Transcript Document

Total Ionizing Dose Effect on
Programmable Input Configurations
J. J. Wang, R. Chan, G. Kuganesan, N. Charest, B. Cronquist
Actel Corporation
JJ Wang
1
2005 MAPLD, Paper 240
Outline
Total Ionizing Dose Testing
Input Threshold TID Testing Data
Annealing Effect
Failure Analysis and Mechanism
Lesson Learned and TID Hardening
JJ Wang
2
2005 MAPLD, Paper 240
Total Ionizing Dose Testing
TM1019 Military Standard for TID testing (Fig 1)
1 Pre-Irradiation Electrical Tests
2 Radiate to Specific Dose
Fail
3 Post-Irradiation Functional Test
Redo Test Using Less
Total Dose
Pass
4 Post-Annealing Electrical Tests
Fig 1 TID testing flow
JJ Wang
3
2005 MAPLD, Paper 240
DUT and Irradiation
0.25µm CMOS technology
Commercial off-shore foundry
VCCI/VCCA = 5V/2.5V
TTL I/O configuration
Defense Microelectronic
Activity (DMEA)
Co-60 Source
Dose Rate = 1 krad(Si)/min
(±5%)
Room temperature irradiation
Static biased irradiation
Fig 2 Picture showing Gamma-ray irradiator
JJ Wang
4
2005 MAPLD, Paper 240
Parameter Measurement
Parameters
Logic Design
1 Functionality
All key architectural functions
2 ICC (ICCA/ICCI)
DUT power supply
3 Input Threshold (VIL/VIH)
Input buffers
4 Output Drive (VOL/VOH)
Output buffers
5 Propagation Delay
String of buffers, Clock to Q
6 Transition Characteristic
D flip-flop output
VIL defined as the start of low to high transition
VIH defined as the start of high to low transition
TTL trip point (average of VIL and VIH) ~ 1.5V,
CMOS ~ 2.5V
JJ Wang
5
2005 MAPLD, Paper 240
Rad-induced Input Threshold Shift
Table 1 Lot B Pre- and Post-Irradiation VT (Net 0)
Pre-Irradiation
DUT
Five (A, B, C, D, E in
chronological order) lots
from foundry X tested
2 lots (B, C) show VIL/VIH
switching from TTL to
CMOS
The number of events
increases with total
accumulated dose and can
be removed by annealing
JJ Wang
Post-Irradiation
Total Dose
VIL (V)
VIH (V)
VIL (V)
VIH (V)
B1
100 krad
1.25
1.48
1.26
1.53
B2
100 krad
1.24
1.48
2.29
2.52
B3
100 krad
1.25
1.47
1.25
1.47
B4
100 krad
1.25
1.49
1.23
1.51
B5
100 krad
1.25
1.47
2.32
2.55
Table 2 Lot C Pre- and Post-Irradiation VT (Net 0)
Pre-Irradiation
DUT
6
Post-Irradiation
Total Dose
VIL (V)
VIH (V)
VIL (V)
VIH (V)
C1
60 krad
1.24
1.51
1.38
1.45
C2
60 krad
1.25
1.52
1.22
1.53
C3
60 krad
1.25
1.51
1.24
1.48
C4
100 krad
1.25
1.52
1.23
1.49
C5
100 krad
1.24
1.51
2.41
2.67
C6
100 krad
1.23
1.50
1.24
1.56
C7
100 krad
1.25
1.51
1.26
1.48
C8
100 krad
1.26
1.52
1.41
1.57
2005 MAPLD, Paper 240
Post-Irradiation Input Threshold
Switching from TTL to CMOS
Lot C is chosen for investigation
More design nets are tested for post-irradiation input
threshold
Part to part and pin to pin dependence observed
Table 3 Lot C Post-Irradiation VT
Net 1
JJ Wang
DUT
Total Dose
C1
Net 2
Net 3
Net 4
VIL (V)
VIH (V)
VIL (V)
VIH (V)
VIL (V)
VIH (V)
VIL (V)
VIH (V)
60krad
1.43
1.46
1.38
1.44
1.39
1.46
NA
NA
C2
60krad
1.39
1.5
1.41
1.47
1.41
1.49
1.21
1.56
C3
60krad
1.39
1.49
1.31
1.54
1.38
1.46
1.38
1.48
C4
100krad
2.59
2.63
1.42
1.48
2.57
2.64
1.4
1.55
C5
100krad
1.47
1.51
1.44
1.47
1.45
1.49
1.43
1.51
C6
100krad
1.39
1.53
1.35
1.49
1.38
1.5
1.4
1.53
C7
100krad
1.41
1.48
1.38
1.48
1.35
1.49
1.36
1.55
C8
100krad
1.42
1.52
1.43
1.49
1.42
1.49
1.41
1.49
7
2005 MAPLD, Paper 240
Post-Irradiation Input Threshold
Switching from TTL to CMOS
Table 3 Lot C Post-Irradiation VT
Net 5
DUT
Total Dose
C1
Net 6
Net 8
VIL (V)
VIH (V)
VIL (V)
VIH (V)
VIL (V)
VIH (V)
VIL (V)
VIH (V)
60krad
1.39
1.45
1.38
1.46
1.38
1.45
1.39
1.45
C2
60krad
1.4
1.52
1.39
1.55
NA
NA
1.38
1.44
C3
60krad
1.37
1.47
1.37
1.47
1.32
1.46
1.36
1.45
C4
100krad
1.37
1.49
1.43
1.49
1.41
1.49
2.61
2.66
C5
100krad
1.47
1.51
1.45
1.51
1.42
1.48
1.42
1.48
C6
100krad
1.39
1.49
NA
NA
1.4
1.62
1.51
1.63
C7
100krad
1.38
1.45
1.43
1.51
1.31
1.56
1.4
1.48
C8
100krad
1.44
1.47
1.44
1.48
1.42
1.47
1.43
1.47
DUT
Total Dose
Net 9
JJ Wang
Net 7
Net 10
Net 11
VIL (V)
VIH (V)
VIL (V)
VIH (V)
VIL (V)
VIH (V)
C1
60krad
1.38
1.45
1.38
1.46
1.4
1.45
C2
60krad
1.38
1.45
1.22
1.58
1.41
1.49
C3
60krad
1.34
1.46
1.38
1.48
1.37
1.44
C4
100krad
1.41
1.47
1.38
1.54
1.36
1.46
C5
100krad
1.44
1.51
1.48
1.54
1.45
1.48
C6
100krad
1.37
1.49
1.37
1.57
1.37
1.48
C7
100krad
1.36
1.43
1.13
1.64
1.39
1.49
C8
100krad
1.42
1.49
1.43
1.5
1.42
1.48
8
2005 MAPLD, Paper 240
Annealing Effect Experiment
Five DUT from lot C are irradiated to 100 krad with a lower dose
rate (1 krad/hr)
No switching from TTL to CMOS observed
DUT
Total
Dose
Net 0
Net 1
Net 2
VIL (V)
VIH (V)
VIL (V)
VIH (V)
VIL (V)
Net 3
VIH (V)
VIL (V)
Net 4
VIH (V)
VIL (V)
Net 5
VIH (V)
VIL (V)
VIH (V)
C9
100krad
1.36
1.49
1.37
1.49
1.37
1.48
1.33
1.46
1.40
1.49
1.39
1.48
C10
100krad
1.36
1.47
1.42
1.52
1.40
1.48
1.38
1.49
1.43
1.51
1.41
1.50
C11
100krad
1.36
1.49
1.40
1.51
1.39
1.49
1.38
1.50
1.40
1.50
1.41
1.50
C12
100krad
1.36
1.61
1.39
1.50
1.38
1.48
1.35
1.48
1.43
1.52
1.33
1.51
C13
100krad
1.34
1.53
1.37
1.52
1.36
1.54
1.35
1.50
1.34
1.54
1.25
1.59
DUT
Total
Dose
Net 6
Net 7
Net 8
VIL (V)
VIH (V)
VIL (V)
VIH (V)
VIL (V)
Net 9
VIH (V)
VIL (V)
Net 10
VIH (V)
VIL (V)
Net 11
VIH (V)
VIL (V)
VIH (V)
C9
100krad
1.41
1.51
1.38
1.46
1.37
1.46
1.35
1.46
1.44
1.53
1.34
1.45
C10
100krad
1.41
1.51
1.38
1.47
1.39
1.49
1.37
1.49
1.42
1.52
1.38
1.46
C11
100krad
1.41
1.52
1.39
1.48
1.39
1.48
1.38
1.50
1.41
1.51
1.37
1.46
C12
100krad
1.41
1.52
1.38
1.48
1.20
1.59
1.35
1.53
1.43
1.53
1.38
1.47
C13
100krad
1.35
1.55
1.43
1.52
1.42
1.52
1.36
1.49
1.41
1.51
1.43
1.53
JJ Wang
9
2005 MAPLD, Paper 240
Focus Ion Beam Experiment
The internal node that is suspected being pulled
down by radiation-induced leakage is FIB’ed for
microprobing
However, the heat generated during the FIB
process annealed the device and hence
recovered the TTL input threshold from CMOS
JJ Wang
10
2005 MAPLD, Paper 240
Fabrication Process Dependence
Foundry X show TTL to CMOS switching in 2 out
of 5 lots, more recent lots show no switching
Foundry Y doesn’t show TTL to CMOS switching
in 3 lots
Variable material characteristics of the
commercial foundry FOX (field oxide) determine
the TID tolerance of this phenomenon
JJ Wang
11
2005 MAPLD, Paper 240
Configurable Input
As shown in Figure below, a popular way to vary the input
threshold is to change the strength of the pull-down by
changing the turn-on number of NMOSFET pull-downs
TTL (1.5V trip point) has more turn-on NMOSFET pulldowns than CMOS (2.5V trip point)
PAD
To core logic
Configuration
Control
Fig 3 showing the simplified schematic of configurable input
JJ Wang
12
2005 MAPLD, Paper 240
Failure Mechanism
For testing the programmable switch, node X is holding
high by a single weak pull-up for TTL configuration
Radiation induced leakage in the NMOS pull-down device
pulls node X down (after certain total dose) and switches
the input configuration from TTL (trip point ~1.5V) to CMOS
(trip point ~2.5V)
PAD
To core logic
Weak pull up
Vref
Node X
Test Control
JJ Wang
Radiation-induced
leakage
Programmable
Switch
13
2005 MAPLD, Paper 240
Physical Mechanism
The charge generation, transport and trapping in a biased
oxide layer. The primary effect in sub-micron device is the hole
trapping near the Si/SiO2 interface.
JJ Wang
14
2005 MAPLD, Paper 240
Physical Mechanism
JJ Wang
15
2005 MAPLD, Paper 240
Physical Mechanism
Total dose induced edge and field leakage
JJ Wang
16
2005 MAPLD, Paper 240
Lesson Learned and TID Hardening
Accelerated testing overestimates the effects caused by
radiation-induced field leakages
Commercial foundries have variable FOX characteristics
Weak pull-up is a weak spot for total dose effect
Commercial design often is not perfectly radiation optimized
due to time to market pressure
Two design options
1. Redesign the logic so there is no weak pull-up
2. Re-layout the leaky NMOSFET to “edgeless” (shown below)
Gate
Drain
Source
JJ Wang
17
2005 MAPLD, Paper 240