Paul van der Heide oral presentation (SN1-Mon2-3-1)
Critical need and future directions of SIMS depth profiling in CMOS device fabrication
imec, Kapeldreef 75, 3001 Leuven, Belgium
Following the year the CMOS patent was filed (1963) to the turn of the century, SIMS played a central role in defining dopant distributions. Dopants are required to define a solid state transistors deletion region. SIMS filled this need as a result of its unparalleled detection limits over localized regions (≥50x50 micron test pads or blanket wafers) and the sub nm depth resolution realized on using ultra low energy primary ions. Within this era, referred to in CMOS circles as the “Dennard era”, Moore’s law was adhered to chiefly through planar structure shrinkage.
To adhere to Moore’s law following the turn of the century required the introduction of multiple new process flows that have allowed the introduction of: High K metal gate stacks, strain engineering, movement from planar to 3D structures and more complex patterning procedures to circumvent the 193 nm lithography diffraction limit (~40 nm). This has increased device fabrication costs while also introducing new SIMS niche areas. Two niche areas include: tuning B and Ge concentrations in the nm scale Si(1-x)Gex films (epitaxially grown on Si to induce strain) and Si-Si(1-x)Gex interface impurity analysis (if not controlled, this can result in crystal relaxation). The increased costs, on the other hand, has driven SIMS toward 24/7 coverage plans with “as precise as possible” and “as fast as possible” data expected.
Future developments of SIMS in supporting CMOS device fabrication may include: further implementation and development of 1.5D SIMS protocols for examining 3D structures, successful implementation of SIMS in the clean room, development of scatterometry like approaches, i.e. comparing SIMS profiles to library profiles, etc. Also of note is the fact that future transistors may not require dopants. One interesting example lies in confinement modulated gap transistors, as these rely on quantum confinement effects to control the band gap as opposed dopant induced control of the Fermi edge within the band gap as used today. Whichever direction CMOS and post-CMOS based devices proceed, there is sure to be a place for SIMS.