drain current in silicon-based bulk and utbb- fdsoi ... · the channel only: quantum resistance 7 2...
TRANSCRIPT
Drain current in Silicon-based Bulk and UTBB-FDSOI devices: what are the key ingredients?
D. Rideau1, F. Monsieur1, O. Nier1,2, Y. M. Niquet3, V. Quenette1, G. Mugny1,5, G. Gouget1, M. Quoirin1, and R. Gonella1
1) STMicroelectronics, 850, rue J. Monnet, BP. 16, 38921 Crolles, France; 2) IMEP-LAHC, MINATEC 3 Parvis Louis Néel, 38016 Grenoble; 3) SP2M, UMR-E CEA/UJF-Grenoble 1, INAC, Grenoble, France
Outlook
• What drives the Linear current in ultra-short MOSFETs?
• NEGF vs Semi classical TCAD • A focus on the Near-LDD region • the link bw Access Resistance and the ‘Apparent’ Channel
Mobility degradation.
• Investigation of the saturation regime • More NEGF simulations • Diffusive Model or Virtual Source Model? • Back to measurements
2
What drives the current reduction in Short Devices? 3
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.610-10
10-9
10-8
10-7
10-6
10-5
10-4
NORM
ALIZ
ED C
URRE
NT (A
)
GATE VOLTAGE (V)
Long device
Short device
• Mobility Degradation • Velocity Saturation • Injection Velocity • Pinch Off
è Focus on Silicon-based Field effect transistors curently in production (2016)
High VDS
Low VDS
• Quantum Transport vs Classical one?
Focus on: è Linear Channel Resistance
What drives the Linear current in ultra-short MOSFETs? 4 Long Channel Length
Short Channel Length (L~20nm)
Total Device Resistance is dominated by the ‘Long channel’ Channel mobility
The access Resistances reduce the current
Investigated: • Do we have additional
mechanisms in the channel? Focus on: è Planar MOSFETs with
homogeneous channel (e.g. no pocket)
Total Device Resistance in the linear regime:
As e.g. [M. Zilli, et al., Electron Device Lett. 28, 1036 (2007).]
IVR D
ON =
Channel Resistance 2: L.(µ0
.COX )-1.(VG-VTH)-1 Access Resistance
Undoped Channel
SOI BOX
Spacer
2
HDD
1
1: R1
L
10 )( RrLRon +⋅= µ?
What drives the current in ultra-short MOSFETs? 5
DDTHGOX VVVVCLI )2/(0 −−⋅⋅=⋅ µDDTHG
DTHGOX
OX VVVVVVV
LCR
CLI )2/())2/()2(1( 00
0 −−−−⋅
⋅⋅⋅+
⋅=⋅
µµ
R0
Impact of R0 only? … or
Measured @ VD=50mV
0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
2
4
6
8
10
12
14
16x 10-6
NORM
ALIZ
ED C
URRE
NT (A
): I*L
/W
GATE VOLTAGE (V)
LCH=0.9 µmLCH=1.8 µm
LCH=0.45 µm
LCH=0.15 µm
LCH=50 nm
LCH=20 nm
µ0(L) due to additional mecanisms?
Access resistance:
Channel Mobility: L-dependent Mobility degradation
Linear Channel Current:
Capacitance
Outlook
• What drives the Linear current in ultra-short MOSFETs?
• NEGF vs Semi classical TCAD • A focus on the Near-LDD region • the link bw Access Resistance and the ‘Apparent’ Channel
Mobility degradation.
• Investigation of the saturation regime • More NEGF simulations • Diffusive Model or Virtual Source Model? • Back to measurements
6
The Channel Only: Quantum Resistance 7
2 4 6 8 10 12
x 1012
0
50
100
150
200
250
QUA
NTUM
RES
ISTA
NCE
[ Ω . µ
m]
INVERSION CHARGE [/cm2]
NEGF: (2nm<TSi<10nm)
Eq. 1
Quantum resistance calculated with NEGF for various channel thicknesses.
Resistance calculated with NEGF in an ideal channel-device.
Calculations TB_SIM
))(/(kTm2 THGQ VVCoxqR −⋅⋅⋅⋅≈ π
è NMOS: σΒ = 8.4 Ω.mm.V.
è 1)2( −⋅⋅= OXBB Cσµ of 23.8 cm2/(V.s.nm)
Mobility
{ }BTHGTHGOX
on VVVVCLR σ
µ⋅⋅
−+
⎭⎬⎫
⎩⎨⎧
−⋅= 2
)(1
)(1
0
Mobility with NEGF in O. Nier, et al., Journal of Computational Electronics 12.4, 675 (2013).
8 Simulation setups
http://www.synopsys.com/Tools/TCAD/ http://inac.cea.fr/L_Sim/TB_Sim
TB_SIM: - self consistent NEGF - Local + remote coulomb - Band structure effects
Local Coulomb Remote Coulomb
PH + SR
Realistic Doping Profile from Process simulation
SYNOPSYS SDEVICE: - Hydrodynamics with effective mobility - Density gradient quantum correction
Varying channel length
Focus on: è Planar MOSFETs with
homogeneous channel (e.g. no pocket)
Access Region: NEGF vs Semi classical 9
0 0.05 0.1 0.15 0.2 0.2550
100
150
200
250
300
Ron Ω
. µ
m
L (µ m.)
VD 0.05V: Mean Error = 2.5e-014 %VD 0.05V: Mean Error = 0.17 %VD 0.05V: Mean Error = 0.22 %
All scatterings (EMA): µ0 =0.04
PHONON Only: µ0 =0.053
VD 0.05V: σ = 31 µ0 =0.028 Cox =0.025
VD 0.05V: σ = 31 µ0 =0.028 Cox =0.025
VD 0.05V: σ = 31 µ0 =0.028 Cox =0.025
All scatterings (kp2): µ0=0.028
VGT=0.8V
Ron = 126 (Ω.µm) = 46 + 40 * 2 (Ω.µm)
{ } 00
2)(2)(
1)(
1 RVVVVC
LR BLDDTHGTHGOX
on ⋅++⋅⋅−
+⎭⎬⎫
⎩⎨⎧
−⋅= σσµ
The Channel represent only 1/3 of the total Resistance in a 30 nm Device:
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90
50
100
150
200
250
300
350
400
450
500
VG - VTH (V)
EXTR
ACTE
D RE
SIST
ANCE
[ Ω . µ
m] TCAD: σ = 33
R0 =18
NEGF(kp2): σ = 31 R0=1.7
0 0.1 0.2 0.3 0.4 0.5 0.6 0.70
500
1000
1500
Ron Ω
. µ
m
L (µ m.)
VD 0.005V: Mean Error = 0.45 %
TCAD: µ 0 = 0.025
NEGF(kp2): µ 0 = 0.028
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90
50
100
150
200
250
300
350
400
450
VG - VTH (V)
EXTR
ACTE
D RE
SIST
ANCE
[ Ω . µ
m] All scatterings (kp2): σ = 31
Phonon Only: σ = 11
All scatterings (EMA):: σ = 25
Quantum (RQ/2): σ = 8.4
10 )(22 −−⋅⋅+⋅= THG VVRR σ
Access Region: Quasi-Fermi Level 10
How can we bridge Quantum transport to semi-classical transport?
Method 1: Current comparison Method 2: Fermi level calculation assuming Fermi Dirac Distribution è Localize the most resistive regions.
NEGF: extraction of quasi Fermi level • Occupied local density-of-states map in energy in linear regime (Vds =50mV)
11
Ballistic transport (no scattering) Interferences of coherent waves
Scattering with phonons. Incoherent transport
-0.5 0 0.510-8
10-6
10-4
10-2
100
ENERGY [eV]
DIST
RIBU
TIO
N [-]
DrainMid-channel
Fermi-Dirac distribution of occupied / unoccupied electrons
𝑓= 1/1+ 𝑒↑𝐸−𝐸↓𝐹 /𝑘↓𝐵 𝑇
à extraction of quasi-Fermi level Ef.
-0.03 -0.02 -0.01 0 0.01 0.02 0.03
0
1
2
3
4
5x 10-3
<FER
MI L
EVEL
> n (eV)
POSITION (µ m)
ChannelAccess Access
Q=1.3e13/cm2
Q=2e12/cm2
NEGF
Hydrodynamics
Access Region: Quasi-Fermi Level 12
1/3 ‘Lost’ in Access
-0.03 -0.02 -0.01 0 0.01 0.02 0.03
1013
1014
<DEN
SITY
> /c
m2
POSITION (µ m)
Q=2e12/cm2
Q=1.3e13/cm2
• The Near-Spacer region exhibts a clear and large Quasi-Fermi-Level voltage drop. • The diffusive part of the channel can be identified both in TCAD and NEGF
simulations • Square resistance can be calculated from the derivative of the QLF vs position.
Outlook
• What drives the Linear current in ultra-short MOSFETs?
• NEGF vs Semi classical TCAD • A focus on the Near-LDD region • the link bw Access Resistance and the ‘Apparent’ Channel
Mobility degradation.
• Investigation of the saturation regime • More NEGF simulations • Diffusive Model or Virtual Source Model? • Back to measurements
13
Theory of Access resistance 14
-0.04 -0.03 -0.02 -0.01 0 0.010
1
2
3
4
5
6x 104
SQUA
RE R
ESIS
TANC
E (Ω
per
squ
are)
POSITION (µ m)
1: Drain
4: Channel
VG-VTH = 0.06V
VG-VTH = 0.6V
2: Drain Corner
3: Spacer
NFDSOI square resistance along current path for VG-VTH=0.06V up to VG-VTH=0.6V; VDS=0.05V.
-0.04 -0.03 -0.02 -0.01 0 0.01 0.020
1000
2000
3000
4000
5000
6000
7000
POSITION (µ m)
SQUA
RE R
ESIS
TANC
E (Ω
per
squ
are)
Spacer
Spacer
Spacer
L=22nm
L=28nm
The near-spacer resistive region does not depend on L but depends on VG
BOX
Gate Drain Source
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90
50
100
150
200
250
300
350
400
450
500
VG - VTH (V)
EXTR
ACTE
D RE
SIST
ANCE
[ Ω . µ
m] TCAD: σ = 33
R0 =18
Quantum (RQ/2): σ = 8.4
{ } 02)(2)(
1 RVV
R BLDDTHG
acc ⋅++⋅⋅−
= σσ
Act as Access resistance
Outlook
• What drives the Linear current in ultra-short MOSFETs?
• NEGF vs Semi classical TCAD • A focus on the Near-LDD region • the link bw Access Resistance and the ‘Apparent’ Channel
Mobility degradation.
• Investigation of the saturation regime • More NEGF simulations • Diffusive Model or Virtual Source Model? • Back to measurements
15
Impact of R0 on linear Drain current 16
DDTHGDTHG
OX VVVVVVV
CLI )2/())2/(1(
0 −−−−⋅+
⋅=⋅
θµ
DDTHG
DTHGOX
OX VVVVVVV
LCRCLI )2/(
))2/()2(1( 0
0 −−−−⋅
⋅⋅⋅++
⋅=⋅
µθ
µ
R0
Impact of R0 only? … or
Measured @ VD=50mV
0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
2
4
6
8
10
12
14
16x 10-6
NORM
ALIZ
ED C
URRE
NT (A
): I*L
/W
GATE VOLTAGE (V)
LCH=0.9 µmLCH=1.8 µm
LCH=0.45 µm
LCH=0.15 µm
LCH=50 nm
LCH=20 nm
µ0(L) due to additional mecanisms?
))2/(1(0
DTHGeff VVV −−⋅+=
θµ
µ
Access resistance:
Channel Mobility: L-dependent Mobility degradation
(θ :Mobility degradation parameter)
Y function[1]: deembeding for θ and R0 17
DDTHG
DTHGOX
OX VVVVVVV
LCRL
CI )2/())2/()2(1( 0
0 −−−−⋅
⋅⋅⋅++⋅
⋅=
µθ
µ
gmIY D=
LCOXeff 0µβ =
)(0THGD
OX VVVLCY −⋅=
µ
Y-function slope
[1] F. Balestra, et al., proc. ESSDERC C4 (1988).
[1]
From slope calculated at a given VG-VTH overdrive one can extract the mobility
Y function measured in NFDSOI @ VD=50mV
0.4 0.5 0.6 0.7 0.8 0.9 10
0.5
1
1.5
2
2.5
3x 10-6
Y .
(L)1/
2 FU
NC
TIO
N [S
.I.]
GATE VOLTAGE [V]
Long Devices
Short Devices
10-2 10-1 100 10160
80
100
120
140
160
180
200
EXTR
ACTE
D µ o [c
m2/V
/s]
L [µ m]
FIT: σ = 31 µ0 =0.019 Cox =0.025
Back to the equation 18
10 )(22 −−⋅⋅+⋅= THG VVRR σ
σµβ 21 0 += OXY CL
DDTHG
DTHGOXOX
OX VVVVVVV
LCR
LCL
CI )2/())2/()2(21( 000
0 −−−−⋅
⋅⋅⋅++
⋅⋅⋅+⋅
⋅=
µθ
µσµ
Access Resistance:
LCOXeff ⋅⋅+= σµµ 211 0
The extracted mobility depends on L
gmIY D=
Y-function slope
)(111 0 Beff L µµµ ⋅+=1)2( −⋅⋅= OXBB Cσµ
or
0 0.5 1 1.5 20
1000
2000
3000
4000
5000
6000
Ron Ω
. µ
m
L (µ m.)
An link bw Racc and Channel mobility?
(Linear) { } 000
22)(
1)(
1 RVVCVVC
LRTHGOXTHGOX
on ⋅+⋅⋅−
+⎭⎬⎫
⎩⎨⎧
+−
⋅= σµθ
µ
Better to think with resistance than currents:
LCOXeff ⋅⋅+= σµµ 211 0
The apparent mobility degradation is linked to the access resistance VG dependence.
10-2 10-1 100 10160
80
100
120
140
160
180
200
EXTR
ACTE
D µ o [c
m2/
V/s]
L [µ m]
FIT: σ = 31 µ0 =0.019 Cox =0.025
0 0.1 0.2 0.3 0.4 0.5 0.6 0.70
100
200
300
400
500
600
700
800
VG - VTH (V)EX
TRAC
TED
RESI
STAN
CE [ Ω
. µ
m]
Racc: σ = 31 R0 =56
RCH (L=0.018 µ m): µ 0 = 0.022
RQ/2
Raccess
TCAD: σ=33 R0=18
Apparent mobility degradation: Measurements 20
Bulk vs FD NFD vs PFD NFD vs T NFD vs Tox
LCOXeff ⋅⋅+= σµµ 211 0 σ ~ 25 - 60
)(111 0 Beff L µµµ ⋅+=
Fit:
µΒ~ 6.7- 8.3 cm2/(V.s.nm) (NMOS) Small Ballistic Mobility per unit length but in-line with other researsh group results
[e.g. µΒ~ 1.8 – 9.9 cm2/(V.s.nm) in M. Zilli, et al., Electron Device Lett. 28, 1036 (2007).]
All Devices have homogenous channel processes (no pockets,…)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7100
200
300
400
500
600
700
VG - VTH (V)
EXTR
ACTE
D RE
SIST
ANCE
[ Ω . µ
m]
FDSOI: σ = 25 R0 =69
Bulk: σ = 46 R0 =33
-0.4 -0.2 0 0.2 0.4 0.6100
200
300
400
500
600
700
800
900
1000
VG - VTH (V)
EXTR
ACTE
D RE
SIST
ANCE
[ Ω . µ
m]
NFDSOI: σ = 25 R0 =69
PFDSOI: σ = 58 R0 =1.6e+002
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7100
150
200
250
300
350
400
450
500
VG - VTH (V)
EXTR
ACTE
D RE
SIST
ANCE
[ Ω . µ
m]
T=25°C: σ = 25 R0 =69
T=85°C: σ = 29 R0 =83
T=-40°C: σ = 24 R0 =66
0 0.2 0.4 0.6 0.8 1 1.2 1.4100
200
300
400
500
600
700
800
900
1000
1100
VG - VTH (V)
EXTR
ACTE
D RE
SIST
ANCE
[ Ω . µ
m]
Cox=0.025: σ = 25 R0 =69
VD 1.8V: σ = 9.2e+002 R0 =2.4e+002VD 1.8V: σ = 9.2e+002 R0 =2.4e+002
Cox=0.01: σ = 67 R0 =1.3e+002
10-2 10-1 100 10160
80
100
120
140
160
180
200
220
EXTR
ACTE
D µ o [c
m2/
V/s]
L [µ m]
FDSOI: σ = 25 µ0 =0.021
Bulk: σ = 46 µ0 =0.02
10-2 10-1 100 10120
40
60
80
100
120
140
160
180
200
220
EXTR
ACTE
D µ o [c
m2/
V/s]
L [µ m]
NFDSOI: σ = 25 µ0 =0.021
PFDSOI: σ = 58 µ0 =0.0069
10-2 10-1 100 10150
100
150
200
250
300
EXTR
ACTE
D µ o [c
m2/
V/s]
L [µ m]
T=25°C: σ = 25 µ0 =0.021
T=85°C: σ = 29 µ0 =0.016
T=-40°C: σ = 24 µ0 =0.03
10-2 10-1 100 10150
100
150
200
250
300
350
400
EXTR
ACTE
D µ o [c
m2/
V/s]
L [µ m]
Cox=0.025: σ = 25 µ0 =0.021
Cox =0.01: σ = 67 µ0 =0.037
10 )(22 −−⋅⋅+⋅= THG VVRR σ
Key ingredients : a realistic picture for the Smaller Devices in Linear Regime
21
1
Channel length
SOI BOX
Spacer
Leff
2 3-4
1: R0
3: spacer σLDD
.(VG-VLDD)-1
3: Quantum σB.(VG-VTH)-1
4: Channel Leff
.(µeff.COX )-1.(VG-VTH)-1
HDD
Racc :Access Resistance RCH: Channel Resistance
Access: 1. Contact Resistance (R0) 2. Quantum Resistance 3. Near-Spacer-region
Resistance
In general in optimized technologies L~ Leff
4/ No need for additional mecanisms in the Channel to account for the measurements:
Ron = Racc + RCH*L
Outlook
• What drives the Linear current in ultra-short MOSFETs?
• NEGF vs Semi classical TCAD • A focus on the Near-LDD region • the link bw Access Resistance and the ‘Apparent’ Channel
Mobility degradation.
• Investigation of the saturation regime • More NEGF simulations • Diffusive Model or Virtual Source Model? • Back to measurements
22
More NEGF simulations: Current 23
0.2 0.4 0.6 0.8 10
1000
2000
3000
4000
5000
6000
Voltage [V]
Cur
rent
[A/m
]
BalisticACAC + OP AC + OP + SR
0.2 0.4 0.6 0.8 10
200
400
600
800
1000
1200
1400
Voltage [V]
Effect of phonons
VDS = 0.9 V VDS = 50 mV
Effect ofscatterings
Effect ofscatterings
L= 20nm
More NEGF simulations: Velocity saturation
• Saturation of velocity in long-channel FDSOI due to optical phonons is well described with NEGF.
24
0 2 4 6 8x 106
0
2
4
6
8
10
12x 104
LONGITUNAL FIELD [V/m]
VELO
CIT
Y [m
/s]
NEGF -- EMANEGF -- KP2
µ=0.055m2/(Vs)ν = 1.1e5m/s
µ=0.07m2/(Vs)ν = 1.4e5m/s
Fit with Canali model
NEGF: # Lattice parameter (nm).ACELL = 0.5431 nm# Molar mass (g/mol).MASS = 28.0855 g/mol# Longitudinal sound velocity (m/s).VL = 9000 m/s.# Transverse sound velocity (m/s).VT = 5400 m/s.# Conduction band acoustic deformation potential (eV).DAC = 14.6 eV# Number of g-type phonons.NGTYP = 3# Energies of g-type phonons (eV).EGTYP1 = 0.012 eVEGTYP2 = 0.019 eVEGTYP3 = 0.062 eV# Deformation potentials of g-type phonons (eV/Ang).DGTYP1 = 0.5 eV/AngDGTYP2 = 0.8 eV/AngDGTYP3 = 11.0 eV/Ang# Number of f-type phonons.NFTYP = 3# Energies of f-type phonons (eV).EFTYP1 = 0.019 eVEFTYP2 = 0.047 eVEFTYP3 = 0.059 eV# Deformation potentials of f-type phonons (eV/Ang).DFTYP1 = 0.3 eV/AngDFTYP2 = 2.0 eV/AngDFTYP3 = 2.0 eV/Ang
More NEGF simulations: High VD Occupied LDOS map in saturation regime (Vds =0.9V) 25
Ballistic (no scattering) Scattering with phonons Incoherent transport
-0.1 0.1 0.3 0.5 0.7 0.9 1.1 1.3
10-10
100
Energy [eV]
Den
sity
[m-2
eV-1
]
AC onlyAC + OP
At the source
At the drain
φs
φd
Opt Ph still play a role in 20nm FDSOI NFet
More NEGF simulations: details 26
-30 -20 -10 0 10 20 300
0.5
1
1.5
2
2.5
3
3.5x 105
X POSITION [nm]
VELO
CIT
Y [m
/s]
LinearSaturation
-30-25-20-15-10 -5 0 5 10 15 20 25 30
-0.8
-0.6
-0.4
-0.2
0
X POSITION [nm]
QU
ASI F
ERM
I LEV
EL [e
V]
BALPHPH+SR
-30-25-20-15-10 -5 0 5 10 15 20 25 30-0.05
-0.04
-0.03
-0.02
-0.01
0
X POSITION [nm]
QU
ASI F
ERM
I LEV
EL [e
V]
PHPH + SRBAL
-30 -20 -10 0 10 20 30
1013
1014
X POSITION [nm]
CH
ARG
E D
ENSI
TY [c
m-2
]
SaturationLinear
Charge and velocity Quasi Fermi Level
50mV
0.9V
NEGF: EMA
Velocity overshoot
Outlook
• What drives the Linear current in ultra-short MOSFETs?
• NEGF vs Semi classical TCAD • A focus on the Near-LDD region • the link bw Access Resistance and the ‘Apparent’ Channel
Mobility degradation.
• Investigation of the saturation regime • More NEGF simulations • Diffusive Model or Virtual Source Model? • Back to measurements
27
P&S in saturation • Current without Resistance
))/(1/()/1/( 00 AVVQQ THGci ϕθµµµ −−⋅+=+=
)/( AVVCQ THGoxi ϕ−−⋅=
)1/( * yveff ∂∂⋅+=ϕµ
µµ
)(1)(1∫∫ ⋅∂=
∂∂
⋅∂=D
s
D
s
ieffieff QLy
QyL
Iϕ
ϕ
µϕϕ
µ
)1())/(1(
)/(
*
0
yvAVV
yAVVC
ITHG
THGox
∂∂⋅+⋅−−⋅+
∂∂⋅−−⋅
=ϕµ
ϕθ
ϕϕµ
))/(1(
)/(1)1(1 0
0*
0 AVVy
AVVCy
LyvIy
L THG
THGoxLL
ϕθ
ϕϕµ
ϕµ−−⋅+
∂∂⋅−−⋅
⋅∂=∂∂⋅+⋅⋅∂ ∫∫
∫ −−⋅+−−
⋅∂=D
sAVV
AVVLCI
THG
THGoxϕ
ϕ ϕθϕ
ϕµ
))/(1()/(0
0
))/(1/( AVV THG
D
s
ϕθϕδϕ
ϕ
−−⋅+∂= ∫
⎥⎦
⎤⎢⎣
⎡
−−⋅+−−⋅+
=−−⋅+∂∫ )/(1)/(1ln))/(1/(AVVVAVVVAAVV
DTHG
STHGTHG
D
sθθ
θϕθϕ
ϕ
ϕ
θδµ )/1(0
0DSDSox VV
LCI −
⋅=
)1(
)/1(
*0
0
δµ
θ
δµ
Lv
VVLCI DSDSox
+
−⋅=
Current accounting for theta, Vsat
yAVVCI THGox ∂
∂⋅−−⋅=ϕ
ϕµ )/(0
DDTHGox VAVVVCI ⋅−−⋅= )2/(0µ
AVVV THGDsat ⋅−= )(
20 )(
2 THGoxsat VVACI −⋅⋅= µ
P&S in saturation • Current with Resistance
)1(
)/1(
*0
0
δµ
θ
δµ
Lv
VVLCI DSDSox
+
−⋅=
R=60*2, sigma=30, theta=0.1 Vsat=2.1e5m/s
IRVIRVV
S
extDD
⋅=
⋅−=
Calculation of Vdsat is numerical
10-2 10-1 100 1010.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
VDSA
T (V
)
CHANNEL LENGH (µ m)
VGT=0.5V
VGT=0.8V
VGT=0.2V
(v*, θ,R)(v*, θ)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.80
2
4
6
8
10
12
14x 104
RESI
STAN
CE (Ω
. µ
m)
CHANNEL LENGH (µ m)
VGT=0.2V
VGT=0.5V
VGT=0.8V
VD=1V
⎥⎦
⎤⎢⎣
⎡
−−⋅+−−⋅+
=)/(1)/(1lnAVVVAVVVA
DTHG
STHG
θθ
θδand
Approximation on VDSAT 30
• Total Resistance @ high VD 2.σeff
⎭⎬⎫
⎩⎨⎧
+−
+⋅−
+⎭⎬⎫
⎩⎨⎧
−⋅
⋅⋅+
−⋅−
⋅⋅=
OXTHGTHG
D
THGOX
D
THGTHGOX
Don CvVV
RVVV
VVCV
VVVVCVLR *0
00
1)()(
2)(
2)()(
2 σµ
θµ
Channel ‘current’
0*00
22)(
1)2/(
1 RCvV
VVCVVVCLR
OX
D
THGOXDTHGOXon ⋅+
⎭⎬⎫
⎩⎨⎧
+⋅⋅−
+⎭⎬⎫
⎩⎨⎧
+−−
⋅= σµθ
µ
• Total Resistance @ low VD
0.2 0.4 0.6 0.8
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
RESI
STAN
CE (Ω
. µ
m)
CHANNEL LENGH (µ m)
VD 0.05V:
VG 0.675
VD 1V
Using P&S ModelUsing Ron Approximation
0.02 0.04 0.06 0.08 0.10
200
400
600
800
1000
1200
1400
1600
1800
RESI
STAN
CE (Ω
. µ
m)
CHANNEL LENGH (µ m)
VD 0.05V:
VG 0.675
VD 1V
Using P&S ModelUsing Ron Approximation
Zoom
Diffusive Model vs NEGF 31
COX=0.025;MU0=0.069;Vsat=2.1e5;R=0; sigma=13; theta=.001;
0 0.05 0.1 0.15 0.2 0.25 0.3
500
1000
1500
2000
2500
3000
3500
RESI
STAN
CE (Ω
. µ
m)
CHANNEL LENGH (µ m)
VD 0.01V: Mean Error = 0.34 %
VD 0.8V: Mean Error = 4 %
Using P&S ModelUsing Ron ApproximationNEGF
VD=10mV
VD=0.8VVGT=0.22V
VD=0.8VVGT=0.15V
VD=0.8VVGT=0.41V
0 0.1 0.2 0.3 0.4 0.50
500
1000
1500
VG - VTH (V)
EXTR
ACTE
D RE
SIST
ANCE
[ Ω . µ
m]
VD 0.01V: σ = 13 R0 =4.9
VD 0.8V: σ = 1.1e+002 R0 =19
⎭⎬⎫
⎩⎨⎧
+−
+⋅−
==OXTHGTHG
D
CvVVR
VVVLR *0
1)()(
2)0( σ
0* 22)(
1)0( RCvV
VVLR
OX
D
THG
⋅+⎭⎬⎫
⎩⎨⎧
+⋅⋅−
== σ
Velocity overshoot (NEGF Vsat (EMA) =1.4e5 m/s)
No L-dependence in Vsat here
Diffusive Model or Virtual Source Model*? 32
0 0.05 0.1 0.15 0.2 0.25 0.3
500
1000
1500
2000
2500
3000
3500
RESI
STAN
CE (Ω
. µ
m)
CHANNEL LENGH (µ m)
VD 0.01V: Mean Error = 0.34 %
VD 0.8V: Mean Error = 4 %
Using MVS ModelNEGF
VD=10mV
VD=0.8VVGT=0.41VVD=0.8V
VGT=0.22V
VD=0.8VVGT=0.15V
COX=0.025;MU0=0.06 ;vsat=1.2e5;R=0; sigma=13;
A fit with VSM seems not so easy: è Any suggestions?
• Injection velocity / Saturation velocity
• Virtual Source / Pinch Off
*Trans on Electron Device Vol 62 Sept 2015
Outlook
• What drives the Linear current in ultra-short MOSFETs?
• NEGF vs Semi classical TCAD • A focus on the Near-LDD region • the link bw Access Resistance and the ‘Apparent’ Channel
Mobility degradation.
• Investigation of the saturation regime • More NEGF simulations • Diffusive Model or Virtual Source Model? • Back to measurements
33
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.80
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2x 104
Ron Ω
. µ
m
L (µ m.)
VD 0.05V: Mean Error = 4.8 %
VD 0.24V: Mean Error = 3 %
VD 0.43V: Mean Error = 1.2 %
VD 0.62V: Mean Error = 1.5 %
VD 0.81V: Mean Error = 1.4 %
VD 1V: Mean Error = 1.4 %
0 0.05 0.1
500
1000
1500
2000
VD 0.05V: Mean Error = 4.8 %VD 0.24V: Mean Error = 3 %
VD 0.43V: Mean Error = 1.2 %
VD 0.62V: Mean Error = 1.5 %
34 • Total Resistance @ high VD
2.σeff
0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.50
500
1000
1500
2000
VG - VTH (V)
CH
AN
NE
L R
ES
ISTA
NC
E (L
=0.0
204 µ
m) [Ω
. µ
m ]
VD 0.24V: µ linext = 0.022 θ =0.0062
VD 0.43 V: µ satext = 0.016 θ ext =0.46
VD 0.62 V: µ satext = 0.023 θ ext =0.18
VD 0.81 V: µ satext = 0.023 θ ext =0.17
VD 1 V: µ satext = 0.023 θ ext =0.17
VD 0.05V: µ linext = 0.024 θ =0.0047
0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.50
500
1000
1500
2000
2500
VG - VTH (V)
EX
TRA
CTE
D R
ES
ISTA
NC
E [ Ω
. µ
m]
VD 0.05V: σ = 30 R0 =53
VD 0.24V: σ = 59 R0 =37
VD 0.43V: σ = 76 R0 =83
VD 0.62V: σ = 1.3e+002 R0 =52
VD 0.81V: σ = 1.7e+002 R0 =72
VD 1V: σ = 2.1e+002 R0 =80
⎭⎬⎫
⎩⎨⎧
+−
+⋅−
+⎭⎬⎫
⎩⎨⎧
−⋅
⋅⋅+
−⋅−
⋅⋅=
OXTHGTHG
D
THGOX
D
THGTHGOX
Don CvVV
RVVV
VVCV
VVVVCVLR *0
00
1)()(
2)(
2)()(
2 σµ
θµ
FDSOI GO1
Measurements
Channel ‘current’
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.80
200
400
600
800
1000
1200
EXTR
ACTE
D PA
RAM
ETER
: σ
DRAIN VOLTAGE (V)
GO1 GO2
Current Break Down: GO1/GO2 25°C 35
0.4 0.5 0.6 0.7 0.8 0.9 1
2
4
6
8
10
x 10-5
NORM
ALIZ
ED C
HANN
EL C
URRE
NT (A
)
GATE VOLTAGE (V)0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.5
1
1.5
2
x 10-5
NORM
ALIZ
ED C
URRE
NT (A
)
GATE VOLTAGE (V)
0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
0
1
2
3
x 10-4
NORM
ALIZ
ED C
HANN
EL C
URRE
NT (A
)
GATE VOLTAGE (V)0.4 0.6 0.8 1 1.2 1.4 1.6 1.80
0.5
1
1.5
2
2.5
3
3.5
4
x 10-5
NORM
ALIZ
ED C
URRE
NT (A
)
GATE VOLTAGE (V)
GO1
GO2
0.1 0.2 0.3 0.4 0.5 0.60
1000
2000
3000
4000
5000
VG - VTH (V)
EXTR
ACTE
D RE
SIST
ANCE
@ L
=0 [ Ω
. µ
m]
Temperature 36
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90
2000
4000
6000
8000
10000
12000
14000
16000
Ron Ω
. µ
m
L (µ m.)
VGT 0.575
T= -40°C
T= 125°C T= 85°C
T= 0°C
T= 25°C
0.02 0.03 0.04 0.05 0.06 0.07
1000
1200
1400
1600
1800
0.5 0.6 0.7 0.8 0.90
2
4
6
8
10
x 10-5
SHO
RT D
EVIC
E NO
RMAL
IZED
CHA
NNEL
CUR
RENT
(A)
GATE VOLTAGE (V)
T= -40°C
T= 85°C
T= 125°C
T= 0°C
T= 25°C
0.2 0.4 0.6 0.8 1
0.5
1
1.5
2
x 10-5
SHO
RT D
EVIC
E NO
RMAL
IZED
CUR
RENT
(A)
GATE VOLTAGE (V)
T= 25°C
T= 125°C
T= 85°C
T= 0°C
T= -40°C
VD=1V
⎭⎬⎫
⎩⎨⎧
+−
+⋅− OXTHGTHG
D
CvVVR
VVV
*01
)()(2 σ
⎭⎬⎫
⎩⎨⎧
−⋅
⋅⋅+
−⋅−
⋅⋅
)(2
)()(2
00 THGOX
D
THGTHGOX
D
VVCV
VVVVCVL
µθ
µ
Velocity Saturation 37
0 0.2 0.4 0.6 0.8 10
100
200
300
400
500
600
EXTR
ACTE
D σ
DRAIN VOLTAGE (V)
T=-40 °CT=-25 °C
T=125 °C
-40 -20 0 20 40 60 80 100 120 1400.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2x 105
TEMPERATURE (deg)
SATU
RATI
ON
VELO
CITY
(m/s
)
Conclusions
1. TCAD and NEGF comparison with a focus on the LDD-region
2. Ron Resistance breakdown: • Ron = Raccess (50-70%) + Rch (30-50%) • Extraction methods should be interpreted with care
3. LDD-region drives a large part of Ron 1. TCAD but also NEGF clear highlight the near LDD region on the total resistance
of the device
4. Saturation Regime 1. P&S or VS model? 2. Velocity Saturation 3. Not accounted for Velocity overshoot
38