Why microfluidics behaves differently
Shrink a pipe and the physics does not scale with it. Some effects become negligible and others take over, and the crossover happens right around the sizes OpenChip is for.
If you have plumbed anything at bench scale, three intuitions are about to stop working.
Surface beats volume
Section titled “Surface beats volume”Volume scales with length cubed; surface area with length squared. Halve every dimension and you have one eighth the fluid touching one quarter the wall — so the surface-to-volume ratio doubles.
Anything that happens at a surface gets stronger relative to anything that happens in the bulk:
- Viscous drag dominates inertia. This is the big one, and it is what the Reynolds number measures.
- Surface tension becomes a structural force rather than a curiosity. It is what holds a droplet together in a droplet generator, and it is why an air bubble in a 0.5 mm channel is a plug you have to push rather than something that floats past.
- Heat crosses walls easily. A microchannel reaches thermal equilibrium with its surroundings fast, which is what makes on-chip heat exchange work at all.
- Adsorption matters. A protein sticking to the wall removes a much larger fraction of what you put in.
There is no turbulence
Section titled “There is no turbulence”At the flow rates and channel sizes used here, the Reynolds number is typically well under 1 — often two or three orders of magnitude below the ~2000 where turbulence begins.
The consequence is not “less mixing”. It is no mixing except by diffusion.
Two streams meeting in a microchannel run side by side down the channel as distinct laminae, with a sharp interface between them, and cross only by molecular motion. Nothing stirs. Making the channel wider or the chamber bigger does not help, because there is no mechanism to help.
This is the single most surprising thing about the field for anyone arriving from ordinary fluid handling, and every mixer geometry in the template library exists because of it. See Mixing without turbulence.
Pressure rises fast
Section titled “Pressure rises fast”Hydraulic resistance scales roughly with the fourth power of the hydraulic diameter. Halve the channel width and the pressure for the same flow goes up about sixteen-fold.
Practical consequences:
- A syringe pump that easily drives a 1 mm channel may stall on a 0.25 mm one.
- Branch balance is fragile. In a gradient generator, a branch printed 10 % wider than its neighbours carries about 46 % more flow. That is why the template’s caveat says to check the flow column after any edit.
- Print tolerance becomes a flow variable. A channel 0.05 mm wider than drawn is a channel with meaningfully less resistance.
See Pressure drop and Hagen–Poiseuille.
Diffusion is slow, but the distances are short
Section titled “Diffusion is slow, but the distances are short”Diffusion time scales with the square of the distance to be crossed.
Across a metre it is hopeless. Across 0.5 mm it takes seconds to minutes for a small molecule — comparable to the residence time in a chip, which is the whole reason microfluidic mixing works at all.
The squaring cuts both ways: halving the channel width cuts the mixing length by four, which is often a better move than making the channel four times longer.
What this means when you are designing
Section titled “What this means when you are designing”| Intuition from bench plumbing | What is true here |
|---|---|
| Turbulence will mix it | Nothing mixes unless the geometry makes it |
| A bigger chamber mixes better | A bigger chamber is a delay line |
| A slightly narrower channel is fine | Fourth-power resistance says otherwise |
| Bubbles rise out | Bubbles form plugs and stay |
| Branches split evenly | Only if their resistances match |
- Laminar flow and the Reynolds number — the number behind all of this.
- Pressure drop and Hagen–Poiseuille
- Mixing without turbulence