In short: A positive photoresist becomes more soluble where light hits it (the exposed pattern washes away), while a negative photoresist becomes less soluble where exposed (the exposed pattern cross-links and stays) — the difference is entirely a photochemical change in how the polymer reacts with developer.
Photoresist is the light-sensitive polymer film that turns an optical image into a physical pattern on a silicon wafer. Everything downstream — etch, implant, deposition — is gated by whether the resist dissolved or survived. Getting the chemistry right is the difference between a 10 nm contact landing on target and a shorted, scrapped lot. This article explains what actually happens inside positive and negative resists when photons arrive, and why the two types invert each other.
Why solubility is the whole game
Lithography never "cuts" anything directly with light. Instead, a thin resist layer (typically 30–500 nm for advanced nodes, spun on at 1,500–5,000 rpm) is exposed through a photomask, then immersed in a chemical developer — usually 2.38% tetramethylammonium hydroxide (TMAH), an aqueous base. The developer dissolves the regions whose solubility the exposure changed, leaving a 3-D relief image. Etch chemistry then attacks the now-bare substrate, and the surviving resist acts as a stencil.
The entire selectivity of lithography rests on a contrast in dissolution rate: unexposed vs. exposed resist can differ by 100× or more in how fast TMAH removes them. That ratio is what a good process window buys you.
Positive resist: light makes it dissolve
The classic g/i-line positive resist is a DNQ–novolac system. Novolac is a phenolic resin that is already mildly soluble in base. Blended into it is a diazonaphthoquinone (DNQ) dissolution inhibitor. In the dark, DNQ is hydrophobic and physically blocks the base from attacking the novolac, cutting the unexposed dissolution rate to near zero.
When UV (365 or 436 nm) strikes DNQ, it undergoes the Wolff rearrangement: the molecule loses nitrogen and, with trace water, converts into a carboxylic acid (indene carboxylic acid). That acid is highly base-soluble. So the exposed region flips from "inhibited" to "accelerated," and TMAH clears it rapidly. The mask image is reproduced one-to-one — clear mask areas become open resist.
At 193 nm and in EUV, the dominant chemistry is the chemically amplified resist (CAR). Here a photoacid generator (PAG) releases a strong acid (H⁺) on exposure. During the post-exposure bake (PEB, ~100–130 °C) that acid catalytically cleaves acid-labile protecting groups (e.g., t-BOC or acetal) off the polymer backbone, exposing carboxylic-acid or phenol groups and making the polymer base-soluble. Because one acid molecule deprotects many sites before being consumed, the quantum yield is "amplified" — essential when EUV delivers few photons per feature.
Negative resist: light makes it harden
Negative resists invert the logic. Exposure triggers cross-linking — the formation of covalent bridges between polymer chains that lock the exposed region into an insoluble network. Classic cyclized-rubber/bis-azide resists generate a nitrene on exposure that inserts into the rubber backbone, knitting chains together. The exposed zone becomes a dense 3-D mesh that the developer cannot penetrate, so the unexposed regions wash away.
Modern negative-tone options include negative CARs, where the photoacid catalyzes cross-linking (e.g., via a melamine cross-linker) during PEB instead of deprotection, and negative-tone development (NTD), a widely used 193i trick in which a normally positive-acting polymer is developed in an organic solvent (like n-butyl acetate) rather than TMAH. The deprotected, polar exposed region now resists the non-polar solvent, so it stays — giving negative-tone imaging of dark-field masks (contacts, trenches) with better aerial-image contrast than bright-field positive tone.
NTD is why many sub-28 nm contact and metal layers are patterned "negative": printing a small dark feature on a bright-field mask is physically easier than printing a small bright feature, so flipping the tone improves the process window directly.
The exposure-to-develop sequence, animated
Side-by-side comparison
| Property | Positive resist | Negative resist |
|---|---|---|
| Effect of exposure | Chain scission / deprotection → soluble | Cross-linking → insoluble |
| What remains | Unexposed areas | Exposed areas |
| Resolution | Historically finer (no swelling) | Older types swell; modern NTD competitive |
| Developer | Aqueous base (TMAH) | Solvent (classic) or TMAH (negative CAR) |
| Best mask tone | Bright-field (lines/spaces) | Dark-field (contacts/trenches via NTD) |
| Adhesion / etch resistance | Moderate | Generally strong |
Where this shows up in the design-to-silicon flow
Resist tone is not just a fab decision — it feeds back to layout and OPC. Resolution-enhancement and optical-proximity-correction tools (Siemens Calibre, Synopsys Proteus, and the lithography models in Synopsys Sentaurus/S-Litho) simulate the aerial image, the acid-diffusion blur, and the develop step to predict the final contour. Cadence Pegasus and Siemens Calibre run the DRC/LVS that assumes those contours. If you flip a layer to negative-tone development, the mask polarity, the sizing bias, and the SRAF (sub-resolution assist feature) placement all change. Engineers who understand the underlying resist chemistry debug "mystery" bridging and necking defects far faster than those who treat the resist as a black box.
For a deeper treatment of material families and vendors, see our companion guide on photoresist types explained, and the step-by-step optics in the photolithography process steps. If you are moving from understanding into building real flows, our partner Avecas VLSI design services runs production tapeout flows where these choices get made every day.
Want to go from "I know what a resist is" to running a full litho model and OPC deck? Our Nano Fab track on CourseTron walks you through exposure, PEB, develop, and defect debug with hands-on tool labs.
Frequently asked questions
Is positive or negative resist "better"?
Neither universally. Positive dominated for decades because it avoided swelling and gave fine lines. But negative-tone development now wins for dark-field layers (contacts, vias, trenches) because printing a small dark feature yields higher image contrast and a wider process window.
What does the post-exposure bake actually do?
In chemically amplified resists the PEB supplies thermal energy for the photoacid to diffuse and catalyze deprotection (positive) or cross-linking (negative). It sets the final solubility contrast — and because diffusion blurs the latent image, PEB temperature and time are tightly controlled to balance sensitivity against line-width roughness.
Why is TMAH the standard developer?
2.38% TMAH is a metal-ion-free aqueous base that cleanly dissolves the deprotected, acidic exposed polymer in positive/positive-CAR systems while leaving the protected polymer intact, and it is compatible with fab contamination limits (no mobile sodium/potassium ions that would poison transistors).
What causes line-width roughness (LWR)?
Photon shot noise (especially in EUV, where each feature sees relatively few photons), random PAG/quencher distribution, and stochastic acid diffusion during PEB. These statistical effects, not mask error, set the LWR floor at advanced nodes.