Search

The Joint Automated Repository for Various Integrated Simulations (JARVIS) is an integrated infrastructure to accelerate materials discovery and design using density functional theory (DFT), classical force-fields (FF), and machine learning (ML) techniques. JARVIS is motivated by the Materials Genome Initiative (MGI) principles of developing open-access databases and tools to reduce the cost and development time of materials discovery, optimization, and deployment. The major features of JARVIS are: JARVIS-DFT, JARVIS-FF, JARVIS-ML, and JARVIS-tools. To date, JARVIS consists of ≈40,000 materials and ≈1 million calculated properties in JARVIS-DFT, ≈500 materials and ≈110 force-fields in JARVIS-FF, and ≈25 ML models for material-property predictions in JARVIS-ML, all of which are continuously expanding. JARVIS-tools provides scripts and workflows for running and analyzing various simulations. We compare our computational data to experiments or high-fidelity computational methods wherever applicable to evaluate error/uncertainty in predictions. In addition to the existing workflows, the infrastructure can support a wide variety of other technologically important applications as part of the data-driven materials design paradigm. The JARVIS datasets and tools are publicly available at the website: https://jarvis.nist.gov.

The current study focuses on the modeling of two-phase γ-γ′ nickel-based superalloys, utilizing multi-scale approaches to simulate and predict the creep behaviors through crystal plasticity finite element (CPFE) platforms. The multi-scale framework links two distinct levels of the spatial spectrum, namely, sub-grain and homogenized scales, capturing the complexity of the system responses as a function of a tractable set of geometric and physical parameters. The model considers two dominant features of γ′ morphology and composition. The γ′ morphology is simulated using three parameters describing the average size, volume fraction, and shape. The sub-grain level is expressed by a size-dependent, dislocation density-based constitutive model in the CPFE framework with the explicit depiction of γ-γ′ morphology as the building block of the homogenized scale. The homogenized scale is developed as an activation energy-based crystal plasticity model reflecting intrinsic composition and morphology effects. The model incorporates the functional configuration of the constitutive parameters characterized over the sub-grain γ-γ′ microstructural morphology. The developed homogenized model significantly expedites the computational processes due to the nature of the parameterized representation of the dominant factors while retains reliable accuracy. Anti-Phase Boundary (APB) shearing and, glide-climb dislocation mechanisms are incorporated in the constitutive model which will become active based on the energies associated with the dislocations. The homogenized constitutive model addresses the thermo-mechanical behavior of nickel-based superalloys for an extensive temperature domain and encompasses orientation dependence as well as the loading condition of tension-compression asymmetry aspects. The model is validated for diverse compositions, temperatures, and orientations based on previously reported data of single crystalline nickel-based superalloy.

The density and configurational changes of crystal dislocations during plastic deformation influence the mechanical properties of materials. These influences have become clearest in nanoscale experiments, in terms of strength, hardness and work hardening size effects in small volumes. The mechanical characterization of a model crystal may be cast as an inverse problem of deducing the defect population characteristics (density, correlations) in small volumes from the mechanical behavior. In this work, we demonstrate how a deep residual network can be used to deduce the dislocation characteristics of a sample of interest using only its surface strain profiles at small deformations, and then statistically predict the mechanical response of size-affected samples at larger deformations. As a testbed of our approach, we utilize high-throughput discrete dislocation simulations for systems of widths that range from nano- to micro- meters. We show that the proposed deep learning model significantly outperforms a traditional machine learning model, as well as accurately produces statistical predictions of the size effects in samples of various widths. By visualizing the filters in convolutional layers and saliency maps, we find that the proposed model is able to learn the significant features of sample strain profiles.

Deep learning has shown its superiority to traditional machine learning methods in various fields, and in general, its success depends on the availability of large amounts of reliable data. However, in some scientific fields such as materials science, such big data is often expensive or even impossible to collect. Thus given relatively small datasets, most of data-driven methods are based on traditional machine learning methods, and it is challenging to apply deep learning for many tasks in these fields. In order to take the advantage of deep learning even for small datasets, a domain knowledge integration approach is proposed in this work. The efficacy of the proposed approach is tested on two materials science datasets with different types of inputs and outputs, for which domain knowledge-aware convolutional neural networks (CNNs) are developed and evaluated against traditional machine learning methods and standard CNN-based approaches. Experiment results demonstrate that integrating domain knowledge into deep learning can not only improve the model’s performance for small datasets, but also make the prediction results more explainable based on domain knowledge.

Systems far from equilibrium respond to probes in a history-dependent manner. The prediction of the system response depends on either knowing the details of that history or being able to characterize all the current system properties. In crystal plasticity, various processing routes contribute to a history dependence that may manifest itself through complex microstructural deformation features with large strain gradients. However, the complete spatial strain correlations may provide further predictive information. In this paper, we demonstrate an explicit example where spatial strain correlations can be used in a statistical manner to infer and classify prior deformation history at various strain levels. The statistical inference is provided by machine-learning (ML) techniques. As source data, we consider uniaxially compressed crystalline thin films generated by two dimensional discrete dislocation plasticity simulations, after prior compression at various levels. Crystalline thin films at the nanoscale demonstrate yield-strength size effects with very noisy mechanical responses that produce a serious challenge to learning techniques. We discuss the influence of size effects and structural uncertainty to the ability of our approach to distinguish different plasticity regimes.

This paper develops non-Schmid crystal plasticity constitutive models at two length scales, and bridges them in a multi-scale framework. The constitutive models address thermo-mechanical behavior of Nickel-based superalloys for a large temperature range, viz. 300 K–1223 K, and include orientation dependencies and tension-compression asymmetry. The orientation dependencies result in tension-compression asymmetry for almost all orientations on the standard unit triangle. However simulations show different trends for the stronger direction (tension or compression) in terms of yield stress and hardening. The multi-scale framework includes two sub-grain and homogenized grain scales. For the sub-grain scale, a size-dependent, dislocation density-based FEM model of the representative volume element (RVE) with explicit depiction of the γ-γ′ morphology is developed as a building block for homogenization. For the next scale, an activation energy based crystal plasticity (AE-CP) model is developed for single crystal Ni-based superalloys. The homogenized AE-CP model develops functional forms of constitutive parameters in terms of characteristics of the sub-grain γ-γ′ microstructural morphology including γ′ shape, volume fraction and γ channel-width in the sub-grain microstructure. This homogenized model can significantly expedite crystal plasticity FE simulations due to the parametrized representation, while retaining accuracy.

We perform high-throughput density functional theory (DFT) calculations for optoelectronic properties (electronic bandgap and frequency dependent dielectric function) using the OptB88vdW functional (OPT) and the Tran-Blaha modified Becke Johnson potential (MBJ). This data is distributed publicly through JARVIS-DFT database. We used this data to evaluate the differences between these two formalisms and quantify their accuracy, comparing to experimental data whenever applicable. At present, we have 17,805 OPT and 7,358 MBJ bandgaps and dielectric functions. MBJ is found to predict better bandgaps and dielectric functions than OPT, so it can be used to improve the well-known bandgap problem of DFT in a relatively inexpensive way. The peak positions in dielectric functions obtained with OPT and MBJ are in comparable agreement with experiments. The data is available on our websites http://www.ctcms.nist.gov/~knc6/JVASP.html and https://jarvis.nist.gov.

Recent advances in experimental techniques (micro-CT scans, automated serial sectioning, electron back-scatter diffraction, and synchrotron radiation X-rays) have made it possible to characterize the full, three-dimensional structure of real materials. Such new experimental techniques have created a need for software tools that can model the response of these materials under various kinds of loads. OOF (Object Oriented Finite Elements) is a desktop software application for studying the relationship between the microstructure of a material and its overall mechanical, dielectric, or thermal properties using finite element models based on real or simulated micrographs. OOF provides methods for segmenting images, creating meshes of complex geometries, solving PDE's using finite element models, and visualizing 3D results. We discuss the challenges involved in implementing OOF in 3D and create a finite element mesh of trabecular bone as an illustrative example.

One way of computing the macroscopic behavior of a material sample with complex microstructure is to construct a finite element model based on a micrograph of a representative slice of the material. The quality of the results produced with such a model obviously depends on the quality of the constructed mesh. In this article, we describe a set of routines that modify and improve the quality of a 2D mesh. Most of the routines are guided by an effective element “energy” functional, which takes into account the shape quality of the elements and the homogeneity of the elements as determined from an underlying segmented image. The interfaces and boundaries in the image arise naturally from the segmentation process. From these routines, we construct a close-to-automatic mesh generator that requires only a few inputs, such as the linear sizes of the largest and smallest features in the micrograph.

This paper develops a framework to obtain the flow stress of nickel-based superalloys as a function of γ-γ′ morphology. The yield strength is a major factor in the design of these alloys. This work provides additional effects of γ′ morphology in the design scope that has been adopted for the model developed by authors. In general, the two-phase γ-γ′ morphology in nickel-based superalloys can be divided into three variables including γ′ shape, γ′ volume fraction and γ′ size in the sub-grain microstructure. In order tfo obtain the flow stress, non-Schmid crystal plasticity constitutive models at two length scales are employed and bridged through a homogenized multi-scale framework. The multi-scale framework includes two sub-grain and homogenized grain scales. For the sub-grain scale, a size-dependent, dislocation-density-based finite element model (FEM) of the representative volume element (RVE) with explicit depiction of the γ-γ′ morphology is developed as a building block for the homogenization. For the next scale, an activation-energy-based crystal plasticity model is developed for the homogenized single crystal of Ni-based superalloys. The constitutive models address the thermo-mechanical behavior of nickel-based superalloys for a large temperature range and include orientation dependencies and tension-compression asymmetry. This homogenized model is used to obtain the morphology dependence on the flow stress in nickel-based superalloys and can significantly expedite crystal plasticity FE simulations in polycrystalline microstructures, as well as higher scale FE models in order to cast and design superalloys.

Recent advances in experimental techniques (micro CT scans, automated serial sectioning, electron back-scatter diffraction, synchrotron radiation x-rays) have made it possible to characterize the full, three dimensional structure of real materials. Such new experimental techniques have created a need for software tools that can model the response of these materials under various kinds of loads. OOF (Object Oriented Finite Elements) is a desktop software application for studying the relationship between the microstructure of a material and its overall mechanical, electromagnetic, or thermal properties using finite element models based on real or simulated micrographs. OOF provides methods for segmenting images, creating meshes of complex geometries, solving PDEs using finite element models, and visualizing 3D results. We discuss the challenges involved in implementing OOF in 3D and use finite element simulations of trabecular bone as an illustrative example.