High Dynamic Range and High Precision Phase Resolution Algorithm Based on Dynamic Frequency Phase Locking

As the core equipment for integrated circuit manufacturing, the multi-degree-of-freedom super-precision positioning technology of the photolithography machine's working table is the key to achieving nanometer-level pattern overlay accuracy. At the 14 -nanometer and below process nodes, the working table needs to maintain sub-nanometer-level positioning stability during high-speed scanning motion, which sets extremely high requirements for the dynamic accuracy and anti-interference ability of the displacement measurement system. Aiming at the phase-solving problem in heterodyne interferometric signal processing, this paper proposes a dynamic frequency-locked phase-locked algorithm. The innovation lies in the use of two groups of orthogonal and equally-spaced reference frequency signals ($3-37 \text{MHz}$, spaced 1 MHz) to mix with the input signal. The difference-frequency component is extracted by a low-pass filter (with a cutoff frequency of 500 kHz). Compared with traditional methods, the core innovation is the introduction of a dynamic frequency-switching mechanism. When the measurement signal frequency exceeds the ±500 kHz range of the current reference channel, it automatically switches to the adjacent reference frequency channel, achieving frequency tracking and maintaining phase continuity. To verify the algorithm's performance, an experimental platform based on a signal generator was built. The static test shows that the algorithm's phase fluctuation peak-to-peak is less than ±0.4 mrad, with a static standard deviation of 0.101 mrad (equivalent displacement of 8.04 pm), which is 4.9 times and 3.1 times better than that of the orthogonal phase-locked method (0.492 mrad) and the frequency-comb phase-locked method (0.318 mrad). In the 0.096 mrad step test, the system resolution reaches 7.63 pm. In the dynamic test, the dynamic standard deviation is below $0.75~\text{mrad} (59.7 \text{pm})$, verifying the algorithm’s robustness in high-speed and frequency-varying scenarios.