2

COMSOL Multiphysics
Software: A review on its
applications
 30

2.1 Overview
This chapter presents a short introduction of the COMSOL Multiphysics
software and the basic methodology to work in COMSOL desktop. It presents a
literature survey on application of COMSOL Multiphysics software for nanoscale
modeling and simulation. The merits and demerits of COMSOL Multiphysics
software along with justification of its application in the present work are also
presented in this chapter.

2.2 Introduction
Computer simulation has become an essential part of science and
engineering. When developing new products or optimizing designs, digital analysis
of components is very important. Today a broad spectrum of options for simulation
is available and the researchers are using everything from basic programming
languages to various high-level packages implementing advanced methods for
investigation of different structures to explore its application in various fields.
Though each of these techniques has its own unique attributes, they all share a
common concern whether the obtained results by these simulation techniques are
reliable or not. A computer simulation environment is simply a translation of real-
world physical laws into their virtual form [1]. A software is said to be reliable when
a model accurately depicts what happens in the real world.

2.3 Basics of COMSOL

COMSOL Multiphysics is a cross-platform finite element analysis, solver and
multiphysics simulation software allowing conventional physics-based user
interfaces and coupled systems of partial differential equations (PDEs) [2].
COMSOL was started in July 1986 by Svante Littmarck and Farhad Saeidi at the
Royal Institute of Technology (KTH) in Stockholm, Sweden as FEMLAB which was
renamed as COMSOL Multiphysics on 5th October, 2005. Finite Element Method
based COMSOL Multiphysics software is a flexible platform that allows users to
model all relevant physical aspects of their designs & to develop customized
solutions applicable to unique circumstances. With this kind of all-inclusive
modeling environment, COMSOL Multiphysics software gives scientists and
 31

engineers the confidence to build different models with real-world precision [3].
Certain characteristics of COMSOL become apparent with use and compatibility is
one such. COMSOL strictly requires that every type of simulation included in the
package has the ability to be combined with any other which actually mirrors what
happens in the real world. For instance, in nature electricity is always accompanied
by some thermal effect; the two are fully compatible [1]. Another noticeable feature
of COMSOL is adaptability. During the modeling when inclusion of another physical
effect is needed, user can just add it or when any of the model input requires a
formula, it can just be entered using tools like parameterized geometry, interactive
meshing, and custom solver sequences. The ability of coupled multiphysics modeling
depicting the real world problems guarantees the reliability of COMSOL
Multiphysics software which has made it one of the most valuable software tools for
engineers and scientists for solving all kinds of scientific and engineering problems.
The ability to test out various geometrical and physical characteristics of a model
using COMSOL Multiphysics software helps to minimize the design challenges and
optimized model can be taken to the production level.

2.3.1 COMSOL Desktop

COMSOL Desktop is a complete and integrated environment for modeling

and simulation of different devices and to investigate the underlying physics of the
systems with the provided tools necessary to build a user-friendly interactive
interface for the models. COMSOL Desktop helps to organise the simulation by
presenting a clear overview of the model at any point which can be customized
according to user’s own needs. The windows can be resized, moved, docked, &
detached and when the session is closed, any changes make to the layout can be
saved which will be available again the next time when COMSOL Multiphysics is
opened. Additional windows and widgets are added when a model is built. An image
of COMSOL desktop with different labelled parts is shown in figure 2.1.
 32

Figure 2.1: Screen shot image of COMSOL desktop [1].

The Model Builder and the Application Builder are the two main components
of COMSOL Desktop environment. In the Model Builder window, every step of the
modeling process, from defining components, global variables & solving steps to the
generation of final report of results, is displayed and presented as a model tree which
act as the flowchart of a model. The model tree reflects the underlying data structure,
the model object, which stores the state of the model including settings for:
geometry, mesh, physics, boundary conditions, studies, solvers, post processing, and
visualizations. A model tree always has a root node (initially labelled Untitled.mph
{model1}), a Global Definitions node, and a Results node. A specialized user
interface is used to quickly create an application in the Application Builder. An
application is based on a model created with the Model Builder.

To create a new model the basic steps to be followed are given below.
1. In the model wizard, by opening a blank model and right clicking the root
node, model component of a certain space dimension (3D, 2D, 2D axis
symmetric, 1D, 1D axis symmetric & 0D) is added first.
 33

2. In the Global Definitions node parameters, variables, functions such as

interpolation, waveform, analytic etc., couplings are defined. These are used
to define the values and functional dependencies of material properties,
forces, geometry, and other relevant features. These sub nodes of Global
Definitions node has plenty of settings according to the model requirements.
3. Right clicking the geometry node, the different required shapes for a system
design can be selected to build the model. As an alternative to building the
geometry in COMSOL, one can import geometry from a Computer Aided
Design (CAD) package. One can use interactive drawing to create a geometry
using the drawing tools from the Work Plane tab in the ribbon while pointing
and clicking in the Graphics window.
4. Materials node stores material properties that can be referenced in the
Component nodes of the model. The functional dependencies of material
properties of a particular domain can be incorporated by the user defined
functions in the material setting window.
5. Right clicking the component node, from the Add Physics window, the
appropriate module for a particular problem is selected. The specific
boundary condition can be selected for the boundary value problems by right
clicking the selected physics modules.
6. Finite element meshing of the modelled geometry is the next step. The mesh
settings determine the resolution of the finite element mesh used to discretize
the model. The finite element method divides the model into small elements
of geometrically simple shapes. A model component can contain different
meshing sequences for generating meshes with different settings. Normally
most of the geometries contain small edges and faces, and thus to improve
computational accuracy, a slightly finer mesh than the default setting is used.
7. By right clicking the root node, Add Study is selected which can be added at
any time in the model tree. In the Study setting window, different Preset
studies such as frequency domain, stationary, time dependent , small signal
analysis (frequency domain) and under the custom studies empty study, Eigen
frequency and Eigen value studies are available from where one or multiple
study types can be selected for different coupled problems.
 34

8. By clicking the compute button in the study setting window, computation is

processed.
9. The results are analysed and obtained in the Result node. After the post
processing of obtained results, final report can be generated and printed it out.

Results visualization can be exported to a number of common image file

types .jpg, .jpeg, .png, .tif, .tiff, .gif & .bmp. Images can also be read and used as
interpolation functions for physics modeling. Animations can be exported to one of
the Animated GIF, Adobe Flash, and AVI file types.

The equations that COMSOL solves are displayed in the Equation section of
the Settings windows of the respective physics nodes. A model can contain multiple
Components & Study nodes and all nodes can be renamed to be descriptive of their
individual purposes. During the solution process, Convergence plots are generated
and are available from tabs next to the Graphics window. These plots show the
convergence progress of the different solver algorithms engaged in the Study. At any
time, users can save their model using any file name as an .mph file and later it can
be opened in exactly the same state in which it was saved.

The MATLAB model files, which have the extension .m, contain a sequence
of COMSOL commands as a MATLAB® M-file. Users can run the model files in
MATLAB® like any other M-file scripts. It is also possible to edit the files in a text
editor to include additional COMSOL commands or general MATLAB® commands.
Running model files in the M-file format requires the COMSOL LiveLink™ for
MATLAB®. Just like MATLAB LiveLink, COMSOL model files can be run in
COMSOL LiveLink™ for SOLIDWORKS which allow users to build simulation
apps based on SOLIDWORKS geometry enabling import of geometry and
synchronize changes back to the CAD software. In fact, users will be able to control
the CAD design parameters within the simulation apps themselves and alternatively,
SOLIDWORKS users will also be able to interface with COMSOL within the CAD
design environment. In addition to these softwares, COMSOL interfaces with many
other softwares like Inventor, Revit, PTC Pro/Engineer, Excel and Solid Edge.
 35

2.3.2 Boundary –value problems

Boundary value problems arise in the mathematical modeling of physical
systems and a typical boundary-value problem can be defined by a governing
differential equation in a domain Ω as

‫ ߔܮ‬ൌ ݂ (2.1)
together with the boundary conditions on the boundary Γ that enclose the domain.
‫ ܮ‬is a differential operator, ݂ is the excitation or forcing function & ߔ is the
unknown quantity.
The boundary conditions range from the simplest Dirichlet and Neumann
conditions, to complicated impedance and radiation conditions, to even more
complicated higher-order conditions [4]. Robin boundary condition which is a
mixture of both Dirichlet and Neumann conditions is the third type boundary
condition. Robin boundary conditions are also called impedance boundary
conditions, from the point of view of their application in electromagnetic problems,
or convective boundary conditions, from their application in heat transfer problems
[5].

2.3.3 Radiation Conditions

A domain is called unbounded or open when the outer boundary of the
domain recedes to infinity. At this outer boundary a condition must be specified to
obtain a unique solution for the problem which is referred to as a radiation
condition.

Assuming that all sources and objects are immersed in free space and
located within a finite distance from the origin of a coordinate system, the electric
and magnetic fields are required to satisfy [4]

ሬԦ ሬԦ
Ž‹௥՜ஶ ‫ ݎ‬൤‫ ׏‬ൈ ൬ ‫ ܧ‬൰ ൅ ݆݇଴ ‫ݎ‬
ෝ ൈ ൬ ‫ ܧ‬൰൨ ൌ0 (2.2)
‫ܪ‬ሬԦ ‫ܪ‬ሬԦ
 36

where‫ ݎ‬ൌ ඥ‫ ݔ‬ଶ ൅ ‫ ݕ‬ଶ ൅ ‫ ݖ‬ଶ . Equation (2.2) is usually referred to as the Sommerfield
radiation condition for general three-dimensional fields and is one of the first
transparent boundary conditions formulated for wave-type problems. This
boundary condition is more formally called the first-order scattering boundary
condition (SBC) and is trivial to implement within COMSOL Multiphysics
software. A significant limitation to this condition is that it is only non-reflecting if
the incident radiation is exactly normally incident to the boundary. Any wave
incident upon the SBC at a non-normal incidence will be partially reflected. To
simulate an antenna accurately, a more transparent boundary condition is required.

2.3.4 Perfectly Matched Layer

When solving open-region scattering/radiation problems using Finite
Element Method based software, the infinite region exterior to the scatterer/radiator
must be truncated with an artificial boundary to limit the size of the computational
domain. Consequently, at this artificial boundary, a boundary condition must be
introduced for a finite element solution such that it minimizes the nonphysical
reflections from that boundary making the boundary appear as transparent as
possible to the scattered/radiated fields.

In three dimensions, in a homogeneous space exterior to the

scatterer/radiator, the scattered/radiated field satisfies the vector wave equation

‫ ׏‬ൈ ‫ ׏‬ൈ ‫ܧ‬ሬԦ െ ݇ ଶ ‫ܧ‬ሬԦ ൌ Ͳ (2.3)

and the Sommerfield radiation condition at infinity. According to Wilcox [6], such
a field can be represented by a convergent infinite series

௘ షೕೖೝ ஺೙ ሺఏǡఝሻ
‫ܧ‬ሬԦ ሺ‫ݎ‬ሻ ൌ σఈ௡ୀ଴ (2.4)
௥ ௥೙

where (r,ߠǡ ߮) are the usual spherical coordinates.

 37

Berenger [7] in 1994 proposed the concept of a perfectly matched layer

(PML) as a boundary which can be applied to a wide frequency band. The PML
can be used to truncate the computational domain for an antenna design using
COMSOL Multiphysics software. The key property of a PML that distinguishes it
from an ordinary absorbing material is that it is designed so that waves incident upon
the PML from a non-PML medium do not reflect at the interface — this property
allows the PML to strongly absorb outgoing waves from the interior of a
computational region without reflecting them back into the interior. Perfectly
matched layer is not a boundary condition, but rather a domain that can be added
along the exterior of the model that should absorb all outgoing waves. A PML is an
interface between two half spaces, one of which is lossy, and it does not reflect a
plane wave for all frequencies and all angles of incidence and polarizations. The
PML interface is widely used to study high frequency antenna design and the
agreement between computed and measured results are excellent [8].This is the
same as the operation of anechoic chambers in practical measurements, where the
accuracy of the results depends primarily on the absorbing properties of the
material[4].
In continuum space, PML is reflection less but this is not the case in the
discretized space. Since in finite element method based COMSOL Multiphysics
software, computation is carried out in the discretized space, care must be taken to
reduce the unwanted reflection due to discretization. A straight forward approach to
reduce unwanted reflection is to reduce the size of mesh elements. Another
approach is to make the PML parameters change smoothly along the PML
thickness so that the change from grid to grid is small. Thus to obtain best
performance, optimization of PML is necessary.
To reduce the numerical error associated with the artificial reflection of the
PML, the best option is to employ complementary perfectly matched layer [4].In
this case a pair of PMLs having same material composition is used where one PML
is backed by a perfect electric conductor (PEC) and another one is backed by a
perfect magnetic conductor (PMC). From the earlier reported study [9], it is found
that by using complementary perfectly matched layer the error can be almost
reduced to obtain an accurate result. In COMSOL Multiphysics software,
 38

complementary perfectly matched layer can be used as a scattering boundary

condition which makes it more realistic to study high frequency antenna design.

2.3.5 Domain Discretization

The discretization of a domain is one of the most important step of any finite
element analysis because the accuracy of a model, computer storage requirements,
the computation time is directly related to the finite element mesh that is used to
discretize a domain. The finite element mesh is used to subdivide the model into
smaller domains called elements (figure 2.2), over which a set of equations are
solved.

Figure 2.2: Node placement and geometry for 1D, 2D and 3D linear elements [4].

Though smaller elements result in more accurate solutions, when the

individual mesh elements start to get very small, a model run to its limits of
numerical precision i.e., the meshes are smaller than can be accurately represented
on a computer which is an inherent problem with all computational methods[10].
But COMSOL Multiphysics software addresses this via Adaptive Mesh
Refinement, which first solves on an initial mesh and iteratively inserts elements
into regions where the error is estimated to be high, and then re-solves the model.
 39

This can be continued for as many iterations as desired with triangular elements in
2D and tetrahedrals in 3D. Customization of the meshing sequence can help to
reduce memory requirements by controlling the number, type, and quality of
elements, thereby creating an efficient and accurate simulation [4].
A basic requirement of the discretization is that there should be no
overlapping or gaps between elements. The second order tetrahedral element used
for three dimensional problems in the COMSOL software give accurate results
[11]. Modifying meshing sequences is a powerful tool for greatly reducing mesh
elements and thus minimizing computational time. Swept meshing is a powerful
technique for minimizing the computational complexity of many classes of
COMSOL Multiphysics models.

2.4 Merits and demerits of COMSOL

Merits:
1. COMSOL features a great deal of flexibility when dealing with any physical
problem. The greatest of them all is the ability to define user’s own physics,
in the form of Partial Differential Equations for any given problem.
2. Comsol is a multiphysics software which can couple & solve almost every
engineering problems.
3. It solves the coupled problem in acceptable run time accurately.
4. It solves the coupled problem for sample of arbitrary shape and size.
COMSOL software offers many boundary conditions that can be used in
lieu of explicitly modeling thin layers of materials.
5. The COMSOL users have complete control on their model. It is possible to
incorporate the user defined parameters and variable definitions, which can
be coupled to COMSOL's predefined physics interfaces.
6. COMSOL adopts finite element meshing along with customizable and
interactive adaptive mesh refinement tool.
7. It offers fairly accurate electromagnetic simulation and it has the capability
to couple the electromagnetic problem to any other problems.
8. COMSOL Desktop has high resolution graphics window.
 40

9. COMSOL supports most forms of parallel computing including shared

memory parallelism for multicore processors and high performance
computing (HPC) for clusters and clouds. All COMSOL licenses are
multicore enabled.
10. It creates a reusable part that can serve as a building block of a more complex
COMSOL model geometry, which is stored in a separate file accessible
through the Part Libraries.
11. It provides the ability to solve multiphysics phenomena accurately.
12. Ability to define any type of physics problems with modification in
governing equations makes COMSOL very suitable for complex scientific
analysis viz. for nanoscale device modeling and design.
Demerits:
1. Refining the mesh size to improve computational accuracy in COMSOL
always involves some sacrifice in speed and typically requires increased
memory usage.
2. Lower upper (LU) factorization is one of the numerical methods used by
COMSOL for solving the large sparse matrix equation system generated by
the finite element method. A Direct solver is a fast and very robust type of
solver that requires little or no manual tuning in order to solve a wide range
of physics problems. The drawback is that it may require large amounts of
RAM.

2.5 Applications of COMSOL Multiphysics software in nanodomain

Jin Liu et al. [12] in 2006 adopting COMSOL Multiphysics software

presented a model for quantifying oxygen diffusion behavior in a nanodevice
fabricated using individual ZnO nanowire/nanobelt.
Xugang Xiong et al. [13] in 2007 adopting COMSOL multiphysics software
simulated the nanowire assembly processes for rapid assembly and organization of
ultra-small gold nanoparticle networks.
Xugang Xiong and Ahmed Busnaina [14] in 2008 showed that using micro
and nanoscale templates, various nanoparticles can be directly assembled into
 41

parallel wires, cross-wires, and many other complex structures using COMSOL
Multiphysics software.
A. Nojeh et al.[15] in 2008 explored the possibility of having an on-chip
wireless communication infrastructure using carbon nanotube antennas operating in
optical frequencies, and the effect of variations in nanotube properties on the
communication behavior adopting COMSOL multiphysics software.
In 2009, Changyao Chen et al. [16] investigated the performance of
monolayer graphene nanomechanical resonators with electrical readout using
COMSOL Multiphysics software.
D. Carrier and J. J. Dubowski [17] in 2009 presented a new model for
nanobiosensor using COMSOL Multiphysics software.
In 2010, F Landolsi and F H Ghorbel [18] proposed a new AFM probe design
suitable for nanomanipulation and simulations using COMSOL Multiphysics and
showed that the largest displacements of the proposed cantilever are experienced by
the tip and tip holder.
Zyad Hassan et al. [19] in 2010 designed ThermalScope for full-chip thermal
analysis of billion-transistor nanometer scale IC designs, with accuracy at the scale
of individual devices and adopted COMSOL Multiphysics software to evaluate the
chip–package- and functional-unit-level modeling accuracy of ThermalScope.
Silvan Schmid et al. [20] in 2010, adopting COMSOL Multiphysics software
presented an analytical model for a transduction scheme based on the Kelvin
polarization force and the presented model facilitates the design of microresonators
and nanoresonators with dielectric actuation, which offers a great freedom in the
choice of structural material.
Mikhail Erdmanis et al. [21] in 2011 presented a surface-plasmon resonance
sensor based on an H-shaped optical fibre adopting COMSOL Multiphysics software
and showed that a refractive-index resolution of up to 5·103 nm/RIU can be
obtained.
S. A. Selvin et al. [22] in 2011 adopting COMSOL Multiphysics software
designed a carbon nanotube (CNT) based piezoresistive pressure sensor with
optimization of design parameters such as diaphragm, nanotube dimensions and
sensitivity.
 42

Pieter Neutens et al. [23] in 2012 reported the design of a Nano-Slit

Plasmonic Detector based on gold nanolayer adopting COMSOL Multiphysics
software.
In 2012, Jyh Ming Wu et al. [24] demonstrated the working principle of the
nanogenerator using the COMSOL Multiphysics software package.
F. Afshinmanesh et al. [25] in 2012 proposed a plasmonic polarimeter
integrated with a Si photodetector that can determine the complete polarization state
of a beam of light and modeled the transmission of light through a linear slit in an Au
film on a Si substrate using COMSOL Multiphysics software.
Vladislav E. Demidov et al. [26] in 2012 characterized auto-oscillations
properties at moderate current densities, at a microwave frequency tunable over a
wide range adopting COMSOL Multiphysics software. These findings suggest a new
route for the implementation of nanoscale microwave sources for next generation
integrated electronics.
Yeongjin Lim [27] in 2013, adopting COMSOL Multiphysics software
studied the effects of geometry of the suspended carbon nanowire structures on the
electrochemical current signal for application as sensor.
Tural Khudiyev et al. [28] in 2014, adopting COMSOL Multiphysics pointed
out a hitherto-undescribed transitory region between nanowire and nanosphere
transformation, where a diverse array of complex quasi one-dimensional
nanostructures is produced by Rayleigh-Plateau instability-mediated deformation
during the progress of a combined ISR/thermal instability technique.
A. Steinbrϋck et al. [29] in 2014 studied local laser-induced heating in fluids
with gold nanodot arrays prepared by electron beam lithography that cover
resonances in the near infrared spectral range from 750 nm to 880 nm adopting
COMSOL Multiphysics software. The two dimensional simulations fit well with the
experimental results.
K. Sathiyamoorthy and Michael C. Kolios [30] in 2015 numerically
investigated plasmonic properties of polystyrene as core and gold as shell in water
and revealed that the electromagnetic field distribution of the blue end plasmonic
mode was associated mainly with core-metal dielectric interface whereas the longer
 43

red end plasmonic mode was associated with surface of core-shell and was strongly
influenced by its surrounding dielectric medium.
In 2016, Xin Huang et al. [31] presented a numerical simulation for guiding
future 2D piezotronic nanodevice design.
Leiming Wang et al. [32] in 2016 studied the plasmon coupling effect at the
metal cathode adopting COMSOL Multiphysics software and the results suggest
that ~50% reduction of plasmon energy loss over a broadband emission is possible
by optimized metal cathode structures.
Han Wang et al. in 2016 [33] proposed a quantum well (QW) – embedded
nanomembrane in a rolled-up geometry faciliting a 3D QW infrared photodetector
(QWIP) device with enhanced responsivity and detectivity adopting COMSOL
Multiphysics software.

Satoshi Ishii et al. [34] in 2017 presented the numerical studies adopting
COMSOL Multiphysics software to understand the optical and thermal properties of
the wavelength-selective spin current generation by combining plasmonic absorbers
with Pt/yttrium-iron-garnet spintronic devices.

In 2017 using COMSOL Multiphysics software, Diego Felipe Santos et al.

[35] proposed a refractive index sensor based on surface plasmon resonance (SPR) in
a gold wire partially incrusted on the surface of a D-type fibre and in contact with the
external medium for increased sensitivity and roughness.

Y. M. El-Toukhy et al. [36] in 2017 adopting COMSOL Multiphysics

software proposed a design of tapered-dipole nanoantenna of Ag nanoparticle for
energy harvesting applications and at 500nm, harvesting efficiency of 60% is
achieved, which is higher than that of conventional dipole component by 34%.

Nicolas Abadia et al. [37] in 2018 analyzed the optical and thermal behavior
of an antenna-based near-field transducer (NFT) using a tapered, hybrid plasmonic
waveguide for heat-assisted magnetic recording (HAMR) adopting COMSOL
Multiphysics Software.
 44

2.6 Comparison of COMSOL and other related softwares

2.6.1 Application of COMSOL Multiphysics for modeling Plasmonic
Nanoantenna and Nanofilter
Though using Finite Difference Time Domain (FDTD) technique (e.g. CST
Microwave Studio), dielectric structures such as photonic crystals can be modeled
reasonably well, it has severe drawbacks for modeling plasmonic structures. The
high localization of the field at the metal-dielectric interfaces requires much finer
grid resolution than dielectric structures and hence the rectangular grid typically used
in FDTD is not suitable for complex plasmonic structures modeling. The piecewise
constant approximation of the fields within grids in FDTD poses difficulties in
applying boundary condition at the metal-dielectric interfaces, which is crucial for
modeling the plasmonic structures [38]. In FDTD and other time domain methods,
the dielectric functions of the material need to be approximated by proper analytical
expressions which give rise to considerable error in broadband calculations [38]. For
other than photonic crystal Local Density of States (LDOS) calculation present
ሬԦ* ሬሬሬԦ
additional challenges for FDTD, because it is difficult to transform ‫ܬ‬ ‫ ܧ‬from time
to frequency domain for arbitrary shaped plasmonic structures as explained by
Koenderink et al. [39]. However, Finite Element Method (FEM) does not have such
challenges due to the more advanced discretization strategy for complex geometric
structures by using a variety of elements of different shapes, and FEM is a frequency
domain method which can handle arbitrary material dispersion [40] with different
types of boundary conditions. LDOS can be efficiently extracted from the FEM
numerical calculations using COMSOL Multiphysics software, without the
numerical difficulties encountered in FDTD, when the boundary condition of the
modeling domain is properly handled [38]. Software users might consider options
like Green’s dyadic function [41] for nanostructures modeling but it encounters
difficulties in constructing Green’s functions for complex geometries.
ANSYS suite of engineering simulation software [42] is a robust and widely
accepted industrial standard Finite Element Analysis tool. Its capabilities are very
strong with the inbuilt physics models but in terms of providing flexibility of user-
defined Partial Differential Equations, it falls behind to COMSOL Multiphysics
software. COMSOL is more user friendly compared to ANSYS. For investigation of
 45

a problem at a fundamental level and deep access to the physics involved, COMSOL
is far better than ANSYS. Mesh generation in COMSOL is smoother compared to
ANSYS. It provides more tools in easier way to use compared to ANSYS.
ANSOFT’s High Frequency Structure Simulator (HFSS) [43], a full wave 3D
electromagnetic simulator basically designed for passive device modeling, has been
used by a variety of researchers for three dimensional structures. However, the
introduction of dispersion into the HFSS models is very problematic because of the
required number of simulations as well as the requisite fineness of the frequency
steps [44]. The axisymmetric ability of COMSOL Multiphysics software has
advantages for dispersion studies associated with the electrically small antenna
systems.
Method of Moment (MOM) [4] is another powerful numerical technique for
solving boundary-value problems in electromagnetics. The method is more efficient
when the object to be dealt with it is perfect electric conductor (PEC), impenetrable
or homogeneous. This method is not well suited for the analysis of complex
inhomogeneous geometries.
When the radiation from antennas is considered, the problem of truncating
infinite space is especially important to know about the far field pattern. The sources
of electromagnetic waves are located on the antennas and in MOM, the sources are
considered as the primary field and only the areas where charges & current present
are discretized. But when lossy materials i.e. non-perfect conductors or non-perfect
dielectrics are considered, MOM codes must discretize the entire domains where
current flow. But software designed for an RF antenna may not be appropriate for
plasmonic antenna because RF antenna exhibits so small material loss which is not
the case for higher frequency antenna. Plasmonic antenna resonates when the
structure is small compared with the wavelength leading to quasi-static effects which
is not the case in the standard RF antenna design. Therefore MOM codes are not well
prepared for the plasmonic analysis. Since mesh refinement is highly desirable, FEM
based COMSOL is the most suited software.
The Wave Optics module of COMSOL is used for nanoantenna design and
ACDC & RF module is used for nanofilter design in the present work.
 46

2.6.2 Application of COMSOL Multiphysics for mem-device modeling

The Semiconductor Module is used for modeling semiconductor devices
with length scales of nanometer or more, which is modeled with semi-classical drift-
diffusion approach using partial differential equations. Within the product, there are
a number of physics interfaces – tools for receiving model inputs to describe a set of
physical equations and boundary conditions. These include interfaces for modeling
the transport of electrons and holes in semiconductor devices, the electrostatic
behavior of such, and an interface for coupling semiconductor simulations to a
SPICE circuit simulation, which is one of the great advantages for memristor
modeling using COMSOL Multiphysics software.
The Semiconductor interface solves Poisson’s equation in conjunction with
the continuity equations for the charge carriers. It solves for both the electron and
hole concentrations explicitly. One can choose to solve the model either with the
finite volume method or the finite element method. The Semiconductor interface
includes material models for semiconducting and insulating materials, in addition to
boundary conditions for ohmic contacts, Schottky contacts, gates, and a wide range
of electrostatics boundary conditions.
The Semiconductor Module includes several predefined mobility models and
the option to create custom, user-defined mobility models. Both these types of
models can be combined in arbitrary ways. Each mobility model defines an output
electron and hole mobility. The output mobility can be used as an input to other
mobility models, while equations can be used to combine mobilities, for example
using Matthiessen's rule. The Semiconductor interface also contains features to add
Auger, Direct, and Shockley-Read Hall recombination to a semiconducting domain
or one can specify own recombination rate.
Specifying the doping distribution is critical for the modeling of
semiconductor devices. The Semiconductor Module of COMSOL Multiphysics
software provides a Doping model feature to do this. Constant and user-defined
doping profiles can be specified, or an approximate Gaussian doping profile can be
used. It is also straightforward to import data from external sources into COMSOL
Multiphysics, which can be treated by built-in interpolation functions. This facility
 47

of COMSOL also makes it a very advantageous and user friendly memristor

modeling simulation tool.
Along with the Semiconductor interface, the Semiconductor module comes
prepared with enhanced electrostatics capabilities, available both within the
Semiconductor interface and in a standalone Electrostatics interface. System level
and mixed device simulations are enabled through a physics interface for electrical
circuits with SPICE import capability. These features make it unique for simulating
the application based memristor models.
Correctness and accuracy of the output results of the simulation can be adhere
as one can change the system geometry, material or other parameters easily to design
prototype of the device to ensure the quality performance. Compared to the widely
studied SPICE models of charge- and flux-controlled memristors, with only
capacitance and resistance forming the input data of such modeling, the memristor
designed by COMSOL Multiphysics software will be more versatile as well as
realistic as it will solve the equations at every nodes of the system geometry. That
guarantees that the models will exactly exhibit all the mem-behavior taking account
of ‫ܮ‬, ‫ܥ‬, ܴ compared to the already reported SPICE model [45].

Therefore, Semiconductor module of COMSOL Multiphysics software is

used for mem-device modeling in the present work.

2.7 Flowchart of the as-modeled nanodevices in the present work

Based on the above study, COMSOL Multiphysics software has been adopted
in the present work to design core-shell nanocomposite based optical nanoantenna, to
study the mem-behaviour of quantum dot and to model a core-shell nanocomposites
based filter for microwave frequency. The flow chart for optical nanoantenna,
nanoscale electronic filter and mem-devices designed in the present work using
COMSOL Multiphysics software is given below (figures 2.3 - 2.5).
 48

*
PML, PEC and PMC stands for Perfectly Mathced Layer, Perfect Electric
Conductor and Perfect Magnetic Conductor, respectively.
Figure 2.3: Flow chart of the optical nanoantenna simulation process in the present
work.
 49

*
PEC stands for Perfect Electric Conductor
Figure 2.4: Flow chart of the nanoscale electronic filter simulation process in the
present work.
 50

Figure 2.5: Flow chart of the mem - devices simulation process in the present work.

The next chapter presents details of theoretical as well as experimental work

done so far on the Optical Yagi-Uda nanoantenna, Nanoscale electronic filter &
Mem-devices.
 51

References
[1] Introduction to COMSOL Multiphysics4.2,www.comsol.com
[2] https://en.wikipedia.org/wiki/COMSOL_Multiphysics
[3] Introduction to COMSOL Multiphysics4.4,www.comsol.com
[4] J. Jin, The Finite Element Method in Electromagnetics, Second
Edition, (John Wiley & Sons, Interscience, 2002).
[5] H. W. David and M. N. Ozisk, Heat Conduction, Third Edition,
(Wiley,2012). ISBN 978-0-470-90293 - 6.
[6] C. H. Wilcox, Commun. Pure Appl. Math. 14, 427(1961).
[7] J. P. Berenger, J. Comput. Phys. 114, 185(1994).
[8] F. L. Teixeria and W.C. Chew, Advances in the theory of perfectly
matched layers, in Fast and Efficient Algorithms in Computational
Electromagnetics, (Artech House, 2001).
[9] J. M. Jin, X. Q. Sheng and W. C. Chew, Microwave Opt. Tech. Lett.
14, 284 (1997).
[10] https://www.comsol.co.in/multiphysics/mesh-refinement
[11] A.Ghahremani and A. E. Fathy, Energy Science and Engineering, 3,
520 (2015).
[12] J. Liu et al., Applied Physics Letters 89, 063125 (2006).
[13] X. Xugang et al., Appl. Phys. Lett. 91, 063101 (2007).
[14] X. Xugang and B. Ahmed, J Nanopart Res 10, 947 (2008).
[15] A. Nojeh et al., in Proceedings of the IEEE 14th International Mixed -
Signals, Sensors, and Systems Test Workshop (2008).
DOI: 10.1109/IMS3TW.2008.4581628.
[16] D. Carrier and J. J. Dubowski, in Proceedings of the Annual
COMSOL Conference (2009).
https://www.comsol.jp/paper/download/44808/Carrier.pdf
[17] C.Chen et al., Nature Nanotechnology 4, 861(2009).
[18] F Landolsi and F H Ghorbel, Smart Mater. Struct. 19, 065028(2010).
[19] Z. Hassan et al., IEEE Transactions on Computer-Aided Design of
Integrated Circuits and Systems 28, 6(2009).
 52

[20] S. Schmid, C. Hierold, and A. Boisen, Journal of Applied Physics

107, 054510 (2010).
[21] M. Erdmanis et al., Optics Express 19, 13980(2011).
[22] S. A. Selvin et al., in Proceedings of the Annual COMSOL
Conference (2011).
https://www.comsol.jp/paper/download/84139/s_paper.pdf
[23] P. Neutens et al., Nanoplasmonic Sensors, Integrated Analytical
Systems, 369 (2012). DOI 10.1007/978-1-4614-3933-2_15.
[24] J. W. Ming et al., Adv. Mater. (2012). DOI: 10.1002/adma.201202445
[25] F. Afshinmanesh et al., Nanophotonics 1, 125 (2012).
[26] V. E. Demidov et al., Nature Materials 11, 1028 (2012).
[27] Y.Lim et al., Nanoscale Research Letters 8,492 (2013).
[28] T. Khudiyev, O. Tobail, and M. Bayindir, Nature Scientific Reports 4,
4864 (2014). DOI: 10.1038/srep04864.
[29] A. Steinbrϋck et al., RSC Advances 4, 61898(2014).
[30] K. Sathiyamoorthy and M. C. Kolios, Proc. of SPIE 9340, 93400V-
1 (2015).
[31] X. Huang et al., Nano Research 9, 282(2016).
[32] L. Wang and J. Amano, KONICA MINOLTA TECHNOLOGY
REPORT 13, 101(2016).
[33] H. Wang et al., Sci. Adv. 2, 1600027(2016).
[34] S. Ishii et al., APL PHOTONICS 2, 106103 (2017).
[35] D. F. Santos, A. Guerreiro, and J. M. Baptista, Optik-International
Journal for Light and Electron Optics 139, 244(2017).
[36] Y. M. El-Toukhy et al., Tapered Plasmonic Nanoantennas for Energy
Harvesting Applications, Chapter 13, Nanoplasmonics -
Fundamentals and Applications, (INTECH publisher, 2017).
[37] N. Abadia et al., Optics Express 26, 1752 (2018).
[38] Y. Chen et al., PHYSICAL REVIEW B 81, 125431(2010).
[39] A. F. Koenderink et al., Optics Letters 30, 3210(2005).
[40] Y. Xu, R. K. Lee, and A. Yariv, PHYSICAL REVIEW A 61,
033807 (2000).
 53

[41] C. H. Papas, Journal of Geophysical Research 68, 4(1963).

[42] https://www.ansys.com/
[43] https://www.ansys.com/en-in/products/electronics/ansys-hfss
[44] R.W. Ziolkowski, OPTO-ELECTRONICS REVIEW 14, 167 (2006).
[45] Z. Biolek, D. Biolek, and V. Biolková, RADIOENGINEERING 18,
2(2009).