The cotton candy effect: why spin coaters make fibres, and how to stop them
Anyone who has spun PMMA, BCB, lift-off resists or thick epoxy resists has probably seen it. You lift the lid after a spin and find fine, wispy strings of dried polymer round the wafer edge and across the bowl and lid, looking for all the world like candyfloss. Some cling to the wafer edge or fold over onto the back. Others drift loose and land on the next substrate.
It’s known in the industry as the cotton candy effect. It’s a nuisance and it causes defects, and the good news is that it’s well understood and can be fixed.
What’s happening
In a normal spin, liquid flows to the edge of the substrate and leaves as droplets, which hit the bowl wall and drain away. The cotton candy effect starts when the liquid at the edge dries before it can break into droplets. Instead of a droplet, a thin jet of liquid is pulled off the edge by the spinning motion. The solvent evaporates from the jet in flight, and the jet dries into a solid thread.
Engineers make polymer nanofibres deliberately in exactly this way, a process known as rotary jet spinning, and the machines were inspired by candyfloss makers. Research on those machines explains what’s going on in a spin coater:
- The pull has to beat surface tension. A jet only forms when the outward pull on the liquid is strong enough to overcome the surface tension that would otherwise round it into a droplet.
- The jet must dry before it breaks up. A jet that stays wet long enough breaks into droplets. Fast evaporation freezes it into a thread.
- The polymer has to hold together. Long polymer chains tangled together make the liquid stretchy, so it pulls into a thread instead of snapping into drops. Above a certain concentration, which depends on the polymer’s molecular weight, the liquid stops forming droplets and starts forming fibres.
So the effect appears when three things coincide at the substrate edge: the material is stringy, the film there is drying quickly, and there’s enough speed to pull a jet out.

Which materials do it
Fibres are most common with materials that combine long polymer chains with fairly high solids content. They’re widely reported with:
- PMMA, particularly higher molecular weight grades
- BCB (benzocyclobutene) dielectrics
- PMGI and LOR lift-off underlayers
- thick epoxy resists such as SU-8
- some thicker positive resists and protective coatings
Thin, low-viscosity resists rarely cause it. Some resist makers have reduced the effect by offering lower molecular weight versions of the same polymer, which is worth asking your supplier about if your process allows it.
Why it matters
- Defects on the substrate. Loose fibres land on the film and leave lumps and streaks.
- Contamination further down the line. Strings that fold under the wafer edge stick to the softbake plate and then to every wafer after it. Fibres also build up in the bowl and lid, need clearing out regularly and can find their way into the exhaust duct.
How to stop it
Work through these in order. The first changes cost nothing, and most processes are fixed before the end of the list.
1. Slow the cast step. High spin speed and fast acceleration stir up the air at the edge and dry the material sooner. Lowering the speed and acceleration of the casting step is the first thing to try. If you need the same final thickness, consider a slightly more dilute grade.
2. Reduce the exhaust while the film forms. Strong exhaust pulls solvent vapour away from the edge and dries it early. Turning the facility exhaust down helps, but it affects every other process on the same duct and may break your site’s safety rules. Programmable exhaust is the better route. It holds the exhaust low during the cast step and opens it fully once the film is set, so the bowl is still cleared of vapour before the lid lifts.
3. Keep the bowl solvent-rich. Running the backside rinse, the edge bead removal nozzle or both during the first part of the spin-off fills the edge region with solvent vapour and keeps the material wet long enough to leave as droplets. Switch them on as the cast step begins, turn them off once the substrate is covered, then finish with the normal spin for your target thickness. One published recipe for a 4 µm resist runs a backside rinse for the first five seconds of a ten-second spin-off at 3,500 rpm. Check with the material supplier which solvent suits the rinse.
4. Keep the material fresh. Resist from a bottle that’s been opened many times loses solvent and grows more viscous, which makes it stringier. Close bottles promptly and decant small quantities for daily use.
5. Use a closed lid. A closed bowl keeps solvent vapour around the substrate and shields the edge from the cleanroom downflow.
6. Keep the bowl clean. Clear fibres out after each session so loose strands can’t fall onto the next substrate. Bowl liners make this quicker, and a splash ring catches material at the rim so less bounces back towards the wafer.
Getting the exhaust right
Lab reports point in both directions on exhaust. Too much exhaust dries the edge and causes fibres. Too little leaves fibres and vapour hanging in the bowl after the spin, ready to settle on the substrate as the lid opens. The answer is to keep the exhaust at the coater maker’s figure for the bowl, then adjust it by recipe step. Low exhaust while the film forms and full exhaust to clear the bowl afterwards deals with both problems.
How SiSTEM can help
Apogee spin coaters have closed bowls and recipe-controlled exhaust, and they take backside rinse, edge bead removal, bowl liners and splash rings. All of these are covered on our dispense and bowl accessories page. If you’re fighting fibres with a particular material, send us the material and your recipe, and we’ll work through it with you and CEE’s process team.