1621

SESSION J:

RADIATION EFFECTS AND DEVICES

IEEE Transactions on Nuclear Science, Vol. NS-34, No. 6, December 1987

COMPARISON OF NEUTRON, PROTON AND GAMMA RAY EFFECTS IN SEMICONDUCTOR DEVICES*

J.P. RaymondMission Research Corporation

San Diego, CA

E.L. PetersenNaval Research Laboratory

Washington, DC

ABSTRACT

Interest in proton radiation effects hasintensified in recent years. A prime focus is therelationship between proton displacement and ioniza-tion effects and the separate consideration ofneutron-induced displacement and gamma-ionizationeffects in TREE characterization. Recent definitivework on proton and neutron displacement damage insilicon in terms of nonionizing energy loss has laidthe groundwork for comparison of proton effects withthe TREE data base. We initiate this comparison witha summary of device radiation susceptibilities inneutron and gamma environments. Proton interactionsin silicon devices are then presented in terms ofdose deposition and nonionizing energy loss. Thisleads to a neutron-proton damage equivalence factorand enables the development of simple correspondence.The device susceptibility charts are then combined soboth displacement damage and ionization-damage can beschematically examined relative to proton dose.These susceptibility charts demonstrate the dominanceof ionization effects for damage in a proton environ-ment for modern silicon microcircuit technologies.This approach is presented as a convenient means ofinterpreting effects for both proton exposures andTREE simulators. It is concluded that TREE characte-rization can be used as a good first-order estimateof proton damage effects.

1. INTRODUCTION

In the assessment of proton-induced damageto silicon microcircuits, it would be very useful todraw from the knowledge gained from the study oftransient radiation effects on electronics (TREE)over the last 25 years. Extensive facilities havebeen built to enable the assessment of performancedegradation effects in semiconductor devices result-ing from neutron-induced atomic displacements andgamma-ray-induced ionization. The existing TREE database would be applicable to proton effects assess-ment, given correlation of displacement damage andionization effects. In addition, given correlation,the number of facilities available for proton damageassessment can be expanded to include the TREE simu-lation facilities as well as available cyclotrons.

Correlation of proton damage effects toboth neutron displacement damage and ionizing radia-tion exposure has been established by a number ofbasic material and device studies [1-4]. In thispaper the correlation will be interpreted in terms ofdevice TREE susceptibility. This interpretationclearly indicates that long-term ionization effectsare the primary damage mechanism for proton exposureof virtually all types of modern silicon microcir-cuits.

2. DEVICE SUSCEPTILBIITY ASSESSMENT

Characterization of displacement and long-term ionization damage effects on semiconductor piece-parts is routine in the support of the development ofsystems which must be hardened to nuclear weapon andspace radiation environments. Typically, it isassumed that the displacement damage and ionizationeffects are independent, so that damage characteriza-tions can be done separately. The facilities used tocharacterize TREE are a pulsed nuclear reactor for theneutron source, and a Cobalt-60 source or electronaccelerator (e.g., Dynamitrons, linear accelerators,Van de Graaff generators) for the ionization source.

2.1 Neutron Displacement Damage

The exposure environment of a pulsed nuclearreactor includes high-energy neutrons and concomitantgamma rays. A comprehensive characterization of semi-conductor device effects includes the time dependentnature of neutron displacement damage and ionizationas potential failure mechanisms. The device(s) undertest would be actively biased during exposure, thetest conditions would be carefully selected, and theresponse(s) would be monitored during and after expo-sure. Fortunately, however, if only characterizationof stable neutron displacement damage effects is ofinterest there is virtually no dependence of theresult on the test conditions during exposure, anddevices are routinely exposed with no electrical con-nection (usually with all external leads shorted inconductive foam).

In terms of the radiation effects data base,neutron damage susceptabil ity is most frequentlyreported in terms of neutron fluence at the failurelevel observed after a series of reactor exposures.Results of these characterizations can be summarizedas a range of device susceptibility for major micro-circuit technologies as shown in Figure 1. Neutronfailure levels for the MOS technologies are estimatedbased only on displacement damage effects. The firstorder effect in n-MOS technologies is the degradationof minority carrier lifetime, which increases criticalleakage currents in high-performance arrays. Theestimate of critical displacement damage effects inbulk CMOS is based on reported limitations in the useof neutron exposure to reduce lifetime and increaselatchup hardness [5,6]. The estimate of critical dis-placement damage effects on CMOS/SOS microcircuits isbased on relatively early evaluations. In recentyears it has been assumed that the hardness is domi-nated by ionization-induced damage [7].

Neutron failure levels for bipolar microcir-cuit technology reflect the sensitivity of circuitperformance to the dominant degradation in transistorcommon-emitter gain, as well as the evolution to

*Work was partially sponsored by the LTH-4.2 Program through Naval Research Laboratory Contract N00014-85-C-2642.

0018-9499/87/1200-1622$01.00 C 1987 IEEE

1622

1623

transistors with higher gain-bandwidth products andthus lower neutron damage susceptibility [8]. Thefailure levels for the TTL microcircuits reflectrelatively old gold-doped devices in which minoritycarrier lifetime degradation is not critical. Theevolution to non-gold-doped Low-Power TTL results ingreater sensitivity to minority carrier lifetimedegradation and an increasing dependence on relative-ly high transistor gains with advances in processingcontrol. The high neutron failure levels of bothJunction-Isolated ECL and all oxide-separated bipolardigital circuit technologies reflects the applicationof very high gain-bandwidth transistor elements. Theneutron susceptibility of ECL is slightly reducedbecause of the trend to depend on relatively hightransistor gains. The neutron susceptibility of ana-log bipolar is shown as a very broad range becausethe failure level reflects both a substantial varia-tion in process technology and depends critically onthe demands of the performance requirements. Thefailure level for a very precise, high performanceanalog device will be much less than that of theanalog device in an applicaiton that allows substan-tial degradation of the performance parameters.

The estimated ranges of microcircuit neu-tron damage susceptibility are shown in the tradi-tional bar chart form in Figure 1 (Refs. 8, 9). Thespecific ranges of the estimated susceptability maybe, and typically are, the source of significantdebate. However, our purpose is to focus attentionon the ranges of susceptibility rather than legislatespecific boundaries.

11 1210 10

TECHNOLOGY 72 47 2

n-MOSBulk Si-Gate CMOS

CommercialHardened

Si-Gate CMOS/SOSCommercialHardened

Junction-Isolated BipolarTTLLow-Power TTLECL

Oxide-SeparatedAnalog Bipolar

13 14 1510 10 10

4 7 2 4 7 2 4 7 2

1610

4 7

Bipolar

2 4 7 2 4 7 2 4 7 2 4 7 2 4 7

011 1012 1013 1014 1015 10Neutron Fluence, n/cm2, 1 MeV equivalent

Figure 1. Estimating ranges of microcircuit neutrondamage susceptibility.

2.2 Long-term Ionization Damage

The characterization of long-term ionizationdamage is typically performed by exposure to gammarays from a Cobalt-60 source or high energy electronsfrom an accelerator. The complexities of determiningthe system-dependent failure level of a given deviceare legion. They include strong dependence on theelectrical bias conditions during exposure, the radi-ation intensity during exposure, the gamma ray/elec-tron energy spectrum at the device, electrical biasafter exposure, and measurement time after exposure.Currently, the techniques and facilities for exposingdevices and measuring a failure level in the labora-tory environments are well known. Unfortunately, itis often difficult to apply the laboratory simulationfailure data to determine the device operationalfailure level. In this paper we will consider theTREE data base as it exists in terms of reportedfailure levels for laboratory exposures.

2.2.1 Cobalt-60 Exposure

Exposure of a semiconductor device to thegamma rays from a Cobalt-60 source is a popular andrelatively inexpensive method of determining ionizingradiation damage susceptibility. Typically, thedevices under test are electrically biased and areeither monitored during exposure (i.e., in situ) orcharacterized by removal from the cell at definedintervals. The reported failure level for the deviceis generally reported in rads(Si), which is inter-preted as essentially equal to the energy depositionby ionization in the silicon-dioxide gate insulatorand surface passivation regions. Gamma ray exposuredoes produce displacment damage in the bulk semicon-ductor through the resultant Compton electrons. Forall practical purposes, in modern semiconductordevices the gamma ray-induced displacment damageeffects are negligible compared to the long-term-ionization effects.

2.2.2 High-Energy Electron Exposure

High-energy electron exposure is a verypopular approach, also used for simulation of theelectron exposure of the natural space environment.The facilities used for simulation are principallyDynamitrons. A major source of radiation effectsdata is the JPL Voyager/Galileo Data Bank thatincludes results of both electron and Cobalt-60 gammaexposure [10,11]. Failure levels resulting from theelectron exposure are reported either in terms ofrads(Si) or electrons/cm2. In the JPL data a conver-sion factor of 4 x 107 e/cm2 per rad(Si) is used forthe 2 MeV electron exposure of the Dynamitron faci-l ity. It should be noted that high-energy photonelectron exposure can cause displacement damage thatmay be important for very sensitive devices such assome bipolar analog microcircuits.

2.2.3 Reported Ranges of Device Susceptibility

The failure levels observed in Cobalt-60and high-energy electron environments can be present-ed as ranges of susceptibility by technology as shownin Figure 2 [8, 10-14]. In addition to all thecaveats noted in the discussion of neutron failurelevel summaries, the long-term-ionization ranges mustinclude the significant variation of failure with

6 bias/time conditions and the variation between high-performance commercial technologies and radiation-hardened technol ogies. The lower limit on long-termionization hardness is for high-performance n-MOSmicrocircuits and is on the the order of 1 krad(Si).The hardness of commercial CMOS is significantlygreater than that of n-MOS because of the increasedtolerance of CMOS circuits to radiation-inducedthreshold voltage shifts. The distinction betweenhigh-performance commercial and hardened CMOS is invariations of fabrication processes, process control,and circuit design which, in total, can increase thehardness by two orders of magnitude or greater. Thehardness of commercial CMOS/SOS is somewhat less thanthat of bulk CMOS because of charge trapping effectsat the additional oxide-semiconductor interfaces.Again, hardened CMOS/SOS is distinguished by processand circuit design techniques for hardness and hard-ness assurance even perhaps at some performance pen-alty.

For many years it had been reasonable toassume that the long-term-ionization hardness ofdigital bipolar technologies was very high and there-fore of little concern. That position is reasonablefor relatively highly-doped junction-isolated digital

1624

bipolar circuits where the failure mechanism is tran-sistor gain degradation. The advent of oxide-separa-tion techniques, which have practically displaced theolder junction-isolated technology, introduced cir-cuit failure due to surface inversion under theoxide-separation isolation regions. As a result, the"total-dose" hardness of the digital and analogoxide-separated bipolar technologies is substantiallyless than that of the earlier junction-isolateddevices. As with other technologies, the lower limitof the susceptibility range represents commercialtechnology optimized for yield and performance, andthe upper range includes circuits that have been fab-ricated with considerations of radiation susceptibi-ltiy.

2 3 4 5 6 7TECHNOLOGY 10 l O 1 O 10 10

24 ,I7 2 4 7 2 4 4 7 2 4 7

n-MOSBulk Si-Gate CMOS

CommercialHardened

Si-Gate CMOS/SOSCommerclalHardened

Junction-Isolated BipolarTTLLow-Power ITLECL

Oxide-Separated BipoicrAnalog Bipolar

10

<: 101

I- r

>

10Qc

\ 1 0r

1 nX l3

Q 1z nC

1'11L102

I

I

I

^ 4 7 2 4 7 2 4 7 2 4 7 2 4 7-l l lL ~lllI lll

02 1T03 1041n05 o61r07Totcl Ionizing Dose, rads(Si)

Figure 2. Estimated ranges of microcircuit long-termionizing radiation damage.

As with neutron damage, the range of sus-ceptiblity for analog bipolar technologies is verywide because of the range of fabrication processesand performance requirements. Significant long-termionizing radiation susceptibility can be observed inany technology for very demanding performancerequirements. The JPL/Galileo Data Base includesdata on a wide variety of analog circuit types, aswell as the observed variations in the hardness of aspecific product with procurement history and withefforts in hardening. As with neutron damageeffects, the ranges of ionizing radiation damage sus-ceptibility can be debated at some length. The pur-pose of this work, however, is to suggest these asrepresentative of the existing data base in order toillustr ate the related effects of high-energy protonexposure.

2.3 Schematic Representation of MicrocircuitSusceptibility

In order to relate proton effects to theTREE data base on neutron and ionization damage itwill be convenient to combine the estimated ranges ofsusceptibility shown in Figures 1 and 2 into a two-dimensional schematic form as shown in Figur e 3. InFigure 3, the ranges of displacement damage suscepti-bility are along the ordinate, and the ranges ofioniziation damage susceptibility are along theabscissa. The one-dimensional susceptibility rangesof the bar char ts now become two-dimensional r egionsof susceptibility. It must be emphasized that thisplot does not imply that the neutron fluence is thedependent variable and that ionization is the inde-pendent variable, but r ather is a schematic conven-ience. In Sections 3 and 4 we will develop the con-cept as a convenience in examining displacement dam-age and ionization in both proton and TREE simulationenvi ronments .

103 1 04 1 05 106 107Total Ionizing Dose, rads(Si)

Figure 3. Schematic representation of microcircuitsusceptibility.

3. HIGH ENERGY PROTON EFFECTS

High energy proton exposure differs fromthe traditional TREE exposur es in that the energydeposition goes into both displacement and ionizationprocesses. Previous studies have carefully consider-ed the potential equivalence of the resultingeffects, both analytically and exper imentally [1-41.

3.1 Proton Ionization

Pr oton energy loss due to ionization domi-nates most proton effects in materials. Incidentprotons lose energy through inelastic collisions withbound electrons in the atoms and molecules of thestopping material and results in their ionization andexcitation. Above 1 MeV the proton energy loss byionization and excitation can be calculated using theBethe equation [15]. The specific energy loss (dE/dx)is ordinarily expressed in units of MeV/g/cm2. Ener-gy less can be directly converted into ionizing dosedelivered. The dose is the amount of energy depo-sited per unit mass, apd is the product of dE/dx timethe fluence, D, (p/cm ). The ionization dose can becalculated from,

D = 1.6 x 1ff8 rads - 9 . dEMeV dx

The ionization dose per unit fluence isplotted as a function of energy in Figur e 4 (rightscale), based on the values of dE/dx calculated byJanni [151.

3.2 Proton Nonionizing Energy Loss

The proton nonionizing energy loss includesfour effects [1]. These are the elastic coulomb scat-tering of protons by the field of the nuclei, nuclearelastic scattering, inelastic nuclear scatter ing andr eactions, and Lindhard energy par tition. The non-ionizing energy loss again has units of Mev/g/cm2.Burke [1] calculated the ener gy dependence of protonnonionizing energy loss, usin gthe most recent data.He then calculated the ratio of proton nonionizingenergy loss to 1 MeV neutron nonionizing enrgy loss(2.04 x 10-3 MeV/g/cm2 obtained from the neutr on dam-age factor, Kn = 95 MeV-mb) [2]. There is

1625

excellent agreement of the calculated ratio of non-ionizing energy with the measured ratio of neutronto proton damage factor (Ko/Kn) [3,16]. The ratiosappear to be identical. Higure 4 shows the energydependence of the displacement damage factor (K /Kn)plotted against the left ordinate. It should benoted from this graph that both ionizing and non-ionizing energy losses have similiar energy depen-dence. They both decrease with increasing protonenergy, and the ratio remains nearly constant.

Figure 5 shows the neutron-proton damageequivalence factor (Rp) as a function of proton energy. Thie neutron fluence corresponding to a givendose delivered at a given proton energy is determinedby the factor Rp at that proton energy. Clearly, Rvaries little over a wide range of proton energietIf one chooses a reference R at 90 (R (90) = 1.54 x107), the geometric meansP of the eitremes, Rp isalways within a factor of two at any other energy.90 MeV is also a convenient reference as 107p/cm2delivers one rad(Si).

1 00-

0)

0

EC)

0)

E0)

0)

0-

a)

10D\

Displacement

0.110

loniza ion

1 00

100

to-o

0

tl-L

N

E

cn

n0

I C

0

N

c-

a

v)

0.1

1 000

Proton Energy, MeV

Figur e 4. Proton energy deposition.

3.3 Neutron/Proton Damage Equivalence Factors

The proton and neutron damage factorsr elate the displacement damage to the particle flu-ence, b and D . It would be convenient to know theneutron fluencA that is equivalent (as far as dis-placement damage is concer ned) to a given protondose. This can be obtained from the information inFigure 4. One curve shows the ratio of proton to neu-tron damage as a function of proton energy. Theother curve shows the dependence of ionizing dose onproton energy. The ratio of the two curves thenrelates neutron to pr oton displacement damage as afunction of dose (D), by a factor (R ) defined as,

KpAnR

P D/bpFor equivalent damage,

KnIn Kp4p

K( = P

n K Pn

K /Kn= p n

* D

D/Dp

on = Rp * D

u-(1a

E

EL

a

z

7

107

Neutron Fluence (n/ cmr 1 Mev equiv) = Rp * Dose (rads(Si))

106 l- l--I1 2 3 5 1 2 3 5 100 2 3 5 1000

Proton Energy, Ep,IeV

Figure 5. Relative proton displacement andionization.

3.4 Proton Effects Schematic Development

It is interesting to plot the neutron flu-ence as a function of proton ionizing dose. This isdone in Figure 6 for 90 MeV protons and for theextremes of 14 MeV (Rp = 8.2 x 106) and 800 MeV (Rp =2.9 x i07). The extremes determine a relatively nar-row band and the neutron-proton equivalence is withinthis band for all proton energies. Figure 6 corres-ponds in form to Figure 3 with neutron fluence on theordinate and dose on the abscissa. This suggeststhat failure data can be directly plotted on Figure 6for susceptibility in a proton environment. Onequestion that needs to be consider ed about totalionizing dose damage is the problem of the equiva-lence of damage by gamma rays (Co-60), pr otons, andother ionizing particles which arises because of thedifferent ionization densities. This has beenstudied by sever al authors [17-23]. The best singleset of experiments was that of Tallon and coworkers[22]. The basic conclusion of these groups was thatunder some conditions protons are less damaging(i.e., cause less transistor threshold shift) thangamma rays for proton energies less than 25 MeV. Thedifferences disappear for devices not under bias orfor protons incident at 45 degrees. The measurementsare greatly complicated by the normal post irradia-tion effects of dependence on dose rate and annealingtime. Ther e cur rently are no gamma-proton compari-sions made with separation of interface states andhole trapping processes. There are still a number ofquestions about the details of dose equivalence.Theseinclude variations in post-irradiation-effects due toexposur e times. The exposur e times of interest rangefrom the high-intensity flash x-ray pulse to the low-intensities of Cobalt-60 exposure 'iith pulsed neu-tron/proton exposures between the limits.

I \~~~~~~~1Il

1626

101> MeV

101

i2 o03 10 105 106 io7 108

Totcl lonizing Dose, rads(Si)

Figure 6. Representation of proton effects.

At the present it appears that a valid con-

servative approach i s to assume that gamma total

ionizing dose damage and proton total ionizing dose

damage are equivalent. The exceptions occur only at

low proton energies for some angles of incidence and

for some operating conditions. With this caveat, the

dose axis on Figure 6 can be assumed to apply for

either proton or gamma dose. In Figure 3, the ranges

of microcircuit susceptibil ity were shown as regions

for each technol ogy in terms of independent charac-

terizations of neutron displacement damage and ioni-

zing radiation damage. In Figure 6, proton energy

deposition is presented in terms of ionization and

the cor responding displ acement damage in terms of

1 MeY neutron fluence. Since Figures 3 and 6 both

have the same scales, they can be overlayed as shown

i n Figure 7. The result i s a schematic in which we

can compare proton effects to the TREE neutron/gamma

data base.

1

100

i 16

I 10

a)

a)a2 1 5

10

0 1 4,1. 0

a)

a) 1 3a 10

0

: 12Q, 1 0z

CMOS/SOS Oxide-Sep. Bipolar

,1 ''0 - Ep = 14 MeVp~~~~

I,7/tV E = 800 MeV1 °o o/11(02 103 104 1 05 106 1 07 1 08

Total Ionizing Dose, rads(Si)

Figure 7. Schematic representation with protoneffects.

4. INTERPRETATION OF DEVICE SUSCEPTIBILITY

4.1 Proton Suceptibility

The failuure level of any par ticular devicecan be shown as a point on Figure 7. For a protonexposure, the increasing fluence corresponds to acursor moving along the equivalance line with corres-ponding ver tical and horizontal lines defining accu-mulated dose and equivalent neutron fluence. When thecursor reaches a point such that either line exceedsthe associated radiation toler ance, the device fails.If the device fails at high gamma dose and low neu-tron fluence, the failure point will be in the regionof susceptibility which is to the lower r ight of theproton equivalence band, and failure in a pr-otonexposur e would be governed by displacement damage.If the device fails at low dose and high neutron flu-ence, the point falling to the upper left of theequivalence band, its failure in a proton exposurewould be due to total ionizing dose effects.

For example, consider the interpretationfor commer-cial CMOS technology as shown in Figure 7.In ter-ms of total ionizing radiation exposure, thefailuure r ange for commercial CMOS is approximatelyfrom 5,000 to 100,000 rads(Si), or at the lower rangeof proton ionization as shown in Figure 7. On theother hand, the displacement damage range for commer-cial CMOS is on the order of 2 x 1014 to 2 x 1015 n/cm2, which corresponds to a neutr on exposure at theupper end of the proton exposure range as shown inFigure 7. In this case, it is clear that the suscep-tibility of the technology in a high energy protonexposure will be dominated by ionization damage. Therole of ionization damage in commercial CMOS hardnessis certainly no surprise, but the same observationfor the other microcircuit technologies shown in Fig-ure 7 is interesting. In gener al, if the region ofmicrocircuit susceptibility as shown in Figur e 7 isabove the proton equivalent line, its hardness isdominated by ionization damage. Conversely, if theregion is below the proton equivalent line, itshardness is dominated by displacement damage.

The increased importance of ionization dam-age as the principal failure mechanism in protonexposures compared to TREE has been noted previously(Ref. 24), particularly for the MOS technologies. Inthe schematic form of Figure 7, however, in additionto the identification of the failure mechanism, onecan estimate the degree of dominance and thus thesensitivity to particular part selection within thetechnology family as well as the sensitivity to pro-ton energy. For example, for the junction-isolatedbipolar technologies, the regions of susceptibilityare near the proton equivalence line, and while ioni-zation damage dominates in many cases, the failurenechanism should be assessed for the specific parttype of interest and for the specific proton energiesof interest. It is interesting to note, however-, theclear dominance of ionization damage in oxide iso-lated bipolar technology. With the relatively highneutron hardness of the technology, it is clear thationization damage will be the primary failure mecha-nism for any specific device in the technology familyand for any proton energy.

4.2 Combined Neutr on/Gamma Susceptibility

It should be noted, however, that theapplication of the TREE data base to proton effectsstill1 contains some impl ici4t assumpteions andunanswer ed questions. In the determination of theneutron damage susceptibility it is generally assumedthat the TREE results are dominated by displacement

1627

damage. In Figure 8, the ionization associated withneutron exposure has been added to the schematic.The line for the "Typical Reactor Exposure" repre-sents the concomitant gamma ionization associatedwith the neutron exposure for a SPR-III reactor faci-lity [25]. Since the estimates of microcircuit sus-ceptibility regions are based on independent charac-terizations of displacement and ionization damage thespecific test conditions of the reactor exposure mustbe considered. The comparison to either the neutron/reactor or proton equivalence lines is valid when thedevices are exposed under active bias test conditionsrepresenting a reasonable worst case for ionizationeffects. With the devices unbiased it is typicallyassumed that the effects are dominated by displace-ment damage. The assumption can be checked by mea-surement of the relative effects in the neutron orproton environment against compar able measurements ina gamma or high-energy electron environment. Forexample, in comparing the reactor exposure line tothe microcircuit susceptibility regions, displacementdamage is the dominant failure mechanism for some ofthe bipolar technologies, ionization damage underactive bias is the dominant failure mechanism for theMOS and some of the bipolar technologies. It is alsointeresting to note that evolution in the developmentof CMOS hardened to ionization damage can result inan increase in displacement damage effects fbr high-energy proton exposures.

:3

CTj

a)

a)

.L

0

:3

ReactorNeutron/Gamma Ep = 90 MeV

11 Exposure

10102 i03 104 105 1o6 107

Total Ionizing Dose, rads(Si); [Cobalt-60]

Figure 8. Scehmatic with proton, reactor effects.

5. SUMMARY

A schematic representation combining TREEdamage susceptibility and high-energy proton effectshas been developed. We have used estimates fr om theTREE data base and recently developed equivalences ofproton and neutron displacement damage utilizing theconcept of nonionizing energy loss. The resultingneutron-proton damage equivalence function is used todirectly relate proton-induced displacement damage to1 MeV equivalent neutron displacement damage. Theschematic representation presenting both ionizationand displacement damage is a useful guide when study-ing proton effects. For example, it is clear that,to a first order, the TREE data base can be useddirectly in evaluating proton induced damage suscep-

tibility. TREE data, however, will generally not beavailable for the evaluation of new or unique silicon

devices. In this case, TREE laboratory simulationfacilities can be used to obtain new test data thatwill give a first order estimate of susceptibility inproton environments. For most current silicon tech-nologies, only gamma or electron exposur es are need-ed, as the total ionization damage effects dominatethe displacement damage effects. In many cases, ade-quate proton damage estimates can be made withoutundertaking a specific proton effects char acteriza-tion.

There ar e a number of situations where directproton testing will still be needed. There are pos-sibilities of heating or synergistic effects at highproton fluxes. The high dose rates also present newproblems with the various classes of effects groupedas post-irradiation effects. The results presentedhere are for the traditional rates of dose applica-tion and do not necessarily apply under extremely lowor extremely high dose rates. Ther e are a large num-ber of uncertainties about relative gammaprotoneffects at low proton energies. These can be impor-tant in the space environment. The results andapproaches developed in this paper allow a quickappraisal of the value of the traditional TREE test-ing and the needs for new proton testing.

ACKNOWLEDGEMENT

The authors wish to express their gratitudefor the encouragement and support of Ed Burke andGeoff Summers.

REFERENCES

1. E.A. Burke, "Energy Dependence of Proton-InducedDisplacement Damage in Silicon, IEEE Trans. onNuclear Science, vol. NS-33, no. 6, pp. 1276-1281; December, 1986.

2. G.P. Summers, E.A. Wolicki, M.A. Xapsos, P. Mar-shall, C.J. Dale, M.A. Gehlhausen and R.D.Blice, "Energy Dependence of Proton DisplacementDamage Factors for Bipolar Transistors," IEEETrans. on Nuclear Science, vol. NS-33, no. 6,pp. 1282-1286; December, 1986.

3. G.P. Summers, E.A. Burke, C.J. Dale, E.A.Wolicki, P. Marshall and M. Gehlhausen, "Corr e-lation of Particle-Induced Displacement Damagein Silicon," presented at the 1987 IEEENuclear and Space Radiation Effects Conference,Snowmass, Colorado; July, 1987.

4. V.A.J. van Lint, G. Gigas and J. Barengoltz,"Correlation of Displacement Effects Produced byElectrons, Protons, and Neutrons in Silicon,"IEEE Trans. on Nuclear Science, vol. NS-22, no.6, pp. 2663-2668; December, 1975.

5 J.R. Adams and R.J. Sokel, "Neutron Irradiationfor Prevention of Latch-up in MOS IntegratedCi rcui ts," IEEE Trans. on Nuclear Science,vol. NS-26, no. 6, pp. 5069-5073; December,1979.

6. J.E. Schroeder, A. Ochoa, Jr., and P.V.Dressendorfer, "Latch-up El imination in BulkCMOS LSI Circuits," IEEE Trans. on NuclearScience, vol. NS-27, no. 6, pp. 1735-1738;December, 1980.

7. G.J. Brucker, "Transient and Steady-StateResponse of CMOS/SOS," IEEE Trans. on Nuclearfor Science, vol. NS-27, no. 6, pp. 1674-1679;December, 1980.

1 08

1628

8. D.M. Long, "Hardness of MOS and Bipolar Inte-grated Ci rcuits," IEEE Trans. on Nucl earScience, vol. NS-27, no. 6, pp. 1674-1679;December, 1980.

9. J.P. Raymond, "MSI/LSI Radiation Response,Characterization and Testing," IEEE Tr ans. onNuclear Science, vol. NS-21, no. 6, pp. 308-314;December, 1974.

10. W.E. Prtice, K.E. Martin, D.K. Nichols, M.K,Gauthier, S.F. Brown, "Total-Dose RadiationEffects Data for Semiconductor Devices," JPLPublication 81-66; December 1, 1981.

11. K.E. Martin, M.K. Gauthier, J.R. Coss, A.R.V.Dantas, W.E. Prtice, "Total-Dose RadiationEffects Data for Semiconductor Devices, 1985Supplement," JPL Publication 85-43, October,15, 1985.

12. D.K. Myers, "RadiationSemiconductor 4 Kilobiton Nuclear Science, vol.1737; December, 1976.

Effects on CommericalMemories," IEEE Trans.NS-23, no. 6, pp. 1732-

13. D.G. Cleveland, "Dose Rate Effects in MOS Mirco-circuits," IEEE Trans. on Nuclear Science,vol. NS-31, no. 6, pp. 1348-1353; December,1984.

14. R.L. Pease, R.M. Tur fler, D. Platteter, D. Emilyand R. Blice, "Total Dose Effects in RecessedOxide Digital Bipolar- Micr ocir cuits," IEEETr ans. on Nuclear Science, vol. NS-30, no. 6,pp. 4216-4223; December, 1983.

15. J.F. Janni, "Proton Range-Energy Tables,"Atomic Data and Nuclear Data Tables, vol. 27,p. 341; 1982.

16. R.L. Pease, E.W. Enlow and G.L. Dinger, "Compar-ison of Proton and Neutron Carr ier RemovalRates," presented at the 1987 IEEENuclear and Space Radiation Effects Conference,Snowmass, Color-ado; July, 1987

17. T.R. Oldham and J.M. McGarrity, "Ionization ofSilicon-Dioxide by Heavy Charged Particles,"IEEE Trans. on Nuclear Science, vol. NS-28, no.6, ppp. 3968-3974; December, 1981.

18. G.J. Brucker, E.G. Stassinopoulos, 0. vanGunten, L.S. August and T.M. Jordan, "The DamageEquivalence of Electrons, Protons, and GammaRays in MOS Devices," IEEE Tr ans. on NuclearScience, vol. NS-29, no. 6, pp. 1966-1969;December, 1982.

19. G.J. Brucker, 0. Van Gunten, E.G.Stassinopoulos, P. Shapiro, L.S. August, T.M.Jordan, "Recovery of Damage in Rad-Hard MOSDevices During and After Irradiation by Elec-trons, Protons, Alphas, and Gamma Rays, " IEEETrans. on Nuclear Science, vol. NS-30, no. 6,pp. 4162-4168; December, 1983.

20. E.G. Stassinopoulos, G.J. Brucker, 0. van

Gunten, "Total-Dose and Dose-Rate Dependence ofProton Damage in MOS Devices Dur ing and AfterI rradi ati on," IEEE Tr ans. on Nuclear Science,vol. NS-31, no. 6, pp. 1444-1447; December,1984.

21. T.R. Oldham, "Analysis of Damage in MOS Devicesfor Several Radiation Environments," IEEETrans. on Nuclear Science, vol . NS-31, no. 6,pp. 1236-1241; December, 1984.

22. R.W. Tallon, M.R. Ackermann, W.T. Kemp, M.H.Owen and D.P. Saunders, "A Compar ison of Ioni-zing Radiation Damage in MOSFETs from Cobalt-60Gamma Rays, 0.5 to 22 MeV Protons and 1 to 7 MeVElectrons," IEEE Trans. on Nuclear Science,vol. NS-32, no. 6, pp. 4393-4398; December,1985.

23. W.J. Stapor, L.S. August and D.H. Wilson, "Pr o-ton and Heavy-Ion Radiation Damage Studies inMOS Transi stors," IEEE Trans. on NuclearScience, vol. NS-32, no. 6, pp. 4399-4404;December, 1985.

24 D.R. Locker, et al., "The Present State of Hard-ness of Electr onic Devices and Sensor Types toHigh Energy Protons," Air Force Weapons Labora-tory Technical Report AFWL-TR-81-81; January,1982.

25. L.M. Choate and T.R. Schmidt, "New Neutron Simu-lation Capabilities Provided by the Sandia PulseReactor-III (SPR-III) and the Upgraded AnnularCore Pulse Reactor (APCR)," IEEE Trans. onNuclear Science, vol. NS-25, no. 6, pp. 1625-1628; December, 1978.