Accelerated search for materials with targeted properties by adaptive design (original) (raw)
Related papers
Materials & Design, 2018
In this study, a framework for the multi-objective materials discovery based on Bayesian approaches is developed. The capabilities of the framework are demonstrated on an example case related to the discovery of precipitation strengthened NiTi shape memory alloys with up to three desired properties. In the presented case the framework is used to carry out an efficient search of the shape memory alloys with desired properties while minimizing the required number of computational experiments. The developed scheme features a Bayesian optimal experimental design process that operates in a closed loop. A Gaussian process regression model is utilized in the framework to emulate the response and uncertainty of the physical/computational data while the sequential exploration of the materials design space is carried out by using an optimal policy based on the expected hyper-volume improvement acquisition function. This scalar metric provides a measure of the utility of querying the materials design space at different locations, irrespective of the number of objectives in the performed task. The framework is deployed for the determination of the composition and microstructure of precipitation-strengthened NiTi shape memory alloys with desired properties, while the materials response as a function of microstructure is determined through a thermodynamically-consistent micromechanical model.
Advanced Functional Materials, 2010
Improving the functional stability of shape memory alloys (SMAs), which undergo a reversible martensitic transformation, is critical for their applications and remains a central research theme driving advances in shape memory technology. By using a thin-film composition-spread technique and high-throughput characterization methods, the lattice parameters of quaternary Ti–Ni–Cu–Pd SMAs and the thermal hysteresis are tailored. Novel alloys with near-zero thermal hysteresis, as predicted by the geometric non-linear theory of martensite, are identified. The thin-film results are successfully transferred to bulk materials and near-zero thermal hysteresis is observed for the phase transformation in bulk alloys using the temperature-dependent alternating current potential drop method. A universal behavior of hysteresis versus the middle eigenvalue of the transformation stretch matrix is observed for different alloy systems. Furthermore, significantly improved functional stability, investigated by thermal cycling using differential scanning calorimetry, is found for the quaternary bulk alloy Ti50.2Ni34.4Cu12.3Pd3.1.
Optimal Experimental Design for Materials Discovery
In this paper, we propose a general experimental design framework for optimally guiding new experiments or simulations in search of new materials with desired properties. The method uses the knowledge of previously completed experiments or simulations to recommend the next that can effectively reduce the pertinent model uncertainty affecting the materials properties. To illustrate the utility of the proposed framework, we focus on a computational problem that utilizes time-dependent Ginzburg-Landau (TDGL) theory for shape memory alloys to calculate the stress-strain profiles for a particular dopant at a given concentration. Our objective is to design materials with the lowest energy dis-sipation at a specific temperature. The aim of experimental design is to suggest the best dopant and its concentration for the next TDGL simulation. Our experimental design utilizes the mean objective cost of uncertainty (MOCU), which is an objective-based uncertainty quantification scheme that measures uncertainty based upon the increased operational cost it induces. In the context of sequential experiments, our method adaptively chooses the next experiment based on the improved model obtained from the previous experiment. We analyze the performance of the proposed method and compare it with other experimental design approaches, namely random selection and pure exploitation.
Data-Driven Strategies for Accelerated Materials Design
Accounts of Chemical Research, 2021
Metrics & More Article Recommendations CONSPECTUS: The ongoing revolution of the natural sciences by the advent of machine learning and artificial intelligence sparked significant interest in the material science community in recent years. The intrinsically high dimensionality of the space of realizable materials makes traditional approaches ineffective for large-scale explorations. Modern data science and machine learning tools developed for increasingly complicated problems are an attractive alternative. An imminent climate catastrophe calls for a clean energy transformation by overhauling current technologies within only several years of possible action available. Tackling this crisis requires the development of new materials at an unprecedented pace and scale. For example, organic photovoltaics have the potential to replace existing silicon-based materials to a large extent and open up new fields of application. In recent years, organic light-emitting diodes have emerged as state-of-the-art technology for digital screens and portable devices and are enabling new applications with flexible displays. Reticular frameworks allow the atom-precise synthesis of nanomaterials and promise to revolutionize the field by the potential to realize multifunctional nanoparticles with applications from gas storage, gas separation, and electrochemical energy storage to nanomedicine. In the recent decade, significant advances in all these fields have been facilitated by the comprehensive application of simulation and machine learning for property prediction, property optimization, and chemical space exploration enabled by considerable advances in computing power and algorithmic efficiency. In this Account, we review the most recent contributions of our group in this thriving field of machine learning for material science. We start with a summary of the most important material classes our group has been involved in, focusing on small molecules as organic electronic materials and crystalline materials. Specifically, we highlight the data-driven approaches we employed to speed up discovery and derive material design strategies. Subsequently, our focus lies on the data-driven methodologies our group has developed and employed, elaborating on high-throughput virtual screening, inverse molecular design, Bayesian optimization, and supervised learning. We discuss the general ideas, their working principles, and their use cases with examples of successful implementations in data-driven material discovery and design efforts. Furthermore, we elaborate on potential pitfalls and remaining challenges of these methods. Finally, we provide a brief outlook for the field as we foresee increasing adaptation and implementation of large scale data-driven approaches in material discovery and design campaigns.
Combined Electronic Structure and Evolutionary Search Approach to Materials Design
Physical Review Letters, 2002
We show that density functional theory calculations have reached an accuracy and speed making it possible to use them in conjunction with an evolutionary algorithm to search for materials with specific properties. The approach is illustrated by finding the most stable four component alloys out of the 192 016 possible fcc and bcc alloys that can be constructed out of 32 different metals. A number of well known and new "super alloys" are identified in this way.
Shape Memory Alloys: A Summary of Recent Achievements
Materials Science Forum, 2008
The Ni-Mn-Ga shape memory alloy displays the largest shape change of all known magnetic Heusler alloys with a strain of the order of 10% in an external magnetic field of less than one Tesla. In addition, the alloys exhibit a sequence of intermediate martensites with the modulated structures usually appearing at c/a < 1 while the low-temperature nonmodulated tetragonal structures have c/a > 1. Typically, in the Ni-based alloys, the martensitic transformation is accompanied by a systematic change of the electronic structure in the vicinity of the Fermi energy, where a peak in the electronic density of states from the non-bonding Ni states is shifted from the occupied region to the unoccupied energy range, which is associated with a reconstruction of the Fermi surface, and, in most cases, by pronounced phonon anomalies. The latter appear in high-temperature cubic austenite, premartensite but also in the modulated phases. In addition, the modulated phases have highly mobile twin boundaries which can be rearranged by an external magnetic field due to the high magnetic anisotropy, which builds up in the martensitic phases and which is the origin of the magnetic shape memory effect. This overall scenario is confirmed by first-principles calculations.
Physical Review Letters, 2014
NiTi is the most used shape-memory alloy, nonetheless, a lack of understanding remains regarding the associated structures and transitions, including their barriers. Using a generalized solid-state nudge elastic band (GSSNEB) method implemented via density-functional theory, we detail the structural transformations in NiTi relevant to shape memory: those between body-centered orthorhombic (BCO) groundstate and a newly identified stable austenite ("glassy" B2-like) structure, including energy barriers (hysteresis) and intermediate structures (observed as a kinetically limited R-phase), and between martensite variants (BCO orientations). All results are in good agreement with available experiment. We contrast the austenite results to those from the often-assumed, but unstable B2. These high-and low-temperature structures and structural transformations provide much needed atomic-scale detail for transitions responsible for NiTi shape-memory effects.
2018
NiTi is the most used shape-memory alloy, nonetheless, a lack of understanding remains regarding the associated structures and transitions, including their barriers. Using a generalized solid-state nudge elastic band (GSSNEB) method implemented via density-functional theory, we detail the structural transformations in NiTi relevant to shape memory: those between body-centered orthorhombic (BCO) groundstate and a newly identified stable austenite (“glassy” B2-like) structure, including energy barriers (hysteresis) and intermediate structures (observed as a kinetically limited R-phase), and between martensite variants (BCO orientations). All results are in good agreement with available experiment. We contrast the austenite results to those from the often-assumed, but unstable B2. These highand low-temperature structures and structural transformations provide much needed atomic-scale detail for transitions responsible for NiTi shape-memory effects.
Neural-Network-Biased Genetic Algorithms for Materials Design: Evolutionary Algorithms that Learn
ACS combinatorial science, 2016
Machine-learning has the potential to dramatically accelerate high-throughput approaches to materials design, as demonstrated by successes in biomolecular design and hard materials design. However, in the search for new soft materials exhibiting properties and performance beyond those previously achieved, machine learning approaches are frequently limited by two shortcomings. First, because they are intrinsically interpolative, they are better suited to the optimization of properties within the known range of accessible behavior than to the discovery of new materials with extremal behavior. Second, they require large pre-existing datasets, which are frequently unavailable and prohibitively expensive to produce. Here we describe a new strategy, the neural-network-biased genetic algorithm (NBGA), for combining genetic algorithms, machine learning, and high-throughput computation or experiment to discover materials with extremal properties in the absence of pre-existing data. Within th...