mRNA / LNP manufacturing equipment

The equipment does not assemble the nanoparticle; chemistry does. Lipids dissolve readily in ethanol, and the ionizable lipid carries a protonatable amine that is partly charged at formulation pH and binds mRNA electrostatically. Merge the ethanolic lipid stream with an aqueous mRNA stream and dilution does the rest: as ethanol concentration falls, the lipids’ solubility collapses and they precipitate — but onto a template, since the polyanionic mRNA nucleates the cationic lipid around itself. Given the right four components, the equilibrium endpoint is a vesicle of roughly a hundred nanometres. The hardware exists because the path to that endpoint, not the endpoint, decides the quality.

Mixing time versus nucleation

Particle size is decided by a race. When the streams first touch, local supersaturation of the lipids rises; nuclei form and then grow by eating the remaining dissolved lipid. If micromixing is fast — faster than nucleation and growth — every particle begins under essentially the same solvent composition and one synchronised burst of nuclei forms: narrow, small distribution. If mixing is slow, particles nucleated early keep growing in a continuously changing environment while new nuclei keep appearing elsewhere: broad sizes, large particles, aggregates. The classic burst-nucleation picture applies, and the controllable levers are the mixing geometry, the ratio of the two flow rates and the total flow rate. Impingement-jet mixers collide the streams head-on in a confined chamber, mixing by chaotic advection within milliseconds; staggered-herringbone microchannels do it by repeatedly folding the lamellae so interfacial area grows exponentially down the channel. In both cases the physics is laminar — viscosity and channel scale forbid turbulence at these rates — so mixing is engineered kinematics, not a stir bar.

What steers the assembly

Two formulation levers matter as much as the mixer. The fraction of PEG-lipid acts as a surface cap: it sterically stabilises the growing particle and cuts growth short, so more of it means smaller particles — which is why particle size responds strongly to composition as well as to flow conditions. The ionizable lipid’s pKa, tuned near endosomal pH, is what lets the particle form at formulation pH and release its payload later; the equipment must not disturb it, which limits shear-driven designs. Downstream, dilution quenches the process below the ethanol threshold so the particles stop restructuring, and tangential-flow filtration removes ethanol and exchanges buffer — but it can only tighten, not rewrite, the size distribution that mixing already fixed.

What breaks at scale

A stirred tank cannot do millisecond micromixing, so there is no scale-up in the classical sense — no bigger vessel doing the same job. Scale comes either from raising flow through the same geometry while holding residence time and flow-ratio constant, or from parallelising identical channels or jets, both of which preserve the local mixing history each fluid element experiences. The failure modes follow: deviate the flow-rate ratio and particle size moves with it; a fouled or eroded chamber changes the effective geometry silently; co-current dead zones nucleate a population of oversized particles that no downstream step removes. The unit operation is unforgiving precisely because it is fast — the entire quality of the batch is decided in the first few millimetres after the streams meet.