Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
University of Bologna
Leveraging on Nano-CMOS
Competences for Diverse
Applications: the Case for
Modeling an Simulation
C. Fiegna1, G. Baccarani1, V. Vyurkov2, I.
Semenikhin2, C. Berti1, M. Zanuccoli1
1 ARCES - University of Bologna, Italy
2 IPT Russian Academy of Sciences Moscow, Russia
University of Bologna
Motivation CMOS Technology has been relentlesslydeveloped during the last 35 years withunprecedented records in terms of bothqualitative and quantitative technologicaladvancement and of pervasive impact onthe human society .It has also significantly stimulated thedevelopment of scientific know -how thatcan be exploited in close -by fields, helpingtheir development in the frame of abeneficial synergy with micro - and nano -electronics itself.
University of Bologna
Contents
This presentation will discuss two examples in the field of modeling and numerical simulation, namely:
1. knowledge and mathematical know -howdeveloped for the quantum -mechanicalsimulation of nano-scale CMOS devices can beexploited to tackle the problem of opticalsimulation for solar cells and optical sensors;
2. the know -how developed in the field of particlebased simulation of transport in small-geometryMOSFETs significantly eases the development ofnumerical simulators for ionic and proteintransport in nano-scale biological systems that,in turn, can be integrated within Si technology.
University of Bologna
Outline
� 3-D Quantum Mechanical Simulation of Nano-MOSFETs vs. Electro-Magnetic simulation
• The transverse mode representation
• Applications to
�nMOS/pMOS I-V characteristics
�Analysis of a super-steep subthreshold slope MOSFET
• Analogies to RCWA EM simulation
�Application to solar-cell analysis
� From Monte Carlo Device Simulation to Ionic Transport through Biological and Synthetic Nano-Scale Channels
University of Bologna
1. Starting from advanced CMOS: the Nano -transistor.
x
University of Bologna
Main strategy of simulation
�Self-consistent solution of
Schrödinger equation
+ Poisson equation.
The main problem here is to solve the 3D Shrödinger equation.
University of Bologna
Solution of 3D Schrödinger equation
U(x,y,z) is the potential inside the channel.Transport along x; confinement on the y-z plane.
2 2 2
2 2 2( , , )
2 2 2x y z
U x y zm x m y m z
∂ ∂ ∂= − − − +∂ ∂ ∂
H
( , , ) ( , , )x y z E x y zψ ψ=H
Where
H is an Hermitian operator.
The direct solution of the stationary 3D Schrödinger equation via a finite difference scheme comes across a well known instability caused by evanescent modes.
University of Bologna
Transverse mode representation
1
( , , ) ( ) ( , )M
i ii
x y z c x y zψ ϕ=
=∑Whereφi(y,z) is the i-th transverse mode wave function,M isthe number of involved modes. The specific set of modewavefunction {φ(y,z)} depends on the detailed mathematicalformulation of the simulation method.Toghether with arbitrary precision arithmetic it circumventsthe evanescent modes problem.
The wave-function is expanded into the sumof x-propagatingtransversal modes
University of Bologna
Channela+
a-
b+
b-
b
b
a
a +
−
+
−
=
T
a
b
b
a
++
− −
=
S
Usually we know a+, b-. Coefficients a-, b+ and function inside the channel are unknown. To solve this problem we can use either T-matrix or S-matrix formulation.
The coefficients ci(x) can be expressed in terms of amplitudes of incoming and outgoing waves
( ) ( ) ( ), 1..i i ic x c x c x i M+ −= + =in vector notation: ( ) ( ) ( )x x x+ −= +c c c
University of Bologna
Example of Transfer-matrix calculation scheme
1, 1,1
1, 1, 2, 2,1
2, 2,1
( ) ( , ), 0
( , , ) ( , , ) ( , , ), 0
( ) ( , ),
n n
n n
Nik x ik x
n n nn
N
n n n n xn
Nik x ik x
n n n xn
c e c e y z x
x y z d x y z d x y z x L
c e c e y z x L
ϕ
χ χ
ϕ
−+ −
=
=
−+ −
=
+ ≤Ψ = + ≤ ≤
+ ≥
∑
∑
∑
whereχn(x,y,z) satisfies the Schrödinger equation withthe following boundary conditions:
1, 1,0
2, 2,0
, 0
0,
Canal
Canal
n n nx x L
n n nx x L
χ ϕ χ
χ χ ϕ= =
= =
= =
= =
University of Bologna
By matching the wave function and its derivative at the left boundaryand the right boundary of the channel we get:
left right
left right
+ +
− −
=
c cT
c c
This scheme has no restrictions on channel length nor potential. It can be easily generalized to N- junction devices with N leads and a coupling region of an arbitrary shape.
The scheme is described in detail in: J. T. Londergan, J. P.Carini, D. P. Murdock, “Binding andscattering in two-dimensional systems: applications to quantum wires,waveguides, and photonic crystals” Springer, 1999.
University of Bologna
Outline
� 3-D Quantum Mechanical Simulation of Nano-MOSFETs vs. Electro-Magnetic simulation
• The transverse mode representation
• Applications to
�nMOS/pMOS I-V characteristics
�Analysis of a super-steep subthreshold slope MOSFET
• Analogies to RCWA EM simulation
�Application to solar-cell analysis
� From Monte Carlo Device Simulation to Ionic Transport through Biological and Synthetic Nano-Scale Channels
University of Bologna
0.0 0.1 0.2 0.30.00
0.25
0.50
0.75
1.00T
rans
ition
Coe
ffic
ient
E, [eV]
pFET No Impurities One Impurity
channel thickness 3 nm
Example: hole transmission in channel
(10nm x 10nm x 5nm)
University of Bologna
Transfer characteristics (log)
(10nm x 10nm x 5nm)
Sub-threshold swing is 71 mV per decade of current.
-0,50 -0,25 0,00 0,25 0,50
10-7
10-6
10-5
10-7
10-6
10-5
Dra
in C
urre
nt, [
A]
Gate Voltage, [V]
Drain Voltage = 1 V nFET pFET
University of Bologna
Output characteristics(10nm x 10nm x 5nm)
-1,0 -0,5 0,0 0,5 1,00,0
4,0x10-6
8,0x10-6
1,2x10-5
0,0
4,0x10-6
8,0x10-6
1,2x10-5
Dra
in C
urre
nt, [
A]
Drain Voltage, [V]
nFET pFET
University of Bologna
Outline
� 3-D Quantum Mechanical Simulation of Nano-MOSFETs vs. Electro-Magnetic simulation
• The transverse mode representation
• Applications to
�nMOS/pMOS I-V characteristics
�Analysis of a super-steep subthreshold slope MOSFET
• Analogies to RCWA EM simulation
�Application to solar-cell analysis
� From Monte Carlo Device Simulation to Ionic Transport through Biological and Synthetic Nano-Scale Channels
University of Bologna
Application: Superlattice-Based Steep-Slope Switch
Advanced Research Center on Electronic Systems (ARCES),Department of Electronics (DEIS) – University of Bologna, Italy
University of Bologna
University of Bologna
Outline
� Superlattice-Based Steep-Slope Switch
� Optimization of the subband structure
� Superlattice-based nanowire FET
� Semiconductor pairs for the superlattice
� Discussion
G. Baccarani
University of Bologna
Device concept
OFF STATE ON STATE
transverse quantization
1D transport problem
Semiclassical and Q.M. turn-on characteristics
University of Bologna
How to filter out high -energy electrons?
superlattice structure between the source and the channel
M. Bjoerk et al., United States Patent Application Publication, no. US 2009/0200540 A1, 13 August 2009.
Krönig-Penney
University of Bologna
Self consistent simulation
� Self-consistent solution of the open-boundary Schrödinger-Poisson problem.
� Effective mass approximation with cylindrical coordinates.
� Every region characterized by its specific transport mass, dielectric constant and electron affinity.
� Energy-adaptive mesh in order to achieve an accurate description of the resonant states generated within the superlattice.
D
C
S
G
G. Baccarani
University of Bologna
Superlattice minibands
Two regions where the transmission probability is close to one
They correspond to the minibands given by the Krönig-Penney model.
This very simple model provides surprisingly good results, despite the inherent assumption of an infinite number of spatial periods.
G. Baccarani
University of Bologna
Superlattice Steep -Slope FET: Discussion
� We have presented an investigation on a novel device concept meant to achieve a steep subthreshold slope by filtering out the high-energy electrons entering the device.
� The filtering function is entrusted to a superlattice in the source extension region.
� The structure could possibly be fabricated by deposition of a number of appropriate semiconductor layers within a manufacturing process of vertical nanowires.
� Simulation results indicate that an SS = 26 mV/dec can be achieved using GaAs/AlGaAs as the constituent materials of the superlattice.
� Major improvements are possible with the appropriate selection of the semiconductor pair, e.g. the GaN/AlGaNsystem
G. Baccarani
University of Bologna
Outline
� 3-D Quantum Mechanical Simulation of Nano-MOSFETs vs. Electro-Magnetic simulation
• The transverse mode representation
• Applications to
�nMOS/pMOS I-V characteristics
�Analysis of a super-steep subthreshold slope MOSFET
• Analogies to RCWA EM simulation
�Application to solar-cell analysis
� From Monte Carlo Device Simulation to Ionic Transport through Biological and Synthetic Nano-Scale Channels
University of Bologna
From Quantum Mechanics for CMOS to
Advanced Optical Simulation
University of Bologna
2.Light in Nano-Scale Optoelectronics:the Macroscopic Maxwell Equations
0
0t t
ρ∇⋅ = ∇⋅ =∂ ∂∇× + = ∇× − =∂ ∂
B D
B DE H J
Assume: D(r ) = ε0ε(r )E(r ), B(r ) = µ0µ(r )H(r ), µ(r )≈1
H(r , t) = H(r )e−iωt, E(r , t) = E(r )e−iωt. ρ=0, J=0
[ ]( ) 0 ( ) 0ε∇⋅ = ∇⋅ =H r r ESo
University of Bologna
0 0( ) ( ) 0 ( ) ( ) ( ) 0i iωµ ωε ε∇ × − = ∇ × + =E r H r H r r E r
21
( ) ( )( ) c
ωε
∇ × ∇ × =
H r H rr
If ε(r ) have no imaginary part, the operator Ξ:
1ˆ ( ) ( )( )ε
Ξ = ∇ × ∇ ×
H r H rr
is Hermitian operator.
University of Bologna
This equation is fully analog to Schrodinger equation. So, to solve Maxwell equations with this conditions we can apply the same formalism and methods as for the Schrodinger equation and vice versa. This fact is widely used for theory and modeling photonic crystals. See for example:
Joannopoulos J., Johnson S.G., Winn J.N., Meade R.D., Photonic Crystals. (Second edition), Princeton Univ. Press, Princeton (2008).
If ε(r ) have nonzero imaginary part, the operator Ξ is not Hermitianbut most of the techniques from quantum theory still work with minor changes. For example T-matrix or S-matrix formalism.
University of Bologna
Advanced Optical Simulation: Introduction• Photon management required to enhance the absorpt ion
• Advanced optical modeling to deal with:
• Multi-layer thin structures (10nm÷1000nm)
• Nano-metrics rough interfaces
1 µm
� Near field optics requires rigorous approaches thatimplies the solution of the Maxwell equations.
� Trade-off between accuracy and computationalresources.
� RCWA (Rigorous Coupled Wave Analysis) leads toefficient and numerically stable solvers of Maxwellequations.
I. Semenikhin, M. Zanuccoli, V. Vyurkov, E. Sangiorgi, C. Fi egna, “ EfficientImplementation of the Fourier Modal Method (RCWA) for the OpticalSimulation of Optoelectronics Devices”, IWCEI 2010
University of Bologna
Applications: Triangular Groove (2 -D)
1
TE, E2
TE, Abs
4TM, Abs
3TM, H
λ=738nm, Mx = 30
University of BolognaPISA - IWCE 2010 31
Applications: triangular groove (2D periodic texturing for solar cells)
0 20 40 60 801.0
1.1
1.2
1.3
1.4 TE TM
Rel
ativ
e A
bsor
ptio
n
Facet angle, [degrees]
• Dependence of relative absorption on facet angle α at λ=738nm.
University of Bologna
Rough Interface Simulation
� 2-D geometry
� Gaussian distribution of heights
� Heigh (R.M.S.) = 50nm, Correlation Length = 75nm
c-Si