Este trabalho apresenta uma avaliação da eficiência do dimensionamento e particionamento (folding) de transistores para a eliminação ou redução de efeitos de radiação. Durante o trabalho foi construído um modelo de transistor tipo-n MOSFET para a tecnologia 90nm, utilizando modelos preditivos. O transistor 3D modelado foi comparado com o modelo de transistor elétrico PTM level 54 da Arizona State University e os resultados mostraram uma coerência entre os dispositivos. Este transistor modelado foi irradiado por uma série de partículas que caracterizam ambientes terrestres e espaciais. Foi descoberto que a técnica de redimensionamento de transistores tem sua eficiência relacionada ao tipo de partícula do ambiente e não é aplicável em ambientes com partículas com alta energia. Descobriu-se também que aplicando o particionamento de transistores é possível reduzir a amplitude e a duração de erros transientes. A combinação do dimensionamento e o particionamento de transistores pode ser utilizada para a redução de efeitos de radiação incluindo partículas leves e pesadas. Por fim um estudo de caso foi realizado com uma célula de memória estática de 6 transistores utilizando as técnicas mencionadas anteriormente. Os resultados da célula de memória indicaram que a combinação das duas técnicas pode de fato reduzir e até impedir a mudança do estado lógico armazenado na célula. / In this work the transistor sizing and folding techniques were evaluated for SET robustness in a 90nm MOSFET technology using a 3D device model. A n-type MOSFET transistor using a 90nm technology predictive profile was modeled and functional behavior compared with PTM level 54 model showing a fit of the device with the PTM. During simulations the modeled device was irradiated in a simulation environment using particles with the profile of sea and space level ions. The radiation effects simulation had indicated that the transistor sizing can be more or less efficient to reduce SET according to the collected charge. It was found that for environments with high energy particle, transistor sizing was not able to reduce soft errors intensity. The use of folding has shown significant reduction of the amplitude and duration of the transient pulse, making this technique very useful to reduce soft errors. For alpha particles and heavy ions the combination of transistor folding and sizing had shown to be an effective combination to enhance the reliability of the circuits. A 6T SRAM cell was modeled to evaluate transistor sizing and folding techniques and the results confirmed the efficiency of folding plus sizing to reduce the effects of radiation.

Galactic-cosmic-rays (GCR) exist in space from unknown origins. A cosmic ray is a very high energy electron, proton, or heavy ion. As a GCR transverses a power semiconductor device, electron-hole-pairs (ehps) are generated along the ion track. Effects from this are referred to as single-event-effects (SEEs). A subset of a SEE is single-event burnout (SEB) which occurs when the parasitic bipolar junction transistor is triggered leading to thermal runaway. The failure mechanism is a complicated mix of photo-generated current, avalanche generated current, and activation of the inherent parasitic bipolar transistor. Current space-borne power systems lack the utility and advantages of terrestrial power systems. Vertical-double-diffused MOSFETs (VDMOS) is by far the most common power semiconductor device and are very susceptible to SEEs by their vertical structure. Modern space power switches typically require system designers to de-rate the power semiconductor switching device to account for this. Consequently, the power system suffers from increased size, cost, and decreased performance. Their switching speed is limited due to their vertical structure and cannot be used for MHz frequency applications limiting the use of modern digital electronics for space missions. Thus, the Power Semiconductor Research Laboratory at the University of Central Florida in conjunction with Sandia National Laboratories is developing a rad-hard by design lateral-double-diffused MOSFET (LDMOS). The study provides a novel in-depth physical analysis of the mechanisms that cause the LDMOS to burnout during an SEE and provides guidelines for making the LDMOS rad-hard to SEB. Total dose radiation, another important radiation effect, can cause threshold voltage shifts but is beyond the scope of this study. The devices presented have been fabricated with a known total dose radiation hard CMOS process. Single-event burnout data from simulations and experiments are presented in the study to prove the viability of using the LDMOS to replace the VDMOS for space power systems. The LDMOS is capable of higher switching speeds due to a reduced drain-gate feedback capacitance (Miller Capacitor). Since the device is lateral it is compatible with complimentary-metal-oxide-semiconductor (CMOS) processes, lowering developing time and fabrication costs. High switching frequencies permit the use of high density point-of-load conversion and provide a fast dynamic response.

Space radiation is significantly harmful to electronic Components. The operating time, duration and orbit of the space mission are affected by the characteristic of the radiation environment. The aging and the performance of the electronic components are modified by radiation. The performance of the space systems such as electronic units, sensors, power and power subsystem units, batteries, payload equipments, communication units, remote sensing instruments, data handling units, externally located units, and propulsion subsystem units is determined by the properly functioning of various electronic systems. Such systems are highly sensitive against space radiation. The space radiation can cause damage to electronic components or functional failure on the electronics. A precisely methodology is needed to ensure that space radiation is not a threat on the functionality and performance of the electronics during their operational lives. This methodology is called as &rdquo / Radiation Hardness Assurance&rdquo / . In this thesis, the hardening of electronics against space radiation is discussed. This thesis describes the space radiation environments, physical mechanisms, effects of space radiation, models of the space radiation environment, simulation of the Total Ionizing Dose, and &rdquo / Radiation Hardness Assurance&rdquo / which covers Total Ionizing Dose and Single Event Effects testing and analyzing of the electronics.

Les particules de l'environnement radiatif naturel sont responsables de dysfonctionnements dans les systèmes électroniques. Dans le cas d'applications critiques nécessitant une très haute fiabilité, il est primordial de répondre aux impératifs de sûreté de fonctionnement. Pour s'en assurer et, le cas échéant, dimensionner les protections de manière adéquate, il est nécessaire de disposer d'outils permettant d'évaluer la sensibilité de l'électronique vis-à-vis de ces perturbations.L'objectif de ce travail est le développement d'outils à destination des ingénieurs pour la prédiction des aléas logiques induits par les radiations dans les mémoires SRAM. Dans un premier temps, des bases de données de réactions nucléaires sont construites à l'aide du code de simulation Geant4. Ces bases de données sont ensuite utilisées par un outil Monte-Carlo dont les prédictions sont comparées avec des résultats d'irradiations que nous avons effectuées sur des mémoires SRAM en technologie 90 et 65 nm. Enfin, des critères simplifiés reposant sur une amélioration de la méthode SIMPA nous permettent de proposer un outil d'ingénieur pour la prédiction de la sensibilité aux protons ou aux neutrons à partir des données expérimentales ions lourds. Cette méthode est validée sur des technologies de SRAM très largement submicroniques et permet l'estimation des évènements multiples, une problématique croissante pour les applications spatiales, avioniques et terrestres. / Particles from natural radiation environment can cause malfunctions in electronic systems. In the case of critical applications involving a very high reliability, it is crucial to fulfill the requirements of dependability. To ensure this and, if necessary, to adequately design mitigations, it is important to get tools for the sensitivity assessment of electronics towards radiations.The purpose of this work is the development of prediction tools for radiation-induced soft errors, which are primarily intended for end users. In a first step, the nuclear reaction databases were built using the Geant4 toolkit. These databases were then used by a pre-existing Monte-Carlo tool which predictions were compared with experimental results performed on 90 and 65 nm SRAM devices. Finally, simplified criteria enabled us to propose an engineering tool for the prediction of the proton or neutron sensitivity from heavy ion data. This method was validated on deep submicron devices and allows the user to estimate multiple events, which are a crucial issue in space, avionic and ground applications.