table of contents – issues, environments, effects

N RADIATION OWNER’S MANUAL

Table of Contents – Issues, Environments, Effects

PageRadiation Needs Today 7Providing a Unique and Cost-Effective Approach to Your Radiation Resistance Needs 8

The Growing Radiation Market 8Incorporating Radiation Design 9Dealing with an Array of Radiation Exposures 9

Product Migration 10Focusing on Military & Space Environments 11

Military Environments 11Space Environments 11

Influence of the Earth’s Magnetic Field 11Solar Activity 12Van Allen Belts 14Atmospheric Neutrons 15Man-Made Particles 16Neutron Single Event Effects (NSEE) 16

Radiation Environments 18Total Dose Environment 18Transient (Dose Rate) Radiation 18Single Event Phenomena (SEP) 19Neutron Radiation 20

Particle Interaction 20Photon Interaction Types 21Charged Particle Interactions 21Neutron Interaction Types 22

Radiation Damage Effects 19Displacement Damage 19Ionization Effects 19

Radiation Design Issues & Considerations 23Incorporating Radiation Design 23Importance of Characterization Data 23Selecting RHA Components 24Ionizing (Total Dose) Radiation & CMOS Devices 24Transient (Dose Rate) Radiation & CMOS Logic Design 25Single Event Effects & CMOS Logic Design 26Plastic Packaging, Temperature & Radiation 26Designing with Linear Bipolar Products 27Low Dose Rate Sensitivity & Bipolar Linear Design 27Shielding 28Changing Technology 30

As we enter the 21st century, significant changes willcontinue to affect the semiconductor industry.Continuous progress in processes, packaging, materials,interconnect, lithography, and many other key areas inthe production of integrated circuits will generate morecomplex devices (e.g., system on a chip), higher perfor-mance, and lower cost. CMOS technology will continueas the dominant technology of choice. Most of the rapidtechnology development is occurring in CMOS and shallcontinue into the first ten years of the new century.

These rapid changes will create many new chal-lenges, including greater susceptibility to radiation effects.In fact, as CMOS technology scales down to the 0.1µmlevel, these effects will become increasingly important.The radiation failure mechanisms that we know and

understand today may no longer be valid as a new setevolves. New sources of radiation will become evident asa result of new lithography equipment, process equip-ment, new techniques, and the use of new materials thatenable manufacture of the 0.1µm semiconductor featuresize.

The Semiconductor Industry Association’s“National Technology Roadmap for SemiconductorTechnology Needs” indicates that by the year 2006 the0.1µm process will be in production, 300mm wafers willbe preferred, and 3.5GHz will enable chip-frequency-on-chip. Because scaling down the feature size also willdecrease the gate dielectric, bits will take less energy toflip. Single-event upset will become a major radiationproblem even at sea level. Reduced power supply volt-

age also will contribute to radia-tion failure. Of continuing con-cern will be total dose degrada-tion and neutron damage in radi-ation environments.

Radiation failures will not dis-appear. In fact, some radiationeffects will become worse.National Semiconductor is a lead-ing of supplier of integrated cir-cuits to the military and aero-space community. Within thisRadiation Owner’s Manual isinformation that you can use toincrease the radiation toleranceof your systems – today and forthe future.

7

N RADIATION OWNER’S MANUAL – Issues, Environments, Effects

Radiation Needs Today

Exposure to high-energy radiation can result in transientand/or permanent changes to a semiconductor’s materialand electrical properties. Use of radiation-resistant prod-ucts, however, can prevent system malfunction or failure.

In harsh environments, such as space, the lack ofaccessibility to institute repairs as well as the space sys-tem’s longevity mandate that semiconductors used inthese systems maintain the highest reliability levels.Through all of this, electronic components must retaincomplete parametric integrity. As man ventures deeperinto space, it is increasingly necessary to harden systemsagainst the natural radiation environments of space.

Radiation effects are also a concern for tacticalapplications. Today’s rapidly changing global political cli-mate is significantly influencing the military strategies ofall countries. Regardless of political changes, nuclearweaponry remains a viable threat. As long as a first-strikecapability exists, radiation-hardened strategic and tacticalsystems will be designed.

Designing and producing radiation-hardenedsurvivable systems is time intensive and financially expen-sive. Rather than riskpremature demise,myriad precautionsmust be taken toensure that satellites,for example, will sur-vive their life expectan-cies. Because of thenecessity to orbit forten or more years,satellites incur highcosts due to emphasison performance, relia-bility, and radiationresistance. Radiationhardening entire sys-tems is a paramountconcern.

National’s Mil/Aero Radiation Effects Laboratories(REL) in South Portland, Maine, and in Santa Clara,California, certify that products are resistant to definedradiation levels. Flexible testing flows meet the needs ofstrategic, space, and tactical applications. Complete radia-tion data can be supplied with each order, providingdetailed radiation information. Customers may chooseQML (MIL-PRF-38535) Class V or Q products, QPL (MIL-PRF-38535, Appendix A), Class S or B products, orspecify SCDs to either of previous classifications of QML.Electrical requirements for post-radiation performance mayalso be specified by customers’ SCDs and may include:❏ Fully meeting pre-radiation electrical parameters❏ Meeting specified drift limits on specified

parameters❏ Meeting relaxed post-radiation electrical parameters

as defined by the customer

The Growing Radiation Market

As man-made radiation environments increase coupledwith the growing worldwide interest in exploring andemploying the knowledge of space, use of semiconduc-tors in radiation-intensive environments is expanding atan unprecedented rate.

Applications that require radiation resistance aregenerally segmented as outlined on page 9. The majorityof radiation-resistant applications are in the tactical arenaand low and high space orbits.

8

N RADIATION OWNER’S MANUAL – Issues, Environments, Effects (cont.)

RadiationResistance Level Application0 - 3 krads Commercial: industrial, robotic,

nuclear, biomedical, space shuttle3 - 30 krads Tactical: submarines, tanks, missiles,

airborne, ground (field radar, communications), space station

20 - 50 krads Space: low orbit50 - 200 krads Space: high orbit200+ krads Deep space

Strategic: military

Radiation Health Hazards❏ Ionizing radiation types

• Alpha• Beta• X-ray• Gamma ray• Neutrons

❏ Natural background• Cosmic rays – 26 mrem/yr• Food/drink – 25 mrem/yr• Medical – 20 mrem/yr• Air travel – 1mrem/2500 miles• Soil – 15 mrem/yr

❏ Man made• Nuclear power plants• Nuclear device tests• Radioactive wast product

Providing a Unique & Cost-Effective Approach to Your Radiation Resistance Needs

Incorporating Radiation Design

Most critical at the conceptional stage is a thoroughunderstanding of the system’s mission relative to itspotential radiation environment. Decisions can be madeon the use of different types of components in differentcircuit applications depending on the mission [i.e., satel-lite (commercial or military), tactical avionics (nuclearweapon), or commercial application (nuclear power plantor medical)]. Occurring at this stage are evaluation ofproper semiconductor technology, determination of theextent of shielding, and evaluation of prototype IC tech-nologies that will offer full availability by the time the sys-tem is in production.

Space systems will be exposed to radiation aswill nuclear power plants. Strategic and tactical systemsmay be exposed, and must be prepared for a nuclearevent. In the past, radiation-sensitive space systems havebeen shielded in a variety of materials. But because thepound-to-thrust cost ratio of the payload is a critical con-cern, better methods, such as radiation-resistant ICs, arerequired to harden a system. Where shielding space sys-tems is very expensive, shielding for underground tacticalradiation environments can be economical. Shielding isnot used for most tactical systems. One exception istanks.

Shielding remains a viable and economicalapproach for today’s industrial, robotic, nuclear, and bio-medical applications.

Dealing with an Array of Radiation Exposures

A system exposed to radiation must resist a variety ofpotentially lethal doses. Gamma rays, cosmic rays, neu-trons, electrons, alpha particles, and prompt dose radia-tion are ever present in space and ground environments.In large doses, all can affect standard off-the-shelf semi-conductors, influencing the dependability and longevityof the system they are in. Exposure to high-energy radia-tion can change the electrical properties of a semiconduc-tor or even destroy the device.

Initial decisions for space system design arebased on whether the use will be military or commercial.Radiation hardness requirements also depend on theorbital inclination and mission duration. Semiconductorsin a satellite experience varying degrees of radiationdegradation depending on whether they reside on thesatellite’s exterior panels or are buried within its body. Inspace environments, major exposure comes from electron/proton irradiation and cosmic rays.

When designing a tactical system (such as foraircraft, shipboard, ground hardware, or equipmenthoused in missile silos or ground bunkers), designersmust know whether that system must operate throughouta nuclear event or shut down until the event has passed.Systems subjected to a nuclear event must withstandgamma ray dose rate irradiation and neutron radiation.

The land-based commercial environment has theeasiest-to-accommodate radiation hardness levels.

9


Market Segment Current Needs

Dose Rate Dose Rate Single Event Single EventSegment Total Dose Upset Latchup Neutron Upset LatchupTactical <30 krad(Si) <1E9 rad(Si)/s <1E9 rad(Si)/s <1E13 neutron/cm2 Not applicable Not applicableAvionics <30 krad(Si) <1E9 rad(Si)/s <1E10 rad(Si)/s <1E13 neutron/cm2 <1GeV* <1GeV*Space Low Orbit (military) 20 - 50 krad(Si) Application dependent Application dependent Application dependent >40 MeV/(mg/cm2) >100 MeV/(mg/cm2)Space High Orbit (military) 100 krad(Si) - 1 Mrad(Si) Application dependent Application dependent Application dependent >40 MeV/(mg/cm2) >100 MeV/(mg/cm2)Commercial Satellites 20 - 100 krad(Si) Not applicable Not applicable Not applicable >40 MeV/(mg/cm2) >100 MeV/(mg/cm2)Strategic Systems (military) <1 Mrad(Si) <1E12 rad(Si)/s <1E12 rad(Si)/s <1E14 neutron/cm2 >40 MeV/(mg/cm2) >100 MeV/(mg/cm2)Nuclear (power plant) LOCA** 5 x 10E8rad(SI) Not applicable Not applicable Not applicable Not applicable Not applicableRobotics & Biomedical Unknown Unknown Unknown Unknown Unknown Unknown

* Single event upsets or single event latchup may occur when neutrons impact ICs at high atmospheric altitudes.** LOCA : Loss of Coolant Accident

10


Over the next ten years, integrated circuit technologywill continue its rapid pace to increase technology perfor-mance and reduce cost. National Semiconductor is com-mitted to this challenging future, particularly in support ofthe military and aerospace industry and RHA products.

The industry’s roadmap shows that by the year2006, products will have a 100nm (or 0.1 microns) featuresize that will use affordable technology design and pro-cessing. Rather than today’s six metal levels, productsusing the 100nm feature size will have fourteen layers.The increased speeds of the new technologies willapproach the fundamental limits of the wavelengths thatcan be used with this chip size and its packaging. At thepresent time, National Semiconductor has the capability ofproducing 350nm and 250nm technology. At the research

and development level, National is committing resourcesin developing 150nm technology.

Over 75% of today’s semiconductors are CMOStechnology. With its high performance, low power con-sumption, and low noise, CMOS technology also will bethe future technology of choice. Advances in CMOS tech-nology will be a driving force as advances are made toother technologies.

Semiconductor manufacturers’ knowledge of radi-ation’s adverse affects will mitigate some adverse effects inupcoming product families. But as technologies are dri-ven to smaller feature sizes and higher performance, semi-conductors will become more susceptible to new radiationfailure mechanisms – both external and local.

Although some equipment parts are exposed to severehostile radiation environments, most parts can be protect-ed with lead shielding or thick cement walls. Major con-cerns here stem from gamma ray total dose and neutronradiation.

Radiation is essentially generated from twosources:❏ Those occurring naturally via the sun, galactic and

extra-galactic, the Earth’s Van Allen Belt, and naturally-occurring radioactive materials found onEarth

❏ Those initiated by man, i.e., nuclear burst, nuclearreactors, and biomedical

System and circuit designers contend with thesemajor radiation environments:❏ Total dose ionization (gamma ray)❏ Transient irradiation (dose rate/gamma ray)❏ Single event phenomena (SEP)❏ Neutron radiation❏ Electromagnetic pulse (EMP)

The two most important space radiation effectsare Total Dose Ionization (at a low dose rate) and SingleEvent Phenomena.

In the tactical environment, transient (dose rate)radiation (primarily associated with nuclear explosion)and neutron effects are the major concerns. Neutron radi-ation is not a concern for CMOS logic radiation design aslong as the fluence is under 1014 neutrons/cm2.

At the present time, EMP environments areaddressed at the system level, not by the component’stechnology.

Product Migration

As noted, each environment contains a different radiationsource. For this discussion, the environments are limitedto those covering military and space radiation.

Military Environment

The military environment is generally equated with thenuclear explosion scenario. In a nuclear explosion at theEarth’s surface, there is a high dose rate gamma flux (typi-cally >108 rads(Si)/s), delayed total ionizing dose [lessthan 104 rad(Si)], and neutron fluence (greater than1014n/cm2). The x-rays generated at the Earth’s surfaceare absorbed by the surrounding air and are responsiblefor the large nuclear fireball. By contrast, these x-rays arenot absorbed in space and travel very long distances.

The predominant radiation effects for the militaryscenario are:❏ Displacement damage❏ Transient dose rate❏ Total ionizing dose (TID)

Additionally, military equipment must addressother detrimental nuclear effects, such as thermal shock,mechanical shock, and EMP (electromagnetic pulse).

In general, bipolar technology is more sensitiveto neutron irradiation than CMOS technology. Bipolartechnology depends upon minority carrier lifetimes andminimal lattice defects in order to meet its electrical per-formance criteria. CMOS technology is a majority carrierand surface-operation device.

Space Environment

Compared with the nuclear space environment, the naturalspace radiation environment is considered to be benign.In truth, however, space contains a number of other pow-erful radiation environments, including solar flares andinterplanetary environments. While knowledge of spaceradiation is increasing, today it remains in its infancy asradiation effects continue to be measured and investigated.

In general, the space radiation environment isaffected by numerous variables and many conditionsexcluding man-made space radiation environments.Therefore, the space radiation environment is always inconstant transformation. This discussion begins with thenear-space radiation of Earth and works outward towardthe geosynchronous/interplanetary environment.

There are fivemajor divisions within thenatural space environment.Each has its own set ofradiation effects andrequirements.❏ Low Earth orbit (LEO)❏ Mid-Earth orbit (MEO)❏ Geosynchronous Earth

orbit (GEO)❏ Geostationary orbit❏ Interplanetary

excursionsThese divisions

are dominated by twotypes of radiation – ionizing radiation andheavy ion particles. Eachtype of radiation involvescharged particles andexhibits different radiation effects that degrade the elec-tronic systems contained on board the satellite.

Influence of the Earth’s Magnetic Field (Ref. No. 4)

The following information is based on a 1997 IEEE NSRECshort course entitled Modeling Space Radiation Environments.

Permission for this use is granted by Janet Barth of the NASA/Goddard Space Flight Center.

The Earth’s magnetic field provides a natural radiationshield from charged particles, heavy ions, cosmic rays,and other minor space effects. It is the sum of two mag-netic fields – the internal magnetic field and the externalmagnetic field.

The internal magnetic field is a quasidipolarmagnetic field generated by the Earth’s core. The mag-netic North Pole of Earth is positioned approximately 76˚ north latitude and 100˚ west longitude; while the mag-netic South Pole is found at approximately 60˚ south lati-tude and 139˚ east longitude. At the higher attitudes, theinterior magnetic field strength decreases and the externalmagnetic field strength becomes a contributor to theEarth’s total magnetic field strength.

11


Focusing on Military & Space EnvironmentsRef. Nos. 2, 5, 6, 16

Trapped Particles in Space

Natural space❏ Protons❏ Electrons❏ Heavy ions

Transient radiation in space❏ Galactic cosmic ray (GCR)

particles❏ Solar events [coronal mass

ejections (CME) and solar flares]

Trapped particles – near-Earth environment❏ Van Allen Belts

• Proton• Electrons• Heavy ions

The external magnetic field strength results fromthe magnetic fields that are transported by the solar windand those fields which the solar wind induced into theEarth’s magnetosphere. The magnetosphere is created by the interaction of the solar wind and the Earth’s mag-netic field (see figure below). The external magnetic fieldis unstable in comparison with the internal magnetic field.The causes for this instability are not fully comprehended.This external magnetic field is comprised of:❏ Magnetopause current❏ Neutral sheet current❏ Ring current

Each of these magnetic currents is affected bythe solar wind activity during any given time.

The Earth’s magnetosphere is defined by itslower boundary (the ionosphere) and its upper boundary(the magnetopause – the interface between the solar windplasma and the Earth’s magnetic field). As charged parti-cles in the solar wind travel around the Earth, some of theparticles cross the Earth’s magnetic field and leak into themagnetosphere. Some charged particles are trapped bythe Earth’s magnetic field and contribute to the Van Allenbelts. Others collect in the magnetic tail and create polesof opposite charge. This produces a generator that trans-ports the charged particles along magnetic field lines atthe poles. The collection of plasma particles in the mag-netotail is the neutral plasma sheet.

Solar Activity

The Sun’s activity impacts near-Earth and interplanetaryradiation levels. The resulting solar magnetic field ishighly variable and unpredictable.

Solar activity is the result of a two-storm eventthat occurs at the Sun – solar flares and coronal massejection. The solar flares are observed as sudden bright-enings in the photosphere layer of the Sun, which islocated near the sunspot. The flares are the release ofhigh energy and involve a tearing and reconnection ofstrong magnetic field lines. The energy released from


12

The Earth’s magnetosphere withexternal field currents indicated.

Courtesy: NASA/Goddard

The Earth’s magnetosphere is formed by the interaction of the solar wind and theEarth’s magnetic field. Courtesy: NASA/Goddard

Dipole field lines calculated with internal and external field models. Courtesy: NASA/Goddard

Protons

Electrons

Aurora

these flares accelerate the particles in the solar plasma tohigher energies.

The coronal mass ejections (CME) occur in thelayer of the Sun that is outside of the photosphere, e.g.,the chromosphere. These ejections are comprised oflarge bubbles of gas and assorted magnetic fields. A plas-ma is created and released into space. While the mecha-nism for the plasma release is not fully understood, it isresponsible for significant increases in the solar windvelocity. Particle acceleration and magnetic storms atEarth are the result of the shock wave of the plasmarelease.

The corona is the outer atmosphere of the Sunand extends several solar diameters into interplanetaryspace. It continuously emits a stream of charged parti-cles, such as protons, electrons, doubly charged heliumions, and small amounts of other heavy ions. This streamis collectively called the solar wind. The high tempera-

ture of the corona causes electrons to escape the gravita-tional pull of the Sun. The electron ejections cause animbalance, which in turn results in the ejection of protonsand heavy ions from the corona. This electrically neutralplasma streams radially from the Sun. The solar windvelocity and density can vary greatly over a short periodof time.

The activity of the Sun is cyclical and maintains acertain frequency. The solar cycle has a range of 9 to 13years. The typical average is 11 years, consisting of a 7-year solar maximum phase and a 4-year solar minimumphase. The Earth’s radiation environment is dominatedby the Sun’s activity. During the solar wind maximumperiod, protons and heavier ions are accelerated that pen-etrate the Earth’s radiation belts. Solar wind particlesbecome trapped in the outer regions of these radiationbelts. Solar cycles impact the galactic cosmic ray (GCR)heavy ions. GCR levels are cyclic and correspond to theSun’s activity. Atmospheric neutrons are secondaryeffects. As they result when GCR heavy ions collide withthe oxygen and nitrogen atoms in the Earth’s atmosphere,their levels are modulated by the solar cycle. The levelsof the trapped particles are modulated by long-term varia-tions in solar activity and solar storm events.

It has been noted that major perturbations in thegeomagnetic field can occur with changes in the solarwind density (solar flares), the solar wind velocity (affect-ed by CME), and the orientation of the embedded solarmagnetic field. Disturbances made by the solar wind andits interaction with the Earth’s magnetosphere cause mag-


13

Bright rim around the Sun is the chromosphere. Courtesy: NASA/Goddard

The corona extends several solar diameters. Courtesy: NASA/GoddardBubble of gas associated with a coronal mass ejection. Courtesy: NASA/Goddard

netic storms and substorms that redistribute particles inthe Earth’s magnetosphere. Magnetic storms can causevariations in the Earth’s magnetic field from a few hoursto 10 days. Under normal magnetospheric conditions,substorms can occur every 2 to 3 hours; however, duringmagnetic storms, they occur with greater frequency andintensity.

Van Allen Belts

The near-Earth radiation environment of the Van AllenBelts consists of trapped particles and transient radiation.These belts are divided into three sections:❏ Inner Belt❏ Slot Region❏ Outer Belt

The figure (above right) depicts these sections aswell as another section – the High Latitude Horns.

The Inner Belt is centered about 1.5 Earth radii (5,973 miles); the Outer Belt, 5.0 Earth radii (19,910 miles). Between these two belts is the SlotRegion. The High Latitude Horns wrap around the InnerBelt.

The Earth’s atmosphere is the lower boundary ofthese radiation belts. The upper boundary is determinedby the strength of the geomagnetic field. This magneticfield decreases to the point where stable trapping can nolonger occur as the distance from the Earth increases.

The trapped charged particles contained in theVan Allen Belts provide a significant source of radiation,particularly Total Ionizing Dose damage. Ever present in

these belts are heavy ions, protons, and electrons. Theelectrons and protons become trapped due to the Earth’smagnetic field. This constrains their motion perpendicu-lar to the magnetic field vector, and keeps them wellwithin a defined region around the Earth.

The Lorentz electromagnetic force is responsiblefor this confinement and is a function of the particle’scharge, its velocity, and the magnetic field. As a result,the charged particles spiral around, oscillating back andforth between the Earth’s northern and southern hemi-spheres. Particles do not mirror back to exactly theiroriginating point due to the non-uniformity of the geo-magnetic field. Therefore, the third motion imposed oncharged particles is a drift vector with protons transvers-ing in the westward direction and electrons drifting in theeastward direction. This opposite drift direction is theresult of the opposing charges producing opposite spiraldirections.


14

Yearly sunspot numbers for the most recent solar cycles. Courtesy: NASA/Goddard

An artist’s representation of the Van Allen belts. Courtesy: NASA/Goddard

A cross-section of the Van Allen Belts showing the location of the trapped heavy ions.Courtesy: NASA/Goddard

The frequency of the spiralmotion (gyration) at 625 miles altitude is500kHz for low energy electrons and300MHz for low energy protons. At higherenergy levels, the period will decreasebecause of the increase in relative mass.The bounce period at 625 miles altitude is1 second for 1MeV protons and 0.1 sec-onds for 1MeV electrons. It requires 1/2hour for 1MeV protons to complete oneazimuthal drift cycle at 625 miles altitudeand about 1 hour for 1MeV electrons. Thethree motions – gyration, bounce, drift –are uncoupled because the frequencies aredifferent orders of magnitude.

Atmospheric Neutrons

In the mid-1980s, single event upsets weredetected in aircraft systems flying at highaltitude (>70,000 feet). At that time, thecause of this phenomena was not clear. Itis now known that upset rates correspond with neutronflux levels. As with protons, the recoil and secondaryproducts resulting from collisions between neutrons in theenvironment and atoms in the material near the sensitiveregion of the device are responsible for neutron singleevent upsets.

Atmospheric neutrons result from cosmic rayparticles interacting at the top of the atmosphere with thenitrogen and oxygen atoms. This complexinteraction from cosmic ray showers causesthe generation of protons, electrons, neutrons,heavy ions, muons, and pions. The neutronlevels are a function of altitude. Studies showthat at altitudes below 60,000 feet, neutronsare the dominant factor in producing SEUs.Above 70,000 feet, heavy ions from cosmicrays begin to dominate. The neutron flux isnot only a function of altitude, but also lati-tude and energy. The energy spectrum of theneutron flux ranges from keV to hundreds ofMeV. For SEU applications, only energiesgreater than 1MeV are significant.

The neutron flux is modulated bysolar events. Because galactic cosmic rays(GCR) are the primary particles that producethe secondary neutrons and protons in the

atmosphere, variations in the GCR’s intensities cause mostof the variations that are observed in the secondary neu-trons and proton levels. Neutron levels rise and fall inthe same 11-year solar cycle that modulates the GCRs.The ability of a heavy ion to penetrate the magnetosphereis determined by its magnetic rigidity. In turn, this isdependent upon the geomagnetic latitude. Magnetic dis-turbances occur more frequently during the active phase


15

A representation of the neutron environment as a function of altitude. Courtesy: NASA/Goddard

Cosmic rays hit the top of the atmosphere and disintegrate into neutrons. Courtesy: NASA/Goddard

of the solar cycle, increasing the ability of GCRs to pene-trate the magnetosphere. Atmospheric conditions, espe-cially barometric pressure, also affect neutron levels.

Man-Made Particles

Man-made particles are generated from atmospheric orexo-atmospheric explosions. The nuclear environment inspace is composed of particles created by the detonation(i.e., fission) of thermonuclear (i.e., fission-fusion)weapons. The products of the primary weapons environ-ment are neutrons, electrons, alpha particles, fission frag-ments, gamma rays, and x-rays. For atmospheric bursts,interaction with the air and ground produce more gammaradiation.

The environment from initial nuclear radiation is transient, but the effect can be transient or permanent.Transient radiation can affect both electronics and opticalmaterials. In a nuclear space explosion, the primary andsecondary gamma rays and x-rays are responsible for thetotal ionizing dose effect on the electronic components.Gamma rays also cause dose rate reactions. Neutronscause parameter degradation of electronics by disruptingthe atomic lattice structures, and can also induce singleevent effects. While shielding can help to mitigate the effect of x-ray radiation, it cannot attenuate thegamma and neutron radiation since they are extremelypenetrating.

During the late 1950s and early 1960s, the UnitedStates of America and the U.S.S.R. detonated nuclearbombs at altitudes above 125 miles. The most dramaticof these high altitude tests was the U. S. detonation onJuly 9, 1962 (program name – Starfish). Ten known satel-lites were lost because of radiation damage – some imme-diately after the explosion. The Starfish explosion inject-ed enough fission spectrum electrons with energies up to7MeV to increase the fluxes in the inner Van Allen Belt byat least a factor of 100. Effects were observed out to 5Earth radii. The Starfish electrons that became trappeddominated the inner zone environment (≈ 2.8 Earth radiiat the equator) for five years and were detectable up toeight years in some regions. The regions where particleswere trapped depends on the latitude of the explosion.At greater than 50˚ latitude, the particles appeared at geo-stationary orbits; at less than 50˚ latitude, it appeared inthe low Earth orbit domains.

Neutron Single Event Effects (NSEE)(Ref. Nos. 9, 10, 11, 12, 13)

The latest radiation environment to be addressed bydesigners of radiation-hardened systems is Neutron SingleEvent Effects (NSEE). This focus is attributable to theintroduction of COTS products into these systems as wellas the reduced feature size of new, state-of-the-art tech-nologies. Because COTS products do not consider theeffects of radiation and because newer technologiesrequire less energy to change the output of devices, NSEEis a growing concern.

The single event effects of the neutron havebeen observed and studied since the late 1970s. By themid-1980s, the effect was being observed in avionics sys-tems at 70,000 feet. NSEE are now observed at altitudesof 40,000 feet. In the near future, NSEE are expected toimpact those ground-level systems that use the latest technologies.

While NSEE generally have the same effects onsystems as protons and heavy ions (i.e., upset andlatchup), the mechanisms differ as both fast neutrons andthermal neutrons must be considered. It is believed thatlow energy neutrons are an important source of NeutronSingle Event Upsets (NSEU) and will generate NSEU moreefficiently than fast neutrons. NSEU occur when neutrons

16


cause a device to upset as a result of the recoil thatoccurs when interactions with the neutron deliver suffi-cient energy into the sensitive area of the device.

The neutron interaction and subsequent primaryrecoil of the atom’s energy result in two types of deliv-ered energy. Part of the energy is transferred to the elec-trons, resulting in direct ionization. The other deliveredenergy is referred to as the nuclear part. This nuclearportion is converted into ionization, vacancy generation,and manufacturing of phonons.

Published literature has reported that neutronscause Single Event Burnout (SEB) in power MOSFETs dueto high energetic neutrons. It also has been reported thatDynamic Random Access Memories (DRAMs) are less sus-ceptible to NSEU than are SRAMs. Therefore, DRAMsmay become the preferred memory product for avionicssystems. NSEE that occur in COTS product have low neu-tron fluxes. This effect does not appear to be an issuewith some specifically hardened product.

Hardening techniques to mitigate NSEE are beinginvestigated. More work is required to better understandthe mechanisms of NSEE, the combined environments ofneutron and total ionizing dose (TID), exposure rate, andthe sensitive parameters that exist in devices that are sub-jected to the neutron environment. It will be important tounderstand the vendor’s process in order to characterize adevice’s NSEE response.

Focus also is needed on Neutron Single EventLatchup (NSEL) since there is very little available informa-tion. While many of the mitigation techniques that weredeveloped for heavy ion susceptibility can be used forNSEE, additional techniques might need to be developedto account for NSEE mechanisms. Neither test methodsnor assurance methods are presently available in order toevaluate both present and future systems’ vulnerabilities.

17


0

0.2

0.4

0.6

0.8

1

1.2

1.4

1 2 3 4 5 6 7 8 9

Altitude (thousands of feet)

Neutron Flux versus Altitude

0 10 20 30 40 50 60 70 80

Courtesy: JaycorRef: IEEE NS, Vol. 43, pp 463, E. Normand, “Single-Event Effects in Avionics”

System Impacts Observed

System NSEE Neutron Source(s) DateICBM Guidance Computer 23 errors or 1 x 10-6 error/p/cm2 Solar flare protons 1989 Los Alamos

> 20MeV in polar regionESA ERS-1 Satellite NEC 64 Kbit CMOS memory latchup Proton AP8 MAX 1991 ESA

caused experiment shutdown SAAMilitary Avionic Units SRAM upsets 1.1 x 10-8 to 5 x 10-9 Natural 29 K ft. 45˚ 1993 BoeingE-31 AWACS NASA ER-2 upset/bit-hr. Project >1 upset/flight 65 K ft. 35˚ to 70˚

by 1998Avionics Units Commercial AC 200 to 400 upsets 19 to 46 fc Natural High Latitude and 40 to 55 K/t 1994 England/Sweden

2 to 4 upsets 0.6 to 1.5 pcAvionics Units ARINC 1 x 10-9 upsets/bit-hr. to Natural 40 K ft. 1995 BoeingReceivers & Gate Arrays 5.2 x 10-10 upsets/bit-hr. 5800 n/cm2-hr at >10MeVFermilab Computer Cray 2.3 error/mod-yr double and Natural >10 MeV 1996 boeing/IBMYMP-8 single bit error Ground levelAvionic Test Equipment (CUTE) 489 upsets in flight Natural at altitude 1998 Sweden

960 upset ground test 5 to 160 MeV, three sourcesImplanted Cardioverter 9.3 x 10-12 upsets/bit-hr. 1 DRAMs Natural, ground 1998 AustraliaDefibrillator 22 upsets, 2 multiple Level 1 >100 MeV Courtesy: Jaycor

Four of the five major radiation environments can causeconcern in the manufacture of an IC – neutron, total ion-izing dose, transient radiation effects (a.k.a., dose rate),and single event effects. An IC manufacturer can mini-mize these effects through layout, design rules, processdesign, and circuit design. At this time, the fifth majorenvironment – electromagnetic pulse (EMP) – is relegatedto system manufacturers and is addressed as part of thesystem design criteria. In the future, this environmentmay be dealt with at the IC device level.

Total Dose Environment

The total dose environ-ment is a composite ofgamma rays, x-ray radi-ation, and other ioniza-tion radiation. Thetotal amount ofabsorbed radiationenergy varies accord-ing to the absorptionmaterial. As applied tosemiconductors, thematerial is silicon diox-ide and total dose isexpressed as rad(Si) orrad(SiO2), respectively.Total dose radiation

can degrade parame-ters to the point wherea circuit’s operation isdetrimentally affected.For example, increasedpropagation time can

result. Parametric concerns include changes to ICC (stand-by current), IOZ (TRI-STATE® leakage current), VIL, VIH,VOL, and VOH. Prevention comes through understandingdegradation, how device parameters are affected, and howto achieve a radiation-hardened design by properly apply-ing the device in the circuit and the system.

Ionizing radiation affects the gate and field oxideof a CMOS semiconductor. When gamma rays strike theoxide, photons generate electron-hole pairs. The elec-trons are swept out of the oxide (gate/field) leaving the

holes behind (trapped charge). This trapped charge caus-es threshold voltages to change. Trapped positive chargeand interface-state generation combine with subsequentannealing effects to constitute time dependent effects(TDE).

Transient (Dose Rate) Radiation

In the military radiation environment, transient (high doserate) radiation is a major concern.

Transient irradiation is associated with nuclearexplosion and is a major concern for designers of tacticalequipment. Characterized by a narrow pulse width (usu-ally 3ns - 10µs) containing a total dose of 1000 rad(Si) orgreater, it represents the amount of total dose irradiationgiven in a specified time interval. Transient radiation isexpressed in rad(Si)/s or rad(SiO2)/s.

A dose rate pulse will generate excess charge ina short period of time. Adverse effects of transient irradi-ation include upset (soft error), latchup, junction burn-out, short transient pulse on the output, and saturatedoutputs which depend on the amount of photocurrent(excess charge) generated and the output loading.

Semiconductors subjected to transient radiationgenerate a large magnitude of photocurrent within each device. When a threshold level of excess charge isattained in a CMOS device, adverse effects can cause:❏ Device latchup – A parasitic SCR (silicon controlled

rectifier) effect that may cause catastrophic failure❏ Junction burnout – Also catastrophic to the device❏ Transient upset – A “soft” error that causes temporary

degradation. Upset error can be addressed through a circumvention approach, by recycling the computer,or with error detection and correction (EDAC) techniques.

Other effects are a short transient pulse on theoutput and saturated outputs. These depend upon theamount of photocurrent (i.e., excessive charge) generatedand the output loading.

The worst-case scenario for transient latchuptesting is a short radiation pulse, high VCC voltage, andhigh temperatures. This assures that the largest amountof photocurrent can be generated in the shortest amountof time and with the greatest efficiency of electron-holepair generation. The high temperature causes changes


Radiation SourcesNeutron Sources❏ Nuclear explosion❏ Nuclear reactors❏ Radioactive contaminants

Total Ionizing Dose Environment❏ Nuclear explosion❏ Space❏ X-ray machines❏ Materials❏ Charged particles❏ IC manufacturing equipment❏ Biomedical

Transient Radiation❏ Nuclear explosion

Single Event Effects❏ Galactic cosmic rays❏ Materials❏ Neutrons❏ Protons

18

Radiation EnvironmentsRef. Nos. 2, 6, 16

that permit a lower triggering currentlevel of the parasitic SCR. In turn, theparasitic SCR turns ON and the devicehas latchup.

The worst-case condition fortransient upset is a wide radiationpulse (greater than 100ns), low tem-perature (generally room tempera-ture), and VCC set at a low voltage.Slower devices have better radiationupset tolerance than faster devices.Sequential logic circuits suffer from“soft” error. Combinatorial logicdevices see a “transient” upset beforereturning to their original output state.The concern with transient upset is thatthe upset spike may be propagated to subsequent circuitswith sufficient amplitude and energy to cause a sequentialcircuit to upset.

Upset level sensitivity is also affected by:❏ The output states of the device❏ The relative position between the occurrence of the

radiation pulse and the device’s clock pulseThin Epi-CMOS, such as used in the manufacture

of FACT logic, is inherently latchup immune to transientradiation.

Single Event Phenomena (SEP)

Cosmic radiation and trapped protons areassociated with SEP. Observed in theearly 1960s, this was not a concern untilthe late 1970s. As technology evolved todecreased geometries, feature sizes, andgate oxide volume while increasingdevice speed, the energy required forgate switching was reduced. As a result,low energies (0.5 pico joules) can nowswitch device gates, making charged par-ticles an important radiation concern.

SEP hardness design is depen-dent on mission requirements and circuitapplication of the part type. Missionrequirements affecting SEP design includeorbit placement (polar or geosynchro-nous), space duration, and orbit inclina-tion. Because the Earth’s geomagneticfield acts as an energy filter, in general

minimal SEP effects occur in low altitude orbits with incli-nation angles lower than 45˚. The South AtlanticAnomaly is an exception since it is less than 45˚; however,it is a concern for SEP, particularly SEU.

Single event phenomena is generated by thesecharged particles:❏ Heavy ions are caused by solar flares and are includ-

ed in the spectrum of galactic cosmic rays. They pro-vide the worst-case SEP environment for electronicspace systems.


19

The relationship between the dipole shell parameter and cut-off rigidity is often used to determine the particleenergy required to penetrate the magnetosphere. Courtesy: NASA/Goddard

The three motions of the trapped particles form drift shells. Courtesy: NASA/Goddard

Particle interaction is categorized as:❏ Photons❏ Charged particles❏ Neutrons

Each category interacts with the target material’satoms through two processes: Excitation and ionization.The particle interaction is a function of the mass, charge,and energy of the incident particle and the atomic mass,atomic number, temperature, and density of the target material.

The photon interaction is complex and consists of:❏ Primary processes: Photoelectric, Compton

Scattering, pair production❏ Secondary processes: Compton electrons, Auger

electrons, photoelectrons, characteristic X-rays,Compton photons, etc.

❏ Tertiary processes: Electrons, photons, X-raysCharged particle interactions are primarily elec-

trons, protons, alpha particles, and heavy ions. Electronsinteract with the target materials through excitation and

20


Particle Interaction

❏ Low-energy alpha particles cause upset in sequentiallogic or memory devices. Low-energy alpha particlesare the result from the decay of radioactive materialsand cause single event effects (SEE). Thorium, aradioactive material used in ceramic packages, is asource for alpha particles.

❏ High-energy protons originate in the Van Allen Beltsor by solar flares. Only protons having energygreater than 10MeV will cause an SEE problem.

Detrimental SEP effects on electronic systemsinclude transients, soft error, and permanent damage.Latchup is the major permanent damage, with burnoutpossible in some devices. This and other effects, such asthe funnel effect, result when a high-energy charged parti-cle passes through a sensitive area.

In general terms, if a device does not suffer fromsingle event upset (SEU) or latchup in a heavy ion envi-ronment, it will be unaffected in a high-energy proton orlow-energy alpha environment. If an alpha particleupsets a component, then upset or, more likely, latchupwill occur in a proton or heavy ion environment.

Neutron Radiation

Neutron radiation-induced defects significantly affect theelectrical behavior of semiconductor devices. In bipolardevices, it causes lattice structure damage; thereby affect-ing a device’s minority carrier lifetime and transistor gain.CMOS technology is immune to neutron irradiation below 1014 neutrons/cm2.

Five basic effects occur in the silicon bandgapdue to neutron radiation:❏ Thermal generation of electron-hole pairs occurs due

to a radiation-induced defect center near the midgap.This causes dark current while increasing leakagecurrents.

❏ Recombination of electron-hole pairs is increased dueto the neutron-induced defect centers. The recombi-nation rate is determined by the defect center density,the free carrier concentration, the electron and holecapture cross-sections, and the energy level position.Recombination centers cause the recombination life-time to decrease and is the major cause for bipolartransistor gain degradation.

❏ There is temporary trapping of carriers by the neutron-induced defect centers. This process isresponsible for increasing the transfer inefficiency incharged-coupled devices.

❏ There is compensation of donors or acceptors by theneutron-induced defect centers. The result of com-pensation is “carrier” removal that causes a change inthe doping concentration. An increase in bipolar col-lector resistance can also occur.

❏ Tunneling of carriers occurs through the potentialbarrier of a bandgap energy state by means of defectenergy levels in the energy bandgap. This processincreases the reverse current in pn-junction diodes orincreases the junction leakage current in a MOSFET.

Other neutron radiation damage causes carriermobility to decrease since neutron radiation-induceddefects also act as scattering centers.

ionization. Some of the elec-trons cause Coulomb Scatteringinteraction. Electron interac-tion also generates two othereffects:❏ Charge deposition by the

electrons when they arestopped in the target material

❏ The attenuation of thepenetrating electrons inthe target material causes agradient of electron-drivencurrent density.

Protons are another source of charged particlesthat interact with the target material in a similar manner asthe electrons (e.g., excitation/ionization); but protons alsolose energy by displacement of atoms due to CoulombScattering.

Neutron interactionwith the target material isthrough either absorption orscattering. “Slow” neutronsare absorbed while “fast”neutrons are associated withthe scattering. A result of“slow” neutron absorptiongenerates the emission of analpha particle or a proton.Fast neutrons undergo two types of scattering – inelasticand elastic – depending on the energy level of the neutron.

Photon Interaction Types

Photons are bundles (i.e., packets) of energy that cover awide range of wavelengths from soft x-rays through ultraviolet (UV) and visible to infrared (IR). Photons havezero rest mass and are electrically neutral. They interactwith the target material’s atoms through three effects thatare based upon the photon’s energy and the atomic num-ber Z of the target material. The following occurs whensilicon is the target material:❏ Photoelectric effect – dominates at energies below 50keV❏ Compton scattering effect – dominates between ener-

gies greater than 50keV and less than 20MeV❏ Electron-hole pair production – dominates when

photon energy is greater than 20MeV

The photoelectric effect process states that inci-dent photon energy is completely absorbed by the emit-ted electron (photoelectron). The Compton Scatteringeffect process states that the incident photon gives uponly a portion of its energy to scatter an atomic electron.It is important to note that incident photon energy ismuch greater than the binding energy of the target materi-al’s atomic electrons. After the impact, the incident pho-ton continues through the target material, giving up a por-tion of its energy to generate another energetic Comptonelectron. This process continues until the photon energyno longer has sufficient energy to generate another ener-getic electron.

The pair production process creates an electron-positron pair. The threshold energy for this process is1.02MeV. Above this energy, an incident photon strikingthe target material can be completely absorbed and cangenerate an electron positron pair. The positron has thesame rest mass as an electron, but its charge is positive.

Charged Particle Interactions

When striking the target material, charged particles cancause displacement and ionization This interaction as well as excitation of the atomic electrons occurs as aresult of Rutherford scattering. Heavy ions are chargedparticles that emulate similar nuclear interactions as neutrons.

Rutherford scattering (also known as Coulombscattering) is caused by an incident charged particleimpinging on the target material. This causes the atomicelectrons to become excited and then liberated. Once anatomic electron is liberated, the ionization activity causesan imbalance in the semiconductor material. Excesscharged particles are created between the densities of

electrons and holes, andbegins a very complexionization process.Secondary electronswith various high ener-gy levels are subse-quently created. Thisfurther increases theionization process, anda cascading effectensues.

21


Particle Interaction withTarget Materials

The nature of interaction isdependent on the following:

❏ Particle property• Mass• Charge• Kinetic energy

❏ Target properties• Mass• Charge• Density

Particle Categories

❏ Photons• X-rays• Gamma rays

❏ Charged particles• Electrons• Protons• Alpha particles• Ions (cosmic rays)

❏ Neutrons

Irradiating Particles

❏ Neutrons❏ Gamma rays❏ X-rays❏ Electrons❏ Protons❏ Alpha particles❏ Ions❏ Cosmic rays

• Protons• Electrons• Alpha particles• Heavy ions

Two types of radiation damage effects occur in solid-stateelectronic products: Displacement damage and ionizationeffects. Displacement damage is the movement of atomsfrom their normal position in the lattice to another place-ment, causing a defect in the lattice material. Ionizationdamage effect is the generation of electron-hole pairs withinthe material that causes radiation effects. A quantity ofexcess carriers in the material results from the interactions ofcharged or uncharged particles. These excess carriers subse-quently impact the electrical performance of the electronicdevice, regardless of its passive or active state. Ionizationdamage generates temporary as well as permanent damage.

Displacement Damage

Displacement damage within the silicon material resultsfrom high-energy particles – such as a neutron – causingcrystal lattice defects. These defects are known as vacan-cies, di-vacancies, interstitials, or defect clusters. Thesedefects generate energy levels in the forbidden bandgap ofthe material lattice, causing degradation or possible failureof the integrated circuit. Bipolar products are most suscep-tible to displacement damage. The presence of the latticedefects in a bipolar device causes the minority carrier life-time to be reduced and the transistor gain to be degraded.

In addition to the displacement damage associat-ed with high energy particles, a secondary effect is causedthrough the interaction of particles (proton, neutron) withthe atomic lattice of the particular material. The resultingionization radiation can cause photocurrent generation ina semiconductor function or trapped charge in an insula-tion material of a semiconductor.

Ionization Effects

When charged particles penetrate the semiconductormaterial, the resulting ionization then causes the majorityof integrated circuit failures. The interaction of an ener-getic charged particle in the semiconductor material orinsulator causes an electron to become excited in thevalance band, then to cross through the forbidden-bandgap into the conduction band. A certain amount ofelectrons and holes that are generated will initially recom-bine. The remainder will drift through the material,assuming an electric field is present. The resulting effectsof this ionization are the generation of photocurrents orradiation-induced trapped centers. These ionizationeffects are responsible for parasitic leakage paths, mobili-ty degradation, threshold voltage shifts, transient radiationeffects, and single event effects.


22

Radiation Damage Effects

Neutron Interactions

In the case of displacement damage, when the neutron par-ticle strikes the target material it will generate defect centersthrough elastic scattering, inelastic scattering, or transmuta-tion reactions. Elastic scattering is the incident neutron hit-ting the lattice atom, causing displacement of the latticeatom and generating a vacancy/interstitial pair-defect. Theneutron continues to strike other atoms, resulting in furtherdisplaced atoms until its energy is less than the displace-ment energy (typically 25eV). Both the displaced atom andincident neutron release energy to ionization.

Inelasticscattering is the sec-ond type of interac-tion associated withdisplacement dam-age and neutronradiation. In this

case, the incident neutron is absorbed by the target atomand subsequently emits another particle at a lower ener-gy. This interaction causes the target atom to be excited,possibly displaced. At some point the excited target atomwill return to its original energy state. In doing so, thetarget atom may emit a gamma ray.

When a transmutation reaction occurs, the inci-dent neutron is captured by the lattice atom’s nucleus andsubsequently emits a charged particle (proton or alpha).The result of transmutation reaction is that the originalmaterial is transmuted into another type of material.

Radiation & Charged Particles

Category Source InteractionX-ray and gamma ray Nuclear weapon, particle collisions Photoelectric, Compton scattering, pair productionCharged particles Electrons, protons, alpha, ions Rutherford scattering, nuclear interactionsNeutrons Nuclear weapon, material Elastic scattering, inelastic scattering, transmutation

The most efficient and cost-effective approach to design-ing a radiation-hardened system is to start at the concep-tual stage. This is where proper evaluation and selectionof semiconductor technology and other factors occur, i.e.,determining the extent of shielding, selecting viable exist-ing technologies, and evaluating prototype IC technolo-gies that will be available when the system is in produc-tion. Most critical at the conceptual stage is a thoroughunderstanding of the system’s mission relative to itspotential radiation environments.

Incorporating Radiation Design

To ensure the best RHA (Radiation Hardness Assured)design, it is necessary to understand the complete radia-tion response of each component in the system circuit,e.g., what electrical parameters are affected by whichradiation environments. This includes variable and func-tionality (attribute) data to the level of radiation failure.Variable data as performed in a step-stress radiationapproach permits observance of non-monotonic behaviorfor each electrical parameter’s radiation response. Forexample, standby current attributed to a field oxide leak-age is non-linear and exhibits significant increases invalue above its pre-radiation value.

When designing a radiation-hardened system,guidelines must be based on the system’s mission andrequired survivability. After the conceptual phase comes selection of the proper components for the inves-tigative Engineering Development Phase. This is oftenthe most costly phase as testing procedures include com-ponents, circuit board, systems assembly, and softwaredocumentation.

Most older logic technologies have low radiation-tolerance limits. Their use, therefore, precipi-tates costly design-assurance techniques. Mature tech-nologies are also subject to the uncertainties associatedwith continued assured supply.

While using customized products can reduce thecost of design-assurance techniques, this one-of-a-kindapproach significantly extends leadtimes and substantiallyescalates product costs.

Once the radiation environment is identified,radiation test procedures must be set and a HardnessAssurance Plan & Program instituted. Establishing a

Change Control Board ensures that circuit design modifi-cations do not impact the system’s hardness assurance.

When a system design is approved to radiation-hardness criteria, other documentation must be initiatedto prevent compromise to the established radiation hard-ness level. Written specifications include radiation testconditions. Acceptance test procedures ensure that com-ponents identified as HCI (Hardness Critical Items) areproperly tested.

To ensure the best Radiation Hardness Assurance(RHA) design, it is necessary to understand how eachcomponent will respond in the system circuit, e.g., whatelectrical parameters are affected by which radiation envi-ronments. This includes variable and functionality(attribute) data to the level of radiation failure. Variabledata (as performed in a step-stress radiation approach)permits observance of non-monotonic behavior for theradiation response of each electrical parameter.

Part procurement drawings, assembly drawingschematics, and purchase orders require complete specifi-cations of radiation requirements, including a worst-casecircuit analysis. Worst-case analysis requires extensivesystem and circuit knowledge with respect to differentradiation environments. Factors to consider include:❏ Analysis of which circuit functions must operate

through a particular radiation environment and thosethat would not

❏ The amount of available radiation shielding❏ Selecting manufacturers with radiation-resistant

componentsPiece-part testing is one of the more costly

efforts when radiation hardening a system. Here eachcomponent’s radiation response is determined by differ-ent radiation environment simulators.

Importance of Characterization Data

By having characterization data, system designers canreview a function’s response to each radiation environ-ment and thereby tailor their approach to hardening thecircuit and system design. This data also identifies radiation-sensitive parameters, enabling designers toadjust system circuitry for minimal parametric degradationor for evaluating the product’s applicability.


23

Radiation Design Issues & Considerations

Device design margins and circuit design mar-gins are determined from characterization data. However, the most important use of characterization data is to establish parameter end-point limits for devicequalification.

For example, a device to be used in the tacticalenvironment is typically specified with a radiation level of3 krad(Si). To eliminate lot acceptance testing, a designmargin of 10x total dose level [30 krad(Si)] would have tobe attained with an acceptable parametric end-point limitand no functional failure.

Selecting RHA Components

Once a technology is selected, the next step is to chooseRHA components. Use of RHA devices reduces cost,improves reliability, and ensures that the system’s radia-tion hardware requirements will be met. RHA componentqualification does not guarantee the devices are impervi-ous to adverse radiation effects, but that they have end-point electrical values which take into account radiationdegradation.

If the system’s manufacturer determines RHAacceptability via its own radiation testing program, thesystem’s cost will significantly increase. In addition to thesystem manufacturer’s involvement in comprehensiveradiation testing, the OEM bears the burden of schedulingand purchasing costly radiation time at test facilities.

A better, less costly method is for OEMs to workwith RHA-qualified IC manufacturers. In general, RHAdevices are qualified for a neutron environment and totaldose irradiation, but are not specified for transient (doserate) environment or single event phenomena.

Concerns associated with RHA componentsinclude examining vendor’s data for outstanding issues,annealing, lot-to-lot variation, wafer-to-wafer variation,and radiation test conditions. It is important to select asemiconductor manufacturer that provides either the spec-ified data or eliminates a particular concern. By choosingthe correct technology and component manufacturer, aradiation-hardened system can be produced with minimalcost, fewer components, and maximum survivability. Thetechnology of choice to meet these radiation requirementsis a CMOS with a thin epitaxial (epi) layer or silicon-on-insulator (SOS or SOI process).

Ionizing (Total Dose) Radiation & CMOS Devices

Total ionizing dose radiation affects CMOS products intwo areas – threshold voltage shifts and radiation-inducedleakage current. The radiation creates trapped positivecharge in the gate and field oxides. A positive trappedcharge in the gate oxide causes enhancement mode n-channel MOSFETs to approach depletion while enhance-ment mode p-channel MOSFETs are driven further intoenhancement. Other generated trapped charges (a.k.a.,interface states) cause the n-channel MOSFET thresholdvoltage to increase after having decreased toward deple-tion operation. This rebound effect is a reliability con-cern for long-term space flight. P-channel MOSFETs areslightly affected by interface states. Trapped-positivecharge and interface-state charge generation combinewith subsequent annealing effects to constitute TDE oftotal dose ionization.

The two major parametric concerns are leakagecurrents and propagation times. Depending on a ven-dor’s CMOS processing, other parameters (such as VIL,VIH, VOL, VOH) may also change.

When designing a radiation-hardened CMOS sys-tem circuit, devices which use NAND gates are more tol-erant than those with NOR gates. Depending on circuitry,CMOS device response may degrade, e.g., a flip-flop composed of NAND gates has a different total dosedegradation than one using inverters and transmissiongates.


24

Radiation-induced leakage paths

As the number of inputs increases for a particu-lar gate, radiation degradation accelerates as total doselevels increase. With increased circuit complexity, radia-tion degradation shifts from parameter failure to circuitfunctionality failure. Therefore, a microprocessor may failfunctionality prior to circuit parameter failure. This is alsotrue for gate array designs. Each gate array has its ownradiation response because of the internal “personaliza-tion” of metal connections to each cell; radiation hardnesscharacteristics change with the design of each gate array.

Design margin also is important in CMOS logicdesign. It is used if a device function fails to meet orexceeds the radiation requirements of a specific applica-tion. Design margin is essentially a ratio of total doseradiation failure level of the design to the specified totaldose radiation level. Design margins are determined bystatistical analysis methods and can be applied as device,circuit, or system criteria. For CMOS devices, possessinga design margin greater than 100 requires minimal radia-tion testing. If the CMOS device’s design margin is lessthan 10, it must have lot acceptance testing and specifiedcontrols.

Transient (Dose Rate) Radiation & CMOS Logic Design

Transient irradiation is primarily associated with a nuclearexplosion and is a major concern for circuit and systemdesigners of tactical equipment. The dose rate is the timerate of change of the total dose irradiation.

The effect of the dose rate pulse is generation ofexcess charge in a short period of time. The quantity of excess charge is dependent on the total ionizing dose inthe pulse. The concentration of these excess carriers isdetermined by the dose rate and carrier lifetime. Excesscharge results when the ionizing pulse occurs at a fasterrate than can be recombined. When a threshold level ofexcess charge is attained in a CMOS device, these radiation-induced effects can cause temporary effects orcatastrophic failures:❏ Upset (soft error)❏ Latchup❏ Junction burn-out

Other effects are a short transient pulse on theoutput or saturated outputs. These depend on theamount of generated photocurrent (excess charge) andoutput loading.

Since there is no permanent damage, upset ofoutput data is a soft error. Combinatorial circuits willupset, then return to their original state. This type ofupset generates a transient voltage at the output pinwhich might or might not affect the next IC. Sequentiallogic circuits upset and remain in this condition untilreset.

While logic upset is acceptable for some pro-jects, the dose rate threshold level is important to design-ers who must work around the upset condition. Latchupoccurs when a sufficiently large quantity of radiation-induced photocurrent initiates a parasitic silicon-controlled rectifier (SCR) action. Once activated, this SCRacts as a low resistance path between ground and powersupply. This usually leads to catastrophic failure, such asblown bond wires or die metalization.

Junction burnout is another catastrophic failurein the dose rate environment. It occurs when sufficientlylarge photocurrent accumulates in the sensitive junctionand cannot be distributed quickly. As a result, thermalenergy increases to a level which causes junctionburnout. For most technologies, the junction area is fairlylarge so that heat can be dissipated. At very high doserate levels, junction burnout is a major concern.


25

When Designing with CMOS . . .

❏ Do not use NMOS technology, if possible.❏ Use enhanced mode MOSFET design.❏ Use CMOS product with thin gate oxide less than 300Å thick.❏ Use low voltage (power supply) product.❏ Use design margin of 2 or greater.❏ Use CMOS product that does not have the rebound effect with

the total ionizing dose level to be used.❏ Use circumvention schemes for dose rate and single event

environments.❏ Use current-limiting resistors when necessary to prevent

latchup. This increases power consumption as well as increas-es propagation time.

❏ Use sequential logic that uses capacitive or resistive elements.This increases RC times to prevent upset.

❏ Employ CMOS product that uses a low resistivity substrate aswell as a very thin, low resistive epi layer.

❏ To minimize leakage current of the n-channel and threshold voltage of the p-channel, do not use NOR gates. Do use NANDgates.

Single Event Effects & CMOS Logic Design

SEP results when a high-energy charged particle passesthrough a sensitive area and deposits energy along itspath. The rate of energy loss in the material is LinearEnergy Transfer (LET). This energy loss generates a plas-ma of electron-hole pairs. If this plasma occurs in adepletion area of the sensitive region, the result is genera-tion of induced current. This induced current is primarilycollected from the depletion region and the funnel region.Latchup is the major permanent damage caused by SEP.

The detrimental results of SEE on electronic sys-tems include transients, soft errors, and permanent dam-age. Single event transient spikes are generally associatedwith combinatorial logic circuits. The transient spikeresulting from a single event strike has a short time dura-tion, but could contain sufficient energy to cause a subse-quent sequential or combinatorial logic input to change.While combinatorial logic outputs have transient upset,the inputs will force the output to its original state. Softerrors are temporary single event upsets and are definedas bit flips.

Induced current is a function of circuit parame-ters, the voltage applied at the sensitive node, and nodecapacitance. The amount of charge required to generatea change of state in a memory cell or sequential logicdevice is defined as the critical charge. Associated withthe critical charge are the sensitive nodes of an IC.Sensitive nodes are the reversed-biased nodes, e.g., theOFF drains of the p- and n-channels of a memory cell.The collected charge at these sensitive nodes causes avoltage transient to develop and be applied to othercross-coupled inverters of the memory cell or sequentiallogic device that generates the change of state at the out-put of the device.

Latchup in the single event effects environmentcan affect devices manufactured on CMOS, bipolar, andECL processes. Because of heavy ions, latchup in CMOStechnologies is generally associated with a bulk technolo-gy or with devices fabricated on a thick Epi substrate.While similar to transient (dose rate), SEE latchup is gen-erated by heavy ions. Since SEE latchup usually has cata-strophic results, designers must select components thatwill be impervious to a single particle strike, e.g., deviceswhich are fabricated with guard rings, built on a very thinEPI, or that use dielectrially-isolated CMOS technologies(SOS, SOI).

Plastic Packaging, Temperature & Radiation

As space and military designers reduce system cost andweight, plastic is more frequently viewed as an alternativeto traditional ceramic packaging. While plastic appearsmore cost-effective, data describing its performance inradiation environments (total ionizing dose, neutron, doserate, and single event) is minimal.

Testing on National Semiconductor’s FACT54AC02 Quad 2-Input NOR Gate circuit was presented atthe 1995 Nuclear Space & Radiation Effects Conference(NSREC) in a paper entitled Plastic Packaging & Burn-InEffects on Ionizing Dose Response in CMOS Microcircuit.(See paper in its entirety on page 107 of this RadiationOwner’s Manual.) This investigation sought to:❏ Study burn-in effects on total ionizing dose

degradation❏ Compare the degradation differences between plastic

and ceramic packagesFor this test, all die were subjected to the same

RHA tests. Ceramic packages were assembled and finaltested to Class Q requirements. Plastic-packaged deviceswere manufactured by National Semiconductor and SEI.(Note: Different companies use varying plastic materialsand encapsulating processes that could yield significantlydifferent total ionizing dose results for the same die.)

MIL-Spec programs require high elevated-temperature-biased stresses be conducted on all military/aerospace product. In 1994 Sandia National Labs firstreported that burn-in screens affect radiation-tolerantCMOS as well as radiation-hardened technologies (“Effectsof Burn-In on Radiation Hardness”, IEEE Trans. NuclearScience, NS-41 2550 – M. R. Shaneyfelt, D. M. Fleetwood, J. R. Schwank, T. L. Meisenheimer, and P. S. Winokur).This is supported by National Semiconductor's tests thatshow non-burned-in CMOS product has less radiationdegradation than burned-in CMOS product. Because reli-ability screens such as burn-in can affect radiationresponses, these affects must be taken into considerationwhen developing the lot acceptance plan.

The effect of burn-in screens on CMOS productis increased standby leakage current over non-burned-inproduct in a total dose environment. It is believed thiseffect is caused by reduced interface trap generation. Themechanism for this reduction is not well understood.While the results stated in the 1995 NSREC paper demon-strated similar results, plastic package burned-in producthad significantly worse results than ceramic burned-in


26

product. This impacts the radiation hardness qualificationtesting when using plastic packages in the radiation environment.

Note that in the manufacture of commercial ICs,these temperature-biased stresses are performed duringthe early qualification stage to detect early mortalityissues. Once production is initiated, burn-in tests oncommercial products are only performed at set intervalswhen production line samples are evaluated under oper-ating-life test requirements. This operating-life test con-firms that the process is unchanged and is under control.

Another important conclusion is that high doserate [>10 rad(Si)/s] is the predominant mechanism forgenerating radiation-induced parasitic leakage pathbetween source and drain of the n-channel MOSFETwhile low dose rate [<1 rad(Si)/s] is responsible for a dif-ferent radiation-induced parasitic path, i.e., n-channelsource to n+ substrate. Again, these results impact RHAtest methodology and application. Parts irradiated at lowdose rate demonstrated a minimal rate of anneal after irra-diation. Parts irradiated at high dose rate had a rapidannealing rate of the standby leakage current.

The 1995 NSREC paper does not determine thephysical mechanism for increased radiation damage inplastic-packaged or burned-in product. However, it doesstate that condition A of MIL-STD-883, Method 1019 isconservative for space applications. Caution is recom-mended when employing plastic/burned-in product in atotal ionizing dose environment. Radiation test samplesshould be representative of the product being used in thesystem design.

Designing with Linear Bipolar Products

In the past, the major radiation concerns for linear bipolarproduct came from environments where neutron anddose rate were issues. The effect of neutron irradiationon linear bipolar basically affects ß (dc gain), drive cur-rent (fan-out), and VCE (saturated). There was little con-cern with total dose and Single Event Effects (SEE).

Recently, it has been noted that linear bipolardevices show total dose sensitivity to low dose rate. Thisis known as the enhanced low dose rate (ELDR) Effect.

In the single event phenomena (SEP) environ-ment, single event transient (SET) and single event burn-out (SEB) are of greatest concern for linear bipolardevices.

The hardening of bipolar circuitry generallyresults by implementing geometrical considerations aswell as adding current sources and resistors to accommo-date different radiation environments. Some radiationhardening can be acquired through a set of radiationhardening design rules. Avoiding certain design tech-niques (i.e., serpentine resistors), minimizing the use oflateral and substrate PNP transistors in the device design,and avoiding Darlington circuit design. Additional para-meters that affect bipolar performance and lead to radia-tion-induced failure are Early Voltage (VA) changes, inputimpedance, and output impedance. All result in adegraded device that has reduced output drive currentand increased load sensitivity.

Low Dose Rate Sensitivity & Bipolar Linear Design

In the past, high dose rate irradiation was used on allbipolar product to simulate the total dose environment.As with CMOS technology, this was considered to be theworst-case scenario. It is now known that low dose rateis sometimes the worst-case condition. The radiation testcommunity has investigated this effect for the past sevenyears. The phenomena is known as the Enhanced LowDose Rate (ELDR) effect. The original discovery of ELDRwas made on new technologies that use poly-emitters,but now include most bipolar technologies – both oldand new. ELDRs are observed when the linear bipolarproduct is operating in a low dose rate environment thatis similar to space or when irradiated at a low dose rate,e.g., less than 1 rad(Si)/s. A low dose rate environmentcan degrade a linear bipolar device by a factor of 2x togreater than 10x for a particular parameter as comparedwith a high dose rate environment.

This makes enhanced low dose rate effect amajor concern for space system designers who use linearbipolar product. To complicate the situation, the ELDReffect does not affect all linear bipolar products or fabrication processes – old or new. All bipolar devicesunder consideration for use in space system designshould be evaluated for enhanced low dose rate degradation.

At a low dose rate, the total dose response rateis affected by two radiation-induced failure mechanismsthat can occur simultaneously in the low dose rate envi-ronment of space [less than 1 rad(Si)/s]:❏ Time-Dependent Effects (TDE)❏ True Dose Rate Effect (TDRE)


27

TDE continues long after the linear part isremoved from the environment. For example, a satellitethat passes in and out of the Van Allen Belts is subject toTDE. However, the true dose rate effect only exists whilethe device/system is within the low dose rate environ-ment. Both failure mechanisms contribute to the degra-dation of linear bipolar parameters.

TDE can be segmented into two mechanisms:Positive oxide trapped charges (high dose rate effect) andinterface trap charges (low dose rate effect). Positiveoxide trapped charges affect standby current, DC gain,slew rate, open-loop gain, etc., and generally are not anissue in space since they are generated as a result of highdose rate irradiation. However, interface trap charge gen-eration is a major concern since it is a result of low doserate irradiation. This detractor causes reduction in DCgain, creates parasitic leakage paths, input offset currentand input bias current failures as well as input offset volt-age failures, etc.

At this time National Semiconductor’s RadiationEngineering Group is investigating the use of a high tem-perature, biased, one-week anneal to determine if lowdose rate Time Dependent Effects (namely, interface trapgeneration) exist when using a high dose rate irradiationtechnique. Essentially, the +100˚C, 168-hour biasedanneal removes the positive oxide trapped charges leav-ing only the interface traps as the sole degradation mech-anism. If parametric failures occur after this anneal, thenusers must be concerned about the system design basedupon space application, space orbit, and shielding. Iffunctional failure occurs at any time, the product must berejected.

The second cause of linear bipolar failure at lowdose rate irradiation is the True Dose Rate Effect.Degradation resulting from this effect only occurs whilethe device is in the actual low dose rate environment.The low dose rate mechanism is the lack of space chargegeneration within the particular oxide area of concern.These oxides are formed directly above the base-emitterjunctions of the transistors and are relatively radiation softto low dose rate effects. The more sensitive transistors tolow dose rate radiation are the lateral pnp and substratepnp. Since there is a lesser amount of space chargedeveloped at low dose rate, the net positive trappedoxide charge is greater at low dose rate due to a reduc-tion in electron trapping caused by the small spacechange. Greater net positive charge at low dose rate

results in increased surface recombination velocity at theemitter-base junction, causing a further increase in Ib(base current) and a reduction in the gain of the device.

Because the ELDR effect is relatively new, it isdifficult to simulate these effects without fully understand-ing the mechanisms that cause them. New approaches tosimulation are in various stages of development.

At this time there is no approved method forsimulating the True Dose Rate effect for linear bipolardevices. An approach that is being considered is to per-form high dose rate irradiation at an elevated temperaturefor simulation of the Enhanced Low Dose Rate Effect.Other methodologies are also being investigated, but nofinal conclusions have been decided.

For the true dose rate effect, a proposed methodis included in the new ASTM document, Standard Guidefor Ionizing Radiation (Total Dose) Effects Testing ofSemiconductor Devices (ASTM-F1892). Total ionizingdose irradiation allows three options:❏ Irradiate at the intended use dose rate❏ Perform the accelerated test method as described in

the document. This approach is used when it isimpractical to irradiate and test within a reasonabletime.

❏ Perform irradiation using one of two methods:1. Irradiate at room temperature and low dose rate

and apply a design margin factor of 2.2. Irradiate at +100˚C, employing a dose rate

between 1 rad(Si)/s and 10 rad(Si)/s, then applya design margin factor of 3.

National Semiconductor is investigating the use of Method 2 of the third option.

Shielding

Sometimes the product that you need just isn’t availablewith inherent radiation resistance. This is where shieldingis a consideration.

Shielding to protect integrated circuits from radi-ation environments has long been used in the militaryand aerospace industries. Over the years, shielding hasdeveloped into a science, replete with the latest simula-tion design tools and techniques. Beginning with a singlelayer material, shielding has evolved into multiple layersof high Z and low Z materials that are integrated into ICpackages. Perhaps even more important than the type ofmaterial used is understanding the radiation environmentsto which the integrated circuits will be exposed. Today’s


28

radiation hardness design techniques enable systemdesigners to use shielding as the last resort. It is only jus-tified when circuit redesign is not possible, a radiationharder part is not available, or the radiation design marginis at its minimal limit.

Shielding is employed in military systems, spacesatellites, and the nuclear commercial business. Eachapplication has unique shielding requirements and associ-ated costs. The purpose of shielding is to reduce radia-tion degradation to an acceptable level based upon theapplication, safety issues, survivability needs, and radia-tion design margins. Shielding analysis begins by evaluat-ing the need for shielding as based upon radiation envi-ronments, design margins, and restrictions. Once shield-ing needs are identified, the next step is to evaluate theavailable (inherent) shielding of the application’s environ-ment as well as additional local shielding, either on-boardor at the part level.

Shielding is easiest with stationary ground facili-ties (i.e., nuclear power plants) where earth, cement, andlead can be used readily and the predominant limitingfactor is cost. Mobile military systems are somewhatharder to shield. This is where weight, size, and cost arerestricting factors — particularly those systems that will be

directly exposed to nuclear blast or must operate throughthe radiation environment. Space systems present thegreatest obstacles for shielding. Here, weight is critical,service life is greater than five years, there is no availablemaintenance to replace failed systems, and size limitationsare essential.

Since fixed military ground installations andcommercial radiation sources can be shielded with readilyavailable material, this discussion focuses on mobile mili-tary and space systems. Note that at this time there is nopractical method to completely shield systems fromgamma rays or neutrons due to the mass of shieldingmaterial that would be required.

To provide insight to the design of the shieldconfiguration, radiation analysis for shielding mustinclude all transport codes for electron and proton envi-ronments. Another key consideration is knowing whatmaterials are located between the radiation-sensitive ICand the radiation environment. It may be possible to usethese materials as a part of the shielding. This approachcan be combined with the possibility of relocating thepart, board, or assembly to another position of the space-craft or aircraft. If these evaluations fail to provide therequired minimal radiation level, a determination may bemade to use local shielding for a sensitive part and itsconfiguration. Or the answer may be for a general massshielding to be applied to the system or LRU.

Recent investigations show multi-layer shieldingis very effective, particularly with the electron and protonenvironments of space. For example, a tri-layer shieldconfiguration can significantly reduce radiation degrada-tion. A shield configuration consisting of a high Z materi-al sandwiched between two layers of low Z material willreduce the electron and proton radiation significantly.The high Z material shields against the penetrating elec-trons while the low Z material facing the part-type or sys-tem electronics shields against the proton environment.The first layer of low Z material could be part of the sys-tem’s chassis cover or the surface of the spacecraft or air-craft. Thin layers of shielding also will have a small effectof reducing cosmic ray low energy spectrum component.

Another option to shield radiation-sensitive partis to use a special package that employs the multi-layershield concept. This shielded package is applicable forthe electron and proton environments, but does not stopthe effects of gamma or cosmic rays (heavy ions). Thesepackages are costly and trade-offs are required.


29

Typical Shielding Evaluation Process

1. Evaluate the radiation environments❏ Electron❏ Proton❏ Neutron❏ Photon❏ Cosmic rays

2. Determine the radiation requirements3. Identify the components, sub-assemblies or assemblies of

each sub-system or system that does not meet the radiationdesign margin or acceptable radiation levels

4. Determine the radiation environment within the satellite orLRU (Line Replaceable Unit) through radiation analysis ormodeling codes

5. Can inherent shielding within the satellite or LRU be used?Or is local shielding at the sensitive sub-assembly or parttype needed?

6. If shielding is required, then determine the shield configura-tion and mass needed. Determination is based upon theremaining radiation environment.

Changing Technology

Most importantly, wherever possible, select vendors thathave radiation-tolerant/radiation-hard product and use theQML approach. This will provide a standardized product,a consistently reliable product base, cost-effectiveness,and diverse product selection.

Integrated circuit technology changes aboutevery 1 1/2 to 2 years. With the decrease in feature sizesand the frequent changes in processing, it is necessary tostay in tune with the industry. Radiation damage is notgoing away with new technology advancements. In fact,in some aspects it may worsen. More testing needs to bedone in the proton and neutron upset environments.


30

table of contents – issues, environments, effects

Documents