1、Designation: F1192 11Standard Guide for theMeasurement of Single Event Phenomena (SEP) Induced byHeavy Ion Irradiation of Semiconductor Devices1This standard is issued under the fixed designation F1192; the number immediately following the designation indicates the year oforiginal adoption or, in th
2、e case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. Asuperscript epsilon () indicates an editorial change since the last revision or reapproval.This standard has been approved for use by agencies of the Department of Defense.1. Scope1.1 This
3、guide defines the requirements and procedures fortesting integrated circuits and other devices for the effects ofsingle event phenomena (SEP) induced by irradiation withheavy ions having an atomic number Z $ 2. This descriptionspecifically excludes the effects of neutrons, protons, and otherlighter
4、particles that may induce SEP via another mechanism.SEP includes any manifestation of upset induced by a singleion strike, including soft errors (one or more simultaneousreversible bit flips), hard errors (irreversible bit flips), latchup(persistent high conducting state), transients induced in com-
5、binatorial devices which may introduce a soft error in nearbycircuits, power field effect transistor (FET) burn-out and gaterupture. This test may be considered to be destructive becauseit often involves the removal of device lids prior to irradiation.Bit flips are usually associated with digital de
6、vices and latchupis usually confined to bulk complementary metal oxide semi-conductor, (CMOS) devices, but heavy ion induced SEP is alsoobserved in combinatorial logic programmable read onlymemory, (PROMs), and certain linear devices that may re-spond to a heavy ion induced charge transient. Power t
7、ransis-tors may be tested by the procedure called out in Method 1080of MIL STD 750.1.2 The procedures described here can be used to simulateand predict SEP arising from the natural space environment,including galactic cosmic rays, planetary trapped ions, andsolar flares. The techniques do not, howev
8、er, simulate heavyion beam effects proposed for military programs. The endproduct of the test is a plot of the SEP cross section (thenumber of upsets per unit fluence) as a function of ion LET(linear energy transfer or ionization deposited along the ionspath through the semiconductor). This data can
9、 be combinedwith the systems heavy ion environment to estimate a systemupset rate.1.3 Although protons can cause SEP, they are not includedin this guide. A separate guide addressing proton induced SEPis being considered.1.4 The values stated in SI units are to be regarded asstandard. No other units
10、of measurement are included in thisstandard.1.5 This standard does not purport to address all of thesafety concerns, if any, associated with its use. It is theresponsibility of the user of this standard to establish appro-priate safety and health practices and determine the applica-bility of regulat
11、ory limitations prior to use.2. Referenced Documents2.1 Military Standard:2750 Method 10803. Terminology3.1 Definitions of Terms Specific to This Standard:3.1.1 DUTdevice under test.3.1.2 fluencethe flux integrated over time, expressed asions/cm2.3.1.3 fluxthe number of ions/s passing through a one
12、cm2area perpendicular to the beam (ions/cm2-s).3.1.4 LETthe linear energy transfer, also known as thestopping power dE/dx, is the amount of energy deposited perunit length along the path of the incident ion, typicallynormalized by the target density and expressed as MeV-cm2/mg.3.1.4.1 DiscussionLET
13、values are obtained by dividingthe energy per unit track length by the density of the irradiatedmedium. Since the energy lost along the track generateselectron-hole pairs, one can also express LET as chargedeposited per unit path length (for example, picocoulombs/micron) if it is known how much ener
14、gy is required to generatean electron-hole pair in the irradiated material. (For silicon,3.62 eV is required per electron-hole pair.)1This guide is under the jurisdiction of ASTM Committee F01 on Electronicsand is the direct responsibility of Subcommittee F01.11 on Nuclear and SpaceRadiation Effects
15、.Current edition approved Oct. 1, 2011. Published October 2011. Originallyapproved in 1988. Last previous edition approved in 2006 as F119200(2006). DOI:10.1520/F1192-11.2Available from Standardization Documents Order Desk, Bldg. 4, Section D,700 Robbins Ave., Philadelphia, PA 191115094.1Copyright A
16、STM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.Acorrection, important for lower energy ions in particular, ismade to allow for the loss of ion energy after it has penetratedoverlayers above the device sensitive volume. Thus the ionsenergy, E, a
17、t the sensitive volume is related to its initial energy,EO, as:Es5 Eo2*ot/cosu!SdEx!dxDdxwhere t is the thickness of the overlayer and u is the angle ofthe incident beam with respect to the surface normal. Theappropriate LET would thus correspond to the modified energy,E.A very important concept, bu
18、t one which is by no meansuniversally true, is the effective LET. The effective LET appliesfor those soft error mechanisms where the device susceptibilitydepends, in reality, on the charge deposited within a sensitivevolume that is thin like a wafer. By equating the chargedeposited at normal inciden
19、ce to that deposited by an ion withincident angle u, we obtain:LETeffective!5LETnormal!/cosuu,60Because of this relationship, one can sometimes test with asingle ion at two different angles to correspond to two different(effective) LETs. Note that the effective LET at high anglesmay not be a realist
20、ic measure (see also 6.6). Note also that theabove relationship breaks down when the lateral dimensions ofthe sensitive volume are comparable to its depth, as is the casewith VLSI and other modern high density ICs.3.1.5 single event burnoutSEB (also known as SEBO)may occur as a result of a single io
21、n strike. Here a powertransistor sustains a high drain-source current condition, whichusually culminates in device destruction.3.1.6 single event effectsSEE is a term used earlier todescribe many of the effects now included in the term SEP.3.1.7 single event gate ruptureSEGR (also known asSEGD) may
22、occur as a result of a single ion strike. Here apower transistor sustains a high gate current as a result ofdamage of the gate oxide.3.1.8 single event functionality interruptSEFI may occuras a result of a single ion striking a special device node, usedfor an electrical functionality test.3.1.9 sing
23、le event hard faultoften called hard error, is apermanent, unalterable change of state that is typically associ-ated with permanent damage to one or more of the materialscomprising the affected device.3.1.10 single event latchupSEL is an abnormal low im-pedance, high-current density state induced in
24、 an integratedcircuit that embodies a parasitic pnpn structure operating as asilicon controlled rectifier.3.1.11 single event phenomenaSEP is the broad categoryof all semiconductor device responses to a single hit from anenergetic particle. This term would also include effects inducedby neutrons and
25、 protons, as well as the response of powertransistorscategories not included in this guide.3.1.12 single event transients, (SET)SETs are SE-causedelectrical transients that are propagated to the outputs ofcombinational logic ICs. Depending upon system applicationof these combinational logic ICs, SET
26、s can cause systemSEU.3.1.13 single event upset, (SEU)comprise soft upsets andhard faults.3.1.14 soft upsetthe change of state of a single latchedlogic state from one to zero, or vice versa. The upset is “soft”if the latch can be rewritten and behave normally thereafter.3.1.15 threshold LETfor a giv
27、en device, the thresholdLET is defined as the minimum LET that a particle must haveto cause a SEU at u = 0 for a specified fluence (for example,106ions/cm2). In some of the literature, the threshold LET isalso sometimes defined as that LET value where the crosssection is some fraction of the “limiti
28、ng” cross section, but thisdefinition is not endorsed herein.3.1.16 SEP cross sectionis a derived quantity equal to thenumber of SEP events per unit fluence.3.1.16.1 DiscussionFor those situations that meet thecriteria described for usage of an effective LET (see 3.1.4), theSEP cross section can be
29、extended to include beams impingingat an oblique angle as follows:s5number of upsetsfluence 3 cosuwhere u = angle of the beam with respect to the perpendicu-larity to the chip. The cross section may have units such ascm2/device or cm2/bit or m2/bit. In the limit of high LET(which depends on the part
30、icular device), the SEP cross sectionwill have an area equal to the sensitive area of the device (withthe boundaries extended to allow for possible diffusion ofcharge from an adjacent ion strike). If any ion causes multipleupsets per strike, the SEP cross section will be proportionallyhigher. If the
31、 thin region waferlike assumption for the shape ofthe sensitive volume does not apply, then the SEP cross sectiondata become a complicated function of incident ion angle. Asa general rule, high angle tests are to be avoided when a normalincident ion of the same LET is available.A limiting or asympto
32、tic cross section is sometimes mea-sured at high LET whenever all particles impinging on asensitive area of the device cause upset. One can establish thisvalue if two measurements, having a different high LET,exhibit the same cross sections.3.2 Abbreviations:3.2.1 ALSadvanced low power Schottky.3.2.
33、2 CMOScomplementary metal oxide semiconductordevice.3.2.3 FETfield effect transistor.3.2.4 ICintegrated circuit.3.2.5 NMOSn-type-channel metal oxide semiconductordevice.3.2.6 PMOSp-type-channel metal oxide semiconductordevice.3.2.7 PROMprogrammable read only memory.3.2.8 RAMrandom access memory.3.2.
34、9 VLSIvery large scale integrated circuit.4. Summary of Guide4.1 The SEP test consists of irradiation of a device with aprescribed heavy ion beam of known energy and flux in such away that the number of single event upsets or other phenomenacan be detected as a function of the beam fluence (particle
35、s/cm2). For the case where latchup is observed, a series ofF1192 112measurements is required in which the fluence is recorded atwhich latchup occurs, in order to obtain an average fluence.4.2 The beam LET, equivalent to the ions stopping power,dE/dx, (energy/distance), is a fundamental measurement v
36、ari-able. A full device characterization requires irradiation withbeams of several different LETs that in turn requires changingthe ion species, energy, or, in some cases, angle of incidencewith respect to the chip surface.4.3 The final useful end product is a plot of the upset rate orcross section
37、as a function of the beam LET or, equivalently, aplot of the average fluence to cause upset as a function of beamLET. These comments presume that LET, independent of Z,isa determinant of SE vulnerability. In cases where chargedensity (or charge density and total charge) per unit distancedetermine de
38、vice response to SEs, results provided solely interms of LET may be incomplete or inaccurate, or both.4.4 Test Conditions and RestrictionsBecause many fac-tors enter into the effects of radiation on the device, parties tothe test should establish and record the test conditions to ensuretest validity
39、 and to facilitate comparison with data obtained byother experimenters testing the same type of device. Importantfactors which must be considered are:4.4.1 Device AppraisalAreview of existing device data toestablish basic test procedures and limits (see 8.1),4.4.2 Radiation SourceThe type and charac
40、teristics of theheavy ion source to be used (see 7.1),4.4.3 Operating ConditionsThe description of the testingprocedure, electrical biases, input vectors, temperature range,current-limiting conditions, clocking rates, reset conditions,etc., must be established (see Sections 6, 7, and 8),4.4.4 Experi
41、mental Set-UpThe physical arrangement ofthe accelerator beam, dosimetry electronics, test device,vacuum chamber, cabling and any other mechanical or electri-cal elements of the test (see Section 7),4.4.5 Upset DetectionThe basis for establishing upsetmust be defined (for example, by comparison of th
42、e test deviceresponse with some reference states, or by comparison ofpost-irradiation bit patterns with the pre-irradiation pattern, andthe like (see 7.4). Tests of heavy ion induced transients requirespecial techniques whose extent depends on the objectives andresources of the experimenter,4.4.6 Do
43、simetryThe techniques to be used to measure ionbeam fluxes and fluence.4.4.7 Flux RangeThe range of heavy ion fluxes (bothaverage and instantaneous) must be established in order toprovide proper dosimetry and ensure the absence of collectiveeffects on device response. For heavy ion SEP tests a norma
44、lflux range will be 102to 105ions/cm2-s. However, higherfluxes are acceptable if it can be established that dosimetry andtester limits, coincident upset effects, device heating, and thelike, are properly accounted for. Such higher limits may beneeded for testing future smaller geometry parts.4.4.8 P
45、article Fluence LevelsThe minimum fluence isthat fluence required to establish that an observance of noupsets corresponds to an acceptable upper bound on the upsetcross section with a given confidence. Sufficient fluence shouldbe provided to also ensure that the measured number of upsetevents provid
46、es an upset cross section whose magnitude lieswithin acceptable error limits (see 8.2.7.2). In practice, afluence of 107ions/cm2will often meet these requirements.4.4.9 Accumulated Total DoseThe total accumulated doseshall be recorded for each device. However, it should be notedthat the average dose
47、 actually represents a few heavy iontracks, 30 m. The U.C.Berkeley 88-inch cyclotron and the Brookhaven NationalLaboratory Van de Graaff have adequate energy for most ions,but not all. Gold data at BNL is frequently too limited in rangeto give consistent results when compared to nearby ions of thepe
48、riodic table. Medium-energy sources, such as the K500cyclotron at TexasAhowever, in general, DUTs may be safely packed and trans-ported without delay after test.6.7 Ion Interaction Effects:6.7.1 The calculation of an effective LET (see discussion in3.1.4) hinges on the thin slab approximation of the
49、 sensitivevolume, which is less likely to hold for high density, smallgeometry devices. This problem can be examined by investi-gating the device SEP response to two different ions having thesame effective LET.6.7.2 The proportion of length to width of the sensitivevolume is also assumed equal to one. Rotating the device alongboth axes of symmetry during the test may provide a moremeaningful characterization.6.7.3 As geometries continue to scale down, the possibilityof multiple bit upsets increases. Hence, the nature of the ionsra
