The rise of Self-Driving Labs: AI-Driven chemistry automation

Milad Abolhasani, Nature Synthesis, 2(6)

Unlike other industries where automation can be relatively straightforward, synthetic chemistry is highly dynamic and often requires a high level of dexterity, perception, and informed judgement. What could possibly be the potential of AI in areas with highly variable physical requirements?

Prototypes of ‘Self-Driving Labs’, which are automated laboratories that do not require human labour or supervision, have already been introduced to the world of chemistry in the past few years.  

These systems are estimated to improve research productivity ‘at least 30-fold’ by improving:

1.    Inefficient and slow experimental space exploration

2.    Physical disconnection between different experimental stages

3.    The time gap between performing an experiment and selecting the next experiment to be tested.

 Now, let’s address the fears of automation reducing the utility of chemists!

Robotic labs can free chemists from repetitive or dangerous lab experiments in order to conduct higher level thinking. This will improve the value of chemists, by increasing the quality and volume of publications they can produce.

Let us take an in-depth look at one of these automated futuristic devices coming to the rescue (or demise) of the synthetic chemist, the ‘Chemputer’ by Professor Lee Cronin. This revolutionary platform combines AI software with chemical hardware to conduct reactions, purification and even solvent evaporation!

While previous chemical automation removed some components of the manual labour process, it still essentially relied on the experimenter’s judgement and instructions. On the other hand, the Chemputer’s analytical AI software allows it to be fully independent in making operational decisions and taking over all manual tasks.

The system can optimise factors such as temperature, pressure, and reagent concentrations to conduct multi-step syntheses which would take a human weeks or months, in a fraction of the time!

An obvious disadvantage of these high-tech labs, however, is the cost of building a self-driving lab, which can range anywhere from US$1000 to US$1000000.

Fortunately, chemistry labs can profit from the lower end of this range, by leveraging open-source automation software such as ChemOS. This automation system can assist with characterisation and data analysis, as well as operate cheap robots such as syringe pumps and liquid handling bots.

The increasing popularity of “cloud labs” is another way that chemists can overcome the hurdle of automation costs. Cloud labs allow researchers to contract a self-driving lab on a pay-as-you go basis to conduct large scale synthesis or smaller scale experiments.

The chemputer does not stand alone! Here are other examples of successful self-driving lab systems in different fields of chemistry:

·       Artificial Chemist – quantum dot synthesis of nanomaterials

·       BEAR – Autonomous experimentation system that results in a 10X reduction in required experiments

·       CAMEO - active learning system for discovery of advanced materials

This is only the earliest stage of innovation, what do you think a synthetic chemist’s role will be in 20 years?

I hope you enjoyed reading, if you want an opportunity to work in this field Professor Milad Albohassani is hiring postdoc students! See you next week.

Valerio - Investing in Science