Pre-Ionization High-Voltage Optimization for Excimer ArF Lasers

The 193 nm ArF excimer laser remains the workhorse of immersion lithography at sub-7 nm nodes, where dose stability below 0.15 % and wavelength stability under 0.07 pm require pre-ionization corona discharge to initiate with microsecond precision and jitter below 800 ps across hundreds of millions of pulses. Optimized pre-ionization high-voltage systems therefore operate in the 18–26 kV range with rise times under 120 ns and amplitude repeatability better than 0.4 % while surviving the aggressive fluorine environment inside the laser chamber.

The generator topology uses a magnetic pulse compression network driven by a compact thyratron-switched primary capacitor, followed by a three-stage saturable inductor compression that delivers 22 kV pulses with 95 ns FWHM into a 5.2 pF effective load presented by the corona pins through ceramic feedthroughs. Energy per pulse is tightly controlled by a resonant charging supply that regulates primary capacitor voltage to ±3 V using a command-charging IGBT buck regulator locked to a precision 0.01 % reference.

Pulse-to-pulse amplitude stability relies on active compensation for thyratron recovery variation. A fast photodiode monitoring pre-ionization UV emission feeds a digital PID that adjusts charging voltage on the subsequent pulse by up to ±45 V, nullifying the 1.2 % amplitude drift previously observed as electrode erosion increased trigger delay. Resulting energy stability remains under ±0.28 % over 80 million pulse intervals without manual recalibration.

Rise-time consistency is maintained through automatic saturation bias adjustment of the final magnetic switch. As core temperature rises during high-duty bursts, saturation flux density decreases; an embedded Hall probe measures B-field collapse timing every 5000 pulses and increases DC reset current in 2 mA steps to restore 110 ns rise time within ±4 ns. This prevents the delayed breakdown that historically caused main discharge timing jitter and subsequent dose errors.

Fluorine resistance is achieved by full encapsulation of the pulse generator in a nitrogen-purged sub-module using only glass, ceramic, and nickel-iron magnetic materials. All organic insulation is eliminated, and high-voltage connections use spring-loaded crown contacts that maintain less than 0.8 mΩ resistance after 18 months of continuous F2 exposure at 70 °C.

Trigger synchronization with the main discharge is locked via a fiber-optic phase-locked loop that references the thyratron trigger to the main PFL charging waveform. Absolute delay is held within ±650 ps over operating temperature, ensuring pre-ionization occurs 420 ns before the main discharge voltage reaches 16 kV, the optimal window for maximum energy extraction and minimum electrode sputtering.

Energy-based dosing replaces voltage-based control entirely. A current-viewing resistor on the corona return path integrates actual transferred charge; when integrated charge reaches target ±0.6 %, the next pulse charging voltage is trimmed accordingly. This compensates for gradual pin erosion and dielectric window coating that previously caused 4–6 % energy drift over gas life.

These optimized pre-ionization systems routinely deliver greater than 99.9998 % trigger success rate and dose stability supporting E0 below 0.12 mJ/cm² across 120 million pulse gas fills, enabling sustained 7 nm and 5 nm layer imaging with CD uniformity under 0.9 nm 3-sigma in volume production.