Session S19: Focus Session: Dopants and Defects in Semiconductors III

Stoichiometry driven impurity configurations in compound semiconductors

As is well known, crystal growth of defect-free compound semiconductors, in contrast to elemental, is inherently limited by non-stoichiometry. High resolution infrared spectroscopy of localized vibrational modes can display unique signatures which reveal the structure of stoichiometry related defect-impurity complexes. The talk will focus on II-VI semiconductors in which group II cations are replaced with a group IIA or a 3d-transition metal ion as an impurity, on the one hand, and a group VI anion replaced with a group VIA impurity, on the other. Incorporation of O replacing Te with a full complement of nearest neighbor Cd's, i.e. $\mathrm{O_{Te}}$, as well as $\mathrm{O_{Te}}$ in association with a Cd vacancy ($\mathrm{V_{Cd}}$) in the zincblende CdTe result in defect centers with unique i.r. signatures. The occurrence of $\mathrm{O_{Te}}$ with $T_d$ symmetry and ($\mathrm{O_{Te}-V_{Cd}}$) with $C_{3v}$ symmetry can be controlled by favoring or suppressing Cd vacancies. In CdSe, with its wurtzite structure, oxygen incorporation occurs in two ways: in one, it is an \textquotedblleft anti-site\textquotedblright ~defect, $\mathrm{O_{Cd}}$, as revealed in its host isotope related fine structure; in the other, oxygen enters in association with Cd vacancies as ($\mathrm{O_{Se}-V_{Cd}}$). The talk will discuss the number of i.r. signatures specific to each center; their polarization characteristics (in CdSe); the striking temperature behavior of the i.r. signatures of ($\mathrm{O_{Te}-V_{Cd}}$) and ($\mathrm{O_{Se}-V_{Cd}}$); and the occurrence of overtones/combinations of the LVMs in CdTe. These investigations provide a wealth of microscopic insights into orientational degeneracy, host isotope effects and acquisition of the temperature averaged higher symmetries by the switchings of the dangling bond of $\mathrm{V_{Cd}}$.   

Carrier compensation in semi-insulating CdTe

Carrier compensation in semi-insulating CdTe has been attributed to the compensation of surplus shallow acceptors by deep donors, usually assumed to be Te antisites. However, our first-principles calculations show that intrinsic defects should not have a significant effect on the carrier compensation due either to lack of deep levels near midgap or to low defect concentration. We demonstrate that an extrinsic defect, O$_{Te}$-H complex, may play an important role in the carrier compensation in CdTe because of its amphoteric character and reasonably high concentration. Our findings have important consequences for improving device performance in CdTe-based radiation detectors.   

Gadolinium Doping in ZnTe

We have investigated, experimentally and theoretically, the effects of Gd doping on the structural and optical properties of ZnTe films grown by pulsed laser deposition. A few {\%} of Gd doping was found to \textit{reduce} the ZnTe lattice constant with \textit{no change} in the fundamental band gap. When the doping level is increased to $>$7{\%}, the lattice constant becomes more or less constant, but the band gap increases abruptly by as much as 50 meV. First principle calculations based on density functional theory using the linearized augmented plane wave method were performed using ZnTe supercells containing either isolated defects or defect complexes. The reduced lattice constant on Gd doping can be attributed to the presence of defect complexes involving substitutional Gd ions and neighboring vacancies. The insensitivity of the band gap at lower Gd concentration can be explained by self-compensation of these defects. The increase in the band gap energy at higher concentration is attributed to band-filling effect.   

First Principles Investigation of H in CdTe

CdTe and alloys particularly (Cd,Zn)Te are of interest for radiation detection and other applications. A key issue is obtaining high mobility compensated material with low concentrations of traps. Hydrogen has been shown to play an important role in various semiconductors with both beneficial and deleterious effects. We investigate the effect of H on vacancies in CdTe using supercells. Our results are obtained from the first principles density functional theory calculations using the full potential linearized augmented planewave method including local orbitals and based on local density approximation. H atoms in Cd vacancies move toward one of Te atoms, which form Td symmetry with Cd in the bulk system. The bond length between H and Te is 1.7 angstrom. This corresponds to a closely bonded Te-H unit similar to OH. We also present H in Te vacancies. We report the electronic structures as well as positron lifetimes. This work was supported by DOE, NA-22.   

Origin of doping bottleneck in semiconductor quantum dots

Doping difficulties in semiconductor nanocrystals have been observed and its origin is currently under debate. It is not clear whether this phenomenon is energetic or depends on the growth kinetics. Using first-principles method, we performed systematic study of defects (donors, acceptors, isovalent defects, etc.) in ZnSe quantum dots. we show that, the transition energies and defect formation energies of the donor and acceptor defects always increase as the quantum dot sizes decrease. However, for isovalent impurities the changes of the defect formation energies are rather small. Our study suggests that for donor and acceptor defects, the doping difficulty is mostly due to energetic effects, whereas for isovalent impurities, the doping difficulty is mostly due to kinetic effects. The origin of the calculated trends is explained using simple band-energy-level models.   

Increased binding energy of impurities near a semiconductor vacuum interface.

We have recently shown that a STM tip can be used as a tool to manipulate the charge state of a single impurity below the cleavage surface of a semiconductor. This manipulation allowed us to determine the binding energy of single donors and acceptors as a function of their depth (up to 1 nm) below the surface. We found that the binding energy strongly increases near the surface. In the case of a Si-donor in GaAs the binding energy increases continuously from 5.6 meV in the bulk to more than 100 meV close to the surface. Our STM techniques also allowed for the determination of the size and shape of the Coulomb field of single ionized donors. We found that the range of the potential is strongly reduced relative to the bulk value. Both the reduced range of the Coulomb potential and the increased binding energy can be related to a reduced dielectric constant and increased effective mass near the surface. We will discuss the implications of these findings.   

First-principles simulations of GaAs defects: The challenge of Ga pseudopotentials

Design of norm-conserving gallium pseudopotentials (PP) has been investigated for density functional theory calculations of defects in GaAs. A converged PP construction is described. We examined the performance in cubic zinc-blende structure phases of GaAs, GaP, and GaN. Computed lattice constants, bulk moduli, and, particularly, band gaps vary greatly depending on PP construction and exchange correlation functional. The Kohn Sham band gaps exhibit a distinctive sensitivity on lattice constant, direct with a strong near-linear dependence at larger lattice constants with crossover to indirect near (within 5\%) of the equilibrium lattice constant. Gradient-corrected functionals with a converged PP give a near-zero GaAs gap, a problem for defect calculations. A 3d-core PP ``fixes'' the band gap, but predicts GaAs defect formation energies different from converged 3d-valence PP. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000.   

Computational studies of a 2D tight-binding model of randomly dispersed hydrogenic centers

The impurity band arising from s-orbitals of randomly dispersed hydrogenic dopants in semiconductors in two dimensions is studied via exact diagonalization of a tight-binding model. Ensemble averaged density of states (DOS) and inverse participation ratio (IPR) of eigenstates are obtained as a function of energy at low to intermediate doping densities, where small clusters of sites are most significant. A similar calculation is done for p-orbitals for comparison. A strong peak in the DOS is seen about the impurity level. Increasing the density of sites weakens this peak and produces asymmetry in the DOS and the IPR. A nearest-neighbour pair approximation qualitatively explains several features in the DOS at low densities but does not reproduce the singularity. This motivates a comparison to a hierarchically constructed pair model, as well as random bipartite systems, which is pursued further via a renormalization group approach in a concurrent study.   

Quenched singularity in the density of states of 2D random hydrogenic systems

Delta-doped hydrogenic dopants in semiconductor heterostructures give rise to an impurity band which can be characterized by a two-dimensional tight-binding model with randomly positioned sites. At low densities, the density of states possesses a singularity about the impurity level, which can be understood in terms of the states of a hierarchically constructed set of impurity pairs. As the density is raised, this singularity is quenched due to further neighbors breaking the electron-hole symmetry. The quenching is accompanied by an asymmetry in the density of states, and pair approximations are insufficient to even qualitatively describe the system at higher densities. We motivate and outline a renormalization group technique that captures the quenched singularity and asymmetry in the density of states. This approach motivates the study of random bipartite systems [1], for which it is particularly suited. We compare the results of both types of systems.\newline [1] M. Inui, S. A. Trugman, and Elihu Abrahams, Phys. Rev. B 49, 3190 (1994).   

Negative magnetocapacitance and associated Schottky barrier height changes in lightly doped GaAs

We investigate the magnetic field response of Schottky barriers formed on GaAs with Si dopant densities of $\sim $3E16 cm$^{-3}$ and $\sim $9E16 cm$^{-3}$. Negative magnetocapacitance of up to 21{\%} at 20K and increasing Schottky barrier height as determined by various impedance measurement techniques are observed. We attribute these effects to a magnetic-field induced increase in the ionization energies of electrons bound to Si impurity atoms, causing shallow impurity electrons to hop to (ionization process) and from (capture process) the conduction band on longer time scales. The effective field at which these effects are seen is a factor of 10$^{5}$ smaller than it would be in vacuum because of the smaller effective mass and larger dielectric constant of the GaAs host. The dependence of interband hopping on magnetic field decreases the dipole response in the depletion region and gives a corresponding decrease in the measured capacitance. Magnetic field dependent Schottky barrier heights are inferred from linear 1/C$^{2}$ versus bias voltage plots. We note that these magnetic field dependent effects are occurring in the absence of magnetic impurities and thus need to be understood before characterizing the magnetic response of diluted magnetic semiconductors.   

Thermally activated persistent photoconductivity {\&} donor binding energy in high mobility AlAs QWs

In AlAs, valley index is important quantum number which can help understand interactions. However, important parameters for growth such as donor binding energy and Si $\delta $-doping efficiency were unknown and AlAs quantum wells (QWs) typically did not conduct in dark. We grew series of (001) and (110) oriented double-sided doped n-type AlAs QWs and deduced Si donor binding energy $\Delta $ in Al$_{0.45}$Ga$_{0.55}$ As and doping efficiency $\eta$. They work in dark possibly because dilute charge traps in substrate are screened by backside doping. From dark saturation density for doping series we deduced $\Delta_{dk}$=65.2 meV [1]. Our studies show thermally activated PPC where sample is illuminated at 4 K and returned to dark without appreciable density increase. As temperature is increased to 30 K, density doubles, indicating shallow binding energy $\Delta_{PIA}$=0 meV post-illumination anneal (PIA). Doping efficiency after illumination for (001) facet was found to be $\eta $=n$_{2D}$/n$_{Si}$=35{\%} and for (110) $\eta $=17{\%}. With this understanding, we designed (001) AlAs QW with PIA density n=2.4 x 10$^{11}$ cm$^{-2}$ and mobility $\mu $=4.3 x 10$^{5}$ cm$^{2}$/Vs(330 mK), improvement of almost an order of magnitude over published results. [1] Dasgupta, \textit{et al.} APL (2007)   

The low-temperature 2D mobility for metallic p-type GaAs Quantum Well

At T $<$ 1.2K the mobility $\mu$(T) is determined by charged trap ionized impurity scattering (iis) and T-dependent screening [1]. $\mu$(T) is calculated with $<$$\tau$(E)$>$ given by an empirical expression $\tau$ = $\tau$$_{o}$x/[x + C tanh($\eta$/2)] [x = E/kT, $\eta$ = T$_{F}$/T and a 2D DOS that features a pseudogap. $\mu$(T) exhibits a minimum at T$_{m}$ = T$_{F}$/2.25 and increases slowly for T $>$ T$_{m}$. The physical reason for this unusual increase in $\mu$(T) is explained. The coefficient C is directly related to $\mu$(0)/$\mu$(T$_{m}$ [4.0 $>$ ratio $>$ 3.6 for p-type GaAs data [2]]. The T-dependent screening $\kappa$$_{2}$(T) = s(T) $\kappa$$_{2}$(0) and s(T) is given by [$\mu$(T)-$\mu$$_{m}$]/ [$\mu$(0)-$\mu$$_{m}$]. This s(T) allows the determination of T* [d$\sigma$/dT = 0] where T* is slightly less than T$_{m}$. The data [2] is an example of ideal 2D behavior. The role of interactions for T $<$ T$_{m}$ and T $>$ T$_{m}$ will be discussed. [1] F. Stern, PRL 44, 1469 (1980); [2] X.P.A. Gao et al., PRL 93, 256402 (2004).