An official website of the United States government
The melting temperature is important for materials design because of its relationship with thermal stability, synthesis, and processing conditions. Current empirical and computational melting point estimation techniques are limited in scope, computational feasibility, or interpretability. We report the development of a machine learning methodology for predicting melting temperatures of binary ionic solid materials. We evaluated different machine-learning models trained on a dataset of the melting points of 476 non-metallic crystalline binary compounds using materials embeddings constructed from elemental properties and density-functional theory calculations as model inputs. A direct supervised-learning approach yields a mean absolute error of around 180 K but suffers from low interpretability. We find that the fidelity of predictions can further be improved by introducing an additional unsupervised-learning step that first classifies the materials before the melting-point regression. Not only does this two-step model exhibit improved accuracy, but the approach also provides a level of interpretability with insights into feature importance and different types of melting that depend on the specific atomic bonding inside a material. Motivated by this finding, we used a symbolic learning approach to find interpretable physical models for the melting temperature, which recovered the best-performing features from both prior models and provided additional interpretability.
more »
« less
Comparing the accuracy of melting temperature prediction methods for high entropy alloys
Refractory complex concentrated alloys (RCCAs) are a relatively new class of materials that can exhibit excellent mechanical properties at high temperatures, and determining their melting temperature (Tm) is critical to assess their range of operation. Unfortunately, the experimental determination of this property is challenging and computational tools to predict the Tm of RCCAs from first-principles calculations are highly desirable. We quantify the uncertainties associated with such predictions for two methods that can be used with density functional theory-based molecular dynamics and apply them to predict the melting temperature of equiatomic NbMoTaW. We find that a combination of free energy calculations of individual phases with a dynamical coexistence method can provide accurate results with the minimum possible computational cost. We predict the melting temperature for the RCCA NbMoTaW to be between 3000 and 3100 K.
more »
« less
- Award ID(s):
- 1922316
- PAR ID:
- 10470685
- Publisher / Repository:
- AIP Publishing
- Date Published:
- Journal Name:
- Journal of Applied Physics
- Volume:
- 132
- Issue:
- 20
- ISSN:
- 0021-8979
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Inferring phase transitions and critical exponents from limited observations with thermodynamic mapsPhase transitions are ubiquitous across life, yet hard to quantify and describe accurately. In this work, we develop an approach for characterizing generic attributes of phase transitions from very limited observations made deep within different phases’ domains of stability. Our approach is called thermodynamic maps (TM), which combines statistical mechanics and molecular simulations with score-based generative models. TM enable learning the temperature dependence of arbitrary thermodynamic observables across a wide range of temperatures. We show its usefulness by calculating phase transition attributes such as melting temperature, temperature-dependent heat capacities, and critical exponents. For instance, we demonstrate the ability of TM to infer the ferromagnetic phase transition of the Ising model, including temperature-dependent heat capacity and critical exponents, despite never having seen samples from the transition region. In addition, we efficiently characterize the temperature-dependent conformational ensemble and compute melting curves of the two RNA systems: a GCAA tetraloop and the HIV-TAR RNA, which are notoriously hard to sample due to glassy-like energy landscapes.more » « less
-
Group contribution (GC) models are powerful, simple, and popular methods for property prediction. However, the most accessible and computationally efficient GC methods, like the Joback and Reid (JR) GC models, often exhibit severe systematic bias. Furthermore, most GC methods do not have uncertainty estimates associated with their predictions. The present work develops a hybrid method for property prediction that integrates GC models with Gaussian process (GP) regression. Predictions from the JR GC method, along with the molecular weight, are used as input features to the GP models, which learn and correct the systematic biases in the GC predictions, resulting in highly accurate property predictions with reliable uncertainty estimates. The method was applied to six properties: normal boiling temperature (Tb), enthalpy of vaporization at Tb (ΔHvap), normal melting temperature (Tm), critical pressure (Pc), critical molar volume (Vc), and critical temperature (Tc). The CRC Handbook of Chemistry and Physics was used as the primary source of experimental data. The final collected experimental data ranged from 485 molecules for ΔHvap to 5640 for Tm. The proposed GCGP method significantly improved property prediction accuracy compared to the GC-only method. The coefficient of determination (R2) values of the testing set predictions are ≥ 0.85 for five out of six and ≥ 0.90 for four out of six properties modeled, and compare favorably with other methods in the literature. Tm was used to demonstrate one way the GCGP method can be tuned for even better predictive accuracy. The GCGP method provides reliable uncertainty estimates and computational efficiency for making new predictions. The GCGP method proved robust to variations in GP model architecture and kernel choice.more » « less
-
Group contribution (GC) models are powerful, simple, and popular methods for property prediction. However, the most accessible and computationally efficient GC methods, like the Joback and Reid (JR) GC models, often exhibit severe systematic bias. Furthermore, most GC methods do not have uncertainty estimates associated with their predictions. The present work develops a hybrid method for property prediction that integrates GC models with Gaussian process (GP) regression. Predictions from the JR GC method, along with the molecular weight, are used as input features to the GP models, which learn and correct the systematic biases in the GC predictions, resulting in highly accurate property predictions with reliable uncertainty estimates. The method was applied to six properties: normal boiling temperature (Tb), enthalpy of vaporization at Tb (ΔHvap), normal melting temperature (Tm), critical pressure (Pc), critical molar volume (Vc), and critical temperature (Tc). The CRC Handbook of Chemistry and Physics was used as the primary source of experimental data. The final collected experimental data ranged from 485 molecules for ΔHvap to 5640 for Tm. The proposed GCGP method significantly improved property prediction accuracy compared to the GC-only method. The coefficient of determination (R2) values of the testing set predictions are ≥0.85 for five out of six and ≥0.90 for four out of six properties modeled, and compare favorably with other methods in the literature. Tm was used to demonstrate one way the GCGP method can be tuned for even better predictive accuracy. The GCGP method provides reliable uncertainty estimates and computational efficiency for making new predictions. The GCGP method proved robust to variations in GP model architecture and kernel choice.more » « less
-
Abstract Number: 381 Working Group: Instrumentation and Methods Abstract The phase state of atmospheric aerosol particles – solid, semi-solid, or liquid – influences their ability to take up water and participate in heterogeneous chemical reactions. Changes in phase state have been predicted by glass transition temperature (Tg) and viscosity; however, direct measurements of these properties is challenging for sub-micron particles. Historically, bulk measurements have been used, but this does not account for particle-to-particle variation or the impacts of particle size. Melting temperature (Tm) is the most significant predictor of Tg, and the two properties can be related through the Boyer-Beaman rule. Herein, we apply a recently developed method utilizing a nano-thermal analysis (nanoTA) module coupled to an atomic force microscope (AFM), to determine the Tm of individual secondary organic aerosol (SOA) particles generated from the reactive uptake of isoprene-derived epoxydiols (IEPOX) onto acidic ammonium sulfate aerosol particles. NanoTA works by using a specialized AFM probe which can be heated while in contact with a particle of interest. As the temperature increases, the probe deflection will first increase due to thermal expansion of the particle followed by a decrease at its Tm. The direct measurements are compared with model predictions based on molecular composition from hydrophilic interaction liquid chromatography coupled to electrospray ionization high-resolution quadrupole time-of-flight mass spectrometry (HILIC/ESI-HR-QTOF-MS) analysis. We compared the Tm of the SOA particles formed from IEPOX uptake onto acidic ammonium sulfate particles created at 30, 65, and 80% relative humidity (RH), and found that increasing RH from 30 to 80% led to an overall decrease in average Tm, indicating less viscous particles at higher RH conditions. Our measurements with this technique will allow for more accurate representations of the phase state of aerosols in the atmosphere.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1063/5.0101548
