23rd Symposium on Microelectronics Technology and Devices: SB Micro 2008 IEEE / EDS Mini Colloquium September 1, 2008, Gramado

, Brazil

1

23rd Symposium on Microelectronics Technology and Devices: SB Micro 2008 IEEE / EDS Mini Colloquium September 1, 2008, Gramado, Brazil

Physics of Nanoscale Transistors:
An Introduction to Electronics from the Bottom Up Mark Lundstrom
Network for Computational Nanotechnology Birck Nanotechnology Center Discovery Park, Purdue University West Lafayette, Indiana USA
NCN
nanoHUB.org

2

technology trends…..

5 μm
↑ log L

‘Moore’s Law’
10 9 ↑ log # chip 10 3
2005

50 nm 5 nm 1975

3

models for devices (conceptual and computational)
Macroscopic dimensions

0.1 mm 10 µm 1µm 0.1 µm 10 nm 1 nm 0.1 nm

drift-diffusion drift-diffusion + velocity saturation Boltzmann for velocity overshoot quasi-ballistic
Atomic dimensions

quantum mechanical

4

21st Century electronic devices
molecular electronics Al Gate
D
HfO2

S

10 nm SiO2

p++ Si
SWNT

nanonets

carbon nanotube electronics

CoFeB (3) MgO (0.85) Insulator CoFeB (3) Ru (0.85) CoFe (2.5) NCN
nanoHUB.org

nanowire PV
spin torque devices
5

nanowire bio-sensors

“Electronics from the Bottom Up”

1) Introduction 2) Generic model of a nanodevice 3) The ballistic MOSFET 4) Scattering in nano-MOSFETs 5) Discussion 6) Summary

6

generic model Gate

EF1

D(E − U )

EF 2

τ1

τ2

S. Datta, Quantum Transport: Atom to Transistor, Cambridge, 2005 7 (“Concepts of Quantum Transport” nanohub.org)

filling states from the left contact Gate
Assumption: Each energy channel is independent

EF1

D( E − U )

includes spin

τ1

N (E) = D(E − U ) f1 (E)
0 1

d N(E) N10 (E) − N = dt τ1 8

filling states from the right contact Gate

D( E − U )

EF 2

0 N 2 (E) = D(E − U) f2 (E)

τ2

0 d N(E) N 2 (E) − N = dt τ2

9

steady-state
0 d N(E) N10 − N N 2 − N = + =0 τ1 τ2 dt

(1 τ 1 )N10 − (1 τ 1 )N + (1 τ 2 )N 20 − (1 τ 2 )N = 0

(1 τ 1 ) N 0 (E ) + (1 τ 2 ) N 0 (E ) N (E ) = (1 τ 1 )+ (1 τ 2 ) 1 (1 τ 1 )+ (1 τ 2 ) 2
N10 ( E ) ≡ D (E − U ) f1 (E )
0 N 2 ( E ) ≡ D (E − U ) f2 (E )
10

γ 1 = h τ1 γ 1 = h τ2

steady-state electron number, N(E)
N (E ) =

γ1 +γ 2

γ1

D (E − U ) f1 (E ) +

γ1 +γ 2

γ2

D (E − U ) f2 (E )

N (E ) = D1 (E − U ) f1 (E ) + D2 (E − U ) f2 (E )
γ1

D1 (E − U ) = D2 (E − U ) =

γ1 +γ 2 γ1 +γ 2 γ2

D (E − U ) D (E − U )

DOS that can be filled by contact 1 DOS that can be filled by contact 2
11

steady-state current, I
Gate

ID

ID

EF1
f1 (E )
N10 (E) − N (E )

D(E − U )

EF 2
f2 (E )

VDS

d N(E) = dt 1

τ1

0 d N(E) N 2 (E) − N = dt 2 τ2

d N(E) d N(E) I D (E ) = +q = −q dt 1 dt 2
12

result
q ⎛ γ 1γ 2 ⎞ I (E ) = ⎜ ⎟ D (E − U )( f1 − f2 ) h⎝ γ1 + γ 2 ⎠

γ1 = γ 2 = γ
2q ⎛ γ ⎞ I D = ∫ I (E )dE = ∫ ⎜ 2 ⎟ π D (E − U )( f1 − f2 )dE h ⎝ ⎠ ⎡ D (E − U ) ⎤ N = ∫ N (E )dE = ∫ ⎢ ( f1 (E ) + f2 (E ))⎥ dE 2 ⎣ ⎦

13

final results

γ1 = γ 2 = γ = h τ
D (E − U ) D′ (E − U ) = 2
density-of-states per spin

2q ID = ∫ γ π D′ (E − U )( f1 − f2 )dE h
N = ∫ D′ (E − U )( f1 + f2 )dE

14

determining τ (γ)
energy channels are independent:
2q I D (E ) = γ π D′ (E − U )( f1 − f2 ) h N (E ) = D′ (E − U )⎡ f1 (E ) + f2 (E )⎤ ⎣ ⎦

if f1 >> f2 (source injects, drain collects), then:
qN h = =τ ID γ

stored charge ID = = τ transit time
15

Q

“Nanoelectronics and the Meaning of Resistance”
1) What and where is the resistance? 2) Microscopic model for electrical resistance 3) Spins and magnets 4) Energy conversion 5) Beyond the one-electron picture

Supriyo Datta
“Electronics from the Bottom Up” on nanoHUB.org
16

outline

1) Introduction 2) Generic model of a nanodevice 3) The ballistic MOSFET 4) Scattering in nano-MOSFETs 5) Discussion 6) Summary

17

controlling current with energy barriers
electron energy vs. position
VGS VDS = 0.05 V

VDS = 1.0 V

E = −qV
VGS

E.O. Johnson, “The Insulated-Gate Field Effect Transistor: A Bipolar Transistor in Disguise,” RCA 18 Review, 34, pp. 80-94, 1973.

“top of the barrier” MOSFET model
U = EC = EC 0 − qψ S

energy

low drain bias:

EF1

‘device’

EC (0 )

LDOS
≈L
contact 1 contact 2

EF 2
EC (x )

γ1 = γ 2 =
19

h

position

τ

“top of the barrier” MOSFET model
U = EC (0 ) = EC 0 (0 ) − qψ S

energy

high drain bias:

EF1

‘device’

EC (0 )

EF 2
<< L
contact 1 contact 2

EC (x )

γ1 = γ 2 =
20

h

position

τ

electron density
U = EC (0)
y

W
contact 1

θ

υ
x
contact 2

N = ∫ D′ (E − U )( f1 + f2 )dE
m* Wl D′ = 2 2π h

f1 (E ) =

1 1 + e(EF1 − EC (0)) kB T
21

f2 (E ) =

1 1 + e(EF1 − qVDS − EC (0)) kB T

electron density
N = ∫ D′ (E − U )( f1 + f2 )dE
N N2D ⎡F 0 (ηF1 ) + F 0 (ηF 2 )⎤ nS (0 ) = = ⎦ 2 ⎣ Wl
N 2 D = m * k BT π h2

ηF1 ≡ [EF1 − EC (0)] kB T
ηF 2 = ηF1 − qVDS kBT
22

current
U = EC (0)
y

W
contact 1

θ

υ
x
contact 2

2q ID = ∫ γ π D′ ( f1 − f2 )dE h

h υx γ = = τ l h
23

m* D′ = Wl 2 2π h

current (cont.)
U = EC (0)
y

W
contact 1

θ

υ
x
contact 2

h υx γ = = τ l h

υ x = υ cosθ = 2 (E − EC ) m* cosθ
υx =

∫π

π /2

− /2

υ cosθ dθ = υ

2

π

=

2 (E − EC ) 2 m*

π

24

current (cont.)
2q ID = ∫ γ π D′ ( f1 − f2 )dE h

h 2 (E − EC ) 2 m* γ π D′ (E ) = ×π × Wl * 2 l m π 2π h

γ π D′ (E ) = W

2m * (E − EC )

πh

= M (E )

W M (E) = (λ 2 )

2q ID = ∫ M (E )( f1 − f2 )dE h
25

current (final result)
2q ID = ∫ M (E )( f1 − f2 )dE h

⎛ N2 D ⎞ I D = Wq ⎜ υT ⎟ ⎡F 1/ 2 (ηF1 ) − F 1/2 (ηF 2 )⎤ ⎦ ⎝ 2 ⎠⎣

N 2 D = m * k BT π h2

ηF1 ≡ [EF1 − EC (0)] kB T
ηF 2 = ηF1 − qVDS kBT
26

υT = 2kBT π m*

re-cap

⎛ N2 D ⎞ I D = Wq ⎜ υT ⎟ ⎡F 1/ 2 (ηF1 ) − F 1/2 (ηF 2 )⎤ ⎦ ⎝ 2 ⎠⎣

(1)

N2 D ⎡F 0 (ηF1 ) + F 0 (ηF 2 )⎤ nS (0 ) = ⎦ 2 ⎣
Solve (2) for N2D, then insert in (1):

(2)

27

I-V characteristic
⎡F 1/2 (ηF1 ) − F 1/2 (ηF 2 )⎤ I D = WqnS (0 )υT ⎢ ⎥ ⎣ F 0 (ηF1 ) + F 0 (ηF 2 ) ⎦ ⎧ F 1/2 (ηF1 )⎫ ⎡ 1 − F 1/2 (ηF 2 ) F 1/2 (ηF1 )⎤ ⎪ ⎪ I D = WQI (0 )⎨υT ⎬⎢ ⎥ ⎪ F 0 (ηF1 ) ⎪ ⎣ 1 + F 0 (ηF 2 ) F 0 (ηF1 ) ⎦ ⎩ ⎭

qnS (0 ) ≈ Cox (VGS − VT )

(simple, 1D MOS electrostatics VGS > VT)
28

final result

⎡ 1 - F 1/2 (ηF 2 ) F 1/2 (ηF1 )⎤ % I D = WCox (VGS − VT )υT ⎢ ⎥ 1 + F 0 (ηF 2 ) F 10 (ηF1 ) ⎦ ⎣

% υT ≡

2k BT F 1/2 (ηF1 ) π m * F 0 (ηF1 )

29

Boltzmann limit

F j (ηF ) → eη

F

⎡ 1 - e− qVDS / kB T ⎤ I D = WCox (VGS − VT )υT ⎢ 1 + e− qVDS / kB T ⎥ ⎣ ⎦

υT =

2kBT π m*
30

on-current
⎡ 1 - F 1/2 (ηF 2 ) F 1/2 (ηF1 )⎤ % (VGS − VT )⎢ I D = WCoxυT ⎥ 1 + F 0 (ηF 2 ) F 10 (ηF1 ) ⎦ ⎣

% I D → WCoxυT (VDD − VT )

I DS
% υT ≡ 2kBT F 1/2 (ηF1 ) π m* F 0 (ηF1 )
VGS = VDD

(100) [110] Si (single subband)
31

VDSAT ≈ k BT q

VDS

velocity saturation in a ballistic MOSFET
⎡ 1 - F 1/2 (ηF 2 ) F 1/2 (ηF1 )⎤ % I D = WCoxυT (VGS − VT )⎢ ⎥ = WQI (0) υ (0 ) ⎣ 1 + F 0 (ηF 2 ) F 10 (ηF1 ) ⎦

υ (0)
υT
‘ballistic injection velocity’

I DS
VGS = VDD

≈ [EF1 − EC (0)] q

VDS
32

VDS

channel resistance of a ballistic MOSFET
⎡ 1 - F 1/2 (ηF 2 ) F 1/2 (ηF1 )⎤ % (VGS − VT )⎢ I D = WCoxυT ⎥ 1 + F 0 (ηF 2 ) F 10 (ηF1 ) ⎦ ⎣

I DS
finite channel resistance
VGS = VDD

as T → 0
GCH 2q 2 1 = → M (EF1 ) h RCH
33

VDSAT ≈ k BT q

VDS

is a nanoscale MOSFET really ballistic?
Typical N-channel MOSFET:
I ON ≈ 1 mA/μm
% I ON (ballistic ) = −WQ I (0 )υ T

−QI (0) q = Cinv (VDD − VT )
about 50% of the ballistic limit

≈ 0.8 × 1013 cm -2

υT ≈ 1.8 × 10 7 cm/s
I ON W (ballistic ) ≈ 2 mA/μm

(Courtesy, Shuji Ikeda, ATDF, Dec. 2007)

about 10% of the ballistic limit.
34

outline

1) Introduction 2) Generic model of a nanodevice 3) The ballistic MOSFET 4) Scattering in nano-MOSFETs 5) Discussion 6) Summary

35

scattering in Si n-channel nano-MOSFETs

ID =

W μnCox (VGS − VT )VDS → ? L
36

“top of the barrier” MOSFET model
(ballistic) low drain bias:
υ=
2 (E − EC ) m
*

energy

ˆ ˆ (cosθ x + sinθ y )

EF1

‘device’

EC (0 )

LDOS
≈L
contact 1 contact 2

EF 2
EC (x )

γ1 = γ 2 =
37

h

position

τ

diffusive current
U = EC (0) W
contact 1

y
x

υ
contact 2

L
2q ID = ∫ γ π D′ ( f1 − f2 )dE h
h υx γ = = τ L h

ballistic

γ =
38

h

τ

= ? diffusive

L2 τ = 2Dn

between ballistic and diffusive
⎛ ⎞ h 1 h γ = =⎜ τB +τD ⎝1+τD τB ⎟ τB ⎠

υ ⎛ 2⎞ Dn (E) = λ0 = ⎜ ⎟ λ0 2 2⎝π⎠
γ π D′ = λ0 + L λ0
× γ B π D′

υx

T=

λ0 + L

λ0

⎛ λ0 ⎞ ⎡ 1 - F 1/2 (ηF 2 ) F 1/ 2 (ηF1 )⎤ % I D = WCox (VGS − VT )⎜ ⎟ υT ⎢ 1 + F (η ) F (η ) ⎥ ⎝ λ0 + L ⎠ ⎣ 0 F2 10 F1 ⎦
39

linear-region current
Boltzmann statistics:

⎛ λ0 ⎞ ⎡ 1 - e− qVDS / kB T ⎤ I D = WCox (VGS − VT )⎜ ⎟ υT ⎢ 1 + eqVDS / kB T ⎥ ⎝ λ0 + L ⎠ ⎣ ⎦
Low VDS:

⎛ λ0 ⎞ υT I D = WCox (VGS − VT )⎜ ⎟ 2 (k T q )VDS ⎝ λ0 + L ⎠ B

W ID = μnCox (VGS − VT )VDS L + λ0
40

scattering in Si n-channel nano-MOSFETs
on-current

W W I D = μn Cox (VGS − VT )VDS → I D = μnCox (VGS − VT )VDS L L + λ0
41

where scattering matters (the most)

EC vs. x for VGS = 0.5V

EC vs. x for VGS = 0.5V

--->

EC (eV)

EC (eV)

--->

Increasing VDS

<< L
-10 -5 0 5 10

-10

-5

0

5

10

X (nm) --->

X (nm) --->

T=

λ0 + L

λ0

T=
42

λ0 + l

λ0

outline

1) Introduction 2) Generic model of a nanodevice 3) The ballistic MOSFET 4) Scattering in nano-MOSFETs
5) Discussion

6) Summary

43

“Physics of Nanoscale Transistors”
1) 2) 3) 4) 5) 6) 7) Review of MOSFET Fundamentals Elementary Theory of the Nanoscale MOSFET Theory of the Ballistic MOSFET Scattering in Nanoscale MOSFETs Application to State-of-the-Art MOSFETs Quantum Transport in Nanoscale MOSFETs Connection to the Bottom Up Approach

Mark Lundstrom

“Electronics from the Bottom Up” on nanoHUB.org
44

S/D quantum mechanical tunneling

4)

3)

2)

1)

45 from M. Luisier, ETH Zurich

generic model to NEGF Gate

EF1

HE Σ [D(] −[US)]

EF 2

[Γ ]τ/ [Σ ]
1

1

1

τ [Γ ]/ [Σ ]
2

2

2

S. Datta, Quantum Transport: Atom to Transistor, Cambridge, 2005 46 (“Concepts of Quantum Transport” nanohub.org)

randomness is the rule - not the exception! Random dopant fluctuations
side view

top view

nanonets
47

“Percolation in Electronic Devices”

1) Percolation in Electronic Devices 2) Thresholds, Islands, and Fractals 3) Nonlinear Electrical Conduction in Percolative Systems 4) Stick Percolation and Nanonet Electronics 5) 2D Nets in 3D World: Sensors, Solar Cells, and Antennas

M. Ashraf Alam
“Electronics from the Bottom Up” on nanoHUB.org
48

outline

1) Introduction 2) Generic model of a nanodevice 3) The ballistic MOSFET 4) Scattering in nano-MOSFETs 5) Discussion
6) Summary

49

summary

1) The bottom-up view provides a simple, but rigorous approach to nanoelectronics. 2) It’s useful for familiar devices, like MOSFETs. 3) It’s also a good starting point for new devices. 4) You can learn more on nanoHUB.org or by attending the annual “Electronics from the Bottom Up” summer schools at Purdue University.
50

nanoHUB.org

“Electronics from the Bottom Up”

51

Questions & Answers

52

Sign up to vote on this title
UsefulNot useful