Predicting the Curie temperature in substitutionally disordered alloys using a first-principles based model

Marian Arale Brännvall, Rickard Armiento, Björn Alling

This paper presents a robust, physics-based approach to a classic problem in materials science: predicting the Curie temperature ($T_C$) of complex alloys. By combining first-principles calculations with a refined analytical model, the authors demonstrate a method that is more physically grounded than "black-box" machine learning and more scalable than traditional atomistic simulations.

TL;DR

Researchers have developed a first-principles model that predicts the Curie temperature ($T_C$) of disordered alloys by calculating the energy cost of flipping magnetic moments. The model successfully captures how $T_C$ evolves across diverse systems like steel alloys and Heusler compounds, even providing a "blind" prediction for the radioactive (and thus hard-to-measure) iron-technetium system.

What Problem Does This Paper Address?

The Curie temperature—the point at which a material loses its permanent magnetism—is a "make-or-break" property for magnetic applications. In modern alloy design, we often mix multiple elements to tailor properties, but predicting how these mixtures affect $T_C$ is notoriously difficult.

Experimental discovery is slow and expensive. On the theoretical side, researchers usually face a trade-off:

  1. High-fidelity Heisenberg models: These require intense manual effort to map out exchange interactions for every new composition.
  2. Machine Learning (ML): While fast, ML models often struggle to capture the continuous, physical variation of $T_C$ as composition changes, especially when they lack a deep "understanding" of the underlying magnetic physics.

The authors bridge this gap with a model that is automated enough for high-throughput screening but remains rooted in fundamental physics.

Methods

The authors utilize a model previously introduced in Phys. Rev. Materials (2024), which defines $T_C$ using three primary physical quantities derived from Density Functional Theory (DFT):

  • $\Delta E$ (Energy Difference): The energy gap between the magnetic ground state (ordered) and the paramagnetic state (disordered).
  • $S^{mag}$ (Magnetic Entropy): A measure of the disorder in the paramagnetic state.
  • $NN$ (Nearest Neighbors): The number of nearest magnetic neighbors.

Key Techniques: * VASP (DFT): Used for the underlying electronic structure calculations. * Special Quasirandom Structures (SQS): A method to simulate "random" alloys within a finite periodic cell. * Disordered Local Moments (DLM): Used to simulate the paramagnetic state by constraining magnetic moments in random directions. * The "Secret Sauce": The model includes a correction factor $(1 - B/NN^C)$ that accounts for Short-Range Order (SRO). This is vital because as you dilute a magnet with non-magnetic atoms, the local environment changes, which simple mean-field theories often miss.

Datasets

The model was validated against several experimentally known substitutional random alloy systems:

  • Iron-based binaries: $Fe_{1-x}Co_x$, $Fe_{1-x}Cr_x$, $Fe_{1-x}V_x$.
  • Heusler alloys: $NiMnSb$-based systems (including $Ni_xCu_{1-x}MnSb$).
  • Nitrides: $Ti_{1-x}Cr_xN$.
  • Itinerant systems: $Co_{1-x}Al_x$.
  • New Prediction: The radioactive $Fe_{1-x}Tc_x$ system.

Experimental data for comparison was pulled from a wide array of historical literature (dating back to 1909) to ensure a rigorous benchmark.

Key Insights and Ideas

  • Physics Captures Trends Better than ML: While ML might struggle with the "kinks" in $T_C$ curves caused by phase transitions (like the bcc-to-fcc transition in Fe-Co), this model naturally handles them because the underlying DFT calculations "see" the structural change.
  • The Dilution Effect: The model correctly predicts that adding non-magnetic elements (like Aluminum or Copper) lowers $T_C$ by reducing the number of magnetic neighbors ($NN$), effectively weakening the "magnetic glue" holding the spins in alignment.
  • Itinerancy is the Enemy: The model performs best on "local" magnets (where spins are tied to specific atoms). It struggles with "itinerant" magnets (where electrons move freely), such as the $Co-Al$ system. In these cases, the size of the magnetic moment changes drastically between the ordered and disordered states, a complexity the current model doesn't fully capture.

Key Results

  • Fe-Co Success: The model showed excellent agreement in the cobalt-rich fcc region (error < 60 K) and accurately tracked the $T_C$ peak in the bcc region.
  • Heusler Precision: For $Ni_xMnSb$, the predictions were very close to experimental values, with the largest error (~100 K) occurring only at high nickel concentrations where structural disorder becomes significant.
  • Technetium Prediction: The model predicts that adding Technetium to Iron will steadily decrease $T_C$, behaving similarly to Molybdenum or Ruthenium. This provides a useful theoretical baseline for a system that is difficult to test in a lab.
  • Lower Bound Limitation: The model has a mathematical lower limit of 124 K. Consequently, it overestimated the $T_C$ for $Ti_{1-x}Cr_xN$ (predicting ~200 K for a system that is experimentally ~50-140 K).

Limitations and Future Work

  • Low-$T_C$ Accuracy: The 124 K floor prevents the model from being used for cryogenic magnetic materials. The authors suggest a future correction term could help differentiate low- and high-$T_C$ regimes.
  • Itinerant Magnetism: To better handle systems like $Co-Al$, the model may need to account for the change in the magnitude of magnetic moments during the transition to paramagnetism, rather than just their orientation.
  • Convergence: For highly itinerant moments (like Cr or V in iron), the DFT calculations sometimes struggle to converge when spins are constrained, suggesting a need for more robust electronic self-consistency strategies.

Summary updated

Abstract

When exploring new magnetic materials, the effect of alloying plays a crucial role for numerous properties. By altering the alloy composition, it is possible to tailor, e.g., the Curie temperature ($T_\text{C}$). In this work, $T_\text{C}$ of various alloys is investigated using a previously developed technique [Brännvall et al. Phys. Rev. Mat. (2024)] designed for robust predictions of $T_\text{C}$ across diverse chemistries and structures. The technique is based on density functional theory calculations and utilizes the energy difference between the magnetic ground state and the magnetically disordered paramagnetic state. It also accounts for the magnetic entropy in the paramagnetic state and the number of nearest magnetic neighbors. The experimentally known systems, Fe$_{1-x}$Co$_x$, Fe$_{1-x}$Cr$_x$, Fe$_{1-x}$V$_x$, NiMnSb-based Heusler alloys, Ti$_{1-x}$Cr$_x$N, and Co$_{1-x}$Al$_x$ are investigated. The experimentally unexplored system Fe$_{1-x}$Tc$_x$ is also tested to demonstrate the usefulness of the developed method in guiding future experimental efforts. This work demonstrates the broad applicability of the developed method across various systems, requiring less hands-on adjustments compared to other theoretical approaches.
BibTeX
@article{brannvall2025predicting,
  author = {Marian Arale Brännvall and Rickard Armiento and Björn Alling},
  title = {{Predicting the Curie temperature in substitutionally disordered alloys using a first-principles based model}},
  journal = {Journal of Magnetism and Magnetic Materials},
  year = {2025},
  doi = {10.1016/j.jmmm.2025.173361},
}

Loading related content...

Figure