The last three layers were perfectly aligned. Yesterday, I could also align the 4th layer very fine. However, today, I observed x, y and theta misalignment by 2-5um, three times, for the same wafer. Alignment marks and backside are ultra-clean.