Carrier scattering in high-κ/metal gate stacks

Abstract

A significant degradation of the mobility has been repeatedly observed at low inversion density in high-κ/metal gate metal-oxide-semiconductor field-effect transistors. However, the scattering mechanisms responsible for this degradation are still debated. It is often assumed that the mobility is limited by remote charges (RCS) at the interface between SiO2 and HfO2. However, the amount of charges needed to reproduce the experimental mobilities is usually very high (a few 1013 cm−2), and does not seem to be consistent with the measured threshold voltages. Scattering by localized dipoles hardly solves these discrepancies. Here, we investigate the contribution from three alternative mechanisms in a non-equilibrium Green's functions framework: (i) scattering by band offset fluctuations at the SiO2/HfO2 interface, (ii) scattering by dielectric constant fluctuations in SiO2 and HfO2, and (iii) scattering by workfunction fluctuations in a granular metal gate. None of these mechanisms significantly shifts the threshold voltage. We show that mechanisms (i) and (iii) efficiently scatter the carriers at low inversion densities. This reduces the amount of RCS charges needed to reproduce the experimental data. RCS and these mechanisms show different dependences on the thickness of the HfO2 layer, which might help to identify the dominant contributions.

Electric-field tuning of the valley splitting in silicon corner dots

“…Here we have modeled this disorder as a distribution of positive and negative charges at the SiO 2 /HfSiON with total density σ = 10 13 cm −2 . This large value is, again, consistent with the mobilities measured in planar as well as nanowire devices, and with their dependence on the thickness of the SiO 2 layer 35,36 . It must be seen as an effective density of charges mimicking all high-κ/metal gate-related disorders described above, which have similar fingerprints on the potential in silicon.…”

“…Room-temperature mobility measurements in high-κ/metal gate devices actually show the fingerprints of "Remote Coulomb Scattering" (RCS 33 ) by charges at the SiO 2 /HfSiON interface 34,35 with apparent densities as large as a few 10 13 cm −2 . In fact, the Coulomb disorder in the gate stack likely results from a combination of charge traps at this interface, local band offset fluctuations (interface dipoles), and possibly from work function fluctuations in granular metal gates 36 . The existence of significant disorder is consistent with the fact that the two corner dots are not fully symmetric, as revealed by the stability map in Fig.…”

Variability of electron and hole spin qubits due to interface roughness and charge traps

“…We then introduce the charged defects at the SiO 2 /HfO 2 instead of the Si/SiO 2 interface, with density n i = 5 × 10 11 cm −2 only chosen 86 for illustrative purposes. 87 The RSDs σ(f L ) and σ(f R ) of the Larmor and Rabi frequencies are plotted as a function of t SiO2 and t HfO2 in Fig. 11b.…”

“…Remarkably, the variability increases rapidly with the thickness of the HfO 2 layer because the screening of the charge traps by the metal gate is softened. 87 In general, surrounding the qubits by materials with higher dielectric constant (SiGe vs SiO 2 ), and by a dense set of gates will reduce the impact of charged defects on variability (and possibly of charge noise on qubit lifetimes). 88 Working in the many electrons/holes regime may also enhance screening, but usually makes the dots larger and more responsive to dis- order.…”

Scaling silicon-based quantum computing using CMOS technology: State-of-the-art, Challenges and Perspectives

“…Tungsten, with a thermal expansion coefficient similar to that of silicon (4.5 × 10 −6 K −1 vs 2.6 × 10 −6 K −1 ), may be an interesting candidate for lower resistivity gate material. The impact of metal gate granularity on the variability of the gate voltage for quantum-dot formation should also be minimised to avoid workfunction fluctuations [98][99][100].…”