With the continuous gate length downscaling, the “access” parts (contacts and doped drain/source) have been playing an increasing role in the device performances. They are now representing a significant part of the total resistance and capacitance of the devices, and are therefore limiting both the ON-current and carrier dynamics.
Injection from silicides to silicon and scattering in highly doped regions therefore need special attention but have been little explored up to now. The geometry of the source/drain is also expected to have a strong impact on the contact (spreading) resistance as the lateral dimensions of the channel are reaching the 10 nm range.
Injection from metallic contacts
In general, carriers are injected in the semiconductor device through a complex contact designed to be as ohmic as possible. The last layer of the contact stack, that actually flows carriers inside the (highly doped) semiconductor, is certainly the most critical. It is very often a “silicide”, an alloy between silicon and a metallic element such as NiSi2. The NiSi2/Si interface will therefore serve as a prototype of a metal/semiconductor interface for this project. The quality of this interface and the match between the silicon and silicide band structures are the key elements controlling the resistance of the contact and its nature (ohmic/Schottky).
We will study the injection through a NiSi2/doped silicon interfaces using self-consistent Green’s functions solvers on top of an atomistic tight-binding description of these materials, along the lines. These atomistic solvers can account for the effects of disorder, which are expected to relax the conditions on parallel momentum conservation at the interface, hence significantly enhancing transmission.
Scattering in highly doped regions
The doped regions between the metal/semiconductor contact and the channel can be a significant part of the total resistance of the devices. The interplay between lateral confinement and impurity bands, the scattering within these regions and the transition with the (lightly doped or even undoped) channel have been little explored.
We will compute the mobility of highly doped (degenerate) silicon nanostructures with self-consistent Green’s functions methods on top of k.p models. In that regime, self-consistent Green’s functions shall provide adequate screening of the impurities. Electron-phonon scattering will also be considered in order to mitigate the disorder-induced localization and get more reliable mobility data.
Optimization of the contact geometry and doping profile
In short channel devices, the channel width and/or height are typically in the 10 nm range and below. The channel therefore makes a pronounced “constriction” with respect to the source and drain access regions. Such a constriction gives rise to a “spreading resistance” which gets larger as the devices get thinner. Also, short channels are much more sensitive to the details of the doping profile that controls carrier density throughout the devices. Contact geometry and doping profiles shape both the current and the capacitances of the devices, which call for a careful analysis and optimization. Here, we will use self-consistent NEGF solvers to address these issues.
- Study the dependence of the resistance and capacitance of trigate devices on the geometry of the source/drain to channel constrictions
- Study the effects of the doping profile on the performances of FDSOI and trigate devices
Study of transient regimes within the NEGF formalism
All NEGF simulations proposed so far are steady state, zero-frequency calculations. Although the switching speed of transistors is mostly limited by parasitic source and drain capacitances today, the “intrinsic” carrier dynamics in the channel and contacts might become an increasingly important limitation as the devices are further scaled down. We will therefore explore options for the modeling of the intrinsic transient dynamics and frequency response of nano-devices within empirical Hamiltonian formalism.