The purpose of this work is to demonstrate equivalent bipolar transistor damage response to neutrons and silicon ions. We report on irradiation tests performed at the White Sands Missile Range Fast Burst Reactor, the Sandia National Laboratories (SNL) Annular Core Research Reactor, the SNL SPHINX accelerator, and the SNL Ion Beam Laboratory using commercial silicon npn bipolar junction transistors (BJTs) and III–V Npn heterojunction bipolar transistors (HBTs). Late time and early time gain metrics as well as defect spectra measurements are reported.

Electrical performance and characterization of deep levels in vertical GaN P-i-N diodes grown on low threading dislocation density (∼104 - 106 cm−2) bulk GaN substrates are investigated. The lightly doped n drift region of these devices is observed to be highly compensated by several prominent deep levels detected using deep level optical spectroscopy at Ec-2.13, 2.92, and 3.2 eV. A combination of steady-state photocapacitance and lighted capacitance-voltage profiling indicates the concentrations of these deep levels to be Nt = 3 × 1012, 2 × 1015, and 5 × 1014 cm−3, respectively. The Ec-2.92 eV level is observed to be the primary compensating defect in as-grown n-type metal-organic chemical vapor deposition GaN, indicating this level acts as a limiting factor for achieving controllably low doping. The device blocking voltage should increase if compensating defects reduce the free carrier concentration of the n drift region. Understanding the incorporation of as-grown and native defects in thick n-GaN is essential for enabling large VBD in the next-generation wide-bandgap power semiconductor devices. Thus, controlling the as-grown defects induced by epitaxial growth conditions is critical to achieve blocking voltage capability above 5 kV.

We studied the effect of light ion and heavy ion irradiations on pnp Si BJTs. A mismatch in DLTS deep peak amplitude for devices with same final gain but irradiated with different ion species was observed. Also, different ions cause different gain degradation when the DLTS spectra are matched. Pre-dosed ion-irradiated samples show that ion induced ionization does not account for the differences in DLTS peak height but isochronal annealing studies suggest that light ions produce more VP defects than heavy ions to compensate for the lack of clusters that heavy ions produce. The creation of defect clusters by heavy ions is evident by the higher content of E4 and V-2* defects compared to light ions.

Electrical performance and defect characterization of vertical GaN P-i-N diodes before and after irradiation with 2.5 MeV protons and neutrons is investigated. Devices exhibit increase in specific on-resistance following irradiation with protons and neutrons, indicating displacement damage introduces defects into the p-GaN and n- drift regions of the device that impact on-state device performance. The breakdown voltage of these devices, initially above 1700 V, is observed to decrease only slightly for particle fluence <; 1013 cm-2. The unipolar figure of merit for power devices indicates that while the on-resistance and breakdown voltage degrade with irradiation, vertical GaN P-i-Ns remain superior to the performance of the best available, unirradiated silicon devices and on-par with unirradiated modern SiC-based power devices.

An improved method for measuring the cross sections for carrier trapping at defects in semiconductors is described. This method, a variation of deep level transient spectroscopy (DLTS) used with bipolar transistors, is applied to hot carrier trapping at vacancy-oxygen, carbon-oxygen, and three charge states of divacancy centers (V-2) in n-and p-type silicon. Unlike standard DLTS, we fill traps by injecting carriers into the depletion region of a bipolar transistor diode using a pulse of forward bias current applied to the adjacent diode. We show that this technique is capable of accurately measuring a wide range of capture cross sections at varying electric fields due to the control of the carrier density it provides. Because this technique can be applied to a variety of carrier energy distributions, it should be valuable in modeling the effect of radiation-induced generation-recombination currents in bipolar devices. (C) 2015 AIP Publishing LLC.

We have measured the temperature and field dependence of emission rates from five traps in electron damaged GaAs. Four of the traps have previously been identified as radiation defects. One of the traps, seen in higher doped diodes, has not been previously identified. We have fit the data to a multiphonon emission theory that allows recombination in GaAs to be characterized over a broad range of temperature and electric field. These results demonstrate an efficient method to calculate field-dependent emission rates in GaAs.

We present an experimental methodology developed to probe the clustered defect formation in GaAs devices under both neutron and ion irradiations. The strengths, limitations, and path forward to gather structural defect information will be addressed.

Ion Beam Induced Charge (IBIC) and Deep Level Transient Spectroscopy (DLTS) were used to investigate displacement damage caused by MeV energy ion beams in Si diodes. The devices were irradiated with 3MeV Si ions to create displacement damage and a 2MeV He ion beam was used for IBIC. The IBIC signal deterioration was measured as the function of the ion fluence and DLTS was used to identify the defects and their quantities. We used a new calculation method based on previous work by Fizzotti et al. [9] and more recently by Vittone [10] to determine the fraction of the active traps that affect the lifetime. The first application of this method is presented in this paper.

The positions of the electronic levels of an intrinsic bistable defect have been measured using deep level transient spectroscopy (DLTS) in n- and p-type damaged silicon bipolar transistor diodes after minority carrier injection and thermal annealing. The kinetic rates observed during conversion of this defect have been determined using both DLTS and transistor gain measurements on devices irradiated with electrons and neutrons. First order conversion kinetics are observed during both injection and thermal annealing of this defect in electron damaged transistors, but more complicated, stretched-out kinetics are seen in neutron irradiated devices. The latter behavior can be successfully modeled as a small spread in the energy barriers for atomic displacements of this defect probably due to strain or electrostatic variations expected in damage clusters. The measured injection bias dependence of the recombination-driven transition to the bistable state of this defect is unlike that seen for generation-recombination currents at deep trap levels, but it is consistent with Shockley—Read–Hall predictions for recombination at shallow states. This latter behavior is expected from the lone shallow peak seen in DLTS measurements of the stable state of this center. Some comparisons of the present results with existing models of this defect are made.

We studied the effect of carbon incorporation on the material and electrical properties of Ta2O5 thin film. We doped the Ta2O5 films with carbon using pulsed-dc reactive and rfmagnetron sputtering of Ta2O5 performed in an Ar/O2/CO2 plasma. In thick (70 nm) films, an optimal amount (0.8 - 1.4 at.%) of carbon doping reduced the leakage current to 10−8 A/cm2 at +3 MV/cm, a four orders of magnitude reduction compared to that in a pure Ta2O5 film grown in similar conditions without CO2 in the plasma. This finding suggests that carbon doping can significantly improve the dielectric leakage property at an optimal concentration. X-ray Photoemission Spectroscopy (XPS) analysis showed the presence of carbonate in these electrically improved carbon-doped films. Analysis by high-resolution transmission electron microscopy (HRTEM) exhibited no morphological or structural changes in these carbon doped films. Carbon doping showed no improvement in the leakage current in thin (10 nm) Ta2O5 films. This phenomenon is explained by a defect compensation mechanism, in which the carbon-related defects remove carriers at low concentrations but form a hopping conduction path at high concentrations.

We find that fast neutron irradiated n- and p-GaAs diodes both show a broad feature in deep level transient spectroscopy (DLTS) previously studied primarily in n-GaAs and termed the “U-band.” The high temperature edge of the broad DLTS feature cuts off at the same temperature in both n- and p-GaAs suggesting that the cut off is due to the DLTS behavior expected for a continuous density of defect states that spans midgap. The band gap implied by the DLTS midgap cut off is 1.36 eV, as compared to the bulk GaAs band gap 1.52 eV. Band gap narrowing is consistent with previous measurements of lattice expansion in neutron irradiated GaAs. This leads to a model of defect cascades that are regions of narrowed band gap with defect levels that are inhomogeneously broadened. We observe, in addition, that the damage cascades are surrounded by large Coulomb barriers that prevent the complete filling of traps in the damaged regions.

Isochronal anneal sequences have been carried out on pnp and npn transistors irradiated with fast neutrons at a variety of fluences. The evolution of base and collector currents was utilized to characterize the annealing behavior of defects in both the emitter-base depletion region and the neutral base. Various annealing biases, theoretical modeling, as well as previous deep level transient spectroscopy (DLTS) data, were used to assign the relative magnitude of each of the important defects to the total recombination current. We find that donor-vacancy pairs in the neutral n-type base of our pnp transistors are responsible for about 1/3 of the postdamage lifetime degradation, while the remaining recombination currents can be largely attributed to a cluster-related divacancylike defect which has no shallow state DLTS emission peak. This latter defect anneals gradually from 350 to 590 K. Generation/recombination currents in the base-emitter junctions in both types of devices were found to anneal in a similar, gradual fashion, suggesting that this same cluster-related intrinsic lattice defect is also responsible for the large, damage-induced base currents.

We have explored defect annealing in radiation damaged silicon in a regime characterized by defect clusters and higher doping. Several types of pnp and npn Si bipolar transistors have been irradiated with ions and neutrons, then isochronally annealed from 300 to 600 K to study the evolution of deep level transient spectroscopy (DLTS) defect signatures. Variations in these data with radiation environment, Fermi level, annealing temperature, and doping density have been used to separate the contributions of three dominant defects to the DLTS defect spectra. We find that the normal Si divacancy and a divacancylike defect with similar properties make similar contributions to a DLTS peak normally associated with transitions from the single minus charge state of the divacancy. However the latter defect is clearly associated with the presence of defect clusters. The vacancy-donor center can also contribute to this high temperature DLTS signature, and its relative importance can be quantitatively assessed by varying doping density and the bias applied to the sample p/n junctions during annealing, and also by the observation that another, donor-related defect grows in as this center anneals. The ratio of vacancy-donor and vacancy-oxygen pairs appears to accurately follow that seen in earlier studies of gamma-irradiated Si. Discussions are presented concerning the effects of defect clustering on the structure, appearance, and evolution of the defects we have identified.

Two deep level transient spectroscopy (DLTS) electron emission signatures, previously labeled E4 and E5, have been shown to be bistable with respect to minority carrier injection at room temperature. These result from two charge state transitions of the same defect. We have performed DLTS measurements as function of annealing between 350 and 680 K, using minority carrier injection after each annealing stage to make E4 and E5 visible. We show that the E4–E5 pair is associated with defect clusters which dominate after neutron or ion damage with annealing characteristics that closely parallel to those of silicon divacancies found in damage clusters. At annealing temperatures above 500 K, the E4–E5 pair ceases to be bistable and exists after anneals in thermal equilibrium. We show that the stable E4 peak appears to be the same emission signature previously labeled the L center. The transformation of the E4–E5 bistable pair into the stable L center and a stable E5 companion level occurs at the same temperature, where it has been suggested that the divacancy becomes mobile. The similarity of the annealing of the E4–E5 pair to that of the divacancy, the dependence of the density of these defects on degree of clustering, and the insensitivity to common impurities combine to suggest that the E4–E5 pair is associated with primary defects located in the defect cluster and closely related to the familiar divacancy.

We report on an early-time inverse gain comparison between ion and neutron irradiated silicon bipolar junction transistors. We find ion irradiations to be an excellent simulator for fast-burst neutrons for early-time behavior and damage creation rates. In addition we report on an experimental to simulation comparison of transient gain annealing response. The simulations are from a physics based modeling approach that is being developed at Sandia National Laboratories as part of the Qualification Alternatives to the Sandia Pulsed Reactor (QASPR) Program. We find excellent agreement between simulation and experiment across a wide range of irradiation conditions.

Following irradiation, the gain of silicon bipolar transistors can be improved by annealing at 350K. We show that both the number of defects measured by deep level transient spectroscopy (DLTS) and the gain can be restored to the post irradiation state by injection of minority carriers. One can cycle between the post irradiation state and the 350K annealed state by alternating minority carrier injection at 300K with zero- or reverse-bias anneals at 350K. The structure of the bistable defects is not known, but we observe that they affect capture kinetics into the shallow charge state of the silicon divacancy defect, V2 (=/−). This suggests that the bistable defects are located within the neutron or ion damage cluster.

In situations where defects are clustered and the number of defects within a cluster approaches the doping level, shallow defects states may be only partially filled in equilibrium. We have modeled both the equilibrium filling and the kinetic capture dynamics of the silicon divacancy, V2, in neutron damaged silicon. A simple electrostatic model that assumes that V2 defects have the same spatial distribution as a simulation of vacancy defects following neutron irradiation is able to account for the observed capture kinetics. This electrostatic model does not, however, explain the large V2 asymmetry typical of neutron damaged silicon.

Using deep level transient spectroscopy, the authors have measured the defect spectrum in the collector of a n-p-n bipolar transistor following fast neutron irradiation as well as the gain on the same device. They show that a slow change observed in both the gain and deep level traps in the n-type collector at 300K are bistable. The transistor gain and the defects can be returned to the postirradiation condition by forward bias at room temperature, i.e., by operating the transistor (gain) or injection through the base-collector diode (defect spectrum).