TUNNELING

D R A G I C A VA S I L E S K A A N D G E R H A R D K L I M E C K
 QUANTUM EFFECTS

• Quantum-mechanical space quantization

• Tunneling
• Quantum Interference

In all but the smallest devices quantum-mechanical

space quantization effects and tunneling play dominant
role and they can be captured with quantum correction
models

Quantum interference dominates the operation of

resonant tunneling diodes and fully quantum transport
approaches are needed to treat this device.
 TREATMENT OF TUNNELING

• WKB Approximation
• Transfer Matrix Approach
• Piece-Wise Constant Potential Barrier Approximation
• Piece-Wise Linear Potential Barrier Approximations
WENTZEL-KRAMERS-BRILLOUIN
(WKB) APPROXIMATION
 IMPORTANT APPLICATIONS IN WHICH
WKB APPROXIMATION IS USED

• Tunneling Breakdown in normal diodes (reverse

biased diode)
• Tunnel (Esaki) diode (forward + reverse bias)
• Scanning Tunneling Microscope
• Gate Leakage in MOSFET Devices
A. BREAKDOWN MECHANISMS IN A DIODE

• Junction breakdown can be due to:

tunneling breakdown

avalanche breakdown

• One can determine which mechanism is responsible for the

breakdown based on the value of the breakdown voltage VBD :

VBD < 4Eg/q → tunneling breakdown

VBD > 6Eg/q → avalanche breakdown

4Eg/q < VBD < 6Eg/q → both tunneling and
avalanche mechanisms are responsible
• Tunneling breakdown occurs in heavily-doped pn-
junctions in which the depletion region width W is about
10 nm.
Zero-bias band diagram: Forward-bias band diagram:

EFn
EF EFp
EC
EC

EV
EV
W W
Reverse-bias band diagram: • Tunneling current (obtained by
using WKB approximation):
* 3  4 2 m
* 3/ 2 
Eg 
It =
2m q FcrVA 
exp −
2 2 1/ 2
4π
E g  3
qFcr 
 
EF Fcr  average electric field in
p
the junction
EFn
EC • The critical voltage for
tunneling breakdown, VBR, is
estimated from:
EV
I t (VBR ) ∝ 10 I S

• With T, Eg and It .

B. TUNNEL (ESAKI) DIODE
Leo Esaki

Nobel Prize in Physics 1973

 (ESAKI) TUNNEL DIODE (TD)

• Simplest tunneling device

• Heavily-doped pn junction
• Leads to overlap of conduction and valence bands
• Carriers are able to tunnel inter-band
• Tunneling goes exponentially with tunneling distance
• Requires junction to be abrupt
 BAND-TO-BAND TUNNELING IN A
TUNNEL DIODE
(c)
(e)

(b) (d)
V
EC
(a)
EV
EF

(a) (b) (c) (d) (e)

 FIGURES OF MERIT

Peak current Peak-to-Valley Ratio (PVR)

100 kA/cm2

V
DIRECT VS. INDIRECT TUNNELING

Direct Indirect

Indirect materials require phonons to tunnel, thus

reducing the probability of a tunneling event
C. SCANNING TUNNELING
MICROSCOPE
 D. WKB APPROXIMATION
EXPLAINED
• The Wentzel-Kramers-Brillouin (WKB) approximation is a
“semiclassical calculation” in quantum mechanics in
which the wavefunction is assumed an exponential
function with amplitude and phase that slowly varies
compared to the de Broglie wavelength, λ, and is then
semiclassically expanded

• While Wentzel, Kramers and Brillouin developed this

approach in 1926, earlier in 1923 Harold Jeffreys had
already developed a more general method of
approximating linear, second-order differential equations
(the Schrodinger equation is a linear second order
differential equation)
WKB APPROXIMATION EXPLAINED,
CONT’D
• While technically this is an “Approximate Method” not an
“Exact solution” to the Schrodinger equation, it is very
close to simple plane wave solutions that we discussed
while describing transmission coefficient calculation in
piece-wise constant potential barriers

• The WKB method is most often applied to 1D problems

but can be applied to 3D Spherically Symmetric
problems as well (see Bohm 1951)

• The WKB approximation is especially useful in deriving

the tunnel current in a tunnel diode
 BASIC IDEA OF THE METHOD

• The WKB approximation states that since in a constant

potential, the wavefunction solutions of the Schrodinger
equation are of the form of simple plane waves, then
± ikx 2 m( E − U )
ψ ( x) = Ae , k = 2π / λ =

2
• Now, if the potential U=U(x) changes slowly with x, the
solution of the Schrodinger equation can also be written
of the general form
iφ ( x )
ψ ( x) = Ae
where φ(x)=xk(x).
- For the constant potential case, φ(x)=±kx so the
phase changes linearly with x
- In a slowly varying potential φ(x) should vary slowly
from the linear case ±kx
 BASIC IDEA OF THE METHOD,
CONT’D
• For the two cases, E>U and E<U, let k(x) be defined as
(so we only have to solve the problem once)

2m( E − U ( x))
k ( x) = 2
, E > U ( x)


2m(U ( x) − E )
k ( x) = −i 2
= −iκ ( x), E < U ( x)


WENTZEL-KRAMERS-BRILLOUIN (WKB)
APPROXIMATION
• Starting from the 1D Schrödinger equation

2 ∂2
− 2
ψ ( x) + U ( x)ψ ( x) = Eψ ( x)
2m ∂x
• And substituting the general solution for slowly-varying
potentials, one gets the following differential equation
2
∂ φ  ∂φ 
2
i 2 −   + k 2 ( x) = 0
∂x  ∂x 
 WENTZEL-KRAMERS-BRILLOUIN
(WKB) APPROXIMATION
• The WKB approximation assumes that the
potentials are slowly varying in space
• Then the 0th order approximation assumes

∂ 2φ ∂φ0
2
= 0, = ± k ( x) → φ0 ( x) = ± ∫ k ( x)dx +C0
∂x ∂x
→ ψ ( x) = exp  ±i ∫ k ( x)dx +C0 
 
 Wentzel-Kramers-Brillouin (WKB)
Approximation
• If a higher order solution is required, then we
solve
2
∂ φ  ∂φ 
2
2 ∂φ 2 ∂ 2
φ
i 2 −   + k ( x) = 0 → = ± k ( x) + i 2
∂x  ∂x  ∂x ∂x
• Then the 1th order approximation assumes

∂φ ∂k
= ± k 2 ( x) ± i
∂x ∂x
 ∂k 
→ ψ ( x) = exp  ±i ∫ k ( x) ± i dx + C1 
2

 ∂x 
 Wentzel-Kramers-Brillouin (WKB)
Approximation
1. In order to apply the WKB approximation we only
need to know the shape of the potential since
 ∂k 
U ( x) → k ( x) → φ ( x) → ψ ( x) = exp  ± ∫ k ( x) ± i dx + C1 
2

 ∂x 
2. For slowly varying U(x) the first order and the
zero order approximation give almost the same
result as
∂ 2
k ( x) << k ( x)
∂x
 Wentzel-Kramers-Brillouin (WKB)
Approximation
3. The WKB approximation breaks down where E~U
(classical turning points) in which case the wavevector
k(x) approaches zero but the derivative does not and
there in fact the argument in (2) does not hold
∂ 2
k ( x) << k ( x)
∂x
Under these circumstances, connection formulas must
be applied to tie together regions on each side of the
classical turning point.
E. EXAMPLE: GATE LEAKAGE
gate leakage
tunnelling current
 GATE LEAKAGE
 For sub-micrometer devices, due to smaller oxide thickness, there is
significant conductance being measured on the gate contact. The finite
gate current gives rise to the following effects:
 Negative => degradation in the device operating characteristics with
time due to oxide charging; larger off-state power dissipation
 Positive => non-volatile memories utilize the gate current to program
and erase charge on the “floating contact” – FLASH, FLOTOX,
EEPROM
 There are two different types of conduction mechanisms to the insulator
layer:
 Tunneling: Fowler-Nordheim or direct tunneling process
 Hot-carrier injection: lucky electron model or Concannon model

Electron is emitted into the oxide • Similar to the lucky electron model, but
when it gains sufficient energy to assumes non-Maxwellian high energy tail on
overcome the insulator/semicon- the distribution function.
ductor barrier. • Requires solution of the energy balance
equation for carrier temperature.
 TUNNELING CURRENTS
 Three types of tunneling processes are schematically shown below
(courtesy of D. K. Schroder)

Vox < φB
Vox = φB
φB Vox > φB

tox

FN FN/Direct Direct

• For tox ≥ 40 Å, Fowler-Nordheim (FN) tunneling dominates

• For tox < 40 Å, direct tunneling becomes important
• Idir > IFN at a given Vox when direct tunneling active
• For given electric field: - IFN independent of oxide thickness
- Idir depends on oxide thickness
SIGNIFICANCE OF GATE LEAKAGE
 As oxide thickness decreases, the gate current becomes more
important. It eventually dominates the off-state leakage current (ID at
VG = 0 V)
 The drain current ID as a function of technology generation is shown
below (courtesy of D. K. Schroder)

10-4 I on
Current (A/µm)

10-6
10-8 I off
10-10
10-12 IG
10-14
10-16
0 50 100 150 200 250
Technology Generation (nm)
 FOWLER-NORDHEIM TUNNELING

ΦB ΦB - eEx

EF EF
a
x-axis
0 0
No applied bias With applied bias
 The difference between the Fermi level and the top of the barrier is
denoted by ΦB
 According to WKB approximation, the tunneling coefficient through this
triangular barrier equals to:
 a  2m *
T ∝ exp − 2 ∫ γ ( x) dx  where: γ ( x ) =
2
(Φ B − eEx )
 0 

 EG-Ez V(z)
EG
z

EZ

W
W

Transmission coefficient:

−2α
T =e
 FOWLER-NORDHEIM TUNNELING
 The final expression for the
Fowler-Nordheim tunneling
coefficient is:

 4 2m *Φ 3B/ 2 
T ∝ exp − 
 3eE

 Important notes:
 The above expression
explains tunneling process
only qualitatively because
Calculated and experimental tunnel
the additional attraction of current characteristics for ultra-thin oxide
the electron back to the plate layers.
is not included (M. Depas et al., Solid State Electronics, Vol.
38, No. 8, pp. 1465-1471, 1995)
 Due to surface
imperfections, the surface
field changes and can make
large difference in the results
TRANSFER MATRIX
APPROACH
 TUNNELING: TRANSFER MATRIX
APPROACH
Within the Transfer Matrix approach one can assume to
have either
• Piece-wise constant potential barrier
• Piecewise-linear potential barrier
 PIECE-WISE CONSTANT
POTENTIAL APPROXIMATION
Piece-Wise Constant Potential Barrier (PCPBT
Tool) installed on the nanoHUB
The Approach at a Glance
The Approach, Continued

Slide property of Sozolenko.

PIECE-WISE LINEAR POTENTIAL
APPROXIMATION
 PIECE-WISE LINEAR APPROXIMATION

• This algorithm is implemented in ASU’s code for modeling

Schottky junction transistors (SJT)
• It approximates real barrier with piece-wise linear
segments for which the solution of the 1D Schrodinger
equation leads to Airy functions and modified Airy functions
• Transfer matrix approach is used to calculate the energy-
dependent transmission coefficient
• Based on the value of the energy of the particle E, T(E) is
looked up and a random number is generated. If r<T(E)
than the tunneling process is allowed, otherwise it is
rejected.
 The Approach at a Glance
• 1D Schrödinger
Vi+1
equation: Vi
2 2

d Ψ
− + V ( x)ψ = Eψ
2m dx 2 E

• Solution for piecewise Vi-1 V(x)

linear potential: ai-1 ai

ai+1
ψ i = Ci(1) Ai (ξ ) + Ci( 2) Bi (ξ ) 1 r1 ' 1 r1 ' 
[
2 iA (0) + Ai (0)] [Bi (0) + Bi (0)]
ik0 2 ik0
• The total transmission MFI = 
1
[ (0) −
r1 '
(0)]
1
[ (0) +
r

1 B' (0)]
 i A Ai Bi i 
matrix: 2 ik0 2 ik0 
π  rNBi (ξN) +ikN+1Bi (ξN) rNBi (ξN) −ikN+1Bi (ξN) 
' '
M T = M FI M1M 2 ........M N −1M BI MBI =  
rn −r A (ξ ) +ik A (ξ ) −r A (ξ ) −ik A (ξ )
' '
 N i N N+1 i N N i N N+1 i N 

• T(E): k
T = N +1
1
π  rB
Mi = 
' 
i i (ξi ) −Bi (ξi )  Ai (ξi )

Bi (ξi ) 
K0 T 2 
m11 ri −rA' (ξ ) A (ξ ) ri+1Ai' (ξi ) ri+1Bi' (ξi )
 i i i i i 
 Simulation Results for Gate Leakage in SJT

-3
10
Drain current
Gate Current
Tunneling Current
-4
10
Current [A/um]

-5
10

-6
10 T. Khan, D. Vasileska and T. J.
Thornton, “Quantum-mechanical
tunneling phenomena in metal-
semiconductor junctions”, NPMS 6-
-7
10 SIMD 4, November 30-December 5,
0.1 0.2 0.3 0.4 0.5 0.6 0.7 2003, Wailea Marriot Resort, Maui,
Gate Voltage [V] Hawaii.
TSU-ESAKI FORMULA FOR THE
CURRENT CALCULATION
DERIVATION OF THE TSU-ESAKI
FORMULA

Energy barrier with two electrodes which

can be used to describe the ECB or HVB
processes.
 ASSUMPTIONS

• Effective mass approximation. The different masses

corresponding to the band structure of the considered
material are lumped into a single value for the effective mass.
This is denoted by meff in the electrodes and mdiel in the
dielectric layer.
• Parabolic bands. The dispersion relation in semiconductors
is approximated by

2k 2
2
E= =
2meff 2meff
( k x2 + k y2 + k z2 )

• Conservation of parallel momentum. Only transitions in the

kx direction are considered, the parallel wavevector k|| = kyey +
kzez is not altered by the tunneling process.
 CURRENT CALCULATION

• The net tunneling current from Electrode 1 to Electrode 2 can be

written as the net difference between current flowing from Side 1 to
Side 2 and vice versa:

J = J1→2 − J 2→1
• The current density through the two interfaces depends on the
perpendicular component of the wavevector kx, the transmission
coefficient Tc, the perpendicular velocity vx, the density of states gc
and the distribution function at both sides of the barrier:

dJ1→2 = qTc ( k x ) vx g1 (k x ) f1 ( E )[1 − f 2 ( E )]dk x

dJ 2→1 = qTc ( k x ) vx g 2 (k x ) f 2 ( E )[1 − f1 ( E )]dk x
 CURRENT CALCULATION, CONT’D

• In this expression it is assumed that the transmission coefficient only

depends upon the momentum perpendicular to the interface. The
density of states g(kx) is:
∞∞
g (k x ) = ∫ ∫ g (k x , k y , k z )dk y dk z
0 0

• Where g(kx,ky,kz) denotes the three-dimensional density of states in

momentum space. Considering the quantized wavevector
components within a cube of side L yields for the density of states
within the cube:

2 1 1 2π
g (k x , k y , k z ) = = , ∆k i =
L3 ∆k x ∆k y ∆k z 4π 3 L
 CURRENT CALCULATION, CONT’D

• For the parabolic dispersion relation, the velocity and

energy components in the tunneling direction obey:
1 ∂E
k x 1
vx = = , and vx dk x = dEx

∂k x meff

• Hence, the expressions for the current density become:

∞∞
q
dJ1→2 = 3
Tc ( Ex ) dEx ∫ ∫ f1 ( E )[1 − f 2 ( E )]dk y dk z
4π
0 0
∞∞
q
dJ 2→1 = 3
Tc ( k x ) dEx ∫ ∫ f 2 ( E )[1 − f1 ( E )]dk y dk z
4π
0 0
 CURRENT CALCULATION, CONT’D

• Using polar coordinates for the parallel wavevector

components
k ρ = k y2 + k z2

• The current density evaluates to:

4π meff q Emax ∞
J1→2 =
h 3 ∫ T ( E ) dE ∫ f ( E )[1 − f
Emin
c x x
0
1 2 ( E )]dEρ

4π meff q Emax ∞
J 2→1 =
h 3 ∫ T ( E ) dE ∫ f
Emin
c x x
0
2 ( E )[1 − f1 ( E )]dEρ

The total energy is sum of longitudinal part Ex and transverse part Eρ.
 CURRENT CALCULATION, CONT’D

• Evaluating the difference, the net current through the

interface equals:
4π meff q Emax ∞
J = J1→2 − J 2→1 =
h 3 ∫
Emin
Tc ( Ex )dEx ∫ [ f1 ( E ) − f 2 ( E )]dEρ
0

• This expression is usually written as an integral over the

product of two independent parts which only depend
upon the energy perpendicular to the interface: The
transmission coefficient Tc(Ex) and the supply function
N(Ex)
4π meff q Emax
J=
h 3 ∫
Emin
Tc ( Ex ) N ( Ex )dEx
 TSU-ESAKI FORMULA

• The expression in the previous slide is known as the Tsu-

Esaki formula.
• The supply function describes the difference in the supply of
carriers at the interfaces of the dielectric layer. Following the
definition of the current, the supply function is given by:
∞
N ( Ex ) = ∫ [ f1 ( E ) − f 2 ( E ) ]dEρ
0

• The occupancy functions f1 and f2 are defined near the

interfaces. Since the exact shape of these distributions is
usually not known, approximate shapes are commonly used.
Furthermore, it is assumed that the distributions are isotropic.
 SUPPLY FUNCTION

• In equilibrium, the energy distribution function of

electrons and holes is given by the FERMI-DIRAC
statistics
1
f (E) =
 E − Ef 
1 + exp  
k
 B T

• Which can be derived from statistical mechanics.

Separating the longitudinal and the transverse energies
E=Ex+Eρ, and splitting the integral N(Ex)=ξ1(Ex)-ξ2(Ex),
the values of ξ1 and ξ2 become:
∞ ∞
1
ξi = ∫ f i ( E )dEρ = ∫ dEρ , i=1,2
0 0  Ex + Eρ − E fi 
1 + exp  
 k BT 
 SUPPLY FUNCTION

• The last expression can be integrated analytically using:

dx  1 
∫ 1 + exp( x)  1 + exp(− x)  + C
= ln

• Then the total supply function is:

  E f 1 − Ex  
 1 + exp  
  k BT  
N ( Ex ) = k BT ln
  E f 2 − Ex  
 1 + exp   
  k BT  