High-aspect-ratio photoresist patterning

1 · What aspect ratio means for lithography

Aspect ratio is the height of a patterned feature divided by its width — a 20 µm-tall wall 2 µm wide is 10:1. In lithography it is the single number that captures how demanding a thick, finely-patterned structure is: the taller and narrower the feature, the harder every process step becomes. The photolithography glossary defines the term alongside the depth-of-focus and exposure-latitude ideas that bound it.

High aspect ratio is not the same as thick. A thick film patterned with wide, gentle features is easy; a moderately thick film patterned with deep, narrow trenches is hard. What follows is about the second case — reaching tall, steep, well-resolved features and getting them to stand.

2 · Why high aspect ratio is hard

Three physical problems get worse as aspect ratio rises.

• —The exposure gradient through the film. Light is absorbed as it travels down through a thick resist, so the top receives more dose than the bottom. Over-expose to clear the base and the top over-develops and the sidewall bulges; under-expose to protect the top and the base under-clears and the feet flare. The window that exposes top and bottom acceptably narrows as the film gets thicker.
• —Development transport into deep trenches. Developer has to reach the bottom of a narrow trench, and dissolved resist has to get back out. In a deep, narrow feature that exchange is slow, so thick HAR resists need long, agitated develops — and the same transport limit is why an incompletely developed foot is a classic HAR failure.
• —Mechanical collapse and stiction. Tall, thin resist walls are fragile. Surface tension during the rinse-and-dry can pull adjacent walls together, and once they touch they stick — the pattern-collapse and stiction that cap how thin a wall of a given height can be.

3 · Resists that reach 10:1 and beyond

A handful of resists are formulated to reach 10:1 and beyond, and the aspect ratios below come from the published literature on each resist, not from a rule of thumb.

• —AZ 125 nXT — an acrylic negative resist Staab et al. demonstrated at 400, 800 and 1400 µm film thickness with a 20:1 aspect ratio for low-cost UV-LIGA, releasing freestanding electroformed metal at 13:1 and 16:1.
• —AZ 9260 — the founding thick-mould positive resist, for which Conédéra et al. reported an aspect ratio of 15–20 at 100 µm on a standard aligner.
• —KMPR 1050 — a thick epoxy negative resist Lee et al. used to make 180 µm, 18:1 electroforming moulds that, unlike SU-8, can be stripped afterwards.
• —SU-8 2025 — the canonical HAR/3D epoxy resist reviewed by del Campo & Greiner; the SU-8 family is the reference material for thick, high-aspect-ratio and 3D microstructures.
• —SUEX — an ultra-thick dry-film epoxy whose processing Johnson et al. optimised specifically for ultra-thick, high-aspect-ratio X-ray LIGA imaging.

A thin resist can also enable a high-aspect-ratio *structure* without being high-aspect-ratio itself. Gerlt et al. used a 1.4 µm S1813 film (with AZ nLOF 2070) as a Bosch deep-reactive-ion-etch mask and reached deep-silicon etch depths beyond 450 µm — here the aspect ratio lives in the etched silicon, and the resist's job is only to survive the etch.

Sources: Staab et al., J. Microelectromech. Syst. 2011; Conédéra et al., J. Micromech. Microeng. 1999; Lee et al., J. Micromech. Microeng. 2008; del Campo & Greiner, J. Micromech. Microeng. 2007; Johnson et al., Proc. SPIE 7972, 2011; Gerlt et al., Micromachines 2021

4 · The thickness trade-off and the process levers

The thickness-versus-feature trade-off. For a given resist there is a floor on the feature width you can hold at a given film thickness — push the film thicker and the smallest wall or trench you can resolve and keep standing grows with it. Read each resist's own published thickness range off its recipe page (SU-8 2025 at 21.7–79.2 µm, SUEX at 20–1000 µm, AZ 125 nXT at 18–120 µm) and treat the biggest published aspect ratio as an optimised result, not a starting point.

Several process levers move HAR results:

• —Bake gradients. Soft-bake and post-exposure-bake a thick film gently, with ramps and holds, so solvent leaves and cross-linking proceeds evenly top-to-bottom rather than skinning the surface; the soft-bake and post-exposure-bake guidelines cover the principle.
• —Development agitation and time. Deep trenches need long, agitated immersion develops so fresh developer reaches the base and dissolved resist clears — see the development guidelines.
• —Rehydration for thick DNQ resists. Thick DNQ–novolak positives such as AZ 9260 need to reabsorb water after a long soft-bake before they will expose and develop cleanly; hold them per the rehydration guidelines before exposure.

Inspection. Aspect ratio is a cross-sectional property, so verify it in cross-section — a cleaved edge or an angled SEM — not from a top-down image, which shows the width but not the wall profile, the foot, or the degree of collapse.

Sources: Conédéra et al., J. Micromech. Microeng. 1999

Common questions

Where to next

1. Staab et al.. Applications of Novel High-Aspect-Ratio Ultrathick UV Photoresist for Microelectroplating. Journal of Microelectromechanical Systems (2011). doi:10.1109/jmems.2011.2159098
2. Conédéra et al.. Potentialities of a new positive photoresist for the realization of thick moulds. Journal of Micromechanics and Microengineering (1999). doi:10.1088/0960-1317/9/2/317
3. Lee et al.. Fabrication of thick electroforming micro mould using a KMPR negative tone photoresist. Journal of Micromechanics and Microengineering (2008). doi:10.1088/0960-1317/18/5/055032
4. Gerlt et al.. Reduced Etch Lag and High Aspect Ratios by Deep Reactive Ion Etching (DRIE). Micromachines (2021). doi:10.3390/mi12050542
5. del Campo, Greiner. SU-8: a photoresist for high-aspect-ratio and 3D submicron lithography. Journal of Micromechanics and Microengineering (2007). doi:10.1088/0960-1317/17/6/R01
6. Johnson et al.. SUEX process optimization for ultra-thick high-aspect ratio LIGA imaging. Proc. SPIE 7972, Advances in Resist Materials and Processing Technology XXVIII (2011). doi:10.1117/12.882872