Session J24: Focus Session: Dopants and Defects in Semiconductors - Si and III-V

Dislocation nucleation in Si -- effect of Ge and vacancies

The presence of dislocations in Si devices is detrimental for their function. However, dislocations may nucleate during device processing from interfaces and from imperfections of the surface. In this work we study the conditions under which this may take place. Specifically, the activation energy for dislocation nucleation in pure Si is evaluated using an atomistic model and considering a corner stress concentrator. At lower temperatures, 60$^{\circ}$ dislocations nucleate in the shuffle set as half-hexagonal loops. The activation energy depends on the step height and on the spacing between steps. The effect of Ge on nucleation is evaluated using the same model and it is observed that the energy barrier decreases slightly in presence of these substitutional atoms. The presence of vacancies in the glide plane also decreases the activation energy. A low concentration of vacancies (0.5{\%}) decreases the barrier for nucleation similarly with a large concentration of Ge (30{\%}).   

Surface Charge Distribution on Hydrogen-Passivated Silicon Measured by UHV Kelvin Probe Microscopy

A few years ago hydrogen-passivated silicon (H:Si) surfaces were used to observe high-mobility in FETs (100,000 cm2/Vs) [1]. This result is expected since chemical passivation of silicon produces an atomically flat surface with low density of defects. As chemical preparation can lead to creation of fixed charges on H:Si surface, one needs a method to get information about their distribution to further improve the quality of a process. This can be done by measuring electrostatic forces created by such charges with Kelvin Probe Microscopy (KPM), which obtains a voltage distribution map with a resolution of a few nm. We have used UHV KPM for our measurements, since UHV conditions are important for avoiding oxidation of H:Si surface as well as achieving its further depassivation/annealing. By annealing samples above depassivation temperature(~500C), we have obtained data that provides more information about depassivation effect on charges that come from chemical residue on hydrogen and silicon itself. Preliminary results show standard deviations of voltage much smaller (by an order of magnitude) than ones measured previously on silicon oxide, consistent with high quality expected from chemical preparation. \\[4pt] [1] Robert N. McFarland et al., Phys. Rev. B 80, 161310(R) (2009)   

Magnetic resonance characterization of silicon nanowires

Silicon nanowires (SiNWs) have been extensively investigated in the last decades. The interest in these nanostructures stems from both fundamental and applied research motivations. The functional properties of one- and zero-dimensional silicon structures are significantly different, at least below a certain critical dimension, from those well known in the bulk. The key and peculiar functional properties of SiNWs find applications in nanoelectronics, classical and quantum information processing and storage, optoelectronics, photovoltaics, thermoelectric, battery technology, nano-biotechnology, and neuroelectronics. We report our work on the characterization by continuous wave (CW) and pulse electron spin resonance (CW, FT-EPR) and electrically detected magnetic resonance (EDMR) measurements of silicon nanowires (SiNWs) produced by different top-down processes. SiNWs were fabricated starting from SOI wafers using standard e-beam lithography and anisotropic wet etching or by metal-assisted chemical etching. Further oxidation was used to reduce the wire cross section. Different EDMR implementations were used to address the electronic wave function of donors (P, As) and to characterize point defects at the SiNWs/SiO$_{2}$ interface.   

Ab initio study of confinement effects and hyperfine structure in chalcogen doped Silicon nanostructures

In recent years, increasing interest has been focused on Si nanostructures as building blocks for ultra-scaled electronic devices where quantum mechanic effects are relevant. We have investigated the atomic and electronic structures of a single chalcogen donor in H passivated Si nanostructures of different size by means of the plane-wave pseudopotential techniques we used in Ref. [1] to study Se doped Si (001) nanowires (NW). Our results showed an increase in the gap with diminishing diameter of Si NW. We studied the size dependence of electronic properties and hyperfine constant of single substitutive chalcogen impurity in Si-NWs (001) and (111) oriented, and Si-dots and their dependence on the distance from the NW axis or the centre of the dot. We show that the hyperfine parameters are strongly dependent on the impurity position: we proved that surface effects can lead to strong differences in the hyperfine parameters depending on the chalcogen location inside the nanowire, suggesting a way to determine experimentally the position of the defect on the basis of electron paramagnatic resonance spectra. Preliminary results on chalcogen doped Ge nanowires complete the work. [1] G. Petretto, A.Debernardi, and M. Fanciulli, Nano Letters, Vol. 11, 4509 (2011).   

Segregation and diffusion of boron dopants in the Si/SiO$_2$ interface

Boron dopants in metal-oxide-semiconductor field-effect transistors exhibit very peculiar behavior such as transient enhanced diffusion, clustering, and segregation. Especially, B segregation to the Si/SiO$_2$ interface significantly affects the dopant distribution and thereby the device performance. However, there is a lack of studies on the mechanism for B segregation and diffusion in the Si/SiO$_2$ interface. In this work, we perform first-principles density-functional calculations to understand how B dopants diffuse and segregate to SiO$_2$. We generate two Si/SiO$_2$ interface structures, in which crystalline alpha-quartz and amorphous SiO$_2$ are placed on Si. Among various B configurations, we find that an interstitial B is energetically more favorable in the oxide, compared with a subsitutional B and a self-interstitial-B complex in Si. We examine the effect of point defects such as a floating bond and an oxygen vacancy in SiO$_2$ on B segregation and also investigate B diffusion pathways across the Si/SiO$_2$ interface.   

Hybrid functional calculations for native defects and dangling bonds in $\alpha$-Al$_2$O$_3$

Al$_2$O$_3$ is a promising material for use as a gate dielectric in III-V-based MOS devices, including in GaN-based transistors. Recent developments indicate that despite the relatively high structural quality, the presence of charge traps and fixed-charge centers near the interface between the oxide and nitride still poses serious problems for device performance. Native defects and dangling bonds in the Al$_2$O$_3$ dielectric or in the vicinity of the interface are the most likely causes. To aid in the identification of these centers, we perform density functional calculations with a hybrid functional for point defects and dangling bonds in $\alpha$-Al$_2$O$_3$. We determine the position of the defect transition levels in the gap of the oxide, and analyze the level positions with respect to the nitride band edges. Our results show that O vacancies and Al dangling-bonds can produce charge traps and Al interstitials act as fixed charges in GaN-based $n$-MOSFETs.   

Role of oxygen-related defects in hafnia

Hafnia has recently received much attention because of its potential application as high-k dielectric material replacing silica in microelectronic devices. Point defects in hafnia - in particular oxygen vacancies and interstitials - can play an important role as traps or sources of fixed charge. In this study, we perform electronic structure calculations on oxygen-related defects in monoclinic hafnia using a combined density functional theory (DFT) and GW formalism. We have previously shown that upon including quasiparticle defect levels and the appropriate electrostatic corrections within a supercell calculation, this formalism corrects for the error in calculating formation energy and charge transition levels using standard DFT. In this study, we calculate the formation energy of these defects as a function of the Fermi level and the chemical potential of oxygen to determine which of these defects are most stable.   

Vibrational spectroscopy of cast Si used to fabricate solar cells: microscopic properties of nitrogen and oxygen impurities

Cast Si with grain sizes from a few mm to a few cm is commonly used for the fabrication of solar cells. Nitrogen impurities are introduced into cast Si by the SiNx coating of the crucible used for casting. Much is known about N and O centers in single-crystal Si used in microelectronics [1]. We have used vibrational; spectroscopy to probe the concentration and defect configurations of nitrogen centers in cast Si used to fabricate solar cells. The interaction of N with O impurities that are present has also been investigated. The dominant N center in cast Si is a N-N interstitial pair. N-O complexes are also formed. Which defect complexes are present depends on the impurity content of the multi-crystalline Si sample, which can vary widely, and its thermal history. [1] H. Ch. Alt and H. E. Wagner, J. Appl. Phys. \textbf{106}, 103511 (2009) and the references contained therein..   

The chemical trends of a new defect cluster: DDX centers

DX center is a major ``killer'' defect limiting n-type doping in group II-VI and III-V semiconductors. It converts a shallow donor to deep one, which is a major reason for the saturation of free-electron carriers in the doping process. Several structure models of isolated DX centers have been proposed in the literatures, such as the broken-bond model (BB-DX), and the $\alpha $ and $\beta $ cation-cation bond model (CCB-DX). All these DX centers can be stabilized with hydrostatic pressure or reduced dimensionality and size. In group III-V and II-VI semiconductors, it has been common believe that cation-site induced DX centers are easier to form than anion-site induced ones. Because DX centers trap an extra electron, therefore, another defect in the system must donate the electron and form a positive charged defect. We show, using GaAs as an example, that in heavily doped semiconductor, the negative charged DX center and positive charged donor can couple strongly through the Coulomb interaction, forming the dominant DDX center. The DDX centers are still deep level defects. However, unlike the DX center, the DDX centers have different chemical trends, i.e., anion-site DDX center is easier to form than cation-site DDX centers. A simple model is proposed to explain the new trends.   

Valence state manipulation of single Fe impurities in GaAs

Cross-sectional STM was used to characterize Fe doped GaAs. We show that, by controlling the tip-induced band bending, Fe atoms can be brought from their isoelectronic state (Fe3+)-3d5 state into their (Fe2+)-3d6 ionized acceptor state. This STM-induced valence manipulation that involves the (de)population of the d-shell of the Fe atom differs from our previous experiments on Mn-acceptors and Si-donors where we changed its charge state by adding or removing a valence band hole or a conduction band electron respectively that is bound by the impurity. In addition for specific tunneling conditions a peculiar contrast is observed under which Fe atoms appear as dark anisotropic features.   

Local Environment Effects on Single Zn and Mn Acceptors in Gallium Arsenide

The ability to precisely control the properties of single dopants in semiconductors is of interest not only for the improvement of current technologies but also for the development of next generation devices. Recent work in our group has exploited the single atom precision afforded by a scanning tunneling microscope to explore how the properties of dopants in gallium arsenide depend on their local environment. Specifically, we have shown that Zn dopants located within the same layer and occupying identical binding sites exhibit dissimilarities that are dependent upon the proximity to neighboring subsurface Zn acceptors.(1) $^{ }$Further, we demonstrated control of the ionization state of single Mn acceptors by both defect- and tip-induced band bending.(2) Finally, we achieved tunable control over the binding energy of Mn acceptors by varying the proximity to charged As vacancies.(3) This talk will review these findings and elaborate on the application of these techniques to characterize defects in wide bandgap materials, where the origin of properties like ferromagnetism are not yet well understood. 1. Appl. Phys. Lett. 99, 053124 (2011). 2. Nano Lett. 11, 2004 (2011). 3. Science 330, 6012 (2010).   

Recombination dynamics of excitons bound to nitrogen isoelectronic centers in GaAs

Using time-resolved photoluminescence, we have studied the radiative recombination dynamics of excitons bound to single nitrogen dyads in GaAs. For in-plane dyads of C$_{2v}$ symmetry considered in this work, the lifetime of all four allowed optical transitions, polarized either along or perpendicular to the dyad, decreases with temperature. Over the temperature range studied, the lifetimes of transitions of same polarization remain highly similar, but they are sensitively longer for transitions polarized along the dyad. These results indicate that 1) the spatial orientation of the exciton wavefunction is an important factor in the recombination dynamic, 2) the transfer dynamics between bright states appears negligible, and 3) the transfer to or from dark states is not significant for temperature over 5 K. These findings enhance the understandings of isoelectronic centers which are promising candidates for the implementation of atomic size memories based on a single charge or spin.   

First-principles study of InX (X=P,Sb) semiconductors

III-V semiconductors are gaining increasing interest for applications due to the possibility of engineering their electronic properties by choosing and combining different elements. Additional design parameters can come from confinement effects which have led to intensified research on nanowires for electronic applications. Among the lesser studied III-V semiconductors with large technological potential are indium-based compounds, where InSb with an extremely small band gap in the infrared range is the staple material for infrared detector devices and InP (with a larger band gap of 1.42 eV) is considered as a promising material for nanowire-based applications. For these materials, many basic questions that have been answered for more-mainstream semiconductors are still unanswered, these include effective masses, optical properties, and the influence of defects on their properties. In order to address some of these questions, we have performed an exhaustive exploration of the defect energetics of InP using first-principles calculations. We also report a detailed comparison of calculated effective masses and optical properties of InSb with experiments. We have used both GGA and HSE06 to treat exchange-correlations. This work was supported by NSF MRSEC DMR-0820414.   

Simple intrinsic defects in GaP and InP

To faithfully simulate evolution of defect chemistry and electrical response in irradiated semiconductor devices requires accurate defect reaction energies and energy levels. In III-Vs, good data is scarce, theory hampered by band gap and supercell problems. I apply density functional theory (DFT) to intrinsic defects in GaP and InP, predicting stable charge states, ground state configurations, defect energy levels, and identifying mobile species. The SeqQuest calculations incorporate rigorous charge boundary conditions removing supercell artifacts, demonstrated converged to the infinite limit. Computed defect levels are not limited by a band gap problem, despite Kohn-Sham gaps much smaller than the experimental gap. As in GaAs, [P.A. Schultz and O.A. von Lilienfeld, Modeling Simul. Mater. Sci. Eng. 17, 084007 (2009)], defects in GaP and InP exhibit great complexity---multitudes of charge states, bistabilities, and negative U systems---but show similarities to each other (and to GaAs). Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Company, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000.   

Defects in GaSb and InAs/GaSb superlattices

Unintentionally doped GaSb is known to be p-type. One possible explanation for the p-type conductivity is due to the existence of Ga on Sb anti-site defects that behave as acceptors. Such an acceptor-like defect state could potentially impact the performance of an IR detector based on a type II superlattice InAs/GaSb. We use pseudopotential density functional theory to investigate this defect state in both bulk GaSb and the superlattice. We calculate the defect levels with and without spin-orbit interaction and with the p-d band separation and bandgaps corrected. Although the defect might be acceptor-like, its energy level does not necessarily follow the GaSb band edge that is shifted in the superlattice due to quantum confinement.