The lithographic process used in semiconductor manufacturing follows a precise sequence of steps: surface preparation, photoresist application, soft baking, alignment and exposure, post-exposure baking, development, etching, and resist stripping. Each step builds on the last, and even small errors at any stage can ruin the final pattern. Here’s what happens at each one and why it matters.
Surface Preparation
Before anything else, the silicon wafer must be perfectly clean and dry. Any moisture, dust, or organic residue on the surface will prevent the next layers from sticking properly. Wafers are first cleaned to remove particles and contaminants, then placed in a vapor prime oven that combines heat with low pressure to drive off not just surface moisture but chemically bound water molecules as well.
Once dehydrated, the wafer is primed with a chemical called HMDS (hexamethyldisilazane). Since photoresist is an organic material and the wafer is not, HMDS acts as a molecular bridge that bonds to both surfaces and dramatically improves adhesion. Without this priming step, the resist can peel or lift during later processing, destroying the pattern.
All of this happens inside a cleanroom, typically rated ISO Class 4 to 6. That means the air contains no more than about 350 to 35,200 particles per cubic meter at 0.5 micrometers in size. For context, normal outdoor air has millions of particles per cubic meter. A single stray particle landing on the wafer during lithography can cause a defect in the final chip.
Applying the Photoresist
Photoresist is a light-sensitive liquid that gets spun onto the wafer in a process called spin coating. A few milliliters of resist are dispensed onto the center of the wafer, which then spins at several thousand revolutions per minute. Centrifugal force spreads the liquid into a thin, uniform film across the entire surface.
The final thickness of the resist depends on the spin speed. Higher speeds produce thinner films: one common resist, for example, yields a 4.0 micrometer film at 2,000 RPM, 3.3 micrometers at 4,000 RPM, and 2.3 micrometers at 6,000 RPM. The relationship follows a predictable curve where thickness decreases with the square root of spin speed, giving engineers fine control over the coating. For very thick films (needed in some specialized applications), multiple coating cycles can stack layers up to 160 micrometers.
Spinning too slowly creates problems. The resist builds up around the wafer’s edge in a thick bead, and the coating becomes uneven and unreliable. To avoid this, the standard approach is to start at a lower speed (around 1,000 RPM) to spread the resist, then ramp quickly up to the final speed that sets the desired thickness.
Soft Bake
After coating, the wafer goes through a soft bake on a hotplate. This step drives off most of the solvent left in the resist film, turning it from a wet coating into a stable solid that won’t stick to the photomask during exposure. The standard recipe is about 100°C for one minute per micrometer of resist thickness. Temperatures can range from about 55°C to 115°C depending on the resist type and the process requirements.
Getting this bake right is a balancing act. Too little heat leaves excess solvent, which makes the resist too soft and degrades the pattern during exposure. Too much heat starts to break down the light-sensitive compounds in the resist before they’ve even been exposed, reducing contrast and resolution.
Alignment and Exposure
This is the core of the process: transferring the desired circuit pattern onto the resist using light. A photomask, essentially a glass plate with the pattern etched in chrome, is positioned above the wafer. Alignment marks on both the mask and the wafer are used to ensure the pattern lands exactly where it needs to, especially when multiple layers must stack on top of each other with nanometer-level accuracy.
The exposure tool shines ultraviolet light through the mask and onto the resist. Two main types of tools handle this job. Steppers expose one rectangular chip area at a time, then “step” to the next position and repeat. Scanners combine stepping with a scanning motion, sweeping a narrow slit of light across both the mask and wafer simultaneously. This scanning approach allows larger exposure fields and higher productivity, which is why scanners dominate modern production lines.
The wavelength of light determines the smallest features you can print. Older systems use deep ultraviolet light at 193 nanometers. The latest technology, extreme ultraviolet (EUV) lithography, uses light at just 13.5 nanometers. At that wavelength, conventional glass lenses absorb the light entirely, so EUV systems use a series of precisely engineered mirrors instead. These mirrors are made from alternating layers of molybdenum and silicon, achieving about 76% reflectivity at that wavelength. Even that small loss compounds across multiple mirrors, making EUV systems extraordinarily complex and expensive.
Post-Exposure Bake
Some resist types, particularly those used with advanced exposure methods, require a bake immediately after exposure. This step activates or completes the chemical reaction that the light started, and it also helps even out any standing wave effects that can create ripples in the resist sidewalls. The temperatures and times vary by resist chemistry, but the goal is consistent: ensure the chemical changes from exposure are complete and uniform before the next step.
Development
Development is where the pattern becomes visible. The wafer is immersed in or sprayed with a developer solution that selectively dissolves part of the resist. What gets dissolved depends on whether you’re using a positive or negative photoresist.
With positive photoresist, the areas that were hit by light become soluble. UV exposure breaks the polymer chains in the resist (a process called chain scission), making those regions dissolve easily in the developer. The unexposed areas remain intact. The result is a resist pattern that matches the clear areas of the mask: light passes through, resist disappears.
Negative photoresist works in reverse. Light causes the polymer chains to cross-link and harden, making the exposed areas insoluble. The developer then washes away the unexposed regions, leaving behind a hardened copy of the pattern. This means you get the opposite tone from the same mask.
The developer solution for modern photoresists is typically a dilute alkaline solution at about 2 to 3% concentration. Development time and technique (puddle, spray, or immersion) are tightly controlled, since over-developing erodes the pattern edges and under-developing leaves residue behind.
Etching
With the resist pattern now sitting on the wafer surface, the exposed areas of the underlying material can be etched away. The remaining resist acts as a protective mask: wherever resist sits, the material underneath is shielded from the etchant.
Etching can be done with liquid chemicals (wet etching) or with reactive gases in a plasma chamber (dry etching). Wet etching is simpler but tends to undercut sideways beneath the resist, which limits how small the features can be. Dry etching, particularly reactive ion etching, can cut nearly straight down, producing the vertical sidewalls needed for modern chip features measured in nanometers.
After etching, the circuit pattern that was designed on the mask now exists as a physical structure in or on the wafer surface.
Resist Stripping
The final step removes all remaining photoresist from the wafer. This can be done with chemical solvents, plasma ashing (using oxygen plasma to burn off the organic resist), or a combination of both. The wafer surface must be completely clean afterward, because this entire sequence will be repeated dozens of times to build up the many layers of a modern chip.
Quality Checks Between Steps
Throughout the process, two measurements are critical. Critical dimension (CD) measurement verifies that the printed features are the correct width. Overlay measurement checks that each new layer aligns precisely with the layers beneath it. Both were traditionally measured on separate tools, but newer techniques can capture both simultaneously using specialized target patterns on the wafer. Overlay targets are typically 2 to 3 micrometers in size, far larger than the circuit features themselves, and are placed in dedicated measurement areas on each chip.
These checks happen at multiple points during the lithographic sequence, not just at the end. Catching an error after exposure but before etching, for instance, allows the resist to be stripped and the step repeated, saving the wafer from being scrapped entirely.