estimating “damage 2 electronics” using fluka m. brugger for the fluka team cern fluka user...

Estimating Estimating “Damage 2 “Damage 2

Electronics” using Electronics” using FLUKAFLUKA

M. Brugger for the FLUKA Team

CERN FLUKA User MeetingJuly 31st 2008

MATERIAL CAUSE RADIATIONEFFECT

Semiconductors Electron-hole pair dose ionizationPhoton interaction photon

absorption Lattice displacement nucleon collision

Polymers Main and side chain rupture dose ionizationcross-linking degradation “ “gas evolution, radical productiondose rate

Ceramics Lattice displacements nucleon collisiontrapped charge carriers dose ionizationcolor centers “ “

Metals Lattice displacements nucleon collisionnuclear reactions producing clusters “

“voids and bubbles “

“

Radiation Effects – Rough Classification

© Lockheed MartinJuly 31st 2008 2CERN FLUKA User Meeting

Radiation Damage Effects

Total Dose DisplacementDamage

Single Event Error

hard SEE soft SEE

clock

data input line

data in register

expected data in register

© T. Wijnands

July 31st 2008 3CERN FLUKA User Meeting

Semiconductors

Polymers

Ceramics

Metals and alloys

10.0 1E2 1E3 1E4 1E5 1E6 1E7 1E8 1E9 1E10 1E11 Gy1E12 1E13 1E14 1E15 1E16 1E17 1E18 1E19 1E20 1E21 1E22 n/cm2

- no damage- mild to severe damage

- destruction

commercial COTS hardened electronics

accelerators

Radiation Damage to Materials/Electronics

!!! Assumption !!!(depends on particle energy spectra)

1 neutron (1MeV) /cm2 ~ 3.3E-11 Gy

Dose & Displacement Damage

© Lockheed MartinJuly 31st 2008 4CERN FLUKA User Meeting

!!! A Rough Overview Only !!!

High-Energy Hadron Fluences

104

e.g., LHC-Levels for Hadrons (E > 20 MeV) per cm2 per nominal year

105 106 107 108 109 1010 1011 1012

Aircraft Altitudes LHC Machine electronics equipment LHC Detectors

sea Level

(Lowest !!!)

Airbus A330

UAs(guess)

UJ76Under

ARC dipoleUnder

ARC quad

RE38

RR53RR77UX85

DS Q8UX45

UJ33

1013

TAN

© T. Wijnands


CERN FLUKA User Meeting 6

Neutron flux (> 20 MeV) in the atmosphere

105 cm-2 y-1

3 105 cm-2 y-1

July 31st 2008

Radiation Physics/Effects/Monitoring

DoseDisplacemen

t

Single Events

EM cascade

h, e,.. > 100 KeV

h > 20 MeV

RadfetPIN

Diodes

SEU counter

nuclear cascade

radiation damage in semiconductors

Radiation Monitor

Radiation Field

Effect in the Device

Measurement

© partly T. Wijnands


Dose in shielded aras (where the electronics is usually lcoated) is mainly due to neutrons (and associated photons)

Dose and neutron fluxes have a very close correlation Cumulative damage comes from

Energy deposition (dose) Lattice displacement (1-MeV n equivalent particle fluxes)

Stochastic failures can occur (SEU) and are mostly due to “high” energy hadrons (“E>20 MeV”)

No safe limit exists, only a risk level can be determined Risk level for commercial electronics is poorly known and varies

by orders of magnitude between different devices and series Only a combination of the following can assure safe operation:

Simulation studies of related radiation levels (Dose, 1MeV, 20MeV) Careful selection and testing of required electronics Shielding and displacement considerations

Important – Damage is Not Only Dose …


How To Calculate it with FLUKA


July 31st 2008 CERN FLUKA User Meeting 10

FLUKA-Implementation – HistoryFrom 1995 up to now:

An linked user code (user defined fluscw.f and comscw.f) had to be used to score the following quantities:

- Dose: Energy deposition in the respective material) -> comscw.f

- SEE: high-energy hadron fluence (>20MeV) -> fluscw.f

- Displacement Damage: folding of fluences with damage functions -> fluscw.fFrom the last release up to now:

Energy dose can be scored directly in FLUKA with the ‘pseudo-particle-type’ DOSE. Please keep in mind that the unit is GeV/g, thus needs to be multiplied with the respective coefficient to get (Gy [J/kg]). The remaining quantities have still to be calculated through user routines.

With the next release (and already in the current pre-release):

Displacement damage and high-energy hadron fluences can be directly scored in FLUKA using the respective ‘pseudo-particle-type’ (SI1MEVNE, HADGT20M)

July 31st 2008 CERN FLUKA User Meeting 11

FLUKA-Implementation – Main features

All important quantities to estimate risks of damage to electronics can be directly scored in FLUKA:Cumulative damage:

Energy deposition (dose) by scoring DOSE with any ‘energy deposition like estimator’ (e.g., USRBIN)

Lattice displacement (1-MeV n equivalent particle fluxes) with any ‘fluence like estimator’ (e.g., USRTRACK)

Stochastic failures (SEU): “high” energy hadron fluences (“E>20 MeV”) with

any ‘fluence like estimator’ (e.g., USRTRACK)

Respective levels can be illustrated through 3D meshes, as well as separately calculated for certain volumes (regions)

This, together with respective particle energy spectra efficient shielding options allows to select best possible locations or efficiently design shielding implementations

USRTRACK scores average d/dE (differential fluence) in a given region (SI1MEVNE, HADGT20M or any particle type)

USRBDX scores for the same quantities average d2/dEd (double-differential fluence or current) on a given surface (between two regions)

USRBIN scores the spatial distribution either of deposited dose, or fluence (1MeV or 20MeV) in a regular mesh (cylindrical or Cartesian) described by the user

USRBIN also scores the same quantites on a region basis

Electronic Damage - Related Scoring

* 1) high-energy hadron fluence spectrumUSRTRACK -1. HADGT20M -31. RADMON1 125. 170.Ust20MeVUSRTRACK 1D3 1D-14 &* 2) displacement damage spectrumUSRBDX 98. SI1MEVNE -41. TAIR RADMON1 150.Usx1MeVUSRBDX 1D3 1D-14 170. &* 3) dose distribution in a regular mesh through the geometryUSRBIN 10. DOSE -21. 100. 20. 200.UsbDoseUSRBIN -100. -20. -100. 100. 20. 150.&* 4) integrated high-energy hadron fluence on a region basisUSRBIN 18.0 HADGT20M -37.0 LSTREG 300.0 10000.0UsbReg20USRBIN FSTREG 0.0 -10000.0 1.0 1.0 1.0 &


A full simulation for all quantities, HADGT20M, DOSE and SI1MEVNE require a full calculation including the electromagnetic cascade and high enough thresholds:

Simulation Settings

* Standard DefaultsDEFAULTS NEW-DEFA

The latter – depending on the complexity of the geometry – might lead to long computing times. If one is interested in SEUs mainly, one can then consider a separate simulation scoring HADGT20M only, switching off the electromagnetic cascade and adjust the thresholds accordingly (with care!): * Standard Defaults

DEFAULTS NEW-DEFA* Switching off the electromagnetic cascadeEMF EMF-OFF** 20 MeV for protons and neutrons ONLY (to be looked at more detail -> see later)!PART-THR -2.0E-2 PROTONPART-THR -2.0E-2 NEUTRON

In addition, if one is interested to get results in heavily shielded locations one will require respective biasing and possibly a Two-Step approach (BIASING, USRDUMP,…)

see some examples later and for more details please refer to the FUM presentations related to biasing as well as CNGS!


Some LHC Applications

-Many Points

Currently under Review


Quite a lot of ‘dangerous’ regions …


LHC Point 7 UJ76 Downstairs

UJ76


UJ76 Upstairs

UJ76


Beam 1 – 20MeV Main Contributors

!!! Simulation over 600 meters !!!July 31st 2008 18CERN FLUKA User Meeting

Beam 1 – 20 MeV Main Contributors

Main Loss (Sampled)

Main ContributorsUJ76

Main ContributorsRR77


UJ76 Results – Beam1 / Beam2 20 MeV

Downstairsopen Geometer Hole

closed Hole Upstairs


IR7 Summary of Expected Levels 4.1 x 1016 as maxium loss assumption

Nominal (both beams): 2.3 x 1016

Ultimate (both beams): 3.7 x 1016


IR6 – UA67 – FLUKA TCDQ, TCDS,…

© B. Goddard, J. UythovenJuly 31st 2008 22CERN FLUKA User Meeting

IR6 – UA67 – FLUKA TCDQ, TCDS,…

TCDQ/TCDS/etc… loss assumptions

Tunnel simulation only

Effect of lateral holes estimated by empirical formulas

90cm hole, 8m long, attenuation factorof ~100

!!! tunnel only !!!

!!! tunnel only !!!


© calculations by S. Roesler

IR6 – UA67 – TCDQ, TCDS,… Fire detection almost in line of

sight of the hole in line with TCS ‘Ducts’ and fully open passage ~8 x 108 for 20MeV ~8 x 109 for 1MeV conservative loss-assumptions

X


IR6 – UA67 – TCDQ, TCDS,… Ethernet switch and power

converters behind hole pointing to TCDM

‘Ducts’ and fully open passage ~4 x 108 for 20MeV ~4 x 109 for 1MeV conservative loss-assumptions

X


IR6 – UA67 – TCDQ, TCDS,…

X X X


TI8 – Injection TestRadMon

Measurements


2 3 4 5 6 7 8

2

4

6

8

10

12x 10

9

Time[Hrs]

Had

ron fl

uen

ce [cm

-2]

RADMON1.2x1010 cm-2

FLUKA0.96 x 1010 cm-2 ± 3.2%

protons: 9.8% neutrons: 34.1% pos. pions: 21.9% neg.pions: 22.4% others: 11.8%

8RM08S, 26.5cm behind dump 2cm off beam axis

High energy hadron fluence

(scoring in a volume of 2 x 2 x 2cm3)

[1.03 x 1010 cm-2 ± 3.2%] (scoring in a volume of 5 x 5 x

5cm3) July 31st 2008 28CERN FLUKA User Meeting


FLUKA5.0 Gy (air) ± 10%

Dose

RADMON 4.73 Gy (Si)

FLUKA2.1 x 1010 cm-2 ± 2.5%

protons: 4.6% neutrons: 81.6% pos. pions: 5.3% neg.pions: 5.6% others: 2.9%

RADMON 2 x 1010 cm-2


1MeV and Dose Levels1MeV


CNGSElectronics Damage


CNGS Target Chamber

Temperature Probes

Ventilation Units

Target Chamber

ServiceGallery

Proton Beam Linetarget

horn He tube 1 reflector

He tube 2 decay tube

Single event upsets in ventilation electronics: caused ventilation control failure and interruption of communication

Electronics

Racks


ElectronicsRacks

Ventilation Units

CV, crane,fire detection

Dose Distribution – Electronics Dose Distribution – Electronics FailuresFailures

Gy/yr for a nominal CNGS year of 4.5 1019 pot


July 31st 2008

CERN FLUKA

User Meeting33

Unit 232

TSG44

Unit 231

Ventilation duct

Balcony TCV4

“Hard” spectrum: ‘In-Line’ with Tunnels

CERN FLUKA User Meeting 34 July 31st 2008

Thermal neutron dominated

“Soft” spectrum: ‘Shadowed’ Part


CNGS: RPL and Alanine Dosimeter Positions

Gy/yr for a nominal CNGS year of 4.5 1019 pot

TCV4 TSG4


FLUKA Comparison: Alanine and RPL dosimeters

Detector type / position description

Exp. Value(Gy/p)

Simulation(Gy/p)

RPL: perp. horn strip lines 1.7·10-15

1.2·10-15Alanine: perp. horn strip lines

9.0·10-16

RPL: perp. refl. strip lines 8.2·10-16

5.4·10-16

Alanine: perp. refl. strip lines 3.8·10-16

RPL: perp. to the target 1.7·10-15

3.7·10-16

Alanine: perp. to the target 3.4·10-16

RPL: top of PMI404 9.4·10-18

7.2·10-18

Alanine: top of PMI404 4.3·10-18

Hole in shielding

not included

in simulatio

n

Alanine and RPL dosimeters are in principle sensitive to both rays and neutrons, RPL are more sensitive to n due to some

Boron content`


Good Agreement

Detector readings thanks to H. Vincke et. al.

CNGS: Holes in the Target Shield (Target Motors)


CNGS: PMI ioniz. Chambers – Only Qualitative

Saturated (correction in red)!!

0.19(x0.8)

0.14(x0.6)

0.32

0.79(x1.2)

0.50(x1.3)

Numbers are exp/FLUKA ratiosafter (rough) correction of PMI data for

saturation , background

0.77(x1.1)0.67

(x24)1.1

(x11)5.2

(x52)2.4

(x12)



TLD’s: Dose Equivalent

Absorbed dose (Gy) for an exposure of 7.2 1016 potJuly 31st 2008 40CERN FLUKA User Meeting

TLD dosimeters

TLD / position descriptionExp. Value

(Sv/p)Simulation

(Sv/p)

n dose: UA232/TSG4 2.3·10-17

1.3·10-17

γ dose: UA232/TSG4 4.3·10-18

n dose: UA231/TSG4 8.1·10-18

2.6·10-18

γ dose: UA231/TSG4 2.5·10-18

n dose: fire detector (TCV4) 4.0·10-18

2.4·10-18

γ dose: fire detector (TCV4) 1.2·10-19

n dose: pump cupboard (TCV4)

1.3·10-18

6.9·10-19

γ dose: pump cupboard (TCV4)

5.0·10-20

Sim. dose equivalent values (Sv) has been obtained out of calculated absorbed dose in air (Gy), correcting for air tissue and weighting for the neutron quality factors with a couple of representative spectra. The related uncertainty is +/- 30% on top of the (large) statistical one



New Shielding Layout


Simulation ApproachBiasing Region importance biasing for

hadrons and neutrons Leading Particle biasing for

electromagnetic showers (inside the target chamber shielding)

“blackhole” where possible

Two-Step Approach First simulation to collect the

spectrum of particles (mainly neutrons) moving towards the additional shielding locations

Separate simulations to evaluate the attenuation due to each plug configuration


The final configuration is a “fake chicane” for one of the ventilation pipes in the movable shielding part and two (real) chicanes in the upper fix shielding part.

Final Shielding Design


Annual Dose (Gy)

Without Shielding

With New Shielding

4.5 x 1019 protons / year


1 MeV Equivalent Neutrons

Without Shielding

With New Shielding



> 20 MeV Hadron Fluence

Without Shielding

With New Shielding



What’s Happening To SEU

Below 20MeV ???


49

SEU Sources & Brief History of SEE Research Cosmic Rays - On ground: neutrons (MeV – GeV) - High altitudes: neutrons, protons, pions (MeV – GeV) - Space programs: high-energy protons & heavy ions (GeV) - Cosmic ray-induced soft fails predicted in 1979 (IBM) Observed in space programs (early 1980s) Neutron-induced SEUs experimentally verified in mainframes (mid-1980s,

IBM) 1st Monte Carlo SEU simulation tool for product designs, SEMM-1 (1986,

IBM) SEMM-2: full BEOL+FEOL analysis & linkage to high-level tools (2001-

present,IBM)

Man-Made Radiation Environments SEUs & other SEE-related problems discovered in high-energy & nuclear

physics experiments in 1990s at BNL, FNL, CERN, …..

Thermal Neutrons BPSG (heavily doped w/ 10B): n(th) + 10B 4He (1.74 MeV) + 7Li (0.84 MeV) Predicted by Fleishman (1980); reported by TI (1995)

Alphas in IC Materials: eg. 206Pb 4He (5.3 MeV) (Bell Labs. & Intel, 1979).

© H. Tang/CERN Talk/20-Sep-2007

July 31st 2008 CERN FLUKA User Meeting

50

Metallization layer 1

Layer 2

Th Foil (Alpha Source)

Alpha particle

Diffusion

Depletion Layer

Funneling Region

Alpha particle

Recoil Ion

Cosmic RayNeutron

++

+

++

+

++

+++

+

++ + +

+ +

+ +

+

--

--

--

--

-

-- - -

--

- --

--

Layer 3

Collision Nucleus

Particle Origin of SEU – Nuclear Physics

SEUs are noise problems triggered by intruding charged particles.

In bulk devices, charge collection in an SEU event is due to a field-assisted funneling mechanism.

In SOIs, SEU can be caused by source-to-drain shunting, or parasitic bipolar gain effect, triggered by charge deposited by an ionizing particle.

Cosmic ray particles like p, n, interact w/ IC materials via spallation reactions. These reactions produce secondary nuclear fragments (light ions, heavy recoils). A charged secondary w/ sufficiently large LET causes a soft fail.



51

Internal Radiation Sources: Alpha particles

Package (Th, U)

Wiring

Silicon

Solder Ball210Pb, 210Po

neutron

Devices

• Alpha particles emitted from radioactive impurities have energies up to 9 MeV and travel distances up to 50 m.

• Alpha particles deposit ~ 1 – 10 fC/um along their path.• Flux = 10-2 alpha / chip / Hr




dE/dx for alpha’s in Silicon:

High ionization density

July 31st 2008


28Si(n,xα) cross section below 20 MeV:

Incident neutron data / ENDF/B-VII.0 / Si28 / /

Incident Energy

Cros

s-se

ctio

n

10.000.000 eV5.000.000 eV4.000.000 eV 6.000.000 eV 8.000.000 eV 20.000.000 eV

0,01 b

0,1 b

0,005 b

0,05 b

0,5 b

MT=22 : (z,na) TotalAlphaMT=107 : (z,a) Cross sectionMT=22 : (z,na) Cross section

A possible guess about the (relative) response to neutrons at E<20 MeV? And if so what is the impact of using this curve below 20 MeV (normalized @ 20 MeV)

and 1 above?

Thresh.: 2.7 MeV

July 31st 2008


Impact on typical CERF n-spectra:

+ 80%

+ 25%

+65 %

July 31st 2008

estimating “damage 2 electronics” using fluka m. brugger for the fluka team cern fluka user...

Documents

damage mild

mev cm2

commercial electronics

particle energy spectra

neutron fluxes

safe operation

determinedrisk level

rough overview