In CMOS fabrication, the bulk/substrate is obtained by crystal growth. Being the first unit step in the process of fabrication, it does not require any controlled doping. Whatever doped will be present in the substrate. Subsequently, the appropriate regions of the MOS device(source and drain) are 'diffused' into the substrate. This has to be done through diligent and controlled doping. Diffusion is preferred over crystal growth because controlled doping is possible in the former whereas, it is not, in the latter.
As a matter of fact, the concentration of carriers being diffused must be at least an order of magnitude (10 times the layer beneath). eg: If an n-substrate is doped with 10^9 atoms/cm^3 then the p+ diffusion layer above it must have greater than 10^10 atoms/cm^3. This is to account for the tolerance range of 10-20% resulting from the possible variations in doping levels. If the p+ diffusion is low, then the carriers in p+ diffusion will be compensated by those in n-substrate, thus leading to an effectively useless device.