Maximizing Yield and Conversion in Chemical Synthesis with IBM Research Europe
Achieving joint yield and conversion rates above 80% in the iodination of terminal alkynes.
Overview
Atinary Technologies and IBM Research Europe collaborated to enhance chemical reaction optimization, addressing the inefficiencies of traditional methods that rely on time-consuming trial and error. By combining Atinary’s no-code AI platform, SDLabs, and Self-Driving Labs technology with IBM’s RoboRXN (Chemspeed robotics), this partnership creates an autonomous workflow that efficiently navigates the large parameter space that is composed of more than 12,000 combinations, accelerating the optimization of chemical synthesis, specifically in the iodination of terminal alkynes.
Graphical abstract from Schilter et al. illustrating the integration of Atinary’s AI platform and Self-driving labs technology, SDLabs, with IBM’s RoboRXN robot.
Challenge
Traditional methods for optimizing chemical reactions are often slow and resource-intensive. Researchers typically rely on trial and error, which can lead to inconsistent results and extended timelines. For complex chemical reactions like the iodination of terminal alkynes, these issues are even more pronounced, making it difficult to achieve reliable outcomes quickly.
Use Case
IBM Research Europe used Atinary’s SDLabs platform for a fully integrated AI, machine learning, and robotics workflow to close the loop in the optimization of iodination of terminal alkynes. This reaction, known for its complexity, was previously a bottleneck due to the extensive experimentation required – a chemical reaction with 2 synthesis routes and multiple optimization parameters to reach the objectives of maximum yield and high conversion rate. By combining their platforms, Atinary and IBM Research Europe demonstrated the power of integrating AI, automation, and robotics to run a data-driven approach that accelerates the speed and efficiency of chemical reaction optimization, produces high yield, and maximizes conversion rates.
By the numbers
| 8 parameters | 2 synthesis routes | 12K+ possible combinations |
Achieved joint yield and conversion rates of over 80% for all four substrates in just 23 experiments, covering ca. 0.2% of the combinatorial space.
Outcome
The integration of ML-driven experiment planning, automated equipment, and interpretability tools demonstrate the potential for expedited reaction condition optimization, allowing for a cleaner and more sustainable industry.
Resources
- Link to Press Release on Atinary-IBM Research Europe Partnership
- Link to Publication: Schilter et al. Combining Bayesian optimization and automation to simultaneously optimize reaction conditions and routes. Chem Sci. 2024 https://doi.org/10.1039/D3SC05607D
- Link to Atinary Webinar
- More about IBM Research Europe in Zurich
- Download the use case as a pdf: brief overview, in detail
“The widespread use of AI and cloud technology will alter the way chemistry is conducted on a global scale and put chemists on the fast track to discovering new materials.
Atinary’s SDLabs and IBM’S RoboRXN are two pioneering technologies with the goal of embodying chemistry in the cloud and conducting real-world impactful research anywhere there is an internet connection.
The synergy of the two complementary capabilities will demonstrate what two innovative groups can do with new technological paradigms.”
Dr. Teodoro Laino, Distinguished Research Scientist of IBM Research
