Benchmarking Charge Trapping Models with NBTI, TDDS ...in4.iue.tuwien.ac.at/pdfs/sispad2020/SISPAD2020_6-3.pdfSharang Bhagdikar and Souvik Mahapatra* Department of Electrical Engineering

Benchmarking Charge Trapping Models with NBTI,

TDDS and RTN Experiments

Sharang Bhagdikar and Souvik Mahapatra*

Department of Electrical Engineering

Indian Institute of Technology Bombay

Mumbai 400076, India

*Phone: +91-222-572-0408, Email: [email protected]

Abstract-- A systematic review and comparison of

existing charge trapping models in literature is performed.

A framework for simulating hole trapping/de-trapping

kinetics is established to compute resultant threshold

voltage degradation (∆VHT) and capture-emission time constants (𝝉𝑪 − 𝝉𝑬 ). The models are analyzed by usingdata from Negative Bias Temperature Instability (NBTI),

Random Telegraph Noise (RTN) and Time Dependent

Defect Spectroscopy (TDDS) experiments.

Keywords—NBTI, RTN, TDDS.

I. INTRODUCTION Charge trapping and detrapping in MOS dielectric

contributes to NBTI, RTN, stress-induced leakage currents

(SILC) and time dependent dielectric breakdown (TDDB) [1]-

[5]. It is established that charge (hole in pFETs) trapping

constitutes the overall threshold voltage (∆𝑉𝑡ℎ ) degradationalong with the interface trap (∆𝑉𝑖𝑡 ) and bulk trap (∆𝑉𝑜𝑡 )generation. Although numerous models have been proposed in

an attempt at modelling the charge trapping time kinetics, the

exact physical mechanism governing the process stays

uncertain [6][7]. The four-state Extended Nonradiative

Multiphonon Model (eNMP) [8] is touted to provide the most

complete description of hole trapping kinetics. Capture and

emission time constants of individual defects are modelled

using eNMP [2]. However, the large number of tuning

parameters make eNMP intractable and limit its practical use.

The Two Well Nonradiative Multiphonon Model (2WNMP) is

an abstraction of the eNMP model that treats the neutral and

charged states of a trap as two energy levels represented as

intersecting parabolic potential wells [8]. The macroscopic

implementation of 2WNMP is used to model hole trapping in

pre-existing defects in large area devices [7][9], and the

stochastic implementation is used for NBTI and TDDS kinetics

in small area devices [10]. The double well thermionic (DWT)

model represents the distinct defect states as energy levels

separated by a thermionic barrier. The original model [11] is

altered in [12] by introducing a temperature activated barrier

for modelling BTI kinetics over a range of temperatures. This

Activated Barrier Double Well Thermionic (ABDWT) model

[12] is invoked to model NBTI stress-recovery transients over

a range of biases and temperatures and across different

technologies [13]. The bias and temperature couplings of

capture (C) and emission (E) time constants measured from

TDDS and RTN studies is modelled using ABDWT [14].

II. MODEL FRAMEWORK & SIMULATION SETUPFig.1 depicts a schematic of the NMP model. The neutral

(E1) and charged (E2) defect energy levels are approximated

as quadratic potential wells. Level E2 is pinned to the energy

of the reservoir which supplies the carriers i.e. either the

substrate conduction band edge (for electrons) or the valence

band edge (for holes). The point where the wells intersect

provide the barrier heights 𝜖12 and 𝜖21 for hole capture (𝑘12)and emission (𝑘21) rates respectively. The bias dependencegets accounted in the fact that upon application of a gate

voltage, level E1 undergoes an electrostatic level shift relative

to E2 (substrate) which would revise the barrier heights and,

hence, the reaction rates. The required energy for the

transition is supplied/dissipated entirely via phonons. The

expressions for the reaction rates are rigorously derived in [8]

and are listed in Fig.1.

The ABDWT model provides transition rates for charge (hole

in p-FET) capture and emission within a trap, Fig.2. A

transition from a reference neutral state (E1) to the charged

state (E2) via a thermally activated barrier (EB) constitutes the

hole capture reaction. A backward transition signals the hole

emission reaction. The barrier EB and state E2 lowers when a

gate bias (VG) is applied to account for the bias dependence on

the reaction rates [13]. The parameters for bias-dependent

barrier lowering and thermal lowering are distinct and allows

for decoupling of bias and T dependence on rate constants.

Defects are distributed uniformly spatially in the dielectric for

performing macroscopic simulations using 2WNMP.

Appropriate model parameters (mean + spread) are assigned

to be consistent with experimental data. Only the defects (E1)

that transition above or below the Fermi level upon application

of bias may take part in the capture-emission reaction and are

otherwise assumed to remain in equilibrium . The setup for the

ABDWT model is similar except that the defects are situated

at the dielectric-substrate interface owing to the fact that

spatial dependence is implicitly captured by its parameters.

For replication of defect-centric data (RTN, TDDS), an

individual defect is placed in the dielectric and assigned

unique model parameters to generate 𝑘12 and 𝑘21, which yield𝜏𝐶 and 𝜏𝐸 respectively.

III. NBTI MODELLING Fig.3-4 shows experimental VHT stress and recovery data

from Gate First HKMG planar MOSFETs [15]. The hole

trapping component VHT is isolated from the measured mean

VTH data using the macroscopic BAT framework [1], which

uses an empirical relation to compute VHT. The VHT is thus

extracted over a range of temperatures and biases using the

6-3

© 2020 The Japan Society of Applied Physics117

BAT framework. The ABDWT model simulations are shown

to map the T activation over the entire range (Fig.5). ABDWT

is shown to accurately model VHT stress data over a range of

VGSTR (Fig.6). Appropriate parameters for 2WNMP are

selected to reproduce the stress data over the entire

temperature range in Fig.7. It is observed that when the same

parameters are used to model VHT stress data over a range of

VGSTR, 2WNMP predicts a stronger bias (VGSTR) activation and

the dataset cannot be matched (Fig.8). In Fig.9, suitable

2WNMP parameters are chosen to model the bias activation.

The same parameters predict a weaker T activation and cannot

map the entire T dataset (Fig.10). Similar analysis is carried

out for VHT extracted from RMG HKMG SOI FinFETs [16].

ABDWT is shown to model ∆𝑉𝐻𝑇 data in Figs.11-12 whereassimultaneous realization of bias and T activation cannot be

achieved using 2WNMP, Figs.13-14. Comparison of VHT

recovery curves over different VGREC (Figs.15-16) and

different stress times (Figs.17-18) is performed. It is observed

that, for concurrent stress curves, the 2WNMP model predicts

consistently slower recovery than ABDWT.

IV. RTN AND TDDS VALIDATION Fig.19(a)-(d) list the distinct types of bias couplings (types

A-D) for capture and emission time constants obtained from

RTN studies [17]. All the different VG couplings can be

reproduced by the ABDWT model upon selection of suitable

parameters. The type-A and type-B bias couplings are

reproduced by 2WNMP. The presence of a bias-dependent

pre-factor (p) in the 2WNMP capture rate expression prevents

C from being bias agnostic. In Fig.19(c), 2WNMP predicts a

weak negative coupling of C as opposed to the zero coupling

observed. A negative coupling of E, as observed in Fig.19(d),

is realized in the weak electron-phonon coupling regime [8] of

2WNMP. C in this regime is also negatively coupled with bias

and it is not possible to achieve type-D coupling using

2WNMP. The dependence of 𝜏𝐶 and 𝜏𝐸 on 𝑉𝐺 at differenttemperatures is recorded from RTN experiments in [18] and

reproduced in Fig. 20. In Fig.21, ABDWT is shown to capture

similar temperature activation trends across all 𝑉𝐺, owing tothe fact that the 𝜏𝐶 and 𝜏𝐸 are not strongly coupled and can betuned independently unlike in 2WNMP.

Fig.22 illustrates the bias dependence of 𝜏C and 𝜏E acquiredfrom TDDS measurements for a non-switching trap [2]. The

bias dependence is reproduced by ABDWT across two

different T using appropriate parameters. The time constants

for the non-switching trap are also modelled using 2WNMP.

Figs.23-24 show TDDS time constants for switching trap with

weak [2] and strong [19] bias activation of 𝜏𝐸 in thesubthreshold region respectively. 2WNMP cannot replicate

the switching behavior i.e. positive coupling of 𝜏E belowthreshold voltage.

V. CONCLUSION Macroscopic frameworks of ABDWT and 2WNMP are

used to model experimental NBTI data over a range of stress

biases, recovery biases and temperatures. The 2WNMP model

in its present form cannot predict T and VGSTR activation as

ABDWT, and is unable to model extended stress dataset.

2WNMP predicts slower VHT recovery than ABDWT.

2WNMP cannot reproduce the available capture and emission

time constant bias couplings observed in RTN experiments.

Switching trap time constants obtained using TDDS are not

modelled by 2WNMP. 𝜏𝐸 and 𝜏𝐶 exhibit weaker correlationcoming from ABDWT as compared to 2WNMP, which makes

the former more versatile in modelling defect-centric data.

REFERENCES [1] N. Parihar, N. Goel, S. Mukhopadhyay and S. Mahapatra, "BTI Analysis

Tool-Modeling of NBTI DC, AC Stress and Recovery Time Kinetics,

Nitrogen Impact, and EOL Estimation," in IEEE Trans. Electron

Devices, vol. 65, no. 2, pp. 392-403, Feb.2018. [2] T. Grasser, et al, “On the Microscopic Origin of the Frequency

Dependence of Hole Trapping in pMOSFETs”, IEDM, Dec. 2012, pp.

19.6.4. [3] A. Kerber, A. Vayshenker, D. Lipp, T. Nigam and E. Cartier, "Impact of

charge trapping on the voltage acceleration of TDDB in metal gate/high-

k n-channel MOSFETs," 2010 IEEE International Reliability Physics Symposium, Anaheim, CA, 2010, pp. 369-372.

[4] W. Goes, M. Waltl, Y. Wimmer, G. Rzepa and T. Grasser, "Advanced

modeling of charge trapping: RTN, 1/f noise, SILC, and BTI," 2014SISPAD, Yokohama, 2014, pp. 77-80.

[5] N. Tega et al., "Increasing threshold voltage variation due to random

telegraph noise in FETs as gate lengths scale to 20 nm," 2009 Symposium on VLSI Technology, Honolulu, HI, 2009, pp. 50-51.

[6] S. Mahapatra and Narendra Parihar, “A review of NBTI mechanisms

and models,” Microelectron. Reliab, Volume 81, 2018, Pages 127-135. [7] Rzepa, G. et al. “Comphy - A compact-physics framework for unified

modeling of BTI.” Microelectron. Reliab. 85 (2018): 49-65.

[8] Grasser, T.. “Stochastic charge trapping in oxides: From random telegraph noise to bias temperature instabilities.” Microelectron.

Reliab. 52 (2012): 39-70.

[9] G. Rzepa et al., "Efficient physical defect model applied to PBTI in high-κ stacks," 2017 IEEE International Reliability Physics Symposium

(IRPS), Monterey, CA, 2017, pp. XT-11.1-XT-11.6.

[10] Anandkrishnan R, et al, “A Stochastic Modeling Framework for NBTI and TDDS in Small Area p-MOSFETs”, Simulation of Semiconductor

Processes and Devices (SISPAD) , Austin TX, Sep. 2018, pp. 181.

[11] D. Ielmini, et al, "A unified model for permanent and recoverable NBTI based on hole trapping and structure relaxation," 2009 IEEE

International Reliability Physics Symposium, Montreal, QC, 2009, pp.

26-32.

[12] S. Desai, et al, "A comprehensive AC / DC NBTI model: Stress,

recovery, frequency, duty cycle and process dependence," 2013 IEEE

International Reliability Physics Symposium (IRPS), Anaheim, CA, 2013, pp. XT.2.1-XT.2.11.

[13] N. Choudhury, N. Parihar, N. Goel, A. Thirunavukkarasu and S.

Mahapatra, "A Model for Hole Trapping-Detrapping Kinetics During NBTI in p-Channel FETs: (Invited paper)," 2020 4th IEEE Electron

Devices Technology & Manufacturing Conference (EDTM), Penang,

Malaysia, 2020, pp. 1-4. [14] S. Bhagdikar and S. Mahapatra, "A Stochastic Hole Trapping-

Detrapping Framework for NBTI, TDDS and RTN," SISPAD, Udine,

Italy, 2019, pp. 1-4. [15] N. Parihar, R. Anandkrishnan, A. Chaudhary and S. Mahapatra, "A

Comparative Analysis of NBTI Variability and TDDS in GF HKMG Planar p-MOSFETs and RMG HKMG p-FinFETs," in IEEE

Transactions on Electron Devices, vol. 66, no. 8, pp. 3273-3278, Aug.

2019. [16] N. Parihar, U. Sharma, R. G. Southwick, M. Wang, J. H. Stathis and S.

Mahapatra, "Ultrafast Measurements and Physical Modeling of NBTI

Stress and Recovery in RMG FinFETs Under Diverse DC–AC Experimental Conditions," in IEEE Transactions on Electron Devices,

vol. 65, no. 1, pp. 23-30, Jan. 2018.

[17] H. Miki, et al, “Understanding short-term BTI behavior through comprehensive observation of gate-voltage dependence of RTN in

highly scaled high-κ / metal-gate pFETs”, Symposium on VLSI

Technology - Digest of Technical Papers, June 2011, pp. 149. [18] H. Miki, et al, “Voltage and temperature dependence of random

telegraph noise in highly scaled HKMG ETSOI nFETs and its impact on

logic delay uncertainty”, in Symposium on VLSI Technology (VLSIT),pp. 138, June 2012.

[19] T. Grasser et al., "Advanced characterization of oxide traps: The

dynamic time-dependent defect spectroscopy," 2013 IEEE International Reliability Physics Symposium (IRPS), Anaheim, CA, 2013, pp. 2D.2.1-

2D.2.7.

118

Fig.1. Schematic of 2WNMP model depicting

potential wells of the neutral (𝑞1) and charged (𝑞2) states.

Fig.2. Schematic of ABDWT model. 𝐸1, 𝐸2 and 𝐸𝑏determine the energetic configuration of the trap.

Fig.5. Modelling of ∆𝑉𝐻𝑇 stress data (symbols) using ABDWT (solid lines) over a range of temperatures.

GF HKMG planar MOSFETs

Fig.6. Modelling of ∆𝑉𝐻𝑇 stress data (symbols) using ABDWT (solid lines) for a range of stress

biases.

Fig.10. Modelling of ∆𝑉𝐻𝑇 stress data using 2WNMP parameters of Fig.9.

2WNMP predicts weaker T activation.

Fig.8. Modelling of ∆𝑉𝐻𝑇 stress data using 2WNMP parameters of Fig.7.

2WNMP predicts stronger bias

activation.

Fig.9. Modelling of ∆𝑉𝐻𝑇 stress data (symbols) using 2WNMP (solid lines)

for a range of biases.


over a range of temperatures.

10-7 10-5 10-3 10-1 101 1030.1

1

10

100

V

HT (

mV

)

Time (s)

VSTR = -1.2 V VSTR = -1.45 V

VSTR = -1.7 V VSTR = -1.9 V

2WNMP

10-7 10-5 10-3 10-1 101 103

0.1

1

10

100

V

HT (

mV

)

Time (s)

-40°C 0°C 25°C

65°C 100°C 125°C

2WNMP

10-7 10-5 10-3 10-1 101 1030.1

1

10

100

V

HT (

mV

)

Time (s)

VSTR = -1.2 V VSTR = -1.45 V

VSTR = -1.7 V VSTR = -1.9 V

2WNMP

10-7 10-5 10-3 10-1 101 1030.01

0.1

1

10

100

V

HT (

mV

)

Time (s)

-40°C 0°C 25°C

65°C 100°C 125°C

2WNMP

10-5

10-4

10-3

10-2

10-1

100

101

102

103

104

10-3

10-2

10-1

VGSTR

/VMEAS

= -1.45/-0.4

T=1250C

W=90nm x 70nm

VT> V

T

VHT

VIT

V

T (

V)

Time (s)

10-7 10-5 10-3 10-1 101 103

0.1

1

10

100

V

HT (

mV

)

Time (s)

-40°C 0°C 25°C

65°C 100°C 125°C

ABDWT

10-7 10-5 10-3 10-1 101 103

1

10

100 VSTR = 1.2 V VSTR = 1.45 V

VSTR = 1.7 V VSTR = 1.9 V

V

HT (

mV

)

Time (s)

T = 125°C

ABDWT

10-6

10-5

10-4

10-3

10-2

10-1

100

101

102

103

104

0.00

0.02

0.04

0.06

0.08

0.10

0.12

V

T (

V)

Time (s)

VSTR

/VMEAS

= -1.45/-0.4

tSTR

=1Ks

T=1250C

W=90nm x 70nm

VT> V

T

VHT

VIT

10-7 10-5 10-3 10-1 101 103

0.1

1

10

100

10-7 10-5 10-3 10-1 101 103

0.1

1

10

100

V

HT (

mV

)

Time (s)

-40°C 0°C 25°C

65°C 100°C 125°C

V

HT (

mV

)

Time (s)

10-7 10-5 10-3 10-1 101 103

1

10

V

HT (

mV

)

Time (s)

1.3 V 1.4 V 1.5 V 1.6 V 1.7 V

10-7 10-5 10-3 10-1 101 103

1

10

V

HT (

mV

)

Time (s)

1.3 V 1.4 V 1.5 V 1.6 V 1.7 V

10-7 10-5 10-3 10-1 101 103

0.1

1

10

10-7 10-5 10-3 10-1 101 103

0.1

1

10

V

HT (

mV

)

Time (s)

-40°C 0°C 25°C

65°C 100°C 125°C

V

HT (

mV

)

Time (s)

Fig.3. Individual (gray) and mean (black) measured

∆𝑉𝑇 traces during stress along with model calculatedmean (red) and decomposition into subcomponents.

Device is GF HKMG planar MOSFET.

Fig.4. Individual (gray) and mean (black) measured ∆𝑉𝑇traces during recovery along with model calculated

mean (red) and decomposition into subcomponents

Fig.11. Modelling of ∆𝑉𝐻𝑇 stress data (symbols) using ABDWT (solid lines)

over a range of temperatures.

Measurements are from RMG HKMG

SOI FinFET.

Fig.12. Modelling of ∆𝑉𝐻𝑇 stress data (symbols) using ABDWT (solid lines)

over a range of stress biases.


for a range of biases.

Fig.14. Modelling of ∆𝑉𝐻𝑇 stress data using 2WNMP parameters used in

Fig.13. 2WNMP predicts much weaker T

activation.

119

𝝉𝑒

𝝉𝒄

Fig.16. Modelling of ∆𝑉𝐻𝑇 recovery data (symbols) using 2WNMP (solid

lines) for a range of recovery biases.

Fig.17. Modelling of ∆𝑉𝐻𝑇 recovery data (symbols) using ABDWT (solid

lines) for a range of stress times.

Fig.18. Modelling of ∆𝑉𝐻𝑇 recovery data (symbols) using 2WNMP (solid lines)

for a range of recovery biases.

-

- - - -

Fig.15. Modelling of ∆𝑉𝐻𝑇 recovery data (symbols) using ABDWT (solid

lines) for a range of recovery biases..

0.88 0.96 1.04 1.120.1

1

10

tE tC

Tim

e c

onsta

nt (s

)

-Vg (V)

0.80 0.84 0.88 0.921

10

100

te tc

Tim

e c

onsta

nt (s

)

-Vg (V)0.88 0.96 1.04 1.12

0.01

1

100

te tc

Tim

e c

onsta

nt (s

)

Vg (V)0.60 0.65 0.70

1

10

te tc

Tim

e c

onsta

nt (s

)

-Vg (V)

10-7 10-5 10-3 10-1 101 103

0

10

20

30

V

HT (

mV

)

Time (s)

VREC= -0.4 V

VREC= -0.8 V

VREC= -1.2 V

2WNMP

10-7 10-5 10-3 10-1 101 103

0

10

20

30

V

HT (

mV

)

Time (s)

tSTR=100ms

tSTR=1s

tSTR=10s

tSTR=100s

2WNMP

10-7 10-5 10-3 10-1 101 103

0

10

20

30

V

HT (

mV

)

Time (s)

tSTR=100ms

tSTR=1s

tSTR=10s

tSTR=100s

VSTR = -1.45 V, VREC = 0 V

T = 125°C

ABDWT

10-7 10-5 10-3 10-1 101 103

0

10

20

30

V

HT (

mV

)

Time (s)

VREC= -0.4 V

VREC= -0.8 V

VREC= -1.2 V

VSTR = -1.45 V, tSTR = 1000s

T = 125°C

ABDWT

Fig.19(a)-(d). Panels depict various couplings of time constants to VG extracted from RTN data (symbols) [12]. (A) 𝜏𝑐 < 0, 𝜏𝑒~0, (B) 𝜏𝑐 < 0, 𝜏𝑒 > 0, (C) 𝜏𝑐~0, 𝜏𝑒~0, (D) 𝜏𝑐~0, 𝜏𝑒 <

0. All the different couplings are reproduced using ABDWT model simulations (solid lines). 2WNMP simulations (dashed lines) can model type A and type B coupling upon selection

of appropriate parameters. Zero coupling of 𝜏𝐶 with bias cannot be reproduced using 2WNMP and a negative coupling is observed in (c) and in (d) where 𝜏𝑐 is not shown (out of

bounds, 𝜏𝐶 ≫ 𝜏𝐸).

Fig.22. Modelling of TDDS capture and emission time constants (symbols) for non-

switching trap [2] using ABDWT (left) and 2WNMP (right). Switching behavior is

evident by the positive bias coupling of 𝜏𝑒 with 𝑉𝐺 .2WNMP in its present form isunable to replicate said coupling.

Fig.21. Modelling of TDDS capture and emission time constants (symbols) for

non-switching trap [2] using ABDWT (left) and 2WNMP (right). Non-switching

nature is evident by the bias agnostic 𝜏𝐸 in the subthreshold regime.

0.0 0.5 1.0 1.5 2.0 2.510-7

10-5

10-3

10-1

101

103 te 125°C

te 175°C

tc 125°C

tc 175°C

Tim

e c

on

sta

nt

(s)

-Vg (V)0.0 0.5 1.0 1.5 2.0 2.5

10-7

10-5

10-3

10-1

101

103 te 125°C

te 175°C

tc 125°C

tc 175°C

Tim

e c

on

sta

nt

(s)

-Vg (V)

0.0 0.5 1.0 1.5 2.0 2.510-7

10-5

10-3

10-1

101

103

Tim

e c

on

sta

nt

(s)

-Vg (V)

te 125°C

te 175°C

tc 125°C

tc 175°C

0.0 0.5 1.0 1.5 2.0 2.510-7

10-5

10-3

10-1

101

103

Tim

e c

on

sta

nt

(s)

-Vg (V)

te 125°C

te 175°C

tc 125°C

tc 175°C

0.5 0.6 0.710-5

10-4

10-3

10-2

10-1

100

Tim

e c

onsta

nts

(s)

VG (V)

0.45 0.50 0.55 0.60 0.65 0.7010-5

10-4

10-3

10-2

10-1

100

tc

Tim

e c

onsta

nt

(s)

Vg (V)

te

-1.0 -0.5 0.0 0.5 1.0 1.5 2.010-7

10-5

10-3

10-1

101

103

Tim

e c

on

sta

nt

(s)

-Vg (V)

te125°C

te175°C

tc125°C

tc175°C

-1.0 -0.5 0.0 0.5 1.0 1.5 2.010-7

10-5

10-3

10-1

101

103 te125°C

te175°C

tc125°C

tc175°C

Tim

e c

on

sta

nt

(s)

-Vg (V)

Fig.23. Modelling of TDDS capture and emission time constants (symbols) for

switching trap B3 [19] using ABDWT (left) and 2WNMP (right).

Fig.20. Time constants as a function of

VG at various temperatures from RTN

experiments [18].

Fig.21. ABDWT model simulated time

constants as a function of VG at various

temperatures.

rap [2] using ABDWT (left) and

2WNMP (right).

120

TitleCopyrightForewordCommitteeCOMMITTEE MEMBERSOrganization:International Steering Committee:Technical Program Committee:

TimeTableProgramAbstracts01 Plenary01-101-201-3

02 Band Structure02-102-202-302-402-502-6

03 Computational Methodology03-103-203-303-403-503-603-7

04 Nanowire04-104-204-304-404-5

05 Material and Geometry Impact05-105-205-305-405-505-6

06 Reliability06-106-206-306-406-5

07 Power and Optoelectronic Devices07-107-207-307-407-507-607-707-8

08 Non-Volatile Memory I08-108-2I. IntroductionII. Simulation MethodIII. Bias Optimization for Read-Out CurrentIV. Impact of Variation on Read-Out CurrentV. ConclusionAcknowledgmentReferences

08-308-4

09 Transport09-1INTRODUCTIONALGORITHMRESULTSCONCLUSIONS

09-209-309-409-5

10 Non-Volatile Memory II10-110-210-310-410-510-6

11 High Speed Switching Devices11-111-211-311-411-511-611-711-8

12 Emerging Devices12-112-212-312-412-512-612-7

13 2D and Nano System I13-113-213-313-413-5

14 FET Devices and Design Technology Co-Optimization14-114-214-314-414-514-614-714-8

15 Machine Learning15-115-215-315-415-515-6

16 2D and Nano System II16-116-216-316-416-515-7

Late News

Author IndexAB

/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 150 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 2.00333 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.00167 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False

/CreateJDFFile false /Description >>> setdistillerparams> setpagedevice























































Documents

Benchmarking Charge Trapping Models with NBTI, TDDS ...in4.iue.tuwien.ac.at/pdfs/sispad2020/SISPAD2020_6-3.pdfSharang Bhagdikar and Souvik Mahapatra* Department of Electrical Engineering