Suppression of local magnetic moment formation and paramagnetic exchange interactions in monolayer Fe 3 GeTe 2

The study investigates the electronic and magnetic properties of monolayer Fe3GeTe2 within a DFT+DMFT approach in its paramagnetic phase, arguing that this compound exhibits non-linear temperature dependencies of magnetic susceptibilities and possesses strong site differentiation between iron atoms, which is crucial for stabilizing long-range ferromagnetic order.

How it works

The research employs a systematic DFT+DMFT study focusing on the itinerant magnetic behavior of monolayer Fe3GeTe2. To describe exchange interactions in this correlated system, the authors utilized an alternative formalism based on nonuniform magnetic susceptibility, which is applicable in the paramagnetic phase and does not require the assumption of a certain magnetic state, providing an unbiased calculation of exchange interactions. This method allows for tracing the temperature evolution of exchange interactions, including self-energy and vertex corrections.

Electronic Properties

The DFT+DMFT calculations reveal significant differences in the occupation of d states among non-equivalent Fe atoms. Specifically, the authors find that the iron atoms located above and below the Ge plane are more strongly correlated than the iron atoms located within the Ge plane, which is consistent with previous considerations. Furthermore, correlations yield a shift of the peak of the density of states for Fe1,2 atoms to larger energies, while tellurium atom density of states shifts to lower energies. The electronic self-energies show that orbitals 1, 3-5 have a "non-quasiparticle form with ∂ImΣ(iνn)/∂νn > 0 (which is related to the formation of the local magnetic moment), while the self-energies of Fe3 atoms exhibit a quasiparticle form."

Magnetic Susceptibilities and Local Moments

The temperature dependence of local susceptibilities shows that the local susceptibilities at the Fe1,2 atoms show Curie-like behavior, corresponding to the formation of local magnetic moments, while Fe3 atoms exhibit Paulilike susceptibilities. The size of the local magnetic moment extracted from these dependencies is found to be in agreement with experimental data when extrapolated to low temperatures. Specifically, for moderate Coulomb interactions (U = 3−4 eV), a calculated moment of µ2loc ≃ 40µB is obtained, which the authors extrapolate to µ2loc ≃ 60µB/Fe1,2 in better agreement with the experimental data.

Exchange Interactions and Spin-Wave Dispersions

The exchange interactions are determined by relating them to the orbital-summed non-local static longitudinal susceptibility. In Figs. 5 and 6, the momentum dependence of susceptibilities shows maxima at the Γ point, indicating a tendency toward ferromagnetic order. The exchange interactions J11,22 (which are determined by the inverse susceptibilities matrix) have a minimum at the Γ point, corresponding to antiferromagnetic exchange between Fe1-Fe2 atoms. Crucially, these interactions are compensated by other types of interactions, which are all ferromagnetic (with the largest Jr3 interaction, r = 1, 2). The resulting spin-wave stiffness is estimated as D ≃ 235 meV·˚A2 at U = 4 eV, which is in reasonable agreement with the experimental data D ≃ 200 meV·˚A2.

Curie Temperature

The finite Curie temperature (TC) arises solely from magnetic anisotropy, as long-range ordering is forbidden in an isotropic two-dimensional system. Using the exchange interactions and spin stiffness, the authors estimate TC =170 K for U = 4 eV, which is in reasonable (good) agreement with the experimental data available for monolayer samples (130 K). The total square of the magnetic moment estimated near T DMFT C is found to be µ2 ≃ 13µ2B/Fe (9µ2B/Fe) for U = 4 eV (2 eV), and extrapolating the temperature dependencies yields a low-temperature estimate of µ2 ≃ 21µ2B/Fe (17µ2B/Fe) for U = 4 eV (2 eV), respectively, which agrees with experimental values. The study concludes that the results highlight the important relationship between the itinerant behavior of Fe3GeTe2 and the suppression of the local magnetic moment at iron atoms located within the Ge plane (Fe3), leading to a strong differentiation of magnetic sites, much larger than previously proposed.

The gist: In monolayer Fe3GeTe2, correlations suppress local magnetic moment formation at Fe3 atoms while enhancing it at Fe1-Fe2 atoms, leading to site-differentiated behavior that explains the observed paramagnetic magnetic properties and Curie temperature.

[1] S. Jiang, J. Shan, and K. F. Mak, Electric-field switching of two-dimensional van der Waals magnets, Nature Materials 17, 406 (2018).

As a fastidious and diligent researcher, I have analyzed the provided scientific paper on the electronic and magnetic properties of monolayer Fe3GeTe2 using DFT+DMFT. The findings provide several high-impact insights into strongly correlated 2D systems, specifically regarding site differentiation in magnetism and the interplay between itinerancy and local moments.

Here are specific improvements that can be made to AI systems, based on the scientific principles detailed in this paper:

)1. Improved Materials Discovery and Screening for Strongly Correlated 2D Magnets

• Improvement: Integrate the concepts of site-differentiated behavior and non-linear temperature dependencies derived from DFT+DMFT into machine learning (ML) potentials or high-throughput screening pipelines. Specifically, train models to predict the magnetic ground state (local moment size, site occupancy) based on local electronic structure features (like orbital occupation differences between Fe1/2 and Fe3 sites).

• What the improved AI system can do: The system could rapidly screen vast chemical spaces for new 2D materials where site-specific correlation effects are expected to be dominant. It would move beyond simple band structure prediction to identify materials with complex magnetic landscapes (e.g., distinguishing between localized vs. itinerant magnetic character across different atomic sites).

)2. Enhanced Predictive Modeling for Magnetic Susceptibility and Exchange Interactions

• Improvement: Implement a surrogate model or a neural network trained on the DFT+DMFT output (specifically, the non-local susceptibility formalism described in Section II) to predict exchange parameters (Jq) and magnetic response functions directly from input structural/electronic descriptors. This bypasses computationally expensive DMFT calculations for rapid property estimation.

• What the improved AI system can do: The system could provide near real-time predictions of macroscopic magnetic properties like the Curie temperature or spin-wave stiffness for novel compounds, significantly accelerating materials characterization workflows that currently rely on lengthy DMFT simulations.

)3. Development of Physics-Informed Machine Learning (PIML) for Correlation Effects

• Improvement: Utilize the explicit mathematical relationships derived from the paper—such as the equation relating exchange interactions to inverse susceptibilities (Eq. 1 in Sec. III C)—as constraints within a PIML framework. This forces the ML model to learn physically consistent representations of electron correlations rather than just pattern matching on raw data.

• What the improved AI system can do: The AI would be capable of performing inverse design for magnetic materials, optimizing structural parameters (lattice constants, interlayer spacing) specifically to maximize or minimize desired correlation effects (e.g., tuning the degree of suppression of local moments or enhancing long-range ferromagnetic order).

)4. Accurate Simulation and Analysis of Itinerant Magnetism

• Improvement: Enhance quantum simulation algorithms (like CT-QMC solvers mentioned in Section II B) by incorporating the findings regarding the non-quasiparticle form of self-energies for Fe1/2 atoms. The AI training data should be explicitly weighted to capture the transition between quasiparticle and non-quasiparticle behavior based on local electronic density near half-filling.

• What the improved AI system can do: This allows for more physically accurate simulation of systems exhibiting itinerant magnetism, enabling better prediction of phenomena like Kondo effect crossovers or temperature-dependent magnetic phase transitions in metallic 2D magnets.

)5. Automated Comparison and Validation Against Experimental Data

• Improvement: Create a validation module that systematically compares the AI's predicted magnetic moment ratios (e.g., the ratio of local moment to saturation moment) and Curie temperatures against experimental benchmarks, specifically focusing on the discrepancies attributed to temperature extrapolation (as discussed in Section III D).

• What the improved AI system can do: The system could automatically flag theoretical predictions that are sensitive to extrapolation errors or overestimated moments, providing a confidence score for its predictions based on how well they align with empirical data, thereby reducing reliance on potentially flawed theoretical extrapolations.

We study the electronic and magnetic properties of monolayer Fe 3 GeTe 2 within the DFT+DMFT approach in the paramagnetic phase. We argue that this compound is sufficiently far from the local magnetic moment limit, demonstrating non-linear temperature dependencies of the partial inverse local and uniform magnetic susceptibilities in a broad temperature range. We find that in the regime of moderate Coulomb interactions (U=3-4 eV), the iron atoms located above and below the Ge plane carry a substantial local magnetic moment (μ 4.5 μ B), while the iron atom located within the Ge plane does not exhibit any pronounced magnetic moment. At the same time, the RKKY-type exchange interactions between these two symmetry-nonequivalent types of atoms turn out to be crucial for stabilizing long-range ferromagnetic order in Fe 3 GeTe 2. The estimated spin-wave stiffness and Curie temperature are in good agreement with the experimental data, indicating that a dynamical treatment of electron correlations in Fe 3 GeTe 2 is essential to properly describe its partially itinerant magnetic behavior.