Journal DOI 10.68956/journal.human-performance-futures

Creative Commons source reading · Data quality and algorithm evaluation

Deep Neural Network-Driven Reliability Assessment Methodology for Total Ionizing Dose Effects in FDSOI MOSFETs

Deng H; Wang Y; Bi J; Jiao J; Ma Y; Chen B; Xuan Y

Originally published in Micromachines (Basel) · 2026-09-30

From the article

Selected source text. JournalOS removes markup and shortens excerpts where indicated. The full article provides the complete context.

Abstract

AbstractThis work investigates the effects of total ionizing dose (TID) on 22 nm FDSOI MOSFET devices through combined technology computer-aided design (TCAD) simulations and actual TID irradiation experiments, establishing a comprehensive dataset of radiation-induced degradation characteristics. By utilizing both simulated and experimental radiation data as training foundations, this research developed and optimized a deep neural network (DNN) capable of conducting efficient and accurate reliability analysis of TID effects under various radiation conditions, with demonstrated effectiveness in predicting device performance degradation across different TID ranges. The methodology integrates physical simulations with deep learning approaches, where TCAD simulations and irradiation experiments capture fundamental device physics under…

2. Analysis of Total Ionizing Dose Effects in FDSOI Devices: Experimental Result

A powered on-wafer X-ray irradiation experiment was performed on a 22 nm-node FDSOI nMOSFET to provide direct evidence of radiation-induced electrical degradation and to supply measurement data for TCAD calibration and validation. The device under test was a 22 nm FDSOI nMOSFET based on the GlobalFoundries 22FDX technology [24]. The X-ray source delivered a dose rate of 166.7 rad(Si) s−1, and the total dose was increased stepwise from 0 to 250 krad(Si). The same device was electrically characterized after each irradiation step, allowing the evolution of device characteristics to be tracked over the accumulated ionizing dose range. The irradiation setup is shown in Figure 1. Therefore, the transfer characteristics and electrical parameters reported in Figure 2 and Table 1 correspond to measurements from this single device, rather than…

5. Conclusions

TID effects in 22 nm FDSOI devices were investigated using X-ray irradiation measurements, a calibrated TCAD model, a regional fixed-charge representation, and a two-stage DNN framework. The TCAD analysis shows that radiation-induced holes accumulate in the BOX and near the BOX/top-silicon interface, where the resulting positive fixed charge modulates the channel potential. With increasing dose, the threshold voltage decreases, the transfer characteristics shift toward negative gate voltage, and the off-state behavior and electrical reliability degrade. Irradiation bias changes the internal electric-field distribution and therefore modifies the direction of hole transport, the location of charge accumulation, and the degree of device degradation. To address the high computational cost of the conventional Radiation model, the…

Editorial reading guide

  • What data created and tested the model, and was evaluation independent?
  • What is the cost of a false positive, false negative or out-of-distribution decision?