WorksPhysics and Astronomy
Accurate Machine Learning Predictions of Coercivity in High-Performance Permanent Magnets
Abstract
Increased demand for high-performance permanent magnets in the electric vehicle and wind turbine industries has prompted the search for cost-effective alternatives.Discovering new magnetic materials with the desired intrinsic and extrinsic permanent magnet properties presents a significant challenge to researchers because of issues with the global supply of rare-earth elements, material stability, and a low maximum magnetic energy product BH.While first-principle density functional theory (DFT) predicts materials' magnetic moments, magneto-crystalline anisotropy constants, and exchange interactions, it cannot compute coercivity ().Although it is possible to calculate theoretically with micromagnetic simulations, the predicted value is larger than the experiment by almost an order of magnitude, due to the Brown paradox.To circumvent these, we employ machine learning (ML) methods on an extensive database obtained from experiments, DFT calculations, and micromagnetic modeling.The use of a large dataset enables realistic predictions for materials such as Ce-doped NdFeB, comparing favorably against micromagnetically simulated coercivities.Remarkably, our ML model accurately identifies uniaxial magneto-crystalline anisotropy as the primary contributor to . With DFT calculations, we predict the Nd-site dependent magnetic anisotropy behavior in NdFeB, confirming that Nd -sites mainly contribute to uniaxial magneto-crystalline anisotropy, and also calculate Curie temperature (T).Both calculated results are in good agreement with experiment.The coupled experimental dataset and ML modeling with DFT input predict with far greater accuracy and speed than was previously possible using micromagnetic modeling.Further, we reverse-engineer the inter-grain exchange coupling with micromagnetic simulations by employing the ML predictions.
Introduction
With the rapid advance of computational capabilities, there is considerable research interest in machine learning (ML) methods for predicting material properties using extensive databases1,2. Of specific interest is the remarkable speed of these techniques, which outperform traditional first-principle methods like density functional theory (DFT) by an order of magnitude. ML methods can deal with complex structures, and are desirable for discovering high-performance permanent magnet materials much needed for the electric vehicle and wind turbine industries. Although recent advances in first-principle methods such as DFT have enabled successful prediction of intrinsic properties, e.g., magnetic moments, magneto-crystalline anisotropy, and exchange interactions, the prediction of coercivity () is a daunting task. Theoretically, can be computed by solving the phenomenological Landau–Lifshitz–Gilbert equation (LLGE) with micromagnetic simulations, but these simulations overestimate the experimental by an order of magnitude due to the Brown paradox3–5. Previous papers employing ML to predict the extrinsic properties6–9 have been limited exclusively to micromagnetic simulated materials.
ML requires datasets that include information about material properties such as crystal structure, micromagnetic grain size and boundaries, saturation magnetization (), the uniaxial magnetocrystalline anisotropy constant (), the exchange stiffness constant (), and the Curie temperature (). To build predictive models based on our dataset, we utilize classical ML and artificial neural network (ANN) algorithms10–12. These models establish patterns and relationships between the independent and dependent () material properties. They are trained on subsets of the known data and tested on the complementary subsets to assess model accuracy and reliability.
An extensive survey of the literature resulted in a dataset of 300 experimentally known materials (see SupplementaryTable I), to our knowledge the largest current experimental ML magnetic dataset. Our second dataset consists of 8770 micromagnetically computed permanent magnet materials. Various predictive techniques, including ML, statistical inference, and micromagnetic modeling (mumax program) are applied to both datasets to predict and compare 14,15. In experimental materials, we find standard non-linear models such as the decision tree (DT), extreme gradient boosting (XGB), and random forest (RF) produce excellent results with (where is a standard statistical measure of accuracy in regression), but tuning the XGB regressor improves the measure to 0.89. Most importantly, ML clearly demonstrates that is related directly to , weakly to , and inversely to .
We predict the of cerium (Ce)-doped NdFeB 2 : 14 : 1 materials to demonstrate the complete pipeline enabled by the new ML toolkit coupled with ab initio calculations. First, for a pure neo-magnet, the site contribution to magnetocrystalline anisotropy is analyzed with DFT calculations showing that 4g-sites mainly contribute to the uniaxial magneto-crystalline anisotropy. Second, our computed for a pure compound using Green’s function in the atomic sphere approximation (ASA) is in good agreement with experiment. Finally, we employ DFT-computed parameters in ML for predicting the of Ce-doped compositions. The ML-predicted matches with experiment, demonstrating that the ab initio computed input parameters and the ML methodology are sufficient to predict experimental , even without access to experimental conditions and advanced internal structural properties. Conversely, the ML prediction for selected candidate materials is used to engineer their grain boundary size (GBS) and inter-grain coupling.
Micromagnetism
Micromagnetics is the study of the behavior of magnetic materials typically in the nanometer range. During its early formulation17,18, the field emphasized qualitative aspects of magnetism: the role of domain structures, domain walls, and magnetic vortices in ferromagnetic materials. The transition to computer simulation in micromagnetics was a significant advance, providing detailed examination of the forces at play inside a magnetic material19.
Naively, the estimate gives an upper bound on 3,20. However, this estimate disregards impurities and multi-grain structures in materials, leading to a gross overestimate of from theory alone: the Brown paradox. Micromagnetic simulations which include demagnetization (shape anisotropy) are crucial for a good understanding of magnetic materials, but the interplay among these complex effects and the Brown paradox still hinder accurate predictions of .
Theory
Magnetodynamics is described by a nonlinear partial differential equation for the spatio-temporal magnetization vector . The time evolution of is given by a phenomenological Landau–Lifshitz–Gilbert (LLG) equation17,18
Here, is the unit vector describing the magnetization of the material with as the saturation magnetic moment per unit volume, while , , and respectively are the effective static magnetic field, the gyromagnetic ratio, and the damping parameter (quantifying the rate at which the magnetization relaxes back to equilibrium). In (1), the first term is the precession of the magnetic moment around the external magnetic field. The second term is the damping, which relaxes the magnetic moment to the equilibrium. For time-independent scenarios, such as the computation of a hysteresis loop, the term is set to 0. Then, . The terms are as follows. is the externally applied field, which is taken as a parameter. is a long-range magnetic field
corresponding to self-interaction of the induced magnetic field with the magnetization across the material21. The exchange field is the Laplacian of the magnetization, which is obtained from the classical Heisenberg model
Although this expression was originally deduced for localized spins, it is still valid for itinerant systems to the first order approximation. is a measure of the strength of magnetic exchange interaction. is the uniaxial anisotropy term, written as
where and are anisotropy constants, and indicates the direction of the anisotropy vector, making it easier for the magnetization to align with the direction of in the case of ferromagnetic materials. Most papers only envisage a single constant: and . Other terms, such as the Dzyaloshinskii-Moriya interaction (DMI), were neglected due to the lack of experimental data for current materials and the relatively weak effect in modern magnetic materials22.
In order to estimate fundamental magnetic parameters in cases of partial experimental information, the following relations were utilized. is inversely proportional to the lattice constant () as given by
where is the spin quantum number, and is the number of magnetic ions per unit cell. Equivalently, can be expressed approximately in terms of as
Here is the number of the nearest neighbors of a magnetic ion. Finally, the equation
is used to estimate from the vacuum permeability and the maximum energy product , which is given by the maximum product of the magnetic flux density and the magnetic field strength at any point on the hysteresis loop23.
The micromagnetic equations are solved in the continuum approximation 24–26. is deduced from the magnetic free energy functional as
where is DMI. Additional terms may be added as needed to count for additional interactions. The is minimized with respect to using the steepest descent algorithm as implemented in the micomagnetic program26. Then the hysteresis loop is obtained by evaluating in each equilibrium magnetic state for different values of the external applied magnetic field.
Experimental materials
An experimental database of 300 magnetic materials was assembled[27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57] [20, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81]. A material was chosen if , , and were explicitly included in the experimental results. For the RFeB family (where R is a rare-earth element) and some other materials, only the experimental anisotropy field () is available, so the estimate was used82.
The exchange strength was determined in one of three different ways. The first was a direct reporting of the stiffness exchange, which was rarely available. For some materials, the value was interpolated from the values for similar alloys and compounds. The third method derived from the Curie-stiffness relation in Eq. (6), using the experimental Curie temperature and known lattice constant83. This latter method was used for RFeB, including Ce-doped NdFeB71, binary alloys69, SmCo, 1 : 5 compositions66, La/Pr/Co-doped hexaferrites61, and other element doped hexaferrites[56, 57] [60, 84].
Curie-stiffness relation in Eq. (6), using the experimental Curie temperature and known lattice constant83. This latter method was used for RFeB, including Ce-doped NdFeB71, binary alloys69, SmCo, 1 : 5 compositions66, La/Pr/Co-doped hexaferrites61, and other element doped hexaferrites[56, 57] [60, 84].
There was significant variation in the reported values, depending on the experimental conditions and fabrication methods. For example, the atmospheric composition during the annealing of iron (Fe) magnets affected as oxygen modified the material composition during the hardening process. Oxygenated crystalline defects, such as nucleation and pinning of the domain walls, play critical roles in the mechanism. The nucleation field85 (the magnetic field at which the atomic spin ceases to align along the magnetic easy axis) lowers the measured , while pinning does the opposite.
Cooling rates have significant effects on the material grain size[27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57] [20, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81]. Several dozen materials (most notably Nd and Fe based ferromagnets) do have grain size recorded in the database. Grain composition, as visualized in Figure 1, is known to be heavily predictive of a material’s 86. However, correlating the grain measurements from different sources made it apparent that a single number cannot completely represent the grain structure, and that the particular measurement techniques employed may induce additional inaccuracies. In any material with multiple measurements for a property, a representative median was selected from a specific experiment (defaulting to papers with more comprehensive materials property estimates). This choice was made before the application of statistical techniques, to ensure that no bias contaminated the dataset.

Figure 1. A magnetic cuboid of size showing different micromagnetic grains. The colors within each grain refer to different magnetization directions. The snapshot of spin texture is captured during the micromagnetic simulation.
For the better training of ML networks, an additional supplementary database was computed, using micromagnetic modeling with , , , , and in the . Using the uniaxial approximation, , , and were supplied as the inputs for stiffness exchange, magnetocrystalline anisotropy, and saturation magnetization. The input parameters were uniformly sampled from the cuboid determined by the ranges of , , and in the 300 experimental materials. There are then two measures of : the experimentally measured coercivity , possessed only by the experimental materials, and the computationally determined , calculated for both the hypothetical and the experimental materials.
For micromagnetic modeling, simple structures were chosen for the magnetic samples. Due to the limited grain structure data, for the majority of the database a single structure composed of multiple grains would have been chosen for the materials. This would have induced scaling on the , which instead may be modeled directly with ML. As no grains or boundaries were involved in the computation, a magnetic cube of at scale was found to be sufficient to avoid loss of precision in the uniform uniaxial anisotropy alignment. For Ce-substituted compositions, we used simulation cells of size for grain boundary engineering as shown in Figure 1.
Micromagnetic results
We compute the for experimentally known systems, and list the comparison of the calculated and experimental values for selected materials in Table I. Theoretically, the upper limit for is the anisotropy field87
however, the true experimental value is an order of magnitude smaller due to the uncertainty in the coercivity mechanism in permanent magnets, commonly known as the Brown paradox3. Experimentally measured values fit well with the empirical Kronmüller equation88 89,90, where is the field to nucleate a reverse domain, and is the demagnetization field, with and being renormalization factors. For example, NdFeB (sintered) has and .89,90
The micromagnetically simulated values differ from the experimental values by a factor of up to . Interestingly, these values are very similar to the experimental . Moreover, for some materials, the use of the estimated values for , , and using the empirical relations as discussed in Eqs. (5)–(7) will result in additional error. In general, depends non-linearly on grain size and domain wall width or particle size in magnetic materials91–95. In modern manufacturing, grain sizes are larger relative to domain sizes. This is especially common in neo-magnets, leading to a decrease of with increasing grain size96, where higher grain surface area hosts more defects. Grain boundary size also affects the demagnetization factor97, further reducing .
The experimental features for 211 of the 300 materials in the dataset are known, while the remaining 90 materials require the use of the aforementioned theoretical models to determine from experimental results. These materials show larger discrepancies in the micromagnetically predicted , which is worth investigating both experimentally and theoretically. We show a comparison of these results (for selected key magnetic materials) in Table I. As all micromagnetically predicted values are overestimated, the only meaningful comparison is obtained with a scaled coercivity , which for fits well with the experimental values obtained for the 210 materials. The estimate is generally a good match for 2 : 14 : 1 compositions, except for Co-based NdCoB and GdCoB. However, the appropriate scaling factor may vary for different compositions. Overall, the variation with independent features is similar in both theory and experiment, although there are some exceptions such as LaFeB. For 1 : 5 compositions, a similar trend is evident.
Machine Learning
Classical machine learning algorithms
Machine learning (ML) encompasses a variety of advanced statistical techniques. It creates a correspondence between a space of independent variables, , and a space of dependent variables, , by taking a ground truth function representing a sequence of observations on a limited subset of the data to construct a more general function which extends the function . This ground truth function may also be represented as a list of observations . The method by which is constructed from is referred to as ML, and it is written symbolically as , where ML is optionally subscripted to indicate the algorithm or hyperparameters in use.
The goal of ML is to represent the structure underlying as . This is usually quantified by splitting the observations into a training set and testing set so that . Then the function is constructed relative to a norm, termed a loss function, so that is minimized in some appropriately defined subspace of to avoid overfitting. The quality of the fit with is measured by evaluating using the same class of norms. In this paper, all ML algorithms use the standard Euclidean norm.
We use the scikit-learn library for the Python programming language for predictive data analysis98. The dataset (both experimental and theoretical) is split into a 70 : 30 train-to-test data set ratio for the model training. Performance does not change drastically by modifying the split ratio to 80 : 20. The classical linear ML models used are:
Linear regression;
Lasso regularization;
Ridge regularization.
The non-linear classical ML models used are:
Decision tree (DT) regression;
Random forest (RF) regression;
Gradient boost (GB) regression;
XGBoost (XGB) regression.
In ML training with cross-validation, the training dataset is further divided into two parts - the training and the validation datasets. Most ML models have arbitrary hyperparameters, which are not modified directly during training. The validation dataset allows a meta-training of the ML models by random or grid search on the training data and evaluation on the validation set without contaminating the testing set by predicting an ML hyperparameter choice for test set evaluation. The most relevant cross-validation method is k-fold validation. The dataset is randomly split into k disjoint subsets, then one set is chosen as the validation dataset, while the others are chosen as training datasets.
| Material | (MA/m) | (pJ/m) | (MJ/m) | (T) | (T) | (T) | (T) |
| LaFeB | 1.09 | 7.41 | 1.4 | 0.50 | 2.11 | 0.53 | 0.03 |
| CeFeB | 0.93 | 6.00 | 1.70 | 0.54 | 3.14 | 0.79 | 0.25 |
| PrFeB | 1.24 | 7.94 | 4.66 | 1.54 | 6.46 | 1.62 | 0.08 |
| NdFeB | 1.27 | 8.23 | 4.65 | 1.24 | 6.25 | 1.56 | 0.32 |
| GdFeB | 0.71 | 9.33 | 0.85 | 0.23 | 2.40 | 0.60 | 0.37 |
| TbFeB | 0.56 | 8.78 | 6.13 | 2.93 | 21.62 | 5.41 | 2.48 |
| DyFeB | 0.56 | 9.42 | 4.24 | 1.70 | 14.79 | 3.70 | 1.99 |
| HoFeB | 0.64 | 8.14 | 2.42 | 0.75 | 7.26 | 1.82 | 1.07 |
| LuFeB | 0.93 | 7.66 | 1.21 | 0.29 | 2.30 | 0.58 | 0.28 |
| YFeB | 1.12 | 8.01 | 1.46 | 0.19 | 2.16 | 0.54 | 0.35 |
| ThFeB | 1.12 | 6.77 | 1.46 | 0.17 | 2.10 | 0.53 | 0.36 |
| LaCoB | 0.79 | 1.50 | 1.19 | 0.34 | 3.06 | 0.77 | 0.42 |
| PrCoB | 1.04 | 1.51 | 5.20 | 2.50 | 9.43 | 2.36 | -0.14 |
| NdCoB | 1.08 | 1.51 | 2.42 | 3.69 | 4.22 | 1.05 | -2.64 |
| GdCoB | 0.23 | 1.52 | 1.03 | 0.30 | 9.48 | 2.37 | 2.07 |
| YCoB | 0.85 | 1.52 | 1.19 | 0.34 | 2.84 | 0.71 | 0.37 |
| SmCo | 0.86 | 1.20 | 1.72 | 7.50 | 38.80 | 9.70 | 2.20 |
| YCo | 0.78 | 7.11 | 5.50 | 3.90 | 13.33 | 3.33 | -0.57 |
| LaCo | 0.71 | 6.47 | 6.30 | 5.25 | 16.89 | 4.22 | -1.03 |
| CeCo | 0.60 | 5.13 | 6.40 | 5.70 | 20.66 | 5.17 | 0.57 |
| PrCo | 0.94 | 6.96 | 8.10 | 5.32 | 16.08 | 4.02 | -1.30 |
| NdCo | 0.93 | 7.09 | 0.24 | 0.15 | 0.47 | 0.12 | -0.03 |
Table 1. Comparison of calculated and experimental for different rare-earth permanent magnets. For realistic comparisons, the mumax values are scaled by a factor of in column . The column presents the difference between and .
Artificial neural networks
An artificial neural network (ANN) is suitable for high complexity problems such as prediction. Figure 2 shows a schematic diagram (a portion used in the calculations) for an ANN consisting of an input layer (6 neurons), two hidden layers (containing 4 and 3 neurons respectively), and the output layer with one neuron. A non-linear activation function leaky-relu [] is used between the input and hidden layers to counter typical convergence problems found in small ANNs. We used the state of the art gradient descent benchmark optimizer Adam99, as implemented in tensorflow100 in the Keras API101.

Figure 2. Schematic diagram showing an artificial neural network (only a portion used in the calculations) with 6 neurons in the input layer, 4 and 3 neurons in the second and the third deep (hidden) layers, and a neuron for regression in the output layer. Two different ANN’s were used in actual calculations, as discussed in the main text.
Performance comparison
For the regression model, given an ML fit, we use two common statistical measures of error applied to the predictions: (i) Mean Square Error (MSE):
which is also used for training all ML models, and (ii) R-squared ():
where is the variance written as
with as the mean of the predicted values. Equivalently, we may write , where is the Residual Sum of Squares , and is the Total Sum of Squares . The loss function is usually written as a sum of the MSE and the first or second-order norms of the target function parameters. This provides convergence of the model, and penalizes overfitting for parameterized machine learning.
Results and discussion
Machine learning on experimental data
Here, we explore the experimental data and the relation between target and independent variables. Figure 3 visualizes as a function of , , and . The actual values of the variables differ by several orders of magnitude. They are scaled using the sci-kit minmaxscalar in which a dimensionless scaled feature variable in the range [0,1] is obtained using the following relation:

Figure 3. The distribution of as a function of , , and in experimental magnetic materials. The physical quantities are rendered as normalized and dimensionless. The color map shows the magnitude of .
where and are minimum and maximum values of the feature variable . Figure 3 shows an increase of with , a decrease with , and very little variation with .
| Model | MSE | |
| Linear regression | 0.62 | 0.64 |
| Lasso | 0.38 | 1.07 |
| Ridge | 0.63 | 0.64 |
| Lasso-CV | 0.61 | 0.68 |
| Ridge-CV | 0.63 | 0.66 |
| Elasticnet | 0.62 | 0.66 |
| DT regressor | 0.89 | 0.19 |
| DT pruned | 0.84 | 0.27 |
| RF regressor | 0.87 | 0.23 |
| gradient boosting (GB) regressor | 0.80 | 0.34 |
| Tuned GB regressor | 0.87 | 0.22 |
| XGB regressor | 0.87 | 0.23 |
| Tuned XGB regressor | 0.89 | 0.18 |
| ANN (Adam) | 0.64 | 0.62 |
| light gradient boosted machine (LGBM) | 0.70 | 0.60 |
| Fine tuned ANN | 0.85 | 0.25 |
Table 2. Comparison of and MSE metrics in different ML models on test data sets for experimental materials. The non-linear regression models show better values of and MSE than the linear models. The error is evaluated over the difference of the logarithm of the coercivity in T.
In gradient-based algorithms, such as ANN, additive relationships are easier to model than multiplicative ones, so the independent and dependent variables are logarithmically scaled. Accordingly, the measures of model performance are given in terms of the logarithm of the . As a result of this change, the absolute measure of is increased by for all gradient-based models.
Figure 4 shows a comparison of ML and experimental for the test dataset. The XGB-tuned model shows better performance than ANN-Adam. Table II shows the performance of various ML models. The linear models show poor performance, with . This demonstrates the complex non-linear relation between the dependent and independent variables. The non-linear models DT, XGB, and RF perform better, with values of 0.89, 0.87, and 0.87, respectively. These values still witness a decrease from the scores of more than 0.9 on the training dataset.

Figure 4. ML predicted and experimental coercivities of magnetic materials obtained with tuned XGB (top) and ANN (bottom). Selected key materials are labeled in the figure. Non-linear regressors all yield similar results.
The ANN model has scores that are similar to other classical models (see Table II). The ANN architecture using (dense layers=64, activation function=leaky_relu, dense layers=32, activation function=leaky_relu, dense layers=16, activation function=leaky_relu, and output layer=1, and learning rate=0.007, epochs=1000, batch size=24) showed similar scores to RF. This weak performance of the ANN architecture is attributed to the smallness of the dataset size.
Feature importance. The model performance can be further validated from the dependence of on the independent features. We compute the importance of RF features by utilizing the model-agnostic interpretive features of scikit-learn on the RF and XGB regression models10,2. The feature importance is computed by the
Gini index
where is the probability of class in the data. Figure 5 depicts the relative importance of the independent parameters for , as computed by the XGB model. ML accurately identifies as the leading contributing feature to . More surprisingly, the effect of is about four times smaller than that of . The uneven distribution of magnetic materials (shown in Figure 3), and a constant noise level, may explain the unequal weighting of and . Finally, we note that has a comparatively minimal effect on .

Figure 5. Histogram of important features contributing to in the XGB model. The clearly leads, followed distantly by and .
Machine learning prediction from micromagnetic data
In order to explore the relation between and the independent variables, we used 8770 micromagnetic simulation-generated data points as a training set. The correlation between variables is given by Pearson’s correlation coefficient, which is defined as
where and denote the variable and its average value. A positive (or negative) value of indicates a positive (or negative) correlation between the two variables: the higher the -value, the higher the correlation. Figure 6 shows a correlation heatmap for the variables. The distribution for the 8770-data set is right-skewed, as shown in Figure 7, with a peak around 0.2–0.4 T. It is similar to the distribution for the experimental materials (not shown here): most of the materials have values around the peak, and a few materials, such as 1 : 5 compositions, have larger values of 2–6 T. For better ML training, logarithmic transformation is appropriate for skewed data, which is employed in our model fits.

Figure 6. Heat map showing the correlation between the , , , and . The color palette corresponds to the correlations between the variables. is strongly correlated with , negatively correlated with , and weakly correlated with . shows a slightly negative correlation with and . shows a slightly positive correlation with .

Figure 7. Distribution of in the training set of 8770 data points. The -axis is scaled by . The data points are right-skewed.
Significantly, the strongest (absolute) correlation is observed between and at 0.66, followed by with a correlation of −0.62, and then by at 0.2. This pattern suggests a high dependence of on . The negative correlation between and aligns with the theoretical relationship . Similarly, the correlation coefficient () between and or validates the proportionalities and . These findings are consistent with the trends observed in actual materials, as illustrated in Figure 3.
Next, we discuss the model performance in micromagnetic simulation data, as given in Table III. Both linear and non-linear models perform well, as the score is above 0.9. This is expected in a linear regression model where multivariate analysis demonstrates very little multicollinearity in randomly generated independent variables. The multicollinearity can be measured with the variance inflation factor (VIF) , where
| Model | ||
| Linear regression | 0.95 | 0.22 |
| Lasso | 0.54 | 2.13 |
| Ridge | 0.95 | 0.22 |
| Lasso-cv | 0.95 | 0.25 |
| Ridge-cv | 0.95 | 0.22 |
| Elasticnet | 0.94 | 0.26 |
| DT regressor | 0.97 | 0.13 |
| DT pruned | 0.97 | 0.15 |
| RF regressor | 0.98 | 0.08 |
| GB regressor | 0.98 | 0.09 |
| Tuned GB regressor | 0.98 | 0.08 |
| XGB regressor | 0.98 | 0.98 |
| Tuned XGB regressor | 0.98 | 0.08 |
| ANN | 0.98 | 0.09 |
Table III. TABLE III. data on hypothetical materials: Comparison of and MSE of for different ML models. All the errors are evaluated over the of the in T. The fine-tuned model hyperparameters are given in Appendix A.
is the [Eq. (10)] value of considered as a function of variable . The computed VIF factors are 1.53 for , 4.36 for , and 1.53 for .
More advanced decision tree regressors improve the performance. XGB, as the most accurate model, was used as a benchmark, with the hyperparameters given in Appendix A for prediction from experimental data. The deviance in the MSE of the training and testing data sets is shown in Figure 14. ANNs exhibit performance similar to other non-linear regressors. The ANN is as follows: (dense layers=64, activation function=leaky_relu, dense layers=32, activation function=leaky_relu, dense layers=16, activation function=leaky_relu, and output layer=1, and learning rate=0.007, epochs=1000, batch size=56). To obtain a deeper insight into ML predictability, the experimental, scaled-micromagnetic, and ML-

Figure 14. Deviance as a function of boosting iterations for the micromagnetic simulated dataset with XGB model. Deviance is MSE between the actual and predicted data set computed in each iteration. Both the training and test data set converge well after 600 boosting iterations.
predicted values are compared with only the known important rare-earth-based materials in Table IV. In general, ML-predicted values are smaller than the micromagnetically computed values. For 2 : 14 : 1 rare-earth magnets, ML improves the results. For 1 : 5 rare-earth magnets, it slightly underestimates the experiment. The full comparison of experimental and ML predicted data is given in Figure 8

Figure 8. Scatter plot showing the comparison of experimental and ML-mumax (ML-micromagntics data) predicted for experimental materials.
| Material | ||||
| LaFeB | 0.52 | 0.52 | 0.48 | 0.53 |
| CeFeB | 0.54 | 0.84 | 0.73 | 0.79 |
| PrFeB | 1.54 | 1.59 | 1.54 | 1.62 |
| NdFeB | 1.24 | 1.50 | 1.45 | 1.56 |
| GdFeB | 0.23 | 0.56 | 0.52 | 0.60 |
| TbFeB | 2.93 | 4.00 | 5.65 | 5.41 |
| DyFeB | 1.70 | 2.76 | 3.80 | 3.70 |
| HoFeB | 0.75 | 1.50 | 1.76 | 1.82 |
| LuFeB | 0.29 | 0.59 | 0.52 | 0.58 |
| YFeB | 0.19 | 0.57 | 0.48 | 0.54 |
| ThFeB | 0.17 | 0.56 | 0.48 | 0.53 |
| LaCoB | 0.34 | 0.65 | 0.68 | 0.77 |
| PrCoB | 2.50 | 2.28 | 2.24 | 2.36 |
| NdCoB | 3.69 | 1.00 | 0.90 | 1.06 |
| GdCoB | 0.30 | 0.67 | 2.79 | 2.37 |
| YCoB | 0.34 | 0.62 | 0.61 | 0.71 |
| NdLaCeFeB | 0.65 | 1.33 | 1.25 | 1.39 |
| LaCeYFeB | 0.41 | 0.99 | 1.26 | 1.39 |
| NdPrFeB | 0.80 | 1.33 | 1.27 | 1.39 |
| SmCo | 1.25 | 1.60 | 1.53 | 1.62 |
| SmFeN | 2.30 | 2.46 | 3.13 | 2.95 |
| SmCo | 7.50 | 6.76 | 6.66 | 9.70 |
| YCo | 3.90 | 2.63 | 2.68 | 3.33 |
| LaCo | 5.25 | 4.18 | 4.14 | 4.22 |
| CeCo | 5.70 | 4.10 | 5.10 | 5.17 |
| PrCo | 5.32 | 4.05 | 3.81 | 4.02 |
| NdCo | 0.15 | 0.12 | 0.10 | 0.12 |
| RMSE | 0.90 | 1.10 | 1.10 | |
| MAE | 0.68 | 0.78 | 0.80 |
Table IV. TABLE IV. (in T) comparisons among experiment, ML prediction, and scaled for different permanent magnetic materials. The ML (tuned XGB and ANN) predictions are obtained by training the hypothetical materials using the random independent variables , , and , and the computed . RMSE and MAE quantify the discrepancy between the experimental ML and the scaled .
Additionally, XGB was trained with different fractions of the dataset, from 200 to 12000, by computing the mean absolute error (MAE) error metric:
Figure 9 shows the convergence of the MAE (obtained with the XGB model) plotted against the training data set size, indicating that the model performs well for datasets with more than 6000 points.

Figure 9. MAE convergence for with respect to training data set size in XGB model. The -axis is in the scale of .
Next, we discuss the loss function RMSE to examine the model performance. We computed the root mean square error (RMSE) as
where and refer to the experimental and ML/ predicted values. Remarkably, the RMSE and MAE between experimental and ML (tuned XGB) predicted values are smaller than the RMSE between experimental and the scaled-, indicating that ML helps to correct the predictions. It also shows how the is correlated with other features. Moreover, ML is an order of magnitude computationally faster than , and we can use the trained model for predicting in new materials without any calculations.
ML application with DFT
Our computational predictions involve two steps: DFT for Ce-doped NdFeB, and then ML trained on micromagnetically generated databases with input parameters from the DFT computations.
DFT methods and crystal structure
We used the Vienna simulation package (VASP)103,104 with the projector augmented wave (PAW) formalism in the generalized gradient approximation (GGA) of Perdew-Burke-Ernzerhof (PBE) semi-local exchange-correlation functionals105,106 including onsite electron-electron correlation Dudarev-Hubbard107 for the states of Nd, and spin-orbit interaction. The kinetic energy cut-off is 520 eV for the plane wave expansion, with a -mesh for the Brillouin zone integration. The experimental structure was used in calculations to avoid overestimating bond lengths lengths and lattice constants in the PBE.
The crystal structure of NdFeB is tetragonal with 68 atoms (4 formula units), having space group (136) and lattice constants and Å, taking at a temperature K108, with experimental parameters for full Ce-doped NdFeB109. The crystal structure of the 2:14:1 neo-magnet is shown in Figure 10, which consists of rare-earth Nd in the 2 inequivalent sites and , transition element Fe in the 6 inequivalent sites at , , , , , and , and finally B at the sites. Fully self-consistent spin-orbit calculations, i.e., PBE++spin-orbit coupling (SOC), were performed to obtain the magnetic anisotropy energy along with the spin and orbital magnetic moments.

Figure 10. Crystal structure of NdFeB occupied by Nd at two non-equivalent sites 4 and 4, Fe at six inequivalent sites 16, 16, 8, 8, 4, and 4 (labeled by Wyckoff positions), and B at the 4 site.
Magnetic properties
Here we discuss the PBE++SOC results with NdFeB, using ML for prediction. Hubbard -correction is necessary for highly correlated 3 and 4 elements110–113 to accurately predict . We performed test calculations for various values of from 4–7 eV for Nd 4 orbitals. Table V shows the spin and orbital magnetic moments of Nd, Fe, and B at different crystallographic sites for different values of . The non-magnetic B atom carries negligible magnetic moments. Nd exhibits an Nd state with spin magnetic moment () of and strongly quenched orbital moment () of due to the crystalline electric field. The spin and orbital moments have opposite signs, consistent with Hund’s rule for a less than half-filled 4 shell. The net magnetic moment is robust, and does not vary significantly with .
Table VI lists , , , and their comparison with experimental values. is positively correlated with . The experimental is 4.5 MJ/m114 at 300 K, which corresponds to the computed value for eV. is computed with the static Green’s function (GF) as implemented in ASA in the local density approximation (LDA)115 for the exchange-correlation functional in the linearized muffin-tin orbital (LMTO) program116,117. The pair exchange interaction between magnetic and ions is computed using the Lichtenstein formula118, from which is estimated. The calculated Weiss mean-field theory119 value of 718.1 K is larger than the experimental value, which is expected in the mean-field approximation (MFA). According to a spin-wave theory by Tyablikov120`` the random-phase approximation (RPA) corrects the value, bringing it down to 539.1 K. The experimental value lies within the MFA and RPA limits.
Next, we discuss the site-resolved spin-orbit anisotropy energy which is computed as 111, where and are PBE++SOC-computed atomic site energies for spin-quantization along the [100] and [001] directions. Generally, 4-elements contribute to , and 3-elements contribute to the magnetic moments in rare-earth-based magnets. The spin moments of the rare-earth ion are anti-parallel to the spin moment of the 3 ion, which is also the case for neo-magnets. The site contribution of the Nd-element to crystalline anisotropy energy is given in Table VII. In our PBE++SOC calculations, we did not impose symmetry, which means that all the atoms are inequivalent under the crystal symmetry. All eight Nd atoms split into eight different sites. The individual on-site energies differ significantly for the atoms of the same crystallographic site. Interestingly, at least one Nd (Nd) at 4 has a negative contribution to crystalline anisotropy energy, as given in Table VII. Theoretically, it can be inferred that Nd has a tendency to be planar at the 4 site, differing from the 4 site, which is strictly uniaxial along the crystalline direction. These results are consistent with experiment121.
Table VIII shows the computed values of the magnetic moments and magnetic anisotropy in the Ce-substituted neo-magnet at Nd-4 sites. The Ce-atom carries a small spin magnetic moment compared to the Nd moment. Moreover, the orbital moment does not entirely cancel the spin moment in Nd, unlike in Ce, where the net spin + orbital moment vanishes. Therefore, there is an increase in the net magnetic moment with Ce. On the other hand, magnetic anisotropy is reduced because the Ce-site contribution is much smaller than that of Nd. The calculated value of slightly underestimates the experiment (see Table I), which is reasonable, given the choice of the -values used for the Ce and Nd 4 states. We used eV for Ce and 4 eV for Nd in the calculations.
ML coercivity prediction
Central to this paper is the hypothesis that macroscopic measures of are not well predicted by theory, while the microscopic magnetic parameters , , and are. However, knowledge of the microscopic parameters is sufficient to produce realistic estimates of by accumulating a database and modeling it with ML. Furthermore, once an ML prediction is made, information about the material grain structure may be reverse-engineered with micromagnetic modeling. Although material fabrication variations have a significant impact on , this section demonstrates the surprising effectiveness of the methodology as applied to NdFeB, with and
| atom | Nd() | Nd() | Fe() | Fe() | Fe() | Fe() | B() | total |
| 4 eV | -3.26 1.52 | -3.28 1.53 | 2.27, 2.36 0.043, 0.049 | 2.29, 2.71 0.044 | 2.03 0.045 | 2.48 0.054 | -0.17 0.00 | 105.24 14.75 |
| 5 eV | -3.26 1.52 | -3.28 1.53 | 2.27, 2.36 0.042, 0.048 | 2.29, 2.70 0.044 | 2.03 0.045 | 2.47 0.052 | -0.17 0.00 | 105.25 14.75 |
| 7 eV | -3.27 1.52 | -3.28 1.53 | 2.27, 2.36 0.042, 0.048 | 2.29, 2.71 0.042 | 2.03 0.044 | 2.48 0.049 | -0.17 0.00 | 105.25 14.68 |
Table V. Spin and orbital magnetic moments ( and ) of individual atoms in different Wyckoff positions, and total magnetic spin and orbital magnetic moments ( and ) per unit cell (in ), in NdFeB computed using the PBE++SOC method with various values of for the Nd () electronic states. There are two and sites, therefore, the pairs are given. The of the pair at -sites do not differ, and only one value is given to represent both atoms.
| 4 eV | 6.16 | 29.98 | 37.7, 32.5 | 718.1, 539.1 | 585 |
| 5 eV | 8.56 | 30.00 | |||
| 7 eV | 11.58 | 30.01 |
Table VI. Calculated in MJ/m, in /f.u. using PBE++SOC, in Kelvin (K) using Green’s function-ASA (LDA), and comparison with available experiment.
| Properties | NdFeB | CeFeB |
| 105.24 | 120.73 | |
| 14.75 | 4.55 | |
| 119.99 | 125.28 | |
| 6.16 | 1.21 | |
| 1.31 | 1.24 |
Table VIII. Total spin, orbital, and spin + orbital magnetic moments , , and in /cell, in MJ/m, and in MA/m of CeFeB and NdFeB.
| Atom | Wyckoff-position | |
| Nd | 1.3780 | |
| Nd | -1.1914 | |
| Nd | 0.5336 | |
| Nd | 6.2204 | |
| Nd | 0.4865 | |
| Nd | 3.8064 | |
| Nd | 6.2094 | |
| Nd | 3.0622 |
Table VII. Site resolved spin-orbit anisotropy energy () in meV of the Nd contribution to magneto-crystalline anisotropy energy. In PBE++SOC calculations, the crystalline symmetry lowers to the point group implying that the atoms (including the Nd) split into eight sites. The negative value for Nd at the site indicates it produces a planar contribution to the crystalline anisotropy energy, consistent with experiment121.
without Ce doping.
The ab initio-calculated values were fed to the tuned XGB ML model to produce the output predictions. Additionally, micromagnetic simulation was used to determine a hysteresis loop. The results are depicted in Table IX. As can be seen, the reduction trend of
| Material | ML-Exp | ML-mumax3 | |
| NdFeB | 3.92 | 2.39 | 2.35 |
| CeNdFeB | 0.88 | 1.27 | 1.29 |
| CeFeB | 0.39 | 0.40 | 0.36 |
| SmCo | 8.87 | 5.88 | 18.12 |
| YCo | 1.45 | 1.51 | 1.60 |
| CeZrFe | 6.25 | 4.93 | 5.99 |
| CeZrFeN | 1.51 | 2.15 | 2.59 |
| CeTiFe | 1.50 | 2.09 | 2.70 |
| CeTiFeN | 1.51 | 1.92 | 2.45 |
Table IX. Comparison of ML predicted and scaled-micromagnetic () in T for materials with calculated DFT parameters. ML-exp/ML-mumax3 represent to predictions made by training experimental/scaled-micromagnetic data.

Figure 11. Hysteresis curve for NdCeFeB (, 1.0, 2.0) calculated from micromagnetic simulation. Ce leads to the reduction in the coercivity. is the reduced magnetization.
with Ce is similar to that found experimentally122. Further, we computed the ML [trained with experiment (ML-Exp) and scaled micromagnetic data (ML-mumax) separately] and scaled micromagnetic simulated for SmCo, YCo, CeZrFe, CeZrFeN, CeTiFe, and CeTiFeN using DFT intrinsic parameters as given in Table IX. The ML-exp predicts inconsistent due to the small data size training, as evident from model performance metrics in Tab II. The ML-mumax trained on big data sets produces more sensible than scaledmicromagnetics and ML-exp. For SmCo, micromagnetic simulation yielded a factor of three larger than the ML-mumax3 model and experiments. For YCo, the ML-mumax3 model produced slightly lower than the experiment, attributed to the DFT underestimate of . Overall the ML-mumax3 model consistently tends to correct the micromagnetically predicted towards experiments for known materials (see Table I).
If the fabricated material is grown as a crystal with well-defined grains, it is possible to reverse-engineer the structure. The dependence of on micromagnetic parameters viz, grain sizes and inter-grain coupling is visualized in Figure 12 in 128 × 128 × 128 nm magnetic cell.

Figure 12. Coercivities for different inter-grain couplings for NdFeB, measured as a function of the ratio of inter-grain exchange stiffness to intra-grain exchange stiffness and different grain sizes. As the material is complex, a cuboid of 128 × 128 × 128 nm was generated for each grain size, resulting in some random undulations according to grain initialization. The plot displays a decreasing coercivity with increasing grain size as observed in experiment122.
Here the grain coupling refers to a reduction factor to the stiffness exchange coupling . We note that we did not find much difference in the results with the use of 64 × 64 × 64 and 128 × 128 × 128 nm simulation cells. Although non-linear, we find a similar trend of with inter-grain exchange coupling and grain sizes. For a fixed value of grain size, qualitatively, decreases with inter-grain exchange coupling, while it is the opposite with grain size. That inter-grain exchange coupling may not be uniform in actual material due to the void or imperfection between the grains, which affects the perfect spin alignment resulting in . A larger inter-grain exchange coupling further reduces and . With an increase in grain size, experiences a decrease, similar to that reported in the Dy-substituted neo-magnet123.
Various combinations of grain size and inter-grain exchange coupling reproduce the actual coercivity. For instance, with grain size nm, and a 0.58 ratio for inter-grain to intra-grain reproduces the realistic of 0.40 T, as demonstrated in Figure 13. Strictly speaking, a larger inter-grain coupling tends to reduce the , regardless of grain size. This suggest that we can explore the other magnetic parameters and grain size by using ML-predicted results employing micromagnetic simulations, which demonstrates the fundamental usefulness of ML, not just as a black box, but also as a pre-processing input to a micromagnetic model to reverse-engineer the micromagnetic structure.

Figure 13. Hysteresis curve for CeFeB calculated naively with micromagnetic simulation with grain size set to 30 nm and intergrain stiffness to intragrain stiffness set at 0.58 to recreate the ML prediction.
Conclusion
Experimental dataset input and ML modeling coupled with DFT predicts with far greater accuracy and speed than was previously possible using micromagnetic modeling. This technique provides a robust computational foundation for predicting novel permanent magnet materials and optimizing their properties. Using micromagnetic simulation, we first studied magnetization as a function of the applied magnetic field for real and hypothetical magnetic materials, mainly focusing on 1 : 5 and 2 : 14 : 1 rare-earth-based permanent magnets such as NdFeB. Calculations of based on the hysteresis loop overestimate by a factor of ; the Brown Paradox. The paradox is side-stepped by the judicious use of ML, obtaining a more accurate prediction of the target variable by learning its dependence on the input features. We find that is directly proportional to , inversely proportional to , and weakly correlated with .
We apply the ML modeling to NdFeB by first computing its independent variables with DFT calculations. These calculations show that the Nd 4g-sites mainly contribute to the uniaxial magneto-crystalline anisotropy. They also yield a value for the , which agrees with the experiment. The DFT predictions suggest the possibility of tuning rare-earth magnetic properties by substituting non-critical elements at specific sites. For instance, Ce-doping at 4f sites shows only a slight reduction in , consistent with the ML prediction. Finally, we engineer the grain boundary size and inter-exchange coupling with the aid of ML-predicted , which indicates that the reduction in inter-grain exchange coupling reduces . On the other hand, reducing the grain size increases qualitatively.
Acknowledgments
This work is supported by the Critical Materials Innovation Hub, an Energy Innovation Hub funded by the US Department of Energy, Office of Energy Efficiency and Renewable Energy, Advanced Materials and Manufacturing Technologies Office. The Ames National Laboratory is operated for the U.S. Department of Energy by Iowa State University of Science and Technology under Contract No. DE-AC02-07CH11358. GNN acknowledges support from the U.S. Department of Energy, Office of Science, Office of Workforce Development for Teachers and Scientists, Office of Science Graduate Student Research (SCGSR) program. The SCGSR program is administered by the Oak Ridge Institute for Science and Education for the DOE under contract number DE-SC0014664.
ml model and hyperparameter tuning
In this appendix we provide information about the hyperparameters used in the fine tuning of the ML models. With the experimental data, we used the hyperparameters given in Table X for tuned XGB model training.
| n_estimators=500 | |
| min_weight_fraction_leaf=0 | |
| max_depth=5 | |
| learning_rate=0.006 |
Table X. Hyperparameters used for XGB model training of experimental magnetic materials.
In mumax data training we use the hyperparameters given in Table XI.
| Model | Hyperparameters |
| Tuned RF | n_estimators=1200, max_depth=5, random_state=1, min_samples_leaf=3, bootstrap=True, max_features=’auto’ |
| Tuned XGB | n_estimators=1200, max_depth=5, min_weight_fraction_leaf=0, learning_rate=0.007 |
| Tuned XGB | n_estimators=1200, max_depth=5, min_child_weight=0, learning_rate=0.007, random_state=42 |
Table XI. Fine-tuned hyperparameters for different ML models used in the training of micromagnetic simulated data.
The hyperparameter used in the fine-tuned GBR model training is shown in Figure 14. The deviance (MSE) converges well in both the test and training datasets after 600 boosting iterations, confirming a very good model performance.
DFT parameters for 1:5 and 1:12 intermetallics
tbl:XII provides the DFT computed parameters for 1 : 5 and 1 : 12 intermetallic compounds.
| Material | (A/m) | (pJ/m) | (MJ/m) |
| SmCo${}^{124}` | 1069768.213 | 7 | 40.27 | |||
| YCo${}^{125}` | 770449.2281 | 7 | 2.65 | |||
| CeZrFe | 597220.2602 | 7 | 7.36 |
| CeZrFeN | 632804.6376 | 7 | 3.42 |
| CeTiFe | 587261.7154 | 7 | 3.29 |
| CeTiFeN | 625792.1247 | 7 | 3.20 |
Table 12. DFT computed intrinsic magnetic parameters for 1 : 12 compositions taken from our previous work113 except for 1 : 5 type materials as referred in Table. is kept fixed for all materials.
References
- [27](1968).
- [28]R. A. Ristau, K. Barmak, L. H. Lewis, K. R. Coffey, and J. K. Howard, On the relationship of high coercivity and L1 ordered phase in CoPt and FePt thin films, J. Appl. Phys. 86, 4527 (1999).DOI
- [29]B. C. Sales, B. Saparov, M. A. McGuire, D. J. Singh, and D. S. Parker, Ferromagnetism of Fe₃Sn and alloys, Sci. Rep. 4, 7024 (2014).DOI
- [30]H. Jaballah, W. Bouzidi, R. Fersi, N. Mliki, and L. Bessais, Structural, magnetic and magnetocaloric properties of (Pr,Sm)₂Fe₁₇ compound at room temperature, J. Phys. Chem. Solids 161, 110438 (2022).DOI
- [31]Y. Iwama, M. Takeuchi, and M. Iwata, New determination of magnetic anisotropy constants of alnico magnet alloys, J. Phys. Colloques 32, C1 (1971).
- [32]Y. Liu, J. Song, G. Liu, J. Chen, C. Wang, H. Wang, J. Wang, and X. Zhang, High strength and low coercivity of cobalt with three-dimensional nanoscale stacking faults, Nano Lett. 21, 6480 (2021), pMID: 34324350.DOI
- [33]H. Sato, T. Yoshioka, H. Tsuchiyura, Y. Mizuno, K. Koike, K. Takahashi, and H. Kato, Rare-earth moment reduction and local magnetic anisotropy in Pr₂Fe₁₄B and Tm₂Fe₁₄B, J. Magn. Magn. Mater. 545, 168684 (2022).DOI
- [34]N. Yu, M. Zhu, L. Song, Y. Fang, K. Song, Q. Wang, and W. Li, Coercivity temperature dependence of Sm₂Co₁₇-type sintered magnets with different cell and cell boundary microchemistry, J. Magn. Magn. Mater. 452, 272 (2018).DOI
- [35]T. Kikuchi, T. Nakamura, T. Yamasaki, M. Nakanishi, T. Fujii, J. Takada, and Y. Ikeda, Magnetic properties of La–Co substituted M-type strontium hexaferrites prepared by polymerizable complex method, J. Magn. Magn. Mater. 322, 2381 (2010).DOI
- [36]K. M. U. Rehman, M. Riaz, X. Liu, M. W. Khan, Y. Yang, K. M. Batoo, S. F. Adil, and M. Khan, Magnetic properties of Ce doped m-type strontium hexaferrites synthesized by ceramic route, J. Magn. Magn. Mater. 474, 83 (2019).DOI
- [37]P. Wang, X. Wang, L. Qiao, J. Zhang, G. Wang, B. Duan, T. Wang, and F. Li, High-frequency magnetic properties and microwave absorption performance of oxidized Pr₂Co₁₇ flakes/epoxy composite in x-band, J. Magn. Magn. Mater. 468, 193 (2018).DOI
- [38]J. Zhang, H. Gao, Y. Yan, X. Bai, F. Su, W. Wang, and X. Du, Morphology and magnetic properties of CeCo₅ submicron flakes prepared by surfactant-assisted high-energy ball milling, J. Magn. Magn. Mater. 324, 3272 (2012).DOI
- [39]A. Gabay, X. Hu, and G. Hadjipanayis, Preparation of YCo₅, PrCo₅ and SmCo₅ anisotropic high-coercivity powders via mechanochemistry, J. Magn. Magn. Mater. 368, 75 (2014).DOI
- [40]Y. Hirayama, A. Panda, T. Ohkubo, and K. Hono, High coercivity Sm₂Fe₁₇N₃ submicron size powder prepared by polymerized-complex and reduction–diffusion process, Scr. Mater. 120, 27 (2016).DOI
- [41]J. Fliegans, Coercivity of NdFeB-based sintered permanent magnets : experimental and numerical approaches, Theses, Université Grenoble Alpes (2019).DOI
- [42]Q. Cheng, J. Lin, and M. Su, On synthesis and magnetic properties of Nd(Fe, Mo, Ti)₁₂Zₓ (Z=N, H), J. Alloys Compd. 280, 310 (1998).
- [43]T. Oyama, ed., The Chemistry of Transition Metal Carbides and Nitrides, 1 (Springer Dordrecht) pp. XXVI, 536.
- [44]T. Wu, S. Cao, M. Kou, Y. Xie, G. Ding, S. Guo, B. Zheng, R. Chen, M. Zhong, and A. Yan, Magnetic performance and microstructure of ndfeb sintered magnet by diffusing Tb₁₀Pr₉₀−ₓ(Cu,Al,Ga)ₓ alloys, J. Alloys Compd. 934, 167888 (2023).DOI
- [45]J. Tsui, K. Strnat, and J. Schweizer, High magnetic coercivity of neodymium- and didymium-cobalt alloys sintered with Pr and Sm additives, APL 21, 446 (2003).
- [46]C. Zhou and F. E. Pinkerton, Magnetic properties of Ce–Nd–Fe–Mo alloys and their nitrides, J. Magn. Magn. Mater. 369, 127 (2014).DOI
- [47]G. Gkouzia, D. Günzing, R. Xie, T. Webels, A. Kovács, A. T. N’Diaye, M. Major, J. P. Palakkal, R. E. Dunin-Borkowski, H. Wende, H. Zhang, K. Ollefs, and L. Alff, Element-specific study of magnetic anisotropy and hardening in SmCo₅−ₓCuₓ thin films, Inorg. Chem. 62, 16354 (2023), pMID: 37739403.DOI
- [48]K. Suresh, R. Gopalan, A. Singh, G. Bhikshamaiah, V. Chandrasekaran, and K. Hono, Coercivity of Sm(Co₀.₉Cu₀.₁)₄.₈ melt-spun ribbons, J. Alloys Compd. 436, 358 (2007).DOI
- [49]J. Téllez-Blanco, R. Grössinger, and R. Sato Turtelli, Structure and magnetic properties of SmCo₅−ₓCuₓ alloys, J. Alloys Compd. 281, 1 (1998).DOI
- [50]M. D. Kuz’min, K. P. Skokov, I. Radulov, C. A. Schwöbel, S. Foro, W. Donner, M. Werwiński, J. Ruzs, E. Delczeg-Czirjak, and O. Gutfleisch, Magnetic anisotropy of La₂Co₇, J. Appl. Phys. 118, 053905 (2015).DOI
- [51]M. Q. Huang, S. G. Sankar, W. E. Wallace, M. E. McHenry, Q. Chen, and B. M. Ma, Structure and magnetic properties of RCo₇−ₓZrₓ (R=Y, Gd, Nd, or Ho, x=0-0.8), J. Appl. Phys. 87, 5305 (2000).DOI
- [52]M. T. Onyszczak, T. N. Lamichhane, S. L. Bud’ko, P. C. Canfield, and A. Palasyuk, Structural and magnetic properties of the CeCo₅−CeZn₅ solid solution and potential improvements upon iron substitution, J. Magn. Magn. Mater. 482, 192 (2019).
- [53]Y. Wong, H. Chang, Y. Lee, W. Chang, C. Chiu, and C. Mo, Coercivity enhancement of thicker sintered ndfeb magnets by grain boundary diffusion with low-melting Tb₇₅−ₓCeₓCu₂₅ (x=0–45) alloys, J. Magn. Magn. Mater. 515, 167287 (2020).DOI
- [54]W. Silva, N. Ferreira, J. Soares, R. da Silva, and M. Macêdo, Investigation of structural and magnetic properties of nanocrystalline mn-doped SrFe₁₂O₁₉ prepared by proteic sol–gel process, J. Magn. Magn. Mater. 395, 263 (2016).DOI
- [55]S. Katlakunta, S. S. Meena, S. Srinath, M. Boududina, R. Sandhya, and K. Praveena, Improved magnetic properties of Cr³⁺ doped SrFe₁₂O₁₉ synthesized via microwave hydrothermal route, Mater. Res. Bull. 63, 58 (2015).DOI
- [56]S. Chawla, S. Meena, P. Kaur, R. Mudsainiyan, and S. Yusuf, Effect of site preferences on structural and magnetic switching properties of Co-Zr doped strontium hexaferrite SrCoₓZrₓFe₍₁₂−₂ₓ₎O₁₉, J. Magn. Magn. Mater. 378, 841 (2015).
- [57]R. Alange, P. P. Khirade, S. D. Birajdar, A. V. Humbe, and K. Jadhav, Structural, magnetic and dielectrical