Introduction

In alloys, the segregation of solute atoms at grain boundaries (GBs) induces structural effects1,2 that include strengthening3,4,5, embrittlement6,7, corrosion resistance8,9, and GB phase transitions10,11. As such, controlling GB segregation is an essential tool for many engineering applications12, including, e.g., the thermodynamic stabilization of nanocrystalline alloys against grain growth13,14,15. And, though most technically relevant alloys are used in a polycrystalline form, there is a very limited understanding of GB segregation in polycrystals16, and a general lack of databases of segregation information relevant to them.

In a polycrystal, the GB network has a variety of site-types that can either promote or inhibit segregation to different degrees, depending on their unique local atomic environments. The drive for a solute atom to segregate to a GB site-type (i) is quantified by the segregation enthalpy \({\Delta}H_i^{{\rm{seg}}}\), which, in solids17, is equivalent to the internal energy difference between the solute atom occupying the GB site, and a bulk (intra-grain) site, \({\Delta}H_i^{{\rm{seg}}} \approx {\Delta}E_i^{{\rm{seg}}} = E_{{\rm{gb}},i}^{{\rm{solute}}} - E_{\rm{c}}^{{\rm{solute}}}\); a negative \({\Delta}E_i^{{\rm{seg}}}\) (i.e. the system reduces energy upon segregation) promotes segregation and vice versa. The spectrum of \({\Delta}E_i^{{\rm{seg}}}\) in a polycrystal will determine the extent of equilibrium GB segregation in an alloy18,19,20. Recently, we have shown this spectrum to be captured by a skew-normal distribution for an Mg solute segregation in an Al polycrystal20. However, the computation of these segregation spectra is a resource-intensive task. For example, a (50 nm)3 Al polycrystal with an average grain size of 10 nm has roughly one million GB sites, which translates to a million atomistic calculations, where a solute atom is placed substitutionally at each GB site independently and allowed to relax. This makes the task of investigating different microstructures (i.e. multiple polycrystalline samples) cost-prohibitive for a given alloy.

Here, we propose a machine learning (ML) framework that can accurately predict the relaxed segregation energy of a solute atom in a GB site, solely based on its undecorated (pre-segregation) atomic environment. Our approach is tiered and offers two models. The first is a high-fidelity model that is trained to accurately capture the variation of segregation energy across a large swath of the GB space, and thus can be used to study an alloy system in detail and instantaneously evaluate segregation for different microstructures. The second is an accelerated model that uses dimensionality reduction to reproduce the high-fidelity model—with a minimal loss in accuracy—using three orders of magnitude fewer data-points for training (only 100 sites). We use the accelerated approach to scan across the alloy space, and build an extensive database giving GB segregation spectra for all aluminum, magnesium, and transition metals-based binary alloys for which an interatomic potential exists in the Interatomic Potentials Repository21,22 of the National Institute of Standards and Technology (NIST) - a total of 259 binary alloys. This database allows us to identify alloys of interest with minimal computational cost, for which high-fidelity models can be trained and used. The proposed ML framework and the resulting spectral segregation database should provide a general and broadly applicable alloy design toolbox relevant to all material properties impacted by solute segregation.

Results

High-Fidelity ML model for GB segregation

If a solute atom is substitutionally placed at a GB site and is allowed to relax, its local neighboring atoms will be affected (and possibly displaced) by the introduced elastic and chemical interactions. Hence it follows that the local atomic environment (LAE) of a GB site will influence its favorability for solute segregation, and thus this environment should be accurately captured in any learning model that aims to correlate the undecorated (pre-segregation) GB site to its final decorated (post-segregation) relaxed state. So far, the state-of-the-art learning models in the literature use simple well-known structural features23,24, such as atomic volume, coordination, and Voronoi parameters, which mostly limit the description of the LAE to its first nearest-neighbor atoms. Instead, we propose using an atom-centered feature extraction method “descriptor” that encodes the local atomic environment around an atom within a cutoff radius25,26. Such descriptors—also known as “fingerprints”—are developed and widely used to construct ML-based interatomic potentials; examples include the atom-centered symmetry functions27, bispectrum components28,29, and smooth overlap of atomic positions (SOAP)30. There are two main advantages to using such atom-centered descriptors. The first is that no a priori knowledge or selection of what constitutes an important structural feature of the LAE (such as volume, coordination, etc.) is required, but rather, by using a complete description of the LAE within a cutoff radius, we relegate the decision of learning the most important features to the ML model. The second is that the use of a large cutoff radius ensures that the most dominant interactions between the solute atom and its LAE are captured. As these descriptors are borrowed from the interatomic potential fitting literature, we can think of our approach as fitting a “pseudo interatomic potential” for solute segregation at GBs.

The proposed high-fidelity ML model is summarized in Fig. 1, which shows two main steps: (a) feature extraction and (b) a learning algorithm. For feature extraction, we use the SOAP method30, as it was recently shown to perform well in describing GB environments (albeit for the different problem of predicting GB energies)31. The SOAP method produces for a given GB site and its LAE within a cutoff radius, a feature vector (descriptor) that is invariant under all physical symmetries (permutation, translation, rotation, etc.). The size of the feature vector is controlled by the SOAP hyperparameters (detailed in the methods section), which, in essence, determines the resolution of the vector and its sensitivity to changes in the LAE. In this work, the SOAP feature vector for each GB site has FSOAP=1,015 features. For the cutoff radius, we use 6 Å, which is a conservative cutoff used in constructing interatomic potentials, as it captures the most dominant atomic interactions for an atom with its LAE25,32. We note that, though we opted to use the same FSOAP and a radial cutoff of 6 Å for all binary alloys (as optimal parameters that require minimal input from the user), this procedure is flexible, and one could, by further optimizing the SOAP hyperparameters to the specific alloy of interest, improve the accuracy of the ML model. (For example, a solute atom that has a large size mismatch with solvent atoms could benefit from a larger radial cutoff.)

Fig. 1: High-fidelity ML model to learn Mg solute36 segregation at GBs in an Al polycrystal.
figure 1

For the (a) 20x20x20 nm3 thermally annealed Al polycrystal with 16 grains (colored by the centro-symmetry parameter74), (b) the LAE of every identified GB atom is transformed into a feature vector, using the SOAP method30 with rcutoff = 6 Å, to construct a feature matrix for the full GB network (NGB atoms x FSOAP features), which is used as the input to the (c) learning algorithm (linear regression) to learn Mg GB segregation energies, using a 50/50 training/testing split. d Predictive performance—mean absolute error (MAE)—of the trained ML model across the full GB network.

Full size image

The product of the first step of the ML framework, feature extraction, is a feature matrix of size (NGB atoms x FSOAP features), which is used as the input to the second step, the learning algorithm, which learns to map the input SOAP features to the target property (segregation energy). For the learning algorithm, we use linear regression for three reasons: first, it is a simple inexpensive model to train and use for predictions; second, it can be automated as it does not require any hyperparameter optimization; and third, it inherently ensures regularization (i.e. is less prone to overfitting)—by simply using a sample size of >10xFSOAP GB sites (following the “one in ten” rule of thumb33, which we further validate in Supplementary Fig. 21) to fit the F + 1 coefficients of the model (F coefficients + intercept), we guard against model overfitting, and selection bias towards a small subset of the population (randomly sampling as few as ~400 points from an infinite population gives a 95% confidence level and 5% margin of error34). We note that although more elaborate learning algorithms could be used, such as support vector machines31, Gaussian process regression28 or neural networks32, our proposed ML framework prioritizes simplicity and minimal input from the user, so that other researchers can adopt it easily. We use this approach to showcase the utility of using atom-centered descriptors for learning GB site segregation energies, without getting lost in the intricate details of fine-tuning more advanced learning algorithms. We note that though the proposed learning framework focuses on segregation spectra in substitutional alloys, it is extensible in principle to interstitial alloys by defining interstitial sites35 at the GB and bulk regions.

Using the high-fidelity approach, we train a model, in Fig. 1c, for Mg solute36 segregation in a thermally annealed 20 × 20 × 20 nm3 Al polycrystal that has 16 grains and ~105 GB sites, using a randomized 50/50 split for training/testing. This simple holdout method is easy/cheap to train and use, and its conservative test ratio will guard against a high variance model in most cases. The trained model is highly accurate, with a mean absolute error (MAE) of 2.4(2.5) kJ/mol for the train(test) datasets, respectively, and a root-mean-square error of RMSE=3.8(4.1) kJ/mol. The model faithfully reproduces the distribution of segregation enthalpies in the polycrystal and has a well-behaved error with normally distributed residuals. This result compares favorably with a more sophisticated ML model by Huber et al.23, which used 19 structural features (volume, coordination, Voronoi analysis parameters, and Steinhardt bond-order parameters) with gradient boosted decision trees, and had a 9-fold cross-validation RMSE=7.7 kJ/mol for Mg solute segregation in a database of 38 low and high-symmetry boundaries in Al. The comparison is not direct, of course, since that work focused on bi-crystals whereas we use polycrystals, but it is also encouraging that the present error is also much lower than the reported error of the interatomic potential as compared to DFT GB segregation energies, which has an RMSE of 8.7 kJ/mol24.

We further validate the efficacy of the high-fidelity ML model for GB solute segregation across the alloy space by training to six more 20 × 20 × 20 nm3 polycrystalline volumes for different alloys: Ag(Ni)37, Cu(Zr)38, Fe(Al)39, Ni(Cu)40, Pt(Au)41, and Zr(Ni)42. As shown in Fig. 2, the ML model accurately reproduces the segregation spectra for the six binary alloys, and has a low MAE typically below ~6 kJ/mol and often below 1 kJ/mol. Alloys with higher absolute values (wider distribution) for the segregation energy will correspondingly have a higher MAE, and the worst of these seen here is MAE = 12.8 kJ/mol for the Zr(Ni) system, but here the segregation spectrum spans about 250 kJ/mol; as a fraction of the total spread of the segregation spectrum, the MAE is uniformly below about 5%. To test the extrapolability of the high-fidelity framework, we report the mean (and standard deviation) absolute errors using 5-fold cross-validation in Supplementary Table 1, which shows that the fitted models are able to generalize well to the unseen folds of the dataset (with similar errors as reported in Fig. 2 for the 50/50 holdout method, and low standard deviation across the folds). We note that although most of the surveyed base-metals have fcc lattice structure, the ML framework seems to be insensitive to the lattice structure, as it similarly performs well for bcc (Fe), and hcp (Zr) metals. Therefore, we conclude that the high-fidelity ML model can be used to accurately model GB segregation across the GB and alloy spaces.

Fig. 2: Validation of the high-fidelity ML framework across the alloy space.
figure 2

The predictive performance—mean absolute error (MAE) –of trained ML models, as outlined in Fig. 1, for solute segregation in six 20x20x20 nm3 polycrystalline alloys: a–f Ag(Ni)37, Cu(Zr)38, Fe(Al)39, Ni(Cu)40, Pt(Au)41, and Zr(Ni)42.

Full size image

Accelerated ML model for GB segregation

In alloy design, it is of interest to be able to quickly scan across the alloy space for interesting combinations. In the context of GB segregation, for example, significant efforts have been conducted to screen for nanocrystalline stabilizing elemental combinations15,43, complexion forming combinations44, or GB embrittling solute additions45,46,47. Though the high-fidelity ML model is highly accurate, it still requires ~104 data points for training and fitting its ~103 coefficients (features). To reduce the training cost and permit a broader scan across the full alloy space, it is appropriate to reduce the dimensions of the input features. We propose the use of unsupervised dimensionality reduction algorithms, which map a high dimensional feature vector into a low-dimensional embedding that captures its main characteristics; “unsupervised” signifies that such mapping is done without a priori knowledge the of the target value (segregation energies). As an illustration, we adopt the simplest of these algorithms, namely principal component analysis48, which we use to transform the FSOAP = 1015 into 10 principal components (PSOAP) that maximize the captured variance of the feature space. We can think of this process as compressing the 1,015 features into 10 components; such compression captures >99% of the variance of the SOAP feature matrix of the Al polycrystal, as shown in Fig. 3.

Fig. 3: Accelerated ML model for GB segregation.
figure 3

For the Al(Mg) alloy, shown in Fig. 1, (a) principal component analysis is used to reduce the dimensionality of the feature matrix by projecting the FSOAP features into P = 10 components that capture >99% of the variance. b k-means clustering is then used to divide the 10-d transformed feature space into P × 10 = 100 similar clusters; the closest GB sites to the cluster centers (shortest Euclidean distance) are used as training data-points for the learning algorithm (linear regression). c Predictive performance—mean absolute error (MAE)—of the accelerated model across the full Al GB network.

Full size image

For an accelerated option of the ML framework, we propose using the 10 principal components obtained from PCA as the input for the linear regression algorithm49,50. As the problem is now reduced to fitting PSOAP + 1 coefficients (instead of FSOAP + 1), we conservatively only need ~P × 10 = 100 data-points for training; 100 molecular statics computations involving the substitution of a single solute atom at a grain boundary site in a polycrystal give insight on the entire segregation spectrum. As for the selection of the 100 training data points, though random selection can be used, this could be a biased approach due to the low number of points accessing only a prevalent subset of the GB feature space in a given polycrystalline structure. Instead, we propose using k-means clustering51,52 to partition the reduced feature space into k=100 clusters that minimize within-cluster variances. We then use the cluster centroids to identify optimal training data-points (i.e., shortest Euclidean distance to the centroids), as shown in Fig. 3, for which GB segregation is computed, and use it to train the accelerated model. Such an approach is computationally inexpensive and ensures the full coverage of the feature space in our training dataset.

Similar to the high-fidelity model, the accelerated one can be fully automated and requires minimal input from the user. To compare the performance of both approaches, an accelerated model for Al(Mg), trained with only 100 GB sites, results in an MAE of 4.2 kJ/mol for predictions of the full ~105 GB sites, compared to an MAE of 2.5 kJ/mol from the high-fidelity model trained with 50% of GB sites. This reduction (two orders of magnitude) in the required training data points, with minimal loss of accuracy, is significant, and showcases the power of the accelerated model to quickly, and accurately, predict the segregation spectra in binary alloys. It also signifies that the full GB space could possibly be reduced to a small number of key GB environments—also known as GB “building blocks”31—that decipher the features of the full space. We expect this to be a significant direction of future work in the context of grain boundary segregation.

Using the accelerated approach, we build ML models to predict solute segregation spectra in polycrystals for every aluminum, and magnesium, and transition metal-based binary alloys (Supplementary Figs. 2–20) that have interatomic potentials in the NIST Interatomic Potentials Repository—a total of 259 alloys (see Supplementary Fig. 1). This segregation database not only allows us to screen the alloy space for segregation “hot-spots” or regions of interest, but also to compare the variation of the spectrum with different interatomic potentials (for alloys where more than one potential exists). To illustrate the utility of the database, we plot in Fig. 4 all solute segregation spectra in a nickel-based alloy; Ni(Ag)37 is predicted to be highly segregating, and the opposite for Ni(Al)53.

Fig. 4: GB segregation spectra in Ni-based alloys.
figure 4

Using the accelerated ML model outlined in Fig. 3, we compute GB segregation energy spectra for 18 solutes37,40,42,53,58,75,76,77,78,79,80,81 in a 20x20x20 nm3 Ni polycrystal of grain size 10 nm; the anti-segregation region (\({\Delta}E_i^{{\rm{seg}}} > 0\)) is shaded. The spectra are fitted to the skew-normal function Eq. (3) (solid line), and the value of the characteristic energy μ (kJ/mol), width σ (kJ/mol), and shape α of the function are listed. Also, the spectra are compared to the “average” segregation energy (dashed vertical line) used to charecterize these alloys in the database of Murdoch and Schuh55.

Full size image

Discussion

There are three key findings to the spectral segregation database (Fig. 4 and Supplementary Figs. 2–20). The first is that all segregation spectra in all binary alloys surveyed, as hypothesized earlier in our study of the Al(Mg) system20, are captured well by a skew-normal function (the fitted probability density function has an R2 > 0.95 in all but one alloy with an R2=0.80; see Supplementary Figs. 2–20). This function involves three parameters—the characteristic energy μ, width σ and shape α of the distribution:

$$F_i^{{\rm{gb}}}({\Delta}E_i^{{\rm{seg}}}) = \frac{1}{{\sqrt {2\pi } \,\sigma }}\exp \left[ { - \frac{{\left( {{\Delta}E_i^{{\rm{seg}}} - \mu } \right)^2}}{{2\sigma ^2}}} \right]{\mathrm{erfc}}\left[ { - \frac{{\alpha \left( {{\Delta}E_i^{{\rm{seg}}} - \mu } \right)}}{{\sqrt 2 \,\sigma }}} \right]$$
(1)

These parameters are provided in the corresponding figure for each alloy considered. The second key finding is that using a McLean “average”54 segregation energy to characterize a binary alloy, which is the segregation literature norm55,56, misses key information about the accessible segregation states at the GB network. For example, the Ni(Ag) system that has a reported “average” segregation energy of −50 kJ/mol55, has approximately 15% of its GB network with segregation energies more than twice that, below −100 kJ/mol, as shown in the first panel of Fig. 4. GB segregation occurs first in the lowest energy states, and before a grain boundary in Ni(Ag) would experience the McLean average segregation energy, it would lie at an extremely high composition of approximately 50 atom% Ag. The knowledge of the full spectrum is thus essential to enable the design of microstructures57 that maximize the desired tail of the segregation spectrum (i.e. either promote or inhibit segregation). The third key finding is that, for alloys with more than one available interatomic potential, the computed segregation spectra can be sensitive to the choice of the potential. For example, potentials for the Al(Ni) system produce completely different segregation spectra, as shown in Supplementary Fig. 3, which range from having almost all GB sites being unfavorable to segregation58 to the complete opposite59; such variation can result in an order of magnitude difference in predictions for GB solute concentration even at low total solute concentrations in the system (see Supplementary Fig. 23). Further work is needed in the future to quantify the accuracy of such potentials for GB segregation studies60, and as always with atomistic models, it is important to remember that the present framework will only return physically reasonable results if the potential is specifically suitable for the problem at hand. For now, we report all of them, and leave the selection step to the judgment of the user.

Though the analysis in Figs. 1–3 shows that the ML models faithfully reproduce most of the details of the GB segregation spectrum, this is not the most critical test for their practical viability; these models are only useful to the extent that they correctly capture GB segregation in some realistic situation. Thus, the most important metric is the prediction for the equilibrium GB segregation state (i.e. extent of segregation). For a spectrum of segregation energies at the GB network, the equilibrium solute distribution among the different sites follows Fermi-Dirac statistics18,19,20. In a closed system with finite grain sizes, the total solute concentration Xtot is fixed and shared by the bulk (intra-grain) and GB solute concentrations, Xc and \(\bar X^{gb}\), respectively, according to the GB site fraction fgb:

$$X^{{\rm{tot}}} = \left( {1 - f^{{\rm{gb}}}} \right)X^c + f^{{\rm{gb}}}\bar X^{{\rm{gb}}}$$
(2)

The equilibrium Xc and \(\bar X^{{\rm{gb}}}\) are a function of the temperature T, the distribution of GB segregation energies \(F_i^{{\rm{gb}}}\)(\({\mathrm{{\Delta}}}E_i^{{\rm{seg}}}\)), and are obtained by numerically solving for Xc that satisfies the expanded form of Eq. (2)20:

$$X^{{\rm{tot}}} = \left( {1 - f^{{\rm{gb}}}} \right)X^c + f^{{\rm{gb}}}\mathop {\sum}\limits_i {F_i^{{\rm{gb}}}\left[ {1 + \frac{{1 - X^{\rm{c}}}}{{X^{\rm{c}}}}\exp \left( {\frac{{{\mathrm{{\Delta}}}E_i^{{\rm{seg}}}}}{{kT}}} \right)} \right]^{ - 1}}$$
(3)

In Fig. 5, we compare the equilibrium GB segregation state obtained using the true computed spectrum versus the ML predicted ones with both high-fidelity and accelerated models, for all seven alloys from Figs. 1 and 2, in a polycrystal of average grain size 15 nm (\(f^{{\rm{gb}}} \approx 10\%\)) at T = 600 K. The predictions of the ML models closely match those of the true spectrum, indicating that the ML models capture the necessary information to correctly predict the equilibrium segregation state. Also, as briefly discussed earlier, though the value of the MAE differs from one system to another, a higher MAE does not necessarily translate to a worse result, when one normalizes to the scale of the segregation energy distribution, e.g. the Zr(Ni) system in Fig. 5. Finally, we note that the difference (deviation) in predictions of the equilibrium segregation state could be even less of an issue if the skew-normal approximation Eq. (1) is used, instead of the full discrete spectra, to quantify GB segregation using the continuous form of the segregation isotherm Eq. (3)20:

$$X^{{\rm{tot}}} = \left( {1 - f^{{\rm{gb}}}} \right)X^c + f^{{\rm{gb}}}{\int}_{ - \infty }^\infty {F_i^{{\rm{gb}}}\left[ {1 + \frac{{1 - X^{\rm{c}}}}{{X^c}}\exp \left( {\frac{{{\Delta}E_i^{{\rm{seg}}}}}{{kT}}} \right)} \right]^{ - 1}d{\Delta}E_i^{{\rm{seg}}}}$$
(4)
Fig. 5: Predictions of the equilibrium segregation state.
figure 5

Predictions of equilibrium \(\bar X^{{\rm{gb}}}\) using the true, and predicted (from both high-fidelity and accelerated ML models) segregation spectra for seven polycrystalline alloys with an average grain size of 15 nm at T= 600 K.

Full size image

as the three fitted parameters (μ, σ, and α) of the skew-normal function for the true and ML predicted spectra should closely match, even for systems with high MAE, as the residuals are well-behaved and normally distributed (with a zero mean, as shown in Figs. 1 and 3).

To motivate further analysis of the spectral segregation database, and visually summarize the segregation tendency across the alloy space, we plot a two-dimensional Pettifor61 map in Fig. 6 (for most alloys in the database) using the 25th percentile value (energy) for the segregation spectra (i.e. 25% of GB sites have lower segregation energies). As the lower tail is the most enthalpically favorable, it will disproportionately influence the segregation tendency in any given alloy, especially at low or dilute solute concentration. The choice of the Pettifor chemical scale (which preserves the Mendeleev-type features of the elements61) is based on its success in pattern clustering (separation) for miscibility62, ordering tendency63,64, and crystal structures of intermetallics61 in binary alloys. Though Fig. 6 shows some clustering, it is not enough to draw concrete conclusions on the segregation tendency across the alloy space; the same finding applies to another two routinely used parameters to characterize the chemical and physical nature of the elements—electronegativity65, and metallic radius66—(see Supplementary Figs. 24–27). It is evident that more effort is needed to formulate (or extract from ML) simple phenomenological parameters (preferably derived from atomic features e.g. Miedema-style parameters67) that better explain these trends. We hope that this preliminary exploration of the data will promote further work on this front.

Fig. 6: Visual summary of the predicted segregation tendency across the alloy space.
figure 6

The value of the first quartile of the segregation spectra (i.e. 25% of GB sites have lower segregation energies) as predicted using the accelerated model for 210 alloys (we removed columns with empty entries for compact viewing); the elements are arranged by their order on the Pettifor chemical scale61. For alloys with multiple interatomic potentials (see Supplementary Fig. 1), we report the least segregating spectra as a conservative choice (see Supplementary Fig. 25 for an alternative version of the figure with the most segregating spectra).

Full size image

In summary, our proposed ML framework, inspired by methods developed for fitting ML-based interatomic potentials, aims to fit a “pseudo interatomic potential” for GB segregation energies in polycrystalline alloys. The framework is designed to require minimal input from the user, and as such, is automatable. As the ML literature is constantly evolving, we look forward to new developments and tools that can further improve the framework. We offered two model options. The first is a high-fidelity model that uses a large SOAP vector (>103 features), a conservative radial cutoff (6 Å), and linear regression. The second, is an accelerated model that uses PCA to transform the original features into a few (10) principal components (which are then used as input features to linear regression); this reduces the dimension of the learning problem to just 100 key GB environments, which are selected by k-means clustering to ensure coverage of the GB space. The accelerated model is used to build an extensive database for segregation spectra in 259 binary alloys, which is included the Supplementary Information. We look forward to applications of this database in alloy design, and hope it motivates more widespread use of spectral approaches to GB segregation in polycrystalline materials.

Methods

GB segregation enthalpies

The atomistic simulation package LAMMPS68,69 is used for all molecular statics and dynamics simulations; OVITO70 is used for visualization and identification of atomic structures.

To generate the base-metal polycrystal, we fill a 20x20x20 nm3 volume with 16 randomly oriented grains using Voronoi tesselations with Atomsk71. The polycrystal is thermally annealed at 0.3–0.5 of the melting temperature under a Nose-Hoover thermostat/barostat for 250 ps using a time step of 1 fs, which relaxes the grain structure and boundaries without permitting exaggerated grain growth; this is followed by slow cooling to 0 K at a cooling rate of 3 K/ps, and a final conjugate gradient energy minimization.

To compute the spectrum of segregation enthalpies in a binary alloy, we follow the procedure in ref. 20. We first relax the base-metal polycrystal using the interatomic potential of that alloy, by applying an external pressure of zero in a conjugate gradient minimization, followed by a second conjugate gradient minimization (with no applied pressure). This is necessary to scale the cell, and correct for minor differences in the equilibrium lattice parameter of the base-metal across the different interatomic potentials (for example, the Ni polycrystal is thermally annealed using an interatomic potential42 that is fitted to Ni lattice parameter of 3.518 Å, but the Ni(Al)53 is fitted to 3.520 Å). Then, every GB site in the annealed polycrystal is identified using adaptive-common neighbor analysis method72; all atoms that have a different atomic structure than the base metal are assumed to be GB atoms. For every GB site (i), its \({\mathrm{{\Delta}}}E_i^{{\rm{seg}}}\) is calculated as the relaxed energy difference between the solute atom occupying the GB site, versus a bulk (intra-grain) site: \({\mathrm{{\Delta}}}E_i^{{\rm{seg}}} = E_{{\rm{gb}},i}^{{\rm{solute}}} - E_{\rm{c}}^{{\rm{solute}}}\); the relaxation of each state is achieved using a conjugate gradient minimization, and the reference bulk site for \(E_c^{solute}\) is chosen as the center of a 6 nm sphere of the pure solvent (in the polycrystal), to avoid any long-range interactions with GB atoms. All calculations are at 0 K, isolating the enthalpic portion of the segregation energy for each site.

Machine Learning

For feature extraction, the LAE of every GB site within a cutoff radius of 6 Å is described using the SOAP method, as implemented in the QUIP/GAP software package28,30. SOAP fits a set of radial basis functions and spherical harmonics to Gaussian particle density functions placed over all neighboring atoms in the LAE. The maximum number of radial basis functions (nmax), degree of spherical harmonics (lmax), and the width of Gaussian functions (σat) control the size and resolution of the SOAP feature vector. We use nmax = lmax = 12 and σat = 1 Å for all alloys, which gives a SOAP vector with 1015 features. As for the other components of the ML framework: linear regression, principal component analysis, and k-means clustering are used as implemented in the Scikit-learn73 python package.