Appendix Three: Space Radiation Effects on Integrated Circuits

Objective:

To describe the radiation environment in space, the physics of radiation interactions with solids, radiation effects on semiconductor devices, and strategies for with standing the space radiation environment.

Semiconductor devices are sensitive to many types of radiation found in the natural space radiation environment. For this reason, ICs intended for use in this environment often require special design and fabrication. Parts so designed are called radiation-hardened, or rad-hard. Although many design strategies can reduce a part's susceptibility to radiation damage, the primary determinants of that susceptibility depend upon the fabrication process. Using special processes and design methodologies, certain vendors can produce ICs that meet severe radiation environmental stresses. Rad-hard standard parts have specifications that indicate radiation exposure limits beyond which they function less efficiently or not at all. When designing ASICs, control over their design provides the opportunity for additional radiation-tolerant design approaches. Military standards such as MIL-M-38510, MIL-I-38535, and MIL-STD- 883 provide a common language for radiation concepts as well as radiation test standards.

Finding Radiation Information in This Guide

This guide provides three sections with information about how radiation affects ASICs. For information on how radiation relates to ASIC qualification, see the management section. For information on design issues regarding radiation tolerance, see Chapter 4 of the design section. Finally, for background information about the physics of radiation and how it applies to ICs in general, continue reading this appendix.

This appendix gives a brief overview of the radiation environment in space, the physics of radiation interactions with solids, radiation effects on semiconductor devices, and strategies for withstanding the space radiation environment.

The Radiation Environment in Space

In space, high-energy particles (in the MeV range) can penetrate devices and cause temporary upsets and permanent damage. Particle sources in space are the cosmic-ray background, solar flare events, and trapped particles in planetary magnetic fields. Interactions with spacecraft components cause secondary particle emissions as well. Finally, components on the spacecraft itself, such as radioactive heaters, radioisotope thermo-elastic generators, and nuclear reactors, can emit particles.

Cosmic rays originate in the outer regions of the galaxy and provide our solar system with a constant onslaught of particles from all directions. Of these particles, 85 percent are protons, 14 percent are alpha particles, and 1 percent are heavy ions. The flux 1 AU from the Earth is about 4/cm-s, with energies from almost 0 to over 10 GeV. Hydrogen, helium, carbon and oxygen ions are the bulk of the heavy ions, peaking at energies around 1 GeV, while ions with atomic numbers above 26 are rarely observed.

Solar flares represent a potential source of radiation damage. These bursts of high-energy charged particles originate from certain regions on the sun. Of these particles, over 90 percent are protons, and the rest are alpha particles, heavy ions, and electrons. These large, sudden fluxes occur sporadically, at about three times per year. The heaviest doses occur 10 to 12 years apart, according to the solar flare cycle. In a typical burst, the flux peak occurs between 2 and 24 hours after its origin and decays over a period of a few days. Heavy ion fluxes from solar flares are usually far below the galactic background, but they can be up to four orders of magnitude above the background during some flares.

Planetary magnetic fields tend to trap electrons and protons from cosmic and solar sources and, in the case of our planet, man-made sources. The strongest known radiation of this type exists in Jupiter's radiation belts. To illustrate planetary magnetic fields, we offer a description of the Earth's magnetic field.

The solar wind causes the Earth's magnetic trapping region facing the sun to be hemispherical, with a radius 10 to 12 times that of Earth, and the trapping region extends cylindrically for hundreds of Earth radii in the direction away from the sun. Van Allen belts result when the Earth catches electrons and protons gyrating around magnetic field lines. Physicists classify trapped electrons into inner and outer zones, divided at a distance of about 2.5 Earth radii from the Earth's surface. The flux of electrons in the outer zone is about 10 times that of the inner zone, with outer zone electron energies of about 7 MeV and inner zone energies of less than 5 MeV. Proton energies vary approximately inversely with altitude and can have energies over 400 MeV close to the Earth. A region in the Atlantic Ocean near Brazil contains magnetic field lines that are depressed relatively low into the Earth's atmosphere. This region, called the South Atlantic Anomaly (SAA), contains the highest concentration of radiation for low- Earth orbits. It is especially rich in proton energies and flux due to the protons' inverse relation with altitude.

The Physics of Radiation Interactions with Solids

Radiation interactions with solids are classified into three groups: photon, charged particle, and neutron interactions. Photons (X-rays and gamma rays) interact with atoms through the photoelectric effect, Compton scattering, and pair production, and in each case they generate energetic free electrons that cause secondary charged particle interactions. Those charged particles, as well as charged particles from other sources, interact through Rutherford (Coulombic) interaction or, if heavy enough, nuclear interactions. Neutrons, which interact through nuclear interactions, are not a problem in space but rather in nuclear explosions and from nuclear reactors.

Rutherford scattering liberates (ionizes) or excites atomic electrons, while nuclear interactions displace atoms in crystal lattices or transmute atoms from one element to another. These two mechanisms change the state of a solid. Two effects of these mechanisms concern the solid state electronics industry: ionization and atomic displacement.

Ionization occurs when an atom loses an excited valence electron to the conduction band. A charged particle ionizes a valence electron by liberating that electron and creating lattice vibrations that may ionize further atoms. The electron absence created by this process is called a hole. If an electric field is present, some electrons and holes immediately recombine with other atoms or become "trapped" in lattice defect sites. Physicists refer to this phenomenon as "initial recombination." While ionization produces electron absences, atomic displacement produces lattice defects.

Atomic displacement occurs when heavy particles interact with atomic nuclei, removing them from their lattice sites to new (interstitial) sites. This upsets the periodicity of the lattice, thus creating lattice defects.

Radiation Effects on Semiconductor Devices

In space, radiation effects on semiconductor devices are classified into two major types: total ionizing dose (TID) and single-event effects (SEE). TID is the accumulated effect of ionizing radiation over the lifetime of a space mission, and SEEs are transient or permanent effects due to single particles. A third type, displacement damage, is less important.

TOTAL IONIZING DOSE

Ionizing radiation primarily effects the oxide layer of MOS structures, silicon dioxide (SiO2). The oxide is a crystal lattice vulnerable to the build-up of injected ions. Engineers measure TID in terms of radiation absorbed dose or rad (1 rad = 100 ergs per gram of material).

Ionizing radiation produces electron-hole pairs in SiO2 which allow holes to cause further damage. These pairs form at the rate of 7.6 x 1012 pairs/rad(SiO2)- cm3. After initial recombination the gate immediately collects the mobile electrons leaving holes to transport to an interface, both via the electric field across the transistor. Holes travel to the Si-SiO2 interface if the field is positive. In modern devices this occurs in microseconds, but it can take seconds to hours in thicker oxides with low fields.

These positively-charged holes can cause a catastrophic negative shift in the threshold voltage of the device. A fraction (a fifth or more for unhardened devices) of the holes are trapped at the Si- SiO2 interface. If this voltage shift is great enough in the right direction, a device which is normally "off" threshold can turn "on" (or vice-versa) with 0 volts applied to it. Also, the positive charge can invert p-type isolation regions, causing leakage currents between adjacent transistors. These traps remain until they are annealed (detrapped) over a period of hours to years via effects such as electron tunneling.

Radiation also induces permanent interface traps that can shift threshold voltages. These interface traps can also hinder carrier mobility, which slows device speeds. These traps are energy levels within the Si bandgap near the Si-SiO2 interface. Their tendency to form, as with trapped holes, depends upon the fabrication process of the device. These extra energy levels form both immediately and over a period of up to a day, causing a net negative charge for n-channel metallic oxide semiconductor field effect transistors (MOSFETs) and a net positive charge for p-channel MOSFETs. Combined with the positive charge due to hole trapping, interface traps offset the hole trapping charge for n-channel devices, and the two effects add for p-channel devices.

A phenomenon known as the rebound effect significantly increases the threshold voltage. Trapped hole charge build-ups tend to dominate over interface trap charge build- ups initially, but trapped holes anneal over time as the interface traps remain. For n-channel devices, this implies a net positive charge that initially reduces the threshold voltage. But as the net positive charge lessens due to annealing of the trapped holes, the smaller effect of negative charge from the interface traps begins to dominate, thus causing "rebound." If there are few interface traps, rebound may favorably readjust the threshold voltages. Using this mechanism rebound may naturally cure devices that previously failed due to trapped hole buildup. However, if there are many interface traps, rebound may also cause the device to fail by shifting the threshold voltage from too low to too high. For p- channel devices, the threshold shifts described here move in opposite directions.

TID depends not only upon total trapped charge, but upon the rate of incoming particles as well. The interplay between hole trapping, hole annealing and interface trap buildup causes this complex dependency.

SINGLE-EVENT EFFECTS

Integrated circuits are susceptible to transient upsets (soft errors) and permanent damage (hard errors) when single high-energy particles impinge on a transistor. A single charged particle can traverse through a transistor's reverse-biased p-n junction and create an ionizing path along its line of trajectory. This causes excess charge along that path, the amount of which is called the Linear Energy Transfer (LET). The electric field across the junction is distorted along the ionizing path (this is called funneling), which increases the electric field's ability to attract the charge to one of the nodes. Eventually, the nodes collect the leftover charge through diffusion.

Soft Errors
Transient errors called single-event upsets (SEUs), which are the most common manifestations of SEEs. An SEU is an ion-induced change in the logical state of a transistor. This will happen if the LET exceeds the critical charge (QCRIT) required to change the logical state of the transistor. SEUs are temporary since they merely cause a logical error--they do not damage the chip. Once any affected registers are reset, the chip will resume correct operation. But that error has the potential to propagate and effect critical functions, causing any number of permanent problems on board the spacecraft.

Hard-Errors
SEEs can also cause errors that permanently damage the transistor. ASICs are sensitive to single-event latchups, snapbacks, and burnouts (although burnouts have not yet been observed in space). Latchups occur when a p-n-p-n or n-p-n-p CMOS transistor is traversed by a high-LET particle. If enough charge is injected into the base of the parasitic bipolars, the normally high- impedance state can become low, thereby turning the cell on permanently. Similarly, a snapback permanently turns a cell on, but without the p-n-p-n structure. For a snapback, charge collection in the p-well surrounding n-channel transistors produces a drain avalanche current and the source-to-drain voltage sustains this current. A single-event burnout (SEB) can occur when a heavy ion traverses a high voltage junction, creating a low resistance path between the gate and the substrate. This burns the cell and renders it inoperative.

DISPLACEMENT DAMAGE

Although rarely a concern for ASICs, high-energy protons--and to some extent heavy ions--can displace atoms in a crystal lattice, creating energy levels in the forbidden band gap. This can reduce the minority carrier lifetimes and thereby reduce gain by introducing new recombination centers. Typically this only concerns solar cells and wide-base bipolar transistors.

Strategies for Withstanding the Radiation Environment

Everyone involved in designing a spacecraft and its mission should be aware of ASIC radiation-tolerance issues. These designers can minimize the effects of radiation by using carefully planned trajectories, shields, rad-hard parts, and fault tolerant design methods.

Currently available tools for mission designers determine flight paths that minimize radiation exposure. These tools can ensure that a spacecraft spends the least possible amount of time in planetary radiation belts.

Radiation shielding is an integral part of any spacecraft design. Low-energy particles are trivial to shield, whereas high-energy particles become much more difficult. The best shields have low atomic number, such as carbon and aluminum. Designers often sculpt shields to work in conjunction with the spacecraft's inherent shielding, thereby minimizing mass. Engineers design many IC packages with shielding as well. Shielding can significantly reduce TID, but it can rarely affect SEEs since particles energetic enough to cause SEEs typically require shields several inches thick to be adequately attenuated. Unfortunately, shielding may also enhance TID and SEEs by slowing fast particles into energy ranges of SEE or TID sensitivity.

Given that flight path considerations and shielding cannot completely shelter electronics from radiation, designers must use rad-hard and fault tolerant subsystems as the final recourse.

Vendors and designers produce rad-hard ASICs by hardening the fabrication process and the design. Hardening the process deals mainly with hardening the oxide layer, since it is the most sensitive to radiation. Hardening the circuit design requires conservative practices such as widening design margins for stability, gain, and noise. Fan-out, slew rate, propagation delays, and other circuit parameters must account for worst-case radiation effects.

Fault tolerance includes built-in self tests, redundancy, and other methods. Built-in self tests constantly check faults so that the system can implement backup procedures immediately. Since SEUs are not permanent, designers often use redundancy techniques to incorporate repetition. By verifying that successive independent repetitions produce the same result, a system can ignore SEU- affected operations. Also, three or more redundant parts may "vote" on the correct operation, avoiding the need for repetition. For more information on "Design for Radiation Tolerance," see Section Three: Chapter 4.

REFERENCES

Kerns, Sherra E., and Shafer, B. D. (editors), "The design of radiation-hardened ICs for space: a compendium of approaches." Proceedings of the IEEE, Vol. 76, No. 11, November, 1988.

McLean, F. Barry, and Oldham, Timothy R., Basic mechanisms of radiation effects in electronic materials and devices. U.S. Army Laboratory Command, Harry Diamond Laboratories, Adelphi, Maryland,1987.

Messenger, George C., and Ash, Milton S., The effects of radiation on electronic systems. Van Nostrand Reinhold, New York, 1986.

Pease, Ronald L., Johnston, Allan H., and Azarewicz, Joseph L., "Radiation testing of semiconductor devices for space electronics." Proceedings of the IEEE, Vol. 76, No. 11, November, 1988.

Rasmussen, Robert D., "Spacecraft electronics design for radiation tolerance." Proceedings of the IEEE, Vol. 76, No. 11, November, 1988.

Stassinopulos, E.G., Brown, Dennis B., McNulty, Peter J., Pease, Ronald L., Johnston, Allan H., and Azarewicz, Joseph L., "Microelectronics for the Natural Radiation Environments of Space." IEEE International Nuclear and Space Radiation Effects Conference Short Course, Reno, Nevada, July 16, 1990.


Now you may jump to: