National Academies Press: OpenBook
« Previous: 2 Designing Materials to Revolutionize and Engineer Our Future: The First 10 Years
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

3

The Current State of Materials Research

This chapter surveys some key aspects of the current state of materials research that are of importance to the National Science Foundation’s (NSF’s) Designing Materials to Revolutionize and Engineer Our Future (DMREF) program’s research landscape. Since this is a forward-looking survey, the discussion of computational methods places particular emphasis on the opportunities and challenges that are emerging through the interaction of materials science with data science. Other areas of focus include new experimental and characterization efforts, and the potential influence of DMREF and the Materials Genome Initiative (MGI) on manufacturing technology. Since the chapter’s goal is to identify bottlenecks as well as opportunities, it focuses on challenges associated with maximizing the impact of DMREF-funded research on the nation’s materials science needs. Additional opportunities for continuing to build on DMREF’s interdisciplinary nature and strengthening broader computational capabilities are discussed in Chapter 6 (see Findings 6.9, 6.19, and 6.20; Key Finding 6.6; Recommendation 6.9; and Key Recommendations 6.1 and 6.6).

MATERIALS RESEARCH IS RAPIDLY EVOLVING

The MGI and DMREF in particular are reshaping materials education and practice in the United States and throughout the world. Materials development has historically been the result of many and long iterations among observations, hypotheses, and testing. Due to the complexity of engineering materials, the relationship between outcomes and design parameters, such as composition and

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

synthesis, has historically been built mostly through empiricism and heuristics. The past two decades have witnessed an increasing trend toward quantitatively predictive models (e.g., by the realization of practical and accurate first-principles density functional theory [DFT] and its extensions to strongly correlated materials) and atomistic-level understanding of materials (e.g., through the development of high-resolution and in-situ, multi-probe characterization methods). In addition, increasing volume and accuracy of thermodynamic data, some of which are also computed through methods such as kinetic Monte Carlo, have made the CALPHAD (CALculation of PHAse Diagrams) approach a critical element in material design in both academia and industry. Phase-field models are used routinely to predict evolving microstructures and crystallographic texture, and mechanical behavior simulations are increasingly taking the underlying dislocation motion and microstructure into account to improve their prediction accuracy. Furthermore, high-throughput computing efforts, such as the Materials Project1 and Open Quantum Materials Database,2 have democratized ab-initio data, enabling tens of thousands of scientists to benefit from more predictive and more quantitative materials information. See Figure 3-1 for the time scale versus length scale representation of various simulation methods.

The next two decades are likely to see even more radical transformations in materials research, for which the community should prepare. The use of mathematical and computational methods, such as artificial intelligence (AI), machine learning (ML), and deep learning, to capture information and turn it into interpretation and decision-making tools foretells a future where scientists’ reasoning is enhanced by such virtual tools. Natural language processing3 can extract information (and potentially reasoning) from the historical body of scientific literature. Data that would be stored in data repositories could capture information in a field more broadly than humans can, and would be used to generate or test hypotheses and to train interpretative algorithms, or used in their most simple form without any duplication. Materials data will directly feed digital engineering, manufacturing, and optimization tools, which will allow extensive analysis of alternatives that reach back to the material’s local structure.4 As such, the importance of data, data-driven

___________________

1 The website for the Materials Project is https://materialsproject.org, accessed September 28, 2022.

2 The Open Quantum Materials Database can be found at https://oqmd.org, accessed September 28, 2022.

3 See, for example, IBM, “What Is Natural Language Processing (NLP)?” https://www.ibm.com/cloud/learn/natural-language-processing, accessed September 28, 2022.

4 National Research Council, 2014, Big Data in Materials Research and Development: Summary of a Workshop, Washington, DC: The National Academies Press, https://doi.org/10.17226/18760; National Academies of Sciences, Engineering, and Medicine, 2018, Data Science: Opportunities to Transform Chemical Sciences and Engineering: Proceedings of a Workshop—in Brief, Washington, DC: The National Academies Press, https://doi.org/10.17226/25191.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Image
FIGURE 3-1 Time scale versus length scale representation of various simulation methods.
SOURCE: Reprinted from A.I. Vakis, V.A. Yastrebov, J. Scheibert, et al., 2018, “Modeling and Simulation in Tribology Across Scales: An Overview,” Tribology International 125:169–199, https://doi.org/10.1016/j.triboint.2018.02.005. Copyright (2018), with permission from Elsevier.

pattern recognition, models, and algorithm quality (i.e., validation and uncertainty quantification) will increase and serve as the core to establish trust with designers and consumers of the MGI approach and its acceleration of emerging materials and manufacturing technologies.

Advanced autonomous and high-throughput materials platforms will be developed to evaluate material properties rapidly. Advanced detectors and highly automated microscopies and spectroscopies will characterize their structure and composition from the macroscale to the atomic scale.5 These autonomous research approaches will enable the establishment of larger-scale, curated archival data sets that are currently missing.

Lower-cost robotics could lead to integrated, self-driving synthesis and characterization laboratories, generating the diverse types of materials data at the scale needed to leverage AI approaches and to more efficiently use the historic body of information to interpret data and offer low-level decision making. Such empowered materials research would lead to faster and more effective theoretical-computational-experimental iterations toward optimized materials. This will ultimately

___________________

5 For example, deep learning methods are useful for examining microscopies and spectroscopies data as the images have more nonlinear properties.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

enable materials to be tailored and manufactured “on-demand” to meet future societal needs.

And Yet, Many Familiar Questions Need Better Answers

The interaction of theory and simulation with experiments lies at the core of the MGI. Most materials science problems have a multiscale character that makes theory and simulation extremely challenging.6

Physics-based modeling has long been at the very foundation of computational and theoretical materials science. Such modeling can be done at various scales: quantum mechanical computations start from an approximated form of the Schrödinger equation; phase transitions are often studied using phase-field type continuum models; the annealing of a polycrystal may be studied using stochastic Potts-type models; and the processing of materials is often studied using continuum models involving heat and mass transfer or elasticity. Such modeling has spurred more quantitative approaches toward materials design and development on both the engineering and the fundamental sides of materials research.

While faster and lower-cost computing and improved algorithms have helped advance physics-based modeling, fundamental scientific advances have improved the understanding of how computable properties could be translated to more desirable engineering behavior. Knowing “what to compute” to make relevant impact on engineering is as essential as knowing “how to compute”—this type of translation has been aided by the interdisciplinary character of DMREF. The emerging importance of data-based methods does not mean that physics-based modeling will become less important. Rather, the challenge will be to integrate physics-based with data-driven approaches in a way that retains the mathematical versatility of data-driven methods, while benefiting from knowledge that is embedded in fundamental physical principles.

COMPUTATIONAL DEVELOPMENTS

Algorithms and Software

Advanced computational tools for materials modeling and simulation continue to play enabling roles in the MGI and the DMREF program. They allow computational scientists and engineers to support experimental synthesis and characterization efforts by providing fundamental insight into materials properties; they are

___________________

6 See, for example, National Academies of Sciences, Engineering, and Medicine, 2019, Frontiers of Materials Research: A Decadal Survey, Washington, DC: The National Academies Press, https://doi.org/10.17226/25244.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

used to identify in silico new materials and new synthetic pathways; and they are the primary tools for creating and maintaining the data ecosystem that underpins data-intensive AI research on materials. As summarized in the National Research Council report Integrated Computational Materials Engineering,7 these computational tools must be integrated in order to fully elucidate the linkage among material chemistry, processing, structure, and properties due to the multiscale nature of materials and the diversity of chemical and physical processes involved in producing materials and in the resulting properties. The National Research Council report8 lists 11 categories of computational tools, 8 of which are material simulation tools, along with their inputs and outputs. Specifically, these categories are electronic structure calculations, atomistic simulations, dislocation dynamics, thermodynamic methods, microstructural evolution simulations, micromechanical and mesoscale property models (e.g., finite element analysis), processing models, and other continuum models. Many of these tools are linked through inputs and outputs; for example, DFT can provide a number of computed properties that could be used as input to computational thermodynamics, mechanical behavior simulations, or optoelectronic properties needed for optical/electronic property predictions. Thermodynamic properties calculated by CALPHAD can be used as input to microstructure evolution simulation and, in turn, can inform property prediction models that consider microstructural effects. To provide the context behind the important role of development and availability of computational tools, further details are provided below.

“Electronic structure calculations” broadly refers to a set of theoretical and computational techniques developed during the past four decades that are capable of predicting certain properties of materials starting from their elemental composition.9 DFT is based on the solution of an approximate version of the quantum-mechanical Schrödinger equation for the atoms of a molecule, solid, or liquid. Its main strength is that it does not rely on empirical parameters; hence, it provides valuable information that is complementary to experimental characterization. In particular, DFT has been a disruptive force in materials science; for example, the scientific publications at the origin of modern DFT feature among the most cited papers of all time, alongside groundbreaking discoveries in life sciences, such as protein assays, deoxyribonucleic acid sequencing, and the polymerase chain reaction.10 Today DFT remains the most popular computational tool for atomistic

___________________

7 National Research Council, 2008, Integrated Computational Materials Engineering: A Transformational Discipline for Improved Competitiveness and National Security, Washington, DC: The National Academies Press, https://doi.org/10.17226/12199.

8 National Research Council, 2008.

9 F. Giustino, 2014, Materials Modelling Using Density Functional Theory: Properties and Predictions, Oxford: Oxford University Press.

10 R. Van Noorden, B. Maher, and R. Nuzzo, 2014, “The Top 100 Papers,” Nature 514:550–553, https://doi.org/10.1038/514550a.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

materials simulations, and on average one scientific manuscript based on DFT is published every 30 minutes around the world.

The computational infrastructure for DFT simulations has made great strides during the past two decades, and considerable community efforts have been devoted to software validation and verification.11 As a result, DFT software is becoming increasingly accessible to non-specialists, and it is not uncommon to find experimental groups who have integrated DFT modeling and simulation capabilities in their research programs. Training in DFT theory and software has also become popular; for example, PARADIM (Platform for the Accelerated Realization, Analysis, and Discovery of Interface Materials), one of four NSF Materials Innovation Platforms (MIPs), offers biannual summer schools designed to broaden adoption of DFT among experimental groups.12

With the exponential increase in popularity of DFT modeling and simulation, important criticalities begin to emerge in materials research. DFT is by design a theory of the total energy of atomistic systems like molecules and crystals, and performs well with predictions of phase stability and compositional phase diagrams. However, DFT has limitations because many important properties of materials, such as electrical, optical, magnetic, thermoelectric, and superconducting properties, are not described correctly by DFT. To predict such properties with reliable accuracy, more sophisticated computational tools are warranted. Efforts in this direction are being pursued by both the Department of Energy (DOE) and NSF, by sponsoring the development of cutting-edge, high-performance computing (HPC) software for advanced materials simulations. Key examples are the Computational Materials Centers sponsored by the DOE Basic Energy Sciences (BES) program, and the NSF Cyberinfrastructure for Sustained Scientific Innovation program. While being mission-critical to the MGI and DMREF, these initiatives are comparatively smaller in resources and scope than similar initiatives in Europe,13 such as the European Union MAX Center of Excellence on Materials at the Exascale,14 which supports large and popular community codes, such as SIESTA and Quantum ESPRESSO.15 To deliver the next generation of integrated experimental/computational materials research within the MGI and DMREF, it will be important to reestablish a competitive software cyberinfrastructure for materials simulations beyond standard DFT, and to invest in the training of the next generation of computational materials

___________________

11 K. Lejaeghere, G. Bihlmayer, T. Björkman, et al., 2016, “Reproducibility in Density Functional Theory Calculations of Solids,” Science 351(6280), https://doi.org/10.1126/science.aad3000.

12 The website for PARADIM is https://www.paradim.org, accessed January 30, 2022.

13 For additional information on activities outside the United States, see Chapter 5.

14 The website for the European Union MAX Center of Excellence on Materials at the Exascale is http://www.max-centre.eu, accessed January 30, 2022.

15 The website for Quantum ESPRESSO is https://www.quantum-espresso.org, accessed January 30, 2022.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

scientists and engineers. Focused efforts in this area are especially critical as materials databases and AI research can only be as good as the data they are built upon; therefore, there is an urgent need for improving the accuracy of DFT calculations16 as well as expanding the array of materials properties, including properties at finite temperature, that can be predicted using supercomputers. Cooperative interactions between computational and experimental scientists and engineers in the research and educational training efforts will be critical to advances in both computation and experiment, in terms of expertise, capabilities, data, and materials.

The preceding paragraphs emphasize DFT and quantum-scale models. There are, however, many problems where mesoscopic or macroscopic heterogeneity is important, and in such settings the useful models are quite different. The field of materials science and engineering has relied on the aforementioned CALPHAD approach. This method takes the thermodynamic free energies of various phases as input and predicts phase behavior (e.g., stable phases, phase fractions, and compositions)17 in multicomponent systems as a function of temperature and composition. This is accomplished by a constrained minimization of the free energy of the system. A phase diagram, which indicates what phases are expected in thermodynamic equilibrium for a given condition and material composition, is a traditional tool in materials science and engineering, but for complex alloys with more than a few components, computation is required to obtain the predictions of the phase behavior. Thanks to both experimental measurements (e.g., thermal analysis and phase analysis) and computational efforts (e.g., DFT and statistical mechanical methods), thermodynamic databases are growing and include pure, binary, and ternary systems in all types of materials (e.g., metals, semiconductors, oxides), which can be used to make predictions for more complex systems by combining the data and extrapolating to systems with more components. CALPHAD software tools are available as commercial software (e.g., Thermo-Calc, Pandat) as well as open-source software (e.g., OpenCalphad). Since this approach relies on the constrained minimization of the free energy based on the input function, its prediction accuracy depends on the accuracy of the input data. Due to the assessed

___________________

16 Where DFT (e.g., VASP [Vienna Ab initio Simulation Package]) is excellent at the quantum scale, molecular dynamics (e.g., LAMMPS) excels at the computation of materials at the nano and small micron scales at varying temperature and pressure conditions. See Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS), “ LAMMPS Molecular Dynamics Simulator,” https://www.lammps.org and Universität Wien, “VASP Homepage,” https://www.vasp.at, accessed January 30, 2022.

17 A typical phase diagram is generally generated at 1 atm and provides information of the equilibrium phases in the alloys. However, for different atmospheric conditions, it could change (e.g., in a vacuum or specific atmosphere, such as an oxygen-rich atmosphere, which can influence the materials’ response).

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

databases that are available with commercial software, industry frequently utilizes the commercial software18 in alloy design and development.19

While the CALPHAD approach predicts what phases are expected to appear and their quantity, it is not able to predict the microstructure or how it evolves. With simplifying assumptions such as spherical or planar geometry, a sharp interface description can be readily applied to predict how the interfaces between phases evolve during phase transformations. When the geometry of the interfaces is complex, the phase-field approach can be applied to predict the microstructure evolution based on the calculated thermodynamic driving force and the resulting flux of the chemical species. The applicability of the phase-field model is broad, from metallic alloys to semiconductors to ceramics. As with CALPHAD, its prediction accuracy depends on the thermodynamic and kinetic input (i.e., the free energy density and mobilities), as well as any physics that influence the evolution of the microstructure (e.g., stress and strain due to lattice misfit).

Numerous other computational techniques are used in materials research to reflect the multiscale nature of materials and capture the different phenomena being simulated. This includes the kinetic Monte Carlo method, which can be used to calculate the free energy function or diffusivity based on DFT-calculated input and can also be used to simulate surface or compositional evolution at atomic scale; dislocation dynamics and crystal plasticity, which predict the plastic deformation of materials under stress at the mesoscopic scale and the continuum scale, respectively; and molecular dynamics, which evolves the atomic positions using inter-atomic potential and has a broad range of applications. Significant advancements have been achieved in all of these approaches through a combination of improved prediction accuracy and increased computational efficiency. As an example, for molecular dynamics, improvements in the potentials, such as the reactive force fields (ReaxFF) that can model chemical reactions by describing the processes of bond formation and bond breaking, have resulted in new applications, such as heterogeneous catalysis, atomic-layer deposition, and phase-change materials.20 However, it remains challenging to effectively link these models and combine data

___________________

18 Commercial databases often lag behind university materials development. Sometimes the software does not include the physics-based models but rather fits the available data. If adding research with CALPHAD, one can also achieve predictive capabilities. CALPHAD is a pervasive methodology that is unique in its ability to incorporate physics-based models and data from experimental and modeled sources.

19 C.E. Campbell, U.R. Kattner, and Z.-K. Liu, 2014, “The Development of Phase-Based Property Data Using the CALPHAD Method and Infrastructure Needs,” Integrating Materials and Manufacturing Innovation 3:158–180, https://doi.org/10.1186/2193-9772-3-12.

20 T.P. Senftle, S. Hong, M.M. Islam, et al., 2016, “The ReaxFF Reactive Force-Field: Development, Applications and Future Directions,” npj Computational Materials 2, https://www.nature.com/articles/npjcompumats201511.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

from these various methods. Part of this challenge is due to the disparate length and time scales, but other challenges exist in aligning underlying modeling assumptions and appropriately accounting for uncertainty. It should also be highlighted that these tools are enhanced by AI, which is described below.

The past decade has also witnessed the rise of AI, ML, and deep learning in materials science. These approaches are used to augment atomistic DFT simulations by extending the simulation length and time scales to find hidden relations between data and to accelerate high-throughput calculations of materials properties. One of the most successful applications of ML in materials simulations consists of using classical force fields trained on DFT quantum simulations.21 These force fields carry the advantage that they incorporate the fine details of quantum-mechanical interactions between atoms with the efficiency of classical simulations. As a result, they allow researchers to perform simulations using very large systems and relatively long length scales. For example, simulations of 100 million atoms with the accuracy of DFT were simply impossible until very recently. Using deep potential molecular dynamics, ML-powered simulations spanning 1 nanosecond per day on the Summit supercomputer have recently been demonstrated.22

It is difficult to predict how AI/ML will change computational materials research in the future, but from the recent explosion of interest in these techniques it seems likely that the integration of these computational tools will deliver another profound paradigm shift. Additional opportunities to enhance computational tools and methods, software, and hardware access are discussed in Chapter 6 (see Findings 6.4, 6.7, and 6.12; Key Finding 6.1; Recommendations 6.2 and 6.13; Key Recommendation 6.4).

Multiscale Challenges

Despite the enormous progress in data science, theory, automation, experimental techniques, and computational materials science during the past decades, there remain many outstanding challenges in the development of new heterogeneous materials that are now barely within quantitative and predictive description. Access to such materials would be enabled by non-equilibrium phenomena (e.g., in materials assembly and processing), dissipative phenomena and open systems, and excited-state energy surfaces in condensed phases. Realistic materials are heterogeneous, and the control of interfaces, defects, and building blocks is critical, but their

___________________

21 J. Behler and M. Parrinello, 2007, “Generalized Neural-Network Representation of High-Dimensional Potential-Energy Surfaces,” Physical Review Letters 98:146401, https://doi.org/10.1103/PhysRevLett.98.146401.

22 W. Jia, H. Wang, M. Chen, et al., 2020, “Pushing the Limit of Molecular Dynamics with Ab Initio Accuracy to 100 Million Atoms with Machine Learning,” SC ‘20: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (5):1–14.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

quantified description with computational and experimental methods has been challenging. The description of these properties further underlines the importance of making data robust and accessible. Moreover, strategies should be developed to implement robust validation and verification strategies for both experimental and computational data.

The challenge of multiscale physics is to connect macroscopic behavior to the atomic scale, governed by the laws of quantum mechanics. Addressing this challenge requires connecting phenomena with widely different intrinsic time scales, and connecting models that describe behavior on widely different length scales. While methods from statistical physics and theoretical chemistry provide a framework for bridging diverse time scales, the design of practical numerical algorithms capable of predictive modeling for real materials remains a research frontier. In connecting behavior on different length scales, a key difficulty is that the phenomena of interest are usually very nonlinear, for example, the effect of polycrystalline texture or the presence of inclusions on the ductility or fracture toughness of a material. In such problems, there are various continuum models, but there is little understanding of how to calibrate them to specific situations since large-scale simulations of the relevant atomic-scale models is infeasible.

An important development in recent years is the emergence of materials whose macroscopic properties have a quantum mechanical character. These “quantum materials” are materials with strong electronic correlations or different types of electronic order, such as superconducting or magnetic order. They also include materials such as topological insulators, Dirac materials such as graphene, Weyl metals, and others with properties governed by genuinely quantum behavior.23 The fundamental degrees of freedom—charge, spin, orbital motion, and lattice—become intertwined, resulting in emergent complex electronic states. Quantum materials also offer opportunities for physical realizations of qubits, the elements of quantum computers, pointing toward the possibility of true quantum simulations of molecules and materials.

High-Performance Computing

HPC is a key driver of progress in materials research and constitutes one of the pillars of the DMREF program alongside materials synthesis and characterization. DMREF teams critically rely on the availability of leadership-class HPC facilities within the United States to foster their materials design and discovery mission. For example, HPC underpins materials simulations from the atomic scale to the mesoscale, high-throughput computational screening of candidate materials prior

___________________

23 Nature Physics Editorial Team, 2016, “The Rise of Quantum Materials,” Nature Physics 12:105, https://doi.org/10.1038/nphys3668.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

to experimental synthesis, and the use of AI to reveal hidden structures in data and to accelerate simulations.

The United States remains a leading player in HPC, but in the past two decades Asia and Europe propelled themselves to the forefront of this field. For example, the United States claimed only three spots in the top-10 ranking of the world’s most powerful supercomputers at the 33rd annual Supercomputing Conference in 2021.24 In comparison, the United States had five supercomputers in the top-10 list in 2011, and eight machines made the cut in 2001. This shifting landscape is the result of increased commitment to HPC leadership in Asia and Europe and signals the global recognition of computational science and engineering as an indispensable engine of scientific progress.

The most powerful U.S. supercomputers are from DOE, namely the Summit supercomputer at Oak Ridge National Laboratory, Sierra at Lawrence Livermore National Laboratory, and Perlmutter at Lawrence Berkeley National Laboratory. Summit is capable of computations at a speed reaching 200 PFlop/s. One PFlop/s is 1 million billion floating-point operations (e.g., multiplications) every second. The United States is expected to be the first country to break through the exascale computing barrier in 2022 with the deployment of the Frontier supercomputer at Oak Ridge National Laboratory and the Aurora supercomputer at Argonne National Laboratory, both supported through the Exascale Computing Project.25 Exascale supercomputers (1 EFlop/s = 1,000 TFlop/s) will be capable of performing 1 billion billion floating-point operations per second and will enable a new generation of materials simulations with unprecedented detail and unrivaled accuracy.

At NSF, the Office of Advanced Cyberinfrastructure supports a distributed HPC infrastructure that complements DOE facilities. The only NSF leadership-class supercomputer is Frontera at the Texas Advanced Computing Center (TACC), which in 2019 claimed the fifth spot among the Top-500 Supercomputers with 40 PFlop/s of peak performance.26 A 10-fold increase in computing power over Frontera will be achieved with the Leadership-Class Computing Facility, which is currently in the conceptual design phase. In addition, NSF operates several medium-scale facilities with the aim to lower the barrier-to-entry in scientific computing, support workforce development and foster academia-industry partnership.27 Medium-scale HPC systems recently funded by NSF include Jetstream 2 at Indiana University, Delta at the University of Illinois at Urbana-Champaign, and Anvil at Purdue University, as well as systems specifically designed for AI and deep

___________________

24 The rankings can be found at https://www.top500.org, accessed January 30, 2022.

25 The website for the Exascale Computing Project is https://www.exascaleproject.org, accessed January 30, 2022.

26 The website for TACC is https://www.tacc.utexas.edu, accessed January 30, 2022.

27 See ACCESS, “Homepage,” https://access-ci.org, accessed October 17, 2022.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

learning, such as Voyager at the San Diego Supercomputing Center and Neocortex at Carnegie Mellon University.28

The research supported by the DMREF program is heavily reliant on these HPC resources. For example, the flagship computational database of the MGI, the Materials Project, is entirely based on atomic-scale quantum-mechanical simulations of materials carried out using the DOE-supported National Energy Research Scientific Computing Center. More broadly, atomic-scale and macromolecular simulations of materials represent a large fraction of the HPC resources employed every year in the United States. At TACC, atomic-scale materials simulations represented about 25 percent of the total workload in 2020,29 and the National Energy Research Scientific Computing Center at Lawrence Berkeley National Laboratory reported a similar fraction for DOE users.30 These figures highlight the strong interdependence between DMREF and HPC and underline the critical role of HPC capabilities in the future of the MGI.

As HPC-powered materials design grows in popularity and the DMREF program expands in scope, new criticalities emerge. For one, access to HPC resources is becoming highly competitive as the materials research community expands. This is seen in the plummeting success rates in recent competitions of the Innovative and Novel Computational Impact on Theory and Experiment program, which is the main pathway for accessing HPC resources at leadership-class DOE supercomputers; and in the low success rates in recent competitions of the Leadership Resource Allocation program, which is the gateway for access to the NSF Frontera supercomputer. These bottlenecks pose a challenge to the DMREF program: while the strength of the program is that DMREF supports graduate students and postdoctoral researchers, it does not provide them with the HPC resources needed to carry out their research. Securing HPC resources has traditionally been considered as a separate challenge that principal investigators (PIs) must address after receiving a DMREF award. For the United States to maintain its competitiveness and leadership in materials research, scientists and engineers will need wider and more streamlined access to leadership-class HPC resources. This could be achieved by continuing to innovate and expand the U.S. HPC infrastructure and by bundling DMREF awards together with guaranteed HPC access.

In addition, HPC offers an extraordinary opportunity for the United States to design products faster, minimize the time to create and test prototypes, streamline production processes, lower the cost of innovation, and develop high-value

___________________

28 See ACCESS, “About,” https://access-ci.org/about, accessed October 17, 2022.

29 The number refers to DFT calculations using VASP, Quantum ESPRESSO, and some quantum chemistry codes.

30 Dan Stanzione, Director of TACC, personal communication, June 20, 2022.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

innovations that would otherwise be impossible.31 The DOE High Performance Computing for Manufacturing program (HPC4Mfg) is an important example of how this can be done. It promotes collaboration between U.S. manufacturers and the DOE national laboratories. Companies submit two-page concept papers describing a project idea that the laboratories help expand to full proposals. The Advanced Manufacturing Office—within DOE’s Office of Energy Efficiency and Renewable Energy—leads the HPC4Mfg program. Manufacturers use advanced modeling, simulation, and analysis to achieve significant energy and cost savings, expand their markets, and grow the economy.32

An additional criticality is emerging from the incredibly rapid evolution of the hardware infrastructure, as many-core architectures and graphics processing units are displacing conventional central processing unit computing. Innovative HPC architectures are disruptive forces in computational science and engineering, but the shifting hardware landscape sometimes requires a drastic overhaul of scientific programming paradigms. Dedicated support for refactoring materials simulation software to effectively and timely leverage new architectures will be key to maintain the competitiveness of computational materials research and to fulfill the mission of the MGI and the DMREF initiatives.

Parallel to the need for wider access to HPC resources is the need for more comprehensive HPC training of graduate students and postdoctoral researchers supported by DMREF projects. Through its Office for Advanced Cyberinfrastructure, NSF offers extensive training on all aspects of HPC, ranging from software and application programming interface (API) development to workflow management, big data, and ML, as well as scientific visualization. This training is delivered across multiple complementary channels. Each HPC facility runs its own training events—for example, the TACC Learning Portal,33 which reports new courses both in-person and online almost on a weekly basis all year. In addition, the NSF Software Institutes have developed strong educational programs. For example, the Molecular Sciences Software Institute offers a variety of educational resources in quantum chemistry, computational materials science, and biomolecular simulation, targeting a broad spectrum of students at all levels as well as postdoctoral researchers and faculty.34 NSF also offers a centralized repository of its educational

___________________

31 DOE, 2021, High Performance Computing for Manufacturing: Using Supercomputers to Improve Energy Efficiency and Performance, Washington, DC: Office of Energy Efficiency & Renewable Energy, https://hpc4energyinnovation.llnl.gov/sites/hpc4energyinnovation/files/2022-03/HPC_Manufacturing_Brochure_MAY_13_2021.pdf.

32 See DOE, p. 5, accessed October 17, 2022.

33 The website for the TACC Learning Portal is https://learn.tacc.utexas.edu, accessed September 29, 2022.

34 The website for the Molecular Sciences Software Institute is education.molssi.org, accessed September 29, 2022.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

offerings in HPC via the XSEDE (Extreme Science and Engineering Discovery Environment) platform.35 This will be replaced by the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support platform starting September 1, 2022. Importantly, all of these online resources, webinars, and in-person courses are provided free or nearly free of charge to students, postdocs, and PIs.

Despite such a strong and diverse educational ecosystem, HPC literacy among DMREF teams remains low. To fill this knowledge gap, it would be useful to increase awareness of available educational resources among DMREF PIs and to place more emphasis on the training and workforce development aspects of the project annual reports.

Additional opportunities to enhance HPC capabilities are discussed in Chapter 6 (see Finding 6.2, Recommendations 6.3 and 6.22).

The Importance of Data in a Changing Materials Research Landscape

Today, the changing role of data is again innovating the way in which materials research is pursued. Historically, data were gathered on materials to verify or build scientific hypotheses, with experimental outcomes mostly living on through the theories established from them. With such a perspective, the role of data is largely complete once facts and insights are established from them. Negative data are almost never reported or stored.

New developments within materials science and at its edge are demanding a re-evaluation of the role of data aggregation in materials research. High-throughput first principles computing can determine a specific property for thousands of compounds in a matter of days but often lacks the structured experimental data sets to make statistically relevant statements about its accuracy. The data requirements for AI/ML are even more substantial. As AI/ML methods proceed with little scientific framework to rely on, their ability to recognize useful patterns and create a predictive model requires a volume of data that is rarely present in a single research study. While the value of ML is well established in commerce and social media, the lack of data and data infrastructure has kept the potential of AI/ML limited in materials research.

There is a need for data aggregation and preservation to be key foci of materials research.36 Perhaps in the future they will be considered as important as publishing,

___________________

35 The website for XSEDE is https://web.archive.org/web/20220820060934/https://www.xsede.org, accessed September 29, 2022.

36 The committee wants to acknowledge the disparate types of data that are necessary to understand materials (e.g., many levels beyond atomistic to the microscale). Two-, three-, and four-dimensional data are needed, and not all data are scalar in character, as assumed by most AI/ML algorithms.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

maybe even linked to the publishing process. Large, aggregated data sets can be leveraged to re-interpret old theories or create new ones. They can form the knowledge base from which automated self-driving laboratories initiate their search for novel materials or new insights. And ultimately, by accessing the historical data record of materials research, AI and ML methods might be able to assist humans in suggesting hypotheses for novel materials, thereby tremendously accelerating the pace of materials research and development (R&D). The National Institute of Standards and Technology (NIST) has created the Materials Data Repository37 as part of an effort in coordination with the MGI to establish data exchange protocols and mechanisms that will foster data sharing and reuse across the wider community of researchers, with the goal of enhancing the quality of materials data and models. Data present on the NIST system are varied and originate from the worldwide materials community; they may not be critically reviewed but are nevertheless made available. NIST’s materials data resources38 include the Materials Resource Registry, which bridges the gap among existing resources, software and repositories, and end users by allowing for the registration of materials resources. The Materials Resource Registry functions as a federated service, making the registered information from multiple institutions available for research to the materials community.39 The NIST Schema Repository and Registry is a service operated to improve discovery, access, and reuse of materials-related schemas, data models, ontologies, and more.40

Data Mining

Data mining is an interdisciplinary exploration that combines tools from statistics, ML, and database systems to uncover patterns and relationships within large data sets, as seen in Figure 3-2.41 It is extensively used in different sectors, such as finance, medicine, marketing, astronomy, and materials science. The increasing availability of experimental and computational materials databases underpins the ability to perform data mining. Examples of experimental databases available

___________________

37 The Materials Data Repository can be found at https://materialsdata.nist.gov, accessed September 29, 2022.

38 See NIST, “Materials Genome Initiative: Materials Data Resources,” https://www.nist.gov/mgi/materials-data-resources, accessed September 29, 2022.

39 See NIST, “Materials Resource Registry,” https://materials.registry.nist.gov, accessed September 29, 2022.

40 The NIST Schema Repository and Registry can be found at https://www.nist.gov/mgi/materials-data-resources, accessed October 17, 2022.

41 For a definition of data mining, see IBM, “What Is Data Mining?” https://www.ibm.com/cloud/learn/data-mining, accessed September 29, 2022.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Image
FIGURE 3-2 The input tools from statistics, machine learning, and database systems coming together in data mining to reveal patterns and relationships of materials properties.

include the Inorganic Crystal Database,42 Cambridge Structural Database,43 the Pauling File,44 CRYSTMET (a computer-readable database of critically evaluated crystallographic data for metals), and Pearson’s Crystal Data.45 Examples of computationally derived materials databases include Materials Project, AFLOWlib,46 NRELMatDB,47 Materials Platform, and Open Quantum Materials Database.48 Citrine Informatics built a commercial platform called Citrination, which offers an open database for both computational and experimental materials and chemicals data, data management, and AI-based tools for data mining.49 International efforts, discussed further in Chapter 5, include the European Union’s NOMAD, a

___________________

42 See the Inorganic Crystal Database at https://icsd.products.fiz-karlsruhe.de, accessed September 29, 2022.

43 The Cambridge Structural Database can be found at https://www.ccdc.cam.ac.uk/solutions/csdcore/components/csd, accessed September 29, 2022.

44 The website for the Pauling File is https://paulingfile.com, accessed September 29, 2022.

45 See Pearson’s Crystal Data database at http://www.crystalimpact.com/pcd, accessed September 29, 2022.

46 AFLOWlib can be found at https://aflowlib.org, accessed September 29, 2022.

47 NRELMatDB can be found at https://materials.nrel.gov, accessed September 29, 2022.

48 See The Open Quantum Materials Database (OQMD), “Homepage,” https://oqmd.org, accessed October 17, 2022, and A. Jain, G. Hautier, S.P. Ong, and K. Persson, 2016, “New Opportunities for Materials Informatics: Resources and Data Mining Techniques for Uncovering Hidden Relationships,” Journal of Materials Research 31(8):977–994.

49 See B. Meredig, 2017, “Industrial Materials Informatics: Analyzing Large-Scale Data to Solve Applied Problems in R&D, Manufacturing, and Supply Chain,” Current Opinion in Solid State and Materials Science 21(3):159–166.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

large data repository for computational materials science data from more than 100 million calculations, FAIR (findable, accessible, interoperable, and reusable) data representation and tools for searching and data mining, and China’s ALKEMIE,50 an open-source computational platform that includes data generation via high-throughput calculations, data management, and data mining.

The increasing availability of materials databases has led to a corresponding rise in data mining capabilities to develop descriptors and relate them to output properties. This has occurred extensively within and outside DMREF, both nationally and internationally. In some cases, these are general-purpose capabilities that can be applied to a wide range of problems, while in other cases, they are tailored to a specific material or application. Two examples of general-purpose data mining platforms are Matminer51 and Pymatgen.52 Pymatgen is the core analysis code for the Materials Project and provides capabilities to extract representation of molecules, structures and supports, electronic structure calculations, and post-calculation analyses. Matminer is an open-source Python-based platform that provides capabilities for retrieval of large data sets from several external databases, community-built feature extraction libraries, and visualization. These capabilities integrate with other ML capabilities that have been developed by the Python data science community.

Despite significant progress in the availability of materials databases and data mining capabilities, several challenges remain. Underpinning data mining is the need to increase data sharing and access using FAIR principles. With multiple databases with API access and data mining tools, it is difficult to search and integrate data across different sources. Integrating computational and experimental data is especially challenging because it is difficult to match conditions under which experiments were performed. Maintenance and updates for these large data platforms are expensive and require dedicated staff and funding.

Machine Learning and Artificial Intelligence

ML is becoming tightly integrated into materials research. ML is focused on the design and analysis of computer algorithms that improve their performance on a task with experience obtained from data.53 The integration of ML with materials research includes the use of existing ML methods, libraries, and code to analyze

___________________

50 G. Wang, L. Peng, K. Li, et al., 2021, “ALKEMIE: An Intelligent Computational Platform for Accelerating Materials Discovery and Design,” Computational Materials Science 186:110064.

51 L. Ward, A. Dunn, A. Faghaninia, et al., 2018, “Matminer: An Open Source Toolkit for Materials Data Mining,” Computational Materials Science 152:60–69.

52 S.P. Ong, W.D. Richards, A. Jain, et al., 2013, “Python Materials Genomics (pymatgen): A Robust, Open-Source Python Library for Materials Analysis,” Computational Materials Science 68:314–319.

53 T.M. Mitchell, 1997, Machine Learning, New York: McGraw-Hill.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

experimental data and material simulations54,55,56,57 and leverage prior knowledge from literature58 and experts.59 Additionally, material applications have led to the creation of a new area of physics-informed ML, which is focused on enhancing extrapolation capabilities of ML using knowledge of physical phenomena to enable good performance, even in domains where no data are available. These approaches use physics knowledge in the form of invariances such as translation, rotation, scale, and permutation often present in material systems;60,61 differential equations or other relations governing material properties;62,63 and physical constraints on material systems.64,65

___________________

54 A. Kusne, T. Mueller, and R. Ramprasad, 2016, “Machine Learning in Materials Science: Recent Progress and Emerging Applications,” Chapter 4 in Reviews in Computational Chemistry, Volume 29 (A.L. Parrill and K.B. Lipkowitz, eds.), https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=915933.

55 K.T. Butler, D.W. Davies, H. Cartwright, O. Isayev, and A. Walsh, 2018, “Machine Learning for Molecular and Materials Science,” Nature 559(7715):547–555.

56 J. Behler and M. Parrinello, 2007, “Generalized Neural-Network Representation of High-Dimensional Potential-Energy Surfaces,” Physical Review Letters 98(14):146401.

57 M. Rupp, A. Tkatchenko, K.-R. Müller, and O.A. von Lilienfeld, 2012, “Fast and Accurate Modeling of Molecular Atomization Energies with Machine Learning,” Physical Review Letters 108(5):058301.

58 E.A. Olivetti, J.M. Cole, E. Kim, et al., 2020, “Data-Driven Materials Research Enabled by Natural Language Processing and Information Extraction,” Applied Physics Reviews 7:041317, https://doi.org/10.1063/5.0021106.

59 R. LeBras, R. Bernstein, C.P. Gomes, B. Selman, and R.B. van Dover, 2012, “Crowdsourcing Backdoor Identification for Combinatorial Optimization,” pp. 2840–2847 in Proceedings of the 23rd International Joint Conference on Artificial Intelligence, International Joint Conference on Artificial Intelligence (AAAI), https://dl.acm.org/doi/10.5555/2540128.2540538.

60 S. Batzner, A. Musaelian, L. Sun, et al., 2022, “E(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentials,” Nature Communications 13:2453, https://doi.org/10.1038/s41467-022-29939-5.

61 N.C. Thomas, T. Smidt, S. Kearnes, et al., 2018, “Tensor Field Networks: Rotation- and Translation-Equivariant Neural Networks for 3D Point Clouds,” ArXiv, https://arxiv.org/abs/1802.08219.

62 M. Raissi, P. Perdikaris, and G.E. Karniadakis, 2019, “Physics-Informed Neural Networks: A Deep Learning Framework for Solving Forward and Inverse Problems Involving Nonlinear Partial Differential Equations,” Journal of Computational Physics 378:686–707, https://doi.org/10.1016/j.jcp.2018.10.045.

63 R.T.Q. Chen, Y. Rubanova, J. Bettencourt, and D.K. Duvenaud, 2018, “Neural Ordinary Differential Equations,” Advances in Neural Information Processing Systems 31.

64 J. Peng, Y. Yamamoto, J.A. Hawk, E. Lara-Curzio, and D. Shin, 2020, “Coupling Physics in Machine Learning to Predict Properties of High-Temperatures Alloys,” npj Computational Materials 6(1):141, https://doi.org/10.1038/s41524-020-00407-2.

65 S. Ermon, R. LeBras, S.K. Suram, et al., 2014, “Pattern Decomposition with Complex Combinatorial Constraints: Application to Materials Discovery,” Proceedings of the 29th AAAI International Conference on Artificial Intelligence (AAAI), https://www.cs.cornell.edu/~lebras/publications/Ermon-2015Pattern.pdf.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

NSF’s DMREF program has embraced ML; 33 percent of funded projects (and 45 percent of currently active awards) have mentioned an ML component.66 Most of these projects involved use of existing ML tools for materials science applications, which clearly points to absorption and adoption of this technology in mainstream materials research. At the same time, DMREF’s ability to enable the reverse impact of materials science on ML algorithm development has not yet been as significant. Here, joint interdisciplinary publications would be a concrete metric on which progress can be targeted in the next 5 years.

This opportunity not only paves the way for convergent research that brings together and amplifies advances in both domains but also embraces the unique set of challenges and opportunities of materials science to advance ML. While classical supervised (classification, regression) and unsupervised (clustering, distribution estimation, and dimensionality reduction) ML tools are increasingly being used in materials research, more complex methods, such as sequential decision making (Bayesian optimization, bandit optimization, active and reinforcement learning), graphical models, and causal inference, have not found a way into materials research to the same extent, despite their obvious potential to enable autonomous materials research, interpretable predictions, and discovery of unknown physics.67 Thus, the potential of ML for materials research is not being leveraged fully. At the same time, materials research has much to offer for advancing ML.68 Specifically, ML applicability is starting to reach limits owing to excessive data requirements,

___________________

66 In the NSF online award database there are 95 of 211 active and 169 of 507 total awards mentioning machine learning.

67 T. Lookman, P.V. Balachandran, D. Xue, and R. Yuan, 2019, “Active Learning in Materials Science with Emphasis on Adaptive Sampling Using Uncertainties for Targeted Design,” npj Computational Materials 5:21; A.G. Kusne, H. Yu, C. Wu, et al., 2020, “On-the-Fly Closed-Loop Materials Discovery via Bayesian Active Learning,” Nature Communications 11:5966; H. Doan and G. Agarwal, 2021, “Active Learning via Bayesian Optimization for Materials Discovery,” https://nanohub.org/resources/35144; H. Abroshan, H.S. Kwak, Y. An, et al., 2022, “Active Learning Accelerates Design and Optimization of Hole-Transporting Materials for Organic Electronics,” Frontiers in Chemistry 9:800371; H. Zhao, M.L. Comer, and M. De Graef, 2014, “A Unified Markov Random Field/Marked Point Process Image Model and Its Application to Computational Materials,” Pp. 6101–6105 in 2014 IEEE International Conference on Image Processing (ICIP), https://doi.org/10.1109/ICIP.2014.7026231;M. Ziatdinov, C.T. Nelson, X. Zhang, et al., 2020, “Causal Analysis of Competing Atomistic Mechanisms in Ferroelectric Materials from High-Resolution Scanning Transmission Electron Microscopy Data,” npj Computational Materials 6:127, https://doi.org/10.1038/s41524-020-00396-2; M.B. Muhlestein, C.F. Sieck, A. Alu, and M.R. Haberman, 2016, “Reciprocity, Passivity and Causality in Willis Materials,” Proceedings of Royal Society A, http://doi.org/10.1098/rspa.2016.0604; S.R. Xie, G.R. Stewart, J.J. Hamlin, P.J. Hirschfeld, and R.G. Hennig, 2019, “Functional Form of the Superconducting Critical Temperature from Machine Learning,” Physical Review B 100:174513, https://doi.org/10.1103/PhysRevB.100.174513.

68 The committee wants to acknowledge length scales and the multimodal character of materials data and how this could motivate development of new AI/ML approaches.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

and leveraging physical phenomena governing data appears key to ensuring continued success of ML. Such physical information is often not readily available in other domains (e.g., biology, neuroscience, climate, and Earth science) as the physical laws behind involved natural phenomena are not well understood. Materials science, on the other hand, involves engineered systems that often follow and can be regulated by established physical laws. Thus, tight-knit integration of materials and ML researchers to jointly explore the challenges and opportunities at this intersection holds a lot of promise, and DMREF can play a significant enabling role.

AI refers to the capacity of machines to imitate human intelligence. AI encompasses ML, as seen in Figure 3-3, but goes beyond the development and deployment of algorithms capable of learning (“on their own”) from data. Recognizing that the boundary between ML and AI is somewhat diffuse (particularly for non-experts in the AI/ML disciplines), AI in the context of materials science can be used to denote the deployment of ML tools in order to solve complex, inverse, and goal-oriented materials problems. In this context, goals could refer to the attainment of specific performance objectives, the gaining of better understanding of a given materials system, or the discovery of fundamental physical and/or chemical principles through the efficient exploration and exploitation of materials spaces.

Among the different types of materials inverse problems amenable to solution via AI, the accelerated discovery of materials is perhaps the one with the most immediate relevance to the MGI’s goal of significantly reducing the time and effort necessary to deploy new technology-enabling materials. The availability of efficient ML models capable of establishing connections between materials attributes and materials behavior is a necessary but insufficient condition to accelerate the materials discovery process. A complete solution to this problem requires the

Image
FIGURE 3-3 The hierarchy of artificial intelligence (AI), machine learning (ML), and deep learning (DL).
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

development of frameworks capable of solving complex, iterative loops, in which (computational and/or experimental) information is acquired with the goal to guide the exploration of the materials design space. Traditional approaches to exploring the materials space are based on the high-throughput and combinatorial exploration of materials systems via modern experimental synthesis platforms or high-performance research computing facilities. Such high-throughput approaches, however, are “open loop” and are incapable of prescribing optimal actions to achieve a goal given data and knowledge previously acquired.

In recent years, the materials science community has focused on moving toward “closed loop” frameworks for materials discovery. In these frameworks, initial data are acquired, ML models are built based on the acquired information, and the resulting ML models are used to decide where else in the materials space to explore to maximize the probability of identifying promising materials solutions. To date, the majority of frameworks seeking to implement these closed loops are based on Bayesian optimization. A characteristic of Bayesian approaches is that they precisely prescribe how prior knowledge can be updated upon acquisition of new information. In the context of materials discovery and design, Bayesian optimization can establish optimal closed-loop frameworks in which each iteration over the materials problem space is carried out optimally.69

Bayesian optimization and other active learning approaches have already demonstrated the ability to dramatically accelerate the process of materials discovery and provide new insights to materials properties. An even more consequential application of such approaches is the development of truly autonomous, self-driven closed-loop materials discovery platforms. The goal of such platforms is to integrate the synthesis, processing, characterization, “on the fly” analysis, and even simulation of materials in feedback loops capable of iteratively improving a set of target material properties with minimal human intervention. The future of accelerated materials discovery may indeed be intertwined with the development of such platforms, in which ML, AI, robotics, natural language processing, and efficient graphical user interfaces enable the autonomous exploration of the materials space.70

DMREF should consider how it can plan for and contribute to this new opportunity. This could be through supporting physical autonomous laboratories as part of larger centers or by supporting projects aimed at demonstrating specific elements of such automated/autonomous closed loops.

___________________

69 R. Arróyave, 2022, “Data Science, Machine Learning and Artificial Intelligence Applied to Metals and Alloys Research: Past, Present, and Future,” Pp. 609–621 in Encyclopedia of Materials: Metals and Alloys, F.G. Caballero, ed., vol. 4, Amsterdam, Netherlands: Elsevier, https://doi.org/10.1016/B978-0-12-819726-4.00078-8.

70 R. Arróyave, pp. 609–621.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

Opportunities relating to the use of ML and AI in DMREF-sponsored communities are further discussed in Chapter 6 (see Key Finding 6.6).

Natural Language Processing

The availability of data has always been crucial to progress in materials science. In recent years, with the development of new tools such as ML for the extraction of information, the availability of data is more important than ever.

The committee discusses elsewhere in this report the need for repositories to which materials scientists can contribute newly created data to facilitate their use by others. But in the past, materials science data have mainly been reported through publications; moreover, this mode of dissemination seems likely to remain dominant in the near future, making published literature an important vehicle by which data are distributed.

The extraction of data from articles has traditionally been done by individual scientists, one article at a time, by reading them. Natural language processing opens the door to an entirely different approach to the literature, involving the extraction of data from many articles at once—leading to insights that are not available from isolated articles and to databases that can be used by entire communities of scientists (see Figure 3-4).

This approach is being pursued by a number of groups in the materials science community.71,72 However, the development of methods for extracting data from the materials science literature is still in its infancy. This is not a routine application of natural language processing technology developed for other purposes. Indeed, published materials science data come in many forms—they appear in the body of an article, in its tables and figures, and often in images. Whatever the location, automated interpretation of published data requires a process for identifying their character (including the type of model in the case of numerical simulations, and the type of measurements in the case of laboratory experiments). The language used to provide such information is highly technical and in many cases domain-specific. Crucial information is often at the beginning of the article (or perhaps in its supplementary material) rather than contiguous with the data, and today’s natural language processing has difficulty connecting various procedures (sometimes discussed in a methods section) to specific outcomes reported in the paper.

___________________

71 E.A. Olivetti, J.M. Cole, E. Kim, et al., 2020, “Data-Driven Materials Research Enabled by Natural Language Processing and Information Extraction,” Applied Physics Reviews 7:041317, https://doi.org/10.1063/5.0021106.

72 O. Kononova, T. He, H. Huo, A. Trewartha, E.A. Olivetti, and G. Ceder, 2021, “Opportunities and Challenges of Text Mining in Materials Research,” iScience 24(3):102155, https://doi.org/10.1016/j.isci.2021.102155.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Image
FIGURE 3-4 Scientific materials publications acting as input to the natural language process using a knowledge base containing source content and data storage for the interaction history and the analytics.

Materials science is a very broad and heterogeneous field. Different subdisciplines focus on different classes of materials, producing and using entirely different types of information. Therefore, one should not envision a one-size-fits-all solution to the challenges associated with automated extraction of data from published literature. Rather, it will be necessary to focus on specific classes of materials that are broad enough to have a large literature but narrow enough to share many scientific characteristics.

The collection of data is not a substitute for understanding; rather, it is a crucial tool for the development of understanding. Materials science data can be used in many different ways—including traditional uses, such as finding discrepancies between theory and experiment or identifying trends, and newer uses, such as the training of ML models. For most purposes, it is useful to have as many data as possible.

It is important that published data do not go to waste. Further development of such techniques is important, since it will permit present and future generations to make better use of the knowledge that is embedded in the existing literature.

EXPERIMENTS AND SYNTHESIS

The MGI aims to accelerate the feedback loop among computation, data, and experiment. To ultimately realize the impact of materials to society, materials must be physically made and manipulated, which requires experimental studies that involve synthesis, processing, and characterization at various stages throughout the materials development continuum (see Figure 1-2). There have been significant advances in experimental techniques that enable more expansive searches of

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

compositional space, accelerated property measurement, higher fidelity characterization, and innovative synthesis. These advances map onto different classes of materials and their functionalities. This section offers some examples of materials classes—these examples were chosen to highlight developments and are by no means complete.

Additional discussion of strategies to enhance material discovery with experimental approaches can be found in Chapter 6 (see Key Recommendation 6.7).

Metallic and Intermetallic Materials

Combinatorial approaches that generate large material libraries enable broad exploration of composition space. For metallic and intermetallic materials, libraries have recently been synthesized by a number of different film deposition approaches that introduce compositional gradients across sample libraries that search for materials that could serve in a wide spectrum of new applications, such as thermoelectrics, catalysts, advanced coatings, and high temperature structural materials.73 Varied deposition approaches, including sputtering, electron beam directed vapor deposition, and ion plasma deposition, generate samples of varying thicknesses, ranging from approximately 50 nm to several hundred microns.74 In parallel, there have been significant advances in microscale characterization that

___________________

73 R.D. Snyder, E.L. Thomas, and A.A. Voevodin, 2015, “Material Optimization via Combinatorial Deposition and Analysis for Thermoelectric Thin Films,” Thin Solid Films 596:233–241; C.C.L. McCrory, S. Jung, I.M. Ferrer, S.M. Chatman, J.C. Peters, and T.F. Jaramillo, 2015, “Benchmarking Hydrogen Evolving Reaction and Oxygen Evolving Reaction Electrocatalysts for Solar Water Splitting Devices,” Journal of the American Chemical Society 137:4347–4357; R.A. Potyrailo, K. Rajan, K. Stoewe, I. Takeuchi, B. Chisholm, and H. Lam, 2011, “ChemInform Abstract: Combinatorial and High-Throughput Screening of Materials Libraries: Review of State of the Art,” ACS Combinatorial Science 13:579–633; C.A. Stewart, S.P. Murray, A. Suzuki, T.M. Pollock, and C.G. Levi, 2020, “Accelerated Discovery of Oxidation-Resistant Co-γ/γ′ Alloys with High L12 Solvus and Low Density,” Materials and Design 189:108445; J. Geng, I.C. Nlebedim, M.F. Besser, E. Simsek, and R.T. Ott, 2016, “Bulk Combinatorial Synthesis and High Throughput Characterization for Rapid Assessment of Magnetic Materials: Application of Laser Engineered Net Shaping (LENS™),” The Journal of the Minerals, Metals & Materials Society 68:1972–1977; R. Emery, O.R. Rios, M.J. Thompson, D. Weiss, and P.D. Rack, 2022, “Thin Film Combinatorial Sputtering of Al-Ce Alloys: Investigating the Phase Separation of As-Deposited Solid Solutions and Determining the Coefficient of Thermal Expansion,” Journal of Alloys and Compounds 913, https://doi.org/10.1016/j.jallcom.2022.165271.

74 Y. Li, K.E. Jensen, Y. Liu, et al., 2016, “Combinatorial Strategies for Synthesis and Characterization of Alloy Microstructures Over Large Compositional Ranges,” ACS Combinatorial Science 18:630–637; P. Nagy, N. Rohbeck, R.N. Widmer, et al., 2022, “Combinatorial Study of Phase Composition, Microstructure and Mechanical Behavior of Co-Cr-Fe-Ni Nanocrystalline Film Processed by Multiple-Beam-Sputtering Physical Vapor Deposition,” Materials 15:2319–2339; T.M. Pollock, D.M. Lipkin, and K.J. Hemker, 2012, “Multifunctional Coating Interlayers for Thermal Barrier Systems,” MRS Bulletin 37:923–931.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

enable measurement of properties on thin and thick film samples. This includes, for example, measurements of hardness by rapid nanoindentation;75 elastic constants with surface acoustic waves;76 tensile properties with in-situ microscopy stages, including microelectromechanical devices;77 thermal conductivity with ultrafast lasers;78 local slip resistance and dislocation glide via in-situ straining with transmission scanning electron microscopy;79 and oxidation behavior with photo-stimulated luminescence spectroscopy.80 While these combinatorial approaches combined with microscale characterization greatly expand knowledge of properties across composition space, they are inherently slow and labor intensive. Fabrication of targets with unique compositions for combinatorial experiments is expensive and subject to trial-and-error processing. Preparation of microscale samples, often using slow focused-ion-beam machining approaches, poses serious challenges to the time scale required for characterization. Innovative and autonomous workflows that speed up the combinatorial design and discovery framework pose a future challenge for DMREF.

Since many metallic and intermetallic materials serve in structural applications in the construction, aerospace, nuclear, and chemical industries, microscale characterization may not provide sufficiently rigorous assessment of critical properties, which are often strongly affected by materials features present from the upper limit of film thickness of several hundred micrometers up to millimeters or centimeters. Fatigue is a well-known example, wherein “rare” intrinsic or extrinsic material features dominate the fatigue life but are not encountered until cubic millimeter volumes of material are sampled with sub-micrometer resolution.81 In this case, exploration demands bulk-scale samples that will capture the most critical properties. Emerging additive manufacturing platforms potentially enable exploration of high

___________________

75 J.M. Wheeler, B. Gan, and R. Spolenak, 2022, “Combinatorial Investigation of the Ni–Ta System via Correlated High-Speed Nanoindentation and EDX Mapping,” Small Methods 6:2101084.

76 R. Patel, S.D. Sharples, M. Clark, M.G. Somekh, and W. Li, 2021, “Single Pixel Camera Methodologies for Spatially Resolved Acoustic Spectroscopy,” Applied Physics Letters 118:051102.

77 L.Y. Chen, S. Terrab, K.F. Murphy, J.P. Sullivan, X. Cheng, and D.S. Gianola, 2014, “Temperature Controlled Tensile Testing of Individual Nanowires,” Review of Scientific Instruments 85:013901.

78 S. Huxtable, D.G. Cahill, V. Fauconnier, J.O. White, and J.C. Zhao, 2004, “Thermal Conductivity Imaging at Micrometre-Scale Resolution for Combinatorial Studies of Materials,” Nature Materials 3:298–304.

79 J.C. Stinville, E.R. Yao, P.G. Callahan, et al., 2019, “Dislocation Dynamics in a Nickel-Based Superalloy via In-Situ Transmission Scanning Electron Microscopy,” Acta Materialia 168:152–166.

80 C.A. Stewart, A. Suzuki, T.M. Pollock, and C.G. Levi, 2018, “Rapid Assessment of Oxidation Behavior in Co-Based γ/γ′ Alloys,” Oxidation of Metals 90:485–498, https://doi.org/10.1007/s11085-018-9849-2.

81 J.C. Stinville, W.C. Lenthe, M.P. Echlin, P.G. Callahan, D. Texier, and T.M. Pollock, 2017, “Micro-structural Statistics for Fatigue Crack Initiation in Polycrystalline Nickel-Base Superalloys,” International Journal of Fracture 208:221–240.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

dimension composition space with synthesis of bulk samples. These bulk additive manufacturing approaches typically employ powders or wire as feedstock.82 Protocols for synthesis of the input feedstock and identification of processes and control parameters that enable such exploration remain to be developed. Some innovative approaches to automated, robotic tensile testing of arrays of three-dimensional (3D) printed test specimens have recently been demonstrated83 (see Figure 3-5). Further development of additive manufacturing protocols that couple with robotic-assisted testing could accelerate discovery, development, and certification of a wide suite of structural materials, representing a unique opportunity for DMREF.

Highly relevant to structural and functional metallic and intermetallic materials is the close coupling of properties with structure, and the processing paths that govern structure development. Fully predictive discovery and design systems need to capture 3D aspects of structure and its evolution with time (4D). This field of materials science has seen intensive progress over the past decade, with the development of automated serial sectioning tomography techniques for acquisition of multimodal (chemical, crystallographic, structural) materials data84 and instrumentation for synchrotron beamline experiments85 that further enables time-resolved observations.86 Collectively, these instrumentation platforms can generate many parallel 2D, 3D, and 4D terabyte-scale data streams in different data formats and with different distortions and noise associated with the disparate classes of detectors. Managing the sheer volume of data and developing methodologies for merging disparate data streams constrains discovery via deep analysis of these large data sets. New high-speed detectors and accompanying tools for planning of optimum data collection strategies, processing data on the fly, and reducing raw

___________________

82 A. Dass and A. Moridi, 2019, “State of the Art in Directed Energy Deposition: From Additive Manufacturing to Materials Design,” Coatings 9:418; H. Knoll, S. Ocylok, A. Weisheit, H. Springer, E. Jagle, and D. Raabe, 2017, “Combinatorial Alloy Design by Laser Additive Manufacturing,” Steel Research International 88:1600416.

83 B.C. Salzbrenner, J.M. Rodelas, B.H. Jared, et al., 2017, “High-Throughput Stochastic Tensile Performance of Additively Manufactured Stainless Steel,” Journal of Materials Processing Technology 241:1–12; N.M. Heckman, T.A. Ivanoff, A.M. Roach, et al., 2020, “Automated High-Throughput Tensile Testing Reveals Stochastic Process Parameter Sensitivity,” Materials Science and Engineering A 772:138632.

84 M.P. Echlin, T.L. Burnett, A.T. Polonsky, T.M. Pollock, and P.J. Withers, 2020, “Serial Sectioning in the SEM for Three Dimensional Materials Science,” Current Opinion on Solid State and Materials Sciences 24(2):100817.

85 M.P. Miller, D.C. Pagan, A.J. Beaudoin, K.E. Nygren, and D.J. Shadle, 2020, “Understanding Micromechanical Material Behavior Using Synchrotron X-rays and In Situ Loading,” Metallurgical and Materials Transactions A 51:4360–4376.

86 K.S. Shanks, H.T. Phillip, J.T. Weizerorick, et al., 2021, “Characterization of a Small-Scale Prototype Detector with Wide Dynamic Range for Time-Resolved High-Energy X-ray Applications,” IEEE Transactions on Nuclear Science 68:2753–2761.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Image
FIGURE 3-5 (a) Build array drawing of 120 tensile specimens. (b) Actual printed array of 120 tensile specimens. (c) A Generation II high-throughput tensile tester.
SOURCES: (a and b) Reprinted from B.C. Salzbrenner, J.M. Rodelas, J.D. Madison, et al., “High-Throughput Stochastic Tensile Performance of Additively Manufactured Stainless Steel,” Journal of Materials Processing Technology 241:1–12, https://www.sciencedirect.com/science/article/abs/pii/S0924013616303727. Copyright (2017), with permission from Elsevier. (c) Reprinted from N.M. Heckman, T.A. Ivanoff, A.M. Roach, et al., “Automated High-Throughput Tensile Testing Reveals Stochastic Process Parameter Sensitivity,” Materials Science and Engineering: A 772:138632, https://www.sciencedirect.com/science/article/abs/pii/S0921509319314182. Copyright (2020), with permission from Elsevier.
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

data volumes are needed. Beyond this, cloud-based systems that store, annotate, and share archival multimodal data sets are missing but would permit community-based development of property models and application of AI/ML approaches and would result in discovery at the data location without moving the large volumes of data.

Soft Matter and Polymers

There have been significant advances in synthetic techniques that allow for the construction and assembly of increasingly complex and sophisticated organic (soft) materials with control over composition, structure, and dimensionality from the molecular to macromolecular to nanoscopic scales, providing opportunities to tune properties, enhance performance, and expand application scopes. One example is dynamic covalent chemistry, discussed in more detail at the end of this section. Such materials classes range from biologically derived natural substances to synthetic materials and include hybridized biologic-synthetic structures. Indeed, there is much overlap between the biologic and synthetic worlds, with biology and biochemistry often serving as inspiration in the design and discovery of synthetic materials, and synthetic or hybrid materials finding applications as biomaterials. In all cases, having abilities to control and determine composition-structure-properties-performance relationships are critical. One example is the DMREF project 2118860/211886187 designing polymer-protein hybrids via ML on a robotic platform.

Synthetic organic chemistry is the foundation discipline for building organic molecules, which over the past several decades has gained exquisite control over the 3D structure of molecular frameworks. The dimensional evolution of synthetic organic chemistry toward large macromolecular structures is accomplished by several different types of interactions of differing levels of strength—from strong covalent bonds to weaker supramolecular interactions. Synthetic biology is at a more emergent level of development, with exciting future possibilities. Also emergent is an expansion of the use of mechanical bonding:88 entangled systems that take advantage of combinations of covalent and supramolecular interactions within architectures that provide for molecular interlocking.

Key parameters governing the properties and applications of macromolecular or polymer materials, in particular, include their chemical composition, topology (architecture), and size. Throughout the latter part of the 20th century, efforts

___________________

87 Specifics of the project can be found in the grant, available at DMREF, “Projects,” https://dmref.org/projects, accessed September 29, 2022.

88 C.J. Bruns and J.F. Stoddart, 2017, The Nature of the Mechanical Bond: From Molecules to Machines, Hoboken, NJ: John Wiley & Sons.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

focused on developing synthetic approaches to control macromolecular architecture, extending from typical linear polymer chains of bi-directionally connected repeating units to multiple interconnections leading to branched, hyperbranched, dendritic, bottlebrush, and other topologies.89 Installation of bonds between repeat units may be accomplished by step-growth condensation, chain-growth addition, or combined mechanisms, which determine the extent of control over the molecular growth, affect the kinetics of polymerization, and define characteristics of the molecular size and size distribution. Significant efforts over the past few decades have led to the development of well-defined polymer structures via several different types of polymerization chemistries. Traditionally, such processes were intended to be controlled to limit the breadth of the distribution of chain lengths (molecular sizes) and afford polymers having narrow dispersities. More recently, it has been recognized that being able to tune the dispersity and the shape of the molecular size distribution provides opportunities to achieve unique behaviors for polymers during processing and in their final intended properties.90 Another traditional process that has also been replaced recently involved the laborious synthesis of individual batches of polymers having differing chain lengths to achieve materials of differing properties. Over the past few years, it has been recognized that simple and automated chromatographic separation may be employed as a pseudo advanced and more selective fractionation process by which a single batch of disperse polymer chains can lead to isolation of discrete materials having different molar masses with their coincidently distinct properties.91 Although interesting advances have been made,92 grand challenges remain in achieving high degrees of sequence control over composition along a polymer framework (e.g., as occurs in proteins, polysaccharides, poly[nucleic acid]s, and other biopolymers of nature).

___________________

89 Z. Qu, S.Z.D. Cheng, and W.-B. Zhang, 2021, “Macromolecular Topology Engineering,” Trends in Chemistry 3:402–415.

90 S.I. Rosenbloom, J.H. Hsu, and B.P. Fors, 2022, “Controlling the Shape of the Molecular Weight Distribution for Tailored Tensile and Rheological Properties in Thermoplastics and Thermoplastic Elastomers,” Journal of Polymer Science 60:1291–1299.

91 J. Lawrence, S.-H. Lee, A. Abdilla, et al., 2016, “A Versatile and Scalable Strategy to Discrete Oligomers,” Journal of the American Chemical Society 138:6306–6310.

92 J.C. Barnes, D.J.C. Ehrlich, A.X. Gao, et al., 2015, “Iterative Exponential Growth of Stereo- and Sequence-Controlled Polymers,” Nature Chemistry 7:810–815; W.R. Gutekunst and C.J. Hawker, 2015, “A General Approach to Sequence-Controlled Polymers Using Macrocyclic Ring Opening Metathesis Polymerization,” Journal of the American Chemical Society 137:8038–8041; J.F. Lutz, 2017, “Defining the Field of Sequence-Controlled Polymers,” Macromolecular Rapid Communications 38:1700582; C. Yang, K.B. Wu, Y. Deng, J. Yuan, and J. Niu, 2021, “Geared Toward Applications: A Perspective on Functional Sequence-Controlled Polymers,” ACS Macro Letters 10:243–257.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

A currently intense forefront in polymer synthesis involves the study of chemically modifiable or “editable” polymer backbone structures.93 For many years, chemical modifications were focused on side chain moieties as a means to alter polymer properties, whereas backbone chemistries were either meant to be robust, to take advantage of the strength and endurance of the plastic-like properties of polymer chains, or cleavable, to allow for polymer degradation once the purpose was complete. Dynamic covalent chemistry has evolved from small molecules to macromolecular systems, allowing for self-healing, reshaping, or recycling of polymers.94 Indeed, polymer recycling and upcycling have become of critical importance as the persistence and accumulation of plastics in the environment is becoming not only a problem of pollution but also a threat to the health and well-being of inhabitants of our planet. Significant efforts are under way not only toward various recycling strategies to address plastics pollution but also with attention to the sourcing of natural building blocks for polymer production, thereby moving beyond petrochemical feedstocks to develop a sustainable circular economy for plastics.95

Synthesis of Nanoscale Materials

The synthesis of advanced materials with well-defined properties, such as controlled size, shape, composition, and morphology, has become foundational to materials science.96 Significant advances have been made in the design and preparation of nanoscale materials that are transforming research and innovation. Typically, materials synthesis involves top-down methods (e.g., nanolithography and mechanical methods) and bottom-up approaches (e.g., chemical and physical techniques). DMREF-funded research has resulted in new synthetic routes for metallic, semiconductor, ceramic, biological, and polymeric materials. For example, top-down lithographic approaches have been developed and include the use of radiant energy applied on films to create patterns on surfaces. These techniques employ photons (photolithography or optical phase-shifting lithography), electrons

___________________

93 R.A.J. Ditzler and A.V. Zhukhovitskiy, 2021, “Sigmatropic Rearrangements of Polymer Backbones: Vinyl Polymers from Polyesters in One Step,” Journal of the American Chemical Society 143:20326–20331; A.D. Fried, B.J. Wilson, N.J. Galan, and J.N. Brantley, 2022, “Electroediting of Soft Polymer Backbones,” Journal of the American Chemical Society 144:8885–8891.

94 N. Zheng, Y. Xu, Q. Zhao, and T. Xie, 2021, “Dynamic Covalent Polymer Networks: A Molecular Platform for Designing Functions Beyond Chemical Recycling and Self-Healing,” Chemical Reviews 121:1716–1745.

95 Ellen MacArthur Foundation, n.d., “A Circular Economy for Plastic: Designing Out Plastic Pollution,” https://ellenmacarthurfoundation.org/topics/plastics/overview, accessed July 8, 2022.

96 P. Zhao, N. Li, and D. Astruc, 2013, “State of the Art in Gold Nanoparticle Synthesis,” Coordination Chemistry Reviews 257:638–665.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

(electron-beam lithography), ions (focused-ion beam, neutral atomic beam), x-rays (usually from a synchrotron source), and nanoimprint lithography (thermal or ultraviolet [UV]-curable). Lithographic techniques have the advantage of leading to devices at the nanoscale; however, the systems required for lithographic processes are expensive. Alternatively, bottom-up approaches have been used to develop materials at relatively lower costs. Examples of chemical methods include wet-chemical synthesis, sol-gel processing, electrospinning, microwave synthesis, chemical vapor deposition, spray pyrolysis, redox processes, co-precipitation, sonochemistry, microemulsion, hydrothermal synthesis, template synthesis, intercalation, in-situ intercalative polymerization, melt intercalation, mixing, and in-situ polymerization. Examples of physical methods include atomic layer deposition, laser ablation, sputtering, mechanical milling, laser pyrolysis, spark discharge, and radiolysis. In addition, there has also been significant progress in the development of sustainable and biological methods for materials synthesis. One example is biologically synthesized metal nanoparticles for cancer theranostics.97

Characterization Instruments and Facilities

The characterization of advanced materials, to gain accurate information of their chemical and physical properties, is critical for both fundamental and applied studies. It is well recognized that institutions of higher education have invested significant funds to support characterization instruments and facilities. General characterization methods98 include the following:

  • Microscopy, including transmission electron microscopy, scanning electron microscopy, atomic force microscopy, helium ion microscopy, structured illumination microscopy, confocal laser scanning microscopy, scanning probe microscopy techniques, stochastic optical reconstruction microscopy, and correlated light and electron microscopy;99
  • Optical methods, including Raman spectroscopy, infrared spectroscopy, UV-visible spectroscopy, and dynamic light scattering; and

___________________

97 See, for example, S. Mukherjee and C.R. Patra, 2017, “Biologically Synthesized Metal Nanoparticles: Recent Advancement and Future Perspectives in Cancer Theranostics,” Future Science OA 3(3):FSO203, https://doi.org/10.4155/fsoa-2017-0035.

98 See, for example, C.R. Brundle, C.A. Evans, Jr., and S. Wilson, eds., 1992, Encyclopedia of Materials Characterization: Surfaces, Interfaces, Thin Films, Amsterdam, Netherlands: Elsevier, https://www.sciencedirect.com/book/9780080523606/encyclopedia-of-materials-characterization, accessed September 29, 2022.

99 P. de Boer, J.P. Hoogenboom, and B.N.G. Giepmans, 2015, “Correlated Light and Electron Microscopy:Ultrastructure Lights Up!” Nature Methods 12:503–513.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
  • Surface techniques, including Auger spectroscopy, thermogravimetric analysis, dynamic mechanical analysis, x-ray diffraction, x-ray photoelectron spectroscopy, and Brunauer-Emmett-Teller surface analysis.

As discussed further in a section below, federal agencies such as NSF, the National Institutes of Health, the Department of Defense (DoD), and the U.S. Department of Agriculture have funding programs that make instrumentation/equipment grants possible. However, the most advanced facilities are the national laboratories, which are funded and administrated by DOE. The facilities housed at the national laboratories provide indispensable capabilities that make the United States competitive globally. Yet, a major challenge exists in researchers having the awareness of the capabilities available at the national laboratories as well as being able to access the instruments in a timely fashion. Here, there are significant barriers that one must overcome to gain access to the facilities. They include identifying the right contacts at a national laboratory, being eligible (with respect to security checks) to enter the laboratories, and gaining access to the facility in a timely manner—a process that depends on having a proposal to use the facility being approved and funded.

AUTOMATED SYNTHESIS AND CHARACTERIZATION

Greater emphasis on the autonomous synthesis of materials is needed to meet the explosive growth in the amount of available information from experiments and the frequent crossover among theory, computation, and data science. While there has been tremendous progress and success in the methods used to synthesize and characterize materials, obtaining well-defined products, while critical to the success of the project, is not often the outcome. The reason is that the synthesis of materials is often a laborious process, requiring the individual expertise of researchers or laboratories that rely strongly on trial-and-error methods, making the goal-oriented optimization and reproducibility of the products challenging.

The selection, design, and discovery aspects of materials synthesis have been drastically enhanced by the use of high-throughput experiments and computer-aided technologies that assist with processes such as data mining or ML algorithms.100 However, despite significant progress, finding optimized synthetic conditions for materials synthesis remains an obstacle. As such, there is a desire to combine automation with AI to enable the autonomous synthesis of materials such

___________________

100 Y. Li, X. Lingling, Y. Fan, Q. Wang, and M. Hu, 2022, “Recent Advances in Autonomous Synthesis of Materials,” ChemPhysMater 1:77–85.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

as nanoparticles or thin films with well-defined structures.101 Some recent successes include the development and use of “Chemputer,” a modular, automated platform, for the automated synthesis of nanoparticles,102,103 and the autonomous research system for the chemical vapor deposition of single-walled nanotubes.

In sum, innovating on-demand, autonomous processes would enable researchers to accelerate the goal-oriented development of advanced materials, something that also will decrease the time it takes for material commercialization, instead of focusing on repetitive optimization and information collection tasks. Thus, researchers must place a greater focus on the autonomous synthesis of materials to meet the goals of the MGI and the DMREF program.

High-Throughput Experimentation for Battery Materials

A key challenge for conducting high-throughput experimentation on battery materials and components is the necessity of electrochemical testing of such materials and components in cells. Analytical characterization is, unfortunately, inadequate to predict successful performance. A high-throughput workflow must therefore include not only synthesis of large numbers of materials but also the means to test those materials in cells.

The simplest approach to high-throughput materials research for batteries uses thin film deposition techniques (e.g., sputtering, physical vapor deposition) to deposit compositional gradients of electrochemically active materials onto a substrate.104,105 A solid-state electrolyte, such as LIPON (lithium phosphate), is then sputtered onto the cathode active material followed by deposition of a lithium metal anode. The resulting cell structure can then be cycled at various spatial positions to evaluate the electrochemistry of specific compositions. However, the resulting cell structures are far from the relevant form factors of a battery, where

___________________

101 E. Egorov, C. Pieters, H. Korach-Rechtman, J. Shklover, and A. Schroeder, 2021, “Robotics, Microfluidics, Nanotechnology and AI in the Synthesis and Evaluation of Liposomes and Polymeric Drug Delivery Systems,” Drug Delivery and Translational Research 11:345–352.

102 P.S. Gromski, J.M. Granda, and L. Cronin, 2020, “Universal Chemical Synthesis and Discovery with ‘The Chemputer,’” TRECHEM 2:4–12.

103 D. Angelone, A.J.S. Hammer, S. Rohrbach, et al., 2021, “Convergence of Multiple Synthetic Paradigms in a Universally Programmable Chemical Synthesis Machine,” Nature Chemistry 13:63–69.

104 J.F. Whitacre, W.C. West, and B.V. Ratnakumar, 2003, “A Combinatorial Study of LiyMnxNi2-xO4 Spinel Cathode Materials Using Microfabricated Solid-State Electrochemical Cells,” Journal of the Electrochemical Society 150:A1676.

105 J.R. Dahn, S. Trussler, T.D. Hatchard, et al., 2002, “Economical Sputtering System to Produce Large-Size Composition-Spread Libraries Having Linear and Orthogonal Stoichiometry Variations,” Chemistry of Materials 14(8):3519–3523; M.D. Fleischauer, T.D. Hatchard, A. Bonakdarpour, and J.R. Dahn, 2005, “Combinatorial Investigations of Advanced Li-Ion Rechargeable Battery Electrode Materials,” Measurement Science and Technology 16(1):212.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

active material particles are formulated as porous electrodes. While cyclic voltammetry can demonstrate promising materials in terms of voltage/capacity curves, other properties like power performance, lifetime, and safety cannot be meaningfully tested. Even the capacity of the material can be affected by the physical form of the active material, which in this approach is only relevant to thin film batteries.

More relevant approaches involve high-throughput synthesis of the active material particles. This allows researchers to test the effect of particle size, particle size distribution, and particle morphology as well as composition. In addition, electrodes can be produced from the active material particles in which performance sensitivity to electrode recipe (e.g., binders, conductive additives) can also be determined. The electrode loading (mg/cm2) and density also have a significant effect on energy density, power, and lifetime. It is, therefore, critical to be able to formulate and produce variations of electrodes from the synthesized active materials. Screening of liquid electrolyte components (e.g., solvents, salts, and additives) is well suited to combinatorial methods. Finally, all of the components must be dried and assembled into sealed cells in dry environments. The cells must then be electrochemically cycled, which requires tens of thousands of test channels and a sophisticated database for data processing and analysis. Figure 3-6 shows some of the variables critical to lithium-ion battery performance that can be studied using a high-throughput workflow.

To date, few groups have the resources and expertise to be successful in high-throughput battery research that is relevant to today’s applications. Thin film approaches cannot determine key properties such as cycle life.

Some groups have been successful in establishing workflows for a single component of the battery. For example, the University of Muenster MEET has published work on high-throughput screening of liquid electrolytes.106 The workflow uses robotic liquid handlers for automated solution preparation. Testing consists of measuring electrolyte properties, such as conductivity, instead of testing in full cells with active material. University of Osaka researchers report electrolyte screening in microarrays using lithium metal anodes.107 In their equipment, the coulombic efficiency of lithium plating and stripping was measured on a nickel substrate. Practically, the electrolyte performance on lithium metal anodes in this type of experiment may not correlate with that in a real cell containing a high energy density cathode.

MTI Corporation published a high-throughput workflow for the synthesis of

___________________

106 A.N. Krishnamoorthy, C. Wölke, D. Diddens, et al., 2022, “Data‐Driven Analysis of High-Throughput Experiments on Liquid Battery Electrolyte Formulations: Unraveling the Impact of Composition on Conductivity,” Chemistry Methods e202200008.

107 S. Matsuda, K. Nishioka, and S. Nakanishi, 2019, “High-Throughput Combinatorial Screening of Multi-Component Electrolyte Additives to Improve the Performance of Li Metal Secondary Batteries,” Scientific Reports 9:6211, https://doi.org/10.1038/s41598-019-42766-x.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Image
FIGURE 3-6 Some of the variables critical to lithium-ion battery performance that can be studied using a high-throughput workflow.
SOURCE: Courtesy of D. Strand, committee member.

active materials.108 However, it is limited to 16 materials in parallel, which cannot effectively probe all of the variables required to truly optimize a material. There are companies that have established a complete high-throughput workflow from synthesis to testing with the ability to synthesize hundreds of materials per week, formulate multiple electrodes from each, formulate hundreds of liquid electrolytes, assemble cells, and test on tens of thousands of channels.109 This workflow has been developed and tested for more than a decade. In 2021, Pacific Northwest National Laboratory described the construction of a high-throughput battery workflow that includes synthesis and testing of materials.110 Any high-throughput workflow needs to demonstrate correlation of material performance in the high-throughput configuration versus relevant lithium-ion cell formats.

LARGE-SCALE NATIONAL FACILITIES

There are a number of highly advanced tools and facilities that can be accessed at no cost by university or academic researchers funded by NSF grants. These

___________________

108 P. Liu, B. Guob, T. An, H. Fang, G. Zhu, C. Jiang, and X. Jiang, 2017, “High Throughput Materials Research and Development for Lithium Ion Batteries,” Journal of Materiomics 3(3):202–208.

109 For example, see Wildcat Discovery Technologies, “High Throughput Platform (HTP),” https://www.wildcatdiscovery.com/high-throughput-process-htp, accessed June 24, 2022.

110 For more information about Pacific Northwest National Laboratory’s work, see https://www.pnnl.gov/projects/energy-storage-materials-initiative-esmi/innovation, accessed June 24, 2022.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

advanced user facilities are typically not available in academia since they can cost many billions of dollars and hence are typically housed in national laboratories.111 Such advanced laboratories exist at various national laboratories (e.g., Oak Ridge National Laboratory, Argonne National Laboratory, Brookhaven National Laboratory, Los Alamos National Laboratory, and Pacific Northwest National Laboratory), the Department of Commerce laboratories (NIST laboratories in Gaithersburg, Maryland, and Boulder, Colorado), and DoD laboratories (Naval Research Laboratory, Air Force Research Laboratory, and Army Research Laboratory).

The vast majority of facility users are granted access via a competitive peer review of their technical proposals in order to select those with the greatest scientific merit. After a research team’s proposal is accepted, that team is granted time at the facility to use its beam and/or specialized instrumentation for scientific experiments. Final research results are reported in the open scientific literature. Each facility has a user program to describe its specific offerings (e.g., characteristics of the specialized instruments available for use) and to administer the proposal submission and review process for prospective users.112

The BES-supported suite of facilities and research centers provides a unique set of analytical tools for studying the atomic structure and functions of complex materials. These facilities provide key capabilities to correlate the microscopic structure of materials with their macroscopic properties. The synchrotron light sources, producing photons largely over a very wide range of photon energies (from the infrared to hard x-rays), shed light on fundamental aspects of the physical world, investigating energy, momentum, and position using the techniques of spectroscopy, scattering, and imaging applied over various time scales. Neutron sources take advantage of the electrical neutrality and special magnetic properties of the neutron to probe atoms and molecules and their assembly into materials. Electron beam instruments provide the spatial resolution needed to observe individual nanostructures and even single atoms by exploiting the strong interactions of electrons with matter and the ability to readily focus beams of charged particles. The Nanoscale Science Research Centers provide the ability to fabricate complex nanostructures using chemical, biological, and other synthesis techniques, and to characterize, assemble, and integrate them into devices.113

Details of the billions of dollars of investments on advanced instrumentation that can be accessed by university researchers at DOE national laboratories are

___________________

111 National Research Council, 2004, Setting Priorities for Large Research Facility Projects Supported by the National Science Foundation, Washington, DC: The National Academies Press, https://doi.org/10.17226/10895.

112 See https://permanent.access.gpo.gov/lps118921/BES_Facilities.pdf, p. 4, accessed October 17, 2022.

113 DOE, n.d., “About the Scientific User Facilities Division,” https://science.osti.gov/bes/suf/About, accessed September 29, 2022.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

published by DOE’s Office of Science.114 NSF also provides a midscale infrastructure program in the Division of Materials Research. The MIPs115 are designed to accelerate advances in materials research. It is envisioned that each MIP is a scientific ecosystem, which includes in-house research scientists, external users, and other contributors who, collectively, form a community of practitioners and share tools, codes, samples, data, and know-how.116

Opportunities to enhance partnerships between researchers in academia and national laboratories (see Key Finding 6.11 and Key Recommendation 6.11) and strengthen the use of available facilities (see Finding 6.12 and Recommendation 6.12) are further discussed in Chapter 6.

FAIR DATA

The FAIR philosophy is trying to generate additional value out of data that are already created; in this process, the first step is to enable data to be found so they can be reused. Once found, anyone should be able to access these data and use them in applications or workflows for analysis, storage, and processing. The ultimate goal of FAIR is to optimize the reuse of data. To achieve this, metadata and data should be well described so that they can be replicated and/or combined in different settings. There are three types of data: the actual data, the metadata, and any indexing of the data—that is, the infrastructure component.

As stated on the GO FAIR website,

In 2016, the “FAIR Guiding Principles for scientific data management and stewardship” were published in Scientific Data. The authors intended to provide guidelines to improve the Findability, Accessibility, Interoperability, and Reuse of digital assets. The principles emphasize machine-actionability (i.e., the capacity of computational systems to find, access, interoperate, and reuse data with no or minimal human intervention) because humans increasingly rely on computational support to deal with data as a result of the increase in volume, complexity, and creation speed of data.117

The FAIR Guiding Principles can be described in the following steps:

F1: (Meta)data are assigned a globally unique and persistent identifier

F2: Data are described with rich metadata

F3: Metadata clearly and explicitly include the identifier of the data they describe

___________________

114 DOE, n.d.

115 The website for the Materials Innovation Platforms is https://beta.nsf.gov/funding/opportunities/materials-innovation-platforms-mip, accessed September 29, 2022.

116 See NSF, “Materials Innovation Platforms (MIP),” https://beta.nsf.gov/funding/opportunities/materials-innovation-platforms-mip, accessed October 17, 2022.

117 GO FAIR, n.d., “FAIR Principles,” go-fair.org/fair-principles, accessed September 29, 2022.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

F4: (Meta)data are registered or indexed in a searchable resource

A1: (Meta)data are retrievable by their identifier using a standardised communications protocol

A1.1: The protocol is open, free, and universally implementable

A1.2: The protocol allows for an authentication and authorisation procedure, where necessary

A2: Metadata are accessible, even when the data are no longer available

I1: (Meta)data use a formal, accessible, shared, and broadly applicable language for knowledge representation

I2: (Meta)data use vocabularies that follow FAIR principles

I3: (Meta)data include qualified references to other (meta)data

R1: (Meta)data are richly described with a plurality of accurate and relevant attributes

R1.1: (Meta)data are released with a clear and accessible data usage license

R1.2: (Meta)data are associated with detailed provenance

R1.3: (Meta)data meet domain-relevant community standards118

In May 2016, Science Europe published a report on funding research data management and related infrastructures,119 and in April 2022, a Nature paper120 argued for FAIR principles so that data mining and AI can extract useful scientific information from the data.

The importance of FAIR data and opportunities to enhance their prevalence in DMREF-sponsored communities is further discussed in Chapter 6 (see Finding 6.13 and Recommendation 6.21).

REPOSITORIES OF EXPERIMENTAL DATA

Materials are characterized by a multitude of different experimental probes and in many sites worldwide. These provide different, non-correlated data sets that are based on different metadata sets owing to the diversity of users and their research objectives and the large suite of instruments (which often has proprietary formats) used to collect data. For this reason, the data are of interest to the individual user and can be evaluated according to an individual research question, but further integration and use of the data are severely limited. The FAIRmat initiative aims to generate generally accessible data archived according to strict quality rules.

As an example, heterogeneous catalysis is a key interdisciplinary field that provides solutions to ensure society’s future energy supply, environmental protection, and sustainable chemistry by closing chemical cycles. It has proven impossible

___________________

118 GO FAIR, n.d.

119 Science Europe and Knowledge Exchange, 2016, Funding Research Data Management and Related Infrastructures, https://www.scienceeurope.org/media/uuqf0i03/se-ke_briefing_paper_funding_rdm.pdf.

120 M. Scheffler, M. Aeschlimann, M. Albrecht, et al., 2022, “FAIR Data Enabling New Horizons for Materials Research,” Nature 604(7907):635–642, https://doi.org/10.1038/s41586-022-04501-x.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

to infer simple relations among process conditions, catalyst properties, and performance as they generally depend on each other in a highly nonlinear fashion. The heterogeneous catalysis pillar within FAIRmat aims to act as an important platform for collecting existing and future data and making these data accessible to the community.

The importance of experimental data is further discussed in Chapter 6 (see Finding 6.2 and Recommendation 6.8).

FROM RESEARCH TO PRODUCT

The successful discovery, manufacture, and deployment of new materials require a close collaboration between academic and industrial researchers. In order to benefit U.S. society at large, the scientific insights and materials discovery tools developed by academic researchers must ultimately become relevant and applicable for industrial needs. Synergistically, many key advances have historically originated in industry, and industry continues to pave directions for important new materials, fundamental study, and practical implementation. Hence, a concerted effort must be made toward bridging the traditional gap between fundamental scientific discoveries and practical technological applications.

The discovery of novel advanced materials necessarily involves the generation of new intellectual property, which must be protected by patents in order to justify the costs associated with research, development, and commercialization. As such, the number of patent applications and granted patents emerging from the MGI is a critical metric to evaluate whether researchers have been able to bridge the gap between fundamental science and the delivery of industrially relevant materials solutions.

For a new material to become a successful product, it must satisfy requirements for both product- and manufacturing-related attributes. Product-related attributes include any property of the material that is required to achieve the desired product performance. To be a successful product, the material must meet the full set of property requirements and exceed the performance of other established materials. The relevant attributes depend on the application of the material and often include properties related to mechanical, chemical, thermal, optical, or electrical performance.121

Successfully attaining the full set of required product-facing attributes is a necessary but insufficient condition to enable broad adoption of the newly developed material. In order to become a successful product, the material must also be manufacturable at an industrial scale. High-quality manufacturing must be achieved with

___________________

121 J.C. Mauro, 2018, “Decoding the Glass Genome,” Current Opinion in Solid State and Materials Science 22(2):58–64, https://doi.org/10.1016/j.cossms.2017.09.001.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

sufficient yield and while minimizing the cost of fabrication. Materials fabrication costs are governed by both the raw material and processing costs, as well as the costs associated with quality control. Materials processing is thus an essential part of ensuring the success of a new product. For the MGI to be successful in achieving its goal, materials process-related issues would need to be treated as equally important to product-related attributes.

The Valley of Death

The “valley of death” 122 is frequently used to describe a discontinuity in product innovation, as shown in Figure 3-7. While this valley is typically attributed to the gap between fundamental science in academia and applied science and engineering in industry, it can occur within any organization where product development is bridging wide ranges of technology readiness levels (TRLs). As shown in Figure 3-7, the resource minimum tends to occur when the primary funding source transitions

Image
FIGURE 3-7 Representation of the “valley of death” in product innovation showing local minimum in resources as technology moves from research to product/prototype development.
NOTES: The phrase “valley of death” is older than the referenced article from 2015. SME, small and mid-size enterprise.
SOURCES: Courtesy of J.L.M. Hensen, R.C.G.M. Loonen, M. Archontiki, and M. Kanellis, 2015, “Using Building Simulation for Moving Innovations Across the ‘Valley of Death,’” REHVA Journal 52(3):58–62, https://www.rehva.eu/rehva-journal/chapter/using-building-simulation-for-moving-innovations-across-the-valley-of-death; Hensen et al., figure 2, https://www.researchgate.net/figure/Availability-of-resources-for-new-product-development-at-various-TRLs-The-gap-in-the_fig4_276205251.

___________________

122 In talking to many researchers during this study, the committee found little to no process describing a handoff from academia to industry—that seems to be the problem.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

from government to the private sector. While this discontinuity often occurs during the handoff between academia (government funding) and large business (private-sector funding), small businesses are more likely to bridge the mid-TRLs. These trends reflect a willingness to take on risk, with government funding used to seed early, risky concepts and large corporations avoiding risk by developing technology that is already demonstrated at higher TRLs. Small companies, especially start-ups, can use funding sources such as Small Business Innovation Research grants or venture capital funds (but often not both) to move technology demonstrated in the laboratory into a product. However, the valley of death does not go away as evidenced by the high failure rate (~68 percent) of technology start-ups.123

An analysis of the 37 DMREF 2021 projects124 shows that only 4 projects had made use of NSF’s Grant Opportunities for Academic Liaison with Industry (GOALI) proposal framework in the description (~11 percent). Of these four projects, only one industrial participant (Bristol Myers Squibb) had significant experience bringing relevant products to market. Of the other three, two advertise product development to aid other companies in bringing products to market. For example, one develops programmable chemistries for a variety of applications. While these chemistries are no doubt useful to achieve the project goals, the industrial partner would not be a development/commercialization partner for the ultimate manufacturing of a CO2 capture product. It should be noted that in a search of all 508 DMREF grants, 26 (~5 percent) were found to be linked to GOALI. This suggests that increasing private sector involvement in early research is an important goal for DMREF moving forward.

The discontinuity in the TRL timeline occurs for all technologies, not just advanced materials. It is not likely that any individual funding agency can solve the problem on its own.125 However, each specific agency, as well as DMREF, could take steps to help ensure that promising technologies get commercialized and benefit society. The DMREF program might consider steps such as those outlined in Recommendation 3.2 below.

FINDING 3.1: It appears that only around 10 percent of the DMREF 2021 projects had a strong connection to industry in such a way that the progression of fundamental materials research toward eventual deployment and manufacturing had a clear path.

___________________

123 K. Kotashev, 2022, “Start-Up Failure Rate: How Many Startups Fail and Why?” https://www.failory.com/blog/startup-failure-rate, last updated January 9, 2022.

124 The list of projects can be found at DMREF, “Projects,” https://dmref.org/projects, accessed September 29, 2022.

125 A number of PIs who the committee interviewed were disappointed that just when they were finding interesting materials and going beyond the initial discovery phase their funding was discontinued.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

RECOMMENDATION 3.1: The DMREF program needs to find pathways to substantially increase collaboration and participation with industry. The DMREF program should also consider pathways to increase academia’s awareness and training in project management and implementation strategies.

RECOMMENDATION 3.2: The DMREF program should consider the following steps to improve commercialization potential:

  • Setting targets for a specific percentage of projects in the DMREF portfolio that receive commercialization support from programs such as Grant Opportunities for Academic Liaison with Industry, Innovation Corps, and Industry-University Cooperative Research Centers;
  • Soliciting advisors from U.S. companies that design, develop, and manufacture products to represent a portion of the reviews for DMREF proposals;
  • Staffing a commercialization director for the program to ensure intellectual property gets filed and technology transfer functions at funded universities are effective;
  • Holding project review sessions specifically for industry to get a snapshot of what is being developed;
  • Keeping a compilation of filed patents, and following licensing activities, to monitor technology transition to commercial products; and
  • Developing a post-DMREF effort that provides support for working with industry on promising DMREF outcomes, including testing at production scale.

Additional opportunities to strengthen engagement with industry are discussed in Chapter 6 (see Recommendation 6.17).

Manufacturing of the Future: Beyond DMREF

Improvements to U.S. manufacturing focus on increasing competitiveness and promoting a sustainable manufacturing infrastructure. These emphases are not specific to a manufacturing segment but are especially valuable for advanced materials. New, novel materials may require expensive elements or precursors or materials that have to be imported to the United States. They may require a highly skilled workforce, which drives up costs. However, lower-cost robotics could empower materials research and lead to faster and more effective theoretical-computational-experimental iterations toward optimized materials. In addition, the new materials may have inherent properties that require more expensive, energy-intensive processes. Finally, regulatory and compliance costs for brand new materials can also be a hindrance to bringing new materials to manufacturing. Therefore, new

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×

manufacturing technologies that reduce waste, minimize energy consumption, and eliminate emissions can make manufacturing of new materials more attractive to U.S. companies.

The DMREF project portfolio already includes a multitude of materials to address global societal issues, such as climate change, water contamination and shortages, and air pollution. For example, DMREF has currently funded projects on active materials and ionic conductors for batteries, high strength/high temperature alloys for improving jet engine efficiencies, and more efficient catalysts for diverse applications. In fact, the entire DMREF concept of accelerating the development of advanced materials aligns with the 2020 Recommendations for Strengthening American Leadership in Industries of the Future.126 These recommendations call out efforts in advanced manufacturing, including AI/ML for generative design, additive manufacturing, advanced robotics for manufacturing, lightweight innovations, advanced composites, and advanced functional fabrics.

The mission of NSF is to promote the progress of science with a vision of capitalizing on new concepts in science and engineering while providing global leadership in advancing research and education. It is, therefore, not realistic for DMREF to directly fund activities at mid- to high TRLs. However, DMREF projects could benefit from industrial advisors in areas such as sustainability and from choosing low-cost or domestically sourced precursors for advanced material synthesis. Factors like low energy processes to make materials should be considered during the review/decision-making process. Finally, it is the committee’s hope that the intellectual property resulting from DMREF projects is later available to U.S. manufacturing companies at low cost.

Additional opportunities to bridge the TRL continuum for DMREF-sponsored research are discussed in Chapter 6 (see Key Finding 6.15).

___________________

126 President’s Council of Advisors on Science and Technology, 2020, Recommendations for Strengthening American Leadership in Industries of the Future, A Report to the President of the United States of America, Washington, DC, https://science.osti.gov/-/media/_/pdf/about/pcast/202006/PCAST_June_2020_Report.pdf.

Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 39
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 40
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 41
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 42
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 43
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 44
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 45
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 46
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 47
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 48
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 49
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 50
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 51
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 52
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 53
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 54
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 55
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 56
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 57
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 58
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 59
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 60
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 61
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 62
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 63
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 64
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 65
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 66
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 67
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 68
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 69
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 70
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 71
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 72
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 73
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 74
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 75
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 76
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 77
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 78
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 79
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 80
Suggested Citation:"3 The Current State of Materials Research." National Academies of Sciences, Engineering, and Medicine. 2023. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF). Washington, DC: The National Academies Press. doi: 10.17226/26723.
×
Page 81
Next: 4 The Talent Pool »
NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF) Get This Book
×
 NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF)
Buy Paperback | $28.00 Buy Ebook | $22.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

The Materials Genome Initiative (MGI) was launched in 2011 by the White House Office of Science and Technology Policy to help accelerate the design, discovery, development and deployment of advanced materials and to reduce costs through the integration of advanced computation and data management with experimental synthesis and characterization. A broad range of federal agencies - including the National Science Foundation (NSF), the Department of Energy, and the Department of Defense - are part of the MGI effort and have invested more than $1 billion in resources and infrastructure accumulative since the start.

The efforts of NSF have been focused largely within the Designing Materials to Revolutionize and Engineer Our Future (DMREF) program, which supports the development of fundamental science, computational and experimental tools for generating and managing data, and workforce that enable industry and other government agencies to develop and deploy materials that meet societal needs and national priorities. At the request of NSF, this report evaluates the goals, progress, and scientific accomplishments of the DMREF program within the context of similar efforts both within the United States and abroad. The recommendations of this report will assist NSF as it continues to increase its engagement with industry and federal agencies to transition the results from fundamental science efforts to reach the MGI goal of deploying advanced materials at least twice as fast as possible today, at a fraction of the cost that meet national priorities.

READ FREE ONLINE

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!