Total Dose, Displacement Damage and Single Event Effects Compared

The three ways space radiation damages electronics, which particles cause each, which parts care, how each is measured and how each is specified

Every radiation requirement, test and mitigation belongs to one of three effect families, and a surprising number of mistakes come from applying the logic of one to another: shielding against single event effects, quoting a dose for a latchup problem, testing an optocoupler with gamma rays. This guide puts the three side by side.

1. The comparison

Total ionizing dose (TID)Displacement damage (DD)Single event effects (SEE)
What happens physicallyIonization creates charge that is trapped in oxides and at interfacesA particle knocks atoms out of the crystal lattice, creating defectsOne particle deposits a dense track of charge in a small volume
Which particlesElectrons, protons, gamma and X rays; anything that ionizesProtons, neutrons, electrons (weakly), heavy ionsHeavy ions from cosmic rays and solar events; protons through nuclear recoils
NatureCumulative, gradualCumulative, gradualRandom, instantaneous, one particle at a time
Unitrad(Si) or gray; 1 gray = 100 radFluence at a reference energy (protons per cm2 at 10 MeV or 1 MeV neutron equivalent) or displacement damage dose in MeV/gEvents per device per day; probability per mission for destructive effects
What it does to a partThreshold shifts, leakage, gain loss, timing change, eventual functional failureReduced minority carrier lifetime: gain loss, dark current, reduced light output, solar cell power lossBit flips, transients, functional hangs, latchup, burnout, gate rupture
Parts that care mostAll MOS and bipolar devices; bipolar linear parts at low dose rate especiallyOptocouplers, LEDs, lasers, photodiodes, imagers, solar cells, precision bipolar referencesMemories, processors, FPGAs, linear parts, power devices, bulk CMOS of every kind
Does shielding helpYes, strongly, for the first few millimetres of aluminiumSomewhat, for lower energy protonsEssentially no for heavy ions; some for trapped protons
Depends on orbit and durationStrongly on bothStrongly on bothRate depends on orbit and solar cycle; the total scales with duration
Ground testCobalt 60 gamma source at a controlled dose rate; low dose rate or accelerated methods for bipolar partsProton beam at one or more energies, converted with NIEL; neutrons for some casesHeavy ion accelerator, proton accelerator, pulsed laser for mechanisms
Governing documentsMIL-STD-883 TM1019, ESCC 22900, ASTM F1892ASTM E722, NASA and ESA guidelines, NIEL tablesJESD57A, ASTM F1192, ESCC 25100, JESD234, MIL-STD-750 TM1080
How the requirement is writtenA dose level with a radiation design margin, typically a factor of twoA fluence or displacement damage dose with marginThreshold LET for destructive effects; tolerable rate or probability per function for the rest

2. Total ionizing dose in a little more depth

Ionizing radiation passing through an oxide creates electron hole pairs. The electrons leave quickly; some holes are trapped in the oxide or create traps at the silicon interface. In a MOS transistor this shifts the threshold voltage and raises off state leakage, which for a large digital device shows up first as rising supply current and eventually as functional failure. In a bipolar transistor the interface traps raise base current and lower gain, which is why precision linear parts drift. Dose is deposited mainly by trapped electrons and protons in Earth orbit, so it depends steeply on altitude, on inclination and on shielding: the dose depth curve for a typical orbit falls by an order of magnitude or more in the first few millimetres of aluminium and then flattens, which is why 100 mil (2.5 mm) is the conventional reference depth. Bipolar linear parts add a complication: many degrade more per unit dose at the low dose rates of space than at the high rates of a convenient laboratory test, the enhanced low dose rate sensitivity that the test methods now address with specific low rate or accelerated procedures. The standard high rate window of 50 to 300 rad per second in the test methods, and the reasons for it, are explained on the standards page.

3. Displacement damage in a little more depth

A proton or neutron colliding with a silicon nucleus can displace it, leaving a vacancy and an interstitial that form electrically active defects. Those defects act as recombination centres, so anything whose performance depends on minority carrier lifetime degrades: the current transfer ratio of an optocoupler, the light output of a LED, the dark current of an imager, the short circuit current of a solar cell, the gain of a bipolar transistor. Most digital CMOS does not care, because MOS transistors are majority carrier devices. Damage scales with the non ionizing energy loss (NIEL) of the particle, which lets a fluence at one energy be converted to an equivalent fluence at another; the NIEL tool does that conversion and computes displacement damage dose. Because trapped protons dominate, displacement damage is worst in medium Earth orbits and matters for any orbit through the belts; it is nearly absent in geostationary orbit.

4. Single event effects in a little more depth

A single heavy ion deposits along its track a charge density thousands of times higher than anything in normal operation. If enough of that charge is collected at a sensitive node before the circuit can restore itself, the state changes; if it turns on a parasitic structure, the part latches or burns out. Because the effect comes from one particle, it does not accumulate and does not depend on how long the mission has already lasted; it happens at a rate set by the particle flux above the part's threshold and by the part's cross section. Shielding is nearly irrelevant because the galactic cosmic ray ions that carry the high LET penetrate any practical structure. The response is characterised by a cross section versus LET curve measured with heavy ions and, for parts sensitive to protons, a cross section versus proton energy. The SEE guide covers the individual effects and the rate guide the arithmetic.

5. Which environment drives which effect

Environment componentTIDDDSEE
Trapped electrons (outer belt, GEO)Dominant contributor in GEO and MEOMinorNone
Trapped protons (inner belt, SAA)Major contributor in LEO and MEODominantUpsets in parts with low thresholds; SAA passes
Galactic cosmic ray heavy ionsNegligibleNegligibleDominant for latchup, burnout and upsets in hard parts; steady background
Solar particle eventsAdds dose in bursts, significant for thin shieldingSignificant for solar cells and opticsRates rise by orders of magnitude for hours to days
Secondary particles from shieldingBremsstrahlung from electrons at depthSecondary neutrons in thick shieldingLow LET secondaries; nuclear reactions in packaging

6. The cross family mistakes

Frequently asked questions

What is a rad, and what is a krad?

A rad is 100 ergs of absorbed energy per gram of material; the material is named, so rad(Si) is dose in silicon. A krad is a thousand rad. The SI unit is the gray, 1 gray = 100 rad. Typical commercial parts fail between a few krad(Si) and tens of krad(Si); hardened parts are specified to hundreds of krad or more.

What total dose will my orbit give?

It depends on altitude, inclination, duration, solar cycle and shielding, and spans four orders of magnitude across common orbits. Behind a typical enclosure a short low Earth orbit mission sees a few krad; a geostationary mission tens of krad over its life; a medium Earth orbit through the proton belt far more. Run an environment model for your orbit; the numbers here are orientation only.

Do I need to worry about displacement damage in a digital design?

Usually not for the digital parts themselves. Check the optocouplers, the LEDs in any optical link, imagers, and precision references, which are where displacement damage shows up.

Why are single event effects specified per function and dose per part?

Dose accumulates in every part regardless of what it does, so a part level requirement makes sense. A single event's consequence depends on what the part was doing at that instant and on the surrounding design, so the requirement belongs to the function; the SEECA workbench is built on that distinction.

Related tools and pages

References

Oldham and McLean, Total Ionizing Dose Effects in MOS Oxides and Devices, IEEE Transactions on Nuclear Science 50(3), 2003. · Srour, Marshall and Marshall, Review of Displacement Damage Effects in Silicon Devices, IEEE Transactions on Nuclear Science 50(3), 2003. · Dodd and Massengill, Basic Mechanisms and Modeling of Single-Event Upset in Digital Microelectronics, IEEE Transactions on Nuclear Science 50(3), 2003. · Barth, Dyer and Stassinopoulos, Space, Atmospheric and Terrestrial Radiation Environments, IEEE Transactions on Nuclear Science 50(3), 2003. · MIL-STD-883 Test Method 1019. · Jun and others, Proton Nonionizing Energy Loss (NIEL) for Device Applications, IEEE Transactions on Nuclear Science 50(6), 2003.

More guides: COTS Parts in Space · Heavy Ion SEE Test Planning · Radiation Hardness Assurance for Commercial Space · How to Calculate a Single Event Effect Rate from Heavy Ion Test Data · Single Event Effects Explained · Glossary · All guides

Questions, corrections or a case you would like written up: the community forum or contact.