Scientists watched a glowing film survive chip lithography

Chips are made with light and chemistry. OLEDs are made with metal vapor through a mask or with ink droplets. For a long time those two factories did not want to talk to each other. In Nature, the team of Chih-Jen Shih and Yinyin Bao at ETH Zürich shows a material that bridges both sides: a photoresist that glows. It can be drawn with ultraviolet and electron-beam lithography like a semiconductor layer, then lit as an OLED pixel.

Anatomy of the problem

A microdisplay for augmented-reality glasses wants pixels that are small, dense and multicolour. On-chip photonics wants a light source right next to silicon electronics. A classic OLED can emit efficiently. Mapping it onto a CMOS grid is harder.

Evaporation through a fine metal mask has limits on resolution and layer overlay. Inkjet also fails to catch up with silicon lithography. When organics are forced into a standard photolithography flow, solvents and developers often quench emission or destroy the layer. Early “electroluminescent photoresists” from the early 2000s typically delivered external quantum efficiencies below 2%. They glowed weakly and did not convince display engineers.

Semiconductor engineering has a different discipline: pattern, develop, next layer, overlay. OLED needs the same discipline if it is to enter monolithic CMOS integration instead of remaining a separate island at the end of the line.

Core, shell, arms

The authors call their system an ELPR — electroluminescent photoresist. The anatomy is deliberate.

At the centre sits a TADF emitter — a molecule that collects energy and returns it as light, including from triplet states. Each polymer chain grows from a single such core. Synthesis uses atom-transfer radical polymerization (ATRP): polymer arms grow outward from the emitter.

Around the core they build a host shell. It is a protective coat. At the very edge sit groups that crosslink with neighbours under UV. Crosslinking chemistry happens at the periphery. The emissive core stays separated from the reactive bustle. That is why after exposure and development the material still glows strongly, instead of turning into a dull network full of quenching traps.

One metaphor and one mechanism: the shell protects the light, and lithography bites only the skin.

Colours come from three families: red, green and blue. Different arm counts and linkers tune chromaticity. Solvent orthogonality lets them stack colours one after another — without washing away the previous pixel.

What the microscope shows

The paper’s numbers are concrete. Fluorescent nanopatterns reach a critical dimension of 110 nm. Average line pitch sits in the same range. Emissive disks run from 150 to 600 nm in diameter. That is no longer “thick” inkjet.

In ultraviolet they show line/space grids in the micrometre range (including L/S 4/2 µm) and sequential colour stacking thanks to resistance to developers in later steps. The device demo builds an RGB array with 30 µm pixels on a shared ITO anode, with pixels defined by a photolithographic SiO₂ mask.

After photocrosslinking, key ELPRs reach a maximum external quantum efficiency of 13.3%. That is a jump versus historic ELPRs below 2%. Intermediate precursor characteristics show even higher peaks — but for a product narrative we keep 13.3% as the hard result after patterning.

Do not stretch the “Moore’s law” metaphor. The authors use it in the abstract as a direction: lithographic scaling of organic optoelectronics. It is a roadmap, not a fab report from Hsinchu.

Who needs this outside the lab

The science layer is hard: ATRP, TADF, process orthogonality, EQE, SEM/AFM, RGB array. The applications layer in the Nature text and ETH framing runs on three tracks — all as potential.

First: AR/VR microdisplays. The smaller the pixel without losing efficiency, the denser the image in glasses. Organics do not need III–V epitaxy and handle dimension scaling better regarding sidewall defects — if the process can draw them at all.

Second: CMOS integration. Back-end-of-line likes low temperature and no lattice mismatch. OLED already promised that. The lithographic bridge was missing. ELPR is a candidate for that bridge, not a finished fab design kit.

Third: on-chip photonics and fast optical links. That horizon is farther. First you need stable, repeatable RGB microsources on a shared platform.

The fact boundary stays sharp. Nobody put these polymers into serial glasses tomorrow. From this one paper we still lack full lifetime, yield on a 300 mm wafer, or monomer cost at tonne scale. We have proof that you can draw a glowing organic layer with chip tools and still get a double-digit EQE.

Where 13.3% comes from

Historic ELPRs suffered two problems at once. First: early fluorescent polymers harvested only singlets. Second: emitter and photoacid sat mixed, so crosslinking made quenching traps.

TADF fixes the first deficit: triplets become usable too. Core–shell architecture fixes the second: UV crosslinking does not happen next to the emitter, only at the edge. That is why after UV and develop EQE does not collapse to a fraction of a percent.

Figures also show that P-series precursors can show very high EQE peaks before a full “production” stack. That matters for polymer chemists. For a TRL09 reader the key number is after lithographic patterning: 13.3%. That number says whether the material survived lithography as a device, not only as a fluorescence cuvette.

Display engineer checklist:

After develop, do luminance and EQE stay in the same order of magnitude?
Can a second and third colour be added without dissolving the first?
Does critical dimension go below the metal-mask limit?
Are anode and pixel definition (here: ITO + SiO₂) compatible with the ELPR process?

ETH answers these questions with lab data. The fab answers are still missing: lifetime at microdisplay brightness, uniformity on a large wafer, feedstock cost.

What is not in this anatomy

ELPR does not remove every OLED barrier. Blue-emitter stability, heat in a dense array, encapsulation, the CMOS driver — those are separate chapters. The paper does not claim organic lithography will replace Korean or Chinese fabs tomorrow.

It also does not claim that 110 nm is already a consumer pixel in glasses. That is an emissive-pattern limit in the experiment. The device demo sits at 30 µm. Between the nanopattern and a production microdisplay lies process engineering, not one paper.

For TRL09 that is good news. The anatomy is honest: you can see which part protects light and which part serves drawing. The reader does not get a magical “new OLED”, but a material cross-section.

Europe reads lithography, not a sales folder

The lead is Swiss–European: ETH Zürich (chemical and bioengineering plus IT/EE), University of Helsinki, with NTUST in Taipei and RMIT in Melbourne. Correspondence: Bao and Shih.

For a reader in Poland and the EU the angle is not “a Polish patent”. It is a European display-materials track that wants to enter semi discipline instead of only buying finished stacks from Asia. Poland has groups in organic electronics and photonics; this paper is not their spin-off. It shows a checklist for later grants: process orthogonality, EQE after patterning, RGB on one anode, critical dimension.

Poland does not need to copy ETH point for point. It can read this anatomy as a brief for its own labs: if you build an organic emitter, design room for a crosslinking group and a host up front, or accept the metal mask. That is an architecture choice, not a slogan.

Anatomy that holds

Back to the cross-section. Emitter in the middle. Host around it. Crosslinker at the rim. UV or e-beam draws the pattern. Developer washes the unexposed. Next colour. Bias. Light.

When someone promises “OLED like a chip”, ask three things. Did the emitter survive develop? Did the second colour eat the first? Is EQE after crosslinking still double-digit? ETH answers yes in lab conditions: 110 nm, 30 µm RGB, 13.3% EQE.

TRL09 sets this at TRL about 3–4. Device demo and nanopatterning exist. Factory and product lifetime do not yet. The material anatomy is clear enough that display and semi engineers can talk in a shared language: lithography, overlay, emissive layer.

Anatomy verdict

The shell protects the core. Lithography bites the skin. RGB stacks layer by layer. Nature is not selling finished glasses. It shows how to draw a glowing organic pixel with chip-line tools — and not quench it on the way.

For an investor and a deep-tech desk the gap between metaphor and product reduces to a checklist. Is there EQE after patterning? Is there colour orthogonality? Does critical dimension fall below the metal-mask limit? This paper passes three lab tests. The fourth — the fab — stays open. And that is fine: TRL09 prefers material anatomy to a sales folder.

Sources

Lo S.-W., Shivarudraiah S.B., Lin Z.-H. et al. Electroluminescent photoresists extending lithographic scaling to OLEDs. Nature (2026). DOI: 10.1038/s41586-026-11042-0 (CC BY 4.0).
Tech Xplore / ETH Zurich framing — Tiny OLED pixels can be manufactured like microchips using chemical-resistant polymers (23.09.2026): https://techxplore.com/news/2026-09-tiny-oled-pixels-microchips-chemical.html
Crossref / OpenAlex / Unpaywall — metadata, OA licence, affiliations.

Narration mode: Anatomia. TRL ~3–4. Working categories: Deep Tech, Hardware, Semiconductors.