DARPA Is Using AI To Develop Better Batteries and Explosives

  • DARPA has developed machine learning-based tools to produce new molecules.
  • The program, named Make-It, can speed up the process of chemical discovery for several military products and applications.

One of the toughest challenges of modern-era organic chemistry is teaching computers how to plan chemical syntheses. Despite years of research, we haven’t been able to achieve complete pathways built by computers and then successfully implemented in the laboratory.

At present, computer software/hardware aren’t capable enough to handle a huge knowledge base of chemical transformation, efficiently navigate vast ‘trees’ of synthetic possibilities, and higher-order logic.

DARPA is currently working on a software tool based on expert-encoded and machine learning techniques to predict synthetic pathways optimized for parameters like time, cost, safety and waste minimization.

The program, named Make-It, aims to free researchers so that they can focus on chemical innovation, instead of testing numerous compound synthesis routes. The team is trying to build automated equipment (based on one-device many molecules concept) that generate desired compounds using their own knowledge base and trial-and-error process.

It can speed up the process of chemical discovery for several military products and applications. The good news is they’ve recently demonstrated a remarkable progress towards building this automated equipment.

It could enable a wide range of next-generation defense products, including smaller and long-lasting batteries, fuel cells, safer-to-handle explosives, effective propellants for rocket engines, better adhesives, paints and coatings.

How This is Going To Help Research Chemists?

Usually, a research chemist spends dozens of hours creating synthetic pathways to a new compound, and months executing and enhancing the synthesis in the laboratory. This new tool will help chemists to use their brain power in other valuable areas like molecular discovery.

Research chemists would be able to generate molecules as per their requirement instead of purchasing them from suppliers and disposing in bulk. The tool could be beneficial not only in the field of chemistry, but also many other technological areas that involve small molecule research and development.

Graphs of synthetic possibilities emerging from the target and expanding with the number. of search iterations

With non-automated methodologies, the small and negligible changes can leave a huge impact on throughput and purity of the compounds generated, which makes it difficult to reproduce prior-reported synthesis. Automated chemical synthesis, on the other hand, generates reproducible procedures that accelerates as well as democratizes the production.

Reference: ScienceDirect | doi: 10.1016/j.chempr.2018.02.002 | DARPA

According to the developers, Make-It ensures accurate reproducibility. Since the pathways are executed by computer instructions, it doesn’t result in lab-to-lab variability.

Make-It synthesis devices offer a cleaner, safer and secure solution. One doesn’t need to physically handle harmful chemicals like explosives. Moreover, these devices require low solvent volumes, producing lesser amounts of waste.

Other Similar Tools

Many other institutes and organizations are exploring different ways to develop tools that can automatically design numerous chemical pathway options. Grzybowski Scientific Inventions’ (GSI) molecular path optimization tool allows scientists to filter million of data points and find the best routes for drug discovery and medicinal chemistry. The tool is commercially available.

Researchers at the University of Glasgow are currently working inexpensive tools that can 3D print portable reactors to manufacturer particular compounds on the go. MIT is also applying expertise on machine learning models to aid drug discovery.

Automated chemical synthesizer | Credit: MIT

The above image shows MIT’s automated chemical synthesizer, in which chemicals flow into the vertically-aligned reactors. A robotic arm can insert, reorder or remove these chemical reactors to carry out computer-designed syntheses.

Purdue University has come up with a high-throughput screening process that uses data processing software, robotics, sensitive detectors and liquid handling devices to rapidly conduct million of genetic and pharmacological tests.

Read: TopoMS Can Accurately Analyze Chemical Bonding In Real-Time

What’s exciting about DARPA’s Make-IT is it’s aimed to generate any molecule, addressing not just pharmaceuticals but all defense fields. In the coming years, researchers will be using advanced analytical instrumentation to integrate real-time classification of a synthesis process and develop more complex compounds.

Written by
Varun Kumar

I am a professional technology and business research analyst with more than a decade of experience in the field. My main areas of expertise include software technologies, business strategies, competitive analysis, and staying up-to-date with market trends.

I hold a Master's degree in computer science from GGSIPU University. If you'd like to learn more about my latest projects and insights, please don't hesitate to reach out to me via email at [email protected].

View all articles
Leave a reply