Characterization and Estimation of Circuit Reliability Degradation under NBTI using OnLine IDDQ Meas - PowerPoint PPT Presentation

Loading...

PPT – Characterization and Estimation of Circuit Reliability Degradation under NBTI using OnLine IDDQ Meas PowerPoint presentation | free to view - id: 20f7ee-YzI2O



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Characterization and Estimation of Circuit Reliability Degradation under NBTI using OnLine IDDQ Meas

Description:

Characterization and Estimation of Circuit Reliability Degradation ... Microphotograph. Layout. Vin. VDD. IDDQ Measurement. 1000 stages. 11. Measurement Flow ... – PowerPoint PPT presentation

Number of Views:346
Avg rating:3.0/5.0
Slides: 22
Provided by: Quic7
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Characterization and Estimation of Circuit Reliability Degradation under NBTI using OnLine IDDQ Meas


1
Characterization and Estimation of Circuit
Reliability Degradation under NBTI using On-Line
IDDQ Measurement
  • Kunhyuk Kang, Keejong Kim, Ahmad E. Islam,
    Muhammad A. Alam, and Kaushik Roy
  • Electrical and Computer Engineering
  • Purdue University

2
Negative Bias Temperature Instability
VGS lt 0V
A. T. Krishnan et al., IEDM05
H2
H2
H2
H2
H2
GATE OXIDE
VD 0V
H
H
VS 0V
H
H
H
x
x
x
x
x
H
H
DRAIN
P
P
Si
Si
Si
Si
n-sub
VBODY 0V
  • PMOS Reliability Issue
  • Reaction-Diffusion (RD) model
  • Fractional Power Law
  • Time exponent n 1/6

?Vt(t) KDC X f(Sp) X t1/6
M. A. Alam., IEDM03
3
Motivation
  • Negative Bias Temperature Instability
  • Challenging Device/Circuit Reliability Issue
  • Degrades end-of-lifetime IC Performance
  • Scaling Impact ? Tox?, N concentration ?, Temp. ?
  • Characterization of NBTI
  • Formal approaches Analytical Tools, fMAX based
    char.
  • of TRs with multiple stress conditions
  • Challenging problem to solve in an analytical
    domain
  • Post-silicon NBTI characterization method based
    on leakage (IDDQ) measurement

Y-H Lee et al., IEDM03, A. T. Krishnan et al.,
IEDM03
4
Outline
  • Introduction
  • NBTI Characterization based on IDDQ Measures
  • Analytical framework
  • Experimental verification measured results
  • Circuit Applications
  • Logic Circuits ? fMAX prediction
  • SRAM Memory Arrays ? SNM prediction
  • Conclusion

5
Introduction NBTI Characterization
NBTI
Multiple Temp. Islands
Temporal Performance Degradation
-3
x 10
100
2
1.5
NBTI
IDS A
10-5
1
NBTI
0.5
J. D. Warnock et al., IBM02
Multiple VDD Islands
10-10
VDD2
0
-0.3
-1.1
-0.9
-0.7
-0.5
-1.3
-1.1
-0.9
-0.7
-0.5
VGS V
VGS V
VDD1
  • NBTI leads to temporal performance degradation
  • Single device-level models available
  • Circuit-level prediction grows as a challenging
    tasks multiple VDD Temp.

VDD1
VDD3
VDD2
VDD1
6
Proposed Characterization Method
Previous Approaches
Proposed Method
  • Combined efforts from Front and Back-end side
  • _at_Back-End
  • Characterize Circuit-Level NBTI degradation
    using IDDQ
  • _at_Front-End
  • Correlate IDDQ trend with other performance
    parameters
  • fMAX, SNM, etc
  • Front-End Techniques
  • Analytical Basis, e.g. RD
  • Cell / Sub-Ckt Models
  • Limitation in its accuracy
  • Back-End Techniques
  • fMAX measurement
  • Limitations in non-timing-critical applications,
    e.g. SRAM
  • Expensive test-cost

B. C. Paul, EDL05, S. Kumar, ICCAD06
What is the temporal trend of IDDQunder NBTI
degradation?
7
Temporal IDDQ Estimation Cell
VDD
Cell IDDQ
102

NOR2 Leakage Reduction
vectors
?Vtp1 f(S1)
M1
01
10
11
?Vtp2 f(S2)
M2
101
n1/6
in1 S1
VOUT
change
in2 S2
PMOS VTh Increase
BPTM 70nm, 125ºC
100

0
2
4
6
8
10
10
10
10
10
Time (s)
Input Vector Dependent Leakage Component
  • Basic leakage analysis for our proposed method
  • Analytical IDDQ Estimation Tool
  • n1/6 at cell level IDDQ

S. Narendra et al., JSSC04
8
Temporal IDDQ Estimation Circuit
75C, 36 input c432
c432 Random Test Pattern Ave.
2
10

102
TEST 1
TEST 2
TEST 3
TEST 4
1
10
TEST 5
n1/6
IDDQ reduction ()
IDDQ reduction ()
101
0
10

n1/6
25
C

75
C

125
C
-1
100
10


0
2
4
6
8
0
2
4
6
8
10
10
10
10
10
10
10
10
10
10
Time (s)
Time (s)
  • Circuit level IDDQ estimation tool VDD temp.
    dependence
  • Fixed Pattern reduction ? n1/6
  • Random Pattern average ? n1/6
  • ? IDDQ degradation trend can be used as an NBTI
    signature

9
Why n1/6 for IDDQ?
Linear dependency term
  • Rleak leakage degradation
  • TR-level relationship can be extended to explain
    n1/6in the circuit level
  • If ?Vt grows comparable to mvT, square term needs
    tobe considered

10
Experimental Verification
Layout
Microphotograph
Inverter Chain
VDD
IDDQ Measurement
Vin
1000 stages
  • Test Circuit Fabricated
  • 1000 stage INV chain
  • DC Stress signal _at_Vin
  • IDDQ measurement _at_GND
  • Controlled Stress Condition to avoid multiple
    reliability degradation sources

11
Measurement Flow
0. Initial (t0) measurement
NBTI
NBTI
1. Stress Condition VDD Vin Vstress Temp
Tstress
1
0
0
1
Measurement delay DM
State Flipping
2. NBTI Stress Applied
NBTI
3. Measure Condition VDD 1.2V, Vin 0V Temp
Tstress
0
1
1
0
Status Flipping betweenStress Measure condition
12
Measurement Results
10
Vstress 1.7V _at_150C
8
101
n gt 1/6
6
S2
n 1/6
IDDQ (Vin0.0) degradation
4
IDDQ (Vin1.7) degradation
S1
2
MAX MIN lt 1.2
0
100
-2

100
101
102
103
104
105
Time (s)
  • DM lt 3ms, Temp125C, Vstress1.7V
  • n1/6 during measurement mode ? IDDQ degradation
    verified
  • Negligible change during stress mode
  • Clear signature of NBTI

13
Measurement Results Cont.
101
101
n 1/6
Vstress 1.7V
n 1/6
Tstress 150C
S1_at_Vstress 1.7V
S1_at_Tstress 150C
IDDQ degradation
IDDQ degradation
100
Vstress
Tstress
S1_at_Vstress 1.5V
1.7V
1.5V
S1_at_Vstress 100C
S1_at_Vstress 1.3V
1.3V
100

0
1
2
3
0
1
2
3
10
10
10
10
10
10
10
10
Time (s)
Time (s)
  • VDD dependency _at_150C, Field Accel. Factor
    1dec/MV/cm
  • Temp. dependency _at_1.7V, Activation Energy 0.7eV
  • n1/6 for all Temp. VDD corners

14
Application of IDDQ Characterization
IDDQ Meas. ? Rleak
Other Factors
IDDQ ? fMAX
IDDQ ? SNM
fMAX Degradation Estimation
SNM Degradation Estimation
  • Transformation of IDDQ characterization constant
    Rleak to other performance factors

15
NBTI Characterization using IDDQ Logic
102
c2670
Linear relationship
c5315
107
c1908
c499
105
c3540
c74181
IDDQ reduction ()
103
101
10
75C, 65nm PTM ISCAS Benchmarks
100

0
10
fMAX reduction ()
IDDQ decrease fMAX increase
n1/6
  • fMAX IDDQ ? n1/6
  • fMAX and IDDQ degradation has linear dependency ?
    K factor
  • K factor Temperature and VDD dependent

16
Simulation Results Logic Circuits
  • 10 ISCAS Benchmark Circuits _at_ BPTM 70nm
  • K3.4 _at_75C, K4.1 _at_125C, VDD1V

17
NBTI in Memory Arrays
WL
?VThL
?VThR
BLB
BL
PDF
RDF
ML
MR
?VTh
Temporal NBTI shift
  • Mismatch between TRs are critical in SRAM cell
  • Two sources of mismatches
  • Spatial Source Process Variation, RDF
  • Temporal Source NBTI in PMOS TR
  • SNM, read write stability, parametric yield is
    affected by NBTI degradation

18
NBTI Characterization using IDDQ SRAM
10
108
1
PTM 65nm, 125C 1MB SRAM
10
PTM 65nm, _at_125C
107
106
6
1/6 trend line
105
Degradation in Leakage
Degradation in SNM ()
104
103
2
0
10
6
2
10
4
6
8
10
10
10
Degradation in SNM
Time (s)
  • ?SNM IDDQ ? n1/6
  • ?SNM and IDDQ degradation has linear dependency ?
    K factor
  • K factor Temperature and VDD dependent

19
IDDQ based Characterization Technique
Design phase
Initial Characterization
  • Circuit-level NBTI Reliability Characterization
  • IDDQ test is used
  • Expensive fMAX testing is avoided (or minimized)
  • Accurate circuit level performance degradation
    can be predicted
  • IC specific burn-in to qualify the target produce
  • Efficient way of field monitoring dynamic local
    signature of produce usage
  • Possible usage in other reliability sources HCI

Compute Riddq, RSNM, RfMAX


Compute KfMAX and KSNM
Reliability
Extrapolation

Measure and sample Rleak
Post-silicon phase
Extrapolate fMAX and SNM using K factor


Estimate degradation
Lifetime Projection

Project fMAX and SNM
Temp
.
Dependency
NBTI Characterization Report
20
Conclusions
  • IDDQ degradation ? n1/6
  • This specific trend of IDDQ degradation can be
    used as a signature of NBTI
  • Verification through a test chip fabricated _at_
    130nm bulk CMOS
  • An efficient NBTI characterization based on IDDQ
    measurement established
  • Proposed method can reduce fMAX test overhead,
    and provide an estimate of lifetime performance
    degradation in IC

21
Decoupling Multiple Reliability Degradation
Sources
  • Different temporal dependency
  • NBTI?1/6, HCI?1/22/3, TDDB?n
  • Multiple VDD and temperature conditions can
    de-convolute the original NBTI-induced IDDQ change
About PowerShow.com