3 Lamba2012

2nd Reading
February 14, 2013 14:38 WSPC/245-JMM 1250007
Journal of Multiscale Modelling

Vol. 4, No. 2 (2012) 1250007 (9 pages)
c Imperial College Press
DOI: 10.1142/S1756973712500072
SCATTERING IN NANO-FILMS
V. K. LAMBA
Department of Electronics and Communications Engineering
Global College of Engineering and Technology, Khanpur Khui, India
J. Multiscale Modelling 2012.04. Downloaded from www.worldscientific.com
O. P. GARG
Department of Electronics
by FUDAN UNIVERSITY on 05/06/15. For personal use only.
RKSD College Kaithal, India
D. ENGLES
Department of Electronics
Guru Nanak Dev University, Amritsar, India
Published 15 February 2013
In this communication, a quantum mechanical technique for treatment of effects of scat-

tering transport at nanoscale in thin films is discussed. We implemented a rigorous treat-
ment of scattering within the NEGF simulation platform. Results obtained by applying
the rigorous scattering model to simulate the devices were used as a benchmark to vali-
date a simple computationally-efficient, phenomenological treatment of scattering. The
NEGF method is used to study the effect of electron confinement on silicon nano-films
and wires. Electron confinement results in almost a factor of 3 decreases in the electrical
conductivity of the 5 nm silicon film compared to the 10 nm film. Increase in the amount
of confinement also leads to a 35% decrease in the conductivity of a 5 nm × 5 nm wire
compared to the 5 nm film. Our simple model provides an excellent tradeoff between
increased computational cost and the physics of scattering that needs to be captured in
devices of the future.
Keywords: Scattering, NEGF; thin films.
1. Introduction
As real devices operate below ballistic limits due to scattering therefore, we focus
on extending the ballistic NEGF simulation scheme, to include the effects of scat-
tering in nanoscale MOSFETS, nano thin films, and two probe systems. We used
the decoupled mode-space solution1–3 to include the effects of scattering on elec-
tron transport since the scattering within a device can be due to several reasons.
Microscopically, electrons are confined within a very narrow channel and the chan-
nel is sandwiched between insulators in SOI devices. These insulator surfaces are
never perfectly smooth and the semiconductor lattice is never defect free. Therefore,
1250007-1
2nd Reading
V. K. Lamba, O. P. Garg & D. Engles
electrons in the channel experience surface roughness scattering.2–7 Both, channel

carrier densities and impurity concentrations, are typically very large. Therefore,
there is significant electron-electron and electron-impurity scattering within the
device. Moreover, devices typically operate at relatively high temperatures result-
ing in strong phonon-electron interactions. So, all these mechanisms need to be
considered for an accurate treatment of scattering.
Currently, solutions to the Boltzmann equation using Monte Carlo methods
offer the best platform for modeling the effects of scattering in detail. As these
simulation platforms do not treat quantum effects rigorously, because of critical
device dimensions shrunk, quantum effects begin to evident strongly. These effects
need careful consideration because they affect critical device performance. Within
the NEGF formalism, a detailed treatment of the various scattering processes is

possible. Due to the large computational cost involved in a detailed simulation,
we use the insights obtained from detailed simulations to motivate an approximate

treatment of scattering using the concept of Büttiker probes.3–12 Apart from its
computational viability, the main advantages of the phenomenological scattering
model are:
(1) The use of a single parameter to treat scattering;
(2) The existence of an analytic relation between this parameter equivalent low-
field mobility.
It has been shown that Coulomb scattering limits effective mobility at most
effective fields, while surface roughness scattering only limits mobility under very
high internal electric fields.13, 14 In case of nanowires with relatively high doping
and large diameters, controlling Coulomb scattering from ionized dopants should
be given precedence over surface engineering when seeking to maximize nanowire
mobility.13–17
2. Methodology
We briefly review the steps for scattering transport models for, thin film sur-
faces.18, 23–29
(1) A 2D solution to the Schrödinger equation is obtained by solving two 1D prob-
lems, one in the direction normal to the channel (z) in case of thin film, which
yields the electron distribution in the confinement direction and the other, along
the channel/surface direction based on the subband profiles, which yields the
electron concentration in the transmission direction.
(2) The 2D electron density for each subband is distributed over the silicon body
(normal to the film) according to the subband wave-functions.
(3) The 2D Poisson’s equation is solved using this electron density to obtain a
new potential profile. This potential profile is used to solve the Schrödinger
equation once again, and the calculation cycle repeated until convergence is
achieved.19–23
1250007-2
2nd Reading
Scattering in Nano-Films
Here we take 2D film with infinite dimensions along the x and y axes (perpen-
dicular to the current flow) while the thickness along the z axis (in the direction
of current flow) has nanoscale dimensions i.e., 5 and 10 nm. The electrons in a 2D
film can be treated as having infinite range of energies along the x and y axes,
while being confined along the z axis, which is also the direction of transport, so
the energy dispersion relation for a 2D thin film is then given by
2
2 kx2 2 ky2 2 nπ
E(k) = Ec + + + ; n = 1, 2, 3 . . . . (1)
2m∗x 2m∗y 2m∗zx Lz
The term nπ/Lz the wave vector corresponding to the confined electrons that
form discrete standing waves when confined to infinite potential well and n is the
subband index. For the film we use the 2D Fermi function given by6

E−µ
f2D (E) = No log 1 + exp − (2)
kB T
that accounts for the infinite states along the x-y plane. Hence, along the transport
direction, the electron energy is given by7
2
2 nπ
E(k) = Ec + . (3)
2m∗zx Lz
In the NEGF formalism, the energy of each subband is calculated by solving

the eigenvalues of the Hamiltonian, and treating each subbands individually. For
ballistic transport (as no scattering in between energy levels), it allows us to model
electron transport in the device separately for each energy level. Accordingly, a
current–voltage characteristic is calculated for the conduction and each additional
subband separately. The current thus obtained is summed over all the bands to
obtain the net channel current. Writing the Schrödinger wave equation along the
z axis as
[H]ψnz = nz ψnz . (4)
The electron energies along the z axis are obtained by substituting a plane wave
basis for the electron wave function and solving the Schrödinger wave equation
along the z axis using the finite-difference method (FDM).
εn (kz ) = Ec + 2t(1 − cos(knz a)) . (5)
Thus, the value of kz for an isolated channel will be equal to nπ/Lz in corre-
spondence to the standing waves in the channel. When the channel is connected
to the contacts, some of the energy density states from the infinite contacts will
spill over into the channel broadening the energy levels around each energy band in
the channel. Wave vectors associated with the broadened energy levels around each
band can be obtained by solving the energy conservation equation, for example, the
1250007-3
2nd Reading
wave vectors corresponding to the broadened energy states around the conduction
band Ec is

−1 E − Ec
kz a = cos 1− . (6)
2t
Similarly, wave vectors for broadened states around the first subband is obtained
as

E − Esub1
kz a = cos−1 1 − (7)
2t
and so on. For each energy band, the channel current-voltage characteristics are
obtained independently until the contribution by any additional subbands is negli-
gible. The final channel current at a particular voltage is obtained as the sum of the
currents of the conduction band and the subbands. In addition to the source and
drain contacts, electrons can scatter into and out of the channel by either phonon
absorption or phonon emission such that εn − εm = ω. The transition of the elec-
trons from εn to εm and vice versa is dependent on the transition rate S(k, k ) which
is obtained using Fermi’s Golden Rule.8 The coupling can be expressed through a
broadening term Γ which is related to the transition rate S
Γ
= S(k, k ) ,

where
em 2 ab 2
Γnn = 2π|Kmn | (N + 1)δ(εn − εm − ω) + 2π|kmn | (N )δ(εn − εm + ω) . (8)
The broadening term obtained through Fermi’s Golden Rule is similar to the broad-
ening term in the NEGF which is expressed as rate S 1
em p n

d(ω) So (Nω + 1)[G (E − ω) + G (E + ω)]
Γs = (9)
2π +Soab (Nω )[Gn (E − ω) + Gp (E + ω)]
Gn (E) is the electronic density and, Gp (E) is the hole density of states, correspond
to the δ(εn − εm ± ω) terms while So corresponds to the value |Kmn |2 terms for
emission and absorption of phonons in Eq. (8). For the case where electrons in
the channel scatter with a phonon of single frequency, the broadening term can be
simplified to

(Nω + 1)[Gp (E − ω) + Gn (E + ω)]
Γs = So . (10)
+(Nω )[Gn (E − ω) + Gp (E + ω)]
Since the imaginary part of the self energy term is responsible for broadening, the
scattering self-energy Σs can be expressed using Eq. (9) as
Γs
= −i . (11)
s
2
1250007-4
2nd Reading
The value of So for a single phonon of energy ωo is obtained as a sum over all
phonon wave vectors β in the Brillouin zone as
D 2 2 Do2 2 π 3
o
So = = , (12)
2ρΩ(ωo ) 12π 2 ρ(ωo ) a
β
ρ is the mass density, ωo the phonon frequency and Do is the optical deformation
potential and a, the lattice constant. An example of single-phonon scattering is g-
type inter-valley longitudinal optical-phonon scattering of electrons from the [001]
valley into the [001̄] valley in silicon.8 The scattering self energy is included in the
Green’s function as an additional contact that scatters electrons into and out of the
channel such that the net current through the scattering contact is zero.
−1

+
G(E) = (E − i0 )I − H − − − . (13)
1 2 s
The NEGF method is used to calculate the current density of silicon films for
varying thickness under a constant applied field of 106 V/m. The current density is
for each film is calculated for two cases:
(1) Ballistic transport through the film having thickness in nano meters, and;
(2) Electrons scattering with longitudinal optical phonons.
We find the one-electron eigenstates by solving the one-electron Schrödinger equa-
tion
H1el ψα (r) = εα ψα (r) . (14)
To solve this differential equation we expand the wave functions in a set of basis
functions

ψα (r) = Cαi φi (r) . (15)
i
This modifies the differential equation into a matrix equation for determining Cαi

Hij Cαj = εα Sij Cαj . (16)
j j
The Hamiltonian matrix, Hij = φi |H1el |φj , and overlap matrix Sij = φi |φj ,
are given by 3D integrals over the basis functions. Thus the electron density of the
system is given by

2 εα − εF
n(r) = |ψα (r)| f , (17)
α
kT
where f (x) = 1/(1 + ex ) is the Fermi function, εF the Fermi energy, and T the
electron temperature. For semi-empirical (SE) cases, the non-self-consistent part of
the tight-binding Hamiltonian is parametrized using a two-center approximation,
i.e., the matrix elements only depend on the distance between two atoms, and
1250007-5
2nd Reading
is independent of the position of the other atoms. The self-consistent part of the
calculation is identical for both SE models.
The density matrix is calculated from the Hamiltonian using non-equilibrium
Green’s functions for device systems, while for molecules and crystals, it is calcu-
lated by diagonalization. The density matrix defines the real-space electron den-
sity, and consequently the Hartree potential can be obtained by solving the Poisson
equation. Further the current is calculated from the transmission coefficient using

e E − EFL + eVR
I(VL , VR , TL , TR ) = T (E) f
h kb TR
L

E − EF + eVL
−f dE , (18)
kb TL
where f is the Fermi function, and TL and TR are the electron temperatures of
the left and right electrodes, respectively. In this equation, T (E) is the transmis-
sion coefficient. In our calculations, the generalized gradient approximation (GGA)

schemes are used with Perdew–Burke–Ernzerhof (PBE) correlation function. The
Hamiltonian overlaps, and electronic densities are evaluated in a real space grid
defined with a plane wave cutoff of 150 Ry to achieve a balance between calcu-
lation efficiency, and accuracy. Core electrons are modeled with Troullier–Martins
nonlocal pseudo potential, and valence electrons are expanded in a SIESTA local-
ized basis set. The method allows complete self-consistent field treatment of an
electrode-molecule-electrode junction for zero bias as well as for applied biases and
incorporates the effect of the external electric field induced by the electrodes on the
molecule.
3. Results and Conclusions

The result is shown in Fig. 1 and reveals the following fundamental aspects of
electron transport in nanoscale devices:
(1) The calculated current is very small for small film thicknesses and increases as
the film thickness increases. This is the effect of electron confinement where
very small film thickness leads to the formation of discrete subband energies.
LOS Scaering Ballisc

Current Density (A/m2)
2.E+09
2.E+07
2.E+05
2.E+03
2.E+01
2 3 5 8 10
Film Thickness (nm)
Fig. 1. Current density versus thickness of silicon films doped to 1018 cm−3 for LOS scattering,
and ballistic electron transport.
1250007-6
2nd Reading
(2) Scattering in the NEGF model is included for very thin films as they have
limited subbands that have a say to current flow, so the effect of scattering is
limited to few subbands only. As a result, the effect of scattering is less than
8% for film thicknesses up to 5 nm. With the increases of film thickness, the
subband energies become closer to each other, thus increasing the current flow.
As a result, there is a higher probability for the electrons to scatter that begins
to gain significance for film thicknesses greater than 5 nm where the decrease
in current is now approximately 7% of the ballistic current.
(3) In film with thickness between 3 to 7 nm, there is a clear impact of scatter-
ing on the current which demonstrates the non-equilibrium nature of electron
transport. As from literature we know that scattering with phonons is a way of
restoring equilibrium in a system, and if the electrons experience limited scat-

tering with phonons, they are in a state of non-equilibrium and Fermi-Dirac
statistics may no longer be applicable to describe electron distribution.
Figure 2 shows the electrical conductivity values as a function of doping varying

from 1016 to 1020 per cm3 , for 5 nm, 10 nm films, 5 × 5 × 9.2 nm wire, and bulk
Silicon using the NEGF formalism. As the number of bulk dimensions is reduced
from 2D to 1D electron confinement is increased from 1D to 2D as in the case of the
5 × 5 × 9.2 wire causing a 35% decrease in conductivity. Confinement of electrons
increases the spacing between adjacent subband energy levels leading to a quadratic
increase in the electrical conductivity with size. Figure 3 shows the subband energies
of 5 and 10 nm silicon films doped to a concentration of 1018 cm−3 corresponding to
Ec of 0.1863 eV. It can be seen from Fig. 3 that the value of the subband energies for
5 nm film are higher than that of the 10 nm film. For the energy range considered,
the 10 nm film has double the number of electron subbands compared to the 5 nm
film leading to a factor of 3.2 decreases in the conductivity of the 5 nm film compared
to the 10 nm film. Also, from Fig. 4, high doping brings the conduction band edge
closer to the fermi level. As the film thickness increases, the number of energy levels
around the fermi energy increases causing the electrical conductivity of the film to
10 nm film 5 nm film 5 X 5nm wire Bulk Si

E. Conducvity (Mho-m)
1.E+05
1.E+04
1.E+03
1.E+02
1.E+01
1.E+00
1.E-01
1.E+16 1.E+17 1.E+18 1.E+19 1.E+20
Doping (cm3)
Fig. 2. Simulated electrical conductivity values for a ballistic silicon 5 nm, 10 nm film; 5 × 5 nm
wire, and bulk silicon.
1250007-7
2nd Reading
10 nm film 5nm film

Channel Subband Energy (ev) 2
1.5
1
0.5
0
1 2 3 4 5 6 7 8 9 10 11 12
Number of Subbands
Fig. 3. Comparison of subband energies for 10 and 5 nm silicon films.

Conducon band Edge Fermi level = 0.1ev

Conducon band Edge (eV)
0.4
0.3
0.2
0.1
0
1.00E+16 1.00E+17 1.00E+18 1.00E+19 1.00E+20
Doping (Cm-3)
Fig. 4. Change in conduction band edge with doping for silicon (Fermi level = 0.1 eV).
increase with film thickness. In reality and as seen from Fig. 4, as the size of the film
increases, transport is no longer ballistic and the conductivity is expected to reach
a maximum before starting to reduce with increase in size due to the dominance of
scattering.
References
1. W. Fichtner, Report, ETH, Integrated System Laboratory (SFIT Zurich, 2008).
2. S. Datta, R. Venugopal, S. Goasguen and M. Lundstrom, J. Appl. Phys. 95 (2004)
292–305.
3. M. Buttiker, Phys. Rev. Lett. 57 (1986) 1761–1764.
4. M. Svizhenko and M. Anantram, IEEE Trans. Elect. Dev. (2002).
5. A. Texier and M. Buttiker, Phys. Rev. B 62 (2000) 7454–7458.
6. S. Datta, Quantum Transport: Atom to Transistor (Cambridge University Press, New
York, 2005).
7. R. Eisberg and R. Resnick, Quantum Physics of Atoms, Molecules, Solids, Nuclei and
Particles (John Wiley & Sons, 1974).
8. M. Lundstrom, Fundamentals of Carrier Transport (Cambridge University Press,
2000).
9. D. A. Broido and T. L. Reinecke, Phys. Rev. B 51 (1995) 13797–13800.
10. S. Takagi, J. Koga and A. Toriumi, IEEE Trans. Int. Elect. Dev. Meeting (1997),
pp. 219–222.
1250007-8
2nd Reading
11. T. Wang, T. H. Hsieh and T. W. Chen, J. Appl. Phys. 74 (1993) 426–430.

12. V. A. Fonoberov and A. A. Balandin, Nano Lett. 6 (2006) 2442–2446.
13. L. Zhang, L. Chen, H. Liu, Y. Hou, A. Hirata, T. Fujita and M. Chen, Phys. Chem.
C 115 (2011) 19583–19587.
14. J. Hao, M.-J. Han, Z. Xu, J. Li and X. Meng, Nanoscale Res. Lett. 6 (2011) 263.
15. V. K. Lamba, R. S. Sawhney and D. Engles, AIP Conf. Proc. 1349 (2011) 671–672.
16. D. R. Khanal, A. X. Levander, K. M. Yu, Z. Liliental-Weber, J. Grandal, E. Calleja
and J. Wu, J. Appl. Phys. 110 (2011) 033705.
17. J. Grandal, M. A. Sa’nchez-Garcı’a, E. Calleja and J. Wu, J. Appl. Phys. 110 (2012)
033705.
18. D. Vashaee and A. Shakouri, J. Appl. Phys. 95 (2004) 1233–1245.
19. T. W. Tang and B. Wu, Semiconductor Science and Technology 19 (2004) 54–60.
20. S. Datta, Electronic Transport in Mesoscopic Systems (Cambridge University Press,
UK, 1997).
21. A. Svizhenko and M. Anantram, IEEE Trans. Elect. Dev. (2002).
22. S. Datta, J. Phys. Condens. Matter 2 (1990) 8023–8052.
23. J. Fernandez, in Lecture Notes in Computer Science (Springer-Verlag, 2002).

24. S. Datta, Electronic Transport in Mesoscopic Systems (Cambridge University Press,
New York, 1995), see, for example, Chapter 8.
25. V. K. Lamba, D. Engles and S. S. Malik, Molecular Simulation 35 (2009) 434–439.
26. L. E. Hall, J. R. Reimers, N. S. Hush and K. J. Silverbrook, J. Chem. Phys. 112
(2000) 1510.
27. P. S. Damle, A. W. Ghost and S. Datta, Phys. Rev. B 64 (2001).
28. T. Numata, T. Irisawa, T. Tezuka, J. Koga, N. Hirashita, K. Usuda, E. Toyoda, Y.
Miyamura, A. Tanabe, N. Sugiyama and S. Takagi, IEEE Trans. Elect. Dev. 53 (2006)
1030–1038.
29. T. Acosta and S. Sood, IEEE Potentials 25 (2006) 31–34.
1250007-9

3 Lamba2012

Uploaded by

Copyright:

Available Formats

You might also like

3 Lamba2012

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

3 Lamba2012

Uploaded by

Copyright:

Available Formats

2nd Reading

February 14, 2013 14:38 WSPC/245-JMM 1250007

Journal of Multiscale Modelling

RKSD College Kaithal, India

Published 15 February 2013

In this communication, a quantum mechanical technique for treatment of eﬀects of scat-

Keywords: Scattering, NEGF; thin ﬁlms.

V. K. Lamba, O. P. Garg & D. Engles

electrons in the channel experience surface roughness scattering.2–7 Both, channel

the NEGF formalism, a detailed treatment of the various scattering processes is

we use the insights obtained from detailed simulations to motivate an approximate

In the NEGF formalism, the energy of each subband is calculated by solving

[H]ψnz = nz ψnz . (4)

εn (kz ) = Ec + 2t(1 − cos(knz a)) . (5)

V. K. Lamba, O. P. Garg & D. Engles

V. K. Lamba, O. P. Garg & D. Engles

sion coeﬃcient. In our calculations, the generalized gradient approximation (GGA)

3. Results and Conclusions

LOS Scaering Ballisc

restoring equilibrium in a system, and if the electrons experience limited scat-

statistics may no longer be applicable to describe electron distribution.

Figure 2 shows the electrical conductivity values as a function of doping varying

10 nm ﬁlm 5 nm ﬁlm 5 X 5nm wire Bulk Si

V. K. Lamba, O. P. Garg & D. Engles

10 nm ﬁlm 5nm ﬁlm

Fig. 3. Comparison of subband energies for 10 and 5 nm silicon ﬁlms.

Conducon band Edge Fermi level = 0.1ev

11. T. Wang, T. H. Hsieh and T. W. Chen, J. Appl. Phys. 74 (1993) 426–430.

23. J. Fernandez, in Lecture Notes in Computer Science (Springer-Verlag, 2002).

You might also like