analytical modeling of nano mosfets in the quasi ballistic … · 2021. 3. 15. · mos-ak workshop...
TRANSCRIPT
MOS-AK Workshop 20101
Analytical modeling of nano MOSFETsin the quasi ballistic regime :
beyond the drift diffusion approximation
IMEP-LAHCGrenoble Institute of Technology
France
DIEGMUniversity of Udine
Italy
R. Clerc, G. Ghibaudo
P. Palestri, L. Selmi
MOS-AK Workshop 20102
Transport in Nano MOSFET (L < 20 nm) challenges Compact Modeling
« Sophisticated computer simulations using techniques such as full band Monte Carlo and full quantum transport approaches are being used to explore the physics of the ultimateMOSFET, but circuit models continue to be based on concepts and approaches developedin the 1960's. »
M. Lundstrom, Int. SOI Conference, 2006
Since the pioneering work of Natori (1994) and Lundstrom (1996), the quasi ballistic regime of transport in nano MOSFETs
has been extensively investigated.
However, existing compact model are still based on Drift Diffusion and saturation velocity concepts
Let us examine the applicability of Quasi Ballistic theories to compact modeling
MOS-AK Workshop 20103
Outline
The “orthodox” Lundstrom theory
Beyond Lundstrom’s theory
Ballisticity extraction from experiments
MOS-AK Workshop 20104
The “orthodox” Lundstrom theory
MOS-AK Workshop 20105
Performance are no longer limited by transport along the channel,
but by injection at the source end
Transistor in ballistic regimeSource Drain
L
Energy
Flu
x of
car
rier (In quasi equilibrium)
Virtual Source
Channel
Net current
Gate(In quasi equilibrium)
Vd
injiBAL d V Q WI ≈
Transistor in diffusive regimeSource Drain
L
Energy
Flu
x of
car
rier
(In quasi equilibrium)
Virtual Source
Channel
Gate(In quasi equilibrium)
Vd
Scattering )y(dy
E d Q W)y(I Fn
effid µ=
Concept of Ballistic Limit
MOS-AK Workshop 20106
Concept of Backscattering Coefficient
Perfect non degenerated
reservoir of carriers“Source like”
x = Lx = 0
thermalisation
thermalisation
rs js+
(1 – rs) js+
(1– rd) jd-
rd jd-
Perfect non degenerated
reservoir of carriers“Drain like”
Distance
jd-
js+
Using Mc Kelvey flux theory of Transport :
J. P. McKelvey, R. L. Longini and T. P. Brody, « Alternative approach to the solution of added carrier transport problems in semiconductors », Phys. Rev., vol123 pp. 51-57 (1961).
Assuming an Inversion Charge Qi at the virtual source controlled by the gate :(like in the Natori’s model)
thid
dd vQ r)(1 kT)/ eV( exp)r1(
r)(1 kT)/ eV( exp)r1(
W
I
−−++−−−−=
thiSat d vQ
r1
r1
W
I
+−=
kT
V e v
2
Q )r1(
W
I dth
iLin d −=
A. Rahman and M. S. Lundstrom, « A compact scattering model for the nanoscale double-gate MOSFET », IEEE. TED, vol. 49 p 481 - 489 (2002)
MOS-AK Workshop 20107
λL
Lr
kT
kTHF +
=
intuited as a generalization of rLF
Gate
Source
Drain
F+
F-kT
LkT
L
rHF = Back Scattering Coefficient at High Field (Saturation Regime)
M. Lundstrom Z. Ren, IEEE TED 49 p.133 (2002)
empirical
λ+=
L
LrLF
M. Lundstrom, « Fundamentals of Carrier Transport », second edition, Cambridge university press, 2000
λ+=
L
LrLF (using Boltzmann or Fermi Statistics)assuming a constant isotropic mean free path :
rLF = Back Scattering Coefficient at Low Field (Ohmic Regime)
When L >>λ, this is consistent with the Drift Diffusion model :
dieffDD Lin d V Q µ
L
1
W
I =e
kT
v
µ 2
th
eff=λif
M. Lundstrom IEEE EDL 22 p. 293 (2001)
kT
V e v
2
Q )r1(
W
I dth
iLin d −=L >>λ
Concept of Backscattering Coefficient
MOS-AK Workshop 20108
In the MOSFET model, we have : withthiSat d vQ
r1
r1
W
I
+−=
λL
Lr
kT
kTHF +
=
When L >>λ, this is consistent with the Drift Diffusion model :
2
)V(VC µ
L
1
W
I2
Tgoxeff
DD Sat d −=
e
kT
v
µ 2
th
eff=λ
if
(same expression than in ohmic regime)
and TgSAT d
kT VV
e/kT 2L
V
e/kT 2LL
−==
which is in fact equal to the LkT layer calculated from the Drift Diffusion potential profile !
1
10
100
1 10 100 1000 10000
Channel Length L (nm)
1000
Dra
in C
urre
nt(µ
A/µ
m)
λ
Id Lin
Id Lin DD Id Lin BAL
Id Sat BAL
Id Sat
Id Sat DD
µ = 200 cm2V-1s-1
V inj = 1.2 × 105 m/sNinv = 1.45×1013 cm-2
M. S. Lundstrom and J. H. Rhew Journal of Computational Electronics, vol. 1 pp 481 - 489 (2002).
• In ohmic regime, Quasi ballistic transport occurs when L ∼ λ• In saturation, Quasi ballistic transport occurs when LkT ∼∼∼∼ λλλλ i.e when L ∼∼∼∼ 10 λλλλ
Backscattering Coefficient Modeling
MOS-AK Workshop 20109
rHF extracted is in fact a function of LkT
λL
Lr
kT
kTHF +
= ??
100 200 300 400 500
0
10
20
0
MC low field mobility (cm2V-1s-1)
Ext
ract
edλ D
ev(n
m)
DG 10 nm
DG 4 nmBulk
Strained Bulk
µev
kT 2
th0 =λ
2
λλ 0=
DevkT
kTMC L
Lr
λ+=
??λ extracted is in fact
proportional to µ
µev
kT 2
th0 =λ
IEDM 2006« Multi Subband Monte Carlo investigation of the mean free path and of the kT layer in degenerated quasi ballistic MOSFETs »P. Palestri, R. Clerc, D. Esseni, L. Lucci, L. Selmi
Validation by Monte Carlo Simulation
MOS-AK Workshop 201010
Strain Silicon : double advantage in QB devices
0
0.5
1
1.5
2
2.5
Bulk DG 12
Inje
ctio
n V
eloc
ity 1
07 cm
/s
Analytical Model
0
0.1
0.2
0.3
0.4
0.5
Bulk DG 12
Bac
k S
catte
ring
rth
effeff
kT
kT
v
kT/e )E(µ 2
2
1
L
Lr
×=λ
λ+=
Multi SubbandMonte Carlo Simulations
Ph. AcousticPh. OpticalSurf. RoughnessTsi Fluctuation
D. P
onto
net
.al.
Pro
c. E
ssde
rc20
06 p
. 166
Application to device performance modeling
MOS-AK Workshop 201011
(An increase of Vinj may also degrade r : is subband engineering a viable strategy ?)
Vth and rHF are valid in non degenerated inversion layers
• How to generalize LkT and λ in degenerated inversion layer ?
λL
Lr
kT
kTHF +
= • is only an empirical formula : where does it come from ?
• How to evaluate LkT ? (V(x) is not known)
(In particular the impact of velocity overshoot on V(x), and thus LkT is not self consistently taken into account)
Limits of Lundstrom’s Model
µv
kT/e 2λ
th
= • OK in low field
µv
kT/e 2λ
th
α=2
1≈αwith • But in high field rather equal to
MOS-AK Workshop 201012
Beyond Lundstrom’s Model
R. Clerc, P. Palestri, L. Selmi« On the Physical Understanding of the kT-Layer Concept in Quasi-Ballistic Regime of Transport in Nanoscale Devices »IEEE TED 53, p 1634 – 1640 (2006)
MOS-AK Workshop 201013
vx
x
B
F1 F2
F4 F5
F3x + ∆x
λΦ−Φ−−=Φ ++
−− )x()x(
m
F e )f(x,0
dx
d LE
)x(dv v)vf(x, F
0
xxx1+
∞
Φ== ∫
BA231 CFFF →+=+
λΦ−Φ∆=
λ−∆=
++∞
→ ∫)x()x(
x dv /v
)v,x(f)vf(x, xC LE
0
xx
xLExBA
1D Balance Equation in the phase space
f+(x,v)
f-(x,v)
Collision Integral (relaxation length approximation ) :
A
Balance Equation for the flux of carrier along the device
Assumption on the shape of f(x,vx) needed
1D flux conservation in the relaxation length approximation
MOS-AK Workshop 201014
)kT 2
vm( exp (x)n
kT π2
m2) v,x(f
2
−= ±±
The quasi ballistic Drift Diffusion
Assuming local Maxwellian distribution function (at each point) :
- 4 - 2 0 2 4Dis
trib
utio
n F
unct
ion
(a.u
)
velocity v/vth
This work
Drift Diffusion Approximation
Application to the backscattering coefficient calculation:
)β1(Lλ
)β(1 Lr
kT
kT
−+−= β = exp( − L / LkT)
0.01 0.1 1
0.2
0.4
0.6
0.8
1
λ+L
L
λ+kT
kT
L
LL = λ / 5
L = 50 λ
This model
Drain Voltage Vd (V)
Bac
kSca
tterin
gC
oeffi
cien
t r
This equation includes both :
kTHF
kT
Lr
λ L=
+LF
Lr
λ L=
+
Similar to Drift Diffusion
MOS-AK Workshop 201015
x thth x th x
λ vd nq (n n ) v λ v 2n F
dx kT / q+ − −Φ = − = − −
Using this approach, if we derive the current flux :
x th
µµ '
1 µF / v=
+th x
d n(n n ) v D' n µ ' F
dx+ −Φ = − = − −
We thus have now a new formalism which includes :
Thermal velocity limitation at the source (Ballistic limit)
Same backscattering coefficient in high and low field (Ballistic mobility)
With no need to calculate LkT …
An alternative to Drift Diffusion ?
n n n+ −= +Using :
Except for the boundary conditions, it is equivalent to Drift Diffusion, including saturation velocity !
Idem for H. Wang, G. Gildenblat, “Scattering matrix based compact MOSFET model”, in IEDM Tech. Dig., pp. 125–128 (2002)
The quasi ballistic Drift Diffusion
MOS-AK Workshop 201016
A more suitable approximated distribution function (at each point) :
m
U(x)2
Ballistic Electron
Velocity
Dis
trib
utio
n F
unct
ion
(a.u
.)
Backscattered Electron
MB
From MC simulations
)kT 2
vm( exp (x)n
kT π2
m2) v,x(f
2
−= ±±
Non Thermal Approach
What is wrong with quasi ballistic Drift Diffusion ?
MOS-AK Workshop 201017
)L,L(A L 2
)L,L(A Lr
kTkT
kTkT
+λ= du)
kT
)u(U1(
2
)u(2 )L,L(A
1L/L
0
kT
kT−
∫
+λ
λ=Backscattering formula :
Non Thermal Approach
1 10 1000.01
0.1
1
kT layer length LkT (nm)
Bac
ksca
tterin
gC
oeffi
cien
t r
MC
This model
Thermal
2 λth = 64 nm
2 λth = 18 nm
2 λth = 36 nm
Linear Pot.
1 10 1000.01
0.1
1
Bac
ksca
tterin
gC
oeffi
cien
t r Parabolic Pot.
2 λth = 64 nm
2 λth = 18 nm
2 λth = 36 nm
kT layer length LkT (nm)
MC
This model
Thermal
MOS-AK Workshop 201018
Non Thermal Approach
Distance (nm)0 2 4 6 8 10 12 14
0
0.5
1
1.5
2
2.5
3
VV+
V−
Linear Pot.
vth
Symbols : MCLines : Model
LkT = 2 nm2 λth = 36 nm
Vel
ocity
(10
5m
/s)
V thermal
vth
Symbols : MCLines : Model
0 5 10 15 20
Distance (nm)
Linear Pot.LkT = 2 nm
2 λth = 36 nm
V+
V
V−
V thermal
0.5
1
1.5
2
2.5
Vel
ocity
(10
5m
/s)
Velocity profile in high field in a non self consistent linear potential profile
With absorbing drain : With real drain (emitting) :
MOS-AK Workshop 201019
Ballisticity extraction from experiments
MOS-AK Workshop 201020
Ballisticity extraction from experiments
Is there any experimental confirmation of the quasi ballistic nature of transport ?
Let us show some results :
Not clear yet !
• let us consider only the ohmic regime (more simple : you do not need to know the charge at the virtual source)
• a standard low field mobility extraction has been performed on :
65 like nm Bulk and undoped Fully Depleted SOI technology
featuring physical gate length down to 40 nm
• according quasi ballistic theory :
d Lin dith app i d
I e VQ 1(1 r) v µ Q V
W 2 kT L= − = app dd
Lµ µ
L=
+ λ
Apparent mobility should decrease with L
in the quasi ballistic regime
MOS-AK Workshop 201021
Bulk nMOS
0.01 0.1 1 10200
300
400
500
600
700
Low
Fie
ld M
obili
tyµ e
xp(c
m2 /
Vs)
Channel length L (µm)
µexp
µdd
Undoped FD-SOI nMOS
0.01 0.1 1 1080
100
120
140
160
180
200
Channel length L (µm)
µdd
µexp
Low
Fie
ld M
obili
tyµ e
xp(c
m2 /
Vs)
Ballisticity extraction from experiments
Bulk 65nm CMOS technologydoped channel (≈1017/cm3) with halos,
SiON gate oxide (CET=2.2nm)polysilicon gate.
Undoped Fully depleted SOIBody thickness of 10 nm
Metal gate TiN, 2.5 nm of HfSiON dielectricRaised source and drain
MOS-AK Workshop 201022
0.01 0.1 1 100.6
0.7
0.8
0.9
1
Channel length L (µm)
Bac
ksca
tterin
gco
effic
ient
r
Bulk nMOS
0.85
0.9
0.95
1
FD-SOI nMOS
0.01 0.1 1 10Channel length L (µm)
Bac
ksca
tterin
gco
effic
ient
r
Bulk nMOS Undoped FD-SOI nMOS
Ballisticity extraction from experiments
rLF ~ 0.7 rLF ~ 0.85
Extracted value of r are very large … Neutral defects ?
app
bal
µr 1
µ= −The backscattering coefficient has been extracted using
MOS-AK Workshop 201023
Conclusions
MOS-AK Workshop 201024
Conclusions
The Lundstrom backscattering theory provides a powerful guidelineto analyse qualitatively device performance in the quasi ballistic regime
However, It is not a compact model (How to compute LkT ?)
If you try to generalize this approach along the channel → quasi ballistic drift diffusion → which is similar to drift diffusion with saturation velocity in high field regime
However, experiments shows a more complex picture : we are not sure yet to operate in the quasi ballistic regime (neutral defects ?)
You need to account for the ballistic distribution of carriers
MOS-AK Workshop 201025
Thank you for your attention !
MOS-AK Workshop 201026
Dra
in C
urre
nt (
a.u)
1 10 100 1000 10000Channel Length (nm)
Ballistic enhancement factor
Drift Diffusion1/L
Drift Diffusion Analytical Model (Velocity Saturation)
Mean FreePath
Ballistic limitMore Exact Quasi Ballistic Model
MOS-AK Workshop 201027
0
500
1000
1500
2000
2500
3000
3500
1 10 100 1000 10000
Bulk Undoped UTB
Strained Undoped UTB
I on
Cur
rent
(µ
A/µ
m)
Channel Length (nm)
DDvsat=105 m/s
BULKµ = 130 cm2V-1s-1
V inj = 1.2 × 105 m/sNinv = 1.45×1013 cm-2
Undoped UTBµ = 200 cm2V-1s-1
V inj = 1.2 × 105 m/sNinv = 1.45×1013 cm-2
Strained Undoped UTBµ = 370 cm2V-1s-1
V inj = 1.3 × 105 m/sNinv = 1.45×1013 cm-2
2 possible strategies to improve Ion :
• improving Vinj
(subband engineering)by DOS reduction
• improving λλλλ, which mean improving µµµµ= effective field mobility
like in pure Drift Diffusion model !
Device Optimisation in the Quasi Ballistic Regime
Still no clear experimental evidence
Application to device performance enhancement
MOS-AK Workshop 201028
Re-investigation of the kT layer Concept
λL
Lr
kT
kTHF +
= has been derived using Quasi Ballistic Drift Diffusion.
key assumptions on which this formula is based :
• Boltzmann statisticsin the contacts• Non Self consistentpotential • use of the relaxation length approximationwith a constant λ
• the population of backscattered carriers (f−) has an equilibrium Maxwellian distribution: at each point x, regardless of the channel length and of the magnitude of the electric field
λ should be energy dependent (especially at high field)
reasonable approximation Only close to the virtual sourceOr along the channel when many collisions are involved