Causal AI for Accelerated Battery Materials Development

点击翻译按钮获取中文摘要

The global chemicals industry stands at US$4.3T, covering a broad application range encapsulating consumer products, speciality chemicals, basic chemicals, pharmaceuticals, etc. The spending on research and development varies from 2% to up to 15% of revenue in companies in this industry.Development timelines and costs across verticals within the chemical industry are notoriously high. In agrochemicals, the lead time for a new protection product stands at over 11.3 years with required investment of $100-250 million. In battery materials, cycling data needed to assess each new material or formulation iteration can take up to months - contributing greatly to the time to market for new chemistries being a minimum of 5-10 years.A key reason for the long iteration cycles and labour intensiveness of industrial chemical research is the reliance on experimental research. Most chemical research and development - be it fundamental research for novel materials, qualification of new suppliers, quality and performance assessment during batch size scale-up for new or modified materials and processes, tweaking of existing platforms as per specific customer requirements - involves laboratory experimentation. This is due to well-understood theoretical chemical reactions being eclipsed by the effects of variations in material properties due to differences in production processes followed by suppliers, in testing and measurement techniques across the value chain and side reactions across complex multi-step production processes.Our goal is to integrate the disruptive technology of causal inference with artificial intelligence models to create a new form of Causal AI that is able to provide insight into the development of chemical products. We will specifically provide actionable insights into the process of battery development with the goal of validating our model in a cutting edge sector that has far reaching impacts on environmentally friendly energy storage. This model will then be extended to other commercial applications within the chemicals industry to accelerate their R&D timescales.