Keith W. Golke

Determination of funnel length from cross section versus LET measurements

Proposes an empirical model and method for determining the funnel length from heavy ion upset cross sections as a function of LET (linear energy transfer) data. It is valid for bulk technologies having a lightly doped epi region over a heavily doped substrate region. The proposed method is applied to 8 K*8 K SRAM (static random-access memory) heavy ion SEU (single event upset) test data. It is shown that the primary cause of discontinuities decreases in the upset cross section even though the surface normal effective LET has increased in the upset cross section versus surface normal effective LET curves. This is due to inappropriate conversion of the ion LET into surface normal effective LET. Previous attempts to resolve these discontinuities focused on modifying the upset cross section. The proposed model focuses on the conversion from ion LET to surface normal effective LET. The discontinuities are significantly reduced by modifying the traditionally used 1/cos conversion equation. >

Single event effects in PDSOI 4 M SRAM fabricated in UNIBOND

Heavy ion and proton single event upsets of 4 M SOI SRAM with a hardened delay element in each memory cell were evaluated. These 4 M SRAM were fabricated in UNIBOND substrates using a radiation hardened partially depleted silicon-on-insulator CMOS technology. Limiting heavy ion upset cross-section of 1.2/spl times/10/sup -10/ cm/sup 2//bit has been achieved. Limiting proton upset cross-section of 1.1/spl times/10/sup -17/ cm/sup 2//bit has also been obtained.

Characterization Summary for a Radiation Hardened 16KX1 SRAM

A radiation-hard 16kx1 SRAM with a typical access time of less than 100 ns and total dose hardness to 1E6 rads(SiO2) has been developed. Extensive radiation characterization has been performed with the aid of a test system developed to support testing in various radiation environments including total dose, dose rate upset, dose rate survival, dose rate photocurrent, and single event upset. Total dose testing of the 16kx1 included characterization of critical ac timing parameters as a function of dose as well as observation of standby current behavior as a function of array pattern. Dose rate tests were designed to explore the SRAMs sensitivities to Vdd, temperature, memory cell resistor value, dose rate pulse width, operational mode, and array pattern. Single event upset testing was also performed under a matrix of test conditions, including variations in supply voltage, temperature, memory cell resistor value, and particle angle of incidence. This paper describes 16kx1 SRAM radiation test procedures and characterization test results.

The Effect of Active Delay Element Resistance on Limiting Heavy Ion SEU Upset Cross-Sections of SOI ADE/SRAMs

This paper discusses the effective limiting heavy-ion induced single-event upset cross section of radiation-hardened 150 nm SOI SRAM and the resistance of the active-delay element (ADE) used in the memory cell. The effective limiting upset cross-section of ADE/SRAM is found to be proportional to the product of the limiting upset cross-section of the 6T SRAM of the same CMOS technology and a dimensionless function of the normalized ADE resistance. In addition, a cross-section higher than this effective limiting cross-section at high effective LET (>250 MeV-cm2/mg) is observed. This additional contribution to the cross-section is attributed to the upset caused by the high effective LET heavy ion hitting the on-transistor in the memory cell.

150nm SOI embedded SRAMs with very low SER

A split word line design technique that improves the soft error rate (SER) of high performance 150nm SOI embedded SRAMs is presented along with SER results.

A new dose rate model for SOI MOSFETs and its implementation in SPICE

A new SPICE based dose rate model is proposed for SOI MOSFETs, which accounts for collections of excess carriers by both source/body and drain/body junctions. It is also identified that the parasitic bipolar transistor does not play a significant role within the range of dose rate of interest. An implementation method for circuit level simulation is described. The validity of this model has been verified by test results.