Silicon germanium interdiffusion in SiGe device fabrication: A calibrated TCAD model

Local modes of hydrogen defects in Si:Ge and Ge:Si

Depositions of Ge, SiGe, and Si atoms onto a Si(001) substrate are simulated by means of the classical molecular-dynamics method. The Tersoff potential is employed to calculate the interatomic interactions. In the first kind of simulation, Ge atoms with the velocity corresponding to an energy of 0.2 eV are deposited one by one onto the equilibrated substrate every 200 ps. In the second kind of simulation, Si and Ge atoms are alternately deposited one by one onto the substrate. In the third kind of simulation, Si atoms are deposited. In each simulation, breakdown of epitaxy is observed after several monolayers have been epitaxially deposited. It is concluded that breakdown of heteroepitaxy is delayed due to the increase of the potential energy of the system. (© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)

Ab initio study of Yb on the Ge(111)–(3×2) and Si(111)–(3×2) surfaces

An analytical model of hole confinement gate voltage range is derived for SiGe-channel p-MOSFETs and verified by SEDAN-3. The dependencies of hole confinement on threshold voltage, gate oxide and Si cap thicknesses, gate material, and Ge mole fraction are discussed. Various bulk and SOI SiGe p-MOSEETs are clarified to have the same hole confinement with threshold voltage adjustment.

Divacancy-related complexes in Si (1−x)Ge (x)

By using first-principles method in the density-functional theory, we clarify the atomic and electronic structures of silicene and germanene on 1×1 GaAs(111). We find stable structures for silicene and germanene on both the As-terminated and Ga-terminated GaAs surfaces. The structures of silicene and germanene are similar to those of the free-standing ones, which present a honeycomb-hexagonal geometry. The cohesive energies of silicene and germanene on both As and Ga sides of GaAs surfaces are comparable to those of their bulk structures and/or those on Ag(111) substrates which have been widely observed in experiment, showing the possibility of synthesizing them on both sides of GaAs surfaces in experiment. The corresponding binding energies are in a range of 0.56-1.37 eV per Si (Ge) atom, 10 times larger than the usual van der Waals interaction, showing the covalent interaction between silicene (germanene) and GaAs surfaces. The band structure calculations show that such a covalent interaction induces the absence of Dirac electrons for silicene and germanene on GaAs surfaces. We then explore the method of recovering the Dirac electrons by using hydrogen (H) intercalation. It is found that the intercalated H atoms are chemically bonded to GaAs surface, and the silicene (germanene) shifts upward distance from GaAs surface increasing from 2.50-2.58 Å to 3.49-3.86 Å, where a covalent van-der-Waals interaction transition happens between silicene (germanene) and GaAs surface. Moreover, the distances between silicene (germanene) and H atoms are 30% and 8% larger than the atomic-radius sum of Si (Ge) and H on As-terminated and Ga-terminated GaAs surfaces, respectively. This shows that the interaction between silicene (germanene) and H on the As-terminated GaAs surface is obviously weaker than the typical covalent interaction, while on the Ga-terminated GaAs surface, it is comparable to the typical covalent interaction. This difference is induced by the difference in electronegativity between As and Ga atoms. We further find that the H intercalation recovers the Dirac electrons well on the As-terminated GaAs(111) due to the weaker Si (Ge)-H interaction, while it does not on the Ga-terminated GaAs(111) due to the stronger Si (Ge)-H interaction. The results are confirmed by performing calculations for silicene (germanene) on larger GaAs(111) surfaces, i.e., the 3×3 GaAs surface. Our study provides the theoretical basis for the preparation and application of silicene and germanene on semiconductor surfaces.

Mixed Si/Ge compounds are of special interest as potential materials for photovoltaic applications. In order to evaluate the usage of soluble precursor compounds, we investigated the synthesis of heteroatomic nine-atom clusters that consist of Si and Ge atoms through dissolution of the ternary Zintl phases K12Si(17-x)Ge(x) (x = 9, 12) and Rb12Si(17-x)Ge(x) (x = 9). Electrospray ionization (ESI) mass spectrometry demonstrates the presence of mixed Si(9-x)Ge(x) clusters in acetonitrile solution. From ammonia solutions of the ternary phases, four compounds that contain 3-fold negatively charged [Si(9-x)Ge(x)](3-) clusters are obtained. The paramagnetic behavior is confirmed by EPR spectroscopy. [E9](3-) Zintl clusters are considered as intermediate structures in the stepwise oxidation of [E9](4-) clusters to novel element allotropes (E = Si-Pb). The structure of Rb[Rb-crypt]2[Si2.3(1)Ge6.7(1)](NH3)7 and the isostructural structures of [Rb-crypt]3[Si2.2(1)Ge6.8(1)](NH3)8, [K-crypt]3[Si2.4(1)Ge6.6(1)](NH3)8.5, and [K-crypt]3[Si4.6(1)Ge4.4(1)](NH3)8.5 are investigated by single-crystal X-ray diffraction (crypt = 4,7,13,16,21,24-hexaoxa-1,10-diazabicyclo[8.8.8]-hexacosane). The Si/Ge ratio of the products correlates with the composition of the ternary precursor phases.

In this work, we address issues pertinent to the understanding of the structural and electronic properties of ${\mathrm{Si}}_{1\ensuremath{-}x}{\mathrm{Ge}}_{x}$ alloys, namely, (i) how does the lattice constant mismatch between bulk Si and bulk Ge manifest itself in the alloy system? and (ii) what are the relevant strain release mechanisms? To provide answers to these questions, we have carried out an in-depth study of the changes in the local geometric and electronic structures arising from the strain relaxation in ${\mathrm{Si}}_{1\ensuremath{-}x}{\mathrm{Ge}}_{x}$ alloys. We first compute the optimized lattice constant for different compositions (x) by fully relaxing the system and by minimizing the total energy with respect to the lattice constant at each composition, using an ab initio molecular dynamics scheme. The optimized lattice constant, while exhibiting a general trend of linear dependence on the composition (Vegard's law), shows a negative deviation from Vegard's law in the vicinity of $x=0.5.$ We delineate the mechanisms responsible for each one of the above features. We show that the radial-strain relaxation through bond stretching is responsible for the overall trend of linear dependence of the lattice constant on the composition. On the other hand, the negative deviation from Vegard's law is shown to arise from the angular-strain relaxation. More specifically, the combined effect of the local bond-angle deviations from the tetrahedral angle and the magnitudes of the corresponding peaks for the partial-angle distribution function determines the negative deviation from Vegard's law. The electronic origin of the changes in the local geometric structure due to strain relaxation is also presented in this work. In particular, the correlation between the bond charges and the bond-lengths for Si-Si, Ge-Ge, and Si-Ge pairs in ${\mathrm{Si}}_{1\ensuremath{-}x}{\mathrm{Ge}}_{x}$ alloys for different compositions is explicitly shown. Our calculation of the average coordination number as a function of composition indicates a random occupation of Si and Ge on the lattice sites, suggesting that Si and Ge atoms are fully miscible in the alloy system.

Si–Ge single crystals, 7 mm in diameter and about 20 mm long with compositions ranging from 0 to 50 at.% Si were grown by the traveling solvent method. The compositions of the crystals were shown to be uniform parallel and perpendicular to the growth axis using EPMA measurement. Raman spectra indicated that the ratio of the numbers of Si–Ge and Ge–Ge bonds agreed with that expected for a random distribution of Si and Ge atoms. The band gap energy of the alloy increased linearly up to 15 at.% Si and increased gradually above 15 at.% Si. From photoluminescence measurements done under uniaxial stress along the <111> direction, it was shown that the dependence of excitonic no-phonon spectrum on the stress changed at about 15 at.% Si. These results indicated that the position of the conduction band minimum of Si–Ge alloy in the k-space changed at about 15 at.% Si.

Local-symmetry effect on light emissivity from SiGe quantum wells

The positions of Ge atoms intermixed in the Si(100) surface at very low concentration are identified using empty-state imaging in scanning tunneling microscopy. A measurable degree of place exchange occurs at temperatures as low as 330 K. Contrary to earlier conclusions, good differentiation between Si atoms and Ge atoms can be achieved by proper imaging conditions.

Raman spectroscopy is often used to determine the composition and strain in Si1-xGex films and structures. It is a fast, non-destructive technique that can give results with a spatial resolution at the micron-scale, and even information on nm-sized structures. In principle, the intensity and frequency of the three Raman peaks associated with vibrations of Ge-Ge, Si-Ge and Si-Si bonds offer information on strain and composition. However, this is only straightforward for compositions where all three, or minimal two of these three Raman peaks can be detected, typically between x = 0.2 and 0.8. And even for that case, although there is consensus about the equations to be used, various publications give different values for the constants to be used in these equations. In addition, the Si-Ge Raman peak makes the evaluation difficult, because its position and shape depends on the number and position of the nearest neighbors of a Ge or Si atom in a Si or Ge host lattice, respectively. There is not one, but there are several different kinds of Si-Ge Raman peaks. If less than 1% Si atoms are present in Ge, there is a localized Si phonon mode related to vibrations of an isolated Si atom surrounded entirely by Ge atoms, and a different method has to be used in Raman spectroscopy to obtain information on composition and strain (see figure). This talk discusses these challenges and proposes potential solutions. Figure 1

The sustained advancement in semiconducting core-shell nanowire technology has unlocked a tantalizing route for making next generation field effect transistor (FET). Understanding how to control carrier mobility of these nanowire channels by applying a gate field is the key to developing a high performance FET. Herein, we have identified the switching mechanism responsible for the superior performance of a Si-Ge core-shell nanowire quantum dot FET over its homogeneous Si counterpart. A quantum transport approach is used to investigate the gate-field modulated switching behavior in electronic current for ultranarrow Si and Si-Ge core-shell nanowire quantum dot FETs. Our calculations reveal that for the ON state, the gate-field induced transverse localization of the wave function restricts the carrier transport to the outer (shell) layer with the pz orbitals providing the pathway for tunneling of electrons in the channels. The higher ON state current in the Si-Ge core-shell nanowire FET is attributed to the pz orbitals that are distributed over the entire channel; in the case of Si nanowire, the participating pz orbital is restricted to a few Si atoms in the channel resulting in a smaller tunneling current. Within the gate bias range considered here, the transconductance is found to be substantially higher in the case of a Si-Ge core-shell nanowire FET than in a Si nanowire FET, which suggests a much higher mobility in the Si-Ge nanowire device.

Electroluminescence and photoconductivity of GeSi heterostructures with self-assembled islands in the wavelength range 1.3–1.55 μm

Point defects in semiconductors play a decisive role for the functionality of semiconductors. A detailed, quantitative understanding of diffusion and defect reactions of dopants is required for advanced modelling of modern nanometer size electronic devices. With isotope heterostructures which consist of epitaxial layers of isotopically pure and deliberately mixed stable isotopes, we have studied the simultaneous self- and dopant diffusion in several major semiconductors such as silicon and germanium. Detailed analysis of the simultaneous diffusion of self- and dopant atoms in Si and Ge yields information about the ionization levels of native defects and about dopant-defect interactions in Si and Ge. The results of these diffusion studies are highlighted in this work.