0.51um: • substrate/well taps require a 0.12um overlap of n+ / p+ diffusion on 1 axis • diffusion <-> diffusion spacing requirement is 0.27um So three options were possible: 1. Cover the entire width of the power rails in diffusion and space contacts every 0.17um. 2. Have individual islands of diffusion, resulting in contacts spaced every 0.27um; required overlap would go N/S 3. Have individual islands of diffusion, resulting in contacts spaced every 0.51um; required overlap would go E/W Option 2 would make our rails far thicker, from 0.61um to 0.61+0.12*2 = 0.85um, so we quickly ruled it out. EDIT: the math in this line has been corrected since the original posting Option 1, we feared, may cause the emergence of parasitic effects due to the existence of this thin layer of diffusion stretching out over the entirety of the layout area. So by process of elimination we arrived at Option 3.