creme96 update/replacement efforts j.h. adams abstract: the … · 2013-04-10 · creme96...

CREME96 Update/Replacement Efforts

J.H. Adams

Abstract:

This talk concerns the plans to update the CREME96 model that is currently available on

the WWW. The talk states the reasons for updating CRI_ME. It describes the updates

that are planned, including the single event prediction paradigm, the method of radiation

transport through the spacecraft to the electronic component of interest and the planned

updates to models for the space radiation environment. It also reviews user suggestions

received do date for the update.

https://ntrs.nasa.gov/search.jsp?R=20070032033 2019-05-07T22:19:58+00:00Z

CREME96 Update/Replacement Efforts

presented by Jim Adams

Single Event Effects Symposium, Long Beach, CA, 11 April 2007 1

Outline

• Why Update CREME?o The CREME paradigm and why it does not work

o Examples

• The CREME Replacemento Technical Approach

~ SEE Rate Predictions using RADSAFE

- RADSAFE Implementation Concept

- Structure of MRED - Core of RHESE Tool

- Modeling a Complex Stack - Nuclear Events

o Using MRED to Predict SEE Rates

• Summary


"Why Update CREME?

+ ·The currently available models are out of dqteo Single Event Effect (SEE) rate predictions are unreliable for

modern devices because:- Components contain features made of heavy metals- The combination of feature sizes are < 1 J.lm, low critical

charge and the high packing density of microcircuits.- So we often do not understand why SEE's occur.

o Improved environmental models are available- More accurate galactic cosmic ray models- Better solar energetic particle models

• Models of recent very large events can be includedo Monte Carlo radiation transport codes are available

- They include transport of heavy ions "- As well as neutron production and transport...- And they correctly treat complex spacecraft designs


"The CREME paradigm

Space Environment Grou,nd TestingIPM-MIN.LET

100 101 102 103 104

LET (MeV-cm'/g)

Direct Ionization: Rectangular Parallelepiped (RPP) modelNuclear Reactions (proton only): Bendel Model

(both are circa 1980)https://creme96.nrl.navy.mil/


Technical Justification (1)Classical On-Orbit SEE Performance Predictions

Ground TestingI

. Space EnvironmentIPM....MIN.LET

On-Orbit SEE Rate?


Xe

R.A. Reed, et. alIEEE Trans. Nuc.Sci., vol. 48, no.6, Dec. 2001, pp.2202 - 2209.

RA Reed, et. ai,IEEE Trans. Nuc.Sci., vol. 49, no. 6,Dec. 2002, pp.3038-304fl

Heavy IoneffectsCMOSSRAM

100

120

Protoneffects in

.-1--801 basedmemones

40

10080

30

lETeff (MeV-cm2/mg)

60

2J 40 ffi 80Argo r:J In::il:lem:l (00grees)

20

40

o

Existing Model (\Would P1jedict

Solid Line IA31I~ __~

20

Proton Angle oflnc:idenc:e (Degrees)

Silicon On InsulatorL..-__--.I

c<>

~en;i!10'"I-----+----~d.--l

~8

~

10'" r;========::;---------,'"!

10-5 ,...------------,~---.....

Protons10.6

Effects inOpticalLinks

K.M. Warren, et.allEEE Trans.Nuc. Sci., vol.48, no. 6, Dec.2005, pp. 21252131.

Heavy IonEffects in

~---, SiGe HBTs

•o

~ 1l~~:~- OJ /.'

N~ 10-8 ~ ~ Ar M1 \!:;{ ,/~ 1~~~:i \ \ \/J~ /' '''~o /. ' I'flo. •.1 .,' i /CIl 10-11 1'/: ~;i .!y·:-V)(/(l •.• '.!: \.

10-12-., ··j.f,'\" ,.10-13 --

o 10

Provided by Vendor A

Single Event Effects Symposium, Long Beach, CA, 11 April 2007

MSOfficel

Examples of Breakdown of Existing SEE Models

No ModelWidelyAvailable!

RA Reed, et. al IEEE Trans. Nuc. Sci., vol. 50,no. 6, Dec. 2003, pp. 2184 - 2190

1.E-07 -r:-------------_1.E-oS

Ne 1.E-09CJ-; 1.E-10o~ 1.E-11 ----------------,--------,--------,-,-"",,-,rn::: 1.E-12~ -=o ---i

/:, ~ -,.....,.,.-"""""-..."j,.".,.",."."-"",,,,,,-------e I; -o 1.E-13 t-I;~••~.~,;;:;;:;;;;;;:;;;:;:::;;-rRADHA.RDCMOS SRAM

1.E-14 .------11.E-15 -h-.-....-...~I:;:::;::=t=;:::;::f Nuclear Reactions

o 20 40 from Heavy IonslET (MeV"

(Z>l)

MSOfficel This is a measurement of the SEU cross section for Radhard part developed for DOD. It was supposed to have threshold at an LETof 40, but obviously it does not. Detailed TCAD simulations of the part could not identify any lower threshold junctions in thecircuit. A RADSAFE analysis showed that the subthreshold SEEs were due to target fragments from heavy ion induced reactions bythe 523 MeV Ne beam.,8/5/2006

Al This figure shows how the data can never be explained by the RPP model in CREME. The purple curve is an attempt to use theRPP moel to explain the SEU cross section versus LET. The data in black, green, pink and orange are from Kr (15 MeV), Ar (15MeV), Ar (20 MeV) and Ne (40 MeVjnuc) beams at 0, 30, 45, 60, and 70 degrees). Ne and Ar beam measurements are notconsistent with the RPP model. This is caused by the sensitive volume shape and heavy ion effects.AdamsJH, 8/6/2006

A2 This shows the angular dependence of the proton-induced SEU cross section on the angle of incidence of the proton in 0.5 micron50S. It is thought that the SEUs in this device are due to nuclear recoils and these are most likely at 0 degrees to the protonbeam. When these recoils are nearly in the plane of the silicon.

. AdamsJH, 8/6/2006

A3 These are proton-induced single event transcient cross section measurements, probably on an optocoupler. Bendel wil predict thepurple line, but if you include direct ionization using the RPP in CREME, you can get the peak at 90 degrees from the directionization of the stopping protons.AdamsJH, 8/6/2006

"The CREME Replacement

+The CREME96 Replacement will have:o Backward compatibility:

- Provide access to the CREME96 methods

- Include the classical SPE models from CREME86

o Provide access to the environment data directly

o Provide total dose estimates

o Provide several new approaches to SEE Predictions- Calorimetric approach

- Device model-dependent approach

o WWW accessible- Have the possibility to be locally hosted by the user


Technical Team

• Marshall Space Flight Center Teamo Scientists

- James H. Adams, Subproject Lead and Cosmic Ray Team Lead at MSFC

- Nasser Barghouty, Astrophysicist, Cosmic Ray Team at MSFC

- Zi-Wei Lin, Nuclear Physicist, Research Professor at Univ. of Alabama in Huntsville

- John Watts, Cosmic Ray Physicist, Cosmic Ray Team at MSFC

• Institute for Space and Defense Electronics (ISDE), Vanderbilt Universityo Principal Engineers

- Robert A. Weller, Prof. of Electrical Engineering, Prof. of Physics

- Robert A. Reed, Research Associate Professor of Electrical Engineering

- Ronald D. Schrimpf, Prof. of Electrical Engineering, Director of ISDE

- Marcus H. Mendenhall, Research Associate Professor of Physics

o Additional Engineering Staff Participants

- Lloyd W. Massengill, Prof. of Electrical Engineering, Director of Engineering, ISDE

- Brian Sierawski, ISDE Engineer

- Kevin Warren, ISDE Senior Engineer

- Dennis Ball, ISDE Engineer

o Graduate Students

- Students from the Radiation Effects and Reliability Group test and use ISDE code


Technical Objectives

+ Improve External Radiation Environment De~criptions

o Galactic cosmic rayso Solar energetic particleso Anomalous cosmic rays

+ Improve Radiation Transport though Spacecrafto Full Monte Carlo radiation transport

+ Provide a Physics-Based Simulation Tool of Single-EventEffects It

o Correctly treat hole-electron plasma creationo Charge collectiono Circuit response

+ Use the Tool to:o Diagnose SEE mechanisms in selected technologies

-' + Provide a WWW site to predict total dose and SEE effectson spacecraft


Technical ApproachThe External Environment

+ Quiet-time Galactic Cosmic Ray (GCR) Environmento We will evaluate the available GCR models against data and choose the

most accurate one.

+ Solar Energetic Particle Environmentso We will offer a variety of models based on real event (probably using

Xapsos spectral fits)

o We will include a probabilistic model based on the work of Joan Feynman

+ The Anomalous Cosmic Ray Environment in the Outer Heliosphere

Radiation Transport+ The external environment will be propagated though the spacecraft

structure to the electronic part of interest using a Monte Carlo transportcode (probably GEANT or FLUKA)o These codes include the capability to calculate the absorbed dose in the

part of interest (thus allowing the total dose to be predicted)


RADSAFE Implementation Concept

..Radiation EnvironmentModels

Approximate__-4.....1JIo<1 Response Models.

Virtual" Irradiator

PYO~SWG ..

REOC++

[Geant4]~ )

LVi~---...-

..Energy

Deposition&

Partitioning

1---.:1 Q(r, t) ~;''''I Poisson +---I:~ DesignSolver Decision

TCADGeometry


Description of RADSAFE

• RADSAFE will extend CREME96 modelso Leverages ongoing NASA/DoD funded development at Vande'rbilt University

• RHESE support will bring key portions of the RADSAFE concept to NASAengineers through the CREME-MC, a world wide web tool

• Examples of new on-orbit prediction capabilitieso Heavy ion nuclear reactionso Realistic geometries associated with modern technologieso Improved charge collection modelso Multiple Bit Upsets (i.e. FPGA based memory)o Proton induced effects in modern technologies

• Extensible to capture new, technology based models

DesignDecision

PoissonSolver

Q(r,t)

reADGeometry

EnergyDeposition

&Partitioning

Radiation Environments 1----+-1 Response Models.------l (CREME96) (CREME96)

RHESE RHESE


SEE Rate Predictions using RADSAFE

Statistical CircuitResponse

Predictable System Error Rates

~

Trackable System EffectsFunctional Errors

ComponentResponseStatistics

Physics of RadiationInteraction


Structure of MRED - Core of RHESE Tool

• MRED8: a Python/Geant4 application• Core: c++ for speed• Interface: SWIG (simplified

wrapper interface generator)• Python: THe language for

software system integration• Target machine is a Linux

cluster with 1k x86 and ppc nodes

Python

SWIG

MREDC++

( Geant4 )


Modeling a Complex Stack - Nuclear Events

~7 __~:-Y;)~ 2x2x2 IJm3

• "\\h " SensitiveVolume

I

Si021.0 IJm

Si-Nitride 0.4 m

High energy secondary particlescreated by nuclear reactions

with W in the Ie overlayer

4------ 50 IJm

2520

- Ne523 MeV- Ne 523 MeV - W plug

10 15

Energy (MeV)

Nuclear Reition Fragments

5

...............LET

I Direct I

• EX9.mple: MRED used to predict SE effects for twostructures:

o Stylized stack to approximate an Ie overlayer withand without W

• Direct and Indirect reactions computed• Energy deposition events included from primary and

secondary particles

(/) 106.....c::

105a.>>

W 104"t-o

103L-

a.>.0

102E:::J

101Za.>

10°>~('0

10-1a.>

0:: 10-2

J


25

/1.0

2010 15Energy (MeV)

• CREME96• Si only: MRED• Si withW layer: MRED

Solar MinimumZ = 1 to 92

Charge (pC)0.5

5

Product Application Example:Using MRED to Predict SEE Rates

......................................:t .

IMSOffice2! -\................. - - - - ~~.. ·k···················-···-·1O~12

10~13 L..-.-l...---'-...l--.I..--'----'----J'---'---'---'--"'il~___...---L...-.....L-'----'-___L......f.J.t___l_--'-----'----'---J

o


MSOffice2 This is a plot of energy deposition rate in a 10 by 10 by 1 micron volume within a block of silicon from the GCR envionment. Theblue case is pure silicon and the red case is with a tungsten layer above the sensitive volume.

1 8/ 6/2006

Summary

+ Single Event Effect (SEE) rate predictions with current models areunreliable for modern devices because: I

- Existing SEE prediction tools do not include effects found in m_oderndevices

- Accelerator tests cannot be intepreted by traditional methods toprovide accurate SEE predictions with existing tools.

+ We will use available physics-based device modeling to diagnose thecauses of SEEs in selected devices.

+ We will provide a WWW-based computational tool that:o supports real-time prediction of component reliability.o Includes improved environmental models that are availableo Uses available Monte Carlo radiation transport codeso Uses calorimetric methods informed by physics-based charge collection

and circuit response codes

+ This tool will provide:o Estimates of mission-specific total dose and SEE rateso To enable the design of safe reliable space electronics.


creme96 update/replacement efforts j.h. adams abstract: the … · 2013-04-10 · creme96...

Documents