Application Analysis of Superconducting Magnetic Energy Storage Pulse Power Supply for High-Voltage Power Supply in Lithography Machines

In the field of semiconductor manufacturing, lithography machines serve as core equipment for breakthroughs in advanced process technologies. Their exposure systems impose stringent requirements on the accuracy, response speed, and stability of power supply. Particularly in processes below 7nm, the microsecond-level regulation of exposure energy for photoresists and the pulse-driven operation of laser light sources both rely on the instantaneous energy output of high-voltage power supplies, and traditional power supply solutions are gradually unable to meet this demand.
Traditional high-voltage power supplies for lithography machines mostly adopt capacitor energy storage or battery energy storage architectures. Although capacitor energy storage has a relatively fast response speed, it features low energy density. Frequent charging is required during continuous pulse output, which easily increases the voltage ripple coefficient (usually greater than 0.5%), thereby causing deviations in lithographic line width. Battery energy storage, on the other hand, suffers from charging and discharging delays, failing to meet the dynamic adjustment needs of microsecond-level pulses. Moreover, its stability decreases during high-current output. These issues restrict the process advancement of lithography machines—for example, processes below 7nm require millisecond-level or even microsecond-level power supply response, which is hardly achievable with traditional solutions.
The introduction of superconducting magnetic energy storage (SMES) technology provides a new solution for high-voltage power supplies in lithography machines. Superconducting materials exhibit zero resistance in low-temperature environments, enabling their energy storage coils to achieve nearly lossless energy storage. With an energy density 5-8 times that of traditional capacitors and a charging/discharging response speed of only microseconds, SMES can accurately match the pulse energy requirements of lithography machine exposure systems. A typical SMES-based high-voltage power supply system consists of four components: a superconducting coil, a cryogenic refrigeration unit, a power conversion module, and an intelligent control system. The cryogenic refrigeration unit maintains an ultra-low temperature environment of 10-20K to ensure the superconducting coil remains stably in a zero-resistance state. The power conversion module converts the DC energy stored in the superconducting coil into high-voltage pulse AC power required by the lithography machine, with a conversion efficiency of over 95%. The intelligent control system dynamically adjusts the pulse amplitude, width, and frequency by real-time collecting load signals from the exposure system, controlling voltage fluctuations within ±0.1% and effectively avoiding lithographic accuracy deviations.
In practical applications, SMES-based high-voltage power supplies need to address two core issues. First is quench protection: when the low-temperature environment is damaged and the superconducting coil quenches, the system must cut off energy output within milliseconds to prevent coil burnout and ensure the safety of the lithography machine. Second is pulse waveform matching: different lithographic processes (such as pre-exposure and main exposure) have significant differences in pulse parameter requirements. The system must have multi-mode output capabilities and realize real-time waveform reconstruction through digital signal processing technology. Currently, this technology has been applied in some prototype advanced lithography machines. Compared with traditional solutions, it not only increases the lithographic yield by 8%-12% but also reduces the overall energy consumption of the equipment by more than 15%, effectively alleviating the high-energy consumption pain point in semiconductor manufacturing.
In the future, with the breakthrough of high-temperature superconducting material technology, the refrigeration cost of SMES-based high-voltage power supplies will be further reduced, and the equipment volume will be significantly reduced. Meanwhile, combined with AI load prediction algorithms, the system can predict exposure demands in advance to achieve more accurate energy scheduling, providing key support for the research and development of lithography machines for 1nm and below processes.