This work presents a systematic study on a commercial high electron mobility transistor based on the AlGaN / GaN heterostructure (GaN HEMT). The study evaluates its robustness to different radiation doses, more specifically, its robustness to the effects of the total ionizing dose (TID) irradiated from an X-ray source with an effective energy of 10 keV. The accumulated dose varies from up to 350 krad (Si). Therefore, for this purpose, three tests were performed on the commercial transistor, GS61008T. First, the acquisition of parameters: threshold voltage (VTH), Transconductance (gm), off current (Ioff) and sub-threshold slope (S) before, during and after exposure to radiation. Then, the switching test, where the rise (tr) and fall (tf) times were aquired, pre and post irradiation, in two different frequency, 100 Hz and 100 kHz. Moreover, the temperature test, where the sample varied from 223 K (-50 ºC) to 348 K (75 ºC) to evaluate its robustness for the temperature variation after having accumulated 350 krad (Si). In addition, for a better understanding of the effects of TID on the sample, all tests were performed in two different polarization modes. The on mode (VGS = 3 V and VDS = 0 V), and the off mode (VGS = VDS = 0 V). The characteristic electrical parameters of the transistor were extracted using the characteristic curves IDxVD, IDxVG and IDxt. Curves, that were obtained using National Instrument’s PXI, with programmable sources and an X-ray diffractometer. The results showed that for the devices analyzed, for this GaN COTS the effects resulting from ionizing radiation (TID), with doses up to 350 krad (Si), are minimal, and also showed a quick and effective recovery of their electrical characteristics after annealing at room temperature, especially when irradiated polarized at on mode. Therefore, indicating that they are good candidates for use in harsh environments, as is the case of aerospace environments, particle accelerators environments and nuclear reactors.

Power MOSFETs in a harsh radiation environment are susceptible to destructive failure modes such as heavy-ion-induced Single-Event Burnout (SEB). The heavy-ion-induced SEB risk in power MOSFETs can be evaluated in ground facilities using non-destructive techniques, but performing experimental measurements with particle accelerators demands high cost, both in time and resources. A complete data analysis of SEBs usually requires the choice of a SEB triggering parameter other than the surface Linear Energy Transfer (LET), usually adopted for the SEE assessment in semiconductor devices. In this work the heavy-ion-induced SEB in low voltage power MOSFETs are studied for several ion-energy combinations and an extensive cross section analysis is carried out. The consequences of adopting distinct SEB triggering parameters, such as surface LET, local LET and accumulated ionized charge, are discussed. A methodology to extract an averaged multiparametric SEB cross section from experimental data for SEB rate prediction is shown. Based on experimental data, the heavy-ion-induced SEB worst-case response is briefly discussed.

The Total Ionizing Dose (TID) is among the most important effects that can happen due to ionizing radiation. It is caused by trapped charges inside the electrical component and can be generated by any type of ionizing radiation. In this effect, the change in behavior depends on the total dose received, which depends on the dose rate and exposure time. At the Ionizing Radiation Effects Laboratory (LERI, Centro Universitário FEI), the X-Ray source used for testing TID effects on electronic devices comes from an X-Ray diffractometer. The sample to be irradiated is fixed with a manual clamp in front of the beam path and can only be adjusted with the equipment turned off. This makes the alignment process difficult, especially when the component is far from the source. This project aims to develop and build a machine capable of moving the sample remotely and provide useful information about the testing environment such as distance and local temperature. The machine will be controlled via a custom-made computer software, where the user can move the sample with the assistance of an alignment camera. The visual system will be able to measure the distance from the sample precisely and simulate the X-Ray beam projection on the sample image.

Space-based observations have transformed our understanding of Earth and has been used to different applications. In the last years particularly, small satellites have caught the attention of the space community due to the rapid development and low to moderate budget. This happens mainly due to the use of COTS components, the possibility to be launched in groups called “swarms,” which allows a lower launch cost. However, satellites operate in a very rigorous and inhospitable environment, due to several factors such as high vacuum, high temperature variation and ionizing radiation. In regard of the ionizing radiation, satellites in Earth orbit are exposed to significant amounts of high-energy electromagnetic radiation, charged particles, X and gamma radiation and neutrons flux mostly coming from the Sun. The effects of ionizing radiation on the electronic components can be summarized in two contributions: Total Ionizing Dose (TID), and Single Events Effects (SEE). These two effects are distinct, as are the mitigation requirements and techniques associated with them. The SPORT project is a CubeSat made by an international collaborative work among institutes in Brazil and United States to investigate the space weather. From Brazil, the ITA Space Center is responsible for developing the spacecraft bus that includes, the satellite structure, the thermal subsystem, the Electrical Power Subsystem (EPS), the Attitude Determination and Control Subsystem (ADCS), the Communication Subsystem - Telemetry. Tracking, and Command (TT&C), and the On-Board Data Handling Subsystem (OBDH). In the SPORT mission there is a peculiarity in which the On-Board Data Handling Subsystem (OBDH) has three on-board computers, something that is not very common in CubeSats. One of the on-board computers is responsible for the management of the Attitude Determination and Control subsystem and housekeeping of spacecraft, the other one is responsible for the management of the payload, and a third computer is necessary for the management, storage and communication of a X-band radio transmitter, responsible to send the science data to the ground station. Because of the large amount of science data the most common microprocessor used in computers with flight heritage was not capable to handle this task, therefore, it was necessary to use a more powerful microprocessor to process the science data of SPORT mission. However, this new computer would imply the use of a component that does not have a flight heritage, which may increase the risk level of the mission. So, it was necessary to carry out a radiation analysis and tests to check out the dose levels and the sensitivity to radiation of candidate components, in order to select the best one that meets the mission requirements. Two alternative microprocessor was tested, the first one was the Atmel microprocessor the AT91SAM3X8E, in which it was tested with the Arduino Due development board kit, and the second one microprocessor was that of Texas Instrument`s which was the TM4C123F through a development board created in the laboratory. The tests carried out were aimed at characterizing the effects of Total Ionizing Dose (TID) by means of a Co-60 source that emits gamma rays, and to observe the sensitivity of these components to Single Events Effects (SEE) through neutrons sources. The objective of this work is to present the results of the tests carried out so far, and the future work which include try to identify which part of circuit or which component in the development board stopped work and try to fix it. Beyond it,a future test more representative with the complete systems close as possible of that one which will flight in the SPORT mission is considered.

The design of an application specific integrated circuit (ASIC) for the detection of ionizing particles is presented. The ASIC is a monolithic and pixelated detector, fabricated in a Silicon-On-Insulator (SOI) fabrication process, and which has the radiation-sensitive pn junction built below the Buried Oxide (BOX) and connected to the readout electronics on the top silicon film. A first low energy X-ray characterization of the detector was carried out using for this purpose fluorescence X-rays emitted be a set of target materials. A pulse-height spectra was obtained, the energy calibration of its response was carried out and the electronic noise was measured. The detector is suitable as a radiation measurement instrument on board small satellites and in harsh environments.

SAMPA is a continuous readout 32 channel chip designed to be the new frontend of CERN ALICE TPC and MCH experiments. Functional tests were positive and showed that the ASIC can be used for its final application, on the other hand, SAMPA V2 irradiation tests triggered current jumps in its digital domain. The events were linked to SELs (Single Event Latch-ups), which the root cause should be found and fixed. This work proposes a method to scan and detect possible vulnerable areas in a way to elect guilt blocks causing SELs. In this way, the DRC latch-up prevention rules were incrementally rewritten using decreasing parameters. The focus was the substrate polarization rules, evidencing possible vulnerable SCR areas prone to SEL. The technique was applied in SAMPA including standard cells and IPs. Major differences were found in a specific SRAM that was replaced. New irradiations did not trigger SELs, indicating the effectiveness of the method, saving time and improving design reliability.

Effects on electronic devices caused by interactions of ionizing radiation are a main concern in several fields, such as aerospace applications and high-energy physics. Among the heavy-ion induced radiation effects are the Single Event Effects, in which a strike of a single ion can be enough to generate an observable effect, as a result of the high energy deposition and thus electron-hole pairs generation. To perform SEE experiments in low-energy particle accelerators, it is necessary to take into account linear energy transfer changes as the particle penetrates in the medium. The SAFIIRA facility at Open Laboratory for Nuclear Physics in Sao Paulo was used for experiments to investigate the effects of isolation and metalization layers in charge collection and observable events generation in an analog and in a digital device, in order to establish proper methodologies for precise studies of radiation effects mechanisms. The analog device studied was a p-MOS transistor, from which the heavy-ion impact induced current signal was analized to obtain cross-section and collected charge, allowing to determine metalization layer thickness to be 1.28(2) microns, and charge collection dependency on particle LET and range, varying from 6.0 to 11.0 microns.

The poster represents a methodology to quickly estimate robustness of flip-flops to radiation using layout-oriented SPICE in 90nm BCD technology. Continuous downscaling of devices has led to increased damaging effects of radiations on electronic storage devices. With reduced node capacitance, the critical charge required to flip the data in a memory element is also shrinking. We present a layout-oriented SPICE tool that can be used to estimate reliability of a design in comparatively less turnaround time. It can be used to compare various mask layouts corresponding to a single design incorporating the effect of radiation on nearby nodes. The radiation modelling has been further improved by including incident angle of the radiation for varying its linear energy of transfer. The proposed method enables fast iterative design of layouts of critical storage elements. Finally, we designed three flip-flops for comparison of their PPA and robustness to select a suitable application domain.

Electronics are increasingly susceptible to energetic particle interactions within the silicon. In order to improve the circuit reliability under radiation effects, several hardening techniques have been adopted in the design flow of VLSI systems. This work proposes a methodology based on the Single-Event Transient (SET) input dependence and signal probability estimation in order to improve the radiation hardness of logic circuits. Signal probability propagation is used to assign the highest probability to the least sensitive input combination of the circuit through pin swapping. The cell optimisation can reach up to 48% reduction on the soft-error rate at no cost in the circuit design area. Additionally, as the pin swapping is a layout-friendly technique, the optimisation does not impact on the cell placement and it can be adopted along with other hardening techniques in the logic and physical synthesis. Additionally, the efficiency of well-know RHBD techniques such as Gate Sizing (GS) and Transistor Stacking (TS) have shown a high input dependence. Thus, signal probability can also be used to enhance the hardening efficiency of RHBD techniques. It was shown that by disregarding the input signal, the hardened designs can show a higher SET cross-section than the unhardened ones. Thus, besides improving the radiation hardening of the techniques, the proposed pin swapping provides a more application-specific hardening analysis. Hence, the hardening technique or circuit selection can be used more efficiently and avoid misleading qualifications. To conclude, the proposed methodology can also provide further information in order to optimise selective Triple-Modular Redundancy (TMR) block insertion strategies in the most critical nodes of a circuit based on its signal probability and the input dependence of the majority voter architectures.

The advancement in the integrated circuit design has provided more robust technologies to radiation effects. However, the soft error susceptibility remains a significant challenge due to the demand for higher operating frequencies and low-power circuits, besides the manufacturing process variability. This work explores different transistor arrangements for a set of logic functions at the layout level to evaluate the SET response under the process variability. Also, an evaluation of the SET sensitivity trends for a circuit-level benchmark, exploring the voltage scaling influence, and adopting decoupling cells, as a radiation-hardening technique, is performed using the 7 nm FinFET technology. Results show that the multi-level logic topology is up to 88% more robust against the SET effects than the conventional complex gate topology for the evaluated cells. All the analyzed logic functions independently of the topology are more sensitive to the SETs considering the process variability impact. Moreover, the near-threshold operation increases the fault rate by at least 53.2%. For the evaluated benchmark, decoupling cells' insertion is more efficient at 0.6V and 0.7V, reaching up to 10.4% on fault rate reduction.

Generally, the radiation robustness of SRAM cells is evaluated considering the cell on the hold state. High-performance systems increase the number of Read and Write access operations in the memories. This, together with the technology scaling, increase the probability of Soft Errors affect the SRAM architecture at any time, including during a reading and writing operations. This work evaluates the Soft Error sensitivity of the complete SRAM architecture. The radiation faults are investigated through propagation in cells, bitlines, and half-selection state, during the Hold, Read, and Write operations. The 6T, 8T, 9T, 8TSER, and DICE SRAM cells are compared. DICE cell presents the highest robustness to the radiation effects, but showed a loss of robustness during the execution of operations. The half-selection situation shows potential to increase the cell sensitivity on 50% compared to the Hold perspective. The 8T and 9T maintained hold robustness during half-selection.

Radiation is a significant threat for reliability of electronic devices, especially those aimed at space applications and human-made radiation environments. Ionizing particle strikes in sensitive areas of the IC may cause a current disturbance, leading to soft errors. Different techniques, as Triple Modular Redundancy (TMR), are used to increase robustness and reliability electronic and computational systems. In this work, a mixed-signal system was tested under heavy ions. Different voting schemes were evaluated in a fault-tolerance data acquisition system with Design Diversity TMR technique. The system was implemented in a Programmable System-on-Chip - PSoC 5LP, from Cypress Semiconductor, manufactured in a 130 nm CMOS technology. The design under test (DUT) was a redundant Data Acquisition System (DAS), comprising three 8-bit ADCs operating in parallel: two Successive Approximation Register (SAR) converters and a ΣΔ converter. With two distinct architectures among the three hardware copies, hardware diversity was achieved. Simultaneously, temporal diversity was implemented by using different sampling rates of the SAR ADCs (740 ksps and 74 ksps). The system also comprises two software-based voters: one main spatial voter and a temporal voter, which also performs the coarse synchronization of the DAS. Two implementations for the temporal voter were tested, one using the bit by bit technique (1st experiment) and the other with a word voter architecture identical to the main voter (2nd experiment). During the irradiation time, only 4 errors were observed in the temporal voter with word voter compared to 18 errors in the first experiment with bit by bit voter. The main voter has a faster voting, and no errors were observed. Different approximate software-based voting schemes presented distinct reliability under heavy ions. Voting time and algorithm complexity were key factors in voter reliability. In this scheme, word voters based on mutual subtraction showed to be more reliable than the bit by bit voter.

Field Programmable Gate Arrays mission reconfigurability capabilities are crucial assets for many critical applications. However, state-of-the-art Radiation-Hardening methods for FPGAs have high costs, limiting the functionalities and environment ionizing particle flux exposure. They consist of logic triplication, memories reinforcement, and bitstream scrubbing with partial reconfiguration. These methods involve a 3x reduction of maximal design capacity and an average time-in-error proportional to design sizes. To reduce these costs, we propose an alternative: Smart-Redundancy, a new method based on the detection of possible events via process and hardware modifications. Thanks to integrated particle sensors, only dual redundancy is required. Results show up to 50% improvement in MDC over actual Rad-Hard methods and an average TIE decrease of at least 10,000x compared to bitstream’s scrubbing, at an avoidable cost of 41% in area using a commercial 40nm technology node.

The use of neural networks in a plethora of critical applications has led to the development of new Neural Network Accelerators (NNA) driving the computation towards the edge. However, the edge consists of highly constrained resources with limited power budget and area footprint. At the same time, in environments with high irradiation flux, reliability has become very crucial and requires a trade-off between area/power due to error mitigation mechanisms. In this paper, we evaluate the vulnerability of the tiny-TPU edge accelerator, which is based on a systolic array design. We use an FPGA-based emulation fault injection framework to execute Artificial Neural Networks (ANNs) using a well-known dataset, and study the vulnerability of each component of the accelerator based on the impact of classification of the ANN, to convey the most critical components which require protection/fault mitigation.

Machine learning (ML) algorithms have been regaining momentum thanks to their ability to analyze substantial and complex data, supporting artificial intelligence decisions in cloud computing but also in near-sensor computing in endpoint devices. Both cloud and near-sensor computing are liable to radiation-induced soft errors, especially in automotive and aerospace safety-critical applications. In this regard, this paper contributes by comparing the accuracy of two prominent machine learning algorithms running on a low-power processor upset by radiation-induced soft errors. Both ML algorithms have been assessed with the help of a fault injection-based method able to natively emulate soft errors directly in a development board. In addition, neutron radiation test results suggest the most critical situations in which mitigation solutions should address.