Microfluidic fabrication
A photoresist film is spin-coated onto a wafer to the channel height, baked, exposed to UV through a mask of the layout and developed. For microfluidic masters a thick negative epoxy resist such as SU-8 is usual, so exposed regions crosslink and remain as ridges.
Our ruleSimulations prepare you for the bench but don't replace hands-on practice.
Photolithography
A photoresist film is spin-coated onto a wafer to the channel height, baked, exposed to UV through a mask of the layout and developed. For microfluidic masters a thick negative epoxy resist such as SU-8 is usual, so exposed regions crosslink and remain as ridges.
How it differs from the standard method
- Spin speed sets thickness; dose and development time must scale with it.
- Film masks are cheap and fine to about 10–20 µm features; chrome masks for finer work.
- Multiple layers of different heights need alignment marks and a fresh bake cycle for each layer.
1Making channels the width of a hair
Most academic microfluidic chips are made in three steps. First a master mould is made by photolithography: a light-sensitive resist is spread on a silicon wafer, exposed to UV through a mask carrying the channel layout, and developed so that the exposed resist stays behind as raised ridges.
Photolithography
Thickness set by spin speed, pattern set by mask and dose. The master decides every dimension of every chip made from it.
PDMS soft lithography
Cheap, transparent, gas-permeable and flexible. Cure, mixing ratio and design aspect ratio decide whether it copies the master faithfully.
Plasma bonding
Activated PDMS and glass bond irreversibly within seconds of contact, but only while the surfaces are still reactive.
2What sets the dimensions
Channel height is the resist thickness, which falls as spin speed rises; channel width is the mask feature, broadened or narrowed by exposure. Thick resists absorb UV on the way down, so the dose has to rise with thickness or the base of each feature stays uncrosslinked, is undercut by the developer and lifts off.
3Why the flow is so well behaved
At these sizes the Reynolds number is far below one hundred, so flow is laminar: streams run side by side and mix only by diffusion, and the pressure needed to drive a flow follows the Hagen–Poiseuille law. For a rectangular channel the hydraulic resistance scales as length divided by width times height cubed, so halving the height needs eight times the pressure for the same flow. That calculation tells you how strong the bond must be before you build anything.
4How chips fail
- Features that float away in the developer: under-exposure for the thickness, or a wafer that was not dehydrated.
- Wide, rounded lines and closed gaps: over-exposure, an edge bead holding the mask off, or an under-baked resist sticking to the mask.
- Cracks in the master: fast temperature changes during the bakes.
- Tacky, torn replicas: under-cured PDMS, too little curing agent, or a master that was not silanised.
- Bubbles in channel walls: PDMS poured without degassing.
- Leaks at the inlet: too long between plasma and contact, dust on the faces, or over-treated, cracked PDMS.
Key terms
Practise these as flashcards →The wafer carrying raised resist features from which PDMS replicas are cast.
An epoxy-based negative photoresist that can be spun tens to hundreds of µm thick; exposed regions crosslink and remain.
The thicker rim of resist at the wafer edge after spinning, which holds the mask off the film unless removed.
The bake that completes acid-catalysed crosslinking in exposed SU-8 before development.
Poly(dimethylsiloxane), the silicone elastomer used for most research chips; usually mixed 10 : 1 base to curing agent.
Coating the master with a fluorosilane so cured PDMS releases cleanly.
Height to width of a feature, or roof span to height of a chamber; extremes collapse or deform.
Pressure drop per unit flow of a channel; for a shallow rectangle about 12μL/(wh³).
The inlet pressure at which a bonded chip delaminates or leaks; the standard test of a bond.
When to use it
- Building custom channel geometries for cell culture, gradients, droplet generation or assays when no commercial chip fits.
- Prototyping quickly: a new design can go from mask to sealed chip in a day or two.
- When optical transparency and gas permeability (for live cells) matter more than chemical resistance.
Compared with alternatives
Uses chips made this way to culture cells under flow; choose it when the device is a means to a biological model.
A downstream application whose junctions need hydrophobic channels and tightly controlled dimensions.
The surface chemistry behind plasma bonding and the coatings that make PDMS wettable or cell-adhesive.
When the structure must be a 3D porous material rather than enclosed channels.
Sub-techniques
Choosing one here switches the whole page, including the protocol and the lab, to it.
A photoresist film is spin-coated onto a wafer to the channel height, baked, exposed to UV through a mask of the layout and developed. For microfluidic masters a thick negative epoxy resist such as SU-8 is usual, so exposed regions crosslink and remain as ridges.
- Spin speed sets thickness; dose and development time must scale with it.
- Film masks are cheap and fine to about 10–20 µm features; chrome masks for finer work.
- Multiple layers of different heights need alignment marks and a fresh bake cycle for each layer.