Effects of Neutrons on Programmable Logic

Recent industry interest in neutron-induced soft errors has focused primarily on data corruption in memory devices. However, neutron-susceptible memory elements are used for configuration storage in SRAM-based FPGAs. There is a significant and growing risk of functional failure in SRAM-based FPGAs due to the corruption of configuration data. Detection and correction of these configuration upsets is not necessarily instantaneous. In fact, several thousands or millions of clock cycles may pass before the functional failure is detected. For this reason, neutron errors that affect FPGA configuration memory are referred to as “firm errors.” Additionally, schemes to detect and correct FPGA firm errors add extra complexity to the system design and significantly increase board space and bill-of-materials cost. The progression in manufacturing processes to ever deeper sub-micron technologies is increasing the risk of system reliability issues due to neutron effects to the extent that manufacturers of telecommunications and networking systems are developing qualification tests designed to identify components that are susceptible to soft errors. Neutron-induced firm errors contribute significantly to the overall system FIT rate for ground-based and airborne equipment. Introduction In the late 1970s and early 1980s, several semiconductor and system companies were experiencing higher than predicted system failure rates due to data corruption in DRAM components. After painstaking research, it was discovered that data corruption resulted from single-event upsets (SEUs) caused by the impact of high-energy neutrons with the SRAM and DRAM components. This paper investigates how these neutrons come into existence, and how they affect semiconductor components and the systems that are built from these components. Additionally, this document offers a detailed analysis on the effect of neutrons on memory cells used to configure FPGAs. Neutrons – How and Why High-energy particles from deep space and our sun (galactic cosmic rays and solar rays) collide with atoms of nitrogen and oxygen in the earth's upper atmosphere. The collisions result in the destruction of the nitrogen and oxygen atoms and the production of a variety of other high-energy particles. Most of these particles are charged and recombine quickly. However, a significant portion of the product of atmospheric collisions are neutrons. These neutrons are emitted from the collisions at very high rates and tend not to recombine with other particles, since they are not charged. They travel at high speed until they collide with atmospheric gases, objects on the earth's surface, or objects traveling through the atmosphere. The number of neutrons, called the "neutron flux," present in the atmosphere depends on several factors, with altitude being the most significant. Neutrons are attenuated by the atmospheric gases, thus decreasing the neutron flux at low altitudes. The peak neutron flux occurs at approximately 60,000 feet. Latitude is also a significant factor in neutron flux. The earth's magnetic field traps cosmic particles and prevents them from colliding with atmospheric gases. The field lines are closer together at the poles, and more cosmic particles are able to penetrate closer to the earth’s surface than at the equator. Hence, there is a significant increase in neutron flux at polar latitudes compared to equatorial latitudes. Additionally, longitude also affects the neutron flux; however, this effect is minor compared to altitude and latitude. Figure 1 on page 6 illustrates the relationship between altitude and neutron flux. Figure 2 on page 6 illustrates the neutron flux at a variety of altitudes and latitudes. It is interesting to note that the flux density is more than three times higher in Denver than it is in New York. Both cities are on approximately the same latitude, but Denver is located at a much higher altitude. Effects of Neutrons on Programmable Logic 5 Figure 1: Neutron Flux as a Function of Altitude1 Figure 2: Neutron Flux as a Function of Altitude and Latitude2 1. Tabor, A. and E. Normand,1993.“Single Event Upsets in Avionics,” IEEE Trans Nuclear Science, NS40 (2): 20. 2. “Measurement and Reporting of Alpha Particles and Terrestrial Cosmic Ray Induced Soft Errors in Semiconductor Devices,” 2001. JEDEC Standard JESD89. 0 0.2 0.4 0.6 0.8 1 1.2 1.4 10 20 30 40 50 60 70 80 Altitude (Thousands of Feet) N eu tr on F lu x 1M eV < E < 1 0M eV (n /c m 2se c)