Simplified Model for Ballistic Current-Voltage

Characteristic in Cylindrical Nanowires

Oka Kurniawan1 , Man-Fai Ng1 , Wee Shing Koh1 , Zuan Yi Leong1 ,
Erping Li1
Institute of High Performance Computing
Singapore, 138632

Abstract
The ballistic regime gives the upper limit of an electron device performance.
This paper proposes a fast and efficient model for calculating the current-
voltage characteristic of a cylindrical nanowire within the framework of the
Non-equilibrium Green’s Function. Under certain assumptions, the calcula-
tion is simplified to a one dimensional problem and the modes due to the
radial confinement are given by an analytical equation. We further derive
an analytical expression for the current-voltage characteristic at tempera-
ture approaching zero Kelvin. The relationship between the radius of the
nanowire and the electrical current is clearly shown in this expression. The
effects of the radius on the current-voltage characteristic curve are also stud-
ied. Furthermore, we plot the trend of the saturation current as the radius
is increased as predicted by both the numerical result and our analytical
model. Our proposed model can be further used to include electron-photon
interaction in the calculation of nanoscale optoelectronic devices.
Keywords:

1. Introduction
Semiconducting nanowires are among the most promising alternatives for
future nanoelectronics and optoelectronics [1, 2]. The diameter of fabricated
nanowires is controlled by the size of the catalyst, and the length is propor-
tional to growth time. Moreover, specific dopants can be incorporated into

Email address: kurniawano@ihpc.a-star.edu.sg (Oka Kurniawan)

Preprint submitted to Microelectronics Journal January 26, 2010

nanowires to tune their electronic properties. Our capability to control these
properties is essential for the development of electronic and optoelectronic
nanodevices.
Nanowires have been fabricated with sizes within the nanometer range.
For example, silicon nanowires with 1-2 nm diameters have been fabricated
[3]. Moreover, the lengths of the nanowires are predicted to shrink as well
[4]. At these length scales for the nanowires, ballistic calculations predict an
upper limit to the device performance. The need for theoretical modeling
and calculation then becomes essential. Theoretical calculations help us to
design the properties of nanowires and to understand the physics behind the
measurement results that are obtained.
In the calculation of nano-scale devices properties, the Non-equilibrium
Green’s function (NEGF) framework has been widely used [5]. By using
NEGF, the phase breaking phenomena such as electron scatterings due to
phonon or photon can be included simply through its self-energy matrices.
In this NEGF framework, the device region is discretized to form the Hamil-
tonian matrix. The size of the matrix can be extremely huge for two- or
three-dimensional calculations. In order to increase the computational effi-
ciency, a mode space approach has been proposed [6, 7]. In this approach,
the wavefunctions are expanded in the mode space. As a result, the trans-
port calculation can often be simplified to a one-dimensional problem along
the transport direction [8].
Even in this mode space approach, we still need to calculate the Schrödinger
equation on every slice perpendicular to the transport direction. To simplify
the calculation, Wang et al. proposed a fast uncoupled method where only
one slice is computed [9]. Perturbation theory is used to obtain the potential
at every grid along the transport direction.
In this work, we further simplify the Non-equilibrium Green’s function
formulation for nanowires. We use an analytical solution to approximate the
solution of the Schrodinger equation perpendicular to the transport direction.
This increases the speed significantly since the numerical calculation is now
only required along the transport direction. The benefits that we achieve
from this simplification is faster calculation for predicting nanowire design.
Our model is able to handle larger nanowire size which would take very long
computational time when using the atomistic approach.
Another necessary step for nanowire design is knowing the effects of cer-
tain parameters on the transport properties. Such parameters include the
diameter of the nanowires and their bandstructure. Yu et al. compared their

2
Figure 1: Schematics of cylindrical nanowire. The source contact is on the
left hand side while the drain is on the right hand side. R and L is the radius
and the length of the nanowire respectively.

ballistic model with experimental results for increasing values of nanowire

diameter [4]. It was shown that the current increases as the radius increases.
In this work we will provide a simple expression that relates the current and
the nanowire diameter under certain assumptions.
It is important to note that the calculation in this paper is made for
nanowires without gate electrode. We assume that the nanowire is isolated
in vacuum and hydrogenated so that the wavefunction will diminish rapidly
when going to vacuum. Applications of these kind of nanowires for optoelec-
tronic comprises solar cells and photodetectors [10, 11, 12].
We will present the formulation of the Non-equilibrium Green’s Function
calculation for the structure shown in Fig. 1. The source contact is con-
nected to the left hand side of the nanowire, and the drain is connected to
the right hand side. In this work, we assume that the contacts are ohmic,
and no barriers are present. For simplicity, we study only the ballistic trans-
port. As mentioned previously, ballistic calculations predict an upper limit
to the device performance. Incorporating phase breaking phenomena into
the calculation will be considered in future works.
If we assume that the potentials in the radial direction do not vary sig-
nificantly, the eigenstates are given by an analytical equation which can be
readily obtained. We will then use this simplified model to compute the
current-voltage characteristic and the transmission curve of the nanowire.
It turns out that this simplified model gives an excellent agreement with a
previously published result.
To understand the effects of the diameter size of the nanowires on the bal-
listic current-voltage characteristic, we derived an analytical equation for an
ideal case of the nanowire shown in Fig. 1 at low temperature. The equation
can be expressed in terms of the electrochemical potential of the contacts
and the radius of the nanowires. In this way, the relationship between the

3
diameter size of the nanowire and its ballistic current is clearly shown.
We will also show the effects of the diameter size of the nanowire on
the current-voltage characteristic curve of the ballistic cylindrical nanowire.
It will be shown that the saturation current is linearly proportional to the
saturation voltage. Moreover, our analytical model predicts pretty well the
trend of the saturation current as the nanowire diameter increases.

2. Ballistic Non-equilibrium Green’s Function Formulation

2.1. Hamiltonian in Cylindrical Coordinate
In this section, we present the formulation of the Non-equilibrium Green’s
function calculation for the cylindrical nanowire shown in Fig. 1. We start
with the Hamiltonian

H3D Ψ(r, θ, z) = EΨ(r, θ, z) (1)

where the wavefunction Ψ is a function of the radial distance r, its angle
θ, and the distance z perpendicular to r. The solution of this eigenvalue
problem can be written as

Ψ(r, θ, z) = Cφ(r, z)eivθ (2)

where C is just a normalization constant and φ(r, z) is the wave function
in two dimensions. Note that the angular wavefunction must satisfy the
following condition:

Ψ(r, θ + 2vπ) = Ψ(r, θ) (3)

where v in (2) and (3) is an integer. This is basically a periodic boundary
condition for the angular wavefunction. Now, let’s return to the Hamiltonian
which is given by

~2 2
H=− ∇ +V (4)
2m∗
where m∗ is the effective mass, and V is the potential. For simplicity in
writing the equations, here we assume that the effective mass is uniform.
The laplacian in cylindrical coordinate is given by [13, 14]

2∂2 1 ∂ 1 ∂2 ∂2
∇ = 2+ + + (5)
∂r r ∂r r 2 ∂θ2 ∂z 2

4
 Substituting (4), (5), and (2) into (1), it can be shown that

~2 1 ∂φ(r, z) ∂ 2 φ(r, z) v 2

− ∗ + − 2 φ(r, z)
2m r ∂r ∂r 2 r
2

∂ φ(r, z)
+ + V (r, z)φ(r, z) = Eφ(r, z) (6)
∂z 2
The wavefunction φ can now be expanded in terms of the eigenfunctions
of the Schrödinger equation in the transverse direction.
X
φ(r, z) = ϕn (z)ξ n (r; z) (7)
n

where ξ n (r; z) is the eigenstate solution of

 2 n
~2 1 dξ n v 2 n

dξ
− ∗ + − 2 ξ + V (z0 )ξ n = Esub
n
(z0 )ξ n (8)
2mr dr 2 r dr r
where m∗r is the effective mass in the radial direction. The above equation is
to be solved on the planes perpendicular to the transport direction. Solving
n
this equation gives the eigenstates ξ(r; z0) and the eigenenergies Esub (z0 ).
In this derivation we assume that the modes in different values of z are
uncoupled.
We can substitute (7) into (6), and then multiply by ξ m and integrate
over r to give

~2 ϕn
m nd
2
Z X Z X
n n
ϕ Esub ξ m ξ n dr − ξ ξ dr
r n
2m∗z r n dz 2
Z X
=E ϕn ξ m ξ n dr (9)
r n

where we have use (8). Note that

Z
ξ m ξ n dr = δm,n (10)
r

so that the equation now becomes

~ 2 d 2 ϕm m
− + Esub ϕm = Eϕm (11)
2m∗z dz 2

5
It is important to note that the superscript m is just an integer index and is
different from the effective mass along the transport direction m∗z . We can
see that the resulting Hamiltonian becomes block diagonal, where for mode
m, the hamiltonian is given by
~2 d 2 m
hmm = − + Esub (12)
2m∗z dz 2
The Green’s function is simply calculated from the following relation:

G = (ES − hmm − Σ1 − Σ2 )−1 (13)

where S is the overlap matrix, and Σi is the self-energy matrix to take into
account contact i. In this calculation we assumed that the i = 1 for the
source contact, and i = 2 for the drain contact.
Note that we have reduced the problem to a one dimensional problem
along the nanowire axis, i.e. z. Moreover, the modes are decoupled, which
can be seen clearly from the block diagonal structure of the Hamiltonian
matrix. In the following section we will make some assumptions to obtain
m
an expression for Esub .

2.2. Solving the Radial Schrödinger Equation

Now, the next thing to do is to solve (8). This is where we will make some
assumptions to simplify the calculations. For a cylindrical nanowire without
a gate electrode, we assume that the potential does not vary significantly
inside the nanowire along the radial direction. Moreover, it is assumed that
the wavefunction terminates at the edge of the nanowire, i.e. at r = R.
With this assumption, the potential profile along the radial direction can be
simplified to a square infinite potential well. This assumption is reasonable
since it was shown that the lattice is almost empty of charge for low gate
biases [15], and the structure that is considered here does not have a gate bias
at all. In this case the potential inside the well is not modified significantly
from the square potential shape. We will further justify this by showing
the results of calculating the energy subbands and the potential from a self-
consistent Schrödinger-Poisson solver. With this assumption, we can write
the potential profile as follows.


0 for r ≤ R
V (r) = (14)
∞ for r > R

6
 The solution of (8) for the given potential in (14) is given by the Bessel
Function of the First Kind, Jv , where the quantum number v is the order
of the Bessel Function and is the same as that found in (2) and (3) [16].
Now, to satisfy the condition that the wavefunction is zero at the edge of the
nanowire, we must have

Jv (znv ) = 0 (15)

where znv is the nth root of the Bessel function of order v, and is related to
the energy as
r
nv
Esub 2m∗r
znv = knv R = R (16)
~2
In this equation, knv is the wave number and is related to the energy through
a parabolic dispersion relation E(k) = ~2 k 2 /2m∗ .
The benefit of making the assumption of an infinite potential well pro-
file is that we obtain an analytical solution for the eigenenergies and the
eigenstates. The eigenenergies that solve the radial Schrödinger equation are
simply given by (16), or can be rewritten as
2 2
nv knv ~
Esub = (17)
2m∗r

It is significant to realize that the roots of the Bessel function are known
once the two quantum numbers n and v are given. Table 1 gives the tabulated
values of the roots of the Bessel function of the First Kind for several values
of n and v.
The four lowest subband energies obtained from solving the radial Schrödinger
equation are given by these quantum numbers (n, v): (1, 0), (1, 1), (1, 2), and
(2, 0). These correspond to the roots znv = 2.4048, 3.8317, 5.1356, and 5.5201
(Table 1).
We have compared the results of obtaining the subband energies using
the Bessel Function approach with that from a self-consistent Schrödinger-
Poisson solver of Atlas, a device simulator from Silvaco [17]. In our Atlas
simulation an oxide was added at the outer layer of the nanowire to simulate
a finite barrier. The Atlas simulation gave us the value of 1.11 eV for the
first subband while our Bessel Function approach gave the value of 1.15
eV. These values are obtained by using the default bulk effective mass in

7
 Table 1: Roots of the Bessel Function of the First Kind, Znv
n=1 n=2 n=3 n=4 n=5
v=0 2.4048 5.5201 8.6537 11.7920 14.9310
v=1 3.8317 7.0156 10.1730 13.3240 16.4710
v=2 5.1356 8.4172 11.6200 14.7960 17.9600
v=3 6.3802 8.7610 13.0150 16.2230 19.4090
v=4 7.5883 11.0650 14.373 17.6160 20.8270

4
3.5
3
Potential (eV)

2.5
2
1.5
1
0.5
0
-0.5
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Distance (nm)

Figure 2: Self-consistent potential along the nanowire radial direction at

different z location calculated using ATLAS. The nanowire is biased with
V=1.0v. The arrow indicates the direction of increasing potential from the
anode due to the applied bias. The potential profile can be approximated as
an infinite potential well along the radial direction.

Atlas. The material used in the simulation was silicon. The small difference
indicates that our assumption of infinite potential well is justified for undoped
nanowires in calculating the subband energies. Fig (2) also shows that the
potential inside the nanowire can be approximated to zero as shown in Eq.
(14).
We can summarize the calculations as follows. We first determine how
many modes or subband energies are to be used in the calculation. The
energies are then obtained from (17) and (16) as well as Table 1. Once the
subband energies are known, the Hamiltonian matrix can be constructed from
(12), and the Green’s function is calculated using (13). With the Green’s
function obtained, the other parameters such as the current, transmission

8
 0

-0.2

Potential (eV)
-0.4

-0.6

-0.8

-1
0 20 40 60 80 100
Position (nm)

Figure 3: Potential along the nanowire axis.

curve, electron density, etc, can be computed.

3. Current-Voltage Characteristic
We work out the calculation using silicon material parameters. The liter-
ature provides quite a number of non-equilibrium Green’s function simulation
for nanowires, but most of them are for field effect transistors with gate elec-
trodes. But, the case we are studying is for a single nanowire suspended in
vacuum which is commonly used for optoelectronic devices. Current-voltage
characteristic for this kind of silicon nanowires has been previously calculated
using atomistic simulation by one of the authors [18]. In this paper, we will
obtain the current-voltage characteristic using the simplified model. In order
to obtain the nanowire current-voltage characteristic, the device shown in
Fig. 1 is biased from 0 V up to 1 V. The source contact is kept at ground
potential, while the drain contact on the right is lowered by the bias. It
is assumed that the potential falls linearly near the contacts and constants
along the nanowire. This non-self-consistent calculation follows that in [19].
Fig. 3 shows the potential applied along the nanowire axis.
The current is then calculated from

∞
2q
Z
I= T (E)[f1 (E) − f2 (E)]dE (18)
h −∞

where h = 2π~, T (E) is the transmission coefficient, fi (E) is the Fermi

function at the contact i, where i = 1 is for the source contact, and i = 2

9
 7

Current (µA)
4

0
0 0.2 0.4 0.6 0.8 1
Voltage (V)

Figure 4: Current-voltage characteristic of the computed cylindrical nanowire

with 3 nm diameter and 100 nm length. We use m∗z = m∗t = 0.19m0 and
m∗r = 0.316m0 .

is for the drain contact. The transmission coefficient is calculated from the
Green’s function of equation (13) as follows.

T (E) = Trace[Γ1 GΓ2 G† ] (19)

where

Γi = [Σi − Σ†i ] (20)

and we use the superscript † to indicate the conjugate transpose of a matrix.

The diameter of the nanowire was set to 3 nm while the length was set
to 100 nm. It has been shown that effective mass approximation is valid for
calculating nanowires with diameter down to about 3 nm [20, 21]. Following
the procedure in [22], we adjusted the Fermi level in the contacts so that the
saturation current was about 6 µA. This method gave us a Fermi level value
of around 0.39 eV. The result is shown in Fig. 4. Our simplified model has
the same profile trend as the results plotted in Fig. 5 (a) of [22].
Two effective mass values should be used for anisotropy bandstructure
material such as silicon. One is for the radial direction and the other one
is for the transport direction. In silicon, the radial effective mass can be
approximated from the transverse and longitudinal effective masses following
the approach used in Atlas, Silvaco. For spherically symmetric bandstructure
such as GaAs only one effective mass value is needed.

10
 3

2.5

Transmission (E)
2

1.5

0.5
VD = 1 V
0 VD = 0 V
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Energy (eV)

Figure 5: Transmission curve at 0 V and 1V biases. The arrow indicates the

position of the Fermi level. The transmission curve has a step-like shape. The
jumps in the curve occur at the subband energies. Note that the transmission
curve does not change significantly under bias.

For this calculation, therefore, we set m∗z = m∗t , which is the transverse
effective mass in silicon and is about m∗t = 0.19m0 . The radial effective mass,
on the other hand, was set as m∗r = 2m∗l m∗t /(m∗l + m∗t ). The longitudinal
effective mass of silicon is m∗l = 0.916m0 . The approximation for the radial
effective mass has been used in the literature [17, 13, 14].
In the current calculation we used two subbands, though effectively only
one subband was needed. The lowest calculated subband energies are around
0.31 eV, 0.79 eV, and 1.42 eV. Since the Fermi level is about 0.39 eV, only
one subband is occupied.
Fig. 5 shows the transmission curve at two biases. Note that the trans-
mission curve does not change when the drain contact is biased. The reason
for this is explained in [8]. The transport for a ballistic device is mainly
determined by the transport at the top of the barrier [23, 24, 25, 26]. Since
there is no barrier present along the nanowire, the top of the barrier for
our device is simply the potential at the source. When the drain contact is
biased, the potential at the drain is lowered by the bias voltage. However,
the potential at the source remains the same. Hence, the transmission curve
does not change at all. The only change is that the Fermi function at the
drain shifts to lower energy levels.
With this understanding, we can further explain the shape of the I-V
curve in Fig. 4. To help in the elaboration, we refer to (18). It appears that
the ballistic current is simply the area under a curve. The curve is obtained

11
 (a) (b)

(c) (d)

Figure 6: Plots of the Fermi functions and the transmission curve. (a) Plot of
the two Fermi functions when the drain is biased. The drain Fermi function
is shifted to the left by the bias voltage. (b) Plot of the difference between
the two Fermi functions. (c) Plot of the transmission curve. (d) Plot of the
term inside the integral in (18). The current is the shaded area under the
curve. We can visualize the current calculation by looking at these figures in
sequence from (a) to (d).

by multiplying the transmission coefficient by the difference of the two Fermi

functions. Fig. 6 shows the current calculation visually.
Note that the only curve that changes is the Fermi function at the drain
contact. This is because the source is kept at the ground potential, and
the transmission curve does not change under bias for this ballistic nanowire
(Fig. 5). So now, the calculation of the current becomes simple and straight-
forward. In the next section, we will derive a simple analytical equation
that relates the radius of the nanowire to the electrical current under certain

12
assumptions.

4. Analytical Equation
In this section we derive an analytical equation for the electrical current
which relates the current to the radius of the nanowire. It is assumed that
the temperature is zero Kelvin. The reason for this assumption is to simplify
the expression since the Fermi function at this temperature can be expressed
by a step function as shown in Fig. 6. In addition, we will only consider one
subband in the derivation. The extension to more subband will be elaborated
at the end.
The analytical equation is derived by considering different regions and
cases in Fig. 6. Let us start with the case when the lowest energy subband
E1 is higher than the electrochemical potential at the source, i.e. E1 > µ1 .
In this case, there is no overlap region between the two Fermi functions and
the transmission curve. In other words, there is no shaded area in Fig. 6d,
and therefore, the current is zero for all bias.
Now consider the case when E1 < µ2 , which means that the lowest sub-
band is lower than the electrochemical potential at the drain. For this case,
the shaded area is fixed and is given by the difference between the two Fermi
functions. Since the magnitude of the Fermi function is unity for energies
lower than the Fermi level, the shaded area is simply given by the width of
the area shown in Fig. 6b. This width is simply the bias voltage VD . Hence
for this region, we can write the current as
2q
I= VD (21)
h
The equation above gives us the linear portion in Fig. 4, which happens
when the drain is biased at a very low voltage. In a latter part, we will give
the expression for the voltage where the current starts to saturate.
In the last case is when the lowest energy subband is between the elec-
trochemical potential at the source and the drain, i.e. µ2 < E1 < µ1 , as is
shown in Fig. 6d. Note that as the drain is biased, the shaded area does
not change. This is because the transmission curve and the source Fermi
function do not change under bias. Hence, the current for this region is a
constant and is given by the width of the shaded area in Fig. 6d. The width
is given by the difference between the electrochemical potential at the source

13
and the first lowest subband energy.
2q
I= (µ1 − E1 ) (22)
h
From (16), (17), and Table 1, we can write the first subband energy as

2.40482 ~2
E1 = (23)
2m∗r R2

Hence, the current in (22) can be expressed in terms of the electrochemical

potential and the radius of the nanowire.
We can summarize the ideal current equation as follows:


 0 , E1 ≥ µ1
2q
I= h D
V , E1 ≤ µ2 (24)
2q
(µ1 − E1 ) , µ2 < E1 < µ1

h

where E1 is given by (23).

What happens if there are more than one subband involved in the trans-
port? We must compute the currents due to each subband using (24) and
sum them up. For example, when a second subband is involved, (24) is used
to calculate I2 by replacing E1 with E2 . The constant in (23) is also replaced
with 3.8317 according to Table 1. Then the total current is simply given by
M
X
Itotal = Ii (25)
i=1

where M is the last subband with energy lower than µ1 , and the current for
each contribution is given by replacing E1 with Ei in (22).

2q
Ii = (µ1 − Ei ) (26)
h
and Ei is the ith subband energy.
Fig. 7 shows the current-voltage characteristic calculated using the an-
alytical model (24) with the same parameters as those used in the previous
section. The agreement between the analytical model and the numerical
calculation is not surprising. The analytical equation is able to model the

14
 7

5 7
6
4

I (µA)
5
3 4
0.05 0.1 0.15 0.2
2

1 Analytical
Numerical
0
0 0.2 0.4 0.6 0.8 1
VD (V)

Figure 7: Current-voltage characteristic calculated using the analytical model

in comparison with the result obtained numerically. We use the same pa-
rameters for both calculation. Note that the analytical model can be used to
model the ballistic nanowire. The inset shows the region where the current
starts to saturate. The difference could be caused by the finite temperature
assumed in the analytical model.

current-voltage characteristic accurately in the linear and in the saturation

regions. The only region where the analytical model differs from the numer-
ical calculation occurs where the current starts to saturate.
The reason for this discrepancy seems to be caused by the assumption
of zero Kelvin temperature in the analytical model. When the temperature
is finite, the Fermi function is no longer a step function. This causes the
numerical results to have a lower value of current as the drain Fermi function
passes through the lowest subband energy E1 .
The point where the current starts to saturate can be obtained by equat-
ing (21) and (22).
2q 2q
I= VS = (µ1 − E1 ) (27)
h h
where VS is the drain voltage at which the current starts to saturate. Solving
the above equation gives

VS = µ1 − E1 (28)

Hence, VS depends on the electrochemical potential as well as on the lowest

subband energy. The lowest subband energy, in turn, depends on the radius
of the nanowire. Recall from (23) that the energy subband is proportional

15
to the inverse square of the nanowire radius. This means that for the same
electrochemical potential level at the source, VS decreases as the radius of
the nanowire decreases.
We can also write the saturation current in terms of this voltage VS . From
(27),

2q
IS = VS (29)
h
This means that the saturation current is proportional to the voltage VS .
Hence, as the radius of the nanowire is decreased, VS and IS will decrease.
Note that the analytical equations presented here are similar to those
in Ref. [27] for low-temperature limits. The derivation of our expression,
however, shows how the radius of the nanowire affects the current.

5. Electrical Current Dependencies on Nanowire Radius

In the previous section, we have derived an analytical equation for the
ballistic current in cylindrical nanowires. Equations (24) and (23) show that
for a given Fermi level in the contact, the current depends on the radius of
the nanowire as well as on the effective mass. In this section, results for
varying the radius of nanowires are presented.
Equation (23) shows that the subband energy depends on the radius of
the nanowires. More exactly, the subband energy is inversely proportional
to the square of the radius of nanowires. This means that as the radius de-
creases, the subband energy increases. An increasing subband energy results
in smaller VS and IS values as shown in (28) and (29).
Fig. 8 (a) shows the current-voltage characteristic when the diameter of
the nanowire is varied. It appears that the current decreases as the diameter
of the nanowire decreases. This is consistent with our previous discussion.
It is interesting to note that the slope in the linear region remains the same
for various values of diameter. From (21), the slope is obviously given by
I 2q
= = G0 (30)
VD h
where G0 is the quantum conductance.
The increase in electrical current can also be explained in this way. As
the subband energy is lowered by increasing the nanowire radius, the trans-
mission curve shifts to the left, and the shaded area in Fig. 6 (d) increases.

16
 50 D = 7 nm

40

Current (µA)
D = 5 nm
30

20

10 D = 3 nm

0
0 0.2 0.4 0.6 0.8 1
Voltage (V)

(a)

10-3
Numerical
Analytical

10-4
Current (A)

10-5

10-6
3 4 5 6 7 8 9 10
Nanowire Diameter (nm)

(b)

Figure 8: (a) Current-voltage curve for three different diameter values. The
current saturates after a certain bias point. (b) The trend of the saturation
current values with respect to the nanowire radius. The data in lines are
calculated using the analytical model while the data in points are calculated
numerically using the simplified model of cylindrical Non-equilibrium Green’s
function. The analytical model gives a good prediction for the saturation
current values.

This, in turns, causes the increase in the current. Since the first subband
energy is the lowest energy where electron can flow, it acts like a barrier for
the electron coming from the source that has energy lower than this first
subband. When the electron barrier decreases, the current starts to flow.
A closer look at Fig. 8 (a) suggests that the current-voltage characteristic
resembles that of a transistor. Indeed, a transistor current is controlled by a
gate voltage which controls the barrier at the channel [28, 23].
Fig. 8 (b) shows the trend of the saturation current values with respect to

17
Figure 9: Schematic of the saturation current calculation for non-zero tem-
perature. The total area of the integratino remains the same due to the
symmetric shape of the Fermi function. The decrease due to the top shaded
area is compensated by the increase due to the bottom shaded area. In this
schematic, only one fermi function is shown.

the nanowire diameter. We also plot the values predicted by the analytical
model. It can be seen that the analytical model predicts pretty well the
saturation current computed numerically. It is to be noted that the analytical
model was derived by assuming zero Kelvin temperature. On the other hand,
the non-equilibrium Green’s function simulation use a room temperature
value. The good prediction of the analytical model is not surprising though.
The reason is that the fermi function is symmetric about the fermi energy.
Hence, even for non-zero temperature, the integration area for the saturation
current calculation remains the same. This is shown in Fig. 9 where the two
shaded areas are equal.
In this calculation, we use up to eight subband energies in both the nu-
merical and the analytical calculation. This is sufficient for the diameter
sizes considered in Fig. 8 (b). As the diameter size increases, the subband
energies decreases and more subbands contribute to electrical current. Using
(17), we compute that the highest subband energy that is still below the
source electrochemical potential is the eight subband, which is about 0.34
eV (Recall that the µ1 = 0.39 eV). Fig. 8, hence, shows the validity of the
analytical solution amidst its simple expression.

6. Conclusion
In this work we have derived a formulation for calculating the ballistic
transport in cylindrical nanowires using a simplified model. The result of

18
this simplification is that the eigenstates due to the confinement along the
radial direction are given by an analytical equation. The three-dimensional
problem of a cylindrical nanowire is reduced to a one-dimensional problem
along the transport direction.
To understand further the ballistic transport in cylindrical nanowires, we
also derived an analytical equation for the current-voltage characteristic by
assuming zero Kelvin temperature. The current is expressed in terms of
known parameters such as the electrochemical potential of the contacts and
the radius of the nanowires. In this expression the relationship between the
current and the radius of the nanowires is clearly shown. The current work
then discusses the effect of the radius on the current voltage characteristic.
Interested readers can obtain the software for the calculation from the
author. In a normal modern PC, the code computes the Non-equilibrium
Green’s function in a few minutes for 31 bias points, and the analytical
equations in just a few seconds. The work presented in this paper yields a
fast and efficient model for designing nanowires, and is useful before a more
elaborate and accurate simulation is performed. The source code used in this
paper can be obtained for free from the author.
We intend to include the interaction between electron and photon into
the calculation. When phase-breaking phenomena is introduced, the Non-
equilibrium Green’s Function calculation generally takes a huge amount of
time. Our proposed model, then, is able to reduce the computational burden
of simulating nano-scale optoelectronic cylindrical nanowires.

References
[1] X. Duan, Y. Huang, Y. Cui, J. Wang, C. M. Lieber, Indium phosphide
nanowires as building blocks for nanoscale electronic and optoelectronic
devices, Nature 409 (6816) (2001) 66–69.

[2] Y. Li, F. Qian, J. Xiang, C. M. Lieber, Nanowire electronic and opto-

electronic devices, Materials Today 9 (10) (2006) 18–27.

[3] D. D. D. Ma, C. S. Lee, F. C. K. Au, S. Y. Tong, S. T. Lee, Small-

Diameter Silicon Nanowire Surfaces, Science 299 (5614) (2003) 1874–
1877.

[4] B. Yu, L. Wang, Y. Yuan, P. M. Asbeck, Y. Taur, Scaling of nanowire

19
 transistors, IEEE Transactions on Electron Devices 55 (11) (2008) 2846–
2858.

[5] S. Datta, Nanoscale device modeling: the green’s function method, Su-
perlattices and Microstructures 28 (4) (2000) 253–278.

[6] R. Venugopal, Z. Ren, S. Datta, M. S. Lundstrom, D. Jovanovic, Sim-

ulating quantum transport in nanoscale transistors: Real versus mode-
space approaches, Journal of Applied Physics 92 (7) (2002) 3730–3739.

[7] M. Luisier, A. Schenk, W. Fichtner, Quantum transport in two- and

three-dimensional nanoscale transistors: Coupled mode effects in the
nonequilibrium green’s function formalism, Journal of Applied Physics
100 (4) (2006) 043713.

[8] O. Kurniawan, P. Bai, E. Li, Ballistic calculation of nonequilibrium

greens function in nanoscale devices using finite element method, Jour-
nal of Physics D: Applied Physics 42 (2009) 105109.

[9] J. Wang, E. Polizzi, M. Lundstrom, A three-dimensional quantum sim-

ulation of silicon nanowire transistors with the effective-mass approxi-
mation, Journal of Applied Physics 96 (4) (2004) 2192–2203.

[10] B. Tian, X. Zheng, T. J. Kempa, Y. Fang, N. Yu, G. Yu, J. Huang,

C. M. Lieber, Coaxial silicon nanowires as solar cells and nanoelectronic
power sources, Nature 449 (7164) (2007) 885–889.

[11] C. Soci, A. Zhang, B. Xiang, S. A. Dayeh, D. P. R. Aplin, J. Park,

X. Y. Bao, Y. H. Lo, D. Wang, Zno nanowire uv photodetectors with
high internal gain, Nano Letters 7 (4) (2007) 1003–1009.

[12] M. D. Kelzenberg, D. B. Turner-Evans, B. M. Kayes, M. A. Filler, M. C.

Putnam, N. S. Lewis, H. A. Atwater, Photovoltaic measurements in
single-nanowire silicon solar cells, Nano Letters 8 (2) (2008) 710–714.

[13] E. Gnani, S. Reggiani, M. Rudan, G. Baccarani, A new approach to the

self-consistent solution of the Schrodinger-Poisson equations in nanowire
MOSFETs, in: Solid-State Device Research conference, 2004. ESS-
DERC 2004. Proceeding of the 34th European, 2004, pp. 177–180.

20
[14] E. Gnani, S. Reggiani, M. Rudan, G. Baccarani, Effects of the Band-
Structure modification in silicon nanowires with small diameters, in:
Solid-State Device Research Conference, 2006. ESSDERC 2006. Pro-
ceeding of the 36th European, 2006, pp. 170–173.

[15] N. Neophytou, A. Paul, M. S. Lundstrom, G. Klimeck, Bandstructure ef-

fects in silicon nanowire electron transport, IEEE Transactions on Elec-
tron Devices 55 (6) (2008) 1286–1297.

[16] L. Wang, D. Wang, P. M. Asbeck, A numerical schrodinger-poisson

solver for radially symmetric nanowire core-shell structures, Solid-State
Electronics 50 (11-12) (2006) 1732–1739.

[17] SILVACO, ATLAS users manual (2006).

[18] M. Ng, L. Zhou, S. Yang, L. Shen, V. B. C. Tan, Geometry dependent

I-V characteristics of silicon nanowires, Nano Letters 8 (11) (2008) 3662–
3667.

[19] A. Svizhenko, M. P. Anantram, Effect of scattering and contacts on

current and electrostatics in carbon nanotubes, Phys. Rev. B 72 (8)
(2005) 085430. doi:10.1103/PhysRevB.72.085430.

[20] A. J. Read, C. C. O. U. K. A. Cavendish Laboratory, University of Cam-

bridge, First-principles calculations of the electronic properties of silicon
quantum wires, Physical Review Letters 69 (8) (1992) 1232.

[21] J. Wang, A. Rahman, A. Ghosh, G. Klimeck, M. Lundstrom, On the

validity of the parabolic effective-mass approximation for the i-v calcu-
lation of silicon nanowire transistors, IEEE Transactions On Electron
Devices 52 (7) (2005) 1589–1595.

[22] A. Svizhenko, P. W. Leu, K. Cho, Effect of growth orientation and

surface roughness on electron transport in silicon nanowires, Physical
Review B 75 (2007) 125417.

[23] A. Rahman, J. Guo, S. Datta, M. S. Lundstrom, Theory of ballistic

nanotransistors, IEEE Transactions on Electron Devices 50 (9) (2003)
1853–1864.

21
[24] J.-H. Rhew, Z. Ren, M. S. Lundstrom, A numerical study of ballistic
transport in a nanoscale mosfet, Solid-State Electronics 46 (11) (2002)
1899–1906.

[25] M. Lundstrom, Z. Ren, Essential physics of carrier transport in

nanoscale mosfets, IEEE Transactions on Electron Devices 49 (1) (2002)
133–141.

[26] M. Lundstrom, Device physics at the scaling limit: what matters?, Elec-
tron Devices Meeting, 2003. IEDM ’03 Technical Digest. IEEE Interna-
tional (2003) 33.1.1–33.1.4.

[27] K. Natori, Compact modeling of ballistic nanowire mosfets, IEEE Trans-

actions on Electron Devices 55 (11) (2008) 2877–2885.

[28] S. M. Sze, Physics of semiconductor devices, 2nd Edition, Wiley-

Interscience, New York, 1981.

22