Replacing the mercury lamp in a mask aligner with UV-LED

The mercury arc lamp has lit mask aligners for decades. It is bright, broadband and well understood, and most resist datasheets assume it. It is also the most frequent consumable on the tool, a source of downtime every time it ages or fails, and a component that contains mercury.

Converting a mask aligner from a mercury lamp to a UV-LED light source removes the lamp changes and the warm-up, and in most cases gives steadier intensity from one exposure to the next. It also changes the spectrum your resists see, so the conversion is a process change as well as a hardware change. This post explains what changes, what to check first and how to bring your recipes across.

What the mercury lamp does today

In a typical aligner lamp house, a short-arc mercury lamp sits at the focus of an ellipsoidal mirror. The light passes through optics that homogenise and collimate it before it reaches the mask. A constant-intensity or constant-power controller holds the output steady as the lamp ages, using a sensor that monitors the beam.

The spectrum contains strong mercury lines, the best known being i-line at 365 nm, h-line at 405 nm and g-line at 436 nm, plus deeper UV lines that some optics pass and some filter out. Users choose filters to select or block parts of that spectrum depending on the resist and the resolution they need.

The lamp's practical problems are familiar to anyone who runs an aligner:

• Warm-up. The lamp needs time to stabilise after ignition, so users leave it on all day whether they expose or not.
• Ageing. Output drops over the lamp's life and the arc can wander, which affects uniformity. Lamps are replaced on hours, often before they fail.
• Heat and ozone. The lamp house needs cooling and, for ozone-producing lamps, exhaust.
• Handling. Lamps run at high internal pressure when hot and contain mercury. Changing them needs care, and spent lamps go to hazardous waste.

What a UV-LED source changes

A UV-LED source replaces the lamp, and sometimes the whole lamp house, with an array of LEDs at one or more fixed wavelengths, plus drive electronics and cooling. Commercial sources for aligners commonly use wavelengths around 365 nm and 405 nm, with 385 nm or a channel near the g-line on some models, and the mix set to suit the resists in use. Some allow each wavelength to be switched or dimmed separately.

The practical effects:

• Instant on. LEDs reach full output immediately. The source is switched on only during exposure, so there is no need to leave it running.
• Stable intensity. LED output is controlled electronically and drifts slowly. Exposure doses become more repeatable from run to run.
• No lamp changes. LEDs degrade over time but do not fail in the abrupt way an arc lamp can. The source still needs periodic intensity checks.
• Less heat at the mask. LEDs emit little infrared, so less heat reaches the mask and wafer. This can help with thermal expansion on long exposures.
• No mercury. No lamp disposal, and no mercury hazard in the cleanroom.

The spectrum question

The part of the conversion that needs real thought is the spectrum. A mercury lamp gives a broad mix of lines; an LED source gives narrow bands at chosen wavelengths. That has consequences:

• Resist sensitivity. Each resist absorbs differently across 365–436 nm. A resist tuned for broadband exposure may need a different dose under a narrower LED spectrum, and some resists, such as thick negative epoxy resists, depend strongly on i-line.
• Deep UV. If any of your processes rely on wavelengths below 365 nm, an LED source built for i-line and longer will not cover them. Check this before converting.
• Resolution and sidewalls. Shorter wavelengths give better resolution in proximity printing. A source weighted towards 365 nm behaves differently from one weighted towards 405 nm, and sidewall angles in thick resists can change.

Before converting, list every resist used on the aligner, its thickness range and the filter currently fitted. That list decides which LED wavelengths the source needs.

Uniformity and optics

The aligner's optics were designed for a small, bright arc at the mirror's focus. An LED array is larger and emits differently. Good conversions handle this in one of two ways: either the LED source is built to imitate the lamp's position and divergence so the existing homogeniser and collimation optics still work, or the conversion replaces the lamp house optics with a matched design.

Either way, measure uniformity across the exposure field before and after. Map intensity at several points across the largest wafer size you use, at each wavelength channel if the source supports it. The result should be at least as good as the tool achieved with a fresh, well-aligned lamp.

On platforms such as the SUSS MicroTec MA6 and the EV Group EVG620, lamp house geometry and controllers differ between generations, so confirm the source is made for your exact configuration.

Bringing your recipes across

Plan the requalification as a short project:

1. Record the baseline. With the mercury lamp still fitted and in good condition, measure intensity at the probe wavelengths and run a dose matrix on each standard resist. Keep the developed patterns and measurements.
2. Install and map. Fit the LED source, check interlocks and cooling, and map uniformity.
3. Calibrate the intensity probe. Make sure your UV meter reads the LED wavelengths correctly. Probes calibrated for a mercury spectrum can read a narrow-band source differently.
4. Repeat the dose matrix. Find the new dose for each resist and thickness, and check feature size, sidewall profile and development time.
5. Update the recipe sheet. Change exposure times or doses, and note the date and source settings so users know which recipes are current.

Expect most positive i-line resists to transfer with a dose adjustment only. Thick and negative resists may need more work.

When the conversion pays off

The case is strongest where the aligner is used intermittently through the day by many users, where lamp costs and downtime are a recurring nuisance, or where your organisation wants to remove mercury from the cleanroom. It is weaker if the tool depends on deep UV exposure, or if a single, heavily optimised process would need full requalification for little gain. The main cost drivers are the source, any lamp house modifications, the time to map uniformity and the requalification of your resists.

What we can do

We survey the aligner, help you choose wavelengths for your resists, install the UV-LED source and hand over a uniformity map and dose results for your standard processes. See our UV-LED conversion service for the scope and steps, or the lithography equipment we refurbish and support.