Upload
phamnhan
View
217
Download
0
Embed Size (px)
Citation preview
component degradation simulation tool
A. Keating, S. Joyce, A. Zadeh, M. Pimenta,
E.Daly, P.Gonçalves
ESA Project:
18121/04/NL/CH
Simulation of Single Event Effects and rate
prediction. CODES an ESA tool
Heliosphere image courtesy of the Cosmic and Heliosphere Learning Centre
23/05/2013 2SPENVIS SUW
2013
Outline
• Simulation of SEE effects
– Environment thru shielding and processing layers
– Energy deposition distribution at SV
• Sensitive Volume Fit (SVFIT) Module
• CODES: the top level integrated tool
• Additional Models developed to be included later
• Conclusions
23/05/2013 4SPENVIS SUW
2013
Single Event Effects
• A transient phenomenon : Decaying with time
• e-h pair generation threshold in the target material (3.6eV Si)
23/05/2013 5SPENVIS SUW
2013
SEE Mechanisms:
Stopping power and nuclear Reactions
• Keating et al. (2012). Modelling the effects of low-LET cosmic rays on electronic
components, Radiat Environ Biophys, DOI 10.1007/s00411-012-0412-2.
• Stopping power
• σ[X(a, b)y] nuclear cross section
• ηc(x) is the charge collection yield
• Z and A
23/05/2013 6SPENVIS SUW
2013
Drift & Diffusion in the semiconductor
• Electron-hole pairs are separated and transported by the electric field
(drift) and by diffusion
Ji= q ni µi Ε + q Di grad(ni);
drift diffusion i=e,h
• Drift: t~0.5ns d<10µm
• Diffusion: t>0.5ns d>10µm
• Part may recombine
(function of injection level, Ma 1989)
• Charges collected by electrodes may propagate
thru the circuit: Transient “ionocurrent”
23/05/2013 7SPENVIS SUW
2013
Drift & Funnelling
• Drift charge : prompt charge collection (~0.5ns)
• Before the ion the Electrical field, E, is limited to the depletion region
• The ion track (or the e-h pair plasma) extends E far down into bulk Si
C. M. Hsieh, P. C. Murley, 1981
23/05/2013 8SPENVIS SUW
2013
Device Response
• The device behaviour depends on the field (ξ), Incoming particle (βγ
∼P/M) and on the angle of incidence (θ)
• The critical charge depends on the node capacitance, voltage and the
response time of the device
eeeee
eeeee
eeee
N4
Word Select line
N1
1
P1
0
N2
P2
1
0N3
eeeee
eeeee
eeee
N4
Word Select line
N1
1
P1
0
N2
P2
1
0N3 N4N4
Word Select line
N1
1
P1
0
N1
1
P1P1
0
N2
P2
1
0
N2
P2
1
P2
1
0N3N3N3
restoreswitch nodenodecrit .I V . C Q τ+=
( ) critcollected Q, , f Q >= θλξ
23/05/2013 9SPENVIS SUW
2013
Which answers can GEANT4 Give?
• Particle physics simulation tool, for particle transport and
interaction with matter used for simulating:
– Shielding
– Packaging Layers
– Sensitive Volume
23/05/2013 10SPENVIS SUW
2013
What are the remaining questions?
• How does the energy/charge deposited is transported
in the device:
– Crystal Lattice?
– Electric fields?
– Time dependent effects?
• What is the electronic behaviour?
• How to get those answers?
23/05/2013 11SPENVIS SUW
2013
How ?
• Experimental data
– Device Response function
– Function of field effects
– Cristal lattice effects
– Experimental conditions
• SVFIT:
– Fits device sensitivity to adjust experimental device response
function
23/05/2013 12SPENVIS SUW
2013
SVFIT: Web interface
• SVFIT_V. 1.5.7
• Meta Information
• SV geometry
• Email account
• Efficiency matrix: No of SV
• Library
• Non Active layers
• Guideline SV dimensions
• Depth uncertainty ->
Interactive Fit
• Experimental Cocktail
• No. Events
• Credits
• Run SVFIT
23/05/2013 15SPENVIS SUW
2013
From Ion data to proton prediction
• Using SVFIT with the Ion test data
• The same SV geometries and Critical Energies
• Proton with energies used in experimental testing were simulated
23/05/2013 16SPENVIS SUW
2013
SVFIT working for different devices
• Tests have been made for the Reference SEU Monitor
and SEL monitor devices
y = 6.12E-05x + 1.18
y = 8.11E-05x + 1.35
y = 1.01E-04x + 1.51
0
2
4
6
8
10
12
14
1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07
No. Events per ion
Run Time [mm:ss]
3 ions cocktail
4 ions cocktail
5 ions cocktail
Linear (3 ions
0
10
20
30
40
50
60
70
80
1.E+02 1.E+03 1.E+04 1.E+05 1.E+06
No. Events per ionAccuracy [%]
3 ions Cocktail
5 ions Cocktail
23/05/2013 18SPENVIS SUW
2013
Geometrical acceptance
Normalization
1.E-15
1.E-14
1.E-13
1.E-12
1.E-11
1.E-10
1.E-09
1.E-08
1.E-07
1.E-06
1.E-05
1.E-04
1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 1.E+06
Primary Fe Energy [MeV]
Flu
x
Integral Flux [/cm2/sr/s]
Differential Flux[/cm2/MeV/sr/s]
Primary Integral Flux Normalization
Theta Max
Total Flux Norm. = PrimFlxuNorm * GeomAccept * SolidAnglNorm
Angular normalization
According to the ICR
• Normalization is based on dMEREM/MARSREM normalization methods
23/05/2013 19SPENVIS SUW
2013
Full spectrum simulation
• CODES pre-processor takes inputs for several ions’
energy spectra in SPENVIS format and others
• Computes individual contributions for SEE rates
• Outputs the total rate prediction
23/05/2013 20SPENVIS SUW
2013
CODES Performance
• The framework is working properly under :– Windows Internet Explorer
– Google Chrome
– Firefox
y = 6.12E-05x + 1.18
y = 8.11E-05x + 1.35
y = 1.01E-04x + 1.51
00:00:00
00:28:48
00:57:36
01:26:24
01:55:12
02:24:00
02:52:48
03:21:36
03:50:24
04:19:12
04:48:00
1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07
No. Events per ion
Run Time [hh:mm:ss]
30 ion species environemnt
60 ion species environment
92 ion species environment
Linear (30 ion species
1.E-05
1.E-04
1.E-03
1.E-02
1.E-01
1.E+00
1.E+01
1.E+02
0 10 20 30 40 50Charge Number [Z]
Accuracy [%]
1E3 Evts
1E4 Evts
1E5 Evts
x10
x 5
x 3
x 130
23/05/2013 22SPENVIS SUW
2013
Area increment above LET knee: Model & Results
• Results published at RADECS 2011
• Doi: 10.1109/RADECS.2011.6131414
23/05/2013 23SPENVIS SUW
2013
Efficiency matrix from Laser maps
• The model of defining acomplex efficiency matrix was developed for
SVFIT
• Objective : robust module for extraction from laser maps the charge
collection efficiency
• SVFIT and CODES : benefit from the inclusion under the user-friendly
interface
Images from of Isabel Lopez
23/05/2013 24SPENVIS SUW
2013
Conclusion
• CODES modules: top level user friendly tool, with a web-interface
• Tests show the robustness of the tool: consistency, accuracy & run time
• Results show :
– Very good reconstruction of device response function (3-5 ion cocktails):
excellent SVFIT
– Accuracy statistics independent (5 ion cocktails)
– Run time : SV shape fit : (3 to 5 ions) - 2 geometries : btw 2-10 min
statistics
– Run time : SV shape and depth fir: (5 ions) - 6 geometries : ~ 30 minutes
– mCODES higher independence from user definitions: stand. statistical
methods and sCODES
23/05/2013 25SPENVIS SUW
2013
– Distribution of the tool
– Inclusion of ready-to-use developed models
– Incrementation of the Device Library
– Further models investigated: TRL needs to be increased
Further Work
23/05/2013 31SPENVIS SUW
2013
User defined NAL
Pre-processor:
� Verification of the total weight percentage in the material is = 1
� Send error messages
� Compute total density
� Compute layer positions in relation to device structure
� Write NALFile.txt with NAL table
� All have been verified
23/05/2013 34SPENVIS SUW
2013
Device Response Function
ID Plateaux With Exp Thres.
n 3.53E-14 4 1 10
p 3.53E-14 4 1 10
hi 1.50E-07 65 1.4
7
1.2
SEU = No. Event * NormF
23/05/2013 35SPENVIS SUW
2013
SEU calculation with mCODES
SEU Rate = No Evts (Edep > Ec)/NGen * Total Flux * Active Area
23/05/2013 36SPENVIS SUW
2013
1.E-06
1.E-05
1.E-04
1.E-03
1.E-02
1.E-01
1E-06 0.0001 0.01 1 100
LET @ SV [MeV.cm2/mg]
To
tal F
lux
[/c
m2
/s]
Differencial
Cumulative
SEU calculation with sCODES
SEU Rate = No Evts (LET> Th)/NGen * Total Flux * <SEU_XS>
2
23/05/2013 37SPENVIS SUW
2013
SVFIT automatic/iterative
• Modular iterative tool
• Microdosimetry Monte-Carlo technique
• Device sensitive volume: charge deposited contributes entirely to prompt charge
collection.
• Input parameters:
– Ion cocktails description
– Irradiation test data
• SEU threshold definition
• SV shape modulation
• Output:
– best SV shape to fit ion test data
– Threshold Energy loss for SEE
– Reconstructed SEE XS curve
N
Ne
Si
Ar
Kr