The scaling down of devices from conventional planar CMOS towards advanced nanodevices – such as fin field-effect transistors (FinFETs) – has emerged specific technological challenges in spite that the standard fabrication techniques (such as ion implantation) has been maintained. In addition, these advanced device architectures have a 3D nature, which require simulation techniques capable of providing information of dopants and defects at atomic scale. Kinetic Monte Carlo atomistic simulations are intrinsically 3D and allow the simulation of complex geometries. Nanoscale feature sizes makes classical molecular dynamics simulations also capable to address this proble
One of the fundamental issues was related to poor dopant incorporation after the source/drain extension implants into the fin (due to its particular geometry and implant tilt angle requirements to avoid shadowing in fin arrays), which it directly affects the parasitic resistance that defines the dynamic performance of FinFETs. Our kinetic Monte Carlo simulations allowed us to identify the different sources of dopant loss which makes possible the optimization of implant parameters in order to maximize dopant incorporation. Thus, we have defined strategies for the optimization of dopant profiles in fin devices by a careful choice of tilt angles (as close to 45º as possible but avoiding shadowing), implant energy (to introduce dopants into the sidewalls but without losing ions through the opposite side of the fin) and fluence (to compensate for inefficient dopant incorporation).
3D schematic of a FinFET and schematic diagram of source-drain implant into the fin, including main contributions to dose loss (backscattering and ion trajectories escaping from fin dimensions).
Cross sectional view of the B concentration in a FinFET implanted with (a) 0.5 keV B 1015 cm-2, 10º tilt (implant angle a in the schematic of the fin), (b) 2 keV B 1015 cm-2, 10º tilt, (c) 0,5 keV B 5·1015 cm-2, 10º tilt, (d) 0,5 keV B 1015 cm-2, 45º tilt (L. Pelaz et al., IEEE International Electron Device Meeting (IEDM 2009), 513 (2009) - Atomistic process modeling based on Kinetic Monte Carlo and Molecular Dynamics for optimization of advanced devices).
Another critical issue was related to the formation of defects during the regrowth of the amorphized fin which degrades device performance and affects variability. By using classical molecular dynamics simulations we have proposed alternative processes to overcome the incomplete regrowth in full amorphized FinFETs oriented along <110> by choosing appropriate implant parameters to achieve partial amorphization or by orienting the FinFET along <100>.
Classical molecular dynamics simulation results showing two technological alternatives to improve regrowth in FinFET devices (L. A. Marqués et al., J. Appl. Phys. 111, 034302 (2012) - Molecular dynamics simulation of the regrowth of nanometric multigate Si devices).