MIT researchers reported in ACS Nano on Sept. 25, 2026, an automated version of a two-step mixing process for producing lipid nanoparticles (LNPs), the fatty particles used to package mRNA and other nucleic acid therapeutics. In the first step, mRNA and lipids are mixed at equal flow rates; after a delay, more buffer is added to stop particle growth. Longer delays yielded larger particles, and changing buffer concentration shifted particles from spheres to elongated shapes. The new system incorporates a commercial dynamic light scattering device that measures particle size as particles form and adjusts the delay to reach a specified size. Shape measurement still occurs outside the automated system.

The researchers used data from these experiments to train a machine-learning model that predicts input combinations for a particular size or shape. They filed for a patent and are working to commercialize the technology through BIZON Labs. Funding came from the U.S. Food and Drug Administration and the Koch Institute Support (core) Grant from the National Cancer Institute.