Multi-physics SEE modeling with MUSCA SEP · Multi-physics SEE modeling with MUSCA SEP 3...
Transcript of Multi-physics SEE modeling with MUSCA SEP · Multi-physics SEE modeling with MUSCA SEP 3...
Outline
• Physical bases of MUSCA SEP3
� Global approach, sequential modeling� Levels description
• Operational rate calculations� Operational calculation� Emerging effects� Anomaly expertise
• Synthesis and perspective
RADPRED2010 1st workshop 14-15 January Toulouse
Introduction
→→→→ MUSCA SEP3
� Pragmatic & global approach based on physical mechanisms & sequential modeling
Why developing a new SEE predictive method ?� Operational rate: SEE sensitivity of the device considered in its material, radiation and operational environments� SEE anomaly analysis and dynamic operational rate
→ “space weather” problematic and forecasting rate� New and relevant methodologies for modern devices
� To investigate the rate trends induced by technological roadmap� To prevent the emerging effects
RADPRED2010 1st workshop 14-15 January Toulouse
Outline: physical bases of MUSCA SEP3
• Physical bases of MUSCA SEP3
� Global approach, sequential modeling� Levels description
• Operational rate calculations� Operational calculation� Emerging effects� Anomaly expertise
• Synthesis and perspective
RADPRED2010 1st workshop 14-15 January Toulouse
Global approach, sequential modeling
Environment
Transport in materials
Interaction in device,e/h generations
e/h transport in SC, collection mechanisms
Circuit effects
SEE sensitivity modeling
Monte-Carlo simulation of radiation events
Electron hole pairs generationTransport carriers and charge collectionCircuit effect
Device and material environment (structure, shielding, package)
System modeling
σσσσ, SER Post processing
RADPRED2010 1st workshop 14-15 January Toulouse
Layer 1: system modeling
Example : 6T SRAM bulk
Translation and symmetry rules
Elementary cell topology Memory (several Mbits)
Full description of the device on its material environment
Structure, shielding and package
Passivation-Metallization
S.C
Wafer
RADPRED2010 1st workshop 14-15 January Toulouse
Layer 2: environment modeling
Cosmic, radiation belt, sporadic event (solar flare), neutron
Neutron (fast and thermal), proton, muons, pions
alpha emitter contamination
Heavy ion, proton, neutron, pion …
Space environment Atmospheric and ground environment
Intrinsic environment
Accelerated (test)environments
Need to model:
Radiationfield
� Isotropic, unidirectional or anisotropic� Ion species, proton, neutron� Mono-energetic or spectrum� Dynamic: sporadic events (solar flare)
RADPRED2010 1st workshop 14-15 January Toulouse
Layer 3 & 4: interaction, transport and e/h generati on
GEANT4 SRIM −+−+−+−+−
+−+−
Particle/system interactions:� Energy loss during transport in structure, shielding and package� Secondary ions induced by nuclear reaction in S.C.� Energy loss in semiconductor by coulombic interaction
−+−+−+−+−
+−+−
+−−
++−−+−+
Radiationfield
Secondary ions induced by a nuclear interaction
Energy loss in matter, e/h pairs created in S.C.
Coulombic database:ion (1 → 92) in Si, Cu, Al, O, Si02, W, Ta (...)
Nuclear databases:Neutron - proton with Si, O, Cu and W (1MeV to 2 GeV)∑
ion
v A, Z,E,r
N Nucl. reac. E, range, LET
RADPRED2010 1st workshop 14-15 January Toulouse
Layer 4: carrier generation in semiconductor
Objective of this layer: To model and quantify the carrier density deposited in the device (active semiconductor)for a given configuration issued from radiation events
n(x,y,z) : e/h pairs density in active semiconductor
−+
−+−+−+−+−
+−+−
+− +−−+−+
∑ion
v A, Z,E,r
Location in S.C. (MC process)
Database from SRIM Database from GEANT4
Nuclear reaction characteristics (MC process)
Next layers: Environment and transport in materials
+ −− ++ −− ++ −− +
heavy ion
RADPRED2010 1st workshop 14-15 January Toulouse
Layer 5: carrier transport and charge collection (1 /3)
OffOn
Off On
Two ways: transient pulse OR charge collection→ SEE sensitivity model adapted for Monte-Carlo approaches
� Ambipolar and drift diffusions� Potential/impedance variations� Charge injection processes� Parasitic structure activation� …
Physical mechanisms
I(t) or Qcoll
simplified model
I
−+
−+−+−+−+−
+−+−
+− +−−+−+
I(t) or Qcoll
I(t) or Qcoll
I(t) or Qcoll
I(t) or Qcoll
Circuit model (layer 6)
Injection in Spice
TCAD & SPICE
I(t) or Qcoll
Multi-cell and multi collection
RADPRED2010 1st workshop 14-15 January Toulouse
Layer 5: carrier transport and charge collection (2 /3)
Qi
Drain
Qcoll = Σ ηi . Qi
� Ion track: series of local charges Qi along the target� Collection efficiency ηηηη: decreases with the distance to the collected zone
SEU charge model� Classical approach: RPP
� MUSCA SEP3 approach: collection efficiency concept
−+−+−+−+−
+−+−
+−−
++−−+−+
Collected charge by eachsensitive zone→ Multi-collection effects
Qcoll
Qcoll Qcoll
Qcoll
RADPRED2010 1st workshop 14-15 January Toulouse
Layer 5: carrier transport and charge collection (3 /3)
Qi
Drain � Ion track: series of local charges Qi along the target� Spherical ambipolar diffusion � Impact on potential variation� Dynamic properties: D(t) and v(t)
Transient model: ADDICT→ Advanced Dynamic DIffusion-Collection Transient model
)v ,E A, Z,,V ,S,N D(t), v(t),fct(t, (t)I ioniondddrainsubdrain
r=
Ion characteristicsStructure/device characteristics
MUSCA SEP3 ADDICTSEU I(t) model (criteria)SET in digital electronics (SPICE)SET in APS
Dynamic Model
RADPRED2010 1st workshop 14-15 January Toulouse
Layer 6: Circuit effects
Objective of this calculation layer: To propose a simplified model accounting for the impact induced by the circuit level
Vdd Masse
I on
NMOS ON PMOS Off
I coll
Vnoeud 1���� Vnoeud 2����
Diff./Coll
nm
µm
Coupling effect
Multi collection
Injection in Spice
∑=zone collected :i
).state on/off type,n/p( icollcell QCQ
Examples: SRAM cell memory
OffOn
Off OnI(t)
I(t)
I(t)
I(t)
Simplified model
RADPRED2010 1st workshop 14-15 January Toulouse
Outline: operational rate calculations
• Physical bases of MUSCA SEP3
� Global approach, sequential modeling� Levels description
• Operational rate calculations� Operational calculation� Emerging effects� Anomaly expertise� Sporadic event analysis
• Synthesis and perspective
RADPRED2010 1st workshop 14-15 January Toulouse
Operational calculation (1/3)
Example : the ICARE equipment on-board SAC-CHitachi 628512 4T SRAM bulk
1E-10
1E-09
1E-08
1E-07
1E-06
0 20 40 60
LET in MeV.cm²/mg
SE
U c
ross
set
ion
in c
m²/
bit
MUSCA SEP3
Experiment
1E-15
1E-14
1E-13
1E-12
0 50 100
Proton energy in MeV
SE
U c
ross
sec
tion
in c
m²/
bit
MUSCA SEP3
Experiment
Structure analysis Technological analysis SEU ground test results
Low (3 mm)
High (3 cm)4Mbit
Technological SEU sensitivity
model
RADPRED2010 1st workshop 14-15 January Toulouse
Operational calculation (2/3)
Thin (3 mm)
Thick (3 cm) 4Mbit
Technological SEU sensitivity model
In-flight rate (#/dev./day)
SEU /day/dev
SAC-C experiment
MUSCA SEP3
Total 1.1 1.24Heavy ion 0.16 0.19
Proton 0.94 1.05
Good agreement between on-board experiment and MUSCA SEP3 calculations� for both heavy ion and proton contributions !
Example : the ICARE equipment on-board SAC-CHitachi 628512 4T SRAM bulk
RADPRED2010 1st workshop 14-15 January Toulouse
1E-15
1E-14
1E-13
1E-12
0 50 100 150 200
Proton energie (MeV)
Cro
ss s
ectio
n cm
²/bi
t
waferwafer + packagewafer + package + 3mmwafer + package + 3cm
Operational calculations (3/3)
Structure (shileding) and environments characteristics (angular distribution) with protons & a 90nm techno
1E-15
1E-14
1E-13
1E-12
0 50 100 150 200
Proton energy (MeV)
Cro
ss s
ectio
n cm
²/bi
t
no shielding
no uniform shielding
uniform shielding (3cm)
uniform shielding (1cm)
Uni-directional flux Isotropic flux
• Very important impact on threshold • Need to describe the non-uniform structure/shielding• Impact induced by the directional properties (threshold and amplitude)
� Factor of 4 on predicted rates between classical and MUSCA approaches (SAC-C orbit considering resp. w/o and w MBU)
RADPRED2010 1st workshop 14-15 January Toulouse
Operational calculation: emerging effect analysis
Direct ionization of proton impacts on SER…. and on the requirements for evaluating the SEU sensitivity
Various material environments
Emerging effect for nanometrictechnologies: the trend
The material environment is the key parameter→ Define the sensitive proton spectrum energy range
1
10
100
Device Dev. + 3mm Dev. + 1cm Dev. + 3cm Dev. + 6cmS
ingl
e E
vent
Rat
e in
SE
U /d
ay/M
bit
AP8-min (nuclear process) + Cosmic
AP8-min (direct ionization process)
Total SER
88% to total SER
58%
Heavy ion
Nuclear proton
Ionizing proton
130nm 25 % 75 % 0 %
90nm 23 % 59 % 18 %
65 nm 1 % 19 % 75 % Classical approach: cosmic + nuclear proton
wafer
Structure/shielding
package
65nm
RADPRED2010 1st workshop 14-15 January Toulouse
1E-17
1E-16
1E-15
1E-14
1E-13
0 200 400 600 800 1000
Proton energy
SE
U c
ross
sec
tion
cm²/b
it
With Plug W
Without Plug W
Operational calculation: anomaly expertise
Example: “hard” devices� Ground tests: heavy ion low SEU sensitivity
→ very low SEU sensitivity to proton
� Operational data: no SEU induced by cosmics but by protons (heavy recoils)
6T SRAM SOI - Qcrit = 100 fC
High energy proton test
requirements !!!
Needs for describing the “system-device” with multi material approach
RADPRED2010 1st workshop 14-15 January Toulouse
Outline: synthesis and perspective
• Physical bases of MUSCA SEP3
� Global approach, sequential modeling� Levels description
• Operational rate calculations� Operational calculation� Emerging effects� Anomaly expertise
• Synthesis and perspective
RADPRED2010 1st workshop 14-15 January Toulouse
Synthesis
Needs for taking into account• An adapted and realistic environment model
� Isotropic/mono-directional, spectrum/mono-energy
� Dynamic and/or static,
• A 3-dimentional structure and device description� Shielding, structure and device (uniform and non uniform)
� Multi-material (Si, SiO2, W, Cu, Al …)
• Nuclear and coulombic interactions� Do not forget the new problematics! (ex: direct ionization of proton)
• Adapted SEE sensitivity models� Specific for each effect and/or device type
� Dynamic and/or static effect
� Thermal effect
• Circuit effect model
RADPRED2010 1st workshop 14-15 January Toulouse
Perspective: post processing and effect calculation s
VHDL⇒
Research activities
Charge or I(t) histogram
Post processing
System modeling
Able to investigate:► Operational conditions► Ground conditions
Output databases
Environment
SER
B Env.cal.
A Env.exp. σσ ⇒
Transport in materials
Interaction in device,e/h generations
e/h transport in SC, collection mechanisms
MUSCA SEP3
SEE sensitivity model from ground test
Proton
Neutron
Heavy ion14 8 2 5 818 6 3 2 962 3 963 8 9 2628363256 325693256 9832561482581863 2962396389 2 6 2 836 3 2 5632 56932 5698 32561 4825 8186329623 96389 2628 36 32563 25 6932 569 83 25 61482 5818632962396 3892628363256325 693 25 69 83 2514 825 8186 3 62396 389 26 2 3632 56 32 5 69 32 61 482581 863 96 2396 389 262 83632563 25 69 32 569 832 561 482 581 8632 962396 3892628 36325 632 69 32 5 69 83 56 14825 818 6329623 96 8926 28 3256 32 56 9 32 5
Multi-physic model
Static and dynamic
SPICE⇒
SEEσ
RADPRED2010 1st worshop 14-15 January Toulouse