1. |
EXECUTIVE SUMMARY |
1.1. |
What is materials informatics? |
1.2. |
AI opportunities at every stage of materials design and development |
1.3. |
Problems with materials science data |
1.4. |
Types of MI algorithms - Supervised vs unsupervised |
1.5. |
Key areas of algorithm advancements in MI |
1.6. |
Simulation data is an important input to MI processes |
1.7. |
Large Language Models and MI: what are the possibilities? (I) |
1.8. |
Large Language Models and MI: what are the possibilities? (II) |
1.9. |
Materials informatics players - categories |
1.10. |
Conclusions and outlook for strategic approaches: approaches for end-users (I) |
1.11. |
Conclusions and outlook for strategic approaches: approaches for end-users (II) |
1.12. |
For MI end-users, there is no one-size-fits-all approach |
1.13. |
Key Partners and Customers of External Providers |
1.14. |
Main industry players (I): Established leaders |
1.15. |
Main industry players (II): Promising challengers |
1.16. |
Notable MI consortia |
1.17. |
Materials Informatics: the state of the industry in 2024 |
1.18. |
Project categories in MI |
1.19. |
Market forecast: external materials informatics players |
1.20. |
Market outlook for external MI companies |
1.21. |
Materials informatics - Market penetration by maturity |
1.22. |
Materials informatics roadmap |
2. |
INTRODUCTION |
2.1. |
What is materials informatics? |
2.2. |
Materials informatics - Why now? |
2.3. |
What can ML/AI do in materials science? |
2.4. |
Materials Informatics - Category definitions |
2.5. |
The broader informatics space in science and engineering |
2.6. |
Key challenges for MI across the full materials spectrum |
2.7. |
Closing the loop on traditional synthetic approaches |
2.8. |
High Throughput Virtual Screening (HTVS) |
2.9. |
Advantages of ML for chemistry and materials science - Acceleration |
2.10. |
Advantages of ML for chemistry and materials science - Scoping and screening |
2.11. |
Advantages of ML for chemistry and materials science - Scoping and screening (2) |
2.12. |
Advantages of ML for chemistry and materials science - New species and relationships |
2.13. |
Data infrastructures for chemistry and materials science |
2.14. |
ELN/LIMS Software and Materials Informatics |
3. |
TECHNOLOGY ASSESSMENT |
3.1. |
Overview |
3.1.1. |
Inputs and outputs of materials informatics algorithms |
3.1.2. |
What is needed for materials informatics? |
3.1.3. |
Summary of technology approaches |
3.1.4. |
Uncertainty in experimental data undermines analysis |
3.1.5. |
QSAR and QSPR: relating structure to properties |
3.2. |
MI algorithms |
3.2.1. |
Overview of MI algorithms |
3.2.2. |
Problems with materials science data |
3.2.3. |
Descriptors and training a model |
3.2.4. |
Describing materials to a computer (I) |
3.2.5. |
Describing materials to a computer (II) |
3.2.6. |
Types of MI algorithms - Supervised vs unsupervised |
3.2.7. |
Problem classes in supervised and unsupervised learning |
3.2.8. |
Reinforcement learning: Learning by trial and error |
3.2.9. |
Automated feature selection |
3.2.10. |
Exploitation vs exploration: Use what you know or look into new areas? |
3.2.11. |
Pure exploitation vs epsilon-greedy policies in materials informatics |
3.2.12. |
Active learning and MI: Choosing experiments to maximize improvement |
3.2.13. |
Supervised learning models: "More sophisticated" is not always better |
3.2.14. |
Bayesian optimization: A versatile tool in machine learning |
3.2.15. |
Genetic algorithms: Mimicking natural selection |
3.2.16. |
Unsupervised learning case study - Mapping phases |
3.2.17. |
Deep learning: Imitating the brain |
3.2.18. |
Deep learning: Types of neural network |
3.2.19. |
Generative vs discriminative algorithms - Explaining vs labelling |
3.2.20. |
Transformer models are at the core of the AI boom |
3.2.21. |
Generative algorithms in materials informatics: case study |
3.2.22. |
Deep learning: An example in MI |
3.2.23. |
Generative models for inorganic compounds (I) |
3.2.24. |
Generative models for inorganic compounds (II): Generative adversarial networks |
3.2.25. |
How to work with small material datasets |
3.2.26. |
Deep learning with small material datasets: examples (I) |
3.2.27. |
Deep learning with small material datasets: examples (II) |
3.2.28. |
Large Language Models (LLMs) and Materials R&D |
3.2.29. |
Capabilities of LLMs in science |
3.2.30. |
Summary: Key areas of algorithm advancements in MI |
3.3. |
Establishing a data infrastructure |
3.3.1. |
A data infrastructure is critical for MI |
3.3.2. |
Developments targeted for chemical and materials science |
3.3.3. |
ELN/LIMS, materials informatics and managing R&D processes |
3.4. |
External databases |
3.4.1. |
Data repositories - Organizations |
3.4.2. |
Leveraging data repositories |
3.4.3. |
Text extraction and analysis |
3.4.4. |
ChemDataExtractor V1.0: Data mining publications and patents |
3.4.5. |
ChemDataExtractor V2.0: Mining relational data |
3.4.6. |
Annotating and extracting the relevant information: The commercial space |
3.5. |
MI with physical experiments and characterization |
3.5.1. |
Achieving high-volumes of physical experimental data |
3.5.2. |
Achieving high-volumes of physical experimental data (2) |
3.5.3. |
High-throughput spectroscopy |
3.5.4. |
In-situ spectroscopy developments |
3.6. |
MI with computational materials science |
3.6.1. |
Simulations for chemistry and materials science R&D |
3.6.2. |
Density functional theory (DFT) - Quantum mechanical modelling for CAMD |
3.6.3. |
Surrogate models reduce the computational expense of atomistic simulation |
3.6.4. |
Simulating matter across the length scale continuum: multiscale modelling |
3.6.5. |
ICME and the role of machine learning |
3.6.6. |
ICME: Why is it important? |
3.6.7. |
QuesTek Innovations and ICME: from service to SaaS |
3.6.8. |
Thermo-Calc and CompuTherm: ICME software provision and QuesTek collaboration |
3.6.9. |
Generating and using the largest computational materials science database |
3.6.10. |
Explorative design utilizing cloud-based simulation |
3.6.11. |
The potential in leveraging quantum computing |
3.6.12. |
Computation autonomy for materials discovery |
3.6.13. |
Summary: simulation data is an important input to MI processes |
3.7. |
Autonomous labs |
3.7.1. |
The future - fully autonomous labs |
3.7.2. |
The future - "Chemputer" |
3.7.3. |
DeepMatter and the Chemputer |
3.7.4. |
Workflow management for laboratory automation |
3.7.5. |
Autonomous high throughput experimentation |
3.7.6. |
Commercial self-driving-laboratories |
3.7.7. |
Gearu: attempting to commercialize mobile autonomous robotic scientists |
3.7.8. |
Retrosynthesis through to robot execution |
3.7.9. |
Technology pillars for chemical autonomy |
4. |
INDUSTRY ANALYSIS |
4.1. |
Overview |
4.1.1. |
Materials Informatics: the state of the industry in 2024 |
4.2. |
Strategic approaches to MI |
4.2.1. |
Materials informatics players - categories |
4.2.2. |
Conclusions and outlook for strategic approaches: approaches for end-users (I) |
4.2.3. |
Conclusions and outlook for strategic approaches: approaches for end-users (II) |
4.2.4. |
Conclusions and outlook for strategic approaches; approaches for external materials informatics companies (I) |
4.2.5. |
Conclusions and outlook for strategic approaches; approaches for external materials informatics companies (II) |
4.3. |
Player analysis |
4.3.1. |
Materials informatics players - overview |
4.3.2. |
Key Partners and Customers of External Providers |
4.3.3. |
Partnerships with engineering simulation software |
4.3.4. |
Funding raised by private companies (I): in-house development leads to high capital requirements |
4.3.5. |
Funding raised by private companies (II): the AI boom may be raising interest in MI |
4.3.6. |
NobleAI: MI, Microsoft, the AI boom and cloud marketplaces |
4.3.7. |
Main industry players (I): Established leaders |
4.3.8. |
Main industry players (II): Promising challengers |
4.3.9. |
Major MI players: on a path to profitability? |
4.3.10. |
What are the barriers to profitability for MI SaaS players? |
4.3.11. |
Taking materials informatics in-house |
4.3.12. |
Offering in-housed operations as a service |
4.3.13. |
Taking the operation in-house: What needs to happen first? |
4.3.14. |
Enthought: Digital transformation in scientific/engineering R&D |
4.3.15. |
Resonac/Showa Denko - from external engagements to in-housed MI strategy? |
4.3.16. |
Retrosynthesis prediction: "Can I make this compound?" |
4.3.17. |
Commercial retrosynthesis predictors |
4.3.18. |
Notable MI consortia (1) - NIMS and Materials Open Platforms |
4.3.19. |
Notable MI consortia (2) - AIST Data-Driven Consortium |
4.3.20. |
Notable MI consortia (3) - Toyota Research Institute and university collaboration |
4.3.21. |
Notable MI consortia (4) - The Global Acceleration Network |
4.3.22. |
Notable past MI consortia (1) - IBM collaborations |
4.3.23. |
Notable past MI consortia (2): CHiMaD and the CMD Network |
4.3.24. |
Public-private collaborations |
4.3.25. |
The Open Catalyst Project: Crowdsourcing MI |
4.3.26. |
Materials Genome Initiative (MGI) |
4.3.27. |
Materials Genome Engineering (MGE) or National Materials Genome Project (China) |
4.3.28. |
Additional key initiatives and research centers around the world (1) |
4.3.29. |
Additional key initiatives and research centers around the world (2) |
4.3.30. |
Conclusion: for MI end-users, there is no one-size-fits-all approach |
4.4. |
Applications of materials informatics |
4.4.1. |
Project categories in MI |
4.4.2. |
Application Progression |
4.4.3. |
Materials informatics roadmap |
4.5. |
Market forecast and outlook |
4.5.1. |
Signs of growth in the MI industry |
4.5.2. |
Market forecast: external materials informatics players |
4.5.3. |
Forecast data and market outlook |
4.6. |
MI industry player data |
4.6.1. |
Lists of MI players |
4.6.2. |
Full player list - Commercial companies (confirmed operational) (1) |
4.6.3. |
Full player list - Commercial companies (confirmed operational) (2) |
4.6.4. |
Full player list - Commercial companies (confirmed operational) (3) |
4.6.5. |
Full player list - Commercial companies (confirmed operational) (4) |
4.6.6. |
Full player list - Commercial companies (confirmed operational) (5) |
4.6.7. |
Full player list - Commercial companies (confirmed operational) (6) |
4.6.8. |
Full player list - Commercial companies (confirmed operational) (7) |
4.6.9. |
Full player list - Commercial companies (confirmed operational) (8) |
4.6.10. |
Full player list - Commercial companies (confirmed operational) (9) |
4.6.11. |
Full player list - Commercial companies (likely industry leavers in 2023) |
4.6.12. |
Player list - Public organizations (I) |
4.6.13. |
Player list - Public organizations (II) |
5. |
COMPANY PROFILES |
5.1. |
Albert Invent |
5.2. |
Alchemy Cloud |
5.3. |
Ansatz AI |
5.4. |
Citrine Informatics (2020) |
5.5. |
Citrine Informatics (2022) |
5.6. |
Citrine Informatics (2023/4 update) |
5.7. |
Copernic Catalysts |
5.8. |
Cynora |
5.9. |
Elix, Inc |
5.10. |
Enthought |
5.11. |
Exomatter |
5.12. |
Exponential Technologies |
5.13. |
FEHRMANN MaterialsX |
5.14. |
Fluence Analytics |
5.15. |
Intellegens |
5.16. |
Kebotix (2020) |
5.17. |
Kebotix (2022) |
5.18. |
Kyulux |
5.19. |
MaterialsIn |
5.20. |
Materials Zone (2020) |
5.21. |
Materials Zone (2022) |
5.22. |
Matmerize |
5.23. |
META |
5.24. |
OTI Lumionics |
5.25. |
Phaseshift Technologies |
5.26. |
Polymerize |
5.27. |
Preferred Computational Chemistry/Matlantis |
5.28. |
QuesTek Innovations LLC |
5.29. |
Schrödinger |
5.30. |
Stoicheia |
5.31. |
Uncountable (2020) |
5.32. |
Uncountable (2022) |
5.33. |
Xinterra |