DigCat 3.0 Launches!
Summary
DigCat is a digital catalysis platform led by Professor Hao Li's team from Tohoku University, Japan, in collaboration with over 20 research teams worldwide. The platform integrates big data and AI for catalysis materials research, featuring >400,000 experimental performance data points for electro-, thermo-, and photocatalysts, as well as >300,000 catalyst structures. DigCat offers dynamic data visualization, precise literature tracking, and an intelligent Q&A assistant. It also includes cutting-edge tools for microkinetic modeling, AI chatbot based on catalytic data and knowledge, machine learning force field training, and regression model development. This is the world’s first digital catalysis platform and comprehensive electrocatalysis experimental database. URL: https://www.digcat.org/
Important Note: After register, please contact Prof. Hao Li ([email protected]) to activate your account with the corresponding functions.
DigCat is a digital catalysis platform led by Professor Hao Li's team from Tohoku University, Japan, in collaboration with over 20 research teams worldwide. The platform integrates big data and AI for catalysis materials research, featuring >400,000 experimental performance data points for electro-, thermo-, and photocatalysts, as well as >300,000 catalyst structures. DigCat offers dynamic data visualization, precise literature tracking, and an intelligent Q&A assistant. It also includes cutting-edge tools for microkinetic modeling, AI chatbot based on catalytic data and knowledge, machine learning force field training, and regression model development. This is the world’s first digital catalysis platform and comprehensive electrocatalysis experimental database. URL: https://www.digcat.org/
Important Note: After register, please contact Prof. Hao Li ([email protected]) to activate your account with the corresponding functions.
The research approach combining "big data platforms + AI methods + precise theoretical modeling" is set to revolutionize the future of catalytic material development. This cutting-edge AI lab frontend provides groundbreaking solutions for catalyst synthesis and performance characterization, holding immense potential to drastically reduce development costs and fast-track the discovery of powerful, low-cost catalysts with exceptional performance.
The DigCat digital catalysis platform, led by Professor Hao Li's team at Tohoku University, Japan, is a pioneering research tool developed in collaboration with over ten research teams globally. Integrating big data and AI, DigCat covers >400,000 experimental performance data points for electro-, thermo-, and photocatalysts, as well as >300,000 catalyst structures. The platform offers powerful features such as dynamic data visualization, precise literature tracking, and an intelligent Q&A assistant. It also includes cutting-edge capabilities like microkinetic modeling, an AI chatbot based on catalytic data and knowledge, machine learning force field training, and regression model development.
As of September 12, 2024, the DigCat platform contains >400,000 experimental data points and >330,000 theoretical structures, covering a wide range of material types. The database includes data from over 120 catalytic reactions and more than 200 performance metrics. These primarily involve reactions such as oxygen reduction, oxygen evolution, hydrogen evolution, ammonia synthesis, carbon dioxide reduction, hydrogen peroxide synthesis, ozone synthesis, hydrogen and ammonia oxidation, nitrogen and ammonia oxidation, electrocatalytic hydrogenation, epoxide synthesis, urea synthesis, methanol/ethanol reforming for hydrogen production, and various organic catalytic reactions. It encompasses a variety of materials, including transition metals and their alloys, metal oxides, nitrides, carbides, sulfides, phosphides, noble metals, single-atom catalysts, 2D materials, and perovskites.
In addition to classic catalytic materials like noble metals, as of September 12, 2024, DigCat has the most comprehensive collection of experimental data from literature on various M-N-C electrocatalysts. This includes novel non-noble metal systems such as single-atom and multi-atom metal cluster-doped defective graphene, single-atom molecular fragment catalysts, covalent organic frameworks (COFs), metal-organic frameworks (MOFs), and metal-doped graphitic carbon nitride (g-C3N4).
Function 1: Experimental data analysis and visualization
Users can select the reaction type and material type, and the platform will automatically display performance data with the year on the horizontal axis by default. By adjusting the axis types or entering specific elements of interest, users can easily achieve dynamic data visualization. The platform also offers categorized data analysis, such as classifying experimental data by testing pH values and displaying performance charts under different pH conditions. For specific requirements, a sliding bar allows users to filter data based on reaction conditions, making the search process more efficient. Additionally, for trending research areas like seawater electrolysis, magnetic catalysis, ammonia synthesis, and epoxidation, the platform supports one-click filtering of relevant literature.
Function 2: Paper Tracking and AI Q&A
After filtering the literature, users can access the platform's extended features, such as original document tracing and AI-powered Q&A. By hovering over a data point, users can view the catalyst ID and DOI, and a dialogue box allows them to jump to the original document link. Additionally, users can view experimentally validated theoretical structures online, and download related computational files or experimental data for building or verifying theoretical or machine learning models. The AI Q&A feature enables users to quickly retrieve detailed information on preparation materials, methods, processes, and performance analysis by simply entering the document ID.
Function 3: Benchmarking Analysis
DigCat’s extensive data analysis also provides strong support for standardizing experiments and ensuring data reliability. By comparing the performance of common benchmark samples across multiple studies, users can easily identify outliers in the data and investigate potential causes. This helps streamline experimental procedures and data processing, ensuring more reliable and reproducible results.
Function 4: AI Catalysis Scientist
By combining standardized literature data with large language models, DigCat has launched a new feature called the AI Catalysis Scientist. Compared to the original version and the chatbot based on PDF-text knowledge, the new DigCat-GPT chatbot, integrated with DigCat's structured data repository, offers more detailed, specific, and insightful answers to expert-level questions in the field of catalysis. Additionally, DigCat has incorporated more material databases as sources, including the Mössbauer Spectroscopy Database, the Solid-State Inorganic and Organic Electrolytes Database, and the Thermoelectric Materials Database. By utilizing large language models to match the fundamental structures and properties of these materials with the demands found in literature, the AI Catalysis Scientist significantly broadens the potential materials range for catalyst prediction.
Function 5: On-the-fly Generation of pH-dependent Microkinetic Volcano
In addition to data-driven and large language model-based predictions, the DigCat platform offers cutting-edge theoretical modeling tools and machine learning model training modules to enhance the reliability of catalyst material predictions. With the pH-electric field coupling microkinetic modeling feature, users can select their desired reaction type and model, and input key computational parameters such as linear scaling relations, electric field response parameters, zero-point energy, entropy and solvation corrections, and electrode potentials. This allows them to generate precise catalytic activity volcano plots, providing more reliable theoretical support for experimental research.
Function 6: On-the-fly Active-Learning Training of Machine Learning Force Field
In addition, DigCat's theoretical modeling capabilities include a vast collection of stable structure models and transition state models related to catalysis, essential for constructing machine learning force fields. Users can quickly download these foundational structures by inputting material types, element types, and specifying whether to include energy, force, or pressure data. These models provide a solid data foundation for machine learning force field training, significantly accelerating DFT calculations in theoretical modeling.
The DigCat development team brings together researchers from academic institutions around the world. We sincerely appreciate your support and interest in DigCat, and we welcome any feedback or inquiries. Your input will help our developers continually improve and enhance the services provided by DigCat.
Project Leaders:
Prof. Hao Li
Dr. Di Zhang
Collaborators:
>100 collaborators from >20 universities worldwide. For the full collaborator list, refer to: www.digcat.org.
