the cosmic ray environment of tactical abms

7
546 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 52, NO. 2, APRIL 2005 The Cosmic Ray Environment of Tactical ABMs John R. Solin, Member, IEEE, Allan J. Tylka, Margaret A. Shea, and Don F. Smart Abstract—The battlespace of tactical antiballistic missiles (TABMs) is mostly exoatmospheric and includes regions with negligible geomagnetic shielding, especially during geomagnetic storms. It is therefore necessary to harden TABMs against galactic cosmic ray (GCR) and solar particle event (SPE) induced single event effects (SEEs) and star tracker and focal plane array (FPA) glitches. The variability of the worldwide GCR and SPE exoatmo- spheric TABM environment is described. Index Terms—Atmospheric neutrons, avionics single event upset (SEU), single event effect, SEU, solar particle event. I. INTRODUCTION T HE galactic cosmic ray (GCR) and solar particle event (SPE) requirements for national missile defense (NMD) tactical antiballistic missiles (TABMs) are in review [1]. NMD satellites are already required to operate through even very large SPEs. This report derives the worldwide GCR and SPE envi- ronments on which a general set of TABM requirements can be based. The region in which a TABM tracks and intercepts a target is termed the battlespace. The battlespace of TABMs is mostly above the atmosphere and includes high latitudes where geo- magnetic shielding is always negligible and mid latitudes where the shielding disappears during large geomagnetic storms. In those regions the GCR and SPE fluxes are nearly the same as at geosynchronous Earth orbit (GEO). TABMs may use single event upset (SEU) sensitive, commercial-off-the-shelf (COTS) ICs. The vulnerability of satellite launch vehicles to SPEs was discovered only after COTS ICs were incorporated. Lockheed Martin now prohibits launches if the forecasted solar proton flux at energies greater than 10 MeV, exceeds 10 cm s sr [2]. The European Space Agency does basically the same [3]. Furthermore, while in the past only major powers could time an attack to coincide with an SPE, now everyone has access to websites that give real-time reports and forecasts of SPEs and geomagnetic storms. Considerable media attention was recently given to those services during a series of SPEs and geomagnetic storms that occurred over October 26 to November 6, 2003. At least 28 satellites were temporarily or permanently disabled Manuscript received October 28, 2004; revised December 18, 2004. This work has been approved for unlimited public release by the Defense Threat Re- duction Agency, DTRA Approval Ref. 04-S-2508. J. R. Solin is with Lockheed Martin Space Systems Co., San Jose, CA 95150- 7871 USA (e-mail: [email protected]). A. J. Tylka is with the U.S. Naval Research Laboratory, Washington, DC 20375-5352 USA (e-mail: [email protected]). M. A. Shea and D. F. Smart are with the U.S. Air Force Research Laboratory, Bedford, MA 01731-3010 USA (e-mail: [email protected]). Digital Object Identifier 10.1109/TNS.2005.846883 [4]. Many satellites were intentionally placed in safemode for protection. Space-weather forecasting capabilities are currently being expanded to support ongoing operations and NASA’s Space Exploration Initiative. II. EFFECTS OF ATMOSPHERIC AND GEOMAGNETIC SHIELDING ON TABM GCR AND SPE FLUXES Atmospheric attenuation of GCR and SPE fluxes is negli- gible above 40 km (the exoatmosphere) and modest down to 30 km [5]–[7]. At lower altitudes single event effects (SEEs) re- sult from the attenuated primary flux and secondary particles produced by the GCR and SPE fluxes. However, we neglect boost phase SEEs here. Maps of the vertical cutoff rigidity at 450 km are shown in Figs. 1 and 2 for quiet geomagnetic con- ditions and for a large geomagnetic storm [8]–[10]. The greater the value of , the greater the shielding against GCR and SPE fluxes. 1 decreases with increasing altitude, so the actual ver- tical cutoffs at any given latitude and longitude will be some- what larger or smaller than the values in the maps, depending upon altitude within the battlespace. Geomagnetic storms occur with or without SPEs. However, very large SPEs and nonrecurrent geomagnetic storms are both caused by fast coronal mass ejections (CMEs). Historically, the largest solar particle increases have occurred when the CME ar- rives at Earth. Thus, one is simultaneously faced with a very se- vere interplanetary particle environment and a maximal reduc- tion in geomagnetic shielding. This is exactly what occurred on October 20, 1989, March 24, 1991, November 6, 2001, and Oc- tober 29, 2003, the four largest increases in MeV protons observed near Earth since the start of nearly-continuous inter- planetary monitoring in 1973. III. CALCULATION OF GCR AND SPE SPECTRA To calculate the GCR and SPE spectra in TABM battlespaces over various regions of the Earth, we made use of a CREME96 [11] option that allows a user to create and upload geomagnetic transmission functions (GTFs). Normally one supplies an orbit and CREME96 calculates the GTF for the orbit. In the compu- tations here we adopt the approximation that an unattenuated energetic particle directed at an exoatmospheric location, pen- etrates to the location if the rigidity of the particle is greater 1 The ability of a particle to penetrate to a location in the Earth’s geomagnetic field is a function of the particle’s rigidity (momentum/charge), shadowing by the Earth, and the geomagnetic cutoff rigidity at that location, in the direction of the particle’s angle of arrival. Cutoffs depend on latitude, longitude, altitude, direction, and also geomagnetic disturbance levels. The vertical cutoff rigidity at a location provides a rough average over arrival directions. 0018-9499/$20.00 © 2005 IEEE

Upload: df

Post on 25-Sep-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The cosmic ray environment of tactical ABMs

546 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 52, NO. 2, APRIL 2005

The Cosmic Ray Environment of Tactical ABMsJohn R. Solin, Member, IEEE, Allan J. Tylka, Margaret A. Shea, and Don F. Smart

Abstract—The battlespace of tactical antiballistic missiles(TABMs) is mostly exoatmospheric and includes regions withnegligible geomagnetic shielding, especially during geomagneticstorms. It is therefore necessary to harden TABMs against galacticcosmic ray (GCR) and solar particle event (SPE) induced singleevent effects (SEEs) and star tracker and focal plane array (FPA)glitches. The variability of the worldwide GCR and SPE exoatmo-spheric TABM environment is described.

Index Terms—Atmospheric neutrons, avionics single event upset(SEU), single event effect, SEU, solar particle event.

I. INTRODUCTION

THE galactic cosmic ray (GCR) and solar particle event(SPE) requirements for national missile defense (NMD)

tactical antiballistic missiles (TABMs) are in review [1]. NMDsatellites are already required to operate through even very largeSPEs. This report derives the worldwide GCR and SPE envi-ronments on which a general set of TABM requirements can bebased.

The region in which a TABM tracks and intercepts a targetis termed the battlespace. The battlespace of TABMs is mostlyabove the atmosphere and includes high latitudes where geo-magnetic shielding is always negligible and mid latitudes wherethe shielding disappears during large geomagnetic storms. Inthose regions the GCR and SPE fluxes are nearly the same asat geosynchronous Earth orbit (GEO). TABMs may use singleevent upset (SEU) sensitive, commercial-off-the-shelf (COTS)ICs. The vulnerability of satellite launch vehicles to SPEs wasdiscovered only after COTS ICs were incorporated. LockheedMartin now prohibits launches if the forecasted solar protonflux at energies greater than 10 MeV, exceeds 10 cm s sr[2]. The European Space Agency does basically the same [3].Furthermore, while in the past only major powers could timean attack to coincide with an SPE, now everyone has access towebsites that give real-time reports and forecasts of SPEs andgeomagnetic storms. Considerable media attention was recentlygiven to those services during a series of SPEs and geomagneticstorms that occurred over October 26 to November 6, 2003.At least 28 satellites were temporarily or permanently disabled

Manuscript received October 28, 2004; revised December 18, 2004. Thiswork has been approved for unlimited public release by the Defense Threat Re-duction Agency, DTRA Approval Ref. 04-S-2508.

J. R. Solin is with Lockheed Martin Space Systems Co., San Jose, CA 95150-7871 USA (e-mail: [email protected]).

A. J. Tylka is with the U.S. Naval Research Laboratory, Washington, DC20375-5352 USA (e-mail: [email protected]).

M. A. Shea and D. F. Smart are with the U.S. Air Force Research Laboratory,Bedford, MA 01731-3010 USA (e-mail: [email protected]).

Digital Object Identifier 10.1109/TNS.2005.846883

[4]. Many satellites were intentionally placed in safemode forprotection. Space-weather forecasting capabilities are currentlybeing expanded to support ongoing operations and NASA’sSpace Exploration Initiative.

II. EFFECTS OF ATMOSPHERIC AND GEOMAGNETIC SHIELDING

ON TABM GCR AND SPE FLUXES

Atmospheric attenuation of GCR and SPE fluxes is negli-gible above 40 km (the exoatmosphere) and modest down to30 km [5]–[7]. At lower altitudes single event effects (SEEs) re-sult from the attenuated primary flux and secondary particlesproduced by the GCR and SPE fluxes. However, we neglectboost phase SEEs here. Maps of the vertical cutoff rigidityat 450 km are shown in Figs. 1 and 2 for quiet geomagnetic con-ditions and for a large geomagnetic storm [8]–[10]. The greaterthe value of , the greater the shielding against GCR and SPEfluxes.1 decreases with increasing altitude, so the actual ver-tical cutoffs at any given latitude and longitude will be some-what larger or smaller than the values in the maps, dependingupon altitude within the battlespace.

Geomagnetic storms occur with or without SPEs. However,very large SPEs and nonrecurrent geomagnetic storms are bothcaused by fast coronal mass ejections (CMEs). Historically, thelargest solar particle increases have occurred when the CME ar-rives at Earth. Thus, one is simultaneously faced with a very se-vere interplanetary particle environment and a maximal reduc-tion in geomagnetic shielding. This is exactly what occurred onOctober 20, 1989, March 24, 1991, November 6, 2001, and Oc-tober 29, 2003, the four largest increases in MeV protonsobserved near Earth since the start of nearly-continuous inter-planetary monitoring in 1973.

III. CALCULATION OF GCR AND SPE SPECTRA

To calculate the GCR and SPE spectra in TABM battlespacesover various regions of the Earth, we made use of a CREME96[11] option that allows a user to create and upload geomagnetictransmission functions (GTFs). Normally one supplies an orbitand CREME96 calculates the GTF for the orbit. In the compu-tations here we adopt the approximation that an unattenuatedenergetic particle directed at an exoatmospheric location, pen-etrates to the location if the rigidity of the particle is greater

1The ability of a particle to penetrate to a location in the Earth’s geomagneticfield is a function of the particle’s rigidity (momentum/charge), shadowing bythe Earth, and the geomagnetic cutoff rigidity at that location, in the directionof the particle’s angle of arrival. Cutoffs depend on latitude, longitude, altitude,direction, and also geomagnetic disturbance levels. The vertical cutoff rigidityR at a location provides a rough average over arrival directions.

0018-9499/$20.00 © 2005 IEEE

Page 2: The cosmic ray environment of tactical ABMs

SOLIN et al.: THE COSMIC RAY ENVIRONMENT OF TACTICAL ABMs 547

Fig. 1. Vertical geomagnetic cutoffs [GV] at 450 km during periods of no disturbance.

Fig. 2. Vertical geomagnetic cutoffs [GV] at 450 km during a large geomagnetic storm (Dst = �500 nT).

than the at the location and the incident path is not shad-owed by the Earth. Otherwise the particle is filtered out.2 Tofind the GCR and very large SPE linear energy transfer (LET)spectra at locations with a given vertical cutoff rigidity , weemployed the CREME96 models for solar-minimum GCRs andpeak 5-minute-averaged solar energetic particles. We filtered

2It should be noted that these calculations are somewhat artificial and intendedto give a general idea of the radiation environment as a function of location ina battlespace. At any given location, the geomagnetic cutoff is not really a stepfunction. The cutoff is somewhat lower than the vertical cutoff for particles ar-riving from westerly directions, and higher for particles arriving from easterlydirections. A complete calculation would involve averaging over all arrival di-rections. For SPE intensities, which drop steeply with increasing energy, it isalso important to average over the actual shielding distribution, rather than justusing a nominal 100 mils, as is customarily done for design studies.

these fluxes with a GTF that was 0 for particle rigidities lessthan the given and equal to 0.5 (for shadowing) for particlerigidities greater than , and then transmitted the flux throughshielding and calculated its LET spectrum in the normal way.Our results for the GCR and very large SPE integral LET spectraof elements 1 to 92 behind 100 mils ( mm) aluminum(Al) shielding are presented in Figs. 3 and 4. The SPE flux inFig. 4 does not include the background GCR flux. The fluxeswere computed using a 1 MeV/nucleon cutoff. Our results forthe integral proton energy spectra of a very large SPE are pre-sented in Fig. 5. In our shielding calculations, we treat the fluxesas isotropic.

Trapped protons can also induce SEEs and glitches. Fig. 6shows the average trapped proton intensities at 450 km [12].

Page 3: The cosmic ray environment of tactical ABMs

548 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 52, NO. 2, APRIL 2005

Fig. 3. Worldwide particle fluxes in TABM battlespaces during periods of no solar activity.

Fig. 4. Worldwide particle fluxes in TABM battlespaces during a very large SPE.

These intensities are for protons above 22.5 MeV, roughly theenergy needed to penetrate 100 mils of Al shielding. From thisfigure, one might infer that trapped protons are irrelevant forTABM operations. However, during large geomagnetic storms,high-energy trapped and quasitrapped protons can appear in re-gions where they are not ordinarily found [13].

IV. SEE RATES OVER NORTH AMERICA AND EUROPE

Tools at the CREME96 website [14] can be used to computethe SEU rate of an IC. Here, however, we only seek to give afeel for the SEU rates; so we will use a well-known analysismethod, the figure of merit (FOM) method [15] and published

Page 4: The cosmic ray environment of tactical ABMs

SOLIN et al.: THE COSMIC RAY ENVIRONMENT OF TACTICAL ABMs 549

Fig. 5. Very large SPE proton fluxes in worldwide TABM battlespaces.

Fig. 6. Worldwide trapped proton fluxes in TABM battlespaces.

results [16], in order to compute SEU rates of generic parts inan environment of general interest, namely, battlespaces wherethere is minimal geomagnetic shielding.

Figs. 1 and 2 show that with or without geomagnetic storms,is GV in battlespaces over much of Canada and

Alaska. A cutoff rigidity of 0.2 GV filters out the same protonsand heavy-ions that would otherwise be stopped in a nominal100-mil Al satellite shield at GEO. Accordingly, as far as singleevent effects behind a 100-mil Al shield are concerned, theeffective GCR and SPE fluxes in battlespaces just north of the

Page 5: The cosmic ray environment of tactical ABMs

550 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 52, NO. 2, APRIL 2005

Fig. 7. SEU cross sections of Hitachi 1- and 4-Mb SRAMs.

U.S. are always nearly the same as at GEO, except for a 0.5 Xreduction due to shadowing by the Earth. In addition, duringlarge geomagnetic storms, the cutoff is reduced to 0.2 GV orless over New York City, Washington DC, London, and Berlin.Thus, during a large geomagnetic storm, the GCR flux overthose cities is also X the GCR flux at GEO; and if thestorm coincides with an SPE, the SPE flux over these cities is

X the SPE flux at GEO.Consider a TABM in a battlespace where is GV. We

assume the TABM has a central memory composed of 50 1-Mbor 50 4-Mb static random access memories (SRAMs) with theSEU cross sections in Fig. 7, [17]. The cross sections are typicalfor fast, COTS SRAMs. The limiting cross section [cm ]of the 1-Mb SRAM is cm ; and

[MeV cm /mg], the LET where the cross section is, is MeV cm /mg. For the 4-Mb SRAM, is

cm and is MeV cm /mg.Few of the SEUs in the 1-Mb SRAM were multi-bit; but manyof the SEUs in the 4-Mb SRAM apparently were [17].

According to the FOM method [15], the SEU rate [er-rors/day] of an IC behind 100 mils Al shielding in the GEOGCR environment, is given approximately by

. For the 1-Mb SRAMerrors/day ( errors/s). For the 4-Mb SRAM

errors/day ( er-rors/s). Assuming 100 mils Al shielding and applying the 0.5X shadowing factor, it follows that in the TABM battlespacethe SEU rate of these SRAMs due to GCRs is . If thecentral memory is composed of the 1-Mb SRAMs, its SEU ratedue to GCRs is errors/s. Fora 500-s battletime the probability the central memory is upsetis %. If the 4-Mb SRAMs areused, the expected number of upsets in the central memory is

per 500 s. All this neglects theupsets in the TABMs clock buffers, processors, analog-to-dig-ital converters, sensors, and navigation equipment.

SPEs greatly increase SEU rates. Comparing Figs. 3 and4 one sees that in a battlespace where is GV, avery large SPE raises the energetic particle flux at the 2- to20-MeV cm /mg SEU threshold of many COTS ICs, byX; and it raises the particle fluxes at lower LETs by up to three

Fig. 8. Number of events per solar active year with peak flux greater than orequal to that shown on the abscissa.

orders of magnitude more. Therefore, in a very large SPE, theSEU rate of an IC with 100 mils Al shielding in a battlespacewhere is GV, is the GCR SEU rate, , plus arate due to heavy ions, plus a rate that weconsider next, due to the vastly increased low LET fluxes.

Peterson has extended the FOM method to protons [15], andusing his results one can show that the additional rate due tothe SPE protons is also for SEU-sensitive ICslike the Hitachi SRAMs. Both on-orbit measurements [16], [18]and careful computations [16] confirm that SPE proton and SPEheavy ion SEU rates can be comparable in sensitive ICs. Thus, arough estimate of the upset rate of a sensitive IC under 100 milsAl shielding, during a very large SPE in a battlespace whereis GV, is .There may also be a significant contribution from SPE alphas;but it is difficult to quantify without precise SEU cross sectionmeasurements at low LET values.

If the 1-Mb SRAMs are used in the central memory, a verylarge SPE will induceupsets per 500 s in the memory. If the 4-Mb ICs are used, thenumber of upsets increases to per 500 s.

V. FPA AND MICROCHANNEL PLATE (MCP) UPSETS

GCRs, SPEs, trapped protons, and trapped and ionosphericelectrons, will all induce impulsive noise (“glitches”) in aTABM’s star trackers, MCPs, FPA pixels, fiber-optic gyros,optocouplers, and detector readout electronics. The glitchesproduced by protons and alphas in the SOHO long-waveinfrared FPA, saturate, on average, 8 pixels; and the glitchesproduced by heavier ions saturate, on average, 50 pixels andcan degrade responsivity for many seconds [19], [20]. A singlecosmic ray can saturate pixels on a charge-coupleddevice (CCD) detector [21]. Secondary particles produced byinteractions of primary particles with the material surroundinga pixel, can increase the particle flux striking a detector by

%. In some science payloads nearly every glitch must besuppressed (with software filtering).

Consider a TABM FPA designed to track targets that generateelectrons/pixel/sample. Assuming silicon (Si) pixels,

Page 6: The cosmic ray environment of tactical ABMs

SOLIN et al.: THE COSMIC RAY ENVIRONMENT OF TACTICAL ABMs 551

Fig. 9. Worldwide trapped electron fluxes in TABM battlespaces.

eV is expended per electron-hole-pair created; soelectrons corresponds to a deposited energy of 0.18 MeV. Typ-ical pixels have dimensions m. The range of a 1-MeVproton in Si is 16.3 m. 1 MeV/nucleon, recall, is the cutoffused in Figs. 3–5. Without glitch suppression software, all theions in the fluxes depicted in Figs. 3–5, can corrupt at least onepixel. As Fig. 4 shows, shielding can greatly reduce the SPEflux. However, even with 400 mils ( mm) Al shielding,the peak integral SPE flux for GV, ism sr s ; and so the glitch rate for a 1-cm FPA will be

( sr) ( m sr s )(1 cm ) s , witha potentially large number of pixels corrupted in each glitch.

VI. SPE FLUXES AND PROBABILITIES

Fig. 8 is an estimate of the rate at which SPEs occur witha given peak flux of greater than 10 MeV protons [22]. Thelargest plotted flux is the August 1972 SPE. The event used incalculating Figs. 4 and 5 is the second-largest event in Fig. 8.An event of that size occurs once per decade; but observe fromthe figure that events that are % as large as that one, occurabout every two years during solar maximum.

VII. TRAPPED ELECTRON GLITCH RATES

TABM battlespaces may be regions of significant trappedelectron content. Fig. 9 depicts the AE8 static model of theworldwide trapped electron environment at 450 km [12].As noted, during large geomagnetic storms, trapped or qua-sitrapped high-energy particles can appear in regions wherethey are not normally found. A 1.25 MeV cutoff was used

because that is roughly the minimum electron energy requiredto penetrate 100 mils Al and then produce a glitch in an FPApixel. The maximum flux is cm s , which isalmost the same as the particle flux of a very large SPE; but400 mils Al shielding will reduce the flux to a value much lessthan the cm s flux of a very large SPE behind400 mils Al. Bremsstrahlung from the electrons will also induceglitches.

VIII. IONOSPHERIC SCINTILLATION

There are other natural radiation effects that can impactTABMs. Irregularities in the electron density in the iono-sphere can cause scintillation (fading), polarization rotation,angle-of-arrival fluctuations, absorption, scattering, and co-herence bandwidth reductions, in communication and radarlinks. The irregularities develop in equatorial and high latituderegions. Geomagnetic storms tend to drive the effects to worstcase. Solar radio bursts can also interfere with communicationand radar links. Forecasts are available on the web of worldwidetotal electron content [23] and scintillation levels [24].

IX. SUMMARY AND CONCLUSION

The worldwide variability of the GCR and SPE environ-ments in TABM battlespaces has been described. The TABMbattlespace includes high latitudes where the environment isnearly the same as that of the NMD satellites.

It has been shown that GCR- and SPE-induced SEEs are aconcern for TABMs, and thus appropriate hardening techniquesshould be applied.

Page 7: The cosmic ray environment of tactical ABMs

552 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 52, NO. 2, APRIL 2005

REFERENCES

[1] J. Sheehy and B. Pugh, “Electronic subsystem hardening,” in HEART1997 Short Course, 2001. (Unclassified-ITAR).

[2] Launch Commit Criteria and Recent Solar Activity Trends, Sept. 4, 2002.Lockheed Martin Lessons Learned Notice 02-20.

[3] M. C. Calvet, R. Mazon, P. Simon, D. Galindo, C. Poivey, P. Garnier, J.Bourrieau, B. Cadot, and R. Ecoffet, “Acknowledgment of the naturalradiation environment upon the ARIANE 5 launcher,” in Proc. RADECS1995 Conf., Arcachon, France, pp. 170–174.

[4] C. S. Dyer, K. Hunter, S. Clucas, and A. Campbell, “Observation of thesolar particle events of October and November 2003 from CREDO andMPTB,” IEEE Trans. Nucl. Sci., vol. 51, no. 6, pp. 3388–3393, Dec.2004.

[5] C. H. Tsao, R. Silberberg, and J. R. Letaw, “Cosmic ray heavy ionsat and above 40,000 feet,” IEEE Trans. Nucl. Sci., vol. 31, no. 6, pp.1066–1068, Dec. 1984.

[6] , “Neutron-generated single event upsets in the atmosphere,” IEEETrans. Nucl. Sci., vol. 31, no. 6, pp. 1183–1185, Dec. 1984.

[7] E. Normand and T. J. Baker, “Altitude and latitude variations in avionicsSEU and atmospheric neutron flux,” IEEE Trans. Nucl. Sci., vol. 40, no.6, pp. 1484–1490, Dec. 1993.

[8] D. F. Smart and M. A. Shea, “The space-developed dynamic verticalcutoff rigidity model and its applicability to aircraft radiation dose,” Adv.Space Res., vol. 32, pp. 103–108, 2003.

[9] D. F. Smart, M. A. Shea, A. J. Tylka, and E. O. Flueckiger, “Calculatedvertical cutoff rigidities for the international space station using the tsy-ganenko model for every two hours in UT,” in Proc. 28th Int. CosmicRay Conf. 7, Tsukuba, Japan, 2003, pp. 4241–4244.

[10] D. F. Smart, M. A. Shea, M. J. Golightly, M. Weyland, and A. S. Johnson,“Evaluation of the dynamic cutoff rigidity model using dosimetry fromSTS-28 flight,” Adv. Space Res., vol. 31, pp. 841–846, 2003.

[11] A. J. Tylka, J. H. Adams Jr., P. R. Bomberg, B. Brownstein, W. F. Diet-rich, E. O. Flucckiger, E. L. Petersen, M. A. Shea, D. F. Smart, and E. C.Smith, “CREME96: A revision of the cosmic ray effects on micro-elec-tronics code,” IEEE Trans. Nucl. Sci., vol. 44, no. 6, pp. 2150–2160,Dec. 1997.

[12] Computed With Tools at the European Space Agency SPENVIS Website[Online]. Available: http://www.spenvis.oma.be/spenvis/

[13] J. B. Blake, M. S. Gussenhoven, E. G. Mullen, and R. W. Fillius, “Iden-tification of an unexpected space radiation hazard,” IEEE Trans. Nucl.Sci, vol. 39, no. 6, pp. 1761–1764, Dec. 1992.

[14] CREME96 Website [Online]. Available: https://creme96.nrl.navy.mil/[15] E. L. Petersen, “The SEU figure of merit and proton upset rate calcula-

tions,” IEEE Trans. Nucl. Sci., vol. 45, no. 6, pp. 2550–2562, Dec. 1998.[16] A. J. Tylka, W. F. Dietrich, P. R. Boberg, E. C. Smith, and J. H. Adams

Jr., “Single event upsets caused by solar energetic heavy ions,” IEEETrans. Nucl. Sci., vol. 43, no. 6, pp. 2758–2766, Dec. 1996.

[17] N. Nemoto, H. Shindou, S. Koboyama, H. Itoh, S. Matsuda, S. Okada,and I. Nashiyama, “Relationship between single-event upset immunityand fabrication processes of recent memories,” in Proc. RADECS 1999Conf., Fontevraud, France, pp. 193–197.

[18] C. S. Dyer, K. Hunter, S. Clucas, D. Rodgers, A. Campbell, and S.Buchner, “Observation of solar particle events from CREDO and MPTBduring the current solar maximum,” IEEE Trans. Nucl. Sci., vol. 49, no.6, pp. 2771–2775, Dec. 2002.

[19] A. Claret, H. Dzitko, J. Engelmann, and J. L. Starck, “Glitch effectsin ISOCAM detectors,” Experimental Astronomy, vol. 10, pp. 305–318,1998. Available: http://www.ipac.caltech.edu/iso/cam/cam_faq.html.

[20] ESA Observer’s Manual for ISOCAM (1996, Feb.). [Online]. Available:http://www.iso.vilspa.esa.es/manuals/iso_cam/cam_om_1.html

[21] T. S. Lomheim, R. M. Shima, J. R Angione, W. F. Woodward, D. J.Asman, R. A. Keller, and L. W. Schumann, “Imaging charge-coupleddevice (CCD) transient response to 17 and 50 MeV proton and heavy-ionirradiation,” IEEE Trans. Nucl. Sci., vol. 37, no. 6, pp. 1876–1885, Dec.1990.

[22] M. A. Xapsos, G. P. Summers, and E. A. Burke, “Probability model forpeak fluxes of solar proton events,” IEEE Trans. Nucl. Sci., vol. 45, no.6, pp. 2948–2953, Dec. 1998.

[23] Ionosphere Maps [Online]. Available: http://www.cx.unibe.ch/aiub/ionosphere.html

[24] Ionospheric Scintillation Predictions [Online]. Available: http://www.nwra-az.com/ionoscint/sp_main.html