Publications

Trapped protons and electrons in the Earth's radiation belts and cosmic rays present significant challenges for electronics that must operate reliably in the natural space environment. Single event effects (SEE) can lead to sudden device or system failure, and total dose effects ran reduce the lifetime of a space-based telecommunications system. One of the greatest sources of uncertainty in developing radiation requirements for a space system is accounting for the small but finite probability that the system will be exposed to a massive solar particle event. Once specifications are decided, standard laboratory tests are available to predict the total dose response of MOS and bipolar components in space, but SEE testing of components can be more challenging. Prospects are discussed for device modeling and for the use of standard commercial electronics in space. © 1999 IEEE.

Measurements of dielectric breakdown during high-field electrical stress are typically performed at or near room temperature via constant voltage or current stress methods. In this summary they explore whether useful information might also be obtained by performing current measurements during a temperature ramp at high electric field.

We have observed discrete random telegraph signals (RTS'S) in the drain voltages of three, observed above 30 K were thermally activated. The switching rate for the only RTS observed below 30 K was thermally activated above 30 K but temperature-independent below 10 K. To our knowledge, this cross-over from thermal activation to tunneling behavior has not been previously observed for RTS's Metal-oxide-semiconductor field-effect transistors (MCEWETS) often exhibit relatively large levels of low-frequency (1/fl noise) [1,2]. Much evidence suggests that this noise is related to the capture all cases, switching rates have been thermally activated, often with different activation energies for capture and/or emission is accompanied by lattice relaxation. Though thermally activated behavior has sufficiently low temperatures [7,9]. While not observed in MOSFETS, cross-over from thermal activation to configurational tunneling has been observed for RTS's in junctions [13]. drain voltage was observed to randomly switch between two discrete levels, designated as Vup and Vdn, similar to RTS's reported by others [2,7'- 11 ]. We have characterized six RTS `S for temperatures above 30 K where thermally activated switching rates are observed. The properties of five of these have been the trap, i.e., the mean time a captured charge carrier spends in the trap before it is emitted. Similarly, we identify the mean time in the low resistance state ( trup in state Vup) as the capture time rc. F@ure 1 shows a typical time trace of the drain-voltage fluctuation &d(t)= Vd(t)+Vd>. This indicate that both the mean capture and emission times become independent of Tat low temperatures and where a= capture or emission, is temperature independent. The solid curve in Figure 3(a) (mean capture time) was obtained using a weighted nonlinear charge carriers are not in thermal equilibrium with the lattice, i.e., that while the lattice is being cooled Instead, we believe that the transition from thermally activated to temperature-independent switching rates is associated with a lattice relaxation mechanism similar to that observed in metal- insulator-metal tunnel junctions [13]. Capture and emission of carriers are mediated by lattice relaxation, which proceeds via a thermally activated process at higher temperatures and a configurational tunneling electron capture rate depended on both lattice and electron temperatures while the emission rate Fkure 2. Arrhenius plot showing the thermally-activated behavior of both the mean capture (triangle) and emission (square) times of the RTS for temperatures above 20 K'.

Our data suggest a correlation between near-interface strain in SiO{sub 2} and the ratio of fixed vs. mobile positive charge generated at the interface during forming gas annealing. A model based on first-principles quantum mechanical calculations supports this correlation.

We report experimental results suggesting that mobile protons are generated at strained Si-O-Si bonds near the Si/SiO₂ interface during annealing in forming gas. Our data further suggest that the presence of the top Si layer plays a crucial role in the mobile H⁺ generation process. Finally, we show that the diffusion of the reactive species (presumably H₂ or H⁰) towards the H⁺ generation sites occurs laterally along the buried oxide layer, and can be impeded significantly due to the presence of trapping sites in the buried oxide.

The isochronal anneal technique used to predict isothermal anneal behavior of MOS devices is analyzed as a function of experimental parameters. The effects of detrapping of trapped holes and compensating electrons are discussed.

It is shown how mobile H{sup +} ions can be generated thermally inside the oxide layer of Si/SiO{sub 2}/Si structures. The technique involves only standard silicon processing steps: the nonvolatile field effect transistor (NVFET) is based on a standard MOSFET with thermally grown SiO{sub 2} capped with a poly-silicon layer. The capped thermal oxide receives an anneal at {approximately}1100 C that enables the incorporation of the mobile protons into the gate oxide. The introduction of the protons is achieved by a subsequent 500-800 C anneal in a hydrogen-containing ambient, such as forming gas (N{sub 2}:H{sub 2} 95:5). The mobile protons are stable and entrapped inside the oxide layer, and unlike alkali ions, their space-charge distribution can be controlled and rapidly rearranged at room temperature by an applied electric field. Using this principle, a standard MOS transistor can be converted into a nonvolatile memory transistor that can be switched between normally on and normally off. Switching speed, retention, endurance, and radiation tolerance data are presented showing that this non-volatile memory technology can be competitive with existing Si-based non-volatile memory technologies such as the floating gate technologies (e.g. Flash memory).

Deep and shallow electron traps form in irradiated thermal SiO{sub 2} as a natural response to hole transport and trapping. The density and stability of these defects are discussed, as are their implications for total-dose modeling.

Thermally stimulated current and capacitance voltage methods are used to investigate the thermal stability of trapped electrons associated with radiation-induced trapped positive charge in metal-oxide-semiconductor capacitors. The density of deeply trapped electrons in radiation-hardened 45 nm oxides exceeds that of shallow electrons by a factor of {approximately}3 after radiation exposure, and by up to a factor of 10 or more during biased annealing. Shallow electron traps anneal faster than deep traps, and seem to be at least qualitatively consistent with the model of Lelis et al. Deeper traps maybe part of a fundamentally distinct dipole complex, and/or have shifted energy levels that inhibit charge exchange with the Si.

Thermally stimulated current (TSC) techniques provide information about oxide-trap charge densities and energy distributions in MOS (metal-oxide-semiconductor) capacitors exposed to ionizing radiation or high-field stress that is difficult or impossible to obtain via standard capacitance-voltage or current-voltage techniques. The precision and reproducibility of measurements through repeated irradiation/TSC cycles on a single capacitor is demonstrated with a radiation-hardened oxide, and small sample-to-sample variations are observed. A small increase in E′δ center density may occur in some non-radiation-hardened oxides during repeated irradiation/TSC measurement cycles. The importance of choosing an appropriate bias to obtain accurate measurements of trapped charge densities and energy distributions is emphasized. A 10 nm deposited oxide with no subsequent annealing above 400 °C shows a different trapped-hole energy distribution than thermally grown oxides, but a similar distribution to thermal oxides is found for deposited oxides annealed at higher temperatures. Charge neutralization during switched-bias irradiation is found to occur both because of hole-electron annihilation and increased electron trapping in the near-interfacial SiO2. Limitations in applying TSC to oxides thinner than approximately 5 nm are discussed.

Mechanisms that control the response of MOS and bipolar devices to ionizing radiation in the natural space environment are briefly reviewed. Standard tests based on room-temperature irradiation and elevated temperature annealing are described for MOS devices to bound the effects of oxide and interface-trap charge in space. For bipolar devices that exhibit enhanced low-dose-rate sensitivity, a standard test equivalent to that developed for MOS devices is not available. However, screening techniques based on room temperature and/or elevated temperature irradiations are described which can minimize the risk to spacecraft and satellite electronics from this phenomenon.

In a number of recent studies the generation of mobile protons in the buried oxide of SOI materials and in thermal oxide buried underneath a poly-Si layer has been discussed. The protons are found to be stable and can be easily rearranged by applying an electric field. The details of the hydrogen reactions leading to the generation of the mobile H{sup +} are still under investigation. In a recent work a dynamic equilibrium model was presented. The forward reaction dominates above {approximately} 500 C and the resulting H{sup +} is mobile and entrapped inside the SiO{sub 2}. The electron is donated to the Si. The H{sup 0} is likely to be formed through H{sub 2} + K {Leftrightarrow} HK + H{sup 0}, where K is a cracking site. In the same work it was shown that the reactive hydrogen species enter the oxide from the device edges. Hence, the amount of the reactive species reaching the oxide by diffusion through the Si overlayer is negligible. These results seem to contradict earlier studies where it is shown that hydrogen can easily diffuse through the top Si layer under the given experimental conditions. The authors present here new details on hydrogen diffusion and chemistry during the protonation anneal that may offer an explanation for the hydrogen diffusion paradox. The new findings suggest that reactions at the ambient/SiO{sub 2} interface play a key role.

Proton implantation into the buried oxide of Si/SiO{sub 2}/Si structures does not introduce mobile protons. The cross section for capture of radiation-induced electrons by mobile protons is two orders of magnitude smaller than for electron capture by trapped holes. The data provide new insights into the atomic mechanisms governing the generation and radiation tolerance of mobile protons in SiO{sub 2}. This can lead to improved techniques for production and radiation hardening of radiation tolerant memory devices.

Radiation-induced degradation of many types of bipolar transistors and circuits is more severe following low dose rate exposure than following high dose rate exposure. Since microelectronic devices in space are generally subjected to low dose rate irradiation, this complicates the hardness assurance testing of linear circuits and can lead to an overestimation of device lifetime in space. Previous work examining the physical mechanisms responsible for this dose rate effect has focused primarily on oxide trapped charge. Reduced net positive oxide trapped charge densities at high dose rates and zero bias have been attributed to space charge effects from slowly transporting holes trapped metastably at O vacancy complexes. Decreasing the dose rate or increasing the irradiation temperature leads to an increase in net positive oxide trapped charge near the Si-SiO{sub 2} interface by reducing these space charge effects. In this work, concentrations of hydrogen transport through two types of bipolar oxides are estimated from dopant passivation measurements in MOS capacitors. For unbiased irradiations, hydrogen passivation of substrate acceptors is greatly reduced at high dose rates compared to that at low dose rates or elevated temperatures. Consistent with other widely accepted models, it is argued that fewer interface traps are formed by high dose rate irradiation under zero bias, because fewer H{sup +} ions can drift to the Si-SiO{sub 2} interface and react with trap precursors. Similar to hole transport in these oxides, drift of the H{sup +} ions is inhibited at high dose rates by space charge accumulated in the oxide bulk.

Oxide trapped charge, field effects from emitter metallization, and high level injection phenomena moderate enhanced gain degradation of lateral pnp transistors at low dose rates. Hardness assurance tests at elevated irradiation temperatures require larger design margins for low power measurement biases.

Radiation response comparisons of lateral PNP bipolar technologies reveal that device hardening may be achieved by extending the emitter contact over the active base. The emitter-tied field plate suppresses recombination of carriers with interface traps.

Capacitance-voltage and thermally-stimulated-current techniques are used to estimate trapped hole and electron densities in MOS oxides as functions of irradiation and isochronal anneal temperature. Trapped-charge annealing and compensation effects are discussed.

A delayed increase in 1/f noise is observed for pMOS transistors showing latent radiation-induced interface-trap buildup. The latent interface traps and increased noise appear to result from the same thermally activated process, likely involving hydrogen.

Dose enhancement and dose rate were measured in more than a dozen gamma sources using pMOS RADFETs and TLDs from two independent sources. ARACOR X-ray dose rates were calibrated using single- and dual-dielectric RADFETs.

The effect of dose rate on radiation-induced gain degradation is compared for verticle npn and lateral pnp bipolar transistors. High dose rate irradiations at elevated temperatures are more effective at simulating low dose rate degradation in the lateral pnp transistors.

A physical model is developed to quantify the contribution of oxide-trapped charge to enhanced low-dose-rate gain degradation in BJTs. Simulations show that space charge limited transport is partially responsible for the low-dose-rate enhancement.

Radiation-induced gain degradation is compared in two types of lateral PNP bipolar devices that are identical except for the emitter doping. The devices with heavily-doped emitters (1×1020 cm-3) degrade less than the devices with lightly-doped emitters (1×1018 cm-3). Both device types are sensitive to interface-trap formation in the oxide above the emitter-base junction and the neutral base region. In addition, the devices with lightly-doped emitters experience spreading of the depletion region into the emitter, increasing their sensitivity to total-dose irradiation. The gain degradation in both device types is due to a combination of increased base current and decreased collector current. The radiation-induced decrease in collector current is more significant for devices from this technology than for other devices studied previously. Increased gain degradation is observed in heavily-doped devices irradiated at low dose rates, but the enhanced degradation appears to be due to time-dependent effects rather than true dose-rate effects. The lightly-doped devices do not exhibit a clear dose-rate trend and the gain of these devices improves during post-irradiation annealing. © 1997 IEEE.

It is shown analytically and experimentally that, when significant densities of positive and/or negative charge are trapped in the bulk of the oxide, standard thermally stimulated current (TSC) measurements at negative gate bias may not provide accurate estimates of MOS oxide-trap charge densities. Combining TSC measurements at negative bias with capacitance-voltage (C-V) measurements allows useful, self-consistent estimates of trapped electron densities in the oxide to be obtained. However, unless one can determine whether most of the trapped electrons lie in the bulk of the oxide or in border traps, unambiguous estimates of trapped positive charge densities cannot be obtained with negative or positive bias TSC, with or without C-V measurements. Implications are discussed for charge trapping in radiation-hardened thermal oxides, SIMOX buried oxides, and bipolar base oxides. © 1997 IEEE.

A room Co-60 source was characterized using thermoluminescent dosimeters (TLDs) and pMOS RADFETs. Measurements were made over a range of dose rates between 0.8 and 100 mrad(Si)/s. Dose enhancement (DE) was measured using RADFETs with and without gold-flashed kovar lids. DE factors ranged from 1.05 to 2.35. A method was developed to predict dose enhancement as a function of position and test configuration. This method involves separation of direct and scattered gamma dose rate contributions. ©1997 IEEE.

The performance, reliability and radiation hardness of modern bipolar/BiCMOS devices and IC`s is limited by changes in surface recombination velocity and surface potential due to oxide-trap charge in the base oxide and near-midgap interface traps at the emitter- base/oxide interface. This report discusses how this charge trapping is enhanced by low-rate radiation as with implantation and annealing.