The influence of the metric of linear energy transfer （LET） on single event upset （SEU）, particularly multiple bit upset （MBU） in a hypothetical 90-nm static random access memory （SRAM） is explored. To explain the odd point of higher LET incident ion but induced lower cross section in the curve of SEU cross section, MBUs induced by incident ions 132Xe and 2~9Bi with the same LET but different energies at oblique incidence are investigated using multi-functional package for single event effect analysis （MUFPSA）. In addition, a comprehensive analytical model of the radial track structure is incorporated into MUFPSA, which is a complementation for assessing and interpreting MBU susceptibility of SRAM. The results show that （i） with the increase of incident angle, MBU multiplicity and probability each present an increasing trend; （ii） due to the higher ion relative velocity and longer range of ~ electrons, higher energy ions trigger the MBU with less probability than lower energy ions.
Experimental evidence is presented relevant to the angular dependences of multiple-bit upset （MBU） rates and patterns in static random access memories （SRAMs） under heavy ion irradiation. The single event upset （SEU） cross sections under tilted ion strikes are overestimated by 23.9%-84.6%, compared with under normally incident ion with the equivalent linear energy transfer （LET） value of 41 MeV/（mg/cm2）, which can be partially explained by the fact that the MBU rate for tilted ions of 30° is 8.5%-9.8% higher than for normally incident ions. While at a lower LET of - 9.5 MeV/（mg/cm2）, no clear discrepancy is observed. Moreover, since the ion trajectories at normal and tilted incidences are different, the predominant double-bit upset （DBU） patterns measured are different in both conditions. Those differences depend on the LET values of heavy ions and devices under test. Thus, effective LET method should be used carefully in ground-based testing of single event effects （SEE） sensitivity, especially in MBU-sensitive devices.
Using a Monte Carlo simulation tool of the multi-functional package for SEEs Analysis （MUFPSA）, we study the temporal characteristics of ion-velocity susceptibility to the single event upset （SEU） effect, including the deposited energy, traversed time within the device, and profile of the current pulse. The results show that the averaged dposited energy decreases with the increase of the ion-velocity, and incident ions of 2~9Bi have a wider distribution of energy deposition than 132Xe at the same ion-velocity. Additionally, the traversed time presents an obvious decreasing trend with the increase of ion-velocity. Concurrently, ion-velocity certainly has an influence on the current pulse and then it presents a particular regularity. The detailed discussion is conducted to estimate the relevant linear energy transfer （LET） of incident ions and the SEU cross section of the testing device from experiment and simulation and to critically consider the metric of LET.
Graphene and thin graphite films deposited on SiO2/Si are irradiated by swift heavy ions(209Bi, 9.5 Me V/u) with the fluences in a range of 1011ions/cm2–1012ions/cm2 at room temperature. Both pristine and irradiated samples are investigated by Raman spectroscopy. For pristine graphite films, the 'blue shift' of 2D bond and the 'red shift' of G bond with the decrease of thickness are found in the Raman spectra. For both irradiated graphene and thin graphite films, the disorder-induced D peak and D' peak are detected at the fluence above a threshold Φth. The thinner the film, the lower the Φthis. In this work, the graphite films thicker than 60 nm reveal defect free via the absence of a D bond signal under the swift heavy ion irradiation till the fluence of 2.6 × 1012ions/cm2. For graphite films thinner than 6 nm, the area ratios between D peak and G peak increase sharply with reducing film thickness. It concludes that it is much easier to induce defects in thinner films than in thicker ones by swift heavy ions. The intensities of the D peak and D' peak increase with increasing ion fluence, which predicts the continuous impacting of swift heavy ions can lead to the increasing of defects in samples. Different defect types are detected in graphite films of different thickness values. The main defect types are discussed via the various intensity ratios between the D peak and D' peak(HD/HD).
Highly oriented pyrolytic graphites are irradiated with 40.5-Me V and 67.7-Me V ^112Sn-ions in a wide range of fluences： 1×10^11 ions/cm^2–1×10^14ions/cm^2. Raman spectra in the region between 1200 cm^-1 and 3500cm^-1 show that the disorder induced by Sn-ions increases with ion fluence increasing. However, for the same fluence, the amount of disorder is greater for 40.5-Me V Sn-ions than that observed for 67.7-Me V Sn-ions, even though the latter has a slightly higher value for electronic energy loss. This is explained by the ion velocity effect. Importantly, ～ 3-cm^-1frequency shift toward lower wavenumber for the D band and ～ 6-cm^-1 shift toward lower wavenumber for the 2D band are observed at a fluence of 1×10^14 ions/cm^2, which is consistent with the scenario of radiation-induced strain. The strain formation is interpreted in the context of inelastic thermal spike model, and the change of the 2D band shape at high ion fluence is explained by the accumulation of stacking faults of the graphene layers activated by radiation-induced strain around ion tracks. Moreover,the hexagonal structure around the ion tracks is observed by scanning tunneling microscopy, which confirms that the strains near the ion tracks locally cause electronic decoupling of neighboring graphene layers.
The limits of previous methods prompt us to design a new approach （named PRESTACE） to predict proton single event effect （SEE） cross-sections using heavy-ion test data. To more realistically simulate the SEE mechanisms, we adopt Geant4 and a location-dependent strategy to describe the physics processes and the sensitivity of the device. Cross-sections predicted by PRESTAGE for over twenty devices are compared with the measured data. Evidence shows that PRESTACE can calculate not only single event upsets induced by indirect proton ionization, but also direct ionization effects and single event latch-ups. Most of the PRESTAGE calculated results agree with the experimental data within a factor of 2 3.
Geant4 tools were used to model the single event upset （SEU） of static random access memory cells induced by heavy ion irradiation. Simulated results obtained in two different regions of incident ion energies have been compared in order to observe the SEU occurrence by energetic ions and their effects on the radial ionization profile of deposited energy density. The disagreement of SEU cross sections of device response and radial distribution of deposited energy density have been observed in both low energy and high energy regions with equal linear energy transfer （LET） which correspond to the both sides of the Bragg peak. In the low energy region, SEUs induced by heavy ions are more dependent upon the incident ion species and radial distribution of deposited energy density, as compared with the high energy region. In addition, the velocity effect of the incident ion in silicon in the high energy region provides valuable feedback for gaining insight into the occurrence of SEU.
Monte Carlo simulation results are reported on the single event upset(SEU) triggered by the direct ionization effect of low-energy proton. The SEU cross-sections on the 45 nm static random access memory(SRAM) were compared with previous research work, which not only validated the simulation approach used herein, but also exposed the existence of saturated cross-section and the multiple bit upsets(MBUs) when the incident energy was less than 1 MeV. Additionally, it was observed that the saturated cross-section and MBUs are involved with energy loss and critical charge. The amount of deposited charge and the distribution with respect to the critical charge as the supplemental evidence are discussed.
We investigate the impact of heavy ion irradiation on a hypothetical static random access memory （SRAM） device. Influences of the irradiation angle, critical charge, drain-drain spacing, and dimension of device structure on the device sensitivity have been studied. These prediction and simulated results are interpreted with MUFPSA, a Monte Carlo code based on Geant4. The results show that the orientation of ion beams and device with different critical charge exert indis- pensable effects on multiple-bit upsets （MBUs）, and that with the decrease in spacing distance between adjacent cells or the dimension of the cells, the device is more susceptible to single event effect, especially to MBUs at oblique incidence.
