KTH's Scaffold Lift-Off Masks Pattern Conformal ALD Films on 3D Structures Down to 27 nm

By NineScrolls Team · 2026-07-29 · 5 min read · Industry

What KTH Actually Demonstrated

Researchers at KTH Royal Institute of Technology have published a method for patterning conformal thin-film coatings on complex three-dimensional structures, with demonstrated features down to 27 nm. The paper, "Lithographic patterning of conformal thin films on 3D structures using Scaffold-architected Lift-off masks," appeared in Nature Communications in July 2026 (Liu, Che, Maj et al., Nat Commun 17, 6201).

The technique — the authors call it Scaffold-Architected Lift-Off, or SALO — targets a gap that has quietly limited 3D nanofabrication: how to define a patterned thin film on a surface that is not flat. Its most striking claim is that it enables lift-off patterning of atomic-layer-deposited (ALD) conformal coatings, something the authors describe as "infeasible for conventional shadowing-based lift-off processes."

Why Lift-Off Breaks on 3D Structures

Lift-off is one of the two classic ways to pattern a thin film. You lay down a sacrificial mask, deposit your film over it, then dissolve the mask so the unwanted film "lifts off" and only the patterned material remains. It works because conventional lift-off relies on a discontinuity: the deposited film must not coat the mask sidewalls, so the solvent can reach underneath and release it.

That assumption collapses in two situations. First, on tall or re-entrant 3D topography, line-of-sight deposition shadows unevenly and the film wraps where it should not. Second — and more fundamentally — with a genuinely conformal process like ALD, the film coats everything uniformly, sidewalls included, sealing the mask so solvent can never get under it. That is precisely why ALD and lift-off have historically been considered incompatible.

How Scaffold-Architected Lift-Off Works

SALO replaces the flat resist stencil with a purpose-built 3D scaffold, written directly by two-photon polymerization. Because the mask is a designed three-dimensional architecture rather than a spun-on layer, it preserves release pathways even after a fully conformal film has been deposited over the entire structure — the solvent still has a route in, so the film lifts off cleanly.

The result is a patterning route that is, in the authors' words, agnostic to the deposition process used and that reaches sub-30 nm features on non-planar surfaces. In other words, the same additive flow can pattern evaporated metals, sputtered films, or conformal ALD dielectrics on the sidewalls and floors of 3D structures where subtractive patterning is awkward or damaging.

Why "Agnostic to the Deposition Process" Is the Real Headline

For a process engineer, the interesting part is not the resolution number — it is that the method decouples how you deposit from how you pattern. Historically, patterning a conformal film on 3D topography meant depositing everywhere and then removing material with a directional plasma. That subtractive route is powerful but constrained: directional etching is line-of-sight, so it struggles to clear film from deep sidewalls, and it exposes the underlying structure to ion bombardment.

The trade-offs between additive lift-off and subtractive plasma removal are exactly the decision a nanofab process owner has to make, and they map onto the etch fundamentals we cover in our reactive ion etching guide and, for the additive-versus-physical-removal question specifically, our comparison of reactive ion etching vs. ion milling. SALO does not replace plasma etch — it gives labs a second option for the specific class of problems where clearing a conformal film off complex 3D topography by etch is the bottleneck.

Where It Lands: MEMS, Photonics, Nanofabrication

The paper positions SALO for MEMS, photonics, and general nanofabrication — fields built on non-planar geometries: released membranes, waveguides, resonators, high-aspect-ratio pillars and trenches. These are research and specialty-device domains rather than high-volume logic, which is where the technique is most likely to be adopted first: in university cleanrooms and R&D lines that already run two-photon lithography and want to add conformal-film patterning without a subtractive etch step.

As always with an academic demonstration, throughput, mask-writing time, and material compatibility will decide how far it travels. But the core contribution is real and specific: it makes ALD films patternable by lift-off, and it does so on structures that flat-resist processes cannot handle.

NineScrolls Niche Angle

SALO is a patterning method, but it lives or dies on deposition and etch. The films it patterns still come from the same tools NineScrolls represents — ALD for conformal dielectrics and barriers, sputtering and evaporation for metals, PECVD for functional layers. A method that finally makes conformal ALD compatible with lift-off raises the value of having a capable conformal-deposition system in the lab, because a whole new class of patterned 3D devices becomes reachable.

It also sharpens, rather than removes, the case for plasma etch. Additive lift-off and subtractive etch are complementary: many 3D flows will still need a directional plasma step to open, clear, or profile features, and choosing between the two routes is a per-layer engineering call. Labs weighing that choice can start with our overview of the differences between PE-RIE and ICP-RIE plasma etching, then read our reactive ion etching vs. ion milling comparison to see where a physical, line-of-sight removal step still wins. For the etch fundamentals underneath all of it, our reactive ion etching guide is the starting point.