G. J. Brucker

Variation in SEU sensitivity of dose-imprinted CMOS SRAMs

The authors report on an experimental study of dose-induced changes in SEU (single-event-upset) sensitivity of CMOS static RAMs. Two time regimes were investigated following exposure of memories to cobalt-60 gamma rays: the near term, within a few hours after exposure, and the long term, after many days. Samples were irradiated both at room and at liquid nitrogen temperatures. The latter procedure was used in order to freeze in the damage state until SEU measurements could be made prior to annealing. Results show that memories damaged by dose are more sensitive to upsets by heavy ions. The induced changes are substantial: threshold linear energy transfer (LET) values decreased by as much as 46% and asymptotic cross sections increased by factors of 2 to 4 (unannealed samples). >

Ionization and Displacement Damage in Silicon Transistors

An investigation of electron-induced ionization and displacement damage in N-P-N and P-N-P transistors at medium and low powers is presented. Measurements of base and collector currents as a function of base-emitter voltage with radiation as a parameter were made. The transistors were irradiated sequentially with 125 keV and 1 MeV electrons. Initial saturation of surface-damage effects with 125 keV made possible the separation of displacement and surface damage. The results show that the dominant displacement damage occurs within the emitter-base transistor region for fluence levels less than or equal to 5×1015 1 MeV e/cm2 and emitter currents less than or equal to 10 milliamperes in the 2N2102 and 2N1132 transistors. Above this fluence level, displacement damage in the bulk of the base region becomes effective. Isochronal and isothermal annealing measurements indicate that the results can be explained by attributing the damage in the transistion region to recombination current losses at K-centers for N-P-N transistors and A-centers for P-N-P transistors.

Recovery of Damage in Rad-Hard MOS Devices during and after Irradiation by Electrons, Protons, Alphas, and Gamma Rays

This paper reports on the recovery properties of rad-hard MOS devices during and after irradiation by electrons, protons, alphas, and gamma rays. The results indicated that complex recovery properties controlled the damage sensitivities of the tested parts. The results also indicated that damage sensitivities depended on dose rate, total dose, supply bias, gate bias, transistor type, radiation source, and particle energy. The complex nature of these dependencies make interpretation of LSI device performance in space (exposure to entire electron and proton spectra) difficult, if not impossible, without respective ground tests and analyses. Complete recovery of n-channel shifts was observed, in some cases within hours after irradiation, with equilibrium values of threshold voltages greater than their pre-irradiation values. This effect depended on total dose, radiation source, and gate bias during exposure. In contrast, the p-channel shifts recovered only 20% within 30 days after irradiation.

Exposure-Dose-Rate-Dependence for a CMOS/SOS Memory

This study of a 1K RAM demonstrated the basic recovery characteristics previously reported in the literature for MOS capacitors and transistors. High exposure-rate-tests (i.e. 5.3 × 105 rads (Si)/(hr) indicated low-failure-doses of 1 kilorad (mean value of Pass/Fail range) in space qualified memories. The most significant failure effect was identified as due to back-channel-leakage which was traceable to starting substrates. Reduction of exposure rate from 5.3 × 105 to 5.9 rads (Si)/hr increased the minimum-failure-dose to 4.5 kilorads. This improvement could not be predicted by the use of Convolution theory.

Total-Dose and Dose-Rate Dependence of Proton Damage in MOS Devices during and after Irradiation

Rad-hard CMOS/Bulk inverters with input gates high and low were irradiated by protons with a nominal energy of 8 MeV. The purpose of this study was to determine the total-dose and doserate dependence of threshold voltage shifts, both during and immediately after irradiation and over extended periods of time thereafter. Post-rad bias conditions were the same as during bombardment. The results suggest that previous measurements [1, 2], which indicated that post irradiation changes of voltage shifts depended on type of radiation source and on energy of particles, were essentially due to the unique combinations of total dose and dose rate applied in those experiments. It now appears, that all of the damage effects observed in the previous experiments for the various sources and energies, during and after irradiation, could be simulated by selected equivalent doses and dose rates with any radiation source or particle energy. The importance of this conclusion is that a key relationship of Co60 damage effects to those from any other radiation source could be experimentally determined in terms of total dose and dose rate alone. Consequently, a valid simulation of specific charged particle and energy effects of the space radiation environment may be accomplished on the ground with appropriately designed Co60 tests.

The Damage Equivalence of Electrons, Protons, Alphas and Gamma Rays in Rad-Hard MOS Devices

This paper reports on a study of damage equivalence in rad-hard MOS devices with 105 rads (SiO2) capability. Damage sensitivities for electrons of 1, 2, 3, 5, and 7 MeV, protons of 1, 3, 7, 22, and 40 MeV, 3.4-MeV alphas, and Co-60 gammas were measured and compared. Results indicated that qualitatively the same charge recombination effects occurred in hard oxide devices for doses of 105 rads (SiO2) as in soft oxide parts for doses of 1 to 4 krads (SiO2). Consequently, damage equivalency or non-equivalency depended on radiation type and energy. However, recovery effects, both during and after irradiation, controlled relative damage sensitivity and its dependency on total dose, dose rate, supply bias, gate bias, radiation type, and energy. Correction factors can be derived from these data or from similar tests of other hard oxide types, so as to properly evaluate the combined effects of the total space environment.

Prediction and Measurement Results of Radiation Damage to CMOS Devices on Board Spacecraft

The final results from the CMOS Radiation Effects Measurement (CREM) experiment, flown on Explorer 55, are being presented and discussed, based on about 15 months of observations and measurements. Conclusions are given relating to (a) long range annealing, (b) effects of operating temperature on semiconductor performance in space, (c) biased and unbiased P-MOS device degradation, (d) unbiased n-channel device performance, (e) changes in device transconductance, and (f) difference in ionization efficiency between Co-60 gamma rays and 1 Mev Van deGraaff electrons. Additionally, the performance of devices in a heavily shielded electronic subsystem box within the spacecraft is evaluated and compared. Finally environment models and computational methods and their impact on device degradation estimates are being reviewed to determine whether they permit cost effective design of spacecraft.

Simulation of Cosmic Ray-Induced Soft Errors in CMOS/SOS Memories

A follow-up series of simulation experiments have been conducted to study cosmic ray-induced soft errors in CMOS/SOS memories. Devices were tested in beams of krypton and argon ions from the Lawrence Berkeley Laboratory (LBL) 88-inch Cyclotron at energies near 2 MeV/nucleon. The SOS test samples consisted of three versions of the TCS 146, 4K, static memory, constructed with buried contact technology. Small differences in processing have a significant effect on the electrical parameters, and these were observed to correlate with the sensitivity to cosmic ray-induced errors of the three versions. Differences in the probability of flips from 0 ¿ 1 and 1 ¿ 0 were observed. These differences are explained in terms of the geometrical details of the memory storage element and the nodal capacitances of the element. The predicted error rate in space was calculated to be 2.6×10-9 errors/ day-bit based on the observed critical charge for soft error failure.

Single Event Upset Vulnerability of Selected 4K and 16K CMOS Static RAM's

Measurements of single event upset probability for several types of prototype bulk CMOS and CMOS/SOS RAMs have been made using beams of 140 MeV krypton, 160 MeV argon and 33 MeV oxygen ions from the Lawrence Berkeley Laboratory 88-in cyclotron. Upset thresholds, determined by varying the ion species and beam-incidence angles, were used in conjunction with manufacturer-supplied information on device design parameters to estimate values of critical charge for upset. These experimental values are in reasonably good agreement with scaling predictions of Petersen et al. and Pickel. Comparison of the critical charge values deduced from experiment with predictions obtained for the 16K CMOS/SOS devices by means of SPICE and RCAP simulation codes shows discrepancies of approximately a factor of two. These discrepancies indicate the need of additional experimental and theoretical work before an adequate understanding of device response to charged particle bombardment is achieved.

High-Energy Radiation Damage in Silicon Transistors

An experimental investigation of electron and gamma ray damage in silicon transistors is presented. At low values of fluence (?e < 1014 electrons/cm2), loss in common-emitter dc current gain of medium frequency n-p-n planar transistors at collector currents of one to 10 milliamperes is attributed to changes in the surface recombination velocity. Displacement-induced recombination centers in the base region cause a reduction in gain when ?e is greater than 1014 electrons/cm2. A technique of saturating the surface damage with low energy electrons (E = 125 kev) so as to permit a separation of surface and bulk damage is demonstrated. The minority-carrier lifetime-damage constant, K?, has been estimated from the separated bulk-damage curve. It agrees with the value determined from electron irradiation of a low frequency (f?b = 1.25 Mc/sec) mesa n-p-n transistor which is shown to suffer degradation in gain only from bulk recombination current losses within the base region. Surface damage from both electrons and gamma rays is annealed at 250°C or by injecting emitter currents of 200 milliamperes which generate a high internal temperature. In contrast to this behavior, electron irradiation of p-n-p transistors caused loss in gain which is attributed to bulk damage. Damage constants, K?, determined from the data show that p-n-p transistors suffer bulk radiation damage about five times greater than n-p-n transistors.