Midgap Energy Research Articles

Voids in Silicon by He Implantation: From Basic to Applications

The mechanism of bubble formation when He is implanted into silicon is described. Many experiments are reviewed and several techniques are considered. During implantation and subsequent annealing, complex Hen–Vm clusters are formed, trapping vacancies, while Si self-interstitials recombine directly at the surface. By increasing temperature He atoms out-diffuse, and the entire process produces a supersaturation of vacancies (void formation). Their evolution is studied during isothermal and isochronal annealing, describing the mechanisms involved; that is, direct coalescence or Ostwald ripening. The internal surface is an efficient trap for self-interstitials and for metals. The gettering mechanism is governed by a surface adsorption at low impurity concentration while at high value a silicide phase is observed. The high getter capability is ensured by the large number of traps introduced (1017–1019 cm−3). Finally, voids introduce mid gap energy levels that act as minority carrier recombination centers, providing a powerful method to control lifetime locally in silicon devices. The reviewed results demonstrate that the trap levels are due to the dangling bonds present on the void surface. This property can be used in many applications.

Impact of neutron irradiation on optical performance of InGaAsP laser diodes

Results are presented of an extended study on the degradation and recovery behavior of optical and electrical performance and on induced lattice defects of 1.3 μm InGaAsP double channel planar buried heterostructure laser diodes with an In0.76Ga0.24As0.55P0.45 multi-quantum well active region, subjected to a 1 MeV fast neutron and 1 MeV electron irradiation. The degradation of the device performance increases with increasing fluence. Two hole capture traps with near midgap energy level in the In0.76Ga0.24As0.55P0.45 multi-quantum well active region are observed after a 1×1016 n/cm2 irradiation. These deep levels are thought to be associated with a Ga vacancy. The decrease of optical power is related to the induced lattice defects, leading to a reduction of the non-radiative recombination lifetime and of the carrier mobility due to scattering. The difference in radiation damage between 1 MeV fast neutrons and 1 MeV electrons is discussed taking into account the non-ionizing energy loss (NIEL). The radiation source dependence of performance degradation is attributed to the difference of mass and the probability of nuclear collision for the formation of lattice defects. The decreased optical power recovers by thermal annealing, and the recovery increases with increasing annealing temperature. The optical power recovers by 45% for 1 MeV neutron irradiation with a fluence of 1×1015 n/cm2 after a 300°C annealing.

Radiation-induced lattice defects in InGaAsP laser diodes and their effects on device performance

Results are presented of an extended study on the degradation and recovery behavior of optical and electrical performance and on induced lattice defects of 1.3μm InGaAsP double channel planar buried heterostructure laser diodes with an In0.76Ga0.24As0.55P0.45 multi-quantum well active region, subjected to 1-MeV fast neutron and 1-MeV electron irradiation. The degradation of the device performance increases with increasing fluence. Two hole capture traps with near midgap energy level in the In0.76Ga0.24As0.55P0.45 multi-quantum well active region are observed after 1×1016n/cm2 irradiation. These deep levels are thought to be associated with a Ga-vacancy. The decrease of optical power is related to the induced lattice defects, leading to reduction of the non-radiative recombination lifetime and of the carrier mobility due to scattering.

Electronic structure of doped fourfold coordinated amorphous semiconductors. Midgap states in amorphous carbon

Midgap states have been studied in phosphorus-doped, tetrahedrally coordinated amorphous carbon by means of the Fragment Self-Consistent Field method (FSCF). It has been found that the midgap energy levels are primarily determined by the relative position of dopants. A simple linear correlation has been obtained between the highest occupied molecular orbital (HOMO) energy of the model clusters and the inverse distance of impurities.

A study of various oxide/silicon interfaces by Ar+ backsurface bombardment

A low-energy (550 eV) argon beam is used to bombard the backsurfaces of 6 kinds of metal–oxide–semiconductor capacitors, and the resulting effects on their interface characteristics are then investigated. The gate oxide of these capacitors includes thermal oxide, trichloroethyene (TCE) oxide, NH3-nitrided oxide, reoxidized-nitrided oxide, rapid-thermal-nitrided oxide, and N2O-nitrided oxide. Measurements show that for bombardment times up to 45 min the interface-state density of all the devices, in general, decreases with increasing bombardment time/dose, and the midgap energy at the silicon surface tends to rise. Moreover, the bombardment is more effective in reducing acceptor-type than donor-type interface states. On the other hand, the change of fixed-charge density is more complex. For TCE, N2O-nitrided and reoxidized-nitrided oxides, fixed-charge density decreases initially with increasing bombardment time, but then increases, while the trend is reversed for the other gate oxides. A model with stress compensation and weak bond breaking is suggested to explain the results.

Theory of dopant pairs in four-fold coordinated amorphous semiconductors

The position of energy levels belonging to dopant pairs was investigated in substitutionally p- or n-type doped four-fold coordinated amorphous carbon and silicon by means of Hartree–Fock ab initio and the Fragment Self-Consistent Field method. Our models contain 45 to 583 carbon or silicon atoms. Boron, phosphorus and nitrogen impurities have been incorporated into the amorphous networks. It has been found that the position of midgap states are primarily determined by the separation of the impurity atoms, and the role of the random network is only subordinate. A general relationship is proposed for the determination of midgap energy levels as a function of the distance between dopant pairs investigated for amorphous carbon and silicon.

Charged bosons in a doped Mott insulator: Electronic properties of domain-wall solitons and meron vortices

We demonstrate from first principles that when a charge carrier is added to a $\mathrm{spin}\ensuremath{-}1/2$ antiferromagnetic Mott insulator in either a one- or two-dimensional lattice, the self-consistent, Hartree-Fock ground state consists of a magnetic soliton texture with a doubly degenerate electronic level at the center of the Mott-Hubbard charge gap. This model is appropriate to systems with weak interchain or interlayer magnetic couplings in which long-range antiferromagnetic order is observed in the absence of charge carriers (doping). These magnetic solitons mediate the destruction of the magnetic order as the charge carrier concentration is increased. In a one-dimensional lattice with nearest-neighbor hopping $t$, on-site Coulomb repulsion $U,$ and self-consistent, antiferromagnetic moment amplitude $S$, we find that a charged, fermionic, magnetic domain wall soliton with a weakly ferromagnetic core, centered between two sites, has lower Hartree-Fock energy than a corresponding charged quasiparticle in one of the Mott-Hubbard bands. However, for $US/tg2$, this soliton is unstable to the formation of a lower energy charged, bosonic domain wall soliton, centered on a single site. For $US/tl2$, both of the above solitons are charged bosons. The self-consistent structure of these solitons exhibits no rotation of the local magnetic moments, but only a local suppression of the local moment amplitude in the vicinity of the hole. In the absence of doping, charge neutral domain wall solitons exhibit spin rotation within their core region. The equilibrium core size $\ensuremath{\rho}$ is determined by the degree of magnetic anisotropy. The ferromagnetic core soliton exhibits a pair of nondegenerate near-midgap electronic states. The antiferromagnetic core soliton exhibits a pair of nondegenerate electronic states that are symmetric about the midgap energy and that merge into the continuum as the anisotropy effects are made small and the soliton core radius $\ensuremath{\rho}$ becomes very large. The two-dimensional antiferromagnetic Mott insulator exhibits analogous behavior to the one-dimensional model. This analogy is precise for a 2D antiferromagnet exhibiting spin flux. For the undoped Mott insulator, the ferromagnetic core meron vortex (``lotus flower'' configuration of local magnetic moments) exhibits a doubly degenerate electronic midgap state in the continuum model and is the analog of the 1D neutral domain wall soliton. We demonstrate that a hole added to the 2D system can form a charged bosonic collective excitation, in which the spin background around the hole forms a planar vortex with local antiferromagnetic correlations at infinity and vanishing local-moment amplitude at the core of the soliton.

Tunneling spectrum into ad-wave superconductor with a time-reversal-symmetry-broken surface state

Within a Blonder-Tinkham-Klapwijk scattering formalism, the effects of a broken-time-reversal-symmetry (BTRS) surface pairing state on the transport properties of a normal-metal--$d$-wave-superconductor junction have been studied by solving the Bogoliubov--de Gennes equations. It has been shown that the zero-bias conductance peak (ZBCP) exhibited in the pure $d$-wave case is splitted due to the shift of the midgap state energy in the presence of the BTRS state. The splitting of the ZBCP is mainly determined by the strength of the subdominant $s$-wave componenet breaking the time-reversal symmetry. If the surface region extends several superconducting coherence lengths, the subgap resonances in conductance take place. Very interestingly, the shift of the subgap resonance peaks is determined by the relative strength of two components of the BTRS surface order parameter, regardless of the absolute strength of them.

Electronic transmission across a metal /conjugated-oligomer/metal structure: role of intrinsic disorder

We have studied electronic transmission across metal/conjugated-oligomer/metal structures: (a) one oligomer chain, and (b) two oligomer chains sandwiched between metal contacts, in the presence of lattice fluctuations. The lattice fluctuations are approximated by static white noise disorder. Resonant transmission occurs when the energy of an incoming electron coincides with a discrete electronic level of the oligomer. Because of disorder there is an enhancement of electronic transmission for energies that lie within the electronic gap of the oligomer. If fluctuations are sufficiently strong, a transmission peak within the gap is found at the midgap energy E = 0 for degenerate conjugated oligomers. For the two-chain case the spatial mirror symmetry is broken and coherence between the wavefunctions of two chains is partly lost when fluctuations are introduced.

Effects of lattice fluctuations on electronic transmission in metal/conjugated-oligomer/metal structures

The electronic transmission across metal/conjugated-oligomer/metal structures in the presence of lattice fluctuations is studied for short oligomer chains. The lattice fluctuations are approximated by static white noise disorder. Resonant transmission occurs when the energy of an incoming electron coincides with a discrete electronic level of the oligomer. The corresponding transmission peak diminishes in intensity with increasing disorder strength. Because of disorder there is an enhancement of the electronic transmission for energies that lie within the electronic gap of the oligomer. If fluctuations are sufficiently strong, a transmission peak within the gap is found at the midgap energy $E=0$ for degenerate conjugated oligomers (e.g., trans-polyacetylene) and $E\ensuremath{\ne}0$ for $\mathrm{AB}$-type degenerate oligomers. These results can be interpreted in terms of soliton-antisoliton states created by lattice fluctuations.

Impurity levels in phosphorus- and boron-doped amorphous silicon

Abstract The AMI semiempirical quantum-chemical method has been used to investigate phosphorus and boron-doped amorphous silicon. A simple relationship for the determination of midgap energy levels is proposed for substitutionally doped amorphous silicon as a function of distances between dopants.

Midgap states in nitrogen doped diamond-like amorphous carbon

Ab initio and semiempirical quantum chemical methods have been used to investigate the effect of nitrogen doping in amorphous carbon. We propose a simple relationship for the determination of midgap energy levels in substitutionally doped diamond-like amorphous carbon in function of distances between N atoms.

Electrical and Optical Properties of Nano-Crystalline GaN and Nano-Crystalline GaN:H thin Films

ABSTRACTNano-crystalline GaN (nc-GaN) and hydrogenated nc-GaN (nc-GaN:H) thin films and thin film transistors (TFT) prepared by a reactive sputtering method have been studied. Hydrogen incorporation in nc-GaN film induces localized states at mid-gap energy. Thermal annealing at 400 °C and 600 °C creates mid-gap states which is detectable by electron spin resonance. Further thermal annealing treatment at 800 °C reduces the deep states in nc-GaN and nc-GaN:H. Photoluminescence spectra of the nc-GaN film have two broad peaks at 2.4 eV and 3.2 eV. The source-drain current voltage characteristics of the nc-GaN TFT is demonstrated for the first time. The obtained field effect mobility is 10−4 cm2/V-s. Thermal annealing at 800 °C improves the field effect mobility to 10−2 cm2/V-s.

Structural and optical properties of strain-relaxed InAsP/InP heterostructures grown by metalorganic vapor phase epitaxy on InP(001) using tertiarybutylarsine

A combination of transmission electron microscopy and high-resolution x-ray diffraction analyses has been used to determine the exact strain in each layer of InAsP/InP multiple-quantum-well structures grown by metalorganic vapor phase epitaxy on InP(001) using trimethylindium, tertiarybutylarsine, and phosphine as precursors. The strain-relaxed structures are characterized by misfit dislocations located exclusively at (i) the interface between the buffer layer and the multilayer, and (ii) the interface between the multilayer and the cap layer. The low-temperature optical absorption spectra show well resolved excitonic transitions that are significantly shifted by strain relaxation. The spectra are analyzed with a solution to the Schrödinger equation in the envelope function formalism using the Bastard–Marzin model. The energies for the major transitions involving light- and heavy-holes are predicted accurately for all samples, allowing the determination of the heterojunction band offset. The heavy- and light-hole exciton binding energies deduced from that analysis range from 5 to 7 meV and 2 to 5 meV, respectively. The absolute values of the conduction band offset (expressed in meV) are consistent with the predictions of the quantum dipole model [J. Tersoff, Phys. Rev. B 30, 4874 (1984)] when calculating the midgap energy using a linear interpolation for the InAsP ternary composition between the values for the binaries InAs and InP. The absolute value of conduction band offset (in meV), which is dictated by the composition of the ternary layer, does not significantly depend on the degree of strain relaxation of the multilayer. However, the effect of this strain-relaxation on the InP and InAsP band gaps causes the conduction band offset to apparently increase from 72 to 82% of the partially-strained band gap difference when the strain-relaxation increases from 0 to 17%.

Influence of backsurface argon bombardment on SiO2–Si interface characteristics

A low-energy (550 eV) argon-ion beam was used to directly bombard the backsurface of polysilicon-gate metal-oxide-semiconductor (MOS) capacitors after the completion of all conventional processing steps. The interface characteristics of the MOS capacitors were investigated. The results show that, as the bombardment dose increases, the active dopant concentration near the oxide-semiconductor interface gets higher; maximum midgap energy increases; and interface-state density becomes lower. This simple technique is compatible with existing integrated-circuit processing, and can easily improve the interface characteristics, and therefore the electrical characteristics of MOS devices.

Zinc sulfide on GaP(110): Characterization of epitaxial growth and electronic structure

Epitaxial layers of ZnS were grown on cleaved GaP(110) surfaces by molecular beam epitaxy in an ultrahigh vacuum photoelectron spectrometer. The growth mode and the structure of the overlayer were studied by means of low-energy electron diffraction (LEED) and core as well as valence level photoemission using synchrotron radiation. The attenuation of substrate core-level intensities with ZnS deposition indicate layerwise growth. LEED demonstrates the growth of the cubic (zinc-blende) phase as expected for substrate-stabilized growth. A minor interface reaction is evident from changes in the appearance of the substrate (Ga 3d) and overlayer (S 2p) core levels with increasing thickness. S–Ga bonding was observed in a thin interfacial layer. The valence band offset for this lattice-matched heterojunction interface system was determined, and found to be of the straddling type (type I); its magnitude is in agreement with predictions based on the dielectric midgap energy model.

Postgrowth annealing effects of TiO2 thin films grown on InP substrate at low-temperature by metal-organic chemical-vapor deposition

The structural and electrical properties of titanium dioxide (TiO2) thin films grown on n-type InP(100) substrate by low-pressure metal-organic chemical-vapor deposition have been studied with postannealing. The thin films of TiO2 were deposited at a low temperature of 350 °C using titanium isopropoxide and oxygen. After a postgrowth annealing by the rapid thermal annealing method at a temperature of 850 °C for 15 s, the TiO2/InP structure of only the anatase phase with (101) and (200) peaks was observed by x-ray diffraction analysis. No interface reaction between TiO2 and InP was detected by Auger electron spectroscopy depth profiling. From capacitance–voltage measurement of the Al/TiO2/n-InP structure, the interface density of states at midgap energy and the dielectric constant were approximately low 1012 eV−1 cm−2 at midgap energy and about 50, respectively.

Changes of the occupied density of defect states of a-Si:H upon illumination.

We study the light-induced transient changes of the near surface density of occupied states g(E) of undoped and boron-doped a-Si:H with photomodulated total photoelectron yield spectroscopy. The data show an increase of g(E) upon illumination between 0.35 eV above ${\mathit{E}}_{\mathit{F}}$ and 0.7 eV below ${\mathit{E}}_{\mathit{F}}$ (towards the valence band) and a decrease in the region of deep valence-band-tail states. The difference signal depends sublinearly on the laser intensity and reaches a maximum of \ensuremath{\Delta}g\ensuremath{\approxeq}${10}^{17}$ ${\mathrm{cm}}^{\mathrm{\ensuremath{-}}3}$ ${\mathrm{eV}}^{\mathrm{\ensuremath{-}}1}$ at a laser intensity of 30 mW ${\mathrm{cm}}^{\mathrm{\ensuremath{-}}2}$ (\ensuremath{\lambda}=532 nm). Time-resolved measurements reveal rise and decay times of the order of milliseconds. The experimental results are explained quantitatively by a recombination model. In the framework of this model, a range of deep defects around mid-gap energy are singly occupied and neutral at probe-light intensities. Additional illumination with a laser leads to double occupation of these defects and a decrease of the valence-band-tail occupation. \textcopyright{} 1996 The American Physical Society.

Surface Recombination in Semiconductors

We propose two general criteria for a surface defect state to act as an efficient, nonradiative recombination center. The first is that the thermal ionization energy should not deviate from the mid-gap energy by more than the relaxation energy of the defect, In this case the activation energy for the recombination is given by the barrier for the capture of the first carrier, whereas the second carrier is captured athermally. The second citerion is related to the position of the average dangling bond energy relative to the band edges. If, as in the cases of InP or InAs, it is located close to a band edge, a low surface recombination velocity is expected. However a much faster recombination is predicated and experimentally observed in the materials with the average dangling bond energy located close to the mid-gap. The relevance of these criteria for the novel wide-gap optoelectronic materials is discussed.

Zincblende–CdSe on GaSb(110): Characterization of epitaxial growth and electronic structure

Substrate-stabilized pseudomorphic growth offers the chance to study the electronic structure of a particular semiconductor in different crystal structures, and to investigate the influence of structural differences on bulk and surface states. We have grown layers of CdSe in the zincblende modification on cleaved GaSb(110) surfaces by molecular beam epitaxy. The growth mode and structure of the overlayer were studied by means of low energy electron diffraction and photoemission using synchrotron radiation. The attenuation of substrate core level intensities with CdSe deposition indicates layerwise growth. Interface reaction leads to the liberation of Sb, which floats on the growth front, and the formation of a Ga–Se compound, as signaled by changes in substrate and overlayer core level line shape. The valence band offset for this lattice-matched interface system is 1.09 eV, such that the heterojunction is of the staggered type, in agreement with predictions based on the dielectric midgap energy model.