This is a top-down cross-section field of the same microfluidic mixer: the horizontal axis is distance down the channel (inlet → outlet), the vertical axis is position across the channel width. Colour encodes the local lipid/aqueous concentration, solved from a genuine advection–diffusion field — not a flattened view of a 3D particle flow.
Re = ρ·v·D_h / μ (Reynolds number, same channel geometry)
∂c/∂t = D_eff · ∂²c/∂x² (cross-channel diffusion, marched downstream)
c → fold(c) at each ridge pair (stretch-and-fold groove recirculation)
Mix = 1 − Var(c_outlet) / Var(c_inlet) (mixing completeness from the field itself)
Straight channels only mix by slow molecular diffusion across the lipid/aqueous interface, so this model runs plain diffusion the whole channel length when the grooves are off. With staggered-herringbone grooves on, the field is periodically remapped by a doubling ("baker's") map at each ridge pair — the same stretch-and-fold action the ridges perform physically — which halves the striation width every pass so diffusion needs far less distance to homogenize it. Flow rate and flow-ratio additionally scale an effective diffusivity (a Taylor-dispersion-style enhancement: faster shear across the channel width speeds up cross-stream homogenization), so the mixing completeness read off the actual field responds live to every slider — independently of the 3D version's closed-form mixing curve.
- Total flow rate — raises Re and the shear-driven effective diffusivity → faster field homogenization → smaller, more uniform particles.
- Flow-rate ratio (aqueous:organic) — more aqueous buffer both dilutes ethanol faster and strengthens the diffusive enhancement, trapping mRNA more efficiently.
- Ionizable-lipid concentration — sets nucleation density; independent of the mixing field, it slightly enlarges particles for the same mixing quality.
- Herringbone grooves toggle — removes the stretch-and-fold remapping entirely, leaving only diffusion (matching the "diffusion alone is too slow" behaviour of a straight channel).
Downstream diameter/PDI/encapsulation-efficiency trends use the same simplified representative curves referenced by the 3D version (calibrated to realistic clinical LNP sizes, ~40–150 nm) — driven here by the mixing completeness computed independently from this field, not reused from the 3D model.