Search

A fundamental issue regarding the introduction of high-mobility Ge channels in CMOS circuits is the electrical passivation of the interface with the high-k gate dielectric. In this paper, we investigate the passivation of p-Ge(001) using molecular H"2S. The modification of the semiconductor surface is monitored in situ by RHEED and the interface is characterized by XPS analyses. MOS capacitors are fabricated to extract interface state density, and finally we demonstrate the efficiency of the passivation scheme using a combination with an ultra thin Al interlayer.

Si and Se implantations have been systematically investigated in In"0"."5"3Ga"0"."4"7As. Different implant doses and various activation anneals with temperatures up to 700^oC have been examined. Raising Si implant dose from 1x10^1^4 to 1x10^1^5cm^-^2 was found to increase the active doping concentration by about a factor of two. As confirmed by Transmission Electron Microscopy (TEM) and electrical measurements, the rest of the implanted Si ions remain as defects in the crystal and degrade the mobility. It was also confirmed from Secondary Ion Mass Spectrometry (SIMS) that the Si diffusivity in InGaAs is negligible up to 700^oC implant activation anneal making Si a suitable option for the formation of shallow junctions in InGaAs. The activation efficiency, sheet resistance, carrier density and mobility data of 25keV Se and Si implanted InGaAs layers are also presented under various activation anneal temperatures.

Numerous metal oxides have been studied worldwide as possible high-k gate dielectric candidates for MOS devices on alternative semiconductor materials (Ge, III/V compounds). We will discuss thermal and plasma-enhanced atomic layer deposition (ALD) of a few materials, HfO"2 and Al"2O"3. We will spend some attention to characteristic features of the growth process and specific growth precursors as this is known to influence strongly the quality of the layer bulk as well as the interface. Detailed electrical characterization of MOS capacitors build on such dielectric layers, before and after forming gas anneals, shows that these interface modifications can lead to a marked decrease of the smaller interface state peaks close to the edges of the bandgap, whereas the larger mid-gap peaks are barely touched. The results of atomistic modeling of the oxidation of a GaAs surface help to understand the origin of these mid-gap electronic states, which are responsible for the apparent pinning of the Fermi level.

The electron energy band alignment at interfaces of In x Ga 1− x As (0 ⩽ x ⩽ 0.53) with atomic-layer deposited insulators Al 2 O 3 and HfO 2 is characterized using combined measurements of internal photoemission of electrons and photoconductivity. The measured energy of the In x Ga 1− x As valence band top is found to be only marginally influenced by the semiconductor composition. This result suggests that the observed bandgap narrowing from 1.42 to 0.75 eV when the In content increases from 0 to 0.53 occurs mostly through downshift of the semiconductor conduction band bottom. Electron states originating from the interfacial oxidation of In x Ga 1− x As lead to reduction of the electron barrier at the semiconductor/oxide interface.

In this paper we compare the interface trap distributions D"i"t(E) of sulfur treated Al"2O"3/In"0"."5"3Ga"0"."4"7As interfaces, which underwent MOS capacitor and transistor fabrication processes. Lower interface trap densities were found close to the conduction band edge for both cases. The inversion channel MOSFET achieves high device performance despite the fact that its oxide-semiconductor interface quality is a notch below that of the MOS capacitor with optimized process.

Photoluminescence intensity (PLI) measurements of GaAs and InGaAs thin films indicate that InGaAs might be inherently easier to passivate than GaAs. The introduction of just 15% of In leads to a reduction of the surface recombination velocity at native oxide interfaces by an order of magnitude. This is more than the effect expected by a reduced bandgap alone. The PLI method applied to thin films can also be used to determine the surface recombination velocity of other IIIV-oxide interfaces.