Searching for Unknown Compounds to Achieve Carbon Neutrality

Co-create with Lumada! Use our proprietary AI to analyze publicly available data

Efficiently search for unknown compounds that have specific desired properties

Currently, research and development (R&D) of materials can take some 10 years from the time a new compound is discovered until the time it is eventually commercialized as a product available on the market.

R&D relying on experience and intuition

Aiming to enhance product performance, many manufacturers pursue long-term materials development. They also face a new challenge. Since the adoption of SDGs in 2015, there have been calls to develop a stronger awareness of the need to prioritize environmental considerations in order to achieve carbon neutrality (a state in which greenhouse emissions are essentially zero).

To develop products that can satisfy practical demands while also being able to easily return to nature, researchers need to seek out compounds that enhance the utility of existing materials and to identify new combinations of compounds used to create new materials. Currently, this research still largely relies on the knowledge, experience, and intuition of the researchers.

Once selected, a compound needs to be synthesized and evaluated. This process is repeated until the researchers find the results they are looking for. If successful evaluation results are obtained, the researchers must conduct checks to prevent potential rights infringements. If a potential infringement is found, the researchers must restart the entire procedure from the compound-selection phase. This repetitious manual labor eats up a lot of the time and increases the cost of R&D of materials.

Utilize vast amounts of information for strategic R&D

In the field of materials development, materials informatics (MI) has recently garnered attention for its potential to use AI and big data to streamline R&D efforts.
The Hitachi product Chemicals Informatics^* (CI) incorporates MI methods. The use of CI in the upstream R&D process of compound selection makes it possible to search vast amounts of information regarding past experiments, research papers, and patent data. Synthesizing and evaluating are performed for selected compounds only, which greatly shortens the time frame and reduces the costs required for R&D.

For example, biodegradable plastics are environmentally friendly plastics that microorganisms in nature return to nature by decomposing them into water and carbon dioxide. The use of such plastics is currently limited to non-reusable items, such as fresh food trays and instant food containers. However, if R&D led by CI finds robust, biodegradable materials, in the near future we could produce low cost products with the required level of durability. As the use of practical biodegradable plastic products catches on, we could reduce the consumption of petroleum, the raw material going into plastics, by reusing plastics and cutting down on waste. Microorganisms can even decompose waste that is not collected because of littering or other reasons, thereby reducing their adverse impact on the environment.

Accelerating the new-product development cycle in the chemical field

Chemicals Informatics—Using AI to discover compounds

Intensifying development competition and technological advances in the chemical industry leads to the expectation that, for a company to remain competitive in this field, the company must be the first to market with a high-value-added product and must capture a high market share.

Ushering in significant changes in materials R&D

To incorporate materials informatics (MI) into materials R&D, vast amounts of data used to explore compounds needs to be prepared in a format that can be analyzed by AI.
To achieve this, the Hitachi product Chemicals Informatics (CI) relies on publicly available data, such as patents and research papers. Hitachi’s proprietary NLP AI (natural language processing AI) reads an enormous amount of publicly available data of a highly technical nature, formats it into data suitable for retrieval by the Exploratory AI, and stores it in a compound database. This eliminates the need for the user to prepare data, allowing the user to search and select compounds by applying MI methods immediately after CI is implemented.

The Novel Compound Generation AI, which uses proprietary software programs, generates information on compounds that have novel structures and also stores this information in the compound database. The search results of the Exploratory AI includes compounds that have novel structures, thereby increasing the likelihood of finding promising compounds.

• Chemicals Informatics

Reducing the time and cost required for materials R&D

The Hitachi product Chemicals Informatics (CI) aids strategic R&D efforts that aim at finding unpatented compound combinations that have specific, desired properties. This comprehensive approach reduces rework, while also shortening the time to market from the planning stage.

• Selecting compounds from vast amounts of data
  An exhaustive search is performed of data on 100 million known compounds, 30 million patent documents, and over 11 million novel compounds generated by the Novel Compound Generation AI. This provides the ability to select combinations of existing compounds, novel compounds, and unpatented compounds that have specific, desired properties.
• Synthesizing and evaluating based on past data
  Past data can be used to reference synthesis methods described in patent documents, and the property values can be used as reference data for evaluations. This narrows down the scope of promising compounds that have the specific, desired properties, effectively reducing the number of experiments (synthesis and evaluation) required.
• R&D that avoids rights infringements
  The ability to search for unpatented compound combinations during the compound selection phase facilitates R&D activities with a low potential of rights infringements.

CI discovers an additive that improves plastic strength and degradation performance

The following actual case shows the efficient and comprehensive use of the Hitachi product Chemicals Informatics (CI) to discover new additives that improve biodegradable plastic strength and more readily degrade in natural environments. This case study was jointly conducted by Hitachi, Ltd., Hitachi High-Tech Corporation, and Hitachi High-Tech Solutions Corporation. The results were as follows.

Conditions

CI was applied to search for polylactic acid, widely used as a biodegradable plastic, and 14 organic materials that were potential additive candidates.

Results

CI found two additives that improved mechanical strength (yield strength) and absorbency (metric used to indicate the ease at which hydrolysis occurs to speed up biodegradation). Molecular dynamics simulations and prototype evaluations were performed to verify validity.

Benefits of implementing CI

The use of publicly available data and proprietary AI technologies were confirmed to have the following benefits compared to before CI was implemented:

• Shortened development times
  The development cycle (including the selection of material candidate compounds, synthesis, evaluation, and rights infringement checks) was reduced to around 1/24th of previous efforts.
• Reduced experiment waste materials
  The number of performed experiments was cut to 1/5th of previous efforts due to improvements in the experiment success rate. This also reduced the amount of experiment waste materials to around 1/5th of previous efforts.
• Increased comprehensiveness of exploratory activities
  The number of searched candidate material compounds was 580 times the number covered when selecting candidate compounds based on researcher knowledge alone.

Future prospects for CI

Earth-friendly materials are needed to resolve social problems such as resource depletion and environmental destruction. The Hitachi product Chemicals Informatics (CI) draws upon vast quantities of data stored in a compound database to aid materials development across a broad range of fields, including plastics, electronic components, batteries, and optical materials. CI contributes to the development of materials that strike a balance between environmental considerations and product performance improvements.

The compound database is updated around once every six months to expand the patent information and other publicly available data, so that customers can use the latest available information to search for compounds. As of August 2022, the database encompasses 61 properties and 119 fields, which are continually being added to, including additions made in response to customer requests. Looking ahead, we hope to continue to expand the scope of CI in our aim to incorporate use across a broad range of applications.

For details on our solutions, see the following webpages.

For details on the evaluation of research related to utilizing CI, see the following webpage.