depth of focus (dof)

Depth of Focus (DOF) is the range of vertical positions (wafer height) over which the projected aerial image remains acceptably sharp and the printed feature dimensions stay within specification, representing a critical process window parameter in semiconductor lithography. DOF determines how much the wafer surface can deviate from the ideal focal plane — due to wafer flatness variation, chuck leveling, topography from underlying layers, and focus control accuracy — while still producing acceptable patterns. The Rayleigh DOF formula is: DOF = k₂ × λ / NA², where λ is the exposure wavelength, NA is the numerical aperture, and k₂ is a process-dependent factor (typically 0.5-1.0). This relationship reveals a fundamental tradeoff: increasing NA improves resolution (proportional to λ/NA) but dramatically reduces DOF (proportional to λ/NA²) — resolution improves linearly with NA while DOF degrades quadratically. For 193nm immersion at NA = 1.35: DOF ≈ 0.5 × 193nm / 1.35² ≈ 53nm — an extraordinarily thin slice requiring sub-50nm focus control accuracy. Factors consuming the DOF budget include: wafer non-flatness (local height variation within the exposure field — specified as focal plane deviation, typically 20-40nm for advanced wafers), topography (height variations from underlying metal, dielectric, and gate layers — can consume 50-100nm or more), lens aberrations (field-dependent focal plane curvature and astigmatism — calibrated and corrected but with residual errors), and environmental factors (pressure and temperature changes affecting the air or immersion medium refractive index). DOF enhancement techniques include: phase-shift masks (improving image contrast allows slightly defocused patterns to still print acceptably), source optimization (specific illumination conditions can improve DOF for targeted feature types), chemical mechanical planarization (CMP — flattening wafer topography to reduce the focus budget consumed by surface height variation), sub-resolution assist features (SRAF — improving process window robustness), and computational lithography (co-optimizing source, mask, and resist processing for maximum DOF).