14.1 Photoresists and lithography

Photoresists and Lithography Fundamentals

Photolithography uses light to transfer geometric patterns onto a substrate, forming the backbone of modern microfabrication. Every microchip, sensor, and display screen you interact with was built through repeated cycles of this process. The photochemistry at its core determines how small, how precise, and how reliable those patterns can be.

Principles of photolithography

The photolithography process follows a strict sequence. Each step builds on the last, so errors early on cascade through the entire pattern transfer.

1. Substrate preparation — the surface is cleaned and primed to ensure uniform coating
2. Photoresist application — a thin film of light-sensitive polymer is spin-coated onto the substrate
3. Soft baking — gentle heating drives off excess solvent and improves adhesion to the substrate
4. Mask alignment — a photomask containing the desired pattern is precisely positioned over the coated substrate
5. UV exposure — light passes through the mask and selectively changes the solubility of the photoresist wherever it strikes
6. Development — a chemical developer dissolves away the more soluble regions, revealing the pattern
7. Hard baking — higher-temperature heating stabilizes the remaining photoresist so it can withstand subsequent processing
8. Etching or deposition — the exposed substrate areas are chemically etched or have material deposited onto them
9. Photoresist removal (stripping) — the remaining resist is dissolved away, leaving only the patterned substrate

Four key components make this work:

• Light source — typically emitting UV wavelengths, since shorter wavelengths allow finer features
• Photomask — a patterned template (usually chrome on glass) that selectively blocks or transmits light
• Photoresist — the light-sensitive polymer whose solubility changes upon exposure
• Substrate — the base material (often a silicon wafer) that receives the final pattern

Photolithography shows up across many industries: integrated circuit production, MEMS fabrication (tiny sensors and actuators), microfluidic lab-on-a-chip devices, and flat panel display manufacturing.

Composition of photoresists

Photoresists fall into two categories based on how they respond to light. The distinction matters because it determines which areas get removed during development.

Positive photoresists become more soluble where exposed to light. The exposed regions wash away in developer, so the pattern on the substrate matches the dark areas of the mask. They contain three main components:

• Base resin (novolac) — a phenolic polymer that provides the structural film
• Photoactive compound (diazonaphthoquinone, or DNQ) — this is the molecule that undergoes a photochemical change, switching from dissolution-inhibiting to dissolution-promoting upon UV absorption
• Solvent (e.g., PGMEA) — allows the resist to be applied as a uniform thin film by spin-coating

Positive resists are favored when you need high resolution and tight process control, which is why they dominate in semiconductor manufacturing.

Negative photoresists become less soluble where exposed to light. The exposed regions harden and remain after development, so the pattern on the substrate matches the clear areas of the mask. Their composition differs accordingly:

• Base polymer (epoxy-based, such as SU-8, or polyisoprene) — forms the film that will crosslink upon exposure
• Photoinitiator — absorbs UV light and generates reactive species (typically free radicals or acids) that trigger crosslinking
• Solvent — enables thin film formation during coating

Negative resists offer better adhesion to substrates and higher chemical resistance, making them well-suited for applications where the patterned resist itself must survive harsh processing, such as electroplating molds or microfluidic channel walls. Their trade-off is lower resolution compared to positive resists, because swelling during development can distort fine features.

Photochemical reactions in patterning

The chemistry that happens during UV exposure is what makes the entire process work. The reactions differ fundamentally between the two resist types.

In positive photoresists, the key reaction is the photodecomposition of DNQ:

1. DNQ absorbs a UV photon and loses a molecule of $N_2$ gas (a Wolff rearrangement occurs)
2. The resulting carbene intermediate reacts with trace water to form a carboxylic acid (indenecarboxylic acid)
3. This carboxylic acid is polar and readily dissolves in the alkaline developer solution
4. Unexposed DNQ remains hydrophobic and acts as a dissolution inhibitor, protecting those regions from the developer

In negative photoresists, the reaction is a light-initiated polymerization:

1. The photoinitiator absorbs UV light and fragments, generating free radicals (or, in chemically amplified resists, a photoacid)
2. These radicals initiate chain reactions that crosslink neighboring polymer chains into a dense network
3. The crosslinked network is insoluble in the organic developer solvent
4. Unexposed regions remain un-crosslinked and dissolve away during development

Quantum yield ($\Phi$) measures how efficiently these photochemical reactions proceed: it's the number of reaction events per photon absorbed. A higher quantum yield means the resist is more sensitive, requiring less exposure energy and shorter exposure times. This directly affects manufacturing throughput.

Advanced Lithography Concepts

Resolution limits in lithography

The smallest feature you can print is governed by the Rayleigh criterion:

$R = k_1 \frac{\lambda}{NA}$

• $R$ = minimum resolvable feature size
• $\lambda$ = wavelength of the exposure light
• $NA$ = numerical aperture of the projection lens (describes how much light the lens can collect and focus)
• $k_1$ = a process-dependent factor (accounts for resist chemistry, mask optimization, and imaging tricks; theoretically bounded at 0.25 on the low end)

The equation tells you there are three levers for shrinking features: use a shorter wavelength, increase the numerical aperture, or reduce $k_1$ through process engineering.

Challenges at small scales:

• Diffraction blurs pattern edges, making it harder to define sharp features
• Depth of focus shrinks as resolution improves, restricting how thick the resist layer can be and how flat the substrate must be
• Proximity effects cause nearby features to influence each other's exposure dose
• Line edge roughness (LER) introduces random variations in feature dimensions that become significant when features are only tens of nanometers wide

Techniques that push beyond traditional limits:

• Extreme ultraviolet (EUV) lithography — uses a 13.5 nm wavelength (compared to 193 nm for conventional deep-UV), dramatically improving resolution. This is the current leading-edge technology for sub-7 nm semiconductor nodes.
• Electron beam (e-beam) lithography — writes patterns directly with a focused electron beam, achieving extremely fine resolution without a mask. It's slow, so it's mainly used for mask making and research rather than mass production.
• Nanoimprint lithography — mechanically presses a mold into the resist to create patterns, bypassing optical diffraction limits entirely.
• Multi-patterning — combines multiple exposure and etch steps to create features smaller than a single exposure could achieve.

Emerging approaches are also under active development:

• Directed self-assembly (DSA) guides block copolymers to spontaneously form regular nanoscale patterns, using the thermodynamic tendency of the two polymer blocks to phase-separate
• Plasmonic lithography exploits near-field optical effects at metal nanostructures to confine light below the diffraction limit