Droplet generation

A droplet generator turns two immiscible streams into a train of uniform droplets — one phase dispersed as discrete volumes inside the other. Each droplet becomes an independent reaction vessel of a few nanolitres, which is the basis of digital PCR, single-cell assays, and a lot of directed-evolution work.

The mechanism

At a flow-focusing junction, the dispersed phase enters from one side and the continuous phase from two others, squeezing it. The combined flow is forced through a narrow orifice.

At the throat, two forces compete:

Interfacial tension holds the thread together. It resists any increase in surface area, and pinching costs surface area.
Viscous shear from the continuous phase stretches the thread and pulls it apart.

Where shear wins, the neck thins, becomes unstable, and pinches off. One droplet. The thread recovers and the cycle repeats — at kilohertz rates, with remarkably uniform volumes.

The relevant dimensionless group is the capillary number, Ca = μv/γ, the ratio of viscous to interfacial forces. Low Ca gives a squeezing regime dominated by the geometry; higher Ca gives a dripping and then a jetting regime.

What sets droplet size

Lever	Effect
Orifice width	The dominant one. Droplet diameter scales with it.
Flow-rate ratio (continuous : dispersed)	More continuous phase, smaller droplets. The easiest knob to turn once printed.
Viscosity ratio	Changes where the regime boundaries sit.
Interfacial tension	Lower tension, smaller droplets. This is what surfactant is for.

Orifice width and flow-rate ratio are the first two things to tune, which is why the droplet generator template builds its throat from dimension markers — you can drag it wider or narrower and watch the neck reflow in 3D.

Practical requirements

Wetting matters more than anything else. The continuous phase must wet the channel walls; the dispersed phase must not. If the dispersed phase wets the wall, it smears along it and you get no droplets at all — just two streams. For water-in-oil this usually means a hydrophobic surface, which cured resin normally is. For oil-in-water it usually means a treatment.

Surfactant. Without it, droplets coalesce on contact. Fluorosurfactants for fluorinated oils, Span/Tween for hydrocarbon oils.

Stable, pulseless flow. Droplet uniformity follows flow uniformity directly. A syringe pump’s stepper ripple shows up as a periodic variation in droplet size.

Clean, sharp geometry. The throat does the work. A printed throat with a rough or rounded edge produces a wider size distribution than a moulded one.

What OpenChip does not model

Be clear about this before designing around a number:

No interfacial tension anywhere in the model.
No multiphase flow. The solver assumes one fluid filling the channel.
No droplet formation, size, or frequency. There is no prediction here at all.
No wetting or contact angle.
No surfactant.

What the Fluid Dynamics tab gives you for a droplet generator is the single-phase behaviour: pressure drop, velocity, shear rate at the throat, residence time. Those are useful — the shear at the throat is what drives the pinch-off, and knowing the pressure tells you whether your pump can deliver the flow — but they are not a droplet prediction.

Sizing droplets is a bench exercise. Print it, run it, measure it, adjust the flow-rate ratio.

Printing one

The throat is the hard part.

The template’s is 0.35 mm, which is near the limit of a standard 4K printer (0.2 mm nominal minimum feature). It will be the first thing the warnings panel flags on a coarse profile, and that warning is correct.

Print a test coupon with the throat alone at several widths before committing a chip. See A first test-print protocol.

Surface finish matters more here than anywhere else on a chip: layer lines at the orifice edge perturb the pinch-off and widen the size distribution. Orienting the chip so the throat sits in-plane rather than across layers helps — see Orientation and supports.