What is photoresist? How are positive and negative resists different?

Photoresist (often just "resist") is a light-sensitive film coated on the wafer. Where light hits it, it changes; rinse it with the right chemical, and the circuit pattern is left on the wafer, like the light-sensitive layer on photographic film. There are two kinds: with positive resist, the parts hit by light wash away; with negative resist it's the other way around, and the parts hit by light stay.

How photolithography (litho for short) works as a whole is in "What is photolithography?", and why that area is lit yellow is in "Why is the lithography area in a chip fab yellow?". This article is only about the resist itself.

What is photoresist made of?

In the bottle, photoresist is a slightly thick liquid with three main ingredients:

• Resin: long, plastic-like molecules. Once coated, they're the backbone of the film.
• The light-sensitive part: it changes chemically when light hits it, and decides whether that spot will dissolve later.
• Solvent: it keeps the resist liquid so it can be spread, and most of it is baked off after coating.

Right after exposure there's no visible pattern yet, like film that hasn't been developed: the exposed parts just dissolve faster or slower in the developer than their neighbors.

How are positive and negative resists different?

	Positive resist	Negative resist
Where light hits	Becomes easy to dissolve and washes away	Molecules link together, it becomes hard to dissolve and stays
The pattern left	Matches the light-blocking parts of the photomask	The reverse of the photomask
Usual developer	A water-based alkaline solution	Older types use organic solvents
Common uses today	Most of the fine patterns on a chip	Very thick films, MEMS (micro-electro-mechanical systems)

Early integrated circuits mostly used negative resist. But the older negative resists were developed in organic solvents, and the resist that stayed soaked up solvent and swelled: lines warped, and lines close together could stick to each other, so they hit their limit at lines a few micrometers wide. Positive resist is developed in a water-based alkaline solution and doesn't swell like that, so as patterns got finer, most of them moved to positive resist; negative resist is now also often used where the coat has to be very thick.

Why does each kind of light need its own resist?

Resist has to absorb the light to react, but not too much, or the bottom of the film gets none. Early resists were tuned for violet-blue and ultraviolet light. When exposure moved to shorter-wavelength deep ultraviolet, the old formulas absorbed far too much, so the materials had to change; when deep ultraviolet then moved from 248 nm to 193 nm, the materials were too opaque again and changed once more. Extreme ultraviolet is another story; see "What is extreme ultraviolet (EUV) light?".

What steps does resist go through, from coating to removal?

Coating, baking and developing usually happen in one coater/developer, which is connected to the scanner.

Step 1: Clean, dry and prime

• What: Clean and dry the wafer, then use the vapor of an adhesion promoter (a common one is called HMDS) to lay down an extremely thin layer on the surface.
• Why: Silicon and oxide surfaces easily hold on to a layer of water, and resist doesn't stick well to them; the primer makes the surface water-repellent, so the resist can grip.
• Otherwise: During developing and later wet steps, water creeps in under the resist and fine lines lift off.

Step 2: Spin coating, flinging it into a thin, flat layer

• What: Put a little resist in the middle of the wafer and spin it at dozens of turns per second so the extra flies off; then wash off the thicker ring that builds up at the edge with solvent.
• Why: Spinning flings the resist out while the solvent evaporates, and the film sets to the same thickness everywhere; faster spinning gives a thinner film, but not in proportion (roughly four times the speed for half the thickness), and a more viscous resist gives a thicker film, so the thickness is mainly chosen by the resist and fine-tuned with the speed.
• Otherwise: With thick and thin spots, the same light is too little where it's thick and too much where it's thin, and the line width drifts; if the edge ring isn't removed, it later flakes off and turns into particles.

Step 3: Soft bake, driving off the solvent

• What: Put the wafer on a hot plate at around 100°C for a minute or two.
• Why: A freshly coated film still holds a lot of solvent; baking most of it off makes the film stable and stick better.
• Otherwise: Underbaked, the resist won't expose or develop as expected; overbaked, the heat destroys some of the light-sensitive part first and the resist becomes less sensitive.

Step 4: Exposure, putting the pattern in with light

• What: Light passes through the photomask onto the resist, and the exposed parts change chemically.
• Why: Only the parts hit by light change, which is what makes a pattern.
• Otherwise: Too much or too little light makes lines too wide or too narrow; if this layer doesn't line up with the one below, the layers don't connect. See "What is overlay?".

Step 5: Post-exposure bake, letting the reaction finish

• What: Right after exposure, the wafer goes back on a hot plate for another bake, the post-exposure bake.
• Why: Most resists for deep ultraviolet are "chemically amplified": the light only makes a tiny amount of acid, and during the bake each acid molecule sets off reaction after reaction, so very little light is enough, but how far the reaction goes depends entirely on temperature and time.
• Otherwise: One degree off on the hot plate can shift the line width by several nanometers, and a plate that is cooler on one side makes the wafer wide on one side and narrow on the other; if the wafer waits too long before the bake, traces of alkaline gas in the air (ammonia, for example) neutralize the acid at the surface, and the lines end up with an undissolved "lid" on top.

Step 6: Developing, the pattern appears

• What: Developer is sprayed on, the parts that should go dissolve, and the wafer is rinsed with water and spun dry. This is developing.
• Why: Exposed and unexposed parts dissolve at different speeds, so after the rinse, what's left is the pattern.
• Otherwise: If the temperature or time is slightly off, openings don't clear fully or the parts that should stay get eaten a little, and the line width changes again.

Step 7: Hard bake and inspection

• What: Many processes bake once more to harden the resist, then measure the line width and alignment and look for defects.
• Why: The resist is about to serve as a shield, and a harder resist holds up better; this is also the last point where a mistake can be undone.
• Otherwise: Baked too hot, the resist softens and flows, corners round off, and it's harder to remove later; skip the inspection, and a wrong pattern gets etched into the wafer. A misprint is stripped and redone, which is called rework.

Step 8: Shielding the wafer during etch or implant

• What: The wafer goes to etch or ion implantation: the parts under resist are protected, and the exposed parts are etched away or have atoms shot into them.
• Why: Etch and implant treat the whole wafer at once, and the resist is the stencil.
• Otherwise: If the resist is too thin, it gets eaten away before the etch is done; if developed wafers wait too long, some layers have to be redone too. See "What is Q-time?".

Step 9: Stripping the resist, job done

• What: The dry way uses an oxygen plasma to "burn" the resist into gas that's pumped away, called ashing; the wet way dissolves it in chemicals such as a strong acid mixed with hydrogen peroxide; the two are often used one after the other, followed by a clean.
• Why: The resist is only a temporary stencil, and left behind it becomes contamination.
• Otherwise: Leftovers travel with the wafer to the next step; resist that has been through ion implantation hardens into a crust on top and is especially hard to remove. For cleaning, see "What is wafer cleaning?".

In fab24hr

In fab24hr, tap "Clock in", choose "A day in the fab", then on the "Your role today" card tap "Production fab", swipe to the process engineer and tap "Litho" under the name: you're the litho process engineer. Late at night, the photoresist bottle on coater/developer TRK-02 is nearly empty, and fresh resist brings tiny bubbles that take hours to clear: swap it while things downstream are stopped, or leave it for the morning shift? In the afternoon, one of the four hot plates TRK-01 bakes on in turn runs cool on one side, the wafers that land on it come out with lines wide on one side and narrow on the other, and you pick which ones to rework.

The R&D line's litho (tap "R&D line" on the card) also gets a new bottle late at night: it came out of the fridge only an hour ago and is from a different batch, the film comes out thicker than usual, and with the usual amount of light, the holes on one side still have resist at the bottom. In "Station by Station", chapter 2 of the main story "The life of a chip", you coat resist, bake, expose and develop by hand to print the first pattern. The companies, people and tool numbers in the game are fictional; how photoresist works and its steps come from public, general sources. For the jargon, see Chip fab jargon in plain English.