Search

${\mathrm{Sb}}_{2}{\mathrm{S}}_{3}$ has attracted great attention recently as a prospective solar cell absorber material. In this work, intrinsic defects, dopants, and their complexes in ${\mathrm{Sb}}_{2}{\mathrm{S}}_{3}$ are systematically investigated by using hybrid functional theory. ${\mathrm{V}}_{\mathrm{Sb}}$ and ${\mathrm{V}}_{\mathrm{S}}$ are dominant native defects and pin the Fermi level near the midgap, which is consistent with the high resistivity observed experimentally. Both ${\mathrm{V}}_{\mathrm{Sb}}$ and ${\mathrm{V}}_{\mathrm{S}}$ introduce deep levels inside the band gap, which can trap free carriers. Our calculated deep transition levels of ${\mathrm{V}}_{\mathrm{Sb}}$ and ${\mathrm{Sb}}_{\mathrm{S}}$ are consistent well with the results of the deep-level transient spectroscopy measurement. We further study dopants (including Cu, Ti, Zn, Br, and Cl) in ${\mathrm{Sb}}_{2}{\mathrm{S}}_{3}$ and find that Zn and Br/Cl are shallow acceptors and donors, respectively, which may be used to control the carrier and trap densities in ${\mathrm{Sb}}_{2}{\mathrm{S}}_{3}$. In addition, the defect complexes, i.e., $\mathrm{Cu}{(\mathrm{Zn})}_{\mathrm{Sb}}+{\mathrm{V}}_{\mathrm{S}}$ and $\mathrm{Cl}{(\mathrm{Br})}_{\mathrm{S}}+{\mathrm{V}}_{\mathrm{Sb}}$ are also investigated. The interaction between the donor and acceptor defects makes the defect levels of complexes shallower and less detrimental to carrier transport.

Monolithically integrated $\mathrm{Ga}\mathrm{As}$ $p$-$i$-$n$ diodes are demonstrated on 300-mm $\mathrm{Si}$ $(001)$ substrates using a nanoridge-engineering approach. Deep-level transient spectroscopy (DLTS) is used to perform defect analysis for nanoridge and planar $\mathrm{Ga}\mathrm{As}$ diodes. The point defect, EL2 with ${N}_{T}\ensuremath{\simeq}3\ifmmode\times\else\texttimes\fi{}{10}^{14}\phantom{\rule{0.2em}{0ex}}{\mathrm{cm}}^{\ensuremath{-}3}$, is observed for nanoridge $p$-$i$-$n$ diodes. A methodology is developed to extract the surface-state density (${N}_{\mathrm{SS}}$) directly from the DLTS spectrum. $\mathrm{Ga}\mathrm{As}$ nanoridge diodes show ${N}_{\mathrm{SS}}\ensuremath{\simeq}2\ifmmode\times\else\texttimes\fi{}{10}^{13}\phantom{\rule{0.2em}{0ex}}{\mathrm{cm}}^{\ensuremath{-}2}$ compared to planar diode approximately $6.5\ifmmode\times\else\texttimes\fi{}{10}^{12}\phantom{\rule{0.2em}{0ex}}{\mathrm{cm}}^{\ensuremath{-}2}$. A clear correlation is observed between dark current and defect density. An investigation on the impact of an in situ and ex situ passivation layers on the leakage current reduction is performed for $\mathrm{Ga}\mathrm{As}$ $p$-$i$-$n$ diodes.

We have considered five different models of charge transfer among coupled defect states in semiconductors where the free-carrier density is limited by the density of unoccupied trap levels, as in the case of defectdominated materials. To determine the time dependence of the trap occupancy features, we formulate a set of coupled differential equations that govern charge capture and emission processes for two defect states. A numerical solution assuming model parameters for traps provides features of the trap occupancy as a function of time. A critical comparison is made in occupancy features for different models, primarily categorized as serial ~hierarchical! and parallel mechanisms of charge transfer. The model predictions are successfully applied to a study of trapping kinetics of defects observed in heavily damaged n-type silicon. We show that, in addition to the occurrence of charge redistribution among multiple traps, the major trap in the damaged silicon exists in two metastable configurations, perhaps with negative U ~Hubbard correlation energy!, and the stable configuration refers to a midgap compensating center related to a small cluster of self-interstitials. The applicability of our model simulations can be extended to more complex defect systems using a combination of these simple models.

The energy barrier at InAs/GaAs interface due to the built-in strain in self-organized system has been determined experimentally. Such a barrier has been predicted by previous theories. From the deep-level transient spectroscopy (DLTS) measurements, we have obtained the electron and hole energy levels of quantum dots E-e(QD-->GaAs) = 0.13 eV and E-h(QD-->GaAs) = 0.09 eV relative to the bulk unstrained GaAs band edges E-c and E-v. DLTS measurements have also provided evidence to the existence of the capture barriers of quantum dots for electron E-eB = 0.30 eV and hole E-hB = 0.26 eV. The barriers can be explained by the apexes appearing in the interface between InAs and GaAs caused by strain. Combining the photoluminescence results, the band structures of InAs and GaAs have been determined.

The interaction of atomic hydrogen with substitutional palladium impurities is studied in nand p-type Si by deep-level transient spectroscopy. After wet-chemical etching, we determine seven different electrically active and at least one passive palladium hydrogen complex. The levels belong to Pd complexes with different number of hydrogen atoms. The PdH1 complex exhibits one level E(200) at EC20.43 eV. PdH2 has two levels E(60) at EC20.10 eV and H~280! at EV10.55 eV. Four levels are assigned to the PdH3 complex E(160) at EC20.29 eV, H~140! at EV10.23 eV, H~55! at EV10.08 eV, and H~45! at EV10.07 eV. An electrically passive complex is associated with a PdH4 complex. There is great similarity with the correspondent complexes in Pt-doped Si. Annealing above 650 K destroys all hydrogen related complexes and restores the original substitutional Pd concentration.

Persistent photoconductivity is studied in ZnSe/(Zn,Cd,Mn)Se two-dimensional electron-gas structures as a function of Mn concentration. Measurements of spectral photoconductivity, photoinduced transient spectroscopy, and deep-level transient Fourier spectroscopy show the presence of a deep trap located 2.1 eV below the conduction band at near room temperature. The buildup of persistent photoconductivity and its decay kinetics are formulated in accordance with the large lattice relaxation model. Implications of the presence of Mn in the two-dimensional electron gas are discussed.

Proton-irradiation induced defects in Ge-containing, $n$-type Si, either grown by molecular-beam epitaxy or by the Czochralski technique, have been studied by on-line deep-level transient spectroscopy (DLTS) after low-temperature irradiations. A level associated to the Ge-V pair has been unambigously identified with an activation enthalpy of 0.29 eV relative to the conduction band. It is argued that this level is most probably the (=/-) level of the Ge-V pair. A DLTS line correlated to a possible (-/0) level is not observed.

Silicon samples of n-type have been implanted with low doses (${10}^{7}$-${10}^{10}$${\mathrm{ncm}}^{\mathrm{\ensuremath{-}}2}$) of $^{11}\mathrm{B}$, $^{12}\mathrm{C}$, $^{16}\mathrm{O}$, $^{28}\mathrm{Si}$, $^{74,76}\mathrm{Ge}$, and $^{120}\mathrm{Sn}$ ions using energies between 0.4 and 8.0 MeV. Because of the low doses employed, single-collision cascades prevail, and an analysis of the implanted samples by deep-level transient spectroscopy (DLTS) reveals two dominant point defects of vacancy type, the divacancy (${\mathrm{V}}_{2}$) and the vacancy-oxygen (VO) center. The generation of ${\mathrm{V}}_{2}$ and VO is studied in detail as a function of ion mass, dose, dose rate, sample depth, and sample temperature. Surface-enhanced annihilation of migrating defects is found to play a significant role in depleting the concentration of ${\mathrm{V}}_{2}$ and VO centers in proximity to the surface. The concentration of ${\mathrm{V}}_{2}$ and VO increases linearly with ion dose, but the generation efficiency per incoming ion decreases at high dose rates after displaying a constant value below \ensuremath{\sim}${10}^{8}$${\mathrm{ncm}}^{\mathrm{\ensuremath{-}}2}$ ${\mathrm{s}}^{\mathrm{\ensuremath{-}}1}$. This dose rate effect exhibits a dependence on ion mass, and is qualitatively predicted by computer simulations of the defect reaction kinetics, applying a model where the interaction between individual collision cascades is primarily due to fast-diffusing Si self-interstitials. At dose rates \ensuremath{\leqslant}${10}^{8}$${\mathrm{ncm}}^{\mathrm{\ensuremath{-}}2}$ ${\mathrm{s}}^{\mathrm{\ensuremath{-}}1}$, the generation of ${\mathrm{V}}_{2}$ centers per ion-induced vacancy in the damage peak region is identical, within \ifmmode\pm\else\textpm\fi{}10%, for all the different types of ions studied. However, the ${\mathrm{V}}_{2}$ centers formed by heavy ions are strongly perturbed, as shown by large deviations from a one-to-one relation between the two DLTS peaks associated with the singly and doubly negative charge state of ${\mathrm{V}}_{2}$, and also by a broadening of the two peaks. In contrast to that for ${\mathrm{V}}_{2}$, the generation of VO centers per ion-induced vacancy decreases with increasing ion mass, consistent with the picture that light ions are more effective in generating point defects than heavy ions. At elevated implantation temperatures, gradual relaxation of lattice strain occurs together with recrystallization of disordered zones, promoting the formation of unperturbed ${\mathrm{V}}_{2}$ centers and increasing the generation efficiency of VO centers.

Deep levels in Cd_(1-x)Zn_xTe have not yet been fully characterized and understood, even though this material is very promising for medical and optoelectronic applications. We have investigated p-type semi-insulating Cd_(0.8)Zn_(0.2)Te with cathodoluminescence (CL) and photoinduced current transient spectroscopy (PICTS) methods. PICTS analyses allow detection of deep levels which are not revealed by other current spectroscopy techniques generally used, as they permit scanning of a wider region of the energy gap. Five levels have been detected (0.16, 0.25, 0.57, 0.78, and 1.1 eV) and, by combining the results obtained with the above-mentioned CL techniques, we were able to advance hypotheses on the character (donor or acceptor) and origin of some of these levels. The key role prayed by the 0.78-eV level in controlling the carrier transport properties has also been confirmed.

Using electron paramagnetic resonance (EPR), energy levels of the carbon vacancy (V(C)) in 4H-SiC and its negative-U properties have been determined. Combining EPR and deep-level transient spectroscopy we show that the two most common defects in as-grown 4H-SiC--the Z(1/2) lifetime-limiting defect and the EH(7) deep defect--are related to the double acceptor (2-|0) and single donor (0|+) levels of V(C), respectively.

Electrical properties of high-dose (1.6\ifmmode\times\else\texttimes\fi{}${10}^{16}$ at./${\mathrm{cm}}^{2}$) ${\mathrm{H}}^{+}$-implanted B-doped silicon have been investigated using transient capacitance spectroscopy, capacitance-voltage, and spreading resistance profiling. The role of hydrogen is twofold: to interact with the defects created by ion implantation, modifying their electrical properties, and to neutralize the shallow-acceptor dopants. The evolution of the defects responsible for the deep levels and the depth of the neutralized region have been investigated after iso- chronal annealing at various temperatures up to 800 \ifmmode^\circ\else\textdegree\fi{}C. Deep-level transient spectroscopy spectra show three hole traps; two of them, H(0.67), H(0.33), have been tentatively identified as vacancy-hydrogen complexes (${\mathrm{VH}}_{2}$, ${\mathrm{VH}}_{3}$) while the attribution of the third, H(0.23), detected in the samples annealed at 400--450 \ifmmode^\circ\else\textdegree\fi{}C, is uncertain. As a function of the heat treatment, the total number of defects is strongly reduced at 300 \ifmmode^\circ\else\textdegree\fi{}C, it increases for Tg300 \ifmmode^\circ\else\textdegree\fi{}C and at the highest temperatures, namely 800 \ifmmode^\circ\else\textdegree\fi{}C, the defects disappear. The first decrease is attributed to the formation of neutral ${\mathrm{VH}}_{4}$ complexes and the disappearance to complete decoration of point defects or their agglomeration. The thickness of the passivated region has a minimum at 300 \ifmmode^\circ\else\textdegree\fi{}C, which corresponds to the formation of the stable ${\mathrm{VH}}_{4}$. At lower temperatures, the hydrogen necessary for the passivation is the unbonded one and presumably comes from the implantation process itself. At higher temperature, H comes from the breaking of the ${\mathrm{VH}}_{4}$ or ${\mathrm{H}}_{\mathit{n}}$ complexes. These results are in good agreement with our previous studies concerning the role played by hydrogen in affecting the crystal properties of silicon.

We present results of correlated electron-paramagnetic-resonance and photoconductivity measurements on the tin donor in ${\mathrm{Ga}}_{1\mathrm{\ensuremath{-}}\mathit{x}}$${\mathrm{Al}}_{\mathit{x}}$As (0.3x0.7), which demonstrate that the photoionization of the DX center leads to the formation of a localized paramagnetic donor state, ${\mathit{DX}}^{0}$(Sn), according to the reaction ${\mathit{DX}}^{\mathrm{\ensuremath{-}}}$\ensuremath{\rightarrow}${\mathit{DX}}^{0}$+${\mathit{e}}^{\mathrm{\ensuremath{-}}}$, in agreement with negative-U models for the DX ground state. The ${\mathit{DX}}^{0}$ state, which has been previously attributed by us to the ${\mathit{A}}_{1}^{\mathrm{ab}}$ antibonding state, is observed only in a limited alloy range 0.3x0.5, where the ${\mathit{DX}}^{0/+}$ level is not resonant with the conduction band. Our results provide a simple explanation for the origin of the intermediate state in the electron-capture process of the single donors, the existence of which has been postulated previously from an analysis of the electron-capture kinetics of the DX center as well as from low-temperature photo-DLTS (deep-level transient spectroscopy) measurements: the one-electron ${\mathit{A}}_{1}^{\mathrm{ab}}$ state. As ${\mathit{A}}_{1}^{\mathrm{ab}}$ states are a fundamental property of substitutional donors, these results are also relevant for the analysis of DX-center properties of the other substitutional donors Si, Se, Te, and S in ${\mathrm{Ga}}_{1\mathrm{\ensuremath{-}}\mathit{x}}$${\mathrm{Al}}_{\mathit{x}}$As.

The effect of electric field on the thermal emission of electrons from the ${\mathrm{Ti}}^{3+}$/${\mathrm{Ti}}^{4+}$ deep donor level in InP has been investigated. Double-correlation deep-level transient spectroscopy as well as differential-isothermal-capacitance transient measurements have been carried out on low-pressure--metalorganic-chemical-vapor-deposition-grown InP:Ti samples. It is found that the emission rates are strongly field dependent increasing by up to a factor of 17 corresponding to an increase of the field by a factor of 3.5 in the measured temperature range of 260 to 340 K. The experimental data are well fitted with a Poole-Frenkel model employing a three-dimensional square-well potential associated with the ${\mathrm{Ti}}^{3+}$/${\mathrm{Ti}}^{4+}$ level with a radius r=4.6 nm. The fit of this model to the experimental data yields variations in the activation energy, \ensuremath{\Delta}E=0.48\ifmmode\pm\else\textpm\fi{}0.02 eV to 0.57\ifmmode\pm\else\textpm\fi{}0.02 eV, depending upon the actual field strength. An extrapolated zero-field \ensuremath{\Delta}E(0)=${\mathit{E}}_{\mathit{C}}$-${\mathit{E}}_{\mathit{T}}$=0.59\ifmmode\pm\else\textpm\fi{}0.02 eV is found. The electron-capture cross section is determined to be \ensuremath{\approxeq}(6.6\ifmmode\pm\else\textpm\fi{}0.3)\ifmmode\times\else\texttimes\fi{}${10}^{\mathrm{\ensuremath{-}}13}$ ${\mathrm{cm}}^{2}$. Thus, the controversy about the previously reported variations in \ensuremath{\Delta}E values is resolved. A comparison of \ensuremath{\Delta}E(0) with the ${\mathrm{Ti}}^{3+}$/${\mathrm{Ti}}^{4+}$ energy position in GaAs and ${\mathrm{In}}_{0.53}$${\mathrm{Ga}}_{0.47}$As shows that the energies are within 20 meV horizontally across the heterojunction, confirming a prediction of the ``internal-reference'' rule for the energy position of transition-metal levels in isoelectronic semiconductors.

Using comparative local-vibrational-mode and deep-level transient spectroscopy, we have studied the off-center ${\mathrm{O}}_{\mathrm{As}}$ defect in high-resistivity, undoped GaAs and neutron-transmutation-doped, n-type samples from the same starting material. The results fully support that this defect has a negative-U character. They also reveal that its deeper, two-electron level corresponds to the chemically unidentified EL3-defect level in GaAs at ${\mathit{E}}_{\mathit{c}}$-0.58 eV.

Using optical-excitation minority-carrier deep-level transient spectroscopy we have studied the thermal annealing in n-GaAs of lattice-defect hole traps produced by 1.0-MeV proton irradiation at 120 K. We find that three hole traps, with electronic levels at 0.16 eV (of low concentration), at about 0.25 eV, and at 0.42 eV above the valence band, are removed in a thermal-annealing stage centered at about 280 K, consistent with the well known resistivity-annealing stage found at this temperature in electron-irradiated n-GaAS

The H3-center with a level at ${E}_{v}+0.535\mathrm{eV}$ is observed in hydrogenated electron-irradiated boron-doped silicon. In samples with boron concentrations of $(2\char21{}20)\ifmmode\times\else\texttimes\fi{}{10}^{15}{\mathrm{cm}}^{\ensuremath{-}3}$ the center is the most abundant among all defects detected by deep-level transient spectroscopy. Electrical properties and thermal stability of the H3-center are studied in detail. It is shown that the H3-center is created by the addition of one hydrogen atom to a radiation-induced defect, the H3 precursor, which escapes direct electrical detection in the samples. The H3 precursor is formed immediately after room-temperature irradiation in oxygen-lean crystals and after heat treatment at $\ensuremath{\sim}200\ifmmode^\circ\else\textdegree\fi{}\mathrm{C}$ in oxygen-rich wafers. The H3 precursor anneals out at $\ensuremath{\sim}375\ifmmode^\circ\else\textdegree\fi{}\mathrm{C}$ in all samples. The introduction rates of H3 precursor and other radiation defects are measured for samples with different boron concentrations. The identification of H3 precursor as the ${\mathrm{B}}_{i}{\mathrm{C}}_{s}$ pair and H3-center as a BCH complex is in good agreement with the available experimental data.

We apply Laplace deep-level transient spectroscopy in situ after low-temperature proton implantation into dilute Si 1 - x Ge x alloys and identify the deep donor state of hydrogen occupying a strained Si-Si bond-center site next to Ge. The activation energy of the electron emission from the donor is ∼158 meV when extrapolated to zero electrical field. We construct a configuration diagram of the Ge-strained site from formation and annealing data and deduce that alloying with ∼1% Ge does not significantly influence the low-temperature migration of hydrogen as compared to elemental Si. We observe two bond-center-type carbon-hydrogen centers and conclude that carbon impurities act as much stronger traps for hydrogen than the alloy Ge atoms.

Positron annihilation spectroscopy has been applied to identify Si and C vacancies as irradiation-induced defects in 6H-SiC. Si vacancies are shown to have ionization levels at E C -0.6 eV and E C -1.1 eV below the conduction-band edge E C by detecting changes of positron trapping under monochromatic illumination. These levels are attributed to (2-/1-) and (1-/0) ionizations of the isolated Si vacancy. In as-grown n-type 6H-SiC, a native defect complex involving V S i is shown to have an ionization level slightly closer to conduction band at roughly E C -0.3 eV. These results are used further to present microscopic interpretations to effects seen in optical-absorption spectra and to electrical levels observed previously by deep-level transient spectroscopy.

It is shown that the divacancy in silicon in the diamagnetic doubly negative charge state has a static trigonal symmetry with inward breathing mode lattice relaxation. There is no measurable Jahn-Teller effect, unlike other charge states of the defect. This conclusion has been drawn from an analysis of the piezoscopic characteristics of the complex derived from high-resolution (Laplace) deep-level transient spectroscopy.

The passivation by reaction with atomic hydrogen of the well-known deep donor and acceptor levels associated with gold in silicon has been observed with the use of deep-level transient spectroscopy. The concentration profiles of the two centers are presented as a function of the duration and temperature of the exposure to atomic hydrogen: When exposed for 2 h at 350\ifmmode^\circ\else\textdegree\fi{}C, less than 5 \ifmmode\times\else\texttimes\fi{} ${10}^{\ensuremath{-}3}$ of the originally electrically active deep donors and acceptors remained to a depth of approximately 7 \ensuremath{\mu}m. Slight but significant differences in the concentration profiles of the electrically active defect centers after reaction with the atomic hydrogen may be additional evidence that the gold donor level and gold acceptor level are not related to the same center.

We report the observation of a defect in Fe-doped InP which can exist in either of two configurations, for the same charge state. Each configuration exhibits distinct electronic and optical properties. Thermally stimulated capacitance, capacitance transient spectroscopy, and photocapacitance were used to study the properties of the defect in each configuration, and the kinetics of the reversible transformations between configurations. The unique properties of the defect are discussed.

Etude du niveau E c -0,17 eV du silicium irradie par des neutrons, par spectroscopie transitoire sous contrainte uniaxiale. Determination de la symetrie des defauts, de la redistribution electronique induite par la contrainte et de l'orientation preferentielle des defauts. Mesure de l'energie d'activation de reorientation

The variation of the potential of a quantum well is similar to that of a deep trap. In that respect a quantum well can capture and emit carriers in much the same way as a trap. The thermal emission energy from a quantum well is closely related to the appropriate band offset. With that in mind, we have carried out deep-level transient spectroscopy measurements on Schottky-barrier diodes containing one or more pseudomorphic ${\mathrm{In}}_{\mathit{x}}$${\mathrm{Ga}}_{1\mathrm{\ensuremath{-}}\mathit{x}}$As/${\mathrm{Al}}_{0.2}$${\mathrm{Ga}}_{0.8}$As (0lx\ensuremath{\le}0.18) quantum wells. The objective was to estimate the conduction-band offset, \ensuremath{\Delta}${E}_{c}$, as a function of x and the resulting strain. From detailed balance between emission and capture, an Arrhenius-type expression was derived to analyze the transient emission data. It is seen that the percentage band offset \ensuremath{\Delta}${E}_{c}$/\ensuremath{\Delta}${E}_{g}$ varies from 62% for x=0.07 to 70% at x=0.18. Furthermore, a linear interpolation of the data leads to \ensuremath{\Delta}${E}_{c}$/\ensuremath{\Delta}${E}_{g}$=58% at x=0, which is close to the widely accepted value. Our results support recent theoretical calculations from which a monotonic increase in \ensuremath{\Delta}${E}_{c}$ with strain in this heterostructure system is predicted.

The possibility of characterizing the traps of the oxide layer near the Si/${\mathrm{SiO}}_{2}$ interface using deep-level transient spectroscopy (DLTS) is pointed out. By use of the so-called saturating-pulse DLTS, an unusual increase of the DLTS spectrum is observed. This behavior is interpreted by taking into account tunneling emission from oxide traps situated near the silicon dioxide--silicon interface (\ensuremath{\approxeq}10 to 30 A\r{}). By introducing a model for the derivation of the DLTS signal amplitude, we show here how the deconvolution between interface states and oxide traps is allowed. Then, from the experimental study of the filling kinetics the concentration of these oxide traps is obtained (\ensuremath{\approxeq}${10}^{7}$ to ${10}^{8}$ ${\mathrm{cm}}^{\mathrm{\ensuremath{-}}2}$) as well as their associated tunneling capture cross section (\ensuremath{\approxeq}${10}^{\mathrm{\ensuremath{-}}22}$ to ${10}^{\mathrm{\ensuremath{-}}21}$ ${\mathrm{cm}}^{2}$). Moreover, by variation of the emission rate window, it is possible to measure the concentration profile of these oxide traps. The validity of the model and measurements are extensively discussed.

Measurements of transient photoinduced optical absorption in $a$-${\mathrm{As}}_{2}$${\mathrm{Se}}_{3}$ show that carriers thermalize within a distribution of localized states in the band gap. Photocurrent as well as photoinduced optical absorption transients are explained in a unified way with the multiple-trapping mechanism. Under such circumstances, transient photoconductivity provides a spectroscopy of the density of localized states.

We report deep-level transient spectroscopy studies on the divacancy formation in electron-bombarded silicon at 80 K. Samples irradiated to a dose D/sub 1/ = 3.5 x 10/sup 16/ e/sup -//cm/sup 2/ and using a beam current i/sub 1/ = 0.33 ..mu..A/cm/sup 2/ give rise to an unstable peak seen to anneal out rapidly below 150 K. The center responsible for this peak seems to act as an intermediate step in the formation of at least a fraction of the neutral divacancies. Under conditions of irradiation with dose D/sub 2/ = 1.7 x 10/sup 16/ e/sup -//cm/sup 2/ and beam current i/sub 2/ = 0.20 ..mu..A/cm/sup 2/ neither the intermediate defect nor the neutral divacancy signal appears in the spectrum. Instead, a divacancy-related E/sub V/+0.13 eV level emerged. An interpretation of the data is attempted in terms of a model postulating charge-state-dependent metastable structures.

A temperature dependence of the electron-capture cross section [${\ensuremath{\sigma}}_{n}(E,T)$] of the gap states in $a\ensuremath{-}\mathrm{S}\mathrm{i}:\mathrm{H}$ was determined for the first time by isothermal capacitance transient spectroscopy. A bump was observed on the gap-state profile of lightly P-doped $a\ensuremath{-}\mathrm{S}\mathrm{i}:\mathrm{H}$ and ${\ensuremath{\sigma}}_{n}(E,T)$ increases as temperature increases. From this temperature dependence as well as the energy dependence of ${\ensuremath{\sigma}}_{n}(E,T)$ reported previously, it is concluded that the bump of gap-state profile is located approximately 0.5 eV below the conduction-band mobility edge and multiphonon emission predominates in the electron-capture process at the gap states near the bump. It is strongly suggested that the bump structure originates from doubly occupied dangling-bond states.