3) mixture are the benchmark against which to assess the mixing rules, and both pressure and internal energy are compared. We find that the mixing rule based on excess-pressure equilibration is overall more accurate than that based on total-pressure equilibration; except for quantum molecular dynamics and a thermodynamic domain characterized by very low or negative excess pressures, it gives pressures which are generally within statistical error or within 1% of the exact ones. Besides, its superiority is amplified in the calculation of a principal Hugoniot.

First-principles local stress in crystalline and amorphous metals

Atomic size is perhaps the most commonly used concept to describe material properties. Advances in the understanding of materials are hindered by the available choices of simplifying concepts that can be used. However, the precise definition of atomic size is not easy, and often controversial. Atomic level stress provides a new interpretive tool that draws on the rich formalism of solid mechanics for use with density functional calculations to advance a deeper understanding of the properties of materials. We discuss atomic level stresses in liquids and glasses and make comparisons with ordered and disordered crystals. Somewhat surprisingly, even ordered compounds that are under no macroscopic stress and whose individual atoms are completely relaxed, i.e., no force acting on them, can have substantial atomic level stresses. On top of concepts such as the ionicity or covalency, the atomic level stresses add to the arsenal of analysis tools that are available to interpret the results of density functional calculations.

Comparison of stress and total energy methods for calculation of elastic properties of semiconductors

We explore the calculation of the elastic properties of zinc-blende and wurtzite semiconductors using two different approaches: one based on stress and the other on total energy as a function of strain. The calculations are carried out within the framework of density functional theory in the local density approximation, with the plane wave-based package VASP. We use AlN as a test system, with some results also shown for selected other materials (C, Si, GaAs and GaN). Differences are found in convergence rate between the two methods, especially in low symmetry cases, where there is a much slower convergence for total energy calculations with respect to the number of plane waves and k points used. The stress method is observed to be more robust than the total energy method with respect to the residual error in the elastic constants calculated for different strain branches in the systems studied.

Numerical convergence of the self-diffusion coefficient and viscosity obtained with Thomas-Fermi-Dirac molecular dynamics

Computations of the self-diffusion coefficient and viscosity in warm dense matter are presented with an emphasis on obtaining numerical convergence and a careful evaluation of the standard deviation. The transport coefficients are computed with the Green-Kubo relation and orbital-free molecular dynamics at the Thomas-Fermi-Dirac level. The numerical parameters are varied until the Green-Kubo integral is equal to a constant in the t→+∞ limit; the transport coefficients are deduced from this constant and not by extrapolation of the Green-Kubo integral. The latter method, which gives rise to an unknown error, is tested for the computation of viscosity; it appears that it should be used with caution. In the large domain of coupling constant considered, both the self-diffusion coefficient and viscosity turn out to be well approximated by simple analytical laws using a single effective atomic number calculated in the average-atom model.

First Principles Calculations of Surface Stress

Only recently have there been fully quantum-mechanical calculations of two-dimensional surface stress tensors. We have calculated total energies and stresses of semiconductor surfaces within the Local Density Approximation, using norm-conserving pseudopotentials. In order to hasten convergence of the stress with respect to basis size, it is useful to remove a fictitious tensile stress. We have calculated surface stress for the relaxed Si (111) 1×1 and 2×2-adatom surfaces, as well as for the relaxed Ge (111) 1×1 and 2×2-adatom surfaces. We have also calculated the surface stress for several chemisorbed systems, including Ga, Ge and As chemisorbed onto Si. We find a dramatic correlation between the electronic structure and chemistry of the surface, and its elastic properties.

Surface stress in d-band metal surfaces

Using calculations within the density functional approach, we make a survey of the surface stresses of relevant d-band metal surfaces. Unlike the case for surface energies, where volcano-shaped dependences have been found across the d period, the surface stress values show a non-trivial dependence on the metal species and surface termination, which cannot be anticipated from homogeneous electron gas models. A qualitative interpretation as regards the origin of this dependence can be given on the basis of the decomposition of the surface stress into repulsive and attractive electronic components.

Extended DFT + U + V method with on-site and inter-site electronic interactions

In this paper we introduce a generalization of the popular DFT + U method based on the extended Hubbard model that includes on-site and inter-site electronic interactions. The novel corrective Hamiltonian is designed to study systems for which electrons are not completely localized on atomic states (according to the general scheme of Mott localization) and hybridization between orbitals from different sites plays an important role. The application of the extended functional to archetypal Mott-charge-transfer (NiO) and covalently bonded insulators (Si and GaAs) demonstrates its accuracy and versatility and the possibility to obtain a unifying and equally accurate description for a broad range of very diverse systems.

An environment-dependent interatomic potential for silicon carbide: calculation of bulk properties, high-pressure phases, point and extended defects, and amorphous structures

An interatomic potential has been developed to describe interactions in silicon, carbon and silicon carbide, based on the environment-dependent interatomic potential (EDIP) (Bazant et al 1997 Phys. Rev. B 56 8542). The functional form of the original EDIP has been generalized and two sets of parameters have been proposed. Tests with these two potentials have been performed for many properties of SiC, including bulk properties, high-pressure phases, point and extended defects, and amorphous structures. One parameter set allows us to keep the original EDIP formulation for silicon, and is shown to be well suited for modelling irradiation-induced effects in silicon carbide, with a very good description of point defects and of the disordered phase. The other set, including a new parametrization for silicon, has been shown to be efficient for modelling point and extended defects, as well as high-pressure phases.

Surface stress of stepped chiral metal surfaces

The use of surface stress as a physical probe for examining chiral effects in surfaces is proposed. First-principles calculations of the surface stress in stepped achiral and chiral bcc metal surfaces (Fe, Mo, and W) are presented. When no mirror symmetry is present, principal stress orientations are unconstrained; nevertheless, we find that the stress is smoothly varying along a suitably chosen stereographic zone of surfaces. Stress ellipses for Fe differ qualitatively from those of Mo and W, suggesting that its surface stress has a distinct origin.

Adsorption-induced surface stresses in alkanethiolate-au self-assembled monolayers

First-principles calculations were employed to elucidate the origin of adsorption-induced surface stresses in alkanethiolate self-assembled monolayers on an Au(111) surface. Our results suggest a mechanism that accounts for the huge relief of the tensile stress compared to the bare surface in terms of a local rearrangement of surface Au atoms accompanying charge removal from the surface towards the Au-S bond. A purely interadsorbate interaction model is shown to be inconsistent with the anisotropy and the magnitude of the calculated stress.

First-Principles Studies on Grain Boundary Energies of [110] Tilt Grain Boundaries in Aluminum

The grain boundary structure and its energy are necessary for the fundamental understanding of the physical properties of materials. In aluminum, three distinct atomic structures of a Σ9(221)[110] tilt grain boundary have been reported in previous studies using atomistic simulations and a high-resolution transmission electron microscopy (HRTEM). In this work, we studied the atomic structure and energy of the Σ9 tilt grain boundary in aluminum using first-principles calculations. A comparison of the grain boundary energies among the three distinct Σ9 tilt grain boundaries determined through first-principles calculations allowed us to identify the most stable atomic structure of Σ9 tilt grain boundary in aluminum.

Structural and vibrational properties of solid nitromethane under high pressure by density functional theory

The structural, vibrational, and electronic properties of solid nitromethane under hydrostatic pressure of up to 20 GPa have been studied using density functional theory. The changes of cell volume, the lattice constants, and the molecular geometry of solid nitromethane under hydrostatic loading are examined, and the bulk modulus B0 and its pressure derivative B0' are fitted from the volume-pressure relation. Our theoretical results are compared with available experiments. The change of electron band gap of nitromethane under high pressure is also discussed. Based on the optimized crystal structures, the vibrational frequencies for the internal and lattice modes of the nitromethane crystal at ambient and high pressures are computed, and the pressure-induced frequency shifts of these modes are discussed.

Very-high-temperature molecular dynamics

It is shown that a modified scheme of density functional theory, using the Thomas-Fermi kinetic energy functional for the electrons, is well suited to perform very-high-temperature molecular dynamics simulations on high-Z elements. As an example, iron on the principal Hugoniot is simulated up to 5 keV and 5 times the normal density, giving an equation of state in agreement with current models. Ionic structure is obtained and is given to an excellent level of precision by the structure of the one-component plasma computed for a coupling parameter corresponding to Thomas-Fermi ionization.

Electronic-enthalpy functional for finite systems under pressure

We introduce the notion of electronic enthalpy for first-principles structural and dynamical calculations of finite systems under pressure. An external pressure field is allowed to act directly on the electronic structure of the system studied via the ground-state minimization of the functional E+PV(q), where V(q) is the quantum volume enclosed by a charge isosurface. The Hellmann-Feynman theorem applies, and assures that the ionic equations of motion follow an isoenthalpic dynamics. No pressurizing medium is explicitly required, while coatings of environmental ions or ligands can be introduced if chemically relevant. We apply this novel approach to the study of group-IV nanoparticles during a shock wave, highlighting the significant differences in the plastic or elastic response of the diamond cage under load, and their potential use as novel nanostructured impact-absorbing materials.

Three-dimensional MgB2-type superconductivity in hole-doped diamond

We substantiate by numerical and analytical calculations that the recently discovered superconductivity below 4 K in 3% boron-doped diamond is caused by electron-phonon coupling of the same type as in MgB2, albeit in three dimensions. Holes at the top of the zone-centered, degenerate sigma-bonding valence-band couple strongly to the optical bond-stretching modes. The increase from two to three dimensions reduces the mode softening crucial for T(c) reaching 40 K in MgB2. Even if diamond had the same bare coupling constant as MgB2, which could be achieved with 10% doping, T(c) would be only 25 K. Superconductivity above 1 K in Si (Ge) requires hole doping beyond 5% (10%).

Anomalous pressure-induced transition(s) in ice XI

The effects of pressure on the structure of ice XI-an ordered form of the phase of ice Ih, which is known to amorphize under pressure-are investigated theoretically using density-functional theory. We find that pressure induces a mechanical instability, which is initiated by the softening of an acoustic phonon occurring at an incommensurate wavelength, followed by the collapse of the entire acoustic band and by the violation of the Born stability criteria. It is argued that phonon collapse may be a quite general feature of pressure-induced amorphization. The implications of our findings for the amorphization of ice Ih are also discussed.

Magnetic stress as a driving force of structural distortions: the case of CrN

We show that the observed transition from rock salt to orthorhombic P(nma) symmetry in CrN can be understood in terms of stress anisotropy. Using local spin density functional theory, we find that the imbalance between stress stored in spin-paired and spin-unpaired Cr nearest neighbors causes the rock-salt structure to be unstable against distortions and justifies the observed antiferromagnetic ordering. This stress has a purely magnetic origin and may be important in any system where the coupling between spin ordering and structure is strong.