comparison of full and analytic simulations

15
L.Dehimi 29-6-2007 1 Comparison of Full and Analytic Simulations Salim Aoulmit, Lakhdar Dehimi, Achour Saadoune (LMSM Laboratory Biskra U) Craig Buttar, Dima Maneuski (Glasgow U) Chris Bowdery, Michal Koziel, André Sopczak (Lancaster U)

Upload: salah-eddine-bekhouche

Post on 10-Mar-2015

43 views

Category:

Documents


3 download

DESCRIPTION

Comparison of Full andAnalytic SimulationsSalim Aoulmit, Lakhdar Dehimi, Achour Saadoune(LMSM Laboratory Biskra U)Craig Buttar, Dima Maneuski(Glasgow U)Chris Bowdery, Michal Koziel, André Sopczak(Lancaster U)

TRANSCRIPT

Page 1: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 1

Comparison of Full and Analytic Simulations

Salim Aoulmit, Lakhdar Dehimi, Achour Saadoune(LMSM Laboratory Biskra U)Craig Buttar, Dima Maneuski

(Glasgow U)Chris Bowdery, Michal Koziel, André Sopczak

(Lancaster U)

Page 2: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 2

Motivation

• Further understanding of the ISE Simulation Results (Full Simulation) for the 2-phase CP-CCD.

• Development of Analytical Model and comparison with Full Simulation.

• Studying the effect of Occupancy.• Longer term: Fast CTI determination over

a larger range of temperatures.

Page 3: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 3

Introduction• Recall: good agreement between

Analytical Model and 3-phase CCD58 Full Simulation.

• Difference between 3-phase and 2-phase: - The shift time is tsh=1/2*f- Factor of 2 (2 nodes) instead of 3 (3-nodes).

• Effect of occupancy on CTI value is introduced by correlation of waiting time and clock frequency.

Page 4: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 4

Model (ref : T. Hardy IEEE 98)

( )

packetparent he t

joincan charges which theduring periode time theis packet previous thefrom imeemission t total theis where

3

join

wemit

tt

s

t

ttt

eenNCTI eemitejoin

=

−= −− ττ

shc t<<τ

● The model considers the effect of a single trapping level and include the

emission time only in the following differential equation :

● Traps initially filled

trapsfilled ofdensity theis n where te

tt ndtdn

τ−=

Page 5: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 5

Hardy paper considers :

● filled traps

( )( )eecsh temittjoint

s

eeenNtCTI τττ −−− −−= 1*2

csht τ>>

csht τ<<Or in our case:

Analytical model is given by:

Page 6: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 6

Input Parameters

• Trap concentration: 1e11/cm3• Occupancy occ: 1% • Signal charge density Ns: 4.5e14/cm3• Frequency f: 50, 25, 10 MHz• Shift time: 1/2f• Waiting time tw: 1/(occ*f)

Page 7: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 7

0.17eV traps. 50MHz

Page 8: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 8

0.44eV traps. 50MHz

Page 9: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 9

Occupancy Study

• Simultaneous change of occupancy occand waiting time tw.

• At very high temperatures, the emission time is very short, and independent of the occupancy traps are empty when the next charge package arrives.

Page 10: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 10

Occupancy 0.17eV traps. 50MHz

Page 11: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 11

Occupancy 0.17eV traps. 25MHz

Page 12: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 12

Occupancy 0.44eV traps. 50MHz

Page 13: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 13

Occupancy 0.44eV traps. 25MHz

Page 14: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 14

Occupancy 0.44eV traps. 10MHz

Page 15: Comparison of Full and Analytic Simulations

L.Dehimi 29-6-2007 15

Conclusions/Outlook• CP-CCD: Glasgow and Lancaster full simulations agree

well.• For 0.17eV traps: analytical and full simulations also

agree well, and for low temperature region analytical model predicts CTI values.

• For 0.44eV traps, analytical simulations agree much better with full simulations compared to CCD-58.

• Reason: Improvement to analytical model: tjoin used for 0.44eV is larger than the one used for 0.17eV. Also improvement for CCD-58 agrrement possible.

• Occupancy study is consistent with Konstantin’s work (PhD thesis pp57-58). Further comparisons with full simulation needed.